1458 files changed, 137708 insertions, 85745 deletions

diff --git a/src/Makefile b/src/Makefile
index fab0069a716..4f65da0e3b7 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -33,7 +33,7 @@ $(TOP)/$(LIB_DIR):
 
 
 clean:
-	@for dir in $(SUBDIRS) ; do \
+	-@for dir in $(SUBDIRS) ; do \
 		if [ -d $$dir ] ; then \
 			(cd $$dir && $(MAKE) clean) ; \
 		fi \
diff --git a/src/driclient/include/driclient.h b/src/driclient/include/driclient.h
new file mode 100644
index 00000000000..d3915250392
--- /dev/null
+++ b/src/driclient/include/driclient.h
@@ -0,0 +1,97 @@
+#ifndef driclient_h
+#define driclient_h
+
+#include <stdint.h>
+#include <X11/Xlib.h>
+#include <drm_sarea.h>
+#include "xf86dri.h"
+
+/* TODO: Bring in DRI XML options */
+
+typedef struct dri_version
+{
+	int major;
+	int minor;
+	int patch;
+} dri_version_t;
+
+typedef struct dri_screen
+{
+	Display			*display;
+	unsigned int		num;
+	dri_version_t		ddx, dri, drm;
+	int			draw_lock_id;
+	int			fd;
+	drm_sarea_t		*sarea;
+	void			*drawable_hash;
+	void			*private;
+} dri_screen_t;
+
+struct dri_context;
+
+typedef struct dri_drawable
+{
+	drm_drawable_t		drm_drawable;
+	Drawable		x_drawable;
+	unsigned int		sarea_index;
+	unsigned int		*sarea_stamp;
+	unsigned int		last_sarea_stamp;
+	int			x, y, w, h;
+	int			back_x, back_y;
+	int			num_cliprects, num_back_cliprects;
+	drm_clip_rect_t		*cliprects, *back_cliprects;
+	dri_screen_t		*dri_screen;
+	unsigned int		refcount;
+	void			*private;
+} dri_drawable_t;
+
+typedef struct dri_context
+{
+	XID			id;
+	drm_context_t		drm_context;
+	dri_screen_t		*dri_screen;
+	void			*private;
+} dri_context_t;
+
+typedef struct dri_framebuffer
+{
+	drm_handle_t		drm_handle;
+	int			base, size, stride;
+	int			private_size;
+	void			*private;
+} dri_framebuffer_t;
+
+int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri_framebuffer_t *dri_framebuf);
+int driDestroyScreen(dri_screen_t *dri_screen);
+int driCreateDrawable(dri_screen_t *dri_screen, Drawable drawable, dri_drawable_t **dri_drawable);
+int driUpdateDrawableInfo(dri_drawable_t *dri_drawable);
+int driDestroyDrawable(dri_drawable_t *dri_drawable);
+int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **dri_context);
+int driDestroyContext(dri_context_t *dri_context);
+
+#define DRI_VALIDATE_DRAWABLE_INFO_ONCE(dri_drawable)					\
+do											\
+{											\
+	if (*(dri_drawable->sarea_stamp) != dri_drawable->last_sarea_stamp)		\
+		driUpdateDrawableInfo(dri_drawable);					\
+} while (0)
+
+#define DRI_VALIDATE_DRAWABLE_INFO(dri_screen, dri_drawable)					\
+do												\
+{												\
+	while (*(dri_drawable->sarea_stamp) != dri_drawable->last_sarea_stamp)			\
+	{											\
+		register unsigned int hwContext = dri_screen->sarea->lock.lock &		\
+		~(DRM_LOCK_HELD | DRM_LOCK_CONT);						\
+		DRM_UNLOCK(dri_screen->fd, &dri_screen->sarea->lock, hwContext);		\
+												\
+		DRM_SPINLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id);	\
+		DRI_VALIDATE_DRAWABLE_INFO_ONCE(dri_drawable);					\
+		DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id);	\
+												\
+		DRM_LIGHT_LOCK(dri_screen->fd, &dri_screen->sarea->lock, hwContext);		\
+	}											\
+} while (0)
+
+#endif
+
diff --git a/src/driclient/include/xf86dri.h b/src/driclient/include/xf86dri.h
new file mode 100644
index 00000000000..baf80a7a9dd
--- /dev/null
+++ b/src/driclient/include/xf86dri.h
@@ -0,0 +1,119 @@
+/**************************************************************************
+
+Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+Copyright 2000 VA Linux Systems, Inc.
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sub license, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
+ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/**
+ * \file xf86dri.h
+ * Protocol numbers and function prototypes for DRI X protocol.
+ *
+ * \author Kevin E. Martin <martin@valinux.com>
+ * \author Jens Owen <jens@tungstengraphics.com>
+ * \author Rickard E. (Rik) Faith <faith@valinux.com>
+ */
+
+#ifndef _XF86DRI_H_
+#define _XF86DRI_H_
+
+#include <X11/Xfuncproto.h>
+#include <xf86drm.h>
+
+#define X_XF86DRIQueryVersion			0
+#define X_XF86DRIQueryDirectRenderingCapable	1
+#define X_XF86DRIOpenConnection			2
+#define X_XF86DRICloseConnection		3
+#define X_XF86DRIGetClientDriverName		4
+#define X_XF86DRICreateContext			5
+#define X_XF86DRIDestroyContext			6
+#define X_XF86DRICreateDrawable			7
+#define X_XF86DRIDestroyDrawable		8
+#define X_XF86DRIGetDrawableInfo		9
+#define X_XF86DRIGetDeviceInfo			10
+#define X_XF86DRIAuthConnection                 11
+#define X_XF86DRIOpenFullScreen                 12   /* Deprecated */
+#define X_XF86DRICloseFullScreen                13   /* Deprecated */
+
+#define XF86DRINumberEvents		0
+
+#define XF86DRIClientNotLocal		0
+#define XF86DRIOperationNotSupported	1
+#define XF86DRINumberErrors		(XF86DRIOperationNotSupported + 1)
+
+#ifndef _XF86DRI_SERVER_
+
+_XFUNCPROTOBEGIN
+
+Bool XF86DRIQueryExtension( Display *dpy, int *event_base, int *error_base );
+
+Bool XF86DRIQueryVersion( Display *dpy, int *majorVersion, int *minorVersion,
+    int *patchVersion );
+
+Bool XF86DRIQueryDirectRenderingCapable( Display *dpy, int screen,
+    Bool *isCapable );
+
+Bool XF86DRIOpenConnection( Display *dpy, int screen, drm_handle_t *hSAREA,
+    char **busIDString );
+
+Bool XF86DRIAuthConnection( Display *dpy, int screen, drm_magic_t magic );
+
+Bool XF86DRICloseConnection( Display *dpy, int screen );
+
+Bool XF86DRIGetClientDriverName( Display *dpy, int screen,
+    int *ddxDriverMajorVersion, int *ddxDriverMinorVersion,
+    int *ddxDriverPatchVersion, char **clientDriverName );
+
+Bool XF86DRICreateContext( Display *dpy, int screen, Visual *visual,
+    XID *ptr_to_returned_context_id, drm_context_t *hHWContext );
+
+Bool XF86DRICreateContextWithConfig( Display *dpy, int screen, int configID,
+    XID *ptr_to_returned_context_id, drm_context_t *hHWContext );
+
+Bool XF86DRIDestroyContext( Display *dpy, int screen,
+    XID context_id );
+
+Bool XF86DRICreateDrawable( Display *dpy, int screen,
+    Drawable drawable, drm_drawable_t *hHWDrawable );
+
+Bool XF86DRIDestroyDrawable( Display *dpy, int screen, 
+    Drawable drawable);
+
+Bool XF86DRIGetDrawableInfo( Display *dpy, int screen, Drawable drawable,
+    unsigned int *index, unsigned int *stamp, 
+    int *X, int *Y, int *W, int *H,
+    int *numClipRects, drm_clip_rect_t ** pClipRects,
+    int *backX, int *backY,
+    int *numBackClipRects, drm_clip_rect_t **pBackClipRects );
+
+Bool XF86DRIGetDeviceInfo( Display *dpy, int screen,
+    drm_handle_t *hFrameBuffer, int *fbOrigin, int *fbSize,
+    int *fbStride, int *devPrivateSize, void **pDevPrivate );
+
+_XFUNCPROTOEND
+
+#endif /* _XF86DRI_SERVER_ */
+
+#endif /* _XF86DRI_H_ */
+
diff --git a/src/driclient/src/Makefile b/src/driclient/src/Makefile
new file mode 100644
index 00000000000..34435a2086e
--- /dev/null
+++ b/src/driclient/src/Makefile
@@ -0,0 +1,19 @@
+TARGET		= libdriclient.a
+OBJECTS		= driclient.o XF86dri.o
+DRMDIR		?= /usr
+
+CFLAGS		+= -g -Wall -fPIC -I../include -I${DRMDIR}/include -I${DRMDIR}/include/drm
+
+#############################################
+
+.PHONY	= all clean
+
+all: ${TARGET}
+
+${TARGET}: ${OBJECTS}
+	ar rcs $@ $^
+	if ! test -d ../lib; then mkdir ../lib; fi
+	cp ${TARGET} ../lib
+
+clean:
+	rm -rf ${OBJECTS} ${TARGET} ../lib/${TARGET}
diff --git a/src/driclient/src/XF86dri.c b/src/driclient/src/XF86dri.c
new file mode 100644
index 00000000000..9e359a92384
--- /dev/null
+++ b/src/driclient/src/XF86dri.c
@@ -0,0 +1,619 @@
+/**************************************************************************
+
+Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+Copyright 2000 VA Linux Systems, Inc.
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sub license, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
+ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/*
+ * Authors:
+ *   Kevin E. Martin <martin@valinux.com>
+ *   Jens Owen <jens@tungstengraphics.com>
+ *   Rickard E. (Rik) Faith <faith@valinux.com>
+ *
+ */
+
+/* THIS IS NOT AN X CONSORTIUM STANDARD */
+
+#define NEED_REPLIES
+#include <X11/Xlibint.h>
+#include <X11/extensions/Xext.h>
+#include <X11/extensions/extutil.h>
+#include "xf86dristr.h"
+
+static XExtensionInfo _xf86dri_info_data;
+static XExtensionInfo *xf86dri_info = &_xf86dri_info_data;
+static char xf86dri_extension_name[] = XF86DRINAME;
+
+#define XF86DRICheckExtension(dpy,i,val) \
+  XextCheckExtension (dpy, i, xf86dri_extension_name, val)
+
+/*****************************************************************************
+ *                                                                           *
+ *			   private utility routines                          *
+ *                                                                           *
+ *****************************************************************************/
+
+static int close_display(Display *dpy, XExtCodes *extCodes);
+static /* const */ XExtensionHooks xf86dri_extension_hooks = {
+    NULL,				/* create_gc */
+    NULL,				/* copy_gc */
+    NULL,				/* flush_gc */
+    NULL,				/* free_gc */
+    NULL,				/* create_font */
+    NULL,				/* free_font */
+    close_display,			/* close_display */
+    NULL,				/* wire_to_event */
+    NULL,				/* event_to_wire */
+    NULL,				/* error */
+    NULL,				/* error_string */
+};
+
+static XEXT_GENERATE_FIND_DISPLAY (find_display, xf86dri_info, 
+				   xf86dri_extension_name, 
+				   &xf86dri_extension_hooks, 
+				   0, NULL)
+
+static XEXT_GENERATE_CLOSE_DISPLAY (close_display, xf86dri_info)
+
+
+/*****************************************************************************
+ *                                                                           *
+ *		    public XFree86-DRI Extension routines                    *
+ *                                                                           *
+ *****************************************************************************/
+
+#if 0
+#include <stdio.h>
+#define TRACE(msg)  fprintf(stderr,"XF86DRI%s\n", msg);
+#else
+#define TRACE(msg)
+#endif
+
+#define PUBLIC
+
+PUBLIC Bool XF86DRIQueryExtension (dpy, event_basep, error_basep)
+    Display *dpy;
+    int *event_basep, *error_basep;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+
+    TRACE("QueryExtension...");
+    if (XextHasExtension(info)) {
+	*event_basep = info->codes->first_event;
+	*error_basep = info->codes->first_error;
+        TRACE("QueryExtension... return True");
+	return True;
+    } else {
+        TRACE("QueryExtension... return False");
+	return False;
+    }
+}
+
+PUBLIC Bool XF86DRIQueryVersion(dpy, majorVersion, minorVersion, patchVersion)
+    Display* dpy;
+    int* majorVersion; 
+    int* minorVersion;
+    int* patchVersion;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIQueryVersionReply rep;
+    xXF86DRIQueryVersionReq *req;
+
+    TRACE("QueryVersion...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIQueryVersion, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIQueryVersion;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("QueryVersion... return False");
+	return False;
+    }
+    *majorVersion = rep.majorVersion;
+    *minorVersion = rep.minorVersion;
+    *patchVersion = rep.patchVersion;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("QueryVersion... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIQueryDirectRenderingCapable(dpy, screen, isCapable)
+    Display* dpy;
+    int screen;
+    Bool* isCapable;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIQueryDirectRenderingCapableReply rep;
+    xXF86DRIQueryDirectRenderingCapableReq *req;
+
+    TRACE("QueryDirectRenderingCapable...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIQueryDirectRenderingCapable, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIQueryDirectRenderingCapable;
+    req->screen = screen;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("QueryDirectRenderingCapable... return False");
+	return False;
+    }
+    *isCapable = rep.isCapable;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("QueryDirectRenderingCapable... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIOpenConnection(dpy, screen, hSAREA, busIdString)
+    Display* dpy;
+    int screen;
+    drm_handle_t * hSAREA;
+    char **busIdString;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIOpenConnectionReply rep;
+    xXF86DRIOpenConnectionReq *req;
+
+    TRACE("OpenConnection...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIOpenConnection, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIOpenConnection;
+    req->screen = screen;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("OpenConnection... return False");
+	return False;
+    }
+
+    *hSAREA = rep.hSAREALow;
+    if (sizeof(drm_handle_t) == 8) {
+       int shift = 32; /* var to prevent warning on next line */
+       *hSAREA |= ((drm_handle_t) rep.hSAREAHigh) << shift;
+    }
+
+    if (rep.length) {
+        if (!(*busIdString = (char *)Xcalloc(rep.busIdStringLength + 1, 1))) {
+            _XEatData(dpy, ((rep.busIdStringLength+3) & ~3));
+            UnlockDisplay(dpy);
+            SyncHandle();
+            TRACE("OpenConnection... return False");
+            return False;
+        }
+	_XReadPad(dpy, *busIdString, rep.busIdStringLength);
+    } else {
+        *busIdString = NULL;
+    }
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("OpenConnection... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIAuthConnection(dpy, screen, magic)
+    Display* dpy;
+    int screen;
+    drm_magic_t magic;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIAuthConnectionReq *req;
+    xXF86DRIAuthConnectionReply rep;
+
+    TRACE("AuthConnection...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIAuthConnection, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIAuthConnection;
+    req->screen = screen;
+    req->magic = magic;
+    rep.authenticated = 0;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse) || !rep.authenticated) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("AuthConnection... return False");
+	return False;
+    }
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("AuthConnection... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRICloseConnection(dpy, screen)
+    Display* dpy;
+    int screen;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRICloseConnectionReq *req;
+
+    TRACE("CloseConnection...");
+
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRICloseConnection, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRICloseConnection;
+    req->screen = screen;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("CloseConnection... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIGetClientDriverName(dpy, screen, ddxDriverMajorVersion, 
+	ddxDriverMinorVersion, ddxDriverPatchVersion, clientDriverName)
+    Display* dpy;
+    int screen;
+    int* ddxDriverMajorVersion;
+    int* ddxDriverMinorVersion;
+    int* ddxDriverPatchVersion;
+    char** clientDriverName;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIGetClientDriverNameReply rep;
+    xXF86DRIGetClientDriverNameReq *req;
+
+    TRACE("GetClientDriverName...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIGetClientDriverName, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIGetClientDriverName;
+    req->screen = screen;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("GetClientDriverName... return False");
+	return False;
+    }
+
+    *ddxDriverMajorVersion = rep.ddxDriverMajorVersion;
+    *ddxDriverMinorVersion = rep.ddxDriverMinorVersion;
+    *ddxDriverPatchVersion = rep.ddxDriverPatchVersion;
+
+    if (rep.length) {
+        if (!(*clientDriverName = (char *)Xcalloc(rep.clientDriverNameLength + 1, 1))) {
+            _XEatData(dpy, ((rep.clientDriverNameLength+3) & ~3));
+            UnlockDisplay(dpy);
+            SyncHandle();
+            TRACE("GetClientDriverName... return False");
+            return False;
+        }
+	_XReadPad(dpy, *clientDriverName, rep.clientDriverNameLength);
+    } else {
+        *clientDriverName = NULL;
+    }
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("GetClientDriverName... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRICreateContextWithConfig(dpy, screen, configID, context,
+	hHWContext)
+    Display* dpy;
+    int screen;
+    int configID;
+    XID* context;
+    drm_context_t * hHWContext;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRICreateContextReply rep;
+    xXF86DRICreateContextReq *req;
+
+    TRACE("CreateContext...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRICreateContext, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRICreateContext;
+    req->visual = configID;
+    req->screen = screen;
+    *context = XAllocID(dpy);
+    req->context = *context;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("CreateContext... return False");
+	return False;
+    }
+    *hHWContext = rep.hHWContext;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("CreateContext... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRICreateContext(dpy, screen, visual, context, hHWContext)
+    Display* dpy;
+    int screen;
+    Visual* visual;
+    XID* context;
+    drm_context_t * hHWContext;
+{
+    return XF86DRICreateContextWithConfig( dpy, screen, visual->visualid,
+					   context, hHWContext );
+}
+
+PUBLIC Bool XF86DRIDestroyContext( Display * ndpy, int screen, 
+    XID context )
+{
+    Display * const dpy = (Display *) ndpy;
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIDestroyContextReq *req;
+
+    TRACE("DestroyContext...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIDestroyContext, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIDestroyContext;
+    req->screen = screen;
+    req->context = context;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("DestroyContext... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRICreateDrawable( Display * ndpy, int screen, 
+    Drawable drawable, drm_drawable_t * hHWDrawable )
+{
+    Display * const dpy = (Display *) ndpy;
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRICreateDrawableReply rep;
+    xXF86DRICreateDrawableReq *req;
+
+    TRACE("CreateDrawable...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRICreateDrawable, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRICreateDrawable;
+    req->screen = screen;
+    req->drawable = drawable;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("CreateDrawable... return False");
+	return False;
+    }
+    *hHWDrawable = rep.hHWDrawable;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("CreateDrawable... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIDestroyDrawable( Display * ndpy, int screen,
+    Drawable drawable )
+{
+    Display * const dpy = (Display *) ndpy;
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIDestroyDrawableReq *req;
+
+    TRACE("DestroyDrawable...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIDestroyDrawable, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIDestroyDrawable;
+    req->screen = screen;
+    req->drawable = drawable;
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("DestroyDrawable... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIGetDrawableInfo(Display* dpy, int screen, Drawable drawable,
+    unsigned int* index, unsigned int* stamp,
+    int* X, int* Y, int* W, int* H,
+    int* numClipRects, drm_clip_rect_t ** pClipRects,
+    int* backX, int* backY,
+    int* numBackClipRects, drm_clip_rect_t ** pBackClipRects )
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIGetDrawableInfoReply rep;
+    xXF86DRIGetDrawableInfoReq *req;
+    int total_rects;
+
+    TRACE("GetDrawableInfo...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIGetDrawableInfo, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIGetDrawableInfo;
+    req->screen = screen;
+    req->drawable = drawable;
+
+    if (!_XReply(dpy, (xReply *)&rep, 1, xFalse)) 
+    {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("GetDrawableInfo... return False");
+	return False;
+    }
+    *index = rep.drawableTableIndex;
+    *stamp = rep.drawableTableStamp;
+    *X = (int)rep.drawableX;
+    *Y = (int)rep.drawableY;
+    *W = (int)rep.drawableWidth;
+    *H = (int)rep.drawableHeight;
+    *numClipRects = rep.numClipRects;
+    total_rects = *numClipRects;
+
+    *backX = rep.backX;
+    *backY = rep.backY;
+    *numBackClipRects = rep.numBackClipRects;
+    total_rects += *numBackClipRects;
+
+#if 0
+    /* Because of the fix in Xserver/GL/dri/xf86dri.c, this check breaks
+     * backwards compatibility (Because of the >> 2 shift) but the fix
+     * enables multi-threaded apps to work.
+     */
+    if (rep.length !=  ((((SIZEOF(xXF86DRIGetDrawableInfoReply) - 
+		       SIZEOF(xGenericReply) + 
+		       total_rects * sizeof(drm_clip_rect_t)) + 3) & ~3) >> 2)) {
+        _XEatData(dpy, rep.length);
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("GetDrawableInfo... return False");
+        return False;
+    }
+#endif
+
+    if (*numClipRects) {
+       int len = sizeof(drm_clip_rect_t) * (*numClipRects);
+
+       *pClipRects = (drm_clip_rect_t *)Xcalloc(len, 1);
+       if (*pClipRects) 
+	  _XRead(dpy, (char*)*pClipRects, len);
+    } else {
+        *pClipRects = NULL;
+    }
+
+    if (*numBackClipRects) {
+       int len = sizeof(drm_clip_rect_t) * (*numBackClipRects);
+
+       *pBackClipRects = (drm_clip_rect_t *)Xcalloc(len, 1);
+       if (*pBackClipRects) 
+	  _XRead(dpy, (char*)*pBackClipRects, len);
+    } else {
+        *pBackClipRects = NULL;
+    }
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("GetDrawableInfo... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIGetDeviceInfo(dpy, screen, hFrameBuffer, 
+	fbOrigin, fbSize, fbStride, devPrivateSize, pDevPrivate)
+    Display* dpy;
+    int screen;
+    drm_handle_t * hFrameBuffer;
+    int* fbOrigin;
+    int* fbSize;
+    int* fbStride;
+    int* devPrivateSize;
+    void** pDevPrivate;
+{
+    XExtDisplayInfo *info = find_display (dpy);
+    xXF86DRIGetDeviceInfoReply rep;
+    xXF86DRIGetDeviceInfoReq *req;
+
+    TRACE("GetDeviceInfo...");
+    XF86DRICheckExtension (dpy, info, False);
+
+    LockDisplay(dpy);
+    GetReq(XF86DRIGetDeviceInfo, req);
+    req->reqType = info->codes->major_opcode;
+    req->driReqType = X_XF86DRIGetDeviceInfo;
+    req->screen = screen;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+        TRACE("GetDeviceInfo... return False");
+	return False;
+    }
+
+    *hFrameBuffer = rep.hFrameBufferLow;
+    if (sizeof(drm_handle_t) == 8) {
+       int shift = 32; /* var to prevent warning on next line */
+       *hFrameBuffer |= ((drm_handle_t) rep.hFrameBufferHigh) << shift;
+    }
+
+    *fbOrigin = rep.framebufferOrigin;
+    *fbSize = rep.framebufferSize;
+    *fbStride = rep.framebufferStride;
+    *devPrivateSize = rep.devPrivateSize;
+
+    if (rep.length) {
+        if (!(*pDevPrivate = (void *)Xcalloc(rep.devPrivateSize, 1))) {
+            _XEatData(dpy, ((rep.devPrivateSize+3) & ~3));
+            UnlockDisplay(dpy);
+            SyncHandle();
+            TRACE("GetDeviceInfo... return False");
+            return False;
+        }
+	_XRead(dpy, (char*)*pDevPrivate, rep.devPrivateSize);
+    } else {
+        *pDevPrivate = NULL;
+    }
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+    TRACE("GetDeviceInfo... return True");
+    return True;
+}
+
+PUBLIC Bool XF86DRIOpenFullScreen(dpy, screen, drawable)
+    Display* dpy;
+    int screen;
+    Drawable drawable;
+{
+    /* This function and the underlying X protocol are deprecated.
+     */
+    (void) dpy;
+    (void) screen;
+    (void) drawable;
+    return False;
+}
+
+PUBLIC Bool XF86DRICloseFullScreen(dpy, screen, drawable)
+    Display* dpy;
+    int screen;
+    Drawable drawable;
+{
+    /* This function and the underlying X protocol are deprecated.
+     */
+    (void) dpy;
+    (void) screen;
+    (void) drawable;
+    return True;
+}
+
diff --git a/src/driclient/src/driclient.c b/src/driclient/src/driclient.c
new file mode 100644
index 00000000000..2d7514166fa
--- /dev/null
+++ b/src/driclient/src/driclient.c
@@ -0,0 +1,295 @@
+#include "driclient.h"
+#include <assert.h>
+#include <stdlib.h>
+
+int driCreateScreen(Display *display, int screen, dri_screen_t **dri_screen, dri_framebuffer_t *dri_framebuf)
+{
+	int		evbase, errbase;
+	char		*driver_name;
+	int		newly_opened;
+	drm_magic_t	magic;
+	drmVersionPtr	drm_version;
+	drm_handle_t	sarea_handle;
+	char		*bus_id;
+	dri_screen_t	*dri_scrn;
+
+	assert(display);
+	assert(dri_screen);
+
+	if (!XF86DRIQueryExtension(display, &evbase, &errbase))
+		return 1;
+
+	dri_scrn = calloc(1, sizeof(dri_screen_t));
+
+	if (!dri_scrn)
+		return 1;
+
+	if (!XF86DRIQueryVersion(display, &dri_scrn->dri.major, &dri_scrn->dri.minor, &dri_scrn->dri.patch))
+		goto free_screen;
+
+	dri_scrn->display = display;
+	dri_scrn->num = screen;
+	dri_scrn->draw_lock_id = 1;
+
+	if (!XF86DRIOpenConnection(display, screen, &sarea_handle, &bus_id))
+		goto free_screen;
+
+	dri_scrn->fd = -1;
+	dri_scrn->fd = drmOpenOnce(NULL, bus_id, &newly_opened);
+	XFree(bus_id);
+
+	if (dri_scrn->fd < 0)
+		goto close_connection;
+
+	if (drmGetMagic(dri_scrn->fd, &magic))
+		goto close_drm;
+
+	drm_version = drmGetVersion(dri_scrn->fd);
+
+	if (!drm_version)
+		goto close_drm;
+
+	dri_scrn->drm.major = drm_version->version_major;
+	dri_scrn->drm.minor = drm_version->version_minor;
+	dri_scrn->drm.patch = drm_version->version_patchlevel;
+	drmFreeVersion(drm_version);
+
+	if (!XF86DRIAuthConnection(display, screen, magic))
+		goto close_drm;
+
+	if (!XF86DRIGetClientDriverName
+	(
+		display,
+		screen,
+		&dri_scrn->ddx.major,
+		&dri_scrn->ddx.minor,
+		&dri_scrn->ddx.patch,
+		&driver_name
+	))
+		goto close_drm;
+
+	if (drmMap(dri_scrn->fd, sarea_handle, SAREA_MAX, (drmAddress)&dri_scrn->sarea))
+		goto close_drm;
+
+	dri_scrn->drawable_hash = drmHashCreate();
+
+	if (!dri_scrn->drawable_hash)
+		goto unmap_sarea;
+
+	if (dri_framebuf)
+	{
+		if (!XF86DRIGetDeviceInfo
+		(
+			display,
+			screen, &dri_framebuf->drm_handle,
+			&dri_framebuf->base,
+			&dri_framebuf->size,
+			&dri_framebuf->stride,
+			&dri_framebuf->private_size,
+			&dri_framebuf->private
+		))
+			goto destroy_hash;
+	}
+
+	*dri_screen = dri_scrn;
+
+	return 0;
+
+destroy_hash:
+	drmHashDestroy(dri_scrn->drawable_hash);
+unmap_sarea:
+	drmUnmap(dri_scrn->sarea, SAREA_MAX);
+close_drm:
+	drmCloseOnce(dri_scrn->fd);
+close_connection:
+	XF86DRICloseConnection(display, screen);
+free_screen:
+	free(dri_scrn);
+
+	return 1;
+}
+
+int driDestroyScreen(dri_screen_t *dri_screen)
+{
+	assert(dri_screen);
+
+	drmHashDestroy(dri_screen->drawable_hash);
+	drmUnmap(dri_screen->sarea, SAREA_MAX);
+	drmCloseOnce(dri_screen->fd);
+	XF86DRICloseConnection(dri_screen->display, dri_screen->num);
+	free(dri_screen);
+
+	return 0;
+}
+
+int driCreateDrawable(dri_screen_t *dri_screen, Drawable drawable, dri_drawable_t **dri_drawable)
+{
+	int		evbase, errbase;
+	dri_drawable_t	*dri_draw;
+
+	assert(dri_screen);
+	assert(dri_drawable);
+
+	if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase))
+		return 1;
+
+	if (!drmHashLookup(dri_screen->drawable_hash, drawable, (void**)dri_drawable))
+	{
+		/* Found */
+		(*dri_drawable)->refcount++;
+		return 0;
+	}
+
+	dri_draw = calloc(1, sizeof(dri_drawable_t));
+
+	if (!dri_draw)
+		return 1;
+
+	if (!XF86DRICreateDrawable(dri_screen->display, 0, drawable, &dri_draw->drm_drawable))
+	{
+		free(dri_draw);
+		return 1;
+	}
+
+	dri_draw->x_drawable = drawable;
+	dri_draw->sarea_index = 0;
+	dri_draw->sarea_stamp = NULL;
+	dri_draw->last_sarea_stamp = 0;
+	dri_draw->dri_screen = dri_screen;
+	dri_draw->refcount = 1;
+
+	if (drmHashInsert(dri_screen->drawable_hash, drawable, dri_draw))
+	{
+		XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable);
+		free(dri_draw);
+		return 1;
+	}
+
+	if (!dri_draw->sarea_stamp || *dri_draw->sarea_stamp != dri_draw->last_sarea_stamp)
+	{
+		DRM_SPINLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id);
+
+		if (driUpdateDrawableInfo(dri_draw))
+		{
+			XF86DRIDestroyDrawable(dri_screen->display, dri_screen->num, drawable);
+			free(dri_draw);
+			DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id);
+			return 1;
+		}
+
+		DRM_SPINUNLOCK(&dri_screen->sarea->drawable_lock, dri_screen->draw_lock_id);
+	}
+
+	*dri_drawable = dri_draw;
+
+	return 0;
+}
+
+int driUpdateDrawableInfo(dri_drawable_t *dri_drawable)
+{
+	assert(dri_drawable);
+
+	if (dri_drawable->cliprects)
+	{
+		XFree(dri_drawable->cliprects);
+		dri_drawable->cliprects = NULL;
+	}
+	if (dri_drawable->back_cliprects)
+	{
+		XFree(dri_drawable->back_cliprects);
+		dri_drawable->back_cliprects = NULL;
+	}
+
+	DRM_SPINUNLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id);
+
+	if (!XF86DRIGetDrawableInfo
+	(
+		dri_drawable->dri_screen->display,
+		dri_drawable->dri_screen->num,
+		dri_drawable->x_drawable,
+    		&dri_drawable->sarea_index,
+    		&dri_drawable->last_sarea_stamp,
+    		&dri_drawable->x,
+    		&dri_drawable->y,
+    		&dri_drawable->w,
+    		&dri_drawable->h,
+		&dri_drawable->num_cliprects,
+		&dri_drawable->cliprects,
+		&dri_drawable->back_x,
+		&dri_drawable->back_y,
+		&dri_drawable->num_back_cliprects,
+		&dri_drawable->back_cliprects
+	))
+	{
+		dri_drawable->sarea_stamp = &dri_drawable->last_sarea_stamp;
+		dri_drawable->num_cliprects = 0;
+		dri_drawable->cliprects = NULL;
+		dri_drawable->num_back_cliprects = 0;
+		dri_drawable->back_cliprects = 0;
+
+		return 1;
+	}
+	else
+		dri_drawable->sarea_stamp = &dri_drawable->dri_screen->sarea->drawableTable[dri_drawable->sarea_index].stamp;
+
+	DRM_SPINLOCK(&dri_drawable->dri_screen->sarea->drawable_lock, dri_drawable->dri_screen->draw_lock_id);
+
+	return 0;
+}
+
+int driDestroyDrawable(dri_drawable_t *dri_drawable)
+{
+	assert(dri_drawable);
+
+	if (--dri_drawable->refcount == 0)
+	{
+		if (dri_drawable->cliprects)
+			XFree(dri_drawable->cliprects);
+		if (dri_drawable->back_cliprects)
+			XFree(dri_drawable->back_cliprects);
+		drmHashDelete(dri_drawable->dri_screen->drawable_hash, dri_drawable->x_drawable);
+		XF86DRIDestroyDrawable(dri_drawable->dri_screen->display, dri_drawable->dri_screen->num, dri_drawable->x_drawable);
+		free(dri_drawable);
+	}
+
+	return 0;
+}
+
+int driCreateContext(dri_screen_t *dri_screen, Visual *visual, dri_context_t **dri_context)
+{
+	int		evbase, errbase;
+	dri_context_t	*dri_ctx;
+
+	assert(dri_screen);
+	assert(visual);
+	assert(dri_context);
+
+	if (!XF86DRIQueryExtension(dri_screen->display, &evbase, &errbase))
+		return 1;
+
+	dri_ctx = calloc(1, sizeof(dri_context_t));
+
+	if (!dri_ctx)
+		return 1;
+
+	if (!XF86DRICreateContext(dri_screen->display, dri_screen->num, visual, &dri_ctx->id, &dri_ctx->drm_context))
+	{
+		free(dri_ctx);
+		return 1;
+	}
+
+	dri_ctx->dri_screen = dri_screen;
+	*dri_context = dri_ctx;
+
+	return 0;
+}
+
+int driDestroyContext(dri_context_t *dri_context)
+{
+	assert(dri_context);
+
+	XF86DRIDestroyContext(dri_context->dri_screen->display, dri_context->dri_screen->num, dri_context->id);
+	free(dri_context);
+
+	return 0;
+}
diff --git a/src/driclient/src/xf86dristr.h b/src/driclient/src/xf86dristr.h
new file mode 100644
index 00000000000..b834bd1a1a0
--- /dev/null
+++ b/src/driclient/src/xf86dristr.h
@@ -0,0 +1,342 @@
+/**************************************************************************
+
+Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+Copyright 2000 VA Linux Systems, Inc.
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sub license, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
+ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/*
+ * Authors:
+ *   Kevin E. Martin <martin@valinux.com>
+ *   Jens Owen <jens@tungstengraphics.com>
+ *   Rickard E. (Rik) Fiath <faith@valinux.com>
+ *
+ */
+
+#ifndef _XF86DRISTR_H_
+#define _XF86DRISTR_H_
+
+#include "xf86dri.h"
+
+#define XF86DRINAME "XFree86-DRI"
+
+/* The DRI version number.  This was originally set to be the same of the
+ * XFree86 version number.  However, this version is really indepedent of
+ * the XFree86 version.
+ *
+ * Version History:
+ *    4.0.0: Original
+ *    4.0.1: Patch to bump clipstamp when windows are destroyed, 28 May 02
+ *    4.1.0: Add transition from single to multi in DRMInfo rec, 24 Jun 02
+ */
+#define XF86DRI_MAJOR_VERSION	4
+#define XF86DRI_MINOR_VERSION	1
+#define XF86DRI_PATCH_VERSION	0
+
+typedef struct _XF86DRIQueryVersion {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIQueryVersion */
+    CARD16	length B16;
+} xXF86DRIQueryVersionReq;
+#define sz_xXF86DRIQueryVersionReq	4
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD16	majorVersion B16;	/* major version of DRI protocol */
+    CARD16	minorVersion B16;	/* minor version of DRI protocol */
+    CARD32	patchVersion B32;       /* patch version of DRI protocol */
+    CARD32	pad3 B32;
+    CARD32	pad4 B32;
+    CARD32	pad5 B32;
+    CARD32	pad6 B32;
+} xXF86DRIQueryVersionReply;
+#define sz_xXF86DRIQueryVersionReply	32
+
+typedef struct _XF86DRIQueryDirectRenderingCapable {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* X_DRIQueryDirectRenderingCapable */
+    CARD16	length B16;
+    CARD32	screen B32;
+} xXF86DRIQueryDirectRenderingCapableReq;
+#define sz_xXF86DRIQueryDirectRenderingCapableReq	8
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    BOOL	isCapable;
+    BOOL	pad2;
+    BOOL	pad3;
+    BOOL	pad4;
+    CARD32	pad5 B32;
+    CARD32	pad6 B32;
+    CARD32	pad7 B32;
+    CARD32	pad8 B32;
+    CARD32	pad9 B32;
+} xXF86DRIQueryDirectRenderingCapableReply;
+#define sz_xXF86DRIQueryDirectRenderingCapableReply	32
+
+typedef struct _XF86DRIOpenConnection {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIOpenConnection */
+    CARD16	length B16;
+    CARD32	screen B32;
+} xXF86DRIOpenConnectionReq;
+#define sz_xXF86DRIOpenConnectionReq	8
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	hSAREALow B32;
+    CARD32	hSAREAHigh B32;
+    CARD32	busIdStringLength B32;
+    CARD32	pad6 B32;
+    CARD32	pad7 B32;
+    CARD32	pad8 B32;
+} xXF86DRIOpenConnectionReply;
+#define sz_xXF86DRIOpenConnectionReply	32
+
+typedef struct _XF86DRIAuthConnection {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRICloseConnection */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32      magic B32;
+} xXF86DRIAuthConnectionReq;
+#define sz_xXF86DRIAuthConnectionReq	12
+
+typedef struct {
+    BYTE        type;
+    BOOL        pad1;
+    CARD16      sequenceNumber B16;
+    CARD32      length B32;
+    CARD32      authenticated B32;
+    CARD32      pad2 B32;
+    CARD32      pad3 B32;
+    CARD32      pad4 B32;
+    CARD32      pad5 B32;
+    CARD32      pad6 B32;
+} xXF86DRIAuthConnectionReply;
+#define zx_xXF86DRIAuthConnectionReply  32
+
+typedef struct _XF86DRICloseConnection {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRICloseConnection */
+    CARD16	length B16;
+    CARD32	screen B32;
+} xXF86DRICloseConnectionReq;
+#define sz_xXF86DRICloseConnectionReq	8
+
+typedef struct _XF86DRIGetClientDriverName {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIGetClientDriverName */
+    CARD16	length B16;
+    CARD32	screen B32;
+} xXF86DRIGetClientDriverNameReq;
+#define sz_xXF86DRIGetClientDriverNameReq	8
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	ddxDriverMajorVersion B32;
+    CARD32	ddxDriverMinorVersion B32;
+    CARD32	ddxDriverPatchVersion B32;
+    CARD32	clientDriverNameLength B32;
+    CARD32	pad5 B32;
+    CARD32	pad6 B32;
+} xXF86DRIGetClientDriverNameReply;
+#define sz_xXF86DRIGetClientDriverNameReply	32
+
+typedef struct _XF86DRICreateContext {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRICreateContext */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32	visual B32;
+    CARD32	context B32;
+} xXF86DRICreateContextReq;
+#define sz_xXF86DRICreateContextReq	16
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	hHWContext B32;
+    CARD32	pad2 B32;
+    CARD32	pad3 B32;
+    CARD32	pad4 B32;
+    CARD32	pad5 B32;
+    CARD32	pad6 B32;
+} xXF86DRICreateContextReply;
+#define sz_xXF86DRICreateContextReply	32
+
+typedef struct _XF86DRIDestroyContext {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIDestroyContext */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32	context B32;
+} xXF86DRIDestroyContextReq;
+#define sz_xXF86DRIDestroyContextReq	12
+
+typedef struct _XF86DRICreateDrawable {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRICreateDrawable */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32	drawable B32;
+} xXF86DRICreateDrawableReq;
+#define sz_xXF86DRICreateDrawableReq	12
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	hHWDrawable B32;
+    CARD32	pad2 B32;
+    CARD32	pad3 B32;
+    CARD32	pad4 B32;
+    CARD32	pad5 B32;
+    CARD32	pad6 B32;
+} xXF86DRICreateDrawableReply;
+#define sz_xXF86DRICreateDrawableReply	32
+
+typedef struct _XF86DRIDestroyDrawable {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIDestroyDrawable */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32	drawable B32;
+} xXF86DRIDestroyDrawableReq;
+#define sz_xXF86DRIDestroyDrawableReq	12
+
+typedef struct _XF86DRIGetDrawableInfo {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIGetDrawableInfo */
+    CARD16	length B16;
+    CARD32	screen B32;
+    CARD32	drawable B32;
+} xXF86DRIGetDrawableInfoReq;
+#define sz_xXF86DRIGetDrawableInfoReq	12
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	drawableTableIndex B32;
+    CARD32	drawableTableStamp B32;
+    INT16	drawableX B16;
+    INT16	drawableY B16;
+    INT16	drawableWidth B16;
+    INT16	drawableHeight B16;
+    CARD32	numClipRects B32;
+    INT16       backX B16;
+    INT16       backY B16;
+    CARD32      numBackClipRects B32;
+} xXF86DRIGetDrawableInfoReply;
+
+#define sz_xXF86DRIGetDrawableInfoReply	36
+
+
+typedef struct _XF86DRIGetDeviceInfo {
+    CARD8	reqType;		/* always DRIReqCode */
+    CARD8	driReqType;		/* always X_DRIGetDeviceInfo */
+    CARD16	length B16;
+    CARD32	screen B32;
+} xXF86DRIGetDeviceInfoReq;
+#define sz_xXF86DRIGetDeviceInfoReq	8
+
+typedef struct {
+    BYTE	type;			/* X_Reply */
+    BOOL	pad1;
+    CARD16	sequenceNumber B16;
+    CARD32	length B32;
+    CARD32	hFrameBufferLow B32;
+    CARD32	hFrameBufferHigh B32;
+    CARD32	framebufferOrigin B32;
+    CARD32	framebufferSize B32;
+    CARD32	framebufferStride B32;
+    CARD32	devPrivateSize B32;
+} xXF86DRIGetDeviceInfoReply;
+#define sz_xXF86DRIGetDeviceInfoReply	32
+
+typedef struct _XF86DRIOpenFullScreen {
+    CARD8       reqType;	/* always DRIReqCode */
+    CARD8       driReqType;	/* always X_DRIOpenFullScreen */
+    CARD16      length B16;
+    CARD32      screen B32;
+    CARD32      drawable B32;
+} xXF86DRIOpenFullScreenReq;
+#define sz_xXF86DRIOpenFullScreenReq    12
+
+typedef struct {
+    BYTE        type;
+    BOOL        pad1;
+    CARD16      sequenceNumber B16;
+    CARD32      length B32;
+    CARD32      isFullScreen B32;
+    CARD32      pad2 B32;
+    CARD32      pad3 B32;
+    CARD32      pad4 B32;
+    CARD32      pad5 B32;
+    CARD32      pad6 B32;
+} xXF86DRIOpenFullScreenReply;
+#define sz_xXF86DRIOpenFullScreenReply  32
+
+typedef struct _XF86DRICloseFullScreen {
+    CARD8       reqType;	/* always DRIReqCode */
+    CARD8       driReqType;	/* always X_DRICloseFullScreen */
+    CARD16      length B16;
+    CARD32      screen B32;
+    CARD32      drawable B32;
+} xXF86DRICloseFullScreenReq;
+#define sz_xXF86DRICloseFullScreenReq   12
+
+typedef struct {
+    BYTE        type;
+    BOOL        pad1;
+    CARD16      sequenceNumber B16;
+    CARD32      length B32;
+    CARD32      pad2 B32;
+    CARD32      pad3 B32;
+    CARD32      pad4 B32;
+    CARD32      pad5 B32;
+    CARD32      pad6 B32;
+    CARD32      pad7 B32;
+} xXF86DRICloseFullScreenReply;
+#define sz_xXF86DRICloseFullScreenReply  32
+
+
+#endif /* _XF86DRISTR_H_ */
diff --git a/src/egl/Makefile b/src/egl/Makefile
index 082d7621838..024453f8bb1 100644
--- a/src/egl/Makefile
+++ b/src/egl/Makefile
@@ -11,14 +11,14 @@ default: subdirs
 subdirs:
 	@for dir in $(SUBDIRS) ; do \
 		if [ -d $$dir ] ; then \
-			(cd $$dir ; $(MAKE)) || exit 1 ; \
+			(cd $$dir && $(MAKE)) || exit 1 ; \
 		fi \
 	done
 
 
 clean:
-	@for dir in $(SUBDIRS) ; do \
+	-@for dir in $(SUBDIRS) ; do \
 		if [ -d $$dir ] ; then \
-			(cd $$dir ; $(MAKE) clean) ; \
+			(cd $$dir && $(MAKE) clean) ; \
 		fi \
 	done
diff --git a/src/egl/drivers/demo/Makefile b/src/egl/drivers/demo/Makefile
index e5a7be24381..6ca25aa5d65 100644
--- a/src/egl/drivers/demo/Makefile
+++ b/src/egl/drivers/demo/Makefile
@@ -21,10 +21,12 @@ default: $(TOP)/$(LIB_DIR)/demodriver.so
 
 
 $(TOP)/$(LIB_DIR)/demodriver.so: $(OBJECTS)
-	$(TOP)/bin/mklib -o demodriver.so -noprefix $(OBJECTS)
+	$(MKLIB) -o demodriver.so -noprefix -linker '$(CC)' \
+		-ldflags '$(LDFLAGS)' -install $(TOP)/$(LIB_DIR) \
+		$(OBJECTS)
 
 
 
 clean:
-	rm -f *.o
-	rm -f *.so
+	-rm -f *.o
+	-rm -f *.so
diff --git a/src/egl/drivers/dri/Makefile b/src/egl/drivers/dri/Makefile
index a41aca43709..d7eba653b2d 100644
--- a/src/egl/drivers/dri/Makefile
+++ b/src/egl/drivers/dri/Makefile
@@ -44,13 +44,14 @@ default: depend library Makefile
 library: $(TOP)/$(LIB_DIR)/libEGLdri.so
 
 $(TOP)/$(LIB_DIR)/libEGLdri.so: $(OBJECTS)
-	$(TOP)/bin/mklib -o EGLdri -major 1 -minor 0 \
-		-install $(TOP)/$(LIB_DIR) -ldl $(OBJECTS) $(DRM_LIB)
+	$(MKLIB) -o EGLdri -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+		-major 1 -minor 0 \
+		-install $(TOP)/$(LIB_DIR) -ldl $(OBJECTS) $(LIBS)
 
 
 clean:
-	rm -f *.o
-	rm -f *.so
+	-rm -f *.o
+	-rm -f *.so
 
 depend: $(SOURCES) $(HEADERS)
 	@ echo "running $(MKDEP)"
diff --git a/src/egl/drivers/glx/Makefile b/src/egl/drivers/glx/Makefile
new file mode 100644
index 00000000000..3cd52765133
--- /dev/null
+++ b/src/egl/drivers/glx/Makefile
@@ -0,0 +1,74 @@
+# src/egl/drivers/glx/Makefile
+
+# Build XEGL DRI driver loader library:  egl_glx.so
+
+
+TOP = ../../../..
+include $(TOP)/configs/current
+
+
+EXTRA_DEFINES = -DDEFAULT_DRIVER_DIR=\"$(DRI_DRIVER_SEARCH_DIR)\"
+
+DRIVER_NAME = egl_glx.so
+
+
+INCLUDE_DIRS = \
+	-I. \
+	-I/usr/include \
+	$(shell pkg-config --cflags-only-I libdrm) \
+	-I$(TOP)/include \
+	-I$(TOP)/include/GL/internal \
+	-I$(TOP)/src/mesa/glapi \
+	-I$(TOP)/src/mesa/drivers/dri/common \
+	-I$(TOP)/src/mesa/main \
+	-I$(TOP)/src/mesa \
+	-I$(TOP)/src/egl/main \
+	-I$(TOP)/src/glx/x11
+
+SOURCES = egl_glx.c
+
+OBJECTS = $(SOURCES:.c=.o)
+
+DRM_LIB = `pkg-config --libs libdrm`
+
+MISC_LIBS = -ldl -lXext -lGL
+
+
+.c.o:
+	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(EXTRA_DEFINES) $< -o $@
+
+
+.PHONY: library
+
+
+default: depend library Makefile
+
+
+library: $(TOP)/$(LIB_DIR)/$(DRIVER_NAME)
+
+
+# Make the egl_glx.so library
+$(TOP)/$(LIB_DIR)/$(DRIVER_NAME): $(OBJECTS)
+	$(TOP)/bin/mklib -o $(DRIVER_NAME) \
+		-noprefix \
+		-major 1 -minor 0 \
+		-L$(TOP)/$(LIB_DIR) \
+		-install $(TOP)/$(LIB_DIR) \
+		$(OBJECTS) $(DRM_LIB) $(MISC_LIBS)
+
+
+clean:
+	rm -f *.o
+	rm -f *.so
+	rm -f depend depend.bak
+
+
+depend: $(SOURCES) $(HEADERS)
+	@ echo "running $(MKDEP)"
+	@ rm -f depend
+	@ touch depend
+	$(MKDEP) $(MKDEP_OPTIONS) $(DEFINES) $(INCLUDE_DIRS) \
+		$(SOURCES) $(HEADERS) >/dev/null 2>/dev/null
+
+include depend
+# DO NOT DELETE
diff --git a/src/egl/drivers/glx/egl_glx.c b/src/egl/drivers/glx/egl_glx.c
new file mode 100644
index 00000000000..9a421b5a80b
--- /dev/null
+++ b/src/egl/drivers/glx/egl_glx.c
@@ -0,0 +1,584 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+/**
+ * This is an EGL driver that wraps GLX. This gives the benefit of being
+ * completely agnostic of the direct rendering implementation.
+ *
+ * Authors: Alan Hourihane <alanh@tungstengraphics.com>
+ */
+
+/*
+ * TODO: 
+ *
+ * Add GLXFBConfig support
+ * Pbuffer & Pixmap support
+ * test eglBind/ReleaseTexImage
+ */
+
+
+#include <assert.h>
+#include <stdio.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include "dlfcn.h"
+#include <X11/Xlib.h>
+#include <GL/gl.h>
+#include "glxclient.h"
+
+#define _EGL_PLATFORM_X
+
+#include "eglconfig.h"
+#include "eglcontext.h"
+#include "egldisplay.h"
+#include "egldriver.h"
+#include "eglglobals.h"
+#include "eglhash.h"
+#include "egllog.h"
+#include "eglsurface.h"
+
+#include <GL/gl.h>
+
+#define CALLOC_STRUCT(T)   (struct T *) calloc(1, sizeof(struct T))
+
+static const EGLint all_apis = (EGL_OPENGL_ES_BIT
+                                | EGL_OPENGL_ES2_BIT
+                                | EGL_OPENVG_BIT
+                                /* | EGL_OPENGL_BIT */); /* can't do */
+
+struct visual_attribs
+{
+   /* X visual attribs */
+   int id;
+   int klass;
+   int depth;
+   int redMask, greenMask, blueMask;
+   int colormapSize;
+   int bitsPerRGB;
+
+   /* GL visual attribs */
+   int supportsGL;
+   int transparentType;
+   int transparentRedValue;
+   int transparentGreenValue;
+   int transparentBlueValue;
+   int transparentAlphaValue;
+   int transparentIndexValue;
+   int bufferSize;
+   int level;
+   int render_type;
+   int doubleBuffer;
+   int stereo;
+   int auxBuffers;
+   int redSize, greenSize, blueSize, alphaSize;
+   int depthSize;
+   int stencilSize;
+   int accumRedSize, accumGreenSize, accumBlueSize, accumAlphaSize;
+   int numSamples, numMultisample;
+   int visualCaveat;
+};
+
+/** subclass of _EGLDriver */
+struct GLX_egl_driver
+{
+   _EGLDriver Base;   /**< base class */
+
+   XVisualInfo *visuals;
+
+   /* GLXFBConfig *fbconfigs - todo */
+};
+
+
+/** subclass of _EGLContext */
+struct GLX_egl_context
+{
+   _EGLContext Base;   /**< base class */
+
+   GLXContext context;
+};
+
+
+/** subclass of _EGLSurface */
+struct GLX_egl_surface
+{
+   _EGLSurface Base;   /**< base class */
+
+   GLXDrawable drawable;
+};
+
+
+/** subclass of _EGLConfig */
+struct GLX_egl_config
+{
+   _EGLConfig Base;   /**< base class */
+};
+
+/** cast wrapper */
+static struct GLX_egl_driver *
+GLX_egl_driver(_EGLDriver *drv)
+{
+   return (struct GLX_egl_driver *) drv;
+}
+
+static struct GLX_egl_context *
+GLX_egl_context(_EGLContext *ctx)
+{
+   return (struct GLX_egl_context *) ctx;
+}
+
+static struct GLX_egl_surface *
+GLX_egl_surface(_EGLSurface *surf)
+{
+   return (struct GLX_egl_surface *) surf;
+}
+
+static GLboolean
+get_visual_attribs(Display *dpy, XVisualInfo *vInfo,
+                   struct visual_attribs *attribs)
+{
+   const char *ext = glXQueryExtensionsString(dpy, vInfo->screen);
+   int rgba;
+
+   memset(attribs, 0, sizeof(struct visual_attribs));
+
+   attribs->id = vInfo->visualid;
+#if defined(__cplusplus) || defined(c_plusplus)
+   attribs->klass = vInfo->c_class;
+#else
+   attribs->klass = vInfo->class;
+#endif
+   attribs->depth = vInfo->depth;
+   attribs->redMask = vInfo->red_mask;
+   attribs->greenMask = vInfo->green_mask;
+   attribs->blueMask = vInfo->blue_mask;
+   attribs->colormapSize = vInfo->colormap_size;
+   attribs->bitsPerRGB = vInfo->bits_per_rgb;
+
+   if (glXGetConfig(dpy, vInfo, GLX_USE_GL, &attribs->supportsGL) != 0 ||
+       !attribs->supportsGL)
+      return GL_FALSE;
+   glXGetConfig(dpy, vInfo, GLX_BUFFER_SIZE, &attribs->bufferSize);
+   glXGetConfig(dpy, vInfo, GLX_LEVEL, &attribs->level);
+   glXGetConfig(dpy, vInfo, GLX_RGBA, &rgba);
+   if (!rgba)
+      return GL_FALSE;
+   attribs->render_type = GLX_RGBA_BIT;
+   
+   glXGetConfig(dpy, vInfo, GLX_DOUBLEBUFFER, &attribs->doubleBuffer);
+   if (!attribs->doubleBuffer)
+      return GL_FALSE;
+
+   glXGetConfig(dpy, vInfo, GLX_STEREO, &attribs->stereo);
+   glXGetConfig(dpy, vInfo, GLX_AUX_BUFFERS, &attribs->auxBuffers);
+   glXGetConfig(dpy, vInfo, GLX_RED_SIZE, &attribs->redSize);
+   glXGetConfig(dpy, vInfo, GLX_GREEN_SIZE, &attribs->greenSize);
+   glXGetConfig(dpy, vInfo, GLX_BLUE_SIZE, &attribs->blueSize);
+   glXGetConfig(dpy, vInfo, GLX_ALPHA_SIZE, &attribs->alphaSize);
+   glXGetConfig(dpy, vInfo, GLX_DEPTH_SIZE, &attribs->depthSize);
+   glXGetConfig(dpy, vInfo, GLX_STENCIL_SIZE, &attribs->stencilSize);
+   glXGetConfig(dpy, vInfo, GLX_ACCUM_RED_SIZE, &attribs->accumRedSize);
+   glXGetConfig(dpy, vInfo, GLX_ACCUM_GREEN_SIZE, &attribs->accumGreenSize);
+   glXGetConfig(dpy, vInfo, GLX_ACCUM_BLUE_SIZE, &attribs->accumBlueSize);
+   glXGetConfig(dpy, vInfo, GLX_ACCUM_ALPHA_SIZE, &attribs->accumAlphaSize);
+
+   /* get transparent pixel stuff */
+   glXGetConfig(dpy, vInfo,GLX_TRANSPARENT_TYPE, &attribs->transparentType);
+   if (attribs->transparentType == GLX_TRANSPARENT_RGB) {
+     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_RED_VALUE, &attribs->transparentRedValue);
+     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_GREEN_VALUE, &attribs->transparentGreenValue);
+     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_BLUE_VALUE, &attribs->transparentBlueValue);
+     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_ALPHA_VALUE, &attribs->transparentAlphaValue);
+   }
+   else if (attribs->transparentType == GLX_TRANSPARENT_INDEX) {
+     glXGetConfig(dpy, vInfo, GLX_TRANSPARENT_INDEX_VALUE, &attribs->transparentIndexValue);
+   }
+
+   /* multisample attribs */
+#ifdef GLX_ARB_multisample
+   if (ext && strstr(ext, "GLX_ARB_multisample")) {
+      glXGetConfig(dpy, vInfo, GLX_SAMPLE_BUFFERS_ARB, &attribs->numMultisample);
+      glXGetConfig(dpy, vInfo, GLX_SAMPLES_ARB, &attribs->numSamples);
+   }
+#endif
+   else {
+      attribs->numSamples = 0;
+      attribs->numMultisample = 0;
+   }
+
+#if defined(GLX_EXT_visual_rating)
+   if (ext && strstr(ext, "GLX_EXT_visual_rating")) {
+      glXGetConfig(dpy, vInfo, GLX_VISUAL_CAVEAT_EXT, &attribs->visualCaveat);
+   }
+   else {
+      attribs->visualCaveat = GLX_NONE_EXT;
+   }
+#else
+   attribs->visualCaveat = 0;
+#endif
+
+   return GL_TRUE;
+}
+
+static EGLBoolean
+create_configs(_EGLDisplay *disp, struct GLX_egl_driver *GLX_drv)
+{
+   XVisualInfo theTemplate;
+   int numVisuals;
+   long mask;
+   int i;
+   struct visual_attribs attribs;
+
+   /* get list of all visuals on this screen */
+   theTemplate.screen = DefaultScreen(disp->Xdpy);
+   mask = VisualScreenMask;
+   GLX_drv->visuals = XGetVisualInfo(disp->Xdpy, mask, &theTemplate, &numVisuals);
+
+   for (i = 0; i < numVisuals; i++) {
+      struct GLX_egl_config *config;
+
+      if (!get_visual_attribs(disp->Xdpy, &GLX_drv->visuals[i], &attribs))
+	 continue;
+
+      config = CALLOC_STRUCT(GLX_egl_config);
+
+      _eglInitConfig(&config->Base, i+1);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_ID, attribs.id);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_BUFFER_SIZE, attribs.bufferSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_RED_SIZE, attribs.redSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_GREEN_SIZE, attribs.greenSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_BLUE_SIZE, attribs.blueSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_ALPHA_SIZE, attribs.alphaSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_DEPTH_SIZE, attribs.depthSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_STENCIL_SIZE, attribs.stencilSize);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLES, attribs.numSamples);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLE_BUFFERS, attribs.numMultisample);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_CONFORMANT, all_apis);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_RENDERABLE_TYPE, all_apis);
+      SET_CONFIG_ATTRIB(&config->Base, EGL_SURFACE_TYPE,
+                        (EGL_WINDOW_BIT /*| EGL_PBUFFER_BIT | EGL_PIXMAP_BIT*/));
+
+      /* XXX possibly other things to init... */
+
+      _eglAddConfig(disp, &config->Base);
+   }
+
+   return EGL_TRUE;
+}
+
+/**
+ * Called via eglInitialize(), GLX_drv->API.Initialize().
+ */
+static EGLBoolean
+GLX_eglInitialize(_EGLDriver *drv, EGLDisplay dpy,
+                   EGLint *minor, EGLint *major)
+{
+   struct GLX_egl_driver *GLX_drv = GLX_egl_driver(drv);
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+
+   _eglLog(_EGL_DEBUG, "XDRI: eglInitialize");
+
+   if (!disp->Xdpy) {
+      disp->Xdpy = XOpenDisplay(NULL);
+      if (!disp->Xdpy) {
+         _eglLog(_EGL_WARNING, "XDRI: XOpenDisplay failed");
+         return EGL_FALSE;
+      }
+   }
+
+   GLX_drv->Base.Initialized = EGL_TRUE;
+
+   GLX_drv->Base.Name = "GLX";
+
+   /* we're supporting EGL 1.4 */
+   *minor = 1;
+   *major = 4;
+
+   create_configs(disp, GLX_drv);
+
+   return EGL_TRUE;
+}
+
+
+/**
+ * Called via eglTerminate(), drv->API.Terminate().
+ */
+static EGLBoolean
+GLX_eglTerminate(_EGLDriver *drv, EGLDisplay dpy)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+
+   _eglLog(_EGL_DEBUG, "XDRI: eglTerminate");
+
+//   XCloseDisplay(disp->Xdpy);
+
+   return EGL_TRUE;
+}
+
+
+/**
+ * Called via eglCreateContext(), drv->API.CreateContext().
+ */
+static EGLContext
+GLX_eglCreateContext(_EGLDriver *drv, EGLDisplay dpy, EGLConfig config,
+                      EGLContext share_list, const EGLint *attrib_list)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   struct GLX_egl_context *GLX_ctx = CALLOC_STRUCT(GLX_egl_context);
+   struct GLX_egl_driver *GLX_drv = GLX_egl_driver(drv);
+   struct GLX_egl_context *GLX_ctx_shared = NULL;
+   _EGLConfig *conf;
+
+   if (!GLX_ctx)
+      return EGL_NO_CONTEXT;
+
+   if (!_eglInitContext(drv, dpy, &GLX_ctx->Base, config, attrib_list)) {
+      free(GLX_ctx);
+      return EGL_NO_CONTEXT;
+   }
+
+   if (share_list != EGL_NO_CONTEXT) {
+      _EGLContext *shareCtx = _eglLookupContext(share_list);
+      if (!shareCtx) {
+         _eglError(EGL_BAD_CONTEXT, "eglCreateContext(share_list)");
+         return EGL_FALSE;
+      }
+      GLX_ctx_shared = GLX_egl_context(shareCtx);
+   }
+
+   conf = _eglLookupConfig(drv, dpy, config);
+   assert(conf);
+
+   GLX_ctx->context = glXCreateContext(disp->Xdpy, &GLX_drv->visuals[(int)config-1], GLX_ctx_shared ? GLX_ctx_shared->context : NULL, GL_TRUE);
+   if (!GLX_ctx->context)
+      return EGL_FALSE;
+
+   /* need to have a direct rendering context */
+   if (!glXIsDirect(disp->Xdpy, GLX_ctx->context))
+      return EGL_FALSE;
+
+   return _eglGetContextHandle(&GLX_ctx->Base);
+}
+
+
+/**
+ * Called via eglMakeCurrent(), drv->API.MakeCurrent().
+ */
+static EGLBoolean
+GLX_eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface d,
+                    EGLSurface r, EGLContext context)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   _EGLContext *ctx = _eglLookupContext(context);
+   _EGLSurface *dsurf = _eglLookupSurface(d);
+   _EGLSurface *rsurf = _eglLookupSurface(r);
+   struct GLX_egl_surface *GLX_dsurf = GLX_egl_surface(dsurf);
+   struct GLX_egl_surface *GLX_rsurf = GLX_egl_surface(rsurf);
+   struct GLX_egl_context *GLX_ctx = GLX_egl_context(ctx);
+
+   if (!_eglMakeCurrent(drv, dpy, d, r, context))
+      return EGL_FALSE;
+
+//   if (!glXMakeContextCurrent(disp->Xdpy, GLX_dsurf->drawable, GLX_rsurf->drawable, GLX_ctx->context))
+   if (!glXMakeCurrent(disp->Xdpy, GLX_dsurf ? GLX_dsurf->drawable : 0, GLX_ctx ? GLX_ctx->context : NULL))
+      return EGL_FALSE;
+
+   return EGL_TRUE;
+}
+
+/** Get size of given window */
+static Status
+get_drawable_size(Display *dpy, Drawable d, uint *width, uint *height)
+{
+   Window root;
+   Status stat;
+   int xpos, ypos;
+   unsigned int w, h, bw, depth;
+   stat = XGetGeometry(dpy, d, &root, &xpos, &ypos, &w, &h, &bw, &depth);
+   *width = w;
+   *height = h;
+   return stat;
+}
+
+/**
+ * Called via eglCreateWindowSurface(), drv->API.CreateWindowSurface().
+ */
+static EGLSurface
+GLX_eglCreateWindowSurface(_EGLDriver *drv, EGLDisplay dpy, EGLConfig config,
+                            NativeWindowType window, const EGLint *attrib_list)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   struct GLX_egl_surface *GLX_surf;
+   uint width, height;
+
+   GLX_surf = CALLOC_STRUCT(GLX_egl_surface);
+   if (!GLX_surf)
+      return EGL_NO_SURFACE;
+
+   if (!_eglInitSurface(drv, dpy, &GLX_surf->Base, EGL_WINDOW_BIT,
+                        config, attrib_list)) {
+      free(GLX_surf);
+      return EGL_FALSE;
+   }
+
+   _eglSaveSurface(&GLX_surf->Base);
+
+   GLX_surf->drawable = window;
+   get_drawable_size(disp->Xdpy, window, &width, &height);
+   GLX_surf->Base.Width = width;
+   GLX_surf->Base.Height = height;
+
+   return _eglGetSurfaceHandle(&GLX_surf->Base);
+}
+
+static EGLBoolean
+GLX_eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   _EGLSurface *surf = _eglLookupSurface(surface);
+   return EGL_TRUE;
+   if (surf) {
+      _eglHashRemove(_eglGlobal.Surfaces, (EGLuint) surface);
+      if (surf->IsBound) {
+         surf->DeletePending = EGL_TRUE;
+      }
+      else {
+         free(surf);
+      }
+
+      return EGL_TRUE;
+   }
+   else {
+      _eglError(EGL_BAD_SURFACE, "eglDestroySurface");
+      return EGL_FALSE;
+   }
+}
+
+
+static EGLBoolean
+GLX_eglBindTexImage(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface,
+                     EGLint buffer)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   _EGLSurface *surf = _eglLookupSurface(surface);
+   struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
+
+   /* buffer ?? */
+   glXBindTexImageEXT(disp->Xdpy, GLX_surf->drawable, GLX_FRONT_LEFT_EXT, NULL);
+
+   return EGL_TRUE;
+}
+
+
+static EGLBoolean
+GLX_eglReleaseTexImage(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface,
+                        EGLint buffer)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   _EGLSurface *surf = _eglLookupSurface(surface);
+   struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
+
+   /* buffer ?? */
+   glXReleaseTexImageEXT(disp->Xdpy, GLX_surf->drawable, GLX_FRONT_LEFT_EXT);
+
+   return EGL_TRUE;
+}
+
+
+static EGLBoolean
+GLX_eglSwapBuffers(_EGLDriver *drv, EGLDisplay dpy, EGLSurface draw)
+{
+   _EGLDisplay *disp = _eglLookupDisplay(dpy);
+   _EGLSurface *surf = _eglLookupSurface(draw);
+   struct GLX_egl_surface *GLX_surf = GLX_egl_surface(surf);
+
+   _eglLog(_EGL_DEBUG, "XDRI: EGL SwapBuffers 0x%x",draw);
+
+   /* error checking step: */
+   if (!_eglSwapBuffers(drv, dpy, draw))
+      return EGL_FALSE;
+
+   glXSwapBuffers(disp->Xdpy, GLX_surf->drawable);
+
+   return EGL_TRUE;
+}
+
+/*
+ * Called from eglGetProcAddress() via drv->API.GetProcAddress().
+ */
+static _EGLProc
+GLX_eglGetProcAddress(const char *procname)
+{
+   return (_EGLProc)glXGetProcAddress((const GLubyte *)procname);   
+}
+
+
+/**
+ * This is the main entrypoint into the driver, called by libEGL.
+ * Create a new _EGLDriver object and init its dispatch table.
+ */
+_EGLDriver *
+_eglMain(_EGLDisplay *disp, const char *args)
+{
+   struct GLX_egl_driver *GLX_drv = CALLOC_STRUCT(GLX_egl_driver);
+   char *env;
+
+   if (!GLX_drv)
+      return NULL;
+
+   _eglInitDriverFallbacks(&GLX_drv->Base);
+   GLX_drv->Base.API.Initialize = GLX_eglInitialize;
+   GLX_drv->Base.API.Terminate = GLX_eglTerminate;
+   GLX_drv->Base.API.CreateContext = GLX_eglCreateContext;
+   GLX_drv->Base.API.MakeCurrent = GLX_eglMakeCurrent;
+   GLX_drv->Base.API.CreateWindowSurface = GLX_eglCreateWindowSurface;
+   GLX_drv->Base.API.DestroySurface = GLX_eglDestroySurface;
+   GLX_drv->Base.API.BindTexImage = GLX_eglBindTexImage;
+   GLX_drv->Base.API.ReleaseTexImage = GLX_eglReleaseTexImage;
+   GLX_drv->Base.API.SwapBuffers = GLX_eglSwapBuffers;
+   GLX_drv->Base.API.GetProcAddress = GLX_eglGetProcAddress;
+
+   GLX_drv->Base.ClientAPIsMask = all_apis;
+   GLX_drv->Base.Name = "GLX";
+
+   _eglLog(_EGL_DEBUG, "GLX: main(%s)", args);
+
+   /* set new DRI path to pick up EGL version (which doesn't contain any mesa 
+    * code), but don't override if one is already set.
+    */
+   env = getenv("LIBGL_DRIVERS_PATH");
+   if (env) {
+      if (!strstr(env, "egl")) {
+         sprintf(env, "%s/egl", env);
+         setenv("LIBGL_DRIVERS_PATH", env, 1);
+      }
+   } else
+      setenv("LIBGL_DRIVERS_PATH", DEFAULT_DRIVER_DIR"/egl", 0);
+
+   return &GLX_drv->Base;
+}
diff --git a/src/egl/drivers/xdri/egl_xdri.c b/src/egl/drivers/xdri/egl_xdri.c
index bb38513e7d0..3b3e312746e 100644
--- a/src/egl/drivers/xdri/egl_xdri.c
+++ b/src/egl/drivers/xdri/egl_xdri.c
@@ -962,7 +962,7 @@ xdri_eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
       }
       else {
          /*
-         st_unreference_framebuffer(&surf->Framebuffer);
+         st_unreference_framebuffer(surf->Framebuffer);
          */
          free(xdri_surf);
       }
diff --git a/src/egl/main/Makefile b/src/egl/main/Makefile
index 96742bb4bc6..8ea60c0199d 100644
--- a/src/egl/main/Makefile
+++ b/src/egl/main/Makefile
@@ -61,14 +61,15 @@ default: depend library
 library: $(TOP)/$(LIB_DIR)/libEGL.so
 
 $(TOP)/$(LIB_DIR)/libEGL.so: $(OBJECTS)
-	$(TOP)/bin/mklib -o EGL -major 1 -minor 0 \
+	$(MKLIB) -o EGL -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+		-major 1 -minor 0 \
 		-install $(TOP)/$(LIB_DIR) -ldl $(OBJECTS) $(LIBS)
 
 
 
 clean:
-	rm -f *.o *.so*
-	rm -f core.*
+	-rm -f *.o *.so*
+	-rm -f core.*
 
 
 depend: $(SOURCES) $(HEADERS)
diff --git a/src/egl/main/eglconfigutil.c b/src/egl/main/eglconfigutil.c
index db2ff1bfe8a..7061df691b2 100644
--- a/src/egl/main/eglconfigutil.c
+++ b/src/egl/main/eglconfigutil.c
@@ -123,7 +123,6 @@ _eglFillInConfigs(_EGLConfig * configs,
                   const GLenum * db_modes, unsigned num_db_modes,
                   int visType)
 {
-#if 0
    static const uint8_t bits_table[3][4] = {
             /* R  G  B  A */
             { 5, 6, 5, 0 },  /* Any GL_UNSIGNED_SHORT_5_6_5 */
@@ -214,6 +213,7 @@ _eglFillInConfigs(_EGLConfig * configs,
       masks = masks_table_rgba[index];
       break;
 
+#if 0
    case GL_BGR:
       bits = (bytes_per_pixel[index] == 2) ? bits_table[0] : bits_table[1];
       masks = masks_table_bgr[index];
@@ -223,6 +223,7 @@ _eglFillInConfigs(_EGLConfig * configs,
       bits = (bytes_per_pixel[index] == 2) ? bits_table[0] : bits_table[2];
       masks = masks_table_bgra[index];
       break;
+#endif
 
    default:
       _eglLog(_EGL_WARNING,
@@ -253,8 +254,5 @@ _eglFillInConfigs(_EGLConfig * configs,
       }
    }
    return GL_TRUE;
-#else
-   return GL_FALSE;
-#endif
 }
 
diff --git a/src/egl/main/eglx.c b/src/egl/main/eglx.c
index 1bba149f702..50acc3a24f9 100644
--- a/src/egl/main/eglx.c
+++ b/src/egl/main/eglx.c
@@ -44,7 +44,7 @@
 #include "eglx.h"
 
 
-static const char *DefaultDRIDriver = "egl_xdri";
+static const char *DefaultGLXDriver = "egl_glx";
 static const char *DefaultSoftDriver = "egl_softpipe";
 
 
@@ -52,7 +52,7 @@ static const char *DefaultSoftDriver = "egl_softpipe";
  * Given an X Display ptr (at dpy->Xdpy) try to determine the appropriate
  * device driver.  Return its name.
  *
- * This boils down to whether to use the egl_xdri.so driver which will
+ * This boils down to whether to use the egl_glx.so driver which will
  * load a DRI driver or the egl_softpipe.so driver that'll do software
  * rendering on Xlib.
  */
@@ -82,9 +82,8 @@ _xeglChooseDriver(_EGLDisplay *dpy)
    /* See if we can choose a DRI/DRM driver */
    driverName = _eglChooseDRMDriver(screen);
    if (driverName) {
-      /* DRI is available */
       free((void *) driverName);
-      driverName = _eglstrdup(DefaultDRIDriver);
+      driverName = _eglstrdup(DefaultGLXDriver);
    }
    else {
       driverName = _eglstrdup(DefaultSoftDriver);
diff --git a/src/gallium/Makefile b/src/gallium/Makefile
index aa77021daf3..36bd3623e7f 100644
--- a/src/gallium/Makefile
+++ b/src/gallium/Makefile
@@ -3,6 +3,7 @@ include $(TOP)/configs/current
 
 
 SUBDIRS = auxiliary drivers
+# Note winsys/ needs to be built after src/mesa
 
 
 default: subdirs
diff --git a/src/gallium/auxiliary/cso_cache/cso_hash.c b/src/gallium/auxiliary/cso_cache/cso_hash.c
index 7f0044c5a7f..4e7664f9bf0 100644
--- a/src/gallium/auxiliary/cso_cache/cso_hash.c
+++ b/src/gallium/auxiliary/cso_cache/cso_hash.c
@@ -431,3 +431,9 @@ struct cso_hash_iter cso_hash_erase(struct cso_hash *hash, struct cso_hash_iter
    --hash->data.d->size;
    return ret;
 }
+
+boolean cso_hash_contains(struct cso_hash *hash, unsigned key)
+{
+   struct cso_node **node = cso_hash_find_node(hash, key);
+   return (*node != hash->data.e);
+}
diff --git a/src/gallium/auxiliary/cso_cache/cso_hash.h b/src/gallium/auxiliary/cso_cache/cso_hash.h
index 85f3e276c6a..5891c325fa5 100644
--- a/src/gallium/auxiliary/cso_cache/cso_hash.h
+++ b/src/gallium/auxiliary/cso_cache/cso_hash.h
@@ -44,6 +44,7 @@
 #ifndef CSO_HASH_H
 #define CSO_HASH_H
 
+#include "pipe/p_compiler.h"
 
 #ifdef	__cplusplus
 extern "C" {
@@ -95,6 +96,11 @@ struct cso_hash_iter cso_hash_first_node(struct cso_hash *hash);
  */
 struct cso_hash_iter cso_hash_find(struct cso_hash *hash, unsigned key);
 
+/**
+ * Returns true if a value with the given key exists in the hash
+ */
+boolean   cso_hash_contains(struct cso_hash *hash, unsigned key);
+
 
 int       cso_hash_iter_is_null(struct cso_hash_iter iter);
 unsigned  cso_hash_iter_key(struct cso_hash_iter iter);
diff --git a/src/gallium/auxiliary/draw/Makefile b/src/gallium/auxiliary/draw/Makefile
index f2e36a89e90..bdbf5a08ede 100644
--- a/src/gallium/auxiliary/draw/Makefile
+++ b/src/gallium/auxiliary/draw/Makefile
@@ -40,6 +40,7 @@ C_SOURCES = \
 	draw_vs_aos_machine.c \
 	draw_vs_exec.c \
 	draw_vs_llvm.c \
+	draw_vs_ppc.c  \
 	draw_vs_sse.c 
 
 
diff --git a/src/gallium/auxiliary/draw/SConscript b/src/gallium/auxiliary/draw/SConscript
index 544a04918b6..5f05aa324a5 100644
--- a/src/gallium/auxiliary/draw/SConscript
+++ b/src/gallium/auxiliary/draw/SConscript
@@ -38,6 +38,7 @@ draw = env.ConvenienceLibrary(
 		'draw_vs_aos_machine.c',
 		'draw_vs_exec.c',
 		'draw_vs_llvm.c',
+		'draw_vs_ppc.c',
 		'draw_vs_sse.c',
 		'draw_vs_varient.c'
 	])
diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index fab8fc95fca..7bd4a2e2218 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -343,6 +343,21 @@ draw_num_vs_outputs(const struct draw_context *draw)
 }
 
 
+/**
+ * Provide TGSI sampler objects for vertex shaders that use texture fetches.
+ * This might only be used by software drivers for the time being.
+ */
+void
+draw_texture_samplers(struct draw_context *draw,
+                      uint num_samplers,
+                      struct tgsi_sampler **samplers)
+{
+   draw->vs.num_samplers = num_samplers;
+   draw->vs.samplers = samplers;
+}
+
+
+
 
 void draw_set_render( struct draw_context *draw, 
 		      struct vbuf_render *render )
diff --git a/src/gallium/auxiliary/draw/draw_context.h b/src/gallium/auxiliary/draw/draw_context.h
index a29bb01d814..d529e4e9a27 100644
--- a/src/gallium/auxiliary/draw/draw_context.h
+++ b/src/gallium/auxiliary/draw/draw_context.h
@@ -45,7 +45,7 @@ struct pipe_context;
 struct draw_context;
 struct draw_stage;
 struct draw_vertex_shader;
-
+struct tgsi_sampler;
 
 
 struct draw_context *draw_create( void );
@@ -92,6 +92,12 @@ uint
 draw_num_vs_outputs(const struct draw_context *draw);
 
 
+void
+draw_texture_samplers(struct draw_context *draw,
+                      uint num_samplers,
+                      struct tgsi_sampler **samplers);
+
+
 
 /*
  * Vertex shader functions
diff --git a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
index 9825e116c32..5ead25efff0 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
@@ -399,6 +399,8 @@ vbuf_alloc_vertices( struct vbuf_stage *vbuf )
     * and it will flush itself if necessary to do so.  If this does
     * fail, we are basically without usable hardware.
     */
+   assert(vbuf->max_vertices < UNDEFINED_VERTEX_ID);
+
    vbuf->vertices = (uint *) vbuf->render->allocate_vertices(vbuf->render,
 							     (ushort) vbuf->vertex_size,
 							     (ushort) vbuf->max_vertices);
diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h
index a16b45d340a..81e4eae401c 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gallium/auxiliary/draw/draw_private.h
@@ -187,6 +187,9 @@ struct draw_context
       /** TGSI program interpreter runtime state */
       struct tgsi_exec_machine machine;
 
+      uint num_samplers;
+      struct tgsi_sampler **samplers;
+
       /* This (and the tgsi_exec_machine struct) probably need to be moved somewhere private.
        */
       struct gallivm_cpu_engine *engine;   
@@ -198,7 +201,7 @@ struct draw_context
 
       const float (*aligned_constants)[4];
 
-      float (*aligned_constant_storage)[4];
+      const float (*aligned_constant_storage)[4];
       unsigned const_storage_size;
 
 
diff --git a/src/gallium/auxiliary/draw/draw_pt_emit.c b/src/gallium/auxiliary/draw/draw_pt_emit.c
index d520b05869b..232dfdaed2d 100644
--- a/src/gallium/auxiliary/draw/draw_pt_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_emit.c
@@ -165,6 +165,11 @@ void draw_pt_emit( struct pt_emit *emit,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (vertex_count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    /* XXX: and work out some way to coordinate the render primitive
     * between vbuf.c and here...
     */
@@ -226,6 +231,11 @@ void draw_pt_emit_linear(struct pt_emit *emit,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    /* XXX: and work out some way to coordinate the render primitive
     * between vbuf.c and here...
     */
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
index 3966ad48ba7..0227652632e 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
@@ -229,6 +229,11 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (fetch_count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)feme->translate->key.output_stride,
                                                (ushort)fetch_count );
@@ -283,6 +288,11 @@ static void fetch_emit_run_linear( struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)feme->translate->key.output_stride,
                                                (ushort)count );
@@ -338,6 +348,9 @@ static boolean fetch_emit_run_linear_elts( struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (count >= UNDEFINED_VERTEX_ID)
+      return FALSE;
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)feme->translate->key.output_stride,
                                                (ushort)count );
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
index f7e6a1a8eeb..1649cdc6cd1 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c
@@ -234,6 +234,11 @@ static void fse_run_linear( struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)fse->key.output_stride,
                                                (ushort)count );
@@ -293,6 +298,11 @@ fse_run(struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (fetch_count >= UNDEFINED_VERTEX_ID) {
+      assert(0);
+      return;
+   }
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)fse->key.output_stride,
                                                (ushort)fetch_count );
@@ -347,6 +357,9 @@ static boolean fse_run_linear_elts( struct draw_pt_middle_end *middle,
     */
    draw_do_flush( draw, DRAW_FLUSH_BACKEND );
 
+   if (count >= UNDEFINED_VERTEX_ID)
+      return FALSE;
+
    hw_verts = draw->render->allocate_vertices( draw->render,
                                                (ushort)fse->key.output_stride,
                                                (ushort)count );
diff --git a/src/gallium/auxiliary/draw/draw_pt_vcache.c b/src/gallium/auxiliary/draw/draw_pt_vcache.c
index 80d7200ca63..5d268a22264 100644
--- a/src/gallium/auxiliary/draw/draw_pt_vcache.c
+++ b/src/gallium/auxiliary/draw/draw_pt_vcache.c
@@ -324,7 +324,7 @@ vcache_check_run( struct draw_pt_front_end *frontend,
    unsigned fetch_count = max_index + 1 - min_index;
    const ushort *transformed_elts;
    ushort *storage = NULL;
-   boolean ok;
+   boolean ok = FALSE;
 
 
    if (0) debug_printf("fetch_count %d fetch_max %d draw_count %d\n", fetch_count, 
@@ -413,11 +413,12 @@ vcache_check_run( struct draw_pt_front_end *frontend,
       transformed_elts = storage;
    }
 
-   ok = vcache->middle->run_linear_elts( vcache->middle,
-                                         min_index, /* start */
-                                         fetch_count,
-                                         transformed_elts,
-                                         draw_count );
+   if (fetch_count < UNDEFINED_VERTEX_ID)
+      ok = vcache->middle->run_linear_elts( vcache->middle,
+                                            min_index, /* start */
+                                            fetch_count,
+                                            transformed_elts,
+                                            draw_count );
    
    FREE(storage);
 
diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c
index 34adbd49b00..7f305304ff7 100644
--- a/src/gallium/auxiliary/draw/draw_vs.c
+++ b/src/gallium/auxiliary/draw/draw_vs.c
@@ -85,7 +85,10 @@ draw_create_vertex_shader(struct draw_context *draw,
    if (!vs) {
       vs = draw_create_vs_sse( draw, shader );
       if (!vs) {
-         vs = draw_create_vs_exec( draw, shader );
+         vs = draw_create_vs_ppc( draw, shader );
+         if (!vs) {
+            vs = draw_create_vs_exec( draw, shader );
+         }
       }
    }
 
diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h
index 68c24abad3b..89ae158751a 100644
--- a/src/gallium/auxiliary/draw/draw_vs.h
+++ b/src/gallium/auxiliary/draw/draw_vs.h
@@ -158,6 +158,10 @@ draw_create_vs_sse(struct draw_context *draw,
 		   const struct pipe_shader_state *templ);
 
 struct draw_vertex_shader *
+draw_create_vs_ppc(struct draw_context *draw,
+		   const struct pipe_shader_state *templ);
+
+struct draw_vertex_shader *
 draw_create_vs_llvm(struct draw_context *draw,
 		    const struct pipe_shader_state *templ);
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c
index 82d27d44934..b3200df8112 100644
--- a/src/gallium/auxiliary/draw/draw_vs_exec.c
+++ b/src/gallium/auxiliary/draw/draw_vs_exec.c
@@ -62,12 +62,15 @@ vs_exec_prepare( struct draw_vertex_shader *shader,
 {
    struct exec_vertex_shader *evs = exec_vertex_shader(shader);
 
-   /* specify the vertex program to interpret/execute */
-   tgsi_exec_machine_bind_shader(evs->machine,
-				 shader->state.tokens,
-				 PIPE_MAX_SAMPLERS,
-				 NULL /*samplers*/ );
-
+   /* Specify the vertex program to interpret/execute.
+    * Avoid rebinding when possible.
+    */
+   if (evs->machine->Tokens != shader->state.tokens) {
+      tgsi_exec_machine_bind_shader(evs->machine,
+                                    shader->state.tokens,
+                                    draw->vs.num_samplers,
+                                    draw->vs.samplers);
+   }
 }
 
 
diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c
new file mode 100644
index 00000000000..8b751361449
--- /dev/null
+++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c
@@ -0,0 +1,239 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+ /*
+  * Authors:
+  *   Keith Whitwell <keith@tungstengraphics.com>
+  *   Brian Paul
+  */
+
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "pipe/p_config.h"
+
+#include "draw_vs.h"
+
+#if defined(PIPE_ARCH_PPC)
+
+#include "pipe/p_shader_tokens.h"
+
+#include "draw_private.h"
+#include "draw_context.h"
+
+#include "rtasm/rtasm_cpu.h"
+#include "rtasm/rtasm_ppc.h"
+#include "tgsi/tgsi_ppc.h"
+#include "tgsi/tgsi_parse.h"
+
+
+
+typedef void (PIPE_CDECL *codegen_function) (float (*inputs)[4][4],
+                                             float (*outputs)[4][4],
+                                             float (*temps)[4][4],
+                                             float (*immeds)[4],
+                                             float (*consts)[4],
+                                             const float *builtins);
+
+
+struct draw_ppc_vertex_shader {
+   struct draw_vertex_shader base;
+   struct ppc_function ppc_program;
+
+   codegen_function func;
+};
+
+
+static void
+vs_ppc_prepare( struct draw_vertex_shader *base,
+		struct draw_context *draw )
+{
+   /* nothing */
+}
+
+
+/**
+ * Simplified vertex shader interface for the pt paths.  Given the
+ * complexity of code-generating all the above operations together,
+ * it's time to try doing all the other stuff separately.
+ */
+static void
+vs_ppc_run_linear( struct draw_vertex_shader *base,
+		   const float (*input)[4],
+		   float (*output)[4],
+		   const float (*constants)[4],
+		   unsigned count,
+		   unsigned input_stride,
+		   unsigned output_stride )
+{
+   struct draw_ppc_vertex_shader *shader = (struct draw_ppc_vertex_shader *)base;
+   unsigned int i;
+
+#define MAX_VERTICES 4
+
+   /* loop over verts */
+   for (i = 0; i < count; i += MAX_VERTICES) {
+      const uint max_vertices = MIN2(MAX_VERTICES, count - i);
+      float inputs_soa[PIPE_MAX_SHADER_INPUTS][4][4] ALIGN16_ATTRIB;
+      float outputs_soa[PIPE_MAX_SHADER_OUTPUTS][4][4] ALIGN16_ATTRIB;
+      float temps_soa[TGSI_EXEC_NUM_TEMPS][4][4] ALIGN16_ATTRIB;
+      uint attr;
+
+      /* convert (up to) four input verts to SoA format */
+      for (attr = 0; attr < base->info.num_inputs; attr++) {
+         const float *vIn = (const float *) input;
+         uint vert;
+         for (vert = 0; vert < max_vertices; vert++) {
+#if 0
+            if (attr==0)
+               printf("Input v%d a%d: %f %f %f %f\n",
+                      vert, attr, vIn[0], vIn[1], vIn[2], vIn[3]);
+#endif
+            inputs_soa[attr][0][vert] = vIn[attr * 4 + 0];
+            inputs_soa[attr][1][vert] = vIn[attr * 4 + 1];
+            inputs_soa[attr][2][vert] = vIn[attr * 4 + 2];
+            inputs_soa[attr][3][vert] = vIn[attr * 4 + 3];
+            vIn += input_stride / 4;
+         }
+      }
+
+      /* run compiled shader
+       */
+      shader->func(inputs_soa, outputs_soa, temps_soa,
+		   (float (*)[4]) shader->base.immediates,
+		   (float (*)[4]) constants,
+                   ppc_builtin_constants);
+
+      /* convert (up to) four output verts from SoA back to AoS format */
+      for (attr = 0; attr < base->info.num_outputs; attr++) {
+         float *vOut = (float *) output;
+         uint vert;
+         for (vert = 0; vert < max_vertices; vert++) {
+            vOut[attr * 4 + 0] = outputs_soa[attr][0][vert];
+            vOut[attr * 4 + 1] = outputs_soa[attr][1][vert];
+            vOut[attr * 4 + 2] = outputs_soa[attr][2][vert];
+            vOut[attr * 4 + 3] = outputs_soa[attr][3][vert];
+#if 0
+            if (attr==0)
+               printf("Output v%d a%d: %f %f %f %f\n",
+                      vert, attr, vOut[0], vOut[1], vOut[2], vOut[3]);
+#endif
+            vOut += output_stride / 4;
+         }
+      }
+
+      /* advance to next group of four input/output verts */
+      input = (const float (*)[4])((const char *)input + input_stride * max_vertices);
+      output = (float (*)[4])((char *)output + output_stride * max_vertices);
+   }
+}
+
+
+static void
+vs_ppc_delete( struct draw_vertex_shader *base )
+{
+   struct draw_ppc_vertex_shader *shader = (struct draw_ppc_vertex_shader *)base;
+   
+   ppc_release_func( &shader->ppc_program );
+
+   align_free( (void *) shader->base.immediates );
+
+   FREE( (void*) shader->base.state.tokens );
+   FREE( shader );
+}
+
+
+struct draw_vertex_shader *
+draw_create_vs_ppc(struct draw_context *draw,
+                   const struct pipe_shader_state *templ)
+{
+   struct draw_ppc_vertex_shader *vs;
+
+   vs = CALLOC_STRUCT( draw_ppc_vertex_shader );
+   if (vs == NULL) 
+      return NULL;
+
+   /* we make a private copy of the tokens */
+   vs->base.state.tokens = tgsi_dup_tokens(templ->tokens);
+   if (!vs->base.state.tokens)
+      goto fail;
+
+   tgsi_scan_shader(templ->tokens, &vs->base.info);
+
+   vs->base.draw = draw;
+#if 0
+   if (1)
+      vs->base.create_varient = draw_vs_varient_aos_ppc;
+   else
+#endif
+      vs->base.create_varient = draw_vs_varient_generic;
+   vs->base.prepare = vs_ppc_prepare;
+   vs->base.run_linear = vs_ppc_run_linear;
+   vs->base.delete = vs_ppc_delete;
+   
+   vs->base.immediates = align_malloc(TGSI_EXEC_NUM_IMMEDIATES * 4 *
+                                      sizeof(float), 16);
+
+   ppc_init_func( &vs->ppc_program );
+
+   if (!tgsi_emit_ppc( (struct tgsi_token *) vs->base.state.tokens,
+			&vs->ppc_program, 
+                       (float (*)[4]) vs->base.immediates, 
+                        TRUE )) 
+      goto fail;
+      
+   vs->func = (codegen_function) ppc_get_func( &vs->ppc_program );
+   if (!vs->func) {
+      goto fail;
+   }
+   
+   return &vs->base;
+
+fail:
+   /*
+   debug_error("tgsi_emit_ppc() failed, falling back to interpreter\n");
+   */
+
+   ppc_release_func( &vs->ppc_program );
+   
+   FREE(vs);
+   return NULL;
+}
+
+
+
+#else /* PIPE_ARCH_PPC */
+
+
+struct draw_vertex_shader *
+draw_create_vs_ppc( struct draw_context *draw,
+		    const struct pipe_shader_state *templ )
+{
+   return (void *) 0;
+}
+
+
+#endif /* PIPE_ARCH_PPC */
diff --git a/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp b/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp
index 3a4a41e5445..93a9748bdb3 100644
--- a/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp
+++ b/src/gallium/auxiliary/gallivm/gallivm_cpu.cpp
@@ -158,8 +158,8 @@ void gallivm_cpu_jit_compile(struct gallivm_cpu_engine *cpu, struct gallivm_prog
    llvm::ExistingModuleProvider *mp = new llvm::ExistingModuleProvider(mod);
    llvm::ExecutionEngine *ee = cpu->engine;
    assert(ee);
-   /*FIXME : remove */
-   ee->DisableLazyCompilation();
+   /*FIXME : why was this disabled ? we need it for pow/sqrt/... */
+   ee->DisableLazyCompilation(false);
    ee->addModuleProvider(mp);
 
    llvm::Function *func = func_for_shader(prog);
@@ -179,8 +179,7 @@ struct gallivm_cpu_engine * gallivm_global_cpu_engine()
 
 typedef void (*vertex_shader_runner)(void *ainputs,
                                      void *dests,
-                                     float (*aconsts)[4],
-                                     void *temps);
+                                     float (*aconsts)[4]);
 
 #define MAX_TGSI_VERTICES 4
 /*!
@@ -202,7 +201,6 @@ int gallivm_cpu_vs_exec(struct gallivm_prog *prog,
    unsigned int i, j;
    unsigned slot;
    vertex_shader_runner runner = reinterpret_cast<vertex_shader_runner>(prog->function);
-
    assert(runner);
 
    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
@@ -224,8 +222,7 @@ int gallivm_cpu_vs_exec(struct gallivm_prog *prog,
       /* run shader */
       runner(machine->Inputs,
              machine->Outputs,
-             (float (*)[4]) constants,
-             machine->Temps);
+             (float (*)[4]) constants);
 
       /* Unswizzle all output results
        */
diff --git a/src/gallium/auxiliary/gallivm/gallivm_p.h b/src/gallium/auxiliary/gallivm/gallivm_p.h
index ebf3e11cd56..d2c5852bdf7 100644
--- a/src/gallium/auxiliary/gallivm/gallivm_p.h
+++ b/src/gallium/auxiliary/gallivm/gallivm_p.h
@@ -101,10 +101,10 @@ static INLINE int gallivm_w_swizzle(int swizzle)
    return w;
 }
 
-#endif /* MESA_LLVM */
-
 #if defined __cplusplus
 }
 #endif
 
+#endif /* MESA_LLVM */
+
 #endif
diff --git a/src/gallium/auxiliary/gallivm/instructionssoa.cpp b/src/gallium/auxiliary/gallivm/instructionssoa.cpp
index a6580725512..d5600fd22da 100644
--- a/src/gallium/auxiliary/gallivm/instructionssoa.cpp
+++ b/src/gallium/auxiliary/gallivm/instructionssoa.cpp
@@ -90,68 +90,11 @@ llvm::Value * InstructionsSoa::vectorFromVals(llvm::Value *x, llvm::Value *y,
    return res;
 }
 
-std::vector<llvm::Value*> InstructionsSoa::arl(const std::vector<llvm::Value*> in)
-{
-   std::vector<llvm::Value*> res(4);
-
-   //Extract x's
-   llvm::Value *x1 = m_builder.CreateExtractElement(in[0],
-                                                    m_storage->constantInt(0),
-                                                    name("extractX"));
-   //cast it to an unsigned int
-   x1 = m_builder.CreateFPToUI(x1, IntegerType::get(32), name("x1IntCast"));
-
-   res[0] = x1;//vectorFromVals(x1, x2, x3, x4);
-   //only x is valid. the others shouldn't be necessary
-   /*
-   res[1] = Constant::getNullValue(m_floatVecType);
-   res[2] = Constant::getNullValue(m_floatVecType);
-   res[3] = Constant::getNullValue(m_floatVecType);
-   */
-
-   return res;
-}
-
-
-std::vector<llvm::Value*> InstructionsSoa::add(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   std::vector<llvm::Value*> res(4);
-
-   res[0] = m_builder.CreateAdd(in1[0], in2[0], name("addx"));
-   res[1] = m_builder.CreateAdd(in1[1], in2[1], name("addy"));
-   res[2] = m_builder.CreateAdd(in1[2], in2[2], name("addz"));
-   res[3] = m_builder.CreateAdd(in1[3], in2[3], name("addw"));
-
-   return res;
-}
-
-std::vector<llvm::Value*> InstructionsSoa::mul(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   std::vector<llvm::Value*> res(4);
-
-   res[0] = m_builder.CreateMul(in1[0], in2[0], name("mulx"));
-   res[1] = m_builder.CreateMul(in1[1], in2[1], name("muly"));
-   res[2] = m_builder.CreateMul(in1[2], in2[2], name("mulz"));
-   res[3] = m_builder.CreateMul(in1[3], in2[3], name("mulw"));
-
-   return res;
-}
-
 void InstructionsSoa::end()
 {
    m_builder.CreateRetVoid();
 }
 
-std::vector<llvm::Value*> InstructionsSoa::madd(const std::vector<llvm::Value*> in1,
-                                                const std::vector<llvm::Value*> in2,
-                                                const std::vector<llvm::Value*> in3)
-{
-   std::vector<llvm::Value*> res = mul(in1, in2);
-   return add(res, in3);
-}
-
 std::vector<llvm::Value*> InstructionsSoa::extractVector(llvm::Value *vector)
 {
    std::vector<llvm::Value*> res(4);
@@ -171,6 +114,11 @@ std::vector<llvm::Value*> InstructionsSoa::extractVector(llvm::Value *vector)
    return res;
 }
 
+llvm::IRBuilder<>* InstructionsSoa::getIRBuilder()
+{
+   return &m_builder;
+}
+
 void InstructionsSoa::createFunctionMap()
 {
    m_functionsMap[TGSI_OPCODE_ABS]   = "abs";
@@ -274,6 +222,41 @@ std::vector<llvm::Value*> InstructionsSoa::abs(const std::vector<llvm::Value*> i
    return callBuiltin(func, in1);
 }
 
+std::vector<llvm::Value*> InstructionsSoa::add(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   std::vector<llvm::Value*> res(4);
+
+   res[0] = m_builder.CreateAdd(in1[0], in2[0], name("addx"));
+   res[1] = m_builder.CreateAdd(in1[1], in2[1], name("addy"));
+   res[2] = m_builder.CreateAdd(in1[2], in2[2], name("addz"));
+   res[3] = m_builder.CreateAdd(in1[3], in2[3], name("addw"));
+
+   return res;
+}
+
+std::vector<llvm::Value*> InstructionsSoa::arl(const std::vector<llvm::Value*> in)
+{
+   std::vector<llvm::Value*> res(4);
+
+   //Extract x's
+   llvm::Value *x1 = m_builder.CreateExtractElement(in[0],
+                                                    m_storage->constantInt(0),
+                                                    name("extractX"));
+   //cast it to an unsigned int
+   x1 = m_builder.CreateFPToUI(x1, IntegerType::get(32), name("x1IntCast"));
+
+   res[0] = x1;//vectorFromVals(x1, x2, x3, x4);
+   //only x is valid. the others shouldn't be necessary
+   /*
+   res[1] = Constant::getNullValue(m_floatVecType);
+   res[2] = Constant::getNullValue(m_floatVecType);
+   res[3] = Constant::getNullValue(m_floatVecType);
+   */
+
+   return res;
+}
+
 std::vector<llvm::Value*> InstructionsSoa::dp3(const std::vector<llvm::Value*> in1,
                                                const std::vector<llvm::Value*> in2)
 {
@@ -281,6 +264,59 @@ std::vector<llvm::Value*> InstructionsSoa::dp3(const std::vector<llvm::Value*> i
    return callBuiltin(func, in1, in2);
 }
 
+std::vector<llvm::Value*> InstructionsSoa::lit(const std::vector<llvm::Value*> in)
+{
+   llvm::Function *func = function(TGSI_OPCODE_LIT);
+   return callBuiltin(func, in);
+}
+
+std::vector<llvm::Value*> InstructionsSoa::madd(const std::vector<llvm::Value*> in1,
+                                                const std::vector<llvm::Value*> in2,
+                                                const std::vector<llvm::Value*> in3)
+{
+   std::vector<llvm::Value*> res = mul(in1, in2);
+   return add(res, in3);
+}
+
+std::vector<llvm::Value*> InstructionsSoa::max(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   llvm::Function *func = function(TGSI_OPCODE_MAX);
+   return callBuiltin(func, in1, in2);
+}
+
+std::vector<llvm::Value*> InstructionsSoa::min(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   llvm::Function *func = function(TGSI_OPCODE_MIN);
+   return callBuiltin(func, in1, in2);
+}
+
+std::vector<llvm::Value*> InstructionsSoa::mul(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   std::vector<llvm::Value*> res(4);
+
+   res[0] = m_builder.CreateMul(in1[0], in2[0], name("mulx"));
+   res[1] = m_builder.CreateMul(in1[1], in2[1], name("muly"));
+   res[2] = m_builder.CreateMul(in1[2], in2[2], name("mulz"));
+   res[3] = m_builder.CreateMul(in1[3], in2[3], name("mulw"));
+
+   return res;
+}
+
+std::vector<llvm::Value*> InstructionsSoa::pow(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   llvm::Function *func = function(TGSI_OPCODE_POWER);
+   return callBuiltin(func, in1, in2);
+}
+
+std::vector<llvm::Value*> InstructionsSoa::rsq(const std::vector<llvm::Value*> in)
+{
+   llvm::Function *func = function(TGSI_OPCODE_RSQ);
+   return callBuiltin(func, in);
+}
 
 std::vector<llvm::Value*> InstructionsSoa::slt(const std::vector<llvm::Value*> in1,
                                                const std::vector<llvm::Value*> in2)
@@ -289,6 +325,37 @@ std::vector<llvm::Value*> InstructionsSoa::slt(const std::vector<llvm::Value*> i
    return callBuiltin(func, in1, in2);
 }
 
+std::vector<llvm::Value*> InstructionsSoa::sub(const std::vector<llvm::Value*> in1,
+                                               const std::vector<llvm::Value*> in2)
+{
+   std::vector<llvm::Value*> res(4);
+
+   res[0] = m_builder.CreateSub(in1[0], in2[0], name("subx"));
+   res[1] = m_builder.CreateSub(in1[1], in2[1], name("suby"));
+   res[2] = m_builder.CreateSub(in1[2], in2[2], name("subz"));
+   res[3] = m_builder.CreateSub(in1[3], in2[3], name("subw"));
+
+   return res;
+}
+
+void checkFunction(Function *func)
+{
+   for (Function::const_iterator BI = func->begin(), BE = func->end();
+        BI != BE; ++BI) {
+      const BasicBlock &BB = *BI;
+      for (BasicBlock::const_iterator II = BB.begin(), IE = BB.end();
+           II != IE; ++II) {
+         const Instruction &I = *II;
+         std::cout<< "Instr = "<<I;
+         for (unsigned op = 0, E = I.getNumOperands(); op != E; ++op) {
+            const Value *Op = I.getOperand(op);
+            std::cout<< "\top = "<<Op<<"("<<op<<")"<<std::endl;
+            //I->setOperand(op, V);
+  }
+      }
+   }
+}
+
 llvm::Value * InstructionsSoa::allocaTemp()
 {
    VectorType *vector   = VectorType::get(Type::FloatTy, 4);
@@ -408,46 +475,6 @@ std::vector<Value*> InstructionsSoa::callBuiltin(llvm::Function *func, const std
    return allocaToResult(allocaPtr);
 }
 
-std::vector<llvm::Value*> InstructionsSoa::pow(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   llvm::Function *func = function(TGSI_OPCODE_POWER);
-   return callBuiltin(func, in1, in2);
-}
-
-std::vector<llvm::Value*> InstructionsSoa::min(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   llvm::Function *func = function(TGSI_OPCODE_MIN);
-   return callBuiltin(func, in1, in2);
-}
-
-
-std::vector<llvm::Value*> InstructionsSoa::max(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   llvm::Function *func = function(TGSI_OPCODE_MAX);
-   return callBuiltin(func, in1, in2);
-}
-
-void checkFunction(Function *func)
-{
-   for (Function::const_iterator BI = func->begin(), BE = func->end();
-        BI != BE; ++BI) {
-      const BasicBlock &BB = *BI;
-      for (BasicBlock::const_iterator II = BB.begin(), IE = BB.end();
-           II != IE; ++II) {
-         const Instruction &I = *II;
-         std::cout<< "Instr = "<<I;
-         for (unsigned op = 0, E = I.getNumOperands(); op != E; ++op) {
-            const Value *Op = I.getOperand(op);
-            std::cout<< "\top = "<<Op<<"("<<op<<")"<<std::endl;
-            //I->setOperand(op, V);
-  }
-      }
-   }
-}
-
 void InstructionsSoa::injectFunction(llvm::Function *originalFunc, int op)
 {
    assert(originalFunc);
@@ -492,28 +519,4 @@ void InstructionsSoa::injectFunction(llvm::Function *originalFunc, int op)
    }
 }
 
-std::vector<llvm::Value*> InstructionsSoa::sub(const std::vector<llvm::Value*> in1,
-                                               const std::vector<llvm::Value*> in2)
-{
-   std::vector<llvm::Value*> res(4);
-
-   res[0] = m_builder.CreateSub(in1[0], in2[0], name("subx"));
-   res[1] = m_builder.CreateSub(in1[1], in2[1], name("suby"));
-   res[2] = m_builder.CreateSub(in1[2], in2[2], name("subz"));
-   res[3] = m_builder.CreateSub(in1[3], in2[3], name("subw"));
-
-   return res;
-}
-
-std::vector<llvm::Value*> InstructionsSoa::lit(const std::vector<llvm::Value*> in)
-{
-   llvm::Function *func = function(TGSI_OPCODE_LIT);
-   return callBuiltin(func, in);
-}
-
-std::vector<llvm::Value*> InstructionsSoa::rsq(const std::vector<llvm::Value*> in)
-{
-   llvm::Function *func = function(TGSI_OPCODE_RSQ);
-   return callBuiltin(func, in);
-}
 
diff --git a/src/gallium/auxiliary/gallivm/instructionssoa.h b/src/gallium/auxiliary/gallivm/instructionssoa.h
index 3817fdc904b..d6831e0a6b9 100644
--- a/src/gallium/auxiliary/gallivm/instructionssoa.h
+++ b/src/gallium/auxiliary/gallivm/instructionssoa.h
@@ -76,6 +76,7 @@ public:
    void         end();
 
    std::vector<llvm::Value*> extractVector(llvm::Value *vector);
+   llvm::IRBuilder<>*  getIRBuilder();
 private:
    const char * name(const char *prefix) const;
    llvm::Value *vectorFromVals(llvm::Value *x, llvm::Value *y,
diff --git a/src/gallium/auxiliary/gallivm/storage.cpp b/src/gallium/auxiliary/gallivm/storage.cpp
index 6f373f6dd5e..73df24c9769 100644
--- a/src/gallium/auxiliary/gallivm/storage.cpp
+++ b/src/gallium/auxiliary/gallivm/storage.cpp
@@ -323,7 +323,7 @@ llvm::Value * Storage::elemIdx(llvm::Value *ptr, int idx,
 
    if (indIdx) {
       getElem = GetElementPtrInst::Create(ptr,
-                                      BinaryOperator::create(Instruction::Add,
+                                      BinaryOperator::Create(Instruction::Add,
                                                              indIdx,
                                                              constantInt(idx),
                                                              name("add"),
diff --git a/src/gallium/auxiliary/gallivm/storagesoa.cpp b/src/gallium/auxiliary/gallivm/storagesoa.cpp
index 78d754371f0..e1e5cabcf55 100644
--- a/src/gallium/auxiliary/gallivm/storagesoa.cpp
+++ b/src/gallium/auxiliary/gallivm/storagesoa.cpp
@@ -48,13 +48,11 @@ using namespace llvm;
 StorageSoa::StorageSoa(llvm::BasicBlock *block,
                        llvm::Value *input,
                        llvm::Value *output,
-                       llvm::Value *consts,
-                       llvm::Value *temps)
+                       llvm::Value *consts)
    : m_block(block),
      m_input(input),
      m_output(output),
      m_consts(consts),
-     m_temps(temps),
      m_immediates(0),
      m_idx(0)
 {
@@ -93,7 +91,7 @@ void StorageSoa::declareImmediates()
       std::vector<float> vals(4);
       std::vector<Constant*> channelArray;
 
-      vals[0] = vec[0]; vals[1] = vec[0]; vals[2] = vec[0]; vals[3] = vec[0];
+      vals[0] = vec[0]; vals[1] = vec[1]; vals[2] = vec[2]; vals[3] = vec[3];
       llvm::Constant *xChannel = createConstGlobalVector(vals);
 
       vals[0] = vec[1]; vals[1] = vec[1]; vals[2] = vec[1]; vals[3] = vec[1];
@@ -144,22 +142,43 @@ std::vector<llvm::Value*> StorageSoa::inputElement(llvm::Value *idx)
    return res;
 }
 
-std::vector<llvm::Value*> StorageSoa::constElement(llvm::Value *idx)
+llvm::Value* StorageSoa::unpackConstElement(llvm::IRBuilder<>* m_builder, llvm::Value* vector, int cc)
 {
-   std::vector<llvm::Value*> res(4);
-   llvm::Value *xChannel, *yChannel, *zChannel, *wChannel;
+   std::vector<llvm::Value*> x(4);
+   x[0] = m_builder->CreateExtractElement(vector,
+                                           constantInt(cc),
+                                           name("x"));
+
+   VectorType  *vectorType = VectorType::get(Type::FloatTy, 4);
+   Constant *constVector = Constant::getNullValue(vectorType);
+   Value *res = m_builder->CreateInsertElement(constVector, x[0],
+                                              constantInt(0),
+                                              name("vecx"));
+   res = m_builder->CreateInsertElement(res, x[0], constantInt(1),
+                               name("vecxx"));
+   res = m_builder->CreateInsertElement(res, x[0], constantInt(2),
+                               name("vecxxx"));
+   res = m_builder->CreateInsertElement(res, x[0], constantInt(3),
+                               name("vecxxxx"));
+   return res;
+}
+
+std::vector<llvm::Value*> StorageSoa::constElement(llvm::IRBuilder<>* m_builder, llvm::Value *idx)
+{
+   llvm::Value* res;
+   std::vector<llvm::Value*> res2(4);
+   llvm::Value *xChannel;
 
    xChannel = elementPointer(m_consts, idx, 0);
-   yChannel = elementPointer(m_consts, idx, 1);
-   zChannel = elementPointer(m_consts, idx, 2);
-   wChannel = elementPointer(m_consts, idx, 3);
 
-   res[0] = alignedArrayLoad(xChannel);
-   res[1] = alignedArrayLoad(yChannel);
-   res[2] = alignedArrayLoad(zChannel);
-   res[3] = alignedArrayLoad(wChannel);
+   res = alignedArrayLoad(xChannel);
 
-   return res;
+   res2[0]=unpackConstElement(m_builder, res,0);
+   res2[1]=unpackConstElement(m_builder, res,1);
+   res2[2]=unpackConstElement(m_builder, res,2);
+   res2[3]=unpackConstElement(m_builder, res,3);
+
+   return res2;
 }
 
 std::vector<llvm::Value*> StorageSoa::outputElement(llvm::Value *idx)
@@ -174,14 +193,15 @@ std::vector<llvm::Value*> StorageSoa::outputElement(llvm::Value *idx)
    return res;
 }
 
-std::vector<llvm::Value*> StorageSoa::tempElement(llvm::Value *idx)
+std::vector<llvm::Value*> StorageSoa::tempElement(llvm::IRBuilder<>* m_builder, int idx)
 {
    std::vector<llvm::Value*> res(4);
+   llvm::Value *temp = m_temps[idx];
 
-   res[0] = element(m_temps, idx, 0);
-   res[1] = element(m_temps, idx, 1);
-   res[2] = element(m_temps, idx, 2);
-   res[3] = element(m_temps, idx, 3);
+   res[0] = element(temp, constantInt(0), 0);
+   res[1] = element(temp, constantInt(0), 1);
+   res[2] = element(temp, constantInt(0), 2);
+   res[3] = element(temp, constantInt(0), 3);
 
    return res;
 }
@@ -260,6 +280,12 @@ llvm::Module * StorageSoa::currentModule() const
     return m_block->getParent()->getParent();
 }
 
+llvm::Constant * StorageSoa::createConstGlobalFloat(const float val)
+{
+   Constant*c = ConstantFP::get(APFloat(val));
+   return c;
+}
+
 llvm::Constant * StorageSoa::createConstGlobalVector(const std::vector<float> &vec)
 {
    VectorType *vectorType = VectorType::get(Type::FloatTy, 4);
@@ -278,7 +304,7 @@ llvm::Constant * StorageSoa::createConstGlobalVector(const std::vector<float> &v
 }
 
 std::vector<llvm::Value*> StorageSoa::load(enum tgsi_file_type type, int idx, int swizzle,
-                                           llvm::Value *indIdx)
+                                           llvm::IRBuilder<>* m_builder,llvm::Value *indIdx)
 {
    std::vector<llvm::Value*> val(4);
 
@@ -299,10 +325,10 @@ std::vector<llvm::Value*> StorageSoa::load(enum tgsi_file_type type, int idx, in
       val = outputElement(realIndex);
       break;
    case TGSI_FILE_TEMPORARY:
-      val = tempElement(realIndex);
+      val = tempElement(m_builder, idx);
       break;
    case TGSI_FILE_CONSTANT:
-      val = constElement(realIndex);
+      val = constElement(m_builder, realIndex);
       break;
    case TGSI_FILE_IMMEDIATE:
       val = immediateElement(realIndex);
@@ -328,19 +354,39 @@ std::vector<llvm::Value*> StorageSoa::load(enum tgsi_file_type type, int idx, in
    return res;
 }
 
+llvm::Value * StorageSoa::allocaTemp(llvm::IRBuilder<>* m_builder)
+{
+   VectorType *vector   = VectorType::get(Type::FloatTy, 4);
+   ArrayType  *vecArray = ArrayType::get(vector, 4);
+   AllocaInst *alloca = new AllocaInst(vecArray, "temp",
+                                       m_builder->GetInsertBlock());
+
+   return alloca;
+}
+
+
 void StorageSoa::store(enum tgsi_file_type type, int idx, const std::vector<llvm::Value*> &val,
-                       int mask)
+                       int mask, llvm::IRBuilder<>* m_builder)
 {
    llvm::Value *out = 0;
+   llvm::Value *realIndex = 0;
    switch(type) {
    case TGSI_FILE_OUTPUT:
       out = m_output;
+      realIndex = constantInt(idx);
       break;
    case TGSI_FILE_TEMPORARY:
-      out = m_temps;
+      // if that temp doesn't already exist, alloca it
+      if (m_temps.find(idx) == m_temps.end())
+         m_temps[idx] = allocaTemp(m_builder);
+
+      out = m_temps[idx];
+
+      realIndex = constantInt(0);
       break;
    case TGSI_FILE_INPUT:
       out = m_input;
+      realIndex = constantInt(idx);
       break;
    case TGSI_FILE_ADDRESS: {
       llvm::Value *addr = m_addresses[idx];
@@ -358,7 +404,6 @@ void StorageSoa::store(enum tgsi_file_type type, int idx, const std::vector<llvm
       assert(0);
       break;
    }
-   llvm::Value *realIndex = constantInt(idx);
    if ((mask & TGSI_WRITEMASK_X)) {
       llvm::Value *xChannel = elementPointer(out, realIndex, 0);
       new StoreInst(val[0], xChannel, false, m_block);
diff --git a/src/gallium/auxiliary/gallivm/storagesoa.h b/src/gallium/auxiliary/gallivm/storagesoa.h
index ae2fc7c6aee..56886f85e7a 100644
--- a/src/gallium/auxiliary/gallivm/storagesoa.h
+++ b/src/gallium/auxiliary/gallivm/storagesoa.h
@@ -29,6 +29,7 @@
 #define STORAGESOA_H
 
 #include <pipe/p_shader_tokens.h>
+#include <llvm/Support/IRBuilder.h>
 
 #include <vector>
 #include <list>
@@ -51,14 +52,13 @@ public:
    StorageSoa(llvm::BasicBlock *block,
               llvm::Value *input,
               llvm::Value *output,
-              llvm::Value *consts,
-              llvm::Value *temps);
+              llvm::Value *consts);
 
 
    std::vector<llvm::Value*> load(enum tgsi_file_type type, int idx, int swizzle, 
-                                  llvm::Value *indIdx =0);
+                                  llvm::IRBuilder<>* m_builder, llvm::Value *indIdx =0);
    void store(enum tgsi_file_type type, int idx, const std::vector<llvm::Value*> &val,
-              int mask);
+              int mask, llvm::IRBuilder<>* m_builder);
 
    void addImmediate(float *vec);
    void declareImmediates();
@@ -76,12 +76,14 @@ private:
    const char *name(const char *prefix) const;
    llvm::Value  *alignedArrayLoad(llvm::Value *val);
    llvm::Module *currentModule() const;
+   llvm::Constant  *createConstGlobalFloat(const float val);
    llvm::Constant  *createConstGlobalVector(const std::vector<float> &vec);
 
    std::vector<llvm::Value*> inputElement(llvm::Value *indIdx);
-   std::vector<llvm::Value*> constElement(llvm::Value *indIdx);
+   llvm::Value* unpackConstElement(llvm::IRBuilder<>* m_builder, llvm::Value *indIdx, int cc);
+   std::vector<llvm::Value*> constElement(llvm::IRBuilder<>* m_builder, llvm::Value *indIdx);
    std::vector<llvm::Value*> outputElement(llvm::Value *indIdx);
-   std::vector<llvm::Value*> tempElement(llvm::Value *indIdx);
+   std::vector<llvm::Value*> tempElement(llvm::IRBuilder<>* m_builder, int idx);
    std::vector<llvm::Value*> immediateElement(llvm::Value *indIdx);
 private:
    llvm::BasicBlock *m_block;
@@ -89,12 +91,13 @@ private:
    llvm::Value *m_input;
    llvm::Value *m_output;
    llvm::Value *m_consts;
-   llvm::Value *m_temps;
+   std::map<int, llvm::Value*> m_temps;
    llvm::GlobalVariable *m_immediates;
 
    std::map<int, llvm::Value*> m_addresses;
 
    std::vector<std::vector<float> > m_immediatesToFlush;
+   llvm::Value * allocaTemp(llvm::IRBuilder<>* m_builder);
 
    mutable std::map<int, llvm::ConstantInt*> m_constInts;
    mutable char        m_name[32];
diff --git a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
index 7292c0e366c..c11b88af9ec 100644
--- a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
+++ b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
@@ -52,8 +52,7 @@ static inline FunctionType *vertexShaderFunctionType()
    // pass are castable to the following:
    // [4 x <4 x float>] inputs,
    // [4 x <4 x float>] output,
-   // [4 x [4 x float]] consts,
-   // [4 x <4 x float>] temps
+   // [4 x [1 x float]] consts,
 
    std::vector<const Type*> funcArgs;
    VectorType *vectorType = VectorType::get(Type::FloatTy, 4);
@@ -61,13 +60,12 @@ static inline FunctionType *vertexShaderFunctionType()
    PointerType *vectorArrayPtr = PointerType::get(vectorArray, 0);
 
    ArrayType   *floatArray     = ArrayType::get(Type::FloatTy, 4);
-   ArrayType   *constsArray    = ArrayType::get(floatArray, 4);
+   ArrayType   *constsArray    = ArrayType::get(floatArray, 1);
    PointerType *constsArrayPtr = PointerType::get(constsArray, 0);
 
    funcArgs.push_back(vectorArrayPtr);//inputs
    funcArgs.push_back(vectorArrayPtr);//output
    funcArgs.push_back(constsArrayPtr);//consts
-   funcArgs.push_back(vectorArrayPtr);//temps
 
    FunctionType *functionType = FunctionType::get(
       /*Result=*/Type::VoidTy,
@@ -707,9 +705,8 @@ translate_instructionir(llvm::Module *module,
       if (src->SrcRegister.Indirect) {
          indIdx = storage->addrElement(src->SrcRegisterInd.Index);
       }
-
       val = storage->load((enum tgsi_file_type)src->SrcRegister.File,
-                          src->SrcRegister.Index, swizzle, indIdx);
+                          src->SrcRegister.Index, swizzle, instr->getIRBuilder(), indIdx);
 
       inputs[i] = val;
    }
@@ -1025,9 +1022,9 @@ translate_instructionir(llvm::Module *module,
    /* store results  */
    for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) {
       struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
-
       storage->store((enum tgsi_file_type)dst->DstRegister.File,
-                     dst->DstRegister.Index, out, dst->DstRegister.WriteMask);
+                     dst->DstRegister.Index, out, dst->DstRegister.WriteMask,
+		     instr->getIRBuilder() );
    }
 }
 
@@ -1122,8 +1119,6 @@ llvm::Module * tgsi_to_llvmir(struct gallivm_ir *ir,
    output->setName("outputs");
    Value *consts = args++;
    consts->setName("consts");
-   Value *temps = args++;
-   temps->setName("temps");
 
    BasicBlock *label_entry = BasicBlock::Create("entry", shader, 0);
 
@@ -1132,7 +1127,7 @@ llvm::Module * tgsi_to_llvmir(struct gallivm_ir *ir,
    fi = tgsi_default_full_instruction();
    fd = tgsi_default_full_declaration();
 
-   StorageSoa storage(label_entry, input, output, consts, temps);
+   StorageSoa storage(label_entry, input, output, consts);
    InstructionsSoa instr(mod, shader, label_entry, &storage);
 
    while(!tgsi_parse_end_of_tokens(&parse)) {
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer.h b/src/gallium/auxiliary/pipebuffer/pb_buffer.h
index 8505d333bda..4f1a8a38942 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer.h
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer.h
@@ -183,6 +183,7 @@ pb_destroy(struct pb_buffer *buf)
    assert(buf);
    if(!buf)
       return;
+   assert(buf->base.refcount == 0);
    buf->vtbl->destroy(buf);
 }
 
@@ -193,11 +194,16 @@ static INLINE void
 pb_reference(struct pb_buffer **dst,
              struct pb_buffer *src)
 {
-   if (src) 
+   if (src) {
+      assert(src->base.refcount);
       src->base.refcount++;
+   }
 
-   if (*dst && --(*dst)->base.refcount == 0)
-      pb_destroy( *dst );
+   if (*dst) {
+      assert((*dst)->base.refcount);
+      if(--(*dst)->base.refcount == 0)
+         pb_destroy( *dst );
+   }
 
    *dst = src;
 }
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c
index 633ee70a75b..e2594ea2369 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_fenced.c
@@ -86,8 +86,7 @@ fenced_bufmgr_create_buffer(struct pb_manager *mgr,
    
    fenced_buf = fenced_buffer_create(fenced_mgr->fenced_list, buf);
    if(!fenced_buf) {
-      assert(buf->base.refcount == 1);
-      pb_destroy(buf);
+      pb_reference(&buf, NULL);
    }
    
    return fenced_buf;
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_mm.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_mm.c
index fe80ca30eea..a976d3041ac 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_mm.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_mm.c
@@ -100,7 +100,7 @@ mm_buffer_destroy(struct pb_buffer *buf)
    assert(buf->base.refcount == 0);
    
    pipe_mutex_lock(mm->mutex);
-   mmFreeMem(mm_buf->block);
+   u_mmFreeMem(mm_buf->block);
    FREE(buf);
    pipe_mutex_unlock(mm->mutex);
 }
@@ -175,14 +175,14 @@ mm_bufmgr_create_buffer(struct pb_manager *mgr,
    
    mm_buf->mgr = mm;
    
-   mm_buf->block = mmAllocMem(mm->heap, size, mm->align2, 0);
+   mm_buf->block = u_mmAllocMem(mm->heap, size, mm->align2, 0);
    if(!mm_buf->block) {
       debug_printf("warning: heap full\n");
 #if 0
       mmDumpMemInfo(mm->heap);
 #endif
       
-      mm_buf->block = mmAllocMem(mm->heap, size, mm->align2, 0);
+      mm_buf->block = u_mmAllocMem(mm->heap, size, mm->align2, 0);
       if(!mm_buf->block) {
          FREE(mm_buf);
          pipe_mutex_unlock(mm->mutex);
@@ -213,7 +213,7 @@ mm_bufmgr_destroy(struct pb_manager *mgr)
    
    pipe_mutex_lock(mm->mutex);
 
-   mmDestroy(mm->heap);
+   u_mmDestroy(mm->heap);
    
    pb_unmap(mm->buffer);
    pb_reference(&mm->buffer, NULL);
@@ -254,7 +254,7 @@ mm_bufmgr_create_from_buffer(struct pb_buffer *buffer,
    if(!mm->map)
       goto failure;
 
-   mm->heap = mmInit(0, size); 
+   mm->heap = u_mmInit(0, size); 
    if (!mm->heap)
       goto failure;
 
@@ -262,7 +262,7 @@ mm_bufmgr_create_from_buffer(struct pb_buffer *buffer,
    
 failure:
 if(mm->heap)
-   mmDestroy(mm->heap);
+   u_mmDestroy(mm->heap);
    if(mm->map)
       pb_unmap(mm->buffer);
    if(mm)
diff --git a/src/gallium/auxiliary/rtasm/Makefile b/src/gallium/auxiliary/rtasm/Makefile
index 39b8a4dbd7a..252dc5274ab 100644
--- a/src/gallium/auxiliary/rtasm/Makefile
+++ b/src/gallium/auxiliary/rtasm/Makefile
@@ -7,6 +7,7 @@ C_SOURCES = \
 	rtasm_cpu.c \
 	rtasm_execmem.c \
 	rtasm_x86sse.c \
+	rtasm_ppc.c \
 	rtasm_ppc_spe.c
 
 include ../../Makefile.template
diff --git a/src/gallium/auxiliary/rtasm/SConscript b/src/gallium/auxiliary/rtasm/SConscript
index 8ea25922aa1..eb48368accb 100644
--- a/src/gallium/auxiliary/rtasm/SConscript
+++ b/src/gallium/auxiliary/rtasm/SConscript
@@ -6,6 +6,7 @@ rtasm = env.ConvenienceLibrary(
 		'rtasm_cpu.c',
 		'rtasm_execmem.c',
 		'rtasm_x86sse.c',
+		'rtasm_ppc.c',
 		'rtasm_ppc_spe.c',
 	])
 
diff --git a/src/gallium/auxiliary/rtasm/rtasm_execmem.c b/src/gallium/auxiliary/rtasm/rtasm_execmem.c
index f16191cb619..be7433baf87 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_execmem.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_execmem.c
@@ -38,12 +38,13 @@
 #include "rtasm_execmem.h"
 
 
-#if defined(__linux__)
+#if defined(PIPE_OS_LINUX)
+
 
 /*
  * Allocate a large block of memory which can hold code then dole it out
  * in pieces by means of the generic memory manager code.
-*/
+ */
 
 #include <unistd.h>
 #include <sys/mman.h>
@@ -62,7 +63,7 @@ static void
 init_heap(void)
 {
    if (!exec_heap)
-      exec_heap = mmInit( 0, EXEC_HEAP_SIZE );
+      exec_heap = u_mmInit( 0, EXEC_HEAP_SIZE );
    
    if (!exec_mem)
       exec_mem = (unsigned char *) mmap(0, EXEC_HEAP_SIZE, 
@@ -83,7 +84,7 @@ rtasm_exec_malloc(size_t size)
 
    if (exec_heap) {
       size = (size + 31) & ~31;  /* next multiple of 32 bytes */
-      block = mmAllocMem( exec_heap, size, 5, 0 ); /* 5 -> 32-byte alignment */
+      block = u_mmAllocMem( exec_heap, size, 5, 0 ); /* 5 -> 32-byte alignment */
    }
 
    if (block)
@@ -103,17 +104,17 @@ rtasm_exec_free(void *addr)
    pipe_mutex_lock(exec_mutex);
 
    if (exec_heap) {
-      struct mem_block *block = mmFindBlock(exec_heap, (unsigned char *)addr - exec_mem);
+      struct mem_block *block = u_mmFindBlock(exec_heap, (unsigned char *)addr - exec_mem);
    
       if (block)
-	 mmFreeMem(block);
+	 u_mmFreeMem(block);
    }
 
    pipe_mutex_unlock(exec_mutex);
 }
 
 
-#else
+#else /* PIPE_OS_LINUX */
 
 /*
  * Just use regular memory.
@@ -133,4 +134,4 @@ rtasm_exec_free(void *addr)
 }
 
 
-#endif
+#endif /* PIPE_OS_LINUX */
diff --git a/src/gallium/auxiliary/rtasm/rtasm_ppc.c b/src/gallium/auxiliary/rtasm/rtasm_ppc.c
new file mode 100644
index 00000000000..b65bfa7bbdf
--- /dev/null
+++ b/src/gallium/auxiliary/rtasm/rtasm_ppc.c
@@ -0,0 +1,959 @@
+/**************************************************************************
+ *
+ * Copyright (C) 2008 Tungsten Graphics, Inc.   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * PPC code generation.
+ * For reference, see http://www.power.org/resources/reading/PowerISA_V2.05.pdf
+ * ABI info: http://www.cs.utsa.edu/~whaley/teach/cs6463FHPO/LEC/lec12_ho.pdf
+ *
+ * Other PPC refs:
+ * http://www-01.ibm.com/chips/techlib/techlib.nsf/techdocs/852569B20050FF778525699600719DF2
+ * http://www.ibm.com/developerworks/eserver/library/es-archguide-v2.html
+ * http://www.freescale.com/files/product/doc/MPCFPE32B.pdf
+ *
+ * \author Brian Paul
+ */
+
+
+#include <stdio.h>
+#include "util/u_memory.h"
+#include "pipe/p_debug.h"
+#include "rtasm_execmem.h"
+#include "rtasm_ppc.h"
+
+
+void
+ppc_init_func(struct ppc_function *p)
+{
+   uint i;
+
+   p->num_inst = 0;
+   p->max_inst = 100; /* first guess at buffer size */
+   p->store = rtasm_exec_malloc(p->max_inst * PPC_INST_SIZE);
+   p->reg_used = 0x0;
+   p->fp_used = 0x0;
+   p->vec_used = 0x0;
+
+   /* only allow using gp registers 3..12 for now */
+   for (i = 0; i < 3; i++)
+      ppc_reserve_register(p, i);
+   for (i = 12; i < PPC_NUM_REGS; i++)
+      ppc_reserve_register(p, i);
+}
+
+
+void
+ppc_release_func(struct ppc_function *p)
+{
+   assert(p->num_inst <= p->max_inst);
+   if (p->store != NULL) {
+      rtasm_exec_free(p->store);
+   }
+   p->store = NULL;
+}
+
+
+uint
+ppc_num_instructions(const struct ppc_function *p)
+{
+   return p->num_inst;
+}
+
+
+void (*ppc_get_func(struct ppc_function *p))(void)
+{
+#if 0
+   DUMP_END();
+   if (DISASSEM && p->store)
+      debug_printf("disassemble %p %p\n", p->store, p->csr);
+
+   if (p->store == p->error_overflow)
+      return (void (*)(void)) NULL;
+   else
+#endif
+      return (void (*)(void)) p->store;
+}
+
+
+void
+ppc_dump_func(const struct ppc_function *p)
+{
+   uint i;
+   for (i = 0; i < p->num_inst; i++) {
+      debug_printf("%3u: 0x%08x\n", i, p->store[i]);
+   }
+}
+
+
+/**
+ * Mark a register as being unavailable.
+ */
+int
+ppc_reserve_register(struct ppc_function *p, int reg)
+{
+   assert(reg < PPC_NUM_REGS);
+   p->reg_used |= (1 << reg);
+   return reg;
+}
+
+
+/**
+ * Allocate a general purpose register.
+ * \return register index or -1 if none left.
+ */
+int
+ppc_allocate_register(struct ppc_function *p)
+{
+   unsigned i;
+   for (i = 0; i < PPC_NUM_REGS; i++) {
+      const uint64_t mask = 1 << i;
+      if ((p->reg_used & mask) == 0) {
+         p->reg_used |= mask;
+         return i;
+      }
+   }
+   return -1;
+}
+
+
+/**
+ * Mark the given general purpose register as "unallocated".
+ */
+void
+ppc_release_register(struct ppc_function *p, int reg)
+{
+   assert(reg < PPC_NUM_REGS);
+   assert(p->reg_used & (1 << reg));
+   p->reg_used &= ~(1 << reg);
+}
+
+
+/**
+ * Allocate a floating point register.
+ * \return register index or -1 if none left.
+ */
+int
+ppc_allocate_fp_register(struct ppc_function *p)
+{
+   unsigned i;
+   for (i = 0; i < PPC_NUM_FP_REGS; i++) {
+      const uint64_t mask = 1 << i;
+      if ((p->fp_used & mask) == 0) {
+         p->fp_used |= mask;
+         return i;
+      }
+   }
+   return -1;
+}
+
+
+/**
+ * Mark the given floating point register as "unallocated".
+ */
+void
+ppc_release_fp_register(struct ppc_function *p, int reg)
+{
+   assert(reg < PPC_NUM_FP_REGS);
+   assert(p->fp_used & (1 << reg));
+   p->fp_used &= ~(1 << reg);
+}
+
+
+/**
+ * Allocate a vector register.
+ * \return register index or -1 if none left.
+ */
+int
+ppc_allocate_vec_register(struct ppc_function *p)
+{
+   unsigned i;
+   for (i = 0; i < PPC_NUM_VEC_REGS; i++) {
+      const uint64_t mask = 1 << i;
+      if ((p->vec_used & mask) == 0) {
+         p->vec_used |= mask;
+         return i;
+      }
+   }
+   return -1;
+}
+
+
+/**
+ * Mark the given vector register as "unallocated".
+ */
+void
+ppc_release_vec_register(struct ppc_function *p, int reg)
+{
+   assert(reg < PPC_NUM_VEC_REGS);
+   assert(p->vec_used & (1 << reg));
+   p->vec_used &= ~(1 << reg);
+}
+
+
+/**
+ * Append instruction to instruction buffer.  Grow buffer if out of room.
+ */
+static void
+emit_instruction(struct ppc_function *p, uint32_t inst_bits)
+{
+   if (!p->store)
+      return;  /* out of memory, drop the instruction */
+
+   if (p->num_inst == p->max_inst) {
+      /* allocate larger buffer */
+      uint32_t *newbuf;
+      p->max_inst *= 2;  /* 2x larger */
+      newbuf = rtasm_exec_malloc(p->max_inst * PPC_INST_SIZE);
+      if (newbuf) {
+         memcpy(newbuf, p->store, p->num_inst * PPC_INST_SIZE);
+      }
+      rtasm_exec_free(p->store);
+      p->store = newbuf;
+      if (!p->store) {
+         /* out of memory */
+         p->num_inst = 0;
+         return;
+      }
+   }
+
+   p->store[p->num_inst++] = inst_bits;
+}
+
+
+union vx_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned vD:5;
+      unsigned vA:5;
+      unsigned vB:5;
+      unsigned op2:11;
+   } inst;
+};
+
+static INLINE void
+emit_vx(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB)
+{
+   union vx_inst inst;
+   inst.inst.op = 4;
+   inst.inst.vD = vD;
+   inst.inst.vA = vA;
+   inst.inst.vB = vB;
+   inst.inst.op2 = op2;
+   emit_instruction(p, inst.bits);
+};
+
+
+union vxr_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned vD:5;
+      unsigned vA:5;
+      unsigned vB:5;
+      unsigned rC:1;
+      unsigned op2:10;
+   } inst;
+};
+
+static INLINE void
+emit_vxr(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB)
+{
+   union vxr_inst inst;
+   inst.inst.op = 4;
+   inst.inst.vD = vD;
+   inst.inst.vA = vA;
+   inst.inst.vB = vB;
+   inst.inst.rC = 0;
+   inst.inst.op2 = op2;
+   emit_instruction(p, inst.bits);
+};
+
+
+union va_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned vD:5;
+      unsigned vA:5;
+      unsigned vB:5;
+      unsigned vC:5;
+      unsigned op2:6;
+   } inst;
+};
+
+static INLINE void
+emit_va(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB, uint vC)
+{
+   union va_inst inst;
+   inst.inst.op = 4;
+   inst.inst.vD = vD;
+   inst.inst.vA = vA;
+   inst.inst.vB = vB;
+   inst.inst.vC = vC;
+   inst.inst.op2 = op2;
+   emit_instruction(p, inst.bits);
+};
+
+
+union i_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned li:24;
+      unsigned aa:1;
+      unsigned lk:1;
+   } inst;
+};
+
+static INLINE void
+emit_i(struct ppc_function *p, uint op, uint li, uint aa, uint lk)
+{
+   union i_inst inst;
+   inst.inst.op = op;
+   inst.inst.li = li;
+   inst.inst.aa = aa;
+   inst.inst.lk = lk;
+   emit_instruction(p, inst.bits);
+}
+
+
+union xl_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned bo:5;
+      unsigned bi:5;
+      unsigned unused:3;
+      unsigned bh:2;
+      unsigned op2:10;
+      unsigned lk:1;
+   } inst;
+};
+
+static INLINE void
+emit_xl(struct ppc_function *p, uint op, uint bo, uint bi, uint bh,
+        uint op2, uint lk)
+{
+   union xl_inst inst;
+   inst.inst.op = op;
+   inst.inst.bo = bo;
+   inst.inst.bi = bi;
+   inst.inst.unused = 0x0;
+   inst.inst.bh = bh;
+   inst.inst.op2 = op2;
+   inst.inst.lk = lk;
+   emit_instruction(p, inst.bits);
+}
+
+static INLINE void
+dump_xl(const char *name, uint inst)
+{
+   union xl_inst i;
+
+   i.bits = inst;
+   debug_printf("%s = 0x%08x\n", name, inst);
+   debug_printf(" op: %d 0x%x\n", i.inst.op, i.inst.op);
+   debug_printf(" bo: %d 0x%x\n", i.inst.bo, i.inst.bo);
+   debug_printf(" bi: %d 0x%x\n", i.inst.bi, i.inst.bi);
+   debug_printf(" unused: %d 0x%x\n", i.inst.unused, i.inst.unused);
+   debug_printf(" bh: %d 0x%x\n", i.inst.bh, i.inst.bh);
+   debug_printf(" op2: %d 0x%x\n", i.inst.op2, i.inst.op2);
+   debug_printf(" lk: %d 0x%x\n", i.inst.lk, i.inst.lk);
+}
+
+
+union x_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned vrs:5;
+      unsigned ra:5;
+      unsigned rb:5;
+      unsigned op2:10;
+      unsigned unused:1;
+   } inst;
+};
+
+static INLINE void
+emit_x(struct ppc_function *p, uint op, uint vrs, uint ra, uint rb, uint op2)
+{
+   union x_inst inst;
+   inst.inst.op = op;
+   inst.inst.vrs = vrs;
+   inst.inst.ra = ra;
+   inst.inst.rb = rb;
+   inst.inst.op2 = op2;
+   inst.inst.unused = 0x0;
+   emit_instruction(p, inst.bits);
+}
+
+
+union d_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned rt:5;
+      unsigned ra:5;
+      unsigned si:16;
+   } inst;
+};
+
+static INLINE void
+emit_d(struct ppc_function *p, uint op, uint rt, uint ra, int si)
+{
+   union d_inst inst;
+   assert(si >= -32768);
+   assert(si <= 32767);
+   inst.inst.op = op;
+   inst.inst.rt = rt;
+   inst.inst.ra = ra;
+   inst.inst.si = (unsigned) (si & 0xffff);
+   emit_instruction(p, inst.bits);
+};
+
+
+union a_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned frt:5;
+      unsigned fra:5;
+      unsigned frb:5;
+      unsigned unused:5;
+      unsigned op2:5;
+      unsigned rc:1;
+   } inst;
+};
+
+static INLINE void
+emit_a(struct ppc_function *p, uint op, uint frt, uint fra, uint frb, uint op2,
+       uint rc)
+{
+   union a_inst inst;
+   inst.inst.op = op;
+   inst.inst.frt = frt;
+   inst.inst.fra = fra;
+   inst.inst.frb = frb;
+   inst.inst.unused = 0x0;
+   inst.inst.op2 = op2;
+   inst.inst.rc = rc;
+   emit_instruction(p, inst.bits);
+};
+
+
+union xo_inst {
+   uint32_t bits;
+   struct {
+      unsigned op:6;
+      unsigned rt:5;
+      unsigned ra:5;
+      unsigned rb:5;
+      unsigned oe:1;
+      unsigned op2:9;
+      unsigned rc:1;
+   } inst;
+};
+
+static INLINE void
+emit_xo(struct ppc_function *p, uint op, uint rt, uint ra, uint rb, uint oe,
+        uint op2, uint rc)
+{
+   union xo_inst inst;
+   inst.inst.op = op;
+   inst.inst.rt = rt;
+   inst.inst.ra = ra;
+   inst.inst.rb = rb;
+   inst.inst.oe = oe;
+   inst.inst.op2 = op2;
+   inst.inst.rc = rc;
+   emit_instruction(p, inst.bits);
+}
+
+
+
+
+
+/**
+ ** float vector arithmetic
+ **/
+
+/** vector float add */
+void
+ppc_vaddfp(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 10, vD, vA, vB);
+}
+
+/** vector float substract */
+void
+ppc_vsubfp(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 74, vD, vA, vB);
+}
+
+/** vector float min */
+void
+ppc_vminfp(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1098, vD, vA, vB);
+}
+
+/** vector float max */
+void
+ppc_vmaxfp(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1034, vD, vA, vB);
+}
+
+/** vector float mult add: vD = vA * vB + vC */
+void
+ppc_vmaddfp(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC)
+{
+   emit_va(p, 46, vD, vA, vC, vB); /* note arg order */
+}
+
+/** vector float negative mult subtract: vD = vA - vB * vC */
+void
+ppc_vnmsubfp(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC)
+{
+   emit_va(p, 47, vD, vB, vA, vC); /* note arg order */
+}
+
+/** vector float compare greater than */
+void
+ppc_vcmpgtfpx(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vxr(p, 710, vD, vA, vB);
+}
+
+/** vector float compare greater than or equal to */
+void
+ppc_vcmpgefpx(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vxr(p, 454, vD, vA, vB);
+}
+
+/** vector float compare equal */
+void
+ppc_vcmpeqfpx(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vxr(p, 198, vD, vA, vB);
+}
+
+/** vector float 2^x */
+void
+ppc_vexptefp(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 394, vD, 0, vB);
+}
+
+/** vector float log2(x) */
+void
+ppc_vlogefp(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 458, vD, 0, vB);
+}
+
+/** vector float reciprocol */
+void
+ppc_vrefp(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 266, vD, 0, vB);
+}
+
+/** vector float reciprocol sqrt estimate */
+void
+ppc_vrsqrtefp(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 330, vD, 0, vB);
+}
+
+/** vector float round to negative infinity */
+void
+ppc_vrfim(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 714, vD, 0, vB);
+}
+
+/** vector float round to positive infinity */
+void
+ppc_vrfip(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 650, vD, 0, vB);
+}
+
+/** vector float round to nearest int */
+void
+ppc_vrfin(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 522, vD, 0, vB);
+}
+
+/** vector float round to int toward zero */
+void
+ppc_vrfiz(struct ppc_function *p, uint vD, uint vB)
+{
+   emit_vx(p, 586, vD, 0, vB);
+}
+
+/** vector store: store vR at mem[vA+vB] */
+void
+ppc_stvx(struct ppc_function *p, uint vR, uint vA, uint vB)
+{
+   emit_x(p, 31, vR, vA, vB, 231);
+}
+
+/** vector load: vR = mem[vA+vB] */
+void
+ppc_lvx(struct ppc_function *p, uint vR, uint vA, uint vB)
+{
+   emit_x(p, 31, vR, vA, vB, 103);
+}
+
+/** load vector element word: vR = mem_word[ra+rb] */
+void
+ppc_lvewx(struct ppc_function *p, uint vr, uint ra, uint rb)
+{
+   emit_x(p, 31, vr, ra, rb, 71);
+}
+
+
+
+
+/**
+ ** vector bitwise operations
+ **/
+
+/** vector and */
+void
+ppc_vand(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1028, vD, vA, vB);
+}
+
+/** vector and complement */
+void
+ppc_vandc(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1092, vD, vA, vB);
+}
+
+/** vector or */
+void
+ppc_vor(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1156, vD, vA, vB);
+}
+
+/** vector nor */
+void
+ppc_vnor(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1284, vD, vA, vB);
+}
+
+/** vector xor */
+void
+ppc_vxor(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 1220, vD, vA, vB);
+}
+
+/** Pseudo-instruction: vector move */
+void
+ppc_vmove(struct ppc_function *p, uint vD, uint vA)
+{
+   ppc_vor(p, vD, vA, vA);
+}
+
+/** Set vector register to {0,0,0,0} */
+void
+ppc_vzero(struct ppc_function *p, uint vr)
+{
+   ppc_vxor(p, vr, vr, vr);
+}
+
+
+
+
+/**
+ ** Vector shuffle / select / splat / etc
+ **/
+
+/** vector permute */
+void
+ppc_vperm(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC)
+{
+   emit_va(p, 43, vD, vA, vB, vC);
+}
+
+/** vector select */
+void
+ppc_vsel(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC)
+{
+   emit_va(p, 42, vD, vA, vB, vC);
+}
+
+/** vector splat byte */
+void
+ppc_vspltb(struct ppc_function *p, uint vD, uint vB, uint imm)
+{
+   emit_vx(p, 42, vD, imm, vB);
+}
+
+/** vector splat half word */
+void
+ppc_vsplthw(struct ppc_function *p, uint vD, uint vB, uint imm)
+{
+   emit_vx(p, 588, vD, imm, vB);
+}
+
+/** vector splat word */
+void
+ppc_vspltw(struct ppc_function *p, uint vD, uint vB, uint imm)
+{
+   emit_vx(p, 652, vD, imm, vB);
+}
+
+/** vector splat signed immediate word */
+void
+ppc_vspltisw(struct ppc_function *p, uint vD, int imm)
+{
+   assert(imm >= -16);
+   assert(imm < 15);
+   emit_vx(p, 908, vD, imm, 0);
+}
+
+/** vector shift left word: vD[word] = vA[word] << (vB[word] & 0x1f) */
+void
+ppc_vslw(struct ppc_function *p, uint vD, uint vA, uint vB)
+{
+   emit_vx(p, 388, vD, vA, vB);
+}
+
+
+
+
+/**
+ ** integer arithmetic
+ **/
+
+/** rt = ra + imm */
+void
+ppc_addi(struct ppc_function *p, uint rt, uint ra, int imm)
+{
+   emit_d(p, 14, rt, ra, imm);
+}
+
+/** rt = ra + (imm << 16) */
+void
+ppc_addis(struct ppc_function *p, uint rt, uint ra, int imm)
+{
+   emit_d(p, 15, rt, ra, imm);
+}
+
+/** rt = ra + rb */
+void
+ppc_add(struct ppc_function *p, uint rt, uint ra, uint rb)
+{
+   emit_xo(p, 31, rt, ra, rb, 0, 266, 0);
+}
+
+/** rt = ra AND ra */
+void
+ppc_and(struct ppc_function *p, uint rt, uint ra, uint rb)
+{
+   emit_x(p, 31, ra, rt, rb, 28);  /* note argument order */
+}
+
+/** rt = ra AND imm */
+void
+ppc_andi(struct ppc_function *p, uint rt, uint ra, int imm)
+{
+   emit_d(p, 28, ra, rt, imm);  /* note argument order */
+}
+
+/** rt = ra OR ra */
+void
+ppc_or(struct ppc_function *p, uint rt, uint ra, uint rb)
+{
+   emit_x(p, 31, ra, rt, rb, 444);  /* note argument order */
+}
+
+/** rt = ra OR imm */
+void
+ppc_ori(struct ppc_function *p, uint rt, uint ra, int imm)
+{
+   emit_d(p, 24, ra, rt, imm);  /* note argument order */
+}
+
+/** rt = ra XOR ra */
+void
+ppc_xor(struct ppc_function *p, uint rt, uint ra, uint rb)
+{
+   emit_x(p, 31, ra, rt, rb, 316);  /* note argument order */
+}
+
+/** rt = ra XOR imm */
+void
+ppc_xori(struct ppc_function *p, uint rt, uint ra, int imm)
+{
+   emit_d(p, 26, ra, rt, imm);  /* note argument order */
+}
+
+/** pseudo instruction: move: rt = ra */
+void
+ppc_mr(struct ppc_function *p, uint rt, uint ra)
+{
+   ppc_or(p, rt, ra, ra);
+}
+
+/** pseudo instruction: load immediate: rt = imm */
+void
+ppc_li(struct ppc_function *p, uint rt, int imm)
+{
+   ppc_addi(p, rt, 0, imm);
+}
+
+/** rt = imm << 16 */
+void
+ppc_lis(struct ppc_function *p, uint rt, int imm)
+{
+   ppc_addis(p, rt, 0, imm);
+}
+
+/** rt = imm */
+void
+ppc_load_int(struct ppc_function *p, uint rt, int imm)
+{
+   ppc_lis(p, rt, (imm >> 16));          /* rt = imm >> 16 */
+   ppc_ori(p, rt, rt, (imm & 0xffff));   /* rt = rt | (imm & 0xffff) */
+}
+
+
+
+
+/**
+ ** integer load/store
+ **/
+
+/** store rs at memory[(ra)+d],
+ * then update ra = (ra)+d
+ */
+void
+ppc_stwu(struct ppc_function *p, uint rs, uint ra, int d)
+{
+   emit_d(p, 37, rs, ra, d);
+}
+
+/** store rs at memory[(ra)+d] */
+void
+ppc_stw(struct ppc_function *p, uint rs, uint ra, int d)
+{
+   emit_d(p, 36, rs, ra, d);
+}
+
+/** Load rt = mem[(ra)+d];  then zero set high 32 bits to zero. */
+void
+ppc_lwz(struct ppc_function *p, uint rt, uint ra, int d)
+{
+   emit_d(p, 32, rt, ra, d);
+}
+
+
+
+/**
+ ** Float (non-vector) arithmetic
+ **/
+
+/** add: frt = fra + frb */
+void
+ppc_fadd(struct ppc_function *p, uint frt, uint fra, uint frb)
+{
+   emit_a(p, 63, frt, fra, frb, 21, 0);
+}
+
+/** sub: frt = fra - frb */
+void
+ppc_fsub(struct ppc_function *p, uint frt, uint fra, uint frb)
+{
+   emit_a(p, 63, frt, fra, frb, 20, 0);
+}
+
+/** convert to int: rt = (int) ra */
+void
+ppc_fctiwz(struct ppc_function *p, uint rt, uint fra)
+{
+   emit_x(p, 63, rt, 0, fra, 15);
+}
+
+/** store frs at mem[(ra)+offset] */
+void
+ppc_stfs(struct ppc_function *p, uint frs, uint ra, int offset)
+{
+   emit_d(p, 52, frs, ra, offset);
+}
+
+/** store frs at mem[(ra)+(rb)] */
+void
+ppc_stfiwx(struct ppc_function *p, uint frs, uint ra, uint rb)
+{
+   emit_x(p, 31, frs, ra, rb, 983);
+}
+
+/** load frt = mem[(ra)+offset] */
+void
+ppc_lfs(struct ppc_function *p, uint frt, uint ra, int offset)
+{
+   emit_d(p, 48, frt, ra, offset);
+}
+
+
+
+
+
+/**
+ ** branch instructions
+ **/
+
+/** BLR: Branch to link register (p. 35) */
+void
+ppc_blr(struct ppc_function *p)
+{
+   emit_i(p, 18, 0, 0, 1);
+}
+
+/** Branch Conditional to Link Register (p. 36) */
+void
+ppc_bclr(struct ppc_function *p, uint condOp, uint branchHint, uint condReg)
+{
+   emit_xl(p, 19, condOp, condReg, branchHint, 16, 0);
+}
+
+/** Pseudo instruction: return from subroutine */
+void
+ppc_return(struct ppc_function *p)
+{
+   ppc_bclr(p, BRANCH_COND_ALWAYS, BRANCH_HINT_SUB_RETURN, 0);
+}
diff --git a/src/gallium/auxiliary/rtasm/rtasm_ppc.h b/src/gallium/auxiliary/rtasm/rtasm_ppc.h
new file mode 100644
index 00000000000..08212a2a253
--- /dev/null
+++ b/src/gallium/auxiliary/rtasm/rtasm_ppc.h
@@ -0,0 +1,335 @@
+/**************************************************************************
+ *
+ * Copyright (C) 2008 Tungsten Graphics, Inc.   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * PPC code generation.
+ * \author Brian Paul
+ */
+
+
+#ifndef RTASM_PPC_H
+#define RTASM_PPC_H
+
+
+#include "pipe/p_compiler.h"
+
+
+#define PPC_INST_SIZE 4  /**< 4 bytes / instruction */
+
+#define PPC_NUM_REGS 32
+#define PPC_NUM_FP_REGS 32
+#define PPC_NUM_VEC_REGS 32
+
+/** Stack pointer register */
+#define PPC_REG_SP 1
+
+/** Branch conditions */
+#define BRANCH_COND_ALWAYS       0x14  /* binary 1z1zz (z=ignored) */
+
+/** Branch hints */
+#define BRANCH_HINT_SUB_RETURN   0x0   /* binary 00 */
+
+
+struct ppc_function
+{
+   uint32_t *store;  /**< instruction buffer */
+   uint num_inst;
+   uint max_inst;
+   uint32_t reg_used;   /** used/free general-purpose registers bitmask */
+   uint32_t fp_used;   /** used/free floating point registers bitmask */
+   uint32_t vec_used;   /** used/free vector registers bitmask */
+};
+
+
+
+extern void ppc_init_func(struct ppc_function *p);
+extern void ppc_release_func(struct ppc_function *p);
+extern uint ppc_num_instructions(const struct ppc_function *p);
+extern void (*ppc_get_func( struct ppc_function *p ))( void );
+extern void ppc_dump_func(const struct ppc_function *p);
+
+extern int ppc_reserve_register(struct ppc_function *p, int reg);
+extern int ppc_allocate_register(struct ppc_function *p);
+extern void ppc_release_register(struct ppc_function *p, int reg);
+extern int ppc_allocate_fp_register(struct ppc_function *p);
+extern void ppc_release_fp_register(struct ppc_function *p, int reg);
+extern int ppc_allocate_vec_register(struct ppc_function *p);
+extern void ppc_release_vec_register(struct ppc_function *p, int reg);
+
+
+
+/**
+ ** float vector arithmetic
+ **/
+
+/** vector float add */
+extern void
+ppc_vaddfp(struct ppc_function *p,uint vD, uint vA, uint vB);
+
+/** vector float substract */
+extern void
+ppc_vsubfp(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float min */
+extern void
+ppc_vminfp(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float max */
+extern void
+ppc_vmaxfp(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float mult add: vD = vA * vB + vC */
+extern void
+ppc_vmaddfp(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC);
+
+/** vector float negative mult subtract: vD = vA - vB * vC */
+extern void
+ppc_vnmsubfp(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC);
+
+/** vector float compare greater than */
+extern void
+ppc_vcmpgtfpx(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float compare greater than or equal to */
+extern void
+ppc_vcmpgefpx(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float compare equal */
+extern void
+ppc_vcmpeqfpx(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector float 2^x */
+extern void
+ppc_vexptefp(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float log2(x) */
+extern void
+ppc_vlogefp(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float reciprocol */
+extern void
+ppc_vrefp(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float reciprocol sqrt estimate */
+extern void
+ppc_vrsqrtefp(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float round to negative infinity */
+extern void
+ppc_vrfim(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float round to positive infinity */
+extern void
+ppc_vrfip(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float round to nearest int */
+extern void
+ppc_vrfin(struct ppc_function *p, uint vD, uint vB);
+
+/** vector float round to int toward zero */
+extern void
+ppc_vrfiz(struct ppc_function *p, uint vD, uint vB);
+
+
+/** vector store: store vR at mem[vA+vB] */
+extern void
+ppc_stvx(struct ppc_function *p, uint vR, uint vA, uint vB);
+
+/** vector load: vR = mem[vA+vB] */
+extern void
+ppc_lvx(struct ppc_function *p, uint vR, uint vA, uint vB);
+
+/** load vector element word: vR = mem_word[vA+vB] */
+extern void
+ppc_lvewx(struct ppc_function *p, uint vR, uint vA, uint vB);
+
+
+
+/**
+ ** vector bitwise operations
+ **/
+
+
+/** vector and */
+extern void
+ppc_vand(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector and complement */
+extern void
+ppc_vandc(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector or */
+extern void
+ppc_vor(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector nor */
+extern void
+ppc_vnor(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** vector xor */
+extern void
+ppc_vxor(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+/** Pseudo-instruction: vector move */
+extern void
+ppc_vmove(struct ppc_function *p, uint vD, uint vA);
+
+/** Set vector register to {0,0,0,0} */
+extern void
+ppc_vzero(struct ppc_function *p, uint vr);
+
+
+
+/**
+ ** Vector shuffle / select / splat / etc
+ **/
+
+/** vector permute */
+extern void
+ppc_vperm(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC);
+
+/** vector select */
+extern void
+ppc_vsel(struct ppc_function *p, uint vD, uint vA, uint vB, uint vC);
+
+/** vector splat byte */
+extern void
+ppc_vspltb(struct ppc_function *p, uint vD, uint vB, uint imm);
+
+/** vector splat half word */
+extern void
+ppc_vsplthw(struct ppc_function *p, uint vD, uint vB, uint imm);
+
+/** vector splat word */
+extern void
+ppc_vspltw(struct ppc_function *p, uint vD, uint vB, uint imm);
+
+/** vector splat signed immediate word */
+extern void
+ppc_vspltisw(struct ppc_function *p, uint vD, int imm);
+
+/** vector shift left word: vD[word] = vA[word] << (vB[word] & 0x1f) */
+extern void
+ppc_vslw(struct ppc_function *p, uint vD, uint vA, uint vB);
+
+
+
+/**
+ ** scalar arithmetic
+ **/
+
+extern void
+ppc_add(struct ppc_function *p, uint rt, uint ra, uint rb);
+
+extern void
+ppc_addi(struct ppc_function *p, uint rt, uint ra, int imm);
+
+extern void
+ppc_addis(struct ppc_function *p, uint rt, uint ra, int imm);
+
+extern void
+ppc_and(struct ppc_function *p, uint rt, uint ra, uint rb);
+
+extern void
+ppc_andi(struct ppc_function *p, uint rt, uint ra, int imm);
+
+extern void
+ppc_or(struct ppc_function *p, uint rt, uint ra, uint rb);
+
+extern void
+ppc_ori(struct ppc_function *p, uint rt, uint ra, int imm);
+
+extern void
+ppc_xor(struct ppc_function *p, uint rt, uint ra, uint rb);
+
+extern void
+ppc_xori(struct ppc_function *p, uint rt, uint ra, int imm);
+
+extern void
+ppc_mr(struct ppc_function *p, uint rt, uint ra);
+
+extern void
+ppc_li(struct ppc_function *p, uint rt, int imm);
+
+extern void
+ppc_lis(struct ppc_function *p, uint rt, int imm);
+
+extern void
+ppc_load_int(struct ppc_function *p, uint rt, int imm);
+
+
+
+/**
+ ** scalar load/store
+ **/
+
+extern void
+ppc_stwu(struct ppc_function *p, uint rs, uint ra, int d);
+
+extern void
+ppc_stw(struct ppc_function *p, uint rs, uint ra, int d);
+
+extern void
+ppc_lwz(struct ppc_function *p, uint rs, uint ra, int d);
+
+
+
+/**
+ ** Float (non-vector) arithmetic
+ **/
+
+extern void
+ppc_fadd(struct ppc_function *p, uint frt, uint fra, uint frb);
+
+extern void
+ppc_fsub(struct ppc_function *p, uint frt, uint fra, uint frb);
+
+extern void
+ppc_fctiwz(struct ppc_function *p, uint rt, uint ra);
+
+extern void
+ppc_stfs(struct ppc_function *p, uint frs, uint ra, int offset);
+
+extern void
+ppc_stfiwx(struct ppc_function *p, uint frs, uint ra, uint rb);
+
+extern void
+ppc_lfs(struct ppc_function *p, uint frt, uint ra, int offset);
+
+
+
+/**
+ ** branch instructions
+ **/
+
+extern void
+ppc_blr(struct ppc_function *p);
+
+void
+ppc_bclr(struct ppc_function *p, uint condOp, uint branchHint, uint condReg);
+
+extern void
+ppc_return(struct ppc_function *p);
+
+
+#endif /* RTASM_PPC_H */
diff --git a/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.c b/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.c
index a04cc6c4ff7..b9a75ae5591 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.c
@@ -27,12 +27,16 @@
  * Real-time assembly generation interface for Cell B.E. SPEs.
  *
  * \author Ian Romanick <idr@us.ibm.com>
+ * \author Brian Paul
  */
 
+
+#include <stdio.h>
 #include "pipe/p_compiler.h"
 #include "util/u_memory.h"
 #include "rtasm_ppc_spe.h"
 
+
 #ifdef GALLIUM_CELL
 /**
  * SPE instruction types
@@ -143,21 +147,91 @@ union spe_inst_RI18 {
 /*@}*/
 
 
+static void
+indent(const struct spe_function *p)
+{
+   int i;
+   for (i = 0; i < p->indent; i++) {
+      putchar(' ');
+   }
+}
+
+
+static const char *
+rem_prefix(const char *longname)
+{
+   return longname + 4;
+}
+
+
+static const char *
+reg_name(int reg)
+{
+   switch (reg) {
+   case SPE_REG_SP:
+      return "$sp";
+   case SPE_REG_RA:
+      return "$lr";
+   default:
+      {
+         /* cycle through four buffers to handle multiple calls per printf */
+         static char buf[4][10];
+         static int b = 0;
+         b = (b + 1) % 4;
+         sprintf(buf[b], "$%d", reg);
+         return buf[b];
+      }
+   }
+}
+
+
+static void
+emit_instruction(struct spe_function *p, uint32_t inst_bits)
+{
+   if (!p->store)
+      return;  /* out of memory, drop the instruction */
+
+   if (p->num_inst == p->max_inst) {
+      /* allocate larger buffer */
+      uint32_t *newbuf;
+      p->max_inst *= 2;  /* 2x larger */
+      newbuf = align_malloc(p->max_inst * SPE_INST_SIZE, 16);
+      if (newbuf) {
+         memcpy(newbuf, p->store, p->num_inst * SPE_INST_SIZE);
+      }
+      align_free(p->store);
+      p->store = newbuf;
+      if (!p->store) {
+         /* out of memory */
+         p->num_inst = 0;
+         return;
+      }
+   }
+
+   p->store[p->num_inst++] = inst_bits;
+}
+
+
+
 static void emit_RR(struct spe_function *p, unsigned op, unsigned rT,
-		    unsigned rA, unsigned rB)
+		    unsigned rA, unsigned rB, const char *name)
 {
     union spe_inst_RR inst;
     inst.inst.op = op;
     inst.inst.rB = rB;
     inst.inst.rA = rA;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, %s, %s\n",
+              rem_prefix(name), reg_name(rT), reg_name(rA), reg_name(rB));
+    }
 }
 
 
 static void emit_RRR(struct spe_function *p, unsigned op, unsigned rT,
-		    unsigned rA, unsigned rB, unsigned rC)
+                     unsigned rA, unsigned rB, unsigned rC, const char *name)
 {
     union spe_inst_RRR inst;
     inst.inst.op = op;
@@ -165,155 +239,212 @@ static void emit_RRR(struct spe_function *p, unsigned op, unsigned rT,
     inst.inst.rB = rB;
     inst.inst.rA = rA;
     inst.inst.rC = rC;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, %s, %s, %s\n", rem_prefix(name), reg_name(rT),
+              reg_name(rA), reg_name(rB), reg_name(rC));
+    }
 }
 
 
 static void emit_RI7(struct spe_function *p, unsigned op, unsigned rT,
-		     unsigned rA, int imm)
+		     unsigned rA, int imm, const char *name)
 {
     union spe_inst_RI7 inst;
     inst.inst.op = op;
     inst.inst.i7 = imm;
     inst.inst.rA = rA;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, %s, 0x%x\n",
+              rem_prefix(name), reg_name(rT), reg_name(rA), imm);
+    }
 }
 
 
 
 static void emit_RI8(struct spe_function *p, unsigned op, unsigned rT,
-		     unsigned rA, int imm)
+		     unsigned rA, int imm, const char *name)
 {
     union spe_inst_RI8 inst;
     inst.inst.op = op;
     inst.inst.i8 = imm;
     inst.inst.rA = rA;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, %s, 0x%x\n",
+              rem_prefix(name), reg_name(rT), reg_name(rA), imm);
+    }
 }
 
 
 
 static void emit_RI10(struct spe_function *p, unsigned op, unsigned rT,
-		      unsigned rA, int imm)
+		      unsigned rA, int imm, const char *name)
 {
     union spe_inst_RI10 inst;
     inst.inst.op = op;
     inst.inst.i10 = imm;
     inst.inst.rA = rA;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, %s, 0x%x\n",
+              rem_prefix(name), reg_name(rT), reg_name(rA), imm);
+    }
+}
+
+
+/** As above, but do range checking on signed immediate value */
+static void emit_RI10s(struct spe_function *p, unsigned op, unsigned rT,
+                       unsigned rA, int imm, const char *name)
+{
+    assert(imm <= 511);
+    assert(imm >= -512);
+    emit_RI10(p, op, rT, rA, imm, name);
 }
 
 
 static void emit_RI16(struct spe_function *p, unsigned op, unsigned rT,
-		      int imm)
+		      int imm, const char *name)
 {
     union spe_inst_RI16 inst;
     inst.inst.op = op;
     inst.inst.i16 = imm;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, 0x%x\n", rem_prefix(name), reg_name(rT), imm);
+    }
 }
 
 
 static void emit_RI18(struct spe_function *p, unsigned op, unsigned rT,
-		      int imm)
+		      int imm, const char *name)
 {
     union spe_inst_RI18 inst;
     inst.inst.op = op;
     inst.inst.i18 = imm;
     inst.inst.rT = rT;
-    p->store[p->num_inst++] = inst.bits;
-    assert(p->num_inst <= p->max_inst);
+    emit_instruction(p, inst.bits);
+    if (p->print) {
+       indent(p);
+       printf("%s\t%s, 0x%x\n", rem_prefix(name), reg_name(rT), imm);
+    }
 }
 
 
-
+#define EMIT(_name, _op) \
+void _name (struct spe_function *p) \
+{ \
+   emit_RR(p, _op, 0, 0, 0, __FUNCTION__); \
+}
 
 #define EMIT_(_name, _op) \
 void _name (struct spe_function *p, unsigned rT) \
 { \
-    emit_RR(p, _op, rT, 0, 0); \
+   emit_RR(p, _op, rT, 0, 0, __FUNCTION__); \
 }
 
 #define EMIT_R(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA) \
 { \
-    emit_RR(p, _op, rT, rA, 0); \
+   emit_RR(p, _op, rT, rA, 0, __FUNCTION__);                 \
 }
 
 #define EMIT_RR(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA, unsigned rB) \
 { \
-    emit_RR(p, _op, rT, rA, rB); \
+   emit_RR(p, _op, rT, rA, rB, __FUNCTION__);                \
 }
 
 #define EMIT_RRR(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA, unsigned rB, unsigned rC) \
 { \
-    emit_RRR(p, _op, rT, rA, rB, rC); \
+   emit_RRR(p, _op, rT, rA, rB, rC, __FUNCTION__);           \
 }
 
 #define EMIT_RI7(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA, int imm) \
 { \
-    emit_RI7(p, _op, rT, rA, imm); \
+   emit_RI7(p, _op, rT, rA, imm, __FUNCTION__);              \
 }
 
 #define EMIT_RI8(_name, _op, bias) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA, int imm) \
 { \
-    emit_RI8(p, _op, rT, rA, bias - imm); \
+   emit_RI8(p, _op, rT, rA, bias - imm, __FUNCTION__);       \
 }
 
 #define EMIT_RI10(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, unsigned rA, int imm) \
 { \
-    emit_RI10(p, _op, rT, rA, imm); \
+   emit_RI10(p, _op, rT, rA, imm, __FUNCTION__);             \
+}
+
+#define EMIT_RI10s(_name, _op) \
+void _name (struct spe_function *p, unsigned rT, unsigned rA, int imm) \
+{ \
+   emit_RI10s(p, _op, rT, rA, imm, __FUNCTION__);             \
 }
 
 #define EMIT_RI16(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, int imm) \
 { \
-    emit_RI16(p, _op, rT, imm); \
+   emit_RI16(p, _op, rT, imm, __FUNCTION__);                 \
 }
 
 #define EMIT_RI18(_name, _op) \
 void _name (struct spe_function *p, unsigned rT, int imm) \
 { \
-    emit_RI18(p, _op, rT, imm); \
+   emit_RI18(p, _op, rT, imm, __FUNCTION__);                 \
 }
 
 #define EMIT_I16(_name, _op) \
 void _name (struct spe_function *p, int imm) \
 { \
-    emit_RI16(p, _op, 0, imm); \
+   emit_RI16(p, _op, 0, imm, __FUNCTION__);                  \
 }
 
 #include "rtasm_ppc_spe.h"
 
 
+
 /**
  * Initialize an spe_function.
- * \param code_size  size of instruction buffer to allocate, in bytes.
+ * \param code_size  initial size of instruction buffer to allocate, in bytes.
+ *                   If zero, use a default.
  */
 void spe_init_func(struct spe_function *p, unsigned code_size)
 {
-    p->store = align_malloc(code_size, 16);
+    unsigned int i;
+
+    if (!code_size)
+       code_size = 64;
+
     p->num_inst = 0;
     p->max_inst = code_size / SPE_INST_SIZE;
+    p->store = align_malloc(code_size, 16);
+
+    p->set_count = 0;
+    memset(p->regs, 0, SPE_NUM_REGS * sizeof(p->regs[0]));
 
     /* Conservatively treat R0 - R2 and R80 - R127 as non-volatile.
      */
-    p->regs[0] = ~7;
-    p->regs[1] = (1U << (80 - 64)) - 1;
+    p->regs[0] = p->regs[1] = p->regs[2] = 1;
+    for (i = 80; i <= 127; i++) {
+      p->regs[i] = 1;
+    }
+
+    p->print = false;
+    p->indent = 0;
 }
 
 
@@ -327,20 +458,23 @@ void spe_release_func(struct spe_function *p)
 }
 
 
+/** Return current code size in bytes. */
+unsigned spe_code_size(const struct spe_function *p)
+{
+   return p->num_inst * SPE_INST_SIZE;
+}
+
+
 /**
- * Alloate a SPE register.
+ * Allocate a SPE register.
  * \return register index or -1 if none left.
  */
 int spe_allocate_available_register(struct spe_function *p)
 {
    unsigned i;
    for (i = 0; i < SPE_NUM_REGS; i++) {
-      const uint64_t mask = (1ULL << (i % 64));
-      const unsigned idx = i / 64;
-
-      assert(idx < 2);
-      if ((p->regs[idx] & mask) != 0) {
-         p->regs[idx] &= ~mask;
+      if (p->regs[i] == 0) {
+         p->regs[i] = 1;
          return i;
       }
    }
@@ -354,31 +488,160 @@ int spe_allocate_available_register(struct spe_function *p)
  */
 int spe_allocate_register(struct spe_function *p, int reg)
 {
-   const unsigned idx = reg / 64;
-   const unsigned bit = reg % 64;
-
    assert(reg < SPE_NUM_REGS);
-   assert((p->regs[idx] & (1ULL << bit)) != 0);
-
-   p->regs[idx] &= ~(1ULL << bit);
+   assert(p->regs[reg] == 0);
+   p->regs[reg] = 1;
    return reg;
 }
 
 
 /**
- * Mark the given SPE register as "unallocated".
+ * Mark the given SPE register as "unallocated".  Note that this should
+ * only be used on registers allocated in the current register set; an
+ * assertion will fail if an attempt is made to deallocate a register
+ * allocated in an earlier register set.
  */
 void spe_release_register(struct spe_function *p, int reg)
 {
-   const unsigned idx = reg / 64;
-   const unsigned bit = reg % 64;
+   assert(reg < SPE_NUM_REGS);
+   assert(p->regs[reg] == 1);
 
-   assert(idx < 2);
+   p->regs[reg] = 0;
+}
+
+/**
+ * Start a new set of registers.  This can be called if
+ * it will be difficult later to determine exactly what
+ * registers were actually allocated during a code generation
+ * sequence, and you really just want to deallocate all of them.
+ */
+void spe_allocate_register_set(struct spe_function *p)
+{
+   unsigned int i;
+
+   /* Keep track of the set count.  If it ever wraps around to 0, 
+    * we're in trouble.
+    */
+   p->set_count++;
+   assert(p->set_count > 0);
+
+   /* Increment the allocation count of all registers currently
+    * allocated.  Then any registers that are allocated in this set
+    * will be the only ones with a count of 1; they'll all be released
+    * when the register set is released.
+    */
+   for (i = 0; i < SPE_NUM_REGS; i++) {
+      if (p->regs[i] > 0)
+         p->regs[i]++;
+   }
+}
+
+void spe_release_register_set(struct spe_function *p)
+{
+   unsigned int i;
+
+   /* If the set count drops below zero, we're in trouble. */
+   assert(p->set_count > 0);
+   p->set_count--;
+
+   /* Drop the allocation level of all registers.  Any allocated
+    * during this register set will drop to 0 and then become
+    * available.
+    */
+   for (i = 0; i < SPE_NUM_REGS; i++) {
+      if (p->regs[i] > 0)
+         p->regs[i]--;
+   }
+}
+
+
+unsigned
+spe_get_registers_used(const struct spe_function *p, ubyte used[])
+{
+   unsigned i, num = 0;
+   /* only count registers in the range available to callers */
+   for (i = 2; i < 80; i++) {
+      if (p->regs[i]) {
+         used[num++] = i;
+      }
+   }
+   return num;
+}
+
+
+void
+spe_print_code(struct spe_function *p, boolean enable)
+{
+   p->print = enable;
+}
 
-   assert(reg < SPE_NUM_REGS);
-   assert((p->regs[idx] & (1ULL << bit)) == 0);
 
-   p->regs[idx] |= (1ULL << bit);
+void
+spe_indent(struct spe_function *p, int spaces)
+{
+   p->indent += spaces;
+}
+
+
+void
+spe_comment(struct spe_function *p, int rel_indent, const char *s)
+{
+   if (p->print) {
+      p->indent += rel_indent;
+      indent(p);
+      p->indent -= rel_indent;
+      printf("# %s\n", s);
+   }
+}
+
+
+/**
+ * Load quad word.
+ * NOTE: offset is in bytes and the least significant 4 bits must be zero!
+ */
+void spe_lqd(struct spe_function *p, unsigned rT, unsigned rA, int offset)
+{
+   const boolean pSave = p->print;
+
+   /* offset must be a multiple of 16 */
+   assert(offset % 16 == 0);
+   /* offset must fit in 10-bit signed int field, after shifting */
+   assert((offset >> 4) <= 511);
+   assert((offset >> 4) >= -512);
+
+   p->print = FALSE;
+   emit_RI10(p, 0x034, rT, rA, offset >> 4, "spe_lqd");
+   p->print = pSave;
+
+   if (p->print) {
+      indent(p);
+      printf("lqd\t%s, %d(%s)\n", reg_name(rT), offset, reg_name(rA));
+   }
+}
+
+
+/**
+ * Store quad word.
+ * NOTE: offset is in bytes and the least significant 4 bits must be zero!
+ */
+void spe_stqd(struct spe_function *p, unsigned rT, unsigned rA, int offset)
+{
+   const boolean pSave = p->print;
+
+   /* offset must be a multiple of 16 */
+   assert(offset % 16 == 0);
+   /* offset must fit in 10-bit signed int field, after shifting */
+   assert((offset >> 4) <= 511);
+   assert((offset >> 4) >= -512);
+
+   p->print = FALSE;
+   emit_RI10(p, 0x024, rT, rA, offset >> 4, "spe_stqd");
+   p->print = pSave;
+
+   if (p->print) {
+      indent(p);
+      printf("stqd\t%s, %d(%s)\n", reg_name(rT), offset, reg_name(rA));
+   }
 }
 
 
@@ -392,51 +655,51 @@ void spe_release_register(struct spe_function *p, int reg)
 /** Branch Indirect to address in rA */
 void spe_bi(struct spe_function *p, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x1a8, 0, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x1a8, 0, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Interupt Return */
 void spe_iret(struct spe_function *p, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x1aa, 0, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x1aa, 0, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect and set link on external data */
 void spe_bisled(struct spe_function *p, unsigned rT, unsigned rA, int d,
 		int e)
 {
-    emit_RI7(p, 0x1ab, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x1ab, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect and set link.  Save PC in rT, jump to rA. */
 void spe_bisl(struct spe_function *p, unsigned rT, unsigned rA, int d,
 		int e)
 {
-    emit_RI7(p, 0x1a9, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x1a9, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect if zero word.  If rT.word[0]==0, jump to rA. */
 void spe_biz(struct spe_function *p, unsigned rT, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x128, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x128, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect if non-zero word.  If rT.word[0]!=0, jump to rA. */
 void spe_binz(struct spe_function *p, unsigned rT, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x129, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x129, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect if zero halfword.  If rT.halfword[1]==0, jump to rA. */
 void spe_bihz(struct spe_function *p, unsigned rT, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x12a, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x12a, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 /** Branch indirect if non-zero halfword.  If rT.halfword[1]!=0, jump to rA. */
 void spe_bihnz(struct spe_function *p, unsigned rT, unsigned rA, int d, int e)
 {
-    emit_RI7(p, 0x12b, rT, rA, (d << 5) | (e << 4));
+   emit_RI7(p, 0x12b, rT, rA, (d << 5) | (e << 4), __FUNCTION__);
 }
 
 
@@ -454,7 +717,6 @@ hbrr;
 #if 0
 stop;
 EMIT_RR  (spe_stopd, 0x140);
-EMIT_    (spe_lnop,  0x001);
 EMIT_    (spe_nop,   0x201);
 sync;
 EMIT_    (spe_dsync, 0x003);
@@ -505,30 +767,226 @@ spe_load_int(struct spe_function *p, unsigned rT, int i)
    }
    else {
       spe_ilhu(p, rT, i >> 16);
-      spe_iohl(p, rT, i & 0xffff);
+      if (i & 0xffff)
+         spe_iohl(p, rT, i & 0xffff);
+   }
+}
+
+void spe_load_uint(struct spe_function *p, unsigned rT, unsigned int ui)
+{
+   /* If the whole value is in the lower 18 bits, use ila, which
+    * doesn't sign-extend.  Otherwise, if the two halfwords of
+    * the constant are identical, use ilh.  Otherwise, if every byte of
+    * the desired value is 0x00 or 0xff, we can use Form Select Mask for
+    * Bytes Immediate (fsmbi) to load the value in a single instruction.
+    * Otherwise, in the general case, we have to use ilhu followed by iohl.
+    */
+   if ((ui & 0x0003ffff) == ui) {
+      spe_ila(p, rT, ui);
+   }
+   else if ((ui >> 16) == (ui & 0xffff)) {
+      spe_ilh(p, rT, ui & 0xffff);
+   }
+   else if (
+      ((ui & 0x000000ff) == 0 || (ui & 0x000000ff) == 0x000000ff) &&
+      ((ui & 0x0000ff00) == 0 || (ui & 0x0000ff00) == 0x0000ff00) &&
+      ((ui & 0x00ff0000) == 0 || (ui & 0x00ff0000) == 0x00ff0000) &&
+      ((ui & 0xff000000) == 0 || (ui & 0xff000000) == 0xff000000)
+   ) {
+      unsigned int mask = 0;
+      /* fsmbi duplicates each bit in the given mask eight times,
+       * using a 16-bit value to initialize a 16-byte quadword.
+       * Each 4-bit nybble of the mask corresponds to a full word
+       * of the result; look at the value and figure out the mask
+       * (replicated for each word in the quadword), and then
+       * form the "select mask" to get the value.
+       */
+      if ((ui & 0x000000ff) == 0x000000ff) mask |= 0x1111;
+      if ((ui & 0x0000ff00) == 0x0000ff00) mask |= 0x2222;
+      if ((ui & 0x00ff0000) == 0x00ff0000) mask |= 0x4444;
+      if ((ui & 0xff000000) == 0xff000000) mask |= 0x8888;
+      spe_fsmbi(p, rT, mask);
+   }
+   else {
+      /* The general case: this usually uses two instructions, but
+       * may use only one if the low-order 16 bits of each word are 0.
+       */
+      spe_ilhu(p, rT, ui >> 16);
+      if (ui & 0xffff)
+         spe_iohl(p, rT, ui & 0xffff);
+   }
+}
+
+/**
+ * This function is constructed identically to spe_xor_uint() below.
+ * Changes to one should be made in the other.
+ */
+void
+spe_and_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui)
+{
+   /* If we can, emit a single instruction, either And Byte Immediate
+    * (which uses the same constant across each byte), And Halfword Immediate
+    * (which sign-extends a 10-bit immediate to 16 bits and uses that
+    * across each halfword), or And Word Immediate (which sign-extends
+    * a 10-bit immediate to 32 bits).
+    *
+    * Otherwise, we'll need to use a temporary register.
+    */
+   unsigned int tmp;
+
+   /* If the upper 23 bits are all 0s or all 1s, sign extension
+    * will work and we can use And Word Immediate
+    */
+   tmp = ui & 0xfffffe00;
+   if (tmp == 0xfffffe00 || tmp  == 0) {
+      spe_andi(p, rT, rA, ui & 0x000003ff);
+      return;
+   }
+   
+   /* If the ui field is symmetric along halfword boundaries and
+    * the upper 7 bits of each halfword are all 0s or 1s, we
+    * can use And Halfword Immediate
+    */
+   tmp = ui & 0xfe00fe00;
+   if ((tmp == 0xfe00fe00 || tmp == 0) && ((ui >> 16) == (ui & 0x0000ffff))) {
+      spe_andhi(p, rT, rA, ui & 0x000003ff);
+      return;
+   }
+
+   /* If the ui field is symmetric in each byte, then we can use
+    * the And Byte Immediate instruction.
+    */
+   tmp = ui & 0x000000ff;
+   if ((ui >> 24) == tmp && ((ui >> 16) & 0xff) == tmp && ((ui >> 8) & 0xff) == tmp) {
+      spe_andbi(p, rT, rA, tmp);
+      return;
+   }
+
+   /* Otherwise, we'll have to use a temporary register. */
+   unsigned int tmp_reg = spe_allocate_available_register(p);
+   spe_load_uint(p, tmp_reg, ui);
+   spe_and(p, rT, rA, tmp_reg);
+   spe_release_register(p, tmp_reg);
+}
+
+
+/**
+ * This function is constructed identically to spe_and_uint() above.
+ * Changes to one should be made in the other.
+ */
+void
+spe_xor_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui)
+{
+   /* If we can, emit a single instruction, either Exclusive Or Byte 
+    * Immediate (which uses the same constant across each byte), Exclusive 
+    * Or Halfword Immediate (which sign-extends a 10-bit immediate to 
+    * 16 bits and uses that across each halfword), or Exclusive Or Word 
+    * Immediate (which sign-extends a 10-bit immediate to 32 bits).
+    *
+    * Otherwise, we'll need to use a temporary register.
+    */
+   unsigned int tmp;
+
+   /* If the upper 23 bits are all 0s or all 1s, sign extension
+    * will work and we can use Exclusive Or Word Immediate
+    */
+   tmp = ui & 0xfffffe00;
+   if (tmp == 0xfffffe00 || tmp  == 0) {
+      spe_xori(p, rT, rA, ui & 0x000003ff);
+      return;
+   }
+   
+   /* If the ui field is symmetric along halfword boundaries and
+    * the upper 7 bits of each halfword are all 0s or 1s, we
+    * can use Exclusive Or Halfword Immediate
+    */
+   tmp = ui & 0xfe00fe00;
+   if ((tmp == 0xfe00fe00 || tmp == 0) && ((ui >> 16) == (ui & 0x0000ffff))) {
+      spe_xorhi(p, rT, rA, ui & 0x000003ff);
+      return;
    }
+
+   /* If the ui field is symmetric in each byte, then we can use
+    * the Exclusive Or Byte Immediate instruction.
+    */
+   tmp = ui & 0x000000ff;
+   if ((ui >> 24) == tmp && ((ui >> 16) & 0xff) == tmp && ((ui >> 8) & 0xff) == tmp) {
+      spe_xorbi(p, rT, rA, tmp);
+      return;
+   }
+
+   /* Otherwise, we'll have to use a temporary register. */
+   unsigned int tmp_reg = spe_allocate_available_register(p);
+   spe_load_uint(p, tmp_reg, ui);
+   spe_xor(p, rT, rA, tmp_reg);
+   spe_release_register(p, tmp_reg);
 }
 
+void
+spe_compare_equal_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui)
+{
+   /* If the comparison value is 9 bits or less, it fits inside a
+    * Compare Equal Word Immediate instruction.
+    */
+   if ((ui & 0x000001ff) == ui) {
+      spe_ceqi(p, rT, rA, ui);
+   }
+   /* Otherwise, we're going to have to load a word first. */
+   else {
+      unsigned int tmp_reg = spe_allocate_available_register(p);
+      spe_load_uint(p, tmp_reg, ui);
+      spe_ceq(p, rT, rA, tmp_reg);
+      spe_release_register(p, tmp_reg);
+   }
+}
+
+void
+spe_compare_greater_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui)
+{
+   /* If the comparison value is 10 bits or less, it fits inside a
+    * Compare Logical Greater Than Word Immediate instruction.
+    */
+   if ((ui & 0x000003ff) == ui) {
+      spe_clgti(p, rT, rA, ui);
+   }
+   /* Otherwise, we're going to have to load a word first. */
+   else {
+      unsigned int tmp_reg = spe_allocate_available_register(p);
+      spe_load_uint(p, tmp_reg, ui);
+      spe_clgt(p, rT, rA, tmp_reg);
+      spe_release_register(p, tmp_reg);
+   }
+}
 
 void
 spe_splat(struct spe_function *p, unsigned rT, unsigned rA)
 {
-   spe_ila(p, rT, 66051);
-   spe_shufb(p, rT, rA, rA, rT);
+   /* Use a temporary, just in case rT == rA */
+   unsigned int tmp_reg = spe_allocate_available_register(p);
+   /* Duplicate bytes 0, 1, 2, and 3 across the whole register */
+   spe_ila(p, tmp_reg, 0x00010203);
+   spe_shufb(p, rT, rA, rA, tmp_reg);
+   spe_release_register(p, tmp_reg);
 }
 
 
 void
-spe_complement(struct spe_function *p, unsigned rT)
+spe_complement(struct spe_function *p, unsigned rT, unsigned rA)
 {
-   spe_nor(p, rT, rT, rT);
+   spe_nor(p, rT, rA, rA);
 }
 
 
 void
 spe_move(struct spe_function *p, unsigned rT, unsigned rA)
 {
-   spe_ori(p, rT, rA, 0);
+   /* Use different instructions depending on the instruction address
+    * to take advantage of the dual pipelines.
+    */
+   if (p->num_inst & 1)
+      spe_shlqbyi(p, rT, rA, 0);  /* odd pipe */
+   else
+      spe_ori(p, rT, rA, 0);  /* even pipe */
 }
 
 
@@ -539,4 +997,70 @@ spe_zero(struct spe_function *p, unsigned rT)
 }
 
 
+void
+spe_splat_word(struct spe_function *p, unsigned rT, unsigned rA, int word)
+{
+   assert(word >= 0);
+   assert(word <= 3);
+
+   if (word == 0) {
+      int tmp1 = rT;
+      spe_ila(p, tmp1, 66051);
+      spe_shufb(p, rT, rA, rA, tmp1);
+   }
+   else {
+      /* XXX review this, we may not need the rotqbyi instruction */
+      int tmp1 = rT;
+      int tmp2 = spe_allocate_available_register(p);
+
+      spe_ila(p, tmp1, 66051);
+      spe_rotqbyi(p, tmp2, rA, 4 * word);
+      spe_shufb(p, rT, tmp2, tmp2, tmp1);
+
+      spe_release_register(p, tmp2);
+   }
+}
+
+/**
+ * For each 32-bit float element of rA and rB, choose the smaller of the
+ * two, compositing them into the rT register.
+ * 
+ * The Float Compare Greater Than (fcgt) instruction will put 1s into
+ * compare_reg where rA > rB, and 0s where rA <= rB.
+ *
+ * Then the Select Bits (selb) instruction will take bits from rA where
+ * compare_reg is 0, and from rB where compare_reg is 1; i.e., from rA
+ * where rA <= rB and from rB where rB > rA, which is exactly the
+ * "min" operation.
+ *
+ * The compare_reg could in many cases be the same as rT, unless
+ * rT == rA || rt == rB.  But since this is common in constructions
+ * like "x = min(x, a)", we always allocate a new register to be safe.
+ */
+void 
+spe_float_min(struct spe_function *p, unsigned rT, unsigned rA, unsigned rB)
+{
+   unsigned int compare_reg = spe_allocate_available_register(p);
+   spe_fcgt(p, compare_reg, rA, rB);
+   spe_selb(p, rT, rA, rB, compare_reg);
+   spe_release_register(p, compare_reg);
+}
+
+/**
+ * For each 32-bit float element of rA and rB, choose the greater of the
+ * two, compositing them into the rT register.
+ * 
+ * The logic is similar to that of spe_float_min() above; the only
+ * difference is that the registers on spe_selb() have been reversed,
+ * so that the larger of the two is selected instead of the smaller.
+ */
+void 
+spe_float_max(struct spe_function *p, unsigned rT, unsigned rA, unsigned rB)
+{
+   unsigned int compare_reg = spe_allocate_available_register(p);
+   spe_fcgt(p, compare_reg, rA, rB);
+   spe_selb(p, rT, rB, rA, compare_reg);
+   spe_release_register(p, compare_reg);
+}
+
 #endif /* GALLIUM_CELL */
diff --git a/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.h b/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.h
index d95e5aace34..f9ad2acacdd 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.h
+++ b/src/gallium/auxiliary/rtasm/rtasm_ppc_spe.h
@@ -28,6 +28,7 @@
  * For details, see /opt/cell/sdk/docs/arch/SPU_ISA_v1.2_27Jan2007_pub.pdf
  *
  * \author Ian Romanick <idr@us.ibm.com>
+ * \author Brian Paul
  */
 
 #ifndef RTASM_PPC_SPE_H
@@ -39,10 +40,10 @@
 /** number of general-purpose SIMD registers */
 #define SPE_NUM_REGS  128
 
-/** Return Address register */
+/** Return Address register (aka $lr / Link Register) */
 #define SPE_REG_RA  0
 
-/** Stack Pointer register */
+/** Stack Pointer register (aka $sp) */
 #define SPE_REG_SP  1
 
 
@@ -52,221 +53,258 @@ struct spe_function
    uint num_inst;
    uint max_inst;
 
-    /**
-     * Mask of used / unused registers
-     *
-     * Each set bit corresponds to an available register.  Each cleared bit
-     * corresponds to an allocated register.
+   /**
+    * The "set count" reflects the number of nested register sets
+    * are allowed.  In the unlikely case that we exceed the set count,
+    * register allocation will start to be confused, which is critical
+    * enough that we check for it.
+    */
+   unsigned char set_count;
+
+   /** 
+    * Flags for used and unused registers.  Each byte corresponds to a
+    * register; a 0 in that byte means that the register is available.
+    * A value of 1 means that the register was allocated in the current
+    * register set.  Any other value N means that the register was allocated
+    * N register sets ago.
      *
      * \sa
      * spe_allocate_register, spe_allocate_available_register,
-     * spe_release_register
+     * spe_allocate_register_set, spe_release_register_set, spe_release_register, 
      */
-    uint64_t regs[SPE_NUM_REGS / 64];
+    unsigned char regs[SPE_NUM_REGS];
+
+    boolean print; /**< print/dump instructions as they're emitted? */
+    int indent;    /**< number of spaces to indent */
 };
 
+
 extern void spe_init_func(struct spe_function *p, unsigned code_size);
 extern void spe_release_func(struct spe_function *p);
+extern unsigned spe_code_size(const struct spe_function *p);
 
 extern int spe_allocate_available_register(struct spe_function *p);
 extern int spe_allocate_register(struct spe_function *p, int reg);
 extern void spe_release_register(struct spe_function *p, int reg);
+extern void spe_allocate_register_set(struct spe_function *p);
+extern void spe_release_register_set(struct spe_function *p);
+
+extern unsigned
+spe_get_registers_used(const struct spe_function *p, ubyte used[]);
+
+extern void spe_print_code(struct spe_function *p, boolean enable);
+extern void spe_indent(struct spe_function *p, int spaces);
+extern void spe_comment(struct spe_function *p, int rel_indent, const char *s);
+
 
 #endif /* RTASM_PPC_SPE_H */
 
-#ifndef EMIT_
-#define EMIT_(name, _op) \
-    extern void _name (struct spe_function *p, unsigned rT)
+#ifndef EMIT
+#define EMIT(_name, _op) \
+    extern void _name (struct spe_function *p);
+#define EMIT_(_name, _op) \
+    extern void _name (struct spe_function *p, unsigned rT);
 #define EMIT_R(_name, _op) \
-    extern void _name (struct spe_function *p, unsigned rT, unsigned rA)
+    extern void _name (struct spe_function *p, unsigned rT, unsigned rA);
 #define EMIT_RR(_name, _op) \
     extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
-			   unsigned rB)
+                       unsigned rB);
 #define EMIT_RRR(_name, _op) \
     extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
-			   unsigned rB, unsigned rC)
+                       unsigned rB, unsigned rC);
 #define EMIT_RI7(_name, _op) \
     extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
-			   int imm)
+                       int imm);
 #define EMIT_RI8(_name, _op, bias) \
     extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
-			   int imm)
+                       int imm);
 #define EMIT_RI10(_name, _op) \
     extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
-			   int imm)
+                       int imm);
+#define EMIT_RI10s(_name, _op) \
+    extern void _name (struct spe_function *p, unsigned rT, unsigned rA, \
+                       int imm);
 #define EMIT_RI16(_name, _op) \
-    extern void _name (struct spe_function *p, unsigned rT, int imm)
+    extern void _name (struct spe_function *p, unsigned rT, int imm);
 #define EMIT_RI18(_name, _op) \
-    extern void _name (struct spe_function *p, unsigned rT, int imm)
+    extern void _name (struct spe_function *p, unsigned rT, int imm);
 #define EMIT_I16(_name, _op) \
-    extern void _name (struct spe_function *p, int imm)
+    extern void _name (struct spe_function *p, int imm);
 #define UNDEF_EMIT_MACROS
-#endif /* EMIT_ */
+#endif /* EMIT */
 
 
 /* Memory load / store instructions
  */
-EMIT_RI10(spe_lqd,  0x034);
-EMIT_RR  (spe_lqx,  0x1c4);
-EMIT_RI16(spe_lqa,  0x061);
-EMIT_RI16(spe_lqr,  0x067);
-EMIT_RI10(spe_stqd, 0x024);
-EMIT_RR  (spe_stqx, 0x144);
-EMIT_RI16(spe_stqa, 0x041);
-EMIT_RI16(spe_stqr, 0x047);
-EMIT_RI7 (spe_cbd,  0x1f4);
-EMIT_RR  (spe_cbx,  0x1d4);
-EMIT_RI7 (spe_chd,  0x1f5);
-EMIT_RI7 (spe_chx,  0x1d5);
-EMIT_RI7 (spe_cwd,  0x1f6);
-EMIT_RI7 (spe_cwx,  0x1d6);
-EMIT_RI7 (spe_cdd,  0x1f7);
-EMIT_RI7 (spe_cdx,  0x1d7);
+EMIT_RR  (spe_lqx,  0x1c4)
+EMIT_RI16(spe_lqa,  0x061)
+EMIT_RI16(spe_lqr,  0x067)
+EMIT_RR  (spe_stqx, 0x144)
+EMIT_RI16(spe_stqa, 0x041)
+EMIT_RI16(spe_stqr, 0x047)
+EMIT_RI7 (spe_cbd,  0x1f4)
+EMIT_RR  (spe_cbx,  0x1d4)
+EMIT_RI7 (spe_chd,  0x1f5)
+EMIT_RI7 (spe_chx,  0x1d5)
+EMIT_RI7 (spe_cwd,  0x1f6)
+EMIT_RI7 (spe_cwx,  0x1d6)
+EMIT_RI7 (spe_cdd,  0x1f7)
+EMIT_RI7 (spe_cdx,  0x1d7)
 
 
 /* Constant formation instructions
  */
-EMIT_RI16(spe_ilh,   0x083);
-EMIT_RI16(spe_ilhu,  0x082);
-EMIT_RI16(spe_il,    0x081);
-EMIT_RI18(spe_ila,   0x021);
-EMIT_RI16(spe_iohl,  0x0c1);
-EMIT_RI16(spe_fsmbi, 0x065);
+EMIT_RI16(spe_ilh,   0x083)
+EMIT_RI16(spe_ilhu,  0x082)
+EMIT_RI16(spe_il,    0x081)
+EMIT_RI18(spe_ila,   0x021)
+EMIT_RI16(spe_iohl,  0x0c1)
+EMIT_RI16(spe_fsmbi, 0x065)
 
 
 
 /* Integer and logical instructions
  */
-EMIT_RR  (spe_ah,      0x0c8);
-EMIT_RI10(spe_ahi,     0x01d);
-EMIT_RR  (spe_a,       0x0c0);
-EMIT_RI10(spe_ai,      0x01c);
-EMIT_RR  (spe_sfh,     0x048);
-EMIT_RI10(spe_sfhi,    0x00d);
-EMIT_RR  (spe_sf,      0x040);
-EMIT_RI10(spe_sfi,     0x00c);
-EMIT_RR  (spe_addx,    0x340);
-EMIT_RR  (spe_cg,      0x0c2);
-EMIT_RR  (spe_cgx,     0x342);
-EMIT_RR  (spe_sfx,     0x341);
-EMIT_RR  (spe_bg,      0x042);
-EMIT_RR  (spe_bgx,     0x343);
-EMIT_RR  (spe_mpy,     0x3c4);
-EMIT_RR  (spe_mpyu,    0x3cc);
-EMIT_RI10(spe_mpyi,    0x074);
-EMIT_RI10(spe_mpyui,   0x075);
-EMIT_RRR (spe_mpya,    0x00c);
-EMIT_RR  (spe_mpyh,    0x3c5);
-EMIT_RR  (spe_mpys,    0x3c7);
-EMIT_RR  (spe_mpyhh,   0x3c6);
-EMIT_RR  (spe_mpyhha,  0x346);
-EMIT_RR  (spe_mpyhhu,  0x3ce);
-EMIT_RR  (spe_mpyhhau, 0x34e);
-EMIT_R   (spe_clz,     0x2a5);
-EMIT_R   (spe_cntb,    0x2b4);
-EMIT_R   (spe_fsmb,    0x1b6);
-EMIT_R   (spe_fsmh,    0x1b5);
-EMIT_R   (spe_fsm,     0x1b4);
-EMIT_R   (spe_gbb,     0x1b2);
-EMIT_R   (spe_gbh,     0x1b1);
-EMIT_R   (spe_gb,      0x1b0);
-EMIT_RR  (spe_avgb,    0x0d3);
-EMIT_RR  (spe_absdb,   0x053);
-EMIT_RR  (spe_sumb,    0x253);
-EMIT_R   (spe_xsbh,    0x2b6);
-EMIT_R   (spe_xshw,    0x2ae);
-EMIT_R   (spe_xswd,    0x2a6);
-EMIT_RR  (spe_and,     0x0c1);
-EMIT_RR  (spe_andc,    0x2c1);
-EMIT_RI10(spe_andbi,   0x016);
-EMIT_RI10(spe_andhi,   0x015);
-EMIT_RI10(spe_andi,    0x014);
-EMIT_RR  (spe_or,      0x041);
-EMIT_RR  (spe_orc,     0x2c9);
-EMIT_RI10(spe_orbi,    0x006);
-EMIT_RI10(spe_orhi,    0x005);
-EMIT_RI10(spe_ori,     0x004);
-EMIT_R   (spe_orx,     0x1f0);
-EMIT_RR  (spe_xor,     0x241);
-EMIT_RI10(spe_xorbi,   0x026);
-EMIT_RI10(spe_xorhi,   0x025);
-EMIT_RI10(spe_xori,    0x024);
-EMIT_RR  (spe_nand,    0x0c9);
-EMIT_RR  (spe_nor,     0x049);
-EMIT_RR  (spe_eqv,     0x249);
-EMIT_RRR (spe_selb,    0x008);
-EMIT_RRR (spe_shufb,   0x00b);
+EMIT_RR  (spe_ah,      0x0c8)
+EMIT_RI10(spe_ahi,     0x01d)
+EMIT_RR  (spe_a,       0x0c0)
+EMIT_RI10s(spe_ai,      0x01c)
+EMIT_RR  (spe_sfh,     0x048)
+EMIT_RI10(spe_sfhi,    0x00d)
+EMIT_RR  (spe_sf,      0x040)
+EMIT_RI10(spe_sfi,     0x00c)
+EMIT_RR  (spe_addx,    0x340)
+EMIT_RR  (spe_cg,      0x0c2)
+EMIT_RR  (spe_cgx,     0x342)
+EMIT_RR  (spe_sfx,     0x341)
+EMIT_RR  (spe_bg,      0x042)
+EMIT_RR  (spe_bgx,     0x343)
+EMIT_RR  (spe_mpy,     0x3c4)
+EMIT_RR  (spe_mpyu,    0x3cc)
+EMIT_RI10(spe_mpyi,    0x074)
+EMIT_RI10(spe_mpyui,   0x075)
+EMIT_RRR (spe_mpya,    0x00c)
+EMIT_RR  (spe_mpyh,    0x3c5)
+EMIT_RR  (spe_mpys,    0x3c7)
+EMIT_RR  (spe_mpyhh,   0x3c6)
+EMIT_RR  (spe_mpyhha,  0x346)
+EMIT_RR  (spe_mpyhhu,  0x3ce)
+EMIT_RR  (spe_mpyhhau, 0x34e)
+EMIT_R   (spe_clz,     0x2a5)
+EMIT_R   (spe_cntb,    0x2b4)
+EMIT_R   (spe_fsmb,    0x1b6)
+EMIT_R   (spe_fsmh,    0x1b5)
+EMIT_R   (spe_fsm,     0x1b4)
+EMIT_R   (spe_gbb,     0x1b2)
+EMIT_R   (spe_gbh,     0x1b1)
+EMIT_R   (spe_gb,      0x1b0)
+EMIT_RR  (spe_avgb,    0x0d3)
+EMIT_RR  (spe_absdb,   0x053)
+EMIT_RR  (spe_sumb,    0x253)
+EMIT_R   (spe_xsbh,    0x2b6)
+EMIT_R   (spe_xshw,    0x2ae)
+EMIT_R   (spe_xswd,    0x2a6)
+EMIT_RR  (spe_and,     0x0c1)
+EMIT_RR  (spe_andc,    0x2c1)
+EMIT_RI10s(spe_andbi,   0x016)
+EMIT_RI10s(spe_andhi,   0x015)
+EMIT_RI10s(spe_andi,    0x014)
+EMIT_RR  (spe_or,      0x041)
+EMIT_RR  (spe_orc,     0x2c9)
+EMIT_RI10s(spe_orbi,    0x006)
+EMIT_RI10s(spe_orhi,    0x005)
+EMIT_RI10s(spe_ori,     0x004)
+EMIT_R   (spe_orx,     0x1f0)
+EMIT_RR  (spe_xor,     0x241)
+EMIT_RI10s(spe_xorbi,   0x046)
+EMIT_RI10s(spe_xorhi,   0x045)
+EMIT_RI10s(spe_xori,    0x044)
+EMIT_RR  (spe_nand,    0x0c9)
+EMIT_RR  (spe_nor,     0x049)
+EMIT_RR  (spe_eqv,     0x249)
+EMIT_RRR (spe_selb,    0x008)
+EMIT_RRR (spe_shufb,   0x00b)
 
 
 /* Shift and rotate instructions
  */
-EMIT_RR  (spe_shlh,      0x05f);
-EMIT_RI7 (spe_shlhi,     0x07f);
-EMIT_RR  (spe_shl,       0x05b);
-EMIT_RI7 (spe_shli,      0x07b);
-EMIT_RR  (spe_shlqbi,    0x1db);
-EMIT_RI7 (spe_shlqbii,   0x1fb);
-EMIT_RR  (spe_shlqby,    0x1df);
-EMIT_RI7 (spe_shlqbyi,   0x1ff);
-EMIT_RR  (spe_shlqbybi,  0x1cf);
-EMIT_RR  (spe_roth,      0x05c);
-EMIT_RI7 (spe_rothi,     0x07c);
-EMIT_RR  (spe_rot,       0x058);
-EMIT_RI7 (spe_roti,      0x078);
-EMIT_RR  (spe_rotqby,    0x1dc);
-EMIT_RI7 (spe_rotqbyi,   0x1fc);
-EMIT_RR  (spe_rotqbybi,  0x1cc);
-EMIT_RR  (spe_rotqbi,    0x1d8);
-EMIT_RI7 (spe_rotqbii,   0x1f8);
-EMIT_RR  (spe_rothm,     0x05d);
-EMIT_RI7 (spe_rothmi,    0x07d);
-EMIT_RR  (spe_rotm,      0x059);
-EMIT_RI7 (spe_rotmi,     0x079);
-EMIT_RR  (spe_rotqmby,   0x1dd);
-EMIT_RI7 (spe_rotqmbyi,  0x1fd);
-EMIT_RR  (spe_rotqmbybi, 0x1cd);
-EMIT_RR  (spe_rotqmbi,   0x1c9);
-EMIT_RI7 (spe_rotqmbii,  0x1f9);
-EMIT_RR  (spe_rotmah,    0x05e);
-EMIT_RI7 (spe_rotmahi,   0x07e);
-EMIT_RR  (spe_rotma,     0x05a);
-EMIT_RI7 (spe_rotmai,    0x07a);
+EMIT_RR  (spe_shlh,      0x05f)
+EMIT_RI7 (spe_shlhi,     0x07f)
+EMIT_RR  (spe_shl,       0x05b)
+EMIT_RI7 (spe_shli,      0x07b)
+EMIT_RR  (spe_shlqbi,    0x1db)
+EMIT_RI7 (spe_shlqbii,   0x1fb)
+EMIT_RR  (spe_shlqby,    0x1df)
+EMIT_RI7 (spe_shlqbyi,   0x1ff)
+EMIT_RR  (spe_shlqbybi,  0x1cf)
+EMIT_RR  (spe_roth,      0x05c)
+EMIT_RI7 (spe_rothi,     0x07c)
+EMIT_RR  (spe_rot,       0x058)
+EMIT_RI7 (spe_roti,      0x078)
+EMIT_RR  (spe_rotqby,    0x1dc)
+EMIT_RI7 (spe_rotqbyi,   0x1fc)
+EMIT_RR  (spe_rotqbybi,  0x1cc)
+EMIT_RR  (spe_rotqbi,    0x1d8)
+EMIT_RI7 (spe_rotqbii,   0x1f8)
+EMIT_RR  (spe_rothm,     0x05d)
+EMIT_RI7 (spe_rothmi,    0x07d)
+EMIT_RR  (spe_rotm,      0x059)
+EMIT_RI7 (spe_rotmi,     0x079)
+EMIT_RR  (spe_rotqmby,   0x1dd)
+EMIT_RI7 (spe_rotqmbyi,  0x1fd)
+EMIT_RR  (spe_rotqmbybi, 0x1cd)
+EMIT_RR  (spe_rotqmbi,   0x1c9)
+EMIT_RI7 (spe_rotqmbii,  0x1f9)
+EMIT_RR  (spe_rotmah,    0x05e)
+EMIT_RI7 (spe_rotmahi,   0x07e)
+EMIT_RR  (spe_rotma,     0x05a)
+EMIT_RI7 (spe_rotmai,    0x07a)
 
 
 /* Compare, branch, and halt instructions
  */
-EMIT_RR  (spe_heq,       0x3d8);
-EMIT_RI10(spe_heqi,      0x07f);
-EMIT_RR  (spe_hgt,       0x258);
-EMIT_RI10(spe_hgti,      0x04f);
-EMIT_RR  (spe_hlgt,      0x2d8);
-EMIT_RI10(spe_hlgti,     0x05f);
-EMIT_RR  (spe_ceqb,      0x3d0);
-EMIT_RI10(spe_ceqbi,     0x07e);
-EMIT_RR  (spe_ceqh,      0x3c8);
-EMIT_RI10(spe_ceqhi,     0x07d);
-EMIT_RR  (spe_ceq,       0x3c0);
-EMIT_RI10(spe_ceqi,      0x07c);
-EMIT_RR  (spe_cgtb,      0x250);
-EMIT_RI10(spe_cgtbi,     0x04e);
-EMIT_RR  (spe_cgth,      0x248);
-EMIT_RI10(spe_cgthi,     0x04d);
-EMIT_RR  (spe_cgt,       0x240);
-EMIT_RI10(spe_cgti,      0x04c);
-EMIT_RR  (spe_clgtb,     0x2d0);
-EMIT_RI10(spe_clgtbi,    0x05e);
-EMIT_RR  (spe_clgth,     0x2c8);
-EMIT_RI10(spe_clgthi,    0x05d);
-EMIT_RR  (spe_clgt,      0x2c0);
-EMIT_RI10(spe_clgti,     0x05c);
-EMIT_I16 (spe_br,        0x064);
-EMIT_I16 (spe_bra,       0x060);
-EMIT_RI16(spe_brsl,      0x066);
-EMIT_RI16(spe_brasl,     0x062);
-EMIT_RI16(spe_brnz,      0x042);
-EMIT_RI16(spe_brz,       0x040);
-EMIT_RI16(spe_brhnz,     0x046);
-EMIT_RI16(spe_brhz,      0x044);
+EMIT_RR  (spe_heq,       0x3d8)
+EMIT_RI10(spe_heqi,      0x07f)
+EMIT_RR  (spe_hgt,       0x258)
+EMIT_RI10(spe_hgti,      0x04f)
+EMIT_RR  (spe_hlgt,      0x2d8)
+EMIT_RI10(spe_hlgti,     0x05f)
+EMIT_RR  (spe_ceqb,      0x3d0)
+EMIT_RI10(spe_ceqbi,     0x07e)
+EMIT_RR  (spe_ceqh,      0x3c8)
+EMIT_RI10(spe_ceqhi,     0x07d)
+EMIT_RR  (spe_ceq,       0x3c0)
+EMIT_RI10(spe_ceqi,      0x07c)
+EMIT_RR  (spe_cgtb,      0x250)
+EMIT_RI10(spe_cgtbi,     0x04e)
+EMIT_RR  (spe_cgth,      0x248)
+EMIT_RI10(spe_cgthi,     0x04d)
+EMIT_RR  (spe_cgt,       0x240)
+EMIT_RI10(spe_cgti,      0x04c)
+EMIT_RR  (spe_clgtb,     0x2d0)
+EMIT_RI10(spe_clgtbi,    0x05e)
+EMIT_RR  (spe_clgth,     0x2c8)
+EMIT_RI10(spe_clgthi,    0x05d)
+EMIT_RR  (spe_clgt,      0x2c0)
+EMIT_RI10(spe_clgti,     0x05c)
+EMIT_I16 (spe_br,        0x064)
+EMIT_I16 (spe_bra,       0x060)
+EMIT_RI16(spe_brsl,      0x066)
+EMIT_RI16(spe_brasl,     0x062)
+EMIT_RI16(spe_brnz,      0x042)
+EMIT_RI16(spe_brz,       0x040)
+EMIT_RI16(spe_brhnz,     0x046)
+EMIT_RI16(spe_brhz,      0x044)
+
+/* Control instructions
+ */
+EMIT     (spe_lnop,      0x001)
+
+extern void
+spe_lqd(struct spe_function *p, unsigned rT, unsigned rA, int offset);
+
+extern void
+spe_stqd(struct spe_function *p, unsigned rT, unsigned rA, int offset);
 
 extern void spe_bi(struct spe_function *p, unsigned rA, int d, int e);
 extern void spe_iret(struct spe_function *p, unsigned rA, int d, int e);
@@ -292,13 +330,33 @@ spe_load_float(struct spe_function *p, unsigned rT, float x);
 extern void
 spe_load_int(struct spe_function *p, unsigned rT, int i);
 
+/** Load/splat immediate unsigned int into rT. */
+extern void
+spe_load_uint(struct spe_function *p, unsigned rT, unsigned int ui);
+
+/** And immediate value into rT. */
+extern void
+spe_and_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui);
+
+/** Xor immediate value into rT. */
+extern void
+spe_xor_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui);
+
+/** Compare equal with immediate value. */
+extern void
+spe_compare_equal_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui);
+
+/** Compare greater with immediate value. */
+extern void
+spe_compare_greater_uint(struct spe_function *p, unsigned rT, unsigned rA, unsigned int ui);
+
 /** Replicate word 0 of rA across rT. */
 extern void
 spe_splat(struct spe_function *p, unsigned rT, unsigned rA);
 
-/** Complement/invert all bits in rT. */
+/** rT = complement_all_bits(rA). */
 extern void
-spe_complement(struct spe_function *p, unsigned rT);
+spe_complement(struct spe_function *p, unsigned rT, unsigned rA);
 
 /** rT = rA. */
 extern void
@@ -308,52 +366,65 @@ spe_move(struct spe_function *p, unsigned rT, unsigned rA);
 extern void
 spe_zero(struct spe_function *p, unsigned rT);
 
+/** rT = splat(rA, word) */
+extern void
+spe_splat_word(struct spe_function *p, unsigned rT, unsigned rA, int word);
+
+/** rT = float min(rA, rB) */
+extern void
+spe_float_min(struct spe_function *p, unsigned rT, unsigned rA, unsigned rB);
+
+/** rT = float max(rA, rB) */
+extern void
+spe_float_max(struct spe_function *p, unsigned rT, unsigned rA, unsigned rB);
+
 
 /* Floating-point instructions
  */
-EMIT_RR  (spe_fa,         0x2c4);
-EMIT_RR  (spe_dfa,        0x2cc);
-EMIT_RR  (spe_fs,         0x2c5);
-EMIT_RR  (spe_dfs,        0x2cd);
-EMIT_RR  (spe_fm,         0x2c6);
-EMIT_RR  (spe_dfm,        0x2ce);
-EMIT_RRR (spe_fma,        0x00e);
-EMIT_RR  (spe_dfma,       0x35c);
-EMIT_RRR (spe_fnms,       0x00d);
-EMIT_RR  (spe_dfnms,      0x35e);
-EMIT_RRR (spe_fms,        0x00f);
-EMIT_RR  (spe_dfms,       0x35d);
-EMIT_RR  (spe_dfnma,      0x35f);
-EMIT_R   (spe_frest,      0x1b8);
-EMIT_R   (spe_frsqest,    0x1b9);
-EMIT_RR  (spe_fi,         0x3d4);
-EMIT_RI8 (spe_csflt,      0x1da, 155);
-EMIT_RI8 (spe_cflts,      0x1d8, 173);
-EMIT_RI8 (spe_cuflt,      0x1db, 155);
-EMIT_RI8 (spe_cfltu,      0x1d9, 173);
-EMIT_R   (spe_frds,       0x3b9);
-EMIT_R   (spe_fesd,       0x3b8);
-EMIT_RR  (spe_dfceq,      0x3c3);
-EMIT_RR  (spe_dfcmeq,     0x3cb);
-EMIT_RR  (spe_dfcgt,      0x2c3);
-EMIT_RR  (spe_dfcmgt,     0x2cb);
-EMIT_RI7 (spe_dftsv,      0x3bf);
-EMIT_RR  (spe_fceq,       0x3c2);
-EMIT_RR  (spe_fcmeq,      0x3ca);
-EMIT_RR  (spe_fcgt,       0x2c2);
-EMIT_RR  (spe_fcmgt,      0x2ca);
-EMIT_R   (spe_fscrwr,     0x3ba);
-EMIT_    (spe_fscrrd,     0x398);
+EMIT_RR  (spe_fa,         0x2c4)
+EMIT_RR  (spe_dfa,        0x2cc)
+EMIT_RR  (spe_fs,         0x2c5)
+EMIT_RR  (spe_dfs,        0x2cd)
+EMIT_RR  (spe_fm,         0x2c6)
+EMIT_RR  (spe_dfm,        0x2ce)
+EMIT_RRR (spe_fma,        0x00e)
+EMIT_RR  (spe_dfma,       0x35c)
+EMIT_RRR (spe_fnms,       0x00d)
+EMIT_RR  (spe_dfnms,      0x35e)
+EMIT_RRR (spe_fms,        0x00f)
+EMIT_RR  (spe_dfms,       0x35d)
+EMIT_RR  (spe_dfnma,      0x35f)
+EMIT_R   (spe_frest,      0x1b8)
+EMIT_R   (spe_frsqest,    0x1b9)
+EMIT_RR  (spe_fi,         0x3d4)
+EMIT_RI8 (spe_csflt,      0x1da, 155)
+EMIT_RI8 (spe_cflts,      0x1d8, 173)
+EMIT_RI8 (spe_cuflt,      0x1db, 155)
+EMIT_RI8 (spe_cfltu,      0x1d9, 173)
+EMIT_R   (spe_frds,       0x3b9)
+EMIT_R   (spe_fesd,       0x3b8)
+EMIT_RR  (spe_dfceq,      0x3c3)
+EMIT_RR  (spe_dfcmeq,     0x3cb)
+EMIT_RR  (spe_dfcgt,      0x2c3)
+EMIT_RR  (spe_dfcmgt,     0x2cb)
+EMIT_RI7 (spe_dftsv,      0x3bf)
+EMIT_RR  (spe_fceq,       0x3c2)
+EMIT_RR  (spe_fcmeq,      0x3ca)
+EMIT_RR  (spe_fcgt,       0x2c2)
+EMIT_RR  (spe_fcmgt,      0x2ca)
+EMIT_R   (spe_fscrwr,     0x3ba)
+EMIT_    (spe_fscrrd,     0x398)
 
 
 /* Channel instructions
  */
-EMIT_R   (spe_rdch,       0x00d);
-EMIT_R   (spe_rdchcnt,    0x00f);
-EMIT_R   (spe_wrch,       0x10d);
+EMIT_R   (spe_rdch,       0x00d)
+EMIT_R   (spe_rdchcnt,    0x00f)
+EMIT_R   (spe_wrch,       0x10d)
 
 
 #ifdef UNDEF_EMIT_MACROS
+#undef EMIT
 #undef EMIT_
 #undef EMIT_R
 #undef EMIT_RR
@@ -361,6 +432,7 @@ EMIT_R   (spe_wrch,       0x10d);
 #undef EMIT_RI7
 #undef EMIT_RI8
 #undef EMIT_RI10
+#undef EMIT_RI10s
 #undef EMIT_RI16
 #undef EMIT_RI18
 #undef EMIT_I16
diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
index ad9d8f8ced9..99ee74cf14b 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
@@ -240,7 +240,8 @@ static void emit_modrm( struct x86_function *p,
    /* Oh-oh we've stumbled into the SIB thing.
     */
    if (regmem.file == file_REG32 &&
-       regmem.idx == reg_SP) {
+       regmem.idx == reg_SP &&
+       regmem.mod != mod_REG) {
       emit_1ub(p, 0x24);		/* simplistic! */
    }
 
@@ -439,25 +440,70 @@ void x86_call( struct x86_function *p, struct x86_reg reg)
 }
 
 
-/* michal:
- * Temporary. As I need immediate operands, and dont want to mess with the codegen,
- * I load the immediate into general purpose register and use it.
- */
 void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm )
 {
    DUMP_RI( dst, imm );
+   assert(dst.file == file_REG32);
    assert(dst.mod == mod_REG);
    emit_1ub(p, 0xb8 + dst.idx);
    emit_1i(p, imm);
 }
 
-void x86_add_reg_imm8( struct x86_function *p, struct x86_reg dst, ubyte imm )
+/**
+ * Immediate group 1 instructions.
+ */
+static INLINE void 
+x86_group1_imm( struct x86_function *p, 
+                unsigned op, struct x86_reg dst, int imm )
 {
-   DUMP_RI( dst, imm );
+   assert(dst.file == file_REG32);
    assert(dst.mod == mod_REG);
-   emit_1ub(p, 0x80);
-   emit_modrm_noreg(p, 0, dst);
-   emit_1ub(p, imm);
+   if(-0x80 <= imm && imm < 0x80) {
+      emit_1ub(p, 0x83);
+      emit_modrm_noreg(p, op, dst);
+      emit_1b(p, (char)imm);
+   }
+   else {
+      emit_1ub(p, 0x81);
+      emit_modrm_noreg(p, op, dst);
+      emit_1i(p, imm);
+   }
+}
+
+void x86_add_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 0, dst, imm);
+}
+
+void x86_or_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 1, dst, imm);
+}
+
+void x86_and_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 4, dst, imm);
+}
+
+void x86_sub_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 5, dst, imm);
+}
+
+void x86_xor_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 6, dst, imm);
+}
+
+void x86_cmp_imm( struct x86_function *p, struct x86_reg dst, int imm )
+{
+   DUMP_RI( dst, imm );
+   x86_group1_imm(p, 7, dst, imm);
 }
 
 
diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
index af79f07dd39..1b5eaaca850 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
+++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
@@ -152,12 +152,13 @@ void x86_jmp( struct x86_function *p, int label );
 /* void x86_call( struct x86_function *p, void (*label)() ); */
 void x86_call( struct x86_function *p, struct x86_reg reg);
 
-/* michal:
- * Temporary. As I need immediate operands, and dont want to mess with the codegen,
- * I load the immediate into general purpose register and use it.
- */
 void x86_mov_reg_imm( struct x86_function *p, struct x86_reg dst, int imm );
-void x86_add_reg_imm8( struct x86_function *p, struct x86_reg dst, ubyte imm );
+void x86_add_imm( struct x86_function *p, struct x86_reg dst, int imm );
+void x86_or_imm( struct x86_function *p, struct x86_reg dst, int imm );
+void x86_and_imm( struct x86_function *p, struct x86_reg dst, int imm );
+void x86_sub_imm( struct x86_function *p, struct x86_reg dst, int imm );
+void x86_xor_imm( struct x86_function *p, struct x86_reg dst, int imm );
+void x86_cmp_imm( struct x86_function *p, struct x86_reg dst, int imm );
 
 
 /* Macro for sse_shufps() and sse2_pshufd():
diff --git a/src/gallium/auxiliary/tgsi/Makefile b/src/gallium/auxiliary/tgsi/Makefile
index c7155a93168..d7df9490cfa 100644
--- a/src/gallium/auxiliary/tgsi/Makefile
+++ b/src/gallium/auxiliary/tgsi/Makefile
@@ -11,6 +11,7 @@ C_SOURCES = \
 	tgsi_info.c \
 	tgsi_iterate.c \
 	tgsi_parse.c \
+	tgsi_ppc.c \
 	tgsi_scan.c \
 	tgsi_sse2.c \
 	tgsi_text.c \
diff --git a/src/gallium/auxiliary/tgsi/SConscript b/src/gallium/auxiliary/tgsi/SConscript
index 45bf3f6d577..8200cce42f5 100644
--- a/src/gallium/auxiliary/tgsi/SConscript
+++ b/src/gallium/auxiliary/tgsi/SConscript
@@ -12,6 +12,7 @@ tgsi = env.ConvenienceLibrary(
 		'tgsi_parse.c',
 		'tgsi_sanity.c',
 		'tgsi_scan.c',
+		'tgsi_ppc.c',
 		'tgsi_sse2.c',
 		'tgsi_text.c',
 		'tgsi_transform.c',
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c
index 74614d36884..fd02c2c87c8 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.c
@@ -119,6 +119,8 @@ tgsi_default_declaration( void )
    declaration.UsageMask = TGSI_WRITEMASK_XYZW;
    declaration.Interpolate = TGSI_INTERPOLATE_CONSTANT;
    declaration.Semantic = 0;
+   declaration.Centroid = 0;
+   declaration.Invariant = 0;
    declaration.Padding = 0;
    declaration.Extended = 0;
 
@@ -131,6 +133,8 @@ tgsi_build_declaration(
    unsigned usage_mask,
    unsigned interpolate,
    unsigned semantic,
+   unsigned centroid,
+   unsigned invariant,
    struct tgsi_header *header )
 {
    struct tgsi_declaration declaration;
@@ -143,6 +147,8 @@ tgsi_build_declaration(
    declaration.UsageMask = usage_mask;
    declaration.Interpolate = interpolate;
    declaration.Semantic = semantic;
+   declaration.Centroid = centroid;
+   declaration.Invariant = invariant;
 
    header_bodysize_grow( header );
 
@@ -194,6 +200,8 @@ tgsi_build_full_declaration(
       full_decl->Declaration.UsageMask,
       full_decl->Declaration.Interpolate,
       full_decl->Declaration.Semantic,
+      full_decl->Declaration.Centroid,
+      full_decl->Declaration.Invariant,
       header );
 
    if (maxsize <= size)
@@ -793,10 +801,14 @@ tgsi_default_instruction_ext_nv( void )
    return instruction_ext_nv;
 }
 
-union token_u32
+
+/** test for inequality of 32-bit values pointed to by a and b */
+static INLINE boolean
+compare32(const void *a, const void *b)
 {
-   unsigned u32;
-};
+   return *((uint32_t *) a) != *((uint32_t *) b);
+}
+
 
 unsigned
 tgsi_compare_instruction_ext_nv(
@@ -805,7 +817,7 @@ tgsi_compare_instruction_ext_nv(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_instruction_ext_nv
@@ -864,7 +876,7 @@ tgsi_compare_instruction_ext_label(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_instruction_ext_label
@@ -905,7 +917,7 @@ tgsi_compare_instruction_ext_texture(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_instruction_ext_texture
@@ -1027,7 +1039,7 @@ tgsi_compare_src_register_ext_swz(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_src_register_ext_swz
@@ -1095,7 +1107,7 @@ tgsi_compare_src_register_ext_mod(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_src_register_ext_mod
@@ -1241,7 +1253,7 @@ tgsi_compare_dst_register_ext_concode(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_dst_register_ext_concode
@@ -1299,7 +1311,7 @@ tgsi_compare_dst_register_ext_modulate(
 {
    a.Padding = b.Padding = 0;
    a.Extended = b.Extended = 0;
-   return ((union token_u32 *) &a)->u32 != ((union token_u32 *) &b)->u32;
+   return compare32(&a, &b);
 }
 
 struct tgsi_dst_register_ext_modulate
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.h b/src/gallium/auxiliary/tgsi/tgsi_build.h
index 7d6234746a2..0fd6fabd83d 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.h
@@ -71,6 +71,8 @@ tgsi_build_declaration(
    unsigned usage_mask,
    unsigned interpolate,
    unsigned semantic,
+   unsigned centroid,
+   unsigned invariant,
    struct tgsi_header *header );
 
 struct tgsi_full_declaration
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index 3177f549523..c2a0ac5aff8 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -246,6 +246,14 @@ iter_declaration(
    TXT( ", " );
    ENM( decl->Declaration.Interpolate, interpolate_names );
 
+   if (decl->Declaration.Centroid) {
+      TXT( ", CENTROID" );
+   }
+
+   if (decl->Declaration.Invariant) {
+      TXT( ", INVARIANT" );
+   }
+
    EOL();
 
    return TRUE;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 65a0f39fdb4..a2d2cfd1fcc 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -133,7 +133,7 @@ tgsi_exec_machine_bind_shader(
    struct tgsi_exec_machine *mach,
    const struct tgsi_token *tokens,
    uint numSamplers,
-   struct tgsi_sampler *samplers)
+   struct tgsi_sampler **samplers)
 {
    uint k;
    struct tgsi_parse_context parse;
@@ -467,17 +467,6 @@ micro_exp2(
 }
 
 static void
-micro_f2it(
-   union tgsi_exec_channel *dst,
-   const union tgsi_exec_channel *src )
-{
-   dst->i[0] = (int) src->f[0];
-   dst->i[1] = (int) src->f[1];
-   dst->i[2] = (int) src->f[2];
-   dst->i[3] = (int) src->f[3];
-}
-
-static void
 micro_f2ut(
    union tgsi_exec_channel *dst,
    const union tgsi_exec_channel *src )
@@ -969,14 +958,22 @@ fetch_src_file_channel(
       switch( file ) {
       case TGSI_FILE_CONSTANT:
          assert(mach->Consts);
-         assert(index->i[0] >= 0);
-         assert(index->i[1] >= 0);
-         assert(index->i[2] >= 0);
-         assert(index->i[3] >= 0);
-         chan->f[0] = mach->Consts[index->i[0]][swizzle];
-         chan->f[1] = mach->Consts[index->i[1]][swizzle];
-         chan->f[2] = mach->Consts[index->i[2]][swizzle];
-         chan->f[3] = mach->Consts[index->i[3]][swizzle];
+         if (index->i[0] < 0)
+            chan->f[0] = 0.0f;
+         else
+            chan->f[0] = mach->Consts[index->i[0]][swizzle];
+         if (index->i[1] < 0)
+            chan->f[1] = 0.0f;
+         else
+            chan->f[1] = mach->Consts[index->i[1]][swizzle];
+         if (index->i[2] < 0)
+            chan->f[2] = 0.0f;
+         else
+            chan->f[2] = mach->Consts[index->i[2]][swizzle];
+         if (index->i[3] < 0)
+            chan->f[3] = 0.0f;
+         else
+            chan->f[3] = mach->Consts[index->i[3]][swizzle];
          break;
 
       case TGSI_FILE_INPUT:
@@ -1048,11 +1045,28 @@ fetch_source(
    union tgsi_exec_channel index;
    uint swizzle;
 
+   /* We start with a direct index into a register file.
+    *
+    *    file[1],
+    *    where:
+    *       file = SrcRegister.File
+    *       [1] = SrcRegister.Index
+    */
    index.i[0] =
    index.i[1] =
    index.i[2] =
    index.i[3] = reg->SrcRegister.Index;
 
+   /* There is an extra source register that indirectly subscripts
+    * a register file. The direct index now becomes an offset
+    * that is being added to the indirect register.
+    *
+    *    file[ind[2].x+1],
+    *    where:
+    *       ind = SrcRegisterInd.File
+    *       [2] = SrcRegisterInd.Index
+    *       .x = SrcRegisterInd.SwizzleX
+    */
    if (reg->SrcRegister.Indirect) {
       union tgsi_exec_channel index2;
       union tgsi_exec_channel indir_index;
@@ -1075,10 +1089,10 @@ fetch_source(
          &indir_index );
 
       /* add value of address register to the offset */
-      index.i[0] += indir_index.i[0];
-      index.i[1] += indir_index.i[1];
-      index.i[2] += indir_index.i[2];
-      index.i[3] += indir_index.i[3];
+      index.i[0] += (int) indir_index.f[0];
+      index.i[1] += (int) indir_index.f[1];
+      index.i[2] += (int) indir_index.f[2];
+      index.i[3] += (int) indir_index.f[3];
 
       /* for disabled execution channels, zero-out the index to
        * avoid using a potential garbage value.
@@ -1089,19 +1103,31 @@ fetch_source(
       }
    }
 
-   if( reg->SrcRegister.Dimension ) {
-      switch( reg->SrcRegister.File ) {
+   /* There is an extra source register that is a second
+    * subscript to a register file. Effectively it means that
+    * the register file is actually a 2D array of registers.
+    *
+    *    file[1][3] == file[1*sizeof(file[1])+3],
+    *    where:
+    *       [3] = SrcRegisterDim.Index
+    */
+   if (reg->SrcRegister.Dimension) {
+      /* The size of the first-order array depends on the register file type.
+       * We need to multiply the index to the first array to get an effective,
+       * "flat" index that points to the beginning of the second-order array.
+       */
+      switch (reg->SrcRegister.File) {
       case TGSI_FILE_INPUT:
-         index.i[0] *= 17;
-         index.i[1] *= 17;
-         index.i[2] *= 17;
-         index.i[3] *= 17;
+         index.i[0] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
+         index.i[1] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
+         index.i[2] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
+         index.i[3] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
          break;
       case TGSI_FILE_CONSTANT:
-         index.i[0] *= 4096;
-         index.i[1] *= 4096;
-         index.i[2] *= 4096;
-         index.i[3] *= 4096;
+         index.i[0] *= TGSI_EXEC_MAX_CONST_BUFFER;
+         index.i[1] *= TGSI_EXEC_MAX_CONST_BUFFER;
+         index.i[2] *= TGSI_EXEC_MAX_CONST_BUFFER;
+         index.i[3] *= TGSI_EXEC_MAX_CONST_BUFFER;
          break;
       default:
          assert( 0 );
@@ -1112,6 +1138,17 @@ fetch_source(
       index.i[2] += reg->SrcRegisterDim.Index;
       index.i[3] += reg->SrcRegisterDim.Index;
 
+      /* Again, the second subscript index can be addressed indirectly
+       * identically to the first one.
+       * Nothing stops us from indirectly addressing the indirect register,
+       * but there is no need for that, so we won't exercise it.
+       *
+       *    file[1][ind[4].y+3],
+       *    where:
+       *       ind = SrcRegisterDimInd.File
+       *       [4] = SrcRegisterDimInd.Index
+       *       .y = SrcRegisterDimInd.SwizzleX
+       */
       if (reg->SrcRegisterDim.Indirect) {
          union tgsi_exec_channel index2;
          union tgsi_exec_channel indir_index;
@@ -1131,10 +1168,10 @@ fetch_source(
             &index2,
             &indir_index );
 
-         index.i[0] += indir_index.i[0];
-         index.i[1] += indir_index.i[1];
-         index.i[2] += indir_index.i[2];
-         index.i[3] += indir_index.i[3];
+         index.i[0] += (int) indir_index.f[0];
+         index.i[1] += (int) indir_index.f[1];
+         index.i[2] += (int) indir_index.f[2];
+         index.i[3] += (int) indir_index.f[3];
 
          /* for disabled execution channels, zero-out the index to
           * avoid using a potential garbage value.
@@ -1144,6 +1181,11 @@ fetch_source(
                index.i[i] = 0;
          }
       }
+
+      /* If by any chance there was a need for a 3D array of register
+       * files, we would have to check whether SrcRegisterDim is followed
+       * by a dimension register and continue the saga.
+       */
    }
 
    swizzle = tgsi_util_get_full_src_register_extswizzle( reg, chan_index );
@@ -1493,7 +1535,7 @@ exec_kilp(struct tgsi_exec_machine *mach,
 
 
 /*
- * Fetch a texel using STR texture coordinates.
+ * Fetch a four texture samples using STR texture coordinates.
  */
 static void
 fetch_texel( struct tgsi_sampler *sampler,
@@ -1527,7 +1569,7 @@ exec_tex(struct tgsi_exec_machine *mach,
          boolean projected)
 {
    const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
-   union tgsi_exec_channel r[8];
+   union tgsi_exec_channel r[4];
    uint chan_index;
    float lodBias;
 
@@ -1550,7 +1592,7 @@ exec_tex(struct tgsi_exec_machine *mach,
       else
          lodBias = 0.0;
 
-      fetch_texel(&mach->Samplers[unit],
+      fetch_texel(mach->Samplers[unit],
                   &r[0], NULL, NULL, lodBias,  /* S, T, P, BIAS */
                   &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
       break;
@@ -1576,7 +1618,7 @@ exec_tex(struct tgsi_exec_machine *mach,
       else
          lodBias = 0.0;
 
-      fetch_texel(&mach->Samplers[unit],
+      fetch_texel(mach->Samplers[unit],
                   &r[0], &r[1], &r[2], lodBias,  /* inputs */
                   &r[0], &r[1], &r[2], &r[3]);  /* outputs */
       break;
@@ -1602,7 +1644,7 @@ exec_tex(struct tgsi_exec_machine *mach,
       else
          lodBias = 0.0;
 
-      fetch_texel(&mach->Samplers[unit],
+      fetch_texel(mach->Samplers[unit],
                   &r[0], &r[1], &r[2], lodBias,
                   &r[0], &r[1], &r[2], &r[3]);
       break;
@@ -1712,6 +1754,7 @@ exec_declaration(
             break;
 
          default:
+            eval = NULL;
             assert( 0 );
          }
 
@@ -1754,7 +1797,7 @@ exec_instruction(
    case TGSI_OPCODE_ARL:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
-         micro_f2it( &r[0], &r[0] );
+         micro_trunc( &r[0], &r[0] );
          STORE( &r[0], 0, chan_index );
       }
       break;
@@ -2044,7 +2087,21 @@ exec_instruction(
 
    case TGSI_OPCODE_DOT2ADD:
       /* TGSI_OPCODE_DP2A */
-      assert (0);
+      FETCH( &r[0], 0, CHAN_X );
+      FETCH( &r[1], 1, CHAN_X );
+      micro_mul( &r[0], &r[0], &r[1] );
+
+      FETCH( &r[1], 0, CHAN_Y );
+      FETCH( &r[2], 1, CHAN_Y );
+      micro_mul( &r[1], &r[1], &r[2] );
+      micro_add( &r[0], &r[0], &r[1] );
+
+      FETCH( &r[2], 2, CHAN_X );
+      micro_add( &r[0], &r[0], &r[2] );
+
+      FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+         STORE( &r[0], 0, chan_index );
+      }
       break;
 
    case TGSI_OPCODE_INDEX:
@@ -2491,7 +2548,8 @@ exec_instruction(
          micro_mul( &dot, &r[2], &r[2] );
          micro_add( &tmp, &tmp, &dot );
 
-         /* tmp = 1 / tmp */
+         /* tmp = 1 / sqrt(tmp) */
+         micro_sqrt( &tmp, &tmp );
          micro_div( &tmp, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &tmp );
 
          /* note: w channel is undefined */
@@ -2524,7 +2582,8 @@ exec_instruction(
          micro_mul( &dot, &r[3], &r[3] );
          micro_add( &tmp, &tmp, &dot );
 
-         /* tmp = 1 / tmp */
+         /* tmp = 1 / sqrt(tmp) */
+         micro_sqrt( &tmp, &tmp );
          micro_div( &tmp, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &tmp );
 
          FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index fc40a25e09f..4ffd4efbffa 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -68,17 +68,12 @@ struct tgsi_interp_coef
    float dady[NUM_CHANNELS];
 };
 
-
-struct softpipe_tile_cache;  /**< Opaque to TGSI */
-
 /**
  * Information for sampling textures, which must be implemented
  * by code outside the TGSI executor.
  */
 struct tgsi_sampler
 {
-   const struct pipe_sampler_state *state;
-   struct pipe_texture *texture;
    /** Get samples for four fragments in a quad */
    void (*get_samples)(struct tgsi_sampler *sampler,
                        const float s[QUAD_SIZE],
@@ -86,8 +81,6 @@ struct tgsi_sampler
                        const float p[QUAD_SIZE],
                        float lodbias,
                        float rgba[NUM_CHANNELS][QUAD_SIZE]);
-   void *pipe; /*XXX temporary*/
-   struct softpipe_tile_cache *cache;
 };
 
 /**
@@ -178,6 +171,16 @@ struct tgsi_exec_labels
 #define TGSI_EXEC_MAX_LOOP_NESTING  20
 #define TGSI_EXEC_MAX_CALL_NESTING  20
 
+/* The maximum number of input attributes per vertex. For 2D
+ * input register files, this is the stride between two 1D
+ * arrays.
+ */
+#define TGSI_EXEC_MAX_INPUT_ATTRIBS 17
+
+/* The maximum number of constant vectors per constant buffer.
+ */
+#define TGSI_EXEC_MAX_CONST_BUFFER  4096
+
 /**
  * Run-time virtual machine state for executing TGSI shader.
  */
@@ -195,7 +198,7 @@ struct tgsi_exec_machine
    struct tgsi_exec_vector       *Temps;
    struct tgsi_exec_vector       *Addrs;
 
-   struct tgsi_sampler           *Samplers;
+   struct tgsi_sampler           **Samplers;
 
    float                         Imms[TGSI_EXEC_NUM_IMMEDIATES][4];
    unsigned                      ImmLimit;
@@ -258,7 +261,7 @@ tgsi_exec_machine_bind_shader(
    struct tgsi_exec_machine *mach,
    const struct tgsi_token *tokens,
    uint numSamplers,
-   struct tgsi_sampler *samplers);
+   struct tgsi_sampler **samplers);
 
 uint
 tgsi_exec_machine_run(
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.c b/src/gallium/auxiliary/tgsi/tgsi_parse.c
index 3757486ba9b..2cd56e413a5 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.c
@@ -88,16 +88,33 @@ tgsi_parse_end_of_tokens(
       1 + ctx->FullHeader.Header.HeaderSize + ctx->FullHeader.Header.BodySize;
 }
 
+
+/**
+ * This function is used to avoid and work-around type punning/aliasing
+ * warnings.  The warnings seem harmless on x86 but on PPC they cause
+ * real failures.
+ */
+static INLINE void
+copy_token(void *dst, const void *src)
+{
+   memcpy(dst, src, 4);
+}
+
+
+/**
+ * Get next 4-byte token, return it at address specified by 'token'
+ */
 static void
 next_token(
    struct tgsi_parse_context *ctx,
    void *token )
 {
    assert( !tgsi_parse_end_of_tokens( ctx ) );
-
-   *(struct tgsi_token *) token = ctx->Tokens[ctx->Position++];
+   copy_token(token, &ctx->Tokens[ctx->Position]);
+   ctx->Position++;
 }
 
+
 void
 tgsi_parse_token(
    struct tgsi_parse_context *ctx )
@@ -116,7 +133,7 @@ tgsi_parse_token(
       struct tgsi_full_declaration *decl = &ctx->FullToken.FullDeclaration;
 
       *decl = tgsi_default_full_declaration();
-      decl->Declaration = *(struct tgsi_declaration *) &token;
+      copy_token(&decl->Declaration, &token);
 
       next_token( ctx, &decl->DeclarationRange );
 
@@ -132,8 +149,7 @@ tgsi_parse_token(
       struct tgsi_full_immediate *imm = &ctx->FullToken.FullImmediate;
 
       *imm = tgsi_default_full_immediate();
-      imm->Immediate = *(struct tgsi_immediate *) &token;
-
+      copy_token(&imm->Immediate, &token);
       assert( !imm->Immediate.Extended );
 
       switch (imm->Immediate.DataType) {
@@ -158,8 +174,7 @@ tgsi_parse_token(
       unsigned extended;
 
       *inst = tgsi_default_full_instruction();
-      inst->Instruction = *(struct tgsi_instruction *) &token;
-
+      copy_token(&inst->Instruction, &token);
       extended = inst->Instruction.Extended;
 
       while( extended ) {
@@ -169,18 +184,15 @@ tgsi_parse_token(
 
          switch( token.Type ) {
          case TGSI_INSTRUCTION_EXT_TYPE_NV:
-            inst->InstructionExtNv =
-               *(struct tgsi_instruction_ext_nv *) &token;
+            copy_token(&inst->InstructionExtNv, &token);
             break;
 
          case TGSI_INSTRUCTION_EXT_TYPE_LABEL:
-            inst->InstructionExtLabel =
-               *(struct tgsi_instruction_ext_label *) &token;
+            copy_token(&inst->InstructionExtLabel, &token);
             break;
 
          case TGSI_INSTRUCTION_EXT_TYPE_TEXTURE:
-            inst->InstructionExtTexture =
-               *(struct tgsi_instruction_ext_texture *) &token;
+            copy_token(&inst->InstructionExtTexture, &token);
             break;
 
          default:
@@ -212,13 +224,13 @@ tgsi_parse_token(
 
             switch( token.Type ) {
             case TGSI_DST_REGISTER_EXT_TYPE_CONDCODE:
-               inst->FullDstRegisters[i].DstRegisterExtConcode =
-                  *(struct tgsi_dst_register_ext_concode *) &token;
+               copy_token(&inst->FullDstRegisters[i].DstRegisterExtConcode,
+                          &token);
                break;
 
             case TGSI_DST_REGISTER_EXT_TYPE_MODULATE:
-               inst->FullDstRegisters[i].DstRegisterExtModulate =
-                  *(struct tgsi_dst_register_ext_modulate *) &token;
+               copy_token(&inst->FullDstRegisters[i].DstRegisterExtModulate,
+                          &token);
                break;
 
             default:
@@ -245,13 +257,13 @@ tgsi_parse_token(
 
             switch( token.Type ) {
             case TGSI_SRC_REGISTER_EXT_TYPE_SWZ:
-               inst->FullSrcRegisters[i].SrcRegisterExtSwz =
-                  *(struct tgsi_src_register_ext_swz *) &token;
+               copy_token(&inst->FullSrcRegisters[i].SrcRegisterExtSwz,
+                          &token);
                break;
 
             case TGSI_SRC_REGISTER_EXT_TYPE_MOD:
-               inst->FullSrcRegisters[i].SrcRegisterExtMod =
-                  *(struct tgsi_src_register_ext_mod *) &token;
+               copy_token(&inst->FullSrcRegisters[i].SrcRegisterExtMod,
+                          &token);
                break;
 
             default:
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ppc.c b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
new file mode 100644
index 00000000000..a92b1902e3d
--- /dev/null
+++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
@@ -0,0 +1,1329 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * TGSI to PowerPC code generation.
+ */
+
+#include "pipe/p_config.h"
+
+#if defined(PIPE_ARCH_PPC)
+
+#include "pipe/p_debug.h"
+#include "pipe/p_shader_tokens.h"
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "util/u_sse.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+#include "tgsi_dump.h"
+#include "tgsi_exec.h"
+#include "tgsi_ppc.h"
+#include "rtasm/rtasm_ppc.h"
+
+
+/**
+ * Since it's pretty much impossible to form PPC vector immediates, load
+ * them from memory here:
+ */
+const float ppc_builtin_constants[] ALIGN16_ATTRIB = {
+   1.0f, -128.0f, 128.0, 0.0
+};
+
+
+#define FOR_EACH_CHANNEL( CHAN )\
+   for (CHAN = 0; CHAN < NUM_CHANNELS; CHAN++)
+
+#define IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
+   ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN)))
+
+#define IF_IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
+   if (IS_DST0_CHANNEL_ENABLED( INST, CHAN ))
+
+#define FOR_EACH_DST0_ENABLED_CHANNEL( INST, CHAN )\
+   FOR_EACH_CHANNEL( CHAN )\
+      IF_IS_DST0_CHANNEL_ENABLED( INST, CHAN )
+
+#define CHAN_X 0
+#define CHAN_Y 1
+#define CHAN_Z 2
+#define CHAN_W 3
+
+
+/**
+ * How many TGSI temps should be implemented with real PPC vector registers
+ * rather than memory.
+ */
+#define MAX_PPC_TEMPS 4
+
+
+struct reg_chan_vec
+{
+   struct tgsi_full_src_register src;
+   uint chan;
+   uint vec;
+};
+
+
+/**
+ * Context/state used during code gen.
+ */
+struct gen_context
+{
+   struct ppc_function *f;
+   int inputs_reg;    /**< GP register pointing to input params */
+   int outputs_reg;   /**< GP register pointing to output params */
+   int temps_reg;     /**< GP register pointing to temporary "registers" */
+   int immed_reg;     /**< GP register pointing to immediates buffer */
+   int const_reg;     /**< GP register pointing to constants buffer */
+   int builtins_reg;  /**< GP register pointint to built-in constants */
+
+   int offset_reg;    /**< used to reduce redundant li instructions */
+   int offset_value;
+
+   int one_vec;       /**< vector register with {1.0, 1.0, 1.0, 1.0} */
+   int bit31_vec;     /**< vector register with {1<<31, 1<<31, 1<<31, 1<<31} */
+
+   /**
+    * Map TGSI temps to PPC vector temps.
+    * We have 32 PPC vector regs.  Use 16 of them for storing 4 TGSI temps.
+    * XXX currently only do this for TGSI temps [0..MAX_PPC_TEMPS-1].
+    */
+   int temps_map[MAX_PPC_TEMPS][4];
+
+   /**
+    * Cache of src registers.
+    * This is used to avoid redundant load instructions.
+    */
+   struct {
+      struct tgsi_full_src_register src;
+      uint chan;
+      uint vec;
+   } regs[12];  /* 3 src regs, 4 channels */
+   uint num_regs;
+};
+
+
+/**
+ * Initialize code generation context.
+ */
+static void
+init_gen_context(struct gen_context *gen, struct ppc_function *func)
+{
+   uint i;
+
+   memset(gen, 0, sizeof(*gen));
+   gen->f = func;
+   gen->inputs_reg = ppc_reserve_register(func, 3);   /* first function param */
+   gen->outputs_reg = ppc_reserve_register(func, 4);  /* second function param */
+   gen->temps_reg = ppc_reserve_register(func, 5);    /* ... */
+   gen->immed_reg = ppc_reserve_register(func, 6);
+   gen->const_reg = ppc_reserve_register(func, 7);
+   gen->builtins_reg = ppc_reserve_register(func, 8);
+   gen->one_vec = -1;
+   gen->bit31_vec = -1;
+   gen->offset_reg = -1;
+   gen->offset_value = -9999999;
+   for (i = 0; i < MAX_PPC_TEMPS; i++) {
+      gen->temps_map[i][0] = ppc_allocate_vec_register(gen->f);
+      gen->temps_map[i][1] = ppc_allocate_vec_register(gen->f);
+      gen->temps_map[i][2] = ppc_allocate_vec_register(gen->f);
+      gen->temps_map[i][3] = ppc_allocate_vec_register(gen->f);
+   }
+}
+
+
+/**
+ * All PPC vector load/store instructions form an effective address
+ * by adding the contents of two registers.  For example:
+ *    lvx v2,r8,r9   # v2 = memory[r8 + r9]
+ *    stvx v2,r8,r9  # memory[r8 + r9] = v2;
+ * So our lvx/stvx instructions are typically preceded by an 'li' instruction
+ * to load r9 (above) with an immediate (an offset).
+ * This code emits that 'li' instruction, but only if the offset value is
+ * different than the previous 'li'.
+ * This optimization seems to save about 10% in the instruction count.
+ * Note that we need to unconditionally emit an 'li' inside basic blocks
+ * (such as inside loops).
+ */
+static int
+emit_li_offset(struct gen_context *gen, int offset)
+{
+   if (gen->offset_reg <= 0) {
+      /* allocate a GP register for storing load/store offset */
+      gen->offset_reg = ppc_allocate_register(gen->f);
+   }
+
+   /* emit new 'li' if offset is changing */
+   if (gen->offset_value < 0 || gen->offset_value != offset) {
+      gen->offset_value = offset;
+      ppc_li(gen->f, gen->offset_reg, offset);
+   }
+
+   return gen->offset_reg;
+}
+
+
+/**
+ * Forces subsequent emit_li_offset() calls to emit an 'li'.
+ * To be called at the top of basic blocks.
+ */
+static void
+reset_li_offset(struct gen_context *gen)
+{
+   gen->offset_value = -9999999;
+}
+
+
+
+/**
+ * Load the given vector register with {value, value, value, value}.
+ * The value must be in the ppu_builtin_constants[] array.
+ * We wouldn't need this if there was a simple way to load PPC vector
+ * registers with immediate values!
+ */
+static void
+load_constant_vec(struct gen_context *gen, int dst_vec, float value)
+{
+   uint pos;
+   for (pos = 0; pos < Elements(ppc_builtin_constants); pos++) {
+      if (ppc_builtin_constants[pos] == value) {
+         int offset = pos * 4;
+         int offset_reg = emit_li_offset(gen, offset);
+
+         /* Load 4-byte word into vector register.
+          * The vector slot depends on the effective address we load from.
+          * We know that our builtins start at a 16-byte boundary so we
+          * know that 'swizzle' tells us which vector slot will have the
+          * loaded word.  The other vector slots will be undefined.
+          */
+         ppc_lvewx(gen->f, dst_vec, gen->builtins_reg, offset_reg);
+         /* splat word[pos % 4] across the vector reg */
+         ppc_vspltw(gen->f, dst_vec, dst_vec, pos % 4);
+         return;
+      }
+   }
+   assert(0 && "Need to add new constant to ppc_builtin_constants array");
+}
+
+
+/**
+ * Return index of vector register containing {1.0, 1.0, 1.0, 1.0}.
+ */
+static int
+gen_one_vec(struct gen_context *gen)
+{
+   if (gen->one_vec < 0) {
+      gen->one_vec = ppc_allocate_vec_register(gen->f);
+      load_constant_vec(gen, gen->one_vec, 1.0f);
+   }
+   return gen->one_vec;
+}
+
+/**
+ * Return index of vector register containing {1<<31, 1<<31, 1<<31, 1<<31}.
+ */
+static int
+gen_get_bit31_vec(struct gen_context *gen)
+{
+   if (gen->bit31_vec < 0) {
+      gen->bit31_vec = ppc_allocate_vec_register(gen->f);
+      ppc_vspltisw(gen->f, gen->bit31_vec, -1);
+      ppc_vslw(gen->f, gen->bit31_vec, gen->bit31_vec, gen->bit31_vec);
+   }
+   return gen->bit31_vec;
+}
+
+
+/**
+ * Register fetch.  Return PPC vector register with result.
+ */
+static int
+emit_fetch(struct gen_context *gen,
+           const struct tgsi_full_src_register *reg,
+           const unsigned chan_index)
+{
+   uint swizzle = tgsi_util_get_full_src_register_extswizzle(reg, chan_index);
+   int dst_vec = -1;
+
+   switch (swizzle) {
+   case TGSI_EXTSWIZZLE_X:
+   case TGSI_EXTSWIZZLE_Y:
+   case TGSI_EXTSWIZZLE_Z:
+   case TGSI_EXTSWIZZLE_W:
+      switch (reg->SrcRegister.File) {
+      case TGSI_FILE_INPUT:
+         {
+            int offset = (reg->SrcRegister.Index * 4 + swizzle) * 16;
+            int offset_reg = emit_li_offset(gen, offset);
+            dst_vec = ppc_allocate_vec_register(gen->f);
+            ppc_lvx(gen->f, dst_vec, gen->inputs_reg, offset_reg);
+         }
+         break;
+      case TGSI_FILE_TEMPORARY:
+         if (reg->SrcRegister.Index < MAX_PPC_TEMPS) {
+            /* use PPC vec register */
+            dst_vec = gen->temps_map[reg->SrcRegister.Index][swizzle];
+         }
+         else {
+            /* use memory-based temp register "file" */
+            int offset = (reg->SrcRegister.Index * 4 + swizzle) * 16;
+            int offset_reg = emit_li_offset(gen, offset);
+            dst_vec = ppc_allocate_vec_register(gen->f);
+            ppc_lvx(gen->f, dst_vec, gen->temps_reg, offset_reg);
+         }
+         break;
+      case TGSI_FILE_IMMEDIATE:
+         {
+            int offset = (reg->SrcRegister.Index * 4 + swizzle) * 4;
+            int offset_reg = emit_li_offset(gen, offset);
+            dst_vec = ppc_allocate_vec_register(gen->f);
+            /* Load 4-byte word into vector register.
+             * The vector slot depends on the effective address we load from.
+             * We know that our immediates start at a 16-byte boundary so we
+             * know that 'swizzle' tells us which vector slot will have the
+             * loaded word.  The other vector slots will be undefined.
+             */
+            ppc_lvewx(gen->f, dst_vec, gen->immed_reg, offset_reg);
+            /* splat word[swizzle] across the vector reg */
+            ppc_vspltw(gen->f, dst_vec, dst_vec, swizzle);
+         }
+         break;
+      case TGSI_FILE_CONSTANT:
+         {
+            int offset = (reg->SrcRegister.Index * 4 + swizzle) * 4;
+            int offset_reg = emit_li_offset(gen, offset);
+            dst_vec = ppc_allocate_vec_register(gen->f);
+            /* Load 4-byte word into vector register.
+             * The vector slot depends on the effective address we load from.
+             * We know that our constants start at a 16-byte boundary so we
+             * know that 'swizzle' tells us which vector slot will have the
+             * loaded word.  The other vector slots will be undefined.
+             */
+            ppc_lvewx(gen->f, dst_vec, gen->const_reg, offset_reg);
+            /* splat word[swizzle] across the vector reg */
+            ppc_vspltw(gen->f, dst_vec, dst_vec, swizzle);
+         }
+         break;
+      default:
+         assert( 0 );
+      }
+      break;
+   case TGSI_EXTSWIZZLE_ZERO:
+      ppc_vzero(gen->f, dst_vec);
+      break;
+   case TGSI_EXTSWIZZLE_ONE:
+      {
+         int one_vec = gen_one_vec(gen);
+         dst_vec = ppc_allocate_vec_register(gen->f);
+         ppc_vmove(gen->f, dst_vec, one_vec);
+      }
+      break;
+   default:
+      assert( 0 );
+   }
+
+   assert(dst_vec >= 0);
+
+   {
+      uint sign_op = tgsi_util_get_full_src_register_sign_mode(reg, chan_index);
+      if (sign_op != TGSI_UTIL_SIGN_KEEP) {
+         int bit31_vec = gen_get_bit31_vec(gen);
+
+         switch (sign_op) {
+         case TGSI_UTIL_SIGN_CLEAR:
+            /* vec = vec & ~bit31 */
+            ppc_vandc(gen->f, dst_vec, dst_vec, bit31_vec);
+            break;
+         case TGSI_UTIL_SIGN_SET:
+            /* vec = vec | bit31 */
+            ppc_vor(gen->f, dst_vec, dst_vec, bit31_vec);
+            break;
+         case TGSI_UTIL_SIGN_TOGGLE:
+            /* vec = vec ^ bit31 */
+            ppc_vxor(gen->f, dst_vec, dst_vec, bit31_vec);
+            break;
+         default:
+            assert(0);
+         }
+      }
+   }
+
+   return dst_vec;
+}
+
+
+
+/**
+ * Test if two TGSI src registers refer to the same memory location.
+ * We use this to avoid redundant register loads.
+ */
+static boolean
+equal_src_locs(const struct tgsi_full_src_register *a, uint chan_a,
+               const struct tgsi_full_src_register *b, uint chan_b)
+{
+   int swz_a, swz_b;
+   int sign_a, sign_b;
+   if (a->SrcRegister.File != b->SrcRegister.File)
+      return FALSE;
+   if (a->SrcRegister.Index != b->SrcRegister.Index)
+      return FALSE;
+   swz_a = tgsi_util_get_full_src_register_extswizzle(a, chan_a);
+   swz_b = tgsi_util_get_full_src_register_extswizzle(b, chan_b);
+   if (swz_a != swz_b)
+      return FALSE;
+   sign_a = tgsi_util_get_full_src_register_sign_mode(a, chan_a);
+   sign_b = tgsi_util_get_full_src_register_sign_mode(b, chan_b);
+   if (sign_a != sign_b)
+      return FALSE;
+   return TRUE;
+}
+
+
+/**
+ * Given a TGSI src register and channel index, return the PPC vector
+ * register containing the value.  We use a cache to prevent re-loading
+ * the same register multiple times.
+ * \return index of PPC vector register with the desired src operand
+ */
+static int
+get_src_vec(struct gen_context *gen,
+            struct tgsi_full_instruction *inst, int src_reg, uint chan)
+{
+   const const struct tgsi_full_src_register *src = 
+      &inst->FullSrcRegisters[src_reg];
+   int vec;
+   uint i;
+
+   /* check the cache */
+   for (i = 0; i < gen->num_regs; i++) {
+      if (equal_src_locs(&gen->regs[i].src, gen->regs[i].chan, src, chan)) {
+         /* cache hit */
+         assert(gen->regs[i].vec >= 0);
+         return gen->regs[i].vec;
+      }
+   }
+
+   /* cache miss: allocate new vec reg and emit fetch/load code */
+   vec = emit_fetch(gen, src, chan);
+   gen->regs[gen->num_regs].src = *src;
+   gen->regs[gen->num_regs].chan = chan;
+   gen->regs[gen->num_regs].vec = vec;
+   gen->num_regs++;
+
+   assert(gen->num_regs <= Elements(gen->regs));
+
+   assert(vec >= 0);
+
+   return vec;
+}
+
+
+/**
+ * Clear the src operand cache.  To be called at the end of each emit function.
+ */
+static void
+release_src_vecs(struct gen_context *gen)
+{
+   uint i;
+   for (i = 0; i < gen->num_regs; i++) {
+      const const struct tgsi_full_src_register src = gen->regs[i].src;
+      if (!(src.SrcRegister.File == TGSI_FILE_TEMPORARY &&
+            src.SrcRegister.Index < MAX_PPC_TEMPS)) {
+         ppc_release_vec_register(gen->f, gen->regs[i].vec);
+      }
+   }
+   gen->num_regs = 0;
+}
+
+
+
+static int
+get_dst_vec(struct gen_context *gen, 
+            const struct tgsi_full_instruction *inst,
+            unsigned chan_index)
+{
+   const struct tgsi_full_dst_register *reg = &inst->FullDstRegisters[0];
+
+   if (reg->DstRegister.File == TGSI_FILE_TEMPORARY &&
+       reg->DstRegister.Index < MAX_PPC_TEMPS) {
+      int vec = gen->temps_map[reg->DstRegister.Index][chan_index];
+      return vec;
+   }
+   else {
+      return ppc_allocate_vec_register(gen->f);
+   }
+}
+
+
+/**
+ * Register store.  Store 'src_vec' at location indicated by 'reg'.
+ * \param free_vec  Should the src_vec be released when done?
+ */
+static void
+emit_store(struct gen_context *gen,
+           int src_vec,
+           const struct tgsi_full_instruction *inst,
+           unsigned chan_index,
+           boolean free_vec)
+{
+   const struct tgsi_full_dst_register *reg = &inst->FullDstRegisters[0];
+
+   switch (reg->DstRegister.File) {
+   case TGSI_FILE_OUTPUT:
+      {
+         int offset = (reg->DstRegister.Index * 4 + chan_index) * 16;
+         int offset_reg = emit_li_offset(gen, offset);
+         ppc_stvx(gen->f, src_vec, gen->outputs_reg, offset_reg);
+      }
+      break;
+   case TGSI_FILE_TEMPORARY:
+      if (reg->DstRegister.Index < MAX_PPC_TEMPS) {
+         if (!free_vec) {
+            int dst_vec = gen->temps_map[reg->DstRegister.Index][chan_index];
+            if (dst_vec != src_vec)
+               ppc_vmove(gen->f, dst_vec, src_vec);
+         }
+         free_vec = FALSE;
+      }
+      else {
+         int offset = (reg->DstRegister.Index * 4 + chan_index) * 16;
+         int offset_reg = emit_li_offset(gen, offset);
+         ppc_stvx(gen->f, src_vec, gen->temps_reg, offset_reg);
+      }
+      break;
+#if 0
+   case TGSI_FILE_ADDRESS:
+      emit_addrs(
+         func,
+         xmm,
+         reg->DstRegister.Index,
+         chan_index );
+      break;
+#endif
+   default:
+      assert( 0 );
+   }
+
+#if 0
+   switch( inst->Instruction.Saturate ) {
+   case TGSI_SAT_NONE:
+      break;
+
+   case TGSI_SAT_ZERO_ONE:
+      /* assert( 0 ); */
+      break;
+
+   case TGSI_SAT_MINUS_PLUS_ONE:
+      assert( 0 );
+      break;
+   }
+#endif
+
+   if (free_vec)
+      ppc_release_vec_register(gen->f, src_vec);
+}
+
+
+static void
+emit_scalar_unaryop(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int v0, v1;
+   uint chan_index;
+
+   v0 = get_src_vec(gen, inst, 0, CHAN_X);
+   v1 = ppc_allocate_vec_register(gen->f);
+
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_RSQ:
+      /* v1 = 1.0 / sqrt(v0) */
+      ppc_vrsqrtefp(gen->f, v1, v0);
+      break;
+   case TGSI_OPCODE_RCP:
+      /* v1 = 1.0 / v0 */
+      ppc_vrefp(gen->f, v1, v0);
+      break;
+   default:
+      assert(0);
+   }
+
+   FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
+      emit_store(gen, v1, inst, chan_index, FALSE);
+   }
+
+   release_src_vecs(gen);
+   ppc_release_vec_register(gen->f, v1);
+}
+
+
+static void
+emit_unaryop(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   uint chan_index;
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan_index) {
+      int v0 = get_src_vec(gen, inst, 0, chan_index);   /* v0 = srcreg[0] */
+      int v1 = get_dst_vec(gen, inst, chan_index);
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_ABS:
+         /* turn off the most significant bit of each vector float word */
+         {
+            int bit31_vec = gen_get_bit31_vec(gen);
+            ppc_vandc(gen->f, v1, v0, bit31_vec); /* v1 = v0 & ~bit31 */
+         }
+         break;
+      case TGSI_OPCODE_FLOOR:
+         ppc_vrfim(gen->f, v1, v0);         /* v1 = floor(v0) */
+         break;
+      case TGSI_OPCODE_FRAC:
+         ppc_vrfim(gen->f, v1, v0);      /* tmp = floor(v0) */
+         ppc_vsubfp(gen->f, v1, v0, v1); /* v1 = v0 - v1 */
+         break;
+      case TGSI_OPCODE_EXPBASE2:
+         ppc_vexptefp(gen->f, v1, v0);     /* v1 = 2^v0 */
+         break;
+      case TGSI_OPCODE_LOGBASE2:
+         /* XXX this may be broken! */
+         ppc_vlogefp(gen->f, v1, v0);      /* v1 = log2(v0) */
+         break;
+      case TGSI_OPCODE_MOV:
+      case TGSI_OPCODE_SWZ:
+         if (v0 != v1)
+            ppc_vmove(gen->f, v1, v0);
+         break;
+      default:
+         assert(0);
+      }
+      emit_store(gen, v1, inst, chan_index, TRUE);  /* store v0 */
+   }
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_binop(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int zero_vec = -1;
+   uint chan;
+
+   if (inst->Instruction.Opcode == TGSI_OPCODE_MUL) {
+      zero_vec = ppc_allocate_vec_register(gen->f);
+      ppc_vzero(gen->f, zero_vec);
+   }
+
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan) {
+      /* fetch src operands */
+      int v0 = get_src_vec(gen, inst, 0, chan);
+      int v1 = get_src_vec(gen, inst, 1, chan);
+      int v2 = get_dst_vec(gen, inst, chan);
+
+      /* emit binop */
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_ADD:
+         ppc_vaddfp(gen->f, v2, v0, v1);
+         break;
+      case TGSI_OPCODE_SUB:
+         ppc_vsubfp(gen->f, v2, v0, v1);
+         break;
+      case TGSI_OPCODE_MUL:
+         ppc_vmaddfp(gen->f, v2, v0, v1, zero_vec);
+         break;
+      case TGSI_OPCODE_MIN:
+         ppc_vminfp(gen->f, v2, v0, v1);
+         break;
+      case TGSI_OPCODE_MAX:
+         ppc_vmaxfp(gen->f, v2, v0, v1);
+         break;
+      default:
+         assert(0);
+      }
+
+      /* store v2 */
+      emit_store(gen, v2, inst, chan, TRUE);
+   }
+
+   if (inst->Instruction.Opcode == TGSI_OPCODE_MUL)
+      ppc_release_vec_register(gen->f, zero_vec);
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_triop(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   uint chan;
+
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan) {
+      /* fetch src operands */
+      int v0 = get_src_vec(gen, inst, 0, chan);
+      int v1 = get_src_vec(gen, inst, 1, chan);
+      int v2 = get_src_vec(gen, inst, 2, chan);
+      int v3 = get_dst_vec(gen, inst, chan);
+
+      /* emit ALU */
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_MAD:
+         ppc_vmaddfp(gen->f, v3, v0, v1, v2);   /* v3 = v0 * v1 + v2 */
+         break;
+      case TGSI_OPCODE_LRP:
+         ppc_vsubfp(gen->f, v3, v1, v2);        /* v3 = v1 - v2 */
+         ppc_vmaddfp(gen->f, v3, v0, v3, v2);   /* v3 = v0 * v3 + v2 */
+         break;
+      default:
+         assert(0);
+      }
+
+      /* store v3 */
+      emit_store(gen, v3, inst, chan, TRUE);
+   }
+
+   release_src_vecs(gen);
+}
+
+
+/**
+ * Vector comparisons, resulting in 1.0 or 0.0 values.
+ */
+static void
+emit_inequality(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   uint chan;
+   int one_vec = gen_one_vec(gen);
+
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan) {
+      /* fetch src operands */
+      int v0 = get_src_vec(gen, inst, 0, chan);
+      int v1 = get_src_vec(gen, inst, 1, chan);
+      int v2 = get_dst_vec(gen, inst, chan);
+      boolean complement = FALSE;
+
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_SNE:
+         complement = TRUE;
+         /* fall-through */
+      case TGSI_OPCODE_SEQ:
+         ppc_vcmpeqfpx(gen->f, v2, v0, v1); /* v2 = v0 == v1 ? ~0 : 0 */
+         break;
+
+      case TGSI_OPCODE_SGE:
+         complement = TRUE;
+         /* fall-through */
+      case TGSI_OPCODE_SLT:
+         ppc_vcmpgtfpx(gen->f, v2, v1, v0); /* v2 = v1 > v0 ? ~0 : 0 */
+         break;
+
+      case TGSI_OPCODE_SLE:
+         complement = TRUE;
+         /* fall-through */
+      case TGSI_OPCODE_SGT:
+         ppc_vcmpgtfpx(gen->f, v2, v0, v1); /* v2 = v0 > v1 ? ~0 : 0 */
+         break;
+      default:
+         assert(0);
+      }
+
+      /* v2 is now {0,0,0,0} or {~0,~0,~0,~0} */
+
+      if (complement)
+         ppc_vandc(gen->f, v2, one_vec, v2);    /* v2 = one_vec & ~v2 */
+      else
+         ppc_vand(gen->f, v2, one_vec, v2);     /* v2 = one_vec & v2 */
+
+      /* store v2 */
+      emit_store(gen, v2, inst, chan, TRUE);
+   }
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_dotprod(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int v0, v1, v2;
+   uint chan_index;
+
+   v2 = ppc_allocate_vec_register(gen->f);
+
+   ppc_vxor(gen->f, v2, v2, v2);           /* v2 = {0, 0, 0, 0} */
+
+   v0 = get_src_vec(gen, inst, 0, CHAN_X); /* v0 = src0.XXXX */
+   v1 = get_src_vec(gen, inst, 1, CHAN_X); /* v1 = src1.XXXX */
+   ppc_vmaddfp(gen->f, v2, v0, v1, v2);    /* v2 = v0 * v1 + v2 */
+
+   v0 = get_src_vec(gen, inst, 0, CHAN_Y); /* v0 = src0.YYYY */
+   v1 = get_src_vec(gen, inst, 1, CHAN_Y); /* v1 = src1.YYYY */
+   ppc_vmaddfp(gen->f, v2, v0, v1, v2);    /* v2 = v0 * v1 + v2 */
+
+   v0 = get_src_vec(gen, inst, 0, CHAN_Z); /* v0 = src0.ZZZZ */
+   v1 = get_src_vec(gen, inst, 1, CHAN_Z); /* v1 = src1.ZZZZ */
+   ppc_vmaddfp(gen->f, v2, v0, v1, v2);    /* v2 = v0 * v1 + v2 */
+
+   if (inst->Instruction.Opcode == TGSI_OPCODE_DP4) {
+      v0 = get_src_vec(gen, inst, 0, CHAN_W); /* v0 = src0.WWWW */
+      v1 = get_src_vec(gen, inst, 1, CHAN_W); /* v1 = src1.WWWW */
+      ppc_vmaddfp(gen->f, v2, v0, v1, v2);    /* v2 = v0 * v1 + v2 */
+   }
+   else if (inst->Instruction.Opcode == TGSI_OPCODE_DPH) {
+      v1 = get_src_vec(gen, inst, 1, CHAN_W); /* v1 = src1.WWWW */
+      ppc_vaddfp(gen->f, v2, v2, v1);         /* v2 = v2 + v1 */
+   }
+
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan_index) {
+      emit_store(gen, v2, inst, chan_index, FALSE);  /* store v2, free v2 later */
+   }
+
+   release_src_vecs(gen);
+
+   ppc_release_vec_register(gen->f, v2);
+}
+
+
+/** Approximation for vr = pow(va, vb) */
+static void
+ppc_vec_pow(struct ppc_function *f, int vr, int va, int vb)
+{
+   /* pow(a,b) ~= exp2(log2(a) * b) */
+   int t_vec = ppc_allocate_vec_register(f);
+   int zero_vec = ppc_allocate_vec_register(f);
+
+   ppc_vzero(f, zero_vec);
+
+   ppc_vlogefp(f, t_vec, va);                   /* t = log2(va) */
+   ppc_vmaddfp(f, t_vec, t_vec, vb, zero_vec);  /* t = t * vb */
+   ppc_vexptefp(f, vr, t_vec);                  /* vr = 2^t */
+
+   ppc_release_vec_register(f, t_vec);
+   ppc_release_vec_register(f, zero_vec);
+}
+
+
+static void
+emit_lit(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int one_vec = gen_one_vec(gen);
+
+   /* Compute X */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X)) {
+      emit_store(gen, one_vec, inst, CHAN_X, FALSE);
+   }
+
+   /* Compute Y, Z */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y) ||
+       IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+      int x_vec;
+      int zero_vec = ppc_allocate_vec_register(gen->f);
+
+      x_vec = get_src_vec(gen, inst, 0, CHAN_X);  /* x_vec = src[0].x */
+
+      ppc_vzero(gen->f, zero_vec);                /* zero = {0,0,0,0} */
+      ppc_vmaxfp(gen->f, x_vec, x_vec, zero_vec); /* x_vec = max(x_vec, 0) */
+
+      if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+         emit_store(gen, x_vec, inst, CHAN_Y, FALSE);
+      }
+
+      if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+         int y_vec, w_vec;
+         int z_vec = ppc_allocate_vec_register(gen->f);
+         int pow_vec = ppc_allocate_vec_register(gen->f);
+         int pos_vec = ppc_allocate_vec_register(gen->f);
+         int p128_vec = ppc_allocate_vec_register(gen->f);
+         int n128_vec = ppc_allocate_vec_register(gen->f);
+
+         y_vec = get_src_vec(gen, inst, 0, CHAN_Y);  /* y_vec = src[0].y */
+         ppc_vmaxfp(gen->f, y_vec, y_vec, zero_vec); /* y_vec = max(y_vec, 0) */
+
+         w_vec = get_src_vec(gen, inst, 0, CHAN_W);  /* w_vec = src[0].w */
+
+         /* clamp W to [-128, 128] */
+         load_constant_vec(gen, p128_vec, 128.0f);
+         load_constant_vec(gen, n128_vec, -128.0f);
+         ppc_vmaxfp(gen->f, w_vec, w_vec, n128_vec); /* w = max(w, -128) */
+         ppc_vminfp(gen->f, w_vec, w_vec, p128_vec); /* w = min(w, 128) */
+
+         /* if temp.x > 0
+          *    z = pow(tmp.y, tmp.w)
+          * else
+          *    z = 0.0
+          */
+         ppc_vec_pow(gen->f, pow_vec, y_vec, w_vec);      /* pow = pow(y, w) */
+         ppc_vcmpgtfpx(gen->f, pos_vec, x_vec, zero_vec); /* pos = x > 0 */
+         ppc_vand(gen->f, z_vec, pow_vec, pos_vec);       /* z = pow & pos */
+
+         emit_store(gen, z_vec, inst, CHAN_Z, FALSE);
+
+         ppc_release_vec_register(gen->f, z_vec);
+         ppc_release_vec_register(gen->f, pow_vec);
+         ppc_release_vec_register(gen->f, pos_vec);
+         ppc_release_vec_register(gen->f, p128_vec);
+         ppc_release_vec_register(gen->f, n128_vec);
+      }
+
+      ppc_release_vec_register(gen->f, zero_vec);
+   }
+
+   /* Compute W */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_W)) {
+      emit_store(gen, one_vec, inst, CHAN_W, FALSE);
+   }
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_exp(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   const int one_vec = gen_one_vec(gen);
+   int src_vec;
+
+   /* get src arg */
+   src_vec = get_src_vec(gen, inst, 0, CHAN_X);
+
+   /* Compute X = 2^floor(src) */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X)) {
+      int dst_vec = get_dst_vec(gen, inst, CHAN_X);
+      int tmp_vec = ppc_allocate_vec_register(gen->f);
+      ppc_vrfim(gen->f, tmp_vec, src_vec);             /* tmp = floor(src); */
+      ppc_vexptefp(gen->f, dst_vec, tmp_vec);          /* dst = 2 ^ tmp */
+      emit_store(gen, dst_vec, inst, CHAN_X, TRUE);
+      ppc_release_vec_register(gen->f, tmp_vec);
+   }
+
+   /* Compute Y = src - floor(src) */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+      int dst_vec = get_dst_vec(gen, inst, CHAN_Y);
+      int tmp_vec = ppc_allocate_vec_register(gen->f);
+      ppc_vrfim(gen->f, tmp_vec, src_vec);             /* tmp = floor(src); */
+      ppc_vsubfp(gen->f, dst_vec, src_vec, tmp_vec);   /* dst = src - tmp */
+      emit_store(gen, dst_vec, inst, CHAN_Y, TRUE);
+      ppc_release_vec_register(gen->f, tmp_vec);
+   }
+
+   /* Compute Z = RoughApprox2ToX(src) */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+      int dst_vec = get_dst_vec(gen, inst, CHAN_Z);
+      ppc_vexptefp(gen->f, dst_vec, src_vec);          /* dst = 2 ^ src */
+      emit_store(gen, dst_vec, inst, CHAN_Z, TRUE);
+   }
+
+   /* Compute W = 1.0 */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_W)) {
+      emit_store(gen, one_vec, inst, CHAN_W, FALSE);
+   }
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_log(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   const int bit31_vec = gen_get_bit31_vec(gen);
+   const int one_vec = gen_one_vec(gen);
+   int src_vec, abs_vec;
+
+   /* get src arg */
+   src_vec = get_src_vec(gen, inst, 0, CHAN_X);
+
+   /* compute abs(src) */
+   abs_vec = ppc_allocate_vec_register(gen->f);
+   ppc_vandc(gen->f, abs_vec, src_vec, bit31_vec);     /* abs = src & ~bit31 */
+
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X) &&
+       IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+
+      /* compute tmp = floor(log2(abs)) */
+      int tmp_vec = ppc_allocate_vec_register(gen->f);
+      ppc_vlogefp(gen->f, tmp_vec, abs_vec);           /* tmp = log2(abs) */
+      ppc_vrfim(gen->f, tmp_vec, tmp_vec);             /* tmp = floor(tmp); */
+
+      /* Compute X = tmp */
+      if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X)) {
+         emit_store(gen, tmp_vec, inst, CHAN_X, FALSE);
+      }
+      
+      /* Compute Y = abs / 2^tmp */
+      if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+         const int zero_vec = ppc_allocate_vec_register(gen->f);
+         ppc_vzero(gen->f, zero_vec);
+         ppc_vexptefp(gen->f, tmp_vec, tmp_vec);       /* tmp = 2 ^ tmp */
+         ppc_vrefp(gen->f, tmp_vec, tmp_vec);          /* tmp = 1 / tmp */
+         /* tmp = abs * tmp + zero */
+         ppc_vmaddfp(gen->f, tmp_vec, abs_vec, tmp_vec, zero_vec);
+         emit_store(gen, tmp_vec, inst, CHAN_Y, FALSE);
+         ppc_release_vec_register(gen->f, zero_vec);
+      }
+
+      ppc_release_vec_register(gen->f, tmp_vec);
+   }
+
+   /* Compute Z = RoughApproxLog2(abs) */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+      int dst_vec = get_dst_vec(gen, inst, CHAN_Z);
+      ppc_vlogefp(gen->f, dst_vec, abs_vec);           /* dst = log2(abs) */
+      emit_store(gen, dst_vec, inst, CHAN_Z, TRUE);
+   }
+
+   /* Compute W = 1.0 */
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_W)) {
+      emit_store(gen, one_vec, inst, CHAN_W, FALSE);
+   }
+
+   ppc_release_vec_register(gen->f, abs_vec);
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_pow(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int s0_vec = get_src_vec(gen, inst, 0, CHAN_X);
+   int s1_vec = get_src_vec(gen, inst, 1, CHAN_X);
+   int pow_vec = ppc_allocate_vec_register(gen->f);
+   int chan;
+
+   ppc_vec_pow(gen->f, pow_vec, s0_vec, s1_vec);
+
+   FOR_EACH_DST0_ENABLED_CHANNEL(*inst, chan) {
+      emit_store(gen, pow_vec, inst, chan, FALSE);
+   }
+
+   ppc_release_vec_register(gen->f, pow_vec);
+
+   release_src_vecs(gen);
+}
+
+
+static void
+emit_xpd(struct gen_context *gen, struct tgsi_full_instruction *inst)
+{
+   int x0_vec, y0_vec, z0_vec;
+   int x1_vec, y1_vec, z1_vec;
+   int zero_vec, tmp_vec;
+   int tmp2_vec;
+
+   zero_vec = ppc_allocate_vec_register(gen->f);
+   ppc_vzero(gen->f, zero_vec);
+
+   tmp_vec = ppc_allocate_vec_register(gen->f);
+   tmp2_vec = ppc_allocate_vec_register(gen->f);
+
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y) ||
+       IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+      x0_vec = get_src_vec(gen, inst, 0, CHAN_X);
+      x1_vec = get_src_vec(gen, inst, 1, CHAN_X);
+   }
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X) ||
+       IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+      y0_vec = get_src_vec(gen, inst, 0, CHAN_Y);
+      y1_vec = get_src_vec(gen, inst, 1, CHAN_Y);
+   }
+   if (IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X) ||
+       IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+      z0_vec = get_src_vec(gen, inst, 0, CHAN_Z);
+      z1_vec = get_src_vec(gen, inst, 1, CHAN_Z);
+   }
+
+   IF_IS_DST0_CHANNEL_ENABLED(*inst, CHAN_X) {
+      /* tmp = y0 * z1 */
+      ppc_vmaddfp(gen->f, tmp_vec, y0_vec, z1_vec, zero_vec);
+      /* tmp = tmp - z0 * y1*/
+      ppc_vnmsubfp(gen->f, tmp_vec, tmp_vec, z0_vec, y1_vec);
+      emit_store(gen, tmp_vec, inst, CHAN_X, FALSE);
+   }
+   IF_IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Y) {
+      /* tmp = z0 * x1 */
+      ppc_vmaddfp(gen->f, tmp_vec, z0_vec, x1_vec, zero_vec);
+      /* tmp = tmp - x0 * z1 */
+      ppc_vnmsubfp(gen->f, tmp_vec, tmp_vec, x0_vec, z1_vec);
+      emit_store(gen, tmp_vec, inst, CHAN_Y, FALSE);
+   }
+   IF_IS_DST0_CHANNEL_ENABLED(*inst, CHAN_Z) {
+      /* tmp = x0 * y1 */
+      ppc_vmaddfp(gen->f, tmp_vec, x0_vec, y1_vec, zero_vec);
+      /* tmp = tmp - y0 * x1 */
+      ppc_vnmsubfp(gen->f, tmp_vec, tmp_vec, y0_vec, x1_vec);
+      emit_store(gen, tmp_vec, inst, CHAN_Z, FALSE);
+   }
+   /* W is undefined */
+
+   ppc_release_vec_register(gen->f, tmp_vec);
+   ppc_release_vec_register(gen->f, zero_vec);
+   release_src_vecs(gen);
+}
+
+static int
+emit_instruction(struct gen_context *gen,
+                 struct tgsi_full_instruction *inst)
+{
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_MOV:
+   case TGSI_OPCODE_SWZ:
+   case TGSI_OPCODE_ABS:
+   case TGSI_OPCODE_FLOOR:
+   case TGSI_OPCODE_FRAC:
+   case TGSI_OPCODE_EXPBASE2:
+   case TGSI_OPCODE_LOGBASE2:
+      emit_unaryop(gen, inst);
+      break;
+   case TGSI_OPCODE_RSQ:
+   case TGSI_OPCODE_RCP:
+      emit_scalar_unaryop(gen, inst);
+      break;
+   case TGSI_OPCODE_ADD:
+   case TGSI_OPCODE_SUB:
+   case TGSI_OPCODE_MUL:
+   case TGSI_OPCODE_MIN:
+   case TGSI_OPCODE_MAX:
+      emit_binop(gen, inst);
+      break;
+   case TGSI_OPCODE_SEQ:
+   case TGSI_OPCODE_SNE:
+   case TGSI_OPCODE_SLT:
+   case TGSI_OPCODE_SGT:
+   case TGSI_OPCODE_SLE:
+   case TGSI_OPCODE_SGE:
+      emit_inequality(gen, inst);
+      break;
+   case TGSI_OPCODE_MAD:
+   case TGSI_OPCODE_LRP:
+      emit_triop(gen, inst);
+      break;
+   case TGSI_OPCODE_DP3:
+   case TGSI_OPCODE_DP4:
+   case TGSI_OPCODE_DPH:
+      emit_dotprod(gen, inst);
+      break;
+   case TGSI_OPCODE_LIT:
+      emit_lit(gen, inst);
+      break;
+   case TGSI_OPCODE_LOG:
+      emit_log(gen, inst);
+      break;
+   case TGSI_OPCODE_EXP:
+      emit_exp(gen, inst);
+      break;
+   case TGSI_OPCODE_POW:
+      emit_pow(gen, inst);
+      break;
+   case TGSI_OPCODE_XPD:
+      emit_xpd(gen, inst);
+      break;
+   case TGSI_OPCODE_END:
+      /* normal end */
+      return 1;
+   default:
+      return 0;
+   }
+   return 1;
+}
+
+
+static void
+emit_declaration(
+   struct ppc_function *func,
+   struct tgsi_full_declaration *decl )
+{
+   if( decl->Declaration.File == TGSI_FILE_INPUT ) {
+#if 0
+      unsigned first, last, mask;
+      unsigned i, j;
+
+      first = decl->DeclarationRange.First;
+      last = decl->DeclarationRange.Last;
+      mask = decl->Declaration.UsageMask;
+
+      for( i = first; i <= last; i++ ) {
+         for( j = 0; j < NUM_CHANNELS; j++ ) {
+            if( mask & (1 << j) ) {
+               switch( decl->Declaration.Interpolate ) {
+               case TGSI_INTERPOLATE_CONSTANT:
+                  emit_coef_a0( func, 0, i, j );
+                  emit_inputs( func, 0, i, j );
+                  break;
+
+               case TGSI_INTERPOLATE_LINEAR:
+                  emit_tempf( func, 0, 0, TGSI_SWIZZLE_X );
+                  emit_coef_dadx( func, 1, i, j );
+                  emit_tempf( func, 2, 0, TGSI_SWIZZLE_Y );
+                  emit_coef_dady( func, 3, i, j );
+                  emit_mul( func, 0, 1 );    /* x * dadx */
+                  emit_coef_a0( func, 4, i, j );
+                  emit_mul( func, 2, 3 );    /* y * dady */
+                  emit_add( func, 0, 4 );    /* x * dadx + a0 */
+                  emit_add( func, 0, 2 );    /* x * dadx + y * dady + a0 */
+                  emit_inputs( func, 0, i, j );
+                  break;
+
+               case TGSI_INTERPOLATE_PERSPECTIVE:
+                  emit_tempf( func, 0, 0, TGSI_SWIZZLE_X );
+                  emit_coef_dadx( func, 1, i, j );
+                  emit_tempf( func, 2, 0, TGSI_SWIZZLE_Y );
+                  emit_coef_dady( func, 3, i, j );
+                  emit_mul( func, 0, 1 );    /* x * dadx */
+                  emit_tempf( func, 4, 0, TGSI_SWIZZLE_W );
+                  emit_coef_a0( func, 5, i, j );
+                  emit_rcp( func, 4, 4 );    /* 1.0 / w */
+                  emit_mul( func, 2, 3 );    /* y * dady */
+                  emit_add( func, 0, 5 );    /* x * dadx + a0 */
+                  emit_add( func, 0, 2 );    /* x * dadx + y * dady + a0 */
+                  emit_mul( func, 0, 4 );    /* (x * dadx + y * dady + a0) / w */
+                  emit_inputs( func, 0, i, j );
+                  break;
+
+               default:
+                  assert( 0 );
+		  break;
+               }
+            }
+         }
+      }
+#endif
+   }
+}
+
+
+
+static void
+emit_prologue(struct ppc_function *func)
+{
+   /* XXX set up stack frame */
+}
+
+
+static void
+emit_epilogue(struct ppc_function *func)
+{
+   ppc_return(func);
+   /* XXX restore prev stack frame */
+   debug_printf("PPC: Emitted %u instructions\n", func->num_inst);
+}
+
+
+
+/**
+ * Translate a TGSI vertex/fragment shader to PPC code.
+ *
+ * \param tokens  the TGSI input shader
+ * \param func  the output PPC code/function
+ * \param immediates  buffer to place immediates, later passed to PPC func
+ * \return TRUE for success, FALSE if translation failed
+ */
+boolean
+tgsi_emit_ppc(const struct tgsi_token *tokens,
+              struct ppc_function *func,
+              float (*immediates)[4],
+              boolean do_swizzles )
+{
+   static int use_ppc_asm = -1;
+   struct tgsi_parse_context parse;
+   /*boolean instruction_phase = FALSE;*/
+   unsigned ok = 1;
+   uint num_immediates = 0;
+   struct gen_context gen;
+
+   if (use_ppc_asm < 0) {
+      /* If GALLIUM_NOPPC is set, don't use PPC codegen */
+      use_ppc_asm = !debug_get_bool_option("GALLIUM_NOPPC", FALSE);
+   }
+   if (!use_ppc_asm)
+      return FALSE;
+
+   if (0) {
+      debug_printf("\n********* TGSI->PPC ********\n");
+      tgsi_dump(tokens, 0);
+   }
+
+   util_init_math();
+
+   init_gen_context(&gen, func);
+
+   emit_prologue(func);
+
+   tgsi_parse_init( &parse, tokens );
+
+   while (!tgsi_parse_end_of_tokens(&parse) && ok) {
+      tgsi_parse_token(&parse);
+
+      switch (parse.FullToken.Token.Type) {
+      case TGSI_TOKEN_TYPE_DECLARATION:
+         if (parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_FRAGMENT) {
+            emit_declaration(func, &parse.FullToken.FullDeclaration );
+         }
+         break;
+
+      case TGSI_TOKEN_TYPE_INSTRUCTION:
+         ok = emit_instruction(&gen, &parse.FullToken.FullInstruction);
+
+	 if (!ok) {
+	    debug_printf("failed to translate tgsi opcode %d to PPC (%s)\n", 
+			 parse.FullToken.FullInstruction.Instruction.Opcode,
+                         parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ?
+                         "vertex shader" : "fragment shader");
+	 }
+         break;
+
+      case TGSI_TOKEN_TYPE_IMMEDIATE:
+         /* splat each immediate component into a float[4] vector for SoA */
+         {
+            const uint size = parse.FullToken.FullImmediate.Immediate.Size - 1;
+            uint i;
+            assert(size <= 4);
+            assert(num_immediates < TGSI_EXEC_NUM_IMMEDIATES);
+            for (i = 0; i < size; i++) {
+               immediates[num_immediates][i] =
+		  parse.FullToken.FullImmediate.u.ImmediateFloat32[i].Float;
+            }
+            num_immediates++;
+         }
+         break;
+
+      default:
+	 ok = 0;
+         assert( 0 );
+      }
+   }
+
+   emit_epilogue(func);
+
+   tgsi_parse_free( &parse );
+
+   if (ppc_num_instructions(func) == 0) {
+      /* ran out of memory for instructions */
+      ok = FALSE;
+   }
+
+   if (!ok)
+      debug_printf("TGSI->PPC translation failed\n");
+
+   return ok;
+}
+
+#endif /* PIPE_ARCH_PPC */
diff --git a/src/mesa/drivers/dri/i965/intel_ioctl.h b/src/gallium/auxiliary/tgsi/tgsi_ppc.h
index df276593626..829ec075e7f 100644
--- a/src/mesa/drivers/dri/i965/intel_ioctl.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,20 +25,27 @@
  * 
  **************************************************************************/
 
-#ifndef INTEL_IOCTL_H
-#define INTEL_IOCTL_H
+#ifndef TGSI_PPC_H
+#define TGSI_PPC_H
 
-#include "intel_context.h"
+#if defined __cplusplus
+extern "C" {
+#endif
+
+struct tgsi_token;
+struct ppc_function;
 
-void intelWaitIrq( struct intel_context *intel, int seq );
-int intelEmitIrqLocked( struct intel_context *intel );
+extern const float ppc_builtin_constants[];
 
-void intel_batch_ioctl( struct intel_context *intel, 
-			GLuint start_offset,
-			GLuint used);
 
-void intel_cmd_ioctl( struct intel_context *intel, 
-		      char *buf,
-		      GLuint used);
+boolean
+tgsi_emit_ppc(const struct tgsi_token *tokens,
+              struct ppc_function *function,
+              float (*immediates)[4],
+              boolean do_swizzles);
 
+#if defined __cplusplus
+}
 #endif
+
+#endif /* TGSI_PPC_H */
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index ff869c8312f..cac44af7f41 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -25,9 +25,16 @@
  * 
  **************************************************************************/
 
+#include "pipe/p_config.h"
+
+#if defined(PIPE_ARCH_X86)
+
 #include "pipe/p_debug.h"
 #include "pipe/p_shader_tokens.h"
 #include "util/u_math.h"
+#if defined(PIPE_ARCH_SSE)
+#include "util/u_sse.h"
+#endif
 #include "tgsi/tgsi_parse.h"
 #include "tgsi/tgsi_util.h"
 #include "tgsi_exec.h"
@@ -35,8 +42,6 @@
 
 #include "rtasm/rtasm_x86sse.h"
 
-#ifdef PIPE_ARCH_X86
-
 /* for 1/sqrt()
  *
  * This costs about 100fps (close to 10%) in gears:
@@ -509,10 +514,31 @@ emit_coef_dady(
  * Function call helpers.
  */
 
+/**
+ * NOTE: In gcc, if the destination uses the SSE intrinsics, then it must be 
+ * defined with __attribute__((force_align_arg_pointer)), as we do not guarantee
+ * that the stack pointer is 16 byte aligned, as expected.
+ */
 static void
-emit_push_gp(
-   struct x86_function *func )
+emit_func_call_dst(
+   struct x86_function *func,
+   unsigned xmm_save,
+   unsigned xmm_dst,
+   void (PIPE_CDECL *code)() )
 {
+   struct x86_reg ecx = x86_make_reg( file_REG32, reg_CX );
+   unsigned i, n;
+   unsigned xmm_mask;
+   
+   /* Bitmask of the xmm registers to save */
+   xmm_mask = (1 << xmm_save) - 1;
+   xmm_mask &= ~(1 << xmm_dst);
+
+   sse_movaps(
+      func,
+      get_temp( TEMP_R0, 0 ),
+      make_xmm( xmm_dst ) );
+
    x86_push(
       func,
       x86_make_reg( file_REG32, reg_AX) );
@@ -522,12 +548,49 @@ emit_push_gp(
    x86_push(
       func,
       x86_make_reg( file_REG32, reg_DX) );
-}
+   
+   for(i = 0, n = 0; i < 8; ++i)
+      if(xmm_mask & (1 << i))
+         ++n;
+   
+   x86_sub_imm(
+      func, 
+      x86_make_reg( file_REG32, reg_SP ),
+      n*16);
+
+   for(i = 0, n = 0; i < 8; ++i)
+      if(xmm_mask & (1 << i)) {
+         sse_movups(
+            func,
+            x86_make_disp( x86_make_reg( file_REG32, reg_SP ), n*16 ),
+            make_xmm( i ) );
+         ++n;
+      }
+   
+   x86_lea(
+      func,
+      ecx,
+      get_temp( TEMP_R0, 0 ) );
+   
+   x86_push( func, ecx );
+   x86_mov_reg_imm( func, ecx, (unsigned long) code );
+   x86_call( func, ecx );
+   x86_pop(func, ecx );
+   
+   for(i = 0, n = 0; i < 8; ++i)
+      if(xmm_mask & (1 << i)) {
+         sse_movups(
+            func,
+            make_xmm( i ),
+            x86_make_disp( x86_make_reg( file_REG32, reg_SP ), n*16 ) );
+         ++n;
+      }
+   
+   x86_add_imm(
+      func, 
+      x86_make_reg( file_REG32, reg_SP ),
+      n*16);
 
-static void
-x86_pop_gp(
-   struct x86_function *func )
-{
    /* Restore GP registers in a reverse order.
     */
    x86_pop(
@@ -539,39 +602,6 @@ x86_pop_gp(
    x86_pop(
       func,
       x86_make_reg( file_REG32, reg_AX) );
-}
-
-static void
-emit_func_call_dst(
-   struct x86_function *func,
-   unsigned xmm_dst,
-   void (PIPE_CDECL *code)() )
-{
-   sse_movaps(
-      func,
-      get_temp( TEMP_R0, 0 ),
-      make_xmm( xmm_dst ) );
-
-   emit_push_gp(
-      func );
-
-   {
-      struct x86_reg ecx = x86_make_reg( file_REG32, reg_CX );
-
-      x86_lea(
-         func,
-         ecx,
-         get_temp( TEMP_R0, 0 ) );
-
-      x86_push( func, ecx );
-      x86_mov_reg_imm( func, ecx, (unsigned long) code );
-      x86_call( func, ecx );
-      x86_pop(func, ecx ); 
-   }
-
-
-   x86_pop_gp(
-      func );
 
    sse_movaps(
       func,
@@ -582,6 +612,7 @@ emit_func_call_dst(
 static void
 emit_func_call_dst_src(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst,
    unsigned xmm_src,
    void (PIPE_CDECL *code)() )
@@ -593,10 +624,119 @@ emit_func_call_dst_src(
 
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       code );
 }
 
+
+#if defined(PIPE_ARCH_SSE)
+
+/*
+ * Fast SSE2 implementation of special math functions.
+ */
+
+#define POLY0(x, c0) _mm_set1_ps(c0)
+#define POLY1(x, c0, c1) _mm_add_ps(_mm_mul_ps(POLY0(x, c1), x), _mm_set1_ps(c0))
+#define POLY2(x, c0, c1, c2) _mm_add_ps(_mm_mul_ps(POLY1(x, c1, c2), x), _mm_set1_ps(c0))
+#define POLY3(x, c0, c1, c2, c3) _mm_add_ps(_mm_mul_ps(POLY2(x, c1, c2, c3), x), _mm_set1_ps(c0))
+#define POLY4(x, c0, c1, c2, c3, c4) _mm_add_ps(_mm_mul_ps(POLY3(x, c1, c2, c3, c4), x), _mm_set1_ps(c0))
+#define POLY5(x, c0, c1, c2, c3, c4, c5) _mm_add_ps(_mm_mul_ps(POLY4(x, c1, c2, c3, c4, c5), x), _mm_set1_ps(c0))
+
+#define EXP_POLY_DEGREE 3
+#define LOG_POLY_DEGREE 5
+
+/**
+ * See http://www.devmaster.net/forums/showthread.php?p=43580
+ */
+static INLINE __m128 
+exp2f4(__m128 x)
+{
+   __m128i ipart;
+   __m128 fpart, expipart, expfpart;
+
+   x = _mm_min_ps(x, _mm_set1_ps( 129.00000f));
+   x = _mm_max_ps(x, _mm_set1_ps(-126.99999f));
+
+   /* ipart = int(x - 0.5) */
+   ipart = _mm_cvtps_epi32(_mm_sub_ps(x, _mm_set1_ps(0.5f)));
+
+   /* fpart = x - ipart */
+   fpart = _mm_sub_ps(x, _mm_cvtepi32_ps(ipart));
+
+   /* expipart = (float) (1 << ipart) */
+   expipart = _mm_castsi128_ps(_mm_slli_epi32(_mm_add_epi32(ipart, _mm_set1_epi32(127)), 23));
+
+   /* minimax polynomial fit of 2**x, in range [-0.5, 0.5[ */
+#if EXP_POLY_DEGREE == 5
+   expfpart = POLY5(fpart, 9.9999994e-1f, 6.9315308e-1f, 2.4015361e-1f, 5.5826318e-2f, 8.9893397e-3f, 1.8775767e-3f);
+#elif EXP_POLY_DEGREE == 4
+   expfpart = POLY4(fpart, 1.0000026f, 6.9300383e-1f, 2.4144275e-1f, 5.2011464e-2f, 1.3534167e-2f);
+#elif EXP_POLY_DEGREE == 3
+   expfpart = POLY3(fpart, 9.9992520e-1f, 6.9583356e-1f, 2.2606716e-1f, 7.8024521e-2f);
+#elif EXP_POLY_DEGREE == 2
+   expfpart = POLY2(fpart, 1.0017247f, 6.5763628e-1f, 3.3718944e-1f);
+#else
+#error
+#endif
+
+   return _mm_mul_ps(expipart, expfpart);
+}
+
+
+/**
+ * See http://www.devmaster.net/forums/showthread.php?p=43580
+ */
+static INLINE __m128 
+log2f4(__m128 x)
+{
+   __m128i expmask = _mm_set1_epi32(0x7f800000);
+   __m128i mantmask = _mm_set1_epi32(0x007fffff);
+   __m128 one = _mm_set1_ps(1.0f);
+
+   __m128i i = _mm_castps_si128(x);
+
+   /* exp = (float) exponent(x) */
+   __m128 exp = _mm_cvtepi32_ps(_mm_sub_epi32(_mm_srli_epi32(_mm_and_si128(i, expmask), 23), _mm_set1_epi32(127)));
+
+   /* mant = (float) mantissa(x) */
+   __m128 mant = _mm_or_ps(_mm_castsi128_ps(_mm_and_si128(i, mantmask)), one);
+
+   __m128 logmant;
+
+   /* Minimax polynomial fit of log2(x)/(x - 1), for x in range [1, 2[ 
+    * These coefficients can be generate with 
+    * http://www.boost.org/doc/libs/1_36_0/libs/math/doc/sf_and_dist/html/math_toolkit/toolkit/internals2/minimax.html
+    */
+#if LOG_POLY_DEGREE == 6
+   logmant = POLY5(mant, 3.11578814719469302614f, -3.32419399085241980044f, 2.59883907202499966007f, -1.23152682416275988241f, 0.318212422185251071475f, -0.0344359067839062357313f);
+#elif LOG_POLY_DEGREE == 5
+   logmant = POLY4(mant, 2.8882704548164776201f, -2.52074962577807006663f, 1.48116647521213171641f, -0.465725644288844778798f, 0.0596515482674574969533f);
+#elif LOG_POLY_DEGREE == 4
+   logmant = POLY3(mant, 2.61761038894603480148f, -1.75647175389045657003f, 0.688243882994381274313f, -0.107254423828329604454f);
+#elif LOG_POLY_DEGREE == 3
+   logmant = POLY2(mant, 2.28330284476918490682f, -1.04913055217340124191f, 0.204446009836232697516f);
+#else
+#error
+#endif
+
+   /* This effectively increases the polynomial degree by one, but ensures that log2(1) == 0*/
+   logmant = _mm_mul_ps(logmant, _mm_sub_ps(mant, one));
+
+   return _mm_add_ps(logmant, exp);
+}
+
+
+static INLINE __m128
+powf4(__m128 x, __m128 y)
+{
+   return exp2f4(_mm_mul_ps(log2f4(x), y));
+}
+
+#endif /* PIPE_ARCH_SSE */
+
+
+
 /**
  * Low-level instruction translators.
  */
@@ -639,38 +779,42 @@ cos4f(
 static void
 emit_cos(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save, 
       xmm_dst,
       cos4f );
 }
 
 static void PIPE_CDECL
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_SSE)
+__attribute__((force_align_arg_pointer))
+#endif
 ex24f(
    float *store )
 {
-#if FAST_MATH
+#if defined(PIPE_ARCH_SSE)
+   _mm_store_ps(&store[0], exp2f4( _mm_load_ps(&store[0]) ));
+#else
    store[0] = util_fast_exp2( store[0] );
    store[1] = util_fast_exp2( store[1] );
    store[2] = util_fast_exp2( store[2] );
    store[3] = util_fast_exp2( store[3] );
-#else
-   store[0] = powf( 2.0f, store[0] );
-   store[1] = powf( 2.0f, store[1] );
-   store[2] = powf( 2.0f, store[2] );
-   store[3] = powf( 2.0f, store[3] );
 #endif
 }
 
 static void
 emit_ex2(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       ex24f );
 }
@@ -710,10 +854,12 @@ flr4f(
 static void
 emit_flr(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       flr4f );
 }
@@ -731,31 +877,42 @@ frc4f(
 static void
 emit_frc(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       frc4f );
 }
 
 static void PIPE_CDECL
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_SSE)
+__attribute__((force_align_arg_pointer))
+#endif
 lg24f(
    float *store )
 {
+#if defined(PIPE_ARCH_SSE)
+   _mm_store_ps(&store[0], log2f4( _mm_load_ps(&store[0]) ));
+#else
    store[0] = util_fast_log2( store[0] );
    store[1] = util_fast_log2( store[1] );
    store[2] = util_fast_log2( store[2] );
    store[3] = util_fast_log2( store[3] );
+#endif
 }
 
 static void
 emit_lg2(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       lg24f );
 }
@@ -797,30 +954,32 @@ emit_neg(
 }
 
 static void PIPE_CDECL
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_SSE)
+__attribute__((force_align_arg_pointer))
+#endif
 pow4f(
    float *store )
 {
-#if FAST_MATH
+#if defined(PIPE_ARCH_SSE)
+   _mm_store_ps(&store[0], powf4( _mm_load_ps(&store[0]), _mm_load_ps(&store[4]) ));
+#else
    store[0] = util_fast_pow( store[0], store[4] );
    store[1] = util_fast_pow( store[1], store[5] );
    store[2] = util_fast_pow( store[2], store[6] );
    store[3] = util_fast_pow( store[3], store[7] );
-#else
-   store[0] = powf( store[0], store[4] );
-   store[1] = powf( store[1], store[5] );
-   store[2] = powf( store[2], store[6] );
-   store[3] = powf( store[3], store[7] );
 #endif
 }
 
 static void
 emit_pow(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst,
    unsigned xmm_src )
 {
    emit_func_call_dst_src(
       func,
+      xmm_save,
       xmm_dst,
       xmm_src,
       pow4f );
@@ -855,10 +1014,12 @@ rnd4f(
 static void
 emit_rnd(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       rnd4f );
 }
@@ -935,10 +1096,12 @@ sgn4f(
 static void
 emit_sgn(
    struct x86_function *func,
+   unsigned xmm_save, 
    unsigned xmm_dst )
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       sgn4f );
 }
@@ -955,10 +1118,12 @@ sin4f(
 
 static void
 emit_sin (struct x86_function *func,
+          unsigned xmm_save, 
           unsigned xmm_dst)
 {
    emit_func_call_dst(
       func,
+      xmm_save,
       xmm_dst,
       sin4f );
 }
@@ -1378,7 +1543,7 @@ emit_instruction(
                get_temp(
                   TGSI_EXEC_TEMP_MINUS_128_I,
                   TGSI_EXEC_TEMP_MINUS_128_C ) );
-            emit_pow( func, 1, 2 );
+            emit_pow( func, 3, 1, 2 );
             FETCH( func, *inst, 0, 0, CHAN_X );
             sse_xorps(
                func,
@@ -1424,11 +1589,11 @@ emit_instruction(
          if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) ||
              IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y )) {
             emit_MOV( func, 1, 0 );
-            emit_flr( func, 1 );
+            emit_flr( func, 2, 1 );
             /* dst.x = ex2(floor(src.x)) */
             if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X )) {
                emit_MOV( func, 2, 1 );
-               emit_ex2( func, 2 );
+               emit_ex2( func, 3, 2 );
                STORE( func, *inst, 2, 0, CHAN_X );
             }
             /* dst.y = src.x - floor(src.x) */
@@ -1440,7 +1605,7 @@ emit_instruction(
          }
          /* dst.z = ex2(src.x) */
          if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z )) {
-            emit_ex2( func, 0 );
+            emit_ex2( func, 3, 0 );
             STORE( func, *inst, 0, 0, CHAN_Z );
          }
       }
@@ -1458,21 +1623,21 @@ emit_instruction(
          FETCH( func, *inst, 0, 0, CHAN_X );
          emit_abs( func, 0 );
          emit_MOV( func, 1, 0 );
-         emit_lg2( func, 1 );
+         emit_lg2( func, 2, 1 );
          /* dst.z = lg2(abs(src.x)) */
          if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z )) {
             STORE( func, *inst, 1, 0, CHAN_Z );
          }
          if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) ||
              IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y )) {
-            emit_flr( func, 1 );
+            emit_flr( func, 2, 1 );
             /* dst.x = floor(lg2(abs(src.x))) */
             if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X )) {
                STORE( func, *inst, 1, 0, CHAN_X );
             }
             /* dst.x = abs(src)/ex2(floor(lg2(abs(src.x)))) */
             if (IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y )) {
-               emit_ex2( func, 1 );
+               emit_ex2( func, 2, 1 );
                emit_rcp( func, 1, 1 );
                emit_mul( func, 0, 1 );
                STORE( func, *inst, 0, 0, CHAN_Y );
@@ -1647,7 +1812,18 @@ emit_instruction(
 
    case TGSI_OPCODE_DOT2ADD:
    /* TGSI_OPCODE_DP2A */
-      return 0;
+      FETCH( func, *inst, 0, 0, CHAN_X );  /* xmm0 = src[0].x */
+      FETCH( func, *inst, 1, 1, CHAN_X );  /* xmm1 = src[1].x */
+      emit_mul( func, 0, 1 );              /* xmm0 = xmm0 * xmm1 */
+      FETCH( func, *inst, 1, 0, CHAN_Y );  /* xmm1 = src[0].y */
+      FETCH( func, *inst, 2, 1, CHAN_Y );  /* xmm2 = src[1].y */
+      emit_mul( func, 1, 2 );              /* xmm1 = xmm1 * xmm2 */
+      emit_add( func, 0, 1 );              /* xmm0 = xmm0 + xmm1 */
+      FETCH( func, *inst, 1, 2, CHAN_X );  /* xmm1 = src[2].x */
+      emit_add( func, 0, 1 );              /* xmm0 = xmm0 + xmm1 */
+      FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
+         STORE( func, *inst, 0, 0, chan_index );  /* dest[ch] = xmm0 */
+      }
       break;
 
    case TGSI_OPCODE_INDEX:
@@ -1662,7 +1838,7 @@ emit_instruction(
    /* TGSI_OPCODE_FRC */
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
-         emit_frc( func, 0 );
+         emit_frc( func, 0, 0 );
          STORE( func, *inst, 0, 0, chan_index );
       }
       break;
@@ -1675,7 +1851,7 @@ emit_instruction(
    /* TGSI_OPCODE_FLR */
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
-         emit_flr( func, 0 );
+         emit_flr( func, 0, 0 );
          STORE( func, *inst, 0, 0, chan_index );
       }
       break;
@@ -1683,7 +1859,7 @@ emit_instruction(
    case TGSI_OPCODE_ROUND:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
-         emit_rnd( func, 0 );
+         emit_rnd( func, 0, 0 );
          STORE( func, *inst, 0, 0, chan_index );
       }
       break;
@@ -1691,7 +1867,7 @@ emit_instruction(
    case TGSI_OPCODE_EXPBASE2:
    /* TGSI_OPCODE_EX2 */
       FETCH( func, *inst, 0, 0, CHAN_X );
-      emit_ex2( func, 0 );
+      emit_ex2( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1700,7 +1876,7 @@ emit_instruction(
    case TGSI_OPCODE_LOGBASE2:
    /* TGSI_OPCODE_LG2 */
       FETCH( func, *inst, 0, 0, CHAN_X );
-      emit_lg2( func, 0 );
+      emit_lg2( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1710,7 +1886,7 @@ emit_instruction(
    /* TGSI_OPCODE_POW */
       FETCH( func, *inst, 0, 0, CHAN_X );
       FETCH( func, *inst, 1, 1, CHAN_X );
-      emit_pow( func, 0, 1 );
+      emit_pow( func, 0, 0, 1 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1801,7 +1977,7 @@ emit_instruction(
 
    case TGSI_OPCODE_COS:
       FETCH( func, *inst, 0, 0, CHAN_X );
-      emit_cos( func, 0 );
+      emit_cos( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1860,7 +2036,7 @@ emit_instruction(
 
    case TGSI_OPCODE_SIN:
       FETCH( func, *inst, 0, 0, CHAN_X );
-      emit_sin( func, 0 );
+      emit_sin( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1927,7 +2103,7 @@ emit_instruction(
    case TGSI_OPCODE_ARR:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
-         emit_rnd( func, 0 );
+         emit_rnd( func, 0, 0 );
          emit_f2it( func, 0 );
          STORE( func, *inst, 0, 0, chan_index );
       }
@@ -1952,7 +2128,7 @@ emit_instruction(
    /* TGSI_OPCODE_SGN */
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
-         emit_sgn( func, 0 );
+         emit_sgn( func, 0, 0 );
          STORE( func, *inst, 0, 0, chan_index );
       }
       break;
@@ -1964,12 +2140,12 @@ emit_instruction(
    case TGSI_OPCODE_SCS:
       IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) {
          FETCH( func, *inst, 0, 0, CHAN_X );
-         emit_cos( func, 0 );
+         emit_cos( func, 0, 0 );
          STORE( func, *inst, 0, 0, CHAN_X );
       }
       IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) {
          FETCH( func, *inst, 0, 0, CHAN_X );
-         emit_sin( func, 0 );
+         emit_sin( func, 0, 0 );
          STORE( func, *inst, 0, 0, CHAN_Y );
       }
       IF_IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Z ) {
@@ -1995,7 +2171,39 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_NRM:
-      return 0;
+      /* fall-through */
+   case TGSI_OPCODE_NRM4:
+      /* 3 or 4-component normalization */
+      {
+         uint dims = (inst->Instruction.Opcode == TGSI_OPCODE_NRM) ? 3 : 4;
+         /* note: cannot use xmm regs 2/3 here (see emit_rsqrt() above) */
+         FETCH( func, *inst, 4, 0, CHAN_X );    /* xmm4 = src[0].x */
+         FETCH( func, *inst, 5, 0, CHAN_Y );    /* xmm5 = src[0].y */
+         FETCH( func, *inst, 6, 0, CHAN_Z );    /* xmm6 = src[0].z */
+         if (dims == 4) {
+            FETCH( func, *inst, 7, 0, CHAN_W ); /* xmm7 = src[0].w */
+         }
+         emit_MOV( func, 0, 4 );                /* xmm0 = xmm3 */
+         emit_mul( func, 0, 4 );                /* xmm0 *= xmm3 */
+         emit_MOV( func, 1, 5 );                /* xmm1 = xmm4 */
+         emit_mul( func, 1, 5 );                /* xmm1 *= xmm4 */
+         emit_add( func, 0, 1 );                /* xmm0 += xmm1 */
+         emit_MOV( func, 1, 6 );                /* xmm1 = xmm5 */
+         emit_mul( func, 1, 6 );                /* xmm1 *= xmm5 */
+         emit_add( func, 0, 1 );                /* xmm0 += xmm1 */
+         if (dims == 4) {
+            emit_MOV( func, 1, 7 );             /* xmm1 = xmm7 */
+            emit_mul( func, 1, 7 );             /* xmm1 *= xmm7 */
+            emit_add( func, 0, 0 );             /* xmm0 += xmm1 */
+         }
+         emit_rsqrt( func, 1, 0 );              /* xmm1 = 1/sqrt(xmm0) */
+         FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
+            if (chan_index < dims) {
+               emit_mul( func, 4+chan_index, 1); /* xmm[4+ch] *= xmm1 */
+               STORE( func, *inst, 4+chan_index, 0, chan_index );
+            }
+         }
+      }
       break;
 
    case TGSI_OPCODE_DIV:
@@ -2003,7 +2211,16 @@ emit_instruction(
       break;
 
    case TGSI_OPCODE_DP2:
-      return 0;
+      FETCH( func, *inst, 0, 0, CHAN_X );  /* xmm0 = src[0].x */
+      FETCH( func, *inst, 1, 1, CHAN_X );  /* xmm1 = src[1].x */
+      emit_mul( func, 0, 1 );              /* xmm0 = xmm0 * xmm1 */
+      FETCH( func, *inst, 1, 0, CHAN_Y );  /* xmm1 = src[0].y */
+      FETCH( func, *inst, 2, 1, CHAN_Y );  /* xmm2 = src[1].y */
+      emit_mul( func, 1, 2 );              /* xmm1 = xmm1 * xmm2 */
+      emit_add( func, 0, 1 );              /* xmm0 = xmm0 + xmm1 */
+      FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
+         STORE( func, *inst, 0, 0, chan_index );  /* dest[ch] = xmm0 */
+      }
       break;
 
    case TGSI_OPCODE_TXL:
diff --git a/src/gallium/auxiliary/util/Makefile b/src/gallium/auxiliary/util/Makefile
index d3951e4e7d7..f611f82773a 100644
--- a/src/gallium/auxiliary/util/Makefile
+++ b/src/gallium/auxiliary/util/Makefile
@@ -10,6 +10,8 @@ C_SOURCES = \
 	u_gen_mipmap.c \
 	u_handle_table.c \
 	u_hash_table.c \
+	u_keymap.c \
+	u_linear.c \
 	u_math.c \
 	u_mm.c \
 	u_rect.c \
diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript
index e65c17b1cc8..8a04955a16e 100644
--- a/src/gallium/auxiliary/util/SConscript
+++ b/src/gallium/auxiliary/util/SConscript
@@ -11,13 +11,14 @@ util = env.ConvenienceLibrary(
 		'u_gen_mipmap.c',
 		'u_handle_table.c',
 		'u_hash_table.c',
+		'u_keymap.c',
 		'u_math.c',
 		'u_mm.c',
 		'u_rect.c',
 		'u_simple_shaders.c',
 		'u_snprintf.c',
-        'u_stream_stdc.c',
-        'u_stream_wd.c',
+		'u_stream_stdc.c',
+		'u_stream_wd.c',
 		'u_tile.c',
 		'u_time.c',
 	])
diff --git a/src/gallium/auxiliary/util/p_debug.c b/src/gallium/auxiliary/util/p_debug.c
index 125f3daf00b..acdfa211c84 100644
--- a/src/gallium/auxiliary/util/p_debug.c
+++ b/src/gallium/auxiliary/util/p_debug.c
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright (c) 2008 VMware, Inc.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -36,6 +37,13 @@
 #include <windows.h>
 #include <winddi.h>
 
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_CE)
+
+#include <stdio.h> 
+#include <stdlib.h> 
+#include <windows.h> 
+#include <types.h> 
+
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
 
 #ifndef WIN32_LEAN_AND_MEAN
@@ -98,7 +106,35 @@ void _debug_vprintf(const char *format, va_list ap)
       OutputDebugStringA(buf);
       buf[0] = '\0';
    }
-#elif defined(PIPE_SUBSYSTEM_WINDOWS_CE) || defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT) 
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_CE)
+   wchar_t *wide_format;
+   long wide_str_len;   
+   char buf[512];   
+   int ret;   
+#if (_WIN32_WCE < 600)
+   ret = vsprintf(buf, format, ap);   
+   if(ret < 0){   
+       sprintf(buf, "Cant handle debug print!");   
+       ret = 25;
+   }
+#else
+   ret = vsprintf_s(buf, 512, format, ap);   
+   if(ret < 0){   
+       sprintf_s(buf, 512, "Cant handle debug print!");   
+       ret = 25;
+   }
+#endif
+   buf[ret] = '\0';   
+   /* Format is ascii - needs to be converted to wchar_t for printing */   
+   wide_str_len = MultiByteToWideChar(CP_ACP, 0, (const char *) buf, -1, NULL, 0);   
+   wide_format = (wchar_t *) malloc((wide_str_len+1) * sizeof(wchar_t));   
+   if (wide_format) {   
+      MultiByteToWideChar(CP_ACP, 0, (const char *) buf, -1,   
+            wide_format, wide_str_len);   
+      NKDbgPrintfW(wide_format, wide_format);   
+      free(wide_format);   
+   } 
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
    /* TODO */
 #else /* !PIPE_SUBSYSTEM_WINDOWS */
 #ifdef DEBUG
@@ -308,6 +344,13 @@ debug_get_flags_option(const char *name,
    str = _debug_get_option(name);
    if(!str)
       result = dfault;
+   else if (!util_strcmp(str, "help")) {
+      result = dfault;
+      while (flags->name) {
+         debug_printf("%s: help for %s: %s [0x%lx]\n", __FUNCTION__, name, flags->name, flags->value);
+         flags++;
+      }
+   }
    else {
       result = 0;
       while( flags->name ) {
@@ -317,7 +360,12 @@ debug_get_flags_option(const char *name,
       }
    }
 
-   debug_printf("%s: %s = 0x%lx\n", __FUNCTION__, name, result);
+   if (str) {
+      debug_printf("%s: %s = 0x%lx (%s)\n", __FUNCTION__, name, result, str);
+   }
+   else {
+      debug_printf("%s: %s = 0x%lx\n", __FUNCTION__, name, result);
+   }
 
    return result;
 }
@@ -486,16 +534,24 @@ static const struct debug_named_value pipe_format_names[] = {
    DEBUG_NAMED_VALUE(PIPE_FORMAT_R8G8B8A8_SSCALED),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_R8G8B8X8_SSCALED),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_L8_SRGB),
-   DEBUG_NAMED_VALUE(PIPE_FORMAT_A8_L8_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_A8L8_SRGB),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_R8G8B8_SRGB),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_R8G8B8A8_SRGB),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_R8G8B8X8_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_A8R8G8B8_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_X8R8G8B8_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_B8G8R8A8_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_B8G8R8X8_SRGB),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_X8UB8UG8SR8S_NORM),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_B6UG5SR5S_NORM),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT1_RGB),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT1_RGBA),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT3_RGBA),
    DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT5_RGBA),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT1_SRGB),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT1_SRGBA),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT3_SRGBA),
+   DEBUG_NAMED_VALUE(PIPE_FORMAT_DXT5_SRGBA),
 #endif
    DEBUG_NAMED_VALUE_END
 };
@@ -627,6 +683,7 @@ void
 debug_dump_surface_bmp(const char *filename,
                        struct pipe_surface *surface)
 {
+#ifndef PIPE_SUBSYSTEM_WINDOWS_MINIPORT
    struct util_stream *stream;
    unsigned surface_usage;
    struct bmp_file_header bmfh;
@@ -693,6 +750,7 @@ error2:
    FREE(rgba);
 error1:
    ;
+#endif
 }
 
 #endif
diff --git a/src/gallium/auxiliary/util/u_keymap.c b/src/gallium/auxiliary/util/u_keymap.c
new file mode 100644
index 00000000000..01b17ddb1b3
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_keymap.c
@@ -0,0 +1,309 @@
+/**************************************************************************
+ *
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * Key lookup/associative container.
+ *
+ * Like Jose's u_hash_table, based on CSO cache code for now.
+ *
+ * Author: Brian Paul
+ */
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_debug.h"
+#include "pipe/p_error.h"
+
+#include "cso_cache/cso_hash.h"
+
+#include "util/u_memory.h"
+#include "util/u_keymap.h"
+
+
+struct keymap
+{
+   struct cso_hash *cso;   
+   unsigned key_size;
+   unsigned max_entries; /* XXX not obeyed net */
+   unsigned num_entries;
+   keymap_delete_func delete_func;
+};
+
+
+struct keymap_item
+{
+   void *key, *value;
+};
+
+
+/**
+ * This the default key-delete function used when the client doesn't
+ * provide one.
+ */
+static void
+default_delete_func(const struct keymap *map,
+                    const void *key, void *data, void *user)
+{
+   FREE((void*) data);
+}
+
+
+static INLINE struct keymap_item *
+hash_table_item(struct cso_hash_iter iter)
+{
+   return (struct keymap_item *) cso_hash_iter_data(iter);
+}
+
+
+/**
+ * Return 4-byte hash key for a block of bytes.
+ */
+static unsigned
+hash(const void *key, unsigned keySize)
+{
+   unsigned i, hash;
+
+   keySize /= 4; /* convert from bytes to uints */
+
+   hash = 0;
+   for (i = 0; i < keySize; i++) {
+      hash ^= (i + 1) * ((const unsigned *) key)[i];
+   }
+
+   /*hash = hash ^ (hash >> 11) ^ (hash >> 22);*/
+
+   return hash;
+}
+
+
+/**
+ * Create a new map.
+ * \param keySize  size of the keys in bytes
+ * \param maxEntries  max number of entries to allow (~0 = infinity)
+ * \param deleteFunc  optional callback to call when entries
+ *                    are deleted/replaced
+ */
+struct keymap *
+util_new_keymap(unsigned keySize, unsigned maxEntries,
+                 keymap_delete_func deleteFunc)
+{
+   struct keymap *map = MALLOC_STRUCT(keymap);
+   if (!map)
+      return NULL;
+   
+   map->cso = cso_hash_create();
+   if (!map->cso) {
+      FREE(map);
+      return NULL;
+   }
+   
+   map->max_entries = maxEntries;
+   map->num_entries = 0;
+   map->key_size = keySize;
+   map->delete_func = deleteFunc ? deleteFunc : default_delete_func;
+
+   return map;
+}
+
+
+/**
+ * Delete/free a keymap and all entries.  The deleteFunc that was given at
+ * create time will be called for each entry.
+ * \param user  user-provided pointer passed through to the delete callback
+ */
+void
+util_delete_keymap(struct keymap *map, void *user)
+{
+   util_keymap_remove_all(map, user);
+   cso_hash_delete(map->cso);
+   FREE(map);
+}
+
+
+static INLINE struct cso_hash_iter
+hash_table_find_iter(const struct keymap *map, const void *key,
+                     unsigned key_hash)
+{
+   struct cso_hash_iter iter;
+   struct keymap_item *item;
+   
+   iter = cso_hash_find(map->cso, key_hash);
+   while (!cso_hash_iter_is_null(iter)) {
+      item = (struct keymap_item *) cso_hash_iter_data(iter);
+      if (!memcmp(item->key, key, map->key_size))
+         break;
+      iter = cso_hash_iter_next(iter);
+   }
+   
+   return iter;
+}
+
+
+static INLINE struct keymap_item *
+hash_table_find_item(const struct keymap *map, const void *key,
+                     unsigned key_hash)
+{
+   struct cso_hash_iter iter = hash_table_find_iter(map, key, key_hash);
+   if (cso_hash_iter_is_null(iter)) {
+      return NULL;
+   }
+   else {
+      return hash_table_item(iter);
+   }
+}
+
+
+/**
+ * Insert a new key + data pointer into the table.
+ * Note: we create a copy of the key, but not the data!
+ * If the key is already present in the table, replace the existing
+ * entry (calling the delete callback on the previous entry).
+ * If the maximum capacity of the map is reached an old entry
+ * will be deleted (the delete callback will be called).
+ */
+boolean
+util_keymap_insert(struct keymap *map, const void *key,
+                   const void *data, void *user)
+{
+   unsigned key_hash;
+   struct keymap_item *item;
+   struct cso_hash_iter iter;
+
+   assert(map);
+
+   key_hash = hash(key, map->key_size);
+
+   item = hash_table_find_item(map, key, key_hash);
+   if (item) {
+      /* call delete callback for old entry/item */
+      map->delete_func(map, item->key, item->value, user);
+      item->value = (void *) data;
+      return TRUE;
+   }
+   
+   item = MALLOC_STRUCT(keymap_item);
+   if (!item)
+      return FALSE;
+
+   item->key = mem_dup(key, map->key_size);
+   item->value = (void *) data;
+   
+   iter = cso_hash_insert(map->cso, key_hash, item);
+   if (cso_hash_iter_is_null(iter)) {
+      FREE(item);
+      return FALSE;
+   }
+
+   map->num_entries++;
+
+   return TRUE;
+}
+
+
+/**
+ * Look up a key in the map and return the associated data pointer.
+ */
+const void *
+util_keymap_lookup(const struct keymap *map, const void *key)
+{
+   unsigned key_hash;
+   struct keymap_item *item;
+
+   assert(map);
+
+   key_hash = hash(key, map->key_size);
+
+   item = hash_table_find_item(map, key, key_hash);
+   if (!item)
+      return NULL;
+   
+   return item->value;
+}
+
+
+/**
+ * Remove an entry from the map.
+ * The delete callback will be called if the given key/entry is found.
+ * \param user  passed to the delete callback as the last param.
+ */
+void
+util_keymap_remove(struct keymap *map, const void *key, void *user)
+{
+   unsigned key_hash;
+   struct cso_hash_iter iter;
+   struct keymap_item *item;
+
+   assert(map);
+
+   key_hash = hash(key, map->key_size);
+
+   iter = hash_table_find_iter(map, key, key_hash);
+   if (cso_hash_iter_is_null(iter))
+      return;
+   
+   item = hash_table_item(iter);
+   assert(item);
+   map->delete_func(map, item->key, item->value, user);
+   FREE(item->key);
+   FREE(item);
+   
+   map->num_entries--;
+
+   cso_hash_erase(map->cso, iter);
+}
+
+
+/**
+ * Remove all entries from the map, calling the delete callback for each.
+ * \param user  passed to the delete callback as the last param.
+ */
+void
+util_keymap_remove_all(struct keymap *map, void *user)
+{
+   struct cso_hash_iter iter;
+   struct keymap_item *item;
+
+   assert(map);
+   
+   iter = cso_hash_first_node(map->cso);
+   while (!cso_hash_iter_is_null(iter)) {
+      item = (struct keymap_item *)
+         cso_hash_take(map->cso, cso_hash_iter_key(iter));
+      map->delete_func(map, item->key, item->value, user);
+      FREE(item->key);
+      FREE(item);
+      iter = cso_hash_first_node(map->cso);
+   }
+}
+
+
+extern void
+util_keymap_info(const struct keymap *map)
+{
+   debug_printf("Keymap %p: %u of max %u entries\n",
+                (void *) map, map->num_entries, map->max_entries);
+}
diff --git a/src/gallium/auxiliary/util/u_keymap.h b/src/gallium/auxiliary/util/u_keymap.h
new file mode 100644
index 00000000000..8d60a76fc3c
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_keymap.h
@@ -0,0 +1,68 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef U_KEYMAP_H
+#define U_KEYMAP_H
+
+#include "pipe/p_compiler.h"
+
+
+/** opaque keymap type */
+struct keymap;
+
+
+/** Delete/callback function type */
+typedef void (*keymap_delete_func)(const struct keymap *map,
+                                   const void *key, void *data,
+                                   void *user);
+
+
+extern struct keymap *
+util_new_keymap(unsigned keySize, unsigned maxEntries,
+                keymap_delete_func deleteFunc);
+
+extern void
+util_delete_keymap(struct keymap *map, void *user);
+
+extern boolean
+util_keymap_insert(struct keymap *map, const void *key,
+                   const void *data, void *user);
+
+extern const void *
+util_keymap_lookup(const struct keymap *map, const void *key);
+
+extern void
+util_keymap_remove(struct keymap *map, const void *key, void *user);
+
+extern void
+util_keymap_remove_all(struct keymap *map, void *user);
+
+extern void
+util_keymap_info(const struct keymap *map);
+
+
+#endif /* U_KEYMAP_H */
diff --git a/src/gallium/auxiliary/util/u_linear.c b/src/gallium/auxiliary/util/u_linear.c
new file mode 100644
index 00000000000..a76704ffc7f
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_linear.c
@@ -0,0 +1,69 @@
+
+#include "pipe/p_debug.h"
+#include "u_linear.h"
+
+void
+pipe_linear_to_tile(size_t src_stride, void *src_ptr,
+		    struct pipe_tile_info *t, void *dst_ptr)
+{
+   int x, y, z;
+   char *ptr;
+   size_t bytes = t->cols * t->block.size;
+
+
+   assert(pipe_linear_check_tile(t));
+
+   /* lets write lineary to the tiled buffer */
+   for (y = 0; y < t->tiles_y; y++) {
+      for (x = 0; x < t->tiles_x; x++) {
+	 /* this inner loop could be replace with SSE magic */
+	 ptr = (char*)src_ptr + src_stride * t->rows * y + bytes * x;
+	 for (z = 0; z < t->rows; z++) {
+	    memcpy(dst_ptr, ptr, bytes);
+	    dst_ptr += bytes;
+	    ptr += src_stride;
+	 }
+      }
+   }
+}
+
+void pipe_linear_from_tile(struct pipe_tile_info *t, void  *src_ptr,
+			   size_t dst_stride, void *dst_ptr)
+{
+   int x, y, z;
+   char *ptr;
+   size_t bytes = t->cols * t->block.size;
+
+   /* lets read lineary from the tiled buffer */
+   for (y = 0; y < t->tiles_y; y++) {
+      for (x = 0; x < t->tiles_x; x++) {
+	 /* this inner loop could be replace with SSE magic */
+	 ptr = (char*)dst_ptr + dst_stride * t->rows * y + bytes * x;
+	 for (z = 0; z < t->rows; z++) {
+	    memcpy(ptr, src_ptr, bytes);
+	    src_ptr += bytes;
+	    ptr += dst_stride;
+	 }
+      }
+   }
+}
+
+void
+pipe_linear_fill_info(struct pipe_tile_info *t,
+		      struct pipe_format_block *block,
+		      unsigned tile_width, unsigned tile_height,
+		      unsigned tiles_x, unsigned tiles_y)
+{
+   t->block = *block;
+
+   t->tile.width = tile_width;
+   t->tile.height = tile_height;
+   t->cols = t->tile.width / t->block.width;
+   t->rows = t->tile.height / t->block.height;
+   t->tile.size = t->cols * t->rows * t->block.size;
+
+   t->tiles_x = tiles_x;
+   t->tiles_y = tiles_y;
+   t->stride = t->cols * t->tiles_x * t->block.size;
+   t->size = t->tiles_x * t->tiles_y * t->tile.size;
+}
diff --git a/src/gallium/auxiliary/util/u_linear.h b/src/gallium/auxiliary/util/u_linear.h
new file mode 100644
index 00000000000..e337cfd7702
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_linear.h
@@ -0,0 +1,60 @@
+
+#ifndef U_LINEAR_H
+#define U_LINEAR_H
+
+#include "pipe/p_format.h"
+struct pipe_tile_info
+{
+   unsigned size;
+   unsigned stride;
+
+   /* The number of tiles */
+   unsigned tiles_x;
+   unsigned tiles_y;
+
+   /* size of each tile expressed in blocks */
+   unsigned cols;
+   unsigned rows;
+
+   /* Describe the tile in pixels */
+   struct pipe_format_block tile;
+
+   /* Describe each block within the tile */
+   struct pipe_format_block block;
+};
+
+void pipe_linear_to_tile(size_t src_stride, void *src_ptr,
+			 struct pipe_tile_info *t, void  *dst_ptr);
+
+void pipe_linear_from_tile(struct pipe_tile_info *t, void  *src_ptr,
+			   size_t dst_stride, void *dst_ptr);
+
+/**
+ * Convenience function to fillout a pipe_tile_info struct.
+ * @t info to fill out.
+ * @block block info about pixel layout
+ * @tile_width the width of the tile in pixels
+ * @tile_height the height of the tile in pixels
+ * @tiles_x number of tiles in x axis
+ * @tiles_y number of tiles in y axis
+ */
+void pipe_linear_fill_info(struct pipe_tile_info *t,
+			   struct pipe_format_block *block,
+			   unsigned tile_width, unsigned tile_height,
+			   unsigned tiles_x, unsigned tiles_y);
+
+static INLINE boolean pipe_linear_check_tile(struct pipe_tile_info *t)
+{
+   if (t->tile.size != t->block.size * t->cols * t->rows)
+      return FALSE;
+
+   if (t->stride != t->block.size * t->cols * t->tiles_x)
+      return FALSE;
+
+   if (t->size < t->stride * t->rows * t->tiles_y)
+      return FALSE;
+
+   return TRUE;
+}
+
+#endif /* U_LINEAR_H */
diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index 1ae3234423a..aa4fa17b596 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -68,7 +68,7 @@ __inline double ceil(double val)
    return ceil_val;
 }
 
-#ifndef PIPE_SUBSYSTEM_WINDOWS_CE
+#ifndef PIPE_SUBSYSTEM_WINDOWS_CE_OGL
 __inline double floor(double val)
 {
    double floor_val;
diff --git a/src/gallium/auxiliary/util/u_memory.h b/src/gallium/auxiliary/util/u_memory.h
index 857102719dc..1a6b596421f 100644
--- a/src/gallium/auxiliary/util/u_memory.h
+++ b/src/gallium/auxiliary/util/u_memory.h
@@ -151,6 +151,8 @@ REALLOC( void *old_ptr, unsigned old_size, unsigned new_size )
 
 #define CALLOC_STRUCT(T)   (struct T *) CALLOC(1, sizeof(struct T))
 
+#define CALLOC_VARIANT_LENGTH_STRUCT(T,more_size)   ((struct T *) CALLOC(1, sizeof(struct T) + more_size))
+
 
 /**
  * Return memory on given byte alignment
diff --git a/src/gallium/auxiliary/util/u_mm.c b/src/gallium/auxiliary/util/u_mm.c
index 01dd67c810b..45ce257b5e5 100644
--- a/src/gallium/auxiliary/util/u_mm.c
+++ b/src/gallium/auxiliary/util/u_mm.c
@@ -31,7 +31,7 @@
 
 
 void
-mmDumpMemInfo(const struct mem_block *heap)
+u_mmDumpMemInfo(const struct mem_block *heap)
 {
    debug_printf("Memory heap %p:\n", (void *)heap);
    if (heap == 0) {
@@ -58,7 +58,7 @@ mmDumpMemInfo(const struct mem_block *heap)
 }
 
 struct mem_block *
-mmInit(int ofs, int size)
+u_mmInit(int ofs, int size)
 {
    struct mem_block *heap, *block;
   
@@ -165,7 +165,7 @@ SliceBlock(struct mem_block *p,
 
 
 struct mem_block *
-mmAllocMem(struct mem_block *heap, int size, int align2, int startSearch)
+u_mmAllocMem(struct mem_block *heap, int size, int align2, int startSearch)
 {
    struct mem_block *p;
    const int mask = (1 << align2)-1;
@@ -202,7 +202,7 @@ mmAllocMem(struct mem_block *heap, int size, int align2, int startSearch)
 
 
 struct mem_block *
-mmFindBlock(struct mem_block *heap, int start)
+u_mmFindBlock(struct mem_block *heap, int start)
 {
    struct mem_block *p;
 
@@ -241,7 +241,7 @@ Join2Blocks(struct mem_block *p)
 }
 
 int
-mmFreeMem(struct mem_block *b)
+u_mmFreeMem(struct mem_block *b)
 {
    if (!b)
       return 0;
@@ -270,7 +270,7 @@ mmFreeMem(struct mem_block *b)
 
 
 void
-mmDestroy(struct mem_block *heap)
+u_mmDestroy(struct mem_block *heap)
 {
    struct mem_block *p;
 
diff --git a/src/gallium/auxiliary/util/u_mm.h b/src/gallium/auxiliary/util/u_mm.h
index b226b101cbe..ce20e487635 100644
--- a/src/gallium/auxiliary/util/u_mm.h
+++ b/src/gallium/auxiliary/util/u_mm.h
@@ -49,7 +49,7 @@ struct mem_block {
  * input: total size in bytes
  * return: a heap pointer if OK, NULL if error
  */
-extern struct mem_block *mmInit(int ofs, int size);
+extern struct mem_block *u_mmInit(int ofs, int size);
 
 /**
  * Allocate 'size' bytes with 2^align2 bytes alignment,
@@ -61,7 +61,7 @@ extern struct mem_block *mmInit(int ofs, int size);
  *		startSearch = linear offset from start of heap to begin search
  * return: pointer to the allocated block, 0 if error
  */
-extern struct mem_block *mmAllocMem(struct mem_block *heap, int size, int align2, 
+extern struct mem_block *u_mmAllocMem(struct mem_block *heap, int size, int align2, 
                             int startSearch);
 
 /**
@@ -69,23 +69,23 @@ extern struct mem_block *mmAllocMem(struct mem_block *heap, int size, int align2
  * input: pointer to a block
  * return: 0 if OK, -1 if error
  */
-extern int mmFreeMem(struct mem_block *b);
+extern int u_mmFreeMem(struct mem_block *b);
 
 /**
  * Free block starts at offset
  * input: pointer to a heap, start offset
  * return: pointer to a block
  */
-extern struct mem_block *mmFindBlock(struct mem_block *heap, int start);
+extern struct mem_block *u_mmFindBlock(struct mem_block *heap, int start);
 
 /**
  * destroy MM
  */
-extern void mmDestroy(struct mem_block *mmInit);
+extern void u_mmDestroy(struct mem_block *mmInit);
 
 /**
  * For debuging purpose.
  */
-extern void mmDumpMemInfo(const struct mem_block *mmInit);
+extern void u_mmDumpMemInfo(const struct mem_block *mmInit);
 
 #endif
diff --git a/src/gallium/auxiliary/util/u_sse.h b/src/gallium/auxiliary/util/u_sse.h
new file mode 100644
index 00000000000..e2a8491e62c
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_sse.h
@@ -0,0 +1,77 @@
+/**************************************************************************
+ *
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * SSE intrinsics portability header.
+ * 
+ * Although the SSE intrinsics are support by all modern x86 and x86-64 
+ * compilers, there are some intrisincs missing in some implementations 
+ * (especially older MSVC versions). This header abstracts that away.
+ */
+
+#ifndef U_SSE_H_
+#define U_SSE_H_
+
+#include "pipe/p_config.h"
+
+#if defined(PIPE_ARCH_SSE)
+
+#include <xmmintrin.h>
+#include <emmintrin.h>
+
+
+/* MSVC before VC8 does not support the _mm_castxxx_yyy */
+#if defined(_MSC_VER) && _MSC_VER < 1500
+
+union __declspec(align(16)) m128_types {
+   __m128 m128;
+   __m128i m128i;
+   __m128d m128d;
+};
+
+static __inline __m128
+_mm_castsi128_ps(__m128i a)
+{
+   union m128_types u;
+   u.m128i = a;
+   return u.m128;
+}
+
+static __inline __m128i
+_mm_castps_si128(__m128 a)
+{
+   union m128_types u;
+   u.m128 = a;
+   return u.m128i;
+}
+
+#endif /* defined(_MSC_VER) && _MSC_VER < 1500 */
+
+#endif /* PIPE_ARCH_X86 || PIPE_ARCH_X86_64 */
+
+#endif /* U_SSE_H_ */
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 853c503f4ff..32f6b072a00 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -460,7 +460,7 @@ l8_put_tile_rgba(ubyte *dst,
       for (j = 0; j < w; j++, pRow += 4) {
          unsigned r;
          r = float_to_ubyte(pRow[0]);
-         *dst++ = r;
+         *dst++ = (ubyte) r;
       }
       p += src_stride;
    }
@@ -504,7 +504,7 @@ a8_put_tile_rgba(ubyte *dst,
       for (j = 0; j < w; j++, pRow += 4) {
          unsigned a;
          a = float_to_ubyte(pRow[3]);
-         *dst++ = a;
+         *dst++ = (ubyte) a;
       }
       p += src_stride;
    }
@@ -634,7 +634,7 @@ i8_put_tile_rgba(ubyte *dst,
       for (j = 0; j < w; j++, pRow += 4) {
          unsigned r;
          r = float_to_ubyte(pRow[0]);
-         *dst++ = r;
+         *dst++ = (ubyte) r;
       }
       p += src_stride;
    }
@@ -769,6 +769,32 @@ z24s8_get_tile_rgba(const unsigned *src,
 }
 
 
+/*** PIPE_FORMAT_Z32_FLOAT ***/
+
+/**
+ * Return each Z value as four floats in [0,1].
+ */
+static void
+z32f_get_tile_rgba(const float *src,
+                   unsigned w, unsigned h,
+                   float *p,
+                   unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         pRow[0] =
+         pRow[1] =
+         pRow[2] =
+         pRow[3] = *src++;
+      }
+      p += dst_stride;
+   }
+}
+
+
 /*** PIPE_FORMAT_YCBCR / PIPE_FORMAT_YCBCR_REV ***/
 
 /**
@@ -913,6 +939,9 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
    case PIPE_FORMAT_Z24S8_UNORM:
       z24s8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
       break;
+   case PIPE_FORMAT_Z32_FLOAT:
+      z32f_get_tile_rgba((float *) src, w, h, dst, dst_stride);
+      break;
    case PIPE_FORMAT_YCBCR:
       ycbcr_get_tile_rgba((ushort *) src, w, h, dst, dst_stride, FALSE);
       break;
diff --git a/src/gallium/auxiliary/util/u_time.c b/src/gallium/auxiliary/util/u_time.c
index bf7d1d1c8d5..57b80e56042 100644
--- a/src/gallium/auxiliary/util/u_time.c
+++ b/src/gallium/auxiliary/util/u_time.c
@@ -200,7 +200,7 @@ util_time_timeout(const struct util_time *start,
 }
 
 
-#if defined(PIPE_SUBSYSYEM_WINDOWS_DISPLAY)
+#if defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
 void util_time_sleep(unsigned usecs)
 {
    LONGLONG start, curr, end;
diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h
index cb0631baf52..98554d7f521 100644
--- a/src/gallium/drivers/cell/common.h
+++ b/src/gallium/drivers/cell/common.h
@@ -64,9 +64,13 @@
 #define ROUNDUP16(k)  (((k) + 0xf) & ~0xf)
 
 
-#define CELL_MAX_SPUS 6
+#define CELL_MAX_SPUS 8
 
 #define CELL_MAX_SAMPLERS 4
+#define CELL_MAX_TEXTURE_LEVELS 12  /* 2k x 2k */
+#define CELL_MAX_CONSTANTS 32  /**< number of float[4] constants */
+#define CELL_MAX_WIDTH 1024    /**< max framebuffer width */
+#define CELL_MAX_HEIGHT 1024   /**< max framebuffer width */
 
 #define TILE_SIZE 32
 
@@ -94,43 +98,94 @@
 #define CELL_CMD_STATE_BIND_VS       18
 #define CELL_CMD_STATE_FRAGMENT_PROGRAM 19
 #define CELL_CMD_STATE_ATTRIB_FETCH  20
-#define CELL_CMD_VS_EXECUTE          22
-#define CELL_CMD_FLUSH_BUFFER_RANGE  23
+#define CELL_CMD_STATE_FS_CONSTANTS  21
+#define CELL_CMD_STATE_RASTERIZER    22
+#define CELL_CMD_VS_EXECUTE          23
+#define CELL_CMD_FLUSH_BUFFER_RANGE  24
+#define CELL_CMD_FENCE               25
 
 
+/** Command/batch buffers */
 #define CELL_NUM_BUFFERS 4
 #define CELL_BUFFER_SIZE (4*1024)  /**< 16KB would be the max */
 
 #define CELL_BUFFER_STATUS_FREE 10
 #define CELL_BUFFER_STATUS_USED 20
 
+/** Debug flags */
+#define CELL_DEBUG_CHECKER              (1 << 0)
+#define CELL_DEBUG_ASM                  (1 << 1)
+#define CELL_DEBUG_SYNC                 (1 << 2)
+#define CELL_DEBUG_FRAGMENT_OPS         (1 << 3)
+#define CELL_DEBUG_FRAGMENT_OP_FALLBACK (1 << 4)
+#define CELL_DEBUG_CMD                  (1 << 5)
+#define CELL_DEBUG_CACHE                (1 << 6)
 
-#define CELL_DEBUG_CHECKER  (1 << 0)
-#define CELL_DEBUG_SYNC     (1 << 1)
+#define CELL_FENCE_IDLE      0
+#define CELL_FENCE_EMITTED   1
+#define CELL_FENCE_SIGNALLED 2
 
+#define CELL_FACING_FRONT    0
+#define CELL_FACING_BACK     1
+
+struct cell_fence
+{
+   /** There's a 16-byte status qword per SPU */
+   volatile uint status[CELL_MAX_SPUS][4];
+};
 
 
-/** Max instructions for doing per-fragment operations */
-#define SPU_MAX_FRAGMENT_OPS_INSTS 64
+/**
+ * Fence command sent to SPUs.  In response, the SPUs will write
+ * CELL_FENCE_STATUS_SIGNALLED back to the fence status word in main memory.
+ */
+struct cell_command_fence
+{
+   uint64_t opcode;      /**< CELL_CMD_FENCE */
+   struct cell_fence *fence;
+};
 
 
 /**
  * Command to specify per-fragment operations state and generated code.
+ * Note that this is a variant-length structure, allocated with as 
+ * much memory as needed to hold the generated code; the "code"
+ * field *must* be the last field in the structure.  Also, the entire
+ * length of the structure (including the variant code field) must be
+ * a multiple of 8 bytes; we require that this structure itself be
+ * a multiple of 8 bytes, and that the generated code also be a multiple
+ * of 8 bytes.
+ *
+ * Also note that the dsa, blend, blend_color fields are really only needed
+ * for the fallback/C per-pixel code.  They're not used when we generate
+ * dynamic SPU fragment code (which is the normal case), and will eventually
+ * be removed from this structure.
  */
 struct cell_command_fragment_ops
 {
    uint64_t opcode;      /**< CELL_CMD_STATE_FRAGMENT_OPS */
+
+   /* Fields for the fallback case */
    struct pipe_depth_stencil_alpha_state dsa;
    struct pipe_blend_state blend;
-   unsigned code[SPU_MAX_FRAGMENT_OPS_INSTS];
+   struct pipe_blend_color blend_color;
+
+   /* Fields for the generated SPU code */
+   unsigned total_code_size;
+   unsigned front_code_index;
+   unsigned back_code_index;
+   /* this field has variant length, and must be the last field in 
+    * the structure
+    */
+   unsigned code[0];
 };
 
 
 /** Max instructions for fragment programs */
-#define SPU_MAX_FRAGMENT_PROGRAM_INSTS 128
+#define SPU_MAX_FRAGMENT_PROGRAM_INSTS 512
 
 /**
- * Command to send a fragment progra to SPUs.
+ * Command to send a fragment program to SPUs.
  */
 struct cell_command_fragment_program
 {
@@ -145,7 +200,7 @@ struct cell_command_fragment_program
  */
 struct cell_command_framebuffer
 {
-   uint64_t opcode;     /**< CELL_CMD_FRAMEBUFFER */
+   uint64_t opcode;     /**< CELL_CMD_STATE_FRAMEBUFFER */
    int width, height;
    void *color_start, *depth_start;
    enum pipe_format color_format, depth_format;
@@ -153,6 +208,16 @@ struct cell_command_framebuffer
 
 
 /**
+ * Tell SPUs about rasterizer state.
+ */
+struct cell_command_rasterizer
+{
+   uint64_t opcode;    /**< CELL_CMD_STATE_RASTERIZER */
+   struct pipe_rasterizer_state rasterizer;
+};
+
+
+/**
  * Clear framebuffer to the given value/color.
  */
 struct cell_command_clear_surface
@@ -248,23 +313,27 @@ struct cell_command_sampler
 struct cell_command_texture
 {
    uint64_t opcode;     /**< CELL_CMD_STATE_TEXTURE */
+   uint target;         /**< PIPE_TEXTURE_x */
    uint unit;
-   void *start;         /**< Address in main memory */
-   ushort width, height;
+   void *start[CELL_MAX_TEXTURE_LEVELS];   /**< Address in main memory */
+   ushort width[CELL_MAX_TEXTURE_LEVELS];
+   ushort height[CELL_MAX_TEXTURE_LEVELS];
+   ushort depth[CELL_MAX_TEXTURE_LEVELS];
 };
 
 
-/** XXX unions don't seem to work */
-/* XXX this should go away; all commands should be placed in batch buffers */
-struct cell_command
+#define MAX_SPU_FUNCTIONS 12
+/**
+ * Used to tell the PPU about the address of particular functions in the
+ * SPU's address space.
+ */
+struct cell_spu_function_info
 {
-#if 0
-   struct cell_command_framebuffer fb;
-   struct cell_command_clear_surface clear;
-   struct cell_command_render render;
-#endif
-   struct cell_command_vs vs;
-} ALIGN16_ATTRIB;
+   uint num;
+   char names[MAX_SPU_FUNCTIONS][16];
+   uint addrs[MAX_SPU_FUNCTIONS];
+   char pad[12];   /**< Pad struct to multiple of 16 bytes (256 currently) */
+};
 
 
 /** This is the object passed to spe_create_thread() */
@@ -273,11 +342,13 @@ struct cell_init_info
    unsigned id;
    unsigned num_spus;
    unsigned debug_flags;  /**< mask of CELL_DEBUG_x flags */
-   struct cell_command *cmd;
+   float inv_timebase;    /**< 1.0/timebase, for perf measurement */
 
    /** Buffers for command batches, vertex/index data */
    ubyte *buffers[CELL_NUM_BUFFERS];
    uint *buffer_status;  /**< points at cell_context->buffer_status */
+
+   struct cell_spu_function_info *spu_functions;
 } ALIGN16_ATTRIB;
 
 
diff --git a/src/gallium/drivers/cell/ppu/Makefile b/src/gallium/drivers/cell/ppu/Makefile
index b28f4c5c31f..9358a47284c 100644
--- a/src/gallium/drivers/cell/ppu/Makefile
+++ b/src/gallium/drivers/cell/ppu/Makefile
@@ -24,6 +24,7 @@ SOURCES = \
 	cell_clear.c \
 	cell_context.c \
 	cell_draw_arrays.c \
+	cell_fence.c \
 	cell_flush.c \
 	cell_gen_fragment.c \
 	cell_gen_fp.c \
diff --git a/src/gallium/drivers/cell/ppu/cell_batch.c b/src/gallium/drivers/cell/ppu/cell_batch.c
index 16882c01295..962775cd335 100644
--- a/src/gallium/drivers/cell/ppu/cell_batch.c
+++ b/src/gallium/drivers/cell/ppu/cell_batch.c
@@ -28,6 +28,7 @@
 
 #include "cell_context.h"
 #include "cell_batch.h"
+#include "cell_fence.h"
 #include "cell_spu.h"
 
 
@@ -42,7 +43,9 @@
 uint
 cell_get_empty_buffer(struct cell_context *cell)
 {
-   uint buf = 0, tries = 0;
+   static uint prev_buffer = 0;
+   uint buf = (prev_buffer + 1) % CELL_NUM_BUFFERS;
+   uint tries = 0;
 
    /* Find a buffer that's marked as free by all SPUs */
    while (1) {
@@ -58,8 +61,13 @@ cell_get_empty_buffer(struct cell_context *cell)
                   cell->buffer_status[spu][buf][0] = CELL_BUFFER_STATUS_USED;
                }
                /*
-               printf("PPU: ALLOC BUFFER %u\n", buf);
+               printf("PPU: ALLOC BUFFER %u, %u tries\n", buf, tries);
                */
+               prev_buffer = buf;
+
+               /* release tex buffer associated w/ prev use of this batch buf */
+               cell_free_fenced_buffers(cell, &cell->fenced_buffers[buf]);
+
                return buf;
             }
          }
@@ -82,6 +90,37 @@ cell_get_empty_buffer(struct cell_context *cell)
 
 
 /**
+ * Append a fence command to the current batch buffer.
+ * Note that we're sure there's always room for this because of the
+ * adjusted size check in cell_batch_free_space().
+ */
+static void
+emit_fence(struct cell_context *cell)
+{
+   const uint batch = cell->cur_batch;
+   const uint size = cell->buffer_size[batch];
+   struct cell_command_fence *fence_cmd;
+   struct cell_fence *fence = &cell->fenced_buffers[batch].fence;
+   uint i;
+
+   /* set fence status to emitted, not yet signalled */
+   for (i = 0; i < cell->num_spus; i++) {
+      fence->status[i][0] = CELL_FENCE_EMITTED;
+   }
+
+   ASSERT(size + sizeof(struct cell_command_fence) <= CELL_BUFFER_SIZE);
+
+   fence_cmd = (struct cell_command_fence *) (cell->buffer[batch] + size);
+   fence_cmd->opcode = CELL_CMD_FENCE;
+   fence_cmd->fence = fence;
+
+   /* update batch buffer size */
+   cell->buffer_size[batch] = size + sizeof(struct cell_command_fence);
+   assert(sizeof(struct cell_command_fence) % 8 == 0);
+}
+
+
+/**
  * Flush the current batch buffer to the SPUs.
  * An empty buffer will be found and set as the new current batch buffer
  * for subsequent commands/data.
@@ -91,7 +130,7 @@ cell_batch_flush(struct cell_context *cell)
 {
    static boolean flushing = FALSE;
    uint batch = cell->cur_batch;
-   const uint size = cell->buffer_size[batch];
+   uint size = cell->buffer_size[batch];
    uint spu, cmd_word;
 
    assert(!flushing);
@@ -99,6 +138,14 @@ cell_batch_flush(struct cell_context *cell)
    if (size == 0)
       return;
 
+   /* Before we use this batch buffer, make sure any fenced texture buffers
+    * are released.
+    */
+   if (cell->fenced_buffers[batch].head) {
+      emit_fence(cell);
+      size = cell->buffer_size[batch];
+   }
+
    flushing = TRUE;
 
    assert(batch < CELL_NUM_BUFFERS);
@@ -139,6 +186,7 @@ uint
 cell_batch_free_space(const struct cell_context *cell)
 {
    uint free = CELL_BUFFER_SIZE - cell->buffer_size[cell->cur_batch];
+   free -= sizeof(struct cell_command_fence);
    return free;
 }
 
@@ -169,7 +217,7 @@ cell_batch_append(struct cell_context *cell, const void *data, uint bytes)
 
    size = cell->buffer_size[cell->cur_batch];
 
-   if (size + bytes > CELL_BUFFER_SIZE) {
+   if (bytes > cell_batch_free_space(cell)) {
       cell_batch_flush(cell);
       size = 0;
    }
@@ -223,7 +271,7 @@ cell_batch_alloc_aligned(struct cell_context *cell, uint bytes,
 
    padbytes = (alignment - (size % alignment)) % alignment;
 
-   if (padbytes + size + bytes > CELL_BUFFER_SIZE) {
+   if (padbytes + bytes > cell_batch_free_space(cell)) {
       cell_batch_flush(cell);
       size = 0;
    }
diff --git a/src/gallium/drivers/cell/ppu/cell_clear.c b/src/gallium/drivers/cell/ppu/cell_clear.c
index c9c0c721bbe..037635e4660 100644
--- a/src/gallium/drivers/cell/ppu/cell_clear.c
+++ b/src/gallium/drivers/cell/ppu/cell_clear.c
@@ -106,4 +106,17 @@ cell_clear_surface(struct pipe_context *pipe, struct pipe_surface *ps,
       clr->surface = surfIndex;
       clr->value = clearValue;
    }
+
+   /* Technically, the surface's contents are now known and cleared,
+    * so we could set the status to PIPE_SURFACE_STATUS_CLEAR.  But
+    * it turns out it's quite painful to recognize when any particular
+    * surface goes from PIPE_SURFACE_STATUS_CLEAR to 
+    * PIPE_SURFACE_STATUS_DEFINED (i.e. with known contents), because
+    * the drawing commands could be operating on numerous draw buffers,
+    * which we'd have to iterate through to set all their stati...
+    * For now, we cheat a bit and set the surface's status to DEFINED
+    * right here.  Later we should revisit this and set the status to
+    * CLEAR here, and find a better place to set the status to DEFINED.
+    */
+   ps->status = PIPE_SURFACE_STATUS_DEFINED;
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_context.c b/src/gallium/drivers/cell/ppu/cell_context.c
index 71f1a3049d1..22d552d8e3d 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.c
+++ b/src/gallium/drivers/cell/ppu/cell_context.c
@@ -47,6 +47,7 @@
 #include "cell_clear.h"
 #include "cell_context.h"
 #include "cell_draw_arrays.h"
+#include "cell_fence.h"
 #include "cell_flush.h"
 #include "cell_state.h"
 #include "cell_surface.h"
@@ -62,6 +63,8 @@ cell_destroy_context( struct pipe_context *pipe )
 {
    struct cell_context *cell = cell_context(pipe);
 
+   util_delete_keymap(cell->fragment_ops_cache, NULL);
+
    cell_spu_exit(cell);
 
    align_free(cell);
@@ -85,13 +88,16 @@ cell_draw_create(struct cell_context *cell)
 }
 
 
-#ifdef DEBUG
 static const struct debug_named_value cell_debug_flags[] = {
    {"checker", CELL_DEBUG_CHECKER},/**< modulate tile clear color by SPU ID */
+   {"asm", CELL_DEBUG_ASM},        /**< dump SPU asm code */
    {"sync", CELL_DEBUG_SYNC},      /**< SPUs do synchronous DMA */
+   {"fragops", CELL_DEBUG_FRAGMENT_OPS}, /**< SPUs emit fragment ops debug messages*/
+   {"fragopfallback", CELL_DEBUG_FRAGMENT_OP_FALLBACK}, /**< SPUs use reference implementation for fragment ops*/
+   {"cmd", CELL_DEBUG_CMD},       /**< SPUs dump command buffer info */
+   {"cache", CELL_DEBUG_CACHE},   /**< report texture cache stats on exit */
    {NULL, 0}
 };
-#endif
 
 
 struct pipe_context *
@@ -99,6 +105,7 @@ cell_create_context(struct pipe_screen *screen,
                     struct cell_winsys *cws)
 {
    struct cell_context *cell;
+   uint i;
 
    /* some fields need to be 16-byte aligned, so align the whole object */
    cell = (struct cell_context*) align_malloc(sizeof(struct cell_context), 16);
@@ -125,11 +132,14 @@ cell_create_context(struct pipe_screen *screen,
    cell_init_state_functions(cell);
    cell_init_shader_functions(cell);
    cell_init_surface_functions(cell);
-   cell_init_texture_functions(cell);
    cell_init_vertex_functions(cell);
 
    cell->draw = cell_draw_create(cell);
 
+   /* Create cache of fragment ops generated code */
+   cell->fragment_ops_cache =
+      util_new_keymap(sizeof(struct cell_fragment_ops_key), ~0, NULL);
+
    cell_init_vbuf(cell);
 
    draw_set_rasterize_stage(cell->draw, cell->vbuf);
@@ -143,17 +153,31 @@ cell_create_context(struct pipe_screen *screen,
                                               cell_debug_flags, 
                                               0 );
 
+   for (i = 0; i < CELL_NUM_BUFFERS; i++)
+      cell_fence_init(&cell->fenced_buffers[i].fence);
+
+
    /*
     * SPU stuff
     */
-   cell->num_spus = 6;
-   /* XXX is this in SDK 3.0 only?
-   cell->num_spus = spe_cpu_info_get(SPE_COUNT_PHYSICAL_SPES, -1);
-   */
+   /* This call only works with SDK 3.0.  Anyone still using 2.1??? */
+   cell->num_cells = spe_cpu_info_get(SPE_COUNT_PHYSICAL_CPU_NODES, -1);
+   cell->num_spus = spe_cpu_info_get(SPE_COUNT_USABLE_SPES, 0);
+   if (cell->debug_flags) {
+      printf("Cell: found %d Cell(s) with %u SPUs\n",
+             cell->num_cells, cell->num_spus);
+   }
+   if (getenv("CELL_NUM_SPUS")) {
+      cell->num_spus = atoi(getenv("CELL_NUM_SPUS"));
+      assert(cell->num_spus > 0);
+   }
 
    cell_start_spus(cell);
 
    cell_init_batch_buffers(cell);
 
+   /* make sure SPU initializations are done before proceeding */
+   cell_flush_int(cell, CELL_FLUSH_WAIT);
+
    return &cell->pipe;
 }
diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h
index 14914b9c6f8..eb1397bb3fa 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.h
+++ b/src/gallium/drivers/cell/ppu/cell_context.h
@@ -38,6 +38,7 @@
 #include "cell/common.h"
 #include "rtasm/rtasm_ppc_spe.h"
 #include "tgsi/tgsi_scan.h"
+#include "util/u_keymap.h"
 
 
 struct cell_vbuf_render;
@@ -67,31 +68,29 @@ struct cell_fragment_shader_state
 
 
 /**
- * Cell blend state atom, subclass of pipe_blend_state.
+ * Key for mapping per-fragment state to cached SPU machine code.
+ *  keymap(cell_fragment_ops_key) => cell_command_fragment_ops
  */
-struct cell_blend_state
+struct cell_fragment_ops_key
 {
-   struct pipe_blend_state base;
-
-   /**
-    * Generated code to perform alpha blending
-    */
-   struct spe_function code;
+   struct pipe_blend_state blend;
+   struct pipe_blend_color blend_color;
+   struct pipe_depth_stencil_alpha_state dsa;
+   enum pipe_format color_format;
+   enum pipe_format zs_format;
 };
 
 
+struct cell_buffer_node;
+
 /**
- * Cell depth/stencil/alpha state atom, subclass of
- * pipe_depth_stencil_alpha_state.
+ * Fenced buffer list.  List of buffers which can be unreferenced after
+ * the fence has been executed/signalled.
  */
-struct cell_depth_stencil_alpha_state
+struct cell_buffer_list
 {
-   struct pipe_depth_stencil_alpha_state base;
-
-   /**
-    * Generated code to perform alpha, stencil, and depth testing on the SPE
-    */
-   struct spe_function code;
+   struct cell_fence fence ALIGN16_ATTRIB;
+   struct cell_buffer_node *head;
 };
 
 
@@ -104,10 +103,10 @@ struct cell_context
 
    struct cell_winsys *winsys;
 
-   const struct cell_blend_state *blend;
+   const struct pipe_blend_state *blend;
    const struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS];
    uint num_samplers;
-   const struct cell_depth_stencil_alpha_state   *depth_stencil;
+   const struct pipe_depth_stencil_alpha_state *depth_stencil;
    const struct pipe_rasterizer_state *rasterizer;
    const struct cell_vertex_shader_state *vs;
    const struct cell_fragment_shader_state *fs;
@@ -135,6 +134,11 @@ struct cell_context
    uint *tex_map;
 
    uint dirty;
+   uint dirty_textures;  /* bitmask of texture units */
+   uint dirty_samplers;  /* bitmask of sampler units */
+
+   /** Cache of code generated for per-fragment ops */
+   struct keymap *fragment_ops_cache;
 
    /** The primitive drawing context */
    struct draw_context *draw;
@@ -149,8 +153,9 @@ struct cell_context
    /** Mapped constant buffers */
    void *mapped_constants[PIPE_SHADER_TYPES];
 
+   struct cell_spu_function_info spu_functions ALIGN16_ATTRIB;
 
-   uint num_spus;
+   uint num_cells, num_spus;
 
    /** Buffers for command batches, vertex/index data */
    uint buffer_size[CELL_NUM_BUFFERS];
@@ -162,6 +167,14 @@ struct cell_context
    uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4] ALIGN16_ATTRIB;
 
 
+   /** Associated with each command/batch buffer is a list of pipe_buffers
+    * that are fenced.  When the last command in a buffer is executed, the
+    * fence will be signalled, indicating that any pipe_buffers preceeding
+    * that fence can be unreferenced (and probably freed).
+    */
+   struct cell_buffer_list fenced_buffers[CELL_NUM_BUFFERS];
+
+
    struct spe_function attrib_fetch;
    unsigned attrib_fetch_offsets[PIPE_MAX_ATTRIBS];
 
diff --git a/src/gallium/drivers/cell/ppu/cell_fence.c b/src/gallium/drivers/cell/ppu/cell_fence.c
new file mode 100644
index 00000000000..867b5dcaa09
--- /dev/null
+++ b/src/gallium/drivers/cell/ppu/cell_fence.c
@@ -0,0 +1,168 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <unistd.h>
+#include "util/u_memory.h"
+#include "pipe/p_inlines.h"
+#include "cell_context.h"
+#include "cell_batch.h"
+#include "cell_fence.h"
+#include "cell_texture.h"
+
+
+void
+cell_fence_init(struct cell_fence *fence)
+{
+   uint i;
+   ASSERT_ALIGN16(fence->status);
+   for (i = 0; i < CELL_MAX_SPUS; i++) {
+      fence->status[i][0] = CELL_FENCE_IDLE;
+   }
+}
+
+
+boolean
+cell_fence_signalled(const struct cell_context *cell,
+                     const struct cell_fence *fence)
+{
+   uint i;
+   for (i = 0; i < cell->num_spus; i++) {
+      if (fence->status[i][0] != CELL_FENCE_SIGNALLED)
+         return FALSE;
+      /*assert(fence->status[i][0] == CELL_FENCE_EMITTED);*/
+   }
+   return TRUE;
+}
+
+
+void
+cell_fence_finish(const struct cell_context *cell,
+                  const struct cell_fence *fence)
+{
+   while (!cell_fence_signalled(cell, fence)) {
+      usleep(10);
+   }
+
+#ifdef DEBUG
+   {
+      uint i;
+      for (i = 0; i < cell->num_spus; i++) {
+         assert(fence->status[i][0] == CELL_FENCE_SIGNALLED);
+      }
+   }
+#endif
+}
+
+
+
+
+struct cell_buffer_node
+{
+   struct pipe_buffer *buffer;
+   struct cell_buffer_node *next;
+};
+
+
+static void
+cell_add_buffer_to_list(struct cell_context *cell,
+                        struct cell_buffer_list *list,
+                        struct pipe_buffer *buffer)
+{
+   struct pipe_screen *ps = cell->pipe.screen;
+   struct cell_buffer_node *node = CALLOC_STRUCT(cell_buffer_node);
+   /* create new list node which references the buffer, insert at head */
+   if (node) {
+      pipe_buffer_reference(ps, &node->buffer, buffer);
+      node->next = list->head;
+      list->head = node;
+   }
+}
+
+
+/**
+ * Wait for completion of the given fence, then unreference any buffers
+ * on the list.
+ * This typically unrefs/frees texture buffers after any rendering which uses
+ * them has completed.
+ */
+void
+cell_free_fenced_buffers(struct cell_context *cell,
+                         struct cell_buffer_list *list)
+{
+   if (list->head) {
+      struct pipe_screen *ps = cell->pipe.screen;
+      struct cell_buffer_node *node;
+
+      cell_fence_finish(cell, &list->fence);
+
+      /* traverse the list, unreferencing buffers, freeing nodes */
+      node = list->head;
+      while (node) {
+         struct cell_buffer_node *next = node->next;
+         assert(node->buffer);
+         pipe_buffer_unmap(ps, node->buffer);
+#if 0
+         printf("Unref buffer %p\n", node->buffer);
+         if (node->buffer->refcount == 1)
+            printf("   Delete!\n");
+#endif
+         pipe_buffer_reference(ps, &node->buffer, NULL);
+         FREE(node);
+         node = next;
+      }
+      list->head = NULL;
+   }
+}
+
+
+/**
+ * This should be called for each render command.
+ * Any texture buffers that are current bound will be added to a fenced
+ * list to be freed later when the fence is executed/signalled.
+ */
+void
+cell_add_fenced_textures(struct cell_context *cell)
+{
+   struct cell_buffer_list *list = &cell->fenced_buffers[cell->cur_batch];
+   uint i;
+
+   for (i = 0; i < cell->num_textures; i++) {
+      struct cell_texture *ct = cell->texture[i];
+      if (ct) {
+         uint level;
+         for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
+            if (ct->tiled_buffer[level]) {
+#if 0
+               printf("Adding texture %p buffer %p to list\n",
+                      ct, ct->tiled_buffer[level]);
+#endif
+               cell_add_buffer_to_list(cell, list, ct->tiled_buffer[level]);
+            }
+         }
+      }
+   }
+}
diff --git a/src/gallium/drivers/cell/ppu/cell_fence.h b/src/gallium/drivers/cell/ppu/cell_fence.h
new file mode 100644
index 00000000000..536b4ba411a
--- /dev/null
+++ b/src/gallium/drivers/cell/ppu/cell_fence.h
@@ -0,0 +1,57 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#ifndef CELL_FENCE_H
+#define CELL_FENCE_H
+
+
+extern void
+cell_fence_init(struct cell_fence *fence);
+
+
+extern boolean
+cell_fence_signalled(const struct cell_context *cell,
+                     const struct cell_fence *fence);
+
+
+extern void
+cell_fence_finish(const struct cell_context *cell,
+                  const struct cell_fence *fence);
+
+
+
+extern void
+cell_free_fenced_buffers(struct cell_context *cell,
+                         struct cell_buffer_list *list);
+
+
+extern void
+cell_add_fenced_textures(struct cell_context *cell);
+
+
+#endif /* CELL_FENCE_H */
diff --git a/src/gallium/drivers/cell/ppu/cell_flush.c b/src/gallium/drivers/cell/ppu/cell_flush.c
index 6596b720101..a64967b4b9e 100644
--- a/src/gallium/drivers/cell/ppu/cell_flush.c
+++ b/src/gallium/drivers/cell/ppu/cell_flush.c
@@ -49,7 +49,7 @@ cell_flush(struct pipe_context *pipe, unsigned flags,
       flags |= CELL_FLUSH_WAIT;
    }
 
-   if (flags & PIPE_FLUSH_SWAPBUFFERS)
+   if (flags & (PIPE_FLUSH_SWAPBUFFERS | PIPE_FLUSH_RENDER_CACHE))
       flags |= CELL_FLUSH_WAIT;
 
    draw_flush( cell->draw );
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
index 6ffe94eb14a..96a1743fc10 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
@@ -37,7 +37,7 @@
  * \author Brian Paul
  */
 
-
+#include <math.h>
 #include "pipe/p_defines.h"
 #include "pipe/p_state.h"
 #include "pipe/p_shader_tokens.h"
@@ -51,25 +51,43 @@
 #include "cell_gen_fp.h"
 
 
-/** Set to 1 to enable debug/disassembly printfs */
-#define DISASSEM 01
+#define MAX_TEMPS 16
+#define MAX_IMMED  8
 
+#define CHAN_X  0
+#define CHAN_Y  1
+#define CHAN_Z  2
+#define CHAN_W  3
 
 /**
  * Context needed during code generation.
  */
 struct codegen
 {
+   struct cell_context *cell;
    int inputs_reg;      /**< 1st function parameter */
    int outputs_reg;     /**< 2nd function parameter */
    int constants_reg;   /**< 3rd function parameter */
-   int temp_regs[8][4]; /**< maps TGSI temps to SPE registers */
+   int temp_regs[MAX_TEMPS][4]; /**< maps TGSI temps to SPE registers */
+   int imm_regs[MAX_IMMED][4];  /**< maps TGSI immediates to SPE registers */
+
+   int num_imm;  /**< number of immediates */
 
    int one_reg;         /**< register containing {1.0, 1.0, 1.0, 1.0} */
 
    /** Per-instruction temps / intermediate temps */
    int num_itemps;
-   int itemps[3];
+   int itemps[12];
+
+   /** Current IF/ELSE/ENDIF nesting level */
+   int if_nesting;
+   /** Index of execution mask register */
+   int exec_mask_reg;
+
+   /** KIL mask: indicates which fragments have been killed */
+   int kill_mask_reg;
+
+   int frame_size;  /**< Stack frame size, in words */
 
    struct spe_function *f;
    boolean error;
@@ -112,19 +130,78 @@ get_const_one_reg(struct codegen *gen)
 {
    if (gen->one_reg <= 0) {
       gen->one_reg = spe_allocate_available_register(gen->f);
-   }
 
-   /* one = {1.0, 1.0, 1.0, 1.0} */
-   spe_load_float(gen->f, gen->one_reg, 1.0f);
-#if DISASSEM
-   printf("il\tr%d, 1.0f\n", gen->one_reg);
-#endif
+      spe_indent(gen->f, 4);
+      spe_comment(gen->f, -4, "INIT CONSTANT 1.0:");
+
+      /* one = {1.0, 1.0, 1.0, 1.0} */
+      spe_load_float(gen->f, gen->one_reg, 1.0f);
+
+      spe_indent(gen->f, -4);
+   }
 
    return gen->one_reg;
 }
 
 
 /**
+ * Return index of the pixel execution mask.
+ * The register is allocated an initialized upon the first call.
+ *
+ * The pixel execution mask controls which pixels in a quad are
+ * modified, according to surrounding conditionals, loops, etc.
+ */
+static int
+get_exec_mask_reg(struct codegen *gen)
+{
+   if (gen->exec_mask_reg <= 0) {
+      gen->exec_mask_reg = spe_allocate_available_register(gen->f);
+
+      spe_indent(gen->f, 4);
+      spe_comment(gen->f, -4, "INIT EXEC MASK = ~0:");
+
+      /* exec_mask = {~0, ~0, ~0, ~0} */
+      spe_load_int(gen->f, gen->exec_mask_reg, ~0);
+
+      spe_indent(gen->f, -4);
+   }
+
+   return gen->exec_mask_reg;
+}
+
+
+static boolean
+is_register_src(struct codegen *gen, int channel,
+                const struct tgsi_full_src_register *src)
+{
+   int swizzle = tgsi_util_get_full_src_register_extswizzle(src, channel);
+   int sign_op = tgsi_util_get_full_src_register_sign_mode(src, channel);
+
+   if (swizzle > TGSI_SWIZZLE_W || sign_op != TGSI_UTIL_SIGN_KEEP) {
+      return FALSE;
+   }
+   if (src->SrcRegister.File == TGSI_FILE_TEMPORARY ||
+       src->SrcRegister.File == TGSI_FILE_IMMEDIATE) {
+      return TRUE;
+   }
+   return FALSE;
+}
+
+  
+static boolean
+is_memory_dst(struct codegen *gen, int channel,
+              const struct tgsi_full_dst_register *dst)
+{
+   if (dst->DstRegister.File == TGSI_FILE_OUTPUT) {
+      return TRUE;
+   }
+   else {
+      return FALSE;
+   }
+}
+
+  
+/**
  * Return the index of the SPU temporary containing the named TGSI
  * source register.  If the TGSI register is a TGSI_FILE_TEMPORARY we
  * just return the corresponding SPE register.  If the TGIS register
@@ -136,35 +213,93 @@ get_src_reg(struct codegen *gen,
             int channel,
             const struct tgsi_full_src_register *src)
 {
-   int reg;
+   int reg = -1;
+   int swizzle = tgsi_util_get_full_src_register_extswizzle(src, channel);
+   boolean reg_is_itemp = FALSE;
+   uint sign_op;
+
+   assert(swizzle >= TGSI_SWIZZLE_X);
+   assert(swizzle <= TGSI_EXTSWIZZLE_ONE);
+
+   if (swizzle == TGSI_EXTSWIZZLE_ONE) {
+      /* Load const one float and early out */
+      reg = get_const_one_reg(gen);
+   }
+   else if (swizzle == TGSI_EXTSWIZZLE_ZERO) {
+      /* Load const zero float and early out */
+      reg = get_itemp(gen);
+      spe_xor(gen->f, reg, reg, reg);
+   }
+   else {
+      assert(swizzle < 4);
 
-   /* XXX need to examine src swizzle info here.
-    * That will involve changing the channel var...
+      switch (src->SrcRegister.File) {
+      case TGSI_FILE_TEMPORARY:
+         reg = gen->temp_regs[src->SrcRegister.Index][swizzle];
+         break;
+      case TGSI_FILE_INPUT:
+         {
+            /* offset is measured in quadwords, not bytes */
+            int offset = src->SrcRegister.Index * 4 + swizzle;
+            reg = get_itemp(gen);
+            reg_is_itemp = TRUE;
+            /* Load:  reg = memory[(machine_reg) + offset] */
+            spe_lqd(gen->f, reg, gen->inputs_reg, offset * 16);
+         }
+         break;
+      case TGSI_FILE_IMMEDIATE:
+         reg = gen->imm_regs[src->SrcRegister.Index][swizzle];
+         break;
+      case TGSI_FILE_CONSTANT:
+         {
+            /* offset is measured in quadwords, not bytes */
+            int offset = src->SrcRegister.Index * 4 + swizzle;
+            reg = get_itemp(gen);
+            reg_is_itemp = TRUE;
+            /* Load:  reg = memory[(machine_reg) + offset] */
+            spe_lqd(gen->f, reg, gen->constants_reg, offset * 16);
+         }
+         break;
+      default:
+         assert(0);
+      }
+   }
+
+   /*
+    * Handle absolute value, negate or set-negative of src register.
     */
+   sign_op = tgsi_util_get_full_src_register_sign_mode(src, channel);
+   if (sign_op != TGSI_UTIL_SIGN_KEEP) {
+      /*
+       * All sign ops are done by manipulating bit 31, the IEEE float sign bit.
+       */
+      const int bit31mask_reg = get_itemp(gen);
+      int result_reg;
+
+      if (reg_is_itemp) {
+         /* re-use 'reg' for the result */
+         result_reg = reg;
+      }
+      else {
+         /* alloc a new reg for the result */
+         result_reg = get_itemp(gen);
+      }
 
+      /* mask with bit 31 set, the rest cleared */
+      spe_load_uint(gen->f, bit31mask_reg, (1 << 31));
 
-   switch (src->SrcRegister.File) {
-   case TGSI_FILE_TEMPORARY:
-      reg = gen->temp_regs[src->SrcRegister.Index][channel];
-      break;
-   case TGSI_FILE_INPUT:
-      {
-         /* offset is measured in quadwords, not bytes */
-         int offset = src->SrcRegister.Index * 4 + channel;
-         reg = get_itemp(gen);
-         /* Load:  reg = memory[(machine_reg) + offset] */
-         spe_lqd(gen->f, reg, gen->inputs_reg, offset);
-#if DISASSEM
-         printf("lqd\tr%d, r%d + %d\n", reg, gen->inputs_reg, offset);
-#endif
+      if (sign_op == TGSI_UTIL_SIGN_CLEAR) {
+         spe_andc(gen->f, result_reg, reg, bit31mask_reg);
       }
-      break;
-   case TGSI_FILE_IMMEDIATE:
-      /* xxx fall-through for now / fix */
-   case TGSI_FILE_CONSTANT:
-      /* xxx fall-through for now / fix */
-   default:
-      assert(0);
+      else if (sign_op == TGSI_UTIL_SIGN_SET) {
+         spe_and(gen->f, result_reg, reg, bit31mask_reg);
+      }
+      else {
+         assert(sign_op == TGSI_UTIL_SIGN_TOGGLE);
+         spe_xor(gen->f, result_reg, reg, bit31mask_reg);
+      }
+
+      reg = result_reg;
    }
 
    return reg;
@@ -183,11 +318,14 @@ get_dst_reg(struct codegen *gen,
             int channel,
             const struct tgsi_full_dst_register *dest)
 {
-   int reg;
+   int reg = -1;
 
    switch (dest->DstRegister.File) {
    case TGSI_FILE_TEMPORARY:
-      reg = gen->temp_regs[dest->DstRegister.Index][channel];
+      if (gen->if_nesting > 0)
+         reg = get_itemp(gen);
+      else
+         reg = gen->temp_regs[dest->DstRegister.Index][channel];
       break;
    case TGSI_FILE_OUTPUT:
       reg = get_itemp(gen);
@@ -211,19 +349,59 @@ store_dest_reg(struct codegen *gen,
                int value_reg, int channel,
                const struct tgsi_full_dst_register *dest)
 {
+   /*
+    * XXX need to implement dst reg clamping/saturation
+    */
+#if 0
+   switch (inst->Instruction.Saturate) {
+   case TGSI_SAT_NONE:
+      break;
+   case TGSI_SAT_ZERO_ONE:
+      break;
+   case TGSI_SAT_MINUS_PLUS_ONE:
+      break;
+   default:
+      assert( 0 );
+   }
+#endif
+
    switch (dest->DstRegister.File) {
    case TGSI_FILE_TEMPORARY:
-      /* no-op */
+      if (gen->if_nesting > 0) {
+         int d_reg = gen->temp_regs[dest->DstRegister.Index][channel];
+         int exec_reg = get_exec_mask_reg(gen);
+         /* Mix d with new value according to exec mask:
+          * d[i] = mask_reg[i] ? value_reg : d_reg
+          */
+         spe_selb(gen->f, d_reg, d_reg, value_reg, exec_reg);
+      }
+      else {
+         /* we're not inside a condition or loop: do nothing special */
+
+      }
       break;
    case TGSI_FILE_OUTPUT:
       {
          /* offset is measured in quadwords, not bytes */
          int offset = dest->DstRegister.Index * 4 + channel;
-         /* Store: memory[(machine_reg) + offset] = reg */
-         spe_stqd(gen->f, value_reg, gen->outputs_reg, offset);
-#if DISASSEM
-         printf("stqd\tr%d, r%d + %d\n", value_reg, gen->outputs_reg, offset);
-#endif
+         if (gen->if_nesting > 0) {
+            int exec_reg = get_exec_mask_reg(gen);
+            int curval_reg = get_itemp(gen);
+            /* First read the current value from memory:
+             * Load:  curval = memory[(machine_reg) + offset]
+             */
+            spe_lqd(gen->f, curval_reg, gen->outputs_reg, offset * 16);
+            /* Mix curval with newvalue according to exec mask:
+             * d[i] = mask_reg[i] ? value_reg : d_reg
+             */
+            spe_selb(gen->f, curval_reg, curval_reg, value_reg, exec_reg);
+            /* Store: memory[(machine_reg) + offset] = curval */
+            spe_stqd(gen->f, curval_reg, gen->outputs_reg, offset * 16);
+         }
+         else {
+            /* Store: memory[(machine_reg) + offset] = reg */
+            spe_stqd(gen->f, value_reg, gen->outputs_reg, offset * 16);
+         }
       }
       break;
    default:
@@ -232,27 +410,114 @@ store_dest_reg(struct codegen *gen,
 }
 
 
+
+static void
+emit_prologue(struct codegen *gen)
+{
+   gen->frame_size = 1024; /* XXX temporary, should be dynamic */
+
+   spe_comment(gen->f, -4, "Function prologue:");
+
+   /* save $lr on stack     # stqd $lr,16($sp) */
+   spe_stqd(gen->f, SPE_REG_RA, SPE_REG_SP, 16);
+
+   if (gen->frame_size >= 512) {
+      /* offset is too large for ai instruction */
+      int offset_reg = spe_allocate_available_register(gen->f);
+      int sp_reg = spe_allocate_available_register(gen->f);
+      /* offset = -framesize */
+      spe_load_int(gen->f, offset_reg, -gen->frame_size);
+      /* sp = $sp */
+      spe_move(gen->f, sp_reg, SPE_REG_SP);
+      /* $sp = $sp + offset_reg */
+      spe_a(gen->f, SPE_REG_SP, SPE_REG_SP, offset_reg);
+      /* save $sp in stack frame */
+      spe_stqd(gen->f, sp_reg, SPE_REG_SP, 0);
+      /* clean up */
+      spe_release_register(gen->f, offset_reg);
+      spe_release_register(gen->f, sp_reg);
+   }
+   else {
+      /* save stack pointer    # stqd $sp,-frameSize($sp) */
+      spe_stqd(gen->f, SPE_REG_SP, SPE_REG_SP, -gen->frame_size);
+
+      /* adjust stack pointer  # ai $sp,$sp,-frameSize */
+      spe_ai(gen->f, SPE_REG_SP, SPE_REG_SP, -gen->frame_size);
+   }
+}
+
+
+static void
+emit_epilogue(struct codegen *gen)
+{
+   const int return_reg = 3;
+
+   spe_comment(gen->f, -4, "Function epilogue:");
+
+   spe_comment(gen->f, 0, "return the killed mask");
+   if (gen->kill_mask_reg > 0) {
+      /* shader called KIL, return the "alive" mask */
+      spe_move(gen->f, return_reg, gen->kill_mask_reg);
+   }
+   else {
+      /* return {0,0,0,0} */
+      spe_load_uint(gen->f, return_reg, 0);
+   }
+
+   spe_comment(gen->f, 0, "restore stack and return");
+   if (gen->frame_size >= 512) {
+      /* offset is too large for ai instruction */
+      int offset_reg = spe_allocate_available_register(gen->f);
+      /* offset = framesize */
+      spe_load_int(gen->f, offset_reg, gen->frame_size);
+      /* $sp = $sp + offset */
+      spe_a(gen->f, SPE_REG_SP, SPE_REG_SP, offset_reg);
+      /* clean up */
+      spe_release_register(gen->f, offset_reg);
+   }
+   else {
+      /* restore stack pointer    # ai $sp,$sp,frameSize */
+      spe_ai(gen->f, SPE_REG_SP, SPE_REG_SP, gen->frame_size);
+   }
+
+   /* restore $lr              # lqd $lr,16($sp) */
+   spe_lqd(gen->f, SPE_REG_RA, SPE_REG_SP, 16);
+
+   /* return from function call */
+   spe_bi(gen->f, SPE_REG_RA, 0, 0);
+}
+
+
 static boolean
 emit_MOV(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, src_reg[4], dst_reg[4];
+
+   spe_comment(gen->f, -4, "MOV:");
    for (ch = 0; ch < 4; ch++) {
       if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int src_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int dst_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         /* XXX we don't always need to actually emit a mov instruction here */
-         spe_move(gen->f, dst_reg, src_reg);
-#if DISASSEM
-         printf("mov\tr%d, r%d\n", dst_reg, src_reg);
-#endif
-         store_dest_reg(gen, dst_reg, ch, &inst->FullDstRegisters[0]);
+         src_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         dst_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         if (is_register_src(gen, ch, &inst->FullSrcRegisters[0]) &&
+             is_memory_dst(gen, ch, &inst->FullDstRegisters[0])) {
+            /* special-case: register to memory store */
+            store_dest_reg(gen, src_reg[ch], ch, &inst->FullDstRegisters[0]);
+         }
+         else {
+            spe_move(gen->f, dst_reg[ch], src_reg[ch]);
+            store_dest_reg(gen, dst_reg[ch], ch, &inst->FullDstRegisters[0]);
+         }
          free_itemps(gen);
       }
    }
    return true;
 }
 
-
 /**
  * Emit addition instructions.  Recall that a single TGSI_OPCODE_ADD
  * becomes (up to) four SPU "fa" instructions because we're doing SOA
@@ -261,24 +526,25 @@ emit_MOV(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_ADD(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
-   /* Loop over Red/Green/Blue/Alpha channels */
+   int ch, s1_reg[4], s2_reg[4], d_reg[4];
+
+   spe_comment(gen->f, -4, "ADD:");
+   /* Loop over Red/Green/Blue/Alpha channels, fetch src operands */
    for (ch = 0; ch < 4; ch++) {
       /* If the dest R, G, B or A writemask is enabled... */
       if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         /* get indexes of the two src, one dest SPE registers */
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+   /* Loop over Red/Green/Blue/Alpha channels, do the add, store results */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
          /* Emit actual SPE instruction: d = s1 + s2 */
-         spe_fa(gen->f, d_reg, s1_reg, s2_reg);
-#if DISASSEM
-         printf("fa\tr%d, r%d, r%d\n", d_reg, s1_reg, s2_reg);
-#endif
-
+         spe_fa(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
          /* Store the result (a no-op for TGSI_FILE_TEMPORARY dests) */
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
          /* Free any intermediate temps we allocated */
          free_itemps(gen);
       }
@@ -286,6 +552,99 @@ emit_ADD(struct codegen *gen, const struct tgsi_full_instruction *inst)
    return true;
 }
 
+/**
+ * Emit subtract.  See emit_ADD for comments.
+ */
+static boolean
+emit_SUB(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch, s1_reg[4], s2_reg[4], d_reg[4];
+   spe_comment(gen->f, -4, "SUB:");
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         /* d = s1 - s2 */
+         spe_fs(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+   return true;
+}
+
+/**
+ * Emit multiply add.  See emit_ADD for comments.
+ */
+static boolean
+emit_MAD(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch, s1_reg[4], s2_reg[4], s3_reg[4], d_reg[4];
+   spe_comment(gen->f, -4, "MAD:");
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         /* d = s1 * s2 + s3 */
+         spe_fma(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch], s3_reg[ch]);
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+   return true;
+}
+
+
+/**
+ * Emit linear interpolate.  See emit_ADD for comments.
+ */
+static boolean
+emit_LERP(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch, s1_reg[4], s2_reg[4], s3_reg[4], d_reg[4], tmp_reg[4];
+   spe_comment(gen->f, -4, "LERP:");
+   /* setup/get src/dst/temp regs */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         tmp_reg[ch] = get_itemp(gen);
+      }
+   }
+
+   /* d = s3 + s1(s2 - s3) */
+   /* do all subtracts, then all fma, then all stores to better pipeline */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_fs(gen->f, tmp_reg[ch], s2_reg[ch], s3_reg[ch]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_fma(gen->f, d_reg[ch], tmp_reg[ch], s1_reg[ch], s3_reg[ch]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      }
+   }
+   free_itemps(gen);
+   return true;
+}
 
 /**
  * Emit multiply.  See emit_ADD for comments.
@@ -293,17 +652,63 @@ emit_ADD(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_MUL(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
+   int ch, s1_reg[4], s2_reg[4], d_reg[4];
+   spe_comment(gen->f, -4, "MUL:");
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         /* d = s1 * s2 */
+         spe_fm(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+   return true;
+}
+
+/**
+ * Emit reciprocal.  See emit_ADD for comments.
+ */
+static boolean
+emit_RCP(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
    int ch;
+   spe_comment(gen->f, -4, "RCP:");
    for (ch = 0; ch < 4; ch++) {
       if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
          int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
          int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         /* d = s1 * s2 */
-         spe_fm(gen->f, d_reg, s1_reg, s2_reg);
-#if DISASSEM
-         printf("fm\tr%d, r%d, r%d\n", d_reg, s1_reg, s2_reg);
-#endif
+         /* d = 1/s1 */
+         spe_frest(gen->f, d_reg, s1_reg);
+         spe_fi(gen->f, d_reg, s1_reg, d_reg);
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+   return true;
+}
+
+/**
+ * Emit reciprocal sqrt.  See emit_ADD for comments.
+ */
+static boolean
+emit_RSQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   spe_comment(gen->f, -4, "RSQ:");
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         /* d = 1/s1 */
+         spe_frsqest(gen->f, d_reg, s1_reg);
+         spe_fi(gen->f, d_reg, s1_reg, d_reg);
          store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
          free_itemps(gen);
       }
@@ -311,6 +716,270 @@ emit_MUL(struct codegen *gen, const struct tgsi_full_instruction *inst)
    return true;
 }
 
+/**
+ * Emit absolute value.  See emit_ADD for comments.
+ */
+static boolean
+emit_ABS(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   spe_comment(gen->f, -4, "ABS:");
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         const int bit31mask_reg = get_itemp(gen);
+
+         /* mask with bit 31 set, the rest cleared */  
+         spe_load_uint(gen->f, bit31mask_reg, (1 << 31));
+
+         /* d = sign bit cleared in s1 */
+         spe_andc(gen->f, d_reg, s1_reg, bit31mask_reg);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+   return true;
+}
+
+/**
+ * Emit 3 component dot product.  See emit_ADD for comments.
+ */
+static boolean
+emit_DP3(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   int s1x_reg, s1y_reg, s1z_reg;
+   int s2x_reg, s2y_reg, s2z_reg;
+   int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
+
+   spe_comment(gen->f, -4, "DP3:");
+
+   s1x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   s2x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   s2y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   s2z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+
+   /* t0 = x0 * x1 */
+   spe_fm(gen->f, t0_reg, s1x_reg, s2x_reg);
+
+   /* t1 = y0 * y1 */
+   spe_fm(gen->f, t1_reg, s1y_reg, s2y_reg);
+
+   /* t0 = z0 * z1 + t0 */
+   spe_fma(gen->f, t0_reg, s1z_reg, s2z_reg, t0_reg);
+
+   /* t0 = t0 + t1 */
+   spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         spe_move(gen->f, d_reg, t0_reg);
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+/**
+ * Emit 4 component dot product.  See emit_ADD for comments.
+ */
+static boolean
+emit_DP4(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   int s0x_reg, s0y_reg, s0z_reg, s0w_reg;
+   int s1x_reg, s1y_reg, s1z_reg, s1w_reg;
+   int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
+
+   spe_comment(gen->f, -4, "DP4:");
+
+   s0x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   s1x_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   s0y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   s1y_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   s0z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   s1z_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   s0w_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[0]);
+   s1w_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[1]);
+
+   /* t0 = x0 * x1 */
+   spe_fm(gen->f, t0_reg, s0x_reg, s1x_reg);
+
+   /* t1 = y0 * y1 */
+   spe_fm(gen->f, t1_reg, s0y_reg, s1y_reg);
+
+   /* t0 = z0 * z1 + t0 */
+   spe_fma(gen->f, t0_reg, s0z_reg, s1z_reg, t0_reg);
+
+   /* t1 = w0 * w1 + t1 */
+   spe_fma(gen->f, t1_reg, s0w_reg, s1w_reg, t1_reg);
+
+   /* t0 = t0 + t1 */
+   spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         spe_move(gen->f, d_reg, t0_reg);
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+/**
+ * Emit homogeneous dot product.  See emit_ADD for comments.
+ */
+static boolean
+emit_DPH(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   /* XXX rewrite this function to look more like DP3/DP4 */
+   int ch;
+   spe_comment(gen->f, -4, "DPH:");
+
+   int s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   int s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   int tmp_reg = get_itemp(gen);
+
+   /* t = x0 * x1 */
+   spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
+
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   /* t = y0 * y1 + t */
+   spe_fma(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
+
+   s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   /* t = z0 * z1 + t */
+   spe_fma(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
+
+   s2_reg = get_src_reg(gen, CHAN_W, &inst->FullSrcRegisters[1]);
+   /* t = w1 + t */
+   spe_fa(gen->f, tmp_reg, s2_reg, tmp_reg);
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         spe_move(gen->f, d_reg, tmp_reg);
+         store_dest_reg(gen, tmp_reg, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+/**
+ * Emit 3-component vector normalize.
+ */
+static boolean
+emit_NRM3(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   int src_reg[3];
+   int t0_reg = get_itemp(gen), t1_reg = get_itemp(gen);
+
+   spe_comment(gen->f, -4, "NRM3:");
+
+   src_reg[0] = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   src_reg[1] = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   src_reg[2] = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+
+   /* t0 = x * x */
+   spe_fm(gen->f, t0_reg, src_reg[0], src_reg[0]);
+
+   /* t1 = y * y */
+   spe_fm(gen->f, t1_reg, src_reg[1], src_reg[1]);
+
+   /* t0 = z * z + t0 */
+   spe_fma(gen->f, t0_reg, src_reg[2], src_reg[2], t0_reg);
+
+   /* t0 = t0 + t1 */
+   spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
+
+   /* t1 = 1.0 / sqrt(t0) */
+   spe_frsqest(gen->f, t1_reg, t0_reg);
+   spe_fi(gen->f, t1_reg, t0_reg, t1_reg);
+
+   for (ch = 0; ch < 3; ch++) {  /* NOTE: omit W channel */
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         /* dst = src[ch] * t1 */
+         spe_fm(gen->f, d_reg, src_reg[ch], t1_reg);
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+
+/**
+ * Emit cross product.  See emit_ADD for comments.
+ */
+static boolean
+emit_XPD(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   spe_comment(gen->f, -4, "XPD:");
+
+   int s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   int s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   int tmp_reg = get_itemp(gen);
+
+   /* t = z0 * y1 */
+   spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
+
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   /* t = y0 * z1 - t */
+   spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
+
+   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_X)) {
+      store_dest_reg(gen, tmp_reg, CHAN_X, &inst->FullDstRegisters[0]);
+   }
+
+   s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[1]);
+   /* t = x0 * z1 */
+   spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
+
+   s1_reg = get_src_reg(gen, CHAN_Z, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   /* t = z0 * x1 - t */
+   spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
+
+   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_Y)) {
+      store_dest_reg(gen, tmp_reg, CHAN_Y, &inst->FullDstRegisters[0]);
+   }
+
+   s1_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[1]);
+   /* t = y0 * x1 */
+   spe_fm(gen->f, tmp_reg, s1_reg, s2_reg);
+
+   s1_reg = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[0]);
+   s2_reg = get_src_reg(gen, CHAN_Y, &inst->FullSrcRegisters[1]);
+   /* t = x0 * y1 - t */
+   spe_fms(gen->f, tmp_reg, s1_reg, s2_reg, tmp_reg);
+
+   if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << CHAN_Z)) {
+      store_dest_reg(gen, tmp_reg, CHAN_Z, &inst->FullDstRegisters[0]);
+   }
+
+   free_itemps(gen);
+   return true;
+}
 
 /**
  * Emit set-if-greater-than.
@@ -323,6 +992,8 @@ emit_SGT(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch;
 
+   spe_comment(gen->f, -4, "SGT:");
+
    for (ch = 0; ch < 4; ch++) {
       if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
          int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
@@ -331,26 +1002,776 @@ emit_SGT(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
          /* d = (s1 > s2) */
          spe_fcgt(gen->f, d_reg, s1_reg, s2_reg);
-#if DISASSEM
-         printf("fcgt\tr%d, r%d, r%d\n", d_reg, s1_reg, s2_reg);
-#endif
 
          /* convert d from 0x0/0xffffffff to 0.0/1.0 */
          /* d = d & one_reg */
          spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
-#if DISASSEM
-         printf("and\tr%d, r%d, r%d\n", d_reg, d_reg, get_const_one_reg(gen));
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit set-if_less-then.  See emit_SGT for comments.
+ */
+static boolean
+emit_SLT(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "SLT:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* d = (s1 < s2) */
+         spe_fcgt(gen->f, d_reg, s2_reg, s1_reg);
+
+         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+         /* d = d & one_reg */
+         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit set-if_greater-then-or-equal.  See emit_SGT for comments.
+ */
+static boolean
+emit_SGE(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "SGE:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* d = (s1 >= s2) */
+         spe_fcgt(gen->f, d_reg, s2_reg, s1_reg);
+
+         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+         /* d = ~d & one_reg */
+         spe_andc(gen->f, d_reg, get_const_one_reg(gen), d_reg);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit set-if_less-then-or-equal.  See emit_SGT for comments.
+ */
+static boolean
+emit_SLE(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "SLE:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* d = (s1 <= s2) */
+         spe_fcgt(gen->f, d_reg, s1_reg, s2_reg);
+
+         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+         /* d = ~d & one_reg */
+         spe_andc(gen->f, d_reg, get_const_one_reg(gen), d_reg);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit set-if_equal.  See emit_SGT for comments.
+ */
+static boolean
+emit_SEQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "SEQ:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* d = (s1 == s2) */
+         spe_fceq(gen->f, d_reg, s1_reg, s2_reg);
+
+         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+         /* d = d & one_reg */
+         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit set-if_not_equal.  See emit_SGT for comments.
+ */
+static boolean
+emit_SNE(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "SNE:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* d = (s1 != s2) */
+         spe_fceq(gen->f, d_reg, s1_reg, s2_reg);
+         spe_nor(gen->f, d_reg, d_reg, d_reg);
+
+         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+         /* d = d & one_reg */
+         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit compare.  See emit_SGT for comments.
+ */
+static boolean
+emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "CMP:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         int s3_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         int zero_reg = get_itemp(gen);
+   
+         spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+
+         /* d = (s1 < 0) ? s2 : s3 */
+         spe_fcgt(gen->f, d_reg, zero_reg, s1_reg);
+         spe_selb(gen->f, d_reg, s3_reg, s2_reg, d_reg);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit trunc.  
+ * Convert float to signed int
+ * Convert signed int to float
+ */
+static boolean
+emit_TRUNC(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "TRUNC:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         /* Convert float to int */
+         spe_cflts(gen->f, d_reg, s1_reg, 0);
+
+         /* Convert int to float */
+         spe_csflt(gen->f, d_reg, d_reg, 0);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Emit floor.  
+ * If negative int subtract one
+ * Convert float to signed int
+ * Convert signed int to float
+ */
+static boolean
+emit_FLR(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "FLR:");
+
+   int zero_reg = get_itemp(gen);
+   spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+   
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         int tmp_reg = get_itemp(gen);
+
+         /* If negative, subtract 1.0 */
+         spe_fcgt(gen->f, tmp_reg, zero_reg, s1_reg);
+         spe_selb(gen->f, tmp_reg, zero_reg, get_const_one_reg(gen), tmp_reg);
+         spe_fs(gen->f, tmp_reg, s1_reg, tmp_reg);
+
+         /* Convert float to int */
+         spe_cflts(gen->f, tmp_reg, tmp_reg, 0);
+
+         /* Convert int to float */
+         spe_csflt(gen->f, d_reg, tmp_reg, 0);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+/**
+ * Compute frac = Input - FLR(Input)
+ */
+static boolean
+emit_FRC(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, "FRC:");
+
+   int zero_reg = get_itemp(gen);
+   spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         int tmp_reg = get_itemp(gen);
+
+         /* If negative, subtract 1.0 */
+         spe_fcgt(gen->f, tmp_reg, zero_reg, s1_reg);
+         spe_selb(gen->f, tmp_reg, zero_reg, get_const_one_reg(gen), tmp_reg);
+         spe_fs(gen->f, tmp_reg, s1_reg, tmp_reg);
+
+         /* Convert float to int */
+         spe_cflts(gen->f, tmp_reg, tmp_reg, 0);
+
+         /* Convert int to float */
+         spe_csflt(gen->f, tmp_reg, tmp_reg, 0);
+
+         /* d = s1 - FLR(s1) */
+         spe_fs(gen->f, d_reg, s1_reg, tmp_reg);
+
+         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+
+#if 0
+static void
+print_functions(struct cell_context *cell)
+{
+   struct cell_spu_function_info *funcs = &cell->spu_functions;
+   uint i;
+   for (i = 0; i < funcs->num; i++) {
+      printf("SPU func %u: %s at %u\n",
+             i, funcs->names[i], funcs->addrs[i]);
+   }
+}
 #endif
 
+
+static uint
+lookup_function(struct cell_context *cell, const char *funcname)
+{
+   const struct cell_spu_function_info *funcs = &cell->spu_functions;
+   uint i, addr = 0;
+   for (i = 0; i < funcs->num; i++) {
+      if (strcmp(funcs->names[i], funcname) == 0) {
+         addr = funcs->addrs[i];
+      }
+   }
+   assert(addr && "spu function not found");
+   return addr / 4;  /* discard 2 least significant bits */
+}
+
+
+/**
+ * Emit code to call a SPU function.
+ * Used to implement instructions like SIN/COS/POW/TEX/etc.
+ * If scalar, only the X components of the src regs are used, and the
+ * result is replicated across the dest register's XYZW components.
+ */
+static boolean
+emit_function_call(struct codegen *gen,
+                   const struct tgsi_full_instruction *inst,
+                   char *funcname, uint num_args, boolean scalar)
+{
+   const uint addr = lookup_function(gen->cell, funcname);
+   char comment[100];
+   int s_regs[3];
+   int func_called = FALSE;
+   uint a, ch;
+   int retval_reg = -1;
+
+   assert(num_args <= 3);
+
+   snprintf(comment, sizeof(comment), "CALL %s:", funcname);
+   spe_comment(gen->f, -4, comment);
+
+   if (scalar) {
+      for (a = 0; a < num_args; a++) {
+         s_regs[a] = get_src_reg(gen, CHAN_X, &inst->FullSrcRegisters[a]);
+      }
+      /* we'll call the function, put the return value in this register,
+       * then replicate it across all write-enabled components in d_reg.
+       */
+      retval_reg = spe_allocate_available_register(gen->f);
+   }
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int d_reg;
+         ubyte usedRegs[SPE_NUM_REGS];
+         uint i, numUsed;
+
+         if (!scalar) {
+            for (a = 0; a < num_args; a++) {
+               s_regs[a] = get_src_reg(gen, ch, &inst->FullSrcRegisters[a]);
+            }
+         }
+
+         d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         if (!scalar || !func_called) {
+            /* for a scalar function, we'll really only call the function once */
+
+            numUsed = spe_get_registers_used(gen->f, usedRegs);
+            assert(numUsed < gen->frame_size / 16 - 2);
+
+            /* save registers to stack */
+            for (i = 0; i < numUsed; i++) {
+               uint reg = usedRegs[i];
+               int offset = 2 + i;
+               spe_stqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+            }
+
+            /* setup function arguments */
+            for (a = 0; a < num_args; a++) {
+               spe_move(gen->f, 3 + a, s_regs[a]);
+            }
+
+            /* branch to function, save return addr */
+            spe_brasl(gen->f, SPE_REG_RA, addr);
+
+            /* save function's return value */
+            if (scalar)
+               spe_move(gen->f, retval_reg, 3);
+            else
+               spe_move(gen->f, d_reg, 3);
+
+            /* restore registers from stack */
+            for (i = 0; i < numUsed; i++) {
+               uint reg = usedRegs[i];
+               if (reg != d_reg && reg != retval_reg) {
+                  int offset = 2 + i;
+                  spe_lqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+               }
+            }
+
+            func_called = TRUE;
+         }
+
+         if (scalar) {
+            spe_move(gen->f, d_reg, retval_reg);
+         }
+
          store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
          free_itemps(gen);
       }
    }
 
+   if (scalar) {
+      spe_release_register(gen->f, retval_reg);
+   }
+
    return true;
 }
 
 
+static boolean
+emit_TEX(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   const uint target = inst->InstructionExtTexture.Texture;
+   const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
+   uint addr;
+   int ch;
+   int coord_regs[4], d_regs[4];
+
+   switch (target) {
+   case TGSI_TEXTURE_1D:
+   case TGSI_TEXTURE_2D:
+      addr = lookup_function(gen->cell, "spu_tex_2d");
+      break;
+   case TGSI_TEXTURE_3D:
+      addr = lookup_function(gen->cell, "spu_tex_3d");
+      break;
+   case TGSI_TEXTURE_CUBE:
+      addr = lookup_function(gen->cell, "spu_tex_cube");
+      break;
+   default:
+      ASSERT(0 && "unsupported texture target");
+      return FALSE;
+   }
+
+   assert(inst->FullSrcRegisters[1].SrcRegister.File == TGSI_FILE_SAMPLER);
+
+   spe_comment(gen->f, -4, "CALL tex:");
+
+   /* get src/dst reg info */
+   for (ch = 0; ch < 4; ch++) {
+      coord_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_regs[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+   }
+
+   {
+      ubyte usedRegs[SPE_NUM_REGS];
+      uint i, numUsed;
+
+      numUsed = spe_get_registers_used(gen->f, usedRegs);
+      assert(numUsed < gen->frame_size / 16 - 2);
+
+      /* save registers to stack */
+      for (i = 0; i < numUsed; i++) {
+         uint reg = usedRegs[i];
+         int offset = 2 + i;
+         spe_stqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+      }
+
+      /* setup function arguments (XXX depends on target) */
+      for (i = 0; i < 4; i++) {
+         spe_move(gen->f, 3 + i, coord_regs[i]);
+      }
+      spe_load_uint(gen->f, 7, unit); /* sampler unit */
+
+      /* branch to function, save return addr */
+      spe_brasl(gen->f, SPE_REG_RA, addr);
+
+      /* save function's return values (four pixel's colors) */
+      for (i = 0; i < 4; i++) {
+         spe_move(gen->f, d_regs[i], 3 + i);
+      }
+
+      /* restore registers from stack */
+      for (i = 0; i < numUsed; i++) {
+         uint reg = usedRegs[i];
+         if (reg != d_regs[0] &&
+             reg != d_regs[1] &&
+             reg != d_regs[2] &&
+             reg != d_regs[3]) {
+            int offset = 2 + i;
+            spe_lqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+         }
+      }
+   }
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         store_dest_reg(gen, d_regs[ch], ch, &inst->FullDstRegisters[0]);
+         free_itemps(gen);
+      }
+   }
+
+   return TRUE;
+}
+
+
+/**
+ * KILL if any of src reg values are less than zero.
+ */
+static boolean
+emit_KIL(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch;
+   int s_regs[4], kil_reg = -1, cmp_reg, zero_reg;
+
+   spe_comment(gen->f, -4, "CALL kil:");
+
+   /* zero = {0,0,0,0} */
+   zero_reg = get_itemp(gen);
+   spe_load_uint(gen->f, zero_reg, 0);
+
+   cmp_reg = get_itemp(gen);
+
+   /* get src regs */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      }
+   }
+
+   /* test if any src regs are < 0 */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         if (kil_reg >= 0) {
+            /* cmp = 0 > src ? : ~0 : 0 */
+            spe_fcgt(gen->f, cmp_reg, zero_reg, s_regs[ch]);
+            /* kil = kil | cmp */
+            spe_or(gen->f, kil_reg, kil_reg, cmp_reg);
+         }
+         else {
+            kil_reg = get_itemp(gen);
+            /* kil = 0 > src ? : ~0 : 0 */
+            spe_fcgt(gen->f, kil_reg, zero_reg, s_regs[ch]);
+         }
+      }
+   }
+
+   if (gen->if_nesting) {
+      /* may have been a conditional kil */
+      spe_and(gen->f, kil_reg, kil_reg, gen->exec_mask_reg);
+   }
+
+   /* allocate the kill mask reg if needed */
+   if (gen->kill_mask_reg <= 0) {
+      gen->kill_mask_reg = spe_allocate_available_register(gen->f);
+      spe_move(gen->f, gen->kill_mask_reg, kil_reg);
+   }
+   else {
+      spe_or(gen->f, gen->kill_mask_reg, gen->kill_mask_reg, kil_reg);
+   }
+
+   free_itemps(gen);
+
+   return TRUE;
+}
+
+
+
+/**
+ * Emit max.  See emit_SGT for comments.
+ */
+static boolean
+emit_MAX(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch, s0_reg[4], s1_reg[4], d_reg[4], tmp_reg[4];
+
+   spe_comment(gen->f, -4, "MAX:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         tmp_reg[ch] = get_itemp(gen);         
+      }
+   }
+
+   /* d = (s0 > s1) ? s0 : s1 */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_fcgt(gen->f, tmp_reg[ch], s0_reg[ch], s1_reg[ch]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_selb(gen->f, d_reg[ch], s1_reg[ch], s0_reg[ch], tmp_reg[ch]);
+      }
+   }
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+/**
+ * Emit max.  See emit_SGT for comments.
+ */
+static boolean
+emit_MIN(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch, s0_reg[4], s1_reg[4], d_reg[4], tmp_reg[4];
+
+   spe_comment(gen->f, -4, "MIN:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+         tmp_reg[ch] = get_itemp(gen);         
+      }
+   }
+
+   /* d = (s1 > s0) ? s0 : s1 */
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_fcgt(gen->f, tmp_reg[ch], s1_reg[ch], s0_reg[ch]);
+      }
+   }
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         spe_selb(gen->f, d_reg[ch], s1_reg[ch], s0_reg[ch], tmp_reg[ch]);
+      }
+   }
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+      }
+   }
+
+   free_itemps(gen);
+   return true;
+}
+
+static boolean
+emit_IF(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   const int channel = 0;
+   const int exec_reg = get_exec_mask_reg(gen);
+
+   spe_comment(gen->f, -4, "IF:");
+
+   /* update execution mask with the predicate register */
+   int tmp_reg = get_itemp(gen);
+   int s1_reg = get_src_reg(gen, channel, &inst->FullSrcRegisters[0]);
+
+   /* tmp = (s1_reg == 0) */
+   spe_ceqi(gen->f, tmp_reg, s1_reg, 0);
+   /* tmp = !tmp */
+   spe_complement(gen->f, tmp_reg, tmp_reg);
+   /* exec_mask = exec_mask & tmp */
+   spe_and(gen->f, exec_reg, exec_reg, tmp_reg);
+
+   gen->if_nesting++;
+
+   free_itemps(gen);
+
+   return true;
+}
+
+
+static boolean
+emit_ELSE(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   const int exec_reg = get_exec_mask_reg(gen);
+
+   spe_comment(gen->f, -4, "ELSE:");
+
+   /* exec_mask = !exec_mask */
+   spe_complement(gen->f, exec_reg, exec_reg);
+
+   return true;
+}
+
+
+static boolean
+emit_ENDIF(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   const int exec_reg = get_exec_mask_reg(gen);
+
+   spe_comment(gen->f, -4, "ENDIF:");
+
+   /* XXX todo: pop execution mask */
+
+   spe_load_int(gen->f, exec_reg, ~0x0);
+
+   gen->if_nesting--;
+   return true;
+}
+
+
+static boolean
+emit_DDX_DDY(struct codegen *gen, const struct tgsi_full_instruction *inst,
+             boolean ddx)
+{
+   int ch;
+
+   spe_comment(gen->f, -4, ddx ? "DDX:" : "DDY:");
+
+   for (ch = 0; ch < 4; ch++) {
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+         int s_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+
+         int t1_reg = get_itemp(gen);
+         int t2_reg = get_itemp(gen);
+
+         spe_splat_word(gen->f, t1_reg, s_reg, 0); /* upper-left pixel */
+         if (ddx) {
+            spe_splat_word(gen->f, t2_reg, s_reg, 1); /* upper-right pixel */
+         }
+         else {
+            spe_splat_word(gen->f, t2_reg, s_reg, 2); /* lower-left pixel */
+         }
+         spe_fs(gen->f, d_reg, t2_reg, t1_reg);
+
+         free_itemps(gen);
+      }
+   }
+
+   return true;
+}
+
+
+
+
 /**
  * Emit END instruction.
  * We just return from the shader function at this point.
@@ -361,11 +1782,8 @@ emit_SGT(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_END(struct codegen *gen)
 {
-   /* return from function call */
-   spe_bi(gen->f, SPE_REG_RA, 0, 0);
-#if DISASSEM
-   printf("bi\trRA\n");
-#endif
+   spe_comment(gen->f, -4, "END:");
+   emit_epilogue(gen);
    return true;
 }
 
@@ -379,19 +1797,101 @@ emit_instruction(struct codegen *gen,
 {
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_MOV:
+   case TGSI_OPCODE_SWZ:
       return emit_MOV(gen, inst);
    case TGSI_OPCODE_MUL:
       return emit_MUL(gen, inst);
    case TGSI_OPCODE_ADD:
       return emit_ADD(gen, inst);
+   case TGSI_OPCODE_SUB:
+      return emit_SUB(gen, inst);
+   case TGSI_OPCODE_MAD:
+      return emit_MAD(gen, inst);
+   case TGSI_OPCODE_LERP:
+      return emit_LERP(gen, inst);
+   case TGSI_OPCODE_DP3:
+      return emit_DP3(gen, inst);
+   case TGSI_OPCODE_DP4:
+      return emit_DP4(gen, inst);
+   case TGSI_OPCODE_DPH:
+      return emit_DPH(gen, inst);
+   case TGSI_OPCODE_NRM:
+      return emit_NRM3(gen, inst);
+   case TGSI_OPCODE_XPD:
+      return emit_XPD(gen, inst);
+   case TGSI_OPCODE_RCP:
+      return emit_RCP(gen, inst);
+   case TGSI_OPCODE_RSQ:
+      return emit_RSQ(gen, inst);
+   case TGSI_OPCODE_ABS:
+      return emit_ABS(gen, inst);
    case TGSI_OPCODE_SGT:
       return emit_SGT(gen, inst);
+   case TGSI_OPCODE_SLT:
+      return emit_SLT(gen, inst);
+   case TGSI_OPCODE_SGE:
+      return emit_SGE(gen, inst);
+   case TGSI_OPCODE_SLE:
+      return emit_SLE(gen, inst);
+   case TGSI_OPCODE_SEQ:
+      return emit_SEQ(gen, inst);
+   case TGSI_OPCODE_SNE:
+      return emit_SNE(gen, inst);
+   case TGSI_OPCODE_CMP:
+      return emit_CMP(gen, inst);
+   case TGSI_OPCODE_MAX:
+      return emit_MAX(gen, inst);
+   case TGSI_OPCODE_MIN:
+      return emit_MIN(gen, inst);
+   case TGSI_OPCODE_TRUNC:
+      return emit_TRUNC(gen, inst);
+   case TGSI_OPCODE_FLR:
+      return emit_FLR(gen, inst);
+   case TGSI_OPCODE_FRC:
+      return emit_FRC(gen, inst);
    case TGSI_OPCODE_END:
       return emit_END(gen);
 
+   case TGSI_OPCODE_COS:
+      return emit_function_call(gen, inst, "spu_cos", 1, TRUE);
+   case TGSI_OPCODE_SIN:
+      return emit_function_call(gen, inst, "spu_sin", 1, TRUE);
+   case TGSI_OPCODE_POW:
+      return emit_function_call(gen, inst, "spu_pow", 2, TRUE);
+   case TGSI_OPCODE_EXPBASE2:
+      return emit_function_call(gen, inst, "spu_exp2", 1, TRUE);
+   case TGSI_OPCODE_LOGBASE2:
+      return emit_function_call(gen, inst, "spu_log2", 1, TRUE);
+   case TGSI_OPCODE_TEX:
+      /* fall-through for now */
+   case TGSI_OPCODE_TXD:
+      /* fall-through for now */
+   case TGSI_OPCODE_TXB:
+      /* fall-through for now */
+   case TGSI_OPCODE_TXL:
+      /* fall-through for now */
+   case TGSI_OPCODE_TXP:
+      return emit_TEX(gen, inst);
+   case TGSI_OPCODE_KIL:
+      return emit_KIL(gen, inst);
+
+   case TGSI_OPCODE_IF:
+      return emit_IF(gen, inst);
+   case TGSI_OPCODE_ELSE:
+      return emit_ELSE(gen, inst);
+   case TGSI_OPCODE_ENDIF:
+      return emit_ENDIF(gen, inst);
+
+   case TGSI_OPCODE_DDX:
+      return emit_DDX_DDY(gen, inst, true);
+   case TGSI_OPCODE_DDY:
+      return emit_DDX_DDY(gen, inst, false);
+
    /* XXX lots more cases to do... */
 
    default:
+      fprintf(stderr, "Cell: unimplemented TGSI instruction %d!\n",
+              inst->Instruction.Opcode);
       return false;
    }
 
@@ -401,48 +1901,93 @@ emit_instruction(struct codegen *gen,
 
 
 /**
+ * Emit code for a TGSI immediate value (vector of four floats).
+ * This involves register allocation and initialization.
+ * XXX the initialization should be done by a "prepare" stage, not
+ * per quad execution!
+ */
+static boolean
+emit_immediate(struct codegen *gen, const struct tgsi_full_immediate *immed)
+{
+   int ch;
+
+   assert(gen->num_imm < MAX_TEMPS);
+
+   spe_comment(gen->f, -4, "IMMEDIATE:");
+
+   for (ch = 0; ch < 4; ch++) {
+      float val = immed->u.ImmediateFloat32[ch].Float;
+
+      if (ch > 0 && val == immed->u.ImmediateFloat32[ch - 1].Float) {
+         /* re-use previous register */
+         gen->imm_regs[gen->num_imm][ch] = gen->imm_regs[gen->num_imm][ch - 1];
+      }
+      else {
+         int reg = spe_allocate_available_register(gen->f);
+
+         if (reg < 0)
+            return false;
+
+         /* update immediate map */
+         gen->imm_regs[gen->num_imm][ch] = reg;
+
+         /* emit initializer instruction */
+         spe_load_float(gen->f, reg, val);
+      }
+   }
+
+   gen->num_imm++;
+
+   return true;
+}
+
+
+
+/**
  * Emit "code" for a TGSI declaration.
  * We only care about TGSI TEMPORARY register declarations at this time.
  * For each TGSI TEMPORARY we allocate four SPE registers.
  */
-static void
-emit_declaration(struct codegen *gen, const struct tgsi_full_declaration *decl)
+static boolean
+emit_declaration(struct cell_context *cell,
+                 struct codegen *gen, const struct tgsi_full_declaration *decl)
 {
    int i, ch;
 
    switch (decl->Declaration.File) {
    case TGSI_FILE_TEMPORARY:
-#if DISASSEM
-      printf("Declare temp reg %d .. %d\n",
-             decl->DeclarationRange.First,
-             decl->DeclarationRange.Last);
-#endif
       for (i = decl->DeclarationRange.First;
            i <= decl->DeclarationRange.Last;
            i++) {
+         assert(i < MAX_TEMPS);
          for (ch = 0; ch < 4; ch++) {
             gen->temp_regs[i][ch] = spe_allocate_available_register(gen->f);
+            if (gen->temp_regs[i][ch] < 0)
+               return false; /* out of regs */
          }
 
          /* XXX if we run out of SPE registers, we need to spill
           * to SPU memory.  someday...
           */
 
-#if DISASSEM
-         printf("  SPE regs: %d %d %d %d\n",
-                gen->temp_regs[i][0],
-                gen->temp_regs[i][1],
-                gen->temp_regs[i][2],
-                gen->temp_regs[i][3]);
-#endif
+         {
+            char buf[100];
+            sprintf(buf, "TGSI temp[%d] maps to SPU regs [$%d $%d $%d $%d]", i,
+                    gen->temp_regs[i][0], gen->temp_regs[i][1],
+                    gen->temp_regs[i][2], gen->temp_regs[i][3]);
+            spe_comment(gen->f, -4, buf);
+         }
       }
       break;
    default:
       ; /* ignore */
    }
+
+   return true;
 }
 
 
+
 /**
  * Translate TGSI shader code to SPE instructions.  This is done when
  * the state tracker gives us a new shader (via pipe->create_fs_state()).
@@ -460,6 +2005,7 @@ cell_gen_fragment_program(struct cell_context *cell,
    struct codegen gen;
 
    memset(&gen, 0, sizeof(gen));
+   gen.cell = cell;
    gen.f = f;
 
    /* For SPE function calls: reg $3 = first param, $4 = second param, etc. */
@@ -472,50 +2018,50 @@ cell_gen_fragment_program(struct cell_context *cell,
    spe_allocate_register(f, gen.outputs_reg);
    spe_allocate_register(f, gen.constants_reg);
 
-#if DISASSEM
-   printf("Begin %s\n", __FUNCTION__);
-   tgsi_dump(tokens, 0);
-#endif
+   if (cell->debug_flags & CELL_DEBUG_ASM) {
+      spe_print_code(f, true);
+      spe_indent(f, 8);
+      printf("Begin %s\n", __FUNCTION__);
+      tgsi_dump(tokens, 0);
+   }
 
    tgsi_parse_init(&parse, tokens);
 
+   emit_prologue(&gen);
+
    while (!tgsi_parse_end_of_tokens(&parse) && !gen.error) {
       tgsi_parse_token(&parse);
 
       switch (parse.FullToken.Token.Type) {
       case TGSI_TOKEN_TYPE_IMMEDIATE:
-#if 0
-         if (!note_immediate(&gen, &parse.FullToken.FullImmediate ))
-            goto fail;
-#endif
+         if (!emit_immediate(&gen, &parse.FullToken.FullImmediate))
+            gen.error = true;
          break;
 
       case TGSI_TOKEN_TYPE_DECLARATION:
-         emit_declaration(&gen, &parse.FullToken.FullDeclaration);
+         if (!emit_declaration(cell, &gen, &parse.FullToken.FullDeclaration))
+            gen.error = true;
          break;
 
       case TGSI_TOKEN_TYPE_INSTRUCTION:
-         if (!emit_instruction(&gen, &parse.FullToken.FullInstruction )) {
+         if (!emit_instruction(&gen, &parse.FullToken.FullInstruction))
             gen.error = true;
-         }
          break;
 
       default:
          assert(0);
-
       }
    }
 
-
    if (gen.error) {
       /* terminate the SPE code */
       return emit_END(&gen);
    }
 
-#if DISASSEM
-   printf("cell_gen_fragment_program nr instructions: %d\n", f->num_inst);
-   printf("End %s\n", __FUNCTION__);
-#endif
+   if (cell->debug_flags & CELL_DEBUG_ASM) {
+      printf("cell_gen_fragment_program nr instructions: %d\n", f->num_inst);
+      printf("End %s\n", __FUNCTION__);
+   }
 
    tgsi_parse_free( &parse );
 
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
index 06219d4e980..2c64eb1bccd 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c
@@ -54,12 +54,17 @@
  * \param ifragZ_reg  register containing integer fragment Z values (in)
  * \param ifbZ_reg    register containing integer frame buffer Z values (in/out)
  * \param zmask_reg   register containing result of Z test/comparison (out)
+ *
+ * Returns true if the Z-buffer needs to be updated.
  */
-static void
-gen_depth_test(const struct pipe_depth_stencil_alpha_state *dsa,
-               struct spe_function *f,
+static boolean
+gen_depth_test(struct spe_function *f,
+               const struct pipe_depth_stencil_alpha_state *dsa,
                int mask_reg, int ifragZ_reg, int ifbZ_reg, int zmask_reg)
 {
+   /* NOTE: we use clgt below, not cgt, because we want to compare _unsigned_
+    * quantities.  This only makes a difference for 32-bit Z values though.
+    */
    ASSERT(dsa->depth.enabled);
 
    switch (dsa->depth.func) {
@@ -79,28 +84,28 @@ gen_depth_test(const struct pipe_depth_stencil_alpha_state *dsa,
 
    case PIPE_FUNC_GREATER:
       /* zmask = (ifragZ > ref) */
-      spe_cgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      spe_clgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
       /* mask = (mask & zmask) */
       spe_and(f, mask_reg, mask_reg, zmask_reg);
       break;
 
    case PIPE_FUNC_LESS:
       /* zmask = (ref > ifragZ) */
-      spe_cgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
+      spe_clgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
       /* mask = (mask & zmask) */
       spe_and(f, mask_reg, mask_reg, zmask_reg);
       break;
 
    case PIPE_FUNC_LEQUAL:
       /* zmask = (ifragZ > ref) */
-      spe_cgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
+      spe_clgt(f, zmask_reg, ifragZ_reg, ifbZ_reg);
       /* mask = (mask & ~zmask) */
       spe_andc(f, mask_reg, mask_reg, zmask_reg);
       break;
 
    case PIPE_FUNC_GEQUAL:
       /* zmask = (ref > ifragZ) */
-      spe_cgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
+      spe_clgt(f, zmask_reg, ifbZ_reg, ifragZ_reg);
       /* mask = (mask & ~zmask) */
       spe_andc(f, mask_reg, mask_reg, zmask_reg);
       break;
@@ -129,7 +134,10 @@ gen_depth_test(const struct pipe_depth_stencil_alpha_state *dsa,
        * framebufferZ = (ztest_passed ? fragmentZ : framebufferZ;
        */
       spe_selb(f, ifbZ_reg, ifbZ_reg, ifragZ_reg, mask_reg);
+      return true;
    }
+
+   return false;
 }
 
 
@@ -229,7 +237,40 @@ gen_alpha_test(const struct pipe_depth_stencil_alpha_state *dsa,
    spe_release_register(f, amask_reg);
 }
 
+/* This pair of functions is used inline to allocate and deallocate
+ * optional constant registers.  Once a constant is discovered to be 
+ * needed, we will likely need it again, so we don't want to deallocate
+ * it and have to allocate and load it again unnecessarily.
+ */
+static inline void
+setup_optional_register(struct spe_function *f, boolean *is_already_set, unsigned int *r)
+{
+   if (*is_already_set) return;
+   *r = spe_allocate_available_register(f);
+   *is_already_set = true;
+}
+
+static inline void
+release_optional_register(struct spe_function *f, boolean *is_already_set, unsigned int r)
+{
+    if (!*is_already_set) return;
+    spe_release_register(f, r);
+    *is_already_set = false;
+}
+
+static inline void
+setup_const_register(struct spe_function *f, boolean *is_already_set, unsigned int *r, float value)
+{
+   if (*is_already_set) return;
+   setup_optional_register(f, is_already_set, r);
+   spe_load_float(f, *r, value);
+}
 
+static inline void
+release_const_register(struct spe_function *f, boolean *is_already_set, unsigned int r)
+{
+    release_optional_register(f, is_already_set, r);
+}
 
 /**
  * Generate SPE code to implement the given blend mode for a quad of pixels.
@@ -242,6 +283,7 @@ gen_alpha_test(const struct pipe_depth_stencil_alpha_state *dsa,
  */
 static void
 gen_blend(const struct pipe_blend_state *blend,
+          const struct pipe_blend_color *blend_color,
           struct spe_function *f,
           enum pipe_format color_format,
           int fragR_reg, int fragG_reg, int fragB_reg, int fragA_reg,
@@ -262,10 +304,17 @@ gen_blend(const struct pipe_blend_state *blend,
    int fbB_reg = spe_allocate_available_register(f);
    int fbA_reg = spe_allocate_available_register(f);
 
-   int one_reg = spe_allocate_available_register(f);
    int tmp_reg = spe_allocate_available_register(f);
 
-   boolean one_reg_set = false; /* avoid setting one_reg more than once */
+   /* Optional constant registers we might or might not end up using;
+    * if we do use them, make sure we only allocate them once by
+    * keeping a flag on each one.
+    */
+   boolean one_reg_set = false;
+   unsigned int one_reg;
+   boolean constR_reg_set = false, constG_reg_set = false, 
+      constB_reg_set = false, constA_reg_set = false;
+   unsigned int constR_reg, constG_reg, constB_reg, constA_reg;
 
    ASSERT(blend->blend_enable);
 
@@ -346,127 +395,449 @@ gen_blend(const struct pipe_blend_state *blend,
       spe_release_register(f, mask_reg);
    }
 
-
    /*
-    * Compute Src RGB terms
+    * Compute Src RGB terms.  We're actually looking for the value
+    * of (the appropriate RGB factors) * (the incoming source RGB color),
+    * because in some cases (like PIPE_BLENDFACTOR_ONE and 
+    * PIPE_BLENDFACTOR_ZERO) we can avoid doing unnecessary math.
     */
    switch (blend->rgb_src_factor) {
    case PIPE_BLENDFACTOR_ONE:
+      /* factors = (1,1,1), so term = (R,G,B) */
       spe_move(f, term1R_reg, fragR_reg);
       spe_move(f, term1G_reg, fragG_reg);
       spe_move(f, term1B_reg, fragB_reg);
       break;
    case PIPE_BLENDFACTOR_ZERO:
-      spe_zero(f, term1R_reg);
-      spe_zero(f, term1G_reg);
-      spe_zero(f, term1B_reg);
+      /* factors = (0,0,0), so term = (0,0,0) */
+      spe_load_float(f, term1R_reg, 0.0f);
+      spe_load_float(f, term1G_reg, 0.0f);
+      spe_load_float(f, term1B_reg, 0.0f);
       break;
    case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* factors = (R,G,B), so term = (R*R, G*G, B*B) */
       spe_fm(f, term1R_reg, fragR_reg, fragR_reg);
       spe_fm(f, term1G_reg, fragG_reg, fragG_reg);
       spe_fm(f, term1B_reg, fragB_reg, fragB_reg);
       break;
    case PIPE_BLENDFACTOR_SRC_ALPHA:
+      /* factors = (A,A,A), so term = (R*A, G*A, B*A) */
       spe_fm(f, term1R_reg, fragR_reg, fragA_reg);
       spe_fm(f, term1G_reg, fragG_reg, fragA_reg);
       spe_fm(f, term1B_reg, fragB_reg, fragA_reg);
       break;
-      /* XXX more cases */
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* factors = (1-R,1-G,1-B), so term = (R*(1-R), G*(1-G), B*(1-B)) 
+       * or in other words term = (R-R*R, G-G*G, B-B*B)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, fragR_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, fragG_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, fragB_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* factors = (Rfb,Gfb,Bfb), so term = (R*Rfb, G*Gfb, B*Bfb) */
+      spe_fm(f, term1R_reg, fragR_reg, fbR_reg);
+      spe_fm(f, term1G_reg, fragG_reg, fbG_reg);
+      spe_fm(f, term1B_reg, fragB_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* factors = (1-Rfb,1-Gfb,1-Bfb), so term = (R*(1-Rfb),G*(1-Gfb),B*(1-Bfb))
+       * or term = (R-R*Rfb, G-G*Gfb, B-B*Bfb)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, fbR_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, fbG_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, fbB_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+      /* factors = (1-A,1-A,1-A), so term = (R*(1-A),G*(1-A),B*(1-A))
+       * or term = (R-R*A,G-G*A,B-B*A)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, fragA_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, fragA_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, fragA_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+      /* factors = (Afb, Afb, Afb), so term = (R*Afb, G*Afb, B*Afb) */
+      spe_fm(f, term1R_reg, fragR_reg, fbA_reg);
+      spe_fm(f, term1G_reg, fragG_reg, fbA_reg);
+      spe_fm(f, term1B_reg, fragB_reg, fbA_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+      /* factors = (1-Afb, 1-Afb, 1-Afb), so term = (R*(1-Afb),G*(1-Afb),B*(1-Afb)) 
+       * or term = (R-R*Afb,G-G*Afb,b-B*Afb)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, fbA_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, fbA_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, fbA_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* now, factor = (Rc,Gc,Bc), so term = (R*Rc,G*Gc,B*Bc) */
+      spe_fm(f, term1R_reg, fragR_reg, constR_reg);
+      spe_fm(f, term1G_reg, fragG_reg, constG_reg);
+      spe_fm(f, term1B_reg, fragB_reg, constB_reg);
+      break;
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+      /* we'll need the optional constant alpha register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = (Ac,Ac,Ac), so term = (R*Ac,G*Ac,B*Ac) */
+      spe_fm(f, term1R_reg, fragR_reg, constA_reg);
+      spe_fm(f, term1G_reg, fragG_reg, constA_reg);
+      spe_fm(f, term1B_reg, fragB_reg, constA_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* factor = (1-Rc,1-Gc,1-Bc), so term = (R*(1-Rc),G*(1-Gc),B*(1-Bc)) 
+       * or term = (R-R*Rc, G-G*Gc, B-B*Bc)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, constR_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, constG_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, constB_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* factor = (1-Ac,1-Ac,1-Ac), so term = (R*(1-Ac),G*(1-Ac),B*(1-Ac))
+       * or term = (R-R*Ac,G-G*Ac,B-B*Ac)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term1R_reg, fragR_reg, constA_reg, fragR_reg);
+      spe_fnms(f, term1G_reg, fragG_reg, constA_reg, fragG_reg);
+      spe_fnms(f, term1B_reg, fragB_reg, constA_reg, fragB_reg);
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+      /* We'll need the optional {1,1,1,1} register */
+      setup_const_register(f, &one_reg_set, &one_reg, 1.0f);
+      /* factor = (min(A,1-Afb),min(A,1-Afb),min(A,1-Afb)), so 
+       * term = (R*min(A,1-Afb), G*min(A,1-Afb), B*min(A,1-Afb))
+       * We could expand the term (as a*min(b,c) == min(a*b,a*c)
+       * as long as a is positive), but then we'd have to do three
+       * spe_float_min() functions instead of one, so this is simpler.
+       */
+      /* tmp = 1 - Afb */
+      spe_fs(f, tmp_reg, one_reg, fbA_reg);
+      /* tmp = min(A,tmp) */
+      spe_float_min(f, tmp_reg, fragA_reg, tmp_reg);
+      /* term = R*tmp */
+      spe_fm(f, term1R_reg, fragR_reg, tmp_reg);
+      spe_fm(f, term1G_reg, fragG_reg, tmp_reg);
+      spe_fm(f, term1B_reg, fragB_reg, tmp_reg);
+      break;
+
+      /* These are special D3D cases involving a second color output
+       * from the fragment shader.  I'm not sure we can support them
+       * yet... XXX
+       */
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
+
    default:
       ASSERT(0);
    }
 
    /*
-    * Compute Src Alpha term
+    * Compute Src Alpha term.  Like the above, we're looking for
+    * the full term A*factor, not just the factor itself, because
+    * in many cases we can avoid doing unnecessary multiplies.
     */
    switch (blend->alpha_src_factor) {
+   case PIPE_BLENDFACTOR_ZERO:
+      /* factor = 0, so term = 0 */
+      spe_load_float(f, term1A_reg, 0.0f);
+      break;
+
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: /* fall through */
    case PIPE_BLENDFACTOR_ONE:
+      /* factor = 1, so term = A */
       spe_move(f, term1A_reg, fragA_reg);
       break;
+
    case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* factor = A, so term = A*A */
       spe_fm(f, term1A_reg, fragA_reg, fragA_reg);
       break;
    case PIPE_BLENDFACTOR_SRC_ALPHA:
       spe_fm(f, term1A_reg, fragA_reg, fragA_reg);
       break;
-      /* XXX more cases */
+
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* factor = 1-A, so term = A*(1-A) = A-A*A */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term1A_reg, fragA_reg, fragA_reg, fragA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_DST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* factor = Afb, so term = A*Afb */
+      spe_fm(f, term1A_reg, fragA_reg, fbA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* factor = 1-Afb, so term = A*(1-Afb) = A - A*Afb */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term1A_reg, fragA_reg, fbA_reg, fragA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_CONST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* We need the optional constA_reg register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = Ac, so term = A*Ac */
+      spe_fm(f, term1A_reg, fragA_reg, constA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* We need the optional constA_reg register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = 1-Ac, so term = A*(1-Ac) = A-A*Ac */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term1A_reg, fragA_reg, constA_reg, fragA_reg);
+      break;
+
+      /* These are special D3D cases involving a second color output
+       * from the fragment shader.  I'm not sure we can support them
+       * yet... XXX
+       */
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
    default:
       ASSERT(0);
    }
 
    /*
-    * Compute Dest RGB terms
+    * Compute Dest RGB term.  Like the above, we're looking for
+    * the full term (Rfb,Gfb,Bfb)*(factor), not just the factor itself, because
+    * in many cases we can avoid doing unnecessary multiplies.
     */
    switch (blend->rgb_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
+      /* factors = (1,1,1), so term = (Rfb,Gfb,Bfb) */
       spe_move(f, term2R_reg, fbR_reg);
       spe_move(f, term2G_reg, fbG_reg);
       spe_move(f, term2B_reg, fbB_reg);
       break;
    case PIPE_BLENDFACTOR_ZERO:
-      spe_zero(f, term2R_reg);
-      spe_zero(f, term2G_reg);
-      spe_zero(f, term2B_reg);
+      /* factor s= (0,0,0), so term = (0,0,0) */
+      spe_load_float(f, term2R_reg, 0.0f);
+      spe_load_float(f, term2G_reg, 0.0f);
+      spe_load_float(f, term2B_reg, 0.0f);
       break;
    case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* factors = (R,G,B), so term = (R*Rfb, G*Gfb, B*Bfb) */
       spe_fm(f, term2R_reg, fbR_reg, fragR_reg);
       spe_fm(f, term2G_reg, fbG_reg, fragG_reg);
       spe_fm(f, term2B_reg, fbB_reg, fragB_reg);
       break;
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* factors = (1-R,1-G,1-B), so term = (Rfb*(1-R), Gfb*(1-G), Bfb*(1-B)) 
+       * or in other words term = (Rfb-Rfb*R, Gfb-Gfb*G, Bfb-Bfb*B)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term2R_reg, fragR_reg, fbR_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fragG_reg, fbG_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fragB_reg, fbB_reg, fbB_reg);
+      break;
    case PIPE_BLENDFACTOR_SRC_ALPHA:
+      /* factors = (A,A,A), so term = (Rfb*A, Gfb*A, Bfb*A) */
       spe_fm(f, term2R_reg, fbR_reg, fragA_reg);
       spe_fm(f, term2G_reg, fbG_reg, fragA_reg);
       spe_fm(f, term2B_reg, fbB_reg, fragA_reg);
       break;
    case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-      /* one = {1.0, 1.0, 1.0, 1.0} */
-      if (!one_reg_set) {
-         spe_load_float(f, one_reg, 1.0f);
-         one_reg_set = true;
-      }
-      /* tmp = one - fragA */
-      spe_fs(f, tmp_reg, one_reg, fragA_reg);
-      /* term = fb * tmp */
-      spe_fm(f, term2R_reg, fbR_reg, tmp_reg);
-      spe_fm(f, term2G_reg, fbG_reg, tmp_reg);
-      spe_fm(f, term2B_reg, fbB_reg, tmp_reg);
-      break;
-      /* XXX more cases */
+      /* factors = (1-A,1-A,1-A) so term = (Rfb-Rfb*A,Gfb-Gfb*A,Bfb-Bfb*A) */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term2R_reg, fbR_reg, fragA_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fbG_reg, fragA_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fbB_reg, fragA_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* factors = (Rfb,Gfb,Bfb), so term = (Rfb*Rfb, Gfb*Gfb, Bfb*Bfb) */
+      spe_fm(f, term2R_reg, fbR_reg, fbR_reg);
+      spe_fm(f, term2G_reg, fbG_reg, fbG_reg);
+      spe_fm(f, term2B_reg, fbB_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* factors = (1-Rfb,1-Gfb,1-Bfb), so term = (Rfb*(1-Rfb),Gfb*(1-Gfb),Bfb*(1-Bfb))
+       * or term = (Rfb-Rfb*Rfb, Gfb-Gfb*Gfb, Bfb-Bfb*Bfb)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term2R_reg, fbR_reg, fbR_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fbG_reg, fbG_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fbB_reg, fbB_reg, fbB_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_DST_ALPHA:
+      /* factors = (Afb, Afb, Afb), so term = (Rfb*Afb, Gfb*Afb, Bfb*Afb) */
+      spe_fm(f, term2R_reg, fbR_reg, fbA_reg);
+      spe_fm(f, term2G_reg, fbG_reg, fbA_reg);
+      spe_fm(f, term2B_reg, fbB_reg, fbA_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+      /* factors = (1-Afb, 1-Afb, 1-Afb), so term = (Rfb*(1-Afb),Gfb*(1-Afb),Bfb*(1-Afb)) 
+       * or term = (Rfb-Rfb*Afb,Gfb-Gfb*Afb,Bfb-Bfb*Afb)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term2R_reg, fbR_reg, fbA_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fbG_reg, fbA_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fbB_reg, fbA_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* now, factor = (Rc,Gc,Bc), so term = (Rfb*Rc,Gfb*Gc,Bfb*Bc) */
+      spe_fm(f, term2R_reg, fbR_reg, constR_reg);
+      spe_fm(f, term2G_reg, fbG_reg, constG_reg);
+      spe_fm(f, term2B_reg, fbB_reg, constB_reg);
+      break;
+   case PIPE_BLENDFACTOR_CONST_ALPHA:
+      /* we'll need the optional constant alpha register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = (Ac,Ac,Ac), so term = (Rfb*Ac,Gfb*Ac,Bfb*Ac) */
+      spe_fm(f, term2R_reg, fbR_reg, constA_reg);
+      spe_fm(f, term2G_reg, fbG_reg, constA_reg);
+      spe_fm(f, term2B_reg, fbB_reg, constA_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* factor = (1-Rc,1-Gc,1-Bc), so term = (Rfb*(1-Rc),Gfb*(1-Gc),Bfb*(1-Bc)) 
+       * or term = (Rfb-Rfb*Rc, Gfb-Gfb*Gc, Bfb-Bfb*Bc)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term2R_reg, fbR_reg, constR_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fbG_reg, constG_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fbB_reg, constB_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+      /* We need the optional constant color registers */
+      setup_const_register(f, &constR_reg_set, &constR_reg, blend_color->color[0]);
+      setup_const_register(f, &constG_reg_set, &constG_reg, blend_color->color[1]);
+      setup_const_register(f, &constB_reg_set, &constB_reg, blend_color->color[2]);
+      /* factor = (1-Ac,1-Ac,1-Ac), so term = (Rfb*(1-Ac),Gfb*(1-Ac),Bfb*(1-Ac))
+       * or term = (Rfb-Rfb*Ac,Gfb-Gfb*Ac,Bfb-Bfb*Ac)
+       * fnms(a,b,c,d) computes a = d - b*c
+       */
+      spe_fnms(f, term2R_reg, fbR_reg, constA_reg, fbR_reg);
+      spe_fnms(f, term2G_reg, fbG_reg, constA_reg, fbG_reg);
+      spe_fnms(f, term2B_reg, fbB_reg, constA_reg, fbB_reg);
+      break;
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: /* not supported for dest RGB */
+      ASSERT(0);
+      break;
+
+      /* These are special D3D cases involving a second color output
+       * from the fragment shader.  I'm not sure we can support them
+       * yet... XXX
+       */
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
+
    default:
       ASSERT(0);
    }
 
    /*
-    * Compute Dest Alpha term
+    * Compute Dest Alpha term.  Like the above, we're looking for
+    * the full term Afb*factor, not just the factor itself, because
+    * in many cases we can avoid doing unnecessary multiplies.
     */
    switch (blend->alpha_dst_factor) {
    case PIPE_BLENDFACTOR_ONE:
+      /* factor = 1, so term = Afb */
       spe_move(f, term2A_reg, fbA_reg);
       break;
    case PIPE_BLENDFACTOR_ZERO:
-      spe_zero(f, term2A_reg);
+      /* factor = 0, so term = 0 */
+      spe_load_float(f, term2A_reg, 0.0f);
       break;
-   case PIPE_BLENDFACTOR_SRC_ALPHA:
+
+   case PIPE_BLENDFACTOR_SRC_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_SRC_COLOR:
+      /* factor = A, so term = Afb*A */
       spe_fm(f, term2A_reg, fbA_reg, fragA_reg);
       break;
-   case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
-      /* one = {1.0, 1.0, 1.0, 1.0} */
-      if (!one_reg_set) {
-         spe_load_float(f, one_reg, 1.0f);
-         one_reg_set = true;
-      }
-      /* tmp = one - fragA */
-      spe_fs(f, tmp_reg, one_reg, fragA_reg);
-      /* termA = fbA * tmp */
-      spe_fm(f, term2A_reg, fbA_reg, tmp_reg);
+
+   case PIPE_BLENDFACTOR_INV_SRC_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+      /* factor = 1-A, so term = Afb*(1-A) = Afb-Afb*A */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term2A_reg, fbA_reg, fragA_reg, fbA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_DST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_DST_COLOR:
+      /* factor = Afb, so term = Afb*Afb */
+      spe_fm(f, term2A_reg, fbA_reg, fbA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_INV_DST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_DST_COLOR:
+      /* factor = 1-Afb, so term = Afb*(1-Afb) = Afb - Afb*Afb */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term2A_reg, fbA_reg, fbA_reg, fbA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_CONST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_CONST_COLOR:
+      /* We need the optional constA_reg register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = Ac, so term = Afb*Ac */
+      spe_fm(f, term2A_reg, fbA_reg, constA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_INV_CONST_ALPHA: /* fall through */
+   case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+      /* We need the optional constA_reg register */
+      setup_const_register(f, &constA_reg_set, &constA_reg, blend_color->color[3]);
+      /* factor = 1-Ac, so term = Afb*(1-Ac) = Afb-Afb*Ac */
+      /* fnms(a,b,c,d) computes a = d - b*c */
+      spe_fnms(f, term2A_reg, fbA_reg, constA_reg, fbA_reg);
+      break;
+
+   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: /* not supported for dest alpha */
+      ASSERT(0);
       break;
-      /* XXX more cases */
+
+      /* These are special D3D cases involving a second color output
+       * from the fragment shader.  I'm not sure we can support them
+       * yet... XXX
+       */
+   case PIPE_BLENDFACTOR_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_SRC1_ALPHA:
+   case PIPE_BLENDFACTOR_INV_SRC1_COLOR:
+   case PIPE_BLENDFACTOR_INV_SRC1_ALPHA:
    default:
       ASSERT(0);
    }
 
    /*
-    * Combine Src/Dest RGB terms
+    * Combine Src/Dest RGB terms as per the blend equation.
     */
    switch (blend->rgb_func) {
    case PIPE_BLEND_ADD:
@@ -479,7 +850,21 @@ gen_blend(const struct pipe_blend_state *blend,
       spe_fs(f, fragG_reg, term1G_reg, term2G_reg);
       spe_fs(f, fragB_reg, term1B_reg, term2B_reg);
       break;
-      /* XXX more cases */
+   case PIPE_BLEND_REVERSE_SUBTRACT:
+      spe_fs(f, fragR_reg, term2R_reg, term1R_reg);
+      spe_fs(f, fragG_reg, term2G_reg, term1G_reg);
+      spe_fs(f, fragB_reg, term2B_reg, term1B_reg);
+      break;
+   case PIPE_BLEND_MIN:
+      spe_float_min(f, fragR_reg, term1R_reg, term2R_reg);
+      spe_float_min(f, fragG_reg, term1G_reg, term2G_reg);
+      spe_float_min(f, fragB_reg, term1B_reg, term2B_reg);
+      break;
+   case PIPE_BLEND_MAX:
+      spe_float_max(f, fragR_reg, term1R_reg, term2R_reg);
+      spe_float_max(f, fragG_reg, term1G_reg, term2G_reg);
+      spe_float_max(f, fragB_reg, term1B_reg, term2B_reg);
+      break;
    default:
       ASSERT(0);
    }
@@ -494,7 +879,15 @@ gen_blend(const struct pipe_blend_state *blend,
    case PIPE_BLEND_SUBTRACT:
       spe_fs(f, fragA_reg, term1A_reg, term2A_reg);
       break;
-      /* XXX more cases */
+   case PIPE_BLEND_REVERSE_SUBTRACT:
+      spe_fs(f, fragA_reg, term2A_reg, term1A_reg);
+      break;
+   case PIPE_BLEND_MIN:
+      spe_float_min(f, fragA_reg, term1A_reg, term2A_reg);
+      break;
+   case PIPE_BLEND_MAX:
+      spe_float_max(f, fragA_reg, term1A_reg, term2A_reg);
+      break;
    default:
       ASSERT(0);
    }
@@ -514,8 +907,14 @@ gen_blend(const struct pipe_blend_state *blend,
    spe_release_register(f, fbB_reg);
    spe_release_register(f, fbA_reg);
 
-   spe_release_register(f, one_reg);
    spe_release_register(f, tmp_reg);
+
+   /* Free any optional registers that actually got used */
+   release_const_register(f, &one_reg_set, one_reg);
+   release_const_register(f, &constR_reg_set, constR_reg);
+   release_const_register(f, &constG_reg_set, constG_reg);
+   release_const_register(f, &constB_reg_set, constB_reg);
+   release_const_register(f, &constA_reg_set, constA_reg);
 }
 
 
@@ -524,24 +923,74 @@ gen_logicop(const struct pipe_blend_state *blend,
             struct spe_function *f,
             int fragRGBA_reg, int fbRGBA_reg)
 {
-   /* XXX to-do */
-   /* operate on 32-bit packed pixels, not float colors */
-}
-
-
-static void
-gen_colormask(uint colormask,
-              struct spe_function *f,
-              int fragRGBA_reg, int fbRGBA_reg)
-{
-   /* XXX to-do */
-   /* operate on 32-bit packed pixels, not float colors */
+   /* We've got four 32-bit RGBA packed pixels in each of
+    * fragRGBA_reg and fbRGBA_reg, not sets of floating-point
+    * reds, greens, blues, and alphas.
+    * */
+   ASSERT(blend->logicop_enable);
+
+   switch(blend->logicop_func) {
+      case PIPE_LOGICOP_CLEAR: /* 0 */
+         spe_zero(f, fragRGBA_reg);
+         break;
+      case PIPE_LOGICOP_NOR: /* ~(s | d) */
+         spe_nor(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_AND_INVERTED: /* ~s & d */
+         /* andc R, A, B computes R = A & ~B */
+         spe_andc(f, fragRGBA_reg, fbRGBA_reg, fragRGBA_reg);
+         break;
+      case PIPE_LOGICOP_COPY_INVERTED: /* ~s */
+         spe_complement(f, fragRGBA_reg, fragRGBA_reg);
+         break;
+      case PIPE_LOGICOP_AND_REVERSE: /* s & ~d */
+         /* andc R, A, B computes R = A & ~B */
+         spe_andc(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_INVERT: /* ~d */
+         /* Note that (A nor A) == ~(A|A) == ~A */
+         spe_nor(f, fragRGBA_reg, fbRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_XOR: /* s ^ d */
+         spe_xor(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_NAND: /* ~(s & d) */
+         spe_nand(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_AND: /* s & d */
+         spe_and(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_EQUIV: /* ~(s ^ d) */
+         spe_xor(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         spe_complement(f, fragRGBA_reg, fragRGBA_reg);
+         break;
+      case PIPE_LOGICOP_NOOP: /* d */
+         spe_move(f, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_OR_INVERTED: /* ~s | d */
+         /* orc R, A, B computes R = A | ~B */
+         spe_orc(f, fragRGBA_reg, fbRGBA_reg, fragRGBA_reg);
+         break;
+      case PIPE_LOGICOP_COPY: /* s */
+         break;
+      case PIPE_LOGICOP_OR_REVERSE: /* s | ~d */
+         /* orc R, A, B computes R = A | ~B */
+         spe_orc(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_OR: /* s | d */
+         spe_or(f, fragRGBA_reg, fragRGBA_reg, fbRGBA_reg);
+         break;
+      case PIPE_LOGICOP_SET: /* 1 */
+         spe_load_int(f, fragRGBA_reg, 0xffffffff);
+         break;
+      default:
+         ASSERT(0);
+   }
 }
 
 
-
 /**
- * Generate code to pack a quad of float colors into a four 32-bit integers.
+ * Generate code to pack a quad of float colors into four 32-bit integers.
  *
  * \param f             SPE function to append instruction onto.
  * \param color_format  the dest color packing format
@@ -557,13 +1006,16 @@ gen_pack_colors(struct spe_function *f,
                 int r_reg, int g_reg, int b_reg, int a_reg,
                 int rgba_reg)
 {
+   int rg_reg = spe_allocate_available_register(f);
+   int ba_reg = spe_allocate_available_register(f);
+
    /* Convert float[4] in [0.0,1.0] to int[4] in [0,~0], with clamping */
    spe_cfltu(f, r_reg, r_reg, 32);
    spe_cfltu(f, g_reg, g_reg, 32);
    spe_cfltu(f, b_reg, b_reg, 32);
    spe_cfltu(f, a_reg, a_reg, 32);
 
-   /* Shift the most significant bytes to least the significant positions.
+   /* Shift the most significant bytes to the least significant positions.
     * I.e.: reg = reg >> 24
     */
    spe_rotmi(f, r_reg, r_reg, -24);
@@ -595,12 +1047,713 @@ gen_pack_colors(struct spe_function *f,
     * OR-ing all those together gives us four packed colors:
     *  RGBA = {0xffffffff, 0xaa114477, 0xbb225588, 0xcc336699}
     */
-   spe_or(f, rgba_reg, r_reg, g_reg);
-   spe_or(f, rgba_reg, rgba_reg, b_reg);
-   spe_or(f, rgba_reg, rgba_reg, a_reg);
+   spe_or(f, rg_reg, r_reg, g_reg);
+   spe_or(f, ba_reg, a_reg, b_reg);
+   spe_or(f, rgba_reg, rg_reg, ba_reg);
+
+   spe_release_register(f, rg_reg);
+   spe_release_register(f, ba_reg);
+}
+
+static void
+gen_colormask(struct spe_function *f,
+              uint colormask,
+              enum pipe_format color_format,
+              int fragRGBA_reg, int fbRGBA_reg)
+{
+   /* We've got four 32-bit RGBA packed pixels in each of
+    * fragRGBA_reg and fbRGBA_reg, not sets of floating-point
+    * reds, greens, blues, and alphas.  Further, the pixels
+    * are packed according to the given color format, not
+    * necessarily RGBA...
+    */
+   unsigned int r_mask;
+   unsigned int g_mask;
+   unsigned int b_mask;
+   unsigned int a_mask;
+
+   /* Calculate exactly where the bits for any particular color
+    * end up, so we can mask them correctly.
+    */
+   switch(color_format) {
+      case PIPE_FORMAT_A8R8G8B8_UNORM:
+         /* ARGB */
+         a_mask = 0xff000000;
+         r_mask = 0x00ff0000;
+         g_mask = 0x0000ff00;
+         b_mask = 0x000000ff;
+         break;
+      case PIPE_FORMAT_B8G8R8A8_UNORM:
+         /* BGRA */
+         b_mask = 0xff000000;
+         g_mask = 0x00ff0000;
+         r_mask = 0x0000ff00;
+         a_mask = 0x000000ff;
+         break;
+      default:
+         ASSERT(0);
+   }
+
+   /* For each R, G, B, and A component we're supposed to mask out, 
+    * clear its bits.   Then our mask operation later will work 
+    * as expected.
+    */
+   if (!(colormask & PIPE_MASK_R)) {
+      r_mask = 0;
+   }
+   if (!(colormask & PIPE_MASK_G)) {
+      g_mask = 0;
+   }
+   if (!(colormask & PIPE_MASK_B)) {
+      b_mask = 0;
+   }
+   if (!(colormask & PIPE_MASK_A)) {
+      a_mask = 0;
+   }
+
+   /* Get a temporary register to hold the mask that will be applied to the fragment */
+   int colormask_reg = spe_allocate_available_register(f);
+
+   /* The actual mask we're going to use is an OR of the remaining R, G, B, and A
+    * masks.  Load the result value into our temporary register.
+    */
+   spe_load_uint(f, colormask_reg, r_mask | g_mask | b_mask | a_mask);
+
+   /* Use the mask register to select between the fragment color
+    * values and the frame buffer color values.  Wherever the
+    * mask has a 0 bit, the current frame buffer color should override
+    * the fragment color.  Wherever the mask has a 1 bit, the 
+    * fragment color should persevere.  The Select Bits (selb rt, rA, rB, rM)
+    * instruction will select bits from its first operand rA wherever the
+    * the mask bits rM are 0, and from its second operand rB wherever the
+    * mask bits rM are 1.  That means that the frame buffer color is the
+    * first operand, and the fragment color the second.
+    */
+    spe_selb(f, fragRGBA_reg, fbRGBA_reg, fragRGBA_reg, colormask_reg);
+
+    /* Release the temporary register and we're done */
+    spe_release_register(f, colormask_reg);
 }
 
+/* This function is annoyingly similar to gen_depth_test(), above, except
+ * that instead of comparing two varying values (i.e. fragment and buffer),
+ * we're comparing a varying value with a static value.  As such, we have
+ * access to the Compare Immediate instructions where we don't in 
+ * gen_depth_test(), which is what makes us very different.
+ *
+ * There's some added complexity if there's a non-trivial state->mask
+ * value; then stencil and reference both must be masked
+ *
+ * The return value in the stencil_pass_reg is a bitmask of valid
+ * fragments that also passed the stencil test.  The bitmask of valid
+ * fragments that failed would be found in (fragment_mask_reg & ~stencil_pass_reg).
+ */
+static void
+gen_stencil_test(struct spe_function *f, const struct pipe_stencil_state *state, 
+                 unsigned int stencil_max_value,
+                 unsigned int fragment_mask_reg, unsigned int fbS_reg, 
+                 unsigned int stencil_pass_reg)
+{
+   /* Generate code that puts the set of passing fragments into the stencil_pass_reg
+    * register, taking into account whether each fragment was active to begin with.
+    */
+   switch (state->func) {
+   case PIPE_FUNC_EQUAL:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & (s == reference) */
+         spe_compare_equal_uint(f, stencil_pass_reg, fbS_reg, state->ref_value);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ((s&mask) == (reference&mask)) */
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_compare_equal_uint(f, stencil_pass_reg, tmp_masked_stencil, state->value_mask & state->ref_value);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
 
+   case PIPE_FUNC_NOTEQUAL:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & ~(s == reference) */
+         spe_compare_equal_uint(f, stencil_pass_reg, fbS_reg, state->ref_value);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ~((s&mask) == (reference&mask)) */
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_compare_equal_uint(f, stencil_pass_reg, tmp_masked_stencil, state->value_mask & state->ref_value);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
+
+   case PIPE_FUNC_LESS:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & (reference < s)  */
+         spe_compare_greater_uint(f, stencil_pass_reg, fbS_reg, state->ref_value);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ((reference&mask) < (s & mask)) */
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_compare_greater_uint(f, stencil_pass_reg, tmp_masked_stencil, state->value_mask & state->ref_value);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
+
+   case PIPE_FUNC_GREATER:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & (reference > s) */
+         /* There's no convenient Compare Less Than Immediate instruction, so
+          * we'll have to do this one the harder way, by loading a register and 
+          * comparing directly.  Compare Logical Greater Than Word (clgt) 
+          * treats its operands as unsigned - no sign extension.
+          */
+         unsigned int tmp_reg = spe_allocate_available_register(f);
+         spe_load_uint(f, tmp_reg, state->ref_value);
+         spe_clgt(f, stencil_pass_reg, tmp_reg, fbS_reg);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ((reference&mask) > (s&mask)) */
+         unsigned int tmp_reg = spe_allocate_available_register(f);
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_load_uint(f, tmp_reg, state->value_mask & state->ref_value);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_clgt(f, stencil_pass_reg, tmp_reg, tmp_masked_stencil);
+         spe_and(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
+
+   case PIPE_FUNC_GEQUAL:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & (reference >= s) 
+          *              = fragment_mask & ~(s > reference) */
+         spe_compare_greater_uint(f, stencil_pass_reg, fbS_reg, state->ref_value);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ~((s&mask) > (reference&mask)) */
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_compare_greater_uint(f, stencil_pass_reg, tmp_masked_stencil, state->value_mask & state->ref_value);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
+
+   case PIPE_FUNC_LEQUAL:
+      if (state->value_mask == stencil_max_value) {
+         /* stencil_pass = fragment_mask & (reference <= s) ]
+          *               = fragment_mask & ~(reference > s) */
+         /* As above, we have to do this by loading a register */
+         unsigned int tmp_reg = spe_allocate_available_register(f);
+         spe_load_uint(f, tmp_reg, state->ref_value);
+         spe_clgt(f, stencil_pass_reg, tmp_reg, fbS_reg);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_reg);
+      }
+      else {
+         /* stencil_pass = fragment_mask & ~((reference&mask) > (s&mask)) */
+         unsigned int tmp_reg = spe_allocate_available_register(f);
+         unsigned int tmp_masked_stencil = spe_allocate_available_register(f);
+         spe_load_uint(f, tmp_reg, state->ref_value & state->value_mask);
+         spe_and_uint(f, tmp_masked_stencil, fbS_reg, state->value_mask);
+         spe_clgt(f, stencil_pass_reg, tmp_reg, tmp_masked_stencil);
+         spe_andc(f, stencil_pass_reg, fragment_mask_reg, stencil_pass_reg);
+         spe_release_register(f, tmp_reg);
+         spe_release_register(f, tmp_masked_stencil);
+      }
+      break;
+
+   case PIPE_FUNC_NEVER:
+      /* stencil_pass = fragment_mask & 0 = 0 */
+      spe_load_uint(f, stencil_pass_reg, 0);
+      break;
+
+   case PIPE_FUNC_ALWAYS:
+      /* stencil_pass = fragment_mask & 1 = fragment_mask */
+      spe_move(f, stencil_pass_reg, fragment_mask_reg);
+      break;
+   }
+
+   /* The fragments that passed the stencil test are now in stencil_pass_reg.
+    * The fragments that failed would be (fragment_mask_reg & ~stencil_pass_reg).
+    */
+}
+
+/* This function generates code that calculates a set of new stencil values
+ * given the earlier values and the operation to apply.  It does not
+ * apply any tests.  It is intended to be called up to 3 times
+ * (for the stencil fail operation, for the stencil pass-z fail operation,
+ * and for the stencil pass-z pass operation) to collect up to three
+ * possible sets of values, and for the caller to combine them based
+ * on the result of the tests.
+ *
+ * stencil_max_value should be (2^n - 1) where n is the number of bits
+ * in the stencil buffer - in other words, it should be usable as a mask.
+ */
+static void
+gen_stencil_values(struct spe_function *f, unsigned int stencil_op,
+                   unsigned int stencil_ref_value, unsigned int stencil_max_value,
+                   unsigned int fbS_reg, unsigned int newS_reg)
+{
+   /* The code below assumes that newS_reg and fbS_reg are not the same
+    * register; if they can be, the calculations below will have to use
+    * an additional temporary register.  For now, mark the assumption
+    * with an assertion that will fail if they are the same.
+    */
+   ASSERT(fbS_reg != newS_reg);
+
+   /* The code also assumes the the stencil_max_value is of the form 
+    * 2^n-1 and can therefore be used as a mask for the valid bits in 
+    * addition to a maximum.  Make sure this is the case as well.
+    * The clever math below exploits the fact that incrementing a 
+    * binary number serves to flip all the bits of a number starting at
+    * the LSB and continuing to (and including) the first zero bit
+    * found.  That means that a number and its increment will always
+    * have at least one bit in common (the high order bit, if nothing
+    * else) *unless* the number is zero, *or* the number is of a form
+    * consisting of some number of 1s in the low-order bits followed
+    * by nothing but 0s in the high-order bits.  The latter case
+    * implies it's of the form 2^n-1.
+    */
+   ASSERT(stencil_max_value > 0 && ((stencil_max_value + 1) & stencil_max_value) == 0);
+
+   switch(stencil_op) {
+   case PIPE_STENCIL_OP_KEEP:
+      /* newS = S */
+      spe_move(f, newS_reg, fbS_reg);
+      break;
+
+   case PIPE_STENCIL_OP_ZERO:
+      /* newS = 0 */
+      spe_zero(f, newS_reg);
+      break;
+
+   case PIPE_STENCIL_OP_REPLACE:
+      /* newS = stencil reference value */
+      spe_load_uint(f, newS_reg, stencil_ref_value);
+      break;
+
+   case PIPE_STENCIL_OP_INCR: {
+      /* newS = (s == max ? max : s + 1) */
+      unsigned int equals_reg = spe_allocate_available_register(f);
+
+      spe_compare_equal_uint(f, equals_reg, fbS_reg, stencil_max_value);
+      /* Add Word Immediate computes rT = rA + 10-bit signed immediate */
+      spe_ai(f, newS_reg, fbS_reg, 1);
+      /* Select from the current value or the new value based on the equality test */
+      spe_selb(f, newS_reg, newS_reg, fbS_reg, equals_reg);
+
+      spe_release_register(f, equals_reg);
+      break;
+   }
+   case PIPE_STENCIL_OP_DECR: {
+      /* newS = (s == 0 ? 0 : s - 1) */
+      unsigned int equals_reg = spe_allocate_available_register(f);
+
+      spe_compare_equal_uint(f, equals_reg, fbS_reg, 0);
+      /* Add Word Immediate with a (-1) value works */
+      spe_ai(f, newS_reg, fbS_reg, -1);
+      /* Select from the current value or the new value based on the equality test */
+      spe_selb(f, newS_reg, newS_reg, fbS_reg, equals_reg);
+
+      spe_release_register(f, equals_reg);
+      break;
+   }
+   case PIPE_STENCIL_OP_INCR_WRAP:
+      /* newS = (s == max ? 0 : s + 1), but since max is 2^n-1, we can
+       * do a normal add and mask off the correct bits 
+       */
+      spe_ai(f, newS_reg, fbS_reg, 1);
+      spe_and_uint(f, newS_reg, newS_reg, stencil_max_value);
+      break;
+
+   case PIPE_STENCIL_OP_DECR_WRAP:
+      /* newS = (s == 0 ? max : s - 1), but we'll pull the same mask trick as above */
+      spe_ai(f, newS_reg, fbS_reg, -1);
+      spe_and_uint(f, newS_reg, newS_reg, stencil_max_value);
+      break;
+
+   case PIPE_STENCIL_OP_INVERT:
+      /* newS = ~s.  We take advantage of the mask/max value to invert only
+       * the valid bits for the field so we don't have to do an extra "and".
+       */
+      spe_xor_uint(f, newS_reg, fbS_reg, stencil_max_value);
+      break;
+
+   default:
+      ASSERT(0);
+   }
+}
+
+
+/* This function generates code to get all the necessary possible
+ * stencil values.  For each of the output registers (fail_reg,
+ * zfail_reg, and zpass_reg), it either allocates a new register
+ * and calculates a new set of values based on the stencil operation,
+ * or it reuses a register allocation and calculation done for an
+ * earlier (matching) operation, or it reuses the fbS_reg register
+ * (if the stencil operation is KEEP, which doesn't change the 
+ * stencil buffer).
+ *
+ * Since this function allocates a variable number of registers,
+ * to avoid incurring complex logic to free them, they should
+ * be allocated after a spe_allocate_register_set() call
+ * and released by the corresponding spe_release_register_set() call.
+ */
+static void
+gen_get_stencil_values(struct spe_function *f, const struct pipe_stencil_state *stencil,
+                       const unsigned int depth_enabled,
+                       unsigned int fbS_reg, 
+                       unsigned int *fail_reg, unsigned int *zfail_reg, 
+                       unsigned int *zpass_reg)
+{
+   unsigned zfail_op;
+
+   /* Stenciling had better be enabled here */
+   ASSERT(stencil->enabled);
+
+   /* If the depth test is not enabled, it is treated as though it always
+    * passes, which means that the zfail_op is not considered - a
+    * failing stencil test triggers the fail_op, and a passing one
+    * triggers the zpass_op
+    *
+    * As an optimization, override calculation of the zfail_op values
+    * if they aren't going to be used.  By setting the value of
+    * the operation to PIPE_STENCIL_OP_KEEP, its value will be assumed
+    * to match the incoming stencil values, and no calculation will
+    * be done.
+    */
+   if (depth_enabled) {
+      zfail_op = stencil->zfail_op;
+   }
+   else {
+      zfail_op = PIPE_STENCIL_OP_KEEP;
+   }
+
+   /* One-sided or front-facing stencil */
+   if (stencil->fail_op == PIPE_STENCIL_OP_KEEP) {
+      *fail_reg = fbS_reg;
+   }
+   else {
+      *fail_reg = spe_allocate_available_register(f);
+      gen_stencil_values(f, stencil->fail_op, stencil->ref_value, 
+         0xff, fbS_reg, *fail_reg);
+   }
+
+   /* Check the possibly overridden value, not the structure value */
+   if (zfail_op == PIPE_STENCIL_OP_KEEP) {
+      *zfail_reg = fbS_reg;
+   }
+   else if (zfail_op == stencil->fail_op) {
+      *zfail_reg = *fail_reg;
+   }
+   else {
+      *zfail_reg = spe_allocate_available_register(f);
+      gen_stencil_values(f, stencil->zfail_op, stencil->ref_value, 
+         0xff, fbS_reg, *zfail_reg);
+   }
+
+   if (stencil->zpass_op == PIPE_STENCIL_OP_KEEP) {
+      *zpass_reg = fbS_reg;
+   }
+   else if (stencil->zpass_op == stencil->fail_op) {
+      *zpass_reg = *fail_reg;
+   }
+   else if (stencil->zpass_op == zfail_op) {
+      *zpass_reg = *zfail_reg;
+   }
+   else {
+      *zpass_reg = spe_allocate_available_register(f);
+      gen_stencil_values(f, stencil->zpass_op, stencil->ref_value, 
+         0xff, fbS_reg, *zpass_reg);
+   }
+}
+
+/* Note that fbZ_reg may *not* be set on entry, if in fact
+ * the depth test is not enabled.  This function must not use
+ * the register if depth is not enabled.
+ */
+static boolean
+gen_stencil_depth_test(struct spe_function *f, 
+                       const struct pipe_depth_stencil_alpha_state *dsa, 
+                       const uint facing,
+                       const int mask_reg, const int fragZ_reg, 
+                       const int fbZ_reg, const int fbS_reg)
+{
+   /* True if we've generated code that could require writeback to the
+    * depth and/or stencil buffers
+    */
+   boolean modified_buffers = false;
+
+   boolean need_to_calculate_stencil_values;
+   boolean need_to_writemask_stencil_values;
+
+   struct pipe_stencil_state *stencil;
+
+   /* Registers.  We may or may not actually allocate these, depending
+    * on whether the state values indicate that we need them.
+    */
+   unsigned int stencil_pass_reg, stencil_fail_reg;
+   unsigned int stencil_fail_values, stencil_pass_depth_fail_values, stencil_pass_depth_pass_values;
+   unsigned int stencil_writemask_reg;
+   unsigned int zmask_reg;
+   unsigned int newS_reg;
+
+   /* Stenciling is quite complex: up to six different configurable stencil 
+    * operations/calculations can be required (three each for front-facing
+    * and back-facing fragments).  Many of those operations will likely 
+    * be identical, so there's good reason to try to avoid calculating 
+    * the same values more than once (which unfortunately makes the code less 
+    * straightforward).
+    *
+    * To make register management easier, we start a new 
+    * register set; we can release all the registers in the set at
+    * once, and avoid having to keep track of exactly which registers
+    * we allocate.  We can still allocate and free registers as 
+    * desired (if we know we no longer need a register), but we don't
+    * have to spend the complexity to track the more difficult variant
+    * register usage scenarios.
+    */
+   spe_comment(f, 0, "Allocating stencil register set");
+   spe_allocate_register_set(f);
+
+   /* The facing we're given is the fragment facing; it doesn't
+    * exactly match the stencil facing.  If stencil is enabled,
+    * but two-sided stencil is *not* enabled, we use the same
+    * stencil settings for both front- and back-facing fragments.
+    * We only use the "back-facing" stencil for backfacing fragments
+    * if two-sided stenciling is enabled.
+    */
+   if (facing == CELL_FACING_BACK && dsa->stencil[1].enabled) {
+      stencil = &dsa->stencil[1];
+   }
+   else {
+      stencil = &dsa->stencil[0];
+   }
+
+   /* Calculate the writemask.  If the writemask is trivial (either
+    * all 0s, meaning that we don't need to calculate any stencil values
+    * because they're not going to change the stencil anyway, or all 1s,
+    * meaning that we have to calculate the stencil values but do not
+    * need to mask them), we can avoid generating code.  Don't forget
+    * that we need to consider backfacing stencil, if enabled.
+    *
+    * Note that if the backface stencil is *not* enabled, the backface
+    * stencil will have the same values as the frontface stencil.
+    */
+   if (stencil->fail_op == PIPE_STENCIL_OP_KEEP &&
+       stencil->zfail_op == PIPE_STENCIL_OP_KEEP &&
+       stencil->zpass_op == PIPE_STENCIL_OP_KEEP) {
+       need_to_calculate_stencil_values = false;
+       need_to_writemask_stencil_values = false;
+    }
+    else if (stencil->write_mask == 0x0) {
+      /* All changes are writemasked out, so no need to calculate
+       * what those changes might be, and no need to write anything back.
+       */
+      need_to_calculate_stencil_values = false;
+      need_to_writemask_stencil_values = false;
+   }
+   else if (stencil->write_mask == 0xff) {
+      /* Still trivial, but a little less so.  We need to write the stencil
+       * values, but we don't need to mask them.
+       */
+      need_to_calculate_stencil_values = true;
+      need_to_writemask_stencil_values = false;
+   }
+   else {
+      /* The general case: calculate, mask, and write */
+      need_to_calculate_stencil_values = true;
+      need_to_writemask_stencil_values = true;
+
+      /* While we're here, generate code that calculates what the
+       * writemask should be.  If backface stenciling is enabled,
+       * and the backface writemask is not the same as the frontface
+       * writemask, we'll have to generate code that merges the
+       * two masks into a single effective mask based on fragment facing.
+       */
+      spe_comment(f, 0, "Computing stencil writemask");
+      stencil_writemask_reg = spe_allocate_available_register(f);
+      spe_load_uint(f, stencil_writemask_reg, dsa->stencil[facing].write_mask);
+   }
+
+   /* At least one-sided stenciling must be on.  Generate code that
+    * runs the stencil test on the basic/front-facing stencil, leaving
+    * the mask of passing stencil bits in stencil_pass_reg.  This mask will
+    * be used both to mask the set of active pixels, and also to
+    * determine how the stencil buffer changes.
+    *
+    * This test will *not* change the value in mask_reg (because we don't
+    * yet know whether to apply the two-sided stencil or one-sided stencil).
+    */
+   spe_comment(f, 0, "Running basic stencil test");
+   stencil_pass_reg = spe_allocate_available_register(f);
+   gen_stencil_test(f, stencil, 0xff, mask_reg, fbS_reg, stencil_pass_reg);
+
+   /* Generate code that, given the mask of valid fragments and the
+    * mask of valid fragments that passed the stencil test, computes
+    * the mask of valid fragments that failed the stencil test.  We
+    * have to do this before we run a depth test (because the
+    * depth test should not be performed on fragments that failed the
+    * stencil test, and because the depth test will update the 
+    * mask of valid fragments based on the results of the depth test).
+    */
+   spe_comment(f, 0, "Computing stencil fail mask and updating fragment mask");
+   stencil_fail_reg = spe_allocate_available_register(f);
+   spe_andc(f, stencil_fail_reg, mask_reg, stencil_pass_reg);
+   /* Now remove the stenciled-out pixels from the valid fragment mask,
+    * so we can later use the valid fragment mask in the depth test.
+    */
+   spe_and(f, mask_reg, mask_reg, stencil_pass_reg);
+
+   /* We may not need to calculate stencil values, if the writemask is off */
+   if (need_to_calculate_stencil_values) {
+      /* Generate code that calculates exactly which stencil values we need,
+       * without calculating the same value twice (say, if two different
+       * stencil ops have the same value).  This code will work for one-sided
+       * and two-sided stenciling (so that we take into account that operations
+       * may match between front and back stencils), and will also take into
+       * account whether the depth test is enabled (if the depth test is off,
+       * we don't need any of the zfail results, because the depth test always
+       * is considered to pass if it is disabled).  Any register value that
+       * does not need to be calculated will come back with the same value
+       * that's in fbS_reg.
+       *
+       * This function will allocate a variant number of registers that
+       * will be released as part of the register set.
+       */
+      spe_comment(f, 0, facing == CELL_FACING_FRONT ? "Computing front-facing stencil values" : "Computing back-facing stencil values");
+      gen_get_stencil_values(f, stencil, dsa->depth.enabled, fbS_reg, 
+         &stencil_fail_values, &stencil_pass_depth_fail_values, 
+         &stencil_pass_depth_pass_values);
+   }  
+
+   /* We now have all the stencil values we need.  We also need 
+    * the results of the depth test to figure out which
+    * stencil values will become the new stencil values.  (Even if
+    * we aren't actually calculating stencil values, we need to apply
+    * the depth test if it's enabled.)
+    *
+    * The code generated by gen_depth_test() returns the results of the
+    * test in the given register, but also alters the mask_reg based
+    * on the results of the test.
+    */
+   if (dsa->depth.enabled) {
+      spe_comment(f, 0, "Running stencil depth test");
+      zmask_reg = spe_allocate_available_register(f);
+      modified_buffers |= gen_depth_test(f, dsa, mask_reg, fragZ_reg, fbZ_reg, zmask_reg);
+   }
+
+   if (need_to_calculate_stencil_values) {
+
+      /* If we need to writemask the stencil values before going into
+       * the stencil buffer, we'll have to use a new register to
+       * hold the new values.  If not, we can just keep using the
+       * current register.
+       */
+      if (need_to_writemask_stencil_values) {
+         newS_reg = spe_allocate_available_register(f);
+         spe_comment(f, 0, "Saving current stencil values for writemasking");
+         spe_move(f, newS_reg, fbS_reg);
+      }
+      else {
+         newS_reg = fbS_reg;
+      }
+
+      /* Merge in the selected stencil fail values */
+      if (stencil_fail_values != fbS_reg) {
+         spe_comment(f, 0, "Loading stencil fail values");
+         spe_selb(f, newS_reg, newS_reg, stencil_fail_values, stencil_fail_reg);
+         modified_buffers = true;
+      }
+
+      /* Same for the stencil pass/depth fail values.  If this calculation
+       * is not needed (say, if depth test is off), then the
+       * stencil_pass_depth_fail_values register will be equal to fbS_reg
+       * and we'll skip the calculation.
+       */
+      if (stencil_pass_depth_fail_values != fbS_reg) {
+         /* We don't actually have a stencil pass/depth fail mask yet.
+          * Calculate it here from the stencil passing mask and the
+          * depth passing mask.  Note that zmask_reg *must* have been
+          * set above if we're here.
+          */
+         unsigned int stencil_pass_depth_fail_mask = spe_allocate_available_register(f);
+         spe_comment(f, 0, "Loading stencil pass/depth fail values");
+         spe_andc(f, stencil_pass_depth_fail_mask, stencil_pass_reg, zmask_reg);
+
+         spe_selb(f, newS_reg, newS_reg, stencil_pass_depth_fail_values, stencil_pass_depth_fail_mask);
+
+         spe_release_register(f, stencil_pass_depth_fail_mask);
+         modified_buffers = true;
+      }
+
+      /* Same for the stencil pass/depth pass mask.  Note that we
+       * *can* get here with zmask_reg being unset (if the depth
+       * test is off but the stencil test is on).  In this case,
+       * we assume the depth test passes, and don't need to mask
+       * the stencil pass mask with the Z mask.
+       */
+      if (stencil_pass_depth_pass_values != fbS_reg) {
+         if (dsa->depth.enabled) {
+            unsigned int stencil_pass_depth_pass_mask = spe_allocate_available_register(f);
+            /* We'll need a separate register */
+            spe_comment(f, 0, "Loading stencil pass/depth pass values");
+            spe_and(f, stencil_pass_depth_pass_mask, stencil_pass_reg, zmask_reg);
+            spe_selb(f, newS_reg, newS_reg, stencil_pass_depth_pass_values, stencil_pass_depth_pass_mask);
+            spe_release_register(f, stencil_pass_depth_pass_mask);
+         }
+         else {
+            /* We can use the same stencil-pass register */
+            spe_comment(f, 0, "Loading stencil pass values");
+            spe_selb(f, newS_reg, newS_reg, stencil_pass_depth_pass_values, stencil_pass_reg);
+         }
+         modified_buffers = true;
+      }
+
+      /* Almost done.  If we need to writemask, do it now, leaving the
+       * results in the fbS_reg register passed in.  If we don't need
+       * to writemask, then the results are *already* in the fbS_reg,
+       * so there's nothing more to do.
+       */
+
+      if (need_to_writemask_stencil_values && modified_buffers) {
+         /* The Select Bytes command makes a fine writemask.  Where
+          * the mask is 0, the first (original) values are retained,
+          * effectively masking out changes.  Where the mask is 1, the
+          * second (new) values are retained, incorporating changes.
+          */
+         spe_comment(f, 0, "Writemasking new stencil values");
+         spe_selb(f, fbS_reg, fbS_reg, newS_reg, stencil_writemask_reg);
+      }
+
+   } /* done calculating stencil values */
+
+   /* The stencil and/or depth values have been applied, and the
+    * mask_reg, fbS_reg, and fbZ_reg values have been updated.
+    * We're all done, except that we've allocated a fair number
+    * of registers that we didn't bother tracking.  Release all
+    * those registers as part of the register set, and go home.
+    */
+   spe_comment(f, 0, "Releasing stencil register set");
+   spe_release_register_set(f);
+
+   /* Return true if we could have modified the stencil and/or
+    * depth buffers.
+    */
+   return modified_buffers;
+}
 
 
 /**
@@ -621,14 +1774,19 @@ gen_pack_colors(struct spe_function *f,
  * should be much faster.
  *
  * \param cell  the rendering context (in)
- * \param f     the generated function (out)
+ * \param facing whether the generated code is for front-facing or 
+ *              back-facing fragments
+ * \param f     the generated function (in/out); on input, the function
+ *              must already have been initialized.  On exit, whatever
+ *              instructions within the generated function have had
+ *              the fragment ops appended.
  */
 void
-cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
+cell_gen_fragment_function(struct cell_context *cell, const uint facing, struct spe_function *f)
 {
-   const struct pipe_depth_stencil_alpha_state *dsa =
-      &cell->depth_stencil->base;
-   const struct pipe_blend_state *blend = &cell->blend->base;
+   const struct pipe_depth_stencil_alpha_state *dsa = cell->depth_stencil;
+   const struct pipe_blend_state *blend = cell->blend;
+   const struct pipe_blend_color *blend_color = &cell->blend_color;
    const enum pipe_format color_format = cell->framebuffer.cbufs[0]->format;
 
    /* For SPE function calls: reg $3 = first param, $4 = second param, etc. */
@@ -643,6 +1801,8 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
    const int fragA_reg = 11;  /* vector float */
    const int mask_reg = 12;   /* vector uint */
 
+   ASSERT(facing == CELL_FACING_FRONT || facing == CELL_FACING_BACK);
+
    /* offset of quad from start of tile
     * XXX assuming 4-byte pixels for color AND Z/stencil!!!!
     */
@@ -651,7 +1811,12 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
    int fbRGBA_reg;  /**< framebuffer's RGBA colors for quad */
    int fbZS_reg;    /**< framebuffer's combined z/stencil values for quad */
 
-   spe_init_func(f, SPU_MAX_FRAGMENT_OPS_INSTS * SPE_INST_SIZE);
+   if (cell->debug_flags & CELL_DEBUG_ASM) {
+      spe_print_code(f, true);
+      spe_indent(f, 8);
+      spe_comment(f, -4, facing == CELL_FACING_FRONT ? "Begin front-facing per-fragment ops": "Begin back-facing per-fragment ops");
+   }
+
    spe_allocate_register(f, x_reg);
    spe_allocate_register(f, y_reg);
    spe_allocate_register(f, color_tile_reg);
@@ -674,8 +1839,9 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
 
       ASSERT(TILE_SIZE == 32);
 
-      spe_rotmi(f, x2_reg, x_reg, -1);  /* x2 = x / 2 */
+      spe_comment(f, 0, "Compute quad offset within tile");
       spe_rotmi(f, y2_reg, y_reg, -1);  /* y2 = y / 2 */
+      spe_rotmi(f, x2_reg, x_reg, -1);  /* x2 = x / 2 */
       spe_shli(f, y2_reg, y2_reg, 4);   /* y2 *= 16 */
       spe_a(f, quad_offset_reg, y2_reg, x2_reg);  /* offset = y2 + x2 */
       spe_shli(f, quad_offset_reg, quad_offset_reg, 4);   /* offset *= 16 */
@@ -684,139 +1850,196 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
       spe_release_register(f, y2_reg);
    }
 
-
+   /* Generate the alpha test, if needed. */
    if (dsa->alpha.enabled) {
       gen_alpha_test(dsa, f, mask_reg, fragA_reg);
    }
 
+   /* If we need the stencil buffers (because one- or two-sided stencil is
+    * enabled) or the depth buffer (because the depth test is enabled),
+    * go grab them.  Note that if either one- or two-sided stencil is
+    * enabled, dsa->stencil[0].enabled will be true.
+    */
    if (dsa->depth.enabled || dsa->stencil[0].enabled) {
       const enum pipe_format zs_format = cell->framebuffer.zsbuf->format;
       boolean write_depth_stencil;
 
-      int fbZ_reg = spe_allocate_available_register(f); /* Z values */
-      int fbS_reg = spe_allocate_available_register(f); /* Stencil values */
+      /* We may or may not need to allocate a register for Z or stencil values */
+      boolean fbS_reg_set = false, fbZ_reg_set = false;
+      unsigned int fbS_reg, fbZ_reg = 0;
+
+      spe_comment(f, 0, "Fetching Z/stencil quad from tile");
 
       /* fetch quad of depth/stencil values from tile at (x,y) */
       /* Load: fbZS_reg = memory[depth_tile_reg + offset_reg] */
+      /* XXX Not sure this is allowed if we've only got a 16-bit Z buffer... */
       spe_lqx(f, fbZS_reg, depth_tile_reg, quad_offset_reg);
 
-      if (dsa->depth.enabled) {
-         /* Extract Z bits from fbZS_reg into fbZ_reg */
-         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
-             zs_format == PIPE_FORMAT_X8Z24_UNORM) {
-            int mask_reg = spe_allocate_available_register(f);
-            spe_fsmbi(f, mask_reg, 0x7777);  /* mask[0,1,2,3] = 0x00ffffff */
-            spe_and(f, fbZ_reg, fbZS_reg, mask_reg);  /* fbZ = fbZS & mask */
-            spe_release_register(f, mask_reg);
-            /* OK, fbZ_reg has four 24-bit Z values now */
-         }
-         else {
-            /* XXX handle other z/stencil formats */
-            ASSERT(0);
-         }
+      /* From the Z/stencil buffer format, pull out the bits we need for
+       * Z and/or stencil.  We'll also convert the incoming fragment Z
+       * value in fragZ_reg from a floating point value in [0.0..1.0] to
+       * an unsigned integer value with the appropriate resolution.
+       * Note that even if depth or stencil is *not* enabled, if it's
+       * present in the buffer, we pull it out and put it back later;
+       * otherwise, we can inadvertently destroy the contents of
+       * buffers we're not supposed to touch (e.g., if the user is
+       * clearing the depth buffer but not the stencil buffer, a
+       * quad of constant depth is drawn over the surface; the stencil
+       * buffer must be maintained).
+       */
+      switch(zs_format) {
 
-         /* Convert fragZ values from float[4] to uint[4] */
-         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
-             zs_format == PIPE_FORMAT_X8Z24_UNORM ||
-             zs_format == PIPE_FORMAT_Z24S8_UNORM ||
-             zs_format == PIPE_FORMAT_Z24X8_UNORM) {
-            /* 24-bit Z values */
-            int scale_reg = spe_allocate_available_register(f);
+         case PIPE_FORMAT_S8Z24_UNORM: /* fall through */
+         case PIPE_FORMAT_X8Z24_UNORM:
+            /* Pull out both Z and stencil */
+            setup_optional_register(f, &fbZ_reg_set, &fbZ_reg);
+            setup_optional_register(f, &fbS_reg_set, &fbS_reg);
 
-            /* scale_reg[0,1,2,3] = float(2^24-1) */
-            spe_load_float(f, scale_reg, (float) 0xffffff);
+            /* four 24-bit Z values in the low-order bits */
+            spe_and_uint(f, fbZ_reg, fbZS_reg, 0x00ffffff);
 
-            /* XXX these two instructions might be combined */
-            spe_fm(f, fragZ_reg, fragZ_reg, scale_reg); /* fragZ *= scale */
-            spe_cfltu(f, fragZ_reg, fragZ_reg, 0);  /* fragZ = (int) fragZ */
+            /* Incoming fragZ_reg value is a float in 0.0...1.0; convert
+             * to a 24-bit unsigned integer
+             */
+            spe_cfltu(f, fragZ_reg, fragZ_reg, 32);
+            spe_rotmi(f, fragZ_reg, fragZ_reg, -8);
 
-            spe_release_register(f, scale_reg);
-         }
-         else {
-            /* XXX handle 16-bit Z format */
-            ASSERT(0);
-         }
-      }
+            /* four 8-bit stencil values in the high-order bits */
+            spe_rotmi(f, fbS_reg, fbZS_reg, -24);
+         break;
 
-      if (dsa->stencil[0].enabled) {
-         /* Extract Stencil bit sfrom fbZS_reg into fbS_reg */
-         if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
-             zs_format == PIPE_FORMAT_X8Z24_UNORM) {
-            /* XXX extract with a shift */
-            ASSERT(0);
-         }
-         else if (zs_format == PIPE_FORMAT_Z24S8_UNORM ||
-                  zs_format == PIPE_FORMAT_Z24X8_UNORM) {
-            /* XXX extract with a mask */
-            ASSERT(0);
-         }
-      }
+         case PIPE_FORMAT_Z24S8_UNORM: /* fall through */
+         case PIPE_FORMAT_Z24X8_UNORM:
+            setup_optional_register(f, &fbZ_reg_set, &fbZ_reg);
+            setup_optional_register(f, &fbS_reg_set, &fbS_reg);
+
+            /* shift by 8 to get the upper 24-bit values */
+            spe_rotmi(f, fbS_reg, fbZS_reg, -8);
+
+            /* Incoming fragZ_reg value is a float in 0.0...1.0; convert
+             * to a 24-bit unsigned integer
+             */
+            spe_cfltu(f, fragZ_reg, fragZ_reg, 32);
+            spe_rotmi(f, fragZ_reg, fragZ_reg, -8);
+
+            /* 8-bit stencil in the low-order bits - mask them out */
+            spe_and_uint(f, fbS_reg, fbZS_reg, 0x000000ff);
+         break;
+
+         case PIPE_FORMAT_Z32_UNORM:
+            setup_optional_register(f, &fbZ_reg_set, &fbZ_reg);
+            /* Copy over 4 32-bit values */
+            spe_move(f, fbZ_reg, fbZS_reg);
+
+            /* Incoming fragZ_reg value is a float in 0.0...1.0; convert
+             * to a 32-bit unsigned integer
+             */
+            spe_cfltu(f, fragZ_reg, fragZ_reg, 32);
+            /* No stencil, so can't do anything there */
+         break;
 
+         case PIPE_FORMAT_Z16_UNORM:
+            /* XXX Not sure this is correct, but it was here before, so we're
+             * going with it for now
+             */
+            setup_optional_register(f, &fbZ_reg_set, &fbZ_reg);
+            /* Copy over 4 32-bit values */
+            spe_move(f, fbZ_reg, fbZS_reg);
+
+            /* Incoming fragZ_reg value is a float in 0.0...1.0; convert
+             * to a 16-bit unsigned integer
+             */
+            spe_cfltu(f, fragZ_reg, fragZ_reg, 32);
+            spe_rotmi(f, fragZ_reg, fragZ_reg, -16);
+            /* No stencil */
+
+         default:
+            ASSERT(0); /* invalid format */
+      }
 
+      /* If stencil is enabled, use the stencil-specific code
+       * generator to generate both the stencil and depth (if needed)
+       * tests.  Otherwise, if only depth is enabled, generate
+       * a quick depth test.  The test generators themselves will
+       * report back whether the depth/stencil buffer has to be
+       * written back.
+       */
       if (dsa->stencil[0].enabled) {
-         /* XXX this may involve depth testing too */
-         // gen_stencil_test(dsa, f, ... );
-         ASSERT(0);
+         /* This will perform the stencil and depth tests, and update
+          * the mask_reg, fbZ_reg, and fbS_reg as required by the
+          * tests.
+          */
+         ASSERT(fbS_reg_set);
+         spe_comment(f, 0, "Perform stencil test");
+
+         /* Note that fbZ_reg may not be set on entry, if stenciling
+          * is enabled but there's no Z-buffer.  The 
+          * gen_stencil_depth_test() function must ignore the
+          * fbZ_reg register if depth is not enabled.
+          */
+         write_depth_stencil = gen_stencil_depth_test(f, dsa, facing, mask_reg, fragZ_reg, fbZ_reg, fbS_reg);
       }
       else if (dsa->depth.enabled) {
          int zmask_reg = spe_allocate_available_register(f);
-         gen_depth_test(dsa, f, mask_reg, fragZ_reg, fbZ_reg, zmask_reg);
+         ASSERT(fbZ_reg_set);
+         spe_comment(f, 0, "Perform depth test");
+         write_depth_stencil = gen_depth_test(f, dsa, mask_reg, fragZ_reg, fbZ_reg, zmask_reg);
          spe_release_register(f, zmask_reg);
       }
-
-      /* do we need to write Z and/or Stencil back into framebuffer? */
-      write_depth_stencil = (dsa->depth.writemask |
-                             dsa->stencil[0].write_mask |
-                             dsa->stencil[1].write_mask);
+      else {
+         write_depth_stencil = false;
+      }
 
       if (write_depth_stencil) {
          /* Merge latest Z and Stencil values into fbZS_reg.
           * fbZ_reg has four Z vals in bits [23..0] or bits [15..0].
           * fbS_reg has four 8-bit Z values in bits [7..0].
           */
+         spe_comment(f, 0, "Store quad's depth/stencil values in tile");
          if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
              zs_format == PIPE_FORMAT_X8Z24_UNORM) {
             spe_shli(f, fbS_reg, fbS_reg, 24); /* fbS = fbS << 24 */
             spe_or(f, fbZS_reg, fbS_reg, fbZ_reg); /* fbZS = fbS | fbZ */
          }
-         else if (zs_format == PIPE_FORMAT_S8Z24_UNORM ||
-                  zs_format == PIPE_FORMAT_X8Z24_UNORM) {
-            /* XXX to do */
-            ASSERT(0);
+         else if (zs_format == PIPE_FORMAT_Z24S8_UNORM ||
+                  zs_format == PIPE_FORMAT_Z24X8_UNORM) {
+            spe_shli(f, fbZ_reg, fbZ_reg, 8); /* fbZ = fbZ << 8 */
+            spe_or(f, fbZS_reg, fbS_reg, fbZ_reg); /* fbZS = fbS | fbZ */
+         }
+         else if (zs_format == PIPE_FORMAT_Z32_UNORM) {
+            spe_move(f, fbZS_reg, fbZ_reg); /* fbZS = fbZ */
          }
          else if (zs_format == PIPE_FORMAT_Z16_UNORM) {
-            /* XXX to do */
-            ASSERT(0);
+            spe_move(f, fbZS_reg, fbZ_reg); /* fbZS = fbZ */
          }
          else if (zs_format == PIPE_FORMAT_S8_UNORM) {
-            /* XXX to do */
-            ASSERT(0);
+            ASSERT(0);   /* XXX to do */
          }
          else {
-            /* bad zs_format */
-            ASSERT(0);
+            ASSERT(0); /* bad zs_format */
          }
 
          /* Store: memory[depth_tile_reg + quad_offset_reg] = fbZS */
          spe_stqx(f, fbZS_reg, depth_tile_reg, quad_offset_reg);
       }
 
-      spe_release_register(f, fbZ_reg);
-      spe_release_register(f, fbS_reg);
+      /* Don't need these any more */
+      release_optional_register(f, &fbZ_reg_set, fbZ_reg);
+      release_optional_register(f, &fbS_reg_set, fbS_reg);
    }
 
-
    /* Get framebuffer quad/colors.  We'll need these for blending,
     * color masking, and to obey the quad/pixel mask.
     * Load: fbRGBA_reg = memory[color_tile + quad_offset]
     * Note: if mask={~0,~0,~0,~0} and we're not blending or colormasking
     * we could skip this load.
     */
+   spe_comment(f, 0, "Fetch quad colors from tile");
    spe_lqx(f, fbRGBA_reg, color_tile_reg, quad_offset_reg);
 
-
    if (blend->blend_enable) {
-      gen_blend(blend, f, color_format,
+      spe_comment(f, 0, "Perform blending");
+      gen_blend(blend, blend_color, f, color_format,
                 fragR_reg, fragG_reg, fragB_reg, fragA_reg, fbRGBA_reg);
    }
 
@@ -829,19 +2052,21 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
       int rgba_reg = spe_allocate_available_register(f);
 
       /* Pack four float colors as four 32-bit int colors */
+      spe_comment(f, 0, "Convert float quad colors to packed int framebuffer colors");
       gen_pack_colors(f, color_format,
                       fragR_reg, fragG_reg, fragB_reg, fragA_reg,
                       rgba_reg);
 
       if (blend->logicop_enable) {
+         spe_comment(f, 0, "Compute logic op");
          gen_logicop(blend, f, rgba_reg, fbRGBA_reg);
       }
 
-      if (blend->colormask != 0xf) {
-         gen_colormask(blend->colormask, f, rgba_reg, fbRGBA_reg);
+      if (blend->colormask != PIPE_MASK_RGBA) {
+         spe_comment(f, 0, "Compute color mask");
+         gen_colormask(f, blend->colormask, color_format, rgba_reg, fbRGBA_reg);
       }
 
-
       /* Mix fragment colors with framebuffer colors using the quad/pixel mask:
        * if (mask[i])
        *    rgba[i] = rgba[i];
@@ -853,6 +2078,7 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
       /* Store updated quad in tile:
        * memory[color_tile + quad_offset] = rgba_reg;
        */
+      spe_comment(f, 0, "Store quad colors into color tile");
       spe_stqx(f, rgba_reg, color_tile_reg, quad_offset_reg);
 
       spe_release_register(f, rgba_reg);
@@ -862,9 +2088,14 @@ cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f)
 
    spe_bi(f, SPE_REG_RA, 0, 0);  /* return from function call */
 
-
    spe_release_register(f, fbRGBA_reg);
    spe_release_register(f, fbZS_reg);
    spe_release_register(f, quad_offset_reg);
-}
 
+   if (cell->debug_flags & CELL_DEBUG_ASM) {
+      char buffer[1024];
+      sprintf(buffer, "End %s-facing per-fragment ops: %d instructions", 
+         facing == CELL_FACING_FRONT ? "front" : "back", f->num_inst);
+      spe_comment(f, -4, buffer);
+   }
+}
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.h b/src/gallium/drivers/cell/ppu/cell_gen_fragment.h
index b59de198dce..21b35d1fafe 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fragment.h
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.h
@@ -31,7 +31,7 @@
 
 
 extern void
-cell_gen_fragment_function(struct cell_context *cell, struct spe_function *f);
+cell_gen_fragment_function(struct cell_context *cell, const uint facing, struct spe_function *f);
 
 
 #endif /* CELL_GEN_FRAGMENT_H */
diff --git a/src/gallium/drivers/cell/ppu/cell_pipe_state.c b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
index 475c6ef0ce6..81efd137c73 100644
--- a/src/gallium/drivers/cell/ppu/cell_pipe_state.c
+++ b/src/gallium/drivers/cell/ppu/cell_pipe_state.c
@@ -35,9 +35,9 @@
 #include "draw/draw_context.h"
 #include "cell_context.h"
 #include "cell_flush.h"
+#include "cell_pipe_state.h"
 #include "cell_state.h"
 #include "cell_texture.h"
-#include "cell_state_per_fragment.h"
 
 
 
@@ -45,24 +45,18 @@ static void *
 cell_create_blend_state(struct pipe_context *pipe,
                         const struct pipe_blend_state *blend)
 {
-   struct cell_blend_state *cb = MALLOC(sizeof(struct cell_blend_state));
-
-   (void) memcpy(cb, blend, sizeof(*blend));
-#if 0
-   cell_generate_alpha_blend(cb);
-#endif
-   return cb;
+   return mem_dup(blend, sizeof(*blend));
 }
 
 
 static void
-cell_bind_blend_state(struct pipe_context *pipe, void *state)
+cell_bind_blend_state(struct pipe_context *pipe, void *blend)
 {
    struct cell_context *cell = cell_context(pipe);
 
    draw_flush(cell->draw);
 
-   cell->blend = (struct cell_blend_state *) state;
+   cell->blend = (struct pipe_blend_state *) blend;
    cell->dirty |= CELL_NEW_BLEND;
 }
 
@@ -70,10 +64,7 @@ cell_bind_blend_state(struct pipe_context *pipe, void *state)
 static void
 cell_delete_blend_state(struct pipe_context *pipe, void *blend)
 {
-   struct cell_blend_state *cb = (struct cell_blend_state *) blend;
-
-   spe_release_func(& cb->code);
-   FREE(cb);
+   FREE(blend);
 }
 
 
@@ -95,41 +86,29 @@ cell_set_blend_color(struct pipe_context *pipe,
 
 static void *
 cell_create_depth_stencil_alpha_state(struct pipe_context *pipe,
-                 const struct pipe_depth_stencil_alpha_state *depth_stencil)
+                 const struct pipe_depth_stencil_alpha_state *dsa)
 {
-   struct cell_depth_stencil_alpha_state *cdsa =
-       MALLOC(sizeof(struct cell_depth_stencil_alpha_state));
-
-   (void) memcpy(cdsa, depth_stencil, sizeof(*depth_stencil));
-#if 0
-   cell_generate_depth_stencil_test(cdsa);
-#endif
-   return cdsa;
+   return mem_dup(dsa, sizeof(*dsa));
 }
 
 
 static void
 cell_bind_depth_stencil_alpha_state(struct pipe_context *pipe,
-                                    void *depth_stencil)
+                                    void *dsa)
 {
    struct cell_context *cell = cell_context(pipe);
 
    draw_flush(cell->draw);
 
-   cell->depth_stencil =
-       (struct cell_depth_stencil_alpha_state *) depth_stencil;
+   cell->depth_stencil = (struct pipe_depth_stencil_alpha_state *) dsa;
    cell->dirty |= CELL_NEW_DEPTH_STENCIL;
 }
 
 
 static void
-cell_delete_depth_stencil_alpha_state(struct pipe_context *pipe, void *depth)
+cell_delete_depth_stencil_alpha_state(struct pipe_context *pipe, void *dsa)
 {
-   struct cell_depth_stencil_alpha_state *cdsa =
-       (struct cell_depth_stencil_alpha_state *) depth;
-
-   spe_release_func(& cdsa->code);
-   FREE(cdsa);
+   FREE(dsa);
 }
 
 
@@ -191,24 +170,23 @@ cell_set_polygon_stipple( struct pipe_context *pipe,
 
 static void *
 cell_create_rasterizer_state(struct pipe_context *pipe,
-                             const struct pipe_rasterizer_state *setup)
+                             const struct pipe_rasterizer_state *rasterizer)
 {
-   struct pipe_rasterizer_state *state
-      = MALLOC(sizeof(struct pipe_rasterizer_state));
-   memcpy(state, setup, sizeof(struct pipe_rasterizer_state));
-   return state;
+   return mem_dup(rasterizer, sizeof(*rasterizer));
 }
 
 
 static void
-cell_bind_rasterizer_state(struct pipe_context *pipe, void *setup)
+cell_bind_rasterizer_state(struct pipe_context *pipe, void *rast)
 {
+   struct pipe_rasterizer_state *rasterizer =
+      (struct pipe_rasterizer_state *) rast;
    struct cell_context *cell = cell_context(pipe);
 
    /* pass-through to draw module */
-   draw_set_rasterizer_state(cell->draw, setup);
+   draw_set_rasterizer_state(cell->draw, rasterizer);
 
-   cell->rasterizer = (struct pipe_rasterizer_state *)setup;
+   cell->rasterizer = rasterizer;
 
    cell->dirty |= CELL_NEW_RASTERIZER;
 }
@@ -235,17 +213,24 @@ cell_bind_sampler_states(struct pipe_context *pipe,
                          unsigned num, void **samplers)
 {
    struct cell_context *cell = cell_context(pipe);
+   uint i, changed = 0x0;
 
    assert(num <= CELL_MAX_SAMPLERS);
 
    draw_flush(cell->draw);
 
-   memcpy(cell->sampler, samplers, num * sizeof(void *));
-   memset(&cell->sampler[num], 0, (CELL_MAX_SAMPLERS - num) *
-          sizeof(void *));
-   cell->num_samplers = num;
+   for (i = 0; i < CELL_MAX_SAMPLERS; i++) {
+      struct pipe_sampler_state *new_samp = i < num ? samplers[i] : NULL;
+      if (cell->sampler[i] != new_samp) {
+         cell->sampler[i] = new_samp;
+         changed |= (1 << i);
+      }
+   }
 
-   cell->dirty |= CELL_NEW_SAMPLER;
+   if (changed) {
+      cell->dirty |= CELL_NEW_SAMPLER;
+      cell->dirty_samplers |= changed;
+   }
 }
 
 
@@ -263,27 +248,25 @@ cell_set_sampler_textures(struct pipe_context *pipe,
                           unsigned num, struct pipe_texture **texture)
 {
    struct cell_context *cell = cell_context(pipe);
-   uint i;
+   uint i, changed = 0x0;
 
    assert(num <= CELL_MAX_SAMPLERS);
 
-   /* Check for no-op */
-   if (num == cell->num_textures &&
-       !memcmp(cell->texture, texture, num * sizeof(struct pipe_texture *)))
-      return;
-
-   draw_flush(cell->draw);
-
    for (i = 0; i < CELL_MAX_SAMPLERS; i++) {
-      struct pipe_texture *tex = i < num ? texture[i] : NULL;
-
-      pipe_texture_reference((struct pipe_texture **) &cell->texture[i], tex);
+      struct pipe_texture *new_tex = i < num ? texture[i] : NULL;
+      if ((struct pipe_texture *) cell->texture[i] != new_tex) {
+         pipe_texture_reference((struct pipe_texture **) &cell->texture[i],
+                                new_tex);
+         changed |= (1 << i);
+      }
    }
-   cell->num_textures = num;
 
-   cell_update_texture_mapping(cell);
+   cell->num_textures = num;
 
-   cell->dirty |= CELL_NEW_TEXTURE;
+   if (changed) {
+      cell->dirty |= CELL_NEW_TEXTURE;
+      cell->dirty_textures |= changed;
+   }
 }
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_render.c b/src/gallium/drivers/cell/ppu/cell_render.c
index dd25ae880e5..79cb8df82fa 100644
--- a/src/gallium/drivers/cell/ppu/cell_render.c
+++ b/src/gallium/drivers/cell/ppu/cell_render.c
@@ -152,6 +152,7 @@ cell_flush_prim_buffer(struct cell_context *cell)
       struct cell_command_render *render = &cell_global.command[i].render;
       render->prim_type = PIPE_PRIM_TRIANGLES;
       render->num_verts = cell->prim_buffer.num_verts;
+      render->front_winding = cell->rasterizer->front_winding;
       render->vertex_size = cell->vertex_info->size * 4;
       render->xmin = cell->prim_buffer.xmin;
       render->ymin = cell->prim_buffer.ymin;
diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c
index 139b3719b62..d2235579507 100644
--- a/src/gallium/drivers/cell/ppu/cell_screen.c
+++ b/src/gallium/drivers/cell/ppu/cell_screen.c
@@ -58,9 +58,9 @@ cell_get_param(struct pipe_screen *screen, int param)
    case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
       return CELL_MAX_SAMPLERS;
    case PIPE_CAP_NPOT_TEXTURES:
-      return 0;
+      return 1;
    case PIPE_CAP_TWO_SIDED_STENCIL:
-      return 0;
+      return 1;
    case PIPE_CAP_GLSL:
       return 1;
    case PIPE_CAP_S3TC:
@@ -68,21 +68,21 @@ cell_get_param(struct pipe_screen *screen, int param)
    case PIPE_CAP_ANISOTROPIC_FILTER:
       return 0;
    case PIPE_CAP_POINT_SPRITE:
-      return 0;
+      return 1;
    case PIPE_CAP_MAX_RENDER_TARGETS:
       return 1;
    case PIPE_CAP_OCCLUSION_QUERY:
-      return 0;
+      return 1;
    case PIPE_CAP_TEXTURE_SHADOW_MAP:
-      return 0;
+      return 10;
    case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
-      return 12; /* max 2Kx2K */
+      return CELL_MAX_TEXTURE_LEVELS;
    case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
       return 8;  /* max 128x128x128 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
-      return 12; /* max 2Kx2K */
+      return CELL_MAX_TEXTURE_LEVELS;
    default:
-      return 0;
+      return 10;
    }
 }
 
@@ -108,7 +108,7 @@ cell_get_paramf(struct pipe_screen *screen, int param)
       return 16.0; /* arbitrary */
 
    default:
-      return 0;
+      return 10;
    }
 }
 
diff --git a/src/gallium/drivers/cell/ppu/cell_spu.c b/src/gallium/drivers/cell/ppu/cell_spu.c
index 9508227e298..28e5e6d706d 100644
--- a/src/gallium/drivers/cell/ppu/cell_spu.c
+++ b/src/gallium/drivers/cell/ppu/cell_spu.c
@@ -36,6 +36,7 @@
 #include "cell_spu.h"
 #include "pipe/p_format.h"
 #include "pipe/p_state.h"
+#include "util/u_memory.h"
 #include "cell/common.h"
 
 
@@ -52,6 +53,35 @@ struct cell_global_info cell_global;
 
 
 /**
+ * Scan /proc/cpuinfo to determine the timebase for the system.
+ * This is used by the SPUs to convert 'decrementer' ticks to seconds.
+ * There may be a better way to get this value...
+ */
+static unsigned
+get_timebase(void)
+{
+   FILE *f = fopen("/proc/cpuinfo", "r");
+   unsigned timebase;
+
+   assert(f);
+   while (!feof(f)) {
+      char line[80];
+      fgets(line, sizeof(line), f);
+      if (strncmp(line, "timebase", 8) == 0) {
+         char *colon = strchr(line, ':');
+         if (colon) {
+            timebase = atoi(colon + 2);
+            break;
+         }
+      }
+   }
+   fclose(f);
+
+   return timebase;
+}
+
+
+/**
  * Write a 1-word message to the given SPE mailbox.
  */
 void
@@ -114,6 +144,7 @@ cell_start_spus(struct cell_context *cell)
 {
    static boolean one_time_init = FALSE;
    uint i, j;
+   uint timebase = get_timebase();
 
    if (one_time_init) {
       fprintf(stderr, "PPU: Multiple rendering contexts not yet supported "
@@ -123,24 +154,29 @@ cell_start_spus(struct cell_context *cell)
 
    one_time_init = TRUE;
 
-   assert(cell->num_spus <= MAX_SPUS);
-
-   ASSERT_ALIGN16(&cell_global.command[0]);
-   ASSERT_ALIGN16(&cell_global.command[1]);
+   assert(cell->num_spus <= CELL_MAX_SPUS);
 
    ASSERT_ALIGN16(&cell_global.inits[0]);
    ASSERT_ALIGN16(&cell_global.inits[1]);
 
+   /*
+    * Initialize the global 'inits' structure for each SPU.
+    * A pointer to the init struct will be passed to each SPU.
+    * The SPUs will then each grab their init info with mfc_get().
+    */
    for (i = 0; i < cell->num_spus; i++) {
       cell_global.inits[i].id = i;
       cell_global.inits[i].num_spus = cell->num_spus;
       cell_global.inits[i].debug_flags = cell->debug_flags;
-      cell_global.inits[i].cmd = &cell_global.command[i];
+      cell_global.inits[i].inv_timebase = 1000.0f / timebase;
+
       for (j = 0; j < CELL_NUM_BUFFERS; j++) {
          cell_global.inits[i].buffers[j] = cell->buffer[j];
       }
       cell_global.inits[i].buffer_status = &cell->buffer_status[0][0][0];
 
+      cell_global.inits[i].spu_functions = &cell->spu_functions;
+
       cell_global.spe_contexts[i] = spe_context_create(0, NULL);
       if (!cell_global.spe_contexts[i]) {
          fprintf(stderr, "spe_context_create() failed\n");
diff --git a/src/gallium/drivers/cell/ppu/cell_spu.h b/src/gallium/drivers/cell/ppu/cell_spu.h
index 137f26612e4..c93958a9ed5 100644
--- a/src/gallium/drivers/cell/ppu/cell_spu.h
+++ b/src/gallium/drivers/cell/ppu/cell_spu.h
@@ -30,14 +30,12 @@
 
 
 #include <libspe2.h>
-#include <libmisc.h>
+#include <pthread.h>
 #include "cell/common.h"
 
 #include "cell_context.h"
 
 
-#define MAX_SPUS 8
-
 /**
  * Global vars, for now anyway.
  */
@@ -46,14 +44,13 @@ struct cell_global_info
    /**
     * SPU/SPE handles, etc
     */
-   spe_context_ptr_t spe_contexts[MAX_SPUS];
-   pthread_t spe_threads[MAX_SPUS];
+   spe_context_ptr_t spe_contexts[CELL_MAX_SPUS];
+   pthread_t spe_threads[CELL_MAX_SPUS];
 
    /**
-    * Data sent to SPUs
+    * Data sent to SPUs at start-up
     */
-   struct cell_init_info inits[MAX_SPUS];
-   struct cell_command command[MAX_SPUS];
+   struct cell_init_info inits[CELL_MAX_SPUS];
 };
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_state.h b/src/gallium/drivers/cell/ppu/cell_state.h
index a7771a55a31..b193170f9ce 100644
--- a/src/gallium/drivers/cell/ppu/cell_state.h
+++ b/src/gallium/drivers/cell/ppu/cell_state.h
@@ -44,8 +44,9 @@
 #define CELL_NEW_TEXTURE       0x800
 #define CELL_NEW_VERTEX        0x1000
 #define CELL_NEW_VS            0x2000
-#define CELL_NEW_CONSTANTS     0x4000
-#define CELL_NEW_VERTEX_INFO   0x8000
+#define CELL_NEW_VS_CONSTANTS  0x4000
+#define CELL_NEW_FS_CONSTANTS  0x8000
+#define CELL_NEW_VERTEX_INFO   0x10000
 
 
 extern void
diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c
index 2da3097983c..0a0af81f53f 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_emit.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c
@@ -25,18 +25,155 @@
  * 
  **************************************************************************/
 
+#include "pipe/p_inlines.h"
 #include "util/u_memory.h"
 #include "cell_context.h"
 #include "cell_gen_fragment.h"
 #include "cell_state.h"
 #include "cell_state_emit.h"
-#include "cell_state_per_fragment.h"
 #include "cell_batch.h"
 #include "cell_texture.h"
 #include "draw/draw_context.h"
 #include "draw/draw_private.h"
 
 
+/**
+ * Find/create a cell_command_fragment_ops object corresponding to the
+ * current blend/stencil/z/colormask/etc. state.
+ */
+static struct cell_command_fragment_ops *
+lookup_fragment_ops(struct cell_context *cell)
+{
+   struct cell_fragment_ops_key key;
+   struct cell_command_fragment_ops *ops;
+
+   /*
+    * Build key
+    */
+   memset(&key, 0, sizeof(key));
+   key.blend = *cell->blend;
+   key.blend_color = cell->blend_color;
+   key.dsa = *cell->depth_stencil;
+
+   if (cell->framebuffer.cbufs[0])
+      key.color_format = cell->framebuffer.cbufs[0]->format;
+   else
+      key.color_format = PIPE_FORMAT_NONE;
+
+   if (cell->framebuffer.zsbuf)
+      key.zs_format = cell->framebuffer.zsbuf->format;
+   else
+      key.zs_format = PIPE_FORMAT_NONE;
+
+   /*
+    * Look up key in cache.
+    */
+   ops = (struct cell_command_fragment_ops *)
+      util_keymap_lookup(cell->fragment_ops_cache, &key);
+
+   /*
+    * If not found, create/save new fragment ops command.
+    */
+   if (!ops) {
+      struct spe_function spe_code_front, spe_code_back;
+      unsigned int facing_dependent, total_code_size;
+
+      if (0)
+         debug_printf("**** Create New Fragment Ops\n");
+
+      /* Prepare the buffer that will hold the generated code.  The
+       * "0" passed in for the size means that the SPE code will
+       * use a default size.
+       */
+      spe_init_func(&spe_code_front, 0);
+      spe_init_func(&spe_code_back, 0);
+
+      /* Generate new code.  Always generate new code for both front-facing
+       * and back-facing fragments, even if it's the same code in both
+       * cases.
+       */
+      cell_gen_fragment_function(cell, CELL_FACING_FRONT, &spe_code_front);
+      cell_gen_fragment_function(cell, CELL_FACING_BACK, &spe_code_back);
+
+      /* Make sure the code is a multiple of 8 bytes long; this is
+       * required to ensure that the dual pipe instruction alignment
+       * is correct.  It's also important for the SPU unpacking,
+       * which assumes 8-byte boundaries.
+       */
+      unsigned int front_code_size = spe_code_size(&spe_code_front);
+      while (front_code_size % 8 != 0) {
+         spe_lnop(&spe_code_front);
+         front_code_size = spe_code_size(&spe_code_front);
+      }
+      unsigned int back_code_size = spe_code_size(&spe_code_back);
+      while (back_code_size % 8 != 0) {
+         spe_lnop(&spe_code_back);
+         back_code_size = spe_code_size(&spe_code_back);
+      }
+
+      /* Determine whether the code we generated is facing-dependent, by
+       * determining whether the generated code is different for the front-
+       * and back-facing fragments.
+       */
+      if (front_code_size == back_code_size && memcmp(spe_code_front.store, spe_code_back.store, front_code_size) == 0) {
+         /* Code is identical; only need one copy. */
+         facing_dependent = 0;
+         total_code_size = front_code_size;
+      }
+      else {
+         /* Code is different for front-facing and back-facing fragments.
+          * Need to send both copies.
+          */
+         facing_dependent = 1;
+         total_code_size = front_code_size + back_code_size;
+      }
+
+      /* alloc new fragment ops command.  Note that this structure
+       * has variant length based on the total code size required.
+       */
+      ops = CALLOC_VARIANT_LENGTH_STRUCT(cell_command_fragment_ops, total_code_size);
+      /* populate the new cell_command_fragment_ops object */
+      ops->opcode = CELL_CMD_STATE_FRAGMENT_OPS;
+      ops->total_code_size = total_code_size;
+      ops->front_code_index = 0;
+      memcpy(ops->code, spe_code_front.store, front_code_size);
+      if (facing_dependent) {
+        /* We have separate front- and back-facing code.  Append the
+         * back-facing code to the buffer.  Be careful because the code
+         * size is in bytes, but the buffer is of unsigned elements.
+         */
+        ops->back_code_index = front_code_size / sizeof(spe_code_front.store[0]);
+        memcpy(ops->code + ops->back_code_index, spe_code_back.store, back_code_size);
+      }
+      else {
+        /* Use the same code for front- and back-facing fragments */
+        ops->back_code_index = ops->front_code_index;
+      }
+
+      /* Set the fields for the fallback case.  Note that these fields
+       * (and the whole fallback case) will eventually go away.
+       */
+      ops->dsa = *cell->depth_stencil;
+      ops->blend = *cell->blend;
+      ops->blend_color = cell->blend_color;
+
+      /* insert cell_command_fragment_ops object into keymap/cache */
+      util_keymap_insert(cell->fragment_ops_cache, &key, ops, NULL);
+
+      /* release rtasm buffer */
+      spe_release_func(&spe_code_front);
+      spe_release_func(&spe_code_back);
+   }
+   else {
+      if (0)
+         debug_printf("**** Re-use Fragment Ops\n");
+   }
+
+   return ops;
+}
+
+
+
 static void
 emit_state_cmd(struct cell_context *cell, uint cmd,
                const void *state, uint state_size)
@@ -73,6 +210,13 @@ cell_emit_state(struct cell_context *cell)
 #endif
    }
 
+   if (cell->dirty & (CELL_NEW_RASTERIZER)) {
+      struct cell_command_rasterizer *rast =
+         cell_batch_alloc(cell, sizeof(*rast));
+      rast->opcode = CELL_CMD_STATE_RASTERIZER;
+      rast->rasterizer = *cell->rasterizer;
+   }
+
    if (cell->dirty & (CELL_NEW_FS)) {
       /* Send new fragment program to SPUs */
       struct cell_command_fragment_program *fp
@@ -90,59 +234,81 @@ cell_emit_state(struct cell_context *cell)
       }
    }
 
+   if (cell->dirty & (CELL_NEW_FS_CONSTANTS)) {
+      const uint shader = PIPE_SHADER_FRAGMENT;
+      const uint num_const = cell->constants[shader].size / sizeof(float);
+      uint i, j;
+      float *buf = cell_batch_alloc(cell, 16 + num_const * sizeof(float));
+      uint64_t *ibuf = (uint64_t *) buf;
+      const float *constants = pipe_buffer_map(cell->pipe.screen,
+                                               cell->constants[shader].buffer,
+                                               PIPE_BUFFER_USAGE_CPU_READ);
+      ibuf[0] = CELL_CMD_STATE_FS_CONSTANTS;
+      ibuf[1] = num_const;
+      j = 4;
+      for (i = 0; i < num_const; i++) {
+         buf[j++] = constants[i];
+      }
+      pipe_buffer_unmap(cell->pipe.screen, cell->constants[shader].buffer);
+   }
+
    if (cell->dirty & (CELL_NEW_FRAMEBUFFER |
                       CELL_NEW_DEPTH_STENCIL |
                       CELL_NEW_BLEND)) {
-      /* XXX we don't want to always do codegen here.  We should have
-       * a hash/lookup table to cache previous results...
-       */
-      struct cell_command_fragment_ops *fops
-            = cell_batch_alloc(cell, sizeof(*fops));
-      struct spe_function spe_code;
-
-      /* generate new code */
-      cell_gen_fragment_function(cell, &spe_code);
-      /* put the new code into the batch buffer */
-      fops->opcode = CELL_CMD_STATE_FRAGMENT_OPS;
-      memcpy(&fops->code, spe_code.store,
-             SPU_MAX_FRAGMENT_OPS_INSTS * SPE_INST_SIZE);
-      fops->dsa = cell->depth_stencil->base;
-      fops->blend = cell->blend->base;
-      /* free codegen buffer */
-      spe_release_func(&spe_code);
+      struct cell_command_fragment_ops *fops, *fops_cmd;
+      /* Note that cell_command_fragment_ops is a variant-sized record */
+      fops = lookup_fragment_ops(cell);
+      fops_cmd = cell_batch_alloc(cell, sizeof(*fops_cmd) + fops->total_code_size);
+      memcpy(fops_cmd, fops, sizeof(*fops) + fops->total_code_size);
    }
 
    if (cell->dirty & CELL_NEW_SAMPLER) {
       uint i;
       for (i = 0; i < CELL_MAX_SAMPLERS; i++) {
-         if (cell->sampler[i]) {
-            struct cell_command_sampler *sampler
-               = cell_batch_alloc(cell, sizeof(*sampler));
-            sampler->opcode = CELL_CMD_STATE_SAMPLER;
-            sampler->unit = i;
-            sampler->state = *cell->sampler[i];
+         if (cell->dirty_samplers & (1 << i)) {
+            if (cell->sampler[i]) {
+               struct cell_command_sampler *sampler
+                  = cell_batch_alloc(cell, sizeof(*sampler));
+               sampler->opcode = CELL_CMD_STATE_SAMPLER;
+               sampler->unit = i;
+               sampler->state = *cell->sampler[i];
+            }
          }
       }
+      cell->dirty_samplers = 0x0;
    }
 
    if (cell->dirty & CELL_NEW_TEXTURE) {
       uint i;
       for (i = 0;i < CELL_MAX_SAMPLERS; i++) {
-         struct cell_command_texture *texture
-            =  cell_batch_alloc(cell, sizeof(*texture));
-         texture->opcode = CELL_CMD_STATE_TEXTURE;
-         texture->unit = i;
-         if (cell->texture[i]) {
-            texture->start = cell->texture[i]->tiled_data;
-            texture->width = cell->texture[i]->base.width[0];
-            texture->height = cell->texture[i]->base.height[0];
-         }
-         else {
-            texture->start = NULL;
-            texture->width = 1;
-            texture->height = 1;
+         if (cell->dirty_textures & (1 << i)) {
+            struct cell_command_texture *texture
+               =  cell_batch_alloc(cell, sizeof(*texture));
+            texture->opcode = CELL_CMD_STATE_TEXTURE;
+            texture->unit = i;
+            if (cell->texture[i]) {
+               uint level;
+               for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
+                  texture->start[level] = cell->texture[i]->tiled_mapped[level];
+                  texture->width[level] = cell->texture[i]->base.width[level];
+                  texture->height[level] = cell->texture[i]->base.height[level];
+                  texture->depth[level] = cell->texture[i]->base.depth[level];
+               }
+               texture->target = cell->texture[i]->base.target;
+            }
+            else {
+               uint level;
+               for (level = 0; level < CELL_MAX_TEXTURE_LEVELS; level++) {
+                  texture->start[level] = NULL;
+                  texture->width[level] = 0;
+                  texture->height[level] = 0;
+                  texture->depth[level] = 0;
+               }
+               texture->target = 0;
+            }
          }
       }
+      cell->dirty_textures = 0x0;
    }
 
    if (cell->dirty & CELL_NEW_VERTEX_INFO) {
diff --git a/src/gallium/drivers/cell/ppu/cell_state_shader.c b/src/gallium/drivers/cell/ppu/cell_state_shader.c
index 3a0d066da2a..cda39f8d592 100644
--- a/src/gallium/drivers/cell/ppu/cell_state_shader.c
+++ b/src/gallium/drivers/cell/ppu/cell_state_shader.c
@@ -191,13 +191,18 @@ cell_set_constant_buffer(struct pipe_context *pipe,
    assert(shader < PIPE_SHADER_TYPES);
    assert(index == 0);
 
+   draw_flush(cell->draw);
+
    /* note: reference counting */
    winsys_buffer_reference(ws,
                         &cell->constants[shader].buffer,
                         buf->buffer);
    cell->constants[shader].size = buf->size;
 
-   cell->dirty |= CELL_NEW_CONSTANTS;
+   if (shader == PIPE_SHADER_VERTEX)
+      cell->dirty |= CELL_NEW_VS_CONSTANTS;
+   else if (shader == PIPE_SHADER_FRAGMENT)
+      cell->dirty |= CELL_NEW_FS_CONSTANTS;
 }
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_surface.c b/src/gallium/drivers/cell/ppu/cell_surface.c
index 732c64082ef..c9203fee087 100644
--- a/src/gallium/drivers/cell/ppu/cell_surface.c
+++ b/src/gallium/drivers/cell/ppu/cell_surface.c
@@ -27,6 +27,7 @@
 
 #include "util/u_rect.h"
 #include "cell_context.h"
+#include "cell_surface.h"
 
 
 void
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index b6590dfb86e..9f83ab8fa4c 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -28,6 +28,7 @@
   * Authors:
   *   Keith Whitwell <keith@tungstengraphics.com>
   *   Michel Dänzer <michel@tungstengraphics.com>
+  *   Brian Paul
   */
 
 #include "pipe/p_context.h"
@@ -42,30 +43,31 @@
 #include "cell_texture.h"
 
 
-/* Simple, maximally packed layout.
- */
 
-static unsigned minify( unsigned d )
+static unsigned
+minify(unsigned d)
 {
    return MAX2(1, d>>1);
 }
 
 
 static void
-cell_texture_layout(struct cell_texture * spt)
+cell_texture_layout(struct cell_texture *ct)
 {
-   struct pipe_texture *pt = &spt->base;
+   struct pipe_texture *pt = &ct->base;
    unsigned level;
    unsigned width = pt->width[0];
    unsigned height = pt->height[0];
    unsigned depth = pt->depth[0];
 
-   spt->buffer_size = 0;
+   ct->buffer_size = 0;
 
    for ( level = 0 ; level <= pt->last_level ; level++ ) {
       unsigned size;
       unsigned w_tile, h_tile;
 
+      assert(level < CELL_MAX_TEXTURE_LEVELS);
+
       /* width, height, rounded up to tile size */
       w_tile = align(width, TILE_SIZE);
       h_tile = align(height, TILE_SIZE);
@@ -76,9 +78,9 @@ cell_texture_layout(struct cell_texture * spt)
       pt->nblocksx[level] = pf_get_nblocksx(&pt->block, w_tile);  
       pt->nblocksy[level] = pf_get_nblocksy(&pt->block, h_tile);  
 
-      spt->stride[level] = pt->nblocksx[level] * pt->block.size;
+      ct->stride[level] = pt->nblocksx[level] * pt->block.size;
 
-      spt->level_offset[level] = spt->buffer_size;
+      ct->level_offset[level] = ct->buffer_size;
 
       size = pt->nblocksx[level] * pt->nblocksy[level] * pt->block.size;
       if (pt->target == PIPE_TEXTURE_CUBE)
@@ -86,7 +88,7 @@ cell_texture_layout(struct cell_texture * spt)
       else
          size *= depth;
 
-      spt->buffer_size += size;
+      ct->buffer_size += size;
 
       width  = minify(width);
       height = minify(height);
@@ -100,26 +102,25 @@ cell_texture_create(struct pipe_screen *screen,
                     const struct pipe_texture *templat)
 {
    struct pipe_winsys *ws = screen->winsys;
-   struct cell_texture *spt = CALLOC_STRUCT(cell_texture);
-   if (!spt)
+   struct cell_texture *ct = CALLOC_STRUCT(cell_texture);
+   if (!ct)
       return NULL;
 
-   spt->base = *templat;
-   spt->base.refcount = 1;
-   spt->base.screen = screen;
+   ct->base = *templat;
+   ct->base.refcount = 1;
+   ct->base.screen = screen;
 
-   cell_texture_layout(spt);
+   cell_texture_layout(ct);
 
-   spt->buffer = ws->buffer_create(ws, 32,
-                                   PIPE_BUFFER_USAGE_PIXEL,
-                                   spt->buffer_size);
+   ct->buffer = ws->buffer_create(ws, 32, PIPE_BUFFER_USAGE_PIXEL,
+                                  ct->buffer_size);
 
-   if (!spt->buffer) {
-      FREE(spt);
+   if (!ct->buffer) {
+      FREE(ct);
       return NULL;
    }
 
-   return &spt->base;
+   return &ct->base;
 }
 
 
@@ -135,177 +136,310 @@ cell_texture_release(struct pipe_screen *screen,
        __FUNCTION__, (void *) *pt, (*pt)->refcount - 1);
    */
    if (--(*pt)->refcount <= 0) {
-      struct cell_texture *spt = cell_texture(*pt);
+      /* Delete this texture now.
+       * But note that the underlying pipe_buffer may linger...
+       */
+      struct cell_texture *ct = cell_texture(*pt);
+      uint i;
 
       /*
-      DBG("%s deleting %p\n", __FUNCTION__, (void *) spt);
+      DBG("%s deleting %p\n", __FUNCTION__, (void *) ct);
       */
 
-      pipe_buffer_reference(screen, &spt->buffer, NULL);
+      pipe_buffer_reference(screen, &ct->buffer, NULL);
 
-      FREE(spt);
+      for (i = 0; i < CELL_MAX_TEXTURE_LEVELS; i++) {
+         /* Unreference the tiled image buffer.
+          * It may not actually be deleted until a fence is hit.
+          */
+         if (ct->tiled_buffer[i]) {
+            ct->tiled_mapped[i] = NULL;
+            winsys_buffer_reference(screen->winsys, &ct->tiled_buffer[i], NULL);
+         }
+      }
+
+      FREE(ct);
    }
    *pt = NULL;
 }
 
 
-#if 0
+
+/**
+ * Convert image from linear layout to tiled layout.  4-byte pixels.
+ */
 static void
-cell_texture_update(struct pipe_context *pipe, struct pipe_texture *texture,
-                    uint face, uint levelsMask)
+twiddle_image_uint(uint w, uint h, uint tile_size, uint *dst,
+                   uint src_stride, const uint *src)
 {
-   /* XXX TO DO:  re-tile the texture data ... */
+   const uint tile_size2 = tile_size * tile_size;
+   const uint h_t = (h + tile_size - 1) / tile_size;
+   const uint w_t = (w + tile_size - 1) / tile_size;
 
-}
-#endif
+   uint it, jt;  /* tile counters */
+   uint i, j;    /* intra-tile counters */
 
+   src_stride /= 4; /* convert from bytes to pixels */
 
-static struct pipe_surface *
-cell_get_tex_surface(struct pipe_screen *screen,
-                     struct pipe_texture *pt,
-                     unsigned face, unsigned level, unsigned zslice,
-                     unsigned usage)
-{
-   struct pipe_winsys *ws = screen->winsys;
-   struct cell_texture *spt = cell_texture(pt);
-   struct pipe_surface *ps;
+   /* loop over dest tiles */
+   for (it = 0; it < h_t; it++) {
+      for (jt = 0; jt < w_t; jt++) {
+         /* start of dest tile: */
+         uint *tdst = dst + (it * w_t + jt) * tile_size2;
 
-   ps = ws->surface_alloc(ws);
-   if (ps) {
-      assert(ps->refcount);
-      assert(ps->winsys);
-      winsys_buffer_reference(ws, &ps->buffer, spt->buffer);
-      ps->format = pt->format;
-      ps->block = pt->block;
-      ps->width = pt->width[level];
-      ps->height = pt->height[level];
-      ps->nblocksx = pt->nblocksx[level];
-      ps->nblocksy = pt->nblocksy[level];
-      ps->stride = spt->stride[level];
-      ps->offset = spt->level_offset[level];
-      ps->usage = usage;
+         /* compute size of this tile (may be smaller than tile_size) */
+         /* XXX note: a compiler bug was found here. That's why the code
+          * looks as it does.
+          */
+         uint tile_width = w - jt * tile_size;
+         tile_width = MIN2(tile_width, tile_size);
+         uint tile_height = h - it * tile_size;
+         tile_height = MIN2(tile_height, tile_size);
 
-      /* XXX may need to override usage flags (see sp_texture.c) */
+         /* loop over texels in the tile */
+         for (i = 0; i < tile_height; i++) {
+            for (j = 0; j < tile_width; j++) {
+               const uint srci = it * tile_size + i;
+               const uint srcj = jt * tile_size + j;
+               ASSERT(srci < h);
+               ASSERT(srcj < w);
+               tdst[i * tile_size + j] = src[srci * src_stride + srcj];
+            }
+         }
+      }
+   }
+}
 
-      pipe_texture_reference(&ps->texture, pt); 
-      ps->face = face;
-      ps->level = level;
-      ps->zslice = zslice;
 
-      if (pt->target == PIPE_TEXTURE_CUBE || pt->target == PIPE_TEXTURE_3D) {
-	 ps->offset += ((pt->target == PIPE_TEXTURE_CUBE) ? face : zslice) *
-		       ps->nblocksy *
-		       ps->stride;
-      }
-      else {
-	 assert(face == 0);
-	 assert(zslice == 0);
+/**
+ * For Cell.  Basically, rearrange the pixels/quads from this layout:
+ *  +--+--+--+--+
+ *  |p0|p1|p2|p3|....
+ *  +--+--+--+--+
+ *
+ * to this layout:
+ *  +--+--+
+ *  |p0|p1|....
+ *  +--+--+
+ *  |p2|p3|
+ *  +--+--+
+ */
+static void
+twiddle_tile(const uint *tileIn, uint *tileOut)
+{
+   int y, x;
+
+   for (y = 0; y < TILE_SIZE; y+=2) {
+      for (x = 0; x < TILE_SIZE; x+=2) {
+         int k = 4 * (y/2 * TILE_SIZE/2 + x/2);
+         tileOut[y * TILE_SIZE + (x + 0)] = tileIn[k];
+         tileOut[y * TILE_SIZE + (x + 1)] = tileIn[k+1];
+         tileOut[(y + 1) * TILE_SIZE + (x + 0)] = tileIn[k+2];
+         tileOut[(y + 1) * TILE_SIZE + (x + 1)] = tileIn[k+3];
       }
    }
-   return ps;
 }
 
 
-
 /**
- * Copy tile data from linear layout to tiled layout.
- * XXX this should be rolled into the future surface-creation code.
- * XXX also need "untile" code...
+ * Convert image from tiled layout to linear layout.  4-byte pixels.
  */
 static void
-tile_copy_data(uint w, uint h, uint tile_size, uint *dst, const uint *src)
+untwiddle_image_uint(uint w, uint h, uint tile_size, uint *dst,
+                     uint dst_stride, const uint *src)
 {
    const uint tile_size2 = tile_size * tile_size;
-   const uint h_t = h / tile_size, w_t = w / tile_size;
-
+   const uint h_t = (h + tile_size - 1) / tile_size;
+   const uint w_t = (w + tile_size - 1) / tile_size;
+   uint *tile_buf;
    uint it, jt;  /* tile counters */
    uint i, j;    /* intra-tile counters */
 
-   /* loop over dest tiles */
+   dst_stride /= 4; /* convert from bytes to pixels */
+
+   tile_buf = align_malloc(tile_size * tile_size * 4, 16);
+   
+   /* loop over src tiles */
    for (it = 0; it < h_t; it++) {
       for (jt = 0; jt < w_t; jt++) {
-         /* start of dest tile: */
-         uint *tdst = dst + (it * w_t + jt) * tile_size2;
+         /* start of src tile: */
+         const uint *tsrc = src + (it * w_t + jt) * tile_size2;
+         
+         twiddle_tile(tsrc, tile_buf);
+         tsrc = tile_buf;
+
+         /* compute size of this tile (may be smaller than tile_size) */
+         /* XXX note: a compiler bug was found here. That's why the code
+          * looks as it does.
+          */
+         uint tile_width = w - jt * tile_size;
+         tile_width = MIN2(tile_width, tile_size);
+         uint tile_height = h - it * tile_size;
+         tile_height = MIN2(tile_height, tile_size);
+
          /* loop over texels in the tile */
-         for (i = 0; i < tile_size; i++) {
-            for (j = 0; j < tile_size; j++) {
-               const uint srci = it * tile_size + i;
-               const uint srcj = jt * tile_size + j;
-               *tdst++ = src[srci * w + srcj];
+         for (i = 0; i < tile_height; i++) {
+            for (j = 0; j < tile_width; j++) {
+               uint dsti = it * tile_size + i;
+               uint dstj = jt * tile_size + j;
+               ASSERT(dsti < h);
+               ASSERT(dstj < w);
+               dst[dsti * dst_stride + dstj] = tsrc[i * tile_size + j];
             }
          }
       }
    }
-}
 
+   align_free(tile_buf);
+}
 
 
 /**
  * Convert linear texture image data to tiled format for SPU usage.
- * XXX recast this in terms of pipe_surfaces (aka texture views).
  */
 static void
-cell_tile_texture(struct cell_context *cell,
-                  struct cell_texture *texture)
+cell_twiddle_texture(struct pipe_screen *screen,
+                     struct pipe_surface *surface)
 {
-   struct pipe_screen *screen = cell->pipe.screen;
-   uint face = 0, level = 0, zslice = 0;
-   struct pipe_surface *surf;
-   const uint w = texture->base.width[0], h = texture->base.height[0];
-   const uint *src;
-
-   /* temporary restrictions: */
-   assert(w >= TILE_SIZE);
-   assert(h >= TILE_SIZE);
-   assert(w % TILE_SIZE == 0);
-   assert(h % TILE_SIZE == 0);
-
-   surf = screen->get_tex_surface(screen, &texture->base, face, level, zslice,
-                                  PIPE_BUFFER_USAGE_CPU_WRITE);
-   ASSERT(surf);
-
-   src = (const uint *) pipe_surface_map(surf, PIPE_BUFFER_USAGE_CPU_WRITE);
-
-   if (texture->tiled_data) {
-      align_free(texture->tiled_data);
+   struct cell_texture *ct = cell_texture(surface->texture);
+   const uint level = surface->level;
+   const uint texWidth = ct->base.width[level];
+   const uint texHeight = ct->base.height[level];
+   const uint bufWidth = align(texWidth, TILE_SIZE);
+   const uint bufHeight = align(texHeight, TILE_SIZE);
+   const void *map = pipe_buffer_map(screen, surface->buffer,
+                                     PIPE_BUFFER_USAGE_CPU_READ);
+   const uint *src = (const uint *) ((const ubyte *) map + surface->offset);
+
+   switch (ct->base.format) {
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_S8Z24_UNORM:
+      {
+         int numFaces = ct->base.target == PIPE_TEXTURE_CUBE ? 6 : 1;
+         int offset = bufWidth * bufHeight * 4 * surface->face;
+         uint *dst;
+
+         if (!ct->tiled_buffer[level]) {
+            /* allocate buffer for tiled data now */
+            struct pipe_winsys *ws = screen->winsys;
+            uint bytes = bufWidth * bufHeight * 4 * numFaces;
+            ct->tiled_buffer[level] = ws->buffer_create(ws, 16,
+                                                        PIPE_BUFFER_USAGE_PIXEL,
+                                                        bytes);
+            /* and map it */
+            ct->tiled_mapped[level] = ws->buffer_map(ws, ct->tiled_buffer[level],
+                                                     PIPE_BUFFER_USAGE_GPU_READ);
+         }
+         dst = (uint *) ((ubyte *) ct->tiled_mapped[level] + offset);
+
+         twiddle_image_uint(texWidth, texHeight, TILE_SIZE, dst,
+                            surface->stride, src);
+      }
+      break;
+   default:
+      printf("Cell: twiddle unsupported texture format %s\n", pf_name(ct->base.format));
+      ;
    }
-   texture->tiled_data = align_malloc(w * h * 4, 16);
 
-   tile_copy_data(w, h, TILE_SIZE, texture->tiled_data, src);
+   pipe_buffer_unmap(screen, surface->buffer);
+}
+
+
+/**
+ * Convert SPU tiled texture image data to linear format for app usage.
+ */
+static void
+cell_untwiddle_texture(struct pipe_screen *screen,
+                     struct pipe_surface *surface)
+{
+   struct cell_texture *ct = cell_texture(surface->texture);
+   const uint level = surface->level;
+   const uint texWidth = ct->base.width[level];
+   const uint texHeight = ct->base.height[level];
+   const void *map = pipe_buffer_map(screen, surface->buffer,
+                                     PIPE_BUFFER_USAGE_CPU_READ);
+   const uint *src = (const uint *) ((const ubyte *) map + surface->offset);
+
+   switch (ct->base.format) {
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_S8Z24_UNORM:
+      {
+         int numFaces = ct->base.target == PIPE_TEXTURE_CUBE ? 6 : 1;
+         int offset = surface->stride * texHeight * 4 * surface->face;
+         uint *dst;
+
+         if (!ct->untiled_data[level]) {
+            ct->untiled_data[level] =
+               align_malloc(surface->stride * texHeight * 4 * numFaces, 16);
+         }
+
+         dst = (uint *) ((ubyte *) ct->untiled_data[level] + offset);
 
-   pipe_surface_unmap(surf);
+         untwiddle_image_uint(texWidth, texHeight, TILE_SIZE, dst,
+                              surface->stride, src);
+      }
+      break;
+   default:
+      {
+         ct->untiled_data[level] = NULL;
+         printf("Cell: untwiddle unsupported texture format %s\n", pf_name(ct->base.format));
+      }
+   }
 
-   pipe_surface_reference(&surf, NULL);
+   pipe_buffer_unmap(screen, surface->buffer);
 }
 
 
-void
-cell_update_texture_mapping(struct cell_context *cell)
+static struct pipe_surface *
+cell_get_tex_surface(struct pipe_screen *screen,
+                     struct pipe_texture *pt,
+                     unsigned face, unsigned level, unsigned zslice,
+                     unsigned usage)
 {
-#if 0
-   uint face = 0, level = 0, zslice = 0;
-#endif
-   uint i;
-
-   for (i = 0; i < CELL_MAX_SAMPLERS; i++) {
-      if (cell->texture[i])
-         cell_tile_texture(cell, cell->texture[i]);
-   }
+   struct pipe_winsys *ws = screen->winsys;
+   struct cell_texture *ct = cell_texture(pt);
+   struct pipe_surface *ps;
 
-#if 0
-   if (cell->tex_surf && cell->tex_map) {
-      pipe_surface_unmap(cell->tex_surf);
-      cell->tex_map = NULL;
-   }
+   ps = ws->surface_alloc(ws);
+   if (ps) {
+      assert(ps->refcount);
+      assert(ps->winsys);
+      winsys_buffer_reference(ws, &ps->buffer, ct->buffer);
+      ps->format = pt->format;
+      ps->block = pt->block;
+      ps->width = pt->width[level];
+      ps->height = pt->height[level];
+      ps->nblocksx = pt->nblocksx[level];
+      ps->nblocksy = pt->nblocksy[level];
+      ps->stride = ct->stride[level];
+      ps->offset = ct->level_offset[level];
+      ps->usage = usage;
 
-   /* XXX free old surface */
+      /* XXX may need to override usage flags (see sp_texture.c) */
 
-   cell->tex_surf = cell_get_tex_surface(&cell->pipe,
-                                         &cell->texture[0]->base,
-                                         face, level, zslice);
+      pipe_texture_reference(&ps->texture, pt); 
+      ps->face = face;
+      ps->level = level;
+      ps->zslice = zslice;
 
-   cell->tex_map = pipe_surface_map(cell->tex_surf);
-#endif
+      if (pt->target == PIPE_TEXTURE_CUBE || pt->target == PIPE_TEXTURE_3D) {
+	         ps->offset += ((pt->target == PIPE_TEXTURE_CUBE) ? face : zslice) *
+		      ps->nblocksy *
+		      ps->stride;
+      }
+      else {
+         assert(face == 0);
+         assert(zslice == 0);
+      }
+
+      if (ps->usage & PIPE_BUFFER_USAGE_CPU_READ) {
+         /* convert from tiled to linear layout */
+         cell_untwiddle_texture(screen, ps);
+      }
+   }
+   return ps;
 }
 
 
@@ -313,11 +447,17 @@ static void
 cell_tex_surface_release(struct pipe_screen *screen, 
                          struct pipe_surface **s)
 {
-   /* Effectively do the texture_update work here - if texture images
-    * needed post-processing to put them into hardware layout, this is
-    * where it would happen.  For softpipe, nothing to do.
-    */
-   assert ((*s)->texture);
+   struct cell_texture *ct = cell_texture((*s)->texture);
+   const uint level = (*s)->level;
+
+   if (((*s)->usage & PIPE_BUFFER_USAGE_CPU_READ) && (ct->untiled_data[level]))
+   {
+      align_free(ct->untiled_data[level]);
+      ct->untiled_data[level] = NULL;
+   }
+
+   /* XXX if done rendering to teximage, re-tile */
+
    pipe_texture_reference(&(*s)->texture, NULL); 
 
    screen->winsys->surface_release(screen->winsys, s);
@@ -325,11 +465,15 @@ cell_tex_surface_release(struct pipe_screen *screen,
 
 
 static void *
-cell_surface_map( struct pipe_screen *screen,
-                  struct pipe_surface *surface,
-                  unsigned flags )
+cell_surface_map(struct pipe_screen *screen,
+                 struct pipe_surface *surface,
+                 unsigned flags)
 {
    ubyte *map;
+   struct cell_texture *ct = cell_texture(surface->texture);
+   const uint level = surface->level;
+
+   assert(ct);
 
    if (flags & ~surface->usage) {
       assert(0);
@@ -339,22 +483,15 @@ cell_surface_map( struct pipe_screen *screen,
    map = pipe_buffer_map( screen, surface->buffer, flags );
    if (map == NULL)
       return NULL;
-
-   /* May want to different things here depending on read/write nature
-    * of the map:
-    */
-   if (surface->texture &&
-       (flags & PIPE_BUFFER_USAGE_CPU_WRITE)) 
+   else
    {
-      /* Do something to notify sharing contexts of a texture change.
-       * In softpipe, that would mean flushing the texture cache.
-       */
-#if 0
-      cell_screen(screen)->timestamp++;
-#endif
+      if ((surface->usage & PIPE_BUFFER_USAGE_CPU_READ) && (ct->untiled_data[level])) {
+         return (void *) ((ubyte *) ct->untiled_data[level] + surface->offset);
+      }
+      else {
+         return (void *) (map + surface->offset);
+      }
    }
-   
-   return map + surface->offset;
 }
 
 
@@ -362,17 +499,21 @@ static void
 cell_surface_unmap(struct pipe_screen *screen,
                    struct pipe_surface *surface)
 {
-   pipe_buffer_unmap( screen, surface->buffer );
-}
+   struct cell_texture *ct = cell_texture(surface->texture);
 
+   assert(ct);
 
-void
-cell_init_texture_functions(struct cell_context *cell)
-{
-   /*cell->pipe.texture_update = cell_texture_update;*/
+   if ((ct->base.tex_usage & PIPE_TEXTURE_USAGE_SAMPLER) &&
+       (surface->usage & PIPE_BUFFER_USAGE_CPU_WRITE)) {
+      /* convert from linear to tiled layout */
+      cell_twiddle_texture(screen, surface);
+   }
+
+   pipe_buffer_unmap( screen, surface->buffer );
 }
 
 
+
 void
 cell_init_screen_texture_funcs(struct pipe_screen *screen)
 {
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.h b/src/gallium/drivers/cell/ppu/cell_texture.h
index 6d37e95ebce..7018b0c9bf7 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.h
+++ b/src/gallium/drivers/cell/ppu/cell_texture.h
@@ -40,15 +40,19 @@ struct cell_texture
 {
    struct pipe_texture base;
 
-   unsigned long level_offset[PIPE_MAX_TEXTURE_LEVELS];
-   unsigned long stride[PIPE_MAX_TEXTURE_LEVELS];
+   unsigned long level_offset[CELL_MAX_TEXTURE_LEVELS];
+   unsigned long stride[CELL_MAX_TEXTURE_LEVELS];
 
    /* The data is held here:
     */
    struct pipe_buffer *buffer;
    unsigned long buffer_size;
 
-   void *tiled_data;  /* XXX this may be temporary */ /*ALIGN16*/
+   /** Texture data in tiled layout is held here */
+   struct pipe_buffer *tiled_buffer[CELL_MAX_TEXTURE_LEVELS];
+   /** Mapped, tiled texture data */
+   void *tiled_mapped[CELL_MAX_TEXTURE_LEVELS];
+   void *untiled_data[CELL_MAX_TEXTURE_LEVELS];
 };
 
 
@@ -62,14 +66,6 @@ cell_texture(struct pipe_texture *pt)
 
 
 extern void
-cell_update_texture_mapping(struct cell_context *cell);
-
-
-extern void
-cell_init_texture_functions(struct cell_context *cell);
-
-
-extern void
 cell_init_screen_texture_funcs(struct pipe_screen *screen);
 
 
diff --git a/src/gallium/drivers/cell/ppu/cell_vbuf.c b/src/gallium/drivers/cell/ppu/cell_vbuf.c
index aa63435b934..65ba51b6bb2 100644
--- a/src/gallium/drivers/cell/ppu/cell_vbuf.c
+++ b/src/gallium/drivers/cell/ppu/cell_vbuf.c
@@ -38,6 +38,7 @@
 
 #include "cell_batch.h"
 #include "cell_context.h"
+#include "cell_fence.h"
 #include "cell_flush.h"
 #include "cell_spu.h"
 #include "cell_vbuf.h"
@@ -108,6 +109,11 @@ cell_vbuf_release_vertices(struct vbuf_render *vbr, void *vertices,
           __FUNCTION__, cvbr->vertex_buf, vertices_used);
    */
 
+   /* Make sure texture buffers aren't released until we're done rendering
+    * with them.
+    */
+   cell_add_fenced_textures(cell);
+
    /* Tell SPUs they can release the vert buf */
    if (cvbr->vertex_buf != ~0U) {
       struct cell_command_release_verts *release
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
index 566df7f59e3..9cba537d9eb 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
@@ -73,8 +73,8 @@ emit_matrix_transpose(struct spe_function *p,
    int col3;
 
 
-   spe_lqd(p, shuf_hi, shuf_ptr, 3);
-   spe_lqd(p, shuf_lo, shuf_ptr, 4);
+   spe_lqd(p, shuf_hi, shuf_ptr, 3*16);
+   spe_lqd(p, shuf_lo, shuf_ptr, 4*16);
    spe_shufb(p, t1, row0, row2, shuf_hi);
    spe_shufb(p, t2, row0, row2, shuf_lo);
 
@@ -122,13 +122,13 @@ emit_matrix_transpose(struct spe_function *p,
     */
    switch (count) {
    case 4:
-      spe_stqd(p, col3, dest_ptr, 3);
+      spe_stqd(p, col3, dest_ptr, 3 * 16);
    case 3:
-      spe_stqd(p, col2, dest_ptr, 2);
+      spe_stqd(p, col2, dest_ptr, 2 * 16);
    case 2:
-      spe_stqd(p, col1, dest_ptr, 1);
+      spe_stqd(p, col1, dest_ptr, 1 * 16);
    case 1:
-      spe_stqd(p, col0, dest_ptr, 0);
+      spe_stqd(p, col0, dest_ptr, 0 * 16);
    }
 
 
@@ -145,6 +145,8 @@ emit_matrix_transpose(struct spe_function *p,
 }
 
 
+#if 0
+/* This appears to not be used currently */
 static void
 emit_fetch(struct spe_function *p,
 	   unsigned in_ptr, unsigned *offset,
@@ -166,17 +168,17 @@ emit_fetch(struct spe_function *p,
    float scale_signed = 0.0;
    float scale_unsigned = 0.0;
 
-   spe_lqd(p, v0, in_ptr, 0 + offset[0]);
-   spe_lqd(p, v1, in_ptr, 1 + offset[0]);
-   spe_lqd(p, v2, in_ptr, 2 + offset[0]);
-   spe_lqd(p, v3, in_ptr, 3 + offset[0]);
+   spe_lqd(p, v0, in_ptr, (0 + offset[0]) * 16);
+   spe_lqd(p, v1, in_ptr, (1 + offset[0]) * 16);
+   spe_lqd(p, v2, in_ptr, (2 + offset[0]) * 16);
+   spe_lqd(p, v3, in_ptr, (3 + offset[0]) * 16);
    offset[0] += 4;
    
    switch (bytes) {
    case 1:
       scale_signed = 1.0f / 127.0f;
       scale_unsigned = 1.0f / 255.0f;
-      spe_lqd(p, tmp, shuf_ptr, 1);
+      spe_lqd(p, tmp, shuf_ptr, 1 * 16);
       spe_shufb(p, v0, v0, v0, tmp);
       spe_shufb(p, v1, v1, v1, tmp);
       spe_shufb(p, v2, v2, v2, tmp);
@@ -185,7 +187,7 @@ emit_fetch(struct spe_function *p,
    case 2:
       scale_signed = 1.0f / 32767.0f;
       scale_unsigned = 1.0f / 65535.0f;
-      spe_lqd(p, tmp, shuf_ptr, 2);
+      spe_lqd(p, tmp, shuf_ptr, 2 * 16);
       spe_shufb(p, v0, v0, v0, tmp);
       spe_shufb(p, v1, v1, v1, tmp);
       spe_shufb(p, v2, v2, v2, tmp);
@@ -241,11 +243,11 @@ emit_fetch(struct spe_function *p,
 
    switch (count) {
    case 1:
-      spe_stqd(p, float_zero, out_ptr, 1);
+      spe_stqd(p, float_zero, out_ptr, 1 * 16);
    case 2:
-      spe_stqd(p, float_zero, out_ptr, 2);
+      spe_stqd(p, float_zero, out_ptr, 2 * 16);
    case 3:
-      spe_stqd(p, float_one, out_ptr, 3);
+      spe_stqd(p, float_one, out_ptr, 3 * 16);
    }
 
    if (float_zero != -1) {
@@ -256,6 +258,7 @@ emit_fetch(struct spe_function *p,
       spe_release_register(p, float_one);
    }
 }
+#endif
 
 
 void cell_update_vertex_fetch(struct draw_context *draw)
diff --git a/src/gallium/drivers/cell/spu/.gitignore b/src/gallium/drivers/cell/spu/.gitignore
new file mode 100644
index 00000000000..2be9a2d3242
--- /dev/null
+++ b/src/gallium/drivers/cell/spu/.gitignore
@@ -0,0 +1 @@
+g3d_spu
diff --git a/src/gallium/drivers/cell/spu/Makefile b/src/gallium/drivers/cell/spu/Makefile
index 1ae0dfb8c10..116453b79c5 100644
--- a/src/gallium/drivers/cell/spu/Makefile
+++ b/src/gallium/drivers/cell/spu/Makefile
@@ -16,8 +16,10 @@ PROG_SPU_EMBED_O = $(PROG)_spu-embed.o
 
 
 SOURCES = \
-	spu_main.c \
+	spu_command.c \
 	spu_dcache.c \
+	spu_funcs.c \
+	spu_main.c \
 	spu_per_fragment_op.c \
 	spu_render.c \
 	spu_texture.c \
diff --git a/src/gallium/drivers/cell/spu/spu_colorpack.h b/src/gallium/drivers/cell/spu/spu_colorpack.h
index fd8dc6ded3e..d7ce0055248 100644
--- a/src/gallium/drivers/cell/spu/spu_colorpack.h
+++ b/src/gallium/drivers/cell/spu/spu_colorpack.h
@@ -31,6 +31,7 @@
 #define SPU_COLORPACK_H
 
 
+#include <transpose_matrix4x4.h>
 #include <spu_intrinsics.h>
 
 
@@ -84,10 +85,10 @@ spu_unpack_B8G8R8A8(uint color)
    vector unsigned int color_u4 = spu_splats(color);
    color_u4 = spu_shuffle(color_u4, color_u4,
                           ((vector unsigned char) {
-                             10, 10, 10, 10,
-                             5, 5, 5, 5,
+                             2, 2, 2, 2,
+                             1, 1, 1, 1,
                              0, 0, 0, 0,
-                             15, 15, 15, 15}) );
+                             3, 3, 3, 3}) );
    return spu_convtf(color_u4, 32);
 }
 
@@ -98,13 +99,47 @@ spu_unpack_A8R8G8B8(uint color)
    vector unsigned int color_u4 = spu_splats(color);
    color_u4 = spu_shuffle(color_u4, color_u4,
                           ((vector unsigned char) {
-                             5, 5, 5, 5,
-                             10, 10, 10, 10,
-                             15, 15, 15, 15,
+                             1, 1, 1, 1,
+                             2, 2, 2, 2,
+                             3, 3, 3, 3,
                              0, 0, 0, 0}) );
-
    return spu_convtf(color_u4, 32);
 }
 
 
+/**
+ * \param color_in - array of 32-bit packed ARGB colors
+ * \param color_out - returns float colors in RRRR, GGGG, BBBB, AAAA order
+ */
+static INLINE void
+spu_unpack_A8R8G8B8_transpose4(const vector unsigned int color_in[4],
+                               vector float color_out[4])
+{
+   vector unsigned int c0;
+
+   c0 = spu_shuffle(color_in[0], color_in[0],
+                    ((vector unsigned char) {
+                       1, 1, 1, 1,  2, 2, 2, 2,  3, 3, 3, 3,  0, 0, 0, 0}) );
+   color_out[0] = spu_convtf(c0, 32);
+
+   c0 = spu_shuffle(color_in[1], color_in[1],
+                    ((vector unsigned char) {
+                       1, 1, 1, 1,  2, 2, 2, 2,  3, 3, 3, 3,  0, 0, 0, 0}) );
+   color_out[1] = spu_convtf(c0, 32);
+
+   c0 = spu_shuffle(color_in[2], color_in[2],
+                    ((vector unsigned char) {
+                       1, 1, 1, 1,  2, 2, 2, 2,  3, 3, 3, 3,  0, 0, 0, 0}) );
+   color_out[2] = spu_convtf(c0, 32);
+
+   c0 = spu_shuffle(color_in[3], color_in[3],
+                    ((vector unsigned char) {
+                       1, 1, 1, 1,  2, 2, 2, 2,  3, 3, 3, 3,  0, 0, 0, 0}) );
+   color_out[3] = spu_convtf(c0, 32);
+
+   _transpose_matrix4x4(color_out, color_out);
+}
+
+
+
 #endif /* SPU_COLORPACK_H */
diff --git a/src/gallium/drivers/cell/spu/spu_command.c b/src/gallium/drivers/cell/spu/spu_command.c
new file mode 100644
index 00000000000..8500d19754e
--- /dev/null
+++ b/src/gallium/drivers/cell/spu/spu_command.c
@@ -0,0 +1,815 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+/**
+ * SPU command processing code
+ */
+
+
+#include <stdio.h>
+#include <libmisc.h>
+
+#include "pipe/p_defines.h"
+
+#include "spu_command.h"
+#include "spu_main.h"
+#include "spu_render.h"
+#include "spu_per_fragment_op.h"
+#include "spu_texture.h"
+#include "spu_tile.h"
+#include "spu_vertex_shader.h"
+#include "spu_dcache.h"
+#include "cell/common.h"
+
+
+struct spu_vs_context draw;
+
+
+/**
+ * Buffers containing dynamically generated SPU code:
+ */
+static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS]
+    ALIGN16_ATTRIB;
+
+
+
+static INLINE int
+align(int value, int alignment)
+{
+   return (value + alignment - 1) & ~(alignment - 1);
+}
+
+
+
+/**
+ * Tell the PPU that this SPU has finished copying a buffer to
+ * local store and that it may be reused by the PPU.
+ * This is done by writting a 16-byte batch-buffer-status block back into
+ * main memory (in cell_context->buffer_status[]).
+ */
+static void
+release_buffer(uint buffer)
+{
+   /* Evidently, using less than a 16-byte status doesn't work reliably */
+   static const vector unsigned int status = {CELL_BUFFER_STATUS_FREE,
+                                              CELL_BUFFER_STATUS_FREE,
+                                              CELL_BUFFER_STATUS_FREE,
+                                              CELL_BUFFER_STATUS_FREE};
+   const uint index = 4 * (spu.init.id * CELL_NUM_BUFFERS + buffer);
+   uint *dst = spu.init.buffer_status + index;
+
+   ASSERT(buffer < CELL_NUM_BUFFERS);
+
+   mfc_put((void *) &status,    /* src in local memory */
+           (unsigned int) dst,  /* dst in main memory */
+           sizeof(status),      /* size */
+           TAG_MISC,            /* tag is unimportant */
+           0, /* tid */
+           0  /* rid */);
+}
+
+
+/**
+ * Write CELL_FENCE_SIGNALLED back to the fence status qword in main memory.
+ * There's a qword of status per SPU.
+ */
+static void
+cmd_fence(struct cell_command_fence *fence_cmd)
+{
+   static const vector unsigned int status = {CELL_FENCE_SIGNALLED,
+                                              CELL_FENCE_SIGNALLED,
+                                              CELL_FENCE_SIGNALLED,
+                                              CELL_FENCE_SIGNALLED};
+   uint *dst = (uint *) fence_cmd->fence;
+   dst += 4 * spu.init.id;  /* main store/memory address, not local store */
+   ASSERT_ALIGN16(dst);
+   mfc_put((void *) &status,    /* src in local memory */
+           (unsigned int) dst,  /* dst in main memory */
+           sizeof(status),      /* size */
+           TAG_FENCE,           /* tag */
+           0, /* tid */
+           0  /* rid */);
+}
+
+
+static void
+cmd_clear_surface(const struct cell_command_clear_surface *clear)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "CLEAR SURF %u to 0x%08x\n", clear->surface, clear->value);
+
+   if (clear->surface == 0) {
+      spu.fb.color_clear_value = clear->value;
+      if (spu.init.debug_flags & CELL_DEBUG_CHECKER) {
+         uint x = (spu.init.id << 4) | (spu.init.id << 12) |
+            (spu.init.id << 20) | (spu.init.id << 28);
+         spu.fb.color_clear_value ^= x;
+      }
+   }
+   else {
+      spu.fb.depth_clear_value = clear->value;
+   }
+
+#define CLEAR_OPT 1
+#if CLEAR_OPT
+
+   /* Simply set all tiles' status to CLEAR.
+    * When we actually begin rendering into a tile, we'll initialize it to
+    * the clear value.  If any tiles go untouched during the frame,
+    * really_clear_tiles() will set them to the clear value.
+    */
+   if (clear->surface == 0) {
+      memset(spu.ctile_status, TILE_STATUS_CLEAR, sizeof(spu.ctile_status));
+   }
+   else {
+      memset(spu.ztile_status, TILE_STATUS_CLEAR, sizeof(spu.ztile_status));
+   }
+
+#else
+
+   /*
+    * This path clears the whole framebuffer to the clear color right now.
+    */
+
+   /*
+   printf("SPU: %s num=%d w=%d h=%d\n",
+          __FUNCTION__, num_tiles, spu.fb.width_tiles, spu.fb.height_tiles);
+   */
+
+   /* init a single tile to the clear value */
+   if (clear->surface == 0) {
+      clear_c_tile(&spu.ctile);
+   }
+   else {
+      clear_z_tile(&spu.ztile);
+   }
+
+   /* walk over my tiles, writing the 'clear' tile's data */
+   {
+      const uint num_tiles = spu.fb.width_tiles * spu.fb.height_tiles;
+      uint i;
+      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
+         uint tx = i % spu.fb.width_tiles;
+         uint ty = i / spu.fb.width_tiles;
+         if (clear->surface == 0)
+            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0);
+         else
+            put_tile(tx, ty, &spu.ztile, TAG_SURFACE_CLEAR, 1);
+      }
+   }
+
+   if (spu.init.debug_flags & CELL_DEBUG_SYNC) {
+      wait_on_mask(1 << TAG_SURFACE_CLEAR);
+   }
+
+#endif /* CLEAR_OPT */
+
+   D_PRINTF(CELL_DEBUG_CMD, "CLEAR SURF done\n");
+}
+
+
+static void
+cmd_release_verts(const struct cell_command_release_verts *release)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "RELEASE VERTS %u\n", release->vertex_buf);
+   ASSERT(release->vertex_buf != ~0U);
+   release_buffer(release->vertex_buf);
+}
+
+
+/**
+ * Process a CELL_CMD_STATE_FRAGMENT_OPS command.
+ * This involves installing new fragment ops SPU code.
+ * If this function is never called, we'll use a regular C fallback function
+ * for fragment processing.
+ */
+static void
+cmd_state_fragment_ops(const struct cell_command_fragment_ops *fops)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "CMD_STATE_FRAGMENT_OPS\n");
+
+   /* Copy state info (for fallback case only - this will eventually
+    * go away when the fallback case goes away)
+    */
+   memcpy(&spu.depth_stencil_alpha, &fops->dsa, sizeof(fops->dsa));
+   memcpy(&spu.blend, &fops->blend, sizeof(fops->blend));
+   memcpy(&spu.blend_color, &fops->blend_color, sizeof(fops->blend_color));
+
+   /* Make sure the SPU knows which buffers it's expected to read when
+    * it's told to pull tiles.
+    */
+   spu.read_depth_stencil = (spu.depth_stencil_alpha.depth.enabled || spu.depth_stencil_alpha.stencil[0].enabled);
+
+   /* If we're forcing the fallback code to be used (for debug purposes),
+    * install that.  Otherwise install the incoming SPU code.
+    */
+   if ((spu.init.debug_flags & CELL_DEBUG_FRAGMENT_OP_FALLBACK) != 0) {
+      static unsigned int warned = 0;
+      if (!warned) {
+         fprintf(stderr, "Cell Warning: using fallback per-fragment code\n");
+         warned = 1;
+      }
+      /* The following two lines aren't really necessary if you
+       * know the debug flags won't change during a run, and if you
+       * know that the function pointers are initialized correctly.
+       * We set them here to allow a person to change the debug
+       * flags during a run (from inside a debugger).
+       */
+      spu.fragment_ops[CELL_FACING_FRONT] = spu_fallback_fragment_ops;
+      spu.fragment_ops[CELL_FACING_BACK] = spu_fallback_fragment_ops;
+      return;
+   }
+
+   /* Make sure the SPU code buffer is large enough to hold the incoming code.
+    * Note that we *don't* use align_malloc() and align_free(), because
+    * those utility functions are *not* available in SPU code.
+    * */
+   if (spu.fragment_ops_code_size < fops->total_code_size) {
+      if (spu.fragment_ops_code != NULL) {
+         free(spu.fragment_ops_code);
+      }
+      spu.fragment_ops_code_size = fops->total_code_size;
+      spu.fragment_ops_code = malloc(fops->total_code_size);
+      if (spu.fragment_ops_code == NULL) {
+         /* Whoops. */
+         fprintf(stderr, "CELL Warning: failed to allocate fragment ops code (%d bytes) - using fallback\n", fops->total_code_size);
+         spu.fragment_ops_code = NULL;
+         spu.fragment_ops_code_size = 0;
+         spu.fragment_ops[CELL_FACING_FRONT] = spu_fallback_fragment_ops;
+         spu.fragment_ops[CELL_FACING_BACK] = spu_fallback_fragment_ops;
+         return;
+      }
+   }
+
+   /* Copy the SPU code from the command buffer to the spu buffer */
+   memcpy(spu.fragment_ops_code, fops->code, fops->total_code_size);
+
+   /* Set the pointers for the front-facing and back-facing fragments
+    * to the specified offsets within the code.  Note that if the
+    * front-facing and back-facing code are the same, they'll have
+    * the same offset.
+    */
+   spu.fragment_ops[CELL_FACING_FRONT] = (spu_fragment_ops_func) &spu.fragment_ops_code[fops->front_code_index];
+   spu.fragment_ops[CELL_FACING_BACK] = (spu_fragment_ops_func) &spu.fragment_ops_code[fops->back_code_index];
+}
+
+static void
+cmd_state_fragment_program(const struct cell_command_fragment_program *fp)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "CMD_STATE_FRAGMENT_PROGRAM\n");
+   /* Copy SPU code from batch buffer to spu buffer */
+   memcpy(spu.fragment_program_code, fp->code,
+          SPU_MAX_FRAGMENT_PROGRAM_INSTS * 4);
+#if 01
+   /* Point function pointer at new code */
+   spu.fragment_program = (spu_fragment_program_func)spu.fragment_program_code;
+#endif
+}
+
+
+static uint
+cmd_state_fs_constants(const uint64_t *buffer, uint pos)
+{
+   const uint num_const = buffer[pos + 1];
+   const float *constants = (const float *) &buffer[pos + 2];
+   uint i;
+
+   D_PRINTF(CELL_DEBUG_CMD, "CMD_STATE_FS_CONSTANTS (%u)\n", num_const);
+
+   /* Expand each float to float[4] for SOA execution */
+   for (i = 0; i < num_const; i++) {
+      D_PRINTF(CELL_DEBUG_CMD, "  const[%u] = %f\n", i, constants[i]);
+      spu.constants[i] = spu_splats(constants[i]);
+   }
+
+   /* return new buffer pos (in 8-byte words) */
+   return pos + 2 + num_const / 2;
+}
+
+
+static void
+cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "FRAMEBUFFER: %d x %d at %p, cformat 0x%x  zformat 0x%x\n",
+             cmd->width,
+             cmd->height,
+             cmd->color_start,
+             cmd->color_format,
+             cmd->depth_format);
+
+   ASSERT_ALIGN16(cmd->color_start);
+   ASSERT_ALIGN16(cmd->depth_start);
+
+   spu.fb.color_start = cmd->color_start;
+   spu.fb.depth_start = cmd->depth_start;
+   spu.fb.color_format = cmd->color_format;
+   spu.fb.depth_format = cmd->depth_format;
+   spu.fb.width = cmd->width;
+   spu.fb.height = cmd->height;
+   spu.fb.width_tiles = (spu.fb.width + TILE_SIZE - 1) / TILE_SIZE;
+   spu.fb.height_tiles = (spu.fb.height + TILE_SIZE - 1) / TILE_SIZE;
+
+   switch (spu.fb.depth_format) {
+   case PIPE_FORMAT_Z32_UNORM:
+      spu.fb.zsize = 4;
+      spu.fb.zscale = (float) 0xffffffffu;
+      break;
+   case PIPE_FORMAT_Z24S8_UNORM:
+   case PIPE_FORMAT_S8Z24_UNORM:
+   case PIPE_FORMAT_Z24X8_UNORM:
+   case PIPE_FORMAT_X8Z24_UNORM:
+      spu.fb.zsize = 4;
+      spu.fb.zscale = (float) 0x00ffffffu;
+      break;
+   case PIPE_FORMAT_Z16_UNORM:
+      spu.fb.zsize = 2;
+      spu.fb.zscale = (float) 0xffffu;
+      break;
+   default:
+      spu.fb.zsize = 0;
+      break;
+   }
+}
+
+
+/**
+ * Tex texture mask_s/t and scale_s/t fields depend on the texture size and
+ * sampler wrap modes.
+ */
+static void
+update_tex_masks(struct spu_texture *texture,
+                 const struct pipe_sampler_state *sampler)
+{
+   uint i;
+
+   for (i = 0; i < CELL_MAX_TEXTURE_LEVELS; i++) {
+      int width = texture->level[i].width;
+      int height = texture->level[i].height;
+
+      if (sampler->wrap_s == PIPE_TEX_WRAP_REPEAT)
+         texture->level[i].mask_s = spu_splats(width - 1);
+      else
+         texture->level[i].mask_s = spu_splats(~0);
+
+      if (sampler->wrap_t == PIPE_TEX_WRAP_REPEAT)
+         texture->level[i].mask_t = spu_splats(height - 1);
+      else
+         texture->level[i].mask_t = spu_splats(~0);
+
+      if (sampler->normalized_coords) {
+         texture->level[i].scale_s = spu_splats((float) width);
+         texture->level[i].scale_t = spu_splats((float) height);
+      }
+      else {
+         texture->level[i].scale_s = spu_splats(1.0f);
+         texture->level[i].scale_t = spu_splats(1.0f);
+      }
+   }
+}
+
+
+static void
+cmd_state_sampler(const struct cell_command_sampler *sampler)
+{
+   uint unit = sampler->unit;
+
+   D_PRINTF(CELL_DEBUG_CMD, "SAMPLER [%u]\n", unit);
+
+   spu.sampler[unit] = sampler->state;
+
+   switch (spu.sampler[unit].min_img_filter) {
+   case PIPE_TEX_FILTER_LINEAR:
+      spu.min_sample_texture_2d[unit] = sample_texture_2d_bilinear;
+      break;
+   case PIPE_TEX_FILTER_ANISO:
+      /* fall-through, for now */
+   case PIPE_TEX_FILTER_NEAREST:
+      spu.min_sample_texture_2d[unit] = sample_texture_2d_nearest;
+      break;
+   default:
+      ASSERT(0);
+   }
+
+   switch (spu.sampler[sampler->unit].mag_img_filter) {
+   case PIPE_TEX_FILTER_LINEAR:
+      spu.mag_sample_texture_2d[unit] = sample_texture_2d_bilinear;
+      break;
+   case PIPE_TEX_FILTER_ANISO:
+      /* fall-through, for now */
+   case PIPE_TEX_FILTER_NEAREST:
+      spu.mag_sample_texture_2d[unit] = sample_texture_2d_nearest;
+      break;
+   default:
+      ASSERT(0);
+   }
+
+   switch (spu.sampler[sampler->unit].min_mip_filter) {
+   case PIPE_TEX_MIPFILTER_NEAREST:
+   case PIPE_TEX_MIPFILTER_LINEAR:
+      spu.sample_texture_2d[unit] = sample_texture_2d_lod;
+      break;
+   case PIPE_TEX_MIPFILTER_NONE:
+      spu.sample_texture_2d[unit] = spu.mag_sample_texture_2d[unit];
+      break;
+   default:
+      ASSERT(0);
+   }
+
+   update_tex_masks(&spu.texture[unit], &spu.sampler[unit]);
+}
+
+
+static void
+cmd_state_texture(const struct cell_command_texture *texture)
+{
+   const uint unit = texture->unit;
+   uint i;
+
+   D_PRINTF(CELL_DEBUG_CMD, "TEXTURE [%u]\n", texture->unit);
+
+   spu.texture[unit].max_level = 0;
+   spu.texture[unit].target = texture->target;
+
+   for (i = 0; i < CELL_MAX_TEXTURE_LEVELS; i++) {
+      uint width = texture->width[i];
+      uint height = texture->height[i];
+      uint depth = texture->depth[i];
+
+      D_PRINTF(CELL_DEBUG_CMD, "  LEVEL %u: at %p  size[0] %u x %u\n", i,
+             texture->start[i], texture->width[i], texture->height[i]);
+
+      spu.texture[unit].level[i].start = texture->start[i];
+      spu.texture[unit].level[i].width = width;
+      spu.texture[unit].level[i].height = height;
+      spu.texture[unit].level[i].depth = depth;
+
+      spu.texture[unit].level[i].tiles_per_row =
+         (width + TILE_SIZE - 1) / TILE_SIZE;
+
+      spu.texture[unit].level[i].bytes_per_image =
+         4 * align(width, TILE_SIZE) * align(height, TILE_SIZE) * depth;
+
+      spu.texture[unit].level[i].max_s = spu_splats((int) width - 1);
+      spu.texture[unit].level[i].max_t = spu_splats((int) height - 1);
+
+      if (texture->start[i])
+         spu.texture[unit].max_level = i;
+   }
+
+   update_tex_masks(&spu.texture[unit], &spu.sampler[unit]);
+}
+
+
+static void
+cmd_state_vertex_info(const struct vertex_info *vinfo)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "VERTEX_INFO num_attribs=%u\n", vinfo->num_attribs);
+   ASSERT(vinfo->num_attribs >= 1);
+   ASSERT(vinfo->num_attribs <= 8);
+   memcpy(&spu.vertex_info, vinfo, sizeof(*vinfo));
+}
+
+
+static void
+cmd_state_vs_array_info(const struct cell_array_info *vs_info)
+{
+   const unsigned attr = vs_info->attr;
+
+   ASSERT(attr < PIPE_MAX_ATTRIBS);
+   draw.vertex_fetch.src_ptr[attr] = vs_info->base;
+   draw.vertex_fetch.pitch[attr] = vs_info->pitch;
+   draw.vertex_fetch.size[attr] = vs_info->size;
+   draw.vertex_fetch.code_offset[attr] = vs_info->function_offset;
+   draw.vertex_fetch.dirty = 1;
+}
+
+
+static void
+cmd_state_attrib_fetch(const struct cell_attribute_fetch_code *code)
+{
+   mfc_get(attribute_fetch_code_buffer,
+           (unsigned int) code->base,  /* src */
+           code->size,
+           TAG_BATCH_BUFFER,
+           0, /* tid */
+           0  /* rid */);
+   wait_on_mask(1 << TAG_BATCH_BUFFER);
+
+   draw.vertex_fetch.code = attribute_fetch_code_buffer;
+}
+
+
+static void
+cmd_finish(void)
+{
+   D_PRINTF(CELL_DEBUG_CMD, "FINISH\n");
+   really_clear_tiles(0);
+   /* wait for all outstanding DMAs to finish */
+   mfc_write_tag_mask(~0);
+   mfc_read_tag_status_all();
+   /* send mbox message to PPU */
+   spu_write_out_mbox(CELL_CMD_FINISH);
+}
+
+
+/**
+ * Execute a batch of commands which was sent to us by the PPU.
+ * See the cell_emit_state.c code to see where the commands come from.
+ *
+ * The opcode param encodes the location of the buffer and its size.
+ */
+static void
+cmd_batch(uint opcode)
+{
+   const uint buf = (opcode >> 8) & 0xff;
+   uint size = (opcode >> 16);
+   uint64_t buffer[CELL_BUFFER_SIZE / 8] ALIGN16_ATTRIB;
+   const unsigned usize = size / sizeof(buffer[0]);
+   uint pos;
+
+   D_PRINTF(CELL_DEBUG_CMD, "BATCH buffer %u, len %u, from %p\n",
+             buf, size, spu.init.buffers[buf]);
+
+   ASSERT((opcode & CELL_CMD_OPCODE_MASK) == CELL_CMD_BATCH);
+
+   ASSERT_ALIGN16(spu.init.buffers[buf]);
+
+   size = ROUNDUP16(size);
+
+   ASSERT_ALIGN16(spu.init.buffers[buf]);
+
+   mfc_get(buffer,  /* dest */
+           (unsigned int) spu.init.buffers[buf],  /* src */
+           size,
+           TAG_BATCH_BUFFER,
+           0, /* tid */
+           0  /* rid */);
+   wait_on_mask(1 << TAG_BATCH_BUFFER);
+
+   /* Tell PPU we're done copying the buffer to local store */
+   D_PRINTF(CELL_DEBUG_CMD, "release batch buf %u\n", buf);
+   release_buffer(buf);
+
+   /*
+    * Loop over commands in the batch buffer
+    */
+   for (pos = 0; pos < usize; /* no incr */) {
+      switch (buffer[pos]) {
+      /*
+       * rendering commands
+       */
+      case CELL_CMD_CLEAR_SURFACE:
+         {
+            struct cell_command_clear_surface *clr
+               = (struct cell_command_clear_surface *) &buffer[pos];
+            cmd_clear_surface(clr);
+            pos += sizeof(*clr) / 8;
+         }
+         break;
+      case CELL_CMD_RENDER:
+         {
+            struct cell_command_render *render
+               = (struct cell_command_render *) &buffer[pos];
+            uint pos_incr;
+            cmd_render(render, &pos_incr);
+            pos += pos_incr;
+         }
+         break;
+      /*
+       * state-update commands
+       */
+      case CELL_CMD_STATE_FRAMEBUFFER:
+         {
+            struct cell_command_framebuffer *fb
+               = (struct cell_command_framebuffer *) &buffer[pos];
+            cmd_state_framebuffer(fb);
+            pos += sizeof(*fb) / 8;
+         }
+         break;
+      case CELL_CMD_STATE_FRAGMENT_OPS:
+         {
+            struct cell_command_fragment_ops *fops
+               = (struct cell_command_fragment_ops *) &buffer[pos];
+            cmd_state_fragment_ops(fops);
+            /* This is a variant-sized command */
+            pos += (sizeof(*fops) + fops->total_code_size)/ 8;
+         }
+         break;
+      case CELL_CMD_STATE_FRAGMENT_PROGRAM:
+         {
+            struct cell_command_fragment_program *fp
+               = (struct cell_command_fragment_program *) &buffer[pos];
+            cmd_state_fragment_program(fp);
+            pos += sizeof(*fp) / 8;
+         }
+         break;
+      case CELL_CMD_STATE_FS_CONSTANTS:
+         pos = cmd_state_fs_constants(buffer, pos);
+         break;
+      case CELL_CMD_STATE_RASTERIZER:
+         {
+            struct cell_command_rasterizer *rast =
+               (struct cell_command_rasterizer *) &buffer[pos];
+            spu.rasterizer = rast->rasterizer;
+            pos += sizeof(*rast) / 8;
+         }
+         break;
+      case CELL_CMD_STATE_SAMPLER:
+         {
+            struct cell_command_sampler *sampler
+               = (struct cell_command_sampler *) &buffer[pos];
+            cmd_state_sampler(sampler);
+            pos += sizeof(*sampler) / 8;
+         }
+         break;
+      case CELL_CMD_STATE_TEXTURE:
+         {
+            struct cell_command_texture *texture
+               = (struct cell_command_texture *) &buffer[pos];
+            cmd_state_texture(texture);
+            pos += sizeof(*texture) / 8;
+         }
+         break;
+      case CELL_CMD_STATE_VERTEX_INFO:
+         cmd_state_vertex_info((struct vertex_info *) &buffer[pos+1]);
+         pos += (1 + ROUNDUP8(sizeof(struct vertex_info)) / 8);
+         break;
+      case CELL_CMD_STATE_VIEWPORT:
+         (void) memcpy(& draw.viewport, &buffer[pos+1],
+                       sizeof(struct pipe_viewport_state));
+         pos += (1 + ROUNDUP8(sizeof(struct pipe_viewport_state)) / 8);
+         break;
+      case CELL_CMD_STATE_UNIFORMS:
+         draw.constants = (const float (*)[4]) (uintptr_t) buffer[pos + 1];
+         pos += 2;
+         break;
+      case CELL_CMD_STATE_VS_ARRAY_INFO:
+         cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]);
+         pos += (1 + ROUNDUP8(sizeof(struct cell_array_info)) / 8);
+         break;
+      case CELL_CMD_STATE_BIND_VS:
+#if 0
+         spu_bind_vertex_shader(&draw,
+                                (struct cell_shader_info *) &buffer[pos+1]);
+#endif
+         pos += (1 + ROUNDUP8(sizeof(struct cell_shader_info)) / 8);
+         break;
+      case CELL_CMD_STATE_ATTRIB_FETCH:
+         cmd_state_attrib_fetch((struct cell_attribute_fetch_code *)
+                                &buffer[pos+1]);
+         pos += (1 + ROUNDUP8(sizeof(struct cell_attribute_fetch_code)) / 8);
+         break;
+      /*
+       * misc commands
+       */
+      case CELL_CMD_FINISH:
+         cmd_finish();
+         pos += 1;
+         break;
+      case CELL_CMD_FENCE:
+         {
+            struct cell_command_fence *fence_cmd =
+               (struct cell_command_fence *) &buffer[pos];
+            cmd_fence(fence_cmd);
+            pos += sizeof(*fence_cmd) / 8;
+         }
+         break;
+      case CELL_CMD_RELEASE_VERTS:
+         {
+            struct cell_command_release_verts *release
+               = (struct cell_command_release_verts *) &buffer[pos];
+            cmd_release_verts(release);
+            pos += sizeof(*release) / 8;
+         }
+         break;
+      case CELL_CMD_FLUSH_BUFFER_RANGE: {
+	 struct cell_buffer_range *br = (struct cell_buffer_range *)
+	     &buffer[pos+1];
+
+	 spu_dcache_mark_dirty((unsigned) br->base, br->size);
+         pos += (1 + ROUNDUP8(sizeof(struct cell_buffer_range)) / 8);
+	 break;
+      }
+      default:
+         printf("SPU %u: bad opcode: 0x%llx\n", spu.init.id, buffer[pos]);
+         ASSERT(0);
+         break;
+      }
+   }
+
+   D_PRINTF(CELL_DEBUG_CMD, "BATCH complete\n");
+}
+
+
+#define PERF 0
+
+
+/**
+ * Main loop for SPEs: Get a command, execute it, repeat.
+ */
+void
+command_loop(void)
+{
+   int exitFlag = 0;
+   uint t0, t1;
+
+   D_PRINTF(CELL_DEBUG_CMD, "Enter command loop\n");
+
+   while (!exitFlag) {
+      unsigned opcode;
+
+      D_PRINTF(CELL_DEBUG_CMD, "Wait for cmd...\n");
+
+      if (PERF)
+         spu_write_decrementer(~0);
+
+      /* read/wait from mailbox */
+      opcode = (unsigned int) spu_read_in_mbox();
+      D_PRINTF(CELL_DEBUG_CMD, "got cmd 0x%x\n", opcode);
+
+      if (PERF)
+         t0 = spu_read_decrementer();
+
+      switch (opcode & CELL_CMD_OPCODE_MASK) {
+      case CELL_CMD_EXIT:
+         D_PRINTF(CELL_DEBUG_CMD, "EXIT\n");
+         exitFlag = 1;
+         break;
+      case CELL_CMD_VS_EXECUTE:
+#if 0
+         spu_execute_vertex_shader(&draw, &cmd.vs);
+#endif
+         break;
+      case CELL_CMD_BATCH:
+         cmd_batch(opcode);
+         break;
+      default:
+         printf("Bad opcode 0x%x!\n", opcode & CELL_CMD_OPCODE_MASK);
+      }
+
+      if (PERF) {
+         t1 = spu_read_decrementer();
+         printf("wait mbox time: %gms   batch time: %gms\n",
+                (~0u - t0) * spu.init.inv_timebase,
+                (t0 - t1) * spu.init.inv_timebase);
+      }
+   }
+
+   D_PRINTF(CELL_DEBUG_CMD, "Exit command loop\n");
+
+   if (spu.init.debug_flags & CELL_DEBUG_CACHE)
+      spu_dcache_report();
+}
+
+/* Initialize this module; we manage the fragment ops buffer here. */
+void
+spu_command_init(void)
+{
+   /* Install default/fallback fragment processing function.
+    * This will normally be overriden by a code-gen'd function
+    * unless CELL_FORCE_FRAGMENT_OPS_FALLBACK is set.
+    */
+   spu.fragment_ops[CELL_FACING_FRONT] = spu_fallback_fragment_ops;
+   spu.fragment_ops[CELL_FACING_BACK] = spu_fallback_fragment_ops;
+
+   /* Set up the basic empty buffer for code-gen'ed fragment ops */
+   spu.fragment_ops_code = NULL;
+   spu.fragment_ops_code_size = 0;
+}
+
+void
+spu_command_close(void)
+{
+   /* Deallocate the code-gen buffer for fragment ops, and reset the
+    * fragment ops functions to their initial setting (just to leave
+    * things in a good state).
+    */
+   if (spu.fragment_ops_code != NULL) {
+      free(spu.fragment_ops_code);
+   }
+   spu_command_init();
+}
diff --git a/src/mesa/drivers/dri/i965/intel_span.h b/src/gallium/drivers/cell/spu/spu_command.h
index 2d4f8589d0f..83dcdade288 100644
--- a/src/mesa/drivers/dri/i965/intel_span.h
+++ b/src/gallium/drivers/cell/spu/spu_command.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,17 +25,11 @@
  * 
  **************************************************************************/
 
-#ifndef _INTEL_SPAN_H
-#define _INTEL_SPAN_H
-
-#include "drirenderbuffer.h"
-
-extern void intelInitSpanFuncs( GLcontext *ctx );
-
-extern void intelSpanRenderFinish( GLcontext *ctx );
-extern void intelSpanRenderStart( GLcontext *ctx );
+extern void
+command_loop(void);
 
 extern void
-intelSetSpanFunctions(driRenderbuffer *rb, const GLvisual *vis);
+spu_command_init(void);
 
-#endif
+extern void
+spu_command_close(void);
diff --git a/src/gallium/drivers/cell/spu/spu_dcache.c b/src/gallium/drivers/cell/spu/spu_dcache.c
index 167404cdc54..a6d67634fd8 100644
--- a/src/gallium/drivers/cell/spu/spu_dcache.c
+++ b/src/gallium/drivers/cell/spu/spu_dcache.c
@@ -36,7 +36,9 @@
 #define CACHE_SET_TAGID(set)  (((set) & 0x03) + TAG_DCACHE0)
 #define CACHE_LOG2NNWAY       2
 #define CACHE_LOG2NSETS       6
-/*#define CACHE_STATS           1*/
+#ifdef DEBUG
+#define CACHE_STATS           1
+#endif
 #include <cache-api.h>
 
 /* Yes folks, this is ugly.
diff --git a/src/gallium/drivers/cell/spu/spu_funcs.c b/src/gallium/drivers/cell/spu/spu_funcs.c
new file mode 100644
index 00000000000..ff3d609d258
--- /dev/null
+++ b/src/gallium/drivers/cell/spu/spu_funcs.c
@@ -0,0 +1,173 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+/**
+ * SPU functions accessed by shaders.
+ *
+ * Authors: Brian Paul
+ */
+
+
+#include <string.h>
+#include <libmisc.h>
+#include <math.h>
+#include <cos14_v.h>
+#include <sin14_v.h>
+#include <simdmath/exp2f4.h>
+#include <simdmath/log2f4.h>
+#include <simdmath/powf4.h>
+
+#include "cell/common.h"
+#include "spu_main.h"
+#include "spu_funcs.h"
+#include "spu_texture.h"
+
+
+/** For "return"-ing four vectors */
+struct vec_4x4
+{
+   vector float v[4];
+};
+
+
+static vector float
+spu_cos(vector float x)
+{
+   return _cos14_v(x);
+}
+
+static vector float
+spu_sin(vector float x)
+{
+   return _sin14_v(x);
+}
+
+static vector float
+spu_pow(vector float x, vector float y)
+{
+   return _powf4(x, y);
+}
+
+static vector float
+spu_exp2(vector float x)
+{
+   return _exp2f4(x);
+}
+
+static vector float
+spu_log2(vector float x)
+{
+   return _log2f4(x);
+}
+
+
+static struct vec_4x4
+spu_tex_2d(vector float s, vector float t, vector float r, vector float q,
+           unsigned unit)
+{
+   struct vec_4x4 colors;
+   (void) r;
+   (void) q;
+   spu.sample_texture_2d[unit](s, t, unit, 0, 0, colors.v);
+   return colors;
+}
+
+static struct vec_4x4
+spu_tex_3d(vector float s, vector float t, vector float r, vector float q,
+           unsigned unit)
+{
+   struct vec_4x4 colors;
+   (void) r;
+   (void) q;
+   spu.sample_texture_2d[unit](s, t, unit, 0, 0, colors.v);
+   return colors;
+}
+
+static struct vec_4x4
+spu_tex_cube(vector float s, vector float t, vector float r, vector float q,
+           unsigned unit)
+{
+   struct vec_4x4 colors;
+   (void) q;
+   sample_texture_cube(s, t, r, unit, colors.v);
+   return colors;
+}
+
+
+/**
+ * Add named function to list of "exported" functions that will be
+ * made available to the PPU-hosted code generator.
+ */
+static void
+export_func(struct cell_spu_function_info *spu_functions,
+            const char *name, void *addr)
+{
+   uint n = spu_functions->num;
+   ASSERT(strlen(name) < 16);
+   strcpy(spu_functions->names[n], name);
+   spu_functions->addrs[n] = (uint) addr;
+   spu_functions->num++;
+   ASSERT(spu_functions->num <= 16);
+}
+
+
+/**
+ * Return info about the SPU's function to the PPU / main memory.
+ * The PPU needs to know the address of some SPU-side functions so
+ * that we can generate shader code with function calls.
+ */
+void
+return_function_info(void)
+{
+   struct cell_spu_function_info funcs ALIGN16_ATTRIB;
+   int tag = TAG_MISC;
+
+   ASSERT(sizeof(funcs) == 256); /* must be multiple of 16 bytes */
+
+   funcs.num = 0;
+   export_func(&funcs, "spu_cos", &spu_cos);
+   export_func(&funcs, "spu_sin", &spu_sin);
+   export_func(&funcs, "spu_pow", &spu_pow);
+   export_func(&funcs, "spu_exp2", &spu_exp2);
+   export_func(&funcs, "spu_log2", &spu_log2);
+   export_func(&funcs, "spu_tex_2d", &spu_tex_2d);
+   export_func(&funcs, "spu_tex_3d", &spu_tex_3d);
+   export_func(&funcs, "spu_tex_cube", &spu_tex_cube);
+
+   /* Send the function info back to the PPU / main memory */
+   mfc_put((void *) &funcs,  /* src in local store */
+           (unsigned int) spu.init.spu_functions, /* dst in main memory */
+           sizeof(funcs),  /* bytes */
+           tag,
+           0, /* tid */
+           0  /* rid */);
+   wait_on_mask(1 << tag);
+}
+
+
+
diff --git a/src/mesa/drivers/dri/i965/intel_tex.h b/src/gallium/drivers/cell/spu/spu_funcs.h
index e389d521461..3adb6ae99f9 100644
--- a/src/mesa/drivers/dri/i965/intel_tex.h
+++ b/src/gallium/drivers/cell/spu/spu_funcs.h
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,18 +25,11 @@
  * 
  **************************************************************************/
 
-#ifndef INTELTEX_INC
-#define INTELTEX_INC
-
-#include "mtypes.h"
-#include "intel_context.h"
-
-
-void intelInitTextureFuncs( struct dd_function_table *functions );
-
-
-GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
-				   struct gl_texture_object *tObj );
+#ifndef SPU_FUNCS_H
+#define SPU_FUNCS_H
 
+extern void
+return_function_info(void);
 
 #endif
+
diff --git a/src/gallium/drivers/cell/spu/spu_main.c b/src/gallium/drivers/cell/spu/spu_main.c
index 78260c4259c..97c86d194da 100644
--- a/src/gallium/drivers/cell/spu/spu_main.c
+++ b/src/gallium/drivers/cell/spu/spu_main.c
@@ -32,16 +32,15 @@
 #include <stdio.h>
 #include <libmisc.h>
 
+#include "pipe/p_defines.h"
+
+#include "spu_funcs.h"
+#include "spu_command.h"
 #include "spu_main.h"
-#include "spu_render.h"
 #include "spu_per_fragment_op.h"
 #include "spu_texture.h"
-#include "spu_tile.h"
 //#include "spu_test.h"
-#include "spu_vertex_shader.h"
-#include "spu_dcache.h"
 #include "cell/common.h"
-#include "pipe/p_defines.h"
 
 
 /*
@@ -50,600 +49,8 @@ helpful headers:
 /opt/cell/sdk/usr/include/libmisc.h
 */
 
-boolean Debug = FALSE;
-
 struct spu_global spu;
 
-struct spu_vs_context draw;
-
-
-/**
- * Buffers containing dynamically generated SPU code:
- */
-static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS]
-    ALIGN16_ATTRIB;
-
-
-
-/**
- * Tell the PPU that this SPU has finished copying a buffer to
- * local store and that it may be reused by the PPU.
- * This is done by writting a 16-byte batch-buffer-status block back into
- * main memory (in cell_context->buffer_status[]).
- */
-static void
-release_buffer(uint buffer)
-{
-   /* Evidently, using less than a 16-byte status doesn't work reliably */
-   static const uint status[4] ALIGN16_ATTRIB
-      = {CELL_BUFFER_STATUS_FREE, 0, 0, 0};
-
-   const uint index = 4 * (spu.init.id * CELL_NUM_BUFFERS + buffer);
-   uint *dst = spu.init.buffer_status + index;
-
-   ASSERT(buffer < CELL_NUM_BUFFERS);
-
-   mfc_put((void *) &status,    /* src in local memory */
-           (unsigned int) dst,  /* dst in main memory */
-           sizeof(status),      /* size */
-           TAG_MISC,            /* tag is unimportant */
-           0, /* tid */
-           0  /* rid */);
-}
-
-
-/**
- * For tiles whose status is TILE_STATUS_CLEAR, write solid-filled
- * tiles back to the main framebuffer.
- */
-static void
-really_clear_tiles(uint surfaceIndex)
-{
-   const uint num_tiles = spu.fb.width_tiles * spu.fb.height_tiles;
-   uint i;
-
-   if (surfaceIndex == 0) {
-      clear_c_tile(&spu.ctile);
-
-      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
-         uint tx = i % spu.fb.width_tiles;
-         uint ty = i / spu.fb.width_tiles;
-         if (spu.ctile_status[ty][tx] == TILE_STATUS_CLEAR) {
-            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0);
-         }
-      }
-   }
-   else {
-      clear_z_tile(&spu.ztile);
-
-      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
-         uint tx = i % spu.fb.width_tiles;
-         uint ty = i / spu.fb.width_tiles;
-         if (spu.ztile_status[ty][tx] == TILE_STATUS_CLEAR)
-            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 1);
-      }
-   }
-
-#if 0
-   wait_on_mask(1 << TAG_SURFACE_CLEAR);
-#endif
-}
-
-
-static void
-cmd_clear_surface(const struct cell_command_clear_surface *clear)
-{
-   if (Debug)
-      printf("SPU %u: CLEAR SURF %u to 0x%08x\n", spu.init.id,
-             clear->surface, clear->value);
-
-   if (clear->surface == 0) {
-      spu.fb.color_clear_value = clear->value;
-      if (spu.init.debug_flags & CELL_DEBUG_CHECKER) {
-         uint x = (spu.init.id << 4) | (spu.init.id << 12) |
-            (spu.init.id << 20) | (spu.init.id << 28);
-         spu.fb.color_clear_value ^= x;
-      }
-   }
-   else {
-      spu.fb.depth_clear_value = clear->value;
-   }
-
-#define CLEAR_OPT 1
-#if CLEAR_OPT
-
-   /* Simply set all tiles' status to CLEAR.
-    * When we actually begin rendering into a tile, we'll initialize it to
-    * the clear value.  If any tiles go untouched during the frame,
-    * really_clear_tiles() will set them to the clear value.
-    */
-   if (clear->surface == 0) {
-      memset(spu.ctile_status, TILE_STATUS_CLEAR, sizeof(spu.ctile_status));
-   }
-   else {
-      memset(spu.ztile_status, TILE_STATUS_CLEAR, sizeof(spu.ztile_status));
-   }
-
-#else
-
-   /*
-    * This path clears the whole framebuffer to the clear color right now.
-    */
-
-   /*
-   printf("SPU: %s num=%d w=%d h=%d\n",
-          __FUNCTION__, num_tiles, spu.fb.width_tiles, spu.fb.height_tiles);
-   */
-
-   /* init a single tile to the clear value */
-   if (clear->surface == 0) {
-      clear_c_tile(&spu.ctile);
-   }
-   else {
-      clear_z_tile(&spu.ztile);
-   }
-
-   /* walk over my tiles, writing the 'clear' tile's data */
-   {
-      const uint num_tiles = spu.fb.width_tiles * spu.fb.height_tiles;
-      uint i;
-      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
-         uint tx = i % spu.fb.width_tiles;
-         uint ty = i / spu.fb.width_tiles;
-         if (clear->surface == 0)
-            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0);
-         else
-            put_tile(tx, ty, &spu.ztile, TAG_SURFACE_CLEAR, 1);
-      }
-   }
-
-   if (spu.init.debug_flags & CELL_DEBUG_SYNC) {
-      wait_on_mask(1 << TAG_SURFACE_CLEAR);
-   }
-
-#endif /* CLEAR_OPT */
-
-   if (Debug)
-      printf("SPU %u: CLEAR SURF done\n", spu.init.id);
-}
-
-
-static void
-cmd_release_verts(const struct cell_command_release_verts *release)
-{
-   if (Debug)
-      printf("SPU %u: RELEASE VERTS %u\n",
-             spu.init.id, release->vertex_buf);
-   ASSERT(release->vertex_buf != ~0U);
-   release_buffer(release->vertex_buf);
-}
-
-
-/**
- * Process a CELL_CMD_STATE_FRAGMENT_OPS command.
- * This involves installing new fragment ops SPU code.
- * If this function is never called, we'll use a regular C fallback function
- * for fragment processing.
- */
-static void
-cmd_state_fragment_ops(const struct cell_command_fragment_ops *fops)
-{
-   if (Debug)
-      printf("SPU %u: CMD_STATE_FRAGMENT_OPS\n", spu.init.id);
-   /* Copy SPU code from batch buffer to spu buffer */
-   memcpy(spu.fragment_ops_code, fops->code, SPU_MAX_FRAGMENT_OPS_INSTS * 4);
-   /* Copy state info (for fallback case only) */
-   memcpy(&spu.depth_stencil_alpha, &fops->dsa, sizeof(fops->dsa));
-   memcpy(&spu.blend, &fops->blend, sizeof(fops->blend));
-
-   /* Point function pointer at new code */
-   spu.fragment_ops = (spu_fragment_ops_func) spu.fragment_ops_code;
-
-   spu.read_depth = spu.depth_stencil_alpha.depth.enabled;
-   spu.read_stencil = spu.depth_stencil_alpha.stencil[0].enabled;
-}
-
-
-static void
-cmd_state_fragment_program(const struct cell_command_fragment_program *fp)
-{
-   if (Debug)
-      printf("SPU %u: CMD_STATE_FRAGMENT_PROGRAM\n", spu.init.id);
-   /* Copy SPU code from batch buffer to spu buffer */
-   memcpy(spu.fragment_program_code, fp->code,
-          SPU_MAX_FRAGMENT_PROGRAM_INSTS * 4);
-#if 01
-   /* Point function pointer at new code */
-   spu.fragment_program = (spu_fragment_program_func)spu.fragment_program_code;
-#endif
-}
-
-
-static void
-cmd_state_framebuffer(const struct cell_command_framebuffer *cmd)
-{
-   if (Debug)
-      printf("SPU %u: FRAMEBUFFER: %d x %d at %p, cformat 0x%x  zformat 0x%x\n",
-             spu.init.id,
-             cmd->width,
-             cmd->height,
-             cmd->color_start,
-             cmd->color_format,
-             cmd->depth_format);
-
-   ASSERT_ALIGN16(cmd->color_start);
-   ASSERT_ALIGN16(cmd->depth_start);
-
-   spu.fb.color_start = cmd->color_start;
-   spu.fb.depth_start = cmd->depth_start;
-   spu.fb.color_format = cmd->color_format;
-   spu.fb.depth_format = cmd->depth_format;
-   spu.fb.width = cmd->width;
-   spu.fb.height = cmd->height;
-   spu.fb.width_tiles = (spu.fb.width + TILE_SIZE - 1) / TILE_SIZE;
-   spu.fb.height_tiles = (spu.fb.height + TILE_SIZE - 1) / TILE_SIZE;
-
-   switch (spu.fb.depth_format) {
-   case PIPE_FORMAT_Z32_UNORM:
-      spu.fb.zsize = 4;
-      spu.fb.zscale = (float) 0xffffffffu;
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM:
-   case PIPE_FORMAT_S8Z24_UNORM:
-   case PIPE_FORMAT_Z24X8_UNORM:
-   case PIPE_FORMAT_X8Z24_UNORM:
-      spu.fb.zsize = 4;
-      spu.fb.zscale = (float) 0x00ffffffu;
-      break;
-   case PIPE_FORMAT_Z16_UNORM:
-      spu.fb.zsize = 2;
-      spu.fb.zscale = (float) 0xffffu;
-      break;
-   default:
-      spu.fb.zsize = 0;
-      break;
-   }
-}
-
-
-static void
-cmd_state_sampler(const struct cell_command_sampler *sampler)
-{
-   if (Debug)
-      printf("SPU %u: SAMPLER [%u]\n",
-             spu.init.id, sampler->unit);
-
-   spu.sampler[sampler->unit] = sampler->state;
-   if (spu.sampler[sampler->unit].min_img_filter == PIPE_TEX_FILTER_LINEAR)
-      spu.sample_texture[sampler->unit] = sample_texture_bilinear;
-   else
-      spu.sample_texture[sampler->unit] = sample_texture_nearest;
-}
-
-
-static void
-cmd_state_texture(const struct cell_command_texture *texture)
-{
-   const uint unit = texture->unit;
-   const uint width = texture->width;
-   const uint height = texture->height;
-
-   if (Debug) {
-      printf("SPU %u: TEXTURE [%u] at %p  size %u x %u\n", spu.init.id,
-             texture->unit, texture->start,
-             texture->width, texture->height);
-   }
-
-   spu.texture[unit].start = texture->start;
-   spu.texture[unit].width = width;
-   spu.texture[unit].height = height;
-
-   spu.texture[unit].tiles_per_row = width / TILE_SIZE;
-
-   spu.texture[unit].tex_size = (vector float) { width, height, 0.0, 0.0};
-   spu.texture[unit].tex_size_mask = (vector unsigned int)
-         { width - 1, height - 1, 0, 0 };
-   spu.texture[unit].tex_size_x_mask = spu_splats(width - 1);
-   spu.texture[unit].tex_size_y_mask = spu_splats(height - 1);
-}
-
-
-static void
-cmd_state_vertex_info(const struct vertex_info *vinfo)
-{
-   if (Debug) {
-      printf("SPU %u: VERTEX_INFO num_attribs=%u\n", spu.init.id,
-             vinfo->num_attribs);
-   }
-   ASSERT(vinfo->num_attribs >= 1);
-   ASSERT(vinfo->num_attribs <= 8);
-   memcpy(&spu.vertex_info, vinfo, sizeof(*vinfo));
-}
-
-
-static void
-cmd_state_vs_array_info(const struct cell_array_info *vs_info)
-{
-   const unsigned attr = vs_info->attr;
-
-   ASSERT(attr < PIPE_MAX_ATTRIBS);
-   draw.vertex_fetch.src_ptr[attr] = vs_info->base;
-   draw.vertex_fetch.pitch[attr] = vs_info->pitch;
-   draw.vertex_fetch.size[attr] = vs_info->size;
-   draw.vertex_fetch.code_offset[attr] = vs_info->function_offset;
-   draw.vertex_fetch.dirty = 1;
-}
-
-
-static void
-cmd_state_attrib_fetch(const struct cell_attribute_fetch_code *code)
-{
-   mfc_get(attribute_fetch_code_buffer,
-           (unsigned int) code->base,  /* src */
-           code->size,
-           TAG_BATCH_BUFFER,
-           0, /* tid */
-           0  /* rid */);
-   wait_on_mask(1 << TAG_BATCH_BUFFER);
-
-   draw.vertex_fetch.code = attribute_fetch_code_buffer;
-}
-
-
-static void
-cmd_finish(void)
-{
-   if (Debug)
-      printf("SPU %u: FINISH\n", spu.init.id);
-   really_clear_tiles(0);
-   /* wait for all outstanding DMAs to finish */
-   mfc_write_tag_mask(~0);
-   mfc_read_tag_status_all();
-   /* send mbox message to PPU */
-   spu_write_out_mbox(CELL_CMD_FINISH);
-}
-
-
-/**
- * Execute a batch of commands which was sent to us by the PPU.
- * See the cell_emit_state.c code to see where the commands come from.
- *
- * The opcode param encodes the location of the buffer and its size.
- */
-static void
-cmd_batch(uint opcode)
-{
-   const uint buf = (opcode >> 8) & 0xff;
-   uint size = (opcode >> 16);
-   uint64_t buffer[CELL_BUFFER_SIZE / 8] ALIGN16_ATTRIB;
-   const unsigned usize = size / sizeof(buffer[0]);
-   uint pos;
-
-   if (Debug)
-      printf("SPU %u: BATCH buffer %u, len %u, from %p\n",
-             spu.init.id, buf, size, spu.init.buffers[buf]);
-
-   ASSERT((opcode & CELL_CMD_OPCODE_MASK) == CELL_CMD_BATCH);
-
-   ASSERT_ALIGN16(spu.init.buffers[buf]);
-
-   size = ROUNDUP16(size);
-
-   ASSERT_ALIGN16(spu.init.buffers[buf]);
-
-   mfc_get(buffer,  /* dest */
-           (unsigned int) spu.init.buffers[buf],  /* src */
-           size,
-           TAG_BATCH_BUFFER,
-           0, /* tid */
-           0  /* rid */);
-   wait_on_mask(1 << TAG_BATCH_BUFFER);
-
-   /* Tell PPU we're done copying the buffer to local store */
-   if (Debug)
-      printf("SPU %u: release batch buf %u\n", spu.init.id, buf);
-   release_buffer(buf);
-
-   /*
-    * Loop over commands in the batch buffer
-    */
-   for (pos = 0; pos < usize; /* no incr */) {
-      switch (buffer[pos]) {
-      /*
-       * rendering commands
-       */
-      case CELL_CMD_CLEAR_SURFACE:
-         {
-            struct cell_command_clear_surface *clr
-               = (struct cell_command_clear_surface *) &buffer[pos];
-            cmd_clear_surface(clr);
-            pos += sizeof(*clr) / 8;
-         }
-         break;
-      case CELL_CMD_RENDER:
-         {
-            struct cell_command_render *render
-               = (struct cell_command_render *) &buffer[pos];
-            uint pos_incr;
-            cmd_render(render, &pos_incr);
-            pos += pos_incr;
-         }
-         break;
-      /*
-       * state-update commands
-       */
-      case CELL_CMD_STATE_FRAMEBUFFER:
-         {
-            struct cell_command_framebuffer *fb
-               = (struct cell_command_framebuffer *) &buffer[pos];
-            cmd_state_framebuffer(fb);
-            pos += sizeof(*fb) / 8;
-         }
-         break;
-      case CELL_CMD_STATE_FRAGMENT_OPS:
-         {
-            struct cell_command_fragment_ops *fops
-               = (struct cell_command_fragment_ops *) &buffer[pos];
-            cmd_state_fragment_ops(fops);
-            pos += sizeof(*fops) / 8;
-         }
-         break;
-      case CELL_CMD_STATE_FRAGMENT_PROGRAM:
-         {
-            struct cell_command_fragment_program *fp
-               = (struct cell_command_fragment_program *) &buffer[pos];
-            cmd_state_fragment_program(fp);
-            pos += sizeof(*fp) / 8;
-         }
-         break;
-      case CELL_CMD_STATE_SAMPLER:
-         {
-            struct cell_command_sampler *sampler
-               = (struct cell_command_sampler *) &buffer[pos];
-            cmd_state_sampler(sampler);
-            pos += sizeof(*sampler) / 8;
-         }
-         break;
-      case CELL_CMD_STATE_TEXTURE:
-         {
-            struct cell_command_texture *texture
-               = (struct cell_command_texture *) &buffer[pos];
-            cmd_state_texture(texture);
-            pos += sizeof(*texture) / 8;
-         }
-         break;
-      case CELL_CMD_STATE_VERTEX_INFO:
-         cmd_state_vertex_info((struct vertex_info *) &buffer[pos+1]);
-         pos += (1 + ROUNDUP8(sizeof(struct vertex_info)) / 8);
-         break;
-      case CELL_CMD_STATE_VIEWPORT:
-         (void) memcpy(& draw.viewport, &buffer[pos+1],
-                       sizeof(struct pipe_viewport_state));
-         pos += (1 + ROUNDUP8(sizeof(struct pipe_viewport_state)) / 8);
-         break;
-      case CELL_CMD_STATE_UNIFORMS:
-         draw.constants = (const float (*)[4]) (uintptr_t) buffer[pos + 1];
-         pos += 2;
-         break;
-      case CELL_CMD_STATE_VS_ARRAY_INFO:
-         cmd_state_vs_array_info((struct cell_array_info *) &buffer[pos+1]);
-         pos += (1 + ROUNDUP8(sizeof(struct cell_array_info)) / 8);
-         break;
-      case CELL_CMD_STATE_BIND_VS:
-#if 0
-         spu_bind_vertex_shader(&draw,
-                                (struct cell_shader_info *) &buffer[pos+1]);
-#endif
-         pos += (1 + ROUNDUP8(sizeof(struct cell_shader_info)) / 8);
-         break;
-      case CELL_CMD_STATE_ATTRIB_FETCH:
-         cmd_state_attrib_fetch((struct cell_attribute_fetch_code *)
-                                &buffer[pos+1]);
-         pos += (1 + ROUNDUP8(sizeof(struct cell_attribute_fetch_code)) / 8);
-         break;
-      /*
-       * misc commands
-       */
-      case CELL_CMD_FINISH:
-         cmd_finish();
-         pos += 1;
-         break;
-      case CELL_CMD_RELEASE_VERTS:
-         {
-            struct cell_command_release_verts *release
-               = (struct cell_command_release_verts *) &buffer[pos];
-            cmd_release_verts(release);
-            pos += sizeof(*release) / 8;
-         }
-         break;
-      case CELL_CMD_FLUSH_BUFFER_RANGE: {
-	 struct cell_buffer_range *br = (struct cell_buffer_range *)
-	     &buffer[pos+1];
-
-	 spu_dcache_mark_dirty((unsigned) br->base, br->size);
-         pos += (1 + ROUNDUP8(sizeof(struct cell_buffer_range)) / 8);
-	 break;
-      }
-      default:
-         printf("SPU %u: bad opcode: 0x%llx\n", spu.init.id, buffer[pos]);
-         ASSERT(0);
-         break;
-      }
-   }
-
-   if (Debug)
-      printf("SPU %u: BATCH complete\n", spu.init.id);
-}
-
-
-/**
- * Temporary/simple main loop for SPEs: Get a command, execute it, repeat.
- */
-static void
-main_loop(void)
-{
-   struct cell_command cmd;
-   int exitFlag = 0;
-
-   if (Debug)
-      printf("SPU %u: Enter main loop\n", spu.init.id);
-
-   ASSERT((sizeof(struct cell_command) & 0xf) == 0);
-   ASSERT_ALIGN16(&cmd);
-
-   while (!exitFlag) {
-      unsigned opcode;
-      int tag = 0;
-
-      if (Debug)
-         printf("SPU %u: Wait for cmd...\n", spu.init.id);
-
-      /* read/wait from mailbox */
-      opcode = (unsigned int) spu_read_in_mbox();
-
-      if (Debug)
-         printf("SPU %u: got cmd 0x%x\n", spu.init.id, opcode);
-
-      /* command payload */
-      mfc_get(&cmd,  /* dest */
-              (unsigned int) spu.init.cmd, /* src */
-              sizeof(struct cell_command), /* bytes */
-              tag,
-              0, /* tid */
-              0  /* rid */);
-      wait_on_mask( 1 << tag );
-
-      /*
-       * NOTE: most commands should be contained in a batch buffer
-       */
-
-      switch (opcode & CELL_CMD_OPCODE_MASK) {
-      case CELL_CMD_EXIT:
-         if (Debug)
-            printf("SPU %u: EXIT\n", spu.init.id);
-         exitFlag = 1;
-         break;
-      case CELL_CMD_VS_EXECUTE:
-#if 0
-         spu_execute_vertex_shader(&draw, &cmd.vs);
-#endif
-         break;
-      case CELL_CMD_BATCH:
-         cmd_batch(opcode);
-         break;
-      default:
-         printf("Bad opcode!\n");
-      }
-
-   }
-
-   if (Debug)
-      printf("SPU %u: Exit main loop\n", spu.init.id);
-
-   spu_dcache_report();
-}
-
-
 
 static void
 one_time_init(void)
@@ -651,15 +58,8 @@ one_time_init(void)
    memset(spu.ctile_status, TILE_STATUS_DEFINED, sizeof(spu.ctile_status));
    memset(spu.ztile_status, TILE_STATUS_DEFINED, sizeof(spu.ztile_status));
    invalidate_tex_cache();
-
-   /* Install default/fallback fragment processing function.
-    * This will normally be overriden by a code-gen'd function.
-    */
-   spu.fragment_ops = spu_fallback_fragment_ops;
 }
 
-
-
 /* In some versions of the SDK the SPE main takes 'unsigned long' as a
  * parameter.  In others it takes 'unsigned long long'.  Use a define to
  * select between the two.
@@ -682,12 +82,16 @@ main(main_param_t speid, main_param_t argp)
 
    ASSERT(sizeof(tile_t) == TILE_SIZE * TILE_SIZE * 4);
    ASSERT(sizeof(struct cell_command_render) % 8 == 0);
+   ASSERT(sizeof(struct cell_command_fragment_ops) % 8 == 0);
+   ASSERT(((unsigned long) &spu.fragment_program_code) % 8 == 0);
 
    one_time_init();
+   spu_command_init();
 
-   if (Debug)
-      printf("SPU: main() speid=%lu\n", (unsigned long) speid);
+   D_PRINTF(CELL_DEBUG_CMD, "main() speid=%lu\n", (unsigned long) speid);
+   D_PRINTF(CELL_DEBUG_FRAGMENT_OP_FALLBACK, "using fragment op fallback\n");
 
+   /* get initialization data */
    mfc_get(&spu.init,  /* dest */
            (unsigned int) argp, /* src */
            sizeof(struct cell_init_info), /* bytes */
@@ -696,12 +100,18 @@ main(main_param_t speid, main_param_t argp)
            0  /* rid */);
    wait_on_mask( 1 << tag );
 
+   if (spu.init.id == 0) {
+      return_function_info();
+   }
+
 #if 0
    if (spu.init.id==0)
-      spu_test_misc();
+      spu_test_misc(spu.init.id);
 #endif
 
-   main_loop();
+   command_loop();
+
+   spu_command_close();
 
    return 0;
 }
diff --git a/src/gallium/drivers/cell/spu/spu_main.h b/src/gallium/drivers/cell/spu/spu_main.h
index 2c7b6258402..33767e7c51d 100644
--- a/src/gallium/drivers/cell/spu/spu_main.h
+++ b/src/gallium/drivers/cell/spu/spu_main.h
@@ -36,9 +36,18 @@
 #include "pipe/p_state.h"
 
 
-
-#define MAX_WIDTH 1024
-#define MAX_HEIGHT 1024
+#if DEBUG
+/* These debug macros use the unusual construction ", ##__VA_ARGS__"
+ * which expands to the expected comma + args if variadic arguments
+ * are supplied, but swallows the comma if there are no variadic
+ * arguments (which avoids syntax errors that would otherwise occur).
+ */
+#define D_PRINTF(flag, format,...) \
+   if (spu.init.debug_flags & (flag)) \
+      printf("SPU %u: " format, spu.init.id, ##__VA_ARGS__)
+#else
+#define D_PRINTF(...)
+#endif
 
 
 /**
@@ -61,8 +70,11 @@ typedef union {
 
 
 /** Function for sampling textures */
-typedef vector float (*spu_sample_texture_func)(uint unit,
-                                                vector float texcoord);
+typedef void (*spu_sample_texture_2d_func)(vector float s,
+                                           vector float t,
+                                           uint unit, uint level, uint face,
+                                           vector float colors[4]);
+
 
 /** Function for performing per-fragment ops */
 typedef void (*spu_fragment_ops_func)(uint x, uint y,
@@ -76,9 +88,9 @@ typedef void (*spu_fragment_ops_func)(uint x, uint y,
                                       vector unsigned int mask);
 
 /** Function for running fragment program */
-typedef void (*spu_fragment_program_func)(vector float *inputs,
-                                          vector float *outputs,
-                                          vector float *constants);
+typedef vector unsigned int (*spu_fragment_program_func)(vector float *inputs,
+                                                         vector float *outputs,
+                                                         vector float *constants);
 
 
 struct spu_framebuffer
@@ -98,15 +110,27 @@ struct spu_framebuffer
 } ALIGN16_ATTRIB;
 
 
-struct spu_texture
+/** per-texture level info */
+struct spu_texture_level
 {
    void *start;
-   ushort width, height;
+   ushort width, height, depth;
    ushort tiles_per_row;
-   vector float tex_size;
-   vector unsigned int tex_size_mask; /**< == int(size - 1) */
-   vector unsigned int tex_size_x_mask; /**< == int(size - 1) */
-   vector unsigned int tex_size_y_mask; /**< == int(size - 1) */
+   uint bytes_per_image;
+   /** texcoord scale factors */
+   vector float scale_s, scale_t, scale_r;
+   /** texcoord masks (if REPEAT then size-1, else ~0) */
+   vector signed int mask_s, mask_t, mask_r;
+   /** texcoord clamp limits */
+   vector signed int max_s, max_t, max_r;
+} ALIGN16_ATTRIB;
+
+
+struct spu_texture
+{
+   struct spu_texture_level level[CELL_MAX_TEXTURE_LEVELS];
+   uint max_level;
+   uint target;  /**< PIPE_TEXTURE_x */
 } ALIGN16_ATTRIB;
 
 
@@ -124,7 +148,9 @@ struct spu_global
    struct spu_framebuffer fb;
    struct pipe_depth_stencil_alpha_state depth_stencil_alpha;
    struct pipe_blend_state blend;
+   struct pipe_blend_color blend_color;
    struct pipe_sampler_state sampler[PIPE_MAX_SAMPLERS];
+   struct pipe_rasterizer_state rasterizer;
    struct spu_texture texture[PIPE_MAX_SAMPLERS];
    struct vertex_info vertex_info;
 
@@ -133,39 +159,38 @@ struct spu_global
    tile_t ztile ALIGN16_ATTRIB;
 
    /** Read depth/stencil tiles? */
-   boolean read_depth;
-   boolean read_stencil;
+   boolean read_depth_stencil;
 
    /** Current tiles' status */
    ubyte cur_ctile_status, cur_ztile_status;
 
    /** Status of all tiles in framebuffer */
-   ubyte ctile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
-   ubyte ztile_status[MAX_HEIGHT/TILE_SIZE][MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
+   ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
+   ubyte ztile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB;
 
-   /** Current fragment ops machine code */
-   uint fragment_ops_code[SPU_MAX_FRAGMENT_OPS_INSTS];
-   /** Current fragment ops function */
-   spu_fragment_ops_func fragment_ops;
+   /** Current fragment ops machine code, at 8-byte boundary */
+   uint *fragment_ops_code;
+   uint fragment_ops_code_size;
+   /** Current fragment ops functions, 0 = frontfacing, 1 = backfacing */
+   spu_fragment_ops_func fragment_ops[2];
 
-   /** Current fragment program machine code */
-   uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS];
+   /** Current fragment program machine code, at 8-byte boundary */
+   uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS] ALIGN8_ATTRIB;
    /** Current fragment ops function */
    spu_fragment_program_func fragment_program;
 
    /** Current texture sampler function */
-   spu_sample_texture_func sample_texture[CELL_MAX_SAMPLERS];
+   spu_sample_texture_2d_func sample_texture_2d[CELL_MAX_SAMPLERS];
+   spu_sample_texture_2d_func min_sample_texture_2d[CELL_MAX_SAMPLERS];
+   spu_sample_texture_2d_func mag_sample_texture_2d[CELL_MAX_SAMPLERS];
 
-   /** Fragment program constants (XXX preliminary/used) */
-#define MAX_CONSTANTS 32
-   vector float constants[MAX_CONSTANTS];
+   /** Fragment program constants */
+   vector float constants[4 * CELL_MAX_CONSTANTS];
 
 } ALIGN16_ATTRIB;
 
 
 extern struct spu_global spu;
-extern boolean Debug;
-
 
 
 
@@ -184,7 +209,7 @@ extern boolean Debug;
 #define TAG_DCACHE1           21
 #define TAG_DCACHE2           22
 #define TAG_DCACHE3           23
-
+#define TAG_FENCE             24
 
 
 static INLINE void
diff --git a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
index 03dd547845b..683664e8a4e 100644
--- a/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
+++ b/src/gallium/drivers/cell/spu/spu_per_fragment_op.c
@@ -40,6 +40,24 @@
 #define LINEAR_QUAD_LAYOUT 1
 
 
+static INLINE vector float
+spu_min(vector float a, vector float b)
+{
+   vector unsigned int m;
+   m = spu_cmpgt(a, b);    /* m = a > b ? ~0 : 0 */
+   return spu_sel(a, b, m);
+}
+
+
+static INLINE vector float
+spu_max(vector float a, vector float b)
+{
+   vector unsigned int m;
+   m = spu_cmpgt(a, b);    /* m = a > b ? ~0 : 0 */
+   return spu_sel(b, a, m);
+}
+
+
 /**
  * Called by rasterizer for each quad after the shader has run.  Do
  * all the per-fragment operations including alpha test, z test,
@@ -60,9 +78,12 @@ spu_fallback_fragment_ops(uint x, uint y,
                           vector unsigned int mask)
 {
    vector float frag_aos[4];
-   unsigned int c0, c1, c2, c3;
+   unsigned int fbc0, fbc1, fbc2, fbc3 ; /* framebuffer/tile colors */
+   unsigned int fragc0, fragc1, fragc2, fragc3;  /* fragment colors */
 
-   /* do alpha test */
+   /*
+    * Do alpha test
+    */
    if (spu.depth_stencil_alpha.alpha.enabled) {
       vector float ref = spu_splats(spu.depth_stencil_alpha.alpha.ref);
       vector unsigned int amask;
@@ -102,7 +123,10 @@ spu_fallback_fragment_ops(uint x, uint y,
       mask = spu_and(mask, amask);
    }
 
-   /* Z and/or stencil testing... */
+
+   /*
+    * Z and/or stencil testing...
+    */
    if (spu.depth_stencil_alpha.depth.enabled ||
        spu.depth_stencil_alpha.stencil[0].enabled) {
 
@@ -178,6 +202,32 @@ spu_fallback_fragment_ops(uint x, uint y,
       }
    }
 
+
+   /*
+    * If we'll need the current framebuffer/tile colors for blending
+    * or logicop or colormask, fetch them now.
+    */
+   if (spu.blend.blend_enable ||
+       spu.blend.logicop_enable ||
+       spu.blend.colormask != 0xf) {
+
+#if LINEAR_QUAD_LAYOUT /* See comments/diagram below */
+      fbc0 = colorTile->ui[y][x*2+0];
+      fbc1 = colorTile->ui[y][x*2+1];
+      fbc2 = colorTile->ui[y][x*2+2];
+      fbc3 = colorTile->ui[y][x*2+3];
+#else
+      fbc0 = colorTile->ui[y+0][x+0];
+      fbc1 = colorTile->ui[y+0][x+1];
+      fbc2 = colorTile->ui[y+1][x+0];
+      fbc3 = colorTile->ui[y+1][x+1];
+#endif
+   }
+
+
+   /*
+    * Do blending
+    */
    if (spu.blend.blend_enable) {
       /* blending terms, misc regs */
       vector float term1r, term1g, term1b, term1a;
@@ -186,43 +236,30 @@ spu_fallback_fragment_ops(uint x, uint y,
 
       vector float fbRGBA[4];  /* current framebuffer colors */
 
-      /* get colors from framebuffer/tile */
+      /* convert framebuffer colors from packed int to vector float */
       {
-         vector float fc[4];
-         uint c0, c1, c2, c3;
-
-#if LINEAR_QUAD_LAYOUT /* See comments/diagram below */
-         c0 = colorTile->ui[y][x*2+0];
-         c1 = colorTile->ui[y][x*2+1];
-         c2 = colorTile->ui[y][x*2+2];
-         c3 = colorTile->ui[y][x*2+3];
-#else
-         c0 = colorTile->ui[y+0][x+0];
-         c1 = colorTile->ui[y+0][x+1];
-         c2 = colorTile->ui[y+1][x+0];
-         c3 = colorTile->ui[y+1][x+1];
-#endif
+         vector float temp[4]; /* float colors in AOS form */
          switch (spu.fb.color_format) {
          case PIPE_FORMAT_B8G8R8A8_UNORM:
-            fc[0] = spu_unpack_B8G8R8A8(c0);
-            fc[1] = spu_unpack_B8G8R8A8(c1);
-            fc[2] = spu_unpack_B8G8R8A8(c2);
-            fc[3] = spu_unpack_B8G8R8A8(c3);
+            temp[0] = spu_unpack_B8G8R8A8(fbc0);
+            temp[1] = spu_unpack_B8G8R8A8(fbc1);
+            temp[2] = spu_unpack_B8G8R8A8(fbc2);
+            temp[3] = spu_unpack_B8G8R8A8(fbc3);
             break;
          case PIPE_FORMAT_A8R8G8B8_UNORM:
-            fc[0] = spu_unpack_A8R8G8B8(c0);
-            fc[1] = spu_unpack_A8R8G8B8(c1);
-            fc[2] = spu_unpack_A8R8G8B8(c2);
-            fc[3] = spu_unpack_A8R8G8B8(c3);
+            temp[0] = spu_unpack_A8R8G8B8(fbc0);
+            temp[1] = spu_unpack_A8R8G8B8(fbc1);
+            temp[2] = spu_unpack_A8R8G8B8(fbc2);
+            temp[3] = spu_unpack_A8R8G8B8(fbc3);
             break;
          default:
             ASSERT(0);
          }
-         _transpose_matrix4x4(fbRGBA, fc);
+         _transpose_matrix4x4(fbRGBA, temp); /* fbRGBA = transpose(temp) */
       }
 
       /*
-       * Compute Src RGB terms
+       * Compute Src RGB terms (fragment color * factor)
        */
       switch (spu.blend.rgb_src_factor) {
       case PIPE_BLENDFACTOR_ONE:
@@ -245,13 +282,33 @@ spu_fallback_fragment_ops(uint x, uint y,
          term1g = spu_mul(fragG, fragA);
          term1b = spu_mul(fragB, fragA);
          break;
+      case PIPE_BLENDFACTOR_DST_COLOR:
+         term1r = spu_mul(fragR, fbRGBA[0]);
+         term1g = spu_mul(fragG, fbRGBA[1]);
+         term1b = spu_mul(fragB, fbRGBA[1]);
+         break;
+      case PIPE_BLENDFACTOR_DST_ALPHA:
+         term1r = spu_mul(fragR, fbRGBA[3]);
+         term1g = spu_mul(fragG, fbRGBA[3]);
+         term1b = spu_mul(fragB, fbRGBA[3]);
+         break;
+      case PIPE_BLENDFACTOR_CONST_COLOR:
+         term1r = spu_mul(fragR, spu_splats(spu.blend_color.color[0]));
+         term1g = spu_mul(fragG, spu_splats(spu.blend_color.color[1]));
+         term1b = spu_mul(fragB, spu_splats(spu.blend_color.color[2]));
+         break;
+      case PIPE_BLENDFACTOR_CONST_ALPHA:
+         term1r = spu_mul(fragR, spu_splats(spu.blend_color.color[3]));
+         term1g = spu_mul(fragG, spu_splats(spu.blend_color.color[3]));
+         term1b = spu_mul(fragB, spu_splats(spu.blend_color.color[3]));
+         break;
       /* XXX more cases */
       default:
          ASSERT(0);
       }
 
       /*
-       * Compute Src Alpha term
+       * Compute Src Alpha term (fragment alpha * factor)
        */
       switch (spu.blend.alpha_src_factor) {
       case PIPE_BLENDFACTOR_ONE:
@@ -263,19 +320,29 @@ spu_fallback_fragment_ops(uint x, uint y,
       case PIPE_BLENDFACTOR_SRC_ALPHA:
          term1a = spu_mul(fragA, fragA);
          break;
+      case PIPE_BLENDFACTOR_DST_COLOR:
+         /* fall-through */
+      case PIPE_BLENDFACTOR_DST_ALPHA:
+         term1a = spu_mul(fragA, fbRGBA[3]);
+         break;
+      case PIPE_BLENDFACTOR_CONST_COLOR:
+         /* fall-through */
+      case PIPE_BLENDFACTOR_CONST_ALPHA:
+         term1a = spu_mul(fragR, spu_splats(spu.blend_color.color[3]));
+         break;
       /* XXX more cases */
       default:
          ASSERT(0);
       }
 
       /*
-       * Compute Dest RGB terms
+       * Compute Dest RGB terms (framebuffer color * factor)
        */
       switch (spu.blend.rgb_dst_factor) {
       case PIPE_BLENDFACTOR_ONE:
-         term2r = fragR;
-         term2g = fragG;
-         term2b = fragB;
+         term2r = fbRGBA[0];
+         term2g = fbRGBA[1];
+         term2b = fbRGBA[2];
          break;
       case PIPE_BLENDFACTOR_ZERO:
          term2r =
@@ -299,17 +366,37 @@ spu_fallback_fragment_ops(uint x, uint y,
          term2g = spu_mul(fbRGBA[1], tmp);
          term2b = spu_mul(fbRGBA[2], tmp);
          break;
-      /* XXX more cases */
+      case PIPE_BLENDFACTOR_DST_COLOR:
+         term2r = spu_mul(fbRGBA[0], fbRGBA[0]);
+         term2g = spu_mul(fbRGBA[1], fbRGBA[1]);
+         term2b = spu_mul(fbRGBA[2], fbRGBA[2]);
+         break;
+      case PIPE_BLENDFACTOR_DST_ALPHA:
+         term2r = spu_mul(fbRGBA[0], fbRGBA[3]);
+         term2g = spu_mul(fbRGBA[1], fbRGBA[3]);
+         term2b = spu_mul(fbRGBA[2], fbRGBA[3]);
+         break;
+      case PIPE_BLENDFACTOR_CONST_COLOR:
+         term2r = spu_mul(fbRGBA[0], spu_splats(spu.blend_color.color[0]));
+         term2g = spu_mul(fbRGBA[1], spu_splats(spu.blend_color.color[1]));
+         term2b = spu_mul(fbRGBA[2], spu_splats(spu.blend_color.color[2]));
+         break;
+      case PIPE_BLENDFACTOR_CONST_ALPHA:
+         term2r = spu_mul(fbRGBA[0], spu_splats(spu.blend_color.color[3]));
+         term2g = spu_mul(fbRGBA[1], spu_splats(spu.blend_color.color[3]));
+         term2b = spu_mul(fbRGBA[2], spu_splats(spu.blend_color.color[3]));
+         break;
+       /* XXX more cases */
       default:
          ASSERT(0);
       }
 
       /*
-       * Compute Dest Alpha term
+       * Compute Dest Alpha term (framebuffer alpha * factor)
        */
       switch (spu.blend.alpha_dst_factor) {
       case PIPE_BLENDFACTOR_ONE:
-         term2a = fragA;
+         term2a = fbRGBA[3];
          break;
       case PIPE_BLENDFACTOR_SRC_COLOR:
          term2a = spu_splats(0.0f);
@@ -322,6 +409,16 @@ spu_fallback_fragment_ops(uint x, uint y,
          tmp = spu_sub(one, fragA);
          term2a = spu_mul(fbRGBA[3], tmp);
          break;
+      case PIPE_BLENDFACTOR_DST_COLOR:
+         /* fall-through */
+      case PIPE_BLENDFACTOR_DST_ALPHA:
+         term2a = spu_mul(fbRGBA[3], fbRGBA[3]);
+         break;
+      case PIPE_BLENDFACTOR_CONST_COLOR:
+         /* fall-through */
+      case PIPE_BLENDFACTOR_CONST_ALPHA:
+         term2a = spu_mul(fbRGBA[3], spu_splats(spu.blend_color.color[3]));
+         break;
       /* XXX more cases */
       default:
          ASSERT(0);
@@ -341,7 +438,21 @@ spu_fallback_fragment_ops(uint x, uint y,
          fragG = spu_sub(term1g, term2g);
          fragB = spu_sub(term1b, term2b);
          break;
-      /* XXX more cases */
+      case PIPE_BLEND_REVERSE_SUBTRACT:
+         fragR = spu_sub(term2r, term1r);
+         fragG = spu_sub(term2g, term1g);
+         fragB = spu_sub(term2b, term1b);
+         break;
+      case PIPE_BLEND_MIN:
+         fragR = spu_min(term1r, term2r);
+         fragG = spu_min(term1g, term2g);
+         fragB = spu_min(term1b, term2b);
+         break;
+      case PIPE_BLEND_MAX:
+         fragR = spu_max(term1r, term2r);
+         fragG = spu_max(term1g, term2g);
+         fragB = spu_max(term1b, term2b);
+         break;
       default:
          ASSERT(0);
       }
@@ -356,7 +467,15 @@ spu_fallback_fragment_ops(uint x, uint y,
       case PIPE_BLEND_SUBTRACT:
          fragA = spu_sub(term1a, term2a);
          break;
-      /* XXX more cases */
+      case PIPE_BLEND_REVERSE_SUBTRACT:
+         fragA = spu_sub(term2a, term1a);
+         break;
+      case PIPE_BLEND_MIN:
+         fragA = spu_min(term1a, term2a);
+         break;
+      case PIPE_BLEND_MAX:
+         fragA = spu_max(term1a, term2a);
+         break;
       default:
          ASSERT(0);
       }
@@ -384,21 +503,20 @@ spu_fallback_fragment_ops(uint x, uint y,
 #endif
 
    /*
-    * Pack float colors into 32-bit RGBA words.
+    * Pack fragment float colors into 32-bit RGBA words.
     */
    switch (spu.fb.color_format) {
    case PIPE_FORMAT_A8R8G8B8_UNORM:
-      c0 = spu_pack_A8R8G8B8(frag_aos[0]);
-      c1 = spu_pack_A8R8G8B8(frag_aos[1]);
-      c2 = spu_pack_A8R8G8B8(frag_aos[2]);
-      c3 = spu_pack_A8R8G8B8(frag_aos[3]);
+      fragc0 = spu_pack_A8R8G8B8(frag_aos[0]);
+      fragc1 = spu_pack_A8R8G8B8(frag_aos[1]);
+      fragc2 = spu_pack_A8R8G8B8(frag_aos[2]);
+      fragc3 = spu_pack_A8R8G8B8(frag_aos[3]);
       break;
-
    case PIPE_FORMAT_B8G8R8A8_UNORM:
-      c0 = spu_pack_B8G8R8A8(frag_aos[0]);
-      c1 = spu_pack_B8G8R8A8(frag_aos[1]);
-      c2 = spu_pack_B8G8R8A8(frag_aos[2]);
-      c3 = spu_pack_B8G8R8A8(frag_aos[3]);
+      fragc0 = spu_pack_B8G8R8A8(frag_aos[0]);
+      fragc1 = spu_pack_B8G8R8A8(frag_aos[1]);
+      fragc2 = spu_pack_B8G8R8A8(frag_aos[2]);
+      fragc3 = spu_pack_B8G8R8A8(frag_aos[3]);
       break;
    default:
       fprintf(stderr, "SPU: Bad pixel format in spu_default_fragment_ops\n");
@@ -407,20 +525,57 @@ spu_fallback_fragment_ops(uint x, uint y,
 
 
    /*
-    * Color masking
+    * Do color masking
     */
    if (spu.blend.colormask != 0xf) {
-      /* XXX to do */
-      /* apply color mask to 32-bit packed colors */
+      uint cmask = 0x0; /* each byte corresponds to a color channel */
+
+      /* Form bitmask depending on color buffer format and colormask bits */
+      switch (spu.fb.color_format) {
+      case PIPE_FORMAT_A8R8G8B8_UNORM:
+         if (spu.blend.colormask & PIPE_MASK_R)
+            cmask |= 0x00ff0000; /* red */
+         if (spu.blend.colormask & PIPE_MASK_G)
+            cmask |= 0x0000ff00; /* green */
+         if (spu.blend.colormask & PIPE_MASK_B)
+            cmask |= 0x000000ff; /* blue */
+         if (spu.blend.colormask & PIPE_MASK_A)
+            cmask |= 0xff000000; /* alpha */
+         break;
+      case PIPE_FORMAT_B8G8R8A8_UNORM:
+         if (spu.blend.colormask & PIPE_MASK_R)
+            cmask |= 0x0000ff00; /* red */
+         if (spu.blend.colormask & PIPE_MASK_G)
+            cmask |= 0x00ff0000; /* green */
+         if (spu.blend.colormask & PIPE_MASK_B)
+            cmask |= 0xff000000; /* blue */
+         if (spu.blend.colormask & PIPE_MASK_A)
+            cmask |= 0x000000ff; /* alpha */
+         break;
+      default:
+         ASSERT(0);
+      }
+
+      /*
+       * Apply color mask to the 32-bit packed colors.
+       * if (cmask[i])
+       *    frag color[i] = frag color[i];
+       * else
+       *    frag color[i] = framebuffer color[i];
+       */
+      fragc0 = (fragc0 & cmask) | (fbc0 & ~cmask);
+      fragc1 = (fragc1 & cmask) | (fbc1 & ~cmask);
+      fragc2 = (fragc2 & cmask) | (fbc2 & ~cmask);
+      fragc3 = (fragc3 & cmask) | (fbc3 & ~cmask);
    }
 
 
    /*
-    * Logic Ops
+    * Do logic ops
     */
    if (spu.blend.logicop_enable) {
       /* XXX to do */
-      /* apply logicop to 32-bit packed colors */
+      /* apply logicop to 32-bit packed colors (fragcx and fbcx) */
    }
 
 
@@ -431,45 +586,46 @@ spu_fallback_fragment_ops(uint x, uint y,
       spu.cur_ctile_status = TILE_STATUS_DIRTY;
    }
    else {
+      /* write no fragments */
       return;
    }
 
 
    /*
-    * Write new quad colors to the framebuffer/tile.
+    * Write new fragment/quad colors to the framebuffer/tile.
     * Only write pixels where the corresponding mask word is set.
     */
 #if LINEAR_QUAD_LAYOUT
    /*
     * Quad layout:
     *  +--+--+--+--+
-    *  |p0|p1|p2|p3|
+    *  |p0|p1|p2|p3|...
     *  +--+--+--+--+
     */
    if (spu_extract(mask, 0))
-      colorTile->ui[y][x*2] = c0;
+      colorTile->ui[y][x*2] = fragc0;
    if (spu_extract(mask, 1))
-      colorTile->ui[y][x*2+1] = c1;
+      colorTile->ui[y][x*2+1] = fragc1;
    if (spu_extract(mask, 2))
-      colorTile->ui[y][x*2+2] = c2;
+      colorTile->ui[y][x*2+2] = fragc2;
    if (spu_extract(mask, 3))
-      colorTile->ui[y][x*2+3] = c3;
+      colorTile->ui[y][x*2+3] = fragc3;
 #else
    /*
     * Quad layout:
     *  +--+--+
-    *  |p0|p1|
+    *  |p0|p1|...
     *  +--+--+
-    *  |p2|p3|
+    *  |p2|p3|...
     *  +--+--+
     */
    if (spu_extract(mask, 0))
-      colorTile->ui[y+0][x+0] = c0;
+      colorTile->ui[y+0][x+0] = fragc0;
    if (spu_extract(mask, 1))
-      colorTile->ui[y+0][x+1] = c1;
+      colorTile->ui[y+0][x+1] = fragc1;
    if (spu_extract(mask, 2))
-      colorTile->ui[y+1][x+0] = c2;
+      colorTile->ui[y+1][x+0] = fragc2;
    if (spu_extract(mask, 3))
-      colorTile->ui[y+1][x+1] = c3;
+      colorTile->ui[y+1][x+1] = fragc3;
 #endif
 }
diff --git a/src/gallium/drivers/cell/spu/spu_render.c b/src/gallium/drivers/cell/spu/spu_render.c
index 305dc988810..7c225e2f27c 100644
--- a/src/gallium/drivers/cell/spu/spu_render.c
+++ b/src/gallium/drivers/cell/spu/spu_render.c
@@ -98,7 +98,7 @@ my_tile(uint tx, uint ty)
 static INLINE void
 get_cz_tiles(uint tx, uint ty)
 {
-   if (spu.read_depth) {
+   if (spu.read_depth_stencil) {
       if (spu.cur_ztile_status != TILE_STATUS_CLEAR) {
          //printf("SPU %u: getting Z tile %u, %u\n", spu.init.id, tx, ty);
          get_tile(tx, ty, &spu.ztile, TAG_READ_TILE_Z, 1);
@@ -153,7 +153,7 @@ static INLINE void
 wait_put_cz_tiles(void)
 {
    wait_on_mask(1 << TAG_WRITE_TILE_COLOR);
-   if (spu.read_depth) {
+   if (spu.read_depth_stencil) {
       wait_on_mask(1 << TAG_WRITE_TILE_Z);
    }
 }
@@ -175,22 +175,14 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr)
    const ubyte *vertices;
    const ushort *indexes;
    uint i, j;
+   uint num_tiles;
 
-
-   if (Debug) {
-      printf("SPU %u: RENDER prim %u, num_vert=%u  num_ind=%u  "
-             "inline_vert=%u\n",
-             spu.init.id,
-             render->prim_type,
-             render->num_verts,
-             render->num_indexes,
-             render->inline_verts);
-
-      /*
-      printf("       bound: %g, %g .. %g, %g\n",
-             render->xmin, render->ymin, render->xmax, render->ymax);
-      */
-   }
+   D_PRINTF(CELL_DEBUG_CMD,
+            "RENDER prim=%u num_vert=%u num_ind=%u inline_vert=%u\n",
+            render->prim_type,
+            render->num_verts,
+            render->num_indexes,
+            render->inline_verts);
 
    ASSERT(sizeof(*render) % 4 == 0);
    ASSERT(total_vertex_bytes % 16 == 0);
@@ -251,6 +243,8 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr)
    wait_on_mask(1 << TAG_SURFACE_CLEAR); /* XXX temporary */
 
 
+   num_tiles = 0;
+
    /**
     ** loop over tiles, rendering tris
     **/
@@ -264,6 +258,8 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr)
       if (!my_tile(tx, ty))
          continue;
 
+      num_tiles++;
+
       spu.cur_ctile_status = spu.ctile_status[ty][tx];
       spu.cur_ztile_status = spu.ztile_status[ty][tx];
 
@@ -293,9 +289,7 @@ cmd_render(const struct cell_command_render *render, uint *pos_incr)
       spu.ztile_status[ty][tx] = spu.cur_ztile_status;
    }
 
-   if (Debug)
-      printf("SPU %u: RENDER done\n",
-             spu.init.id);
+   D_PRINTF(CELL_DEBUG_CMD,
+            "RENDER done (%u tiles hit)\n",
+            num_tiles);
 }
-
-
diff --git a/src/gallium/drivers/cell/spu/spu_texture.c b/src/gallium/drivers/cell/spu/spu_texture.c
index 117b8a36f80..69784c89788 100644
--- a/src/gallium/drivers/cell/spu/spu_texture.c
+++ b/src/gallium/drivers/cell/spu/spu_texture.c
@@ -26,6 +26,8 @@
  **************************************************************************/
 
 
+#include <math.h>
+
 #include "pipe/p_compiler.h"
 #include "spu_main.h"
 #include "spu_texture.h"
@@ -40,37 +42,19 @@
 void
 invalidate_tex_cache(void)
 {
-   uint unit = 0;
-   uint bytes = 4 * spu.texture[unit].width
-      * spu.texture[unit].height;
-
-   spu_dcache_mark_dirty((unsigned) spu.texture[unit].start, bytes);
-}
+   uint lvl;
+   for (lvl = 0; lvl < CELL_MAX_TEXTURE_LEVELS; lvl++) {
+      uint unit = 0;
+      uint bytes = 4 * spu.texture[unit].level[lvl].width
+         * spu.texture[unit].level[lvl].height;
 
+      if (spu.texture[unit].target == PIPE_TEXTURE_CUBE)
+         bytes *= 6;
+      else if (spu.texture[unit].target == PIPE_TEXTURE_3D)
+         bytes *= spu.texture[unit].level[lvl].depth;
 
-/**
- * XXX look into getting texels for all four pixels in a quad at once.
- */
-static uint
-get_texel(uint unit, vec_uint4 coordinate)
-{
-   /*
-    * XXX we could do the "/ TILE_SIZE" and "% TILE_SIZE" operations as
-    * SIMD since X and Y are already in a SIMD register.
-    */
-   const unsigned texture_ea = (uintptr_t) spu.texture[unit].start;
-   ushort x = spu_extract(coordinate, 0);
-   ushort y = spu_extract(coordinate, 1);
-   unsigned tile_offset = sizeof(tile_t)
-      * ((y / TILE_SIZE * spu.texture[unit].tiles_per_row) + (x / TILE_SIZE));
-   ushort texel_offset = (ushort) 4
-      * (ushort) (((ushort) (y % TILE_SIZE) * (ushort) TILE_SIZE) + (x % TILE_SIZE));
-   vec_uint4 tmp;
-
-   spu_dcache_fetch_unaligned((qword *) & tmp,
-                              texture_ea + tile_offset + texel_offset,
-                              4);
-   return spu_extract(tmp, 0);
+      spu_dcache_mark_dirty((unsigned) spu.texture[unit].level[lvl].start, bytes);
+   }
 }
 
 
@@ -88,15 +72,17 @@ get_texel(uint unit, vec_uint4 coordinate)
  * a time.
  */
 static void
-get_four_texels(uint unit, vec_uint4 x, vec_uint4 y, vec_uint4 *texels)
+get_four_texels(const struct spu_texture_level *tlevel, uint face,
+                vec_int4 x, vec_int4 y,
+                vec_uint4 *texels)
 {
-   const unsigned texture_ea = (uintptr_t) spu.texture[unit].start;
-   vec_uint4 tile_x = spu_rlmask(x, -5);
-   vec_uint4 tile_y = spu_rlmask(y, -5);
-   const qword offset_x = si_andi((qword) x, 0x1f);
-   const qword offset_y = si_andi((qword) y, 0x1f);
+   unsigned texture_ea = (uintptr_t) tlevel->start;
+   const vec_int4 tile_x = spu_rlmask(x, -5);  /* tile_x = x / 32 */
+   const vec_int4 tile_y = spu_rlmask(y, -5);  /* tile_y = y / 32 */
+   const qword offset_x = si_andi((qword) x, 0x1f); /* offset_x = x & 0x1f */
+   const qword offset_y = si_andi((qword) y, 0x1f); /* offset_y = y & 0x1f */
 
-   const qword tiles_per_row = (qword) spu_splats(spu.texture[unit].tiles_per_row);
+   const qword tiles_per_row = (qword) spu_splats(tlevel->tiles_per_row);
    const qword tile_size = (qword) spu_splats((unsigned) sizeof(tile_t));
 
    qword tile_offset = si_mpya((qword) tile_y, tiles_per_row, (qword) tile_x);
@@ -107,6 +93,8 @@ get_four_texels(uint unit, vec_uint4 x, vec_uint4 y, vec_uint4 *texels)
    
    vec_uint4 offset = (vec_uint4) si_a(tile_offset, texel_offset);
    
+   texture_ea = texture_ea + face * tlevel->bytes_per_image;
+
    spu_dcache_fetch_unaligned((qword *) & texels[0],
                               texture_ea + spu_extract(offset, 0), 4);
    spu_dcache_fetch_unaligned((qword *) & texels[1],
@@ -118,83 +106,536 @@ get_four_texels(uint unit, vec_uint4 x, vec_uint4 y, vec_uint4 *texels)
 }
 
 
+/** clamp vec to [0, max] */
+static INLINE vector signed int
+spu_clamp(vector signed int vec, vector signed int max)
+{
+   static const vector signed int zero = {0,0,0,0};
+   vector unsigned int c;
+   c = spu_cmpgt(vec, zero);    /* c = vec > zero ? ~0 : 0 */
+   vec = spu_sel(zero, vec, c);
+   c = spu_cmpgt(vec, max);    /* c = vec > max ? ~0 : 0 */
+   vec = spu_sel(vec, max, c);
+   return vec;
+}
+
+
+
 /**
- * Get texture sample at texcoord.
+ * Do nearest texture sampling for four pixels.
+ * \param colors  returned colors in SOA format (rrrr, gggg, bbbb, aaaa).
  */
-vector float
-sample_texture_nearest(uint unit, vector float texcoord)
+void
+sample_texture_2d_nearest(vector float s, vector float t,
+                          uint unit, uint level, uint face,
+                          vector float colors[4])
 {
-   vector float tc = spu_mul(texcoord, spu.texture[unit].tex_size);
-   vector unsigned int itc = spu_convtu(tc, 0);  /* convert to int */
-   itc = spu_and(itc, spu.texture[unit].tex_size_mask); /* mask (GL_REPEAT) */
-   uint texel = get_texel(unit, itc);
-   return spu_unpack_A8R8G8B8(texel);
+   const struct spu_texture_level *tlevel = &spu.texture[unit].level[level];
+   vector float ss = spu_mul(s, tlevel->scale_s);
+   vector float tt = spu_mul(t, tlevel->scale_t);
+   vector signed int is = spu_convts(ss, 0);
+   vector signed int it = spu_convts(tt, 0);
+   vec_uint4 texels[4];
+
+   /* PIPE_TEX_WRAP_REPEAT */
+   is = spu_and(is, tlevel->mask_s);
+   it = spu_and(it, tlevel->mask_t);
+
+   /* PIPE_TEX_WRAP_CLAMP */
+   is = spu_clamp(is, tlevel->max_s);
+   it = spu_clamp(it, tlevel->max_t);
+
+   get_four_texels(tlevel, face, is, it, texels);
+
+   /* convert four packed ARGBA pixels to float RRRR,GGGG,BBBB,AAAA */
+   spu_unpack_A8R8G8B8_transpose4(texels, colors);
 }
 
 
-vector float
-sample_texture_bilinear(uint unit, vector float texcoord)
+/**
+ * Do bilinear texture sampling for four pixels.
+ * \param colors  returned colors in SOA format (rrrr, gggg, bbbb, aaaa).
+ */
+void
+sample_texture_2d_bilinear(vector float s, vector float t,
+                           uint unit, uint level, uint face,
+                           vector float colors[4])
 {
-   static const vec_uint4 offset_x = {0, 0, 1, 1};
-   static const vec_uint4 offset_y = {0, 1, 0, 1};
+   const struct spu_texture_level *tlevel = &spu.texture[unit].level[level];
+   static const vector float half = {-0.5f, -0.5f, -0.5f, -0.5f};
 
-   vector float tc = spu_mul(texcoord, spu.texture[unit].tex_size);
-   tc = spu_add(tc, spu_splats(-0.5f));  /* half texel bias */
+   vector float ss = spu_madd(s, tlevel->scale_s, half);
+   vector float tt = spu_madd(t, tlevel->scale_t, half);
 
-   /* integer texcoords S,T: */
-   vec_uint4 itc = spu_convtu(tc, 0);  /* convert to int */
+   vector signed int is0 = spu_convts(ss, 0);
+   vector signed int it0 = spu_convts(tt, 0);
 
-   vec_uint4 texels[4];
-   
-   /* setup texcoords for quad:
-    *  +-----+-----+
-    *  |x0,y0|x1,y1|
-    *  +-----+-----+
-    *  |x2,y2|x3,y3|
-    *  +-----+-----+
-    */
-   vec_uint4 x = spu_splats(spu_extract(itc, 0));
-   vec_uint4 y = spu_splats(spu_extract(itc, 1));
-   x = spu_add(x, offset_x);
-   y = spu_add(y, offset_y);
+   /* is + 1, it + 1 */
+   vector signed int is1 = spu_add(is0, 1);
+   vector signed int it1 = spu_add(it0, 1);
 
-   /* GL_REPEAT wrap mode: */
-   x = spu_and(x, spu.texture[unit].tex_size_x_mask);
-   y = spu_and(y, spu.texture[unit].tex_size_y_mask);
+   /* PIPE_TEX_WRAP_REPEAT */
+   is0 = spu_and(is0, tlevel->mask_s);
+   it0 = spu_and(it0, tlevel->mask_t);
+   is1 = spu_and(is1, tlevel->mask_s);
+   it1 = spu_and(it1, tlevel->mask_t);
 
-   get_four_texels(unit, x, y, texels);
+   /* PIPE_TEX_WRAP_CLAMP */
+   is0 = spu_clamp(is0, tlevel->max_s);
+   it0 = spu_clamp(it0, tlevel->max_t);
+   is1 = spu_clamp(is1, tlevel->max_s);
+   it1 = spu_clamp(it1, tlevel->max_t);
 
-   /* integer A8R8G8B8 to float texel conversion */
-   vector float texel00 = spu_unpack_A8R8G8B8(spu_extract(texels[0], 0));
-   vector float texel01 = spu_unpack_A8R8G8B8(spu_extract(texels[1], 0));
-   vector float texel10 = spu_unpack_A8R8G8B8(spu_extract(texels[2], 0));
-   vector float texel11 = spu_unpack_A8R8G8B8(spu_extract(texels[3], 0));
+   /* get packed int texels */
+   vector unsigned int texels[16];
+   get_four_texels(tlevel, face, is0, it0, texels + 0);  /* upper-left */
+   get_four_texels(tlevel, face, is1, it0, texels + 4);  /* upper-right */
+   get_four_texels(tlevel, face, is0, it1, texels + 8);  /* lower-left */
+   get_four_texels(tlevel, face, is1, it1, texels + 12); /* lower-right */
 
+   /* convert packed int texels to float colors */
+   vector float ftexels[16];
+   spu_unpack_A8R8G8B8_transpose4(texels + 0, ftexels + 0);
+   spu_unpack_A8R8G8B8_transpose4(texels + 4, ftexels + 4);
+   spu_unpack_A8R8G8B8_transpose4(texels + 8, ftexels + 8);
+   spu_unpack_A8R8G8B8_transpose4(texels + 12, ftexels + 12);
 
    /* Compute weighting factors in [0,1]
     * Multiply texcoord by 1024, AND with 1023, convert back to float.
     */
-   vector float tc1024 = spu_mul(tc, spu_splats(1024.0f));
-   vector signed int itc1024 = spu_convts(tc1024, 0);
-   itc1024 = spu_and(itc1024, spu_splats((1 << 10) - 1));
-   vector float weight = spu_convtf(itc1024, 10);
-
-   /* smeared frac and 1-frac */
-   vector float sfrac = spu_splats(spu_extract(weight, 0));
-   vector float tfrac = spu_splats(spu_extract(weight, 1));
-   vector float sfrac1 = spu_sub(spu_splats(1.0f), sfrac);
-   vector float tfrac1 = spu_sub(spu_splats(1.0f), tfrac);
-
-   /* multiply the samples (colors) by the S/T weights */
-   texel00 = spu_mul(spu_mul(texel00, sfrac1), tfrac1);
-   texel10 = spu_mul(spu_mul(texel10, sfrac ), tfrac1);
-   texel01 = spu_mul(spu_mul(texel01, sfrac1), tfrac );
-   texel11 = spu_mul(spu_mul(texel11, sfrac ), tfrac );
-
-   /* compute sum of weighted samples */
-   vector float texel_sum = spu_add(texel00, texel01);
-   texel_sum = spu_add(texel_sum, texel10);
-   texel_sum = spu_add(texel_sum, texel11);
-
-   return texel_sum;
+   vector float ss1024 = spu_mul(ss, spu_splats(1024.0f));
+   vector signed int iss1024 = spu_convts(ss1024, 0);
+   iss1024 = spu_and(iss1024, 1023);
+   vector float sWeights0 = spu_convtf(iss1024, 10);
+
+   vector float tt1024 = spu_mul(tt, spu_splats(1024.0f));
+   vector signed int itt1024 = spu_convts(tt1024, 0);
+   itt1024 = spu_and(itt1024, 1023);
+   vector float tWeights0 = spu_convtf(itt1024, 10);
+
+   /* 1 - sWeight and 1 - tWeight */
+   vector float sWeights1 = spu_sub(spu_splats(1.0f), sWeights0);
+   vector float tWeights1 = spu_sub(spu_splats(1.0f), tWeights0);
+
+   /* reds, for four pixels */
+   ftexels[ 0] = spu_mul(ftexels[ 0], spu_mul(sWeights1, tWeights1)); /*ul*/
+   ftexels[ 4] = spu_mul(ftexels[ 4], spu_mul(sWeights0, tWeights1)); /*ur*/
+   ftexels[ 8] = spu_mul(ftexels[ 8], spu_mul(sWeights1, tWeights0)); /*ll*/
+   ftexels[12] = spu_mul(ftexels[12], spu_mul(sWeights0, tWeights0)); /*lr*/
+   colors[0] = spu_add(spu_add(ftexels[0], ftexels[4]),
+                       spu_add(ftexels[8], ftexels[12]));
+
+   /* greens, for four pixels */
+   ftexels[ 1] = spu_mul(ftexels[ 1], spu_mul(sWeights1, tWeights1)); /*ul*/
+   ftexels[ 5] = spu_mul(ftexels[ 5], spu_mul(sWeights0, tWeights1)); /*ur*/
+   ftexels[ 9] = spu_mul(ftexels[ 9], spu_mul(sWeights1, tWeights0)); /*ll*/
+   ftexels[13] = spu_mul(ftexels[13], spu_mul(sWeights0, tWeights0)); /*lr*/
+   colors[1] = spu_add(spu_add(ftexels[1], ftexels[5]),
+                       spu_add(ftexels[9], ftexels[13]));
+
+   /* blues, for four pixels */
+   ftexels[ 2] = spu_mul(ftexels[ 2], spu_mul(sWeights1, tWeights1)); /*ul*/
+   ftexels[ 6] = spu_mul(ftexels[ 6], spu_mul(sWeights0, tWeights1)); /*ur*/
+   ftexels[10] = spu_mul(ftexels[10], spu_mul(sWeights1, tWeights0)); /*ll*/
+   ftexels[14] = spu_mul(ftexels[14], spu_mul(sWeights0, tWeights0)); /*lr*/
+   colors[2] = spu_add(spu_add(ftexels[2], ftexels[6]),
+                       spu_add(ftexels[10], ftexels[14]));
+
+   /* alphas, for four pixels */
+   ftexels[ 3] = spu_mul(ftexels[ 3], spu_mul(sWeights1, tWeights1)); /*ul*/
+   ftexels[ 7] = spu_mul(ftexels[ 7], spu_mul(sWeights0, tWeights1)); /*ur*/
+   ftexels[11] = spu_mul(ftexels[11], spu_mul(sWeights1, tWeights0)); /*ll*/
+   ftexels[15] = spu_mul(ftexels[15], spu_mul(sWeights0, tWeights0)); /*lr*/
+   colors[3] = spu_add(spu_add(ftexels[3], ftexels[7]),
+                       spu_add(ftexels[11], ftexels[15]));
+}
+
+
+
+/**
+ * Adapted from /opt/cell/sdk/usr/spu/include/transpose_matrix4x4.h
+ */
+static INLINE void
+transpose(vector unsigned int *mOut0,
+          vector unsigned int *mOut1,
+          vector unsigned int *mOut2,
+          vector unsigned int *mOut3,
+          vector unsigned int *mIn)
+{
+  vector unsigned int abcd, efgh, ijkl, mnop;	/* input vectors */
+  vector unsigned int aeim, bfjn, cgko, dhlp;	/* output vectors */
+  vector unsigned int aibj, ckdl, emfn, gohp;	/* intermediate vectors */
+
+  vector unsigned char shufflehi = ((vector unsigned char) {
+					       0x00, 0x01, 0x02, 0x03,
+					       0x10, 0x11, 0x12, 0x13,
+					       0x04, 0x05, 0x06, 0x07,
+					       0x14, 0x15, 0x16, 0x17});
+  vector unsigned char shufflelo = ((vector unsigned char) {
+					       0x08, 0x09, 0x0A, 0x0B,
+					       0x18, 0x19, 0x1A, 0x1B,
+					       0x0C, 0x0D, 0x0E, 0x0F,
+					       0x1C, 0x1D, 0x1E, 0x1F});
+  abcd = *(mIn+0);
+  efgh = *(mIn+1);
+  ijkl = *(mIn+2);
+  mnop = *(mIn+3);
+
+  aibj = spu_shuffle(abcd, ijkl, shufflehi);
+  ckdl = spu_shuffle(abcd, ijkl, shufflelo);
+  emfn = spu_shuffle(efgh, mnop, shufflehi);
+  gohp = spu_shuffle(efgh, mnop, shufflelo);
+
+  aeim = spu_shuffle(aibj, emfn, shufflehi);
+  bfjn = spu_shuffle(aibj, emfn, shufflelo);
+  cgko = spu_shuffle(ckdl, gohp, shufflehi);
+  dhlp = spu_shuffle(ckdl, gohp, shufflelo);
+
+  *mOut0 = aeim;
+  *mOut1 = bfjn;
+  *mOut2 = cgko;
+  *mOut3 = dhlp;
+}
+
+
+/**
+ * Bilinear filtering, using int instead of float arithmetic for computing
+ * sample weights.
+ */
+void
+sample_texture_2d_bilinear_int(vector float s, vector float t,
+                               uint unit, uint level, uint face,
+                               vector float colors[4])
+{
+   const struct spu_texture_level *tlevel = &spu.texture[unit].level[level];
+   static const vector float half = {-0.5f, -0.5f, -0.5f, -0.5f};
+
+   /* Scale texcoords by size of texture, and add half pixel bias */
+   vector float ss = spu_madd(s, tlevel->scale_s, half);
+   vector float tt = spu_madd(t, tlevel->scale_t, half);
+
+   /* convert float coords to fixed-pt coords with 7 fraction bits */
+   vector signed int is = spu_convts(ss, 7);  /* XXX really need floor() here */
+   vector signed int it = spu_convts(tt, 7);  /* XXX really need floor() here */
+
+   /* compute integer texel weights in [0, 127] */
+   vector signed int sWeights0 = spu_and(is, 127);
+   vector signed int tWeights0 = spu_and(it, 127);
+   vector signed int sWeights1 = spu_sub(127, sWeights0);
+   vector signed int tWeights1 = spu_sub(127, tWeights0);
+
+   /* texel coords: is0 = is / 128, it0 = is / 128 */
+   vector signed int is0 = spu_rlmask(is, -7);
+   vector signed int it0 = spu_rlmask(it, -7);
+
+   /* texel coords: i1 = is0 + 1, it1 = it0 + 1 */
+   vector signed int is1 = spu_add(is0, 1);
+   vector signed int it1 = spu_add(it0, 1);
+
+   /* PIPE_TEX_WRAP_REPEAT */
+   is0 = spu_and(is0, tlevel->mask_s);
+   it0 = spu_and(it0, tlevel->mask_t);
+   is1 = spu_and(is1, tlevel->mask_s);
+   it1 = spu_and(it1, tlevel->mask_t);
+
+   /* PIPE_TEX_WRAP_CLAMP */
+   is0 = spu_clamp(is0, tlevel->max_s);
+   it0 = spu_clamp(it0, tlevel->max_t);
+   is1 = spu_clamp(is1, tlevel->max_s);
+   it1 = spu_clamp(it1, tlevel->max_t);
+
+   /* get packed int texels */
+   vector unsigned int texels[16];
+   get_four_texels(tlevel, face, is0, it0, texels + 0);  /* upper-left */
+   get_four_texels(tlevel, face, is1, it0, texels + 4);  /* upper-right */
+   get_four_texels(tlevel, face, is0, it1, texels + 8);  /* lower-left */
+   get_four_texels(tlevel, face, is1, it1, texels + 12); /* lower-right */
+
+   /* twiddle packed 32-bit BGRA pixels into RGBA as four unsigned ints */
+   {
+      static const unsigned char ZERO = 0x80;
+      int i;
+      for (i = 0; i < 16; i++) {
+         texels[i] = spu_shuffle(texels[i], texels[i],
+                                 ((vector unsigned char) {
+                                    ZERO, ZERO, ZERO, 1,
+                                    ZERO, ZERO, ZERO, 2,
+                                    ZERO, ZERO, ZERO, 3,
+                                    ZERO, ZERO, ZERO, 0}));
+      }
+   }
+
+   /* convert RGBA,RGBA,RGBA,RGBA to RRRR,GGGG,BBBB,AAAA */
+   vector unsigned int texel0, texel1, texel2, texel3, texel4, texel5, texel6, texel7,
+      texel8, texel9, texel10, texel11, texel12, texel13, texel14, texel15;
+   transpose(&texel0, &texel1, &texel2, &texel3, texels + 0);
+   transpose(&texel4, &texel5, &texel6, &texel7, texels + 4);
+   transpose(&texel8, &texel9, &texel10, &texel11, texels + 8);
+   transpose(&texel12, &texel13, &texel14, &texel15, texels + 12);
+
+   /* computed weighted colors */
+   vector unsigned int c0, c1, c2, c3, cSum;
+
+   /* red */
+   c0 = (vector unsigned int) si_mpy((qword) texel0, si_mpy((qword) sWeights1, (qword) tWeights1)); /*ul*/
+   c1 = (vector unsigned int) si_mpy((qword) texel4, si_mpy((qword) sWeights0, (qword) tWeights1)); /*ur*/
+   c2 = (vector unsigned int) si_mpy((qword) texel8, si_mpy((qword) sWeights1, (qword) tWeights0)); /*ll*/
+   c3 = (vector unsigned int) si_mpy((qword) texel12, si_mpy((qword) sWeights0, (qword) tWeights0)); /*lr*/
+   cSum = spu_add(spu_add(c0, c1), spu_add(c2, c3));
+   colors[0] = spu_convtf(cSum, 22);
+
+   /* green */
+   c0 = (vector unsigned int) si_mpy((qword) texel1, si_mpy((qword) sWeights1, (qword) tWeights1)); /*ul*/
+   c1 = (vector unsigned int) si_mpy((qword) texel5, si_mpy((qword) sWeights0, (qword) tWeights1)); /*ur*/
+   c2 = (vector unsigned int) si_mpy((qword) texel9, si_mpy((qword) sWeights1, (qword) tWeights0)); /*ll*/
+   c3 = (vector unsigned int) si_mpy((qword) texel13, si_mpy((qword) sWeights0, (qword) tWeights0)); /*lr*/
+   cSum = spu_add(spu_add(c0, c1), spu_add(c2, c3));
+   colors[1] = spu_convtf(cSum, 22);
+
+   /* blue */
+   c0 = (vector unsigned int) si_mpy((qword) texel2, si_mpy((qword) sWeights1, (qword) tWeights1)); /*ul*/
+   c1 = (vector unsigned int) si_mpy((qword) texel6, si_mpy((qword) sWeights0, (qword) tWeights1)); /*ur*/
+   c2 = (vector unsigned int) si_mpy((qword) texel10, si_mpy((qword) sWeights1, (qword) tWeights0)); /*ll*/
+   c3 = (vector unsigned int) si_mpy((qword) texel14, si_mpy((qword) sWeights0, (qword) tWeights0)); /*lr*/
+   cSum = spu_add(spu_add(c0, c1), spu_add(c2, c3));
+   colors[2] = spu_convtf(cSum, 22);
+
+   /* alpha */
+   c0 = (vector unsigned int) si_mpy((qword) texel3, si_mpy((qword) sWeights1, (qword) tWeights1)); /*ul*/
+   c1 = (vector unsigned int) si_mpy((qword) texel7, si_mpy((qword) sWeights0, (qword) tWeights1)); /*ur*/
+   c2 = (vector unsigned int) si_mpy((qword) texel11, si_mpy((qword) sWeights1, (qword) tWeights0)); /*ll*/
+   c3 = (vector unsigned int) si_mpy((qword) texel15, si_mpy((qword) sWeights0, (qword) tWeights0)); /*lr*/
+   cSum = spu_add(spu_add(c0, c1), spu_add(c2, c3));
+   colors[3] = spu_convtf(cSum, 22);
+}
+
+
+
+/**
+ * Compute level of detail factor from texcoords.
+ */
+static INLINE float
+compute_lambda_2d(uint unit, vector float s, vector float t)
+{
+   uint baseLevel = 0;
+   float width = spu.texture[unit].level[baseLevel].width;
+   float height = spu.texture[unit].level[baseLevel].width;
+   float dsdx = width * (spu_extract(s, 1) - spu_extract(s, 0));
+   float dsdy = width * (spu_extract(s, 2) - spu_extract(s, 0));
+   float dtdx = height * (spu_extract(t, 1) - spu_extract(t, 0));
+   float dtdy = height * (spu_extract(t, 2) - spu_extract(t, 0));
+#if 0
+   /* ideal value */
+   float x = dsdx * dsdx + dtdx * dtdx;
+   float y = dsdy * dsdy + dtdy * dtdy;
+   float rho = x > y ? x : y;
+   rho = sqrtf(rho);
+#else
+   /* approximation */
+   dsdx = fabsf(dsdx);
+   dsdy = fabsf(dsdy);
+   dtdx = fabsf(dtdx);
+   dtdy = fabsf(dtdy);
+   float rho = (dsdx + dsdy + dtdx + dtdy) * 0.5;
+#endif
+   float lambda = logf(rho) * 1.442695f; /* compute logbase2(rho) */
+   return lambda;
+}
+
+
+/**
+ * Blend two sets of colors according to weight.
+ */
+static void
+blend_colors(vector float c0[4], const vector float c1[4], float weight)
+{
+   vector float t = spu_splats(weight);
+   vector float dc0 = spu_sub(c1[0], c0[0]);
+   vector float dc1 = spu_sub(c1[1], c0[1]);
+   vector float dc2 = spu_sub(c1[2], c0[2]);
+   vector float dc3 = spu_sub(c1[3], c0[3]);
+   c0[0] = spu_madd(dc0, t, c0[0]);
+   c0[1] = spu_madd(dc1, t, c0[1]);
+   c0[2] = spu_madd(dc2, t, c0[2]);
+   c0[3] = spu_madd(dc3, t, c0[3]);
+}
+
+
+/**
+ * Texture sampling with level of detail selection and possibly mipmap
+ * interpolation.
+ */
+void
+sample_texture_2d_lod(vector float s, vector float t,
+                      uint unit, uint level_ignored, uint face,
+                      vector float colors[4])
+{
+   /*
+    * Note that we're computing a lambda/lod here that's used for all
+    * four pixels in the quad.
+    */
+   float lambda = compute_lambda_2d(unit, s, t);
+
+   (void) face;
+   (void) level_ignored;
+
+   /* apply lod bias */
+   lambda += spu.sampler[unit].lod_bias;
+
+   /* clamp */
+   if (lambda < spu.sampler[unit].min_lod)
+      lambda = spu.sampler[unit].min_lod;
+   else if (lambda > spu.sampler[unit].max_lod)
+      lambda = spu.sampler[unit].max_lod;
+
+   if (lambda <= 0.0f) {
+      /* magnify */
+      spu.mag_sample_texture_2d[unit](s, t, unit, 0, face, colors);
+   }
+   else {
+      /* minify */
+      if (spu.sampler[unit].min_img_filter == PIPE_TEX_FILTER_LINEAR) {
+         /* sample two mipmap levels and interpolate */
+         int level = (int) lambda;
+         if (level > (int) spu.texture[unit].max_level)
+            level = spu.texture[unit].max_level;
+         spu.min_sample_texture_2d[unit](s, t, unit, level, face, colors);
+         if (spu.sampler[unit].min_img_filter == PIPE_TEX_FILTER_LINEAR) {
+            /* sample second mipmap level */
+            float weight = lambda - (float) level;
+            level++;
+            if (level <= (int) spu.texture[unit].max_level) {
+               vector float colors2[4];
+               spu.min_sample_texture_2d[unit](s, t, unit, level, face, colors2);
+               blend_colors(colors, colors2, weight);
+            }
+         }
+      }
+      else {
+         /* sample one mipmap level */
+         int level = (int) (lambda + 0.5f);
+         if (level > (int) spu.texture[unit].max_level)
+            level = spu.texture[unit].max_level;
+         spu.min_sample_texture_2d[unit](s, t, unit, level, face, colors);
+      }
+   }
+}
+
+
+/** XXX need a SIMD version of this */
+static unsigned
+choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
+{
+   /*
+      major axis
+      direction     target                             sc     tc    ma
+      ----------    -------------------------------    ---    ---   ---
+       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
+       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
+       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
+       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
+       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
+       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
+   */
+   const float arx = fabsf(rx);
+   const float ary = fabsf(ry);
+   const float arz = fabsf(rz);
+   unsigned face;
+   float sc, tc, ma;
+
+   if (arx > ary && arx > arz) {
+      if (rx >= 0.0F) {
+         face = PIPE_TEX_FACE_POS_X;
+         sc = -rz;
+         tc = -ry;
+         ma = arx;
+      }
+      else {
+         face = PIPE_TEX_FACE_NEG_X;
+         sc = rz;
+         tc = -ry;
+         ma = arx;
+      }
+   }
+   else if (ary > arx && ary > arz) {
+      if (ry >= 0.0F) {
+         face = PIPE_TEX_FACE_POS_Y;
+         sc = rx;
+         tc = rz;
+         ma = ary;
+      }
+      else {
+         face = PIPE_TEX_FACE_NEG_Y;
+         sc = rx;
+         tc = -rz;
+         ma = ary;
+      }
+   }
+   else {
+      if (rz > 0.0F) {
+         face = PIPE_TEX_FACE_POS_Z;
+         sc = rx;
+         tc = -ry;
+         ma = arz;
+      }
+      else {
+         face = PIPE_TEX_FACE_NEG_Z;
+         sc = -rx;
+         tc = -ry;
+         ma = arz;
+      }
+   }
+
+   *newS = (sc / ma + 1.0F) * 0.5F;
+   *newT = (tc / ma + 1.0F) * 0.5F;
+
+   return face;
+}
+
+
+
+void
+sample_texture_cube(vector float s, vector float t, vector float r,
+                    uint unit, vector float colors[4])
+{
+   uint p, faces[4], level = 0;
+   float newS[4], newT[4];
+
+   /* Compute cube faces referenced by the four sets of texcoords.
+    * XXX we should SIMD-ize this.
+    */
+   for (p = 0; p < 4; p++) {      
+      float rx = spu_extract(s, p);
+      float ry = spu_extract(t, p);
+      float rz = spu_extract(r, p);
+      faces[p] = choose_cube_face(rx, ry, rz, &newS[p], &newT[p]);
+   }
+
+   if (faces[0] == faces[1] &&
+       faces[0] == faces[2] &&
+       faces[0] == faces[3]) {
+      /* GOOD!  All four texcoords refer to the same cube face */
+      s = (vector float) {newS[0], newS[1], newS[2], newS[3]};
+      t = (vector float) {newT[0], newT[1], newT[2], newT[3]};
+      spu.sample_texture_2d[unit](s, t, unit, level, faces[0], colors);
+   }
+   else {
+      /* BAD!  The four texcoords refer to different faces */
+      for (p = 0; p < 4; p++) {      
+         vector float c[4];
+
+         spu.sample_texture_2d[unit](spu_splats(newS[p]), spu_splats(newT[p]),
+                                     unit, level, faces[p], c);
+
+         float red = spu_extract(c[0], p);
+         float green = spu_extract(c[1], p);
+         float blue = spu_extract(c[2], p);
+         float alpha = spu_extract(c[3], p);
+
+         colors[0] = spu_insert(red,   colors[0], p);
+         colors[1] = spu_insert(green, colors[1], p);
+         colors[2] = spu_insert(blue,  colors[2], p);
+         colors[3] = spu_insert(alpha, colors[3], p);
+      }
+   }
 }
diff --git a/src/gallium/drivers/cell/spu/spu_texture.h b/src/gallium/drivers/cell/spu/spu_texture.h
index f7c9738be88..7b75b007b5a 100644
--- a/src/gallium/drivers/cell/spu/spu_texture.h
+++ b/src/gallium/drivers/cell/spu/spu_texture.h
@@ -36,12 +36,32 @@ extern void
 invalidate_tex_cache(void);
 
 
-extern vector float
-sample_texture_nearest(uint unit, vector float texcoord);
+extern void
+sample_texture_2d_nearest(vector float s, vector float t,
+                          uint unit, uint level, uint face,
+                          vector float colors[4]);
+
+
+extern void
+sample_texture_2d_bilinear(vector float s, vector float t,
+                           uint unit, uint level, uint face,
+                           vector float colors[4]);
+
+extern void
+sample_texture_2d_bilinear_int(vector float s, vector float t,
+                               uint unit, uint level, uint face,
+                               vector float colors[4]);
+
 
+extern void
+sample_texture_2d_lod(vector float s, vector float t,
+                      uint unit, uint level, uint face,
+                      vector float colors[4]);
 
-extern vector float
-sample_texture_bilinear(uint unit, vector float texcoord);
+
+extern void
+sample_texture_cube(vector float s, vector float t, vector float r,
+                    uint unit, vector float colors[4]);
 
 
 #endif /* SPU_TEXTURE_H */
diff --git a/src/gallium/drivers/cell/spu/spu_tile.c b/src/gallium/drivers/cell/spu/spu_tile.c
index 216a33126b7..6905015a483 100644
--- a/src/gallium/drivers/cell/spu/spu_tile.c
+++ b/src/gallium/drivers/cell/spu/spu_tile.c
@@ -87,3 +87,40 @@ put_tile(uint tx, uint ty, const tile_t *tile, int tag, int zBuf)
            0  /* rid */);
 }
 
+
+/**
+ * For tiles whose status is TILE_STATUS_CLEAR, write solid-filled
+ * tiles back to the main framebuffer.
+ */
+void
+really_clear_tiles(uint surfaceIndex)
+{
+   const uint num_tiles = spu.fb.width_tiles * spu.fb.height_tiles;
+   uint i;
+
+   if (surfaceIndex == 0) {
+      clear_c_tile(&spu.ctile);
+
+      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
+         uint tx = i % spu.fb.width_tiles;
+         uint ty = i / spu.fb.width_tiles;
+         if (spu.ctile_status[ty][tx] == TILE_STATUS_CLEAR) {
+            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 0);
+         }
+      }
+   }
+   else {
+      clear_z_tile(&spu.ztile);
+
+      for (i = spu.init.id; i < num_tiles; i += spu.init.num_spus) {
+         uint tx = i % spu.fb.width_tiles;
+         uint ty = i / spu.fb.width_tiles;
+         if (spu.ztile_status[ty][tx] == TILE_STATUS_CLEAR)
+            put_tile(tx, ty, &spu.ctile, TAG_SURFACE_CLEAR, 1);
+      }
+   }
+
+#if 0
+   wait_on_mask(1 << TAG_SURFACE_CLEAR);
+#endif
+}
diff --git a/src/gallium/drivers/cell/spu/spu_tile.h b/src/gallium/drivers/cell/spu/spu_tile.h
index 1b5491112db..7bfb52be8f3 100644
--- a/src/gallium/drivers/cell/spu/spu_tile.h
+++ b/src/gallium/drivers/cell/spu/spu_tile.h
@@ -36,12 +36,14 @@
 
 
 
-void
+extern void
 get_tile(uint tx, uint ty, tile_t *tile, int tag, int zBuf);
 
-void
+extern void
 put_tile(uint tx, uint ty, const tile_t *tile, int tag, int zBuf);
 
+extern void
+really_clear_tiles(uint surfaceIndex);
 
 
 static INLINE void
diff --git a/src/gallium/drivers/cell/spu/spu_tri.c b/src/gallium/drivers/cell/spu/spu_tri.c
index 8b938781920..22e51a86ae5 100644
--- a/src/gallium/drivers/cell/spu/spu_tri.c
+++ b/src/gallium/drivers/cell/spu/spu_tri.c
@@ -43,11 +43,6 @@
 /** Masks are uint[4] vectors with each element being 0 or 0xffffffff */
 typedef vector unsigned int mask_t;
 
-typedef union
-{
-   vector float v;
-   float f[4];
-} float4;
 
 
 /**
@@ -91,9 +86,9 @@ struct edge {
 
 struct interp_coef
 {
-   float4 a0;
-   float4 dadx;
-   float4 dady;
+   vector float a0;
+   vector float dadx;
+   vector float dady;
 };
 
 
@@ -116,21 +111,15 @@ struct setup_stage {
    struct edge etop;
    struct edge emaj;
 
-   float oneoverarea;
+   float oneOverArea;  /* XXX maybe make into vector? */
+
+   uint facing;
 
-   uint tx, ty;
+   uint tx, ty;  /**< position of current tile (x, y) */
 
    int cliprect_minx, cliprect_maxx, cliprect_miny, cliprect_maxy;
 
-#if 0
-   struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS];
-#else
    struct interp_coef coef[PIPE_MAX_SHADER_INPUTS];
-#endif
-
-#if 0
-   struct quad_header quad; 
-#endif
 
    struct {
       int left[2];   /**< [0] = row0, [1] = row1 */
@@ -142,118 +131,105 @@ struct setup_stage {
 };
 
 
-
 static struct setup_stage setup;
 
 
-
-
-#if 0
-/**
- * Basically a cast wrapper.
- */
-static INLINE struct setup_stage *setup_stage( struct draw_stage *stage )
-{
-   return (struct setup_stage *)stage;
-}
-#endif
-
-#if 0
-/**
- * Clip setup.quad against the scissor/surface bounds.
- */
-static INLINE void
-quad_clip(struct setup_stage *setup)
-{
-   const struct pipe_scissor_state *cliprect = &setup.softpipe->cliprect;
-   const int minx = (int) cliprect->minx;
-   const int maxx = (int) cliprect->maxx;
-   const int miny = (int) cliprect->miny;
-   const int maxy = (int) cliprect->maxy;
-
-   if (setup.quad.x0 >= maxx ||
-       setup.quad.y0 >= maxy ||
-       setup.quad.x0 + 1 < minx ||
-       setup.quad.y0 + 1 < miny) {
-      /* totally clipped */
-      setup.quad.mask = 0x0;
-      return;
-   }
-   if (setup.quad.x0 < minx)
-      setup.quad.mask &= (MASK_BOTTOM_RIGHT | MASK_TOP_RIGHT);
-   if (setup.quad.y0 < miny)
-      setup.quad.mask &= (MASK_BOTTOM_LEFT | MASK_BOTTOM_RIGHT);
-   if (setup.quad.x0 == maxx - 1)
-      setup.quad.mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT);
-   if (setup.quad.y0 == maxy - 1)
-      setup.quad.mask &= (MASK_TOP_LEFT | MASK_TOP_RIGHT);
-}
-#endif
-
-#if 0
-/**
- * Emit a quad (pass to next stage) with clipping.
- */
-static INLINE void
-clip_emit_quad(struct setup_stage *setup)
-{
-   quad_clip(setup);
-   if (setup.quad.mask) {
-      struct softpipe_context *sp = setup.softpipe;
-      sp->quad.first->run(sp->quad.first, &setup.quad);
-   }
-}
-#endif
-
 /**
  * Evaluate attribute coefficients (plane equations) to compute
  * attribute values for the four fragments in a quad.
  * Eg: four colors will be computed (in AoS format).
  */
 static INLINE void
-eval_coeff(uint slot, float x, float y, vector float result[4])
+eval_coeff(uint slot, float x, float y, vector float w, vector float result[4])
 {
-   switch (spu.vertex_info.interp_mode[slot]) {
+   switch (spu.vertex_info.attrib[slot].interp_mode) {
    case INTERP_CONSTANT:
       result[QUAD_TOP_LEFT] =
       result[QUAD_TOP_RIGHT] =
       result[QUAD_BOTTOM_LEFT] =
-      result[QUAD_BOTTOM_RIGHT] = setup.coef[slot].a0.v;
+      result[QUAD_BOTTOM_RIGHT] = setup.coef[slot].a0;
       break;
-
    case INTERP_LINEAR:
-      /* fall-through, for now */
-   default:
       {
-         register vector float dadx = setup.coef[slot].dadx.v;
-         register vector float dady = setup.coef[slot].dady.v;
-         register vector float topLeft
-            = spu_add(setup.coef[slot].a0.v,
-                      spu_add(spu_mul(spu_splats(x), dadx),
-                              spu_mul(spu_splats(y), dady)));
+         vector float dadx = setup.coef[slot].dadx;
+         vector float dady = setup.coef[slot].dady;
+         vector float topLeft =
+            spu_add(setup.coef[slot].a0,
+                    spu_add(spu_mul(spu_splats(x), dadx),
+                            spu_mul(spu_splats(y), dady)));
 
          result[QUAD_TOP_LEFT] = topLeft;
          result[QUAD_TOP_RIGHT] = spu_add(topLeft, dadx);
          result[QUAD_BOTTOM_LEFT] = spu_add(topLeft, dady);
          result[QUAD_BOTTOM_RIGHT] = spu_add(spu_add(topLeft, dadx), dady);
       }
+      break;
+   case INTERP_PERSPECTIVE:
+      {
+         vector float dadx = setup.coef[slot].dadx;
+         vector float dady = setup.coef[slot].dady;
+         vector float topLeft =
+            spu_add(setup.coef[slot].a0,
+                    spu_add(spu_mul(spu_splats(x), dadx),
+                            spu_mul(spu_splats(y), dady)));
+
+         vector float wInv = spu_re(w);  /* 1.0 / w */
+
+         result[QUAD_TOP_LEFT] = spu_mul(topLeft, wInv);
+         result[QUAD_TOP_RIGHT] = spu_mul(spu_add(topLeft, dadx), wInv);
+         result[QUAD_BOTTOM_LEFT] = spu_mul(spu_add(topLeft, dady), wInv);
+         result[QUAD_BOTTOM_RIGHT] = spu_mul(spu_add(spu_add(topLeft, dadx), dady), wInv);
+      }
+      break;
+   case INTERP_POS:
+   case INTERP_NONE:
+      break;
+   default:
+      ASSERT(0);
    }
 }
 
 
+/**
+ * As above, but return 4 vectors in SOA format.
+ * XXX this will all be re-written someday.
+ */
+static INLINE void
+eval_coeff_soa(uint slot, float x, float y, vector float w, vector float result[4])
+{
+   eval_coeff(slot, x, y, w, result);
+   _transpose_matrix4x4(result, result);
+}
+
+
+/** Evalute coefficients to get Z for four pixels in a quad */
 static INLINE vector float
 eval_z(float x, float y)
 {
    const uint slot = 0;
-   const float dzdx = setup.coef[slot].dadx.f[2];
-   const float dzdy = setup.coef[slot].dady.f[2];
-   const float topLeft = setup.coef[slot].a0.f[2] + x * dzdx + y * dzdy;
+   const float dzdx = spu_extract(setup.coef[slot].dadx, 2);
+   const float dzdy = spu_extract(setup.coef[slot].dady, 2);
+   const float topLeft = spu_extract(setup.coef[slot].a0, 2) + x * dzdx + y * dzdy;
    const vector float topLeftv = spu_splats(topLeft);
    const vector float derivs = (vector float) { 0.0, dzdx, dzdy, dzdx + dzdy };
    return spu_add(topLeftv, derivs);
 }
 
 
+/** Evalute coefficients to get W for four pixels in a quad */
+static INLINE vector float
+eval_w(float x, float y)
+{
+   const uint slot = 0;
+   const float dwdx = spu_extract(setup.coef[slot].dadx, 3);
+   const float dwdy = spu_extract(setup.coef[slot].dady, 3);
+   const float topLeft = spu_extract(setup.coef[slot].a0, 3) + x * dwdx + y * dwdy;
+   const vector float topLeftv = spu_splats(topLeft);
+   const vector float derivs = (vector float) { 0.0, dwdx, dwdy, dwdx + dwdy };
+   return spu_add(topLeftv, derivs);
+}
+
+
 /**
  * Emit a quad (pass to next stage).  No clipping is done.
  * Note: about 1/5 to 1/7 of the time, mask is zero and this function
@@ -261,120 +237,59 @@ eval_z(float x, float y)
  * overall.
  */
 static INLINE void
-emit_quad( int x, int y, mask_t mask )
+emit_quad( int x, int y, mask_t mask)
 {
    /* If any bits in mask are set... */
    if (spu_extract(spu_orx(mask), 0)) {
       const int ix = x - setup.cliprect_minx;
       const int iy = y - setup.cliprect_miny;
-      vector float colors[4];
 
       spu.cur_ctile_status = TILE_STATUS_DIRTY;
       spu.cur_ztile_status = TILE_STATUS_DIRTY;
 
-      if (spu.texture[0].start) {
-         /* texture mapping */
-         const uint unit = 0;
-         vector float texcoords[4];
-         eval_coeff(2, (float) x, (float) y, texcoords);
-
-         if (spu_extract(mask, 0))
-            colors[0] = spu.sample_texture[unit](unit, texcoords[0]);
-         if (spu_extract(mask, 1))
-            colors[1] = spu.sample_texture[unit](unit, texcoords[1]);
-         if (spu_extract(mask, 2))
-            colors[2] = spu.sample_texture[unit](unit, texcoords[2]);
-         if (spu_extract(mask, 3))
-            colors[3] = spu.sample_texture[unit](unit, texcoords[3]);
-
-
-         if (spu.texture[1].start) {
-            /* multi-texture mapping */
-            const uint unit = 1;
-            vector float colors1[4];
-
-            eval_coeff(2, (float) x, (float) y, texcoords);
-
-            if (spu_extract(mask, 0))
-               colors1[0] = spu.sample_texture[unit](unit, texcoords[0]);
-            if (spu_extract(mask, 1))
-               colors1[1] = spu.sample_texture[unit](unit, texcoords[1]);
-            if (spu_extract(mask, 2))
-               colors1[2] = spu.sample_texture[unit](unit, texcoords[2]);
-            if (spu_extract(mask, 3))
-               colors1[3] = spu.sample_texture[unit](unit, texcoords[3]);
-
-            /* hack: modulate first texture by second */
-            colors[0] = spu_mul(colors[0], colors1[0]);
-            colors[1] = spu_mul(colors[1], colors1[1]);
-            colors[2] = spu_mul(colors[2], colors1[2]);
-            colors[3] = spu_mul(colors[3], colors1[3]);
-         }
+      {
+         /*
+          * Run fragment shader, execute per-fragment ops, update fb/tile.
+          */
+         vector float inputs[4*4], outputs[2*4];
+         vector float fragZ = eval_z((float) x, (float) y);
+         vector float fragW = eval_w((float) x, (float) y);
+         vector unsigned int kill_mask;
 
-      }
-      else {
-         /* simple shading */
+         /* setup inputs */
 #if 0
-         eval_coeff(1, (float) x, (float) y, colors);
-
+         eval_coeff_soa(1, (float) x, (float) y, fragW, inputs);
 #else
-         /* XXX new fragment program code */
-
-         if (spu.fragment_program) {
-            vector float inputs[4*4], outputs[2*4];
-
-            /* setup inputs */
-            eval_coeff(1, (float) x, (float) y, inputs);
-
-            /* Execute the current fragment program */
-            spu.fragment_program(inputs, outputs, spu.constants);
-
-            /* Copy outputs */
-            colors[0] = outputs[0*4+0];
-            colors[1] = outputs[0*4+1];
-            colors[2] = outputs[0*4+2];
-            colors[3] = outputs[0*4+3];
-
-            if (0 && spu.init.id==0 && y == 48) {
-               printf("colors[0] = %f %f %f %f\n",
-                      spu_extract(colors[0], 0),
-                      spu_extract(colors[0], 1),
-                      spu_extract(colors[0], 2),
-                      spu_extract(colors[0], 3));
-               printf("colors[1] = %f %f %f %f\n",
-                      spu_extract(colors[1], 0),
-                      spu_extract(colors[1], 1),
-                      spu_extract(colors[1], 2),
-                      spu_extract(colors[1], 3));
-            }
-
+         uint i;
+         for (i = 0; i < spu.vertex_info.num_attribs; i++) {
+            eval_coeff_soa(i+1, (float) x, (float) y, fragW, inputs + i * 4);
          }
 #endif
-      }
-
-
-      {
-         /* Convert fragment data from AoS to SoA format.
-          * I.e. (RGBA,RGBA,RGBA,RGBA) -> (RRRR,GGGG,BBBB,AAAA)
-          * This is temporary!
+         ASSERT(spu.fragment_program);
+         ASSERT(spu.fragment_ops);
+
+         /* Execute the current fragment program */
+         kill_mask = spu.fragment_program(inputs, outputs, spu.constants);
+
+         mask = spu_andc(mask, kill_mask);
+
+         /* Execute per-fragment/quad operations, including:
+          * alpha test, z test, stencil test, blend and framebuffer writing.
+          * Note that there are two different fragment operations functions
+          * that can be called, one for front-facing fragments, and one
+          * for back-facing fragments.  (Often the two are the same;
+          * but in some cases, like two-sided stenciling, they can be
+          * very different.)  So choose the correct function depending
+          * on the calculated facing.
           */
-         vector float soa_frag[4];
-         _transpose_matrix4x4(soa_frag, colors);
-
-         float4 fragZ;
-
-         fragZ.v = eval_z((float) x, (float) y);
-
-         /* Do all per-fragment/quad operations here, including:
-          *  alpha test, z test, stencil test, blend and framebuffer writing.
-          */
-         spu.fragment_ops(ix, iy, &spu.ctile, &spu.ztile,
-                          fragZ.v,
-                          soa_frag[0], soa_frag[1],
-                          soa_frag[2], soa_frag[3],
+         spu.fragment_ops[setup.facing](ix, iy, &spu.ctile, &spu.ztile,
+                          fragZ,
+                          outputs[0*4+0],
+                          outputs[0*4+1],
+                          outputs[0*4+2],
+                          outputs[0*4+3],
                           mask);
       }
-
    }
 }
 
@@ -383,7 +298,8 @@ emit_quad( int x, int y, mask_t mask )
  * Given an X or Y coordinate, return the block/quad coordinate that it
  * belongs to.
  */
-static INLINE int block( int x )
+static INLINE int
+block(int x)
 {
    return x & ~1;
 }
@@ -394,7 +310,8 @@ static INLINE int block( int x )
  * the triangle's bounds.
  * The mask is a uint4 vector and each element will be 0 or 0xffffffff.
  */
-static INLINE mask_t calculate_mask( int x )
+static INLINE mask_t
+calculate_mask(int x)
 {
    /* This is a little tricky.
     * Use & instead of && to avoid branches.
@@ -412,7 +329,8 @@ static INLINE mask_t calculate_mask( int x )
 /**
  * Render a horizontal span of quads
  */
-static void flush_spans( void )
+static void
+flush_spans(void)
 {
    int minleft, maxright;
    int x;
@@ -440,7 +358,6 @@ static void flush_spans( void )
       return;
    }
 
-
    /* OK, we're very likely to need the tile data now.
     * clear or finish waiting if needed.
     */
@@ -457,7 +374,7 @@ static void flush_spans( void )
    }
    ASSERT(spu.cur_ctile_status != TILE_STATUS_DEFINED);
 
-   if (spu.read_depth) {
+   if (spu.read_depth_stencil) {
       if (spu.cur_ztile_status == TILE_STATUS_GETTING) {
          /* wait for mfc_get() to complete */
          //printf("SPU: %u: waiting for ztile\n", spu.init.id);
@@ -476,9 +393,7 @@ static void flush_spans( void )
     * calculate_mask() could be simplified a bit...
     */
    for (x = block(minleft); x <= block(maxright); x += 2) {
-#if 1
-      emit_quad( x, setup.span.y, calculate_mask( x ) );
-#endif
+      emit_quad( x, setup.span.y, calculate_mask( x ));
    }
 
    setup.span.y = 0;
@@ -487,33 +402,46 @@ static void flush_spans( void )
    setup.span.right[1] = 0;
 }
 
+
 #if DEBUG_VERTS
-static void print_vertex(const struct vertex_header *v)
+static void
+print_vertex(const struct vertex_header *v)
 {
-   int i;
-   fprintf(stderr, "Vertex: (%p)\n", v);
-   for (i = 0; i < setup.quad.nr_attrs; i++) {
-      fprintf(stderr, "  %d: %f %f %f %f\n",  i, 
-              v->data[i][0], v->data[i][1], v->data[i][2], v->data[i][3]);
+   uint i;
+   fprintf(stderr, "  Vertex: (%p)\n", v);
+   for (i = 0; i < spu.vertex_info.num_attribs; i++) {
+      fprintf(stderr, "    %d: %f %f %f %f\n",  i, 
+              spu_extract(v->data[i], 0),
+              spu_extract(v->data[i], 1),
+              spu_extract(v->data[i], 2),
+              spu_extract(v->data[i], 3));
    }
 }
 #endif
 
 
-static boolean setup_sort_vertices(const struct vertex_header *v0,
-                                   const struct vertex_header *v1,
-                                   const struct vertex_header *v2)
+/**
+ * Sort vertices from top to bottom.
+ * Compute area and determine front vs. back facing.
+ * Do coarse clip test against tile bounds
+ * \return  FALSE if tri is totally outside tile, TRUE otherwise
+ */
+static boolean
+setup_sort_vertices(const struct vertex_header *v0,
+                    const struct vertex_header *v1,
+                    const struct vertex_header *v2)
 {
+   float area, sign;
 
 #if DEBUG_VERTS
-   fprintf(stderr, "Triangle:\n");
-   print_vertex(v0);
-   print_vertex(v1);
-   print_vertex(v2);
+   if (spu.init.id==0) {
+      fprintf(stderr, "SPU %u: Triangle:\n", spu.init.id);
+      print_vertex(v0);
+      print_vertex(v1);
+      print_vertex(v2);
+   }
 #endif
 
-   setup.vprovoke = v2;
-
    /* determine bottom to top order of vertices */
    {
       float y0 = spu_extract(v0->data[0], 1);
@@ -525,18 +453,21 @@ static boolean setup_sort_vertices(const struct vertex_header *v0,
 	    setup.vmin = v0;   
 	    setup.vmid = v1;   
 	    setup.vmax = v2;
+            sign = -1.0f;
 	 }
 	 else if (y2 <= y0) {
 	    /* y2<=y0<=y1 */
 	    setup.vmin = v2;   
 	    setup.vmid = v0;   
 	    setup.vmax = v1;   
+            sign = -1.0f;
 	 }
 	 else {
 	    /* y0<=y2<=y1 */
 	    setup.vmin = v0;   
 	    setup.vmid = v2;   
 	    setup.vmax = v1;  
+            sign = 1.0f;
 	 }
       }
       else {
@@ -545,18 +476,21 @@ static boolean setup_sort_vertices(const struct vertex_header *v0,
 	    setup.vmin = v1;   
 	    setup.vmid = v0;   
 	    setup.vmax = v2;  
+            sign = 1.0f;
 	 }
 	 else if (y2 <= y1) {
 	    /* y2<=y1<=y0 */
 	    setup.vmin = v2;   
 	    setup.vmid = v1;   
 	    setup.vmax = v0;  
+            sign = 1.0f;
 	 }
 	 else {
 	    /* y1<=y2<=y0 */
 	    setup.vmin = v1;   
 	    setup.vmid = v2;   
 	    setup.vmax = v0;
+            sign = -1.0f;
 	 }
       }
    }
@@ -585,31 +519,23 @@ static boolean setup_sort_vertices(const struct vertex_header *v0,
    /*
     * Compute triangle's area.  Use 1/area to compute partial
     * derivatives of attributes later.
-    *
-    * The area will be the same as prim->det, but the sign may be
-    * different depending on how the vertices get sorted above.
-    *
-    * To determine whether the primitive is front or back facing we
-    * use the prim->det value because its sign is correct.
     */
-   {
-      const float area = (setup.emaj.dx * setup.ebot.dy - 
-			    setup.ebot.dx * setup.emaj.dy);
-
-      setup.oneoverarea = 1.0f / area;
-      /*
-      _mesa_printf("%s one-over-area %f  area %f  det %f\n",
-                   __FUNCTION__, setup.oneoverarea, area, prim->det );
-      */
-   }
+   area = setup.emaj.dx * setup.ebot.dy - setup.ebot.dx * setup.emaj.dy;
 
-#if 0
-   /* We need to know if this is a front or back-facing triangle for:
-    *  - the GLSL gl_FrontFacing fragment attribute (bool)
-    *  - two-sided stencil test
+   setup.oneOverArea = 1.0f / area;
+
+   /* The product of area * sign indicates front/back orientation (0/1).
+    * Just in case someone gets the bright idea of switching the front
+    * and back constants without noticing that we're assuming their
+    * values in this operation, also assert that the values are
+    * what we think they are.
     */
-   setup.quad.facing = (prim->det > 0.0) ^ (setup.softpipe->rasterizer->front_winding == PIPE_WINDING_CW);
-#endif
+   ASSERT(CELL_FACING_FRONT == 0);
+   ASSERT(CELL_FACING_BACK == 1);
+   setup.facing = (area * sign > 0.0f)
+      ^ (spu.rasterizer.front_winding == PIPE_WINDING_CW);
+
+   setup.vprovoke = v2;
 
    return TRUE;
 }
@@ -622,63 +548,11 @@ static boolean setup_sort_vertices(const struct vertex_header *v0,
  * \param slot  which attribute slot 
  */
 static INLINE void
-const_coeff(uint slot)
+const_coeff4(uint slot)
 {
-   setup.coef[slot].dadx.v = (vector float) {0.0, 0.0, 0.0, 0.0};
-   setup.coef[slot].dady.v = (vector float) {0.0, 0.0, 0.0, 0.0};
-   setup.coef[slot].a0.v = setup.vprovoke->data[slot];
-}
-
-
-/**
- * Compute a0, dadx and dady for a linearly interpolated coefficient,
- * for a triangle.
- */
-static INLINE void
-tri_linear_coeff(uint slot, uint firstComp, uint lastComp)
-{
-   uint i;
-   const float *vmin_d = (float *) &setup.vmin->data[slot];
-   const float *vmid_d = (float *) &setup.vmid->data[slot];
-   const float *vmax_d = (float *) &setup.vmax->data[slot];
-   const float x = spu_extract(setup.vmin->data[0], 0) - 0.5f;
-   const float y = spu_extract(setup.vmin->data[0], 1) - 0.5f;
-
-   for (i = firstComp; i < lastComp; i++) {
-      float botda = vmid_d[i] - vmin_d[i];
-      float majda = vmax_d[i] - vmin_d[i];
-      float a = setup.ebot.dy * majda - botda * setup.emaj.dy;
-      float b = setup.emaj.dx * botda - majda * setup.ebot.dx;
-   
-      ASSERT(slot < PIPE_MAX_SHADER_INPUTS);
-
-      setup.coef[slot].dadx.f[i] = a * setup.oneoverarea;
-      setup.coef[slot].dady.f[i] = b * setup.oneoverarea;
-
-      /* calculate a0 as the value which would be sampled for the
-       * fragment at (0,0), taking into account that we want to sample at
-       * pixel centers, in other words (0.5, 0.5).
-       *
-       * this is neat but unfortunately not a good way to do things for
-       * triangles with very large values of dadx or dady as it will
-       * result in the subtraction and re-addition from a0 of a very
-       * large number, which means we'll end up loosing a lot of the
-       * fractional bits and precision from a0.  the way to fix this is
-       * to define a0 as the sample at a pixel center somewhere near vmin
-       * instead - i'll switch to this later.
-       */
-      setup.coef[slot].a0.f[i] = (vmin_d[i] - 
-                                 (setup.coef[slot].dadx.f[i] * x + 
-                                  setup.coef[slot].dady.f[i] * y));
-   }
-
-   /*
-   _mesa_printf("attr[%d].%c: %f dx:%f dy:%f\n",
-		slot, "xyzw"[i], 
-		setup.coef[slot].a0[i],
-		setup.coef[slot].dadx.f[i],
-		setup.coef[slot].dady.f[i]);
-   */
+   setup.coef[slot].dadx = (vector float) {0.0, 0.0, 0.0, 0.0};
+   setup.coef[slot].dady = (vector float) {0.0, 0.0, 0.0, 0.0};
+   setup.coef[slot].a0 = setup.vprovoke->data[slot];
 }
 
 
@@ -702,18 +576,16 @@ tri_linear_coeff4(uint slot)
    vector float b = spu_sub(spu_mul(spu_splats(setup.emaj.dx), botda),
                             spu_mul(majda, spu_splats(setup.ebot.dx)));
 
-   setup.coef[slot].dadx.v = spu_mul(a, spu_splats(setup.oneoverarea));
-   setup.coef[slot].dady.v = spu_mul(b, spu_splats(setup.oneoverarea));
+   setup.coef[slot].dadx = spu_mul(a, spu_splats(setup.oneOverArea));
+   setup.coef[slot].dady = spu_mul(b, spu_splats(setup.oneOverArea));
 
-   vector float tempx = spu_mul(setup.coef[slot].dadx.v, xxxx);
-   vector float tempy = spu_mul(setup.coef[slot].dady.v, yyyy);
+   vector float tempx = spu_mul(setup.coef[slot].dadx, xxxx);
+   vector float tempy = spu_mul(setup.coef[slot].dady, yyyy);
                          
-   setup.coef[slot].a0.v = spu_sub(vmin_d, spu_add(tempx, tempy));
+   setup.coef[slot].a0 = spu_sub(vmin_d, spu_add(tempx, tempy));
 }
 
 
-
-#if 0
 /**
  * Compute a0, dadx and dady for a perspective-corrected interpolant,
  * for a triangle.
@@ -722,82 +594,76 @@ tri_linear_coeff4(uint slot)
  * Later, when we compute the value at a particular fragment position we'll
  * divide the interpolated value by the interpolated W at that fragment.
  */
-static void tri_persp_coeff( unsigned slot,
-                             unsigned i )
+static void
+tri_persp_coeff4(uint slot)
 {
-   /* premultiply by 1/w:
-    */
-   float mina = setup.vmin->data[slot][i] * setup.vmin->data[0][3];
-   float mida = setup.vmid->data[slot][i] * setup.vmid->data[0][3];
-   float maxa = setup.vmax->data[slot][i] * setup.vmax->data[0][3];
-
-   float botda = mida - mina;
-   float majda = maxa - mina;
-   float a = setup.ebot.dy * majda - botda * setup.emaj.dy;
-   float b = setup.emaj.dx * botda - majda * setup.ebot.dx;
-      
-   /*
-   printf("tri persp %d,%d: %f %f %f\n", slot, i,
-          setup.vmin->data[slot][i],
-          setup.vmid->data[slot][i],
-          setup.vmax->data[slot][i]
-          );
-   */
+   const vector float xxxx = spu_splats(spu_extract(setup.vmin->data[0], 0) - 0.5f);
+   const vector float yyyy = spu_splats(spu_extract(setup.vmin->data[0], 1) - 0.5f);
+
+   const vector float vmin_w = spu_splats(spu_extract(setup.vmin->data[0], 3));
+   const vector float vmid_w = spu_splats(spu_extract(setup.vmid->data[0], 3));
+   const vector float vmax_w = spu_splats(spu_extract(setup.vmax->data[0], 3));
+
+   vector float vmin_d = setup.vmin->data[slot];
+   vector float vmid_d = setup.vmid->data[slot];
+   vector float vmax_d = setup.vmax->data[slot];
+
+   vmin_d = spu_mul(vmin_d, vmin_w);
+   vmid_d = spu_mul(vmid_d, vmid_w);
+   vmax_d = spu_mul(vmax_d, vmax_w);
+
+   vector float botda = vmid_d - vmin_d;
+   vector float majda = vmax_d - vmin_d;
 
-   assert(slot < PIPE_MAX_SHADER_INPUTS);
-   assert(i <= 3);
+   vector float a = spu_sub(spu_mul(spu_splats(setup.ebot.dy), majda),
+                            spu_mul(botda, spu_splats(setup.emaj.dy)));
+   vector float b = spu_sub(spu_mul(spu_splats(setup.emaj.dx), botda),
+                            spu_mul(majda, spu_splats(setup.ebot.dx)));
 
-   setup.coef[slot].dadx.f[i] = a * setup.oneoverarea;
-   setup.coef[slot].dady.f[i] = b * setup.oneoverarea;
-   setup.coef[slot].a0.f[i] = (mina - 
-			    (setup.coef[slot].dadx.f[i] * (setup.vmin->data[0][0] - 0.5f) + 
-			     setup.coef[slot].dady.f[i] * (setup.vmin->data[0][1] - 0.5f)));
+   setup.coef[slot].dadx = spu_mul(a, spu_splats(setup.oneOverArea));
+   setup.coef[slot].dady = spu_mul(b, spu_splats(setup.oneOverArea));
+
+   vector float tempx = spu_mul(setup.coef[slot].dadx, xxxx);
+   vector float tempy = spu_mul(setup.coef[slot].dady, yyyy);
+                         
+   setup.coef[slot].a0 = spu_sub(vmin_d, spu_add(tempx, tempy));
 }
-#endif
+
 
 
 /**
  * Compute the setup.coef[] array dadx, dady, a0 values.
  * Must be called after setup.vmin,vmid,vmax,vprovoke are initialized.
  */
-static void setup_tri_coefficients(void)
+static void
+setup_tri_coefficients(void)
 {
-#if 1
    uint i;
 
    for (i = 0; i < spu.vertex_info.num_attribs; i++) {
-      switch (spu.vertex_info.interp_mode[i]) {
+      switch (spu.vertex_info.attrib[i].interp_mode) {
       case INTERP_NONE:
          break;
-      case INTERP_POS:
-         /*tri_linear_coeff(i, 2, 3);*/
-         /* XXX interp W if PERSPECTIVE... */
-         tri_linear_coeff4(i);
-         break;
       case INTERP_CONSTANT:
-         const_coeff(i);
+         const_coeff4(i);
          break;
+      case INTERP_POS:
+         /* fall-through */
       case INTERP_LINEAR:
          tri_linear_coeff4(i);
          break;
       case INTERP_PERSPECTIVE:
-         tri_linear_coeff4(i);  /* temporary */
+         tri_persp_coeff4(i);
          break;
       default:
          ASSERT(0);
       }
    }
-#else
-   ASSERT(spu.vertex_info.interp_mode[0] == INTERP_POS);
-   ASSERT(spu.vertex_info.interp_mode[1] == INTERP_LINEAR ||
-          spu.vertex_info.interp_mode[1] == INTERP_CONSTANT);
-   tri_linear_coeff(0, 2, 3);  /* slot 0, z */
-   tri_linear_coeff(1, 0, 4);  /* slot 1, color */
-#endif
 }
 
 
-static void setup_tri_edges(void)
+static void
+setup_tri_edges(void)
 {
    float vmin_x = spu_extract(setup.vmin->data[0], 0) + 0.5f;
    float vmid_x = spu_extract(setup.vmid->data[0], 0) + 0.5f;
@@ -827,9 +693,8 @@ static void setup_tri_edges(void)
  * Render the upper or lower half of a triangle.
  * Scissoring/cliprect is applied here too.
  */
-static void subtriangle( struct edge *eleft,
-			 struct edge *eright,
-			 unsigned lines )
+static void
+subtriangle(struct edge *eleft, struct edge *eright, unsigned lines)
 {
    const int minx = setup.cliprect_minx;
    const int maxx = setup.cliprect_maxx;
@@ -902,7 +767,8 @@ static void subtriangle( struct edge *eleft,
  * The tile data should have already been fetched.
  */
 boolean
-tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty)
+tri_draw(const float *v0, const float *v1, const float *v2,
+         uint tx, uint ty)
 {
    setup.tx = tx;
    setup.ty = ty;
@@ -926,19 +792,14 @@ tri_draw(const float *v0, const float *v1, const float *v2, uint tx, uint ty)
    setup.span.y_flags = 0;
    setup.span.right[0] = 0;
    setup.span.right[1] = 0;
-   /*   setup.span.z_mode = tri_z_mode( setup.ctx ); */
 
-   /*   init_constant_attribs( setup ); */
-      
-   if (setup.oneoverarea < 0.0) {
-      /* emaj on left:
-       */
+   if (setup.oneOverArea < 0.0) {
+      /* emaj on left */
       subtriangle( &setup.emaj, &setup.ebot, setup.ebot.lines );
       subtriangle( &setup.emaj, &setup.etop, setup.etop.lines );
    }
    else {
-      /* emaj on right:
-       */
+      /* emaj on right */
       subtriangle( &setup.ebot, &setup.emaj, setup.ebot.lines );
       subtriangle( &setup.etop, &setup.emaj, setup.etop.lines );
    }
diff --git a/src/gallium/drivers/i915simple/i915_debug.c b/src/gallium/drivers/i915simple/i915_debug.c
index 5e26d1b9057..7a4e7051d20 100644
--- a/src/gallium/drivers/i915simple/i915_debug.c
+++ b/src/gallium/drivers/i915simple/i915_debug.c
@@ -210,6 +210,7 @@ BITS(
    PRINTF(stream, ": 0x%x\n", ((dw) & himask) >> (lo));
 }
 
+#ifdef DEBUG
 #define MBZ( dw, hi, lo) do {							\
    unsigned x = (dw) >> (lo);				\
    unsigned lomask = (1 << (lo)) - 1;			\
@@ -217,6 +218,10 @@ BITS(
    himask = (1UL << (hi)) - 1;				\
    assert ((x & himask & ~lomask) == 0);	\
 } while (0)
+#else
+#define MBZ( dw, hi, lo) do {							\
+} while (0)
+#endif
 
 static void
 FLAG(
diff --git a/src/gallium/drivers/i915simple/i915_fpc_translate.c b/src/gallium/drivers/i915simple/i915_fpc_translate.c
index 34b4a846c11..43d62c51765 100644
--- a/src/gallium/drivers/i915simple/i915_fpc_translate.c
+++ b/src/gallium/drivers/i915simple/i915_fpc_translate.c
@@ -144,7 +144,7 @@ src_vector(struct i915_fp_compile *p,
            const struct tgsi_full_src_register *source)
 {
    uint index = source->SrcRegister.Index;
-   uint src, sem_name, sem_ind;
+   uint src = 0, sem_name, sem_ind;
 
    switch (source->SrcRegister.File) {
    case TGSI_FILE_TEMPORARY:
diff --git a/src/gallium/drivers/i915simple/i915_texture.c b/src/gallium/drivers/i915simple/i915_texture.c
index bd87217063c..2f5459af67f 100644
--- a/src/gallium/drivers/i915simple/i915_texture.c
+++ b/src/gallium/drivers/i915simple/i915_texture.c
@@ -178,7 +178,9 @@ i915_displaytarget_layout(struct i915_texture *tex)
    if (tex->base.width[0] >= 128) {
       tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size);
       tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8);
+#if 0 /* used for tiled display targets */
       tex->tiled = 1;
+#endif
    } else {
       tex->stride = round_up(tex->base.nblocksx[0] * pt->block.size, 64);
       tex->total_nblocksy = tex->base.nblocksy[0];
@@ -206,11 +208,10 @@ i945_miptree_layout_2d( struct i915_texture *tex )
    unsigned nblocksx = pt->nblocksx[0];
    unsigned nblocksy = pt->nblocksy[0];
 
-#if 0 /* used for tiled display targets */
-   if (pt->last_level == 0 && pt->block.size == 4)
+   /* used for tiled display targets */
+   if (0)
       if (i915_displaytarget_layout(tex))
 	 return;
-#endif
 
    tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4);
 
diff --git a/src/gallium/drivers/nouveau/nouveau_bo.h b/src/gallium/drivers/nouveau/nouveau_bo.h
new file mode 100644
index 00000000000..65b138283c4
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_bo.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_BO_H__
+#define __NOUVEAU_BO_H__
+
+/* Relocation/Buffer type flags */
+#define NOUVEAU_BO_VRAM  (1 << 0)
+#define NOUVEAU_BO_GART  (1 << 1)
+#define NOUVEAU_BO_RD    (1 << 2)
+#define NOUVEAU_BO_WR    (1 << 3)
+#define NOUVEAU_BO_RDWR  (NOUVEAU_BO_RD | NOUVEAU_BO_WR)
+#define NOUVEAU_BO_MAP   (1 << 4)
+#define NOUVEAU_BO_PIN   (1 << 5)
+#define NOUVEAU_BO_LOW   (1 << 6)
+#define NOUVEAU_BO_HIGH  (1 << 7)
+#define NOUVEAU_BO_OR    (1 << 8)
+#define NOUVEAU_BO_LOCAL (1 << 9)
+#define NOUVEAU_BO_TILED (1 << 10)
+#define NOUVEAU_BO_ZTILE (1 << 11)
+#define NOUVEAU_BO_DUMMY (1 << 31)
+
+struct nouveau_bo {
+	struct nouveau_device *device;
+	uint64_t handle;
+
+	uint64_t size;
+	void *map;
+
+	uint32_t flags;
+	uint64_t offset;
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_channel.h b/src/gallium/drivers/nouveau/nouveau_channel.h
new file mode 100644
index 00000000000..cd99a676bdc
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_channel.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_CHANNEL_H__
+#define __NOUVEAU_CHANNEL_H__
+
+struct nouveau_channel {
+	struct nouveau_device *device;
+	int id;
+
+	struct nouveau_pushbuf *pushbuf;
+
+	struct nouveau_grobj *nullobj;
+	struct nouveau_grobj *vram;
+	struct nouveau_grobj *gart;
+
+	void *user_private;
+	void (*hang_notify)(struct nouveau_channel *);
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_class.h b/src/gallium/drivers/nouveau/nouveau_class.h
new file mode 100644
index 00000000000..3df3d7b2b83
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_class.h
@@ -0,0 +1,8006 @@
+/*************************************************************************
+
+   Autogenerated file, do not edit !
+
+**************************************************************************
+
+   Copyright (C) 2006-2008 :
+   Dmitry Baryshkov,
+   Laurent Carlier,
+   Matthieu Castet,
+   Dawid Gajownik,
+   Jeremy Kolb,
+   Stephane Loeuillet,
+   Patrice Mandin,
+   Stephane Marchesin,
+   Serge Martin,
+   Sylvain Munaut,
+   Simon Raffeiner,
+   Ben Skeggs,
+   Erik Waling,
+   koala_br,
+
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial
+portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+*************************************************************************/
+
+
+#ifndef NOUVEAU_REG_H
+#define NOUVEAU_REG_H 1
+
+
+#define NV01_ROOT									0x00000001
+
+
+
+#define NV01_CONTEXT_DMA								0x00000002
+
+
+
+#define NV01_DEVICE									0x00000003
+
+
+
+#define NV01_TIMER									0x00000004
+
+#define  NV01_TIMER_SYNCHRONIZE								0x00000100
+#define  NV01_TIMER_STOP_ALARM								0x00000104
+#define  NV01_TIMER_DMA_NOTIFY								0x00000180
+#define  NV01_TIMER_TIME(x)								(0x00000300+((x)*4))
+#define  NV01_TIMER_TIME__SIZE								0x00000002
+#define  NV01_TIMER_ALARM_NOTIFY							0x00000308
+
+
+#define NV_IMAGE_STENCIL								0x00000010
+
+#define  NV_IMAGE_STENCIL_NOTIFY							0x00000104
+#define  NV_IMAGE_STENCIL_DMA_NOTIFY							0x00000180
+#define  NV_IMAGE_STENCIL_IMAGE_OUTPUT							0x00000200
+#define  NV_IMAGE_STENCIL_IMAGE_INPUT(x)						(0x00000204+((x)*4))
+#define  NV_IMAGE_STENCIL_IMAGE_INPUT__SIZE						0x00000002
+
+
+#define NV_IMAGE_BLEND_AND								0x00000011
+
+#define  NV_IMAGE_BLEND_AND_NOP								0x00000100
+#define  NV_IMAGE_BLEND_AND_NOTIFY							0x00000104
+#define  NV_IMAGE_BLEND_AND_DMA_NOTIFY							0x00000180
+#define  NV_IMAGE_BLEND_AND_IMAGE_OUTPUT						0x00000200
+#define  NV_IMAGE_BLEND_AND_BETA_INPUT							0x00000204
+#define  NV_IMAGE_BLEND_AND_IMAGE_INPUT							0x00000208
+
+
+#define NV01_CONTEXT_BETA1								0x00000012
+
+#define  NV01_CONTEXT_BETA1_NOP								0x00000100
+#define  NV01_CONTEXT_BETA1_NOTIFY							0x00000104
+#define  NV01_CONTEXT_BETA1_DMA_NOTIFY							0x00000180
+#define  NV01_CONTEXT_BETA1_BETA_1D31							0x00000300
+
+
+#define NV_IMAGE_ROP_AND								0x00000013
+
+#define  NV_IMAGE_ROP_AND_NOTIFY							0x00000104
+#define  NV_IMAGE_ROP_AND_DMA_NOTIFY							0x00000180
+#define  NV_IMAGE_ROP_AND_IMAGE_OUTPUT							0x00000200
+#define  NV_IMAGE_ROP_AND_ROP_INPUT							0x00000204
+#define  NV_IMAGE_ROP_AND_IMAGE_INPUT(x)						(0x00000208+((x)*4))
+#define  NV_IMAGE_ROP_AND_IMAGE_INPUT__SIZE						0x00000002
+
+
+#define NV_IMAGE_COLOR_KEY								0x00000015
+
+
+
+#define NV01_CONTEXT_COLOR_KEY								0x00000017
+
+#define  NV01_CONTEXT_COLOR_KEY_NOP							0x00000100
+#define  NV01_CONTEXT_COLOR_KEY_NOTIFY							0x00000104
+#define  NV01_CONTEXT_COLOR_KEY_DMA_NOTIFY						0x00000180
+#define  NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT						0x00000300
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A8Y8					0x00000001
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X24Y8					0x00000002
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16A1R5G5B5				0x00000003
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X17R5G5B5					0x00000004
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A8R8G8B8					0x00000005
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X8R8G8B8					0x00000006
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_A16Y16					0x00000007
+#define   NV01_CONTEXT_COLOR_KEY_COLOR_FORMAT_X16Y16					0x00000008
+#define  NV01_CONTEXT_COLOR_KEY_COLOR							0x00000304
+
+
+#define NV01_CONTEXT_PATTERN								0x00000018
+
+#define  NV01_CONTEXT_PATTERN_NOP							0x00000100
+#define  NV01_CONTEXT_PATTERN_NOTIFY							0x00000104
+#define  NV01_CONTEXT_PATTERN_DMA_NOTIFY						0x00000180
+#define  NV01_CONTEXT_PATTERN_COLOR_FORMAT						0x00000300
+#define  NV01_CONTEXT_PATTERN_MONOCHROME_FORMAT						0x00000304
+#define  NV01_CONTEXT_PATTERN_SHAPE							0x00000308
+#define  NV01_CONTEXT_PATTERN_COLOR(x)							(0x00000310+((x)*4))
+#define  NV01_CONTEXT_PATTERN_COLOR__SIZE						0x00000002
+#define  NV01_CONTEXT_PATTERN_PATTERN(x)						(0x00000318+((x)*4))
+#define  NV01_CONTEXT_PATTERN_PATTERN__SIZE						0x00000002
+
+
+#define NV01_CONTEXT_CLIP_RECTANGLE							0x00000019
+
+#define  NV01_CONTEXT_CLIP_RECTANGLE_NOP						0x00000100
+#define  NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY						0x00000104
+#define  NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY						0x00000180
+#define  NV01_CONTEXT_CLIP_RECTANGLE_POINT						0x00000300
+#define   NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_SHIFT					0
+#define   NV01_CONTEXT_CLIP_RECTANGLE_POINT_X_MASK					0x0000ffff
+#define   NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_SHIFT					16
+#define   NV01_CONTEXT_CLIP_RECTANGLE_POINT_Y_MASK					0xffff0000
+#define  NV01_CONTEXT_CLIP_RECTANGLE_SIZE						0x00000304
+#define   NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_SHIFT					0
+#define   NV01_CONTEXT_CLIP_RECTANGLE_SIZE_W_MASK					0x0000ffff
+#define   NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_SHIFT					16
+#define   NV01_CONTEXT_CLIP_RECTANGLE_SIZE_H_MASK					0xffff0000
+
+
+#define NV01_RENDER_SOLID_LINE								0x0000001c
+
+#define  NV01_RENDER_SOLID_LINE_NOP							0x00000100
+#define  NV01_RENDER_SOLID_LINE_NOTIFY							0x00000104
+#define  NV01_RENDER_SOLID_LINE_PATCH							0x0000010c
+#define  NV01_RENDER_SOLID_LINE_DMA_NOTIFY						0x00000180
+#define  NV01_RENDER_SOLID_LINE_CLIP_RECTANGLE						0x00000184
+#define  NV01_RENDER_SOLID_LINE_PATTERN							0x00000188
+#define  NV01_RENDER_SOLID_LINE_ROP							0x0000018c
+#define  NV01_RENDER_SOLID_LINE_BETA1							0x00000190
+#define  NV01_RENDER_SOLID_LINE_SURFACE							0x00000194
+#define  NV01_RENDER_SOLID_LINE_OPERATION						0x000002fc
+#define   NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_AND					0x00000000
+#define   NV01_RENDER_SOLID_LINE_OPERATION_ROP_AND					0x00000001
+#define   NV01_RENDER_SOLID_LINE_OPERATION_BLEND_AND					0x00000002
+#define   NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY					0x00000003
+#define   NV01_RENDER_SOLID_LINE_OPERATION_SRCCOPY_PREMULT				0x00000004
+#define   NV01_RENDER_SOLID_LINE_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV01_RENDER_SOLID_LINE_COLOR_FORMAT						0x00000300
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A8Y8					0x00000001
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X24Y8					0x00000002
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16A1R5G5B5				0x00000003
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X17R5G5B5					0x00000004
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A8R8G8B8					0x00000005
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X8R8G8B8					0x00000006
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_A16Y16					0x00000007
+#define   NV01_RENDER_SOLID_LINE_COLOR_FORMAT_X16Y16					0x00000008
+#define  NV01_RENDER_SOLID_LINE_COLOR							0x00000304
+#define  NV01_RENDER_SOLID_LINE_LINE_POINT0(x)						(0x00000400+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_LINE_POINT0__SIZE					0x00000010
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT0_X_SHIFT					0
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT0_X_MASK					0x0000ffff
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_SHIFT					16
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT0_Y_MASK					0xffff0000
+#define  NV01_RENDER_SOLID_LINE_LINE_POINT1(x)						(0x00000404+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_LINE_POINT1__SIZE					0x00000010
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT1_X_SHIFT					0
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT1_X_MASK					0x0000ffff
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_SHIFT					16
+#define   NV01_RENDER_SOLID_LINE_LINE_POINT1_Y_MASK					0xffff0000
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT0_X(x)					(0x00000480+((x)*16))
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT0_X__SIZE					0x00000010
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y(x)					(0x00000484+((x)*16))
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT0_Y__SIZE					0x00000010
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT1_X(x)					(0x00000488+((x)*16))
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT1_X__SIZE					0x00000010
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y(x)					(0x0000048c+((x)*16))
+#define  NV01_RENDER_SOLID_LINE_LINE32_POINT1_Y__SIZE					0x00000010
+#define  NV01_RENDER_SOLID_LINE_POLYLINE(x)						(0x00000500+((x)*4))
+#define  NV01_RENDER_SOLID_LINE_POLYLINE__SIZE						0x00000020
+#define   NV01_RENDER_SOLID_LINE_POLYLINE_X_SHIFT					0
+#define   NV01_RENDER_SOLID_LINE_POLYLINE_X_MASK					0x0000ffff
+#define   NV01_RENDER_SOLID_LINE_POLYLINE_Y_SHIFT					16
+#define   NV01_RENDER_SOLID_LINE_POLYLINE_Y_MASK					0xffff0000
+#define  NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X(x)					(0x00000580+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_X__SIZE				0x00000010
+#define  NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y(x)					(0x00000584+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_POLYLINE32_POINT_Y__SIZE				0x00000010
+#define  NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR(x)					(0x00000600+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_CPOLYLINE_COLOR__SIZE					0x00000010
+#define  NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT(x)					(0x00000604+((x)*8))
+#define  NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT__SIZE					0x00000010
+#define   NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_SHIFT				0
+#define   NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_X_MASK					0x0000ffff
+#define   NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_LINE_CPOLYLINE_POINT_Y_MASK					0xffff0000
+
+
+#define NV01_RENDER_SOLID_TRIANGLE							0x0000001d
+
+#define  NV01_RENDER_SOLID_TRIANGLE_NOP							0x00000100
+#define  NV01_RENDER_SOLID_TRIANGLE_NOTIFY						0x00000104
+#define  NV01_RENDER_SOLID_TRIANGLE_PATCH						0x0000010c
+#define  NV01_RENDER_SOLID_TRIANGLE_DMA_NOTIFY						0x00000180
+#define  NV01_RENDER_SOLID_TRIANGLE_CLIP_RECTANGLE					0x00000184
+#define  NV01_RENDER_SOLID_TRIANGLE_PATTERN						0x00000188
+#define  NV01_RENDER_SOLID_TRIANGLE_ROP							0x0000018c
+#define  NV01_RENDER_SOLID_TRIANGLE_BETA1						0x00000190
+#define  NV01_RENDER_SOLID_TRIANGLE_SURFACE						0x00000194
+#define  NV01_RENDER_SOLID_TRIANGLE_OPERATION						0x000002fc
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_AND				0x00000000
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_ROP_AND					0x00000001
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_AND				0x00000002
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY					0x00000003
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_SRCCOPY_PREMULT				0x00000004
+#define   NV01_RENDER_SOLID_TRIANGLE_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV01_RENDER_SOLID_TRIANGLE_COLOR_FORMAT					0x00000300
+#define  NV01_RENDER_SOLID_TRIANGLE_COLOR						0x00000304
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0					0x00000310
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT0_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1					0x00000314
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT1_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2					0x00000318
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIANGLE_POINT2_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_X					0x00000320
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT0_Y					0x00000324
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_X					0x00000328
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT1_Y					0x0000032c
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_X					0x00000330
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIANGLE32_POINT2_Y					0x00000334
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH(x)						(0x00000400+((x)*4))
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH__SIZE					0x00000020
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_SHIFT					0
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIMESH_X_MASK					0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_SHIFT					16
+#define   NV01_RENDER_SOLID_TRIANGLE_TRIMESH_Y_MASK					0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X(x)				(0x00000480+((x)*8))
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_X__SIZE				0x00000010
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y(x)				(0x00000484+((x)*8))
+#define  NV01_RENDER_SOLID_TRIANGLE_TRIMESH32_POINT_Y__SIZE				0x00000010
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR(x)					(0x00000500+((x)*16))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_COLOR__SIZE				0x00000008
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0(x)					(0x00000504+((x)*16))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0__SIZE				0x00000008
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT0_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1(x)					(0x00000508+((x)*16))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1__SIZE				0x00000008
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT1_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2(x)					(0x0000050c+((x)*16))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2__SIZE				0x00000008
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIANGLE_POINT2_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR(x)					(0x00000580+((x)*8))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_COLOR__SIZE				0x00000010
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT(x)					(0x00000584+((x)*8))
+#define  NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT__SIZE				0x00000010
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_SHIFT				0
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_TRIANGLE_CTRIMESH_POINT_Y_MASK				0xffff0000
+
+
+#define NV01_RENDER_SOLID_RECTANGLE							0x0000001e
+
+#define  NV01_RENDER_SOLID_RECTANGLE_NOP						0x00000100
+#define  NV01_RENDER_SOLID_RECTANGLE_NOTIFY						0x00000104
+#define  NV01_RENDER_SOLID_RECTANGLE_PATCH						0x0000010c
+#define  NV01_RENDER_SOLID_RECTANGLE_DMA_NOTIFY						0x00000180
+#define  NV01_RENDER_SOLID_RECTANGLE_CLIP_RECTANGLE					0x00000184
+#define  NV01_RENDER_SOLID_RECTANGLE_PATTERN						0x00000188
+#define  NV01_RENDER_SOLID_RECTANGLE_ROP						0x0000018c
+#define  NV01_RENDER_SOLID_RECTANGLE_BETA1						0x00000190
+#define  NV01_RENDER_SOLID_RECTANGLE_SURFACE						0x00000194
+#define  NV01_RENDER_SOLID_RECTANGLE_OPERATION						0x000002fc
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_AND				0x00000000
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_ROP_AND					0x00000001
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_AND				0x00000002
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY					0x00000003
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_SRCCOPY_PREMULT				0x00000004
+#define   NV01_RENDER_SOLID_RECTANGLE_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV01_RENDER_SOLID_RECTANGLE_COLOR_FORMAT					0x00000300
+#define  NV01_RENDER_SOLID_RECTANGLE_COLOR						0x00000304
+#define  NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT(x)					(0x00000400+((x)*8))
+#define  NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT__SIZE				0x00000010
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_SHIFT				0
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_X_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_SHIFT				16
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_POINT_Y_MASK				0xffff0000
+#define  NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE(x)					(0x00000404+((x)*8))
+#define  NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE__SIZE				0x00000010
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_SHIFT				0
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_W_MASK				0x0000ffff
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_SHIFT				16
+#define   NV01_RENDER_SOLID_RECTANGLE_RECTANGLE_SIZE_H_MASK				0xffff0000
+
+
+#define NV01_IMAGE_BLIT									0x0000001f
+
+#define  NV01_IMAGE_BLIT_NOP								0x00000100
+#define  NV01_IMAGE_BLIT_NOTIFY								0x00000104
+#define  NV01_IMAGE_BLIT_PATCH								0x0000010c
+#define  NV01_IMAGE_BLIT_DMA_NOTIFY							0x00000180
+#define  NV01_IMAGE_BLIT_COLOR_KEY							0x00000184
+#define  NV01_IMAGE_BLIT_CLIP_RECTANGLE							0x00000188
+#define  NV01_IMAGE_BLIT_PATTERN							0x0000018c
+#define  NV01_IMAGE_BLIT_ROP								0x00000190
+#define  NV01_IMAGE_BLIT_BETA1								0x00000194
+#define  NV01_IMAGE_BLIT_SURFACE							0x0000019c
+#define  NV01_IMAGE_BLIT_OPERATION							0x000002fc
+#define  NV01_IMAGE_BLIT_IMAGE_INPUT							0x00000204
+#define  NV01_IMAGE_BLIT_POINT_IN							0x00000300
+#define   NV01_IMAGE_BLIT_POINT_IN_X_SHIFT						0
+#define   NV01_IMAGE_BLIT_POINT_IN_X_MASK						0x0000ffff
+#define   NV01_IMAGE_BLIT_POINT_IN_Y_SHIFT						16
+#define   NV01_IMAGE_BLIT_POINT_IN_Y_MASK						0xffff0000
+#define  NV01_IMAGE_BLIT_POINT_OUT							0x00000304
+#define   NV01_IMAGE_BLIT_POINT_OUT_X_SHIFT						0
+#define   NV01_IMAGE_BLIT_POINT_OUT_X_MASK						0x0000ffff
+#define   NV01_IMAGE_BLIT_POINT_OUT_Y_SHIFT						16
+#define   NV01_IMAGE_BLIT_POINT_OUT_Y_MASK						0xffff0000
+#define  NV01_IMAGE_BLIT_SIZE								0x00000308
+#define   NV01_IMAGE_BLIT_SIZE_W_SHIFT							0
+#define   NV01_IMAGE_BLIT_SIZE_W_MASK							0x0000ffff
+#define   NV01_IMAGE_BLIT_SIZE_H_SHIFT							16
+#define   NV01_IMAGE_BLIT_SIZE_H_MASK							0xffff0000
+
+
+#define NV01_IMAGE_FROM_CPU								0x00000021
+
+#define  NV01_IMAGE_FROM_CPU_NOP							0x00000100
+#define  NV01_IMAGE_FROM_CPU_NOTIFY							0x00000104
+#define  NV01_IMAGE_FROM_CPU_PATCH							0x0000010c
+#define  NV01_IMAGE_FROM_CPU_DMA_NOTIFY							0x00000180
+#define  NV01_IMAGE_FROM_CPU_COLOR_KEY							0x00000184
+#define  NV01_IMAGE_FROM_CPU_CLIP_RECTANGLE						0x00000188
+#define  NV01_IMAGE_FROM_CPU_PATTERN							0x0000018c
+#define  NV01_IMAGE_FROM_CPU_ROP							0x00000190
+#define  NV01_IMAGE_FROM_CPU_BETA1							0x00000194
+#define  NV01_IMAGE_FROM_CPU_SURFACE							0x00000198
+#define  NV01_IMAGE_FROM_CPU_OPERATION							0x000002fc
+#define   NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_AND					0x00000000
+#define   NV01_IMAGE_FROM_CPU_OPERATION_ROP_AND						0x00000001
+#define   NV01_IMAGE_FROM_CPU_OPERATION_BLEND_AND					0x00000002
+#define   NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY						0x00000003
+#define   NV01_IMAGE_FROM_CPU_OPERATION_SRCCOPY_PREMULT					0x00000004
+#define   NV01_IMAGE_FROM_CPU_OPERATION_BLEND_PREMULT					0x00000005
+#define  NV01_IMAGE_FROM_CPU_COLOR_FORMAT						0x00000300
+#define   NV01_IMAGE_FROM_CPU_COLOR_FORMAT_Y8						0x00000001
+#define   NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A1R5G5B5					0x00000002
+#define   NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X1R5G5B5					0x00000003
+#define   NV01_IMAGE_FROM_CPU_COLOR_FORMAT_A8R8G8B8					0x00000004
+#define   NV01_IMAGE_FROM_CPU_COLOR_FORMAT_X8R8G8B8					0x00000005
+#define  NV01_IMAGE_FROM_CPU_POINT							0x00000304
+#define   NV01_IMAGE_FROM_CPU_POINT_X_SHIFT						0
+#define   NV01_IMAGE_FROM_CPU_POINT_X_MASK						0x0000ffff
+#define   NV01_IMAGE_FROM_CPU_POINT_Y_SHIFT						16
+#define   NV01_IMAGE_FROM_CPU_POINT_Y_MASK						0xffff0000
+#define  NV01_IMAGE_FROM_CPU_SIZE_OUT							0x00000308
+#define   NV01_IMAGE_FROM_CPU_SIZE_OUT_W_SHIFT						0
+#define   NV01_IMAGE_FROM_CPU_SIZE_OUT_W_MASK						0x0000ffff
+#define   NV01_IMAGE_FROM_CPU_SIZE_OUT_H_SHIFT						16
+#define   NV01_IMAGE_FROM_CPU_SIZE_OUT_H_MASK						0xffff0000
+#define  NV01_IMAGE_FROM_CPU_SIZE_IN							0x0000030c
+#define   NV01_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT						0
+#define   NV01_IMAGE_FROM_CPU_SIZE_IN_W_MASK						0x0000ffff
+#define   NV01_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT						16
+#define   NV01_IMAGE_FROM_CPU_SIZE_IN_H_MASK						0xffff0000
+#define  NV01_IMAGE_FROM_CPU_COLOR(x)							(0x00000400+((x)*4))
+#define  NV01_IMAGE_FROM_CPU_COLOR__SIZE						0x00000020
+
+
+#define NV01_NULL									0x00000030
+
+
+
+#define NV03_STRETCHED_IMAGE_FROM_CPU							0x00000036
+
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_NOP						0x00000100
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_NOTIFY						0x00000104
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_PATCH						0x0000010c
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_DMA_NOTIFY					0x00000180
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_KEY					0x00000184
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_PATTERN						0x00000188
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_ROP						0x0000018c
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_BETA1						0x00000190
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_SURFACE						0x00000194
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_OPERATION					0x000002fc
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_COLOR_FORMAT					0x00000300
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN						0x00000304
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_SHIFT					0
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_W_MASK					0x0000ffff
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_SHIFT					16
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_SIZE_IN_H_MASK					0xffff0000
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_DX_DU						0x00000308
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_DY_DV						0x0000030c
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT					0x00000310
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_SHIFT				0
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_X_MASK				0x0000ffff
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_SHIFT				16
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_POINT_Y_MASK				0xffff0000
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE					0x00000314
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_SHIFT				0
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_W_MASK				0x0000ffff
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_SHIFT				16
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_CLIP_SIZE_H_MASK				0xffff0000
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4					0x00000318
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_SHIFT				0
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_X_MASK				0x0000ffff
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_SHIFT				16
+#define   NV03_STRETCHED_IMAGE_FROM_CPU_POINT12D4_Y_MASK				0xffff0000
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_COLOR(x)						(0x00000400+((x)*4))
+#define  NV03_STRETCHED_IMAGE_FROM_CPU_COLOR__SIZE					0x00000020
+
+
+#define NV03_SCALED_IMAGE_FROM_MEMORY							0x00000037
+
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_NOP						0x00000100
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_NOTIFY						0x00000104
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY					0x00000180
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE					0x00000184
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_PATTERN						0x00000188
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_ROP						0x0000018c
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_BETA1						0x00000190
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_SURFACE						0x00000194
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT					0x00000300
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5				0x00000001
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5				0x00000002
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8				0x00000003
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8				0x00000004
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8				0x00000005
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8				0x00000006
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5				0x00000007
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8					0x00000008
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8				0x00000009
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION					0x00000304
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND				0x00000000
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND				0x00000001
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND				0x00000002
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY				0x00000003
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT			0x00000004
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT					0x00000308
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK				0xffff0000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE					0x0000030c
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK				0xffff0000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT					0x00000310
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_X_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_POINT_Y_MASK				0xffff0000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE					0x00000314
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_W_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_OUT_SIZE_H_MASK				0xffff0000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DU_DX					0x00000318
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_DELTA_DV_DY					0x0000031c
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE					0x00000400
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_W_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_SIZE_H_MASK				0xffff0000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT					0x00000404
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_SHIFT			0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_PITCH_MASK			0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_SHIFT			16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_MASK			0x00ff0000
+#define    NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CENTER			0x00010000
+#define    NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_ORIGIN_CORNER			0x00020000
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_SHIFT		24
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_FORMAT_INTERPOLATOR_MASK		0xff000000
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_OFFSET					0x00000408
+#define  NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT					0x0000040c
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_SHIFT				0
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_U_MASK				0x0000ffff
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_SHIFT				16
+#define   NV03_SCALED_IMAGE_FROM_MEMORY_IMAGE_IN_POINT_V_MASK				0xffff0000
+
+
+#define NV04_DVD_SUBPICTURE								0x00000038
+
+#define  NV04_DVD_SUBPICTURE_NOP							0x00000100
+#define  NV04_DVD_SUBPICTURE_NOTIFY							0x00000104
+#define  NV04_DVD_SUBPICTURE_WAIT_FOR_IDLE						0x00000108
+#define  NV04_DVD_SUBPICTURE_DMA_NOTIFY							0x00000180
+#define  NV04_DVD_SUBPICTURE_DMA_OVERLAY						0x00000184
+#define  NV04_DVD_SUBPICTURE_DMA_IMAGEIN						0x00000188
+#define  NV04_DVD_SUBPICTURE_DMA_IMAGEOUT						0x0000018c
+#define  NV04_DVD_SUBPICTURE_IMAGEOUT_POINT						0x00000300
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_X_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_POINT_Y_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE						0x00000304
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_W_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_SIZE_H_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT						0x00000308
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_SHIFT				0
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_PITCH_MASK				0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_SHIFT				16
+#define   NV04_DVD_SUBPICTURE_IMAGEOUT_FORMAT_COLOR_MASK				0xffff0000
+#define  NV04_DVD_SUBPICTURE_IMAGEOUT_OFFSET						0x0000030c
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DU_DX					0x00000310
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_DELTA_DV_DY					0x00000314
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_SIZE						0x00000318
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_W_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_SIZE_H_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT						0x0000031c
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_SHIFT				0
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_PITCH_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_SHIFT				16
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_FORMAT_COLOR_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_OFFSET						0x00000320
+#define  NV04_DVD_SUBPICTURE_IMAGEIN_POINT						0x00000324
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_POINT_U_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_IMAGEIN_POINT_V_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DU_DX					0x00000328
+#define  NV04_DVD_SUBPICTURE_OVERLAY_DELTA_DV_DY					0x0000032c
+#define  NV04_DVD_SUBPICTURE_OVERLAY_SIZE						0x00000330
+#define   NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_OVERLAY_SIZE_W_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_OVERLAY_SIZE_H_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_OVERLAY_FORMAT						0x00000334
+#define   NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_SHIFT				0
+#define   NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_PITCH_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_SHIFT				16
+#define   NV04_DVD_SUBPICTURE_OVERLAY_FORMAT_COLOR_MASK					0xffff0000
+#define  NV04_DVD_SUBPICTURE_OVERLAY_OFFSET						0x00000338
+#define  NV04_DVD_SUBPICTURE_OVERLAY_POINT						0x0000033c
+#define   NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_SHIFT					0
+#define   NV04_DVD_SUBPICTURE_OVERLAY_POINT_U_MASK					0x0000ffff
+#define   NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_SHIFT					16
+#define   NV04_DVD_SUBPICTURE_OVERLAY_POINT_V_MASK					0xffff0000
+
+
+#define NV04_MEMORY_TO_MEMORY_FORMAT							0x00000039
+
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_NOP						0x00000100
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_NOTIFY						0x00000104
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY					0x00000180
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN					0x00000184
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_OUT					0x00000188
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN						0x0000030c
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT					0x00000310
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_IN						0x00000314
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT						0x00000318
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN					0x0000031c
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT					0x00000320
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT						0x00000324
+#define   NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_SHIFT				0
+#define   NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_INPUT_INC_MASK				0x0000000f
+#define   NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_SHIFT				8
+#define   NV04_MEMORY_TO_MEMORY_FORMAT_FORMAT_OUTPUT_INC_MASK				0x00000f00
+#define  NV04_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY					0x00000328
+
+
+#define NV01_MEMORY_LOCAL_BANKED							0x0000003d
+
+
+
+#define NV01_MAPPING_SYSTEM								0x0000003e
+
+
+
+#define NV03_MEMORY_LOCAL_CURSOR							0x0000003f
+
+
+
+#define NV01_MEMORY_LOCAL_LINEAR							0x00000040
+
+
+
+#define NV01_MAPPING_LOCAL								0x00000041
+
+
+
+#define NV04_CONTEXT_SURFACES_2D							0x00000042
+
+#define  NV04_CONTEXT_SURFACES_2D_NOP							0x00000100
+#define  NV04_CONTEXT_SURFACES_2D_NOTIFY						0x00000104
+#define  NV04_CONTEXT_SURFACES_2D_PM_TRIGGER						0x00000140
+#define  NV04_CONTEXT_SURFACES_2D_DMA_NOTIFY						0x00000180
+#define  NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE					0x00000184
+#define  NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_DESTIN					0x00000188
+#define  NV04_CONTEXT_SURFACES_2D_FORMAT						0x00000300
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_Y8						0x00000001
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_Z1R5G5B5				0x00000002
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X1R5G5B5_X1R5G5B5				0x00000003
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5					0x00000004
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_Y16						0x00000005
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_Z8R8G8B8				0x00000006
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X8R8G8B8_X8R8G8B8				0x00000007
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_Z1A7R8G8B8				0x00000008
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_X1A7R8G8B8_X1A7R8G8B8				0x00000009
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8					0x0000000a
+#define   NV04_CONTEXT_SURFACES_2D_FORMAT_Y32						0x0000000b
+#define  NV04_CONTEXT_SURFACES_2D_PITCH							0x00000304
+#define   NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT					0
+#define   NV04_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK					0x0000ffff
+#define   NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT					16
+#define   NV04_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK					0xffff0000
+#define  NV04_CONTEXT_SURFACES_2D_OFFSET_SOURCE						0x00000308
+#define  NV04_CONTEXT_SURFACES_2D_OFFSET_DESTIN						0x0000030c
+
+
+#define NV03_CONTEXT_ROP								0x00000043
+
+#define  NV03_CONTEXT_ROP_NOP								0x00000100
+#define  NV03_CONTEXT_ROP_NOTIFY							0x00000104
+#define  NV03_CONTEXT_ROP_DMA_NOTIFY							0x00000180
+#define  NV03_CONTEXT_ROP_ROP								0x00000300
+#define   NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SHIFT					0
+#define   NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_MASK					0x0000000f
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_CLEAR					0x00000000
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOR					0x00000001
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_INVERTED				0x00000002
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY_INVERTED				0x00000003
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND_REVERSE				0x00000004
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_INVERT					0x00000005
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_XOR					0x00000006
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NAND					0x00000007
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_AND					0x00000008
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_EQUI					0x00000009
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_NOOP					0x0000000a
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_INVERTED				0x0000000b
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_COPY					0x0000000c
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR_REVERSE					0x0000000d
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_OR						0x0000000e
+#define    NV03_CONTEXT_ROP_ROP_DST_LOGIC_OP_SET					0x0000000f
+#define   NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SHIFT					4
+#define   NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_MASK					0x000000f0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_CLEAR					0x00000000
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOR					0x00000010
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_INVERTED				0x00000020
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY_INVERTED				0x00000030
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND_REVERSE				0x00000040
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_INVERT					0x00000050
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_XOR					0x00000060
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NAND					0x00000070
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_AND					0x00000080
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_EQUI					0x00000090
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_NOOP					0x000000a0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_INVERTED				0x000000b0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_COPY					0x000000c0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR_REVERSE					0x000000d0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_OR						0x000000e0
+#define    NV03_CONTEXT_ROP_ROP_SRC_LOGIC_OP_SET					0x000000f0
+
+
+#define NV04_IMAGE_PATTERN								0x00000044
+
+#define  NV04_IMAGE_PATTERN_NOP								0x00000100
+#define  NV04_IMAGE_PATTERN_NOTIFY							0x00000104
+#define  NV04_IMAGE_PATTERN_DMA_NOTIFY							0x00000180
+#define  NV04_IMAGE_PATTERN_COLOR_FORMAT						0x00000300
+#define   NV04_IMAGE_PATTERN_COLOR_FORMAT_A16R5G6B5					0x00000001
+#define   NV04_IMAGE_PATTERN_COLOR_FORMAT_X16A1R5G5B5					0x00000002
+#define   NV04_IMAGE_PATTERN_COLOR_FORMAT_A8R8G8B8					0x00000003
+#define  NV04_IMAGE_PATTERN_MONOCHROME_FORMAT						0x00000304
+#define   NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_CGA6					0x00000001
+#define   NV04_IMAGE_PATTERN_MONOCHROME_FORMAT_LE					0x00000002
+#define  NV04_IMAGE_PATTERN_MONOCHROME_SHAPE						0x00000308
+#define   NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_8X8					0x00000000
+#define   NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_64X1					0x00000001
+#define   NV04_IMAGE_PATTERN_MONOCHROME_SHAPE_1X64					0x00000002
+#define  NV04_IMAGE_PATTERN_PATTERN_SELECT						0x0000030c
+#define   NV04_IMAGE_PATTERN_PATTERN_SELECT_MONO					0x00000001
+#define   NV04_IMAGE_PATTERN_PATTERN_SELECT_COLOR					0x00000002
+#define  NV04_IMAGE_PATTERN_MONOCHROME_COLOR0						0x00000310
+#define  NV04_IMAGE_PATTERN_MONOCHROME_COLOR1						0x00000314
+#define  NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0						0x00000318
+#define  NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1						0x0000031c
+#define  NV04_IMAGE_PATTERN_PATTERN_Y8(x)						(0x00000400+((x)*4))
+#define  NV04_IMAGE_PATTERN_PATTERN_Y8__SIZE						0x00000010
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_SHIFT					0
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y0_MASK						0x000000ff
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_SHIFT					8
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y1_MASK						0x0000ff00
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_SHIFT					16
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y2_MASK						0x00ff0000
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_SHIFT					24
+#define   NV04_IMAGE_PATTERN_PATTERN_Y8_Y3_MASK						0xff000000
+#define  NV04_IMAGE_PATTERN_PATTERN_R5G6B5(x)						(0x00000500+((x)*4))
+#define  NV04_IMAGE_PATTERN_PATTERN_R5G6B5__SIZE					0x00000020
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_SHIFT					0
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B0_MASK					0x0000001f
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_SHIFT					5
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G0_MASK					0x000007e0
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_SHIFT					11
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R0_MASK					0x0000f800
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_SHIFT					16
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_B1_MASK					0x001f0000
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_SHIFT					21
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_G1_MASK					0x07e00000
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_SHIFT					27
+#define   NV04_IMAGE_PATTERN_PATTERN_R5G6B5_R1_MASK					0xf8000000
+#define  NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5(x)						(0x00000600+((x)*4))
+#define  NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5__SIZE					0x00000020
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_SHIFT					0
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B0_MASK					0x0000001f
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_SHIFT					5
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G0_MASK					0x000003e0
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_SHIFT					10
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R0_MASK					0x00007c00
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_SHIFT					16
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_B1_MASK					0x001f0000
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_SHIFT					21
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_G1_MASK					0x03e00000
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_SHIFT					26
+#define   NV04_IMAGE_PATTERN_PATTERN_X1R5G5B5_R1_MASK					0x7c000000
+#define  NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8(x)						(0x00000700+((x)*4))
+#define  NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8__SIZE					0x00000040
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_SHIFT					0
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_B_MASK					0x000000ff
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_SHIFT					8
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_G_MASK					0x0000ff00
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_SHIFT					16
+#define   NV04_IMAGE_PATTERN_PATTERN_X8R8G8B8_R_MASK					0x00ff0000
+
+
+#define NV03_VIDEO_LUT_CURSOR_DAC							0x00000046
+
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SYNCHRONIZE						0x00000100
+#define  NV03_VIDEO_LUT_CURSOR_DAC_STOP_IMAGE						0x00000104
+#define  NV03_VIDEO_LUT_CURSOR_DAC_STOP_CURSOR						0x00000108
+#define  NV03_VIDEO_LUT_CURSOR_DAC_STOP_DAC						0x0000010c
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_NOTIFY						0x00000180
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE(x)						(0x00000184+((x)*4))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_IMAGE__SIZE					0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT(x)						(0x0000018c+((x)*4))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_LUT__SIZE					0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR(x)					(0x00000194+((x)*4))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_DMA_CURSOR__SIZE					0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_GET							0x000002fc
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET(x)					(0x00000300+((x)*8))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_OFFSET__SIZE				0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT(x)					(0x00000304+((x)*8))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT__SIZE				0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_SHIFT			0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_PITCH_MASK				0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_SHIFT			16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_COLOR_MASK				0x0fff0000
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_SHIFT			28
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_IMAGE_FORMAT_NOTIFY_MASK			0xf0000000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET(x)					(0x00000340+((x)*12))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_OFFSET__SIZE				0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT(x)				(0x00000344+((x)*12))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT__SIZE				0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_SHIFT			0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_X_MASK				0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_SHIFT			16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_Y_MASK				0xffff0000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT(x)					(0x00000348+((x)*12))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_FORMAT__SIZE				0x00000002
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A				0x00000358
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_SHIFT			0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_X_MASK			0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_SHIFT			16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_CURSOR_POINT_OUT_A_Y_MASK			0xffff0000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE(x)				(0x00000380+((x)*16))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE__SIZE				0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_SHIFT				0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_W_MASK				0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_SHIFT				16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_IMAGE_SIZE_H_MASK				0xffff0000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC(x)					(0x00000384+((x)*16))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC__SIZE					0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_SHIFT				0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_START_MASK				0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_SHIFT				16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_WIDTH_MASK				0x0fff0000
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_SHIFT			28
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_HSYNC_POLARITY_MASK				0xf0000000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC(x)					(0x00000388+((x)*16))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC__SIZE					0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_SHIFT				0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_START_MASK				0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_SHIFT				16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_WIDTH_MASK				0x0fff0000
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_SHIFT			28
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_VSYNC_POLARITY_MASK				0xf0000000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE(x)				(0x0000038c+((x)*16))
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE__SIZE				0x00000002
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_SHIFT			0
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_WIDTH_MASK			0x0000ffff
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_SHIFT			16
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_HEIGHT_MASK			0x0fff0000
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_SHIFT			28
+#define   NV03_VIDEO_LUT_CURSOR_DAC_SET_DAC_TOTAL_SIZE_NOTIFY_MASK			0xf0000000
+#define  NV03_VIDEO_LUT_CURSOR_DAC_SET_PIXEL_CLOCK					0x000003a0
+
+
+#define NV03_DX3_TEXTURED_TRIANGLE							0x00000048
+
+#define  NV03_DX3_TEXTURED_TRIANGLE_NOP							0x00000100
+#define  NV03_DX3_TEXTURED_TRIANGLE_NOTIFY						0x00000104
+#define  NV03_DX3_TEXTURED_TRIANGLE_PATCH						0x0000010c
+#define  NV03_DX3_TEXTURED_TRIANGLE_DMA_NOTIFY						0x00000180
+#define  NV03_DX3_TEXTURED_TRIANGLE_DMA_TEXTURE						0x00000184
+#define  NV03_DX3_TEXTURED_TRIANGLE_CLIP_RECTANGLE					0x00000188
+#define  NV03_DX3_TEXTURED_TRIANGLE_SURFACE						0x0000018c
+#define  NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_OFFSET					0x00000304
+#define  NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT					0x00000308
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_SHIFT		0
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_MASK_MASK			0x0000ffff
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_SHIFT		16
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_KEY_ENABLE_MASK		0x000f0000
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT				20
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK				0x00f00000
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_SHIFT			24
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MIN_MASK			0x0f000000
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_SHIFT			28
+#define   NV03_DX3_TEXTURED_TRIANGLE_TEXTURE_FORMAT_SIZE_MAX_MASK			0xf0000000
+#define  NV03_DX3_TEXTURED_TRIANGLE_FILTER						0x0000030c
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_SHIFT				0
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_X_MASK				0x0000001f
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_SHIFT				8
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SPREAD_Y_MASK				0x00001f00
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_SHIFT				16
+#define   NV03_DX3_TEXTURED_TRIANGLE_FILTER_SIZE_ADJUST_MASK				0x00ff0000
+#define  NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR						0x00000310
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_SHIFT					0
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_B_MASK					0x000000ff
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_SHIFT					8
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_G_MASK					0x0000ff00
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_SHIFT					16
+#define   NV03_DX3_TEXTURED_TRIANGLE_FOG_COLOR_R_MASK					0x00ff0000
+#define  NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT						0x00000314
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_SHIFT			0
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_INTERPOLATOR_MASK			0x0000000f
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_SHIFT				4
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_U_MASK				0x00000030
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_SHIFT				6
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_WRAP_V_MASK				0x000000c0
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_SHIFT			8
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SOURCE_COLOR_MASK			0x00000f00
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_SHIFT				12
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_CULLING_MASK				0x00007000
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_PERSPECTIVE_ENABLE			(1 << 15)
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_SHIFT				16
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_FUNC_MASK				0x000f0000
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_SHIFT			20
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_Z_WRITE_ENABLE_MASK			0x00f00000
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_SHIFT		24
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_COLOR_WRITE_ENABLE_MASK		0x07000000
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_SHIFT				27
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_ROP_MASK				0x18000000
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_BETA					(1 << 29)
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_DST_BLEND				(1 << 30)
+#define   NV03_DX3_TEXTURED_TRIANGLE_CONTROL_OUT_SRC_BLEND				(1 << 31)
+#define  NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL					0x00000318
+#define   NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_SHIFT			0
+#define   NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_REF_MASK			0x000000ff
+#define   NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_SHIFT			8
+#define   NV03_DX3_TEXTURED_TRIANGLE_ALPHA_CONTROL_ALPHA_FUNC_MASK			0xffffff00
+#define  NV03_DX3_TEXTURED_TRIANGLE_SPECULAR(x)						(0x00001000+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_SPECULAR__SIZE					0x00000040
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_SHIFT					0
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I0_MASK					0x0000000f
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_SHIFT					4
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I1_MASK					0x000000f0
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_SHIFT					8
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I2_MASK					0x00000f00
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_SHIFT					12
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I3_MASK					0x0000f000
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_SHIFT					16
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I4_MASK					0x000f0000
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_SHIFT					20
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_I5_MASK					0x00f00000
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_SHIFT					24
+#define   NV03_DX3_TEXTURED_TRIANGLE_SPECULAR_FOG_MASK					0xff000000
+#define  NV03_DX3_TEXTURED_TRIANGLE_COLOR(x)						(0x00001004+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_COLOR__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_X(x)						(0x00001008+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_X__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_Y(x)						(0x0000100c+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_Y__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_Z(x)						(0x00001010+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_Z__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_M(x)						(0x00001014+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_M__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_U(x)						(0x00001018+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_U__SIZE						0x00000040
+#define  NV03_DX3_TEXTURED_TRIANGLE_V(x)						(0x0000101c+((x)*32))
+#define  NV03_DX3_TEXTURED_TRIANGLE_V__SIZE						0x00000040
+
+
+#define NV04_GDI_RECTANGLE_TEXT								0x0000004a
+
+#define  NV04_GDI_RECTANGLE_TEXT_NOP							0x00000100
+#define  NV04_GDI_RECTANGLE_TEXT_NOTIFY							0x00000104
+#define  NV04_GDI_RECTANGLE_TEXT_PATCH							0x0000010c
+#define  NV04_GDI_RECTANGLE_TEXT_PM_TRIGGER						0x00000140
+#define  NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY						0x00000180
+#define  NV04_GDI_RECTANGLE_TEXT_DMA_FONTS						0x00000184
+#define  NV04_GDI_RECTANGLE_TEXT_PATTERN						0x00000188
+#define  NV04_GDI_RECTANGLE_TEXT_ROP							0x0000018c
+#define  NV04_GDI_RECTANGLE_TEXT_BETA1							0x00000190
+#define  NV04_GDI_RECTANGLE_TEXT_BETA4							0x00000194
+#define  NV04_GDI_RECTANGLE_TEXT_SURFACE						0x00000198
+#define  NV04_GDI_RECTANGLE_TEXT_OPERATION						0x000002fc
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_AND					0x00000000
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_ROP_AND					0x00000001
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_AND					0x00000002
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY					0x00000003
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY_PREMULT				0x00000004
+#define   NV04_GDI_RECTANGLE_TEXT_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT						0x00000300
+#define   NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5				0x00000001
+#define   NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_X16A1R5G5B5				0x00000002
+#define   NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8					0x00000003
+#define  NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT					0x00000304
+#define   NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_CGA6				0x00000001
+#define   NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE					0x00000002
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_A						0x000003fc
+#define  NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(x)				(0x00000400+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT__SIZE			0x00000020
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK			0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT			16
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK			0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE(x)				(0x00000404+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE__SIZE				0x00000020
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK			0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT			16
+#define   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK			0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0						0x000005f4
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_L_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT0_T_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1						0x000005f8
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_R_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_B_POINT1_B_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_B						0x000005fc
+#define  NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0(x)				(0x00000600+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0__SIZE			0x00000020
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK			0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT			16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK			0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1(x)				(0x00000604+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1__SIZE			0x00000020
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK			0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT			16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK			0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0						0x000007ec
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1						0x000007f0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_C						0x000007f4
+#define  NV04_GDI_RECTANGLE_TEXT_SIZE_C							0x000007f8
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK						0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK						0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_POINT_C						0x000007fc
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_C_X_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x)					(0x00000800+((x)*4))
+#define  NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE				0x00000080
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0						0x00000be4
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1						0x00000be8
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR0_E						0x00000bec
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_E						0x00000bf0
+#define  NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E						0x00000bf4
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E						0x00000bf8
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_POINT_E						0x00000bfc
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_E_X_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x)				(0x00000c00+((x)*4))
+#define  NV04_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE				0x00000080
+#define  NV04_GDI_RECTANGLE_TEXT_FONT_F							0x00000ff0
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_F_OFFSET_MASK					0x0fffffff
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_SHIFT					28
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_F_PITCH_MASK					0xf0000000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0						0x00000ff4
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_L_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT0_T_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1						0x00000ff8
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_R_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_F_POINT1_B_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_F						0x00000ffc
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F(x)					(0x00001000+((x)*4))
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F__SIZE				0x00000100
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_INDEX_MASK				0x000000ff
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_SHIFT				8
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_X_MASK				0x000fff00
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_SHIFT				20
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_F_Y_MASK				0xfff00000
+#define  NV04_GDI_RECTANGLE_TEXT_FONT_G							0x000017f0
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_G_OFFSET_MASK					0x0fffffff
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_SHIFT					28
+#define   NV04_GDI_RECTANGLE_TEXT_FONT_G_PITCH_MASK					0xf0000000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0						0x000017f4
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_L_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT0_T_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1						0x000017f8
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_SHIFT					0
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_R_MASK					0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_SHIFT					16
+#define   NV04_GDI_RECTANGLE_TEXT_CLIP_G_POINT1_B_MASK					0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_COLOR1_G						0x000017fc
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT(x)				(0x00001800+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT__SIZE				0x00000100
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_SHIFT			0
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_X_MASK			0x0000ffff
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_SHIFT			16
+#define   NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_POINT_Y_MASK			0xffff0000
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX(x)				(0x00001804+((x)*8))
+#define  NV04_GDI_RECTANGLE_TEXT_CHARACTER_COLOR1_G_INDEX__SIZE				0x00000100
+
+
+#define NV03_GDI_RECTANGLE_TEXT								0x0000004b
+
+#define  NV03_GDI_RECTANGLE_TEXT_NOP							0x00000100
+#define  NV03_GDI_RECTANGLE_TEXT_NOTIFY							0x00000104
+#define  NV03_GDI_RECTANGLE_TEXT_DMA_NOTIFY						0x00000180
+#define  NV03_GDI_RECTANGLE_TEXT_PATTERN						0x00000184
+#define  NV03_GDI_RECTANGLE_TEXT_ROP							0x00000188
+#define  NV03_GDI_RECTANGLE_TEXT_BETA1							0x0000018c
+#define  NV03_GDI_RECTANGLE_TEXT_SURFACE						0x00000190
+#define  NV03_GDI_RECTANGLE_TEXT_OPERATION						0x000002fc
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR_FORMAT						0x00000300
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT					0x00000304
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR1_A						0x000003fc
+#define  NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT				0x00000400
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_SHIFT			0
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_Y_MASK			0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_SHIFT			16
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT_X_MASK			0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE				0x00000404
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_SHIFT			0
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_H_MASK			0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_SHIFT			16
+#define   NV03_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_SIZE_W_MASK			0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B						0x000007f4
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_L_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT0_B_T_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B						0x000007f8
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_R_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_POINT1_B_B_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR1_B						0x000007fc
+#define  NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0				0x00000800
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_SHIFT			0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_L_MASK			0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_SHIFT			16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_0_T_MASK			0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1				0x00000804
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_SHIFT			0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_R_MASK			0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_SHIFT			16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIPPED_RECTANGLE_POINT_1_B_MASK			0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0						0x00000bec
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_L_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT0_T_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1						0x00000bf0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_R_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_C_POINT1_B_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR1_C						0x00000bf4
+#define  NV03_GDI_RECTANGLE_TEXT_SIZE_C							0x00000bf8
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_C_W_MASK						0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_C_H_MASK						0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_POINT_C						0x00000bfc
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_C_X_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_C_X_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_C_Y_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C(x)					(0x00000c00+((x)*4))
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_C__SIZE				0x00000020
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0						0x00000fe8
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_L_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT0_T_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1						0x00000fec
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_R_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_D_POINT1_B_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR1_D						0x00000ff0
+#define  NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D						0x00000ff4
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_W_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_D_H_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D						0x00000ff8
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_W_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_D_H_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_POINT_D						0x00000ffc
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_D_X_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_D_X_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_D_Y_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D(x)					(0x00001000+((x)*4))
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR1_D__SIZE				0x00000020
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0						0x000013e4
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_L_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT0_T_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1						0x000013e8
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_R_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_CLIP_E_POINT1_B_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR0_E						0x000013ec
+#define  NV03_GDI_RECTANGLE_TEXT_COLOR1_E						0x000013f0
+#define  NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E						0x000013f4
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_W_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_IN_E_H_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E						0x000013f8
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_W_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_SIZE_OUT_E_H_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_POINT_E						0x000013fc
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_E_X_SHIFT					0
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_E_X_MASK					0x0000ffff
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_SHIFT					16
+#define   NV03_GDI_RECTANGLE_TEXT_POINT_E_Y_MASK					0xffff0000
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E(x)				(0x00001400+((x)*4))
+#define  NV03_GDI_RECTANGLE_TEXT_MONOCHROME_COLOR01_E__SIZE				0x00000020
+
+
+#define NV04_SWIZZLED_SURFACE								0x00000052
+
+#define  NV04_SWIZZLED_SURFACE_NOP							0x00000100
+#define  NV04_SWIZZLED_SURFACE_NOTIFY							0x00000104
+#define  NV04_SWIZZLED_SURFACE_DMA_NOTIFY						0x00000180
+#define  NV04_SWIZZLED_SURFACE_DMA_IMAGE						0x00000184
+#define  NV04_SWIZZLED_SURFACE_FORMAT							0x00000300
+#define   NV04_SWIZZLED_SURFACE_FORMAT_COLOR_SHIFT					0
+#define   NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK					0x000000ff
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y8					0x00000001
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5				0x00000002
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1R5G5B5_X1R5G5B5				0x00000003
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_R5G6B5					0x00000004
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y16					0x00000005
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8				0x00000006
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X8R8G8B8_X8R8G8B8				0x00000007
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8			0x00000008
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8			0x00000009
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_A8R8G8B8					0x0000000a
+#define    NV04_SWIZZLED_SURFACE_FORMAT_COLOR_Y32					0x0000000b
+#define   NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT				16
+#define   NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_MASK					0x00ff0000
+#define   NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT				24
+#define   NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_MASK					0xff000000
+#define  NV04_SWIZZLED_SURFACE_OFFSET							0x00000304
+
+
+#define NV04_CONTEXT_SURFACES_3D							0x00000053
+
+#define  NV04_CONTEXT_SURFACES_3D_NOP							0x00000100
+#define  NV04_CONTEXT_SURFACES_3D_NOTIFY						0x00000104
+#define  NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY						0x00000180
+#define  NV04_CONTEXT_SURFACES_3D_DMA_COLOR						0x00000184
+#define  NV04_CONTEXT_SURFACES_3D_DMA_ZETA						0x00000188
+#define  NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL					0x000002f8
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_SHIFT				0
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK				0x0000ffff
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_SHIFT				16
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_W_MASK				0xffff0000
+#define  NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL						0x000002fc
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_SHIFT				0
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK					0x0000ffff
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_SHIFT				16
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_H_MASK					0xffff0000
+#define  NV04_CONTEXT_SURFACES_3D_FORMAT						0x00000300
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_SHIFT					0
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK					0x000000ff
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_Z1R5G5B5			0x00000001
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1R5G5B5_X1R5G5B5			0x00000002
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_R5G6B5					0x00000003
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_Z8R8G8B8			0x00000004
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X8R8G8B8_X8R8G8B8			0x00000005
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_Z1A7R8G8B8			0x00000006
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_X1A7R8G8B8_X1A7R8G8B8			0x00000007
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_A8R8G8B8				0x00000008
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT					8
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK					0x0000ff00
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_PITCH					0x00000100
+#define    NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SWIZZLE					0x00000200
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_SHIFT				16
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_U_MASK				0x00ff0000
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_SHIFT				24
+#define   NV04_CONTEXT_SURFACES_3D_FORMAT_BASE_SIZE_V_MASK				0xff000000
+#define  NV04_CONTEXT_SURFACES_3D_CLIP_SIZE						0x00000304
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_SHIFT					0
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_W_MASK					0x0000ffff
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_SHIFT					16
+#define   NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_H_MASK					0xffff0000
+#define  NV04_CONTEXT_SURFACES_3D_PITCH							0x00000308
+#define   NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_SHIFT					0
+#define   NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK					0x0000ffff
+#define   NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT					16
+#define   NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK					0xffff0000
+#define  NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR						0x0000030c
+#define  NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA						0x00000310
+
+
+#define NV04_DX5_TEXTURED_TRIANGLE							0x00000054
+
+#define  NV04_DX5_TEXTURED_TRIANGLE_NOP							0x00000100
+#define  NV04_DX5_TEXTURED_TRIANGLE_NOTIFY						0x00000104
+#define  NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY						0x00000180
+#define  NV04_DX5_TEXTURED_TRIANGLE_DMA_A						0x00000184
+#define  NV04_DX5_TEXTURED_TRIANGLE_DMA_B						0x00000188
+#define  NV04_DX5_TEXTURED_TRIANGLE_SURFACE						0x0000018c
+#define  NV04_DX5_TEXTURED_TRIANGLE_COLORKEY						0x00000300
+#define  NV04_DX5_TEXTURED_TRIANGLE_OFFSET						0x00000304
+#define  NV04_DX5_TEXTURED_TRIANGLE_FORMAT						0x00000308
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_SHIFT					0
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_DMA_MASK					0x00000003
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_SHIFT			2
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_KEY_MATCH_MASK			0x0000000c
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT				4
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK				0x00000030
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CENTER				0x00000010
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER				0x00000020
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT				6
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_MASK				0x000000c0
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CENTER				0x00000040
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER				0x00000080
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_SHIFT					8
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_MASK					0x00000f00
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_Y8					0x00000100
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A1R5G5B5				0x00000200
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X1R5G5B5				0x00000300
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A4R4G4B4				0x00000400
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_R5G6B5				0x00000500
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_A8R8G8B8				0x00000600
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_X8R8G8B8				0x00000700
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT				12
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK				0x0000f000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT				16
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_MASK				0x000f0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT				20
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_MASK				0x00f00000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT				24
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MASK				0x07000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT				0x01000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT			0x02000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE			0x03000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER			0x04000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP				0x05000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPU					(1 << 27)
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT				28
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MASK				0x70000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_REPEAT				0x10000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_MIRRORED_REPEAT			0x20000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE			0x30000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_BORDER			0x40000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP				0x50000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FORMAT_WRAPV					(1 << 31)
+#define  NV04_DX5_TEXTURED_TRIANGLE_FILTER						0x0000030c
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT				0
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK				0x000000ff
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT				8
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK				0x00007f00
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE			(1 << 15)
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT			16
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK				0x00ff0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_SHIFT				24
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_MASK					0x07000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST				0x01000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR				0x02000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST		0x03000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST		0x04000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR		0x05000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR		0x06000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE			(1 << 27)
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_SHIFT				28
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_MASK				0x70000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST				0x10000000
+#define    NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR				0x20000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE			(1 << 31)
+#define  NV04_DX5_TEXTURED_TRIANGLE_BLEND						0x00000310
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_SHIFT				0
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP_MASK				0x0000000f
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_SHIFT				4
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_MASK_BIT_MASK				0x00000030
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_SHIFT				6
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_MASK				0x000000c0
+#define    NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT				0x00000040
+#define    NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD				0x00000080
+#define    NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_PHONG				0x000000c0
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT		8
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK		0x00000f00
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT			12
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK				0x0000f000
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_SHIFT				16
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_FOG_ENABLE_MASK				0x000f0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT				20
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_ALPHA_ENABLE_MASK				0x00f00000
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT					24
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK					0x0f000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT					28
+#define   NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK					0xf0000000
+#define  NV04_DX5_TEXTURED_TRIANGLE_CONTROL						0x00000314
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_SHIFT				0
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_REF_MASK				0x000000ff
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT				8
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_MASK				0x00000f00
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE				(1 << 12)
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN					(1 << 13)
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT				14
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_MASK				0x0000c000
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT				16
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_MASK				0x000f0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT				20
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_MASK				0x00300000
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE				(1 << 22)
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE			(1 << 23)
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT			24
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_MASK			0x3f000000
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT				30
+#define   NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_MASK				0xc0000000
+#define  NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR						0x00000318
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_SHIFT					0
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_B_MASK					0x000000ff
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_SHIFT					8
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_G_MASK					0x0000ff00
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_SHIFT					16
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_R_MASK					0x00ff0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_SHIFT					24
+#define   NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR_A_MASK					0xff000000
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(x)					(0x00000400+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY(x)					(0x00000404+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ(x)					(0x00000408+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW(x)					(0x0000040c+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_RHW__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR(x)					(0x00000410+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR__SIZE				0x00000010
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_SHIFT				0
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_B_MASK				0x000000ff
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_SHIFT				8
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_G_MASK				0x0000ff00
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_SHIFT				16
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_R_MASK				0x00ff0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_SHIFT				24
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_COLOR_A_MASK				0xff000000
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR(x)				(0x00000414+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR__SIZE				0x00000010
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_SHIFT				0
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_B_MASK				0x000000ff
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_SHIFT				8
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_G_MASK				0x0000ff00
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_SHIFT				16
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_R_MASK				0x00ff0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_SHIFT			24
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SPECULAR_FOG_MASK				0xff000000
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU(x)					(0x00000418+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TU__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV(x)					(0x0000041c+((x)*32))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_TV__SIZE					0x00000010
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(x)				(0x00000600+((x)*4))
+#define  NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE__SIZE			0x00000040
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_SHIFT			0
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I0_MASK			0x0000000f
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_SHIFT			4
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I1_MASK			0x000000f0
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_SHIFT			8
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I2_MASK			0x00000f00
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_SHIFT			12
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I3_MASK			0x0000f000
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_SHIFT			16
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I4_MASK			0x000f0000
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_SHIFT			20
+#define   NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE_I5_MASK			0x00f00000
+
+
+#define NV04_DX6_MULTITEX_TRIANGLE							0x00000055
+
+#define  NV04_DX6_MULTITEX_TRIANGLE_NOP							0x00000100
+#define  NV04_DX6_MULTITEX_TRIANGLE_NOTIFY						0x00000104
+#define  NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY						0x00000180
+#define  NV04_DX6_MULTITEX_TRIANGLE_DMA_A						0x00000184
+#define  NV04_DX6_MULTITEX_TRIANGLE_DMA_B						0x00000188
+#define  NV04_DX6_MULTITEX_TRIANGLE_SURFACE						0x0000018c
+#define  NV04_DX6_MULTITEX_TRIANGLE_OFFSET(x)						(0x00000308+((x)*4))
+#define  NV04_DX6_MULTITEX_TRIANGLE_OFFSET__SIZE					0x00000002
+#define  NV04_DX6_MULTITEX_TRIANGLE_FORMAT(x)						(0x00000310+((x)*4))
+#define  NV04_DX6_MULTITEX_TRIANGLE_FORMAT__SIZE					0x00000002
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_SHIFT					0
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_DMA_MASK					0x0000000f
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_SHIFT				4
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_ZOH_MASK				0x00000030
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_SHIFT				6
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ORIGIN_FOH_MASK				0x000000c0
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_SHIFT					8
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_COLOR_MASK					0x00000f00
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT				12
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_MIPMAP_LEVELS_MASK				0x0000f000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT				16
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_U_MASK				0x000f0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT				20
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_BASE_SIZE_V_MASK				0x00f00000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_SHIFT				24
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSU_MASK				0x07000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPU					(1 << 27)
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_SHIFT				28
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_ADDRESSV_MASK				0x70000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FORMAT_WRAPV					(1 << 31)
+#define  NV04_DX6_MULTITEX_TRIANGLE_FILTER(x)						(0x00000318+((x)*4))
+#define  NV04_DX6_MULTITEX_TRIANGLE_FILTER__SIZE					0x00000002
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_SHIFT				0
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_X_MASK				0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_SHIFT				8
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_KERNEL_SIZE_Y_MASK				0x00007f00
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_DITHER_ENABLE			(1 << 15)
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_SHIFT			16
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MIPMAP_LODBIAS_MASK				0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_SHIFT				24
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MINIFY_MASK					0x07000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE			(1 << 27)
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_SHIFT				28
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_MAGNIFY_MASK				0x70000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE			(1 << 31)
+#define  NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA					0x00000320
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE0				(1 <<  0)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA0				(1 <<  1)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_SHIFT			2
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT0_MASK			0x000000fc
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE1				(1 <<  8)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA1				(1 <<  9)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_SHIFT			10
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT1_MASK			0x0000fc00
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE2				(1 << 16)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA2				(1 << 17)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_SHIFT			18
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT2_MASK			0x00fc0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_INVERSE3				(1 << 24)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ALPHA3				(1 << 25)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_SHIFT			26
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_ARGUMENT3_MASK			0x1c000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_SHIFT			29
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA_OPERATION_MASK			0xe0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR					0x00000324
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE0				(1 <<  0)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA0				(1 <<  1)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_SHIFT			2
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT0_MASK			0x000000fc
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE1				(1 <<  8)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA1				(1 <<  9)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_SHIFT			10
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT1_MASK			0x0000fc00
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE2				(1 << 16)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA2				(1 << 17)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_SHIFT			18
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT2_MASK			0x00fc0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_INVERSE3				(1 << 24)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ALPHA3				(1 << 25)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_SHIFT			26
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_ARGUMENT3_MASK			0x1c000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_SHIFT			29
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR_OPERATION_MASK			0xe0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA					0x0000032c
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE0				(1 <<  0)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA0				(1 <<  1)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_SHIFT			2
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT0_MASK			0x000000fc
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE1				(1 <<  8)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA1				(1 <<  9)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_SHIFT			10
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT1_MASK			0x0000fc00
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE2				(1 << 16)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA2				(1 << 17)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_SHIFT			18
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT2_MASK			0x00fc0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_INVERSE3				(1 << 24)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ALPHA3				(1 << 25)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_SHIFT			26
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_ARGUMENT3_MASK			0x1c000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_SHIFT			29
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA_OPERATION_MASK			0xe0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR					0x00000330
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE0				(1 <<  0)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA0				(1 <<  1)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_SHIFT			2
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT0_MASK			0x000000fc
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE1				(1 <<  8)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA1				(1 <<  9)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_SHIFT			10
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT1_MASK			0x0000fc00
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE2				(1 << 16)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA2				(1 << 17)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_SHIFT			18
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT2_MASK			0x00fc0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_INVERSE3				(1 << 24)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ALPHA3				(1 << 25)
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_SHIFT			26
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_ARGUMENT3_MASK			0x1c000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_SHIFT			29
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR_OPERATION_MASK			0xe0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR					0x00000334
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_SHIFT				0
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_B_MASK				0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_SHIFT				8
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_G_MASK				0x0000ff00
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_SHIFT				16
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_R_MASK				0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_SHIFT				24
+#define   NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR_A_MASK				0xff000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_BLEND						0x00000338
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_SHIFT				4
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_MASK_BIT_MASK				0x00000030
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_SHIFT				6
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SHADE_MODE_MASK				0x000000c0
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_SHIFT		8
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_TEXTURE_PERSPECTIVE_ENABLE_MASK		0x00000f00
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_SHIFT			12
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SPECULAR_ENABLE_MASK				0x0000f000
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_SHIFT				16
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_FOG_ENABLE_MASK				0x000f0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_SHIFT				20
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_ALPHA_ENABLE_MASK				0x00f00000
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT					24
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK					0x0f000000
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT					28
+#define   NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK					0xf0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_CONTROL0						0x0000033c
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_SHIFT				0
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_REF_MASK				0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_SHIFT				8
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_FUNC_MASK				0x00000f00
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_TEST_ENABLE				(1 << 12)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ORIGIN					(1 << 13)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_SHIFT				14
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_ENABLE_MASK				0x0000c000
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_SHIFT				16
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FUNC_MASK				0x000f0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_SHIFT				20
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULL_MODE_MASK				0x00300000
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_DITHER_ENABLE				(1 << 22)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_PERSPECTIVE_ENABLE			(1 << 23)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_WRITE_ENABLE				(1 << 24)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_ENABLE			(1 << 25)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_ENABLE			(1 << 26)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_ENABLE				(1 << 27)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_ENABLE			(1 << 28)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_ENABLE				(1 << 29)
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_SHIFT				30
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_Z_FORMAT_MASK				0xc0000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_CONTROL1						0x00000340
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_SHIFT			0
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_TEST_ENABLE_MASK			0x0000000f
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_SHIFT			4
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_FUNC_MASK				0x000000f0
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_SHIFT				8
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_REF_MASK				0x0000ff00
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT			16
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK			0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT			24
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK			0xff000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_CONTROL2						0x00000344
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_SHIFT			0
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_FAIL_MASK			0x0000000f
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_SHIFT			4
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZFAIL_MASK			0x000000f0
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_SHIFT			8
+#define   NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_OP_ZPASS_MASK			0x00000f00
+#define  NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR						0x00000348
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_SHIFT					0
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_B_MASK					0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_SHIFT					8
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_G_MASK					0x0000ff00
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_SHIFT					16
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_R_MASK					0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_SHIFT					24
+#define   NV04_DX6_MULTITEX_TRIANGLE_FOGCOLOR_A_MASK					0xff000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX(x)					(0x00000400+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SX__SIZE					0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY(x)					(0x00000404+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SY__SIZE					0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ(x)					(0x00000408+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SZ__SIZE					0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW(x)					(0x0000040c+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_RHW__SIZE				0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR(x)					(0x00000410+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR__SIZE				0x00000008
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_SHIFT				0
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_B_MASK				0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_SHIFT				8
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_G_MASK				0x0000ff00
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_SHIFT				16
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_R_MASK				0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_SHIFT				24
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_COLOR_A_MASK				0xff000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR(x)				(0x00000414+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR__SIZE				0x00000008
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_SHIFT			0
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_B_MASK				0x000000ff
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_SHIFT			8
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_G_MASK				0x0000ff00
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_SHIFT			16
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_R_MASK				0x00ff0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_SHIFT			24
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_SPECULAR_FOG_MASK			0xff000000
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0(x)					(0x00000418+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU0__SIZE				0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0(x)					(0x0000041c+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV0__SIZE				0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1(x)					(0x00000420+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TU1__SIZE				0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1(x)					(0x00000424+((x)*40))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_TV1__SIZE				0x00000008
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE(x)				(0x00000540+((x)*4))
+#define  NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE__SIZE			0x00000030
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_SHIFT			0
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I0_MASK			0x0000000f
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_SHIFT			4
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I1_MASK			0x000000f0
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_SHIFT			8
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I2_MASK			0x00000f00
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_SHIFT			12
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I3_MASK			0x0000f000
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_SHIFT			16
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I4_MASK			0x000f0000
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_SHIFT			20
+#define   NV04_DX6_MULTITEX_TRIANGLE_TLMTVERTEX_DRAWPRIMITIVE_I5_MASK			0x00f00000
+
+
+#define NV10_DX5_TEXTURED_TRIANGLE							0x00000094
+
+
+
+#define NV10TCL										0x00000056
+
+#define  NV10TCL_NOP									0x00000100
+#define  NV10TCL_NOTIFY									0x00000104
+#define  NV10TCL_DMA_NOTIFY								0x00000180
+#define  NV10TCL_DMA_IN_MEMORY0								0x00000184
+#define  NV10TCL_DMA_IN_MEMORY1								0x00000188
+#define  NV10TCL_DMA_VTXBUF0								0x0000018c
+#define  NV10TCL_DMA_IN_MEMORY2								0x00000194
+#define  NV10TCL_DMA_IN_MEMORY3								0x00000198
+#define  NV10TCL_RT_HORIZ								0x00000200
+#define   NV10TCL_RT_HORIZ_X_SHIFT							0
+#define   NV10TCL_RT_HORIZ_X_MASK							0x0000ffff
+#define   NV10TCL_RT_HORIZ_W_SHIFT							16
+#define   NV10TCL_RT_HORIZ_W_MASK							0xffff0000
+#define  NV10TCL_RT_VERT								0x00000204
+#define   NV10TCL_RT_VERT_Y_SHIFT							0
+#define   NV10TCL_RT_VERT_Y_MASK							0x0000ffff
+#define   NV10TCL_RT_VERT_H_SHIFT							16
+#define   NV10TCL_RT_VERT_H_MASK							0xffff0000
+#define  NV10TCL_RT_FORMAT								0x00000208
+#define   NV10TCL_RT_FORMAT_TYPE_SHIFT							8
+#define   NV10TCL_RT_FORMAT_TYPE_MASK							0x00000f00
+#define    NV10TCL_RT_FORMAT_TYPE_LINEAR						0x00000100
+#define    NV10TCL_RT_FORMAT_TYPE_SWIZZLED						0x00000200
+#define   NV10TCL_RT_FORMAT_COLOR_SHIFT							0
+#define   NV10TCL_RT_FORMAT_COLOR_MASK							0x0000001f
+#define    NV10TCL_RT_FORMAT_COLOR_R5G6B5						0x00000003
+#define    NV10TCL_RT_FORMAT_COLOR_X8R8G8B8						0x00000005
+#define    NV10TCL_RT_FORMAT_COLOR_A8R8G8B8						0x00000008
+#define    NV10TCL_RT_FORMAT_COLOR_B8							0x00000009
+#define    NV10TCL_RT_FORMAT_COLOR_UNKNOWN						0x0000000d
+#define    NV10TCL_RT_FORMAT_COLOR_X8B8G8R8						0x0000000f
+#define    NV10TCL_RT_FORMAT_COLOR_A8B8G8R8						0x00000010
+#define  NV10TCL_RT_PITCH								0x0000020c
+#define   NV10TCL_RT_PITCH_COLOR_PITCH_SHIFT						0
+#define   NV10TCL_RT_PITCH_COLOR_PITCH_MASK						0x0000ffff
+#define   NV10TCL_RT_PITCH_ZETA_PITCH_SHIFT						16
+#define   NV10TCL_RT_PITCH_ZETA_PITCH_MASK						0xffff0000
+#define  NV10TCL_COLOR_OFFSET								0x00000210
+#define  NV10TCL_ZETA_OFFSET								0x00000214
+#define  NV10TCL_TX_OFFSET(x)								(0x00000218+((x)*4))
+#define  NV10TCL_TX_OFFSET__SIZE							0x00000002
+#define  NV10TCL_TX_FORMAT(x)								(0x00000220+((x)*4))
+#define  NV10TCL_TX_FORMAT__SIZE							0x00000002
+#define   NV10TCL_TX_FORMAT_DMA0							(1 <<  0)
+#define   NV10TCL_TX_FORMAT_DMA1							(1 <<  1)
+#define   NV10TCL_TX_FORMAT_CUBE_MAP							(1 <<  2)
+#define   NV10TCL_TX_FORMAT_FORMAT_SHIFT						7
+#define   NV10TCL_TX_FORMAT_FORMAT_MASK							0x00000780
+#define    NV10TCL_TX_FORMAT_FORMAT_L8							0x00000000
+#define    NV10TCL_TX_FORMAT_FORMAT_A8							0x00000080
+#define    NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5						0x00000100
+#define    NV10TCL_TX_FORMAT_FORMAT_A8_RECT						0x00000180
+#define    NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4						0x00000200
+#define    NV10TCL_TX_FORMAT_FORMAT_R5G6B5						0x00000280
+#define    NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8						0x00000300
+#define    NV10TCL_TX_FORMAT_FORMAT_X8R8G8B8						0x00000380
+#define    NV10TCL_TX_FORMAT_FORMAT_INDEX8						0x00000580
+#define    NV10TCL_TX_FORMAT_FORMAT_DXT1						0x00000600
+#define    NV10TCL_TX_FORMAT_FORMAT_DXT3						0x00000700
+#define    NV10TCL_TX_FORMAT_FORMAT_DXT5						0x00000780
+#define    NV10TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT					0x00000800
+#define    NV10TCL_TX_FORMAT_FORMAT_R5G6B5_RECT						0x00000880
+#define    NV10TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT					0x00000900
+#define    NV10TCL_TX_FORMAT_FORMAT_L8_RECT						0x00000980
+#define    NV10TCL_TX_FORMAT_FORMAT_A8L8						0x00000d00
+#define    NV10TCL_TX_FORMAT_FORMAT_A8_RECT2						0x00000d80
+#define    NV10TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT					0x00000e80
+#define    NV10TCL_TX_FORMAT_FORMAT_R8G8B8_RECT						0x00000f00
+#define    NV10TCL_TX_FORMAT_FORMAT_L8A8_RECT						0x00001000
+#define    NV10TCL_TX_FORMAT_FORMAT_DSDT						0x00001400
+#define    NV10TCL_TX_FORMAT_FORMAT_A16							0x00001900
+#define    NV10TCL_TX_FORMAT_FORMAT_HILO16						0x00001980
+#define    NV10TCL_TX_FORMAT_FORMAT_A16_RECT						0x00001a80
+#define    NV10TCL_TX_FORMAT_FORMAT_HILO16_RECT						0x00001b00
+#define    NV10TCL_TX_FORMAT_FORMAT_HILO8						0x00002200
+#define    NV10TCL_TX_FORMAT_FORMAT_SIGNED_HILO8					0x00002280
+#define    NV10TCL_TX_FORMAT_FORMAT_HILO8_RECT						0x00002300
+#define    NV10TCL_TX_FORMAT_FORMAT_SIGNED_HILO8_RECT					0x00002380
+#define    NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV					0x00002500
+#define    NV10TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV					0x00002580
+#define    NV10TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV					0x00002600
+#define   NV10TCL_TX_FORMAT_NPOT							(1 << 11)
+#define   NV10TCL_TX_FORMAT_MIPMAP							(1 << 15)
+#define   NV10TCL_TX_FORMAT_BASE_SIZE_U_SHIFT						16
+#define   NV10TCL_TX_FORMAT_BASE_SIZE_U_MASK						0x000f0000
+#define   NV10TCL_TX_FORMAT_BASE_SIZE_V_SHIFT						20
+#define   NV10TCL_TX_FORMAT_BASE_SIZE_V_MASK						0x00f00000
+#define   NV10TCL_TX_FORMAT_WRAP_S_SHIFT						24
+#define   NV10TCL_TX_FORMAT_WRAP_S_MASK							0x0f000000
+#define    NV10TCL_TX_FORMAT_WRAP_S_REPEAT						0x01000000
+#define    NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT					0x02000000
+#define    NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE					0x03000000
+#define    NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER					0x04000000
+#define    NV10TCL_TX_FORMAT_WRAP_S_CLAMP						0x05000000
+#define   NV10TCL_TX_FORMAT_WRAP_T_SHIFT						28
+#define   NV10TCL_TX_FORMAT_WRAP_T_MASK							0xf0000000
+#define    NV10TCL_TX_FORMAT_WRAP_T_REPEAT						0x10000000
+#define    NV10TCL_TX_FORMAT_WRAP_T_MIRRORED_REPEAT					0x20000000
+#define    NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE					0x30000000
+#define    NV10TCL_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER					0x40000000
+#define    NV10TCL_TX_FORMAT_WRAP_T_CLAMP						0x50000000
+#define  NV10TCL_TX_ENABLE(x)								(0x00000228+((x)*4))
+#define  NV10TCL_TX_ENABLE__SIZE							0x00000002
+#define   NV10TCL_TX_ENABLE_ANISOTROPY_SHIFT						4
+#define   NV10TCL_TX_ENABLE_ANISOTROPY_MASK						0x00000030
+#define   NV10TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT					14
+#define   NV10TCL_TX_ENABLE_MIPMAP_MAX_LOD_MASK						0x0003c000
+#define   NV10TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT					26
+#define   NV10TCL_TX_ENABLE_MIPMAP_MIN_LOD_MASK						0x3c000000
+#define   NV10TCL_TX_ENABLE_ENABLE							(1 << 30)
+#define  NV10TCL_TX_NPOT_PITCH(x)							(0x00000230+((x)*4))
+#define  NV10TCL_TX_NPOT_PITCH__SIZE							0x00000002
+#define   NV10TCL_TX_NPOT_PITCH_PITCH_SHIFT						16
+#define   NV10TCL_TX_NPOT_PITCH_PITCH_MASK						0xffff0000
+#define  NV10TCL_TX_NPOT_SIZE(x)							(0x00000240+((x)*4))
+#define  NV10TCL_TX_NPOT_SIZE__SIZE							0x00000002
+#define   NV10TCL_TX_NPOT_SIZE_H_SHIFT							0
+#define   NV10TCL_TX_NPOT_SIZE_H_MASK							0x0000ffff
+#define   NV10TCL_TX_NPOT_SIZE_W_SHIFT							16
+#define   NV10TCL_TX_NPOT_SIZE_W_MASK							0xffff0000
+#define  NV10TCL_TX_FILTER(x)								(0x00000248+((x)*4))
+#define  NV10TCL_TX_FILTER__SIZE							0x00000002
+#define   NV10TCL_TX_FILTER_LOD_BIAS_SHIFT						8
+#define   NV10TCL_TX_FILTER_LOD_BIAS_MASK						0x00000f00
+#define   NV10TCL_TX_FILTER_MINIFY_SHIFT						24
+#define   NV10TCL_TX_FILTER_MINIFY_MASK							0x0f000000
+#define    NV10TCL_TX_FILTER_MINIFY_NEAREST						0x01000000
+#define    NV10TCL_TX_FILTER_MINIFY_LINEAR						0x02000000
+#define    NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST				0x03000000
+#define    NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST				0x04000000
+#define    NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR				0x05000000
+#define    NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR				0x06000000
+#define   NV10TCL_TX_FILTER_MAGNIFY_SHIFT						28
+#define   NV10TCL_TX_FILTER_MAGNIFY_MASK						0xf0000000
+#define    NV10TCL_TX_FILTER_MAGNIFY_NEAREST						0x10000000
+#define    NV10TCL_TX_FILTER_MAGNIFY_LINEAR						0x20000000
+#define  NV10TCL_TX_PALETTE_OFFSET(x)							(0x00000250+((x)*4))
+#define  NV10TCL_TX_PALETTE_OFFSET__SIZE						0x00000002
+#define  NV10TCL_RC_IN_ALPHA(x)								(0x00000260+((x)*4))
+#define  NV10TCL_RC_IN_ALPHA__SIZE							0x00000002
+#define   NV10TCL_RC_IN_ALPHA_D_INPUT_SHIFT						0
+#define   NV10TCL_RC_IN_ALPHA_D_INPUT_MASK						0x0000000f
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_FOG						0x00000003
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV					0x0000000c
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV					0x0000000d
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV10TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV10TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV10TCL_RC_IN_ALPHA_D_MAPPING_SHIFT						5
+#define   NV10TCL_RC_IN_ALPHA_D_MAPPING_MASK						0x000000e0
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV				0x00000040
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV				0x00000060
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV10TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV				0x000000e0
+#define   NV10TCL_RC_IN_ALPHA_C_INPUT_SHIFT						8
+#define   NV10TCL_RC_IN_ALPHA_C_INPUT_MASK						0x00000f00
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_FOG						0x00000300
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV					0x00000c00
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV					0x00000d00
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV10TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE						(1 << 12)
+#define    NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV10TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV10TCL_RC_IN_ALPHA_C_MAPPING_SHIFT						13
+#define   NV10TCL_RC_IN_ALPHA_C_MAPPING_MASK						0x0000e000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV				0x00004000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV				0x00006000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV10TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV				0x0000e000
+#define   NV10TCL_RC_IN_ALPHA_B_INPUT_SHIFT						16
+#define   NV10TCL_RC_IN_ALPHA_B_INPUT_MASK						0x000f0000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV				0x00010000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV				0x00020000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_FOG						0x00030000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV				0x00050000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV					0x000c0000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV					0x000d0000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV10TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE						(1 << 20)
+#define    NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV10TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV10TCL_RC_IN_ALPHA_B_MAPPING_SHIFT						21
+#define   NV10TCL_RC_IN_ALPHA_B_MAPPING_MASK						0x00e00000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV				0x00400000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV				0x00600000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV10TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV				0x00e00000
+#define   NV10TCL_RC_IN_ALPHA_A_INPUT_SHIFT						24
+#define   NV10TCL_RC_IN_ALPHA_A_INPUT_MASK						0x0f000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV				0x01000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV				0x02000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_FOG						0x03000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV				0x05000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV					0x0c000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV					0x0d000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV10TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE						(1 << 28)
+#define    NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV10TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV10TCL_RC_IN_ALPHA_A_MAPPING_SHIFT						29
+#define   NV10TCL_RC_IN_ALPHA_A_MAPPING_MASK						0xe0000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV				0x40000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV				0x60000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV10TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV				0xe0000000
+#define  NV10TCL_RC_IN_RGB(x)								(0x00000268+((x)*4))
+#define  NV10TCL_RC_IN_RGB__SIZE							0x00000002
+#define   NV10TCL_RC_IN_RGB_D_INPUT_SHIFT						0
+#define   NV10TCL_RC_IN_RGB_D_INPUT_MASK						0x0000000f
+#define    NV10TCL_RC_IN_RGB_D_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV10TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV10TCL_RC_IN_RGB_D_INPUT_FOG						0x00000003
+#define    NV10TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV10TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV10TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV10TCL_RC_IN_RGB_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV10TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV10TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV10TCL_RC_IN_RGB_D_MAPPING_SHIFT						5
+#define   NV10TCL_RC_IN_RGB_D_MAPPING_MASK						0x000000e0
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV10TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV10TCL_RC_IN_RGB_C_INPUT_SHIFT						8
+#define   NV10TCL_RC_IN_RGB_C_INPUT_MASK						0x00000f00
+#define    NV10TCL_RC_IN_RGB_C_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV10TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV10TCL_RC_IN_RGB_C_INPUT_FOG						0x00000300
+#define    NV10TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV10TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV10TCL_RC_IN_RGB_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV10TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV10TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE						(1 << 12)
+#define    NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV10TCL_RC_IN_RGB_C_MAPPING_SHIFT						13
+#define   NV10TCL_RC_IN_RGB_C_MAPPING_MASK						0x0000e000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV10TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV10TCL_RC_IN_RGB_B_INPUT_SHIFT						16
+#define   NV10TCL_RC_IN_RGB_B_INPUT_MASK						0x000f0000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_FOG						0x00030000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV10TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE						(1 << 20)
+#define    NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV10TCL_RC_IN_RGB_B_MAPPING_SHIFT						21
+#define   NV10TCL_RC_IN_RGB_B_MAPPING_MASK						0x00e00000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV10TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV10TCL_RC_IN_RGB_A_INPUT_SHIFT						24
+#define   NV10TCL_RC_IN_RGB_A_INPUT_MASK						0x0f000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_FOG						0x03000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV10TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE						(1 << 28)
+#define    NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV10TCL_RC_IN_RGB_A_MAPPING_SHIFT						29
+#define   NV10TCL_RC_IN_RGB_A_MAPPING_MASK						0xe0000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV10TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV10TCL_RC_COLOR(x)								(0x00000270+((x)*4))
+#define  NV10TCL_RC_COLOR__SIZE								0x00000002
+#define   NV10TCL_RC_COLOR_B_SHIFT							0
+#define   NV10TCL_RC_COLOR_B_MASK							0x000000ff
+#define   NV10TCL_RC_COLOR_G_SHIFT							8
+#define   NV10TCL_RC_COLOR_G_MASK							0x0000ff00
+#define   NV10TCL_RC_COLOR_R_SHIFT							16
+#define   NV10TCL_RC_COLOR_R_MASK							0x00ff0000
+#define   NV10TCL_RC_COLOR_A_SHIFT							24
+#define   NV10TCL_RC_COLOR_A_MASK							0xff000000
+#define  NV10TCL_RC_OUT_ALPHA(x)							(0x00000278+((x)*4))
+#define  NV10TCL_RC_OUT_ALPHA__SIZE							0x00000002
+#define   NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT						0
+#define   NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK						0x0000000f
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG						0x00000003
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x0000000e
+#define    NV10TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT						4
+#define   NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK						0x000000f0
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG						0x00000030
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x000000e0
+#define    NV10TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT						8
+#define   NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK						0x00000f00
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG						0x00000300
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x00000e00
+#define    NV10TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT						(1 << 12)
+#define   NV10TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT						(1 << 13)
+#define   NV10TCL_RC_OUT_ALPHA_MUX_SUM							(1 << 14)
+#define   NV10TCL_RC_OUT_ALPHA_BIAS							(1 << 15)
+#define    NV10TCL_RC_OUT_ALPHA_BIAS_NONE						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV			0x00008000
+#define   NV10TCL_RC_OUT_ALPHA_SCALE_SHIFT						17
+#define   NV10TCL_RC_OUT_ALPHA_SCALE_MASK						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_SCALE_NONE						0x00000000
+#define    NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV10TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define  NV10TCL_RC_OUT_RGB(x)								(0x00000280+((x)*4))
+#define  NV10TCL_RC_OUT_RGB__SIZE							0x00000002
+#define   NV10TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT						0
+#define   NV10TCL_RC_OUT_RGB_CD_OUTPUT_MASK						0x0000000f
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_FOG						0x00000003
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV10TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV10TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT						4
+#define   NV10TCL_RC_OUT_RGB_AB_OUTPUT_MASK						0x000000f0
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_FOG						0x00000030
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000000e0
+#define    NV10TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT						8
+#define   NV10TCL_RC_OUT_RGB_SUM_OUTPUT_MASK						0x00000f00
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO						0x00000000
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_FOG						0x00000300
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV10TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_OUT_RGB_CD_DOT_PRODUCT						(1 << 12)
+#define   NV10TCL_RC_OUT_RGB_AB_DOT_PRODUCT						(1 << 13)
+#define   NV10TCL_RC_OUT_RGB_MUX_SUM							(1 << 14)
+#define   NV10TCL_RC_OUT_RGB_BIAS							(1 << 15)
+#define    NV10TCL_RC_OUT_RGB_BIAS_NONE							0x00000000
+#define    NV10TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV				0x00008000
+#define   NV10TCL_RC_OUT_RGB_SCALE_SHIFT						17
+#define   NV10TCL_RC_OUT_RGB_SCALE_MASK							0x00000000
+#define    NV10TCL_RC_OUT_RGB_SCALE_NONE						0x00000000
+#define    NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV10TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define   NV10TCL_RC_OUT_RGB_OPERATION_SHIFT						27
+#define   NV10TCL_RC_OUT_RGB_OPERATION_MASK						0x38000000
+#define  NV10TCL_RC_FINAL0								0x00000288
+#define   NV10TCL_RC_FINAL0_D_INPUT_SHIFT						0
+#define   NV10TCL_RC_FINAL0_D_INPUT_MASK						0x0000000f
+#define    NV10TCL_RC_FINAL0_D_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV10TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV10TCL_RC_FINAL0_D_INPUT_FOG						0x00000003
+#define    NV10TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV10TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV10TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV10TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV10TCL_RC_FINAL0_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV10TCL_RC_FINAL0_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV10TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV10TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV10TCL_RC_FINAL0_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV10TCL_RC_FINAL0_D_MAPPING_SHIFT						5
+#define   NV10TCL_RC_FINAL0_D_MAPPING_MASK						0x000000e0
+#define    NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV10TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV10TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV10TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV10TCL_RC_FINAL0_C_INPUT_SHIFT						8
+#define   NV10TCL_RC_FINAL0_C_INPUT_MASK						0x00000f00
+#define    NV10TCL_RC_FINAL0_C_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV10TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV10TCL_RC_FINAL0_C_INPUT_FOG						0x00000300
+#define    NV10TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV10TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV10TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_FINAL0_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV10TCL_RC_FINAL0_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV10TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV10TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_FINAL0_C_COMPONENT_USAGE						(1 << 12)
+#define    NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV10TCL_RC_FINAL0_C_MAPPING_SHIFT						13
+#define   NV10TCL_RC_FINAL0_C_MAPPING_MASK						0x0000e000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV10TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV10TCL_RC_FINAL0_B_INPUT_SHIFT						16
+#define   NV10TCL_RC_FINAL0_B_INPUT_MASK						0x000f0000
+#define    NV10TCL_RC_FINAL0_B_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV10TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV10TCL_RC_FINAL0_B_INPUT_FOG						0x00030000
+#define    NV10TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV10TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV10TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV10TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV10TCL_RC_FINAL0_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV10TCL_RC_FINAL0_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV10TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV10TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV10TCL_RC_FINAL0_B_COMPONENT_USAGE						(1 << 20)
+#define    NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV10TCL_RC_FINAL0_B_MAPPING_SHIFT						21
+#define   NV10TCL_RC_FINAL0_B_MAPPING_MASK						0x00e00000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV10TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV10TCL_RC_FINAL0_A_INPUT_SHIFT						24
+#define   NV10TCL_RC_FINAL0_A_INPUT_MASK						0x0f000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_FOG						0x03000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV10TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV10TCL_RC_FINAL0_A_COMPONENT_USAGE						(1 << 28)
+#define    NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV10TCL_RC_FINAL0_A_MAPPING_SHIFT						29
+#define   NV10TCL_RC_FINAL0_A_MAPPING_MASK						0xe0000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV10TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV10TCL_RC_FINAL1								0x0000028c
+#define   NV10TCL_RC_FINAL1_COLOR_SUM_CLAMP						(1 <<  7)
+#define   NV10TCL_RC_FINAL1_G_INPUT_SHIFT						8
+#define   NV10TCL_RC_FINAL1_G_INPUT_MASK						0x00000f00
+#define    NV10TCL_RC_FINAL1_G_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV10TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV10TCL_RC_FINAL1_G_INPUT_FOG						0x00000300
+#define    NV10TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV10TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV10TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV10TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV10TCL_RC_FINAL1_G_INPUT_SPARE0_NV						0x00000c00
+#define    NV10TCL_RC_FINAL1_G_INPUT_SPARE1_NV						0x00000d00
+#define    NV10TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV10TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV10TCL_RC_FINAL1_G_COMPONENT_USAGE						(1 << 12)
+#define    NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV10TCL_RC_FINAL1_G_MAPPING_SHIFT						13
+#define   NV10TCL_RC_FINAL1_G_MAPPING_MASK						0x0000e000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV10TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV10TCL_RC_FINAL1_F_INPUT_SHIFT						16
+#define   NV10TCL_RC_FINAL1_F_INPUT_MASK						0x000f0000
+#define    NV10TCL_RC_FINAL1_F_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV10TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV10TCL_RC_FINAL1_F_INPUT_FOG						0x00030000
+#define    NV10TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV10TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV10TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV10TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV10TCL_RC_FINAL1_F_INPUT_SPARE0_NV						0x000c0000
+#define    NV10TCL_RC_FINAL1_F_INPUT_SPARE1_NV						0x000d0000
+#define    NV10TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV10TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV10TCL_RC_FINAL1_F_COMPONENT_USAGE						(1 << 20)
+#define    NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV10TCL_RC_FINAL1_F_MAPPING_SHIFT						21
+#define   NV10TCL_RC_FINAL1_F_MAPPING_MASK						0x00e00000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV10TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV10TCL_RC_FINAL1_E_INPUT_SHIFT						24
+#define   NV10TCL_RC_FINAL1_E_INPUT_MASK						0x0f000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_ZERO						0x00000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_FOG						0x03000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_SPARE0_NV						0x0c000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_SPARE1_NV						0x0d000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV10TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV10TCL_RC_FINAL1_E_COMPONENT_USAGE						(1 << 28)
+#define    NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB					0x00000000
+#define    NV10TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV10TCL_RC_FINAL1_E_MAPPING_SHIFT						29
+#define   NV10TCL_RC_FINAL1_E_MAPPING_MASK						0xe0000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV10TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV10TCL_LIGHT_MODEL								0x00000294
+#define   NV10TCL_LIGHT_MODEL_COLOR_CONTROL						(1 <<  1)
+#define   NV10TCL_LIGHT_MODEL_LOCAL_VIEWER						(1 << 16)
+#define  NV10TCL_COLOR_MATERIAL_ENABLE							0x00000298
+#define   NV10TCL_COLOR_MATERIAL_ENABLE_SPECULAR					(1 <<  0)
+#define   NV10TCL_COLOR_MATERIAL_ENABLE_DIFFUSE						(1 <<  1)
+#define   NV10TCL_COLOR_MATERIAL_ENABLE_AMBIENT						(1 <<  2)
+#define   NV10TCL_COLOR_MATERIAL_ENABLE_EMISSION					(1 <<  3)
+#define  NV10TCL_FOG_MODE								0x0000029c
+#define   NV10TCL_FOG_MODE_EXP								0x00000800
+#define   NV10TCL_FOG_MODE_EXP_2							0x00000802
+#define   NV10TCL_FOG_MODE_EXP2								0x00000803
+#define   NV10TCL_FOG_MODE_LINEAR							0x00000804
+#define   NV10TCL_FOG_MODE_LINEAR_2							0x00002601
+#define  NV10TCL_FOG_COORD_DIST								0x000002a0
+#define   NV10TCL_FOG_COORD_DIST_COORD_FALSE						0x00000000
+#define   NV10TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_RADIAL_NV		0x00000001
+#define   NV10TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_PLANE_ABSOLUTE_NV	0x00000002
+#define   NV10TCL_FOG_COORD_DIST_COORD_FOG						0x00000003
+#define  NV10TCL_FOG_ENABLE								0x000002a4
+#define  NV10TCL_FOG_COLOR								0x000002a8
+#define   NV10TCL_FOG_COLOR_R_SHIFT							0
+#define   NV10TCL_FOG_COLOR_R_MASK							0x000000ff
+#define   NV10TCL_FOG_COLOR_G_SHIFT							8
+#define   NV10TCL_FOG_COLOR_G_MASK							0x0000ff00
+#define   NV10TCL_FOG_COLOR_B_SHIFT							16
+#define   NV10TCL_FOG_COLOR_B_MASK							0x00ff0000
+#define   NV10TCL_FOG_COLOR_A_SHIFT							24
+#define   NV10TCL_FOG_COLOR_A_MASK							0xff000000
+#define  NV10TCL_VIEWPORT_CLIP_MODE							0x000002b4
+#define  NV10TCL_VIEWPORT_CLIP_HORIZ(x)							(0x000002c0+((x)*4))
+#define  NV10TCL_VIEWPORT_CLIP_HORIZ__SIZE						0x00000008
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_SHIFT					0
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_L_MASK					0x000007ff
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_LEFT_ENABLE					(1 << 11)
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_SHIFT					16
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_R_MASK					0x07ff0000
+#define   NV10TCL_VIEWPORT_CLIP_HORIZ_CLIP_RIGHT_ENABLE					(1 << 27)
+#define  NV10TCL_VIEWPORT_CLIP_VERT(x)							(0x000002e0+((x)*4))
+#define  NV10TCL_VIEWPORT_CLIP_VERT__SIZE						0x00000008
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_SHIFT					0
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_T_MASK					0x000007ff
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_TOP_ENABLE					(1 << 11)
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_SHIFT					16
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_B_MASK					0x07ff0000
+#define   NV10TCL_VIEWPORT_CLIP_VERT_CLIP_BOTTOM_ENABLE					(1 << 27)
+#define  NV10TCL_ALPHA_FUNC_ENABLE							0x00000300
+#define  NV10TCL_BLEND_FUNC_ENABLE							0x00000304
+#define  NV10TCL_CULL_FACE_ENABLE							0x00000308
+#define  NV10TCL_DEPTH_TEST_ENABLE							0x0000030c
+#define  NV10TCL_DITHER_ENABLE								0x00000310
+#define  NV10TCL_LIGHTING_ENABLE							0x00000314
+#define  NV10TCL_POINT_PARAMETERS_ENABLE						0x00000318
+#define  NV10TCL_POINT_SMOOTH_ENABLE							0x0000031c
+#define  NV10TCL_LINE_SMOOTH_ENABLE							0x00000320
+#define  NV10TCL_POLYGON_SMOOTH_ENABLE							0x00000324
+#define  NV10TCL_VERTEX_WEIGHT_ENABLE							0x00000328
+#define  NV10TCL_STENCIL_ENABLE								0x0000032c
+#define  NV10TCL_POLYGON_OFFSET_POINT_ENABLE						0x00000330
+#define  NV10TCL_POLYGON_OFFSET_LINE_ENABLE						0x00000334
+#define  NV10TCL_POLYGON_OFFSET_FILL_ENABLE						0x00000338
+#define  NV10TCL_ALPHA_FUNC_FUNC							0x0000033c
+#define   NV10TCL_ALPHA_FUNC_FUNC_NEVER							0x00000200
+#define   NV10TCL_ALPHA_FUNC_FUNC_LESS							0x00000201
+#define   NV10TCL_ALPHA_FUNC_FUNC_EQUAL							0x00000202
+#define   NV10TCL_ALPHA_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV10TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV10TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV10TCL_ALPHA_FUNC_FUNC_NOTEQUAL						0x00000205
+#define   NV10TCL_ALPHA_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV10TCL_ALPHA_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV10TCL_ALPHA_FUNC_REF								0x00000340
+#define  NV10TCL_BLEND_FUNC_SRC								0x00000344
+#define   NV10TCL_BLEND_FUNC_SRC_ZERO							0x00000000
+#define   NV10TCL_BLEND_FUNC_SRC_ONE							0x00000001
+#define   NV10TCL_BLEND_FUNC_SRC_SRC_COLOR						0x00000300
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR					0x00000301
+#define   NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA						0x00000302
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA					0x00000303
+#define   NV10TCL_BLEND_FUNC_SRC_DST_ALPHA						0x00000304
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA					0x00000305
+#define   NV10TCL_BLEND_FUNC_SRC_DST_COLOR						0x00000306
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR					0x00000307
+#define   NV10TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE					0x00000308
+#define   NV10TCL_BLEND_FUNC_SRC_CONSTANT_COLOR						0x00008001
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV10TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA						0x00008003
+#define   NV10TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV10TCL_BLEND_FUNC_DST								0x00000348
+#define   NV10TCL_BLEND_FUNC_DST_ZERO							0x00000000
+#define   NV10TCL_BLEND_FUNC_DST_ONE							0x00000001
+#define   NV10TCL_BLEND_FUNC_DST_SRC_COLOR						0x00000300
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR					0x00000301
+#define   NV10TCL_BLEND_FUNC_DST_SRC_ALPHA						0x00000302
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA					0x00000303
+#define   NV10TCL_BLEND_FUNC_DST_DST_ALPHA						0x00000304
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA					0x00000305
+#define   NV10TCL_BLEND_FUNC_DST_DST_COLOR						0x00000306
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR					0x00000307
+#define   NV10TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE					0x00000308
+#define   NV10TCL_BLEND_FUNC_DST_CONSTANT_COLOR						0x00008001
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV10TCL_BLEND_FUNC_DST_CONSTANT_ALPHA						0x00008003
+#define   NV10TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV10TCL_BLEND_COLOR								0x0000034c
+#define   NV10TCL_BLEND_COLOR_B_SHIFT							0
+#define   NV10TCL_BLEND_COLOR_B_MASK							0x000000ff
+#define   NV10TCL_BLEND_COLOR_G_SHIFT							8
+#define   NV10TCL_BLEND_COLOR_G_MASK							0x0000ff00
+#define   NV10TCL_BLEND_COLOR_R_SHIFT							16
+#define   NV10TCL_BLEND_COLOR_R_MASK							0x00ff0000
+#define   NV10TCL_BLEND_COLOR_A_SHIFT							24
+#define   NV10TCL_BLEND_COLOR_A_MASK							0xff000000
+#define  NV10TCL_BLEND_EQUATION								0x00000350
+#define   NV10TCL_BLEND_EQUATION_FUNC_ADD						0x00008006
+#define   NV10TCL_BLEND_EQUATION_MIN							0x00008007
+#define   NV10TCL_BLEND_EQUATION_MAX							0x00008008
+#define   NV10TCL_BLEND_EQUATION_FUNC_SUBTRACT						0x0000800a
+#define   NV10TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT					0x0000800b
+#define  NV10TCL_DEPTH_FUNC								0x00000354
+#define   NV10TCL_DEPTH_FUNC_NEVER							0x00000200
+#define   NV10TCL_DEPTH_FUNC_LESS							0x00000201
+#define   NV10TCL_DEPTH_FUNC_EQUAL							0x00000202
+#define   NV10TCL_DEPTH_FUNC_LEQUAL							0x00000203
+#define   NV10TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV10TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV10TCL_DEPTH_FUNC_NOTEQUAL							0x00000205
+#define   NV10TCL_DEPTH_FUNC_GEQUAL							0x00000206
+#define   NV10TCL_DEPTH_FUNC_ALWAYS							0x00000207
+#define  NV10TCL_COLOR_MASK								0x00000358
+#define   NV10TCL_COLOR_MASK_B								(1 <<  0)
+#define   NV10TCL_COLOR_MASK_G								(1 <<  8)
+#define   NV10TCL_COLOR_MASK_R								(1 << 16)
+#define   NV10TCL_COLOR_MASK_A								(1 << 24)
+#define  NV10TCL_DEPTH_WRITE_ENABLE							0x0000035c
+#define  NV10TCL_STENCIL_MASK								0x00000360
+#define  NV10TCL_STENCIL_FUNC_FUNC							0x00000364
+#define   NV10TCL_STENCIL_FUNC_FUNC_NEVER						0x00000200
+#define   NV10TCL_STENCIL_FUNC_FUNC_LESS						0x00000201
+#define   NV10TCL_STENCIL_FUNC_FUNC_EQUAL						0x00000202
+#define   NV10TCL_STENCIL_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV10TCL_STENCIL_FUNC_FUNC_GREATER						0x00000204
+#define   NV10TCL_STENCIL_FUNC_FUNC_GREATER						0x00000204
+#define   NV10TCL_STENCIL_FUNC_FUNC_NOTEQUAL						0x00000205
+#define   NV10TCL_STENCIL_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV10TCL_STENCIL_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV10TCL_STENCIL_FUNC_REF							0x00000368
+#define  NV10TCL_STENCIL_FUNC_MASK							0x0000036c
+#define  NV10TCL_STENCIL_OP_FAIL							0x00000370
+#define   NV10TCL_STENCIL_OP_FAIL_ZERO							0x00000000
+#define   NV10TCL_STENCIL_OP_FAIL_INVERT						0x0000150a
+#define   NV10TCL_STENCIL_OP_FAIL_KEEP							0x00001e00
+#define   NV10TCL_STENCIL_OP_FAIL_REPLACE						0x00001e01
+#define   NV10TCL_STENCIL_OP_FAIL_INCR							0x00001e02
+#define   NV10TCL_STENCIL_OP_FAIL_DECR							0x00001e03
+#define   NV10TCL_STENCIL_OP_FAIL_INCR_WRAP						0x00008507
+#define   NV10TCL_STENCIL_OP_FAIL_DECR_WRAP						0x00008508
+#define  NV10TCL_STENCIL_OP_ZFAIL							0x00000374
+#define   NV10TCL_STENCIL_OP_ZFAIL_ZERO							0x00000000
+#define   NV10TCL_STENCIL_OP_ZFAIL_INVERT						0x0000150a
+#define   NV10TCL_STENCIL_OP_ZFAIL_KEEP							0x00001e00
+#define   NV10TCL_STENCIL_OP_ZFAIL_REPLACE						0x00001e01
+#define   NV10TCL_STENCIL_OP_ZFAIL_INCR							0x00001e02
+#define   NV10TCL_STENCIL_OP_ZFAIL_DECR							0x00001e03
+#define   NV10TCL_STENCIL_OP_ZFAIL_INCR_WRAP						0x00008507
+#define   NV10TCL_STENCIL_OP_ZFAIL_DECR_WRAP						0x00008508
+#define  NV10TCL_STENCIL_OP_ZPASS							0x00000378
+#define   NV10TCL_STENCIL_OP_ZPASS_ZERO							0x00000000
+#define   NV10TCL_STENCIL_OP_ZPASS_INVERT						0x0000150a
+#define   NV10TCL_STENCIL_OP_ZPASS_KEEP							0x00001e00
+#define   NV10TCL_STENCIL_OP_ZPASS_REPLACE						0x00001e01
+#define   NV10TCL_STENCIL_OP_ZPASS_INCR							0x00001e02
+#define   NV10TCL_STENCIL_OP_ZPASS_DECR							0x00001e03
+#define   NV10TCL_STENCIL_OP_ZPASS_INCR_WRAP						0x00008507
+#define   NV10TCL_STENCIL_OP_ZPASS_DECR_WRAP						0x00008508
+#define  NV10TCL_SHADE_MODEL								0x0000037c
+#define   NV10TCL_SHADE_MODEL_FLAT							0x00001d00
+#define   NV10TCL_SHADE_MODEL_SMOOTH							0x00001d01
+#define  NV10TCL_LINE_WIDTH								0x00000380
+#define  NV10TCL_POLYGON_OFFSET_FACTOR							0x00000384
+#define  NV10TCL_POLYGON_OFFSET_UNITS							0x00000388
+#define  NV10TCL_POLYGON_MODE_FRONT							0x0000038c
+#define   NV10TCL_POLYGON_MODE_FRONT_POINT						0x00001b00
+#define   NV10TCL_POLYGON_MODE_FRONT_LINE						0x00001b01
+#define   NV10TCL_POLYGON_MODE_FRONT_FILL						0x00001b02
+#define  NV10TCL_POLYGON_MODE_BACK							0x00000390
+#define   NV10TCL_POLYGON_MODE_BACK_POINT						0x00001b00
+#define   NV10TCL_POLYGON_MODE_BACK_LINE						0x00001b01
+#define   NV10TCL_POLYGON_MODE_BACK_FILL						0x00001b02
+#define  NV10TCL_DEPTH_RANGE_NEAR							0x00000394
+#define  NV10TCL_DEPTH_RANGE_FAR							0x00000398
+#define  NV10TCL_CULL_FACE								0x0000039c
+#define   NV10TCL_CULL_FACE_FRONT							0x00000404
+#define   NV10TCL_CULL_FACE_BACK							0x00000405
+#define   NV10TCL_CULL_FACE_FRONT_AND_BACK						0x00000408
+#define  NV10TCL_FRONT_FACE								0x000003a0
+#define   NV10TCL_FRONT_FACE_CW								0x00000900
+#define   NV10TCL_FRONT_FACE_CCW							0x00000901
+#define  NV10TCL_NORMALIZE_ENABLE							0x000003a4
+#define  NV10TCL_COLOR_MATERIAL_R							0x000003a8
+#define  NV10TCL_COLOR_MATERIAL_G							0x000003ac
+#define  NV10TCL_COLOR_MATERIAL_B							0x000003b0
+#define  NV10TCL_COLOR_MATERIAL_A							0x000003b4
+#define  NV10TCL_COLOR_CONTROL								0x000003b8
+#define  NV10TCL_ENABLED_LIGHTS								0x000003bc
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT0							(1 <<  0)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT1							(1 <<  2)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT2							(1 <<  4)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT3							(1 <<  6)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT4							(1 <<  8)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT5							(1 << 10)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT6							(1 << 12)
+#define   NV10TCL_ENABLED_LIGHTS_LIGHT7							(1 << 14)
+#define  NV10TCL_TX_GEN_S(x)								(0x000003c0+((x)*16))
+#define  NV10TCL_TX_GEN_S__SIZE								0x00000002
+#define   NV10TCL_TX_GEN_S_FALSE							0x00000000
+#define   NV10TCL_TX_GEN_S_EYE_LINEAR							0x00002400
+#define   NV10TCL_TX_GEN_S_OBJECT_LINEAR						0x00002401
+#define   NV10TCL_TX_GEN_S_SPHERE_MAP							0x00002402
+#define   NV10TCL_TX_GEN_S_NORMAL_MAP							0x00008511
+#define   NV10TCL_TX_GEN_S_REFLECTION_MAP						0x00008512
+#define  NV10TCL_TX_GEN_T(x)								(0x000003c4+((x)*16))
+#define  NV10TCL_TX_GEN_T__SIZE								0x00000002
+#define   NV10TCL_TX_GEN_T_FALSE							0x00000000
+#define   NV10TCL_TX_GEN_T_EYE_LINEAR							0x00002400
+#define   NV10TCL_TX_GEN_T_OBJECT_LINEAR						0x00002401
+#define   NV10TCL_TX_GEN_T_SPHERE_MAP							0x00002402
+#define   NV10TCL_TX_GEN_T_NORMAL_MAP							0x00008511
+#define   NV10TCL_TX_GEN_T_REFLECTION_MAP						0x00008512
+#define  NV10TCL_TX_GEN_R(x)								(0x000003c8+((x)*16))
+#define  NV10TCL_TX_GEN_R__SIZE								0x00000002
+#define   NV10TCL_TX_GEN_R_FALSE							0x00000000
+#define   NV10TCL_TX_GEN_R_EYE_LINEAR							0x00002400
+#define   NV10TCL_TX_GEN_R_OBJECT_LINEAR						0x00002401
+#define   NV10TCL_TX_GEN_R_SPHERE_MAP							0x00002402
+#define   NV10TCL_TX_GEN_R_NORMAL_MAP							0x00008511
+#define   NV10TCL_TX_GEN_R_REFLECTION_MAP						0x00008512
+#define  NV10TCL_TX_GEN_Q(x)								(0x000003cc+((x)*16))
+#define  NV10TCL_TX_GEN_Q__SIZE								0x00000002
+#define   NV10TCL_TX_GEN_Q_FALSE							0x00000000
+#define   NV10TCL_TX_GEN_Q_EYE_LINEAR							0x00002400
+#define   NV10TCL_TX_GEN_Q_OBJECT_LINEAR						0x00002401
+#define   NV10TCL_TX_GEN_Q_SPHERE_MAP							0x00002402
+#define   NV10TCL_TX_GEN_Q_NORMAL_MAP							0x00008511
+#define   NV10TCL_TX_GEN_Q_REFLECTION_MAP						0x00008512
+#define  NV10TCL_TX_MATRIX_ENABLE(x)							(0x000003e0+((x)*4))
+#define  NV10TCL_TX_MATRIX_ENABLE__SIZE							0x00000002
+#define  NV10TCL_VIEW_MATRIX_ENABLE							0x000003e8
+#define   NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW1						(1 <<  0)
+#define   NV10TCL_VIEW_MATRIX_ENABLE_MODELVIEW0						(1 <<  1)
+#define   NV10TCL_VIEW_MATRIX_ENABLE_PROJECTION						(1 <<  2)
+#define  NV10TCL_POINT_SIZE								0x000003ec
+#define  NV10TCL_MODELVIEW0_MATRIX(x)							(0x00000400+((x)*4))
+#define  NV10TCL_MODELVIEW0_MATRIX__SIZE						0x00000010
+#define  NV10TCL_MODELVIEW1_MATRIX(x)							(0x00000440+((x)*4))
+#define  NV10TCL_MODELVIEW1_MATRIX__SIZE						0x00000010
+#define  NV10TCL_INVERSE_MODELVIEW0_MATRIX(x)						(0x00000480+((x)*4))
+#define  NV10TCL_INVERSE_MODELVIEW0_MATRIX__SIZE					0x00000010
+#define  NV10TCL_INVERSE_MODELVIEW1_MATRIX(x)						(0x000004c0+((x)*4))
+#define  NV10TCL_INVERSE_MODELVIEW1_MATRIX__SIZE					0x00000010
+#define  NV10TCL_PROJECTION_MATRIX(x)							(0x00000500+((x)*4))
+#define  NV10TCL_PROJECTION_MATRIX__SIZE						0x00000010
+#define  NV10TCL_TX0_MATRIX(x)								(0x00000540+((x)*4))
+#define  NV10TCL_TX0_MATRIX__SIZE							0x00000010
+#define  NV10TCL_TX1_MATRIX(x)								(0x00000580+((x)*4))
+#define  NV10TCL_TX1_MATRIX__SIZE							0x00000010
+#define  NV10TCL_CLIP_PLANE_A(x)							(0x00000600+((x)*16))
+#define  NV10TCL_CLIP_PLANE_A__SIZE							0x00000008
+#define  NV10TCL_CLIP_PLANE_B(x)							(0x00000604+((x)*16))
+#define  NV10TCL_CLIP_PLANE_B__SIZE							0x00000008
+#define  NV10TCL_CLIP_PLANE_C(x)							(0x00000608+((x)*16))
+#define  NV10TCL_CLIP_PLANE_C__SIZE							0x00000008
+#define  NV10TCL_CLIP_PLANE_D(x)							(0x0000060c+((x)*16))
+#define  NV10TCL_CLIP_PLANE_D__SIZE							0x00000008
+#define  NV10TCL_FOG_EQUATION_CONSTANT							0x00000680
+#define  NV10TCL_FOG_EQUATION_LINEAR							0x00000684
+#define  NV10TCL_FOG_EQUATION_QUADRATIC							0x00000688
+#define  NV10TCL_FRONT_MATERIAL_SHININESS(x)						(0x000006a0+((x)*4))
+#define  NV10TCL_FRONT_MATERIAL_SHININESS__SIZE						0x00000006
+#define  NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R			0x000006c4
+#define  NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G			0x000006c8
+#define  NV10TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B			0x000006cc
+#define  NV10TCL_VIEWPORT_SCALE_X							0x000006e8
+#define  NV10TCL_VIEWPORT_SCALE_Y							0x000006ec
+#define  NV10TCL_VIEWPORT_SCALE_Z							0x000006f0
+#define  NV10TCL_VIEWPORT_SCALE_W							0x000006f4
+#define  NV10TCL_POINT_PARAMETER(x)							(0x000006f8+((x)*4))
+#define  NV10TCL_POINT_PARAMETER__SIZE							0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x)					(0x00000800+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x)					(0x00000804+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x)					(0x00000808+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x)					(0x0000080c+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x)					(0x00000810+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x)					(0x00000814+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x)					(0x00000818+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x)					(0x0000081c+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE				0x00000008
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x)					(0x00000820+((x)*128))
+#define  NV10TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE				0x00000008
+#define  NV10TCL_LIGHT_HALF_VECTOR_X(x)							(0x00000828+((x)*128))
+#define  NV10TCL_LIGHT_HALF_VECTOR_X__SIZE						0x00000008
+#define  NV10TCL_LIGHT_HALF_VECTOR_Y(x)							(0x0000082c+((x)*128))
+#define  NV10TCL_LIGHT_HALF_VECTOR_Y__SIZE						0x00000008
+#define  NV10TCL_LIGHT_HALF_VECTOR_Z(x)							(0x00000830+((x)*128))
+#define  NV10TCL_LIGHT_HALF_VECTOR_Z__SIZE						0x00000008
+#define  NV10TCL_LIGHT_DIRECTION_X(x)							(0x00000834+((x)*128))
+#define  NV10TCL_LIGHT_DIRECTION_X__SIZE						0x00000008
+#define  NV10TCL_LIGHT_DIRECTION_Y(x)							(0x00000838+((x)*128))
+#define  NV10TCL_LIGHT_DIRECTION_Y__SIZE						0x00000008
+#define  NV10TCL_LIGHT_DIRECTION_Z(x)							(0x0000083c+((x)*128))
+#define  NV10TCL_LIGHT_DIRECTION_Z__SIZE						0x00000008
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_A(x)							(0x00000840+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_A__SIZE						0x00000008
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_B(x)							(0x00000844+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_B__SIZE						0x00000008
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_C(x)							(0x00000848+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_C__SIZE						0x00000008
+#define  NV10TCL_LIGHT_SPOT_DIR_X(x)							(0x0000084c+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_DIR_X__SIZE							0x00000008
+#define  NV10TCL_LIGHT_SPOT_DIR_Y(x)							(0x00000850+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_DIR_Y__SIZE							0x00000008
+#define  NV10TCL_LIGHT_SPOT_DIR_Z(x)							(0x00000854+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_DIR_Z__SIZE							0x00000008
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_D(x)							(0x00000858+((x)*128))
+#define  NV10TCL_LIGHT_SPOT_CUTOFF_D__SIZE						0x00000008
+#define  NV10TCL_LIGHT_POSITION_X(x)							(0x0000085c+((x)*128))
+#define  NV10TCL_LIGHT_POSITION_X__SIZE							0x00000008
+#define  NV10TCL_LIGHT_POSITION_Y(x)							(0x00000860+((x)*128))
+#define  NV10TCL_LIGHT_POSITION_Y__SIZE							0x00000008
+#define  NV10TCL_LIGHT_POSITION_Z(x)							(0x00000864+((x)*128))
+#define  NV10TCL_LIGHT_POSITION_Z__SIZE							0x00000008
+#define  NV10TCL_LIGHT_ATTENUATION_CONSTANT(x)						(0x00000868+((x)*128))
+#define  NV10TCL_LIGHT_ATTENUATION_CONSTANT__SIZE					0x00000008
+#define  NV10TCL_LIGHT_ATTENUATION_LINEAR(x)						(0x0000086c+((x)*128))
+#define  NV10TCL_LIGHT_ATTENUATION_LINEAR__SIZE						0x00000008
+#define  NV10TCL_LIGHT_ATTENUATION_QUADRATIC(x)						(0x00000870+((x)*128))
+#define  NV10TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE					0x00000008
+#define  NV10TCL_VERTEX_POS_3F_X							0x00000c00
+#define  NV10TCL_VERTEX_POS_3F_Y							0x00000c04
+#define  NV10TCL_VERTEX_POS_3F_Z							0x00000c08
+#define  NV10TCL_VERTEX_POS_4F_X							0x00000c18
+#define  NV10TCL_VERTEX_POS_4F_Y							0x00000c1c
+#define  NV10TCL_VERTEX_POS_4F_Z							0x00000c20
+#define  NV10TCL_VERTEX_POS_4F_W							0x00000c24
+#define  NV10TCL_VERTEX_NOR_3F_X							0x00000c30
+#define  NV10TCL_VERTEX_NOR_3F_Y							0x00000c34
+#define  NV10TCL_VERTEX_NOR_3F_Z							0x00000c38
+#define  NV10TCL_VERTEX_NOR_3I_XY							0x00000c40
+#define   NV10TCL_VERTEX_NOR_3I_XY_X_SHIFT						0
+#define   NV10TCL_VERTEX_NOR_3I_XY_X_MASK						0x0000ffff
+#define   NV10TCL_VERTEX_NOR_3I_XY_Y_SHIFT						16
+#define   NV10TCL_VERTEX_NOR_3I_XY_Y_MASK						0xffff0000
+#define  NV10TCL_VERTEX_NOR_3I_Z							0x00000c44
+#define   NV10TCL_VERTEX_NOR_3I_Z_Z_SHIFT						0
+#define   NV10TCL_VERTEX_NOR_3I_Z_Z_MASK						0x0000ffff
+#define  NV10TCL_VERTEX_COL_4F_R							0x00000c50
+#define  NV10TCL_VERTEX_COL_4F_G							0x00000c54
+#define  NV10TCL_VERTEX_COL_4F_B							0x00000c58
+#define  NV10TCL_VERTEX_COL_4F_A							0x00000c5c
+#define  NV10TCL_VERTEX_COL_3F_R							0x00000c60
+#define  NV10TCL_VERTEX_COL_3F_G							0x00000c64
+#define  NV10TCL_VERTEX_COL_3F_B							0x00000c68
+#define  NV10TCL_VERTEX_COL_4I								0x00000c6c
+#define   NV10TCL_VERTEX_COL_4I_R_SHIFT							0
+#define   NV10TCL_VERTEX_COL_4I_R_MASK							0x000000ff
+#define   NV10TCL_VERTEX_COL_4I_G_SHIFT							8
+#define   NV10TCL_VERTEX_COL_4I_G_MASK							0x0000ff00
+#define   NV10TCL_VERTEX_COL_4I_B_SHIFT							16
+#define   NV10TCL_VERTEX_COL_4I_B_MASK							0x00ff0000
+#define   NV10TCL_VERTEX_COL_4I_A_SHIFT							24
+#define   NV10TCL_VERTEX_COL_4I_A_MASK							0xff000000
+#define  NV10TCL_VERTEX_COL2_3F_R							0x00000c80
+#define  NV10TCL_VERTEX_COL2_3F_G							0x00000c84
+#define  NV10TCL_VERTEX_COL2_3F_B							0x00000c88
+#define  NV10TCL_VERTEX_COL2_3I								0x00000c8c
+#define   NV10TCL_VERTEX_COL2_3I_R_SHIFT						0
+#define   NV10TCL_VERTEX_COL2_3I_R_MASK							0x000000ff
+#define   NV10TCL_VERTEX_COL2_3I_G_SHIFT						8
+#define   NV10TCL_VERTEX_COL2_3I_G_MASK							0x0000ff00
+#define   NV10TCL_VERTEX_COL2_3I_B_SHIFT						16
+#define   NV10TCL_VERTEX_COL2_3I_B_MASK							0x00ff0000
+#define  NV10TCL_VERTEX_TX0_2F_S							0x00000c90
+#define  NV10TCL_VERTEX_TX0_2F_T							0x00000c94
+#define  NV10TCL_VERTEX_TX0_2I								0x00000c98
+#define   NV10TCL_VERTEX_TX0_2I_S_SHIFT							0
+#define   NV10TCL_VERTEX_TX0_2I_S_MASK							0x0000ffff
+#define   NV10TCL_VERTEX_TX0_2I_T_SHIFT							16
+#define   NV10TCL_VERTEX_TX0_2I_T_MASK							0xffff0000
+#define  NV10TCL_VERTEX_TX0_4F_S							0x00000ca0
+#define  NV10TCL_VERTEX_TX0_4F_T							0x00000ca4
+#define  NV10TCL_VERTEX_TX0_4F_R							0x00000ca8
+#define  NV10TCL_VERTEX_TX0_4F_Q							0x00000cac
+#define  NV10TCL_VERTEX_TX0_4I_ST							0x00000cb0
+#define   NV10TCL_VERTEX_TX0_4I_ST_S_SHIFT						0
+#define   NV10TCL_VERTEX_TX0_4I_ST_S_MASK						0x0000ffff
+#define   NV10TCL_VERTEX_TX0_4I_ST_T_SHIFT						16
+#define   NV10TCL_VERTEX_TX0_4I_ST_T_MASK						0xffff0000
+#define  NV10TCL_VERTEX_TX0_4I_RQ							0x00000cb4
+#define   NV10TCL_VERTEX_TX0_4I_RQ_R_SHIFT						0
+#define   NV10TCL_VERTEX_TX0_4I_RQ_R_MASK						0x0000ffff
+#define   NV10TCL_VERTEX_TX0_4I_RQ_Q_SHIFT						16
+#define   NV10TCL_VERTEX_TX0_4I_RQ_Q_MASK						0xffff0000
+#define  NV10TCL_VERTEX_TX1_2F_S							0x00000cb8
+#define  NV10TCL_VERTEX_TX1_2F_T							0x00000cbc
+#define  NV10TCL_VERTEX_TX1_2I								0x00000cc0
+#define   NV10TCL_VERTEX_TX1_2I_S_SHIFT							0
+#define   NV10TCL_VERTEX_TX1_2I_S_MASK							0x0000ffff
+#define   NV10TCL_VERTEX_TX1_2I_T_SHIFT							16
+#define   NV10TCL_VERTEX_TX1_2I_T_MASK							0xffff0000
+#define  NV10TCL_VERTEX_TX1_4F_S							0x00000cc8
+#define  NV10TCL_VERTEX_TX1_4F_T							0x00000ccc
+#define  NV10TCL_VERTEX_TX1_4F_R							0x00000cd0
+#define  NV10TCL_VERTEX_TX1_4F_Q							0x00000cd4
+#define  NV10TCL_VERTEX_TX1_4I_ST							0x00000cd8
+#define   NV10TCL_VERTEX_TX1_4I_ST_S_SHIFT						0
+#define   NV10TCL_VERTEX_TX1_4I_ST_S_MASK						0x0000ffff
+#define   NV10TCL_VERTEX_TX1_4I_ST_T_SHIFT						16
+#define   NV10TCL_VERTEX_TX1_4I_ST_T_MASK						0xffff0000
+#define  NV10TCL_VERTEX_TX1_4I_RQ							0x00000cdc
+#define   NV10TCL_VERTEX_TX1_4I_RQ_R_SHIFT						0
+#define   NV10TCL_VERTEX_TX1_4I_RQ_R_MASK						0x0000ffff
+#define   NV10TCL_VERTEX_TX1_4I_RQ_Q_SHIFT						16
+#define   NV10TCL_VERTEX_TX1_4I_RQ_Q_MASK						0xffff0000
+#define  NV10TCL_VERTEX_FOG_1F								0x00000ce0
+#define  NV10TCL_VERTEX_WGH_1F								0x00000ce4
+#define  NV10TCL_EDGEFLAG_ENABLE							0x00000cec
+#define  NV10TCL_VERTEX_ARRAY_VALIDATE							0x00000cf0
+#define  NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(x)						(0x00000d00+((x)*8))
+#define  NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET__SIZE					0x00000008
+#define  NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(x)						(0x00000d04+((x)*8))
+#define  NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT__SIZE					0x00000008
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_FIELDS_SHIFT				4
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_FIELDS_MASK				0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_STRIDE_SHIFT				8
+#define   NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT_STRIDE_MASK				0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_POS						0x00000d00
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_POS						0x00000d04
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_COL						0x00000d08
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_COL						0x00000d0c
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_COL2						0x00000d10
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_COL2						0x00000d14
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_TX0						0x00000d18
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_TX0						0x00000d1c
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_TX1						0x00000d20
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_TX1						0x00000d24
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_NOR						0x00000d28
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_NOR						0x00000d2c
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_WGH						0x00000d30
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_WGH						0x00000d34
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_ARRAY_OFFSET_FOG						0x00000d38
+#define  NV10TCL_VERTEX_ARRAY_FORMAT_FOG						0x00000d3c
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_SHIFT					0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK					0x0000000f
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT					4
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK					0x000000f0
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT					8
+#define   NV10TCL_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK					0x0000ff00
+#define  NV10TCL_VERTEX_BEGIN_END							0x00000dfc
+#define   NV10TCL_VERTEX_BEGIN_END_STOP							0x00000000
+#define   NV10TCL_VERTEX_BEGIN_END_POINTS						0x00000001
+#define   NV10TCL_VERTEX_BEGIN_END_LINES						0x00000002
+#define   NV10TCL_VERTEX_BEGIN_END_LINE_LOOP						0x00000003
+#define   NV10TCL_VERTEX_BEGIN_END_LINE_STRIP						0x00000004
+#define   NV10TCL_VERTEX_BEGIN_END_TRIANGLES						0x00000005
+#define   NV10TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP					0x00000006
+#define   NV10TCL_VERTEX_BEGIN_END_TRIANGLE_FAN						0x00000007
+#define   NV10TCL_VERTEX_BEGIN_END_QUADS						0x00000008
+#define   NV10TCL_VERTEX_BEGIN_END_QUAD_STRIP						0x00000009
+#define   NV10TCL_VERTEX_BEGIN_END_POLYGON						0x0000000a
+#define  NV10TCL_VB_ELEMENT_U16								0x00000e00
+#define   NV10TCL_VB_ELEMENT_U16_I0_SHIFT						0
+#define   NV10TCL_VB_ELEMENT_U16_I0_MASK						0x0000ffff
+#define   NV10TCL_VB_ELEMENT_U16_I1_SHIFT						16
+#define   NV10TCL_VB_ELEMENT_U16_I1_MASK						0xffff0000
+#define  NV10TCL_VB_ELEMENT_U32								0x00001100
+#define  NV10TCL_VERTEX_BUFFER_BEGIN_END						0x000013fc
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_STOP						0x00000000
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_POINTS					0x00000001
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_LINES						0x00000002
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_LOOP					0x00000003
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_LINE_STRIP					0x00000004
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLES					0x00000005
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_STRIP				0x00000006
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_TRIANGLE_FAN					0x00000007
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_QUADS						0x00000008
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_QUAD_STRIP					0x00000009
+#define   NV10TCL_VERTEX_BUFFER_BEGIN_END_POLYGON					0x0000000a
+#define  NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS						0x00001400
+#define   NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_SHIFT					0
+#define   NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK					0x0000ffff
+#define   NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_SHIFT					24
+#define   NV10TCL_VERTEX_BUFFER_DRAW_ARRAYS_LAST_MASK					0xff000000
+#define  NV10TCL_VERTEX_ARRAY_DATA							0x00001800
+
+
+#define NV04_CONTEXT_COLOR_KEY								0x00000057
+
+
+
+#define NV03_CONTEXT_SURFACES_2D							0x00000058
+
+#define  NV03_CONTEXT_SURFACES_2D_SYNCHRONIZE						0x00000100
+#define  NV03_CONTEXT_SURFACES_2D_DMA_NOTIFY						0x00000180
+#define  NV03_CONTEXT_SURFACES_2D_DMA_SOURCE						0x00000184
+#define  NV03_CONTEXT_SURFACES_2D_DMA_DESTIN						0x00000188
+#define  NV03_CONTEXT_SURFACES_2D_COLOR_FORMAT						0x00000300
+#define  NV03_CONTEXT_SURFACES_2D_PITCH							0x00000304
+#define   NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_SHIFT					0
+#define   NV03_CONTEXT_SURFACES_2D_PITCH_SOURCE_MASK					0x0000ffff
+#define   NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_SHIFT					16
+#define   NV03_CONTEXT_SURFACES_2D_PITCH_DESTIN_MASK					0xffff0000
+#define  NV03_CONTEXT_SURFACES_2D_OFFSET_SOURCE						0x00000308
+#define  NV03_CONTEXT_SURFACES_2D_OFFSET_DESTIN						0x0000030c
+
+
+#define NV03_CONTEXT_SURFACES_3D							0x0000005a
+
+#define  NV03_CONTEXT_SURFACES_3D_SYNCHRONIZE						0x00000100
+#define  NV03_CONTEXT_SURFACES_3D_DMA_NOTIFY						0x00000180
+#define  NV03_CONTEXT_SURFACES_3D_DMA_SURFACE						0x00000184
+#define  NV03_CONTEXT_SURFACES_3D_PITCH							0x00000300
+#define  NV03_CONTEXT_SURFACES_3D_OFFSET_COLOR						0x00000304
+#define  NV03_CONTEXT_SURFACES_3D_OFFSET_ZETA						0x00000308
+
+
+#define NV04_RENDER_SOLID_LINE								0x0000005c
+
+#define  NV04_RENDER_SOLID_LINE_SURFACE							0x00000198
+
+
+#define NV04_RENDER_SOLID_TRIANGLE							0x0000005d
+
+
+
+#define NV04_RENDER_SOLID_RECTANGLE							0x0000005e
+
+#define  NV04_RENDER_SOLID_RECTANGLE_SURFACE						0x00000198
+
+
+#define NV04_IMAGE_BLIT									0x0000005f
+
+#define  NV04_IMAGE_BLIT_NOP								0x00000100
+#define  NV04_IMAGE_BLIT_NOTIFY								0x00000104
+#define  NV04_IMAGE_BLIT_DMA_NOTIFY							0x00000180
+#define  NV04_IMAGE_BLIT_COLOR_KEY							0x00000184
+#define  NV04_IMAGE_BLIT_CLIP_RECTANGLE							0x00000188
+#define  NV04_IMAGE_BLIT_PATTERN							0x0000018c
+#define  NV04_IMAGE_BLIT_ROP								0x00000190
+#define  NV04_IMAGE_BLIT_BETA4								0x00000198
+#define  NV04_IMAGE_BLIT_SURFACE							0x0000019c
+#define  NV04_IMAGE_BLIT_OPERATION							0x000002fc
+#define   NV04_IMAGE_BLIT_OPERATION_SRCCOPY_AND						0x00000000
+#define   NV04_IMAGE_BLIT_OPERATION_ROP_AND						0x00000001
+#define   NV04_IMAGE_BLIT_OPERATION_BLEND_AND						0x00000002
+#define   NV04_IMAGE_BLIT_OPERATION_SRCCOPY						0x00000003
+#define   NV04_IMAGE_BLIT_OPERATION_SRCCOPY_PREMULT					0x00000004
+#define   NV04_IMAGE_BLIT_OPERATION_BLEND_PREMULT					0x00000005
+
+
+#define NV04_INDEXED_IMAGE_FROM_CPU							0x00000060
+
+#define  NV04_INDEXED_IMAGE_FROM_CPU_NOP						0x00000100
+#define  NV04_INDEXED_IMAGE_FROM_CPU_NOTIFY						0x00000104
+#define  NV04_INDEXED_IMAGE_FROM_CPU_PATCH						0x0000010c
+#define  NV04_INDEXED_IMAGE_FROM_CPU_DMA_NOTIFY						0x00000180
+#define  NV04_INDEXED_IMAGE_FROM_CPU_DMA_LUT						0x00000184
+#define  NV04_INDEXED_IMAGE_FROM_CPU_COLOR_FORMAT					0x000003e8
+#define  NV04_INDEXED_IMAGE_FROM_CPU_INDEX_FORMAT					0x000003ec
+#define  NV04_INDEXED_IMAGE_FROM_CPU_LUT_OFFSET						0x000003f0
+#define  NV04_INDEXED_IMAGE_FROM_CPU_POINT						0x000003f4
+#define  NV04_INDEXED_IMAGE_FROM_CPU_SIZE_OUT						0x000003f8
+#define  NV04_INDEXED_IMAGE_FROM_CPU_SIZE_IN						0x000003fc
+#define  NV04_INDEXED_IMAGE_FROM_CPU_COLOR						0x00000400
+
+
+#define NV04_IMAGE_FROM_CPU								0x00000061
+
+#define  NV04_IMAGE_FROM_CPU_BETA4							0x00000198
+#define  NV04_IMAGE_FROM_CPU_SURFACE							0x0000019c
+
+
+#define NV10_CONTEXT_SURFACES_2D							0x00000062
+
+
+
+#define NV05_SCALED_IMAGE_FROM_MEMORY							0x00000063
+
+#define  NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION					0x000002fc
+#define   NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER				0x00000000
+#define   NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE			0x00000001
+#define   NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE			0x00000002
+
+
+#define NV01_IMAGE_SRCCOPY_AND								0x00000064
+
+#define  NV01_IMAGE_SRCCOPY_AND_NOTIFY							0x00000104
+#define  NV01_IMAGE_SRCCOPY_AND_DMA_NOTIFY						0x00000180
+#define  NV01_IMAGE_SRCCOPY_AND_IMAGE_OUTPUT						0x00000200
+#define  NV01_IMAGE_SRCCOPY_AND_IMAGE_INPUT						0x00000204
+
+
+#define NV05_INDEXED_IMAGE_FROM_CPU							0x00000064
+
+#define  NV05_INDEXED_IMAGE_FROM_CPU_COLOR_KEY						0x00000188
+#define  NV05_INDEXED_IMAGE_FROM_CPU_CLIP_RECTANGLE					0x0000018c
+#define  NV05_INDEXED_IMAGE_FROM_CPU_PATTERN						0x00000190
+#define  NV05_INDEXED_IMAGE_FROM_CPU_ROP						0x00000194
+#define  NV05_INDEXED_IMAGE_FROM_CPU_BETA1						0x00000198
+#define  NV05_INDEXED_IMAGE_FROM_CPU_BETA4						0x0000019c
+#define  NV05_INDEXED_IMAGE_FROM_CPU_SURFACE						0x000001a0
+#define  NV05_INDEXED_IMAGE_FROM_CPU_COLOR_CONVERSION					0x000003e0
+#define  NV05_INDEXED_IMAGE_FROM_CPU_OPERATION						0x000003e4
+#define  NV05_INDEXED_IMAGE_FROM_CPU_INDICES						0x00000400
+
+
+#define NV05_IMAGE_FROM_CPU								0x00000065
+
+#define  NV05_IMAGE_FROM_CPU_BETA4							0x00000198
+#define  NV05_IMAGE_FROM_CPU_SURFACE							0x0000019c
+
+
+#define NV05_STRETCHED_IMAGE_FROM_CPU							0x00000066
+
+#define  NV05_STRETCHED_IMAGE_FROM_CPU_BETA4						0x00000194
+#define  NV05_STRETCHED_IMAGE_FROM_CPU_SURFACE						0x00000198
+#define  NV05_STRETCHED_IMAGE_FROM_CPU_COLOR_CONVERSION					0x000002f8
+
+
+#define NV04_IMAGE_BLEND_PREMULT							0x00000067
+
+#define  NV04_IMAGE_BLEND_PREMULT_NOP							0x00000100
+#define  NV04_IMAGE_BLEND_PREMULT_NOTIFY						0x00000104
+#define  NV04_IMAGE_BLEND_PREMULT_DMA_NOTIFY						0x00000180
+#define  NV04_IMAGE_BLEND_PREMULT_IMAGE_OUTPUT						0x00000200
+#define  NV04_IMAGE_BLEND_PREMULT_BETA_INPUT						0x00000204
+#define  NV04_IMAGE_BLEND_PREMULT_IMAGE_INPUT						0x00000208
+
+
+#define NV03_CHANNEL_PIO								0x0000006a
+
+
+
+#define NV03_CHANNEL_DMA								0x0000006b
+
+
+
+#define NV04_BETA_SOLID									0x00000072
+
+#define  NV04_BETA_SOLID_NOP								0x00000100
+#define  NV04_BETA_SOLID_NOTIFY								0x00000104
+#define  NV04_BETA_SOLID_DMA_NOTIFY							0x00000180
+#define  NV04_BETA_SOLID_BETA_OUTPUT							0x00000200
+#define  NV04_BETA_SOLID_BETA_FACTOR							0x00000300
+
+
+#define NV04_STRETCHED_IMAGE_FROM_CPU							0x00000076
+
+
+
+#define NV04_SCALED_IMAGE_FROM_MEMORY							0x00000077
+
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_NOP						0x00000100
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_NOTIFY						0x00000104
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY					0x00000180
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE					0x00000184
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN						0x00000188
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_ROP						0x0000018c
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_BETA1						0x00000190
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_BETA4						0x00000194
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE						0x00000198
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION					0x000002fc
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_DITHER				0x00000000
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE			0x00000001
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_SUBTR_TRUNCATE			0x00000002
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT					0x00000300
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5				0x00000001
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X1R5G5B5				0x00000002
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8				0x00000003
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8				0x00000004
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_V8YB8U8YA8				0x00000005
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_YB8V8YA8U8				0x00000006
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5				0x00000007
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_Y8					0x00000008
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_AY8				0x00000009
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION					0x00000304
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_AND				0x00000000
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_ROP_AND				0x00000001
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_AND				0x00000002
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY				0x00000003
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY_PREMULT			0x00000004
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_BLEND_PREMULT				0x00000005
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT					0x00000308
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_SHIFT				0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_X_MASK				0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_SHIFT				16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POINT_Y_MASK				0xffff0000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE					0x0000030c
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_SHIFT				0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_W_MASK				0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_SHIFT				16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_H_MASK				0xffff0000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT					0x00000310
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_SHIFT				0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_X_MASK				0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_SHIFT				16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POINT_Y_MASK				0xffff0000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE						0x00000314
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_SHIFT				0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_W_MASK					0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_SHIFT				16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_H_MASK					0xffff0000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX						0x00000318
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY						0x0000031c
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_SIZE						0x00000400
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_SHIFT					0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_W_MASK					0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_SHIFT					16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_H_MASK					0xffff0000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT						0x00000404
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_SHIFT				0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK				0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_SHIFT				16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_MASK				0x00ff0000
+#define    NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER				0x00010000
+#define    NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CORNER				0x00020000
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_SHIFT				24
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_MASK				0xff000000
+#define    NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE			0x00000000
+#define    NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_BILINEAR				0x01000000
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_ADDRESS						0x00000408
+#define  NV04_SCALED_IMAGE_FROM_MEMORY_POINT						0x0000040c
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_SHIFT					0
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_POINT_X_MASK					0x0000ffff
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_SHIFT					16
+#define   NV04_SCALED_IMAGE_FROM_MEMORY_POINT_Y_MASK					0xffff0000
+
+
+#define NV10_TEXTURE_FROM_CPU								0x0000007b
+
+#define  NV10_TEXTURE_FROM_CPU_NOP							0x00000100
+#define  NV10_TEXTURE_FROM_CPU_NOTIFY							0x00000104
+#define  NV10_TEXTURE_FROM_CPU_WAIT_FOR_IDLE						0x00000108
+#define  NV10_TEXTURE_FROM_CPU_PM_TRIGGER						0x00000140
+#define  NV10_TEXTURE_FROM_CPU_DMA_NOTIFY						0x00000180
+#define  NV10_TEXTURE_FROM_CPU_SURFACE							0x00000184
+#define  NV10_TEXTURE_FROM_CPU_COLOR_FORMAT						0x00000300
+#define  NV10_TEXTURE_FROM_CPU_POINT							0x00000304
+#define   NV10_TEXTURE_FROM_CPU_POINT_X_SHIFT						0
+#define   NV10_TEXTURE_FROM_CPU_POINT_X_MASK						0x0000ffff
+#define   NV10_TEXTURE_FROM_CPU_POINT_Y_SHIFT						16
+#define   NV10_TEXTURE_FROM_CPU_POINT_Y_MASK						0xffff0000
+#define  NV10_TEXTURE_FROM_CPU_SIZE							0x00000308
+#define   NV10_TEXTURE_FROM_CPU_SIZE_W_SHIFT						0
+#define   NV10_TEXTURE_FROM_CPU_SIZE_W_MASK						0x0000ffff
+#define   NV10_TEXTURE_FROM_CPU_SIZE_H_SHIFT						16
+#define   NV10_TEXTURE_FROM_CPU_SIZE_H_MASK						0xffff0000
+#define  NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL						0x0000030c
+#define   NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_SHIFT					0
+#define   NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_X_MASK					0x0000ffff
+#define   NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_SHIFT					16
+#define   NV10_TEXTURE_FROM_CPU_CLIP_HORIZONTAL_W_MASK					0xffff0000
+#define  NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL						0x00000310
+#define   NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_SHIFT					0
+#define   NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_Y_MASK					0x0000ffff
+#define   NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_SHIFT					16
+#define   NV10_TEXTURE_FROM_CPU_CLIP_VERTICAL_H_MASK					0xffff0000
+#define  NV10_TEXTURE_FROM_CPU_COLOR(x)							(0x00000400+((x)*4))
+#define  NV10_TEXTURE_FROM_CPU_COLOR__SIZE						0x00000700
+
+
+#define NV10_VIDEO_DISPLAY								0x0000007c
+
+
+
+#define NV10_DVD_SUBPICTURE								0x00000088
+
+
+
+#define NV10_SCALED_IMAGE_FROM_MEMORY							0x00000089
+
+#define  NV10_SCALED_IMAGE_FROM_MEMORY_WAIT_FOR_IDLE					0x00000108
+
+
+#define NV10_IMAGE_FROM_CPU								0x0000008a
+
+#define  NV10_IMAGE_FROM_CPU_COLOR_CONVERSION						0x000002f8
+
+
+#define NV10_CONTEXT_SURFACES_3D							0x00000093
+
+
+
+#define NV10_DX5_TEXTURE_TRIANGLE							0x00000094
+
+
+
+#define NV10_DX6_MULTI_TEXTURE_TRIANGLE							0x00000095
+
+
+
+#define NV11TCL										0x00000096
+
+#define  NV11TCL_COLOR_LOGIC_OP_ENABLE							0x00000d40
+#define  NV11TCL_COLOR_LOGIC_OP_OP							0x00000d44
+#define   NV11TCL_COLOR_LOGIC_OP_OP_CLEAR						0x00001500
+#define   NV11TCL_COLOR_LOGIC_OP_OP_AND							0x00001501
+#define   NV11TCL_COLOR_LOGIC_OP_OP_AND_REVERSE						0x00001502
+#define   NV11TCL_COLOR_LOGIC_OP_OP_COPY						0x00001503
+#define   NV11TCL_COLOR_LOGIC_OP_OP_AND_INVERTED					0x00001504
+#define   NV11TCL_COLOR_LOGIC_OP_OP_NOOP						0x00001505
+#define   NV11TCL_COLOR_LOGIC_OP_OP_XOR							0x00001506
+#define   NV11TCL_COLOR_LOGIC_OP_OP_OR							0x00001507
+#define   NV11TCL_COLOR_LOGIC_OP_OP_NOR							0x00001508
+#define   NV11TCL_COLOR_LOGIC_OP_OP_EQUIV						0x00001509
+#define   NV11TCL_COLOR_LOGIC_OP_OP_INVERT						0x0000150a
+#define   NV11TCL_COLOR_LOGIC_OP_OP_OR_REVERSE						0x0000150b
+#define   NV11TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED					0x0000150c
+#define   NV11TCL_COLOR_LOGIC_OP_OP_OR_INVERTED						0x0000150d
+#define   NV11TCL_COLOR_LOGIC_OP_OP_NAND						0x0000150e
+#define   NV11TCL_COLOR_LOGIC_OP_OP_SET							0x0000150f
+
+
+#define NV20TCL										0x00000097
+
+#define  NV20TCL_NOP									0x00000100
+#define  NV20TCL_NOTIFY									0x00000104
+#define  NV20TCL_DMA_NOTIFY								0x00000180
+#define  NV20TCL_DMA_TEXTURE0								0x00000184
+#define  NV20TCL_DMA_TEXTURE1								0x00000188
+#define  NV20TCL_DMA_COLOR								0x00000194
+#define  NV20TCL_DMA_ZETA								0x00000198
+#define  NV20TCL_DMA_VTXBUF0								0x0000019c
+#define  NV20TCL_DMA_VTXBUF1								0x000001a0
+#define  NV20TCL_DMA_FENCE								0x000001a4
+#define  NV20TCL_DMA_QUERY								0x000001a8
+#define  NV20TCL_RT_HORIZ								0x00000200
+#define   NV20TCL_RT_HORIZ_X_SHIFT							0
+#define   NV20TCL_RT_HORIZ_X_MASK							0x0000ffff
+#define   NV20TCL_RT_HORIZ_W_SHIFT							16
+#define   NV20TCL_RT_HORIZ_W_MASK							0xffff0000
+#define  NV20TCL_RT_VERT								0x00000204
+#define   NV20TCL_RT_VERT_Y_SHIFT							0
+#define   NV20TCL_RT_VERT_Y_MASK							0x0000ffff
+#define   NV20TCL_RT_VERT_H_SHIFT							16
+#define   NV20TCL_RT_VERT_H_MASK							0xffff0000
+#define  NV20TCL_RT_FORMAT								0x00000208
+#define   NV20TCL_RT_FORMAT_TYPE_SHIFT							8
+#define   NV20TCL_RT_FORMAT_TYPE_MASK							0x00000f00
+#define    NV20TCL_RT_FORMAT_TYPE_LINEAR						0x00000100
+#define    NV20TCL_RT_FORMAT_TYPE_SWIZZLED						0x00000200
+#define   NV20TCL_RT_FORMAT_COLOR_SHIFT							0
+#define   NV20TCL_RT_FORMAT_COLOR_MASK							0x0000001f
+#define    NV20TCL_RT_FORMAT_COLOR_R5G6B5						0x00000003
+#define    NV20TCL_RT_FORMAT_COLOR_X8R8G8B8						0x00000005
+#define    NV20TCL_RT_FORMAT_COLOR_A8R8G8B8						0x00000008
+#define    NV20TCL_RT_FORMAT_COLOR_B8							0x00000009
+#define    NV20TCL_RT_FORMAT_COLOR_UNKNOWN						0x0000000d
+#define    NV20TCL_RT_FORMAT_COLOR_X8B8G8R8						0x0000000f
+#define    NV20TCL_RT_FORMAT_COLOR_A8B8G8R8						0x00000010
+#define  NV20TCL_RT_PITCH								0x0000020c
+#define   NV20TCL_RT_PITCH_COLOR_PITCH_SHIFT						0
+#define   NV20TCL_RT_PITCH_COLOR_PITCH_MASK						0x0000ffff
+#define   NV20TCL_RT_PITCH_ZETA_PITCH_SHIFT						16
+#define   NV20TCL_RT_PITCH_ZETA_PITCH_MASK						0xffff0000
+#define  NV20TCL_COLOR_OFFSET								0x00000210
+#define  NV20TCL_ZETA_OFFSET								0x00000214
+#define  NV20TCL_RC_IN_ALPHA(x)								(0x00000260+((x)*4))
+#define  NV20TCL_RC_IN_ALPHA__SIZE							0x00000008
+#define   NV20TCL_RC_IN_ALPHA_D_INPUT_SHIFT						0
+#define   NV20TCL_RC_IN_ALPHA_D_INPUT_MASK						0x0000000f
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_FOG						0x00000003
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV					0x0000000c
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV					0x0000000d
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV20TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV20TCL_RC_IN_ALPHA_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV20TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV20TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV20TCL_RC_IN_ALPHA_D_MAPPING_SHIFT						5
+#define   NV20TCL_RC_IN_ALPHA_D_MAPPING_MASK						0x000000e0
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV				0x00000040
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV				0x00000060
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV20TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV				0x000000e0
+#define   NV20TCL_RC_IN_ALPHA_C_INPUT_SHIFT						8
+#define   NV20TCL_RC_IN_ALPHA_C_INPUT_MASK						0x00000f00
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_FOG						0x00000300
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV					0x00000c00
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV					0x00000d00
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV20TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_IN_ALPHA_C_COMPONENT_USAGE						(1 << 12)
+#define    NV20TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV20TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV20TCL_RC_IN_ALPHA_C_MAPPING_SHIFT						13
+#define   NV20TCL_RC_IN_ALPHA_C_MAPPING_MASK						0x0000e000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV				0x00004000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV				0x00006000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV20TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV				0x0000e000
+#define   NV20TCL_RC_IN_ALPHA_B_INPUT_SHIFT						16
+#define   NV20TCL_RC_IN_ALPHA_B_INPUT_MASK						0x000f0000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV				0x00010000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV				0x00020000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_FOG						0x00030000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV				0x00050000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV					0x000c0000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV					0x000d0000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV20TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV20TCL_RC_IN_ALPHA_B_COMPONENT_USAGE						(1 << 20)
+#define    NV20TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV20TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV20TCL_RC_IN_ALPHA_B_MAPPING_SHIFT						21
+#define   NV20TCL_RC_IN_ALPHA_B_MAPPING_MASK						0x00e00000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV				0x00400000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV				0x00600000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV20TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV				0x00e00000
+#define   NV20TCL_RC_IN_ALPHA_A_INPUT_SHIFT						24
+#define   NV20TCL_RC_IN_ALPHA_A_INPUT_MASK						0x0f000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV				0x01000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV				0x02000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_FOG						0x03000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV				0x05000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV					0x0c000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV					0x0d000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV20TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV20TCL_RC_IN_ALPHA_A_COMPONENT_USAGE						(1 << 28)
+#define    NV20TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV20TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV20TCL_RC_IN_ALPHA_A_MAPPING_SHIFT						29
+#define   NV20TCL_RC_IN_ALPHA_A_MAPPING_MASK						0xe0000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV				0x40000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV				0x60000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV20TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV				0xe0000000
+#define  NV20TCL_RC_FINAL0								0x00000288
+#define   NV20TCL_RC_FINAL0_D_INPUT_SHIFT						0
+#define   NV20TCL_RC_FINAL0_D_INPUT_MASK						0x0000000f
+#define    NV20TCL_RC_FINAL0_D_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV20TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV20TCL_RC_FINAL0_D_INPUT_FOG						0x00000003
+#define    NV20TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV20TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV20TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV20TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV20TCL_RC_FINAL0_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV20TCL_RC_FINAL0_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV20TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV20TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV20TCL_RC_FINAL0_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV20TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV20TCL_RC_FINAL0_D_MAPPING_SHIFT						5
+#define   NV20TCL_RC_FINAL0_D_MAPPING_MASK						0x000000e0
+#define    NV20TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV20TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV20TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV20TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV20TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV20TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV20TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV20TCL_RC_FINAL0_C_INPUT_SHIFT						8
+#define   NV20TCL_RC_FINAL0_C_INPUT_MASK						0x00000f00
+#define    NV20TCL_RC_FINAL0_C_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV20TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV20TCL_RC_FINAL0_C_INPUT_FOG						0x00000300
+#define    NV20TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV20TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV20TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_FINAL0_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV20TCL_RC_FINAL0_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV20TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV20TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_FINAL0_C_COMPONENT_USAGE						(1 << 12)
+#define    NV20TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV20TCL_RC_FINAL0_C_MAPPING_SHIFT						13
+#define   NV20TCL_RC_FINAL0_C_MAPPING_MASK						0x0000e000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV20TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV20TCL_RC_FINAL0_B_INPUT_SHIFT						16
+#define   NV20TCL_RC_FINAL0_B_INPUT_MASK						0x000f0000
+#define    NV20TCL_RC_FINAL0_B_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV20TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV20TCL_RC_FINAL0_B_INPUT_FOG						0x00030000
+#define    NV20TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV20TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV20TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV20TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV20TCL_RC_FINAL0_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV20TCL_RC_FINAL0_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV20TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV20TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV20TCL_RC_FINAL0_B_COMPONENT_USAGE						(1 << 20)
+#define    NV20TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV20TCL_RC_FINAL0_B_MAPPING_SHIFT						21
+#define   NV20TCL_RC_FINAL0_B_MAPPING_MASK						0x00e00000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV20TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV20TCL_RC_FINAL0_A_INPUT_SHIFT						24
+#define   NV20TCL_RC_FINAL0_A_INPUT_MASK						0x0f000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_FOG						0x03000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV20TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV20TCL_RC_FINAL0_A_COMPONENT_USAGE						(1 << 28)
+#define    NV20TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV20TCL_RC_FINAL0_A_MAPPING_SHIFT						29
+#define   NV20TCL_RC_FINAL0_A_MAPPING_MASK						0xe0000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV20TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV20TCL_RC_FINAL1								0x0000028c
+#define   NV20TCL_RC_FINAL1_COLOR_SUM_CLAMP						(1 <<  7)
+#define   NV20TCL_RC_FINAL1_G_INPUT_SHIFT						8
+#define   NV20TCL_RC_FINAL1_G_INPUT_MASK						0x00000f00
+#define    NV20TCL_RC_FINAL1_G_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV20TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV20TCL_RC_FINAL1_G_INPUT_FOG						0x00000300
+#define    NV20TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV20TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV20TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_FINAL1_G_INPUT_SPARE0_NV						0x00000c00
+#define    NV20TCL_RC_FINAL1_G_INPUT_SPARE1_NV						0x00000d00
+#define    NV20TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV20TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_FINAL1_G_COMPONENT_USAGE						(1 << 12)
+#define    NV20TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV20TCL_RC_FINAL1_G_MAPPING_SHIFT						13
+#define   NV20TCL_RC_FINAL1_G_MAPPING_MASK						0x0000e000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV20TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV20TCL_RC_FINAL1_F_INPUT_SHIFT						16
+#define   NV20TCL_RC_FINAL1_F_INPUT_MASK						0x000f0000
+#define    NV20TCL_RC_FINAL1_F_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV20TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV20TCL_RC_FINAL1_F_INPUT_FOG						0x00030000
+#define    NV20TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV20TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV20TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV20TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV20TCL_RC_FINAL1_F_INPUT_SPARE0_NV						0x000c0000
+#define    NV20TCL_RC_FINAL1_F_INPUT_SPARE1_NV						0x000d0000
+#define    NV20TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV20TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV20TCL_RC_FINAL1_F_COMPONENT_USAGE						(1 << 20)
+#define    NV20TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV20TCL_RC_FINAL1_F_MAPPING_SHIFT						21
+#define   NV20TCL_RC_FINAL1_F_MAPPING_MASK						0x00e00000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV20TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV20TCL_RC_FINAL1_E_INPUT_SHIFT						24
+#define   NV20TCL_RC_FINAL1_E_INPUT_MASK						0x0f000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_FOG						0x03000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_SPARE0_NV						0x0c000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_SPARE1_NV						0x0d000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV20TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV20TCL_RC_FINAL1_E_COMPONENT_USAGE						(1 << 28)
+#define    NV20TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV20TCL_RC_FINAL1_E_MAPPING_SHIFT						29
+#define   NV20TCL_RC_FINAL1_E_MAPPING_MASK						0xe0000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV20TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV20TCL_LIGHT_CONTROL								0x00000294
+#define  NV20TCL_FOG_MODE								0x0000029c
+#define   NV20TCL_FOG_MODE_EXP								0x00000800
+#define   NV20TCL_FOG_MODE_EXP_2							0x00000802
+#define   NV20TCL_FOG_MODE_EXP2								0x00000803
+#define   NV20TCL_FOG_MODE_LINEAR							0x00000804
+#define   NV20TCL_FOG_MODE_LINEAR_2							0x00002601
+#define  NV20TCL_FOG_COORD_DIST								0x000002a0
+#define   NV20TCL_FOG_COORD_DIST_COORD_FALSE						0x00000000
+#define   NV20TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_RADIAL_NV		0x00000001
+#define   NV20TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_PLANE_ABSOLUTE_NV	0x00000002
+#define   NV20TCL_FOG_COORD_DIST_COORD_FOG						0x00000003
+#define  NV20TCL_FOG_ENABLE								0x000002a4
+#define  NV20TCL_FOG_COLOR								0x000002a8
+#define   NV20TCL_FOG_COLOR_R_SHIFT							0
+#define   NV20TCL_FOG_COLOR_R_MASK							0x000000ff
+#define   NV20TCL_FOG_COLOR_G_SHIFT							8
+#define   NV20TCL_FOG_COLOR_G_MASK							0x0000ff00
+#define   NV20TCL_FOG_COLOR_B_SHIFT							16
+#define   NV20TCL_FOG_COLOR_B_MASK							0x00ff0000
+#define   NV20TCL_FOG_COLOR_A_SHIFT							24
+#define   NV20TCL_FOG_COLOR_A_MASK							0xff000000
+#define  NV20TCL_VIEWPORT_CLIP_MODE							0x000002b4
+#define  NV20TCL_VIEWPORT_CLIP_HORIZ(x)							(0x000002c0+((x)*4))
+#define  NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE						0x00000008
+#define  NV20TCL_VIEWPORT_CLIP_VERT(x)							(0x000002e0+((x)*4))
+#define  NV20TCL_VIEWPORT_CLIP_VERT__SIZE						0x00000008
+#define  NV20TCL_ALPHA_FUNC_ENABLE							0x00000300
+#define  NV20TCL_BLEND_FUNC_ENABLE							0x00000304
+#define  NV20TCL_CULL_FACE_ENABLE							0x00000308
+#define  NV20TCL_DEPTH_TEST_ENABLE							0x0000030c
+#define  NV20TCL_DITHER_ENABLE								0x00000310
+#define  NV20TCL_LIGHTING_ENABLE							0x00000314
+#define  NV20TCL_POINT_PARAMETERS_ENABLE						0x00000318
+#define  NV20TCL_POINT_SMOOTH_ENABLE							0x0000031c
+#define  NV20TCL_LINE_SMOOTH_ENABLE							0x00000320
+#define  NV20TCL_POLYGON_SMOOTH_ENABLE							0x00000324
+#define  NV20TCL_STENCIL_ENABLE								0x0000032c
+#define  NV20TCL_POLYGON_OFFSET_POINT_ENABLE						0x00000330
+#define  NV20TCL_POLYGON_OFFSET_LINE_ENABLE						0x00000334
+#define  NV20TCL_POLYGON_OFFSET_FILL_ENABLE						0x00000338
+#define  NV20TCL_ALPHA_FUNC_FUNC							0x0000033c
+#define   NV20TCL_ALPHA_FUNC_FUNC_NEVER							0x00000200
+#define   NV20TCL_ALPHA_FUNC_FUNC_LESS							0x00000201
+#define   NV20TCL_ALPHA_FUNC_FUNC_EQUAL							0x00000202
+#define   NV20TCL_ALPHA_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV20TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV20TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV20TCL_ALPHA_FUNC_FUNC_NOTEQUAL						0x00000205
+#define   NV20TCL_ALPHA_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV20TCL_ALPHA_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV20TCL_ALPHA_FUNC_REF								0x00000340
+#define  NV20TCL_BLEND_FUNC_SRC								0x00000344
+#define   NV20TCL_BLEND_FUNC_SRC_ZERO							0x00000000
+#define   NV20TCL_BLEND_FUNC_SRC_ONE							0x00000001
+#define   NV20TCL_BLEND_FUNC_SRC_SRC_COLOR						0x00000300
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_COLOR					0x00000301
+#define   NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA						0x00000302
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_SRC_ALPHA					0x00000303
+#define   NV20TCL_BLEND_FUNC_SRC_DST_ALPHA						0x00000304
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_ALPHA					0x00000305
+#define   NV20TCL_BLEND_FUNC_SRC_DST_COLOR						0x00000306
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_DST_COLOR					0x00000307
+#define   NV20TCL_BLEND_FUNC_SRC_SRC_ALPHA_SATURATE					0x00000308
+#define   NV20TCL_BLEND_FUNC_SRC_CONSTANT_COLOR						0x00008001
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV20TCL_BLEND_FUNC_SRC_CONSTANT_ALPHA						0x00008003
+#define   NV20TCL_BLEND_FUNC_SRC_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV20TCL_BLEND_FUNC_DST								0x00000348
+#define   NV20TCL_BLEND_FUNC_DST_ZERO							0x00000000
+#define   NV20TCL_BLEND_FUNC_DST_ONE							0x00000001
+#define   NV20TCL_BLEND_FUNC_DST_SRC_COLOR						0x00000300
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_COLOR					0x00000301
+#define   NV20TCL_BLEND_FUNC_DST_SRC_ALPHA						0x00000302
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_SRC_ALPHA					0x00000303
+#define   NV20TCL_BLEND_FUNC_DST_DST_ALPHA						0x00000304
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_ALPHA					0x00000305
+#define   NV20TCL_BLEND_FUNC_DST_DST_COLOR						0x00000306
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_DST_COLOR					0x00000307
+#define   NV20TCL_BLEND_FUNC_DST_SRC_ALPHA_SATURATE					0x00000308
+#define   NV20TCL_BLEND_FUNC_DST_CONSTANT_COLOR						0x00008001
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV20TCL_BLEND_FUNC_DST_CONSTANT_ALPHA						0x00008003
+#define   NV20TCL_BLEND_FUNC_DST_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV20TCL_BLEND_COLOR								0x0000034c
+#define   NV20TCL_BLEND_COLOR_B_SHIFT							0
+#define   NV20TCL_BLEND_COLOR_B_MASK							0x000000ff
+#define   NV20TCL_BLEND_COLOR_G_SHIFT							8
+#define   NV20TCL_BLEND_COLOR_G_MASK							0x0000ff00
+#define   NV20TCL_BLEND_COLOR_R_SHIFT							16
+#define   NV20TCL_BLEND_COLOR_R_MASK							0x00ff0000
+#define   NV20TCL_BLEND_COLOR_A_SHIFT							24
+#define   NV20TCL_BLEND_COLOR_A_MASK							0xff000000
+#define  NV20TCL_BLEND_EQUATION								0x00000350
+#define   NV20TCL_BLEND_EQUATION_FUNC_ADD						0x00008006
+#define   NV20TCL_BLEND_EQUATION_MIN							0x00008007
+#define   NV20TCL_BLEND_EQUATION_MAX							0x00008008
+#define   NV20TCL_BLEND_EQUATION_FUNC_SUBTRACT						0x0000800a
+#define   NV20TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT					0x0000800b
+#define  NV20TCL_DEPTH_FUNC								0x00000354
+#define   NV20TCL_DEPTH_FUNC_NEVER							0x00000200
+#define   NV20TCL_DEPTH_FUNC_LESS							0x00000201
+#define   NV20TCL_DEPTH_FUNC_EQUAL							0x00000202
+#define   NV20TCL_DEPTH_FUNC_LEQUAL							0x00000203
+#define   NV20TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV20TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV20TCL_DEPTH_FUNC_NOTEQUAL							0x00000205
+#define   NV20TCL_DEPTH_FUNC_GEQUAL							0x00000206
+#define   NV20TCL_DEPTH_FUNC_ALWAYS							0x00000207
+#define  NV20TCL_COLOR_MASK								0x00000358
+#define   NV20TCL_COLOR_MASK_B								(1 <<  0)
+#define   NV20TCL_COLOR_MASK_G								(1 <<  8)
+#define   NV20TCL_COLOR_MASK_R								(1 << 16)
+#define   NV20TCL_COLOR_MASK_A								(1 << 24)
+#define  NV20TCL_DEPTH_WRITE_ENABLE							0x0000035c
+#define  NV20TCL_STENCIL_MASK								0x00000360
+#define  NV20TCL_STENCIL_FUNC_FUNC							0x00000364
+#define   NV20TCL_STENCIL_FUNC_FUNC_NEVER						0x00000200
+#define   NV20TCL_STENCIL_FUNC_FUNC_LESS						0x00000201
+#define   NV20TCL_STENCIL_FUNC_FUNC_EQUAL						0x00000202
+#define   NV20TCL_STENCIL_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV20TCL_STENCIL_FUNC_FUNC_GREATER						0x00000204
+#define   NV20TCL_STENCIL_FUNC_FUNC_GREATER						0x00000204
+#define   NV20TCL_STENCIL_FUNC_FUNC_NOTEQUAL						0x00000205
+#define   NV20TCL_STENCIL_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV20TCL_STENCIL_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV20TCL_STENCIL_FUNC_REF							0x00000368
+#define  NV20TCL_STENCIL_FUNC_MASK							0x0000036c
+#define  NV20TCL_STENCIL_OP_FAIL							0x00000370
+#define   NV20TCL_STENCIL_OP_FAIL_ZERO							0x00000000
+#define   NV20TCL_STENCIL_OP_FAIL_INVERT						0x0000150a
+#define   NV20TCL_STENCIL_OP_FAIL_KEEP							0x00001e00
+#define   NV20TCL_STENCIL_OP_FAIL_REPLACE						0x00001e01
+#define   NV20TCL_STENCIL_OP_FAIL_INCR							0x00001e02
+#define   NV20TCL_STENCIL_OP_FAIL_DECR							0x00001e03
+#define   NV20TCL_STENCIL_OP_FAIL_INCR_WRAP						0x00008507
+#define   NV20TCL_STENCIL_OP_FAIL_DECR_WRAP						0x00008508
+#define  NV20TCL_STENCIL_OP_ZFAIL							0x00000374
+#define   NV20TCL_STENCIL_OP_ZFAIL_ZERO							0x00000000
+#define   NV20TCL_STENCIL_OP_ZFAIL_INVERT						0x0000150a
+#define   NV20TCL_STENCIL_OP_ZFAIL_KEEP							0x00001e00
+#define   NV20TCL_STENCIL_OP_ZFAIL_REPLACE						0x00001e01
+#define   NV20TCL_STENCIL_OP_ZFAIL_INCR							0x00001e02
+#define   NV20TCL_STENCIL_OP_ZFAIL_DECR							0x00001e03
+#define   NV20TCL_STENCIL_OP_ZFAIL_INCR_WRAP						0x00008507
+#define   NV20TCL_STENCIL_OP_ZFAIL_DECR_WRAP						0x00008508
+#define  NV20TCL_STENCIL_OP_ZPASS							0x00000378
+#define   NV20TCL_STENCIL_OP_ZPASS_ZERO							0x00000000
+#define   NV20TCL_STENCIL_OP_ZPASS_INVERT						0x0000150a
+#define   NV20TCL_STENCIL_OP_ZPASS_KEEP							0x00001e00
+#define   NV20TCL_STENCIL_OP_ZPASS_REPLACE						0x00001e01
+#define   NV20TCL_STENCIL_OP_ZPASS_INCR							0x00001e02
+#define   NV20TCL_STENCIL_OP_ZPASS_DECR							0x00001e03
+#define   NV20TCL_STENCIL_OP_ZPASS_INCR_WRAP						0x00008507
+#define   NV20TCL_STENCIL_OP_ZPASS_DECR_WRAP						0x00008508
+#define  NV20TCL_SHADE_MODEL								0x0000037c
+#define   NV20TCL_SHADE_MODEL_FLAT							0x00001d00
+#define   NV20TCL_SHADE_MODEL_SMOOTH							0x00001d01
+#define  NV20TCL_LINE_WIDTH								0x00000380
+#define  NV20TCL_POLYGON_OFFSET_FACTOR							0x00000384
+#define  NV20TCL_POLYGON_OFFSET_UNITS							0x00000388
+#define  NV20TCL_POLYGON_MODE_FRONT							0x0000038c
+#define   NV20TCL_POLYGON_MODE_FRONT_POINT						0x00001b00
+#define   NV20TCL_POLYGON_MODE_FRONT_LINE						0x00001b01
+#define   NV20TCL_POLYGON_MODE_FRONT_FILL						0x00001b02
+#define  NV20TCL_POLYGON_MODE_BACK							0x00000390
+#define   NV20TCL_POLYGON_MODE_BACK_POINT						0x00001b00
+#define   NV20TCL_POLYGON_MODE_BACK_LINE						0x00001b01
+#define   NV20TCL_POLYGON_MODE_BACK_FILL						0x00001b02
+#define  NV20TCL_DEPTH_RANGE_NEAR							0x00000394
+#define  NV20TCL_DEPTH_RANGE_FAR							0x00000398
+#define  NV20TCL_CULL_FACE								0x0000039c
+#define   NV20TCL_CULL_FACE_FRONT							0x00000404
+#define   NV20TCL_CULL_FACE_BACK							0x00000405
+#define   NV20TCL_CULL_FACE_FRONT_AND_BACK						0x00000408
+#define  NV20TCL_FRONT_FACE								0x000003a0
+#define   NV20TCL_FRONT_FACE_CW								0x00000900
+#define   NV20TCL_FRONT_FACE_CCW							0x00000901
+#define  NV20TCL_NORMALIZE_ENABLE							0x000003a4
+#define  NV20TCL_COLOR_MATERIAL_FRONT_R							0x000003a8
+#define  NV20TCL_COLOR_MATERIAL_FRONT_G							0x000003ac
+#define  NV20TCL_COLOR_MATERIAL_FRONT_B							0x000003b0
+#define  NV20TCL_COLOR_MATERIAL_FRONT_A							0x000003b4
+#define  NV20TCL_SEPARATE_SPECULAR_ENABLE						0x000003b8
+#define  NV20TCL_ENABLED_LIGHTS								0x000003bc
+#define  NV20TCL_TX_GEN_S(x)								(0x000003c0+((x)*16))
+#define  NV20TCL_TX_GEN_S__SIZE								0x00000004
+#define   NV20TCL_TX_GEN_S_FALSE							0x00000000
+#define   NV20TCL_TX_GEN_S_EYE_LINEAR							0x00002400
+#define   NV20TCL_TX_GEN_S_OBJECT_LINEAR						0x00002401
+#define   NV20TCL_TX_GEN_S_SPHERE_MAP							0x00002402
+#define   NV20TCL_TX_GEN_S_NORMAL_MAP							0x00008511
+#define   NV20TCL_TX_GEN_S_REFLECTION_MAP						0x00008512
+#define  NV20TCL_TX_GEN_T(x)								(0x000003c4+((x)*16))
+#define  NV20TCL_TX_GEN_T__SIZE								0x00000004
+#define   NV20TCL_TX_GEN_T_FALSE							0x00000000
+#define   NV20TCL_TX_GEN_T_EYE_LINEAR							0x00002400
+#define   NV20TCL_TX_GEN_T_OBJECT_LINEAR						0x00002401
+#define   NV20TCL_TX_GEN_T_SPHERE_MAP							0x00002402
+#define   NV20TCL_TX_GEN_T_NORMAL_MAP							0x00008511
+#define   NV20TCL_TX_GEN_T_REFLECTION_MAP						0x00008512
+#define  NV20TCL_TX_GEN_R(x)								(0x000003c8+((x)*16))
+#define  NV20TCL_TX_GEN_R__SIZE								0x00000004
+#define   NV20TCL_TX_GEN_R_FALSE							0x00000000
+#define   NV20TCL_TX_GEN_R_EYE_LINEAR							0x00002400
+#define   NV20TCL_TX_GEN_R_OBJECT_LINEAR						0x00002401
+#define   NV20TCL_TX_GEN_R_SPHERE_MAP							0x00002402
+#define   NV20TCL_TX_GEN_R_NORMAL_MAP							0x00008511
+#define   NV20TCL_TX_GEN_R_REFLECTION_MAP						0x00008512
+#define  NV20TCL_TX_GEN_Q(x)								(0x000003cc+((x)*16))
+#define  NV20TCL_TX_GEN_Q__SIZE								0x00000004
+#define   NV20TCL_TX_GEN_Q_FALSE							0x00000000
+#define   NV20TCL_TX_GEN_Q_EYE_LINEAR							0x00002400
+#define   NV20TCL_TX_GEN_Q_OBJECT_LINEAR						0x00002401
+#define   NV20TCL_TX_GEN_Q_SPHERE_MAP							0x00002402
+#define   NV20TCL_TX_GEN_Q_NORMAL_MAP							0x00008511
+#define   NV20TCL_TX_GEN_Q_REFLECTION_MAP						0x00008512
+#define  NV20TCL_TX_MATRIX_ENABLE(x)							(0x00000420+((x)*4))
+#define  NV20TCL_TX_MATRIX_ENABLE__SIZE							0x00000004
+#define  NV20TCL_POINT_SIZE								0x0000043c
+#define  NV20TCL_MODELVIEW0_MATRIX(x)							(0x00000480+((x)*4))
+#define  NV20TCL_MODELVIEW0_MATRIX__SIZE						0x00000010
+#define  NV20TCL_MODELVIEW1_MATRIX(x)							(0x000004c0+((x)*4))
+#define  NV20TCL_MODELVIEW1_MATRIX__SIZE						0x00000010
+#define  NV20TCL_MODELVIEW2_MATRIX(x)							(0x00000500+((x)*4))
+#define  NV20TCL_MODELVIEW2_MATRIX__SIZE						0x00000010
+#define  NV20TCL_MODELVIEW3_MATRIX(x)							(0x00000540+((x)*4))
+#define  NV20TCL_MODELVIEW3_MATRIX__SIZE						0x00000010
+#define  NV20TCL_INVERSE_MODELVIEW0_MATRIX(x)						(0x00000580+((x)*4))
+#define  NV20TCL_INVERSE_MODELVIEW0_MATRIX__SIZE					0x00000010
+#define  NV20TCL_INVERSE_MODELVIEW1_MATRIX(x)						(0x000005c0+((x)*4))
+#define  NV20TCL_INVERSE_MODELVIEW1_MATRIX__SIZE					0x00000010
+#define  NV20TCL_INVERSE_MODELVIEW2_MATRIX(x)						(0x00000600+((x)*4))
+#define  NV20TCL_INVERSE_MODELVIEW2_MATRIX__SIZE					0x00000010
+#define  NV20TCL_INVERSE_MODELVIEW3_MATRIX(x)						(0x00000640+((x)*4))
+#define  NV20TCL_INVERSE_MODELVIEW3_MATRIX__SIZE					0x00000010
+#define  NV20TCL_PROJECTION_MATRIX(x)							(0x00000680+((x)*4))
+#define  NV20TCL_PROJECTION_MATRIX__SIZE						0x00000010
+#define  NV20TCL_TX0_MATRIX(x)								(0x000006c0+((x)*4))
+#define  NV20TCL_TX0_MATRIX__SIZE							0x00000010
+#define  NV20TCL_TX1_MATRIX(x)								(0x00000700+((x)*4))
+#define  NV20TCL_TX1_MATRIX__SIZE							0x00000010
+#define  NV20TCL_TX2_MATRIX(x)								(0x00000740+((x)*4))
+#define  NV20TCL_TX2_MATRIX__SIZE							0x00000010
+#define  NV20TCL_TX3_MATRIX(x)								(0x00000780+((x)*4))
+#define  NV20TCL_TX3_MATRIX__SIZE							0x00000010
+#define  NV20TCL_TX0_CLIP_PLANE_A(x)							(0x00000840+((x)*16))
+#define  NV20TCL_TX0_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV20TCL_TX0_CLIP_PLANE_B(x)							(0x00000844+((x)*16))
+#define  NV20TCL_TX0_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV20TCL_TX0_CLIP_PLANE_C(x)							(0x00000848+((x)*16))
+#define  NV20TCL_TX0_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV20TCL_TX0_CLIP_PLANE_D(x)							(0x0000084c+((x)*16))
+#define  NV20TCL_TX0_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV20TCL_TX1_CLIP_PLANE_A(x)							(0x00000880+((x)*16))
+#define  NV20TCL_TX1_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV20TCL_TX1_CLIP_PLANE_B(x)							(0x00000884+((x)*16))
+#define  NV20TCL_TX1_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV20TCL_TX1_CLIP_PLANE_C(x)							(0x00000888+((x)*16))
+#define  NV20TCL_TX1_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV20TCL_TX1_CLIP_PLANE_D(x)							(0x0000088c+((x)*16))
+#define  NV20TCL_TX1_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV20TCL_TX2_CLIP_PLANE_A(x)							(0x000008c0+((x)*16))
+#define  NV20TCL_TX2_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV20TCL_TX2_CLIP_PLANE_B(x)							(0x000008c4+((x)*16))
+#define  NV20TCL_TX2_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV20TCL_TX2_CLIP_PLANE_C(x)							(0x000008c8+((x)*16))
+#define  NV20TCL_TX2_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV20TCL_TX2_CLIP_PLANE_D(x)							(0x000008cc+((x)*16))
+#define  NV20TCL_TX2_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV20TCL_TX3_CLIP_PLANE_A(x)							(0x00000900+((x)*16))
+#define  NV20TCL_TX3_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV20TCL_TX3_CLIP_PLANE_B(x)							(0x00000904+((x)*16))
+#define  NV20TCL_TX3_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV20TCL_TX3_CLIP_PLANE_C(x)							(0x00000908+((x)*16))
+#define  NV20TCL_TX3_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV20TCL_TX3_CLIP_PLANE_D(x)							(0x0000090c+((x)*16))
+#define  NV20TCL_TX3_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV20TCL_FOG_EQUATION_CONSTANT							0x000009c0
+#define  NV20TCL_FOG_EQUATION_LINEAR							0x000009c4
+#define  NV20TCL_FOG_EQUATION_QUADRATIC							0x000009c8
+#define  NV20TCL_FRONT_MATERIAL_SHININESS(x)						(0x000009e0+((x)*4))
+#define  NV20TCL_FRONT_MATERIAL_SHININESS__SIZE						0x00000006
+#define  NV20TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R			0x00000a10
+#define  NV20TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G			0x00000a14
+#define  NV20TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B			0x00000a18
+#define  NV20TCL_VIEWPORT_SCALE0_X							0x00000a20
+#define  NV20TCL_VIEWPORT_SCALE0_Y							0x00000a24
+#define  NV20TCL_VIEWPORT_SCALE0_Z							0x00000a28
+#define  NV20TCL_VIEWPORT_SCALE0_W							0x00000a2c
+#define  NV20TCL_POINT_PARAMETER(x)							(0x00000a30+((x)*4))
+#define  NV20TCL_POINT_PARAMETER__SIZE							0x00000008
+#define  NV20TCL_RC_CONSTANT_COLOR0(x)							(0x00000a60+((x)*4))
+#define  NV20TCL_RC_CONSTANT_COLOR0__SIZE						0x00000008
+#define   NV20TCL_RC_CONSTANT_COLOR0_B_SHIFT						0
+#define   NV20TCL_RC_CONSTANT_COLOR0_B_MASK						0x000000ff
+#define   NV20TCL_RC_CONSTANT_COLOR0_G_SHIFT						8
+#define   NV20TCL_RC_CONSTANT_COLOR0_G_MASK						0x0000ff00
+#define   NV20TCL_RC_CONSTANT_COLOR0_R_SHIFT						16
+#define   NV20TCL_RC_CONSTANT_COLOR0_R_MASK						0x00ff0000
+#define   NV20TCL_RC_CONSTANT_COLOR0_A_SHIFT						24
+#define   NV20TCL_RC_CONSTANT_COLOR0_A_MASK						0xff000000
+#define  NV20TCL_RC_CONSTANT_COLOR1(x)							(0x00000a80+((x)*4))
+#define  NV20TCL_RC_CONSTANT_COLOR1__SIZE						0x00000008
+#define   NV20TCL_RC_CONSTANT_COLOR1_B_SHIFT						0
+#define   NV20TCL_RC_CONSTANT_COLOR1_B_MASK						0x000000ff
+#define   NV20TCL_RC_CONSTANT_COLOR1_G_SHIFT						8
+#define   NV20TCL_RC_CONSTANT_COLOR1_G_MASK						0x0000ff00
+#define   NV20TCL_RC_CONSTANT_COLOR1_R_SHIFT						16
+#define   NV20TCL_RC_CONSTANT_COLOR1_R_MASK						0x00ff0000
+#define   NV20TCL_RC_CONSTANT_COLOR1_A_SHIFT						24
+#define   NV20TCL_RC_CONSTANT_COLOR1_A_MASK						0xff000000
+#define  NV20TCL_RC_OUT_ALPHA(x)							(0x00000aa0+((x)*4))
+#define  NV20TCL_RC_OUT_ALPHA__SIZE							0x00000008
+#define   NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT						0
+#define   NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK						0x0000000f
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG						0x00000003
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x0000000e
+#define    NV20TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT						4
+#define   NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK						0x000000f0
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG						0x00000030
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x000000e0
+#define    NV20TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT						8
+#define   NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK						0x00000f00
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG						0x00000300
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x00000e00
+#define    NV20TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT						(1 << 12)
+#define   NV20TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT						(1 << 13)
+#define   NV20TCL_RC_OUT_ALPHA_MUX_SUM							(1 << 14)
+#define   NV20TCL_RC_OUT_ALPHA_BIAS							(1 << 15)
+#define    NV20TCL_RC_OUT_ALPHA_BIAS_NONE						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV			0x00008000
+#define   NV20TCL_RC_OUT_ALPHA_SCALE_SHIFT						17
+#define   NV20TCL_RC_OUT_ALPHA_SCALE_MASK						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_SCALE_NONE						0x00000000
+#define    NV20TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV20TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV20TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define  NV20TCL_RC_IN_RGB(x)								(0x00000ac0+((x)*4))
+#define  NV20TCL_RC_IN_RGB__SIZE							0x00000008
+#define   NV20TCL_RC_IN_RGB_D_INPUT_SHIFT						0
+#define   NV20TCL_RC_IN_RGB_D_INPUT_MASK						0x0000000f
+#define    NV20TCL_RC_IN_RGB_D_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV20TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV20TCL_RC_IN_RGB_D_INPUT_FOG						0x00000003
+#define    NV20TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV20TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV20TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV20TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV20TCL_RC_IN_RGB_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV20TCL_RC_IN_RGB_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV20TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV20TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV20TCL_RC_IN_RGB_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV20TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV20TCL_RC_IN_RGB_D_MAPPING_SHIFT						5
+#define   NV20TCL_RC_IN_RGB_D_MAPPING_MASK						0x000000e0
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV20TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV20TCL_RC_IN_RGB_C_INPUT_SHIFT						8
+#define   NV20TCL_RC_IN_RGB_C_INPUT_MASK						0x00000f00
+#define    NV20TCL_RC_IN_RGB_C_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV20TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV20TCL_RC_IN_RGB_C_INPUT_FOG						0x00000300
+#define    NV20TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV20TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV20TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_IN_RGB_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV20TCL_RC_IN_RGB_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV20TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV20TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_IN_RGB_C_COMPONENT_USAGE						(1 << 12)
+#define    NV20TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV20TCL_RC_IN_RGB_C_MAPPING_SHIFT						13
+#define   NV20TCL_RC_IN_RGB_C_MAPPING_MASK						0x0000e000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV20TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV20TCL_RC_IN_RGB_B_INPUT_SHIFT						16
+#define   NV20TCL_RC_IN_RGB_B_INPUT_MASK						0x000f0000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_FOG						0x00030000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV20TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV20TCL_RC_IN_RGB_B_COMPONENT_USAGE						(1 << 20)
+#define    NV20TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV20TCL_RC_IN_RGB_B_MAPPING_SHIFT						21
+#define   NV20TCL_RC_IN_RGB_B_MAPPING_MASK						0x00e00000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV20TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV20TCL_RC_IN_RGB_A_INPUT_SHIFT						24
+#define   NV20TCL_RC_IN_RGB_A_INPUT_MASK						0x0f000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_ZERO						0x00000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_FOG						0x03000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV20TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV20TCL_RC_IN_RGB_A_COMPONENT_USAGE						(1 << 28)
+#define    NV20TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV20TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV20TCL_RC_IN_RGB_A_MAPPING_SHIFT						29
+#define   NV20TCL_RC_IN_RGB_A_MAPPING_MASK						0xe0000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV20TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV20TCL_VIEWPORT_SCALE1_X							0x00000af0
+#define  NV20TCL_VIEWPORT_SCALE1_Y							0x00000af4
+#define  NV20TCL_VIEWPORT_SCALE1_Z							0x00000af8
+#define  NV20TCL_VIEWPORT_SCALE1_W							0x00000afc
+#define  NV20TCL_VP_UPLOAD_INST(x)							(0x00000b00+((x)*4))
+#define  NV20TCL_VP_UPLOAD_INST__SIZE							0x00000004
+#define  NV20TCL_VP_UPLOAD_CONST(x)							(0x00000b80+((x)*4))
+#define  NV20TCL_VP_UPLOAD_CONST__SIZE							0x00000004
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_R(x)					(0x00000c00+((x)*64))
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_R__SIZE				0x00000008
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_G(x)					(0x00000c04+((x)*64))
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_G__SIZE				0x00000008
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_B(x)					(0x00000c08+((x)*64))
+#define  NV20TCL_LIGHT_BACK_SIDE_PRODUCT_AMBIENT_B__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x)					(0x00001000+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x)					(0x00001004+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x)					(0x00001008+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x)					(0x0000100c+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x)					(0x00001010+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x)					(0x00001014+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x)					(0x00001018+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x)					(0x0000101c+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE				0x00000008
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x)					(0x00001020+((x)*128))
+#define  NV20TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE				0x00000008
+#define  NV20TCL_LIGHT_HALF_VECTOR_X(x)							(0x00001028+((x)*128))
+#define  NV20TCL_LIGHT_HALF_VECTOR_X__SIZE						0x00000008
+#define  NV20TCL_LIGHT_HALF_VECTOR_Y(x)							(0x0000102c+((x)*128))
+#define  NV20TCL_LIGHT_HALF_VECTOR_Y__SIZE						0x00000008
+#define  NV20TCL_LIGHT_HALF_VECTOR_Z(x)							(0x00001030+((x)*128))
+#define  NV20TCL_LIGHT_HALF_VECTOR_Z__SIZE						0x00000008
+#define  NV20TCL_LIGHT_DIRECTION_X(x)							(0x00001034+((x)*128))
+#define  NV20TCL_LIGHT_DIRECTION_X__SIZE						0x00000008
+#define  NV20TCL_LIGHT_DIRECTION_Y(x)							(0x00001038+((x)*128))
+#define  NV20TCL_LIGHT_DIRECTION_Y__SIZE						0x00000008
+#define  NV20TCL_LIGHT_DIRECTION_Z(x)							(0x0000103c+((x)*128))
+#define  NV20TCL_LIGHT_DIRECTION_Z__SIZE						0x00000008
+#define  NV20TCL_LIGHT_POSITION_X(x)							(0x0000105c+((x)*128))
+#define  NV20TCL_LIGHT_POSITION_X__SIZE							0x00000008
+#define  NV20TCL_LIGHT_POSITION_Y(x)							(0x00001060+((x)*128))
+#define  NV20TCL_LIGHT_POSITION_Y__SIZE							0x00000008
+#define  NV20TCL_LIGHT_POSITION_Z(x)							(0x00001064+((x)*128))
+#define  NV20TCL_LIGHT_POSITION_Z__SIZE							0x00000008
+#define  NV20TCL_LIGHT_CONSTANT_ATTENUATION(x)						(0x00001068+((x)*128))
+#define  NV20TCL_LIGHT_CONSTANT_ATTENUATION__SIZE					0x00000008
+#define  NV20TCL_LIGHT_LINEAR_ATTENUATION(x)						(0x0000106c+((x)*128))
+#define  NV20TCL_LIGHT_LINEAR_ATTENUATION__SIZE						0x00000008
+#define  NV20TCL_LIGHT_QUADRATIC_ATTENUATION(x)						(0x00001070+((x)*128))
+#define  NV20TCL_LIGHT_QUADRATIC_ATTENUATION__SIZE					0x00000008
+#define  NV20TCL_POLYGON_STIPPLE_ENABLE							0x0000147c
+#define  NV20TCL_POLYGON_STIPPLE_PATTERN(x)						(0x00001480+((x)*4))
+#define  NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE						0x00000020
+#define  NV20TCL_VERTEX_POS_3F_X							0x00001500
+#define  NV20TCL_VERTEX_POS_3F_Y							0x00001504
+#define  NV20TCL_VERTEX_POS_3F_Z							0x00001508
+#define  NV20TCL_VERTEX_POS_4F_X							0x00001518
+#define  NV20TCL_VERTEX_POS_4F_Y							0x0000151c
+#define  NV20TCL_VERTEX_POS_4F_Z							0x00001520
+#define  NV20TCL_VERTEX_POS_3I_XY							0x00001528
+#define   NV20TCL_VERTEX_POS_3I_XY_X_SHIFT						0
+#define   NV20TCL_VERTEX_POS_3I_XY_X_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_POS_3I_XY_Y_SHIFT						16
+#define   NV20TCL_VERTEX_POS_3I_XY_Y_MASK						0xffff0000
+#define  NV20TCL_VERTEX_POS_3I_Z							0x0000152c
+#define   NV20TCL_VERTEX_POS_3I_Z_Z_SHIFT						0
+#define   NV20TCL_VERTEX_POS_3I_Z_Z_MASK						0x0000ffff
+#define  NV20TCL_VERTEX_NOR_3F_X							0x00001530
+#define  NV20TCL_VERTEX_NOR_3F_Y							0x00001534
+#define  NV20TCL_VERTEX_NOR_3F_Z							0x00001538
+#define  NV20TCL_VERTEX_NOR_3I_XY							0x00001540
+#define   NV20TCL_VERTEX_NOR_3I_XY_X_SHIFT						0
+#define   NV20TCL_VERTEX_NOR_3I_XY_X_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_NOR_3I_XY_Y_SHIFT						16
+#define   NV20TCL_VERTEX_NOR_3I_XY_Y_MASK						0xffff0000
+#define  NV20TCL_VERTEX_NOR_3I_Z							0x00001544
+#define   NV20TCL_VERTEX_NOR_3I_Z_Z_SHIFT						0
+#define   NV20TCL_VERTEX_NOR_3I_Z_Z_MASK						0x0000ffff
+#define  NV20TCL_VERTEX_COL_4F_X							0x00001550
+#define  NV20TCL_VERTEX_COL_4F_Y							0x00001554
+#define  NV20TCL_VERTEX_COL_4F_Z							0x00001558
+#define  NV20TCL_VERTEX_COL_4F_W							0x0000155c
+#define  NV20TCL_VERTEX_COL_3F_X							0x00001560
+#define  NV20TCL_VERTEX_COL_3F_Y							0x00001564
+#define  NV20TCL_VERTEX_COL_3F_Z							0x00001568
+#define  NV20TCL_VERTEX_COL_4I								0x0000156c
+#define   NV20TCL_VERTEX_COL_4I_R_SHIFT							0
+#define   NV20TCL_VERTEX_COL_4I_R_MASK							0x000000ff
+#define   NV20TCL_VERTEX_COL_4I_G_SHIFT							8
+#define   NV20TCL_VERTEX_COL_4I_G_MASK							0x0000ff00
+#define   NV20TCL_VERTEX_COL_4I_B_SHIFT							16
+#define   NV20TCL_VERTEX_COL_4I_B_MASK							0x00ff0000
+#define   NV20TCL_VERTEX_COL_4I_A_SHIFT							24
+#define   NV20TCL_VERTEX_COL_4I_A_MASK							0xff000000
+#define  NV20TCL_VERTEX_COL2_3F_X							0x00001580
+#define  NV20TCL_VERTEX_COL2_3F_Y							0x00001584
+#define  NV20TCL_VERTEX_COL2_3F_Z							0x00001588
+#define  NV20TCL_VERTEX_COL2_4I								0x0000158c
+#define   NV20TCL_VERTEX_COL2_4I_R_SHIFT						0
+#define   NV20TCL_VERTEX_COL2_4I_R_MASK							0x000000ff
+#define   NV20TCL_VERTEX_COL2_4I_G_SHIFT						8
+#define   NV20TCL_VERTEX_COL2_4I_G_MASK							0x0000ff00
+#define   NV20TCL_VERTEX_COL2_4I_B_SHIFT						16
+#define   NV20TCL_VERTEX_COL2_4I_B_MASK							0x00ff0000
+#define   NV20TCL_VERTEX_COL2_4I_A_SHIFT						24
+#define   NV20TCL_VERTEX_COL2_4I_A_MASK							0xff000000
+#define  NV20TCL_VERTEX_TX0_2F_S							0x00001590
+#define  NV20TCL_VERTEX_TX0_2F_T							0x00001594
+#define  NV20TCL_VERTEX_TX0_2I								0x00001598
+#define   NV20TCL_VERTEX_TX0_2I_S_SHIFT							0
+#define   NV20TCL_VERTEX_TX0_2I_S_MASK							0x0000ffff
+#define   NV20TCL_VERTEX_TX0_2I_T_SHIFT							16
+#define   NV20TCL_VERTEX_TX0_2I_T_MASK							0xffff0000
+#define  NV20TCL_VERTEX_TX0_4F_S							0x000015a0
+#define  NV20TCL_VERTEX_TX0_4F_T							0x000015a4
+#define  NV20TCL_VERTEX_TX0_4F_R							0x000015a8
+#define  NV20TCL_VERTEX_TX0_4F_Q							0x000015ac
+#define  NV20TCL_VERTEX_TX0_4I_ST							0x000015b0
+#define   NV20TCL_VERTEX_TX0_4I_ST_S_SHIFT						0
+#define   NV20TCL_VERTEX_TX0_4I_ST_S_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX0_4I_ST_T_SHIFT						16
+#define   NV20TCL_VERTEX_TX0_4I_ST_T_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX0_4I_RQ							0x000015b4
+#define   NV20TCL_VERTEX_TX0_4I_RQ_R_SHIFT						0
+#define   NV20TCL_VERTEX_TX0_4I_RQ_R_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX0_4I_RQ_Q_SHIFT						16
+#define   NV20TCL_VERTEX_TX0_4I_RQ_Q_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX1_2F_S							0x000015b8
+#define  NV20TCL_VERTEX_TX1_2F_T							0x000015bc
+#define  NV20TCL_VERTEX_TX1_2I								0x000015c0
+#define   NV20TCL_VERTEX_TX1_2I_S_SHIFT							0
+#define   NV20TCL_VERTEX_TX1_2I_S_MASK							0x0000ffff
+#define   NV20TCL_VERTEX_TX1_2I_T_SHIFT							16
+#define   NV20TCL_VERTEX_TX1_2I_T_MASK							0xffff0000
+#define  NV20TCL_VERTEX_TX1_4F_S							0x000015c8
+#define  NV20TCL_VERTEX_TX1_4F_T							0x000015cc
+#define  NV20TCL_VERTEX_TX1_4F_R							0x000015d0
+#define  NV20TCL_VERTEX_TX1_4F_Q							0x000015d4
+#define  NV20TCL_VERTEX_TX1_4I_ST							0x000015d8
+#define   NV20TCL_VERTEX_TX1_4I_ST_S_SHIFT						0
+#define   NV20TCL_VERTEX_TX1_4I_ST_S_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX1_4I_ST_T_SHIFT						16
+#define   NV20TCL_VERTEX_TX1_4I_ST_T_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX1_4I_RQ							0x000015dc
+#define   NV20TCL_VERTEX_TX1_4I_RQ_R_SHIFT						0
+#define   NV20TCL_VERTEX_TX1_4I_RQ_R_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX1_4I_RQ_Q_SHIFT						16
+#define   NV20TCL_VERTEX_TX1_4I_RQ_Q_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX2_2F_S							0x000015e0
+#define  NV20TCL_VERTEX_TX2_2F_T							0x000015e4
+#define  NV20TCL_VERTEX_TX2_2I								0x000015e8
+#define   NV20TCL_VERTEX_TX2_2I_S_SHIFT							0
+#define   NV20TCL_VERTEX_TX2_2I_S_MASK							0x0000ffff
+#define   NV20TCL_VERTEX_TX2_2I_T_SHIFT							16
+#define   NV20TCL_VERTEX_TX2_2I_T_MASK							0xffff0000
+#define  NV20TCL_VERTEX_TX2_4F_S							0x000015f0
+#define  NV20TCL_VERTEX_TX2_4F_T							0x000015f4
+#define  NV20TCL_VERTEX_TX2_4F_R							0x000015f8
+#define  NV20TCL_VERTEX_TX2_4F_Q							0x000015fc
+#define  NV20TCL_VERTEX_TX2_4I_ST							0x00001600
+#define   NV20TCL_VERTEX_TX2_4I_ST_S_SHIFT						0
+#define   NV20TCL_VERTEX_TX2_4I_ST_S_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX2_4I_ST_T_SHIFT						16
+#define   NV20TCL_VERTEX_TX2_4I_ST_T_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX2_4I_RQ							0x00001604
+#define   NV20TCL_VERTEX_TX2_4I_RQ_R_SHIFT						0
+#define   NV20TCL_VERTEX_TX2_4I_RQ_R_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX2_4I_RQ_Q_SHIFT						16
+#define   NV20TCL_VERTEX_TX2_4I_RQ_Q_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX3_2F_S							0x00001608
+#define  NV20TCL_VERTEX_TX3_2F_T							0x0000160c
+#define  NV20TCL_VERTEX_TX3_2I								0x00001610
+#define   NV20TCL_VERTEX_TX3_2I_S_SHIFT							0
+#define   NV20TCL_VERTEX_TX3_2I_S_MASK							0x0000ffff
+#define   NV20TCL_VERTEX_TX3_2I_T_SHIFT							16
+#define   NV20TCL_VERTEX_TX3_2I_T_MASK							0xffff0000
+#define  NV20TCL_VERTEX_TX3_4F_S							0x00001620
+#define  NV20TCL_VERTEX_TX3_4F_T							0x00001624
+#define  NV20TCL_VERTEX_TX3_4F_R							0x00001628
+#define  NV20TCL_VERTEX_TX3_4F_Q							0x0000162c
+#define  NV20TCL_VERTEX_TX3_4I_ST							0x00001630
+#define   NV20TCL_VERTEX_TX3_4I_ST_S_SHIFT						0
+#define   NV20TCL_VERTEX_TX3_4I_ST_S_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX3_4I_ST_T_SHIFT						16
+#define   NV20TCL_VERTEX_TX3_4I_ST_T_MASK						0xffff0000
+#define  NV20TCL_VERTEX_TX3_4I_RQ							0x00001634
+#define   NV20TCL_VERTEX_TX3_4I_RQ_R_SHIFT						0
+#define   NV20TCL_VERTEX_TX3_4I_RQ_R_MASK						0x0000ffff
+#define   NV20TCL_VERTEX_TX3_4I_RQ_Q_SHIFT						16
+#define   NV20TCL_VERTEX_TX3_4I_RQ_Q_MASK						0xffff0000
+#define  NV20TCL_VERTEX_FOG_1F								0x00001698
+#define  NV20TCL_EDGEFLAG_ENABLE							0x000016bc
+#define  NV20TCL_VTXBUF_ADDRESS(x)							(0x00001720+((x)*4))
+#define  NV20TCL_VTXBUF_ADDRESS__SIZE							0x00000010
+#define   NV20TCL_VTXBUF_ADDRESS_DMA1							(1 << 31)
+#define   NV20TCL_VTXBUF_ADDRESS_OFFSET_SHIFT						0
+#define   NV20TCL_VTXBUF_ADDRESS_OFFSET_MASK						0x0fffffff
+#define  NV20TCL_VTXFMT(x)								(0x00001760+((x)*4))
+#define  NV20TCL_VTXFMT__SIZE								0x00000010
+#define   NV20TCL_VTXFMT_TYPE_SHIFT							0
+#define   NV20TCL_VTXFMT_TYPE_MASK							0x0000000f
+#define    NV20TCL_VTXFMT_TYPE_FLOAT							0x00000002
+#define    NV20TCL_VTXFMT_TYPE_UBYTE							0x00000004
+#define    NV20TCL_VTXFMT_TYPE_USHORT							0x00000005
+#define   NV20TCL_VTXFMT_SIZE_SHIFT							4
+#define   NV20TCL_VTXFMT_SIZE_MASK							0x000000f0
+#define   NV20TCL_VTXFMT_STRIDE_SHIFT							8
+#define   NV20TCL_VTXFMT_STRIDE_MASK							0x0000ff00
+#define  NV20TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R			0x000017a0
+#define  NV20TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G			0x000017a4
+#define  NV20TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B			0x000017a8
+#define  NV20TCL_COLOR_MATERIAL_BACK_A							0x000017ac
+#define  NV20TCL_COLOR_MATERIAL_BACK_R							0x000017b0
+#define  NV20TCL_COLOR_MATERIAL_BACK_G							0x000017b4
+#define  NV20TCL_COLOR_MATERIAL_BACK_B							0x000017b8
+#define  NV20TCL_COLOR_LOGIC_OP_ENABLE							0x000017bc
+#define  NV20TCL_COLOR_LOGIC_OP_OP							0x000017c0
+#define   NV20TCL_COLOR_LOGIC_OP_OP_CLEAR						0x00001500
+#define   NV20TCL_COLOR_LOGIC_OP_OP_AND							0x00001501
+#define   NV20TCL_COLOR_LOGIC_OP_OP_AND_REVERSE						0x00001502
+#define   NV20TCL_COLOR_LOGIC_OP_OP_COPY						0x00001503
+#define   NV20TCL_COLOR_LOGIC_OP_OP_AND_INVERTED					0x00001504
+#define   NV20TCL_COLOR_LOGIC_OP_OP_NOOP						0x00001505
+#define   NV20TCL_COLOR_LOGIC_OP_OP_XOR							0x00001506
+#define   NV20TCL_COLOR_LOGIC_OP_OP_OR							0x00001507
+#define   NV20TCL_COLOR_LOGIC_OP_OP_NOR							0x00001508
+#define   NV20TCL_COLOR_LOGIC_OP_OP_EQUIV						0x00001509
+#define   NV20TCL_COLOR_LOGIC_OP_OP_INVERT						0x0000150a
+#define   NV20TCL_COLOR_LOGIC_OP_OP_OR_REVERSE						0x0000150b
+#define   NV20TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED					0x0000150c
+#define   NV20TCL_COLOR_LOGIC_OP_OP_OR_INVERTED						0x0000150d
+#define   NV20TCL_COLOR_LOGIC_OP_OP_NAND						0x0000150e
+#define   NV20TCL_COLOR_LOGIC_OP_OP_SET							0x0000150f
+#define  NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE						0x000017c4
+#define  NV20TCL_TX_SHADER_CULL_MODE							0x000017f8
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX0_S						(1 <<  0)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_S_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_S_LESS					0x00000001
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX0_T						(1 <<  1)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_T_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_T_LESS					0x00000002
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX0_R						(1 <<  2)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_R_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_R_LESS					0x00000004
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX0_Q						(1 <<  3)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_Q_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX0_Q_LESS					0x00000008
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX1_S						(1 <<  4)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_S_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_S_LESS					0x00000010
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX1_T						(1 <<  5)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_T_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_T_LESS					0x00000020
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX1_R						(1 <<  6)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_R_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_R_LESS					0x00000040
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX1_Q						(1 <<  7)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_Q_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX1_Q_LESS					0x00000080
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX2_S						(1 <<  8)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_S_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_S_LESS					0x00000100
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX2_T						(1 <<  9)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_T_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_T_LESS					0x00000200
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX2_R						(1 << 10)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_R_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_R_LESS					0x00000400
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX2_Q						(1 << 11)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_Q_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX2_Q_LESS					0x00000800
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX3_S						(1 << 12)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_S_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_S_LESS					0x00001000
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX3_T						(1 << 13)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_T_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_T_LESS					0x00002000
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX3_R						(1 << 14)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_R_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_R_LESS					0x00004000
+#define   NV20TCL_TX_SHADER_CULL_MODE_TX3_Q						(1 << 15)
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_Q_GEQUAL					0x00000000
+#define    NV20TCL_TX_SHADER_CULL_MODE_TX3_Q_LESS					0x00008000
+#define  NV20TCL_VERTEX_BEGIN_END							0x000017fc
+#define   NV20TCL_VERTEX_BEGIN_END_STOP							0x00000000
+#define   NV20TCL_VERTEX_BEGIN_END_POINTS						0x00000001
+#define   NV20TCL_VERTEX_BEGIN_END_LINES						0x00000002
+#define   NV20TCL_VERTEX_BEGIN_END_LINE_LOOP						0x00000003
+#define   NV20TCL_VERTEX_BEGIN_END_LINE_STRIP						0x00000004
+#define   NV20TCL_VERTEX_BEGIN_END_TRIANGLES						0x00000005
+#define   NV20TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP					0x00000006
+#define   NV20TCL_VERTEX_BEGIN_END_TRIANGLE_FAN						0x00000007
+#define   NV20TCL_VERTEX_BEGIN_END_QUADS						0x00000008
+#define   NV20TCL_VERTEX_BEGIN_END_QUAD_STRIP						0x00000009
+#define   NV20TCL_VERTEX_BEGIN_END_POLYGON						0x0000000a
+#define  NV20TCL_VB_ELEMENT_U16								0x00001800
+#define   NV20TCL_VB_ELEMENT_U16_I0_SHIFT						0
+#define   NV20TCL_VB_ELEMENT_U16_I0_MASK						0x0000ffff
+#define   NV20TCL_VB_ELEMENT_U16_I1_SHIFT						16
+#define   NV20TCL_VB_ELEMENT_U16_I1_MASK						0xffff0000
+#define  NV20TCL_VB_VERTEX_BATCH							0x00001810
+#define   NV20TCL_VB_VERTEX_BATCH_OFFSET_SHIFT						0
+#define   NV20TCL_VB_VERTEX_BATCH_OFFSET_MASK						0x00ffffff
+#define   NV20TCL_VB_VERTEX_BATCH_COUNT_SHIFT						24
+#define   NV20TCL_VB_VERTEX_BATCH_COUNT_MASK						0xff000000
+#define  NV20TCL_VERTEX_DATA								0x00001818
+#define  NV20TCL_TX_SHADER_CONST_EYE_X							0x0000181c
+#define  NV20TCL_TX_SHADER_CONST_EYE_Y							0x00001820
+#define  NV20TCL_TX_SHADER_CONST_EYE_Z							0x00001824
+#define  NV20TCL_VTX_ATTR_4F_X(x)							(0x00001a00+((x)*16))
+#define  NV20TCL_VTX_ATTR_4F_X__SIZE							0x00000010
+#define  NV20TCL_VTX_ATTR_4F_Y(x)							(0x00001a04+((x)*16))
+#define  NV20TCL_VTX_ATTR_4F_Y__SIZE							0x00000010
+#define  NV20TCL_VTX_ATTR_4F_Z(x)							(0x00001a08+((x)*16))
+#define  NV20TCL_VTX_ATTR_4F_Z__SIZE							0x00000010
+#define  NV20TCL_VTX_ATTR_4F_W(x)							(0x00001a0c+((x)*16))
+#define  NV20TCL_VTX_ATTR_4F_W__SIZE							0x00000010
+#define  NV20TCL_TX_OFFSET(x)								(0x00001b00+((x)*64))
+#define  NV20TCL_TX_OFFSET__SIZE							0x00000004
+#define  NV20TCL_TX_FORMAT(x)								(0x00001b04+((x)*64))
+#define  NV20TCL_TX_FORMAT__SIZE							0x00000004
+#define   NV20TCL_TX_FORMAT_DMA0							(1 <<  0)
+#define   NV20TCL_TX_FORMAT_DMA1							(1 <<  1)
+#define   NV20TCL_TX_FORMAT_CUBIC							(1 <<  2)
+#define   NV20TCL_TX_FORMAT_NO_BORDER							(1 <<  3)
+#define   NV20TCL_TX_FORMAT_DIMS_SHIFT							4
+#define   NV20TCL_TX_FORMAT_DIMS_MASK							0x000000f0
+#define    NV20TCL_TX_FORMAT_DIMS_1D							0x00000010
+#define    NV20TCL_TX_FORMAT_DIMS_2D							0x00000020
+#define    NV20TCL_TX_FORMAT_DIMS_3D							0x00000030
+#define   NV20TCL_TX_FORMAT_FORMAT_SHIFT						8
+#define   NV20TCL_TX_FORMAT_FORMAT_MASK							0x0000ff00
+#define    NV20TCL_TX_FORMAT_FORMAT_L8							0x00000000
+#define    NV20TCL_TX_FORMAT_FORMAT_A8							0x00000100
+#define    NV20TCL_TX_FORMAT_FORMAT_A1R5G5B5						0x00000200
+#define    NV20TCL_TX_FORMAT_FORMAT_A8_RECT						0x00000300
+#define    NV20TCL_TX_FORMAT_FORMAT_A4R4G4B4						0x00000400
+#define    NV20TCL_TX_FORMAT_FORMAT_R5G6B5						0x00000500
+#define    NV20TCL_TX_FORMAT_FORMAT_A8R8G8B8						0x00000600
+#define    NV20TCL_TX_FORMAT_FORMAT_X8R8G8B8						0x00000700
+#define    NV20TCL_TX_FORMAT_FORMAT_INDEX8						0x00000b00
+#define    NV20TCL_TX_FORMAT_FORMAT_DXT1						0x00000c00
+#define    NV20TCL_TX_FORMAT_FORMAT_DXT3						0x00000e00
+#define    NV20TCL_TX_FORMAT_FORMAT_DXT5						0x00000f00
+#define    NV20TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT					0x00001000
+#define    NV20TCL_TX_FORMAT_FORMAT_R5G6B5_RECT						0x00001100
+#define    NV20TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT					0x00001200
+#define    NV20TCL_TX_FORMAT_FORMAT_L8_RECT						0x00001300
+#define    NV20TCL_TX_FORMAT_FORMAT_A8L8						0x00001a00
+#define    NV20TCL_TX_FORMAT_FORMAT_A8_RECT2						0x00001b00
+#define    NV20TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT					0x00001d00
+#define    NV20TCL_TX_FORMAT_FORMAT_R8G8B8_RECT						0x00001e00
+#define    NV20TCL_TX_FORMAT_FORMAT_L8A8_RECT						0x00002000
+#define    NV20TCL_TX_FORMAT_FORMAT_DSDT						0x00002800
+#define    NV20TCL_TX_FORMAT_FORMAT_A16							0x00003200
+#define    NV20TCL_TX_FORMAT_FORMAT_HILO16						0x00003300
+#define    NV20TCL_TX_FORMAT_FORMAT_A16_RECT						0x00003500
+#define    NV20TCL_TX_FORMAT_FORMAT_HILO16_RECT						0x00003600
+#define    NV20TCL_TX_FORMAT_FORMAT_HILO8						0x00004400
+#define    NV20TCL_TX_FORMAT_FORMAT_SIGNED_HILO8					0x00004500
+#define    NV20TCL_TX_FORMAT_FORMAT_HILO8_RECT						0x00004600
+#define    NV20TCL_TX_FORMAT_FORMAT_SIGNED_HILO8_RECT					0x00004700
+#define    NV20TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV					0x00004a00
+#define    NV20TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV					0x00004b00
+#define    NV20TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV					0x00004c00
+#define   NV20TCL_TX_FORMAT_MIPMAP							(1 << 19)
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_U_SHIFT						20
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_U_MASK						0x00f00000
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_V_SHIFT						24
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_V_MASK						0x0f000000
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_W_SHIFT						28
+#define   NV20TCL_TX_FORMAT_BASE_SIZE_W_MASK						0xf0000000
+#define  NV20TCL_TX_WRAP(x)								(0x00001b08+((x)*64))
+#define  NV20TCL_TX_WRAP__SIZE								0x00000004
+#define   NV20TCL_TX_WRAP_S_SHIFT							0
+#define   NV20TCL_TX_WRAP_S_MASK							0x000000ff
+#define    NV20TCL_TX_WRAP_S_REPEAT							0x00000001
+#define    NV20TCL_TX_WRAP_S_MIRRORED_REPEAT						0x00000002
+#define    NV20TCL_TX_WRAP_S_CLAMP_TO_EDGE						0x00000003
+#define    NV20TCL_TX_WRAP_S_CLAMP_TO_BORDER						0x00000004
+#define    NV20TCL_TX_WRAP_S_CLAMP							0x00000005
+#define   NV20TCL_TX_WRAP_T_SHIFT							8
+#define   NV20TCL_TX_WRAP_T_MASK							0x00000f00
+#define    NV20TCL_TX_WRAP_T_REPEAT							0x00000100
+#define    NV20TCL_TX_WRAP_T_MIRRORED_REPEAT						0x00000200
+#define    NV20TCL_TX_WRAP_T_CLAMP_TO_EDGE						0x00000300
+#define    NV20TCL_TX_WRAP_T_CLAMP_TO_BORDER						0x00000400
+#define    NV20TCL_TX_WRAP_T_CLAMP							0x00000500
+#define   NV20TCL_TX_WRAP_R_SHIFT							16
+#define   NV20TCL_TX_WRAP_R_MASK							0x000f0000
+#define    NV20TCL_TX_WRAP_R_REPEAT							0x00010000
+#define    NV20TCL_TX_WRAP_R_MIRRORED_REPEAT						0x00020000
+#define    NV20TCL_TX_WRAP_R_CLAMP_TO_EDGE						0x00030000
+#define    NV20TCL_TX_WRAP_R_CLAMP_TO_BORDER						0x00040000
+#define    NV20TCL_TX_WRAP_R_CLAMP							0x00050000
+#define  NV20TCL_TX_ENABLE(x)								(0x00001b0c+((x)*64))
+#define  NV20TCL_TX_ENABLE__SIZE							0x00000004
+#define   NV20TCL_TX_ENABLE_ANISO_SHIFT							4
+#define   NV20TCL_TX_ENABLE_ANISO_MASK							0x00000030
+#define    NV20TCL_TX_ENABLE_ANISO_NONE							0x00000000
+#define    NV20TCL_TX_ENABLE_ANISO_2X							0x00000010
+#define    NV20TCL_TX_ENABLE_ANISO_4X							0x00000020
+#define    NV20TCL_TX_ENABLE_ANISO_8X							0x00000030
+#define   NV20TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT					14
+#define   NV20TCL_TX_ENABLE_MIPMAP_MAX_LOD_MASK						0x0003c000
+#define   NV20TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT					26
+#define   NV20TCL_TX_ENABLE_MIPMAP_MIN_LOD_MASK						0x3c000000
+#define   NV20TCL_TX_ENABLE_ENABLE							(1 << 30)
+#define  NV20TCL_TX_SWIZZLE(x)								(0x00001b10+((x)*64))
+#define  NV20TCL_TX_SWIZZLE__SIZE							0x00000004
+#define   NV20TCL_TX_SWIZZLE_RECT_PITCH_SHIFT						16
+#define   NV20TCL_TX_SWIZZLE_RECT_PITCH_MASK						0xffff0000
+#define  NV20TCL_TX_FILTER(x)								(0x00001b14+((x)*64))
+#define  NV20TCL_TX_FILTER__SIZE							0x00000004
+#define   NV20TCL_TX_FILTER_LOD_BIAS_SHIFT						8
+#define   NV20TCL_TX_FILTER_LOD_BIAS_MASK						0x00000f00
+#define   NV20TCL_TX_FILTER_MINIFY_SHIFT						16
+#define   NV20TCL_TX_FILTER_MINIFY_MASK							0x000f0000
+#define    NV20TCL_TX_FILTER_MINIFY_NEAREST						0x00010000
+#define    NV20TCL_TX_FILTER_MINIFY_LINEAR						0x00020000
+#define    NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST				0x00030000
+#define    NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST				0x00040000
+#define    NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR				0x00050000
+#define    NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR				0x00060000
+#define   NV20TCL_TX_FILTER_MAGNIFY_SHIFT						24
+#define   NV20TCL_TX_FILTER_MAGNIFY_MASK						0x0f000000
+#define    NV20TCL_TX_FILTER_MAGNIFY_NEAREST						0x01000000
+#define    NV20TCL_TX_FILTER_MAGNIFY_LINEAR						0x02000000
+#define  NV20TCL_TX_NPOT_SIZE(x)							(0x00001b1c+((x)*64))
+#define  NV20TCL_TX_NPOT_SIZE__SIZE							0x00000004
+#define   NV20TCL_TX_NPOT_SIZE_H_SHIFT							0
+#define   NV20TCL_TX_NPOT_SIZE_H_MASK							0x0000ffff
+#define   NV20TCL_TX_NPOT_SIZE_W_SHIFT							16
+#define   NV20TCL_TX_NPOT_SIZE_W_MASK							0xffff0000
+#define  NV20TCL_TX_PALETTE_OFFSET(x)							(0x00001b20+((x)*64))
+#define  NV20TCL_TX_PALETTE_OFFSET__SIZE						0x00000004
+#define  NV20TCL_TX_BORDER_COLOR(x)							(0x00001b24+((x)*64))
+#define  NV20TCL_TX_BORDER_COLOR__SIZE							0x00000004
+#define   NV20TCL_TX_BORDER_COLOR_B_SHIFT						0
+#define   NV20TCL_TX_BORDER_COLOR_B_MASK						0x000000ff
+#define   NV20TCL_TX_BORDER_COLOR_G_SHIFT						8
+#define   NV20TCL_TX_BORDER_COLOR_G_MASK						0x0000ff00
+#define   NV20TCL_TX_BORDER_COLOR_R_SHIFT						16
+#define   NV20TCL_TX_BORDER_COLOR_R_MASK						0x00ff0000
+#define   NV20TCL_TX_BORDER_COLOR_A_SHIFT						24
+#define   NV20TCL_TX_BORDER_COLOR_A_MASK						0xff000000
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX00(x)						(0x00001b28+((x)*64))
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX00__SIZE					0x00000004
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX01(x)						(0x00001b2c+((x)*64))
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX01__SIZE					0x00000004
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX11(x)						(0x00001b30+((x)*64))
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX11__SIZE					0x00000004
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX10(x)						(0x00001b34+((x)*64))
+#define  NV20TCL_TX_SHADER_OFFSET_MATRIX10__SIZE					0x00000004
+#define  NV20TCL_DEPTH_UNK17D8								0x00001d78
+#define   NV20TCL_DEPTH_UNK17D8_CLAMP_SHIFT						4
+#define   NV20TCL_DEPTH_UNK17D8_CLAMP_MASK						0x000000f0
+#define  NV20TCL_MULTISAMPLE_CONTROL							0x00001d7c
+#define  NV20TCL_CLEAR_DEPTH_VALUE							0x00001d8c
+#define  NV20TCL_CLEAR_VALUE								0x00001d90
+#define  NV20TCL_CLEAR_BUFFERS								0x00001d94
+#define   NV20TCL_CLEAR_BUFFERS_COLOR_A							(1 <<  7)
+#define   NV20TCL_CLEAR_BUFFERS_COLOR_B							(1 <<  6)
+#define   NV20TCL_CLEAR_BUFFERS_COLOR_G							(1 <<  5)
+#define   NV20TCL_CLEAR_BUFFERS_COLOR_R							(1 <<  4)
+#define   NV20TCL_CLEAR_BUFFERS_STENCIL							(1 <<  1)
+#define   NV20TCL_CLEAR_BUFFERS_DEPTH							(1 <<  0)
+#define  NV20TCL_RC_COLOR0								0x00001e20
+#define   NV20TCL_RC_COLOR0_B_SHIFT							0
+#define   NV20TCL_RC_COLOR0_B_MASK							0x000000ff
+#define   NV20TCL_RC_COLOR0_G_SHIFT							8
+#define   NV20TCL_RC_COLOR0_G_MASK							0x0000ff00
+#define   NV20TCL_RC_COLOR0_R_SHIFT							16
+#define   NV20TCL_RC_COLOR0_R_MASK							0x00ff0000
+#define   NV20TCL_RC_COLOR0_A_SHIFT							24
+#define   NV20TCL_RC_COLOR0_A_MASK							0xff000000
+#define  NV20TCL_RC_COLOR1								0x00001e24
+#define   NV20TCL_RC_COLOR1_B_SHIFT							0
+#define   NV20TCL_RC_COLOR1_B_MASK							0x000000ff
+#define   NV20TCL_RC_COLOR1_G_SHIFT							8
+#define   NV20TCL_RC_COLOR1_G_MASK							0x0000ff00
+#define   NV20TCL_RC_COLOR1_R_SHIFT							16
+#define   NV20TCL_RC_COLOR1_R_MASK							0x00ff0000
+#define   NV20TCL_RC_COLOR1_A_SHIFT							24
+#define   NV20TCL_RC_COLOR1_A_MASK							0xff000000
+#define  NV20TCL_BACK_MATERIAL_SHININESS(x)						(0x00001e28+((x)*4))
+#define  NV20TCL_BACK_MATERIAL_SHININESS__SIZE						0x00000006
+#define  NV20TCL_RC_OUT_RGB(x)								(0x00001e40+((x)*4))
+#define  NV20TCL_RC_OUT_RGB__SIZE							0x00000008
+#define   NV20TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT						0
+#define   NV20TCL_RC_OUT_RGB_CD_OUTPUT_MASK						0x0000000f
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_FOG						0x00000003
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV20TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV20TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT						4
+#define   NV20TCL_RC_OUT_RGB_AB_OUTPUT_MASK						0x000000f0
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_FOG						0x00000030
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000000e0
+#define    NV20TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV20TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT						8
+#define   NV20TCL_RC_OUT_RGB_SUM_OUTPUT_MASK						0x00000f00
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO						0x00000000
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_FOG						0x00000300
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV20TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV20TCL_RC_OUT_RGB_CD_DOT_PRODUCT						(1 << 12)
+#define   NV20TCL_RC_OUT_RGB_AB_DOT_PRODUCT						(1 << 13)
+#define   NV20TCL_RC_OUT_RGB_MUX_SUM							(1 << 14)
+#define   NV20TCL_RC_OUT_RGB_BIAS							(1 << 15)
+#define    NV20TCL_RC_OUT_RGB_BIAS_NONE							0x00000000
+#define    NV20TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV				0x00008000
+#define   NV20TCL_RC_OUT_RGB_SCALE_SHIFT						17
+#define   NV20TCL_RC_OUT_RGB_SCALE_MASK							0x00000000
+#define    NV20TCL_RC_OUT_RGB_SCALE_NONE						0x00000000
+#define    NV20TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV20TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV20TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define  NV20TCL_RC_ENABLE								0x00001e60
+#define   NV20TCL_RC_ENABLE_NUM_COMBINERS_SHIFT						0
+#define   NV20TCL_RC_ENABLE_NUM_COMBINERS_MASK						0x0000000f
+#define  NV20TCL_TX_RCOMP								0x00001e6c
+#define   NV20TCL_TX_RCOMP_NEVER							0x00000000
+#define   NV20TCL_TX_RCOMP_GREATER							0x00000001
+#define   NV20TCL_TX_RCOMP_EQUAL							0x00000002
+#define   NV20TCL_TX_RCOMP_GEQUAL							0x00000003
+#define   NV20TCL_TX_RCOMP_LESS								0x00000004
+#define   NV20TCL_TX_RCOMP_NOTEQUAL							0x00000005
+#define   NV20TCL_TX_RCOMP_LEQUAL							0x00000006
+#define   NV20TCL_TX_RCOMP_ALWAYS							0x00000007
+#define  NV20TCL_TX_SHADER_OP								0x00001e70
+#define   NV20TCL_TX_SHADER_OP_TX0_SHIFT						0
+#define   NV20TCL_TX_SHADER_OP_TX0_MASK							0x0000001f
+#define    NV20TCL_TX_SHADER_OP_TX0_NONE						0x00000000
+#define    NV20TCL_TX_SHADER_OP_TX0_TEXTURE_2D						0x00000001
+#define    NV20TCL_TX_SHADER_OP_TX0_PASS_THROUGH					0x00000004
+#define    NV20TCL_TX_SHADER_OP_TX0_CULL_FRAGMENT					0x00000005
+#define    NV20TCL_TX_SHADER_OP_TX0_OFFSET_TEXTURE_2D					0x00000006
+#define    NV20TCL_TX_SHADER_OP_TX0_DOT_PRODUCT_TEXTURE_2D				0x00000009
+#define    NV20TCL_TX_SHADER_OP_TX0_DOT_PRODUCT_DEPTH_REPLACE				0x0000000a
+#define    NV20TCL_TX_SHADER_OP_TX0_DEPENDANT_AR_TEXTURE_2D				0x0000000f
+#define    NV20TCL_TX_SHADER_OP_TX0_DEPENDANT_GB_TEXTURE_2D				0x00000010
+#define    NV20TCL_TX_SHADER_OP_TX0_DOT_PRODUCT						0x00000011
+#define   NV20TCL_TX_SHADER_OP_TX1_SHIFT						5
+#define   NV20TCL_TX_SHADER_OP_TX1_MASK							0x000003e0
+#define    NV20TCL_TX_SHADER_OP_TX1_NONE						0x00000000
+#define    NV20TCL_TX_SHADER_OP_TX1_TEXTURE_2D						0x00000020
+#define    NV20TCL_TX_SHADER_OP_TX1_PASS_THROUGH					0x00000080
+#define    NV20TCL_TX_SHADER_OP_TX1_CULL_FRAGMENT					0x000000a0
+#define    NV20TCL_TX_SHADER_OP_TX1_OFFSET_TEXTURE_2D					0x000000c0
+#define    NV20TCL_TX_SHADER_OP_TX1_DOT_PRODUCT_TEXTURE_2D				0x00000120
+#define    NV20TCL_TX_SHADER_OP_TX1_DOT_PRODUCT_DEPTH_REPLACE				0x00000140
+#define    NV20TCL_TX_SHADER_OP_TX1_DEPENDANT_AR_TEXTURE_2D				0x000001e0
+#define    NV20TCL_TX_SHADER_OP_TX1_DEPENDANT_GB_TEXTURE_2D				0x00000200
+#define    NV20TCL_TX_SHADER_OP_TX1_DOT_PRODUCT						0x00000220
+#define   NV20TCL_TX_SHADER_OP_TX2_SHIFT						10
+#define   NV20TCL_TX_SHADER_OP_TX2_MASK							0x00007c00
+#define    NV20TCL_TX_SHADER_OP_TX2_NONE						0x00000000
+#define    NV20TCL_TX_SHADER_OP_TX2_TEXTURE_2D						0x00000400
+#define    NV20TCL_TX_SHADER_OP_TX2_PASS_THROUGH					0x00001000
+#define    NV20TCL_TX_SHADER_OP_TX2_CULL_FRAGMENT					0x00001400
+#define    NV20TCL_TX_SHADER_OP_TX2_OFFSET_TEXTURE_2D					0x00001800
+#define    NV20TCL_TX_SHADER_OP_TX2_DOT_PRODUCT_TEXTURE_2D				0x00002400
+#define    NV20TCL_TX_SHADER_OP_TX2_DOT_PRODUCT_DEPTH_REPLACE				0x00002800
+#define    NV20TCL_TX_SHADER_OP_TX2_DEPENDANT_AR_TEXTURE_2D				0x00003c00
+#define    NV20TCL_TX_SHADER_OP_TX2_DEPENDANT_GB_TEXTURE_2D				0x00004000
+#define    NV20TCL_TX_SHADER_OP_TX2_DOT_PRODUCT						0x00004400
+#define   NV20TCL_TX_SHADER_OP_TX3_SHIFT						15
+#define   NV20TCL_TX_SHADER_OP_TX3_MASK							0x000f8000
+#define    NV20TCL_TX_SHADER_OP_TX3_NONE						0x00000000
+#define    NV20TCL_TX_SHADER_OP_TX3_TEXTURE_2D						0x00008000
+#define    NV20TCL_TX_SHADER_OP_TX3_PASS_THROUGH					0x00020000
+#define    NV20TCL_TX_SHADER_OP_TX3_CULL_FRAGMENT					0x00028000
+#define    NV20TCL_TX_SHADER_OP_TX3_OFFSET_TEXTURE_2D					0x00030000
+#define    NV20TCL_TX_SHADER_OP_TX3_DOT_PRODUCT_TEXTURE_2D				0x00048000
+#define    NV20TCL_TX_SHADER_OP_TX3_DOT_PRODUCT_DEPTH_REPLACE				0x00050000
+#define    NV20TCL_TX_SHADER_OP_TX3_DEPENDANT_AR_TEXTURE_2D				0x00078000
+#define    NV20TCL_TX_SHADER_OP_TX3_DEPENDANT_GB_TEXTURE_2D				0x00080000
+#define    NV20TCL_TX_SHADER_OP_TX3_DOT_PRODUCT						0x00088000
+#define  NV20TCL_TX_SHADER_DOTMAPPING							0x00001e74
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX0_SHIFT					0
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX0_MASK						0x0000000f
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX1_SHIFT					4
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX1_MASK						0x000000f0
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX2_SHIFT					8
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX2_MASK						0x00000f00
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX3_SHIFT					12
+#define   NV20TCL_TX_SHADER_DOTMAPPING_TX3_MASK						0x0000f000
+#define  NV20TCL_TX_SHADER_PREVIOUS							0x00001e78
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX0_SHIFT						8
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX0_MASK						0x00000f00
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX1_SHIFT						12
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX1_MASK						0x0000f000
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX2_SHIFT						16
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX2_MASK						0x00030000
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX3_SHIFT						20
+#define   NV20TCL_TX_SHADER_PREVIOUS_TX3_MASK						0x00300000
+#define  NV20TCL_ENGINE									0x00001e94
+#define   NV20TCL_ENGINE_VP								(1 <<  1)
+#define   NV20TCL_ENGINE_FIXED								(1 <<  2)
+#define  NV20TCL_VP_UPLOAD_FROM_ID							0x00001e9c
+#define  NV20TCL_VP_START_FROM_ID							0x00001ea0
+#define  NV20TCL_VP_UPLOAD_CONST_ID							0x00001ea4
+#define  NV20TCL_VIEWPORT_TRANSLATE_X							0x00001f00
+#define  NV20TCL_VIEWPORT_TRANSLATE_Y							0x00001f04
+#define  NV20TCL_VIEWPORT_TRANSLATE_Z							0x00001f08
+#define  NV20TCL_VIEWPORT_TRANSLATE_W							0x00001f0c
+
+
+#define NV17TCL										0x00000099
+
+#define  NV17TCL_DMA_IN_MEMORY4								0x000001ac
+#define  NV17TCL_DMA_IN_MEMORY5								0x000001b0
+#define  NV17TCL_COLOR_MASK_ENABLE							0x000002bc
+#define  NV17TCL_LMA_DEPTH_BUFFER_PITCH							0x00000d5c
+#define  NV17TCL_LMA_DEPTH_BUFFER_OFFSET						0x00000d60
+#define  NV17TCL_LMA_DEPTH_FILL_VALUE							0x00000d68
+#define  NV17TCL_LMA_DEPTH_BUFFER_CLEAR							0x00000d6c
+#define  NV17TCL_LMA_DEPTH_ENABLE							0x00001658
+
+
+#define NV20_SWIZZLED_SURFACE								0x0000009e
+
+
+
+#define NV12_IMAGE_BLIT									0x0000009f
+
+
+
+#define NV30_CONTEXT_SURFACES_2D							0x00000362
+
+
+
+#define NV30_STRETCHED_IMAGE_FROM_CPU							0x00000366
+
+
+
+#define NV30_TEXTURE_FROM_CPU								0x0000037b
+
+
+
+#define NV30_SCALED_IMAGE_FROM_MEMORY							0x00000389
+
+
+
+#define NV30_IMAGE_FROM_CPU								0x0000038a
+
+
+
+#define NV30TCL										0x00000397
+
+
+
+#define NV30_SWIZZLED_SURFACE								0x0000039e
+
+
+
+#define NV35TCL										0x00000497
+
+
+
+#define NV25TCL										0x00000597
+
+#define  NV25TCL_DMA_IN_MEMORY4								0x0000019c
+#define  NV25TCL_DMA_IN_MEMORY5								0x000001a0
+#define  NV25TCL_DMA_IN_MEMORY8								0x000001ac
+#define  NV25TCL_DMA_IN_MEMORY9								0x000001b0
+
+
+#define NV34TCL										0x00000697
+
+#define  NV34TCL_NOP									0x00000100
+#define  NV34TCL_NOTIFY									0x00000104
+#define  NV34TCL_DMA_NOTIFY								0x00000180
+#define  NV34TCL_DMA_TEXTURE0								0x00000184
+#define  NV34TCL_DMA_TEXTURE1								0x00000188
+#define  NV34TCL_DMA_COLOR1								0x0000018c
+#define  NV34TCL_DMA_COLOR0								0x00000194
+#define  NV34TCL_DMA_ZETA								0x00000198
+#define  NV34TCL_DMA_VTXBUF0								0x0000019c
+#define  NV34TCL_DMA_VTXBUF1								0x000001a0
+#define  NV34TCL_DMA_FENCE								0x000001a4
+#define  NV34TCL_DMA_QUERY								0x000001a8
+#define  NV34TCL_DMA_IN_MEMORY7								0x000001ac
+#define  NV34TCL_DMA_IN_MEMORY8								0x000001b0
+#define  NV34TCL_RT_HORIZ								0x00000200
+#define   NV34TCL_RT_HORIZ_X_SHIFT							0
+#define   NV34TCL_RT_HORIZ_X_MASK							0x0000ffff
+#define   NV34TCL_RT_HORIZ_W_SHIFT							16
+#define   NV34TCL_RT_HORIZ_W_MASK							0xffff0000
+#define  NV34TCL_RT_VERT								0x00000204
+#define   NV34TCL_RT_VERT_Y_SHIFT							0
+#define   NV34TCL_RT_VERT_Y_MASK							0x0000ffff
+#define   NV34TCL_RT_VERT_H_SHIFT							16
+#define   NV34TCL_RT_VERT_H_MASK							0xffff0000
+#define  NV34TCL_RT_FORMAT								0x00000208
+#define   NV34TCL_RT_FORMAT_TYPE_SHIFT							8
+#define   NV34TCL_RT_FORMAT_TYPE_MASK							0x00000f00
+#define    NV34TCL_RT_FORMAT_TYPE_LINEAR						0x00000100
+#define    NV34TCL_RT_FORMAT_TYPE_SWIZZLED						0x00000200
+#define   NV34TCL_RT_FORMAT_ZETA_SHIFT							5
+#define   NV34TCL_RT_FORMAT_ZETA_MASK							0x000000e0
+#define    NV34TCL_RT_FORMAT_ZETA_Z16							0x00000020
+#define    NV34TCL_RT_FORMAT_ZETA_Z24S8							0x00000040
+#define   NV34TCL_RT_FORMAT_COLOR_SHIFT							0
+#define   NV34TCL_RT_FORMAT_COLOR_MASK							0x0000001f
+#define    NV34TCL_RT_FORMAT_COLOR_R5G6B5						0x00000003
+#define    NV34TCL_RT_FORMAT_COLOR_X8R8G8B8						0x00000005
+#define    NV34TCL_RT_FORMAT_COLOR_A8R8G8B8						0x00000008
+#define    NV34TCL_RT_FORMAT_COLOR_B8							0x00000009
+#define    NV34TCL_RT_FORMAT_COLOR_UNKNOWN						0x0000000d
+#define    NV34TCL_RT_FORMAT_COLOR_X8B8G8R8						0x0000000f
+#define    NV34TCL_RT_FORMAT_COLOR_A8B8G8R8						0x00000010
+#define  NV34TCL_COLOR0_PITCH								0x0000020c
+#define   NV34TCL_COLOR0_PITCH_COLOR0_SHIFT						0
+#define   NV34TCL_COLOR0_PITCH_COLOR0_MASK						0x0000ffff
+#define   NV34TCL_COLOR0_PITCH_ZETA_SHIFT						16
+#define   NV34TCL_COLOR0_PITCH_ZETA_MASK						0xffff0000
+#define  NV34TCL_COLOR0_OFFSET								0x00000210
+#define  NV34TCL_ZETA_OFFSET								0x00000214
+#define  NV34TCL_COLOR1_OFFSET								0x00000218
+#define  NV34TCL_COLOR1_PITCH								0x0000021c
+#define  NV34TCL_RT_ENABLE								0x00000220
+#define   NV34TCL_RT_ENABLE_MRT								(1 <<  4)
+#define   NV34TCL_RT_ENABLE_COLOR1							(1 <<  1)
+#define   NV34TCL_RT_ENABLE_COLOR0							(1 <<  0)
+#define  NV34TCL_LMA_DEPTH_PITCH							0x0000022c
+#define  NV34TCL_LMA_DEPTH_OFFSET							0x00000230
+#define  NV34TCL_TX_UNITS_ENABLE							0x0000023c
+#define   NV34TCL_TX_UNITS_ENABLE_TX0							(1 <<  0)
+#define   NV34TCL_TX_UNITS_ENABLE_TX1							(1 <<  1)
+#define   NV34TCL_TX_UNITS_ENABLE_TX2							(1 <<  2)
+#define   NV34TCL_TX_UNITS_ENABLE_TX3							(1 <<  3)
+#define   NV34TCL_TX_UNITS_ENABLE_TX4							(1 <<  4)
+#define   NV34TCL_TX_UNITS_ENABLE_TX5							(1 <<  5)
+#define   NV34TCL_TX_UNITS_ENABLE_TX6							(1 <<  6)
+#define   NV34TCL_TX_UNITS_ENABLE_TX7							(1 <<  7)
+#define  NV34TCL_TX_MATRIX_ENABLE(x)							(0x00000240+((x)*4))
+#define  NV34TCL_TX_MATRIX_ENABLE__SIZE							0x00000008
+#define  NV34TCL_VIEWPORT_TX_ORIGIN							0x000002b8
+#define   NV34TCL_VIEWPORT_TX_ORIGIN_X_SHIFT						0
+#define   NV34TCL_VIEWPORT_TX_ORIGIN_X_MASK						0x0000ffff
+#define   NV34TCL_VIEWPORT_TX_ORIGIN_Y_SHIFT						16
+#define   NV34TCL_VIEWPORT_TX_ORIGIN_Y_MASK						0xffff0000
+#define  NV34TCL_VIEWPORT_CLIP_MODE							0x000002bc
+#define  NV34TCL_VIEWPORT_CLIP_HORIZ(x)							(0x000002c0+((x)*8))
+#define  NV34TCL_VIEWPORT_CLIP_HORIZ__SIZE						0x00000008
+#define   NV34TCL_VIEWPORT_CLIP_HORIZ_L_SHIFT						0
+#define   NV34TCL_VIEWPORT_CLIP_HORIZ_L_MASK						0x0000ffff
+#define   NV34TCL_VIEWPORT_CLIP_HORIZ_R_SHIFT						16
+#define   NV34TCL_VIEWPORT_CLIP_HORIZ_R_MASK						0xffff0000
+#define  NV34TCL_VIEWPORT_CLIP_VERT(x)							(0x000002c4+((x)*8))
+#define  NV34TCL_VIEWPORT_CLIP_VERT__SIZE						0x00000008
+#define   NV34TCL_VIEWPORT_CLIP_VERT_T_SHIFT						0
+#define   NV34TCL_VIEWPORT_CLIP_VERT_T_MASK						0x0000ffff
+#define   NV34TCL_VIEWPORT_CLIP_VERT_D_SHIFT						16
+#define   NV34TCL_VIEWPORT_CLIP_VERT_D_MASK						0xffff0000
+#define  NV34TCL_DITHER_ENABLE								0x00000300
+#define  NV34TCL_ALPHA_FUNC_ENABLE							0x00000304
+#define  NV34TCL_ALPHA_FUNC_FUNC							0x00000308
+#define   NV34TCL_ALPHA_FUNC_FUNC_NEVER							0x00000200
+#define   NV34TCL_ALPHA_FUNC_FUNC_LESS							0x00000201
+#define   NV34TCL_ALPHA_FUNC_FUNC_EQUAL							0x00000202
+#define   NV34TCL_ALPHA_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV34TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV34TCL_ALPHA_FUNC_FUNC_GREATER						0x00000204
+#define   NV34TCL_ALPHA_FUNC_FUNC_NOTEQUAL						0x00000205
+#define   NV34TCL_ALPHA_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV34TCL_ALPHA_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV34TCL_ALPHA_FUNC_REF								0x0000030c
+#define  NV34TCL_BLEND_FUNC_ENABLE							0x00000310
+#define  NV34TCL_BLEND_FUNC_SRC								0x00000314
+#define   NV34TCL_BLEND_FUNC_SRC_RGB_SHIFT						0
+#define   NV34TCL_BLEND_FUNC_SRC_RGB_MASK						0x0000ffff
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ZERO						0x00000000
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE						0x00000001
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR						0x00000300
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA						0x00000302
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA						0x00000304
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_DST_COLOR						0x00000306
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE				0x00000308
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR					0x00008001
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA					0x00008003
+#define    NV34TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define   NV34TCL_BLEND_FUNC_SRC_ALPHA_SHIFT						16
+#define   NV34TCL_BLEND_FUNC_SRC_ALPHA_MASK						0xffff0000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ZERO						0x00000000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE						0x00010000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR					0x03000000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR				0x03010000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA					0x03020000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA				0x03030000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA					0x03040000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA				0x03050000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR					0x03060000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR				0x03070000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE				0x03080000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR					0x80010000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR			0x80020000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA					0x80030000
+#define    NV34TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA			0x80040000
+#define  NV34TCL_BLEND_FUNC_DST								0x00000318
+#define   NV34TCL_BLEND_FUNC_DST_RGB_SHIFT						0
+#define   NV34TCL_BLEND_FUNC_DST_RGB_MASK						0x0000ffff
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ZERO						0x00000000
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE						0x00000001
+#define    NV34TCL_BLEND_FUNC_DST_RGB_SRC_COLOR						0x00000300
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define    NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA						0x00000302
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define    NV34TCL_BLEND_FUNC_DST_RGB_DST_ALPHA						0x00000304
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define    NV34TCL_BLEND_FUNC_DST_RGB_DST_COLOR						0x00000306
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define    NV34TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE				0x00000308
+#define    NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR					0x00008001
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define    NV34TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA					0x00008003
+#define    NV34TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define   NV34TCL_BLEND_FUNC_DST_ALPHA_SHIFT						16
+#define   NV34TCL_BLEND_FUNC_DST_ALPHA_MASK						0xffff0000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ZERO						0x00000000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE						0x00010000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR					0x03000000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR				0x03010000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA					0x03020000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA				0x03030000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA					0x03040000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA				0x03050000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR					0x03060000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR				0x03070000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE				0x03080000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR					0x80010000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR			0x80020000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA					0x80030000
+#define    NV34TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA			0x80040000
+#define  NV34TCL_BLEND_COLOR								0x0000031c
+#define   NV34TCL_BLEND_COLOR_B_SHIFT							0
+#define   NV34TCL_BLEND_COLOR_B_MASK							0x000000ff
+#define   NV34TCL_BLEND_COLOR_G_SHIFT							8
+#define   NV34TCL_BLEND_COLOR_G_MASK							0x0000ff00
+#define   NV34TCL_BLEND_COLOR_R_SHIFT							16
+#define   NV34TCL_BLEND_COLOR_R_MASK							0x00ff0000
+#define   NV34TCL_BLEND_COLOR_A_SHIFT							24
+#define   NV34TCL_BLEND_COLOR_A_MASK							0xff000000
+#define  NV34TCL_BLEND_EQUATION								0x00000320
+#define   NV34TCL_BLEND_EQUATION_FUNC_ADD						0x00008006
+#define   NV34TCL_BLEND_EQUATION_MIN							0x00008007
+#define   NV34TCL_BLEND_EQUATION_MAX							0x00008008
+#define   NV34TCL_BLEND_EQUATION_FUNC_SUBTRACT						0x0000800a
+#define   NV34TCL_BLEND_EQUATION_FUNC_REVERSE_SUBTRACT					0x0000800b
+#define  NV34TCL_COLOR_MASK								0x00000324
+#define   NV34TCL_COLOR_MASK_B_SHIFT							0
+#define   NV34TCL_COLOR_MASK_B_MASK							0x000000ff
+#define   NV34TCL_COLOR_MASK_G_SHIFT							8
+#define   NV34TCL_COLOR_MASK_G_MASK							0x0000ff00
+#define   NV34TCL_COLOR_MASK_R_SHIFT							16
+#define   NV34TCL_COLOR_MASK_R_MASK							0x00ff0000
+#define   NV34TCL_COLOR_MASK_A_SHIFT							24
+#define   NV34TCL_COLOR_MASK_A_MASK							0xff000000
+#define  NV34TCL_STENCIL_BACK_ENABLE							0x00000328
+#define  NV34TCL_STENCIL_BACK_MASK							0x0000032c
+#define  NV34TCL_STENCIL_BACK_FUNC_FUNC							0x00000330
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_NEVER						0x00000200
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_LESS						0x00000201
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_EQUAL						0x00000202
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV34TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV34TCL_STENCIL_BACK_FUNC_REF							0x00000334
+#define  NV34TCL_STENCIL_BACK_FUNC_MASK							0x00000338
+#define  NV34TCL_STENCIL_BACK_OP_FAIL							0x0000033c
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_ZERO						0x00000000
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_REPLACE						0x00001e01
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_INCR						0x00001e02
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_DECR						0x00001e03
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV34TCL_STENCIL_BACK_OP_ZFAIL							0x00000340
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_ZERO						0x00000000
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_REPLACE						0x00001e01
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR						0x00001e02
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR						0x00001e03
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV34TCL_STENCIL_BACK_OP_ZPASS							0x00000344
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_ZERO						0x00000000
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_REPLACE						0x00001e01
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_INCR						0x00001e02
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_DECR						0x00001e03
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV34TCL_STENCIL_FRONT_ENABLE							0x00000348
+#define  NV34TCL_STENCIL_FRONT_MASK							0x0000034c
+#define  NV34TCL_STENCIL_FRONT_FUNC_FUNC						0x00000350
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_NEVER						0x00000200
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_LESS						0x00000201
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL						0x00000202
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL					0x00000203
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL					0x00000206
+#define   NV34TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS					0x00000207
+#define  NV34TCL_STENCIL_FRONT_FUNC_REF							0x00000354
+#define  NV34TCL_STENCIL_FRONT_FUNC_MASK						0x00000358
+#define  NV34TCL_STENCIL_FRONT_OP_FAIL							0x0000035c
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_ZERO						0x00000000
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_REPLACE						0x00001e01
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_INCR						0x00001e02
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_DECR						0x00001e03
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV34TCL_STENCIL_FRONT_OP_ZFAIL							0x00000360
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_ZERO						0x00000000
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE					0x00001e01
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR						0x00001e02
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR						0x00001e03
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV34TCL_STENCIL_FRONT_OP_ZPASS							0x00000364
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_ZERO						0x00000000
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_INVERT						0x0000150a
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_KEEP						0x00001e00
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_REPLACE					0x00001e01
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR						0x00001e02
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR						0x00001e03
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV34TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV34TCL_SHADE_MODEL								0x00000368
+#define   NV34TCL_SHADE_MODEL_FLAT							0x00001d00
+#define   NV34TCL_SHADE_MODEL_SMOOTH							0x00001d01
+#define  NV34TCL_FOG_ENABLE								0x0000036c
+#define  NV34TCL_FOG_COLOR								0x00000370
+#define   NV34TCL_FOG_COLOR_R_SHIFT							0
+#define   NV34TCL_FOG_COLOR_R_MASK							0x000000ff
+#define   NV34TCL_FOG_COLOR_G_SHIFT							8
+#define   NV34TCL_FOG_COLOR_G_MASK							0x0000ff00
+#define   NV34TCL_FOG_COLOR_B_SHIFT							16
+#define   NV34TCL_FOG_COLOR_B_MASK							0x00ff0000
+#define   NV34TCL_FOG_COLOR_A_SHIFT							24
+#define   NV34TCL_FOG_COLOR_A_MASK							0xff000000
+#define  NV34TCL_COLOR_LOGIC_OP_ENABLE							0x00000374
+#define  NV34TCL_COLOR_LOGIC_OP_OP							0x00000378
+#define   NV34TCL_COLOR_LOGIC_OP_OP_CLEAR						0x00001500
+#define   NV34TCL_COLOR_LOGIC_OP_OP_AND							0x00001501
+#define   NV34TCL_COLOR_LOGIC_OP_OP_AND_REVERSE						0x00001502
+#define   NV34TCL_COLOR_LOGIC_OP_OP_COPY						0x00001503
+#define   NV34TCL_COLOR_LOGIC_OP_OP_AND_INVERTED					0x00001504
+#define   NV34TCL_COLOR_LOGIC_OP_OP_NOOP						0x00001505
+#define   NV34TCL_COLOR_LOGIC_OP_OP_XOR							0x00001506
+#define   NV34TCL_COLOR_LOGIC_OP_OP_OR							0x00001507
+#define   NV34TCL_COLOR_LOGIC_OP_OP_NOR							0x00001508
+#define   NV34TCL_COLOR_LOGIC_OP_OP_EQUIV						0x00001509
+#define   NV34TCL_COLOR_LOGIC_OP_OP_INVERT						0x0000150a
+#define   NV34TCL_COLOR_LOGIC_OP_OP_OR_REVERSE						0x0000150b
+#define   NV34TCL_COLOR_LOGIC_OP_OP_COPY_INVERTED					0x0000150c
+#define   NV34TCL_COLOR_LOGIC_OP_OP_OR_INVERTED						0x0000150d
+#define   NV34TCL_COLOR_LOGIC_OP_OP_NAND						0x0000150e
+#define   NV34TCL_COLOR_LOGIC_OP_OP_SET							0x0000150f
+#define  NV34TCL_NORMALIZE_ENABLE							0x0000037c
+#define  NV34TCL_COLOR_MATERIAL								0x00000390
+#define   NV34TCL_COLOR_MATERIAL_FRONT_EMISSION_ENABLE					(1 <<  0)
+#define   NV34TCL_COLOR_MATERIAL_FRONT_AMBIENT_ENABLE					(1 <<  2)
+#define   NV34TCL_COLOR_MATERIAL_FRONT_DIFFUSE_ENABLE					(1 <<  4)
+#define   NV34TCL_COLOR_MATERIAL_FRONT_SPECULAR_ENABLE					(1 <<  6)
+#define   NV34TCL_COLOR_MATERIAL_BACK_EMISSION_ENABLE					(1 <<  8)
+#define   NV34TCL_COLOR_MATERIAL_BACK_AMBIENT_ENABLE					(1 << 10)
+#define   NV34TCL_COLOR_MATERIAL_BACK_DIFFUSE_ENABLE					(1 << 12)
+#define   NV34TCL_COLOR_MATERIAL_BACK_SPECULAR_ENABLE					(1 << 14)
+#define  NV34TCL_DEPTH_RANGE_NEAR							0x00000394
+#define  NV34TCL_DEPTH_RANGE_FAR							0x00000398
+#define  NV34TCL_COLOR_MATERIAL_FRONT_R							0x000003a0
+#define  NV34TCL_COLOR_MATERIAL_FRONT_G							0x000003a4
+#define  NV34TCL_COLOR_MATERIAL_FRONT_B							0x000003a8
+#define  NV34TCL_COLOR_MATERIAL_FRONT_A							0x000003b4
+#define  NV34TCL_LINE_WIDTH								0x000003b8
+#define  NV34TCL_LINE_SMOOTH_ENABLE							0x000003bc
+#define  NV34TCL_TX_GEN_S(x)								(0x00000400+((x)*16))
+#define  NV34TCL_TX_GEN_S__SIZE								0x00000008
+#define   NV34TCL_TX_GEN_S_FALSE							0x00000000
+#define   NV34TCL_TX_GEN_S_EYE_LINEAR							0x00002400
+#define   NV34TCL_TX_GEN_S_OBJECT_LINEAR						0x00002401
+#define   NV34TCL_TX_GEN_S_SPHERE_MAP							0x00002402
+#define   NV34TCL_TX_GEN_S_NORMAL_MAP							0x00008511
+#define   NV34TCL_TX_GEN_S_REFLECTION_MAP						0x00008512
+#define  NV34TCL_TX_GEN_T(x)								(0x00000404+((x)*16))
+#define  NV34TCL_TX_GEN_T__SIZE								0x00000008
+#define   NV34TCL_TX_GEN_T_FALSE							0x00000000
+#define   NV34TCL_TX_GEN_T_EYE_LINEAR							0x00002400
+#define   NV34TCL_TX_GEN_T_OBJECT_LINEAR						0x00002401
+#define   NV34TCL_TX_GEN_T_SPHERE_MAP							0x00002402
+#define   NV34TCL_TX_GEN_T_NORMAL_MAP							0x00008511
+#define   NV34TCL_TX_GEN_T_REFLECTION_MAP						0x00008512
+#define  NV34TCL_TX_GEN_R(x)								(0x00000408+((x)*16))
+#define  NV34TCL_TX_GEN_R__SIZE								0x00000008
+#define   NV34TCL_TX_GEN_R_FALSE							0x00000000
+#define   NV34TCL_TX_GEN_R_EYE_LINEAR							0x00002400
+#define   NV34TCL_TX_GEN_R_OBJECT_LINEAR						0x00002401
+#define   NV34TCL_TX_GEN_R_SPHERE_MAP							0x00002402
+#define   NV34TCL_TX_GEN_R_NORMAL_MAP							0x00008511
+#define   NV34TCL_TX_GEN_R_REFLECTION_MAP						0x00008512
+#define  NV34TCL_TX_GEN_Q(x)								(0x0000040c+((x)*16))
+#define  NV34TCL_TX_GEN_Q__SIZE								0x00000008
+#define   NV34TCL_TX_GEN_Q_FALSE							0x00000000
+#define   NV34TCL_TX_GEN_Q_EYE_LINEAR							0x00002400
+#define   NV34TCL_TX_GEN_Q_OBJECT_LINEAR						0x00002401
+#define   NV34TCL_TX_GEN_Q_SPHERE_MAP							0x00002402
+#define   NV34TCL_TX_GEN_Q_NORMAL_MAP							0x00008511
+#define   NV34TCL_TX_GEN_Q_REFLECTION_MAP						0x00008512
+#define  NV34TCL_MODELVIEW_MATRIX(x)							(0x00000480+((x)*4))
+#define  NV34TCL_MODELVIEW_MATRIX__SIZE							0x00000010
+#define  NV34TCL_INVERSE_MODELVIEW_MATRIX(x)						(0x00000580+((x)*4))
+#define  NV34TCL_INVERSE_MODELVIEW_MATRIX__SIZE						0x0000000c
+#define  NV34TCL_PROJECTION_MATRIX(x)							(0x00000680+((x)*4))
+#define  NV34TCL_PROJECTION_MATRIX__SIZE						0x00000010
+#define  NV34TCL_TX0_MATRIX(x)								(0x000006c0+((x)*4))
+#define  NV34TCL_TX0_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX1_MATRIX(x)								(0x00000700+((x)*4))
+#define  NV34TCL_TX1_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX2_MATRIX(x)								(0x00000740+((x)*4))
+#define  NV34TCL_TX2_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX3_MATRIX(x)								(0x00000780+((x)*4))
+#define  NV34TCL_TX3_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX4_MATRIX(x)								(0x000007c0+((x)*4))
+#define  NV34TCL_TX4_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX5_MATRIX(x)								(0x00000800+((x)*4))
+#define  NV34TCL_TX5_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX6_MATRIX(x)								(0x00000840+((x)*4))
+#define  NV34TCL_TX6_MATRIX__SIZE							0x00000010
+#define  NV34TCL_TX7_MATRIX(x)								(0x00000880+((x)*4))
+#define  NV34TCL_TX7_MATRIX__SIZE							0x00000010
+#define  NV34TCL_SCISSOR_HORIZ								0x000008c0
+#define   NV34TCL_SCISSOR_HORIZ_X_SHIFT							0
+#define   NV34TCL_SCISSOR_HORIZ_X_MASK							0x0000ffff
+#define   NV34TCL_SCISSOR_HORIZ_W_SHIFT							16
+#define   NV34TCL_SCISSOR_HORIZ_W_MASK							0xffff0000
+#define  NV34TCL_SCISSOR_VERT								0x000008c4
+#define   NV34TCL_SCISSOR_VERT_Y_SHIFT							0
+#define   NV34TCL_SCISSOR_VERT_Y_MASK							0x0000ffff
+#define   NV34TCL_SCISSOR_VERT_H_SHIFT							16
+#define   NV34TCL_SCISSOR_VERT_H_MASK							0xffff0000
+#define  NV34TCL_FOG_COORD_DIST								0x000008c8
+#define   NV34TCL_FOG_COORD_DIST_COORD_FALSE						0x00000000
+#define   NV34TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_RADIAL_NV		0x00000001
+#define   NV34TCL_FOG_COORD_DIST_COORD_FRAGMENT_DEPTH_DISTANCE_EYE_PLANE_ABSOLUTE_NV	0x00000002
+#define   NV34TCL_FOG_COORD_DIST_COORD_FOG						0x00000003
+#define  NV34TCL_FOG_MODE								0x000008cc
+#define   NV34TCL_FOG_MODE_EXP								0x00000800
+#define   NV34TCL_FOG_MODE_EXP_2							0x00000802
+#define   NV34TCL_FOG_MODE_EXP2								0x00000803
+#define   NV34TCL_FOG_MODE_LINEAR							0x00000804
+#define   NV34TCL_FOG_MODE_LINEAR_2							0x00002601
+#define  NV34TCL_FOG_EQUATION_CONSTANT							0x000008d0
+#define  NV34TCL_FOG_EQUATION_LINEAR							0x000008d4
+#define  NV34TCL_FOG_EQUATION_QUADRATIC							0x000008d8
+#define  NV34TCL_FP_ACTIVE_PROGRAM							0x000008e4
+#define   NV34TCL_FP_ACTIVE_PROGRAM_DMA0						(1 <<  0)
+#define   NV34TCL_FP_ACTIVE_PROGRAM_DMA1						(1 <<  1)
+#define   NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_SHIFT					2
+#define   NV34TCL_FP_ACTIVE_PROGRAM_OFFSET_MASK						0xfffffffc
+#define  NV34TCL_RC_COLOR0								0x000008ec
+#define   NV34TCL_RC_COLOR0_B_SHIFT							0
+#define   NV34TCL_RC_COLOR0_B_MASK							0x000000ff
+#define   NV34TCL_RC_COLOR0_G_SHIFT							8
+#define   NV34TCL_RC_COLOR0_G_MASK							0x0000ff00
+#define   NV34TCL_RC_COLOR0_R_SHIFT							16
+#define   NV34TCL_RC_COLOR0_R_MASK							0x00ff0000
+#define   NV34TCL_RC_COLOR0_A_SHIFT							24
+#define   NV34TCL_RC_COLOR0_A_MASK							0xff000000
+#define  NV34TCL_RC_COLOR1								0x000008f0
+#define   NV34TCL_RC_COLOR1_B_SHIFT							0
+#define   NV34TCL_RC_COLOR1_B_MASK							0x000000ff
+#define   NV34TCL_RC_COLOR1_G_SHIFT							8
+#define   NV34TCL_RC_COLOR1_G_MASK							0x0000ff00
+#define   NV34TCL_RC_COLOR1_R_SHIFT							16
+#define   NV34TCL_RC_COLOR1_R_MASK							0x00ff0000
+#define   NV34TCL_RC_COLOR1_A_SHIFT							24
+#define   NV34TCL_RC_COLOR1_A_MASK							0xff000000
+#define  NV34TCL_RC_FINAL0								0x000008f4
+#define   NV34TCL_RC_FINAL0_D_INPUT_SHIFT						0
+#define   NV34TCL_RC_FINAL0_D_INPUT_MASK						0x0000000f
+#define    NV34TCL_RC_FINAL0_D_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV34TCL_RC_FINAL0_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV34TCL_RC_FINAL0_D_INPUT_FOG						0x00000003
+#define    NV34TCL_RC_FINAL0_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV34TCL_RC_FINAL0_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV34TCL_RC_FINAL0_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV34TCL_RC_FINAL0_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV34TCL_RC_FINAL0_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV34TCL_RC_FINAL0_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV34TCL_RC_FINAL0_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV34TCL_RC_FINAL0_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV34TCL_RC_FINAL0_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV34TCL_RC_FINAL0_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL0_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV34TCL_RC_FINAL0_D_MAPPING_SHIFT						5
+#define   NV34TCL_RC_FINAL0_D_MAPPING_MASK						0x000000e0
+#define    NV34TCL_RC_FINAL0_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL0_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV34TCL_RC_FINAL0_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV34TCL_RC_FINAL0_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV34TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV34TCL_RC_FINAL0_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV34TCL_RC_FINAL0_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV34TCL_RC_FINAL0_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV34TCL_RC_FINAL0_C_INPUT_SHIFT						8
+#define   NV34TCL_RC_FINAL0_C_INPUT_MASK						0x00000f00
+#define    NV34TCL_RC_FINAL0_C_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV34TCL_RC_FINAL0_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV34TCL_RC_FINAL0_C_INPUT_FOG						0x00000300
+#define    NV34TCL_RC_FINAL0_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV34TCL_RC_FINAL0_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV34TCL_RC_FINAL0_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_FINAL0_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_FINAL0_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV34TCL_RC_FINAL0_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV34TCL_RC_FINAL0_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV34TCL_RC_FINAL0_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_FINAL0_C_COMPONENT_USAGE						(1 << 12)
+#define    NV34TCL_RC_FINAL0_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL0_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV34TCL_RC_FINAL0_C_MAPPING_SHIFT						13
+#define   NV34TCL_RC_FINAL0_C_MAPPING_MASK						0x0000e000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV34TCL_RC_FINAL0_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV34TCL_RC_FINAL0_B_INPUT_SHIFT						16
+#define   NV34TCL_RC_FINAL0_B_INPUT_MASK						0x000f0000
+#define    NV34TCL_RC_FINAL0_B_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV34TCL_RC_FINAL0_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV34TCL_RC_FINAL0_B_INPUT_FOG						0x00030000
+#define    NV34TCL_RC_FINAL0_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV34TCL_RC_FINAL0_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV34TCL_RC_FINAL0_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV34TCL_RC_FINAL0_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV34TCL_RC_FINAL0_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV34TCL_RC_FINAL0_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV34TCL_RC_FINAL0_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV34TCL_RC_FINAL0_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV34TCL_RC_FINAL0_B_COMPONENT_USAGE						(1 << 20)
+#define    NV34TCL_RC_FINAL0_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL0_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV34TCL_RC_FINAL0_B_MAPPING_SHIFT						21
+#define   NV34TCL_RC_FINAL0_B_MAPPING_MASK						0x00e00000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV34TCL_RC_FINAL0_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV34TCL_RC_FINAL0_A_INPUT_SHIFT						24
+#define   NV34TCL_RC_FINAL0_A_INPUT_MASK						0x0f000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_FOG						0x03000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV34TCL_RC_FINAL0_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV34TCL_RC_FINAL0_A_COMPONENT_USAGE						(1 << 28)
+#define    NV34TCL_RC_FINAL0_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL0_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV34TCL_RC_FINAL0_A_MAPPING_SHIFT						29
+#define   NV34TCL_RC_FINAL0_A_MAPPING_MASK						0xe0000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV34TCL_RC_FINAL0_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV34TCL_RC_FINAL1								0x000008f8
+#define   NV34TCL_RC_FINAL1_COLOR_SUM_CLAMP						(1 <<  7)
+#define   NV34TCL_RC_FINAL1_G_INPUT_SHIFT						8
+#define   NV34TCL_RC_FINAL1_G_INPUT_MASK						0x00000f00
+#define    NV34TCL_RC_FINAL1_G_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV34TCL_RC_FINAL1_G_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV34TCL_RC_FINAL1_G_INPUT_FOG						0x00000300
+#define    NV34TCL_RC_FINAL1_G_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV34TCL_RC_FINAL1_G_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV34TCL_RC_FINAL1_G_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_FINAL1_G_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_FINAL1_G_INPUT_SPARE0_NV						0x00000c00
+#define    NV34TCL_RC_FINAL1_G_INPUT_SPARE1_NV						0x00000d00
+#define    NV34TCL_RC_FINAL1_G_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV34TCL_RC_FINAL1_G_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_FINAL1_G_COMPONENT_USAGE						(1 << 12)
+#define    NV34TCL_RC_FINAL1_G_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL1_G_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV34TCL_RC_FINAL1_G_MAPPING_SHIFT						13
+#define   NV34TCL_RC_FINAL1_G_MAPPING_MASK						0x0000e000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV34TCL_RC_FINAL1_G_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV34TCL_RC_FINAL1_F_INPUT_SHIFT						16
+#define   NV34TCL_RC_FINAL1_F_INPUT_MASK						0x000f0000
+#define    NV34TCL_RC_FINAL1_F_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV34TCL_RC_FINAL1_F_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV34TCL_RC_FINAL1_F_INPUT_FOG						0x00030000
+#define    NV34TCL_RC_FINAL1_F_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV34TCL_RC_FINAL1_F_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV34TCL_RC_FINAL1_F_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV34TCL_RC_FINAL1_F_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV34TCL_RC_FINAL1_F_INPUT_SPARE0_NV						0x000c0000
+#define    NV34TCL_RC_FINAL1_F_INPUT_SPARE1_NV						0x000d0000
+#define    NV34TCL_RC_FINAL1_F_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV34TCL_RC_FINAL1_F_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV34TCL_RC_FINAL1_F_COMPONENT_USAGE						(1 << 20)
+#define    NV34TCL_RC_FINAL1_F_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL1_F_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV34TCL_RC_FINAL1_F_MAPPING_SHIFT						21
+#define   NV34TCL_RC_FINAL1_F_MAPPING_MASK						0x00e00000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV34TCL_RC_FINAL1_F_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV34TCL_RC_FINAL1_E_INPUT_SHIFT						24
+#define   NV34TCL_RC_FINAL1_E_INPUT_MASK						0x0f000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_FOG						0x03000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_SPARE0_NV						0x0c000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_SPARE1_NV						0x0d000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV34TCL_RC_FINAL1_E_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV34TCL_RC_FINAL1_E_COMPONENT_USAGE						(1 << 28)
+#define    NV34TCL_RC_FINAL1_E_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_FINAL1_E_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV34TCL_RC_FINAL1_E_MAPPING_SHIFT						29
+#define   NV34TCL_RC_FINAL1_E_MAPPING_MASK						0xe0000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV34TCL_RC_FINAL1_E_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV34TCL_RC_ENABLE								0x000008fc
+#define   NV34TCL_RC_ENABLE_NUM_COMBINERS_SHIFT						0
+#define   NV34TCL_RC_ENABLE_NUM_COMBINERS_MASK						0x0000000f
+#define   NV34TCL_RC_ENABLE_STAGE_CONSTANT_COLOR0_SHIFT					12
+#define   NV34TCL_RC_ENABLE_STAGE_CONSTANT_COLOR0_MASK					0x0000f000
+#define   NV34TCL_RC_ENABLE_STAGE_CONSTANT_COLOR1_SHIFT					16
+#define   NV34TCL_RC_ENABLE_STAGE_CONSTANT_COLOR1_MASK					0x000f0000
+#define  NV34TCL_RC_IN_ALPHA(x)								(0x00000900+((x)*32))
+#define  NV34TCL_RC_IN_ALPHA__SIZE							0x00000008
+#define   NV34TCL_RC_IN_ALPHA_D_INPUT_SHIFT						0
+#define   NV34TCL_RC_IN_ALPHA_D_INPUT_MASK						0x0000000f
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_FOG						0x00000003
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_SPARE0_NV					0x0000000c
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_SPARE1_NV					0x0000000d
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV34TCL_RC_IN_ALPHA_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV34TCL_RC_IN_ALPHA_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV34TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV34TCL_RC_IN_ALPHA_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV34TCL_RC_IN_ALPHA_D_MAPPING_SHIFT						5
+#define   NV34TCL_RC_IN_ALPHA_D_MAPPING_MASK						0x000000e0
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NORMAL_NV				0x00000040
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_EXPAND_NEGATE_NV				0x00000060
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV34TCL_RC_IN_ALPHA_D_MAPPING_SIGNED_NEGATE_NV				0x000000e0
+#define   NV34TCL_RC_IN_ALPHA_C_INPUT_SHIFT						8
+#define   NV34TCL_RC_IN_ALPHA_C_INPUT_MASK						0x00000f00
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_FOG						0x00000300
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_SPARE0_NV					0x00000c00
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_SPARE1_NV					0x00000d00
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV34TCL_RC_IN_ALPHA_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_IN_ALPHA_C_COMPONENT_USAGE						(1 << 12)
+#define    NV34TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV34TCL_RC_IN_ALPHA_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV34TCL_RC_IN_ALPHA_C_MAPPING_SHIFT						13
+#define   NV34TCL_RC_IN_ALPHA_C_MAPPING_MASK						0x0000e000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NORMAL_NV				0x00004000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_EXPAND_NEGATE_NV				0x00006000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV34TCL_RC_IN_ALPHA_C_MAPPING_SIGNED_NEGATE_NV				0x0000e000
+#define   NV34TCL_RC_IN_ALPHA_B_INPUT_SHIFT						16
+#define   NV34TCL_RC_IN_ALPHA_B_INPUT_MASK						0x000f0000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR0_NV				0x00010000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_CONSTANT_COLOR1_NV				0x00020000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_FOG						0x00030000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_SECONDARY_COLOR_NV				0x00050000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_SPARE0_NV					0x000c0000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_SPARE1_NV					0x000d0000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV34TCL_RC_IN_ALPHA_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV34TCL_RC_IN_ALPHA_B_COMPONENT_USAGE						(1 << 20)
+#define    NV34TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV34TCL_RC_IN_ALPHA_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV34TCL_RC_IN_ALPHA_B_MAPPING_SHIFT						21
+#define   NV34TCL_RC_IN_ALPHA_B_MAPPING_MASK						0x00e00000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NORMAL_NV				0x00400000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_EXPAND_NEGATE_NV				0x00600000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV34TCL_RC_IN_ALPHA_B_MAPPING_SIGNED_NEGATE_NV				0x00e00000
+#define   NV34TCL_RC_IN_ALPHA_A_INPUT_SHIFT						24
+#define   NV34TCL_RC_IN_ALPHA_A_INPUT_MASK						0x0f000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR0_NV				0x01000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_CONSTANT_COLOR1_NV				0x02000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_FOG						0x03000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_SECONDARY_COLOR_NV				0x05000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_SPARE0_NV					0x0c000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_SPARE1_NV					0x0d000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV34TCL_RC_IN_ALPHA_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV34TCL_RC_IN_ALPHA_A_COMPONENT_USAGE						(1 << 28)
+#define    NV34TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_BLUE					0x00000000
+#define    NV34TCL_RC_IN_ALPHA_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV34TCL_RC_IN_ALPHA_A_MAPPING_SHIFT						29
+#define   NV34TCL_RC_IN_ALPHA_A_MAPPING_MASK						0xe0000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NORMAL_NV				0x40000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_EXPAND_NEGATE_NV				0x60000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV34TCL_RC_IN_ALPHA_A_MAPPING_SIGNED_NEGATE_NV				0xe0000000
+#define  NV34TCL_RC_IN_RGB(x)								(0x00000904+((x)*32))
+#define  NV34TCL_RC_IN_RGB__SIZE							0x00000008
+#define   NV34TCL_RC_IN_RGB_D_INPUT_SHIFT						0
+#define   NV34TCL_RC_IN_RGB_D_INPUT_MASK						0x0000000f
+#define    NV34TCL_RC_IN_RGB_D_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR0_NV					0x00000001
+#define    NV34TCL_RC_IN_RGB_D_INPUT_CONSTANT_COLOR1_NV					0x00000002
+#define    NV34TCL_RC_IN_RGB_D_INPUT_FOG						0x00000003
+#define    NV34TCL_RC_IN_RGB_D_INPUT_PRIMARY_COLOR_NV					0x00000004
+#define    NV34TCL_RC_IN_RGB_D_INPUT_SECONDARY_COLOR_NV					0x00000005
+#define    NV34TCL_RC_IN_RGB_D_INPUT_TEXTURE0_ARB					0x00000008
+#define    NV34TCL_RC_IN_RGB_D_INPUT_TEXTURE1_ARB					0x00000009
+#define    NV34TCL_RC_IN_RGB_D_INPUT_SPARE0_NV						0x0000000c
+#define    NV34TCL_RC_IN_RGB_D_INPUT_SPARE1_NV						0x0000000d
+#define    NV34TCL_RC_IN_RGB_D_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV34TCL_RC_IN_RGB_D_INPUT_E_TIMES_F_NV					0x0000000f
+#define   NV34TCL_RC_IN_RGB_D_COMPONENT_USAGE						(1 <<  4)
+#define    NV34TCL_RC_IN_RGB_D_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_IN_RGB_D_COMPONENT_USAGE_ALPHA					0x00000010
+#define   NV34TCL_RC_IN_RGB_D_MAPPING_SHIFT						5
+#define   NV34TCL_RC_IN_RGB_D_MAPPING_MASK						0x000000e0
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_UNSIGNED_INVERT_NV				0x00000020
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_EXPAND_NORMAL_NV					0x00000040
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_EXPAND_NEGATE_NV					0x00000060
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NORMAL_NV				0x00000080
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_HALF_BIAS_NEGATE_NV				0x000000a0
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_SIGNED_IDENTITY_NV				0x000000c0
+#define    NV34TCL_RC_IN_RGB_D_MAPPING_SIGNED_NEGATE_NV					0x000000e0
+#define   NV34TCL_RC_IN_RGB_C_INPUT_SHIFT						8
+#define   NV34TCL_RC_IN_RGB_C_INPUT_MASK						0x00000f00
+#define    NV34TCL_RC_IN_RGB_C_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR0_NV					0x00000100
+#define    NV34TCL_RC_IN_RGB_C_INPUT_CONSTANT_COLOR1_NV					0x00000200
+#define    NV34TCL_RC_IN_RGB_C_INPUT_FOG						0x00000300
+#define    NV34TCL_RC_IN_RGB_C_INPUT_PRIMARY_COLOR_NV					0x00000400
+#define    NV34TCL_RC_IN_RGB_C_INPUT_SECONDARY_COLOR_NV					0x00000500
+#define    NV34TCL_RC_IN_RGB_C_INPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_IN_RGB_C_INPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_IN_RGB_C_INPUT_SPARE0_NV						0x00000c00
+#define    NV34TCL_RC_IN_RGB_C_INPUT_SPARE1_NV						0x00000d00
+#define    NV34TCL_RC_IN_RGB_C_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV34TCL_RC_IN_RGB_C_INPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_IN_RGB_C_COMPONENT_USAGE						(1 << 12)
+#define    NV34TCL_RC_IN_RGB_C_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_IN_RGB_C_COMPONENT_USAGE_ALPHA					0x00001000
+#define   NV34TCL_RC_IN_RGB_C_MAPPING_SHIFT						13
+#define   NV34TCL_RC_IN_RGB_C_MAPPING_MASK						0x0000e000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_UNSIGNED_INVERT_NV				0x00002000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_EXPAND_NORMAL_NV					0x00004000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_EXPAND_NEGATE_NV					0x00006000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NORMAL_NV				0x00008000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_HALF_BIAS_NEGATE_NV				0x0000a000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_SIGNED_IDENTITY_NV				0x0000c000
+#define    NV34TCL_RC_IN_RGB_C_MAPPING_SIGNED_NEGATE_NV					0x0000e000
+#define   NV34TCL_RC_IN_RGB_B_INPUT_SHIFT						16
+#define   NV34TCL_RC_IN_RGB_B_INPUT_MASK						0x000f0000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR0_NV					0x00010000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_CONSTANT_COLOR1_NV					0x00020000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_FOG						0x00030000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_PRIMARY_COLOR_NV					0x00040000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_SECONDARY_COLOR_NV					0x00050000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_TEXTURE0_ARB					0x00080000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_TEXTURE1_ARB					0x00090000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_SPARE0_NV						0x000c0000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_SPARE1_NV						0x000d0000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000e0000
+#define    NV34TCL_RC_IN_RGB_B_INPUT_E_TIMES_F_NV					0x000f0000
+#define   NV34TCL_RC_IN_RGB_B_COMPONENT_USAGE						(1 << 20)
+#define    NV34TCL_RC_IN_RGB_B_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_IN_RGB_B_COMPONENT_USAGE_ALPHA					0x00100000
+#define   NV34TCL_RC_IN_RGB_B_MAPPING_SHIFT						21
+#define   NV34TCL_RC_IN_RGB_B_MAPPING_MASK						0x00e00000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_UNSIGNED_INVERT_NV				0x00200000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_EXPAND_NORMAL_NV					0x00400000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_EXPAND_NEGATE_NV					0x00600000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NORMAL_NV				0x00800000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_HALF_BIAS_NEGATE_NV				0x00a00000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_SIGNED_IDENTITY_NV				0x00c00000
+#define    NV34TCL_RC_IN_RGB_B_MAPPING_SIGNED_NEGATE_NV					0x00e00000
+#define   NV34TCL_RC_IN_RGB_A_INPUT_SHIFT						24
+#define   NV34TCL_RC_IN_RGB_A_INPUT_MASK						0x0f000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_ZERO						0x00000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR0_NV					0x01000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_CONSTANT_COLOR1_NV					0x02000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_FOG						0x03000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_PRIMARY_COLOR_NV					0x04000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_SECONDARY_COLOR_NV					0x05000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_TEXTURE0_ARB					0x08000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_TEXTURE1_ARB					0x09000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_SPARE0_NV						0x0c000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_SPARE1_NV						0x0d000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0e000000
+#define    NV34TCL_RC_IN_RGB_A_INPUT_E_TIMES_F_NV					0x0f000000
+#define   NV34TCL_RC_IN_RGB_A_COMPONENT_USAGE						(1 << 28)
+#define    NV34TCL_RC_IN_RGB_A_COMPONENT_USAGE_RGB					0x00000000
+#define    NV34TCL_RC_IN_RGB_A_COMPONENT_USAGE_ALPHA					0x10000000
+#define   NV34TCL_RC_IN_RGB_A_MAPPING_SHIFT						29
+#define   NV34TCL_RC_IN_RGB_A_MAPPING_MASK						0xe0000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_IDENTITY_NV				0x00000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_UNSIGNED_INVERT_NV				0x20000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_EXPAND_NORMAL_NV					0x40000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_EXPAND_NEGATE_NV					0x60000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NORMAL_NV				0x80000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_HALF_BIAS_NEGATE_NV				0xa0000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_SIGNED_IDENTITY_NV				0xc0000000
+#define    NV34TCL_RC_IN_RGB_A_MAPPING_SIGNED_NEGATE_NV					0xe0000000
+#define  NV34TCL_RC_CONSTANT_COLOR0(x)							(0x00000908+((x)*32))
+#define  NV34TCL_RC_CONSTANT_COLOR0__SIZE						0x00000008
+#define   NV34TCL_RC_CONSTANT_COLOR0_B_SHIFT						0
+#define   NV34TCL_RC_CONSTANT_COLOR0_B_MASK						0x000000ff
+#define   NV34TCL_RC_CONSTANT_COLOR0_G_SHIFT						8
+#define   NV34TCL_RC_CONSTANT_COLOR0_G_MASK						0x0000ff00
+#define   NV34TCL_RC_CONSTANT_COLOR0_R_SHIFT						16
+#define   NV34TCL_RC_CONSTANT_COLOR0_R_MASK						0x00ff0000
+#define   NV34TCL_RC_CONSTANT_COLOR0_A_SHIFT						24
+#define   NV34TCL_RC_CONSTANT_COLOR0_A_MASK						0xff000000
+#define  NV34TCL_RC_CONSTANT_COLOR1(x)							(0x0000090c+((x)*32))
+#define  NV34TCL_RC_CONSTANT_COLOR1__SIZE						0x00000008
+#define   NV34TCL_RC_CONSTANT_COLOR1_B_SHIFT						0
+#define   NV34TCL_RC_CONSTANT_COLOR1_B_MASK						0x000000ff
+#define   NV34TCL_RC_CONSTANT_COLOR1_G_SHIFT						8
+#define   NV34TCL_RC_CONSTANT_COLOR1_G_MASK						0x0000ff00
+#define   NV34TCL_RC_CONSTANT_COLOR1_R_SHIFT						16
+#define   NV34TCL_RC_CONSTANT_COLOR1_R_MASK						0x00ff0000
+#define   NV34TCL_RC_CONSTANT_COLOR1_A_SHIFT						24
+#define   NV34TCL_RC_CONSTANT_COLOR1_A_MASK						0xff000000
+#define  NV34TCL_RC_OUT_ALPHA(x)							(0x00000910+((x)*32))
+#define  NV34TCL_RC_OUT_ALPHA__SIZE							0x00000008
+#define   NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_SHIFT						0
+#define   NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_MASK						0x0000000f
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_FOG						0x00000003
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x0000000e
+#define    NV34TCL_RC_OUT_ALPHA_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_SHIFT						4
+#define   NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_MASK						0x000000f0
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_FOG						0x00000030
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x000000e0
+#define    NV34TCL_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT						8
+#define   NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_MASK						0x00000f00
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_FOG						0x00000300
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV		0x00000e00
+#define    NV34TCL_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_OUT_ALPHA_CD_DOT_PRODUCT						(1 << 12)
+#define   NV34TCL_RC_OUT_ALPHA_AB_DOT_PRODUCT						(1 << 13)
+#define   NV34TCL_RC_OUT_ALPHA_MUX_SUM							(1 << 14)
+#define   NV34TCL_RC_OUT_ALPHA_BIAS							(1 << 15)
+#define    NV34TCL_RC_OUT_ALPHA_BIAS_NONE						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV			0x00008000
+#define   NV34TCL_RC_OUT_ALPHA_SCALE_SHIFT						17
+#define   NV34TCL_RC_OUT_ALPHA_SCALE_MASK						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_SCALE_NONE						0x00000000
+#define    NV34TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV34TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV34TCL_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define  NV34TCL_RC_OUT_RGB(x)								(0x00000914+((x)*32))
+#define  NV34TCL_RC_OUT_RGB__SIZE							0x00000008
+#define   NV34TCL_RC_OUT_RGB_CD_OUTPUT_SHIFT						0
+#define   NV34TCL_RC_OUT_RGB_CD_OUTPUT_MASK						0x0000000f
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR0_NV				0x00000001
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_CONSTANT_COLOR1_NV				0x00000002
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_FOG						0x00000003
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_PRIMARY_COLOR_NV				0x00000004
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_SECONDARY_COLOR_NV				0x00000005
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE0_ARB					0x00000008
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_TEXTURE1_ARB					0x00000009
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_NV					0x0000000c
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_SPARE1_NV					0x0000000d
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x0000000e
+#define    NV34TCL_RC_OUT_RGB_CD_OUTPUT_E_TIMES_F_NV					0x0000000f
+#define   NV34TCL_RC_OUT_RGB_AB_OUTPUT_SHIFT						4
+#define   NV34TCL_RC_OUT_RGB_AB_OUTPUT_MASK						0x000000f0
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV				0x00000010
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV				0x00000020
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_FOG						0x00000030
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV				0x00000040
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV				0x00000050
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB					0x00000080
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB					0x00000090
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV					0x000000c0
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV					0x000000d0
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x000000e0
+#define    NV34TCL_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV					0x000000f0
+#define   NV34TCL_RC_OUT_RGB_SUM_OUTPUT_SHIFT						8
+#define   NV34TCL_RC_OUT_RGB_SUM_OUTPUT_MASK						0x00000f00
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_ZERO						0x00000000
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV				0x00000100
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV				0x00000200
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_FOG						0x00000300
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV				0x00000400
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV				0x00000500
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB					0x00000800
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB					0x00000900
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV					0x00000c00
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV					0x00000d00
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV			0x00000e00
+#define    NV34TCL_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV					0x00000f00
+#define   NV34TCL_RC_OUT_RGB_CD_DOT_PRODUCT						(1 << 12)
+#define   NV34TCL_RC_OUT_RGB_AB_DOT_PRODUCT						(1 << 13)
+#define   NV34TCL_RC_OUT_RGB_MUX_SUM							(1 << 14)
+#define   NV34TCL_RC_OUT_RGB_BIAS							(1 << 15)
+#define    NV34TCL_RC_OUT_RGB_BIAS_NONE							0x00000000
+#define    NV34TCL_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV				0x00008000
+#define   NV34TCL_RC_OUT_RGB_SCALE_SHIFT						17
+#define   NV34TCL_RC_OUT_RGB_SCALE_MASK							0x00000000
+#define    NV34TCL_RC_OUT_RGB_SCALE_NONE						0x00000000
+#define    NV34TCL_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV					0x00020000
+#define    NV34TCL_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV					0x00040000
+#define    NV34TCL_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV				0x00060000
+#define  NV34TCL_VIEWPORT_HORIZ								0x00000a00
+#define   NV34TCL_VIEWPORT_HORIZ_X_SHIFT						0
+#define   NV34TCL_VIEWPORT_HORIZ_X_MASK							0x0000ffff
+#define   NV34TCL_VIEWPORT_HORIZ_W_SHIFT						16
+#define   NV34TCL_VIEWPORT_HORIZ_W_MASK							0xffff0000
+#define  NV34TCL_VIEWPORT_VERT								0x00000a04
+#define   NV34TCL_VIEWPORT_VERT_Y_SHIFT							0
+#define   NV34TCL_VIEWPORT_VERT_Y_MASK							0x0000ffff
+#define   NV34TCL_VIEWPORT_VERT_H_SHIFT							16
+#define   NV34TCL_VIEWPORT_VERT_H_MASK							0xffff0000
+#define  NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R			0x00000a10
+#define  NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G			0x00000a14
+#define  NV34TCL_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B			0x00000a18
+#define  NV34TCL_VIEWPORT_TRANSLATE_X							0x00000a20
+#define  NV34TCL_VIEWPORT_TRANSLATE_Y							0x00000a24
+#define  NV34TCL_VIEWPORT_TRANSLATE_Z							0x00000a28
+#define  NV34TCL_VIEWPORT_TRANSLATE_W							0x00000a2c
+#define  NV34TCL_VIEWPORT_SCALE_X							0x00000a30
+#define  NV34TCL_VIEWPORT_SCALE_Y							0x00000a34
+#define  NV34TCL_VIEWPORT_SCALE_Z							0x00000a38
+#define  NV34TCL_VIEWPORT_SCALE_W							0x00000a3c
+#define  NV34TCL_POLYGON_OFFSET_POINT_ENABLE						0x00000a60
+#define  NV34TCL_POLYGON_OFFSET_LINE_ENABLE						0x00000a64
+#define  NV34TCL_POLYGON_OFFSET_FILL_ENABLE						0x00000a68
+#define  NV34TCL_DEPTH_FUNC								0x00000a6c
+#define   NV34TCL_DEPTH_FUNC_NEVER							0x00000200
+#define   NV34TCL_DEPTH_FUNC_LESS							0x00000201
+#define   NV34TCL_DEPTH_FUNC_EQUAL							0x00000202
+#define   NV34TCL_DEPTH_FUNC_LEQUAL							0x00000203
+#define   NV34TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV34TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV34TCL_DEPTH_FUNC_NOTEQUAL							0x00000205
+#define   NV34TCL_DEPTH_FUNC_GEQUAL							0x00000206
+#define   NV34TCL_DEPTH_FUNC_ALWAYS							0x00000207
+#define  NV34TCL_DEPTH_WRITE_ENABLE							0x00000a70
+#define  NV34TCL_DEPTH_TEST_ENABLE							0x00000a74
+#define  NV34TCL_POLYGON_OFFSET_FACTOR							0x00000a78
+#define  NV34TCL_POLYGON_OFFSET_UNITS							0x00000a7c
+#define  NV34TCL_VTX_ATTR_3I_XY(x)							(0x00000a80+((x)*8))
+#define  NV34TCL_VTX_ATTR_3I_XY__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_3I_XY_X_SHIFT						0
+#define   NV34TCL_VTX_ATTR_3I_XY_X_MASK							0x0000ffff
+#define   NV34TCL_VTX_ATTR_3I_XY_Y_SHIFT						16
+#define   NV34TCL_VTX_ATTR_3I_XY_Y_MASK							0xffff0000
+#define  NV34TCL_VTX_ATTR_3I_Z(x)							(0x00000a84+((x)*8))
+#define  NV34TCL_VTX_ATTR_3I_Z__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_3I_Z_Z_SHIFT							0
+#define   NV34TCL_VTX_ATTR_3I_Z_Z_MASK							0x0000ffff
+#define  NV34TCL_VP_UPLOAD_INST(x)							(0x00000b80+((x)*4))
+#define  NV34TCL_VP_UPLOAD_INST__SIZE							0x00000004
+#define  NV34TCL_TX0_CLIP_PLANE_A(x)							(0x00000e00+((x)*16))
+#define  NV34TCL_TX0_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX0_CLIP_PLANE_B(x)							(0x00000e04+((x)*16))
+#define  NV34TCL_TX0_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX0_CLIP_PLANE_C(x)							(0x00000e08+((x)*16))
+#define  NV34TCL_TX0_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX0_CLIP_PLANE_D(x)							(0x00000e0c+((x)*16))
+#define  NV34TCL_TX0_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX1_CLIP_PLANE_A(x)							(0x00000e40+((x)*16))
+#define  NV34TCL_TX1_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX1_CLIP_PLANE_B(x)							(0x00000e44+((x)*16))
+#define  NV34TCL_TX1_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX1_CLIP_PLANE_C(x)							(0x00000e48+((x)*16))
+#define  NV34TCL_TX1_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX1_CLIP_PLANE_D(x)							(0x00000e4c+((x)*16))
+#define  NV34TCL_TX1_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX2_CLIP_PLANE_A(x)							(0x00000e80+((x)*16))
+#define  NV34TCL_TX2_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX2_CLIP_PLANE_B(x)							(0x00000e84+((x)*16))
+#define  NV34TCL_TX2_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX2_CLIP_PLANE_C(x)							(0x00000e88+((x)*16))
+#define  NV34TCL_TX2_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX2_CLIP_PLANE_D(x)							(0x00000e8c+((x)*16))
+#define  NV34TCL_TX2_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX3_CLIP_PLANE_A(x)							(0x00000ec0+((x)*16))
+#define  NV34TCL_TX3_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX3_CLIP_PLANE_B(x)							(0x00000ec4+((x)*16))
+#define  NV34TCL_TX3_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX3_CLIP_PLANE_C(x)							(0x00000ec8+((x)*16))
+#define  NV34TCL_TX3_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX3_CLIP_PLANE_D(x)							(0x00000ecc+((x)*16))
+#define  NV34TCL_TX3_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX4_CLIP_PLANE_A(x)							(0x00000f00+((x)*16))
+#define  NV34TCL_TX4_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX4_CLIP_PLANE_B(x)							(0x00000f04+((x)*16))
+#define  NV34TCL_TX4_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX4_CLIP_PLANE_C(x)							(0x00000f08+((x)*16))
+#define  NV34TCL_TX4_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX4_CLIP_PLANE_D(x)							(0x00000f0c+((x)*16))
+#define  NV34TCL_TX4_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX5_CLIP_PLANE_A(x)							(0x00000f40+((x)*16))
+#define  NV34TCL_TX5_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX5_CLIP_PLANE_B(x)							(0x00000f44+((x)*16))
+#define  NV34TCL_TX5_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX5_CLIP_PLANE_C(x)							(0x00000f48+((x)*16))
+#define  NV34TCL_TX5_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX5_CLIP_PLANE_D(x)							(0x00000f4c+((x)*16))
+#define  NV34TCL_TX5_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX6_CLIP_PLANE_A(x)							(0x00000f80+((x)*16))
+#define  NV34TCL_TX6_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX6_CLIP_PLANE_B(x)							(0x00000f84+((x)*16))
+#define  NV34TCL_TX6_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX6_CLIP_PLANE_C(x)							(0x00000f88+((x)*16))
+#define  NV34TCL_TX6_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX6_CLIP_PLANE_D(x)							(0x00000f8c+((x)*16))
+#define  NV34TCL_TX6_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_TX7_CLIP_PLANE_A(x)							(0x00000fc0+((x)*16))
+#define  NV34TCL_TX7_CLIP_PLANE_A__SIZE							0x00000004
+#define  NV34TCL_TX7_CLIP_PLANE_B(x)							(0x00000fc4+((x)*16))
+#define  NV34TCL_TX7_CLIP_PLANE_B__SIZE							0x00000004
+#define  NV34TCL_TX7_CLIP_PLANE_C(x)							(0x00000fc8+((x)*16))
+#define  NV34TCL_TX7_CLIP_PLANE_C__SIZE							0x00000004
+#define  NV34TCL_TX7_CLIP_PLANE_D(x)							(0x00000fcc+((x)*16))
+#define  NV34TCL_TX7_CLIP_PLANE_D__SIZE							0x00000004
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(x)					(0x00001000+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(x)					(0x00001004+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(x)					(0x00001008+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(x)					(0x0000100c+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(x)					(0x00001010+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(x)					(0x00001014+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(x)					(0x00001018+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(x)					(0x0000101c+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G__SIZE				0x00000008
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(x)					(0x00001020+((x)*64))
+#define  NV34TCL_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B__SIZE				0x00000008
+#define  NV34TCL_LIGHT_HALF_VECTOR_X(x)							(0x00001028+((x)*64))
+#define  NV34TCL_LIGHT_HALF_VECTOR_X__SIZE						0x00000008
+#define  NV34TCL_LIGHT_HALF_VECTOR_Y(x)							(0x0000102c+((x)*64))
+#define  NV34TCL_LIGHT_HALF_VECTOR_Y__SIZE						0x00000008
+#define  NV34TCL_LIGHT_HALF_VECTOR_Z(x)							(0x00001030+((x)*64))
+#define  NV34TCL_LIGHT_HALF_VECTOR_Z__SIZE						0x00000008
+#define  NV34TCL_LIGHT_DIRECTION_X(x)							(0x00001034+((x)*64))
+#define  NV34TCL_LIGHT_DIRECTION_X__SIZE						0x00000008
+#define  NV34TCL_LIGHT_DIRECTION_Y(x)							(0x00001038+((x)*64))
+#define  NV34TCL_LIGHT_DIRECTION_Y__SIZE						0x00000008
+#define  NV34TCL_LIGHT_DIRECTION_Z(x)							(0x0000103c+((x)*64))
+#define  NV34TCL_LIGHT_DIRECTION_Z__SIZE						0x00000008
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_A(x)							(0x00001200+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_A__SIZE						0x00000008
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_B(x)							(0x00001204+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_B__SIZE						0x00000008
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_C(x)							(0x00001208+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_C__SIZE						0x00000008
+#define  NV34TCL_LIGHT_SPOT_DIR_X(x)							(0x0000120c+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_DIR_X__SIZE							0x00000008
+#define  NV34TCL_LIGHT_SPOT_DIR_Y(x)							(0x00001210+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_DIR_Y__SIZE							0x00000008
+#define  NV34TCL_LIGHT_SPOT_DIR_Z(x)							(0x00001214+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_DIR_Z__SIZE							0x00000008
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_D(x)							(0x00001218+((x)*64))
+#define  NV34TCL_LIGHT_SPOT_CUTOFF_D__SIZE						0x00000008
+#define  NV34TCL_LIGHT_POSITION_X(x)							(0x0000121c+((x)*64))
+#define  NV34TCL_LIGHT_POSITION_X__SIZE							0x00000008
+#define  NV34TCL_LIGHT_POSITION_Y(x)							(0x00001220+((x)*64))
+#define  NV34TCL_LIGHT_POSITION_Y__SIZE							0x00000008
+#define  NV34TCL_LIGHT_POSITION_Z(x)							(0x00001224+((x)*64))
+#define  NV34TCL_LIGHT_POSITION_Z__SIZE							0x00000008
+#define  NV34TCL_LIGHT_ATTENUATION_CONSTANT(x)						(0x00001228+((x)*64))
+#define  NV34TCL_LIGHT_ATTENUATION_CONSTANT__SIZE					0x00000008
+#define  NV34TCL_LIGHT_ATTENUATION_LINEAR(x)						(0x0000122c+((x)*64))
+#define  NV34TCL_LIGHT_ATTENUATION_LINEAR__SIZE						0x00000008
+#define  NV34TCL_LIGHT_ATTENUATION_QUADRATIC(x)						(0x00001230+((x)*64))
+#define  NV34TCL_LIGHT_ATTENUATION_QUADRATIC__SIZE					0x00000008
+#define  NV34TCL_FRONT_MATERIAL_SHININESS(x)						(0x00001400+((x)*4))
+#define  NV34TCL_FRONT_MATERIAL_SHININESS__SIZE						0x00000006
+#define  NV34TCL_ENABLED_LIGHTS								0x00001420
+#define  NV34TCL_FP_REG_CONTROL								0x00001450
+#define   NV34TCL_FP_REG_CONTROL_UNK1_SHIFT						16
+#define   NV34TCL_FP_REG_CONTROL_UNK1_MASK						0xffff0000
+#define   NV34TCL_FP_REG_CONTROL_UNK0_SHIFT						0
+#define   NV34TCL_FP_REG_CONTROL_UNK0_MASK						0x0000ffff
+#define  NV34TCL_VP_CLIP_PLANES_ENABLE							0x00001478
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE0						(1 <<  1)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE1						(1 <<  5)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE2						(1 <<  9)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE3						(1 << 13)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE4						(1 << 17)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE5						(1 << 21)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE6						(1 << 25)
+#define   NV34TCL_VP_CLIP_PLANES_ENABLE_PLANE7						(1 << 29)
+#define  NV34TCL_POLYGON_STIPPLE_ENABLE							0x0000147c
+#define  NV34TCL_POLYGON_STIPPLE_PATTERN(x)						(0x00001480+((x)*4))
+#define  NV34TCL_POLYGON_STIPPLE_PATTERN__SIZE						0x00000020
+#define  NV34TCL_VTX_ATTR_3F_X(x)							(0x00001500+((x)*16))
+#define  NV34TCL_VTX_ATTR_3F_X__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_3F_Y(x)							(0x00001504+((x)*16))
+#define  NV34TCL_VTX_ATTR_3F_Y__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_3F_Z(x)							(0x00001508+((x)*16))
+#define  NV34TCL_VTX_ATTR_3F_Z__SIZE							0x00000010
+#define  NV34TCL_VP_CLIP_PLANE_A(x)							(0x00001600+((x)*16))
+#define  NV34TCL_VP_CLIP_PLANE_A__SIZE							0x00000006
+#define  NV34TCL_VP_CLIP_PLANE_B(x)							(0x00001604+((x)*16))
+#define  NV34TCL_VP_CLIP_PLANE_B__SIZE							0x00000006
+#define  NV34TCL_VP_CLIP_PLANE_C(x)							(0x00001608+((x)*16))
+#define  NV34TCL_VP_CLIP_PLANE_C__SIZE							0x00000006
+#define  NV34TCL_VP_CLIP_PLANE_D(x)							(0x0000160c+((x)*16))
+#define  NV34TCL_VP_CLIP_PLANE_D__SIZE							0x00000006
+#define  NV34TCL_VTXBUF_ADDRESS(x)							(0x00001680+((x)*4))
+#define  NV34TCL_VTXBUF_ADDRESS__SIZE							0x00000010
+#define   NV34TCL_VTXBUF_ADDRESS_DMA1							(1 << 31)
+#define   NV34TCL_VTXBUF_ADDRESS_OFFSET_SHIFT						0
+#define   NV34TCL_VTXBUF_ADDRESS_OFFSET_MASK						0x0fffffff
+#define  NV34TCL_VTXFMT(x)								(0x00001740+((x)*4))
+#define  NV34TCL_VTXFMT__SIZE								0x00000010
+#define   NV34TCL_VTXFMT_TYPE_SHIFT							0
+#define   NV34TCL_VTXFMT_TYPE_MASK							0x0000000f
+#define    NV34TCL_VTXFMT_TYPE_FLOAT							0x00000002
+#define    NV34TCL_VTXFMT_TYPE_UBYTE							0x00000004
+#define    NV34TCL_VTXFMT_TYPE_USHORT							0x00000005
+#define   NV34TCL_VTXFMT_SIZE_SHIFT							4
+#define   NV34TCL_VTXFMT_SIZE_MASK							0x000000f0
+#define   NV34TCL_VTXFMT_STRIDE_SHIFT							8
+#define   NV34TCL_VTXFMT_STRIDE_MASK							0x0000ff00
+#define  NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R			0x000017a0
+#define  NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G			0x000017a4
+#define  NV34TCL_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B			0x000017a8
+#define  NV34TCL_COLOR_MATERIAL_BACK_R							0x000017b0
+#define  NV34TCL_COLOR_MATERIAL_BACK_G							0x000017b4
+#define  NV34TCL_COLOR_MATERIAL_BACK_B							0x000017b8
+#define  NV34TCL_COLOR_MATERIAL_BACK_A							0x000017c0
+#define  NV34TCL_QUERY_RESET								0x000017c8
+#define  NV34TCL_QUERY_UNK17CC								0x000017cc
+#define  NV34TCL_QUERY_GET								0x00001800
+#define   NV34TCL_QUERY_GET_UNK24_SHIFT							24
+#define   NV34TCL_QUERY_GET_UNK24_MASK							0xff000000
+#define   NV34TCL_QUERY_GET_OFFSET_SHIFT						0
+#define   NV34TCL_QUERY_GET_OFFSET_MASK							0x00ffffff
+#define  NV34TCL_VERTEX_BEGIN_END							0x00001808
+#define   NV34TCL_VERTEX_BEGIN_END_STOP							0x00000000
+#define   NV34TCL_VERTEX_BEGIN_END_POINTS						0x00000001
+#define   NV34TCL_VERTEX_BEGIN_END_LINES						0x00000002
+#define   NV34TCL_VERTEX_BEGIN_END_LINE_LOOP						0x00000003
+#define   NV34TCL_VERTEX_BEGIN_END_LINE_STRIP						0x00000004
+#define   NV34TCL_VERTEX_BEGIN_END_TRIANGLES						0x00000005
+#define   NV34TCL_VERTEX_BEGIN_END_TRIANGLE_STRIP					0x00000006
+#define   NV34TCL_VERTEX_BEGIN_END_TRIANGLE_FAN						0x00000007
+#define   NV34TCL_VERTEX_BEGIN_END_QUADS						0x00000008
+#define   NV34TCL_VERTEX_BEGIN_END_QUAD_STRIP						0x00000009
+#define   NV34TCL_VERTEX_BEGIN_END_POLYGON						0x0000000a
+#define  NV34TCL_VB_ELEMENT_U16								0x0000180c
+#define   NV34TCL_VB_ELEMENT_U16_I0_SHIFT						0
+#define   NV34TCL_VB_ELEMENT_U16_I0_MASK						0x0000ffff
+#define   NV34TCL_VB_ELEMENT_U16_I1_SHIFT						16
+#define   NV34TCL_VB_ELEMENT_U16_I1_MASK						0xffff0000
+#define  NV34TCL_VB_ELEMENT_U32								0x00001810
+#define  NV34TCL_VB_VERTEX_BATCH							0x00001814
+#define   NV34TCL_VB_VERTEX_BATCH_OFFSET_SHIFT						0
+#define   NV34TCL_VB_VERTEX_BATCH_OFFSET_MASK						0x00ffffff
+#define   NV34TCL_VB_VERTEX_BATCH_COUNT_SHIFT						24
+#define   NV34TCL_VB_VERTEX_BATCH_COUNT_MASK						0xff000000
+#define  NV34TCL_VERTEX_DATA								0x00001818
+#define  NV34TCL_IDXBUF_ADDRESS								0x0000181c
+#define  NV34TCL_IDXBUF_FORMAT								0x00001820
+#define   NV34TCL_IDXBUF_FORMAT_TYPE_SHIFT						4
+#define   NV34TCL_IDXBUF_FORMAT_TYPE_MASK						0x000000f0
+#define    NV34TCL_IDXBUF_FORMAT_TYPE_U32						0x00000000
+#define    NV34TCL_IDXBUF_FORMAT_TYPE_U16						0x00000010
+#define   NV34TCL_IDXBUF_FORMAT_DMA1							(1 <<  0)
+#define  NV34TCL_VB_INDEX_BATCH								0x00001824
+#define   NV34TCL_VB_INDEX_BATCH_COUNT_SHIFT						24
+#define   NV34TCL_VB_INDEX_BATCH_COUNT_MASK						0xff000000
+#define   NV34TCL_VB_INDEX_BATCH_START_SHIFT						0
+#define   NV34TCL_VB_INDEX_BATCH_START_MASK						0x00ffffff
+#define  NV34TCL_POLYGON_MODE_FRONT							0x00001828
+#define   NV34TCL_POLYGON_MODE_FRONT_POINT						0x00001b00
+#define   NV34TCL_POLYGON_MODE_FRONT_LINE						0x00001b01
+#define   NV34TCL_POLYGON_MODE_FRONT_FILL						0x00001b02
+#define  NV34TCL_POLYGON_MODE_BACK							0x0000182c
+#define   NV34TCL_POLYGON_MODE_BACK_POINT						0x00001b00
+#define   NV34TCL_POLYGON_MODE_BACK_LINE						0x00001b01
+#define   NV34TCL_POLYGON_MODE_BACK_FILL						0x00001b02
+#define  NV34TCL_CULL_FACE								0x00001830
+#define   NV34TCL_CULL_FACE_FRONT							0x00000404
+#define   NV34TCL_CULL_FACE_BACK							0x00000405
+#define   NV34TCL_CULL_FACE_FRONT_AND_BACK						0x00000408
+#define  NV34TCL_FRONT_FACE								0x00001834
+#define   NV34TCL_FRONT_FACE_CW								0x00000900
+#define   NV34TCL_FRONT_FACE_CCW							0x00000901
+#define  NV34TCL_POLYGON_SMOOTH_ENABLE							0x00001838
+#define  NV34TCL_CULL_FACE_ENABLE							0x0000183c
+#define  NV34TCL_TX_PALETTE_OFFSET(x)							(0x00001840+((x)*4))
+#define  NV34TCL_TX_PALETTE_OFFSET__SIZE						0x00000004
+#define  NV34TCL_VTX_ATTR_2F_X(x)							(0x00001880+((x)*8))
+#define  NV34TCL_VTX_ATTR_2F_X__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_2F_Y(x)							(0x00001884+((x)*8))
+#define  NV34TCL_VTX_ATTR_2F_Y__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_2I(x)								(0x00001900+((x)*4))
+#define  NV34TCL_VTX_ATTR_2I__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_2I_X_SHIFT							0
+#define   NV34TCL_VTX_ATTR_2I_X_MASK							0x0000ffff
+#define   NV34TCL_VTX_ATTR_2I_Y_SHIFT							16
+#define   NV34TCL_VTX_ATTR_2I_Y_MASK							0xffff0000
+#define  NV34TCL_VTX_ATTR_4UB(x)							(0x00001940+((x)*4))
+#define  NV34TCL_VTX_ATTR_4UB__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_4UB_X_SHIFT							0
+#define   NV34TCL_VTX_ATTR_4UB_X_MASK							0x000000ff
+#define   NV34TCL_VTX_ATTR_4UB_Y_SHIFT							8
+#define   NV34TCL_VTX_ATTR_4UB_Y_MASK							0x0000ff00
+#define   NV34TCL_VTX_ATTR_4UB_Z_SHIFT							16
+#define   NV34TCL_VTX_ATTR_4UB_Z_MASK							0x00ff0000
+#define   NV34TCL_VTX_ATTR_4UB_W_SHIFT							24
+#define   NV34TCL_VTX_ATTR_4UB_W_MASK							0xff000000
+#define  NV34TCL_VTX_ATTR_4I_XY(x)							(0x00001980+((x)*8))
+#define  NV34TCL_VTX_ATTR_4I_XY__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_4I_XY_X_SHIFT						0
+#define   NV34TCL_VTX_ATTR_4I_XY_X_MASK							0x0000ffff
+#define   NV34TCL_VTX_ATTR_4I_XY_Y_SHIFT						16
+#define   NV34TCL_VTX_ATTR_4I_XY_Y_MASK							0xffff0000
+#define  NV34TCL_VTX_ATTR_4I_ZW(x)							(0x00001984+((x)*8))
+#define  NV34TCL_VTX_ATTR_4I_ZW__SIZE							0x00000010
+#define   NV34TCL_VTX_ATTR_4I_ZW_Z_SHIFT						0
+#define   NV34TCL_VTX_ATTR_4I_ZW_Z_MASK							0x0000ffff
+#define   NV34TCL_VTX_ATTR_4I_ZW_W_SHIFT						16
+#define   NV34TCL_VTX_ATTR_4I_ZW_W_MASK							0xffff0000
+#define  NV34TCL_TX_OFFSET(x)								(0x00001a00+((x)*32))
+#define  NV34TCL_TX_OFFSET__SIZE							0x00000004
+#define  NV34TCL_TX_FORMAT(x)								(0x00001a04+((x)*32))
+#define  NV34TCL_TX_FORMAT__SIZE							0x00000004
+#define   NV34TCL_TX_FORMAT_DMA0							(1 <<  0)
+#define   NV34TCL_TX_FORMAT_DMA1							(1 <<  1)
+#define   NV34TCL_TX_FORMAT_CUBIC							(1 <<  2)
+#define   NV34TCL_TX_FORMAT_NO_BORDER							(1 <<  3)
+#define   NV34TCL_TX_FORMAT_DIMS_SHIFT							4
+#define   NV34TCL_TX_FORMAT_DIMS_MASK							0x000000f0
+#define    NV34TCL_TX_FORMAT_DIMS_1D							0x00000010
+#define    NV34TCL_TX_FORMAT_DIMS_2D							0x00000020
+#define    NV34TCL_TX_FORMAT_DIMS_3D							0x00000030
+#define   NV34TCL_TX_FORMAT_FORMAT_SHIFT						8
+#define   NV34TCL_TX_FORMAT_FORMAT_MASK							0x0000ff00
+#define    NV34TCL_TX_FORMAT_FORMAT_L8							0x00000000
+#define    NV34TCL_TX_FORMAT_FORMAT_A8							0x00000100
+#define    NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5						0x00000200
+#define    NV34TCL_TX_FORMAT_FORMAT_A8_RECT						0x00000300
+#define    NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4						0x00000400
+#define    NV34TCL_TX_FORMAT_FORMAT_R5G6B5						0x00000500
+#define    NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8						0x00000600
+#define    NV34TCL_TX_FORMAT_FORMAT_X8R8G8B8						0x00000700
+#define    NV34TCL_TX_FORMAT_FORMAT_INDEX8						0x00000b00
+#define    NV34TCL_TX_FORMAT_FORMAT_DXT1						0x00000c00
+#define    NV34TCL_TX_FORMAT_FORMAT_DXT3						0x00000e00
+#define    NV34TCL_TX_FORMAT_FORMAT_DXT5						0x00000f00
+#define    NV34TCL_TX_FORMAT_FORMAT_A1R5G5B5_RECT					0x00001000
+#define    NV34TCL_TX_FORMAT_FORMAT_R5G6B5_RECT						0x00001100
+#define    NV34TCL_TX_FORMAT_FORMAT_A8R8G8B8_RECT					0x00001200
+#define    NV34TCL_TX_FORMAT_FORMAT_L8_RECT						0x00001300
+#define    NV34TCL_TX_FORMAT_FORMAT_A8L8						0x00001a00
+#define    NV34TCL_TX_FORMAT_FORMAT_A8_RECT2						0x00001b00
+#define    NV34TCL_TX_FORMAT_FORMAT_A4R4G4B4_RECT					0x00001d00
+#define    NV34TCL_TX_FORMAT_FORMAT_R8G8B8_RECT						0x00001e00
+#define    NV34TCL_TX_FORMAT_FORMAT_L8A8_RECT						0x00002000
+#define    NV34TCL_TX_FORMAT_FORMAT_DSDT						0x00002800
+#define    NV34TCL_TX_FORMAT_FORMAT_A16							0x00003200
+#define    NV34TCL_TX_FORMAT_FORMAT_HILO16						0x00003300
+#define    NV34TCL_TX_FORMAT_FORMAT_A16_RECT						0x00003500
+#define    NV34TCL_TX_FORMAT_FORMAT_HILO16_RECT						0x00003600
+#define    NV34TCL_TX_FORMAT_FORMAT_HILO8						0x00004400
+#define    NV34TCL_TX_FORMAT_FORMAT_SIGNED_HILO8					0x00004500
+#define    NV34TCL_TX_FORMAT_FORMAT_HILO8_RECT						0x00004600
+#define    NV34TCL_TX_FORMAT_FORMAT_SIGNED_HILO8_RECT					0x00004700
+#define    NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA16_NV					0x00004a00
+#define    NV34TCL_TX_FORMAT_FORMAT_FLOAT_RGBA32_NV					0x00004b00
+#define    NV34TCL_TX_FORMAT_FORMAT_FLOAT_R32_NV					0x00004c00
+#define   NV34TCL_TX_FORMAT_MIPMAP							(1 << 19)
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_U_SHIFT						20
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_U_MASK						0x00f00000
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_V_SHIFT						24
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_V_MASK						0x0f000000
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_W_SHIFT						28
+#define   NV34TCL_TX_FORMAT_BASE_SIZE_W_MASK						0xf0000000
+#define  NV34TCL_TX_WRAP(x)								(0x00001a08+((x)*32))
+#define  NV34TCL_TX_WRAP__SIZE								0x00000004
+#define   NV34TCL_TX_WRAP_S_SHIFT							0
+#define   NV34TCL_TX_WRAP_S_MASK							0x000000ff
+#define    NV34TCL_TX_WRAP_S_REPEAT							0x00000001
+#define    NV34TCL_TX_WRAP_S_MIRRORED_REPEAT						0x00000002
+#define    NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE						0x00000003
+#define    NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER						0x00000004
+#define    NV34TCL_TX_WRAP_S_CLAMP							0x00000005
+#define   NV34TCL_TX_WRAP_T_SHIFT							8
+#define   NV34TCL_TX_WRAP_T_MASK							0x00000f00
+#define    NV34TCL_TX_WRAP_T_REPEAT							0x00000100
+#define    NV34TCL_TX_WRAP_T_MIRRORED_REPEAT						0x00000200
+#define    NV34TCL_TX_WRAP_T_CLAMP_TO_EDGE						0x00000300
+#define    NV34TCL_TX_WRAP_T_CLAMP_TO_BORDER						0x00000400
+#define    NV34TCL_TX_WRAP_T_CLAMP							0x00000500
+#define   NV34TCL_TX_WRAP_EXPAND_NORMAL_SHIFT						12
+#define   NV34TCL_TX_WRAP_EXPAND_NORMAL_MASK						0x0000f000
+#define   NV34TCL_TX_WRAP_R_SHIFT							16
+#define   NV34TCL_TX_WRAP_R_MASK							0x000f0000
+#define    NV34TCL_TX_WRAP_R_REPEAT							0x00010000
+#define    NV34TCL_TX_WRAP_R_MIRRORED_REPEAT						0x00020000
+#define    NV34TCL_TX_WRAP_R_CLAMP_TO_EDGE						0x00030000
+#define    NV34TCL_TX_WRAP_R_CLAMP_TO_BORDER						0x00040000
+#define    NV34TCL_TX_WRAP_R_CLAMP							0x00050000
+#define   NV34TCL_TX_WRAP_RCOMP_SHIFT							28
+#define   NV34TCL_TX_WRAP_RCOMP_MASK							0xf0000000
+#define    NV34TCL_TX_WRAP_RCOMP_NEVER							0x00000000
+#define    NV34TCL_TX_WRAP_RCOMP_GREATER						0x10000000
+#define    NV34TCL_TX_WRAP_RCOMP_EQUAL							0x20000000
+#define    NV34TCL_TX_WRAP_RCOMP_GEQUAL							0x30000000
+#define    NV34TCL_TX_WRAP_RCOMP_LESS							0x40000000
+#define    NV34TCL_TX_WRAP_RCOMP_NOTEQUAL						0x50000000
+#define    NV34TCL_TX_WRAP_RCOMP_LEQUAL							0x60000000
+#define    NV34TCL_TX_WRAP_RCOMP_ALWAYS							0x70000000
+#define  NV34TCL_TX_ENABLE(x)								(0x00001a0c+((x)*32))
+#define  NV34TCL_TX_ENABLE__SIZE							0x00000004
+#define   NV34TCL_TX_ENABLE_ANISO_SHIFT							4
+#define   NV34TCL_TX_ENABLE_ANISO_MASK							0x00000030
+#define    NV34TCL_TX_ENABLE_ANISO_NONE							0x00000000
+#define    NV34TCL_TX_ENABLE_ANISO_2X							0x00000010
+#define    NV34TCL_TX_ENABLE_ANISO_4X							0x00000020
+#define    NV34TCL_TX_ENABLE_ANISO_8X							0x00000030
+#define   NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT					14
+#define   NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_MASK						0x0003c000
+#define   NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT					26
+#define   NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_MASK						0x3c000000
+#define   NV34TCL_TX_ENABLE_ENABLE							(1 << 30)
+#define  NV34TCL_TX_SWIZZLE(x)								(0x00001a10+((x)*32))
+#define  NV34TCL_TX_SWIZZLE__SIZE							0x00000004
+#define   NV34TCL_TX_SWIZZLE_S0_X_SHIFT							14
+#define   NV34TCL_TX_SWIZZLE_S0_X_MASK							0x0000c000
+#define    NV34TCL_TX_SWIZZLE_S0_X_ZERO							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S0_X_ONE							0x00004000
+#define    NV34TCL_TX_SWIZZLE_S0_X_S1							0x00008000
+#define   NV34TCL_TX_SWIZZLE_S0_Y_SHIFT							12
+#define   NV34TCL_TX_SWIZZLE_S0_Y_MASK							0x00003000
+#define    NV34TCL_TX_SWIZZLE_S0_Y_ZERO							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S0_Y_ONE							0x00001000
+#define    NV34TCL_TX_SWIZZLE_S0_Y_S1							0x00002000
+#define   NV34TCL_TX_SWIZZLE_S0_Z_SHIFT							10
+#define   NV34TCL_TX_SWIZZLE_S0_Z_MASK							0x00000c00
+#define    NV34TCL_TX_SWIZZLE_S0_Z_ZERO							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S0_Z_ONE							0x00000400
+#define    NV34TCL_TX_SWIZZLE_S0_Z_S1							0x00000800
+#define   NV34TCL_TX_SWIZZLE_S0_W_SHIFT							8
+#define   NV34TCL_TX_SWIZZLE_S0_W_MASK							0x00000300
+#define    NV34TCL_TX_SWIZZLE_S0_W_ZERO							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S0_W_ONE							0x00000100
+#define    NV34TCL_TX_SWIZZLE_S0_W_S1							0x00000200
+#define   NV34TCL_TX_SWIZZLE_S1_X_SHIFT							6
+#define   NV34TCL_TX_SWIZZLE_S1_X_MASK							0x000000c0
+#define    NV34TCL_TX_SWIZZLE_S1_X_W							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S1_X_Z							0x00000040
+#define    NV34TCL_TX_SWIZZLE_S1_X_Y							0x00000080
+#define    NV34TCL_TX_SWIZZLE_S1_X_X							0x000000c0
+#define   NV34TCL_TX_SWIZZLE_S1_Y_SHIFT							4
+#define   NV34TCL_TX_SWIZZLE_S1_Y_MASK							0x00000030
+#define    NV34TCL_TX_SWIZZLE_S1_Y_W							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S1_Y_Z							0x00000010
+#define    NV34TCL_TX_SWIZZLE_S1_Y_Y							0x00000020
+#define    NV34TCL_TX_SWIZZLE_S1_Y_X							0x00000030
+#define   NV34TCL_TX_SWIZZLE_S1_Z_SHIFT							2
+#define   NV34TCL_TX_SWIZZLE_S1_Z_MASK							0x0000000c
+#define    NV34TCL_TX_SWIZZLE_S1_Z_W							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S1_Z_Z							0x00000004
+#define    NV34TCL_TX_SWIZZLE_S1_Z_Y							0x00000008
+#define    NV34TCL_TX_SWIZZLE_S1_Z_X							0x0000000c
+#define   NV34TCL_TX_SWIZZLE_S1_W_SHIFT							0
+#define   NV34TCL_TX_SWIZZLE_S1_W_MASK							0x00000003
+#define    NV34TCL_TX_SWIZZLE_S1_W_W							0x00000000
+#define    NV34TCL_TX_SWIZZLE_S1_W_Z							0x00000001
+#define    NV34TCL_TX_SWIZZLE_S1_W_Y							0x00000002
+#define    NV34TCL_TX_SWIZZLE_S1_W_X							0x00000003
+#define   NV34TCL_TX_SWIZZLE_RECT_PITCH_SHIFT						16
+#define   NV34TCL_TX_SWIZZLE_RECT_PITCH_MASK						0xffff0000
+#define  NV34TCL_TX_FILTER(x)								(0x00001a14+((x)*32))
+#define  NV34TCL_TX_FILTER__SIZE							0x00000004
+#define   NV34TCL_TX_FILTER_LOD_BIAS_SHIFT						8
+#define   NV34TCL_TX_FILTER_LOD_BIAS_MASK						0x00000f00
+#define   NV34TCL_TX_FILTER_MINIFY_SHIFT						16
+#define   NV34TCL_TX_FILTER_MINIFY_MASK							0x000f0000
+#define    NV34TCL_TX_FILTER_MINIFY_NEAREST						0x00010000
+#define    NV34TCL_TX_FILTER_MINIFY_LINEAR						0x00020000
+#define    NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST				0x00030000
+#define    NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST				0x00040000
+#define    NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR				0x00050000
+#define    NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR				0x00060000
+#define   NV34TCL_TX_FILTER_MAGNIFY_SHIFT						24
+#define   NV34TCL_TX_FILTER_MAGNIFY_MASK						0x0f000000
+#define    NV34TCL_TX_FILTER_MAGNIFY_NEAREST						0x01000000
+#define    NV34TCL_TX_FILTER_MAGNIFY_LINEAR						0x02000000
+#define   NV34TCL_TX_FILTER_SIGNED_BLUE							(1 << 28)
+#define   NV34TCL_TX_FILTER_SIGNED_GREEN						(1 << 29)
+#define   NV34TCL_TX_FILTER_SIGNED_RED							(1 << 30)
+#define   NV34TCL_TX_FILTER_SIGNED_ALPHA						(1 << 31)
+#define  NV34TCL_TX_NPOT_SIZE(x)							(0x00001a18+((x)*32))
+#define  NV34TCL_TX_NPOT_SIZE__SIZE							0x00000004
+#define   NV34TCL_TX_NPOT_SIZE_H_SHIFT							0
+#define   NV34TCL_TX_NPOT_SIZE_H_MASK							0x0000ffff
+#define   NV34TCL_TX_NPOT_SIZE_W_SHIFT							16
+#define   NV34TCL_TX_NPOT_SIZE_W_MASK							0xffff0000
+#define  NV34TCL_TX_BORDER_COLOR(x)							(0x00001a1c+((x)*32))
+#define  NV34TCL_TX_BORDER_COLOR__SIZE							0x00000004
+#define   NV34TCL_TX_BORDER_COLOR_B_SHIFT						0
+#define   NV34TCL_TX_BORDER_COLOR_B_MASK						0x000000ff
+#define   NV34TCL_TX_BORDER_COLOR_G_SHIFT						8
+#define   NV34TCL_TX_BORDER_COLOR_G_MASK						0x0000ff00
+#define   NV34TCL_TX_BORDER_COLOR_R_SHIFT						16
+#define   NV34TCL_TX_BORDER_COLOR_R_MASK						0x00ff0000
+#define   NV34TCL_TX_BORDER_COLOR_A_SHIFT						24
+#define   NV34TCL_TX_BORDER_COLOR_A_MASK						0xff000000
+#define  NV34TCL_VTX_ATTR_4F_X(x)							(0x00001c00+((x)*16))
+#define  NV34TCL_VTX_ATTR_4F_X__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_4F_Y(x)							(0x00001c04+((x)*16))
+#define  NV34TCL_VTX_ATTR_4F_Y__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_4F_Z(x)							(0x00001c08+((x)*16))
+#define  NV34TCL_VTX_ATTR_4F_Z__SIZE							0x00000010
+#define  NV34TCL_VTX_ATTR_4F_W(x)							(0x00001c0c+((x)*16))
+#define  NV34TCL_VTX_ATTR_4F_W__SIZE							0x00000010
+#define  NV34TCL_FP_CONTROL								0x00001d60
+#define   NV34TCL_FP_CONTROL_USES_KIL							(1 <<  7)
+#define   NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_SHIFT				0
+#define   NV34TCL_FP_CONTROL_USED_REGS_MINUS1_DIV2_MASK					0x0000000f
+#define  NV34TCL_DEPTH_UNK17D8								0x00001d78
+#define   NV34TCL_DEPTH_UNK17D8_CLAMP_SHIFT						4
+#define   NV34TCL_DEPTH_UNK17D8_CLAMP_MASK						0x000000f0
+#define  NV34TCL_MULTISAMPLE_CONTROL							0x00001d7c
+#define   NV34TCL_MULTISAMPLE_CONTROL_ENABLE						(1 <<  0)
+#define   NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_ALPHA_TO_COVERAGE				(1 <<  4)
+#define   NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_ALPHA_TO_ONE				(1 <<  8)
+#define   NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_COVERAGE_SHIFT				16
+#define   NV34TCL_MULTISAMPLE_CONTROL_SAMPLE_COVERAGE_MASK				0xffff0000
+#define  NV34TCL_CLEAR_DEPTH_VALUE							0x00001d8c
+#define  NV34TCL_CLEAR_COLOR_VALUE							0x00001d90
+#define   NV34TCL_CLEAR_COLOR_VALUE_B_SHIFT						0
+#define   NV34TCL_CLEAR_COLOR_VALUE_B_MASK						0x000000ff
+#define   NV34TCL_CLEAR_COLOR_VALUE_G_SHIFT						8
+#define   NV34TCL_CLEAR_COLOR_VALUE_G_MASK						0x0000ff00
+#define   NV34TCL_CLEAR_COLOR_VALUE_R_SHIFT						16
+#define   NV34TCL_CLEAR_COLOR_VALUE_R_MASK						0x00ff0000
+#define   NV34TCL_CLEAR_COLOR_VALUE_A_SHIFT						24
+#define   NV34TCL_CLEAR_COLOR_VALUE_A_MASK						0xff000000
+#define  NV34TCL_CLEAR_BUFFERS								0x00001d94
+#define   NV34TCL_CLEAR_BUFFERS_COLOR_A							(1 <<  7)
+#define   NV34TCL_CLEAR_BUFFERS_COLOR_B							(1 <<  6)
+#define   NV34TCL_CLEAR_BUFFERS_COLOR_G							(1 <<  5)
+#define   NV34TCL_CLEAR_BUFFERS_COLOR_R							(1 <<  4)
+#define   NV34TCL_CLEAR_BUFFERS_STENCIL							(1 <<  1)
+#define   NV34TCL_CLEAR_BUFFERS_DEPTH							(1 <<  0)
+#define  NV34TCL_DO_VERTICES								0x00001dac
+#define  NV34TCL_LINE_STIPPLE_ENABLE							0x00001db4
+#define  NV34TCL_LINE_STIPPLE_PATTERN							0x00001db8
+#define   NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT					0
+#define   NV34TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK					0x0000ffff
+#define   NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT					16
+#define   NV34TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK					0xffff0000
+#define  NV34TCL_BACK_MATERIAL_SHININESS(x)						(0x00001e20+((x)*4))
+#define  NV34TCL_BACK_MATERIAL_SHININESS__SIZE						0x00000006
+#define  NV34TCL_VTX_ATTR_1F(x)								(0x00001e40+((x)*4))
+#define  NV34TCL_VTX_ATTR_1F__SIZE							0x00000010
+#define  NV34TCL_ENGINE									0x00001e94
+#define   NV34TCL_ENGINE_FP								(1 <<  0)
+#define   NV34TCL_ENGINE_VP								(1 <<  1)
+#define   NV34TCL_ENGINE_FIXED								(1 <<  2)
+#define  NV34TCL_VP_UPLOAD_FROM_ID							0x00001e9c
+#define  NV34TCL_VP_START_FROM_ID							0x00001ea0
+#define  NV34TCL_POINT_PARAMETERS(x)							(0x00001ec0+((x)*4))
+#define  NV34TCL_POINT_PARAMETERS__SIZE							0x00000008
+#define  NV34TCL_POINT_SIZE								0x00001ee0
+#define  NV34TCL_POINT_PARAMETERS_ENABLE						0x00001ee4
+#define  NV34TCL_POINT_SPRITE								0x00001ee8
+#define   NV34TCL_POINT_SPRITE_ENABLE							(1 <<  0)
+#define   NV34TCL_POINT_SPRITE_R_MODE_SHIFT						1
+#define   NV34TCL_POINT_SPRITE_R_MODE_MASK						0x00000006
+#define    NV34TCL_POINT_SPRITE_R_MODE_ZERO						0x00000000
+#define    NV34TCL_POINT_SPRITE_R_MODE_R						0x00000002
+#define    NV34TCL_POINT_SPRITE_R_MODE_S						0x00000004
+#define   NV34TCL_POINT_SPRITE_COORD_REPLACE						(1 << 11)
+#define  NV34TCL_VP_UPLOAD_CONST_ID							0x00001efc
+#define  NV34TCL_VP_UPLOAD_CONST_X(x)							(0x00001f00+((x)*16))
+#define  NV34TCL_VP_UPLOAD_CONST_X__SIZE						0x00000004
+#define  NV34TCL_VP_UPLOAD_CONST_Y(x)							(0x00001f04+((x)*16))
+#define  NV34TCL_VP_UPLOAD_CONST_Y__SIZE						0x00000004
+#define  NV34TCL_VP_UPLOAD_CONST_Z(x)							(0x00001f08+((x)*16))
+#define  NV34TCL_VP_UPLOAD_CONST_Z__SIZE						0x00000004
+#define  NV34TCL_VP_UPLOAD_CONST_W(x)							(0x00001f0c+((x)*16))
+#define  NV34TCL_VP_UPLOAD_CONST_W__SIZE						0x00000004
+#define  NV34TCL_UNK1f80(x)								(0x00001f80+((x)*4))
+#define  NV34TCL_UNK1f80__SIZE								0x00000010
+
+
+#define NV40_CONTEXT_SURFACES_2D							0x00003062
+
+
+
+#define NV40_STRETCHED_IMAGE_FROM_CPU							0x00003066
+
+
+
+#define NV40_TEXTURE_FROM_CPU								0x0000307b
+
+
+
+#define NV40_SCALED_IMAGE_FROM_MEMORY							0x00003089
+
+
+
+#define NV40_IMAGE_FROM_CPU								0x0000308a
+
+
+
+#define NV40_SWIZZLED_SURFACE								0x0000309e
+
+
+
+#define NV40TCL										0x00004097
+
+#define  NV40TCL_REF_CNT								0x00000050
+#define  NV40TCL_NOP									0x00000100
+#define  NV40TCL_NOTIFY									0x00000104
+#define  NV40TCL_DMA_NOTIFY								0x00000180
+#define  NV40TCL_DMA_TEXTURE0								0x00000184
+#define  NV40TCL_DMA_TEXTURE1								0x00000188
+#define  NV40TCL_DMA_COLOR1								0x0000018c
+#define  NV40TCL_DMA_COLOR0								0x00000194
+#define  NV40TCL_DMA_ZETA								0x00000198
+#define  NV40TCL_DMA_VTXBUF0								0x0000019c
+#define  NV40TCL_DMA_VTXBUF1								0x000001a0
+#define  NV40TCL_DMA_FENCE								0x000001a4
+#define  NV40TCL_DMA_QUERY								0x000001a8
+#define  NV40TCL_DMA_UNK01AC								0x000001ac
+#define  NV40TCL_DMA_UNK01B0								0x000001b0
+#define  NV40TCL_DMA_COLOR2								0x000001b4
+#define  NV40TCL_DMA_COLOR3								0x000001b8
+#define  NV40TCL_RT_HORIZ								0x00000200
+#define   NV40TCL_RT_HORIZ_W_SHIFT							16
+#define   NV40TCL_RT_HORIZ_W_MASK							0xffff0000
+#define   NV40TCL_RT_HORIZ_X_SHIFT							0
+#define   NV40TCL_RT_HORIZ_X_MASK							0x0000ffff
+#define  NV40TCL_RT_VERT								0x00000204
+#define   NV40TCL_RT_VERT_H_SHIFT							16
+#define   NV40TCL_RT_VERT_H_MASK							0xffff0000
+#define   NV40TCL_RT_VERT_Y_SHIFT							0
+#define   NV40TCL_RT_VERT_Y_MASK							0x0000ffff
+#define  NV40TCL_RT_FORMAT								0x00000208
+#define   NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT						24
+#define   NV40TCL_RT_FORMAT_LOG2_HEIGHT_MASK						0xff000000
+#define   NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT						16
+#define   NV40TCL_RT_FORMAT_LOG2_WIDTH_MASK						0x00ff0000
+#define   NV40TCL_RT_FORMAT_TYPE_SHIFT							8
+#define   NV40TCL_RT_FORMAT_TYPE_MASK							0x00000f00
+#define    NV40TCL_RT_FORMAT_TYPE_LINEAR						0x00000100
+#define    NV40TCL_RT_FORMAT_TYPE_SWIZZLED						0x00000200
+#define   NV40TCL_RT_FORMAT_ZETA_SHIFT							5
+#define   NV40TCL_RT_FORMAT_ZETA_MASK							0x000000e0
+#define    NV40TCL_RT_FORMAT_ZETA_Z16							0x00000020
+#define    NV40TCL_RT_FORMAT_ZETA_Z24S8							0x00000040
+#define   NV40TCL_RT_FORMAT_COLOR_SHIFT							0
+#define   NV40TCL_RT_FORMAT_COLOR_MASK							0x0000001f
+#define    NV40TCL_RT_FORMAT_COLOR_R5G6B5						0x00000003
+#define    NV40TCL_RT_FORMAT_COLOR_X8R8G8B8						0x00000005
+#define    NV40TCL_RT_FORMAT_COLOR_A8R8G8B8						0x00000008
+#define    NV40TCL_RT_FORMAT_COLOR_B8							0x00000009
+#define    NV40TCL_RT_FORMAT_COLOR_UNKNOWN						0x0000000d
+#define    NV40TCL_RT_FORMAT_COLOR_X8B8G8R8						0x0000000f
+#define    NV40TCL_RT_FORMAT_COLOR_A8B8G8R8						0x00000010
+#define  NV40TCL_COLOR0_PITCH								0x0000020c
+#define  NV40TCL_COLOR0_OFFSET								0x00000210
+#define  NV40TCL_ZETA_OFFSET								0x00000214
+#define  NV40TCL_COLOR1_OFFSET								0x00000218
+#define  NV40TCL_COLOR1_PITCH								0x0000021c
+#define  NV40TCL_RT_ENABLE								0x00000220
+#define   NV40TCL_RT_ENABLE_MRT								(1 <<  4)
+#define   NV40TCL_RT_ENABLE_COLOR3							(1 <<  3)
+#define   NV40TCL_RT_ENABLE_COLOR2							(1 <<  2)
+#define   NV40TCL_RT_ENABLE_COLOR1							(1 <<  1)
+#define   NV40TCL_RT_ENABLE_COLOR0							(1 <<  0)
+#define  NV40TCL_ZETA_PITCH								0x0000022c
+#define  NV40TCL_COLOR2_PITCH								0x00000280
+#define  NV40TCL_COLOR3_PITCH								0x00000284
+#define  NV40TCL_COLOR2_OFFSET								0x00000288
+#define  NV40TCL_COLOR3_OFFSET								0x0000028c
+#define  NV40TCL_VIEWPORT_CLIP_HORIZ(x)							(0x000002c0+((x)*8))
+#define  NV40TCL_VIEWPORT_CLIP_HORIZ__SIZE						0x00000008
+#define  NV40TCL_VIEWPORT_CLIP_VERT(x)							(0x000002c4+((x)*8))
+#define  NV40TCL_VIEWPORT_CLIP_VERT__SIZE						0x00000008
+#define  NV40TCL_DITHER_ENABLE								0x00000300
+#define  NV40TCL_ALPHA_TEST_ENABLE							0x00000304
+#define  NV40TCL_ALPHA_TEST_FUNC							0x00000308
+#define   NV40TCL_ALPHA_TEST_FUNC_NEVER							0x00000200
+#define   NV40TCL_ALPHA_TEST_FUNC_LESS							0x00000201
+#define   NV40TCL_ALPHA_TEST_FUNC_EQUAL							0x00000202
+#define   NV40TCL_ALPHA_TEST_FUNC_LEQUAL						0x00000203
+#define   NV40TCL_ALPHA_TEST_FUNC_GREATER						0x00000204
+#define   NV40TCL_ALPHA_TEST_FUNC_GREATER						0x00000204
+#define   NV40TCL_ALPHA_TEST_FUNC_NOTEQUAL						0x00000205
+#define   NV40TCL_ALPHA_TEST_FUNC_GEQUAL						0x00000206
+#define   NV40TCL_ALPHA_TEST_FUNC_ALWAYS						0x00000207
+#define  NV40TCL_ALPHA_TEST_REF								0x0000030c
+#define  NV40TCL_BLEND_ENABLE								0x00000310
+#define  NV40TCL_BLEND_FUNC_SRC								0x00000314
+#define   NV40TCL_BLEND_FUNC_SRC_RGB_SHIFT						0
+#define   NV40TCL_BLEND_FUNC_SRC_RGB_MASK						0x0000ffff
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ZERO						0x00000000
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE						0x00000001
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR						0x00000300
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA						0x00000302
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA						0x00000304
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_DST_COLOR						0x00000306
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE				0x00000308
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR					0x00008001
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA					0x00008003
+#define    NV40TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define   NV40TCL_BLEND_FUNC_SRC_ALPHA_SHIFT						16
+#define   NV40TCL_BLEND_FUNC_SRC_ALPHA_MASK						0xffff0000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ZERO						0x00000000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE						0x00010000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR					0x03000000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR				0x03010000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA					0x03020000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA				0x03030000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA					0x03040000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA				0x03050000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR					0x03060000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR				0x03070000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE				0x03080000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR					0x80010000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR			0x80020000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA					0x80030000
+#define    NV40TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA			0x80040000
+#define  NV40TCL_BLEND_FUNC_DST								0x00000318
+#define   NV40TCL_BLEND_FUNC_DST_RGB_SHIFT						0
+#define   NV40TCL_BLEND_FUNC_DST_RGB_MASK						0x0000ffff
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ZERO						0x00000000
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE						0x00000001
+#define    NV40TCL_BLEND_FUNC_DST_RGB_SRC_COLOR						0x00000300
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define    NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA						0x00000302
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define    NV40TCL_BLEND_FUNC_DST_RGB_DST_ALPHA						0x00000304
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define    NV40TCL_BLEND_FUNC_DST_RGB_DST_COLOR						0x00000306
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define    NV40TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE				0x00000308
+#define    NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR					0x00008001
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define    NV40TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA					0x00008003
+#define    NV40TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define   NV40TCL_BLEND_FUNC_DST_ALPHA_SHIFT						16
+#define   NV40TCL_BLEND_FUNC_DST_ALPHA_MASK						0xffff0000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ZERO						0x00000000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE						0x00010000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR					0x03000000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR				0x03010000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA					0x03020000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA				0x03030000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA					0x03040000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA				0x03050000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR					0x03060000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR				0x03070000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE				0x03080000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR					0x80010000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR			0x80020000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA					0x80030000
+#define    NV40TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA			0x80040000
+#define  NV40TCL_BLEND_COLOR								0x0000031c
+#define  NV40TCL_BLEND_EQUATION								0x00000320
+#define   NV40TCL_BLEND_EQUATION_RGB_SHIFT						0
+#define   NV40TCL_BLEND_EQUATION_RGB_MASK						0x0000ffff
+#define    NV40TCL_BLEND_EQUATION_RGB_FUNC_ADD						0x00008006
+#define    NV40TCL_BLEND_EQUATION_RGB_MIN						0x00008007
+#define    NV40TCL_BLEND_EQUATION_RGB_MAX						0x00008008
+#define    NV40TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT					0x0000800a
+#define    NV40TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT				0x0000800b
+#define   NV40TCL_BLEND_EQUATION_ALPHA_SHIFT						16
+#define   NV40TCL_BLEND_EQUATION_ALPHA_MASK						0xffff0000
+#define    NV40TCL_BLEND_EQUATION_ALPHA_FUNC_ADD					0x80060000
+#define    NV40TCL_BLEND_EQUATION_ALPHA_MIN						0x80070000
+#define    NV40TCL_BLEND_EQUATION_ALPHA_MAX						0x80080000
+#define    NV40TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT					0x800a0000
+#define    NV40TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT				0x800b0000
+#define  NV40TCL_COLOR_MASK								0x00000324
+#define   NV40TCL_COLOR_MASK_BUFFER0_B_SHIFT						0
+#define   NV40TCL_COLOR_MASK_BUFFER0_B_MASK						0x000000ff
+#define   NV40TCL_COLOR_MASK_BUFFER0_G_SHIFT						8
+#define   NV40TCL_COLOR_MASK_BUFFER0_G_MASK						0x0000ff00
+#define   NV40TCL_COLOR_MASK_BUFFER0_R_SHIFT						16
+#define   NV40TCL_COLOR_MASK_BUFFER0_R_MASK						0x00ff0000
+#define   NV40TCL_COLOR_MASK_BUFFER0_A_SHIFT						24
+#define   NV40TCL_COLOR_MASK_BUFFER0_A_MASK						0xff000000
+#define  NV40TCL_STENCIL_FRONT_ENABLE							0x00000328
+#define  NV40TCL_STENCIL_FRONT_MASK							0x0000032c
+#define  NV40TCL_STENCIL_FRONT_FUNC_FUNC						0x00000330
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_NEVER						0x00000200
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_LESS						0x00000201
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL						0x00000202
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL					0x00000203
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL					0x00000206
+#define   NV40TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS					0x00000207
+#define  NV40TCL_STENCIL_FRONT_FUNC_REF							0x00000334
+#define  NV40TCL_STENCIL_FRONT_FUNC_MASK						0x00000338
+#define  NV40TCL_STENCIL_FRONT_OP_FAIL							0x0000033c
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_ZERO						0x00000000
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_REPLACE						0x00001e01
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_INCR						0x00001e02
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_DECR						0x00001e03
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV40TCL_STENCIL_FRONT_OP_ZFAIL							0x00000340
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_ZERO						0x00000000
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE					0x00001e01
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR						0x00001e02
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR						0x00001e03
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV40TCL_STENCIL_FRONT_OP_ZPASS							0x00000344
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_ZERO						0x00000000
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_REPLACE					0x00001e01
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR						0x00001e02
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR						0x00001e03
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV40TCL_STENCIL_BACK_ENABLE							0x00000348
+#define  NV40TCL_STENCIL_BACK_MASK							0x0000034c
+#define  NV40TCL_STENCIL_BACK_FUNC_FUNC							0x00000350
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_NEVER						0x00000200
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_LESS						0x00000201
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_EQUAL						0x00000202
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV40TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV40TCL_STENCIL_BACK_FUNC_REF							0x00000354
+#define  NV40TCL_STENCIL_BACK_FUNC_MASK							0x00000358
+#define  NV40TCL_STENCIL_BACK_OP_FAIL							0x0000035c
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_ZERO						0x00000000
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_REPLACE						0x00001e01
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_INCR						0x00001e02
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_DECR						0x00001e03
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV40TCL_STENCIL_BACK_OP_ZFAIL							0x00000360
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_ZERO						0x00000000
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_REPLACE						0x00001e01
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR						0x00001e02
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR						0x00001e03
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV40TCL_STENCIL_BACK_OP_ZPASS							0x00000364
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_ZERO						0x00000000
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_INVERT						0x0000150a
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_KEEP						0x00001e00
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_REPLACE						0x00001e01
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_INCR						0x00001e02
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_DECR						0x00001e03
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV40TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV40TCL_SHADE_MODEL								0x00000368
+#define   NV40TCL_SHADE_MODEL_FLAT							0x00001d00
+#define   NV40TCL_SHADE_MODEL_SMOOTH							0x00001d01
+#define  NV40TCL_MRT_COLOR_MASK								0x00000370
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER1_A						(1 <<  4)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER1_R						(1 <<  5)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER1_G						(1 <<  6)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER1_B						(1 <<  7)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER2_A						(1 <<  8)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER2_R						(1 <<  9)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER2_G						(1 << 10)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER2_B						(1 << 11)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER3_A						(1 << 12)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER3_R						(1 << 13)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER3_G						(1 << 14)
+#define   NV40TCL_MRT_COLOR_MASK_BUFFER3_B						(1 << 15)
+#define  NV40TCL_COLOR_LOGIC_OP_ENABLE							0x00000374
+#define  NV40TCL_COLOR_LOGIC_OP								0x00000378
+#define   NV40TCL_COLOR_LOGIC_OP_CLEAR							0x00001500
+#define   NV40TCL_COLOR_LOGIC_OP_AND							0x00001501
+#define   NV40TCL_COLOR_LOGIC_OP_AND_REVERSE						0x00001502
+#define   NV40TCL_COLOR_LOGIC_OP_COPY							0x00001503
+#define   NV40TCL_COLOR_LOGIC_OP_AND_INVERTED						0x00001504
+#define   NV40TCL_COLOR_LOGIC_OP_NOOP							0x00001505
+#define   NV40TCL_COLOR_LOGIC_OP_XOR							0x00001506
+#define   NV40TCL_COLOR_LOGIC_OP_OR							0x00001507
+#define   NV40TCL_COLOR_LOGIC_OP_NOR							0x00001508
+#define   NV40TCL_COLOR_LOGIC_OP_EQUIV							0x00001509
+#define   NV40TCL_COLOR_LOGIC_OP_INVERT							0x0000150a
+#define   NV40TCL_COLOR_LOGIC_OP_OR_REVERSE						0x0000150b
+#define   NV40TCL_COLOR_LOGIC_OP_COPY_INVERTED						0x0000150c
+#define   NV40TCL_COLOR_LOGIC_OP_OR_INVERTED						0x0000150d
+#define   NV40TCL_COLOR_LOGIC_OP_NAND							0x0000150e
+#define   NV40TCL_COLOR_LOGIC_OP_SET							0x0000150f
+#define  NV40TCL_DEPTH_RANGE_NEAR							0x00000394
+#define  NV40TCL_DEPTH_RANGE_FAR							0x00000398
+#define  NV40TCL_LINE_WIDTH								0x000003b8
+#define  NV40TCL_LINE_SMOOTH_ENABLE							0x000003bc
+#define  NV40TCL_UNK03C0(x)								(0x000003c0+((x)*4))
+#define  NV40TCL_UNK03C0__SIZE								0x00000010
+#define  NV40TCL_UNK0400(x)								(0x00000400+((x)*4))
+#define  NV40TCL_UNK0400__SIZE								0x00000010
+#define  NV40TCL_UNK0440(x)								(0x00000440+((x)*4))
+#define  NV40TCL_UNK0440__SIZE								0x00000020
+#define  NV40TCL_SCISSOR_HORIZ								0x000008c0
+#define   NV40TCL_SCISSOR_HORIZ_X_SHIFT							0
+#define   NV40TCL_SCISSOR_HORIZ_X_MASK							0x0000ffff
+#define   NV40TCL_SCISSOR_HORIZ_W_SHIFT							16
+#define   NV40TCL_SCISSOR_HORIZ_W_MASK							0xffff0000
+#define  NV40TCL_SCISSOR_VERT								0x000008c4
+#define   NV40TCL_SCISSOR_VERT_Y_SHIFT							0
+#define   NV40TCL_SCISSOR_VERT_Y_MASK							0x0000ffff
+#define   NV40TCL_SCISSOR_VERT_H_SHIFT							16
+#define   NV40TCL_SCISSOR_VERT_H_MASK							0xffff0000
+#define  NV40TCL_FOG_MODE								0x000008cc
+#define  NV40TCL_FOG_EQUATION_CONSTANT							0x000008d0
+#define  NV40TCL_FOG_EQUATION_LINEAR							0x000008d4
+#define  NV40TCL_FOG_EQUATION_QUADRATIC							0x000008d8
+#define  NV40TCL_FP_ADDRESS								0x000008e4
+#define   NV40TCL_FP_ADDRESS_OFFSET_SHIFT						8
+#define   NV40TCL_FP_ADDRESS_OFFSET_MASK						0xffffff00
+#define   NV40TCL_FP_ADDRESS_DMA1							(1 <<  1)
+#define   NV40TCL_FP_ADDRESS_DMA0							(1 <<  0)
+#define  NV40TCL_VIEWPORT_HORIZ								0x00000a00
+#define   NV40TCL_VIEWPORT_HORIZ_W_SHIFT						16
+#define   NV40TCL_VIEWPORT_HORIZ_W_MASK							0xffff0000
+#define   NV40TCL_VIEWPORT_HORIZ_X_SHIFT						0
+#define   NV40TCL_VIEWPORT_HORIZ_X_MASK							0x0000ffff
+#define  NV40TCL_VIEWPORT_VERT								0x00000a04
+#define   NV40TCL_VIEWPORT_VERT_H_SHIFT							16
+#define   NV40TCL_VIEWPORT_VERT_H_MASK							0xffff0000
+#define   NV40TCL_VIEWPORT_VERT_Y_SHIFT							0
+#define   NV40TCL_VIEWPORT_VERT_Y_MASK							0x0000ffff
+#define  NV40TCL_VIEWPORT_TRANSLATE_X							0x00000a20
+#define  NV40TCL_VIEWPORT_TRANSLATE_Y							0x00000a24
+#define  NV40TCL_VIEWPORT_TRANSLATE_Z							0x00000a28
+#define  NV40TCL_VIEWPORT_TRANSLATE_W							0x00000a2c
+#define  NV40TCL_VIEWPORT_SCALE_X							0x00000a30
+#define  NV40TCL_VIEWPORT_SCALE_Y							0x00000a34
+#define  NV40TCL_VIEWPORT_SCALE_Z							0x00000a38
+#define  NV40TCL_VIEWPORT_SCALE_W							0x00000a3c
+#define  NV40TCL_POLYGON_OFFSET_POINT_ENABLE						0x00000a60
+#define  NV40TCL_POLYGON_OFFSET_LINE_ENABLE						0x00000a64
+#define  NV40TCL_POLYGON_OFFSET_FILL_ENABLE						0x00000a68
+#define  NV40TCL_DEPTH_FUNC								0x00000a6c
+#define   NV40TCL_DEPTH_FUNC_NEVER							0x00000200
+#define   NV40TCL_DEPTH_FUNC_LESS							0x00000201
+#define   NV40TCL_DEPTH_FUNC_EQUAL							0x00000202
+#define   NV40TCL_DEPTH_FUNC_LEQUAL							0x00000203
+#define   NV40TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV40TCL_DEPTH_FUNC_GREATER							0x00000204
+#define   NV40TCL_DEPTH_FUNC_NOTEQUAL							0x00000205
+#define   NV40TCL_DEPTH_FUNC_GEQUAL							0x00000206
+#define   NV40TCL_DEPTH_FUNC_ALWAYS							0x00000207
+#define  NV40TCL_DEPTH_WRITE_ENABLE							0x00000a70
+#define  NV40TCL_DEPTH_TEST_ENABLE							0x00000a74
+#define  NV40TCL_POLYGON_OFFSET_FACTOR							0x00000a78
+#define  NV40TCL_POLYGON_OFFSET_UNITS							0x00000a7c
+#define  NV40TCL_VTX_ATTR_3I_XY(x)							(0x00000a80+((x)*8))
+#define  NV40TCL_VTX_ATTR_3I_XY__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_3I_XY_X_SHIFT						0
+#define   NV40TCL_VTX_ATTR_3I_XY_X_MASK							0x0000ffff
+#define   NV40TCL_VTX_ATTR_3I_XY_Y_SHIFT						16
+#define   NV40TCL_VTX_ATTR_3I_XY_Y_MASK							0xffff0000
+#define  NV40TCL_VTX_ATTR_3I_Z(x)							(0x00000a84+((x)*8))
+#define  NV40TCL_VTX_ATTR_3I_Z__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_3I_Z_Z_SHIFT							0
+#define   NV40TCL_VTX_ATTR_3I_Z_Z_MASK							0x0000ffff
+#define  NV40TCL_UNK0B40(x)								(0x00000b40+((x)*4))
+#define  NV40TCL_UNK0B40__SIZE								0x00000008
+#define  NV40TCL_VP_UPLOAD_INST(x)							(0x00000b80+((x)*4))
+#define  NV40TCL_VP_UPLOAD_INST__SIZE							0x00000004
+#define  NV40TCL_CLIP_PLANE_ENABLE							0x00001478
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE0						(1 <<  1)
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE1						(1 <<  5)
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE2						(1 <<  9)
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE3						(1 << 13)
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE4						(1 << 17)
+#define   NV40TCL_CLIP_PLANE_ENABLE_PLANE5						(1 << 21)
+#define  NV40TCL_POLYGON_STIPPLE_ENABLE							0x0000147c
+#define  NV40TCL_POLYGON_STIPPLE_PATTERN(x)						(0x00001480+((x)*4))
+#define  NV40TCL_POLYGON_STIPPLE_PATTERN__SIZE						0x00000020
+#define  NV40TCL_VTX_ATTR_3F_X(x)							(0x00001500+((x)*16))
+#define  NV40TCL_VTX_ATTR_3F_X__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_3F_Y(x)							(0x00001504+((x)*16))
+#define  NV40TCL_VTX_ATTR_3F_Y__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_3F_Z(x)							(0x00001508+((x)*16))
+#define  NV40TCL_VTX_ATTR_3F_Z__SIZE							0x00000010
+#define  NV40TCL_VTXBUF_ADDRESS(x)							(0x00001680+((x)*4))
+#define  NV40TCL_VTXBUF_ADDRESS__SIZE							0x00000010
+#define   NV40TCL_VTXBUF_ADDRESS_DMA1							(1 << 31)
+#define   NV40TCL_VTXBUF_ADDRESS_OFFSET_SHIFT						0
+#define   NV40TCL_VTXBUF_ADDRESS_OFFSET_MASK						0x0fffffff
+#define  NV40TCL_VTX_CACHE_INVALIDATE							0x00001714
+#define  NV40TCL_VTXFMT(x)								(0x00001740+((x)*4))
+#define  NV40TCL_VTXFMT__SIZE								0x00000010
+#define   NV40TCL_VTXFMT_TYPE_SHIFT							0
+#define   NV40TCL_VTXFMT_TYPE_MASK							0x0000000f
+#define    NV40TCL_VTXFMT_TYPE_FLOAT							0x00000002
+#define    NV40TCL_VTXFMT_TYPE_UBYTE							0x00000004
+#define    NV40TCL_VTXFMT_TYPE_USHORT							0x00000005
+#define   NV40TCL_VTXFMT_SIZE_SHIFT							4
+#define   NV40TCL_VTXFMT_SIZE_MASK							0x000000f0
+#define   NV40TCL_VTXFMT_STRIDE_SHIFT							8
+#define   NV40TCL_VTXFMT_STRIDE_MASK							0x0000ff00
+#define  NV40TCL_QUERY_RESET								0x000017c8
+#define  NV40TCL_QUERY_UNK17CC								0x000017cc
+#define  NV40TCL_QUERY_GET								0x00001800
+#define   NV40TCL_QUERY_GET_UNK24_SHIFT							24
+#define   NV40TCL_QUERY_GET_UNK24_MASK							0xff000000
+#define   NV40TCL_QUERY_GET_OFFSET_SHIFT						0
+#define   NV40TCL_QUERY_GET_OFFSET_MASK							0x00ffffff
+#define  NV40TCL_BEGIN_END								0x00001808
+#define   NV40TCL_BEGIN_END_STOP							0x00000000
+#define   NV40TCL_BEGIN_END_POINTS							0x00000001
+#define   NV40TCL_BEGIN_END_LINES							0x00000002
+#define   NV40TCL_BEGIN_END_LINE_LOOP							0x00000003
+#define   NV40TCL_BEGIN_END_LINE_STRIP							0x00000004
+#define   NV40TCL_BEGIN_END_TRIANGLES							0x00000005
+#define   NV40TCL_BEGIN_END_TRIANGLE_STRIP						0x00000006
+#define   NV40TCL_BEGIN_END_TRIANGLE_FAN						0x00000007
+#define   NV40TCL_BEGIN_END_QUADS							0x00000008
+#define   NV40TCL_BEGIN_END_QUAD_STRIP							0x00000009
+#define   NV40TCL_BEGIN_END_POLYGON							0x0000000a
+#define  NV40TCL_VB_ELEMENT_U16								0x0000180c
+#define   NV40TCL_VB_ELEMENT_U16_1_SHIFT						16
+#define   NV40TCL_VB_ELEMENT_U16_1_MASK							0xffff0000
+#define   NV40TCL_VB_ELEMENT_U16_0_SHIFT						0
+#define   NV40TCL_VB_ELEMENT_U16_0_MASK							0x0000ffff
+#define  NV40TCL_VB_ELEMENT_U32								0x00001810
+#define  NV40TCL_VB_VERTEX_BATCH							0x00001814
+#define   NV40TCL_VB_VERTEX_BATCH_COUNT_SHIFT						24
+#define   NV40TCL_VB_VERTEX_BATCH_COUNT_MASK						0xff000000
+#define   NV40TCL_VB_VERTEX_BATCH_START_SHIFT						0
+#define   NV40TCL_VB_VERTEX_BATCH_START_MASK						0x00ffffff
+#define  NV40TCL_VERTEX_DATA								0x00001818
+#define  NV40TCL_IDXBUF_ADDRESS								0x0000181c
+#define  NV40TCL_IDXBUF_FORMAT								0x00001820
+#define   NV40TCL_IDXBUF_FORMAT_TYPE_SHIFT						4
+#define   NV40TCL_IDXBUF_FORMAT_TYPE_MASK						0x000000f0
+#define    NV40TCL_IDXBUF_FORMAT_TYPE_U32						0x00000000
+#define    NV40TCL_IDXBUF_FORMAT_TYPE_U16						0x00000010
+#define   NV40TCL_IDXBUF_FORMAT_DMA1							(1 <<  0)
+#define  NV40TCL_VB_INDEX_BATCH								0x00001824
+#define   NV40TCL_VB_INDEX_BATCH_COUNT_SHIFT						24
+#define   NV40TCL_VB_INDEX_BATCH_COUNT_MASK						0xff000000
+#define   NV40TCL_VB_INDEX_BATCH_START_SHIFT						0
+#define   NV40TCL_VB_INDEX_BATCH_START_MASK						0x00ffffff
+#define  NV40TCL_POLYGON_MODE_FRONT							0x00001828
+#define   NV40TCL_POLYGON_MODE_FRONT_POINT						0x00001b00
+#define   NV40TCL_POLYGON_MODE_FRONT_LINE						0x00001b01
+#define   NV40TCL_POLYGON_MODE_FRONT_FILL						0x00001b02
+#define  NV40TCL_POLYGON_MODE_BACK							0x0000182c
+#define   NV40TCL_POLYGON_MODE_BACK_POINT						0x00001b00
+#define   NV40TCL_POLYGON_MODE_BACK_LINE						0x00001b01
+#define   NV40TCL_POLYGON_MODE_BACK_FILL						0x00001b02
+#define  NV40TCL_CULL_FACE								0x00001830
+#define   NV40TCL_CULL_FACE_FRONT							0x00000404
+#define   NV40TCL_CULL_FACE_BACK							0x00000405
+#define   NV40TCL_CULL_FACE_FRONT_AND_BACK						0x00000408
+#define  NV40TCL_FRONT_FACE								0x00001834
+#define   NV40TCL_FRONT_FACE_CW								0x00000900
+#define   NV40TCL_FRONT_FACE_CCW							0x00000901
+#define  NV40TCL_POLYGON_SMOOTH_ENABLE							0x00001838
+#define  NV40TCL_CULL_FACE_ENABLE							0x0000183c
+#define  NV40TCL_TEX_SIZE1(x)								(0x00001840+((x)*4))
+#define  NV40TCL_TEX_SIZE1__SIZE							0x00000008
+#define   NV40TCL_TEX_SIZE1_DEPTH_SHIFT							20
+#define   NV40TCL_TEX_SIZE1_DEPTH_MASK							0xfff00000
+#define   NV40TCL_TEX_SIZE1_PITCH_SHIFT							0
+#define   NV40TCL_TEX_SIZE1_PITCH_MASK							0x0000ffff
+#define  NV40TCL_VTX_ATTR_2F_X(x)							(0x00001880+((x)*8))
+#define  NV40TCL_VTX_ATTR_2F_X__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_2F_Y(x)							(0x00001884+((x)*8))
+#define  NV40TCL_VTX_ATTR_2F_Y__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_2I(x)								(0x00001900+((x)*4))
+#define  NV40TCL_VTX_ATTR_2I__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_2I_X_SHIFT							0
+#define   NV40TCL_VTX_ATTR_2I_X_MASK							0x0000ffff
+#define   NV40TCL_VTX_ATTR_2I_Y_SHIFT							16
+#define   NV40TCL_VTX_ATTR_2I_Y_MASK							0xffff0000
+#define  NV40TCL_VTX_ATTR_4UB(x)							(0x00001940+((x)*4))
+#define  NV40TCL_VTX_ATTR_4UB__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_4UB_X_SHIFT							0
+#define   NV40TCL_VTX_ATTR_4UB_X_MASK							0x000000ff
+#define   NV40TCL_VTX_ATTR_4UB_Y_SHIFT							8
+#define   NV40TCL_VTX_ATTR_4UB_Y_MASK							0x0000ff00
+#define   NV40TCL_VTX_ATTR_4UB_Z_SHIFT							16
+#define   NV40TCL_VTX_ATTR_4UB_Z_MASK							0x00ff0000
+#define   NV40TCL_VTX_ATTR_4UB_W_SHIFT							24
+#define   NV40TCL_VTX_ATTR_4UB_W_MASK							0xff000000
+#define  NV40TCL_VTX_ATTR_4I_XY(x)							(0x00001980+((x)*8))
+#define  NV40TCL_VTX_ATTR_4I_XY__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_4I_XY_X_SHIFT						0
+#define   NV40TCL_VTX_ATTR_4I_XY_X_MASK							0x0000ffff
+#define   NV40TCL_VTX_ATTR_4I_XY_Y_SHIFT						16
+#define   NV40TCL_VTX_ATTR_4I_XY_Y_MASK							0xffff0000
+#define  NV40TCL_VTX_ATTR_4I_ZW(x)							(0x00001984+((x)*8))
+#define  NV40TCL_VTX_ATTR_4I_ZW__SIZE							0x00000010
+#define   NV40TCL_VTX_ATTR_4I_ZW_Z_SHIFT						0
+#define   NV40TCL_VTX_ATTR_4I_ZW_Z_MASK							0x0000ffff
+#define   NV40TCL_VTX_ATTR_4I_ZW_W_SHIFT						16
+#define   NV40TCL_VTX_ATTR_4I_ZW_W_MASK							0xffff0000
+#define  NV40TCL_TEX_OFFSET(x)								(0x00001a00+((x)*32))
+#define  NV40TCL_TEX_OFFSET__SIZE							0x00000010
+#define  NV40TCL_TEX_FORMAT(x)								(0x00001a04+((x)*32))
+#define  NV40TCL_TEX_FORMAT__SIZE							0x00000010
+#define   NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT						16
+#define   NV40TCL_TEX_FORMAT_MIPMAP_COUNT_MASK						0x000f0000
+#define   NV40TCL_TEX_FORMAT_RECT							(1 << 14)
+#define   NV40TCL_TEX_FORMAT_LINEAR							(1 << 13)
+#define   NV40TCL_TEX_FORMAT_FORMAT_SHIFT						8
+#define   NV40TCL_TEX_FORMAT_FORMAT_MASK						0x00001f00
+#define    NV40TCL_TEX_FORMAT_FORMAT_L8							0x00000100
+#define    NV40TCL_TEX_FORMAT_FORMAT_A1R5G5B5						0x00000200
+#define    NV40TCL_TEX_FORMAT_FORMAT_A4R4G4B4						0x00000300
+#define    NV40TCL_TEX_FORMAT_FORMAT_R5G6B5						0x00000400
+#define    NV40TCL_TEX_FORMAT_FORMAT_A8R8G8B8						0x00000500
+#define    NV40TCL_TEX_FORMAT_FORMAT_DXT1						0x00000600
+#define    NV40TCL_TEX_FORMAT_FORMAT_DXT3						0x00000700
+#define    NV40TCL_TEX_FORMAT_FORMAT_DXT5						0x00000800
+#define    NV40TCL_TEX_FORMAT_FORMAT_A8L8						0x00000b00
+#define    NV40TCL_TEX_FORMAT_FORMAT_Z24						0x00001000
+#define    NV40TCL_TEX_FORMAT_FORMAT_Z16						0x00001200
+#define    NV40TCL_TEX_FORMAT_FORMAT_A16						0x00001400
+#define    NV40TCL_TEX_FORMAT_FORMAT_A16L16						0x00001500
+#define    NV40TCL_TEX_FORMAT_FORMAT_HILO8						0x00001800
+#define    NV40TCL_TEX_FORMAT_FORMAT_RGBA16F						0x00001a00
+#define    NV40TCL_TEX_FORMAT_FORMAT_RGBA32F						0x00001b00
+#define   NV40TCL_TEX_FORMAT_DIMS_SHIFT							4
+#define   NV40TCL_TEX_FORMAT_DIMS_MASK							0x000000f0
+#define    NV40TCL_TEX_FORMAT_DIMS_1D							0x00000010
+#define    NV40TCL_TEX_FORMAT_DIMS_2D							0x00000020
+#define    NV40TCL_TEX_FORMAT_DIMS_3D							0x00000030
+#define   NV40TCL_TEX_FORMAT_NO_BORDER							(1 <<  3)
+#define   NV40TCL_TEX_FORMAT_CUBIC							(1 <<  2)
+#define   NV40TCL_TEX_FORMAT_DMA1							(1 <<  1)
+#define   NV40TCL_TEX_FORMAT_DMA0							(1 <<  0)
+#define  NV40TCL_TEX_WRAP(x)								(0x00001a08+((x)*32))
+#define  NV40TCL_TEX_WRAP__SIZE								0x00000010
+#define   NV40TCL_TEX_WRAP_S_SHIFT							0
+#define   NV40TCL_TEX_WRAP_S_MASK							0x000000ff
+#define    NV40TCL_TEX_WRAP_S_REPEAT							0x00000001
+#define    NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT						0x00000002
+#define    NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE						0x00000003
+#define    NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER						0x00000004
+#define    NV40TCL_TEX_WRAP_S_CLAMP							0x00000005
+#define    NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE					0x00000006
+#define    NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER					0x00000007
+#define    NV40TCL_TEX_WRAP_S_MIRROR_CLAMP						0x00000008
+#define   NV40TCL_TEX_WRAP_T_SHIFT							8
+#define   NV40TCL_TEX_WRAP_T_MASK							0x00000f00
+#define    NV40TCL_TEX_WRAP_T_REPEAT							0x00000100
+#define    NV40TCL_TEX_WRAP_T_MIRRORED_REPEAT						0x00000200
+#define    NV40TCL_TEX_WRAP_T_CLAMP_TO_EDGE						0x00000300
+#define    NV40TCL_TEX_WRAP_T_CLAMP_TO_BORDER						0x00000400
+#define    NV40TCL_TEX_WRAP_T_CLAMP							0x00000500
+#define    NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_EDGE					0x00000600
+#define    NV40TCL_TEX_WRAP_T_MIRROR_CLAMP_TO_BORDER					0x00000700
+#define    NV40TCL_TEX_WRAP_T_MIRROR_CLAMP						0x00000800
+#define   NV40TCL_TEX_WRAP_EXPAND_NORMAL_SHIFT						12
+#define   NV40TCL_TEX_WRAP_EXPAND_NORMAL_MASK						0x0000f000
+#define   NV40TCL_TEX_WRAP_R_SHIFT							16
+#define   NV40TCL_TEX_WRAP_R_MASK							0x00ff0000
+#define    NV40TCL_TEX_WRAP_R_REPEAT							0x00010000
+#define    NV40TCL_TEX_WRAP_R_MIRRORED_REPEAT						0x00020000
+#define    NV40TCL_TEX_WRAP_R_CLAMP_TO_EDGE						0x00030000
+#define    NV40TCL_TEX_WRAP_R_CLAMP_TO_BORDER						0x00040000
+#define    NV40TCL_TEX_WRAP_R_CLAMP							0x00050000
+#define    NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_EDGE					0x00060000
+#define    NV40TCL_TEX_WRAP_R_MIRROR_CLAMP_TO_BORDER					0x00070000
+#define    NV40TCL_TEX_WRAP_R_MIRROR_CLAMP						0x00080000
+#define   NV40TCL_TEX_WRAP_RCOMP_SHIFT							28
+#define   NV40TCL_TEX_WRAP_RCOMP_MASK							0xf0000000
+#define    NV40TCL_TEX_WRAP_RCOMP_NEVER							0x00000000
+#define    NV40TCL_TEX_WRAP_RCOMP_GREATER						0x10000000
+#define    NV40TCL_TEX_WRAP_RCOMP_EQUAL							0x20000000
+#define    NV40TCL_TEX_WRAP_RCOMP_GEQUAL						0x30000000
+#define    NV40TCL_TEX_WRAP_RCOMP_LESS							0x40000000
+#define    NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL						0x50000000
+#define    NV40TCL_TEX_WRAP_RCOMP_LEQUAL						0x60000000
+#define    NV40TCL_TEX_WRAP_RCOMP_ALWAYS						0x70000000
+#define  NV40TCL_TEX_ENABLE(x)								(0x00001a0c+((x)*32))
+#define  NV40TCL_TEX_ENABLE__SIZE							0x00000010
+#define   NV40TCL_TEX_ENABLE_ENABLE							(1 << 31)
+#define   NV40TCL_TEX_ENABLE_MIPMAP_MIN_LOD_SHIFT					27
+#define   NV40TCL_TEX_ENABLE_MIPMAP_MIN_LOD_MASK					0x38000000
+#define   NV40TCL_TEX_ENABLE_MIPMAP_MAX_LOD_SHIFT					15
+#define   NV40TCL_TEX_ENABLE_MIPMAP_MAX_LOD_MASK					0x00038000
+#define   NV40TCL_TEX_ENABLE_ANISO_SHIFT						4
+#define   NV40TCL_TEX_ENABLE_ANISO_MASK							0x000000f0
+#define    NV40TCL_TEX_ENABLE_ANISO_NONE						0x00000000
+#define    NV40TCL_TEX_ENABLE_ANISO_2X							0x00000010
+#define    NV40TCL_TEX_ENABLE_ANISO_4X							0x00000020
+#define    NV40TCL_TEX_ENABLE_ANISO_6X							0x00000030
+#define    NV40TCL_TEX_ENABLE_ANISO_8X							0x00000040
+#define    NV40TCL_TEX_ENABLE_ANISO_10X							0x00000050
+#define    NV40TCL_TEX_ENABLE_ANISO_12X							0x00000060
+#define    NV40TCL_TEX_ENABLE_ANISO_16X							0x00000070
+#define  NV40TCL_TEX_SWIZZLE(x)								(0x00001a10+((x)*32))
+#define  NV40TCL_TEX_SWIZZLE__SIZE							0x00000010
+#define   NV40TCL_TEX_SWIZZLE_S0_X_SHIFT						14
+#define   NV40TCL_TEX_SWIZZLE_S0_X_MASK							0x0000c000
+#define    NV40TCL_TEX_SWIZZLE_S0_X_ZERO						0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S0_X_ONE							0x00004000
+#define    NV40TCL_TEX_SWIZZLE_S0_X_S1							0x00008000
+#define   NV40TCL_TEX_SWIZZLE_S0_Y_SHIFT						12
+#define   NV40TCL_TEX_SWIZZLE_S0_Y_MASK							0x00003000
+#define    NV40TCL_TEX_SWIZZLE_S0_Y_ZERO						0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S0_Y_ONE							0x00001000
+#define    NV40TCL_TEX_SWIZZLE_S0_Y_S1							0x00002000
+#define   NV40TCL_TEX_SWIZZLE_S0_Z_SHIFT						10
+#define   NV40TCL_TEX_SWIZZLE_S0_Z_MASK							0x00000c00
+#define    NV40TCL_TEX_SWIZZLE_S0_Z_ZERO						0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S0_Z_ONE							0x00000400
+#define    NV40TCL_TEX_SWIZZLE_S0_Z_S1							0x00000800
+#define   NV40TCL_TEX_SWIZZLE_S0_W_SHIFT						8
+#define   NV40TCL_TEX_SWIZZLE_S0_W_MASK							0x00000300
+#define    NV40TCL_TEX_SWIZZLE_S0_W_ZERO						0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S0_W_ONE							0x00000100
+#define    NV40TCL_TEX_SWIZZLE_S0_W_S1							0x00000200
+#define   NV40TCL_TEX_SWIZZLE_S1_X_SHIFT						6
+#define   NV40TCL_TEX_SWIZZLE_S1_X_MASK							0x000000c0
+#define    NV40TCL_TEX_SWIZZLE_S1_X_W							0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S1_X_Z							0x00000040
+#define    NV40TCL_TEX_SWIZZLE_S1_X_Y							0x00000080
+#define    NV40TCL_TEX_SWIZZLE_S1_X_X							0x000000c0
+#define   NV40TCL_TEX_SWIZZLE_S1_Y_SHIFT						4
+#define   NV40TCL_TEX_SWIZZLE_S1_Y_MASK							0x00000030
+#define    NV40TCL_TEX_SWIZZLE_S1_Y_W							0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S1_Y_Z							0x00000010
+#define    NV40TCL_TEX_SWIZZLE_S1_Y_Y							0x00000020
+#define    NV40TCL_TEX_SWIZZLE_S1_Y_X							0x00000030
+#define   NV40TCL_TEX_SWIZZLE_S1_Z_SHIFT						2
+#define   NV40TCL_TEX_SWIZZLE_S1_Z_MASK							0x0000000c
+#define    NV40TCL_TEX_SWIZZLE_S1_Z_W							0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S1_Z_Z							0x00000004
+#define    NV40TCL_TEX_SWIZZLE_S1_Z_Y							0x00000008
+#define    NV40TCL_TEX_SWIZZLE_S1_Z_X							0x0000000c
+#define   NV40TCL_TEX_SWIZZLE_S1_W_SHIFT						0
+#define   NV40TCL_TEX_SWIZZLE_S1_W_MASK							0x00000003
+#define    NV40TCL_TEX_SWIZZLE_S1_W_W							0x00000000
+#define    NV40TCL_TEX_SWIZZLE_S1_W_Z							0x00000001
+#define    NV40TCL_TEX_SWIZZLE_S1_W_Y							0x00000002
+#define    NV40TCL_TEX_SWIZZLE_S1_W_X							0x00000003
+#define  NV40TCL_TEX_FILTER(x)								(0x00001a14+((x)*32))
+#define  NV40TCL_TEX_FILTER__SIZE							0x00000010
+#define   NV40TCL_TEX_FILTER_SIGNED_ALPHA						(1 << 31)
+#define   NV40TCL_TEX_FILTER_SIGNED_RED							(1 << 30)
+#define   NV40TCL_TEX_FILTER_SIGNED_GREEN						(1 << 29)
+#define   NV40TCL_TEX_FILTER_SIGNED_BLUE						(1 << 28)
+#define   NV40TCL_TEX_FILTER_MIN_SHIFT							16
+#define   NV40TCL_TEX_FILTER_MIN_MASK							0x000f0000
+#define    NV40TCL_TEX_FILTER_MIN_NEAREST						0x00010000
+#define    NV40TCL_TEX_FILTER_MIN_LINEAR						0x00020000
+#define    NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST				0x00030000
+#define    NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST					0x00040000
+#define    NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR					0x00050000
+#define    NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR					0x00060000
+#define   NV40TCL_TEX_FILTER_MAG_SHIFT							24
+#define   NV40TCL_TEX_FILTER_MAG_MASK							0x0f000000
+#define    NV40TCL_TEX_FILTER_MAG_NEAREST						0x01000000
+#define    NV40TCL_TEX_FILTER_MAG_LINEAR						0x02000000
+#define  NV40TCL_TEX_SIZE0(x)								(0x00001a18+((x)*32))
+#define  NV40TCL_TEX_SIZE0__SIZE							0x00000010
+#define   NV40TCL_TEX_SIZE0_H_SHIFT							0
+#define   NV40TCL_TEX_SIZE0_H_MASK							0x0000ffff
+#define   NV40TCL_TEX_SIZE0_W_SHIFT							16
+#define   NV40TCL_TEX_SIZE0_W_MASK							0xffff0000
+#define  NV40TCL_TEX_BORDER_COLOR(x)							(0x00001a1c+((x)*32))
+#define  NV40TCL_TEX_BORDER_COLOR__SIZE							0x00000010
+#define   NV40TCL_TEX_BORDER_COLOR_B_SHIFT						0
+#define   NV40TCL_TEX_BORDER_COLOR_B_MASK						0x000000ff
+#define   NV40TCL_TEX_BORDER_COLOR_G_SHIFT						8
+#define   NV40TCL_TEX_BORDER_COLOR_G_MASK						0x0000ff00
+#define   NV40TCL_TEX_BORDER_COLOR_R_SHIFT						16
+#define   NV40TCL_TEX_BORDER_COLOR_R_MASK						0x00ff0000
+#define   NV40TCL_TEX_BORDER_COLOR_A_SHIFT						24
+#define   NV40TCL_TEX_BORDER_COLOR_A_MASK						0xff000000
+#define  NV40TCL_VTX_ATTR_4F_X(x)							(0x00001c00+((x)*16))
+#define  NV40TCL_VTX_ATTR_4F_X__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_4F_Y(x)							(0x00001c04+((x)*16))
+#define  NV40TCL_VTX_ATTR_4F_Y__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_4F_Z(x)							(0x00001c08+((x)*16))
+#define  NV40TCL_VTX_ATTR_4F_Z__SIZE							0x00000010
+#define  NV40TCL_VTX_ATTR_4F_W(x)							(0x00001c0c+((x)*16))
+#define  NV40TCL_VTX_ATTR_4F_W__SIZE							0x00000010
+#define  NV40TCL_FP_CONTROL								0x00001d60
+#define   NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT						24
+#define   NV40TCL_FP_CONTROL_TEMP_COUNT_MASK						0xff000000
+#define   NV40TCL_FP_CONTROL_KIL							(1 <<  7)
+#define  NV40TCL_MULTISAMPLE_CONTROL							0x00001d7c
+#define  NV40TCL_CLEAR_VALUE_DEPTH							0x00001d8c
+#define  NV40TCL_CLEAR_VALUE_COLOR							0x00001d90
+#define  NV40TCL_CLEAR_BUFFERS								0x00001d94
+#define   NV40TCL_CLEAR_BUFFERS_COLOR_A							(1 <<  7)
+#define   NV40TCL_CLEAR_BUFFERS_COLOR_B							(1 <<  6)
+#define   NV40TCL_CLEAR_BUFFERS_COLOR_G							(1 <<  5)
+#define   NV40TCL_CLEAR_BUFFERS_COLOR_R							(1 <<  4)
+#define   NV40TCL_CLEAR_BUFFERS_STENCIL							(1 <<  1)
+#define   NV40TCL_CLEAR_BUFFERS_DEPTH							(1 <<  0)
+#define  NV40TCL_LINE_STIPPLE_ENABLE							0x00001db4
+#define  NV40TCL_LINE_STIPPLE_PATTERN							0x00001db8
+#define   NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_SHIFT					0
+#define   NV40TCL_LINE_STIPPLE_PATTERN_FACTOR_MASK					0x0000ffff
+#define   NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_SHIFT					16
+#define   NV40TCL_LINE_STIPPLE_PATTERN_PATTERN_MASK					0xffff0000
+#define  NV40TCL_VTX_ATTR_1F(x)								(0x00001e40+((x)*4))
+#define  NV40TCL_VTX_ATTR_1F__SIZE							0x00000010
+#define  NV40TCL_VP_UPLOAD_FROM_ID							0x00001e9c
+#define  NV40TCL_VP_START_FROM_ID							0x00001ea0
+#define  NV40TCL_POINT_SIZE								0x00001ee0
+#define  NV40TCL_POINT_SPRITE								0x00001ee8
+#define  NV40TCL_VP_UPLOAD_CONST_ID							0x00001efc
+#define  NV40TCL_VP_UPLOAD_CONST_X(x)							(0x00001f00+((x)*16))
+#define  NV40TCL_VP_UPLOAD_CONST_X__SIZE						0x00000004
+#define  NV40TCL_VP_UPLOAD_CONST_Y(x)							(0x00001f04+((x)*16))
+#define  NV40TCL_VP_UPLOAD_CONST_Y__SIZE						0x00000004
+#define  NV40TCL_VP_UPLOAD_CONST_Z(x)							(0x00001f08+((x)*16))
+#define  NV40TCL_VP_UPLOAD_CONST_Z__SIZE						0x00000004
+#define  NV40TCL_VP_UPLOAD_CONST_W(x)							(0x00001f0c+((x)*16))
+#define  NV40TCL_VP_UPLOAD_CONST_W__SIZE						0x00000004
+#define  NV40TCL_TEX_CACHE_CTL								0x00001fd8
+#define  NV40TCL_VP_ATTRIB_EN								0x00001ff0
+#define  NV40TCL_VP_RESULT_EN								0x00001ff4
+
+
+#define NV44TCL										0x00004497
+
+
+
+#define NV50_2D										0x0000502d
+
+#define  NV50_2D_NOP									0x00000100
+#define  NV50_2D_NOTIFY									0x00000104
+#define  NV50_2D_DMA_NOTIFY								0x00000180
+#define  NV50_2D_DMA_IN_MEMORY0								0x00000184
+#define  NV50_2D_DMA_IN_MEMORY1								0x00000188
+#define  NV50_2D_DMA_IN_MEMORY2								0x0000018c
+#define  NV50_2D_DST_FORMAT								0x00000200
+#define   NV50_2D_DST_FORMAT_32BPP							0x000000cf
+#define   NV50_2D_DST_FORMAT_24BPP							0x000000e6
+#define   NV50_2D_DST_FORMAT_16BPP							0x000000e8
+#define   NV50_2D_DST_FORMAT_8BPP							0x000000f3
+#define   NV50_2D_DST_FORMAT_15BPP							0x000000f8
+#define  NV50_2D_DST_PITCH								0x00000214
+#define  NV50_2D_DST_WIDTH								0x00000218
+#define  NV50_2D_DST_HEIGHT								0x0000021c
+#define  NV50_2D_DST_ADDRESS_HIGH							0x00000220
+#define  NV50_2D_DST_ADDRESS_LOW							0x00000224
+#define  NV50_2D_SRC_FORMAT								0x00000230
+#define   NV50_2D_SRC_FORMAT_32BPP							0x000000cf
+#define   NV50_2D_SRC_FORMAT_24BPP							0x000000e6
+#define   NV50_2D_SRC_FORMAT_16BPP							0x000000e8
+#define   NV50_2D_SRC_FORMAT_8BPP							0x000000f3
+#define   NV50_2D_SRC_FORMAT_15BPP							0x000000f8
+#define  NV50_2D_SRC_PITCH								0x00000244
+#define  NV50_2D_SRC_WIDTH								0x00000248
+#define  NV50_2D_SRC_HEIGHT								0x0000024c
+#define  NV50_2D_SRC_ADDRESS_HIGH							0x00000250
+#define  NV50_2D_SRC_ADDRESS_LOW							0x00000254
+#define  NV50_2D_CLIP_X									0x00000280
+#define  NV50_2D_CLIP_Y									0x00000284
+#define  NV50_2D_CLIP_Z									0x00000288
+#define  NV50_2D_CLIP_W									0x0000028c
+#define  NV50_2D_ROP									0x000002a0
+#define  NV50_2D_OPERATION								0x000002ac
+#define   NV50_2D_OPERATION_SRCCOPY_AND							0x00000000
+#define   NV50_2D_OPERATION_ROP_AND							0x00000001
+#define   NV50_2D_OPERATION_BLEND_AND							0x00000002
+#define   NV50_2D_OPERATION_SRCCOPY							0x00000003
+#define   NV50_2D_OPERATION_SRCCOPY_PREMULT						0x00000004
+#define   NV50_2D_OPERATION_BLEND_PREMULT						0x00000005
+#define  NV50_2D_PATTERN_FORMAT								0x000002e8
+#define   NV50_2D_PATTERN_FORMAT_16BPP							0x00000000
+#define   NV50_2D_PATTERN_FORMAT_15BPP							0x00000001
+#define   NV50_2D_PATTERN_FORMAT_32BPP							0x00000002
+#define   NV50_2D_PATTERN_FORMAT_8BPP							0x00000003
+#define  NV50_2D_PATTERN_COLOR(x)							(0x000002f0+((x)*4))
+#define  NV50_2D_PATTERN_COLOR__SIZE							0x00000002
+#define  NV50_2D_PATTERN_BITMAP(x)							(0x000002f8+((x)*4))
+#define  NV50_2D_PATTERN_BITMAP__SIZE							0x00000002
+#define  NV50_2D_RECT_FORMAT								0x00000584
+#define   NV50_2D_RECT_FORMAT_32BPP							0x000000cf
+#define   NV50_2D_RECT_FORMAT_24BPP							0x000000e6
+#define   NV50_2D_RECT_FORMAT_16BPP							0x000000e8
+#define   NV50_2D_RECT_FORMAT_8BPP							0x000000f3
+#define   NV50_2D_RECT_FORMAT_15BPP							0x000000f8
+#define  NV50_2D_RECT_COLOR								0x00000588
+#define  NV50_2D_RECT_X1								0x00000600
+#define  NV50_2D_RECT_Y1								0x00000604
+#define  NV50_2D_RECT_X2								0x00000608
+#define  NV50_2D_RECT_Y2								0x0000060c
+#define  NV50_2D_SIFC_UNK0800								0x00000800
+#define  NV50_2D_SIFC_FORMAT								0x00000804
+#define   NV50_2D_SIFC_FORMAT_32BPP							0x000000cf
+#define   NV50_2D_SIFC_FORMAT_24BPP							0x000000e6
+#define   NV50_2D_SIFC_FORMAT_16BPP							0x000000e8
+#define   NV50_2D_SIFC_FORMAT_8BPP							0x000000f3
+#define   NV50_2D_SIFC_FORMAT_15BPP							0x000000f8
+#define  NV50_2D_SIFC_WIDTH								0x00000838
+#define  NV50_2D_SIFC_HEIGHT								0x0000083c
+#define  NV50_2D_SIFC_SCALE_UNK0840							0x00000840
+#define  NV50_2D_SIFC_SCALE_UNK0844							0x00000844
+#define  NV50_2D_SIFC_SCALE_UNK0848							0x00000848
+#define  NV50_2D_SIFC_SCALE_UNK084C							0x0000084c
+#define  NV50_2D_SIFC_UNK0850								0x00000850
+#define  NV50_2D_SIFC_DST_X								0x00000854
+#define  NV50_2D_SIFC_UNK0858								0x00000858
+#define  NV50_2D_SIFC_DST_Y								0x0000085c
+#define  NV50_2D_SIFC_DATA								0x00000860
+#define  NV50_2D_BLIT_DST_X								0x000008b0
+#define  NV50_2D_BLIT_DST_Y								0x000008b4
+#define  NV50_2D_BLIT_DST_W								0x000008b8
+#define  NV50_2D_BLIT_DST_H								0x000008bc
+#define  NV50_2D_BLIT_SRC_X								0x000008d4
+#define  NV50_2D_BLIT_SRC_Y								0x000008dc
+
+
+#define NV50_MEMORY_TO_MEMORY_FORMAT							0x00005039
+
+#define  NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN_HIGH					0x00000238
+#define  NV50_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT_HIGH					0x0000023c
+
+
+#define NV50TCL										0x00005097
+
+#define  NV50TCL_NOP									0x00000100
+#define  NV50TCL_NOTIFY									0x00000104
+#define  NV50TCL_DMA_NOTIFY								0x00000180
+#define  NV50TCL_DMA_UNK0(x)								(0x00000184+((x)*4))
+#define  NV50TCL_DMA_UNK0__SIZE								0x0000000b
+#define  NV50TCL_DMA_UNK1(x)								(0x000001c0+((x)*4))
+#define  NV50TCL_DMA_UNK1__SIZE								0x00000008
+#define  NV50TCL_RT_ADDRESS_HIGH(x)							(0x00000200+((x)*32))
+#define  NV50TCL_RT_ADDRESS_HIGH__SIZE							0x00000008
+#define  NV50TCL_RT_ADDRESS_LOW(x)							(0x00000204+((x)*32))
+#define  NV50TCL_RT_ADDRESS_LOW__SIZE							0x00000008
+#define  NV50TCL_RT_FORMAT(x)								(0x00000208+((x)*32))
+#define  NV50TCL_RT_FORMAT__SIZE							0x00000008
+#define   NV50TCL_RT_FORMAT_32BPP							0x000000cf
+#define   NV50TCL_RT_FORMAT_24BPP							0x000000e6
+#define   NV50TCL_RT_FORMAT_16BPP							0x000000e8
+#define   NV50TCL_RT_FORMAT_8BPP							0x000000f3
+#define   NV50TCL_RT_FORMAT_15BPP							0x000000f8
+#define  NV50TCL_RT_TILE_UNK(x)								(0x0000020c+((x)*32))
+#define  NV50TCL_RT_TILE_UNK__SIZE							0x00000008
+#define  NV50TCL_RT_UNK4(x)								(0x00000210+((x)*32))
+#define  NV50TCL_RT_UNK4__SIZE								0x00000008
+#define  NV50TCL_VTX_ATTR_1F(x)								(0x00000300+((x)*4))
+#define  NV50TCL_VTX_ATTR_1F__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_2F_X(x)							(0x00000380+((x)*8))
+#define  NV50TCL_VTX_ATTR_2F_X__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_2F_Y(x)							(0x00000384+((x)*8))
+#define  NV50TCL_VTX_ATTR_2F_Y__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_3F_X(x)							(0x00000400+((x)*16))
+#define  NV50TCL_VTX_ATTR_3F_X__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_3F_Y(x)							(0x00000404+((x)*16))
+#define  NV50TCL_VTX_ATTR_3F_Y__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_3F_Z(x)							(0x00000408+((x)*16))
+#define  NV50TCL_VTX_ATTR_3F_Z__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_3F_W(x)							(0x0000040c+((x)*16))
+#define  NV50TCL_VTX_ATTR_3F_W__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_4F_X(x)							(0x00000500+((x)*16))
+#define  NV50TCL_VTX_ATTR_4F_X__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_4F_Y(x)							(0x00000504+((x)*16))
+#define  NV50TCL_VTX_ATTR_4F_Y__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_4F_Z(x)							(0x00000508+((x)*16))
+#define  NV50TCL_VTX_ATTR_4F_Z__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_4F_W(x)							(0x0000050c+((x)*16))
+#define  NV50TCL_VTX_ATTR_4F_W__SIZE							0x00000010
+#define  NV50TCL_VTX_ATTR_2I(x)								(0x00000680+((x)*4))
+#define  NV50TCL_VTX_ATTR_2I__SIZE							0x00000010
+#define   NV50TCL_VTX_ATTR_2I_X_SHIFT							0
+#define   NV50TCL_VTX_ATTR_2I_X_MASK							0x0000ffff
+#define   NV50TCL_VTX_ATTR_2I_Y_SHIFT							16
+#define   NV50TCL_VTX_ATTR_2I_Y_MASK							0xffff0000
+#define  NV50TCL_VTX_ATTR_4I_0(x)							(0x00000700+((x)*8))
+#define  NV50TCL_VTX_ATTR_4I_0__SIZE							0x00000010
+#define   NV50TCL_VTX_ATTR_4I_0_X_SHIFT							0
+#define   NV50TCL_VTX_ATTR_4I_0_X_MASK							0x0000ffff
+#define   NV50TCL_VTX_ATTR_4I_0_Y_SHIFT							16
+#define   NV50TCL_VTX_ATTR_4I_0_Y_MASK							0xffff0000
+#define  NV50TCL_VTX_ATTR_4I_1(x)							(0x00000704+((x)*8))
+#define  NV50TCL_VTX_ATTR_4I_1__SIZE							0x00000010
+#define   NV50TCL_VTX_ATTR_4I_1_Z_SHIFT							0
+#define   NV50TCL_VTX_ATTR_4I_1_Z_MASK							0x0000ffff
+#define   NV50TCL_VTX_ATTR_4I_1_W_SHIFT							16
+#define   NV50TCL_VTX_ATTR_4I_1_W_MASK							0xffff0000
+#define  NV50TCL_VTX_ATTR_4NI_0(x)							(0x00000780+((x)*8))
+#define  NV50TCL_VTX_ATTR_4NI_0__SIZE							0x00000010
+#define   NV50TCL_VTX_ATTR_4NI_0_X_SHIFT						0
+#define   NV50TCL_VTX_ATTR_4NI_0_X_MASK							0x0000ffff
+#define   NV50TCL_VTX_ATTR_4NI_0_Y_SHIFT						16
+#define   NV50TCL_VTX_ATTR_4NI_0_Y_MASK							0xffff0000
+#define  NV50TCL_VTX_ATTR_4NI_1(x)							(0x00000784+((x)*8))
+#define  NV50TCL_VTX_ATTR_4NI_1__SIZE							0x00000010
+#define   NV50TCL_VTX_ATTR_4NI_1_Z_SHIFT						0
+#define   NV50TCL_VTX_ATTR_4NI_1_Z_MASK							0x0000ffff
+#define   NV50TCL_VTX_ATTR_4NI_1_W_SHIFT						16
+#define   NV50TCL_VTX_ATTR_4NI_1_W_MASK							0xffff0000
+#define  NV50TCL_VERTEX_ARRAY_FORMAT(x)							(0x00000900+((x)*16))
+#define  NV50TCL_VERTEX_ARRAY_FORMAT__SIZE						0x00000010
+#define  NV50TCL_VIEWPORT_UNK0(x)							(0x00000a00+((x)*4))
+#define  NV50TCL_VIEWPORT_UNK0__SIZE							0x00000003
+#define  NV50TCL_VIEWPORT_UNK1(x)							(0x00000a0c+((x)*4))
+#define  NV50TCL_VIEWPORT_UNK1__SIZE							0x00000003
+#define  NV50TCL_VIEWPORT_HORIZ								0x00000c00
+#define   NV50TCL_VIEWPORT_HORIZ_X_SHIFT						0
+#define   NV50TCL_VIEWPORT_HORIZ_X_MASK							0x0000ffff
+#define   NV50TCL_VIEWPORT_HORIZ_W_SHIFT						16
+#define   NV50TCL_VIEWPORT_HORIZ_W_MASK							0xffff0000
+#define  NV50TCL_VIEWPORT_VERT								0x00000c04
+#define   NV50TCL_VIEWPORT_VERT_Y_SHIFT							0
+#define   NV50TCL_VIEWPORT_VERT_Y_MASK							0x0000ffff
+#define   NV50TCL_VIEWPORT_VERT_H_SHIFT							16
+#define   NV50TCL_VIEWPORT_VERT_H_MASK							0xffff0000
+#define  NV50TCL_DEPTH_RANGE_NEAR							0x00000c08
+#define  NV50TCL_DEPTH_RANGE_FAR							0x00000c0c
+#define  NV50TCL_VIEWPORT_CLIP_HORIZ(x)							(0x00000d00+((x)*8))
+#define  NV50TCL_VIEWPORT_CLIP_HORIZ__SIZE						0x00000008
+#define  NV50TCL_VIEWPORT_CLIP_VERT(x)							(0x00000d04+((x)*8))
+#define  NV50TCL_VIEWPORT_CLIP_VERT__SIZE						0x00000008
+#define  NV50TCL_VERTEX_BUFFER_FIRST							0x00000d74
+#define  NV50TCL_VERTEX_BUFFER_COUNT							0x00000d78
+#define  NV50TCL_CLEAR_COLOR(x)								(0x00000d80+((x)*4))
+#define  NV50TCL_CLEAR_COLOR__SIZE							0x00000004
+#define  NV50TCL_CLEAR_DEPTH								0x00000d90
+#define  NV50TCL_CLEAR_STENCIL								0x00000da0
+#define  NV50TCL_POLYGON_MODE_FRONT							0x00000dac
+#define   NV50TCL_POLYGON_MODE_FRONT_POINT						0x00001b00
+#define   NV50TCL_POLYGON_MODE_FRONT_LINE						0x00001b01
+#define   NV50TCL_POLYGON_MODE_FRONT_FILL						0x00001b02
+#define  NV50TCL_POLYGON_MODE_BACK							0x00000db0
+#define   NV50TCL_POLYGON_MODE_BACK_POINT						0x00001b00
+#define   NV50TCL_POLYGON_MODE_BACK_LINE						0x00001b01
+#define   NV50TCL_POLYGON_MODE_BACK_FILL						0x00001b02
+#define  NV50TCL_POLYGON_SMOOTH_ENABLE							0x00000db4
+#define  NV50TCL_POLYGON_OFFSET_POINT_ENABLE						0x00000dc0
+#define  NV50TCL_POLYGON_OFFSET_LINE_ENABLE						0x00000dc4
+#define  NV50TCL_POLYGON_OFFSET_FILL_ENABLE						0x00000dc8
+#define  NV50TCL_SCISSOR_HORIZ								0x00000e04
+#define   NV50TCL_SCISSOR_HORIZ_L_SHIFT							0
+#define   NV50TCL_SCISSOR_HORIZ_L_MASK							0x0000ffff
+#define   NV50TCL_SCISSOR_HORIZ_R_SHIFT							16
+#define   NV50TCL_SCISSOR_HORIZ_R_MASK							0xffff0000
+#define  NV50TCL_SCISSOR_VERT								0x00000e08
+#define   NV50TCL_SCISSOR_VERT_T_SHIFT							0
+#define   NV50TCL_SCISSOR_VERT_T_MASK							0x0000ffff
+#define   NV50TCL_SCISSOR_VERT_B_SHIFT							16
+#define   NV50TCL_SCISSOR_VERT_B_MASK							0xffff0000
+#define  NV50TCL_CB_ADDR								0x00000f00
+#define   NV50TCL_CB_ADDR_ID_SHIFT							8
+#define   NV50TCL_CB_ADDR_ID_MASK							0xffffff00
+#define   NV50TCL_CB_ADDR_BUFFER_SHIFT							0
+#define   NV50TCL_CB_ADDR_BUFFER_MASK							0x000000ff
+#define  NV50TCL_CB_DATA(x)								(0x00000f04+((x)*4))
+#define  NV50TCL_CB_DATA__SIZE								0x00000010
+#define  NV50TCL_STENCIL_FRONT_FUNC_REF							0x00000f54
+#define  NV50TCL_STENCIL_FRONT_MASK							0x00000f58
+#define  NV50TCL_STENCIL_FRONT_FUNC_MASK						0x00000f5c
+#define  NV50TCL_GP_ADDRESS_HIGH							0x00000f70
+#define  NV50TCL_GP_ADDRESS_LOW								0x00000f74
+#define  NV50TCL_VP_ADDRESS_HIGH							0x00000f7c
+#define  NV50TCL_VP_ADDRESS_LOW								0x00000f80
+#define  NV50TCL_FP_ADDRESS_HIGH							0x00000fa4
+#define  NV50TCL_FP_ADDRESS_LOW								0x00000fa8
+#define  NV50TCL_ZETA_ADDRESS_HIGH							0x00000fe0
+#define  NV50TCL_ZETA_ADDRESS_LOW							0x00000fe4
+#define  NV50TCL_UNKFF4									0x00000ff4
+#define   NV50TCL_UNKFF4_W_SHIFT							16
+#define   NV50TCL_UNKFF4_W_MASK								0xffff0000
+#define  NV50TCL_UNKFF8									0x00000ff8
+#define   NV50TCL_UNKFF8_H_SHIFT							16
+#define   NV50TCL_UNKFF8_H_MASK								0xffff0000
+#define  NV50TCL_RT_HORIZ(x)								(0x00001240+((x)*8))
+#define  NV50TCL_RT_HORIZ__SIZE								0x00000008
+#define  NV50TCL_RT_VERT(x)								(0x00001244+((x)*8))
+#define  NV50TCL_RT_VERT__SIZE								0x00000008
+#define  NV50TCL_CB_DEF_ADDRESS_HIGH							0x00001280
+#define  NV50TCL_CB_DEF_ADDRESS_LOW							0x00001284
+#define  NV50TCL_CB_DEF_SET								0x00001288
+#define   NV50TCL_CB_DEF_SET_SIZE_SHIFT							0
+#define   NV50TCL_CB_DEF_SET_SIZE_MASK							0x0000ffff
+#define   NV50TCL_CB_DEF_SET_BUFFER_SHIFT						16
+#define   NV50TCL_CB_DEF_SET_BUFFER_MASK						0xffff0000
+#define  NV50TCL_DEPTH_TEST_ENABLE							0x000012cc
+#define  NV50TCL_SHADE_MODEL								0x000012d4
+#define   NV50TCL_SHADE_MODEL_FLAT							0x00001d00
+#define   NV50TCL_SHADE_MODEL_SMOOTH							0x00001d01
+#define  NV50TCL_DEPTH_WRITE_ENABLE							0x000012e8
+#define  NV50TCL_ALPHA_TEST_ENABLE							0x000012ec
+#define  NV50TCL_DEPTH_TEST_FUNC							0x0000130c
+#define   NV50TCL_DEPTH_TEST_FUNC_NEVER							0x00000200
+#define   NV50TCL_DEPTH_TEST_FUNC_LESS							0x00000201
+#define   NV50TCL_DEPTH_TEST_FUNC_EQUAL							0x00000202
+#define   NV50TCL_DEPTH_TEST_FUNC_LEQUAL						0x00000203
+#define   NV50TCL_DEPTH_TEST_FUNC_GREATER						0x00000204
+#define   NV50TCL_DEPTH_TEST_FUNC_GREATER						0x00000204
+#define   NV50TCL_DEPTH_TEST_FUNC_NOTEQUAL						0x00000205
+#define   NV50TCL_DEPTH_TEST_FUNC_GEQUAL						0x00000206
+#define   NV50TCL_DEPTH_TEST_FUNC_ALWAYS						0x00000207
+#define  NV50TCL_ALPHA_TEST_REF								0x00001310
+#define  NV50TCL_ALPHA_TEST_FUNC							0x00001314
+#define   NV50TCL_ALPHA_TEST_FUNC_NEVER							0x00000200
+#define   NV50TCL_ALPHA_TEST_FUNC_LESS							0x00000201
+#define   NV50TCL_ALPHA_TEST_FUNC_EQUAL							0x00000202
+#define   NV50TCL_ALPHA_TEST_FUNC_LEQUAL						0x00000203
+#define   NV50TCL_ALPHA_TEST_FUNC_GREATER						0x00000204
+#define   NV50TCL_ALPHA_TEST_FUNC_GREATER						0x00000204
+#define   NV50TCL_ALPHA_TEST_FUNC_NOTEQUAL						0x00000205
+#define   NV50TCL_ALPHA_TEST_FUNC_GEQUAL						0x00000206
+#define   NV50TCL_ALPHA_TEST_FUNC_ALWAYS						0x00000207
+#define  NV50TCL_BLEND_COLOR(x)								(0x0000131c+((x)*4))
+#define  NV50TCL_BLEND_COLOR__SIZE							0x00000004
+#define  NV50TCL_BLEND_EQUATION_RGB							0x00001340
+#define   NV50TCL_BLEND_EQUATION_RGB_FUNC_ADD						0x00008006
+#define   NV50TCL_BLEND_EQUATION_RGB_MIN						0x00008007
+#define   NV50TCL_BLEND_EQUATION_RGB_MAX						0x00008008
+#define   NV50TCL_BLEND_EQUATION_RGB_FUNC_SUBTRACT					0x0000800a
+#define   NV50TCL_BLEND_EQUATION_RGB_FUNC_REVERSE_SUBTRACT				0x0000800b
+#define  NV50TCL_BLEND_FUNC_SRC_RGB							0x00001344
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ZERO						0x00000000
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE						0x00000001
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_SRC_COLOR						0x00000300
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA						0x00000302
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_DST_ALPHA						0x00000304
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_DST_COLOR						0x00000306
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_SRC_ALPHA_SATURATE					0x00000308
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_COLOR					0x00008001
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_CONSTANT_ALPHA					0x00008003
+#define   NV50TCL_BLEND_FUNC_SRC_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV50TCL_BLEND_FUNC_DST_RGB							0x00001348
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ZERO						0x00000000
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE						0x00000001
+#define   NV50TCL_BLEND_FUNC_DST_RGB_SRC_COLOR						0x00000300
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_COLOR				0x00000301
+#define   NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA						0x00000302
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_SRC_ALPHA				0x00000303
+#define   NV50TCL_BLEND_FUNC_DST_RGB_DST_ALPHA						0x00000304
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_ALPHA				0x00000305
+#define   NV50TCL_BLEND_FUNC_DST_RGB_DST_COLOR						0x00000306
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_DST_COLOR				0x00000307
+#define   NV50TCL_BLEND_FUNC_DST_RGB_SRC_ALPHA_SATURATE					0x00000308
+#define   NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_COLOR					0x00008001
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV50TCL_BLEND_FUNC_DST_RGB_CONSTANT_ALPHA					0x00008003
+#define   NV50TCL_BLEND_FUNC_DST_RGB_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV50TCL_BLEND_EQUATION_ALPHA							0x0000134c
+#define   NV50TCL_BLEND_EQUATION_ALPHA_FUNC_ADD						0x00008006
+#define   NV50TCL_BLEND_EQUATION_ALPHA_MIN						0x00008007
+#define   NV50TCL_BLEND_EQUATION_ALPHA_MAX						0x00008008
+#define   NV50TCL_BLEND_EQUATION_ALPHA_FUNC_SUBTRACT					0x0000800a
+#define   NV50TCL_BLEND_EQUATION_ALPHA_FUNC_REVERSE_SUBTRACT				0x0000800b
+#define  NV50TCL_BLEND_FUNC_SRC_ALPHA							0x00001350
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ZERO						0x00000000
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE						0x00000001
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_COLOR					0x00000300
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_COLOR				0x00000301
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA					0x00000302
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_SRC_ALPHA				0x00000303
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_ALPHA					0x00000304
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_ALPHA				0x00000305
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_DST_COLOR					0x00000306
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_DST_COLOR				0x00000307
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_SRC_ALPHA_SATURATE				0x00000308
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_COLOR					0x00008001
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_CONSTANT_ALPHA					0x00008003
+#define   NV50TCL_BLEND_FUNC_SRC_ALPHA_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV50TCL_BLEND_FUNC_DST_ALPHA							0x00001358
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ZERO						0x00000000
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE						0x00000001
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_COLOR					0x00000300
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_COLOR				0x00000301
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA					0x00000302
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_SRC_ALPHA				0x00000303
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_DST_ALPHA					0x00000304
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_ALPHA				0x00000305
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_DST_COLOR					0x00000306
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_DST_COLOR				0x00000307
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_SRC_ALPHA_SATURATE				0x00000308
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_COLOR					0x00008001
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_COLOR				0x00008002
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_CONSTANT_ALPHA					0x00008003
+#define   NV50TCL_BLEND_FUNC_DST_ALPHA_ONE_MINUS_CONSTANT_ALPHA				0x00008004
+#define  NV50TCL_BLEND_ENABLE(x)							(0x00001360+((x)*4))
+#define  NV50TCL_BLEND_ENABLE__SIZE							0x00000008
+#define  NV50TCL_STENCIL_BACK_ENABLE							0x00001380
+#define  NV50TCL_STENCIL_BACK_OP_FAIL							0x00001384
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_ZERO						0x00000000
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_REPLACE						0x00001e01
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_INCR						0x00001e02
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_DECR						0x00001e03
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_BACK_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_BACK_OP_ZFAIL							0x00001388
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_ZERO						0x00000000
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_REPLACE						0x00001e01
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR						0x00001e02
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR						0x00001e03
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_BACK_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_BACK_OP_ZPASS							0x0000138c
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_ZERO						0x00000000
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_REPLACE						0x00001e01
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_INCR						0x00001e02
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_DECR						0x00001e03
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_BACK_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_BACK_FUNC_FUNC							0x00001390
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_NEVER						0x00000200
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_LESS						0x00000201
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_EQUAL						0x00000202
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_LEQUAL						0x00000203
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_GREATER					0x00000204
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_GEQUAL						0x00000206
+#define   NV50TCL_STENCIL_BACK_FUNC_FUNC_ALWAYS						0x00000207
+#define  NV50TCL_STENCIL_BACK_FUNC_REF							0x00001394
+#define  NV50TCL_STENCIL_BACK_MASK							0x00001398
+#define  NV50TCL_STENCIL_BACK_FUNC_MASK							0x0000139c
+#define  NV50TCL_LINE_WIDTH								0x000013b0
+#define  NV50TCL_VP_START_ID								0x0000140c
+#define  NV50TCL_GP_START_ID								0x00001410
+#define  NV50TCL_FP_START_ID								0x00001414
+#define  NV50TCL_POINT_SIZE								0x00001518
+#define  NV50TCL_TSC_ADDRESS_HIGH							0x0000155c
+#define  NV50TCL_TSC_ADDRESS_LOW							0x00001560
+#define  NV50TCL_POLYGON_OFFSET_FACTOR							0x0000156c
+#define  NV50TCL_LINE_SMOOTH_ENABLE							0x00001570
+#define  NV50TCL_TIC_ADDRESS_HIGH							0x00001574
+#define  NV50TCL_TIC_ADDRESS_LOW							0x00001578
+#define  NV50TCL_STENCIL_FRONT_ENABLE							0x00001594
+#define  NV50TCL_STENCIL_FRONT_OP_FAIL							0x00001598
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_ZERO						0x00000000
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_REPLACE						0x00001e01
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_INCR						0x00001e02
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_DECR						0x00001e03
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_FRONT_OP_FAIL_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_FRONT_OP_ZFAIL							0x0000159c
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_ZERO						0x00000000
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_REPLACE					0x00001e01
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR						0x00001e02
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR						0x00001e03
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_FRONT_OP_ZFAIL_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_FRONT_OP_ZPASS							0x000015a0
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_ZERO						0x00000000
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_INVERT						0x0000150a
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_KEEP						0x00001e00
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_REPLACE					0x00001e01
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR						0x00001e02
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR						0x00001e03
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_INCR_WRAP					0x00008507
+#define   NV50TCL_STENCIL_FRONT_OP_ZPASS_DECR_WRAP					0x00008508
+#define  NV50TCL_STENCIL_FRONT_FUNC_FUNC						0x000015a4
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_NEVER						0x00000200
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_LESS						0x00000201
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_EQUAL						0x00000202
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_LEQUAL					0x00000203
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_GREATER					0x00000204
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_NOTEQUAL					0x00000205
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_GEQUAL					0x00000206
+#define   NV50TCL_STENCIL_FRONT_FUNC_FUNC_ALWAYS					0x00000207
+#define  NV50TCL_POLYGON_OFFSET_UNITS							0x000015bc
+#define  NV50TCL_VERTEX_BEGIN								0x000015dc
+#define   NV50TCL_VERTEX_BEGIN_POINTS							0x00000000
+#define   NV50TCL_VERTEX_BEGIN_LINES							0x00000001
+#define   NV50TCL_VERTEX_BEGIN_LINE_LOOP						0x00000002
+#define   NV50TCL_VERTEX_BEGIN_LINE_STRIP						0x00000003
+#define   NV50TCL_VERTEX_BEGIN_TRIANGLES						0x00000004
+#define   NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP						0x00000005
+#define   NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN						0x00000006
+#define   NV50TCL_VERTEX_BEGIN_QUADS							0x00000007
+#define   NV50TCL_VERTEX_BEGIN_QUAD_STRIP						0x00000008
+#define   NV50TCL_VERTEX_BEGIN_POLYGON							0x00000009
+#define  NV50TCL_VERTEX_END								0x000015e0
+#define  NV50TCL_VERTEX_DATA								0x00001640
+#define  NV50TCL_VP_ATTR_EN_0								0x00001650
+#define   NV50TCL_VP_ATTR_EN_0_7_SHIFT							28
+#define   NV50TCL_VP_ATTR_EN_0_7_MASK							0xf0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XNNN							0x10000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NYNN							0x20000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XYNN							0x30000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NNZN							0x40000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XNZN							0x50000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NYZN							0x60000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XYZN							0x70000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NNNW							0x80000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XNNW							0x90000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NYNW							0xa0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XYNW							0xb0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NNZW							0xc0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XNZW							0xd0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_NYZW							0xe0000000
+#define    NV50TCL_VP_ATTR_EN_0_7_XYZW							0xf0000000
+#define   NV50TCL_VP_ATTR_EN_0_6_SHIFT							24
+#define   NV50TCL_VP_ATTR_EN_0_6_MASK							0x0f000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XNNN							0x01000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NYNN							0x02000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XYNN							0x03000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NNZN							0x04000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XNZN							0x05000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NYZN							0x06000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XYZN							0x07000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NNNW							0x08000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XNNW							0x09000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NYNW							0x0a000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XYNW							0x0b000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NNZW							0x0c000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XNZW							0x0d000000
+#define    NV50TCL_VP_ATTR_EN_0_6_NYZW							0x0e000000
+#define    NV50TCL_VP_ATTR_EN_0_6_XYZW							0x0f000000
+#define   NV50TCL_VP_ATTR_EN_0_5_SHIFT							20
+#define   NV50TCL_VP_ATTR_EN_0_5_MASK							0x00f00000
+#define    NV50TCL_VP_ATTR_EN_0_5_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_5_XNNN							0x00100000
+#define    NV50TCL_VP_ATTR_EN_0_5_NYNN							0x00200000
+#define    NV50TCL_VP_ATTR_EN_0_5_XYNN							0x00300000
+#define    NV50TCL_VP_ATTR_EN_0_5_NNZN							0x00400000
+#define    NV50TCL_VP_ATTR_EN_0_5_XNZN							0x00500000
+#define    NV50TCL_VP_ATTR_EN_0_5_NYZN							0x00600000
+#define    NV50TCL_VP_ATTR_EN_0_5_XYZN							0x00700000
+#define    NV50TCL_VP_ATTR_EN_0_5_NNNW							0x00800000
+#define    NV50TCL_VP_ATTR_EN_0_5_XNNW							0x00900000
+#define    NV50TCL_VP_ATTR_EN_0_5_NYNW							0x00a00000
+#define    NV50TCL_VP_ATTR_EN_0_5_XYNW							0x00b00000
+#define    NV50TCL_VP_ATTR_EN_0_5_NNZW							0x00c00000
+#define    NV50TCL_VP_ATTR_EN_0_5_XNZW							0x00d00000
+#define    NV50TCL_VP_ATTR_EN_0_5_NYZW							0x00e00000
+#define    NV50TCL_VP_ATTR_EN_0_5_XYZW							0x00f00000
+#define   NV50TCL_VP_ATTR_EN_0_4_SHIFT							16
+#define   NV50TCL_VP_ATTR_EN_0_4_MASK							0x000f0000
+#define    NV50TCL_VP_ATTR_EN_0_4_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_4_XNNN							0x00010000
+#define    NV50TCL_VP_ATTR_EN_0_4_NYNN							0x00020000
+#define    NV50TCL_VP_ATTR_EN_0_4_XYNN							0x00030000
+#define    NV50TCL_VP_ATTR_EN_0_4_NNZN							0x00040000
+#define    NV50TCL_VP_ATTR_EN_0_4_XNZN							0x00050000
+#define    NV50TCL_VP_ATTR_EN_0_4_NYZN							0x00060000
+#define    NV50TCL_VP_ATTR_EN_0_4_XYZN							0x00070000
+#define    NV50TCL_VP_ATTR_EN_0_4_NNNW							0x00080000
+#define    NV50TCL_VP_ATTR_EN_0_4_XNNW							0x00090000
+#define    NV50TCL_VP_ATTR_EN_0_4_NYNW							0x000a0000
+#define    NV50TCL_VP_ATTR_EN_0_4_XYNW							0x000b0000
+#define    NV50TCL_VP_ATTR_EN_0_4_NNZW							0x000c0000
+#define    NV50TCL_VP_ATTR_EN_0_4_XNZW							0x000d0000
+#define    NV50TCL_VP_ATTR_EN_0_4_NYZW							0x000e0000
+#define    NV50TCL_VP_ATTR_EN_0_4_XYZW							0x000f0000
+#define   NV50TCL_VP_ATTR_EN_0_3_SHIFT							12
+#define   NV50TCL_VP_ATTR_EN_0_3_MASK							0x0000f000
+#define    NV50TCL_VP_ATTR_EN_0_3_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_3_XNNN							0x00001000
+#define    NV50TCL_VP_ATTR_EN_0_3_NYNN							0x00002000
+#define    NV50TCL_VP_ATTR_EN_0_3_XYNN							0x00003000
+#define    NV50TCL_VP_ATTR_EN_0_3_NNZN							0x00004000
+#define    NV50TCL_VP_ATTR_EN_0_3_XNZN							0x00005000
+#define    NV50TCL_VP_ATTR_EN_0_3_NYZN							0x00006000
+#define    NV50TCL_VP_ATTR_EN_0_3_XYZN							0x00007000
+#define    NV50TCL_VP_ATTR_EN_0_3_NNNW							0x00008000
+#define    NV50TCL_VP_ATTR_EN_0_3_XNNW							0x00009000
+#define    NV50TCL_VP_ATTR_EN_0_3_NYNW							0x0000a000
+#define    NV50TCL_VP_ATTR_EN_0_3_XYNW							0x0000b000
+#define    NV50TCL_VP_ATTR_EN_0_3_NNZW							0x0000c000
+#define    NV50TCL_VP_ATTR_EN_0_3_XNZW							0x0000d000
+#define    NV50TCL_VP_ATTR_EN_0_3_NYZW							0x0000e000
+#define    NV50TCL_VP_ATTR_EN_0_3_XYZW							0x0000f000
+#define   NV50TCL_VP_ATTR_EN_0_2_SHIFT							8
+#define   NV50TCL_VP_ATTR_EN_0_2_MASK							0x00000f00
+#define    NV50TCL_VP_ATTR_EN_0_2_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_2_XNNN							0x00000100
+#define    NV50TCL_VP_ATTR_EN_0_2_NYNN							0x00000200
+#define    NV50TCL_VP_ATTR_EN_0_2_XYNN							0x00000300
+#define    NV50TCL_VP_ATTR_EN_0_2_NNZN							0x00000400
+#define    NV50TCL_VP_ATTR_EN_0_2_XNZN							0x00000500
+#define    NV50TCL_VP_ATTR_EN_0_2_NYZN							0x00000600
+#define    NV50TCL_VP_ATTR_EN_0_2_XYZN							0x00000700
+#define    NV50TCL_VP_ATTR_EN_0_2_NNNW							0x00000800
+#define    NV50TCL_VP_ATTR_EN_0_2_XNNW							0x00000900
+#define    NV50TCL_VP_ATTR_EN_0_2_NYNW							0x00000a00
+#define    NV50TCL_VP_ATTR_EN_0_2_XYNW							0x00000b00
+#define    NV50TCL_VP_ATTR_EN_0_2_NNZW							0x00000c00
+#define    NV50TCL_VP_ATTR_EN_0_2_XNZW							0x00000d00
+#define    NV50TCL_VP_ATTR_EN_0_2_NYZW							0x00000e00
+#define    NV50TCL_VP_ATTR_EN_0_2_XYZW							0x00000f00
+#define   NV50TCL_VP_ATTR_EN_0_1_SHIFT							4
+#define   NV50TCL_VP_ATTR_EN_0_1_MASK							0x000000f0
+#define    NV50TCL_VP_ATTR_EN_0_1_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_1_XNNN							0x00000010
+#define    NV50TCL_VP_ATTR_EN_0_1_NYNN							0x00000020
+#define    NV50TCL_VP_ATTR_EN_0_1_XYNN							0x00000030
+#define    NV50TCL_VP_ATTR_EN_0_1_NNZN							0x00000040
+#define    NV50TCL_VP_ATTR_EN_0_1_XNZN							0x00000050
+#define    NV50TCL_VP_ATTR_EN_0_1_NYZN							0x00000060
+#define    NV50TCL_VP_ATTR_EN_0_1_XYZN							0x00000070
+#define    NV50TCL_VP_ATTR_EN_0_1_NNNW							0x00000080
+#define    NV50TCL_VP_ATTR_EN_0_1_XNNW							0x00000090
+#define    NV50TCL_VP_ATTR_EN_0_1_NYNW							0x000000a0
+#define    NV50TCL_VP_ATTR_EN_0_1_XYNW							0x000000b0
+#define    NV50TCL_VP_ATTR_EN_0_1_NNZW							0x000000c0
+#define    NV50TCL_VP_ATTR_EN_0_1_XNZW							0x000000d0
+#define    NV50TCL_VP_ATTR_EN_0_1_NYZW							0x000000e0
+#define    NV50TCL_VP_ATTR_EN_0_1_XYZW							0x000000f0
+#define   NV50TCL_VP_ATTR_EN_0_0_SHIFT							0
+#define   NV50TCL_VP_ATTR_EN_0_0_MASK							0x0000000f
+#define    NV50TCL_VP_ATTR_EN_0_0_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_0_0_XNNN							0x00000001
+#define    NV50TCL_VP_ATTR_EN_0_0_NYNN							0x00000002
+#define    NV50TCL_VP_ATTR_EN_0_0_XYNN							0x00000003
+#define    NV50TCL_VP_ATTR_EN_0_0_NNZN							0x00000004
+#define    NV50TCL_VP_ATTR_EN_0_0_XNZN							0x00000005
+#define    NV50TCL_VP_ATTR_EN_0_0_NYZN							0x00000006
+#define    NV50TCL_VP_ATTR_EN_0_0_XYZN							0x00000007
+#define    NV50TCL_VP_ATTR_EN_0_0_NNNW							0x00000008
+#define    NV50TCL_VP_ATTR_EN_0_0_XNNW							0x00000009
+#define    NV50TCL_VP_ATTR_EN_0_0_NYNW							0x0000000a
+#define    NV50TCL_VP_ATTR_EN_0_0_XYNW							0x0000000b
+#define    NV50TCL_VP_ATTR_EN_0_0_NNZW							0x0000000c
+#define    NV50TCL_VP_ATTR_EN_0_0_XNZW							0x0000000d
+#define    NV50TCL_VP_ATTR_EN_0_0_NYZW							0x0000000e
+#define    NV50TCL_VP_ATTR_EN_0_0_XYZW							0x0000000f
+#define  NV50TCL_VP_ATTR_EN_1								0x00001654
+#define   NV50TCL_VP_ATTR_EN_1_15_SHIFT							28
+#define   NV50TCL_VP_ATTR_EN_1_15_MASK							0xf0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XNNN							0x10000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NYNN							0x20000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XYNN							0x30000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NNZN							0x40000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XNZN							0x50000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NYZN							0x60000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XYZN							0x70000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NNNW							0x80000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XNNW							0x90000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NYNW							0xa0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XYNW							0xb0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NNZW							0xc0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XNZW							0xd0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_NYZW							0xe0000000
+#define    NV50TCL_VP_ATTR_EN_1_15_XYZW							0xf0000000
+#define   NV50TCL_VP_ATTR_EN_1_14_SHIFT							24
+#define   NV50TCL_VP_ATTR_EN_1_14_MASK							0x0f000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XNNN							0x01000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NYNN							0x02000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XYNN							0x03000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NNZN							0x04000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XNZN							0x05000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NYZN							0x06000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XYZN							0x07000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NNNW							0x08000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XNNW							0x09000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NYNW							0x0a000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XYNW							0x0b000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NNZW							0x0c000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XNZW							0x0d000000
+#define    NV50TCL_VP_ATTR_EN_1_14_NYZW							0x0e000000
+#define    NV50TCL_VP_ATTR_EN_1_14_XYZW							0x0f000000
+#define   NV50TCL_VP_ATTR_EN_1_13_SHIFT							20
+#define   NV50TCL_VP_ATTR_EN_1_13_MASK							0x00f00000
+#define    NV50TCL_VP_ATTR_EN_1_13_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_13_XNNN							0x00100000
+#define    NV50TCL_VP_ATTR_EN_1_13_NYNN							0x00200000
+#define    NV50TCL_VP_ATTR_EN_1_13_XYNN							0x00300000
+#define    NV50TCL_VP_ATTR_EN_1_13_NNZN							0x00400000
+#define    NV50TCL_VP_ATTR_EN_1_13_XNZN							0x00500000
+#define    NV50TCL_VP_ATTR_EN_1_13_NYZN							0x00600000
+#define    NV50TCL_VP_ATTR_EN_1_13_XYZN							0x00700000
+#define    NV50TCL_VP_ATTR_EN_1_13_NNNW							0x00800000
+#define    NV50TCL_VP_ATTR_EN_1_13_XNNW							0x00900000
+#define    NV50TCL_VP_ATTR_EN_1_13_NYNW							0x00a00000
+#define    NV50TCL_VP_ATTR_EN_1_13_XYNW							0x00b00000
+#define    NV50TCL_VP_ATTR_EN_1_13_NNZW							0x00c00000
+#define    NV50TCL_VP_ATTR_EN_1_13_XNZW							0x00d00000
+#define    NV50TCL_VP_ATTR_EN_1_13_NYZW							0x00e00000
+#define    NV50TCL_VP_ATTR_EN_1_13_XYZW							0x00f00000
+#define   NV50TCL_VP_ATTR_EN_1_12_SHIFT							16
+#define   NV50TCL_VP_ATTR_EN_1_12_MASK							0x000f0000
+#define    NV50TCL_VP_ATTR_EN_1_12_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_12_XNNN							0x00010000
+#define    NV50TCL_VP_ATTR_EN_1_12_NYNN							0x00020000
+#define    NV50TCL_VP_ATTR_EN_1_12_XYNN							0x00030000
+#define    NV50TCL_VP_ATTR_EN_1_12_NNZN							0x00040000
+#define    NV50TCL_VP_ATTR_EN_1_12_XNZN							0x00050000
+#define    NV50TCL_VP_ATTR_EN_1_12_NYZN							0x00060000
+#define    NV50TCL_VP_ATTR_EN_1_12_XYZN							0x00070000
+#define    NV50TCL_VP_ATTR_EN_1_12_NNNW							0x00080000
+#define    NV50TCL_VP_ATTR_EN_1_12_XNNW							0x00090000
+#define    NV50TCL_VP_ATTR_EN_1_12_NYNW							0x000a0000
+#define    NV50TCL_VP_ATTR_EN_1_12_XYNW							0x000b0000
+#define    NV50TCL_VP_ATTR_EN_1_12_NNZW							0x000c0000
+#define    NV50TCL_VP_ATTR_EN_1_12_XNZW							0x000d0000
+#define    NV50TCL_VP_ATTR_EN_1_12_NYZW							0x000e0000
+#define    NV50TCL_VP_ATTR_EN_1_12_XYZW							0x000f0000
+#define   NV50TCL_VP_ATTR_EN_1_11_SHIFT							12
+#define   NV50TCL_VP_ATTR_EN_1_11_MASK							0x0000f000
+#define    NV50TCL_VP_ATTR_EN_1_11_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_11_XNNN							0x00001000
+#define    NV50TCL_VP_ATTR_EN_1_11_NYNN							0x00002000
+#define    NV50TCL_VP_ATTR_EN_1_11_XYNN							0x00003000
+#define    NV50TCL_VP_ATTR_EN_1_11_NNZN							0x00004000
+#define    NV50TCL_VP_ATTR_EN_1_11_XNZN							0x00005000
+#define    NV50TCL_VP_ATTR_EN_1_11_NYZN							0x00006000
+#define    NV50TCL_VP_ATTR_EN_1_11_XYZN							0x00007000
+#define    NV50TCL_VP_ATTR_EN_1_11_NNNW							0x00008000
+#define    NV50TCL_VP_ATTR_EN_1_11_XNNW							0x00009000
+#define    NV50TCL_VP_ATTR_EN_1_11_NYNW							0x0000a000
+#define    NV50TCL_VP_ATTR_EN_1_11_XYNW							0x0000b000
+#define    NV50TCL_VP_ATTR_EN_1_11_NNZW							0x0000c000
+#define    NV50TCL_VP_ATTR_EN_1_11_XNZW							0x0000d000
+#define    NV50TCL_VP_ATTR_EN_1_11_NYZW							0x0000e000
+#define    NV50TCL_VP_ATTR_EN_1_11_XYZW							0x0000f000
+#define   NV50TCL_VP_ATTR_EN_1_10_SHIFT							8
+#define   NV50TCL_VP_ATTR_EN_1_10_MASK							0x00000f00
+#define    NV50TCL_VP_ATTR_EN_1_10_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_10_XNNN							0x00000100
+#define    NV50TCL_VP_ATTR_EN_1_10_NYNN							0x00000200
+#define    NV50TCL_VP_ATTR_EN_1_10_XYNN							0x00000300
+#define    NV50TCL_VP_ATTR_EN_1_10_NNZN							0x00000400
+#define    NV50TCL_VP_ATTR_EN_1_10_XNZN							0x00000500
+#define    NV50TCL_VP_ATTR_EN_1_10_NYZN							0x00000600
+#define    NV50TCL_VP_ATTR_EN_1_10_XYZN							0x00000700
+#define    NV50TCL_VP_ATTR_EN_1_10_NNNW							0x00000800
+#define    NV50TCL_VP_ATTR_EN_1_10_XNNW							0x00000900
+#define    NV50TCL_VP_ATTR_EN_1_10_NYNW							0x00000a00
+#define    NV50TCL_VP_ATTR_EN_1_10_XYNW							0x00000b00
+#define    NV50TCL_VP_ATTR_EN_1_10_NNZW							0x00000c00
+#define    NV50TCL_VP_ATTR_EN_1_10_XNZW							0x00000d00
+#define    NV50TCL_VP_ATTR_EN_1_10_NYZW							0x00000e00
+#define    NV50TCL_VP_ATTR_EN_1_10_XYZW							0x00000f00
+#define   NV50TCL_VP_ATTR_EN_1_9_SHIFT							4
+#define   NV50TCL_VP_ATTR_EN_1_9_MASK							0x000000f0
+#define    NV50TCL_VP_ATTR_EN_1_9_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_9_XNNN							0x00000010
+#define    NV50TCL_VP_ATTR_EN_1_9_NYNN							0x00000020
+#define    NV50TCL_VP_ATTR_EN_1_9_XYNN							0x00000030
+#define    NV50TCL_VP_ATTR_EN_1_9_NNZN							0x00000040
+#define    NV50TCL_VP_ATTR_EN_1_9_XNZN							0x00000050
+#define    NV50TCL_VP_ATTR_EN_1_9_NYZN							0x00000060
+#define    NV50TCL_VP_ATTR_EN_1_9_XYZN							0x00000070
+#define    NV50TCL_VP_ATTR_EN_1_9_NNNW							0x00000080
+#define    NV50TCL_VP_ATTR_EN_1_9_XNNW							0x00000090
+#define    NV50TCL_VP_ATTR_EN_1_9_NYNW							0x000000a0
+#define    NV50TCL_VP_ATTR_EN_1_9_XYNW							0x000000b0
+#define    NV50TCL_VP_ATTR_EN_1_9_NNZW							0x000000c0
+#define    NV50TCL_VP_ATTR_EN_1_9_XNZW							0x000000d0
+#define    NV50TCL_VP_ATTR_EN_1_9_NYZW							0x000000e0
+#define    NV50TCL_VP_ATTR_EN_1_9_XYZW							0x000000f0
+#define   NV50TCL_VP_ATTR_EN_1_8_SHIFT							0
+#define   NV50TCL_VP_ATTR_EN_1_8_MASK							0x0000000f
+#define    NV50TCL_VP_ATTR_EN_1_8_NONE							0x00000000
+#define    NV50TCL_VP_ATTR_EN_1_8_XNNN							0x00000001
+#define    NV50TCL_VP_ATTR_EN_1_8_NYNN							0x00000002
+#define    NV50TCL_VP_ATTR_EN_1_8_XYNN							0x00000003
+#define    NV50TCL_VP_ATTR_EN_1_8_NNZN							0x00000004
+#define    NV50TCL_VP_ATTR_EN_1_8_XNZN							0x00000005
+#define    NV50TCL_VP_ATTR_EN_1_8_NYZN							0x00000006
+#define    NV50TCL_VP_ATTR_EN_1_8_XYZN							0x00000007
+#define    NV50TCL_VP_ATTR_EN_1_8_NNNW							0x00000008
+#define    NV50TCL_VP_ATTR_EN_1_8_XNNW							0x00000009
+#define    NV50TCL_VP_ATTR_EN_1_8_NYNW							0x0000000a
+#define    NV50TCL_VP_ATTR_EN_1_8_XYNW							0x0000000b
+#define    NV50TCL_VP_ATTR_EN_1_8_NNZW							0x0000000c
+#define    NV50TCL_VP_ATTR_EN_1_8_XNZW							0x0000000d
+#define    NV50TCL_VP_ATTR_EN_1_8_NYZW							0x0000000e
+#define    NV50TCL_VP_ATTR_EN_1_8_XYZW							0x0000000f
+#define  NV50TCL_LINE_STIPPLE_ENABLE							0x0000166c
+#define  NV50TCL_LINE_STIPPLE_PATTERN							0x00001680
+#define  NV50TCL_POLYGON_STIPPLE_ENABLE							0x0000168c
+#define  NV50TCL_VP_REG_HPOS								0x000016bc
+#define   NV50TCL_VP_REG_HPOS_X_SHIFT							0
+#define   NV50TCL_VP_REG_HPOS_X_MASK							0x000000ff
+#define   NV50TCL_VP_REG_HPOS_Y_SHIFT							8
+#define   NV50TCL_VP_REG_HPOS_Y_MASK							0x0000ff00
+#define   NV50TCL_VP_REG_HPOS_Z_SHIFT							16
+#define   NV50TCL_VP_REG_HPOS_Z_MASK							0x00ff0000
+#define   NV50TCL_VP_REG_HPOS_W_SHIFT							24
+#define   NV50TCL_VP_REG_HPOS_W_MASK							0xff000000
+#define  NV50TCL_VP_REG_COL0								0x000016c0
+#define   NV50TCL_VP_REG_COL0_X_SHIFT							0
+#define   NV50TCL_VP_REG_COL0_X_MASK							0x000000ff
+#define   NV50TCL_VP_REG_COL0_Y_SHIFT							8
+#define   NV50TCL_VP_REG_COL0_Y_MASK							0x0000ff00
+#define   NV50TCL_VP_REG_COL0_Z_SHIFT							16
+#define   NV50TCL_VP_REG_COL0_Z_MASK							0x00ff0000
+#define   NV50TCL_VP_REG_COL0_W_SHIFT							24
+#define   NV50TCL_VP_REG_COL0_W_MASK							0xff000000
+#define  NV50TCL_POLYGON_STIPPLE_PATTERN(x)						(0x00001700+((x)*4))
+#define  NV50TCL_POLYGON_STIPPLE_PATTERN__SIZE						0x00000020
+#define  NV50TCL_CULL_FACE_ENABLE							0x00001918
+#define  NV50TCL_FRONT_FACE								0x0000191c
+#define   NV50TCL_FRONT_FACE_CW								0x00000900
+#define   NV50TCL_FRONT_FACE_CCW							0x00000901
+#define  NV50TCL_CULL_FACE								0x00001920
+#define   NV50TCL_CULL_FACE_FRONT							0x00000404
+#define   NV50TCL_CULL_FACE_BACK							0x00000405
+#define   NV50TCL_CULL_FACE_FRONT_AND_BACK						0x00000408
+#define  NV50TCL_LOGIC_OP_ENABLE							0x000019c4
+#define  NV50TCL_LOGIC_OP								0x000019c8
+#define   NV50TCL_LOGIC_OP_CLEAR							0x00001500
+#define   NV50TCL_LOGIC_OP_AND								0x00001501
+#define   NV50TCL_LOGIC_OP_AND_REVERSE							0x00001502
+#define   NV50TCL_LOGIC_OP_COPY								0x00001503
+#define   NV50TCL_LOGIC_OP_AND_INVERTED							0x00001504
+#define   NV50TCL_LOGIC_OP_NOOP								0x00001505
+#define   NV50TCL_LOGIC_OP_XOR								0x00001506
+#define   NV50TCL_LOGIC_OP_OR								0x00001507
+#define   NV50TCL_LOGIC_OP_NOR								0x00001508
+#define   NV50TCL_LOGIC_OP_EQUIV							0x00001509
+#define   NV50TCL_LOGIC_OP_INVERT							0x0000150a
+#define   NV50TCL_LOGIC_OP_OR_REVERSE							0x0000150b
+#define   NV50TCL_LOGIC_OP_COPY_INVERTED						0x0000150c
+#define   NV50TCL_LOGIC_OP_OR_INVERTED							0x0000150d
+#define   NV50TCL_LOGIC_OP_NAND								0x0000150e
+#define   NV50TCL_LOGIC_OP_SET								0x0000150f
+#define  NV50TCL_CLEAR_BUFFERS								0x000019d0
+#define  NV50TCL_COLOR_MASK(x)								(0x00001a00+((x)*4))
+#define  NV50TCL_COLOR_MASK__SIZE							0x00000008
+#define   NV50TCL_COLOR_MASK_R_SHIFT							0
+#define   NV50TCL_COLOR_MASK_R_MASK							0x0000000f
+#define   NV50TCL_COLOR_MASK_G_SHIFT							4
+#define   NV50TCL_COLOR_MASK_G_MASK							0x000000f0
+#define   NV50TCL_COLOR_MASK_B_SHIFT							8
+#define   NV50TCL_COLOR_MASK_B_MASK							0x00000f00
+#define   NV50TCL_COLOR_MASK_A_SHIFT							12
+#define   NV50TCL_COLOR_MASK_A_MASK							0x0000f000
+
+
+#define NV50_COMPUTE									0x000050c0
+
+#define  NV50_COMPUTE_DMA_UNK0								0x000001a0
+#define  NV50_COMPUTE_DMA_STATUS							0x000001a4
+#define  NV50_COMPUTE_DMA_UNK1								0x000001b8
+#define  NV50_COMPUTE_DMA_UNK2								0x000001bc
+#define  NV50_COMPUTE_DMA_UNK3								0x000001c0
+#define  NV50_COMPUTE_UNK4_HIGH								0x00000210
+#define  NV50_COMPUTE_UNK4_LOW								0x00000214
+#define  NV50_COMPUTE_UNK5_HIGH								0x00000218
+#define  NV50_COMPUTE_UNK5_LOW								0x0000021c
+#define  NV50_COMPUTE_UNK6_HIGH								0x00000294
+#define  NV50_COMPUTE_UNK6_LOW								0x00000298
+#define  NV50_COMPUTE_CONST_BASE_HIGH							0x000002a4
+#define  NV50_COMPUTE_CONST_BASE_LO							0x000002a8
+#define  NV50_COMPUTE_CONST_SIZE_SEG							0x000002ac
+#define  NV50_COMPUTE_REG_COUNT								0x000002c0
+#define  NV50_COMPUTE_STATUS_HIGH							0x00000310
+#define  NV50_COMPUTE_STATUS_LOW							0x00000314
+#define  NV50_COMPUTE_EXECUTE								0x0000031c
+#define  NV50_COMPUTE_USER_PARAM_COUNT							0x00000374
+#define  NV50_COMPUTE_GRIDDIM_YX							0x000003a4
+#define  NV50_COMPUTE_SHARED_SIZE							0x000003a8
+#define  NV50_COMPUTE_BLOCKDIM_YX							0x000003ac
+#define  NV50_COMPUTE_BLOCKDIM_Z							0x000003b0
+#define  NV50_COMPUTE_CALL_ADDRESS							0x000003b4
+#define  NV50_COMPUTE_GLOBAL_BASE_HIGH(x)						(0x00000400+((x)*32))
+#define  NV50_COMPUTE_GLOBAL_BASE_HIGH__SIZE						0x00000010
+#define  NV50_COMPUTE_GLOBAL_BASE_LOW(x)						(0x00000404+((x)*32))
+#define  NV50_COMPUTE_GLOBAL_BASE_LOW__SIZE						0x00000010
+#define  NV50_COMPUTE_GLOBAL_LIMIT_HIGH(x)						(0x00000408+((x)*32))
+#define  NV50_COMPUTE_GLOBAL_LIMIT_HIGH__SIZE						0x00000010
+#define  NV50_COMPUTE_GLOBAL_LIMIT_LOW(x)						(0x0000040c+((x)*32))
+#define  NV50_COMPUTE_GLOBAL_LIMIT_LOW__SIZE						0x00000010
+#define  NV50_COMPUTE_GLOBAL_UNK(x)							(0x00000410+((x)*32))
+#define  NV50_COMPUTE_GLOBAL_UNK__SIZE							0x00000010
+#define  NV50_COMPUTE_USER_PARAM(x)							(0x00000600+((x)*4))
+#define  NV50_COMPUTE_USER_PARAM__SIZE							0x00000040
+
+
+#define NV54TCL										0x00008297
+
+
+
+#endif /* NOUVEAU_REG_H */
diff --git a/src/gallium/drivers/nouveau/nouveau_device.h b/src/gallium/drivers/nouveau/nouveau_device.h
new file mode 100644
index 00000000000..e25e89fedda
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_device.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_DEVICE_H__
+#define __NOUVEAU_DEVICE_H__
+
+struct nouveau_device {
+	unsigned chipset;
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_gldefs.h b/src/gallium/drivers/nouveau/nouveau_gldefs.h
new file mode 100644
index 00000000000..ff97aaa9af0
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_gldefs.h
@@ -0,0 +1,196 @@
+#ifndef __NOUVEAU_GLDEFS_H__
+#define __NOUVEAU_GLDEFS_H__
+
+static INLINE unsigned
+nvgl_blend_func(unsigned factor)
+{
+	switch (factor) {
+	case PIPE_BLENDFACTOR_ZERO:
+		return 0x0000;
+	case PIPE_BLENDFACTOR_ONE:
+		return 0x0001;
+	case PIPE_BLENDFACTOR_SRC_COLOR:
+		return 0x0300;
+	case PIPE_BLENDFACTOR_INV_SRC_COLOR:
+		return 0x0301;
+	case PIPE_BLENDFACTOR_SRC_ALPHA:
+		return 0x0302;
+	case PIPE_BLENDFACTOR_INV_SRC_ALPHA:
+		return 0x0303;
+	case PIPE_BLENDFACTOR_DST_ALPHA:
+		return 0x0304;
+	case PIPE_BLENDFACTOR_INV_DST_ALPHA:
+		return 0x0305;
+	case PIPE_BLENDFACTOR_DST_COLOR:
+		return 0x0306;
+	case PIPE_BLENDFACTOR_INV_DST_COLOR:
+		return 0x0307;
+	case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
+		return 0x0308;
+	case PIPE_BLENDFACTOR_CONST_COLOR:
+		return 0x8001;
+	case PIPE_BLENDFACTOR_INV_CONST_COLOR:
+		return 0x8002;
+	case PIPE_BLENDFACTOR_CONST_ALPHA:
+		return 0x8003;
+	case PIPE_BLENDFACTOR_INV_CONST_ALPHA:
+		return 0x8004;
+	default:
+		return 0x0000;
+	}
+}
+
+static INLINE unsigned
+nvgl_blend_eqn(unsigned func)
+{
+	switch (func) {
+	case PIPE_BLEND_ADD:
+		return 0x8006;
+	case PIPE_BLEND_MIN:
+		return 0x8007;
+	case PIPE_BLEND_MAX:
+		return 0x8008;
+	case PIPE_BLEND_SUBTRACT:
+		return 0x800a;
+	case PIPE_BLEND_REVERSE_SUBTRACT:
+		return 0x800b;
+	default:
+		return 0x8006;
+	}
+}
+
+static INLINE unsigned
+nvgl_logicop_func(unsigned func)
+{
+	switch (func) {
+	case PIPE_LOGICOP_CLEAR:
+		return 0x1500;
+	case PIPE_LOGICOP_NOR:
+		return 0x1508;
+	case PIPE_LOGICOP_AND_INVERTED:
+		return 0x1504;
+	case PIPE_LOGICOP_COPY_INVERTED:
+		return 0x150c;
+	case PIPE_LOGICOP_AND_REVERSE:
+		return 0x1502;
+	case PIPE_LOGICOP_INVERT:
+		return 0x150a;
+	case PIPE_LOGICOP_XOR:
+		return 0x1506;
+	case PIPE_LOGICOP_NAND:
+		return 0x150e;
+	case PIPE_LOGICOP_AND:
+		return 0x1501;
+	case PIPE_LOGICOP_EQUIV:
+		return 0x1509;
+	case PIPE_LOGICOP_NOOP:
+		return 0x1505;
+	case PIPE_LOGICOP_OR_INVERTED:
+		return 0x150d;
+	case PIPE_LOGICOP_COPY:
+		return 0x1503;
+	case PIPE_LOGICOP_OR_REVERSE:
+		return 0x150b;
+	case PIPE_LOGICOP_OR:
+		return 0x1507;
+	case PIPE_LOGICOP_SET:
+		return 0x150f;
+	default:
+		return 0x1505;
+	}
+}
+
+static INLINE unsigned
+nvgl_comparison_op(unsigned op)
+{
+	switch (op) {
+	case PIPE_FUNC_NEVER:
+		return 0x0200;
+	case PIPE_FUNC_LESS:
+		return 0x0201;
+	case PIPE_FUNC_EQUAL:
+		return 0x0202;
+	case PIPE_FUNC_LEQUAL:
+		return 0x0203;
+	case PIPE_FUNC_GREATER:
+		return 0x0204;
+	case PIPE_FUNC_NOTEQUAL:
+		return 0x0205;
+	case PIPE_FUNC_GEQUAL:
+		return 0x0206;
+	case PIPE_FUNC_ALWAYS:
+		return 0x0207;
+	default:
+		return 0x0207;
+	}
+}
+
+static INLINE unsigned
+nvgl_polygon_mode(unsigned mode)
+{
+	switch (mode) {
+	case PIPE_POLYGON_MODE_POINT:
+		return 0x1b00;
+	case PIPE_POLYGON_MODE_LINE:
+		return 0x1b01;
+	case PIPE_POLYGON_MODE_FILL:
+		return 0x1b02;
+	default:
+		return 0x1b02;
+	}
+}
+
+static INLINE unsigned
+nvgl_stencil_op(unsigned op)
+{
+	switch (op) {
+	case PIPE_STENCIL_OP_ZERO:
+		return 0x0000;
+	case PIPE_STENCIL_OP_INVERT:
+		return 0x150a;
+	case PIPE_STENCIL_OP_KEEP:
+		return 0x1e00;
+	case PIPE_STENCIL_OP_REPLACE:
+		return 0x1e01;
+	case PIPE_STENCIL_OP_INCR:
+		return 0x1e02;
+	case PIPE_STENCIL_OP_DECR:
+		return 0x1e03;
+	case PIPE_STENCIL_OP_INCR_WRAP:
+		return 0x8507;
+	case PIPE_STENCIL_OP_DECR_WRAP:
+		return 0x8508;
+	default:
+		return 0x1e00;
+	}
+}
+
+static INLINE unsigned
+nvgl_primitive(unsigned prim) {
+	switch (prim) {
+	case PIPE_PRIM_POINTS:
+		return 0x0001;
+	case PIPE_PRIM_LINES:
+		return 0x0002;
+	case PIPE_PRIM_LINE_LOOP:
+		return 0x0003;
+	case PIPE_PRIM_LINE_STRIP:
+		return 0x0004;
+	case PIPE_PRIM_TRIANGLES:
+		return 0x0005;
+	case PIPE_PRIM_TRIANGLE_STRIP:
+		return 0x0006;
+	case PIPE_PRIM_TRIANGLE_FAN:
+		return 0x0007;
+	case PIPE_PRIM_QUADS:
+		return 0x0008;
+	case PIPE_PRIM_QUAD_STRIP:
+		return 0x0009;
+	case PIPE_PRIM_POLYGON:
+		return 0x000a;
+	default:
+		return 0;
+	}
+}
+
+#endif
diff --git a/src/mesa/swrast/s_drawpix.h b/src/gallium/drivers/nouveau/nouveau_grobj.h
index 7882a799663..8f5abf90514 100644
--- a/src/mesa/swrast/s_drawpix.h
+++ b/src/gallium/drivers/nouveau/nouveau_grobj.h
@@ -1,9 +1,5 @@
-
 /*
- * Mesa 3-D graphics library
- * Version:  3.5
- *
- * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ * Copyright 2007 Nouveau Project
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -12,25 +8,28 @@
  * and/or sell copies of the Software, and to permit persons to whom the
  * Software is furnished to do so, subject to the following conditions:
  *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
  *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
  */
 
+#ifndef __NOUVEAU_GROBJ_H__
+#define __NOUVEAU_GROBJ_H__
 
-#ifndef S_DRAWPIXELS_H
-#define S_DRAWPIXELS_H
-
-
-#include "main/mtypes.h"
-#include "swrast.h"
+#include "nouveau_channel.h"
 
-/* XXX kill this header? */
+struct nouveau_grobj {
+	struct nouveau_channel *channel;
+	int grclass;
+	uint32_t handle;
+	int subc;
+};
 
 #endif
diff --git a/src/gallium/drivers/nouveau/nouveau_notifier.h b/src/gallium/drivers/nouveau/nouveau_notifier.h
new file mode 100644
index 00000000000..35adde1e324
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_notifier.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_NOTIFIER_H__
+#define __NOUVEAU_NOTIFIER_H__
+
+#define NV_NOTIFIER_SIZE                                                      32
+#define NV_NOTIFY_TIME_0                                              0x00000000
+#define NV_NOTIFY_TIME_1                                              0x00000004
+#define NV_NOTIFY_RETURN_VALUE                                        0x00000008
+#define NV_NOTIFY_STATE                                               0x0000000C
+#define NV_NOTIFY_STATE_STATUS_MASK                                   0xFF000000
+#define NV_NOTIFY_STATE_STATUS_SHIFT                                          24
+#define NV_NOTIFY_STATE_STATUS_COMPLETED                                    0x00
+#define NV_NOTIFY_STATE_STATUS_IN_PROCESS                                   0x01
+#define NV_NOTIFY_STATE_ERROR_CODE_MASK                               0x0000FFFF
+#define NV_NOTIFY_STATE_ERROR_CODE_SHIFT                                       0
+
+struct nouveau_notifier {
+	struct nouveau_channel *channel;
+	uint32_t handle;
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_push.h b/src/gallium/drivers/nouveau/nouveau_push.h
new file mode 100644
index 00000000000..54ef1c1291e
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_push.h
@@ -0,0 +1,82 @@
+#ifndef __NOUVEAU_PUSH_H__
+#define __NOUVEAU_PUSH_H__
+
+#include "nouveau/nouveau_winsys.h"
+
+#ifndef NOUVEAU_PUSH_CONTEXT
+#error undefined push context
+#endif
+
+#define OUT_RING(data) do {                                                    \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	(*pc->nvws->channel->pushbuf->cur++) = (data);                         \
+} while(0)
+
+#define OUT_RINGp(src,size) do {                                               \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	memcpy(pc->nvws->channel->pushbuf->cur, (src), (size) * 4);            \
+	pc->nvws->channel->pushbuf->cur += (size);                             \
+} while(0)
+
+#define OUT_RINGf(data) do {                                                   \
+	union { float v; uint32_t u; } c;                                      \
+	c.v = (data);                                                          \
+	OUT_RING(c.u);                                                         \
+} while(0)
+
+#define BEGIN_RING(obj,mthd,size) do {                                         \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	if (pc->nvws->channel->pushbuf->remaining < ((size) + 1))              \
+		pc->nvws->push_flush(pc->nvws, ((size) + 1), NULL);            \
+	OUT_RING((pc->obj->subc << 13) | ((size) << 18) | (mthd));             \
+	pc->nvws->channel->pushbuf->remaining -= ((size) + 1);                 \
+} while(0)
+
+#define BEGIN_RING_NI(obj,mthd,size) do {                                      \
+	BEGIN_RING(obj, (mthd) | 0x40000000, (size));                          \
+} while(0)
+
+#define FIRE_RING(fence) do {                                                  \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	pc->nvws->push_flush(pc->nvws, 0, fence);                              \
+} while(0)
+
+#define OUT_RELOC(bo,data,flags,vor,tor) do {                                  \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	pc->nvws->push_reloc(pc->nvws, pc->nvws->channel->pushbuf->cur++,      \
+			     (bo), (data), (flags), (vor), (tor));             \
+} while(0)
+
+/* Raw data + flags depending on FB/TT buffer */
+#define OUT_RELOCd(bo,data,flags,vor,tor) do {                                 \
+	OUT_RELOC((bo), (data), (flags) | NOUVEAU_BO_OR, (vor), (tor));        \
+} while(0)
+
+/* FB/TT object handle */
+#define OUT_RELOCo(bo,flags) do {                                              \
+	OUT_RELOC((bo), 0, (flags) | NOUVEAU_BO_OR,                            \
+		  pc->nvws->channel->vram->handle,                             \
+		  pc->nvws->channel->gart->handle);                            \
+} while(0)
+
+/* Low 32-bits of offset */
+#define OUT_RELOCl(bo,delta,flags) do {                                        \
+	OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_LOW, 0, 0);              \
+} while(0)
+
+/* High 32-bits of offset */
+#define OUT_RELOCh(bo,delta,flags) do {                                        \
+	OUT_RELOC((bo), (delta), (flags) | NOUVEAU_BO_HIGH, 0, 0);             \
+} while(0)
+
+/* A reloc which'll recombine into a NV_DMA_METHOD packet header */
+#define OUT_RELOCm(bo, flags, obj, mthd, size) do {                            \
+	NOUVEAU_PUSH_CONTEXT(pc);                                              \
+	if (pc->nvws->channel->pushbuf->remaining < ((size) + 1))              \
+		pc->nvws->push_flush(pc->nvws->channel, ((size) + 1), NULL);   \
+	OUT_RELOCd((bo), (pc->obj->subc << 13) | ((size) << 18) | (mthd),      \
+		   (flags), 0, 0);                                             \
+	pc->nvws->channel->pushbuf->remaining -= ((size) + 1);                 \
+} while(0)
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_pushbuf.h b/src/gallium/drivers/nouveau/nouveau_pushbuf.h
new file mode 100644
index 00000000000..19097650982
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_pushbuf.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_PUSHBUF_H__
+#define __NOUVEAU_PUSHBUF_H__
+
+struct nouveau_pushbuf {
+	struct nouveau_channel *channel;
+	unsigned remaining;
+	uint32_t *cur;
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_resource.h b/src/gallium/drivers/nouveau/nouveau_resource.h
new file mode 100644
index 00000000000..1af7961d301
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_resource.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_RESOURCE_H__
+#define __NOUVEAU_RESOURCE_H__
+
+struct nouveau_resource {
+	struct nouveau_resource *prev;
+	struct nouveau_resource *next;
+
+	int in_use;
+	void *priv;
+
+	unsigned int start;
+	unsigned int size;
+};
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_stateobj.h b/src/gallium/drivers/nouveau/nouveau_stateobj.h
new file mode 100644
index 00000000000..729988b095e
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_stateobj.h
@@ -0,0 +1,158 @@
+#ifndef __NOUVEAU_STATEOBJ_H__
+#define __NOUVEAU_STATEOBJ_H__
+
+#include "pipe/p_debug.h"
+
+struct nouveau_stateobj_reloc {
+	struct pipe_buffer *bo;
+
+	unsigned offset;
+	unsigned packet;
+
+	unsigned data;
+	unsigned flags;
+	unsigned vor;
+	unsigned tor;
+};
+
+struct nouveau_stateobj {
+	int refcount;
+
+	unsigned *push;
+	struct nouveau_stateobj_reloc *reloc;
+
+	unsigned *cur;
+	unsigned cur_packet;
+	unsigned cur_reloc;
+};
+
+static INLINE struct nouveau_stateobj *
+so_new(unsigned push, unsigned reloc)
+{
+	struct nouveau_stateobj *so;
+
+	so = MALLOC(sizeof(struct nouveau_stateobj));
+	so->refcount = 0;
+	so->push = MALLOC(sizeof(unsigned) * push);
+	so->reloc = MALLOC(sizeof(struct nouveau_stateobj_reloc) * reloc);
+
+	so->cur = so->push;
+	so->cur_reloc = so->cur_packet = 0;
+
+	return so;
+}
+
+static INLINE void
+so_ref(struct nouveau_stateobj *ref, struct nouveau_stateobj **pso)
+{
+	struct nouveau_stateobj *so = *pso;
+
+	if (ref) {
+		ref->refcount++;
+	}
+
+	if (so && --so->refcount <= 0) {
+		free(so->push);
+		free(so->reloc);
+		free(so);
+	}
+
+	*pso = ref;
+}
+
+static INLINE void
+so_data(struct nouveau_stateobj *so, unsigned data)
+{
+	(*so->cur++) = (data);
+	so->cur_packet += 4;
+}
+
+static INLINE void
+so_datap(struct nouveau_stateobj *so, unsigned *data, unsigned size)
+{
+	so->cur_packet += (4 * size);
+	while (size--)
+		(*so->cur++) = (*data++);
+}
+
+static INLINE void
+so_method(struct nouveau_stateobj *so, struct nouveau_grobj *gr,
+	  unsigned mthd, unsigned size)
+{
+	so->cur_packet = (gr->subc << 13) | (1 << 18) | (mthd - 4);
+	so_data(so, (gr->subc << 13) | (size << 18) | mthd);
+}
+
+static INLINE void
+so_reloc(struct nouveau_stateobj *so, struct pipe_buffer *bo,
+	 unsigned data, unsigned flags, unsigned vor, unsigned tor)
+{
+	struct nouveau_stateobj_reloc *r = &so->reloc[so->cur_reloc++];
+	
+	r->bo = bo;
+	r->offset = so->cur - so->push;
+	r->packet = so->cur_packet;
+	r->data = data;
+	r->flags = flags;
+	r->vor = vor;
+	r->tor = tor;
+	so_data(so, data);
+}
+
+static INLINE void
+so_dump(struct nouveau_stateobj *so)
+{
+	unsigned i, nr = so->cur - so->push;
+
+	for (i = 0; i < nr; i++)
+		debug_printf("+0x%04x: 0x%08x\n", i, so->push[i]);
+}
+
+static INLINE void
+so_emit(struct nouveau_winsys *nvws, struct nouveau_stateobj *so)
+{
+	struct nouveau_pushbuf *pb = nvws->channel->pushbuf;
+	unsigned nr, i;
+
+	nr = so->cur - so->push;
+	if (pb->remaining < nr)
+		nvws->push_flush(nvws, nr, NULL);
+	pb->remaining -= nr;
+
+	memcpy(pb->cur, so->push, nr * 4);
+	for (i = 0; i < so->cur_reloc; i++) {
+		struct nouveau_stateobj_reloc *r = &so->reloc[i];
+
+		nvws->push_reloc(nvws, pb->cur + r->offset, r->bo,
+				 r->data, r->flags, r->vor, r->tor);
+	}
+	pb->cur += nr;
+}
+
+static INLINE void
+so_emit_reloc_markers(struct nouveau_winsys *nvws, struct nouveau_stateobj *so)
+{
+	struct nouveau_pushbuf *pb = nvws->channel->pushbuf;
+	unsigned i;
+
+	if (!so)
+		return;
+
+	i = so->cur_reloc << 1;
+	if (nvws->channel->pushbuf->remaining < i)
+		nvws->push_flush(nvws, i, NULL);
+	nvws->channel->pushbuf->remaining -= i;
+
+	for (i = 0; i < so->cur_reloc; i++) {
+		struct nouveau_stateobj_reloc *r = &so->reloc[i];
+
+		nvws->push_reloc(nvws, pb->cur++, r->bo, r->packet,
+				 (r->flags &
+				  (NOUVEAU_BO_VRAM | NOUVEAU_BO_GART)) |
+				 NOUVEAU_BO_DUMMY, 0, 0);
+		nvws->push_reloc(nvws, pb->cur++, r->bo, r->data,
+				 r->flags | NOUVEAU_BO_DUMMY, r->vor, r->tor);
+	}
+}
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_util.h b/src/gallium/drivers/nouveau/nouveau_util.h
new file mode 100644
index 00000000000..a10114beab9
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_util.h
@@ -0,0 +1,91 @@
+#ifndef __NOUVEAU_UTIL_H__
+#define __NOUVEAU_UTIL_H__
+
+/* Determine how many vertices can be pushed into the command stream.
+ * Where the remaining space isn't large enough to represent all verices,
+ * split the buffer at primitive boundaries.
+ *
+ * Returns a count of vertices that can be rendered, and an index to
+ * restart drawing at after a flush.
+ */
+static INLINE unsigned
+nouveau_vbuf_split(unsigned remaining, unsigned overhead, unsigned vpp,
+		   unsigned mode, unsigned start, unsigned count,
+		   unsigned *restart)
+{
+	int max, adj = 0;
+
+	max  = remaining - overhead;
+	if (max < 0)
+		return 0;
+
+	max *= vpp;
+	if (max >= count)
+		return count;
+
+	switch (mode) {
+	case PIPE_PRIM_POINTS:
+		break;
+	case PIPE_PRIM_LINES:
+		max = max & 1;
+		break;
+	case PIPE_PRIM_TRIANGLES:
+		max = max - (max % 3);
+		break;
+	case PIPE_PRIM_QUADS:
+		max = max & 3;
+		break;
+	case PIPE_PRIM_LINE_LOOP:
+	case PIPE_PRIM_LINE_STRIP:
+		if (max < 2)
+			max = 0;
+		adj = 1;
+		break;
+	case PIPE_PRIM_POLYGON:
+	case PIPE_PRIM_TRIANGLE_STRIP:
+	case PIPE_PRIM_TRIANGLE_FAN:
+		if (max < 3)
+			max = 0;
+		adj = 2;
+		break;
+	case PIPE_PRIM_QUAD_STRIP:
+		if (max < 4)
+			max = 0;
+		adj = 3;
+		break;
+	default:
+		assert(0);
+	}
+
+	*restart = start + max - adj;
+	return max;
+}
+
+/* Integer base-2 logarithm, rounded towards zero. */
+static INLINE unsigned log2i(unsigned i)
+{
+	unsigned r = 0;
+
+	if (i & 0xffff0000) {
+		i >>= 16;
+		r += 16;
+	}
+	if (i & 0x0000ff00) {
+		i >>= 8;
+		r += 8;
+	}
+	if (i & 0x000000f0) {
+		i >>= 4;
+		r += 4;
+	}
+	if (i & 0x0000000c) {
+		i >>= 2;
+		r += 2;
+	}
+	if (i & 0x00000002) {
+		r += 1;
+	}
+	return r;
+}
+
+#endif
diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h
new file mode 100644
index 00000000000..5535ebb6a99
--- /dev/null
+++ b/src/gallium/drivers/nouveau/nouveau_winsys.h
@@ -0,0 +1,98 @@
+#ifndef NOUVEAU_WINSYS_H
+#define NOUVEAU_WINSYS_H
+
+#include <stdint.h>
+#include "pipe/p_winsys.h"
+#include "pipe/p_defines.h"
+
+#include "nouveau/nouveau_bo.h"
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_class.h"
+#include "nouveau/nouveau_device.h"
+#include "nouveau/nouveau_grobj.h"
+#include "nouveau/nouveau_notifier.h"
+#include "nouveau/nouveau_resource.h"
+#include "nouveau/nouveau_pushbuf.h"
+
+#define NOUVEAU_CAP_HW_VTXBUF (0xbeef0000)
+#define NOUVEAU_CAP_HW_IDXBUF (0xbeef0001)
+
+#define NOUVEAU_TEXTURE_USAGE_LINEAR (1 << 16)
+
+#define NOUVEAU_BUFFER_USAGE_TEXTURE (1 << 16)
+#define NOUVEAU_BUFFER_USAGE_ZETA    (1 << 17)
+
+struct nouveau_winsys {
+	struct nouveau_context *nv;
+
+	struct nouveau_channel *channel;
+
+	int  (*res_init)(struct nouveau_resource **heap, unsigned start,
+			 unsigned size);
+	int  (*res_alloc)(struct nouveau_resource *heap, int size, void *priv,
+			  struct nouveau_resource **);
+	void (*res_free)(struct nouveau_resource **);
+
+	int  (*push_reloc)(struct nouveau_winsys *, void *ptr,
+			   struct pipe_buffer *, uint32_t data,
+			   uint32_t flags, uint32_t vor, uint32_t tor);
+	int  (*push_flush)(struct nouveau_winsys *, unsigned size,
+			   struct pipe_fence_handle **fence);
+			       
+	int       (*grobj_alloc)(struct nouveau_winsys *, int grclass,
+				 struct nouveau_grobj **);
+	void      (*grobj_free)(struct nouveau_grobj **);
+
+	int       (*notifier_alloc)(struct nouveau_winsys *, int count,
+				    struct nouveau_notifier **);
+	void      (*notifier_free)(struct nouveau_notifier **);
+	void      (*notifier_reset)(struct nouveau_notifier *, int id);
+	uint32_t  (*notifier_status)(struct nouveau_notifier *, int id);
+	uint32_t  (*notifier_retval)(struct nouveau_notifier *, int id);
+	int       (*notifier_wait)(struct nouveau_notifier *, int id,
+				   int status, int timeout);
+
+	int (*surface_copy)(struct nouveau_winsys *, struct pipe_surface *,
+			    unsigned, unsigned, struct pipe_surface *,
+			    unsigned, unsigned, unsigned, unsigned);
+	int (*surface_fill)(struct nouveau_winsys *, struct pipe_surface *,
+			    unsigned, unsigned, unsigned, unsigned, unsigned);
+};
+
+extern struct pipe_screen *
+nv04_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv04_create(struct pipe_screen *, unsigned pctx_id);
+
+extern struct pipe_screen *
+nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv10_create(struct pipe_screen *, unsigned pctx_id);
+
+extern struct pipe_screen *
+nv20_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv20_create(struct pipe_screen *, unsigned pctx_id);
+
+extern struct pipe_screen *
+nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv30_create(struct pipe_screen *, unsigned pctx_id);
+
+extern struct pipe_screen *
+nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv40_create(struct pipe_screen *, unsigned pctx_id);
+
+extern struct pipe_screen *
+nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *);
+
+extern struct pipe_context *
+nv50_create(struct pipe_screen *, unsigned pctx_id);
+
+#endif
diff --git a/src/gallium/drivers/nv04/Makefile b/src/gallium/drivers/nv04/Makefile
new file mode 100644
index 00000000000..5ea51a2f420
--- /dev/null
+++ b/src/gallium/drivers/nv04/Makefile
@@ -0,0 +1,28 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv04
+
+DRIVER_SOURCES = \
+	nv04_clear.c \
+	nv04_context.c \
+	nv04_fragprog.c \
+	nv04_fragtex.c \
+	nv04_miptree.c \
+	nv04_prim_vbuf.c \
+	nv04_screen.c \
+	nv04_state.c \
+	nv04_state_emit.c \
+	nv04_surface.c \
+	nv04_vbo.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv04/nv04_clear.c b/src/gallium/drivers/nv04/nv04_clear.c
new file mode 100644
index 00000000000..01cacd36fe1
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_clear.c
@@ -0,0 +1,12 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv04_context.h"
+
+void
+nv04_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue);
+}
diff --git a/src/gallium/drivers/nv04/nv04_context.c b/src/gallium/drivers/nv04/nv04_context.c
new file mode 100644
index 00000000000..9f75253363f
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_context.c
@@ -0,0 +1,106 @@
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv04_context.h"
+#include "nv04_screen.h"
+
+static void
+nv04_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	draw_flush(nv04->draw);
+
+	FIRE_RING(fence);
+}
+
+static void
+nv04_destroy(struct pipe_context *pipe)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	if (nv04->draw)
+		draw_destroy(nv04->draw);
+
+	FREE(nv04);
+}
+
+static void
+nv04_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+}
+
+static boolean
+nv04_init_hwctx(struct nv04_context *nv04)
+{
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_NOTIFY, 1);
+	OUT_RING(0);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_NOP, 1);
+	OUT_RING(0);
+
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1);
+	OUT_RING(0x40182800);
+//	OUT_RING(1<<20/*no cull*/);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1);
+//	OUT_RING(0x24|(1<<6)|(1<<8));
+	OUT_RING(0x120001a4);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FORMAT, 1);
+	OUT_RING(0x332213a1);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FILTER, 1);
+	OUT_RING(0x11001010);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_COLORKEY, 1);
+	OUT_RING(0x0);
+//	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 1);
+//	OUT_RING(SCREEN_OFFSET);
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_FOGCOLOR, 1);
+	OUT_RING(0xff000000);
+
+
+
+	FIRE_RING (NULL);
+	return TRUE;
+}
+
+struct pipe_context *
+nv04_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct nv04_screen *screen = nv04_screen(pscreen);
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv04_context *nv04;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nv04 = CALLOC(1, sizeof(struct nv04_context));
+	if (!nv04)
+		return NULL;
+	nv04->screen = screen;
+	nv04->pctx_id = pctx_id;
+
+	nv04->nvws = nvws;
+
+	nv04->pipe.winsys = ws;
+	nv04->pipe.screen = pscreen;
+	nv04->pipe.destroy = nv04_destroy;
+	nv04->pipe.set_edgeflags = nv04_set_edgeflags;
+	nv04->pipe.draw_arrays = nv04_draw_arrays;
+	nv04->pipe.draw_elements = nv04_draw_elements;
+	nv04->pipe.clear = nv04_clear;
+	nv04->pipe.flush = nv04_flush;
+
+	nv04_init_surface_functions(nv04);
+	nv04_init_state_functions(nv04);
+
+	nv04->draw = draw_create();
+	assert(nv04->draw);
+	draw_wide_point_threshold(nv04->draw, 0.0);
+	draw_wide_line_threshold(nv04->draw, 0.0);
+	draw_enable_line_stipple(nv04->draw, FALSE);
+	draw_enable_point_sprites(nv04->draw, FALSE);
+	draw_set_rasterize_stage(nv04->draw, nv04_draw_vbuf_stage(nv04));
+
+	nv04_init_hwctx(nv04);
+
+	return &nv04->pipe;
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_context.h b/src/gallium/drivers/nv04/nv04_context.h
new file mode 100644
index 00000000000..3e6a0852702
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_context.h
@@ -0,0 +1,146 @@
+#ifndef __NV04_CONTEXT_H__
+#define __NV04_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv04_screen *ctx = nv04->screen
+#include "nouveau/nouveau_push.h"
+
+#include "nv04_state.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+#include "nv04_screen.h"
+
+#define NV04_NEW_VERTPROG	(1 << 1)
+#define NV04_NEW_FRAGPROG	(1 << 2)
+#define NV04_NEW_BLEND		(1 << 3)
+#define NV04_NEW_RAST		(1 << 4)
+#define NV04_NEW_CONTROL	(1 << 5)
+#define NV04_NEW_VIEWPORT	(1 << 6)
+#define NV04_NEW_SAMPLER	(1 << 7)
+
+struct nv04_context {
+	struct pipe_context pipe;
+
+	struct nouveau_winsys *nvws;
+	struct nv04_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	int chipset;
+	struct nouveau_notifier *sync;
+
+	uint32_t dirty;
+
+	struct nv04_blend_state *blend;
+	struct nv04_sampler_state *sampler[PIPE_MAX_SAMPLERS];
+	struct nv04_fragtex_state fragtex;
+	struct nv04_rasterizer_state *rast;
+	struct nv04_depth_stencil_alpha_state *dsa;
+
+	struct nv04_miptree *tex_miptree[PIPE_MAX_SAMPLERS];
+	unsigned dirty_samplers;
+	unsigned fp_samplers;
+	unsigned vp_samplers;
+
+	uint32_t rt_enable;
+	struct pipe_buffer *rt[4];
+	struct pipe_buffer *zeta;
+
+	struct {
+		struct pipe_buffer *buffer;
+		uint32_t format;
+	} tex[16];
+
+	unsigned vb_enable;
+	struct {
+		struct pipe_buffer *buffer;
+		unsigned delta;
+	} vb[16];
+
+	struct vertex_info vertex_info;
+	struct {
+	
+		struct nouveau_resource *exec_heap;
+		struct nouveau_resource *data_heap;
+
+		struct nv04_vertex_program *active;
+
+		struct nv04_vertex_program *current;
+		struct pipe_buffer *constant_buf;
+	} vertprog;
+
+	struct {
+		struct nv04_fragment_program *active;
+
+		struct nv04_fragment_program *current;
+		struct pipe_buffer *constant_buf;
+	} fragprog;
+
+	struct pipe_vertex_buffer  vertex_buffer[PIPE_MAX_ATTRIBS];
+	unsigned num_vertex_buffers;
+	unsigned num_vertex_elements;
+
+	struct pipe_viewport_state viewport;
+};
+
+static INLINE struct nv04_context *
+nv04_context(struct pipe_context *pipe)
+{
+	return (struct nv04_context *)pipe;
+}
+
+extern void nv04_init_state_functions(struct nv04_context *nv04);
+extern void nv04_init_surface_functions(struct nv04_context *nv04);
+extern void nv04_init_miptree_functions(struct pipe_screen *screen);
+
+/* nv04_clear.c */
+extern void nv04_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+/* nv04_draw.c */
+extern struct draw_stage *nv04_draw_render_stage(struct nv04_context *nv04);
+
+/* nv04_fragprog.c */
+extern void nv04_fragprog_bind(struct nv04_context *,
+			       struct nv04_fragment_program *);
+extern void nv04_fragprog_destroy(struct nv04_context *,
+				  struct nv04_fragment_program *);
+
+/* nv04_fragtex.c */
+extern void nv04_fragtex_bind(struct nv04_context *);
+
+/* nv04_prim_vbuf.c */
+struct draw_stage *nv04_draw_vbuf_stage( struct nv04_context *nv04 );
+
+/* nv04_state.c and friends */
+extern void nv04_emit_hw_state(struct nv04_context *nv04);
+extern void nv04_state_tex_update(struct nv04_context *nv04);
+
+/* nv04_vbo.c */
+extern boolean nv04_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv04_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count);
+
+
+#endif
diff --git a/src/gallium/drivers/nv04/nv04_fragprog.c b/src/gallium/drivers/nv04/nv04_fragprog.c
new file mode 100644
index 00000000000..8a2af41fe06
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_fragprog.c
@@ -0,0 +1,21 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv04_context.h"
+
+void
+nv04_fragprog_bind(struct nv04_context *nv04, struct nv04_fragment_program *fp)
+{
+}
+
+void
+nv04_fragprog_destroy(struct nv04_context *nv04,
+		      struct nv04_fragment_program *fp)
+{
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_fragtex.c b/src/gallium/drivers/nv04/nv04_fragtex.c
new file mode 100644
index 00000000000..21f990fd536
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_fragtex.c
@@ -0,0 +1,73 @@
+#include "nv04_context.h"
+#include "nouveau/nouveau_util.h"
+
+#define _(m,tf)                                                                \
+{                                                                              \
+  PIPE_FORMAT_##m,                                                             \
+  NV04_DX5_TEXTURED_TRIANGLE_FORMAT_COLOR_##tf,                                               \
+}
+
+struct nv04_texture_format {
+	uint	pipe;
+	int     format;
+};
+
+static struct nv04_texture_format
+nv04_texture_formats[] = {
+	_(A8R8G8B8_UNORM, A8R8G8B8),
+	_(X8R8G8B8_UNORM, X8R8G8B8),
+	_(A1R5G5B5_UNORM, A1R5G5B5),
+	_(A4R4G4B4_UNORM, A4R4G4B4),
+	_(L8_UNORM,       Y8      ),
+	_(A8_UNORM,       Y8      ),
+};
+
+static uint32_t
+nv04_fragtex_format(uint pipe_format)
+{
+	struct nv04_texture_format *tf = nv04_texture_formats;
+	int i;
+
+	for (i=0; i< sizeof(nv04_texture_formats)/sizeof(nv04_texture_formats[0]); i++) {
+		if (tf->pipe == pipe_format)
+			return tf->format;
+		tf++;
+	}
+
+	NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format));
+	return 0;
+}
+
+
+static void
+nv04_fragtex_build(struct nv04_context *nv04, int unit)
+{
+	struct nv04_miptree *nv04mt = nv04->tex_miptree[unit];
+	struct pipe_texture *pt = &nv04mt->base;
+
+	switch (pt->target) {
+	case PIPE_TEXTURE_2D:
+		break;
+	default:
+		NOUVEAU_ERR("Unknown target %d\n", pt->target);
+		return;
+	}
+
+	nv04->fragtex.format = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_ZOH_CORNER 
+		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ORIGIN_FOH_CORNER
+		| nv04_fragtex_format(pt->format)
+		| ( (pt->last_level + 1) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_MIPMAP_LEVELS_SHIFT )
+		| ( log2i(pt->width[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_U_SHIFT )
+		| ( log2i(pt->height[0]) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_BASE_SIZE_V_SHIFT )
+		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE
+		| NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_CLAMP_TO_EDGE
+		;
+}
+
+
+void
+nv04_fragtex_bind(struct nv04_context *nv04)
+{
+	nv04_fragtex_build(nv04, 0);
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_miptree.c b/src/gallium/drivers/nv04/nv04_miptree.c
new file mode 100644
index 00000000000..0cbb91e187b
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_miptree.c
@@ -0,0 +1,138 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv04_context.h"
+#include "nv04_screen.h"
+
+static void
+nv04_miptree_layout(struct nv04_miptree *nv04mt)
+{
+	struct pipe_texture *pt = &nv04mt->base;
+	uint width = pt->width[0], height = pt->height[0];
+	uint offset = 0;
+	int nr_faces, l, f;
+
+	nr_faces = 1;
+
+	for (l = 0; l <= pt->last_level; l++) {
+		pt->width[l] = width;
+		pt->height[l] = height;
+
+		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
+		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
+		
+		nv04mt->level[l].pitch = pt->width[0] * pt->block.size;
+		nv04mt->level[l].pitch = (nv04mt->level[l].pitch + 63) & ~63;
+
+		nv04mt->level[l].image_offset =
+			CALLOC(nr_faces, sizeof(unsigned));
+
+		width  = MAX2(1, width  >> 1);
+		height = MAX2(1, height >> 1);
+
+	}
+
+	for (f = 0; f < nr_faces; f++) {
+		for (l = 0; l <= pt->last_level; l++) {
+			nv04mt->level[l].image_offset[f] = offset;
+			offset += nv04mt->level[l].pitch * pt->height[l];
+		}
+	}
+
+	nv04mt->total_size = offset;
+}
+
+static struct pipe_texture *
+nv04_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = screen->winsys;
+	struct nv04_miptree *mt;
+
+	mt = MALLOC(sizeof(struct nv04_miptree));
+	if (!mt)
+		return NULL;
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = screen;
+
+	nv04_miptree_layout(mt);
+
+	mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL,
+					   mt->total_size);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+	
+	return &mt->base;
+}
+
+static void
+nv04_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt)
+{
+	struct pipe_texture *mt = *pt;
+
+	*pt = NULL;
+	if (--mt->refcount <= 0) {
+		struct nv04_miptree *nv04mt = (struct nv04_miptree *)mt;
+		int l;
+
+		pipe_buffer_reference(screen, &nv04mt->buffer, NULL);
+		for (l = 0; l <= mt->last_level; l++) {
+			if (nv04mt->level[l].image_offset)
+				FREE(nv04mt->level[l].image_offset);
+		}
+		FREE(nv04mt);
+	}
+}
+
+static struct pipe_surface *
+nv04_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv04_miptree *nv04mt = (struct nv04_miptree *)pt;
+	struct pipe_surface *ps;
+
+	ps = ws->surface_alloc(ws);
+	if (!ps)
+		return NULL;
+	pipe_buffer_reference(pscreen, &ps->buffer, nv04mt->buffer);
+	ps->format = pt->format;
+		ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = nv04mt->level[level].pitch;
+	ps->refcount = 1;
+	ps->winsys = pscreen->winsys;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		ps->offset = nv04mt->level[level].image_offset[face];
+	} else {
+		ps->offset = nv04mt->level[level].image_offset[0];
+	}
+
+	return ps;
+}
+
+static void
+nv04_miptree_surface_del(struct pipe_screen *pscreen,
+			 struct pipe_surface **psurface)
+{
+}
+
+void
+nv04_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv04_miptree_create;
+	pscreen->texture_release = nv04_miptree_release;
+	pscreen->get_tex_surface = nv04_miptree_surface_new;
+	pscreen->tex_surface_release = nv04_miptree_surface_del;
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_prim_vbuf.c b/src/gallium/drivers/nv04/nv04_prim_vbuf.c
new file mode 100644
index 00000000000..19979fff795
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_prim_vbuf.c
@@ -0,0 +1,309 @@
+
+#include "pipe/p_debug.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_compiler.h"
+
+#include "draw/draw_vbuf.h"
+
+#include "nv04_context.h"
+#include "nv04_state.h"
+
+#define VERTEX_SIZE 40
+#define VERTEX_BUFFER_SIZE (4096*VERTEX_SIZE) // 4096 vertices of 40 bytes each
+
+/**
+ * Primitive renderer for nv04.
+ */
+struct nv04_vbuf_render {
+	struct vbuf_render base;
+
+	struct nv04_context *nv04;   
+
+	/** Vertex buffer */
+	unsigned char* buffer;
+
+	/** Vertex size in bytes */
+	unsigned vertex_size;
+
+	/** Current primitive */
+	unsigned prim;
+};
+
+
+/**
+ * Basically a cast wrapper.
+ */
+static INLINE struct nv04_vbuf_render *
+nv04_vbuf_render( struct vbuf_render *render )
+{
+	assert(render);
+	return (struct nv04_vbuf_render *)render;
+}
+
+
+static const struct vertex_info *
+nv04_vbuf_render_get_vertex_info( struct vbuf_render *render )
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+	struct nv04_context *nv04 = nv04_render->nv04;
+	return &nv04->vertex_info;
+}
+
+
+static void *
+nv04_vbuf_render_allocate_vertices( struct vbuf_render *render,
+		ushort vertex_size,
+		ushort nr_vertices )
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+
+	nv04_render->buffer = (unsigned char*) MALLOC(VERTEX_BUFFER_SIZE);
+	assert(!nv04_render->buffer);
+
+	return nv04_render->buffer;
+}
+
+
+static boolean 
+nv04_vbuf_render_set_primitive( struct vbuf_render *render, 
+		unsigned prim )
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+
+	if (prim <= PIPE_PRIM_LINE_STRIP)
+		return FALSE;
+
+	nv04_render->prim = prim;
+	return TRUE;
+}
+
+static INLINE void nv04_2triangles(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2, ushort v3, ushort v4, ushort v5)
+{
+	BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xA),49);
+	OUT_RINGp(buffer + VERTEX_SIZE * v0,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v1,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v2,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v3,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v4,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v5,8);
+	OUT_RING(0xFEDCBA);
+}
+
+static INLINE void nv04_1triangle(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2)
+{
+	BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xD),25);
+	OUT_RINGp(buffer + VERTEX_SIZE * v0,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v1,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v2,8);
+	OUT_RING(0xFED);
+}
+
+static INLINE void nv04_1quad(struct nv04_context* nv04, unsigned char* buffer, ushort v0, ushort v1, ushort v2, ushort v3)
+{
+	BEGIN_RING(fahrenheit,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xC),33);
+	OUT_RINGp(buffer + VERTEX_SIZE * v0,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v1,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v2,8);
+	OUT_RINGp(buffer + VERTEX_SIZE * v3,8);
+	OUT_RING(0xFECEDC);
+}
+
+static void nv04_vbuf_render_triangles_elts(struct nv04_vbuf_render * render, const ushort * indices, uint nr_indices)
+{
+	unsigned char* buffer = render->buffer;
+	struct nv04_context* nv04 = render->nv04;
+	int i;
+
+	for( i=0; i< nr_indices-5; i+=6)
+		nv04_2triangles(nv04,
+				buffer,
+				indices[i+0],
+				indices[i+1],
+				indices[i+2],
+				indices[i+3],
+				indices[i+4],
+				indices[i+5]
+			       );
+	if (i != nr_indices)
+	{
+		nv04_1triangle(nv04,
+				buffer,
+				indices[i+0],
+				indices[i+1],
+				indices[i+2]
+			       );
+		i+=3;
+	}
+	if (i != nr_indices)
+		NOUVEAU_ERR("Houston, we have lost some vertices\n");
+}
+
+static void nv04_vbuf_render_tri_strip_elts(struct nv04_vbuf_render* render, const ushort* indices, uint nr_indices)
+{
+	const uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC};
+	unsigned char* buffer = render->buffer;
+	struct nv04_context* nv04 = render->nv04;
+	int i,j;
+
+	for(i = 0; i<nr_indices; i+=14) 
+	{
+		int numvert = MIN2(16, nr_indices - i);
+		int numtri = numvert - 2;
+		if (numvert<3)
+			break;
+
+		BEGIN_RING( fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0), numvert*8 );
+		for(j = 0; j<numvert; j++)
+			OUT_RINGp( buffer + VERTEX_SIZE * indices [i+j], 8 );
+
+		BEGIN_RING_NI( fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(0), (numtri+1)/2 );
+		for(j = 0; j<numtri/2; j++ )
+			OUT_RING(striptbl[j]);
+		if (numtri%2)
+			OUT_RING(striptbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_vbuf_render_tri_fan_elts(struct nv04_vbuf_render* render, const ushort* indices, uint nr_indices)
+{
+	const uint32_t fantbl[]={0x320210,0x540430,0x760650,0x980870,0xBA0A90,0xDC0CB0,0xFE0ED0};
+	unsigned char* buffer = render->buffer;
+	struct nv04_context* nv04 = render->nv04;
+	int i,j;
+
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0), 8);
+	OUT_RINGp(buffer + VERTEX_SIZE * indices[0], 8);
+
+	for(i = 1; i<nr_indices; i+=14)
+	{
+		int numvert=MIN2(15, nr_indices - i);
+		int numtri=numvert-2;
+		if (numvert < 3)
+			break;
+
+		BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x1), numvert*8);
+
+		for(j=0;j<numvert;j++)
+			OUT_RINGp( buffer + VERTEX_SIZE * indices[ i+j ], 8 );
+
+		BEGIN_RING_NI(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_DRAWPRIMITIVE(0), (numtri+1)/2);
+		for(j = 0; j<numtri/2; j++)
+			OUT_RING(fantbl[j]);
+		if (numtri%2)
+			OUT_RING(fantbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_vbuf_render_quads_elts(struct nv04_vbuf_render* render, const ushort* indices, uint nr_indices)
+{
+	unsigned char* buffer = render->buffer;
+	struct nv04_context* nv04 = render->nv04;
+	int i;
+
+	for(i = 0; i < nr_indices; i += 4)
+		nv04_1quad(nv04,
+				buffer,
+				indices[i+0],
+				indices[i+1],
+				indices[i+2],
+				indices[i+3]
+			       );
+}
+
+
+static void 
+nv04_vbuf_render_draw( struct vbuf_render *render,
+		const ushort *indices,
+		uint nr_indices)
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+
+	// emit the indices
+	switch( nv04_render->prim )
+	{
+		case PIPE_PRIM_TRIANGLES:
+			nv04_vbuf_render_triangles_elts(nv04_render, indices, nr_indices);
+			break;
+		case PIPE_PRIM_QUAD_STRIP:
+		case PIPE_PRIM_TRIANGLE_STRIP:
+			nv04_vbuf_render_tri_strip_elts(nv04_render, indices, nr_indices);
+			break;
+		case PIPE_PRIM_TRIANGLE_FAN:
+		case PIPE_PRIM_POLYGON:
+			nv04_vbuf_render_tri_fan_elts(nv04_render, indices, nr_indices);
+			break;
+		case PIPE_PRIM_QUADS:
+			nv04_vbuf_render_quads_elts(nv04_render, indices, nr_indices);
+			break;
+		default:
+			NOUVEAU_ERR("You have to implement primitive %d, young padawan\n", nv04_render->prim);
+			break;
+	}
+}
+
+
+static void
+nv04_vbuf_render_release_vertices( struct vbuf_render *render,
+		void *vertices, 
+		unsigned vertex_size,
+		unsigned vertices_used )
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+
+	free(nv04_render->buffer);
+	nv04_render->buffer = NULL;
+}
+
+
+static void
+nv04_vbuf_render_destroy( struct vbuf_render *render )
+{
+	struct nv04_vbuf_render *nv04_render = nv04_vbuf_render(render);
+	FREE(nv04_render);
+}
+
+
+/**
+ * Create a new primitive render.
+ */
+static struct vbuf_render *
+nv04_vbuf_render_create( struct nv04_context *nv04 )
+{
+	struct nv04_vbuf_render *nv04_render = CALLOC_STRUCT(nv04_vbuf_render);
+
+	nv04_render->nv04 = nv04;
+
+	nv04_render->base.max_vertex_buffer_bytes = VERTEX_BUFFER_SIZE;
+	nv04_render->base.max_indices = 65536; 
+	nv04_render->base.get_vertex_info = nv04_vbuf_render_get_vertex_info;
+	nv04_render->base.allocate_vertices = nv04_vbuf_render_allocate_vertices;
+	nv04_render->base.set_primitive = nv04_vbuf_render_set_primitive;
+	nv04_render->base.draw = nv04_vbuf_render_draw;
+	nv04_render->base.release_vertices = nv04_vbuf_render_release_vertices;
+	nv04_render->base.destroy = nv04_vbuf_render_destroy;
+
+	return &nv04_render->base;
+}
+
+
+/**
+ * Create a new primitive vbuf/render stage.
+ */
+struct draw_stage *nv04_draw_vbuf_stage( struct nv04_context *nv04 )
+{
+	struct vbuf_render *render;
+	struct draw_stage *stage;
+
+	render = nv04_vbuf_render_create(nv04);
+	if(!render)
+		return NULL;
+
+	stage = draw_vbuf_stage( nv04->draw, render );
+	if(!stage) {
+		render->destroy(render);
+		return NULL;
+	}
+
+	return stage;
+}
diff --git a/src/gallium/drivers/nv04/nv04_screen.c b/src/gallium/drivers/nv04/nv04_screen.c
new file mode 100644
index 00000000000..3966a29ffa9
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_screen.c
@@ -0,0 +1,212 @@
+#include "pipe/p_screen.h"
+
+#include "nv04_context.h"
+#include "nv04_screen.h"
+
+static const char *
+nv04_screen_get_name(struct pipe_screen *screen)
+{
+	struct nv04_screen *nv04screen = nv04_screen(screen);
+	struct nouveau_device *dev = nv04screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv04_screen_get_vendor(struct pipe_screen *screen)
+{
+	return "nouveau";
+}
+
+static int
+nv04_screen_get_param(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 1;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 0;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 0;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 0;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 0;
+	case PIPE_CAP_POINT_SPRITE:
+		return 0;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 1;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 0;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 10;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv04_screen_get_paramf(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 0.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 0.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 0.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 0.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static boolean
+nv04_screen_is_format_supported(struct pipe_screen *screen,
+				enum pipe_format format,
+				enum pipe_texture_target target,
+				unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static void *
+nv04_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
+		 unsigned flags )
+{
+	struct pipe_winsys *ws = screen->winsys;
+	void *map;
+
+	map = ws->buffer_map(ws, surface->buffer, flags);
+	if (!map)
+		return NULL;
+
+	return map + surface->offset;
+}
+
+static void
+nv04_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
+{
+	struct pipe_winsys *ws = screen->winsys;
+
+	ws->buffer_unmap(ws, surface->buffer);
+}
+
+static void
+nv04_screen_destroy(struct pipe_screen *pscreen)
+{
+	struct nv04_screen *screen = nv04_screen(pscreen);
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nvws->notifier_free(&screen->sync);
+	nvws->grobj_free(&screen->fahrenheit);
+
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv04_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv04_screen *screen = CALLOC_STRUCT(nv04_screen);
+	unsigned fahrenheit_class = 0, sub3d_class = 0;
+	unsigned chipset = nvws->channel->device->chipset;
+	int ret;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	if (chipset>=0x20) {
+		fahrenheit_class = 0;
+		sub3d_class = 0;
+	} else if (chipset>=0x10) {
+		fahrenheit_class = NV10_DX5_TEXTURED_TRIANGLE;
+		sub3d_class = NV10_CONTEXT_SURFACES_3D;
+	} else {
+		fahrenheit_class=NV04_DX5_TEXTURED_TRIANGLE;
+		sub3d_class = NV04_CONTEXT_SURFACES_3D;
+	}
+
+	if (!fahrenheit_class) {
+		NOUVEAU_ERR("Unknown nv04 chipset: nv%02x\n", chipset);
+		return NULL;
+	}
+
+	/* 3D object */
+	ret = nvws->grobj_alloc(nvws, fahrenheit_class, &screen->fahrenheit);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		return NULL;
+	}
+
+	/* 3D surface object */
+	ret = nvws->grobj_alloc(nvws, sub3d_class, &screen->context_surfaces_3d);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D surface object: %d\n", ret);
+		return NULL;
+	}
+
+	/* Notifier for sync purposes */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv04_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	screen->pipe.winsys = ws;
+	screen->pipe.destroy = nv04_screen_destroy;
+
+	screen->pipe.get_name = nv04_screen_get_name;
+	screen->pipe.get_vendor = nv04_screen_get_vendor;
+	screen->pipe.get_param = nv04_screen_get_param;
+	screen->pipe.get_paramf = nv04_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv04_screen_is_format_supported;
+
+	screen->pipe.surface_map = nv04_surface_map;
+	screen->pipe.surface_unmap = nv04_surface_unmap;
+
+	nv04_screen_init_miptree_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_screen.h b/src/gallium/drivers/nv04/nv04_screen.h
new file mode 100644
index 00000000000..99a49cdf7a9
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_screen.h
@@ -0,0 +1,25 @@
+#ifndef __NV04_SCREEN_H__
+#define __NV04_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv04_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+	unsigned chipset;
+
+	/* HW graphics objects */
+	struct nouveau_grobj *fahrenheit;
+	struct nouveau_grobj *context_surfaces_3d;
+	struct nouveau_notifier *sync;
+
+};
+
+static INLINE struct nv04_screen *
+nv04_screen(struct pipe_screen *screen)
+{
+	return (struct nv04_screen *)screen;
+}
+
+#endif
diff --git a/src/gallium/drivers/nv04/nv04_state.c b/src/gallium/drivers/nv04/nv04_state.c
new file mode 100644
index 00000000000..ff1933b5508
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_state.c
@@ -0,0 +1,495 @@
+#include "draw/draw_context.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv04_context.h"
+#include "nv04_state.h"
+
+static void *
+nv04_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv04_blend_state *cb;
+
+	cb = MALLOC(sizeof(struct nv04_blend_state));
+
+	cb->b_enable = cso->blend_enable ? 1 : 0;
+	cb->b_src = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_src_factor)));
+	cb->b_dst = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_dst_factor)));
+	
+
+	return (void *)cb;
+}
+
+static void
+nv04_blend_state_bind(struct pipe_context *pipe, void *blend)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	nv04->blend = (struct nv04_blend_state*)blend;
+
+	nv04->dirty |= NV04_NEW_BLEND;
+}
+
+static void
+nv04_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	free(hwcso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_CLAMP;
+	}
+	return ret >> NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT;
+}
+
+static void *
+nv04_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+
+	struct nv04_sampler_state *ss;
+	uint32_t filter = 0;
+
+	ss = MALLOC(sizeof(struct nv04_sampler_state));
+
+	ss->format = ((wrap_mode(cso->wrap_s) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSU_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV04_DX5_TEXTURED_TRIANGLE_FORMAT_ADDRESSV_SHIFT));
+
+	if (cso->max_anisotropy > 1.0) {
+		filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MINIFY_ENABLE | NV04_DX5_TEXTURED_TRIANGLE_FILTER_ANISOTROPIC_MAGNIFY_ENABLE;
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV04_DX5_TEXTURED_TRIANGLE_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ss->filter = filter;
+
+	return (void *)ss;
+}
+
+static void
+nv04_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv04->sampler[unit] = sampler[unit];
+		nv04->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void
+nv04_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	free(hwcso);
+}
+
+static void
+nv04_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv04->tex_miptree[unit] = (struct nv04_miptree *)miptree[unit];
+		nv04->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void *
+nv04_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv04_rasterizer_state *rs;
+
+	/*XXX: ignored:
+	 * 	scissor
+	 * 	points/lines (no hw support, emulated with tris in gallium)
+	 */
+	rs = MALLOC(sizeof(struct nv04_rasterizer_state));
+
+	rs->blend = cso->flatshade ? NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_FLAT : NV04_DX5_TEXTURED_TRIANGLE_BLEND_SHADE_MODE_GOURAUD;
+
+	return (void *)rs;
+}
+
+static void
+nv04_rasterizer_state_bind(struct pipe_context *pipe, void *rast)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	nv04->rast = (struct nv04_rasterizer_state*)rast;
+
+	draw_set_rasterizer_state(nv04->draw, (nv04->rast ? nv04->rast->templ : NULL));
+
+	nv04->dirty |= NV04_NEW_RAST | NV04_NEW_BLEND;
+}
+
+static void
+nv04_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	free(hwcso);
+}
+
+static INLINE uint32_t nv04_compare_func(uint32_t f)
+{
+	switch ( f ) {
+		case PIPE_FUNC_NEVER:		return 1;
+		case PIPE_FUNC_LESS:		return 2;
+		case PIPE_FUNC_EQUAL:		return 3;
+		case PIPE_FUNC_LEQUAL:		return 4;
+		case PIPE_FUNC_GREATER:		return 5;
+		case PIPE_FUNC_NOTEQUAL:	return 6;
+		case PIPE_FUNC_GEQUAL:		return 7;
+		case PIPE_FUNC_ALWAYS:		return 8;
+	}
+	NOUVEAU_MSG("Unable to find the function\n");
+	return 0;
+}
+
+static void *
+nv04_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv04_depth_stencil_alpha_state *hw;
+
+	hw = MALLOC(sizeof(struct nv04_depth_stencil_alpha_state));
+
+	hw->control = float_to_ubyte(cso->alpha.ref);
+	hw->control |= ( nv04_compare_func(cso->alpha.func) << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_FUNC_SHIFT );
+	hw->control |= cso->alpha.enabled ? NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHA_TEST_ENABLE : 0;
+	hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ORIGIN;
+	hw->control |= cso->depth.enabled ? (1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_ENABLE_SHIFT) : 0;
+	hw->control |= ( nv04_compare_func(cso->depth.func)<< NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FUNC_SHIFT );
+	hw->control |= 1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULL_MODE_SHIFT; // no culling, handled by the draw module
+	hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_DITHER_ENABLE;
+	hw->control |= NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_PERSPECTIVE_ENABLE;
+	hw->control |= cso->depth.writemask ? (1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_WRITE_ENABLE_SHIFT) : 0;
+	hw->control |= 1 << NV04_DX5_TEXTURED_TRIANGLE_CONTROL_Z_FORMAT_SHIFT; // integer zbuffer format
+
+	return (void *)hw;
+}
+
+static void
+nv04_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	nv04->dsa = hwcso;
+	nv04->dirty |= NV04_NEW_CONTROL;
+}
+
+static void
+nv04_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	free(hwcso);
+}
+
+static void *
+nv04_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *templ)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	return draw_create_vertex_shader(nv04->draw, templ);
+}
+
+static void
+nv04_vp_state_bind(struct pipe_context *pipe, void *shader)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	draw_bind_vertex_shader(nv04->draw, (struct draw_vertex_shader *) shader);
+
+	nv04->dirty |= NV04_NEW_VERTPROG;
+}
+
+static void
+nv04_vp_state_delete(struct pipe_context *pipe, void *shader)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	draw_delete_vertex_shader(nv04->draw, (struct draw_vertex_shader *) shader);
+}
+
+static void *
+nv04_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv04_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv04_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+
+	return (void *)fp;
+}
+
+static void
+nv04_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct nv04_fragment_program *fp = hwcso;
+
+	nv04->fragprog.current = fp;
+	nv04->dirty |= NV04_NEW_FRAGPROG;
+}
+
+static void
+nv04_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct nv04_fragment_program *fp = hwcso;
+
+	nv04_fragprog_destroy(nv04, fp);
+	free((void*)fp->pipe.tokens);
+	free(fp);
+}
+
+static void
+nv04_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+}
+
+static void
+nv04_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+}
+
+static void
+nv04_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	if (shader == PIPE_SHADER_VERTEX) {
+		nv04->vertprog.constant_buf = buf->buffer;
+		nv04->dirty |= NV04_NEW_VERTPROG;
+	} else
+	if (shader == PIPE_SHADER_FRAGMENT) {
+		nv04->fragprog.constant_buf = buf->buffer;
+		nv04->dirty |= NV04_NEW_FRAGPROG;
+	}
+}
+
+static void
+nv04_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct pipe_surface *rt, *zeta;
+	uint32_t rt_format, w, h;
+	int colour_format = 0, zeta_format = 0;
+
+	w = fb->cbufs[0]->width;
+	h = fb->cbufs[0]->height;
+	colour_format = fb->cbufs[0]->format;
+	rt = fb->cbufs[0];
+
+	if (fb->zsbuf) {
+		if (colour_format) {
+			assert(w == fb->zsbuf->width);
+			assert(h == fb->zsbuf->height);
+		} else {
+			w = fb->zsbuf->width;
+			h = fb->zsbuf->height;
+		}
+
+		zeta_format = fb->zsbuf->format;
+		zeta = fb->zsbuf;
+	}
+
+	switch (colour_format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case 0:
+		rt_format = 0x108;
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		rt_format = 0x103;
+		break;
+	default:
+		assert(0);
+	}
+
+	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_FORMAT, 1);
+	OUT_RING(rt_format);
+
+	/* FIXME pitches have to be aligned ! */
+	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
+	OUT_RING(rt->stride|(zeta->stride<<16));
+	OUT_RELOCl(rt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	if (fb->zsbuf) {
+		BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
+		OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	}
+}
+
+static void
+nv04_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	NOUVEAU_ERR("line stipple hahaha\n");
+}
+
+static void
+nv04_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+/*	struct nv04_context *nv04 = nv04_context(pipe);
+
+	// XXX
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_SCISSOR_HORIZ, 2);
+	OUT_RING  (((s->maxx - s->minx) << 16) | s->minx);
+	OUT_RING  (((s->maxy - s->miny) << 16) | s->miny);*/
+}
+
+static void
+nv04_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *viewport)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	nv04->viewport = *viewport;
+
+	draw_set_viewport_state(nv04->draw, &nv04->viewport);
+}
+
+static void
+nv04_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+		       const struct pipe_vertex_buffer *buffers)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	draw_flush(nv04->draw);
+
+	memcpy(nv04->vertex_buffer, buffers, count * sizeof(buffers[0]));
+	nv04->num_vertex_buffers = count;
+
+	draw_set_vertex_buffers(nv04->draw, count, buffers);
+}
+
+static void
+nv04_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_element *elements)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+
+	draw_flush(nv04->draw);
+
+	nv04->num_vertex_elements = count;
+	draw_set_vertex_elements(nv04->draw, count, elements);
+}
+
+void
+nv04_init_state_functions(struct nv04_context *nv04)
+{
+	nv04->pipe.create_blend_state = nv04_blend_state_create;
+	nv04->pipe.bind_blend_state = nv04_blend_state_bind;
+	nv04->pipe.delete_blend_state = nv04_blend_state_delete;
+
+	nv04->pipe.create_sampler_state = nv04_sampler_state_create;
+	nv04->pipe.bind_sampler_states = nv04_sampler_state_bind;
+	nv04->pipe.delete_sampler_state = nv04_sampler_state_delete;
+	nv04->pipe.set_sampler_textures = nv04_set_sampler_texture;
+
+	nv04->pipe.create_rasterizer_state = nv04_rasterizer_state_create;
+	nv04->pipe.bind_rasterizer_state = nv04_rasterizer_state_bind;
+	nv04->pipe.delete_rasterizer_state = nv04_rasterizer_state_delete;
+
+	nv04->pipe.create_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_create;
+	nv04->pipe.bind_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_bind;
+	nv04->pipe.delete_depth_stencil_alpha_state = nv04_depth_stencil_alpha_state_delete;
+
+	nv04->pipe.create_vs_state = nv04_vp_state_create;
+	nv04->pipe.bind_vs_state = nv04_vp_state_bind;
+	nv04->pipe.delete_vs_state = nv04_vp_state_delete;
+
+	nv04->pipe.create_fs_state = nv04_fp_state_create;
+	nv04->pipe.bind_fs_state = nv04_fp_state_bind;
+	nv04->pipe.delete_fs_state = nv04_fp_state_delete;
+
+	nv04->pipe.set_blend_color = nv04_set_blend_color;
+	nv04->pipe.set_clip_state = nv04_set_clip_state;
+	nv04->pipe.set_constant_buffer = nv04_set_constant_buffer;
+	nv04->pipe.set_framebuffer_state = nv04_set_framebuffer_state;
+	nv04->pipe.set_polygon_stipple = nv04_set_polygon_stipple;
+	nv04->pipe.set_scissor_state = nv04_set_scissor_state;
+	nv04->pipe.set_viewport_state = nv04_set_viewport_state;
+
+	nv04->pipe.set_vertex_buffers = nv04_set_vertex_buffers;
+	nv04->pipe.set_vertex_elements = nv04_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_state.h b/src/gallium/drivers/nv04/nv04_state.h
new file mode 100644
index 00000000000..399f750dbe7
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_state.h
@@ -0,0 +1,71 @@
+#ifndef __NV04_STATE_H__
+#define __NV04_STATE_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv04_blend_state {
+	uint32_t b_enable;
+	uint32_t b_src;
+	uint32_t b_dst;
+};
+
+struct nv04_fragtex_state {
+	uint32_t format;
+};
+
+struct nv04_sampler_state {
+	uint32_t filter;
+	uint32_t format;
+};
+
+struct nv04_depth_stencil_alpha_state {
+	uint32_t control;
+};
+
+struct nv04_rasterizer_state {
+	uint32_t blend;
+
+	const struct pipe_rasterizer_state *templ;
+};
+
+struct nv04_miptree {
+	struct pipe_texture base;
+
+	struct pipe_buffer *buffer;
+	uint total_size;
+
+	struct {
+		uint pitch;
+		uint *image_offset;
+	} level[PIPE_MAX_TEXTURE_LEVELS];
+};
+
+struct nv04_fragment_program_data {
+	unsigned offset;
+	unsigned index;
+};
+
+struct nv04_fragment_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+
+	boolean translated;
+	boolean on_hw;
+	unsigned samplers;
+
+	uint32_t *insn;
+	int       insn_len;
+
+	struct nv04_fragment_program_data *consts;
+	unsigned nr_consts;
+
+	struct pipe_buffer *buffer;
+
+	uint32_t fp_control;
+	uint32_t fp_reg_control;
+};
+
+
+
+#endif
diff --git a/src/gallium/drivers/nv04/nv04_state_emit.c b/src/gallium/drivers/nv04/nv04_state_emit.c
new file mode 100644
index 00000000000..0ad40a092ea
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_state_emit.c
@@ -0,0 +1,156 @@
+#include "nv04_context.h"
+#include "nv04_state.h"
+
+static void nv04_vertex_layout(struct pipe_context* pipe)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct nv04_fragment_program *fp = nv04->fragprog.current;
+	uint32_t src = 0;
+	int i;
+	struct vertex_info vinfo;
+
+	memset(&vinfo, 0, sizeof(vinfo));
+
+	for (i = 0; i < fp->info.num_inputs; i++) {
+		switch (i) {
+			case TGSI_SEMANTIC_POSITION:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++);
+				break;
+			case TGSI_SEMANTIC_COLOR:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++);
+				break;
+			default:
+			case TGSI_SEMANTIC_GENERIC:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++);
+				break;
+			case TGSI_SEMANTIC_FOG:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++);
+				break;
+		}
+	}
+	draw_compute_vertex_size(&vinfo);
+}
+
+static uint32_t nv04_blend_func(uint32_t f)
+{
+	switch ( f ) {
+		case PIPE_BLENDFACTOR_ZERO:			return 0x1;
+		case PIPE_BLENDFACTOR_ONE:			return 0x2;
+		case PIPE_BLENDFACTOR_SRC_COLOR:		return 0x3;
+		case PIPE_BLENDFACTOR_INV_SRC_COLOR:		return 0x4;
+		case PIPE_BLENDFACTOR_SRC_ALPHA:		return 0x5;
+		case PIPE_BLENDFACTOR_INV_SRC_ALPHA:		return 0x6;
+		case PIPE_BLENDFACTOR_DST_ALPHA:		return 0x7;
+		case PIPE_BLENDFACTOR_INV_DST_ALPHA:		return 0x8;
+		case PIPE_BLENDFACTOR_DST_COLOR:		return 0x9;
+		case PIPE_BLENDFACTOR_INV_DST_COLOR:		return 0xA;
+		case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:	return 0xB;
+	}
+	NOUVEAU_MSG("Unable to find the blend function 0x%x\n",f);
+	return 0;
+}
+
+static void nv04_emit_control(struct nv04_context* nv04)
+{
+	uint32_t control = nv04->dsa->control;
+
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1);
+	OUT_RING(control);
+}
+
+static void nv04_emit_blend(struct nv04_context* nv04)
+{
+	uint32_t blend;
+
+	blend=0x4; // texture MODULATE_ALPHA
+	blend|=0x20; // alpha is MSB
+	blend|=(2<<6); // flat shading
+	blend|=(1<<8); // persp correct
+	blend|=(0<<16); // no fog
+	blend|=(nv04->blend->b_enable<<20);
+	blend|=(nv04_blend_func(nv04->blend->b_src)<<24);
+	blend|=(nv04_blend_func(nv04->blend->b_dst)<<28);
+
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1);
+	OUT_RING(blend);
+}
+
+static void nv04_emit_sampler(struct nv04_context *nv04, int unit)
+{
+	struct nv04_miptree *nv04mt = nv04->tex_miptree[unit];
+	struct pipe_texture *pt = &nv04mt->base;
+
+	BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 3);
+	OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RELOCd(nv04mt->buffer, (nv04->fragtex.format | nv04->sampler[unit]->format), NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
+	OUT_RING(nv04->sampler[unit]->filter);
+}
+
+void
+nv04_emit_hw_state(struct nv04_context *nv04)
+{
+	int i;
+
+	if (nv04->dirty & NV04_NEW_VERTPROG) {
+		//nv04_vertprog_bind(nv04, nv04->vertprog.current);
+		nv04->dirty &= ~NV04_NEW_VERTPROG;
+	}
+
+	if (nv04->dirty & NV04_NEW_FRAGPROG) {
+		nv04_fragprog_bind(nv04, nv04->fragprog.current);
+		/*XXX: clear NV04_NEW_FRAGPROG if no new program uploaded */
+		nv04->dirty_samplers |= (1<<10);
+		nv04->dirty_samplers = 0;
+	}
+
+	if (nv04->dirty & NV04_NEW_CONTROL) {
+		nv04->dirty &= ~NV04_NEW_CONTROL;
+
+		BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1);
+		OUT_RING(nv04->dsa->control);
+	}
+
+	if (nv04->dirty & NV04_NEW_BLEND) {
+		nv04->dirty &= ~NV04_NEW_BLEND;
+
+		nv04_emit_blend(nv04);
+	}
+
+	if (nv04->dirty & NV04_NEW_SAMPLER) {
+		nv04->dirty &= ~NV04_NEW_SAMPLER;
+
+		nv04_emit_sampler(nv04, 0);
+	}
+
+	if (nv04->dirty & NV04_NEW_VIEWPORT) {
+		nv04->dirty &= ~NV04_NEW_VIEWPORT;
+//		nv04_state_emit_viewport(nv04);
+	}
+
+	/* Emit relocs for every referenced buffer.
+	 * This is to ensure the bufmgr has an accurate idea of how
+	 * the buffer is used.  This isn't very efficient, but we don't
+	 * seem to take a significant performance hit.  Will be improved
+	 * at some point.  Vertex arrays are emitted by nv04_vbo.c
+	 */
+
+	/* Render target */
+/*	BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
+	OUT_RING(rt->stride|(zeta->stride<<16));
+	OUT_RELOCl(rt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	if (fb->zsbuf) {
+		BEGIN_RING(context_surfaces_3d, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
+		OUT_RELOCl(zeta->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	}*/
+
+	/* Texture images */
+	for (i = 0; i < 1; i++) {
+		if (!(nv04->fp_samplers & (1 << i)))
+			continue;
+		struct nv04_miptree *nv04mt = nv04->tex_miptree[i];
+		BEGIN_RING(fahrenheit, NV04_DX5_TEXTURED_TRIANGLE_OFFSET, 2);
+		OUT_RELOCl(nv04mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+		OUT_RELOCd(nv04mt->buffer, (nv04->fragtex.format | nv04->sampler[i]->format), NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
+	}
+}
+
diff --git a/src/gallium/drivers/nv04/nv04_surface.c b/src/gallium/drivers/nv04/nv04_surface.c
new file mode 100644
index 00000000000..9d9943ed4e4
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_surface.c
@@ -0,0 +1,64 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "nv04_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+#include "util/u_tile.h"
+
+static void
+nv04_surface_copy(struct pipe_context *pipe, boolean do_flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct nouveau_winsys *nvws = nv04->nvws;
+
+	nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+			   width, height);
+}
+
+static void
+nv04_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv04_context *nv04 = nv04_context(pipe);
+	struct nouveau_winsys *nvws = nv04->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+void
+nv04_init_surface_functions(struct nv04_context *nv04)
+{
+	nv04->pipe.surface_copy = nv04_surface_copy;
+	nv04->pipe.surface_fill = nv04_surface_fill;
+}
diff --git a/src/gallium/drivers/nv04/nv04_vbo.c b/src/gallium/drivers/nv04/nv04_vbo.c
new file mode 100644
index 00000000000..91f919d48ec
--- /dev/null
+++ b/src/gallium/drivers/nv04/nv04_vbo.c
@@ -0,0 +1,71 @@
+#include "draw/draw_context.h"
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv04_context.h"
+#include "nv04_state.h"
+
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_pushbuf.h"
+
+boolean nv04_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count)
+{
+	struct nv04_context *nv04 = nv04_context( pipe );
+	struct draw_context *draw = nv04->draw;
+	unsigned i;
+
+	/*
+	 * Map vertex buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv04->vertex_buffer[i].buffer) {
+			void *buf
+				= pipe->winsys->buffer_map(pipe->winsys,
+						nv04->vertex_buffer[i].buffer,
+						PIPE_BUFFER_USAGE_CPU_READ);
+			draw_set_mapped_vertex_buffer(draw, i, buf);
+		}
+	}
+	/* Map index buffer, if present */
+	if (indexBuffer) {
+		void *mapped_indexes
+			= pipe->winsys->buffer_map(pipe->winsys, indexBuffer,
+					PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes);
+	}
+	else {
+		/* no index/element buffer */
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	/* draw! */
+	draw_arrays(nv04->draw, prim, start, count);
+
+	/*
+	 * unmap vertex/index buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv04->vertex_buffer[i].buffer) {
+			pipe->winsys->buffer_unmap(pipe->winsys, nv04->vertex_buffer[i].buffer);
+			draw_set_mapped_vertex_buffer(draw, i, NULL);
+		}
+	}
+	if (indexBuffer) {
+		pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer);
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	return TRUE;
+}
+
+boolean nv04_draw_arrays( struct pipe_context *pipe,
+				 unsigned prim, unsigned start, unsigned count)
+{
+	return nv04_draw_elements(pipe, NULL, 0, prim, start, count);
+}
+
+
+
diff --git a/src/gallium/drivers/nv10/Makefile b/src/gallium/drivers/nv10/Makefile
new file mode 100644
index 00000000000..4ba7ce586d6
--- /dev/null
+++ b/src/gallium/drivers/nv10/Makefile
@@ -0,0 +1,28 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv10
+
+DRIVER_SOURCES = \
+	nv10_clear.c \
+	nv10_context.c \
+	nv10_fragprog.c \
+	nv10_fragtex.c \
+	nv10_miptree.c \
+	nv10_prim_vbuf.c \
+	nv10_screen.c \
+	nv10_state.c \
+	nv10_state_emit.c \
+	nv10_surface.c \
+	nv10_vbo.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv10/nv10_clear.c b/src/gallium/drivers/nv10/nv10_clear.c
new file mode 100644
index 00000000000..be7e09cf4b0
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_clear.c
@@ -0,0 +1,12 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv10_context.h"
+
+void
+nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue);
+}
diff --git a/src/gallium/drivers/nv10/nv10_context.c b/src/gallium/drivers/nv10/nv10_context.c
new file mode 100644
index 00000000000..4eb4ed9185e
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_context.c
@@ -0,0 +1,296 @@
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv10_context.h"
+#include "nv10_screen.h"
+
+static void
+nv10_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_flush(nv10->draw);
+
+	FIRE_RING(fence);
+}
+
+static void
+nv10_destroy(struct pipe_context *pipe)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	if (nv10->draw)
+		draw_destroy(nv10->draw);
+
+	FREE(nv10);
+}
+
+static void nv10_init_hwctx(struct nv10_context *nv10)
+{
+	struct nv10_screen *screen = nv10->screen;
+	struct nouveau_winsys *nvws = screen->nvws;
+	int i;
+	float projectionmatrix[16];
+
+	BEGIN_RING(celsius, NV10TCL_DMA_NOTIFY, 1);
+	OUT_RING  (screen->sync->handle);
+	BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY0, 2);
+	OUT_RING  (nvws->channel->vram->handle);
+	OUT_RING  (nvws->channel->gart->handle);
+	BEGIN_RING(celsius, NV10TCL_DMA_IN_MEMORY2, 2);
+	OUT_RING  (nvws->channel->vram->handle);
+	OUT_RING  (nvws->channel->vram->handle);
+
+	BEGIN_RING(celsius, NV10TCL_NOP, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+
+	BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 1);
+	OUT_RING  ((0x7ff<<16)|0x800);
+	BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(0), 1);
+	OUT_RING  ((0x7ff<<16)|0x800);
+
+	for (i=1;i<8;i++) {
+		BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(i), 1);
+		OUT_RING  (0);
+		BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_VERT(i), 1);
+		OUT_RING  (0);
+	}
+
+	BEGIN_RING(celsius, 0x290, 1);
+	OUT_RING  ((0x10<<16)|1);
+	BEGIN_RING(celsius, 0x3f4, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(celsius, NV10TCL_NOP, 1);
+	OUT_RING  (0);
+
+	if (nv10->screen->celsius->grclass != NV10TCL) {
+		/* For nv11, nv17 */
+		BEGIN_RING(celsius, 0x120, 3);
+		OUT_RING  (0);
+		OUT_RING  (1);
+		OUT_RING  (2);
+
+		BEGIN_RING(celsius, NV10TCL_NOP, 1);
+		OUT_RING  (0);
+	}
+
+	BEGIN_RING(celsius, NV10TCL_NOP, 1);
+	OUT_RING  (0);
+
+	/* Set state */
+	BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 2);
+	OUT_RING  (0x207);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_TX_ENABLE(0), 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+
+	BEGIN_RING(celsius, NV10TCL_RC_IN_ALPHA(0), 12);
+	OUT_RING  (0x30141010);
+	OUT_RING  (0);
+	OUT_RING  (0x20040000);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0x18000000);
+	OUT_RING  (0x300e0300);
+	OUT_RING  (0x0c091c80);
+
+	BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 2);
+	OUT_RING  (1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_WEIGHT_ENABLE, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_SRC, 4);
+	OUT_RING  (1);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0x8006);
+	BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 8);
+	OUT_RING  (0xff);
+	OUT_RING  (0x207);
+	OUT_RING  (0);
+	OUT_RING  (0xff);
+	OUT_RING  (0x1e00);
+	OUT_RING  (0x1e00);
+	OUT_RING  (0x1e00);
+	OUT_RING  (0x1d01);
+	BEGIN_RING(celsius, NV10TCL_NORMALIZE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_FOG_ENABLE, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_LIGHT_MODEL, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_COLOR_CONTROL, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_ENABLED_LIGHTS, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1);
+	OUT_RING  (0x201);
+	BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_POLYGON_OFFSET_FACTOR, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1);
+	OUT_RING  (8);
+	BEGIN_RING(celsius, NV10TCL_POINT_PARAMETERS_ENABLE, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_LINE_WIDTH, 1);
+	OUT_RING  (8);
+	BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (0x1b02);
+	OUT_RING  (0x1b02);
+	BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2);
+	OUT_RING  (0x405);
+	OUT_RING  (0x901);
+	BEGIN_RING(celsius, NV10TCL_POLYGON_SMOOTH_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_TX_GEN_S(0), 8);
+	for (i=0;i<8;i++) {
+		OUT_RING  (0);
+	}
+	BEGIN_RING(celsius, NV10TCL_FOG_EQUATION_CONSTANT, 3);
+	OUT_RING  (0x3fc00000);	/* -1.50 */
+	OUT_RING  (0xbdb8aa0a);	/* -0.09 */
+	OUT_RING  (0);		/*  0.00 */
+
+	BEGIN_RING(celsius, NV10TCL_NOP, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(celsius, NV10TCL_FOG_MODE, 2);
+	OUT_RING  (0x802);
+	OUT_RING  (2);
+	/* for some reason VIEW_MATRIX_ENABLE need to be 6 instead of 4 when
+	 * using texturing, except when using the texture matrix
+	 */
+	BEGIN_RING(celsius, NV10TCL_VIEW_MATRIX_ENABLE, 1);
+	OUT_RING  (6);
+	BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1);
+	OUT_RING  (0x01010101);
+
+	/* Set vertex component */
+	BEGIN_RING(celsius, NV10TCL_VERTEX_COL_4F_R, 4);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_COL2_3F_R, 3);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_NOR_3F_X, 3);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RINGf (1.0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_TX0_4F_S, 4);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_TX1_4F_S, 4);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING(celsius, NV10TCL_VERTEX_FOG_1F, 1);
+	OUT_RINGf (0.0);
+	BEGIN_RING(celsius, NV10TCL_EDGEFLAG_ENABLE, 1);
+	OUT_RING  (1);
+
+	memset(projectionmatrix, 0, sizeof(projectionmatrix));
+	BEGIN_RING(celsius, NV10TCL_PROJECTION_MATRIX(0), 16);
+	projectionmatrix[0*4+0] = 1.0;
+	projectionmatrix[1*4+1] = 1.0;
+	projectionmatrix[2*4+2] = 1.0;
+	projectionmatrix[3*4+3] = 1.0;
+	for (i=0;i<16;i++) {
+		OUT_RINGf  (projectionmatrix[i]);
+	}
+
+	BEGIN_RING(celsius, NV10TCL_DEPTH_RANGE_NEAR, 2);
+	OUT_RING  (0.0);
+	OUT_RINGf  (16777216.0);
+
+	BEGIN_RING(celsius, NV10TCL_VIEWPORT_SCALE_X, 4);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (16777215.0 * 0.5);
+	OUT_RING  (0);
+
+	FIRE_RING (NULL);
+}
+
+static void
+nv10_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+}
+
+struct pipe_context *
+nv10_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct nv10_screen *screen = nv10_screen(pscreen);
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv10_context *nv10;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nv10 = CALLOC(1, sizeof(struct nv10_context));
+	if (!nv10)
+		return NULL;
+	nv10->screen = screen;
+	nv10->pctx_id = pctx_id;
+
+	nv10->nvws = nvws;
+
+	nv10->pipe.winsys = ws;
+	nv10->pipe.screen = pscreen;
+	nv10->pipe.destroy = nv10_destroy;
+	nv10->pipe.set_edgeflags = nv10_set_edgeflags;
+	nv10->pipe.draw_arrays = nv10_draw_arrays;
+	nv10->pipe.draw_elements = nv10_draw_elements;
+	nv10->pipe.clear = nv10_clear;
+	nv10->pipe.flush = nv10_flush;
+
+	nv10_init_surface_functions(nv10);
+	nv10_init_state_functions(nv10);
+
+	nv10->draw = draw_create();
+	assert(nv10->draw);
+	draw_set_rasterize_stage(nv10->draw, nv10_draw_vbuf_stage(nv10));
+
+	nv10_init_hwctx(nv10);
+
+	return &nv10->pipe;
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_context.h b/src/gallium/drivers/nv10/nv10_context.h
new file mode 100644
index 00000000000..f3b56de25a7
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_context.h
@@ -0,0 +1,153 @@
+#ifndef __NV10_CONTEXT_H__
+#define __NV10_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv10_screen *ctx = nv10->screen
+#include "nouveau/nouveau_push.h"
+
+#include "nv10_state.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+#define NV10_NEW_VERTPROG	(1 << 0)
+#define NV10_NEW_FRAGPROG	(1 << 1)
+#define NV10_NEW_VTXARRAYS	(1 << 2)
+#define NV10_NEW_BLEND		(1 << 3)
+#define NV10_NEW_BLENDCOL	(1 << 4)
+#define NV10_NEW_RAST 		(1 << 5)
+#define NV10_NEW_DSA  		(1 << 6)
+#define NV10_NEW_VIEWPORT	(1 << 7)
+#define NV10_NEW_SCISSOR	(1 << 8)
+#define NV10_NEW_FRAMEBUFFER	(1 << 9)
+
+#include "nv10_screen.h"
+
+struct nv10_context {
+	struct pipe_context pipe;
+
+	struct nouveau_winsys *nvws;
+	struct nv10_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	uint32_t dirty;
+
+	struct nv10_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS];
+	struct nv10_miptree *tex_miptree[PIPE_MAX_SAMPLERS];
+	unsigned dirty_samplers;
+	unsigned fp_samplers;
+	unsigned vp_samplers;
+
+	uint32_t rt_enable;
+	struct pipe_buffer *rt[4];
+	struct pipe_buffer *zeta;
+	uint32_t lma_offset;
+
+	struct nv10_blend_state *blend;
+	struct pipe_blend_color *blend_color;
+	struct nv10_rasterizer_state *rast;
+	struct nv10_depth_stencil_alpha_state *dsa;
+	struct pipe_viewport_state *viewport;
+	struct pipe_scissor_state *scissor;
+	struct pipe_framebuffer_state *framebuffer;
+
+	//struct pipe_buffer *constbuf[PIPE_SHADER_TYPES];
+	float *constbuf[PIPE_SHADER_TYPES][32][4];
+	unsigned constbuf_nr[PIPE_SHADER_TYPES];
+
+	struct vertex_info vertex_info;
+
+	struct {
+		struct pipe_buffer *buffer;
+		uint32_t format;
+	} tex[2];
+
+	unsigned vb_enable;
+	struct {
+		struct pipe_buffer *buffer;
+		unsigned delta;
+	} vb[16];
+
+/*	struct {
+	
+		struct nouveau_resource *exec_heap;
+		struct nouveau_resource *data_heap;
+
+		struct nv10_vertex_program *active;
+
+		struct nv10_vertex_program *current;
+	} vertprog;
+*/
+	struct {
+		struct nv10_fragment_program *active;
+
+		struct nv10_fragment_program *current;
+		struct pipe_buffer *constant_buf;
+	} fragprog;
+
+	struct pipe_vertex_buffer  vtxbuf[PIPE_MAX_ATTRIBS];
+	struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS];
+};
+
+static INLINE struct nv10_context *
+nv10_context(struct pipe_context *pipe)
+{
+	return (struct nv10_context *)pipe;
+}
+
+extern void nv10_init_state_functions(struct nv10_context *nv10);
+extern void nv10_init_surface_functions(struct nv10_context *nv10);
+
+extern void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen);
+
+/* nv10_clear.c */
+extern void nv10_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+/* nv10_draw.c */
+extern struct draw_stage *nv10_draw_render_stage(struct nv10_context *nv10);
+
+/* nv10_fragprog.c */
+extern void nv10_fragprog_bind(struct nv10_context *,
+			       struct nv10_fragment_program *);
+extern void nv10_fragprog_destroy(struct nv10_context *,
+				  struct nv10_fragment_program *);
+
+/* nv10_fragtex.c */
+extern void nv10_fragtex_bind(struct nv10_context *);
+
+/* nv10_prim_vbuf.c */
+struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 );
+extern void nv10_vtxbuf_bind(struct nv10_context* nv10);
+
+/* nv10_state.c and friends */
+extern void nv10_emit_hw_state(struct nv10_context *nv10);
+extern void nv10_state_tex_update(struct nv10_context *nv10);
+
+/* nv10_vbo.c */
+extern boolean nv10_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv10_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count);
+
+
+#endif
diff --git a/src/gallium/drivers/nv10/nv10_fragprog.c b/src/gallium/drivers/nv10/nv10_fragprog.c
new file mode 100644
index 00000000000..698db5a16a9
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_fragprog.c
@@ -0,0 +1,21 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv10_context.h"
+
+void
+nv10_fragprog_bind(struct nv10_context *nv10, struct nv10_fragment_program *fp)
+{
+}
+
+void
+nv10_fragprog_destroy(struct nv10_context *nv10,
+		      struct nv10_fragment_program *fp)
+{
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_fragtex.c b/src/gallium/drivers/nv10/nv10_fragtex.c
new file mode 100644
index 00000000000..27f2f875847
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_fragtex.c
@@ -0,0 +1,124 @@
+#include "nv10_context.h"
+#include "nouveau/nouveau_util.h"
+
+#define _(m,tf)                                                                \
+{                                                                              \
+  TRUE,                                                                        \
+  PIPE_FORMAT_##m,                                                             \
+  NV10TCL_TX_FORMAT_FORMAT_##tf,                                               \
+}
+
+struct nv10_texture_format {
+	boolean defined;
+	uint	pipe;
+	int     format;
+};
+
+static struct nv10_texture_format
+nv10_texture_formats[] = {
+	_(A8R8G8B8_UNORM, A8R8G8B8),
+	_(A1R5G5B5_UNORM, A1R5G5B5),
+	_(A4R4G4B4_UNORM, A4R4G4B4),
+	_(L8_UNORM      , L8      ),
+	_(A8_UNORM      , A8      ),
+	_(A8L8_UNORM    , A8L8    ),
+//	_(RGB_DXT1      , DXT1,   ),
+//	_(RGBA_DXT1     , DXT1,   ),
+//	_(RGBA_DXT3     , DXT3,   ),
+//	_(RGBA_DXT5     , DXT5,   ),
+	{},
+};
+
+static struct nv10_texture_format *
+nv10_fragtex_format(uint pipe_format)
+{
+	struct nv10_texture_format *tf = nv10_texture_formats;
+
+	while (tf->defined) {
+		if (tf->pipe == pipe_format)
+			return tf;
+		tf++;
+	}
+
+	return NULL;
+}
+
+
+static void
+nv10_fragtex_build(struct nv10_context *nv10, int unit)
+{
+#if 0
+	struct nv10_sampler_state *ps = nv10->tex_sampler[unit];
+	struct nv10_miptree *nv10mt = nv10->tex_miptree[unit];
+	struct pipe_texture *pt = &nv10mt->base;
+	struct nv10_texture_format *tf;
+	uint32_t txf, txs, txp;
+
+	tf = nv10_fragtex_format(pt->format);
+	if (!tf || !tf->defined) {
+		NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format);
+		return;
+	}
+
+	txf  = tf->format << 8;
+	txf |= (pt->last_level + 1) << 16;
+	txf |= log2i(pt->width[0]) << 20;
+	txf |= log2i(pt->height[0]) << 24;
+	txf |= log2i(pt->depth[0]) << 28;
+	txf |= 8;
+
+	switch (pt->target) {
+	case PIPE_TEXTURE_CUBE:
+		txf |= NV10TCL_TX_FORMAT_CUBE_MAP;
+		/* fall-through */
+	case PIPE_TEXTURE_2D:
+		txf |= (2<<4);
+		break;
+	case PIPE_TEXTURE_1D:
+		txf |= (1<<4);
+		break;
+	default:
+		NOUVEAU_ERR("Unknown target %d\n", pt->target);
+		return;
+	}
+
+	BEGIN_RING(celsius, NV10TCL_TX_OFFSET(unit), 8);
+	OUT_RELOCl(nv10mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RELOCd(nv10mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
+	OUT_RING  (ps->wrap);
+	OUT_RING  (0x40000000); /* enable */
+	OUT_RING  (txs);
+	OUT_RING  (ps->filt | 0x2000 /* magic */);
+	OUT_RING  ((pt->width[0] << 16) | pt->height[0]);
+	OUT_RING  (ps->bcol);
+#endif
+}
+
+void
+nv10_fragtex_bind(struct nv10_context *nv10)
+{
+#if 0
+	struct nv10_fragment_program *fp = nv10->fragprog.active;
+	unsigned samplers, unit;
+
+	samplers = nv10->fp_samplers & ~fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		BEGIN_RING(celsius, NV10TCL_TX_ENABLE(unit), 1);
+		OUT_RING  (0);
+	}
+
+	samplers = nv10->dirty_samplers & fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		nv10_fragtex_build(nv10, unit);
+	}
+
+	nv10->fp_samplers = fp->samplers;
+#endif
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_miptree.c b/src/gallium/drivers/nv10/nv10_miptree.c
new file mode 100644
index 00000000000..943f9e21e98
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_miptree.c
@@ -0,0 +1,149 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv10_context.h"
+#include "nv10_screen.h"
+
+static void
+nv10_miptree_layout(struct nv10_miptree *nv10mt)
+{
+	struct pipe_texture *pt = &nv10mt->base;
+	boolean swizzled = FALSE;
+	uint width = pt->width[0], height = pt->height[0];
+	uint offset = 0;
+	int nr_faces, l, f;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		nr_faces = 6;
+	} else {
+		nr_faces = 1;
+	}
+	
+	for (l = 0; l <= pt->last_level; l++) {
+		pt->width[l] = width;
+		pt->height[l] = height;
+		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
+		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
+
+		if (swizzled)
+			nv10mt->level[l].pitch = pt->nblocksx[l] * pt->block.size;
+		else
+			nv10mt->level[l].pitch = pt->nblocksx[0] * pt->block.size;
+		nv10mt->level[l].pitch = (nv10mt->level[l].pitch + 63) & ~63;
+
+		nv10mt->level[l].image_offset =
+			CALLOC(nr_faces, sizeof(unsigned));
+
+		width  = MAX2(1, width  >> 1);
+		height = MAX2(1, height >> 1);
+
+	}
+
+	for (f = 0; f < nr_faces; f++) {
+		for (l = 0; l <= pt->last_level; l++) {
+			nv10mt->level[l].image_offset[f] = offset;
+			offset += nv10mt->level[l].pitch * pt->height[l];
+		}
+	}
+
+	nv10mt->total_size = offset;
+}
+
+static struct pipe_texture *
+nv10_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = screen->winsys;
+	struct nv10_miptree *mt;
+
+	mt = MALLOC(sizeof(struct nv10_miptree));
+	if (!mt)
+		return NULL;
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = screen;
+
+	nv10_miptree_layout(mt);
+
+	mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL,
+					   mt->total_size);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+	
+	return &mt->base;
+}
+
+static void
+nv10_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt)
+{
+	struct pipe_texture *mt = *pt;
+
+	*pt = NULL;
+	if (--mt->refcount <= 0) {
+		struct nv10_miptree *nv10mt = (struct nv10_miptree *)mt;
+		int l;
+
+		pipe_buffer_reference(screen, &nv10mt->buffer, NULL);
+		for (l = 0; l <= mt->last_level; l++) {
+			if (nv10mt->level[l].image_offset)
+				FREE(nv10mt->level[l].image_offset);
+		}
+		FREE(nv10mt);
+	}
+}
+
+static void
+nv10_miptree_update(struct pipe_context *pipe, struct pipe_texture *mt,
+		    uint face, uint levels)
+{
+}
+
+
+static struct pipe_surface *
+nv10_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct pipe_winsys *ws = screen->winsys;
+	struct nv10_miptree *nv10mt = (struct nv10_miptree *)pt;
+	struct pipe_surface *ps;
+
+	ps = ws->surface_alloc(ws);
+	if (!ps)
+		return NULL;
+	pipe_buffer_reference(screen, &ps->buffer, nv10mt->buffer);
+	ps->format = pt->format;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = nv10mt->level[level].pitch;
+	ps->refcount = 1;
+	ps->winsys = screen->winsys;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		ps->offset = nv10mt->level[level].image_offset[face];
+	} else {
+		ps->offset = nv10mt->level[level].image_offset[0];
+	}
+
+	return ps;
+}
+
+static void
+nv10_miptree_surface_release(struct pipe_screen *screen,
+			     struct pipe_surface **surface)
+{
+}
+
+void nv10_screen_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv10_miptree_create;
+	pscreen->texture_release = nv10_miptree_release;
+	pscreen->get_tex_surface = nv10_miptree_surface_get;
+	pscreen->tex_surface_release = nv10_miptree_surface_release;
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_prim_vbuf.c b/src/gallium/drivers/nv10/nv10_prim_vbuf.c
new file mode 100644
index 00000000000..e7e81d3dff8
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_prim_vbuf.c
@@ -0,0 +1,245 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * \file
+ * Build post-transformation, post-clipping vertex buffers and element
+ * lists by hooking into the end of the primitive pipeline and
+ * manipulating the vertex_id field in the vertex headers.
+ *
+ * XXX: work in progress 
+ * 
+ * \author José Fonseca <jrfonseca@tungstengraphics.com>
+ * \author Keith Whitwell <keith@tungstengraphics.com>
+ */
+
+
+#include "pipe/p_debug.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv10_context.h"
+#include "nv10_state.h"
+
+#include "draw/draw_vbuf.h"
+
+/**
+ * Primitive renderer for nv10.
+ */
+struct nv10_vbuf_render {
+	struct vbuf_render base;
+
+	struct nv10_context *nv10;   
+
+	/** Vertex buffer */
+	struct pipe_buffer* buffer;
+
+	/** Vertex size in bytes */
+	unsigned vertex_size;
+
+	/** Hardware primitive */
+	unsigned hwprim;
+};
+
+
+void nv10_vtxbuf_bind( struct nv10_context* nv10 )
+{
+	int i;
+	for(i = 0; i < 8; i++) {
+		BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_OFFSET(i), 1);
+		OUT_RING(0/*nv10->vtxbuf*/);
+		BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_ATTRIB_FORMAT(i) ,1);
+		OUT_RING(0/*XXX*/);
+	}
+}
+
+/**
+ * Basically a cast wrapper.
+ */
+static INLINE struct nv10_vbuf_render *
+nv10_vbuf_render( struct vbuf_render *render )
+{
+	assert(render);
+	return (struct nv10_vbuf_render *)render;
+}
+
+
+static const struct vertex_info *
+nv10_vbuf_render_get_vertex_info( struct vbuf_render *render )
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	struct nv10_context *nv10 = nv10_render->nv10;
+
+	nv10_emit_hw_state(nv10);
+
+	return &nv10->vertex_info;
+}
+
+
+static void *
+nv10_vbuf_render_allocate_vertices( struct vbuf_render *render,
+		ushort vertex_size,
+		ushort nr_vertices )
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	struct nv10_context *nv10 = nv10_render->nv10;
+	struct pipe_winsys *winsys = nv10->pipe.winsys;
+	size_t size = (size_t)vertex_size * (size_t)nr_vertices;
+
+	assert(!nv10_render->buffer);
+	nv10_render->buffer = winsys->buffer_create(winsys, 64, PIPE_BUFFER_USAGE_VERTEX, size);
+
+	nv10->dirty |= NV10_NEW_VTXARRAYS;
+
+	return winsys->buffer_map(winsys, 
+			nv10_render->buffer, 
+			PIPE_BUFFER_USAGE_CPU_WRITE);
+}
+
+
+static boolean
+nv10_vbuf_render_set_primitive( struct vbuf_render *render, 
+		unsigned prim )
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	unsigned hwp = nvgl_primitive(prim);
+	if (hwp == 0)
+		return FALSE;
+
+	nv10_render->hwprim = hwp;
+	return TRUE;
+}
+
+
+static void 
+nv10_vbuf_render_draw( struct vbuf_render *render,
+		const ushort *indices,
+		uint nr_indices)
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	struct nv10_context *nv10 = nv10_render->nv10;
+	int push, i;
+
+	nv10_emit_hw_state(nv10);
+
+	BEGIN_RING(celsius, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1);
+	OUT_RELOCl(nv10_render->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+
+	BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING(nv10_render->hwprim);
+
+	if (nr_indices & 1) {
+		BEGIN_RING(celsius, NV10TCL_VB_ELEMENT_U32, 1);
+		OUT_RING  (indices[0]);
+		indices++; nr_indices--;
+	}
+
+	while (nr_indices) {
+		// XXX too big/small ? check the size
+		push = MIN2(nr_indices, 1200 * 2);
+
+		BEGIN_RING_NI(celsius, NV10TCL_VB_ELEMENT_U16, push >> 1);
+		for (i = 0; i < push; i+=2)
+			OUT_RING((indices[i+1] << 16) | indices[i]);
+
+		nr_indices -= push;
+		indices  += push;
+	}
+
+	BEGIN_RING(celsius, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING  (0);
+}
+
+
+static void
+nv10_vbuf_render_release_vertices( struct vbuf_render *render,
+		void *vertices, 
+		unsigned vertex_size,
+		unsigned vertices_used )
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	struct nv10_context *nv10 = nv10_render->nv10;
+	struct pipe_winsys *winsys = nv10->pipe.winsys;
+	struct pipe_screen *pscreen = &nv10->screen->pipe;
+
+	assert(nv10_render->buffer);
+	winsys->buffer_unmap(winsys, nv10_render->buffer);
+	pipe_buffer_reference(pscreen, &nv10_render->buffer, NULL);
+}
+
+
+static void
+nv10_vbuf_render_destroy( struct vbuf_render *render )
+{
+	struct nv10_vbuf_render *nv10_render = nv10_vbuf_render(render);
+	FREE(nv10_render);
+}
+
+
+/**
+ * Create a new primitive render.
+ */
+static struct vbuf_render *
+nv10_vbuf_render_create( struct nv10_context *nv10 )
+{
+	struct nv10_vbuf_render *nv10_render = CALLOC_STRUCT(nv10_vbuf_render);
+
+	nv10_render->nv10 = nv10;
+
+	nv10_render->base.max_vertex_buffer_bytes = 16*1024;
+	nv10_render->base.max_indices = 1024;
+	nv10_render->base.get_vertex_info = nv10_vbuf_render_get_vertex_info;
+	nv10_render->base.allocate_vertices = nv10_vbuf_render_allocate_vertices;
+	nv10_render->base.set_primitive = nv10_vbuf_render_set_primitive;
+	nv10_render->base.draw = nv10_vbuf_render_draw;
+	nv10_render->base.release_vertices = nv10_vbuf_render_release_vertices;
+	nv10_render->base.destroy = nv10_vbuf_render_destroy;
+
+	return &nv10_render->base;
+}
+
+
+/**
+ * Create a new primitive vbuf/render stage.
+ */
+struct draw_stage *nv10_draw_vbuf_stage( struct nv10_context *nv10 )
+{
+	struct vbuf_render *render;
+	struct draw_stage *stage;
+
+	render = nv10_vbuf_render_create(nv10);
+	if(!render)
+		return NULL;
+
+	stage = draw_vbuf_stage( nv10->draw, render );
+	if(!stage) {
+		render->destroy(render);
+		return NULL;
+	}
+
+	return stage;
+}
diff --git a/src/gallium/drivers/nv10/nv10_screen.c b/src/gallium/drivers/nv10/nv10_screen.c
new file mode 100644
index 00000000000..27a9edf9bba
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_screen.c
@@ -0,0 +1,208 @@
+#include "pipe/p_screen.h"
+
+#include "nv10_context.h"
+#include "nv10_screen.h"
+
+static const char *
+nv10_screen_get_name(struct pipe_screen *screen)
+{
+	struct nv10_screen *nv10screen = nv10_screen(screen);
+	struct nouveau_device *dev = nv10screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv10_screen_get_vendor(struct pipe_screen *screen)
+{
+	return "nouveau";
+}
+
+static int
+nv10_screen_get_param(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 2;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 0;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 0;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 0;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 1;
+	case PIPE_CAP_POINT_SPRITE:
+		return 0;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 1;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 0;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 12;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 12;
+	case NOUVEAU_CAP_HW_VTXBUF:
+	case NOUVEAU_CAP_HW_IDXBUF:
+		return 0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv10_screen_get_paramf(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 10.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 64.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 2.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 4.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static boolean
+nv10_screen_is_format_supported(struct pipe_screen *screen,
+				enum pipe_format format,
+				enum pipe_texture_target target,
+				unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_A4R4G4B4_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static void *
+nv10_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
+		 unsigned flags )
+{
+	struct pipe_winsys *ws = screen->winsys;
+	void *map;
+
+	map = ws->buffer_map(ws, surface->buffer, flags);
+	if (!map)
+		return NULL;
+
+	return map + surface->offset;
+}
+
+static void
+nv10_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
+{
+	struct pipe_winsys *ws = screen->winsys;
+
+	ws->buffer_unmap(ws, surface->buffer);
+}
+
+static void
+nv10_screen_destroy(struct pipe_screen *pscreen)
+{
+	struct nv10_screen *screen = nv10_screen(pscreen);
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nvws->notifier_free(&screen->sync);
+	nvws->grobj_free(&screen->celsius);
+
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv10_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv10_screen *screen = CALLOC_STRUCT(nv10_screen);
+	unsigned celsius_class;
+	unsigned chipset = nvws->channel->device->chipset;
+	int ret;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	/* 3D object */
+	if (chipset>=0x20)
+		celsius_class=NV11TCL;
+	else if (chipset>=0x17)
+		celsius_class=NV17TCL;
+	else if (chipset>=0x11)
+		celsius_class=NV11TCL;
+	else
+		celsius_class=NV10TCL;
+
+	if (!celsius_class) {
+		NOUVEAU_ERR("Unknown nv1x chipset: nv%02x\n", chipset);
+		return NULL;
+	}
+
+	ret = nvws->grobj_alloc(nvws, celsius_class, &screen->celsius);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		return FALSE;
+	}
+
+	/* Notifier for sync purposes */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv10_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	screen->pipe.winsys = ws;
+	screen->pipe.destroy = nv10_screen_destroy;
+
+	screen->pipe.get_name = nv10_screen_get_name;
+	screen->pipe.get_vendor = nv10_screen_get_vendor;
+	screen->pipe.get_param = nv10_screen_get_param;
+	screen->pipe.get_paramf = nv10_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv10_screen_is_format_supported;
+
+	screen->pipe.surface_map = nv10_surface_map;
+	screen->pipe.surface_unmap = nv10_surface_unmap;
+
+	nv10_screen_init_miptree_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_screen.h b/src/gallium/drivers/nv10/nv10_screen.h
new file mode 100644
index 00000000000..3f8750a13f7
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_screen.h
@@ -0,0 +1,22 @@
+#ifndef __NV10_SCREEN_H__
+#define __NV10_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv10_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+
+	/* HW graphics objects */
+	struct nouveau_grobj *celsius;
+	struct nouveau_notifier *sync;
+};
+
+static INLINE struct nv10_screen *
+nv10_screen(struct pipe_screen *screen)
+{
+	return (struct nv10_screen *)screen;
+}
+
+#endif
diff --git a/src/gallium/drivers/nv10/nv10_state.c b/src/gallium/drivers/nv10/nv10_state.c
new file mode 100644
index 00000000000..d2375aa2f64
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_state.c
@@ -0,0 +1,588 @@
+#include "draw/draw_context.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv10_context.h"
+#include "nv10_state.h"
+
+static void *
+nv10_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv10_blend_state *cb;
+
+	cb = MALLOC(sizeof(struct nv10_blend_state));
+
+	cb->b_enable = cso->blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_dst_factor)));
+
+	cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
+
+	cb->d_enable = cso->dither ? 1 : 0;
+
+	return (void *)cb;
+}
+
+static void
+nv10_blend_state_bind(struct pipe_context *pipe, void *blend)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->blend = (struct nv10_blend_state*)blend;
+
+	nv10->dirty |= NV10_NEW_BLEND;
+}
+
+static void
+nv10_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV10TCL_TX_FORMAT_WRAP_S_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV10TCL_TX_FORMAT_WRAP_S_REPEAT;
+		break;
+	}
+
+	return ret >> NV10TCL_TX_FORMAT_WRAP_S_SHIFT;
+}
+
+static void *
+nv10_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv10_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv10_sampler_state));
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV10TCL_TX_FORMAT_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV10TCL_TX_FORMAT_WRAP_T_SHIFT));
+
+	ps->en = 0;
+	if (cso->max_anisotropy > 1.0) {
+		/* no idea, binary driver sets it, works without it.. meh.. */
+		ps->wrap |= (1 << 5);
+
+/*		if (cso->max_anisotropy >= 16.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_16X;
+		} else
+		if (cso->max_anisotropy >= 12.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_12X;
+		} else
+		if (cso->max_anisotropy >= 10.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_10X;
+		} else
+		if (cso->max_anisotropy >= 8.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_8X;
+		} else
+		if (cso->max_anisotropy >= 6.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_6X;
+		} else
+		if (cso->max_anisotropy >= 4.0) {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_4X;
+		} else {
+			ps->en |= NV10TCL_TX_ENABLE_ANISO_2X;
+		}*/
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV10TCL_TX_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV10TCL_TX_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV10TCL_TX_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV10TCL_TX_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+/*	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}*/
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv10_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv10->tex_sampler[unit] = sampler[unit];
+		nv10->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void
+nv10_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv10_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv10->tex_miptree[unit] = (struct nv10_miptree *)miptree[unit];
+		nv10->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void *
+nv10_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv10_rasterizer_state *rs;
+	int i;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	offset_cw/ccw -nohw
+	 * 	scissor
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 * 	offset_units / offset_scale
+	 */
+	rs = MALLOC(sizeof(struct nv10_rasterizer_state));
+
+	rs->templ = cso;
+	
+	rs->shade_model = cso->flatshade ? 0x1d00 : 0x1d01;
+
+	rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff;
+	rs->line_smooth_en = cso->line_smooth ? 1 : 0;
+
+	rs->point_size = *(uint32_t*)&cso->point_size;
+
+	rs->poly_smooth_en = cso->poly_smooth ? 1 : 0;
+
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		rs->front_face = NV10TCL_FRONT_FACE_CCW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_cw);
+	} else {
+		rs->front_face = NV10TCL_FRONT_FACE_CW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_ccw);
+	}
+
+	switch (cso->cull_mode) {
+	case PIPE_WINDING_CCW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CCW)
+			rs->cull_face    = NV10TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV10TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_CW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CW)
+			rs->cull_face    = NV10TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV10TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_BOTH:
+		rs->cull_face_en = 1;
+		rs->cull_face    = NV10TCL_CULL_FACE_FRONT_AND_BACK;
+		break;
+	case PIPE_WINDING_NONE:
+	default:
+		rs->cull_face_en = 0;
+		rs->cull_face    = 0;
+		break;
+	}
+
+	if (cso->point_sprite) {
+		rs->point_sprite = (1 << 0);
+		for (i = 0; i < 8; i++) {
+			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+				rs->point_sprite |= (1 << (8 + i));
+		}
+	} else {
+		rs->point_sprite = 0;
+	}
+
+	return (void *)rs;
+}
+
+static void
+nv10_rasterizer_state_bind(struct pipe_context *pipe, void *rast)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->rast = (struct nv10_rasterizer_state*)rast;
+
+	draw_set_rasterizer_state(nv10->draw, (nv10->rast ? nv10->rast->templ : NULL));
+
+	nv10->dirty |= NV10_NEW_RAST;
+}
+
+static void
+nv10_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv10_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv10_depth_stencil_alpha_state *hw;
+
+	hw = MALLOC(sizeof(struct nv10_depth_stencil_alpha_state));
+
+	hw->depth.func		= nvgl_comparison_op(cso->depth.func);
+	hw->depth.write_enable	= cso->depth.writemask ? 1 : 0;
+	hw->depth.test_enable	= cso->depth.enabled ? 1 : 0;
+
+	hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0;
+	hw->stencil.wmask = cso->stencil[0].write_mask;
+	hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func);
+	hw->stencil.ref	= cso->stencil[0].ref_value;
+	hw->stencil.vmask = cso->stencil[0].value_mask;
+	hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op);
+	hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op);
+	hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op);
+
+	hw->alpha.enabled = cso->alpha.enabled ? 1 : 0;
+	hw->alpha.func = nvgl_comparison_op(cso->alpha.func);
+	hw->alpha.ref  = float_to_ubyte(cso->alpha.ref);
+
+	return (void *)hw;
+}
+
+static void
+nv10_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->dsa = (struct nv10_depth_stencil_alpha_state*)dsa;
+
+	nv10->dirty |= NV10_NEW_DSA;
+}
+
+static void
+nv10_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv10_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *templ)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	return draw_create_vertex_shader(nv10->draw, templ);
+}
+
+static void
+nv10_vp_state_bind(struct pipe_context *pipe, void *shader)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_bind_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader);
+
+	nv10->dirty |= NV10_NEW_VERTPROG;
+}
+
+static void
+nv10_vp_state_delete(struct pipe_context *pipe, void *shader)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_delete_vertex_shader(nv10->draw, (struct draw_vertex_shader *) shader);
+}
+
+static void *
+nv10_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv10_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv10_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	
+	tgsi_scan_shader(cso->tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv10_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nv10_fragment_program *fp = hwcso;
+
+	nv10->fragprog.current = fp;
+	nv10->dirty |= NV10_NEW_FRAGPROG;
+}
+
+static void
+nv10_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nv10_fragment_program *fp = hwcso;
+
+	nv10_fragprog_destroy(nv10, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv10_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->blend_color = (struct pipe_blend_color*)bcol;
+
+	nv10->dirty |= NV10_NEW_BLENDCOL;
+}
+
+static void
+nv10_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	draw_set_clip_state(nv10->draw, clip);
+}
+
+static void
+nv10_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+
+	assert(shader < PIPE_SHADER_TYPES);
+	assert(index == 0);
+
+	if (buf) {
+		void *mapped;
+		if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ)))
+		{
+			memcpy(nv10->constbuf[shader], mapped, buf->size);
+			nv10->constbuf_nr[shader] =
+				buf->size / (4 * sizeof(float));
+			ws->buffer_unmap(ws, buf->buffer);
+		}
+	}
+}
+
+static void
+nv10_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->framebuffer = (struct pipe_framebuffer_state*)fb;
+
+	nv10->dirty |= NV10_NEW_FRAMEBUFFER;
+}
+
+static void
+nv10_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	NOUVEAU_ERR("line stipple hahaha\n");
+}
+
+static void
+nv10_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->scissor = (struct pipe_scissor_state*)s;
+
+	nv10->dirty |= NV10_NEW_SCISSOR;
+}
+
+static void
+nv10_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	nv10->viewport = (struct pipe_viewport_state*)vpt;
+
+	draw_set_viewport_state(nv10->draw, nv10->viewport);
+
+	nv10->dirty |= NV10_NEW_VIEWPORT;
+}
+
+static void
+nv10_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	memcpy(nv10->vtxbuf, vb, sizeof(*vb) * count);
+	nv10->dirty |= NV10_NEW_VTXARRAYS;
+
+	draw_set_vertex_buffers(nv10->draw, count, vb);
+}
+
+static void
+nv10_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+
+	memcpy(nv10->vtxelt, ve, sizeof(*ve) * count);
+	nv10->dirty |= NV10_NEW_VTXARRAYS;
+
+	draw_set_vertex_elements(nv10->draw, count, ve);
+}
+
+void
+nv10_init_state_functions(struct nv10_context *nv10)
+{
+	nv10->pipe.create_blend_state = nv10_blend_state_create;
+	nv10->pipe.bind_blend_state = nv10_blend_state_bind;
+	nv10->pipe.delete_blend_state = nv10_blend_state_delete;
+
+	nv10->pipe.create_sampler_state = nv10_sampler_state_create;
+	nv10->pipe.bind_sampler_states = nv10_sampler_state_bind;
+	nv10->pipe.delete_sampler_state = nv10_sampler_state_delete;
+	nv10->pipe.set_sampler_textures = nv10_set_sampler_texture;
+
+	nv10->pipe.create_rasterizer_state = nv10_rasterizer_state_create;
+	nv10->pipe.bind_rasterizer_state = nv10_rasterizer_state_bind;
+	nv10->pipe.delete_rasterizer_state = nv10_rasterizer_state_delete;
+
+	nv10->pipe.create_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_create;
+	nv10->pipe.bind_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_bind;
+	nv10->pipe.delete_depth_stencil_alpha_state =
+		nv10_depth_stencil_alpha_state_delete;
+
+	nv10->pipe.create_vs_state = nv10_vp_state_create;
+	nv10->pipe.bind_vs_state = nv10_vp_state_bind;
+	nv10->pipe.delete_vs_state = nv10_vp_state_delete;
+
+	nv10->pipe.create_fs_state = nv10_fp_state_create;
+	nv10->pipe.bind_fs_state = nv10_fp_state_bind;
+	nv10->pipe.delete_fs_state = nv10_fp_state_delete;
+
+	nv10->pipe.set_blend_color = nv10_set_blend_color;
+	nv10->pipe.set_clip_state = nv10_set_clip_state;
+	nv10->pipe.set_constant_buffer = nv10_set_constant_buffer;
+	nv10->pipe.set_framebuffer_state = nv10_set_framebuffer_state;
+	nv10->pipe.set_polygon_stipple = nv10_set_polygon_stipple;
+	nv10->pipe.set_scissor_state = nv10_set_scissor_state;
+	nv10->pipe.set_viewport_state = nv10_set_viewport_state;
+
+	nv10->pipe.set_vertex_buffers = nv10_set_vertex_buffers;
+	nv10->pipe.set_vertex_elements = nv10_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_state.h b/src/gallium/drivers/nv10/nv10_state.h
new file mode 100644
index 00000000000..3a3fd0d4f4f
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_state.h
@@ -0,0 +1,139 @@
+#ifndef __NV10_STATE_H__
+#define __NV10_STATE_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv10_blend_state {
+	uint32_t b_enable;
+	uint32_t b_srcfunc;
+	uint32_t b_dstfunc;
+
+	uint32_t c_mask;
+
+	uint32_t d_enable;
+};
+
+struct nv10_sampler_state {
+	uint32_t wrap;
+	uint32_t en;
+	uint32_t filt;
+	uint32_t bcol;
+};
+
+struct nv10_rasterizer_state {
+	uint32_t shade_model;
+
+	uint32_t line_width;
+	uint32_t line_smooth_en;
+
+	uint32_t point_size;
+
+	uint32_t poly_smooth_en;
+	
+	uint32_t poly_mode_front;
+	uint32_t poly_mode_back;
+
+	uint32_t front_face;
+	uint32_t cull_face;
+	uint32_t cull_face_en;
+
+	uint32_t point_sprite;
+
+	const struct pipe_rasterizer_state *templ;
+};
+
+struct nv10_vertex_program_exec {
+	uint32_t data[4];
+	boolean has_branch_offset;
+	int const_index;
+};
+
+struct nv10_vertex_program_data {
+	int index; /* immediates == -1 */
+	float value[4];
+};
+
+struct nv10_vertex_program {
+	const struct pipe_shader_state *pipe;
+
+	boolean translated;
+	struct nv10_vertex_program_exec *insns;
+	unsigned nr_insns;
+	struct nv10_vertex_program_data *consts;
+	unsigned nr_consts;
+
+	struct nouveau_resource *exec;
+	unsigned exec_start;
+	struct nouveau_resource *data;
+	unsigned data_start;
+	unsigned data_start_min;
+
+	uint32_t ir;
+	uint32_t or;
+};
+
+struct nv10_fragment_program_data {
+	unsigned offset;
+	unsigned index;
+};
+
+struct nv10_fragment_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+
+	boolean translated;
+	boolean on_hw;
+	unsigned samplers;
+
+	uint32_t *insn;
+	int       insn_len;
+
+	struct nv10_fragment_program_data *consts;
+	unsigned nr_consts;
+
+	struct pipe_buffer *buffer;
+
+	uint32_t fp_control;
+	uint32_t fp_reg_control;
+};
+
+
+struct nv10_depth_stencil_alpha_state {
+	struct {
+		uint32_t func;
+		uint32_t write_enable;
+		uint32_t test_enable;
+	} depth;
+
+	struct {
+		uint32_t enable;
+		uint32_t wmask;
+		uint32_t func;
+		uint32_t ref;
+		uint32_t vmask;
+		uint32_t fail;
+		uint32_t zfail;
+		uint32_t zpass;
+	} stencil;
+
+	struct {
+		uint32_t enabled;
+		uint32_t func;
+		uint32_t ref;
+	} alpha;
+};
+
+struct nv10_miptree {
+	struct pipe_texture base;
+
+	struct pipe_buffer *buffer;
+	uint total_size;
+
+	struct {
+		uint pitch;
+		uint *image_offset;
+	} level[PIPE_MAX_TEXTURE_LEVELS];
+};
+
+#endif
diff --git a/src/gallium/drivers/nv10/nv10_state_emit.c b/src/gallium/drivers/nv10/nv10_state_emit.c
new file mode 100644
index 00000000000..46c7e1d7536
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_state_emit.c
@@ -0,0 +1,303 @@
+#include "nv10_context.h"
+#include "nv10_state.h"
+
+static void nv10_state_emit_blend(struct nv10_context* nv10)
+{
+	struct nv10_blend_state *b = nv10->blend;
+
+	BEGIN_RING(celsius, NV10TCL_DITHER_ENABLE, 1);
+	OUT_RING  (b->d_enable);
+
+	BEGIN_RING(celsius, NV10TCL_BLEND_FUNC_ENABLE, 3);
+	OUT_RING  (b->b_enable);
+	OUT_RING  (b->b_srcfunc);
+	OUT_RING  (b->b_dstfunc);
+
+	BEGIN_RING(celsius, NV10TCL_COLOR_MASK, 1);
+	OUT_RING  (b->c_mask);
+}
+
+static void nv10_state_emit_blend_color(struct nv10_context* nv10)
+{
+	struct pipe_blend_color *c = nv10->blend_color;
+
+	BEGIN_RING(celsius, NV10TCL_BLEND_COLOR, 1);
+	OUT_RING  ((float_to_ubyte(c->color[3]) << 24)|
+		   (float_to_ubyte(c->color[0]) << 16)|
+		   (float_to_ubyte(c->color[1]) << 8) |
+		   (float_to_ubyte(c->color[2]) << 0));
+}
+
+static void nv10_state_emit_rast(struct nv10_context* nv10)
+{
+	struct nv10_rasterizer_state *r = nv10->rast;
+
+	BEGIN_RING(celsius, NV10TCL_SHADE_MODEL, 2);
+	OUT_RING  (r->shade_model);
+	OUT_RING  (r->line_width);
+
+
+	BEGIN_RING(celsius, NV10TCL_POINT_SIZE, 1);
+	OUT_RING  (r->point_size);
+
+	BEGIN_RING(celsius, NV10TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (r->poly_mode_front);
+	OUT_RING  (r->poly_mode_back);
+
+
+	BEGIN_RING(celsius, NV10TCL_CULL_FACE, 2);
+	OUT_RING  (r->cull_face);
+	OUT_RING  (r->front_face);
+
+	BEGIN_RING(celsius, NV10TCL_LINE_SMOOTH_ENABLE, 2);
+	OUT_RING  (r->line_smooth_en);
+	OUT_RING  (r->poly_smooth_en);
+
+	BEGIN_RING(celsius, NV10TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (r->cull_face_en);
+}
+
+static void nv10_state_emit_dsa(struct nv10_context* nv10)
+{
+	struct nv10_depth_stencil_alpha_state *d = nv10->dsa;
+
+	BEGIN_RING(celsius, NV10TCL_DEPTH_FUNC, 1);
+	OUT_RING (d->depth.func);
+
+	BEGIN_RING(celsius, NV10TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING (d->depth.write_enable);
+
+	BEGIN_RING(celsius, NV10TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING (d->depth.test_enable);
+
+#if 0
+	BEGIN_RING(celsius, NV10TCL_STENCIL_ENABLE, 1);
+	OUT_RING (d->stencil.enable);
+	BEGIN_RING(celsius, NV10TCL_STENCIL_MASK, 7);
+	OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7);
+#endif
+
+	BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING (d->alpha.enabled);
+
+	BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_FUNC, 1);
+	OUT_RING (d->alpha.func);
+
+	BEGIN_RING(celsius, NV10TCL_ALPHA_FUNC_REF, 1);
+	OUT_RING (d->alpha.ref);
+}
+
+static void nv10_state_emit_viewport(struct nv10_context* nv10)
+{
+}
+
+static void nv10_state_emit_scissor(struct nv10_context* nv10)
+{
+	// XXX this is so not working
+/*	struct pipe_scissor_state *s = nv10->scissor;
+	BEGIN_RING(celsius, NV10TCL_SCISSOR_HORIZ, 2);
+	OUT_RING  (((s->maxx - s->minx) << 16) | s->minx);
+	OUT_RING  (((s->maxy - s->miny) << 16) | s->miny);*/
+}
+
+static void nv10_state_emit_framebuffer(struct nv10_context* nv10)
+{
+	struct pipe_framebuffer_state* fb = nv10->framebuffer;
+	struct pipe_surface *rt, *zeta = NULL;
+	uint32_t rt_format, w, h;
+	int colour_format = 0, zeta_format = 0;
+
+	w = fb->cbufs[0]->width;
+	h = fb->cbufs[0]->height;
+	colour_format = fb->cbufs[0]->format;
+	rt = fb->cbufs[0];
+
+	if (fb->zsbuf) {
+		if (colour_format) {
+			assert(w == fb->zsbuf->width);
+			assert(h == fb->zsbuf->height);
+		} else {
+			w = fb->zsbuf->width;
+			h = fb->zsbuf->height;
+		}
+
+		zeta_format = fb->zsbuf->format;
+		zeta = fb->zsbuf;
+	}
+
+	rt_format = NV10TCL_RT_FORMAT_TYPE_LINEAR;
+
+	switch (colour_format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case 0:
+		rt_format |= NV10TCL_RT_FORMAT_COLOR_A8R8G8B8;
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		rt_format |= NV10TCL_RT_FORMAT_COLOR_R5G6B5;
+		break;
+	default:
+		assert(0);
+	}
+
+	if (zeta) {
+		BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1);
+		OUT_RING  (rt->stride | (zeta->stride << 16));
+	} else {
+		BEGIN_RING(celsius, NV10TCL_RT_PITCH, 1);
+		OUT_RING  (rt->stride | (rt->stride << 16));
+	}
+
+	nv10->rt[0] = rt->buffer;
+
+	if (zeta_format)
+	{
+		nv10->zeta = zeta->buffer;
+	}
+
+	BEGIN_RING(celsius, NV10TCL_RT_HORIZ, 3);
+	OUT_RING  ((w << 16) | 0);
+	OUT_RING  ((h << 16) | 0);
+	OUT_RING  (rt_format);
+	BEGIN_RING(celsius, NV10TCL_VIEWPORT_CLIP_HORIZ(0), 2);
+	OUT_RING  (((w - 1) << 16) | 0 | 0x08000800);
+	OUT_RING  (((h - 1) << 16) | 0 | 0x08000800);
+}
+
+static void nv10_vertex_layout(struct nv10_context *nv10)
+{
+	struct nv10_fragment_program *fp = nv10->fragprog.current;
+	uint32_t src = 0;
+	int i;
+	struct vertex_info vinfo;
+
+	memset(&vinfo, 0, sizeof(vinfo));
+
+	for (i = 0; i < fp->info.num_inputs; i++) {
+		switch (fp->info.input_semantic_name[i]) {
+			case TGSI_SEMANTIC_POSITION:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++);
+				break;
+			case TGSI_SEMANTIC_COLOR:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_LINEAR, src++);
+				break;
+			default:
+			case TGSI_SEMANTIC_GENERIC:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++);
+				break;
+			case TGSI_SEMANTIC_FOG:
+				draw_emit_vertex_attr(&vinfo, EMIT_4F, INTERP_PERSPECTIVE, src++);
+				break;
+		}
+	}
+	draw_compute_vertex_size(&vinfo);
+}
+
+void
+nv10_emit_hw_state(struct nv10_context *nv10)
+{
+	int i;
+
+	if (nv10->dirty & NV10_NEW_VERTPROG) {
+		//nv10_vertprog_bind(nv10, nv10->vertprog.current);
+		nv10->dirty &= ~NV10_NEW_VERTPROG;
+	}
+
+	if (nv10->dirty & NV10_NEW_FRAGPROG) {
+		nv10_fragprog_bind(nv10, nv10->fragprog.current);
+		/*XXX: clear NV10_NEW_FRAGPROG if no new program uploaded */
+		nv10->dirty_samplers |= (1<<10);
+		nv10->dirty_samplers = 0;
+	}
+
+	if (nv10->dirty_samplers || (nv10->dirty & NV10_NEW_FRAGPROG)) {
+		nv10_fragtex_bind(nv10);
+		nv10->dirty &= ~NV10_NEW_FRAGPROG;
+	}
+
+	if (nv10->dirty & NV10_NEW_VTXARRAYS) {
+		nv10->dirty &= ~NV10_NEW_VTXARRAYS;
+		nv10_vertex_layout(nv10);
+		nv10_vtxbuf_bind(nv10);
+	}
+
+	if (nv10->dirty & NV10_NEW_BLEND) {
+		nv10->dirty &= ~NV10_NEW_BLEND;
+		nv10_state_emit_blend(nv10);
+	}
+
+	if (nv10->dirty & NV10_NEW_BLENDCOL) {
+		nv10->dirty &= ~NV10_NEW_BLENDCOL;
+		nv10_state_emit_blend_color(nv10);
+	}
+
+	if (nv10->dirty & NV10_NEW_RAST) {
+		nv10->dirty &= ~NV10_NEW_RAST;
+		nv10_state_emit_rast(nv10);
+	}
+
+	if (nv10->dirty & NV10_NEW_DSA) {
+		nv10->dirty &= ~NV10_NEW_DSA;
+		nv10_state_emit_dsa(nv10);
+	}
+
+ 	if (nv10->dirty & NV10_NEW_VIEWPORT) {
+		nv10->dirty &= ~NV10_NEW_VIEWPORT;
+		nv10_state_emit_viewport(nv10);
+	}
+
+ 	if (nv10->dirty & NV10_NEW_SCISSOR) {
+		nv10->dirty &= ~NV10_NEW_SCISSOR;
+		nv10_state_emit_scissor(nv10);
+	}
+
+ 	if (nv10->dirty & NV10_NEW_FRAMEBUFFER) {
+		nv10->dirty &= ~NV10_NEW_FRAMEBUFFER;
+		nv10_state_emit_framebuffer(nv10);
+	}
+
+	/* Emit relocs for every referenced buffer.
+	 * This is to ensure the bufmgr has an accurate idea of how
+	 * the buffer is used.  This isn't very efficient, but we don't
+	 * seem to take a significant performance hit.  Will be improved
+	 * at some point.  Vertex arrays are emitted by nv10_vbo.c
+	 */
+
+	/* Render target */
+// XXX figre out who's who for NV10TCL_DMA_* and fill accordingly
+//	BEGIN_RING(celsius, NV10TCL_DMA_COLOR0, 1);
+//	OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	if (nv10->zeta) {
+// XXX
+//		BEGIN_RING(celsius, NV10TCL_DMA_ZETA, 1);
+//		OUT_RELOCo(nv10->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+		BEGIN_RING(celsius, NV10TCL_ZETA_OFFSET, 1);
+		OUT_RELOCl(nv10->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+		/* XXX for when we allocate LMA on nv17 */
+/*		BEGIN_RING(celsius, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1);
+		OUT_RELOCl(nv10->zeta + lma_offset);*/
+	}
+
+	/* Vertex buffer */
+	BEGIN_RING(celsius, NV10TCL_DMA_VTXBUF0, 1);
+	OUT_RELOCo(nv10->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(celsius, NV10TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(nv10->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	/* Texture images */
+	for (i = 0; i < 2; i++) {
+		if (!(nv10->fp_samplers & (1 << i)))
+			continue;
+		BEGIN_RING(celsius, NV10TCL_TX_OFFSET(i), 1);
+		OUT_RELOCl(nv10->tex[i].buffer, 0, NOUVEAU_BO_VRAM |
+			   NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+		BEGIN_RING(celsius, NV10TCL_TX_FORMAT(i), 1);
+		OUT_RELOCd(nv10->tex[i].buffer, nv10->tex[i].format,
+			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD |
+			   NOUVEAU_BO_OR, NV10TCL_TX_FORMAT_DMA0,
+			   NV10TCL_TX_FORMAT_DMA1);
+	}
+}
+
diff --git a/src/gallium/drivers/nv10/nv10_surface.c b/src/gallium/drivers/nv10/nv10_surface.c
new file mode 100644
index 00000000000..be44c7bed50
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_surface.c
@@ -0,0 +1,64 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "nv10_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+#include "util/u_tile.h"
+
+static void
+nv10_surface_copy(struct pipe_context *pipe, boolean do_flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nouveau_winsys *nvws = nv10->nvws;
+
+	nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+			   width, height);
+}
+
+static void
+nv10_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv10_context *nv10 = nv10_context(pipe);
+	struct nouveau_winsys *nvws = nv10->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+void
+nv10_init_surface_functions(struct nv10_context *nv10)
+{
+	nv10->pipe.surface_copy = nv10_surface_copy;
+	nv10->pipe.surface_fill = nv10_surface_fill;
+}
diff --git a/src/gallium/drivers/nv10/nv10_vbo.c b/src/gallium/drivers/nv10/nv10_vbo.c
new file mode 100644
index 00000000000..d0e788ac036
--- /dev/null
+++ b/src/gallium/drivers/nv10/nv10_vbo.c
@@ -0,0 +1,77 @@
+#include "draw/draw_context.h"
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv10_context.h"
+#include "nv10_state.h"
+
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_pushbuf.h"
+
+boolean nv10_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count)
+{
+	struct nv10_context *nv10 = nv10_context( pipe );
+	struct draw_context *draw = nv10->draw;
+	unsigned i;
+
+	nv10_emit_hw_state(nv10);
+
+	/*
+	 * Map vertex buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv10->vtxbuf[i].buffer) {
+			void *buf
+				= pipe->winsys->buffer_map(pipe->winsys,
+						nv10->vtxbuf[i].buffer,
+						PIPE_BUFFER_USAGE_CPU_READ);
+			draw_set_mapped_vertex_buffer(draw, i, buf);
+		}
+	}
+	/* Map index buffer, if present */
+	if (indexBuffer) {
+		void *mapped_indexes
+			= pipe->winsys->buffer_map(pipe->winsys, indexBuffer,
+					PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes);
+	}
+	else {
+		/* no index/element buffer */
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	draw_set_mapped_constant_buffer(draw,
+					nv10->constbuf[PIPE_SHADER_VERTEX],
+					nv10->constbuf_nr[PIPE_SHADER_VERTEX]);
+
+	/* draw! */
+	draw_arrays(nv10->draw, prim, start, count);
+
+	/*
+	 * unmap vertex/index buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv10->vtxbuf[i].buffer) {
+			pipe->winsys->buffer_unmap(pipe->winsys, nv10->vtxbuf[i].buffer);
+			draw_set_mapped_vertex_buffer(draw, i, NULL);
+		}
+	}
+	if (indexBuffer) {
+		pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer);
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	return TRUE;
+}
+
+boolean nv10_draw_arrays( struct pipe_context *pipe,
+				 unsigned prim, unsigned start, unsigned count)
+{
+	return nv10_draw_elements(pipe, NULL, 0, prim, start, count);
+}
+
+
+
diff --git a/src/gallium/drivers/nv20/Makefile b/src/gallium/drivers/nv20/Makefile
new file mode 100644
index 00000000000..d777fd3d8b4
--- /dev/null
+++ b/src/gallium/drivers/nv20/Makefile
@@ -0,0 +1,29 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv20
+
+DRIVER_SOURCES = \
+	nv20_clear.c \
+	nv20_context.c \
+	nv20_fragprog.c \
+	nv20_fragtex.c \
+	nv20_miptree.c \
+	nv20_prim_vbuf.c \
+	nv20_screen.c \
+	nv20_state.c \
+	nv20_state_emit.c \
+	nv20_surface.c \
+	nv20_vbo.c
+#	nv20_vertprog.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv20/nv20_clear.c b/src/gallium/drivers/nv20/nv20_clear.c
new file mode 100644
index 00000000000..81b6f3e78ac
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_clear.c
@@ -0,0 +1,12 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv20_context.h"
+
+void
+nv20_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue);
+}
diff --git a/src/gallium/drivers/nv20/nv20_context.c b/src/gallium/drivers/nv20/nv20_context.c
new file mode 100644
index 00000000000..9a17f4af571
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_context.c
@@ -0,0 +1,419 @@
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv20_context.h"
+#include "nv20_screen.h"
+
+static void
+nv20_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_flush(nv20->draw);
+
+	FIRE_RING(fence);
+}
+
+static void
+nv20_destroy(struct pipe_context *pipe)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	if (nv20->draw)
+		draw_destroy(nv20->draw);
+
+	FREE(nv20);
+}
+
+static void nv20_init_hwctx(struct nv20_context *nv20)
+{
+	struct nv20_screen *screen = nv20->screen;
+	struct nouveau_winsys *nvws = screen->nvws;
+	int i;
+	float projectionmatrix[16];
+	const boolean is_nv25tcl = (nv20->screen->kelvin->grclass == NV25TCL);
+
+	BEGIN_RING(kelvin, NV20TCL_DMA_NOTIFY, 1);
+	OUT_RING  (screen->sync->handle);
+	BEGIN_RING(kelvin, NV20TCL_DMA_TEXTURE0, 2);
+	OUT_RING  (nvws->channel->vram->handle);
+	OUT_RING  (nvws->channel->gart->handle); /* TEXTURE1 */
+	BEGIN_RING(kelvin, NV20TCL_DMA_COLOR, 2);
+	OUT_RING  (nvws->channel->vram->handle);
+	OUT_RING  (nvws->channel->vram->handle); /* ZETA */
+
+	BEGIN_RING(kelvin, NV20TCL_DMA_QUERY, 1);
+	OUT_RING  (0); /* renouveau: beef0351, unique */
+
+	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 1);
+	OUT_RING  ((0xfff << 16) | 0x0);
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_VERT(0), 1);
+	OUT_RING  ((0xfff << 16) | 0x0);
+
+	for (i = 1; i < NV20TCL_VIEWPORT_CLIP_HORIZ__SIZE; i++) {
+		BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(i), 1);
+		OUT_RING  (0);
+		BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_VERT(i), 1);
+		OUT_RING  (0);
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_MODE, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(kelvin, 0x17e0, 3);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+
+	if (is_nv25tcl) {
+		BEGIN_RING(kelvin, NV20TCL_TX_RCOMP, 1);
+		OUT_RING  (NV20TCL_TX_RCOMP_LEQUAL | 0xdb0);
+	} else {
+		BEGIN_RING(kelvin, 0x1e68, 1);
+		OUT_RING  (0x4b800000); /* 16777216.000000 */
+		BEGIN_RING(kelvin, NV20TCL_TX_RCOMP, 1);
+		OUT_RING  (NV20TCL_TX_RCOMP_LEQUAL);
+	}
+
+	BEGIN_RING(kelvin, 0x290, 1);
+	OUT_RING  ((0x10 << 16) | 1);
+	BEGIN_RING(kelvin, 0x9fc, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, 0x1d80, 1);
+	OUT_RING  (1);
+	BEGIN_RING(kelvin, 0x9f8, 1);
+	OUT_RING  (4);
+	BEGIN_RING(kelvin, 0x17ec, 3);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	OUT_RINGf (0.0);
+
+	if (is_nv25tcl) {
+		BEGIN_RING(kelvin, 0x1d88, 1);
+		OUT_RING  (3);
+
+		BEGIN_RING(kelvin, NV25TCL_DMA_IN_MEMORY9, 1);
+		OUT_RING  (nvws->channel->vram->handle);
+		BEGIN_RING(kelvin, NV25TCL_DMA_IN_MEMORY8, 1);
+		OUT_RING  (nvws->channel->vram->handle);
+	}
+	BEGIN_RING(kelvin, NV20TCL_DMA_FENCE, 1);
+	OUT_RING  (0);	/* renouveau: beef1e10 */
+
+	BEGIN_RING(kelvin, 0x1e98, 1);
+	OUT_RING  (0);
+#if 0
+	if (is_nv25tcl) {
+		BEGIN_RING(NvSub3D, NV25TCL_DMA_IN_MEMORY4, 2);
+		OUT_RING  (NvDmaTT);	/* renouveau: beef0202 */
+		OUT_RING  (NvDmaFB);	/* renouveau: beef0201 */
+
+		BEGIN_RING(NvSub3D, NV20TCL_DMA_TEXTURE1, 1);
+		OUT_RING  (NvDmaTT);	/* renouveau: beef0202 */
+	}
+#endif
+	BEGIN_RING(kelvin, NV20TCL_NOTIFY, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(kelvin, 0x120, 3);
+	OUT_RING  (0);
+	OUT_RING  (1);
+	OUT_RING  (2);
+
+/* error: ILLEGAL_MTHD, PROTECTION_FAULT
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_TRANSLATE_X, 4);
+	OUT_RINGf (0.0);
+	OUT_RINGf (512.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+*/
+
+	if (is_nv25tcl) {
+		BEGIN_RING(kelvin, 0x022c, 2);
+		OUT_RING  (0x280);
+		OUT_RING  (0x07d28000);
+	}
+
+/* * illegal method, protection fault
+	BEGIN_RING(NvSub3D, 0x1c2c, 1);
+	OUT_RING  (0); */
+
+	if (is_nv25tcl) {
+		BEGIN_RING(kelvin, 0x1da4, 1);
+		OUT_RING  (0);
+	}
+
+/* * crashes with illegal method, protection fault
+	BEGIN_RING(NvSub3D, 0x1c18, 1);
+	OUT_RING  (0x200); */
+
+	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 2);
+	OUT_RING  ((0 << 16) | 0);
+	OUT_RING  ((0 << 16) | 0);
+
+	/* *** Set state *** */
+
+	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_FUNC, 2);
+	OUT_RING  (NV20TCL_ALPHA_FUNC_FUNC_ALWAYS);
+	OUT_RING  (0);			/* NV20TCL_ALPHA_FUNC_REF */
+
+	for (i = 0; i < NV20TCL_TX_ENABLE__SIZE; ++i) {
+		BEGIN_RING(kelvin, NV20TCL_TX_ENABLE(i), 1);
+		OUT_RING  (0);
+	}
+	BEGIN_RING(kelvin, NV20TCL_TX_SHADER_OP, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_TX_SHADER_CULL_MODE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_IN_ALPHA(0), 4);
+	OUT_RING  (0x30d410d0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_OUT_RGB(0), 4);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_ENABLE, 1);
+	OUT_RING  (0x00011101);
+	BEGIN_RING(kelvin, NV20TCL_RC_FINAL0, 2);
+	OUT_RING  (0x130e0300);
+	OUT_RING  (0x0c091c80);
+	BEGIN_RING(kelvin, NV20TCL_RC_OUT_ALPHA(0), 4);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_IN_RGB(0), 4);
+	OUT_RING  (0x20c400c0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_COLOR0, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_RC_CONSTANT_COLOR0(0), 4);
+	OUT_RING  (0x035125a0);
+	OUT_RING  (0);
+	OUT_RING  (0x40002000);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_MULTISAMPLE_CONTROL, 1);
+	OUT_RING  (0xffff0000);
+
+	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_DITHER_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_STENCIL_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_SRC, 4);
+	OUT_RING  (NV20TCL_BLEND_FUNC_SRC_ONE);
+	OUT_RING  (NV20TCL_BLEND_FUNC_DST_ZERO);
+	OUT_RING  (0);			/* NV20TCL_BLEND_COLOR */
+	OUT_RING  (NV20TCL_BLEND_EQUATION_FUNC_ADD);
+	BEGIN_RING(kelvin, NV20TCL_STENCIL_MASK, 7);
+	OUT_RING  (0xff);
+	OUT_RING  (NV20TCL_STENCIL_FUNC_FUNC_ALWAYS);
+	OUT_RING  (0);			/* NV20TCL_STENCIL_FUNC_REF */
+	OUT_RING  (0xff);		/* NV20TCL_STENCIL_FUNC_MASK */
+	OUT_RING  (NV20TCL_STENCIL_OP_FAIL_KEEP);
+	OUT_RING  (NV20TCL_STENCIL_OP_ZFAIL_KEEP);
+	OUT_RING  (NV20TCL_STENCIL_OP_ZPASS_KEEP);
+
+	BEGIN_RING(kelvin, NV20TCL_COLOR_LOGIC_OP_ENABLE, 2);
+	OUT_RING  (0);
+	OUT_RING  (NV20TCL_COLOR_LOGIC_OP_OP_COPY);
+	BEGIN_RING(kelvin, 0x17cc, 1);
+	OUT_RING  (0);
+	if (is_nv25tcl) {
+		BEGIN_RING(kelvin, 0x1d84, 1);
+		OUT_RING  (1);
+	}
+	BEGIN_RING(kelvin, NV20TCL_LIGHTING_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_LIGHT_CONTROL, 1);
+	OUT_RING  (0x00020000);
+	BEGIN_RING(kelvin, NV20TCL_SEPARATE_SPECULAR_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_LIGHT_MODEL_TWO_SIDE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_ENABLED_LIGHTS, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_NORMALIZE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_STIPPLE_PATTERN(0),
+					NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE);
+	for (i = 0; i < NV20TCL_POLYGON_STIPPLE_PATTERN__SIZE; ++i) {
+		OUT_RING(0xffffffff);
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	OUT_RING  (0);
+	OUT_RING  (0);		/* NV20TCL.POLYGON_OFFSET_LINE_ENABLE */
+	OUT_RING  (0);		/* NV20TCL.POLYGON_OFFSET_FILL_ENABLE */
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_FUNC, 1);
+	OUT_RING  (NV20TCL_DEPTH_FUNC_LESS);
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING  (1);
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_OFFSET_FACTOR, 2);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);	/* NV20TCL.POLYGON_OFFSET_UNITS */
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_UNK17D8, 1);
+	OUT_RING  (1);
+	if (!is_nv25tcl) {
+		BEGIN_RING(kelvin, 0x1d84, 1);
+		OUT_RING  (3);
+	}
+	BEGIN_RING(kelvin, NV20TCL_POINT_SIZE, 1);
+	if (!is_nv25tcl) {
+		OUT_RING  (8);
+	} else {
+		OUT_RINGf (1.0);
+	}
+	if (!is_nv25tcl) {
+		BEGIN_RING(kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 2);
+		OUT_RING  (0);
+		OUT_RING  (0);		/* NV20TCL.POINT_SMOOTH_ENABLE */
+	} else {
+		BEGIN_RING(kelvin, NV20TCL_POINT_PARAMETERS_ENABLE, 1);
+		OUT_RING  (0);
+		BEGIN_RING(kelvin, 0x0a1c, 1);
+		OUT_RING  (0x800);
+	}
+	BEGIN_RING(kelvin, NV20TCL_LINE_WIDTH, 1);
+	OUT_RING  (8);
+	BEGIN_RING(kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (NV20TCL_POLYGON_MODE_FRONT_FILL);
+	OUT_RING  (NV20TCL_POLYGON_MODE_BACK_FILL);
+	BEGIN_RING(kelvin, NV20TCL_CULL_FACE, 2);
+	OUT_RING  (NV20TCL_CULL_FACE_BACK);
+	OUT_RING  (NV20TCL_FRONT_FACE_CCW);
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_SMOOTH_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_SHADE_MODEL, 1);
+	OUT_RING  (NV20TCL_SHADE_MODEL_SMOOTH);
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_STIPPLE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING(kelvin, NV20TCL_TX_GEN_S(0), 4 * NV20TCL_TX_GEN_S__SIZE);
+	for (i=0; i < 4 * NV20TCL_TX_GEN_S__SIZE; ++i) {
+		OUT_RING(0);
+	}
+	BEGIN_RING(kelvin, NV20TCL_FOG_EQUATION_CONSTANT, 3);
+	OUT_RINGf (1.5);
+	OUT_RINGf (-0.090168);		/* NV20TCL.FOG_EQUATION_LINEAR */
+	OUT_RINGf (0.0);		/* NV20TCL.FOG_EQUATION_QUADRATIC */
+	BEGIN_RING(kelvin, NV20TCL_FOG_MODE, 2);
+	OUT_RING  (NV20TCL_FOG_MODE_EXP_2);
+	OUT_RING  (NV20TCL_FOG_COORD_DIST_COORD_FOG);
+	BEGIN_RING(kelvin, NV20TCL_FOG_ENABLE, 2);
+	OUT_RING  (0);
+	OUT_RING  (0);			/* NV20TCL.FOG_COLOR */
+	BEGIN_RING(kelvin, NV20TCL_ENGINE, 1);
+	OUT_RING  (NV20TCL_ENGINE_FIXED);
+
+	for (i = 0; i < NV20TCL_TX_MATRIX_ENABLE__SIZE; ++i) {
+		BEGIN_RING(kelvin, NV20TCL_TX_MATRIX_ENABLE(i), 1);
+		OUT_RING  (0);
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_VTX_ATTR_4F_X(1), 4 * 15);
+	OUT_RINGf(1.0); OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(1.0);
+	OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(1.0); OUT_RINGf(1.0);
+	OUT_RINGf(1.0); OUT_RINGf(1.0); OUT_RINGf(1.0); OUT_RINGf(1.0);
+	for (i = 4; i < 16; ++i) {
+		OUT_RINGf(0.0); OUT_RINGf(0.0); OUT_RINGf(0.0);	OUT_RINGf(1.0);
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_EDGEFLAG_ENABLE, 1);
+	OUT_RING  (1);
+	BEGIN_RING(kelvin, NV20TCL_COLOR_MASK, 1);
+	OUT_RING (0x00010101);
+	BEGIN_RING(kelvin, NV20TCL_CLEAR_VALUE, 1);
+	OUT_RING (0);
+
+	memset(projectionmatrix, 0, sizeof(projectionmatrix));
+	projectionmatrix[0*4+0] = 1.0;
+	projectionmatrix[1*4+1] = 1.0;
+	projectionmatrix[2*4+2] = 1.0;
+	projectionmatrix[3*4+3] = 1.0;
+	BEGIN_RING(kelvin, NV20TCL_PROJECTION_MATRIX(0), 16);
+	for (i = 0; i < 16; i++) {
+		OUT_RINGf  (projectionmatrix[i]);
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_RANGE_NEAR, 2);
+	OUT_RINGf  (0.0);
+	OUT_RINGf  (16777216.0); /* bpp dependant? */
+
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_SCALE0_X, 4);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (16777215.0 * 0.5);
+	OUT_RING  (0);
+
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_SCALE1_X, 4);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (-2048.0);
+	OUT_RINGf  (16777215.0 * 0.5);
+	OUT_RING  (0);
+
+	FIRE_RING (NULL);
+}
+
+static void
+nv20_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+}
+
+struct pipe_context *
+nv20_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct nv20_screen *screen = nv20_screen(pscreen);
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv20_context *nv20;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nv20 = CALLOC(1, sizeof(struct nv20_context));
+	if (!nv20)
+		return NULL;
+	nv20->screen = screen;
+	nv20->pctx_id = pctx_id;
+
+	nv20->nvws = nvws;
+
+	nv20->pipe.winsys = ws;
+	nv20->pipe.screen = pscreen;
+	nv20->pipe.destroy = nv20_destroy;
+	nv20->pipe.set_edgeflags = nv20_set_edgeflags;
+	nv20->pipe.draw_arrays = nv20_draw_arrays;
+	nv20->pipe.draw_elements = nv20_draw_elements;
+	nv20->pipe.clear = nv20_clear;
+	nv20->pipe.flush = nv20_flush;
+
+	nv20_init_surface_functions(nv20);
+	nv20_init_state_functions(nv20);
+
+	nv20->draw = draw_create();
+	assert(nv20->draw);
+	draw_set_rasterize_stage(nv20->draw, nv20_draw_vbuf_stage(nv20));
+
+	nv20_init_hwctx(nv20);
+
+	return &nv20->pipe;
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_context.h b/src/gallium/drivers/nv20/nv20_context.h
new file mode 100644
index 00000000000..8ad926db20a
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_context.h
@@ -0,0 +1,153 @@
+#ifndef __NV20_CONTEXT_H__
+#define __NV20_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv20_screen *ctx = nv20->screen
+#include "nouveau/nouveau_push.h"
+
+#include "nv20_state.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+#define NV20_NEW_VERTPROG	(1 << 0)
+#define NV20_NEW_FRAGPROG	(1 << 1)
+#define NV20_NEW_VTXARRAYS	(1 << 2)
+#define NV20_NEW_BLEND		(1 << 3)
+#define NV20_NEW_BLENDCOL	(1 << 4)
+#define NV20_NEW_RAST 		(1 << 5)
+#define NV20_NEW_DSA  		(1 << 6)
+#define NV20_NEW_VIEWPORT	(1 << 7)
+#define NV20_NEW_SCISSOR	(1 << 8)
+#define NV20_NEW_FRAMEBUFFER	(1 << 9)
+
+#include "nv20_screen.h"
+
+struct nv20_context {
+	struct pipe_context pipe;
+
+	struct nouveau_winsys *nvws;
+	struct nv20_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	uint32_t dirty;
+
+	struct nv20_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS];
+	struct nv20_miptree *tex_miptree[PIPE_MAX_SAMPLERS];
+	unsigned dirty_samplers;
+	unsigned fp_samplers;
+	unsigned vp_samplers;
+
+	uint32_t rt_enable;
+	struct pipe_buffer *rt[4];
+	struct pipe_buffer *zeta;
+	uint32_t lma_offset;
+
+	struct nv20_blend_state *blend;
+	struct pipe_blend_color *blend_color;
+	struct nv20_rasterizer_state *rast;
+	struct nv20_depth_stencil_alpha_state *dsa;
+	struct pipe_viewport_state *viewport;
+	struct pipe_scissor_state *scissor;
+	struct pipe_framebuffer_state *framebuffer;
+
+	//struct pipe_buffer *constbuf[PIPE_SHADER_TYPES];
+	float *constbuf[PIPE_SHADER_TYPES][32][4];
+	unsigned constbuf_nr[PIPE_SHADER_TYPES];
+
+	struct vertex_info vertex_info;
+
+	struct {
+		struct pipe_buffer *buffer;
+		uint32_t format;
+	} tex[2];
+
+	unsigned vb_enable;
+	struct {
+		struct pipe_buffer *buffer;
+		unsigned delta;
+	} vb[16];
+
+/*	struct {
+	
+		struct nouveau_resource *exec_heap;
+		struct nouveau_resource *data_heap;
+
+		struct nv20_vertex_program *active;
+
+		struct nv20_vertex_program *current;
+	} vertprog;
+*/
+	struct {
+		struct nv20_fragment_program *active;
+
+		struct nv20_fragment_program *current;
+		struct pipe_buffer *constant_buf;
+	} fragprog;
+
+	struct pipe_vertex_buffer  vtxbuf[PIPE_MAX_ATTRIBS];
+	struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS];
+};
+
+static INLINE struct nv20_context *
+nv20_context(struct pipe_context *pipe)
+{
+	return (struct nv20_context *)pipe;
+}
+
+extern void nv20_init_state_functions(struct nv20_context *nv20);
+extern void nv20_init_surface_functions(struct nv20_context *nv20);
+
+extern void nv20_screen_init_miptree_functions(struct pipe_screen *pscreen);
+
+/* nv20_clear.c */
+extern void nv20_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+/* nv20_draw.c */
+extern struct draw_stage *nv20_draw_render_stage(struct nv20_context *nv20);
+
+/* nv20_fragprog.c */
+extern void nv20_fragprog_bind(struct nv20_context *,
+			       struct nv20_fragment_program *);
+extern void nv20_fragprog_destroy(struct nv20_context *,
+				  struct nv20_fragment_program *);
+
+/* nv20_fragtex.c */
+extern void nv20_fragtex_bind(struct nv20_context *);
+
+/* nv20_prim_vbuf.c */
+struct draw_stage *nv20_draw_vbuf_stage( struct nv20_context *nv20 );
+extern void nv20_vtxbuf_bind(struct nv20_context* nv20);
+
+/* nv20_state.c and friends */
+extern void nv20_emit_hw_state(struct nv20_context *nv20);
+extern void nv20_state_tex_update(struct nv20_context *nv20);
+
+/* nv20_vbo.c */
+extern boolean nv20_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv20_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count);
+
+
+#endif
diff --git a/src/gallium/drivers/nv20/nv20_fragprog.c b/src/gallium/drivers/nv20/nv20_fragprog.c
new file mode 100644
index 00000000000..4f496369dd3
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_fragprog.c
@@ -0,0 +1,21 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv20_context.h"
+
+void
+nv20_fragprog_bind(struct nv20_context *nv20, struct nv20_fragment_program *fp)
+{
+}
+
+void
+nv20_fragprog_destroy(struct nv20_context *nv20,
+		      struct nv20_fragment_program *fp)
+{
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_fragtex.c b/src/gallium/drivers/nv20/nv20_fragtex.c
new file mode 100644
index 00000000000..495a7be9127
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_fragtex.c
@@ -0,0 +1,124 @@
+#include "nv20_context.h"
+#include "nouveau/nouveau_util.h"
+
+#define _(m,tf)                                                                \
+{                                                                              \
+  TRUE,                                                                        \
+  PIPE_FORMAT_##m,                                                             \
+  NV20TCL_TX_FORMAT_FORMAT_##tf,                                               \
+}
+
+struct nv20_texture_format {
+	boolean defined;
+	uint	pipe;
+	int     format;
+};
+
+static struct nv20_texture_format
+nv20_texture_formats[] = {
+	_(A8R8G8B8_UNORM, A8R8G8B8),
+	_(A1R5G5B5_UNORM, A1R5G5B5),
+	_(A4R4G4B4_UNORM, A4R4G4B4),
+	_(L8_UNORM      , L8      ),
+	_(A8_UNORM      , A8      ),
+	_(A8L8_UNORM    , A8L8    ),
+/*	_(RGB_DXT1      , DXT1,   ), */
+/*	_(RGBA_DXT1     , DXT1,   ), */
+/*	_(RGBA_DXT3     , DXT3,   ), */
+/*	_(RGBA_DXT5     , DXT5,   ), */
+	{},
+};
+
+static struct nv20_texture_format *
+nv20_fragtex_format(uint pipe_format)
+{
+	struct nv20_texture_format *tf = nv20_texture_formats;
+
+	while (tf->defined) {
+		if (tf->pipe == pipe_format)
+			return tf;
+		tf++;
+	}
+
+	return NULL;
+}
+
+
+static void
+nv20_fragtex_build(struct nv20_context *nv20, int unit)
+{
+#if 0
+	struct nv20_sampler_state *ps = nv20->tex_sampler[unit];
+	struct nv20_miptree *nv20mt = nv20->tex_miptree[unit];
+	struct pipe_texture *pt = &nv20mt->base;
+	struct nv20_texture_format *tf;
+	uint32_t txf, txs, txp;
+
+	tf = nv20_fragtex_format(pt->format);
+	if (!tf || !tf->defined) {
+		NOUVEAU_ERR("Unsupported texture format: 0x%x\n", pt->format);
+		return;
+	}
+
+	txf  = tf->format << 8;
+	txf |= (pt->last_level + 1) << 16;
+	txf |= log2i(pt->width[0]) << 20;
+	txf |= log2i(pt->height[0]) << 24;
+	txf |= log2i(pt->depth[0]) << 28;
+	txf |= 8;
+
+	switch (pt->target) {
+	case PIPE_TEXTURE_CUBE:
+		txf |= NV10TCL_TX_FORMAT_CUBE_MAP;
+		/* fall-through */
+	case PIPE_TEXTURE_2D:
+		txf |= (2<<4);
+		break;
+	case PIPE_TEXTURE_1D:
+		txf |= (1<<4);
+		break;
+	default:
+		NOUVEAU_ERR("Unknown target %d\n", pt->target);
+		return;
+	}
+
+	BEGIN_RING(kelvin, NV10TCL_TX_OFFSET(unit), 8);
+	OUT_RELOCl(nv20mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+	OUT_RELOCd(nv20mt->buffer,txf,NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_OR | NOUVEAU_BO_RD, 1/*VRAM*/,2/*TT*/);
+	OUT_RING  (ps->wrap);
+	OUT_RING  (0x40000000); /* enable */
+	OUT_RING  (txs);
+	OUT_RING  (ps->filt | 0x2000 /* magic */);
+	OUT_RING  ((pt->width[0] << 16) | pt->height[0]);
+	OUT_RING  (ps->bcol);
+#endif
+}
+
+void
+nv20_fragtex_bind(struct nv20_context *nv20)
+{
+#if 0
+	struct nv20_fragment_program *fp = nv20->fragprog.active;
+	unsigned samplers, unit;
+
+	samplers = nv20->fp_samplers & ~fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		BEGIN_RING(kelvin, NV10TCL_TX_ENABLE(unit), 1);
+		OUT_RING  (0);
+	}
+
+	samplers = nv20->dirty_samplers & fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		nv20_fragtex_build(nv20, unit);
+	}
+
+	nv20->fp_samplers = fp->samplers;
+#endif
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_miptree.c b/src/gallium/drivers/nv20/nv20_miptree.c
new file mode 100644
index 00000000000..c6106d58c43
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_miptree.c
@@ -0,0 +1,156 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv20_context.h"
+#include "nv20_screen.h"
+
+static void
+nv20_miptree_layout(struct nv20_miptree *nv20mt)
+{
+	struct pipe_texture *pt = &nv20mt->base;
+	boolean swizzled = FALSE;
+	uint width = pt->width[0], height = pt->height[0];
+	uint offset = 0;
+	int nr_faces, l, f;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		nr_faces = 6;
+	} else {
+		nr_faces = 1;
+	}
+	
+	for (l = 0; l <= pt->last_level; l++) {
+		pt->width[l] = width;
+		pt->height[l] = height;
+		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
+		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
+
+		if (swizzled)
+			nv20mt->level[l].pitch = pt->nblocksx[l] * pt->block.size;
+		else
+			nv20mt->level[l].pitch = pt->nblocksx[0] * pt->block.size;
+		nv20mt->level[l].pitch = (nv20mt->level[l].pitch + 63) & ~63;
+
+		nv20mt->level[l].image_offset =
+			CALLOC(nr_faces, sizeof(unsigned));
+
+		width  = MAX2(1, width  >> 1);
+		height = MAX2(1, height >> 1);
+
+	}
+
+	for (f = 0; f < nr_faces; f++) {
+		for (l = 0; l <= pt->last_level; l++) {
+			nv20mt->level[l].image_offset[f] = offset;
+			offset += nv20mt->level[l].pitch * pt->height[l];
+		}
+	}
+
+	nv20mt->total_size = offset;
+}
+
+static struct pipe_texture *
+nv20_miptree_create(struct pipe_screen *screen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = screen->winsys;
+	struct nv20_miptree *mt;
+
+	mt = MALLOC(sizeof(struct nv20_miptree));
+	if (!mt)
+		return NULL;
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = screen;
+
+	nv20_miptree_layout(mt);
+
+	mt->buffer = ws->buffer_create(ws, 256, PIPE_BUFFER_USAGE_PIXEL,
+					   mt->total_size);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+	
+	return &mt->base;
+}
+
+static void
+nv20_miptree_release(struct pipe_screen *screen, struct pipe_texture **pt)
+{
+	struct pipe_texture *mt = *pt;
+
+	*pt = NULL;
+	if (--mt->refcount <= 0) {
+		struct nv20_miptree *nv20mt = (struct nv20_miptree *)mt;
+		int l;
+
+		pipe_buffer_reference(screen, &nv20mt->buffer, NULL);
+		for (l = 0; l <= mt->last_level; l++) {
+			if (nv20mt->level[l].image_offset)
+				FREE(nv20mt->level[l].image_offset);
+		}
+		FREE(nv20mt);
+	}
+}
+
+static struct pipe_surface *
+nv20_miptree_surface_get(struct pipe_screen *screen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct nv20_miptree *nv20mt = (struct nv20_miptree *)pt;
+	struct pipe_surface *ps;
+
+	ps = CALLOC_STRUCT(pipe_surface);
+	if (!ps)
+		return NULL;
+	pipe_texture_reference(&ps->texture, pt);
+	pipe_buffer_reference(screen, &ps->buffer, nv20mt->buffer);
+	ps->format = pt->format;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = nv20mt->level[level].pitch;
+	ps->usage = flags;
+	ps->status = PIPE_SURFACE_STATUS_DEFINED;
+	ps->refcount = 1;
+	ps->winsys = screen->winsys;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		ps->offset = nv20mt->level[level].image_offset[face];
+	} else
+	if (pt->target == PIPE_TEXTURE_3D) {
+		ps->offset = nv20mt->level[level].image_offset[zslice];
+	} else {
+		ps->offset = nv20mt->level[level].image_offset[0];
+	}
+
+	return ps;
+}
+
+static void
+nv20_miptree_surface_release(struct pipe_screen *pscreen,
+			     struct pipe_surface **psurface)
+{
+	struct pipe_surface *ps = *psurface;
+
+	*psurface = NULL;
+	if (--ps->refcount > 0)
+		return;
+
+	pipe_texture_reference(&ps->texture, NULL);
+	pipe_buffer_reference(pscreen, &ps->buffer, NULL);
+	FREE(ps);
+}
+
+void nv20_screen_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv20_miptree_create;
+	pscreen->texture_release = nv20_miptree_release;
+	pscreen->get_tex_surface = nv20_miptree_surface_get;
+	pscreen->tex_surface_release = nv20_miptree_surface_release;
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_prim_vbuf.c b/src/gallium/drivers/nv20/nv20_prim_vbuf.c
new file mode 100644
index 00000000000..74540845a87
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_prim_vbuf.c
@@ -0,0 +1,402 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * \file
+ * Build post-transformation, post-clipping vertex buffers and element
+ * lists by hooking into the end of the primitive pipeline and
+ * manipulating the vertex_id field in the vertex headers.
+ *
+ * XXX: work in progress 
+ * 
+ * \author José Fonseca <jrfonseca@tungstengraphics.com>
+ * \author Keith Whitwell <keith@tungstengraphics.com>
+ */
+
+
+#include "pipe/p_debug.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv20_context.h"
+#include "nv20_state.h"
+
+#include "draw/draw_vbuf.h"
+
+/**
+ * Primitive renderer for nv20.
+ */
+struct nv20_vbuf_render {
+	struct vbuf_render base;
+
+	struct nv20_context *nv20;   
+
+	/** Vertex buffer in VRAM */
+	struct pipe_buffer *pbuffer;
+
+	/** Vertex buffer in normal memory */
+	void *mbuffer;
+
+	/** Vertex size in bytes */
+	/*unsigned vertex_size;*/
+
+	/** Hardware primitive */
+	unsigned hwprim;
+};
+
+/**
+ * Basically a cast wrapper.
+ */
+static INLINE struct nv20_vbuf_render *
+nv20_vbuf_render(struct vbuf_render *render)
+{
+	assert(render);
+	return (struct nv20_vbuf_render *)render;
+}
+
+void nv20_vtxbuf_bind( struct nv20_context* nv20 )
+{
+#if 0
+	int i;
+	for(i = 0; i < NV20TCL_VTXBUF_ADDRESS__SIZE; i++) {
+		BEGIN_RING(kelvin, NV20TCL_VTXBUF_ADDRESS(i), 1);
+		OUT_RING(0/*nv20->vtxbuf*/);
+		BEGIN_RING(kelvin, NV20TCL_VTXFMT(i) ,1);
+		OUT_RING(0/*XXX*/);
+	}
+#endif
+}
+
+static const struct vertex_info *
+nv20_vbuf_render_get_vertex_info( struct vbuf_render *render )
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+	struct nv20_context *nv20 = nv20_render->nv20;
+
+	nv20_emit_hw_state(nv20);
+
+	return &nv20->vertex_info;
+}
+
+static void *
+nv20__allocate_mbuffer(struct nv20_vbuf_render *nv20_render, size_t size)
+{
+	nv20_render->mbuffer = MALLOC(size);
+	return nv20_render->mbuffer;
+}
+
+static void *
+nv20__allocate_pbuffer(struct nv20_vbuf_render *nv20_render, size_t size)
+{
+	struct pipe_winsys *winsys = nv20_render->nv20->pipe.winsys;
+	nv20_render->pbuffer = winsys->buffer_create(winsys, 64,
+					PIPE_BUFFER_USAGE_VERTEX, size);
+	return winsys->buffer_map(winsys,
+			nv20_render->pbuffer,
+			PIPE_BUFFER_USAGE_CPU_WRITE);
+}
+
+static void *
+nv20_vbuf_render_allocate_vertices( struct vbuf_render *render,
+		ushort vertex_size,
+		ushort nr_vertices )
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+	size_t size = (size_t)vertex_size * (size_t)nr_vertices;
+	void *buf;
+
+	assert(!nv20_render->pbuffer);
+	assert(!nv20_render->mbuffer);
+
+	/*
+	 * For small amount of vertices, don't bother with pipe vertex
+	 * buffer, the data will be passed directly via the fifo.
+	 */
+	/* XXX: Pipe vertex buffers don't work. */
+	if (0 && size > 16 * 1024)
+		buf = nv20__allocate_pbuffer(nv20_render, size);
+	else
+		buf = nv20__allocate_mbuffer(nv20_render, size);
+
+	if (buf)
+		nv20_render->nv20->dirty |= NV20_NEW_VTXARRAYS;
+
+	return buf;
+}
+
+static boolean
+nv20_vbuf_render_set_primitive( struct vbuf_render *render, 
+		unsigned prim )
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+	unsigned hwp = nvgl_primitive(prim);
+	if (hwp == 0)
+		return FALSE;
+
+	nv20_render->hwprim = hwp;
+	return TRUE;
+}
+
+static uint32_t
+nv20__vtxhwformat(unsigned stride, unsigned fields, unsigned type)
+{
+	return (stride << NV20TCL_VTXFMT_STRIDE_SHIFT) |
+		(fields << NV20TCL_VTXFMT_SIZE_SHIFT) |
+		(type << NV20TCL_VTXFMT_TYPE_SHIFT);
+}
+
+static unsigned
+nv20__emit_format(struct nv20_context *nv20, enum attrib_emit type, int hwattr)
+{
+	uint32_t hwfmt = 0;
+	unsigned fields;
+
+	switch (type) {
+	case EMIT_OMIT:
+		hwfmt = nv20__vtxhwformat(0, 0, 2);
+		fields = 0;
+		break;
+	case EMIT_1F:
+		hwfmt = nv20__vtxhwformat(4, 1, 2);
+		fields = 1;
+		break;
+	case EMIT_2F:
+		hwfmt = nv20__vtxhwformat(8, 2, 2);
+		fields = 2;
+		break;
+	case EMIT_3F:
+		hwfmt = nv20__vtxhwformat(12, 3, 2);
+		fields = 3;
+		break;
+	case EMIT_4F:
+		hwfmt = nv20__vtxhwformat(16, 4, 2);
+		fields = 4;
+		break;
+	default:
+		NOUVEAU_ERR("unhandled attrib_emit %d\n", type);
+		return 0;
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_VTXFMT(hwattr), 1);
+	OUT_RING(hwfmt);
+	return fields;
+}
+
+static unsigned
+nv20__emit_vertex_array_format(struct nv20_context *nv20)
+{
+	struct vertex_info *vinfo = &nv20->vertex_info;
+	int hwattr = NV20TCL_VTXFMT__SIZE;
+	int attr = 0;
+	unsigned nr_fields = 0;
+
+	while (hwattr-- > 0) {
+		if (vinfo->hwfmt[0] & (1 << hwattr)) {
+			nr_fields += nv20__emit_format(nv20,
+					vinfo->attrib[attr].emit, hwattr);
+			attr++;
+		} else
+			nv20__emit_format(nv20, EMIT_OMIT, hwattr);
+	}
+
+	return nr_fields;
+}
+
+static void
+nv20__draw_mbuffer(struct nv20_vbuf_render *nv20_render,
+		const ushort *indices,
+		uint nr_indices)
+{
+	struct nv20_context *nv20 = nv20_render->nv20;
+	struct vertex_info *vinfo = &nv20->vertex_info;
+	unsigned nr_fields;
+	int max_push;
+	ubyte *data = nv20_render->mbuffer;
+	int vsz = 4 * vinfo->size;
+
+	nr_fields = nv20__emit_vertex_array_format(nv20);
+
+	BEGIN_RING(kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
+	OUT_RING(nv20_render->hwprim);
+
+	max_push = 1200 / nr_fields;
+	while (nr_indices) {
+		int i;
+		int push = MIN2(nr_indices, max_push);
+
+		BEGIN_RING_NI(kelvin, NV20TCL_VERTEX_DATA, push * nr_fields);
+		for (i = 0; i < push; i++) {
+			/* XXX: fixme to handle other than floats? */
+			int f = nr_fields;
+			float *attrv = (float*)&data[indices[i] * vsz];
+			while (f-- > 0)
+				OUT_RINGf(*attrv++);
+		}
+
+		nr_indices -= push;
+		indices += push;
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_VERTEX_BEGIN_END, 1);
+	OUT_RING(NV20TCL_VERTEX_BEGIN_END_STOP);
+}
+
+static void
+nv20__draw_pbuffer(struct nv20_vbuf_render *nv20_render,
+		const ushort *indices,
+		uint nr_indices)
+{
+	struct nv20_context *nv20 = nv20_render->nv20;
+	int push, i;
+
+	NOUVEAU_ERR("nv20__draw_pbuffer: this path is broken.\n");
+
+	BEGIN_RING(kelvin, NV10TCL_VERTEX_ARRAY_OFFSET_POS, 1);
+	OUT_RELOCl(nv20_render->pbuffer, 0,
+			NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+
+	BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING(nv20_render->hwprim);
+
+	if (nr_indices & 1) {
+		BEGIN_RING(kelvin, NV10TCL_VB_ELEMENT_U32, 1);
+		OUT_RING  (indices[0]);
+		indices++; nr_indices--;
+	}
+
+	while (nr_indices) {
+		// XXX too big/small ? check the size
+		push = MIN2(nr_indices, 1200 * 2);
+
+		BEGIN_RING_NI(kelvin, NV10TCL_VB_ELEMENT_U16, push >> 1);
+		for (i = 0; i < push; i+=2)
+			OUT_RING((indices[i+1] << 16) | indices[i]);
+
+		nr_indices -= push;
+		indices  += push;
+	}
+
+	BEGIN_RING(kelvin, NV10TCL_VERTEX_BUFFER_BEGIN_END, 1);
+	OUT_RING  (0);
+}
+
+static void
+nv20_vbuf_render_draw( struct vbuf_render *render,
+		const ushort *indices,
+		uint nr_indices)
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+
+	nv20_emit_hw_state(nv20_render->nv20);
+
+	if (nv20_render->pbuffer)
+		nv20__draw_pbuffer(nv20_render, indices, nr_indices);
+	else if (nv20_render->mbuffer)
+		nv20__draw_mbuffer(nv20_render, indices, nr_indices);
+	else
+		assert(0);
+}
+
+
+static void
+nv20_vbuf_render_release_vertices( struct vbuf_render *render,
+		void *vertices, 
+		unsigned vertex_size,
+		unsigned vertices_used )
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+	struct nv20_context *nv20 = nv20_render->nv20;
+	struct pipe_winsys *winsys = nv20->pipe.winsys;
+	struct pipe_screen *pscreen = &nv20->screen->pipe;
+
+	if (nv20_render->pbuffer) {
+		winsys->buffer_unmap(winsys, nv20_render->pbuffer);
+		pipe_buffer_reference(pscreen, &nv20_render->pbuffer, NULL);
+	} else if (nv20_render->mbuffer) {
+		FREE(nv20_render->mbuffer);
+		nv20_render->mbuffer = NULL;
+	} else
+		assert(0);
+}
+
+
+static void
+nv20_vbuf_render_destroy( struct vbuf_render *render )
+{
+	struct nv20_vbuf_render *nv20_render = nv20_vbuf_render(render);
+
+	assert(!nv20_render->pbuffer);
+	assert(!nv20_render->mbuffer);
+
+	FREE(nv20_render);
+}
+
+
+/**
+ * Create a new primitive render.
+ */
+static struct vbuf_render *
+nv20_vbuf_render_create( struct nv20_context *nv20 )
+{
+	struct nv20_vbuf_render *nv20_render = CALLOC_STRUCT(nv20_vbuf_render);
+
+	nv20_render->nv20 = nv20;
+
+	nv20_render->base.max_vertex_buffer_bytes = 16*1024;
+	nv20_render->base.max_indices = 1024;
+	nv20_render->base.get_vertex_info = nv20_vbuf_render_get_vertex_info;
+	nv20_render->base.allocate_vertices =
+					nv20_vbuf_render_allocate_vertices;
+	nv20_render->base.set_primitive = nv20_vbuf_render_set_primitive;
+	nv20_render->base.draw = nv20_vbuf_render_draw;
+	nv20_render->base.release_vertices = nv20_vbuf_render_release_vertices;
+	nv20_render->base.destroy = nv20_vbuf_render_destroy;
+
+	return &nv20_render->base;
+}
+
+
+/**
+ * Create a new primitive vbuf/render stage.
+ */
+struct draw_stage *nv20_draw_vbuf_stage( struct nv20_context *nv20 )
+{
+	struct vbuf_render *render;
+	struct draw_stage *stage;
+
+	render = nv20_vbuf_render_create(nv20);
+	if(!render)
+		return NULL;
+
+	stage = draw_vbuf_stage( nv20->draw, render );
+	if(!stage) {
+		render->destroy(render);
+		return NULL;
+	}
+
+	return stage;
+}
diff --git a/src/gallium/drivers/nv20/nv20_screen.c b/src/gallium/drivers/nv20/nv20_screen.c
new file mode 100644
index 00000000000..c0a90f6c584
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_screen.c
@@ -0,0 +1,204 @@
+#include "pipe/p_screen.h"
+
+#include "nv20_context.h"
+#include "nv20_screen.h"
+
+static const char *
+nv20_screen_get_name(struct pipe_screen *screen)
+{
+	struct nv20_screen *nv20screen = nv20_screen(screen);
+	struct nouveau_device *dev = nv20screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv20_screen_get_vendor(struct pipe_screen *screen)
+{
+	return "nouveau";
+}
+
+static int
+nv20_screen_get_param(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 2;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 0;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 0;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 0;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 1;
+	case PIPE_CAP_POINT_SPRITE:
+		return 0;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 1;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 0;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 12;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 12;
+	case NOUVEAU_CAP_HW_VTXBUF:
+	case NOUVEAU_CAP_HW_IDXBUF:
+		return 0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv20_screen_get_paramf(struct pipe_screen *screen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 10.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 64.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 2.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 4.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static boolean
+nv20_screen_is_format_supported(struct pipe_screen *screen,
+				enum pipe_format format,
+				enum pipe_texture_target target,
+				unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_A4R4G4B4_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static void *
+nv20_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
+		 unsigned flags )
+{
+	struct pipe_winsys *ws = screen->winsys;
+	void *map;
+
+	map = ws->buffer_map(ws, surface->buffer, flags);
+	if (!map)
+		return NULL;
+
+	return map + surface->offset;
+}
+
+static void
+nv20_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
+{
+	struct pipe_winsys *ws = screen->winsys;
+
+	ws->buffer_unmap(ws, surface->buffer);
+}
+
+static void
+nv20_screen_destroy(struct pipe_screen *pscreen)
+{
+	struct nv20_screen *screen = nv20_screen(pscreen);
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nvws->notifier_free(&screen->sync);
+	nvws->grobj_free(&screen->kelvin);
+
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv20_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv20_screen *screen = CALLOC_STRUCT(nv20_screen);
+	unsigned kelvin_class = 0;
+	unsigned chipset = nvws->channel->device->chipset;
+	int ret;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	/* 3D object */
+	if (chipset >= 0x25)
+		kelvin_class = NV25TCL;
+	else if (chipset >= 0x20)
+		kelvin_class = NV20TCL;
+
+	if (!kelvin_class || chipset >= 0x30) {
+		NOUVEAU_ERR("Unknown nv2x chipset: nv%02x\n", chipset);
+		return NULL;
+	}
+
+	ret = nvws->grobj_alloc(nvws, kelvin_class, &screen->kelvin);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		return FALSE;
+	}
+
+	/* Notifier for sync purposes */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv20_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	screen->pipe.winsys = ws;
+	screen->pipe.destroy = nv20_screen_destroy;
+
+	screen->pipe.get_name = nv20_screen_get_name;
+	screen->pipe.get_vendor = nv20_screen_get_vendor;
+	screen->pipe.get_param = nv20_screen_get_param;
+	screen->pipe.get_paramf = nv20_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv20_screen_is_format_supported;
+
+	screen->pipe.surface_map = nv20_surface_map;
+	screen->pipe.surface_unmap = nv20_surface_unmap;
+
+	nv20_screen_init_miptree_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_screen.h b/src/gallium/drivers/nv20/nv20_screen.h
new file mode 100644
index 00000000000..8f2f2e341db
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_screen.h
@@ -0,0 +1,22 @@
+#ifndef __NV20_SCREEN_H__
+#define __NV20_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv20_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+
+	/* HW graphics objects */
+	struct nouveau_grobj *kelvin;
+	struct nouveau_notifier *sync;
+};
+
+static INLINE struct nv20_screen *
+nv20_screen(struct pipe_screen *screen)
+{
+	return (struct nv20_screen *)screen;
+}
+
+#endif
diff --git a/src/gallium/drivers/nv20/nv20_state.c b/src/gallium/drivers/nv20/nv20_state.c
new file mode 100644
index 00000000000..21bde5b81f9
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_state.c
@@ -0,0 +1,581 @@
+#include "draw/draw_context.h"
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv20_context.h"
+#include "nv20_state.h"
+
+static void *
+nv20_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv20_blend_state *cb;
+
+	cb = MALLOC(sizeof(struct nv20_blend_state));
+
+	cb->b_enable = cso->blend_enable ? 1 : 0;
+	cb->b_srcfunc = ((nvgl_blend_func(cso->alpha_src_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_src_factor)));
+	cb->b_dstfunc = ((nvgl_blend_func(cso->alpha_dst_factor)<<16) |
+			 (nvgl_blend_func(cso->rgb_dst_factor)));
+
+	cb->c_mask = (((cso->colormask & PIPE_MASK_A) ? (0x01<<24) : 0) |
+		      ((cso->colormask & PIPE_MASK_R) ? (0x01<<16) : 0) |
+		      ((cso->colormask & PIPE_MASK_G) ? (0x01<< 8) : 0) |
+		      ((cso->colormask & PIPE_MASK_B) ? (0x01<< 0) : 0));
+
+	cb->d_enable = cso->dither ? 1 : 0;
+
+	return (void *)cb;
+}
+
+static void
+nv20_blend_state_bind(struct pipe_context *pipe, void *blend)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->blend = (struct nv20_blend_state*)blend;
+
+	nv20->dirty |= NV20_NEW_BLEND;
+}
+
+static void
+nv20_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV20TCL_TX_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV20TCL_TX_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV20TCL_TX_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV20TCL_TX_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV20TCL_TX_WRAP_S_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV20TCL_TX_WRAP_S_REPEAT;
+		break;
+	}
+
+	return (ret >> NV20TCL_TX_WRAP_S_SHIFT);
+}
+
+static void *
+nv20_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv20_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv20_sampler_state));
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV20TCL_TX_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV20TCL_TX_WRAP_T_SHIFT));
+
+	ps->en = 0;
+	if (cso->max_anisotropy > 1.0) {
+		/* no idea, binary driver sets it, works without it.. meh.. */
+		ps->wrap |= (1 << 5);
+
+/*		if (cso->max_anisotropy >= 8.0) {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_8X;
+		} else
+		if (cso->max_anisotropy >= 4.0) {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_4X;
+		} else {
+			ps->en |= NV20TCL_TX_ENABLE_ANISO_2X;
+		}*/
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV20TCL_TX_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV20TCL_TX_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV20TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV20TCL_TX_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |=
+				NV20TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV20TCL_TX_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+/*	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV10TCL_TX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}*/
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv20_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv20->tex_sampler[unit] = sampler[unit];
+		nv20->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void
+nv20_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv20_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv20->tex_miptree[unit] = (struct nv20_miptree *)miptree[unit];
+		nv20->dirty_samplers |= (1 << unit);
+	}
+}
+
+static void *
+nv20_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv20_rasterizer_state *rs;
+	int i;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	offset_cw/ccw -nohw
+	 * 	scissor
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 * 	offset_units / offset_scale
+	 */
+	rs = MALLOC(sizeof(struct nv20_rasterizer_state));
+
+	rs->templ = cso;
+	
+	rs->shade_model = cso->flatshade ? NV20TCL_SHADE_MODEL_FLAT :
+						NV20TCL_SHADE_MODEL_SMOOTH;
+
+	rs->line_width = (unsigned char)(cso->line_width * 8.0) & 0xff;
+	rs->line_smooth_en = cso->line_smooth ? 1 : 0;
+
+	/* XXX: nv20 and nv25 different! */
+	rs->point_size = *(uint32_t*)&cso->point_size;
+
+	rs->poly_smooth_en = cso->poly_smooth ? 1 : 0;
+
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		rs->front_face = NV20TCL_FRONT_FACE_CCW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_ccw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_cw);
+	} else {
+		rs->front_face = NV20TCL_FRONT_FACE_CW;
+		rs->poly_mode_front = nvgl_polygon_mode(cso->fill_cw);
+		rs->poly_mode_back  = nvgl_polygon_mode(cso->fill_ccw);
+	}
+
+	switch (cso->cull_mode) {
+	case PIPE_WINDING_CCW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CCW)
+			rs->cull_face    = NV20TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV20TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_CW:
+		rs->cull_face_en = 1;
+		if (cso->front_winding == PIPE_WINDING_CW)
+			rs->cull_face    = NV20TCL_CULL_FACE_FRONT;
+		else
+			rs->cull_face    = NV20TCL_CULL_FACE_BACK;
+		break;
+	case PIPE_WINDING_BOTH:
+		rs->cull_face_en = 1;
+		rs->cull_face    = NV20TCL_CULL_FACE_FRONT_AND_BACK;
+		break;
+	case PIPE_WINDING_NONE:
+	default:
+		rs->cull_face_en = 0;
+		rs->cull_face    = 0;
+		break;
+	}
+
+	if (cso->point_sprite) {
+		rs->point_sprite = (1 << 0);
+		for (i = 0; i < 8; i++) {
+			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+				rs->point_sprite |= (1 << (8 + i));
+		}
+	} else {
+		rs->point_sprite = 0;
+	}
+
+	return (void *)rs;
+}
+
+static void
+nv20_rasterizer_state_bind(struct pipe_context *pipe, void *rast)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->rast = (struct nv20_rasterizer_state*)rast;
+
+	draw_set_rasterizer_state(nv20->draw, (nv20->rast ? nv20->rast->templ : NULL));
+
+	nv20->dirty |= NV20_NEW_RAST;
+}
+
+static void
+nv20_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv20_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv20_depth_stencil_alpha_state *hw;
+
+	hw = MALLOC(sizeof(struct nv20_depth_stencil_alpha_state));
+
+	hw->depth.func		= nvgl_comparison_op(cso->depth.func);
+	hw->depth.write_enable	= cso->depth.writemask ? 1 : 0;
+	hw->depth.test_enable	= cso->depth.enabled ? 1 : 0;
+
+	hw->stencil.enable = cso->stencil[0].enabled ? 1 : 0;
+	hw->stencil.wmask = cso->stencil[0].write_mask;
+	hw->stencil.func = nvgl_comparison_op(cso->stencil[0].func);
+	hw->stencil.ref	= cso->stencil[0].ref_value;
+	hw->stencil.vmask = cso->stencil[0].value_mask;
+	hw->stencil.fail = nvgl_stencil_op(cso->stencil[0].fail_op);
+	hw->stencil.zfail = nvgl_stencil_op(cso->stencil[0].zfail_op);
+	hw->stencil.zpass = nvgl_stencil_op(cso->stencil[0].zpass_op);
+
+	hw->alpha.enabled = cso->alpha.enabled ? 1 : 0;
+	hw->alpha.func = nvgl_comparison_op(cso->alpha.func);
+	hw->alpha.ref  = float_to_ubyte(cso->alpha.ref);
+
+	return (void *)hw;
+}
+
+static void
+nv20_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *dsa)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->dsa = (struct nv20_depth_stencil_alpha_state*)dsa;
+
+	nv20->dirty |= NV20_NEW_DSA;
+}
+
+static void
+nv20_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void *
+nv20_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *templ)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	return draw_create_vertex_shader(nv20->draw, templ);
+}
+
+static void
+nv20_vp_state_bind(struct pipe_context *pipe, void *shader)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_bind_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader);
+
+	nv20->dirty |= NV20_NEW_VERTPROG;
+}
+
+static void
+nv20_vp_state_delete(struct pipe_context *pipe, void *shader)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_delete_vertex_shader(nv20->draw, (struct draw_vertex_shader *) shader);
+}
+
+static void *
+nv20_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv20_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv20_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	
+	tgsi_scan_shader(cso->tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv20_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nv20_fragment_program *fp = hwcso;
+
+	nv20->fragprog.current = fp;
+	nv20->dirty |= NV20_NEW_FRAGPROG;
+}
+
+static void
+nv20_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nv20_fragment_program *fp = hwcso;
+
+	nv20_fragprog_destroy(nv20, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv20_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->blend_color = (struct pipe_blend_color*)bcol;
+
+	nv20->dirty |= NV20_NEW_BLENDCOL;
+}
+
+static void
+nv20_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	draw_set_clip_state(nv20->draw, clip);
+}
+
+static void
+nv20_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+
+	assert(shader < PIPE_SHADER_TYPES);
+	assert(index == 0);
+
+	if (buf) {
+		void *mapped;
+		if (buf->size && (mapped = ws->buffer_map(ws, buf->buffer, PIPE_BUFFER_USAGE_CPU_READ)))
+		{
+			memcpy(nv20->constbuf[shader], mapped, buf->size);
+			nv20->constbuf_nr[shader] =
+				buf->size / (4 * sizeof(float));
+			ws->buffer_unmap(ws, buf->buffer);
+		}
+	}
+}
+
+static void
+nv20_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->framebuffer = (struct pipe_framebuffer_state*)fb;
+
+	nv20->dirty |= NV20_NEW_FRAMEBUFFER;
+}
+
+static void
+nv20_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	NOUVEAU_ERR("line stipple hahaha\n");
+}
+
+static void
+nv20_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->scissor = (struct pipe_scissor_state*)s;
+
+	nv20->dirty |= NV20_NEW_SCISSOR;
+}
+
+static void
+nv20_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	nv20->viewport = (struct pipe_viewport_state*)vpt;
+
+	draw_set_viewport_state(nv20->draw, nv20->viewport);
+
+	nv20->dirty |= NV20_NEW_VIEWPORT;
+}
+
+static void
+nv20_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	memcpy(nv20->vtxbuf, vb, sizeof(*vb) * count);
+	nv20->dirty |= NV20_NEW_VTXARRAYS;
+
+	draw_set_vertex_buffers(nv20->draw, count, vb);
+}
+
+static void
+nv20_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+
+	memcpy(nv20->vtxelt, ve, sizeof(*ve) * count);
+	nv20->dirty |= NV20_NEW_VTXARRAYS;
+
+	draw_set_vertex_elements(nv20->draw, count, ve);
+}
+
+void
+nv20_init_state_functions(struct nv20_context *nv20)
+{
+	nv20->pipe.create_blend_state = nv20_blend_state_create;
+	nv20->pipe.bind_blend_state = nv20_blend_state_bind;
+	nv20->pipe.delete_blend_state = nv20_blend_state_delete;
+
+	nv20->pipe.create_sampler_state = nv20_sampler_state_create;
+	nv20->pipe.bind_sampler_states = nv20_sampler_state_bind;
+	nv20->pipe.delete_sampler_state = nv20_sampler_state_delete;
+	nv20->pipe.set_sampler_textures = nv20_set_sampler_texture;
+
+	nv20->pipe.create_rasterizer_state = nv20_rasterizer_state_create;
+	nv20->pipe.bind_rasterizer_state = nv20_rasterizer_state_bind;
+	nv20->pipe.delete_rasterizer_state = nv20_rasterizer_state_delete;
+
+	nv20->pipe.create_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_create;
+	nv20->pipe.bind_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_bind;
+	nv20->pipe.delete_depth_stencil_alpha_state =
+		nv20_depth_stencil_alpha_state_delete;
+
+	nv20->pipe.create_vs_state = nv20_vp_state_create;
+	nv20->pipe.bind_vs_state = nv20_vp_state_bind;
+	nv20->pipe.delete_vs_state = nv20_vp_state_delete;
+
+	nv20->pipe.create_fs_state = nv20_fp_state_create;
+	nv20->pipe.bind_fs_state = nv20_fp_state_bind;
+	nv20->pipe.delete_fs_state = nv20_fp_state_delete;
+
+	nv20->pipe.set_blend_color = nv20_set_blend_color;
+	nv20->pipe.set_clip_state = nv20_set_clip_state;
+	nv20->pipe.set_constant_buffer = nv20_set_constant_buffer;
+	nv20->pipe.set_framebuffer_state = nv20_set_framebuffer_state;
+	nv20->pipe.set_polygon_stipple = nv20_set_polygon_stipple;
+	nv20->pipe.set_scissor_state = nv20_set_scissor_state;
+	nv20->pipe.set_viewport_state = nv20_set_viewport_state;
+
+	nv20->pipe.set_vertex_buffers = nv20_set_vertex_buffers;
+	nv20->pipe.set_vertex_elements = nv20_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_state.h b/src/gallium/drivers/nv20/nv20_state.h
new file mode 100644
index 00000000000..34f402fdcbf
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_state.h
@@ -0,0 +1,139 @@
+#ifndef __NV20_STATE_H__
+#define __NV20_STATE_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv20_blend_state {
+	uint32_t b_enable;
+	uint32_t b_srcfunc;
+	uint32_t b_dstfunc;
+
+	uint32_t c_mask;
+
+	uint32_t d_enable;
+};
+
+struct nv20_sampler_state {
+	uint32_t wrap;
+	uint32_t en;
+	uint32_t filt;
+	uint32_t bcol;
+};
+
+struct nv20_rasterizer_state {
+	uint32_t shade_model;
+
+	uint32_t line_width;
+	uint32_t line_smooth_en;
+
+	uint32_t point_size;
+
+	uint32_t poly_smooth_en;
+	
+	uint32_t poly_mode_front;
+	uint32_t poly_mode_back;
+
+	uint32_t front_face;
+	uint32_t cull_face;
+	uint32_t cull_face_en;
+
+	uint32_t point_sprite;
+
+	const struct pipe_rasterizer_state *templ;
+};
+
+struct nv20_vertex_program_exec {
+	uint32_t data[4];
+	boolean has_branch_offset;
+	int const_index;
+};
+
+struct nv20_vertex_program_data {
+	int index; /* immediates == -1 */
+	float value[4];
+};
+
+struct nv20_vertex_program {
+	const struct pipe_shader_state *pipe;
+
+	boolean translated;
+	struct nv20_vertex_program_exec *insns;
+	unsigned nr_insns;
+	struct nv20_vertex_program_data *consts;
+	unsigned nr_consts;
+
+	struct nouveau_resource *exec;
+	unsigned exec_start;
+	struct nouveau_resource *data;
+	unsigned data_start;
+	unsigned data_start_min;
+
+	uint32_t ir;
+	uint32_t or;
+};
+
+struct nv20_fragment_program_data {
+	unsigned offset;
+	unsigned index;
+};
+
+struct nv20_fragment_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+
+	boolean translated;
+	boolean on_hw;
+	unsigned samplers;
+
+	uint32_t *insn;
+	int       insn_len;
+
+	struct nv20_fragment_program_data *consts;
+	unsigned nr_consts;
+
+	struct pipe_buffer *buffer;
+
+	uint32_t fp_control;
+	uint32_t fp_reg_control;
+};
+
+
+struct nv20_depth_stencil_alpha_state {
+	struct {
+		uint32_t func;
+		uint32_t write_enable;
+		uint32_t test_enable;
+	} depth;
+
+	struct {
+		uint32_t enable;
+		uint32_t wmask;
+		uint32_t func;
+		uint32_t ref;
+		uint32_t vmask;
+		uint32_t fail;
+		uint32_t zfail;
+		uint32_t zpass;
+	} stencil;
+
+	struct {
+		uint32_t enabled;
+		uint32_t func;
+		uint32_t ref;
+	} alpha;
+};
+
+struct nv20_miptree {
+	struct pipe_texture base;
+
+	struct pipe_buffer *buffer;
+	uint total_size;
+
+	struct {
+		uint pitch;
+		uint *image_offset;
+	} level[PIPE_MAX_TEXTURE_LEVELS];
+};
+
+#endif
diff --git a/src/gallium/drivers/nv20/nv20_state_emit.c b/src/gallium/drivers/nv20/nv20_state_emit.c
new file mode 100644
index 00000000000..5265bf3a314
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_state_emit.c
@@ -0,0 +1,359 @@
+#include "nv20_context.h"
+#include "nv20_state.h"
+#include "draw/draw_context.h"
+
+static void nv20_state_emit_blend(struct nv20_context* nv20)
+{
+	struct nv20_blend_state *b = nv20->blend;
+
+	BEGIN_RING(kelvin, NV20TCL_DITHER_ENABLE, 1);
+	OUT_RING  (b->d_enable);
+
+	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_ENABLE, 1);
+	OUT_RING  (b->b_enable);
+
+	BEGIN_RING(kelvin, NV20TCL_BLEND_FUNC_SRC, 2);
+	OUT_RING  (b->b_srcfunc);
+	OUT_RING  (b->b_dstfunc);
+
+	BEGIN_RING(kelvin, NV20TCL_COLOR_MASK, 1);
+	OUT_RING  (b->c_mask);
+}
+
+static void nv20_state_emit_blend_color(struct nv20_context* nv20)
+{
+	struct pipe_blend_color *c = nv20->blend_color;
+
+	BEGIN_RING(kelvin, NV20TCL_BLEND_COLOR, 1);
+	OUT_RING  ((float_to_ubyte(c->color[3]) << 24)|
+		   (float_to_ubyte(c->color[0]) << 16)|
+		   (float_to_ubyte(c->color[1]) << 8) |
+		   (float_to_ubyte(c->color[2]) << 0));
+}
+
+static void nv20_state_emit_rast(struct nv20_context* nv20)
+{
+	struct nv20_rasterizer_state *r = nv20->rast;
+
+	BEGIN_RING(kelvin, NV20TCL_SHADE_MODEL, 2);
+	OUT_RING  (r->shade_model);
+	OUT_RING  (r->line_width);
+
+
+	BEGIN_RING(kelvin, NV20TCL_POINT_SIZE, 1);
+	OUT_RING  (r->point_size);
+
+	BEGIN_RING(kelvin, NV20TCL_POLYGON_MODE_FRONT, 2);
+	OUT_RING  (r->poly_mode_front);
+	OUT_RING  (r->poly_mode_back);
+
+
+	BEGIN_RING(kelvin, NV20TCL_CULL_FACE, 2);
+	OUT_RING  (r->cull_face);
+	OUT_RING  (r->front_face);
+
+	BEGIN_RING(kelvin, NV20TCL_LINE_SMOOTH_ENABLE, 2);
+	OUT_RING  (r->line_smooth_en);
+	OUT_RING  (r->poly_smooth_en);
+
+	BEGIN_RING(kelvin, NV20TCL_CULL_FACE_ENABLE, 1);
+	OUT_RING  (r->cull_face_en);
+}
+
+static void nv20_state_emit_dsa(struct nv20_context* nv20)
+{
+	struct nv20_depth_stencil_alpha_state *d = nv20->dsa;
+
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_FUNC, 1);
+	OUT_RING (d->depth.func);
+
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING (d->depth.write_enable);
+
+	BEGIN_RING(kelvin, NV20TCL_DEPTH_TEST_ENABLE, 1);
+	OUT_RING (d->depth.test_enable);
+
+#if 0
+	BEGIN_RING(kelvin, NV20TCL_STENCIL_ENABLE, 1);
+	OUT_RING (d->stencil.enable);
+	BEGIN_RING(kelvin, NV20TCL_STENCIL_MASK, 7);
+	OUT_RINGp ((uint32_t *)&(d->stencil.wmask), 7);
+#endif
+
+	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_ENABLE, 1);
+	OUT_RING (d->alpha.enabled);
+
+	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_FUNC, 1);
+	OUT_RING (d->alpha.func);
+
+	BEGIN_RING(kelvin, NV20TCL_ALPHA_FUNC_REF, 1);
+	OUT_RING (d->alpha.ref);
+}
+
+static void nv20_state_emit_viewport(struct nv20_context* nv20)
+{
+}
+
+static void nv20_state_emit_scissor(struct nv20_context* nv20)
+{
+	/* NV20TCL_SCISSOR_* is probably a software method */
+/*	struct pipe_scissor_state *s = nv20->scissor;
+	BEGIN_RING(kelvin, NV20TCL_SCISSOR_HORIZ, 2);
+	OUT_RING  (((s->maxx - s->minx) << 16) | s->minx);
+	OUT_RING  (((s->maxy - s->miny) << 16) | s->miny);*/
+}
+
+static void nv20_state_emit_framebuffer(struct nv20_context* nv20)
+{
+	struct pipe_framebuffer_state* fb = nv20->framebuffer;
+	struct pipe_surface *rt, *zeta = NULL;
+	uint32_t rt_format, w, h;
+	int colour_format = 0, zeta_format = 0;
+
+	w = fb->cbufs[0]->width;
+	h = fb->cbufs[0]->height;
+	colour_format = fb->cbufs[0]->format;
+	rt = fb->cbufs[0];
+
+	if (fb->zsbuf) {
+		if (colour_format) {
+			assert(w == fb->zsbuf->width);
+			assert(h == fb->zsbuf->height);
+		} else {
+			w = fb->zsbuf->width;
+			h = fb->zsbuf->height;
+		}
+
+		zeta_format = fb->zsbuf->format;
+		zeta = fb->zsbuf;
+	}
+
+	rt_format = NV20TCL_RT_FORMAT_TYPE_LINEAR | 0x20;
+
+	switch (colour_format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case 0:
+		rt_format |= NV20TCL_RT_FORMAT_COLOR_A8R8G8B8;
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		rt_format |= NV20TCL_RT_FORMAT_COLOR_R5G6B5;
+		break;
+	default:
+		assert(0);
+	}
+
+	if (zeta) {
+		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
+		OUT_RING  (rt->stride | (zeta->stride << 16));
+	} else {
+		BEGIN_RING(kelvin, NV20TCL_RT_PITCH, 1);
+		OUT_RING  (rt->stride | (rt->stride << 16));
+	}
+
+	nv20->rt[0] = rt->buffer;
+
+	if (zeta_format)
+	{
+		nv20->zeta = zeta->buffer;
+	}
+
+	BEGIN_RING(kelvin, NV20TCL_RT_HORIZ, 3);
+	OUT_RING  ((w << 16) | 0);
+	OUT_RING  ((h << 16) | 0);
+	OUT_RING  (rt_format);
+	BEGIN_RING(kelvin, NV20TCL_VIEWPORT_CLIP_HORIZ(0), 2);
+	OUT_RING  (((w - 1) << 16) | 0);
+	OUT_RING  (((h - 1) << 16) | 0);
+}
+
+static void nv20_vertex_layout(struct nv20_context *nv20)
+{
+	struct nv20_fragment_program *fp = nv20->fragprog.current;
+	struct draw_context *dc = nv20->draw;
+	uint32_t src;
+	int i;
+	struct vertex_info *vinfo = &nv20->vertex_info;
+	const enum interp_mode colorInterp = INTERP_LINEAR;
+	boolean colors[2] = { FALSE };
+	boolean generics[4] = { FALSE };
+	boolean fog = FALSE;
+
+	memset(vinfo, 0, sizeof(*vinfo));
+
+	/*
+	 * NV10 hardware vertex attribute order:
+	 * fog, weight, normal, tex1, tex0, 2nd color, color, position
+	 * vinfo->hwfmt[0] has a used-bit corresponding to each of these.
+	 * relation to TGSI_SEMANTIC_*:
+	 * - POSITION: position (always used)
+	 * - COLOR: 2nd color, color
+	 * - GENERIC: weight, normal, tex1, tex0
+	 * - FOG: fog
+	 */
+
+	for (i = 0; i < fp->info.num_inputs; i++) {
+		int isn = fp->info.input_semantic_name[i];
+		int isi = fp->info.input_semantic_index[i];
+		switch (isn) {
+		case TGSI_SEMANTIC_POSITION:
+			break;
+		case TGSI_SEMANTIC_COLOR:
+			assert(isi < 2);
+			colors[isi] = TRUE;
+			break;
+		case TGSI_SEMANTIC_GENERIC:
+			assert(isi < 4);
+			generics[isi] = TRUE;
+			break;
+		case TGSI_SEMANTIC_FOG:
+			fog = TRUE;
+			break;
+		default:
+			assert(0 && "unknown input_semantic_name");
+		}
+	}
+
+	if (fog) {
+		int src = draw_find_vs_output(dc, TGSI_SEMANTIC_FOG, 0);
+		draw_emit_vertex_attr(vinfo, EMIT_1F, INTERP_PERSPECTIVE, src);
+		vinfo->hwfmt[0] |= (1 << 7);
+	}
+
+	for (i = 3; i >= 0; i--) {
+		int src;
+		if (!generics[i])
+			continue;
+		src = draw_find_vs_output(dc, TGSI_SEMANTIC_GENERIC, i);
+		draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src);
+		vinfo->hwfmt[0] |= (1 << (i + 3));
+	}
+
+	if (colors[1]) {
+		int src = draw_find_vs_output(dc, TGSI_SEMANTIC_COLOR, 1);
+		draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
+		vinfo->hwfmt[0] |= (1 << 2);
+	}
+
+	if (colors[0]) {
+		int src = draw_find_vs_output(dc, TGSI_SEMANTIC_COLOR, 0);
+		draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src);
+		vinfo->hwfmt[0] |= (1 << 1);
+	}
+
+	/* always do position */
+	src = draw_find_vs_output(dc, TGSI_SEMANTIC_POSITION, 0);
+	draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_LINEAR, src);
+	vinfo->hwfmt[0] |= (1 << 0);
+
+	draw_compute_vertex_size(vinfo);
+}
+
+void
+nv20_emit_hw_state(struct nv20_context *nv20)
+{
+	int i;
+
+	if (nv20->dirty & NV20_NEW_VERTPROG) {
+		//nv20_vertprog_bind(nv20, nv20->vertprog.current);
+		nv20->dirty &= ~NV20_NEW_VERTPROG;
+	}
+
+	if (nv20->dirty & NV20_NEW_FRAGPROG) {
+		nv20_fragprog_bind(nv20, nv20->fragprog.current);
+		/*XXX: clear NV20_NEW_FRAGPROG if no new program uploaded */
+		nv20->dirty_samplers |= (1<<10);
+		nv20->dirty_samplers = 0;
+	}
+
+	if (nv20->dirty_samplers || (nv20->dirty & NV20_NEW_FRAGPROG)) {
+		nv20_fragtex_bind(nv20);
+		nv20->dirty &= ~NV20_NEW_FRAGPROG;
+	}
+
+	if (nv20->dirty & NV20_NEW_VTXARRAYS) {
+		nv20->dirty &= ~NV20_NEW_VTXARRAYS;
+		nv20_vertex_layout(nv20);
+		nv20_vtxbuf_bind(nv20);
+	}
+
+	if (nv20->dirty & NV20_NEW_BLEND) {
+		nv20->dirty &= ~NV20_NEW_BLEND;
+		nv20_state_emit_blend(nv20);
+	}
+
+	if (nv20->dirty & NV20_NEW_BLENDCOL) {
+		nv20->dirty &= ~NV20_NEW_BLENDCOL;
+		nv20_state_emit_blend_color(nv20);
+	}
+
+	if (nv20->dirty & NV20_NEW_RAST) {
+		nv20->dirty &= ~NV20_NEW_RAST;
+		nv20_state_emit_rast(nv20);
+	}
+
+	if (nv20->dirty & NV20_NEW_DSA) {
+		nv20->dirty &= ~NV20_NEW_DSA;
+		nv20_state_emit_dsa(nv20);
+	}
+
+ 	if (nv20->dirty & NV20_NEW_VIEWPORT) {
+		nv20->dirty &= ~NV20_NEW_VIEWPORT;
+		nv20_state_emit_viewport(nv20);
+	}
+
+ 	if (nv20->dirty & NV20_NEW_SCISSOR) {
+		nv20->dirty &= ~NV20_NEW_SCISSOR;
+		nv20_state_emit_scissor(nv20);
+	}
+
+ 	if (nv20->dirty & NV20_NEW_FRAMEBUFFER) {
+		nv20->dirty &= ~NV20_NEW_FRAMEBUFFER;
+		nv20_state_emit_framebuffer(nv20);
+	}
+
+	/* Emit relocs for every referenced buffer.
+	 * This is to ensure the bufmgr has an accurate idea of how
+	 * the buffer is used.  This isn't very efficient, but we don't
+	 * seem to take a significant performance hit.  Will be improved
+	 * at some point.  Vertex arrays are emitted by nv20_vbo.c
+	 */
+
+	/* Render target */
+/* XXX figre out who's who for NV10TCL_DMA_* and fill accordingly
+ *	BEGIN_RING(kelvin, NV20TCL_DMA_COLOR0, 1);
+ *	OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); */
+	BEGIN_RING(kelvin, NV20TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	if (nv20->zeta) {
+/* XXX
+ *		BEGIN_RING(kelvin, NV20TCL_DMA_ZETA, 1);
+ *		OUT_RELOCo(nv20->zeta, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR); */
+		BEGIN_RING(kelvin, NV20TCL_ZETA_OFFSET, 1);
+		OUT_RELOCl(nv20->zeta, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+		/* XXX for when we allocate LMA on nv17 */
+/*		BEGIN_RING(kelvin, NV10TCL_LMA_DEPTH_BUFFER_OFFSET, 1);
+		OUT_RELOCl(nv20->zeta + lma_offset);*/
+	}
+
+	/* Vertex buffer */
+	BEGIN_RING(kelvin, NV20TCL_DMA_VTXBUF0, 1);
+	OUT_RELOCo(nv20->rt[0], NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(kelvin, NV20TCL_COLOR_OFFSET, 1);
+	OUT_RELOCl(nv20->rt[0], 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	/* Texture images */
+	for (i = 0; i < 2; i++) {
+		if (!(nv20->fp_samplers & (1 << i)))
+			continue;
+		BEGIN_RING(kelvin, NV20TCL_TX_OFFSET(i), 1);
+		OUT_RELOCl(nv20->tex[i].buffer, 0, NOUVEAU_BO_VRAM |
+			   NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+		BEGIN_RING(kelvin, NV20TCL_TX_FORMAT(i), 1);
+		OUT_RELOCd(nv20->tex[i].buffer, nv20->tex[i].format,
+			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD |
+			   NOUVEAU_BO_OR, NV20TCL_TX_FORMAT_DMA0,
+			   NV20TCL_TX_FORMAT_DMA1);
+	}
+}
+
diff --git a/src/gallium/drivers/nv20/nv20_surface.c b/src/gallium/drivers/nv20/nv20_surface.c
new file mode 100644
index 00000000000..7bc68d0ca26
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_surface.c
@@ -0,0 +1,64 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "nv20_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+#include "util/u_tile.h"
+
+static void
+nv20_surface_copy(struct pipe_context *pipe, boolean do_flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nouveau_winsys *nvws = nv20->nvws;
+
+	nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+			   width, height);
+}
+
+static void
+nv20_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv20_context *nv20 = nv20_context(pipe);
+	struct nouveau_winsys *nvws = nv20->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+void
+nv20_init_surface_functions(struct nv20_context *nv20)
+{
+	nv20->pipe.surface_copy = nv20_surface_copy;
+	nv20->pipe.surface_fill = nv20_surface_fill;
+}
diff --git a/src/gallium/drivers/nv20/nv20_vbo.c b/src/gallium/drivers/nv20/nv20_vbo.c
new file mode 100644
index 00000000000..4edc4efebd8
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_vbo.c
@@ -0,0 +1,77 @@
+#include "draw/draw_context.h"
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv20_context.h"
+#include "nv20_state.h"
+
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_pushbuf.h"
+
+boolean nv20_draw_elements( struct pipe_context *pipe,
+                    struct pipe_buffer *indexBuffer,
+                    unsigned indexSize,
+                    unsigned prim, unsigned start, unsigned count)
+{
+	struct nv20_context *nv20 = nv20_context( pipe );
+	struct draw_context *draw = nv20->draw;
+	unsigned i;
+
+	nv20_emit_hw_state(nv20);
+
+	/*
+	 * Map vertex buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv20->vtxbuf[i].buffer) {
+			void *buf
+				= pipe->winsys->buffer_map(pipe->winsys,
+						nv20->vtxbuf[i].buffer,
+						PIPE_BUFFER_USAGE_CPU_READ);
+			draw_set_mapped_vertex_buffer(draw, i, buf);
+		}
+	}
+	/* Map index buffer, if present */
+	if (indexBuffer) {
+		void *mapped_indexes
+			= pipe->winsys->buffer_map(pipe->winsys, indexBuffer,
+					PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_element_buffer(draw, indexSize, mapped_indexes);
+	}
+	else {
+		/* no index/element buffer */
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	draw_set_mapped_constant_buffer(draw,
+					nv20->constbuf[PIPE_SHADER_VERTEX],
+					nv20->constbuf_nr[PIPE_SHADER_VERTEX]);
+
+	/* draw! */
+	draw_arrays(nv20->draw, prim, start, count);
+
+	/*
+	 * unmap vertex/index buffers
+	 */
+	for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+		if (nv20->vtxbuf[i].buffer) {
+			pipe->winsys->buffer_unmap(pipe->winsys, nv20->vtxbuf[i].buffer);
+			draw_set_mapped_vertex_buffer(draw, i, NULL);
+		}
+	}
+	if (indexBuffer) {
+		pipe->winsys->buffer_unmap(pipe->winsys, indexBuffer);
+		draw_set_mapped_element_buffer(draw, 0, NULL);
+	}
+
+	return TRUE;
+}
+
+boolean nv20_draw_arrays( struct pipe_context *pipe,
+				 unsigned prim, unsigned start, unsigned count)
+{
+	return nv20_draw_elements(pipe, NULL, 0, prim, start, count);
+}
+
+
+
diff --git a/src/gallium/drivers/nv20/nv20_vertprog.c b/src/gallium/drivers/nv20/nv20_vertprog.c
new file mode 100644
index 00000000000..a885fcd7a56
--- /dev/null
+++ b/src/gallium/drivers/nv20/nv20_vertprog.c
@@ -0,0 +1,838 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_dump.h"
+
+#include "nv20_context.h"
+#include "nv20_state.h"
+
+/* TODO (at least...):
+ *  1. Indexed consts  + ARL
+ *  2. Arb. swz/negation
+ *  3. NV_vp11, NV_vp2, NV_vp3 features
+ *       - extra arith opcodes
+ *       - branching
+ *       - texture sampling
+ *       - indexed attribs
+ *       - indexed results
+ *  4. bugs
+ */
+
+#define SWZ_X 0
+#define SWZ_Y 1
+#define SWZ_Z 2
+#define SWZ_W 3
+#define MASK_X 8
+#define MASK_Y 4
+#define MASK_Z 2
+#define MASK_W 1
+#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W)
+#define DEF_SCALE 0
+#define DEF_CTEST 0
+#include "nv20_shader.h"
+
+#define swz(s,x,y,z,w) nv20_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w)
+#define neg(s) nv20_sr_neg((s))
+#define abs(s) nv20_sr_abs((s))
+
+struct nv20_vpc {
+	struct nv20_vertex_program *vp;
+
+	struct nv20_vertex_program_exec *vpi;
+
+	unsigned output_map[PIPE_MAX_SHADER_OUTPUTS];
+
+	int high_temp;
+	int temp_temp_count;
+
+	struct nv20_sreg *imm;
+	unsigned nr_imm;
+};
+
+static struct nv20_sreg
+temp(struct nv20_vpc *vpc)
+{
+	int idx;
+
+	idx  = vpc->temp_temp_count++;
+	idx += vpc->high_temp + 1;
+	return nv20_sr(NV30SR_TEMP, idx);
+}
+
+static struct nv20_sreg
+constant(struct nv20_vpc *vpc, int pipe, float x, float y, float z, float w)
+{
+	struct nv20_vertex_program *vp = vpc->vp;
+	struct nv20_vertex_program_data *vpd;
+	int idx;
+
+	if (pipe >= 0) {
+		for (idx = 0; idx < vp->nr_consts; idx++) {
+			if (vp->consts[idx].index == pipe)
+				return nv20_sr(NV30SR_CONST, idx);
+		}
+	}
+
+	idx = vp->nr_consts++;
+	vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts);
+	vpd = &vp->consts[idx];
+
+	vpd->index = pipe;
+	vpd->value[0] = x;
+	vpd->value[1] = y;
+	vpd->value[2] = z;
+	vpd->value[3] = w;
+	return nv20_sr(NV30SR_CONST, idx);
+}
+
+#define arith(cc,s,o,d,m,s0,s1,s2) \
+	nv20_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2))
+
+static void
+emit_src(struct nv20_vpc *vpc, uint32_t *hw, int pos, struct nv20_sreg src)
+{
+	struct nv20_vertex_program *vp = vpc->vp;
+	uint32_t sr = 0;
+
+	switch (src.type) {
+	case NV30SR_TEMP:
+		sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT);
+		sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT);
+		break;
+	case NV30SR_INPUT:
+		sr |= (NV30_VP_SRC_REG_TYPE_INPUT <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		vp->ir |= (1 << src.index);
+		hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT);
+		break;
+	case NV30SR_CONST:
+		sr |= (NV30_VP_SRC_REG_TYPE_CONST <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		assert(vpc->vpi->const_index == -1 ||
+		       vpc->vpi->const_index == src.index);
+		vpc->vpi->const_index = src.index;
+		break;
+	case NV30SR_NONE:
+		sr |= (NV30_VP_SRC_REG_TYPE_INPUT <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		break;
+	default:
+		assert(0);
+	}
+
+	if (src.negate)
+		sr |= NV30_VP_SRC_NEGATE;
+
+	if (src.abs)
+		hw[0] |= (1 << (21 + pos));
+
+	sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) |
+	       (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) |
+	       (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) |
+	       (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT));
+
+/*
+ * |VVV|
+ * d�.�b
+ *  \u/
+ *
+ */
+
+	switch (pos) {
+	case 0:
+		hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >>
+			  NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT;
+		hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) <<
+			  NV30_VP_INST_SRC0L_SHIFT;
+		break;
+	case 1:
+		hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT;
+		break;
+	case 2:
+		hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >>
+			  NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT;
+		hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) <<
+			  NV30_VP_INST_SRC2L_SHIFT;
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+emit_dst(struct nv20_vpc *vpc, uint32_t *hw, int slot, struct nv20_sreg dst)
+{
+	struct nv20_vertex_program *vp = vpc->vp;
+
+	switch (dst.type) {
+	case NV30SR_TEMP:
+		hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT);
+		break;
+	case NV30SR_OUTPUT:
+		switch (dst.index) {
+		case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break;
+		case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break;
+		case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break;
+		case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break;
+		case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break;
+		case NV30_VP_INST_DEST_PSZ  : vp->or |= (1 << 5); break;
+		case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break;
+		case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break;
+		case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break;
+		case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break;
+		case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break;
+		case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break;
+		case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break;
+		case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break;
+		default:
+			break;
+		}
+
+		hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT);
+		hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20);
+
+		/*XXX: no way this is entirely correct, someone needs to
+		 *     figure out what exactly it is.
+		 */
+		hw[3] |= 0x800;
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+nv20_vp_arith(struct nv20_vpc *vpc, int slot, int op,
+	      struct nv20_sreg dst, int mask,
+	      struct nv20_sreg s0, struct nv20_sreg s1,
+	      struct nv20_sreg s2)
+{
+	struct nv20_vertex_program *vp = vpc->vp;
+	uint32_t *hw;
+
+	vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi));
+	vpc->vpi = &vp->insns[vp->nr_insns - 1];
+	memset(vpc->vpi, 0, sizeof(*vpc->vpi));
+	vpc->vpi->const_index = -1;
+
+	hw = vpc->vpi->data;
+
+	hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT);
+	hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) |
+		  (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) |
+		  (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) |
+		  (3 << NV30_VP_INST_COND_SWZ_W_SHIFT));
+
+	hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT);
+//	hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK;
+//	hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT);
+
+	if (dst.type == NV30SR_OUTPUT) {
+		if (slot)
+			hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT);
+		else
+			hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT);
+	} else {
+		if (slot)
+			hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT);
+		else
+			hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT);
+	}
+
+	emit_dst(vpc, hw, slot, dst);
+	emit_src(vpc, hw, 0, s0);
+	emit_src(vpc, hw, 1, s1);
+	emit_src(vpc, hw, 2, s2);
+}
+
+static INLINE struct nv20_sreg
+tgsi_src(struct nv20_vpc *vpc, const struct tgsi_full_src_register *fsrc) {
+	struct nv20_sreg src;
+
+	switch (fsrc->SrcRegister.File) {
+	case TGSI_FILE_INPUT:
+		src = nv20_sr(NV30SR_INPUT, fsrc->SrcRegister.Index);
+		break;
+	case TGSI_FILE_CONSTANT:
+		src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0);
+		break;
+	case TGSI_FILE_IMMEDIATE:
+		src = vpc->imm[fsrc->SrcRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		if (vpc->high_temp < fsrc->SrcRegister.Index)
+			vpc->high_temp = fsrc->SrcRegister.Index;
+		src = nv20_sr(NV30SR_TEMP, fsrc->SrcRegister.Index);
+		break;
+	default:
+		NOUVEAU_ERR("bad src file\n");
+		break;
+	}
+
+	src.abs = fsrc->SrcRegisterExtMod.Absolute;
+	src.negate = fsrc->SrcRegister.Negate;
+	src.swz[0] = fsrc->SrcRegister.SwizzleX;
+	src.swz[1] = fsrc->SrcRegister.SwizzleY;
+	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
+	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	return src;
+}
+
+static INLINE struct nv20_sreg
+tgsi_dst(struct nv20_vpc *vpc, const struct tgsi_full_dst_register *fdst) {
+	struct nv20_sreg dst;
+
+	switch (fdst->DstRegister.File) {
+	case TGSI_FILE_OUTPUT:
+		dst = nv20_sr(NV30SR_OUTPUT,
+			      vpc->output_map[fdst->DstRegister.Index]);
+
+		break;
+	case TGSI_FILE_TEMPORARY:
+		dst = nv20_sr(NV30SR_TEMP, fdst->DstRegister.Index);
+		if (vpc->high_temp < dst.index)
+			vpc->high_temp = dst.index;
+		break;
+	default:
+		NOUVEAU_ERR("bad dst file\n");
+		break;
+	}
+
+	return dst;
+}
+
+static INLINE int
+tgsi_mask(uint tgsi)
+{
+	int mask = 0;
+
+	if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X;
+	if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y;
+	if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z;
+	if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W;
+	return mask;
+}
+
+static boolean
+nv20_vertprog_parse_instruction(struct nv20_vpc *vpc,
+				const struct tgsi_full_instruction *finst)
+{
+	struct nv20_sreg src[3], dst, tmp;
+	struct nv20_sreg none = nv20_sr(NV30SR_NONE, 0);
+	int mask;
+	int ai = -1, ci = -1;
+	int i;
+
+	if (finst->Instruction.Opcode == TGSI_OPCODE_END)
+		return TRUE;
+
+	vpc->temp_temp_count = 0;
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+			src[i] = tgsi_src(vpc, fsrc);
+		}
+	}
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
+				ai = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		/*XXX: index comparison is broken now that consts come from
+		 *     two different register files.
+		 */
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_IMMEDIATE:
+			if (ci == -1 || ci == fsrc->SrcRegister.Index) {
+				ci = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_TEMPORARY:
+			/* handled above */
+			break;
+		default:
+			NOUVEAU_ERR("bad src file\n");
+			return FALSE;
+		}
+	}
+
+	dst  = tgsi_dst(vpc, &finst->FullDstRegisters[0]);
+	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+
+	switch (finst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none);
+		break;
+	case TGSI_OPCODE_ADD:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]);
+		break;
+	case TGSI_OPCODE_ARL:
+		arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_DP3:
+		arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DP4:
+		arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DPH:
+		arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DST:
+		arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_EX2:
+		arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_EXP:
+		arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_FLR:
+		arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FRC:
+		arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_LG2:
+		arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LIT:
+		arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LOG:
+		arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_MAD:
+		arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAX:
+		arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MIN:
+		arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MOV:
+		arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_MUL:
+		arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_POW:
+		tmp = temp(vpc);
+		arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none,
+		      swz(src[0], X, X, X, X));
+		arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X),
+		      swz(src[1], X, X, X, X), none);
+		arith(vpc, 1, OP_EX2, dst, mask, none, none,
+		      swz(tmp, X, X, X, X));
+		break;
+	case TGSI_OPCODE_RCP:
+		arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_RET:
+		break;
+	case TGSI_OPCODE_RSQ:
+		arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_SGE:
+		arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SGT:
+		arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SLT:
+		arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SUB:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1]));
+		break;
+	case TGSI_OPCODE_XPD:
+		tmp = temp(vpc);
+		arith(vpc, 0, OP_MUL, tmp, mask,
+		      swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none);
+		arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W),
+		      swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y),
+		      neg(tmp));
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv20_vertprog_parse_decl_output(struct nv20_vpc *vpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	int hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = NV30_VP_INST_DEST_POS;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV30_VP_INST_DEST_COL0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV30_VP_INST_DEST_COL1;
+		} else {
+			NOUVEAU_ERR("bad colour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_BCOLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV30_VP_INST_DEST_BFC0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV30_VP_INST_DEST_BFC1;
+		} else {
+			NOUVEAU_ERR("bad bcolour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_FOG:
+		hw = NV30_VP_INST_DEST_FOGC;
+		break;
+	case TGSI_SEMANTIC_PSIZE:
+		hw = NV30_VP_INST_DEST_PSZ;
+		break;
+	case TGSI_SEMANTIC_GENERIC:
+		if (fdec->Semantic.SemanticIndex <= 7) {
+			hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex);
+		} else {
+			NOUVEAU_ERR("bad generic semantic index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad output semantic\n");
+		return FALSE;
+	}
+
+	vpc->output_map[fdec->DeclarationRange.First] = hw;
+	return TRUE;
+}
+
+static boolean
+nv20_vertprog_prepare(struct nv20_vpc *vpc)
+{
+	struct tgsi_parse_context p;
+	int nr_imm = 0;
+
+	tgsi_parse_init(&p, vpc->vp->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch(tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+			nr_imm++;
+			break;
+		default:
+			break;
+		}
+	}
+	tgsi_parse_free(&p);
+
+	if (nr_imm) {
+		vpc->imm = CALLOC(nr_imm, sizeof(struct nv20_sreg));
+		assert(vpc->imm);
+	}
+
+	return TRUE;
+}
+
+static void
+nv20_vertprog_translate(struct nv20_context *nv20,
+			struct nv20_vertex_program *vp)
+{
+	struct tgsi_parse_context parse;
+	struct nv20_vpc *vpc = NULL;
+
+	tgsi_dump(vp->pipe.tokens,0);
+
+	vpc = CALLOC(1, sizeof(struct nv20_vpc));
+	if (!vpc)
+		return;
+	vpc->vp = vp;
+	vpc->high_temp = -1;
+
+	if (!nv20_vertprog_prepare(vpc)) {
+		FREE(vpc);
+		return;
+	}
+
+	tgsi_parse_init(&parse, vp->pipe.tokens);
+
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		tgsi_parse_token(&parse);
+
+		switch (parse.FullToken.Token.Type) {
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *fdec;
+			fdec = &parse.FullToken.FullDeclaration;
+			switch (fdec->Declaration.File) {
+			case TGSI_FILE_OUTPUT:
+				if (!nv20_vertprog_parse_decl_output(vpc, fdec))
+					goto out_err;
+				break;
+			default:
+				break;
+			}
+		}
+			break;
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			const struct tgsi_full_immediate *imm;
+
+			imm = &parse.FullToken.FullImmediate;
+			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
+//			assert(imm->Immediate.Size == 4);
+			vpc->imm[vpc->nr_imm++] =
+				constant(vpc, -1,
+					 imm->u.ImmediateFloat32[0].Float,
+					 imm->u.ImmediateFloat32[1].Float,
+					 imm->u.ImmediateFloat32[2].Float,
+					 imm->u.ImmediateFloat32[3].Float);
+		}
+			break;
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+			finst = &parse.FullToken.FullInstruction;
+			if (!nv20_vertprog_parse_instruction(vpc, finst))
+				goto out_err;
+		}
+			break;
+		default:
+			break;
+		}
+	}
+
+	vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST;
+	vp->translated = TRUE;
+out_err:
+	tgsi_parse_free(&parse);
+	FREE(vpc);
+}
+
+static boolean
+nv20_vertprog_validate(struct nv20_context *nv20)
+{ 
+	struct nouveau_winsys *nvws = nv20->nvws;
+	struct pipe_winsys *ws = nv20->pipe.winsys;
+	struct nouveau_grobj *rankine = nv20->screen->rankine;
+	struct nv20_vertex_program *vp;
+	struct pipe_buffer *constbuf;
+	boolean upload_code = FALSE, upload_data = FALSE;
+	int i;
+
+	vp = nv20->vertprog;
+	constbuf = nv20->constbuf[PIPE_SHADER_VERTEX];
+
+	/* Translate TGSI shader into hw bytecode */
+	if (!vp->translated) {
+		nv20_vertprog_translate(nv20, vp);
+		if (!vp->translated)
+			return FALSE;
+	}
+
+	/* Allocate hw vtxprog exec slots */
+	if (!vp->exec) {
+		struct nouveau_resource *heap = nv20->screen->vp_exec_heap;
+		struct nouveau_stateobj *so;
+		uint vplen = vp->nr_insns;
+
+		if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) {
+			while (heap->next && heap->size < vplen) {
+				struct nv20_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->exec);
+			}
+
+			if (nvws->res_alloc(heap, vplen, vp, &vp->exec))
+				assert(0);
+		}
+
+		so = so_new(2, 0);
+		so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1);
+		so_data  (so, vp->exec->start);
+		so_ref(so, &vp->so);
+
+		upload_code = TRUE;
+	}
+
+	/* Allocate hw vtxprog const slots */
+	if (vp->nr_consts && !vp->data) {
+		struct nouveau_resource *heap = nv20->screen->vp_data_heap;
+
+		if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) {
+			while (heap->next && heap->size < vp->nr_consts) {
+				struct nv20_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->data);
+			}
+
+			if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data))
+				assert(0);
+		}
+
+		/*XXX: handle this some day */
+		assert(vp->data->start >= vp->data_start_min);
+
+		upload_data = TRUE;
+		if (vp->data_start != vp->data->start)
+			upload_code = TRUE;
+	}
+
+	/* If exec or data segments moved we need to patch the program to
+	 * fixup offsets and register IDs.
+	 */
+	if (vp->exec_start != vp->exec->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv20_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->has_branch_offset) {
+				assert(0);
+			}
+		}
+
+		vp->exec_start = vp->exec->start;
+	}
+
+	if (vp->nr_consts && vp->data_start != vp->data->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv20_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->const_index >= 0) {
+				vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK;
+				vpi->data[1] |=
+					(vpi->const_index + vp->data->start) <<
+					NV30_VP_INST_CONST_SRC_SHIFT;
+
+			}
+		}
+
+		vp->data_start = vp->data->start;
+	}
+
+	/* Update + Upload constant values */
+	if (vp->nr_consts) {
+		float *map = NULL;
+
+		if (constbuf) {
+			map = ws->buffer_map(ws, constbuf,
+					     PIPE_BUFFER_USAGE_CPU_READ);
+		}
+
+		for (i = 0; i < vp->nr_consts; i++) {
+			struct nv20_vertex_program_data *vpd = &vp->consts[i];
+
+			if (vpd->index >= 0) {
+				if (!upload_data &&
+				    !memcmp(vpd->value, &map[vpd->index * 4],
+					    4 * sizeof(float)))
+					continue;
+				memcpy(vpd->value, &map[vpd->index * 4],
+				       4 * sizeof(float));
+			}
+
+			BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5);
+			OUT_RING  (i + vp->data->start);
+			OUT_RINGp ((uint32_t *)vpd->value, 4);
+		}
+
+		if (constbuf) {
+			ws->buffer_unmap(ws, constbuf);
+		}
+	}
+
+	/* Upload vtxprog */
+	if (upload_code) {
+#if 0
+		for (i = 0; i < vp->nr_insns; i++) {
+			NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n",
+				i, vp->insns[i].data[0], vp->insns[i].data[1],
+				vp->insns[i].data[2], vp->insns[i].data[3]);
+		}
+#endif
+		BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1);
+		OUT_RING  (vp->exec->start);
+		for (i = 0; i < vp->nr_insns; i++) {
+			BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4);
+			OUT_RINGp (vp->insns[i].data, 4);
+		}
+	}
+
+	if (vp->so != nv20->state.hw[NV30_STATE_VERTPROG]) {
+		so_ref(vp->so, &nv20->state.hw[NV30_STATE_VERTPROG]);
+		return TRUE;
+	}
+
+	return FALSE;
+}
+
+void
+nv20_vertprog_destroy(struct nv20_context *nv20, struct nv20_vertex_program *vp)
+{
+	struct nouveau_winsys *nvws = nv20->screen->nvws;
+
+	vp->translated = FALSE;
+
+	if (vp->nr_insns) {
+		FREE(vp->insns);
+		vp->insns = NULL;
+		vp->nr_insns = 0;
+	}
+
+	if (vp->nr_consts) {
+		FREE(vp->consts);
+		vp->consts = NULL;
+		vp->nr_consts = 0;
+	}
+
+	nvws->res_free(&vp->exec);
+	vp->exec_start = 0;
+	nvws->res_free(&vp->data);
+	vp->data_start = 0;
+	vp->data_start_min = 0;
+
+	vp->ir = vp->or = 0;
+	so_ref(NULL, &vp->so);
+}
+
+struct nv20_state_entry nv20_state_vertprog = {
+	.validate = nv20_vertprog_validate,
+	.dirty = {
+		.pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/,
+		.hw = NV30_STATE_VERTPROG,
+	}
+};
diff --git a/src/gallium/drivers/nv30/Makefile b/src/gallium/drivers/nv30/Makefile
new file mode 100644
index 00000000000..69f2790dfe2
--- /dev/null
+++ b/src/gallium/drivers/nv30/Makefile
@@ -0,0 +1,37 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv30
+
+DRIVER_SOURCES = \
+	nv30_clear.c \
+	nv30_context.c \
+	nv30_draw.c \
+	nv30_fragprog.c \
+	nv30_fragtex.c \
+	nv30_miptree.c \
+	nv30_query.c \
+	nv30_screen.c \
+	nv30_state.c \
+	nv30_state_blend.c \
+	nv30_state_emit.c \
+	nv30_state_fb.c \
+	nv30_state_rasterizer.c \
+	nv30_state_scissor.c \
+	nv30_state_stipple.c \
+	nv30_state_viewport.c \
+	nv30_state_zsa.c \
+	nv30_surface.c \
+	nv30_vbo.c \
+	nv30_vertprog.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv30/nv30_clear.c b/src/gallium/drivers/nv30/nv30_clear.c
new file mode 100644
index 00000000000..8c3ca204d58
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_clear.c
@@ -0,0 +1,13 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv30_context.h"
+
+void
+nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue);
+	ps->status = PIPE_SURFACE_STATUS_CLEAR;
+}
diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c
new file mode 100644
index 00000000000..2bff28aca9c
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_context.c
@@ -0,0 +1,72 @@
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv30_context.h"
+#include "nv30_screen.h"
+
+static void
+nv30_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	
+	if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
+		BEGIN_RING(rankine, 0x1fd8, 1);
+		OUT_RING  (2);
+		BEGIN_RING(rankine, 0x1fd8, 1);
+		OUT_RING  (1);
+	}
+
+	FIRE_RING(fence);
+}
+
+static void
+nv30_destroy(struct pipe_context *pipe)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	if (nv30->draw)
+		draw_destroy(nv30->draw);
+	FREE(nv30);
+}
+
+struct pipe_context *
+nv30_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct nv30_screen *screen = nv30_screen(pscreen);
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv30_context *nv30;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nv30 = CALLOC(1, sizeof(struct nv30_context));
+	if (!nv30)
+		return NULL;
+	nv30->screen = screen;
+	nv30->pctx_id = pctx_id;
+
+	nv30->nvws = nvws;
+
+	nv30->pipe.winsys = ws;
+	nv30->pipe.screen = pscreen;
+	nv30->pipe.destroy = nv30_destroy;
+	nv30->pipe.draw_arrays = nv30_draw_arrays;
+	nv30->pipe.draw_elements = nv30_draw_elements;
+	nv30->pipe.clear = nv30_clear;
+	nv30->pipe.flush = nv30_flush;
+
+	nv30_init_query_functions(nv30);
+	nv30_init_surface_functions(nv30);
+	nv30_init_state_functions(nv30);
+
+	/* Create, configure, and install fallback swtnl path */
+	nv30->draw = draw_create();
+	draw_wide_point_threshold(nv30->draw, 9999999.0);
+	draw_wide_line_threshold(nv30->draw, 9999999.0);
+	draw_enable_line_stipple(nv30->draw, FALSE);
+	draw_enable_point_sprites(nv30->draw, FALSE);
+	draw_set_rasterize_stage(nv30->draw, nv30_draw_render_stage(nv30));
+
+	return &nv30->pipe;
+}
+	
diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h
new file mode 100644
index 00000000000..b9337697008
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_context.h
@@ -0,0 +1,212 @@
+#ifndef __NV30_CONTEXT_H__
+#define __NV30_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv30_screen *ctx = nv30->screen
+#include "nouveau/nouveau_push.h"
+#include "nouveau/nouveau_stateobj.h"
+
+#include "nv30_state.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+enum nv30_state_index {
+	NV30_STATE_FB = 0,
+	NV30_STATE_VIEWPORT = 1,
+	NV30_STATE_BLEND = 2,
+	NV30_STATE_RAST = 3,
+	NV30_STATE_ZSA = 4,
+	NV30_STATE_BCOL = 5,
+	NV30_STATE_CLIP = 6,
+	NV30_STATE_SCISSOR = 7,
+	NV30_STATE_STIPPLE = 8,
+	NV30_STATE_FRAGPROG = 9,
+	NV30_STATE_VERTPROG = 10,
+	NV30_STATE_FRAGTEX0 = 11,
+	NV30_STATE_FRAGTEX1 = 12,
+	NV30_STATE_FRAGTEX2 = 13,
+	NV30_STATE_FRAGTEX3 = 14,
+	NV30_STATE_FRAGTEX4 = 15,
+	NV30_STATE_FRAGTEX5 = 16,
+	NV30_STATE_FRAGTEX6 = 17,
+	NV30_STATE_FRAGTEX7 = 18,
+	NV30_STATE_FRAGTEX8 = 19,
+	NV30_STATE_FRAGTEX9 = 20,
+	NV30_STATE_FRAGTEX10 = 21,
+	NV30_STATE_FRAGTEX11 = 22,
+	NV30_STATE_FRAGTEX12 = 23,
+	NV30_STATE_FRAGTEX13 = 24,
+	NV30_STATE_FRAGTEX14 = 25,
+	NV30_STATE_FRAGTEX15 = 26,
+	NV30_STATE_VERTTEX0 = 27,
+	NV30_STATE_VERTTEX1 = 28,
+	NV30_STATE_VERTTEX2 = 29,
+	NV30_STATE_VERTTEX3 = 30,
+	NV30_STATE_VTXBUF = 31,
+	NV30_STATE_VTXFMT = 32,
+	NV30_STATE_VTXATTR = 33,
+	NV30_STATE_MAX = 34
+};
+
+#include "nv30_screen.h"
+
+#define NV30_NEW_BLEND		(1 <<  0)
+#define NV30_NEW_RAST		(1 <<  1)
+#define NV30_NEW_ZSA		(1 <<  2)
+#define NV30_NEW_SAMPLER	(1 <<  3)
+#define NV30_NEW_FB		(1 <<  4)
+#define NV30_NEW_STIPPLE	(1 <<  5)
+#define NV30_NEW_SCISSOR	(1 <<  6)
+#define NV30_NEW_VIEWPORT	(1 <<  7)
+#define NV30_NEW_BCOL		(1 <<  8)
+#define NV30_NEW_VERTPROG	(1 <<  9)
+#define NV30_NEW_FRAGPROG	(1 << 10)
+#define NV30_NEW_ARRAYS		(1 << 11)
+#define NV30_NEW_UCP		(1 << 12)
+
+struct nv30_rasterizer_state {
+	struct pipe_rasterizer_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv30_zsa_state {
+	struct pipe_depth_stencil_alpha_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv30_blend_state {
+	struct pipe_blend_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+
+struct nv30_state {
+	unsigned scissor_enabled;
+	unsigned stipple_enabled;
+	unsigned viewport_bypass;
+	unsigned fp_samplers;
+
+	uint64_t dirty;
+	struct nouveau_stateobj *hw[NV30_STATE_MAX];
+};
+
+struct nv30_context {
+	struct pipe_context pipe;
+
+	struct nouveau_winsys *nvws;
+	struct nv30_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	/* HW state derived from pipe states */
+	struct nv30_state state;
+
+	/* Context state */
+	unsigned dirty;
+	struct pipe_scissor_state scissor;
+	unsigned stipple[32];
+	struct nv30_vertex_program *vertprog;
+	struct nv30_fragment_program *fragprog;
+	struct pipe_buffer *constbuf[PIPE_SHADER_TYPES];
+	unsigned constbuf_nr[PIPE_SHADER_TYPES];
+	struct nv30_rasterizer_state *rasterizer;
+	struct nv30_zsa_state *zsa;
+	struct nv30_blend_state *blend;
+	struct pipe_blend_color blend_colour;
+	struct pipe_viewport_state viewport;
+	struct pipe_framebuffer_state framebuffer;
+	struct pipe_buffer *idxbuf;
+	unsigned idxbuf_format;
+	struct nv30_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS];
+	struct nv30_miptree *tex_miptree[PIPE_MAX_SAMPLERS];
+	unsigned nr_samplers;
+	unsigned nr_textures;
+	unsigned dirty_samplers;
+	struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS];
+	unsigned vtxbuf_nr;
+	struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS];
+	unsigned vtxelt_nr;
+	const unsigned *edgeflags;
+};
+
+static INLINE struct nv30_context *
+nv30_context(struct pipe_context *pipe)
+{
+	return (struct nv30_context *)pipe;
+}
+
+struct nv30_state_entry {
+	boolean (*validate)(struct nv30_context *nv30);
+	struct {
+		unsigned pipe;
+		unsigned hw;
+	} dirty;
+};
+
+extern void nv30_init_state_functions(struct nv30_context *nv30);
+extern void nv30_init_surface_functions(struct nv30_context *nv30);
+extern void nv30_init_query_functions(struct nv30_context *nv30);
+
+extern void nv30_screen_init_miptree_functions(struct pipe_screen *pscreen);
+
+/* nv30_draw.c */
+extern struct draw_stage *nv30_draw_render_stage(struct nv30_context *nv30);
+
+/* nv30_vertprog.c */
+extern void nv30_vertprog_destroy(struct nv30_context *,
+				  struct nv30_vertex_program *);
+
+/* nv30_fragprog.c */
+extern void nv30_fragprog_destroy(struct nv30_context *,
+				  struct nv30_fragment_program *);
+
+/* nv30_fragtex.c */
+extern void nv30_fragtex_bind(struct nv30_context *);
+
+/* nv30_state.c and friends */
+extern boolean nv30_state_validate(struct nv30_context *nv30);
+extern void nv30_state_emit(struct nv30_context *nv30);
+extern struct nv30_state_entry nv30_state_rasterizer;
+extern struct nv30_state_entry nv30_state_scissor;
+extern struct nv30_state_entry nv30_state_stipple;
+extern struct nv30_state_entry nv30_state_fragprog;
+extern struct nv30_state_entry nv30_state_vertprog;
+extern struct nv30_state_entry nv30_state_blend;
+extern struct nv30_state_entry nv30_state_blend_colour;
+extern struct nv30_state_entry nv30_state_zsa;
+extern struct nv30_state_entry nv30_state_viewport;
+extern struct nv30_state_entry nv30_state_framebuffer;
+extern struct nv30_state_entry nv30_state_fragtex;
+extern struct nv30_state_entry nv30_state_vbo;
+
+/* nv30_vbo.c */
+extern boolean nv30_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv30_draw_elements(struct pipe_context *pipe,
+				  struct pipe_buffer *indexBuffer,
+				  unsigned indexSize,
+				  unsigned mode, unsigned start,
+				  unsigned count);
+
+/* nv30_clear.c */
+extern void nv30_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+#endif
diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c
new file mode 100644
index 00000000000..74fc138c051
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_draw.c
@@ -0,0 +1,61 @@
+#include "draw/draw_pipe.h"
+
+#include "nv30_context.h"
+
+struct nv30_draw_stage {
+	struct draw_stage draw;
+	struct nv30_context *nv30;
+};
+
+static void
+nv30_draw_point(struct draw_stage *draw, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv30_draw_line(struct draw_stage *draw, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv30_draw_tri(struct draw_stage *draw, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv30_draw_flush(struct draw_stage *draw, unsigned flags)
+{
+}
+
+static void
+nv30_draw_reset_stipple_counter(struct draw_stage *draw)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv30_draw_destroy(struct draw_stage *draw)
+{
+	FREE(draw);
+}
+
+struct draw_stage *
+nv30_draw_render_stage(struct nv30_context *nv30)
+{
+	struct nv30_draw_stage *nv30draw = CALLOC_STRUCT(nv30_draw_stage);
+
+	nv30draw->nv30 = nv30;
+	nv30draw->draw.draw = nv30->draw;
+	nv30draw->draw.point = nv30_draw_point;
+	nv30draw->draw.line = nv30_draw_line;
+	nv30draw->draw.tri = nv30_draw_tri;
+	nv30draw->draw.flush = nv30_draw_flush;
+	nv30draw->draw.reset_stipple_counter = nv30_draw_reset_stipple_counter;
+	nv30draw->draw.destroy = nv30_draw_destroy;
+
+	return &nv30draw->draw;
+}
+
diff --git a/src/gallium/drivers/nv30/nv30_fragprog.c b/src/gallium/drivers/nv30/nv30_fragprog.c
new file mode 100644
index 00000000000..320ba3f4bf4
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_fragprog.c
@@ -0,0 +1,911 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv30_context.h"
+
+#define SWZ_X 0
+#define SWZ_Y 1
+#define SWZ_Z 2
+#define SWZ_W 3
+#define MASK_X 1
+#define MASK_Y 2
+#define MASK_Z 4
+#define MASK_W 8
+#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W)
+#define DEF_SCALE NV30_FP_OP_DST_SCALE_1X
+#define DEF_CTEST NV30_FP_OP_COND_TR
+#include "nv30_shader.h"
+
+#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w)
+#define neg(s) nv30_sr_neg((s))
+#define abs(s) nv30_sr_abs((s))
+#define scale(s,v) nv30_sr_scale((s), NV30_FP_OP_DST_SCALE_##v)
+
+#define MAX_CONSTS 128
+#define MAX_IMM 32
+struct nv30_fpc {
+	struct nv30_fragment_program *fp;
+
+	uint attrib_map[PIPE_MAX_SHADER_INPUTS];
+
+	int high_temp;
+	int temp_temp_count;
+	int num_regs;
+
+	uint depth_id;
+	uint colour_id;
+
+	unsigned inst_offset;
+
+	struct {
+		int pipe;
+		float vals[4];
+	} consts[MAX_CONSTS];
+	int nr_consts;
+
+	struct nv30_sreg imm[MAX_IMM];
+	unsigned nr_imm;
+};
+
+static INLINE struct nv30_sreg
+temp(struct nv30_fpc *fpc)
+{
+	int idx;
+
+	idx  = fpc->temp_temp_count++;
+	idx += fpc->high_temp + 1;
+	return nv30_sr(NV30SR_TEMP, idx);
+}
+
+static INLINE struct nv30_sreg
+constant(struct nv30_fpc *fpc, int pipe, float vals[4])
+{
+	int idx;
+
+	if (fpc->nr_consts == MAX_CONSTS)
+		assert(0);
+	idx = fpc->nr_consts++;
+
+	fpc->consts[idx].pipe = pipe;
+	if (pipe == -1)
+		memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float));
+	return nv30_sr(NV30SR_CONST, idx);
+}
+
+#define arith(cc,s,o,d,m,s0,s1,s2) \
+	nv30_fp_arith((cc), (s), NV30_FP_OP_OPCODE_##o, \
+			(d), (m), (s0), (s1), (s2))
+#define tex(cc,s,o,u,d,m,s0,s1,s2) \
+	nv30_fp_tex((cc), (s), NV30_FP_OP_OPCODE_##o, (u), \
+		    (d), (m), (s0), none, none)
+
+static void
+grow_insns(struct nv30_fpc *fpc, int size)
+{
+	struct nv30_fragment_program *fp = fpc->fp;
+
+	fp->insn_len += size;
+	fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len);
+}
+
+static void
+emit_src(struct nv30_fpc *fpc, int pos, struct nv30_sreg src)
+{
+	struct nv30_fragment_program *fp = fpc->fp;
+	uint32_t *hw = &fp->insn[fpc->inst_offset];
+	uint32_t sr = 0;
+
+	switch (src.type) {
+	case NV30SR_INPUT:
+		sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT);
+		hw[0] |= (src.index << NV30_FP_OP_INPUT_SRC_SHIFT);
+		break;
+	case NV30SR_OUTPUT:
+		sr |= NV30_FP_REG_SRC_HALF;
+		/* fall-through */
+	case NV30SR_TEMP:
+		sr |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT);
+		sr |= (src.index << NV30_FP_REG_SRC_SHIFT);
+		break;
+	case NV30SR_CONST:
+		grow_insns(fpc, 4);
+		hw = &fp->insn[fpc->inst_offset];
+		if (fpc->consts[src.index].pipe >= 0) {
+			struct nv30_fragment_program_data *fpd;
+
+			fp->consts = realloc(fp->consts, ++fp->nr_consts *
+					     sizeof(*fpd));
+			fpd = &fp->consts[fp->nr_consts - 1];
+			fpd->offset = fpc->inst_offset + 4;
+			fpd->index = fpc->consts[src.index].pipe;
+			memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4);
+		} else {
+			memcpy(&fp->insn[fpc->inst_offset + 4],
+				fpc->consts[src.index].vals,
+				sizeof(uint32_t) * 4);
+		}
+
+		sr |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT);	
+		break;
+	case NV30SR_NONE:
+		sr |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT);
+		break;
+	default:
+		assert(0);
+	}
+
+	if (src.negate)
+		sr |= NV30_FP_REG_NEGATE;
+
+	if (src.abs)
+		hw[1] |= (1 << (29 + pos));
+
+	sr |= ((src.swz[0] << NV30_FP_REG_SWZ_X_SHIFT) |
+	       (src.swz[1] << NV30_FP_REG_SWZ_Y_SHIFT) |
+	       (src.swz[2] << NV30_FP_REG_SWZ_Z_SHIFT) |
+	       (src.swz[3] << NV30_FP_REG_SWZ_W_SHIFT));
+
+	hw[pos + 1] |= sr;
+}
+
+static void
+emit_dst(struct nv30_fpc *fpc, struct nv30_sreg dst)
+{
+	struct nv30_fragment_program *fp = fpc->fp;
+	uint32_t *hw = &fp->insn[fpc->inst_offset];
+
+	switch (dst.type) {
+	case NV30SR_TEMP:
+		if (fpc->num_regs < (dst.index + 1))
+			fpc->num_regs = dst.index + 1;
+		break;
+	case NV30SR_OUTPUT:
+		if (dst.index == 1) {
+			fp->fp_control |= 0xe;
+		} else {
+			hw[0] |= NV30_FP_OP_OUT_REG_HALF;
+		}
+		break;
+	case NV30SR_NONE:
+		hw[0] |= (1 << 30);
+		break;
+	default:
+		assert(0);
+	}
+
+	hw[0] |= (dst.index << NV30_FP_OP_OUT_REG_SHIFT);
+}
+
+static void
+nv30_fp_arith(struct nv30_fpc *fpc, int sat, int op,
+	      struct nv30_sreg dst, int mask,
+	      struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2)
+{
+	struct nv30_fragment_program *fp = fpc->fp;
+	uint32_t *hw;
+
+	fpc->inst_offset = fp->insn_len;
+	grow_insns(fpc, 4);
+	hw = &fp->insn[fpc->inst_offset];
+	memset(hw, 0, sizeof(uint32_t) * 4);
+
+	if (op == NV30_FP_OP_OPCODE_KIL)
+		fp->fp_control |= NV34TCL_FP_CONTROL_USES_KIL;
+	hw[0] |= (op << NV30_FP_OP_OPCODE_SHIFT);
+	hw[0] |= (mask << NV30_FP_OP_OUTMASK_SHIFT);
+	hw[2] |= (dst.dst_scale << NV30_FP_OP_DST_SCALE_SHIFT);
+
+	if (sat)
+		hw[0] |= NV30_FP_OP_OUT_SAT;
+
+	if (dst.cc_update)
+		hw[0] |= NV30_FP_OP_COND_WRITE_ENABLE;
+	hw[1] |= (dst.cc_test << NV30_FP_OP_COND_SHIFT);
+	hw[1] |= ((dst.cc_swz[0] << NV30_FP_OP_COND_SWZ_X_SHIFT) |
+		  (dst.cc_swz[1] << NV30_FP_OP_COND_SWZ_Y_SHIFT) |
+		  (dst.cc_swz[2] << NV30_FP_OP_COND_SWZ_Z_SHIFT) |
+		  (dst.cc_swz[3] << NV30_FP_OP_COND_SWZ_W_SHIFT));
+
+	emit_dst(fpc, dst);
+	emit_src(fpc, 0, s0);
+	emit_src(fpc, 1, s1);
+	emit_src(fpc, 2, s2);
+}
+
+static void
+nv30_fp_tex(struct nv30_fpc *fpc, int sat, int op, int unit,
+	    struct nv30_sreg dst, int mask,
+	    struct nv30_sreg s0, struct nv30_sreg s1, struct nv30_sreg s2)
+{
+	struct nv30_fragment_program *fp = fpc->fp;
+
+	nv30_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2);
+
+	fp->insn[fpc->inst_offset] |= (unit << NV30_FP_OP_TEX_UNIT_SHIFT);
+	fp->samplers |= (1 << unit);
+}
+
+static INLINE struct nv30_sreg
+tgsi_src(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc)
+{
+	struct nv30_sreg src;
+
+	switch (fsrc->SrcRegister.File) {
+	case TGSI_FILE_INPUT:
+		src = nv30_sr(NV30SR_INPUT,
+			      fpc->attrib_map[fsrc->SrcRegister.Index]);
+		break;
+	case TGSI_FILE_CONSTANT:
+		src = constant(fpc, fsrc->SrcRegister.Index, NULL);
+		break;
+	case TGSI_FILE_IMMEDIATE:
+		assert(fsrc->SrcRegister.Index < fpc->nr_imm);
+		src = fpc->imm[fsrc->SrcRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index + 1);
+		if (fpc->high_temp < src.index)
+			fpc->high_temp = src.index;
+		break;
+	/* This is clearly insane, but gallium hands us shaders like this.
+	 * Luckily fragprog results are just temp regs..
+	 */
+	case TGSI_FILE_OUTPUT:
+		if (fsrc->SrcRegister.Index == fpc->colour_id)
+			return nv30_sr(NV30SR_OUTPUT, 0);
+		else
+			return nv30_sr(NV30SR_OUTPUT, 1);
+		break;
+	default:
+		NOUVEAU_ERR("bad src file\n");
+		break;
+	}
+
+	src.abs = fsrc->SrcRegisterExtMod.Absolute;
+	src.negate = fsrc->SrcRegister.Negate;
+	src.swz[0] = fsrc->SrcRegister.SwizzleX;
+	src.swz[1] = fsrc->SrcRegister.SwizzleY;
+	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
+	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	return src;
+}
+
+static INLINE struct nv30_sreg
+tgsi_dst(struct nv30_fpc *fpc, const struct tgsi_full_dst_register *fdst) {
+	int idx;
+
+	switch (fdst->DstRegister.File) {
+	case TGSI_FILE_OUTPUT:
+		if (fdst->DstRegister.Index == fpc->colour_id)
+			return nv30_sr(NV30SR_OUTPUT, 0);
+		else
+			return nv30_sr(NV30SR_OUTPUT, 1);
+		break;
+	case TGSI_FILE_TEMPORARY:
+		idx = fdst->DstRegister.Index + 1;
+		if (fpc->high_temp < idx)
+			fpc->high_temp = idx;
+		return nv30_sr(NV30SR_TEMP, idx);
+	case TGSI_FILE_NULL:
+		return nv30_sr(NV30SR_NONE, 0);
+	default:
+		NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File);
+		return nv30_sr(NV30SR_NONE, 0);
+	}
+}
+
+static INLINE int
+tgsi_mask(uint tgsi)
+{
+	int mask = 0;
+
+	if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X;
+	if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y;
+	if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z;
+	if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W;
+	return mask;
+}
+
+static boolean
+src_native_swz(struct nv30_fpc *fpc, const struct tgsi_full_src_register *fsrc,
+	       struct nv30_sreg *src)
+{
+	const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0);
+	struct nv30_sreg tgsi = tgsi_src(fpc, fsrc);
+	uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0;
+	uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX,
+			fsrc->SrcRegisterExtSwz.NegateY,
+			fsrc->SrcRegisterExtSwz.NegateZ,
+			fsrc->SrcRegisterExtSwz.NegateW };
+	uint c;
+
+	for (c = 0; c < 4; c++) {
+		switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) {
+		case TGSI_EXTSWIZZLE_X:
+		case TGSI_EXTSWIZZLE_Y:
+		case TGSI_EXTSWIZZLE_Z:
+		case TGSI_EXTSWIZZLE_W:
+			mask |= (1 << c);
+			break;
+		case TGSI_EXTSWIZZLE_ZERO:
+			zero_mask |= (1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		case TGSI_EXTSWIZZLE_ONE:
+			one_mask |= (1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		default:
+			assert(0);
+		}
+
+		if (!tgsi.negate && neg[c])
+			neg_mask |= (1 << c);
+	}
+
+	if (mask == MASK_ALL && !neg_mask)
+		return TRUE;
+
+	*src = temp(fpc);
+
+	if (mask)
+		arith(fpc, 0, MOV, *src, mask, tgsi, none, none);
+
+	if (zero_mask)
+		arith(fpc, 0, SFL, *src, zero_mask, *src, none, none);
+
+	if (one_mask)
+		arith(fpc, 0, STR, *src, one_mask, *src, none, none);
+
+	if (neg_mask) {
+		struct nv30_sreg one = temp(fpc);
+		arith(fpc, 0, STR, one, neg_mask, one, none, none);
+		arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none);
+	}
+
+	return FALSE;
+}
+
+static boolean
+nv30_fragprog_parse_instruction(struct nv30_fpc *fpc,
+				const struct tgsi_full_instruction *finst)
+{
+	const struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0);
+	struct nv30_sreg src[3], dst, tmp;
+	int mask, sat, unit = 0;
+	int ai = -1, ci = -1;
+	int i;
+
+	if (finst->Instruction.Opcode == TGSI_OPCODE_END)
+		return TRUE;
+
+	fpc->temp_temp_count = 0;
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+			src[i] = tgsi_src(fpc, fsrc);
+		}
+	}
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_TEMPORARY:
+			if (!src_native_swz(fpc, fsrc, &src[i]))
+				continue;
+			break;
+		default:
+			break;
+		}
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
+				ai = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(fpc, fsrc);
+			} else {
+				NOUVEAU_MSG("extra src attr %d\n",
+					 fsrc->SrcRegister.Index);
+				src[i] = temp(fpc);
+				arith(fpc, 0, MOV, src[i], MASK_ALL,
+				      tgsi_src(fpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_IMMEDIATE:
+			if (ci == -1 || ci == fsrc->SrcRegister.Index) {
+				ci = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(fpc, fsrc);
+			} else {
+				src[i] = temp(fpc);
+				arith(fpc, 0, MOV, src[i], MASK_ALL,
+				      tgsi_src(fpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_TEMPORARY:
+			/* handled above */
+			break;
+		case TGSI_FILE_SAMPLER:
+			unit = fsrc->SrcRegister.Index;
+			break;
+		case TGSI_FILE_OUTPUT:
+			break;
+		default:
+			NOUVEAU_ERR("bad src file\n");
+			return FALSE;
+		}
+	}
+
+	dst  = tgsi_dst(fpc, &finst->FullDstRegisters[0]);
+	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+	sat  = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE);
+
+	switch (finst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none);
+		break;
+	case TGSI_OPCODE_ADD:
+		arith(fpc, sat, ADD, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_CMP:
+		tmp = temp(fpc);
+		arith(fpc, sat, MOV, dst, mask, src[2], none, none);
+		tmp.cc_update = 1;
+		arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none);
+		dst.cc_test = NV30_VP_INST_COND_LT;
+		arith(fpc, sat, MOV, dst, mask, src[1], none, none);
+		break;
+	case TGSI_OPCODE_COS:
+		arith(fpc, sat, COS, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_DP3:
+		arith(fpc, sat, DP3, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DP4:
+		arith(fpc, sat, DP4, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DPH:
+		tmp = temp(fpc);
+		arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none);
+		arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X),
+		      swz(src[1], W, W, W, W), none);
+		break;
+	case TGSI_OPCODE_DST:
+		arith(fpc, sat, DST, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_EX2:
+		arith(fpc, sat, EX2, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FLR:
+		arith(fpc, sat, FLR, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FRC:
+		arith(fpc, sat, FRC, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_KILP:
+		arith(fpc, 0, KIL, none, 0, none, none, none);
+		break;
+	case TGSI_OPCODE_KIL:
+		dst = nv30_sr(NV30SR_NONE, 0);
+		dst.cc_update = 1;
+		arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none);
+		dst.cc_update = 0; dst.cc_test = NV30_FP_OP_COND_LT;
+		arith(fpc, 0, KIL, dst, 0, none, none, none);
+		break;
+	case TGSI_OPCODE_LG2:
+		arith(fpc, sat, LG2, dst, mask, src[0], none, none);
+		break;
+//	case TGSI_OPCODE_LIT:
+	case TGSI_OPCODE_LRP:
+		arith(fpc, sat, LRP, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAD:
+		arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAX:
+		arith(fpc, sat, MAX, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MIN:
+		arith(fpc, sat, MIN, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MOV:
+		arith(fpc, sat, MOV, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_MUL:
+		arith(fpc, sat, MUL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_NOISE1:
+	case TGSI_OPCODE_NOISE2:
+	case TGSI_OPCODE_NOISE3:
+	case TGSI_OPCODE_NOISE4:
+		arith(fpc, sat, SFL, dst, mask, none, none, none);
+		break;
+	case TGSI_OPCODE_POW:
+		arith(fpc, sat, POW, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_RCP:
+		arith(fpc, sat, RCP, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_RET:
+		assert(0);
+		break;
+	case TGSI_OPCODE_RFL:
+		arith(fpc, 0, RFL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_RSQ:
+		arith(fpc, sat, RSQ, dst, mask, abs(swz(src[0], X, X, X, X)), none, none);
+		break;
+	case TGSI_OPCODE_SCS:
+		if (mask & MASK_X) {
+			arith(fpc, sat, COS, dst, MASK_X,
+			      swz(src[0], X, X, X, X), none, none);
+		}
+		if (mask & MASK_Y) {
+			arith(fpc, sat, SIN, dst, MASK_Y,
+			      swz(src[0], X, X, X, X), none, none);
+		}
+		break;
+	case TGSI_OPCODE_SIN:
+		arith(fpc, sat, SIN, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_SGE:
+		arith(fpc, sat, SGE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SGT:
+		arith(fpc, sat, SGT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SLT:
+		arith(fpc, sat, SLT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SUB:
+		arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none);
+		break;
+	case TGSI_OPCODE_TEX:
+		tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_TXB:
+		tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_TXP:
+		tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_XPD:
+		tmp = temp(fpc);
+		arith(fpc, 0, MUL, tmp, mask,
+		      swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none);
+		arith(fpc, sat, MAD, dst, (mask & ~MASK_W),
+		      swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y),
+		      neg(tmp));
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv30_fragprog_parse_decl_attrib(struct nv30_fpc *fpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	int hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = NV30_FP_OP_INPUT_SRC_POSITION;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV30_FP_OP_INPUT_SRC_COL0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV30_FP_OP_INPUT_SRC_COL1;
+		} else {
+			NOUVEAU_ERR("bad colour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_FOG:
+		hw = NV30_FP_OP_INPUT_SRC_FOGC;
+		break;
+	case TGSI_SEMANTIC_GENERIC:
+		if (fdec->Semantic.SemanticIndex <= 7) {
+			hw = NV30_FP_OP_INPUT_SRC_TC(fdec->Semantic.
+						     SemanticIndex);
+		} else {
+			NOUVEAU_ERR("bad generic semantic index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad input semantic\n");
+		return FALSE;
+	}
+
+	fpc->attrib_map[fdec->DeclarationRange.First] = hw;
+	return TRUE;
+}
+
+static boolean
+nv30_fragprog_parse_decl_output(struct nv30_fpc *fpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		fpc->depth_id = fdec->DeclarationRange.First;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		fpc->colour_id = fdec->DeclarationRange.First;
+		break;
+	default:
+		NOUVEAU_ERR("bad output semantic\n");
+		return FALSE;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv30_fragprog_prepare(struct nv30_fpc *fpc)
+{
+	struct tgsi_parse_context p;
+	/*int high_temp = -1, i;*/
+
+	tgsi_parse_init(&p, fpc->fp->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch(tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *fdec;
+			fdec = &p.FullToken.FullDeclaration;
+			switch (fdec->Declaration.File) {
+			case TGSI_FILE_INPUT:
+				if (!nv30_fragprog_parse_decl_attrib(fpc, fdec))
+					goto out_err;
+				break;
+			case TGSI_FILE_OUTPUT:
+				if (!nv30_fragprog_parse_decl_output(fpc, fdec))
+					goto out_err;
+				break;
+			/*case TGSI_FILE_TEMPORARY:
+				if (fdec->DeclarationRange.Last > high_temp) {
+					high_temp =
+						fdec->DeclarationRange.Last;
+				}
+				break;*/
+			default:
+				break;
+			}
+		}
+			break;
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			struct tgsi_full_immediate *imm;
+			float vals[4];
+			
+			imm = &p.FullToken.FullImmediate;
+			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
+			assert(fpc->nr_imm < MAX_IMM);
+
+			vals[0] = imm->u.ImmediateFloat32[0].Float;
+			vals[1] = imm->u.ImmediateFloat32[1].Float;
+			vals[2] = imm->u.ImmediateFloat32[2].Float;
+			vals[3] = imm->u.ImmediateFloat32[3].Float;
+			fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals);
+		}
+			break;
+		default:
+			break;
+		}
+	}
+	tgsi_parse_free(&p);
+
+	/*if (++high_temp) {
+		fpc->r_temp = CALLOC(high_temp, sizeof(struct nv30_sreg));
+		for (i = 0; i < high_temp; i++)
+			fpc->r_temp[i] = temp(fpc);
+		fpc->r_temps_discard = 0;
+	}*/
+
+	return TRUE;
+
+out_err:
+	/*if (fpc->r_temp)
+		FREE(fpc->r_temp);*/
+	tgsi_parse_free(&p);
+	return FALSE;
+}
+
+static void
+nv30_fragprog_translate(struct nv30_context *nv30,
+			struct nv30_fragment_program *fp)
+{
+	struct tgsi_parse_context parse;
+	struct nv30_fpc *fpc = NULL;
+
+	tgsi_dump(fp->pipe.tokens,0);
+
+	fpc = CALLOC(1, sizeof(struct nv30_fpc));
+	if (!fpc)
+		return;
+	fpc->fp = fp;
+	fpc->high_temp = -1;
+	fpc->num_regs = 2;
+
+	if (!nv30_fragprog_prepare(fpc)) {
+		FREE(fpc);
+		return;
+	}
+
+	tgsi_parse_init(&parse, fp->pipe.tokens);
+
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		tgsi_parse_token(&parse);
+
+		switch (parse.FullToken.Token.Type) {
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+
+			finst = &parse.FullToken.FullInstruction;
+			if (!nv30_fragprog_parse_instruction(fpc, finst))
+				goto out_err;
+		}
+			break;
+		default:
+			break;
+		}
+	}
+
+	fp->fp_control |= (fpc->num_regs-1)/2;
+	fp->fp_reg_control = (1<<16)|0x4;
+
+	/* Terminate final instruction */
+	fp->insn[fpc->inst_offset] |= 0x00000001;
+
+	/* Append NOP + END instruction, may or may not be necessary. */
+	fpc->inst_offset = fp->insn_len;
+	grow_insns(fpc, 4);
+	fp->insn[fpc->inst_offset + 0] = 0x00000001;
+	fp->insn[fpc->inst_offset + 1] = 0x00000000;
+	fp->insn[fpc->inst_offset + 2] = 0x00000000;
+	fp->insn[fpc->inst_offset + 3] = 0x00000000;
+	
+	fp->translated = TRUE;
+	fp->on_hw = FALSE;
+out_err:
+	tgsi_parse_free(&parse);
+	FREE(fpc);
+}
+
+static void
+nv30_fragprog_upload(struct nv30_context *nv30,
+		     struct nv30_fragment_program *fp)
+{
+	struct pipe_winsys *ws = nv30->pipe.winsys;
+	const uint32_t le = 1;
+	uint32_t *map;
+	int i;
+
+	map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_USAGE_CPU_WRITE);
+
+#if 0
+	for (i = 0; i < fp->insn_len; i++) {
+		fflush(stdout); fflush(stderr);
+		NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]);
+		fflush(stdout); fflush(stderr);
+	}
+#endif
+
+	if ((*(const uint8_t *)&le)) {
+		for (i = 0; i < fp->insn_len; i++) {
+			map[i] = fp->insn[i];
+		}
+	} else {
+		/* Weird swapping for big-endian chips */
+		for (i = 0; i < fp->insn_len; i++) {
+			map[i] = ((fp->insn[i] & 0xffff) << 16) |
+				  ((fp->insn[i] >> 16) & 0xffff);
+		}
+	}
+
+	ws->buffer_unmap(ws, fp->buffer);
+}
+
+static boolean
+nv30_fragprog_validate(struct nv30_context *nv30)
+{
+	struct nv30_fragment_program *fp = nv30->fragprog;
+	struct pipe_buffer *constbuf =
+		nv30->constbuf[PIPE_SHADER_FRAGMENT];
+	struct pipe_winsys *ws = nv30->pipe.winsys;
+	struct nouveau_stateobj *so;
+	boolean new_consts = FALSE;
+	int i;
+
+	if (fp->translated)
+		goto update_constants;
+
+	/*nv30->fallback_swrast &= ~NV30_NEW_FRAGPROG;*/
+	nv30_fragprog_translate(nv30, fp);
+	if (!fp->translated) {
+		/*nv30->fallback_swrast |= NV30_NEW_FRAGPROG;*/
+		return FALSE;
+	}
+
+	fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4);
+	nv30_fragprog_upload(nv30, fp);
+
+	so = so_new(8, 1);
+	so_method(so, nv30->screen->rankine, NV34TCL_FP_ACTIVE_PROGRAM, 1);
+	so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR,
+		  NV34TCL_FP_ACTIVE_PROGRAM_DMA0, NV34TCL_FP_ACTIVE_PROGRAM_DMA1);
+	so_method(so, nv30->screen->rankine, NV34TCL_FP_CONTROL, 1);
+	so_data  (so, fp->fp_control);
+	so_method(so, nv30->screen->rankine, NV34TCL_FP_REG_CONTROL, 1);
+	so_data  (so, fp->fp_reg_control);
+	so_method(so, nv30->screen->rankine, NV34TCL_TX_UNITS_ENABLE, 1);
+	so_data  (so, fp->samplers);
+	so_ref(so, &fp->so);
+
+update_constants:
+	if (fp->nr_consts) {
+		float *map;
+		
+		map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ);
+		for (i = 0; i < fp->nr_consts; i++) {
+			struct nv30_fragment_program_data *fpd = &fp->consts[i];
+			uint32_t *p = &fp->insn[fpd->offset];
+			uint32_t *cb = (uint32_t *)&map[fpd->index * 4];
+
+			if (!memcmp(p, cb, 4 * sizeof(float)))
+				continue;
+			memcpy(p, cb, 4 * sizeof(float));
+			new_consts = TRUE;
+		}
+		ws->buffer_unmap(ws, constbuf);
+
+		if (new_consts)
+			nv30_fragprog_upload(nv30, fp);
+	}
+
+	if (new_consts || fp->so != nv30->state.hw[NV30_STATE_FRAGPROG]) {
+		so_ref(fp->so, &nv30->state.hw[NV30_STATE_FRAGPROG]);
+		return TRUE;
+	}
+
+	return FALSE;
+}
+
+void
+nv30_fragprog_destroy(struct nv30_context *nv30,
+		      struct nv30_fragment_program *fp)
+{
+	if (fp->insn_len)
+		FREE(fp->insn);
+}
+
+struct nv30_state_entry nv30_state_fragprog = {
+	.validate = nv30_fragprog_validate,
+	.dirty = {
+		.pipe = NV30_NEW_FRAGPROG,
+		.hw = NV30_STATE_FRAGPROG
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_fragtex.c b/src/gallium/drivers/nv30/nv30_fragtex.c
new file mode 100644
index 00000000000..b1d2663af38
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_fragtex.c
@@ -0,0 +1,163 @@
+#include "nv30_context.h"
+#include "nouveau/nouveau_util.h"
+
+#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w)                        \
+{                                                                              \
+  TRUE,                                                                        \
+  PIPE_FORMAT_##m,                                                             \
+  NV34TCL_TX_FORMAT_FORMAT_##tf,                                               \
+  (NV34TCL_TX_SWIZZLE_S0_X_##ts0x | NV34TCL_TX_SWIZZLE_S0_Y_##ts0y |           \
+   NV34TCL_TX_SWIZZLE_S0_Z_##ts0z | NV34TCL_TX_SWIZZLE_S0_W_##ts0w |           \
+   NV34TCL_TX_SWIZZLE_S1_X_##ts1x | NV34TCL_TX_SWIZZLE_S1_Y_##ts1y |           \
+   NV34TCL_TX_SWIZZLE_S1_Z_##ts1z | NV34TCL_TX_SWIZZLE_S1_W_##ts1w)            \
+}
+
+struct nv30_texture_format {
+	boolean defined;
+	uint	pipe;
+	int     format;
+	int     swizzle;
+};
+
+static struct nv30_texture_format
+nv30_texture_formats[] = {
+	_(A8R8G8B8_UNORM, A8R8G8B8,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	_(A1R5G5B5_UNORM, A1R5G5B5,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	_(A4R4G4B4_UNORM, A4R4G4B4,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	_(R5G6B5_UNORM  , R5G6B5  ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
+	_(L8_UNORM      , L8      ,   S1,   S1,   S1,  ONE, X, X, X, X),
+	_(A8_UNORM      , L8      , ZERO, ZERO, ZERO,   S1, X, X, X, X),
+	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X),
+	_(A8L8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y),
+//	_(Z16_UNORM     , Z16     ,   S1,   S1,   S1,  ONE, X, X, X, X),
+//	_(Z24S8_UNORM   , Z24     ,   S1,   S1,   S1,  ONE, X, X, X, X),
+	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W),
+	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	_(DXT5_RGBA     , DXT5    ,   S1,   S1,   S1,   S1, X, Y, Z, W),
+	{},
+};
+
+static struct nv30_texture_format *
+nv30_fragtex_format(uint pipe_format)
+{
+	struct nv30_texture_format *tf = nv30_texture_formats;
+	char fs[128];
+
+	while (tf->defined) {
+		if (tf->pipe == pipe_format)
+			return tf;
+		tf++;
+	}
+
+	NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format));
+	return NULL;
+}
+
+
+static struct nouveau_stateobj *
+nv30_fragtex_build(struct nv30_context *nv30, int unit)
+{
+	struct nv30_sampler_state *ps = nv30->tex_sampler[unit];
+	struct nv30_miptree *nv30mt = nv30->tex_miptree[unit];
+	struct pipe_texture *pt = &nv30mt->base;
+	struct nv30_texture_format *tf;
+	struct nouveau_stateobj *so;
+	uint32_t txf, txs , txp;
+	unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+
+	tf = nv30_fragtex_format(pt->format);
+	if (!tf)
+		assert(0);
+
+	txf  = tf->format;
+	txf |= ((pt->last_level>0) ? NV34TCL_TX_FORMAT_MIPMAP : 0);
+	txf |= log2i(pt->width[0]) << 20;
+	txf |= log2i(pt->height[0]) << 24;
+	txf |= log2i(pt->depth[0]) << 28;
+	txf |= NV34TCL_TX_FORMAT_NO_BORDER | 0x10000;
+
+	switch (pt->target) {
+	case PIPE_TEXTURE_CUBE:
+		txf |= NV34TCL_TX_FORMAT_CUBIC;
+		/* fall-through */
+	case PIPE_TEXTURE_2D:
+		txf |= NV34TCL_TX_FORMAT_DIMS_2D;
+		break;
+	case PIPE_TEXTURE_3D:
+		txf |= NV34TCL_TX_FORMAT_DIMS_3D;
+		break;
+	case PIPE_TEXTURE_1D:
+		txf |= NV34TCL_TX_FORMAT_DIMS_1D;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown target %d\n", pt->target);
+		return NULL;
+	}
+
+	if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		txp = 0;
+	} else {
+		txp  = nv30mt->level[0].pitch;
+		txf |= (1<<13) /*FIXME: NV34TCL_TX_FORMAT_LINEAR ? */;
+	}
+
+	txs = tf->swizzle;
+
+	so = so_new(16, 2);
+	so_method(so, nv30->screen->rankine, NV34TCL_TX_OFFSET(unit), 8);
+	so_reloc (so, nv30mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0);
+	so_reloc (so, nv30mt->buffer, txf, tex_flags | NOUVEAU_BO_OR,
+		  NV34TCL_TX_FORMAT_DMA0, NV34TCL_TX_FORMAT_DMA1);
+	so_data  (so, ps->wrap);
+	so_data  (so, NV34TCL_TX_ENABLE_ENABLE | ps->en);
+	so_data  (so, txs);
+	so_data  (so, ps->filt | 0x2000 /*voodoo*/);
+	so_data  (so, (pt->width[0] << NV34TCL_TX_NPOT_SIZE_W_SHIFT) |
+		       pt->height[0]);
+	so_data  (so, ps->bcol);
+
+	return so;
+}
+
+static boolean
+nv30_fragtex_validate(struct nv30_context *nv30)
+{
+	struct nv30_fragment_program *fp = nv30->fragprog;
+	struct nv30_state *state = &nv30->state;
+	struct nouveau_stateobj *so;
+	unsigned samplers, unit;
+
+	samplers = state->fp_samplers & ~fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		so = so_new(2, 0);
+		so_method(so, nv30->screen->rankine, NV34TCL_TX_ENABLE(unit), 1);
+		so_data  (so, 0);
+		so_ref(so, &nv30->state.hw[NV30_STATE_FRAGTEX0 + unit]);
+		state->dirty |= (1ULL << (NV30_STATE_FRAGTEX0 + unit));
+	}
+
+	samplers = nv30->dirty_samplers & fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		so = nv30_fragtex_build(nv30, unit);
+		so_ref(so, &nv30->state.hw[NV30_STATE_FRAGTEX0 + unit]);
+		state->dirty |= (1ULL << (NV30_STATE_FRAGTEX0 + unit));
+	}
+
+	nv30->state.fp_samplers = fp->samplers;
+	return FALSE;
+}
+
+struct nv30_state_entry nv30_state_fragtex = {
+	.validate = nv30_fragtex_validate,
+	.dirty = {
+		.pipe = NV30_NEW_SAMPLER | NV30_NEW_FRAGPROG,
+		.hw = 0
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c
new file mode 100644
index 00000000000..aa670b9a45b
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_miptree.c
@@ -0,0 +1,195 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv30_context.h"
+
+static void
+nv30_miptree_layout(struct nv30_miptree *nv30mt)
+{
+	struct pipe_texture *pt = &nv30mt->base;
+	boolean swizzled = FALSE;
+	uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0];
+	uint offset = 0;
+	int nr_faces, l, f, pitch;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		nr_faces = 6;
+	} else
+	if (pt->target == PIPE_TEXTURE_3D) {
+		nr_faces = pt->depth[0];
+	} else {
+		nr_faces = 1;
+	}
+	
+	pitch = pt->width[0];
+	for (l = 0; l <= pt->last_level; l++) {
+		pt->width[l] = width;
+		pt->height[l] = height;
+		pt->depth[l] = depth;
+		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
+		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
+
+		if (swizzled)
+			pitch = pt->nblocksx[l];
+		pitch = align(pitch, 64);
+
+		nv30mt->level[l].pitch = pitch * pt->block.size;
+		nv30mt->level[l].image_offset =
+			CALLOC(nr_faces, sizeof(unsigned));
+
+		width  = MAX2(1, width  >> 1);
+		height = MAX2(1, height >> 1);
+		depth  = MAX2(1, depth  >> 1);
+	}
+
+	for (f = 0; f < nr_faces; f++) {
+		for (l = 0; l <= pt->last_level; l++) {
+			nv30mt->level[l].image_offset[f] = offset;
+			offset += nv30mt->level[l].pitch * pt->height[l];
+		}
+	}
+
+	nv30mt->total_size = offset;
+}
+
+static struct pipe_texture *
+nv30_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv30_miptree *mt;
+
+	mt = MALLOC(sizeof(struct nv30_miptree));
+	if (!mt)
+		return NULL;
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = pscreen;
+	mt->shadow_tex = NULL;
+	mt->shadow_surface = NULL;
+
+	/* Swizzled textures must be POT */
+	if (pt->width[0] & (pt->width[0] - 1) ||
+	    pt->height[0] & (pt->height[0] - 1))
+		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+	else
+	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
+	                     PIPE_TEXTURE_USAGE_DISPLAY_TARGET))
+		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+	else {
+		switch (pt->format) {
+		/* TODO: Figure out which formats can be swizzled */
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
+		/* XXX: Re-enable when SIFM size limits are fixed */
+		/*case PIPE_FORMAT_R16_SNORM:*/
+			break;
+		default:
+			mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+		}
+	}
+
+	nv30_miptree_layout(mt);
+
+	mt->buffer = ws->buffer_create(ws, 256,
+				       PIPE_BUFFER_USAGE_PIXEL |
+				       NOUVEAU_BUFFER_USAGE_TEXTURE,
+				       mt->total_size);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+
+	return &mt->base;
+}
+
+static void
+nv30_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
+{
+	struct pipe_texture *pt = *ppt;
+	struct nv30_miptree *mt = (struct nv30_miptree *)pt;
+	int l;
+
+	*ppt = NULL;
+	if (--pt->refcount)
+		return;
+
+	pipe_buffer_reference(pscreen, &mt->buffer, NULL);
+	for (l = 0; l <= pt->last_level; l++) {
+		if (mt->level[l].image_offset)
+			FREE(mt->level[l].image_offset);
+	}
+
+	if (mt->shadow_tex) {
+		assert(mt->shadow_surface);
+		pscreen->tex_surface_release(pscreen, &mt->shadow_surface);
+		nv30_miptree_release(pscreen, &mt->shadow_tex);
+	}
+
+	FREE(mt);
+}
+
+static struct pipe_surface *
+nv30_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct nv30_miptree *nv30mt = (struct nv30_miptree *)pt;
+	struct pipe_surface *ps;
+
+	ps = CALLOC_STRUCT(pipe_surface);
+	if (!ps)
+		return NULL;
+	pipe_texture_reference(&ps->texture, pt);
+	pipe_buffer_reference(pscreen, &ps->buffer, nv30mt->buffer);
+	ps->format = pt->format;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = nv30mt->level[level].pitch;
+	ps->usage = flags;
+	ps->status = PIPE_SURFACE_STATUS_DEFINED;
+	ps->refcount = 1;
+	ps->winsys = pscreen->winsys;
+	ps->face = face;
+	ps->level = level;
+	ps->zslice = zslice;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		ps->offset = nv30mt->level[level].image_offset[face];
+	} else
+	if (pt->target == PIPE_TEXTURE_3D) {
+		ps->offset = nv30mt->level[level].image_offset[zslice];
+	} else {
+		ps->offset = nv30mt->level[level].image_offset[0];
+	}
+
+	return ps;
+}
+
+static void
+nv30_miptree_surface_del(struct pipe_screen *pscreen,
+			 struct pipe_surface **psurface)
+{
+	struct pipe_surface *ps = *psurface;
+
+	*psurface = NULL;
+	if (--ps->refcount > 0)
+		return;
+
+	pipe_texture_reference(&ps->texture, NULL);
+	pipe_buffer_reference(pscreen->winsys, &ps->buffer, NULL);
+	FREE(ps);
+}
+
+void
+nv30_screen_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv30_miptree_create;
+	pscreen->texture_release = nv30_miptree_release;
+	pscreen->get_tex_surface = nv30_miptree_surface_new;
+	pscreen->tex_surface_release = nv30_miptree_surface_del;
+}
+
diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c
new file mode 100644
index 00000000000..d40d75f2640
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_query.c
@@ -0,0 +1,122 @@
+#include "pipe/p_context.h"
+
+#include "nv30_context.h"
+
+struct nv30_query {
+	struct nouveau_resource *object;
+	unsigned type;
+	boolean ready;
+	uint64_t result;
+};
+
+static INLINE struct nv30_query *
+nv30_query(struct pipe_query *pipe)
+{
+	return (struct nv30_query *)pipe;
+}
+
+static struct pipe_query *
+nv30_query_create(struct pipe_context *pipe, unsigned query_type)
+{
+	struct nv30_query *q;
+
+	q = CALLOC(1, sizeof(struct nv30_query));
+	q->type = query_type;
+
+	return (struct pipe_query *)q;
+}
+
+static void
+nv30_query_destroy(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_query *q = nv30_query(pq);
+
+	if (q->object)
+		nv30->nvws->res_free(&q->object);
+	FREE(q);
+}
+
+static void
+nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_query *q = nv30_query(pq);
+
+	assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER);
+
+	/* Happens when end_query() is called, then another begin_query()
+	 * without querying the result in-between.  For now we'll wait for
+	 * the existing query to notify completion, but it could be better.
+	 */
+	if (q->object) {
+		uint64 tmp;
+		pipe->get_query_result(pipe, pq, 1, &tmp);
+	}
+
+	if (nv30->nvws->res_alloc(nv30->screen->query_heap, 1, NULL, &q->object))
+		assert(0);
+	nv30->nvws->notifier_reset(nv30->screen->query, q->object->start);
+
+	BEGIN_RING(rankine, NV34TCL_QUERY_RESET, 1);
+	OUT_RING  (1);
+	BEGIN_RING(rankine, NV34TCL_QUERY_UNK17CC, 1);
+	OUT_RING  (1);
+
+	q->ready = FALSE;
+}
+
+static void
+nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_query *q = nv30_query(pq);
+
+	BEGIN_RING(rankine, NV34TCL_QUERY_GET, 1);
+	OUT_RING  ((0x01 << NV34TCL_QUERY_GET_UNK24_SHIFT) |
+		   ((q->object->start * 32) << NV34TCL_QUERY_GET_OFFSET_SHIFT));
+	FIRE_RING(NULL);
+}
+
+static boolean
+nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq,
+		  boolean wait, uint64 *result)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_query *q = nv30_query(pq);
+	struct nouveau_winsys *nvws = nv30->nvws;
+
+	assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER);
+
+	if (!q->ready) {
+		unsigned status;
+
+		status = nvws->notifier_status(nv30->screen->query,
+					       q->object->start);
+		if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) {
+			if (wait == FALSE)
+				return FALSE;
+			nvws->notifier_wait(nv30->screen->query, q->object->start,
+					    NV_NOTIFY_STATE_STATUS_COMPLETED,
+					    0);
+		}
+
+		q->result = nvws->notifier_retval(nv30->screen->query,
+						  q->object->start);
+		q->ready = TRUE;
+		nvws->res_free(&q->object);
+	}
+
+	*result = q->result;
+	return TRUE;
+}
+
+void
+nv30_init_query_functions(struct nv30_context *nv30)
+{
+	nv30->pipe.create_query = nv30_query_create;
+	nv30->pipe.destroy_query = nv30_query_destroy;
+	nv30->pipe.begin_query = nv30_query_begin;
+	nv30->pipe.end_query = nv30_query_end;
+	nv30->pipe.get_query_result = nv30_query_result;
+}
diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c
new file mode 100644
index 00000000000..910a3c456dd
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_screen.c
@@ -0,0 +1,383 @@
+#include "pipe/p_screen.h"
+
+#include "nv30_context.h"
+#include "nv30_screen.h"
+
+#define NV30TCL_CHIPSET_3X_MASK 0x00000003
+#define NV34TCL_CHIPSET_3X_MASK 0x00000010
+#define NV35TCL_CHIPSET_3X_MASK 0x000001e0
+
+static const char *
+nv30_screen_get_name(struct pipe_screen *pscreen)
+{
+	struct nv30_screen *screen = nv30_screen(pscreen);
+	struct nouveau_device *dev = screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv30_screen_get_vendor(struct pipe_screen *pscreen)
+{
+	return "nouveau";
+}
+
+static int
+nv30_screen_get_param(struct pipe_screen *pscreen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 16;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 0;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 1;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 0;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 1;
+	case PIPE_CAP_POINT_SPRITE:
+		return 1;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 2;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 1;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 1;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 13;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 10;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 13;
+	case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
+		return 0;
+	case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
+		return 1;
+	case NOUVEAU_CAP_HW_VTXBUF:
+	case NOUVEAU_CAP_HW_IDXBUF:
+		return 1;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv30_screen_get_paramf(struct pipe_screen *pscreen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 10.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 64.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 8.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 4.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static boolean
+nv30_screen_surface_format_supported(struct pipe_screen *pscreen,
+				     enum pipe_format format,
+				     enum pipe_texture_target target,
+				     unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_A4R4G4B4_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
+		case PIPE_FORMAT_A8L8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+		case PIPE_FORMAT_Z24S8_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static void *
+nv30_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
+		 unsigned flags )
+{
+	struct pipe_winsys	*ws = screen->winsys;
+	struct pipe_surface	*surface_to_map;
+	void			*map;
+
+	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		struct nv30_miptree *mt = (struct nv30_miptree *)surface->texture;
+
+		if (!mt->shadow_tex) {
+			unsigned old_tex_usage = surface->texture->tex_usage;
+			surface->texture->tex_usage = NOUVEAU_TEXTURE_USAGE_LINEAR;
+			mt->shadow_tex = screen->texture_create(screen, surface->texture);
+			surface->texture->tex_usage = old_tex_usage;
+
+			assert(mt->shadow_tex->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR);
+			mt->shadow_surface = screen->get_tex_surface
+			(
+				screen, mt->shadow_tex,
+				surface->face, surface->level, surface->zslice,
+				surface->usage
+			);
+		}
+
+		surface_to_map = mt->shadow_surface;
+	}
+	else
+		surface_to_map = surface;
+
+	assert(surface_to_map);
+
+	map = ws->buffer_map(ws, surface_to_map->buffer, flags);
+	if (!map)
+		return NULL;
+
+	return map + surface_to_map->offset;
+}
+
+static void
+nv30_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
+{
+	struct pipe_winsys	*ws = screen->winsys;
+	struct pipe_surface	*surface_to_unmap;
+
+	/* TODO: Copy from shadow just before push buffer is flushed instead.
+	         There are probably some programs that map/unmap excessively
+	         before rendering. */
+	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		struct nv30_miptree *mt = (struct nv30_miptree *)surface->texture;
+
+		assert(mt->shadow_tex);
+
+		surface_to_unmap = mt->shadow_surface;
+	}
+	else
+		surface_to_unmap = surface;
+
+	assert(surface_to_unmap);
+
+	ws->buffer_unmap(ws, surface_to_unmap->buffer);
+
+	if (surface_to_unmap != surface) {
+		struct nv30_screen *nvscreen = nv30_screen(screen);
+
+		nvscreen->nvws->surface_copy(nvscreen->nvws,
+		                             surface, 0, 0,
+		                             surface_to_unmap, 0, 0,
+		                             surface->width, surface->height);
+	}
+}
+
+static void
+nv30_screen_destroy(struct pipe_screen *pscreen)
+{
+	struct nv30_screen *screen = nv30_screen(pscreen);
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nvws->res_free(&screen->vp_exec_heap);
+	nvws->res_free(&screen->vp_data_heap);
+	nvws->res_free(&screen->query_heap);
+	nvws->notifier_free(&screen->query);
+	nvws->notifier_free(&screen->sync);
+	nvws->grobj_free(&screen->rankine);
+
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv30_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv30_screen *screen = CALLOC_STRUCT(nv30_screen);
+	struct nouveau_stateobj *so;
+	unsigned rankine_class = 0;
+	unsigned chipset = nvws->channel->device->chipset;
+	int ret, i;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	/* 3D object */
+	switch (chipset & 0xf0) {
+	case 0x30:
+		if (NV30TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f)))
+			rankine_class = 0x0397;
+		else
+		if (NV34TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f)))
+			rankine_class = 0x0697;
+		else
+		if (NV35TCL_CHIPSET_3X_MASK & (1 << (chipset & 0x0f)))
+			rankine_class = 0x0497;
+		break;
+	default:
+		break;
+	}
+
+	if (!rankine_class) {
+		NOUVEAU_ERR("Unknown nv3x chipset: nv%02x\n", chipset);
+		return NULL;
+	}
+
+	ret = nvws->grobj_alloc(nvws, rankine_class, &screen->rankine);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		return FALSE;
+	}
+
+	/* Notifier for sync purposes */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv30_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Query objects */
+	ret = nvws->notifier_alloc(nvws, 32, &screen->query);
+	if (ret) {
+		NOUVEAU_ERR("Error initialising query objects: %d\n", ret);
+		nv30_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	ret = nvws->res_init(&screen->query_heap, 0, 32);
+	if (ret) {
+		NOUVEAU_ERR("Error initialising query object heap: %d\n", ret);
+		nv30_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Vtxprog resources */
+	if (nvws->res_init(&screen->vp_exec_heap, 0, 256) ||
+	    nvws->res_init(&screen->vp_data_heap, 0, 256)) {
+		nv30_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Static rankine initialisation */
+	so = so_new(128, 0);
+	so_method(so, screen->rankine, NV34TCL_DMA_NOTIFY, 1);
+	so_data  (so, screen->sync->handle);
+	so_method(so, screen->rankine, NV34TCL_DMA_TEXTURE0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->gart->handle);
+	so_method(so, screen->rankine, NV34TCL_DMA_COLOR1, 1);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->rankine, NV34TCL_DMA_COLOR0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->rankine, NV34TCL_DMA_VTXBUF0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->gart->handle);
+/*	so_method(so, screen->rankine, NV34TCL_DMA_FENCE, 2);
+	so_data  (so, 0);
+	so_data  (so, screen->query->handle);*/
+	so_method(so, screen->rankine, NV34TCL_DMA_IN_MEMORY7, 1);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->rankine, NV34TCL_DMA_IN_MEMORY8, 1);
+	so_data  (so, nvws->channel->vram->handle);
+
+	for (i=1; i<8; i++) {
+		so_method(so, screen->rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(i), 1);
+		so_data  (so, 0);
+		so_method(so, screen->rankine, NV34TCL_VIEWPORT_CLIP_VERT(i), 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, screen->rankine, 0x220, 1);
+	so_data  (so, 1);
+
+	so_method(so, screen->rankine, 0x03b0, 1);
+	so_data  (so, 0x00100000);
+	so_method(so, screen->rankine, 0x1454, 1);
+	so_data  (so, 0);
+	so_method(so, screen->rankine, 0x1d80, 1);
+	so_data  (so, 3);
+	so_method(so, screen->rankine, 0x1450, 1);
+	so_data  (so, 0x00030004);
+	
+	/* NEW */
+	so_method(so, screen->rankine, 0x1e98, 1);
+	so_data  (so, 0);
+	so_method(so, screen->rankine, 0x17e0, 3);
+	so_data  (so, fui(0.0));
+	so_data  (so, fui(0.0));
+	so_data  (so, fui(1.0));
+	so_method(so, screen->rankine, 0x1f80, 16);
+	for (i=0; i<16; i++) {
+		so_data  (so, (i==8) ? 0x0000ffff : 0);
+	}
+
+	so_method(so, screen->rankine, 0x120, 3);
+	so_data  (so, 0);
+	so_data  (so, 1);
+	so_data  (so, 2);
+
+	so_method(so, screen->rankine, 0x1d88, 1);
+	so_data  (so, 0x00001200);
+
+	so_method(so, screen->rankine, NV34TCL_RC_ENABLE, 1);
+	so_data  (so, 0);
+
+	so_method(so, screen->rankine, NV34TCL_DEPTH_RANGE_NEAR, 2);
+	so_data  (so, fui(0.0));
+	so_data  (so, fui(1.0));
+
+	so_method(so, screen->rankine, NV34TCL_MULTISAMPLE_CONTROL, 1);
+	so_data  (so, 0xffff0000);
+
+	/* enables use of vp rather than fixed-function somehow */
+	so_method(so, screen->rankine, 0x1e94, 1);
+	so_data  (so, 0x13);
+
+	so_emit(nvws, so);
+	so_ref(NULL, &so);
+	nvws->push_flush(nvws, 0, NULL);
+
+	screen->pipe.winsys = ws;
+	screen->pipe.destroy = nv30_screen_destroy;
+
+	screen->pipe.get_name = nv30_screen_get_name;
+	screen->pipe.get_vendor = nv30_screen_get_vendor;
+	screen->pipe.get_param = nv30_screen_get_param;
+	screen->pipe.get_paramf = nv30_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv30_screen_surface_format_supported;
+
+	screen->pipe.surface_map = nv30_surface_map;
+	screen->pipe.surface_unmap = nv30_surface_unmap;
+
+	nv30_screen_init_miptree_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h
new file mode 100644
index 00000000000..b7ddc2a9594
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_screen.h
@@ -0,0 +1,35 @@
+#ifndef __NV30_SCREEN_H__
+#define __NV30_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv30_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+
+	unsigned cur_pctx;
+
+	/* HW graphics objects */
+	struct nouveau_grobj *rankine;
+	struct nouveau_notifier *sync;
+
+	/* Query object resources */
+	struct nouveau_notifier *query;
+	struct nouveau_resource *query_heap;
+
+	/* Vtxprog resources */
+	struct nouveau_resource *vp_exec_heap;
+	struct nouveau_resource *vp_data_heap;
+
+	/* Current 3D state of channel */
+	struct nouveau_stateobj *state[NV30_STATE_MAX];
+};
+
+static INLINE struct nv30_screen *
+nv30_screen(struct pipe_screen *screen)
+{
+	return (struct nv30_screen *)screen;
+}
+
+#endif
diff --git a/src/gallium/drivers/nv30/nv30_shader.h b/src/gallium/drivers/nv30/nv30_shader.h
new file mode 100644
index 00000000000..dd3a36f78f3
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_shader.h
@@ -0,0 +1,490 @@
+#ifndef __NV30_SHADER_H__
+#define __NV30_SHADER_H__
+
+/* Vertex programs instruction set
+ *
+ * 128bit opcodes, split into 4 32-bit ones for ease of use.
+ *
+ * Non-native instructions
+ *   ABS - MOV + NV40_VP_INST0_DEST_ABS
+ *   POW - EX2 + MUL + LG2
+ *   SUB - ADD, second source negated
+ *   SWZ - MOV
+ *   XPD -  
+ *
+ * Register access
+ *   - Only one INPUT can be accessed per-instruction (move extras into TEMPs)
+ *   - Only one CONST can be accessed per-instruction (move extras into TEMPs)
+ *
+ * Relative Addressing
+ *   According to the value returned for
+ *   MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB
+ *
+ *   there are only two address registers available.  The destination in the
+ *   ARL instruction is set to TEMP <n> (The temp isn't actually written).
+ *
+ *   When using vanilla ARB_v_p, the proprietary driver will squish both the
+ *   available ADDRESS regs into the first hardware reg in the X and Y
+ *   components.
+ *
+ *   To use an address reg as an index into consts, the CONST_SRC is set to
+ *   (const_base + offset) and INDEX_CONST is set.
+ *
+ *   To access the second address reg use ADDR_REG_SELECT_1. A particular
+ *   component of the address regs is selected with ADDR_SWZ.
+ *
+ *   Only one address register can be accessed per instruction.
+ *
+ * Conditional execution (see NV_vertex_program{2,3} for details) Conditional
+ * execution of an instruction is enabled by setting COND_TEST_ENABLE, and
+ * selecting the condition which will allow the test to pass with
+ * COND_{FL,LT,...}.  It is possible to swizzle the values in the condition
+ * register, which allows for testing against an individual component.
+ *
+ * Branching:
+ *
+ *   The BRA/CAL instructions seem to follow a slightly different opcode
+ *   layout.  The destination instruction ID (IADDR) overlaps a source field.
+ *   Instruction ID's seem to be numbered based on the UPLOAD_FROM_ID FIFO
+ *   command, and is incremented automatically on each UPLOAD_INST FIFO
+ *   command.
+ *
+ *   Conditional branching is achieved by using the condition tests described
+ *   above.  There doesn't appear to be dedicated looping instructions, but
+ *   this can be done using a temp reg + conditional branching.
+ *
+ *   Subroutines may be uploaded before the main program itself, but the first
+ *   executed instruction is determined by the PROGRAM_START_ID FIFO command.
+ *
+ */
+
+/* DWORD 0 */
+
+#define NV30_VP_INST_ADDR_REG_SELECT_1        (1 << 24)
+#define NV30_VP_INST_SRC2_ABS           (1 << 23) /* guess */
+#define NV30_VP_INST_SRC1_ABS           (1 << 22) /* guess */
+#define NV30_VP_INST_SRC0_ABS           (1 << 21) /* guess */
+#define NV30_VP_INST_VEC_RESULT         (1 << 20)
+#define NV30_VP_INST_DEST_TEMP_ID_SHIFT        16
+#define NV30_VP_INST_DEST_TEMP_ID_MASK        (0x0F << 16)
+#define NV30_VP_INST_COND_UPDATE_ENABLE        (1<<15)
+#define NV30_VP_INST_VEC_DEST_TEMP_MASK      (0xF << 16)
+#define NV30_VP_INST_COND_TEST_ENABLE        (1<<14)
+#define NV30_VP_INST_COND_SHIFT          11
+#define NV30_VP_INST_COND_MASK          (0x07 << 11)
+#  define NV30_VP_INST_COND_FL  0 /* guess */  
+#  define NV30_VP_INST_COND_LT  1  
+#  define NV30_VP_INST_COND_EQ  2
+#  define NV30_VP_INST_COND_LE  3
+#  define NV30_VP_INST_COND_GT  4
+#  define NV30_VP_INST_COND_NE  5
+#  define NV30_VP_INST_COND_GE  6
+#  define NV30_VP_INST_COND_TR  7 /* guess */
+#define NV30_VP_INST_COND_SWZ_X_SHIFT        9
+#define NV30_VP_INST_COND_SWZ_X_MASK        (0x03 <<  9)
+#define NV30_VP_INST_COND_SWZ_Y_SHIFT        7
+#define NV30_VP_INST_COND_SWZ_Y_MASK        (0x03 <<  7)
+#define NV30_VP_INST_COND_SWZ_Z_SHIFT        5
+#define NV30_VP_INST_COND_SWZ_Z_MASK        (0x03 <<  5)
+#define NV30_VP_INST_COND_SWZ_W_SHIFT        3
+#define NV30_VP_INST_COND_SWZ_W_MASK        (0x03 <<  3)
+#define NV30_VP_INST_COND_SWZ_ALL_SHIFT        3
+#define NV30_VP_INST_COND_SWZ_ALL_MASK        (0xFF <<  3)
+#define NV30_VP_INST_ADDR_SWZ_SHIFT        1
+#define NV30_VP_INST_ADDR_SWZ_MASK        (0x03 <<  1)
+#define NV30_VP_INST_SCA_OPCODEH_SHIFT        0
+#define NV30_VP_INST_SCA_OPCODEH_MASK        (0x01 <<  0)
+
+/* DWORD 1 */
+#define NV30_VP_INST_SCA_OPCODEL_SHIFT        28
+#define NV30_VP_INST_SCA_OPCODEL_MASK        (0x0F << 28)
+#  define NV30_VP_INST_OP_NOP  0x00
+#  define NV30_VP_INST_OP_RCP  0x02
+#  define NV30_VP_INST_OP_RCC  0x03
+#  define NV30_VP_INST_OP_RSQ  0x04
+#  define NV30_VP_INST_OP_EXP  0x05
+#  define NV30_VP_INST_OP_LOG  0x06
+#  define NV30_VP_INST_OP_LIT  0x07
+#  define NV30_VP_INST_OP_BRA  0x09
+#  define NV30_VP_INST_OP_CAL  0x0B
+#  define NV30_VP_INST_OP_RET  0x0C
+#  define NV30_VP_INST_OP_LG2  0x0D
+#  define NV30_VP_INST_OP_EX2  0x0E
+#  define NV30_VP_INST_OP_SIN  0x0F
+#  define NV30_VP_INST_OP_COS  0x10
+#define NV30_VP_INST_VEC_OPCODE_SHIFT        23
+#define NV30_VP_INST_VEC_OPCODE_MASK        (0x1F << 23)
+#  define NV30_VP_INST_OP_NOPV  0x00
+#  define NV30_VP_INST_OP_MOV  0x01
+#  define NV30_VP_INST_OP_MUL  0x02
+#  define NV30_VP_INST_OP_ADD  0x03
+#  define NV30_VP_INST_OP_MAD  0x04
+#  define NV30_VP_INST_OP_DP3  0x05
+#  define NV30_VP_INST_OP_DP4  0x07
+#  define NV30_VP_INST_OP_DPH  0x06
+#  define NV30_VP_INST_OP_DST  0x08
+#  define NV30_VP_INST_OP_MIN  0x09
+#  define NV30_VP_INST_OP_MAX  0x0A
+#  define NV30_VP_INST_OP_SLT  0x0B
+#  define NV30_VP_INST_OP_SGE  0x0C
+#  define NV30_VP_INST_OP_ARL  0x0D
+#  define NV30_VP_INST_OP_FRC  0x0E
+#  define NV30_VP_INST_OP_FLR  0x0F
+#  define NV30_VP_INST_OP_SEQ  0x10
+#  define NV30_VP_INST_OP_SFL  0x11
+#  define NV30_VP_INST_OP_SGT  0x12
+#  define NV30_VP_INST_OP_SLE  0x13
+#  define NV30_VP_INST_OP_SNE  0x14
+#  define NV30_VP_INST_OP_STR  0x15
+#  define NV30_VP_INST_OP_SSG  0x16
+#  define NV30_VP_INST_OP_ARR  0x17
+#  define NV30_VP_INST_OP_ARA  0x18
+#define NV30_VP_INST_CONST_SRC_SHIFT        14
+#define NV30_VP_INST_CONST_SRC_MASK        (0xFF << 14)
+#define NV30_VP_INST_INPUT_SRC_SHIFT        9    /*NV20*/
+#define NV30_VP_INST_INPUT_SRC_MASK        (0x0F <<  9)  /*NV20*/
+#  define NV30_VP_INST_IN_POS  0    /* These seem to match the bindings specified in */
+#  define NV30_VP_INST_IN_WEIGHT  1    /* the ARB_v_p spec (2.14.3.1) */
+#  define NV30_VP_INST_IN_NORMAL  2    
+#  define NV30_VP_INST_IN_COL0  3    /* Should probably confirm them all though */
+#  define NV30_VP_INST_IN_COL1  4
+#  define NV30_VP_INST_IN_FOGC  5
+#  define NV30_VP_INST_IN_TC0  8
+#  define NV30_VP_INST_IN_TC(n)  (8+n)
+#define NV30_VP_INST_SRC0H_SHIFT        0    /*NV20*/
+#define NV30_VP_INST_SRC0H_MASK          (0x1FF << 0)  /*NV20*/
+
+/* Please note: the IADDR fields overlap other fields because they are used
+ * only for branch instructions.  See Branching: label above
+ *
+ * DWORD 2
+ */
+#define NV30_VP_INST_SRC0L_SHIFT        26    /*NV20*/
+#define NV30_VP_INST_SRC0L_MASK         (0x3F  <<26)  /* NV30_VP_SRC0_LOW_MASK << 26 */
+#define NV30_VP_INST_SRC1_SHIFT         11    /*NV20*/
+#define NV30_VP_INST_SRC1_MASK          (0x7FFF<<11)  /*NV20*/
+#define NV30_VP_INST_SRC2H_SHIFT        0    /*NV20*/
+#define NV30_VP_INST_SRC2H_MASK          (0x7FF << 0)  /* NV30_VP_SRC2_HIGH_MASK >> 4*/
+#define NV30_VP_INST_IADDR_SHIFT        2
+#define NV30_VP_INST_IADDR_MASK          (0xF <<  28)   /* NV30_VP_SRC2_LOW_MASK << 28 */
+
+/* DWORD 3 */
+#define NV30_VP_INST_SRC2L_SHIFT        28    /*NV20*/
+#define NV30_VP_INST_SRC2L_MASK          (0x0F  <<28)  /*NV20*/
+#define NV30_VP_INST_STEMP_WRITEMASK_SHIFT      24
+#define NV30_VP_INST_STEMP_WRITEMASK_MASK      (0x0F << 24)
+#define NV30_VP_INST_VTEMP_WRITEMASK_SHIFT      20
+#define NV30_VP_INST_VTEMP_WRITEMASK_MASK      (0x0F << 20)
+#define NV30_VP_INST_SDEST_WRITEMASK_SHIFT      16
+#define NV30_VP_INST_SDEST_WRITEMASK_MASK      (0x0F << 16)
+#define NV30_VP_INST_VDEST_WRITEMASK_SHIFT      12    /*NV20*/
+#define NV30_VP_INST_VDEST_WRITEMASK_MASK      (0x0F << 12)  /*NV20*/
+#define NV30_VP_INST_DEST_SHIFT        2
+#define NV30_VP_INST_DEST_MASK        (0x0F <<  2)
+#  define NV30_VP_INST_DEST_POS  0
+#  define NV30_VP_INST_DEST_BFC0  1
+#  define NV30_VP_INST_DEST_BFC1  2
+#  define NV30_VP_INST_DEST_COL0  3
+#  define NV30_VP_INST_DEST_COL1  4
+#  define NV30_VP_INST_DEST_FOGC  5
+#  define NV30_VP_INST_DEST_PSZ   6
+#  define NV30_VP_INST_DEST_TC(n)  (8+n)
+
+#define NV30_VP_INST_LAST                           (1 << 0)
+
+/* Useful to split the source selection regs into their pieces */
+#define NV30_VP_SRC0_HIGH_SHIFT                                                6
+#define NV30_VP_SRC0_HIGH_MASK                                        0x00007FC0
+#define NV30_VP_SRC0_LOW_MASK                                         0x0000003F
+#define NV30_VP_SRC2_HIGH_SHIFT                                                4
+#define NV30_VP_SRC2_HIGH_MASK                                        0x00007FF0
+#define NV30_VP_SRC2_LOW_MASK                                         0x0000000F
+
+
+/* Source-register definition - matches NV20 exactly */
+#define NV30_VP_SRC_NEGATE          (1<<14)
+#define NV30_VP_SRC_SWZ_X_SHIFT        12
+#define NV30_VP_SRC_REG_SWZ_X_MASK        (0x03  <<12)
+#define NV30_VP_SRC_SWZ_Y_SHIFT        10
+#define NV30_VP_SRC_REG_SWZ_Y_MASK        (0x03  <<10)
+#define NV30_VP_SRC_SWZ_Z_SHIFT        8
+#define NV30_VP_SRC_REG_SWZ_Z_MASK        (0x03  << 8)
+#define NV30_VP_SRC_SWZ_W_SHIFT        6
+#define NV30_VP_SRC_REG_SWZ_W_MASK        (0x03  << 6)
+#define NV30_VP_SRC_REG_SWZ_ALL_SHIFT        6
+#define NV30_VP_SRC_REG_SWZ_ALL_MASK        (0xFF  << 6)
+#define NV30_VP_SRC_TEMP_SRC_SHIFT        2
+#define NV30_VP_SRC_REG_TEMP_ID_MASK        (0x0F  << 0)
+#define NV30_VP_SRC_REG_TYPE_SHIFT        0
+#define NV30_VP_SRC_REG_TYPE_MASK        (0x03  << 0)
+#define NV30_VP_SRC_REG_TYPE_TEMP  1
+#define NV30_VP_SRC_REG_TYPE_INPUT  2
+#define NV30_VP_SRC_REG_TYPE_CONST  3 /* guess */
+
+/*
+ * Each fragment program opcode appears to be comprised of 4 32-bit values.
+ *
+ *   0 - Opcode, output reg/mask, ATTRIB source
+ *   1 - Source 0
+ *   2 - Source 1
+ *   3 - Source 2
+ *
+ * There appears to be no special difference between result regs and temp regs.
+ *     result.color == R0.xyzw
+ *     result.depth == R1.z
+ * When the fragprog contains instructions to write depth, NV30_TCL_PRIMITIVE_3D_UNK1D78=0
+ * otherwise it is set to 1.
+ *
+ * Constants are inserted directly after the instruction that uses them.
+ * 
+ * It appears that it's not possible to use two input registers in one
+ * instruction as the input sourcing is done in the instruction dword
+ * and not the source selection dwords.  As such instructions such as:
+ * 
+ *     ADD result.color, fragment.color, fragment.texcoord[0];
+ *
+ * must be split into two MOV's and then an ADD (nvidia does this) but
+ * I'm not sure why it's not just one MOV and then source the second input
+ * in the ADD instruction..
+ *
+ * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary
+ * negation requires multiplication with a const.
+ *
+ * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO/SWIZZLE_ONE
+ * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as SWIZZLE_ZERO
+ * is implemented simply by not writing to the relevant components of the destination.
+ *
+ * Conditional execution
+ *   TODO
+ * 
+ * Non-native instructions:
+ *   LIT
+ *   LRP - MAD+MAD
+ *   SUB - ADD, negate second source
+ *   RSQ - LG2 + EX2
+ *   POW - LG2 + MUL + EX2
+ *   SCS - COS + SIN
+ *   XPD
+ */
+
+//== Opcode / Destination selection ==
+#define NV30_FP_OP_PROGRAM_END          (1 << 0)
+#define NV30_FP_OP_OUT_REG_SHIFT        1
+#define NV30_FP_OP_OUT_REG_MASK          (31 << 1)  /* uncertain */
+/* Needs to be set when writing outputs to get expected result.. */
+#define NV30_FP_OP_OUT_REG_HALF          (1 << 7)
+#define NV30_FP_OP_COND_WRITE_ENABLE        (1 << 8)
+#define NV30_FP_OP_OUTMASK_SHIFT        9
+#define NV30_FP_OP_OUTMASK_MASK          (0xF << 9)
+#  define NV30_FP_OP_OUT_X  (1<<9)
+#  define NV30_FP_OP_OUT_Y  (1<<10)
+#  define NV30_FP_OP_OUT_Z  (1<<11)
+#  define NV30_FP_OP_OUT_W  (1<<12)
+/* Uncertain about these, especially the input_src values.. it's possible that
+ * they can be dynamically changed.
+ */
+#define NV30_FP_OP_INPUT_SRC_SHIFT        13
+#define NV30_FP_OP_INPUT_SRC_MASK        (15 << 13)
+#  define NV30_FP_OP_INPUT_SRC_POSITION  0x0
+#  define NV30_FP_OP_INPUT_SRC_COL0  0x1
+#  define NV30_FP_OP_INPUT_SRC_COL1  0x2
+#  define NV30_FP_OP_INPUT_SRC_FOGC  0x3
+#  define NV30_FP_OP_INPUT_SRC_TC0    0x4
+#  define NV30_FP_OP_INPUT_SRC_TC(n)  (0x4 + n)
+#define NV30_FP_OP_TEX_UNIT_SHIFT        17
+#define NV30_FP_OP_TEX_UNIT_MASK        (0xF << 17) /* guess */
+#define NV30_FP_OP_PRECISION_SHIFT        22
+#define NV30_FP_OP_PRECISION_MASK        (3 << 22)
+#   define NV30_FP_PRECISION_FP32  0
+#   define NV30_FP_PRECISION_FP16  1
+#   define NV30_FP_PRECISION_FX12  2
+#define NV30_FP_OP_OPCODE_SHIFT          24
+#define NV30_FP_OP_OPCODE_MASK          (0x3F << 24)
+#  define NV30_FP_OP_OPCODE_NOP  0x00
+#  define NV30_FP_OP_OPCODE_MOV  0x01
+#  define NV30_FP_OP_OPCODE_MUL  0x02
+#  define NV30_FP_OP_OPCODE_ADD  0x03
+#  define NV30_FP_OP_OPCODE_MAD  0x04
+#  define NV30_FP_OP_OPCODE_DP3  0x05
+#  define NV30_FP_OP_OPCODE_DP4  0x06
+#  define NV30_FP_OP_OPCODE_DST  0x07
+#  define NV30_FP_OP_OPCODE_MIN  0x08
+#  define NV30_FP_OP_OPCODE_MAX  0x09
+#  define NV30_FP_OP_OPCODE_SLT  0x0A
+#  define NV30_FP_OP_OPCODE_SGE  0x0B
+#  define NV30_FP_OP_OPCODE_SLE  0x0C
+#  define NV30_FP_OP_OPCODE_SGT  0x0D
+#  define NV30_FP_OP_OPCODE_SNE  0x0E
+#  define NV30_FP_OP_OPCODE_SEQ  0x0F
+#  define NV30_FP_OP_OPCODE_FRC  0x10
+#  define NV30_FP_OP_OPCODE_FLR  0x11
+#  define NV30_FP_OP_OPCODE_KIL  0x12
+#  define NV30_FP_OP_OPCODE_PK4B   0x13
+#  define NV30_FP_OP_OPCODE_UP4B   0x14
+#  define NV30_FP_OP_OPCODE_DDX  0x15 /* can only write XY */
+#  define NV30_FP_OP_OPCODE_DDY  0x16 /* can only write XY */
+#  define NV30_FP_OP_OPCODE_TEX  0x17
+#  define NV30_FP_OP_OPCODE_TXP  0x18
+#  define NV30_FP_OP_OPCODE_TXD  0x19
+#  define NV30_FP_OP_OPCODE_RCP  0x1A
+#  define NV30_FP_OP_OPCODE_RSQ  0x1B
+#  define NV30_FP_OP_OPCODE_EX2  0x1C
+#  define NV30_FP_OP_OPCODE_LG2  0x1D
+#  define NV30_FP_OP_OPCODE_LIT  0x1E
+#  define NV30_FP_OP_OPCODE_LRP  0x1F
+#  define NV30_FP_OP_OPCODE_STR  0x20 
+#  define NV30_FP_OP_OPCODE_SFL  0x21
+#  define NV30_FP_OP_OPCODE_COS  0x22
+#  define NV30_FP_OP_OPCODE_SIN  0x23
+#  define NV30_FP_OP_OPCODE_PK2H   0x24
+#  define NV30_FP_OP_OPCODE_UP2H   0x25
+#  define NV30_FP_OP_OPCODE_POW  0x26
+#  define NV30_FP_OP_OPCODE_PK4UB  0x27
+#  define NV30_FP_OP_OPCODE_UP4UB  0x28
+#  define NV30_FP_OP_OPCODE_PK2US  0x29
+#  define NV30_FP_OP_OPCODE_UP2US  0x2A
+#  define NV30_FP_OP_OPCODE_DP2A   0x2E
+#  define NV30_FP_OP_OPCODE_TXB  0x31
+#  define NV30_FP_OP_OPCODE_RFL  0x36
+#  define NV30_FP_OP_OPCODE_DIV  0x3A
+#define NV30_FP_OP_OUT_SAT          (1 << 31)
+
+/* high order bits of SRC0 */
+#define NV30_FP_OP_OUT_ABS          (1 << 29)
+#define NV30_FP_OP_COND_SWZ_W_SHIFT        27
+#define NV30_FP_OP_COND_SWZ_W_MASK        (3 << 27)
+#define NV30_FP_OP_COND_SWZ_Z_SHIFT        25
+#define NV30_FP_OP_COND_SWZ_Z_MASK        (3 << 25)
+#define NV30_FP_OP_COND_SWZ_Y_SHIFT        23
+#define NV30_FP_OP_COND_SWZ_Y_MASK        (3 << 23)
+#define NV30_FP_OP_COND_SWZ_X_SHIFT        21
+#define NV30_FP_OP_COND_SWZ_X_MASK        (3 << 21)
+#define NV30_FP_OP_COND_SWZ_ALL_SHIFT        21
+#define NV30_FP_OP_COND_SWZ_ALL_MASK        (0xFF << 21)
+#define NV30_FP_OP_COND_SHIFT          18
+#define NV30_FP_OP_COND_MASK          (0x07 << 18)
+#  define NV30_FP_OP_COND_FL  0
+#  define NV30_FP_OP_COND_LT  1
+#  define NV30_FP_OP_COND_EQ  2
+#  define NV30_FP_OP_COND_LE  3
+#  define NV30_FP_OP_COND_GT  4
+#  define NV30_FP_OP_COND_NE  5
+#  define NV30_FP_OP_COND_GE  6
+#  define NV30_FP_OP_COND_TR  7
+
+/* high order bits of SRC1 */
+#define NV30_FP_OP_DST_SCALE_SHIFT        28
+#define NV30_FP_OP_DST_SCALE_MASK        (3 << 28)
+#define NV30_FP_OP_DST_SCALE_1X                                                0
+#define NV30_FP_OP_DST_SCALE_2X                                                1
+#define NV30_FP_OP_DST_SCALE_4X                                                2
+#define NV30_FP_OP_DST_SCALE_8X                                                3
+#define NV30_FP_OP_DST_SCALE_INV_2X                                            5
+#define NV30_FP_OP_DST_SCALE_INV_4X                                            6
+#define NV30_FP_OP_DST_SCALE_INV_8X                                            7
+
+
+/* high order bits of SRC2 */
+#define NV30_FP_OP_INDEX_INPUT          (1 << 30)
+
+//== Register selection ==
+#define NV30_FP_REG_TYPE_SHIFT          0
+#define NV30_FP_REG_TYPE_MASK          (3 << 0)
+#  define NV30_FP_REG_TYPE_TEMP  0
+#  define NV30_FP_REG_TYPE_INPUT  1
+#  define NV30_FP_REG_TYPE_CONST  2
+#define NV30_FP_REG_SRC_SHIFT          2 /* uncertain */
+#define NV30_FP_REG_SRC_MASK          (31 << 2)
+#define NV30_FP_REG_SRC_HALF          (1 << 8)
+#define NV30_FP_REG_SWZ_ALL_SHIFT        9
+#define NV30_FP_REG_SWZ_ALL_MASK        (255 << 9)
+#define NV30_FP_REG_SWZ_X_SHIFT          9
+#define NV30_FP_REG_SWZ_X_MASK          (3 << 9)
+#define NV30_FP_REG_SWZ_Y_SHIFT          11
+#define NV30_FP_REG_SWZ_Y_MASK          (3 << 11)
+#define NV30_FP_REG_SWZ_Z_SHIFT          13
+#define NV30_FP_REG_SWZ_Z_MASK          (3 << 13)
+#define NV30_FP_REG_SWZ_W_SHIFT          15
+#define NV30_FP_REG_SWZ_W_MASK          (3 << 15)
+#  define NV30_FP_SWIZZLE_X  0
+#  define NV30_FP_SWIZZLE_Y  1
+#  define NV30_FP_SWIZZLE_Z  2
+#  define NV30_FP_SWIZZLE_W  3
+#define NV30_FP_REG_NEGATE          (1 << 17)
+
+#define NV30SR_NONE	0
+#define NV30SR_OUTPUT	1
+#define NV30SR_INPUT	2
+#define NV30SR_TEMP	3
+#define NV30SR_CONST	4
+
+struct nv30_sreg {
+	int type;
+	int index;
+
+	int dst_scale;
+
+	int negate;
+	int abs;
+	int swz[4];
+
+	int cc_update;
+	int cc_update_reg;
+	int cc_test;
+	int cc_test_reg;
+	int cc_swz[4];
+};
+
+static INLINE struct nv30_sreg
+nv30_sr(int type, int index)
+{
+	struct nv30_sreg temp = {
+		.type = type,
+		.index = index,
+		.dst_scale = DEF_SCALE,
+		.abs = 0,
+		.negate = 0,
+		.swz = { 0, 1, 2, 3 },
+		.cc_update = 0,
+		.cc_update_reg = 0,
+		.cc_test = DEF_CTEST,
+		.cc_test_reg = 0,
+		.cc_swz = { 0, 1, 2, 3 },
+	};
+	return temp;
+}
+
+static INLINE struct nv30_sreg
+nv30_sr_swz(struct nv30_sreg src, int x, int y, int z, int w)
+{
+	struct nv30_sreg dst = src;
+
+	dst.swz[SWZ_X] = src.swz[x];
+	dst.swz[SWZ_Y] = src.swz[y];
+	dst.swz[SWZ_Z] = src.swz[z];
+	dst.swz[SWZ_W] = src.swz[w];
+	return dst;
+}
+
+static INLINE struct nv30_sreg
+nv30_sr_neg(struct nv30_sreg src)
+{
+	src.negate = !src.negate;
+	return src;
+}
+
+static INLINE struct nv30_sreg
+nv30_sr_abs(struct nv30_sreg src)
+{
+	src.abs = 1;
+	return src;
+}
+
+static INLINE struct nv30_sreg
+nv30_sr_scale(struct nv30_sreg src, int scale)
+{
+	src.dst_scale = scale;
+	return src;
+}
+
+#endif
diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c
new file mode 100644
index 00000000000..fc66075c83f
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state.c
@@ -0,0 +1,724 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv30_context.h"
+#include "nv30_state.h"
+
+static void *
+nv30_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+	struct nv30_blend_state *bso = CALLOC(1, sizeof(*bso));
+	struct nouveau_stateobj *so = so_new(16, 0);
+
+	if (cso->blend_enable) {
+		so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 3);
+		so_data  (so, 1);
+		so_data  (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) |
+			       nvgl_blend_func(cso->rgb_src_factor));
+		so_data  (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 |
+			      nvgl_blend_func(cso->rgb_dst_factor));
+		so_method(so, rankine, NV34TCL_BLEND_EQUATION, 1);
+		so_data  (so, nvgl_blend_eqn(cso->alpha_func) << 16 |
+			      nvgl_blend_eqn(cso->rgb_func));
+	} else {
+		so_method(so, rankine, NV34TCL_BLEND_FUNC_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, rankine, NV34TCL_COLOR_MASK, 1);
+	so_data  (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) |
+		       ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) |
+		       ((cso->colormask & PIPE_MASK_G) ? (0x01 <<  8) : 0) |
+		       ((cso->colormask & PIPE_MASK_B) ? (0x01 <<  0) : 0)));
+
+	if (cso->logicop_enable) {
+		so_method(so, rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 2);
+		so_data  (so, 1);
+		so_data  (so, nvgl_logicop_func(cso->logicop_func));
+	} else {
+		so_method(so, rankine, NV34TCL_COLOR_LOGIC_OP_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, rankine, NV34TCL_DITHER_ENABLE, 1);
+	so_data  (so, cso->dither ? 1 : 0);
+
+	so_ref(so, &bso->so);
+	bso->pipe = *cso;
+	return (void *)bso;
+}
+
+static void
+nv30_blend_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->blend = hwcso;
+	nv30->dirty |= NV30_NEW_BLEND;
+}
+
+static void
+nv30_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_blend_state *bso = hwcso;
+
+	so_ref(NULL, &bso->so);
+	FREE(bso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV34TCL_TX_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV34TCL_TX_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV34TCL_TX_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV34TCL_TX_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV34TCL_TX_WRAP_S_CLAMP;
+		break;
+/*	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+		ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+		ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+		ret = NV34TCL_TX_WRAP_S_MIRROR_CLAMP;
+		break;*/
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV34TCL_TX_WRAP_S_REPEAT;
+		break;
+	}
+
+	return ret >> NV34TCL_TX_WRAP_S_SHIFT;
+}
+
+static void *
+nv30_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv30_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv30_sampler_state));
+
+	ps->fmt = 0;
+	/* TODO: Not all RECTs formats have this bit set, bits 15-8 of format
+	   are the tx format to use. We should store normalized coord flag
+	   in sampler state structure, and set appropriate format in
+	   nvxx_fragtex_build()
+	 */
+	/*NV34TCL_TX_FORMAT_RECT*/
+	/*if (!cso->normalized_coords) {
+		ps->fmt |= (1<<14) ;
+	}*/
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV34TCL_TX_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV34TCL_TX_WRAP_T_SHIFT) |
+		    (wrap_mode(cso->wrap_r) << NV34TCL_TX_WRAP_R_SHIFT));
+
+	ps->en = 0;
+
+	if (cso->max_anisotropy >= 8.0) {
+		ps->en |= NV34TCL_TX_ENABLE_ANISO_8X;
+	} else
+	if (cso->max_anisotropy >= 4.0) {
+		ps->en |= NV34TCL_TX_ENABLE_ANISO_4X;
+	} else
+	if (cso->max_anisotropy >= 2.0) {
+		ps->en |= NV34TCL_TX_ENABLE_ANISO_2X;
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV34TCL_TX_FILTER_MAGNIFY_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV34TCL_TX_FILTER_MAGNIFY_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV34TCL_TX_FILTER_MINIFY_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV34TCL_TX_FILTER_MINIFY_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+	{
+		float limit;
+
+		limit = CLAMP(cso->lod_bias, -16.0, 15.0);
+		ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff;
+
+		limit = CLAMP(cso->max_lod, 0.0, 15.0);
+		ps->en |= (int)(limit) << 14 /*NV34TCL_TX_ENABLE_MIPMAP_MAX_LOD_SHIFT*/;
+
+		limit = CLAMP(cso->min_lod, 0.0, 15.0);
+		ps->en |= (int)(limit) << 26 /*NV34TCL_TX_ENABLE_MIPMAP_MIN_LOD_SHIFT*/;
+	}
+
+	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV34TCL_TX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv30_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv30->tex_sampler[unit] = sampler[unit];
+		nv30->dirty_samplers |= (1 << unit);
+	}
+
+	for (unit = nr; unit < nv30->nr_samplers; unit++) {
+		nv30->tex_sampler[unit] = NULL;
+		nv30->dirty_samplers |= (1 << unit);
+	}
+
+	nv30->nr_samplers = nr;
+	nv30->dirty |= NV30_NEW_SAMPLER;
+}
+
+static void
+nv30_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv30_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		pipe_texture_reference((struct pipe_texture **)
+				       &nv30->tex_miptree[unit], miptree[unit]);
+		nv30->dirty_samplers |= (1 << unit);
+	}
+
+	for (unit = nr; unit < nv30->nr_textures; unit++) {
+		pipe_texture_reference((struct pipe_texture **)
+				       &nv30->tex_miptree[unit], NULL);
+		nv30->dirty_samplers |= (1 << unit);
+	}
+
+	nv30->nr_textures = nr;
+	nv30->dirty |= NV30_NEW_SAMPLER;
+}
+
+static void *
+nv30_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso));
+	struct nouveau_stateobj *so = so_new(32, 0);
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 */
+
+	so_method(so, rankine, NV34TCL_SHADE_MODEL, 1);
+	so_data  (so, cso->flatshade ? NV34TCL_SHADE_MODEL_FLAT :
+				       NV34TCL_SHADE_MODEL_SMOOTH);
+
+	so_method(so, rankine, NV34TCL_LINE_WIDTH, 2);
+	so_data  (so, (unsigned char)(cso->line_width * 8.0) & 0xff);
+	so_data  (so, cso->line_smooth ? 1 : 0);
+	so_method(so, rankine, NV34TCL_LINE_STIPPLE_ENABLE, 2);
+	so_data  (so, cso->line_stipple_enable ? 1 : 0);
+	so_data  (so, (cso->line_stipple_pattern << 16) |
+		       cso->line_stipple_factor);
+
+	so_method(so, rankine, NV34TCL_POINT_SIZE, 1);
+	so_data  (so, fui(cso->point_size));
+
+	so_method(so, rankine, NV34TCL_POLYGON_MODE_FRONT, 6);
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV34TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV34TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV34TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV34TCL_CULL_FACE_BACK);
+			break;
+		}
+		so_data(so, NV34TCL_FRONT_FACE_CCW);
+	} else {
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV34TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV34TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV34TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV34TCL_CULL_FACE_BACK);
+			break;
+		}
+		so_data(so, NV34TCL_FRONT_FACE_CW);
+	}
+	so_data(so, cso->poly_smooth ? 1 : 0);
+	so_data(so, (cso->cull_mode != PIPE_WINDING_NONE) ? 1 : 0);
+
+	so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1);
+	so_data  (so, cso->poly_stipple_enable ? 1 : 0);
+
+	so_method(so, rankine, NV34TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if (cso->offset_cw || cso->offset_ccw) {
+		so_method(so, rankine, NV34TCL_POLYGON_OFFSET_FACTOR, 2);
+		so_data  (so, fui(cso->offset_scale));
+		so_data  (so, fui(cso->offset_units * 2));
+	}
+
+	so_method(so, rankine, NV34TCL_POINT_SPRITE, 1);
+	if (cso->point_sprite) {
+		unsigned psctl = (1 << 0), i;
+
+		for (i = 0; i < 8; i++) {
+			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+				psctl |= (1 << (8 + i));
+		}
+
+		so_data(so, psctl);
+	} else {
+		so_data(so, 0);
+	}
+
+	so_ref(so, &rsso->so);
+	rsso->pipe = *cso;
+	return (void *)rsso;
+}
+
+static void
+nv30_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->rasterizer = hwcso;
+	nv30->dirty |= NV30_NEW_RAST;
+	/*nv30->draw_dirty |= NV30_NEW_RAST;*/
+}
+
+static void
+nv30_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_rasterizer_state *rsso = hwcso;
+
+	so_ref(NULL, &rsso->so);
+	FREE(rsso);
+}
+
+static void *
+nv30_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso));
+	struct nouveau_stateobj *so = so_new(32, 0);
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+
+	so_method(so, rankine, NV34TCL_DEPTH_FUNC, 3);
+	so_data  (so, nvgl_comparison_op(cso->depth.func));
+	so_data  (so, cso->depth.writemask ? 1 : 0);
+	so_data  (so, cso->depth.enabled ? 1 : 0);
+
+	so_method(so, rankine, NV34TCL_ALPHA_FUNC_ENABLE, 3);
+	so_data  (so, cso->alpha.enabled ? 1 : 0);
+	so_data  (so, nvgl_comparison_op(cso->alpha.func));
+	so_data  (so, float_to_ubyte(cso->alpha.ref));
+
+	if (cso->stencil[0].enabled) {
+		so_method(so, rankine, NV34TCL_STENCIL_FRONT_ENABLE, 8);
+		so_data  (so, cso->stencil[0].enabled ? 1 : 0);
+		so_data  (so, cso->stencil[0].write_mask);
+		so_data  (so, nvgl_comparison_op(cso->stencil[0].func));
+		so_data  (so, cso->stencil[0].ref_value);
+		so_data  (so, cso->stencil[0].value_mask);
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zpass_op));
+	} else {
+		so_method(so, rankine, NV34TCL_STENCIL_FRONT_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	if (cso->stencil[1].enabled) {
+		so_method(so, rankine, NV34TCL_STENCIL_BACK_ENABLE, 8);
+		so_data  (so, cso->stencil[1].enabled ? 1 : 0);
+		so_data  (so, cso->stencil[1].write_mask);
+		so_data  (so, nvgl_comparison_op(cso->stencil[1].func));
+		so_data  (so, cso->stencil[1].ref_value);
+		so_data  (so, cso->stencil[1].value_mask);
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zpass_op));
+	} else {
+		so_method(so, rankine, NV34TCL_STENCIL_BACK_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_ref(so, &zsaso->so);
+	zsaso->pipe = *cso;
+	return (void *)zsaso;
+}
+
+static void
+nv30_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->zsa = hwcso;
+	nv30->dirty |= NV30_NEW_ZSA;
+}
+
+static void
+nv30_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_zsa_state *zsaso = hwcso;
+
+	so_ref(NULL, &zsaso->so);
+	FREE(zsaso);
+}
+
+static void *
+nv30_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	/*struct nv30_context *nv30 = nv30_context(pipe);*/
+	struct nv30_vertex_program *vp;
+
+	vp = CALLOC(1, sizeof(struct nv30_vertex_program));
+	vp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	/*vp->draw = draw_create_vertex_shader(nv30->draw, &vp->pipe);*/
+
+	return (void *)vp;
+}
+
+static void
+nv30_vp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->vertprog = hwcso;
+	nv30->dirty |= NV30_NEW_VERTPROG;
+	/*nv30->draw_dirty |= NV30_NEW_VERTPROG;*/
+}
+
+static void
+nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_vertex_program *vp = hwcso;
+
+	/*draw_delete_vertex_shader(nv30->draw, vp->draw);*/
+	nv30_vertprog_destroy(nv30, vp);
+	FREE((void*)vp->pipe.tokens);
+	FREE(vp);
+}
+
+static void *
+nv30_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv30_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv30_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+
+	tgsi_scan_shader(fp->pipe.tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv30_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->fragprog = hwcso;
+	nv30->dirty |= NV30_NEW_FRAGPROG;
+}
+
+static void
+nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nv30_fragment_program *fp = hwcso;
+
+	nv30_fragprog_destroy(nv30, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv30_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->blend_colour = *bcol;
+	nv30->dirty |= NV30_NEW_BCOL;
+}
+
+static void
+nv30_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+}
+
+static void
+nv30_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->constbuf[shader] = buf->buffer;
+	nv30->constbuf_nr[shader] = buf->size / (4 * sizeof(float));
+
+	if (shader == PIPE_SHADER_VERTEX) {
+		nv30->dirty |= NV30_NEW_VERTPROG;
+	} else
+	if (shader == PIPE_SHADER_FRAGMENT) {
+		nv30->dirty |= NV30_NEW_FRAGPROG;
+	}
+}
+
+static void
+nv30_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->framebuffer = *fb;
+	nv30->dirty |= NV30_NEW_FB;
+}
+
+static void
+nv30_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	memcpy(nv30->stipple, stipple->stipple, 4 * 32);
+	nv30->dirty |= NV30_NEW_STIPPLE;
+}
+
+static void
+nv30_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->scissor = *s;
+	nv30->dirty |= NV30_NEW_SCISSOR;
+}
+
+static void
+nv30_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->viewport = *vpt;
+	nv30->dirty |= NV30_NEW_VIEWPORT;
+	/*nv30->draw_dirty |= NV30_NEW_VIEWPORT;*/
+}
+
+static void
+nv30_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	memcpy(nv30->vtxbuf, vb, sizeof(*vb) * count);
+	nv30->vtxbuf_nr = count;
+
+	nv30->dirty |= NV30_NEW_ARRAYS;
+	/*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/
+}
+
+static void
+nv30_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	memcpy(nv30->vtxelt, ve, sizeof(*ve) * count);
+	nv30->vtxelt_nr = count;
+
+	nv30->dirty |= NV30_NEW_ARRAYS;
+	/*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/
+}
+
+static void
+nv30_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+
+	nv30->edgeflags = bitfield;
+	nv30->dirty |= NV30_NEW_ARRAYS;
+	/*nv30->draw_dirty |= NV30_NEW_ARRAYS;*/
+}
+
+void
+nv30_init_state_functions(struct nv30_context *nv30)
+{
+	nv30->pipe.create_blend_state = nv30_blend_state_create;
+	nv30->pipe.bind_blend_state = nv30_blend_state_bind;
+	nv30->pipe.delete_blend_state = nv30_blend_state_delete;
+
+	nv30->pipe.create_sampler_state = nv30_sampler_state_create;
+	nv30->pipe.bind_sampler_states = nv30_sampler_state_bind;
+	nv30->pipe.delete_sampler_state = nv30_sampler_state_delete;
+	nv30->pipe.set_sampler_textures = nv30_set_sampler_texture;
+
+	nv30->pipe.create_rasterizer_state = nv30_rasterizer_state_create;
+	nv30->pipe.bind_rasterizer_state = nv30_rasterizer_state_bind;
+	nv30->pipe.delete_rasterizer_state = nv30_rasterizer_state_delete;
+
+	nv30->pipe.create_depth_stencil_alpha_state =
+		nv30_depth_stencil_alpha_state_create;
+	nv30->pipe.bind_depth_stencil_alpha_state =
+		nv30_depth_stencil_alpha_state_bind;
+	nv30->pipe.delete_depth_stencil_alpha_state =
+		nv30_depth_stencil_alpha_state_delete;
+
+	nv30->pipe.create_vs_state = nv30_vp_state_create;
+	nv30->pipe.bind_vs_state = nv30_vp_state_bind;
+	nv30->pipe.delete_vs_state = nv30_vp_state_delete;
+
+	nv30->pipe.create_fs_state = nv30_fp_state_create;
+	nv30->pipe.bind_fs_state = nv30_fp_state_bind;
+	nv30->pipe.delete_fs_state = nv30_fp_state_delete;
+
+	nv30->pipe.set_blend_color = nv30_set_blend_color;
+	nv30->pipe.set_clip_state = nv30_set_clip_state;
+	nv30->pipe.set_constant_buffer = nv30_set_constant_buffer;
+	nv30->pipe.set_framebuffer_state = nv30_set_framebuffer_state;
+	nv30->pipe.set_polygon_stipple = nv30_set_polygon_stipple;
+	nv30->pipe.set_scissor_state = nv30_set_scissor_state;
+	nv30->pipe.set_viewport_state = nv30_set_viewport_state;
+
+	nv30->pipe.set_edgeflags = nv30_set_edgeflags;
+	nv30->pipe.set_vertex_buffers = nv30_set_vertex_buffers;
+	nv30->pipe.set_vertex_elements = nv30_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv30/nv30_state.h b/src/gallium/drivers/nv30/nv30_state.h
new file mode 100644
index 00000000000..2023278e377
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state.h
@@ -0,0 +1,88 @@
+#ifndef __NV30_STATE_H__
+#define __NV30_STATE_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv30_sampler_state {
+	uint32_t fmt;
+	uint32_t wrap;
+	uint32_t en;
+	uint32_t filt;
+	uint32_t bcol;
+};
+
+struct nv30_vertex_program_exec {
+	uint32_t data[4];
+	boolean has_branch_offset;
+	int const_index;
+};
+
+struct nv30_vertex_program_data {
+	int index; /* immediates == -1 */
+	float value[4];
+};
+
+struct nv30_vertex_program {
+	struct pipe_shader_state pipe;
+
+	boolean translated;
+
+	struct nv30_vertex_program_exec *insns;
+	unsigned nr_insns;
+	struct nv30_vertex_program_data *consts;
+	unsigned nr_consts;
+
+	struct nouveau_resource *exec;
+	unsigned exec_start;
+	struct nouveau_resource *data;
+	unsigned data_start;
+	unsigned data_start_min;
+
+	uint32_t ir;
+	uint32_t or;
+	struct nouveau_stateobj *so;
+};
+
+struct nv30_fragment_program_data {
+	unsigned offset;
+	unsigned index;
+};
+
+struct nv30_fragment_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+
+	boolean translated;
+	boolean on_hw;
+	unsigned samplers;
+
+	uint32_t *insn;
+	int       insn_len;
+
+	struct nv30_fragment_program_data *consts;
+	unsigned nr_consts;
+
+	struct pipe_buffer *buffer;
+
+	uint32_t fp_control;
+	uint32_t fp_reg_control;
+	struct nouveau_stateobj *so;
+};
+
+struct nv30_miptree {
+	struct pipe_texture base;
+
+	struct pipe_buffer *buffer;
+	uint total_size;
+
+	struct pipe_texture *shadow_tex;
+	struct pipe_surface *shadow_surface;
+
+	struct {
+		uint pitch;
+		uint *image_offset;
+	} level[PIPE_MAX_TEXTURE_LEVELS];
+};
+
+#endif
diff --git a/src/gallium/drivers/nv30/nv30_state_blend.c b/src/gallium/drivers/nv30/nv30_state_blend.c
new file mode 100644
index 00000000000..44d43e132a5
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_blend.c
@@ -0,0 +1,40 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_blend_validate(struct nv30_context *nv30)
+{
+	so_ref(nv30->blend->so, &nv30->state.hw[NV30_STATE_BLEND]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_blend = {
+	.validate = nv30_state_blend_validate,
+	.dirty = {
+		.pipe = NV30_NEW_BLEND,
+		.hw = NV30_STATE_BLEND
+	}
+};
+
+static boolean
+nv30_state_blend_colour_validate(struct nv30_context *nv30)
+{
+	struct nouveau_stateobj *so = so_new(2, 0);
+	struct pipe_blend_color *bcol = &nv30->blend_colour;
+
+	so_method(so, nv30->screen->rankine, NV34TCL_BLEND_COLOR, 1);
+	so_data  (so, ((float_to_ubyte(bcol->color[3]) << 24) |
+		       (float_to_ubyte(bcol->color[0]) << 16) |
+		       (float_to_ubyte(bcol->color[1]) <<  8) |
+		       (float_to_ubyte(bcol->color[2]) <<  0)));
+
+	so_ref(so, &nv30->state.hw[NV30_STATE_BCOL]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_blend_colour = {
+	.validate = nv30_state_blend_colour_validate,
+	.dirty = {
+		.pipe = NV30_NEW_BCOL,
+		.hw = NV30_STATE_BCOL
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c
new file mode 100644
index 00000000000..40fed621b24
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_emit.c
@@ -0,0 +1,118 @@
+#include "nv30_context.h"
+#include "nv30_state.h"
+
+static struct nv30_state_entry *render_states[] = {
+	&nv30_state_framebuffer,
+	&nv30_state_rasterizer,
+	&nv30_state_scissor,
+	&nv30_state_stipple,
+	&nv30_state_fragprog,
+	&nv30_state_fragtex,
+	&nv30_state_vertprog,
+	&nv30_state_blend,
+	&nv30_state_blend_colour,
+	&nv30_state_zsa,
+	&nv30_state_viewport,
+	&nv30_state_vbo,
+	NULL
+};
+
+static void
+nv30_state_do_validate(struct nv30_context *nv30,
+		       struct nv30_state_entry **states)
+{
+	const struct pipe_framebuffer_state *fb = &nv30->framebuffer;
+	unsigned i;
+
+	for (i = 0; i < fb->num_cbufs; i++)
+		fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED;
+	if (fb->zsbuf)
+		fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED;
+
+	while (*states) {
+		struct nv30_state_entry *e = *states;
+
+		if (nv30->dirty & e->dirty.pipe) {
+			if (e->validate(nv30)) {
+				nv30->state.dirty |= (1ULL << e->dirty.hw);
+			}
+		}
+
+		states++;
+	}
+	nv30->dirty = 0;
+}
+
+void
+nv30_state_emit(struct nv30_context *nv30)
+{
+	struct nv30_state *state = &nv30->state;
+	struct nv30_screen *screen = nv30->screen;
+	unsigned i, samplers;
+	uint64 states;
+
+	if (nv30->pctx_id != screen->cur_pctx) {
+		for (i = 0; i < NV30_STATE_MAX; i++) {
+			if (state->hw[i] && screen->state[i] != state->hw[i])
+				state->dirty |= (1ULL << i);
+		}
+
+		screen->cur_pctx = nv30->pctx_id;
+	}
+
+	for (i = 0, states = state->dirty; states; i++) {
+		if (!(states & (1ULL << i)))
+			continue;
+		so_ref (state->hw[i], &nv30->screen->state[i]);
+		if (state->hw[i])
+			so_emit(nv30->nvws, nv30->screen->state[i]);
+		states &= ~(1ULL << i);
+	}
+
+	state->dirty = 0;
+
+	so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FB]);
+	for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) {
+		if (!(samplers & (1 << i)))
+			continue;
+		so_emit_reloc_markers(nv30->nvws,
+				      state->hw[NV30_STATE_FRAGTEX0+i]);
+		samplers &= ~(1ULL << i);
+	}
+	so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_FRAGPROG]);
+	if (state->hw[NV30_STATE_VTXBUF] /*&& nv30->render_mode == HW*/)
+		so_emit_reloc_markers(nv30->nvws, state->hw[NV30_STATE_VTXBUF]);
+}
+
+boolean
+nv30_state_validate(struct nv30_context *nv30)
+{
+#if 0
+	boolean was_sw = nv30->fallback_swtnl ? TRUE : FALSE;
+
+	if (nv30->render_mode != HW) {
+		/* Don't even bother trying to go back to hw if none
+		 * of the states that caused swtnl previously have changed.
+		 */
+		if ((nv30->fallback_swtnl & nv30->dirty)
+				!= nv30->fallback_swtnl)
+			return FALSE;
+
+		/* Attempt to go to hwtnl again */
+		nv30->pipe.flush(&nv30->pipe, 0, NULL);
+		nv30->dirty |= (NV30_NEW_VIEWPORT |
+				NV30_NEW_VERTPROG |
+				NV30_NEW_ARRAYS);
+		nv30->render_mode = HW;
+	}
+#endif
+	nv30_state_do_validate(nv30, render_states);
+#if 0
+	if (nv30->fallback_swtnl || nv30->fallback_swrast)
+		return FALSE;
+	
+	if (was_sw)
+		NOUVEAU_ERR("swtnl->hw\n");
+#endif
+	return TRUE;
+}
diff --git a/src/gallium/drivers/nv30/nv30_state_fb.c b/src/gallium/drivers/nv30/nv30_state_fb.c
new file mode 100644
index 00000000000..73bdf7e56c5
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_fb.c
@@ -0,0 +1,140 @@
+#include "nv30_context.h"
+#include "nouveau/nouveau_util.h"
+
+static boolean
+nv30_state_framebuffer_validate(struct nv30_context *nv30)
+{
+	struct pipe_framebuffer_state *fb = &nv30->framebuffer;
+	struct pipe_surface *rt[2], *zeta = NULL;
+	uint32_t rt_enable, rt_format;
+	int i, colour_format = 0, zeta_format = 0;
+	struct nouveau_stateobj *so = so_new(64, 10);
+	unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM;
+	unsigned w = fb->width;
+	unsigned h = fb->height;
+
+	rt_enable = 0;
+	for (i = 0; i < fb->num_cbufs; i++) {
+		if (colour_format) {
+			assert(colour_format == fb->cbufs[i]->format);
+		} else {
+			colour_format = fb->cbufs[i]->format;
+			rt_enable |= (NV34TCL_RT_ENABLE_COLOR0 << i);
+			rt[i] = fb->cbufs[i];
+		}
+	}
+
+	if (rt_enable & NV34TCL_RT_ENABLE_COLOR1)
+		rt_enable |= NV34TCL_RT_ENABLE_MRT;
+
+	if (fb->zsbuf) {
+		zeta_format = fb->zsbuf->format;
+		zeta = fb->zsbuf;
+	}
+
+	if (!(rt[0]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
+		for (i = 1; i < fb->num_cbufs; i++)
+			assert(!(rt[i]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
+
+		/* FIXME: NV34TCL_RT_FORMAT_LOG2_[WIDTH/HEIGHT] */
+		rt_format = NV34TCL_RT_FORMAT_TYPE_SWIZZLED |
+		log2i(fb->width) << 16 /*NV34TCL_RT_FORMAT_LOG2_WIDTH_SHIFT*/ |
+		log2i(fb->height) << 24 /*NV34TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT*/;
+	}
+	else
+		rt_format = NV34TCL_RT_FORMAT_TYPE_LINEAR;
+
+	switch (colour_format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case 0:
+		rt_format |= NV34TCL_RT_FORMAT_COLOR_A8R8G8B8;
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		rt_format |= NV34TCL_RT_FORMAT_COLOR_R5G6B5;
+		break;
+	default:
+		assert(0);
+	}
+
+	switch (zeta_format) {
+	case PIPE_FORMAT_Z16_UNORM:
+		rt_format |= NV34TCL_RT_FORMAT_ZETA_Z16;
+		break;
+	case PIPE_FORMAT_Z24S8_UNORM:
+	case 0:
+		rt_format |= NV34TCL_RT_FORMAT_ZETA_Z24S8;
+		break;
+	default:
+		assert(0);
+	}
+
+	if (rt_enable & NV34TCL_RT_ENABLE_COLOR0) {
+		uint32_t pitch = rt[0]->stride;
+		if (zeta) {
+			pitch |= (zeta->stride << 16);
+		} else {
+			pitch |= (pitch << 16);
+		}
+
+		so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR0, 1);
+		so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv30->nvws->channel->vram->handle,
+			  nv30->nvws->channel->gart->handle);
+		so_method(so, nv30->screen->rankine, NV34TCL_COLOR0_PITCH, 2);
+		so_data  (so, pitch);
+		so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+	}
+
+	if (rt_enable & NV34TCL_RT_ENABLE_COLOR1) {
+		so_method(so, nv30->screen->rankine, NV34TCL_DMA_COLOR1, 1);
+		so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv30->nvws->channel->vram->handle,
+			  nv30->nvws->channel->gart->handle);
+		so_method(so, nv30->screen->rankine, NV34TCL_COLOR1_OFFSET, 2);
+		so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		so_data  (so, rt[1]->stride);
+	}
+
+	if (zeta_format) {
+		so_method(so, nv30->screen->rankine, NV34TCL_DMA_ZETA, 1);
+		so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv30->nvws->channel->vram->handle,
+			  nv30->nvws->channel->gart->handle);
+		so_method(so, nv30->screen->rankine, NV34TCL_ZETA_OFFSET, 1);
+		so_reloc (so, zeta->buffer, zeta->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		/* TODO: allocate LMA depth buffer */
+	}
+
+	so_method(so, nv30->screen->rankine, NV34TCL_RT_ENABLE, 1);
+	so_data  (so, rt_enable);
+	so_method(so, nv30->screen->rankine, NV34TCL_RT_HORIZ, 3);
+	so_data  (so, (w << 16) | 0);
+	so_data  (so, (h << 16) | 0);
+	so_data  (so, rt_format);
+	so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_HORIZ, 2);
+	so_data  (so, (w << 16) | 0);
+	so_data  (so, (h << 16) | 0);
+	so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_CLIP_HORIZ(0), 2);
+	so_data  (so, ((w - 1) << 16) | 0);
+	so_data  (so, ((h - 1) << 16) | 0);
+	so_method(so, nv30->screen->rankine, 0x1d88, 1);
+	so_data  (so, (1 << 12) | h);
+	/* Wonder why this is needed, context should all be set to zero on init */
+	so_method(so, nv30->screen->rankine, NV34TCL_VIEWPORT_TX_ORIGIN, 1);
+	so_data  (so, 0);
+
+	so_ref(so, &nv30->state.hw[NV30_STATE_FB]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_framebuffer = {
+	.validate = nv30_state_framebuffer_validate,
+	.dirty = {
+		.pipe = NV30_NEW_FB,
+		.hw = NV30_STATE_FB
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_rasterizer.c b/src/gallium/drivers/nv30/nv30_state_rasterizer.c
new file mode 100644
index 00000000000..6d1b60e043d
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_rasterizer.c
@@ -0,0 +1,17 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_rasterizer_validate(struct nv30_context *nv30)
+{
+	so_ref(nv30->rasterizer->so,
+	       &nv30->state.hw[NV30_STATE_RAST]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_rasterizer = {
+	.validate = nv30_state_rasterizer_validate,
+	.dirty = {
+		.pipe = NV30_NEW_RAST,
+		.hw = NV30_STATE_RAST
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_scissor.c b/src/gallium/drivers/nv30/nv30_state_scissor.c
new file mode 100644
index 00000000000..1db9bc17955
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_scissor.c
@@ -0,0 +1,35 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_scissor_validate(struct nv30_context *nv30)
+{
+	struct pipe_rasterizer_state *rast = &nv30->rasterizer->pipe;
+	struct pipe_scissor_state *s = &nv30->scissor;
+	struct nouveau_stateobj *so;
+
+	if (nv30->state.hw[NV30_STATE_SCISSOR] &&
+	    (rast->scissor == 0 && nv30->state.scissor_enabled == 0))
+		return FALSE;
+	nv30->state.scissor_enabled = rast->scissor;
+
+	so = so_new(3, 0);
+	so_method(so, nv30->screen->rankine, NV34TCL_SCISSOR_HORIZ, 2);
+	if (nv30->state.scissor_enabled) {
+		so_data  (so, ((s->maxx - s->minx) << 16) | s->minx);
+		so_data  (so, ((s->maxy - s->miny) << 16) | s->miny);
+	} else {
+		so_data  (so, 4096 << 16);
+		so_data  (so, 4096 << 16);
+	}
+
+	so_ref(so, &nv30->state.hw[NV30_STATE_SCISSOR]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_scissor = {
+	.validate = nv30_state_scissor_validate,
+	.dirty = {
+		.pipe = NV30_NEW_SCISSOR | NV30_NEW_RAST,
+		.hw = NV30_STATE_SCISSOR
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_stipple.c b/src/gallium/drivers/nv30/nv30_state_stipple.c
new file mode 100644
index 00000000000..41b42813b49
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_stipple.c
@@ -0,0 +1,39 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_stipple_validate(struct nv30_context *nv30)
+{
+	struct pipe_rasterizer_state *rast = &nv30->rasterizer->pipe;
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+	struct nouveau_stateobj *so;
+
+	if (nv30->state.hw[NV30_STATE_STIPPLE] &&
+	   (rast->poly_stipple_enable == 0 && nv30->state.stipple_enabled == 0))
+		return FALSE;
+
+	if (rast->poly_stipple_enable) {
+		unsigned i;
+
+		so = so_new(35, 0);
+		so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1);
+		so_data  (so, 1);
+		so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_PATTERN(0), 32);
+		for (i = 0; i < 32; i++)
+			so_data(so, nv30->stipple[i]);
+	} else {
+		so = so_new(2, 0);
+		so_method(so, rankine, NV34TCL_POLYGON_STIPPLE_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_ref(so, &nv30->state.hw[NV30_STATE_STIPPLE]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_stipple = {
+	.validate = nv30_state_stipple_validate,
+	.dirty = {
+		.pipe = NV30_NEW_STIPPLE | NV30_NEW_RAST,
+		.hw = NV30_STATE_STIPPLE,
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_viewport.c b/src/gallium/drivers/nv30/nv30_state_viewport.c
new file mode 100644
index 00000000000..951d40ebfdd
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_viewport.c
@@ -0,0 +1,70 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_viewport_validate(struct nv30_context *nv30)
+{
+	struct pipe_viewport_state *vpt = &nv30->viewport;
+	struct nouveau_stateobj *so;
+	unsigned bypass;
+
+	if (/*nv30->render_mode == HW &&*/ !nv30->rasterizer->pipe.bypass_clipping)
+		bypass = 0;
+	else
+		bypass = 1;
+
+	if (nv30->state.hw[NV30_STATE_VIEWPORT] &&
+	    (bypass || !(nv30->dirty & NV30_NEW_VIEWPORT)) &&
+	    nv30->state.viewport_bypass == bypass)
+		return FALSE;
+	nv30->state.viewport_bypass = bypass;
+
+	so = so_new(11, 0);
+	if (!bypass) {
+		so_method(so, nv30->screen->rankine,
+			  NV34TCL_VIEWPORT_TRANSLATE_X, 8);
+		so_data  (so, fui(vpt->translate[0]));
+		so_data  (so, fui(vpt->translate[1]));
+		so_data  (so, fui(vpt->translate[2]));
+		so_data  (so, fui(vpt->translate[3]));
+		so_data  (so, fui(vpt->scale[0]));
+		so_data  (so, fui(vpt->scale[1]));
+		so_data  (so, fui(vpt->scale[2]));
+		so_data  (so, fui(vpt->scale[3]));
+/*		so_method(so, nv30->screen->rankine, 0x1d78, 1);
+		so_data  (so, 1);
+*/	} else {
+		so_method(so, nv30->screen->rankine,
+			  NV34TCL_VIEWPORT_TRANSLATE_X, 8);
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(0.0));
+		/* Not entirely certain what this is yet.  The DDX uses this
+		 * value also as it fixes rendering when you pass
+		 * pre-transformed vertices to the GPU.  My best gusss is that
+		 * this bypasses some culling/clipping stage.  Might be worth
+		 * noting that points/lines are uneffected by whatever this
+		 * value fixes, only filled polygons are effected.
+		 */
+/*		so_method(so, nv30->screen->rankine, 0x1d78, 1);
+		so_data  (so, 0x110);
+*/	}
+	/* TODO/FIXME: never saw value 0x0110 in renouveau dumps, only 0x0001 */
+	so_method(so, nv30->screen->rankine, 0x1d78, 1);
+	so_data  (so, 1);
+
+	so_ref(so, &nv30->state.hw[NV30_STATE_VIEWPORT]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_viewport = {
+	.validate = nv30_state_viewport_validate,
+	.dirty = {
+		.pipe = NV30_NEW_VIEWPORT | NV30_NEW_RAST,
+		.hw = NV30_STATE_VIEWPORT
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_state_zsa.c b/src/gallium/drivers/nv30/nv30_state_zsa.c
new file mode 100644
index 00000000000..0940b7269b2
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_state_zsa.c
@@ -0,0 +1,17 @@
+#include "nv30_context.h"
+
+static boolean
+nv30_state_zsa_validate(struct nv30_context *nv30)
+{
+	so_ref(nv30->zsa->so,
+	       &nv30->state.hw[NV30_STATE_ZSA]);
+	return TRUE;
+}
+
+struct nv30_state_entry nv30_state_zsa = {
+	.validate = nv30_state_zsa_validate,
+	.dirty = {
+		.pipe = NV30_NEW_ZSA,
+		.hw = NV30_STATE_ZSA
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_surface.c b/src/gallium/drivers/nv30/nv30_surface.c
new file mode 100644
index 00000000000..d3376a73bf5
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_surface.c
@@ -0,0 +1,77 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "nv30_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+
+#include "util/u_tile.h"
+
+static void
+nv30_surface_copy(struct pipe_context *pipe, boolean do_flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nouveau_winsys *nvws = nv30->nvws;
+
+	if (do_flip) {
+		/*XXX: This dodgyness will do for now for correctness.  But,
+		 *     need to investigate whether the 2D engine is able to
+		 *     manage a flip (perhaps SIFM?), if not, use the 3D engine
+		 */
+		desty += height;
+		while (height--) {
+			nvws->surface_copy(nvws, dest, destx, desty--, src,
+					   srcx, srcy++, width, 1);
+		}
+	} else {
+		nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+				   width, height);
+	}
+}
+
+static void
+nv30_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nouveau_winsys *nvws = nv30->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+void
+nv30_init_surface_functions(struct nv30_context *nv30)
+{
+	nv30->pipe.surface_copy = nv30_surface_copy;
+	nv30->pipe.surface_fill = nv30_surface_fill;
+}
diff --git a/src/gallium/drivers/nv30/nv30_vbo.c b/src/gallium/drivers/nv30/nv30_vbo.c
new file mode 100644
index 00000000000..556f981d4a5
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_vbo.c
@@ -0,0 +1,556 @@
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv30_context.h"
+#include "nv30_state.h"
+
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_pushbuf.h"
+#include "nouveau/nouveau_util.h"
+
+#define FORCE_SWTNL 0
+
+static INLINE int
+nv30_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp)
+{
+	switch (pipe) {
+	case PIPE_FORMAT_R32_FLOAT:
+	case PIPE_FORMAT_R32G32_FLOAT:
+	case PIPE_FORMAT_R32G32B32_FLOAT:
+	case PIPE_FORMAT_R32G32B32A32_FLOAT:
+		*fmt = NV34TCL_VTXFMT_TYPE_FLOAT;
+		break;
+	case PIPE_FORMAT_R8_UNORM:
+	case PIPE_FORMAT_R8G8_UNORM:
+	case PIPE_FORMAT_R8G8B8_UNORM:
+	case PIPE_FORMAT_R8G8B8A8_UNORM:
+		*fmt = NV34TCL_VTXFMT_TYPE_UBYTE;
+		break;
+	case PIPE_FORMAT_R16_SSCALED:
+	case PIPE_FORMAT_R16G16_SSCALED:
+	case PIPE_FORMAT_R16G16B16_SSCALED:
+	case PIPE_FORMAT_R16G16B16A16_SSCALED:
+		*fmt = NV34TCL_VTXFMT_TYPE_USHORT;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe));
+		return 1;
+	}
+
+	switch (pipe) {
+	case PIPE_FORMAT_R8_UNORM:
+	case PIPE_FORMAT_R32_FLOAT:
+	case PIPE_FORMAT_R16_SSCALED:
+		*ncomp = 1;
+		break;
+	case PIPE_FORMAT_R8G8_UNORM:
+	case PIPE_FORMAT_R32G32_FLOAT:
+	case PIPE_FORMAT_R16G16_SSCALED:
+		*ncomp = 2;
+		break;
+	case PIPE_FORMAT_R8G8B8_UNORM:
+	case PIPE_FORMAT_R32G32B32_FLOAT:
+	case PIPE_FORMAT_R16G16B16_SSCALED:
+		*ncomp = 3;
+		break;
+	case PIPE_FORMAT_R8G8B8A8_UNORM:
+	case PIPE_FORMAT_R32G32B32A32_FLOAT:
+	case PIPE_FORMAT_R16G16B16A16_SSCALED:
+		*ncomp = 4;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe));
+		return 1;
+	}
+
+	return 0;
+}
+
+static boolean
+nv30_vbo_set_idxbuf(struct nv30_context *nv30, struct pipe_buffer *ib,
+		    unsigned ib_size)
+{
+	struct pipe_screen *pscreen = &nv30->screen->pipe;
+	unsigned type;
+
+	if (!ib) {
+		nv30->idxbuf = NULL;
+		nv30->idxbuf_format = 0xdeadbeef;
+		return FALSE;
+	}
+
+	if (!pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF) || ib_size == 1)
+		return FALSE;
+
+	switch (ib_size) {
+	case 2:
+		type = NV34TCL_IDXBUF_FORMAT_TYPE_U16;
+		break;
+	case 4:
+		type = NV34TCL_IDXBUF_FORMAT_TYPE_U32;
+		break;
+	default:
+		return FALSE;
+	}
+
+	if (ib != nv30->idxbuf ||
+	    type != nv30->idxbuf_format) {
+		nv30->dirty |= NV30_NEW_ARRAYS;
+		nv30->idxbuf = ib;
+		nv30->idxbuf_format = type;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv30_vbo_static_attrib(struct nv30_context *nv30, struct nouveau_stateobj *so,
+		       int attrib, struct pipe_vertex_element *ve,
+		       struct pipe_vertex_buffer *vb)
+{
+	struct pipe_winsys *ws = nv30->pipe.winsys;
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+	unsigned type, ncomp;
+	void *map;
+
+	if (nv30_vbo_format_to_hw(ve->src_format, &type, &ncomp))
+		return FALSE;
+
+	map  = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ);
+	map += vb->buffer_offset + ve->src_offset;
+
+	switch (type) {
+	case NV34TCL_VTXFMT_TYPE_FLOAT:
+	{
+		float *v = map;
+
+		switch (ncomp) {
+		case 4:
+			so_method(so, rankine, NV34TCL_VTX_ATTR_4F_X(attrib), 4);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			so_data  (so, fui(v[2]));
+			so_data  (so, fui(v[3]));
+			break;
+		case 3:
+			so_method(so, rankine, NV34TCL_VTX_ATTR_3F_X(attrib), 3);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			so_data  (so, fui(v[2]));
+			break;
+		case 2:
+			so_method(so, rankine, NV34TCL_VTX_ATTR_2F_X(attrib), 2);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			break;
+		case 1:
+			so_method(so, rankine, NV34TCL_VTX_ATTR_1F(attrib), 1);
+			so_data  (so, fui(v[0]));
+			break;
+		default:
+			ws->buffer_unmap(ws, vb->buffer);
+			return FALSE;
+		}
+	}
+		break;
+	default:
+		ws->buffer_unmap(ws, vb->buffer);
+		return FALSE;
+	}
+
+	ws->buffer_unmap(ws, vb->buffer);
+
+	return TRUE;
+}
+
+boolean
+nv30_draw_arrays(struct pipe_context *pipe,
+		 unsigned mode, unsigned start, unsigned count)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nouveau_channel *chan = nv30->nvws->channel;
+	unsigned restart = 0;
+
+	nv30_vbo_set_idxbuf(nv30, NULL, 0);
+	if (FORCE_SWTNL || !nv30_state_validate(nv30)) {
+		/*return nv30_draw_elements_swtnl(pipe, NULL, 0,
+						mode, start, count);*/
+		return FALSE;
+	}
+
+	while (count) {
+		unsigned vc, nr;
+
+		nv30_state_emit(nv30);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256,
+					mode, start, count, &restart);
+		if (!vc) {
+			FIRE_RING(NULL);
+			continue;
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		nr = (vc & 0xff);
+		if (nr) {
+			BEGIN_RING(rankine, NV34TCL_VB_VERTEX_BATCH, 1);
+			OUT_RING  (((nr - 1) << 24) | start);
+			start += nr;
+		}
+
+		nr = vc >> 8;
+		while (nr) {
+			unsigned push = nr > 2047 ? 2047 : nr;
+
+			nr -= push;
+
+			BEGIN_RING_NI(rankine, NV34TCL_VB_VERTEX_BATCH, push);
+			while (push--) {
+				OUT_RING(((0x100 - 1) << 24) | start);
+				start += 0x100;
+			}
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		count -= vc;
+		start = restart;
+	}
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+static INLINE void
+nv30_draw_elements_u08(struct nv30_context *nv30, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv30->nvws->channel;
+
+	while (count) {
+		uint8_t *elts = (uint8_t *)ib + start;
+		unsigned vc, push, restart = 0;
+
+		nv30_state_emit(nv30);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		if (vc & 1) {
+			BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1);
+			OUT_RING  (elts[0]);
+			elts++; vc--;
+		}
+
+		while (vc) {
+			unsigned i;
+
+			push = MIN2(vc, 2047 * 2);
+
+			BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1);
+			for (i = 0; i < push; i+=2)
+				OUT_RING((elts[i+1] << 16) | elts[i]);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static INLINE void
+nv30_draw_elements_u16(struct nv30_context *nv30, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv30->nvws->channel;
+
+	while (count) {
+		uint16_t *elts = (uint16_t *)ib + start;
+		unsigned vc, push, restart = 0;
+
+		nv30_state_emit(nv30);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		if (vc & 1) {
+			BEGIN_RING(rankine, NV34TCL_VB_ELEMENT_U32, 1);
+			OUT_RING  (elts[0]);
+			elts++; vc--;
+		}
+
+		while (vc) {
+			unsigned i;
+
+			push = MIN2(vc, 2047 * 2);
+
+			BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U16, push >> 1);
+			for (i = 0; i < push; i+=2)
+				OUT_RING((elts[i+1] << 16) | elts[i]);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static INLINE void
+nv30_draw_elements_u32(struct nv30_context *nv30, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv30->nvws->channel;
+
+	while (count) {
+		uint32_t *elts = (uint32_t *)ib + start;
+		unsigned vc, push, restart = 0;
+
+		nv30_state_emit(nv30);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 5, 1,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		while (vc) {
+			push = MIN2(vc, 2047);
+
+			BEGIN_RING_NI(rankine, NV34TCL_VB_ELEMENT_U32, push);
+			OUT_RINGp    (elts, push);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static boolean
+nv30_draw_elements_inline(struct pipe_context *pipe,
+			  struct pipe_buffer *ib, unsigned ib_size,
+			  unsigned mode, unsigned start, unsigned count)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+	void *map;
+
+	map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ);
+	if (!ib) {
+		NOUVEAU_ERR("failed mapping ib\n");
+		return FALSE;
+	}
+
+	switch (ib_size) {
+	case 1:
+		nv30_draw_elements_u08(nv30, map, mode, start, count);
+		break;
+	case 2:
+		nv30_draw_elements_u16(nv30, map, mode, start, count);
+		break;
+	case 4:
+		nv30_draw_elements_u32(nv30, map, mode, start, count);
+		break;
+	default:
+		NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size);
+		break;
+	}
+
+	ws->buffer_unmap(ws, ib);
+	return TRUE;
+}
+
+static boolean
+nv30_draw_elements_vbo(struct pipe_context *pipe,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	struct nouveau_channel *chan = nv30->nvws->channel;
+	unsigned restart = 0;
+
+	while (count) {
+		unsigned nr, vc;
+
+		nv30_state_emit(nv30);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256,
+					mode, start, count, &restart);
+		if (!vc) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		nr = (vc & 0xff);
+		if (nr) {
+			BEGIN_RING(rankine, NV34TCL_VB_INDEX_BATCH, 1);
+			OUT_RING  (((nr - 1) << 24) | start);
+			start += nr;
+		}
+
+		nr = vc >> 8;
+		while (nr) {
+			unsigned push = nr > 2047 ? 2047 : nr;
+
+			nr -= push;
+
+			BEGIN_RING_NI(rankine, NV34TCL_VB_INDEX_BATCH, push);
+			while (push--) {
+				OUT_RING(((0x100 - 1) << 24) | start);
+				start += 0x100;
+			}
+		}
+
+		BEGIN_RING(rankine, NV34TCL_VERTEX_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		count -= vc;
+		start = restart;
+	}
+
+	return TRUE;
+}
+
+boolean
+nv30_draw_elements(struct pipe_context *pipe,
+		   struct pipe_buffer *indexBuffer, unsigned indexSize,
+		   unsigned mode, unsigned start, unsigned count)
+{
+	struct nv30_context *nv30 = nv30_context(pipe);
+	boolean idxbuf;
+
+	idxbuf = nv30_vbo_set_idxbuf(nv30, indexBuffer, indexSize);
+	if (FORCE_SWTNL || !nv30_state_validate(nv30)) {
+		/*return nv30_draw_elements_swtnl(pipe, NULL, 0,
+						mode, start, count);*/
+		return FALSE;	
+	}
+
+	if (idxbuf) {
+		nv30_draw_elements_vbo(pipe, mode, start, count);
+	} else {
+		nv30_draw_elements_inline(pipe, indexBuffer, indexSize,
+					  mode, start, count);
+	}
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+static boolean
+nv30_vbo_validate(struct nv30_context *nv30)
+{
+	struct nouveau_stateobj *vtxbuf, *vtxfmt, *sattr = NULL;
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+	struct pipe_buffer *ib = nv30->idxbuf;
+	unsigned ib_format = nv30->idxbuf_format;
+	unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+	int hw;
+
+	if (nv30->edgeflags) {
+		/*nv30->fallback_swtnl |= NV30_NEW_ARRAYS;*/
+		return FALSE;
+	}
+
+	vtxbuf = so_new(20, 18);
+	so_method(vtxbuf, rankine, NV34TCL_VTXBUF_ADDRESS(0), nv30->vtxelt_nr);
+	vtxfmt = so_new(17, 0);
+	so_method(vtxfmt, rankine, NV34TCL_VTXFMT(0), nv30->vtxelt_nr);
+
+	for (hw = 0; hw < nv30->vtxelt_nr; hw++) {
+		struct pipe_vertex_element *ve;
+		struct pipe_vertex_buffer *vb;
+		unsigned type, ncomp;
+
+		ve = &nv30->vtxelt[hw];
+		vb = &nv30->vtxbuf[ve->vertex_buffer_index];
+
+		if (!vb->pitch) {
+			if (!sattr)
+				sattr = so_new(16 * 5, 0);
+
+			if (nv30_vbo_static_attrib(nv30, sattr, hw, ve, vb)) {
+				so_data(vtxbuf, 0);
+				so_data(vtxfmt, NV34TCL_VTXFMT_TYPE_FLOAT);
+				continue;
+			}
+		}
+
+		if (nv30_vbo_format_to_hw(ve->src_format, &type, &ncomp)) {
+			/*nv30->fallback_swtnl |= NV30_NEW_ARRAYS;*/
+			so_ref(NULL, &vtxbuf);
+			so_ref(NULL, &vtxfmt);
+			return FALSE;
+		}
+
+		so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset,
+			 vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR,
+			 0, NV34TCL_VTXBUF_ADDRESS_DMA1);
+		so_data (vtxfmt, ((vb->pitch << NV34TCL_VTXFMT_STRIDE_SHIFT) |
+				  (ncomp << NV34TCL_VTXFMT_SIZE_SHIFT) | type));
+	}
+
+	if (ib) {
+		so_method(vtxbuf, rankine, NV34TCL_IDXBUF_ADDRESS, 2);
+		so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0);
+		so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR,
+			  0, NV34TCL_IDXBUF_FORMAT_DMA1);
+	}
+
+	so_method(vtxbuf, rankine, 0x1710, 1);
+	so_data  (vtxbuf, 0);
+
+	so_ref(vtxbuf, &nv30->state.hw[NV30_STATE_VTXBUF]);
+	nv30->state.dirty |= (1ULL << NV30_STATE_VTXBUF);
+	so_ref(vtxfmt, &nv30->state.hw[NV30_STATE_VTXFMT]);
+	nv30->state.dirty |= (1ULL << NV30_STATE_VTXFMT);
+	so_ref(sattr, &nv30->state.hw[NV30_STATE_VTXATTR]);
+	nv30->state.dirty |= (1ULL << NV30_STATE_VTXATTR);
+	return FALSE;
+}
+
+struct nv30_state_entry nv30_state_vbo = {
+	.validate = nv30_vbo_validate,
+	.dirty = {
+		.pipe = NV30_NEW_ARRAYS,
+		.hw = 0,
+	}
+};
diff --git a/src/gallium/drivers/nv30/nv30_vertprog.c b/src/gallium/drivers/nv30/nv30_vertprog.c
new file mode 100644
index 00000000000..72824559e8b
--- /dev/null
+++ b/src/gallium/drivers/nv30/nv30_vertprog.c
@@ -0,0 +1,838 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_dump.h"
+
+#include "nv30_context.h"
+#include "nv30_state.h"
+
+/* TODO (at least...):
+ *  1. Indexed consts  + ARL
+ *  2. Arb. swz/negation
+ *  3. NV_vp11, NV_vp2, NV_vp3 features
+ *       - extra arith opcodes
+ *       - branching
+ *       - texture sampling
+ *       - indexed attribs
+ *       - indexed results
+ *  4. bugs
+ */
+
+#define SWZ_X 0
+#define SWZ_Y 1
+#define SWZ_Z 2
+#define SWZ_W 3
+#define MASK_X 8
+#define MASK_Y 4
+#define MASK_Z 2
+#define MASK_W 1
+#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W)
+#define DEF_SCALE 0
+#define DEF_CTEST 0
+#include "nv30_shader.h"
+
+#define swz(s,x,y,z,w) nv30_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w)
+#define neg(s) nv30_sr_neg((s))
+#define abs(s) nv30_sr_abs((s))
+
+struct nv30_vpc {
+	struct nv30_vertex_program *vp;
+
+	struct nv30_vertex_program_exec *vpi;
+
+	unsigned output_map[PIPE_MAX_SHADER_OUTPUTS];
+
+	int high_temp;
+	int temp_temp_count;
+
+	struct nv30_sreg *imm;
+	unsigned nr_imm;
+};
+
+static struct nv30_sreg
+temp(struct nv30_vpc *vpc)
+{
+	int idx;
+
+	idx  = vpc->temp_temp_count++;
+	idx += vpc->high_temp + 1;
+	return nv30_sr(NV30SR_TEMP, idx);
+}
+
+static struct nv30_sreg
+constant(struct nv30_vpc *vpc, int pipe, float x, float y, float z, float w)
+{
+	struct nv30_vertex_program *vp = vpc->vp;
+	struct nv30_vertex_program_data *vpd;
+	int idx;
+
+	if (pipe >= 0) {
+		for (idx = 0; idx < vp->nr_consts; idx++) {
+			if (vp->consts[idx].index == pipe)
+				return nv30_sr(NV30SR_CONST, idx);
+		}
+	}
+
+	idx = vp->nr_consts++;
+	vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts);
+	vpd = &vp->consts[idx];
+
+	vpd->index = pipe;
+	vpd->value[0] = x;
+	vpd->value[1] = y;
+	vpd->value[2] = z;
+	vpd->value[3] = w;
+	return nv30_sr(NV30SR_CONST, idx);
+}
+
+#define arith(cc,s,o,d,m,s0,s1,s2) \
+	nv30_vp_arith((cc), (s), NV30_VP_INST_##o, (d), (m), (s0), (s1), (s2))
+
+static void
+emit_src(struct nv30_vpc *vpc, uint32_t *hw, int pos, struct nv30_sreg src)
+{
+	struct nv30_vertex_program *vp = vpc->vp;
+	uint32_t sr = 0;
+
+	switch (src.type) {
+	case NV30SR_TEMP:
+		sr |= (NV30_VP_SRC_REG_TYPE_TEMP << NV30_VP_SRC_REG_TYPE_SHIFT);
+		sr |= (src.index << NV30_VP_SRC_TEMP_SRC_SHIFT);
+		break;
+	case NV30SR_INPUT:
+		sr |= (NV30_VP_SRC_REG_TYPE_INPUT <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		vp->ir |= (1 << src.index);
+		hw[1] |= (src.index << NV30_VP_INST_INPUT_SRC_SHIFT);
+		break;
+	case NV30SR_CONST:
+		sr |= (NV30_VP_SRC_REG_TYPE_CONST <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		assert(vpc->vpi->const_index == -1 ||
+		       vpc->vpi->const_index == src.index);
+		vpc->vpi->const_index = src.index;
+		break;
+	case NV30SR_NONE:
+		sr |= (NV30_VP_SRC_REG_TYPE_INPUT <<
+		       NV30_VP_SRC_REG_TYPE_SHIFT);
+		break;
+	default:
+		assert(0);
+	}
+
+	if (src.negate)
+		sr |= NV30_VP_SRC_NEGATE;
+
+	if (src.abs)
+		hw[0] |= (1 << (21 + pos));
+
+	sr |= ((src.swz[0] << NV30_VP_SRC_SWZ_X_SHIFT) |
+	       (src.swz[1] << NV30_VP_SRC_SWZ_Y_SHIFT) |
+	       (src.swz[2] << NV30_VP_SRC_SWZ_Z_SHIFT) |
+	       (src.swz[3] << NV30_VP_SRC_SWZ_W_SHIFT));
+
+/*
+ * |VVV|
+ * d�.�b
+ *  \u/
+ *
+ */
+
+	switch (pos) {
+	case 0:
+		hw[1] |= ((sr & NV30_VP_SRC0_HIGH_MASK) >>
+			  NV30_VP_SRC0_HIGH_SHIFT) << NV30_VP_INST_SRC0H_SHIFT;
+		hw[2] |= (sr & NV30_VP_SRC0_LOW_MASK) <<
+			  NV30_VP_INST_SRC0L_SHIFT;
+		break;
+	case 1:
+		hw[2] |= sr << NV30_VP_INST_SRC1_SHIFT;
+		break;
+	case 2:
+		hw[2] |= ((sr & NV30_VP_SRC2_HIGH_MASK) >>
+			  NV30_VP_SRC2_HIGH_SHIFT) << NV30_VP_INST_SRC2H_SHIFT;
+		hw[3] |= (sr & NV30_VP_SRC2_LOW_MASK) <<
+			  NV30_VP_INST_SRC2L_SHIFT;
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+emit_dst(struct nv30_vpc *vpc, uint32_t *hw, int slot, struct nv30_sreg dst)
+{
+	struct nv30_vertex_program *vp = vpc->vp;
+
+	switch (dst.type) {
+	case NV30SR_TEMP:
+		hw[0] |= (dst.index << NV30_VP_INST_DEST_TEMP_ID_SHIFT);
+		break;
+	case NV30SR_OUTPUT:
+		switch (dst.index) {
+		case NV30_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break;
+		case NV30_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break;
+		case NV30_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break;
+		case NV30_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break;
+		case NV30_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break;
+		case NV30_VP_INST_DEST_PSZ  : vp->or |= (1 << 5); break;
+		case NV30_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break;
+		case NV30_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break;
+		case NV30_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break;
+		case NV30_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break;
+		case NV30_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break;
+		case NV30_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break;
+		case NV30_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break;
+		case NV30_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break;
+		default:
+			break;
+		}
+
+		hw[3] |= (dst.index << NV30_VP_INST_DEST_SHIFT);
+		hw[0] |= NV30_VP_INST_VEC_DEST_TEMP_MASK | (1<<20);
+
+		/*XXX: no way this is entirely correct, someone needs to
+		 *     figure out what exactly it is.
+		 */
+		hw[3] |= 0x800;
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+nv30_vp_arith(struct nv30_vpc *vpc, int slot, int op,
+	      struct nv30_sreg dst, int mask,
+	      struct nv30_sreg s0, struct nv30_sreg s1,
+	      struct nv30_sreg s2)
+{
+	struct nv30_vertex_program *vp = vpc->vp;
+	uint32_t *hw;
+
+	vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi));
+	vpc->vpi = &vp->insns[vp->nr_insns - 1];
+	memset(vpc->vpi, 0, sizeof(*vpc->vpi));
+	vpc->vpi->const_index = -1;
+
+	hw = vpc->vpi->data;
+
+	hw[0] |= (NV30_VP_INST_COND_TR << NV30_VP_INST_COND_SHIFT);
+	hw[0] |= ((0 << NV30_VP_INST_COND_SWZ_X_SHIFT) |
+		  (1 << NV30_VP_INST_COND_SWZ_Y_SHIFT) |
+		  (2 << NV30_VP_INST_COND_SWZ_Z_SHIFT) |
+		  (3 << NV30_VP_INST_COND_SWZ_W_SHIFT));
+
+	hw[1] |= (op << NV30_VP_INST_VEC_OPCODE_SHIFT);
+//	hw[3] |= NV30_VP_INST_SCA_DEST_TEMP_MASK;
+//	hw[3] |= (mask << NV30_VP_INST_VEC_WRITEMASK_SHIFT);
+
+	if (dst.type == NV30SR_OUTPUT) {
+		if (slot)
+			hw[3] |= (mask << NV30_VP_INST_SDEST_WRITEMASK_SHIFT);
+		else
+			hw[3] |= (mask << NV30_VP_INST_VDEST_WRITEMASK_SHIFT);
+	} else {
+		if (slot)
+			hw[3] |= (mask << NV30_VP_INST_STEMP_WRITEMASK_SHIFT);
+		else
+			hw[3] |= (mask << NV30_VP_INST_VTEMP_WRITEMASK_SHIFT);
+	}
+
+	emit_dst(vpc, hw, slot, dst);
+	emit_src(vpc, hw, 0, s0);
+	emit_src(vpc, hw, 1, s1);
+	emit_src(vpc, hw, 2, s2);
+}
+
+static INLINE struct nv30_sreg
+tgsi_src(struct nv30_vpc *vpc, const struct tgsi_full_src_register *fsrc) {
+	struct nv30_sreg src;
+
+	switch (fsrc->SrcRegister.File) {
+	case TGSI_FILE_INPUT:
+		src = nv30_sr(NV30SR_INPUT, fsrc->SrcRegister.Index);
+		break;
+	case TGSI_FILE_CONSTANT:
+		src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0);
+		break;
+	case TGSI_FILE_IMMEDIATE:
+		src = vpc->imm[fsrc->SrcRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		if (vpc->high_temp < fsrc->SrcRegister.Index)
+			vpc->high_temp = fsrc->SrcRegister.Index;
+		src = nv30_sr(NV30SR_TEMP, fsrc->SrcRegister.Index);
+		break;
+	default:
+		NOUVEAU_ERR("bad src file\n");
+		break;
+	}
+
+	src.abs = fsrc->SrcRegisterExtMod.Absolute;
+	src.negate = fsrc->SrcRegister.Negate;
+	src.swz[0] = fsrc->SrcRegister.SwizzleX;
+	src.swz[1] = fsrc->SrcRegister.SwizzleY;
+	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
+	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	return src;
+}
+
+static INLINE struct nv30_sreg
+tgsi_dst(struct nv30_vpc *vpc, const struct tgsi_full_dst_register *fdst) {
+	struct nv30_sreg dst;
+
+	switch (fdst->DstRegister.File) {
+	case TGSI_FILE_OUTPUT:
+		dst = nv30_sr(NV30SR_OUTPUT,
+			      vpc->output_map[fdst->DstRegister.Index]);
+
+		break;
+	case TGSI_FILE_TEMPORARY:
+		dst = nv30_sr(NV30SR_TEMP, fdst->DstRegister.Index);
+		if (vpc->high_temp < dst.index)
+			vpc->high_temp = dst.index;
+		break;
+	default:
+		NOUVEAU_ERR("bad dst file\n");
+		break;
+	}
+
+	return dst;
+}
+
+static INLINE int
+tgsi_mask(uint tgsi)
+{
+	int mask = 0;
+
+	if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X;
+	if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y;
+	if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z;
+	if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W;
+	return mask;
+}
+
+static boolean
+nv30_vertprog_parse_instruction(struct nv30_vpc *vpc,
+				const struct tgsi_full_instruction *finst)
+{
+	struct nv30_sreg src[3], dst, tmp;
+	struct nv30_sreg none = nv30_sr(NV30SR_NONE, 0);
+	int mask;
+	int ai = -1, ci = -1;
+	int i;
+
+	if (finst->Instruction.Opcode == TGSI_OPCODE_END)
+		return TRUE;
+
+	vpc->temp_temp_count = 0;
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+			src[i] = tgsi_src(vpc, fsrc);
+		}
+	}
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
+				ai = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		/*XXX: index comparison is broken now that consts come from
+		 *     two different register files.
+		 */
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_IMMEDIATE:
+			if (ci == -1 || ci == fsrc->SrcRegister.Index) {
+				ci = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_TEMPORARY:
+			/* handled above */
+			break;
+		default:
+			NOUVEAU_ERR("bad src file\n");
+			return FALSE;
+		}
+	}
+
+	dst  = tgsi_dst(vpc, &finst->FullDstRegisters[0]);
+	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+
+	switch (finst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none);
+		break;
+	case TGSI_OPCODE_ADD:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]);
+		break;
+	case TGSI_OPCODE_ARL:
+		arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_DP3:
+		arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DP4:
+		arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DPH:
+		arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DST:
+		arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_EX2:
+		arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_EXP:
+		arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_FLR:
+		arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FRC:
+		arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_LG2:
+		arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LIT:
+		arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LOG:
+		arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_MAD:
+		arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAX:
+		arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MIN:
+		arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MOV:
+		arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_MUL:
+		arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_POW:
+		tmp = temp(vpc);
+		arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none,
+		      swz(src[0], X, X, X, X));
+		arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X),
+		      swz(src[1], X, X, X, X), none);
+		arith(vpc, 1, OP_EX2, dst, mask, none, none,
+		      swz(tmp, X, X, X, X));
+		break;
+	case TGSI_OPCODE_RCP:
+		arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_RET:
+		break;
+	case TGSI_OPCODE_RSQ:
+		arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_SGE:
+		arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SGT:
+		arith(vpc, 0, OP_SGT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SLT:
+		arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SUB:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1]));
+		break;
+	case TGSI_OPCODE_XPD:
+		tmp = temp(vpc);
+		arith(vpc, 0, OP_MUL, tmp, mask,
+		      swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none);
+		arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W),
+		      swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y),
+		      neg(tmp));
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv30_vertprog_parse_decl_output(struct nv30_vpc *vpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	int hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = NV30_VP_INST_DEST_POS;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV30_VP_INST_DEST_COL0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV30_VP_INST_DEST_COL1;
+		} else {
+			NOUVEAU_ERR("bad colour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_BCOLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV30_VP_INST_DEST_BFC0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV30_VP_INST_DEST_BFC1;
+		} else {
+			NOUVEAU_ERR("bad bcolour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_FOG:
+		hw = NV30_VP_INST_DEST_FOGC;
+		break;
+	case TGSI_SEMANTIC_PSIZE:
+		hw = NV30_VP_INST_DEST_PSZ;
+		break;
+	case TGSI_SEMANTIC_GENERIC:
+		if (fdec->Semantic.SemanticIndex <= 7) {
+			hw = NV30_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex);
+		} else {
+			NOUVEAU_ERR("bad generic semantic index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad output semantic\n");
+		return FALSE;
+	}
+
+	vpc->output_map[fdec->DeclarationRange.First] = hw;
+	return TRUE;
+}
+
+static boolean
+nv30_vertprog_prepare(struct nv30_vpc *vpc)
+{
+	struct tgsi_parse_context p;
+	int nr_imm = 0;
+
+	tgsi_parse_init(&p, vpc->vp->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch(tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+			nr_imm++;
+			break;
+		default:
+			break;
+		}
+	}
+	tgsi_parse_free(&p);
+
+	if (nr_imm) {
+		vpc->imm = CALLOC(nr_imm, sizeof(struct nv30_sreg));
+		assert(vpc->imm);
+	}
+
+	return TRUE;
+}
+
+static void
+nv30_vertprog_translate(struct nv30_context *nv30,
+			struct nv30_vertex_program *vp)
+{
+	struct tgsi_parse_context parse;
+	struct nv30_vpc *vpc = NULL;
+
+	tgsi_dump(vp->pipe.tokens,0);
+
+	vpc = CALLOC(1, sizeof(struct nv30_vpc));
+	if (!vpc)
+		return;
+	vpc->vp = vp;
+	vpc->high_temp = -1;
+
+	if (!nv30_vertprog_prepare(vpc)) {
+		FREE(vpc);
+		return;
+	}
+
+	tgsi_parse_init(&parse, vp->pipe.tokens);
+
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		tgsi_parse_token(&parse);
+
+		switch (parse.FullToken.Token.Type) {
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *fdec;
+			fdec = &parse.FullToken.FullDeclaration;
+			switch (fdec->Declaration.File) {
+			case TGSI_FILE_OUTPUT:
+				if (!nv30_vertprog_parse_decl_output(vpc, fdec))
+					goto out_err;
+				break;
+			default:
+				break;
+			}
+		}
+			break;
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			const struct tgsi_full_immediate *imm;
+
+			imm = &parse.FullToken.FullImmediate;
+			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
+//			assert(imm->Immediate.Size == 4);
+			vpc->imm[vpc->nr_imm++] =
+				constant(vpc, -1,
+					 imm->u.ImmediateFloat32[0].Float,
+					 imm->u.ImmediateFloat32[1].Float,
+					 imm->u.ImmediateFloat32[2].Float,
+					 imm->u.ImmediateFloat32[3].Float);
+		}
+			break;
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+			finst = &parse.FullToken.FullInstruction;
+			if (!nv30_vertprog_parse_instruction(vpc, finst))
+				goto out_err;
+		}
+			break;
+		default:
+			break;
+		}
+	}
+
+	vp->insns[vp->nr_insns - 1].data[3] |= NV30_VP_INST_LAST;
+	vp->translated = TRUE;
+out_err:
+	tgsi_parse_free(&parse);
+	FREE(vpc);
+}
+
+static boolean
+nv30_vertprog_validate(struct nv30_context *nv30)
+{ 
+	struct nouveau_winsys *nvws = nv30->nvws;
+	struct pipe_winsys *ws = nv30->pipe.winsys;
+	struct nouveau_grobj *rankine = nv30->screen->rankine;
+	struct nv30_vertex_program *vp;
+	struct pipe_buffer *constbuf;
+	boolean upload_code = FALSE, upload_data = FALSE;
+	int i;
+
+	vp = nv30->vertprog;
+	constbuf = nv30->constbuf[PIPE_SHADER_VERTEX];
+
+	/* Translate TGSI shader into hw bytecode */
+	if (!vp->translated) {
+		nv30_vertprog_translate(nv30, vp);
+		if (!vp->translated)
+			return FALSE;
+	}
+
+	/* Allocate hw vtxprog exec slots */
+	if (!vp->exec) {
+		struct nouveau_resource *heap = nv30->screen->vp_exec_heap;
+		struct nouveau_stateobj *so;
+		uint vplen = vp->nr_insns;
+
+		if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) {
+			while (heap->next && heap->size < vplen) {
+				struct nv30_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->exec);
+			}
+
+			if (nvws->res_alloc(heap, vplen, vp, &vp->exec))
+				assert(0);
+		}
+
+		so = so_new(2, 0);
+		so_method(so, rankine, NV34TCL_VP_START_FROM_ID, 1);
+		so_data  (so, vp->exec->start);
+		so_ref(so, &vp->so);
+
+		upload_code = TRUE;
+	}
+
+	/* Allocate hw vtxprog const slots */
+	if (vp->nr_consts && !vp->data) {
+		struct nouveau_resource *heap = nv30->screen->vp_data_heap;
+
+		if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) {
+			while (heap->next && heap->size < vp->nr_consts) {
+				struct nv30_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->data);
+			}
+
+			if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data))
+				assert(0);
+		}
+
+		/*XXX: handle this some day */
+		assert(vp->data->start >= vp->data_start_min);
+
+		upload_data = TRUE;
+		if (vp->data_start != vp->data->start)
+			upload_code = TRUE;
+	}
+
+	/* If exec or data segments moved we need to patch the program to
+	 * fixup offsets and register IDs.
+	 */
+	if (vp->exec_start != vp->exec->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv30_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->has_branch_offset) {
+				assert(0);
+			}
+		}
+
+		vp->exec_start = vp->exec->start;
+	}
+
+	if (vp->nr_consts && vp->data_start != vp->data->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv30_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->const_index >= 0) {
+				vpi->data[1] &= ~NV30_VP_INST_CONST_SRC_MASK;
+				vpi->data[1] |=
+					(vpi->const_index + vp->data->start) <<
+					NV30_VP_INST_CONST_SRC_SHIFT;
+
+			}
+		}
+
+		vp->data_start = vp->data->start;
+	}
+
+	/* Update + Upload constant values */
+	if (vp->nr_consts) {
+		float *map = NULL;
+
+		if (constbuf) {
+			map = ws->buffer_map(ws, constbuf,
+					     PIPE_BUFFER_USAGE_CPU_READ);
+		}
+
+		for (i = 0; i < vp->nr_consts; i++) {
+			struct nv30_vertex_program_data *vpd = &vp->consts[i];
+
+			if (vpd->index >= 0) {
+				if (!upload_data &&
+				    !memcmp(vpd->value, &map[vpd->index * 4],
+					    4 * sizeof(float)))
+					continue;
+				memcpy(vpd->value, &map[vpd->index * 4],
+				       4 * sizeof(float));
+			}
+
+			BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_CONST_ID, 5);
+			OUT_RING  (i + vp->data->start);
+			OUT_RINGp ((uint32_t *)vpd->value, 4);
+		}
+
+		if (constbuf) {
+			ws->buffer_unmap(ws, constbuf);
+		}
+	}
+
+	/* Upload vtxprog */
+	if (upload_code) {
+#if 0
+		for (i = 0; i < vp->nr_insns; i++) {
+			NOUVEAU_MSG("VP inst %d: 0x%08x 0x%08x 0x%08x 0x%08x\n",
+				i, vp->insns[i].data[0], vp->insns[i].data[1],
+				vp->insns[i].data[2], vp->insns[i].data[3]);
+		}
+#endif
+		BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_FROM_ID, 1);
+		OUT_RING  (vp->exec->start);
+		for (i = 0; i < vp->nr_insns; i++) {
+			BEGIN_RING(rankine, NV34TCL_VP_UPLOAD_INST(0), 4);
+			OUT_RINGp (vp->insns[i].data, 4);
+		}
+	}
+
+	if (vp->so != nv30->state.hw[NV30_STATE_VERTPROG]) {
+		so_ref(vp->so, &nv30->state.hw[NV30_STATE_VERTPROG]);
+		return TRUE;
+	}
+
+	return FALSE;
+}
+
+void
+nv30_vertprog_destroy(struct nv30_context *nv30, struct nv30_vertex_program *vp)
+{
+	struct nouveau_winsys *nvws = nv30->screen->nvws;
+
+	vp->translated = FALSE;
+
+	if (vp->nr_insns) {
+		FREE(vp->insns);
+		vp->insns = NULL;
+		vp->nr_insns = 0;
+	}
+
+	if (vp->nr_consts) {
+		FREE(vp->consts);
+		vp->consts = NULL;
+		vp->nr_consts = 0;
+	}
+
+	nvws->res_free(&vp->exec);
+	vp->exec_start = 0;
+	nvws->res_free(&vp->data);
+	vp->data_start = 0;
+	vp->data_start_min = 0;
+
+	vp->ir = vp->or = 0;
+	so_ref(NULL, &vp->so);
+}
+
+struct nv30_state_entry nv30_state_vertprog = {
+	.validate = nv30_vertprog_validate,
+	.dirty = {
+		.pipe = NV30_NEW_VERTPROG /*| NV30_NEW_UCP*/,
+		.hw = NV30_STATE_VERTPROG,
+	}
+};
diff --git a/src/gallium/drivers/nv40/Makefile b/src/gallium/drivers/nv40/Makefile
new file mode 100644
index 00000000000..9c8eadf7e44
--- /dev/null
+++ b/src/gallium/drivers/nv40/Makefile
@@ -0,0 +1,37 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv40
+
+DRIVER_SOURCES = \
+	nv40_clear.c \
+	nv40_context.c \
+	nv40_draw.c \
+	nv40_fragprog.c \
+	nv40_fragtex.c \
+	nv40_miptree.c \
+	nv40_query.c \
+	nv40_screen.c \
+	nv40_state.c \
+	nv40_state_blend.c \
+	nv40_state_emit.c \
+	nv40_state_fb.c \
+	nv40_state_rasterizer.c \
+	nv40_state_scissor.c \
+	nv40_state_stipple.c \
+	nv40_state_viewport.c \
+	nv40_state_zsa.c \
+	nv40_surface.c \
+	nv40_vbo.c \
+	nv40_vertprog.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv40/nv40_clear.c b/src/gallium/drivers/nv40/nv40_clear.c
new file mode 100644
index 00000000000..59efd620e32
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_clear.c
@@ -0,0 +1,13 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv40_context.h"
+
+void
+nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, clearValue);
+	ps->status = PIPE_SURFACE_STATUS_CLEAR;
+}
diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c
new file mode 100644
index 00000000000..cc63dd734bc
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_context.c
@@ -0,0 +1,72 @@
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv40_context.h"
+#include "nv40_screen.h"
+
+static void
+nv40_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	
+	if (flags & PIPE_FLUSH_TEXTURE_CACHE) {
+		BEGIN_RING(curie, 0x1fd8, 1);
+		OUT_RING  (2);
+		BEGIN_RING(curie, 0x1fd8, 1);
+		OUT_RING  (1);
+	}
+
+	FIRE_RING(fence);
+}
+
+static void
+nv40_destroy(struct pipe_context *pipe)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	if (nv40->draw)
+		draw_destroy(nv40->draw);
+	FREE(nv40);
+}
+
+struct pipe_context *
+nv40_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct nv40_screen *screen = nv40_screen(pscreen);
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv40_context *nv40;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nv40 = CALLOC(1, sizeof(struct nv40_context));
+	if (!nv40)
+		return NULL;
+	nv40->screen = screen;
+	nv40->pctx_id = pctx_id;
+
+	nv40->nvws = nvws;
+
+	nv40->pipe.winsys = ws;
+	nv40->pipe.screen = pscreen;
+	nv40->pipe.destroy = nv40_destroy;
+	nv40->pipe.draw_arrays = nv40_draw_arrays;
+	nv40->pipe.draw_elements = nv40_draw_elements;
+	nv40->pipe.clear = nv40_clear;
+	nv40->pipe.flush = nv40_flush;
+
+	nv40_init_query_functions(nv40);
+	nv40_init_surface_functions(nv40);
+	nv40_init_state_functions(nv40);
+
+	/* Create, configure, and install fallback swtnl path */
+	nv40->draw = draw_create();
+	draw_wide_point_threshold(nv40->draw, 9999999.0);
+	draw_wide_line_threshold(nv40->draw, 9999999.0);
+	draw_enable_line_stipple(nv40->draw, FALSE);
+	draw_enable_point_sprites(nv40->draw, FALSE);
+	draw_set_rasterize_stage(nv40->draw, nv40_draw_render_stage(nv40));
+
+	return &nv40->pipe;
+}
+	
diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h
new file mode 100644
index 00000000000..adcfbdd85a8
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_context.h
@@ -0,0 +1,233 @@
+#ifndef __NV40_CONTEXT_H__
+#define __NV40_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv40_screen *ctx = nv40->screen
+#include "nouveau/nouveau_push.h"
+#include "nouveau/nouveau_stateobj.h"
+
+#include "nv40_state.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+enum nv40_state_index {
+	NV40_STATE_FB = 0,
+	NV40_STATE_VIEWPORT = 1,
+	NV40_STATE_BLEND = 2,
+	NV40_STATE_RAST = 3,
+	NV40_STATE_ZSA = 4,
+	NV40_STATE_BCOL = 5,
+	NV40_STATE_CLIP = 6,
+	NV40_STATE_SCISSOR = 7,
+	NV40_STATE_STIPPLE = 8,
+	NV40_STATE_FRAGPROG = 9,
+	NV40_STATE_VERTPROG = 10,
+	NV40_STATE_FRAGTEX0 = 11,
+	NV40_STATE_FRAGTEX1 = 12,
+	NV40_STATE_FRAGTEX2 = 13,
+	NV40_STATE_FRAGTEX3 = 14,
+	NV40_STATE_FRAGTEX4 = 15,
+	NV40_STATE_FRAGTEX5 = 16,
+	NV40_STATE_FRAGTEX6 = 17,
+	NV40_STATE_FRAGTEX7 = 18,
+	NV40_STATE_FRAGTEX8 = 19,
+	NV40_STATE_FRAGTEX9 = 20,
+	NV40_STATE_FRAGTEX10 = 21,
+	NV40_STATE_FRAGTEX11 = 22,
+	NV40_STATE_FRAGTEX12 = 23,
+	NV40_STATE_FRAGTEX13 = 24,
+	NV40_STATE_FRAGTEX14 = 25,
+	NV40_STATE_FRAGTEX15 = 26,
+	NV40_STATE_VERTTEX0 = 27,
+	NV40_STATE_VERTTEX1 = 28,
+	NV40_STATE_VERTTEX2 = 29,
+	NV40_STATE_VERTTEX3 = 30,
+	NV40_STATE_VTXBUF = 31,
+	NV40_STATE_VTXFMT = 32,
+	NV40_STATE_VTXATTR = 33,
+	NV40_STATE_MAX = 34
+};
+
+#include "nv40_screen.h"
+
+#define NV40_NEW_BLEND		(1 <<  0)
+#define NV40_NEW_RAST		(1 <<  1)
+#define NV40_NEW_ZSA		(1 <<  2)
+#define NV40_NEW_SAMPLER	(1 <<  3)
+#define NV40_NEW_FB		(1 <<  4)
+#define NV40_NEW_STIPPLE	(1 <<  5)
+#define NV40_NEW_SCISSOR	(1 <<  6)
+#define NV40_NEW_VIEWPORT	(1 <<  7)
+#define NV40_NEW_BCOL		(1 <<  8)
+#define NV40_NEW_VERTPROG	(1 <<  9)
+#define NV40_NEW_FRAGPROG	(1 << 10)
+#define NV40_NEW_ARRAYS		(1 << 11)
+#define NV40_NEW_UCP		(1 << 12)
+
+struct nv40_rasterizer_state {
+	struct pipe_rasterizer_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv40_zsa_state {
+	struct pipe_depth_stencil_alpha_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv40_blend_state {
+	struct pipe_blend_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+
+struct nv40_state {
+	unsigned scissor_enabled;
+	unsigned stipple_enabled;
+	unsigned viewport_bypass;
+	unsigned fp_samplers;
+
+	uint64_t dirty;
+	struct nouveau_stateobj *hw[NV40_STATE_MAX];
+};
+
+struct nv40_context {
+	struct pipe_context pipe;
+
+	struct nouveau_winsys *nvws;
+	struct nv40_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	/* HW state derived from pipe states */
+	struct nv40_state state;
+	struct {
+		struct nv40_vertex_program *vertprog;
+
+		unsigned nr_attribs;
+		unsigned hw[PIPE_MAX_SHADER_INPUTS];
+		unsigned draw[PIPE_MAX_SHADER_INPUTS];
+		unsigned emit[PIPE_MAX_SHADER_INPUTS];
+	} swtnl;
+
+	enum {
+		HW, SWTNL, SWRAST
+	} render_mode;
+	unsigned fallback_swtnl;
+	unsigned fallback_swrast;
+
+	/* Context state */
+	unsigned dirty, draw_dirty;
+	struct pipe_scissor_state scissor;
+	unsigned stipple[32];
+	struct pipe_clip_state clip;
+	struct nv40_vertex_program *vertprog;
+	struct nv40_fragment_program *fragprog;
+	struct pipe_buffer *constbuf[PIPE_SHADER_TYPES];
+	unsigned constbuf_nr[PIPE_SHADER_TYPES];
+	struct nv40_rasterizer_state *rasterizer;
+	struct nv40_zsa_state *zsa;
+	struct nv40_blend_state *blend;
+	struct pipe_blend_color blend_colour;
+	struct pipe_viewport_state viewport;
+	struct pipe_framebuffer_state framebuffer;
+	struct pipe_buffer *idxbuf;
+	unsigned idxbuf_format;
+	struct nv40_sampler_state *tex_sampler[PIPE_MAX_SAMPLERS];
+	struct nv40_miptree *tex_miptree[PIPE_MAX_SAMPLERS];
+	unsigned nr_samplers;
+	unsigned nr_textures;
+	unsigned dirty_samplers;
+	struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS];
+	unsigned vtxbuf_nr;
+	struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS];
+	unsigned vtxelt_nr;
+	const unsigned *edgeflags;
+};
+
+static INLINE struct nv40_context *
+nv40_context(struct pipe_context *pipe)
+{
+	return (struct nv40_context *)pipe;
+}
+
+struct nv40_state_entry {
+	boolean (*validate)(struct nv40_context *nv40);
+	struct {
+		unsigned pipe;
+		unsigned hw;
+	} dirty;
+};
+
+extern void nv40_init_state_functions(struct nv40_context *nv40);
+extern void nv40_init_surface_functions(struct nv40_context *nv40);
+extern void nv40_init_query_functions(struct nv40_context *nv40);
+
+extern void nv40_screen_init_miptree_functions(struct pipe_screen *pscreen);
+
+/* nv40_draw.c */
+extern struct draw_stage *nv40_draw_render_stage(struct nv40_context *nv40);
+extern boolean nv40_draw_elements_swtnl(struct pipe_context *pipe,
+					struct pipe_buffer *idxbuf,
+					unsigned ib_size, unsigned mode,
+					unsigned start, unsigned count);
+
+/* nv40_vertprog.c */
+extern void nv40_vertprog_destroy(struct nv40_context *,
+				  struct nv40_vertex_program *);
+
+/* nv40_fragprog.c */
+extern void nv40_fragprog_destroy(struct nv40_context *,
+				  struct nv40_fragment_program *);
+
+/* nv40_fragtex.c */
+extern void nv40_fragtex_bind(struct nv40_context *);
+
+/* nv40_state.c and friends */
+extern boolean nv40_state_validate(struct nv40_context *nv40);
+extern boolean nv40_state_validate_swtnl(struct nv40_context *nv40);
+extern void nv40_state_emit(struct nv40_context *nv40);
+extern struct nv40_state_entry nv40_state_rasterizer;
+extern struct nv40_state_entry nv40_state_scissor;
+extern struct nv40_state_entry nv40_state_stipple;
+extern struct nv40_state_entry nv40_state_fragprog;
+extern struct nv40_state_entry nv40_state_vertprog;
+extern struct nv40_state_entry nv40_state_blend;
+extern struct nv40_state_entry nv40_state_blend_colour;
+extern struct nv40_state_entry nv40_state_zsa;
+extern struct nv40_state_entry nv40_state_viewport;
+extern struct nv40_state_entry nv40_state_framebuffer;
+extern struct nv40_state_entry nv40_state_fragtex;
+extern struct nv40_state_entry nv40_state_vbo;
+extern struct nv40_state_entry nv40_state_vtxfmt;
+
+/* nv40_vbo.c */
+extern boolean nv40_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv40_draw_elements(struct pipe_context *pipe,
+				  struct pipe_buffer *indexBuffer,
+				  unsigned indexSize,
+				  unsigned mode, unsigned start,
+				  unsigned count);
+
+/* nv40_clear.c */
+extern void nv40_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+#endif
diff --git a/src/gallium/drivers/nv40/nv40_draw.c b/src/gallium/drivers/nv40/nv40_draw.c
new file mode 100644
index 00000000000..8e56cdc2fe0
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_draw.c
@@ -0,0 +1,349 @@
+#include "pipe/p_shader_tokens.h"
+
+#include "util/u_pack_color.h"
+
+#include "draw/draw_context.h"
+#include "draw/draw_vertex.h"
+#include "draw/draw_pipe.h"
+
+#include "nv40_context.h"
+#define NV40_SHADER_NO_FUCKEDNESS
+#include "nv40_shader.h"
+
+/* Simple, but crappy, swtnl path, hopefully we wont need to hit this very
+ * often at all.  Uses "quadro style" vertex submission + a fixed vertex
+ * layout to avoid the need to generate a vertex program or vtxfmt.
+ */
+
+struct nv40_render_stage {
+	struct draw_stage stage;
+	struct nv40_context *nv40;
+	unsigned prim;
+};
+
+static INLINE struct nv40_render_stage *
+nv40_render_stage(struct draw_stage *stage)
+{
+	return (struct nv40_render_stage *)stage;
+}
+
+static INLINE void
+nv40_render_vertex(struct nv40_context *nv40, const struct vertex_header *v)
+{
+	unsigned i;
+
+	for (i = 0; i < nv40->swtnl.nr_attribs; i++) {
+		unsigned idx = nv40->swtnl.draw[i];
+		unsigned hw = nv40->swtnl.hw[i];
+
+		switch (nv40->swtnl.emit[i]) {
+		case EMIT_OMIT:
+			break;
+		case EMIT_1F:
+			BEGIN_RING(curie, NV40TCL_VTX_ATTR_1F(hw), 1);
+			OUT_RING  (fui(v->data[idx][0]));
+			break;
+		case EMIT_2F:
+			BEGIN_RING(curie, NV40TCL_VTX_ATTR_2F_X(hw), 2);
+			OUT_RING  (fui(v->data[idx][0]));
+			OUT_RING  (fui(v->data[idx][1]));
+			break;
+		case EMIT_3F:
+			BEGIN_RING(curie, NV40TCL_VTX_ATTR_3F_X(hw), 3);
+			OUT_RING  (fui(v->data[idx][0]));
+			OUT_RING  (fui(v->data[idx][1]));
+			OUT_RING  (fui(v->data[idx][2]));
+			break;
+		case EMIT_4F:
+			BEGIN_RING(curie, NV40TCL_VTX_ATTR_4F_X(hw), 4);
+			OUT_RING  (fui(v->data[idx][0]));
+			OUT_RING  (fui(v->data[idx][1]));
+			OUT_RING  (fui(v->data[idx][2]));
+			OUT_RING  (fui(v->data[idx][3]));
+			break;
+		case EMIT_4UB:
+			BEGIN_RING(curie, NV40TCL_VTX_ATTR_4UB(hw), 1);
+			OUT_RING  (pack_ub4(float_to_ubyte(v->data[idx][0]),
+					    float_to_ubyte(v->data[idx][1]),
+					    float_to_ubyte(v->data[idx][2]),
+					    float_to_ubyte(v->data[idx][3])));
+			break;
+		default:
+			assert(0);
+			break;
+		}
+	}
+}
+
+static INLINE void
+nv40_render_prim(struct draw_stage *stage, struct prim_header *prim,
+	       unsigned mode, unsigned count)
+{
+	struct nv40_render_stage *rs = nv40_render_stage(stage);
+	struct nv40_context *nv40 = rs->nv40;
+	struct nouveau_pushbuf *pb = nv40->nvws->channel->pushbuf;
+	unsigned i;
+
+	/* Ensure there's room for 4xfloat32 + potentially 3 begin/end */
+	if (pb->remaining < ((count * 20) + 6)) {
+		if (rs->prim != NV40TCL_BEGIN_END_STOP) {
+			NOUVEAU_ERR("AIII, missed flush\n");
+			assert(0);
+		}
+		FIRE_RING(NULL);
+		nv40_state_emit(nv40);
+	}
+
+	/* Switch primitive modes if necessary */
+	if (rs->prim != mode) {
+		if (rs->prim != NV40TCL_BEGIN_END_STOP) {
+			BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+			OUT_RING  (NV40TCL_BEGIN_END_STOP);	
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (mode);
+		rs->prim = mode;
+	}
+
+	/* Emit vertex data */
+	for (i = 0; i < count; i++)
+		nv40_render_vertex(nv40, prim->v[i]);
+
+	/* If it's likely we'll need to empty the push buffer soon, finish
+	 * off the primitive now.
+	 */
+	if (pb->remaining < ((count * 20) + 6)) {
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (NV40TCL_BEGIN_END_STOP);
+		rs->prim = NV40TCL_BEGIN_END_STOP;
+	}
+}
+
+static void
+nv40_render_point(struct draw_stage *draw, struct prim_header *prim)
+{
+	nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_POINTS, 1);
+}
+
+static void
+nv40_render_line(struct draw_stage *draw, struct prim_header *prim)
+{
+	nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_LINES, 2);
+}
+
+static void
+nv40_render_tri(struct draw_stage *draw, struct prim_header *prim)
+{
+	nv40_render_prim(draw, prim, NV40TCL_BEGIN_END_TRIANGLES, 3);
+}
+
+static void
+nv40_render_flush(struct draw_stage *draw, unsigned flags)
+{
+	struct nv40_render_stage *rs = nv40_render_stage(draw);
+	struct nv40_context *nv40 = rs->nv40;
+
+	if (rs->prim != NV40TCL_BEGIN_END_STOP) {
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (NV40TCL_BEGIN_END_STOP);
+		rs->prim = NV40TCL_BEGIN_END_STOP;
+	}
+}
+
+static void
+nv40_render_reset_stipple_counter(struct draw_stage *draw)
+{
+}
+
+static void
+nv40_render_destroy(struct draw_stage *draw)
+{
+	FREE(draw);
+}
+
+static INLINE void
+emit_mov(struct nv40_vertex_program *vp,
+	 unsigned dst, unsigned src, unsigned vor, unsigned mask)
+{
+	struct nv40_vertex_program_exec *inst;
+
+	vp->insns = realloc(vp->insns,
+			    sizeof(struct nv40_vertex_program_exec) *
+			    ++vp->nr_insns);
+	inst = &vp->insns[vp->nr_insns - 1];
+
+	inst->data[0] = 0x401f9c6c;
+	inst->data[1] = 0x0040000d | (src << 8);
+	inst->data[2] = 0x8106c083;
+	inst->data[3] = 0x6041ff80 | (dst << 2) | (mask << 13);
+	inst->const_index = -1;
+	inst->has_branch_offset = FALSE;
+
+	vp->ir |= (1 << src);
+	if (vor != ~0)
+		vp->or |= (1 << vor);
+}
+
+static struct nv40_vertex_program *
+create_drawvp(struct nv40_context *nv40)
+{
+	struct nv40_vertex_program *vp = CALLOC_STRUCT(nv40_vertex_program);
+	unsigned i;
+
+	emit_mov(vp, NV40_VP_INST_DEST_POS, 0, ~0, 0xf);
+	emit_mov(vp, NV40_VP_INST_DEST_COL0, 3, 0, 0xf);
+	emit_mov(vp, NV40_VP_INST_DEST_COL1, 4, 1, 0xf);
+	emit_mov(vp, NV40_VP_INST_DEST_BFC0, 3, 2, 0xf);
+	emit_mov(vp, NV40_VP_INST_DEST_BFC1, 4, 3, 0xf);
+	emit_mov(vp, NV40_VP_INST_DEST_FOGC, 5, 4, 0x8);
+	for (i = 0; i < 8; i++)
+		emit_mov(vp, NV40_VP_INST_DEST_TC(i), 8 + i, 14 + i, 0xf);
+
+	vp->insns[vp->nr_insns - 1].data[3] |= 1;
+	vp->translated = TRUE;
+	return vp;
+}
+
+struct draw_stage *
+nv40_draw_render_stage(struct nv40_context *nv40)
+{
+	struct nv40_render_stage *render = CALLOC_STRUCT(nv40_render_stage);
+
+	if (!nv40->swtnl.vertprog)
+		nv40->swtnl.vertprog = create_drawvp(nv40);
+
+	render->nv40 = nv40;
+	render->stage.draw = nv40->draw;
+	render->stage.point = nv40_render_point;
+	render->stage.line = nv40_render_line;
+	render->stage.tri = nv40_render_tri;
+	render->stage.flush = nv40_render_flush;
+	render->stage.reset_stipple_counter = nv40_render_reset_stipple_counter;
+	render->stage.destroy = nv40_render_destroy;
+
+	return &render->stage;
+}
+
+boolean
+nv40_draw_elements_swtnl(struct pipe_context *pipe,
+			 struct pipe_buffer *idxbuf, unsigned idxbuf_size,
+			 unsigned mode, unsigned start, unsigned count)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+	unsigned i;
+	void *map;
+
+	if (!nv40_state_validate_swtnl(nv40))
+		return FALSE;
+	nv40->state.dirty &= ~(1ULL << NV40_STATE_VTXBUF);
+	nv40_state_emit(nv40);
+
+	for (i = 0; i < nv40->vtxbuf_nr; i++) {
+		map = ws->buffer_map(ws, nv40->vtxbuf[i].buffer,
+				     PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_vertex_buffer(nv40->draw, i, map);
+	}
+
+	if (idxbuf) {
+		map = ws->buffer_map(ws, idxbuf, PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_element_buffer(nv40->draw, idxbuf_size, map);
+	} else {
+		draw_set_mapped_element_buffer(nv40->draw, 0, NULL);
+	}
+
+	if (nv40->constbuf[PIPE_SHADER_VERTEX]) {
+		const unsigned nr = nv40->constbuf_nr[PIPE_SHADER_VERTEX];
+
+		map = ws->buffer_map(ws, nv40->constbuf[PIPE_SHADER_VERTEX],
+				     PIPE_BUFFER_USAGE_CPU_READ);
+		draw_set_mapped_constant_buffer(nv40->draw, map, nr);
+	}
+
+	draw_arrays(nv40->draw, mode, start, count);
+
+	for (i = 0; i < nv40->vtxbuf_nr; i++)
+		ws->buffer_unmap(ws, nv40->vtxbuf[i].buffer);
+
+	if (idxbuf)
+		ws->buffer_unmap(ws, idxbuf);
+
+	if (nv40->constbuf[PIPE_SHADER_VERTEX])
+		ws->buffer_unmap(ws, nv40->constbuf[PIPE_SHADER_VERTEX]);
+
+	draw_flush(nv40->draw);
+	pipe->flush(pipe, 0, NULL);
+
+	return TRUE;
+}
+
+static INLINE void
+emit_attrib(struct nv40_context *nv40, unsigned hw, unsigned emit,
+	    unsigned semantic, unsigned index)
+{
+	unsigned draw_out = draw_find_vs_output(nv40->draw, semantic, index);
+	unsigned a = nv40->swtnl.nr_attribs++;
+
+	nv40->swtnl.hw[a] = hw;
+	nv40->swtnl.emit[a] = emit;
+	nv40->swtnl.draw[a] = draw_out;
+}
+
+static boolean
+nv40_state_vtxfmt_validate(struct nv40_context *nv40)
+{
+	struct nv40_fragment_program *fp = nv40->fragprog;
+	unsigned colour = 0, texcoords = 0, fog = 0, i;
+
+	/* Determine needed fragprog inputs */
+	for (i = 0; i < fp->info.num_inputs; i++) {
+		switch (fp->info.input_semantic_name[i]) {
+		case TGSI_SEMANTIC_POSITION:
+			break;
+		case TGSI_SEMANTIC_COLOR:
+			colour |= (1 << fp->info.input_semantic_index[i]);
+			break;
+		case TGSI_SEMANTIC_GENERIC:
+			texcoords |= (1 << fp->info.input_semantic_index[i]);
+			break;
+		case TGSI_SEMANTIC_FOG:
+			fog = 1;
+			break;
+		default:
+			assert(0);
+		}
+	}
+
+	nv40->swtnl.nr_attribs = 0;
+
+	/* Map draw vtxprog output to hw attribute IDs */
+	for (i = 0; i < 2; i++) {
+		if (!(colour & (1 << i)))
+			continue;
+		emit_attrib(nv40, 3 + i, EMIT_4UB, TGSI_SEMANTIC_COLOR, i);
+	}
+
+	for (i = 0; i < 8; i++) {
+		if (!(texcoords & (1 << i)))
+			continue;
+		emit_attrib(nv40, 8 + i, EMIT_4F, TGSI_SEMANTIC_GENERIC, i);
+	}
+
+	if (fog) {
+		emit_attrib(nv40, 5, EMIT_1F, TGSI_SEMANTIC_FOG, 0);
+	}
+
+	emit_attrib(nv40, 0, EMIT_3F, TGSI_SEMANTIC_POSITION, 0);
+
+	return FALSE;
+}
+
+struct nv40_state_entry nv40_state_vtxfmt = {
+	.validate = nv40_state_vtxfmt_validate,
+	.dirty = {
+		.pipe = NV40_NEW_ARRAYS | NV40_NEW_FRAGPROG,
+		.hw = 0
+	}
+};
+
diff --git a/src/gallium/drivers/nv40/nv40_fragprog.c b/src/gallium/drivers/nv40/nv40_fragprog.c
new file mode 100644
index 00000000000..91dcbebda0d
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_fragprog.c
@@ -0,0 +1,991 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv40_context.h"
+
+#define SWZ_X 0
+#define SWZ_Y 1
+#define SWZ_Z 2
+#define SWZ_W 3
+#define MASK_X 1
+#define MASK_Y 2
+#define MASK_Z 4
+#define MASK_W 8
+#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W)
+#define DEF_SCALE NV40_FP_OP_DST_SCALE_1X
+#define DEF_CTEST NV40_FP_OP_COND_TR
+#include "nv40_shader.h"
+
+#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w)
+#define neg(s) nv40_sr_neg((s))
+#define abs(s) nv40_sr_abs((s))
+#define scale(s,v) nv40_sr_scale((s), NV40_FP_OP_DST_SCALE_##v)
+
+#define MAX_CONSTS 128
+#define MAX_IMM 32
+struct nv40_fpc {
+	struct nv40_fragment_program *fp;
+
+	uint attrib_map[PIPE_MAX_SHADER_INPUTS];
+
+	unsigned r_temps;
+	unsigned r_temps_discard;
+	struct nv40_sreg r_result[PIPE_MAX_SHADER_OUTPUTS];
+	struct nv40_sreg *r_temp;
+
+	int num_regs;
+
+	unsigned inst_offset;
+	unsigned have_const;
+
+	struct {
+		int pipe;
+		float vals[4];
+	} consts[MAX_CONSTS];
+	int nr_consts;
+
+	struct nv40_sreg imm[MAX_IMM];
+	unsigned nr_imm;
+};
+
+static INLINE struct nv40_sreg
+temp(struct nv40_fpc *fpc)
+{
+	int idx = ffs(~fpc->r_temps) - 1;
+
+	if (idx < 0) {
+		NOUVEAU_ERR("out of temps!!\n");
+		assert(0);
+		return nv40_sr(NV40SR_TEMP, 0);
+	}
+
+	fpc->r_temps |= (1 << idx);
+	fpc->r_temps_discard |= (1 << idx);
+	return nv40_sr(NV40SR_TEMP, idx);
+}
+
+static INLINE void
+release_temps(struct nv40_fpc *fpc)
+{
+	fpc->r_temps &= ~fpc->r_temps_discard;
+	fpc->r_temps_discard = 0;
+}
+
+static INLINE struct nv40_sreg
+constant(struct nv40_fpc *fpc, int pipe, float vals[4])
+{
+	int idx;
+
+	if (fpc->nr_consts == MAX_CONSTS)
+		assert(0);
+	idx = fpc->nr_consts++;
+
+	fpc->consts[idx].pipe = pipe;
+	if (pipe == -1)
+		memcpy(fpc->consts[idx].vals, vals, 4 * sizeof(float));
+	return nv40_sr(NV40SR_CONST, idx);
+}
+
+#define arith(cc,s,o,d,m,s0,s1,s2) \
+	nv40_fp_arith((cc), (s), NV40_FP_OP_OPCODE_##o, \
+			(d), (m), (s0), (s1), (s2))
+#define tex(cc,s,o,u,d,m,s0,s1,s2) \
+	nv40_fp_tex((cc), (s), NV40_FP_OP_OPCODE_##o, (u), \
+		    (d), (m), (s0), none, none)
+
+static void
+grow_insns(struct nv40_fpc *fpc, int size)
+{
+	struct nv40_fragment_program *fp = fpc->fp;
+
+	fp->insn_len += size;
+	fp->insn = realloc(fp->insn, sizeof(uint32_t) * fp->insn_len);
+}
+
+static void
+emit_src(struct nv40_fpc *fpc, int pos, struct nv40_sreg src)
+{
+	struct nv40_fragment_program *fp = fpc->fp;
+	uint32_t *hw = &fp->insn[fpc->inst_offset];
+	uint32_t sr = 0;
+
+	switch (src.type) {
+	case NV40SR_INPUT:
+		sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT);
+		hw[0] |= (src.index << NV40_FP_OP_INPUT_SRC_SHIFT);
+		break;
+	case NV40SR_OUTPUT:
+		sr |= NV40_FP_REG_SRC_HALF;
+		/* fall-through */
+	case NV40SR_TEMP:
+		sr |= (NV40_FP_REG_TYPE_TEMP << NV40_FP_REG_TYPE_SHIFT);
+		sr |= (src.index << NV40_FP_REG_SRC_SHIFT);
+		break;
+	case NV40SR_CONST:
+		if (!fpc->have_const) {
+			grow_insns(fpc, 4);
+			fpc->have_const = 1;
+		}
+
+		hw = &fp->insn[fpc->inst_offset];
+		if (fpc->consts[src.index].pipe >= 0) {
+			struct nv40_fragment_program_data *fpd;
+
+			fp->consts = realloc(fp->consts, ++fp->nr_consts *
+					     sizeof(*fpd));
+			fpd = &fp->consts[fp->nr_consts - 1];
+			fpd->offset = fpc->inst_offset + 4;
+			fpd->index = fpc->consts[src.index].pipe;
+			memset(&fp->insn[fpd->offset], 0, sizeof(uint32_t) * 4);
+		} else {
+			memcpy(&fp->insn[fpc->inst_offset + 4],
+				fpc->consts[src.index].vals,
+				sizeof(uint32_t) * 4);
+		}
+
+		sr |= (NV40_FP_REG_TYPE_CONST << NV40_FP_REG_TYPE_SHIFT);	
+		break;
+	case NV40SR_NONE:
+		sr |= (NV40_FP_REG_TYPE_INPUT << NV40_FP_REG_TYPE_SHIFT);
+		break;
+	default:
+		assert(0);
+	}
+
+	if (src.negate)
+		sr |= NV40_FP_REG_NEGATE;
+
+	if (src.abs)
+		hw[1] |= (1 << (29 + pos));
+
+	sr |= ((src.swz[0] << NV40_FP_REG_SWZ_X_SHIFT) |
+	       (src.swz[1] << NV40_FP_REG_SWZ_Y_SHIFT) |
+	       (src.swz[2] << NV40_FP_REG_SWZ_Z_SHIFT) |
+	       (src.swz[3] << NV40_FP_REG_SWZ_W_SHIFT));
+
+	hw[pos + 1] |= sr;
+}
+
+static void
+emit_dst(struct nv40_fpc *fpc, struct nv40_sreg dst)
+{
+	struct nv40_fragment_program *fp = fpc->fp;
+	uint32_t *hw = &fp->insn[fpc->inst_offset];
+
+	switch (dst.type) {
+	case NV40SR_TEMP:
+		if (fpc->num_regs < (dst.index + 1))
+			fpc->num_regs = dst.index + 1;
+		break;
+	case NV40SR_OUTPUT:
+		if (dst.index == 1) {
+			fp->fp_control |= 0xe;
+		} else {
+			hw[0] |= NV40_FP_OP_OUT_REG_HALF;
+		}
+		break;
+	case NV40SR_NONE:
+		hw[0] |= (1 << 30);
+		break;
+	default:
+		assert(0);
+	}
+
+	hw[0] |= (dst.index << NV40_FP_OP_OUT_REG_SHIFT);
+}
+
+static void
+nv40_fp_arith(struct nv40_fpc *fpc, int sat, int op,
+	      struct nv40_sreg dst, int mask,
+	      struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2)
+{
+	struct nv40_fragment_program *fp = fpc->fp;
+	uint32_t *hw;
+
+	fpc->inst_offset = fp->insn_len;
+	fpc->have_const = 0;
+	grow_insns(fpc, 4);
+	hw = &fp->insn[fpc->inst_offset];
+	memset(hw, 0, sizeof(uint32_t) * 4);
+
+	if (op == NV40_FP_OP_OPCODE_KIL)
+		fp->fp_control |= NV40TCL_FP_CONTROL_KIL;
+	hw[0] |= (op << NV40_FP_OP_OPCODE_SHIFT);
+	hw[0] |= (mask << NV40_FP_OP_OUTMASK_SHIFT);
+	hw[2] |= (dst.dst_scale << NV40_FP_OP_DST_SCALE_SHIFT);
+
+	if (sat)
+		hw[0] |= NV40_FP_OP_OUT_SAT;
+
+	if (dst.cc_update)
+		hw[0] |= NV40_FP_OP_COND_WRITE_ENABLE;
+	hw[1] |= (dst.cc_test << NV40_FP_OP_COND_SHIFT);
+	hw[1] |= ((dst.cc_swz[0] << NV40_FP_OP_COND_SWZ_X_SHIFT) |
+		  (dst.cc_swz[1] << NV40_FP_OP_COND_SWZ_Y_SHIFT) |
+		  (dst.cc_swz[2] << NV40_FP_OP_COND_SWZ_Z_SHIFT) |
+		  (dst.cc_swz[3] << NV40_FP_OP_COND_SWZ_W_SHIFT));
+
+	emit_dst(fpc, dst);
+	emit_src(fpc, 0, s0);
+	emit_src(fpc, 1, s1);
+	emit_src(fpc, 2, s2);
+}
+
+static void
+nv40_fp_tex(struct nv40_fpc *fpc, int sat, int op, int unit,
+	    struct nv40_sreg dst, int mask,
+	    struct nv40_sreg s0, struct nv40_sreg s1, struct nv40_sreg s2)
+{
+	struct nv40_fragment_program *fp = fpc->fp;
+
+	nv40_fp_arith(fpc, sat, op, dst, mask, s0, s1, s2);
+
+	fp->insn[fpc->inst_offset] |= (unit << NV40_FP_OP_TEX_UNIT_SHIFT);
+	fp->samplers |= (1 << unit);
+}
+
+static INLINE struct nv40_sreg
+tgsi_src(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc)
+{
+	struct nv40_sreg src;
+
+	switch (fsrc->SrcRegister.File) {
+	case TGSI_FILE_INPUT:
+		src = nv40_sr(NV40SR_INPUT,
+			      fpc->attrib_map[fsrc->SrcRegister.Index]);
+		break;
+	case TGSI_FILE_CONSTANT:
+		src = constant(fpc, fsrc->SrcRegister.Index, NULL);
+		break;
+	case TGSI_FILE_IMMEDIATE:
+		assert(fsrc->SrcRegister.Index < fpc->nr_imm);
+		src = fpc->imm[fsrc->SrcRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		src = fpc->r_temp[fsrc->SrcRegister.Index];
+		break;
+	/* NV40 fragprog result regs are just temps, so this is simple */
+	case TGSI_FILE_OUTPUT:
+		src = fpc->r_result[fsrc->SrcRegister.Index];
+		break;
+	default:
+		NOUVEAU_ERR("bad src file\n");
+		break;
+	}
+
+	src.abs = fsrc->SrcRegisterExtMod.Absolute;
+	src.negate = fsrc->SrcRegister.Negate;
+	src.swz[0] = fsrc->SrcRegister.SwizzleX;
+	src.swz[1] = fsrc->SrcRegister.SwizzleY;
+	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
+	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	return src;
+}
+
+static INLINE struct nv40_sreg
+tgsi_dst(struct nv40_fpc *fpc, const struct tgsi_full_dst_register *fdst) {
+	switch (fdst->DstRegister.File) {
+	case TGSI_FILE_OUTPUT:
+		return fpc->r_result[fdst->DstRegister.Index];
+	case TGSI_FILE_TEMPORARY:
+		return fpc->r_temp[fdst->DstRegister.Index];
+	case TGSI_FILE_NULL:
+		return nv40_sr(NV40SR_NONE, 0);
+	default:
+		NOUVEAU_ERR("bad dst file %d\n", fdst->DstRegister.File);
+		return nv40_sr(NV40SR_NONE, 0);
+	}
+}
+
+static INLINE int
+tgsi_mask(uint tgsi)
+{
+	int mask = 0;
+
+	if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X;
+	if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y;
+	if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z;
+	if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W;
+	return mask;
+}
+
+static boolean
+src_native_swz(struct nv40_fpc *fpc, const struct tgsi_full_src_register *fsrc,
+	       struct nv40_sreg *src)
+{
+	const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0);
+	struct nv40_sreg tgsi = tgsi_src(fpc, fsrc);
+	uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0;
+	uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX,
+			fsrc->SrcRegisterExtSwz.NegateY,
+			fsrc->SrcRegisterExtSwz.NegateZ,
+			fsrc->SrcRegisterExtSwz.NegateW };
+	uint c;
+
+	for (c = 0; c < 4; c++) {
+		switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) {
+		case TGSI_EXTSWIZZLE_X:
+		case TGSI_EXTSWIZZLE_Y:
+		case TGSI_EXTSWIZZLE_Z:
+		case TGSI_EXTSWIZZLE_W:
+			mask |= (1 << c);
+			break;
+		case TGSI_EXTSWIZZLE_ZERO:
+			zero_mask |= (1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		case TGSI_EXTSWIZZLE_ONE:
+			one_mask |= (1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		default:
+			assert(0);
+		}
+
+		if (!tgsi.negate && neg[c])
+			neg_mask |= (1 << c);
+	}
+
+	if (mask == MASK_ALL && !neg_mask)
+		return TRUE;
+
+	*src = temp(fpc);
+
+	if (mask)
+		arith(fpc, 0, MOV, *src, mask, tgsi, none, none);
+
+	if (zero_mask)
+		arith(fpc, 0, SFL, *src, zero_mask, *src, none, none);
+
+	if (one_mask)
+		arith(fpc, 0, STR, *src, one_mask, *src, none, none);
+
+	if (neg_mask) {
+		struct nv40_sreg one = temp(fpc);
+		arith(fpc, 0, STR, one, neg_mask, one, none, none);
+		arith(fpc, 0, MUL, *src, neg_mask, *src, neg(one), none);
+	}
+
+	return FALSE;
+}
+
+static boolean
+nv40_fragprog_parse_instruction(struct nv40_fpc *fpc,
+				const struct tgsi_full_instruction *finst)
+{
+	const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0);
+	struct nv40_sreg src[3], dst, tmp;
+	int mask, sat, unit;
+	int ai = -1, ci = -1, ii = -1;
+	int i;
+
+	if (finst->Instruction.Opcode == TGSI_OPCODE_END)
+		return TRUE;
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+			src[i] = tgsi_src(fpc, fsrc);
+		}
+	}
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_TEMPORARY:
+			if (!src_native_swz(fpc, fsrc, &src[i]))
+				continue;
+			break;
+		default:
+			break;
+		}
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
+				ai = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(fpc, fsrc);
+			} else {
+				src[i] = temp(fpc);
+				arith(fpc, 0, MOV, src[i], MASK_ALL,
+				      tgsi_src(fpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_CONSTANT:
+			if ((ci == -1 && ii == -1) ||
+			    ci == fsrc->SrcRegister.Index) {
+				ci = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(fpc, fsrc);
+			} else {
+				src[i] = temp(fpc);
+				arith(fpc, 0, MOV, src[i], MASK_ALL,
+				      tgsi_src(fpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_IMMEDIATE:
+			if ((ci == -1 && ii == -1) ||
+			    ii == fsrc->SrcRegister.Index) {
+				ii = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(fpc, fsrc);
+			} else {
+				src[i] = temp(fpc);
+				arith(fpc, 0, MOV, src[i], MASK_ALL,
+				      tgsi_src(fpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_TEMPORARY:
+			/* handled above */
+			break;
+		case TGSI_FILE_SAMPLER:
+			unit = fsrc->SrcRegister.Index;
+			break;
+		case TGSI_FILE_OUTPUT:
+			break;
+		default:
+			NOUVEAU_ERR("bad src file\n");
+			return FALSE;
+		}
+	}
+
+	dst  = tgsi_dst(fpc, &finst->FullDstRegisters[0]);
+	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+	sat  = (finst->Instruction.Saturate == TGSI_SAT_ZERO_ONE);
+
+	switch (finst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		arith(fpc, sat, MOV, dst, mask, abs(src[0]), none, none);
+		break;
+	case TGSI_OPCODE_ADD:
+		arith(fpc, sat, ADD, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_CMP:
+		tmp = temp(fpc);
+		arith(fpc, sat, MOV, dst, mask, src[2], none, none);
+		tmp.cc_update = 1;
+		arith(fpc, 0, MOV, tmp, 0xf, src[0], none, none);
+		dst.cc_test = NV40_VP_INST_COND_LT;
+		arith(fpc, sat, MOV, dst, mask, src[1], none, none);
+		break;
+	case TGSI_OPCODE_COS:
+		arith(fpc, sat, COS, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_DDX:
+		if (mask & (MASK_Z | MASK_W)) {
+			tmp = temp(fpc);
+			arith(fpc, sat, DDX, tmp, MASK_X | MASK_Y,
+			      swz(src[0], Z, W, Z, W), none, none);
+			arith(fpc, 0, MOV, tmp, MASK_Z | MASK_W,
+			      swz(tmp, X, Y, X, Y), none, none);
+			arith(fpc, sat, DDX, tmp, MASK_X | MASK_Y, src[0],
+			      none, none);
+			arith(fpc, 0, MOV, dst, mask, tmp, none, none);
+		} else {
+			arith(fpc, sat, DDX, dst, mask, src[0], none, none);
+		}
+		break;
+	case TGSI_OPCODE_DDY:
+		if (mask & (MASK_Z | MASK_W)) {
+			tmp = temp(fpc);
+			arith(fpc, sat, DDY, tmp, MASK_X | MASK_Y,
+			      swz(src[0], Z, W, Z, W), none, none);
+			arith(fpc, 0, MOV, tmp, MASK_Z | MASK_W,
+			      swz(tmp, X, Y, X, Y), none, none);
+			arith(fpc, sat, DDY, tmp, MASK_X | MASK_Y, src[0],
+			      none, none);
+			arith(fpc, 0, MOV, dst, mask, tmp, none, none);
+		} else {
+			arith(fpc, sat, DDY, dst, mask, src[0], none, none);
+		}
+		break;
+	case TGSI_OPCODE_DP3:
+		arith(fpc, sat, DP3, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DP4:
+		arith(fpc, sat, DP4, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DPH:
+		tmp = temp(fpc);
+		arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[1], none);
+		arith(fpc, sat, ADD, dst, mask, swz(tmp, X, X, X, X),
+		      swz(src[1], W, W, W, W), none);
+		break;
+	case TGSI_OPCODE_DST:
+		arith(fpc, sat, DST, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_EX2:
+		arith(fpc, sat, EX2, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FLR:
+		arith(fpc, sat, FLR, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FRC:
+		arith(fpc, sat, FRC, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_KILP:
+		arith(fpc, 0, KIL, none, 0, none, none, none);
+		break;
+	case TGSI_OPCODE_KIL:
+		dst = nv40_sr(NV40SR_NONE, 0);
+		dst.cc_update = 1;
+		arith(fpc, 0, MOV, dst, MASK_ALL, src[0], none, none);
+		dst.cc_update = 0; dst.cc_test = NV40_FP_OP_COND_LT;
+		arith(fpc, 0, KIL, dst, 0, none, none, none);
+		break;
+	case TGSI_OPCODE_LG2:
+		arith(fpc, sat, LG2, dst, mask, src[0], none, none);
+		break;
+//	case TGSI_OPCODE_LIT:
+	case TGSI_OPCODE_LRP:
+		tmp = temp(fpc);
+		arith(fpc, 0, MAD, tmp, mask, neg(src[0]), src[2], src[2]);
+		arith(fpc, sat, MAD, dst, mask, src[0], src[1], tmp);
+		break;
+	case TGSI_OPCODE_MAD:
+		arith(fpc, sat, MAD, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAX:
+		arith(fpc, sat, MAX, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MIN:
+		arith(fpc, sat, MIN, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MOV:
+		arith(fpc, sat, MOV, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_MUL:
+		arith(fpc, sat, MUL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_NOISE1:
+	case TGSI_OPCODE_NOISE2:
+	case TGSI_OPCODE_NOISE3:
+	case TGSI_OPCODE_NOISE4:
+		arith(fpc, sat, SFL, dst, mask, none, none, none);
+		break;
+	case TGSI_OPCODE_POW:
+		tmp = temp(fpc);
+		arith(fpc, 0, LG2, tmp, MASK_X,
+		      swz(src[0], X, X, X, X), none, none);
+		arith(fpc, 0, MUL, tmp, MASK_X, swz(tmp, X, X, X, X),
+		      swz(src[1], X, X, X, X), none);
+		arith(fpc, sat, EX2, dst, mask,
+		      swz(tmp, X, X, X, X), none, none);
+		break;
+	case TGSI_OPCODE_RCP:
+		arith(fpc, sat, RCP, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_RET:
+		assert(0);
+		break;
+	case TGSI_OPCODE_RFL:
+		tmp = temp(fpc);
+		arith(fpc, 0, DP3, tmp, MASK_X, src[0], src[0], none);
+		arith(fpc, 0, DP3, tmp, MASK_Y, src[0], src[1], none);
+		arith(fpc, 0, DIV, scale(tmp, 2X), MASK_Z,
+		      swz(tmp, Y, Y, Y, Y), swz(tmp, X, X, X, X), none);
+		arith(fpc, sat, MAD, dst, mask,
+		      swz(tmp, Z, Z, Z, Z), src[0], neg(src[1]));
+		break;
+	case TGSI_OPCODE_RSQ:
+		tmp = temp(fpc);
+		arith(fpc, 0, LG2, scale(tmp, INV_2X), MASK_X,
+		      abs(swz(src[0], X, X, X, X)), none, none);
+		arith(fpc, sat, EX2, dst, mask,
+		      neg(swz(tmp, X, X, X, X)), none, none);
+		break;
+	case TGSI_OPCODE_SCS:
+		if (mask & MASK_X) {
+			arith(fpc, sat, COS, dst, MASK_X,
+			      swz(src[0], X, X, X, X), none, none);
+		}
+		if (mask & MASK_Y) {
+			arith(fpc, sat, SIN, dst, MASK_Y,
+			      swz(src[0], X, X, X, X), none, none);
+		}
+		break;
+	case TGSI_OPCODE_SEQ:
+		arith(fpc, sat, SEQ, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SFL:
+		arith(fpc, sat, SFL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SGE:
+		arith(fpc, sat, SGE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SGT:
+		arith(fpc, sat, SGT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SIN:
+		arith(fpc, sat, SIN, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_SLE:
+		arith(fpc, sat, SLE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SLT:
+		arith(fpc, sat, SLT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SNE:
+		arith(fpc, sat, SNE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_STR:
+		arith(fpc, sat, STR, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SUB:
+		arith(fpc, sat, ADD, dst, mask, src[0], neg(src[1]), none);
+		break;
+	case TGSI_OPCODE_TEX:
+		tex(fpc, sat, TEX, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_TXB:
+		tex(fpc, sat, TXB, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_TXP:
+		tex(fpc, sat, TXP, unit, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_XPD:
+		tmp = temp(fpc);
+		arith(fpc, 0, MUL, tmp, mask,
+		      swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none);
+		arith(fpc, sat, MAD, dst, (mask & ~MASK_W),
+		      swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y),
+		      neg(tmp));
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	release_temps(fpc);
+	return TRUE;
+}
+
+static boolean
+nv40_fragprog_parse_decl_attrib(struct nv40_fpc *fpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	int hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = NV40_FP_OP_INPUT_SRC_POSITION;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV40_FP_OP_INPUT_SRC_COL0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV40_FP_OP_INPUT_SRC_COL1;
+		} else {
+			NOUVEAU_ERR("bad colour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_FOG:
+		hw = NV40_FP_OP_INPUT_SRC_FOGC;
+		break;
+	case TGSI_SEMANTIC_GENERIC:
+		if (fdec->Semantic.SemanticIndex <= 7) {
+			hw = NV40_FP_OP_INPUT_SRC_TC(fdec->Semantic.
+						     SemanticIndex);
+		} else {
+			NOUVEAU_ERR("bad generic semantic index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad input semantic\n");
+		return FALSE;
+	}
+
+	fpc->attrib_map[fdec->DeclarationRange.First] = hw;
+	return TRUE;
+}
+
+static boolean
+nv40_fragprog_parse_decl_output(struct nv40_fpc *fpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	unsigned idx = fdec->DeclarationRange.First;
+	unsigned hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = 1;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		switch (fdec->Semantic.SemanticIndex) {
+		case 0: hw = 0; break;
+		case 1: hw = 2; break;
+		case 2: hw = 3; break;
+		case 3: hw = 4; break;
+		default:
+			NOUVEAU_ERR("bad rcol index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad output semantic\n");
+		return FALSE;
+	}
+
+	fpc->r_result[idx] = nv40_sr(NV40SR_OUTPUT, hw);
+	fpc->r_temps |= (1 << hw);
+	return TRUE;
+}
+
+static boolean
+nv40_fragprog_prepare(struct nv40_fpc *fpc)
+{
+	struct tgsi_parse_context p;
+	int high_temp = -1, i;
+
+	tgsi_parse_init(&p, fpc->fp->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch(tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *fdec;
+			fdec = &p.FullToken.FullDeclaration;
+			switch (fdec->Declaration.File) {
+			case TGSI_FILE_INPUT:
+				if (!nv40_fragprog_parse_decl_attrib(fpc, fdec))
+					goto out_err;
+				break;
+			case TGSI_FILE_OUTPUT:
+				if (!nv40_fragprog_parse_decl_output(fpc, fdec))
+					goto out_err;
+				break;
+			case TGSI_FILE_TEMPORARY:
+				if (fdec->DeclarationRange.Last > high_temp) {
+					high_temp =
+						fdec->DeclarationRange.Last;
+				}
+				break;
+			default:
+				break;
+			}
+		}
+			break;
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			struct tgsi_full_immediate *imm;
+			float vals[4];
+			
+			imm = &p.FullToken.FullImmediate;
+			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
+			assert(fpc->nr_imm < MAX_IMM);
+
+			vals[0] = imm->u.ImmediateFloat32[0].Float;
+			vals[1] = imm->u.ImmediateFloat32[1].Float;
+			vals[2] = imm->u.ImmediateFloat32[2].Float;
+			vals[3] = imm->u.ImmediateFloat32[3].Float;
+			fpc->imm[fpc->nr_imm++] = constant(fpc, -1, vals);
+		}
+			break;
+		default:
+			break;
+		}
+	}
+	tgsi_parse_free(&p);
+
+	if (++high_temp) {
+		fpc->r_temp = CALLOC(high_temp, sizeof(struct nv40_sreg));
+		for (i = 0; i < high_temp; i++)
+			fpc->r_temp[i] = temp(fpc);
+		fpc->r_temps_discard = 0;
+	}
+
+	return TRUE;
+
+out_err:
+	if (fpc->r_temp)
+		FREE(fpc->r_temp);
+	tgsi_parse_free(&p);
+	return FALSE;
+}
+
+static void
+nv40_fragprog_translate(struct nv40_context *nv40,
+			struct nv40_fragment_program *fp)
+{
+	struct tgsi_parse_context parse;
+	struct nv40_fpc *fpc = NULL;
+
+	fpc = CALLOC(1, sizeof(struct nv40_fpc));
+	if (!fpc)
+		return;
+	fpc->fp = fp;
+	fpc->num_regs = 2;
+
+	if (!nv40_fragprog_prepare(fpc)) {
+		FREE(fpc);
+		return;
+	}
+
+	tgsi_parse_init(&parse, fp->pipe.tokens);
+
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		tgsi_parse_token(&parse);
+
+		switch (parse.FullToken.Token.Type) {
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+
+			finst = &parse.FullToken.FullInstruction;
+			if (!nv40_fragprog_parse_instruction(fpc, finst))
+				goto out_err;
+		}
+			break;
+		default:
+			break;
+		}
+	}
+
+	fp->fp_control |= fpc->num_regs << NV40TCL_FP_CONTROL_TEMP_COUNT_SHIFT;
+
+	/* Terminate final instruction */
+	fp->insn[fpc->inst_offset] |= 0x00000001;
+
+	/* Append NOP + END instruction, may or may not be necessary. */
+	fpc->inst_offset = fp->insn_len;
+	grow_insns(fpc, 4);
+	fp->insn[fpc->inst_offset + 0] = 0x00000001;
+	fp->insn[fpc->inst_offset + 1] = 0x00000000;
+	fp->insn[fpc->inst_offset + 2] = 0x00000000;
+	fp->insn[fpc->inst_offset + 3] = 0x00000000;
+	
+	fp->translated = TRUE;
+out_err:
+	tgsi_parse_free(&parse);
+	if (fpc->r_temp)
+		FREE(fpc->r_temp);
+	FREE(fpc);
+}
+
+static void
+nv40_fragprog_upload(struct nv40_context *nv40,
+		     struct nv40_fragment_program *fp)
+{
+	struct pipe_winsys *ws = nv40->pipe.winsys;
+	const uint32_t le = 1;
+	uint32_t *map;
+	int i;
+
+	map = ws->buffer_map(ws, fp->buffer, PIPE_BUFFER_USAGE_CPU_WRITE);
+
+#if 0
+	for (i = 0; i < fp->insn_len; i++) {
+		fflush(stdout); fflush(stderr);
+		NOUVEAU_ERR("%d 0x%08x\n", i, fp->insn[i]);
+		fflush(stdout); fflush(stderr);
+	}
+#endif
+
+	if ((*(const uint8_t *)&le)) {
+		for (i = 0; i < fp->insn_len; i++) {
+			map[i] = fp->insn[i];
+		}
+	} else {
+		/* Weird swapping for big-endian chips */
+		for (i = 0; i < fp->insn_len; i++) {
+			map[i] = ((fp->insn[i] & 0xffff) << 16) |
+				  ((fp->insn[i] >> 16) & 0xffff);
+		}
+	}
+
+	ws->buffer_unmap(ws, fp->buffer);
+}
+
+static boolean
+nv40_fragprog_validate(struct nv40_context *nv40)
+{
+	struct nv40_fragment_program *fp = nv40->fragprog;
+	struct pipe_buffer *constbuf =
+		nv40->constbuf[PIPE_SHADER_FRAGMENT];
+	struct pipe_winsys *ws = nv40->pipe.winsys;
+	struct nouveau_stateobj *so;
+	boolean new_consts = FALSE;
+	int i;
+
+	if (fp->translated)
+		goto update_constants;
+
+	nv40->fallback_swrast &= ~NV40_NEW_FRAGPROG;
+	nv40_fragprog_translate(nv40, fp);
+	if (!fp->translated) {
+		nv40->fallback_swrast |= NV40_NEW_FRAGPROG;
+		return FALSE;
+	}
+
+	fp->buffer = ws->buffer_create(ws, 0x100, 0, fp->insn_len * 4);
+	nv40_fragprog_upload(nv40, fp);
+
+	so = so_new(4, 1);
+	so_method(so, nv40->screen->curie, NV40TCL_FP_ADDRESS, 1);
+	so_reloc (so, fp->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW | NOUVEAU_BO_OR,
+		  NV40TCL_FP_ADDRESS_DMA0, NV40TCL_FP_ADDRESS_DMA1);
+	so_method(so, nv40->screen->curie, NV40TCL_FP_CONTROL, 1);
+	so_data  (so, fp->fp_control);
+	so_ref(so, &fp->so);
+
+update_constants:
+	if (fp->nr_consts) {
+		float *map;
+		
+		map = ws->buffer_map(ws, constbuf, PIPE_BUFFER_USAGE_CPU_READ);
+		for (i = 0; i < fp->nr_consts; i++) {
+			struct nv40_fragment_program_data *fpd = &fp->consts[i];
+			uint32_t *p = &fp->insn[fpd->offset];
+			uint32_t *cb = (uint32_t *)&map[fpd->index * 4];
+
+			if (!memcmp(p, cb, 4 * sizeof(float)))
+				continue;
+			memcpy(p, cb, 4 * sizeof(float));
+			new_consts = TRUE;
+		}
+		ws->buffer_unmap(ws, constbuf);
+
+		if (new_consts)
+			nv40_fragprog_upload(nv40, fp);
+	}
+
+	if (new_consts || fp->so != nv40->state.hw[NV40_STATE_FRAGPROG]) {
+		so_ref(fp->so, &nv40->state.hw[NV40_STATE_FRAGPROG]);
+		return TRUE;
+	}
+
+	return FALSE;
+}
+
+void
+nv40_fragprog_destroy(struct nv40_context *nv40,
+		      struct nv40_fragment_program *fp)
+{
+	if (fp->insn_len)
+		FREE(fp->insn);
+}
+
+struct nv40_state_entry nv40_state_fragprog = {
+	.validate = nv40_fragprog_validate,
+	.dirty = {
+		.pipe = NV40_NEW_FRAGPROG,
+		.hw = NV40_STATE_FRAGPROG
+	}
+};
+
diff --git a/src/gallium/drivers/nv40/nv40_fragtex.c b/src/gallium/drivers/nv40/nv40_fragtex.c
new file mode 100644
index 00000000000..0227d22620d
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_fragtex.c
@@ -0,0 +1,168 @@
+#include "nv40_context.h"
+
+#define _(m,tf,ts0x,ts0y,ts0z,ts0w,ts1x,ts1y,ts1z,ts1w,sx,sy,sz,sw)            \
+{                                                                              \
+  TRUE,                                                                        \
+  PIPE_FORMAT_##m,                                                             \
+  NV40TCL_TEX_FORMAT_FORMAT_##tf,                                              \
+  (NV40TCL_TEX_SWIZZLE_S0_X_##ts0x | NV40TCL_TEX_SWIZZLE_S0_Y_##ts0y |         \
+   NV40TCL_TEX_SWIZZLE_S0_Z_##ts0z | NV40TCL_TEX_SWIZZLE_S0_W_##ts0w |         \
+   NV40TCL_TEX_SWIZZLE_S1_X_##ts1x | NV40TCL_TEX_SWIZZLE_S1_Y_##ts1y |         \
+   NV40TCL_TEX_SWIZZLE_S1_Z_##ts1z | NV40TCL_TEX_SWIZZLE_S1_W_##ts1w),         \
+  ((NV40TCL_TEX_FILTER_SIGNED_RED*sx) | (NV40TCL_TEX_FILTER_SIGNED_GREEN*sy) |       \
+   (NV40TCL_TEX_FILTER_SIGNED_BLUE*sz) | (NV40TCL_TEX_FILTER_SIGNED_ALPHA*sw))       \
+}
+
+struct nv40_texture_format {
+	boolean defined;
+	uint	pipe;
+	int     format;
+	int     swizzle;
+	int     sign;
+};
+
+static struct nv40_texture_format
+nv40_texture_formats[] = {
+	_(A8R8G8B8_UNORM, A8R8G8B8,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	_(A1R5G5B5_UNORM, A1R5G5B5,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	_(A4R4G4B4_UNORM, A4R4G4B4,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	_(R5G6B5_UNORM  , R5G6B5  ,   S1,   S1,   S1,  ONE, X, Y, Z, W, 0, 0, 0, 0),
+	_(L8_UNORM      , L8      ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
+	_(A8_UNORM      , L8      , ZERO, ZERO, ZERO,   S1, X, X, X, X, 0, 0, 0, 0),
+	_(R16_SNORM     , A16     , ZERO, ZERO,   S1,  ONE, X, X, X, Y, 1, 1, 1, 1),
+	_(I8_UNORM      , L8      ,   S1,   S1,   S1,   S1, X, X, X, X, 0, 0, 0, 0),
+	_(A8L8_UNORM    , A8L8    ,   S1,   S1,   S1,   S1, X, X, X, Y, 0, 0, 0, 0),
+	_(Z16_UNORM     , Z16     ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
+	_(Z24S8_UNORM   , Z24     ,   S1,   S1,   S1,  ONE, X, X, X, X, 0, 0, 0, 0),
+	_(DXT1_RGB      , DXT1    ,   S1,   S1,   S1,  ONE, X, Y, Z, W, 0, 0, 0, 0),
+	_(DXT1_RGBA     , DXT1    ,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	_(DXT3_RGBA     , DXT3    ,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	_(DXT5_RGBA     , DXT5    ,   S1,   S1,   S1,   S1, X, Y, Z, W, 0, 0, 0, 0),
+	{},
+};
+
+static struct nv40_texture_format *
+nv40_fragtex_format(uint pipe_format)
+{
+	struct nv40_texture_format *tf = nv40_texture_formats;
+
+	while (tf->defined) {
+		if (tf->pipe == pipe_format)
+			return tf;
+		tf++;
+	}
+
+	NOUVEAU_ERR("unknown texture format %s\n", pf_name(pipe_format));
+	return NULL;
+}
+
+
+static struct nouveau_stateobj *
+nv40_fragtex_build(struct nv40_context *nv40, int unit)
+{
+	struct nv40_sampler_state *ps = nv40->tex_sampler[unit];
+	struct nv40_miptree *nv40mt = nv40->tex_miptree[unit];
+	struct pipe_texture *pt = &nv40mt->base;
+	struct nv40_texture_format *tf;
+	struct nouveau_stateobj *so;
+	uint32_t txf, txs, txp;
+	unsigned tex_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+
+	tf = nv40_fragtex_format(pt->format);
+	if (!tf)
+		assert(0);
+
+	txf  = ps->fmt;
+	txf |= tf->format | 0x8000;
+	txf |= ((pt->last_level + 1) << NV40TCL_TEX_FORMAT_MIPMAP_COUNT_SHIFT);
+
+	if (1) /* XXX */
+		txf |= NV40TCL_TEX_FORMAT_NO_BORDER;
+
+	switch (pt->target) {
+	case PIPE_TEXTURE_CUBE:
+		txf |= NV40TCL_TEX_FORMAT_CUBIC;
+		/* fall-through */
+	case PIPE_TEXTURE_2D:
+		txf |= NV40TCL_TEX_FORMAT_DIMS_2D;
+		break;
+	case PIPE_TEXTURE_3D:
+		txf |= NV40TCL_TEX_FORMAT_DIMS_3D;
+		break;
+	case PIPE_TEXTURE_1D:
+		txf |= NV40TCL_TEX_FORMAT_DIMS_1D;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown target %d\n", pt->target);
+		return NULL;
+	}
+
+	if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		txp = 0;
+	} else {
+		txp  = nv40mt->level[0].pitch;
+		txf |= NV40TCL_TEX_FORMAT_LINEAR;
+	}
+
+	txs = tf->swizzle;
+
+	so = so_new(16, 2);
+	so_method(so, nv40->screen->curie, NV40TCL_TEX_OFFSET(unit), 8);
+	so_reloc (so, nv40mt->buffer, 0, tex_flags | NOUVEAU_BO_LOW, 0, 0);
+	so_reloc (so, nv40mt->buffer, txf, tex_flags | NOUVEAU_BO_OR,
+		  NV40TCL_TEX_FORMAT_DMA0, NV40TCL_TEX_FORMAT_DMA1);
+	so_data  (so, ps->wrap);
+	so_data  (so, NV40TCL_TEX_ENABLE_ENABLE | ps->en);
+	so_data  (so, txs);
+	so_data  (so, ps->filt | tf->sign | 0x2000 /*voodoo*/);
+	so_data  (so, (pt->width[0] << NV40TCL_TEX_SIZE0_W_SHIFT) |
+		       pt->height[0]);
+	so_data  (so, ps->bcol);
+	so_method(so, nv40->screen->curie, NV40TCL_TEX_SIZE1(unit), 1);
+	so_data  (so, (pt->depth[0] << NV40TCL_TEX_SIZE1_DEPTH_SHIFT) | txp);
+
+	return so;
+}
+
+static boolean
+nv40_fragtex_validate(struct nv40_context *nv40)
+{
+	struct nv40_fragment_program *fp = nv40->fragprog;
+	struct nv40_state *state = &nv40->state;
+	struct nouveau_stateobj *so;
+	unsigned samplers, unit;
+
+	samplers = state->fp_samplers & ~fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		so = so_new(2, 0);
+		so_method(so, nv40->screen->curie, NV40TCL_TEX_ENABLE(unit), 1);
+		so_data  (so, 0);
+		so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]);
+		state->dirty |= (1ULL << (NV40_STATE_FRAGTEX0 + unit));
+	}
+
+	samplers = nv40->dirty_samplers & fp->samplers;
+	while (samplers) {
+		unit = ffs(samplers) - 1;
+		samplers &= ~(1 << unit);
+
+		so = nv40_fragtex_build(nv40, unit);
+		so_ref(so, &nv40->state.hw[NV40_STATE_FRAGTEX0 + unit]);
+		state->dirty |= (1ULL << (NV40_STATE_FRAGTEX0 + unit));
+	}
+
+	nv40->state.fp_samplers = fp->samplers;
+	return FALSE;
+}
+
+struct nv40_state_entry nv40_state_fragtex = {
+	.validate = nv40_fragtex_validate,
+	.dirty = {
+		.pipe = NV40_NEW_SAMPLER | NV40_NEW_FRAGPROG,
+		.hw = 0
+	}
+};
+
diff --git a/src/gallium/drivers/nv40/nv40_miptree.c b/src/gallium/drivers/nv40/nv40_miptree.c
new file mode 100644
index 00000000000..b68967c07fd
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_miptree.c
@@ -0,0 +1,194 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv40_context.h"
+
+static void
+nv40_miptree_layout(struct nv40_miptree *mt)
+{
+	struct pipe_texture *pt = &mt->base;
+	uint width = pt->width[0], height = pt->height[0], depth = pt->depth[0];
+	uint offset = 0;
+	int nr_faces, l, f, pitch;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		nr_faces = 6;
+	} else
+	if (pt->target == PIPE_TEXTURE_3D) {
+		nr_faces = pt->depth[0];
+	} else {
+		nr_faces = 1;
+	}
+
+	pitch = pt->width[0];
+	for (l = 0; l <= pt->last_level; l++) {
+		pt->width[l] = width;
+		pt->height[l] = height;
+		pt->depth[l] = depth;
+		pt->nblocksx[l] = pf_get_nblocksx(&pt->block, width);
+		pt->nblocksy[l] = pf_get_nblocksy(&pt->block, height);
+
+		if (!(pt->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR))
+			pitch = pt->nblocksx[l];
+		pitch = align(pitch, 64);
+
+		mt->level[l].pitch = pitch * pt->block.size;
+		mt->level[l].image_offset =
+			CALLOC(nr_faces, sizeof(unsigned));
+
+		width  = MAX2(1, width  >> 1);
+		height = MAX2(1, height >> 1);
+		depth  = MAX2(1, depth  >> 1);
+	}
+
+	for (f = 0; f < nr_faces; f++) {
+		for (l = 0; l <= pt->last_level; l++) {
+			mt->level[l].image_offset[f] = offset;
+			offset += mt->level[l].pitch * pt->height[l];
+		}
+	}
+
+	mt->total_size = offset;
+}
+
+static struct pipe_texture *
+nv40_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv40_miptree *mt;
+
+	mt = MALLOC(sizeof(struct nv40_miptree));
+	if (!mt)
+		return NULL;
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = pscreen;
+	mt->shadow_tex = NULL;
+	mt->shadow_surface = NULL;
+
+	/* Swizzled textures must be POT */
+	if (pt->width[0] & (pt->width[0] - 1) ||
+	    pt->height[0] & (pt->height[0] - 1))
+		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+	else
+	if (pt->tex_usage & (PIPE_TEXTURE_USAGE_PRIMARY |
+	                     PIPE_TEXTURE_USAGE_DISPLAY_TARGET))
+		mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+	else {
+		switch (pt->format) {
+		/* TODO: Figure out which formats can be swizzled */
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_X8R8G8B8_UNORM:
+		/* XXX: Re-enable when SIFM size limits are fixed */
+		/*case PIPE_FORMAT_R16_SNORM:*/
+			break;
+		default:
+			mt->base.tex_usage |= NOUVEAU_TEXTURE_USAGE_LINEAR;
+		}
+	}
+
+	nv40_miptree_layout(mt);
+
+	mt->buffer = ws->buffer_create(ws, 256,
+				       PIPE_BUFFER_USAGE_PIXEL |
+				       NOUVEAU_BUFFER_USAGE_TEXTURE,
+				       mt->total_size);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+
+	return &mt->base;
+}
+
+static void
+nv40_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
+{
+	struct pipe_texture *pt = *ppt;
+	struct nv40_miptree *mt = (struct nv40_miptree *)pt;
+	int l;
+
+	*ppt = NULL;
+	if (--pt->refcount)
+		return;
+
+	pipe_buffer_reference(pscreen, &mt->buffer, NULL);
+	for (l = 0; l <= pt->last_level; l++) {
+		if (mt->level[l].image_offset)
+			FREE(mt->level[l].image_offset);
+	}
+
+	if (mt->shadow_tex) {
+		assert(mt->shadow_surface);
+		pscreen->tex_surface_release(pscreen, &mt->shadow_surface);
+		nv40_miptree_release(pscreen, &mt->shadow_tex);
+	}
+
+	FREE(mt);
+}
+
+static struct pipe_surface *
+nv40_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct nv40_miptree *mt = (struct nv40_miptree *)pt;
+	struct pipe_surface *ps;
+
+	ps = CALLOC_STRUCT(pipe_surface);
+	if (!ps)
+		return NULL;
+	pipe_texture_reference(&ps->texture, pt);
+	pipe_buffer_reference(pscreen, &ps->buffer, mt->buffer);
+	ps->format = pt->format;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = mt->level[level].pitch;
+	ps->usage = flags;
+	ps->status = PIPE_SURFACE_STATUS_DEFINED;
+	ps->refcount = 1;
+	ps->winsys = pscreen->winsys;
+	ps->face = face;
+	ps->level = level;
+	ps->zslice = zslice;
+
+	if (pt->target == PIPE_TEXTURE_CUBE) {
+		ps->offset = mt->level[level].image_offset[face];
+	} else
+	if (pt->target == PIPE_TEXTURE_3D) {
+		ps->offset = mt->level[level].image_offset[zslice];
+	} else {
+		ps->offset = mt->level[level].image_offset[0];
+	}
+
+	return ps;
+}
+
+static void
+nv40_miptree_surface_del(struct pipe_screen *pscreen,
+			 struct pipe_surface **psurface)
+{
+	struct pipe_surface *ps = *psurface;
+
+	*psurface = NULL;
+	if (--ps->refcount > 0)
+		return;
+
+	pipe_texture_reference(&ps->texture, NULL);
+	pipe_buffer_reference(pscreen, &ps->buffer, NULL);
+	FREE(ps);
+}
+
+void
+nv40_screen_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv40_miptree_create;
+	pscreen->texture_release = nv40_miptree_release;
+	pscreen->get_tex_surface = nv40_miptree_surface_new;
+	pscreen->tex_surface_release = nv40_miptree_surface_del;
+}
+
diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c
new file mode 100644
index 00000000000..57f39cfab0c
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_query.c
@@ -0,0 +1,122 @@
+#include "pipe/p_context.h"
+
+#include "nv40_context.h"
+
+struct nv40_query {
+	struct nouveau_resource *object;
+	unsigned type;
+	boolean ready;
+	uint64_t result;
+};
+
+static INLINE struct nv40_query *
+nv40_query(struct pipe_query *pipe)
+{
+	return (struct nv40_query *)pipe;
+}
+
+static struct pipe_query *
+nv40_query_create(struct pipe_context *pipe, unsigned query_type)
+{
+	struct nv40_query *q;
+
+	q = CALLOC(1, sizeof(struct nv40_query));
+	q->type = query_type;
+
+	return (struct pipe_query *)q;
+}
+
+static void
+nv40_query_destroy(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_query *q = nv40_query(pq);
+
+	if (q->object)
+		nv40->nvws->res_free(&q->object);
+	FREE(q);
+}
+
+static void
+nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_query *q = nv40_query(pq);
+
+	assert(q->type == PIPE_QUERY_OCCLUSION_COUNTER);
+
+	/* Happens when end_query() is called, then another begin_query()
+	 * without querying the result in-between.  For now we'll wait for
+	 * the existing query to notify completion, but it could be better.
+	 */
+	if (q->object) {
+		uint64 tmp;
+		pipe->get_query_result(pipe, pq, 1, &tmp);
+	}
+
+	if (nv40->nvws->res_alloc(nv40->screen->query_heap, 1, NULL, &q->object))
+		assert(0);
+	nv40->nvws->notifier_reset(nv40->screen->query, q->object->start);
+
+	BEGIN_RING(curie, NV40TCL_QUERY_RESET, 1);
+	OUT_RING  (1);
+	BEGIN_RING(curie, NV40TCL_QUERY_UNK17CC, 1);
+	OUT_RING  (1);
+
+	q->ready = FALSE;
+}
+
+static void
+nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_query *q = nv40_query(pq);
+
+	BEGIN_RING(curie, NV40TCL_QUERY_GET, 1);
+	OUT_RING  ((0x01 << NV40TCL_QUERY_GET_UNK24_SHIFT) |
+		   ((q->object->start * 32) << NV40TCL_QUERY_GET_OFFSET_SHIFT));
+	FIRE_RING(NULL);
+}
+
+static boolean
+nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq,
+		  boolean wait, uint64 *result)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_query *q = nv40_query(pq);
+	struct nouveau_winsys *nvws = nv40->nvws;
+
+	assert(q->object && q->type == PIPE_QUERY_OCCLUSION_COUNTER);
+
+	if (!q->ready) {
+		unsigned status;
+
+		status = nvws->notifier_status(nv40->screen->query,
+					       q->object->start);
+		if (status != NV_NOTIFY_STATE_STATUS_COMPLETED) {
+			if (wait == FALSE)
+				return FALSE;
+			nvws->notifier_wait(nv40->screen->query, q->object->start,
+					    NV_NOTIFY_STATE_STATUS_COMPLETED,
+					    0);
+		}
+
+		q->result = nvws->notifier_retval(nv40->screen->query,
+						  q->object->start);
+		q->ready = TRUE;
+		nvws->res_free(&q->object);
+	}
+
+	*result = q->result;
+	return TRUE;
+}
+
+void
+nv40_init_query_functions(struct nv40_context *nv40)
+{
+	nv40->pipe.create_query = nv40_query_create;
+	nv40->pipe.destroy_query = nv40_query_destroy;
+	nv40->pipe.begin_query = nv40_query_begin;
+	nv40->pipe.end_query = nv40_query_end;
+	nv40->pipe.get_query_result = nv40_query_result;
+}
diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c
new file mode 100644
index 00000000000..25c78682961
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_screen.c
@@ -0,0 +1,365 @@
+#include "pipe/p_screen.h"
+
+#include "nv40_context.h"
+#include "nv40_screen.h"
+
+#define NV4X_GRCLASS4097_CHIPSETS 0x00000baf
+#define NV4X_GRCLASS4497_CHIPSETS 0x00005450
+#define NV6X_GRCLASS4497_CHIPSETS 0x00000088
+
+static const char *
+nv40_screen_get_name(struct pipe_screen *pscreen)
+{
+	struct nv40_screen *screen = nv40_screen(pscreen);
+	struct nouveau_device *dev = screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv40_screen_get_vendor(struct pipe_screen *pscreen)
+{
+	return "nouveau";
+}
+
+static int
+nv40_screen_get_param(struct pipe_screen *pscreen, int param)
+{
+	struct nv40_screen *screen = nv40_screen(pscreen);
+
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 16;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 1;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 1;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 1;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 1;
+	case PIPE_CAP_POINT_SPRITE:
+		return 1;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 4;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 1;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 1;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 13;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 10;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 13;
+	case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
+	case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
+		return 1;
+	case NOUVEAU_CAP_HW_VTXBUF:
+		return 1;
+	case NOUVEAU_CAP_HW_IDXBUF:
+		if (screen->curie->grclass == NV40TCL)
+			return 1;
+		return 0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv40_screen_get_paramf(struct pipe_screen *pscreen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 10.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 64.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 16.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 16.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static boolean
+nv40_screen_surface_format_supported(struct pipe_screen *pscreen,
+				     enum pipe_format format,
+				     enum pipe_texture_target target,
+				     unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM: 
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_A4R4G4B4_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM:
+		case PIPE_FORMAT_R16_SNORM:
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
+		case PIPE_FORMAT_A8L8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_DXT1_RGB:
+		case PIPE_FORMAT_DXT1_RGBA:
+		case PIPE_FORMAT_DXT3_RGBA:
+		case PIPE_FORMAT_DXT5_RGBA:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static void *
+nv40_surface_map(struct pipe_screen *screen, struct pipe_surface *surface,
+		 unsigned flags )
+{
+	struct pipe_winsys	*ws = screen->winsys;
+	struct pipe_surface	*surface_to_map;
+	void			*map;
+
+	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		struct nv40_miptree *mt = (struct nv40_miptree *)surface->texture;
+
+		if (!mt->shadow_tex) {
+			unsigned old_tex_usage = surface->texture->tex_usage;
+			surface->texture->tex_usage = NOUVEAU_TEXTURE_USAGE_LINEAR;
+			mt->shadow_tex = screen->texture_create(screen, surface->texture);
+			surface->texture->tex_usage = old_tex_usage;
+
+			assert(mt->shadow_tex->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR);
+			mt->shadow_surface = screen->get_tex_surface
+			(
+				screen, mt->shadow_tex,
+				surface->face, surface->level, surface->zslice,
+				surface->usage
+			);
+		}
+
+		surface_to_map = mt->shadow_surface;
+	}
+	else
+		surface_to_map = surface;
+
+	assert(surface_to_map);
+
+	map = ws->buffer_map(ws, surface_to_map->buffer, flags);
+	if (!map)
+		return NULL;
+
+	return map + surface_to_map->offset;
+}
+
+static void
+nv40_surface_unmap(struct pipe_screen *screen, struct pipe_surface *surface)
+{
+	struct pipe_winsys	*ws = screen->winsys;
+	struct pipe_surface	*surface_to_unmap;
+
+	/* TODO: Copy from shadow just before push buffer is flushed instead.
+	         There are probably some programs that map/unmap excessively
+	         before rendering. */
+	if (!(surface->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		struct nv40_miptree *mt = (struct nv40_miptree *)surface->texture;
+
+		assert(mt->shadow_tex);
+
+		surface_to_unmap = mt->shadow_surface;
+	}
+	else
+		surface_to_unmap = surface;
+
+	assert(surface_to_unmap);
+
+	ws->buffer_unmap(ws, surface_to_unmap->buffer);
+
+	if (surface_to_unmap != surface) {
+		struct nv40_screen *nvscreen = nv40_screen(screen);
+
+		nvscreen->nvws->surface_copy(nvscreen->nvws,
+		                             surface, 0, 0,
+		                             surface_to_unmap, 0, 0,
+		                             surface->width, surface->height);
+	}
+}
+
+static void
+nv40_screen_destroy(struct pipe_screen *pscreen)
+{
+	struct nv40_screen *screen = nv40_screen(pscreen);
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	nvws->res_free(&screen->vp_exec_heap);
+	nvws->res_free(&screen->vp_data_heap);
+	nvws->res_free(&screen->query_heap);
+	nvws->notifier_free(&screen->query);
+	nvws->notifier_free(&screen->sync);
+	nvws->grobj_free(&screen->curie);
+
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv40_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv40_screen *screen = CALLOC_STRUCT(nv40_screen);
+	struct nouveau_stateobj *so;
+	unsigned curie_class;
+	unsigned chipset = nvws->channel->device->chipset;
+	int ret;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	/* 3D object */
+	switch (chipset & 0xf0) {
+	case 0x40:
+		if (NV4X_GRCLASS4097_CHIPSETS & (1 << (chipset & 0x0f)))
+			curie_class = NV40TCL;
+		else
+		if (NV4X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f)))
+			curie_class = NV44TCL;
+		break;
+	case 0x60:
+		if (NV6X_GRCLASS4497_CHIPSETS & (1 << (chipset & 0x0f)))
+			curie_class = NV44TCL;
+		break;
+	default:
+		break;
+	}
+
+	if (!curie_class) {
+		NOUVEAU_ERR("Unknown nv4x chipset: nv%02x\n", chipset);
+		return NULL;
+	}
+
+	ret = nvws->grobj_alloc(nvws, curie_class, &screen->curie);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		return FALSE;
+	}
+
+	/* Notifier for sync purposes */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv40_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Query objects */
+	ret = nvws->notifier_alloc(nvws, 32, &screen->query);
+	if (ret) {
+		NOUVEAU_ERR("Error initialising query objects: %d\n", ret);
+		nv40_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	ret = nvws->res_init(&screen->query_heap, 0, 32);
+	if (ret) {
+		NOUVEAU_ERR("Error initialising query object heap: %d\n", ret);
+		nv40_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Vtxprog resources */
+	if (nvws->res_init(&screen->vp_exec_heap, 0, 512) ||
+	    nvws->res_init(&screen->vp_data_heap, 0, 256)) {
+		nv40_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Static curie initialisation */
+	so = so_new(128, 0);
+	so_method(so, screen->curie, NV40TCL_DMA_NOTIFY, 1);
+	so_data  (so, screen->sync->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_TEXTURE0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->gart->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_COLOR1, 1);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_COLOR0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_VTXBUF0, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->gart->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_FENCE, 2);
+	so_data  (so, 0);
+	so_data  (so, screen->query->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_UNK01AC, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->vram->handle);
+	so_method(so, screen->curie, NV40TCL_DMA_COLOR2, 2);
+	so_data  (so, nvws->channel->vram->handle);
+	so_data  (so, nvws->channel->vram->handle);
+
+	so_method(so, screen->curie, 0x1ea4, 3);
+	so_data  (so, 0x00000010);
+	so_data  (so, 0x01000100);
+	so_data  (so, 0xff800006);
+
+	/* vtxprog output routing */
+	so_method(so, screen->curie, 0x1fc4, 1);
+	so_data  (so, 0x06144321);
+	so_method(so, screen->curie, 0x1fc8, 2);
+	so_data  (so, 0xedcba987);
+	so_data  (so, 0x00000021);
+	so_method(so, screen->curie, 0x1fd0, 1);
+	so_data  (so, 0x00171615);
+	so_method(so, screen->curie, 0x1fd4, 1);
+	so_data  (so, 0x001b1a19);
+
+	so_method(so, screen->curie, 0x1ef8, 1);
+	so_data  (so, 0x0020ffff);
+	so_method(so, screen->curie, 0x1d64, 1);
+	so_data  (so, 0x00d30000);
+	so_method(so, screen->curie, 0x1e94, 1);
+	so_data  (so, 0x00000001);
+
+	so_emit(nvws, so);
+	so_ref(NULL, &so);
+	nvws->push_flush(nvws, 0, NULL);
+
+	screen->pipe.winsys = ws;
+	screen->pipe.destroy = nv40_screen_destroy;
+
+	screen->pipe.get_name = nv40_screen_get_name;
+	screen->pipe.get_vendor = nv40_screen_get_vendor;
+	screen->pipe.get_param = nv40_screen_get_param;
+	screen->pipe.get_paramf = nv40_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv40_screen_surface_format_supported;
+
+	screen->pipe.surface_map = nv40_surface_map;
+	screen->pipe.surface_unmap = nv40_surface_unmap;
+
+	nv40_screen_init_miptree_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h
new file mode 100644
index 00000000000..c04a1275a00
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_screen.h
@@ -0,0 +1,35 @@
+#ifndef __NV40_SCREEN_H__
+#define __NV40_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv40_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+
+	unsigned cur_pctx;
+
+	/* HW graphics objects */
+	struct nouveau_grobj *curie;
+	struct nouveau_notifier *sync;
+
+	/* Query object resources */
+	struct nouveau_notifier *query;
+	struct nouveau_resource *query_heap;
+
+	/* Vtxprog resources */
+	struct nouveau_resource *vp_exec_heap;
+	struct nouveau_resource *vp_data_heap;
+
+	/* Current 3D state of channel */
+	struct nouveau_stateobj *state[NV40_STATE_MAX];
+};
+
+static INLINE struct nv40_screen *
+nv40_screen(struct pipe_screen *screen)
+{
+	return (struct nv40_screen *)screen;
+}
+
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nv40_shader.h b/src/gallium/drivers/nv40/nv40_shader.h
index 584f4c23e08..854dccf5486 100644
--- a/src/mesa/drivers/dri/nouveau/nv40_shader.h
+++ b/src/gallium/drivers/nv40/nv40_shader.h
@@ -90,8 +90,8 @@
 #    define NV40_VP_INST_OP_ADD                                             0x03
 #    define NV40_VP_INST_OP_MAD                                             0x04
 #    define NV40_VP_INST_OP_DP3                                             0x05
-#    define NV40_VP_INST_OP_DP4                                             0x07
 #    define NV40_VP_INST_OP_DPH                                             0x06
+#    define NV40_VP_INST_OP_DP4                                             0x07
 #    define NV40_VP_INST_OP_DST                                             0x08
 #    define NV40_VP_INST_OP_MIN                                             0x09
 #    define NV40_VP_INST_OP_MAX                                             0x0A
@@ -109,9 +109,11 @@
 #    define NV40_VP_INST_OP_SSG                                             0x16
 #    define NV40_VP_INST_OP_ARR                                             0x17
 #    define NV40_VP_INST_OP_ARA                                             0x18
-#    define NV40_VP_INST_OP_TXWHAT                                          0x19
+#    define NV40_VP_INST_OP_TXL                                             0x19
 #define NV40_VP_INST_SCA_OPCODE_SHIFT                                         27
 #define NV40_VP_INST_SCA_OPCODE_MASK                                (0x1F << 27)
+#    define NV40_VP_INST_OP_NOP                                             0x00
+#    define NV40_VP_INST_OP_MOV                                             0x01
 #    define NV40_VP_INST_OP_RCP                                             0x02
 #    define NV40_VP_INST_OP_RCC                                             0x03
 #    define NV40_VP_INST_OP_RSQ                                             0x04
@@ -292,9 +294,9 @@
 //== Opcode / Destination selection ==
 #define NV40_FP_OP_PROGRAM_END                                          (1 << 0)
 #define NV40_FP_OP_OUT_REG_SHIFT                                               1
-#define NV40_FP_OP_OUT_REG_MASK                                        (31 << 1)
+#define NV40_FP_OP_OUT_REG_MASK                                        (63 << 1)
 /* Needs to be set when writing outputs to get expected result.. */
-#define NV40_FP_OP_UNK0_7                                               (1 << 7)
+#define NV40_FP_OP_OUT_REG_HALF                                         (1 << 7)
 #define NV40_FP_OP_COND_WRITE_ENABLE                                    (1 << 8)
 #define NV40_FP_OP_OUTMASK_SHIFT                                               9
 #define NV40_FP_OP_OUTMASK_MASK                                       (0xF << 9)
@@ -353,6 +355,8 @@
 #        define NV40_FP_OP_OPCODE_RCP                                       0x1A
 #        define NV40_FP_OP_OPCODE_EX2                                       0x1C
 #        define NV40_FP_OP_OPCODE_LG2                                       0x1D
+#        define NV40_FP_OP_OPCODE_STR                                       0x20
+#        define NV40_FP_OP_OPCODE_SFL                                       0x21
 #        define NV40_FP_OP_OPCODE_COS                                       0x22
 #        define NV40_FP_OP_OPCODE_SIN                                       0x23
 #        define NV40_FP_OP_OPCODE_PK2H                                      0x24
@@ -365,6 +369,7 @@
 #        define NV40_FP_OP_OPCODE_TXL                                       0x2F
 #        define NV40_FP_OP_OPCODE_TXB                                       0x31
 #        define NV40_FP_OP_OPCODE_DIV                                       0x3A
+#        define NV40_FP_OP_OPCODE_UNK_LIT                                   0x3C
 /* The use of these instructions appears to be indicated by bit 31 of DWORD 2.*/
 #        define NV40_FP_OP_BRA_OPCODE_BRK                                    0x0
 #        define NV40_FP_OP_BRA_OPCODE_CAL                                    0x1
@@ -401,6 +406,13 @@
 #define NV40_FP_OP_OPCODE_IS_BRANCH                                      (1<<31)
 #define NV40_FP_OP_DST_SCALE_SHIFT                                            28
 #define NV40_FP_OP_DST_SCALE_MASK                                      (3 << 28)
+#define NV40_FP_OP_DST_SCALE_1X                                                0
+#define NV40_FP_OP_DST_SCALE_2X                                                1
+#define NV40_FP_OP_DST_SCALE_4X                                                2
+#define NV40_FP_OP_DST_SCALE_8X                                                3
+#define NV40_FP_OP_DST_SCALE_INV_2X                                            5
+#define NV40_FP_OP_DST_SCALE_INV_4X                                            6
+#define NV40_FP_OP_DST_SCALE_INV_8X                                            7
 
 /* SRC1 LOOP */
 #define NV40_FP_OP_LOOP_INCR_SHIFT                                            19
@@ -446,8 +458,8 @@
 #        define NV40_FP_REG_TYPE_INPUT                                         1
 #        define NV40_FP_REG_TYPE_CONST                                         2
 #define NV40_FP_REG_SRC_SHIFT                                                  2
-#define NV40_FP_REG_SRC_MASK                                           (31 << 2)
-#define NV40_FP_REG_UNK_0                                               (1 << 8)
+#define NV40_FP_REG_SRC_MASK                                           (63 << 2)
+#define NV40_FP_REG_SRC_HALF                                            (1 << 8)
 #define NV40_FP_REG_SWZ_ALL_SHIFT                                              9
 #define NV40_FP_REG_SWZ_ALL_MASK                                      (255 << 9)
 #define NV40_FP_REG_SWZ_X_SHIFT                                                9
@@ -464,4 +476,81 @@
 #        define NV40_FP_SWIZZLE_W                                              3
 #define NV40_FP_REG_NEGATE                                             (1 << 17)
 
+#ifndef NV40_SHADER_NO_FUCKEDNESS
+#define NV40SR_NONE	0
+#define NV40SR_OUTPUT	1
+#define NV40SR_INPUT	2
+#define NV40SR_TEMP	3
+#define NV40SR_CONST	4
+
+struct nv40_sreg {
+	int type;
+	int index;
+
+	int dst_scale;
+
+	int negate;
+	int abs;
+	int swz[4];
+
+	int cc_update;
+	int cc_update_reg;
+	int cc_test;
+	int cc_test_reg;
+	int cc_swz[4];
+};
+
+static INLINE struct nv40_sreg
+nv40_sr(int type, int index)
+{
+	struct nv40_sreg temp = {
+		.type = type,
+		.index = index,
+		.dst_scale = DEF_SCALE,
+		.abs = 0,
+		.negate = 0,
+		.swz = { 0, 1, 2, 3 },
+		.cc_update = 0,
+		.cc_update_reg = 0,
+		.cc_test = DEF_CTEST,
+		.cc_test_reg = 0,
+		.cc_swz = { 0, 1, 2, 3 },
+	};
+	return temp;
+}
+
+static INLINE struct nv40_sreg
+nv40_sr_swz(struct nv40_sreg src, int x, int y, int z, int w)
+{
+	struct nv40_sreg dst = src;
+
+	dst.swz[SWZ_X] = src.swz[x];
+	dst.swz[SWZ_Y] = src.swz[y];
+	dst.swz[SWZ_Z] = src.swz[z];
+	dst.swz[SWZ_W] = src.swz[w];
+	return dst;
+}
+
+static INLINE struct nv40_sreg
+nv40_sr_neg(struct nv40_sreg src)
+{
+	src.negate = !src.negate;
+	return src;
+}
+
+static INLINE struct nv40_sreg
+nv40_sr_abs(struct nv40_sreg src)
+{
+	src.abs = 1;
+	return src;
+}
+
+static INLINE struct nv40_sreg
+nv40_sr_scale(struct nv40_sreg src, int scale)
+{
+	src.dst_scale = scale;
+	return src;
+}
+#endif
+
 #endif
diff --git a/src/gallium/drivers/nv40/nv40_state.c b/src/gallium/drivers/nv40/nv40_state.c
new file mode 100644
index 00000000000..255c4b294d1
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state.c
@@ -0,0 +1,740 @@
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "draw/draw_context.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv40_context.h"
+#include "nv40_state.h"
+
+static void *
+nv40_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nouveau_grobj *curie = nv40->screen->curie;
+	struct nv40_blend_state *bso = CALLOC(1, sizeof(*bso));
+	struct nouveau_stateobj *so = so_new(16, 0);
+
+	if (cso->blend_enable) {
+		so_method(so, curie, NV40TCL_BLEND_ENABLE, 3);
+		so_data  (so, 1);
+		so_data  (so, (nvgl_blend_func(cso->alpha_src_factor) << 16) |
+			       nvgl_blend_func(cso->rgb_src_factor));
+		so_data  (so, nvgl_blend_func(cso->alpha_dst_factor) << 16 |
+			      nvgl_blend_func(cso->rgb_dst_factor));
+		so_method(so, curie, NV40TCL_BLEND_EQUATION, 1);
+		so_data  (so, nvgl_blend_eqn(cso->alpha_func) << 16 |
+			      nvgl_blend_eqn(cso->rgb_func));
+	} else {
+		so_method(so, curie, NV40TCL_BLEND_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, curie, NV40TCL_COLOR_MASK, 1);
+	so_data  (so, (((cso->colormask & PIPE_MASK_A) ? (0x01 << 24) : 0) |
+		       ((cso->colormask & PIPE_MASK_R) ? (0x01 << 16) : 0) |
+		       ((cso->colormask & PIPE_MASK_G) ? (0x01 <<  8) : 0) |
+		       ((cso->colormask & PIPE_MASK_B) ? (0x01 <<  0) : 0)));
+
+	if (cso->logicop_enable) {
+		so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 2);
+		so_data  (so, 1);
+		so_data  (so, nvgl_logicop_func(cso->logicop_func));
+	} else {
+		so_method(so, curie, NV40TCL_COLOR_LOGIC_OP_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, curie, NV40TCL_DITHER_ENABLE, 1);
+	so_data  (so, cso->dither ? 1 : 0);
+
+	so_ref(so, &bso->so);
+	bso->pipe = *cso;
+	return (void *)bso;
+}
+
+static void
+nv40_blend_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->blend = hwcso;
+	nv40->dirty |= NV40_NEW_BLEND;
+}
+
+static void
+nv40_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_blend_state *bso = hwcso;
+
+	so_ref(NULL, &bso->so);
+	FREE(bso);
+}
+
+
+static INLINE unsigned
+wrap_mode(unsigned wrap) {
+	unsigned ret;
+
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		ret = NV40TCL_TEX_WRAP_S_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		ret = NV40TCL_TEX_WRAP_S_MIRRORED_REPEAT;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		ret = NV40TCL_TEX_WRAP_S_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_CLAMP:
+		ret = NV40TCL_TEX_WRAP_S_CLAMP;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+		ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_EDGE;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+		ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP_TO_BORDER;
+		break;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+		ret = NV40TCL_TEX_WRAP_S_MIRROR_CLAMP;
+		break;
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		ret = NV40TCL_TEX_WRAP_S_REPEAT;
+		break;
+	}
+
+	return ret >> NV40TCL_TEX_WRAP_S_SHIFT;
+}
+
+static void *
+nv40_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	struct nv40_sampler_state *ps;
+	uint32_t filter = 0;
+
+	ps = MALLOC(sizeof(struct nv40_sampler_state));
+
+	ps->fmt = 0;
+	if (!cso->normalized_coords)
+		ps->fmt |= NV40TCL_TEX_FORMAT_RECT;
+
+	ps->wrap = ((wrap_mode(cso->wrap_s) << NV40TCL_TEX_WRAP_S_SHIFT) |
+		    (wrap_mode(cso->wrap_t) << NV40TCL_TEX_WRAP_T_SHIFT) |
+		    (wrap_mode(cso->wrap_r) << NV40TCL_TEX_WRAP_R_SHIFT));
+
+	ps->en = 0;
+	if (cso->max_anisotropy >= 2.0) {
+		/* no idea, binary driver sets it, works without it.. meh.. */
+		ps->wrap |= (1 << 5);
+
+		if (cso->max_anisotropy >= 16.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_16X;
+		} else
+		if (cso->max_anisotropy >= 12.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_12X;
+		} else
+		if (cso->max_anisotropy >= 10.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_10X;
+		} else
+		if (cso->max_anisotropy >= 8.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_8X;
+		} else
+		if (cso->max_anisotropy >= 6.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_6X;
+		} else
+		if (cso->max_anisotropy >= 4.0) {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_4X;
+		} else {
+			ps->en |= NV40TCL_TEX_ENABLE_ANISO_2X;
+		}
+	}
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		filter |= NV40TCL_TEX_FILTER_MAG_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		filter |= NV40TCL_TEX_FILTER_MAG_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_NEAREST;
+			break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV40TCL_TEX_FILTER_MIN_LINEAR_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV40TCL_TEX_FILTER_MIN_LINEAR;
+			break;
+		}
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		switch (cso->min_mip_filter) {
+		case PIPE_TEX_MIPFILTER_NEAREST:
+			filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_NEAREST;
+		break;
+		case PIPE_TEX_MIPFILTER_LINEAR:
+			filter |= NV40TCL_TEX_FILTER_MIN_NEAREST_MIPMAP_LINEAR;
+			break;
+		case PIPE_TEX_MIPFILTER_NONE:
+		default:
+			filter |= NV40TCL_TEX_FILTER_MIN_NEAREST;
+			break;
+		}
+		break;
+	}
+
+	ps->filt = filter;
+
+	{
+		float limit;
+
+		limit = CLAMP(cso->lod_bias, -16.0, 15.0);
+		ps->filt |= (int)(cso->lod_bias * 256.0) & 0x1fff;
+
+		limit = CLAMP(cso->max_lod, 0.0, 15.0);
+		ps->en |= (int)(limit * 256.0) << 7;
+
+		limit = CLAMP(cso->min_lod, 0.0, 15.0);
+		ps->en |= (int)(limit * 256.0) << 19;
+	}
+
+
+	if (cso->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+		switch (cso->compare_func) {
+		case PIPE_FUNC_NEVER:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NEVER;
+			break;
+		case PIPE_FUNC_GREATER:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GREATER;
+			break;
+		case PIPE_FUNC_EQUAL:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_EQUAL;
+			break;
+		case PIPE_FUNC_GEQUAL:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_GEQUAL;
+			break;
+		case PIPE_FUNC_LESS:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LESS;
+			break;
+		case PIPE_FUNC_NOTEQUAL:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_NOTEQUAL;
+			break;
+		case PIPE_FUNC_LEQUAL:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_LEQUAL;
+			break;
+		case PIPE_FUNC_ALWAYS:
+			ps->wrap |= NV40TCL_TEX_WRAP_RCOMP_ALWAYS;
+			break;
+		default:
+			break;
+		}
+	}
+
+	ps->bcol = ((float_to_ubyte(cso->border_color[3]) << 24) |
+		    (float_to_ubyte(cso->border_color[0]) << 16) |
+		    (float_to_ubyte(cso->border_color[1]) <<  8) |
+		    (float_to_ubyte(cso->border_color[2]) <<  0));
+
+	return (void *)ps;
+}
+
+static void
+nv40_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		nv40->tex_sampler[unit] = sampler[unit];
+		nv40->dirty_samplers |= (1 << unit);
+	}
+
+	for (unit = nr; unit < nv40->nr_samplers; unit++) {
+		nv40->tex_sampler[unit] = NULL;
+		nv40->dirty_samplers |= (1 << unit);
+	}
+
+	nv40->nr_samplers = nr;
+	nv40->dirty |= NV40_NEW_SAMPLER;
+}
+
+static void
+nv40_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv40_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **miptree)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	unsigned unit;
+
+	for (unit = 0; unit < nr; unit++) {
+		pipe_texture_reference((struct pipe_texture **)
+				       &nv40->tex_miptree[unit], miptree[unit]);
+		nv40->dirty_samplers |= (1 << unit);
+	}
+
+	for (unit = nr; unit < nv40->nr_textures; unit++) {
+		pipe_texture_reference((struct pipe_texture **)
+				       &nv40->tex_miptree[unit], NULL);
+		nv40->dirty_samplers |= (1 << unit);
+	}
+
+	nv40->nr_textures = nr;
+	nv40->dirty |= NV40_NEW_SAMPLER;
+}
+
+static void *
+nv40_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_rasterizer_state *rsso = CALLOC(1, sizeof(*rsso));
+	struct nouveau_stateobj *so = so_new(32, 0);
+	struct nouveau_grobj *curie = nv40->screen->curie;
+
+	/*XXX: ignored:
+	 * 	light_twoside
+	 * 	point_smooth -nohw
+	 * 	multisample
+	 */
+
+	so_method(so, curie, NV40TCL_SHADE_MODEL, 1);
+	so_data  (so, cso->flatshade ? NV40TCL_SHADE_MODEL_FLAT :
+				       NV40TCL_SHADE_MODEL_SMOOTH);
+
+	so_method(so, curie, NV40TCL_LINE_WIDTH, 2);
+	so_data  (so, (unsigned char)(cso->line_width * 8.0) & 0xff);
+	so_data  (so, cso->line_smooth ? 1 : 0);
+	so_method(so, curie, NV40TCL_LINE_STIPPLE_ENABLE, 2);
+	so_data  (so, cso->line_stipple_enable ? 1 : 0);
+	so_data  (so, (cso->line_stipple_pattern << 16) |
+		       cso->line_stipple_factor);
+
+	so_method(so, curie, NV40TCL_POINT_SIZE, 1);
+	so_data  (so, fui(cso->point_size));
+
+	so_method(so, curie, NV40TCL_POLYGON_MODE_FRONT, 6);
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV40TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV40TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV40TCL_CULL_FACE_BACK);
+			break;
+		}
+		so_data(so, NV40TCL_FRONT_FACE_CCW);
+	} else {
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV40TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV40TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV40TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV40TCL_CULL_FACE_BACK);
+			break;
+		}
+		so_data(so, NV40TCL_FRONT_FACE_CW);
+	}
+	so_data(so, cso->poly_smooth ? 1 : 0);
+	so_data(so, (cso->cull_mode != PIPE_WINDING_NONE) ? 1 : 0);
+
+	so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1);
+	so_data  (so, cso->poly_stipple_enable ? 1 : 0);
+
+	so_method(so, curie, NV40TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if (cso->offset_cw || cso->offset_ccw) {
+		so_method(so, curie, NV40TCL_POLYGON_OFFSET_FACTOR, 2);
+		so_data  (so, fui(cso->offset_scale));
+		so_data  (so, fui(cso->offset_units * 2));
+	}
+
+	so_method(so, curie, NV40TCL_POINT_SPRITE, 1);
+	if (cso->point_sprite) {
+		unsigned psctl = (1 << 0), i;
+
+		for (i = 0; i < 8; i++) {
+			if (cso->sprite_coord_mode[i] != PIPE_SPRITE_COORD_NONE)
+				psctl |= (1 << (8 + i));
+		}
+
+		so_data(so, psctl);
+	} else {
+		so_data(so, 0);
+	}
+
+	so_ref(so, &rsso->so);
+	rsso->pipe = *cso;
+	return (void *)rsso;
+}
+
+static void
+nv40_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->rasterizer = hwcso;
+	nv40->dirty |= NV40_NEW_RAST;
+	nv40->draw_dirty |= NV40_NEW_RAST;
+}
+
+static void
+nv40_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_rasterizer_state *rsso = hwcso;
+
+	so_ref(NULL, &rsso->so);
+	FREE(rsso);
+}
+
+static void *
+nv40_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_zsa_state *zsaso = CALLOC(1, sizeof(*zsaso));
+	struct nouveau_stateobj *so = so_new(32, 0);
+	struct nouveau_grobj *curie = nv40->screen->curie;
+
+	so_method(so, curie, NV40TCL_DEPTH_FUNC, 3);
+	so_data  (so, nvgl_comparison_op(cso->depth.func));
+	so_data  (so, cso->depth.writemask ? 1 : 0);
+	so_data  (so, cso->depth.enabled ? 1 : 0);
+
+	so_method(so, curie, NV40TCL_ALPHA_TEST_ENABLE, 3);
+	so_data  (so, cso->alpha.enabled ? 1 : 0);
+	so_data  (so, nvgl_comparison_op(cso->alpha.func));
+	so_data  (so, float_to_ubyte(cso->alpha.ref));
+
+	if (cso->stencil[0].enabled) {
+		so_method(so, curie, NV40TCL_STENCIL_FRONT_ENABLE, 8);
+		so_data  (so, cso->stencil[0].enabled ? 1 : 0);
+		so_data  (so, cso->stencil[0].write_mask);
+		so_data  (so, nvgl_comparison_op(cso->stencil[0].func));
+		so_data  (so, cso->stencil[0].ref_value);
+		so_data  (so, cso->stencil[0].value_mask);
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zpass_op));
+	} else {
+		so_method(so, curie, NV40TCL_STENCIL_FRONT_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	if (cso->stencil[1].enabled) {
+		so_method(so, curie, NV40TCL_STENCIL_BACK_ENABLE, 8);
+		so_data  (so, cso->stencil[1].enabled ? 1 : 0);
+		so_data  (so, cso->stencil[1].write_mask);
+		so_data  (so, nvgl_comparison_op(cso->stencil[1].func));
+		so_data  (so, cso->stencil[1].ref_value);
+		so_data  (so, cso->stencil[1].value_mask);
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zpass_op));
+	} else {
+		so_method(so, curie, NV40TCL_STENCIL_BACK_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_ref(so, &zsaso->so);
+	zsaso->pipe = *cso;
+	return (void *)zsaso;
+}
+
+static void
+nv40_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->zsa = hwcso;
+	nv40->dirty |= NV40_NEW_ZSA;
+}
+
+static void
+nv40_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_zsa_state *zsaso = hwcso;
+
+	so_ref(NULL, &zsaso->so);
+	FREE(zsaso);
+}
+
+static void *
+nv40_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_vertex_program *vp;
+
+	vp = CALLOC(1, sizeof(struct nv40_vertex_program));
+	vp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	vp->draw = draw_create_vertex_shader(nv40->draw, &vp->pipe);
+
+	return (void *)vp;
+}
+
+static void
+nv40_vp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->vertprog = hwcso;
+	nv40->dirty |= NV40_NEW_VERTPROG;
+	nv40->draw_dirty |= NV40_NEW_VERTPROG;
+}
+
+static void
+nv40_vp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_vertex_program *vp = hwcso;
+
+	draw_delete_vertex_shader(nv40->draw, vp->draw);
+	nv40_vertprog_destroy(nv40, vp);
+	FREE((void*)vp->pipe.tokens);
+	FREE(vp);
+}
+
+static void *
+nv40_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv40_fragment_program *fp;
+
+	fp = CALLOC(1, sizeof(struct nv40_fragment_program));
+	fp->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+
+	tgsi_scan_shader(fp->pipe.tokens, &fp->info);
+
+	return (void *)fp;
+}
+
+static void
+nv40_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->fragprog = hwcso;
+	nv40->dirty |= NV40_NEW_FRAGPROG;
+}
+
+static void
+nv40_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nv40_fragment_program *fp = hwcso;
+
+	nv40_fragprog_destroy(nv40, fp);
+	FREE((void*)fp->pipe.tokens);
+	FREE(fp);
+}
+
+static void
+nv40_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->blend_colour = *bcol;
+	nv40->dirty |= NV40_NEW_BCOL;
+}
+
+static void
+nv40_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->clip = *clip;
+	nv40->dirty |= NV40_NEW_UCP;
+	nv40->draw_dirty |= NV40_NEW_UCP;
+}
+
+static void
+nv40_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->constbuf[shader] = buf->buffer;
+	nv40->constbuf_nr[shader] = buf->size / (4 * sizeof(float));
+
+	if (shader == PIPE_SHADER_VERTEX) {
+		nv40->dirty |= NV40_NEW_VERTPROG;
+	} else
+	if (shader == PIPE_SHADER_FRAGMENT) {
+		nv40->dirty |= NV40_NEW_FRAGPROG;
+	}
+}
+
+static void
+nv40_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->framebuffer = *fb;
+	nv40->dirty |= NV40_NEW_FB;
+}
+
+static void
+nv40_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	memcpy(nv40->stipple, stipple->stipple, 4 * 32);
+	nv40->dirty |= NV40_NEW_STIPPLE;
+}
+
+static void
+nv40_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->scissor = *s;
+	nv40->dirty |= NV40_NEW_SCISSOR;
+}
+
+static void
+nv40_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->viewport = *vpt;
+	nv40->dirty |= NV40_NEW_VIEWPORT;
+	nv40->draw_dirty |= NV40_NEW_VIEWPORT;
+}
+
+static void
+nv40_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	memcpy(nv40->vtxbuf, vb, sizeof(*vb) * count);
+	nv40->vtxbuf_nr = count;
+
+	nv40->dirty |= NV40_NEW_ARRAYS;
+	nv40->draw_dirty |= NV40_NEW_ARRAYS;
+}
+
+static void
+nv40_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	memcpy(nv40->vtxelt, ve, sizeof(*ve) * count);
+	nv40->vtxelt_nr = count;
+
+	nv40->dirty |= NV40_NEW_ARRAYS;
+	nv40->draw_dirty |= NV40_NEW_ARRAYS;
+}
+
+static void
+nv40_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+
+	nv40->edgeflags = bitfield;
+	nv40->dirty |= NV40_NEW_ARRAYS;
+	nv40->draw_dirty |= NV40_NEW_ARRAYS;
+}
+
+void
+nv40_init_state_functions(struct nv40_context *nv40)
+{
+	nv40->pipe.create_blend_state = nv40_blend_state_create;
+	nv40->pipe.bind_blend_state = nv40_blend_state_bind;
+	nv40->pipe.delete_blend_state = nv40_blend_state_delete;
+
+	nv40->pipe.create_sampler_state = nv40_sampler_state_create;
+	nv40->pipe.bind_sampler_states = nv40_sampler_state_bind;
+	nv40->pipe.delete_sampler_state = nv40_sampler_state_delete;
+	nv40->pipe.set_sampler_textures = nv40_set_sampler_texture;
+
+	nv40->pipe.create_rasterizer_state = nv40_rasterizer_state_create;
+	nv40->pipe.bind_rasterizer_state = nv40_rasterizer_state_bind;
+	nv40->pipe.delete_rasterizer_state = nv40_rasterizer_state_delete;
+
+	nv40->pipe.create_depth_stencil_alpha_state =
+		nv40_depth_stencil_alpha_state_create;
+	nv40->pipe.bind_depth_stencil_alpha_state =
+		nv40_depth_stencil_alpha_state_bind;
+	nv40->pipe.delete_depth_stencil_alpha_state =
+		nv40_depth_stencil_alpha_state_delete;
+
+	nv40->pipe.create_vs_state = nv40_vp_state_create;
+	nv40->pipe.bind_vs_state = nv40_vp_state_bind;
+	nv40->pipe.delete_vs_state = nv40_vp_state_delete;
+
+	nv40->pipe.create_fs_state = nv40_fp_state_create;
+	nv40->pipe.bind_fs_state = nv40_fp_state_bind;
+	nv40->pipe.delete_fs_state = nv40_fp_state_delete;
+
+	nv40->pipe.set_blend_color = nv40_set_blend_color;
+	nv40->pipe.set_clip_state = nv40_set_clip_state;
+	nv40->pipe.set_constant_buffer = nv40_set_constant_buffer;
+	nv40->pipe.set_framebuffer_state = nv40_set_framebuffer_state;
+	nv40->pipe.set_polygon_stipple = nv40_set_polygon_stipple;
+	nv40->pipe.set_scissor_state = nv40_set_scissor_state;
+	nv40->pipe.set_viewport_state = nv40_set_viewport_state;
+
+	nv40->pipe.set_edgeflags = nv40_set_edgeflags;
+	nv40->pipe.set_vertex_buffers = nv40_set_vertex_buffers;
+	nv40->pipe.set_vertex_elements = nv40_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv40/nv40_state.h b/src/gallium/drivers/nv40/nv40_state.h
new file mode 100644
index 00000000000..9c55903ae30
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state.h
@@ -0,0 +1,91 @@
+#ifndef __NV40_STATE_H__
+#define __NV40_STATE_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv40_sampler_state {
+	uint32_t fmt;
+	uint32_t wrap;
+	uint32_t en;
+	uint32_t filt;
+	uint32_t bcol;
+};
+
+struct nv40_vertex_program_exec {
+	uint32_t data[4];
+	boolean has_branch_offset;
+	int const_index;
+};
+
+struct nv40_vertex_program_data {
+	int index; /* immediates == -1 */
+	float value[4];
+};
+
+struct nv40_vertex_program {
+	struct pipe_shader_state pipe;
+
+	struct draw_vertex_shader *draw;
+
+	boolean translated;
+
+	struct pipe_clip_state ucp;
+
+	struct nv40_vertex_program_exec *insns;
+	unsigned nr_insns;
+	struct nv40_vertex_program_data *consts;
+	unsigned nr_consts;
+
+	struct nouveau_resource *exec;
+	unsigned exec_start;
+	struct nouveau_resource *data;
+	unsigned data_start;
+	unsigned data_start_min;
+
+	uint32_t ir;
+	uint32_t or;
+	uint32_t clip_ctrl;
+	struct nouveau_stateobj *so;
+};
+
+struct nv40_fragment_program_data {
+	unsigned offset;
+	unsigned index;
+};
+
+struct nv40_fragment_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+
+	boolean translated;
+	unsigned samplers;
+
+	uint32_t *insn;
+	int       insn_len;
+
+	struct nv40_fragment_program_data *consts;
+	unsigned nr_consts;
+
+	struct pipe_buffer *buffer;
+
+	uint32_t fp_control;
+	struct nouveau_stateobj *so;
+};
+
+struct nv40_miptree {
+	struct pipe_texture base;
+
+	struct pipe_buffer *buffer;
+	uint total_size;
+
+	struct pipe_texture *shadow_tex;
+	struct pipe_surface *shadow_surface;
+
+	struct {
+		uint pitch;
+		uint *image_offset;
+	} level[PIPE_MAX_TEXTURE_LEVELS];
+};
+
+#endif
diff --git a/src/gallium/drivers/nv40/nv40_state_blend.c b/src/gallium/drivers/nv40/nv40_state_blend.c
new file mode 100644
index 00000000000..95e6d7394f4
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_blend.c
@@ -0,0 +1,40 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_blend_validate(struct nv40_context *nv40)
+{
+	so_ref(nv40->blend->so, &nv40->state.hw[NV40_STATE_BLEND]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_blend = {
+	.validate = nv40_state_blend_validate,
+	.dirty = {
+		.pipe = NV40_NEW_BLEND,
+		.hw = NV40_STATE_BLEND
+	}
+};
+
+static boolean
+nv40_state_blend_colour_validate(struct nv40_context *nv40)
+{
+	struct nouveau_stateobj *so = so_new(2, 0);
+	struct pipe_blend_color *bcol = &nv40->blend_colour;
+
+	so_method(so, nv40->screen->curie, NV40TCL_BLEND_COLOR, 1);
+	so_data  (so, ((float_to_ubyte(bcol->color[3]) << 24) |
+		       (float_to_ubyte(bcol->color[0]) << 16) |
+		       (float_to_ubyte(bcol->color[1]) <<  8) |
+		       (float_to_ubyte(bcol->color[2]) <<  0)));
+
+	so_ref(so, &nv40->state.hw[NV40_STATE_BCOL]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_blend_colour = {
+	.validate = nv40_state_blend_colour_validate,
+	.dirty = {
+		.pipe = NV40_NEW_BCOL,
+		.hw = NV40_STATE_BCOL
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c
new file mode 100644
index 00000000000..ab88dc416e5
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_emit.c
@@ -0,0 +1,184 @@
+#include "nv40_context.h"
+#include "nv40_state.h"
+#include "draw/draw_context.h"
+
+static struct nv40_state_entry *render_states[] = {
+	&nv40_state_framebuffer,
+	&nv40_state_rasterizer,
+	&nv40_state_scissor,
+	&nv40_state_stipple,
+	&nv40_state_fragprog,
+	&nv40_state_fragtex,
+	&nv40_state_vertprog,
+	&nv40_state_blend,
+	&nv40_state_blend_colour,
+	&nv40_state_zsa,
+	&nv40_state_viewport,
+	&nv40_state_vbo,
+	NULL
+};
+
+static struct nv40_state_entry *swtnl_states[] = {
+	&nv40_state_framebuffer,
+	&nv40_state_rasterizer,
+	&nv40_state_scissor,
+	&nv40_state_stipple,
+	&nv40_state_fragprog,
+	&nv40_state_fragtex,
+	&nv40_state_vertprog,
+	&nv40_state_blend,
+	&nv40_state_blend_colour,
+	&nv40_state_zsa,
+	&nv40_state_viewport,
+	&nv40_state_vtxfmt,
+	NULL
+};
+
+static void
+nv40_state_do_validate(struct nv40_context *nv40,
+		       struct nv40_state_entry **states)
+{
+	const struct pipe_framebuffer_state *fb = &nv40->framebuffer;
+	unsigned i;
+
+	for (i = 0; i < fb->num_cbufs; i++)
+		fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED;
+	if (fb->zsbuf)
+		fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED;
+
+	while (*states) {
+		struct nv40_state_entry *e = *states;
+
+		if (nv40->dirty & e->dirty.pipe) {
+			if (e->validate(nv40))
+				nv40->state.dirty |= (1ULL << e->dirty.hw);
+		}
+
+		states++;
+	}
+	nv40->dirty = 0;
+}
+
+void
+nv40_state_emit(struct nv40_context *nv40)
+{
+	struct nv40_state *state = &nv40->state;
+	struct nv40_screen *screen = nv40->screen;
+	unsigned i, samplers;
+	uint64 states;
+
+	if (nv40->pctx_id != screen->cur_pctx) {
+		for (i = 0; i < NV40_STATE_MAX; i++) {
+			if (state->hw[i] && screen->state[i] != state->hw[i])
+				state->dirty |= (1ULL << i);
+		}
+
+		screen->cur_pctx = nv40->pctx_id;
+	}
+
+	for (i = 0, states = state->dirty; states; i++) {
+		if (!(states & (1ULL << i)))
+			continue;
+		so_ref (state->hw[i], &nv40->screen->state[i]);
+		if (state->hw[i])
+			so_emit(nv40->nvws, nv40->screen->state[i]);
+		states &= ~(1ULL << i);
+	}
+
+	if (state->dirty & ((1ULL << NV40_STATE_FRAGPROG) |
+			    (1ULL << NV40_STATE_FRAGTEX0))) {
+		BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1);
+		OUT_RING  (2);
+		BEGIN_RING(curie, NV40TCL_TEX_CACHE_CTL, 1);
+		OUT_RING  (1);
+	}
+
+	state->dirty = 0;
+
+	so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FB]);
+	for (i = 0, samplers = state->fp_samplers; i < 16 && samplers; i++) {
+		if (!(samplers & (1 << i)))
+			continue;
+		so_emit_reloc_markers(nv40->nvws,
+				      state->hw[NV40_STATE_FRAGTEX0+i]);
+		samplers &= ~(1ULL << i);
+	}
+	so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_FRAGPROG]);
+	if (state->hw[NV40_STATE_VTXBUF] && nv40->render_mode == HW)
+		so_emit_reloc_markers(nv40->nvws, state->hw[NV40_STATE_VTXBUF]);
+}
+
+boolean
+nv40_state_validate(struct nv40_context *nv40)
+{
+	boolean was_sw = nv40->fallback_swtnl ? TRUE : FALSE;
+
+	if (nv40->render_mode != HW) {
+		/* Don't even bother trying to go back to hw if none
+		 * of the states that caused swtnl previously have changed.
+		 */
+		if ((nv40->fallback_swtnl & nv40->dirty)
+				!= nv40->fallback_swtnl)
+			return FALSE;
+
+		/* Attempt to go to hwtnl again */
+		nv40->pipe.flush(&nv40->pipe, 0, NULL);
+		nv40->dirty |= (NV40_NEW_VIEWPORT |
+				NV40_NEW_VERTPROG |
+				NV40_NEW_ARRAYS);
+		nv40->render_mode = HW;
+	}
+
+	nv40_state_do_validate(nv40, render_states);
+	if (nv40->fallback_swtnl || nv40->fallback_swrast)
+		return FALSE;
+	
+	if (was_sw)
+		NOUVEAU_ERR("swtnl->hw\n");
+
+	return TRUE;
+}
+
+boolean
+nv40_state_validate_swtnl(struct nv40_context *nv40)
+{
+	struct draw_context *draw = nv40->draw;
+
+	/* Setup for swtnl */
+	if (nv40->render_mode == HW) {
+		NOUVEAU_ERR("hw->swtnl 0x%08x\n", nv40->fallback_swtnl);
+		nv40->pipe.flush(&nv40->pipe, 0, NULL);
+		nv40->dirty |= (NV40_NEW_VIEWPORT |
+				NV40_NEW_VERTPROG |
+				NV40_NEW_ARRAYS);
+		nv40->render_mode = SWTNL;
+	}
+
+	if (nv40->draw_dirty & NV40_NEW_VERTPROG)
+		draw_bind_vertex_shader(draw, nv40->vertprog->draw);
+
+	if (nv40->draw_dirty & NV40_NEW_RAST)
+		draw_set_rasterizer_state(draw, &nv40->rasterizer->pipe);
+
+	if (nv40->draw_dirty & NV40_NEW_UCP)
+		draw_set_clip_state(draw, &nv40->clip);
+
+	if (nv40->draw_dirty & NV40_NEW_VIEWPORT)
+		draw_set_viewport_state(draw, &nv40->viewport);
+
+	if (nv40->draw_dirty & NV40_NEW_ARRAYS) {
+		draw_set_edgeflags(draw, nv40->edgeflags);
+		draw_set_vertex_buffers(draw, nv40->vtxbuf_nr, nv40->vtxbuf);
+		draw_set_vertex_elements(draw, nv40->vtxelt_nr, nv40->vtxelt);	
+	}
+
+	nv40_state_do_validate(nv40, swtnl_states);
+	if (nv40->fallback_swrast) {
+		NOUVEAU_ERR("swtnl->swrast 0x%08x\n", nv40->fallback_swrast);
+		return FALSE;
+	}
+
+	nv40->draw_dirty = 0;
+	return TRUE;
+}
+
diff --git a/src/gallium/drivers/nv40/nv40_state_fb.c b/src/gallium/drivers/nv40/nv40_state_fb.c
new file mode 100644
index 00000000000..28592d71c37
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_fb.c
@@ -0,0 +1,155 @@
+#include "nv40_context.h"
+#include "nouveau/nouveau_util.h"
+
+static boolean
+nv40_state_framebuffer_validate(struct nv40_context *nv40)
+{
+	struct pipe_framebuffer_state *fb = &nv40->framebuffer;
+	struct pipe_surface *rt[4], *zeta;
+	uint32_t rt_enable, rt_format;
+	int i, colour_format = 0, zeta_format = 0;
+	struct nouveau_stateobj *so = so_new(64, 10);
+	unsigned rt_flags = NOUVEAU_BO_RDWR | NOUVEAU_BO_VRAM;
+	unsigned w = fb->width;
+	unsigned h = fb->height;
+
+	rt_enable = 0;
+	for (i = 0; i < fb->num_cbufs; i++) {
+		if (colour_format) {
+			assert(colour_format == fb->cbufs[i]->format);
+		} else {
+			colour_format = fb->cbufs[i]->format;
+			rt_enable |= (NV40TCL_RT_ENABLE_COLOR0 << i);
+			rt[i] = fb->cbufs[i];
+		}
+	}
+
+	if (rt_enable & (NV40TCL_RT_ENABLE_COLOR1 | NV40TCL_RT_ENABLE_COLOR2 |
+			 NV40TCL_RT_ENABLE_COLOR3))
+		rt_enable |= NV40TCL_RT_ENABLE_MRT;
+
+	if (fb->zsbuf) {
+		zeta_format = fb->zsbuf->format;
+		zeta = fb->zsbuf;
+	}
+
+	if (!(rt[0]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR)) {
+		assert(!(fb->width & (fb->width - 1)) && !(fb->height & (fb->height - 1)));
+		for (i = 1; i < fb->num_cbufs; i++)
+			assert(!(rt[i]->texture->tex_usage & NOUVEAU_TEXTURE_USAGE_LINEAR));
+
+		rt_format = NV40TCL_RT_FORMAT_TYPE_SWIZZLED |
+		            log2i(fb->width) << NV40TCL_RT_FORMAT_LOG2_WIDTH_SHIFT |
+		            log2i(fb->height) << NV40TCL_RT_FORMAT_LOG2_HEIGHT_SHIFT;
+	}
+	else
+		rt_format = NV40TCL_RT_FORMAT_TYPE_LINEAR;
+
+	switch (colour_format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case 0:
+		rt_format |= NV40TCL_RT_FORMAT_COLOR_A8R8G8B8;
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		rt_format |= NV40TCL_RT_FORMAT_COLOR_R5G6B5;
+		break;
+	default:
+		assert(0);
+	}
+
+	switch (zeta_format) {
+	case PIPE_FORMAT_Z16_UNORM:
+		rt_format |= NV40TCL_RT_FORMAT_ZETA_Z16;
+		break;
+	case PIPE_FORMAT_Z24S8_UNORM:
+	case 0:
+		rt_format |= NV40TCL_RT_FORMAT_ZETA_Z24S8;
+		break;
+	default:
+		assert(0);
+	}
+
+	if (rt_enable & NV40TCL_RT_ENABLE_COLOR0) {
+		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR0, 1);
+		so_reloc (so, rt[0]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv40->nvws->channel->vram->handle,
+			  nv40->nvws->channel->gart->handle);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR0_PITCH, 2);
+		so_data  (so, rt[0]->stride);
+		so_reloc (so, rt[0]->buffer, rt[0]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+	}
+
+	if (rt_enable & NV40TCL_RT_ENABLE_COLOR1) {
+		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR1, 1);
+		so_reloc (so, rt[1]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv40->nvws->channel->vram->handle,
+			  nv40->nvws->channel->gart->handle);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR1_OFFSET, 2);
+		so_reloc (so, rt[1]->buffer, rt[1]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		so_data  (so, rt[1]->stride);
+	}
+
+	if (rt_enable & NV40TCL_RT_ENABLE_COLOR2) {
+		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR2, 1);
+		so_reloc (so, rt[2]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv40->nvws->channel->vram->handle,
+			  nv40->nvws->channel->gart->handle);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR2_OFFSET, 1);
+		so_reloc (so, rt[2]->buffer, rt[2]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR2_PITCH, 1);
+		so_data  (so, rt[2]->stride);
+	}
+
+	if (rt_enable & NV40TCL_RT_ENABLE_COLOR3) {
+		so_method(so, nv40->screen->curie, NV40TCL_DMA_COLOR3, 1);
+		so_reloc (so, rt[3]->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv40->nvws->channel->vram->handle,
+			  nv40->nvws->channel->gart->handle);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR3_OFFSET, 1);
+		so_reloc (so, rt[3]->buffer, rt[3]->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		so_method(so, nv40->screen->curie, NV40TCL_COLOR3_PITCH, 1);
+		so_data  (so, rt[3]->stride);
+	}
+
+	if (zeta_format) {
+		so_method(so, nv40->screen->curie, NV40TCL_DMA_ZETA, 1);
+		so_reloc (so, zeta->buffer, 0, rt_flags | NOUVEAU_BO_OR,
+			  nv40->nvws->channel->vram->handle,
+			  nv40->nvws->channel->gart->handle);
+		so_method(so, nv40->screen->curie, NV40TCL_ZETA_OFFSET, 1);
+		so_reloc (so, zeta->buffer, zeta->offset, rt_flags |
+			  NOUVEAU_BO_LOW, 0, 0);
+		so_method(so, nv40->screen->curie, NV40TCL_ZETA_PITCH, 1);
+		so_data  (so, zeta->stride);
+	}
+
+	so_method(so, nv40->screen->curie, NV40TCL_RT_ENABLE, 1);
+	so_data  (so, rt_enable);
+	so_method(so, nv40->screen->curie, NV40TCL_RT_HORIZ, 3);
+	so_data  (so, (w << 16) | 0);
+	so_data  (so, (h << 16) | 0);
+	so_data  (so, rt_format);
+	so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_HORIZ, 2);
+	so_data  (so, (w << 16) | 0);
+	so_data  (so, (h << 16) | 0);
+	so_method(so, nv40->screen->curie, NV40TCL_VIEWPORT_CLIP_HORIZ(0), 2);
+	so_data  (so, ((w - 1) << 16) | 0);
+	so_data  (so, ((h - 1) << 16) | 0);
+	so_method(so, nv40->screen->curie, 0x1d88, 1);
+	so_data  (so, (1 << 12) | h);
+
+	so_ref(so, &nv40->state.hw[NV40_STATE_FB]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_framebuffer = {
+	.validate = nv40_state_framebuffer_validate,
+	.dirty = {
+		.pipe = NV40_NEW_FB,
+		.hw = NV40_STATE_FB
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_rasterizer.c b/src/gallium/drivers/nv40/nv40_state_rasterizer.c
new file mode 100644
index 00000000000..9ecda5990f0
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_rasterizer.c
@@ -0,0 +1,17 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_rasterizer_validate(struct nv40_context *nv40)
+{
+	so_ref(nv40->rasterizer->so,
+	       &nv40->state.hw[NV40_STATE_RAST]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_rasterizer = {
+	.validate = nv40_state_rasterizer_validate,
+	.dirty = {
+		.pipe = NV40_NEW_RAST,
+		.hw = NV40_STATE_RAST
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_scissor.c b/src/gallium/drivers/nv40/nv40_state_scissor.c
new file mode 100644
index 00000000000..285239ef419
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_scissor.c
@@ -0,0 +1,35 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_scissor_validate(struct nv40_context *nv40)
+{
+	struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe;
+	struct pipe_scissor_state *s = &nv40->scissor;
+	struct nouveau_stateobj *so;
+
+	if (nv40->state.hw[NV40_STATE_SCISSOR] &&
+	    (rast->scissor == 0 && nv40->state.scissor_enabled == 0))
+		return FALSE;
+	nv40->state.scissor_enabled = rast->scissor;
+
+	so = so_new(3, 0);
+	so_method(so, nv40->screen->curie, NV40TCL_SCISSOR_HORIZ, 2);
+	if (nv40->state.scissor_enabled) {
+		so_data  (so, ((s->maxx - s->minx) << 16) | s->minx);
+		so_data  (so, ((s->maxy - s->miny) << 16) | s->miny);
+	} else {
+		so_data  (so, 4096 << 16);
+		so_data  (so, 4096 << 16);
+	}
+
+	so_ref(so, &nv40->state.hw[NV40_STATE_SCISSOR]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_scissor = {
+	.validate = nv40_state_scissor_validate,
+	.dirty = {
+		.pipe = NV40_NEW_SCISSOR | NV40_NEW_RAST,
+		.hw = NV40_STATE_SCISSOR
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_stipple.c b/src/gallium/drivers/nv40/nv40_state_stipple.c
new file mode 100644
index 00000000000..b51024ad9b2
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_stipple.c
@@ -0,0 +1,39 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_stipple_validate(struct nv40_context *nv40)
+{
+	struct pipe_rasterizer_state *rast = &nv40->rasterizer->pipe;
+	struct nouveau_grobj *curie = nv40->screen->curie;
+	struct nouveau_stateobj *so;
+
+	if (nv40->state.hw[NV40_STATE_STIPPLE] &&
+	   (rast->poly_stipple_enable == 0 && nv40->state.stipple_enabled == 0))
+		return FALSE;
+
+	if (rast->poly_stipple_enable) {
+		unsigned i;
+
+		so = so_new(35, 0);
+		so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1);
+		so_data  (so, 1);
+		so_method(so, curie, NV40TCL_POLYGON_STIPPLE_PATTERN(0), 32);
+		for (i = 0; i < 32; i++)
+			so_data(so, nv40->stipple[i]);
+	} else {
+		so = so_new(2, 0);
+		so_method(so, curie, NV40TCL_POLYGON_STIPPLE_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_ref(so, &nv40->state.hw[NV40_STATE_STIPPLE]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_stipple = {
+	.validate = nv40_state_stipple_validate,
+	.dirty = {
+		.pipe = NV40_NEW_STIPPLE | NV40_NEW_RAST,
+		.hw = NV40_STATE_STIPPLE,
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_viewport.c b/src/gallium/drivers/nv40/nv40_state_viewport.c
new file mode 100644
index 00000000000..869a55b4053
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_viewport.c
@@ -0,0 +1,67 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_viewport_validate(struct nv40_context *nv40)
+{
+	struct pipe_viewport_state *vpt = &nv40->viewport;
+	struct nouveau_stateobj *so;
+	unsigned bypass;
+
+	if (nv40->render_mode == HW && !nv40->rasterizer->pipe.bypass_clipping)
+		bypass = 0;
+	else
+		bypass = 1;
+
+	if (nv40->state.hw[NV40_STATE_VIEWPORT] &&
+	    (bypass || !(nv40->dirty & NV40_NEW_VIEWPORT)) &&
+	    nv40->state.viewport_bypass == bypass)
+		return FALSE;
+	nv40->state.viewport_bypass = bypass;
+
+	so = so_new(11, 0);
+	if (!bypass) {
+		so_method(so, nv40->screen->curie,
+			  NV40TCL_VIEWPORT_TRANSLATE_X, 8);
+		so_data  (so, fui(vpt->translate[0]));
+		so_data  (so, fui(vpt->translate[1]));
+		so_data  (so, fui(vpt->translate[2]));
+		so_data  (so, fui(vpt->translate[3]));
+		so_data  (so, fui(vpt->scale[0]));
+		so_data  (so, fui(vpt->scale[1]));
+		so_data  (so, fui(vpt->scale[2]));
+		so_data  (so, fui(vpt->scale[3]));
+		so_method(so, nv40->screen->curie, 0x1d78, 1);
+		so_data  (so, 1);
+	} else {
+		so_method(so, nv40->screen->curie,
+			  NV40TCL_VIEWPORT_TRANSLATE_X, 8);
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(0.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(1.0));
+		so_data  (so, fui(0.0));
+		/* Not entirely certain what this is yet.  The DDX uses this
+		 * value also as it fixes rendering when you pass
+		 * pre-transformed vertices to the GPU.  My best gusss is that
+		 * this bypasses some culling/clipping stage.  Might be worth
+		 * noting that points/lines are uneffected by whatever this
+		 * value fixes, only filled polygons are effected.
+		 */
+		so_method(so, nv40->screen->curie, 0x1d78, 1);
+		so_data  (so, 0x110);
+	}
+
+	so_ref(so, &nv40->state.hw[NV40_STATE_VIEWPORT]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_viewport = {
+	.validate = nv40_state_viewport_validate,
+	.dirty = {
+		.pipe = NV40_NEW_VIEWPORT | NV40_NEW_RAST,
+		.hw = NV40_STATE_VIEWPORT
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_state_zsa.c b/src/gallium/drivers/nv40/nv40_state_zsa.c
new file mode 100644
index 00000000000..fb760677c88
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_state_zsa.c
@@ -0,0 +1,17 @@
+#include "nv40_context.h"
+
+static boolean
+nv40_state_zsa_validate(struct nv40_context *nv40)
+{
+	so_ref(nv40->zsa->so,
+	       &nv40->state.hw[NV40_STATE_ZSA]);
+	return TRUE;
+}
+
+struct nv40_state_entry nv40_state_zsa = {
+	.validate = nv40_state_zsa_validate,
+	.dirty = {
+		.pipe = NV40_NEW_ZSA,
+		.hw = NV40_STATE_ZSA
+	}
+};
diff --git a/src/gallium/drivers/nv40/nv40_surface.c b/src/gallium/drivers/nv40/nv40_surface.c
new file mode 100644
index 00000000000..576af7c59ee
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_surface.c
@@ -0,0 +1,77 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "nv40_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+
+#include "util/u_tile.h"
+
+static void
+nv40_surface_copy(struct pipe_context *pipe, boolean do_flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nouveau_winsys *nvws = nv40->nvws;
+
+	if (do_flip) {
+		/*XXX: This dodgyness will do for now for correctness.  But,
+		 *     need to investigate whether the 2D engine is able to
+		 *     manage a flip (perhaps SIFM?), if not, use the 3D engine
+		 */
+		desty += height;
+		while (height--) {
+			nvws->surface_copy(nvws, dest, destx, desty--, src,
+					   srcx, srcy++, width, 1);
+		}
+	} else {
+		nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+				   width, height);
+	}
+}
+
+static void
+nv40_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nouveau_winsys *nvws = nv40->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+void
+nv40_init_surface_functions(struct nv40_context *nv40)
+{
+	nv40->pipe.surface_copy = nv40_surface_copy;
+	nv40->pipe.surface_fill = nv40_surface_fill;
+}
diff --git a/src/gallium/drivers/nv40/nv40_vbo.c b/src/gallium/drivers/nv40/nv40_vbo.c
new file mode 100644
index 00000000000..09f6e79d32a
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_vbo.c
@@ -0,0 +1,555 @@
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv40_context.h"
+#include "nv40_state.h"
+
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_pushbuf.h"
+#include "nouveau/nouveau_util.h"
+
+#define FORCE_SWTNL 0
+
+static INLINE int
+nv40_vbo_format_to_hw(enum pipe_format pipe, unsigned *fmt, unsigned *ncomp)
+{
+	switch (pipe) {
+	case PIPE_FORMAT_R32_FLOAT:
+	case PIPE_FORMAT_R32G32_FLOAT:
+	case PIPE_FORMAT_R32G32B32_FLOAT:
+	case PIPE_FORMAT_R32G32B32A32_FLOAT:
+		*fmt = NV40TCL_VTXFMT_TYPE_FLOAT;
+		break;
+	case PIPE_FORMAT_R8_UNORM:
+	case PIPE_FORMAT_R8G8_UNORM:
+	case PIPE_FORMAT_R8G8B8_UNORM:
+	case PIPE_FORMAT_R8G8B8A8_UNORM:
+		*fmt = NV40TCL_VTXFMT_TYPE_UBYTE;
+		break;
+	case PIPE_FORMAT_R16_SSCALED:
+	case PIPE_FORMAT_R16G16_SSCALED:
+	case PIPE_FORMAT_R16G16B16_SSCALED:
+	case PIPE_FORMAT_R16G16B16A16_SSCALED:
+		*fmt = NV40TCL_VTXFMT_TYPE_USHORT;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe));
+		return 1;
+	}
+
+	switch (pipe) {
+	case PIPE_FORMAT_R8_UNORM:
+	case PIPE_FORMAT_R32_FLOAT:
+	case PIPE_FORMAT_R16_SSCALED:
+		*ncomp = 1;
+		break;
+	case PIPE_FORMAT_R8G8_UNORM:
+	case PIPE_FORMAT_R32G32_FLOAT:
+	case PIPE_FORMAT_R16G16_SSCALED:
+		*ncomp = 2;
+		break;
+	case PIPE_FORMAT_R8G8B8_UNORM:
+	case PIPE_FORMAT_R32G32B32_FLOAT:
+	case PIPE_FORMAT_R16G16B16_SSCALED:
+		*ncomp = 3;
+		break;
+	case PIPE_FORMAT_R8G8B8A8_UNORM:
+	case PIPE_FORMAT_R32G32B32A32_FLOAT:
+	case PIPE_FORMAT_R16G16B16A16_SSCALED:
+		*ncomp = 4;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown format %s\n", pf_name(pipe));
+		return 1;
+	}
+
+	return 0;
+}
+
+static boolean
+nv40_vbo_set_idxbuf(struct nv40_context *nv40, struct pipe_buffer *ib,
+		    unsigned ib_size)
+{
+	struct pipe_screen *pscreen = &nv40->screen->pipe;
+	unsigned type;
+
+	if (!ib) {
+		nv40->idxbuf = NULL;
+		nv40->idxbuf_format = 0xdeadbeef;
+		return FALSE;
+	}
+
+	if (!pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF) || ib_size == 1)
+		return FALSE;
+
+	switch (ib_size) {
+	case 2:
+		type = NV40TCL_IDXBUF_FORMAT_TYPE_U16;
+		break;
+	case 4:
+		type = NV40TCL_IDXBUF_FORMAT_TYPE_U32;
+		break;
+	default:
+		return FALSE;
+	}
+
+	if (ib != nv40->idxbuf ||
+	    type != nv40->idxbuf_format) {
+		nv40->dirty |= NV40_NEW_ARRAYS;
+		nv40->idxbuf = ib;
+		nv40->idxbuf_format = type;
+	}
+
+	return TRUE;
+}
+
+static boolean
+nv40_vbo_static_attrib(struct nv40_context *nv40, struct nouveau_stateobj *so,
+		       int attrib, struct pipe_vertex_element *ve,
+		       struct pipe_vertex_buffer *vb)
+{
+	struct pipe_winsys *ws = nv40->pipe.winsys;
+	struct nouveau_grobj *curie = nv40->screen->curie;
+	unsigned type, ncomp;
+	void *map;
+
+	if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp))
+		return FALSE;
+
+	map  = ws->buffer_map(ws, vb->buffer, PIPE_BUFFER_USAGE_CPU_READ);
+	map += vb->buffer_offset + ve->src_offset;
+
+	switch (type) {
+	case NV40TCL_VTXFMT_TYPE_FLOAT:
+	{
+		float *v = map;
+
+		switch (ncomp) {
+		case 4:
+			so_method(so, curie, NV40TCL_VTX_ATTR_4F_X(attrib), 4);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			so_data  (so, fui(v[2]));
+			so_data  (so, fui(v[3]));
+			break;
+		case 3:
+			so_method(so, curie, NV40TCL_VTX_ATTR_3F_X(attrib), 3);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			so_data  (so, fui(v[2]));
+			break;
+		case 2:
+			so_method(so, curie, NV40TCL_VTX_ATTR_2F_X(attrib), 2);
+			so_data  (so, fui(v[0]));
+			so_data  (so, fui(v[1]));
+			break;
+		case 1:
+			so_method(so, curie, NV40TCL_VTX_ATTR_1F(attrib), 1);
+			so_data  (so, fui(v[0]));
+			break;
+		default:
+			ws->buffer_unmap(ws, vb->buffer);
+			return FALSE;
+		}
+	}
+		break;
+	default:
+		ws->buffer_unmap(ws, vb->buffer);
+		return FALSE;
+	}
+
+	ws->buffer_unmap(ws, vb->buffer);
+
+	return TRUE;
+}
+
+boolean
+nv40_draw_arrays(struct pipe_context *pipe,
+		 unsigned mode, unsigned start, unsigned count)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nouveau_channel *chan = nv40->nvws->channel;
+	unsigned restart;
+
+	nv40_vbo_set_idxbuf(nv40, NULL, 0);
+	if (FORCE_SWTNL || !nv40_state_validate(nv40)) {
+		return nv40_draw_elements_swtnl(pipe, NULL, 0,
+						mode, start, count);
+	}
+
+	while (count) {
+		unsigned vc, nr;
+
+		nv40_state_emit(nv40);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256,
+					mode, start, count, &restart);
+		if (!vc) {
+			FIRE_RING(NULL);
+			continue;
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		nr = (vc & 0xff);
+		if (nr) {
+			BEGIN_RING(curie, NV40TCL_VB_VERTEX_BATCH, 1);
+			OUT_RING  (((nr - 1) << 24) | start);
+			start += nr;
+		}
+
+		nr = vc >> 8;
+		while (nr) {
+			unsigned push = nr > 2047 ? 2047 : nr;
+
+			nr -= push;
+
+			BEGIN_RING_NI(curie, NV40TCL_VB_VERTEX_BATCH, push);
+			while (push--) {
+				OUT_RING(((0x100 - 1) << 24) | start);
+				start += 0x100;
+			}
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		count -= vc;
+		start = restart;
+	}
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+static INLINE void
+nv40_draw_elements_u08(struct nv40_context *nv40, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv40->nvws->channel;
+
+	while (count) {
+		uint8_t *elts = (uint8_t *)ib + start;
+		unsigned vc, push, restart;
+
+		nv40_state_emit(nv40);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		if (vc & 1) {
+			BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1);
+			OUT_RING  (elts[0]);
+			elts++; vc--;
+		}
+
+		while (vc) {
+			unsigned i;
+
+			push = MIN2(vc, 2047 * 2);
+
+			BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1);
+			for (i = 0; i < push; i+=2)
+				OUT_RING((elts[i+1] << 16) | elts[i]);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static INLINE void
+nv40_draw_elements_u16(struct nv40_context *nv40, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv40->nvws->channel;
+
+	while (count) {
+		uint16_t *elts = (uint16_t *)ib + start;
+		unsigned vc, push, restart;
+
+		nv40_state_emit(nv40);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 2,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		if (vc & 1) {
+			BEGIN_RING(curie, NV40TCL_VB_ELEMENT_U32, 1);
+			OUT_RING  (elts[0]);
+			elts++; vc--;
+		}
+
+		while (vc) {
+			unsigned i;
+
+			push = MIN2(vc, 2047 * 2);
+
+			BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U16, push >> 1);
+			for (i = 0; i < push; i+=2)
+				OUT_RING((elts[i+1] << 16) | elts[i]);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static INLINE void
+nv40_draw_elements_u32(struct nv40_context *nv40, void *ib,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nouveau_channel *chan = nv40->nvws->channel;
+
+	while (count) {
+		uint32_t *elts = (uint32_t *)ib + start;
+		unsigned vc, push, restart;
+
+		nv40_state_emit(nv40);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 5, 1,
+					mode, start, count, &restart);
+		if (vc == 0) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		count -= vc;
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		while (vc) {
+			push = MIN2(vc, 2047);
+
+			BEGIN_RING_NI(curie, NV40TCL_VB_ELEMENT_U32, push);
+			OUT_RINGp    (elts, push);
+
+			vc -= push;
+			elts += push;
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		start = restart;
+	}
+}
+
+static boolean
+nv40_draw_elements_inline(struct pipe_context *pipe,
+			  struct pipe_buffer *ib, unsigned ib_size,
+			  unsigned mode, unsigned start, unsigned count)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+	void *map;
+
+	map = ws->buffer_map(ws, ib, PIPE_BUFFER_USAGE_CPU_READ);
+	if (!ib) {
+		NOUVEAU_ERR("failed mapping ib\n");
+		return FALSE;
+	}
+
+	switch (ib_size) {
+	case 1:
+		nv40_draw_elements_u08(nv40, map, mode, start, count);
+		break;
+	case 2:
+		nv40_draw_elements_u16(nv40, map, mode, start, count);
+		break;
+	case 4:
+		nv40_draw_elements_u32(nv40, map, mode, start, count);
+		break;
+	default:
+		NOUVEAU_ERR("invalid idxbuf fmt %d\n", ib_size);
+		break;
+	}
+
+	ws->buffer_unmap(ws, ib);
+	return TRUE;
+}
+
+static boolean
+nv40_draw_elements_vbo(struct pipe_context *pipe,
+		       unsigned mode, unsigned start, unsigned count)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	struct nouveau_channel *chan = nv40->nvws->channel;
+	unsigned restart;
+
+	while (count) {
+		unsigned nr, vc;
+
+		nv40_state_emit(nv40);
+
+		vc = nouveau_vbuf_split(chan->pushbuf->remaining, 6, 256,
+					mode, start, count, &restart);
+		if (!vc) {
+			FIRE_RING(NULL);
+			continue;
+		}
+		
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (nvgl_primitive(mode));
+
+		nr = (vc & 0xff);
+		if (nr) {
+			BEGIN_RING(curie, NV40TCL_VB_INDEX_BATCH, 1);
+			OUT_RING  (((nr - 1) << 24) | start);
+			start += nr;
+		}
+
+		nr = vc >> 8;
+		while (nr) {
+			unsigned push = nr > 2047 ? 2047 : nr;
+
+			nr -= push;
+
+			BEGIN_RING_NI(curie, NV40TCL_VB_INDEX_BATCH, push);
+			while (push--) {
+				OUT_RING(((0x100 - 1) << 24) | start);
+				start += 0x100;
+			}
+		}
+
+		BEGIN_RING(curie, NV40TCL_BEGIN_END, 1);
+		OUT_RING  (0);
+
+		count -= vc;
+		start = restart;
+	}
+
+	return TRUE;
+}
+
+boolean
+nv40_draw_elements(struct pipe_context *pipe,
+		   struct pipe_buffer *indexBuffer, unsigned indexSize,
+		   unsigned mode, unsigned start, unsigned count)
+{
+	struct nv40_context *nv40 = nv40_context(pipe);
+	boolean idxbuf;
+
+	idxbuf = nv40_vbo_set_idxbuf(nv40, indexBuffer, indexSize);
+	if (FORCE_SWTNL || !nv40_state_validate(nv40)) {
+		return nv40_draw_elements_swtnl(pipe, NULL, 0,
+						mode, start, count);
+	}
+
+	if (idxbuf) {
+		nv40_draw_elements_vbo(pipe, mode, start, count);
+	} else {
+		nv40_draw_elements_inline(pipe, indexBuffer, indexSize,
+					  mode, start, count);
+	}
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+static boolean
+nv40_vbo_validate(struct nv40_context *nv40)
+{
+	struct nouveau_stateobj *vtxbuf, *vtxfmt, *sattr = NULL;
+	struct nouveau_grobj *curie = nv40->screen->curie;
+	struct pipe_buffer *ib = nv40->idxbuf;
+	unsigned ib_format = nv40->idxbuf_format;
+	unsigned vb_flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD;
+	int hw;
+
+	if (nv40->edgeflags) {
+		nv40->fallback_swtnl |= NV40_NEW_ARRAYS;
+		return FALSE;
+	}
+
+	vtxbuf = so_new(20, 18);
+	so_method(vtxbuf, curie, NV40TCL_VTXBUF_ADDRESS(0), nv40->vtxelt_nr);
+	vtxfmt = so_new(17, 0);
+	so_method(vtxfmt, curie, NV40TCL_VTXFMT(0), nv40->vtxelt_nr);
+
+	for (hw = 0; hw < nv40->vtxelt_nr; hw++) {
+		struct pipe_vertex_element *ve;
+		struct pipe_vertex_buffer *vb;
+		unsigned type, ncomp;
+
+		ve = &nv40->vtxelt[hw];
+		vb = &nv40->vtxbuf[ve->vertex_buffer_index];
+
+		if (!vb->pitch) {
+			if (!sattr)
+				sattr = so_new(16 * 5, 0);
+
+			if (nv40_vbo_static_attrib(nv40, sattr, hw, ve, vb)) {
+				so_data(vtxbuf, 0);
+				so_data(vtxfmt, NV40TCL_VTXFMT_TYPE_FLOAT);
+				continue;
+			}
+		}
+
+		if (nv40_vbo_format_to_hw(ve->src_format, &type, &ncomp)) {
+			nv40->fallback_swtnl |= NV40_NEW_ARRAYS;
+			so_ref(NULL, &vtxbuf);
+			so_ref(NULL, &vtxfmt);
+			return FALSE;
+		}
+
+		so_reloc(vtxbuf, vb->buffer, vb->buffer_offset + ve->src_offset,
+			 vb_flags | NOUVEAU_BO_LOW | NOUVEAU_BO_OR,
+			 0, NV40TCL_VTXBUF_ADDRESS_DMA1);
+		so_data (vtxfmt, ((vb->pitch << NV40TCL_VTXFMT_STRIDE_SHIFT) |
+				  (ncomp << NV40TCL_VTXFMT_SIZE_SHIFT) | type));
+	}
+
+	if (ib) {
+		so_method(vtxbuf, curie, NV40TCL_IDXBUF_ADDRESS, 2);
+		so_reloc (vtxbuf, ib, 0, vb_flags | NOUVEAU_BO_LOW, 0, 0);
+		so_reloc (vtxbuf, ib, ib_format, vb_flags | NOUVEAU_BO_OR,
+			  0, NV40TCL_IDXBUF_FORMAT_DMA1);
+	}
+
+	so_method(vtxbuf, curie, 0x1710, 1);
+	so_data  (vtxbuf, 0);
+
+	so_ref(vtxbuf, &nv40->state.hw[NV40_STATE_VTXBUF]);
+	nv40->state.dirty |= (1ULL << NV40_STATE_VTXBUF);
+	so_ref(vtxfmt, &nv40->state.hw[NV40_STATE_VTXFMT]);
+	nv40->state.dirty |= (1ULL << NV40_STATE_VTXFMT);
+	so_ref(sattr, &nv40->state.hw[NV40_STATE_VTXATTR]);
+	nv40->state.dirty |= (1ULL << NV40_STATE_VTXATTR);
+	return FALSE;
+}
+
+struct nv40_state_entry nv40_state_vbo = {
+	.validate = nv40_vbo_validate,
+	.dirty = {
+		.pipe = NV40_NEW_ARRAYS,
+		.hw = 0,
+	}
+};
+
diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c
new file mode 100644
index 00000000000..ff988e6a5f4
--- /dev/null
+++ b/src/gallium/drivers/nv40/nv40_vertprog.c
@@ -0,0 +1,1070 @@
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv40_context.h"
+#include "nv40_state.h"
+
+/* TODO (at least...):
+ *  1. Indexed consts  + ARL
+ *  3. NV_vp11, NV_vp2, NV_vp3 features
+ *       - extra arith opcodes
+ *       - branching
+ *       - texture sampling
+ *       - indexed attribs
+ *       - indexed results
+ *  4. bugs
+ */
+
+#define SWZ_X 0
+#define SWZ_Y 1
+#define SWZ_Z 2
+#define SWZ_W 3
+#define MASK_X 8
+#define MASK_Y 4
+#define MASK_Z 2
+#define MASK_W 1
+#define MASK_ALL (MASK_X|MASK_Y|MASK_Z|MASK_W)
+#define DEF_SCALE 0
+#define DEF_CTEST 0
+#include "nv40_shader.h"
+
+#define swz(s,x,y,z,w) nv40_sr_swz((s), SWZ_##x, SWZ_##y, SWZ_##z, SWZ_##w)
+#define neg(s) nv40_sr_neg((s))
+#define abs(s) nv40_sr_abs((s))
+
+#define NV40_VP_INST_DEST_CLIP(n) ((~0 - 6) + (n))
+
+struct nv40_vpc {
+	struct nv40_vertex_program *vp;
+
+	struct nv40_vertex_program_exec *vpi;
+
+	unsigned r_temps;
+	unsigned r_temps_discard;
+	struct nv40_sreg r_result[PIPE_MAX_SHADER_OUTPUTS];
+	struct nv40_sreg *r_address;
+	struct nv40_sreg *r_temp;
+
+	struct nv40_sreg *imm;
+	unsigned nr_imm;
+
+	unsigned hpos_idx;
+};
+
+static struct nv40_sreg
+temp(struct nv40_vpc *vpc)
+{
+	int idx = ffs(~vpc->r_temps) - 1;
+
+	if (idx < 0) {
+		NOUVEAU_ERR("out of temps!!\n");
+		assert(0);
+		return nv40_sr(NV40SR_TEMP, 0);
+	}
+
+	vpc->r_temps |= (1 << idx);
+	vpc->r_temps_discard |= (1 << idx);
+	return nv40_sr(NV40SR_TEMP, idx);
+}
+
+static INLINE void
+release_temps(struct nv40_vpc *vpc)
+{
+	vpc->r_temps &= ~vpc->r_temps_discard;
+	vpc->r_temps_discard = 0;
+}
+
+static struct nv40_sreg
+constant(struct nv40_vpc *vpc, int pipe, float x, float y, float z, float w)
+{
+	struct nv40_vertex_program *vp = vpc->vp;
+	struct nv40_vertex_program_data *vpd;
+	int idx;
+
+	if (pipe >= 0) {
+		for (idx = 0; idx < vp->nr_consts; idx++) {
+			if (vp->consts[idx].index == pipe)
+				return nv40_sr(NV40SR_CONST, idx);
+		}
+	}
+
+	idx = vp->nr_consts++;
+	vp->consts = realloc(vp->consts, sizeof(*vpd) * vp->nr_consts);
+	vpd = &vp->consts[idx];
+
+	vpd->index = pipe;
+	vpd->value[0] = x;
+	vpd->value[1] = y;
+	vpd->value[2] = z;
+	vpd->value[3] = w;
+	return nv40_sr(NV40SR_CONST, idx);
+}
+
+#define arith(cc,s,o,d,m,s0,s1,s2) \
+	nv40_vp_arith((cc), (s), NV40_VP_INST_##o, (d), (m), (s0), (s1), (s2))
+
+static void
+emit_src(struct nv40_vpc *vpc, uint32_t *hw, int pos, struct nv40_sreg src)
+{
+	struct nv40_vertex_program *vp = vpc->vp;
+	uint32_t sr = 0;
+
+	switch (src.type) {
+	case NV40SR_TEMP:
+		sr |= (NV40_VP_SRC_REG_TYPE_TEMP << NV40_VP_SRC_REG_TYPE_SHIFT);
+		sr |= (src.index << NV40_VP_SRC_TEMP_SRC_SHIFT);
+		break;
+	case NV40SR_INPUT:
+		sr |= (NV40_VP_SRC_REG_TYPE_INPUT <<
+		       NV40_VP_SRC_REG_TYPE_SHIFT);
+		vp->ir |= (1 << src.index);
+		hw[1] |= (src.index << NV40_VP_INST_INPUT_SRC_SHIFT);
+		break;
+	case NV40SR_CONST:
+		sr |= (NV40_VP_SRC_REG_TYPE_CONST <<
+		       NV40_VP_SRC_REG_TYPE_SHIFT);
+		assert(vpc->vpi->const_index == -1 ||
+		       vpc->vpi->const_index == src.index);
+		vpc->vpi->const_index = src.index;
+		break;
+	case NV40SR_NONE:
+		sr |= (NV40_VP_SRC_REG_TYPE_INPUT <<
+		       NV40_VP_SRC_REG_TYPE_SHIFT);
+		break;
+	default:
+		assert(0);
+	}
+
+	if (src.negate)
+		sr |= NV40_VP_SRC_NEGATE;
+
+	if (src.abs)
+		hw[0] |= (1 << (21 + pos));
+
+	sr |= ((src.swz[0] << NV40_VP_SRC_SWZ_X_SHIFT) |
+	       (src.swz[1] << NV40_VP_SRC_SWZ_Y_SHIFT) |
+	       (src.swz[2] << NV40_VP_SRC_SWZ_Z_SHIFT) |
+	       (src.swz[3] << NV40_VP_SRC_SWZ_W_SHIFT));
+
+	switch (pos) {
+	case 0:
+		hw[1] |= ((sr & NV40_VP_SRC0_HIGH_MASK) >>
+			  NV40_VP_SRC0_HIGH_SHIFT) << NV40_VP_INST_SRC0H_SHIFT;
+		hw[2] |= (sr & NV40_VP_SRC0_LOW_MASK) <<
+			  NV40_VP_INST_SRC0L_SHIFT;
+		break;
+	case 1:
+		hw[2] |= sr << NV40_VP_INST_SRC1_SHIFT;
+		break;
+	case 2:
+		hw[2] |= ((sr & NV40_VP_SRC2_HIGH_MASK) >>
+			  NV40_VP_SRC2_HIGH_SHIFT) << NV40_VP_INST_SRC2H_SHIFT;
+		hw[3] |= (sr & NV40_VP_SRC2_LOW_MASK) <<
+			  NV40_VP_INST_SRC2L_SHIFT;
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+emit_dst(struct nv40_vpc *vpc, uint32_t *hw, int slot, struct nv40_sreg dst)
+{
+	struct nv40_vertex_program *vp = vpc->vp;
+
+	switch (dst.type) {
+	case NV40SR_TEMP:
+		hw[3] |= NV40_VP_INST_DEST_MASK;
+		if (slot == 0) {
+			hw[0] |= (dst.index <<
+				  NV40_VP_INST_VEC_DEST_TEMP_SHIFT);
+		} else {
+			hw[3] |= (dst.index << 
+				  NV40_VP_INST_SCA_DEST_TEMP_SHIFT);
+		}
+		break;
+	case NV40SR_OUTPUT:
+		switch (dst.index) {
+		case NV40_VP_INST_DEST_COL0 : vp->or |= (1 << 0); break;
+		case NV40_VP_INST_DEST_COL1 : vp->or |= (1 << 1); break;
+		case NV40_VP_INST_DEST_BFC0 : vp->or |= (1 << 2); break;
+		case NV40_VP_INST_DEST_BFC1 : vp->or |= (1 << 3); break;
+		case NV40_VP_INST_DEST_FOGC : vp->or |= (1 << 4); break;
+		case NV40_VP_INST_DEST_PSZ  : vp->or |= (1 << 5); break;
+		case NV40_VP_INST_DEST_TC(0): vp->or |= (1 << 14); break;
+		case NV40_VP_INST_DEST_TC(1): vp->or |= (1 << 15); break;
+		case NV40_VP_INST_DEST_TC(2): vp->or |= (1 << 16); break;
+		case NV40_VP_INST_DEST_TC(3): vp->or |= (1 << 17); break;
+		case NV40_VP_INST_DEST_TC(4): vp->or |= (1 << 18); break;
+		case NV40_VP_INST_DEST_TC(5): vp->or |= (1 << 19); break;
+		case NV40_VP_INST_DEST_TC(6): vp->or |= (1 << 20); break;
+		case NV40_VP_INST_DEST_TC(7): vp->or |= (1 << 21); break;
+		case NV40_VP_INST_DEST_CLIP(0):
+			vp->or |= (1 << 6);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE0;
+			dst.index = NV40_VP_INST_DEST_FOGC;
+			break;
+		case NV40_VP_INST_DEST_CLIP(1):
+			vp->or |= (1 << 7);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE1;
+			dst.index = NV40_VP_INST_DEST_FOGC;
+			break;
+		case NV40_VP_INST_DEST_CLIP(2):
+			vp->or |= (1 << 8);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE2;
+			dst.index = NV40_VP_INST_DEST_FOGC;
+			break;
+		case NV40_VP_INST_DEST_CLIP(3):
+			vp->or |= (1 << 9);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE3;
+			dst.index = NV40_VP_INST_DEST_PSZ;
+			break;
+		case NV40_VP_INST_DEST_CLIP(4):
+			vp->or |= (1 << 10);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE4;
+			dst.index = NV40_VP_INST_DEST_PSZ;
+			break;
+		case NV40_VP_INST_DEST_CLIP(5):
+			vp->or |= (1 << 11);
+			vp->clip_ctrl |= NV40TCL_CLIP_PLANE_ENABLE_PLANE5;
+			dst.index = NV40_VP_INST_DEST_PSZ;
+			break;
+		default:
+			break;
+		}
+
+		hw[3] |= (dst.index << NV40_VP_INST_DEST_SHIFT);
+		if (slot == 0) {
+			hw[0] |= NV40_VP_INST_VEC_RESULT;
+			hw[0] |= NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20);
+		} else {
+			hw[3] |= NV40_VP_INST_SCA_RESULT;
+			hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK;
+		}
+		break;
+	default:
+		assert(0);
+	}
+}
+
+static void
+nv40_vp_arith(struct nv40_vpc *vpc, int slot, int op,
+	      struct nv40_sreg dst, int mask,
+	      struct nv40_sreg s0, struct nv40_sreg s1,
+	      struct nv40_sreg s2)
+{
+	struct nv40_vertex_program *vp = vpc->vp;
+	uint32_t *hw;
+
+	vp->insns = realloc(vp->insns, ++vp->nr_insns * sizeof(*vpc->vpi));
+	vpc->vpi = &vp->insns[vp->nr_insns - 1];
+	memset(vpc->vpi, 0, sizeof(*vpc->vpi));
+	vpc->vpi->const_index = -1;
+
+	hw = vpc->vpi->data;
+
+	hw[0] |= (NV40_VP_INST_COND_TR << NV40_VP_INST_COND_SHIFT);
+	hw[0] |= ((0 << NV40_VP_INST_COND_SWZ_X_SHIFT) |
+		  (1 << NV40_VP_INST_COND_SWZ_Y_SHIFT) |
+		  (2 << NV40_VP_INST_COND_SWZ_Z_SHIFT) |
+		  (3 << NV40_VP_INST_COND_SWZ_W_SHIFT));
+
+	if (slot == 0) {
+		hw[1] |= (op << NV40_VP_INST_VEC_OPCODE_SHIFT);
+		hw[3] |= NV40_VP_INST_SCA_DEST_TEMP_MASK;
+		hw[3] |= (mask << NV40_VP_INST_VEC_WRITEMASK_SHIFT);
+	} else {
+		hw[1] |= (op << NV40_VP_INST_SCA_OPCODE_SHIFT);
+		hw[0] |= (NV40_VP_INST_VEC_DEST_TEMP_MASK | (1 << 20));
+		hw[3] |= (mask << NV40_VP_INST_SCA_WRITEMASK_SHIFT);
+	}
+
+	emit_dst(vpc, hw, slot, dst);
+	emit_src(vpc, hw, 0, s0);
+	emit_src(vpc, hw, 1, s1);
+	emit_src(vpc, hw, 2, s2);
+}
+
+static INLINE struct nv40_sreg
+tgsi_src(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc) {
+	struct nv40_sreg src;
+
+	switch (fsrc->SrcRegister.File) {
+	case TGSI_FILE_INPUT:
+		src = nv40_sr(NV40SR_INPUT, fsrc->SrcRegister.Index);
+		break;
+	case TGSI_FILE_CONSTANT:
+		src = constant(vpc, fsrc->SrcRegister.Index, 0, 0, 0, 0);
+		break;
+	case TGSI_FILE_IMMEDIATE:
+		src = vpc->imm[fsrc->SrcRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		src = vpc->r_temp[fsrc->SrcRegister.Index];
+		break;
+	default:
+		NOUVEAU_ERR("bad src file\n");
+		break;
+	}
+
+	src.abs = fsrc->SrcRegisterExtMod.Absolute;
+	src.negate = fsrc->SrcRegister.Negate;
+	src.swz[0] = fsrc->SrcRegister.SwizzleX;
+	src.swz[1] = fsrc->SrcRegister.SwizzleY;
+	src.swz[2] = fsrc->SrcRegister.SwizzleZ;
+	src.swz[3] = fsrc->SrcRegister.SwizzleW;
+	return src;
+}
+
+static INLINE struct nv40_sreg
+tgsi_dst(struct nv40_vpc *vpc, const struct tgsi_full_dst_register *fdst) {
+	struct nv40_sreg dst;
+
+	switch (fdst->DstRegister.File) {
+	case TGSI_FILE_OUTPUT:
+		dst = vpc->r_result[fdst->DstRegister.Index];
+		break;
+	case TGSI_FILE_TEMPORARY:
+		dst = vpc->r_temp[fdst->DstRegister.Index];
+		break;
+	case TGSI_FILE_ADDRESS:
+		dst = vpc->r_address[fdst->DstRegister.Index];
+		break;
+	default:
+		NOUVEAU_ERR("bad dst file\n");
+		break;
+	}
+
+	return dst;
+}
+
+static INLINE int
+tgsi_mask(uint tgsi)
+{
+	int mask = 0;
+
+	if (tgsi & TGSI_WRITEMASK_X) mask |= MASK_X;
+	if (tgsi & TGSI_WRITEMASK_Y) mask |= MASK_Y;
+	if (tgsi & TGSI_WRITEMASK_Z) mask |= MASK_Z;
+	if (tgsi & TGSI_WRITEMASK_W) mask |= MASK_W;
+	return mask;
+}
+
+static boolean
+src_native_swz(struct nv40_vpc *vpc, const struct tgsi_full_src_register *fsrc,
+	       struct nv40_sreg *src)
+{
+	const struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0);
+	struct nv40_sreg tgsi = tgsi_src(vpc, fsrc);
+	uint mask = 0, zero_mask = 0, one_mask = 0, neg_mask = 0;
+	uint neg[4] = { fsrc->SrcRegisterExtSwz.NegateX,
+			fsrc->SrcRegisterExtSwz.NegateY,
+			fsrc->SrcRegisterExtSwz.NegateZ,
+			fsrc->SrcRegisterExtSwz.NegateW };
+	uint c;
+
+	for (c = 0; c < 4; c++) {
+		switch (tgsi_util_get_full_src_register_extswizzle(fsrc, c)) {
+		case TGSI_EXTSWIZZLE_X:
+		case TGSI_EXTSWIZZLE_Y:
+		case TGSI_EXTSWIZZLE_Z:
+		case TGSI_EXTSWIZZLE_W:
+			mask |= tgsi_mask(1 << c);
+			break;
+		case TGSI_EXTSWIZZLE_ZERO:
+			zero_mask |= tgsi_mask(1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		case TGSI_EXTSWIZZLE_ONE:
+			one_mask |= tgsi_mask(1 << c);
+			tgsi.swz[c] = SWZ_X;
+			break;
+		default:
+			assert(0);
+		}
+
+		if (!tgsi.negate && neg[c])
+			neg_mask |= tgsi_mask(1 << c);
+	}
+
+	if (mask == MASK_ALL && !neg_mask)
+		return TRUE;
+
+	*src = temp(vpc);
+
+	if (mask)
+		arith(vpc, 0, OP_MOV, *src, mask, tgsi, none, none);
+
+	if (zero_mask)
+		arith(vpc, 0, OP_SFL, *src, zero_mask, *src, none, none);
+
+	if (one_mask)
+		arith(vpc, 0, OP_STR, *src, one_mask, *src, none, none);
+
+	if (neg_mask) {
+		struct nv40_sreg one = temp(vpc);
+		arith(vpc, 0, OP_STR, one, neg_mask, one, none, none);
+		arith(vpc, 0, OP_MUL, *src, neg_mask, *src, neg(one), none);
+	}
+
+	return FALSE;
+}
+
+static boolean
+nv40_vertprog_parse_instruction(struct nv40_vpc *vpc,
+				const struct tgsi_full_instruction *finst)
+{
+	struct nv40_sreg src[3], dst, tmp;
+	struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0);
+	int mask;
+	int ai = -1, ci = -1, ii = -1;
+	int i;
+
+	if (finst->Instruction.Opcode == TGSI_OPCODE_END)
+		return TRUE;
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+		if (fsrc->SrcRegister.File == TGSI_FILE_TEMPORARY) {
+			src[i] = tgsi_src(vpc, fsrc);
+		}
+	}
+
+	for (i = 0; i < finst->Instruction.NumSrcRegs; i++) {
+		const struct tgsi_full_src_register *fsrc;
+
+		fsrc = &finst->FullSrcRegisters[i];
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+		case TGSI_FILE_CONSTANT:
+		case TGSI_FILE_TEMPORARY:
+			if (!src_native_swz(vpc, fsrc, &src[i]))
+				continue;
+			break;
+		default:
+			break;
+		}
+
+		switch (fsrc->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			if (ai == -1 || ai == fsrc->SrcRegister.Index) {
+				ai = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_CONSTANT:
+			if ((ci == -1 && ii == -1) ||
+			    ci == fsrc->SrcRegister.Index) {
+				ci = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_IMMEDIATE:
+			if ((ci == -1 && ii == -1) ||
+			    ii == fsrc->SrcRegister.Index) {
+				ii = fsrc->SrcRegister.Index;
+				src[i] = tgsi_src(vpc, fsrc);
+			} else {
+				src[i] = temp(vpc);
+				arith(vpc, 0, OP_MOV, src[i], MASK_ALL,
+				      tgsi_src(vpc, fsrc), none, none);
+			}
+			break;
+		case TGSI_FILE_TEMPORARY:
+			/* handled above */
+			break;
+		default:
+			NOUVEAU_ERR("bad src file\n");
+			return FALSE;
+		}
+	}
+
+	dst  = tgsi_dst(vpc, &finst->FullDstRegisters[0]);
+	mask = tgsi_mask(finst->FullDstRegisters[0].DstRegister.WriteMask);
+
+	switch (finst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		arith(vpc, 0, OP_MOV, dst, mask, abs(src[0]), none, none);
+		break;
+	case TGSI_OPCODE_ADD:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, src[1]);
+		break;
+	case TGSI_OPCODE_ARL:
+		arith(vpc, 0, OP_ARL, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_DP3:
+		arith(vpc, 0, OP_DP3, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DP4:
+		arith(vpc, 0, OP_DP4, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DPH:
+		arith(vpc, 0, OP_DPH, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_DST:
+		arith(vpc, 0, OP_DST, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_EX2:
+		arith(vpc, 1, OP_EX2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_EXP:
+		arith(vpc, 1, OP_EXP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_FLR:
+		arith(vpc, 0, OP_FLR, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_FRC:
+		arith(vpc, 0, OP_FRC, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_LG2:
+		arith(vpc, 1, OP_LG2, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LIT:
+		arith(vpc, 1, OP_LIT, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_LOG:
+		arith(vpc, 1, OP_LOG, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_MAD:
+		arith(vpc, 0, OP_MAD, dst, mask, src[0], src[1], src[2]);
+		break;
+	case TGSI_OPCODE_MAX:
+		arith(vpc, 0, OP_MAX, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MIN:
+		arith(vpc, 0, OP_MIN, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_MOV:
+		arith(vpc, 0, OP_MOV, dst, mask, src[0], none, none);
+		break;
+	case TGSI_OPCODE_MUL:
+		arith(vpc, 0, OP_MUL, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_POW:
+		tmp = temp(vpc);
+		arith(vpc, 1, OP_LG2, tmp, MASK_X, none, none,
+		      swz(src[0], X, X, X, X));
+		arith(vpc, 0, OP_MUL, tmp, MASK_X, swz(tmp, X, X, X, X),
+		      swz(src[1], X, X, X, X), none);
+		arith(vpc, 1, OP_EX2, dst, mask, none, none,
+		      swz(tmp, X, X, X, X));
+		break;
+	case TGSI_OPCODE_RCP:
+		arith(vpc, 1, OP_RCP, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_RET:
+		break;
+	case TGSI_OPCODE_RSQ:
+		arith(vpc, 1, OP_RSQ, dst, mask, none, none, src[0]);
+		break;
+	case TGSI_OPCODE_SGE:
+		arith(vpc, 0, OP_SGE, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SLT:
+		arith(vpc, 0, OP_SLT, dst, mask, src[0], src[1], none);
+		break;
+	case TGSI_OPCODE_SUB:
+		arith(vpc, 0, OP_ADD, dst, mask, src[0], none, neg(src[1]));
+		break;
+	case TGSI_OPCODE_XPD:
+		tmp = temp(vpc);
+		arith(vpc, 0, OP_MUL, tmp, mask,
+		      swz(src[0], Z, X, Y, Y), swz(src[1], Y, Z, X, X), none);
+		arith(vpc, 0, OP_MAD, dst, (mask & ~MASK_W),
+		      swz(src[0], Y, Z, X, X), swz(src[1], Z, X, Y, Y),
+		      neg(tmp));
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", finst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	release_temps(vpc);
+	return TRUE;
+}
+
+static boolean
+nv40_vertprog_parse_decl_output(struct nv40_vpc *vpc,
+				const struct tgsi_full_declaration *fdec)
+{
+	unsigned idx = fdec->DeclarationRange.First;
+	int hw;
+
+	switch (fdec->Semantic.SemanticName) {
+	case TGSI_SEMANTIC_POSITION:
+		hw = NV40_VP_INST_DEST_POS;
+		vpc->hpos_idx = idx;
+		break;
+	case TGSI_SEMANTIC_COLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV40_VP_INST_DEST_COL0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV40_VP_INST_DEST_COL1;
+		} else {
+			NOUVEAU_ERR("bad colour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_BCOLOR:
+		if (fdec->Semantic.SemanticIndex == 0) {
+			hw = NV40_VP_INST_DEST_BFC0;
+		} else
+		if (fdec->Semantic.SemanticIndex == 1) {
+			hw = NV40_VP_INST_DEST_BFC1;
+		} else {
+			NOUVEAU_ERR("bad bcolour semantic index\n");
+			return FALSE;
+		}
+		break;
+	case TGSI_SEMANTIC_FOG:
+		hw = NV40_VP_INST_DEST_FOGC;
+		break;
+	case TGSI_SEMANTIC_PSIZE:
+		hw = NV40_VP_INST_DEST_PSZ;
+		break;
+	case TGSI_SEMANTIC_GENERIC:
+		if (fdec->Semantic.SemanticIndex <= 7) {
+			hw = NV40_VP_INST_DEST_TC(fdec->Semantic.SemanticIndex);
+		} else {
+			NOUVEAU_ERR("bad generic semantic index\n");
+			return FALSE;
+		}
+		break;
+	default:
+		NOUVEAU_ERR("bad output semantic\n");
+		return FALSE;
+	}
+
+	vpc->r_result[idx] = nv40_sr(NV40SR_OUTPUT, hw);
+	return TRUE;
+}
+
+static boolean
+nv40_vertprog_prepare(struct nv40_vpc *vpc)
+{
+	struct tgsi_parse_context p;
+	int high_temp = -1, high_addr = -1, nr_imm = 0, i;
+
+	tgsi_parse_init(&p, vpc->vp->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch(tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+			nr_imm++;
+			break;
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *fdec;
+
+			fdec = &p.FullToken.FullDeclaration;
+			switch (fdec->Declaration.File) {
+			case TGSI_FILE_TEMPORARY:
+				if (fdec->DeclarationRange.Last > high_temp) {
+					high_temp =
+						fdec->DeclarationRange.Last;
+				}
+				break;
+#if 0 /* this would be nice.. except gallium doesn't track it */
+			case TGSI_FILE_ADDRESS:
+				if (fdec->DeclarationRange.Last > high_addr) {
+					high_addr =
+						fdec->DeclarationRange.Last;
+				}
+				break;
+#endif
+			case TGSI_FILE_OUTPUT:
+				if (!nv40_vertprog_parse_decl_output(vpc, fdec))
+					return FALSE;
+				break;
+			default:
+				break;
+			}
+		}
+			break;
+#if 1 /* yay, parse instructions looking for address regs instead */
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+			const struct tgsi_full_dst_register *fdst;
+
+			finst = &p.FullToken.FullInstruction;
+			fdst = &finst->FullDstRegisters[0];
+
+			if (fdst->DstRegister.File == TGSI_FILE_ADDRESS) {
+				if (fdst->DstRegister.Index > high_addr)
+					high_addr = fdst->DstRegister.Index;
+			}
+		
+		}
+			break;
+#endif
+		default:
+			break;
+		}
+	}
+	tgsi_parse_free(&p);
+
+	if (nr_imm) {
+		vpc->imm = CALLOC(nr_imm, sizeof(struct nv40_sreg));
+		assert(vpc->imm);
+	}
+
+	if (++high_temp) {
+		vpc->r_temp = CALLOC(high_temp, sizeof(struct nv40_sreg));
+		for (i = 0; i < high_temp; i++)
+			vpc->r_temp[i] = temp(vpc);
+	}
+
+	if (++high_addr) {
+		vpc->r_address = CALLOC(high_addr, sizeof(struct nv40_sreg));
+		for (i = 0; i < high_addr; i++)
+			vpc->r_address[i] = temp(vpc);
+	}
+
+	vpc->r_temps_discard = 0;
+	return TRUE;
+}
+
+static void
+nv40_vertprog_translate(struct nv40_context *nv40,
+			struct nv40_vertex_program *vp)
+{
+	struct tgsi_parse_context parse;
+	struct nv40_vpc *vpc = NULL;
+	struct nv40_sreg none = nv40_sr(NV40SR_NONE, 0);
+	int i;
+
+	vpc = CALLOC(1, sizeof(struct nv40_vpc));
+	if (!vpc)
+		return;
+	vpc->vp = vp;
+
+	if (!nv40_vertprog_prepare(vpc)) {
+		FREE(vpc);
+		return;
+	}
+
+	/* Redirect post-transform vertex position to a temp if user clip
+	 * planes are enabled.  We need to append code the the vtxprog
+	 * to handle clip planes later.
+	 */
+	if (vp->ucp.nr)  {
+		vpc->r_result[vpc->hpos_idx] = temp(vpc);
+		vpc->r_temps_discard = 0;
+	}
+
+	tgsi_parse_init(&parse, vp->pipe.tokens);
+
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		tgsi_parse_token(&parse);
+
+		switch (parse.FullToken.Token.Type) {
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			const struct tgsi_full_immediate *imm;
+
+			imm = &parse.FullToken.FullImmediate;
+			assert(imm->Immediate.DataType == TGSI_IMM_FLOAT32);
+//			assert(imm->Immediate.Size == 4);
+			vpc->imm[vpc->nr_imm++] =
+				constant(vpc, -1,
+					 imm->u.ImmediateFloat32[0].Float,
+					 imm->u.ImmediateFloat32[1].Float,
+					 imm->u.ImmediateFloat32[2].Float,
+					 imm->u.ImmediateFloat32[3].Float);
+		}
+			break;
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+		{
+			const struct tgsi_full_instruction *finst;
+			finst = &parse.FullToken.FullInstruction;
+			if (!nv40_vertprog_parse_instruction(vpc, finst))
+				goto out_err;
+		}
+			break;
+		default:
+			break;
+		}
+	}
+
+	/* Write out HPOS if it was redirected to a temp earlier */
+	if (vpc->r_result[vpc->hpos_idx].type != NV40SR_OUTPUT) {
+		struct nv40_sreg hpos = nv40_sr(NV40SR_OUTPUT,
+						NV40_VP_INST_DEST_POS);
+		struct nv40_sreg htmp = vpc->r_result[vpc->hpos_idx];
+
+		arith(vpc, 0, OP_MOV, hpos, MASK_ALL, htmp, none, none);
+	}
+
+	/* Insert code to handle user clip planes */
+	for (i = 0; i < vp->ucp.nr; i++) {
+		struct nv40_sreg cdst = nv40_sr(NV40SR_OUTPUT,
+						NV40_VP_INST_DEST_CLIP(i));
+		struct nv40_sreg ceqn = constant(vpc, -1,
+						 nv40->clip.ucp[i][0],
+						 nv40->clip.ucp[i][1],
+						 nv40->clip.ucp[i][2],
+						 nv40->clip.ucp[i][3]);
+		struct nv40_sreg htmp = vpc->r_result[vpc->hpos_idx];
+		unsigned mask;
+
+		switch (i) {
+		case 0: case 3: mask = MASK_Y; break;
+		case 1: case 4: mask = MASK_Z; break;
+		case 2: case 5: mask = MASK_W; break;
+		default:
+			NOUVEAU_ERR("invalid clip dist #%d\n", i);
+			goto out_err;
+		}
+
+		arith(vpc, 0, OP_DP4, cdst, mask, htmp, ceqn, none);
+	}
+
+	vp->insns[vp->nr_insns - 1].data[3] |= NV40_VP_INST_LAST;
+	vp->translated = TRUE;
+out_err:
+	tgsi_parse_free(&parse);
+	if (vpc->r_temp)
+		FREE(vpc->r_temp); 
+	if (vpc->r_address)
+		FREE(vpc->r_address); 
+	if (vpc->imm)	
+		FREE(vpc->imm); 
+	FREE(vpc);
+}
+
+static boolean
+nv40_vertprog_validate(struct nv40_context *nv40)
+{ 
+	struct nouveau_winsys *nvws = nv40->nvws;
+	struct pipe_winsys *ws = nv40->pipe.winsys;
+	struct nouveau_grobj *curie = nv40->screen->curie;
+	struct nv40_vertex_program *vp;
+	struct pipe_buffer *constbuf;
+	boolean upload_code = FALSE, upload_data = FALSE;
+	int i;
+
+	if (nv40->render_mode == HW) {
+		vp = nv40->vertprog;
+		constbuf = nv40->constbuf[PIPE_SHADER_VERTEX];
+
+		if ((nv40->dirty & NV40_NEW_UCP) ||
+		    memcmp(&nv40->clip, &vp->ucp, sizeof(vp->ucp))) {
+			nv40_vertprog_destroy(nv40, vp);
+			memcpy(&vp->ucp, &nv40->clip, sizeof(vp->ucp));
+		}
+	} else {
+		vp = nv40->swtnl.vertprog;
+		constbuf = NULL;
+	}
+
+	/* Translate TGSI shader into hw bytecode */
+	if (vp->translated)
+		goto check_gpu_resources;
+
+	nv40->fallback_swtnl &= ~NV40_NEW_VERTPROG;
+	nv40_vertprog_translate(nv40, vp);
+	if (!vp->translated) {
+		nv40->fallback_swtnl |= NV40_NEW_VERTPROG;
+		return FALSE;
+	}
+
+check_gpu_resources:
+	/* Allocate hw vtxprog exec slots */
+	if (!vp->exec) {
+		struct nouveau_resource *heap = nv40->screen->vp_exec_heap;
+		struct nouveau_stateobj *so;
+		uint vplen = vp->nr_insns;
+
+		if (nvws->res_alloc(heap, vplen, vp, &vp->exec)) {
+			while (heap->next && heap->size < vplen) {
+				struct nv40_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->exec);
+			}
+
+			if (nvws->res_alloc(heap, vplen, vp, &vp->exec))
+				assert(0);
+		}
+
+		so = so_new(7, 0);
+		so_method(so, curie, NV40TCL_VP_START_FROM_ID, 1);
+		so_data  (so, vp->exec->start);
+		so_method(so, curie, NV40TCL_VP_ATTRIB_EN, 2);
+		so_data  (so, vp->ir);
+		so_data  (so, vp->or);
+		so_method(so, curie,  NV40TCL_CLIP_PLANE_ENABLE, 1);
+		so_data  (so, vp->clip_ctrl);
+		so_ref(so, &vp->so);
+
+		upload_code = TRUE;
+	}
+
+	/* Allocate hw vtxprog const slots */
+	if (vp->nr_consts && !vp->data) {
+		struct nouveau_resource *heap = nv40->screen->vp_data_heap;
+
+		if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data)) {
+			while (heap->next && heap->size < vp->nr_consts) {
+				struct nv40_vertex_program *evict;
+				
+				evict = heap->next->priv;
+				nvws->res_free(&evict->data);
+			}
+
+			if (nvws->res_alloc(heap, vp->nr_consts, vp, &vp->data))
+				assert(0);
+		}
+
+		/*XXX: handle this some day */
+		assert(vp->data->start >= vp->data_start_min);
+
+		upload_data = TRUE;
+		if (vp->data_start != vp->data->start)
+			upload_code = TRUE;
+	}
+
+	/* If exec or data segments moved we need to patch the program to
+	 * fixup offsets and register IDs.
+	 */
+	if (vp->exec_start != vp->exec->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv40_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->has_branch_offset) {
+				assert(0);
+			}
+		}
+
+		vp->exec_start = vp->exec->start;
+	}
+
+	if (vp->nr_consts && vp->data_start != vp->data->start) {
+		for (i = 0; i < vp->nr_insns; i++) {
+			struct nv40_vertex_program_exec *vpi = &vp->insns[i];
+
+			if (vpi->const_index >= 0) {
+				vpi->data[1] &= ~NV40_VP_INST_CONST_SRC_MASK;
+				vpi->data[1] |=
+					(vpi->const_index + vp->data->start) <<
+					NV40_VP_INST_CONST_SRC_SHIFT;
+
+			}
+		}
+
+		vp->data_start = vp->data->start;
+	}
+
+	/* Update + Upload constant values */
+	if (vp->nr_consts) {
+		float *map = NULL;
+
+		if (constbuf) {
+			map = ws->buffer_map(ws, constbuf,
+					     PIPE_BUFFER_USAGE_CPU_READ);
+		}
+
+		for (i = 0; i < vp->nr_consts; i++) {
+			struct nv40_vertex_program_data *vpd = &vp->consts[i];
+
+			if (vpd->index >= 0) {
+				if (!upload_data &&
+				    !memcmp(vpd->value, &map[vpd->index * 4],
+					    4 * sizeof(float)))
+					continue;
+				memcpy(vpd->value, &map[vpd->index * 4],
+				       4 * sizeof(float));
+			}
+
+			BEGIN_RING(curie, NV40TCL_VP_UPLOAD_CONST_ID, 5);
+			OUT_RING  (i + vp->data->start);
+			OUT_RINGp ((uint32_t *)vpd->value, 4);
+		}
+
+		if (constbuf)
+			ws->buffer_unmap(ws, constbuf);
+	}
+
+	/* Upload vtxprog */
+	if (upload_code) {
+#if 0
+		for (i = 0; i < vp->nr_insns; i++) {
+			NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[0]);
+			NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[1]);
+			NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[2]);
+			NOUVEAU_MSG("VP %d: 0x%08x\n", i, vp->insns[i].data[3]);
+		}
+#endif
+		BEGIN_RING(curie, NV40TCL_VP_UPLOAD_FROM_ID, 1);
+		OUT_RING  (vp->exec->start);
+		for (i = 0; i < vp->nr_insns; i++) {
+			BEGIN_RING(curie, NV40TCL_VP_UPLOAD_INST(0), 4);
+			OUT_RINGp (vp->insns[i].data, 4);
+		}
+	}
+
+	if (vp->so != nv40->state.hw[NV40_STATE_VERTPROG]) {
+		so_ref(vp->so, &nv40->state.hw[NV40_STATE_VERTPROG]);
+		return TRUE;
+	}
+
+	return FALSE;
+}
+
+void
+nv40_vertprog_destroy(struct nv40_context *nv40, struct nv40_vertex_program *vp)
+{
+	struct nouveau_winsys *nvws = nv40->screen->nvws;
+
+	vp->translated = FALSE;
+
+	if (vp->nr_insns) {
+		FREE(vp->insns);
+		vp->insns = NULL;
+		vp->nr_insns = 0;
+	}
+
+	if (vp->nr_consts) {
+		FREE(vp->consts);
+		vp->consts = NULL;
+		vp->nr_consts = 0;
+	}
+
+	nvws->res_free(&vp->exec);
+	vp->exec_start = 0;
+	nvws->res_free(&vp->data);
+	vp->data_start = 0;
+	vp->data_start_min = 0;
+
+	vp->ir = vp->or = vp->clip_ctrl = 0;
+	so_ref(NULL, &vp->so);
+}
+
+struct nv40_state_entry nv40_state_vertprog = {
+	.validate = nv40_vertprog_validate,
+	.dirty = {
+		.pipe = NV40_NEW_VERTPROG | NV40_NEW_UCP,
+		.hw = NV40_STATE_VERTPROG,
+	}
+};
+
diff --git a/src/gallium/drivers/nv50/Makefile b/src/gallium/drivers/nv50/Makefile
new file mode 100644
index 00000000000..be30400c039
--- /dev/null
+++ b/src/gallium/drivers/nv50/Makefile
@@ -0,0 +1,29 @@
+TOP = ../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nv50
+
+DRIVER_SOURCES = \
+	nv50_clear.c \
+	nv50_context.c \
+	nv50_draw.c \
+	nv50_miptree.c \
+	nv50_query.c \
+	nv50_program.c \
+	nv50_screen.c \
+	nv50_state.c \
+	nv50_state_validate.c \
+	nv50_surface.c \
+	nv50_tex.c \
+	nv50_vbo.c
+
+C_SOURCES = \
+	$(COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../../Makefile.template
+
+symlinks:
+
diff --git a/src/gallium/drivers/nv50/nv50_clear.c b/src/gallium/drivers/nv50/nv50_clear.c
new file mode 100644
index 00000000000..a31a42d6b54
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_clear.c
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+
+#include "nv50_context.h"
+
+void
+nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+	   unsigned clearValue)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct pipe_framebuffer_state fb, s_fb = nv50->framebuffer;
+	struct pipe_scissor_state sc, s_sc = nv50->scissor;
+	unsigned dirty = nv50->dirty;
+
+	nv50->dirty = 0;
+
+	if (ps->format == PIPE_FORMAT_Z24S8_UNORM ||
+	    ps->format == PIPE_FORMAT_Z16_UNORM) {
+		fb.num_cbufs = 0;
+		fb.zsbuf = ps;
+	} else {
+		fb.num_cbufs = 1;
+		fb.cbufs[0] = ps;
+		fb.zsbuf = NULL;
+	}
+	fb.width = ps->width;
+	fb.height = ps->height;
+	pipe->set_framebuffer_state(pipe, &fb);
+
+	sc.minx = sc.miny = 0;
+	sc.maxx = fb.width;
+	sc.maxy = fb.height;
+	pipe->set_scissor_state(pipe, &sc);
+
+	nv50_state_validate(nv50);
+
+	switch (ps->format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+		BEGIN_RING(tesla, 0x0d80, 4);
+		OUT_RINGf (ubyte_to_float((clearValue >> 16) & 0xff));
+		OUT_RINGf (ubyte_to_float((clearValue >>  8) & 0xff));
+		OUT_RINGf (ubyte_to_float((clearValue >>  0) & 0xff));
+		OUT_RINGf (ubyte_to_float((clearValue >> 24) & 0xff));
+		BEGIN_RING(tesla, 0x19d0, 1);
+		OUT_RING  (0x3c);
+		break;
+	case PIPE_FORMAT_Z24S8_UNORM:
+		BEGIN_RING(tesla, 0x0d90, 1);
+		OUT_RINGf ((float)(clearValue >> 8) * (1.0 / 16777215.0));
+		BEGIN_RING(tesla, 0x0da0, 1);
+		OUT_RING  (clearValue & 0xff);
+		BEGIN_RING(tesla, 0x19d0, 1);
+		OUT_RING  (0x03);
+		break;
+	default:
+		pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height,
+				   clearValue);
+		break;
+	}
+
+	pipe->set_framebuffer_state(pipe, &s_fb);
+	pipe->set_scissor_state(pipe, &s_sc);
+	nv50->dirty |= dirty;
+
+	ps->status = PIPE_SURFACE_STATUS_CLEAR;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c
new file mode 100644
index 00000000000..b02c53f2095
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_context.c
@@ -0,0 +1,90 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "draw/draw_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+
+#include "nv50_context.h"
+#include "nv50_screen.h"
+
+static void
+nv50_flush(struct pipe_context *pipe, unsigned flags,
+	   struct pipe_fence_handle **fence)
+{
+	struct nv50_context *nv50 = (struct nv50_context *)pipe;
+	
+	FIRE_RING(fence);
+}
+
+static void
+nv50_destroy(struct pipe_context *pipe)
+{
+	struct nv50_context *nv50 = (struct nv50_context *)pipe;
+
+	draw_destroy(nv50->draw);
+	FREE(nv50);
+}
+
+
+static void
+nv50_set_edgeflags(struct pipe_context *pipe, const unsigned *bitfield)
+{
+}
+
+struct pipe_context *
+nv50_create(struct pipe_screen *pscreen, unsigned pctx_id)
+{
+	struct pipe_winsys *pipe_winsys = pscreen->winsys;
+	struct nv50_screen *screen = nv50_screen(pscreen);
+	struct nv50_context *nv50;
+
+	nv50 = CALLOC_STRUCT(nv50_context);
+	if (!nv50)
+		return NULL;
+	nv50->screen = screen;
+	nv50->pctx_id = pctx_id;
+
+	nv50->pipe.winsys = pipe_winsys;
+	nv50->pipe.screen = pscreen;
+
+	nv50->pipe.destroy = nv50_destroy;
+
+	nv50->pipe.set_edgeflags = nv50_set_edgeflags;
+	nv50->pipe.draw_arrays = nv50_draw_arrays;
+	nv50->pipe.draw_elements = nv50_draw_elements;
+	nv50->pipe.clear = nv50_clear;
+
+	nv50->pipe.flush = nv50_flush;
+
+	nv50_init_surface_functions(nv50);
+	nv50_init_state_functions(nv50);
+	nv50_init_query_functions(nv50);
+
+	nv50->draw = draw_create();
+	assert(nv50->draw);
+	draw_set_rasterize_stage(nv50->draw, nv50_draw_render_stage(nv50));
+
+	return &nv50->pipe;
+}
+
+		
diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h
new file mode 100644
index 00000000000..5d377f2d067
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_context.h
@@ -0,0 +1,185 @@
+#ifndef __NV50_CONTEXT_H__
+#define __NV50_CONTEXT_H__
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_compiler.h"
+
+#include "util/u_memory.h"
+#include "util/u_math.h"
+
+#include "draw/draw_vertex.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau/nouveau_gldefs.h"
+#include "nouveau/nouveau_stateobj.h"
+
+#define NOUVEAU_PUSH_CONTEXT(ctx)                                              \
+	struct nv50_screen *ctx = nv50->screen
+#include "nouveau/nouveau_push.h"
+
+#include "nv50_screen.h"
+#include "nv50_program.h"
+
+#define NOUVEAU_ERR(fmt, args...) \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);
+#define NOUVEAU_MSG(fmt, args...) \
+	fprintf(stderr, "nouveau: "fmt, ##args);
+
+/* Constant buffer assignment */
+#define NV50_CB_PMISC		0
+#define NV50_CB_PVP		1
+#define NV50_CB_PFP		2
+#define NV50_CB_PGP		3
+#define NV50_CB_TIC		4
+#define NV50_CB_TSC		5
+#define NV50_CB_PUPLOAD         6
+
+#define NV50_NEW_BLEND		(1 << 0)
+#define NV50_NEW_ZSA		(1 << 1)
+#define NV50_NEW_BLEND_COLOUR	(1 << 2)
+#define NV50_NEW_STIPPLE	(1 << 3)
+#define NV50_NEW_SCISSOR	(1 << 4)
+#define NV50_NEW_VIEWPORT	(1 << 5)
+#define NV50_NEW_RASTERIZER	(1 << 6)
+#define NV50_NEW_FRAMEBUFFER	(1 << 7)
+#define NV50_NEW_VERTPROG	(1 << 8)
+#define NV50_NEW_VERTPROG_CB	(1 << 9)
+#define NV50_NEW_FRAGPROG	(1 << 10)
+#define NV50_NEW_FRAGPROG_CB	(1 << 11)
+#define NV50_NEW_ARRAYS		(1 << 12)
+#define NV50_NEW_SAMPLER	(1 << 13)
+#define NV50_NEW_TEXTURE	(1 << 14)
+
+struct nv50_blend_stateobj {
+	struct pipe_blend_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv50_zsa_stateobj {
+	struct pipe_depth_stencil_alpha_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv50_rasterizer_stateobj {
+	struct pipe_rasterizer_state pipe;
+	struct nouveau_stateobj *so;
+};
+
+struct nv50_miptree {
+	struct pipe_texture base;
+	struct pipe_buffer *buffer;
+};
+
+static INLINE struct nv50_miptree *
+nv50_miptree(struct pipe_texture *pt)
+{
+	return (struct nv50_miptree *)pt;
+}
+
+struct nv50_surface {
+	struct pipe_surface base;
+	struct pipe_buffer *untiled;
+};
+
+static INLINE struct nv50_surface *
+nv50_surface(struct pipe_surface *pt)
+{
+	return (struct nv50_surface *)pt;
+}
+
+struct nv50_state {
+	unsigned dirty;
+
+	struct nouveau_stateobj *fb;
+	struct nouveau_stateobj *blend;
+	struct nouveau_stateobj *blend_colour;
+	struct nouveau_stateobj *zsa;
+	struct nouveau_stateobj *rast;
+	struct nouveau_stateobj *stipple;
+	struct nouveau_stateobj *scissor;
+	unsigned scissor_enabled;
+	struct nouveau_stateobj *viewport;
+	unsigned viewport_bypass;
+	struct nouveau_stateobj *tsc_upload;
+	struct nouveau_stateobj *tic_upload;
+	struct nouveau_stateobj *vertprog;
+	struct nouveau_stateobj *fragprog;
+	struct nouveau_stateobj *vtxfmt;
+	struct nouveau_stateobj *vtxbuf;
+};
+
+struct nv50_context {
+	struct pipe_context pipe;
+
+	struct nv50_screen *screen;
+	unsigned pctx_id;
+
+	struct draw_context *draw;
+
+	struct nv50_state state;
+
+	unsigned dirty;
+	struct nv50_blend_stateobj *blend;
+	struct nv50_zsa_stateobj *zsa;
+	struct nv50_rasterizer_stateobj *rasterizer;
+	struct pipe_blend_color blend_colour;
+	struct pipe_poly_stipple stipple;
+	struct pipe_scissor_state scissor;
+	struct pipe_viewport_state viewport;
+	struct pipe_framebuffer_state framebuffer;
+	struct nv50_program *vertprog;
+	struct nv50_program *fragprog;
+	struct pipe_buffer *constbuf[PIPE_SHADER_TYPES];
+	struct pipe_vertex_buffer vtxbuf[PIPE_MAX_ATTRIBS];
+	unsigned vtxbuf_nr;
+	struct pipe_vertex_element vtxelt[PIPE_MAX_ATTRIBS];
+	unsigned vtxelt_nr;
+	unsigned *sampler[PIPE_MAX_SAMPLERS];
+	unsigned sampler_nr;
+	struct nv50_miptree *miptree[PIPE_MAX_SAMPLERS];
+	unsigned miptree_nr;
+};
+
+static INLINE struct nv50_context *
+nv50_context(struct pipe_context *pipe)
+{
+	return (struct nv50_context *)pipe;
+}
+
+extern void nv50_init_surface_functions(struct nv50_context *nv50);
+extern void nv50_init_state_functions(struct nv50_context *nv50);
+extern void nv50_init_query_functions(struct nv50_context *nv50);
+
+extern void nv50_screen_init_miptree_functions(struct pipe_screen *pscreen);
+
+/* nv50_draw.c */
+extern struct draw_stage *nv50_draw_render_stage(struct nv50_context *nv50);
+
+/* nv50_vbo.c */
+extern boolean nv50_draw_arrays(struct pipe_context *, unsigned mode,
+				unsigned start, unsigned count);
+extern boolean nv50_draw_elements(struct pipe_context *pipe,
+				  struct pipe_buffer *indexBuffer,
+				  unsigned indexSize,
+				  unsigned mode, unsigned start,
+				  unsigned count);
+extern void nv50_vbo_validate(struct nv50_context *nv50);
+
+/* nv50_clear.c */
+extern void nv50_clear(struct pipe_context *pipe, struct pipe_surface *ps,
+		       unsigned clearValue);
+
+/* nv50_program.c */
+extern void nv50_vertprog_validate(struct nv50_context *nv50);
+extern void nv50_fragprog_validate(struct nv50_context *nv50);
+extern void nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p);
+
+/* nv50_state_validate.c */
+extern boolean nv50_state_validate(struct nv50_context *nv50);
+
+/* nv50_tex.c */
+extern void nv50_tex_validate(struct nv50_context *);
+
+#endif
diff --git a/src/gallium/drivers/nv50/nv50_draw.c b/src/gallium/drivers/nv50/nv50_draw.c
new file mode 100644
index 00000000000..2f6f607261e
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_draw.c
@@ -0,0 +1,89 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "draw/draw_pipe.h"
+
+#include "nv50_context.h"
+
+struct nv50_render_stage {
+	struct draw_stage stage;
+	struct nv50_context *nv50;
+};
+
+static INLINE struct nv50_render_stage *
+nv50_render_stage(struct draw_stage *stage)
+{
+	return (struct nv50_render_stage *)stage;
+}
+
+static void
+nv50_render_point(struct draw_stage *stage, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv50_render_line(struct draw_stage *stage, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv50_render_tri(struct draw_stage *stage, struct prim_header *prim)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv50_render_flush(struct draw_stage *stage, unsigned flags)
+{
+}
+
+static void
+nv50_render_reset_stipple_counter(struct draw_stage *stage)
+{
+	NOUVEAU_ERR("\n");
+}
+
+static void
+nv50_render_destroy(struct draw_stage *stage)
+{
+	FREE(stage);
+}
+
+struct draw_stage *
+nv50_draw_render_stage(struct nv50_context *nv50)
+{
+	struct nv50_render_stage *rs = CALLOC_STRUCT(nv50_render_stage);
+
+	rs->nv50 = nv50;
+	rs->stage.draw = nv50->draw;
+	rs->stage.destroy = nv50_render_destroy;
+	rs->stage.point = nv50_render_point;
+	rs->stage.line = nv50_render_line;
+	rs->stage.tri = nv50_render_tri;
+	rs->stage.flush = nv50_render_flush;
+	rs->stage.reset_stipple_counter = nv50_render_reset_stipple_counter;
+
+	return &rs->stage;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c
new file mode 100644
index 00000000000..28a8bdc0fab
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_miptree.c
@@ -0,0 +1,135 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "nv50_context.h"
+
+static struct pipe_texture *
+nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
+{
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree);
+	unsigned usage, pitch;
+
+	mt->base = *pt;
+	mt->base.refcount = 1;
+	mt->base.screen = pscreen;
+
+	usage = PIPE_BUFFER_USAGE_PIXEL;
+	switch (pt->format) {
+	case PIPE_FORMAT_Z24S8_UNORM:
+	case PIPE_FORMAT_Z16_UNORM:
+		usage |= NOUVEAU_BUFFER_USAGE_ZETA;
+		break;
+	default:
+		break;
+	}
+
+	pitch = ((pt->width[0] + 63) & ~63) * pt->block.size;
+	/*XXX*/
+	pitch *= 2;
+
+	mt->buffer = ws->buffer_create(ws, 256, usage, pitch * pt->height[0]);
+	if (!mt->buffer) {
+		FREE(mt);
+		return NULL;
+	}
+
+	return &mt->base;
+}
+
+static void
+nv50_miptree_release(struct pipe_screen *pscreen, struct pipe_texture **ppt)
+{
+	struct pipe_texture *pt = *ppt;
+
+	*ppt = NULL;
+
+	if (--pt->refcount <= 0) {
+		struct nv50_miptree *mt = nv50_miptree(pt);
+
+		pipe_buffer_reference(pscreen, &mt->buffer, NULL);
+		FREE(mt);
+	}
+}
+
+static struct pipe_surface *
+nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
+			 unsigned face, unsigned level, unsigned zslice,
+			 unsigned flags)
+{
+	struct nv50_miptree *mt = nv50_miptree(pt);
+	struct nv50_surface *s;
+	struct pipe_surface *ps;
+
+	s = CALLOC_STRUCT(nv50_surface);
+	if (!s)
+		return NULL;
+	ps = &s->base;
+
+	ps->refcount = 1;
+	ps->winsys = pscreen->winsys;
+	ps->format = pt->format;
+	ps->width = pt->width[level];
+	ps->height = pt->height[level];
+	ps->block = pt->block;
+	ps->nblocksx = pt->nblocksx[level];
+	ps->nblocksy = pt->nblocksy[level];
+	ps->stride = ps->width * ps->block.size;
+	ps->offset = 0;
+	ps->usage = flags;
+	ps->status = PIPE_SURFACE_STATUS_DEFINED;
+
+	pipe_texture_reference(&ps->texture, pt);
+	pipe_buffer_reference(pscreen, &ps->buffer, mt->buffer);
+
+	return ps;
+}
+
+static void
+nv50_miptree_surface_del(struct pipe_screen *pscreen,
+			 struct pipe_surface **psurface)
+{
+	struct pipe_surface *ps = *psurface;
+	struct nv50_surface *s = nv50_surface(ps);
+
+	*psurface = NULL;
+
+	if (--ps->refcount <= 0) {
+		pipe_texture_reference(&ps->texture, NULL);
+		pipe_buffer_reference(pscreen, &ps->buffer, NULL);
+		FREE(s);
+	}
+}
+
+void
+nv50_screen_init_miptree_functions(struct pipe_screen *pscreen)
+{
+	pscreen->texture_create = nv50_miptree_create;
+	pscreen->texture_release = nv50_miptree_release;
+	pscreen->get_tex_surface = nv50_miptree_surface_new;
+	pscreen->tex_surface_release = nv50_miptree_surface_del;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
new file mode 100644
index 00000000000..d6fbdd18243
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -0,0 +1,1708 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_state.h"
+#include "pipe/p_inlines.h"
+
+#include "pipe/p_shader_tokens.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
+
+#include "nv50_context.h"
+
+#define NV50_SU_MAX_TEMP 64
+#define NV50_PROGRAM_DUMP
+
+/* ARL - gallium craps itself on progs/vp/arl.txt
+ *
+ * MSB - Like MAD, but MUL+SUB
+ * 	- Fuck it off, introduce a way to negate args for ops that
+ * 	  support it.
+ *
+ * Look into inlining IMMD for ops other than MOV (make it general?)
+ * 	- Maybe even relax restrictions a bit, can't do P_RESULT + P_IMMD,
+ * 	  but can emit to P_TEMP first - then MOV later. NVIDIA does this
+ *
+ * In ops such as ADD it's possible to construct a bad opcode in the !is_long()
+ * case, if the emit_src() causes the inst to suddenly become long.
+ *
+ * Verify half-insns work where expected - and force disable them where they
+ * don't work - MUL has it forcibly disabled atm as it fixes POW..
+ *
+ * FUCK! watch dst==src vectors, can overwrite components that are needed.
+ * 	ie. SUB R0, R0.yzxw, R0
+ *
+ * Things to check with renouveau:
+ * 	FP attr/result assignment - how?
+ * 		attrib
+ * 			- 0x16bc maps vp output onto fp hpos
+ * 			- 0x16c0 maps vp output onto fp col0
+ * 		result
+ * 			- colr always 0-3
+ * 			- depr always 4
+ * 0x16bc->0x16e8 --> some binding between vp/fp regs
+ * 0x16b8 --> VP output count
+ *
+ * 0x1298 --> "MOV rcol.x, fcol.y" "MOV depr, fcol.y" = 0x00000005
+ * 	      "MOV rcol.x, fcol.y" = 0x00000004
+ * 0x19a8 --> as above but 0x00000100 and 0x00000000
+ * 	- 0x00100000 used when KIL used
+ * 0x196c --> as above but 0x00000011 and 0x00000000
+ *
+ * 0x1988 --> 0xXXNNNNNN
+ * 	- XX == FP high something
+ */
+struct nv50_reg {
+	enum {
+		P_TEMP,
+		P_ATTR,
+		P_RESULT,
+		P_CONST,
+		P_IMMD
+	} type;
+	int index;
+
+	int hw;
+	int neg;
+};
+
+struct nv50_pc {
+	struct nv50_program *p;
+
+	/* hw resources */
+	struct nv50_reg *r_temp[NV50_SU_MAX_TEMP];
+
+	/* tgsi resources */
+	struct nv50_reg *temp;
+	int temp_nr;
+	struct nv50_reg *attr;
+	int attr_nr;
+	struct nv50_reg *result;
+	int result_nr;
+	struct nv50_reg *param;
+	int param_nr;
+	struct nv50_reg *immd;
+	float *immd_buf;
+	int immd_nr;
+
+	struct nv50_reg *temp_temp[16];
+	unsigned temp_temp_nr;
+};
+
+static void
+alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg)
+{
+	int i;
+
+	if (reg->type == P_RESULT) {
+		if (pc->p->cfg.high_result < (reg->hw + 1))
+			pc->p->cfg.high_result = reg->hw + 1;
+	}
+
+	if (reg->type != P_TEMP)
+		return;
+
+	if (reg->hw >= 0) {
+		/*XXX: do this here too to catch FP temp-as-attr usage..
+		 *     not clean, but works */
+		if (pc->p->cfg.high_temp < (reg->hw + 1))
+			pc->p->cfg.high_temp = reg->hw + 1;
+		return;
+	}
+
+	for (i = 0; i < NV50_SU_MAX_TEMP; i++) {
+		if (!(pc->r_temp[i])) {
+			pc->r_temp[i] = reg;
+			reg->hw = i;
+			if (pc->p->cfg.high_temp < (i + 1))
+				pc->p->cfg.high_temp = i + 1;
+			return;
+		}
+	}
+
+	assert(0);
+}
+
+static struct nv50_reg *
+alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst)
+{
+	struct nv50_reg *r;
+	int i;
+
+	if (dst && dst->type == P_TEMP && dst->hw == -1)
+		return dst;
+
+	for (i = 0; i < NV50_SU_MAX_TEMP; i++) {
+		if (!pc->r_temp[i]) {
+			r = CALLOC_STRUCT(nv50_reg);
+			r->type = P_TEMP;
+			r->index = -1;
+			r->hw = i;
+			pc->r_temp[i] = r;
+			return r;
+		}
+	}
+
+	assert(0);
+	return NULL;
+}
+
+static void
+free_temp(struct nv50_pc *pc, struct nv50_reg *r)
+{
+	if (r->index == -1) {
+		unsigned hw = r->hw;
+
+		FREE(pc->r_temp[hw]);
+		pc->r_temp[hw] = NULL;
+	}
+}
+
+static struct nv50_reg *
+temp_temp(struct nv50_pc *pc)
+{
+	if (pc->temp_temp_nr >= 16)
+		assert(0);
+
+	pc->temp_temp[pc->temp_temp_nr] = alloc_temp(pc, NULL);
+	return pc->temp_temp[pc->temp_temp_nr++];
+}
+
+static void
+kill_temp_temp(struct nv50_pc *pc)
+{
+	int i;
+	
+	for (i = 0; i < pc->temp_temp_nr; i++)
+		free_temp(pc, pc->temp_temp[i]);
+	pc->temp_temp_nr = 0;
+}
+
+static int
+ctor_immd(struct nv50_pc *pc, float x, float y, float z, float w)
+{
+	pc->immd_buf = REALLOC(pc->immd_buf, (pc->immd_nr * r * sizeof(float)),
+			       (pc->immd_nr + 1) * 4 * sizeof(float));
+	pc->immd_buf[(pc->immd_nr * 4) + 0] = x;
+	pc->immd_buf[(pc->immd_nr * 4) + 1] = y;
+	pc->immd_buf[(pc->immd_nr * 4) + 2] = z;
+	pc->immd_buf[(pc->immd_nr * 4) + 3] = w;
+	
+	return pc->immd_nr++;
+}
+
+static struct nv50_reg *
+alloc_immd(struct nv50_pc *pc, float f)
+{
+	struct nv50_reg *r = CALLOC_STRUCT(nv50_reg);
+	unsigned hw;
+
+	hw = ctor_immd(pc, f, 0, 0, 0) * 4;
+	r->type = P_IMMD;
+	r->hw = hw;
+	r->index = -1;
+	return r;
+}
+
+static struct nv50_program_exec *
+exec(struct nv50_pc *pc)
+{
+	struct nv50_program_exec *e = CALLOC_STRUCT(nv50_program_exec);
+
+	e->param.index = -1;
+	return e;
+}
+
+static void
+emit(struct nv50_pc *pc, struct nv50_program_exec *e)
+{
+	struct nv50_program *p = pc->p;
+
+	if (p->exec_tail)
+		p->exec_tail->next = e;
+	if (!p->exec_head)
+		p->exec_head = e;
+	p->exec_tail = e;
+	p->exec_size += (e->inst[0] & 1) ? 2 : 1;
+}
+
+static INLINE void set_long(struct nv50_pc *, struct nv50_program_exec *);
+
+static boolean
+is_long(struct nv50_program_exec *e)
+{
+	if (e->inst[0] & 1)
+		return TRUE;
+	return FALSE;
+}
+
+static boolean
+is_immd(struct nv50_program_exec *e)
+{
+	if (is_long(e) && (e->inst[1] & 3) == 3)
+		return TRUE;
+	return FALSE;
+}
+
+static INLINE void
+set_pred(struct nv50_pc *pc, unsigned pred, unsigned idx,
+	 struct nv50_program_exec *e)
+{
+	set_long(pc, e);
+	e->inst[1] &= ~((0x1f << 7) | (0x3 << 12));
+	e->inst[1] |= (pred << 7) | (idx << 12);
+}
+
+static INLINE void
+set_pred_wr(struct nv50_pc *pc, unsigned on, unsigned idx,
+	    struct nv50_program_exec *e)
+{
+	set_long(pc, e);
+	e->inst[1] &= ~((0x3 << 4) | (1 << 6));
+	e->inst[1] |= (idx << 4) | (on << 6);
+}
+
+static INLINE void
+set_long(struct nv50_pc *pc, struct nv50_program_exec *e)
+{
+	if (is_long(e))
+		return;
+
+	e->inst[0] |= 1;
+	set_pred(pc, 0xf, 0, e);
+	set_pred_wr(pc, 0, 0, e);
+}
+
+static INLINE void
+set_dst(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_program_exec *e)
+{
+	if (dst->type == P_RESULT) {
+		set_long(pc, e);
+		e->inst[1] |= 0x00000008;
+	}
+
+	alloc_reg(pc, dst);
+	e->inst[0] |= (dst->hw << 2);
+}
+
+static INLINE void
+set_immd(struct nv50_pc *pc, struct nv50_reg *imm, struct nv50_program_exec *e)
+{
+	unsigned val = fui(pc->immd_buf[imm->hw]); /* XXX */
+
+	set_long(pc, e);
+	/*XXX: can't be predicated - bits overlap.. catch cases where both
+	 *     are required and avoid them. */
+	set_pred(pc, 0, 0, e);
+	set_pred_wr(pc, 0, 0, e);
+
+	e->inst[1] |= 0x00000002 | 0x00000001;
+	e->inst[0] |= (val & 0x3f) << 16;
+	e->inst[1] |= (val >> 6) << 2;
+}
+
+static void
+emit_interp(struct nv50_pc *pc, struct nv50_reg *dst,
+	    struct nv50_reg *src, struct nv50_reg *iv)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0x80000000;
+	set_dst(pc, dst, e);
+	alloc_reg(pc, src);
+	e->inst[0] |= (src->hw << 16);
+	if (iv) {
+		e->inst[0] |= (1 << 25);
+		alloc_reg(pc, iv);
+		e->inst[0] |= (iv->hw << 9);
+	}
+
+	emit(pc, e);
+}
+
+static void
+set_data(struct nv50_pc *pc, struct nv50_reg *src, unsigned m, unsigned s,
+	 struct nv50_program_exec *e)
+{
+	set_long(pc, e);
+#if 1
+	e->inst[1] |= (1 << 22);
+#else
+	if (src->type == P_IMMD) {
+		e->inst[1] |= (NV50_CB_PMISC << 22);
+	} else {
+		if (pc->p->type == PIPE_SHADER_VERTEX)
+			e->inst[1] |= (NV50_CB_PVP << 22);
+		else
+			e->inst[1] |= (NV50_CB_PFP << 22);
+	}
+#endif
+
+	e->param.index = src->hw;
+	e->param.shift = s;
+	e->param.mask = m << (s % 32);
+}
+
+static void
+emit_mov(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0x10000000;
+
+	set_dst(pc, dst, e);
+
+	if (0 && dst->type != P_RESULT && src->type == P_IMMD) {
+		set_immd(pc, src, e);
+		/*XXX: 32-bit, but steals part of "half" reg space - need to
+		 *     catch and handle this case if/when we do half-regs
+		 */
+		e->inst[0] |= 0x00008000;
+	} else
+	if (src->type == P_IMMD || src->type == P_CONST) {
+		set_long(pc, e);
+		set_data(pc, src, 0x7f, 9, e);
+		e->inst[1] |= 0x20000000; /* src0 const? */
+	} else {
+		if (src->type == P_ATTR) {
+			set_long(pc, e);
+			e->inst[1] |= 0x00200000;
+		}
+
+		alloc_reg(pc, src);
+		e->inst[0] |= (src->hw << 9);
+	}
+
+	/* We really should support "half" instructions here at some point,
+	 * but I don't feel confident enough about them yet.
+	 */
+	set_long(pc, e);
+	if (is_long(e) && !is_immd(e)) {
+		e->inst[1] |= 0x04000000; /* 32-bit */
+		e->inst[1] |= 0x0003c000; /* "subsubop" 0xf == mov */
+	}
+
+	emit(pc, e);
+}
+
+static boolean
+check_swap_src_0_1(struct nv50_pc *pc,
+		   struct nv50_reg **s0, struct nv50_reg **s1)
+{
+	struct nv50_reg *src0 = *s0, *src1 = *s1;
+
+	if (src0->type == P_CONST) {
+		if (src1->type != P_CONST) {
+			*s0 = src1;
+			*s1 = src0;
+			return TRUE;
+		}
+	} else
+	if (src1->type == P_ATTR) {
+		if (src0->type != P_ATTR) {
+			*s0 = src1;
+			*s1 = src0;
+			return TRUE;
+		}
+	}
+
+	return FALSE;
+}
+
+static void
+set_src_0(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
+{
+	if (src->type == P_ATTR) {
+		set_long(pc, e);
+		e->inst[1] |= 0x00200000;
+	} else
+	if (src->type == P_CONST || src->type == P_IMMD) {
+		struct nv50_reg *temp = temp_temp(pc);
+
+		emit_mov(pc, temp, src);
+		src = temp;
+	}
+
+	alloc_reg(pc, src);
+	e->inst[0] |= (src->hw << 9);
+}
+
+static void
+set_src_1(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
+{
+	if (src->type == P_ATTR) {
+		struct nv50_reg *temp = temp_temp(pc);
+
+		emit_mov(pc, temp, src);
+		src = temp;
+	} else
+	if (src->type == P_CONST || src->type == P_IMMD) {
+		assert(!(e->inst[0] & 0x00800000));
+		if (e->inst[0] & 0x01000000) {
+			struct nv50_reg *temp = temp_temp(pc);
+
+			emit_mov(pc, temp, src);
+			src = temp;
+		} else {
+			set_data(pc, src, 0x7f, 16, e);
+			e->inst[0] |= 0x00800000;
+		}
+	}
+
+	alloc_reg(pc, src);
+	e->inst[0] |= (src->hw << 16);
+}
+
+static void
+set_src_2(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e)
+{
+	set_long(pc, e);
+
+	if (src->type == P_ATTR) {
+		struct nv50_reg *temp = temp_temp(pc);
+
+		emit_mov(pc, temp, src);
+		src = temp;
+	} else
+	if (src->type == P_CONST || src->type == P_IMMD) {
+		assert(!(e->inst[0] & 0x01000000));
+		if (e->inst[0] & 0x00800000) {
+			struct nv50_reg *temp = temp_temp(pc);
+
+			emit_mov(pc, temp, src);
+			src = temp;
+		} else {
+			set_data(pc, src, 0x7f, 32+14, e);
+			e->inst[0] |= 0x01000000;
+		}
+	}
+
+	alloc_reg(pc, src);
+	e->inst[1] |= (src->hw << 14);
+}
+
+static void
+emit_mul(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0,
+	 struct nv50_reg *src1)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xc0000000;
+	set_long(pc, e);
+
+	check_swap_src_0_1(pc, &src0, &src1);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_1(pc, src1, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_add(struct nv50_pc *pc, struct nv50_reg *dst,
+	 struct nv50_reg *src0, struct nv50_reg *src1)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xb0000000;
+
+	check_swap_src_0_1(pc, &src0, &src1);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	if (is_long(e))
+		set_src_2(pc, src1, e);
+	else
+		set_src_1(pc, src1, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_minmax(struct nv50_pc *pc, unsigned sub, struct nv50_reg *dst,
+	    struct nv50_reg *src0, struct nv50_reg *src1)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	set_long(pc, e);
+	e->inst[0] |= 0xb0000000;
+	e->inst[1] |= (sub << 29);
+
+	check_swap_src_0_1(pc, &src0, &src1);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_1(pc, src1, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_sub(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0,
+	 struct nv50_reg *src1)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xb0000000;
+
+	set_long(pc, e);
+	if (check_swap_src_0_1(pc, &src0, &src1))
+		e->inst[1] |= 0x04000000;
+	else
+		e->inst[1] |= 0x08000000;
+
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_2(pc, src1, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_mad(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0,
+	 struct nv50_reg *src1, struct nv50_reg *src2)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xe0000000;
+
+	check_swap_src_0_1(pc, &src0, &src1);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_1(pc, src1, e);
+	set_src_2(pc, src2, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_msb(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src0,
+	 struct nv50_reg *src1, struct nv50_reg *src2)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xe0000000;
+	set_long(pc, e);
+	e->inst[1] |= 0x08000000; /* src0 * src1 - src2 */
+
+	check_swap_src_0_1(pc, &src0, &src1);
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_1(pc, src1, e);
+	set_src_2(pc, src2, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_flop(struct nv50_pc *pc, unsigned sub,
+	  struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0x90000000;
+	if (sub) {
+		set_long(pc, e);
+		e->inst[1] |= (sub << 29);
+	}
+
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_preex2(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xb0000000;
+
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+	set_long(pc, e);
+	e->inst[1] |= (6 << 29) | 0x00004000;
+
+	emit(pc, e);
+}
+
+static void
+emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] |= 0xb0000000;
+
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+	set_long(pc, e);
+	e->inst[1] |= (6 << 29);
+
+	emit(pc, e);
+}
+
+static void
+emit_set(struct nv50_pc *pc, unsigned c_op, struct nv50_reg *dst,
+	 struct nv50_reg *src0, struct nv50_reg *src1)
+{
+	struct nv50_program_exec *e = exec(pc);
+	unsigned inv_cop[8] = { 0, 4, 2, 6, 1, 5, 3, 7 };
+	struct nv50_reg *rdst;
+
+	assert(c_op <= 7);
+	if (check_swap_src_0_1(pc, &src0, &src1))
+		c_op = inv_cop[c_op];
+
+	rdst = dst;
+	if (dst->type != P_TEMP)
+		dst = alloc_temp(pc, NULL);
+
+	/* set.u32 */
+	set_long(pc, e);
+	e->inst[0] |= 0xb0000000;
+	e->inst[1] |= (3 << 29);
+	e->inst[1] |= (c_op << 14);
+	/*XXX: breaks things, .u32 by default?
+	 *     decuda will disasm as .u16 and use .lo/.hi regs, but this
+	 *     doesn't seem to match what the hw actually does.
+	inst[1] |= 0x04000000; << breaks things.. .u32 by default?
+	 */
+	set_dst(pc, dst, e);
+	set_src_0(pc, src0, e);
+	set_src_1(pc, src1, e);
+	emit(pc, e);
+
+	/* cvt.f32.u32 */
+	e = exec(pc);
+	e->inst[0] = 0xa0000001;
+	e->inst[1] = 0x64014780;
+	set_dst(pc, rdst, e);
+	set_src_0(pc, dst, e);
+	emit(pc, e);
+
+	if (dst != rdst)
+		free_temp(pc, dst);
+}
+
+static void
+emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] = 0xa0000000; /* cvt */
+	set_long(pc, e);
+	e->inst[1] |= (6 << 29); /* cvt */
+	e->inst[1] |= 0x08000000; /* integer mode */
+	e->inst[1] |= 0x04000000; /* 32 bit */
+	e->inst[1] |= ((0x1 << 3)) << 14; /* .rn */
+	e->inst[1] |= (1 << 14); /* src .f32 */
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_pow(struct nv50_pc *pc, struct nv50_reg *dst,
+	 struct nv50_reg *v, struct nv50_reg *e)
+{
+	struct nv50_reg *temp = alloc_temp(pc, NULL);
+
+	emit_flop(pc, 3, temp, v);
+	emit_mul(pc, temp, temp, e);
+	emit_preex2(pc, temp, temp);
+	emit_flop(pc, 6, dst, temp);
+
+	free_temp(pc, temp);
+}
+
+static void
+emit_abs(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	e->inst[0] = 0xa0000000; /* cvt */
+	set_long(pc, e);
+	e->inst[1] |= (6 << 29); /* cvt */
+	e->inst[1] |= 0x04000000; /* 32 bit */
+	e->inst[1] |= (1 << 14); /* src .f32 */
+	e->inst[1] |= ((1 << 6) << 14); /* .abs */
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+
+	emit(pc, e);
+}
+
+static void
+emit_lit(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask,
+	 struct nv50_reg **src)
+{
+	struct nv50_reg *one = alloc_immd(pc, 1.0);
+	struct nv50_reg *zero = alloc_immd(pc, 0.0);
+	struct nv50_reg *neg128 = alloc_immd(pc, -127.999999);
+	struct nv50_reg *pos128 = alloc_immd(pc,  127.999999);
+	struct nv50_reg *tmp[4];
+
+	if (mask & (1 << 0))
+		emit_mov(pc, dst[0], one);
+
+	if (mask & (1 << 3))
+		emit_mov(pc, dst[3], one);
+
+	if (mask & (3 << 1)) {
+		if (mask & (1 << 1))
+			tmp[0] = dst[1];
+		else
+			tmp[0] = temp_temp(pc);
+		emit_minmax(pc, 4, tmp[0], src[0], zero);
+	}
+
+	if (mask & (1 << 2)) {
+		set_pred_wr(pc, 1, 0, pc->p->exec_tail);
+
+		tmp[1] = temp_temp(pc);
+		emit_minmax(pc, 4, tmp[1], src[1], zero);
+
+		tmp[3] = temp_temp(pc);
+		emit_minmax(pc, 4, tmp[3], src[3], neg128);
+		emit_minmax(pc, 5, tmp[3], tmp[3], pos128);
+
+		emit_pow(pc, dst[2], tmp[1], tmp[3]);
+		emit_mov(pc, dst[2], zero);
+		set_pred(pc, 3, 0, pc->p->exec_tail);
+	}
+}
+
+static void
+emit_neg(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src)
+{
+	struct nv50_program_exec *e = exec(pc);
+
+	set_long(pc, e);
+	e->inst[0] |= 0xa0000000; /* delta */
+	e->inst[1] |= (7 << 29); /* delta */
+	e->inst[1] |= 0x04000000; /* negate arg0? probably not */
+	e->inst[1] |= (1 << 14); /* src .f32 */
+	set_dst(pc, dst, e);
+	set_src_0(pc, src, e);
+
+	emit(pc, e);
+}
+
+static struct nv50_reg *
+tgsi_dst(struct nv50_pc *pc, int c, const struct tgsi_full_dst_register *dst)
+{
+	switch (dst->DstRegister.File) {
+	case TGSI_FILE_TEMPORARY:
+		return &pc->temp[dst->DstRegister.Index * 4 + c];
+	case TGSI_FILE_OUTPUT:
+		return &pc->result[dst->DstRegister.Index * 4 + c];
+	case TGSI_FILE_NULL:
+		return NULL;
+	default:
+		break;
+	}
+
+	return NULL;
+}
+
+static struct nv50_reg *
+tgsi_src(struct nv50_pc *pc, int chan, const struct tgsi_full_src_register *src)
+{
+	struct nv50_reg *r = NULL;
+	struct nv50_reg *temp;
+	unsigned c;
+
+	c = tgsi_util_get_full_src_register_extswizzle(src, chan);
+	switch (c) {
+	case TGSI_EXTSWIZZLE_X:
+	case TGSI_EXTSWIZZLE_Y:
+	case TGSI_EXTSWIZZLE_Z:
+	case TGSI_EXTSWIZZLE_W:
+		switch (src->SrcRegister.File) {
+		case TGSI_FILE_INPUT:
+			r = &pc->attr[src->SrcRegister.Index * 4 + c];
+			break;
+		case TGSI_FILE_TEMPORARY:
+			r = &pc->temp[src->SrcRegister.Index * 4 + c];
+			break;
+		case TGSI_FILE_CONSTANT:
+			r = &pc->param[src->SrcRegister.Index * 4 + c];
+			break;
+		case TGSI_FILE_IMMEDIATE:
+			r = &pc->immd[src->SrcRegister.Index * 4 + c];
+			break;
+		case TGSI_FILE_SAMPLER:
+			break;
+		default:
+			assert(0);
+			break;
+		}
+		break;
+	case TGSI_EXTSWIZZLE_ZERO:
+		r = alloc_immd(pc, 0.0);
+		break;
+	case TGSI_EXTSWIZZLE_ONE:
+		r = alloc_immd(pc, 1.0);
+		break;
+	default:
+		assert(0);
+		break;
+	}
+
+	switch (tgsi_util_get_full_src_register_sign_mode(src, chan)) {
+	case TGSI_UTIL_SIGN_KEEP:
+		break;
+	case TGSI_UTIL_SIGN_CLEAR:
+		temp = temp_temp(pc);
+		emit_abs(pc, temp, r);
+		r = temp;
+		break;
+	case TGSI_UTIL_SIGN_TOGGLE:
+		temp = temp_temp(pc);
+		emit_neg(pc, temp, r);
+		r = temp;
+		break;
+	case TGSI_UTIL_SIGN_SET:
+		temp = temp_temp(pc);
+		emit_abs(pc, temp, r);
+		emit_neg(pc, temp, r);
+		r = temp;
+		break;
+	default:
+		assert(0);
+		break;
+	}
+
+	return r;
+}
+
+static boolean
+nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
+{
+	const struct tgsi_full_instruction *inst = &tok->FullInstruction;
+	struct nv50_reg *rdst[4], *dst[4], *src[3][4], *temp;
+	unsigned mask, sat;
+	int i, c;
+
+	mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
+	sat = inst->Instruction.Saturate == TGSI_SAT_ZERO_ONE;
+
+	for (c = 0; c < 4; c++) {
+		if (mask & (1 << c))
+			dst[c] = tgsi_dst(pc, c, &inst->FullDstRegisters[0]);
+		else
+			dst[c] = NULL;
+	}
+
+	for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
+		for (c = 0; c < 4; c++)
+			src[i][c] = tgsi_src(pc, c, &inst->FullSrcRegisters[i]);
+	}
+
+	if (sat) {
+		for (c = 0; c < 4; c++) {
+			rdst[c] = dst[c];
+			dst[c] = temp_temp(pc);
+		}
+	}
+
+	switch (inst->Instruction.Opcode) {
+	case TGSI_OPCODE_ABS:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_abs(pc, dst[c], src[0][c]);
+		}
+		break;
+	case TGSI_OPCODE_ADD:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_add(pc, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_COS:
+		temp = alloc_temp(pc, NULL);
+		emit_precossin(pc, temp, src[0][0]);
+		emit_flop(pc, 5, temp, temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		break;
+	case TGSI_OPCODE_DP3:
+		temp = alloc_temp(pc, NULL);
+		emit_mul(pc, temp, src[0][0], src[1][0]);
+		emit_mad(pc, temp, src[0][1], src[1][1], temp);
+		emit_mad(pc, temp, src[0][2], src[1][2], temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_DP4:
+		temp = alloc_temp(pc, NULL);
+		emit_mul(pc, temp, src[0][0], src[1][0]);
+		emit_mad(pc, temp, src[0][1], src[1][1], temp);
+		emit_mad(pc, temp, src[0][2], src[1][2], temp);
+		emit_mad(pc, temp, src[0][3], src[1][3], temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_DPH:
+		temp = alloc_temp(pc, NULL);
+		emit_mul(pc, temp, src[0][0], src[1][0]);
+		emit_mad(pc, temp, src[0][1], src[1][1], temp);
+		emit_mad(pc, temp, src[0][2], src[1][2], temp);
+		emit_add(pc, temp, src[1][3], temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_DST:
+	{
+		struct nv50_reg *one = alloc_immd(pc, 1.0);
+		if (mask & (1 << 0))
+			emit_mov(pc, dst[0], one);
+		if (mask & (1 << 1))
+			emit_mul(pc, dst[1], src[0][1], src[1][1]);
+		if (mask & (1 << 2))
+			emit_mov(pc, dst[2], src[0][2]);
+		if (mask & (1 << 3))
+			emit_mov(pc, dst[3], src[1][3]);
+		FREE(one);
+	}
+		break;
+	case TGSI_OPCODE_EX2:
+		temp = alloc_temp(pc, NULL);
+		emit_preex2(pc, temp, src[0][0]);
+		emit_flop(pc, 6, temp, temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_FLR:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_flr(pc, dst[c], src[0][c]);
+		}
+		break;
+	case TGSI_OPCODE_FRC:
+		temp = alloc_temp(pc, NULL);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_flr(pc, temp, src[0][c]);
+			emit_sub(pc, dst[c], src[0][c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_LIT:
+		emit_lit(pc, &dst[0], mask, &src[0][0]);
+		break;
+	case TGSI_OPCODE_LG2:
+		temp = alloc_temp(pc, NULL);
+		emit_flop(pc, 3, temp, src[0][0]);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		break;
+	case TGSI_OPCODE_LRP:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			/*XXX: we can do better than this */
+			temp = alloc_temp(pc, NULL);
+			emit_neg(pc, temp, src[0][c]);
+			emit_mad(pc, temp, temp, src[2][c], src[2][c]);
+			emit_mad(pc, dst[c], src[0][c], src[1][c], temp);
+			free_temp(pc, temp);
+		}
+		break;
+	case TGSI_OPCODE_MAD:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mad(pc, dst[c], src[0][c], src[1][c], src[2][c]);
+		}
+		break;
+	case TGSI_OPCODE_MAX:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_minmax(pc, 4, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_MIN:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_minmax(pc, 5, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_MOV:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], src[0][c]);
+		}
+		break;
+	case TGSI_OPCODE_MUL:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mul(pc, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_POW:
+		temp = alloc_temp(pc, NULL);
+		emit_pow(pc, temp, src[0][0], src[1][0]);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_RCP:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_flop(pc, 0, dst[c], src[0][0]);
+		}
+		break;
+	case TGSI_OPCODE_RSQ:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_flop(pc, 2, dst[c], src[0][0]);
+		}
+		break;
+	case TGSI_OPCODE_SCS:
+		temp = alloc_temp(pc, NULL);
+		emit_precossin(pc, temp, src[0][0]);
+		if (mask & (1 << 0))
+			emit_flop(pc, 5, dst[0], temp);
+		if (mask & (1 << 1))
+			emit_flop(pc, 4, dst[1], temp);
+		break;
+	case TGSI_OPCODE_SGE:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_set(pc, 6, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_SIN:
+		temp = alloc_temp(pc, NULL);
+		emit_precossin(pc, temp, src[0][0]);
+		emit_flop(pc, 4, temp, temp);
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_mov(pc, dst[c], temp);
+		}
+		break;
+	case TGSI_OPCODE_SLT:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_set(pc, 1, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_SUB:
+		for (c = 0; c < 4; c++) {
+			if (!(mask & (1 << c)))
+				continue;
+			emit_sub(pc, dst[c], src[0][c], src[1][c]);
+		}
+		break;
+	case TGSI_OPCODE_TEX:
+		{
+			struct nv50_reg *t0, *t1, *t2, *t3;
+			struct nv50_program_exec *e;
+
+			t0 = alloc_temp(pc, NULL);
+			t0 = alloc_temp(pc, NULL);
+			t1 = alloc_temp(pc, NULL);
+			t2 = alloc_temp(pc, NULL);
+			t3 = alloc_temp(pc, NULL);
+			emit_mov(pc, t0, src[0][0]);
+			emit_mov(pc, t1, src[0][1]);
+
+			e = exec(pc);
+			e->inst[0] = 0xf6400000;
+			set_long(pc, e);
+			e->inst[1] |= 0x0000c004;
+			set_dst(pc, t0, e);
+			emit(pc, e);
+
+			if (mask & (1 << 0)) emit_mov(pc, dst[0], t0);
+			if (mask & (1 << 1)) emit_mov(pc, dst[1], t1);
+			if (mask & (1 << 2)) emit_mov(pc, dst[2], t2);
+			if (mask & (1 << 3)) emit_mov(pc, dst[3], t3);
+
+			free_temp(pc, t0);
+			free_temp(pc, t1);
+			free_temp(pc, t2);
+			free_temp(pc, t3);
+		}
+		break;
+	case TGSI_OPCODE_XPD:
+		temp = alloc_temp(pc, NULL);
+		if (mask & (1 << 0)) {
+			emit_mul(pc, temp, src[0][2], src[1][1]);
+			emit_msb(pc, dst[0], src[0][1], src[1][2], temp);
+		}
+		if (mask & (1 << 1)) {
+			emit_mul(pc, temp, src[0][0], src[1][2]);
+			emit_msb(pc, dst[1], src[0][2], src[1][0], temp);
+		}
+		if (mask & (1 << 2)) {
+			emit_mul(pc, temp, src[0][1], src[1][0]);
+			emit_msb(pc, dst[2], src[0][0], src[1][1], temp);
+		}
+		free_temp(pc, temp);
+		break;
+	case TGSI_OPCODE_END:
+		break;
+	default:
+		NOUVEAU_ERR("invalid opcode %d\n", inst->Instruction.Opcode);
+		return FALSE;
+	}
+
+	if (sat) {
+		for (c = 0; c < 4; c++) {
+			struct nv50_program_exec *e;
+
+			if (!(mask & (1 << c)))
+				continue;
+			e = exec(pc);
+
+			e->inst[0] = 0xa0000000; /* cvt */
+			set_long(pc, e);
+			e->inst[1] |= (6 << 29); /* cvt */
+			e->inst[1] |= 0x04000000; /* 32 bit */
+			e->inst[1] |= (1 << 14); /* src .f32 */
+			e->inst[1] |= ((1 << 5) << 14); /* .sat */
+			set_dst(pc, rdst[c], e);
+			set_src_0(pc, dst[c], e);
+			emit(pc, e);
+		}
+	}
+
+	kill_temp_temp(pc);
+	return TRUE;
+}
+
+static boolean
+nv50_program_tx_prep(struct nv50_pc *pc)
+{
+	struct tgsi_parse_context p;
+	boolean ret = FALSE;
+	unsigned i, c;
+
+	tgsi_parse_init(&p, pc->p->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&p)) {
+		const union tgsi_full_token *tok = &p.FullToken;
+
+		tgsi_parse_token(&p);
+		switch (tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_IMMEDIATE:
+		{
+			const struct tgsi_full_immediate *imm =
+				&p.FullToken.FullImmediate;
+
+			ctor_immd(pc, imm->u.ImmediateFloat32[0].Float,
+				      imm->u.ImmediateFloat32[1].Float,
+				      imm->u.ImmediateFloat32[2].Float,
+				      imm->u.ImmediateFloat32[3].Float);
+		}
+			break;
+		case TGSI_TOKEN_TYPE_DECLARATION:
+		{
+			const struct tgsi_full_declaration *d;
+			unsigned last;
+
+			d = &p.FullToken.FullDeclaration;
+			last = d->DeclarationRange.Last;
+
+			switch (d->Declaration.File) {
+			case TGSI_FILE_TEMPORARY:
+				if (pc->temp_nr < (last + 1))
+					pc->temp_nr = last + 1;
+				break;
+			case TGSI_FILE_OUTPUT:
+				if (pc->result_nr < (last + 1))
+					pc->result_nr = last + 1;
+				break;
+			case TGSI_FILE_INPUT:
+				if (pc->attr_nr < (last + 1))
+					pc->attr_nr = last + 1;
+				break;
+			case TGSI_FILE_CONSTANT:
+				if (pc->param_nr < (last + 1))
+					pc->param_nr = last + 1;
+				break;
+			case TGSI_FILE_SAMPLER:
+				break;
+			default:
+				NOUVEAU_ERR("bad decl file %d\n",
+					    d->Declaration.File);
+				goto out_err;
+			}
+		}
+			break;
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (pc->temp_nr) {
+		pc->temp = CALLOC(pc->temp_nr * 4, sizeof(struct nv50_reg));
+		if (!pc->temp)
+			goto out_err;
+
+		for (i = 0; i < pc->temp_nr; i++) {
+			for (c = 0; c < 4; c++) {
+				pc->temp[i*4+c].type = P_TEMP;
+				pc->temp[i*4+c].hw = -1;
+				pc->temp[i*4+c].index = i;
+			}
+		}
+	}
+
+	if (pc->attr_nr) {
+		struct nv50_reg *iv = NULL;
+		int aid = 0;
+
+		pc->attr = CALLOC(pc->attr_nr * 4, sizeof(struct nv50_reg));
+		if (!pc->attr)
+			goto out_err;
+
+		if (pc->p->type == PIPE_SHADER_FRAGMENT) {
+			iv = alloc_temp(pc, NULL);
+			emit_interp(pc, iv, iv, NULL);
+			emit_flop(pc, 0, iv, iv);
+			aid++;
+		}
+
+		for (i = 0; i < pc->attr_nr; i++) {
+			struct nv50_reg *a = &pc->attr[i*4];
+
+			for (c = 0; c < 4; c++) {
+				if (pc->p->type == PIPE_SHADER_FRAGMENT) {
+					struct nv50_reg *at =
+						alloc_temp(pc, NULL);
+					pc->attr[i*4+c].type = at->type;
+					pc->attr[i*4+c].hw = at->hw;
+					pc->attr[i*4+c].index = at->index;
+				} else {
+					pc->p->cfg.vp.attr[aid/32] |=
+						(1 << (aid % 32));
+					pc->attr[i*4+c].type = P_ATTR;
+					pc->attr[i*4+c].hw = aid++;
+					pc->attr[i*4+c].index = i;
+				}
+			}
+
+			if (pc->p->type != PIPE_SHADER_FRAGMENT)
+				continue;
+
+			emit_interp(pc, &a[0], &a[0], iv);
+			emit_interp(pc, &a[1], &a[1], iv);
+			emit_interp(pc, &a[2], &a[2], iv);
+			emit_interp(pc, &a[3], &a[3], iv);
+		}
+
+		if (iv)
+			free_temp(pc, iv);
+	}
+
+	if (pc->result_nr) {
+		int rid = 0;
+
+		pc->result = CALLOC(pc->result_nr * 4, sizeof(struct nv50_reg));
+		if (!pc->result)
+			goto out_err;
+
+		for (i = 0; i < pc->result_nr; i++) {
+			for (c = 0; c < 4; c++) {
+				if (pc->p->type == PIPE_SHADER_FRAGMENT) {
+					pc->result[i*4+c].type = P_TEMP;
+					pc->result[i*4+c].hw = -1;
+				} else {
+					pc->result[i*4+c].type = P_RESULT;
+					pc->result[i*4+c].hw = rid++;
+				}
+				pc->result[i*4+c].index = i;
+			}
+		}
+	}
+
+	if (pc->param_nr) {
+		int rid = 0;
+
+		pc->param = CALLOC(pc->param_nr * 4, sizeof(struct nv50_reg));
+		if (!pc->param)
+			goto out_err;
+
+		for (i = 0; i < pc->param_nr; i++) {
+			for (c = 0; c < 4; c++) {
+				pc->param[i*4+c].type = P_CONST;
+				pc->param[i*4+c].hw = rid++;
+				pc->param[i*4+c].index = i;
+			}
+		}
+	}
+
+	if (pc->immd_nr) {
+		int rid = pc->param_nr * 4;
+
+		pc->immd = CALLOC(pc->immd_nr * 4, sizeof(struct nv50_reg));
+		if (!pc->immd)
+			goto out_err;
+
+		for (i = 0; i < pc->immd_nr; i++) {
+			for (c = 0; c < 4; c++) {
+				pc->immd[i*4+c].type = P_IMMD;
+				pc->immd[i*4+c].hw = rid++;
+				pc->immd[i*4+c].index = i;
+			}
+		}
+	}
+
+	ret = TRUE;
+out_err:
+	tgsi_parse_free(&p);
+	return ret;
+}
+
+static boolean
+nv50_program_tx(struct nv50_program *p)
+{
+	struct tgsi_parse_context parse;
+	struct nv50_pc *pc;
+	boolean ret;
+
+	pc = CALLOC_STRUCT(nv50_pc);
+	if (!pc)
+		return FALSE;
+	pc->p = p;
+	pc->p->cfg.high_temp = 4;
+
+	ret = nv50_program_tx_prep(pc);
+	if (ret == FALSE)
+		goto out_cleanup;
+
+	tgsi_parse_init(&parse, pc->p->pipe.tokens);
+	while (!tgsi_parse_end_of_tokens(&parse)) {
+		const union tgsi_full_token *tok = &parse.FullToken;
+
+		tgsi_parse_token(&parse);
+
+		switch (tok->Token.Type) {
+		case TGSI_TOKEN_TYPE_INSTRUCTION:
+			ret = nv50_program_tx_insn(pc, tok);
+			if (ret == FALSE)
+				goto out_err;
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (p->type == PIPE_SHADER_FRAGMENT) {
+		struct nv50_reg out;
+
+		out.type = P_TEMP;
+		for (out.hw = 0; out.hw < pc->result_nr * 4; out.hw++)
+			emit_mov(pc, &out, &pc->result[out.hw]);
+	}
+
+	assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head));
+	pc->p->exec_tail->inst[1] |= 0x00000001;
+
+	p->param_nr = pc->param_nr * 4;
+	p->immd_nr = pc->immd_nr * 4;
+	p->immd = pc->immd_buf;
+
+out_err:
+	tgsi_parse_free(&parse);
+
+out_cleanup:
+	return ret;
+}
+
+static void
+nv50_program_validate(struct nv50_context *nv50, struct nv50_program *p)
+{
+	if (nv50_program_tx(p) == FALSE)
+		assert(0);
+	p->translated = TRUE;
+}
+
+static void
+nv50_program_upload_data(struct nv50_context *nv50, float *map,
+			 unsigned start, unsigned count)
+{
+	while (count) {
+		unsigned nr = count > 2047 ? 2047 : count;
+
+		BEGIN_RING(tesla, 0x00000f00, 1);
+		OUT_RING  ((NV50_CB_PMISC << 0) | (start << 8));
+		BEGIN_RING(tesla, 0x40000f04, nr);
+		OUT_RINGp (map, nr);
+
+		map += nr;
+		start += nr;
+		count -= nr;
+	}
+}
+
+static void
+nv50_program_validate_data(struct nv50_context *nv50, struct nv50_program *p)
+{
+	struct nouveau_winsys *nvws = nv50->screen->nvws;
+	struct pipe_winsys *ws = nv50->pipe.winsys;
+	unsigned nr = p->param_nr + p->immd_nr;
+
+	if (!p->data && nr) {
+		struct nouveau_resource *heap = nv50->screen->vp_data_heap;
+
+		if (nvws->res_alloc(heap, nr, p, &p->data)) {
+			while (heap->next && heap->size < nr) {
+				struct nv50_program *evict = heap->next->priv;
+				nvws->res_free(&evict->data);
+			}
+
+			if (nvws->res_alloc(heap, nr, p, &p->data))
+				assert(0);
+		}
+	}
+
+	if (p->param_nr) {
+		float *map = ws->buffer_map(ws, nv50->constbuf[p->type],
+					    PIPE_BUFFER_USAGE_CPU_READ);
+		nv50_program_upload_data(nv50, map, p->data->start,
+					 p->param_nr);
+		ws->buffer_unmap(ws, nv50->constbuf[p->type]);
+	}
+
+	if (p->immd_nr) {
+		nv50_program_upload_data(nv50, p->immd,
+					 p->data->start + p->param_nr,
+					 p->immd_nr);
+	}
+}
+
+static void
+nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p)
+{
+	struct pipe_winsys *ws = nv50->pipe.winsys;
+	struct nv50_program_exec *e;
+	struct nouveau_stateobj *so;
+	const unsigned flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_WR;
+	unsigned start, count, *up, *ptr;
+	boolean upload = FALSE;
+
+	if (!p->buffer) {
+		p->buffer = ws->buffer_create(ws, 0x100, 0, p->exec_size * 4);
+		upload = TRUE;
+	}
+
+	if (p->data && p->data->start != p->data_start) {
+		for (e = p->exec_head; e; e = e->next) {
+			unsigned ei, ci;
+
+			if (e->param.index < 0)
+				continue;
+			ei = e->param.shift >> 5;
+			ci = e->param.index + p->data->start;
+
+			e->inst[ei] &= ~e->param.mask;
+			e->inst[ei] |= (ci << e->param.shift);
+		}
+
+		p->data_start = p->data->start;
+		upload = TRUE;
+	}
+
+	if (!upload)
+		return;
+
+	up = ptr = MALLOC(p->exec_size * 4);
+	for (e = p->exec_head; e; e = e->next) {
+		*(ptr++) = e->inst[0];
+		if (is_long(e))
+			*(ptr++) = e->inst[1];
+	}
+
+	so = so_new(4,2);
+	so_method(so, nv50->screen->tesla, 0x1280, 3);
+	so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, p->buffer, 0, flags | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, (NV50_CB_PUPLOAD << 16) | 0x0800); //(p->exec_size * 4));
+
+	start = 0; count = p->exec_size;
+	while (count) {
+		struct nouveau_winsys *nvws = nv50->screen->nvws;
+		unsigned nr;
+
+		so_emit(nvws, so);
+
+		nr = MIN2(count, 2047);
+		nr = MIN2(nvws->channel->pushbuf->remaining, nr);
+		if (nvws->channel->pushbuf->remaining < (nr + 3)) {
+			FIRE_RING(NULL);
+			continue;
+		}
+
+		BEGIN_RING(tesla, 0x0f00, 1);
+		OUT_RING  ((start << 8) | NV50_CB_PUPLOAD);
+		BEGIN_RING(tesla, 0x40000f04, nr);	
+		OUT_RINGp (up + start, nr);
+
+		start += nr;
+		count -= nr;
+	}
+
+	FREE(up);
+	so_ref(NULL, &so);
+}
+
+void
+nv50_vertprog_validate(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_program *p = nv50->vertprog;
+	struct nouveau_stateobj *so;
+
+	if (!p->translated) {
+		nv50_program_validate(nv50, p);
+		if (!p->translated)
+			assert(0);
+	}
+
+	nv50_program_validate_data(nv50, p);
+	nv50_program_validate_code(nv50, p);
+
+	so = so_new(13, 2);
+	so_method(so, tesla, NV50TCL_VP_ADDRESS_HIGH, 2);
+	so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
+		  NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
+		  NOUVEAU_BO_LOW, 0, 0);
+	so_method(so, tesla, 0x1650, 2);
+	so_data  (so, p->cfg.vp.attr[0]);
+	so_data  (so, p->cfg.vp.attr[1]);
+	so_method(so, tesla, 0x16b8, 1);
+	so_data  (so, p->cfg.high_result);
+	so_method(so, tesla, 0x16ac, 2);
+	so_data  (so, p->cfg.high_result); //8);
+	so_data  (so, p->cfg.high_temp);
+	so_method(so, tesla, 0x140c, 1);
+	so_data  (so, 0); /* program start offset */
+	so_ref(so, &nv50->state.vertprog);
+}
+
+void
+nv50_fragprog_validate(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nv50_program *p = nv50->fragprog;
+	struct nouveau_stateobj *so;
+
+	if (!p->translated) {
+		nv50_program_validate(nv50, p);
+		if (!p->translated)
+			assert(0);
+	}
+
+	nv50_program_validate_data(nv50, p);
+	nv50_program_validate_code(nv50, p);
+
+	so = so_new(64, 2);
+	so_method(so, tesla, NV50TCL_FP_ADDRESS_HIGH, 2);
+	so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
+		  NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, p->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RD |
+		  NOUVEAU_BO_LOW, 0, 0);
+	so_method(so, tesla, 0x1904, 4);
+	so_data  (so, 0x01040404); /* p: 0x01000404 */
+	so_data  (so, 0x00000004);
+	so_data  (so, 0x00000000);
+	so_data  (so, 0x00000000);
+	so_method(so, tesla, 0x16bc, 3); /*XXX: fixme */
+	so_data  (so, 0x03020100);
+	so_data  (so, 0x07060504);
+	so_data  (so, 0x0b0a0908);
+	so_method(so, tesla, 0x1988, 2);
+	so_data  (so, 0x08080408); //0x08040404); /* p: 0x0f000401 */
+	so_data  (so, p->cfg.high_temp);
+	so_method(so, tesla, 0x1414, 1);
+	so_data  (so, 0); /* program start offset */
+	so_ref(so, &nv50->state.fragprog);
+}
+
+void
+nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
+{
+	struct pipe_winsys *ws = nv50->pipe.winsys;
+
+	while (p->exec_head) {
+		struct nv50_program_exec *e = p->exec_head;
+
+		p->exec_head = e->next;
+		FREE(e);
+	}
+	p->exec_tail = NULL;
+	p->exec_size = 0;
+
+	if (p->buffer)
+		pipe_buffer_reference(ws, &p->buffer, NULL);
+
+	nv50->screen->nvws->res_free(&p->data);
+
+	p->translated = 0;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_program.h b/src/gallium/drivers/nv50/nv50_program.h
new file mode 100644
index 00000000000..78deed6a384
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_program.h
@@ -0,0 +1,45 @@
+#ifndef __NV50_PROGRAM_H__
+#define __NV50_PROGRAM_H__
+
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_scan.h"
+
+struct nv50_program_exec {
+	struct nv50_program_exec *next;
+
+	unsigned inst[2];
+	struct {
+		int index;
+		unsigned mask;
+		unsigned shift;
+	} param;
+};
+
+struct nv50_program {
+	struct pipe_shader_state pipe;
+	struct tgsi_shader_info info;
+	boolean translated;
+
+	unsigned type;
+	struct nv50_program_exec *exec_head;
+	struct nv50_program_exec *exec_tail;
+	unsigned exec_size;
+	struct nouveau_resource *data;
+	unsigned data_start;
+
+	struct pipe_buffer *buffer;
+
+	float *immd;
+	unsigned immd_nr;
+	unsigned param_nr;
+
+	struct {
+		unsigned high_temp;
+		unsigned high_result;
+		struct {
+			unsigned attr[2];
+		} vp;
+	} cfg;
+};
+
+#endif
diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c
new file mode 100644
index 00000000000..26bd90ccc5e
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_query.c
@@ -0,0 +1,69 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_context.h"
+
+#include "nv50_context.h"
+
+static struct pipe_query *
+nv50_query_create(struct pipe_context *pipe, unsigned type)
+{
+	NOUVEAU_ERR("unimplemented\n");
+	return NULL;
+}
+
+static void
+nv50_query_destroy(struct pipe_context *pipe, struct pipe_query *q)
+{
+	NOUVEAU_ERR("unimplemented\n");
+}
+
+static void
+nv50_query_begin(struct pipe_context *pipe, struct pipe_query *q)
+{
+	NOUVEAU_ERR("unimplemented\n");
+}
+
+static void
+nv50_query_end(struct pipe_context *pipe, struct pipe_query *q)
+{
+	NOUVEAU_ERR("unimplemented\n");
+}
+
+static boolean
+nv50_query_result(struct pipe_context *pipe, struct pipe_query *q,
+		  boolean wait, uint64 *result)
+{
+	NOUVEAU_ERR("unimplemented\n");
+	*result = 0xdeadcafe;
+	return TRUE;
+}
+
+void
+nv50_init_query_functions(struct nv50_context *nv50)
+{
+	nv50->pipe.create_query = nv50_query_create;
+	nv50->pipe.destroy_query = nv50_query_destroy;
+	nv50->pipe.begin_query = nv50_query_begin;
+	nv50->pipe.end_query = nv50_query_end;
+	nv50->pipe.get_query_result = nv50_query_result;
+}
diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c
new file mode 100644
index 00000000000..b5aef7dadd8
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_screen.c
@@ -0,0 +1,323 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_screen.h"
+
+#include "nv50_context.h"
+#include "nv50_screen.h"
+
+#include "nouveau/nouveau_stateobj.h"
+
+#define NV5X_GRCLASS5097_CHIPSETS 0x00000001
+#define NV8X_GRCLASS8297_CHIPSETS 0x00000050
+#define NV9X_GRCLASS8297_CHIPSETS 0x00000014
+
+static boolean
+nv50_screen_is_format_supported(struct pipe_screen *pscreen,
+				enum pipe_format format,
+				enum pipe_texture_target target,
+				unsigned tex_usage, unsigned geom_flags)
+{
+	if (tex_usage & PIPE_TEXTURE_USAGE_RENDER_TARGET) {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM:
+		case PIPE_FORMAT_Z24S8_UNORM:
+		case PIPE_FORMAT_Z16_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	} else {
+		switch (format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+		case PIPE_FORMAT_A1R5G5B5_UNORM:
+		case PIPE_FORMAT_A4R4G4B4_UNORM:
+		case PIPE_FORMAT_R5G6B5_UNORM:
+		case PIPE_FORMAT_L8_UNORM:
+		case PIPE_FORMAT_A8_UNORM:
+		case PIPE_FORMAT_I8_UNORM:
+		case PIPE_FORMAT_A8L8_UNORM:
+			return TRUE;
+		default:
+			break;
+		}
+	}
+
+	return FALSE;
+}
+
+static const char *
+nv50_screen_get_name(struct pipe_screen *pscreen)
+{
+	struct nv50_screen *screen = nv50_screen(pscreen);
+	struct nouveau_device *dev = screen->nvws->channel->device;
+	static char buffer[128];
+
+	snprintf(buffer, sizeof(buffer), "NV%02X", dev->chipset);
+	return buffer;
+}
+
+static const char *
+nv50_screen_get_vendor(struct pipe_screen *pscreen)
+{
+	return "nouveau";
+}
+
+static int
+nv50_screen_get_param(struct pipe_screen *pscreen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
+		return 32;
+	case PIPE_CAP_NPOT_TEXTURES:
+		return 0;
+	case PIPE_CAP_TWO_SIDED_STENCIL:
+		return 1;
+	case PIPE_CAP_GLSL:
+		return 0;
+	case PIPE_CAP_S3TC:
+		return 0;
+	case PIPE_CAP_ANISOTROPIC_FILTER:
+		return 0;
+	case PIPE_CAP_POINT_SPRITE:
+		return 0;
+	case PIPE_CAP_MAX_RENDER_TARGETS:
+		return 8;
+	case PIPE_CAP_OCCLUSION_QUERY:
+		return 0;
+	case PIPE_CAP_TEXTURE_SHADOW_MAP:
+		return 0;
+	case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
+		return 13;
+	case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
+		return 10;
+	case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
+		return 13;
+	case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
+	case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
+		return 1;
+	case NOUVEAU_CAP_HW_VTXBUF:	
+		return 1;
+	case NOUVEAU_CAP_HW_IDXBUF:	
+		return 0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0;
+	}
+}
+
+static float
+nv50_screen_get_paramf(struct pipe_screen *pscreen, int param)
+{
+	switch (param) {
+	case PIPE_CAP_MAX_LINE_WIDTH:
+	case PIPE_CAP_MAX_LINE_WIDTH_AA:
+		return 10.0;
+	case PIPE_CAP_MAX_POINT_WIDTH:
+	case PIPE_CAP_MAX_POINT_WIDTH_AA:
+		return 64.0;
+	case PIPE_CAP_MAX_TEXTURE_ANISOTROPY:
+		return 16.0;
+	case PIPE_CAP_MAX_TEXTURE_LOD_BIAS:
+		return 4.0;
+	default:
+		NOUVEAU_ERR("Unknown PIPE_CAP %d\n", param);
+		return 0.0;
+	}
+}
+
+static void
+nv50_screen_destroy(struct pipe_screen *pscreen)
+{
+	FREE(pscreen);
+}
+
+struct pipe_screen *
+nv50_screen_create(struct pipe_winsys *ws, struct nouveau_winsys *nvws)
+{
+	struct nv50_screen *screen = CALLOC_STRUCT(nv50_screen);
+	struct nouveau_stateobj *so;
+	unsigned tesla_class = 0, ret;
+	unsigned chipset = nvws->channel->device->chipset;
+	int i;
+
+	if (!screen)
+		return NULL;
+	screen->nvws = nvws;
+
+	/* 3D object */
+	if ((chipset & 0xf0) != 0x50 && (chipset & 0xf0) != 0x80) {
+		NOUVEAU_ERR("Not a G8x chipset\n");
+		nv50_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	switch (chipset & 0xf0) {
+	case 0x50:
+		if (NV5X_GRCLASS5097_CHIPSETS & (1 << (chipset & 0x0f)))
+			tesla_class = 0x5097;
+		break;
+	case 0x80:
+		if (NV8X_GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f)))
+			tesla_class = 0x8297;
+		break;
+	case 0x90:
+		if (NV9X_GRCLASS8297_CHIPSETS & (1 << (chipset & 0x0f)))
+			tesla_class = 0x8297;
+		break;
+	default:
+		break;
+	}
+
+	if (tesla_class == 0) {
+		NOUVEAU_ERR("Unknown G8x chipset: NV%02x\n", chipset);
+		nv50_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	ret = nvws->grobj_alloc(nvws, tesla_class, &screen->tesla);
+	if (ret) {
+		NOUVEAU_ERR("Error creating 3D object: %d\n", ret);
+		nv50_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Sync notifier */
+	ret = nvws->notifier_alloc(nvws, 1, &screen->sync);
+	if (ret) {
+		NOUVEAU_ERR("Error creating notifier object: %d\n", ret);
+		nv50_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	/* Static tesla init */
+	so = so_new(256, 20);
+
+	so_method(so, screen->tesla, 0x1558, 1);
+	so_data  (so, 1);
+	so_method(so, screen->tesla, NV50TCL_DMA_NOTIFY, 1);
+	so_data  (so, screen->sync->handle);
+	so_method(so, screen->tesla, NV50TCL_DMA_UNK0(0),
+				     NV50TCL_DMA_UNK0__SIZE);
+	for (i = 0; i < NV50TCL_DMA_UNK0__SIZE; i++)
+		so_data(so, nvws->channel->vram->handle);
+	so_method(so, screen->tesla, NV50TCL_DMA_UNK1(0),
+				     NV50TCL_DMA_UNK1__SIZE);
+	for (i = 0; i < NV50TCL_DMA_UNK1__SIZE; i++)
+		so_data(so, nvws->channel->vram->handle);
+	so_method(so, screen->tesla, 0x121c, 1);
+	so_data  (so, 1);
+
+	so_method(so, screen->tesla, 0x13bc, 1);
+	so_data  (so, 0x54);
+	so_method(so, screen->tesla, 0x13ac, 1);
+	so_data  (so, 1);
+	so_method(so, screen->tesla, 0x16b8, 1);
+	so_data  (so, 8);
+
+	/* Shared constant buffer */
+	screen->constbuf = ws->buffer_create(ws, 0, 0, 128 * 4 * 4);
+	if (nvws->res_init(&screen->vp_data_heap, 0, 128)) {
+		NOUVEAU_ERR("Error initialising constant buffer\n");
+		nv50_screen_destroy(&screen->pipe);
+		return NULL;
+	}
+
+	so_method(so, screen->tesla, 0x1280, 3);
+	so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->constbuf, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, (NV50_CB_PMISC << 16) | 0x00001000);
+
+	/* Texture sampler/image unit setup - we abuse the constant buffer
+	 * upload mechanism for the moment to upload data to the tex config
+	 * blocks.  At some point we *may* want to go the NVIDIA way of doing
+	 * things?
+	 */
+	screen->tic = ws->buffer_create(ws, 0, 0, 32 * 8 * 4);
+	so_method(so, screen->tesla, 0x1280, 3);
+	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, (NV50_CB_TIC << 16) | 0x0800);
+	so_method(so, screen->tesla, 0x1574, 3);
+	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->tic, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, 0x00000800);
+
+	screen->tsc = ws->buffer_create(ws, 0, 0, 32 * 8 * 4);
+	so_method(so, screen->tesla, 0x1280, 3);
+	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, (NV50_CB_TSC << 16) | 0x0800);
+	so_method(so, screen->tesla, 0x155c, 3);
+	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+	so_reloc (so, screen->tsc, 0, NOUVEAU_BO_VRAM |
+		  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	so_data  (so, 0x00000800);
+
+
+	/* Vertex array limits - max them out */
+	for (i = 0; i < 16; i++) {
+		so_method(so, screen->tesla, 0x1080 + (i * 8), 2);
+		so_data  (so, 0x000000ff);
+		so_data  (so, 0xffffffff);
+	}
+
+	so_method(so, screen->tesla, NV50TCL_DEPTH_RANGE_NEAR, 2);
+	so_data  (so, fui(0.0));
+	so_data  (so, fui(1.0));
+
+	so_method(so, screen->tesla, 0x1234, 1);
+	so_data  (so, 1);
+	so_method(so, screen->tesla, 0x1458, 1);
+	so_data  (so, 1);
+
+	so_emit(nvws, so);
+	so_ref(so, &screen->static_init);
+	nvws->push_flush(nvws, 0, NULL);
+
+	screen->pipe.winsys = ws;
+
+	screen->pipe.destroy = nv50_screen_destroy;
+
+	screen->pipe.get_name = nv50_screen_get_name;
+	screen->pipe.get_vendor = nv50_screen_get_vendor;
+	screen->pipe.get_param = nv50_screen_get_param;
+	screen->pipe.get_paramf = nv50_screen_get_paramf;
+
+	screen->pipe.is_format_supported = nv50_screen_is_format_supported;
+
+	nv50_screen_init_miptree_functions(&screen->pipe);
+	nv50_surface_init_screen_functions(&screen->pipe);
+
+	return &screen->pipe;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h
new file mode 100644
index 00000000000..400ddcef06d
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_screen.h
@@ -0,0 +1,33 @@
+#ifndef __NV50_SCREEN_H__
+#define __NV50_SCREEN_H__
+
+#include "pipe/p_screen.h"
+
+struct nv50_screen {
+	struct pipe_screen pipe;
+
+	struct nouveau_winsys *nvws;
+
+	unsigned cur_pctx;
+
+	struct nouveau_grobj *tesla;
+	struct nouveau_notifier *sync;
+
+	struct pipe_buffer *constbuf;
+	struct nouveau_resource *vp_data_heap;
+
+	struct pipe_buffer *tic;
+	struct pipe_buffer *tsc;
+
+	struct nouveau_stateobj *static_init;
+};
+
+static INLINE struct nv50_screen *
+nv50_screen(struct pipe_screen *screen)
+{
+	return (struct nv50_screen *)screen;
+}
+
+void nv50_surface_init_screen_functions(struct pipe_screen *);
+
+#endif
diff --git a/src/gallium/drivers/nv50/nv50_state.c b/src/gallium/drivers/nv50/nv50_state.c
new file mode 100644
index 00000000000..95f9d408b5e
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_state.c
@@ -0,0 +1,638 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_state.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "tgsi/tgsi_parse.h"
+
+#include "nv50_context.h"
+#include "nv50_texture.h"
+
+#include "nouveau/nouveau_stateobj.h"
+
+static void *
+nv50_blend_state_create(struct pipe_context *pipe,
+			const struct pipe_blend_state *cso)
+{
+	struct nouveau_stateobj *so = so_new(64, 0);
+	struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla;
+	struct nv50_blend_stateobj *bso = CALLOC_STRUCT(nv50_blend_stateobj);
+	unsigned cmask = 0, i;
+
+	/*XXX ignored:
+	 * 	- dither
+	 */
+
+	if (cso->blend_enable == 0) {
+		so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
+		for (i = 0; i < 8; i++)
+			so_data(so, 0);
+	} else {
+		so_method(so, tesla, NV50TCL_BLEND_ENABLE(0), 8);
+		for (i = 0; i < 8; i++)
+			so_data(so, 1);
+		so_method(so, tesla, NV50TCL_BLEND_EQUATION_RGB, 5);
+		so_data  (so, nvgl_blend_eqn(cso->rgb_func));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rgb_src_factor));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->rgb_dst_factor));
+		so_data  (so, nvgl_blend_eqn(cso->alpha_func));
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->alpha_src_factor));
+		so_method(so, tesla, NV50TCL_BLEND_FUNC_DST_ALPHA, 1);
+		so_data  (so, 0x4000 | nvgl_blend_func(cso->alpha_dst_factor));
+	}
+
+	if (cso->logicop_enable == 0 ) {
+		so_method(so, tesla, NV50TCL_LOGIC_OP_ENABLE, 1);
+		so_data  (so, 0);
+	} else {
+		so_method(so, tesla, NV50TCL_LOGIC_OP_ENABLE, 2);
+		so_data  (so, 1);
+		so_data  (so, nvgl_logicop_func(cso->logicop_func));
+	}
+
+	if (cso->colormask & PIPE_MASK_R)
+		cmask |= (1 << 0);
+	if (cso->colormask & PIPE_MASK_G)
+		cmask |= (1 << 4);
+	if (cso->colormask & PIPE_MASK_B)
+		cmask |= (1 << 8);
+	if (cso->colormask & PIPE_MASK_A)
+		cmask |= (1 << 12);
+	so_method(so, tesla, NV50TCL_COLOR_MASK(0), 8);
+	for (i = 0; i < 8; i++)
+		so_data(so, cmask);
+
+	bso->pipe = *cso;
+	so_ref(so, &bso->so);
+	return (void *)bso;
+}
+
+static void
+nv50_blend_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->blend = hwcso;
+	nv50->dirty |= NV50_NEW_BLEND;
+}
+
+static void
+nv50_blend_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_blend_stateobj *bso = hwcso;
+
+	so_ref(NULL, &bso->so);
+	FREE(bso);
+}
+
+static INLINE unsigned
+wrap_mode(unsigned wrap)
+{
+	switch (wrap) {
+	case PIPE_TEX_WRAP_REPEAT:
+		return NV50TSC_1_0_WRAPS_REPEAT;
+	case PIPE_TEX_WRAP_MIRROR_REPEAT:
+		return NV50TSC_1_0_WRAPS_MIRROR_REPEAT;
+	case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
+		return NV50TSC_1_0_WRAPS_CLAMP_TO_EDGE;
+	case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
+		return NV50TSC_1_0_WRAPS_CLAMP_TO_BORDER;
+	case PIPE_TEX_WRAP_CLAMP:
+		return NV50TSC_1_0_WRAPS_CLAMP;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
+		return NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_EDGE;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
+		return NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_BORDER;
+	case PIPE_TEX_WRAP_MIRROR_CLAMP:
+		return NV50TSC_1_0_WRAPS_MIRROR_CLAMP;
+	default:
+		NOUVEAU_ERR("unknown wrap mode: %d\n", wrap);
+		return NV50TSC_1_0_WRAPS_REPEAT;
+	}
+}
+static void *
+nv50_sampler_state_create(struct pipe_context *pipe,
+			  const struct pipe_sampler_state *cso)
+{
+	unsigned *tsc = CALLOC(8, sizeof(unsigned));
+
+	tsc[0] = (0x00024000 |
+		  (wrap_mode(cso->wrap_s) << 0) |
+		  (wrap_mode(cso->wrap_t) << 3) |
+		  (wrap_mode(cso->wrap_r) << 6));
+
+	switch (cso->mag_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		tsc[1] |= NV50TSC_1_1_MAGF_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		tsc[1] |= NV50TSC_1_1_MAGF_NEAREST;
+		break;
+	}
+
+	switch (cso->min_img_filter) {
+	case PIPE_TEX_FILTER_LINEAR:
+		tsc[1] |= NV50TSC_1_1_MINF_LINEAR;
+		break;
+	case PIPE_TEX_FILTER_NEAREST:
+	default:
+		tsc[1] |= NV50TSC_1_1_MINF_NEAREST;
+		break;
+	}
+
+	switch (cso->min_mip_filter) {
+	case PIPE_TEX_MIPFILTER_LINEAR:
+		tsc[1] |= NV50TSC_1_1_MIPF_LINEAR;
+		break;
+	case PIPE_TEX_MIPFILTER_NEAREST:
+		tsc[1] |= NV50TSC_1_1_MIPF_NEAREST;
+		break;
+	case PIPE_TEX_MIPFILTER_NONE:
+	default:
+		tsc[1] |= NV50TSC_1_1_MIPF_NONE;
+		break;
+	}
+
+	return (void *)tsc;
+}
+
+static void
+nv50_sampler_state_bind(struct pipe_context *pipe, unsigned nr, void **sampler)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	int i;
+
+	nv50->sampler_nr = nr;
+	for (i = 0; i < nv50->sampler_nr; i++)
+		nv50->sampler[i] = sampler[i];
+
+	nv50->dirty |= NV50_NEW_SAMPLER;
+}
+
+static void
+nv50_sampler_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	FREE(hwcso);
+}
+
+static void
+nv50_set_sampler_texture(struct pipe_context *pipe, unsigned nr,
+			 struct pipe_texture **pt)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	int i;
+
+	for (i = 0; i < nr; i++)
+		pipe_texture_reference((void *)&nv50->miptree[i], pt[i]);
+	for (i = nr; i < nv50->miptree_nr; i++)
+		pipe_texture_reference((void *)&nv50->miptree[i], NULL);
+
+	nv50->miptree_nr = nr;
+	nv50->dirty |= NV50_NEW_TEXTURE;
+}
+
+static void *
+nv50_rasterizer_state_create(struct pipe_context *pipe,
+			     const struct pipe_rasterizer_state *cso)
+{
+	struct nouveau_stateobj *so = so_new(64, 0);
+	struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla;
+	struct nv50_rasterizer_stateobj *rso =
+		CALLOC_STRUCT(nv50_rasterizer_stateobj);
+
+	/*XXX: ignored
+	 * 	- light_twosize
+	 * 	- point_smooth
+	 * 	- multisample
+	 * 	- point_sprite / sprite_coord_mode
+	 */
+
+	so_method(so, tesla, NV50TCL_SHADE_MODEL, 1);
+	so_data  (so, cso->flatshade ? NV50TCL_SHADE_MODEL_FLAT :
+				       NV50TCL_SHADE_MODEL_SMOOTH);
+
+	so_method(so, tesla, NV50TCL_LINE_WIDTH, 1);
+	so_data  (so, fui(cso->line_width));
+	so_method(so, tesla, NV50TCL_LINE_SMOOTH_ENABLE, 1);
+	so_data  (so, cso->line_smooth ? 1 : 0);
+	if (cso->line_stipple_enable) {
+		so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1);
+		so_data  (so, 1);
+		so_method(so, tesla, NV50TCL_LINE_STIPPLE_PATTERN, 1);
+		so_data  (so, (cso->line_stipple_pattern << 8) |
+			       cso->line_stipple_factor);
+	} else {
+		so_method(so, tesla, NV50TCL_LINE_STIPPLE_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	so_method(so, tesla, NV50TCL_POINT_SIZE, 1);
+	so_data  (so, fui(cso->point_size));
+
+	so_method(so, tesla, NV50TCL_POLYGON_MODE_FRONT, 3);
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+	} else {
+		so_data(so, nvgl_polygon_mode(cso->fill_cw));
+		so_data(so, nvgl_polygon_mode(cso->fill_ccw));
+	}
+	so_data(so, cso->poly_smooth ? 1 : 0);
+
+	so_method(so, tesla, NV50TCL_CULL_FACE_ENABLE, 3);
+	so_data  (so, cso->cull_mode != PIPE_WINDING_NONE);
+	if (cso->front_winding == PIPE_WINDING_CCW) {
+		so_data(so, NV50TCL_FRONT_FACE_CCW);
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV50TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV50TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV50TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV50TCL_CULL_FACE_BACK);
+			break;
+		}
+	} else {
+		so_data(so, NV50TCL_FRONT_FACE_CW);
+		switch (cso->cull_mode) {
+		case PIPE_WINDING_CCW:
+			so_data(so, NV50TCL_CULL_FACE_BACK);
+			break;
+		case PIPE_WINDING_CW:
+			so_data(so, NV50TCL_CULL_FACE_FRONT);
+			break;
+		case PIPE_WINDING_BOTH:
+			so_data(so, NV50TCL_CULL_FACE_FRONT_AND_BACK);
+			break;
+		default:
+			so_data(so, NV50TCL_CULL_FACE_BACK);
+			break;
+		}
+	}
+
+	so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_ENABLE, 1);
+	so_data  (so, cso->poly_stipple_enable ? 1 : 0);
+
+	so_method(so, tesla, NV50TCL_POLYGON_OFFSET_POINT_ENABLE, 3);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_POINT) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_POINT))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_LINE) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_LINE))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+	if ((cso->offset_cw && cso->fill_cw == PIPE_POLYGON_MODE_FILL) ||
+	    (cso->offset_ccw && cso->fill_ccw == PIPE_POLYGON_MODE_FILL))
+		so_data(so, 1);
+	else
+		so_data(so, 0);
+
+	if (cso->offset_cw || cso->offset_ccw) {
+		so_method(so, tesla, NV50TCL_POLYGON_OFFSET_FACTOR, 1);
+		so_data  (so, fui(cso->offset_scale));
+		so_method(so, tesla, NV50TCL_POLYGON_OFFSET_UNITS, 1);
+		so_data  (so, fui(cso->offset_units));
+	}
+
+	rso->pipe = *cso;
+	so_ref(so, &rso->so);
+	return (void *)rso;
+}
+
+static void
+nv50_rasterizer_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->rasterizer = hwcso;
+	nv50->dirty |= NV50_NEW_RASTERIZER;
+}
+
+static void
+nv50_rasterizer_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_rasterizer_stateobj *rso = hwcso;
+
+	so_ref(NULL, &rso->so);
+	FREE(rso);
+}
+
+static void *
+nv50_depth_stencil_alpha_state_create(struct pipe_context *pipe,
+			const struct pipe_depth_stencil_alpha_state *cso)
+{
+	struct nouveau_grobj *tesla = nv50_context(pipe)->screen->tesla;
+	struct nv50_zsa_stateobj *zsa = CALLOC_STRUCT(nv50_zsa_stateobj);
+	struct nouveau_stateobj *so = so_new(64, 0);
+
+	so_method(so, tesla, NV50TCL_DEPTH_WRITE_ENABLE, 1);
+	so_data  (so, cso->depth.writemask ? 1 : 0);
+	if (cso->depth.enabled) {
+		so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1);
+		so_data  (so, 1);
+		so_method(so, tesla, NV50TCL_DEPTH_TEST_FUNC, 1);
+		so_data  (so, nvgl_comparison_op(cso->depth.func));
+	} else {
+		so_method(so, tesla, NV50TCL_DEPTH_TEST_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	/*XXX: yes, I know they're backwards.. header needs fixing */
+	if (cso->stencil[0].enabled) {
+		so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 5);
+		so_data  (so, 1);
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[0].zpass_op));
+		so_data  (so, nvgl_comparison_op(cso->stencil[0].func));
+		so_method(so, tesla, NV50TCL_STENCIL_BACK_FUNC_REF, 3);
+		so_data  (so, cso->stencil[0].ref_value);
+		so_data  (so, cso->stencil[0].write_mask);
+		so_data  (so, cso->stencil[0].value_mask);
+	} else {
+		so_method(so, tesla, NV50TCL_STENCIL_BACK_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	if (cso->stencil[1].enabled) {
+		so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 8);
+		so_data  (so, 1);
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].fail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zfail_op));
+		so_data  (so, nvgl_stencil_op(cso->stencil[1].zpass_op));
+		so_data  (so, nvgl_comparison_op(cso->stencil[1].func));
+		so_data  (so, cso->stencil[1].ref_value);
+		so_data  (so, cso->stencil[1].write_mask);
+		so_data  (so, cso->stencil[1].value_mask);
+	} else {
+		so_method(so, tesla, NV50TCL_STENCIL_FRONT_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	if (cso->alpha.enabled) {
+		so_method(so, tesla, NV50TCL_ALPHA_TEST_ENABLE, 1);
+		so_data  (so, 1);
+		so_method(so, tesla, NV50TCL_ALPHA_TEST_REF, 2);
+		so_data  (so, fui(cso->alpha.ref));
+		so_data  (so, nvgl_comparison_op(cso->alpha.func));
+	} else {
+		so_method(so, tesla, NV50TCL_ALPHA_TEST_ENABLE, 1);
+		so_data  (so, 0);
+	}
+
+	zsa->pipe = *cso;
+	so_ref(so, &zsa->so);
+	return (void *)zsa;
+}
+
+static void
+nv50_depth_stencil_alpha_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->zsa = hwcso;
+	nv50->dirty |= NV50_NEW_ZSA;
+}
+
+static void
+nv50_depth_stencil_alpha_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_zsa_stateobj *zsa = hwcso;
+
+	so_ref(NULL, &zsa->so);
+	FREE(zsa);
+}
+
+static void *
+nv50_vp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv50_program *p = CALLOC_STRUCT(nv50_program);
+
+	p->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	p->type = PIPE_SHADER_VERTEX;
+	tgsi_scan_shader(p->pipe.tokens, &p->info);
+	return (void *)p;
+}
+
+static void
+nv50_vp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->vertprog = hwcso;
+	nv50->dirty |= NV50_NEW_VERTPROG;
+}
+
+static void
+nv50_vp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct nv50_program *p = hwcso;
+
+	nv50_program_destroy(nv50, p);
+	FREE((void*)p->pipe.tokens);
+	FREE(p);
+}
+
+static void *
+nv50_fp_state_create(struct pipe_context *pipe,
+		     const struct pipe_shader_state *cso)
+{
+	struct nv50_program *p = CALLOC_STRUCT(nv50_program);
+
+	p->pipe.tokens = tgsi_dup_tokens(cso->tokens);
+	p->type = PIPE_SHADER_FRAGMENT;
+	tgsi_scan_shader(p->pipe.tokens, &p->info);
+	return (void *)p;
+}
+
+static void
+nv50_fp_state_bind(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->fragprog = hwcso;
+	nv50->dirty |= NV50_NEW_FRAGPROG;
+}
+
+static void
+nv50_fp_state_delete(struct pipe_context *pipe, void *hwcso)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct nv50_program *p = hwcso;
+
+	nv50_program_destroy(nv50, p);
+	FREE((void*)p->pipe.tokens);
+	FREE(p);
+}
+
+static void
+nv50_set_blend_color(struct pipe_context *pipe,
+		     const struct pipe_blend_color *bcol)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->blend_colour = *bcol;
+	nv50->dirty |= NV50_NEW_BLEND_COLOUR;
+}
+
+static void
+nv50_set_clip_state(struct pipe_context *pipe,
+		    const struct pipe_clip_state *clip)
+{
+}
+
+static void
+nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
+			 const struct pipe_constant_buffer *buf )
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	if (shader == PIPE_SHADER_VERTEX) {
+		nv50->constbuf[PIPE_SHADER_VERTEX] = buf->buffer;
+		nv50->dirty |= NV50_NEW_VERTPROG_CB;
+	} else
+	if (shader == PIPE_SHADER_FRAGMENT) {
+		nv50->constbuf[PIPE_SHADER_FRAGMENT] = buf->buffer;
+		nv50->dirty |= NV50_NEW_FRAGPROG_CB;
+	}
+}
+
+static void
+nv50_set_framebuffer_state(struct pipe_context *pipe,
+			   const struct pipe_framebuffer_state *fb)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->framebuffer = *fb;
+	nv50->dirty |= NV50_NEW_FRAMEBUFFER;
+}
+
+static void
+nv50_set_polygon_stipple(struct pipe_context *pipe,
+			 const struct pipe_poly_stipple *stipple)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->stipple = *stipple;
+	nv50->dirty |= NV50_NEW_STIPPLE;
+}
+
+static void
+nv50_set_scissor_state(struct pipe_context *pipe,
+		       const struct pipe_scissor_state *s)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->scissor = *s;
+	nv50->dirty |= NV50_NEW_SCISSOR;
+}
+
+static void
+nv50_set_viewport_state(struct pipe_context *pipe,
+			const struct pipe_viewport_state *vpt)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50->viewport = *vpt;
+	nv50->dirty |= NV50_NEW_VIEWPORT;
+}
+
+static void
+nv50_set_vertex_buffers(struct pipe_context *pipe, unsigned count,
+			const struct pipe_vertex_buffer *vb)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	memcpy(nv50->vtxbuf, vb, sizeof(*vb) * count);
+	nv50->vtxbuf_nr = count;
+
+	nv50->dirty |= NV50_NEW_ARRAYS;
+}
+
+static void
+nv50_set_vertex_elements(struct pipe_context *pipe, unsigned count,
+			 const struct pipe_vertex_element *ve)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	memcpy(nv50->vtxelt, ve, sizeof(*ve) * count);
+	nv50->vtxelt_nr = count;
+
+	nv50->dirty |= NV50_NEW_ARRAYS;
+}
+
+void
+nv50_init_state_functions(struct nv50_context *nv50)
+{
+	nv50->pipe.create_blend_state = nv50_blend_state_create;
+	nv50->pipe.bind_blend_state = nv50_blend_state_bind;
+	nv50->pipe.delete_blend_state = nv50_blend_state_delete;
+
+	nv50->pipe.create_sampler_state = nv50_sampler_state_create;
+	nv50->pipe.bind_sampler_states = nv50_sampler_state_bind;
+	nv50->pipe.delete_sampler_state = nv50_sampler_state_delete;
+	nv50->pipe.set_sampler_textures = nv50_set_sampler_texture;
+
+	nv50->pipe.create_rasterizer_state = nv50_rasterizer_state_create;
+	nv50->pipe.bind_rasterizer_state = nv50_rasterizer_state_bind;
+	nv50->pipe.delete_rasterizer_state = nv50_rasterizer_state_delete;
+
+	nv50->pipe.create_depth_stencil_alpha_state =
+		nv50_depth_stencil_alpha_state_create;
+	nv50->pipe.bind_depth_stencil_alpha_state =
+		nv50_depth_stencil_alpha_state_bind;
+	nv50->pipe.delete_depth_stencil_alpha_state =
+		nv50_depth_stencil_alpha_state_delete;
+
+	nv50->pipe.create_vs_state = nv50_vp_state_create;
+	nv50->pipe.bind_vs_state = nv50_vp_state_bind;
+	nv50->pipe.delete_vs_state = nv50_vp_state_delete;
+
+	nv50->pipe.create_fs_state = nv50_fp_state_create;
+	nv50->pipe.bind_fs_state = nv50_fp_state_bind;
+	nv50->pipe.delete_fs_state = nv50_fp_state_delete;
+
+	nv50->pipe.set_blend_color = nv50_set_blend_color;
+	nv50->pipe.set_clip_state = nv50_set_clip_state;
+	nv50->pipe.set_constant_buffer = nv50_set_constant_buffer;
+	nv50->pipe.set_framebuffer_state = nv50_set_framebuffer_state;
+	nv50->pipe.set_polygon_stipple = nv50_set_polygon_stipple;
+	nv50->pipe.set_scissor_state = nv50_set_scissor_state;
+	nv50->pipe.set_viewport_state = nv50_set_viewport_state;
+
+	nv50->pipe.set_vertex_buffers = nv50_set_vertex_buffers;
+	nv50->pipe.set_vertex_elements = nv50_set_vertex_elements;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c
new file mode 100644
index 00000000000..198e25f4482
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_state_validate.c
@@ -0,0 +1,309 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "nv50_context.h"
+#include "nouveau/nouveau_stateobj.h"
+
+static void
+nv50_state_validate_fb(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_stateobj *so = so_new(128, 18);
+	struct pipe_framebuffer_state *fb = &nv50->framebuffer;
+	unsigned i, w, h, gw = 0;
+
+	for (i = 0; i < fb->num_cbufs; i++) {
+		if (!gw) {
+			w = fb->cbufs[i]->width;
+			h = fb->cbufs[i]->height;
+			gw = 1;
+		} else {
+			assert(w == fb->cbufs[i]->width);
+			assert(h == fb->cbufs[i]->height);
+		}
+
+		so_method(so, tesla, NV50TCL_RT_HORIZ(i), 2);
+		so_data  (so, fb->cbufs[i]->width);
+		so_data  (so, fb->cbufs[i]->height);
+
+		so_method(so, tesla, NV50TCL_RT_ADDRESS_HIGH(i), 5);
+		so_reloc (so, fb->cbufs[i]->buffer, fb->cbufs[i]->offset,
+			  NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0);
+		so_reloc (so, fb->cbufs[i]->buffer, fb->cbufs[i]->offset,
+			  NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0);
+		switch (fb->cbufs[i]->format) {
+		case PIPE_FORMAT_A8R8G8B8_UNORM:
+			so_data(so, 0xcf);
+			break;
+		case PIPE_FORMAT_R5G6B5_UNORM:
+			so_data(so, 0xe8);
+			break;
+		default:
+			NOUVEAU_ERR("AIIII unknown format %s\n",
+				    pf_name(fb->cbufs[i]->format));
+			so_data(so, 0xe6);
+			break;
+		}
+		so_data(so, 0x00000000);
+		so_data(so, 0x00000000);
+
+		so_method(so, tesla, 0x1224, 1);
+		so_data  (so, 1);
+	}
+
+	if (fb->zsbuf) {
+		if (!gw) {
+			w = fb->zsbuf->width;
+			h = fb->zsbuf->height;
+			gw = 1;
+		} else {
+			assert(w == fb->zsbuf->width);
+			assert(h == fb->zsbuf->height);
+		}
+
+		so_method(so, tesla, NV50TCL_ZETA_ADDRESS_HIGH, 5);
+		so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset,
+			  NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0);
+		so_reloc (so, fb->zsbuf->buffer, fb->zsbuf->offset,
+			  NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0);
+		switch (fb->zsbuf->format) {
+		case PIPE_FORMAT_Z24S8_UNORM:
+			so_data(so, 0x16);
+			break;
+		case PIPE_FORMAT_Z16_UNORM:
+			so_data(so, 0x15);
+			break;
+		default:
+			NOUVEAU_ERR("AIIII unknown format %s\n",
+				    pf_name(fb->zsbuf->format));
+			so_data(so, 0x16);
+			break;
+		}
+		so_data(so, 0x00000000);
+		so_data(so, 0x00000000);
+
+		so_method(so, tesla, 0x1538, 1);
+		so_data  (so, 1);
+		so_method(so, tesla, 0x1228, 3);
+		so_data  (so, fb->zsbuf->width);
+		so_data  (so, fb->zsbuf->height);
+		so_data  (so, 0x00010001);
+	}
+
+	so_method(so, tesla, NV50TCL_VIEWPORT_HORIZ, 2);
+	so_data  (so, w << 16);
+	so_data  (so, h << 16);
+	so_method(so, tesla, 0x0e04, 2);
+	so_data  (so, w << 16);
+	so_data  (so, h << 16);
+	so_method(so, tesla, 0xdf8, 2);
+	so_data  (so, 0);
+	so_data  (so, h);
+
+	so_ref(so, &nv50->state.fb);
+}
+
+static void
+nv50_state_emit(struct nv50_context *nv50)
+{
+	struct nv50_screen *screen = nv50->screen;
+	struct nouveau_winsys *nvws = screen->nvws;
+
+	if (nv50->pctx_id != screen->cur_pctx) {
+		nv50->state.dirty |= 0xffffffff;
+		screen->cur_pctx = nv50->pctx_id;
+	}
+
+	if (nv50->state.dirty & NV50_NEW_FRAMEBUFFER)
+		so_emit(nvws, nv50->state.fb);
+	if (nv50->state.dirty & NV50_NEW_BLEND)
+		so_emit(nvws, nv50->state.blend);
+	if (nv50->state.dirty & NV50_NEW_ZSA)
+		so_emit(nvws, nv50->state.zsa);
+	if (nv50->state.dirty & NV50_NEW_VERTPROG)
+		so_emit(nvws, nv50->state.vertprog);
+	if (nv50->state.dirty & NV50_NEW_FRAGPROG)
+		so_emit(nvws, nv50->state.fragprog);
+	if (nv50->state.dirty & NV50_NEW_RASTERIZER)
+		so_emit(nvws, nv50->state.rast);
+	if (nv50->state.dirty & NV50_NEW_BLEND_COLOUR)
+		so_emit(nvws, nv50->state.blend_colour);
+	if (nv50->state.dirty & NV50_NEW_STIPPLE)
+		so_emit(nvws, nv50->state.stipple);
+	if (nv50->state.dirty & NV50_NEW_SCISSOR)
+		so_emit(nvws, nv50->state.scissor);
+	if (nv50->state.dirty & NV50_NEW_VIEWPORT)
+		so_emit(nvws, nv50->state.viewport);
+	if (nv50->state.dirty & NV50_NEW_SAMPLER)
+		so_emit(nvws, nv50->state.tsc_upload);
+	if (nv50->state.dirty & NV50_NEW_TEXTURE)
+		so_emit(nvws, nv50->state.tic_upload);
+	if (nv50->state.dirty & NV50_NEW_ARRAYS) {
+		so_emit(nvws, nv50->state.vtxfmt);
+		so_emit(nvws, nv50->state.vtxbuf);
+	}
+	nv50->state.dirty = 0;
+
+	so_emit_reloc_markers(nvws, nv50->state.fb);
+	so_emit_reloc_markers(nvws, nv50->state.vertprog);
+	so_emit_reloc_markers(nvws, nv50->state.fragprog);
+	so_emit_reloc_markers(nvws, nv50->state.vtxbuf);
+	so_emit_reloc_markers(nvws, nv50->screen->static_init);
+}
+
+boolean
+nv50_state_validate(struct nv50_context *nv50)
+{
+	const struct pipe_framebuffer_state *fb = &nv50->framebuffer;
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_stateobj *so;
+	unsigned i;
+
+	for (i = 0; i < fb->num_cbufs; i++)
+		fb->cbufs[i]->status = PIPE_SURFACE_STATUS_DEFINED;
+
+	if (fb->zsbuf)
+		fb->zsbuf->status = PIPE_SURFACE_STATUS_DEFINED;
+
+	if (nv50->dirty & NV50_NEW_FRAMEBUFFER)
+		nv50_state_validate_fb(nv50);
+
+	if (nv50->dirty & NV50_NEW_BLEND)
+		so_ref(nv50->blend->so, &nv50->state.blend);
+
+	if (nv50->dirty & NV50_NEW_ZSA)
+		so_ref(nv50->zsa->so, &nv50->state.zsa);
+
+	if (nv50->dirty & (NV50_NEW_VERTPROG | NV50_NEW_VERTPROG_CB))
+		nv50_vertprog_validate(nv50);
+
+	if (nv50->dirty & (NV50_NEW_FRAGPROG | NV50_NEW_FRAGPROG_CB))
+		nv50_fragprog_validate(nv50);
+
+	if (nv50->dirty & NV50_NEW_RASTERIZER)
+		so_ref(nv50->rasterizer->so, &nv50->state.rast);
+
+	if (nv50->dirty & NV50_NEW_BLEND_COLOUR) {
+		so = so_new(5, 0);
+		so_method(so, tesla, NV50TCL_BLEND_COLOR(0), 4);
+		so_data  (so, fui(nv50->blend_colour.color[0]));
+		so_data  (so, fui(nv50->blend_colour.color[1]));
+		so_data  (so, fui(nv50->blend_colour.color[2]));
+		so_data  (so, fui(nv50->blend_colour.color[3]));
+		so_ref(so, &nv50->state.blend_colour);
+	}
+
+	if (nv50->dirty & NV50_NEW_STIPPLE) {
+		so = so_new(33, 0);
+		so_method(so, tesla, NV50TCL_POLYGON_STIPPLE_PATTERN(0), 32);
+		for (i = 0; i < 32; i++)
+			so_data(so, nv50->stipple.stipple[i]);
+		so_ref(so, &nv50->state.stipple);
+	}
+
+	if (nv50->dirty & (NV50_NEW_SCISSOR | NV50_NEW_RASTERIZER)) {
+		struct pipe_rasterizer_state *rast = &nv50->rasterizer->pipe;
+		struct pipe_scissor_state *s = &nv50->scissor;
+
+		if (nv50->state.scissor &&
+		    (rast->scissor == 0 && nv50->state.scissor_enabled == 0))
+			goto scissor_uptodate;
+		nv50->state.scissor_enabled = rast->scissor;
+
+		so = so_new(3, 0);
+		so_method(so, tesla, 0x0ff4, 2);
+		if (nv50->state.scissor_enabled) {
+			so_data(so, ((s->maxx - s->minx) << 16) | s->minx);
+			so_data(so, ((s->maxy - s->miny) << 16) | s->miny);
+		} else {
+			so_data(so, (8192 << 16));
+			so_data(so, (8192 << 16));
+		}
+		so_ref(so, &nv50->state.scissor);
+		nv50->state.dirty |= NV50_NEW_SCISSOR;
+	}
+scissor_uptodate:
+
+	if (nv50->dirty & NV50_NEW_VIEWPORT) {
+		unsigned bypass;
+
+		if (!nv50->rasterizer->pipe.bypass_clipping)
+			bypass = 0;
+		else
+			bypass = 1;
+
+		if (nv50->state.viewport &&
+		    (bypass || !(nv50->dirty & NV50_NEW_VIEWPORT)) &&
+		    nv50->state.viewport_bypass == bypass)
+			goto viewport_uptodate;
+		nv50->state.viewport_bypass = bypass;
+
+		so = so_new(12, 0);
+		if (!bypass) {
+			so_method(so, tesla, NV50TCL_VIEWPORT_UNK1(0), 3);
+			so_data  (so, fui(nv50->viewport.translate[0]));
+			so_data  (so, fui(nv50->viewport.translate[1]));
+			so_data  (so, fui(nv50->viewport.translate[2]));
+			so_method(so, tesla, NV50TCL_VIEWPORT_UNK0(0), 3);
+			so_data  (so, fui(nv50->viewport.scale[0]));
+			so_data  (so, fui(-nv50->viewport.scale[1]));
+			so_data  (so, fui(nv50->viewport.scale[2]));
+			so_method(so, tesla, 0x192c, 1);
+			so_data  (so, 1);
+			so_method(so, tesla, 0x0f90, 1);
+			so_data  (so, 0);
+		} else {
+			so_method(so, tesla, 0x192c, 1);
+			so_data  (so, 0);
+			so_method(so, tesla, 0x0f90, 1);
+			so_data  (so, 1);
+		}
+
+		so_ref(so, &nv50->state.viewport);
+	}
+viewport_uptodate:
+
+	if (nv50->dirty & NV50_NEW_SAMPLER) {
+		int i;
+
+		so = so_new(nv50->sampler_nr * 8 + 3, 0);
+		so_method(so, tesla, 0x0f00, 1);
+		so_data  (so, NV50_CB_TSC);
+		so_method(so, tesla, 0x40000f04, nv50->sampler_nr * 8);
+		for (i = 0; i < nv50->sampler_nr; i++)
+			so_datap (so, nv50->sampler[i], 8);
+		so_ref(so, &nv50->state.tsc_upload);
+	}
+
+	if (nv50->dirty & NV50_NEW_TEXTURE)
+		nv50_tex_validate(nv50);
+
+	if (nv50->dirty & NV50_NEW_ARRAYS)
+		nv50_vbo_validate(nv50);
+
+	nv50->state.dirty |= nv50->dirty;
+	nv50->dirty = 0;
+	nv50_state_emit(nv50);
+
+	return TRUE;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c
new file mode 100644
index 00000000000..5bf97d3a6bf
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_surface.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "nv50_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_winsys.h"
+#include "pipe/p_inlines.h"
+
+#include "util/u_tile.h"
+
+static void
+nv50_surface_copy(struct pipe_context *pipe, boolean flip,
+		  struct pipe_surface *dest, unsigned destx, unsigned desty,
+		  struct pipe_surface *src, unsigned srcx, unsigned srcy,
+		  unsigned width, unsigned height)
+{
+	struct nv50_context *nv50 = (struct nv50_context *)pipe;
+	struct nouveau_winsys *nvws = nv50->screen->nvws;
+
+	if (flip) {
+		desty += height;
+		while (height--) {
+			nvws->surface_copy(nvws, dest, destx, desty--, src,
+					   srcx, srcy++, width, 1);
+		}
+	} else {
+		nvws->surface_copy(nvws, dest, destx, desty, src, srcx, srcy,
+				   width, height);
+	}
+}
+
+static void
+nv50_surface_fill(struct pipe_context *pipe, struct pipe_surface *dest,
+		  unsigned destx, unsigned desty, unsigned width,
+		  unsigned height, unsigned value)
+{
+	struct nv50_context *nv50 = (struct nv50_context *)pipe;
+	struct nouveau_winsys *nvws = nv50->screen->nvws;
+
+	nvws->surface_fill(nvws, dest, destx, desty, width, height, value);
+}
+
+static void *
+nv50_surface_map(struct pipe_screen *screen, struct pipe_surface *ps,
+		 unsigned flags )
+{
+	struct nouveau_winsys *nvws = nv50_screen(screen)->nvws;
+	struct pipe_winsys *ws = screen->winsys;
+	struct nv50_surface *s = nv50_surface(ps);
+	struct nv50_surface m = *s;
+	void *map;
+
+	if (!s->untiled) {
+		s->untiled = ws->buffer_create(ws, 0, 0, ps->buffer->size);
+
+		m.base.buffer = s->untiled;
+		nvws->surface_copy(nvws, &m.base, 0, 0, &s->base, 0, 0,
+					 ps->width, ps->height);
+	}
+
+	/* Map original tiled surface to disallow it being validated while
+	 * untiled mirror is mapped.
+	 */
+	ws->buffer_map(ws, ps->buffer, flags);
+
+	map = ws->buffer_map(ws, s->untiled, flags);
+	if (!map)
+		return NULL;
+
+	return map;
+}
+
+static void
+nv50_surface_unmap(struct pipe_screen *pscreen, struct pipe_surface *ps)
+{
+	struct nouveau_winsys *nvws = nv50_screen(pscreen)->nvws;
+	struct pipe_winsys *ws = pscreen->winsys;
+	struct nv50_surface *s = nv50_surface(ps);
+	struct nv50_surface m = *s;
+
+	ws->buffer_unmap(ws, s->untiled);
+	ws->buffer_unmap(ws, ps->buffer);
+
+	m.base.buffer = s->untiled;
+	nvws->surface_copy(nvws, &s->base, 0, 0, &m.base, 0, 0,
+				 ps->width, ps->height);
+
+	pipe_buffer_reference(pscreen, &s->untiled, NULL);
+}
+
+void
+nv50_init_surface_functions(struct nv50_context *nv50)
+{
+	nv50->pipe.surface_copy = nv50_surface_copy;
+	nv50->pipe.surface_fill = nv50_surface_fill;
+}
+
+void
+nv50_surface_init_screen_functions(struct pipe_screen *pscreen)
+{
+	pscreen->surface_map = nv50_surface_map;
+	pscreen->surface_unmap = nv50_surface_unmap;
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_tex.c b/src/gallium/drivers/nv50/nv50_tex.c
new file mode 100644
index 00000000000..fde3c97c059
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_tex.c
@@ -0,0 +1,124 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "nv50_context.h"
+#include "nv50_texture.h"
+
+#include "nouveau/nouveau_stateobj.h"
+
+static int
+nv50_tex_construct(struct nouveau_stateobj *so, struct nv50_miptree *mt)
+{
+	switch (mt->base.format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_8_8_8_8);
+		break;
+	case PIPE_FORMAT_A1R5G5B5_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_1_5_5_5);
+		break;
+	case PIPE_FORMAT_A4R4G4B4_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C3 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_4_4_4_4);
+		break;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C1 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C2 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_5_6_5);
+		break;
+	case PIPE_FORMAT_L8_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_ONE | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_8);
+		break;
+	case PIPE_FORMAT_A8_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_ZERO | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_ZERO | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_ZERO | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_8);
+		break;
+	case PIPE_FORMAT_I8_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C0 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_8);
+		break;
+	case PIPE_FORMAT_A8L8_UNORM:
+		so_data(so, NV50TIC_0_0_MAPA_C1 | NV50TIC_0_0_TYPEA_UNORM |
+			    NV50TIC_0_0_MAPR_C0 | NV50TIC_0_0_TYPER_UNORM |
+			    NV50TIC_0_0_MAPG_C0 | NV50TIC_0_0_TYPEG_UNORM |
+			    NV50TIC_0_0_MAPB_C0 | NV50TIC_0_0_TYPEB_UNORM |
+			    NV50TIC_0_0_FMT_8_8);
+		break;
+	default:
+		return 1;
+	}
+
+	so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_LOW, 0, 0);
+	so_data (so, 0xd0005000);
+	so_data (so, 0x00300000);
+	so_data (so, mt->base.width[0]);
+	so_data (so, (mt->base.depth[0] << 16) | mt->base.height[0]);
+	so_data (so, 0x03000000);
+	so_reloc(so, mt->buffer, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_HIGH, 0, 0);
+
+	return 0;
+}
+
+void
+nv50_tex_validate(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_stateobj *so;
+	int i;
+
+	so = so_new(nv50->miptree_nr * 8 + 3, nv50->miptree_nr * 2);
+	so_method(so, tesla, 0x0f00, 1);
+	so_data  (so, NV50_CB_TIC);
+	so_method(so, tesla, 0x40000f04, nv50->miptree_nr * 8);
+	for (i = 0; i < nv50->miptree_nr; i++) {
+		if (nv50_tex_construct(so, nv50->miptree[i])) {
+			NOUVEAU_ERR("failed tex validate\n");
+			so_ref(NULL, &so);
+			return;
+		}
+	}
+
+	so_ref(so, &nv50->state.tic_upload);
+}
+
diff --git a/src/gallium/drivers/nv50/nv50_texture.h b/src/gallium/drivers/nv50/nv50_texture.h
new file mode 100644
index 00000000000..6861d67b4d5
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_texture.h
@@ -0,0 +1,126 @@
+#ifndef __NV50_TEXTURE_H__
+#define __NV50_TEXTURE_H__
+
+/* It'd be really nice to have these in nouveau_class.h generated by
+ * renouveau like the rest of the object header - but not sure it can
+ * handle non-object stuff nicely - need to look into it.
+ */
+
+/* Texture image control block */
+#define NV50TIC_0_0_MAPA_MASK                                     0x38000000
+#define NV50TIC_0_0_MAPA_ZERO                                     0x00000000
+#define NV50TIC_0_0_MAPA_C0                                       0x10000000
+#define NV50TIC_0_0_MAPA_C1                                       0x18000000
+#define NV50TIC_0_0_MAPA_C2                                       0x20000000
+#define NV50TIC_0_0_MAPA_C3                                       0x28000000
+#define NV50TIC_0_0_MAPA_ONE                                      0x38000000
+#define NV50TIC_0_0_MAPR_MASK                                     0x07000000
+#define NV50TIC_0_0_MAPR_ZERO                                     0x00000000
+#define NV50TIC_0_0_MAPR_C0                                       0x02000000
+#define NV50TIC_0_0_MAPR_C1                                       0x03000000
+#define NV50TIC_0_0_MAPR_C2                                       0x04000000
+#define NV50TIC_0_0_MAPR_C3                                       0x05000000
+#define NV50TIC_0_0_MAPR_ONE                                      0x07000000
+#define NV50TIC_0_0_MAPG_MASK                                     0x00e00000
+#define NV50TIC_0_0_MAPG_ZERO                                     0x00000000
+#define NV50TIC_0_0_MAPG_C0                                       0x00400000
+#define NV50TIC_0_0_MAPG_C1                                       0x00600000
+#define NV50TIC_0_0_MAPG_C2                                       0x00800000
+#define NV50TIC_0_0_MAPG_C3                                       0x00a00000
+#define NV50TIC_0_0_MAPG_ONE                                      0x00e00000
+#define NV50TIC_0_0_MAPB_MASK                                     0x001c0000
+#define NV50TIC_0_0_MAPB_ZERO                                     0x00000000
+#define NV50TIC_0_0_MAPB_C0                                       0x00080000
+#define NV50TIC_0_0_MAPB_C1                                       0x000c0000
+#define NV50TIC_0_0_MAPB_C2                                       0x00100000
+#define NV50TIC_0_0_MAPB_C3                                       0x00140000
+#define NV50TIC_0_0_MAPB_ONE                                      0x001c0000
+#define NV50TIC_0_0_TYPEA_MASK                                    0x00038000
+#define NV50TIC_0_0_TYPEA_UNORM                                   0x00010000
+#define NV50TIC_0_0_TYPER_MASK                                    0x00007000
+#define NV50TIC_0_0_TYPER_UNORM                                   0x00002000
+#define NV50TIC_0_0_TYPEG_MASK                                    0x00000e00
+#define NV50TIC_0_0_TYPEG_UNORM                                   0x00000400
+#define NV50TIC_0_0_TYPEB_MASK                                    0x000001c0
+#define NV50TIC_0_0_TYPEB_UNORM                                   0x00000080
+#define NV50TIC_0_0_FMT_MASK                                      0x0000003c
+#define NV50TIC_0_0_FMT_8_8_8_8                                   0x00000008
+#define NV50TIC_0_0_FMT_4_4_4_4                                   0x00000012
+#define NV50TIC_0_0_FMT_1_5_5_5                                   0x00000013
+#define NV50TIC_0_0_FMT_5_6_5                                     0x00000015
+#define NV50TIC_0_0_FMT_8_8                                       0x00000018
+#define NV50TIC_0_0_FMT_8                                         0x0000001d
+
+#define NV50TIC_0_1_OFFSET_LOW_MASK                               0xffffffff
+#define NV50TIC_0_1_OFFSET_LOW_SHIFT                                       0
+
+#define NV50TIC_0_2_UNKNOWN_MASK                                  0xffffffff
+
+#define NV50TIC_0_3_UNKNOWN_MASK                                  0xffffffff
+
+#define NV50TIC_0_4_WIDTH_MASK                                    0x0000ffff
+#define NV50TIC_0_4_WIDTH_SHIFT                                            0
+
+#define NV50TIC_0_5_DEPTH_MASK                                    0xffff0000
+#define NV50TIC_0_5_DEPTH_SHIFT                                           16
+#define NV50TIC_0_5_HEIGHT_MASK                                   0x0000ffff
+#define NV50TIC_0_5_HEIGHT_SHIFT                                           0
+
+#define NV50TIC_0_6_UNKNOWN_MASK                                  0xffffffff
+
+#define NV50TIC_0_7_OFFSET_HIGH_MASK                              0xffffffff
+#define NV50TIC_0_7_OFFSET_HIGH_SHIFT                                      0
+
+/* Texture sampler control block */
+#define NV50TSC_1_0_WRAPS_MASK                                   0x00000007
+#define NV50TSC_1_0_WRAPS_REPEAT                                 0x00000000
+#define NV50TSC_1_0_WRAPS_MIRROR_REPEAT                          0x00000001
+#define NV50TSC_1_0_WRAPS_CLAMP_TO_EDGE                          0x00000002
+#define NV50TSC_1_0_WRAPS_CLAMP_TO_BORDER                        0x00000003
+#define NV50TSC_1_0_WRAPS_CLAMP                                  0x00000004
+#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_EDGE                   0x00000005
+#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP_TO_BORDER                 0x00000006
+#define NV50TSC_1_0_WRAPS_MIRROR_CLAMP                           0x00000007
+#define NV50TSC_1_0_WRAPT_MASK                                   0x00000038
+#define NV50TSC_1_0_WRAPT_REPEAT                                 0x00000000
+#define NV50TSC_1_0_WRAPT_MIRROR_REPEAT                          0x00000008
+#define NV50TSC_1_0_WRAPT_CLAMP_TO_EDGE                          0x00000010
+#define NV50TSC_1_0_WRAPT_CLAMP_TO_BORDER                        0x00000018
+#define NV50TSC_1_0_WRAPT_CLAMP                                  0x00000020
+#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP_TO_EDGE                   0x00000028
+#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP_TO_BORDER                 0x00000030
+#define NV50TSC_1_0_WRAPT_MIRROR_CLAMP                           0x00000038
+#define NV50TSC_1_0_WRAPR_MASK                                   0x000001c0
+#define NV50TSC_1_0_WRAPR_REPEAT                                 0x00000000
+#define NV50TSC_1_0_WRAPR_MIRROR_REPEAT                          0x00000040
+#define NV50TSC_1_0_WRAPR_CLAMP_TO_EDGE                          0x00000080
+#define NV50TSC_1_0_WRAPR_CLAMP_TO_BORDER                        0x000000c0
+#define NV50TSC_1_0_WRAPR_CLAMP                                  0x00000100
+#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP_TO_EDGE                   0x00000140
+#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP_TO_BORDER                 0x00000180
+#define NV50TSC_1_0_WRAPR_MIRROR_CLAMP                           0x000001c0
+
+#define NV50TSC_1_1_MAGF_MASK                                    0x00000003
+#define NV50TSC_1_1_MAGF_NEAREST                                 0x00000001
+#define NV50TSC_1_1_MAGF_LINEAR                                  0x00000002
+#define NV50TSC_1_1_MINF_MASK                                    0x00000030
+#define NV50TSC_1_1_MINF_NEAREST                                 0x00000010
+#define NV50TSC_1_1_MINF_LINEAR                                  0x00000020
+#define NV50TSC_1_1_MIPF_MASK                                    0x000000c0
+#define NV50TSC_1_1_MIPF_NONE                                    0x00000040
+#define NV50TSC_1_1_MIPF_NEAREST                                 0x00000080
+#define NV50TSC_1_1_MIPF_LINEAR                                  0x000000c0
+
+#define NV50TSC_1_2_UNKNOWN_MASK                                 0xffffffff
+
+#define NV50TSC_1_3_UNKNOWN_MASK                                 0xffffffff
+
+#define NV50TSC_1_4_UNKNOWN_MASK                                 0xffffffff
+
+#define NV50TSC_1_5_UNKNOWN_MASK                                 0xffffffff
+
+#define NV50TSC_1_6_UNKNOWN_MASK                                 0xffffffff
+
+#define NV50TSC_1_7_UNKNOWN_MASK                                 0xffffffff
+
+#endif
diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c
new file mode 100644
index 00000000000..584336682e4
--- /dev/null
+++ b/src/gallium/drivers/nv50/nv50_vbo.c
@@ -0,0 +1,237 @@
+/*
+ * Copyright 2008 Ben Skeggs
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+
+#include "nv50_context.h"
+
+static INLINE unsigned
+nv50_prim(unsigned mode)
+{
+	switch (mode) {
+	case PIPE_PRIM_POINTS: return NV50TCL_VERTEX_BEGIN_POINTS;
+	case PIPE_PRIM_LINES: return NV50TCL_VERTEX_BEGIN_LINES;
+	case PIPE_PRIM_LINE_LOOP: return NV50TCL_VERTEX_BEGIN_LINE_LOOP;
+	case PIPE_PRIM_LINE_STRIP: return NV50TCL_VERTEX_BEGIN_LINE_STRIP;
+	case PIPE_PRIM_TRIANGLES: return NV50TCL_VERTEX_BEGIN_TRIANGLES;
+	case PIPE_PRIM_TRIANGLE_STRIP:
+		return NV50TCL_VERTEX_BEGIN_TRIANGLE_STRIP;
+	case PIPE_PRIM_TRIANGLE_FAN: return NV50TCL_VERTEX_BEGIN_TRIANGLE_FAN;
+	case PIPE_PRIM_QUADS: return NV50TCL_VERTEX_BEGIN_QUADS;
+	case PIPE_PRIM_QUAD_STRIP: return NV50TCL_VERTEX_BEGIN_QUAD_STRIP;
+	case PIPE_PRIM_POLYGON: return NV50TCL_VERTEX_BEGIN_POLYGON;
+	default:
+		break;
+	}
+
+	NOUVEAU_ERR("invalid primitive type %d\n", mode);
+	return NV50TCL_VERTEX_BEGIN_POINTS;
+}
+
+boolean
+nv50_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start,
+		 unsigned count)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+
+	nv50_state_validate(nv50);
+
+	BEGIN_RING(tesla, 0x142c, 1);
+	OUT_RING  (0);
+	BEGIN_RING(tesla, 0x142c, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(tesla, NV50TCL_VERTEX_BEGIN, 1);
+	OUT_RING  (nv50_prim(mode));
+	BEGIN_RING(tesla, NV50TCL_VERTEX_BUFFER_FIRST, 2);
+	OUT_RING  (start);
+	OUT_RING  (count);
+	BEGIN_RING(tesla, NV50TCL_VERTEX_END, 1);
+	OUT_RING  (0);
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+static INLINE void
+nv50_draw_elements_inline_u08(struct nv50_context *nv50, uint8_t *map,
+			      unsigned start, unsigned count)
+{
+	map += start;
+
+	if (count & 1) {
+		BEGIN_RING(tesla, 0x15e8, 1);
+		OUT_RING  (map[0]);
+		map++;
+		count--;
+	}
+
+	while (count) {
+		unsigned nr = count > 2046 ? 2046 : count;
+		int i;
+
+		BEGIN_RING(tesla, 0x400015f0, nr >> 1);
+		for (i = 0; i < nr; i += 2)
+			OUT_RING  ((map[1] << 16) | map[0]);
+
+		count -= nr;
+		map += nr;
+	}
+}
+
+static INLINE void
+nv50_draw_elements_inline_u16(struct nv50_context *nv50, uint16_t *map,
+			      unsigned start, unsigned count)
+{
+	map += start;
+
+	if (count & 1) {
+		BEGIN_RING(tesla, 0x15e8, 1);
+		OUT_RING  (map[0]);
+		map++;
+		count--;
+	}
+
+	while (count) {
+		unsigned nr = count > 2046 ? 2046 : count;
+		int i;
+
+		BEGIN_RING(tesla, 0x400015f0, nr >> 1);
+		for (i = 0; i < nr; i += 2)
+			OUT_RING  ((map[1] << 16) | map[0]);
+
+		count -= nr;
+		map += nr;
+	}
+}
+
+static INLINE void
+nv50_draw_elements_inline_u32(struct nv50_context *nv50, uint8_t *map,
+			      unsigned start, unsigned count)
+{
+	map += start;
+
+	while (count) {
+		unsigned nr = count > 2047 ? 2047 : count;
+
+		BEGIN_RING(tesla, 0x400015e8, nr);
+		OUT_RINGp (map, nr);
+
+		count -= nr;
+		map += nr;
+	}
+}
+
+boolean
+nv50_draw_elements(struct pipe_context *pipe,
+		   struct pipe_buffer *indexBuffer, unsigned indexSize,
+		   unsigned mode, unsigned start, unsigned count)
+{
+	struct nv50_context *nv50 = nv50_context(pipe);
+	struct pipe_winsys *ws = pipe->winsys;
+	void *map = ws->buffer_map(ws, indexBuffer, PIPE_BUFFER_USAGE_CPU_READ);
+
+	nv50_state_validate(nv50);
+
+	BEGIN_RING(tesla, 0x142c, 1);
+	OUT_RING  (0);
+	BEGIN_RING(tesla, 0x142c, 1);
+	OUT_RING  (0);
+
+	BEGIN_RING(tesla, NV50TCL_VERTEX_BEGIN, 1);
+	OUT_RING  (nv50_prim(mode));
+	switch (indexSize) {
+	case 1:
+		nv50_draw_elements_inline_u08(nv50, map, start, count);
+		break;
+	case 2:
+		nv50_draw_elements_inline_u16(nv50, map, start, count);
+		break;
+	case 4:
+		nv50_draw_elements_inline_u32(nv50, map, start, count);
+		break;
+	default:
+		assert(0);
+	}
+	BEGIN_RING(tesla, NV50TCL_VERTEX_END, 1);
+	OUT_RING  (0);
+
+	pipe->flush(pipe, 0, NULL);
+	return TRUE;
+}
+
+void
+nv50_vbo_validate(struct nv50_context *nv50)
+{
+	struct nouveau_grobj *tesla = nv50->screen->tesla;
+	struct nouveau_stateobj *vtxbuf, *vtxfmt;
+	int i, vpi = 0;
+
+	vtxbuf = so_new(nv50->vtxelt_nr * 4, nv50->vtxelt_nr * 2);
+	vtxfmt = so_new(nv50->vtxelt_nr + 1, 0);
+	so_method(vtxfmt, tesla, 0x1ac0, nv50->vtxelt_nr);
+
+	for (i = 0; i < nv50->vtxelt_nr; i++) {
+		struct pipe_vertex_element *ve = &nv50->vtxelt[i];
+		struct pipe_vertex_buffer *vb =
+			&nv50->vtxbuf[ve->vertex_buffer_index];
+
+		switch (ve->src_format) {
+		case PIPE_FORMAT_R32G32B32A32_FLOAT:
+			so_data(vtxfmt, 0x7e080000 | i);
+			break;
+		case PIPE_FORMAT_R32G32B32_FLOAT:
+			so_data(vtxfmt, 0x7e100000 | i);
+			break;
+		case PIPE_FORMAT_R32G32_FLOAT:
+			so_data(vtxfmt, 0x7e200000 | i);
+			break;
+		case PIPE_FORMAT_R32_FLOAT:
+			so_data(vtxfmt, 0x7e900000 | i);
+			break;
+		case PIPE_FORMAT_R8G8B8A8_UNORM:
+			so_data(vtxfmt, 0x24500000 | i);
+			break;
+		default:
+		{
+			NOUVEAU_ERR("invalid vbo format %s\n",
+				    pf_name(ve->src_format));
+			assert(0);
+			return;
+		}
+		}
+
+		so_method(vtxbuf, tesla, 0x900 + (i * 16), 3);
+		so_data  (vtxbuf, 0x20000000 | vb->pitch);
+		so_reloc (vtxbuf, vb->buffer, vb->buffer_offset +
+			  ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART |
+			  NOUVEAU_BO_RD | NOUVEAU_BO_HIGH, 0, 0);
+		so_reloc (vtxbuf, vb->buffer, vb->buffer_offset +
+			  ve->src_offset, NOUVEAU_BO_VRAM | NOUVEAU_BO_GART |
+			  NOUVEAU_BO_RD | NOUVEAU_BO_LOW, 0, 0);
+	}
+
+	so_ref (vtxfmt, &nv50->state.vtxfmt);
+	so_ref (vtxbuf, &nv50->state.vtxbuf);
+}
+
diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c
index cd1e6663d86..99b52748575 100644
--- a/src/gallium/drivers/softpipe/sp_context.c
+++ b/src/gallium/drivers/softpipe/sp_context.c
@@ -221,6 +221,14 @@ softpipe_create( struct pipe_screen *screen,
       softpipe->quad[i].output = sp_quad_output_stage(softpipe);
    }
 
+   for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
+      softpipe->tgsi.samplers[i].base.get_samples = sp_get_samples;
+      softpipe->tgsi.samplers[i].unit = i;
+      softpipe->tgsi.samplers[i].sp = softpipe;
+      softpipe->tgsi.samplers[i].cache = softpipe->tex_cache[i];
+      softpipe->tgsi.samplers_list[i] = &softpipe->tgsi.samplers[i];
+   }
+
    /*
     * Create drawing context and plug our rendering stage into it.
     */
@@ -228,6 +236,9 @@ softpipe_create( struct pipe_screen *screen,
    if (!softpipe->draw) 
       goto fail;
 
+   draw_texture_samplers(softpipe->draw,
+                         PIPE_MAX_SAMPLERS, softpipe->tgsi.samplers_list);
+
    softpipe->setup = sp_draw_render_stage(softpipe);
    if (!softpipe->setup)
       goto fail;
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index 2b9a2a8ee52..790143aecc9 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -37,6 +37,7 @@
 #include "draw/draw_vertex.h"
 
 #include "sp_quad.h"
+#include "sp_tex_sample.h"
 
 
 /**
@@ -139,6 +140,12 @@ struct softpipe_context {
       struct quad_stage *first; /**< points to one of the above stages */
    } quad[SP_NUM_QUAD_THREADS];
 
+   /** TGSI exec things */
+   struct {
+      struct sp_shader_sampler samplers[PIPE_MAX_SAMPLERS];
+      struct sp_shader_sampler *samplers_list[PIPE_MAX_SAMPLERS];
+   } tgsi;
+
    /** The primitive drawing context */
    struct draw_context *draw;
    struct draw_stage *setup;
diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c
index 701ee4c72f2..453b0373f0f 100644
--- a/src/gallium/drivers/softpipe/sp_fs_exec.c
+++ b/src/gallium/drivers/softpipe/sp_fs_exec.c
@@ -39,11 +39,19 @@
 #include "tgsi/tgsi_exec.h"
 #include "tgsi/tgsi_parse.h"
 
-struct sp_exec_fragment_shader {
+struct sp_exec_fragment_shader
+{
    struct sp_fragment_shader base;
 };
 
 
+/** cast wrapper */
+static INLINE struct sp_exec_fragment_shader *
+sp_exec_fragment_shader(const struct sp_fragment_shader *base)
+{
+   return (struct sp_exec_fragment_shader *) base;
+}
+
 
 /**
  * Compute quad X,Y,Z,W for the four fragments in a quad.
@@ -84,12 +92,18 @@ sp_setup_pos_vector(const struct tgsi_interp_coef *coef,
 static void
 exec_prepare( const struct sp_fragment_shader *base,
 	      struct tgsi_exec_machine *machine,
-	      struct tgsi_sampler *samplers )
+	      struct tgsi_sampler **samplers )
 {
-   tgsi_exec_machine_bind_shader( machine,
-				  base->shader.tokens,
-				  PIPE_MAX_SAMPLERS,
-				  samplers );
+   /*
+    * Bind tokens/shader to the interpreter's machine state.
+    * Avoid redundant binding.
+    */
+   if (machine->Tokens != base->shader.tokens) {
+      tgsi_exec_machine_bind_shader( machine,
+                                     base->shader.tokens,
+                                     PIPE_MAX_SAMPLERS,
+                                     samplers );
+   }
 }
 
 
diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c
index 50eb2c07bcb..9a273c87643 100644
--- a/src/gallium/drivers/softpipe/sp_fs_sse.c
+++ b/src/gallium/drivers/softpipe/sp_fs_sse.c
@@ -40,7 +40,7 @@
 #include "tgsi/tgsi_sse2.h"
 
 
-#ifdef PIPE_ARCH_X86
+#if defined(PIPE_ARCH_X86)
 
 #include "rtasm/rtasm_x86sse.h"
 
@@ -69,7 +69,7 @@ struct sp_sse_fragment_shader {
 static void
 fs_sse_prepare( const struct sp_fragment_shader *base,
 		struct tgsi_exec_machine *machine,
-		struct tgsi_sampler *samplers )
+		struct tgsi_sampler **samplers )
 {
 }
 
diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
index 425e13cd283..9cd5784e5bc 100644
--- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c
+++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c
@@ -26,10 +26,10 @@
  **************************************************************************/
 
 /**
- * Post-transform vertex buffering.  This is an optional part of the
- * softpipe rendering pipeline.
- * Probably not desired in general, but useful for testing/debuggin.
- * Enabled/Disabled with SP_VBUF env var.
+ * Interface between 'draw' module's output and the softpipe rasterizer/setup
+ * code.  When the 'draw' module has finished filling a vertex buffer, the
+ * draw_arrays() functions below will be called.  Loop over the vertices and
+ * call the point/line/tri setup functions.
  *
  * Authors
  *  Brian Paul
@@ -131,21 +131,23 @@ static INLINE cptrf4 get_vert( const void *vertex_buffer,
 }
 
 
+/**
+ * draw elements / indexed primitives
+ */
 static void
 sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
    struct softpipe_context *softpipe = cvbr->softpipe;
-   unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float);
-   unsigned i;
+   const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float);
    const void *vertex_buffer = cvbr->vertex_buffer;
+   unsigned i;
 
    /* XXX: break this dependency - make setup_context live under
     * softpipe, rename the old "setup" draw stage to something else.
     */
    struct draw_stage *setup = softpipe->setup;
-   struct setup_context *setup_ctx = sp_draw_setup_context(softpipe->setup);
-
+   struct setup_context *setup_ctx = sp_draw_setup_context(setup);
 
    switch (cvbr->prim) {
    case PIPE_PRIM_POINTS:
@@ -258,13 +260,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr)
 {
    struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr);
    struct softpipe_context *softpipe = cvbr->softpipe;
-   struct draw_stage *setup = softpipe->setup;
-   const void *vertex_buffer = NULL;
    const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float);
+   const void *vertex_buffer =
+      (void *) get_vert(cvbr->vertex_buffer, start, stride);
    unsigned i;
-   struct setup_context *setup_ctx = sp_draw_setup_context(setup);
 
-   vertex_buffer = (void *)get_vert(cvbr->vertex_buffer, start, stride);
+   /* XXX: break this dependency - make setup_context live under
+    * softpipe, rename the old "setup" draw stage to something else.
+    */
+   struct draw_stage *setup = softpipe->setup;
+   struct setup_context *setup_ctx = sp_draw_setup_context(setup);
 
    switch (cvbr->prim) {
    case PIPE_PRIM_POINTS:
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 1f0cb3e0355..963a2b44f55 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -50,8 +50,9 @@
 
 struct quad_shade_stage
 {
-   struct quad_stage stage;
-   struct tgsi_sampler samplers[PIPE_MAX_SAMPLERS];
+   struct quad_stage stage;  /**< base class */
+   struct sp_shader_sampler samplers[PIPE_MAX_SAMPLERS];
+   struct sp_shader_sampler *samplers_list[PIPE_MAX_SAMPLERS];
    struct tgsi_exec_machine machine;
    struct tgsi_exec_vector *inputs, *outputs;
 };
@@ -147,18 +148,10 @@ static void shade_begin(struct quad_stage *qs)
 {
    struct quad_shade_stage *qss = quad_shade_stage(qs);
    struct softpipe_context *softpipe = qs->softpipe;
-   unsigned i;
-   unsigned num = MAX2(softpipe->num_textures, softpipe->num_samplers);
-
-   /* set TGSI sampler state that varies */
-   for (i = 0; i < num; i++) {
-      qss->samplers[i].state = softpipe->sampler[i];
-      qss->samplers[i].texture = softpipe->texture[i];
-   }
 
    softpipe->fs->prepare( softpipe->fs, 
 			  &qss->machine,
-			  qss->samplers );
+			  (struct tgsi_sampler **) qss->samplers_list );
 
    qs->next->begin(qs->next);
 }
@@ -191,12 +184,14 @@ struct quad_stage *sp_quad_shade_stage( struct softpipe_context *softpipe )
    qss->stage.run = shade_quad;
    qss->stage.destroy = shade_destroy;
 
-   /* set TGSI sampler state that's constant */
+   /* setup TGSI sampler state */
    for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
       assert(softpipe->tex_cache[i]);
-      qss->samplers[i].get_samples = sp_get_samples;
-      qss->samplers[i].pipe = &softpipe->pipe;
+      qss->samplers[i].base.get_samples = sp_get_samples;
+      qss->samplers[i].unit = i;
+      qss->samplers[i].sp = softpipe;
       qss->samplers[i].cache = softpipe->tex_cache[i];
+      qss->samplers_list[i] = &qss->samplers[i];
    }
 
    tgsi_exec_machine_init( &qss->machine );
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index 9644dbd168f..12f98c32f53 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -56,6 +56,8 @@ softpipe_get_param(struct pipe_screen *screen, int param)
    switch (param) {
    case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
       return 8;
+   case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS:
+      return 8;
    case PIPE_CAP_NPOT_TEXTURES:
       return 1;
    case PIPE_CAP_TWO_SIDED_STENCIL:
diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h
index 476ef3dc8fb..3eff41ffa5f 100644
--- a/src/gallium/drivers/softpipe/sp_state.h
+++ b/src/gallium/drivers/softpipe/sp_state.h
@@ -69,7 +69,7 @@ struct sp_fragment_shader {
 
    void (*prepare)( const struct sp_fragment_shader *shader,
 		    struct tgsi_exec_machine *machine,
-		    struct tgsi_sampler *samplers);
+		    struct tgsi_sampler **samplers);
 
    /* Run the shader - this interface will get cleaned up in the
     * future:
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c
index 49250ec084c..631c60966c2 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.c
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.c
@@ -35,6 +35,7 @@
 #include "sp_context.h"
 #include "sp_headers.h"
 #include "sp_surface.h"
+#include "sp_texture.h"
 #include "sp_tex_sample.h"
 #include "sp_tile_cache.h"
 #include "pipe/p_context.h"
@@ -57,7 +58,11 @@
 /**
  * Linear interpolation macro
  */
-#define LERP(T, A, B)  ( (A) + (T) * ((B) - (A)) )
+static INLINE float
+lerp(float a, float v0, float v1)
+{
+   return v0 + a * (v1 - v0);
+}
 
 
 /**
@@ -72,13 +77,28 @@ static INLINE float
 lerp_2d(float a, float b,
         float v00, float v10, float v01, float v11)
 {
-   const float temp0 = LERP(a, v00, v10);
-   const float temp1 = LERP(a, v01, v11);
-   return LERP(b, temp0, temp1);
+   const float temp0 = lerp(a, v00, v10);
+   const float temp1 = lerp(a, v01, v11);
+   return lerp(b, temp0, temp1);
 }
 
 
 /**
+ * As above, but 3D interpolation of 8 values.
+ */
+static INLINE float
+lerp_3d(float a, float b, float c,
+        float v000, float v100, float v010, float v110,
+        float v001, float v101, float v011, float v111)
+{
+   const float temp0 = lerp_2d(a, b, v000, v100, v010, v110);
+   const float temp1 = lerp_2d(a, b, v001, v101, v011, v111);
+   return lerp(c, temp0, temp1);
+}
+
+
+
+/**
  * If A is a signed integer, A % B doesn't give the right value for A < 0
  * (in terms of texture repeat).  Just casting to unsigned fixes that.
  */
@@ -86,250 +106,275 @@ lerp_2d(float a, float b,
 
 
 /**
- * Apply texture coord wrapping mode and return integer texture index.
+ * Apply texture coord wrapping mode and return integer texture indexes
+ * for a vector of four texcoords (S or T or P).
  * \param wrapMode  PIPE_TEX_WRAP_x
- * \param s  the texcoord
+ * \param s  the incoming texcoords
  * \param size  the texture image size
+ * \param icoord  returns the integer texcoords
  * \return  integer texture index
  */
-static INLINE int
-nearest_texcoord(unsigned wrapMode, float s, unsigned size)
+static INLINE void
+nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
+                   int icoord[4])
 {
-   int i;
+   uint ch;
    switch (wrapMode) {
    case PIPE_TEX_WRAP_REPEAT:
       /* s limited to [0,1) */
       /* i limited to [0,size-1] */
-      i = util_ifloor(s * size);
-      i = REMAINDER(i, size);
-      return i;
+      for (ch = 0; ch < 4; ch++) {
+         int i = util_ifloor(s[ch] * size);
+         icoord[ch] = REMAINDER(i, size);
+      }
+      return;
    case PIPE_TEX_WRAP_CLAMP:
       /* s limited to [0,1] */
       /* i limited to [0,size-1] */
-      if (s <= 0.0F)
-         i = 0;
-      else if (s >= 1.0F)
-         i = size - 1;
-      else
-         i = util_ifloor(s * size);
-      return i;
+      for (ch = 0; ch < 4; ch++) {
+         if (s[ch] <= 0.0F)
+            icoord[ch] = 0;
+         else if (s[ch] >= 1.0F)
+            icoord[ch] = size - 1;
+         else
+            icoord[ch] = util_ifloor(s[ch] * size);
+      }
+      return;
    case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
       {
          /* s limited to [min,max] */
          /* i limited to [0, size-1] */
          const float min = 1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         if (s < min)
-            i = 0;
-         else if (s > max)
-            i = size - 1;
-         else
-            i = util_ifloor(s * size);
+         for (ch = 0; ch < 4; ch++) {
+            if (s[ch] < min)
+               icoord[ch] = 0;
+            else if (s[ch] > max)
+               icoord[ch] = size - 1;
+            else
+               icoord[ch] = util_ifloor(s[ch] * size);
+         }
       }
-      return i;
+      return;
    case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
       {
          /* s limited to [min,max] */
          /* i limited to [-1, size] */
          const float min = -1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         if (s <= min)
-            i = -1;
-         else if (s >= max)
-            i = size;
-         else
-            i = util_ifloor(s * size);
+         for (ch = 0; ch < 4; ch++) {
+            if (s[ch] <= min)
+               icoord[ch] = -1;
+            else if (s[ch] >= max)
+               icoord[ch] = size;
+            else
+               icoord[ch] = util_ifloor(s[ch] * size);
+         }
       }
-      return i;
+      return;
    case PIPE_TEX_WRAP_MIRROR_REPEAT:
       {
          const float min = 1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         const int flr = util_ifloor(s);
-         float u;
-         if (flr & 1)
-            u = 1.0F - (s - (float) flr);
-         else
-            u = s - (float) flr;
-         if (u < min)
-            i = 0;
-         else if (u > max)
-            i = size - 1;
-         else
-            i = util_ifloor(u * size);
+         for (ch = 0; ch < 4; ch++) {
+            const int flr = util_ifloor(s[ch]);
+            float u;
+            if (flr & 1)
+               u = 1.0F - (s[ch] - (float) flr);
+            else
+               u = s[ch] - (float) flr;
+            if (u < min)
+               icoord[ch] = 0;
+            else if (u > max)
+               icoord[ch] = size - 1;
+            else
+               icoord[ch] = util_ifloor(u * size);
+         }
       }
-      return i;
+      return;
    case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      {
+      for (ch = 0; ch < 4; ch++) {
          /* s limited to [0,1] */
          /* i limited to [0,size-1] */
-         const float u = fabsf(s);
+         const float u = fabsf(s[ch]);
          if (u <= 0.0F)
-            i = 0;
+            icoord[ch] = 0;
          else if (u >= 1.0F)
-            i = size - 1;
+            icoord[ch] = size - 1;
          else
-            i = util_ifloor(u * size);
+            icoord[ch] = util_ifloor(u * size);
       }
-      return i;
+      return;
    case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
       {
          /* s limited to [min,max] */
          /* i limited to [0, size-1] */
          const float min = 1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         const float u = fabsf(s);
-         if (u < min)
-            i = 0;
-         else if (u > max)
-            i = size - 1;
-         else
-            i = util_ifloor(u * size);
+         for (ch = 0; ch < 4; ch++) {
+            const float u = fabsf(s[ch]);
+            if (u < min)
+               icoord[ch] = 0;
+            else if (u > max)
+               icoord[ch] = size - 1;
+            else
+               icoord[ch] = util_ifloor(u * size);
+         }
       }
-      return i;
+      return;
    case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
       {
          /* s limited to [min,max] */
          /* i limited to [0, size-1] */
          const float min = -1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         const float u = fabsf(s);
-         if (u < min)
-            i = -1;
-         else if (u > max)
-            i = size;
-         else
-            i = util_ifloor(u * size);
+         for (ch = 0; ch < 4; ch++) {
+            const float u = fabsf(s[ch]);
+            if (u < min)
+               icoord[ch] = -1;
+            else if (u > max)
+               icoord[ch] = size;
+            else
+               icoord[ch] = util_ifloor(u * size);
+         }
       }
-      return i;
+      return;
    default:
       assert(0);
-      return 0;
    }
 }
 
 
 /**
- * Used to compute texel locations for linear sampling.
+ * Used to compute texel locations for linear sampling for four texcoords.
  * \param wrapMode  PIPE_TEX_WRAP_x
- * \param s  the texcoord
+ * \param s  the texcoords
  * \param size  the texture image size
- * \param i0  returns first texture index
- * \param i1  returns second texture index (usually *i0 + 1)
- * \param a  returns blend factor/weight between texture indexes
+ * \param icoord0  returns first texture indexes
+ * \param icoord1  returns second texture indexes (usually icoord0 + 1)
+ * \param w  returns blend factor/weight between texture indexes
+ * \param icoord  returns the computed integer texture coords
  */
 static INLINE void
-linear_texcoord(unsigned wrapMode, float s, unsigned size,
-                int *i0, int *i1, float *a)
+linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size,
+                  int icoord0[4], int icoord1[4], float w[4])
 {
-   float u;
+   uint ch;
+
    switch (wrapMode) {
    case PIPE_TEX_WRAP_REPEAT:
-      u = s * size - 0.5F;
-      *i0 = REMAINDER(util_ifloor(u), size);
-      *i1 = REMAINDER(*i0 + 1, size);
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         float u = s[ch] * size - 0.5F;
+         icoord0[ch] = REMAINDER(util_ifloor(u), size);
+         icoord1[ch] = REMAINDER(icoord0[ch] + 1, size);
+         w[ch] = FRAC(u);
+      }
+      break;;
    case PIPE_TEX_WRAP_CLAMP:
-      if (s <= 0.0F)
-         u = 0.0F;
-      else if (s >= 1.0F)
-         u = (float) size;
-      else
-         u = s * size;
-      u -= 0.5F;
-      *i0 = util_ifloor(u);
-      *i1 = *i0 + 1;
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         float u = CLAMP(s[ch], 0.0F, 1.0F);
+         u = u * size - 0.5f;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         w[ch] = FRAC(u);
+      }
+      break;;
    case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
-      if (s <= 0.0F)
-         u = 0.0F;
-      else if (s >= 1.0F)
-         u = (float) size;
-      else
-         u = s * size;
-      u -= 0.5F;
-      *i0 = util_ifloor(u);
-      *i1 = *i0 + 1;
-      if (*i0 < 0)
-         *i0 = 0;
-      if (*i1 >= (int) size)
-         *i1 = size - 1;
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         float u = CLAMP(s[ch], 0.0F, 1.0F);
+         u = u * size - 0.5f;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         if (icoord0[ch] < 0)
+            icoord0[ch] = 0;
+         if (icoord1[ch] >= (int) size)
+            icoord1[ch] = size - 1;
+         w[ch] = FRAC(u);
+      }
+      break;;
    case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
       {
          const float min = -1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         if (s <= min)
-            u = min * size;
-         else if (s >= max)
-            u = max * size;
-         else
-            u = s * size;
-         u -= 0.5F;
-         *i0 = util_ifloor(u);
-         *i1 = *i0 + 1;
+         for (ch = 0; ch < 4; ch++) {
+            float u = CLAMP(s[ch], min, max);
+            u = u * size - 0.5f;
+            icoord0[ch] = util_ifloor(u);
+            icoord1[ch] = icoord0[ch] + 1;
+            w[ch] = FRAC(u);
+         }
       }
-      break;
+      break;;
    case PIPE_TEX_WRAP_MIRROR_REPEAT:
-      {
-         const int flr = util_ifloor(s);
+      for (ch = 0; ch < 4; ch++) {
+         const int flr = util_ifloor(s[ch]);
+         float u;
          if (flr & 1)
-            u = 1.0F - (s - (float) flr);
+            u = 1.0F - (s[ch] - (float) flr);
          else
-            u = s - (float) flr;
-         u = (u * size) - 0.5F;
-         *i0 = util_ifloor(u);
-         *i1 = *i0 + 1;
-         if (*i0 < 0)
-            *i0 = 0;
-         if (*i1 >= (int) size)
-            *i1 = size - 1;
+            u = s[ch] - (float) flr;
+         u = u * size - 0.5F;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         if (icoord0[ch] < 0)
+            icoord0[ch] = 0;
+         if (icoord1[ch] >= (int) size)
+            icoord1[ch] = size - 1;
+         w[ch] = FRAC(u);
       }
-      break;
+      break;;
    case PIPE_TEX_WRAP_MIRROR_CLAMP:
-      u = fabsf(s);
-      if (u >= 1.0F)
-         u = (float) size;
-      else
-         u *= size;
-      u -= 0.5F;
-      *i0 = util_ifloor(u);
-      *i1 = *i0 + 1;
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         float u = fabsf(s[ch]);
+         if (u >= 1.0F)
+            u = (float) size;
+         else
+            u *= size;
+         u -= 0.5F;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         w[ch] = FRAC(u);
+      }
+      break;;
    case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE:
-      u = fabsf(s);
-      if (u >= 1.0F)
-         u = (float) size;
-      else
-         u *= size;
-      u -= 0.5F;
-      *i0 = util_ifloor(u);
-      *i1 = *i0 + 1;
-      if (*i0 < 0)
-         *i0 = 0;
-      if (*i1 >= (int) size)
-         *i1 = size - 1;
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         float u = fabsf(s[ch]);
+         if (u >= 1.0F)
+            u = (float) size;
+         else
+            u *= size;
+         u -= 0.5F;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         if (icoord0[ch] < 0)
+            icoord0[ch] = 0;
+         if (icoord1[ch] >= (int) size)
+            icoord1[ch] = size - 1;
+         w[ch] = FRAC(u);
+      }
+      break;;
    case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER:
       {
          const float min = -1.0F / (2.0F * size);
          const float max = 1.0F - min;
-         u = fabsf(s);
-         if (u <= min)
-            u = min * size;
-         else if (u >= max)
-            u = max * size;
-         else
-            u *= size;
-         u -= 0.5F;
-         *i0 = util_ifloor(u);
-         *i1 = *i0 + 1;
+         for (ch = 0; ch < 4; ch++) {
+            float u = fabsf(s[ch]);
+            if (u <= min)
+               u = min * size;
+            else if (u >= max)
+               u = max * size;
+            else
+               u *= size;
+            u -= 0.5F;
+            icoord0[ch] = util_ifloor(u);
+            icoord1[ch] = icoord0[ch] + 1;
+            w[ch] = FRAC(u);
+         }
       }
-      break;
+      break;;
    default:
       assert(0);
    }
-   *a = FRAC(u);
 }
 
 
@@ -337,21 +382,27 @@ linear_texcoord(unsigned wrapMode, float s, unsigned size,
  * For RECT textures / unnormalized texcoords
  * Only a subset of wrap modes supported.
  */
-static INLINE int
-nearest_texcoord_unnorm(unsigned wrapMode, float s, unsigned size)
+static INLINE void
+nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
+                          int icoord[4])
 {
-   int i;
+   uint ch;
    switch (wrapMode) {
    case PIPE_TEX_WRAP_CLAMP:
-      i = util_ifloor(s);
-      return CLAMP(i, 0, (int) size-1);
+      for (ch = 0; ch < 4; ch++) {
+         int i = util_ifloor(s[ch]);
+         icoord[ch]= CLAMP(i, 0, (int) size-1);
+      }
+      return;
    case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
       /* fall-through */
    case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      return util_ifloor( CLAMP(s, 0.5F, (float) size - 0.5F) );
+      for (ch = 0; ch < 4; ch++) {
+         icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) );
+      }
+      return;
    default:
       assert(0);
-      return 0;
    }
 }
 
@@ -361,30 +412,36 @@ nearest_texcoord_unnorm(unsigned wrapMode, float s, unsigned size)
  * Only a subset of wrap modes supported.
  */
 static INLINE void
-linear_texcoord_unnorm(unsigned wrapMode, float s, unsigned size,
-                       int *i0, int *i1, float *a)
+linear_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size,
+                         int icoord0[4], int icoord1[4], float w[4])
 {
+   uint ch;
    switch (wrapMode) {
    case PIPE_TEX_WRAP_CLAMP:
-      /* Not exactly what the spec says, but it matches NVIDIA output */
-      s = CLAMP(s - 0.5F, 0.0f, (float) size - 1.0f);
-      *i0 = util_ifloor(s);
-      *i1 = *i0 + 1;
-      break;
+      for (ch = 0; ch < 4; ch++) {
+         /* Not exactly what the spec says, but it matches NVIDIA output */
+         float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f);
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         w[ch] = FRAC(u);
+      }
+      return;
    case PIPE_TEX_WRAP_CLAMP_TO_EDGE:
       /* fall-through */
    case PIPE_TEX_WRAP_CLAMP_TO_BORDER:
-      s = CLAMP(s, 0.5F, (float) size - 0.5F);
-      s -= 0.5F;
-      *i0 = util_ifloor(s);
-      *i1 = *i0 + 1;
-      if (*i1 > (int) size - 1)
-         *i1 = size - 1;
+      for (ch = 0; ch < 4; ch++) {
+         float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F);
+         u -= 0.5F;
+         icoord0[ch] = util_ifloor(u);
+         icoord1[ch] = icoord0[ch] + 1;
+         if (icoord1[ch] > (int) size - 1)
+            icoord1[ch] = size - 1;
+         w[ch] = FRAC(u);
+      }
       break;
    default:
       assert(0);
    }
-   *a = FRAC(s);
 }
 
 
@@ -463,7 +520,8 @@ choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
  * This is only done for fragment shaders, not vertex shaders.
  */
 static float
-compute_lambda(struct tgsi_sampler *sampler,
+compute_lambda(const struct pipe_texture *tex,
+               const struct pipe_sampler_state *sampler,
                const float s[QUAD_SIZE],
                const float t[QUAD_SIZE],
                const float p[QUAD_SIZE],
@@ -471,7 +529,7 @@ compute_lambda(struct tgsi_sampler *sampler,
 {
    float rho, lambda;
 
-   assert(sampler->state->normalized_coords);
+   assert(sampler->normalized_coords);
 
    assert(s);
    {
@@ -479,7 +537,7 @@ compute_lambda(struct tgsi_sampler *sampler,
       float dsdy = s[QUAD_TOP_LEFT]     - s[QUAD_BOTTOM_LEFT];
       dsdx = fabsf(dsdx);
       dsdy = fabsf(dsdy);
-      rho = MAX2(dsdx, dsdy) * sampler->texture->width[0];
+      rho = MAX2(dsdx, dsdy) * tex->width[0];
    }
    if (t) {
       float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT];
@@ -487,7 +545,7 @@ compute_lambda(struct tgsi_sampler *sampler,
       float max;
       dtdx = fabsf(dtdx);
       dtdy = fabsf(dtdy);
-      max = MAX2(dtdx, dtdy) * sampler->texture->height[0];
+      max = MAX2(dtdx, dtdy) * tex->height[0];
       rho = MAX2(rho, max);
    }
    if (p) {
@@ -496,13 +554,13 @@ compute_lambda(struct tgsi_sampler *sampler,
       float max;
       dpdx = fabsf(dpdx);
       dpdy = fabsf(dpdy);
-      max = MAX2(dpdx, dpdy) * sampler->texture->depth[0];
+      max = MAX2(dpdx, dpdy) * tex->depth[0];
       rho = MAX2(rho, max);
    }
 
    lambda = util_fast_log2(rho);
-   lambda += lodbias + sampler->state->lod_bias;
-   lambda = CLAMP(lambda, sampler->state->min_lod, sampler->state->max_lod);
+   lambda += lodbias + sampler->lod_bias;
+   lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod);
 
    return lambda;
 }
@@ -514,9 +572,14 @@ compute_lambda(struct tgsi_sampler *sampler,
  * 2. Determine if we're minifying or magnifying
  * 3. If minifying, choose mipmap levels
  * 4. Return image filter to use within mipmap images
+ * \param level0  Returns first mipmap level to sample from
+ * \param level1  Returns second mipmap level to sample from
+ * \param levelBlend  Returns blend factor between levels, in [0,1]
+ * \param imgFilter  Returns either the min or mag filter, depending on lambda
  */
 static void
-choose_mipmap_levels(struct tgsi_sampler *sampler,
+choose_mipmap_levels(const struct pipe_texture *texture,
+                     const struct pipe_sampler_state *sampler,
                      const float s[QUAD_SIZE],
                      const float t[QUAD_SIZE],
                      const float p[QUAD_SIZE],
@@ -524,25 +587,25 @@ choose_mipmap_levels(struct tgsi_sampler *sampler,
                      unsigned *level0, unsigned *level1, float *levelBlend,
                      unsigned *imgFilter)
 {
-   if (sampler->state->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
+   if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) {
       /* no mipmap selection needed */
-      *level0 = *level1 = CLAMP((int) sampler->state->min_lod,
-                                0, (int) sampler->texture->last_level);
+      *level0 = *level1 = CLAMP((int) sampler->min_lod,
+                                0, (int) texture->last_level);
 
-      if (sampler->state->min_img_filter != sampler->state->mag_img_filter) {
+      if (sampler->min_img_filter != sampler->mag_img_filter) {
          /* non-mipmapped texture, but still need to determine if doing
           * minification or magnification.
           */
-         float lambda = compute_lambda(sampler, s, t, p, lodbias);
+         float lambda = compute_lambda(texture, sampler, s, t, p, lodbias);
          if (lambda <= 0.0) {
-            *imgFilter = sampler->state->mag_img_filter;
+            *imgFilter = sampler->mag_img_filter;
          }
          else {
-            *imgFilter = sampler->state->min_img_filter;
+            *imgFilter = sampler->min_img_filter;
          }
       }
       else {
-         *imgFilter = sampler->state->mag_img_filter;
+         *imgFilter = sampler->mag_img_filter;
       }
    }
    else {
@@ -550,32 +613,32 @@ choose_mipmap_levels(struct tgsi_sampler *sampler,
 
       if (1)
          /* fragment shader */
-         lambda = compute_lambda(sampler, s, t, p, lodbias);
+         lambda = compute_lambda(texture, sampler, s, t, p, lodbias);
       else
          /* vertex shader */
          lambda = lodbias; /* not really a bias, but absolute LOD */
 
       if (lambda <= 0.0) { /* XXX threshold depends on the filter */
          /* magnifying */
-         *imgFilter = sampler->state->mag_img_filter;
+         *imgFilter = sampler->mag_img_filter;
          *level0 = *level1 = 0;
       }
       else {
          /* minifying */
-         *imgFilter = sampler->state->min_img_filter;
+         *imgFilter = sampler->min_img_filter;
 
          /* choose mipmap level(s) and compute the blend factor between them */
-         if (sampler->state->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) {
+         if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) {
             /* Nearest mipmap level */
             const int lvl = (int) (lambda + 0.5);
             *level0 =
-            *level1 = CLAMP(lvl, 0, (int) sampler->texture->last_level);
+            *level1 = CLAMP(lvl, 0, (int) texture->last_level);
          }
          else {
             /* Linear interpolation between mipmap levels */
             const int lvl = (int) lambda;
-            *level0 = CLAMP(lvl,     0, (int) sampler->texture->last_level);
-            *level1 = CLAMP(lvl + 1, 0, (int) sampler->texture->last_level);
+            *level0 = CLAMP(lvl,     0, (int) texture->last_level);
+            *level1 = CLAMP(lvl + 1, 0, (int) texture->last_level);
             *levelBlend = FRAC(lambda);  /* blending weight between levels */
          }
       }
@@ -598,23 +661,29 @@ choose_mipmap_levels(struct tgsi_sampler *sampler,
  * sp_get_cached_tile_tex() function.  Also, get 4 texels instead of 1...
  */
 static void
-get_texel(struct tgsi_sampler *sampler,
+get_texel(const struct tgsi_sampler *tgsi_sampler,
           unsigned face, unsigned level, int x, int y, int z,
           float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j)
 {
-   if (x < 0 || x >= (int) sampler->texture->width[level] ||
-       y < 0 || y >= (int) sampler->texture->height[level] ||
-       z < 0 || z >= (int) sampler->texture->depth[level]) {
-      rgba[0][j] = sampler->state->border_color[0];
-      rgba[1][j] = sampler->state->border_color[1];
-      rgba[2][j] = sampler->state->border_color[2];
-      rgba[3][j] = sampler->state->border_color[3];
+   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
+   struct softpipe_context *sp = samp->sp;
+   const uint unit = samp->unit;
+   const struct pipe_texture *texture = sp->texture[unit];
+   const struct pipe_sampler_state *sampler = sp->sampler[unit];
+
+   if (x < 0 || x >= (int) texture->width[level] ||
+       y < 0 || y >= (int) texture->height[level] ||
+       z < 0 || z >= (int) texture->depth[level]) {
+      rgba[0][j] = sampler->border_color[0];
+      rgba[1][j] = sampler->border_color[1];
+      rgba[2][j] = sampler->border_color[2];
+      rgba[3][j] = sampler->border_color[3];
    }
    else {
       const int tx = x % TILE_SIZE;
       const int ty = y % TILE_SIZE;
       const struct softpipe_cached_tile *tile
-         = sp_get_cached_tile_tex(sampler->pipe, sampler->cache,
+         = sp_get_cached_tile_tex(sp, samp->cache,
                                   x, y, z, face, level);
       rgba[0][j] = tile->data.color[ty][tx][0];
       rgba[1][j] = tile->data.color[ty][tx][1];
@@ -624,7 +693,7 @@ get_texel(struct tgsi_sampler *sampler,
       {
          debug_printf("Get texel %f %f %f %f from %s\n",
                       rgba[0][j], rgba[1][j], rgba[2][j], rgba[3][j],
-                      pf_name(sampler->texture->format));
+                      pf_name(texture->format));
       }
    }
 }
@@ -682,7 +751,7 @@ shadow_compare(uint compare_func,
  * Could probably extend for 3D...
  */
 static void
-sp_get_samples_2d_common(struct tgsi_sampler *sampler,
+sp_get_samples_2d_common(const struct tgsi_sampler *tgsi_sampler,
                          const float s[QUAD_SIZE],
                          const float t[QUAD_SIZE],
                          const float p[QUAD_SIZE],
@@ -690,95 +759,115 @@ sp_get_samples_2d_common(struct tgsi_sampler *sampler,
                          float rgba[NUM_CHANNELS][QUAD_SIZE],
                          const unsigned faces[4])
 {
-   const uint compare_func = sampler->state->compare_func;
+   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
+   const struct softpipe_context *sp = samp->sp;
+   const uint unit = samp->unit;
+   const struct pipe_texture *texture = sp->texture[unit];
+   const struct pipe_sampler_state *sampler = sp->sampler[unit];
+   const uint compare_func = sampler->compare_func;
    unsigned level0, level1, j, imgFilter;
    int width, height;
    float levelBlend;
 
-   choose_mipmap_levels(sampler, s, t, p, lodbias,
+   choose_mipmap_levels(texture, sampler, s, t, p, lodbias,
                         &level0, &level1, &levelBlend, &imgFilter);
 
-   assert(sampler->state->normalized_coords);
+   assert(sampler->normalized_coords);
 
-   width = sampler->texture->width[level0];
-   height = sampler->texture->height[level0];
+   width = texture->width[level0];
+   height = texture->height[level0];
 
    assert(width > 0);
 
    switch (imgFilter) {
    case PIPE_TEX_FILTER_NEAREST:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = nearest_texcoord(sampler->state->wrap_s, s[j], width);
-         int y = nearest_texcoord(sampler->state->wrap_t, t[j], height);
-         get_texel(sampler, faces[j], level0, x, y, 0, rgba, j);
-         if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-            shadow_compare(compare_func, rgba, p, j);
-         }
-
-         if (level0 != level1) {
-            /* get texels from second mipmap level and blend */
-            float rgba2[4][4];
-            unsigned c;
-            x = x / 2;
-            y = y / 2;
-            get_texel(sampler, faces[j], level1, x, y, 0, rgba2, j);
-            if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
-               shadow_compare(compare_func, rgba2, p, j);
+      {
+         int x[4], y[4];
+         nearest_texcoord_4(sampler->wrap_s, s, width, x);
+         nearest_texcoord_4(sampler->wrap_t, t, height, y);
+
+         for (j = 0; j < QUAD_SIZE; j++) {
+            get_texel(tgsi_sampler, faces[j], level0, x[j], y[j], 0, rgba, j);
+            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+               shadow_compare(compare_func, rgba, p, j);
             }
 
-            for (c = 0; c < NUM_CHANNELS; c++) {
-               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
+            if (level0 != level1) {
+               /* get texels from second mipmap level and blend */
+               float rgba2[4][4];
+               unsigned c;
+               x[j] /= 2;
+               y[j] /= 2;
+               get_texel(tgsi_sampler, faces[j], level1, x[j], y[j], 0,
+                         rgba2, j);
+               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
+                  shadow_compare(compare_func, rgba2, p, j);
+               }
+
+               for (c = 0; c < NUM_CHANNELS; c++) {
+                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
+               }
             }
          }
       }
       break;
    case PIPE_TEX_FILTER_LINEAR:
    case PIPE_TEX_FILTER_ANISO:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         float tx[4][4], a, b;
-         int x0, y0, x1, y1, c;
-         linear_texcoord(sampler->state->wrap_s, s[j], width,  &x0, &x1, &a);
-         linear_texcoord(sampler->state->wrap_t, t[j], height, &y0, &y1, &b);
-         get_texel(sampler, faces[j], level0, x0, y0, 0, tx, 0);
-         get_texel(sampler, faces[j], level0, x1, y0, 0, tx, 1);
-         get_texel(sampler, faces[j], level0, x0, y1, 0, tx, 2);
-         get_texel(sampler, faces[j], level0, x1, y1, 0, tx, 3);
-         if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-            shadow_compare(compare_func, tx, p, 0);
-            shadow_compare(compare_func, tx, p, 1);
-            shadow_compare(compare_func, tx, p, 2);
-            shadow_compare(compare_func, tx, p, 3);
-         }
-
-         for (c = 0; c < 4; c++) {
-            rgba[c][j] = lerp_2d(a, b, tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
-         }
-
-         if (level0 != level1) {
-            /* get texels from second mipmap level and blend */
-            float rgba2[4][4];
-            x0 = x0 / 2;
-            y0 = y0 / 2;
-            x1 = x1 / 2;
-            y1 = y1 / 2;
-            get_texel(sampler, faces[j], level1, x0, y0, 0, tx, 0);
-            get_texel(sampler, faces[j], level1, x1, y0, 0, tx, 1);
-            get_texel(sampler, faces[j], level1, x0, y1, 0, tx, 2);
-            get_texel(sampler, faces[j], level1, x1, y1, 0, tx, 3);
-            if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
+      {
+         int x0[4], y0[4], x1[4], y1[4];
+         float xw[4], yw[4]; /* weights */
+
+         linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw);
+         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
+
+         for (j = 0; j < QUAD_SIZE; j++) {
+            float tx[4][4]; /* texels */
+            int c;
+            get_texel(tgsi_sampler, faces[j], level0, x0[j], y0[j], 0, tx, 0);
+            get_texel(tgsi_sampler, faces[j], level0, x1[j], y0[j], 0, tx, 1);
+            get_texel(tgsi_sampler, faces[j], level0, x0[j], y1[j], 0, tx, 2);
+            get_texel(tgsi_sampler, faces[j], level0, x1[j], y1[j], 0, tx, 3);
+            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
                shadow_compare(compare_func, tx, p, 0);
                shadow_compare(compare_func, tx, p, 1);
                shadow_compare(compare_func, tx, p, 2);
                shadow_compare(compare_func, tx, p, 3);
             }
 
+            /* interpolate R, G, B, A */
             for (c = 0; c < 4; c++) {
-               rgba2[c][j] = lerp_2d(a, b,
-                                     tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
+               rgba[c][j] = lerp_2d(xw[j], yw[j],
+                                    tx[c][0], tx[c][1],
+                                    tx[c][2], tx[c][3]);
             }
 
-            for (c = 0; c < NUM_CHANNELS; c++) {
-               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
+            if (level0 != level1) {
+               /* get texels from second mipmap level and blend */
+               float rgba2[4][4];
+               x0[j] /= 2;
+               y0[j] /= 2;
+               x1[j] /= 2;
+               y1[j] /= 2;
+               get_texel(tgsi_sampler, faces[j], level1, x0[j], y0[j], 0, tx, 0);
+               get_texel(tgsi_sampler, faces[j], level1, x1[j], y0[j], 0, tx, 1);
+               get_texel(tgsi_sampler, faces[j], level1, x0[j], y1[j], 0, tx, 2);
+               get_texel(tgsi_sampler, faces[j], level1, x1[j], y1[j], 0, tx, 3);
+               if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){
+                  shadow_compare(compare_func, tx, p, 0);
+                  shadow_compare(compare_func, tx, p, 1);
+                  shadow_compare(compare_func, tx, p, 2);
+                  shadow_compare(compare_func, tx, p, 3);
+               }
+
+               /* interpolate R, G, B, A */
+               for (c = 0; c < 4; c++) {
+                  rgba2[c][j] = lerp_2d(xw[j], yw[j],
+                                        tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
+               }
+
+               for (c = 0; c < NUM_CHANNELS; c++) {
+                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
+               }
             }
          }
       }
@@ -789,8 +878,8 @@ sp_get_samples_2d_common(struct tgsi_sampler *sampler,
 }
 
 
-static void
-sp_get_samples_1d(struct tgsi_sampler *sampler,
+static INLINE void
+sp_get_samples_1d(const struct tgsi_sampler *sampler,
                   const float s[QUAD_SIZE],
                   const float t[QUAD_SIZE],
                   const float p[QUAD_SIZE],
@@ -803,8 +892,8 @@ sp_get_samples_1d(struct tgsi_sampler *sampler,
 }
 
 
-static void
-sp_get_samples_2d(struct tgsi_sampler *sampler,
+static INLINE void
+sp_get_samples_2d(const struct tgsi_sampler *sampler,
                   const float s[QUAD_SIZE],
                   const float t[QUAD_SIZE],
                   const float p[QUAD_SIZE],
@@ -816,28 +905,33 @@ sp_get_samples_2d(struct tgsi_sampler *sampler,
 }
 
 
-static void
-sp_get_samples_3d(struct tgsi_sampler *sampler,
+static INLINE void
+sp_get_samples_3d(const struct tgsi_sampler *tgsi_sampler,
                   const float s[QUAD_SIZE],
                   const float t[QUAD_SIZE],
                   const float p[QUAD_SIZE],
                   float lodbias,
                   float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
+   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
+   const struct softpipe_context *sp = samp->sp;
+   const uint unit = samp->unit;
+   const struct pipe_texture *texture = sp->texture[unit];
+   const struct pipe_sampler_state *sampler = sp->sampler[unit];
    /* get/map pipe_surfaces corresponding to 3D tex slices */
    unsigned level0, level1, j, imgFilter;
    int width, height, depth;
    float levelBlend;
    const uint face = 0;
 
-   choose_mipmap_levels(sampler, s, t, p, lodbias,
+   choose_mipmap_levels(texture, sampler, s, t, p, lodbias,
                         &level0, &level1, &levelBlend, &imgFilter);
 
-   assert(sampler->state->normalized_coords);
+   assert(sampler->normalized_coords);
 
-   width = sampler->texture->width[level0];
-   height = sampler->texture->height[level0];
-   depth = sampler->texture->depth[level0];
+   width = texture->width[level0];
+   height = texture->height[level0];
+   depth = texture->depth[level0];
 
    assert(width > 0);
    assert(height > 0);
@@ -845,89 +939,89 @@ sp_get_samples_3d(struct tgsi_sampler *sampler,
 
    switch (imgFilter) {
    case PIPE_TEX_FILTER_NEAREST:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = nearest_texcoord(sampler->state->wrap_s, s[j], width);
-         int y = nearest_texcoord(sampler->state->wrap_t, t[j], height);
-         int z = nearest_texcoord(sampler->state->wrap_r, p[j], depth);
-         get_texel(sampler, face, level0, x, y, z, rgba, j);
-
-         if (level0 != level1) {
-            /* get texels from second mipmap level and blend */
-            float rgba2[4][4];
-            unsigned c;
-            x /= 2;
-            y /= 2;
-            z /= 2;
-            get_texel(sampler, face, level1, x, y, z, rgba2, j);
-            for (c = 0; c < NUM_CHANNELS; c++) {
-               rgba[c][j] = LERP(levelBlend, rgba2[c][j], rgba[c][j]);
+      {
+         int x[4], y[4], z[4];
+         nearest_texcoord_4(sampler->wrap_s, s, width, x);
+         nearest_texcoord_4(sampler->wrap_t, t, height, y);
+         nearest_texcoord_4(sampler->wrap_r, p, depth, z);
+         for (j = 0; j < QUAD_SIZE; j++) {
+            get_texel(tgsi_sampler, face, level0, x[j], y[j], z[j], rgba, j);
+            if (level0 != level1) {
+               /* get texels from second mipmap level and blend */
+               float rgba2[4][4];
+               unsigned c;
+               x[j] /= 2;
+               y[j] /= 2;
+               z[j] /= 2;
+               get_texel(tgsi_sampler, face, level1, x[j], y[j], z[j], rgba2, j);
+               for (c = 0; c < NUM_CHANNELS; c++) {
+                  rgba[c][j] = lerp(levelBlend, rgba2[c][j], rgba[c][j]);
+               }
             }
          }
       }
       break;
    case PIPE_TEX_FILTER_LINEAR:
    case PIPE_TEX_FILTER_ANISO:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         float texel0[4][4], texel1[4][4];
-         float xw, yw, zw; /* interpolation weights */
-         int x0, x1, y0, y1, z0, z1, c;
-         linear_texcoord(sampler->state->wrap_s, s[j], width,  &x0, &x1, &xw);
-         linear_texcoord(sampler->state->wrap_t, t[j], height, &y0, &y1, &yw);
-         linear_texcoord(sampler->state->wrap_r, p[j], depth,  &z0, &z1, &zw);
-         get_texel(sampler, face, level0, x0, y0, z0, texel0, 0);
-         get_texel(sampler, face, level0, x1, y0, z0, texel0, 1);
-         get_texel(sampler, face, level0, x0, y1, z0, texel0, 2);
-         get_texel(sampler, face, level0, x1, y1, z0, texel0, 3);
-         get_texel(sampler, face, level0, x0, y0, z1, texel1, 0);
-         get_texel(sampler, face, level0, x1, y0, z1, texel1, 1);
-         get_texel(sampler, face, level0, x0, y1, z1, texel1, 2);
-         get_texel(sampler, face, level0, x1, y1, z1, texel1, 3);
-
-         /* 3D lerp */
-         for (c = 0; c < 4; c++) {
-            float ctemp0[4][4], ctemp1[4][4];
-            ctemp0[c][j] = lerp_2d(xw, yw,
-                                   texel0[c][0], texel0[c][1],
-                                   texel0[c][2], texel0[c][3]);
-            ctemp1[c][j] = lerp_2d(xw, yw,
-                                   texel1[c][0], texel1[c][1],
-                                   texel1[c][2], texel1[c][3]);
-            rgba[c][j] = LERP(zw, ctemp0[c][j], ctemp1[c][j]);
-         }
-
-         if (level0 != level1) {
-            /* get texels from second mipmap level and blend */
-            float rgba2[4][4];
-            x0 /= 2;
-            y0 /= 2;
-            z0 /= 2;
-            x1 /= 2;
-            y1 /= 2;
-            z1 /= 2;
-            get_texel(sampler, face, level1, x0, y0, z0, texel0, 0);
-            get_texel(sampler, face, level1, x1, y0, z0, texel0, 1);
-            get_texel(sampler, face, level1, x0, y1, z0, texel0, 2);
-            get_texel(sampler, face, level1, x1, y1, z0, texel0, 3);
-            get_texel(sampler, face, level1, x0, y0, z1, texel1, 0);
-            get_texel(sampler, face, level1, x1, y0, z1, texel1, 1);
-            get_texel(sampler, face, level1, x0, y1, z1, texel1, 2);
-            get_texel(sampler, face, level1, x1, y1, z1, texel1, 3);
-
-            /* 3D lerp */
+      {
+         int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4];
+         float xw[4], yw[4], zw[4]; /* interpolation weights */
+         linear_texcoord_4(sampler->wrap_s, s, width,  x0, x1, xw);
+         linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw);
+         linear_texcoord_4(sampler->wrap_r, p, depth,  z0, z1, zw);
+
+         for (j = 0; j < QUAD_SIZE; j++) {
+            int c;
+            float tx0[4][4], tx1[4][4];
+            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z0[j], tx0, 0);
+            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z0[j], tx0, 1);
+            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z0[j], tx0, 2);
+            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z0[j], tx0, 3);
+            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z1[j], tx1, 0);
+            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z1[j], tx1, 1);
+            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z1[j], tx1, 2);
+            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z1[j], tx1, 3);
+
+            /* interpolate R, G, B, A */
             for (c = 0; c < 4; c++) {
-               float ctemp0[4][4], ctemp1[4][4];
-               ctemp0[c][j] = lerp_2d(xw, yw,
-                                      texel0[c][0], texel0[c][1],
-                                      texel0[c][2], texel0[c][3]);
-               ctemp1[c][j] = lerp_2d(xw, yw,
-                                      texel1[c][0], texel1[c][1],
-                                      texel1[c][2], texel1[c][3]);
-               rgba2[c][j] = LERP(zw, ctemp0[c][j], ctemp1[c][j]);
+               rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j],
+                                    tx0[c][0], tx0[c][1],
+                                    tx0[c][2], tx0[c][3],
+                                    tx1[c][0], tx1[c][1],
+                                    tx1[c][2], tx1[c][3]);
             }
 
-            /* blend mipmap levels */
-            for (c = 0; c < NUM_CHANNELS; c++) {
-               rgba[c][j] = LERP(levelBlend, rgba[c][j], rgba2[c][j]);
+            if (level0 != level1) {
+               /* get texels from second mipmap level and blend */
+               float rgba2[4][4];
+               x0[j] /= 2;
+               y0[j] /= 2;
+               z0[j] /= 2;
+               x1[j] /= 2;
+               y1[j] /= 2;
+               z1[j] /= 2;
+               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z0[j], tx0, 0);
+               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z0[j], tx0, 1);
+               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z0[j], tx0, 2);
+               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z0[j], tx0, 3);
+               get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z1[j], tx1, 0);
+               get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z1[j], tx1, 1);
+               get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z1[j], tx1, 2);
+               get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z1[j], tx1, 3);
+
+               /* interpolate R, G, B, A */
+               for (c = 0; c < 4; c++) {
+                  rgba2[c][j] = lerp_3d(xw[j], yw[j], zw[j],
+                                        tx0[c][0], tx0[c][1],
+                                        tx0[c][2], tx0[c][3],
+                                        tx1[c][0], tx1[c][1],
+                                        tx1[c][2], tx1[c][3]);
+               }
+
+               /* blend mipmap levels */
+               for (c = 0; c < NUM_CHANNELS; c++) {
+                  rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]);
+               }
             }
          }
       }
@@ -939,7 +1033,7 @@ sp_get_samples_3d(struct tgsi_sampler *sampler,
 
 
 static void
-sp_get_samples_cube(struct tgsi_sampler *sampler,
+sp_get_samples_cube(const struct tgsi_sampler *sampler,
                     const float s[QUAD_SIZE],
                     const float t[QUAD_SIZE],
                     const float p[QUAD_SIZE],
@@ -956,62 +1050,73 @@ sp_get_samples_cube(struct tgsi_sampler *sampler,
 
 
 static void
-sp_get_samples_rect(struct tgsi_sampler *sampler,
+sp_get_samples_rect(const struct tgsi_sampler *tgsi_sampler,
                     const float s[QUAD_SIZE],
                     const float t[QUAD_SIZE],
                     const float p[QUAD_SIZE],
                     float lodbias,
                     float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
-   //sp_get_samples_2d_common(sampler, s, t, p, lodbias, rgba, faces);
-   static const uint face = 0;
-   const uint compare_func = sampler->state->compare_func;
+   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
+   const struct softpipe_context *sp = samp->sp;
+   const uint unit = samp->unit;
+   const struct pipe_texture *texture = sp->texture[unit];
+   const struct pipe_sampler_state *sampler = sp->sampler[unit];
+   const uint face = 0;
+   const uint compare_func = sampler->compare_func;
    unsigned level0, level1, j, imgFilter;
    int width, height;
    float levelBlend;
 
-   choose_mipmap_levels(sampler, s, t, p, lodbias,
+   choose_mipmap_levels(texture, sampler, s, t, p, lodbias,
                         &level0, &level1, &levelBlend, &imgFilter);
 
    /* texture RECTS cannot be mipmapped */
    assert(level0 == level1);
 
-   width = sampler->texture->width[level0];
-   height = sampler->texture->height[level0];
+   width = texture->width[level0];
+   height = texture->height[level0];
 
    assert(width > 0);
 
    switch (imgFilter) {
    case PIPE_TEX_FILTER_NEAREST:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         int x = nearest_texcoord_unnorm(sampler->state->wrap_s, s[j], width);
-         int y = nearest_texcoord_unnorm(sampler->state->wrap_t, t[j], height);
-         get_texel(sampler, face, level0, x, y, 0, rgba, j);
-         if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-            shadow_compare(compare_func, rgba, p, j);
+      {
+         int x[4], y[4];
+         nearest_texcoord_unnorm_4(sampler->wrap_s, s, width, x);
+         nearest_texcoord_unnorm_4(sampler->wrap_t, t, height, y);
+         for (j = 0; j < QUAD_SIZE; j++) {
+            get_texel(tgsi_sampler, face, level0, x[j], y[j], 0, rgba, j);
+            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+               shadow_compare(compare_func, rgba, p, j);
+            }
          }
       }
       break;
    case PIPE_TEX_FILTER_LINEAR:
    case PIPE_TEX_FILTER_ANISO:
-      for (j = 0; j < QUAD_SIZE; j++) {
-         float tx[4][4], a, b;
-         int x0, y0, x1, y1, c;
-         linear_texcoord_unnorm(sampler->state->wrap_s, s[j], width,  &x0, &x1, &a);
-         linear_texcoord_unnorm(sampler->state->wrap_t, t[j], height, &y0, &y1, &b);
-         get_texel(sampler, face, level0, x0, y0, 0, tx, 0);
-         get_texel(sampler, face, level0, x1, y0, 0, tx, 1);
-         get_texel(sampler, face, level0, x0, y1, 0, tx, 2);
-         get_texel(sampler, face, level0, x1, y1, 0, tx, 3);
-         if (sampler->state->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
-            shadow_compare(compare_func, tx, p, 0);
-            shadow_compare(compare_func, tx, p, 1);
-            shadow_compare(compare_func, tx, p, 2);
-            shadow_compare(compare_func, tx, p, 3);
-         }
-
-         for (c = 0; c < 4; c++) {
-            rgba[c][j] = lerp_2d(a, b, tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
+      {
+         int x0[4], y0[4], x1[4], y1[4];
+         float xw[4], yw[4]; /* weights */
+         linear_texcoord_unnorm_4(sampler->wrap_s, s, width,  x0, x1, xw);
+         linear_texcoord_unnorm_4(sampler->wrap_t, t, height, y0, y1, yw);
+         for (j = 0; j < QUAD_SIZE; j++) {
+            float tx[4][4]; /* texels */
+            int c;
+            get_texel(tgsi_sampler, face, level0, x0[j], y0[j], 0, tx, 0);
+            get_texel(tgsi_sampler, face, level0, x1[j], y0[j], 0, tx, 1);
+            get_texel(tgsi_sampler, face, level0, x0[j], y1[j], 0, tx, 2);
+            get_texel(tgsi_sampler, face, level0, x1[j], y1[j], 0, tx, 3);
+            if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) {
+               shadow_compare(compare_func, tx, p, 0);
+               shadow_compare(compare_func, tx, p, 1);
+               shadow_compare(compare_func, tx, p, 2);
+               shadow_compare(compare_func, tx, p, 3);
+            }
+            for (c = 0; c < 4; c++) {
+               rgba[c][j] = lerp_2d(xw[j], yw[j],
+                                    tx[c][0], tx[c][1], tx[c][2], tx[c][3]);
+            }
          }
       }
       break;
@@ -1021,49 +1126,45 @@ sp_get_samples_rect(struct tgsi_sampler *sampler,
 }
 
 
-
-
 /**
  * Called via tgsi_sampler::get_samples()
- * Use the sampler's state setting to get a filtered RGBA value
- * from the sampler's texture.
- *
- * XXX we can implement many versions of this function, each
- * tightly coded for a specific combination of sampler state
- * (nearest + repeat), (bilinear mipmap + clamp), etc.
- *
- * The update_samplers() function in st_atom_sampler.c could create
- * a new tgsi_sampler object for each state combo it finds....
+ * Get four filtered RGBA values from the sampler's texture.
  */
 void
-sp_get_samples(struct tgsi_sampler *sampler,
+sp_get_samples(struct tgsi_sampler *tgsi_sampler,
                const float s[QUAD_SIZE],
                const float t[QUAD_SIZE],
                const float p[QUAD_SIZE],
                float lodbias,
                float rgba[NUM_CHANNELS][QUAD_SIZE])
 {
-   if (!sampler->texture)
+   const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler);
+   const struct softpipe_context *sp = samp->sp;
+   const uint unit = samp->unit;
+   const struct pipe_texture *texture = sp->texture[unit];
+   const struct pipe_sampler_state *sampler = sp->sampler[unit];
+
+   if (!texture)
       return;
 
-   switch (sampler->texture->target) {
+   switch (texture->target) {
    case PIPE_TEXTURE_1D:
-      assert(sampler->state->normalized_coords);
-      sp_get_samples_1d(sampler, s, t, p, lodbias, rgba);
+      assert(sampler->normalized_coords);
+      sp_get_samples_1d(tgsi_sampler, s, t, p, lodbias, rgba);
       break;
    case PIPE_TEXTURE_2D:
-      if (sampler->state->normalized_coords)
-         sp_get_samples_2d(sampler, s, t, p, lodbias, rgba);
+      if (sampler->normalized_coords)
+         sp_get_samples_2d(tgsi_sampler, s, t, p, lodbias, rgba);
       else
-         sp_get_samples_rect(sampler, s, t, p, lodbias, rgba);
+         sp_get_samples_rect(tgsi_sampler, s, t, p, lodbias, rgba);
       break;
    case PIPE_TEXTURE_3D:
-      assert(sampler->state->normalized_coords);
-      sp_get_samples_3d(sampler, s, t, p, lodbias, rgba);
+      assert(sampler->normalized_coords);
+      sp_get_samples_3d(tgsi_sampler, s, t, p, lodbias, rgba);
       break;
    case PIPE_TEXTURE_CUBE:
-      assert(sampler->state->normalized_coords);
-      sp_get_samples_cube(sampler, s, t, p, lodbias, rgba);
+      assert(sampler->normalized_coords);
+      sp_get_samples_cube(tgsi_sampler, s, t, p, lodbias, rgba);
       break;
    default:
       assert(0);
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.h b/src/gallium/drivers/softpipe/sp_tex_sample.h
index 404bfd0c365..f0a8a787782 100644
--- a/src/gallium/drivers/softpipe/sp_tex_sample.h
+++ b/src/gallium/drivers/softpipe/sp_tex_sample.h
@@ -1,8 +1,56 @@
+/**************************************************************************
+ * 
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
 #ifndef SP_TEX_SAMPLE_H
 #define SP_TEX_SAMPLE_H
 
 
-struct tgsi_sampler;
+#include "tgsi/tgsi_exec.h"
+
+
+/**
+ * Subclass of tgsi_sampler
+ */
+struct sp_shader_sampler
+{
+   struct tgsi_sampler base;  /**< base class */
+
+   uint unit;
+   struct softpipe_context *sp;
+   struct softpipe_tile_cache *cache;
+};
+
+
+
+static INLINE const struct sp_shader_sampler *
+sp_shader_sampler(const struct tgsi_sampler *sampler)
+{
+   return (const struct sp_shader_sampler *) sampler;
+}
 
 
 extern void
diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c
index b50c9845133..78b0efa46d2 100644
--- a/src/gallium/drivers/softpipe/sp_tile_cache.c
+++ b/src/gallium/drivers/softpipe/sp_tile_cache.c
@@ -494,11 +494,11 @@ tex_cache_pos(int x, int y, int z, int face, int level)
  * Tiles are read-only and indexed with more params.
  */
 const struct softpipe_cached_tile *
-sp_get_cached_tile_tex(struct pipe_context *pipe,
+sp_get_cached_tile_tex(struct softpipe_context *sp,
                        struct softpipe_tile_cache *tc, int x, int y, int z,
                        int face, int level)
 {
-   struct pipe_screen *screen = pipe->screen;
+   struct pipe_screen *screen = sp->pipe.screen;
    /* tile pos in framebuffer: */
    const int tile_x = x & ~(TILE_SIZE - 1);
    const int tile_y = y & ~(TILE_SIZE - 1);
diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.h b/src/gallium/drivers/softpipe/sp_tile_cache.h
index bc96c941f61..a66bb50bcc1 100644
--- a/src/gallium/drivers/softpipe/sp_tile_cache.h
+++ b/src/gallium/drivers/softpipe/sp_tile_cache.h
@@ -96,7 +96,7 @@ sp_get_cached_tile(struct softpipe_context *softpipe,
                    struct softpipe_tile_cache *tc, int x, int y);
 
 extern const struct softpipe_cached_tile *
-sp_get_cached_tile_tex(struct pipe_context *pipe,
+sp_get_cached_tile_tex(struct softpipe_context *softpipe,
                        struct softpipe_tile_cache *tc, int x, int y, int z,
                        int face, int level);
 
diff --git a/src/gallium/include/pipe/p_compiler.h b/src/gallium/include/pipe/p_compiler.h
index 4d64c74a4aa..7bcebd3d6b6 100644
--- a/src/gallium/include/pipe/p_compiler.h
+++ b/src/gallium/include/pipe/p_compiler.h
@@ -144,10 +144,12 @@ typedef unsigned char boolean;
 #define ALIGN16_DECL(TYPE, NAME, SIZE)  TYPE NAME##___aligned[SIZE] __attribute__(( aligned( 16 ) ))
 #define ALIGN16_ASSIGN(NAME) NAME##___aligned
 #define ALIGN16_ATTRIB  __attribute__(( aligned( 16 ) ))
+#define ALIGN8_ATTRIB  __attribute__(( aligned( 8 ) ))
 #else
 #define ALIGN16_DECL(TYPE, NAME, SIZE)  TYPE NAME##___unaligned[SIZE + 1]
 #define ALIGN16_ASSIGN(NAME) align16(NAME##___unaligned)
 #define ALIGN16_ATTRIB
+#define ALIGN8_ATTRIB
 #endif
 
 
diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h
index af3746c0265..05cbd2fc4df 100644
--- a/src/gallium/include/pipe/p_config.h
+++ b/src/gallium/include/pipe/p_config.h
@@ -85,8 +85,19 @@
 #define PIPE_ARCH_X86_64
 #endif
 
-#if 0 /* FIXME */
+#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+#if defined(PIPE_CC_GCC) && !defined(__SSE2__)
+/* #warning SSE2 support requires -msse -msse2 compiler options */
+#else
+#define PIPE_ARCH_SSE
+#endif
+#endif
+
+#if defined(__PPC__)
 #define PIPE_ARCH_PPC
+#if defined(__PPC64__)
+#define PIPE_ARCH_PPC_64
+#endif
 #endif
 
 
diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h
index dc8a92dccb0..6bfac581f9f 100644
--- a/src/gallium/include/pipe/p_defines.h
+++ b/src/gallium/include/pipe/p_defines.h
@@ -171,6 +171,8 @@ enum pipe_texture_target {
 #define PIPE_TEXTURE_USAGE_PRIMARY         0x4 /* ie a frontbuffer */
 #define PIPE_TEXTURE_USAGE_DEPTH_STENCIL   0x8
 #define PIPE_TEXTURE_USAGE_SAMPLER         0x10
+/** Pipe driver custom usage flags should be greater or equal to this value */
+#define PIPE_TEXTURE_USAGE_CUSTOM          (1 << 16)
 
 #define PIPE_TEXTURE_GEOM_NON_SQUARE       0x1
 #define PIPE_TEXTURE_GEOM_NON_POWER_OF_TWO 0x2
@@ -292,6 +294,7 @@ enum pipe_texture_target {
 #define PIPE_CAP_GUARD_BAND_BOTTOM       23  /*< float */
 #define PIPE_CAP_TEXTURE_MIRROR_CLAMP    24
 #define PIPE_CAP_TEXTURE_MIRROR_REPEAT   25
+#define PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS 26
 
 
 
diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h
index 97a4c8c510b..6bd55d77358 100644
--- a/src/gallium/include/pipe/p_format.h
+++ b/src/gallium/include/pipe/p_format.h
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright (c) 2008 VMware, Inc.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -245,13 +246,14 @@ static INLINE uint pf_rev(pipe_format_ycbcr_t f)
 /**
   * Compresssed format layouts (this will probably change)
   */
-#define _PIPE_FORMAT_DXT( LEVEL, RSIZE, GSIZE, BSIZE, ASIZE ) \
+#define _PIPE_FORMAT_DXT( LEVEL, RSIZE, GSIZE, BSIZE, ASIZE, TYPE ) \
    ((PIPE_FORMAT_LAYOUT_DXT << 0) | \
     ((LEVEL) << 2) | \
     ((RSIZE) << 5) | \
     ((GSIZE) << 8) | \
     ((BSIZE) << 11) | \
-    ((ASIZE) << 14) )
+    ((ASIZE) << 14) | \
+    ((TYPE) << 29))
 
 
 
@@ -360,20 +362,30 @@ enum pipe_format {
    PIPE_FORMAT_R32G32B32A32_FIXED    = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 4, 4, 4, 4, PIPE_FORMAT_TYPE_FIXED ),
    /* sRGB formats */
    PIPE_FORMAT_L8_SRGB               = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRR1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SRGB ),
-   PIPE_FORMAT_A8_L8_SRGB            = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_A8L8_SRGB             = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RRRG, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
    PIPE_FORMAT_R8G8B8_SRGB           = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 0, PIPE_FORMAT_TYPE_SRGB ),
    PIPE_FORMAT_R8G8B8A8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGBA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
    PIPE_FORMAT_R8G8B8X8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_RGB1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_A8R8G8B8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_ARGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_X8R8G8B8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_1RGB, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_B8G8R8A8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGRA, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_B8G8R8X8_SRGB         = _PIPE_FORMAT_RGBAZS_8 ( _PIPE_FORMAT_BGR1, 1, 1, 1, 1, PIPE_FORMAT_TYPE_SRGB ),
 
    /* mixed formats */
    PIPE_FORMAT_X8UB8UG8SR8S_NORM     = _PIPE_FORMAT_MIXED( _PIPE_FORMAT_1BGR, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1 ),
    PIPE_FORMAT_B6UG5SR5S_NORM        = _PIPE_FORMAT_MIXED( _PIPE_FORMAT_BGR1, 6, 5, 5, 0, 0, 1, 1, 0, 1, 0 ),
 
    /* compressed formats */
-   PIPE_FORMAT_DXT1_RGB              = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 0 ),
-   PIPE_FORMAT_DXT1_RGBA             = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 8 ),
-   PIPE_FORMAT_DXT3_RGBA             = _PIPE_FORMAT_DXT( 3, 8, 8, 8, 8 ),
-   PIPE_FORMAT_DXT5_RGBA             = _PIPE_FORMAT_DXT( 5, 8, 8, 8, 8 )
+   PIPE_FORMAT_DXT1_RGB              = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 0, PIPE_FORMAT_TYPE_UNORM ),
+   PIPE_FORMAT_DXT1_RGBA             = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 8, PIPE_FORMAT_TYPE_UNORM ),
+   PIPE_FORMAT_DXT3_RGBA             = _PIPE_FORMAT_DXT( 3, 8, 8, 8, 8, PIPE_FORMAT_TYPE_UNORM ),
+   PIPE_FORMAT_DXT5_RGBA             = _PIPE_FORMAT_DXT( 5, 8, 8, 8, 8, PIPE_FORMAT_TYPE_UNORM ),
+
+   /* sRGB, compressed */
+   PIPE_FORMAT_DXT1_SRGB             = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 0, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_DXT1_SRGBA            = _PIPE_FORMAT_DXT( 1, 8, 8, 8, 8, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_DXT3_SRGBA            = _PIPE_FORMAT_DXT( 3, 8, 8, 8, 8, PIPE_FORMAT_TYPE_SRGB ),
+   PIPE_FORMAT_DXT5_SRGBA            = _PIPE_FORMAT_DXT( 5, 8, 8, 8, 8, PIPE_FORMAT_TYPE_SRGB )
 };
 
 /**
@@ -477,12 +489,16 @@ pf_get_block(enum pipe_format format, struct pipe_format_block *block)
    switch(format) {
    case PIPE_FORMAT_DXT1_RGBA:
    case PIPE_FORMAT_DXT1_RGB:
+   case PIPE_FORMAT_DXT1_SRGBA:
+   case PIPE_FORMAT_DXT1_SRGB:
       block->size = 8;
       block->width = 4;
       block->height = 4;
       break;
    case PIPE_FORMAT_DXT3_RGBA:
    case PIPE_FORMAT_DXT5_RGBA:
+   case PIPE_FORMAT_DXT3_SRGBA:
+   case PIPE_FORMAT_DXT5_SRGBA:
       block->size = 16;
       block->width = 4;
       block->height = 4;
@@ -540,7 +556,7 @@ pf_has_alpha( enum pipe_format format )
       /* FIXME: pf_get_component_bits( PIPE_FORMAT_A8L8_UNORM, PIPE_FORMAT_COMP_A ) should not return 0 right? */
       if(format == PIPE_FORMAT_A8_UNORM || 
          format == PIPE_FORMAT_A8L8_UNORM || 
-         format == PIPE_FORMAT_A8_L8_SRGB)
+         format == PIPE_FORMAT_A8L8_SRGB)
          return TRUE;
       return pf_get_component_bits( format, PIPE_FORMAT_COMP_A ) ? TRUE : FALSE;
    case PIPE_FORMAT_LAYOUT_YCBCR:
@@ -550,6 +566,9 @@ pf_has_alpha( enum pipe_format format )
       case PIPE_FORMAT_DXT1_RGBA:
       case PIPE_FORMAT_DXT3_RGBA:
       case PIPE_FORMAT_DXT5_RGBA:
+      case PIPE_FORMAT_DXT1_SRGBA:
+      case PIPE_FORMAT_DXT3_SRGBA:
+      case PIPE_FORMAT_DXT5_SRGBA:
          return TRUE;
       default:
          return FALSE;
diff --git a/src/gallium/include/pipe/p_inlines.h b/src/gallium/include/pipe/p_inlines.h
index d70de8e3011..5e79b7f485a 100644
--- a/src/gallium/include/pipe/p_inlines.h
+++ b/src/gallium/include/pipe/p_inlines.h
@@ -82,11 +82,14 @@ static INLINE void
 pipe_surface_reference(struct pipe_surface **ptr, struct pipe_surface *surf)
 {
    /* bump the refcount first */
-   if (surf) 
+   if (surf) {
+      assert(surf->refcount);
       surf->refcount++;
+   }
 
    if (*ptr) {
-
+      assert((*ptr)->refcount);
+      
       /* There are currently two sorts of surfaces... This needs to be
        * fixed so that all surfaces are views into a texture.
        */
@@ -113,11 +116,16 @@ winsys_buffer_reference(struct pipe_winsys *winsys,
 		      struct pipe_buffer **ptr,
 		      struct pipe_buffer *buf)
 {
-   if (buf) 
+   if (buf) {
+      assert(buf->refcount);
       buf->refcount++;
+   }
 
-   if (*ptr && --(*ptr)->refcount == 0)
-      winsys->buffer_destroy( winsys, *ptr );
+   if (*ptr) {
+      assert((*ptr)->refcount);
+      if(--(*ptr)->refcount == 0)
+         winsys->buffer_destroy( winsys, *ptr );
+   }
 
    *ptr = buf;
 }
@@ -133,12 +141,15 @@ pipe_texture_reference(struct pipe_texture **ptr,
 {
    assert(ptr);
 
-   if (pt) 
+   if (pt) { 
+      assert(pt->refcount);
       pt->refcount++;
+   }
 
    if (*ptr) {
       struct pipe_screen *screen = (*ptr)->screen;
       assert(screen);
+      assert((*ptr)->refcount);
       screen->texture_release(screen, ptr);
 
       assert(!*ptr);
@@ -154,6 +165,7 @@ pipe_texture_release(struct pipe_texture **ptr)
    struct pipe_screen *screen;
    assert(ptr);
    screen = (*ptr)->screen;
+   assert((*ptr)->refcount);
    screen->texture_release(screen, ptr);
    *ptr = NULL;
 }
@@ -176,12 +188,6 @@ pipe_user_buffer_create( struct pipe_screen *screen, void *ptr, unsigned size )
    return screen->winsys->user_buffer_create(screen->winsys, ptr, size);
 }
 
-static INLINE void
-pipe_buffer_destroy( struct pipe_screen *screen, struct pipe_buffer *buf )
-{
-   screen->winsys->buffer_destroy(screen->winsys, buf);
-}
-
 static INLINE void *
 pipe_buffer_map(struct pipe_screen *screen,
                 struct pipe_buffer *buf,
diff --git a/src/gallium/include/pipe/p_screen.h b/src/gallium/include/pipe/p_screen.h
index b15affef7a5..3bedc752947 100644
--- a/src/gallium/include/pipe/p_screen.h
+++ b/src/gallium/include/pipe/p_screen.h
@@ -26,6 +26,8 @@
  **************************************************************************/
 
 /**
+ * @file
+ * 
  * Screen, Adapter or GPU
  *
  * These are driver functions/facilities that are context independent.
diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h
index a562e3a6b14..ad43d799f32 100644
--- a/src/gallium/include/pipe/p_shader_tokens.h
+++ b/src/gallium/include/pipe/p_shader_tokens.h
@@ -1,4 +1,31 @@
-#if !defined TGSI_TOKEN_H
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef TGSI_TOKEN_H
 #define TGSI_TOKEN_H
 
 #ifdef __cplusplus
@@ -36,10 +63,10 @@ struct tgsi_processor
 
 struct tgsi_token
 {
-   unsigned Type       : 4;  /* TGSI_TOKEN_TYPE_ */
-   unsigned Size       : 8;  /* UINT */
+   unsigned Type       : 4;  /**< TGSI_TOKEN_TYPE_x */
+   unsigned Size       : 8;  /**< UINT */
    unsigned Padding    : 19;
-   unsigned Extended   : 1;  /* BOOL */
+   unsigned Extended   : 1;  /**< BOOL */
 };
 
 enum tgsi_file_type {
@@ -79,20 +106,22 @@ enum tgsi_file_type {
 
 struct tgsi_declaration
 {
-   unsigned Type        : 4;  /* TGSI_TOKEN_TYPE_DECLARATION */
-   unsigned Size        : 8;  /* UINT */
-   unsigned File        : 4;  /* one of TGSI_FILE_x */
-   unsigned UsageMask   : 4;  /* bitmask of TGSI_WRITEMASK_x flags */
-   unsigned Interpolate : 4;  /* TGSI_INTERPOLATE_ */
-   unsigned Semantic    : 1;  /* BOOL, any semantic info? */
-   unsigned Padding     : 6;
-   unsigned Extended    : 1;  /* BOOL */
+   unsigned Type        : 4;  /**< TGSI_TOKEN_TYPE_DECLARATION */
+   unsigned Size        : 8;  /**< UINT */
+   unsigned File        : 4;  /**< one of TGSI_FILE_x */
+   unsigned UsageMask   : 4;  /**< bitmask of TGSI_WRITEMASK_x flags */
+   unsigned Interpolate : 4;  /**< one of TGSI_INTERPOLATE_x */
+   unsigned Semantic    : 1;  /**< BOOL, any semantic info? */
+   unsigned Centroid    : 1;  /**< centroid sampling? */
+   unsigned Invariant   : 1;  /**< invariant optimization? */
+   unsigned Padding     : 4;
+   unsigned Extended    : 1;  /**< BOOL */
 };
 
 struct tgsi_declaration_range
 {
-   unsigned First   : 16; /* UINT */
-   unsigned Last    : 16; /* UINT */
+   unsigned First   : 16; /**< UINT */
+   unsigned Last    : 16; /**< UINT */
 };
 
 #define TGSI_SEMANTIC_POSITION 0
@@ -106,8 +135,8 @@ struct tgsi_declaration_range
 
 struct tgsi_declaration_semantic
 {
-   unsigned SemanticName   : 8;  /* one of TGSI_SEMANTIC_ */
-   unsigned SemanticIndex  : 16; /* UINT */
+   unsigned SemanticName   : 8;  /**< one of TGSI_SEMANTIC_x */
+   unsigned SemanticIndex  : 16; /**< UINT */
    unsigned Padding        : 8;
 };
 
@@ -115,11 +144,11 @@ struct tgsi_declaration_semantic
 
 struct tgsi_immediate
 {
-   unsigned Type       : 4;  /* TGSI_TOKEN_TYPE_IMMEDIATE */
-   unsigned Size       : 8;  /* UINT */
-   unsigned DataType   : 4;  /* TGSI_IMM_ */
+   unsigned Type       : 4;  /**< TGSI_TOKEN_TYPE_IMMEDIATE */
+   unsigned Size       : 8;  /**< UINT */
+   unsigned DataType   : 4;  /**< one of TGSI_IMM_x */
    unsigned Padding    : 15;
-   unsigned Extended   : 1;  /* BOOL */
+   unsigned Extended   : 1;  /**< BOOL */
 };
 
 struct tgsi_immediate_float32
@@ -396,7 +425,7 @@ struct tgsi_immediate_float32
 #define TGSI_SAT_ZERO_ONE        1  /* clamp to [0,1] */
 #define TGSI_SAT_MINUS_PLUS_ONE  2  /* clamp to [-1,1] */
 
-/*
+/**
  * Opcode is the operation code to execute. A given operation defines the
  * semantics how the source registers (if any) are interpreted and what is
  * written to the destination registers (if any) as a result of execution.
@@ -481,7 +510,7 @@ struct tgsi_instruction_ext
 #define TGSI_SWIZZLE_Z      2
 #define TGSI_SWIZZLE_W      3
 
-/*
+/**
  * Precision controls the precision at which the operation should be executed.
  *
  * CondDstUpdate enables condition code register writes. When this field is
@@ -548,7 +577,7 @@ struct tgsi_instruction_ext_predicate
    unsigned Extended         : 1;    /* BOOL */
 };
 
-/*
+/**
  * File specifies the register array to access.
  *
  * Index specifies the element number of a register in the register file.
@@ -580,7 +609,7 @@ struct tgsi_src_register
    unsigned Extended    : 1;  /* BOOL */
 };
 
-/*
+/**
  * If tgsi_src_register::Extended is TRUE, tgsi_src_register_ext follows.
  * 
  * Then, if tgsi_src_register::Indirect is TRUE, another tgsi_src_register
@@ -599,7 +628,7 @@ struct tgsi_src_register_ext
    unsigned Extended : 1;    /* BOOL */
 };
 
-/*
+/**
  * If tgsi_src_register_ext::Type is TGSI_SRC_REGISTER_EXT_TYPE_SWZ,
  * it should be cast to tgsi_src_register_ext_swz.
  * 
@@ -617,7 +646,7 @@ struct tgsi_src_register_ext
 #define TGSI_EXTSWIZZLE_ZERO    4
 #define TGSI_EXTSWIZZLE_ONE     5
 
-/*
+/**
  * ExtSwizzleX, ExtSwizzleY, ExtSwizzleZ and ExtSwizzleW swizzle the source
  * register in an extended manner.
  *
diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
index da783389dae..342f17260a7 100644
--- a/src/gallium/include/pipe/p_state.h
+++ b/src/gallium/include/pipe/p_state.h
@@ -27,6 +27,8 @@
 
 
 /**
+ * @file
+ * 
  * Abstract graphics pipe state objects.
  *
  * Basic notes:
diff --git a/src/gallium/include/pipe/p_thread.h b/src/gallium/include/pipe/p_thread.h
index e01d5a602b8..8af3cd958b0 100644
--- a/src/gallium/include/pipe/p_thread.h
+++ b/src/gallium/include/pipe/p_thread.h
@@ -25,6 +25,8 @@
 
 
 /**
+ * @file
+ * 
  * Thread, mutex, condition var and thread-specific data functions.
  */
 
diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile
new file mode 100644
index 00000000000..4f7a9534847
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/Makefile
@@ -0,0 +1,18 @@
+TARGET		= libg3dvl.a
+OBJECTS		= vl_display.o vl_screen.o vl_context.o vl_surface.o vl_shader_build.o vl_util.o vl_basic_csc.o	\
+		  vl_r16snorm_mc.o vl_r16snorm_mc_buf.o
+GALLIUMDIR	= ../..
+
+CFLAGS		+= -g -Wall -fPIC -I${GALLIUMDIR}/include -I${GALLIUMDIR}/auxiliary -I${GALLIUMDIR}/winsys/g3dvl
+
+#############################################
+
+.PHONY	= all clean
+
+all: ${TARGET}
+
+${TARGET}: ${OBJECTS}
+	ar rcs $@ $^
+
+clean:
+	rm -rf ${OBJECTS} ${TARGET}
diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c
new file mode 100644
index 00000000000..e3b3d032566
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c
@@ -0,0 +1,712 @@
+#define VL_INTERNAL
+#include "vl_basic_csc.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <pipe/p_context.h>
+#include <pipe/p_winsys.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+#include "vl_csc.h"
+#include "vl_surface.h"
+#include "vl_shader_build.h"
+#include "vl_types.h"
+
+struct vlVertexShaderConsts
+{
+	struct vlVertex4f	dst_scale;
+	struct vlVertex4f	dst_trans;
+	struct vlVertex4f	src_scale;
+	struct vlVertex4f	src_trans;
+};
+
+struct vlFragmentShaderConsts
+{
+	struct vlVertex4f	bias;
+	float			matrix[16];
+};
+
+struct vlBasicCSC
+{
+	struct vlCSC				base;
+
+	struct pipe_context			*pipe;
+	struct pipe_viewport_state		viewport;
+	struct pipe_framebuffer_state		framebuffer;
+	struct pipe_texture			*framebuffer_tex;
+	void					*sampler;
+	void					*vertex_shader, *fragment_shader;
+	struct pipe_vertex_buffer 		vertex_bufs[2];
+	struct pipe_vertex_element		vertex_elems[2];
+	struct pipe_constant_buffer		vs_const_buf, fs_const_buf;
+};
+
+static int vlResizeFrameBuffer
+(
+	struct vlCSC *csc,
+	unsigned int width,
+	unsigned int height
+)
+{
+	struct vlBasicCSC	*basic_csc;
+	struct pipe_context	*pipe;
+	struct pipe_texture	template;
+
+	assert(csc);
+
+	basic_csc = (struct vlBasicCSC*)csc;
+	pipe = basic_csc->pipe;
+
+	if (basic_csc->framebuffer.width == width && basic_csc->framebuffer.height == height)
+		return 0;
+
+	basic_csc->viewport.scale[0] = width;
+	basic_csc->viewport.scale[1] = height;
+	basic_csc->viewport.scale[2] = 1;
+	basic_csc->viewport.scale[3] = 1;
+	basic_csc->viewport.translate[0] = 0;
+	basic_csc->viewport.translate[1] = 0;
+	basic_csc->viewport.translate[2] = 0;
+	basic_csc->viewport.translate[3] = 0;
+	
+	if (basic_csc->framebuffer_tex)
+		pipe_texture_release(&basic_csc->framebuffer_tex);
+	
+	memset(&template, 0, sizeof(struct pipe_texture));
+	template.target = PIPE_TEXTURE_2D;
+	template.format = PIPE_FORMAT_A8R8G8B8_UNORM;
+	template.last_level = 0;
+	template.width[0] = width;
+	template.height[0] = height;
+	template.depth[0] = 1;
+	template.compressed = 0;
+	pf_get_block(template.format, &template.block);
+	template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET;
+
+	basic_csc->framebuffer_tex = pipe->screen->texture_create(pipe->screen, &template);
+
+	basic_csc->framebuffer.width = width;
+	basic_csc->framebuffer.height = height;
+	basic_csc->framebuffer.cbufs[0] = pipe->screen->get_tex_surface
+	(
+		pipe->screen,
+		basic_csc->framebuffer_tex,
+		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+	);
+
+	/* Clear to black, in case video doesn't fill the entire window */
+	pipe->clear(pipe, basic_csc->framebuffer.cbufs[0], 0);
+
+	return 0;
+}
+
+static int vlBegin
+(
+	struct vlCSC *csc
+)
+{
+	struct vlBasicCSC	*basic_csc;
+	struct pipe_context	*pipe;
+
+	assert(csc);
+
+	basic_csc = (struct vlBasicCSC*)csc;
+	pipe = basic_csc->pipe;
+
+	pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer);
+	pipe->set_viewport_state(pipe, &basic_csc->viewport);
+	pipe->bind_sampler_states(pipe, 1, (void**)&basic_csc->sampler);
+	/* Source texture set in vlPutPictureCSC() */
+	pipe->bind_vs_state(pipe, basic_csc->vertex_shader);
+	pipe->bind_fs_state(pipe, basic_csc->fragment_shader);
+	pipe->set_vertex_buffers(pipe, 2, basic_csc->vertex_bufs);
+	pipe->set_vertex_elements(pipe, 2, basic_csc->vertex_elems);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &basic_csc->vs_const_buf);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &basic_csc->fs_const_buf);
+
+	return 0;
+}
+
+static int vlPutPictureCSC
+(
+	struct vlCSC *csc,
+	struct vlSurface *surface,
+	int srcx,
+	int srcy,
+	int srcw,
+	int srch,
+	int destx,
+	int desty,
+	int destw,
+	int desth,
+	enum vlPictureType picture_type
+)
+{
+	struct vlBasicCSC		*basic_csc;
+	struct pipe_context		*pipe;
+	struct vlVertexShaderConsts	*vs_consts;
+
+	assert(csc);
+	assert(surface);
+
+	basic_csc = (struct vlBasicCSC*)csc;
+	pipe = basic_csc->pipe;
+
+	vs_consts = pipe->winsys->buffer_map
+	(
+		pipe->winsys,
+		basic_csc->vs_const_buf.buffer,
+		PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	vs_consts->dst_scale.x = destw / (float)basic_csc->framebuffer.cbufs[0]->width;
+	vs_consts->dst_scale.y = desth / (float)basic_csc->framebuffer.cbufs[0]->height;
+	vs_consts->dst_scale.z = 1;
+	vs_consts->dst_scale.w = 1;
+	vs_consts->dst_trans.x = destx / (float)basic_csc->framebuffer.cbufs[0]->width;
+	vs_consts->dst_trans.y = desty / (float)basic_csc->framebuffer.cbufs[0]->height;
+	vs_consts->dst_trans.z = 0;
+	vs_consts->dst_trans.w = 0;
+
+	vs_consts->src_scale.x = srcw / (float)surface->texture->width[0];
+	vs_consts->src_scale.y = srch / (float)surface->texture->height[0];
+	vs_consts->src_scale.z = 1;
+	vs_consts->src_scale.w = 1;
+	vs_consts->src_trans.x = srcx / (float)surface->texture->width[0];
+	vs_consts->src_trans.y = srcy / (float)surface->texture->height[0];
+	vs_consts->src_trans.z = 0;
+	vs_consts->src_trans.w = 0;
+
+	pipe->winsys->buffer_unmap(pipe->winsys, basic_csc->vs_const_buf.buffer);
+
+	pipe->set_sampler_textures(pipe, 1, &surface->texture);
+	pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4);
+
+	return 0;
+}
+
+static int vlEnd
+(
+	struct vlCSC *csc
+)
+{
+	assert(csc);
+
+	return 0;
+}
+
+static struct pipe_surface* vlGetFrameBuffer
+(
+	struct vlCSC *csc
+)
+{
+	struct vlBasicCSC	*basic_csc;
+
+	assert(csc);
+
+	basic_csc = (struct vlBasicCSC*)csc;
+
+	return basic_csc->framebuffer.cbufs[0];
+}
+
+static int vlDestroy
+(
+	struct vlCSC *csc
+)
+{
+	struct vlBasicCSC	*basic_csc;
+	struct pipe_context	*pipe;
+	unsigned int		i;
+
+	assert(csc);
+
+	basic_csc = (struct vlBasicCSC*)csc;
+	pipe = basic_csc->pipe;
+
+	if (basic_csc->framebuffer_tex)
+		pipe_texture_release(&basic_csc->framebuffer_tex);
+
+	pipe->delete_sampler_state(pipe, basic_csc->sampler);
+	pipe->delete_vs_state(pipe, basic_csc->vertex_shader);
+	pipe->delete_fs_state(pipe, basic_csc->fragment_shader);
+
+	for (i = 0; i < 2; ++i)
+		pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->vertex_bufs[i].buffer);
+
+	pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->vs_const_buf.buffer);
+	pipe->winsys->buffer_destroy(pipe->winsys, basic_csc->fs_const_buf.buffer);
+
+	free(basic_csc);
+
+	return 0;
+}
+
+/*
+ * Represents 2 triangles in a strip in normalized coords.
+ * Used to render the surface onto the frame buffer.
+ */
+static const struct vlVertex2f surface_verts[4] =
+{
+	{0.0f, 0.0f},
+	{0.0f, 1.0f},
+	{1.0f, 0.0f},
+	{1.0f, 1.0f}
+};
+
+/*
+ * Represents texcoords for the above. We can use the position values directly.
+ * TODO: Duplicate these in the shader, no need to create a buffer.
+ */
+static const struct vlVertex2f *surface_texcoords = surface_verts;
+
+/*
+ * Identity color conversion constants, for debugging
+ */
+static const struct vlFragmentShaderConsts identity =
+{
+	{
+		0.0f, 0.0f, 0.0f, 0.0f
+	},
+	{
+		1.0f, 0.0f, 0.0f, 0.0f,
+		0.0f, 1.0f, 0.0f, 0.0f,
+		0.0f, 0.0f, 1.0f, 0.0f,
+		0.0f, 0.0f, 0.0f, 1.0f
+	}
+};
+
+/*
+ * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [16,235]
+ */
+static const struct vlFragmentShaderConsts bt_601 =
+{
+	{
+		0.0f,		0.501960784f,	0.501960784f,	0.0f
+	},
+	{
+		1.0f,		0.0f,		1.371f,		0.0f,
+		1.0f,		-0.336f,	-0.698f,	0.0f,
+		1.0f,		1.732f,		0.0f,		0.0f,
+		0.0f,		0.0f,		0.0f,		1.0f
+	}
+};
+
+/*
+ * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [0,255]
+ */
+static const struct vlFragmentShaderConsts bt_601_full =
+{
+	{
+		0.062745098f,	0.501960784f,	0.501960784f,	0.0f
+	},
+	{
+		1.164f,		0.0f,		1.596f,		0.0f,
+		1.164f,		-0.391f,	-0.813f,	0.0f,
+		1.164f,		2.018f,		0.0f,		0.0f,
+		0.0f,		0.0f,		0.0f,		1.0f
+	}
+};
+
+/*
+ * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [16,235]
+ */
+static const struct vlFragmentShaderConsts bt_709 =
+{
+	{
+		0.0f,		0.501960784f,	0.501960784f,	0.0f
+	},
+	{
+		1.0f,		0.0f,		1.540f,		0.0f,
+		1.0f,		-0.183f,	-0.459f,	0.0f,
+		1.0f,		1.816f,		0.0f,		0.0f,
+		0.0f,		0.0f,		0.0f,		1.0f
+	}
+};
+
+/*
+ * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where:
+ * Y is in [16,235], Cb and Cr are in [16,240]
+ * R, G, and B are in [0,255]
+ */
+const struct vlFragmentShaderConsts bt_709_full =
+{
+	{
+		0.062745098f,	0.501960784f,	0.501960784f,	0.0f
+	},
+	{
+		1.164f,		0.0f,		1.793f,		0.0f,
+		1.164f,		-0.213f,	-0.534f,	0.0f,
+		1.164f,		2.115f,		0.0f,		0.0f,
+		0.0f,		0.0f,		0.0f,		1.0f
+	}
+};
+
+static int vlCreateVertexShader
+(
+	struct vlBasicCSC *csc
+)
+{
+	const unsigned int		max_tokens = 50;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(context);
+
+	pipe = csc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Vertex texcoords
+	 */
+	for (i = 0; i < 2; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale vertex pos rect to destination size
+	 * decl c1		; Translation vector to move vertex pos rect into position
+	 * decl c2		; Scaling vector to scale texcoord rect to source size
+	 * decl c3		; Translation vector to move texcoord rect into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Vertex texcoords
+	 */
+	for (i = 0; i < 2; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * madd o0, i0, c0, c1	; Scale and translate unit output rect to destination size and pos
+	 * madd o1, i1, c2, c3	; Scale and translate unit texcoord rect to source size and pos
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_inst4(TGSI_OPCODE_MADD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	csc->vertex_shader = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShader
+(
+	struct vlBasicCSC *csc
+)
+{
+	const unsigned int		max_tokens = 50;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(context);
+
+	pipe = csc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/* decl i0		; Texcoords for s0 */
+	decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl c0		; Bias vector for CSC
+	 * decl c1-c4		; CSC matrix c1-c4
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0		; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0 */
+	decl = vl_decl_temps(0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl s0		; Sampler for tex containing picture to display */
+	decl = vl_decl_samplers(0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* tex2d t0, i0, s0	; Read src pixel */
+	inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* sub t0, t0, c0	; Subtract bias vector from pixel */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * dp4 o0.x, t0, c1	; Multiply pixel by the color conversion matrix
+	 * dp4 o0.y, t0, c2
+	 * dp4 o0.z, t0, c3
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1);
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	csc->fragment_shader = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateDataBufs
+(
+	struct vlBasicCSC *csc
+)
+{
+	struct pipe_context *pipe;
+
+	assert(csc);
+
+	pipe = csc->pipe;
+
+	/*
+	 * Create our vertex buffer and vertex buffer element
+	 * VB contains 4 vertices that render a quad covering the entire window
+	 * to display a rendered surface
+	 * Quad is rendered as a tri strip
+	 */
+	csc->vertex_bufs[0].pitch = sizeof(struct vlVertex2f);
+	csc->vertex_bufs[0].max_index = 3;
+	csc->vertex_bufs[0].buffer_offset = 0;
+	csc->vertex_bufs[0].buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_VERTEX,
+		sizeof(struct vlVertex2f) * 4
+	);
+
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, csc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		surface_verts,
+		sizeof(struct vlVertex2f) * 4
+	);
+
+	pipe->winsys->buffer_unmap(pipe->winsys, csc->vertex_bufs[0].buffer);
+
+	csc->vertex_elems[0].src_offset = 0;
+	csc->vertex_elems[0].vertex_buffer_index = 0;
+	csc->vertex_elems[0].nr_components = 2;
+	csc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/*
+	 * Create our texcoord buffer and texcoord buffer element
+	 * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices
+	 */
+	csc->vertex_bufs[1].pitch = sizeof(struct vlVertex2f);
+	csc->vertex_bufs[1].max_index = 3;
+	csc->vertex_bufs[1].buffer_offset = 0;
+	csc->vertex_bufs[1].buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_VERTEX,
+		sizeof(struct vlVertex2f) * 4
+	);
+
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, csc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		surface_texcoords,
+		sizeof(struct vlVertex2f) * 4
+	);
+
+	pipe->winsys->buffer_unmap(pipe->winsys, csc->vertex_bufs[1].buffer);
+
+	csc->vertex_elems[1].src_offset = 0;
+	csc->vertex_elems[1].vertex_buffer_index = 1;
+	csc->vertex_elems[1].nr_components = 2;
+	csc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/*
+	 * Create our vertex shader's constant buffer
+	 * Const buffer contains scaling and translation vectors
+	 */
+	csc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts);
+	csc->vs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		csc->vs_const_buf.size
+	);
+
+	/*
+	 * Create our fragment shader's constant buffer
+	 * Const buffer contains the color conversion matrix and bias vectors
+	 */
+	csc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts);
+	csc->fs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		csc->fs_const_buf.size
+	);
+
+	/*
+	 * TODO: Refactor this into a seperate function,
+	 * allow changing the CSC matrix at runtime to switch between regular & full versions
+	 */
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, csc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		&bt_601,
+		sizeof(struct vlFragmentShaderConsts)
+	);
+
+	pipe->winsys->buffer_unmap(pipe->winsys, csc->fs_const_buf.buffer);
+
+	return 0;
+}
+
+static int vlInit
+(
+	struct vlBasicCSC *csc
+)
+{
+	struct pipe_context		*pipe;
+	struct pipe_sampler_state	sampler;
+
+	assert(csc);
+
+	pipe = csc->pipe;
+
+	/* Delay creating the FB until vlPutPictureCSC() so we know window size */
+	csc->framebuffer_tex = NULL;
+	csc->framebuffer.width = 0;
+	csc->framebuffer.height = 0;
+	csc->framebuffer.num_cbufs = 1;
+	csc->framebuffer.cbufs[0] = NULL;
+	csc->framebuffer.zsbuf = NULL;
+
+	sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+	sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+	sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+	sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR;
+	sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+	sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR;
+	sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
+	sampler.compare_func = PIPE_FUNC_ALWAYS;
+	sampler.normalized_coords = 1;
+	/*sampler.prefilter = ;*/
+	/*sampler.shadow_ambient = ;*/
+	/*sampler.lod_bias = ;*/
+	/*sampler.min_lod = ;*/
+	/*sampler.max_lod = ;*/
+	/*sampler.border_color[i] = ;*/
+	/*sampler.max_anisotropy = ;*/
+	csc->sampler = pipe->create_sampler_state(pipe, &sampler);
+
+	vlCreateVertexShader(csc);
+	vlCreateFragmentShader(csc);
+	vlCreateDataBufs(csc);
+
+	return 0;
+}
+
+int vlCreateBasicCSC
+(
+	struct pipe_context *pipe,
+	struct vlCSC **csc
+)
+{
+	struct vlBasicCSC *basic_csc;
+
+	assert(pipe);
+	assert(csc);
+
+	basic_csc = calloc(1, sizeof(struct vlBasicCSC));
+
+	if (!basic_csc)
+		return 1;
+
+	basic_csc->base.vlResizeFrameBuffer = &vlResizeFrameBuffer;
+	basic_csc->base.vlBegin = &vlBegin;
+	basic_csc->base.vlPutPicture = &vlPutPictureCSC;
+	basic_csc->base.vlEnd = &vlEnd;
+	basic_csc->base.vlGetFrameBuffer = &vlGetFrameBuffer;
+	basic_csc->base.vlDestroy = &vlDestroy;
+	basic_csc->pipe = pipe;
+
+	vlInit(basic_csc);
+
+	*csc = &basic_csc->base;
+
+	return 0;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h
new file mode 100644
index 00000000000..2e17f1d814a
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h
@@ -0,0 +1,13 @@
+#ifndef vl_basic_csc_h
+#define vl_basic_csc_h
+
+struct pipe_context;
+struct vlCSC;
+
+int vlCreateBasicCSC
+(
+	struct pipe_context *pipe,
+	struct vlCSC **csc
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c
new file mode 100644
index 00000000000..fe107e406da
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_context.c
@@ -0,0 +1,210 @@
+#define VL_INTERNAL
+#include "vl_context.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <pipe/p_context.h>
+#include <pipe/p_state.h>
+#include "vl_render.h"
+#include "vl_r16snorm_mc.h"
+#include "vl_r16snorm_mc_buf.h"
+#include "vl_csc.h"
+#include "vl_basic_csc.h"
+
+static int vlInitCommon(struct vlContext *context)
+{
+	struct pipe_context			*pipe;
+	struct pipe_rasterizer_state		rast;
+	struct pipe_blend_state			blend;
+	struct pipe_depth_stencil_alpha_state	dsa;
+	unsigned int				i;
+
+	assert(context);
+
+	pipe = context->pipe;
+
+	rast.flatshade = 1;
+	rast.flatshade_first = 0;
+	rast.light_twoside = 0;
+	rast.front_winding = PIPE_WINDING_CCW;
+	rast.cull_mode = PIPE_WINDING_CW;
+	rast.fill_cw = PIPE_POLYGON_MODE_FILL;
+	rast.fill_ccw = PIPE_POLYGON_MODE_FILL;
+	rast.offset_cw = 0;
+	rast.offset_ccw = 0;
+	rast.scissor = 0;
+	rast.poly_smooth = 0;
+	rast.poly_stipple_enable = 0;
+	rast.point_sprite = 0;
+	rast.point_size_per_vertex = 0;
+	rast.multisample = 0;
+	rast.line_smooth = 0;
+	rast.line_stipple_enable = 0;
+	rast.line_stipple_factor = 0;
+	rast.line_stipple_pattern = 0;
+	rast.line_last_pixel = 0;
+	/* Don't need clipping, but viewport mapping done here */
+	rast.bypass_clipping = 0;
+	rast.bypass_vs = 0;
+	rast.origin_lower_left = 0;
+	rast.line_width = 1;
+	rast.point_smooth = 0;
+	rast.point_size = 1;
+	rast.offset_units = 1;
+	rast.offset_scale = 1;
+	/*rast.sprite_coord_mode[i] = ;*/
+	context->raster = pipe->create_rasterizer_state(pipe, &rast);
+	pipe->bind_rasterizer_state(pipe, context->raster);
+
+	blend.blend_enable = 0;
+	blend.rgb_func = PIPE_BLEND_ADD;
+	blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE;
+	blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE;
+	blend.alpha_func = PIPE_BLEND_ADD;
+	blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE;
+	blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE;
+	blend.logicop_enable = 0;
+	blend.logicop_func = PIPE_LOGICOP_CLEAR;
+	/* Needed to allow color writes to FB, even if blending disabled */
+	blend.colormask = PIPE_MASK_RGBA;
+	blend.dither = 0;
+	context->blend = pipe->create_blend_state(pipe, &blend);
+	pipe->bind_blend_state(pipe, context->blend);
+
+	dsa.depth.enabled = 0;
+	dsa.depth.writemask = 0;
+	dsa.depth.func = PIPE_FUNC_ALWAYS;
+	dsa.depth.occlusion_count = 0;
+	for (i = 0; i < 2; ++i)
+	{
+		dsa.stencil[i].enabled = 0;
+		dsa.stencil[i].func = PIPE_FUNC_ALWAYS;
+		dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP;
+		dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP;
+		dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP;
+		dsa.stencil[i].ref_value = 0;
+		dsa.stencil[i].value_mask = 0;
+		dsa.stencil[i].write_mask = 0;
+	}
+	dsa.alpha.enabled = 0;
+	dsa.alpha.func = PIPE_FUNC_ALWAYS;
+	dsa.alpha.ref = 0;
+	context->dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa);
+	pipe->bind_depth_stencil_alpha_state(pipe, context->dsa);
+
+	return 0;
+}
+
+int vlCreateContext
+(
+	struct vlScreen *screen,
+	struct pipe_context *pipe,
+	unsigned int picture_width,
+	unsigned int picture_height,
+	enum vlFormat picture_format,
+	enum vlProfile profile,
+	enum vlEntryPoint entry_point,
+	struct vlContext **context
+)
+{
+	struct vlContext *ctx;
+
+	assert(screen);
+	assert(context);
+	assert(pipe);
+
+	ctx = calloc(1, sizeof(struct vlContext));
+
+	if (!ctx)
+		return 1;
+
+	ctx->screen = screen;
+	ctx->pipe = pipe;
+	ctx->picture_width = picture_width;
+	ctx->picture_height = picture_height;
+	ctx->picture_format = picture_format;
+	ctx->profile = profile;
+	ctx->entry_point = entry_point;
+
+	vlInitCommon(ctx);
+
+	/*vlCreateR16SNormMC(pipe, picture_width, picture_height, picture_format, &ctx->render);*/
+	vlCreateR16SNormBufferedMC(pipe, picture_width, picture_height, picture_format, &ctx->render);
+	vlCreateBasicCSC(pipe, &ctx->csc);
+
+	*context = ctx;
+
+	return 0;
+}
+
+int vlDestroyContext
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	/* XXX: Must unbind shaders before we can delete them for some reason */
+	context->pipe->bind_vs_state(context->pipe, NULL);
+	context->pipe->bind_fs_state(context->pipe, NULL);
+
+	context->render->vlDestroy(context->render);
+	context->csc->vlDestroy(context->csc);
+
+	context->pipe->delete_blend_state(context->pipe, context->blend);
+	context->pipe->delete_rasterizer_state(context->pipe, context->raster);
+	context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->dsa);
+
+	free(context);
+
+	return 0;
+}
+
+struct vlScreen* vlContextGetScreen
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	return context->screen;
+}
+
+struct pipe_context* vlGetPipeContext
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	return context->pipe;
+}
+
+unsigned int vlGetPictureWidth
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	return context->picture_width;
+}
+
+unsigned int vlGetPictureHeight
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	return context->picture_height;
+}
+
+enum vlFormat vlGetPictureFormat
+(
+	struct vlContext *context
+)
+{
+	assert(context);
+
+	return context->picture_format;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h
new file mode 100644
index 00000000000..3d14634c44e
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_context.h
@@ -0,0 +1,73 @@
+#ifndef vl_context_h
+#define vl_context_h
+
+#include "vl_types.h"
+
+struct pipe_context;
+
+#ifdef VL_INTERNAL
+struct vlRender;
+struct vlCSC;
+
+struct vlContext
+{
+	struct vlScreen		*screen;
+	struct pipe_context	*pipe;
+	unsigned int		picture_width;
+	unsigned int		picture_height;
+	enum vlFormat		picture_format;
+	enum vlProfile		profile;
+	enum vlEntryPoint	entry_point;
+
+	void			*raster;
+	void			*dsa;
+	void			*blend;
+
+	struct vlRender		*render;
+	struct vlCSC		*csc;
+};
+#endif
+
+int vlCreateContext
+(
+	struct vlScreen *screen,
+	struct pipe_context *pipe,
+	unsigned int picture_width,
+	unsigned int picture_height,
+	enum vlFormat picture_format,
+	enum vlProfile profile,
+	enum vlEntryPoint entry_point,
+	struct vlContext **context
+);
+
+int vlDestroyContext
+(
+	struct vlContext *context
+);
+
+struct vlScreen* vlContextGetScreen
+(
+	struct vlContext *context
+);
+
+struct pipe_context* vlGetPipeContext
+(
+	struct vlContext *context
+);
+
+unsigned int vlGetPictureWidth
+(
+	struct vlContext *context
+);
+
+unsigned int vlGetPictureHeight
+(
+	struct vlContext *context
+);
+
+enum vlFormat vlGetPictureFormat
+(
+	struct vlContext *context
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_csc.h b/src/gallium/state_trackers/g3dvl/vl_csc.h
new file mode 100644
index 00000000000..36417a27929
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_csc.h
@@ -0,0 +1,53 @@
+#ifndef vl_csc_h
+#define vl_csc_h
+
+#include "vl_types.h"
+
+struct pipe_surface;
+
+struct vlCSC
+{
+	int (*vlResizeFrameBuffer)
+	(
+		struct vlCSC *csc,
+		unsigned int width,
+		unsigned int height
+	);
+
+	int (*vlBegin)
+	(
+		struct vlCSC *csc
+	);
+
+	int (*vlPutPicture)
+	(
+		struct vlCSC *csc,
+		struct vlSurface *surface,
+		int srcx,
+		int srcy,
+		int srcw,
+		int srch,
+		int destx,
+		int desty,
+		int destw,
+		int desth,
+		enum vlPictureType picture_type
+	);
+
+	int (*vlEnd)
+	(
+		struct vlCSC *csc
+	);
+
+	struct pipe_surface* (*vlGetFrameBuffer)
+	(
+		struct vlCSC *csc
+	);
+
+	int (*vlDestroy)
+	(
+		struct vlCSC *csc
+	);
+};
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_defs.h b/src/gallium/state_trackers/g3dvl/vl_defs.h
new file mode 100644
index 00000000000..d612d02502f
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_defs.h
@@ -0,0 +1,11 @@
+#ifndef vl_defs_h
+#define vl_defs_h
+
+#define VL_BLOCK_WIDTH		8
+#define VL_BLOCK_HEIGHT		8
+#define VL_BLOCK_SIZE		(VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT)
+#define VL_MACROBLOCK_WIDTH	16
+#define VL_MACROBLOCK_HEIGHT	16
+#define VL_MACROBLOCK_SIZE	(VL_MACROBLOCK_WIDTH * VL_MACROBLOCK_HEIGHT)
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_display.c b/src/gallium/state_trackers/g3dvl/vl_display.c
new file mode 100644
index 00000000000..af80faa7f52
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_display.c
@@ -0,0 +1,48 @@
+#define VL_INTERNAL
+#include "vl_display.h"
+#include <assert.h>
+#include <stdlib.h>
+
+int vlCreateDisplay
+(
+	vlNativeDisplay native_display,
+	struct vlDisplay **display
+)
+{
+	struct vlDisplay *dpy;
+
+	assert(native_display);
+	assert(display);
+
+	dpy = calloc(1, sizeof(struct vlDisplay));
+
+	if (!dpy)
+		return 1;
+
+	dpy->native = native_display;
+	*display = dpy;
+
+	return 0;
+}
+
+int vlDestroyDisplay
+(
+	struct vlDisplay *display
+)
+{
+	assert(display);
+
+	free(display);
+
+	return 0;
+}
+
+vlNativeDisplay vlGetNativeDisplay
+(
+	struct vlDisplay *display
+)
+{
+	assert(display);
+
+	return display->native;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_display.h b/src/gallium/state_trackers/g3dvl/vl_display.h
new file mode 100644
index 00000000000..e11fd407993
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_display.h
@@ -0,0 +1,29 @@
+#ifndef vl_display_h
+#define vl_display_h
+
+#include "vl_types.h"
+
+#ifdef VL_INTERNAL
+struct vlDisplay
+{
+	vlNativeDisplay native;
+};
+#endif
+
+int vlCreateDisplay
+(
+	vlNativeDisplay native_display,
+	struct vlDisplay **display
+);
+
+int vlDestroyDisplay
+(
+	struct vlDisplay *display
+);
+
+vlNativeDisplay vlGetNativeDisplay
+(
+	struct vlDisplay *display
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c
new file mode 100644
index 00000000000..3272220ef83
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.c
@@ -0,0 +1,2344 @@
+#define VL_INTERNAL
+#include "vl_r16snorm_mc.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <pipe/p_context.h>
+#include <pipe/p_winsys.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+#include "vl_render.h"
+#include "vl_shader_build.h"
+#include "vl_surface.h"
+#include "vl_util.h"
+#include "vl_types.h"
+#include "vl_defs.h"
+
+#define NUM_BUFS 4	/* Number of rotating buffers to use */
+
+struct vlVertexShaderConsts
+{
+	/*struct vlVertex4f scale;
+	struct vlVertex4f denorm;*/
+	struct vlVertex4f	scale;
+	struct vlVertex4f	mb_pos_trans;
+	struct vlVertex4f	denorm;
+	struct
+	{
+		struct vlVertex4f	top_field;
+		struct vlVertex4f	bottom_field;
+	} mb_tc_trans[2];
+};
+
+struct vlFragmentShaderConsts
+{
+	struct vlVertex4f multiplier;
+	struct vlVertex4f div;
+};
+
+struct vlR16SnormMC
+{
+	struct vlRender				base;
+
+	unsigned int				video_width, video_height;
+	enum vlFormat				video_format;
+	unsigned int				cur_buf;
+
+	struct pipe_context			*pipe;
+	struct pipe_viewport_state		viewport;
+	struct pipe_framebuffer_state		render_target;
+	struct pipe_sampler_state		*samplers[5];
+	struct pipe_texture			*textures[NUM_BUFS][5];
+	void					*i_vs, *p_vs[2], *b_vs[2];
+	void					*i_fs, *p_fs[2], *b_fs[2];
+	struct pipe_vertex_buffer 		vertex_bufs[3];
+	struct pipe_vertex_element		vertex_elems[3];
+	struct pipe_constant_buffer		vs_const_buf, fs_const_buf;
+};
+
+static int vlBegin
+(
+	struct vlRender *render
+)
+{
+	struct vlR16SnormMC	*mc;
+	struct pipe_context	*pipe;
+
+	assert(render);
+
+	mc = (struct vlR16SnormMC*)render;
+	pipe = mc->pipe;
+
+	/* Frame buffer set in vlRender*Macroblock() */
+	/* Shaders, samplers, textures set in vlRender*Macroblock() */
+	pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs);
+	pipe->set_vertex_elements(pipe, 3, mc->vertex_elems);
+	pipe->set_viewport_state(pipe, &mc->viewport);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf);
+
+	return 0;
+}
+
+/*static int vlGrabMacroBlock
+(
+	struct vlR16SnormMC *mc,
+	struct vlMpeg2MacroBlock *macroblock
+)
+{
+	assert(mc);
+	assert(macroblock);
+
+
+
+	return 0;
+}*/
+
+/*#define DO_IDCT*/
+
+#ifdef DO_IDCT
+static int vlTransformBlock(short *src, short *dst, short bias)
+{
+	static const float basis[8][8] =
+	{
+		{0.3536,   0.4904,   0.4619,   0.4157,   0.3536,   0.2778,   0.1913,   0.0975},
+		{0.3536,   0.4157,   0.1913,  -0.0975,  -0.3536,  -0.4904,  -0.4619,  -0.2778},
+		{0.3536,   0.2778,  -0.1913,  -0.4904,  -0.3536,   0.0975,   0.4619,   0.4157},
+		{0.3536,   0.0975,  -0.4619,  -0.2778,   0.3536,   0.4157,  -0.1913,  -0.4904},
+		{0.3536,  -0.0975,  -0.4619,   0.2778,   0.3536,  -0.4157,  -0.1913,   0.4904},
+		{0.3536,  -0.2778,  -0.1913,   0.4904,  -0.3536,  -0.0975,   0.4619,  -0.4157},
+		{0.3536,  -0.4157,   0.1913,   0.0975,  -0.3536,   0.4904,  -0.4619,   0.2778},
+		{0.3536,  -0.4904,   0.4619,  -0.4157,   0.3536,  -0.2778,   0.1913,  -0.0975}
+	};
+
+	unsigned int	x, y;
+	short		tmp[64];
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		for (x = 0; x < VL_BLOCK_WIDTH; ++x)
+			tmp[y * VL_BLOCK_WIDTH + x] = (short)
+			(
+				src[y * VL_BLOCK_WIDTH + 0] * basis[x][0] +
+				src[y * VL_BLOCK_WIDTH + 1] * basis[x][1] +
+				src[y * VL_BLOCK_WIDTH + 2] * basis[x][2] +
+				src[y * VL_BLOCK_WIDTH + 3] * basis[x][3] +
+				src[y * VL_BLOCK_WIDTH + 4] * basis[x][4] +
+				src[y * VL_BLOCK_WIDTH + 5] * basis[x][5] +
+				src[y * VL_BLOCK_WIDTH + 6] * basis[x][6] +
+				src[y * VL_BLOCK_WIDTH + 7] * basis[x][7]
+			);
+
+	for (x = 0; x < VL_BLOCK_WIDTH; ++x)
+		for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		{
+			dst[y * VL_BLOCK_WIDTH + x] = bias + (short)
+			(
+				tmp[0 * VL_BLOCK_WIDTH + x] * basis[y][0] +
+				tmp[1 * VL_BLOCK_WIDTH + x] * basis[y][1] +
+				tmp[2 * VL_BLOCK_WIDTH + x] * basis[y][2] +
+				tmp[3 * VL_BLOCK_WIDTH + x] * basis[y][3] +
+				tmp[4 * VL_BLOCK_WIDTH + x] * basis[y][4] +
+				tmp[5 * VL_BLOCK_WIDTH + x] * basis[y][5] +
+				tmp[6 * VL_BLOCK_WIDTH + x] * basis[y][6] +
+				tmp[7 * VL_BLOCK_WIDTH + x] * basis[y][7]
+			);
+			if (dst[y * VL_BLOCK_WIDTH + x] > 255)
+				dst[y * VL_BLOCK_WIDTH + x] = 255;
+			else if (bias > 0 && dst[y * VL_BLOCK_WIDTH + x] < 0)
+				dst[y * VL_BLOCK_WIDTH + x] = 0;
+		}
+	return 0;
+}
+#endif
+
+static int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		memcpy
+		(
+			dst + y * dst_pitch,
+			src + y * VL_BLOCK_WIDTH,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+static int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT / 2; ++y)
+		memcpy
+		(
+			dst + y * dst_pitch * 2,
+			src + y * VL_BLOCK_WIDTH,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	dst += VL_BLOCK_HEIGHT * dst_pitch;
+
+	for (; y < VL_BLOCK_HEIGHT; ++y)
+		memcpy
+		(
+			dst + y * dst_pitch * 2,
+			src + y * VL_BLOCK_WIDTH,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+static int vlGrabNoBlock(short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		memset
+		(
+			dst + y * dst_pitch,
+			0,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+enum vlSampleType
+{
+	vlSampleTypeFull,
+	vlSampleTypeDiff
+};
+
+static int vlGrabBlocks
+(
+	struct vlR16SnormMC *mc,
+	unsigned int coded_block_pattern,
+	enum vlDCTType dct_type,
+	enum vlSampleType sample_type,
+	short *blocks
+)
+{
+	struct pipe_surface	*tex_surface;
+	short			*texels;
+	unsigned int		tex_pitch;
+	unsigned int		tb, sb = 0;
+
+	assert(mc);
+	assert(blocks);
+
+	tex_surface = mc->pipe->screen->get_tex_surface
+	(
+		mc->pipe->screen,
+		mc->textures[mc->cur_buf % NUM_BUFS][0],
+		0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE);
+	tex_pitch = tex_surface->stride / tex_surface->block.size;
+
+	for (tb = 0; tb < 4; ++tb)
+	{
+		if ((coded_block_pattern >> (5 - tb)) & 1)
+		{
+			short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
+
+#ifdef DO_IDCT
+			vlTransformBlock(cur_block, cur_block, sample_type == vlSampleTypeFull ? 128 : 0);
+#endif
+
+			if (dct_type == vlDCTTypeFrameCoded)
+				vlGrabFrameCodedBlock
+				(
+					cur_block,
+					texels + tb * tex_pitch * VL_BLOCK_HEIGHT,
+					tex_pitch
+				);
+			else
+				vlGrabFieldCodedBlock
+				(
+					cur_block,
+					texels + (tb % 2) * tex_pitch * VL_BLOCK_HEIGHT + (tb / 2) * tex_pitch,
+					tex_pitch
+				);
+
+			++sb;
+		}
+		else
+			vlGrabNoBlock(texels + tb * tex_pitch * VL_BLOCK_HEIGHT, tex_pitch);
+	}
+
+	pipe_surface_unmap(tex_surface);
+
+	/* TODO: Implement 422, 444 */
+	for (tb = 0; tb < 2; ++tb)
+	{
+		tex_surface = mc->pipe->screen->get_tex_surface
+		(
+			mc->pipe->screen,
+			mc->textures[mc->cur_buf % NUM_BUFS][tb + 1],
+			0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE
+		);
+
+		texels = pipe_surface_map(tex_surface, PIPE_BUFFER_USAGE_CPU_WRITE);
+		tex_pitch = tex_surface->stride / tex_surface->block.size;
+
+		if ((coded_block_pattern >> (1 - tb)) & 1)
+		{
+			short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
+
+#ifdef DO_IDCT
+			vlTransformBlock(cur_block, cur_block, sample_type == vlSampleTypeFull ? 128 : 0);
+#endif
+
+			vlGrabFrameCodedBlock
+			(
+				cur_block,
+				texels,
+				tex_pitch
+			);
+
+			++sb;
+		}
+		else
+			vlGrabNoBlock(texels, tex_pitch);
+
+		pipe_surface_unmap(tex_surface);
+	}
+
+	return 0;
+}
+
+static int vlRenderIMacroBlock
+(
+	struct vlR16SnormMC *mc,
+	enum vlPictureType picture_type,
+	enum vlFieldOrder field_order,
+	unsigned int mbx,
+	unsigned int mby,
+	unsigned int coded_block_pattern,
+	enum vlDCTType dct_type,
+	short *blocks,
+	struct vlSurface *surface
+)
+{
+	struct pipe_context		*pipe;
+	struct vlVertexShaderConsts	*vs_consts;
+
+	assert(blocks);
+	assert(surface);
+
+	/* TODO: Implement interlaced rendering */
+	if (picture_type != vlPictureTypeFrame)
+		return 0;
+
+	vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeFull, blocks);
+
+	pipe = mc->pipe;
+
+	vs_consts = pipe->winsys->buffer_map
+	(
+		pipe->winsys,
+		mc->vs_const_buf.buffer,
+		PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0];
+	vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0];
+	vs_consts->scale.z = 1.0f;
+	vs_consts->scale.w = 1.0f;
+	vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0];
+	vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0];
+	vs_consts->mb_pos_trans.z = 0.0f;
+	vs_consts->mb_pos_trans.w = 0.0f;
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer);
+
+	mc->render_target.cbufs[0] = pipe->screen->get_tex_surface
+	(
+		pipe->screen,
+		surface->texture,
+		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+	);
+	pipe->set_framebuffer_state(pipe, &mc->render_target);
+	pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUFS]);
+	pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers);
+	pipe->bind_vs_state(pipe, mc->i_vs);
+	pipe->bind_fs_state(pipe, mc->i_fs);
+
+	pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24);
+
+	mc->cur_buf++;
+
+	return 0;
+}
+
+static int vlRenderPMacroBlock
+(
+	struct vlR16SnormMC *mc,
+	enum vlPictureType picture_type,
+	enum vlFieldOrder field_order,
+	unsigned int mbx,
+	unsigned int mby,
+	enum vlMotionType mc_type,
+	short top_x,
+	short top_y,
+	short bottom_x,
+	short bottom_y,
+	unsigned int coded_block_pattern,
+	enum vlDCTType dct_type,
+	short *blocks,
+	struct vlSurface *ref_surface,
+	struct vlSurface *surface
+)
+{
+	struct pipe_context		*pipe;
+	struct vlVertexShaderConsts	*vs_consts;
+
+	assert(motion_vectors);
+	assert(blocks);
+	assert(ref_surface);
+	assert(surface);
+
+	/* TODO: Implement interlaced rendering */
+	if (picture_type != vlPictureTypeFrame)
+		return 0;
+	/* TODO: Implement other MC types */
+	if (mc_type != vlMotionTypeFrame && mc_type != vlMotionTypeField)
+		return 0;
+
+	vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeDiff, blocks);
+
+	pipe = mc->pipe;
+
+	vs_consts = pipe->winsys->buffer_map
+	(
+		pipe->winsys,
+		mc->vs_const_buf.buffer,
+		PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0];
+	vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0];
+	vs_consts->scale.z = 1.0f;
+	vs_consts->scale.w = 1.0f;
+	vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0];
+	vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0];
+	vs_consts->mb_pos_trans.z = 0.0f;
+	vs_consts->mb_pos_trans.w = 0.0f;
+	vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_x * 0.5f) / (float)surface->texture->width[0];
+	vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_y * 0.5f) / (float)surface->texture->height[0];
+	vs_consts->mb_tc_trans[0].top_field.z = 0.0f;
+	vs_consts->mb_tc_trans[0].top_field.w = 0.0f;
+
+	if (mc_type == vlMotionTypeField)
+	{
+		vs_consts->denorm.x = (float)surface->texture->width[0];
+		vs_consts->denorm.y = (float)surface->texture->height[0];
+
+		vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_x * 0.5f) / (float)surface->texture->width[0];
+		vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_y * 0.5f) / (float)surface->texture->height[0];
+		vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f;
+		vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f;
+
+		pipe->bind_vs_state(pipe, mc->p_vs[1]);
+		pipe->bind_fs_state(pipe, mc->p_fs[1]);
+	}
+	else
+	{
+		pipe->bind_vs_state(pipe, mc->p_vs[0]);
+		pipe->bind_fs_state(pipe, mc->p_fs[0]);
+	}
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer);
+
+	mc->render_target.cbufs[0] = pipe->screen->get_tex_surface
+	(
+		pipe->screen,
+		surface->texture,
+		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+	);
+	pipe->set_framebuffer_state(pipe, &mc->render_target);
+
+	mc->textures[mc->cur_buf % NUM_BUFS][3] = ref_surface->texture;
+	pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUFS]);
+	pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers);
+
+	pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24);
+
+	mc->cur_buf++;
+
+	return 0;
+}
+
+static int vlRenderBMacroBlock
+(
+	struct vlR16SnormMC *mc,
+	enum vlPictureType picture_type,
+	enum vlFieldOrder field_order,
+	unsigned int mbx,
+	unsigned int mby,
+	enum vlMotionType mc_type,
+	short top_past_x,
+	short top_past_y,
+	short bottom_past_x,
+	short bottom_past_y,
+	short top_future_x,
+	short top_future_y,
+	short bottom_future_x,
+	short bottom_future_y,
+	unsigned int coded_block_pattern,
+	enum vlDCTType dct_type,
+	short *blocks,
+	struct vlSurface *past_surface,
+	struct vlSurface *future_surface,
+	struct vlSurface *surface
+)
+{
+	struct pipe_context		*pipe;
+	struct vlVertexShaderConsts	*vs_consts;
+
+	assert(motion_vectors);
+	assert(blocks);
+	assert(ref_surface);
+	assert(surface);
+
+	/* TODO: Implement interlaced rendering */
+	if (picture_type != vlPictureTypeFrame)
+		return 0;
+	/* TODO: Implement other MC types */
+	if (mc_type != vlMotionTypeFrame && mc_type != vlMotionTypeField)
+		return 0;
+
+	vlGrabBlocks(mc, coded_block_pattern, dct_type, vlSampleTypeDiff, blocks);
+
+	pipe = mc->pipe;
+
+	vs_consts = pipe->winsys->buffer_map
+	(
+		pipe->winsys,
+		mc->vs_const_buf.buffer,
+		PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	vs_consts->scale.x = VL_MACROBLOCK_WIDTH / (float)surface->texture->width[0];
+	vs_consts->scale.y = VL_MACROBLOCK_HEIGHT / (float)surface->texture->height[0];
+	vs_consts->scale.z = 1.0f;
+	vs_consts->scale.w = 1.0f;
+	vs_consts->mb_pos_trans.x = (mbx * VL_MACROBLOCK_WIDTH) / (float)surface->texture->width[0];
+	vs_consts->mb_pos_trans.y = (mby * VL_MACROBLOCK_HEIGHT) / (float)surface->texture->height[0];
+	vs_consts->mb_pos_trans.z = 0.0f;
+	vs_consts->mb_pos_trans.w = 0.0f;
+	vs_consts->mb_tc_trans[0].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_past_x * 0.5f) / (float)surface->texture->width[0];
+	vs_consts->mb_tc_trans[0].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_past_y * 0.5f) / (float)surface->texture->height[0];
+	vs_consts->mb_tc_trans[0].top_field.z = 0.0f;
+	vs_consts->mb_tc_trans[0].top_field.w = 0.0f;
+	vs_consts->mb_tc_trans[1].top_field.x = (mbx * VL_MACROBLOCK_WIDTH + top_future_x * 0.5f) / (float)surface->texture->width[0];
+	vs_consts->mb_tc_trans[1].top_field.y = (mby * VL_MACROBLOCK_HEIGHT + top_future_y * 0.5f) / (float)surface->texture->height[0];
+	vs_consts->mb_tc_trans[1].top_field.z = 0.0f;
+	vs_consts->mb_tc_trans[1].top_field.w = 0.0f;
+
+	if (mc_type == vlMotionTypeField)
+	{
+		vs_consts->denorm.x = (float)surface->texture->width[0];
+		vs_consts->denorm.y = (float)surface->texture->height[0];
+
+		vs_consts->mb_tc_trans[0].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_past_x * 0.5f) / (float)surface->texture->width[0];
+		vs_consts->mb_tc_trans[0].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_past_y * 0.5f) / (float)surface->texture->height[0];
+		vs_consts->mb_tc_trans[0].bottom_field.z = 0.0f;
+		vs_consts->mb_tc_trans[0].bottom_field.w = 0.0f;
+		vs_consts->mb_tc_trans[1].bottom_field.x = (mbx * VL_MACROBLOCK_WIDTH + bottom_future_x * 0.5f) / (float)surface->texture->width[0];
+		vs_consts->mb_tc_trans[1].bottom_field.y = (mby * VL_MACROBLOCK_HEIGHT + bottom_future_y * 0.5f) / (float)surface->texture->height[0];
+		vs_consts->mb_tc_trans[1].bottom_field.z = 0.0f;
+		vs_consts->mb_tc_trans[1].bottom_field.w = 0.0f;
+
+		pipe->bind_vs_state(pipe, mc->b_vs[1]);
+		pipe->bind_fs_state(pipe, mc->b_fs[1]);
+	}
+	else
+	{
+		pipe->bind_vs_state(pipe, mc->b_vs[0]);
+		pipe->bind_fs_state(pipe, mc->b_fs[0]);
+	}
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer);
+
+	mc->render_target.cbufs[0] = pipe->screen->get_tex_surface
+	(
+		pipe->screen,
+		surface->texture,
+		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+	);
+	pipe->set_framebuffer_state(pipe, &mc->render_target);
+
+	mc->textures[mc->cur_buf % NUM_BUFS][3] = past_surface->texture;
+	mc->textures[mc->cur_buf % NUM_BUFS][4] = future_surface->texture;
+	pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUFS]);
+	pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers);
+
+	pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, 0, 24);
+
+	mc->cur_buf++;
+
+	return 0;
+}
+
+static int vlRenderMacroBlocksMpeg2R16Snorm
+(
+	struct vlRender *render,
+	struct vlMpeg2MacroBlockBatch *batch,
+	struct vlSurface *surface
+)
+{
+	struct vlR16SnormMC	*mc;
+	unsigned int		i;
+
+	assert(render);
+
+	mc = (struct vlR16SnormMC*)render;
+
+	/*for (i = 0; i < batch->num_macroblocks; ++i)
+		vlGrabMacroBlock(batch->macroblocks[i]);*/
+
+	for (i = 0; i < batch->num_macroblocks; ++i)
+	{
+		switch (batch->macroblocks[i].mb_type)
+		{
+			case vlMacroBlockTypeIntra:
+			{
+				vlRenderIMacroBlock
+				(
+					mc,
+					batch->picture_type,
+					batch->field_order,
+					batch->macroblocks[i].mbx,
+					batch->macroblocks[i].mby,
+					batch->macroblocks[i].cbp,
+					batch->macroblocks[i].dct_type,
+					batch->macroblocks[i].blocks,
+					surface
+				);
+				break;
+			}
+			case vlMacroBlockTypeFwdPredicted:
+			{
+				vlRenderPMacroBlock
+				(
+					mc,
+					batch->picture_type,
+					batch->field_order,
+					batch->macroblocks[i].mbx,
+					batch->macroblocks[i].mby,
+					batch->macroblocks[i].mo_type,
+					batch->macroblocks[i].PMV[0][0][0],
+					batch->macroblocks[i].PMV[0][0][1],
+					batch->macroblocks[i].PMV[1][0][0],
+					batch->macroblocks[i].PMV[1][0][1],
+					batch->macroblocks[i].cbp,
+					batch->macroblocks[i].dct_type,
+					batch->macroblocks[i].blocks,
+					batch->past_surface,
+					surface
+				);
+				break;
+			}
+			case vlMacroBlockTypeBkwdPredicted:
+			{
+				vlRenderPMacroBlock
+				(
+					mc,
+					batch->picture_type,
+					batch->field_order,
+					batch->macroblocks[i].mbx,
+					batch->macroblocks[i].mby,
+					batch->macroblocks[i].mo_type,
+					batch->macroblocks[i].PMV[0][1][0],
+					batch->macroblocks[i].PMV[0][1][1],
+					batch->macroblocks[i].PMV[1][1][0],
+					batch->macroblocks[i].PMV[1][1][1],
+					batch->macroblocks[i].cbp,
+					batch->macroblocks[i].dct_type,
+					batch->macroblocks[i].blocks,
+					batch->future_surface,
+					surface
+				);
+				break;
+			}
+			case vlMacroBlockTypeBiPredicted:
+			{
+				vlRenderBMacroBlock
+				(
+					mc,
+					batch->picture_type,
+					batch->field_order,
+					batch->macroblocks[i].mbx,
+					batch->macroblocks[i].mby,
+					batch->macroblocks[i].mo_type,
+					batch->macroblocks[i].PMV[0][0][0],
+					batch->macroblocks[i].PMV[0][0][1],
+					batch->macroblocks[i].PMV[1][0][0],
+					batch->macroblocks[i].PMV[1][0][1],
+					batch->macroblocks[i].PMV[0][1][0],
+					batch->macroblocks[i].PMV[0][1][1],
+					batch->macroblocks[i].PMV[1][1][0],
+					batch->macroblocks[i].PMV[1][1][1],
+					batch->macroblocks[i].cbp,
+					batch->macroblocks[i].dct_type,
+					batch->macroblocks[i].blocks,
+					batch->past_surface,
+					batch->future_surface,
+					surface
+				);
+				break;
+			}
+			default:
+				assert(0);
+		}
+	}
+
+	return 0;
+}
+
+static int vlEnd
+(
+	struct vlRender *render
+)
+{
+	assert(render);
+
+	return 0;
+}
+
+static int vlFlush
+(
+	struct vlRender *render
+)
+{
+	assert(render);
+
+	return 0;
+}
+
+static int vlDestroy
+(
+	struct vlRender *render
+)
+{
+	struct vlR16SnormMC	*mc;
+	struct pipe_context	*pipe;
+	unsigned int		i;
+
+	assert(render);
+
+	mc = (struct vlR16SnormMC*)render;
+	pipe = mc->pipe;
+
+	for (i = 0; i < 5; ++i)
+		pipe->delete_sampler_state(pipe, mc->samplers[i]);
+
+	for (i = 0; i < 3; ++i)
+		pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[i].buffer);
+
+	/* Textures 3 & 4 are not created directly, no need to release them here */
+	for (i = 0; i < NUM_BUFS; ++i)
+	{
+		pipe_texture_release(&mc->textures[i][0]);
+		pipe_texture_release(&mc->textures[i][1]);
+		pipe_texture_release(&mc->textures[i][2]);
+	}
+
+	pipe->delete_vs_state(pipe, mc->i_vs);
+	pipe->delete_fs_state(pipe, mc->i_fs);
+
+	for (i = 0; i < 2; ++i)
+	{
+		pipe->delete_vs_state(pipe, mc->p_vs[i]);
+		pipe->delete_fs_state(pipe, mc->p_fs[i]);
+		pipe->delete_vs_state(pipe, mc->b_vs[i]);
+		pipe->delete_fs_state(pipe, mc->b_fs[i]);
+	}
+
+	pipe->winsys->buffer_destroy(pipe->winsys, mc->vs_const_buf.buffer);
+	pipe->winsys->buffer_destroy(pipe->winsys, mc->fs_const_buf.buffer);
+
+	free(mc);
+
+	return 0;
+}
+
+/*
+ * Represents 8 triangles (4 quads, 1 per block) in noormalized coords
+ * that render a macroblock.
+ * Need to be scaled to cover mbW*mbH macroblock pixels and translated into
+ * position on target surface.
+ */
+static const struct vlVertex2f macroblock_verts[24] =
+{
+	{0.0f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.0f},
+	{0.5f, 0.0f}, {0.0f, 0.5f}, {0.5f, 0.5f},
+
+	{0.5f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.0f},
+	{1.0f, 0.0f}, {0.5f, 0.5f}, {1.0f, 0.5f},
+
+	{0.0f, 0.5f}, {0.0f, 1.0f}, {0.5f, 0.5f},
+	{0.5f, 0.5f}, {0.0f, 1.0f}, {0.5f, 1.0f},
+
+	{0.5f, 0.5f}, {0.5f, 1.0f}, {1.0f, 0.5f},
+	{1.0f, 0.5f}, {0.5f, 1.0f}, {1.0f, 1.0f}
+};
+
+/*
+ * Represents texcoords for the above for rendering 4 luma blocks arranged
+ * in a bW*(bH*4) texture. First luma block located at 0,0->bW,bH; second at
+ * 0,bH->bW,2bH; third at 0,2bH->bW,3bH; fourth at 0,3bH->bW,4bH.
+ */
+static const struct vlVertex2f macroblock_luma_texcoords[24] =
+{
+	{0.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.0f},
+	{1.0f, 0.0f}, {0.0f, 0.25f}, {1.0f, 0.25f},
+
+	{0.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.25f},
+	{1.0f, 0.25f}, {0.0f, 0.5f}, {1.0f, 0.5f},
+
+	{0.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.5f},
+	{1.0f, 0.5f}, {0.0f, 0.75f}, {1.0f, 0.75f},
+
+	{0.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 0.75f},
+	{1.0f, 0.75f}, {0.0f, 1.0f}, {1.0f, 1.0f}
+};
+
+/*
+ * Represents texcoords for the above for rendering 1 chroma block.
+ * Straight forward 0,0->1,1 mapping so we can reuse the MB pos vectors.
+ */
+static const struct vlVertex2f *macroblock_chroma_420_texcoords = macroblock_verts;
+
+/*
+ * Represents texcoords for the above for rendering 2 chroma blocks arranged
+ * in a bW*(bH*2) texture. First chroma block located at 0,0->bW,bH; second at
+ * 0,bH->bW,2bH. We can render this with 0,0->1,1 mapping.
+ * Straight forward 0,0->1,1 mapping so we can reuse MB pos vectors.
+ */
+static const struct vlVertex2f *macroblock_chroma_422_texcoords = macroblock_verts;
+
+/*
+ * Represents texcoords for the above for rendering 4 chroma blocks.
+ * Same case as 4 luma blocks.
+ */
+static const struct vlVertex2f *macroblock_chroma_444_texcoords = macroblock_luma_texcoords;
+
+/*
+ * Used when rendering P and B macroblocks, multiplier is applied to the A channel,
+ * which is then added to the L channel, then the bias is subtracted from that to
+ * get back the differential. The differential is then added to the samples from the
+ * reference surface(s).
+ */
+static const struct vlFragmentShaderConsts fs_consts =
+{
+	{32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f},
+	{0.5f, 2.0f, 0.0f, 0.0f}
+};
+
+static int vlCreateVertexShaderIMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 50;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale unit rect to macroblock size
+	 * decl c1		; Translation vector to move macroblock into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0 */
+	decl = vl_decl_temps(0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t0, i0, c0	; Scale unit rect to normalized MB size */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, c1	; Translate rect into position */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma texcoords to output
+	 */
+	for (i = 1; i < 3; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->i_vs = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderIMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Texcoords for s0
+	 * decl i1			; Texcoords for s1, s2
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i1, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	}
+
+	/* mul o0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->i_fs = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFramePMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale unit rect to macroblock size
+	 * decl c1		; Translation vector to move macroblock into position
+	 * decl c2		; Unused
+	 * decl c3		; Translation vector to move ref macroblock texcoords into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma texcoords
+	 * decl o3		; Ref macroblock texcoords
+	 */
+	for (i = 0; i < 4; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0 */
+	decl = vl_decl_temps(0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t0, i0, c0	; Scale unit rect to normalized MB size */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, c1	; Translate rect into position */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma texcoords to output
+	 */
+	for (i = 1; i < 3; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* add o3, t0, c3	; Translate rect into position on ref macroblock */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->p_vs[0] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFieldPMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration
+		(
+			&decl,
+			&tokens[ti],
+			header,
+			max_tokens - ti
+		);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale unit rect to macroblock size
+	 * decl c1		; Translation vector to move macroblock into position
+	 * decl c2		; Denorm coefficients
+	 * decl c3		; Translation vector to move top field ref macroblock texcoords into position
+	 * decl c4		; Translation vector to move bottom field ref macroblock texcoords into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma texcoords
+	 * decl o3		; Top field ref macroblock texcoords
+	 * decl o4		; Bottom field ref macroblock texcoords
+	 * decl o5		; Denormalized vertex pos
+	 */
+	for (i = 0; i < 6; i++)
+	{
+		decl = vl_decl_output((i == 0 || i == 5) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t0, i0, c0	; Scale unit rect to normalized MB size */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add t1, t0, c1	; Translate rect into position */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mov o0, t1		; Move vertex pos to output */
+	inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	mov o1, i1		; Move input luma texcoords to output
+	mov o2, i2		; Move input chroma texcoords to output
+	*/
+	for (i = 1; i < 3; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* add o3, t0, c3	; Translate top field rect into position on ref macroblock
+	   add o4, t0, c4	; Translate bottom field rect into position on ref macroblock */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul o5, t1, c2	; Denorm vertex pos */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 5, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->p_vs[1] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFramePMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Texcoords for s0
+	 * decl i1			; Texcoords for s1, s2
+	 * decl i2			; Texcoords for s3
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for ref surface texture
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i1, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* tex2d t1, i2, s3		; Read texel from ref macroblock */
+	inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 2, TGSI_FILE_SAMPLER, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->p_fs[0] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFieldPMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 200;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Texcoords for s0
+	 * decl i1			; Texcoords for s1, s2
+	 * decl i2			; Texcoords for s3
+	 * decl i3			; Texcoords for s3
+	 * decl i4			; Denormalized vertex pos
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t4 */
+	decl = vl_decl_temps(0, 4);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for ref surface texture
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i1, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i2, s3		; Read texel from ref macroblock top field
+	 * tex2d t2, i3, s3		; Read texel from ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* XXX: Pos values off by 0.5? */
+	/* sub t4, i4.y, c1.x		; Sub 0.5 from denormalized pos */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 4, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* floor t3, t3			; Get rid of fractional part */
+	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t3, c1.y		; Multiply by 2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->p_fs[1] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFrameBMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale unit rect to macroblock size
+	 * decl c1		; Translation vector to move macroblock into position
+	 * decl c2		; Unused
+	 * decl c3		; Translation vector to move past ref macroblock texcoords into position
+	 * decl c4		; Unused
+	 * decl c5		; Translation vector to move future ref macroblock texcoords into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 5);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma texcoords
+	 * decl o3		; Past ref macroblock texcoords
+	 * decl o4		; Future ref macroblock texcoords
+	 */
+	for (i = 0; i < 5; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0 */
+	decl = vl_decl_temps(0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t0, i0, c0	; Scale unit rect to normalized MB size */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, c1	; Translate rect into position */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma texcoords to output
+	 */
+	for (i = 1; i < 3; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* add o3, t0, c3	; Translate rect into position on past ref macroblock
+	   add o4, t0, c5	; Translate rect into position on future ref macroblock */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i * 2 + 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->b_vs[0] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFieldBMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma texcoords
+	 */
+	for (i = 0; i < 3; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0		; Scaling vector to scale unit rect to macroblock size
+	 * decl c1		; Translation vector to move macroblock into position
+	 * decl c2		; Denorm coefficients
+	 * decl c3		; Translation vector to move top field past ref macroblock texcoords into position
+	 * decl c4		; Translation vector to move bottom field past ref macroblock texcoords into position
+	 * decl c5		; Translation vector to move top field future ref macroblock texcoords into position
+	 * decl c6		; Translation vector to move bottom field future ref macroblock texcoords into position
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 6);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma texcoords
+	 * decl o3		; Top field past ref macroblock texcoords
+	 * decl o4		; Bottom field past ref macroblock texcoords
+	 * decl o5		; Top field future ref macroblock texcoords
+	 * decl o6		; Bottom field future ref macroblock texcoords
+	 * decl o7		; Denormalized vertex pos
+	 */
+	for (i = 0; i < 8; i++)
+	{
+		decl = vl_decl_output((i == 0 || i == 7) ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t0, i0, c0	; Scale unit rect to normalized MB size */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add t1, t0, c1	; Translate rect into position */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mov o0, t1		; Move vertex pos to output */
+	inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma texcoords to output
+	 */
+	for (i = 1; i < 3; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * add o3, t0, c3	; Translate top field rect into position on past ref macroblock
+	 * add o4, t0, c4	; Translate bottom field rect into position on past ref macroblock
+	 * add o5, t0, c5	; Translate top field rect into position on future ref macroblock
+	 * add o6, t0, c6	; Translate bottom field rect into position on future ref macroblock
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 3, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul o7, t1, c2	; Denorm vertex pos */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 7, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->b_vs[1] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFrameBMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Texcoords for s0
+	 * decl i1			; Texcoords for s1, s2
+	 * decl i2			; Texcoords for s3
+	 * decl i3			; Texcoords for s4
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constant 1/2 in .x channel to use as weight to blend past and future texels
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t2 */
+	decl = vl_decl_temps(0, 2);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for past ref surface texture
+	 * decl s4			; Sampler for future ref surface texture
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i1, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i2, s3		; Read texel from past ref macroblock
+	 * tex2d t2, i3, s4		; Read texel from future ref macroblock
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, i + 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->b_fs[0] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFieldBMB
+(
+	struct vlR16SnormMC *mc
+)
+{
+	const unsigned int		max_tokens = 200;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Texcoords for s0
+	 * decl i1			; Texcoords for s1, s2
+	 * decl i2			; Texcoords for s3
+	 * decl i3			; Texcoords for s3
+	 * decl i4			; Texcoords for s4
+	 * decl i5			; Texcoords for s4
+	 * decl i6			; Denormalized vertex pos
+	 */
+	for (i = 0; i < 7; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels
+	 *				; and for Y-mod-2 top/bottom field selection
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t5 */
+	decl = vl_decl_temps(0, 5);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for past ref surface texture
+	 * decl s4			; Sampler for future ref surface texture
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i1, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i > 0 ? 1 : 0, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* XXX: Pos values off by 0.5? */
+	/* sub t4, i6.y, c1.x		; Sub 0.5 from denormalized pos */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 6, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* floor t3, t3			; Get rid of fractional part */
+	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t3, c1.y		; Multiply by 2 */
+	inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i2, s3		; Read texel from past ref macroblock top field
+	 * tex2d t2, i3, s3		; Read texel from past ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 2, TGSI_FILE_SAMPLER, 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t4, i4, s4		; Read texel from future ref macroblock top field
+	 * tex2d t5, i5, s4		; Read texel from future ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 4, TGSI_FILE_SAMPLER, 4);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t2, t3, t4, t5		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->b_fs[1] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateDataBufs
+(
+	struct vlR16SnormMC *mc
+)
+{
+	struct pipe_context	*pipe;
+	unsigned int		i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+
+	/* Create our vertex buffer and vertex buffer element */
+	mc->vertex_bufs[0].pitch = sizeof(struct vlVertex2f);
+	mc->vertex_bufs[0].max_index = 23;
+	mc->vertex_bufs[0].buffer_offset = 0;
+	mc->vertex_bufs[0].buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_VERTEX,
+		sizeof(struct vlVertex2f) * 24
+	);
+
+	mc->vertex_elems[0].src_offset = 0;
+	mc->vertex_elems[0].vertex_buffer_index = 0;
+	mc->vertex_elems[0].nr_components = 2;
+	mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Create our texcoord buffers and texcoord buffer elements */
+	for (i = 1; i < 3; ++i)
+	{
+		mc->vertex_bufs[i].pitch = sizeof(struct vlVertex2f);
+		mc->vertex_bufs[i].max_index = 23;
+		mc->vertex_bufs[i].buffer_offset = 0;
+		mc->vertex_bufs[i].buffer = pipe->winsys->buffer_create
+		(
+			pipe->winsys,
+			1,
+			PIPE_BUFFER_USAGE_VERTEX,
+			sizeof(struct vlVertex2f) * 24
+		);
+
+		mc->vertex_elems[i].src_offset = 0;
+		mc->vertex_elems[i].vertex_buffer_index = i;
+		mc->vertex_elems[i].nr_components = 2;
+		mc->vertex_elems[i].src_format = PIPE_FORMAT_R32G32_FLOAT;
+	}
+
+	/* Fill buffers */
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		macroblock_verts,
+		sizeof(struct vlVertex2f) * 24
+	);
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		macroblock_luma_texcoords,
+		sizeof(struct vlVertex2f) * 24
+	);
+	/* TODO: Accomodate 422, 444 */
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, mc->vertex_bufs[2].buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		macroblock_chroma_420_texcoords,
+		sizeof(struct vlVertex2f) * 24
+	);
+
+	for (i = 0; i < 3; ++i)
+		pipe->winsys->buffer_unmap(pipe->winsys, mc->vertex_bufs[i].buffer);
+
+	/* Create our constant buffer */
+	mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts);
+	mc->vs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		mc->vs_const_buf.size
+	);
+
+	mc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts);
+	mc->fs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		mc->fs_const_buf.size
+	);
+
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		&fs_consts,
+		sizeof(struct vlFragmentShaderConsts)
+	);
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->fs_const_buf.buffer);
+
+	return 0;
+}
+
+static int vlInit
+(
+	struct vlR16SnormMC *mc
+)
+{
+	struct pipe_context		*pipe;
+	struct pipe_sampler_state	sampler;
+	struct pipe_texture		template;
+	unsigned int			filters[5];
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+
+	/* For MC we render to textures, which are rounded up to nearest POT */
+	mc->viewport.scale[0] = vlRoundUpPOT(mc->video_width);
+	mc->viewport.scale[1] = vlRoundUpPOT(mc->video_height);
+	mc->viewport.scale[2] = 1;
+	mc->viewport.scale[3] = 1;
+	mc->viewport.translate[0] = 0;
+	mc->viewport.translate[1] = 0;
+	mc->viewport.translate[2] = 0;
+	mc->viewport.translate[3] = 0;
+
+	mc->render_target.width = vlRoundUpPOT(mc->video_width);
+	mc->render_target.height = vlRoundUpPOT(mc->video_height);
+	mc->render_target.num_cbufs = 1;
+	/* FB for MC stage is a vlSurface, set in vlSetRenderSurface() */
+	mc->render_target.zsbuf = NULL;
+
+	filters[0] = PIPE_TEX_FILTER_NEAREST;
+	filters[1] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR;
+	filters[2] = mc->video_format == vlFormatYCbCr444 ? PIPE_TEX_FILTER_NEAREST : PIPE_TEX_FILTER_LINEAR;
+	filters[3] = PIPE_TEX_FILTER_LINEAR;
+	filters[4] = PIPE_TEX_FILTER_LINEAR;
+
+	for (i = 0; i < 5; ++i)
+	{
+		sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.min_img_filter = filters[i];
+		sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+		sampler.mag_img_filter = filters[i];
+		sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
+		sampler.compare_func = PIPE_FUNC_ALWAYS;
+		sampler.normalized_coords = 1;
+		/*sampler.prefilter = ;*/
+		/*sampler.shadow_ambient = ;*/
+		/*sampler.lod_bias = ;*/
+		sampler.min_lod = 0;
+		/*sampler.max_lod = ;*/
+		/*sampler.border_color[i] = ;*/
+		/*sampler.max_anisotropy = ;*/
+		mc->samplers[i] = pipe->create_sampler_state(pipe, &sampler);
+	}
+
+	memset(&template, 0, sizeof(struct pipe_texture));
+	template.target = PIPE_TEXTURE_2D;
+	template.format = PIPE_FORMAT_R16_SNORM;
+	template.last_level = 0;
+	template.width[0] = 8;
+	template.height[0] = 8 * 4;
+	template.depth[0] = 1;
+	template.compressed = 0;
+	pf_get_block(template.format, &template.block);
+
+	for (i = 0; i < NUM_BUFS; ++i)
+		mc->textures[i][0] = pipe->screen->texture_create(pipe->screen, &template);
+
+	if (mc->video_format == vlFormatYCbCr420)
+		template.height[0] = 8;
+	else if (mc->video_format == vlFormatYCbCr422)
+		template.height[0] = 8 * 2;
+	else if (mc->video_format == vlFormatYCbCr444)
+		template.height[0] = 8 * 4;
+	else
+		assert(0);
+
+	for (i = 0; i < NUM_BUFS; ++i)
+	{
+		mc->textures[i][1] = pipe->screen->texture_create(pipe->screen, &template);
+		mc->textures[i][2] = pipe->screen->texture_create(pipe->screen, &template);
+	}
+
+	/* textures[3] & textures[4] are assigned from vlSurfaces for P and B macroblocks at render time */
+
+	vlCreateVertexShaderIMB(mc);
+	vlCreateFragmentShaderIMB(mc);
+	vlCreateVertexShaderFramePMB(mc);
+	vlCreateVertexShaderFieldPMB(mc);
+	vlCreateFragmentShaderFramePMB(mc);
+	vlCreateFragmentShaderFieldPMB(mc);
+	vlCreateVertexShaderFrameBMB(mc);
+	vlCreateVertexShaderFieldBMB(mc);
+	vlCreateFragmentShaderFrameBMB(mc);
+	vlCreateFragmentShaderFieldBMB(mc);
+	vlCreateDataBufs(mc);
+
+	return 0;
+}
+
+int vlCreateR16SNormMC
+(
+	struct pipe_context *pipe,
+	unsigned int video_width,
+	unsigned int video_height,
+	enum vlFormat video_format,
+	struct vlRender **render
+)
+{
+	struct vlR16SnormMC *mc;
+
+	assert(pipe);
+	assert(render);
+
+	mc = calloc(1, sizeof(struct vlR16SnormMC));
+
+	mc->base.vlBegin = &vlBegin;
+	mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16Snorm;
+	mc->base.vlEnd = &vlEnd;
+	mc->base.vlFlush = &vlFlush;
+	mc->base.vlDestroy = &vlDestroy;
+	mc->pipe = pipe;
+	mc->video_width = video_width;
+	mc->video_height = video_height;
+	mc->cur_buf = 0;
+
+	vlInit(mc);
+
+	*render = &mc->base;
+
+	return 0;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h
new file mode 100644
index 00000000000..9842926bf70
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc.h
@@ -0,0 +1,18 @@
+#ifndef vl_r16snorm_mc_h
+#define vl_r16snorm_mc_h
+
+#include "vl_types.h"
+
+struct pipe_context;
+struct vlRender;
+
+int vlCreateR16SNormMC
+(
+	struct pipe_context *pipe,
+	unsigned int video_width,
+	unsigned int video_height,
+	enum vlFormat video_format,
+	struct vlRender **render
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c
new file mode 100644
index 00000000000..650528ed8fa
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c
@@ -0,0 +1,1147 @@
+#define VL_INTERNAL
+#include "vl_r16snorm_mc_buf.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <pipe/p_context.h>
+#include <pipe/p_winsys.h>
+#include <pipe/p_screen.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+#include <util/u_math.h>
+#include "vl_render.h"
+#include "vl_shader_build.h"
+#include "vl_surface.h"
+#include "vl_util.h"
+#include "vl_types.h"
+#include "vl_defs.h"
+
+/*
+ * TODO: Dynamically determine number of buf sets to use, based on
+ * video size and available mem, since we can easily run out of memory
+ * for high res videos.
+ * Note: Destroying previous frame's buffers and creating new ones
+ * doesn't work, since the buffer are not actually destroyed until their
+ * fence is signalled, and if we render fast enough we will create faster
+ * than we destroy.
+ */
+#define NUM_BUF_SETS 4	/* Number of rotating buffer sets to use */
+
+enum vlMacroBlockTypeEx
+{
+	vlMacroBlockExTypeIntra,
+	vlMacroBlockExTypeFwdPredictedFrame,
+	vlMacroBlockExTypeFwdPredictedField,
+	vlMacroBlockExTypeBkwdPredictedFrame,
+	vlMacroBlockExTypeBkwdPredictedField,
+	vlMacroBlockExTypeBiPredictedFrame,
+	vlMacroBlockExTypeBiPredictedField,
+
+	vlNumMacroBlockExTypes
+};
+
+struct vlVertexShaderConsts
+{
+	struct vlVertex4f denorm;
+};
+
+struct vlFragmentShaderConsts
+{
+	struct vlVertex4f multiplier;
+	struct vlVertex4f div;
+};
+
+struct vlR16SnormBufferedMC
+{
+	struct vlRender				base;
+
+	unsigned int				picture_width, picture_height;
+	enum vlFormat				picture_format;
+
+	unsigned int				cur_buf;
+	struct vlSurface			*buffered_surface;
+	struct vlSurface			*past_surface, *future_surface;
+	struct vlVertex2f			surface_tex_inv_size;
+	struct vlVertex2f			zero_block[3];
+	unsigned int				num_macroblocks;
+	struct vlMpeg2MacroBlock		*macroblocks;
+
+	struct pipe_context			*pipe;
+	struct pipe_viewport_state		viewport;
+	struct pipe_framebuffer_state		render_target;
+	struct pipe_sampler_state		*samplers[5];
+	struct pipe_texture			*textures[NUM_BUF_SETS][5];
+	struct pipe_surface			*tex_surface[3];
+	short					*texels[3];
+	void					*i_vs, *p_vs[2], *b_vs[2];
+	void					*i_fs, *p_fs[2], *b_fs[2];
+	struct pipe_vertex_buffer 		vertex_bufs[NUM_BUF_SETS][3];
+	struct pipe_vertex_element		vertex_elems[8];
+	struct pipe_constant_buffer		vs_const_buf, fs_const_buf;
+};
+
+static int vlBegin
+(
+	struct vlRender *render
+)
+{
+	assert(render);
+
+	return 0;
+}
+
+static inline int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		memcpy
+		(
+			dst + y * dst_pitch,
+			src + y * VL_BLOCK_WIDTH,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+static inline int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		memcpy
+		(
+			dst + y * dst_pitch * 2,
+			src + y * VL_BLOCK_WIDTH,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+static inline int vlGrabNoBlock(short *dst, unsigned int dst_pitch)
+{
+	unsigned int y;
+
+	for (y = 0; y < VL_BLOCK_HEIGHT; ++y)
+		memset
+		(
+			dst + y * dst_pitch,
+			0,
+			VL_BLOCK_WIDTH * 2
+		);
+
+	return 0;
+}
+
+static inline int vlGrabBlocks
+(
+	struct vlR16SnormBufferedMC *mc,
+	unsigned int mbx,
+	unsigned int mby,
+	enum vlDCTType dct_type,
+	unsigned int coded_block_pattern,
+	short *blocks
+)
+{
+	short			*texels;
+	unsigned int		tex_pitch;
+	unsigned int		x, y, tb = 0, sb = 0;
+	unsigned int		mbpx = mbx * VL_MACROBLOCK_WIDTH, mbpy = mby * VL_MACROBLOCK_HEIGHT;
+
+	assert(mc);
+	assert(blocks);
+
+	tex_pitch = mc->tex_surface[0]->stride / mc->tex_surface[0]->block.size;
+	texels = mc->texels[0] + mbpy * tex_pitch + mbpx;
+
+	for (y = 0; y < 2; ++y)
+	{
+		for (x = 0; x < 2; ++x, ++tb)
+		{
+			if ((coded_block_pattern >> (5 - tb)) & 1)
+			{
+				short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
+
+				if (dct_type == vlDCTTypeFrameCoded)
+				{
+					vlGrabFrameCodedBlock
+					(
+						cur_block,
+						texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH,
+						tex_pitch
+					);
+				}
+				else
+				{
+					vlGrabFieldCodedBlock
+					(
+						cur_block,
+						texels + y * tex_pitch + x * VL_BLOCK_WIDTH,
+						tex_pitch
+					);
+				}
+
+				++sb;
+			}
+			else if (mc->zero_block[0].x < 0.0f)
+			{
+				vlGrabNoBlock(texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, tex_pitch);
+
+				mc->zero_block[0].x = (mbpx + x * 8) * mc->surface_tex_inv_size.x;
+				mc->zero_block[0].y = (mbpy + y * 8) * mc->surface_tex_inv_size.y;
+			}
+		}
+	}
+
+	/* TODO: Implement 422, 444 */
+	mbpx >>= 1;
+	mbpy >>= 1;
+
+	for (tb = 0; tb < 2; ++tb)
+	{
+		tex_pitch = mc->tex_surface[tb + 1]->stride / mc->tex_surface[tb + 1]->block.size;
+		texels = mc->texels[tb + 1] + mbpy * tex_pitch + mbpx;
+
+		if ((coded_block_pattern >> (1 - tb)) & 1)
+		{
+			short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT;
+
+			vlGrabFrameCodedBlock
+			(
+				cur_block,
+				texels,
+				tex_pitch
+			);
+
+			++sb;
+		}
+		else if (mc->zero_block[tb + 1].x < 0.0f)
+		{
+			vlGrabNoBlock(texels, tex_pitch);
+
+			mc->zero_block[tb + 1].x = (mbpx << 1) * mc->surface_tex_inv_size.x;
+			mc->zero_block[tb + 1].y = (mbpy << 1) * mc->surface_tex_inv_size.y;
+		}
+	}
+
+	return 0;
+}
+
+static inline enum vlMacroBlockTypeEx vlGetMacroBlockTypeEx(struct vlMpeg2MacroBlock *mb)
+{
+	assert(mb);
+
+	switch (mb->mb_type)
+	{
+		case vlMacroBlockTypeIntra:
+			return vlMacroBlockExTypeIntra;
+		case vlMacroBlockTypeFwdPredicted:
+			return mb->mo_type == vlMotionTypeFrame ?
+				vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField;
+		case vlMacroBlockTypeBkwdPredicted:
+			return mb->mo_type == vlMotionTypeFrame ?
+				vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField;
+		case vlMacroBlockTypeBiPredicted:
+			return mb->mo_type == vlMotionTypeFrame ?
+				vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField;
+		default:
+			assert(0);
+	}
+
+	/* Unreachable */
+	return -1;
+}
+
+static inline int vlGrabMacroBlock
+(
+	struct vlR16SnormBufferedMC *mc,
+	struct vlMpeg2MacroBlock *macroblock
+)
+{
+	assert(mc);
+	assert(macroblock);
+
+	mc->macroblocks[mc->num_macroblocks].mbx = macroblock->mbx;
+	mc->macroblocks[mc->num_macroblocks].mby = macroblock->mby;
+	mc->macroblocks[mc->num_macroblocks].mb_type = macroblock->mb_type;
+	mc->macroblocks[mc->num_macroblocks].mo_type = macroblock->mo_type;
+	mc->macroblocks[mc->num_macroblocks].dct_type = macroblock->dct_type;
+	mc->macroblocks[mc->num_macroblocks].PMV[0][0][0] = macroblock->PMV[0][0][0];
+	mc->macroblocks[mc->num_macroblocks].PMV[0][0][1] = macroblock->PMV[0][0][1];
+	mc->macroblocks[mc->num_macroblocks].PMV[0][1][0] = macroblock->PMV[0][1][0];
+	mc->macroblocks[mc->num_macroblocks].PMV[0][1][1] = macroblock->PMV[0][1][1];
+	mc->macroblocks[mc->num_macroblocks].PMV[1][0][0] = macroblock->PMV[1][0][0];
+	mc->macroblocks[mc->num_macroblocks].PMV[1][0][1] = macroblock->PMV[1][0][1];
+	mc->macroblocks[mc->num_macroblocks].PMV[1][1][0] = macroblock->PMV[1][1][0];
+	mc->macroblocks[mc->num_macroblocks].PMV[1][1][1] = macroblock->PMV[1][1][1];
+	mc->macroblocks[mc->num_macroblocks].cbp = macroblock->cbp;
+	mc->macroblocks[mc->num_macroblocks].blocks = macroblock->blocks;
+
+	vlGrabBlocks
+	(
+		mc,
+		macroblock->mbx,
+		macroblock->mby,
+		macroblock->dct_type,
+		macroblock->cbp,
+		macroblock->blocks
+	);
+
+	mc->num_macroblocks++;
+
+	return 0;
+}
+
+#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zb)					\
+	(vb)[0].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[0].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[1].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+	(vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].pos.y = (mby) * (unity) + (ofsy);			\
+	(vb)[4].pos.x = (mbx) * (unitx) + (ofsx);		(vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+	(vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy);		\
+																\
+	if ((cbp) & (lm))													\
+	{															\
+		(vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].luma_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].luma_tc.x = (zb)[0].x;		(vb)[0].luma_tc.y = (zb)[0].y;						\
+		(vb)[1].luma_tc.x = (zb)[0].x;		(vb)[1].luma_tc.y = (zb)[0].y + (hy);					\
+		(vb)[2].luma_tc.x = (zb)[0].x + (hx);	(vb)[2].luma_tc.y = (zb)[0].y;						\
+		(vb)[3].luma_tc.x = (zb)[0].x + (hx);	(vb)[3].luma_tc.y = (zb)[0].y;						\
+		(vb)[4].luma_tc.x = (zb)[0].x;		(vb)[4].luma_tc.y = (zb)[0].y + (hy);					\
+		(vb)[5].luma_tc.x = (zb)[0].x + (hx);	(vb)[5].luma_tc.y = (zb)[0].y + (hy);					\
+	}															\
+																\
+	if ((cbp) & (cbm))													\
+	{															\
+		(vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cb_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].cb_tc.x = (zb)[1].x;		(vb)[0].cb_tc.y = (zb)[1].y;						\
+		(vb)[1].cb_tc.x = (zb)[1].x;		(vb)[1].cb_tc.y = (zb)[1].y + (hy);					\
+		(vb)[2].cb_tc.x = (zb)[1].x + (hx);	(vb)[2].cb_tc.y = (zb)[1].y;						\
+		(vb)[3].cb_tc.x = (zb)[1].x + (hx);	(vb)[3].cb_tc.y = (zb)[1].y;						\
+		(vb)[4].cb_tc.x = (zb)[1].x;		(vb)[4].cb_tc.y = (zb)[1].y + (hy);					\
+		(vb)[5].cb_tc.x = (zb)[1].x + (hx);	(vb)[5].cb_tc.y = (zb)[1].y + (hy);					\
+	}															\
+																\
+	if ((cbp) & (crm))													\
+	{															\
+		(vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[0].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[2].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[3].cr_tc.y = (mby) * (unity) + (ofsy);		\
+		(vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx);		(vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+		(vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx);	(vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy);	\
+	}															\
+	else															\
+	{															\
+		(vb)[0].cr_tc.x = (zb)[2].x;		(vb)[0].cr_tc.y = (zb)[2].y;						\
+		(vb)[1].cr_tc.x = (zb)[2].x;		(vb)[1].cr_tc.y = (zb)[2].y + (hy);					\
+		(vb)[2].cr_tc.x = (zb)[2].x + (hx);	(vb)[2].cr_tc.y = (zb)[2].y;						\
+		(vb)[3].cr_tc.x = (zb)[2].x + (hx);	(vb)[3].cr_tc.y = (zb)[2].y;						\
+		(vb)[4].cr_tc.x = (zb)[2].x;		(vb)[4].cr_tc.y = (zb)[2].y + (hy);					\
+		(vb)[5].cr_tc.x = (zb)[2].x + (hx);	(vb)[5].cr_tc.y = (zb)[2].y + (hy);					\
+	}
+
+static inline int vlGrabMacroBlockVB
+(
+	struct vlR16SnormBufferedMC *mc,
+	struct vlMpeg2MacroBlock *macroblock,
+	unsigned int pos
+)
+{
+	struct vlVertex2f	mo_vec[2];
+	unsigned int		i;
+
+	assert(mc);
+	assert(macroblock);
+
+	switch (macroblock->mb_type)
+	{
+		case vlMacroBlockTypeBiPredicted:
+		{
+			struct vlVertex2f *vb;
+
+			vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map
+			(
+				mc->pipe->winsys,
+				mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][2].buffer,
+				PIPE_BUFFER_USAGE_CPU_WRITE
+			) + pos * 2 * 24;
+
+			mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x;
+			mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y;
+
+			if (macroblock->mo_type == vlMotionTypeFrame)
+			{
+				for (i = 0; i < 24 * 2; i += 2)
+				{
+					vb[i].x = mo_vec[0].x;
+					vb[i].y = mo_vec[0].y;
+				}
+			}
+			else
+			{
+				mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x;
+				mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y;
+
+				for (i = 0; i < 24 * 2; i += 2)
+				{
+					vb[i].x = mo_vec[0].x;
+					vb[i].y = mo_vec[0].y;
+					vb[i + 1].x = mo_vec[1].x;
+					vb[i + 1].y = mo_vec[1].y;
+				}
+			}
+
+			mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][2].buffer);
+
+			/* fall-through */
+		}
+		case vlMacroBlockTypeFwdPredicted:
+		case vlMacroBlockTypeBkwdPredicted:
+		{
+			struct vlVertex2f *vb;
+
+			vb = (struct vlVertex2f*)mc->pipe->winsys->buffer_map
+			(
+				mc->pipe->winsys,
+				mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][1].buffer,
+				PIPE_BUFFER_USAGE_CPU_WRITE
+			) + pos * 2 * 24;
+
+			if (macroblock->mb_type == vlMacroBlockTypeBkwdPredicted)
+			{
+				mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x;
+				mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y;
+
+				if (macroblock->mo_type == vlMotionTypeField)
+				{
+					mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x;
+					mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y;
+				}
+			}
+			else
+			{
+				mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x;
+				mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y;
+
+				if (macroblock->mo_type == vlMotionTypeField)
+				{
+					mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x;
+					mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y;
+				}
+			}
+
+			if (macroblock->mo_type == vlMotionTypeFrame)
+			{
+				for (i = 0; i < 24 * 2; i += 2)
+				{
+					vb[i].x = mo_vec[0].x;
+					vb[i].y = mo_vec[0].y;
+				}
+			}
+			else
+			{
+				for (i = 0; i < 24 * 2; i += 2)
+				{
+					vb[i].x = mo_vec[0].x;
+					vb[i].y = mo_vec[0].y;
+					vb[i + 1].x = mo_vec[1].x;
+					vb[i + 1].y = mo_vec[1].y;
+				}
+			}
+
+			mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][1].buffer);
+
+			/* fall-through */
+		}
+		case vlMacroBlockTypeIntra:
+		{
+			const struct vlVertex2f	unit =
+			{
+				mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH,
+				mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT
+			};
+			const struct vlVertex2f half =
+			{
+				mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2),
+				mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2)
+			};
+
+			struct vlMacroBlockVertexStream0
+			{
+				struct vlVertex2f pos;
+				struct vlVertex2f luma_tc;
+				struct vlVertex2f cb_tc;
+				struct vlVertex2f cr_tc;
+			} *vb;
+
+			vb = (struct vlMacroBlockVertexStream0*)mc->pipe->winsys->buffer_map
+			(
+				mc->pipe->winsys,
+				mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer,
+				PIPE_BUFFER_USAGE_CPU_WRITE
+			) + pos * 24;
+
+			SET_BLOCK
+			(
+				vb,
+				macroblock->cbp, macroblock->mbx, macroblock->mby,
+				unit.x, unit.y, 0, 0, half.x, half.y,
+				32, 2, 1, mc->zero_block
+			);
+
+			SET_BLOCK
+			(
+				vb + 6,
+				macroblock->cbp, macroblock->mbx, macroblock->mby,
+				unit.x, unit.y, half.x, 0, half.x, half.y,
+				16, 2, 1, mc->zero_block
+			);
+
+			SET_BLOCK
+			(
+				vb + 12,
+				macroblock->cbp, macroblock->mbx, macroblock->mby,
+				unit.x, unit.y, 0, half.y, half.x, half.y,
+				8, 2, 1, mc->zero_block
+			);
+
+			SET_BLOCK
+			(
+				vb + 18,
+				macroblock->cbp, macroblock->mbx, macroblock->mby,
+				unit.x, unit.y, half.x, half.y, half.x, half.y,
+				4, 2, 1, mc->zero_block
+			);
+
+			mc->pipe->winsys->buffer_unmap(mc->pipe->winsys, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS][0].buffer);
+
+			break;
+		}
+		default:
+			assert(0);
+	}
+
+	return 0;
+}
+
+static int vlFlush
+(
+	struct vlRender *render
+)
+{
+	struct vlR16SnormBufferedMC	*mc;
+	struct pipe_context		*pipe;
+	struct vlVertexShaderConsts	*vs_consts;
+	unsigned int			num_macroblocks[vlNumMacroBlockExTypes] = {0};
+	unsigned int			offset[vlNumMacroBlockExTypes];
+	unsigned int			vb_start = 0;
+	unsigned int			mbw;
+	unsigned int			mbh;
+	unsigned int			num_mb_per_frame;
+	unsigned int			i;
+
+	assert(render);
+
+	mc = (struct vlR16SnormBufferedMC*)render;
+
+	if (!mc->buffered_surface)
+		return 0;
+
+	mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH;
+	mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT;
+	num_mb_per_frame = mbw * mbh;
+
+	if (mc->num_macroblocks < num_mb_per_frame)
+		return 0;
+
+	pipe = mc->pipe;
+
+	for (i = 0; i < mc->num_macroblocks; ++i)
+	{
+		enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]);
+
+		num_macroblocks[mb_type_ex]++;
+	}
+
+	offset[0] = 0;
+
+	for (i = 1; i < vlNumMacroBlockExTypes; ++i)
+		offset[i] = offset[i - 1] + num_macroblocks[i - 1];
+
+	for (i = 0; i < mc->num_macroblocks; ++i)
+	{
+		enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]);
+
+		vlGrabMacroBlockVB(mc, &mc->macroblocks[i], offset[mb_type_ex]);
+
+		offset[mb_type_ex]++;
+	}
+	
+	for (i = 0; i < 3; ++i)
+	{
+		pipe_surface_unmap(mc->tex_surface[i]);
+		mc->pipe->screen->tex_surface_release(mc->pipe->screen, &mc->tex_surface[i]);
+	}
+
+	mc->render_target.cbufs[0] = pipe->screen->get_tex_surface
+	(
+		pipe->screen,
+		mc->buffered_surface->texture,
+		0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE
+	);
+
+	pipe->set_framebuffer_state(pipe, &mc->render_target);
+	pipe->set_viewport_state(pipe, &mc->viewport);
+	vs_consts = pipe->winsys->buffer_map
+	(
+		pipe->winsys,
+		mc->vs_const_buf.buffer,
+		PIPE_BUFFER_USAGE_CPU_WRITE
+	);
+
+	vs_consts->denorm.x = mc->buffered_surface->texture->width[0];
+	vs_consts->denorm.y = mc->buffered_surface->texture->height[0];
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->vs_const_buf.buffer);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf);
+	pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf);
+
+	if (num_macroblocks[vlMacroBlockExTypeIntra] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 4, mc->vertex_elems);
+		pipe->set_sampler_textures(pipe, 3, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 3, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->i_vs);
+		pipe->bind_fs_state(pipe, mc->i_fs);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeIntra] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeIntra] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture;
+		pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->p_vs[0]);
+		pipe->bind_fs_state(pipe, mc->p_fs[0]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture;
+		pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->p_vs[1]);
+		pipe->bind_fs_state(pipe, mc->p_fs[1]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture;
+		pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->p_vs[0]);
+		pipe->bind_fs_state(pipe, mc->p_fs[0]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 6, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->future_surface->texture;
+		pipe->set_sampler_textures(pipe, 4, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 4, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->p_vs[1]);
+		pipe->bind_fs_state(pipe, mc->p_fs[1]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 8, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture;
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture;
+		pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->b_vs[0]);
+		pipe->bind_fs_state(pipe, mc->b_fs[0]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24;
+	}
+
+	if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0)
+	{
+		pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->set_vertex_elements(pipe, 8, mc->vertex_elems);
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][3] = mc->past_surface->texture;
+		mc->textures[mc->cur_buf % NUM_BUF_SETS][4] = mc->future_surface->texture;
+		pipe->set_sampler_textures(pipe, 5, mc->textures[mc->cur_buf % NUM_BUF_SETS]);
+		pipe->bind_sampler_states(pipe, 5, (void**)mc->samplers);
+		pipe->bind_vs_state(pipe, mc->b_vs[1]);
+		pipe->bind_fs_state(pipe, mc->b_fs[1]);
+
+		pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24);
+		vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24;
+	}
+
+	pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &mc->buffered_surface->render_fence);
+
+	for (i = 0; i < 3; ++i)
+		mc->zero_block[i].x = -1.0f;
+
+	mc->buffered_surface = NULL;
+	mc->num_macroblocks = 0;
+	mc->cur_buf++;
+
+	return 0;
+}
+
+static int vlRenderMacroBlocksMpeg2R16SnormBuffered
+(
+	struct vlRender *render,
+	struct vlMpeg2MacroBlockBatch *batch,
+	struct vlSurface *surface
+)
+{
+	struct vlR16SnormBufferedMC	*mc;
+	unsigned int			i;
+
+	assert(render);
+
+	mc = (struct vlR16SnormBufferedMC*)render;
+
+	if (mc->buffered_surface)
+	{
+		if (mc->buffered_surface != surface)
+		{
+			vlFlush(&mc->base);
+			mc->buffered_surface = surface;
+			mc->past_surface = batch->past_surface;
+			mc->future_surface = batch->future_surface;
+			mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0];
+			mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0];
+			
+			for (i = 0; i < 3; ++i)
+			{
+				mc->tex_surface[i] = mc->pipe->screen->get_tex_surface
+				(
+					mc->pipe->screen,
+					mc->textures[mc->cur_buf % NUM_BUF_SETS][i],
+					0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE
+				);
+
+				mc->texels[i] = pipe_surface_map(mc->tex_surface[i], PIPE_BUFFER_USAGE_CPU_WRITE);
+			}
+		}
+	}
+	else
+	{
+		mc->buffered_surface = surface;
+		mc->past_surface = batch->past_surface;
+		mc->future_surface = batch->future_surface;
+		mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0];
+		mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0];
+		
+		for (i = 0; i < 3; ++i)
+		{
+			mc->tex_surface[i] = mc->pipe->screen->get_tex_surface
+			(
+				mc->pipe->screen,
+				mc->textures[mc->cur_buf % NUM_BUF_SETS][i],
+				0, 0, 0, PIPE_BUFFER_USAGE_CPU_WRITE
+			);
+
+			mc->texels[i] = pipe_surface_map(mc->tex_surface[i], PIPE_BUFFER_USAGE_CPU_WRITE);
+		}
+	}
+
+	for (i = 0; i < batch->num_macroblocks; ++i)
+		vlGrabMacroBlock(mc, &batch->macroblocks[i]);
+
+	return 0;
+}
+
+static int vlEnd
+(
+	struct vlRender *render
+)
+{
+	assert(render);
+
+	return 0;
+}
+
+static int vlDestroy
+(
+	struct vlRender *render
+)
+{
+	struct vlR16SnormBufferedMC	*mc;
+	struct pipe_context		*pipe;
+	unsigned int			h, i;
+
+	assert(render);
+
+	mc = (struct vlR16SnormBufferedMC*)render;
+	pipe = mc->pipe;
+
+	for (i = 0; i < 5; ++i)
+		pipe->delete_sampler_state(pipe, mc->samplers[i]);
+
+	for (h = 0; h < NUM_BUF_SETS; ++h)
+			for (i = 0; i < 3; ++i)
+				pipe->winsys->buffer_destroy(pipe->winsys, mc->vertex_bufs[h][i].buffer);
+
+	/* Textures 3 & 4 are not created directly, no need to release them here */
+	for (i = 0; i < NUM_BUF_SETS; ++i)
+	{
+		pipe_texture_release(&mc->textures[i][0]);
+		pipe_texture_release(&mc->textures[i][1]);
+		pipe_texture_release(&mc->textures[i][2]);
+	}
+
+	pipe->delete_vs_state(pipe, mc->i_vs);
+	pipe->delete_fs_state(pipe, mc->i_fs);
+
+	for (i = 0; i < 2; ++i)
+	{
+		pipe->delete_vs_state(pipe, mc->p_vs[i]);
+		pipe->delete_fs_state(pipe, mc->p_fs[i]);
+		pipe->delete_vs_state(pipe, mc->b_vs[i]);
+		pipe->delete_fs_state(pipe, mc->b_fs[i]);
+	}
+
+	pipe->winsys->buffer_destroy(pipe->winsys, mc->vs_const_buf.buffer);
+	pipe->winsys->buffer_destroy(pipe->winsys, mc->fs_const_buf.buffer);
+
+	free(mc->macroblocks);
+	free(mc);
+
+	return 0;
+}
+
+/*
+ * Muliplier renormalizes block samples from 16 bits to 12 bits.
+ * Divider is used when calculating Y % 2 for choosing top or bottom
+ * field for P or B macroblocks.
+ * TODO: Use immediates.
+ */
+static const struct vlFragmentShaderConsts fs_consts =
+{
+	{32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f},
+	{0.5f, 2.0f, 0.0f, 0.0f}
+};
+
+#include "vl_r16snorm_mc_buf_shaders.inc"
+
+static int vlCreateDataBufs
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int	mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH;
+	const unsigned int	mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT;
+	const unsigned int	num_mb_per_frame = mbw * mbh;
+
+	struct pipe_context	*pipe;
+	unsigned int		h, i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+
+	/* Create our vertex buffers */
+	for (h = 0; h < NUM_BUF_SETS; ++h)
+	{
+		mc->vertex_bufs[h][0].pitch = sizeof(struct vlVertex2f) * 4;
+		mc->vertex_bufs[h][0].max_index = 24 * num_mb_per_frame - 1;
+		mc->vertex_bufs[h][0].buffer_offset = 0;
+		mc->vertex_bufs[h][0].buffer = pipe->winsys->buffer_create
+		(
+			pipe->winsys,
+			1,
+			PIPE_BUFFER_USAGE_VERTEX,
+			sizeof(struct vlVertex2f) * 4 * 24 * num_mb_per_frame
+		);
+
+		for (i = 1; i < 3; ++i)
+		{
+			mc->vertex_bufs[h][i].pitch = sizeof(struct vlVertex2f) * 2;
+			mc->vertex_bufs[h][i].max_index = 24 * num_mb_per_frame - 1;
+			mc->vertex_bufs[h][i].buffer_offset = 0;
+			mc->vertex_bufs[h][i].buffer = pipe->winsys->buffer_create
+			(
+				pipe->winsys,
+				1,
+				PIPE_BUFFER_USAGE_VERTEX,
+				sizeof(struct vlVertex2f) * 2 * 24 * num_mb_per_frame
+			);
+		}
+	}
+
+	/* Position element */
+	mc->vertex_elems[0].src_offset = 0;
+	mc->vertex_elems[0].vertex_buffer_index = 0;
+	mc->vertex_elems[0].nr_components = 2;
+	mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Luma, texcoord element */
+	mc->vertex_elems[1].src_offset = sizeof(struct vlVertex2f);
+	mc->vertex_elems[1].vertex_buffer_index = 0;
+	mc->vertex_elems[1].nr_components = 2;
+	mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Chroma Cr texcoord element */
+	mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f) * 2;
+	mc->vertex_elems[2].vertex_buffer_index = 0;
+	mc->vertex_elems[2].nr_components = 2;
+	mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Chroma Cb texcoord element */
+	mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f) * 3;
+	mc->vertex_elems[3].vertex_buffer_index = 0;
+	mc->vertex_elems[3].nr_components = 2;
+	mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* First ref surface top field texcoord element */
+	mc->vertex_elems[4].src_offset = 0;
+	mc->vertex_elems[4].vertex_buffer_index = 1;
+	mc->vertex_elems[4].nr_components = 2;
+	mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* First ref surface bottom field texcoord element */
+	mc->vertex_elems[5].src_offset = sizeof(struct vlVertex2f);
+	mc->vertex_elems[5].vertex_buffer_index = 1;
+	mc->vertex_elems[5].nr_components = 2;
+	mc->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Second ref surface top field texcoord element */
+	mc->vertex_elems[6].src_offset = 0;
+	mc->vertex_elems[6].vertex_buffer_index = 2;
+	mc->vertex_elems[6].nr_components = 2;
+	mc->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Second ref surface bottom field texcoord element */
+	mc->vertex_elems[7].src_offset = sizeof(struct vlVertex2f);
+	mc->vertex_elems[7].vertex_buffer_index = 2;
+	mc->vertex_elems[7].nr_components = 2;
+	mc->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT;
+
+	/* Create our constant buffer */
+	mc->vs_const_buf.size = sizeof(struct vlVertexShaderConsts);
+	mc->vs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		mc->vs_const_buf.size
+	);
+
+	mc->fs_const_buf.size = sizeof(struct vlFragmentShaderConsts);
+	mc->fs_const_buf.buffer = pipe->winsys->buffer_create
+	(
+		pipe->winsys,
+		1,
+		PIPE_BUFFER_USAGE_CONSTANT,
+		mc->fs_const_buf.size
+	);
+
+	memcpy
+	(
+		pipe->winsys->buffer_map(pipe->winsys, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE),
+		&fs_consts,
+		sizeof(struct vlFragmentShaderConsts)
+	);
+
+	pipe->winsys->buffer_unmap(pipe->winsys, mc->fs_const_buf.buffer);
+
+	mc->macroblocks = malloc(sizeof(struct vlMpeg2MacroBlock) * num_mb_per_frame);
+
+	return 0;
+}
+
+static int vlInit
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	struct pipe_context		*pipe;
+	struct pipe_sampler_state	sampler;
+	struct pipe_texture		template;
+	unsigned int			filters[5];
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+
+	/* For MC we render to textures, which are rounded up to nearest POT */
+	mc->viewport.scale[0] = vlRoundUpPOT(mc->picture_width);
+	mc->viewport.scale[1] = vlRoundUpPOT(mc->picture_height);
+	mc->viewport.scale[2] = 1;
+	mc->viewport.scale[3] = 1;
+	mc->viewport.translate[0] = 0;
+	mc->viewport.translate[1] = 0;
+	mc->viewport.translate[2] = 0;
+	mc->viewport.translate[3] = 0;
+
+	mc->render_target.width = vlRoundUpPOT(mc->picture_width);
+	mc->render_target.height = vlRoundUpPOT(mc->picture_height);
+	mc->render_target.num_cbufs = 1;
+	/* FB for MC stage is a vlSurface created by the user, set at render time */
+	mc->render_target.zsbuf = NULL;
+
+	filters[0] = PIPE_TEX_FILTER_NEAREST;
+	/* FIXME: Linear causes discoloration around block edges */
+	filters[1] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/;
+	filters[2] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/;
+	filters[3] = PIPE_TEX_FILTER_LINEAR;
+	filters[4] = PIPE_TEX_FILTER_LINEAR;
+
+	for (i = 0; i < 5; ++i)
+	{
+		sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+		sampler.min_img_filter = filters[i];
+		sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
+		sampler.mag_img_filter = filters[i];
+		sampler.compare_mode = PIPE_TEX_COMPARE_NONE;
+		sampler.compare_func = PIPE_FUNC_ALWAYS;
+		sampler.normalized_coords = 1;
+		/*sampler.prefilter = ;*/
+		/*sampler.shadow_ambient = ;*/
+		/*sampler.lod_bias = ;*/
+		sampler.min_lod = 0;
+		/*sampler.max_lod = ;*/
+		/*sampler.border_color[i] = ;*/
+		/*sampler.max_anisotropy = ;*/
+		mc->samplers[i] = pipe->create_sampler_state(pipe, &sampler);
+	}
+
+	memset(&template, 0, sizeof(struct pipe_texture));
+	template.target = PIPE_TEXTURE_2D;
+	template.format = PIPE_FORMAT_R16_SNORM;
+	template.last_level = 0;
+	template.width[0] = vlRoundUpPOT(mc->picture_width);
+	template.height[0] = vlRoundUpPOT(mc->picture_height);
+	template.depth[0] = 1;
+	template.compressed = 0;
+	pf_get_block(template.format, &template.block);
+	template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER;
+
+	for (i = 0; i < NUM_BUF_SETS; ++i)
+		mc->textures[i][0] = pipe->screen->texture_create(pipe->screen, &template);
+
+	if (mc->picture_format == vlFormatYCbCr420)
+	{
+		template.width[0] = vlRoundUpPOT(mc->picture_width / 2);
+		template.height[0] = vlRoundUpPOT(mc->picture_height / 2);
+	}
+	else if (mc->picture_format == vlFormatYCbCr422)
+		template.height[0] = vlRoundUpPOT(mc->picture_height / 2);
+
+	for (i = 0; i < NUM_BUF_SETS; ++i)
+	{
+		mc->textures[i][1] = pipe->screen->texture_create(pipe->screen, &template);
+		mc->textures[i][2] = pipe->screen->texture_create(pipe->screen, &template);
+	}
+
+	/* textures[3] & textures[4] are assigned from vlSurfaces for P and B macroblocks at render time */
+
+	vlCreateVertexShaderIMB(mc);
+	vlCreateFragmentShaderIMB(mc);
+	vlCreateVertexShaderFramePMB(mc);
+	vlCreateVertexShaderFieldPMB(mc);
+	vlCreateFragmentShaderFramePMB(mc);
+	vlCreateFragmentShaderFieldPMB(mc);
+	vlCreateVertexShaderFrameBMB(mc);
+	vlCreateVertexShaderFieldBMB(mc);
+	vlCreateFragmentShaderFrameBMB(mc);
+	vlCreateFragmentShaderFieldBMB(mc);
+	vlCreateDataBufs(mc);
+
+	return 0;
+}
+
+int vlCreateR16SNormBufferedMC
+(
+	struct pipe_context *pipe,
+	unsigned int picture_width,
+	unsigned int picture_height,
+	enum vlFormat picture_format,
+	struct vlRender **render
+)
+{
+	struct vlR16SnormBufferedMC	*mc;
+	unsigned int			i;
+
+	assert(pipe);
+	assert(render);
+
+	mc = calloc(1, sizeof(struct vlR16SnormBufferedMC));
+
+	mc->base.vlBegin = &vlBegin;
+	mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16SnormBuffered;
+	mc->base.vlEnd = &vlEnd;
+	mc->base.vlFlush = &vlFlush;
+	mc->base.vlDestroy = &vlDestroy;
+	mc->pipe = pipe;
+	mc->picture_width = picture_width;
+	mc->picture_height = picture_height;
+
+	mc->cur_buf = 0;
+	mc->buffered_surface = NULL;
+	mc->past_surface = NULL;
+	mc->future_surface = NULL;
+	for (i = 0; i < 3; ++i)
+		mc->zero_block[i].x = -1.0f;
+	mc->num_macroblocks = 0;
+
+	vlInit(mc);
+
+	*render = &mc->base;
+
+	return 0;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h
new file mode 100644
index 00000000000..27177d64cad
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h
@@ -0,0 +1,18 @@
+#ifndef vl_r16snorm_mc_buf_h
+#define vl_r16snorm_mc_buf_h
+
+#include "vl_types.h"
+
+struct pipe_context;
+struct vlRender;
+
+int vlCreateR16SNormBufferedMC
+(
+	struct pipe_context *pipe,
+	unsigned int picture_width,
+	unsigned int picture_height,
+	enum vlFormat picture_format,
+	struct vlRender **render
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc
new file mode 100644
index 00000000000..ef4a4b2add9
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc
@@ -0,0 +1,1185 @@
+static int vlCreateVertexShaderIMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 50;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma Cb texcoords
+	 * decl i3		; Chroma Cr texcoords
+	 */
+	for (i = 0; i < 4; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma Cb texcoords
+	 * decl o3		; Chroma Cr texcoords
+	 */
+	for (i = 0; i < 4; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * mov o0, i0		; Move input vertex pos to output
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma Cb texcoords to output
+	 * mov o3, i3		; Move input chroma Cr texcoords to output
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->i_vs = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderIMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Luma texcoords
+	 * decl i1			; Chroma Cb texcoords
+	 * decl i2			; Chroma Cr texcoords
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul o0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->i_fs = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFramePMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma Cb texcoords
+	 * decl i3		; Chroma Cr texcoords
+	 * decl i4		; Ref surface top field texcoords
+	 * decl i5		; Ref surface bottom field texcoords (unused, packed in the same stream)
+	 */
+	for (i = 0; i < 6; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma Cb texcoords
+	 * decl o3		; Chroma Cr texcoords
+	 * decl o4		; Ref macroblock texcoords
+	 */
+	for (i = 0; i < 5; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * mov o0, i0		; Move input vertex pos to output
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma Cb texcoords to output
+	 * mov o3, i3		; Move input chroma Cr texcoords to output
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* add o4, i0, i4	; Translate vertex pos by motion vec to form ref macroblock texcoords */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->p_vs[0] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFieldPMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma Cb texcoords
+	 * decl i3		; Chroma Cr texcoords
+	 * decl i4              ; Ref macroblock top field texcoords
+	 * decl i5              ; Ref macroblock bottom field texcoords
+	 */
+	for (i = 0; i < 6; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0		; Render target dimensions */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma Cb texcoords
+	 * decl o3		; Chroma Cr texcoords
+	 * decl o4		; Ref macroblock top field texcoords
+	 * decl o5		; Ref macroblock bottom field texcoords
+	 * decl o6		; Denormalized vertex pos
+	 */
+	for (i = 0; i < 7; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * mov o0, i0		; Move input vertex pos to output
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma Cb texcoords to output
+	 * mov o3, i3		; Move input chroma Cr texcoords to output
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * add o4, i0, i4	; Translate vertex pos by motion vec to form top field macroblock texcoords
+	 * add o5, i0, i5	; Translate vertex pos by motion vec to form bottom field macroblock texcoords
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul o6, i0, c0	; Denorm vertex pos */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->p_vs[1] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFramePMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Luma texcoords
+	 * decl i1			; Chroma Cb texcoords
+	 * decl i2			; Chroma Cr texcoords
+	 * decl i3			; Ref macroblock texcoords
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for ref surface texture
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* tex2d t1, i3, s3		; Read texel from ref macroblock */
+	inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->p_fs[0] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFieldPMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 200;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Luma texcoords
+	 * decl i1			; Chroma Cb texcoords
+	 * decl i2			; Chroma Cr texcoords
+	 * decl i3			; Ref macroblock top field texcoords
+	 * decl i4			; Ref macroblock bottom field texcoords
+	 * decl i5			; Denormalized vertex pos
+	 */
+	for (i = 0; i < 6; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t4 */
+	decl = vl_decl_temps(0, 4);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for ref surface texture
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i3, s3		; Read texel from ref macroblock top field
+	 * tex2d t2, i4, s3		; Read texel from ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* XXX: Pos values off by 0.5? */
+	/* sub t4, i5.y, c1.x		; Sub 0.5 from denormalized pos */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* floor t3, t3			; Get rid of fractional part */
+	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t3, c1.y		; Multiply by 2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->p_fs[1] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFrameBMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma Cb texcoords
+	 * decl i3		; Chroma Cr texcoords
+	 * decl i4              ; First ref macroblock top field texcoords
+	 * decl i5              ; First ref macroblock bottom field texcoords (unused, packed in the same stream)
+	 * decl i6		; Second ref macroblock top field texcoords
+	 * decl i7		; Second ref macroblock bottom field texcoords (unused, packed in the same stream)
+	 */
+	for (i = 0; i < 8; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma Cb texcoords
+	 * decl o3		; Chroma Cr texcoords
+	 * decl o4		; First ref macroblock texcoords
+	 * decl o5		; Second ref macroblock texcoords
+	 */
+	for (i = 0; i < 6; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * mov o0, i0		; Move input vertex pos to output
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma Cb texcoords to output
+	 * mov o3, i3		; Move input chroma Cr texcoords to output
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * add o4, i0, i4	; Translate vertex pos by motion vec to form first ref macroblock texcoords
+	 * add o5, i0, i6	; Translate vertex pos by motion vec to form second ref macroblock texcoords
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->b_vs[0] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateVertexShaderFieldBMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	vs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0		; Vertex pos
+	 * decl i1		; Luma texcoords
+	 * decl i2		; Chroma Cb texcoords
+	 * decl i3		; Chroma Cr texcoords
+	 * decl i4              ; First ref macroblock top field texcoords
+	 * decl i5              ; First ref macroblock bottom field texcoords
+	 * decl i6              ; Second ref macroblock top field texcoords
+	 * decl i7              ; Second ref macroblock bottom field texcoords
+	 */
+	for (i = 0; i < 8; i++)
+	{
+		decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl c0		; Render target dimensions */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl o0		; Vertex pos
+	 * decl o1		; Luma texcoords
+	 * decl o2		; Chroma Cb texcoords
+	 * decl o3		; Chroma Cr texcoords
+	 * decl o4		; First ref macroblock top field texcoords
+	 * decl o5		; First ref macroblock Bottom field texcoords
+	 * decl o6		; Second ref macroblock top field texcoords
+	 * decl o7		; Second ref macroblock Bottom field texcoords
+	 * decl o8		; Denormalized vertex pos
+	 */
+	for (i = 0; i < 9; i++)
+	{
+		decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* decl t0, t1 */
+	decl = vl_decl_temps(0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * mov o0, i0		; Move input vertex pos to output
+	 * mov o1, i1		; Move input luma texcoords to output
+	 * mov o2, i2		; Move input chroma Cb texcoords to output
+	 * mov o3, i3		; Move input chroma Cr texcoords to output
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * add o4, i0, i4	; Translate vertex pos by motion vec to form first top field macroblock texcoords
+	 * add o5, i0, i5	; Translate vertex pos by motion vec to form first bottom field macroblock texcoords
+	 * add o6, i0, i6	; Translate vertex pos by motion vec to form second top field macroblock texcoords
+	 * add o7, i0, i7	; Translate vertex pos by motion vec to form second bottom field macroblock texcoords
+	 */
+	for (i = 0; i < 4; ++i)
+	{
+		inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul o8, i0, c0	; Denorm vertex pos */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 8, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	vs.tokens = tokens;
+	mc->b_vs[1] = pipe->create_vs_state(pipe, &vs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFrameBMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 100;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Luma texcoords
+	 * decl i1			; Chroma Cb texcoords
+	 * decl i2			; Chroma Cr texcoords
+	 * decl i3			; First ref macroblock texcoords
+	 * decl i4			; Second ref macroblock texcoords
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constant 1/2 in .x channel to use as weight to blend past and future texels
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t2 */
+	decl = vl_decl_temps(0, 2);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for first ref surface texture
+	 * decl s4			; Sampler for second ref surface texture
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i3, s3		; Read texel from first ref macroblock
+	 * tex2d t2, i4, s4		; Read texel from second ref macroblock
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->b_fs[0] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
+
+static int vlCreateFragmentShaderFieldBMB
+(
+	struct vlR16SnormBufferedMC *mc
+)
+{
+	const unsigned int		max_tokens = 200;
+
+	struct pipe_context		*pipe;
+	struct pipe_shader_state	fs;
+	struct tgsi_token		*tokens;
+	struct tgsi_header		*header;
+
+	struct tgsi_full_declaration	decl;
+	struct tgsi_full_instruction	inst;
+
+	unsigned int			ti;
+	unsigned int			i;
+
+	assert(mc);
+
+	pipe = mc->pipe;
+	tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token));
+
+	/* Version */
+	*(struct tgsi_version*)&tokens[0] = tgsi_build_version();
+	/* Header */
+	header = (struct tgsi_header*)&tokens[1];
+	*header = tgsi_build_header();
+	/* Processor */
+	*(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+
+	ti = 3;
+
+	/*
+	 * decl i0			; Luma texcoords
+	 * decl i1			; Chroma Cb texcoords
+	 * decl i2			; Chroma Cr texcoords
+	 * decl i3			; First ref macroblock top field texcoords
+	 * decl i4			; First ref macroblock bottom field texcoords
+	 * decl i5			; Second ref macroblock top field texcoords
+	 * decl i6			; Second ref macroblock bottom field texcoords
+	 * decl i7			; Denormalized vertex pos
+	 */
+	for (i = 0; i < 8; ++i)
+	{
+		decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * decl c0			; Scaling factor, rescales 16-bit snorm to 9-bit snorm
+	 * decl c1			; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels
+	 *				; and for Y-mod-2 top/bottom field selection
+	 */
+	decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl o0			; Fragment color */
+	decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/* decl t0-t5 */
+	decl = vl_decl_temps(0, 5);
+	ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * decl s0			; Sampler for luma texture
+	 * decl s1			; Sampler for chroma Cb texture
+	 * decl s2			; Sampler for chroma Cr texture
+	 * decl s3			; Sampler for first ref surface texture
+	 * decl s4			; Sampler for second ref surface texture
+	 */
+	for (i = 0; i < 5; ++i)
+	{
+		decl = vl_decl_samplers(i, i);
+		ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/*
+	 * tex2d t1, i0, s0		; Read texel from luma texture
+	 * mov t0.x, t1.x		; Move luma sample into .x component
+	 * tex2d t1, i1, s1		; Read texel from chroma Cb texture
+	 * mov t0.y, t1.x		; Move Cb sample into .y component
+	 * tex2d t1, i2, s2		; Read texel from chroma Cr texture
+	 * mov t0.z, t1.x		; Move Cr sample into .z component
+	 */
+	for (i = 0; i < 3; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+		inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+		inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+		inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* mul t0, t0, c0		; Rescale texel to correct range */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* XXX: Pos values off by 0.5? */
+	/* sub t4, i7.y, c1.x		; Sub 0.5 from denormalized pos */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 7, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t4, c1.x		; Multiply pos Y-coord by 1/2 */
+	inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* floor t3, t3			; Get rid of fractional part */
+	inst = vl_inst2(TGSI_OPCODE_FLOOR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* mul t3, t3, c1.y		; Multiply by 2 */
+	inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1);
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
+	inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* sub t3, t4, t3		; Subtract from original Y to get Y % 2 */
+	inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t1, i3, s3		; Read texel from past ref macroblock top field
+	 * tex2d t2, i4, s3		; Read texel from past ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t1, t3, t1, t2		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/*
+	 * tex2d t4, i5, s4		; Read texel from future ref macroblock top field
+	 * tex2d t5, i6, s4		; Read texel from future ref macroblock bottom field
+	 */
+	for (i = 0; i < 2; ++i)
+	{
+		inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 5, TGSI_FILE_SAMPLER, 4);
+		ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+	}
+
+	/* TODO: Move to conditional tex fetch on t3 instead of lerp */
+	/* lerp t2, t3, t4, t5		; Choose between top and bottom fields based on Y % 2 */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* lerp t1, c1.x, t1, t2	; Blend past and future texels */
+	inst = vl_inst4(TGSI_OPCODE_LERP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
+	inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* add o0, t0, t1		; Add past/future ref and differential to form final output */
+	inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	/* end */
+	inst = vl_end();
+	ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
+
+	fs.tokens = tokens;
+	mc->b_fs[1] = pipe->create_fs_state(pipe, &fs);
+	free(tokens);
+
+	return 0;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_render.h b/src/gallium/state_trackers/g3dvl/vl_render.h
new file mode 100644
index 00000000000..166030b4988
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_render.h
@@ -0,0 +1,38 @@
+#ifndef vl_render_h
+#define vl_render_h
+
+#include "vl_types.h"
+
+struct pipe_surface;
+
+struct vlRender
+{
+	int (*vlBegin)
+	(
+		struct vlRender *render
+	);
+
+	int (*vlRenderMacroBlocksMpeg2)
+	(
+		struct vlRender *render,
+		struct vlMpeg2MacroBlockBatch *batch,
+		struct vlSurface *surface
+	);
+
+	int (*vlEnd)
+	(
+		struct vlRender *render
+	);
+
+	int (*vlFlush)
+	(
+		struct vlRender *render
+	);
+
+	int (*vlDestroy)
+	(
+		struct vlRender *render
+	);
+};
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.c b/src/gallium/state_trackers/g3dvl/vl_screen.c
new file mode 100644
index 00000000000..484f63b0d4d
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_screen.c
@@ -0,0 +1,115 @@
+#define VL_INTERNAL
+#include "vl_screen.h"
+#include <assert.h>
+#include <stdlib.h>
+
+int vlCreateScreen
+(
+	struct vlDisplay *display,
+	int screen,
+	struct pipe_screen *pscreen,
+	struct vlScreen **vl_screen
+)
+{
+	struct vlScreen *scrn;
+
+	assert(display);
+	assert(pscreen);
+	assert(vl_screen);
+
+	scrn = calloc(1, sizeof(struct vlScreen));
+
+	if (!scrn)
+		return 1;
+
+	scrn->display = display;
+	scrn->ordinal = screen;
+	scrn->pscreen = pscreen;
+	*vl_screen = scrn;
+
+	return 0;
+}
+
+int vlDestroyScreen
+(
+	struct vlScreen *screen
+)
+{
+	assert(screen);
+
+	free(screen);
+
+	return 0;
+}
+
+struct vlDisplay* vlGetDisplay
+(
+	struct vlScreen *screen
+)
+{
+	assert(screen);
+
+	return screen->display;
+}
+
+struct pipe_screen* vlGetPipeScreen
+(
+	struct vlScreen *screen
+)
+{
+	assert(screen);
+
+	return screen->pscreen;
+}
+
+unsigned int vlGetMaxProfiles
+(
+	struct vlScreen *screen
+)
+{
+	assert(screen);
+
+	return vlProfileCount;
+}
+
+int vlQueryProfiles
+(
+	struct vlScreen *screen,
+	enum vlProfile *profiles
+)
+{
+	assert(screen);
+	assert(profiles);
+
+	profiles[0] = vlProfileMpeg2Simple;
+	profiles[1] = vlProfileMpeg2Main;
+
+	return 0;
+}
+
+unsigned int vlGetMaxEntryPoints
+(
+	struct vlScreen *screen
+)
+{
+	assert(screen);
+
+	return vlEntryPointCount;
+}
+
+int vlQueryEntryPoints
+(
+	struct vlScreen *screen,
+	enum vlProfile profile,
+	enum vlEntryPoint *entry_points
+)
+{
+	assert(screen);
+	assert(entry_points);
+
+	entry_points[0] = vlEntryPointIDCT;
+	entry_points[1] = vlEntryPointMC;
+	entry_points[2] = vlEntryPointCSC;
+
+	return 0;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.h b/src/gallium/state_trackers/g3dvl/vl_screen.h
new file mode 100644
index 00000000000..98f3d429b61
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_screen.h
@@ -0,0 +1,63 @@
+#ifndef vl_screen_h
+#define vl_screen_h
+
+#include "vl_types.h"
+
+struct pipe_screen;
+
+#ifdef VL_INTERNAL
+struct vlScreen
+{
+	struct vlDisplay	*display;
+	unsigned int		ordinal;
+	struct pipe_screen	*pscreen;
+};
+#endif
+
+int vlCreateScreen
+(
+	struct vlDisplay *display,
+	int screen,
+	struct pipe_screen *pscreen,
+	struct vlScreen **vl_screen
+);
+
+int vlDestroyScreen
+(
+	struct vlScreen *screen
+);
+
+struct vlDisplay* vlGetDisplay
+(
+	struct vlScreen *screen
+);
+
+struct pipe_screen* vlGetPipeScreen
+(
+	struct vlScreen *screen
+);
+
+unsigned int vlGetMaxProfiles
+(
+	struct vlScreen *screen
+);
+
+int vlQueryProfiles
+(
+	struct vlScreen *screen,
+	enum vlProfile *profiles
+);
+
+unsigned int vlGetMaxEntryPoints
+(
+	struct vlScreen *screen
+);
+
+int vlQueryEntryPoints
+(
+	struct vlScreen *screen,
+	enum vlProfile profile,
+	enum vlEntryPoint *entry_points
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c
new file mode 100644
index 00000000000..51f1721a332
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.c
@@ -0,0 +1,215 @@
+#include "vl_shader_build.h"
+#include <assert.h>
+#include <tgsi/tgsi_parse.h>
+#include <tgsi/tgsi_build.h>
+
+struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	decl.Declaration.File = TGSI_FILE_INPUT;
+	decl.Declaration.Semantic = 1;
+	decl.Semantic.SemanticName = name;
+	decl.Semantic.SemanticIndex = index;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_declaration vl_decl_interpolated_input
+(
+	unsigned int name,
+	unsigned int index,
+	unsigned int first,
+	unsigned int last,
+	int interpolation
+)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	assert
+	(
+		interpolation == TGSI_INTERPOLATE_CONSTANT ||
+		interpolation == TGSI_INTERPOLATE_LINEAR ||
+		interpolation == TGSI_INTERPOLATE_PERSPECTIVE
+	);
+
+	decl.Declaration.File = TGSI_FILE_INPUT;
+	decl.Declaration.Semantic = 1;
+	decl.Semantic.SemanticName = name;
+	decl.Semantic.SemanticIndex = index;
+	decl.Declaration.Interpolate = interpolation;;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	decl.Declaration.File = TGSI_FILE_CONSTANT;
+	decl.Declaration.Semantic = 1;
+	decl.Semantic.SemanticName = name;
+	decl.Semantic.SemanticIndex = index;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	decl.Declaration.File = TGSI_FILE_OUTPUT;
+	decl.Declaration.Semantic = 1;
+	decl.Semantic.SemanticName = name;
+	decl.Semantic.SemanticIndex = index;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	decl = tgsi_default_full_declaration();
+	decl.Declaration.File = TGSI_FILE_TEMPORARY;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last)
+{
+	struct tgsi_full_declaration decl = tgsi_default_full_declaration();
+
+	decl = tgsi_default_full_declaration();
+	decl.Declaration.File = TGSI_FILE_SAMPLER;
+	decl.DeclarationRange.First = first;
+	decl.DeclarationRange.Last = last;
+
+	return decl;
+}
+
+struct tgsi_full_instruction vl_inst2
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src_file,
+	unsigned int src_index
+)
+{
+	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+	inst.Instruction.Opcode = opcode;
+	inst.Instruction.NumDstRegs = 1;
+	inst.FullDstRegisters[0].DstRegister.File = dst_file;
+	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+	inst.Instruction.NumSrcRegs = 1;
+	inst.FullSrcRegisters[0].SrcRegister.File = src_file;
+	inst.FullSrcRegisters[0].SrcRegister.Index = src_index;
+
+	return inst;
+}
+
+struct tgsi_full_instruction vl_inst3
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index
+)
+{
+	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+	inst.Instruction.Opcode = opcode;
+	inst.Instruction.NumDstRegs = 1;
+	inst.FullDstRegisters[0].DstRegister.File = dst_file;
+	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+	inst.Instruction.NumSrcRegs = 2;
+	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+
+	return inst;
+}
+
+struct tgsi_full_instruction vl_tex
+(
+	int tex,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index
+)
+{
+	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+	inst.Instruction.Opcode = TGSI_OPCODE_TEX;
+	inst.Instruction.NumDstRegs = 1;
+	inst.FullDstRegisters[0].DstRegister.File = dst_file;
+	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+	inst.Instruction.NumSrcRegs = 2;
+	inst.InstructionExtTexture.Texture = tex;
+	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+
+	return inst;
+}
+
+struct tgsi_full_instruction vl_inst4
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index,
+	enum tgsi_file_type src3_file,
+	unsigned int src3_index
+)
+{
+	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+	inst.Instruction.Opcode = opcode;
+	inst.Instruction.NumDstRegs = 1;
+	inst.FullDstRegisters[0].DstRegister.File = dst_file;
+	inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+	inst.Instruction.NumSrcRegs = 3;
+	inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
+	inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
+	inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
+	inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+	inst.FullSrcRegisters[2].SrcRegister.File = src3_file;
+	inst.FullSrcRegisters[2].SrcRegister.Index = src3_index;
+
+	return inst;
+}
+
+struct tgsi_full_instruction vl_end(void)
+{
+	struct tgsi_full_instruction inst = tgsi_default_full_instruction();
+
+	inst.Instruction.Opcode = TGSI_OPCODE_END;
+	inst.Instruction.NumDstRegs = 0;
+	inst.Instruction.NumSrcRegs = 0;
+
+	return inst;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h
new file mode 100644
index 00000000000..dc615cb1566
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_shader_build.h
@@ -0,0 +1,61 @@
+#ifndef vl_shader_build_h
+#define vl_shader_build_h
+
+#include <pipe/p_shader_tokens.h>
+
+struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_interpolated_input
+(
+	unsigned int name,
+	unsigned int index,
+	unsigned int first,
+	unsigned int last,
+	int interpolation
+);
+struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last);
+struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last);
+struct tgsi_full_instruction vl_inst2
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src_file,
+	unsigned int src_index
+);
+struct tgsi_full_instruction vl_inst3
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index
+);
+struct tgsi_full_instruction vl_tex
+(
+	int tex,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index
+);
+struct tgsi_full_instruction vl_inst4
+(
+	int opcode,
+	enum tgsi_file_type dst_file,
+	unsigned int dst_index,
+	enum tgsi_file_type src1_file,
+	unsigned int src1_index,
+	enum tgsi_file_type src2_file,
+	unsigned int src2_index,
+	enum tgsi_file_type src3_file,
+	unsigned int src3_index
+);
+struct tgsi_full_instruction vl_end(void);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c
new file mode 100644
index 00000000000..076bd40d414
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_surface.c
@@ -0,0 +1,237 @@
+#define VL_INTERNAL
+#include "vl_surface.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <pipe/p_screen.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <vl_winsys.h>
+#include "vl_screen.h"
+#include "vl_context.h"
+#include "vl_render.h"
+#include "vl_csc.h"
+#include "vl_util.h"
+
+int vlCreateSurface
+(
+	struct vlScreen *screen,
+	unsigned int width,
+	unsigned int height,
+	enum vlFormat format,
+	struct vlSurface **surface
+)
+{
+	struct vlSurface	*sfc;
+	struct pipe_texture	template;
+
+	assert(screen);
+	assert(surface);
+
+	sfc = calloc(1, sizeof(struct vlSurface));
+
+	if (!sfc)
+		return 1;
+
+	sfc->screen = screen;
+	sfc->width = width;
+	sfc->height = height;
+	sfc->format = format;
+
+	memset(&template, 0, sizeof(struct pipe_texture));
+	template.target = PIPE_TEXTURE_2D;
+	template.format = PIPE_FORMAT_A8R8G8B8_UNORM;
+	template.last_level = 0;
+	template.width[0] = vlRoundUpPOT(sfc->width);
+	template.height[0] = vlRoundUpPOT(sfc->height);
+	template.depth[0] = 1;
+	template.compressed = 0;
+	pf_get_block(template.format, &template.block);
+	template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+	sfc->texture = vlGetPipeScreen(screen)->texture_create(vlGetPipeScreen(screen), &template);
+
+	*surface = sfc;
+
+	return 0;
+}
+
+int vlDestroySurface
+(
+	struct vlSurface *surface
+)
+{
+	assert(surface);
+
+	pipe_texture_release(&surface->texture);
+	free(surface);
+
+	return 0;
+}
+
+int vlRenderMacroBlocksMpeg2
+(
+	struct vlMpeg2MacroBlockBatch *batch,
+	struct vlSurface *surface
+)
+{
+	assert(batch);
+	assert(surface);
+	assert(surface->context);
+
+	surface->context->render->vlBegin(surface->context->render);
+
+	surface->context->render->vlRenderMacroBlocksMpeg2
+	(
+		surface->context->render,
+		batch,
+		surface
+	);
+
+	surface->context->render->vlEnd(surface->context->render);
+
+	return 0;
+}
+
+int vlPutPicture
+(
+	struct vlSurface *surface,
+	vlNativeDrawable drawable,
+	int srcx,
+	int srcy,
+	int srcw,
+	int srch,
+	int destx,
+	int desty,
+	int destw,
+	int desth,
+	int drawable_w,
+	int drawable_h,
+	enum vlPictureType picture_type
+)
+{
+	struct vlCSC		*csc;
+	struct pipe_context	*pipe;
+
+	assert(surface);
+	assert(surface->context);
+
+	surface->context->render->vlFlush(surface->context->render);
+
+	csc = surface->context->csc;
+	pipe = surface->context->pipe;
+
+	csc->vlResizeFrameBuffer(csc, drawable_w, drawable_h);
+
+	csc->vlBegin(csc);
+
+	csc->vlPutPicture
+	(
+		csc,
+		surface,
+		srcx,
+		srcy,
+		srcw,
+		srch,
+		destx,
+		desty,
+		destw,
+		desth,
+		picture_type
+	);
+
+	csc->vlEnd(csc);
+
+	pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &surface->disp_fence);
+
+	bind_pipe_drawable(pipe, drawable);
+
+	pipe->winsys->flush_frontbuffer
+	(
+		pipe->winsys,
+		csc->vlGetFrameBuffer(csc),
+		pipe->priv
+	);
+
+	return 0;
+}
+
+int vlSurfaceGetStatus
+(
+	struct vlSurface *surface,
+	enum vlResourceStatus *status
+)
+{
+	assert(surface);
+	assert(surface->context);
+	assert(status);
+
+	if (surface->render_fence && !surface->context->pipe->winsys->fence_signalled(surface->context->pipe->winsys, surface->render_fence, 0))
+	{
+		*status = vlResourceStatusRendering;
+		return 0;
+	}
+
+	if (surface->disp_fence && !surface->context->pipe->winsys->fence_signalled(surface->context->pipe->winsys, surface->disp_fence, 0))
+	{
+		*status = vlResourceStatusDisplaying;
+		return 0;
+	}
+
+	*status = vlResourceStatusFree;
+
+	return 0;
+}
+
+int vlSurfaceFlush
+(
+	struct vlSurface *surface
+)
+{
+	assert(surface);
+	assert(surface->context);
+
+	surface->context->render->vlFlush(surface->context->render);
+
+	return 0;
+}
+
+int vlSurfaceSync
+(
+	struct vlSurface *surface
+)
+{
+	assert(surface);
+	assert(surface->context);
+	assert(surface->render_fence);
+
+	surface->context->pipe->winsys->fence_finish(surface->context->pipe->winsys, surface->render_fence, 0);
+
+	return 0;
+}
+
+struct vlScreen* vlSurfaceGetScreen
+(
+	struct vlSurface *surface
+)
+{
+	assert(surface);
+
+	return surface->screen;
+}
+
+struct vlContext* vlBindToContext
+(
+	struct vlSurface *surface,
+	struct vlContext *context
+)
+{
+	struct vlContext *old;
+
+	assert(surface);
+
+	old = surface->context;
+	surface->context = context;
+
+	return old;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h
new file mode 100644
index 00000000000..133e1515ef3
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_surface.h
@@ -0,0 +1,86 @@
+#ifndef vl_surface_h
+#define vl_surface_h
+
+#include "vl_types.h"
+
+#ifdef VL_INTERNAL
+struct pipe_texture;
+
+struct vlSurface
+{
+	struct vlScreen			*screen;
+	struct vlContext		*context;
+	unsigned int			width;
+	unsigned int			height;
+	enum vlFormat			format;
+	struct pipe_texture		*texture;
+	struct pipe_fence_handle	*render_fence;
+	struct pipe_fence_handle	*disp_fence;
+};
+#endif
+
+int vlCreateSurface
+(
+	struct vlScreen *screen,
+	unsigned int width,
+	unsigned int height,
+	enum vlFormat format,
+	struct vlSurface **surface
+);
+
+int vlDestroySurface
+(
+	struct vlSurface *surface
+);
+
+int vlRenderMacroBlocksMpeg2
+(
+	struct vlMpeg2MacroBlockBatch *batch,
+	struct vlSurface *surface
+);
+
+int vlPutPicture
+(
+	struct vlSurface *surface,
+	vlNativeDrawable drawable,
+	int srcx,
+	int srcy,
+	int srcw,
+	int srch,
+	int destx,
+	int desty,
+	int destw,
+	int desth,
+	int drawable_w,
+	int drawable_h,
+	enum vlPictureType picture_type
+);
+
+int vlSurfaceGetStatus
+(
+	struct vlSurface *surface,
+	enum vlResourceStatus *status
+);
+
+int vlSurfaceFlush
+(
+	struct vlSurface *surface
+);
+
+int vlSurfaceSync
+(
+	struct vlSurface *surface
+);
+
+struct vlScreen* vlSurfaceGetScreen
+(
+	struct vlSurface *surface
+);
+
+struct vlContext* vlBindToContext
+(
+	struct vlSurface *surface,
+	struct vlContext *context
+);
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h
new file mode 100644
index 00000000000..274e1f74377
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_types.h
@@ -0,0 +1,115 @@
+#ifndef vl_types_h
+#define vl_types_h
+
+#if 1 /*#ifdef X11*/
+#include <X11/Xlib.h>
+
+typedef Display* vlNativeDisplay;
+typedef Drawable vlNativeDrawable;
+#endif
+
+struct vlDisplay;
+struct vlScreen;
+struct vlContext;
+struct vlSurface;
+
+enum vlResourceStatus
+{
+	vlResourceStatusFree,
+	vlResourceStatusRendering,
+	vlResourceStatusDisplaying
+};
+
+enum vlProfile
+{
+	vlProfileMpeg2Simple,
+	vlProfileMpeg2Main,
+
+	vlProfileCount
+};
+
+enum vlEntryPoint
+{
+	vlEntryPointIDCT,
+	vlEntryPointMC,
+	vlEntryPointCSC,
+
+	vlEntryPointCount
+};
+
+enum vlFormat
+{
+	vlFormatYCbCr420,
+	vlFormatYCbCr422,
+	vlFormatYCbCr444
+};
+
+enum vlPictureType
+{
+	vlPictureTypeTopField,
+	vlPictureTypeBottomField,
+	vlPictureTypeFrame
+};
+
+enum vlMotionType
+{
+	vlMotionTypeField,
+	vlMotionTypeFrame,
+	vlMotionTypeDualPrime,
+	vlMotionType16x8
+};
+
+enum vlFieldOrder
+{
+	vlFieldOrderFirst,
+	vlFieldOrderSecond
+};
+
+enum vlDCTType
+{
+	vlDCTTypeFrameCoded,
+	vlDCTTypeFieldCoded
+};
+
+struct vlVertex2f
+{
+	float x, y;
+};
+
+struct vlVertex4f
+{
+	float x, y, z, w;
+};
+
+enum vlMacroBlockType
+{
+	vlMacroBlockTypeIntra,
+	vlMacroBlockTypeFwdPredicted,
+	vlMacroBlockTypeBkwdPredicted,
+	vlMacroBlockTypeBiPredicted,
+
+	vlNumMacroBlockTypes
+};
+
+struct vlMpeg2MacroBlock
+{
+	unsigned int		mbx, mby;
+	enum vlMacroBlockType	mb_type;
+	enum vlMotionType	mo_type;
+	enum vlDCTType		dct_type;
+	int			PMV[2][2][2];
+	unsigned int		cbp;
+	short			*blocks;
+};
+
+struct vlMpeg2MacroBlockBatch
+{
+	struct vlSurface		*past_surface;
+	struct vlSurface		*future_surface;
+	enum vlPictureType		picture_type;
+	enum vlFieldOrder		field_order;
+	unsigned int			num_macroblocks;
+	struct vlMpeg2MacroBlock	*macroblocks;
+};
+
+#endif
diff --git a/src/gallium/state_trackers/g3dvl/vl_util.c b/src/gallium/state_trackers/g3dvl/vl_util.c
new file mode 100644
index 00000000000..50aa9af66f2
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_util.c
@@ -0,0 +1,16 @@
+#include "vl_util.h"
+#include <assert.h>
+
+unsigned int vlRoundUpPOT(unsigned int x)
+{
+	unsigned int i;
+
+	assert(x > 0);
+
+	--x;
+
+	for (i = 1; i < sizeof(unsigned int) * 8; i <<= 1)
+		x |= x >> i;
+
+	return x + 1;
+}
diff --git a/src/gallium/state_trackers/g3dvl/vl_util.h b/src/gallium/state_trackers/g3dvl/vl_util.h
new file mode 100644
index 00000000000..bc98e79df47
--- /dev/null
+++ b/src/gallium/state_trackers/g3dvl/vl_util.h
@@ -0,0 +1,6 @@
+#ifndef vl_util_h
+#define vl_util_h
+
+unsigned int vlRoundUpPOT(unsigned int x);
+
+#endif
diff --git a/src/gallium/state_trackers/python/p_format.i b/src/gallium/state_trackers/python/p_format.i
index 51ad4bebcdc..26fb12b387f 100644
--- a/src/gallium/state_trackers/python/p_format.i
+++ b/src/gallium/state_trackers/python/p_format.i
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright (c) 2008 VMware, Inc.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -128,10 +129,14 @@ enum pipe_format {
    PIPE_FORMAT_R32G32B32A32_FIXED,
 
    PIPE_FORMAT_L8_SRGB,
-   PIPE_FORMAT_A8_L8_SRGB,
+   PIPE_FORMAT_A8L8_SRGB,
    PIPE_FORMAT_R8G8B8_SRGB,
    PIPE_FORMAT_R8G8B8A8_SRGB,
    PIPE_FORMAT_R8G8B8X8_SRGB,
+   PIPE_FORMAT_A8R8G8B8_SRGB,
+   PIPE_FORMAT_X8R8G8B8_SRGB,
+   PIPE_FORMAT_B8G8R8A8_SRGB,
+   PIPE_FORMAT_B8G8R8X8_SRGB,
 
    PIPE_FORMAT_X8UB8UG8SR8S_NORM,
    PIPE_FORMAT_B6UG5SR5S_NORM,
@@ -140,6 +145,11 @@ enum pipe_format {
    PIPE_FORMAT_DXT1_RGBA,
    PIPE_FORMAT_DXT3_RGBA,
    PIPE_FORMAT_DXT5_RGBA,
+
+   PIPE_FORMAT_DXT1_SRGB,
+   PIPE_FORMAT_DXT1_SRGBA,
+   PIPE_FORMAT_DXT3_SRGBA,
+   PIPE_FORMAT_DXT5_SRGBA,
 };
 
 
diff --git a/src/gallium/winsys/drm/intel/dri/intel_screen.c b/src/gallium/winsys/drm/intel/dri/intel_screen.c
index 3a486481f56..ed753689829 100644
--- a/src/gallium/winsys/drm/intel/dri/intel_screen.c
+++ b/src/gallium/winsys/drm/intel/dri/intel_screen.c
@@ -113,7 +113,118 @@ static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
 
 extern const struct dri_extension card_extensions[];
 
+static GLboolean
+intel_get_param(__DRIscreenPrivate *psp, int param, int *value)
+{
+   int ret;
+   struct drm_i915_getparam gp;
+
+   gp.param = param;
+   gp.value = value;
+
+   ret = drmCommandWriteRead(psp->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
+   if (ret) {
+      fprintf(stderr, "drm_i915_getparam: %d\n", ret);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+static void
+intelSetTexOffset(__DRIcontext *pDRICtx, int texname,
+		  unsigned long long offset, int depth, uint pitch)
+{
+   abort();
+#if 0
+   struct intel_context *intel = (struct intel_context*)
+      ((__DRIcontextPrivate*)pDRICtx->private)->driverPrivate;
+   struct gl_texture_object *tObj = _mesa_lookup_texture(&intel->ctx, texname);
+   struct st_texture_object *stObj = st_texture_object(tObj);
+
+   if (!stObj)
+      return;
+
+   if (stObj->pt)
+      st->pipe->texture_release(intel->st->pipe, &stObj->pt);
+
+   stObj->imageOverride = GL_TRUE;
+   stObj->depthOverride = depth;
+   stObj->pitchOverride = pitch;
+
+   if (offset)
+      stObj->textureOffset = offset;
+#endif
+}
+
+
+#if 0
+static void
+intelHandleDrawableConfig(__DRIdrawablePrivate *dPriv,
+			  __DRIcontextPrivate *pcp,
+			  __DRIDrawableConfigEvent *event)
+{
+   (void) dPriv;
+   (void) pcp;
+   (void) event;
+}
+#endif
+
+#if 0
+static void
+intelHandleBufferAttach(__DRIdrawablePrivate *dPriv,
+			__DRIcontextPrivate *pcp,
+			__DRIBufferAttachEvent *ba)
+{
+   struct intel_screen *intelScreen = intel_screen(dPriv->driScreenPriv);
+
+   switch (ba->buffer.attachment) {
+   case DRI_DRAWABLE_BUFFER_FRONT_LEFT:
+      intelScreen->front.width = dPriv->w;
+      intelScreen->front.height = dPriv->h;
+      intelScreen->front.cpp = ba->buffer.cpp;
+      intelScreen->front.pitch = ba->buffer.pitch;
+      driGenBuffers(intelScreen->base.staticPool, "front", 1, &intelScreen->front.buffer, 0, 0, 0);
+      driBOSetReferenced(intelScreen->front.buffer, ba->buffer.handle);
+      break;
+
+   case DRI_DRAWABLE_BUFFER_BACK_LEFT:
+   case DRI_DRAWABLE_BUFFER_DEPTH:
+   case DRI_DRAWABLE_BUFFER_STENCIL:
+   case DRI_DRAWABLE_BUFFER_ACCUM:
+      /* anything ?? */
+      break;
+
+   default:
+      fprintf(stderr, "unhandled buffer attach event, attachment type %d\n",
+	      ba->buffer.attachment);
+      return;
+   }
+}
+#endif
+
+static const __DRItexOffsetExtension intelTexOffsetExtension = {
+   { __DRI_TEX_OFFSET },
+   intelSetTexOffset,
+};
+
+#if 0
+static const __DRItexBufferExtension intelTexBufferExtension = {
+    { __DRI_TEX_BUFFER, __DRI_TEX_BUFFER_VERSION },
+   intelSetTexBuffer,
+};
+#endif
 
+static const __DRIextension *intelScreenExtensions[] = {
+    &driReadDrawableExtension,
+    &driCopySubBufferExtension.base,
+    &driSwapControlExtension.base,
+    &driFrameTrackingExtension.base,
+    &driMediaStreamCounterExtension.base,
+    &intelTexOffsetExtension.base,
+//    &intelTexBufferExtension.base,
+    NULL
+};
 
 
 static void
@@ -232,7 +343,8 @@ intelCreatePools(__DRIscreenPrivate * sPriv)
 
    intelScreen->havePools = GL_TRUE;
 
-   intelUpdateScreenRotation(sPriv, intelScreen->sarea);
+   if (intelScreen->sarea)
+	intelUpdateScreenRotation(sPriv, intelScreen->sarea);
 
    return GL_TRUE;
 }
@@ -265,11 +377,6 @@ intelInitDriver(__DRIscreenPrivate * sPriv)
    struct intel_screen *intelScreen;
    I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
 
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-      (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->
-                                      getProcAddress("glxEnableExtension"));
-   void *const psc = sPriv->psc->screenConfigs;
-
    if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
       fprintf(stderr,
               "\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
@@ -286,28 +393,19 @@ intelInitDriver(__DRIscreenPrivate * sPriv)
                       __driConfigOptions, __driNConfigOptions);
 
    sPriv->private = (void *) intelScreen;
-
    intelScreen->sarea = (drmI830Sarea *) (((GLubyte *) sPriv->pSAREA) +
-					  gDRIPriv->sarea_priv_offset);
-   intelScreen->deviceID = gDRIPriv->deviceID;
-   intelScreen->front.cpp = gDRIPriv->cpp;
-   intelScreen->drmMinor = sPriv->drmMinor;
+                                            gDRIPriv->sarea_priv_offset);
 
-   assert(gDRIPriv->bitsPerPixel == 16 ||
-	  gDRIPriv->bitsPerPixel == 32);
+   intelScreen->deviceID = gDRIPriv->deviceID;
 
+   intelScreen->front.cpp = gDRIPriv->cpp;
+   intelScreen->drmMinor = sPriv->drm_version.minor;
    intelUpdateScreenRotation(sPriv, intelScreen->sarea);
 
    if (0)
       intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
 
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension) (psc, "GLX_SGI_swap_control");
-      (*glx_enable_extension) (psc, "GLX_SGI_video_sync");
-      (*glx_enable_extension) (psc, "GLX_MESA_swap_control");
-      (*glx_enable_extension) (psc, "GLX_MESA_swap_frame_usage");
-      (*glx_enable_extension) (psc, "GLX_SGI_make_current_read");
-   }
+   sPriv->extensions = intelScreenExtensions;
 
    intelScreen->base.base.flush_frontbuffer = intel_flush_frontbuffer;
    intelScreen->base.base.get_name = intel_get_name;
@@ -387,7 +485,7 @@ intelDestroyBuffer(__DRIdrawablePrivate * driDrawPriv)
 {
    struct intel_framebuffer *intelfb = intel_framebuffer(driDrawPriv);
    assert(intelfb->stfb);
-   st_unreference_framebuffer(&intelfb->stfb);
+   st_unreference_framebuffer(intelfb->stfb);
    free(intelfb);
 }
 
@@ -406,65 +504,19 @@ intelGetSwapInfo(__DRIdrawablePrivate * dPriv, __DRIswapInfo * sInfo)
    return 0;
 }
 
-
-static void
-intelSetTexOffset(__DRIcontext *pDRICtx, int texname,
-		  unsigned long long offset, int depth, uint pitch)
-{
-   abort();
-#if 0
-   struct intel_context *intel = (struct intel_context*)
-      ((__DRIcontextPrivate*)pDRICtx->private)->driverPrivate;
-   struct gl_texture_object *tObj = _mesa_lookup_texture(&intel->ctx, texname);
-   struct st_texture_object *stObj = st_texture_object(tObj);
-
-   if (!stObj)
-      return;
-
-   if (stObj->pt)
-      st->pipe->texture_release(intel->st->pipe, &stObj->pt);
-
-   stObj->imageOverride = GL_TRUE;
-   stObj->depthOverride = depth;
-   stObj->pitchOverride = pitch;
-
-   if (offset)
-      stObj->textureOffset = offset;
-#endif
-}
-
-
-static const struct __DriverAPIRec intelAPI = {
-   .InitDriver = intelInitDriver,
-   .DestroyScreen = intelDestroyScreen,
-   .CreateContext = intelCreateContext,
-   .DestroyContext = intelDestroyContext,
-   .CreateBuffer = intelCreateBuffer,
-   .DestroyBuffer = intelDestroyBuffer,
-   .SwapBuffers = intelSwapBuffers,
-   .MakeCurrent = intelMakeCurrent,
-   .UnbindContext = intelUnbindContext,
-   .GetSwapInfo = intelGetSwapInfo,
-   .GetMSC = driGetMSC32,
-   .WaitForMSC = driWaitForMSC32,
-   .WaitForSBC = NULL,
-   .SwapBuffersMSC = NULL,
-   .CopySubBuffer = intelCopySubBuffer,
-   .setTexOffset = intelSetTexOffset,
-};
-
-
-static __GLcontextModes *
-intelFillInModes(unsigned pixel_bits, unsigned depth_bits,
-                 unsigned stencil_bits, boolean have_back_buffer)
+static __DRIconfig **
+intelFillInModes(__DRIscreenPrivate *psp,
+		 unsigned pixel_bits, unsigned depth_bits,
+                 unsigned stencil_bits, GLboolean have_back_buffer)
 {
-   __GLcontextModes *modes;
+   __DRIconfig **configs;
    __GLcontextModes *m;
    unsigned num_modes;
    unsigned depth_buffer_factor;
    unsigned back_buffer_factor;
    GLenum fb_format;
    GLenum fb_type;
+   int i;
 
    /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
     * support pageflipping at all.
@@ -508,100 +560,144 @@ intelFillInModes(unsigned pixel_bits, unsigned depth_bits,
       fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
    }
 
-   modes =
-      (*dri_interface->createContextModes) (num_modes,
-                                            sizeof(__GLcontextModes));
-   m = modes;
-   if (!driFillInModes(&m, fb_format, fb_type,
-                       depth_bits_array, stencil_bits_array,
-                       depth_buffer_factor, back_buffer_modes,
-                       back_buffer_factor, msaa_samples_array, 1, GLX_TRUE_COLOR)) {
-      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
-              __LINE__);
-      return NULL;
-   }
-   if (!driFillInModes(&m, fb_format, fb_type,
-                       depth_bits_array, stencil_bits_array,
-                       depth_buffer_factor, back_buffer_modes,
-                       back_buffer_factor, msaa_samples_array, 1, GLX_DIRECT_COLOR)) {
-      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+   configs = driCreateConfigs(fb_format, fb_type,
+			      depth_bits_array, stencil_bits_array,
+			      depth_buffer_factor, back_buffer_modes,
+			      back_buffer_factor, msaa_samples_array, 1);
+   if (configs == NULL) {
+    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
               __LINE__);
       return NULL;
    }
 
    /* Mark the visual as slow if there are "fake" stencil bits.
     */
-   for (m = modes; m != NULL; m = m->next) {
+   for (i = 0; configs[i]; i++) {
+      m = &configs[i]->modes;
       if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
          m->visualRating = GLX_SLOW_CONFIG;
       }
    }
 
-   return modes;
+   return configs;
 }
 
-
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \todo maybe fold this into intelInitDriver
  *
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on
- *         failure.
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC void *
-__driCreateNewScreen_20050727(__DRInativeDisplay * dpy, int scrn,
-                              __DRIscreen * psc,
-                              const __GLcontextModes * modes,
-                              const __DRIversion * ddx_version,
-                              const __DRIversion * dri_version,
-                              const __DRIversion * drm_version,
-                              const __DRIframebuffer * frame_buffer,
-                              drmAddress pSAREA, int fd,
-                              int internal_api_version,
-                              const __DRIinterfaceMethods * interface,
-                              __GLcontextModes ** driver_modes)
+static const __DRIconfig **intelInitScreen(__DRIscreenPrivate *psp)
 {
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 1, 7, 0 };
+#ifdef I915
+   static const __DRIversion ddx_expected = { 1, 5, 0 };
+#else
+   static const __DRIversion ddx_expected = { 1, 6, 0 };
+#endif
    static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 7, 0 };
-
-   dri_interface = interface;
+   static const __DRIversion drm_expected = { 1, 5, 0 };
+   I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
 
    if (!driCheckDriDdxDrmVersions2("i915",
-                                   dri_version, &dri_expected,
-                                   ddx_version, &ddx_expected,
-                                   drm_version, &drm_expected)) {
+                                   &psp->dri_version, &dri_expected,
+                                   &psp->ddx_version, &ddx_expected,
+                                   &psp->drm_version, &drm_expected)) {
       return NULL;
    }
 
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-                                  ddx_version, dri_version, drm_version,
-                                  frame_buffer, pSAREA, fd,
-                                  internal_api_version, &intelAPI);
-
-   if (psp != NULL) {
-      I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-      *driver_modes = intelFillInModes(dri_priv->cpp * 8,
-                                       (dri_priv->cpp == 2) ? 16 : 24,
-                                       (dri_priv->cpp == 2) ? 0 : 8, 1);
-
-      /* Calling driInitExtensions here, with a NULL context pointer,
-       * does not actually enable the extensions.  It just makes sure
-       * that all the dispatch offsets for all the extensions that
-       * *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create
-       * is called, but we can't enable the extensions until we have a
-       * context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions(NULL, card_extensions, GL_FALSE);
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+   //intelInitExtensions(NULL, GL_TRUE);
+	   
+   if (!intelInitDriver(psp))
+       return NULL;
+
+   psp->extensions = intelScreenExtensions;
+
+   return (const __DRIconfig **)
+       intelFillInModes(psp, dri_priv->cpp * 8,
+			(dri_priv->cpp == 2) ? 16 : 24,
+			(dri_priv->cpp == 2) ? 0  : 8, 1);
+}
+
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \return the __GLcontextModes supported by this driver
+ */
+static const
+__DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
+{
+   struct intel_screen *intelScreen;
+
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   //intelInitExtensions(NULL, GL_TRUE);
+
+   /* Allocate the private area */
+   intelScreen = CALLOC_STRUCT(intel_screen);
+   if (!intelScreen) {
+      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
+      return GL_FALSE;
    }
+   /* parse information in __driConfigOptions */
+   driParseOptionInfo(&intelScreen->optionCache,
+                      __driConfigOptions, __driNConfigOptions);
+
+   psp->private = (void *) intelScreen;
+
+   intelScreen->drmMinor = psp->drm_version.minor;
 
-   return (void *) psp;
+   /* Determine chipset ID? */
+   if (!intel_get_param(psp, I915_PARAM_CHIPSET_ID,
+			&intelScreen->deviceID))
+      return GL_FALSE;
+
+   psp->extensions = intelScreenExtensions;
+
+   intel_be_init_device(&intelScreen->base, psp->fd, intelScreen->deviceID);
+   intelScreen->base.base.flush_frontbuffer = intel_flush_frontbuffer;
+   intelScreen->base.base.get_name = intel_get_name;
+
+   return driConcatConfigs(intelFillInModes(psp, 16, 16, 0, 1),
+			   intelFillInModes(psp, 32, 24, 8, 1));
 }
 
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen		 = intelInitScreen,
+   .DestroyScreen	 = intelDestroyScreen,
+   .CreateContext	 = intelCreateContext,
+   .DestroyContext	 = intelDestroyContext,
+   .CreateBuffer	 = intelCreateBuffer,
+   .DestroyBuffer	 = intelDestroyBuffer,
+   .SwapBuffers		 = intelSwapBuffers,
+   .MakeCurrent		 = intelMakeCurrent,
+   .UnbindContext	 = intelUnbindContext,
+   .GetSwapInfo		 = intelGetSwapInfo,
+   .GetDrawableMSC	 = driDrawableGetMSC32,
+   .WaitForMSC		 = driWaitForMSC32,
+   .CopySubBuffer	 = intelCopySubBuffer,
+
+   //.InitScreen2		 = intelInitScreen2,
+   //.HandleDrawableConfig = intelHandleDrawableConfig,
+   //.HandleBufferAttach	 = intelHandleBufferAttach,
+};
diff --git a/src/gallium/winsys/drm/intel/egl/intel_device.c b/src/gallium/winsys/drm/intel/egl/intel_device.c
index b9649cbec71..1964745c994 100644
--- a/src/gallium/winsys/drm/intel/egl/intel_device.c
+++ b/src/gallium/winsys/drm/intel/egl/intel_device.c
@@ -131,7 +131,7 @@ intel_destroy_drawable(struct egl_drm_drawable *drawable)
 	drawable->priv = NULL;
 
 	assert(intelfb->stfb);
-	st_unreference_framebuffer(&intelfb->stfb);
+	st_unreference_framebuffer(intelfb->stfb);
 	free(intelfb);
 	return TRUE;
 }
diff --git a/src/gallium/winsys/drm/intel/egl/intel_egl.c b/src/gallium/winsys/drm/intel/egl/intel_egl.c
index 3204ed3131c..ed464076ee3 100644
--- a/src/gallium/winsys/drm/intel/egl/intel_egl.c
+++ b/src/gallium/winsys/drm/intel/egl/intel_egl.c
@@ -565,7 +565,7 @@ static struct drm_mode_modeinfo *
 drm_find_mode(drmModeConnectorPtr connector, _EGLMode *mode)
 {
 	int i;
-	struct drm_mode_modeinfo *m;
+	struct drm_mode_modeinfo *m = NULL;
 
 	for (i = 0; i < connector->count_modes; i++) {
 		m = &connector->modes[i];
diff --git a/src/gallium/winsys/drm/nouveau/Makefile b/src/gallium/winsys/drm/nouveau/Makefile
new file mode 100644
index 00000000000..81562ca78d3
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/Makefile
@@ -0,0 +1,46 @@
+
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = nouveau_dri.so
+
+MINIGLX_SOURCES =
+
+PIPE_DRIVERS = \
+	$(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \
+	$(TOP)/src/gallium/drivers/nv04/libnv04.a \
+	$(TOP)/src/gallium/drivers/nv10/libnv10.a \
+	$(TOP)/src/gallium/drivers/nv20/libnv20.a \
+	$(TOP)/src/gallium/drivers/nv30/libnv30.a \
+	$(TOP)/src/gallium/drivers/nv40/libnv40.a \
+	$(TOP)/src/gallium/drivers/nv50/libnv50.a
+
+DRIVER_SOURCES = \
+	nouveau_bo.c \
+	nouveau_channel.c \
+	nouveau_context.c \
+	nouveau_device.c \
+	nouveau_dma.c \
+	nouveau_fence.c \
+	nouveau_grobj.c \
+	nouveau_lock.c \
+	nouveau_notifier.c \
+	nouveau_pushbuf.c \
+	nouveau_resource.c \
+	nouveau_screen.c \
+	nouveau_swapbuffers.c \
+	nouveau_winsys.c \
+	nouveau_winsys_pipe.c \
+	nouveau_winsys_softpipe.c \
+	nv04_surface.c \
+	nv50_surface.c
+
+C_SOURCES = \
+	$(COMMON_GALLIUM_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES = 
+
+include ../Makefile.template
+
+symlinks:
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_bo.c b/src/gallium/winsys/drm/nouveau/nouveau_bo.c
new file mode 100644
index 00000000000..b5942994d94
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_bo.c
@@ -0,0 +1,470 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <assert.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+#include "nouveau_local.h"
+
+static void
+nouveau_mem_free(struct nouveau_device *dev, struct drm_nouveau_mem_alloc *ma,
+		 void **map)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	struct drm_nouveau_mem_free mf;
+
+	if (map && *map) {
+		drmUnmap(*map, ma->size);
+		*map = NULL;
+	}
+
+	if (ma->size) {
+		mf.offset = ma->offset;
+		mf.flags = ma->flags;
+		drmCommandWrite(nvdev->fd, DRM_NOUVEAU_MEM_FREE,
+				&mf, sizeof(mf));
+		ma->size = 0;
+	}
+}
+
+static int
+nouveau_mem_alloc(struct nouveau_device *dev, unsigned size, unsigned align,
+		  uint32_t flags, struct drm_nouveau_mem_alloc *ma, void **map)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	int ret;
+
+	ma->alignment = align;
+	ma->size = size;
+	ma->flags = flags;
+	if (map)
+		ma->flags |= NOUVEAU_MEM_MAPPED;
+	ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_MEM_ALLOC, ma,
+				  sizeof(struct drm_nouveau_mem_alloc));
+	if (ret)
+		return ret;
+
+	if (map) {
+		ret = drmMap(nvdev->fd, ma->map_handle, ma->size, map);
+		if (ret) {
+			*map = NULL;
+			nouveau_mem_free(dev, ma, map);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
+static void
+nouveau_bo_tmp_del(void *priv)
+{
+	struct nouveau_resource *r = priv;
+
+	nouveau_fence_ref(NULL, (struct nouveau_fence **)&r->priv);
+	nouveau_resource_free(&r);
+}
+
+static unsigned
+nouveau_bo_tmp_max(struct nouveau_device_priv *nvdev)
+{
+	struct nouveau_resource *r = nvdev->sa_heap;
+	unsigned max = 0;
+
+	while (r) {
+		if (r->in_use && !nouveau_fence(r->priv)->emitted) {
+			r = r->next;
+			continue;
+		}
+
+		if (max < r->size)
+			max = r->size;
+		r = r->next;
+	}
+
+	return max;
+}
+
+static struct nouveau_resource *
+nouveau_bo_tmp(struct nouveau_channel *chan, unsigned size,
+	       struct nouveau_fence *fence)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(chan->device);
+	struct nouveau_resource *r = NULL;
+	struct nouveau_fence *ref = NULL;
+
+	if (fence)
+		nouveau_fence_ref(fence, &ref);
+	else
+		nouveau_fence_new(chan, &ref);
+	assert(ref);
+
+	while (nouveau_resource_alloc(nvdev->sa_heap, size, ref, &r)) {
+		if (nouveau_bo_tmp_max(nvdev) < size) {
+			nouveau_fence_ref(NULL, &ref);
+			return NULL;
+		}
+
+		nouveau_fence_flush(chan);
+	}
+	nouveau_fence_signal_cb(ref, nouveau_bo_tmp_del, r);
+
+	return r;
+}
+
+int
+nouveau_bo_init(struct nouveau_device *dev)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	int ret;
+
+	ret = nouveau_mem_alloc(dev, 128*1024, 0, NOUVEAU_MEM_AGP |
+				NOUVEAU_MEM_PCI, &nvdev->sa, &nvdev->sa_map);
+	if (ret)
+		return ret;
+
+	ret = nouveau_resource_init(&nvdev->sa_heap, 0, nvdev->sa.size);
+	if (ret) {
+		nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map);
+		return ret;
+	}
+
+	return 0;
+}
+
+void
+nouveau_bo_takedown(struct nouveau_device *dev)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+
+	nouveau_mem_free(dev, &nvdev->sa, &nvdev->sa_map);
+}
+
+int
+nouveau_bo_new(struct nouveau_device *dev, uint32_t flags, int align,
+	       int size, struct nouveau_bo **bo)
+{
+	struct nouveau_bo_priv *nvbo;
+	int ret;
+
+	if (!dev || !bo || *bo)
+		return -EINVAL;
+
+	nvbo = calloc(1, sizeof(struct nouveau_bo_priv));
+	if (!nvbo)
+		return -ENOMEM;
+	nvbo->base.device = dev;
+	nvbo->base.size = size;
+	nvbo->base.handle = bo_to_ptr(nvbo);
+	nvbo->drm.alignment = align;
+	nvbo->refcount = 1;
+
+	if (flags & NOUVEAU_BO_TILED) {
+		nvbo->tiled = 1;
+		if (flags & NOUVEAU_BO_ZTILE)
+			nvbo->tiled |= 2;
+		flags &= ~NOUVEAU_BO_TILED;
+	}
+
+	ret = nouveau_bo_set_status(&nvbo->base, flags);
+	if (ret) {
+		free(nvbo);
+		return ret;
+	}
+
+	*bo = &nvbo->base;
+	return 0;
+}
+
+int
+nouveau_bo_user(struct nouveau_device *dev, void *ptr, int size,
+		struct nouveau_bo **bo)
+{
+	struct nouveau_bo_priv *nvbo;
+
+	if (!dev || !bo || *bo)
+		return -EINVAL;
+
+	nvbo = calloc(1, sizeof(*nvbo));
+	if (!nvbo)
+		return -ENOMEM;
+	nvbo->base.device = dev;
+	
+	nvbo->sysmem = ptr;
+	nvbo->user = 1;
+
+	nvbo->base.size = size;
+	nvbo->base.offset = nvbo->drm.offset;
+	nvbo->base.handle = bo_to_ptr(nvbo);
+	nvbo->refcount = 1;
+	*bo = &nvbo->base;
+	return 0;
+}
+
+int
+nouveau_bo_ref(struct nouveau_device *dev, uint64_t handle,
+	       struct nouveau_bo **bo)
+{
+	struct nouveau_bo_priv *nvbo = ptr_to_bo(handle);
+
+	if (!dev || !bo || *bo)
+		return -EINVAL;
+
+	nvbo->refcount++;
+	*bo = &nvbo->base;
+	return 0;
+}
+
+static void
+nouveau_bo_del_cb(void *priv)
+{
+	struct nouveau_bo_priv *nvbo = priv;
+
+	nouveau_fence_ref(NULL, &nvbo->fence);
+	nouveau_mem_free(nvbo->base.device, &nvbo->drm, &nvbo->map);
+	if (nvbo->sysmem && !nvbo->user)
+		free(nvbo->sysmem);
+	free(nvbo);
+}
+
+void
+nouveau_bo_del(struct nouveau_bo **bo)
+{
+	struct nouveau_bo_priv *nvbo;
+
+	if (!bo || !*bo)
+		return;
+	nvbo = nouveau_bo(*bo);
+	*bo = NULL;
+
+	if (--nvbo->refcount)
+		return;
+
+	if (nvbo->pending)
+		nouveau_pushbuf_flush(nvbo->pending->channel, 0);
+
+	if (nvbo->fence)
+		nouveau_fence_signal_cb(nvbo->fence, nouveau_bo_del_cb, nvbo);
+	else
+		nouveau_bo_del_cb(nvbo);
+}
+
+int
+nouveau_bo_map(struct nouveau_bo *bo, uint32_t flags)
+{
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+
+	if (!nvbo)
+		return -EINVAL;
+
+	if (nvbo->pending &&
+	    (nvbo->pending->flags & NOUVEAU_BO_WR || flags & NOUVEAU_BO_WR)) {
+		nouveau_pushbuf_flush(nvbo->pending->channel, 0);
+	}
+
+	if (flags & NOUVEAU_BO_WR)
+		nouveau_fence_wait(&nvbo->fence);
+	else
+		nouveau_fence_wait(&nvbo->wr_fence);
+
+	if (nvbo->sysmem)
+		bo->map = nvbo->sysmem;
+	else
+		bo->map = nvbo->map;
+	return 0;
+}
+
+void
+nouveau_bo_unmap(struct nouveau_bo *bo)
+{
+	bo->map = NULL;
+}
+
+static int
+nouveau_bo_upload(struct nouveau_bo_priv *nvbo)
+{
+	if (nvbo->fence)
+		nouveau_fence_wait(&nvbo->fence);
+	memcpy(nvbo->map, nvbo->sysmem, nvbo->drm.size);
+	return 0;
+}
+
+int
+nouveau_bo_set_status(struct nouveau_bo *bo, uint32_t flags)
+{
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+	struct drm_nouveau_mem_alloc new;
+	void *new_map = NULL, *new_sysmem = NULL;
+	unsigned new_flags = 0, ret;
+
+	assert(!bo->map);
+
+	/* Check current memtype vs requested, if they match do nothing */
+	if ((nvbo->drm.flags & NOUVEAU_MEM_FB) && (flags & NOUVEAU_BO_VRAM))
+		return 0;
+	if ((nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI)) &&
+	    (flags & NOUVEAU_BO_GART))
+		return 0;
+	if (nvbo->drm.size == 0 && nvbo->sysmem && (flags & NOUVEAU_BO_LOCAL))
+		return 0;
+
+	memset(&new, 0x00, sizeof(new));
+
+	/* Allocate new memory */
+	if (flags & NOUVEAU_BO_VRAM)
+		new_flags |= NOUVEAU_MEM_FB;
+	else
+	if (flags & NOUVEAU_BO_GART)
+		new_flags |= (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI);
+	
+	if (nvbo->tiled && flags) {
+		new_flags |= NOUVEAU_MEM_TILE;
+		if (nvbo->tiled & 2)
+			new_flags |= NOUVEAU_MEM_TILE_ZETA;
+	}
+
+	if (new_flags) {
+		ret = nouveau_mem_alloc(bo->device, bo->size,
+					nvbo->drm.alignment, new_flags,
+					&new, &new_map);
+		if (ret)
+			return ret;
+	} else
+	if (!nvbo->user) {
+		new_sysmem = malloc(bo->size);
+	}
+
+	/* Copy old -> new */
+	/*XXX: use M2MF */
+	if (nvbo->sysmem || nvbo->map) {
+		struct nouveau_pushbuf_bo *pbo = nvbo->pending;
+		nvbo->pending = NULL;
+		nouveau_bo_map(bo, NOUVEAU_BO_RD);
+		memcpy(new_map, bo->map, bo->size);
+		nouveau_bo_unmap(bo);
+		nvbo->pending = pbo;
+	}
+
+	/* Free old memory */
+	if (nvbo->fence)
+		nouveau_fence_wait(&nvbo->fence);
+	nouveau_mem_free(bo->device, &nvbo->drm, &nvbo->map);
+	if (nvbo->sysmem && !nvbo->user)
+		free(nvbo->sysmem);
+
+	nvbo->drm = new;
+	nvbo->map = new_map;
+	if (!nvbo->user)
+		nvbo->sysmem = new_sysmem;
+	bo->flags = flags;
+	bo->offset = nvbo->drm.offset;
+	return 0;
+}
+
+static int
+nouveau_bo_validate_user(struct nouveau_channel *chan, struct nouveau_bo *bo,
+			 struct nouveau_fence *fence, uint32_t flags)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_device_priv *nvdev = nouveau_device(chan->device);
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+	struct nouveau_resource *r;
+
+	if (nvchan->user_charge + bo->size > nvdev->sa.size)
+		return 1;
+
+	if (!(flags & NOUVEAU_BO_GART))
+		return 1;
+
+	r = nouveau_bo_tmp(chan, bo->size, fence);
+	if (!r)
+		return 1;
+	nvchan->user_charge += bo->size;
+
+	memcpy(nvdev->sa_map + r->start, nvbo->sysmem, bo->size);
+
+	nvbo->offset = nvdev->sa.offset + r->start;
+	nvbo->flags = NOUVEAU_BO_GART;
+	return 0;
+}
+
+static int
+nouveau_bo_validate_bo(struct nouveau_channel *chan, struct nouveau_bo *bo,
+		       struct nouveau_fence *fence, uint32_t flags)
+{
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+	int ret;
+
+	ret = nouveau_bo_set_status(bo, flags);
+	if (ret) {
+		nouveau_fence_flush(chan);
+
+		ret = nouveau_bo_set_status(bo, flags);
+		if (ret)
+			return ret;
+	}
+
+	if (nvbo->user)
+		nouveau_bo_upload(nvbo);
+
+	nvbo->offset = nvbo->drm.offset;
+	if (nvbo->drm.flags & (NOUVEAU_MEM_AGP | NOUVEAU_MEM_PCI))
+		nvbo->flags = NOUVEAU_BO_GART;
+	else
+		nvbo->flags = NOUVEAU_BO_VRAM;
+
+	return 0;
+}
+
+int
+nouveau_bo_validate(struct nouveau_channel *chan, struct nouveau_bo *bo,
+		    uint32_t flags)
+{
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+	struct nouveau_fence *fence = nouveau_pushbuf(chan->pushbuf)->fence;
+	int ret;
+
+	assert(bo->map == NULL);
+
+	if (nvbo->user) {
+		ret = nouveau_bo_validate_user(chan, bo, fence, flags);
+		if (ret) {
+			ret = nouveau_bo_validate_bo(chan, bo, fence, flags);
+			if (ret)
+				return ret;
+		}
+	} else {
+		ret = nouveau_bo_validate_bo(chan, bo, fence, flags);
+		if (ret)
+			return ret;
+	}
+
+	if (flags & NOUVEAU_BO_WR)
+		nouveau_fence_ref(fence, &nvbo->wr_fence);
+	nouveau_fence_ref(fence, &nvbo->fence);
+	return 0;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_channel.c b/src/gallium/winsys/drm/nouveau/nouveau_channel.c
new file mode 100644
index 00000000000..3b4dcd1ecf2
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_channel.c
@@ -0,0 +1,126 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+
+int
+nouveau_channel_alloc(struct nouveau_device *dev, uint32_t fb_ctxdma,
+		      uint32_t tt_ctxdma, struct nouveau_channel **chan)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	struct nouveau_channel_priv *nvchan;
+	int ret;
+
+	if (!nvdev || !chan || *chan)
+	    return -EINVAL;
+
+	nvchan = calloc(1, sizeof(struct nouveau_channel_priv));
+	if (!nvchan)
+		return -ENOMEM;
+	nvchan->base.device = dev;
+
+	nvchan->drm.fb_ctxdma_handle = fb_ctxdma;
+	nvchan->drm.tt_ctxdma_handle = tt_ctxdma;
+	ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_CHANNEL_ALLOC,
+				  &nvchan->drm, sizeof(nvchan->drm));
+	if (ret) {
+		free(nvchan);
+		return ret;
+	}
+
+	nvchan->base.id = nvchan->drm.channel;
+	if (nouveau_grobj_ref(&nvchan->base, nvchan->drm.fb_ctxdma_handle,
+			      &nvchan->base.vram) ||
+	    nouveau_grobj_ref(&nvchan->base, nvchan->drm.tt_ctxdma_handle,
+		    	      &nvchan->base.gart)) {
+		nouveau_channel_free((void *)&nvchan);
+		return -EINVAL;
+	}
+
+	ret = drmMap(nvdev->fd, nvchan->drm.ctrl, nvchan->drm.ctrl_size,
+		     (void*)&nvchan->user);
+	if (ret) {
+		nouveau_channel_free((void *)&nvchan);
+		return ret;
+	}
+	nvchan->put     = &nvchan->user[0x40/4];
+	nvchan->get     = &nvchan->user[0x44/4];
+	nvchan->ref_cnt = &nvchan->user[0x48/4];
+
+	ret = drmMap(nvdev->fd, nvchan->drm.notifier, nvchan->drm.notifier_size,
+		     (drmAddressPtr)&nvchan->notifier_block);
+	if (ret) {
+		nouveau_channel_free((void *)&nvchan);
+		return ret;
+	}
+
+	ret = drmMap(nvdev->fd, nvchan->drm.cmdbuf, nvchan->drm.cmdbuf_size,
+		     (void*)&nvchan->pushbuf);
+	if (ret) {
+		nouveau_channel_free((void *)&nvchan);
+		return ret;
+	}
+
+	ret = nouveau_grobj_alloc(&nvchan->base, 0x00000000, 0x0030,
+				  &nvchan->base.nullobj);
+	if (ret) {
+		nouveau_channel_free((void *)&nvchan);
+		return ret;
+	}
+
+	nouveau_dma_channel_init(&nvchan->base);
+	nouveau_pushbuf_init(&nvchan->base);
+
+	*chan = &nvchan->base;
+	return 0;
+}
+
+void
+nouveau_channel_free(struct nouveau_channel **chan)
+{
+	struct nouveau_channel_priv *nvchan;
+	struct nouveau_device_priv *nvdev;
+	struct drm_nouveau_channel_free cf;
+
+	if (!chan || !*chan)
+		return;
+	nvchan = nouveau_channel(*chan);
+	*chan = NULL;
+	nvdev = nouveau_device(nvchan->base.device);
+	
+	FIRE_RING_CH(&nvchan->base);
+
+	nouveau_grobj_free(&nvchan->base.vram);
+	nouveau_grobj_free(&nvchan->base.gart);
+	nouveau_grobj_free(&nvchan->base.nullobj);
+
+	cf.channel = nvchan->drm.channel;
+	drmCommandWrite(nvdev->fd, DRM_NOUVEAU_CHANNEL_FREE, &cf, sizeof(cf));
+	free(nvchan);
+}
+
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_context.c b/src/gallium/winsys/drm/nouveau/nouveau_context.c
new file mode 100644
index 00000000000..74413c408f3
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_context.c
@@ -0,0 +1,346 @@
+#include "main/glheader.h"
+#include "glapi/glthread.h"
+#include <GL/internal/glcore.h>
+#include "utils.h"
+
+#include "state_tracker/st_public.h"
+#include "state_tracker/st_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_context.h"
+#include "pipe/p_screen.h"
+
+#include "nouveau_context.h"
+#include "nouveau_dri.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_winsys_pipe.h"
+
+#ifdef DEBUG
+static const struct dri_debug_control debug_control[] = {
+	{ "bo", DEBUG_BO },
+	{ NULL, 0 }
+};
+int __nouveau_debug = 0;
+#endif
+
+static void
+nouveau_channel_context_destroy(struct nouveau_channel_context *nvc)
+{
+	nouveau_grobj_free(&nvc->NvCtxSurf2D);
+	nouveau_grobj_free(&nvc->NvImageBlit);
+	nouveau_grobj_free(&nvc->NvGdiRect);
+	nouveau_grobj_free(&nvc->NvM2MF);
+	nouveau_grobj_free(&nvc->Nv2D);
+	nouveau_grobj_free(&nvc->NvSwzSurf);
+	nouveau_grobj_free(&nvc->NvSIFM);
+
+	nouveau_notifier_free(&nvc->sync_notifier);
+
+	nouveau_channel_free(&nvc->channel);
+
+	FREE(nvc);
+}
+
+static struct nouveau_channel_context *
+nouveau_channel_context_create(struct nouveau_device *dev)
+{
+	struct nouveau_channel_context *nvc;
+	int ret;
+
+	nvc = CALLOC_STRUCT(nouveau_channel_context);
+	if (!nvc)
+		return NULL;
+
+	if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002,
+					 &nvc->channel))) {
+		NOUVEAU_ERR("Error creating GPU channel: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	nvc->next_handle = 0x80000000;
+
+	if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1,
+					  &nvc->sync_notifier))) {
+		NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	switch (dev->chipset & 0xf0) {
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		ret = nouveau_surface_channel_create_nv50(nvc);
+		break;
+	default:
+		ret = nouveau_surface_channel_create_nv04(nvc);
+		break;
+	}
+
+	if (ret) {
+		NOUVEAU_ERR("Error initialising surface objects: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	return nvc;
+}
+
+GLboolean
+nouveau_context_create(const __GLcontextModes *glVis,
+		       __DRIcontextPrivate *driContextPriv,
+		       void *sharedContextPrivate)
+{
+	__DRIscreenPrivate *driScrnPriv = driContextPriv->driScreenPriv;
+	struct nouveau_screen  *nv_screen = driScrnPriv->private;
+	struct nouveau_context *nv = CALLOC_STRUCT(nouveau_context);
+	struct pipe_context *pipe = NULL;
+	struct st_context *st_share = NULL;
+	struct nouveau_channel_context *nvc = NULL;
+	struct nouveau_device *dev = nv_screen->device;
+	int i;
+
+	if (sharedContextPrivate) {
+		st_share = ((struct nouveau_context *)sharedContextPrivate)->st;
+	}
+
+	switch (dev->chipset & 0xf0) {
+	case 0x10:
+	case 0x20:
+		/* NV10 */
+	case 0x30:
+		/* NV30 */
+	case 0x40:
+	case 0x60:
+		/* NV40 */
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		/* G80 */
+		break;
+	default:
+		NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset);
+		return GL_FALSE;
+	}
+
+	driContextPriv->driverPrivate = (void *)nv;
+	nv->nv_screen  = nv_screen;
+	nv->dri_screen = driScrnPriv;
+
+	{
+		struct nouveau_device_priv *nvdev = nouveau_device(dev);
+
+		nvdev->ctx  = driContextPriv->hHWContext;
+		nvdev->lock = (drmLock *)&driScrnPriv->pSAREA->lock;
+	}
+
+	driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache,
+			    nv->dri_screen->myNum, "nouveau");
+#ifdef DEBUG
+	__nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"),
+					      debug_control);
+#endif
+
+	/*XXX: Hack up a fake region and buffer object for front buffer.
+	 *     This will go away with TTM, replaced with a simple reference
+	 *     of the front buffer handle passed to us by the DDX.
+	 */
+	{
+		struct pipe_surface *fb_surf;
+		struct nouveau_pipe_buffer *fb_buf;
+		struct nouveau_bo_priv *fb_bo;
+
+		fb_bo = calloc(1, sizeof(struct nouveau_bo_priv));
+		fb_bo->drm.offset = nv_screen->front_offset;
+		fb_bo->drm.flags = NOUVEAU_MEM_FB;
+		fb_bo->drm.size = nv_screen->front_pitch * 
+				  nv_screen->front_height;
+		fb_bo->refcount = 1;
+		fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM;
+		fb_bo->base.offset = fb_bo->drm.offset;
+		fb_bo->base.handle = (unsigned long)fb_bo;
+		fb_bo->base.size = fb_bo->drm.size;
+		fb_bo->base.device = nv_screen->device;
+
+		fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer));
+		fb_buf->bo = &fb_bo->base;
+
+		fb_surf = calloc(1, sizeof(struct pipe_surface));
+		if (nv_screen->front_cpp == 2)
+			fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM;
+		else
+			fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM;
+		pf_get_block(fb_surf->format, &fb_surf->block);
+		fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp;
+		fb_surf->height = nv_screen->front_height;
+		fb_surf->stride = fb_surf->width * fb_surf->block.size;
+		fb_surf->refcount = 1;
+		fb_surf->buffer = &fb_buf->base;
+
+		nv->frontbuffer = fb_surf;
+	}
+
+	/* Attempt to share a single channel between multiple contexts from
+	 * a single process.
+	 */
+	nvc = nv_screen->nvc;
+	if (!nvc && st_share) {
+		struct nouveau_context *snv = st_share->pipe->priv;
+		if (snv) {
+			nvc = snv->nvc;
+		}
+	}
+
+	/*XXX: temporary - disable multi-context/single-channel on pre-NV4x */
+	switch (dev->chipset & 0xf0) {
+	case 0x40:
+	case 0x60:
+		/* NV40 class */
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		/* G80 class */
+		break;
+	default:
+		nvc = NULL;
+		break;
+	}
+
+	if (!nvc) {
+		nvc = nouveau_channel_context_create(dev);
+		if (!nvc) {
+			NOUVEAU_ERR("Failed initialising GPU context\n");
+			return GL_FALSE;
+		}
+		nv_screen->nvc = nvc;
+	}
+
+	nvc->refcount++;
+	nv->nvc = nvc;
+
+	/* Find a free slot for a pipe context, allocate a new one if needed */
+	nv->pctx_id = -1;
+	for (i = 0; i < nvc->nr_pctx; i++) {
+		if (nvc->pctx[i] == NULL) {
+			nv->pctx_id = i;
+			break;
+		}
+	}
+
+	if (nv->pctx_id < 0) {
+		nv->pctx_id = nvc->nr_pctx++;
+		nvc->pctx =
+			realloc(nvc->pctx,
+				sizeof(struct pipe_context *) * nvc->nr_pctx);
+	}
+
+	/* Create pipe */
+	switch (dev->chipset & 0xf0) {
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		if (nouveau_surface_init_nv50(nv))
+			return GL_FALSE;
+		break;
+	default:
+		if (nouveau_surface_init_nv04(nv))
+			return GL_FALSE;
+		break;
+	}
+
+	if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) {
+		struct pipe_screen *pscreen;
+
+		pipe = nouveau_pipe_create(nv);
+		if (!pipe)
+			NOUVEAU_ERR("Couldn't create hw pipe\n");
+		pscreen = nvc->pscreen;
+
+		nv->cap.hw_vertex_buffer =
+			pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF);
+		nv->cap.hw_index_buffer =
+			pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF);
+	}
+
+	if (!pipe) {
+		NOUVEAU_MSG("Using softpipe\n");
+		pipe = nouveau_create_softpipe(nv);
+		if (!pipe) {
+			NOUVEAU_ERR("Error creating pipe, bailing\n");
+			return GL_FALSE;
+		}
+	}
+
+	pipe->priv = nv;
+	nv->st = st_create_context(pipe, glVis, st_share);
+	return GL_TRUE;
+}
+
+void
+nouveau_context_destroy(__DRIcontextPrivate *driContextPriv)
+{
+	struct nouveau_context *nv = driContextPriv->driverPrivate;
+	struct nouveau_channel_context *nvc = nv->nvc;
+
+	assert(nv);
+
+	st_finish(nv->st);
+	st_destroy_context(nv->st);
+
+	if (nv->pctx_id >= 0) {
+		nvc->pctx[nv->pctx_id] = NULL;
+		if (--nvc->refcount <= 0) {
+			nouveau_channel_context_destroy(nvc);
+			nv->nv_screen->nvc = NULL;
+		}
+	}
+
+	free(nv);
+}
+
+GLboolean
+nouveau_context_bind(__DRIcontextPrivate *driContextPriv,
+		     __DRIdrawablePrivate *driDrawPriv,
+		     __DRIdrawablePrivate *driReadPriv)
+{
+	struct nouveau_context *nv;
+	struct nouveau_framebuffer *draw, *read;
+
+	if (!driContextPriv) {
+		st_make_current(NULL, NULL, NULL);
+		return GL_TRUE;
+	}
+
+	nv = driContextPriv->driverPrivate;
+	draw = driDrawPriv->driverPrivate;
+	read = driReadPriv->driverPrivate;
+
+	st_make_current(nv->st, draw->stfb, read->stfb);
+
+	if ((nv->dri_drawable != driDrawPriv) ||
+	    (nv->last_stamp != driDrawPriv->lastStamp)) {
+		nv->dri_drawable = driDrawPriv;
+		st_resize_framebuffer(draw->stfb, driDrawPriv->w,
+				      driDrawPriv->h);
+		nv->last_stamp = driDrawPriv->lastStamp;
+	}
+
+	if (driDrawPriv != driReadPriv) {
+		st_resize_framebuffer(read->stfb, driReadPriv->w,
+				      driReadPriv->h);
+	}
+
+	return GL_TRUE;
+}
+
+GLboolean
+nouveau_context_unbind(__DRIcontextPrivate *driContextPriv)
+{
+	struct nouveau_context *nv = driContextPriv->driverPrivate;
+	(void)nv;
+
+	st_flush(nv->st, 0, NULL);
+	return GL_TRUE;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_context.h b/src/gallium/winsys/drm/nouveau/nouveau_context.h
new file mode 100644
index 00000000000..77e2147a2c7
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_context.h
@@ -0,0 +1,113 @@
+#ifndef __NOUVEAU_CONTEXT_H__
+#define __NOUVEAU_CONTEXT_H__
+
+#include "dri_util.h"
+#include "xmlconfig.h"
+
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+
+struct nouveau_framebuffer {
+	struct st_framebuffer *stfb;
+};
+
+struct nouveau_channel_context {
+	struct pipe_screen *pscreen;
+	int refcount;
+
+	unsigned cur_pctx;
+	unsigned nr_pctx;
+	struct pipe_context **pctx;
+
+	struct nouveau_channel  *channel;
+
+	struct nouveau_notifier *sync_notifier;
+
+	/* Common */
+	struct nouveau_grobj    *NvM2MF;
+	/* NV04-NV40 */
+	struct nouveau_grobj    *NvCtxSurf2D;
+	struct nouveau_grobj	*NvSwzSurf;
+	struct nouveau_grobj    *NvImageBlit;
+	struct nouveau_grobj    *NvGdiRect;
+	struct nouveau_grobj	*NvSIFM;
+	/* G80 */
+	struct nouveau_grobj    *Nv2D;
+
+	uint32_t                 next_handle;
+	uint32_t                 next_subchannel;
+	uint32_t                 next_sequence;
+};
+
+struct nouveau_context {
+	struct st_context *st;
+
+	/* DRI stuff */
+	__DRIscreenPrivate    *dri_screen;
+	__DRIdrawablePrivate  *dri_drawable;
+	unsigned int           last_stamp;
+	driOptionCache         dri_option_cache;
+	drm_context_t          drm_context;
+	drmLock                drm_lock;
+	GLboolean              locked;
+	struct nouveau_screen *nv_screen;
+	struct pipe_surface *frontbuffer;
+
+	struct {
+		int hw_vertex_buffer;
+		int hw_index_buffer;
+	} cap;
+
+	/* Hardware context */
+	struct nouveau_channel_context *nvc;
+	int pctx_id;
+
+	/* pipe_surface accel */
+	struct pipe_surface *surf_src, *surf_dst;
+	unsigned surf_src_offset, surf_dst_offset;
+	int  (*surface_copy_prep)(struct nouveau_context *,
+				  struct pipe_surface *dst,
+				  struct pipe_surface *src);
+	void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy,
+			     unsigned sx, unsigned sy, unsigned w, unsigned h);
+	void (*surface_copy_done)(struct nouveau_context *);
+	int (*surface_fill)(struct nouveau_context *, struct pipe_surface *,
+			    unsigned, unsigned, unsigned, unsigned, unsigned);
+};
+
+extern GLboolean nouveau_context_create(const __GLcontextModes *,
+					__DRIcontextPrivate *, void *);
+extern void nouveau_context_destroy(__DRIcontextPrivate *);
+extern GLboolean nouveau_context_bind(__DRIcontextPrivate *,
+				      __DRIdrawablePrivate *draw,
+				      __DRIdrawablePrivate *read);
+extern GLboolean nouveau_context_unbind(__DRIcontextPrivate *);
+
+#ifdef DEBUG
+extern int __nouveau_debug;
+
+#define DEBUG_BO (1 << 0)
+
+#define DBG(flag, ...) do {                   \
+	if (__nouveau_debug & (DEBUG_##flag)) \
+		NOUVEAU_ERR(__VA_ARGS__);     \
+} while(0)
+#else
+#define DBG(flag, ...)
+#endif
+
+extern void LOCK_HARDWARE(struct nouveau_context *);
+extern void UNLOCK_HARDWARE(struct nouveau_context *);
+
+extern int
+nouveau_surface_channel_create_nv04(struct nouveau_channel_context *);
+extern int
+nouveau_surface_channel_create_nv50(struct nouveau_channel_context *);
+extern int nouveau_surface_init_nv04(struct nouveau_context *);
+extern int nouveau_surface_init_nv50(struct nouveau_context *);
+
+extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int);
+extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *);
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_device.c b/src/gallium/winsys/drm/nouveau/nouveau_device.c
new file mode 100644
index 00000000000..0b452fcd02d
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_device.c
@@ -0,0 +1,159 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+
+int
+nouveau_device_open_existing(struct nouveau_device **dev, int close,
+			     int fd, drm_context_t ctx)
+{
+	struct nouveau_device_priv *nvdev;
+	int ret;
+
+	if (!dev || *dev)
+	    return -EINVAL;
+
+	nvdev = calloc(1, sizeof(*nvdev));
+	if (!nvdev)
+	    return -ENOMEM;
+	nvdev->fd = fd;
+	nvdev->ctx = ctx;
+	nvdev->needs_close = close;
+
+	drmCommandNone(nvdev->fd, DRM_NOUVEAU_CARD_INIT);
+
+	if ((ret = nouveau_bo_init(&nvdev->base))) {
+		nouveau_device_close((void *)&nvdev);
+		return ret;
+	}
+
+	{
+		uint64_t value;
+
+		ret = nouveau_device_get_param(&nvdev->base,
+					       NOUVEAU_GETPARAM_CHIPSET_ID,
+					       &value);
+		if (ret) {
+			nouveau_device_close((void *)&nvdev);
+			return ret;
+		}
+		nvdev->base.chipset = value;
+	}
+
+	*dev = &nvdev->base;
+	return 0;
+}
+
+int
+nouveau_device_open(struct nouveau_device **dev, const char *busid)
+{
+	drm_context_t ctx;
+	int fd, ret;
+
+	if (!dev || *dev)
+		return -EINVAL;
+
+	fd = drmOpen("nouveau", busid);
+	if (fd < 0)
+		return -EINVAL;
+
+	ret = drmCreateContext(fd, &ctx);
+	if (ret) {
+		drmClose(fd);
+		return ret;
+	}
+
+	ret = nouveau_device_open_existing(dev, 1, fd, ctx);
+	if (ret) {
+	    drmDestroyContext(fd, ctx);
+	    drmClose(fd);
+	    return ret;
+	}
+
+	return 0;
+}
+
+void
+nouveau_device_close(struct nouveau_device **dev)
+{
+	struct nouveau_device_priv *nvdev;
+
+	if (dev || !*dev)
+		return;
+	nvdev = nouveau_device(*dev);
+	*dev = NULL;
+
+	nouveau_bo_takedown(&nvdev->base);
+
+	if (nvdev->needs_close) {
+		drmDestroyContext(nvdev->fd, nvdev->ctx);
+		drmClose(nvdev->fd);
+	}
+	free(nvdev);
+}
+
+int
+nouveau_device_get_param(struct nouveau_device *dev,
+			 uint64_t param, uint64_t *value)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	struct drm_nouveau_getparam g;
+	int ret;
+
+	if (!nvdev || !value)
+		return -EINVAL;
+
+	g.param = param;
+	ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_GETPARAM,
+				  &g, sizeof(g));
+	if (ret)
+		return ret;
+
+	*value = g.value;
+	return 0;
+}
+
+int
+nouveau_device_set_param(struct nouveau_device *dev,
+			 uint64_t param, uint64_t value)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	struct drm_nouveau_setparam s;
+	int ret;
+
+	if (!nvdev)
+		return -EINVAL;
+
+	s.param = param;
+	s.value = value;
+	ret = drmCommandWriteRead(nvdev->fd, DRM_NOUVEAU_SETPARAM,
+				  &s, sizeof(s));
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_dma.c b/src/gallium/winsys/drm/nouveau/nouveau_dma.c
new file mode 100644
index 00000000000..f8a8ba04f6d
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_dma.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdint.h>
+#include <assert.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+#include "nouveau_local.h"
+
+static inline uint32_t
+READ_GET(struct nouveau_channel_priv *nvchan)
+{
+	return *nvchan->get;
+}
+
+static inline void
+WRITE_PUT(struct nouveau_channel_priv *nvchan, uint32_t val)
+{
+	uint32_t put = ((val << 2) + nvchan->dma->base);
+	volatile int dum;
+
+	NOUVEAU_DMA_BARRIER;
+	dum = READ_GET(nvchan);
+
+	*nvchan->put = put;
+	nvchan->dma->put = val;
+#ifdef NOUVEAU_DMA_TRACE
+	NOUVEAU_MSG("WRITE_PUT %d/0x%08x\n", nvchan->drm.channel, put);
+#endif
+
+	NOUVEAU_DMA_BARRIER;
+}
+
+static inline int
+LOCAL_GET(struct nouveau_dma_priv *dma, uint32_t *val)
+{
+	uint32_t get = *val;
+
+	if (get >= dma->base && get <= (dma->base + (dma->max << 2))) {
+		*val = (get - dma->base) >> 2;
+		return 1;
+	}
+
+	return 0;
+}
+
+void
+nouveau_dma_channel_init(struct nouveau_channel *chan)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	int i;
+
+	nvchan->dma = &nvchan->dma_master;
+	nvchan->dma->base = nvchan->drm.put_base;
+	nvchan->dma->cur  = nvchan->dma->put = 0;
+	nvchan->dma->max  = (nvchan->drm.cmdbuf_size >> 2) - 2;
+	nvchan->dma->free = nvchan->dma->max - nvchan->dma->cur;
+
+	RING_SPACE_CH(chan, RING_SKIPS);
+	for (i = 0; i < RING_SKIPS; i++)
+		OUT_RING_CH(chan, 0);
+}
+
+#define CHECK_TIMEOUT() do {                                                   \
+	if ((NOUVEAU_TIME_MSEC() - t_start) > NOUVEAU_DMA_TIMEOUT)             \
+		return - EBUSY;                                                \
+} while(0)
+
+int
+nouveau_dma_wait(struct nouveau_channel *chan, int size)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+	uint32_t get, t_start;
+
+	FIRE_RING_CH(chan);
+
+	t_start = NOUVEAU_TIME_MSEC();
+	while (dma->free < size) {
+		CHECK_TIMEOUT();
+
+		get = READ_GET(nvchan);
+		if (!LOCAL_GET(dma, &get))
+			continue;
+
+		if (dma->put >= get) {
+			dma->free = dma->max - dma->cur;
+
+			if (dma->free < size) {
+#ifdef NOUVEAU_DMA_DEBUG
+				dma->push_free = 1;
+#endif
+				OUT_RING_CH(chan, 0x20000000 | dma->base);
+				if (get <= RING_SKIPS) {
+					/*corner case - will be idle*/
+					if (dma->put <= RING_SKIPS)
+						WRITE_PUT(nvchan,
+							  RING_SKIPS + 1);
+
+					do {
+						CHECK_TIMEOUT();
+						get = READ_GET(nvchan);
+						if (!LOCAL_GET(dma, &get))
+							get = 0;
+					} while (get <= RING_SKIPS);
+				}
+
+				WRITE_PUT(nvchan, RING_SKIPS);
+				dma->cur  = dma->put = RING_SKIPS;
+				dma->free = get - (RING_SKIPS + 1);
+			}
+		} else {
+			dma->free = get - dma->cur - 1;
+		}
+	}
+
+	return 0;
+}
+
+#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF
+static void
+nouveau_dma_parse_pushbuf(struct nouveau_channel *chan, int get, int put)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	unsigned mthd_count = 0;
+	
+	while (get != put) {
+		uint32_t gpuget = (get << 2) + nvchan->drm.put_base;
+		uint32_t data;
+
+		if (get < 0 || get >= nvchan->drm.cmdbuf_size) {
+			NOUVEAU_ERR("DMA_PT 0x%08x\n", gpuget);
+			assert(0);
+		}
+		data = nvchan->pushbuf[get++];
+
+		if (mthd_count) {
+			NOUVEAU_MSG("0x%08x 0x%08x\n", gpuget, data);
+			mthd_count--;
+			continue;
+		}
+
+		switch (data & 0x60000000) {
+		case 0x00000000:
+			mthd_count = (data >> 18) & 0x7ff;
+			NOUVEAU_MSG("0x%08x 0x%08x MTHD "
+				    "Sc %d Mthd 0x%04x Size %d\n",
+				    gpuget, data, (data>>13) & 7, data & 0x1ffc,
+				    mthd_count);
+			break;
+		case 0x20000000:
+			get = (data & 0x1ffffffc) >> 2;
+			NOUVEAU_MSG("0x%08x 0x%08x JUMP 0x%08x\n",
+				    gpuget, data, data & 0x1ffffffc);
+			continue;
+		case 0x40000000:
+			mthd_count = (data >> 18) & 0x7ff;
+			NOUVEAU_MSG("0x%08x 0x%08x NINC "
+				    "Sc %d Mthd 0x%04x Size %d\n",
+				    gpuget, data, (data>>13) & 7, data & 0x1ffc,
+				    mthd_count);
+			break;
+		case 0x60000000:
+			/* DMA_OPCODE_CALL apparently, doesn't seem to work on
+			 * my NV40 at least..
+			 */
+			/* fall-through */
+		default:
+			NOUVEAU_MSG("DMA_PUSHER 0x%08x 0x%08x\n",
+				    gpuget, data);
+			assert(0);
+		}
+	}
+}
+#endif
+
+void
+nouveau_dma_kickoff(struct nouveau_channel *chan)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+
+	if (dma->cur == dma->put)
+		return;
+
+#ifdef NOUVEAU_DMA_DEBUG
+	if (dma->push_free) {
+		NOUVEAU_ERR("Packet incomplete: %d left\n", dma->push_free);
+		return;
+	}
+#endif
+
+#ifdef NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF
+	nouveau_dma_parse_pushbuf(chan, dma->put, dma->cur);
+#endif
+
+	WRITE_PUT(nvchan, dma->cur);
+}
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_dma.h b/src/gallium/winsys/drm/nouveau/nouveau_dma.h
new file mode 100644
index 00000000000..cfa6d26e828
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_dma.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_DMA_H__
+#define __NOUVEAU_DMA_H__
+
+#include <string.h>
+#include "nouveau_drmif.h"
+#include "nouveau_local.h"
+
+#define RING_SKIPS 8
+
+extern int  nouveau_dma_wait(struct nouveau_channel *chan, int size);
+extern void nouveau_dma_subc_bind(struct nouveau_grobj *);
+extern void nouveau_dma_channel_init(struct nouveau_channel *);
+extern void nouveau_dma_kickoff(struct nouveau_channel *);
+
+#ifdef NOUVEAU_DMA_DEBUG
+static char faulty[1024];
+#endif
+
+static inline void
+nouveau_dma_out(struct nouveau_channel *chan, uint32_t data)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+
+#ifdef NOUVEAU_DMA_DEBUG
+	if (dma->push_free == 0) {
+		NOUVEAU_ERR("No space left in packet at %s\n", faulty);
+		return;
+	}
+	dma->push_free--;
+#endif
+#ifdef NOUVEAU_DMA_TRACE
+	{
+		uint32_t offset = (dma->cur << 2) + dma->base;
+		NOUVEAU_MSG("\tOUT_RING %d/0x%08x -> 0x%08x\n",
+			    nvchan->drm.channel, offset, data);
+	}
+#endif
+	nvchan->pushbuf[dma->cur + (dma->base - nvchan->drm.put_base)/4] = data;
+	dma->cur++;
+}
+
+static inline void
+nouveau_dma_outp(struct nouveau_channel *chan, uint32_t *ptr, int size)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+	(void)dma;
+
+#ifdef NOUVEAU_DMA_DEBUG
+	if (dma->push_free < size) {
+		NOUVEAU_ERR("Packet too small.  Free=%d, Need=%d\n",
+			    dma->push_free, size);
+		return;
+	}
+#endif
+#ifdef NOUVEAU_DMA_TRACE
+	while (size--) {
+		nouveau_dma_out(chan, *ptr);
+		ptr++;
+	}
+#else
+	memcpy(&nvchan->pushbuf[dma->cur], ptr, size << 2);
+#ifdef NOUVEAU_DMA_DEBUG
+	dma->push_free -= size;
+#endif
+	dma->cur += size;
+#endif
+}
+
+static inline void
+nouveau_dma_space(struct nouveau_channel *chan, int size)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+
+	if (dma->free < size) {
+		if (nouveau_dma_wait(chan, size) && chan->hang_notify)
+			chan->hang_notify(chan);
+	}
+	dma->free -= size;
+#ifdef NOUVEAU_DMA_DEBUG
+	dma->push_free = size;
+#endif
+}
+
+static inline void
+nouveau_dma_begin(struct nouveau_channel *chan, struct nouveau_grobj *grobj,
+		  int method, int size, const char* file, int line)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *dma = nvchan->dma;
+	(void)dma;
+
+#ifdef NOUVEAU_DMA_TRACE
+	NOUVEAU_MSG("BEGIN_RING %d/%08x/%d/0x%04x/%d\n", nvchan->drm.channel,
+		    grobj->handle, grobj->subc, method, size);
+#endif
+
+#ifdef NOUVEAU_DMA_DEBUG
+	if (dma->push_free) {
+		NOUVEAU_ERR("Previous packet incomplete: %d left at %s\n",
+			    dma->push_free, faulty);
+		return;
+	}
+	sprintf(faulty,"%s:%d",file,line);
+#endif
+
+	nouveau_dma_space(chan, (size + 1));
+	nouveau_dma_out(chan, (size << 18) | (grobj->subc << 13) | method);
+}
+
+#define RING_SPACE_CH(ch,sz)         nouveau_dma_space((ch), (sz))
+#define BEGIN_RING_CH(ch,gr,m,sz)    nouveau_dma_begin((ch), (gr), (m), (sz), __FUNCTION__, __LINE__ )
+#define OUT_RING_CH(ch, data)        nouveau_dma_out((ch), (data))
+#define OUT_RINGp_CH(ch,ptr,dwords)  nouveau_dma_outp((ch), (void*)(ptr),      \
+						      (dwords))
+#define FIRE_RING_CH(ch)             nouveau_dma_kickoff((ch))
+#define WAIT_RING_CH(ch,sz)          nouveau_dma_wait((ch), (sz))
+		
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_dri.h b/src/gallium/winsys/drm/nouveau/nouveau_dri.h
index ce3c3fb9cc4..1207c2d609c 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_dri.h
+++ b/src/gallium/winsys/drm/nouveau/nouveau_dri.h
@@ -5,7 +5,7 @@
 #include "drm.h"
 #include "nouveau_drm.h"
 
-typedef struct {
+struct nouveau_dri {
 	uint32_t device_id;	/**< \brief PCI device ID */
 	uint32_t width;		/**< \brief width in pixels of display */
 	uint32_t height;	/**< \brief height in scanlines of display */
@@ -22,7 +22,7 @@ typedef struct {
 	uint32_t depth_offset;	/**< \brief private depth buffer offset */
 	uint32_t depth_pitch;	/**< \brief private depth buffer pitch */
 
-} NOUVEAUDRIRec, *NOUVEAUDRIPtr;
+};
 
 #endif
 
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_drmif.h b/src/gallium/winsys/drm/nouveau/nouveau_drmif.h
new file mode 100644
index 00000000000..dcd6a5eb0a4
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_drmif.h
@@ -0,0 +1,310 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __NOUVEAU_DRMIF_H__
+#define __NOUVEAU_DRMIF_H__
+
+#include <stdint.h>
+#include <xf86drm.h>
+#include <nouveau_drm.h>
+
+#include "nouveau/nouveau_device.h"
+#include "nouveau/nouveau_channel.h"
+#include "nouveau/nouveau_grobj.h"
+#include "nouveau/nouveau_notifier.h"
+#include "nouveau/nouveau_bo.h"
+#include "nouveau/nouveau_resource.h"
+#include "nouveau/nouveau_pushbuf.h"
+
+struct nouveau_device_priv {
+	struct nouveau_device base;
+
+	int fd;
+	drm_context_t ctx;
+	drmLock *lock;
+	int needs_close;
+
+	struct drm_nouveau_mem_alloc sa;
+	void *sa_map;
+	struct nouveau_resource *sa_heap;
+};
+#define nouveau_device(n) ((struct nouveau_device_priv *)(n))
+
+extern int
+nouveau_device_open_existing(struct nouveau_device **, int close,
+			     int fd, drm_context_t ctx);
+
+extern int
+nouveau_device_open(struct nouveau_device **, const char *busid);
+
+extern void
+nouveau_device_close(struct nouveau_device **);
+
+extern int
+nouveau_device_get_param(struct nouveau_device *, uint64_t param, uint64_t *v);
+
+extern int
+nouveau_device_set_param(struct nouveau_device *, uint64_t param, uint64_t val);
+
+struct nouveau_fence {
+	struct nouveau_channel *channel;
+};
+
+struct nouveau_fence_cb {
+	struct nouveau_fence_cb *next;
+	void (*func)(void *);
+	void *priv;
+};
+
+struct nouveau_fence_priv {
+	struct nouveau_fence base;
+	int refcount;
+
+	struct nouveau_fence *next;
+	struct nouveau_fence_cb *signal_cb;
+
+	uint32_t sequence;
+	int emitted;
+	int signalled;
+};
+#define nouveau_fence(n) ((struct nouveau_fence_priv *)(n))
+
+extern int
+nouveau_fence_new(struct nouveau_channel *, struct nouveau_fence **);
+
+extern int
+nouveau_fence_ref(struct nouveau_fence *, struct nouveau_fence **);
+
+extern int
+nouveau_fence_signal_cb(struct nouveau_fence *, void (*)(void *), void *);
+
+extern void
+nouveau_fence_emit(struct nouveau_fence *);
+
+extern int
+nouveau_fence_wait(struct nouveau_fence **);
+
+extern void
+nouveau_fence_flush(struct nouveau_channel *);
+
+struct nouveau_pushbuf_reloc {
+	struct nouveau_pushbuf_bo *pbbo;
+	uint32_t *ptr;
+	uint32_t flags;
+	uint32_t data;
+	uint32_t vor;
+	uint32_t tor;
+};
+
+struct nouveau_pushbuf_bo {
+	struct nouveau_channel *channel;
+	struct nouveau_bo *bo;
+	unsigned flags;
+	unsigned handled;
+};
+
+#define NOUVEAU_PUSHBUF_MAX_BUFFERS 1024
+#define NOUVEAU_PUSHBUF_MAX_RELOCS 1024
+struct nouveau_pushbuf_priv {
+	struct nouveau_pushbuf base;
+
+	struct nouveau_fence *fence;
+
+	unsigned nop_jump;
+	unsigned start;
+	unsigned size;
+
+	struct nouveau_pushbuf_bo *buffers;
+	unsigned nr_buffers;
+	struct nouveau_pushbuf_reloc *relocs;
+	unsigned nr_relocs;
+};
+#define nouveau_pushbuf(n) ((struct nouveau_pushbuf_priv *)(n))
+
+#define pbbo_to_ptr(o) ((uint64_t)(unsigned long)(o))
+#define ptr_to_pbbo(h) ((struct nouveau_pushbuf_bo *)(unsigned long)(h))
+#define pbrel_to_ptr(o) ((uint64_t)(unsigned long)(o))
+#define ptr_to_pbrel(h) ((struct nouveau_pushbuf_reloc *)(unsigned long)(h))
+#define bo_to_ptr(o) ((uint64_t)(unsigned long)(o))
+#define ptr_to_bo(h) ((struct nouveau_bo_priv *)(unsigned long)(h))
+
+extern int
+nouveau_pushbuf_init(struct nouveau_channel *);
+
+extern int
+nouveau_pushbuf_flush(struct nouveau_channel *, unsigned min);
+
+extern int
+nouveau_pushbuf_emit_reloc(struct nouveau_channel *, void *ptr,
+			   struct nouveau_bo *, uint32_t data, uint32_t flags,
+			   uint32_t vor, uint32_t tor);
+
+struct nouveau_dma_priv {
+	uint32_t base;
+	uint32_t max;
+	uint32_t cur;
+	uint32_t put;
+	uint32_t free;
+
+	int push_free;
+} dma;
+
+struct nouveau_channel_priv {
+	struct nouveau_channel base;
+
+	struct drm_nouveau_channel_alloc drm;
+
+	uint32_t *pushbuf;
+	void     *notifier_block;
+
+	volatile uint32_t *user;
+	volatile uint32_t *put;
+	volatile uint32_t *get;
+	volatile uint32_t *ref_cnt;
+
+	struct nouveau_dma_priv dma_master;
+	struct nouveau_dma_priv dma_bufmgr;
+	struct nouveau_dma_priv *dma;
+
+	struct nouveau_fence *fence_head;
+	struct nouveau_fence *fence_tail;
+	uint32_t fence_sequence;
+
+	struct nouveau_pushbuf_priv pb;
+
+	unsigned user_charge;
+};
+#define nouveau_channel(n) ((struct nouveau_channel_priv *)(n))
+
+extern int
+nouveau_channel_alloc(struct nouveau_device *, uint32_t fb, uint32_t tt,
+		      struct nouveau_channel **);
+
+extern void
+nouveau_channel_free(struct nouveau_channel **);
+
+struct nouveau_grobj_priv {
+	struct nouveau_grobj base;
+};
+#define nouveau_grobj(n) ((struct nouveau_grobj_priv *)(n))
+
+extern int nouveau_grobj_alloc(struct nouveau_channel *, uint32_t handle,
+			       int class, struct nouveau_grobj **);
+extern int nouveau_grobj_ref(struct nouveau_channel *, uint32_t handle,
+			     struct nouveau_grobj **);
+extern void nouveau_grobj_free(struct nouveau_grobj **);
+
+
+struct nouveau_notifier_priv {
+	struct nouveau_notifier base;
+
+	struct drm_nouveau_notifierobj_alloc drm;
+	volatile void *map;
+};
+#define nouveau_notifier(n) ((struct nouveau_notifier_priv *)(n))
+
+extern int
+nouveau_notifier_alloc(struct nouveau_channel *, uint32_t handle, int count,
+		       struct nouveau_notifier **);
+
+extern void
+nouveau_notifier_free(struct nouveau_notifier **);
+
+extern void
+nouveau_notifier_reset(struct nouveau_notifier *, int id);
+
+extern uint32_t
+nouveau_notifier_status(struct nouveau_notifier *, int id);
+
+extern uint32_t
+nouveau_notifier_return_val(struct nouveau_notifier *, int id);
+
+extern int
+nouveau_notifier_wait_status(struct nouveau_notifier *, int id, int status,
+			     int timeout);
+
+struct nouveau_bo_priv {
+	struct nouveau_bo base;
+
+	struct nouveau_pushbuf_bo *pending;
+	struct nouveau_fence *fence;
+	struct nouveau_fence *wr_fence;
+
+	struct drm_nouveau_mem_alloc drm;
+	void *map;
+
+	void *sysmem;
+	int user;
+
+	int refcount;
+
+	uint64_t offset;
+	uint64_t flags;
+	int tiled;
+};
+#define nouveau_bo(n) ((struct nouveau_bo_priv *)(n))
+
+extern int
+nouveau_bo_init(struct nouveau_device *);
+
+extern void
+nouveau_bo_takedown(struct nouveau_device *);
+
+extern int
+nouveau_bo_new(struct nouveau_device *, uint32_t flags, int align, int size,
+	       struct nouveau_bo **);
+
+extern int
+nouveau_bo_user(struct nouveau_device *, void *ptr, int size,
+		struct nouveau_bo **);
+
+extern int
+nouveau_bo_ref(struct nouveau_device *, uint64_t handle, struct nouveau_bo **);
+
+extern int
+nouveau_bo_set_status(struct nouveau_bo *, uint32_t flags);
+
+extern void
+nouveau_bo_del(struct nouveau_bo **);
+
+extern int
+nouveau_bo_map(struct nouveau_bo *, uint32_t flags);
+
+extern void
+nouveau_bo_unmap(struct nouveau_bo *);
+
+extern int
+nouveau_bo_validate(struct nouveau_channel *, struct nouveau_bo *,
+		    uint32_t flags);
+
+extern int
+nouveau_resource_init(struct nouveau_resource **heap, unsigned start,
+		      unsigned size);
+
+extern int
+nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv,
+		       struct nouveau_resource **);
+
+extern void
+nouveau_resource_free(struct nouveau_resource **);
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_fence.c b/src/gallium/winsys/drm/nouveau/nouveau_fence.c
new file mode 100644
index 00000000000..e7b0b4ff079
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_fence.c
@@ -0,0 +1,214 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <errno.h>
+#include <assert.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+#include "nouveau_local.h"
+
+static void
+nouveau_fence_del_unsignalled(struct nouveau_fence *fence)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel);
+	struct nouveau_fence *le;
+
+	if (nvchan->fence_head == fence) {
+		nvchan->fence_head = nouveau_fence(fence)->next;
+		if (nvchan->fence_head == NULL)
+			nvchan->fence_tail = NULL;
+		return;
+	}
+
+	le = nvchan->fence_head;
+	while (le && nouveau_fence(le)->next != fence)
+		le = nouveau_fence(le)->next;
+	assert(le && nouveau_fence(le)->next == fence);
+	nouveau_fence(le)->next = nouveau_fence(fence)->next;
+	if (nvchan->fence_tail == fence)
+		nvchan->fence_tail = le;
+}
+
+static void
+nouveau_fence_del(struct nouveau_fence **fence)
+{
+	struct nouveau_fence_priv *nvfence;
+
+	if (!fence || !*fence)
+		return;
+	nvfence = nouveau_fence(*fence);
+	*fence = NULL;
+
+	if (--nvfence->refcount)
+		return;
+
+	if (nvfence->emitted && !nvfence->signalled) {
+		if (nvfence->signal_cb) {
+			nvfence->refcount++;
+			nouveau_fence_wait((void *)&nvfence);
+			return;
+		}
+
+		nouveau_fence_del_unsignalled(&nvfence->base);
+	}
+	free(nvfence);
+}
+
+int
+nouveau_fence_new(struct nouveau_channel *chan, struct nouveau_fence **fence)
+{
+	struct nouveau_fence_priv *nvfence;
+
+	if (!chan || !fence || *fence)
+		return -EINVAL;
+	
+	nvfence = calloc(1, sizeof(struct nouveau_fence_priv));
+	if (!nvfence)
+		return -ENOMEM;
+	nvfence->base.channel = chan;
+	nvfence->refcount = 1;
+
+	*fence = &nvfence->base;
+	return 0;
+}
+
+int
+nouveau_fence_ref(struct nouveau_fence *ref, struct nouveau_fence **fence)
+{
+	struct nouveau_fence_priv *nvfence;
+
+	if (!fence)
+		return -EINVAL;
+
+	if (*fence) {
+		nouveau_fence_del(fence);
+		*fence = NULL;
+	}
+
+	if (ref) {
+		nvfence = nouveau_fence(ref);
+		nvfence->refcount++;	
+		*fence = &nvfence->base;
+	}
+
+	return 0;
+}
+
+int
+nouveau_fence_signal_cb(struct nouveau_fence *fence, void (*func)(void *),
+			void *priv)
+{
+	struct nouveau_fence_priv *nvfence = nouveau_fence(fence);
+	struct nouveau_fence_cb *cb;
+
+	if (!nvfence || !func)
+		return -EINVAL;
+
+	cb = malloc(sizeof(struct nouveau_fence_cb));
+	if (!cb)
+		return -ENOMEM;
+
+	cb->func = func;
+	cb->priv = priv;
+	cb->next = nvfence->signal_cb;
+	nvfence->signal_cb = cb;
+	return 0;
+}
+
+void
+nouveau_fence_emit(struct nouveau_fence *fence)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(fence->channel);
+	struct nouveau_fence_priv *nvfence = nouveau_fence(fence);
+
+	nvfence->emitted = 1;
+	nvfence->sequence = ++nvchan->fence_sequence;
+	if (nvfence->sequence == 0xffffffff)
+		NOUVEAU_ERR("AII wrap unhandled\n");
+
+	/*XXX: assumes subc 0 is populated */
+	RING_SPACE_CH(fence->channel, 2);
+	OUT_RING_CH  (fence->channel, 0x00040050);
+	OUT_RING_CH  (fence->channel, nvfence->sequence);
+
+	if (nvchan->fence_tail) {
+		nouveau_fence(nvchan->fence_tail)->next = fence;
+	} else {
+		nvchan->fence_head = fence;
+	}
+	nvchan->fence_tail = fence;
+}
+
+void
+nouveau_fence_flush(struct nouveau_channel *chan)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	uint32_t sequence = *nvchan->ref_cnt;
+
+	while (nvchan->fence_head) {
+		struct nouveau_fence_priv *nvfence;
+	
+		nvfence = nouveau_fence(nvchan->fence_head);
+		if (nvfence->sequence > sequence)
+			break;
+		nouveau_fence_del_unsignalled(&nvfence->base);
+		nvfence->signalled = 1;
+
+		if (nvfence->signal_cb) {
+			struct nouveau_fence *fence = NULL;
+
+			nouveau_fence_ref(&nvfence->base, &fence);
+
+			while (nvfence->signal_cb) {
+				struct nouveau_fence_cb *cb;
+				
+				cb = nvfence->signal_cb;
+				nvfence->signal_cb = cb->next;
+				cb->func(cb->priv);
+				free(cb);
+			}
+
+			nouveau_fence_ref(NULL, &fence);
+		}
+	}
+}
+
+int
+nouveau_fence_wait(struct nouveau_fence **fence)
+{
+	struct nouveau_fence_priv *nvfence;
+	
+	if (!fence || !*fence)
+		return -EINVAL;
+	nvfence = nouveau_fence(*fence);
+
+	if (nvfence->emitted) {
+		while (!nvfence->signalled)
+			nouveau_fence_flush(nvfence->base.channel);
+	}
+	nouveau_fence_ref(NULL, fence);
+
+	return 0;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_grobj.c b/src/gallium/winsys/drm/nouveau/nouveau_grobj.c
new file mode 100644
index 00000000000..51523897d58
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_grobj.c
@@ -0,0 +1,107 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+
+int
+nouveau_grobj_alloc(struct nouveau_channel *chan, uint32_t handle,
+		    int class, struct nouveau_grobj **grobj)
+{
+	struct nouveau_device_priv *nvdev = nouveau_device(chan->device);
+	struct nouveau_grobj_priv *nvgrobj;
+	struct drm_nouveau_grobj_alloc g;
+	int ret;
+
+	if (!nvdev || !grobj || *grobj)
+		return -EINVAL;
+
+	nvgrobj = calloc(1, sizeof(*nvgrobj));
+	if (!nvgrobj)
+		return -ENOMEM;
+	nvgrobj->base.channel = chan;
+	nvgrobj->base.handle  = handle;
+	nvgrobj->base.grclass = class;
+
+	g.channel = chan->id;
+	g.handle  = handle;
+	g.class   = class;
+	ret = drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GROBJ_ALLOC,
+			      &g, sizeof(g));
+	if (ret) {
+		nouveau_grobj_free((void *)&nvgrobj);
+		return ret;
+	}
+
+	*grobj = &nvgrobj->base;
+	return 0;
+}
+
+int
+nouveau_grobj_ref(struct nouveau_channel *chan, uint32_t handle,
+		  struct nouveau_grobj **grobj)
+{
+	struct nouveau_grobj_priv *nvgrobj;
+
+	if (!chan || !grobj || *grobj)
+		return -EINVAL;
+
+	nvgrobj = calloc(1, sizeof(struct nouveau_grobj_priv));
+	if (!nvgrobj)
+		return -ENOMEM;
+	nvgrobj->base.channel = chan;
+	nvgrobj->base.handle = handle;
+	nvgrobj->base.grclass = 0;
+
+	*grobj = &nvgrobj->base;
+	return 0;
+}
+
+void
+nouveau_grobj_free(struct nouveau_grobj **grobj)
+{
+	struct nouveau_device_priv *nvdev;
+	struct nouveau_channel_priv *chan;
+	struct nouveau_grobj_priv *nvgrobj;
+
+	if (!grobj || !*grobj)
+		return;
+	nvgrobj = nouveau_grobj(*grobj);
+	*grobj = NULL;
+
+
+	chan = nouveau_channel(nvgrobj->base.channel);
+	nvdev = nouveau_device(chan->base.device);
+
+	if (nvgrobj->base.grclass) {
+		struct drm_nouveau_gpuobj_free f;
+
+		f.channel = chan->drm.channel;
+		f.handle  = nvgrobj->base.handle;
+		drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE,
+				&f, sizeof(f));	
+	}
+	free(nvgrobj);
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_local.h b/src/gallium/winsys/drm/nouveau/nouveau_local.h
new file mode 100644
index 00000000000..e878a408037
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_local.h
@@ -0,0 +1,117 @@
+#ifndef __NOUVEAU_LOCAL_H__
+#define __NOUVEAU_LOCAL_H__
+
+#include "pipe/p_compiler.h"
+#include "nouveau_winsys_pipe.h"
+#include <stdio.h>
+
+struct pipe_buffer;
+
+/* Debug output */
+#define NOUVEAU_MSG(fmt, args...) do {                                         \
+	fprintf(stdout, "nouveau: "fmt, ##args);                               \
+	fflush(stdout);                                                        \
+} while(0)
+
+#define NOUVEAU_ERR(fmt, args...) do {                                         \
+	fprintf(stderr, "%s:%d -  "fmt, __func__, __LINE__, ##args);           \
+	fflush(stderr);                                                        \
+} while(0)
+
+#define NOUVEAU_TIME_MSEC() 0
+
+/* User FIFO control */
+//#define NOUVEAU_DMA_TRACE
+//#define NOUVEAU_DMA_DEBUG
+//#define NOUVEAU_DMA_DUMP_POSTRELOC_PUSHBUF
+#define NOUVEAU_DMA_BARRIER 
+#define NOUVEAU_DMA_TIMEOUT 2000
+
+/* Push buffer access macros */
+static INLINE void
+OUT_RING(struct nouveau_channel *chan, unsigned data)
+{
+	*(chan->pushbuf->cur++) = (data);
+}
+
+static INLINE void
+OUT_RINGp(struct nouveau_channel *chan, uint32_t *data, unsigned size)
+{
+	memcpy(chan->pushbuf->cur, data, size * 4);
+	chan->pushbuf->cur += size;
+}
+
+static INLINE void
+OUT_RINGf(struct nouveau_channel *chan, float f)
+{
+	union { uint32_t i; float f; } c;
+	c.f = f;
+	OUT_RING(chan, c.i);
+}
+
+static INLINE void
+BEGIN_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr,
+	   unsigned mthd, unsigned size)
+{
+	if (chan->pushbuf->remaining < (size + 1))
+		nouveau_pushbuf_flush(chan, (size + 1));
+	OUT_RING(chan, (gr->subc << 13) | (size << 18) | mthd);
+	chan->pushbuf->remaining -= (size + 1);
+}
+
+static INLINE void
+FIRE_RING(struct nouveau_channel *chan)
+{
+	nouveau_pushbuf_flush(chan, 0);
+}
+
+static INLINE void
+BIND_RING(struct nouveau_channel *chan, struct nouveau_grobj *gr, unsigned subc)
+{
+	gr->subc = subc;
+	BEGIN_RING(chan, gr, 0x0000, 1);
+	OUT_RING  (chan, gr->handle);
+}
+
+static INLINE void
+OUT_RELOC(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	  unsigned data, unsigned flags, unsigned vor, unsigned tor)
+{
+	nouveau_pushbuf_emit_reloc(chan, chan->pushbuf->cur++, bo,
+				   data, flags, vor, tor);
+}
+
+/* Raw data + flags depending on FB/TT buffer */
+static INLINE void
+OUT_RELOCd(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	   unsigned data, unsigned flags, unsigned vor, unsigned tor)
+{
+	OUT_RELOC(chan, bo, data, flags | NOUVEAU_BO_OR, vor, tor);
+}
+
+/* FB/TT object handle */
+static INLINE void
+OUT_RELOCo(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	   unsigned flags)
+{
+	OUT_RELOC(chan, bo, 0, flags | NOUVEAU_BO_OR,
+		  chan->vram->handle, chan->gart->handle);
+}
+
+/* Low 32-bits of offset */
+static INLINE void
+OUT_RELOCl(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	   unsigned delta, unsigned flags)
+{
+	OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_LOW, 0, 0);
+}
+
+/* High 32-bits of offset */
+static INLINE void
+OUT_RELOCh(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	   unsigned delta, unsigned flags)
+{
+	OUT_RELOC(chan, bo, delta, flags | NOUVEAU_BO_HIGH, 0, 0);
+}
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_lock.c b/src/gallium/winsys/drm/nouveau/nouveau_lock.c
new file mode 100644
index 00000000000..9adb9ac8547
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_lock.c
@@ -0,0 +1,94 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "glapi/glthread.h"
+#include <GL/internal/glcore.h>
+
+#include "nouveau_context.h"
+#include "nouveau_screen.h"
+
+_glthread_DECLARE_STATIC_MUTEX( lockMutex );
+
+static void
+nouveau_contended_lock(struct nouveau_context *nv, GLuint flags)
+{
+	__DRIdrawablePrivate *dPriv = nv->dri_drawable;
+	__DRIscreenPrivate *sPriv = nv->dri_screen;
+	struct nouveau_screen *nv_screen = nv->nv_screen;
+	struct nouveau_device *dev = nv_screen->device;
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+
+	drmGetLock(nvdev->fd, nvdev->ctx, flags);
+
+	/* If the window moved, may need to set a new cliprect now.
+	 *
+	 * NOTE: This releases and regains the hw lock, so all state
+	 * checking must be done *after* this call:
+	 */
+	if (dPriv)
+		DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
+}
+
+/* Lock the hardware and validate our state.
+ */
+void
+LOCK_HARDWARE(struct nouveau_context *nv)
+{
+	struct nouveau_screen *nv_screen = nv->nv_screen;
+	struct nouveau_device *dev = nv_screen->device;
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+	char __ret=0;
+
+	_glthread_LOCK_MUTEX(lockMutex);
+	assert(!nv->locked);
+	
+	DRM_CAS(nvdev->lock, nvdev->ctx,
+		(DRM_LOCK_HELD | nvdev->ctx), __ret);
+	
+	if (__ret)
+		nouveau_contended_lock(nv, 0);
+	nv->locked = GL_TRUE;
+}
+
+
+  /* Unlock the hardware using the global current context 
+   */
+void
+UNLOCK_HARDWARE(struct nouveau_context *nv)
+{
+	struct nouveau_screen *nv_screen = nv->nv_screen;
+	struct nouveau_device *dev = nv_screen->device;
+	struct nouveau_device_priv *nvdev = nouveau_device(dev);
+
+	assert(nv->locked);
+	nv->locked = GL_FALSE;
+
+	DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx);
+
+	_glthread_UNLOCK_MUTEX(lockMutex);
+} 
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_notifier.c b/src/gallium/winsys/drm/nouveau/nouveau_notifier.c
new file mode 100644
index 00000000000..01e8f38440e
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_notifier.c
@@ -0,0 +1,137 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_local.h"
+
+#define NOTIFIER(__v)                                                          \
+	struct nouveau_notifier_priv *nvnotify = nouveau_notifier(notifier);   \
+	volatile uint32_t *__v = (void*)nvnotify->map + (id * 32)
+
+int
+nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle,
+		       int count, struct nouveau_notifier **notifier)
+{
+	struct nouveau_notifier_priv *nvnotify;
+	int ret;
+
+	if (!chan || !notifier || *notifier)
+		return -EINVAL;
+
+	nvnotify = calloc(1, sizeof(struct nouveau_notifier_priv));
+	if (!nvnotify)
+		return -ENOMEM;
+	nvnotify->base.channel = chan;
+	nvnotify->base.handle  = handle;
+
+	nvnotify->drm.channel = chan->id;
+	nvnotify->drm.handle  = handle;
+	nvnotify->drm.count   = count;
+	if ((ret = drmCommandWriteRead(nouveau_device(chan->device)->fd,
+				       DRM_NOUVEAU_NOTIFIEROBJ_ALLOC,
+				       &nvnotify->drm,
+				       sizeof(nvnotify->drm)))) {
+		nouveau_notifier_free((void *)&nvnotify);
+		return ret;
+	}
+
+	nvnotify->map = (void *)nouveau_channel(chan)->notifier_block +
+				nvnotify->drm.offset;
+	*notifier = &nvnotify->base;
+	return 0;
+}
+
+void
+nouveau_notifier_free(struct nouveau_notifier **notifier)
+{
+
+	struct nouveau_notifier_priv *nvnotify;
+	struct nouveau_channel_priv *nvchan;
+	struct nouveau_device_priv *nvdev;
+	struct drm_nouveau_gpuobj_free f;
+
+	if (!notifier || !*notifier)
+		return;
+	nvnotify = nouveau_notifier(*notifier);
+	*notifier = NULL;
+
+	nvchan = nouveau_channel(nvnotify->base.channel);
+	nvdev   = nouveau_device(nvchan->base.device);
+
+	f.channel = nvchan->drm.channel;
+	f.handle  = nvnotify->base.handle;
+	drmCommandWrite(nvdev->fd, DRM_NOUVEAU_GPUOBJ_FREE, &f, sizeof(f));		
+	free(nvnotify);
+}
+
+void
+nouveau_notifier_reset(struct nouveau_notifier *notifier, int id)
+{
+	NOTIFIER(n);
+
+	n[NV_NOTIFY_TIME_0      /4] = 0x00000000;
+	n[NV_NOTIFY_TIME_1      /4] = 0x00000000;
+	n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000;
+	n[NV_NOTIFY_STATE       /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS <<
+				       NV_NOTIFY_STATE_STATUS_SHIFT);
+}
+
+uint32_t
+nouveau_notifier_status(struct nouveau_notifier *notifier, int id)
+{
+	NOTIFIER(n);
+
+	return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT;
+}
+
+uint32_t
+nouveau_notifier_return_val(struct nouveau_notifier *notifier, int id)
+{
+	NOTIFIER(n);
+
+	return n[NV_NOTIFY_RETURN_VALUE/4];
+}
+
+int
+nouveau_notifier_wait_status(struct nouveau_notifier *notifier, int id,
+			     int status, int timeout)
+{
+	NOTIFIER(n);
+	uint32_t time = 0, t_start = NOUVEAU_TIME_MSEC();
+
+	while (time <= timeout) {
+		uint32_t v;
+
+		v = n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT;
+		if (v == status)
+			return 0;
+
+		if (timeout)
+			time = NOUVEAU_TIME_MSEC() - t_start;
+	}
+
+	return -EBUSY;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c
new file mode 100644
index 00000000000..815046ba85f
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_pushbuf.c
@@ -0,0 +1,271 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <errno.h>
+#include <assert.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+
+#define PB_BUFMGR_DWORDS   (4096 / 2)
+#define PB_MIN_USER_DWORDS  2048
+
+static int
+nouveau_pushbuf_space(struct nouveau_channel *chan, unsigned min)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_pushbuf_priv *nvpb = &nvchan->pb;
+
+	assert((min + 1) <= nvchan->dma->max);
+
+	/* Wait for enough space in push buffer */
+	min = min < PB_MIN_USER_DWORDS ? PB_MIN_USER_DWORDS : min;
+	min += 1; /* a bit extra for the NOP */
+	if (nvchan->dma->free < min)
+		WAIT_RING_CH(chan, min);
+
+	/* Insert NOP, may turn into a jump later */
+	RING_SPACE_CH(chan, 1);
+	nvpb->nop_jump = nvchan->dma->cur;
+	OUT_RING_CH(chan, 0);
+
+	/* Any remaining space is available to the user */
+	nvpb->start = nvchan->dma->cur;
+	nvpb->size = nvchan->dma->free;
+	nvpb->base.channel = chan;
+	nvpb->base.remaining = nvpb->size;
+	nvpb->base.cur = &nvchan->pushbuf[nvpb->start];
+
+	/* Create a new fence object for this "frame" */
+	nouveau_fence_ref(NULL, &nvpb->fence);
+	nouveau_fence_new(chan, &nvpb->fence);
+
+	return 0;
+}
+
+int
+nouveau_pushbuf_init(struct nouveau_channel *chan)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_dma_priv *m = &nvchan->dma_master;
+	struct nouveau_dma_priv *b = &nvchan->dma_bufmgr;
+	int i;
+
+	if (!nvchan)
+		return -EINVAL;
+
+	/* Reassign last bit of push buffer for a "separate" bufmgr
+	 * ring buffer
+	 */
+	m->max -= PB_BUFMGR_DWORDS;
+	m->free -= PB_BUFMGR_DWORDS;
+
+	b->base = m->base + ((m->max + 2) << 2);
+	b->max = PB_BUFMGR_DWORDS - 2;
+	b->cur = b->put = 0;
+	b->free = b->max - b->cur;
+
+	/* Some NOPs just to be safe
+	 *XXX: RING_SKIPS
+	 */
+	nvchan->dma = b;
+	RING_SPACE_CH(chan, 8);
+	for (i = 0; i < 8; i++)
+		OUT_RING_CH(chan, 0);
+	nvchan->dma = m;
+
+	nouveau_pushbuf_space(chan, 0);
+	chan->pushbuf = &nvchan->pb.base;
+
+	nvchan->pb.buffers = calloc(NOUVEAU_PUSHBUF_MAX_BUFFERS,
+				    sizeof(struct nouveau_pushbuf_bo));
+	nvchan->pb.relocs = calloc(NOUVEAU_PUSHBUF_MAX_RELOCS,
+				   sizeof(struct nouveau_pushbuf_reloc));
+	return 0;
+}
+
+static uint32_t
+nouveau_pushbuf_calc_reloc(struct nouveau_bo *bo,
+			   struct nouveau_pushbuf_reloc *r)
+{
+	uint32_t push;
+
+	if (r->flags & NOUVEAU_BO_LOW) {
+		push = bo->offset + r->data;
+	} else
+	if (r->flags & NOUVEAU_BO_HIGH) {
+		push = (bo->offset + r->data) >> 32;
+	} else {
+		push = r->data;
+	}
+
+	if (r->flags & NOUVEAU_BO_OR) {
+		if (bo->flags & NOUVEAU_BO_VRAM)
+			push |= r->vor;
+		else
+			push |= r->tor;
+	}
+
+	return push;
+}
+
+/* This would be our TTM "superioctl" */
+int
+nouveau_pushbuf_flush(struct nouveau_channel *chan, unsigned min)
+{
+	struct nouveau_channel_priv *nvchan = nouveau_channel(chan);
+	struct nouveau_pushbuf_priv *nvpb = &nvchan->pb;
+	int ret, i;
+
+	if (nvpb->base.remaining == nvpb->size)
+		return 0;
+
+	nouveau_fence_flush(chan);
+
+	nvpb->size -= nvpb->base.remaining;
+	nvchan->dma->cur += nvpb->size;
+	nvchan->dma->free -= nvpb->size;
+	assert(nvchan->dma->cur <= nvchan->dma->max);
+
+	nvchan->dma = &nvchan->dma_bufmgr;
+	nvchan->pushbuf[nvpb->nop_jump] = 0x20000000 |
+		(nvchan->dma->base + (nvchan->dma->cur << 2));
+
+	/* Validate buffers + apply relocations */
+	nvchan->user_charge = 0;
+	for (i = 0; i < nvpb->nr_relocs; i++) {
+		struct nouveau_pushbuf_reloc *r = &nvpb->relocs[i];
+		struct nouveau_pushbuf_bo *pbbo = r->pbbo;
+		struct nouveau_bo *bo = pbbo->bo;
+
+		/* Validated, mem matches presumed, no relocation necessary */
+		if (pbbo->handled & 2) {
+			if (!(pbbo->handled & 1))
+				assert(0);
+			continue;
+		}
+
+		/* Not yet validated, do it now */
+		if (!(pbbo->handled & 1)) {
+			ret = nouveau_bo_validate(chan, bo, pbbo->flags);
+			if (ret) {
+				assert(0);
+				return ret;
+			}
+			pbbo->handled |= 1;
+
+			if (bo->offset == nouveau_bo(bo)->offset &&
+			    bo->flags == nouveau_bo(bo)->flags) {
+				pbbo->handled |= 2;
+				continue;
+			}
+			bo->offset = nouveau_bo(bo)->offset;
+			bo->flags = nouveau_bo(bo)->flags;
+		}
+
+		/* Apply the relocation */
+		*r->ptr = nouveau_pushbuf_calc_reloc(bo, r);
+	}
+	nvpb->nr_relocs = 0;
+
+	/* Dereference all buffers on validate list */
+	for (i = 0; i < nvpb->nr_buffers; i++) {
+		struct nouveau_pushbuf_bo *pbbo = &nvpb->buffers[i];
+
+		nouveau_bo(pbbo->bo)->pending = NULL;
+		nouveau_bo_del(&pbbo->bo);
+	}
+	nvpb->nr_buffers = 0;
+
+	/* Switch back to user's ring */
+	RING_SPACE_CH(chan, 1);
+	OUT_RING_CH(chan, 0x20000000 | ((nvpb->start << 2) +
+					nvchan->dma_master.base));
+	nvchan->dma = &nvchan->dma_master;
+
+	/* Fence + kickoff */
+	nouveau_fence_emit(nvpb->fence);
+	FIRE_RING_CH(chan);
+
+	/* Allocate space for next push buffer */
+	ret = nouveau_pushbuf_space(chan, min);
+	assert(!ret);
+
+	return 0;
+}
+
+static struct nouveau_pushbuf_bo *
+nouveau_pushbuf_emit_buffer(struct nouveau_channel *chan, struct nouveau_bo *bo)
+{
+	struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf);
+	struct nouveau_bo_priv *nvbo = nouveau_bo(bo);
+	struct nouveau_pushbuf_bo *pbbo;
+
+	if (nvbo->pending)
+		return nvbo->pending;
+
+	if (nvpb->nr_buffers >= NOUVEAU_PUSHBUF_MAX_BUFFERS)
+		return NULL;
+	pbbo = nvpb->buffers + nvpb->nr_buffers++;
+	nvbo->pending = pbbo;
+
+	nouveau_bo_ref(bo->device, bo->handle, &pbbo->bo);
+	pbbo->channel = chan;
+	pbbo->flags = NOUVEAU_BO_VRAM | NOUVEAU_BO_GART;
+	pbbo->handled = 0;
+	return pbbo;
+}
+
+int
+nouveau_pushbuf_emit_reloc(struct nouveau_channel *chan, void *ptr,
+			   struct nouveau_bo *bo, uint32_t data, uint32_t flags,
+			   uint32_t vor, uint32_t tor)
+{
+	struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(chan->pushbuf);
+	struct nouveau_pushbuf_bo *pbbo;
+	struct nouveau_pushbuf_reloc *r;
+
+	if (nvpb->nr_relocs >= NOUVEAU_PUSHBUF_MAX_RELOCS)
+		return -ENOMEM;
+
+	pbbo = nouveau_pushbuf_emit_buffer(chan, bo);
+	if (!pbbo)
+		return -ENOMEM;
+	pbbo->flags |= (flags & NOUVEAU_BO_RDWR);
+	pbbo->flags &= (flags | NOUVEAU_BO_RDWR);
+
+	r = nvpb->relocs + nvpb->nr_relocs++;
+	r->pbbo = pbbo;
+	r->ptr = ptr;
+	r->flags = flags;
+	r->data = data;
+	r->vor = vor;
+	r->tor = tor;
+
+	if (flags & NOUVEAU_BO_DUMMY)
+		*(uint32_t *)ptr = 0;
+	else
+		*(uint32_t *)ptr = nouveau_pushbuf_calc_reloc(bo, r);
+	return 0;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_resource.c b/src/gallium/winsys/drm/nouveau/nouveau_resource.c
new file mode 100644
index 00000000000..3bbcb5c45e0
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_resource.c
@@ -0,0 +1,116 @@
+/*
+ * Copyright 2007 Nouveau Project
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <errno.h>
+
+#include "nouveau_drmif.h"
+#include "nouveau_local.h"
+
+int
+nouveau_resource_init(struct nouveau_resource **heap,
+		      unsigned start, unsigned size)
+{
+	struct nouveau_resource *r;
+
+	r = calloc(1, sizeof(struct nouveau_resource));
+	if (!r)
+		return 1;
+
+	r->start = start;
+	r->size  = size;
+	*heap = r;
+	return 0;
+}
+
+int
+nouveau_resource_alloc(struct nouveau_resource *heap, int size, void *priv,
+		       struct nouveau_resource **res)
+{
+	struct nouveau_resource *r;
+
+	if (!heap || !size || !res || *res)
+		return 1;
+
+	while (heap) {
+		if (!heap->in_use && heap->size >= size) {
+			r = calloc(1, sizeof(struct nouveau_resource));
+			if (!r)
+				return 1;
+
+			r->start  = (heap->start + heap->size) - size;
+			r->size   = size;
+			r->in_use = 1;
+			r->priv   = priv;
+
+			heap->size -= size;
+
+			r->next = heap->next;
+			if (heap->next)
+				heap->next->prev = r;
+			r->prev = heap;
+			heap->next = r;
+
+			*res = r;
+			return 0;
+		}
+			
+		heap = heap->next;
+	}
+
+	return 1;
+}
+
+void
+nouveau_resource_free(struct nouveau_resource **res)
+{
+	struct nouveau_resource *r;
+
+	if (!res || !*res)
+		return;
+	r = *res;
+	*res = NULL;
+
+	r->in_use = 0;
+
+	if (r->next && !r->next->in_use) {
+		struct nouveau_resource *new = r->next;
+
+		new->prev = r->prev;
+		if (r->prev)
+			r->prev->next = new;
+		new->size += r->size;
+		new->start = r->start;
+
+		free(r);
+		r = new;
+	}
+
+	if (r->prev && !r->prev->in_use) {
+		r->prev->next = r->next;
+		if (r->next)
+			r->next->prev = r->prev;
+		r->prev->size += r->size;
+		free(r);
+	}
+	
+}
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.c b/src/gallium/winsys/drm/nouveau/nouveau_screen.c
new file mode 100644
index 00000000000..c6d0c535887
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.c
@@ -0,0 +1,265 @@
+#include "utils.h"
+#include "vblank.h"
+#include "xmlpool.h"
+
+#include "pipe/p_context.h"
+#include "state_tracker/st_public.h"
+#include "state_tracker/st_cb_fbo.h"
+
+#include "nouveau_context.h"
+#include "nouveau_drm.h"
+#include "nouveau_dri.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+
+#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11
+#error nouveau_drm.h version does not match expected version
+#endif
+
+/* Extension stuff, enabling of extensions handled by Gallium's GL state
+ * tracker.  But, we still need to define the entry points we want.
+ */
+#define need_GL_ARB_fragment_program
+#define need_GL_ARB_multisample
+#define need_GL_ARB_occlusion_query
+#define need_GL_ARB_point_parameters
+#define need_GL_ARB_shader_objects
+#define need_GL_ARB_texture_compression
+#define need_GL_ARB_vertex_program
+#define need_GL_ARB_vertex_shader
+#define need_GL_ARB_vertex_buffer_object
+#define need_GL_EXT_compiled_vertex_array
+#define need_GL_EXT_fog_coord
+#define need_GL_EXT_secondary_color
+#define need_GL_EXT_framebuffer_object
+#define need_GL_VERSION_2_0
+#define need_GL_VERSION_2_1
+#include "extension_helper.h"
+
+const struct dri_extension card_extensions[] =
+{
+	{ "GL_ARB_multisample", GL_ARB_multisample_functions },
+	{ "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions },
+	{ "GL_ARB_point_parameters", GL_ARB_point_parameters_functions },
+	{ "GL_ARB_shader_objects", GL_ARB_shader_objects_functions },
+	{ "GL_ARB_shading_language_100", GL_VERSION_2_0_functions },
+	{ "GL_ARB_shading_language_120", GL_VERSION_2_1_functions },
+	{ "GL_ARB_texture_compression", GL_ARB_texture_compression_functions },
+	{ "GL_ARB_vertex_program", GL_ARB_vertex_program_functions },
+	{ "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions },
+	{ "GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions },
+	{ "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions },
+	{ "GL_EXT_fog_coord", GL_EXT_fog_coord_functions },
+	{ "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions },
+	{ "GL_EXT_secondary_color", GL_EXT_secondary_color_functions },
+	{ NULL, 0 }
+};
+
+PUBLIC const char __driConfigOptions[] =
+DRI_CONF_BEGIN
+DRI_CONF_END;
+static const GLuint __driNConfigOptions = 0;
+
+extern const struct dri_extension common_extensions[];
+extern const struct dri_extension nv40_extensions[];
+
+static GLboolean
+nouveau_create_buffer(__DRIscreenPrivate * driScrnPriv,
+		      __DRIdrawablePrivate * driDrawPriv,
+		      const __GLcontextModes *glVis, GLboolean pixmapBuffer)
+{
+	struct nouveau_framebuffer *nvfb;
+	enum pipe_format colour, depth, stencil;
+
+	if (pixmapBuffer)
+		return GL_FALSE;
+
+	nvfb = CALLOC_STRUCT(nouveau_framebuffer);
+	if (!nvfb)
+		return GL_FALSE;
+
+	if (glVis->redBits == 5)
+		colour = PIPE_FORMAT_R5G6B5_UNORM;
+	else
+		colour = PIPE_FORMAT_A8R8G8B8_UNORM;
+
+	if (glVis->depthBits == 16)
+		depth = PIPE_FORMAT_Z16_UNORM;
+	else if (glVis->depthBits == 24)
+		depth = PIPE_FORMAT_Z24S8_UNORM;
+	else
+		depth = PIPE_FORMAT_NONE;
+
+	if (glVis->stencilBits == 8)
+		stencil = PIPE_FORMAT_Z24S8_UNORM;
+	else
+		stencil = PIPE_FORMAT_NONE;
+
+	nvfb->stfb = st_create_framebuffer(glVis, colour, depth, stencil,
+					   driDrawPriv->w, driDrawPriv->h,
+					   (void*)nvfb);
+	if (!nvfb->stfb) {
+		free(nvfb);
+		return  GL_FALSE;
+	}
+
+	driDrawPriv->driverPrivate = (void *)nvfb;
+	return GL_TRUE;
+}
+
+static void
+nouveau_destroy_buffer(__DRIdrawablePrivate * driDrawPriv)
+{
+	struct nouveau_framebuffer *nvfb;
+	
+	nvfb = (struct nouveau_framebuffer *)driDrawPriv->driverPrivate;
+	st_unreference_framebuffer(nvfb->stfb);
+	free(nvfb);
+}
+
+static __DRIconfig **
+nouveau_fill_in_modes(__DRIscreenPrivate *psp,
+		      unsigned pixel_bits, unsigned depth_bits,
+		      unsigned stencil_bits, GLboolean have_back_buffer)
+{
+	__DRIconfig **configs;
+	unsigned depth_buffer_factor;
+	unsigned back_buffer_factor;
+	GLenum fb_format;
+	GLenum fb_type;
+
+	static const GLenum back_buffer_modes[] = {
+		GLX_NONE, GLX_SWAP_UNDEFINED_OML,
+	};
+
+	uint8_t depth_bits_array[3];
+	uint8_t stencil_bits_array[3];
+	uint8_t msaa_samples_array[1];
+
+	depth_bits_array[0] = 0;
+	depth_bits_array[1] = depth_bits;
+	depth_bits_array[2] = depth_bits;
+
+	/* Just like with the accumulation buffer, always provide some modes
+	 * with a stencil buffer.  It will be a sw fallback, but some apps won't
+	 * care about that.
+	 */
+	stencil_bits_array[0] = 0;
+	stencil_bits_array[1] = 0;
+	if (depth_bits == 24)
+		stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
+	stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
+
+	msaa_samples_array[0] = 0;
+
+	depth_buffer_factor =
+		((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
+	back_buffer_factor = (have_back_buffer) ? 3 : 1;
+
+	if (pixel_bits == 16) {
+		fb_format = GL_RGB;
+		fb_type = GL_UNSIGNED_SHORT_5_6_5;
+	}
+	else {
+		fb_format = GL_BGRA;
+		fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+	}
+
+	configs = driCreateConfigs(fb_format, fb_type,
+				   depth_bits_array, stencil_bits_array,
+				   depth_buffer_factor, back_buffer_modes,
+				   back_buffer_factor, msaa_samples_array, 1);
+	if (configs == NULL) {
+	 fprintf(stderr, "[%s:%u] Error creating FBConfig!\n",
+			 __func__, __LINE__);
+		return NULL;
+	}
+
+	return configs;
+}
+
+static const __DRIconfig **
+nouveau_screen_create(__DRIscreenPrivate *psp)
+{
+	struct nouveau_dri *nv_dri = psp->pDevPriv;
+	struct nouveau_screen *nv_screen;
+	static const __DRIversion ddx_expected =
+		{ 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL };
+	static const __DRIversion dri_expected = { 4, 0, 0 };
+	static const __DRIversion drm_expected =
+		{ 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL };
+	int ret;
+
+	if (!driCheckDriDdxDrmVersions2("nouveau",
+					&psp->dri_version, &dri_expected,
+					&psp->ddx_version, &ddx_expected,
+					&psp->drm_version, &drm_expected)) {
+		return NULL;
+	}
+
+	if (drm_expected.patch != psp->drm_version.patch) {
+		fprintf(stderr, "Incompatible DRM patch level.\n"
+				"Expected: %d\n" "Current : %d\n",
+			drm_expected.patch, psp->drm_version.patch);
+		return NULL;
+	}
+
+	driInitExtensions(NULL, card_extensions, GL_FALSE);
+
+	if (psp->devPrivSize != sizeof(struct nouveau_dri)) {
+		NOUVEAU_ERR("DRI struct mismatch between DDX/DRI\n");
+		return GL_FALSE;
+	}
+
+	nv_screen = CALLOC_STRUCT(nouveau_screen);
+	if (!nv_screen)
+		return GL_FALSE;
+	nv_screen->driScrnPriv = psp;
+	psp->private = (void *)nv_screen;
+
+	driParseOptionInfo(&nv_screen->option_cache,
+			   __driConfigOptions, __driNConfigOptions);
+
+	if ((ret = nouveau_device_open_existing(&nv_screen->device, 0,
+						psp->fd, 0))) {
+		NOUVEAU_ERR("Failed opening nouveau device: %d\n", ret);
+		return GL_FALSE;
+	}
+
+	nv_screen->front_offset = nv_dri->front_offset;
+	nv_screen->front_pitch  = nv_dri->front_pitch * (nv_dri->bpp / 8);
+	nv_screen->front_cpp = nv_dri->bpp / 8;
+	nv_screen->front_height = nv_dri->height;
+
+	return (const __DRIconfig **)
+		nouveau_fill_in_modes(psp, nv_dri->bpp,
+				      (nv_dri->bpp == 16) ? 16 : 24,
+				      (nv_dri->bpp == 16) ? 0 : 8, 1);
+}
+
+static void
+nouveau_screen_destroy(__DRIscreenPrivate *driScrnPriv)
+{
+	struct nouveau_screen *nv_screen = driScrnPriv->private;
+
+	driScrnPriv->private = NULL;
+	FREE(nv_screen);
+}
+
+const struct __DriverAPIRec
+driDriverAPI = {
+	.InitScreen	= nouveau_screen_create,
+	.DestroyScreen	= nouveau_screen_destroy,
+	.CreateContext	= nouveau_context_create,
+	.DestroyContext	= nouveau_context_destroy,
+	.CreateBuffer	= nouveau_create_buffer,
+	.DestroyBuffer	= nouveau_destroy_buffer,
+	.SwapBuffers	= nouveau_swap_buffers,
+	.MakeCurrent	= nouveau_context_bind,
+	.UnbindContext	= nouveau_context_unbind,
+	.CopySubBuffer	= nouveau_copy_sub_buffer,
+
+	.InitScreen2	= NULL, /* one day, I promise! */
+};
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_screen.h b/src/gallium/winsys/drm/nouveau/nouveau_screen.h
new file mode 100644
index 00000000000..388d6be9bbc
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_screen.h
@@ -0,0 +1,20 @@
+#ifndef __NOUVEAU_SCREEN_H__
+#define __NOUVEAU_SCREEN_H__
+
+#include "xmlconfig.h"
+
+struct nouveau_screen {
+	__DRIscreenPrivate *driScrnPriv;
+	driOptionCache      option_cache;
+
+	struct nouveau_device *device;
+
+	uint32_t front_offset;
+	uint32_t front_pitch;
+	uint32_t front_cpp;
+	uint32_t front_height;
+
+	void *nvc;
+};
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c
new file mode 100644
index 00000000000..70e0104e83b
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.c
@@ -0,0 +1,86 @@
+#include "main/glheader.h"
+#include "glapi/glthread.h"
+#include <GL/internal/glcore.h>
+
+#include "pipe/p_context.h"
+#include "state_tracker/st_public.h"
+#include "state_tracker/st_context.h"
+#include "state_tracker/st_cb_fbo.h"
+
+#include "nouveau_context.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+
+void
+nouveau_copy_buffer(__DRIdrawablePrivate *dPriv, struct pipe_surface *surf,
+		    const drm_clip_rect_t *rect)
+{
+	struct nouveau_context *nv = dPriv->driContextPriv->driverPrivate;
+	drm_clip_rect_t *pbox;
+	int nbox, i;
+
+	LOCK_HARDWARE(nv);
+	if (!dPriv->numClipRects) {
+		UNLOCK_HARDWARE(nv);
+		return;
+	}
+	pbox = dPriv->pClipRects;
+	nbox = dPriv->numClipRects;
+
+	nv->surface_copy_prep(nv, nv->frontbuffer, surf);
+	for (i = 0; i < nbox; i++, pbox++) {
+		int sx, sy, dx, dy, w, h;
+
+		sx = pbox->x1 - dPriv->x;
+		sy = pbox->y1 - dPriv->y;
+		dx = pbox->x1;
+		dy = pbox->y1;
+		w  = pbox->x2 - pbox->x1;
+		h  = pbox->y2 - pbox->y1;
+
+		nv->surface_copy(nv, dx, dy, sx, sy, w, h);
+	}
+
+	FIRE_RING(nv->nvc->channel);
+	UNLOCK_HARDWARE(nv);
+
+	if (nv->last_stamp != dPriv->lastStamp) {
+		struct nouveau_framebuffer *nvfb = dPriv->driverPrivate;
+		st_resize_framebuffer(nvfb->stfb, dPriv->w, dPriv->h);
+		nv->last_stamp = dPriv->lastStamp;
+	}
+}
+
+void
+nouveau_copy_sub_buffer(__DRIdrawablePrivate *dPriv, int x, int y, int w, int h)
+{
+	struct nouveau_framebuffer *nvfb = dPriv->driverPrivate;
+	struct pipe_surface *surf;
+
+	surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT);
+	if (surf) {
+		drm_clip_rect_t rect;
+		rect.x1 = x;
+		rect.y1 = y;
+		rect.x2 = x + w;
+		rect.y2 = y + h;
+
+		st_notify_swapbuffers(nvfb->stfb);
+		nouveau_copy_buffer(dPriv, surf, &rect);
+	}
+}
+
+void
+nouveau_swap_buffers(__DRIdrawablePrivate *dPriv)
+{
+	struct nouveau_framebuffer *nvfb = dPriv->driverPrivate;
+	struct pipe_surface *surf;
+
+	surf = st_get_framebuffer_surface(nvfb->stfb, ST_SURFACE_BACK_LEFT);
+	if (surf) {
+		st_notify_swapbuffers(nvfb->stfb);
+		nouveau_copy_buffer(dPriv, surf, NULL);
+	}
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h
new file mode 100644
index 00000000000..825d3da6da5
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_swapbuffers.h
@@ -0,0 +1,10 @@
+#ifndef __NOUVEAU_SWAPBUFFERS_H__
+#define __NOUVEAU_SWAPBUFFERS_H__
+
+extern void nouveau_copy_buffer(__DRIdrawablePrivate *, struct pipe_surface *,
+				const drm_clip_rect_t *);
+extern void nouveau_copy_sub_buffer(__DRIdrawablePrivate *,
+				    int x, int y, int w, int h);
+extern void nouveau_swap_buffers(__DRIdrawablePrivate *);
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c
new file mode 100644
index 00000000000..364340e1d3d
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys.c
@@ -0,0 +1,161 @@
+#include "util/u_memory.h"
+
+#include "nouveau_context.h"
+#include "nouveau_screen.h"
+#include "nouveau_winsys_pipe.h"
+
+#include "nouveau/nouveau_winsys.h"
+
+static int
+nouveau_pipe_notifier_alloc(struct nouveau_winsys *nvws, int count,
+			    struct nouveau_notifier **notify)
+{
+	struct nouveau_context *nv = nvws->nv;
+
+	return nouveau_notifier_alloc(nv->nvc->channel, nv->nvc->next_handle++,
+				      count, notify);
+}
+
+static int
+nouveau_pipe_grobj_alloc(struct nouveau_winsys *nvws, int grclass,
+			 struct nouveau_grobj **grobj)
+{
+	struct nouveau_context *nv = nvws->nv;
+	struct nouveau_channel *chan = nv->nvc->channel;
+	int ret;
+
+	ret = nouveau_grobj_alloc(chan, nv->nvc->next_handle++,
+				  grclass, grobj);
+	if (ret)
+		return ret;
+
+	assert(nv->nvc->next_subchannel < 7);
+	BIND_RING(chan, *grobj, nv->nvc->next_subchannel++);
+	return 0;
+}
+
+static int
+nouveau_pipe_surface_copy(struct nouveau_winsys *nvws, struct pipe_surface *dst,
+			  unsigned dx, unsigned dy, struct pipe_surface *src,
+			  unsigned sx, unsigned sy, unsigned w, unsigned h)
+{
+	struct nouveau_context *nv = nvws->nv;
+
+	if (nv->surface_copy_prep(nv, dst, src))
+		return 1;
+	nv->surface_copy(nv, dx, dy, sx, sy, w, h);
+	nv->surface_copy_done(nv);
+
+	return 0;
+}
+
+static int
+nouveau_pipe_surface_fill(struct nouveau_winsys *nvws, struct pipe_surface *dst,
+			  unsigned dx, unsigned dy, unsigned w, unsigned h,
+			  unsigned value)
+{
+	if (nvws->nv->surface_fill(nvws->nv, dst, dx, dy, w, h, value))
+		return 1;
+	return 0;
+}
+
+static int
+nouveau_pipe_push_reloc(struct nouveau_winsys *nvws, void *ptr,
+			struct pipe_buffer *buf, uint32_t data,
+			uint32_t flags, uint32_t vor, uint32_t tor)
+{
+	return nouveau_pushbuf_emit_reloc(nvws->channel, ptr,
+					  nouveau_buffer(buf)->bo,
+					  data, flags, vor, tor);
+}
+
+static int
+nouveau_pipe_push_flush(struct nouveau_winsys *nvws, unsigned size,
+			struct pipe_fence_handle **fence)
+{
+	if (fence) {
+		struct nouveau_pushbuf *pb = nvws->channel->pushbuf;
+		struct nouveau_pushbuf_priv *nvpb = nouveau_pushbuf(pb);
+		struct nouveau_fence *ref = NULL;
+
+		nouveau_fence_ref(nvpb->fence, &ref);
+		*fence = (struct pipe_fence_handle *)ref;
+	}
+
+	return nouveau_pushbuf_flush(nvws->channel, size);
+}
+
+struct pipe_context *
+nouveau_pipe_create(struct nouveau_context *nv)
+{
+	struct nouveau_channel_context *nvc = nv->nvc;
+	struct nouveau_winsys *nvws = CALLOC_STRUCT(nouveau_winsys);
+	struct pipe_screen *(*hws_create)(struct pipe_winsys *,
+					  struct nouveau_winsys *);
+	struct pipe_context *(*hw_create)(struct pipe_screen *, unsigned);
+	struct pipe_winsys *ws;
+	unsigned chipset = nv->nv_screen->device->chipset;
+
+	if (!nvws)
+		return NULL;
+
+	switch (chipset & 0xf0) {
+	case 0x10:
+		hws_create = nv10_screen_create;
+		hw_create = nv10_create;
+		break;
+	case 0x20:
+		hws_create = nv20_screen_create;
+		hw_create = nv20_create;
+		break;
+	case 0x30:
+		hws_create = nv30_screen_create;
+		hw_create = nv30_create;
+		break;
+	case 0x40:
+	case 0x60:
+		hws_create = nv40_screen_create;
+		hw_create = nv40_create;
+		break;
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		hws_create = nv50_screen_create;
+		hw_create = nv50_create;
+		break;
+	default:
+		NOUVEAU_ERR("Unknown chipset NV%02x\n", chipset);
+		return NULL;
+	}
+
+	nvws->nv		= nv;
+	nvws->channel		= nv->nvc->channel;
+
+	nvws->res_init		= nouveau_resource_init;
+	nvws->res_alloc		= nouveau_resource_alloc;
+	nvws->res_free		= nouveau_resource_free;
+
+	nvws->push_reloc        = nouveau_pipe_push_reloc;
+	nvws->push_flush	= nouveau_pipe_push_flush;
+
+	nvws->grobj_alloc	= nouveau_pipe_grobj_alloc;
+	nvws->grobj_free	= nouveau_grobj_free;
+
+	nvws->notifier_alloc	= nouveau_pipe_notifier_alloc;
+	nvws->notifier_free	= nouveau_notifier_free;
+	nvws->notifier_reset	= nouveau_notifier_reset;
+	nvws->notifier_status	= nouveau_notifier_status;
+	nvws->notifier_retval	= nouveau_notifier_return_val;
+	nvws->notifier_wait	= nouveau_notifier_wait_status;
+
+	nvws->surface_copy	= nouveau_pipe_surface_copy;
+	nvws->surface_fill	= nouveau_pipe_surface_fill;
+
+	ws = nouveau_create_pipe_winsys(nv);
+
+	if (!nvc->pscreen)
+		nvc->pscreen = hws_create(ws, nvws);
+	nvc->pctx[nv->pctx_id] = hw_create(nvc->pscreen, nv->pctx_id);
+	return nvc->pctx[nv->pctx_id];
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c
new file mode 100644
index 00000000000..5276806de6b
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.c
@@ -0,0 +1,206 @@
+#include "pipe/p_winsys.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+
+#include "util/u_memory.h"
+
+#include "nouveau_context.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+#include "nouveau_winsys_pipe.h"
+
+static void
+nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf,
+			  void *context_private)
+{
+	struct nouveau_context *nv = context_private;
+	__DRIdrawablePrivate *dPriv = nv->dri_drawable;
+
+	nouveau_copy_buffer(dPriv, surf, NULL);
+}
+
+static const char *
+nouveau_get_name(struct pipe_winsys *pws)
+{
+	return "Nouveau/DRI";
+}
+
+static struct pipe_buffer *
+nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment,
+		       unsigned usage, unsigned size)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws;
+	struct nouveau_context *nv = nvpws->nv;
+	struct nouveau_device *dev = nv->nv_screen->device;
+	struct nouveau_pipe_buffer *nvbuf;
+	uint32_t flags;
+
+	nvbuf = calloc(1, sizeof(*nvbuf));
+	if (!nvbuf)
+		return NULL;
+	nvbuf->base.refcount = 1;
+	nvbuf->base.alignment = alignment;
+	nvbuf->base.usage = usage;
+	nvbuf->base.size = size;
+
+	flags = NOUVEAU_BO_LOCAL;
+
+	if (usage & PIPE_BUFFER_USAGE_PIXEL) {
+		if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE)
+			flags |= NOUVEAU_BO_GART;
+		flags |= NOUVEAU_BO_VRAM;
+
+		switch (dev->chipset & 0xf0) {
+		case 0x50:
+		case 0x80:
+		case 0x90:
+			flags |= NOUVEAU_BO_TILED;
+			if (usage & NOUVEAU_BUFFER_USAGE_ZETA)
+				flags |= NOUVEAU_BO_ZTILE;
+			break;
+		default:
+			break;
+		}
+	}
+
+	if (usage & PIPE_BUFFER_USAGE_VERTEX) {
+		if (nv->cap.hw_vertex_buffer)
+			flags |= NOUVEAU_BO_GART;
+	}
+
+	if (usage & PIPE_BUFFER_USAGE_INDEX) {
+		if (nv->cap.hw_index_buffer)
+			flags |= NOUVEAU_BO_GART;
+	}
+
+	if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) {
+		free(nvbuf);
+		return NULL;
+	}
+
+	return &nvbuf->base;
+}
+
+static struct pipe_buffer *
+nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws;
+	struct nouveau_device *dev = nvpws->nv->nv_screen->device;
+	struct nouveau_pipe_buffer *nvbuf;
+
+	nvbuf = calloc(1, sizeof(*nvbuf));
+	if (!nvbuf)
+		return NULL;
+	nvbuf->base.refcount = 1;
+	nvbuf->base.size = bytes;
+
+	if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) {
+		free(nvbuf);
+		return NULL;
+	}
+
+	return &nvbuf->base;
+}
+
+static void
+nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+
+	nouveau_bo_del(&nvbuf->bo);
+	free(nvbuf);
+}
+
+static void *
+nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf,
+		    unsigned flags)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+	uint32_t map_flags = 0;
+
+	if (flags & PIPE_BUFFER_USAGE_CPU_READ)
+		map_flags |= NOUVEAU_BO_RD;
+	if (flags & PIPE_BUFFER_USAGE_CPU_WRITE)
+		map_flags |= NOUVEAU_BO_WR;
+
+	if (nouveau_bo_map(nvbuf->bo, map_flags))
+		return NULL;
+	return nvbuf->bo->map;
+}
+
+static void
+nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+
+	nouveau_bo_unmap(nvbuf->bo);
+}
+
+static INLINE struct nouveau_fence *
+nouveau_pipe_fence(struct pipe_fence_handle *pfence)
+{
+	return (struct nouveau_fence *)pfence;
+}
+
+static void
+nouveau_pipe_fence_reference(struct pipe_winsys *ws,
+			     struct pipe_fence_handle **ptr,
+			     struct pipe_fence_handle *pfence)
+{
+	nouveau_fence_ref((void *)pfence, (void *)ptr);
+}
+
+static int
+nouveau_pipe_fence_signalled(struct pipe_winsys *ws,
+			     struct pipe_fence_handle *pfence, unsigned flag)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws;
+	struct nouveau_fence *fence = nouveau_pipe_fence(pfence);
+
+	if (nouveau_fence(fence)->signalled == 0)
+		nouveau_fence_flush(nvpws->nv->nvc->channel);
+
+	return !nouveau_fence(fence)->signalled;
+}
+
+static int
+nouveau_pipe_fence_finish(struct pipe_winsys *ws,
+			  struct pipe_fence_handle *pfence, unsigned flag)
+{
+	struct nouveau_fence *fence = nouveau_pipe_fence(pfence);
+	struct nouveau_fence *ref = NULL;
+
+	nouveau_fence_ref(fence, &ref);
+	return nouveau_fence_wait(&ref);
+}
+
+struct pipe_winsys *
+nouveau_create_pipe_winsys(struct nouveau_context *nv)
+{
+	struct nouveau_pipe_winsys *nvpws;
+	struct pipe_winsys *pws;
+
+	nvpws = CALLOC_STRUCT(nouveau_pipe_winsys);
+	if (!nvpws)
+		return NULL;
+	nvpws->nv = nv;
+	pws = &nvpws->pws;
+
+	pws->flush_frontbuffer = nouveau_flush_frontbuffer;
+
+	pws->buffer_create = nouveau_pipe_bo_create;
+	pws->buffer_destroy = nouveau_pipe_bo_del;
+	pws->user_buffer_create = nouveau_pipe_bo_user_create;
+	pws->buffer_map = nouveau_pipe_bo_map;
+	pws->buffer_unmap = nouveau_pipe_bo_unmap;
+
+	pws->fence_reference = nouveau_pipe_fence_reference;
+	pws->fence_signalled = nouveau_pipe_fence_signalled;
+	pws->fence_finish = nouveau_pipe_fence_finish;
+
+	pws->get_name = nouveau_get_name;
+
+	return &nvpws->pws;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h
new file mode 100644
index 00000000000..6a03ac0d773
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_pipe.h
@@ -0,0 +1,34 @@
+#ifndef NOUVEAU_PIPE_WINSYS_H
+#define NOUVEAU_PIPE_WINSYS_H
+
+#include "pipe/p_context.h"
+#include "pipe/p_winsys.h"
+#include "nouveau_context.h"
+
+struct nouveau_pipe_buffer {
+	struct pipe_buffer base;
+	struct nouveau_bo *bo;
+};
+
+static inline struct nouveau_pipe_buffer *
+nouveau_buffer(struct pipe_buffer *buf)
+{
+	return (struct nouveau_pipe_buffer *)buf;
+}
+
+struct nouveau_pipe_winsys {
+	struct pipe_winsys pws;
+
+	struct nouveau_context *nv;
+};
+
+extern struct pipe_winsys *
+nouveau_create_pipe_winsys(struct nouveau_context *nv);
+
+struct pipe_context *
+nouveau_create_softpipe(struct nouveau_context *nv);
+
+struct pipe_context *
+nouveau_pipe_create(struct nouveau_context *nv);
+
+#endif
diff --git a/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c
new file mode 100644
index 00000000000..68aade829d5
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nouveau_winsys_softpipe.c
@@ -0,0 +1,86 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * 
+ **************************************************************************/
+/*
+ * Authors: Keith Whitwell <keithw-at-tungstengraphics-dot-com>
+ */
+
+#include "imports.h"
+
+#include "pipe/p_defines.h"
+#include "pipe/p_format.h"
+#include "softpipe/sp_winsys.h"
+
+#include "nouveau_context.h"
+#include "nouveau_winsys_pipe.h"
+
+struct nouveau_softpipe_winsys {
+   struct softpipe_winsys sws;
+   struct nouveau_context *nv;
+};
+
+/**
+ * Return list of surface formats supported by this driver.
+ */
+static boolean
+nouveau_is_format_supported(struct softpipe_winsys *sws,
+						enum pipe_format format)
+{
+	switch (format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case PIPE_FORMAT_R5G6B5_UNORM:
+	case PIPE_FORMAT_Z24S8_UNORM:
+		return TRUE;
+	default:
+		break;
+	};
+
+	return FALSE;
+}
+
+struct pipe_context *
+nouveau_create_softpipe(struct nouveau_context *nv)
+{
+	struct nouveau_softpipe_winsys *nvsws;
+	struct pipe_screen *pscreen;
+	struct pipe_winsys *ws;
+
+	ws = nouveau_create_pipe_winsys(nv);
+	if (!ws)
+		return NULL;
+	pscreen = softpipe_create_screen(ws);
+
+	nvsws = CALLOC_STRUCT(nouveau_softpipe_winsys);
+	if (!nvsws)
+		return NULL;
+
+	nvsws->sws.is_format_supported = nouveau_is_format_supported;
+	nvsws->nv = nv;
+
+	return softpipe_create(pscreen, ws, &nvsws->sws);
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nv04_surface.c b/src/gallium/winsys/drm/nouveau/nv04_surface.c
new file mode 100644
index 00000000000..68338eb814a
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nv04_surface.c
@@ -0,0 +1,420 @@
+#include "pipe/p_context.h"
+#include "pipe/p_format.h"
+
+#include "nouveau_context.h"
+
+static INLINE int log2i(int i)
+{
+	int r = 0;
+
+	if (i & 0xffff0000) {
+		i >>= 16;
+		r += 16;
+	}
+	if (i & 0x0000ff00) {
+		i >>= 8;
+		r += 8;
+	}
+	if (i & 0x000000f0) {
+		i >>= 4;
+		r += 4;
+	}
+	if (i & 0x0000000c) {
+		i >>= 2;
+		r += 2;
+	}
+	if (i & 0x00000002) {
+		r += 1;
+	}
+	return r;
+}
+
+static INLINE int
+nv04_surface_format(enum pipe_format format)
+{
+	switch (format) {
+	case PIPE_FORMAT_A8_UNORM:
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_Y8;
+	case PIPE_FORMAT_R16_SNORM:
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_R5G6B5;
+	case PIPE_FORMAT_X8R8G8B8_UNORM:
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_A8R8G8B8;
+	case PIPE_FORMAT_Z24S8_UNORM:
+		return NV04_CONTEXT_SURFACES_2D_FORMAT_Y32;
+	default:
+		return -1;
+	}
+}
+
+static INLINE int
+nv04_rect_format(enum pipe_format format)
+{
+	switch (format) {
+	case PIPE_FORMAT_A8_UNORM:
+		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A16R5G6B5;
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case PIPE_FORMAT_Z24S8_UNORM:
+		return NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT_A8R8G8B8;
+	default:
+		return -1;
+	}
+}
+
+static INLINE int
+nv04_scaled_image_format(enum pipe_format format)
+{
+	switch (format) {
+	case PIPE_FORMAT_A1R5G5B5_UNORM:
+		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A1R5G5B5;
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_A8R8G8B8;
+	case PIPE_FORMAT_X8R8G8B8_UNORM:
+		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_X8R8G8B8;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+	case PIPE_FORMAT_R16_SNORM:
+		return NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT_R5G6B5;
+	default:
+		return -1;
+	}
+}
+
+static void
+nv04_surface_copy_swizzle(struct nouveau_context *nv, unsigned dx, unsigned dy,
+                          unsigned sx, unsigned sy, unsigned w, unsigned h)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct pipe_surface *dst = nv->surf_dst;
+	struct pipe_surface *src = nv->surf_src;
+
+	/* POT or GTFO */
+	assert(!(w & (w - 1)) && !(h & (h - 1)));
+
+	BEGIN_RING(chan, nv->nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_DMA_IMAGE, 1);
+	OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo,
+	                 NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	BEGIN_RING(chan, nv->nvc->NvSwzSurf, NV04_SWIZZLED_SURFACE_FORMAT, 2);
+	OUT_RING  (chan, nv04_surface_format(dst->format) |
+	                 log2i(w) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_U_SHIFT |
+	                 log2i(h) << NV04_SWIZZLED_SURFACE_FORMAT_BASE_SIZE_V_SHIFT);
+	OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset,
+	                 NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	BEGIN_RING(chan, nv->nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE, 1);
+	OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo,
+	                 NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	BEGIN_RING(chan, nv->nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE, 1);
+	OUT_RING  (chan, nv->nvc->NvSwzSurf->handle);
+
+	BEGIN_RING(chan, nv->nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION, 9);
+	OUT_RING  (chan, NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION_TRUNCATE);
+	OUT_RING  (chan, nv04_scaled_image_format(src->format));
+	OUT_RING  (chan, NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION_SRCCOPY);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, h << 16 | w);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, h << 16 | w);
+	OUT_RING  (chan, 1 << 20);
+	OUT_RING  (chan, 1 << 20);
+	BEGIN_RING(chan, nv->nvc->NvSIFM, NV04_SCALED_IMAGE_FROM_MEMORY_SIZE, 4);
+	OUT_RING  (chan, h << 16 | w);
+	OUT_RING  (chan, src->stride |
+	                 NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER |
+	                 NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE);
+	OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset,
+	                 NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RING  (chan, 0);
+}
+
+static void
+nv04_surface_copy_m2mf(struct nouveau_context *nv, unsigned dx, unsigned dy,
+		       unsigned sx, unsigned sy, unsigned w, unsigned h)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct pipe_surface *dst = nv->surf_dst;
+	struct pipe_surface *src = nv->surf_src;
+	unsigned dst_offset, src_offset;
+
+	dst_offset = dst->offset + (dy * dst->stride) + (dx * dst->block.size);
+	src_offset = src->offset + (sy * src->stride) + (sx * src->block.size);
+
+	while (h) {
+		int count = (h > 2047) ? 2047 : h;
+
+		BEGIN_RING(chan, nv->nvc->NvM2MF,
+			   NV04_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
+		OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src_offset,
+			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_RD);
+		OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst_offset,
+			   NOUVEAU_BO_VRAM | NOUVEAU_BO_GART | NOUVEAU_BO_WR);
+		OUT_RING  (chan, src->stride);
+		OUT_RING  (chan, dst->stride);
+		OUT_RING  (chan, w * src->block.size);
+		OUT_RING  (chan, count);
+		OUT_RING  (chan, 0x0101);
+		OUT_RING  (chan, 0);
+
+		h -= count;
+		src_offset += src->stride * count;
+		dst_offset += dst->stride * count;
+	}
+}
+
+static void
+nv04_surface_copy_blit(struct nouveau_context *nv, unsigned dx, unsigned dy,
+		       unsigned sx, unsigned sy, unsigned w, unsigned h)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+
+	BEGIN_RING(chan, nv->nvc->NvImageBlit, 0x0300, 3);
+	OUT_RING  (chan, (sy << 16) | sx);
+	OUT_RING  (chan, (dy << 16) | dx);
+	OUT_RING  (chan, ( h << 16) |  w);
+}
+
+static int
+nv04_surface_copy_prep(struct nouveau_context *nv, struct pipe_surface *dst,
+		       struct pipe_surface *src)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	int format;
+
+	if (src->format != dst->format)
+		return 1;
+
+	/* Setup transfer to swizzle the texture to vram if needed */
+	/* FIXME/TODO: check proper limits of this operation */
+	if (src->texture && dst->texture) {
+		unsigned int src_linear = src->texture->tex_usage &
+		                          NOUVEAU_TEXTURE_USAGE_LINEAR;
+		unsigned int dst_linear = dst->texture->tex_usage &
+		                          NOUVEAU_TEXTURE_USAGE_LINEAR;
+		if (src_linear ^ dst_linear) {
+			nv->surface_copy = nv04_surface_copy_swizzle;
+			nv->surf_dst = dst;
+			nv->surf_src = src;
+			return 0;
+		}
+	}
+
+	/* NV_CONTEXT_SURFACES_2D has buffer alignment restrictions, fallback
+	 * to NV_MEMORY_TO_MEMORY_FORMAT in this case.
+	 */
+	if ((src->offset & 63) || (dst->offset & 63)) {
+		BEGIN_RING(nv->nvc->channel, nv->nvc->NvM2MF,
+			   NV04_MEMORY_TO_MEMORY_FORMAT_DMA_BUFFER_IN, 2);
+		OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo,
+			   NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+		OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo,
+			   NOUVEAU_BO_GART | NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+		nv->surface_copy = nv04_surface_copy_m2mf;
+		nv->surf_dst = dst;
+		nv->surf_src = src;
+		return 0;
+
+	}
+
+	if ((format = nv04_surface_format(dst->format)) < 0) {
+		NOUVEAU_ERR("Bad surface format 0x%x\n", dst->format);
+		return 1;
+	}
+	nv->surface_copy = nv04_surface_copy_blit;
+
+	BEGIN_RING(chan, nv->nvc->NvCtxSurf2D,
+		   NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2);
+	OUT_RELOCo(chan, nouveau_buffer(src->buffer)->bo,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	BEGIN_RING(chan, nv->nvc->NvCtxSurf2D,
+		   NV04_CONTEXT_SURFACES_2D_FORMAT, 4);
+	OUT_RING  (chan, format);
+	OUT_RING  (chan, (dst->stride << 16) | src->stride);
+	OUT_RELOCl(chan, nouveau_buffer(src->buffer)->bo, src->offset,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_RD);
+	OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	return 0;
+}
+
+static void
+nv04_surface_copy_done(struct nouveau_context *nv)
+{
+	FIRE_RING(nv->nvc->channel);
+}
+
+static int
+nv04_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst,
+		  unsigned dx, unsigned dy, unsigned w, unsigned h,
+		  unsigned value)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct nouveau_grobj *surf2d = nv->nvc->NvCtxSurf2D;
+	struct nouveau_grobj *rect = nv->nvc->NvGdiRect;
+	int cs2d_format, gdirect_format;
+
+	if ((cs2d_format = nv04_surface_format(dst->format)) < 0) {
+		NOUVEAU_ERR("Bad format = %d\n", dst->format);
+		return 1;
+	}
+
+	if ((gdirect_format = nv04_rect_format(dst->format)) < 0) {
+		NOUVEAU_ERR("Bad format = %d\n", dst->format);
+		return 1;
+	}
+
+	BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2);
+	OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	OUT_RELOCo(chan, nouveau_buffer(dst->buffer)->bo,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	BEGIN_RING(chan, surf2d, NV04_CONTEXT_SURFACES_2D_FORMAT, 4);
+	OUT_RING  (chan, cs2d_format);
+	OUT_RING  (chan, (dst->stride << 16) | dst->stride);
+	OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+	OUT_RELOCl(chan, nouveau_buffer(dst->buffer)->bo, dst->offset,
+		   NOUVEAU_BO_VRAM | NOUVEAU_BO_WR);
+
+	BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR_FORMAT, 1);
+	OUT_RING  (chan, gdirect_format);
+	BEGIN_RING(chan, rect, NV04_GDI_RECTANGLE_TEXT_COLOR1_A, 1);
+	OUT_RING  (chan, value);
+	BEGIN_RING(chan, rect,
+		   NV04_GDI_RECTANGLE_TEXT_UNCLIPPED_RECTANGLE_POINT(0), 2);
+	OUT_RING  (chan, (dx << 16) | dy);
+	OUT_RING  (chan, ( w << 16) |  h);
+
+	FIRE_RING(chan);
+	return 0;
+}
+
+int
+nouveau_surface_channel_create_nv04(struct nouveau_channel_context *nvc)
+{
+	struct nouveau_channel *chan = nvc->channel;
+	unsigned chipset = nvc->channel->device->chipset, class;
+	int ret;
+
+	if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, 0x39,
+				       &nvc->NvM2MF))) {
+		NOUVEAU_ERR("Error creating m2mf object: %d\n", ret);
+		return 1;
+	}
+	BIND_RING (chan, nvc->NvM2MF, nvc->next_subchannel++);
+	BEGIN_RING(chan, nvc->NvM2MF,
+		   NV04_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1);
+	OUT_RING  (chan, nvc->sync_notifier->handle);
+
+	class = chipset < 0x10 ? NV04_CONTEXT_SURFACES_2D :
+				 NV10_CONTEXT_SURFACES_2D;
+	if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class,
+				       &nvc->NvCtxSurf2D))) {
+		NOUVEAU_ERR("Error creating 2D surface object: %d\n", ret);
+		return 1;
+	}
+	BIND_RING (chan, nvc->NvCtxSurf2D, nvc->next_subchannel++);
+	BEGIN_RING(chan, nvc->NvCtxSurf2D,
+		   NV04_CONTEXT_SURFACES_2D_DMA_IMAGE_SOURCE, 2);
+	OUT_RING  (chan, nvc->channel->vram->handle);
+	OUT_RING  (chan, nvc->channel->vram->handle);
+
+	class = chipset < 0x10 ? NV04_IMAGE_BLIT : NV12_IMAGE_BLIT;
+	if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class,
+				       &nvc->NvImageBlit))) {
+		NOUVEAU_ERR("Error creating blit object: %d\n", ret);
+		return 1;
+	}
+	BIND_RING (chan, nvc->NvImageBlit, nvc->next_subchannel++);
+	BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_DMA_NOTIFY, 1);
+	OUT_RING  (chan, nvc->sync_notifier->handle);
+	BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_SURFACE, 1);
+	OUT_RING  (chan, nvc->NvCtxSurf2D->handle);
+	BEGIN_RING(chan, nvc->NvImageBlit, NV04_IMAGE_BLIT_OPERATION, 1);
+	OUT_RING  (chan, NV04_IMAGE_BLIT_OPERATION_SRCCOPY);
+
+	class = NV04_GDI_RECTANGLE_TEXT;
+	if ((ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class,
+				       &nvc->NvGdiRect))) {
+		NOUVEAU_ERR("Error creating rect object: %d\n", ret);
+		return 1;
+	}
+	BIND_RING (chan, nvc->NvGdiRect, nvc->next_subchannel++);
+	BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_DMA_NOTIFY, 1);
+	OUT_RING  (chan, nvc->sync_notifier->handle);
+	BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1);
+	OUT_RING  (chan, nvc->NvCtxSurf2D->handle);
+	BEGIN_RING(chan, nvc->NvGdiRect, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1);
+	OUT_RING  (chan, NV04_GDI_RECTANGLE_TEXT_OPERATION_SRCCOPY);
+	BEGIN_RING(chan, nvc->NvGdiRect,
+		   NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT, 1);
+	OUT_RING  (chan, NV04_GDI_RECTANGLE_TEXT_MONOCHROME_FORMAT_LE);
+
+	switch (chipset & 0xf0) {
+	case 0x00:
+	case 0x10:
+		class = NV04_SWIZZLED_SURFACE;
+		break;
+	case 0x20:
+		class = NV20_SWIZZLED_SURFACE;
+		break;
+	case 0x30:
+		class = NV30_SWIZZLED_SURFACE;
+		break;
+	case 0x40:
+	case 0x60:
+		class = NV40_SWIZZLED_SURFACE;
+		break;
+	default:
+		/* Famous last words: this really can't happen.. */
+		assert(0);
+		break;
+	}
+
+	ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class,
+				  &nvc->NvSwzSurf);
+	if (ret) {
+		NOUVEAU_ERR("Error creating swizzled surface: %d\n", ret);
+		return 1;
+	}
+
+	BIND_RING (chan, nvc->NvSwzSurf, nvc->next_subchannel++);
+
+	if (chipset < 0x10) {
+		class = NV04_SCALED_IMAGE_FROM_MEMORY;
+	} else
+	if (chipset < 0x40) {
+		class = NV10_SCALED_IMAGE_FROM_MEMORY;
+	} else {
+		class = NV40_SCALED_IMAGE_FROM_MEMORY;
+	}
+
+	ret = nouveau_grobj_alloc(chan, nvc->next_handle++, class,
+				  &nvc->NvSIFM);
+	if (ret) {
+		NOUVEAU_ERR("Error creating scaled image object: %d\n", ret);
+		return 1;
+	}
+
+	BIND_RING (chan, nvc->NvSIFM, nvc->next_subchannel++);
+
+	return 0;
+}
+
+int
+nouveau_surface_init_nv04(struct nouveau_context *nv)
+{
+	nv->surface_copy_prep = nv04_surface_copy_prep;
+	nv->surface_copy = nv04_surface_copy_blit;
+	nv->surface_copy_done = nv04_surface_copy_done;
+	nv->surface_fill = nv04_surface_fill;
+	return 0;
+}
+
diff --git a/src/gallium/winsys/drm/nouveau/nv50_surface.c b/src/gallium/winsys/drm/nouveau/nv50_surface.c
new file mode 100644
index 00000000000..c8ab7f690f3
--- /dev/null
+++ b/src/gallium/winsys/drm/nouveau/nv50_surface.c
@@ -0,0 +1,194 @@
+#include "pipe/p_context.h"
+#include "pipe/p_format.h"
+
+#include "nouveau_context.h"
+
+static INLINE int
+nv50_format(enum pipe_format format)
+{
+	switch (format) {
+	case PIPE_FORMAT_A8R8G8B8_UNORM:
+	case PIPE_FORMAT_Z24S8_UNORM:
+		return NV50_2D_DST_FORMAT_32BPP;
+	case PIPE_FORMAT_X8R8G8B8_UNORM:
+		return NV50_2D_DST_FORMAT_24BPP;
+	case PIPE_FORMAT_R5G6B5_UNORM:
+		return NV50_2D_DST_FORMAT_16BPP;
+	case PIPE_FORMAT_A8_UNORM:
+		return NV50_2D_DST_FORMAT_8BPP;
+	default:
+		return -1;
+	}
+}
+
+static int
+nv50_surface_set(struct nouveau_context *nv, struct pipe_surface *surf, int dst)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct nouveau_grobj *eng2d = nv->nvc->Nv2D;
+ 	struct nouveau_bo *bo = nouveau_buffer(surf->buffer)->bo;
+ 	int surf_format, mthd = dst ? NV50_2D_DST_FORMAT : NV50_2D_SRC_FORMAT;
+ 	int flags = NOUVEAU_BO_VRAM | (dst ? NOUVEAU_BO_WR : NOUVEAU_BO_RD);
+  
+ 	surf_format = nv50_format(surf->format);
+ 	if (surf_format < 0)
+ 		return 1;
+  
+ 	if (!nouveau_bo(bo)->tiled) {
+ 		BEGIN_RING(chan, eng2d, mthd, 2);
+ 		OUT_RING  (chan, surf_format);
+ 		OUT_RING  (chan, 1);
+ 		BEGIN_RING(chan, eng2d, mthd + 0x14, 5);
+ 		OUT_RING  (chan, surf->stride);
+ 		OUT_RING  (chan, surf->width);
+ 		OUT_RING  (chan, surf->height);
+ 		OUT_RELOCh(chan, bo, surf->offset, flags);
+ 		OUT_RELOCl(chan, bo, surf->offset, flags);
+ 	} else {
+ 		BEGIN_RING(chan, eng2d, mthd, 5);
+ 		OUT_RING  (chan, surf_format);
+ 		OUT_RING  (chan, 0);
+ 		OUT_RING  (chan, 0);
+ 		OUT_RING  (chan, 1);
+ 		OUT_RING  (chan, 0);
+ 		BEGIN_RING(chan, eng2d, mthd + 0x18, 4);
+ 		OUT_RING  (chan, surf->width);
+ 		OUT_RING  (chan, surf->height);
+ 		OUT_RELOCh(chan, bo, surf->offset, flags);
+ 		OUT_RELOCl(chan, bo, surf->offset, flags);
+ 	}
+ 
+#if 0
+ 	if (dst) {
+ 		BEGIN_RING(chan, eng2d, NV50_2D_CLIP_X, 4);
+ 		OUT_RING  (chan, 0);
+ 		OUT_RING  (chan, 0);
+ 		OUT_RING  (chan, surf->width);
+ 		OUT_RING  (chan, surf->height);
+ 	}
+#endif
+  
+ 	return 0;
+}
+
+static int
+nv50_surface_copy_prep(struct nouveau_context *nv,
+		       struct pipe_surface *dst, struct pipe_surface *src)
+{
+	int ret;
+
+	assert(src->format == dst->format);
+
+	ret = nv50_surface_set(nv, dst, 1);
+	if (ret)
+		return ret;
+
+	ret = nv50_surface_set(nv, src, 0);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void
+nv50_surface_copy(struct nouveau_context *nv, unsigned dx, unsigned dy,
+		  unsigned sx, unsigned sy, unsigned w, unsigned h)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct nouveau_grobj *eng2d = nv->nvc->Nv2D;
+
+	BEGIN_RING(chan, eng2d, 0x088c, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, eng2d, NV50_2D_BLIT_DST_X, 4);
+	OUT_RING  (chan, dx);
+	OUT_RING  (chan, dy);
+	OUT_RING  (chan, w);
+	OUT_RING  (chan, h);
+	BEGIN_RING(chan, eng2d, 0x08c0, 4);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 1);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, 1);
+	BEGIN_RING(chan, eng2d, 0x08d0, 4);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, sx);
+	OUT_RING  (chan, 0);
+	OUT_RING  (chan, sy);
+}
+
+static void
+nv50_surface_copy_done(struct nouveau_context *nv)
+{
+	FIRE_RING(nv->nvc->channel);
+}
+
+static int
+nv50_surface_fill(struct nouveau_context *nv, struct pipe_surface *dst,
+		  unsigned dx, unsigned dy, unsigned w, unsigned h,
+		  unsigned value)
+{
+	struct nouveau_channel *chan = nv->nvc->channel;
+	struct nouveau_grobj *eng2d = nv->nvc->Nv2D;
+	int rect_format, ret;
+
+	rect_format = nv50_format(dst->format);
+	if (rect_format < 0)
+		return 1;
+
+	ret = nv50_surface_set(nv, dst, 1);
+	if (ret)
+		return ret;
+
+	BEGIN_RING(chan, eng2d, 0x0580, 3);
+	OUT_RING  (chan, 4);
+	OUT_RING  (chan, rect_format);
+	OUT_RING  (chan, value);
+
+	BEGIN_RING(chan, eng2d, NV50_2D_RECT_X1, 4);
+	OUT_RING  (chan, dx);
+	OUT_RING  (chan, dy);
+	OUT_RING  (chan, dx + w);
+	OUT_RING  (chan, dy + h);
+
+	FIRE_RING(chan);
+	return 0;
+}
+
+int
+nouveau_surface_channel_create_nv50(struct nouveau_channel_context *nvc)
+{
+	struct nouveau_channel *chan = nvc->channel;
+	struct nouveau_grobj *eng2d = NULL;
+	int ret;
+
+	ret = nouveau_grobj_alloc(chan, nvc->next_handle++, NV50_2D, &eng2d);
+	if (ret)
+		return ret;
+	nvc->Nv2D = eng2d;
+
+	BIND_RING (chan, eng2d, nvc->next_subchannel++);
+	BEGIN_RING(chan, eng2d, NV50_2D_DMA_NOTIFY, 4);
+	OUT_RING  (chan, nvc->sync_notifier->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	OUT_RING  (chan, chan->vram->handle);
+	BEGIN_RING(chan, eng2d, NV50_2D_OPERATION, 1);
+	OUT_RING  (chan, NV50_2D_OPERATION_SRCCOPY);
+	BEGIN_RING(chan, eng2d, 0x0290, 1);
+	OUT_RING  (chan, 0);
+	BEGIN_RING(chan, eng2d, 0x0888, 1);
+	OUT_RING  (chan, 1);
+
+	return 0;
+}
+
+int
+nouveau_surface_init_nv50(struct nouveau_context *nv)
+{
+	nv->surface_copy_prep = nv50_surface_copy_prep;
+	nv->surface_copy = nv50_surface_copy;
+	nv->surface_copy_done = nv50_surface_copy_done;
+	nv->surface_fill = nv50_surface_fill;
+	return 0;
+}
+
diff --git a/src/gallium/winsys/egl_xlib/egl_xlib.c b/src/gallium/winsys/egl_xlib/egl_xlib.c
index e9f821d2764..477d766925c 100644
--- a/src/gallium/winsys/egl_xlib/egl_xlib.c
+++ b/src/gallium/winsys/egl_xlib/egl_xlib.c
@@ -537,7 +537,7 @@ xlib_eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
       }
       else {
          XFreeGC(surf->Dpy, surf->Gc);
-         st_unreference_framebuffer(&surf->Framebuffer);
+         st_unreference_framebuffer(surf->Framebuffer);
          free(surf);
       }
       return EGL_TRUE;
diff --git a/src/gallium/winsys/g3dvl/nouveau/Makefile b/src/gallium/winsys/g3dvl/nouveau/Makefile
new file mode 100644
index 00000000000..ff43327778a
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/Makefile
@@ -0,0 +1,50 @@
+TARGET		= libnouveau_dri.so
+GALLIUMDIR	= ../../..
+DRMDIR		?= /usr
+DRIDIR		= ../../../../driclient
+
+OBJECTS		= nouveau_bo.o nouveau_fence.o nouveau_swapbuffers.o nouveau_channel.o		\
+		  nouveau_grobj.o nouveau_context.o nouveau_winsys.o nouveau_lock.o		\
+		  nouveau_winsys_pipe.o nouveau_device.o nouveau_notifier.o nouveau_dma.o	\
+		  nouveau_pushbuf.o nouveau_resource.o nouveau_screen.o nv04_surface.o		\
+		  nv50_surface.o #nouveau_winsys_softpipe.o
+
+CFLAGS		+= -g -Wall -fPIC			\
+		   -I${GALLIUMDIR}/include		\
+		   -I${GALLIUMDIR}/winsys/g3dvl		\
+		   -I${DRMDIR}/include			\
+		   -I${DRMDIR}/include/drm		\
+		   -I${GALLIUMDIR}/drivers		\
+		   -I${GALLIUMDIR}/auxiliary		\
+		   -I${DRIDIR}/include
+
+LDFLAGS		+= -L${DRMDIR}/lib			\
+		   -L${DRIDIR}/lib			\
+		   -L${GALLIUMDIR}/auxiliary/draw	\
+		   -L${GALLIUMDIR}/auxiliary/tgsi	\
+		   -L${GALLIUMDIR}/auxiliary/translate	\
+		   -L${GALLIUMDIR}/auxiliary/rtasm	\
+		   -L${GALLIUMDIR}/auxiliary/cso_cache	\
+		   -L${GALLIUMDIR}/drivers/nv10		\
+		   -L${GALLIUMDIR}/drivers/nv20		\
+		   -L${GALLIUMDIR}/drivers/nv30		\
+		   -L${GALLIUMDIR}/drivers/nv40		\
+		   -L${GALLIUMDIR}/drivers/nv50
+
+LIBS		+= -ldriclient -ldrm -lnv10 -lnv20 -lnv30 -lnv40 -lnv50 -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lm
+
+#############################################
+
+.PHONY	= all clean libdriclient
+
+all: ${TARGET}
+
+${TARGET}: ${OBJECTS} libdriclient
+	$(CC) ${LDFLAGS} -shared -o $@ ${OBJECTS} ${LIBS}
+
+libdriclient:
+	cd ${DRIDIR}/src; ${MAKE}
+
+clean:
+	cd ${DRIDIR}/src; ${MAKE} clean
+	rm -rf ${OBJECTS} ${TARGET}
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c
new file mode 120000
index 00000000000..1005282dd4d
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_bo.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_bo.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c
new file mode 120000
index 00000000000..5af82029508
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_channel.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_channel.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c
new file mode 100644
index 00000000000..06a61fcda3e
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.c
@@ -0,0 +1,370 @@
+#include "pipe/p_defines.h"
+#include "pipe/p_context.h"
+#include "pipe/p_screen.h"
+#include "util/u_memory.h"
+
+#include "nouveau_context.h"
+#include "nouveau_dri.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_winsys_pipe.h"
+
+/*
+#ifdef DEBUG
+static const struct dri_debug_control debug_control[] = {
+	{ "bo", DEBUG_BO },
+	{ NULL, 0 }
+};
+int __nouveau_debug = 0;
+#endif
+*/
+
+/*
+ * TODO: Re-examine dri_screen, dri_context, nouveau_screen, nouveau_context
+ * relationships, seems like there is a lot of room for simplification there.
+ */
+
+static void
+nouveau_channel_context_destroy(struct nouveau_channel_context *nvc)
+{
+	nouveau_grobj_free(&nvc->NvCtxSurf2D);
+	nouveau_grobj_free(&nvc->NvImageBlit);
+	nouveau_grobj_free(&nvc->NvGdiRect);
+	nouveau_grobj_free(&nvc->NvM2MF);
+	nouveau_grobj_free(&nvc->Nv2D);
+	nouveau_grobj_free(&nvc->NvSwzSurf);
+	nouveau_grobj_free(&nvc->NvSIFM);
+
+	nouveau_notifier_free(&nvc->sync_notifier);
+
+	nouveau_channel_free(&nvc->channel);
+
+	FREE(nvc);
+}
+
+static struct nouveau_channel_context *
+nouveau_channel_context_create(struct nouveau_device *dev)
+{
+	struct nouveau_channel_context *nvc;
+	int ret;
+
+	nvc = CALLOC_STRUCT(nouveau_channel_context);
+	if (!nvc)
+		return NULL;
+
+	if ((ret = nouveau_channel_alloc(dev, 0x8003d001, 0x8003d002,
+					 &nvc->channel))) {
+		NOUVEAU_ERR("Error creating GPU channel: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	nvc->next_handle = 0x80000000;
+
+	if ((ret = nouveau_notifier_alloc(nvc->channel, nvc->next_handle++, 1,
+					  &nvc->sync_notifier))) {
+		NOUVEAU_ERR("Error creating channel sync notifier: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	switch (dev->chipset & 0xf0) {
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		ret = nouveau_surface_channel_create_nv50(nvc);
+		break;
+	default:
+		ret = nouveau_surface_channel_create_nv04(nvc);
+		break;
+	}
+
+	if (ret) {
+		NOUVEAU_ERR("Error initialising surface objects: %d\n", ret);
+		nouveau_channel_context_destroy(nvc);
+		return NULL;
+	}
+
+	return nvc;
+}
+
+int
+nouveau_context_create(dri_context_t *dri_context)
+{
+	dri_screen_t			*dri_screen = dri_context->dri_screen;
+	struct nouveau_screen 		*nv_screen = dri_screen->private;
+	struct nouveau_context		*nv = CALLOC_STRUCT(nouveau_context);
+	struct pipe_context		*pipe = NULL;
+	struct nouveau_channel_context	*nvc = NULL;
+	struct nouveau_device		*dev = nv_screen->device;
+	int				i;
+
+	switch (dev->chipset & 0xf0) {
+	case 0x10:
+	case 0x20:
+		/* NV10 */
+	case 0x30:
+		/* NV30 */
+	case 0x40:
+	case 0x60:
+		/* NV40 */
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		/* G80 */
+		break;
+	default:
+		NOUVEAU_ERR("Unsupported chipset: NV%02x\n", dev->chipset);
+		return 1;
+	}
+
+	dri_context->private = (void*)nv;
+	nv->dri_context = dri_context;
+	nv->nv_screen  = nv_screen;
+
+	{
+		struct nouveau_device_priv *nvdev = nouveau_device(dev);
+
+		nvdev->ctx  = dri_context->drm_context;
+		nvdev->lock = (drmLock*)&dri_screen->sarea->lock;
+	}
+
+	/*
+	driParseConfigFiles(&nv->dri_option_cache, &nv_screen->option_cache,
+			    nv->dri_screen->myNum, "nouveau");
+#ifdef DEBUG
+	__nouveau_debug = driParseDebugString(getenv("NOUVEAU_DEBUG"),
+					      debug_control);
+#endif
+	*/
+
+	/*XXX: Hack up a fake region and buffer object for front buffer.
+	 *     This will go away with TTM, replaced with a simple reference
+	 *     of the front buffer handle passed to us by the DDX.
+	 */
+	{
+		struct pipe_surface *fb_surf;
+		struct nouveau_pipe_buffer *fb_buf;
+		struct nouveau_bo_priv *fb_bo;
+
+		fb_bo = calloc(1, sizeof(struct nouveau_bo_priv));
+		fb_bo->drm.offset = nv_screen->front_offset;
+		fb_bo->drm.flags = NOUVEAU_MEM_FB;
+		fb_bo->drm.size = nv_screen->front_pitch *
+				  nv_screen->front_height;
+		fb_bo->refcount = 1;
+		fb_bo->base.flags = NOUVEAU_BO_PIN | NOUVEAU_BO_VRAM;
+		fb_bo->base.offset = fb_bo->drm.offset;
+		fb_bo->base.handle = (unsigned long)fb_bo;
+		fb_bo->base.size = fb_bo->drm.size;
+		fb_bo->base.device = nv_screen->device;
+
+		fb_buf = calloc(1, sizeof(struct nouveau_pipe_buffer));
+		fb_buf->bo = &fb_bo->base;
+
+		fb_surf = calloc(1, sizeof(struct pipe_surface));
+		if (nv_screen->front_cpp == 2)
+			fb_surf->format = PIPE_FORMAT_R5G6B5_UNORM;
+		else
+			fb_surf->format = PIPE_FORMAT_A8R8G8B8_UNORM;
+		pf_get_block(fb_surf->format, &fb_surf->block);
+		fb_surf->width = nv_screen->front_pitch / nv_screen->front_cpp;
+		fb_surf->height = nv_screen->front_height;
+		fb_surf->stride = fb_surf->width * fb_surf->block.size;
+		fb_surf->refcount = 1;
+		fb_surf->buffer = &fb_buf->base;
+
+		nv->frontbuffer = fb_surf;
+	}
+
+	nvc = nv_screen->nvc;
+
+	if (!nvc) {
+		nvc = nouveau_channel_context_create(dev);
+		if (!nvc) {
+			NOUVEAU_ERR("Failed initialising GPU context\n");
+			return 1;
+		}
+		nv_screen->nvc = nvc;
+	}
+
+	nvc->refcount++;
+	nv->nvc = nvc;
+
+	/* Find a free slot for a pipe context, allocate a new one if needed */
+	nv->pctx_id = -1;
+	for (i = 0; i < nvc->nr_pctx; i++) {
+		if (nvc->pctx[i] == NULL) {
+			nv->pctx_id = i;
+			break;
+		}
+	}
+
+	if (nv->pctx_id < 0) {
+		nv->pctx_id = nvc->nr_pctx++;
+		nvc->pctx =
+			realloc(nvc->pctx,
+				sizeof(struct pipe_context *) * nvc->nr_pctx);
+	}
+
+	/* Create pipe */
+	switch (dev->chipset & 0xf0) {
+	case 0x50:
+	case 0x80:
+	case 0x90:
+		if (nouveau_surface_init_nv50(nv))
+			return 1;
+		break;
+	default:
+		if (nouveau_surface_init_nv04(nv))
+			return 1;
+		break;
+	}
+
+	if (!getenv("NOUVEAU_FORCE_SOFTPIPE")) {
+		struct pipe_screen *pscreen;
+
+		pipe = nouveau_pipe_create(nv);
+		if (!pipe)
+			NOUVEAU_ERR("Couldn't create hw pipe\n");
+		pscreen = nvc->pscreen;
+
+		nv->cap.hw_vertex_buffer =
+			pscreen->get_param(pscreen, NOUVEAU_CAP_HW_VTXBUF);
+		nv->cap.hw_index_buffer =
+			pscreen->get_param(pscreen, NOUVEAU_CAP_HW_IDXBUF);
+	}
+
+	/* XXX: nouveau_winsys_softpipe needs a mesa header removed before we can compile it. */
+	/*
+	if (!pipe) {
+		NOUVEAU_MSG("Using softpipe\n");
+		pipe = nouveau_create_softpipe(nv);
+		if (!pipe) {
+			NOUVEAU_ERR("Error creating pipe, bailing\n");
+			return 1;
+		}
+	}
+	*/
+	if (!pipe) {
+		NOUVEAU_ERR("Error creating pipe, bailing\n");
+		return 1;
+	}
+
+	pipe->priv = nv;
+
+	return 0;
+}
+
+void
+nouveau_context_destroy(dri_context_t *dri_context)
+{
+	struct nouveau_context *nv = dri_context->private;
+	struct nouveau_channel_context *nvc = nv->nvc;
+
+	assert(nv);
+
+	if (nv->pctx_id >= 0) {
+		nvc->pctx[nv->pctx_id] = NULL;
+		if (--nvc->refcount <= 0) {
+			nouveau_channel_context_destroy(nvc);
+			nv->nv_screen->nvc = NULL;
+		}
+	}
+
+	free(nv);
+}
+
+int
+nouveau_context_bind(struct nouveau_context *nv, dri_drawable_t *dri_drawable)
+{
+	assert(nv);
+	assert(dri_drawable);
+
+	if (nv->dri_drawable != dri_drawable)
+	{
+		nv->dri_drawable = dri_drawable;
+		dri_drawable->private = nv;
+	}
+
+	return 0;
+}
+
+int
+nouveau_context_unbind(struct nouveau_context *nv)
+{
+	assert(nv);
+
+	nv->dri_drawable = NULL;
+
+	return 0;
+}
+
+/* Show starts here */
+
+int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable)
+{
+	struct nouveau_context	*nv;
+	dri_drawable_t		*dri_drawable;
+
+	nv = pipe->priv;
+
+	driCreateDrawable(nv->nv_screen->dri_screen, drawable, &dri_drawable);
+
+	nouveau_context_bind(nv, dri_drawable);
+
+	return 0;
+}
+
+int unbind_pipe_drawable(struct pipe_context *pipe)
+{
+	nouveau_context_unbind(pipe->priv);
+
+	return 0;
+}
+
+struct pipe_context* create_pipe_context(Display *display, int screen)
+{
+	dri_screen_t		*dri_screen;
+	dri_framebuffer_t	dri_framebuf;
+	dri_context_t		*dri_context;
+	struct nouveau_context	*nv;
+
+	driCreateScreen(display, screen, &dri_screen, &dri_framebuf);
+	driCreateContext(dri_screen, XDefaultVisual(display, screen), &dri_context);
+
+	nouveau_screen_create(dri_screen, &dri_framebuf);
+	nouveau_context_create(dri_context);
+
+	nv = dri_context->private;
+
+	return nv->nvc->pctx[nv->pctx_id];
+}
+
+int destroy_pipe_context(struct pipe_context *pipe)
+{
+	struct pipe_screen	*screen;
+	struct pipe_winsys	*winsys;
+	struct nouveau_context	*nv;
+	dri_screen_t		*dri_screen;
+	dri_context_t		*dri_context;
+
+	assert(pipe);
+
+	screen = pipe->screen;
+	winsys = pipe->winsys;
+	nv = pipe->priv;
+	dri_context = nv->dri_context;
+	dri_screen = dri_context->dri_screen;
+
+	pipe->destroy(pipe);
+	screen->destroy(screen);
+	free(winsys);
+
+	nouveau_context_destroy(dri_context);
+	nouveau_screen_destroy(dri_screen);
+	driDestroyContext(dri_context);
+	driDestroyScreen(dri_screen);
+
+	return 0;
+}
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h
new file mode 100644
index 00000000000..395a3ab7903
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_context.h
@@ -0,0 +1,105 @@
+#ifndef __NOUVEAU_CONTEXT_H__
+#define __NOUVEAU_CONTEXT_H__
+
+/*#include "xmlconfig.h"*/
+
+#include <driclient.h>
+#include "nouveau/nouveau_winsys.h"
+#include "nouveau_drmif.h"
+#include "nouveau_dma.h"
+
+struct nouveau_channel_context {
+	struct pipe_screen *pscreen;
+	int refcount;
+
+	unsigned cur_pctx;
+	unsigned nr_pctx;
+	struct pipe_context **pctx;
+
+	struct nouveau_channel  *channel;
+
+	struct nouveau_notifier *sync_notifier;
+
+	/* Common */
+	struct nouveau_grobj    *NvM2MF;
+	/* NV04-NV40 */
+	struct nouveau_grobj    *NvCtxSurf2D;
+	struct nouveau_grobj	*NvSwzSurf;
+	struct nouveau_grobj    *NvImageBlit;
+	struct nouveau_grobj    *NvGdiRect;
+	struct nouveau_grobj	*NvSIFM;
+	/* G80 */
+	struct nouveau_grobj    *Nv2D;
+
+	uint32_t                 next_handle;
+	uint32_t                 next_subchannel;
+	uint32_t                 next_sequence;
+};
+
+struct nouveau_context {
+	/* DRI stuff */
+	dri_context_t		*dri_context;
+	dri_drawable_t		*dri_drawable;
+	unsigned int		last_stamp;
+	/*driOptionCache	dri_option_cache;*/
+	drm_context_t		drm_context;
+	drmLock			drm_lock;
+	int			locked;
+	struct nouveau_screen	*nv_screen;
+	struct pipe_surface	*frontbuffer;
+
+	struct {
+		int hw_vertex_buffer;
+		int hw_index_buffer;
+	} cap;
+
+	/* Hardware context */
+	struct nouveau_channel_context	*nvc;
+	int				pctx_id;
+
+	/* pipe_surface accel */
+	struct pipe_surface		*surf_src, *surf_dst;
+	unsigned			surf_src_offset, surf_dst_offset;
+	
+	int  (*surface_copy_prep)(struct nouveau_context *,
+				  struct pipe_surface *dst,
+				  struct pipe_surface *src);
+	void (*surface_copy)(struct nouveau_context *, unsigned dx, unsigned dy,
+			     unsigned sx, unsigned sy, unsigned w, unsigned h);
+	void (*surface_copy_done)(struct nouveau_context *);
+	int (*surface_fill)(struct nouveau_context *, struct pipe_surface *,
+			    unsigned, unsigned, unsigned, unsigned, unsigned);
+};
+
+extern int nouveau_context_create(dri_context_t *);
+extern void nouveau_context_destroy(dri_context_t *);
+extern int nouveau_context_bind(struct nouveau_context *, dri_drawable_t *);
+extern int nouveau_context_unbind(struct nouveau_context *);
+
+#ifdef DEBUG
+extern int __nouveau_debug;
+
+#define DEBUG_BO (1 << 0)
+
+#define DBG(flag, ...) do {                   \
+	if (__nouveau_debug & (DEBUG_##flag)) \
+		NOUVEAU_ERR(__VA_ARGS__);     \
+} while(0)
+#else
+#define DBG(flag, ...)
+#endif
+
+extern void LOCK_HARDWARE(struct nouveau_context *);
+extern void UNLOCK_HARDWARE(struct nouveau_context *);
+
+extern int
+nouveau_surface_channel_create_nv04(struct nouveau_channel_context *);
+extern int
+nouveau_surface_channel_create_nv50(struct nouveau_channel_context *);
+extern int nouveau_surface_init_nv04(struct nouveau_context *);
+extern int nouveau_surface_init_nv50(struct nouveau_context *);
+
+extern uint32_t *nouveau_pipe_dma_beginp(struct nouveau_grobj *, int, int);
+extern void nouveau_pipe_dma_kickoff(struct nouveau_channel *);
+
+#endif
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c
new file mode 120000
index 00000000000..63f1fa040c7
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_device.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_device.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c
new file mode 120000
index 00000000000..cd0d32e5373
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_dma.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h
new file mode 120000
index 00000000000..e6c7d4bc965
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dma.h
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_dma.h
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h
new file mode 120000
index 00000000000..c8f9dbdc3a8
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_dri.h
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_dri.h
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h
new file mode 120000
index 00000000000..27082c9d34c
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_drmif.h
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_drmif.h
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c
new file mode 120000
index 00000000000..51a50527e6d
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_fence.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_fence.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c
new file mode 120000
index 00000000000..db17c72e6d7
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_grobj.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_grobj.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h
new file mode 120000
index 00000000000..4e9d3042cb5
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_local.h
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_local.h
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c
new file mode 100644
index 00000000000..375634bd059
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_lock.c
@@ -0,0 +1,92 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <pthread.h>
+#include <driclient.h>
+#include "nouveau_context.h"
+#include "nouveau_screen.h"
+
+static pthread_mutex_t lockMutex = PTHREAD_MUTEX_INITIALIZER;
+
+static void
+nouveau_contended_lock(struct nouveau_context *nv, unsigned int flags)
+{
+	dri_drawable_t			*dri_drawable = nv->dri_drawable;
+	dri_screen_t			*dri_screen = nv->dri_context->dri_screen;
+	struct nouveau_screen		*nv_screen = nv->nv_screen;
+	struct nouveau_device		*dev = nv_screen->device;
+	struct nouveau_device_priv	*nvdev = nouveau_device(dev);
+
+	drmGetLock(nvdev->fd, nvdev->ctx, flags);
+
+	/* If the window moved, may need to set a new cliprect now.
+	 *
+	 * NOTE: This releases and regains the hw lock, so all state
+	 * checking must be done *after* this call:
+	 */
+	if (dri_drawable)
+		DRI_VALIDATE_DRAWABLE_INFO(dri_screen, dri_drawable);
+}
+
+/* Lock the hardware and validate our state.
+ */
+void
+LOCK_HARDWARE(struct nouveau_context *nv)
+{
+	struct nouveau_screen		*nv_screen = nv->nv_screen;
+	struct nouveau_device		*dev = nv_screen->device;
+	struct nouveau_device_priv	*nvdev = nouveau_device(dev);
+	char				__ret=0;
+
+	pthread_mutex_lock(&lockMutex);
+	assert(!nv->locked);
+	
+	DRM_CAS(nvdev->lock, nvdev->ctx,
+		(DRM_LOCK_HELD | nvdev->ctx), __ret);
+	
+	if (__ret)
+		nouveau_contended_lock(nv, 0);
+	nv->locked = 1;
+}
+
+
+/* Unlock the hardware using the global current context 
+ */
+void
+UNLOCK_HARDWARE(struct nouveau_context *nv)
+{
+	struct nouveau_screen		*nv_screen = nv->nv_screen;
+	struct nouveau_device		*dev = nv_screen->device;
+	struct nouveau_device_priv	*nvdev = nouveau_device(dev);
+
+	assert(nv->locked);
+	nv->locked = 0;
+
+	DRM_UNLOCK(nvdev->fd, nvdev->lock, nvdev->ctx);
+
+	pthread_mutex_unlock(&lockMutex);
+} 
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c
new file mode 120000
index 00000000000..703bc3ce4ab
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_notifier.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_notifier.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c
new file mode 120000
index 00000000000..4ac137cadad
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_pushbuf.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_pushbuf.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c
new file mode 120000
index 00000000000..2241af328ff
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_resource.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_resource.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c
new file mode 100644
index 00000000000..f80d00050c8
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.c
@@ -0,0 +1,91 @@
+#include "pipe/p_context.h"
+#include "util/u_memory.h"
+#include "nouveau_context.h"
+#include <nouveau_drm.h>
+#include "nouveau_dri.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+
+#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 11
+#error nouveau_drm.h version does not match expected version
+#endif
+
+/*
+PUBLIC const char __driConfigOptions[] =
+DRI_CONF_BEGIN
+DRI_CONF_END;
+static const GLuint __driNConfigOptions = 0;
+*/
+
+int nouveau_check_dri_drm_ddx(dri_version_t *dri, dri_version_t *drm, dri_version_t *ddx)
+{
+	static const dri_version_t ddx_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL};
+	static const dri_version_t dri_expected = {4, 0, 0};
+	static const dri_version_t drm_expected = {0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL};
+
+	assert(dri);
+	assert(drm);
+	assert(ddx);
+
+	if (dri->major != dri_expected.major || dri->minor < dri_expected.minor)
+	{
+		NOUVEAU_ERR("Unexpected DRI version.\n");
+		return 1;
+	}
+	if (drm->major != drm_expected.major || drm->minor < drm_expected.minor)
+	{
+		NOUVEAU_ERR("Unexpected DRM version.\n");
+		return 1;
+	}
+	if (ddx->major != ddx_expected.major || ddx->minor < ddx_expected.minor)
+	{
+		NOUVEAU_ERR("Unexpected DDX version.\n");
+		return 1;
+	}
+
+	return 0;
+}
+
+int
+nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf)
+{
+	struct nouveau_dri	*nv_dri = dri_framebuf->private;
+	struct nouveau_screen	*nv_screen;
+	int			ret;
+
+	if (nouveau_check_dri_drm_ddx(&dri_screen->dri, &dri_screen->drm, &dri_screen->ddx))
+		return 1;
+
+	nv_screen = CALLOC_STRUCT(nouveau_screen);
+	if (!nv_screen)
+		return 1;
+	nv_screen->dri_screen = dri_screen;
+	dri_screen->private = (void*)nv_screen;
+
+	/*
+	driParseOptionInfo(&nv_screen->option_cache,
+			   __driConfigOptions, __driNConfigOptions);
+	*/
+
+	if ((ret = nouveau_device_open_existing(&nv_screen->device, 0,
+						dri_screen->fd, 0))) {
+		NOUVEAU_ERR("Failed opening nouveau device: %d.\n", ret);
+		return 1;
+	}
+
+	nv_screen->front_offset = nv_dri->front_offset;
+	nv_screen->front_pitch  = nv_dri->front_pitch * (nv_dri->bpp / 8);
+	nv_screen->front_cpp = nv_dri->bpp / 8;
+	nv_screen->front_height = nv_dri->height;
+
+	return 0;
+}
+
+void
+nouveau_screen_destroy(dri_screen_t *dri_screen)
+{
+	struct nouveau_screen *nv_screen = dri_screen->private;
+
+	FREE(nv_screen);
+}
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h
new file mode 100644
index 00000000000..8a58bb75564
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_screen.h
@@ -0,0 +1,24 @@
+#ifndef __NOUVEAU_SCREEN_H__
+#define __NOUVEAU_SCREEN_H__
+
+/* TODO: Investigate using DRI options for interesting things */
+/*#include "xmlconfig.h"*/
+
+struct nouveau_screen {
+	dri_screen_t			*dri_screen;
+	struct nouveau_device		*device;
+	struct nouveau_channel_context	*nvc;
+
+	uint32_t			front_offset;
+	uint32_t			front_pitch;
+	uint32_t			front_cpp;
+	uint32_t			front_height;
+	
+	/*driOptionCache		option_cache;*/
+};
+
+int nouveau_screen_create(dri_screen_t *dri_screen, dri_framebuffer_t *dri_framebuf);
+void nouveau_screen_destroy(dri_screen_t *dri_screen);
+
+#endif
+
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c
new file mode 100644
index 00000000000..7916c806151
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.c
@@ -0,0 +1,64 @@
+#include "pipe/p_context.h"
+#include "nouveau_context.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+
+void
+nouveau_copy_buffer(dri_drawable_t *dri_drawable, struct pipe_surface *surf,
+		    const drm_clip_rect_t *rect)
+{
+	struct nouveau_context	*nv = dri_drawable->private;
+	drm_clip_rect_t		*pbox;
+	int			nbox, i;
+
+	LOCK_HARDWARE(nv);
+	if (!dri_drawable->num_cliprects) {
+		UNLOCK_HARDWARE(nv);
+		return;
+	}
+	pbox = dri_drawable->cliprects;
+	nbox = dri_drawable->num_cliprects;
+
+	nv->surface_copy_prep(nv, nv->frontbuffer, surf);
+	for (i = 0; i < nbox; i++, pbox++) {
+		int sx, sy, dx, dy, w, h;
+
+		sx = pbox->x1 - dri_drawable->x;
+		sy = pbox->y1 - dri_drawable->y;
+		dx = pbox->x1;
+		dy = pbox->y1;
+		w  = pbox->x2 - pbox->x1;
+		h  = pbox->y2 - pbox->y1;
+
+		nv->surface_copy(nv, dx, dy, sx, sy, w, h);
+	}
+
+	FIRE_RING(nv->nvc->channel);
+	UNLOCK_HARDWARE(nv);
+
+	//if (nv->last_stamp != dri_drawable->last_sarea_stamp)
+		//nv->last_stamp = dri_drawable->last_sarea_stamp;
+}
+
+void
+nouveau_copy_sub_buffer(dri_drawable_t *dri_drawable, struct pipe_surface *surf, int x, int y, int w, int h)
+{
+	if (surf) {
+		drm_clip_rect_t rect;
+		rect.x1 = x;
+		rect.y1 = y;
+		rect.x2 = x + w;
+		rect.y2 = y + h;
+
+		nouveau_copy_buffer(dri_drawable, surf, &rect);
+	}
+}
+
+void
+nouveau_swap_buffers(dri_drawable_t *dri_drawable, struct pipe_surface *surf)
+{
+	if (surf)
+		nouveau_copy_buffer(dri_drawable, surf, NULL);
+}
+
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h
new file mode 100644
index 00000000000..35e934adba8
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_swapbuffers.h
@@ -0,0 +1,10 @@
+#ifndef __NOUVEAU_SWAPBUFFERS_H__
+#define __NOUVEAU_SWAPBUFFERS_H__
+
+extern void nouveau_copy_buffer(dri_drawable_t *, struct pipe_surface *,
+				const drm_clip_rect_t *);
+extern void nouveau_copy_sub_buffer(dri_drawable_t *, struct pipe_surface *,
+				    int x, int y, int w, int h);
+extern void nouveau_swap_buffers(dri_drawable_t *, struct pipe_surface *);
+
+#endif
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c
new file mode 120000
index 00000000000..ce4052d5572
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_winsys.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c
new file mode 100644
index 00000000000..4f6ac9cad0b
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.c
@@ -0,0 +1,260 @@
+#include "pipe/p_winsys.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+#include "util/u_memory.h"
+
+#include "nouveau_context.h"
+#include "nouveau_local.h"
+#include "nouveau_screen.h"
+#include "nouveau_swapbuffers.h"
+#include "nouveau_winsys_pipe.h"
+
+static void
+nouveau_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surf,
+			  void *context_private)
+{
+	struct nouveau_context *nv = context_private;
+	dri_drawable_t *dri_drawable = nv->dri_drawable;
+
+	nouveau_copy_buffer(dri_drawable, surf, NULL);
+}
+
+static const char *
+nouveau_get_name(struct pipe_winsys *pws)
+{
+	return "Nouveau/DRI";
+}
+
+static struct pipe_surface *
+nouveau_surface_alloc(struct pipe_winsys *ws)
+{
+	struct pipe_surface *surf;
+
+	surf = CALLOC_STRUCT(pipe_surface);
+	if (!surf)
+		return NULL;
+
+	surf->refcount = 1;
+	surf->winsys = ws;
+	return surf;
+}
+
+/* Borrowed from Mesa's xm_winsys */
+static unsigned int
+round_up(unsigned n, unsigned multiple)
+{
+   return (n + multiple - 1) & ~(multiple - 1);
+}
+
+static int
+nouveau_surface_alloc_storage
+(
+	struct pipe_winsys *pws,
+	struct pipe_surface *surface,
+	unsigned width,
+	unsigned height,
+	enum pipe_format format,
+	unsigned flags,
+	unsigned tex_usage
+)
+{
+	const unsigned int ALIGNMENT = 256;
+
+	assert(pws);
+	assert(surface);
+
+	surface->width = width;
+	surface->height = height;
+	surface->format = format;
+	pf_get_block(format, &surface->block);
+	surface->nblocksx = pf_get_nblocksx(&surface->block, width);
+	surface->nblocksy = pf_get_nblocksy(&surface->block, height);
+	surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT);
+	surface->usage = flags;
+	surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy);
+
+	return 0;
+}
+
+static void
+nouveau_surface_release(struct pipe_winsys *ws, struct pipe_surface **s)
+{
+	struct pipe_surface *surf = *s;
+
+	*s = NULL;
+	if (--surf->refcount <= 0) {
+		if (surf->buffer)
+			winsys_buffer_reference(ws, &surf->buffer, NULL);
+		free(surf);
+	}
+}
+
+static struct pipe_buffer *
+nouveau_pipe_bo_create(struct pipe_winsys *pws, unsigned alignment,
+		       unsigned usage, unsigned size)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws;
+	struct nouveau_context *nv = nvpws->nv;
+	struct nouveau_device *dev = nv->nv_screen->device;
+	struct nouveau_pipe_buffer *nvbuf;
+	uint32_t flags;
+
+	nvbuf = calloc(1, sizeof(*nvbuf));
+	if (!nvbuf)
+		return NULL;
+	nvbuf->base.refcount = 1;
+	nvbuf->base.alignment = alignment;
+	nvbuf->base.usage = usage;
+	nvbuf->base.size = size;
+
+	flags = NOUVEAU_BO_LOCAL;
+
+	if (usage & PIPE_BUFFER_USAGE_PIXEL) {
+		if (usage & NOUVEAU_BUFFER_USAGE_TEXTURE)
+			flags |= NOUVEAU_BO_GART;
+		flags |= NOUVEAU_BO_VRAM;
+	}
+
+	if (usage & PIPE_BUFFER_USAGE_VERTEX) {
+		if (nv->cap.hw_vertex_buffer)
+			flags |= NOUVEAU_BO_GART;
+	}
+
+	if (usage & PIPE_BUFFER_USAGE_INDEX) {
+		if (nv->cap.hw_index_buffer)
+			flags |= NOUVEAU_BO_GART;
+	}
+
+	if (nouveau_bo_new(dev, flags, alignment, size, &nvbuf->bo)) {
+		free(nvbuf);
+		return NULL;
+	}
+
+	return &nvbuf->base;
+}
+
+static struct pipe_buffer *
+nouveau_pipe_bo_user_create(struct pipe_winsys *pws, void *ptr, unsigned bytes)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)pws;
+	struct nouveau_device *dev = nvpws->nv->nv_screen->device;
+	struct nouveau_pipe_buffer *nvbuf;
+
+	nvbuf = calloc(1, sizeof(*nvbuf));
+	if (!nvbuf)
+		return NULL;
+	nvbuf->base.refcount = 1;
+	nvbuf->base.size = bytes;
+
+	if (nouveau_bo_user(dev, ptr, bytes, &nvbuf->bo)) {
+		free(nvbuf);
+		return NULL;
+	}
+
+	return &nvbuf->base;
+}
+
+static void
+nouveau_pipe_bo_del(struct pipe_winsys *ws, struct pipe_buffer *buf)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+
+	nouveau_bo_del(&nvbuf->bo);
+	free(nvbuf);
+}
+
+static void *
+nouveau_pipe_bo_map(struct pipe_winsys *pws, struct pipe_buffer *buf,
+		    unsigned flags)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+	uint32_t map_flags = 0;
+
+	if (flags & PIPE_BUFFER_USAGE_CPU_READ)
+		map_flags |= NOUVEAU_BO_RD;
+	if (flags & PIPE_BUFFER_USAGE_CPU_WRITE)
+		map_flags |= NOUVEAU_BO_WR;
+
+	if (nouveau_bo_map(nvbuf->bo, map_flags))
+		return NULL;
+	return nvbuf->bo->map;
+}
+
+static void
+nouveau_pipe_bo_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf)
+{
+	struct nouveau_pipe_buffer *nvbuf = nouveau_buffer(buf);
+
+	nouveau_bo_unmap(nvbuf->bo);
+}
+
+static INLINE struct nouveau_fence *
+nouveau_pipe_fence(struct pipe_fence_handle *pfence)
+{
+	return (struct nouveau_fence *)pfence;
+}
+
+static void
+nouveau_pipe_fence_reference(struct pipe_winsys *ws,
+			     struct pipe_fence_handle **ptr,
+			     struct pipe_fence_handle *pfence)
+{
+	nouveau_fence_ref((void *)pfence, (void *)ptr);
+}
+
+static int
+nouveau_pipe_fence_signalled(struct pipe_winsys *ws,
+			     struct pipe_fence_handle *pfence, unsigned flag)
+{
+	struct nouveau_pipe_winsys *nvpws = (struct nouveau_pipe_winsys *)ws;
+	struct nouveau_fence *fence = nouveau_pipe_fence(pfence);
+
+	if (nouveau_fence(fence)->signalled == 0)
+		nouveau_fence_flush(nvpws->nv->nvc->channel);
+
+	return !nouveau_fence(fence)->signalled;
+}
+
+static int
+nouveau_pipe_fence_finish(struct pipe_winsys *ws,
+			  struct pipe_fence_handle *pfence, unsigned flag)
+{
+	struct nouveau_fence *fence = nouveau_pipe_fence(pfence);
+	struct nouveau_fence *ref = NULL;
+
+	nouveau_fence_ref(fence, &ref);
+	return nouveau_fence_wait(&ref);
+}
+
+struct pipe_winsys *
+nouveau_create_pipe_winsys(struct nouveau_context *nv)
+{
+	struct nouveau_pipe_winsys *nvpws;
+	struct pipe_winsys *pws;
+
+	nvpws = CALLOC_STRUCT(nouveau_pipe_winsys);
+	if (!nvpws)
+		return NULL;
+	nvpws->nv = nv;
+	pws = &nvpws->pws;
+
+	pws->flush_frontbuffer = nouveau_flush_frontbuffer;
+
+	pws->surface_alloc = nouveau_surface_alloc;
+	pws->surface_alloc_storage = nouveau_surface_alloc_storage;
+	pws->surface_release = nouveau_surface_release;
+
+	pws->buffer_create = nouveau_pipe_bo_create;
+	pws->buffer_destroy = nouveau_pipe_bo_del;
+	pws->user_buffer_create = nouveau_pipe_bo_user_create;
+	pws->buffer_map = nouveau_pipe_bo_map;
+	pws->buffer_unmap = nouveau_pipe_bo_unmap;
+
+	pws->fence_reference = nouveau_pipe_fence_reference;
+	pws->fence_signalled = nouveau_pipe_fence_signalled;
+	pws->fence_finish = nouveau_pipe_fence_finish;
+
+	pws->get_name = nouveau_get_name;
+
+	return &nvpws->pws;
+}
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h
new file mode 120000
index 00000000000..9d9460883e0
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_pipe.h
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_winsys_pipe.h
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c
new file mode 120000
index 00000000000..ec613ecbf65
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nouveau_winsys_softpipe.c
@@ -0,0 +1 @@
+../../drm/nouveau/nouveau_winsys_softpipe.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c
new file mode 120000
index 00000000000..4455d8f9243
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nv04_surface.c
@@ -0,0 +1 @@
+../../drm/nouveau/nv04_surface.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c
new file mode 120000
index 00000000000..19f102001e9
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/nouveau/nv50_surface.c
@@ -0,0 +1 @@
+../../drm/nouveau/nv50_surface.c
\ No newline at end of file
diff --git a/src/gallium/winsys/g3dvl/vl_winsys.h b/src/gallium/winsys/g3dvl/vl_winsys.h
new file mode 100644
index 00000000000..c83db28dd98
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/vl_winsys.h
@@ -0,0 +1,14 @@
+#ifndef vl_winsys_h
+#define vl_winsys_h
+
+#include <X11/Xlib.h>
+
+struct pipe_context;
+
+struct pipe_context* create_pipe_context(Display *display, int screen);
+int destroy_pipe_context(struct pipe_context *pipe);
+int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable);
+int unbind_pipe_drawable(struct pipe_context *pipe);
+
+#endif
+
diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c
new file mode 100644
index 00000000000..68be2c2ea3a
--- /dev/null
+++ b/src/gallium/winsys/g3dvl/xsp_winsys.c
@@ -0,0 +1,336 @@
+#include "vl_winsys.h"
+#include <X11/Xutil.h>
+#include <pipe/p_winsys.h>
+#include <pipe/p_state.h>
+#include <pipe/p_inlines.h>
+#include <util/u_memory.h>
+#include <softpipe/sp_winsys.h>
+
+/* pipe_winsys implementation */
+
+struct xsp_pipe_winsys
+{
+	struct pipe_winsys	base;
+	XImage			fbimage;
+};
+
+struct xsp_context
+{
+	Display			*display;
+	int			screen;
+	Drawable		drawable;
+	int			drawable_bound;
+};
+
+struct xsp_buffer
+{
+	struct pipe_buffer	base;
+	boolean			is_user_buffer;
+	void			*data;
+	void			*mapped_data;
+};
+
+static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size)
+{
+	struct xsp_buffer *buffer;
+
+	assert(pws);
+
+	buffer = calloc(1, sizeof(struct xsp_buffer));
+	buffer->base.refcount = 1;
+	buffer->base.alignment = alignment;
+	buffer->base.usage = usage;
+	buffer->base.size = size;
+	buffer->data = align_malloc(size, alignment);
+
+	return (struct pipe_buffer*)buffer;
+}
+
+static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size)
+{
+	struct xsp_buffer *buffer;
+
+	assert(pws);
+
+	buffer = calloc(1, sizeof(struct xsp_buffer));
+	buffer->base.refcount = 1;
+	buffer->base.size = size;
+	buffer->is_user_buffer = TRUE;
+	buffer->data = data;
+
+	return (struct pipe_buffer*)buffer;
+}
+
+static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags)
+{
+	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+	assert(pws);
+	assert(buffer);
+
+	xsp_buf->mapped_data = xsp_buf->data;
+
+	return xsp_buf->mapped_data;
+}
+
+static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer)
+{
+	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+	assert(pws);
+	assert(buffer);
+
+	xsp_buf->mapped_data = NULL;
+}
+
+static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buffer)
+{
+	struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer;
+
+	assert(pws);
+	assert(buffer);
+
+	if (!xsp_buf->is_user_buffer)
+		align_free(xsp_buf->data);
+
+	free(xsp_buf);
+}
+
+static struct pipe_surface* xsp_surface_alloc(struct pipe_winsys *pws)
+{
+	struct pipe_surface *surface;
+
+	assert(pws);
+
+	surface = calloc(1, sizeof(struct pipe_surface));
+	surface->refcount = 1;
+	surface->winsys = pws;
+
+	return surface;
+}
+
+/* Borrowed from Mesa's xm_winsys */
+static unsigned int round_up(unsigned n, unsigned multiple)
+{
+   return (n + multiple - 1) & ~(multiple - 1);
+}
+
+static int xsp_surface_alloc_storage
+(
+	struct pipe_winsys *pws,
+	struct pipe_surface *surface,
+	unsigned width,
+	unsigned height,
+	enum pipe_format format,
+	unsigned flags,
+	unsigned tex_usage
+)
+{
+	const unsigned int ALIGNMENT = 1;
+
+	assert(pws);
+	assert(surface);
+
+	surface->width = width;
+	surface->height = height;
+	surface->format = format;
+	pf_get_block(format, &surface->block);
+	surface->nblocksx = pf_get_nblocksx(&surface->block, width);
+	surface->nblocksy = pf_get_nblocksy(&surface->block, height);
+	surface->stride = round_up(surface->nblocksx * surface->block.size, ALIGNMENT);
+	surface->usage = flags;
+	surface->buffer = pws->buffer_create(pws, ALIGNMENT, PIPE_BUFFER_USAGE_PIXEL, surface->stride * surface->nblocksy);
+
+	return 0;
+}
+
+static void xsp_surface_release(struct pipe_winsys *pws, struct pipe_surface **surface)
+{
+	struct pipe_surface *s;
+
+	assert(pws);
+	assert(surface);
+	assert(*surface);
+
+	s = *surface;
+
+	s->refcount--;
+
+	if (s->refcount == 0)
+	{
+		winsys_buffer_reference(pws, &s->buffer, NULL);
+		free(s);
+	}
+
+	*surface = NULL;
+}
+
+static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence)
+{
+	assert(pws);
+	assert(ptr);
+	assert(fence);
+}
+
+static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
+{
+	assert(pws);
+	assert(fence);
+
+	return 0;
+}
+
+static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag)
+{
+	assert(pws);
+	assert(fence);
+
+	return 0;
+}
+
+static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private)
+{
+	struct xsp_pipe_winsys	*xsp_winsys;
+	struct xsp_context	*xsp_context;
+
+	assert(pws);
+	assert(surface);
+	assert(context_private);
+
+	xsp_winsys = (struct xsp_pipe_winsys*)pws;
+	xsp_context = (struct xsp_context*)context_private;
+
+	if (!xsp_context->drawable_bound)
+		return;
+
+	xsp_winsys->fbimage.width = surface->width;
+	xsp_winsys->fbimage.height = surface->height;
+	xsp_winsys->fbimage.bytes_per_line = surface->width * (xsp_winsys->fbimage.bits_per_pixel >> 3);
+	xsp_winsys->fbimage.data = pipe_surface_map(surface, 0);
+
+	XPutImage
+	(
+		xsp_context->display,
+		xsp_context->drawable,
+		XDefaultGC(xsp_context->display, xsp_context->screen),
+		&xsp_winsys->fbimage,
+		0,
+		0,
+		0,
+		0,
+		surface->width,
+		surface->height
+	);
+	XFlush(xsp_context->display);
+	pipe_surface_unmap(surface);
+}
+
+static const char* xsp_get_name(struct pipe_winsys *pws)
+{
+	assert(pws);
+	return "X11 SoftPipe";
+}
+
+/* Show starts here */
+
+int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable)
+{
+	struct xsp_context *xsp_context;
+
+	assert(pipe);
+
+	xsp_context = pipe->priv;
+	xsp_context->drawable = drawable;
+	xsp_context->drawable_bound = 1;
+
+	return 0;
+}
+
+int unbind_pipe_drawable(struct pipe_context *pipe)
+{
+	struct xsp_context *xsp_context;
+
+	assert(pipe);
+
+	xsp_context = pipe->priv;
+	xsp_context->drawable_bound = 0;
+
+	return 0;
+}
+
+struct pipe_context* create_pipe_context(Display *display, int screen)
+{
+	struct xsp_pipe_winsys	*xsp_winsys;
+	struct xsp_context	*xsp_context;
+	struct pipe_screen	*sp_screen;
+	struct pipe_context	*sp_pipe;
+
+	assert(display);
+
+	xsp_winsys = calloc(1, sizeof(struct xsp_pipe_winsys));
+	xsp_winsys->base.buffer_create = xsp_buffer_create;
+	xsp_winsys->base.user_buffer_create = xsp_user_buffer_create;
+	xsp_winsys->base.buffer_map = xsp_buffer_map;
+	xsp_winsys->base.buffer_unmap = xsp_buffer_unmap;
+	xsp_winsys->base.buffer_destroy = xsp_buffer_destroy;
+	xsp_winsys->base.surface_alloc = xsp_surface_alloc;
+	xsp_winsys->base.surface_alloc_storage = xsp_surface_alloc_storage;
+	xsp_winsys->base.surface_release = xsp_surface_release;
+	xsp_winsys->base.fence_reference = xsp_fence_reference;
+	xsp_winsys->base.fence_signalled = xsp_fence_signalled;
+	xsp_winsys->base.fence_finish = xsp_fence_finish;
+	xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer;
+	xsp_winsys->base.get_name = xsp_get_name;
+
+	{
+		/* XXX: Can't use the returned XImage* directly,
+		since we don't have control over winsys destruction
+		and we wouldn't be able to free it */
+		XImage *template = XCreateImage
+		(
+			display,
+			XDefaultVisual(display, XDefaultScreen(display)),
+			XDefaultDepth(display, XDefaultScreen(display)),
+			ZPixmap,
+			0,
+			NULL,
+			0,	/* Don't know the width and height until flush_frontbuffer */
+			0,
+			32,
+			0
+		);
+
+		memcpy(&xsp_winsys->fbimage, template, sizeof(XImage));
+		XInitImage(&xsp_winsys->fbimage);
+
+		XDestroyImage(template);
+	}
+
+	sp_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys);
+	sp_pipe = softpipe_create(sp_screen, (struct pipe_winsys*)xsp_winsys, NULL);
+
+	xsp_context = calloc(1, sizeof(struct xsp_context));
+	xsp_context->display = display;
+	xsp_context->screen = screen;
+
+	sp_pipe->priv = xsp_context;
+
+	return sp_pipe;
+}
+
+int destroy_pipe_context(struct pipe_context *pipe)
+{
+	struct pipe_screen *screen;
+	struct pipe_winsys *winsys;
+
+	assert(pipe);
+
+	screen = pipe->screen;
+	winsys = pipe->winsys;
+	free(pipe->priv);
+	pipe->destroy(pipe);
+	screen->destroy(screen);
+	free(winsys);
+
+	return 0;
+}
diff --git a/src/gallium/winsys/xlib/SConscript b/src/gallium/winsys/xlib/SConscript
index 324fbef306a..3aef3b6cedd 100644
--- a/src/gallium/winsys/xlib/SConscript
+++ b/src/gallium/winsys/xlib/SConscript
@@ -5,8 +5,7 @@ Import('*')
 
 if env['platform'] == 'linux' \
         and 'mesa' in env['statetrackers'] \
-        and 'softpipe' in env['drivers'] \
-        and 'i965simple' in env['drivers'] \
+        and ('softpipe' or 'i915simple' or 'trace') in env['drivers'] \
         and not env['dri']:
 
     env = env.Clone()
@@ -22,15 +21,20 @@ if env['platform'] == 'linux' \
         'xfonts.c',
         'xm_api.c',
         'xm_winsys.c',
-        'xm_winsys_aub.c',
-        'brw_aub.c',
     ]
+
+    drivers = [];
+        
+    if 'softpipe' in env['drivers']:
+        drivers += [softpipe]
+
+    if 'i965simple' in env['drivers']:
+        drivers += [i965simple]
+        sources += [
+            'brw_aub.c',
+            'xm_winsys_aub.c',
+            ]
         
-    drivers = [
-        softpipe,
-        i965simple,
-    ]
-    
     if 'trace' in env['drivers']:
         env.Append(CPPDEFINES = 'GALLIUM_TRACE')
         drivers += [trace]
diff --git a/src/gallium/winsys/xlib/xm_winsys.c b/src/gallium/winsys/xlib/xm_winsys.c
index 3334af175bc..acb5ad8f714 100644
--- a/src/gallium/winsys/xlib/xm_winsys.c
+++ b/src/gallium/winsys/xlib/xm_winsys.c
@@ -352,7 +352,7 @@ xmesa_display_surface_tiled(XMesaBuffer b, const struct pipe_surface *surf)
          /* offset in pixels */
          offset *= TILE_SIZE * TILE_SIZE;
 
-         if (XSHM_ENABLED(xm_buf)) {
+         if (0 && XSHM_ENABLED(xm_buf)) {
             ximage->data = (char *) xm_buf->data + 4 * offset;
             /* make copy of tile data */
             memcpy(tmpTile, (uint *) ximage->data, sizeof(tmpTile));
@@ -365,7 +365,7 @@ xmesa_display_surface_tiled(XMesaBuffer b, const struct pipe_surface *surf)
 #endif
          }
          else {
-            /* twiddel from ximage buffer to temp tile */
+            /* twiddle from ximage buffer to temp tile */
             twiddle_tile((uint *) xm_buf->data + offset, tmpTile);
             /* display temp tile data */
             ximage->data = (char *) tmpTile;
diff --git a/src/glu/Makefile b/src/glu/Makefile
index b8c55db6d0a..e519dfeec49 100644
--- a/src/glu/Makefile
+++ b/src/glu/Makefile
@@ -10,14 +10,30 @@ SUBDIRS = $(GLU_DIRS)
 
 default: $(TOP)/configs/current
 	@for dir in $(SUBDIRS) ; do \
-		(cd $$dir ; $(MAKE)) ; \
+		(cd $$dir && $(MAKE)) || exit 1 ; \
 	done
 
-install:
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) $(TOP)/$(LIB_DIR)/libGLU.* $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
+# GLU pkg-config file
+pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
+	-e 's,@GLU_PC_REQ@,$(GLU_PC_REQ),' \
+	-e 's,@GLU_PC_REQ_PRIV@,$(GLU_PC_REQ_PRIV),' \
+	-e 's,@GLU_PC_LIB_PRIV@,$(GLU_PC_LIB_PRIV),' \
+	-e 's,@GLU_PC_CFLAGS@,$(GLU_PC_CFLAGS),'
+glu.pc: glu.pc.in
+	$(pcedit) $< > $@
+
+install: glu.pc
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) $(TOP)/$(LIB_DIR)/$(GLU_LIB_GLOB) $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 glu.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 clean:
-	@for dir in $(SUBDIRS) ; do \
-		(cd $$dir ; $(MAKE) clean) ; \
+	-@for dir in $(SUBDIRS) ; do \
+		(cd $$dir && $(MAKE) clean) ; \
 	done
+	-rm -f *.pc
diff --git a/src/glu/glu.pc.in b/src/glu/glu.pc.in
new file mode 100644
index 00000000000..bc2517e90ed
--- /dev/null
+++ b/src/glu/glu.pc.in
@@ -0,0 +1,13 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: glu
+Description: Mesa OpenGL Utility library
+Requires: @GLU_PC_REQ@
+Requires.private: @GLU_PC_REQ_PRIV@
+Version: @VERSION@
+Libs: -L${libdir} -lGLU
+Libs.private: @GLU_PC_LIB_PRIV@
+Cflags: -I${includedir} @GLU_PC_CFLAGS@
diff --git a/src/glu/mesa/Makefile b/src/glu/mesa/Makefile
index 47f95ef0240..c468ce62100 100644
--- a/src/glu/mesa/Makefile
+++ b/src/glu/mesa/Makefile
@@ -23,7 +23,7 @@ C_SOURCES = \
 
 OBJECTS = $(C_SOURCES:.c=.o)
 
-INCLUDES = -I. -I- -I$(TOP)/include
+INCLUDES = -I. -I$(TOP)/include
 
 ##### RULES #####
 
@@ -37,7 +37,7 @@ default:
 		echo "$(GLU_LIB_NAME) not build under BeOS, but integrated into ${GL_LIB_NAME}." ; \
 		exit 0 ; \
 	else \
-		$(MAKE) $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) ; \
+		$(MAKE) $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) || exit 1 ; \
 	fi
 
 $(TOP)/$(LIB_DIR):
@@ -45,7 +45,7 @@ $(TOP)/$(LIB_DIR):
 
 # Make the library:
 $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME): $(OBJECTS)
-	@ $(TOP)/bin/mklib -o $(GLU_LIB) -linker '$(CC)' \
+	@ $(MKLIB) -o $(GLU_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLU_MAJOR) -minor $(GLU_MINOR) -patch $(GLU_TINY) \
 		$(MKLIB_OPTIONS) -install $(TOP)/$(LIB_DIR) \
 		$(GLU_LIB_DEPS) $(OBJECTS)
diff --git a/src/glu/mesa/project.c b/src/glu/mesa/project.c
index 6fa03267e53..2e79cdf0840 100644
--- a/src/glu/mesa/project.c
+++ b/src/glu/mesa/project.c
@@ -346,7 +346,8 @@ gluUnProject(GLdouble winx, GLdouble winy, GLdouble winz,
 
    /* calcul transformation inverse */
    matmul(A, proj, model);
-   invert_matrix(A, m);
+   if (!invert_matrix(A, m))
+      return GL_FALSE;
 
    /* d'ou les coordonnees objets */
    transform_point(out, m, in);
@@ -386,7 +387,8 @@ gluUnProject4(GLdouble winx, GLdouble winy, GLdouble winz, GLdouble clipw,
 
    /* calcul transformation inverse */
    matmul(A, projMatrix, modelMatrix);
-   invert_matrix(A, m);
+   if (!invert_matrix(A, m))
+      return GL_FALSE;
 
    /* d'ou les coordonnees objets */
    transform_point(out, m, in);
diff --git a/src/glu/mini/project.c b/src/glu/mini/project.c
index 6fa03267e53..2e79cdf0840 100644
--- a/src/glu/mini/project.c
+++ b/src/glu/mini/project.c
@@ -346,7 +346,8 @@ gluUnProject(GLdouble winx, GLdouble winy, GLdouble winz,
 
    /* calcul transformation inverse */
    matmul(A, proj, model);
-   invert_matrix(A, m);
+   if (!invert_matrix(A, m))
+      return GL_FALSE;
 
    /* d'ou les coordonnees objets */
    transform_point(out, m, in);
@@ -386,7 +387,8 @@ gluUnProject4(GLdouble winx, GLdouble winy, GLdouble winz, GLdouble clipw,
 
    /* calcul transformation inverse */
    matmul(A, projMatrix, modelMatrix);
-   invert_matrix(A, m);
+   if (!invert_matrix(A, m))
+      return GL_FALSE;
 
    /* d'ou les coordonnees objets */
    transform_point(out, m, in);
diff --git a/src/glu/sgi/.gitignore b/src/glu/sgi/.gitignore
new file mode 100644
index 00000000000..279ea7d4345
--- /dev/null
+++ b/src/glu/sgi/.gitignore
@@ -0,0 +1 @@
+exptmp
diff --git a/src/glu/sgi/Makefile b/src/glu/sgi/Makefile
index cdd267a025c..20c3bed0c93 100644
--- a/src/glu/sgi/Makefile
+++ b/src/glu/sgi/Makefile
@@ -128,7 +128,7 @@ default:
 		echo "$(GLU_LIB_NAME) not build under BeOS, but integrated into ${GL_LIB_NAME}." ; \
 		exit 0 ; \
 	else \
-		$(MAKE) $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) ; \
+		$(MAKE) $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME) || exit 1 ; \
 	fi
 
 $(TOP)/$(LIB_DIR):
@@ -136,10 +136,10 @@ $(TOP)/$(LIB_DIR):
 
 # Make the library:
 $(TOP)/$(LIB_DIR)/$(GLU_LIB_NAME): $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLU_LIB) -linker '$(CXX)' \
+	$(MKLIB) -o $(GLU_LIB) -linker '$(CXX)' -ldflags '$(LDFLAGS)' \
 		-major $(GLU_MAJOR) -minor $(GLU_MINOR) -patch $(GLU_TINY) \
 		-cplusplus $(MKLIB_OPTIONS) -install $(TOP)/$(LIB_DIR) \
-		-exports glu.exports \
+		-exports glu.exports -id $(INSTALL_LIB_DIR)/lib$(GLU_LIB).$(GLU_MAJOR).dylib \
 		$(GLU_LIB_DEPS) $(OBJECTS)
 
 
diff --git a/src/glu/sgi/Makefile.mgw b/src/glu/sgi/Makefile.mgw
new file mode 100644
index 00000000000..d00d97a3b61
--- /dev/null
+++ b/src/glu/sgi/Makefile.mgw
@@ -0,0 +1,230 @@
+# Mesa 3-D graphics library
+# Version:  5.1
+# 
+# Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+# 
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+# 
+# The above copyright notice and this permission notice shall be included
+# in all copies or substantial portions of the Software.
+# 
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+# OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+# MinGW core makefile v1.4 for Mesa
+#
+#  Copyright (C) 2002 - Daniel Borca
+#  Email : dborca@users.sourceforge.net
+#  Web   : http://www.geocities.com/dborca
+
+# MinGW core-glu makefile updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+#
+#  Available options:
+#
+#    Environment variables:
+#	CFLAGS
+#
+#	GLIDE		path to Glide3 SDK; used with FX.
+#			default = $(TOP)/glide3
+#	FX=1		build for 3dfx Glide3. Note that this disables
+#			compilation of most WMesa code and requires fxMesa.
+#			As a consequence, you'll need the Win32 Glide3
+#			library to build any application.
+#			default = no
+#	ICD=1		build the installable client driver interface
+#			(windows opengl driver interface)
+#			default = no
+#	X86=1		optimize for x86 (if possible, use MMX, SSE, 3DNow).
+#			default = no
+#
+#    Targets:
+#	all:		build GL
+#	clean:		remove object files
+#
+
+
+
+.PHONY: all clean
+.INTERMEDIATE: x86/gen_matypes.exe
+.SUFFIXES: .rc .res
+
+# Set this to the prefix of your build tools, i.e. mingw32-
+TOOLS_PREFIX = mingw32-
+
+TOP = ../../..
+LIBDIR = $(TOP)/lib
+
+LIB_NAME = glu32
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+GLU_DLL = $(LIB_NAME)$(DLL_EXT)
+GLU_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+GLU_DEF = $(LIB_NAME).def
+
+LDLIBS = -L$(LIBDIR) -lopengl32
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GLU32 -D_DLL
+
+
+CC = $(TOOLS_PREFIX)gcc
+CFLAGS += -DNDEBUG -DLIBRARYBUILD -I$(TOP)/include -Iinclude
+CXX = $(TOOLS_PREFIX)g++
+CXXFLAGS = $(CFLAGS) -Ilibnurbs/internals -Ilibnurbs/interface -Ilibnurbs/nurbtess
+
+AR = ar
+ARFLAGS = crus
+
+UNLINK = del $(subst /,\,$(1))
+ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+ifneq ($(wildcard $(addsuffix /rm,$(subst :, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+
+C_SOURCES = \
+	libutil/error.c \
+	libutil/glue.c \
+	libutil/mipmap.c \
+	libutil/project.c \
+	libutil/quad.c \
+	libutil/registry.c \
+	libtess/dict.c \
+	libtess/geom.c \
+	libtess/memalloc.c \
+	libtess/mesh.c \
+	libtess/normal.c \
+	libtess/priorityq.c \
+	libtess/render.c \
+	libtess/sweep.c \
+	libtess/tess.c \
+	libtess/tessmono.c
+
+CC_SOURCES = \
+	libnurbs/interface/bezierEval.cc \
+	libnurbs/interface/bezierPatch.cc \
+	libnurbs/interface/bezierPatchMesh.cc \
+	libnurbs/interface/glcurveval.cc \
+	libnurbs/interface/glinterface.cc \
+	libnurbs/interface/glrenderer.cc \
+	libnurbs/interface/glsurfeval.cc \
+	libnurbs/interface/incurveeval.cc \
+	libnurbs/interface/insurfeval.cc \
+	libnurbs/internals/arc.cc \
+	libnurbs/internals/arcsorter.cc \
+	libnurbs/internals/arctess.cc \
+	libnurbs/internals/backend.cc \
+	libnurbs/internals/basiccrveval.cc \
+	libnurbs/internals/basicsurfeval.cc \
+	libnurbs/internals/bin.cc \
+	libnurbs/internals/bufpool.cc \
+	libnurbs/internals/cachingeval.cc \
+	libnurbs/internals/ccw.cc \
+	libnurbs/internals/coveandtiler.cc \
+	libnurbs/internals/curve.cc \
+	libnurbs/internals/curvelist.cc \
+	libnurbs/internals/curvesub.cc \
+	libnurbs/internals/dataTransform.cc \
+	libnurbs/internals/displaylist.cc \
+	libnurbs/internals/flist.cc \
+	libnurbs/internals/flistsorter.cc \
+	libnurbs/internals/hull.cc \
+	libnurbs/internals/intersect.cc \
+	libnurbs/internals/knotvector.cc \
+	libnurbs/internals/mapdesc.cc \
+	libnurbs/internals/mapdescv.cc \
+	libnurbs/internals/maplist.cc \
+	libnurbs/internals/mesher.cc \
+	libnurbs/internals/monoTriangulationBackend.cc \
+	libnurbs/internals/monotonizer.cc \
+	libnurbs/internals/mycode.cc \
+	libnurbs/internals/nurbsinterfac.cc \
+	libnurbs/internals/nurbstess.cc \
+	libnurbs/internals/patch.cc \
+	libnurbs/internals/patchlist.cc \
+	libnurbs/internals/quilt.cc \
+	libnurbs/internals/reader.cc \
+	libnurbs/internals/renderhints.cc \
+	libnurbs/internals/slicer.cc \
+	libnurbs/internals/sorter.cc \
+	libnurbs/internals/splitarcs.cc \
+	libnurbs/internals/subdivider.cc \
+	libnurbs/internals/tobezier.cc \
+	libnurbs/internals/trimline.cc \
+	libnurbs/internals/trimregion.cc \
+	libnurbs/internals/trimvertpool.cc \
+	libnurbs/internals/uarray.cc \
+	libnurbs/internals/varray.cc \
+	libnurbs/nurbtess/directedLine.cc \
+	libnurbs/nurbtess/gridWrap.cc \
+	libnurbs/nurbtess/monoChain.cc \
+	libnurbs/nurbtess/monoPolyPart.cc \
+	libnurbs/nurbtess/monoTriangulation.cc \
+	libnurbs/nurbtess/partitionX.cc \
+	libnurbs/nurbtess/partitionY.cc \
+	libnurbs/nurbtess/polyDBG.cc \
+	libnurbs/nurbtess/polyUtil.cc \
+	libnurbs/nurbtess/primitiveStream.cc \
+	libnurbs/nurbtess/quicksort.cc \
+	libnurbs/nurbtess/rectBlock.cc \
+	libnurbs/nurbtess/sampleComp.cc \
+	libnurbs/nurbtess/sampleCompBot.cc \
+	libnurbs/nurbtess/sampleCompRight.cc \
+	libnurbs/nurbtess/sampleCompTop.cc \
+	libnurbs/nurbtess/sampleMonoPoly.cc \
+	libnurbs/nurbtess/sampledLine.cc \
+	libnurbs/nurbtess/searchTree.cc
+
+SOURCES = $(C_SOURCES) $(CC_SOURCES)
+
+OBJECTS = $(addsuffix .o,$(basename $(SOURCES)))
+
+.c.o:
+	$(CC) -o $@ $(CFLAGS) -c $<
+.cc.o:
+	$(CXX) -o $@ $(CXXFLAGS) -c $<
+
+
+all: $(LIBDIR) $(LIBDIR)/$(GLU_DLL) $(LIBDIR)/$(GLU_IMP)
+
+$(LIBDIR):
+	mkdir -p $(LIBDIR)
+
+$(LIBDIR)/$(GLU_DLL) $(LIBDIR)/$(GLU_IMP): $(OBJECTS)
+	$(CXX) $(LDFLAGS) -o $(LIBDIR)/$(GLU_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GLU_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GLU_IMP) --def $(LIBDIR)/$(GLU_DEF)
+
+
+clean:
+	-$(call UNLINK,libutil/*.o)
+	-$(call UNLINK,libtess/*.o)
+	-$(call UNLINK,libnurbs/interface/*.o)
+	-$(call UNLINK,libnurbs/internals/*.o)
+	-$(call UNLINK,libnurbs/nurbtess/*.o)
diff --git a/src/glu/sgi/glu.exports.darwin b/src/glu/sgi/glu.exports.darwin
new file mode 100644
index 00000000000..62d20ae1fbc
--- /dev/null
+++ b/src/glu/sgi/glu.exports.darwin
@@ -0,0 +1,59 @@
+_gluBeginCurve
+_gluBeginPolygon
+_gluBeginSurface
+_gluBeginTrim
+_gluBuild1DMipmapLevels
+_gluBuild1DMipmaps
+_gluBuild2DMipmapLevels
+_gluBuild2DMipmaps
+_gluBuild3DMipmapLevels
+_gluBuild3DMipmaps
+_gluCheckExtension
+_gluCylinder
+_gluDeleteNurbsRenderer
+_gluDeleteQuadric
+_gluDeleteTess
+_gluDisk
+_gluEndCurve
+_gluEndPolygon
+_gluEndSurface
+_gluEndTrim
+_gluErrorString
+_gluGetNurbsProperty
+_gluGetString
+_gluGetTessProperty
+_gluLoadSamplingMatrices
+_gluLookAt
+_gluNewNurbsRenderer
+_gluNewQuadric
+_gluNewTess
+_gluNextContour
+_gluNurbsCallback
+_gluNurbsCallbackData
+_gluNurbsCallbackDataEXT
+_gluNurbsCurve
+_gluNurbsProperty
+_gluNurbsSurface
+_gluOrtho2D
+_gluPartialDisk
+_gluPerspective
+_gluPickMatrix
+_gluProject
+_gluPwlCurve
+_gluQuadricCallback
+_gluQuadricDrawStyle
+_gluQuadricNormals
+_gluQuadricOrientation
+_gluQuadricTexture
+_gluScaleImage
+_gluSphere
+_gluTessBeginContour
+_gluTessBeginPolygon
+_gluTessCallback
+_gluTessEndContour
+_gluTessEndPolygon
+_gluTessNormal
+_gluTessProperty
+_gluTessVertex
+_gluUnProject
+_gluUnProject4
diff --git a/src/glu/sgi/libnurbs/interface/bezierEval.h b/src/glu/sgi/libnurbs/interface/bezierEval.h
index adecfe9b2fb..93e8c69d12e 100644
--- a/src/glu/sgi/libnurbs/interface/bezierEval.h
+++ b/src/glu/sgi/libnurbs/interface/bezierEval.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/interface/bezierPatch.cc b/src/glu/sgi/libnurbs/interface/bezierPatch.cc
index fa1daed52e9..dbab3a1a2b7 100644
--- a/src/glu/sgi/libnurbs/interface/bezierPatch.cc
+++ b/src/glu/sgi/libnurbs/interface/bezierPatch.cc
@@ -109,8 +109,11 @@ void bezierPatchDelete(bezierPatch *b)
 void bezierPatchDeleteList(bezierPatch *b)
 {
   bezierPatch *temp;
-  for(temp = b; temp != NULL; temp = temp->next)
-    bezierPatchDelete(temp);
+  while (b != NULL) {
+    temp = b;
+    b = b->next;
+    bezierPatchDelete(temp);    
+  }
 }
 
 bezierPatch* bezierPatchInsert(bezierPatch *list, bezierPatch *b)
diff --git a/src/glu/sgi/libnurbs/interface/bezierPatch.h b/src/glu/sgi/libnurbs/interface/bezierPatch.h
index ad0f8b0d2a4..981183afaea 100644
--- a/src/glu/sgi/libnurbs/interface/bezierPatch.h
+++ b/src/glu/sgi/libnurbs/interface/bezierPatch.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/interface/bezierPatchMesh.cc b/src/glu/sgi/libnurbs/interface/bezierPatchMesh.cc
index 3dc16313fff..ac7ff84fc4a 100644
--- a/src/glu/sgi/libnurbs/interface/bezierPatchMesh.cc
+++ b/src/glu/sgi/libnurbs/interface/bezierPatchMesh.cc
@@ -103,23 +103,22 @@ bezierPatchMesh *bezierPatchMeshMake(int maptype, float umin, float umax, int us
   int dimension;
   int the_ustride;
   int the_vstride;
-  
-  bezierPatchMesh *ret = (bezierPatchMesh*) malloc(sizeof(bezierPatchMesh));
-  assert(ret);
 
-  ret->bpatch = NULL;
-  ret->bpatch_normal = NULL;
-  ret->bpatch_color  = NULL;
-  ret->bpatch_texcoord = NULL;
- 
   if(maptype == GL_MAP2_VERTEX_3) dimension = 3;
   else if (maptype==GL_MAP2_VERTEX_4) dimension = 4;
   else {
     fprintf(stderr, "error in inMap2f, maptype=%i is wrong, maptype,map is invalid\n", maptype);
     return NULL;
   }
-  
+
+  bezierPatchMesh *ret = (bezierPatchMesh*) malloc(sizeof(bezierPatchMesh));
+  assert(ret);
+
+  ret->bpatch_normal = NULL;
+  ret->bpatch_color  = NULL;
+  ret->bpatch_texcoord = NULL;
   ret->bpatch = bezierPatchMake(umin, vmin, umax, vmax, uorder, vorder, dimension);
+
   /*copy the control points there*/
   the_ustride = vorder * dimension;
   the_vstride = dimension;
diff --git a/src/glu/sgi/libnurbs/interface/bezierPatchMesh.h b/src/glu/sgi/libnurbs/interface/bezierPatchMesh.h
index 2ab24dff5b0..449329665cc 100644
--- a/src/glu/sgi/libnurbs/interface/bezierPatchMesh.h
+++ b/src/glu/sgi/libnurbs/interface/bezierPatchMesh.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/interface/glcurveval.h b/src/glu/sgi/libnurbs/interface/glcurveval.h
index a09a74d04c3..8fa493034b2 100644
--- a/src/glu/sgi/libnurbs/interface/glcurveval.h
+++ b/src/glu/sgi/libnurbs/interface/glcurveval.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * glcurveval.h
diff --git a/src/glu/sgi/libnurbs/interface/glimports.h b/src/glu/sgi/libnurbs/interface/glimports.h
index 2c307f63e87..6e69feb405e 100644
--- a/src/glu/sgi/libnurbs/interface/glimports.h
+++ b/src/glu/sgi/libnurbs/interface/glimports.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * glimports.h
diff --git a/src/glu/sgi/libnurbs/interface/glrenderer.h b/src/glu/sgi/libnurbs/interface/glrenderer.h
index 8fc23125e0f..3b72d44d1fd 100644
--- a/src/glu/sgi/libnurbs/interface/glrenderer.h
+++ b/src/glu/sgi/libnurbs/interface/glrenderer.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * glrenderer.h
diff --git a/src/glu/sgi/libnurbs/interface/glsurfeval.h b/src/glu/sgi/libnurbs/interface/glsurfeval.h
index b7a88069f5a..1567c6b098c 100644
--- a/src/glu/sgi/libnurbs/interface/glsurfeval.h
+++ b/src/glu/sgi/libnurbs/interface/glsurfeval.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * glsurfeval.h
diff --git a/src/glu/sgi/libnurbs/interface/insurfeval.cc b/src/glu/sgi/libnurbs/interface/insurfeval.cc
index 78d8bece13e..9d0c82a91c0 100644
--- a/src/glu/sgi/libnurbs/interface/insurfeval.cc
+++ b/src/glu/sgi/libnurbs/interface/insurfeval.cc
@@ -1531,8 +1531,8 @@ void OpenGLSurfaceEvaluator::inEvalVStrip(int n_left, REAL u_left, REAL* left_va
     }
   //clean up 
   free(leftXYZ);
-  free(leftXYZ);
-  free(rightNormal);
+  free(rightXYZ);
+  free(leftNormal);
   free(rightNormal);
 }
 
diff --git a/src/glu/sgi/libnurbs/interface/mystdio.h b/src/glu/sgi/libnurbs/interface/mystdio.h
index e9947ea3937..5ab49effd1a 100644
--- a/src/glu/sgi/libnurbs/interface/mystdio.h
+++ b/src/glu/sgi/libnurbs/interface/mystdio.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mystdio.h
@@ -41,7 +37,7 @@
 #define __glumystdio_h_
 
 #ifdef STANDALONE
-inline void _glu_dprintf( char *, ... ) { }
+inline void _glu_dprintf( const char *, ... ) { }
 #endif
 
 #ifdef LIBRARYBUILD
@@ -49,12 +45,12 @@ inline void _glu_dprintf( char *, ... ) { }
 #include <stdio.h>
 #define _glu_dprintf printf
 #else
-inline void _glu_dprintf( char *, ... ) { }
+inline void _glu_dprintf( const char *, ... ) { }
 #endif
 #endif
 
 #ifdef GLBUILD
-inline void _glu_dprintf( char *, ... ) { }
+inline void _glu_dprintf( const char *, ... ) { }
 #endif
 
 #ifndef NULL
diff --git a/src/glu/sgi/libnurbs/interface/mystdlib.h b/src/glu/sgi/libnurbs/interface/mystdlib.h
index 2520b41e0af..ab7a3b273a3 100644
--- a/src/glu/sgi/libnurbs/interface/mystdlib.h
+++ b/src/glu/sgi/libnurbs/interface/mystdlib.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mystdlib.h
diff --git a/src/glu/sgi/libnurbs/internals/arc.h b/src/glu/sgi/libnurbs/internals/arc.h
index bbed33c6497..e986019c3a7 100644
--- a/src/glu/sgi/libnurbs/internals/arc.h
+++ b/src/glu/sgi/libnurbs/internals/arc.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * arc.h
diff --git a/src/glu/sgi/libnurbs/internals/arcsorter.h b/src/glu/sgi/libnurbs/internals/arcsorter.h
index 1025d30b5db..d72dd1e86ce 100644
--- a/src/glu/sgi/libnurbs/internals/arcsorter.h
+++ b/src/glu/sgi/libnurbs/internals/arcsorter.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * arcsorter.h
diff --git a/src/glu/sgi/libnurbs/internals/arctess.h b/src/glu/sgi/libnurbs/internals/arctess.h
index d3ea2071ea1..7802af90c0a 100644
--- a/src/glu/sgi/libnurbs/internals/arctess.h
+++ b/src/glu/sgi/libnurbs/internals/arctess.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * arctess.h
diff --git a/src/glu/sgi/libnurbs/internals/backend.cc b/src/glu/sgi/libnurbs/internals/backend.cc
index 69c46b2d52a..88dc3f51680 100644
--- a/src/glu/sgi/libnurbs/internals/backend.cc
+++ b/src/glu/sgi/libnurbs/internals/backend.cc
@@ -244,7 +244,7 @@ Backend::evalVStrip(int n_left, REAL u_left, REAL* left_val,
  *-------------------------------------------------------------------------
  */
 void
-Backend::bgntmesh( char * )		
+Backend::bgntmesh( const char * )		
 {
 #ifndef NOWIREFRAME
 
diff --git a/src/glu/sgi/libnurbs/internals/backend.h b/src/glu/sgi/libnurbs/internals/backend.h
index fb03859f277..6840cb19e92 100644
--- a/src/glu/sgi/libnurbs/internals/backend.h
+++ b/src/glu/sgi/libnurbs/internals/backend.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * backend.h
@@ -63,7 +59,7 @@ public:
     void		surfbbox( long, REAL *, REAL * );
     void		surfgrid( REAL, REAL, long, REAL, REAL, long ); 
     void		surfmesh( long, long, long, long ); 
-    void		bgntmesh( char * );
+    void		bgntmesh( const char * );
     void		endtmesh( void );
     void		swaptmesh( void );
     void		tmeshvert( GridTrimVertex * );
diff --git a/src/glu/sgi/libnurbs/internals/basiccrveval.h b/src/glu/sgi/libnurbs/internals/basiccrveval.h
index 41abedbb200..277dc8dc967 100644
--- a/src/glu/sgi/libnurbs/internals/basiccrveval.h
+++ b/src/glu/sgi/libnurbs/internals/basiccrveval.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * basiccurveeval.h
diff --git a/src/glu/sgi/libnurbs/internals/basicsurfeval.h b/src/glu/sgi/libnurbs/internals/basicsurfeval.h
index 2fe76ad67df..8fe18a67fb0 100644
--- a/src/glu/sgi/libnurbs/internals/basicsurfeval.h
+++ b/src/glu/sgi/libnurbs/internals/basicsurfeval.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * basicsurfeval.h
diff --git a/src/glu/sgi/libnurbs/internals/bezierarc.h b/src/glu/sgi/libnurbs/internals/bezierarc.h
index a6d5a13ee68..4acfb094881 100644
--- a/src/glu/sgi/libnurbs/internals/bezierarc.h
+++ b/src/glu/sgi/libnurbs/internals/bezierarc.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * bezierarc.h
diff --git a/src/glu/sgi/libnurbs/internals/bin.h b/src/glu/sgi/libnurbs/internals/bin.h
index ecdf9b83b83..2f976eb9bd2 100644
--- a/src/glu/sgi/libnurbs/internals/bin.h
+++ b/src/glu/sgi/libnurbs/internals/bin.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * bin.h
diff --git a/src/glu/sgi/libnurbs/internals/bufpool.cc b/src/glu/sgi/libnurbs/internals/bufpool.cc
index f60f7dc7b17..8cc847ab225 100644
--- a/src/glu/sgi/libnurbs/internals/bufpool.cc
+++ b/src/glu/sgi/libnurbs/internals/bufpool.cc
@@ -46,7 +46,7 @@
  * Pool - allocate a new pool of buffers
  *-----------------------------------------------------------------------------
  */
-Pool::Pool( int _buffersize, int initpoolsize, char *n )
+Pool::Pool( int _buffersize, int initpoolsize, const char *n )
 {
     if((unsigned)_buffersize < sizeof(Buffer))
         buffersize = sizeof(Buffer);
diff --git a/src/glu/sgi/libnurbs/internals/bufpool.h b/src/glu/sgi/libnurbs/internals/bufpool.h
index 8eaafc4fd03..0fa9433750c 100644
--- a/src/glu/sgi/libnurbs/internals/bufpool.h
+++ b/src/glu/sgi/libnurbs/internals/bufpool.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * bufpool.h
@@ -53,7 +49,7 @@ class Buffer {
 
 class Pool {
 public:
-			Pool( int, int, char * );
+			Pool( int, int, const char * );
 			~Pool( void );
     inline void*	new_buffer( void );
     inline void		free_buffer( void * );
@@ -72,7 +68,7 @@ protected:
     int			nextfree;		/* byte offset past next free buffer */
     int			initsize;
     enum Magic { is_allocated = 0xf3a1, is_free = 0xf1a2 };
-    char		*name;			/* name of the pool */
+    const char		*name;			/* name of the pool */
     Magic		magic;			/* marker for valid pool */
 };
 
diff --git a/src/glu/sgi/libnurbs/internals/cachingeval.h b/src/glu/sgi/libnurbs/internals/cachingeval.h
index cb4c83501ad..b390067a281 100644
--- a/src/glu/sgi/libnurbs/internals/cachingeval.h
+++ b/src/glu/sgi/libnurbs/internals/cachingeval.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * cachingeval.h
diff --git a/src/glu/sgi/libnurbs/internals/coveandtiler.h b/src/glu/sgi/libnurbs/internals/coveandtiler.h
index bb682b75c75..d454202e6d3 100644
--- a/src/glu/sgi/libnurbs/internals/coveandtiler.h
+++ b/src/glu/sgi/libnurbs/internals/coveandtiler.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * coveandtiler.h
diff --git a/src/glu/sgi/libnurbs/internals/curve.h b/src/glu/sgi/libnurbs/internals/curve.h
index 6f7b1de9c09..0da81c72168 100644
--- a/src/glu/sgi/libnurbs/internals/curve.h
+++ b/src/glu/sgi/libnurbs/internals/curve.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * curve.h
diff --git a/src/glu/sgi/libnurbs/internals/curvelist.h b/src/glu/sgi/libnurbs/internals/curvelist.h
index afbaa353ece..733a5115d01 100644
--- a/src/glu/sgi/libnurbs/internals/curvelist.h
+++ b/src/glu/sgi/libnurbs/internals/curvelist.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * curvelist.h
diff --git a/src/glu/sgi/libnurbs/internals/dataTransform.h b/src/glu/sgi/libnurbs/internals/dataTransform.h
index 08730e174e0..7898df754d7 100644
--- a/src/glu/sgi/libnurbs/internals/dataTransform.h
+++ b/src/glu/sgi/libnurbs/internals/dataTransform.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/internals/defines.h b/src/glu/sgi/libnurbs/internals/defines.h
index aae1682e398..32950d613ba 100644
--- a/src/glu/sgi/libnurbs/internals/defines.h
+++ b/src/glu/sgi/libnurbs/internals/defines.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * defines.h
diff --git a/src/glu/sgi/libnurbs/internals/displaylist.h b/src/glu/sgi/libnurbs/internals/displaylist.h
index 4bd6d763845..22cbec37870 100644
--- a/src/glu/sgi/libnurbs/internals/displaylist.h
+++ b/src/glu/sgi/libnurbs/internals/displaylist.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * displaylist.h
diff --git a/src/glu/sgi/libnurbs/internals/displaymode.h b/src/glu/sgi/libnurbs/internals/displaymode.h
index 9289b99b895..53968b2f1c9 100644
--- a/src/glu/sgi/libnurbs/internals/displaymode.h
+++ b/src/glu/sgi/libnurbs/internals/displaymode.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/internals/flist.h b/src/glu/sgi/libnurbs/internals/flist.h
index a643db52b84..014273ebe51 100644
--- a/src/glu/sgi/libnurbs/internals/flist.h
+++ b/src/glu/sgi/libnurbs/internals/flist.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * flist.h
diff --git a/src/glu/sgi/libnurbs/internals/flistsorter.h b/src/glu/sgi/libnurbs/internals/flistsorter.h
index d9fe81a85fe..65e6a572e30 100644
--- a/src/glu/sgi/libnurbs/internals/flistsorter.h
+++ b/src/glu/sgi/libnurbs/internals/flistsorter.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * flistsorter.h
diff --git a/src/glu/sgi/libnurbs/internals/gridline.h b/src/glu/sgi/libnurbs/internals/gridline.h
index eaa87972171..1a92e3b13b5 100644
--- a/src/glu/sgi/libnurbs/internals/gridline.h
+++ b/src/glu/sgi/libnurbs/internals/gridline.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * gridline.h
diff --git a/src/glu/sgi/libnurbs/internals/gridtrimvertex.h b/src/glu/sgi/libnurbs/internals/gridtrimvertex.h
index 72f737a9dcf..707f649d20b 100644
--- a/src/glu/sgi/libnurbs/internals/gridtrimvertex.h
+++ b/src/glu/sgi/libnurbs/internals/gridtrimvertex.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * gridtrimvertex.h
diff --git a/src/glu/sgi/libnurbs/internals/gridvertex.h b/src/glu/sgi/libnurbs/internals/gridvertex.h
index 23035a00c55..36a65c7bdd3 100644
--- a/src/glu/sgi/libnurbs/internals/gridvertex.h
+++ b/src/glu/sgi/libnurbs/internals/gridvertex.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * gridvertex.h
diff --git a/src/glu/sgi/libnurbs/internals/hull.h b/src/glu/sgi/libnurbs/internals/hull.h
index 30ffd6bac3b..dfe865267f7 100644
--- a/src/glu/sgi/libnurbs/internals/hull.h
+++ b/src/glu/sgi/libnurbs/internals/hull.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * hull.h
diff --git a/src/glu/sgi/libnurbs/internals/jarcloc.h b/src/glu/sgi/libnurbs/internals/jarcloc.h
index 3582a607a72..5297036fa55 100644
--- a/src/glu/sgi/libnurbs/internals/jarcloc.h
+++ b/src/glu/sgi/libnurbs/internals/jarcloc.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * jarcloc.h
diff --git a/src/glu/sgi/libnurbs/internals/knotvector.cc b/src/glu/sgi/libnurbs/internals/knotvector.cc
index 610046674cd..9eb5cbace94 100644
--- a/src/glu/sgi/libnurbs/internals/knotvector.cc
+++ b/src/glu/sgi/libnurbs/internals/knotvector.cc
@@ -126,7 +126,7 @@ int Knotvector::validate( void )
     return 0;
 }
 
-void Knotvector::show( char *msg )
+void Knotvector::show( const char *msg )
 {
 #ifndef NDEBUG
     _glu_dprintf( "%s\n", msg );
diff --git a/src/glu/sgi/libnurbs/internals/knotvector.h b/src/glu/sgi/libnurbs/internals/knotvector.h
index 508fc4f3450..85b162f6f11 100644
--- a/src/glu/sgi/libnurbs/internals/knotvector.h
+++ b/src/glu/sgi/libnurbs/internals/knotvector.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * knotvector.h
@@ -47,7 +43,7 @@ struct Knotvector { /* a knot vector */
 			~Knotvector( void );
     void		init( long, long, long, INREAL * );
     int			validate( void );
-    void 		show( char * );
+    void 		show( const char * );
 
     long		order;		/* order of spline  */
     long		knotcount;	/* number of knots  */
diff --git a/src/glu/sgi/libnurbs/internals/mapdesc.h b/src/glu/sgi/libnurbs/internals/mapdesc.h
index fe5d650a2af..6f06154f92a 100644
--- a/src/glu/sgi/libnurbs/internals/mapdesc.h
+++ b/src/glu/sgi/libnurbs/internals/mapdesc.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mapdesc.h
diff --git a/src/glu/sgi/libnurbs/internals/maplist.h b/src/glu/sgi/libnurbs/internals/maplist.h
index e86253966d0..49720e49b42 100644
--- a/src/glu/sgi/libnurbs/internals/maplist.h
+++ b/src/glu/sgi/libnurbs/internals/maplist.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * maplist.h
diff --git a/src/glu/sgi/libnurbs/internals/mesher.h b/src/glu/sgi/libnurbs/internals/mesher.h
index b9f74f3819d..bb65d0f1c7d 100644
--- a/src/glu/sgi/libnurbs/internals/mesher.h
+++ b/src/glu/sgi/libnurbs/internals/mesher.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mesher.h
diff --git a/src/glu/sgi/libnurbs/internals/monotonizer.h b/src/glu/sgi/libnurbs/internals/monotonizer.h
index 7282a8c4913..e951084ff03 100644
--- a/src/glu/sgi/libnurbs/internals/monotonizer.h
+++ b/src/glu/sgi/libnurbs/internals/monotonizer.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * monotonizer.h
diff --git a/src/glu/sgi/libnurbs/internals/myassert.h b/src/glu/sgi/libnurbs/internals/myassert.h
index 9b5ee0f353b..cbd6006f55c 100644
--- a/src/glu/sgi/libnurbs/internals/myassert.h
+++ b/src/glu/sgi/libnurbs/internals/myassert.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * myassert.h
diff --git a/src/glu/sgi/libnurbs/internals/mymath.h b/src/glu/sgi/libnurbs/internals/mymath.h
index 73078251dda..17ef7237fe2 100644
--- a/src/glu/sgi/libnurbs/internals/mymath.h
+++ b/src/glu/sgi/libnurbs/internals/mymath.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mymath.h
diff --git a/src/glu/sgi/libnurbs/internals/mysetjmp.h b/src/glu/sgi/libnurbs/internals/mysetjmp.h
index bfb9cea98ba..2a3f6a87b3a 100644
--- a/src/glu/sgi/libnurbs/internals/mysetjmp.h
+++ b/src/glu/sgi/libnurbs/internals/mysetjmp.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mysetjmp.h
diff --git a/src/glu/sgi/libnurbs/internals/mystring.h b/src/glu/sgi/libnurbs/internals/mystring.h
index fedf32f114c..d2952da894e 100644
--- a/src/glu/sgi/libnurbs/internals/mystring.h
+++ b/src/glu/sgi/libnurbs/internals/mystring.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mystring.h
diff --git a/src/glu/sgi/libnurbs/internals/nurbsconsts.h b/src/glu/sgi/libnurbs/internals/nurbsconsts.h
index 30277d6892e..acc5d7f31f6 100644
--- a/src/glu/sgi/libnurbs/internals/nurbsconsts.h
+++ b/src/glu/sgi/libnurbs/internals/nurbsconsts.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * nurbsconsts.h
diff --git a/src/glu/sgi/libnurbs/internals/nurbstess.cc b/src/glu/sgi/libnurbs/internals/nurbstess.cc
index a5bd060fb0e..68dfd95f394 100644
--- a/src/glu/sgi/libnurbs/internals/nurbstess.cc
+++ b/src/glu/sgi/libnurbs/internals/nurbstess.cc
@@ -675,7 +675,7 @@ NurbsTessellator::do_nurbserror( int msg )
 }
 
 int 
-NurbsTessellator::do_check_knots( Knotvector *knots, char *msg )
+NurbsTessellator::do_check_knots( Knotvector *knots, const char *msg )
 {
     int status = knots->validate();
     if( status ) {
diff --git a/src/glu/sgi/libnurbs/internals/nurbstess.h b/src/glu/sgi/libnurbs/internals/nurbstess.h
index 3577088a9c2..a6869002fb6 100644
--- a/src/glu/sgi/libnurbs/internals/nurbstess.h
+++ b/src/glu/sgi/libnurbs/internals/nurbstess.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * nurbstess.h
@@ -112,7 +108,7 @@ protected:
 private:
 
     void		resetObjects( void );
-    int			do_check_knots( Knotvector *, char * );
+    int			do_check_knots( Knotvector *, const char * );
     void		do_nurbserror( int );
     void		do_bgncurve( O_curve * );
     void		do_endcurve( void );
diff --git a/src/glu/sgi/libnurbs/internals/patch.h b/src/glu/sgi/libnurbs/internals/patch.h
index d42613b67e0..324a25938b4 100644
--- a/src/glu/sgi/libnurbs/internals/patch.h
+++ b/src/glu/sgi/libnurbs/internals/patch.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * patch.h
diff --git a/src/glu/sgi/libnurbs/internals/patchlist.h b/src/glu/sgi/libnurbs/internals/patchlist.h
index 04701c292b2..9bfc5b47a24 100644
--- a/src/glu/sgi/libnurbs/internals/patchlist.h
+++ b/src/glu/sgi/libnurbs/internals/patchlist.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * patchlist.h
diff --git a/src/glu/sgi/libnurbs/internals/pwlarc.h b/src/glu/sgi/libnurbs/internals/pwlarc.h
index b0422b4ded6..2e5175a6f33 100644
--- a/src/glu/sgi/libnurbs/internals/pwlarc.h
+++ b/src/glu/sgi/libnurbs/internals/pwlarc.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * pwlarc.h
diff --git a/src/glu/sgi/libnurbs/internals/quilt.h b/src/glu/sgi/libnurbs/internals/quilt.h
index a23c3c11b1f..fe1e3309ead 100644
--- a/src/glu/sgi/libnurbs/internals/quilt.h
+++ b/src/glu/sgi/libnurbs/internals/quilt.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * quilt.h
diff --git a/src/glu/sgi/libnurbs/internals/reader.h b/src/glu/sgi/libnurbs/internals/reader.h
index c826d3812f2..8a8dcebb501 100644
--- a/src/glu/sgi/libnurbs/internals/reader.h
+++ b/src/glu/sgi/libnurbs/internals/reader.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * reader.h
diff --git a/src/glu/sgi/libnurbs/internals/renderhints.h b/src/glu/sgi/libnurbs/internals/renderhints.h
index e2484229250..444f649be23 100644
--- a/src/glu/sgi/libnurbs/internals/renderhints.h
+++ b/src/glu/sgi/libnurbs/internals/renderhints.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * renderhints.h
diff --git a/src/glu/sgi/libnurbs/internals/simplemath.h b/src/glu/sgi/libnurbs/internals/simplemath.h
index 195471e23e6..0a060c57ead 100644
--- a/src/glu/sgi/libnurbs/internals/simplemath.h
+++ b/src/glu/sgi/libnurbs/internals/simplemath.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * simplemath.h
diff --git a/src/glu/sgi/libnurbs/internals/slicer.h b/src/glu/sgi/libnurbs/internals/slicer.h
index 6700024ba2c..e1f5d22ff10 100644
--- a/src/glu/sgi/libnurbs/internals/slicer.h
+++ b/src/glu/sgi/libnurbs/internals/slicer.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * slicer.h
diff --git a/src/glu/sgi/libnurbs/internals/sorter.h b/src/glu/sgi/libnurbs/internals/sorter.h
index 0f6b43be37b..57ea1212907 100644
--- a/src/glu/sgi/libnurbs/internals/sorter.h
+++ b/src/glu/sgi/libnurbs/internals/sorter.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/internals/subdivider.h b/src/glu/sgi/libnurbs/internals/subdivider.h
index 37970d6942f..4d539518186 100644
--- a/src/glu/sgi/libnurbs/internals/subdivider.h
+++ b/src/glu/sgi/libnurbs/internals/subdivider.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * subdivider.h
diff --git a/src/glu/sgi/libnurbs/internals/trimline.h b/src/glu/sgi/libnurbs/internals/trimline.h
index 5d52e30aba6..ecc884284cc 100644
--- a/src/glu/sgi/libnurbs/internals/trimline.h
+++ b/src/glu/sgi/libnurbs/internals/trimline.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * trimline.h
diff --git a/src/glu/sgi/libnurbs/internals/trimregion.h b/src/glu/sgi/libnurbs/internals/trimregion.h
index 6534a8c1da9..ee15d7b9888 100644
--- a/src/glu/sgi/libnurbs/internals/trimregion.h
+++ b/src/glu/sgi/libnurbs/internals/trimregion.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * trimregion.h
diff --git a/src/glu/sgi/libnurbs/internals/trimvertex.h b/src/glu/sgi/libnurbs/internals/trimvertex.h
index 85f11621670..2780b764a54 100644
--- a/src/glu/sgi/libnurbs/internals/trimvertex.h
+++ b/src/glu/sgi/libnurbs/internals/trimvertex.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * trimvertex.h
diff --git a/src/glu/sgi/libnurbs/internals/trimvertpool.h b/src/glu/sgi/libnurbs/internals/trimvertpool.h
index 2420e8cca48..ebd3c04b1f1 100644
--- a/src/glu/sgi/libnurbs/internals/trimvertpool.h
+++ b/src/glu/sgi/libnurbs/internals/trimvertpool.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * trimvertexpool.h
diff --git a/src/glu/sgi/libnurbs/internals/types.h b/src/glu/sgi/libnurbs/internals/types.h
index 3f89e525935..8b59625cff5 100644
--- a/src/glu/sgi/libnurbs/internals/types.h
+++ b/src/glu/sgi/libnurbs/internals/types.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * types.h
diff --git a/src/glu/sgi/libnurbs/internals/uarray.h b/src/glu/sgi/libnurbs/internals/uarray.h
index 908b8ccfc86..cc6fae1b754 100644
--- a/src/glu/sgi/libnurbs/internals/uarray.h
+++ b/src/glu/sgi/libnurbs/internals/uarray.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * uarray.h
diff --git a/src/glu/sgi/libnurbs/internals/varray.h b/src/glu/sgi/libnurbs/internals/varray.h
index 8408f27baee..b12ff837c85 100644
--- a/src/glu/sgi/libnurbs/internals/varray.h
+++ b/src/glu/sgi/libnurbs/internals/varray.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * varray.h
diff --git a/src/glu/sgi/libnurbs/nurbtess/definitions.h b/src/glu/sgi/libnurbs/nurbtess/definitions.h
index 8dcbf200503..962921da10d 100644
--- a/src/glu/sgi/libnurbs/nurbtess/definitions.h
+++ b/src/glu/sgi/libnurbs/nurbtess/definitions.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/directedLine.h b/src/glu/sgi/libnurbs/nurbtess/directedLine.h
index 9d68183ad30..4ed0128ca6b 100644
--- a/src/glu/sgi/libnurbs/nurbtess/directedLine.h
+++ b/src/glu/sgi/libnurbs/nurbtess/directedLine.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/glimports.h b/src/glu/sgi/libnurbs/nurbtess/glimports.h
index 2c307f63e87..6e69feb405e 100644
--- a/src/glu/sgi/libnurbs/nurbtess/glimports.h
+++ b/src/glu/sgi/libnurbs/nurbtess/glimports.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * glimports.h
diff --git a/src/glu/sgi/libnurbs/nurbtess/gridWrap.h b/src/glu/sgi/libnurbs/nurbtess/gridWrap.h
index 723988d2d07..b6428484f02 100644
--- a/src/glu/sgi/libnurbs/nurbtess/gridWrap.h
+++ b/src/glu/sgi/libnurbs/nurbtess/gridWrap.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/monoChain.h b/src/glu/sgi/libnurbs/nurbtess/monoChain.h
index 0302ff9ce28..999f1631902 100644
--- a/src/glu/sgi/libnurbs/nurbtess/monoChain.h
+++ b/src/glu/sgi/libnurbs/nurbtess/monoChain.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/monoPolyPart.h b/src/glu/sgi/libnurbs/nurbtess/monoPolyPart.h
index 51a664de347..6877a59c27a 100644
--- a/src/glu/sgi/libnurbs/nurbtess/monoPolyPart.h
+++ b/src/glu/sgi/libnurbs/nurbtess/monoPolyPart.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
  *monoPolyPart.h
  */
diff --git a/src/glu/sgi/libnurbs/nurbtess/monoTriangulation.h b/src/glu/sgi/libnurbs/nurbtess/monoTriangulation.h
index 86b8b7164be..f78b8f402e3 100644
--- a/src/glu/sgi/libnurbs/nurbtess/monoTriangulation.h
+++ b/src/glu/sgi/libnurbs/nurbtess/monoTriangulation.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/mystdio.h b/src/glu/sgi/libnurbs/nurbtess/mystdio.h
index e9947ea3937..a394e0ff95d 100644
--- a/src/glu/sgi/libnurbs/nurbtess/mystdio.h
+++ b/src/glu/sgi/libnurbs/nurbtess/mystdio.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mystdio.h
diff --git a/src/glu/sgi/libnurbs/nurbtess/mystdlib.h b/src/glu/sgi/libnurbs/nurbtess/mystdlib.h
index 2520b41e0af..ab7a3b273a3 100644
--- a/src/glu/sgi/libnurbs/nurbtess/mystdlib.h
+++ b/src/glu/sgi/libnurbs/nurbtess/mystdlib.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * mystdlib.h
diff --git a/src/glu/sgi/libnurbs/nurbtess/partitionX.h b/src/glu/sgi/libnurbs/nurbtess/partitionX.h
index bef724fe1f3..69f2b2df212 100644
--- a/src/glu/sgi/libnurbs/nurbtess/partitionX.h
+++ b/src/glu/sgi/libnurbs/nurbtess/partitionX.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/partitionY.h b/src/glu/sgi/libnurbs/nurbtess/partitionY.h
index 7e62aeaa9d5..8dda409de1a 100644
--- a/src/glu/sgi/libnurbs/nurbtess/partitionY.h
+++ b/src/glu/sgi/libnurbs/nurbtess/partitionY.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
  *partitionY.h:
  *partition a polygon into a Y-monotone polygon:
diff --git a/src/glu/sgi/libnurbs/nurbtess/polyDBG.h b/src/glu/sgi/libnurbs/nurbtess/polyDBG.h
index 832fe05093e..74843d88879 100644
--- a/src/glu/sgi/libnurbs/nurbtess/polyDBG.h
+++ b/src/glu/sgi/libnurbs/nurbtess/polyDBG.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/polyUtil.h b/src/glu/sgi/libnurbs/nurbtess/polyUtil.h
index 010838a9cc9..1ca2ebcc074 100644
--- a/src/glu/sgi/libnurbs/nurbtess/polyUtil.h
+++ b/src/glu/sgi/libnurbs/nurbtess/polyUtil.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/primitiveStream.h b/src/glu/sgi/libnurbs/nurbtess/primitiveStream.h
index 8063dcd6220..b4174f1466f 100644
--- a/src/glu/sgi/libnurbs/nurbtess/primitiveStream.h
+++ b/src/glu/sgi/libnurbs/nurbtess/primitiveStream.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/quicksort.h b/src/glu/sgi/libnurbs/nurbtess/quicksort.h
index 1a32188ee13..3a8dcd6a2cf 100644
--- a/src/glu/sgi/libnurbs/nurbtess/quicksort.h
+++ b/src/glu/sgi/libnurbs/nurbtess/quicksort.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/rectBlock.h b/src/glu/sgi/libnurbs/nurbtess/rectBlock.h
index ce546442df2..8dbd7eb819d 100644
--- a/src/glu/sgi/libnurbs/nurbtess/rectBlock.h
+++ b/src/glu/sgi/libnurbs/nurbtess/rectBlock.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampleComp.h b/src/glu/sgi/libnurbs/nurbtess/sampleComp.h
index e35e5e291db..a3c1790254e 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampleComp.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampleComp.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampleCompBot.h b/src/glu/sgi/libnurbs/nurbtess/sampleCompBot.h
index 6debef91194..7b982794c04 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampleCompBot.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampleCompBot.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampleCompRight.h b/src/glu/sgi/libnurbs/nurbtess/sampleCompRight.h
index b4b0e0732ea..4670b87ca80 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampleCompRight.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampleCompRight.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampleCompTop.h b/src/glu/sgi/libnurbs/nurbtess/sampleCompTop.h
index 695092c586c..95598d63a8b 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampleCompTop.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampleCompTop.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampleMonoPoly.h b/src/glu/sgi/libnurbs/nurbtess/sampleMonoPoly.h
index 777a28fa2b2..3a5fa6c91ba 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampleMonoPoly.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampleMonoPoly.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/sampledLine.h b/src/glu/sgi/libnurbs/nurbtess/sampledLine.h
index 147b8a5e124..2cc726948ec 100644
--- a/src/glu/sgi/libnurbs/nurbtess/sampledLine.h
+++ b/src/glu/sgi/libnurbs/nurbtess/sampledLine.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/searchTree.h b/src/glu/sgi/libnurbs/nurbtess/searchTree.h
index 246099fbac0..d5e33314bb0 100644
--- a/src/glu/sgi/libnurbs/nurbtess/searchTree.h
+++ b/src/glu/sgi/libnurbs/nurbtess/searchTree.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libnurbs/nurbtess/zlassert.h b/src/glu/sgi/libnurbs/nurbtess/zlassert.h
index 68913851967..b9c07325768 100644
--- a/src/glu/sgi/libnurbs/nurbtess/zlassert.h
+++ b/src/glu/sgi/libnurbs/nurbtess/zlassert.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 */
 
diff --git a/src/glu/sgi/libtess/dict-list.h b/src/glu/sgi/libtess/dict-list.h
index 8cc1069c52a..11331a76edf 100644
--- a/src/glu/sgi/libtess/dict-list.h
+++ b/src/glu/sgi/libtess/dict-list.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/dict.c b/src/glu/sgi/libtess/dict.c
index f42565f2ff3..49d4f759e7e 100644
--- a/src/glu/sgi/libtess/dict.c
+++ b/src/glu/sgi/libtess/dict.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/dict.h b/src/glu/sgi/libtess/dict.h
index 8cc1069c52a..11331a76edf 100644
--- a/src/glu/sgi/libtess/dict.h
+++ b/src/glu/sgi/libtess/dict.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/geom.c b/src/glu/sgi/libtess/geom.c
index 45515317517..7d3b8d8a6a5 100644
--- a/src/glu/sgi/libtess/geom.c
+++ b/src/glu/sgi/libtess/geom.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/geom.h b/src/glu/sgi/libtess/geom.h
index 7e1e86528a5..5cb76c7d154 100644
--- a/src/glu/sgi/libtess/geom.h
+++ b/src/glu/sgi/libtess/geom.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/memalloc.c b/src/glu/sgi/libtess/memalloc.c
index 11fd33b3d7e..81879ef782f 100644
--- a/src/glu/sgi/libtess/memalloc.c
+++ b/src/glu/sgi/libtess/memalloc.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/memalloc.h b/src/glu/sgi/libtess/memalloc.h
index 5cbdb5342ea..c2f969b8bea 100644
--- a/src/glu/sgi/libtess/memalloc.h
+++ b/src/glu/sgi/libtess/memalloc.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/mesh.c b/src/glu/sgi/libtess/mesh.c
index 4ffe1a61429..ae861f86428 100644
--- a/src/glu/sgi/libtess/mesh.c
+++ b/src/glu/sgi/libtess/mesh.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/mesh.h b/src/glu/sgi/libtess/mesh.h
index 299b1bacb08..690c5f2f696 100644
--- a/src/glu/sgi/libtess/mesh.h
+++ b/src/glu/sgi/libtess/mesh.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/normal.c b/src/glu/sgi/libtess/normal.c
index f69544618cf..0a2494be343 100644
--- a/src/glu/sgi/libtess/normal.c
+++ b/src/glu/sgi/libtess/normal.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/normal.h b/src/glu/sgi/libtess/normal.h
index 9a805d56f90..c376ca44523 100644
--- a/src/glu/sgi/libtess/normal.h
+++ b/src/glu/sgi/libtess/normal.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/priorityq-heap.c b/src/glu/sgi/libtess/priorityq-heap.c
index 3f8a6f5f0d5..e3a6c6068a0 100644
--- a/src/glu/sgi/libtess/priorityq-heap.c
+++ b/src/glu/sgi/libtess/priorityq-heap.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/priorityq-heap.h b/src/glu/sgi/libtess/priorityq-heap.h
index 095cc3456cf..dc9aaef87f9 100644
--- a/src/glu/sgi/libtess/priorityq-heap.h
+++ b/src/glu/sgi/libtess/priorityq-heap.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/priorityq-sort.h b/src/glu/sgi/libtess/priorityq-sort.h
index 9e62e983e99..746cf5fa69a 100644
--- a/src/glu/sgi/libtess/priorityq-sort.h
+++ b/src/glu/sgi/libtess/priorityq-sort.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/priorityq.c b/src/glu/sgi/libtess/priorityq.c
index 7eac424e96a..11f0263ac93 100644
--- a/src/glu/sgi/libtess/priorityq.c
+++ b/src/glu/sgi/libtess/priorityq.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/priorityq.h b/src/glu/sgi/libtess/priorityq.h
index 9e62e983e99..746cf5fa69a 100644
--- a/src/glu/sgi/libtess/priorityq.h
+++ b/src/glu/sgi/libtess/priorityq.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/render.c b/src/glu/sgi/libtess/render.c
index c2b12b35c3a..4f376f7f425 100644
--- a/src/glu/sgi/libtess/render.c
+++ b/src/glu/sgi/libtess/render.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/render.h b/src/glu/sgi/libtess/render.h
index 271e616c192..a298c9a9482 100644
--- a/src/glu/sgi/libtess/render.h
+++ b/src/glu/sgi/libtess/render.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/sweep.c b/src/glu/sgi/libtess/sweep.c
index 1cfd7788f78..744be6b47d7 100644
--- a/src/glu/sgi/libtess/sweep.c
+++ b/src/glu/sgi/libtess/sweep.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/sweep.h b/src/glu/sgi/libtess/sweep.h
index 74c375c9b32..feb68b0ffe5 100644
--- a/src/glu/sgi/libtess/sweep.h
+++ b/src/glu/sgi/libtess/sweep.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/tess.c b/src/glu/sgi/libtess/tess.c
index 0489350a027..029a02c3ae4 100644
--- a/src/glu/sgi/libtess/tess.c
+++ b/src/glu/sgi/libtess/tess.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/tess.h b/src/glu/sgi/libtess/tess.h
index d705d04c6e7..16249608814 100644
--- a/src/glu/sgi/libtess/tess.h
+++ b/src/glu/sgi/libtess/tess.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/tessmono.c b/src/glu/sgi/libtess/tessmono.c
index a2b6eccbacf..4d084400594 100644
--- a/src/glu/sgi/libtess/tessmono.c
+++ b/src/glu/sgi/libtess/tessmono.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libtess/tessmono.h b/src/glu/sgi/libtess/tessmono.h
index cbe8950d202..8ee1b2fe341 100644
--- a/src/glu/sgi/libtess/tessmono.h
+++ b/src/glu/sgi/libtess/tessmono.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 /*
 ** Author: Eric Veach, July 1994.
 **
diff --git a/src/glu/sgi/libutil/error.c b/src/glu/sgi/libutil/error.c
index 24d8b70f88c..e734818ed6d 100644
--- a/src/glu/sgi/libutil/error.c
+++ b/src/glu/sgi/libutil/error.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "gluos.h"
 #include "gluint.h"
@@ -80,7 +75,7 @@ gluErrorString(GLenum errorCode)
     if ((errorCode >= GLU_NURBS_ERROR1) && (errorCode <= GLU_NURBS_ERROR37)) {
 	return (const GLubyte *) __gluNURBSErrorString(errorCode - (GLU_NURBS_ERROR1 - 1));
     }
-    if ((errorCode >= GLU_TESS_ERROR1) && (errorCode <= GLU_TESS_ERROR8)) {
+    if ((errorCode >= GLU_TESS_ERROR1) && (errorCode <= GLU_TESS_ERROR6)) {
 	return (const GLubyte *) __gluTessErrorString(errorCode - (GLU_TESS_ERROR1 - 1));
     }
     return (const GLubyte *) 0;
diff --git a/src/glu/sgi/libutil/glue.c b/src/glu/sgi/libutil/glue.c
index 6a4e6c7c6f9..cd65d65ad28 100644
--- a/src/glu/sgi/libutil/glue.c
+++ b/src/glu/sgi/libutil/glue.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include <stdlib.h>
 #include "gluint.h"
diff --git a/src/glu/sgi/libutil/gluint.h b/src/glu/sgi/libutil/gluint.h
index cd2a56fed9c..fd513caa693 100644
--- a/src/glu/sgi/libutil/gluint.h
+++ b/src/glu/sgi/libutil/gluint.h
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #ifndef __gluint_h__
 #define __gluint_h__
diff --git a/src/glu/sgi/libutil/mipmap.c b/src/glu/sgi/libutil/mipmap.c
index 44f519a4e21..af647af73c9 100644
--- a/src/glu/sgi/libutil/mipmap.c
+++ b/src/glu/sgi/libutil/mipmap.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "gluos.h"
 #include <assert.h>
@@ -6627,7 +6622,7 @@ typedef void (GLAPIENTRY *TexImage3Dproc)( GLenum target, GLint level,
 
 static TexImage3Dproc pTexImage3D = 0;
 
-#ifndef _WIN32
+#if !defined(_WIN32) && !defined(__WIN32__)
 #  include <dlfcn.h>
 #  include <sys/types.h>
 #else
@@ -6642,7 +6637,7 @@ static void gluTexImage3D( GLenum target, GLint level,
 			   const GLvoid *pixels )
 {
    if (!pTexImage3D) {
-#ifdef _WIN32
+#if defined(_WIN32) || defined(__WIN32__)
       pTexImage3D = (TexImage3Dproc) wglGetProcAddress("glTexImage3D");
       if (!pTexImage3D)
 	 pTexImage3D = (TexImage3Dproc) wglGetProcAddress("glTexImage3DEXT");
diff --git a/src/glu/sgi/libutil/project.c b/src/glu/sgi/libutil/project.c
index 5ba396ca1c0..f4bf8839b66 100644
--- a/src/glu/sgi/libutil/project.c
+++ b/src/glu/sgi/libutil/project.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "gluos.h"
 #include <math.h>
@@ -166,74 +161,57 @@ static void __gluMultMatrixVecd(const GLdouble matrix[16], const GLdouble in[4],
 }
 
 /*
-** inverse = invert(src)
-** New, faster implementation by Shan Hao Bo, April 2006.
+** Invert 4x4 matrix.
+** Contributed by David Moore (See Mesa bug #6748)
 */
-static int __gluInvertMatrixd(const GLdouble src[16], GLdouble inverse[16])
+static int __gluInvertMatrixd(const GLdouble m[16], GLdouble invOut[16])
 {
-	int i, j, k;
-	double t;
-	GLdouble temp[4][4];
-	 
-	for (i=0; i<4; i++) {
-		for (j=0; j<4; j++) {
-		    temp[i][j] = src[i*4+j];
-		}
-	}
-	__gluMakeIdentityd(inverse);
-	
-	for (i = 0; i < 4; i++) {
-		if (temp[i][i] == 0.0f) {
-		    /*
-		    ** Look for non-zero element in column
-		    */
-		    for (j = i + 1; j < 4; j++) {
-				if (temp[j][i] != 0.0f) {
-				    break;
-				}
-		    }
-		
-		    if (j != 4) {
-				/*
-				 ** Swap rows.
-				 */
-				for (k = 0; k < 4; k++) {
-				    t = temp[i][k];
-				    temp[i][k] = temp[j][k];
-				    temp[j][k] = t;
-			
-				    t = inverse[i*4+k];
-				    inverse[i*4+k] = inverse[j*4+k];
-				    inverse[j*4+k] = t;
-				}
-		    }
-		    else {
-				/*
-				** No non-zero pivot.  The matrix is singular, 
-which shouldn't
-				** happen.  This means the user gave us a bad 
-matrix.
-				*/
-				return GL_FALSE;
-		    }
-		}
-		
-		t = 1.0f / temp[i][i];
-		for (k = 0; k < 4; k++) {
-		    temp[i][k] *= t;
-		    inverse[i*4+k] *= t;
-		}
-		for (j = 0; j < 4; j++) {
-		    if (j != i) {
-				t = temp[j][i];
-				for (k = 0; k < 4; k++) {
-					    temp[j][k] -= temp[i][k]*t;
-					    inverse[j*4+k] -= inverse[i*4+k]*t;
-				}
-		    }
-		}
-	}
-	return GL_TRUE;
+    double inv[16], det;
+    int i;
+
+    inv[0] =   m[5]*m[10]*m[15] - m[5]*m[11]*m[14] - m[9]*m[6]*m[15]
+             + m[9]*m[7]*m[14] + m[13]*m[6]*m[11] - m[13]*m[7]*m[10];
+    inv[4] =  -m[4]*m[10]*m[15] + m[4]*m[11]*m[14] + m[8]*m[6]*m[15]
+             - m[8]*m[7]*m[14] - m[12]*m[6]*m[11] + m[12]*m[7]*m[10];
+    inv[8] =   m[4]*m[9]*m[15] - m[4]*m[11]*m[13] - m[8]*m[5]*m[15]
+             + m[8]*m[7]*m[13] + m[12]*m[5]*m[11] - m[12]*m[7]*m[9];
+    inv[12] = -m[4]*m[9]*m[14] + m[4]*m[10]*m[13] + m[8]*m[5]*m[14]
+             - m[8]*m[6]*m[13] - m[12]*m[5]*m[10] + m[12]*m[6]*m[9];
+    inv[1] =  -m[1]*m[10]*m[15] + m[1]*m[11]*m[14] + m[9]*m[2]*m[15]
+             - m[9]*m[3]*m[14] - m[13]*m[2]*m[11] + m[13]*m[3]*m[10];
+    inv[5] =   m[0]*m[10]*m[15] - m[0]*m[11]*m[14] - m[8]*m[2]*m[15]
+             + m[8]*m[3]*m[14] + m[12]*m[2]*m[11] - m[12]*m[3]*m[10];
+    inv[9] =  -m[0]*m[9]*m[15] + m[0]*m[11]*m[13] + m[8]*m[1]*m[15]
+             - m[8]*m[3]*m[13] - m[12]*m[1]*m[11] + m[12]*m[3]*m[9];
+    inv[13] =  m[0]*m[9]*m[14] - m[0]*m[10]*m[13] - m[8]*m[1]*m[14]
+             + m[8]*m[2]*m[13] + m[12]*m[1]*m[10] - m[12]*m[2]*m[9];
+    inv[2] =   m[1]*m[6]*m[15] - m[1]*m[7]*m[14] - m[5]*m[2]*m[15]
+             + m[5]*m[3]*m[14] + m[13]*m[2]*m[7] - m[13]*m[3]*m[6];
+    inv[6] =  -m[0]*m[6]*m[15] + m[0]*m[7]*m[14] + m[4]*m[2]*m[15]
+             - m[4]*m[3]*m[14] - m[12]*m[2]*m[7] + m[12]*m[3]*m[6];
+    inv[10] =  m[0]*m[5]*m[15] - m[0]*m[7]*m[13] - m[4]*m[1]*m[15]
+             + m[4]*m[3]*m[13] + m[12]*m[1]*m[7] - m[12]*m[3]*m[5];
+    inv[14] = -m[0]*m[5]*m[14] + m[0]*m[6]*m[13] + m[4]*m[1]*m[14]
+             - m[4]*m[2]*m[13] - m[12]*m[1]*m[6] + m[12]*m[2]*m[5];
+    inv[3] =  -m[1]*m[6]*m[11] + m[1]*m[7]*m[10] + m[5]*m[2]*m[11]
+             - m[5]*m[3]*m[10] - m[9]*m[2]*m[7] + m[9]*m[3]*m[6];
+    inv[7] =   m[0]*m[6]*m[11] - m[0]*m[7]*m[10] - m[4]*m[2]*m[11]
+             + m[4]*m[3]*m[10] + m[8]*m[2]*m[7] - m[8]*m[3]*m[6];
+    inv[11] = -m[0]*m[5]*m[11] + m[0]*m[7]*m[9] + m[4]*m[1]*m[11]
+             - m[4]*m[3]*m[9] - m[8]*m[1]*m[7] + m[8]*m[3]*m[5];
+    inv[15] =  m[0]*m[5]*m[10] - m[0]*m[6]*m[9] - m[4]*m[1]*m[10]
+             + m[4]*m[2]*m[9] + m[8]*m[1]*m[6] - m[8]*m[2]*m[5];
+
+    det = m[0]*inv[0] + m[1]*inv[4] + m[2]*inv[8] + m[3]*inv[12];
+    if (det == 0)
+        return GL_FALSE;
+
+    det = 1.0 / det;
+
+    for (i = 0; i < 16; i++)
+        invOut[i] = inv[i] * det;
+
+    return GL_TRUE;
 }
 
 static void __gluMultMatricesd(const GLdouble a[16], const GLdouble b[16],
diff --git a/src/glu/sgi/libutil/quad.c b/src/glu/sgi/libutil/quad.c
index e604539c82f..d88b20f5566 100644
--- a/src/glu/sgi/libutil/quad.c
+++ b/src/glu/sgi/libutil/quad.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-**
-** http://oss.sgi.com/projects/FreeB
-**
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-**
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-**
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "gluos.h"
 #include "gluint.h"
@@ -713,8 +708,8 @@ gluSphere(GLUquadric *qobj, GLdouble radius, GLint slices, GLint stacks)
     GLfloat cosCache3b[CACHE_SIZE];
     GLfloat angle;
     GLfloat zLow, zHigh;
-    GLfloat sintemp1, sintemp2, sintemp3 = 0.0, sintemp4 = 0.0;
-    GLfloat costemp1, costemp2 = 0.0, costemp3 = 0.0, costemp4 = 0.0;
+    GLfloat sintemp1 = 0.0, sintemp2 = 0.0, sintemp3 = 0.0, sintemp4 = 0.0;
+    GLfloat costemp1 = 0.0, costemp2 = 0.0, costemp3 = 0.0, costemp4 = 0.0;
     GLboolean needCache2, needCache3;
     GLint start, finish;
 
diff --git a/src/glu/sgi/libutil/registry.c b/src/glu/sgi/libutil/registry.c
index d83d2fef117..2e3b574fb04 100644
--- a/src/glu/sgi/libutil/registry.c
+++ b/src/glu/sgi/libutil/registry.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "gluos.h"
 #include <GL/glu.h>
diff --git a/src/glut/beos/Makefile b/src/glut/beos/Makefile
index a7ba512cb11..a9f8684c9bd 100644
--- a/src/glut/beos/Makefile
+++ b/src/glut/beos/Makefile
@@ -83,9 +83,10 @@ $(TOP)/$(LIB_DIR):
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	@$(TOP)/bin/mklib -o $(GLUT_LIB) -major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
-		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) $(GLUT_LIB_DEPS) \
-		$(OBJECTS)
+	@$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
+		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
+		$(GLUT_LIB_DEPS) $(OBJECTS)
 
 clean:
 	-rm -f *.o
diff --git a/src/glut/directfb/Makefile b/src/glut/directfb/Makefile
index e881b8aacdd..bf93d63e305 100644
--- a/src/glut/directfb/Makefile
+++ b/src/glut/directfb/Makefile
@@ -61,7 +61,7 @@ default: depend $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLUT_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
 		$(GLUT_LIB_DEPS) -install $(TOP)/$(LIB_DIR) \
 		$(MKLIB_OPTIONS) $(OBJECTS)
@@ -81,7 +81,7 @@ tags:
 
 # Remove .o and backup files
 clean: depend
-	-rm -f depend
+	-rm -f depend depend.bak
 	-rm -f *.o *~ *.o *~ *.so libglut.so.3.7
 
 include depend
diff --git a/src/glut/directfb/callback.c b/src/glut/directfb/callback.c
index 4f23441167a..7c9768aac0a 100644
--- a/src/glut/directfb/callback.c
+++ b/src/glut/directfb/callback.c
@@ -278,7 +278,7 @@ __glutGetTimeout( int *ret_msec )
      gettimeofday( &now, NULL );
      
      *ret_msec = (time->tv_sec  - now.tv_sec) * 1000 +
-                 (time->tv_usec - now.tv_usec + 500) / 1000;
+                 (time->tv_usec - now.tv_usec + 999) / 1000;
                  
      return GL_TRUE;
 }
diff --git a/src/glut/directfb/internal.h b/src/glut/directfb/internal.h
index 47311c9342b..55c4b5ce280 100644
--- a/src/glut/directfb/internal.h
+++ b/src/glut/directfb/internal.h
@@ -62,6 +62,9 @@ typedef struct __GlutWindow_s {
      IDirectFBSurface      *surface;
      IDirectFBGL           *gl;
 
+     /* display mode */
+     GLenum                 mode;
+
      /* cursor position in fullscreen mode */
      int                    cx;
      int                    cy;
diff --git a/src/glut/directfb/window.c b/src/glut/directfb/window.c
index c8174d795e7..62cf8bb33e0 100644
--- a/src/glut/directfb/window.c
+++ b/src/glut/directfb/window.c
@@ -66,6 +66,9 @@ __glutCreateWindow( GLboolean fullscreen )
                     case 8:
                          config.pixelformat = DSPF_RGB332;
                          break;
+                    case 12:
+                         config.pixelformat = DSPF_ARGB4444;
+                         break;
                     case 15:
                          config.pixelformat = DSPF_ARGB1555;
                          break;
@@ -109,7 +112,7 @@ __glutCreateWindow( GLboolean fullscreen )
           if (joystick)
                joystick->AttachEventBuffer( joystick, events );
                
-          new->visible = GL_TRUE;        
+          new->visible = GL_TRUE;    
      }
      else {
           DFBWindowDescription dsc;
@@ -160,11 +163,12 @@ __glutCreateWindow( GLboolean fullscreen )
                                                   DWET_BUTTONDOWN | DWET_BUTTONUP |
                                                   DWET_ENTER      | DWET_LEAVE    |
                                                   DWET_MOTION     | DWET_SIZE );
-                                                  
           
           new->req.flags |= WINDOW_REQUEST_SHOW;
      }
 
+     new->mode = g_display_mode;
+     
      new->reshape    = GL_TRUE;
      new->visibility = GL_TRUE;
      new->redisplay  = GL_TRUE;
@@ -220,7 +224,8 @@ __glutHandleWindows( void )
      __GlutWindow *cur = g_stack;
      
      while (cur) {
-          __GlutWindow *next = cur->next;
+          __GlutWindow *next      = cur->next;
+          GLboolean     displayed = GL_FALSE;
           
           if (cur->window && cur->req.flags) {
                if (cur == g_current)
@@ -291,18 +296,29 @@ __glutHandleWindows( void )
                cur->surface->GetSize( cur->surface, &w, &h ); 
                __glutSetWindow( cur );
                reshape_func( w, h );
+               displayed = GL_TRUE;
           }
           
           if (cur->visibility && visibility_func) {
                g_idle = GL_FALSE;
                __glutSetWindow( cur );
                visibility_func( cur->visible ? GLUT_VISIBLE : GLUT_NOT_VISIBLE );
+               displayed = GL_TRUE;
           }
 
           if (cur->redisplay && display_func) {
                g_idle = GL_FALSE;
                __glutSetWindow( cur );
                display_func();
+               displayed = GL_TRUE;
+          }
+          
+          if (displayed && cur->window && cur->visible) {
+               if (!(cur->mode & GLUT_DOUBLE)) {
+                    cur->gl->Unlock( cur->gl );
+                    cur->surface->Flip( cur->surface, NULL, 0 );
+                    cur->gl->Lock( cur->gl );
+               }
           }
                
           cur->reshape    = GL_FALSE;
diff --git a/src/glut/fbdev/Makefile b/src/glut/fbdev/Makefile
index b32b4141e8d..199d8c390a6 100644
--- a/src/glut/fbdev/Makefile
+++ b/src/glut/fbdev/Makefile
@@ -63,16 +63,16 @@ default: depend $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLUT_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
 		$(GLUT_LIB_DEPS) -install $(TOP)/$(LIB_DIR) \
 		$(MKLIB_OPTIONS) $(OBJECTS)
 
 install:
-	$(INSTALL) -d $(INSTALL_DIR)/include/GL
-	$(INSTALL) -d $(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) -m 644 $(TOP)/include/GL/glut.h $(INSTALL_DIR)/include/GL
-	$(INSTALL) $(TOP)/$(LIB_DIR)/libglut* $(INSTALL_DIR)/$(LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 $(TOP)/include/GL/glut.h $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) $(TOP)/$(LIB_DIR)/libglut* $(DESTDIR)$(INSTALL_LIB_DIR)
 
 # Run 'make -f Makefile.solo dep' to update the dependencies if you change
 # what's included by any source file.
@@ -88,7 +88,7 @@ tags:
 
 # Remove .o and backup files
 clean: depend
-	-rm -f depend
+	-rm -f depend depend.bak
 	-rm -f *.o *~ *.o *~ *.so libglut.so.3.7
 
 include depend
diff --git a/src/glut/ggi/Makefile b/src/glut/ggi/Makefile
index 748c83c2286..c23dfcac105 100644
--- a/src/glut/ggi/Makefile
+++ b/src/glut/ggi/Makefile
@@ -32,7 +32,7 @@ default: $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLUT_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) \
 		-patch $(GLUT_TINY) $(MKLIB_OPTIONS) -install $(TOP)/$(LIB_DIR) \
 		$(GLUT_LIB_DEPS) $(OBJECTS)
diff --git a/src/glut/glx/Makefile b/src/glut/glx/Makefile
index 910c6fa28a4..c956c5f99c9 100644
--- a/src/glut/glx/Makefile
+++ b/src/glut/glx/Makefile
@@ -36,6 +36,7 @@ SOURCES = \
 	glut_dstr.c \
 	glut_event.c \
 	glut_ext.c \
+	glut_fcb.c \
 	glut_fullscrn.c \
 	glut_gamemode.c \
 	glut_get.c \
@@ -90,25 +91,41 @@ default: $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLUT_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
 		$(MKLIB_OPTIONS) -install $(TOP)/$(LIB_DIR) \
+		-id $(INSTALL_LIB_DIR)/lib$(GLUT_LIB).$(GLUT_MAJOR).dylib \
 		$(GLUT_LIB_DEPS) $(OBJECTS)
 
 
-install:
-	$(INSTALL) -d $(INSTALL_DIR)/include/GL
-	$(INSTALL) -d $(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) -m 644 $(TOP)/include/GL/glut.h $(INSTALL_DIR)/include/GL
-	$(INSTALL) $(TOP)/$(LIB_DIR)/libglut* $(INSTALL_DIR)/$(LIB_DIR)
+# glut pkgconfig file
+pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(GLUT_MAJOR).$(GLUT_MINOR).$(GLUT_TINY),' \
+	-e 's,@GLUT_PC_REQ_PRIV@,$(GLUT_PC_REQ_PRIV),' \
+	-e 's,@GLUT_PC_LIB_PRIV@,$(GLUT_PC_LIB_PRIV),' \
+	-e 's,@GLUT_PC_CFLAGS@,$(GLUT_PC_CFLAGS),'
+glut.pc: glut.pc.in
+	$(pcedit) $< > $@
+
+install: glut.pc
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) -m 644 $(TOP)/include/GL/glut.h $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) $(TOP)/$(LIB_DIR)/$(GLUT_LIB_GLOB) $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 glut.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 
 clean:
 	-rm -f *.o *~
 	-rm -f *.lo
 	-rm -f *.la
+	-rm -f *.pc
 	-rm -rf .libs
-	-rm -rf depend depend.bak
+	-rm -f depend depend.bak
 
 
 depend: $(SOURCES)
@@ -116,6 +133,6 @@ depend: $(SOURCES)
 	@ rm -f depend
 	@ touch depend
 	@ $(MKDEP) $(MKDEP_OPTIONS) -I$(TOP)/include $(SOURCES) \
-		> /dev/null 
+		$(X11_INCLUDES) > /dev/null 
 
 include depend
diff --git a/src/glut/glx/Makefile.mgw b/src/glut/glx/Makefile.mgw
new file mode 100644
index 00000000000..5e1fde03de2
--- /dev/null
+++ b/src/glut/glx/Makefile.mgw
@@ -0,0 +1,191 @@
+# Mesa 3-D graphics library
+# Version:  5.1
+# 
+# Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+# 
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+# 
+# The above copyright notice and this permission notice shall be included
+# in all copies or substantial portions of the Software.
+# 
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+# OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+# MinGW core makefile v1.4 for Mesa
+#
+#  Copyright (C) 2002 - Daniel Borca
+#  Email : dborca@users.sourceforge.net
+#  Web   : http://www.geocities.com/dborca
+
+# MinGW core-glut makefile updated for Mesa 7.0
+#
+#  Updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+
+#
+#  Available options:
+#
+#    Environment variables:
+#	CFLAGS
+#
+#	GLIDE		path to Glide3 SDK; used with FX.
+#			default = $(TOP)/glide3
+#	FX=1		build for 3dfx Glide3. Note that this disables
+#			compilation of most WMesa code and requires fxMesa.
+#			As a consequence, you'll need the Win32 Glide3
+#			library to build any application.
+#			default = no
+#	ICD=1		build the installable client driver interface
+#			(windows opengl driver interface)
+#			default = no
+#	X86=1		optimize for x86 (if possible, use MMX, SSE, 3DNow).
+#			default = no
+#
+#    Targets:
+#	all:		build GL
+#	clean:		remove object files
+#
+
+
+
+.PHONY: all clean
+.INTERMEDIATE: x86/gen_matypes.exe
+.SUFFIXES: .rc .res
+
+# Set this to the prefix of your build tools, i.e. mingw32-
+TOOLS_PREFIX = mingw32-
+
+TOP = ../../..
+
+LIBDIR = $(TOP)/lib
+
+LIB_NAME = glut32
+
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+GLUT_DLL = $(LIB_NAME)$(DLL_EXT)
+GLUT_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+GLUT_DEF = $(LIB_NAME).def
+
+LDLIBS = -L$(LIBDIR) -lwinmm -lgdi32 -luser32 -lopengl32 -lglu32 
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GLUT32 -DGLUT_BUILDING_LIB -DMESA -D_DLL
+CFLAGS += -DNDEBUG -DLIBRARYBUILD -I$(TOP)/include 
+
+CC = $(TOOLS_PREFIX)gcc
+CXX = $(TOOLS_PREFIX)g++
+CXXFLAGS = $(CFLAGS)
+
+AR = ar
+ARFLAGS = crus
+
+UNLINK = del $(subst /,\,$(1))
+ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+ifneq ($(wildcard $(addsuffix /rm,$(subst :, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+
+HDRS = glutint.h glutstroke.h glutbitmap.h glutwin32.h stroke.h win32_glx.h win32_x11.h
+
+SRCS = \
+	glut_bitmap.c \
+	glut_bwidth.c \
+	glut_cindex.c \
+	glut_cmap.c \
+	glut_cursor.c \
+	glut_dials.c \
+	glut_dstr.c \
+	glut_event.c \
+	glut_ext.c \
+	glut_fcb.c \
+	glut_fullscrn.c \
+	glut_gamemode.c \
+	glut_get.c \
+	glut_init.c \
+	glut_input.c \
+	glut_joy.c \
+	glut_key.c \
+	glut_keyctrl.c \
+	glut_keyup.c \
+	glut_mesa.c \
+	glut_modifier.c \
+	glut_overlay.c \
+	glut_shapes.c \
+	glut_space.c \
+	glut_stroke.c \
+	glut_swap.c \
+	glut_swidth.c \
+	glut_tablet.c \
+	glut_teapot.c \
+	glut_util.c \
+	glut_vidresize.c \
+	glut_warp.c \
+	glut_win.c \
+	glut_winmisc.c \
+	win32_glx.c \
+	win32_menu.c \
+	win32_util.c \
+	win32_winproc.c \
+	win32_x11.c
+	
+
+SRCSSEMIGENS = \
+	glut_8x13.c \
+	glut_9x15.c \
+	glut_hel10.c \
+	glut_hel12.c \
+	glut_hel18.c \
+	glut_mroman.c \
+	glut_roman.c \
+	glut_tr10.c \
+	glut_tr24.c
+
+
+
+SOURCES = $(SRCS) $(SRCSSEMIGENS)
+
+OBJECTS = $(addsuffix .o,$(basename $(SOURCES)))
+
+.c.o:
+	$(CC) -o $@ $(CFLAGS) -c $<
+.cc.o:
+	$(CXX) -o $@ $(CXXFLAGS) -c $<
+
+
+all: $(LIBDIR) $(LIBDIR)/$(GLUT_DLL) $(LIBDIR)/$(GLUT_IMP)
+
+$(LIBDIR):
+	mkdir -p $(LIBDIR)
+
+$(LIBDIR)/$(GLUT_DLL) $(LIBDIR)/$(GLUT_IMP): $(OBJECTS)
+	$(CXX) $(LDFLAGS) -o $(LIBDIR)/$(GLUT_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GLUT_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GLUT_IMP) --def $(LIBDIR)/$(GLUT_DEF) 
+
+clean:
+	-$(call UNLINK,*.o)
\ No newline at end of file
diff --git a/src/glut/glx/glut.pc.in b/src/glut/glx/glut.pc.in
new file mode 100644
index 00000000000..ae0689d7e81
--- /dev/null
+++ b/src/glut/glx/glut.pc.in
@@ -0,0 +1,13 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: glut
+Description: Mesa OpenGL Utility Toolkit library
+Requires: gl glu
+Requires.private: @GLUT_PC_REQ_PRIV@
+Version: @VERSION@
+Libs: -L${libdir} -lglut
+Libs.private: @GLUT_PC_LIB_PRIV@
+Cflags: -I${includedir} @GLUT_PC_CFLAGS@
diff --git a/src/glut/glx/glut_dstr.c b/src/glut/glx/glut_dstr.c
index ec9386f3e50..2513af45394 100644
--- a/src/glut/glx/glut_dstr.c
+++ b/src/glut/glx/glut_dstr.c
@@ -291,6 +291,7 @@ loadVisuals(int *nitems_return)
   fbmodes = (FrameBufferMode *) malloc(n * sizeof(FrameBufferMode));
   if (fbmodes == NULL) {
     *nitems_return = -1;
+    free(vlist);
     return NULL;
   }
   for (i = 0; i < n; i++) {
diff --git a/src/glut/glx/glut_event.c b/src/glut/glx/glut_event.c
index 443f9a85740..cf34e05ca98 100644
--- a/src/glut/glx/glut_event.c
+++ b/src/glut/glx/glut_event.c
@@ -24,7 +24,7 @@
 # ifdef __sgi
 #  include <bstring.h>    /* prototype for bzero used by FD_ZERO */
 # endif
-# if (defined(SVR4) || defined(CRAY) || defined(AIXV3)) && !defined(FD_SETSIZE)
+# if (defined(__FreeBSD__) || defined(SVR4) || defined(CRAY) || defined(AIXV3)) && !defined(FD_SETSIZE)
 #  include <sys/select.h> /* select system call interface */
 #  ifdef luna
 #   include <sysent.h>
@@ -172,10 +172,14 @@ handleTimeouts(void)
   GETTIMEOFDAY(&now);
   while (IS_AT_OR_AFTER(__glutTimerList->timeout, now)) {
     timer = __glutTimerList;
-    __glutTimerList = timer->next;
+    /* call the timer function */
     timer->func(timer->value);
+    /* remove from the linked list */
+    __glutTimerList = timer->next;
+    /* put this timer on the "free" list */
     timer->next = freeTimerList;
     freeTimerList = timer;
+
     if (!__glutTimerList)
       break;
   }
diff --git a/src/glut/glx/glut_fbc.c b/src/glut/glx/glut_fcb.c
index e93188b8622..c8a3422b360 100644
--- a/src/glut/glx/glut_fbc.c
+++ b/src/glut/glx/glut_fcb.c
@@ -19,7 +19,7 @@
 /* Set a Fortran callback function. */
 
 void APIENTRY
-__glutSetFCB(int which, void *func)
+__glutSetFCB(int which, GLUTproc func)
 {
 #ifdef SUPPORT_FORTRAN
   switch (which) {
@@ -100,61 +100,61 @@ __glutSetFCB(int which, void *func)
 
 /* Get a Fortran callback function. */
 
-void* APIENTRY
+GLUTproc APIENTRY
 __glutGetFCB(int which)
 {
 #ifdef SUPPORT_FORTRAN
   switch (which) {
   case GLUT_FCB_DISPLAY:
-    return (void *) __glutCurrentWindow->fdisplay;
+    return __glutCurrentWindow->fdisplay;
   case GLUT_FCB_RESHAPE:
-    return (void *) __glutCurrentWindow->freshape;
+    return __glutCurrentWindow->freshape;
   case GLUT_FCB_MOUSE:
-    return (void *) __glutCurrentWindow->fmouse;
+    return __glutCurrentWindow->fmouse;
   case GLUT_FCB_MOTION:
-    return (void *) __glutCurrentWindow->fmotion;
+    return __glutCurrentWindow->fmotion;
   case GLUT_FCB_PASSIVE:
-    return (void *) __glutCurrentWindow->fpassive;
+    return __glutCurrentWindow->fpassive;
   case GLUT_FCB_ENTRY:
-    return (void *) __glutCurrentWindow->fentry;
+    return __glutCurrentWindow->fentry;
   case GLUT_FCB_KEYBOARD:
-    return (void *) __glutCurrentWindow->fkeyboard;
+    return __glutCurrentWindow->fkeyboard;
   case GLUT_FCB_KEYBOARD_UP:
-    return (void *) __glutCurrentWindow->fkeyboardUp;
+    return __glutCurrentWindow->fkeyboardUp;
   case GLUT_FCB_WINDOW_STATUS:
-    return (void *) __glutCurrentWindow->fwindowStatus;
+    return __glutCurrentWindow->fwindowStatus;
   case GLUT_FCB_VISIBILITY:
-    return (void *) __glutCurrentWindow->fvisibility;
+    return __glutCurrentWindow->fvisibility;
   case GLUT_FCB_SPECIAL:
-    return (void *) __glutCurrentWindow->fspecial;
+    return __glutCurrentWindow->fspecial;
   case GLUT_FCB_SPECIAL_UP:
-    return (void *) __glutCurrentWindow->fspecialUp;
+    return __glutCurrentWindow->fspecialUp;
   case GLUT_FCB_BUTTON_BOX:
-    return (void *) __glutCurrentWindow->fbuttonBox;
+    return __glutCurrentWindow->fbuttonBox;
   case GLUT_FCB_DIALS:
-    return (void *) __glutCurrentWindow->fdials;
+    return __glutCurrentWindow->fdials;
   case GLUT_FCB_SPACE_MOTION:
-    return (void *) __glutCurrentWindow->fspaceMotion;
+    return __glutCurrentWindow->fspaceMotion;
   case GLUT_FCB_SPACE_ROTATE:
-    return (void *) __glutCurrentWindow->fspaceRotate;
+    return __glutCurrentWindow->fspaceRotate;
   case GLUT_FCB_SPACE_BUTTON:
-    return (void *) __glutCurrentWindow->fspaceButton;
+    return __glutCurrentWindow->fspaceButton;
   case GLUT_FCB_TABLET_MOTION:
-    return (void *) __glutCurrentWindow->ftabletMotion;
+    return __glutCurrentWindow->ftabletMotion;
   case GLUT_FCB_TABLET_BUTTON:
-    return (void *) __glutCurrentWindow->ftabletButton;
+    return __glutCurrentWindow->ftabletButton;
   case GLUT_FCB_JOYSTICK:
 #ifdef _WIN32
-    return (void *) __glutCurrentWindow->fjoystick;
+    return __glutCurrentWindow->fjoystick;
 #else
     return NULL;
 #endif
   case GLUT_FCB_OVERLAY_DISPLAY:
-    return (void *) __glutCurrentWindow->overlay->fdisplay;
+    return __glutCurrentWindow->overlay->fdisplay;
   case GLUT_FCB_SELECT:
-    return (void *) __glutCurrentMenu->fselect;
+    return __glutCurrentMenu->fselect;
   case GLUT_FCB_TIMER:
-    return (void *) __glutTimerList->ffunc;
+     return __glutTimerList ? __glutTimerList->ffunc : NULL;
   default:
     return NULL;
   }
diff --git a/src/glut/mini/Makefile b/src/glut/mini/Makefile
index 980306d0e28..841a473f891 100644
--- a/src/glut/mini/Makefile
+++ b/src/glut/mini/Makefile
@@ -55,7 +55,7 @@ default: depend $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME)
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLUT_LIB_NAME): depend $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLUT_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLUT_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GLUT_MAJOR) -minor $(GLUT_MINOR) -patch $(GLUT_TINY) \
 		$(GLUT_LIB_DEPS) -install $(TOP)/$(LIB_DIR) \
 		$(MKLIB_OPTIONS) $(OBJECTS)
@@ -73,9 +73,30 @@ tags:
 	etags `find . -name \*.[ch]` `find ../include`
 
 
+# glut pkgconfig file
+pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(GLUT_MAJOR).$(GLUT_MINOR).$(GLUT_TINY),' \
+	-e 's,@GLUT_PC_REQ_PRIV@,$(GLUT_PC_REQ_PRIV),' \
+	-e 's,@GLUT_PC_LIB_PRIV@,$(GLUT_PC_LIB_PRIV),' \
+	-e 's,@GLUT_PC_CFLAGS@,$(GLUT_PC_CFLAGS),'
+glut.pc: glut.pc.in
+	$(pcedit) $< > $@
+
+install: glut.pc
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) -m 644 $(TOP)/include/GL/glut.h $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) $(TOP)/$(LIB_DIR)/libglut* $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 glut.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+
+
 # Remove .o and backup files
 clean: depend
-	-rm -f depend
-	-rm -f *.o *~ *.o *~ *.so libglut.so.3.7
+	-rm -f depend depend.bak
+	-rm -f *.o *~ *.o *~ *.so *.pc libglut.so.3.7
 
 include depend
diff --git a/src/glut/mini/glut.pc.in b/src/glut/mini/glut.pc.in
new file mode 100644
index 00000000000..ae0689d7e81
--- /dev/null
+++ b/src/glut/mini/glut.pc.in
@@ -0,0 +1,13 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: glut
+Description: Mesa OpenGL Utility Toolkit library
+Requires: gl glu
+Requires.private: @GLUT_PC_REQ_PRIV@
+Version: @VERSION@
+Libs: -L${libdir} -lglut
+Libs.private: @GLUT_PC_LIB_PRIV@
+Cflags: -I${includedir} @GLUT_PC_CFLAGS@
diff --git a/src/glw/Makefile b/src/glw/Makefile
index 86a0bbaeace..b153a6df751 100644
--- a/src/glw/Makefile
+++ b/src/glw/Makefile
@@ -7,7 +7,7 @@ MAJOR = 1
 MINOR = 0
 TINY = 0
 
-INCDIRS = -I$(TOP)/include -I/usr/include/Motif1.2 $(X11_INCLUDES)
+INCDIRS = -I$(TOP)/include $(MOTIF_CFLAGS) $(X11_INCLUDES)
 
 
 OBJECTS = $(GLW_SOURCES:.c=.o)
@@ -25,22 +25,37 @@ OBJECTS = $(GLW_SOURCES:.c=.o)
 
 default: $(TOP)/$(LIB_DIR)/$(GLW_LIB_NAME)
 
-install:
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/include/GL
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) -m 644 *.h $(DESTDIR)$(INSTALL_DIR)/include/GL
-	$(INSTALL) $(TOP)/$(LIB_DIR)/libGLw.* $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
+# GLU pkg-config file
+pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(MAJOR).$(MINOR).$(TINY),' \
+	-e 's,@GLW_PC_REQ_PRIV@,$(GLW_PC_REQ_PRIV),' \
+	-e 's,@GLW_PC_LIB_PRIV@,$(GLW_PC_LIB_PRIV),' \
+	-e 's,@GLW_PC_CFLAGS@,$(GLW_PC_CFLAGS),'
+glw.pc: glw.pc.in
+	$(pcedit) $< > $@
+
+install: glw.pc
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) -m 644 *.h $(DESTDIR)$(INSTALL_INC_DIR)/GL
+	$(INSTALL) $(TOP)/$(LIB_DIR)/$(GLW_LIB_GLOB) $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 glw.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 clean:
-	-rm depend depend.bak
-	-rm -f *.o *~
+	-rm -f depend depend.bak
+	-rm -f *.o *.pc *~
 
 
 # Make the library
 $(TOP)/$(LIB_DIR)/$(GLW_LIB_NAME): $(OBJECTS)
-	$(TOP)/bin/mklib -o $(GLW_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GLW_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MAJOR) -minor $(MINOR) -patch $(TINY) \
 		$(MKLIB_OPTIONS) -install $(TOP)/$(LIB_DIR) \
+		-id $(INSTALL_LIB_DIR)/lib$(GLW_LIB).$(MAJOR).dylib \
 		$(GLW_LIB_DEPS) $(OBJECTS)
 
 
@@ -52,7 +67,7 @@ depend: $(GLW_SOURCES)
 	rm -f depend
 	touch depend
 	$(MKDEP) $(MKDEP_OPTIONS) -I$(TOP)/include $(GLW_SOURCES) \
-		> /dev/null 
+		$(X11_INCLUDES) > /dev/null
 
 
 include depend
diff --git a/src/glw/glw.pc.in b/src/glw/glw.pc.in
new file mode 100644
index 00000000000..5493093be19
--- /dev/null
+++ b/src/glw/glw.pc.in
@@ -0,0 +1,13 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: glw
+Description: Mesa OpenGL widget library
+Requires: gl
+Requires.private: @GLW_PC_REQ_PRIV@
+Version: @VERSION@
+Libs: -L${libdir} -lGLw
+Libs.private: @GLW_PC_LIB_PRIV@
+Cflags: -I${includedir} @GLW_PC_CFLAGS@
diff --git a/src/glx/Makefile b/src/glx/Makefile
index bd486cf81b3..e0ab4a02401 100644
--- a/src/glx/Makefile
+++ b/src/glx/Makefile
@@ -4,9 +4,9 @@ include $(TOP)/configs/current
 
 
 default:
-	cd mini ; $(MAKE)
+	cd mini && $(MAKE)
 
 
 clean:
-	cd mini ; $(MAKE) clean
+	-@cd mini && $(MAKE) clean
 
diff --git a/src/glx/mini/Makefile b/src/glx/mini/Makefile
index 9545505cbff..6b5a3c76d74 100644
--- a/src/glx/mini/Makefile
+++ b/src/glx/mini/Makefile
@@ -49,7 +49,7 @@ default: depend $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) $(TOP)/$(LIB_DIR)/miniglx.conf
 
 # Make libGL
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME):  $(OBJECTS) Makefile
-	@ $(TOP)/bin/mklib -o $(GL_LIB) -linker '$(CC)' \
+	@ $(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major 1 -minor 2 $(MKLIB_OPTIONS) \
 		-install $(TOP)/$(LIB_DIR) $(GL_LIB_DEPS) $(OBJECTS) \
 		$(LIBDRM_LIB) $(PCIACCESS_LIB)
@@ -76,10 +76,14 @@ tags:
 	etags `find . -name \*.[ch]` `find ../include`
 
 
+# Dummy install target
+install:
+
+
 # Remove .o and backup files
 clean:
 	-rm -f drmtest $(TOP)/$(LIB_DIR)/libGL.so*
 	-rm -f *.o *~
-	-rm -f depend
+	-rm -f depend depend.bak
 
 include depend
diff --git a/src/glx/x11/.gitignore b/src/glx/x11/.gitignore
deleted file mode 100644
index ffc14ab7660..00000000000
--- a/src/glx/x11/.gitignore
+++ /dev/null
@@ -1 +0,0 @@
-glcontextmodes.c
diff --git a/src/glx/x11/Makefile b/src/glx/x11/Makefile
index c99bcf26a90..86d84d4b9ff 100644
--- a/src/glx/x11/Makefile
+++ b/src/glx/x11/Makefile
@@ -10,12 +10,14 @@ SOURCES = \
 	  compsize.c \
 	  eval.c \
 	  glxcmds.c \
+	  glxcurrent.c \
 	  glxext.c \
 	  glxextensions.c \
 	  indirect.c \
 	  indirect_init.c \
 	  indirect_size.c \
 	  indirect_window_pos.c \
+	  indirect_texture_compression.c \
 	  indirect_transpose_matrix.c \
 	  indirect_vertex_array.c \
 	  indirect_vertex_program.c \
@@ -29,9 +31,13 @@ SOURCES = \
 	  xfont.c \
 	  glx_pbuffer.c \
 	  glx_query.c \
-	  glx_texture_compression.c \
+	  drisw_glx.c \
+	  dri_common.c \
 	  dri_glx.c \
-	  XF86dri.c
+	  XF86dri.c \
+	  glxhash.c \
+	  dri2_glx.c \
+	  dri2.c
 
 include $(TOP)/src/mesa/sources.mak
 
@@ -45,9 +51,9 @@ INCLUDES = -I. \
 	-I$(TOP)/include \
 	-I$(TOP)/include/GL/internal \
 	-I$(TOP)/src/mesa \
-	-I$(TOP)/src/mesa/main \
 	-I$(TOP)/src/mesa/glapi \
 	$(LIBDRM_CFLAGS) \
+	$(DRI2PROTO_CFLAGS) \
 	$(X11_INCLUDES)
 
 
@@ -65,9 +71,10 @@ default: depend $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)
 
 # Make libGL
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME):  $(OBJECTS) Makefile
-	$(TOP)/bin/mklib -o $(GL_LIB) -linker '$(CC)' \
+	$(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major 1 -minor 2 $(MKLIB_OPTIONS) \
-		-install $(TOP)/$(LIB_DIR) $(GL_LIB_DEPS) $(OBJECTS)
+		-install $(TOP)/$(LIB_DIR) -id $(INSTALL_LIB_DIR)/lib$(GL_LIB).1.dylib \
+		$(GL_LIB_DEPS) $(OBJECTS)
 
 
 depend: $(SOURCES) $(MESA_GLAPI_SOURCES) $(MESA_GLAPI_ASM_SOURCES) Makefile
@@ -81,13 +88,13 @@ depend: $(SOURCES) $(MESA_GLAPI_SOURCES) $(MESA_GLAPI_ASM_SOURCES) Makefile
 tags:
 	etags `find . -name \*.[ch]` `find $(TOP)/include`
 
-# Dummy install target
-install:
+install: $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)
+	$(MAKE) -C $(TOP)/src/mesa install-libgl
 
 # Remove .o and backup files
 clean:
 	-rm -f $(TOP)/$(LIB_DIR)/libGL.so*
 	-rm -f *.o *~
-	-rm -f depend
+	-rm -f depend depend.bak
 
-include depend
+-include depend
diff --git a/src/glx/x11/XF86dri.c b/src/glx/x11/XF86dri.c
index 9919a409771..90a5dd81eab 100644
--- a/src/glx/x11/XF86dri.c
+++ b/src/glx/x11/XF86dri.c
@@ -42,9 +42,19 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <X11/Xlibint.h>
 #include <X11/extensions/Xext.h>
 #include <X11/extensions/extutil.h>
-#include "glheader.h"
 #include "xf86dristr.h"
 
+
+#if defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 303
+#  define PUBLIC __attribute__((visibility("default")))
+#  define USED __attribute__((used))
+#else
+#  define PUBLIC
+#  define USED
+#endif
+
+
+
 static XExtensionInfo _xf86dri_info_data;
 static XExtensionInfo *xf86dri_info = &_xf86dri_info_data;
 static char xf86dri_extension_name[] = XF86DRINAME;
@@ -58,27 +68,28 @@ static char xf86dri_extension_name[] = XF86DRINAME;
  *                                                                           *
  *****************************************************************************/
 
-static int close_display(Display *dpy, XExtCodes *extCodes);
+static int close_display(Display * dpy, XExtCodes * extCodes);
 static /* const */ XExtensionHooks xf86dri_extension_hooks = {
-    NULL,				/* create_gc */
-    NULL,				/* copy_gc */
-    NULL,				/* flush_gc */
-    NULL,				/* free_gc */
-    NULL,				/* create_font */
-    NULL,				/* free_font */
-    close_display,			/* close_display */
-    NULL,				/* wire_to_event */
-    NULL,				/* event_to_wire */
-    NULL,				/* error */
-    NULL,				/* error_string */
+   NULL,                        /* create_gc */
+   NULL,                        /* copy_gc */
+   NULL,                        /* flush_gc */
+   NULL,                        /* free_gc */
+   NULL,                        /* create_font */
+   NULL,                        /* free_font */
+   close_display,               /* close_display */
+   NULL,                        /* wire_to_event */
+   NULL,                        /* event_to_wire */
+   NULL,                        /* error */
+   NULL,                        /* error_string */
 };
 
-static XEXT_GENERATE_FIND_DISPLAY (find_display, xf86dri_info, 
-				   xf86dri_extension_name, 
-				   &xf86dri_extension_hooks, 
-				   0, NULL)
+static
+XEXT_GENERATE_FIND_DISPLAY(find_display, xf86dri_info,
+                           xf86dri_extension_name,
+                           &xf86dri_extension_hooks, 0, NULL)
 
-static XEXT_GENERATE_CLOSE_DISPLAY (close_display, xf86dri_info)
+static
+XEXT_GENERATE_CLOSE_DISPLAY(close_display, xf86dri_info)
 
 
 /*****************************************************************************
@@ -86,7 +97,6 @@ static XEXT_GENERATE_CLOSE_DISPLAY (close_display, xf86dri_info)
  *		    public XFree86-DRI Extension routines                    *
  *                                                                           *
  *****************************************************************************/
-
 #if 0
 #include <stdio.h>
 #define TRACE(msg)  fprintf(stderr,"XF86DRI%s\n", msg);
@@ -94,529 +104,534 @@ static XEXT_GENERATE_CLOSE_DISPLAY (close_display, xf86dri_info)
 #define TRACE(msg)
 #endif
 
+PUBLIC Bool
+XF86DRIQueryExtension(Display * dpy, int *event_basep,
+                      int *error_basep)
+{
+   XExtDisplayInfo *info = find_display(dpy);
+
+   TRACE("QueryExtension...");
+   if (XextHasExtension(info)) {
+      *event_basep = info->codes->first_event;
+      *error_basep = info->codes->first_error;
+      TRACE("QueryExtension... return True");
+      return True;
+   }
+   else {
+      TRACE("QueryExtension... return False");
+      return False;
+   }
+}
 
-PUBLIC Bool XF86DRIQueryExtension (dpy, event_basep, error_basep)
-    Display *dpy;
-    int *event_basep, *error_basep;
+PUBLIC Bool
+XF86DRIQueryVersion(Display * dpy, int *majorVersion, int *minorVersion,
+                    int *patchVersion)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-
-    TRACE("QueryExtension...");
-    if (XextHasExtension(info)) {
-	*event_basep = info->codes->first_event;
-	*error_basep = info->codes->first_error;
-        TRACE("QueryExtension... return True");
-	return True;
-    } else {
-        TRACE("QueryExtension... return False");
-	return False;
-    }
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIQueryVersionReply rep;
+   xXF86DRIQueryVersionReq *req;
+
+   TRACE("QueryVersion...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIQueryVersion, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIQueryVersion;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("QueryVersion... return False");
+      return False;
+   }
+   *majorVersion = rep.majorVersion;
+   *minorVersion = rep.minorVersion;
+   *patchVersion = rep.patchVersion;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("QueryVersion... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIQueryVersion(dpy, majorVersion, minorVersion, patchVersion)
-    Display* dpy;
-    int* majorVersion; 
-    int* minorVersion;
-    int* patchVersion;
+PUBLIC Bool
+XF86DRIQueryDirectRenderingCapable(Display * dpy, int screen,
+                                   Bool * isCapable)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIQueryVersionReply rep;
-    xXF86DRIQueryVersionReq *req;
-
-    TRACE("QueryVersion...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIQueryVersion, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIQueryVersion;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("QueryVersion... return False");
-	return False;
-    }
-    *majorVersion = rep.majorVersion;
-    *minorVersion = rep.minorVersion;
-    *patchVersion = rep.patchVersion;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("QueryVersion... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIQueryDirectRenderingCapableReply rep;
+   xXF86DRIQueryDirectRenderingCapableReq *req;
+
+   TRACE("QueryDirectRenderingCapable...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIQueryDirectRenderingCapable, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIQueryDirectRenderingCapable;
+   req->screen = screen;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("QueryDirectRenderingCapable... return False");
+      return False;
+   }
+   *isCapable = rep.isCapable;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("QueryDirectRenderingCapable... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIQueryDirectRenderingCapable(dpy, screen, isCapable)
-    Display* dpy;
-    int screen;
-    Bool* isCapable;
+PUBLIC Bool
+XF86DRIOpenConnection(Display * dpy, int screen, drm_handle_t * hSAREA,
+                      char **busIdString)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIQueryDirectRenderingCapableReply rep;
-    xXF86DRIQueryDirectRenderingCapableReq *req;
-
-    TRACE("QueryDirectRenderingCapable...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIQueryDirectRenderingCapable, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIQueryDirectRenderingCapable;
-    req->screen = screen;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("QueryDirectRenderingCapable... return False");
-	return False;
-    }
-    *isCapable = rep.isCapable;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("QueryDirectRenderingCapable... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIOpenConnectionReply rep;
+   xXF86DRIOpenConnectionReq *req;
+
+   TRACE("OpenConnection...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIOpenConnection, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIOpenConnection;
+   req->screen = screen;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("OpenConnection... return False");
+      return False;
+   }
+
+   *hSAREA = rep.hSAREALow;
+   if (sizeof(drm_handle_t) == 8) {
+      int shift = 32;           /* var to prevent warning on next line */
+      *hSAREA |= ((drm_handle_t) rep.hSAREAHigh) << shift;
+   }
+
+   if (rep.length) {
+      if (!(*busIdString = (char *) Xcalloc(rep.busIdStringLength + 1, 1))) {
+         _XEatData(dpy, ((rep.busIdStringLength + 3) & ~3));
+         UnlockDisplay(dpy);
+         SyncHandle();
+         TRACE("OpenConnection... return False");
+         return False;
+      }
+      _XReadPad(dpy, *busIdString, rep.busIdStringLength);
+   }
+   else {
+      *busIdString = NULL;
+   }
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("OpenConnection... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIOpenConnection(dpy, screen, hSAREA, busIdString)
-    Display* dpy;
-    int screen;
-    drm_handle_t * hSAREA;
-    char **busIdString;
+PUBLIC Bool
+XF86DRIAuthConnection(Display * dpy, int screen, drm_magic_t magic)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIOpenConnectionReply rep;
-    xXF86DRIOpenConnectionReq *req;
-
-    TRACE("OpenConnection...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIOpenConnection, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIOpenConnection;
-    req->screen = screen;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("OpenConnection... return False");
-	return False;
-    }
-
-    *hSAREA = rep.hSAREALow;
-    if (sizeof(drm_handle_t) == 8) {
-       int shift = 32; /* var to prevent warning on next line */
-       *hSAREA |= ((drm_handle_t) rep.hSAREAHigh) << shift;
-    }
-
-    if (rep.length) {
-        if (!(*busIdString = (char *)Xcalloc(rep.busIdStringLength + 1, 1))) {
-            _XEatData(dpy, ((rep.busIdStringLength+3) & ~3));
-            UnlockDisplay(dpy);
-            SyncHandle();
-            TRACE("OpenConnection... return False");
-            return False;
-        }
-	_XReadPad(dpy, *busIdString, rep.busIdStringLength);
-    } else {
-        *busIdString = NULL;
-    }
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("OpenConnection... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIAuthConnectionReq *req;
+   xXF86DRIAuthConnectionReply rep;
+
+   TRACE("AuthConnection...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIAuthConnection, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIAuthConnection;
+   req->screen = screen;
+   req->magic = magic;
+   rep.authenticated = 0;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse) || !rep.authenticated) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("AuthConnection... return False");
+      return False;
+   }
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("AuthConnection... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIAuthConnection(dpy, screen, magic)
-    Display* dpy;
-    int screen;
-    drm_magic_t magic;
+PUBLIC Bool
+XF86DRICloseConnection(Display * dpy, int screen)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIAuthConnectionReq *req;
-    xXF86DRIAuthConnectionReply rep;
-
-    TRACE("AuthConnection...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIAuthConnection, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIAuthConnection;
-    req->screen = screen;
-    req->magic = magic;
-    rep.authenticated = 0;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse) || !rep.authenticated) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("AuthConnection... return False");
-	return False;
-    }
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("AuthConnection... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRICloseConnectionReq *req;
+
+   TRACE("CloseConnection...");
+
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRICloseConnection, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRICloseConnection;
+   req->screen = screen;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("CloseConnection... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRICloseConnection(dpy, screen)
-    Display* dpy;
-    int screen;
+PUBLIC Bool
+XF86DRIGetClientDriverName(Display * dpy, int screen,
+                           int *ddxDriverMajorVersion,
+                           int *ddxDriverMinorVersion,
+                           int *ddxDriverPatchVersion,
+                           char **clientDriverName)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRICloseConnectionReq *req;
-
-    TRACE("CloseConnection...");
-
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRICloseConnection, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRICloseConnection;
-    req->screen = screen;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("CloseConnection... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIGetClientDriverNameReply rep;
+   xXF86DRIGetClientDriverNameReq *req;
+
+   TRACE("GetClientDriverName...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIGetClientDriverName, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIGetClientDriverName;
+   req->screen = screen;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("GetClientDriverName... return False");
+      return False;
+   }
+
+   *ddxDriverMajorVersion = rep.ddxDriverMajorVersion;
+   *ddxDriverMinorVersion = rep.ddxDriverMinorVersion;
+   *ddxDriverPatchVersion = rep.ddxDriverPatchVersion;
+
+   if (rep.length) {
+      if (!
+          (*clientDriverName =
+           (char *) Xcalloc(rep.clientDriverNameLength + 1, 1))) {
+         _XEatData(dpy, ((rep.clientDriverNameLength + 3) & ~3));
+         UnlockDisplay(dpy);
+         SyncHandle();
+         TRACE("GetClientDriverName... return False");
+         return False;
+      }
+      _XReadPad(dpy, *clientDriverName, rep.clientDriverNameLength);
+   }
+   else {
+      *clientDriverName = NULL;
+   }
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("GetClientDriverName... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIGetClientDriverName(dpy, screen, ddxDriverMajorVersion, 
-	ddxDriverMinorVersion, ddxDriverPatchVersion, clientDriverName)
-    Display* dpy;
-    int screen;
-    int* ddxDriverMajorVersion;
-    int* ddxDriverMinorVersion;
-    int* ddxDriverPatchVersion;
-    char** clientDriverName;
+PUBLIC Bool
+XF86DRICreateContextWithConfig(Display * dpy, int screen, int configID,
+                               XID * context, drm_context_t * hHWContext)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIGetClientDriverNameReply rep;
-    xXF86DRIGetClientDriverNameReq *req;
-
-    TRACE("GetClientDriverName...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIGetClientDriverName, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIGetClientDriverName;
-    req->screen = screen;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("GetClientDriverName... return False");
-	return False;
-    }
-
-    *ddxDriverMajorVersion = rep.ddxDriverMajorVersion;
-    *ddxDriverMinorVersion = rep.ddxDriverMinorVersion;
-    *ddxDriverPatchVersion = rep.ddxDriverPatchVersion;
-
-    if (rep.length) {
-        if (!(*clientDriverName = (char *)Xcalloc(rep.clientDriverNameLength + 1, 1))) {
-            _XEatData(dpy, ((rep.clientDriverNameLength+3) & ~3));
-            UnlockDisplay(dpy);
-            SyncHandle();
-            TRACE("GetClientDriverName... return False");
-            return False;
-        }
-	_XReadPad(dpy, *clientDriverName, rep.clientDriverNameLength);
-    } else {
-        *clientDriverName = NULL;
-    }
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("GetClientDriverName... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRICreateContextReply rep;
+   xXF86DRICreateContextReq *req;
+
+   TRACE("CreateContext...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRICreateContext, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRICreateContext;
+   req->visual = configID;
+   req->screen = screen;
+   *context = XAllocID(dpy);
+   req->context = *context;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("CreateContext... return False");
+      return False;
+   }
+   *hHWContext = rep.hHWContext;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("CreateContext... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRICreateContextWithConfig(dpy, screen, configID, context,
-	hHWContext)
-    Display* dpy;
-    int screen;
-    int configID;
-    XID* context;
-    drm_context_t * hHWContext;
+PUBLIC Bool
+XF86DRICreateContext(Display * dpy, int screen, Visual * visual,
+                     XID * context, drm_context_t * hHWContext)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRICreateContextReply rep;
-    xXF86DRICreateContextReq *req;
-
-    TRACE("CreateContext...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRICreateContext, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRICreateContext;
-    req->visual = configID;
-    req->screen = screen;
-    *context = XAllocID(dpy);
-    req->context = *context;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("CreateContext... return False");
-	return False;
-    }
-    *hHWContext = rep.hHWContext;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("CreateContext... return True");
-    return True;
+   return XF86DRICreateContextWithConfig(dpy, screen, visual->visualid,
+                                         context, hHWContext);
 }
 
-PUBLIC Bool XF86DRICreateContext(dpy, screen, visual, context, hHWContext)
-    Display* dpy;
-    int screen;
-    Visual* visual;
-    XID* context;
-    drm_context_t * hHWContext;
+PUBLIC Bool
+XF86DRIDestroyContext(Display * dpy, int screen, XID context)
 {
-    return XF86DRICreateContextWithConfig( dpy, screen, visual->visualid,
-					   context, hHWContext );
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIDestroyContextReq *req;
+
+   TRACE("DestroyContext...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIDestroyContext, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIDestroyContext;
+   req->screen = screen;
+   req->context = context;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("DestroyContext... return True");
+   return True;
 }
 
-PUBLIC GLboolean XF86DRIDestroyContext( __DRInativeDisplay * ndpy, int screen, 
-    __DRIid context )
+PUBLIC Bool
+XF86DRICreateDrawable(Display * dpy, int screen,
+                      XID drawable, drm_drawable_t * hHWDrawable)
 {
-    Display * const dpy = (Display *) ndpy;
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIDestroyContextReq *req;
-
-    TRACE("DestroyContext...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIDestroyContext, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIDestroyContext;
-    req->screen = screen;
-    req->context = context;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("DestroyContext... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRICreateDrawableReply rep;
+   xXF86DRICreateDrawableReq *req;
+
+   TRACE("CreateDrawable...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRICreateDrawable, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRICreateDrawable;
+   req->screen = screen;
+   req->drawable = drawable;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("CreateDrawable... return False");
+      return False;
+   }
+   *hHWDrawable = rep.hHWDrawable;
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("CreateDrawable... return True");
+   return True;
 }
 
-PUBLIC GLboolean XF86DRICreateDrawable( __DRInativeDisplay * ndpy, int screen, 
-    __DRIid drawable, drm_drawable_t * hHWDrawable )
+static int
+noopErrorHandler(Display * dpy, XErrorEvent * xerr)
 {
-    Display * const dpy = (Display *) ndpy;
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRICreateDrawableReply rep;
-    xXF86DRICreateDrawableReq *req;
-
-    TRACE("CreateDrawable...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRICreateDrawable, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRICreateDrawable;
-    req->screen = screen;
-    req->drawable = drawable;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("CreateDrawable... return False");
-	return False;
-    }
-    *hHWDrawable = rep.hHWDrawable;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("CreateDrawable... return True");
-    return True;
+   return 0;
 }
 
-PUBLIC GLboolean XF86DRIDestroyDrawable( __DRInativeDisplay * ndpy, int screen,
-    __DRIid drawable )
+PUBLIC Bool
+XF86DRIDestroyDrawable(Display * dpy, int screen, XID drawable)
 {
-    Display * const dpy = (Display *) ndpy;
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIDestroyDrawableReq *req;
-
-    TRACE("DestroyDrawable...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIDestroyDrawable, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIDestroyDrawable;
-    req->screen = screen;
-    req->drawable = drawable;
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("DestroyDrawable... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIDestroyDrawableReq *req;
+   int (*oldXErrorHandler) (Display *, XErrorEvent *);
+
+   TRACE("DestroyDrawable...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   /* This is called from the DRI driver, which used call it like this
+    *
+    *   if (windowExists(drawable))
+    *     destroyDrawable(drawable);
+    *
+    * which is a textbook race condition - the window may disappear
+    * from the server between checking for its existance and
+    * destroying it.  Instead we change the semantics of
+    * __DRIinterfaceMethodsRec::destroyDrawable() to succeed even if
+    * the windows is gone, by wrapping the destroy call in an error
+    * handler. */
+
+   XSync(dpy, False);
+   oldXErrorHandler = XSetErrorHandler(noopErrorHandler);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIDestroyDrawable, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIDestroyDrawable;
+   req->screen = screen;
+   req->drawable = drawable;
+   UnlockDisplay(dpy);
+   SyncHandle();
+
+   XSetErrorHandler(oldXErrorHandler);
+
+   TRACE("DestroyDrawable... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIGetDrawableInfo(Display* dpy, int screen, Drawable drawable,
-    unsigned int* index, unsigned int* stamp,
-    int* X, int* Y, int* W, int* H,
-    int* numClipRects, drm_clip_rect_t ** pClipRects,
-    int* backX, int* backY,
-    int* numBackClipRects, drm_clip_rect_t ** pBackClipRects )
+PUBLIC Bool
+XF86DRIGetDrawableInfo(Display * dpy, int screen, Drawable drawable,
+                       unsigned int *index, unsigned int *stamp,
+                       int *X, int *Y, int *W, int *H,
+                       int *numClipRects, drm_clip_rect_t ** pClipRects,
+                       int *backX, int *backY,
+                       int *numBackClipRects,
+                       drm_clip_rect_t ** pBackClipRects)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIGetDrawableInfoReply rep;
-    xXF86DRIGetDrawableInfoReq *req;
-    int total_rects;
-
-    TRACE("GetDrawableInfo...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIGetDrawableInfo, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIGetDrawableInfo;
-    req->screen = screen;
-    req->drawable = drawable;
-
-    if (!_XReply(dpy, (xReply *)&rep, 1, xFalse)) 
-    {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("GetDrawableInfo... return False");
-	return False;
-    }
-    *index = rep.drawableTableIndex;
-    *stamp = rep.drawableTableStamp;
-    *X = (int)rep.drawableX;
-    *Y = (int)rep.drawableY;
-    *W = (int)rep.drawableWidth;
-    *H = (int)rep.drawableHeight;
-    *numClipRects = rep.numClipRects;
-    total_rects = *numClipRects;
-
-    *backX = rep.backX;
-    *backY = rep.backY;
-    *numBackClipRects = rep.numBackClipRects;
-    total_rects += *numBackClipRects;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIGetDrawableInfoReply rep;
+   xXF86DRIGetDrawableInfoReq *req;
+   int total_rects;
+
+   TRACE("GetDrawableInfo...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIGetDrawableInfo, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIGetDrawableInfo;
+   req->screen = screen;
+   req->drawable = drawable;
+
+   if (!_XReply(dpy, (xReply *) & rep, 1, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("GetDrawableInfo... return False");
+      return False;
+   }
+   *index = rep.drawableTableIndex;
+   *stamp = rep.drawableTableStamp;
+   *X = (int) rep.drawableX;
+   *Y = (int) rep.drawableY;
+   *W = (int) rep.drawableWidth;
+   *H = (int) rep.drawableHeight;
+   *numClipRects = rep.numClipRects;
+   total_rects = *numClipRects;
+
+   *backX = rep.backX;
+   *backY = rep.backY;
+   *numBackClipRects = rep.numBackClipRects;
+   total_rects += *numBackClipRects;
 
 #if 0
-    /* Because of the fix in Xserver/GL/dri/xf86dri.c, this check breaks
-     * backwards compatibility (Because of the >> 2 shift) but the fix
-     * enables multi-threaded apps to work.
-     */
-    if (rep.length !=  ((((SIZEOF(xXF86DRIGetDrawableInfoReply) - 
-		       SIZEOF(xGenericReply) + 
-		       total_rects * sizeof(drm_clip_rect_t)) + 3) & ~3) >> 2)) {
-        _XEatData(dpy, rep.length);
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("GetDrawableInfo... return False");
-        return False;
-    }
+   /* Because of the fix in Xserver/GL/dri/xf86dri.c, this check breaks
+    * backwards compatibility (Because of the >> 2 shift) but the fix
+    * enables multi-threaded apps to work.
+    */
+   if (rep.length != ((((SIZEOF(xXF86DRIGetDrawableInfoReply) -
+                         SIZEOF(xGenericReply) +
+                         total_rects * sizeof(drm_clip_rect_t)) +
+                        3) & ~3) >> 2)) {
+      _XEatData(dpy, rep.length);
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("GetDrawableInfo... return False");
+      return False;
+   }
 #endif
 
-    if (*numClipRects) {
-       int len = sizeof(drm_clip_rect_t) * (*numClipRects);
-
-       *pClipRects = (drm_clip_rect_t *)Xcalloc(len, 1);
-       if (*pClipRects) 
-	  _XRead(dpy, (char*)*pClipRects, len);
-    } else {
-        *pClipRects = NULL;
-    }
-
-    if (*numBackClipRects) {
-       int len = sizeof(drm_clip_rect_t) * (*numBackClipRects);
-
-       *pBackClipRects = (drm_clip_rect_t *)Xcalloc(len, 1);
-       if (*pBackClipRects) 
-	  _XRead(dpy, (char*)*pBackClipRects, len);
-    } else {
-        *pBackClipRects = NULL;
-    }
-
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("GetDrawableInfo... return True");
-    return True;
+   if (*numClipRects) {
+      int len = sizeof(drm_clip_rect_t) * (*numClipRects);
+
+      *pClipRects = (drm_clip_rect_t *) Xcalloc(len, 1);
+      if (*pClipRects)
+         _XRead(dpy, (char *) *pClipRects, len);
+   }
+   else {
+      *pClipRects = NULL;
+   }
+
+   if (*numBackClipRects) {
+      int len = sizeof(drm_clip_rect_t) * (*numBackClipRects);
+
+      *pBackClipRects = (drm_clip_rect_t *) Xcalloc(len, 1);
+      if (*pBackClipRects)
+         _XRead(dpy, (char *) *pBackClipRects, len);
+   }
+   else {
+      *pBackClipRects = NULL;
+   }
+
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("GetDrawableInfo... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIGetDeviceInfo(dpy, screen, hFrameBuffer, 
-	fbOrigin, fbSize, fbStride, devPrivateSize, pDevPrivate)
-    Display* dpy;
-    int screen;
-    drm_handle_t * hFrameBuffer;
-    int* fbOrigin;
-    int* fbSize;
-    int* fbStride;
-    int* devPrivateSize;
-    void** pDevPrivate;
+PUBLIC Bool
+XF86DRIGetDeviceInfo(Display * dpy, int screen, drm_handle_t * hFrameBuffer,
+                     int *fbOrigin, int *fbSize, int *fbStride,
+                     int *devPrivateSize, void **pDevPrivate)
 {
-    XExtDisplayInfo *info = find_display (dpy);
-    xXF86DRIGetDeviceInfoReply rep;
-    xXF86DRIGetDeviceInfoReq *req;
-
-    TRACE("GetDeviceInfo...");
-    XF86DRICheckExtension (dpy, info, False);
-
-    LockDisplay(dpy);
-    GetReq(XF86DRIGetDeviceInfo, req);
-    req->reqType = info->codes->major_opcode;
-    req->driReqType = X_XF86DRIGetDeviceInfo;
-    req->screen = screen;
-    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-        TRACE("GetDeviceInfo... return False");
-	return False;
-    }
-
-    *hFrameBuffer = rep.hFrameBufferLow;
-    if (sizeof(drm_handle_t) == 8) {
-       int shift = 32; /* var to prevent warning on next line */
-       *hFrameBuffer |= ((drm_handle_t) rep.hFrameBufferHigh) << shift;
-    }
-
-    *fbOrigin = rep.framebufferOrigin;
-    *fbSize = rep.framebufferSize;
-    *fbStride = rep.framebufferStride;
-    *devPrivateSize = rep.devPrivateSize;
-
-    if (rep.length) {
-        if (!(*pDevPrivate = (void *)Xcalloc(rep.devPrivateSize, 1))) {
-            _XEatData(dpy, ((rep.devPrivateSize+3) & ~3));
-            UnlockDisplay(dpy);
-            SyncHandle();
-            TRACE("GetDeviceInfo... return False");
-            return False;
-        }
-	_XRead(dpy, (char*)*pDevPrivate, rep.devPrivateSize);
-    } else {
-        *pDevPrivate = NULL;
-    }
-
-    UnlockDisplay(dpy);
-    SyncHandle();
-    TRACE("GetDeviceInfo... return True");
-    return True;
+   XExtDisplayInfo *info = find_display(dpy);
+   xXF86DRIGetDeviceInfoReply rep;
+   xXF86DRIGetDeviceInfoReq *req;
+
+   TRACE("GetDeviceInfo...");
+   XF86DRICheckExtension(dpy, info, False);
+
+   LockDisplay(dpy);
+   GetReq(XF86DRIGetDeviceInfo, req);
+   req->reqType = info->codes->major_opcode;
+   req->driReqType = X_XF86DRIGetDeviceInfo;
+   req->screen = screen;
+   if (!_XReply(dpy, (xReply *) & rep, 0, xFalse)) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      TRACE("GetDeviceInfo... return False");
+      return False;
+   }
+
+   *hFrameBuffer = rep.hFrameBufferLow;
+   if (sizeof(drm_handle_t) == 8) {
+      int shift = 32;           /* var to prevent warning on next line */
+      *hFrameBuffer |= ((drm_handle_t) rep.hFrameBufferHigh) << shift;
+   }
+
+   *fbOrigin = rep.framebufferOrigin;
+   *fbSize = rep.framebufferSize;
+   *fbStride = rep.framebufferStride;
+   *devPrivateSize = rep.devPrivateSize;
+
+   if (rep.length) {
+      if (!(*pDevPrivate = (void *) Xcalloc(rep.devPrivateSize, 1))) {
+         _XEatData(dpy, ((rep.devPrivateSize + 3) & ~3));
+         UnlockDisplay(dpy);
+         SyncHandle();
+         TRACE("GetDeviceInfo... return False");
+         return False;
+      }
+      _XRead(dpy, (char *) *pDevPrivate, rep.devPrivateSize);
+   }
+   else {
+      *pDevPrivate = NULL;
+   }
+
+   UnlockDisplay(dpy);
+   SyncHandle();
+   TRACE("GetDeviceInfo... return True");
+   return True;
 }
 
-PUBLIC Bool XF86DRIOpenFullScreen(dpy, screen, drawable)
-    Display* dpy;
-    int screen;
-    Drawable drawable;
+PUBLIC Bool
+XF86DRIOpenFullScreen(Display * dpy, int screen, Drawable drawable)
 {
-    /* This function and the underlying X protocol are deprecated.
-     */
-    (void) dpy;
-    (void) screen;
-    (void) drawable;
-    return False;
+   /* This function and the underlying X protocol are deprecated.
+    */
+   (void) dpy;
+   (void) screen;
+   (void) drawable;
+   return False;
 }
 
-PUBLIC Bool XF86DRICloseFullScreen(dpy, screen, drawable)
-    Display* dpy;
-    int screen;
-    Drawable drawable;
+PUBLIC Bool
+XF86DRICloseFullScreen(Display * dpy, int screen, Drawable drawable)
 {
-    /* This function and the underlying X protocol are deprecated.
-     */
-    (void) dpy;
-    (void) screen;
-    (void) drawable;
-    return True;
+   /* This function and the underlying X protocol are deprecated.
+    */
+   (void) dpy;
+   (void) screen;
+   (void) drawable;
+   return True;
 }
 
 #endif /* GLX_DIRECT_RENDERING */
diff --git a/src/glx/x11/clientattrib.c b/src/glx/x11/clientattrib.c
index 888f8e3187e..a7dfb534860 100644
--- a/src/glx/x11/clientattrib.c
+++ b/src/glx/x11/clientattrib.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include <assert.h>
 #include "glxclient.h"
@@ -41,100 +36,107 @@
 /*****************************************************************************/
 
 static void
-do_enable_disable(GLenum array, GLboolean val )
+do_enable_disable(GLenum array, GLboolean val)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    unsigned index = 0;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   unsigned index = 0;
 
-    if ( array == GL_TEXTURE_COORD_ARRAY ) {
-	index = __glXGetActiveTextureUnit( state );
-    }
+   if (array == GL_TEXTURE_COORD_ARRAY) {
+      index = __glXGetActiveTextureUnit(state);
+   }
 
-    if ( ! __glXSetArrayEnable( state, array, index, val ) ) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-    }
+   if (!__glXSetArrayEnable(state, array, index, val)) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+   }
 }
 
-void __indirect_glEnableClientState(GLenum array)
+void
+__indirect_glEnableClientState(GLenum array)
 {
-    do_enable_disable( array, GL_TRUE );
+   do_enable_disable(array, GL_TRUE);
 }
 
-void __indirect_glDisableClientState(GLenum array)
+void
+__indirect_glDisableClientState(GLenum array)
 {
-    do_enable_disable( array, GL_FALSE );
+   do_enable_disable(array, GL_FALSE);
 }
 
 /************************************************************************/
 
-void __indirect_glPushClientAttrib(GLuint mask)
+void
+__indirect_glPushClientAttrib(GLuint mask)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    __GLXattribute **spp = gc->attributes.stackPointer, *sp;
-
-    if (spp < &gc->attributes.stack[__GL_CLIENT_ATTRIB_STACK_DEPTH]) {
-	if (!(sp = *spp)) {
-	    sp = (__GLXattribute *)Xmalloc(sizeof(__GLXattribute));
-	    *spp = sp;
-	}
-	sp->mask = mask;
-	gc->attributes.stackPointer = spp + 1;
-	if (mask & GL_CLIENT_PIXEL_STORE_BIT) {
-	    sp->storePack = state->storePack;
-	    sp->storeUnpack = state->storeUnpack;
-	}
-	if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) {
-	    __glXPushArrayState( state );
-	}
-    } else {
-	__glXSetError(gc, GL_STACK_OVERFLOW);
-	return;
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   __GLXattribute **spp = gc->attributes.stackPointer, *sp;
+
+   if (spp < &gc->attributes.stack[__GL_CLIENT_ATTRIB_STACK_DEPTH]) {
+      if (!(sp = *spp)) {
+         sp = (__GLXattribute *) Xmalloc(sizeof(__GLXattribute));
+         *spp = sp;
+      }
+      sp->mask = mask;
+      gc->attributes.stackPointer = spp + 1;
+      if (mask & GL_CLIENT_PIXEL_STORE_BIT) {
+         sp->storePack = state->storePack;
+         sp->storeUnpack = state->storeUnpack;
+      }
+      if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) {
+         __glXPushArrayState(state);
+      }
+   }
+   else {
+      __glXSetError(gc, GL_STACK_OVERFLOW);
+      return;
+   }
 }
 
-void __indirect_glPopClientAttrib(void)
+void
+__indirect_glPopClientAttrib(void)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    __GLXattribute **spp = gc->attributes.stackPointer, *sp;
-    GLuint mask;
-
-    if (spp > &gc->attributes.stack[0]) {
-	--spp;
-	sp = *spp;
-	assert(sp != 0);
-	mask = sp->mask;
-	gc->attributes.stackPointer = spp;
-
-	if (mask & GL_CLIENT_PIXEL_STORE_BIT) {
-	    state->storePack = sp->storePack;
-	    state->storeUnpack = sp->storeUnpack;
-	}
-	if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) {
-	    __glXPopArrayState( state );
-	}
-
-	sp->mask = 0;
-    } else {
-	__glXSetError(gc, GL_STACK_UNDERFLOW);
-	return;
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   __GLXattribute **spp = gc->attributes.stackPointer, *sp;
+   GLuint mask;
+
+   if (spp > &gc->attributes.stack[0]) {
+      --spp;
+      sp = *spp;
+      assert(sp != 0);
+      mask = sp->mask;
+      gc->attributes.stackPointer = spp;
+
+      if (mask & GL_CLIENT_PIXEL_STORE_BIT) {
+         state->storePack = sp->storePack;
+         state->storeUnpack = sp->storeUnpack;
+      }
+      if (mask & GL_CLIENT_VERTEX_ARRAY_BIT) {
+         __glXPopArrayState(state);
+      }
+
+      sp->mask = 0;
+   }
+   else {
+      __glXSetError(gc, GL_STACK_UNDERFLOW);
+      return;
+   }
 }
 
-void __glFreeAttributeState(__GLXcontext *gc)
+void
+__glFreeAttributeState(__GLXcontext * gc)
 {
-    __GLXattribute *sp, **spp;
-
-    for (spp = &gc->attributes.stack[0];
-         spp < &gc->attributes.stack[__GL_CLIENT_ATTRIB_STACK_DEPTH];
-	 spp++) {
-	sp = *spp;
-        if (sp) {
-            XFree((char *)sp);
-        } else {
-            break;
-	}
-    }
+   __GLXattribute *sp, **spp;
+
+   for (spp = &gc->attributes.stack[0];
+        spp < &gc->attributes.stack[__GL_CLIENT_ATTRIB_STACK_DEPTH]; spp++) {
+      sp = *spp;
+      if (sp) {
+         XFree((char *) sp);
+      }
+      else {
+         break;
+      }
+   }
 }
diff --git a/src/glx/x11/compsize.c b/src/glx/x11/compsize.c
index 2d124573ef0..eca572feb3c 100644
--- a/src/glx/x11/compsize.c
+++ b/src/glx/x11/compsize.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include <GL/gl.h>
 #include "indirect_size.h"
@@ -40,151 +35,156 @@
 /*
 ** Return the number of elements per group of a specified format
 */
-GLint __glElementsPerGroup(GLenum format, GLenum type) 
+GLint
+__glElementsPerGroup(GLenum format, GLenum type)
 {
-    /*
+   /*
     ** To make row length computation valid for image extraction,
     ** packed pixel types assume elements per group equals one.
     */
-    switch(type) {
-    case GL_UNSIGNED_BYTE_3_3_2:
-    case GL_UNSIGNED_BYTE_2_3_3_REV:
-    case GL_UNSIGNED_SHORT_5_6_5:
-    case GL_UNSIGNED_SHORT_5_6_5_REV:
-    case GL_UNSIGNED_SHORT_4_4_4_4:
-    case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-    case GL_UNSIGNED_SHORT_5_5_5_1:
-    case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-    case GL_UNSIGNED_SHORT_8_8_APPLE:
-    case GL_UNSIGNED_SHORT_8_8_REV_APPLE:
-    case GL_UNSIGNED_SHORT_15_1_MESA:
-    case GL_UNSIGNED_SHORT_1_15_REV_MESA:
-    case GL_UNSIGNED_INT_8_8_8_8:
-    case GL_UNSIGNED_INT_8_8_8_8_REV:
-    case GL_UNSIGNED_INT_10_10_10_2:
-    case GL_UNSIGNED_INT_2_10_10_10_REV:
-    case GL_UNSIGNED_INT_24_8_NV:
-    case GL_UNSIGNED_INT_24_8_MESA:
-    case GL_UNSIGNED_INT_8_24_REV_MESA:
+   switch (type) {
+   case GL_UNSIGNED_BYTE_3_3_2:
+   case GL_UNSIGNED_BYTE_2_3_3_REV:
+   case GL_UNSIGNED_SHORT_5_6_5:
+   case GL_UNSIGNED_SHORT_5_6_5_REV:
+   case GL_UNSIGNED_SHORT_4_4_4_4:
+   case GL_UNSIGNED_SHORT_4_4_4_4_REV:
+   case GL_UNSIGNED_SHORT_5_5_5_1:
+   case GL_UNSIGNED_SHORT_1_5_5_5_REV:
+   case GL_UNSIGNED_SHORT_8_8_APPLE:
+   case GL_UNSIGNED_SHORT_8_8_REV_APPLE:
+   case GL_UNSIGNED_SHORT_15_1_MESA:
+   case GL_UNSIGNED_SHORT_1_15_REV_MESA:
+   case GL_UNSIGNED_INT_8_8_8_8:
+   case GL_UNSIGNED_INT_8_8_8_8_REV:
+   case GL_UNSIGNED_INT_10_10_10_2:
+   case GL_UNSIGNED_INT_2_10_10_10_REV:
+   case GL_UNSIGNED_INT_24_8_NV:
+   case GL_UNSIGNED_INT_24_8_MESA:
+   case GL_UNSIGNED_INT_8_24_REV_MESA:
       return 1;
-    default:
+   default:
       break;
-    }
+   }
 
-    switch(format) {
-      case GL_RGB:
-      case GL_BGR:
-	return 3;
-      case GL_422_EXT:
-      case GL_422_REV_EXT:
-      case GL_422_AVERAGE_EXT:
-      case GL_422_REV_AVERAGE_EXT:
-      case GL_YCBCR_422_APPLE:
-      case GL_LUMINANCE_ALPHA:
-	return 2;
-      case GL_RGBA:
-      case GL_BGRA:
-      case GL_ABGR_EXT:
-	return 4;
-      case GL_COLOR_INDEX:
-      case GL_STENCIL_INDEX:
-      case GL_DEPTH_COMPONENT:
-      case GL_RED:
-      case GL_GREEN:
-      case GL_BLUE:
-      case GL_ALPHA:
-      case GL_LUMINANCE:
-      case GL_INTENSITY:
-	return 1;
-      default:
-	return 0;
-    }
+   switch (format) {
+   case GL_RGB:
+   case GL_BGR:
+      return 3;
+   case GL_422_EXT:
+   case GL_422_REV_EXT:
+   case GL_422_AVERAGE_EXT:
+   case GL_422_REV_AVERAGE_EXT:
+   case GL_YCBCR_422_APPLE:
+   case GL_LUMINANCE_ALPHA:
+      return 2;
+   case GL_RGBA:
+   case GL_BGRA:
+   case GL_ABGR_EXT:
+      return 4;
+   case GL_COLOR_INDEX:
+   case GL_STENCIL_INDEX:
+   case GL_DEPTH_COMPONENT:
+   case GL_RED:
+   case GL_GREEN:
+   case GL_BLUE:
+   case GL_ALPHA:
+   case GL_LUMINANCE:
+   case GL_INTENSITY:
+      return 1;
+   default:
+      return 0;
+   }
 }
 
 /*
 ** Return the number of bytes per element, based on the element type (other
 ** than GL_BITMAP).
 */
-GLint __glBytesPerElement(GLenum type) 
+GLint
+__glBytesPerElement(GLenum type)
 {
-    switch(type) {
-      case GL_UNSIGNED_SHORT:
-      case GL_SHORT:
-      case GL_UNSIGNED_SHORT_5_6_5:
-      case GL_UNSIGNED_SHORT_5_6_5_REV:
-      case GL_UNSIGNED_SHORT_4_4_4_4:
-      case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-      case GL_UNSIGNED_SHORT_5_5_5_1:
-      case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-      case GL_UNSIGNED_SHORT_8_8_APPLE:
-      case GL_UNSIGNED_SHORT_8_8_REV_APPLE:
-      case GL_UNSIGNED_SHORT_15_1_MESA:
-      case GL_UNSIGNED_SHORT_1_15_REV_MESA:
-	return 2;
-      case GL_UNSIGNED_BYTE:
-      case GL_BYTE:
-      case GL_UNSIGNED_BYTE_3_3_2:
-      case GL_UNSIGNED_BYTE_2_3_3_REV:
-	return 1;
-      case GL_INT:
-      case GL_UNSIGNED_INT:
-      case GL_FLOAT:
-      case GL_UNSIGNED_INT_8_8_8_8:
-      case GL_UNSIGNED_INT_8_8_8_8_REV:
-      case GL_UNSIGNED_INT_10_10_10_2:
-      case GL_UNSIGNED_INT_2_10_10_10_REV:
-      case GL_UNSIGNED_INT_24_8_NV:
-      case GL_UNSIGNED_INT_24_8_MESA:
-      case GL_UNSIGNED_INT_8_24_REV_MESA:
-	return 4;
-      default:
-	return 0;
-    }
+   switch (type) {
+   case GL_UNSIGNED_SHORT:
+   case GL_SHORT:
+   case GL_UNSIGNED_SHORT_5_6_5:
+   case GL_UNSIGNED_SHORT_5_6_5_REV:
+   case GL_UNSIGNED_SHORT_4_4_4_4:
+   case GL_UNSIGNED_SHORT_4_4_4_4_REV:
+   case GL_UNSIGNED_SHORT_5_5_5_1:
+   case GL_UNSIGNED_SHORT_1_5_5_5_REV:
+   case GL_UNSIGNED_SHORT_8_8_APPLE:
+   case GL_UNSIGNED_SHORT_8_8_REV_APPLE:
+   case GL_UNSIGNED_SHORT_15_1_MESA:
+   case GL_UNSIGNED_SHORT_1_15_REV_MESA:
+      return 2;
+   case GL_UNSIGNED_BYTE:
+   case GL_BYTE:
+   case GL_UNSIGNED_BYTE_3_3_2:
+   case GL_UNSIGNED_BYTE_2_3_3_REV:
+      return 1;
+   case GL_INT:
+   case GL_UNSIGNED_INT:
+   case GL_FLOAT:
+   case GL_UNSIGNED_INT_8_8_8_8:
+   case GL_UNSIGNED_INT_8_8_8_8_REV:
+   case GL_UNSIGNED_INT_10_10_10_2:
+   case GL_UNSIGNED_INT_2_10_10_10_REV:
+   case GL_UNSIGNED_INT_24_8_NV:
+   case GL_UNSIGNED_INT_24_8_MESA:
+   case GL_UNSIGNED_INT_8_24_REV_MESA:
+      return 4;
+   default:
+      return 0;
+   }
 }
 
 /*
 ** Compute memory required for internal packed array of data of given type
 ** and format.
 */
-GLint __glImageSize(GLsizei width, GLsizei height, GLsizei depth,
-		    GLenum format, GLenum type, GLenum target)
+GLint
+__glImageSize(GLsizei width, GLsizei height, GLsizei depth,
+              GLenum format, GLenum type, GLenum target)
 {
-    int bytes_per_row;
-    int components;
+   int bytes_per_row;
+   int components;
 
-    switch( target ) {
-    case GL_PROXY_TEXTURE_1D:
-    case GL_PROXY_TEXTURE_2D:
-    case GL_PROXY_TEXTURE_3D:
-    case GL_PROXY_TEXTURE_4D_SGIS:
-    case GL_PROXY_TEXTURE_CUBE_MAP:
-    case GL_PROXY_TEXTURE_RECTANGLE_ARB:
-    case GL_PROXY_HISTOGRAM:
-    case GL_PROXY_COLOR_TABLE:
-    case GL_PROXY_TEXTURE_COLOR_TABLE_SGI:
-    case GL_PROXY_POST_CONVOLUTION_COLOR_TABLE:
-    case GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE:
-    case GL_PROXY_POST_IMAGE_TRANSFORM_COLOR_TABLE_HP:
-	return 0;
-    }
+   switch (target) {
+   case GL_PROXY_TEXTURE_1D:
+   case GL_PROXY_TEXTURE_2D:
+   case GL_PROXY_TEXTURE_3D:
+   case GL_PROXY_TEXTURE_4D_SGIS:
+   case GL_PROXY_TEXTURE_CUBE_MAP:
+   case GL_PROXY_TEXTURE_RECTANGLE_ARB:
+   case GL_PROXY_HISTOGRAM:
+   case GL_PROXY_COLOR_TABLE:
+   case GL_PROXY_TEXTURE_COLOR_TABLE_SGI:
+   case GL_PROXY_POST_CONVOLUTION_COLOR_TABLE:
+   case GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE:
+   case GL_PROXY_POST_IMAGE_TRANSFORM_COLOR_TABLE_HP:
+      return 0;
+   }
 
-    if (width < 0 || height < 0 || depth < 0) {
-	return 0;
-    }
+   if (width < 0 || height < 0 || depth < 0) {
+      return 0;
+   }
 
-    /*
+   /*
     ** Zero is returned if either format or type are invalid.
     */
-    components = __glElementsPerGroup(format,type);
-    if (type == GL_BITMAP) {
-	if (format == GL_COLOR_INDEX || format == GL_STENCIL_INDEX) {
-	    bytes_per_row = (width + 7) >> 3;
-	} else {
-	    return 0;
-	}
-    } else {
-	bytes_per_row = __glBytesPerElement(type) * width;
-    }
+   components = __glElementsPerGroup(format, type);
+   if (type == GL_BITMAP) {
+      if (format == GL_COLOR_INDEX || format == GL_STENCIL_INDEX) {
+         bytes_per_row = (width + 7) >> 3;
+      }
+      else {
+         return 0;
+      }
+   }
+   else {
+      bytes_per_row = __glBytesPerElement(type) * width;
+   }
 
-    return bytes_per_row * height * depth * components;
+   return bytes_per_row * height * depth * components;
 }
diff --git a/src/glx/x11/dri2.c b/src/glx/x11/dri2.c
new file mode 100644
index 00000000000..3b48cd9c1ca
--- /dev/null
+++ b/src/glx/x11/dri2.c
@@ -0,0 +1,301 @@
+/*
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Soft-
+ * ware"), to deal in the Software without restriction, including without
+ * limitation the rights to use, copy, modify, merge, publish, distribute,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, provided that the above copyright
+ * notice(s) and this permission notice appear in all copies of the Soft-
+ * ware and that both the above copyright notice(s) and this permission
+ * notice appear in supporting documentation.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ * ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY
+ * RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN
+ * THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSE-
+ * QUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+ * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFOR-
+ * MANCE OF THIS SOFTWARE.
+ *
+ * Except as contained in this notice, the name of a copyright holder shall
+ * not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization of
+ * the copyright holder.
+ *
+ * Authors:
+ *   Kristian Høgsberg (krh@redhat.com)
+ */
+
+
+#define NEED_REPLIES
+#include <X11/Xlibint.h>
+#include <X11/extensions/Xext.h>
+#include <X11/extensions/extutil.h>
+#include <X11/extensions/dri2proto.h>
+#include "xf86drm.h"
+#include "dri2.h"
+
+static char dri2ExtensionName[] = DRI2_NAME;
+static XExtensionInfo *dri2Info;
+static XEXT_GENERATE_CLOSE_DISPLAY (DRI2CloseDisplay, dri2Info)
+static /* const */ XExtensionHooks dri2ExtensionHooks = {
+    NULL,				/* create_gc */
+    NULL,				/* copy_gc */
+    NULL,				/* flush_gc */
+    NULL,				/* free_gc */
+    NULL,				/* create_font */
+    NULL,				/* free_font */
+    DRI2CloseDisplay,			/* close_display */
+    NULL,				/* wire_to_event */
+    NULL,				/* event_to_wire */
+    NULL,				/* error */
+    NULL,				/* error_string */
+};
+
+static XEXT_GENERATE_FIND_DISPLAY (DRI2FindDisplay, dri2Info, 
+				   dri2ExtensionName, 
+				   &dri2ExtensionHooks, 
+				   0, NULL)
+
+Bool DRI2QueryExtension(Display *dpy, int *eventBase, int *errorBase)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+
+    if (XextHasExtension(info)) {
+	*eventBase = info->codes->first_event;
+	*errorBase = info->codes->first_error;
+	return True;
+    }
+
+    return False;
+}
+
+Bool DRI2QueryVersion(Display *dpy, int *major, int *minor)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay (dpy);
+    xDRI2QueryVersionReply rep;
+    xDRI2QueryVersionReq *req;
+
+    XextCheckExtension (dpy, info, dri2ExtensionName, False);
+
+    LockDisplay(dpy);
+    GetReq(DRI2QueryVersion, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2QueryVersion;
+    req->majorVersion = DRI2_MAJOR;
+    req->minorVersion = DRI2_MINOR;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+    *major = rep.majorVersion;
+    *minor = rep.minorVersion;
+    UnlockDisplay(dpy);
+    SyncHandle();
+
+    return True;
+}
+
+Bool DRI2Connect(Display *dpy, XID window,
+		 char **driverName, char **deviceName)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2ConnectReply rep;
+    xDRI2ConnectReq *req;
+
+    XextCheckExtension (dpy, info, dri2ExtensionName, False);
+
+    LockDisplay(dpy);
+    GetReq(DRI2Connect, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2Connect;
+    req->window = window;
+    req->driverType = DRI2DriverDRI;
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+
+    if (rep.driverNameLength == 0 && rep.deviceNameLength == 0) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+
+    *driverName = Xmalloc(rep.driverNameLength + 1);
+    if (*driverName == NULL) {
+	_XEatData(dpy, 
+		  ((rep.driverNameLength + 3) & ~3) +
+		  ((rep.deviceNameLength + 3) & ~3));
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+    _XReadPad(dpy, *driverName, rep.driverNameLength);
+    (*driverName)[rep.driverNameLength] = '\0';
+
+    *deviceName = Xmalloc(rep.deviceNameLength + 1);
+    if (*deviceName == NULL) {
+	Xfree(*driverName);
+	_XEatData(dpy, ((rep.deviceNameLength + 3) & ~3));
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+    _XReadPad(dpy, *deviceName, rep.deviceNameLength);
+    (*deviceName)[rep.deviceNameLength] = '\0';
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+
+    return True;
+}
+
+Bool DRI2Authenticate(Display *dpy, XID window, drm_magic_t magic)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2AuthenticateReq *req;
+    xDRI2AuthenticateReply rep;
+
+    XextCheckExtension (dpy, info, dri2ExtensionName, False);
+
+    LockDisplay(dpy);
+    GetReq(DRI2Authenticate, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2Authenticate;
+    req->window = window;
+    req->magic = magic;
+
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return False;
+    }
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+
+    return rep.authenticated;
+}
+
+void DRI2CreateDrawable(Display *dpy, XID drawable)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2CreateDrawableReq *req;
+
+    XextSimpleCheckExtension (dpy, info, dri2ExtensionName);
+
+    LockDisplay(dpy);
+    GetReq(DRI2CreateDrawable, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2CreateDrawable;
+    req->drawable = drawable;
+    UnlockDisplay(dpy);
+    SyncHandle();
+}
+
+void DRI2DestroyDrawable(Display *dpy, XID drawable)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2DestroyDrawableReq *req;
+
+    XextSimpleCheckExtension (dpy, info, dri2ExtensionName);
+
+    XSync(dpy, False);
+
+    LockDisplay(dpy);
+    GetReq(DRI2DestroyDrawable, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2DestroyDrawable;
+    req->drawable = drawable;
+    UnlockDisplay(dpy);
+    SyncHandle();
+}
+
+DRI2Buffer *DRI2GetBuffers(Display *dpy, XID drawable,
+			   int *width, int *height,
+			   unsigned int *attachments, int count,
+			   int *outCount)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2GetBuffersReply rep;
+    xDRI2GetBuffersReq *req;
+    DRI2Buffer *buffers;
+    xDRI2Buffer repBuffer;
+    CARD32 *p;
+    int i;
+
+    XextCheckExtension (dpy, info, dri2ExtensionName, False);
+
+    LockDisplay(dpy);
+    GetReqExtra(DRI2GetBuffers, count * 4, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2GetBuffers;
+    req->drawable = drawable;
+    req->count = count;
+    p = (CARD32 *) &req[1];
+    for (i = 0; i < count; i++)
+	p[i] = attachments[i];
+
+    if (!_XReply(dpy, (xReply *)&rep, 0, xFalse)) {
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return NULL;
+    }
+
+    *width = rep.width;
+    *height = rep.height;
+    *outCount = rep.count;
+
+    buffers = Xmalloc(count * sizeof buffers[0]);
+    if (buffers == NULL) {
+	_XEatData(dpy, rep.count * sizeof repBuffer);
+	UnlockDisplay(dpy);
+	SyncHandle();
+	return NULL;
+    }
+
+    for (i = 0; i < rep.count; i++) {
+	_XReadPad(dpy, (char *) &repBuffer, sizeof repBuffer);
+	buffers[i].attachment = repBuffer.attachment;
+	buffers[i].name = repBuffer.name;
+	buffers[i].pitch = repBuffer.pitch;
+	buffers[i].cpp = repBuffer.cpp;
+	buffers[i].flags = repBuffer.flags;
+    }
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+
+    return buffers;
+}
+
+void DRI2CopyRegion(Display *dpy, XID drawable, XserverRegion region,
+		    CARD32 dest, CARD32 src)
+{
+    XExtDisplayInfo *info = DRI2FindDisplay(dpy);
+    xDRI2CopyRegionReq *req;
+    xDRI2CopyRegionReply rep;
+
+    XextSimpleCheckExtension (dpy, info, dri2ExtensionName);
+
+    LockDisplay(dpy);
+    GetReq(DRI2CopyRegion, req);
+    req->reqType = info->codes->major_opcode;
+    req->dri2ReqType = X_DRI2CopyRegion;
+    req->drawable = drawable;
+    req->region = region;
+    req->dest = dest;
+    req->src = src;
+
+    _XReply(dpy, (xReply *)&rep, 0, xFalse);
+
+    UnlockDisplay(dpy);
+    SyncHandle();
+}
diff --git a/src/glx/x11/dri2.h b/src/glx/x11/dri2.h
new file mode 100644
index 00000000000..356c6bcb55d
--- /dev/null
+++ b/src/glx/x11/dri2.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright © 2007,2008 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Soft-
+ * ware"), to deal in the Software without restriction, including without
+ * limitation the rights to use, copy, modify, merge, publish, distribute,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, provided that the above copyright
+ * notice(s) and this permission notice appear in all copies of the Soft-
+ * ware and that both the above copyright notice(s) and this permission
+ * notice appear in supporting documentation.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ * ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY
+ * RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN
+ * THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSE-
+ * QUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+ * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFOR-
+ * MANCE OF THIS SOFTWARE.
+ *
+ * Except as contained in this notice, the name of a copyright holder shall
+ * not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization of
+ * the copyright holder.
+ *
+ * Authors:
+ *   Kristian Høgsberg (krh@redhat.com)
+ */
+
+#ifndef _DRI2_H_
+#define _DRI2_H_
+
+#include <X11/extensions/Xfixes.h>
+#include <X11/extensions/dri2tokens.h>
+
+typedef struct {
+    unsigned int attachment;
+    unsigned int name;
+    unsigned int pitch;
+    unsigned int cpp;
+    unsigned int flags;
+} DRI2Buffer;
+
+extern Bool
+DRI2QueryExtension(Display *display, int *eventBase, int *errorBase);
+extern Bool
+DRI2QueryVersion(Display *display, int *major, int *minor);
+extern Bool
+DRI2Connect(Display *display, XID window,
+	    char **driverName, char **deviceName);
+extern Bool
+DRI2Authenticate(Display *display, XID window, drm_magic_t magic);
+extern void
+DRI2CreateDrawable(Display *display, XID drawable);
+extern void
+DRI2DestroyDrawable(Display *display, XID handle);
+extern DRI2Buffer *
+DRI2GetBuffers(Display *dpy, XID drawable,
+	       int *width, int *height,
+	       unsigned int *attachments, int count,
+	       int *outCount);
+
+extern void
+DRI2CopyRegion(Display *dpy, XID drawable, XserverRegion region,
+	       CARD32 dest, CARD32 src);
+
+#endif
diff --git a/src/glx/x11/dri2_glx.c b/src/glx/x11/dri2_glx.c
new file mode 100644
index 00000000000..2bee67780bd
--- /dev/null
+++ b/src/glx/x11/dri2_glx.c
@@ -0,0 +1,392 @@
+/*
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Soft-
+ * ware"), to deal in the Software without restriction, including without
+ * limitation the rights to use, copy, modify, merge, publish, distribute,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, provided that the above copyright
+ * notice(s) and this permission notice appear in all copies of the Soft-
+ * ware and that both the above copyright notice(s) and this permission
+ * notice appear in supporting documentation.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ * ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY
+ * RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN
+ * THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSE-
+ * QUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+ * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFOR-
+ * MANCE OF THIS SOFTWARE.
+ *
+ * Except as contained in this notice, the name of a copyright holder shall
+ * not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization of
+ * the copyright holder.
+ *
+ * Authors:
+ *   Kristian Høgsberg (krh@redhat.com)
+ */
+
+#ifdef GLX_DIRECT_RENDERING
+
+#include <X11/Xlib.h>
+#include <X11/extensions/Xfixes.h>
+#include <X11/extensions/Xdamage.h>
+#include "glxclient.h"
+#include "glcontextmodes.h"
+#include "xf86dri.h"
+#include <dlfcn.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/mman.h>
+#include "xf86drm.h"
+#include "dri2.h"
+#include "dri_common.h"
+
+typedef struct __GLXDRIdisplayPrivateRec __GLXDRIdisplayPrivate;
+typedef struct __GLXDRIcontextPrivateRec __GLXDRIcontextPrivate;
+typedef struct __GLXDRIdrawablePrivateRec __GLXDRIdrawablePrivate;
+
+struct __GLXDRIdisplayPrivateRec {
+    __GLXDRIdisplay base;
+
+    /*
+    ** XFree86-DRI version information
+    */
+    int driMajor;
+    int driMinor;
+    int driPatch;
+};
+
+struct __GLXDRIcontextPrivateRec {
+    __GLXDRIcontext base;
+    __DRIcontext *driContext;
+    __GLXscreenConfigs *psc;
+};
+
+struct __GLXDRIdrawablePrivateRec {
+    __GLXDRIdrawable base;
+    __DRIbuffer buffers[5];
+    int bufferCount;
+    int width, height;
+};
+
+static void dri2DestroyContext(__GLXDRIcontext *context,
+			      __GLXscreenConfigs *psc, Display *dpy)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+    const __DRIcoreExtension *core = pcp->psc->core;
+
+    (*core->destroyContext)(pcp->driContext);
+
+    Xfree(pcp);
+}
+
+static Bool dri2BindContext(__GLXDRIcontext *context,
+			   __GLXDRIdrawable *draw, __GLXDRIdrawable *read)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+    const __DRIcoreExtension *core = pcp->psc->core;
+
+    return (*core->bindContext)(pcp->driContext,
+				draw->driDrawable,
+				read->driDrawable);
+}
+
+static void dri2UnbindContext(__GLXDRIcontext *context)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+    const __DRIcoreExtension *core = pcp->psc->core;
+
+    (*core->unbindContext)(pcp->driContext);
+}
+
+static __GLXDRIcontext *dri2CreateContext(__GLXscreenConfigs *psc,
+					 const __GLcontextModes *mode,
+					 GLXContext gc,
+					 GLXContext shareList, int renderType)
+{
+    __GLXDRIcontextPrivate *pcp, *pcp_shared;
+    __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) mode;
+    __DRIcontext *shared = NULL;
+
+    if (shareList) {
+	pcp_shared = (__GLXDRIcontextPrivate *) shareList->driContext;
+	shared = pcp_shared->driContext;
+    }
+
+    pcp = Xmalloc(sizeof *pcp);
+    if (pcp == NULL)
+	return NULL;
+
+    pcp->psc = psc;
+    pcp->driContext = 
+	(*psc->dri2->createNewContext)(psc->__driScreen,
+				       config->driConfig, shared, pcp);
+    gc->__driContext = pcp->driContext;
+
+    if (pcp->driContext == NULL) {
+	Xfree(pcp);
+	return NULL;
+    }
+
+    pcp->base.destroyContext = dri2DestroyContext;
+    pcp->base.bindContext = dri2BindContext;
+    pcp->base.unbindContext = dri2UnbindContext;
+
+    return &pcp->base;
+}
+
+static void dri2DestroyDrawable(__GLXDRIdrawable *pdraw)
+{
+    const __DRIcoreExtension *core = pdraw->psc->core;
+
+    (*core->destroyDrawable)(pdraw->driDrawable);
+    DRI2DestroyDrawable(pdraw->psc->dpy, pdraw->drawable);
+    Xfree(pdraw);
+}
+
+static __GLXDRIdrawable *dri2CreateDrawable(__GLXscreenConfigs *psc,
+					    XID xDrawable,
+					    GLXDrawable drawable,
+					    const __GLcontextModes *modes)
+{
+    __GLXDRIdrawablePrivate *pdraw;
+    __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) modes;
+
+    pdraw = Xmalloc(sizeof(*pdraw));
+    if (!pdraw)
+	return NULL;
+
+    pdraw->base.destroyDrawable = dri2DestroyDrawable;
+    pdraw->base.xDrawable = xDrawable;
+    pdraw->base.drawable = drawable;
+    pdraw->base.psc = psc;
+
+    DRI2CreateDrawable(psc->dpy, xDrawable);
+
+    /* Create a new drawable */
+    pdraw->base.driDrawable =
+	(*psc->dri2->createNewDrawable)(psc->__driScreen,
+					config->driConfig, pdraw);
+
+    if (!pdraw->base.driDrawable) {
+	DRI2DestroyDrawable(psc->dpy, drawable);
+	Xfree(pdraw);
+	return NULL;
+    }
+
+    return &pdraw->base;
+}
+
+static void dri2CopySubBuffer(__GLXDRIdrawable *pdraw,
+			      int x, int y, int width, int height)
+{
+    __GLXDRIdrawablePrivate *priv = (__GLXDRIdrawablePrivate *) pdraw;
+    XRectangle xrect;
+    XserverRegion region;
+
+    xrect.x = x;
+    xrect.y = priv->height - y - height;
+    xrect.width = width;
+    xrect.height = height;
+
+    region = XFixesCreateRegion(pdraw->psc->dpy, &xrect, 1);
+    DRI2CopyRegion(pdraw->psc->dpy, pdraw->drawable, region,
+		   DRI2BufferFrontLeft, DRI2BufferBackLeft);
+    XFixesDestroyRegion(pdraw->psc->dpy, region);
+}
+
+static void dri2SwapBuffers(__GLXDRIdrawable *pdraw)
+{
+    __GLXDRIdrawablePrivate *priv = (__GLXDRIdrawablePrivate *) pdraw;
+
+    dri2CopySubBuffer(pdraw, 0, 0, priv->width, priv->height);
+}
+
+static void dri2DestroyScreen(__GLXscreenConfigs *psc)
+{
+    /* Free the direct rendering per screen data */
+    (*psc->core->destroyScreen)(psc->__driScreen);
+    close(psc->fd);
+    psc->__driScreen = NULL;
+}
+
+static __DRIbuffer *
+dri2GetBuffers(__DRIdrawable *driDrawable,
+	       int *width, int *height,
+	       unsigned int *attachments, int count,
+	       int *out_count, void *loaderPrivate)
+{
+    __GLXDRIdrawablePrivate *pdraw = loaderPrivate;
+    DRI2Buffer *buffers;
+    int i;
+
+    buffers = DRI2GetBuffers(pdraw->base.psc->dpy, pdraw->base.xDrawable,
+			     width, height, attachments, count, out_count);
+    if (buffers == NULL)
+       return NULL;
+
+    pdraw->width = *width;
+    pdraw->height = *height;
+
+    /* This assumes the DRI2 buffer attachment tokens matches the
+     * __DRIbuffer tokens. */
+    for (i = 0; i < *out_count; i++) {
+	pdraw->buffers[i].attachment = buffers[i].attachment;
+	pdraw->buffers[i].name = buffers[i].name;
+	pdraw->buffers[i].pitch = buffers[i].pitch;
+	pdraw->buffers[i].cpp = buffers[i].cpp;
+	pdraw->buffers[i].flags = buffers[i].flags;
+    }
+
+    Xfree(buffers);
+
+    return pdraw->buffers;
+}
+
+static const __DRIdri2LoaderExtension dri2LoaderExtension = {
+    { __DRI_DRI2_LOADER, __DRI_DRI2_LOADER_VERSION },
+    dri2GetBuffers,
+};
+
+static const __DRIextension *loader_extensions[] = {
+    &dri2LoaderExtension.base,
+    &systemTimeExtension.base,
+    NULL
+};
+
+static __GLXDRIscreen *dri2CreateScreen(__GLXscreenConfigs *psc, int screen,
+					__GLXdisplayPrivate *priv)
+{
+    const __DRIconfig **driver_configs;
+    const __DRIextension **extensions;
+    __GLXDRIscreen *psp;
+    char *driverName, *deviceName;
+    drm_magic_t magic;
+    int i;
+
+    psp = Xmalloc(sizeof *psp);
+    if (psp == NULL)
+	return NULL;
+
+    /* Initialize per screen dynamic client GLX extensions */
+    psc->ext_list_first_time = GL_TRUE;
+
+    if (!DRI2Connect(psc->dpy, RootWindow(psc->dpy, screen),
+		     &driverName, &deviceName))
+	return NULL;
+
+    psc->driver = driOpenDriver(driverName);
+    if (psc->driver == NULL)
+	goto handle_error;
+
+    extensions = dlsym(psc->driver, __DRI_DRIVER_EXTENSIONS);
+    if (extensions == NULL) {
+	ErrorMessageF("driver exports no extensions (%s)\n", dlerror());
+	goto handle_error;
+    }
+    
+    for (i = 0; extensions[i]; i++) {
+	if (strcmp(extensions[i]->name, __DRI_CORE) == 0)
+	    psc->core = (__DRIcoreExtension *) extensions[i];
+	if (strcmp(extensions[i]->name, __DRI_DRI2) == 0)
+	    psc->dri2 = (__DRIdri2Extension *) extensions[i];
+    }
+
+    if (psc->core == NULL || psc->dri2 == NULL) {
+	ErrorMessageF("core dri or dri2 extension not found\n");
+	goto handle_error;
+    }
+
+    psc->fd = open(deviceName, O_RDWR);
+    if (psc->fd < 0) {
+	ErrorMessageF("failed to open drm device: %s\n", strerror(errno));
+	return NULL;
+    }
+
+    if (drmGetMagic(psc->fd, &magic))
+	return NULL;
+
+    if (!DRI2Authenticate(psc->dpy, RootWindow(psc->dpy, screen), magic))
+	return NULL;
+
+    psc->__driScreen = 
+	psc->dri2->createNewScreen(screen, psc->fd,
+				   loader_extensions, &driver_configs, psc);
+    if (psc->__driScreen == NULL) {
+	ErrorMessageF("failed to create dri screen\n");
+	return NULL;
+    }
+
+    driBindExtensions(psc, 1);
+
+    psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
+    psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
+
+    psp->destroyScreen = dri2DestroyScreen;
+    psp->createContext = dri2CreateContext;
+    psp->createDrawable = dri2CreateDrawable;
+    psp->swapBuffers = dri2SwapBuffers;
+
+    /* DRI2 suports SubBuffer through DRI2CopyRegion, so it's always
+     * available.*/
+    psp->copySubBuffer = dri2CopySubBuffer;
+    __glXEnableDirectExtension(psc, "GLX_MESA_copy_sub_buffer");
+
+    Xfree(driverName);
+    Xfree(deviceName);
+
+    return psp;
+
+ handle_error:
+    Xfree(driverName);
+    Xfree(deviceName);
+
+    /* FIXME: clean up here */
+
+    return NULL;
+}
+
+/* Called from __glXFreeDisplayPrivate.
+ */
+static void dri2DestroyDisplay(__GLXDRIdisplay *dpy)
+{
+    Xfree(dpy);
+}
+
+/*
+ * Allocate, initialize and return a __DRIdisplayPrivate object.
+ * This is called from __glXInitialize() when we are given a new
+ * display pointer.
+ */
+_X_HIDDEN __GLXDRIdisplay *dri2CreateDisplay(Display *dpy)
+{
+    __GLXDRIdisplayPrivate *pdp;
+    int eventBase, errorBase;
+
+    if (!DRI2QueryExtension(dpy, &eventBase, &errorBase))
+	return NULL;
+
+    pdp = Xmalloc(sizeof *pdp);
+    if (pdp == NULL)
+	return NULL;
+
+    if (!DRI2QueryVersion(dpy, &pdp->driMajor, &pdp->driMinor)) {
+	Xfree(pdp);
+	return NULL;
+    }
+
+    pdp->driPatch = 0;
+
+    pdp->base.destroyDisplay = dri2DestroyDisplay;
+    pdp->base.createScreen = dri2CreateScreen;
+
+    return &pdp->base;
+}
+
+#endif /* GLX_DIRECT_RENDERING */
diff --git a/src/glx/x11/dri_common.c b/src/glx/x11/dri_common.c
new file mode 100644
index 00000000000..4fda649e59a
--- /dev/null
+++ b/src/glx/x11/dri_common.c
@@ -0,0 +1,399 @@
+/*
+ * Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Soft-
+ * ware"), to deal in the Software without restriction, including without
+ * limitation the rights to use, copy, modify, merge, publish, distribute,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, provided that the above copyright
+ * notice(s) and this permission notice appear in all copies of the Soft-
+ * ware and that both the above copyright notice(s) and this permission
+ * notice appear in supporting documentation.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ * ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY
+ * RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN
+ * THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSE-
+ * QUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+ * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFOR-
+ * MANCE OF THIS SOFTWARE.
+ *
+ * Except as contained in this notice, the name of a copyright holder shall
+ * not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization of
+ * the copyright holder.
+ *
+ * Authors:
+ *   Kevin E. Martin <kevin@precisioninsight.com>
+ *   Brian Paul <brian@precisioninsight.com>
+ *   Kristian Høgsberg (krh@redhat.com)
+ */
+
+#ifdef GLX_DIRECT_RENDERING
+
+#include <unistd.h>
+#include <dlfcn.h>
+#include <stdarg.h>
+#include "glxclient.h"
+#include "glcontextmodes.h"
+#include "dri_common.h"
+
+#ifndef RTLD_NOW
+#define RTLD_NOW 0
+#endif
+#ifndef RTLD_GLOBAL
+#define RTLD_GLOBAL 0
+#endif
+
+_X_HIDDEN void InfoMessageF(const char *f, ...)
+{
+    va_list args;
+    const char *env;
+
+    if ((env = getenv("LIBGL_DEBUG")) && strstr(env, "verbose")) {
+	fprintf(stderr, "libGL: ");
+	va_start(args, f);
+	vfprintf(stderr, f, args);
+	va_end(args);
+    }
+}
+
+/**
+ * Print error to stderr, unless LIBGL_DEBUG=="quiet".
+ */
+_X_HIDDEN void ErrorMessageF(const char *f, ...)
+{
+    va_list args;
+    const char *env;
+
+    if ((env = getenv("LIBGL_DEBUG")) && !strstr(env, "quiet")) {
+	fprintf(stderr, "libGL error: ");
+	va_start(args, f);
+	vfprintf(stderr, f, args);
+	va_end(args);
+    }
+}
+
+#ifndef DEFAULT_DRIVER_DIR
+/* this is normally defined in Mesa/configs/default with DRI_DRIVER_SEARCH_PATH */
+#define DEFAULT_DRIVER_DIR "/usr/local/lib/dri"
+#endif
+
+/**
+ * Try to \c dlopen the named driver.
+ *
+ * This function adds the "_dri.so" suffix to the driver name and searches the
+ * directories specified by the \c LIBGL_DRIVERS_PATH environment variable in
+ * order to find the driver.
+ *
+ * \param driverName - a name like "tdfx", "i810", "mga", etc.
+ *
+ * \returns
+ * A handle from \c dlopen, or \c NULL if driver file not found.
+ */
+_X_HIDDEN void *driOpenDriver(const char *driverName)
+{
+   void *glhandle, *handle;
+   const char *libPaths, *p, *next;
+   char realDriverName[200];
+   int len;
+
+   /* Attempt to make sure libGL symbols will be visible to the driver */
+   glhandle = dlopen("libGL.so.1", RTLD_NOW | RTLD_GLOBAL);
+
+   libPaths = NULL;
+   if (geteuid() == getuid()) {
+      /* don't allow setuid apps to use LIBGL_DRIVERS_PATH */
+      libPaths = getenv("LIBGL_DRIVERS_PATH");
+      if (!libPaths)
+         libPaths = getenv("LIBGL_DRIVERS_DIR"); /* deprecated */
+   }
+   if (libPaths == NULL)
+       libPaths = DEFAULT_DRIVER_DIR;
+
+   handle = NULL;
+   for (p = libPaths; *p; p = next) {
+       next = strchr(p, ':');
+       if (next == NULL) {
+	   len = strlen(p);
+	   next = p + len;
+       } else {
+	   len = next - p;
+	   next++;
+       }
+
+#ifdef GLX_USE_TLS
+      snprintf(realDriverName, sizeof realDriverName,
+	       "%.*s/tls/%s_dri.so", len, p, driverName);
+      InfoMessageF("OpenDriver: trying %s\n", realDriverName);
+      handle = dlopen(realDriverName, RTLD_NOW | RTLD_GLOBAL);
+#endif
+
+      if ( handle == NULL ) {
+	 snprintf(realDriverName, sizeof realDriverName,
+		  "%.*s/%s_dri.so", len, p, driverName);
+	 InfoMessageF("OpenDriver: trying %s\n", realDriverName);
+	 handle = dlopen(realDriverName, RTLD_NOW | RTLD_GLOBAL);
+      }
+
+      if ( handle != NULL )
+	  break;
+      else
+	 ErrorMessageF("dlopen %s failed (%s)\n", realDriverName, dlerror());
+   }
+
+   if (!handle)
+      ErrorMessageF("unable to load driver: %s_dri.so\n", driverName);
+
+   if (glhandle)
+      dlclose(glhandle);
+
+   return handle;
+}
+
+_X_HIDDEN const __DRIsystemTimeExtension systemTimeExtension = {
+    { __DRI_SYSTEM_TIME, __DRI_SYSTEM_TIME_VERSION },
+    __glXGetUST,
+    __driGetMscRateOML
+};
+
+#define __ATTRIB(attrib, field) \
+    { attrib, offsetof(__GLcontextModes, field) }
+
+static const struct { unsigned int attrib, offset; } attribMap[] = {
+    __ATTRIB(__DRI_ATTRIB_BUFFER_SIZE,			rgbBits),
+    __ATTRIB(__DRI_ATTRIB_LEVEL,			level),
+    __ATTRIB(__DRI_ATTRIB_RED_SIZE,			redBits),
+    __ATTRIB(__DRI_ATTRIB_GREEN_SIZE,			greenBits),
+    __ATTRIB(__DRI_ATTRIB_BLUE_SIZE,			blueBits),
+    __ATTRIB(__DRI_ATTRIB_ALPHA_SIZE,			alphaBits),
+    __ATTRIB(__DRI_ATTRIB_DEPTH_SIZE,			depthBits),
+    __ATTRIB(__DRI_ATTRIB_STENCIL_SIZE,			stencilBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_RED_SIZE,		accumRedBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_GREEN_SIZE,		accumGreenBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_BLUE_SIZE,		accumBlueBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_ALPHA_SIZE,		accumAlphaBits),
+    __ATTRIB(__DRI_ATTRIB_SAMPLE_BUFFERS,		sampleBuffers),
+    __ATTRIB(__DRI_ATTRIB_SAMPLES,			samples),
+    __ATTRIB(__DRI_ATTRIB_DOUBLE_BUFFER,		doubleBufferMode),
+    __ATTRIB(__DRI_ATTRIB_STEREO,			stereoMode),
+    __ATTRIB(__DRI_ATTRIB_AUX_BUFFERS,			numAuxBuffers),
+#if 0
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_TYPE,		transparentPixel),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_INDEX_VALUE,	transparentIndex),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_RED_VALUE,	transparentRed),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_GREEN_VALUE,	transparentGreen),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_BLUE_VALUE,	transparentBlue),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_ALPHA_VALUE,	transparentAlpha),
+    __ATTRIB(__DRI_ATTRIB_RED_MASK,			redMask),
+    __ATTRIB(__DRI_ATTRIB_GREEN_MASK,			greenMask),
+    __ATTRIB(__DRI_ATTRIB_BLUE_MASK,			blueMask),
+    __ATTRIB(__DRI_ATTRIB_ALPHA_MASK,			alphaMask),
+#endif
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_WIDTH,		maxPbufferWidth),
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_HEIGHT,		maxPbufferHeight),
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_PIXELS,		maxPbufferPixels),
+    __ATTRIB(__DRI_ATTRIB_OPTIMAL_PBUFFER_WIDTH,	optimalPbufferWidth),
+    __ATTRIB(__DRI_ATTRIB_OPTIMAL_PBUFFER_HEIGHT,	optimalPbufferHeight),
+#if 0
+    __ATTRIB(__DRI_ATTRIB_SWAP_METHOD,			swapMethod),
+#endif
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_RGB,		bindToTextureRgb),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_RGBA,		bindToTextureRgba),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_MIPMAP_TEXTURE,	bindToMipmapTexture),
+    __ATTRIB(__DRI_ATTRIB_YINVERTED,			yInverted),
+};
+
+#define ARRAY_SIZE(a) (sizeof (a) / sizeof ((a)[0]))
+
+static int
+scalarEqual(__GLcontextModes *mode, unsigned int attrib, unsigned int value)
+{
+    unsigned int glxValue;
+    int i;
+
+    for (i = 0; i < ARRAY_SIZE(attribMap); i++)
+	if (attribMap[i].attrib == attrib) {
+	    glxValue = *(unsigned int *) ((char *) mode + attribMap[i].offset);
+	    return glxValue == GLX_DONT_CARE || glxValue == value;
+	}
+
+    return GL_TRUE; /* Is a non-existing attribute equal to value? */
+}
+
+static int
+driConfigEqual(const __DRIcoreExtension *core,
+	       __GLcontextModes *modes, const __DRIconfig *driConfig)
+{
+    unsigned int attrib, value, glxValue;
+    int i;
+
+    i = 0;
+    while (core->indexConfigAttrib(driConfig, i++, &attrib, &value)) {
+	switch (attrib) {
+	case __DRI_ATTRIB_RENDER_TYPE:
+	    glxValue = 0;
+	    if (value & __DRI_ATTRIB_RGBA_BIT) {
+		glxValue |= GLX_RGBA_BIT;
+	    } else if (value & __DRI_ATTRIB_COLOR_INDEX_BIT) {
+		glxValue |= GLX_COLOR_INDEX_BIT;
+	    }
+	    if (glxValue != modes->renderType)
+		return GL_FALSE;
+	    break;
+
+	case __DRI_ATTRIB_CONFIG_CAVEAT:
+	    if (value & __DRI_ATTRIB_NON_CONFORMANT_CONFIG)
+		glxValue = GLX_NON_CONFORMANT_CONFIG;
+	    else if (value & __DRI_ATTRIB_SLOW_BIT)
+		glxValue = GLX_SLOW_CONFIG;
+	    else
+		glxValue = GLX_NONE;
+	    if (glxValue != modes->visualRating)
+		return GL_FALSE;
+	    break;
+
+	case __DRI_ATTRIB_BIND_TO_TEXTURE_TARGETS:
+	    glxValue = 0;
+	    if (value & __DRI_ATTRIB_TEXTURE_1D_BIT)
+		glxValue |= GLX_TEXTURE_1D_BIT_EXT;
+	    if (value & __DRI_ATTRIB_TEXTURE_2D_BIT)
+		glxValue |= GLX_TEXTURE_2D_BIT_EXT;
+	    if (value & __DRI_ATTRIB_TEXTURE_RECTANGLE_BIT)
+		glxValue |= GLX_TEXTURE_RECTANGLE_BIT_EXT;
+	    if (modes->bindToTextureTargets != GLX_DONT_CARE &&
+		glxValue != modes->bindToTextureTargets)
+		return GL_FALSE;
+	    break;	
+
+	default:
+	    if (!scalarEqual(modes, attrib, value))
+		return GL_FALSE;
+	}
+    }
+
+    return GL_TRUE;
+}
+
+static __GLcontextModes *
+createDriMode(const __DRIcoreExtension *core,
+	      __GLcontextModes *modes, const __DRIconfig **driConfigs)
+{
+    __GLXDRIconfigPrivate *config;
+    int i;
+
+    for (i = 0; driConfigs[i]; i++) {
+	if (driConfigEqual(core, modes, driConfigs[i]))
+	    break;
+    }
+
+    if (driConfigs[i] == NULL)
+	return NULL;
+
+    config = Xmalloc(sizeof *config);
+    if (config == NULL)
+	return NULL;
+
+    config->modes = *modes;
+    config->driConfig = driConfigs[i];
+
+    return &config->modes;
+}
+
+_X_HIDDEN __GLcontextModes *
+driConvertConfigs(const __DRIcoreExtension *core,
+		  __GLcontextModes *modes, const __DRIconfig **configs)
+{
+    __GLcontextModes head, *tail, *m;
+
+    tail = &head;
+    head.next = NULL;
+    for (m = modes; m; m = m->next) {
+	tail->next = createDriMode(core, m, configs);
+	if (tail->next == NULL) {
+	    /* no matching dri config for m */
+	    continue;
+	}
+
+
+	tail = tail->next;
+    }
+
+    _gl_context_modes_destroy(modes);
+
+    return head.next;
+}
+
+_X_HIDDEN void
+driBindExtensions(__GLXscreenConfigs *psc, int dri2)
+{
+    const __DRIextension **extensions;
+    int i;
+
+    extensions = psc->core->getExtensions(psc->__driScreen);
+
+    for (i = 0; extensions[i]; i++) {
+#ifdef __DRI_COPY_SUB_BUFFER
+	if (strcmp(extensions[i]->name, __DRI_COPY_SUB_BUFFER) == 0) {
+	    psc->driCopySubBuffer = (__DRIcopySubBufferExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_MESA_copy_sub_buffer");
+	}
+#endif
+
+#ifdef __DRI_SWAP_CONTROL
+	if (strcmp(extensions[i]->name, __DRI_SWAP_CONTROL) == 0) {
+	    psc->swapControl = (__DRIswapControlExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_SGI_swap_control");
+	    __glXEnableDirectExtension(psc, "GLX_MESA_swap_control");
+	}
+#endif
+
+#ifdef __DRI_ALLOCATE
+	if (strcmp(extensions[i]->name, __DRI_ALLOCATE) == 0) {
+	    psc->allocate = (__DRIallocateExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_MESA_allocate_memory");
+	}
+#endif
+
+#ifdef __DRI_FRAME_TRACKING
+	if (strcmp(extensions[i]->name, __DRI_FRAME_TRACKING) == 0) {
+	    psc->frameTracking = (__DRIframeTrackingExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_MESA_swap_frame_usage");
+	}
+#endif
+
+#ifdef __DRI_MEDIA_STREAM_COUNTER
+	if (strcmp(extensions[i]->name, __DRI_MEDIA_STREAM_COUNTER) == 0) {
+	    psc->msc = (__DRImediaStreamCounterExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_SGI_video_sync");
+	}
+#endif
+
+#ifdef __DRI_SWAP_BUFFER_COUNTER
+	/* No driver supports this at this time and the extension is
+	 * not defined in dri_interface.h.  Will enable
+	 * GLX_OML_sync_control if implemented. */
+#endif
+
+#ifdef __DRI_READ_DRAWABLE
+	if (strcmp(extensions[i]->name, __DRI_READ_DRAWABLE) == 0) {
+	    __glXEnableDirectExtension(psc, "GLX_SGI_make_current_read");
+	}
+#endif
+
+#ifdef __DRI_TEX_BUFFER
+	if ((strcmp(extensions[i]->name, __DRI_TEX_BUFFER) == 0) && dri2) {
+	    psc->texBuffer = (__DRItexBufferExtension *) extensions[i];
+	    __glXEnableDirectExtension(psc, "GLX_EXT_texture_from_pixmap");
+	}
+#endif
+
+	/* Ignore unknown extensions */
+    }
+}
+
+#endif /* GLX_DIRECT_RENDERING */
diff --git a/src/glx/x11/dri_common.h b/src/glx/x11/dri_common.h
new file mode 100644
index 00000000000..61ac9c64160
--- /dev/null
+++ b/src/glx/x11/dri_common.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Soft-
+ * ware"), to deal in the Software without restriction, including without
+ * limitation the rights to use, copy, modify, merge, publish, distribute,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, provided that the above copyright
+ * notice(s) and this permission notice appear in all copies of the Soft-
+ * ware and that both the above copyright notice(s) and this permission
+ * notice appear in supporting documentation.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABIL-
+ * ITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY
+ * RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN
+ * THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSE-
+ * QUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+ * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFOR-
+ * MANCE OF THIS SOFTWARE.
+ *
+ * Except as contained in this notice, the name of a copyright holder shall
+ * not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization of
+ * the copyright holder.
+ *
+ * Authors:
+ *   Kevin E. Martin <kevin@precisioninsight.com>
+ *   Brian Paul <brian@precisioninsight.com>
+ *   Kristian Høgsberg (krh@redhat.com)
+ */
+
+#ifndef _DRI_COMMON_H
+#define _DRI_COMMON_H
+
+typedef struct __GLXDRIconfigPrivateRec __GLXDRIconfigPrivate;
+
+struct __GLXDRIconfigPrivateRec
+{
+   __GLcontextModes modes;
+   const __DRIconfig *driConfig;
+};
+
+extern __GLcontextModes *driConvertConfigs(const __DRIcoreExtension * core,
+                                           __GLcontextModes * modes,
+                                           const __DRIconfig ** configs);
+
+extern const __DRIsystemTimeExtension systemTimeExtension;
+
+extern void InfoMessageF(const char *f, ...);
+
+extern void ErrorMessageF(const char *f, ...);
+
+extern void *driOpenDriver(const char *driverName);
+
+extern void driBindExtensions(__GLXscreenConfigs * psc, int dri2);
+
+#endif /* _DRI_COMMON_H */
diff --git a/src/glx/x11/dri_glx.c b/src/glx/x11/dri_glx.c
index e637c96ac9b..44724d2c7d0 100644
--- a/src/glx/x11/dri_glx.c
+++ b/src/glx/x11/dri_glx.c
@@ -34,294 +34,46 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #ifdef GLX_DIRECT_RENDERING
 
-#include <unistd.h>
-#include <X11/Xlibint.h>
-#include <X11/extensions/Xext.h>
-#include <X11/extensions/extutil.h>
-#include "glheader.h"
+#include <X11/Xlib.h>
+#include <X11/extensions/Xfixes.h>
+#include <X11/extensions/Xdamage.h>
 #include "glxclient.h"
+#include "glcontextmodes.h"
 #include "xf86dri.h"
 #include "sarea.h"
-#include <stdio.h>
 #include <dlfcn.h>
-#include "dri_glx.h"
 #include <sys/types.h>
-#include <stdarg.h>
-
-#ifndef RTLD_NOW
-#define RTLD_NOW 0
-#endif
-#ifndef RTLD_GLOBAL
-#define RTLD_GLOBAL 0
-#endif
-
-
-#ifndef DEFAULT_DRIVER_DIR
-/* this is normally defined in Mesa/configs/default with DRI_DRIVER_SEARCH_PATH */
-#define DEFAULT_DRIVER_DIR "/usr/X11R6/lib/modules/dri"
-#endif
-
-static __DRIdriver *Drivers = NULL;
-
-
-/** If non-zero, prefer an "egl_foo_dri.so" driver over "foo_dri.so" */
-static int PreferEGL = 0;
-
-
-/**
- * This may be called by libEGL.so to tell libGL to prefer "egl_" drivers
- * over regular DRI drivers".
- */
-void __glXPreferEGL(int state)
-{
-   PreferEGL = state;
-}
-
-
-/*
- * printf wrappers
- */
-
-static void InfoMessageF(const char *f, ...)
-{
-    va_list args;
-    const char *env;
-
-    if ((env = getenv("LIBGL_DEBUG")) && strstr(env, "verbose")) {
-	fprintf(stderr, "libGL: ");
-	va_start(args, f);
-	vfprintf(stderr, f, args);
-	va_end(args);
-    }
-}
-
-/**
- * Print error to stderr, unless LIBGL_DEBUG=="quiet".
- */
-static void ErrorMessageF(const char *f, ...)
-{
-    va_list args;
-    const char *env;
-
-    if ((env = getenv("LIBGL_DEBUG")) && !strstr(env, "quiet")) {
-	fprintf(stderr, "libGL error: ");
-	va_start(args, f);
-	vfprintf(stderr, f, args);
-	va_end(args);
-    }
-}
-
-
-/**
- * Extract the ith directory path out of a colon-separated list of paths.  No
- * more than \c dirLen characters, including the terminating \c NUL, will be
- * written to \c dir.
- *
- * \param index  Index of path to extract (starting at zero)
- * \param paths  The colon-separated list of paths
- * \param dirLen Maximum length of result to store in \c dir
- * \param dir    Buffer to hold the extracted directory path
- *
- * \returns
- * The number of characters that would have been written to \c dir had there
- * been enough room.  This does not include the terminating \c NUL.  When
- * extraction fails, zero will be returned.
- * 
- * \todo
- * It seems like this function could be rewritten to use \c strchr.
- */
-static size_t
-ExtractDir(int index, const char *paths, int dirLen, char *dir)
-{
-   int i, len;
-   const char *start, *end;
-
-   /* find ith colon */
-   start = paths;
-   i = 0;
-   while (i < index) {
-      if (*start == ':') {
-         i++;
-         start++;
-      }
-      else if (*start == 0) {
-         /* end of string and couldn't find ith colon */
-         dir[0] = 0;
-         return 0;
-      }
-      else {
-         start++;
-      }
-   }
-
-   while (*start == ':')
-      start++;
-
-   /* find next colon, or end of string */
-   end = start + 1;
-   while (*end != ':' && *end != 0) {
-      end++;
-   }
-
-   /* copy string between <start> and <end> into result string */
-   len = end - start;
-   if (len > dirLen - 1)
-      len = dirLen - 1;
-   strncpy(dir, start, len);
-   dir[len] = 0;
-
-   return( end - start );
-}
-
-
-/**
- * Try to dlopen() a driver.
- * \param libDir  the directory to search
- * \param driverName  the name of the driver (such as "r300").
- * \param tls  if true, try to find TLS version of driver
- * \param preferEGL  if true, try to find EGL-specific driver
- * \return  handle from dlopen(), will be NULL if fails.
- */
-static void *
-try_open(const char *libDir, const char *driverName,
-         GLboolean tls, GLboolean preferEGL)
-{
-   const char *tlsDir = tls ? "/tls/" : "";
-   const char *eglPrefix = preferEGL ? "egl_" : "";
-   char fullPathName[200];
-   void *handle = NULL;
-
-   snprintf(fullPathName, 200, "%s%s/%s%s_dri.so",
-            libDir, tlsDir, eglPrefix, driverName);
-   InfoMessageF("OpenDriver: trying %s\n", fullPathName);
-   handle = dlopen(fullPathName, RTLD_NOW | RTLD_GLOBAL);
-
-   return handle;
-}
-
-
-
-/**
- * Versioned name of the expected \c __driCreateNewScreen function.
- * 
- * The version of the last incompatible loader/driver inteface change is
- * appended to the name of the \c __driCreateNewScreen function.  This
- * prevents loaders from trying to load drivers that are too old.
- * 
- * \todo
- * Create a macro or something so that this is automatically updated.
- */
-static const char createNewScreenName[] = "__driCreateNewScreen_20050727";
-
-
-/**
- * Try to \c dlopen the named driver.
- *
- * This function adds the "_dri.so" suffix to the driver name and searches the
- * directories specified by the \c LIBGL_DRIVERS_PATH environment variable in
- * order to find the driver.
- *
- * \param driverName - a name like "tdfx", "i810", "mga", etc.
- *
- * \returns
- * A handle from \c dlopen, or \c NULL if driver file not found.
- */
-static __DRIdriver *OpenDriver(const char *driverName)
-{
-   void *glhandle = NULL;
-   char *libPaths = NULL;
-   char libDir[1000];
-   int i;
-   __DRIdriver *driver;
-
-   /* First, search Drivers list to see if we've already opened this driver */
-   for (driver = Drivers; driver; driver = driver->next) {
-      if (strcmp(driver->name, driverName) == 0) {
-         /* found it */
-         return driver;
-      }
-   }
-
-   /* Attempt to make sure libGL symbols will be visible to the driver */
-   glhandle = dlopen("libGL.so.1", RTLD_NOW | RTLD_GLOBAL);
-
-   if (geteuid() == getuid()) {
-      /* don't allow setuid apps to use LIBGL_DRIVERS_PATH */
-      libPaths = getenv("LIBGL_DRIVERS_PATH");
-      if (!libPaths)
-         libPaths = getenv("LIBGL_DRIVERS_DIR"); /* deprecated */
-   }
-   if (!libPaths)
-      libPaths = DEFAULT_DRIVER_DIR;
-
-   for ( i = 0 ; ExtractDir(i, libPaths, 1000, libDir) != 0 ; i++ ) {
-      void *handle = NULL;
-
-      if (PreferEGL)
-         handle = try_open(libDir, driverName, GL_FALSE, GL_TRUE);
-#ifdef GLX_USE_TLS
-      if (!handle)
-         handle = try_open(libDir, driverName, GL_TRUE, GL_FALSE);
-#endif
-      if (!handle)
-         handle = try_open(libDir, driverName, GL_FALSE, GL_FALSE);
-
-      if ( handle != NULL ) {
-         /* allocate __DRIdriver struct */
-         driver = (__DRIdriver *) Xmalloc(sizeof(__DRIdriver));
-         if (!driver)
-            break; /* out of memory! */
-         /* init the struct */
-         driver->name = __glXstrdup(driverName);
-         if (!driver->name) {
-            Xfree(driver);
-            driver = NULL;
-            break; /* out of memory! */
-         }
-
-         driver->createNewScreenFunc = (PFNCREATENEWSCREENFUNC)
-            dlsym(handle, createNewScreenName);
-
-         if ( driver->createNewScreenFunc == NULL ) {
-            /* If the driver doesn't have this symbol then something's
-             * really, really wrong.
-             */
-            ErrorMessageF("%s not defined in %s_dri.so!\n"
-			  "Your driver may be too old for this libGL.\n",
-			  createNewScreenName, driverName);
-            Xfree(driver);
-            driver = NULL;
-            dlclose(handle);
-            continue;
-         }
-         driver->handle = handle;
-         /* put at head of linked list */
-         driver->next = Drivers;
-         Drivers = driver;
-         break;
-      }
-      else {
-	 ErrorMessageF("Unable to find/open driver for %s (%s)\n",
-                       driverName, dlerror());
-      }
-   }
-
-   if (!driver)
-      ErrorMessageF("unable to load driver: %s_dri.so\n", driverName);
-
-   if (glhandle)
-      dlclose(glhandle);
-
-   return driver;
-}
-
+#include <sys/mman.h>
+#include "xf86drm.h"
+#include "dri_common.h"
+
+typedef struct __GLXDRIdisplayPrivateRec __GLXDRIdisplayPrivate;
+typedef struct __GLXDRIcontextPrivateRec __GLXDRIcontextPrivate;
+
+struct __GLXDRIdisplayPrivateRec {
+    __GLXDRIdisplay base;
+
+    /*
+    ** XFree86-DRI version information
+    */
+    int driMajor;
+    int driMinor;
+    int driPatch;
+};
+
+struct __GLXDRIcontextPrivateRec {
+    __GLXDRIcontext base;
+    __DRIcontext *driContext;
+    XID hwContextID;
+    __GLXscreenConfigs *psc;
+};
 
 /*
  * Given a display pointer and screen number, determine the name of
  * the DRI driver for the screen. (I.e. "r128", "tdfx", etc).
  * Return True for success, False for failure.
  */
-static Bool GetDriverName(Display *dpy, int scrNum, char **driverName)
+static Bool driGetDriverName(Display *dpy, int scrNum, char **driverName)
 {
    int directCapable;
    Bool b;
@@ -351,25 +103,6 @@ static Bool GetDriverName(Display *dpy, int scrNum, char **driverName)
    return True;
 }
 
-
-/*
- * Given a display pointer and screen number, return a __DRIdriver handle.
- * Return NULL if anything goes wrong.
- */
-__DRIdriver *driGetDriver(Display *dpy, int scrNum)
-{
-   char *driverName;
-   if (GetDriverName(dpy, scrNum, &driverName)) {
-      __DRIdriver *ret;
-      ret = OpenDriver(driverName);
-      if (driverName)
-     	 Xfree(driverName);
-      return ret;
-   }
-   return NULL;
-}
-
-
 /*
  * Exported function for querying the DRI driver for a given screen.
  *
@@ -379,7 +112,7 @@ __DRIdriver *driGetDriver(Display *dpy, int scrNum)
 PUBLIC const char *glXGetScreenDriver (Display *dpy, int scrNum) {
    static char ret[32];
    char *driverName;
-   if (GetDriverName(dpy, scrNum, &driverName)) {
+   if (driGetDriverName(dpy, scrNum, &driverName)) {
       int len;
       if (!driverName)
 	 return NULL;
@@ -393,7 +126,6 @@ PUBLIC const char *glXGetScreenDriver (Display *dpy, int scrNum) {
    return NULL;
 }
 
-
 /*
  * Exported function for obtaining a driver's option list (UTF-8 encoded XML).
  *
@@ -405,71 +137,545 @@ PUBLIC const char *glXGetScreenDriver (Display *dpy, int scrNum) {
  *
  * Note: The driver remains opened after this function returns.
  */
-PUBLIC const char *glXGetDriverConfig (const char *driverName) {
-   __DRIdriver *driver = OpenDriver (driverName);
-   if (driver)
-      return dlsym (driver->handle, "__driConfigOptions");
+PUBLIC const char *glXGetDriverConfig (const char *driverName)
+{
+   void *handle = driOpenDriver (driverName);
+   if (handle)
+      return dlsym (handle, "__driConfigOptions");
    else
       return NULL;
 }
 
+#ifdef XDAMAGE_1_1_INTERFACE
 
-/* Called from __glXFreeDisplayPrivate.
+static GLboolean has_damage_post(Display *dpy)
+{
+    static GLboolean inited = GL_FALSE;
+    static GLboolean has_damage;
+
+    if (!inited) {
+	int major, minor;
+
+	if (XDamageQueryVersion(dpy, &major, &minor) &&
+	    major == 1 && minor >= 1)
+	{
+	    has_damage = GL_TRUE;
+	} else {
+	    has_damage = GL_FALSE;
+	}
+	inited = GL_TRUE;
+    }
+
+    return has_damage;
+}
+
+static void __glXReportDamage(__DRIdrawable *driDraw,
+			      int x, int y,
+			      drm_clip_rect_t *rects, int num_rects,
+			      GLboolean front_buffer,
+			      void *loaderPrivate)
+{
+    XRectangle *xrects;
+    XserverRegion region;
+    int i;
+    int x_off, y_off;
+    __GLXDRIdrawable *glxDraw = loaderPrivate;
+    __GLXscreenConfigs *psc = glxDraw->psc;
+    Display *dpy = psc->dpy;
+    Drawable drawable;
+
+    if (!has_damage_post(dpy))
+	return;
+
+    if (front_buffer) {
+	x_off = x;
+	y_off = y;
+	drawable = RootWindow(dpy, psc->scr);
+    } else{
+	x_off = 0;
+	y_off = 0;
+	drawable = glxDraw->xDrawable;
+    }
+
+    xrects = malloc(sizeof(XRectangle) * num_rects);
+    if (xrects == NULL)
+	return;
+
+    for (i = 0; i < num_rects; i++) {
+	xrects[i].x = rects[i].x1 + x_off;
+	xrects[i].y = rects[i].y1 + y_off;
+	xrects[i].width = rects[i].x2 - rects[i].x1;
+	xrects[i].height = rects[i].y2 - rects[i].y1;
+    }
+    region = XFixesCreateRegion(dpy, xrects, num_rects);
+    free(xrects);
+    XDamageAdd(dpy, drawable, region);
+    XFixesDestroyRegion(dpy, region);
+}
+
+static const __DRIdamageExtension damageExtension = {
+    { __DRI_DAMAGE, __DRI_DAMAGE_VERSION },
+    __glXReportDamage,
+};
+
+#endif
+
+static GLboolean
+__glXDRIGetDrawableInfo(__DRIdrawable *drawable,
+			unsigned int *index, unsigned int *stamp, 
+			int *X, int *Y, int *W, int *H,
+			int *numClipRects, drm_clip_rect_t ** pClipRects,
+			int *backX, int *backY,
+			int *numBackClipRects, drm_clip_rect_t **pBackClipRects,
+			void *loaderPrivate)
+{
+    __GLXDRIdrawable *glxDraw = loaderPrivate;
+    __GLXscreenConfigs *psc = glxDraw->psc;
+    Display *dpy = psc->dpy;
+
+    return XF86DRIGetDrawableInfo(dpy, psc->scr, glxDraw->drawable,
+				  index, stamp, X, Y, W, H,
+				  numClipRects, pClipRects,
+				  backX, backY,
+				  numBackClipRects, pBackClipRects);
+}
+
+static const __DRIgetDrawableInfoExtension getDrawableInfoExtension = {
+    { __DRI_GET_DRAWABLE_INFO, __DRI_GET_DRAWABLE_INFO_VERSION },
+    __glXDRIGetDrawableInfo
+};
+
+static const __DRIextension *loader_extensions[] = {
+    &systemTimeExtension.base,
+    &getDrawableInfoExtension.base,
+#ifdef XDAMAGE_1_1_INTERFACE
+    &damageExtension.base,
+#endif
+    NULL
+};
+
+#ifndef GLX_USE_APPLEGL
+
+/**
+ * Perform the required libGL-side initialization and call the client-side
+ * driver's \c __driCreateNewScreen function.
+ * 
+ * \param dpy    Display pointer.
+ * \param scrn   Screen number on the display.
+ * \param psc    DRI screen information.
+ * \param driDpy DRI display information.
+ * \param createNewScreen  Pointer to the client-side driver's
+ *               \c __driCreateNewScreen function.
+ * \returns A pointer to the \c __DRIscreenPrivate structure returned by
+ *          the client-side driver on success, or \c NULL on failure.
  */
-static void driDestroyDisplay(Display *dpy, void *private)
+static void *
+CallCreateNewScreen(Display *dpy, int scrn, __GLXscreenConfigs *psc,
+		    __GLXDRIdisplayPrivate * driDpy)
 {
-    __DRIdisplayPrivate *pdpyp = (__DRIdisplayPrivate *)private;
-
-    if (pdpyp) {
-        const int numScreens = ScreenCount(dpy);
-        int i;
-        for (i = 0; i < numScreens; i++) {
-	   if (pdpyp->libraryHandles[i]) {
-	      __DRIdriver *driver, *prev;
-
-	      /* Remove driver from Drivers list */
-	      for (prev = NULL, driver = Drivers; driver;
-		   prev = driver, driver = driver->next) {
-		 if (driver->handle == pdpyp->libraryHandles[i]) {
-		    if (prev)
-		       prev->next = driver->next;
-		    else
-		       Drivers = driver->next;
-
-		    Xfree((void *) driver->name);
-		    Xfree(driver);
-		    break;
-		 }
-	      }
-
-	      dlclose(pdpyp->libraryHandles[i]);
-	   }
-        }
-        Xfree(pdpyp->libraryHandles);
-	Xfree(pdpyp);
+    void *psp = NULL;
+    drm_handle_t hSAREA;
+    drmAddress pSAREA = MAP_FAILED;
+    char *BusID;
+    __DRIversion   ddx_version;
+    __DRIversion   dri_version;
+    __DRIversion   drm_version;
+    __DRIframebuffer  framebuffer;
+    int   fd = -1;
+    int   status;
+
+    drm_magic_t magic;
+    drmVersionPtr version;
+    int newlyopened;
+    char *driverName;
+    drm_handle_t  hFB;
+    int        junk;
+    const __DRIconfig **driver_configs;
+
+    /* DRI protocol version. */
+    dri_version.major = driDpy->driMajor;
+    dri_version.minor = driDpy->driMinor;
+    dri_version.patch = driDpy->driPatch;
+
+    framebuffer.base = MAP_FAILED;
+    framebuffer.dev_priv = NULL;
+
+    if (!XF86DRIOpenConnection(dpy, scrn, &hSAREA, &BusID)) {
+	ErrorMessageF("XF86DRIOpenConnection failed\n");
+	goto handle_error;
     }
+
+    fd = drmOpenOnce(NULL, BusID, &newlyopened);
+
+    Xfree(BusID); /* No longer needed */
+
+    if (fd < 0) {
+	ErrorMessageF("drmOpenOnce failed (%s)\n", strerror(-fd));
+	goto handle_error;
+    }
+
+    if (drmGetMagic(fd, &magic)) {
+	ErrorMessageF("drmGetMagic failed\n");
+	goto handle_error;
+    }
+
+    version = drmGetVersion(fd);
+    if (version) {
+	drm_version.major = version->version_major;
+	drm_version.minor = version->version_minor;
+	drm_version.patch = version->version_patchlevel;
+	drmFreeVersion(version);
+    }
+    else {
+	drm_version.major = -1;
+	drm_version.minor = -1;
+	drm_version.patch = -1;
+    }
+
+    if (newlyopened && !XF86DRIAuthConnection(dpy, scrn, magic)) {
+	ErrorMessageF("XF86DRIAuthConnection failed\n");
+	goto handle_error;
+    }
+
+    /* Get device name (like "tdfx") and the ddx version numbers.
+     * We'll check the version in each DRI driver's "createNewScreen"
+     * function. */
+    if (!XF86DRIGetClientDriverName(dpy, scrn,
+				    &ddx_version.major,
+				    &ddx_version.minor,
+				    &ddx_version.patch,
+				    &driverName)) {
+	ErrorMessageF("XF86DRIGetClientDriverName failed\n");
+	goto handle_error;
+    }
+
+    Xfree(driverName); /* No longer needed. */
+
+    /*
+     * Get device-specific info.  pDevPriv will point to a struct
+     * (such as DRIRADEONRec in xfree86/driver/ati/radeon_dri.h) that
+     * has information about the screen size, depth, pitch, ancilliary
+     * buffers, DRM mmap handles, etc.
+     */
+    if (!XF86DRIGetDeviceInfo(dpy, scrn, &hFB, &junk,
+			      &framebuffer.size, &framebuffer.stride,
+			      &framebuffer.dev_priv_size, &framebuffer.dev_priv)) {
+	ErrorMessageF("XF86DRIGetDeviceInfo failed");
+	goto handle_error;
+    }
+
+    framebuffer.width = DisplayWidth(dpy, scrn);
+    framebuffer.height = DisplayHeight(dpy, scrn);
+
+    /* Map the framebuffer region. */
+    status = drmMap(fd, hFB, framebuffer.size, 
+		    (drmAddressPtr)&framebuffer.base);
+    if (status != 0) {
+	ErrorMessageF("drmMap of framebuffer failed (%s)", strerror(-status));
+	goto handle_error;
+    }
+
+    /* Map the SAREA region.  Further mmap regions may be setup in
+     * each DRI driver's "createNewScreen" function.
+     */
+    status = drmMap(fd, hSAREA, SAREA_MAX, &pSAREA);
+    if (status != 0) {
+	ErrorMessageF("drmMap of SAREA failed (%s)", strerror(-status));
+	goto handle_error;
+    }
+
+    psp = (*psc->legacy->createNewScreen)(scrn,
+					  &ddx_version,
+					  &dri_version,
+					  &drm_version,
+					  &framebuffer,
+					  pSAREA,
+					  fd,
+					  loader_extensions,
+					  &driver_configs,
+					  psc);
+
+    if (psp == NULL) {
+	ErrorMessageF("Calling driver entry point failed");
+	goto handle_error;
+    }
+
+    psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
+    psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
+
+    return psp;
+
+ handle_error:
+    if (pSAREA != MAP_FAILED)
+	drmUnmap(pSAREA, SAREA_MAX);
+
+    if (framebuffer.base != MAP_FAILED)
+	drmUnmap((drmAddress)framebuffer.base, framebuffer.size);
+
+    if (framebuffer.dev_priv != NULL)
+	Xfree(framebuffer.dev_priv);
+
+    if (fd >= 0)
+	drmCloseOnce(fd);
+
+    XF86DRICloseConnection(dpy, scrn);
+
+    ErrorMessageF("reverting to software direct rendering\n");
+
+    return NULL;
+}
+
+#else /* !GLX_USE_APPLEGL */
+
+static void *
+CallCreateNewScreen(Display *dpy, int scrn, __GLXscreenConfigs *psc,
+		    __GLXDRIdisplayPrivate * driDpy)
+{
+    return NULL;
+}
+
+#endif /* !GLX_USE_APPLEGL */
+
+static void driDestroyContext(__GLXDRIcontext *context,
+			      __GLXscreenConfigs *psc, Display *dpy)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+			
+    (*psc->core->destroyContext)(pcp->driContext);
+
+    XF86DRIDestroyContext(psc->dpy, psc->scr, pcp->hwContextID);
+    Xfree(pcp);
+}
+
+static Bool driBindContext(__GLXDRIcontext *context,
+			   __GLXDRIdrawable *draw, __GLXDRIdrawable *read)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+    const __DRIcoreExtension *core = pcp->psc->core;
+
+    return (*core->bindContext)(pcp->driContext,
+				draw->driDrawable,
+				read->driDrawable);
+}
+
+static void driUnbindContext(__GLXDRIcontext *context)
+{
+    __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+    const __DRIcoreExtension *core = pcp->psc->core;
+
+    (*core->unbindContext)(pcp->driContext);
+}
+
+static __GLXDRIcontext *driCreateContext(__GLXscreenConfigs *psc,
+					 const __GLcontextModes *mode,
+					 GLXContext gc,
+					 GLXContext shareList, int renderType)
+{
+    __GLXDRIcontextPrivate *pcp, *pcp_shared;
+    drm_context_t hwContext;
+    __DRIcontext *shared = NULL;
+    __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) mode;
+
+    if (!psc || !psc->driScreen)
+	return NULL;
+
+    if (shareList) {
+	pcp_shared = (__GLXDRIcontextPrivate *) shareList->driContext;
+	shared = pcp_shared->driContext;
+    }
+
+    pcp = Xmalloc(sizeof *pcp);
+    if (pcp == NULL)
+	return NULL;
+
+    pcp->psc = psc;
+    if (!XF86DRICreateContextWithConfig(psc->dpy, psc->scr,
+					mode->visualID,
+					&pcp->hwContextID, &hwContext)) {
+	Xfree(pcp);
+	return NULL;
+    }
+
+    pcp->driContext =
+        (*psc->legacy->createNewContext)(psc->__driScreen,
+					 config->driConfig,
+					 renderType,
+					 shared,
+					 hwContext,
+					 pcp);
+    if (pcp->driContext == NULL) {
+	XF86DRIDestroyContext(psc->dpy, psc->scr, pcp->hwContextID);
+	Xfree(pcp);
+	return NULL;
+    }
+
+    pcp->base.destroyContext = driDestroyContext;
+    pcp->base.bindContext = driBindContext;
+    pcp->base.unbindContext = driUnbindContext;
+
+    return &pcp->base;
+}
+
+static void driDestroyDrawable(__GLXDRIdrawable *pdraw)
+{
+    __GLXscreenConfigs *psc = pdraw->psc;
+
+    (*psc->core->destroyDrawable)(pdraw->driDrawable);
+    XF86DRIDestroyDrawable(psc->dpy, psc->scr, pdraw->drawable);
+    Xfree(pdraw);
+}
+
+static __GLXDRIdrawable *driCreateDrawable(__GLXscreenConfigs *psc,
+					   XID xDrawable,
+					   GLXDrawable drawable,
+					   const __GLcontextModes *modes)
+{
+    __GLXDRIdrawable *pdraw;
+    drm_drawable_t hwDrawable;
+    void *empty_attribute_list = NULL;
+    __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) modes;
+
+    /* Old dri can't handle GLX 1.3+ drawable constructors. */
+    if (xDrawable != drawable)
+	return NULL;
+
+    pdraw = Xmalloc(sizeof(*pdraw));
+    if (!pdraw)
+	return NULL;
+
+    pdraw->drawable = drawable;
+    pdraw->psc = psc;
+
+    if (!XF86DRICreateDrawable(psc->dpy, psc->scr, drawable, &hwDrawable))
+	return NULL;
+
+    /* Create a new drawable */
+    pdraw->driDrawable =
+	(*psc->legacy->createNewDrawable)(psc->__driScreen,
+					  config->driConfig,
+					  hwDrawable,
+					  GLX_WINDOW_BIT,
+					  empty_attribute_list,
+					  pdraw);
+
+    if (!pdraw->driDrawable) {
+	XF86DRIDestroyDrawable(psc->dpy, psc->scr, drawable);
+	Xfree(pdraw);
+	return NULL;
+    }
+
+    pdraw->destroyDrawable = driDestroyDrawable;
+
+    return pdraw;
+}
+
+static void driSwapBuffers(__GLXDRIdrawable *pdraw)
+{
+   (*pdraw->psc->core->swapBuffers)(pdraw->driDrawable);
+}
+
+static void driCopySubBuffer(__GLXDRIdrawable *pdraw,
+			     int x, int y, int width, int height)
+{
+    (*pdraw->psc->driCopySubBuffer->copySubBuffer)(pdraw->driDrawable,
+						   x, y, width, height);
 }
 
+static void driDestroyScreen(__GLXscreenConfigs *psc)
+{
+    /* Free the direct rendering per screen data */
+    if (psc->__driScreen)
+	(*psc->core->destroyScreen)(psc->__driScreen);
+    psc->__driScreen = NULL;
+    if (psc->driver)
+	dlclose(psc->driver);
+}
+
+static __GLXDRIscreen *driCreateScreen(__GLXscreenConfigs *psc, int screen,
+				       __GLXdisplayPrivate *priv)
+{
+    __GLXDRIdisplayPrivate *pdp;
+    __GLXDRIscreen *psp;
+    const __DRIextension **extensions;
+    char *driverName;
+    int i;
+
+    psp = Xmalloc(sizeof *psp);
+    if (psp == NULL)
+	return NULL;
+
+    /* Initialize per screen dynamic client GLX extensions */
+    psc->ext_list_first_time = GL_TRUE;
+
+    if (!driGetDriverName(priv->dpy, screen, &driverName)) {
+	Xfree(psp);
+	return NULL;
+    }
+
+    psc->driver = driOpenDriver(driverName);
+    Xfree(driverName);
+    if (psc->driver == NULL) {
+	Xfree(psp);
+	return NULL;
+    }
+
+    extensions = dlsym(psc->driver, __DRI_DRIVER_EXTENSIONS);
+    if (extensions == NULL) {
+ 	ErrorMessageF("driver exports no extensions (%s)\n", dlerror());
+	Xfree(psp);
+	return NULL;
+    }
+
+    for (i = 0; extensions[i]; i++) {
+	if (strcmp(extensions[i]->name, __DRI_CORE) == 0)
+	    psc->core = (__DRIcoreExtension *) extensions[i];
+	if (strcmp(extensions[i]->name, __DRI_LEGACY) == 0)
+	    psc->legacy = (__DRIlegacyExtension *) extensions[i];
+    }
+
+    if (psc->core == NULL || psc->legacy == NULL) {
+	Xfree(psp);
+	return NULL;
+    }
+  
+    pdp = (__GLXDRIdisplayPrivate *) priv->driDisplay;
+    psc->__driScreen =
+ 	CallCreateNewScreen(psc->dpy, screen, psc, pdp);
+    if (psc->__driScreen == NULL) {
+ 	dlclose(psc->driver);
+ 	Xfree(psp);
+ 	return NULL;
+    }
+
+    driBindExtensions(psc, 0);
+    if (psc->driCopySubBuffer)
+	psp->copySubBuffer = driCopySubBuffer;
+
+    psp->destroyScreen = driDestroyScreen;
+    psp->createContext = driCreateContext;
+    psp->createDrawable = driCreateDrawable;
+    psp->swapBuffers = driSwapBuffers;
+
+    return psp;
+}
+
+/* Called from __glXFreeDisplayPrivate.
+ */
+static void driDestroyDisplay(__GLXDRIdisplay *dpy)
+{
+    Xfree(dpy);
+}
 
 /*
  * Allocate, initialize and return a __DRIdisplayPrivate object.
  * This is called from __glXInitialize() when we are given a new
  * display pointer.
  */
-void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp)
+_X_HIDDEN __GLXDRIdisplay *driCreateDisplay(Display *dpy)
 {
-    const int numScreens = ScreenCount(dpy);
-    __DRIdisplayPrivate *pdpyp;
+    __GLXDRIdisplayPrivate *pdpyp;
     int eventBase, errorBase;
     int major, minor, patch;
-    int scrn;
-
-    /* Initialize these fields to NULL in case we fail.
-     * If we don't do this we may later get segfaults trying to free random
-     * addresses when the display is closed.
-     */
-    pdisp->private = NULL;
-    pdisp->destroyDisplay = NULL;
 
     if (!XF86DRIQueryExtension(dpy, &eventBase, &errorBase)) {
 	return NULL;
@@ -479,7 +685,7 @@ void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp)
 	return NULL;
     }
 
-    pdpyp = (__DRIdisplayPrivate *)Xmalloc(sizeof(__DRIdisplayPrivate));
+    pdpyp = Xmalloc(sizeof *pdpyp);
     if (!pdpyp) {
 	return NULL;
     }
@@ -488,41 +694,10 @@ void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp)
     pdpyp->driMinor = minor;
     pdpyp->driPatch = patch;
 
-    pdisp->destroyDisplay = driDestroyDisplay;
-
-    /* allocate array of pointers to createNewScreen funcs */
-    pdisp->createNewScreen = (PFNCREATENEWSCREENFUNC *)
-      Xmalloc(numScreens * sizeof(void *));
-    if (!pdisp->createNewScreen) {
-       Xfree(pdpyp);
-       return NULL;
-    }
-
-    /* allocate array of library handles */
-    pdpyp->libraryHandles = (void **) Xmalloc(numScreens * sizeof(void*));
-    if (!pdpyp->libraryHandles) {
-       Xfree(pdisp->createNewScreen);
-       Xfree(pdpyp);
-       return NULL;
-    }
-
-    /* dynamically discover DRI drivers for all screens, saving each
-     * driver's "__driCreateScreen" function pointer.  That's the bootstrap
-     * entrypoint for all DRI drivers.
-     */
-    for (scrn = 0; scrn < numScreens; scrn++) {
-        __DRIdriver *driver = driGetDriver(dpy, scrn);
-        if (driver) {
-           pdisp->createNewScreen[scrn] = driver->createNewScreenFunc;
-           pdpyp->libraryHandles[scrn] = driver->handle;
-        }
-        else {
-           pdisp->createNewScreen[scrn] = NULL;
-           pdpyp->libraryHandles[scrn] = NULL;
-        }
-    }
+    pdpyp->base.destroyDisplay = driDestroyDisplay;
+    pdpyp->base.createScreen = driCreateScreen;
 
-    return (void *)pdpyp;
+    return &pdpyp->base;
 }
 
 #endif /* GLX_DIRECT_RENDERING */
diff --git a/src/glx/x11/dri_glx.h b/src/glx/x11/dri_glx.h
deleted file mode 100644
index 75561685c81..00000000000
--- a/src/glx/x11/dri_glx.h
+++ /dev/null
@@ -1,61 +0,0 @@
-/**************************************************************************
-
-Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sub license, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial portions
-of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
-OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
-IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
-ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
-TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
-SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Kevin E. Martin <kevin@precisioninsight.com>
- *   Brian Paul <brian@precisioninsight.com>
- *
- */
-
-#ifndef _DRI_GLX_H_
-#define _DRI_GLX_H_
-
-#ifdef GLX_DIRECT_RENDERING
-
-struct __DRIdisplayPrivateRec {
-    /*
-    ** XFree86-DRI version information
-    */
-    int driMajor;
-    int driMinor;
-    int driPatch;
-
-    /*
-    ** Array of library handles [indexed by screen number]
-    */
-    void **libraryHandles;
-};
-
-typedef struct __DRIdisplayPrivateRec  __DRIdisplayPrivate;
-typedef struct __DRIscreenPrivateRec   __DRIscreenPrivate;
-typedef struct __DRIvisualPrivateRec   __DRIvisualPrivate;
-typedef struct __DRIcontextPrivateRec  __DRIcontextPrivate;
-typedef struct __DRIdrawablePrivateRec __DRIdrawablePrivate;
-
-#endif
-#endif /* _DRI_GLX_H_ */
diff --git a/src/glx/x11/drisw_glx.c b/src/glx/x11/drisw_glx.c
new file mode 100644
index 00000000000..fee45952b38
--- /dev/null
+++ b/src/glx/x11/drisw_glx.c
@@ -0,0 +1,450 @@
+/*
+ * Copyright 2008 George Sapountzis
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifdef GLX_DIRECT_RENDERING
+
+#include <X11/Xlib.h>
+#include "glxclient.h"
+#include "glcontextmodes.h"
+#include <dlfcn.h>
+#include "dri_common.h"
+
+typedef struct __GLXDRIdisplayPrivateRec __GLXDRIdisplayPrivate;
+typedef struct __GLXDRIcontextPrivateRec __GLXDRIcontextPrivate;
+typedef struct __GLXDRIdrawablePrivateRec __GLXDRIdrawablePrivate;
+
+struct __GLXDRIdisplayPrivateRec
+{
+   __GLXDRIdisplay base;
+};
+
+struct __GLXDRIcontextPrivateRec
+{
+   __GLXDRIcontext base;
+   __DRIcontext *driContext;
+   __GLXscreenConfigs *psc;
+};
+
+struct __GLXDRIdrawablePrivateRec
+{
+   __GLXDRIdrawable base;
+
+   GC gc;
+   GC swapgc;
+
+   XVisualInfo *visinfo;
+   XImage *ximage;
+   int bpp;
+};
+
+/**
+ * swrast loader functions
+ */
+
+static Bool
+XCreateDrawable(__GLXDRIdrawablePrivate * pdp,
+                Display * dpy, XID drawable, int visualid)
+{
+   XGCValues gcvalues;
+   long visMask;
+   XVisualInfo visTemp;
+   int num_visuals;
+
+   /* create GC's */
+   pdp->gc = XCreateGC(dpy, drawable, 0, NULL);
+   pdp->swapgc = XCreateGC(dpy, drawable, 0, NULL);
+
+   gcvalues.function = GXcopy;
+   gcvalues.graphics_exposures = False;
+   XChangeGC(dpy, pdp->gc, GCFunction, &gcvalues);
+   XChangeGC(dpy, pdp->swapgc, GCFunction, &gcvalues);
+   XChangeGC(dpy, pdp->swapgc, GCGraphicsExposures, &gcvalues);
+
+   /* create XImage  */
+   visTemp.screen = DefaultScreen(dpy);
+   visTemp.visualid = visualid;
+   visMask = (VisualScreenMask | VisualIDMask);
+   pdp->visinfo = XGetVisualInfo(dpy, visMask, &visTemp, &num_visuals);
+
+   pdp->ximage = XCreateImage(dpy, pdp->visinfo->visual, pdp->visinfo->depth, ZPixmap, 0,       /* format, offset */
+                              NULL,     /* data */
+                              0, 0,     /* size */
+                              32,       /* bitmap_pad */
+                              0);       /* bytes_per_line */
+
+   /* get the true number of bits per pixel */
+   pdp->bpp = pdp->ximage->bits_per_pixel;
+
+   return True;
+}
+
+static void
+XDestroyDrawable(__GLXDRIdrawablePrivate * pdp, Display * dpy, XID drawable)
+{
+   XDestroyImage(pdp->ximage);
+   XFree(pdp->visinfo);
+
+   XFreeGC(dpy, pdp->gc);
+   XFreeGC(dpy, pdp->swapgc);
+}
+
+static void
+swrastGetDrawableInfo(__DRIdrawable * draw,
+                      int *x, int *y, int *w, int *h, void *loaderPrivate)
+{
+   __GLXDRIdrawablePrivate *pdp = loaderPrivate;
+   __GLXDRIdrawable *pdraw = &(pdp->base);;
+   Display *dpy = pdraw->psc->dpy;
+   Drawable drawable;
+
+   Window root;
+   Status stat;
+   unsigned int bw, depth;
+
+   drawable = pdraw->xDrawable;
+
+   stat = XGetGeometry(dpy, drawable, &root,
+                       x, y, (unsigned int *) w, (unsigned int *) h,
+                       &bw, &depth);
+}
+
+static inline int
+bytes_per_line(int w, int bpp, unsigned mul)
+{
+   unsigned mask = mul - 1;
+
+   return ((w * bpp + mask) & ~mask) / 8;
+}
+
+static void
+swrastPutImage(__DRIdrawable * draw, int op,
+               int x, int y, int w, int h, char *data, void *loaderPrivate)
+{
+   __GLXDRIdrawablePrivate *pdp = loaderPrivate;
+   __GLXDRIdrawable *pdraw = &(pdp->base);;
+   Display *dpy = pdraw->psc->dpy;
+   Drawable drawable;
+   XImage *ximage;
+   GC gc;
+
+   switch (op) {
+   case __DRI_SWRAST_IMAGE_OP_DRAW:
+      gc = pdp->gc;
+      break;
+   case __DRI_SWRAST_IMAGE_OP_SWAP:
+      gc = pdp->swapgc;
+      break;
+   default:
+      return;
+   }
+
+   drawable = pdraw->xDrawable;
+
+   ximage = pdp->ximage;
+   ximage->data = data;
+   ximage->width = w;
+   ximage->height = h;
+   ximage->bytes_per_line = bytes_per_line(w, pdp->bpp, 32);
+
+   XPutImage(dpy, drawable, gc, ximage, 0, 0, x, y, w, h);
+
+   ximage->data = NULL;
+}
+
+static void
+swrastGetImage(__DRIdrawable * draw,
+               int x, int y, int w, int h, char *data, void *loaderPrivate)
+{
+   __GLXDRIdrawablePrivate *pdp = loaderPrivate;
+   __GLXDRIdrawable *pdraw = &(pdp->base);;
+   Display *dpy = pdraw->psc->dpy;
+   Drawable drawable;
+   XImage *ximage;
+
+   drawable = pdraw->xDrawable;
+
+   ximage = pdp->ximage;
+   ximage->data = data;
+   ximage->width = w;
+   ximage->height = h;
+   ximage->bytes_per_line = bytes_per_line(w, pdp->bpp, 32);
+
+   XGetSubImage(dpy, drawable, x, y, w, h, ~0L, ZPixmap, ximage, 0, 0);
+
+   ximage->data = NULL;
+}
+
+static const __DRIswrastLoaderExtension swrastLoaderExtension = {
+   {__DRI_SWRAST_LOADER, __DRI_SWRAST_LOADER_VERSION},
+   swrastGetDrawableInfo,
+   swrastPutImage,
+   swrastGetImage
+};
+
+static const __DRIextension *loader_extensions[] = {
+   &systemTimeExtension.base,
+   &swrastLoaderExtension.base,
+   NULL
+};
+
+/**
+ * GLXDRI functions
+ */
+
+static void
+driDestroyContext(__GLXDRIcontext * context,
+                  __GLXscreenConfigs * psc, Display * dpy)
+{
+   __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+   const __DRIcoreExtension *core = pcp->psc->core;
+
+   (*core->destroyContext) (pcp->driContext);
+
+   Xfree(pcp);
+}
+
+static Bool
+driBindContext(__GLXDRIcontext * context,
+               __GLXDRIdrawable * draw, __GLXDRIdrawable * read)
+{
+   __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+   const __DRIcoreExtension *core = pcp->psc->core;
+
+   return (*core->bindContext) (pcp->driContext,
+                                draw->driDrawable, read->driDrawable);
+}
+
+static void
+driUnbindContext(__GLXDRIcontext * context)
+{
+   __GLXDRIcontextPrivate *pcp = (__GLXDRIcontextPrivate *) context;
+   const __DRIcoreExtension *core = pcp->psc->core;
+
+   (*core->unbindContext) (pcp->driContext);
+}
+
+static __GLXDRIcontext *
+driCreateContext(__GLXscreenConfigs * psc,
+                 const __GLcontextModes * mode,
+                 GLXContext gc, GLXContext shareList, int renderType)
+{
+   __GLXDRIcontextPrivate *pcp, *pcp_shared;
+   __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) mode;
+   const __DRIcoreExtension *core = psc->core;
+   __DRIcontext *shared = NULL;
+
+   if (!psc || !psc->driScreen)
+      return NULL;
+
+   if (shareList) {
+      pcp_shared = (__GLXDRIcontextPrivate *) shareList->driContext;
+      shared = pcp_shared->driContext;
+   }
+
+   pcp = Xmalloc(sizeof *pcp);
+   if (pcp == NULL)
+      return NULL;
+
+   pcp->psc = psc;
+   pcp->driContext =
+      (*core->createNewContext) (psc->__driScreen,
+                                 config->driConfig, shared, pcp);
+   if (pcp->driContext == NULL) {
+      Xfree(pcp);
+      return NULL;
+   }
+
+   pcp->base.destroyContext = driDestroyContext;
+   pcp->base.bindContext = driBindContext;
+   pcp->base.unbindContext = driUnbindContext;
+
+   return &pcp->base;
+}
+
+static void
+driDestroyDrawable(__GLXDRIdrawable * pdraw)
+{
+   __GLXDRIdrawablePrivate *pdp = (__GLXDRIdrawablePrivate *) pdraw;
+   const __DRIcoreExtension *core = pdraw->psc->core;
+
+   (*core->destroyDrawable) (pdraw->driDrawable);
+
+   XDestroyDrawable(pdp, pdraw->psc->dpy, pdraw->drawable);
+   Xfree(pdp);
+}
+
+static __GLXDRIdrawable *
+driCreateDrawable(__GLXscreenConfigs * psc,
+                  XID xDrawable,
+                  GLXDrawable drawable, const __GLcontextModes * modes)
+{
+   __GLXDRIdrawable *pdraw;
+   __GLXDRIdrawablePrivate *pdp;
+   __GLXDRIconfigPrivate *config = (__GLXDRIconfigPrivate *) modes;
+   const __DRIswrastExtension *swrast = psc->swrast;
+
+   /* Old dri can't handle GLX 1.3+ drawable constructors. */
+   if (xDrawable != drawable)
+      return NULL;
+
+   pdp = Xmalloc(sizeof(*pdp));
+   if (!pdp)
+      return NULL;
+
+   pdraw = &(pdp->base);
+   pdraw->xDrawable = xDrawable;
+   pdraw->drawable = drawable;
+   pdraw->psc = psc;
+
+   XCreateDrawable(pdp, psc->dpy, xDrawable, modes->visualID);
+
+   /* Create a new drawable */
+   pdraw->driDrawable =
+      (*swrast->createNewDrawable) (psc->__driScreen, config->driConfig, pdp);
+
+   if (!pdraw->driDrawable) {
+      XDestroyDrawable(pdp, psc->dpy, xDrawable);
+      Xfree(pdp);
+      return NULL;
+   }
+
+   pdraw->destroyDrawable = driDestroyDrawable;
+
+   return pdraw;
+}
+
+static void
+driSwapBuffers(__GLXDRIdrawable * pdraw)
+{
+   (*pdraw->psc->core->swapBuffers) (pdraw->driDrawable);
+}
+
+static void
+driDestroyScreen(__GLXscreenConfigs * psc)
+{
+   /* Free the direct rendering per screen data */
+   (*psc->core->destroyScreen) (psc->__driScreen);
+   psc->__driScreen = NULL;
+   if (psc->driver)
+      dlclose(psc->driver);
+}
+
+static __GLXDRIscreen *
+driCreateScreen(__GLXscreenConfigs * psc, int screen,
+                __GLXdisplayPrivate * priv)
+{
+   __GLXDRIscreen *psp;
+   const __DRIconfig **driver_configs;
+   const __DRIextension **extensions;
+   const char *driverName = "swrast";
+   int i;
+
+   psp = Xmalloc(sizeof *psp);
+   if (psp == NULL)
+      return NULL;
+
+   /* Initialize per screen dynamic client GLX extensions */
+   psc->ext_list_first_time = GL_TRUE;
+
+   psc->driver = driOpenDriver(driverName);
+   if (psc->driver == NULL)
+      goto handle_error;
+
+   extensions = dlsym(psc->driver, __DRI_DRIVER_EXTENSIONS);
+   if (extensions == NULL) {
+      ErrorMessageF("driver exports no extensions (%s)\n", dlerror());
+      goto handle_error;
+   }
+
+   for (i = 0; extensions[i]; i++) {
+      if (strcmp(extensions[i]->name, __DRI_CORE) == 0)
+         psc->core = (__DRIcoreExtension *) extensions[i];
+      if (strcmp(extensions[i]->name, __DRI_SWRAST) == 0)
+         psc->swrast = (__DRIswrastExtension *) extensions[i];
+   }
+
+   if (psc->core == NULL || psc->swrast == NULL) {
+      ErrorMessageF("core dri extension not found\n");
+      goto handle_error;
+   }
+
+   psc->__driScreen =
+      psc->swrast->createNewScreen(screen,
+                                   loader_extensions, &driver_configs, psc);
+   if (psc->__driScreen == NULL) {
+      ErrorMessageF("failed to create dri screen\n");
+      goto handle_error;
+   }
+
+   driBindExtensions(psc, 0);
+
+   psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
+   psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
+
+   psp->destroyScreen = driDestroyScreen;
+   psp->createContext = driCreateContext;
+   psp->createDrawable = driCreateDrawable;
+   psp->swapBuffers = driSwapBuffers;
+
+   return psp;
+
+ handle_error:
+   Xfree(psp);
+
+   if (psc->driver)
+      dlclose(psc->driver);
+
+   ErrorMessageF("reverting to indirect rendering\n");
+
+   return NULL;
+}
+
+/* Called from __glXFreeDisplayPrivate.
+ */
+static void
+driDestroyDisplay(__GLXDRIdisplay * dpy)
+{
+   Xfree(dpy);
+}
+
+/*
+ * Allocate, initialize and return a __DRIdisplayPrivate object.
+ * This is called from __glXInitialize() when we are given a new
+ * display pointer.
+ */
+_X_HIDDEN __GLXDRIdisplay *
+driswCreateDisplay(Display * dpy)
+{
+   __GLXDRIdisplayPrivate *pdpyp;
+
+   pdpyp = Xmalloc(sizeof *pdpyp);
+   if (pdpyp == NULL)
+      return NULL;
+
+   pdpyp->base.destroyDisplay = driDestroyDisplay;
+   pdpyp->base.createScreen = driCreateScreen;
+
+   return &pdpyp->base;
+}
+
+#endif /* GLX_DIRECT_RENDERING */
diff --git a/src/glx/x11/eval.c b/src/glx/x11/eval.c
index 2544c50fceb..226fb7df2e8 100644
--- a/src/glx/x11/eval.c
+++ b/src/glx/x11/eval.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "packrender.h"
 
@@ -41,91 +36,97 @@
 ** the information that the GL needs.
 */
 
-void __glFillMap1f(GLint k, GLint order, GLint stride, 
-		   const GLfloat *points, GLubyte *pc)
+void
+__glFillMap1f(GLint k, GLint order, GLint stride,
+              const GLfloat * points, GLubyte * pc)
 {
-    if (stride == k) {
-	/* Just copy the data */
-	__GLX_PUT_FLOAT_ARRAY(0, points, order * k);
-    } else {
-	GLint i;
+   if (stride == k) {
+      /* Just copy the data */
+      __GLX_PUT_FLOAT_ARRAY(0, points, order * k);
+   }
+   else {
+      GLint i;
 
-	for (i = 0; i < order; i++) {
-	    __GLX_PUT_FLOAT_ARRAY(0, points, k);
-	    points += stride;
-	    pc += k * __GLX_SIZE_FLOAT32;
-	}
-    }
+      for (i = 0; i < order; i++) {
+         __GLX_PUT_FLOAT_ARRAY(0, points, k);
+         points += stride;
+         pc += k * __GLX_SIZE_FLOAT32;
+      }
+   }
 }
 
-void __glFillMap1d(GLint k, GLint order, GLint stride, 
-		   const GLdouble *points, GLubyte *pc)
+void
+__glFillMap1d(GLint k, GLint order, GLint stride,
+              const GLdouble * points, GLubyte * pc)
 {
-    if (stride == k) {
-	/* Just copy the data */
-	__GLX_PUT_DOUBLE_ARRAY(0, points, order * k);
-    } else {
-	GLint i;
-	for (i = 0; i < order; i++) {
-            __GLX_PUT_DOUBLE_ARRAY(0, points, k);
-	    points += stride;
-	    pc += k * __GLX_SIZE_FLOAT64;
-	}
-    }
+   if (stride == k) {
+      /* Just copy the data */
+      __GLX_PUT_DOUBLE_ARRAY(0, points, order * k);
+   }
+   else {
+      GLint i;
+      for (i = 0; i < order; i++) {
+         __GLX_PUT_DOUBLE_ARRAY(0, points, k);
+         points += stride;
+         pc += k * __GLX_SIZE_FLOAT64;
+      }
+   }
 }
 
-void __glFillMap2f(GLint k, GLint majorOrder, GLint minorOrder, 
-		   GLint majorStride, GLint minorStride,
-		   const GLfloat *points, GLfloat *data)
+void
+__glFillMap2f(GLint k, GLint majorOrder, GLint minorOrder,
+              GLint majorStride, GLint minorStride,
+              const GLfloat * points, GLfloat * data)
 {
-    GLint i, j, x;
+   GLint i, j, x;
 
-    if ((minorStride == k) && (majorStride == minorOrder*k)) {
-	/* Just copy the data */
-	__GLX_MEM_COPY(data, points, majorOrder * majorStride *
-		       __GLX_SIZE_FLOAT32);
-	return;
-    }
-    for (i = 0; i < majorOrder; i++) {
-	for (j = 0; j < minorOrder; j++) {
-	    for (x = 0; x < k; x++) {
-		data[x] = points[x];
-	    }
-	    points += minorStride;
-	    data += k;
-	}
-	points += majorStride - minorStride * minorOrder;
-    }
+   if ((minorStride == k) && (majorStride == minorOrder * k)) {
+      /* Just copy the data */
+      __GLX_MEM_COPY(data, points, majorOrder * majorStride *
+                     __GLX_SIZE_FLOAT32);
+      return;
+   }
+   for (i = 0; i < majorOrder; i++) {
+      for (j = 0; j < minorOrder; j++) {
+         for (x = 0; x < k; x++) {
+            data[x] = points[x];
+         }
+         points += minorStride;
+         data += k;
+      }
+      points += majorStride - minorStride * minorOrder;
+   }
 }
 
-void __glFillMap2d(GLint k, GLint majorOrder, GLint minorOrder, 
-		   GLint majorStride, GLint minorStride,
-		   const GLdouble *points, GLdouble *data)
+void
+__glFillMap2d(GLint k, GLint majorOrder, GLint minorOrder,
+              GLint majorStride, GLint minorStride,
+              const GLdouble * points, GLdouble * data)
 {
-    int i,j,x;
+   int i, j, x;
 
-    if ((minorStride == k) && (majorStride == minorOrder*k)) {
-	/* Just copy the data */
-	__GLX_MEM_COPY(data, points, majorOrder * majorStride *
-		       __GLX_SIZE_FLOAT64);
-	return;
-    }
+   if ((minorStride == k) && (majorStride == minorOrder * k)) {
+      /* Just copy the data */
+      __GLX_MEM_COPY(data, points, majorOrder * majorStride *
+                     __GLX_SIZE_FLOAT64);
+      return;
+   }
 
 #ifdef __GLX_ALIGN64
-    x = k * __GLX_SIZE_FLOAT64;
+   x = k * __GLX_SIZE_FLOAT64;
 #endif
-    for (i = 0; i<majorOrder; i++) {
-	for (j = 0; j<minorOrder; j++) {
+   for (i = 0; i < majorOrder; i++) {
+      for (j = 0; j < minorOrder; j++) {
 #ifdef __GLX_ALIGN64
-	    __GLX_MEM_COPY(data, points, x);
+         __GLX_MEM_COPY(data, points, x);
 #else
-	    for (x = 0; x<k; x++) {
-		data[x] = points[x];
-	    }
+         for (x = 0; x < k; x++) {
+            data[x] = points[x];
+         }
 #endif
-	    points += minorStride;
-	    data += k;
-	}
-	points += majorStride - minorStride * minorOrder;
-    }
+         points += minorStride;
+         data += k;
+      }
+      points += majorStride - minorStride * minorOrder;
+   }
 }
diff --git a/src/glx/x11/glcontextmodes.c b/src/glx/x11/glcontextmodes.c
index 788ecf6a3aa..232031c2ca8 100644
--- a/src/glx/x11/glcontextmodes.c
+++ b/src/glx/x11/glcontextmodes.c
@@ -78,16 +78,16 @@
  *         be returned.  Otherwise \c GLX_NONE will be returned.
  */
 GLint
-_gl_convert_from_x_visual_type( int visualType )
+_gl_convert_from_x_visual_type(int visualType)
 {
-    static const int glx_visual_types[ NUM_VISUAL_TYPES ] = {
-	GLX_STATIC_GRAY,  GLX_GRAY_SCALE,
-	GLX_STATIC_COLOR, GLX_PSEUDO_COLOR,
-	GLX_TRUE_COLOR,   GLX_DIRECT_COLOR
-    };
-
-    return ( (unsigned) visualType < NUM_VISUAL_TYPES )
-	? glx_visual_types[ visualType ] : GLX_NONE;
+   static const int glx_visual_types[NUM_VISUAL_TYPES] = {
+      GLX_STATIC_GRAY, GLX_GRAY_SCALE,
+      GLX_STATIC_COLOR, GLX_PSEUDO_COLOR,
+      GLX_TRUE_COLOR, GLX_DIRECT_COLOR
+   };
+
+   return ((unsigned) visualType < NUM_VISUAL_TYPES)
+      ? glx_visual_types[visualType] : GLX_NONE;
 }
 
 
@@ -100,16 +100,16 @@ _gl_convert_from_x_visual_type( int visualType )
  *         be returned.  Otherwise -1 will be returned.
  */
 GLint
-_gl_convert_to_x_visual_type( int visualType )
+_gl_convert_to_x_visual_type(int visualType)
 {
-    static const int x_visual_types[ NUM_VISUAL_TYPES ] = {
-	TrueColor,   DirectColor,
-	PseudoColor, StaticColor,
-	GrayScale,   StaticGray
-    };
-
-    return ( (unsigned) (visualType - GLX_TRUE_COLOR) < NUM_VISUAL_TYPES )
-	? x_visual_types[ visualType - GLX_TRUE_COLOR ] : -1;
+   static const int x_visual_types[NUM_VISUAL_TYPES] = {
+      TrueColor, DirectColor,
+      PseudoColor, StaticColor,
+      GrayScale, StaticGray
+   };
+
+   return ((unsigned) (visualType - GLX_TRUE_COLOR) < NUM_VISUAL_TYPES)
+      ? x_visual_types[visualType - GLX_TRUE_COLOR] : -1;
 }
 
 
@@ -128,76 +128,76 @@ _gl_convert_to_x_visual_type( int visualType )
  * structure will be set to the \c vid of the \c __GLXvisualConfig structure.
  */
 void
-_gl_copy_visual_to_context_mode( __GLcontextModes * mode,
-				 const __GLXvisualConfig * config )
+_gl_copy_visual_to_context_mode(__GLcontextModes * mode,
+                                const __GLXvisualConfig * config)
 {
-    __GLcontextModes * const next = mode->next;
-
-    (void) _mesa_memset( mode, 0, sizeof( __GLcontextModes ) );
-    mode->next = next;
-
-    mode->visualID = config->vid;
-    mode->visualType = _gl_convert_from_x_visual_type( config->class );
-    mode->xRenderable = GL_TRUE;
-    mode->fbconfigID = config->vid;
-    mode->drawableType = GLX_WINDOW_BIT | GLX_PIXMAP_BIT;
-
-    mode->rgbMode = (config->rgba != 0);
-    mode->renderType = (mode->rgbMode) ? GLX_RGBA_BIT : GLX_COLOR_INDEX_BIT;
-
-    mode->colorIndexMode = !(mode->rgbMode);
-    mode->doubleBufferMode = (config->doubleBuffer != 0);
-    mode->stereoMode = (config->stereo != 0);
-
-    mode->haveAccumBuffer = ((config->accumRedSize +
-			      config->accumGreenSize +
-			      config->accumBlueSize +
-			      config->accumAlphaSize) > 0);
-    mode->haveDepthBuffer = (config->depthSize > 0);
-    mode->haveStencilBuffer = (config->stencilSize > 0);
-
-    mode->redBits = config->redSize;
-    mode->greenBits = config->greenSize;
-    mode->blueBits = config->blueSize;
-    mode->alphaBits = config->alphaSize;
-    mode->redMask = config->redMask;
-    mode->greenMask = config->greenMask;
-    mode->blueMask = config->blueMask;
-    mode->alphaMask = config->alphaMask;
-    mode->rgbBits = mode->rgbMode ? config->bufferSize : 0;
-    mode->indexBits = mode->colorIndexMode ? config->bufferSize : 0;
-
-    mode->accumRedBits = config->accumRedSize;
-    mode->accumGreenBits = config->accumGreenSize;
-    mode->accumBlueBits = config->accumBlueSize;
-    mode->accumAlphaBits = config->accumAlphaSize;
-    mode->depthBits = config->depthSize;
-    mode->stencilBits = config->stencilSize;
-
-    mode->numAuxBuffers = config->auxBuffers;
-    mode->level = config->level;
-
-    mode->visualRating = config->visualRating;
-    mode->transparentPixel = config->transparentPixel;
-    mode->transparentRed   = config->transparentRed;
-    mode->transparentGreen = config->transparentGreen;
-    mode->transparentBlue  = config->transparentBlue;
-    mode->transparentAlpha = config->transparentAlpha;
-    mode->transparentIndex = config->transparentIndex;
-    mode->samples = config->multiSampleSize;
-    mode->sampleBuffers = config->nMultiSampleBuffers;
-    /* mode->visualSelectGroup = config->visualSelectGroup; ? */
-
-    mode->swapMethod = GLX_SWAP_UNDEFINED_OML;
-
-    mode->bindToTextureRgb = (mode->rgbMode) ? GL_TRUE : GL_FALSE;
-    mode->bindToTextureRgba = (mode->rgbMode && mode->alphaBits) ?
-	GL_TRUE : GL_FALSE;
-    mode->bindToMipmapTexture = mode->rgbMode ? GL_TRUE : GL_FALSE;
-    mode->bindToTextureTargets = mode->rgbMode ?
-	GLX_TEXTURE_1D_BIT_EXT | GLX_TEXTURE_2D_BIT_EXT |
-	GLX_TEXTURE_RECTANGLE_BIT_EXT : 0;
-    mode->yInverted = GL_FALSE;
+   __GLcontextModes *const next = mode->next;
+
+   (void) _mesa_memset(mode, 0, sizeof(__GLcontextModes));
+   mode->next = next;
+
+   mode->visualID = config->vid;
+   mode->visualType = _gl_convert_from_x_visual_type(config->class);
+   mode->xRenderable = GL_TRUE;
+   mode->fbconfigID = config->vid;
+   mode->drawableType = GLX_WINDOW_BIT | GLX_PIXMAP_BIT;
+
+   mode->rgbMode = (config->rgba != 0);
+   mode->renderType = (mode->rgbMode) ? GLX_RGBA_BIT : GLX_COLOR_INDEX_BIT;
+
+   mode->colorIndexMode = !(mode->rgbMode);
+   mode->doubleBufferMode = (config->doubleBuffer != 0);
+   mode->stereoMode = (config->stereo != 0);
+
+   mode->haveAccumBuffer = ((config->accumRedSize +
+                             config->accumGreenSize +
+                             config->accumBlueSize +
+                             config->accumAlphaSize) > 0);
+   mode->haveDepthBuffer = (config->depthSize > 0);
+   mode->haveStencilBuffer = (config->stencilSize > 0);
+
+   mode->redBits = config->redSize;
+   mode->greenBits = config->greenSize;
+   mode->blueBits = config->blueSize;
+   mode->alphaBits = config->alphaSize;
+   mode->redMask = config->redMask;
+   mode->greenMask = config->greenMask;
+   mode->blueMask = config->blueMask;
+   mode->alphaMask = config->alphaMask;
+   mode->rgbBits = mode->rgbMode ? config->bufferSize : 0;
+   mode->indexBits = mode->colorIndexMode ? config->bufferSize : 0;
+
+   mode->accumRedBits = config->accumRedSize;
+   mode->accumGreenBits = config->accumGreenSize;
+   mode->accumBlueBits = config->accumBlueSize;
+   mode->accumAlphaBits = config->accumAlphaSize;
+   mode->depthBits = config->depthSize;
+   mode->stencilBits = config->stencilSize;
+
+   mode->numAuxBuffers = config->auxBuffers;
+   mode->level = config->level;
+
+   mode->visualRating = config->visualRating;
+   mode->transparentPixel = config->transparentPixel;
+   mode->transparentRed = config->transparentRed;
+   mode->transparentGreen = config->transparentGreen;
+   mode->transparentBlue = config->transparentBlue;
+   mode->transparentAlpha = config->transparentAlpha;
+   mode->transparentIndex = config->transparentIndex;
+   mode->samples = config->multiSampleSize;
+   mode->sampleBuffers = config->nMultiSampleBuffers;
+   /* mode->visualSelectGroup = config->visualSelectGroup; ? */
+
+   mode->swapMethod = GLX_SWAP_UNDEFINED_OML;
+
+   mode->bindToTextureRgb = (mode->rgbMode) ? GL_TRUE : GL_FALSE;
+   mode->bindToTextureRgba = (mode->rgbMode && mode->alphaBits) ?
+      GL_TRUE : GL_FALSE;
+   mode->bindToMipmapTexture = mode->rgbMode ? GL_TRUE : GL_FALSE;
+   mode->bindToTextureTargets = mode->rgbMode ?
+      GLX_TEXTURE_1D_BIT_EXT | GLX_TEXTURE_2D_BIT_EXT |
+      GLX_TEXTURE_RECTANGLE_BIT_EXT : 0;
+   mode->yInverted = GL_FALSE;
 }
 
 
@@ -211,148 +211,149 @@ _gl_copy_visual_to_context_mode( __GLcontextModes * mode,
  *          returned.  Otherwise \c GLX_BAD_ATTRIBUTE is returned.
  */
 int
-_gl_get_context_mode_data(const __GLcontextModes *mode, int attribute,
-			  int *value_return)
+_gl_get_context_mode_data(const __GLcontextModes * mode, int attribute,
+                          int *value_return)
 {
-    switch (attribute) {
-      case GLX_USE_GL:
-	*value_return = GL_TRUE;
-	return 0;
-      case GLX_BUFFER_SIZE:
-	*value_return = mode->rgbBits;
-	return 0;
-      case GLX_RGBA:
-	*value_return = mode->rgbMode;
-	return 0;
-      case GLX_RED_SIZE:
-	*value_return = mode->redBits;
-	return 0;
-      case GLX_GREEN_SIZE:
-	*value_return = mode->greenBits;
-	return 0;
-      case GLX_BLUE_SIZE:
-	*value_return = mode->blueBits;
-	return 0;
-      case GLX_ALPHA_SIZE:
-	*value_return = mode->alphaBits;
-	return 0;
-      case GLX_DOUBLEBUFFER:
-	*value_return = mode->doubleBufferMode;
-	return 0;
-      case GLX_STEREO:
-	*value_return = mode->stereoMode;
-	return 0;
-      case GLX_AUX_BUFFERS:
-	*value_return = mode->numAuxBuffers;
-	return 0;
-      case GLX_DEPTH_SIZE:
-	*value_return = mode->depthBits;
-	return 0;
-      case GLX_STENCIL_SIZE:
-	*value_return = mode->stencilBits;
-	return 0;
-      case GLX_ACCUM_RED_SIZE:
-	*value_return = mode->accumRedBits;
-	return 0;
-      case GLX_ACCUM_GREEN_SIZE:
-	*value_return = mode->accumGreenBits;
-	return 0;
-      case GLX_ACCUM_BLUE_SIZE:
-	*value_return = mode->accumBlueBits;
-	return 0;
-      case GLX_ACCUM_ALPHA_SIZE:
-	*value_return = mode->accumAlphaBits;
-	return 0;
-      case GLX_LEVEL:
-	*value_return = mode->level;
-	return 0;
-      case GLX_TRANSPARENT_TYPE_EXT:
-	*value_return = mode->transparentPixel;
-	return 0;
-      case GLX_TRANSPARENT_RED_VALUE:
-	*value_return = mode->transparentRed;
-	return 0;
-      case GLX_TRANSPARENT_GREEN_VALUE:
-	*value_return = mode->transparentGreen;
-	return 0;
-      case GLX_TRANSPARENT_BLUE_VALUE:
-	*value_return = mode->transparentBlue;
-	return 0;
-      case GLX_TRANSPARENT_ALPHA_VALUE:
-	*value_return = mode->transparentAlpha;
-	return 0;
-      case GLX_TRANSPARENT_INDEX_VALUE:
-	*value_return = mode->transparentIndex;
-	return 0;
-      case GLX_X_VISUAL_TYPE:
-	*value_return = mode->visualType;
-	return 0;
-      case GLX_CONFIG_CAVEAT:
-	*value_return = mode->visualRating;
-	return 0;
-      case GLX_VISUAL_ID:
-	*value_return = mode->visualID;
-	return 0;
-      case GLX_DRAWABLE_TYPE:
-	*value_return = mode->drawableType;
-	return 0;
-      case GLX_RENDER_TYPE:
-	*value_return = mode->renderType;
-	return 0;
-      case GLX_X_RENDERABLE:
-	*value_return = mode->xRenderable;
-	return 0;
-      case GLX_FBCONFIG_ID:
-	*value_return = mode->fbconfigID;
-	return 0;
-      case GLX_MAX_PBUFFER_WIDTH:
-	*value_return = mode->maxPbufferWidth;
-	return 0;
-      case GLX_MAX_PBUFFER_HEIGHT:
-	*value_return = mode->maxPbufferHeight;
-	return 0;
-      case GLX_MAX_PBUFFER_PIXELS:
-	*value_return = mode->maxPbufferPixels;
-	return 0;
-      case GLX_OPTIMAL_PBUFFER_WIDTH_SGIX:
-	*value_return = mode->optimalPbufferWidth;
-	return 0;
-      case GLX_OPTIMAL_PBUFFER_HEIGHT_SGIX:
-	*value_return = mode->optimalPbufferHeight;
-	return 0;
-      case GLX_SWAP_METHOD_OML:
-	*value_return = mode->swapMethod;
-	return 0;
-      case GLX_SAMPLE_BUFFERS_SGIS:
-	*value_return = mode->sampleBuffers;
-	return 0;
-      case GLX_SAMPLES_SGIS:
-	*value_return = mode->samples;
-	return 0;
-      case GLX_BIND_TO_TEXTURE_RGB_EXT:
-	*value_return = mode->bindToTextureRgb;
-	return 0;
-      case GLX_BIND_TO_TEXTURE_RGBA_EXT:
-	*value_return = mode->bindToTextureRgba;
-	return 0;
-      case GLX_BIND_TO_MIPMAP_TEXTURE_EXT:
-	*value_return = mode->bindToMipmapTexture;
-	return 0;
-      case GLX_BIND_TO_TEXTURE_TARGETS_EXT:
-	*value_return = mode->bindToTextureTargets;
-	return 0;
-      case GLX_Y_INVERTED_EXT:
-	*value_return = mode->yInverted;
-	return 0;
+   switch (attribute) {
+   case GLX_USE_GL:
+      *value_return = GL_TRUE;
+      return 0;
+   case GLX_BUFFER_SIZE:
+      *value_return = mode->rgbBits;
+      return 0;
+   case GLX_RGBA:
+      *value_return = mode->rgbMode;
+      return 0;
+   case GLX_RED_SIZE:
+      *value_return = mode->redBits;
+      return 0;
+   case GLX_GREEN_SIZE:
+      *value_return = mode->greenBits;
+      return 0;
+   case GLX_BLUE_SIZE:
+      *value_return = mode->blueBits;
+      return 0;
+   case GLX_ALPHA_SIZE:
+      *value_return = mode->alphaBits;
+      return 0;
+   case GLX_DOUBLEBUFFER:
+      *value_return = mode->doubleBufferMode;
+      return 0;
+   case GLX_STEREO:
+      *value_return = mode->stereoMode;
+      return 0;
+   case GLX_AUX_BUFFERS:
+      *value_return = mode->numAuxBuffers;
+      return 0;
+   case GLX_DEPTH_SIZE:
+      *value_return = mode->depthBits;
+      return 0;
+   case GLX_STENCIL_SIZE:
+      *value_return = mode->stencilBits;
+      return 0;
+   case GLX_ACCUM_RED_SIZE:
+      *value_return = mode->accumRedBits;
+      return 0;
+   case GLX_ACCUM_GREEN_SIZE:
+      *value_return = mode->accumGreenBits;
+      return 0;
+   case GLX_ACCUM_BLUE_SIZE:
+      *value_return = mode->accumBlueBits;
+      return 0;
+   case GLX_ACCUM_ALPHA_SIZE:
+      *value_return = mode->accumAlphaBits;
+      return 0;
+   case GLX_LEVEL:
+      *value_return = mode->level;
+      return 0;
+   case GLX_TRANSPARENT_TYPE_EXT:
+      *value_return = mode->transparentPixel;
+      return 0;
+   case GLX_TRANSPARENT_RED_VALUE:
+      *value_return = mode->transparentRed;
+      return 0;
+   case GLX_TRANSPARENT_GREEN_VALUE:
+      *value_return = mode->transparentGreen;
+      return 0;
+   case GLX_TRANSPARENT_BLUE_VALUE:
+      *value_return = mode->transparentBlue;
+      return 0;
+   case GLX_TRANSPARENT_ALPHA_VALUE:
+      *value_return = mode->transparentAlpha;
+      return 0;
+   case GLX_TRANSPARENT_INDEX_VALUE:
+      *value_return = mode->transparentIndex;
+      return 0;
+   case GLX_X_VISUAL_TYPE:
+      *value_return = mode->visualType;
+      return 0;
+   case GLX_CONFIG_CAVEAT:
+      *value_return = mode->visualRating;
+      return 0;
+   case GLX_VISUAL_ID:
+      *value_return = mode->visualID;
+      return 0;
+   case GLX_DRAWABLE_TYPE:
+      *value_return = mode->drawableType;
+      return 0;
+   case GLX_RENDER_TYPE:
+      *value_return = mode->renderType;
+      return 0;
+   case GLX_X_RENDERABLE:
+      *value_return = mode->xRenderable;
+      return 0;
+   case GLX_FBCONFIG_ID:
+      *value_return = mode->fbconfigID;
+      return 0;
+   case GLX_MAX_PBUFFER_WIDTH:
+      *value_return = mode->maxPbufferWidth;
+      return 0;
+   case GLX_MAX_PBUFFER_HEIGHT:
+      *value_return = mode->maxPbufferHeight;
+      return 0;
+   case GLX_MAX_PBUFFER_PIXELS:
+      *value_return = mode->maxPbufferPixels;
+      return 0;
+   case GLX_OPTIMAL_PBUFFER_WIDTH_SGIX:
+      *value_return = mode->optimalPbufferWidth;
+      return 0;
+   case GLX_OPTIMAL_PBUFFER_HEIGHT_SGIX:
+      *value_return = mode->optimalPbufferHeight;
+      return 0;
+   case GLX_SWAP_METHOD_OML:
+      *value_return = mode->swapMethod;
+      return 0;
+   case GLX_SAMPLE_BUFFERS_SGIS:
+      *value_return = mode->sampleBuffers;
+      return 0;
+   case GLX_SAMPLES_SGIS:
+      *value_return = mode->samples;
+      return 0;
+   case GLX_BIND_TO_TEXTURE_RGB_EXT:
+      *value_return = mode->bindToTextureRgb;
+      return 0;
+   case GLX_BIND_TO_TEXTURE_RGBA_EXT:
+      *value_return = mode->bindToTextureRgba;
+      return 0;
+   case GLX_BIND_TO_MIPMAP_TEXTURE_EXT:
+      *value_return = mode->bindToMipmapTexture == GL_TRUE ? GL_TRUE :
+         GL_FALSE;
+      return 0;
+   case GLX_BIND_TO_TEXTURE_TARGETS_EXT:
+      *value_return = mode->bindToTextureTargets;
+      return 0;
+   case GLX_Y_INVERTED_EXT:
+      *value_return = mode->yInverted;
+      return 0;
 
       /* Applications are NOT allowed to query GLX_VISUAL_SELECT_GROUP_SGIX.
        * It is ONLY for communication between the GLX client and the GLX
        * server.
        */
-      case GLX_VISUAL_SELECT_GROUP_SGIX:
-      default:
-	return GLX_BAD_ATTRIBUTE;
-    }
+   case GLX_VISUAL_SELECT_GROUP_SGIX:
+   default:
+      return GLX_BAD_ATTRIBUTE;
+   }
 }
 #endif /* !defined(IN_MINI_GLX) */
 
@@ -384,24 +385,24 @@ _gl_get_context_mode_data(const __GLcontextModes *mode, int attribute,
  *          extend the \c __GLcontextModes data-structure.
  */
 __GLcontextModes *
-_gl_context_modes_create( unsigned count, size_t minimum_size )
+_gl_context_modes_create(unsigned count, size_t minimum_size)
 {
-   const size_t size = (minimum_size > sizeof( __GLcontextModes ))
-       ? minimum_size : sizeof( __GLcontextModes );
-   __GLcontextModes * base = NULL;
-   __GLcontextModes ** next;
-   unsigned   i;
-
-   next = & base;
-   for ( i = 0 ; i < count ; i++ ) {
-      *next = (__GLcontextModes *) _mesa_malloc( size );
-      if ( *next == NULL ) {
-	 _gl_context_modes_destroy( base );
-	 base = NULL;
-	 break;
+   const size_t size = (minimum_size > sizeof(__GLcontextModes))
+      ? minimum_size : sizeof(__GLcontextModes);
+   __GLcontextModes *base = NULL;
+   __GLcontextModes **next;
+   unsigned i;
+
+   next = &base;
+   for (i = 0; i < count; i++) {
+      *next = (__GLcontextModes *) _mesa_malloc(size);
+      if (*next == NULL) {
+         _gl_context_modes_destroy(base);
+         base = NULL;
+         break;
       }
-      
-      (void) _mesa_memset( *next, 0, size );
+
+      (void) _mesa_memset(*next, 0, size);
       (*next)->visualID = GLX_DONT_CARE;
       (*next)->visualType = GLX_DONT_CARE;
       (*next)->visualRating = GLX_NONE;
@@ -417,10 +418,10 @@ _gl_context_modes_create( unsigned count, size_t minimum_size )
       (*next)->bindToTextureRgb = GLX_DONT_CARE;
       (*next)->bindToTextureRgba = GLX_DONT_CARE;
       (*next)->bindToMipmapTexture = GLX_DONT_CARE;
-      (*next)->bindToTextureTargets = 0;
+      (*next)->bindToTextureTargets = GLX_DONT_CARE;
       (*next)->yInverted = GLX_DONT_CARE;
 
-      next = & ((*next)->next);
+      next = &((*next)->next);
    }
 
    return base;
@@ -435,12 +436,12 @@ _gl_context_modes_create( unsigned count, size_t minimum_size )
  *               in the list will be freed.
  */
 void
-_gl_context_modes_destroy( __GLcontextModes * modes )
+_gl_context_modes_destroy(__GLcontextModes * modes)
 {
-   while ( modes != NULL ) {
-      __GLcontextModes * const next = modes->next;
+   while (modes != NULL) {
+      __GLcontextModes *const next = modes->next;
 
-      _mesa_free( modes );
+      _mesa_free(modes);
       modes = next;
    }
 }
@@ -456,19 +457,28 @@ _gl_context_modes_destroy( __GLcontextModes * modes )
  */
 
 __GLcontextModes *
-_gl_context_modes_find_visual( __GLcontextModes * modes, int vid )
+_gl_context_modes_find_visual(__GLcontextModes * modes, int vid)
 {
-    while ( modes != NULL ) {
-	if ( modes->visualID == vid ) {
-	    break;
-	}
+   __GLcontextModes *m;
 
-	modes = modes->next;
-    }
+   for (m = modes; m != NULL; m = m->next)
+      if (m->visualID == vid)
+         return m;
 
-    return modes;
+   return NULL;
 }
 
+__GLcontextModes *
+_gl_context_modes_find_fbconfig(__GLcontextModes * modes, int fbid)
+{
+   __GLcontextModes *m;
+
+   for (m = modes; m != NULL; m = m->next)
+      if (m->fbconfigID == fbid)
+         return m;
+
+   return NULL;
+}
 
 /**
  * Determine if two context-modes are the same.  This is intended to be used
@@ -480,61 +490,55 @@ _gl_context_modes_find_visual( __GLcontextModes * modes, int vid )
  *          returned otherwise.
  */
 GLboolean
-_gl_context_modes_are_same( const __GLcontextModes * a,
-			    const __GLcontextModes * b )
+_gl_context_modes_are_same(const __GLcontextModes * a,
+                           const __GLcontextModes * b)
 {
-    return( (a->rgbMode == b->rgbMode) &&
-	    (a->floatMode == b->floatMode) &&
-	    (a->colorIndexMode == b->colorIndexMode) &&
-	    (a->doubleBufferMode == b->doubleBufferMode) &&
-	    (a->stereoMode == b->stereoMode) &&
-	    (a->redBits == b->redBits) &&
-	    (a->greenBits == b->greenBits) &&
-	    (a->blueBits == b->blueBits) &&
-	    (a->alphaBits == b->alphaBits) &&
-#if 0 /* For some reason these don't get set on the client-side in libGL. */
-	    (a->redMask == b->redMask) &&
-	    (a->greenMask == b->greenMask) &&
-	    (a->blueMask == b->blueMask) &&
-	    (a->alphaMask == b->alphaMask) &&
+   return ((a->rgbMode == b->rgbMode) &&
+           (a->floatMode == b->floatMode) &&
+           (a->colorIndexMode == b->colorIndexMode) &&
+           (a->doubleBufferMode == b->doubleBufferMode) &&
+           (a->stereoMode == b->stereoMode) &&
+           (a->redBits == b->redBits) &&
+           (a->greenBits == b->greenBits) &&
+           (a->blueBits == b->blueBits) && (a->alphaBits == b->alphaBits) &&
+#if 0                           /* For some reason these don't get set on the client-side in libGL. */
+           (a->redMask == b->redMask) &&
+           (a->greenMask == b->greenMask) &&
+           (a->blueMask == b->blueMask) && (a->alphaMask == b->alphaMask) &&
 #endif
-	    (a->rgbBits == b->rgbBits) &&
-	    (a->indexBits == b->indexBits) &&
-	    (a->accumRedBits == b->accumRedBits) &&
-	    (a->accumGreenBits == b->accumGreenBits) &&
-	    (a->accumBlueBits == b->accumBlueBits) &&
-	    (a->accumAlphaBits == b->accumAlphaBits) &&
-	    (a->depthBits == b->depthBits) &&
-	    (a->stencilBits == b->stencilBits) &&
-	    (a->numAuxBuffers == b->numAuxBuffers) &&
-	    (a->level == b->level) &&
-	    (a->pixmapMode == b->pixmapMode) &&
-	    (a->visualRating == b->visualRating) &&
-
-	    (a->transparentPixel == b->transparentPixel) &&
-
-	    ((a->transparentPixel != GLX_TRANSPARENT_RGB) ||
-	     ((a->transparentRed == b->transparentRed) &&
-	      (a->transparentGreen == b->transparentGreen) &&
-	      (a->transparentBlue == b->transparentBlue) &&
-	      (a->transparentAlpha == b->transparentAlpha))) &&
-
-	    ((a->transparentPixel != GLX_TRANSPARENT_INDEX) || 
-	     (a->transparentIndex == b->transparentIndex)) &&
-
-	    (a->sampleBuffers == b->sampleBuffers) &&
-	    (a->samples == b->samples) &&
-	    ((a->drawableType & b->drawableType) != 0) &&
-	    (a->renderType == b->renderType) &&
-	    (a->maxPbufferWidth == b->maxPbufferWidth) &&
-	    (a->maxPbufferHeight == b->maxPbufferHeight) &&
-	    (a->maxPbufferPixels == b->maxPbufferPixels) &&
-	    (a->optimalPbufferWidth == b->optimalPbufferWidth) &&
-	    (a->optimalPbufferHeight == b->optimalPbufferHeight) &&
-	    (a->swapMethod == b->swapMethod) &&
-	    (a->bindToTextureRgb == b->bindToTextureRgb) &&
-	    (a->bindToTextureRgba == b->bindToTextureRgba) &&
-	    (a->bindToMipmapTexture == b->bindToMipmapTexture) &&
-	    (a->bindToTextureTargets == b->bindToTextureTargets) &&
-	    (a->yInverted == b->yInverted) );
+           (a->rgbBits == b->rgbBits) &&
+           (a->indexBits == b->indexBits) &&
+           (a->accumRedBits == b->accumRedBits) &&
+           (a->accumGreenBits == b->accumGreenBits) &&
+           (a->accumBlueBits == b->accumBlueBits) &&
+           (a->accumAlphaBits == b->accumAlphaBits) &&
+           (a->depthBits == b->depthBits) &&
+           (a->stencilBits == b->stencilBits) &&
+           (a->numAuxBuffers == b->numAuxBuffers) &&
+           (a->level == b->level) &&
+           (a->pixmapMode == b->pixmapMode) &&
+           (a->visualRating == b->visualRating) &&
+           (a->transparentPixel == b->transparentPixel) &&
+           ((a->transparentPixel != GLX_TRANSPARENT_RGB) ||
+            ((a->transparentRed == b->transparentRed) &&
+             (a->transparentGreen == b->transparentGreen) &&
+             (a->transparentBlue == b->transparentBlue) &&
+             (a->transparentAlpha == b->transparentAlpha))) &&
+           ((a->transparentPixel != GLX_TRANSPARENT_INDEX) ||
+            (a->transparentIndex == b->transparentIndex)) &&
+           (a->sampleBuffers == b->sampleBuffers) &&
+           (a->samples == b->samples) &&
+           ((a->drawableType & b->drawableType) != 0) &&
+           (a->renderType == b->renderType) &&
+           (a->maxPbufferWidth == b->maxPbufferWidth) &&
+           (a->maxPbufferHeight == b->maxPbufferHeight) &&
+           (a->maxPbufferPixels == b->maxPbufferPixels) &&
+           (a->optimalPbufferWidth == b->optimalPbufferWidth) &&
+           (a->optimalPbufferHeight == b->optimalPbufferHeight) &&
+           (a->swapMethod == b->swapMethod) &&
+           (a->bindToTextureRgb == b->bindToTextureRgb) &&
+           (a->bindToTextureRgba == b->bindToTextureRgba) &&
+           (a->bindToMipmapTexture == b->bindToMipmapTexture) &&
+           (a->bindToTextureTargets == b->bindToTextureTargets) &&
+           (a->yInverted == b->yInverted));
 }
diff --git a/src/glx/x11/glcontextmodes.h b/src/glx/x11/glcontextmodes.h
index 4b5c6f68b8f..6676ae306c2 100644
--- a/src/glx/x11/glcontextmodes.h
+++ b/src/glx/x11/glcontextmodes.h
@@ -33,20 +33,22 @@
 #include "GL/internal/glcore.h"
 
 #if !defined(IN_MINI_GLX)
-extern GLint _gl_convert_from_x_visual_type( int visualType );
-extern GLint _gl_convert_to_x_visual_type( int visualType );
-extern void _gl_copy_visual_to_context_mode( __GLcontextModes * mode,
-    const __GLXvisualConfig * config );
-extern int _gl_get_context_mode_data( const __GLcontextModes *mode,
-    int attribute, int *value_return );
+extern GLint _gl_convert_from_x_visual_type(int visualType);
+extern GLint _gl_convert_to_x_visual_type(int visualType);
+extern void _gl_copy_visual_to_context_mode(__GLcontextModes * mode,
+                                            const __GLXvisualConfig * config);
+extern int _gl_get_context_mode_data(const __GLcontextModes * mode,
+                                     int attribute, int *value_return);
 #endif /* !defined(IN_MINI_GLX) */
 
-extern __GLcontextModes * _gl_context_modes_create( unsigned count,
-    size_t minimum_size );
-extern void _gl_context_modes_destroy( __GLcontextModes * modes );
-extern  __GLcontextModes * _gl_context_modes_find_visual(
-    __GLcontextModes * modes, int vid );
-extern GLboolean _gl_context_modes_are_same( const __GLcontextModes * a,
-    const __GLcontextModes * b );
+extern __GLcontextModes *_gl_context_modes_create(unsigned count,
+                                                  size_t minimum_size);
+extern void _gl_context_modes_destroy(__GLcontextModes * modes);
+extern __GLcontextModes *_gl_context_modes_find_visual(__GLcontextModes *
+                                                       modes, int vid);
+extern __GLcontextModes *_gl_context_modes_find_fbconfig(__GLcontextModes *
+                                                         modes, int fbid);
+extern GLboolean _gl_context_modes_are_same(const __GLcontextModes * a,
+                                            const __GLcontextModes * b);
 
 #endif /* GLCONTEXTMODES_H */
diff --git a/src/glx/x11/glx_pbuffer.c b/src/glx/x11/glx_pbuffer.c
index 1df2d0f342a..c63d53439d5 100644
--- a/src/glx/x11/glx_pbuffer.c
+++ b/src/glx/x11/glx_pbuffer.c
@@ -38,7 +38,6 @@
 #include "glapi.h"
 #include "glxextensions.h"
 #include "glcontextmodes.h"
-#include "glheader.h"
 
 
 /**
@@ -55,14 +54,14 @@
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static void
-ChangeDrawableAttribute( Display * dpy, GLXDrawable drawable,
-			 const CARD32 * attribs, size_t num_attribs )
+ChangeDrawableAttribute(Display * dpy, GLXDrawable drawable,
+                        const CARD32 * attribs, size_t num_attribs)
 {
    __GLXdisplayPrivate *priv = __glXInitialize(dpy);
-   CARD32 * output;
+   CARD32 *output;
    CARD8 opcode;
 
-   if ( (dpy == NULL) || (drawable == 0) ) {
+   if ((dpy == NULL) || (drawable == 0)) {
       return;
    }
 
@@ -72,10 +71,10 @@ ChangeDrawableAttribute( Display * dpy, GLXDrawable drawable,
 
    LockDisplay(dpy);
 
-   if ( (priv->majorVersion > 1) || (priv->minorVersion >= 3) ) {
+   if ((priv->majorVersion > 1) || (priv->minorVersion >= 3)) {
       xGLXChangeDrawableAttributesReq *req;
 
-      GetReqExtra( GLXChangeDrawableAttributes, 8 + (8 * num_attribs), req );
+      GetReqExtra(GLXChangeDrawableAttributes, 8 + (8 * num_attribs), req);
       output = (CARD32 *) (req + 1);
 
       req->reqType = opcode;
@@ -86,7 +85,7 @@ ChangeDrawableAttribute( Display * dpy, GLXDrawable drawable,
    else {
       xGLXVendorPrivateWithReplyReq *vpreq;
 
-      GetReqExtra( GLXVendorPrivateWithReply, 4 + (8 * num_attribs), vpreq );
+      GetReqExtra(GLXVendorPrivateWithReply, 4 + (8 * num_attribs), vpreq);
       output = (CARD32 *) (vpreq + 1);
 
       vpreq->reqType = opcode;
@@ -97,7 +96,7 @@ ChangeDrawableAttribute( Display * dpy, GLXDrawable drawable,
       output++;
    }
 
-   (void) memcpy( output, attribs, sizeof( CARD32 ) * 2 * num_attribs );
+   (void) memcpy(output, attribs, sizeof(CARD32) * 2 * num_attribs);
 
    UnlockDisplay(dpy);
    SyncHandle();
@@ -120,12 +119,12 @@ ChangeDrawableAttribute( Display * dpy, GLXDrawable drawable,
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static void
-DestroyPbuffer( Display * dpy, GLXDrawable drawable )
+DestroyPbuffer(Display * dpy, GLXDrawable drawable)
 {
    __GLXdisplayPrivate *priv = __glXInitialize(dpy);
    CARD8 opcode;
 
-   if ( (dpy == NULL) || (drawable == 0) ) {
+   if ((dpy == NULL) || (drawable == 0)) {
       return;
    }
 
@@ -135,19 +134,19 @@ DestroyPbuffer( Display * dpy, GLXDrawable drawable )
 
    LockDisplay(dpy);
 
-   if ( (priv->majorVersion > 1) || (priv->minorVersion >= 3) ) {
-      xGLXDestroyPbufferReq * req;
+   if ((priv->majorVersion > 1) || (priv->minorVersion >= 3)) {
+      xGLXDestroyPbufferReq *req;
 
-      GetReq( GLXDestroyPbuffer, req );
+      GetReq(GLXDestroyPbuffer, req);
       req->reqType = opcode;
       req->glxCode = X_GLXDestroyPbuffer;
       req->pbuffer = (GLXPbuffer) drawable;
    }
    else {
       xGLXVendorPrivateWithReplyReq *vpreq;
-      CARD32 * data;
+      CARD32 *data;
 
-      GetReqExtra( GLXVendorPrivateWithReply, 4, vpreq );
+      GetReqExtra(GLXVendorPrivateWithReply, 4, vpreq);
       data = (CARD32 *) (vpreq + 1);
 
       data[0] = (CARD32) drawable;
@@ -164,6 +163,34 @@ DestroyPbuffer( Display * dpy, GLXDrawable drawable )
 }
 
 
+#ifdef GLX_DIRECT_RENDERING
+extern __GLXDRIdrawable *GetGLXDRIDrawable(Display * dpy,
+                                           GLXDrawable drawable,
+                                           int *const scrn_num);
+
+static GLenum
+determineTextureTarget(const int *attribs, int numAttribs)
+{
+   GLenum target = 0;
+   int i;
+
+   for (i = 0; i < numAttribs; i++) {
+      if (attribs[2 * i] == GLX_TEXTURE_TARGET_EXT) {
+         switch (attribs[2 * i + 1]) {
+         case GLX_TEXTURE_2D_EXT:
+            target = GL_TEXTURE_2D;
+            break;
+         case GLX_TEXTURE_RECTANGLE_EXT:
+            target = GL_TEXTURE_RECTANGLE_ARB;
+            break;
+         }
+      }
+   }
+
+   return target;
+}
+#endif
+
 /**
  * Get a drawable's attribute.
  *
@@ -183,24 +210,24 @@ DestroyPbuffer( Display * dpy, GLXDrawable drawable )
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static int
-GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
-		      int attribute, unsigned int *value )
+GetDrawableAttribute(Display * dpy, GLXDrawable drawable,
+                     int attribute, unsigned int *value)
 {
    __GLXdisplayPrivate *priv;
    xGLXGetDrawableAttributesReply reply;
-   CARD32 * data;
+   CARD32 *data;
    CARD8 opcode;
    unsigned int length;
    unsigned int i;
    unsigned int num_attributes;
 
-   if ( (dpy == NULL) || (drawable == 0) ) {
+   if ((dpy == NULL) || (drawable == 0)) {
       return 0;
    }
 
    priv = __glXInitialize(dpy);
    GLboolean use_glx_1_3 = ((priv->majorVersion > 1)
-			    || (priv->minorVersion >= 3));
+                            || (priv->minorVersion >= 3));
 
    *value = 0;
 
@@ -211,10 +238,10 @@ GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
 
    LockDisplay(dpy);
 
-   if ( use_glx_1_3 ) {
+   if (use_glx_1_3) {
       xGLXGetDrawableAttributesReq *req;
 
-      GetReqExtra( GLXGetDrawableAttributes, 4, req );
+      GetReqExtra(GLXGetDrawableAttributes, 4, req);
       req->reqType = opcode;
       req->glxCode = X_GLXGetDrawableAttributes;
       req->drawable = drawable;
@@ -222,7 +249,7 @@ GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
    else {
       xGLXVendorPrivateWithReplyReq *vpreq;
 
-      GetReqExtra( GLXVendorPrivateWithReply, 4, vpreq );
+      GetReqExtra(GLXVendorPrivateWithReply, 4, vpreq);
       data = (CARD32 *) (vpreq + 1);
       data[0] = (CARD32) drawable;
 
@@ -231,38 +258,47 @@ GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
       vpreq->vendorCode = X_GLXvop_GetDrawableAttributesSGIX;
    }
 
-   _XReply(dpy, (xReply*) &reply, 0, False);
+   _XReply(dpy, (xReply *) & reply, 0, False);
 
-   if (reply.type == X_Error)
-   {
-       UnlockDisplay(dpy);
-       SyncHandle();
-       return 0;
+   if (reply.type == X_Error) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+      return 0;
    }
 
    length = reply.length;
-   if (length)
-   {
-       num_attributes = (use_glx_1_3) ? reply.numAttribs : length / 2;
-       data = (CARD32 *) Xmalloc( length * sizeof(CARD32) );
-       if ( data == NULL ) {
-	   /* Throw data on the floor */
-	   _XEatData(dpy, length);
-       } else {
-	   _XRead(dpy, (char *)data, length * sizeof(CARD32) );
-
-	   /* Search the set of returned attributes for the attribute requested by
-	    * the caller.
-	    */
-	   for ( i = 0 ; i < num_attributes ; i++ ) {
-	       if ( data[i*2] == attribute ) {
-		   *value = data[ (i*2) + 1 ];
-		   break;
-	       }
-	   }
-
-	   Xfree( data );
-       }
+   if (length) {
+      num_attributes = (use_glx_1_3) ? reply.numAttribs : length / 2;
+      data = (CARD32 *) Xmalloc(length * sizeof(CARD32));
+      if (data == NULL) {
+         /* Throw data on the floor */
+         _XEatData(dpy, length);
+      }
+      else {
+         _XRead(dpy, (char *) data, length * sizeof(CARD32));
+
+         /* Search the set of returned attributes for the attribute requested by
+          * the caller.
+          */
+         for (i = 0; i < num_attributes; i++) {
+            if (data[i * 2] == attribute) {
+               *value = data[(i * 2) + 1];
+               break;
+            }
+         }
+
+#ifdef GLX_DIRECT_RENDERING
+         {
+            __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, NULL);
+
+            if (pdraw != NULL && !pdraw->textureTarget)
+               pdraw->textureTarget =
+                  determineTextureTarget((const int *) data, num_attributes);
+         }
+#endif
+
+         Xfree(data);
+      }
    }
 
    UnlockDisplay(dpy);
@@ -271,7 +307,6 @@ GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
    return 0;
 }
 
-
 /**
  * Create a non-pbuffer GLX drawable.
  *
@@ -279,19 +314,18 @@ GetDrawableAttribute( Display *dpy, GLXDrawable drawable,
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static GLXDrawable
-CreateDrawable( Display *dpy, const __GLcontextModes * fbconfig,
-		Drawable drawable, const int *attrib_list,
-		CARD8 glxCode )
+CreateDrawable(Display * dpy, const __GLcontextModes * fbconfig,
+               Drawable drawable, const int *attrib_list, CARD8 glxCode)
 {
-   xGLXCreateWindowReq * req;
-   CARD32 * data;
+   xGLXCreateWindowReq *req;
+   CARD32 *data;
    unsigned int i;
    CARD8 opcode;
 
    i = 0;
    if (attrib_list) {
-       while (attrib_list[i * 2] != None)
-	   i++;
+      while (attrib_list[i * 2] != None)
+         i++;
    }
 
    opcode = __glXSetupForCommand(dpy);
@@ -299,23 +333,51 @@ CreateDrawable( Display *dpy, const __GLcontextModes * fbconfig,
       return None;
 
    LockDisplay(dpy);
-   GetReqExtra( GLXCreateWindow, 8 * i, req );
+   GetReqExtra(GLXCreateWindow, 8 * i, req);
    data = (CARD32 *) (req + 1);
 
    req->reqType = opcode;
    req->glxCode = glxCode;
    req->screen = (CARD32) fbconfig->screen;
    req->fbconfig = fbconfig->fbconfigID;
-   req->window = (GLXPbuffer) drawable;
+   req->window = (CARD32) drawable;
    req->glxwindow = (GLXWindow) XAllocID(dpy);
    req->numAttribs = (CARD32) i;
 
-   memcpy( data, attrib_list, 8 * i );
-   
+   memcpy(data, attrib_list, 8 * i);
+
    UnlockDisplay(dpy);
    SyncHandle();
-   
-   return (GLXDrawable)req->glxwindow;
+
+#ifdef GLX_DIRECT_RENDERING
+   do {
+      /* FIXME: Maybe delay __DRIdrawable creation until the drawable
+       * is actually bound to a context... */
+
+      __GLXdisplayPrivate *const priv = __glXInitialize(dpy);
+      __GLXDRIdrawable *pdraw;
+      __GLXscreenConfigs *psc;
+
+      psc = &priv->screenConfigs[fbconfig->screen];
+      if (psc->driScreen == NULL)
+         break;
+      pdraw = psc->driScreen->createDrawable(psc, drawable,
+                                             req->glxwindow, fbconfig);
+      if (pdraw == NULL) {
+         fprintf(stderr, "failed to create drawable\n");
+         break;
+      }
+
+      if (__glxHashInsert(psc->drawHash, req->glxwindow, pdraw)) {
+         (*pdraw->destroyDrawable) (pdraw);
+         return None;           /* FIXME: Check what we're supposed to do here... */
+      }
+
+      pdraw->textureTarget = determineTextureTarget(attrib_list, i);
+   } while (0);
+#endif
+
+   return (GLXDrawable) req->glxwindow;
 }
 
 
@@ -326,12 +388,12 @@ CreateDrawable( Display *dpy, const __GLcontextModes * fbconfig,
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static void
-DestroyDrawable( Display * dpy, GLXDrawable drawable, CARD32 glxCode )
+DestroyDrawable(Display * dpy, GLXDrawable drawable, CARD32 glxCode)
 {
-   xGLXDestroyPbufferReq * req;
+   xGLXDestroyPbufferReq *req;
    CARD8 opcode;
 
-   if ( (dpy == NULL) || (drawable == 0) ) {
+   if ((dpy == NULL) || (drawable == 0)) {
       return;
    }
 
@@ -342,7 +404,7 @@ DestroyDrawable( Display * dpy, GLXDrawable drawable, CARD32 glxCode )
 
    LockDisplay(dpy);
 
-   GetReqExtra( GLXDestroyPbuffer, 4, req );
+   GetReqExtra(GLXDestroyPbuffer, 4, req);
    req->reqType = opcode;
    req->glxCode = glxCode;
    req->pbuffer = (GLXPbuffer) drawable;
@@ -350,6 +412,20 @@ DestroyDrawable( Display * dpy, GLXDrawable drawable, CARD32 glxCode )
    UnlockDisplay(dpy);
    SyncHandle();
 
+#ifdef GLX_DIRECT_RENDERING
+   {
+      int screen;
+      __GLXdisplayPrivate *const priv = __glXInitialize(dpy);
+      __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
+      __GLXscreenConfigs *psc = &priv->screenConfigs[screen];
+
+      if (pdraw != NULL) {
+         (*pdraw->destroyDrawable) (pdraw);
+         __glxHashDelete(psc->drawHash, drawable);
+      }
+   }
+#endif
+
    return;
 }
 
@@ -368,20 +444,20 @@ DestroyDrawable( Display * dpy, GLXDrawable drawable, CARD32 glxCode )
  * This function needs to be modified to work with direct-rendering drivers.
  */
 static GLXDrawable
-CreatePbuffer( Display *dpy, const __GLcontextModes * fbconfig,
-	       unsigned int width, unsigned int height, 
-	       const int *attrib_list, GLboolean size_in_attribs )
+CreatePbuffer(Display * dpy, const __GLcontextModes * fbconfig,
+              unsigned int width, unsigned int height,
+              const int *attrib_list, GLboolean size_in_attribs)
 {
    __GLXdisplayPrivate *priv = __glXInitialize(dpy);
    GLXDrawable id = 0;
-   CARD32 * data;
+   CARD32 *data;
    CARD8 opcode;
-   unsigned int  i;
+   unsigned int i;
 
    i = 0;
    if (attrib_list) {
-       while (attrib_list[i * 2])
-	   i++;
+      while (attrib_list[i * 2])
+         i++;
    }
 
    opcode = __glXSetupForCommand(dpy);
@@ -391,11 +467,11 @@ CreatePbuffer( Display *dpy, const __GLcontextModes * fbconfig,
    LockDisplay(dpy);
    id = XAllocID(dpy);
 
-   if ( (priv->majorVersion > 1) || (priv->minorVersion >= 3) ) {
-      xGLXCreatePbufferReq * req;
+   if ((priv->majorVersion > 1) || (priv->minorVersion >= 3)) {
+      xGLXCreatePbufferReq *req;
       unsigned int extra = (size_in_attribs) ? 0 : 2;
 
-      GetReqExtra( GLXCreatePbuffer, (8 * (i + extra)), req );
+      GetReqExtra(GLXCreatePbuffer, (8 * (i + extra)), req);
       data = (CARD32 *) (req + 1);
 
       req->reqType = opcode;
@@ -405,18 +481,18 @@ CreatePbuffer( Display *dpy, const __GLcontextModes * fbconfig,
       req->pbuffer = (GLXPbuffer) id;
       req->numAttribs = (CARD32) (i + extra);
 
-      if ( ! size_in_attribs ) {
-	 data[(2 * i) + 0] = GLX_PBUFFER_WIDTH;
-	 data[(2 * i) + 1] = width;
-	 data[(2 * i) + 2] = GLX_PBUFFER_HEIGHT;
-	 data[(2 * i) + 3] = height;
-	 data += 4;
+      if (!size_in_attribs) {
+         data[(2 * i) + 0] = GLX_PBUFFER_WIDTH;
+         data[(2 * i) + 1] = width;
+         data[(2 * i) + 2] = GLX_PBUFFER_HEIGHT;
+         data[(2 * i) + 3] = height;
+         data += 4;
       }
    }
    else {
       xGLXVendorPrivateReq *vpreq;
 
-      GetReqExtra( GLXVendorPrivate, 20 + (8 * i), vpreq );
+      GetReqExtra(GLXVendorPrivate, 20 + (8 * i), vpreq);
       data = (CARD32 *) (vpreq + 1);
 
       vpreq->reqType = opcode;
@@ -431,7 +507,7 @@ CreatePbuffer( Display *dpy, const __GLcontextModes * fbconfig,
       data += 5;
    }
 
-   (void) memcpy( data, attrib_list, sizeof(CARD32) * 2 * i );
+   (void) memcpy(data, attrib_list, sizeof(CARD32) * 2 * i);
 
    UnlockDisplay(dpy);
    SyncHandle();
@@ -444,13 +520,13 @@ CreatePbuffer( Display *dpy, const __GLcontextModes * fbconfig,
  * Create a new pbuffer.
  */
 PUBLIC GLXPbufferSGIX
-glXCreateGLXPbufferSGIX(Display *dpy, GLXFBConfigSGIX config,
-			unsigned int width, unsigned int height,
-			int *attrib_list)
+glXCreateGLXPbufferSGIX(Display * dpy, GLXFBConfigSGIX config,
+                        unsigned int width, unsigned int height,
+                        int *attrib_list)
 {
-   return (GLXPbufferSGIX) CreatePbuffer( dpy, (__GLcontextModes *) config,
-					  width, height,
-					  attrib_list, GL_FALSE );
+   return (GLXPbufferSGIX) CreatePbuffer(dpy, (__GLcontextModes *) config,
+                                         width, height,
+                                         attrib_list, GL_FALSE);
 }
 
 
@@ -458,11 +534,26 @@ glXCreateGLXPbufferSGIX(Display *dpy, GLXFBConfigSGIX config,
  * Create a new pbuffer.
  */
 PUBLIC GLXPbuffer
-glXCreatePbuffer(Display *dpy, GLXFBConfig config, const int *attrib_list)
+glXCreatePbuffer(Display * dpy, GLXFBConfig config, const int *attrib_list)
 {
-   return (GLXPbuffer) CreatePbuffer( dpy, (__GLcontextModes *) config,
-				      0, 0,
-				      attrib_list, GL_TRUE );
+   int i, width, height;
+
+   width = 0;
+   height = 0;
+
+   for (i = 0; attrib_list[i * 2]; i++) {
+      switch (attrib_list[i * 2]) {
+      case GLX_PBUFFER_WIDTH:
+         width = attrib_list[i * 2 + 1];
+         break;
+      case GLX_PBUFFER_HEIGHT:
+         height = attrib_list[i * 2 + 1];
+         break;
+      }
+   }
+
+   return (GLXPbuffer) CreatePbuffer(dpy, (__GLcontextModes *) config,
+                                     width, height, attrib_list, GL_TRUE);
 }
 
 
@@ -470,9 +561,9 @@ glXCreatePbuffer(Display *dpy, GLXFBConfig config, const int *attrib_list)
  * Destroy an existing pbuffer.
  */
 PUBLIC void
-glXDestroyPbuffer(Display *dpy, GLXPbuffer pbuf)
+glXDestroyPbuffer(Display * dpy, GLXPbuffer pbuf)
 {
-   DestroyPbuffer( dpy, pbuf );
+   DestroyPbuffer(dpy, pbuf);
 }
 
 
@@ -480,10 +571,10 @@ glXDestroyPbuffer(Display *dpy, GLXPbuffer pbuf)
  * Query an attribute of a drawable.
  */
 PUBLIC void
-glXQueryDrawable(Display *dpy, GLXDrawable drawable,
-		 int attribute, unsigned int *value)
+glXQueryDrawable(Display * dpy, GLXDrawable drawable,
+                 int attribute, unsigned int *value)
 {
-   GetDrawableAttribute( dpy, drawable, attribute, value );
+   GetDrawableAttribute(dpy, drawable, attribute, value);
 }
 
 
@@ -491,10 +582,10 @@ glXQueryDrawable(Display *dpy, GLXDrawable drawable,
  * Query an attribute of a pbuffer.
  */
 PUBLIC int
-glXQueryGLXPbufferSGIX(Display *dpy, GLXPbufferSGIX drawable,
-		       int attribute, unsigned int *value)
+glXQueryGLXPbufferSGIX(Display * dpy, GLXPbufferSGIX drawable,
+                       int attribute, unsigned int *value)
 {
-   return GetDrawableAttribute( dpy, drawable, attribute, value );
+   return GetDrawableAttribute(dpy, drawable, attribute, value);
 }
 
 
@@ -502,14 +593,14 @@ glXQueryGLXPbufferSGIX(Display *dpy, GLXPbufferSGIX drawable,
  * Select the event mask for a drawable.
  */
 PUBLIC void
-glXSelectEvent(Display *dpy, GLXDrawable drawable, unsigned long mask)
+glXSelectEvent(Display * dpy, GLXDrawable drawable, unsigned long mask)
 {
    CARD32 attribs[2];
 
    attribs[0] = (CARD32) GLX_EVENT_MASK;
    attribs[1] = (CARD32) mask;
 
-   ChangeDrawableAttribute( dpy, drawable, attribs, 1 );
+   ChangeDrawableAttribute(dpy, drawable, attribs, 1);
 }
 
 
@@ -517,7 +608,7 @@ glXSelectEvent(Display *dpy, GLXDrawable drawable, unsigned long mask)
  * Get the selected event mask for a drawable.
  */
 PUBLIC void
-glXGetSelectedEvent(Display *dpy, GLXDrawable drawable, unsigned long *mask)
+glXGetSelectedEvent(Display * dpy, GLXDrawable drawable, unsigned long *mask)
 {
    unsigned int value;
 
@@ -527,56 +618,56 @@ glXGetSelectedEvent(Display *dpy, GLXDrawable drawable, unsigned long *mask)
     * we could just type-cast the pointer, but why?
     */
 
-   GetDrawableAttribute( dpy, drawable, GLX_EVENT_MASK_SGIX, & value );
+   GetDrawableAttribute(dpy, drawable, GLX_EVENT_MASK_SGIX, &value);
    *mask = value;
 }
 
 
 PUBLIC GLXPixmap
-glXCreatePixmap( Display *dpy, GLXFBConfig config, Pixmap pixmap,
-		 const int *attrib_list )
+glXCreatePixmap(Display * dpy, GLXFBConfig config, Pixmap pixmap,
+                const int *attrib_list)
 {
-   return CreateDrawable( dpy, (__GLcontextModes *) config,
-			  (Drawable) pixmap, attrib_list,
-			  X_GLXCreatePixmap );
+   return CreateDrawable(dpy, (__GLcontextModes *) config,
+                         (Drawable) pixmap, attrib_list, X_GLXCreatePixmap);
 }
 
 
 PUBLIC GLXWindow
-glXCreateWindow( Display *dpy, GLXFBConfig config, Window win,
-		 const int *attrib_list )
+glXCreateWindow(Display * dpy, GLXFBConfig config, Window win,
+                const int *attrib_list)
 {
-   return CreateDrawable( dpy, (__GLcontextModes *) config,
-			  (Drawable) win, attrib_list,
-			  X_GLXCreateWindow );
+   return CreateDrawable(dpy, (__GLcontextModes *) config,
+                         (Drawable) win, attrib_list, X_GLXCreateWindow);
 }
 
 
 PUBLIC void
-glXDestroyPixmap(Display *dpy, GLXPixmap pixmap)
+glXDestroyPixmap(Display * dpy, GLXPixmap pixmap)
 {
-   DestroyDrawable( dpy, (GLXDrawable) pixmap, X_GLXDestroyPixmap );
+   DestroyDrawable(dpy, (GLXDrawable) pixmap, X_GLXDestroyPixmap);
 }
 
 
 PUBLIC void
-glXDestroyWindow(Display *dpy, GLXWindow win)
+glXDestroyWindow(Display * dpy, GLXWindow win)
 {
-   DestroyDrawable( dpy, (GLXDrawable) win, X_GLXDestroyWindow );
+   DestroyDrawable(dpy, (GLXDrawable) win, X_GLXDestroyWindow);
 }
 
 
-PUBLIC GLX_ALIAS_VOID(glXDestroyGLXPbufferSGIX,
-	  (Display *dpy, GLXPbufferSGIX pbuf),
-	  (dpy, pbuf),
-	  glXDestroyPbuffer)
+PUBLIC
+GLX_ALIAS_VOID(glXDestroyGLXPbufferSGIX,
+               (Display * dpy, GLXPbufferSGIX pbuf),
+               (dpy, pbuf), glXDestroyPbuffer)
 
-PUBLIC GLX_ALIAS_VOID(glXSelectEventSGIX,
-	  (Display *dpy, GLXDrawable drawable, unsigned long mask),
-	  (dpy, drawable, mask),
-	  glXSelectEvent)
+PUBLIC
+GLX_ALIAS_VOID(glXSelectEventSGIX,
+               (Display * dpy, GLXDrawable drawable,
+                unsigned long mask), (dpy, drawable, mask),
+               glXSelectEvent)
 
-PUBLIC GLX_ALIAS_VOID(glXGetSelectedEventSGIX,
-	  (Display *dpy, GLXDrawable drawable, unsigned long *mask),
-	  (dpy, drawable, mask),
-	  glXGetSelectedEvent)
+PUBLIC
+GLX_ALIAS_VOID(glXGetSelectedEventSGIX,
+               (Display * dpy, GLXDrawable drawable,
+                unsigned long *mask), (dpy, drawable, mask),
+               glXGetSelectedEvent)
diff --git a/src/glx/x11/glx_query.c b/src/glx/x11/glx_query.c
index e93cd2afd43..2789b841b1b 100644
--- a/src/glx/x11/glx_query.c
+++ b/src/glx/x11/glx_query.c
@@ -31,6 +31,70 @@
 
 #include "glxclient.h"
 
+#if defined(USE_XCB)
+# include <X11/Xlib-xcb.h>
+# include <xcb/xcb.h>
+# include <xcb/glx.h>
+#endif
+
+#ifdef USE_XCB
+
+/**
+ * Exchange a protocol request for glXQueryServerString.
+ */
+char *
+__glXQueryServerString(Display* dpy,
+                       int opcode,
+                       CARD32 screen,
+                       CARD32 name)
+{
+   xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_glx_query_server_string_reply_t* reply =
+      xcb_glx_query_server_string_reply(c,
+                                        xcb_glx_query_server_string(c,
+                                                                    screen,
+                                                                    name),
+                                        NULL);
+
+   /* The spec doesn't mention this, but the Xorg server replies with
+    * a string already terminated with '\0'. */
+   uint32_t len = xcb_glx_query_server_string_string_length(reply);
+   char* buf = Xmalloc(len);
+   memcpy(buf, xcb_glx_query_server_string_string(reply), len);
+   free(reply);
+
+   return buf;
+}
+
+/**
+ * Exchange a protocol request for glGetString.
+ */
+char *
+__glXGetString(Display* dpy,
+               int opcode,
+               CARD32 contextTag,
+               CARD32 name)
+{
+   xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_glx_get_string_reply_t* reply =
+      xcb_glx_get_string_reply(c,
+                               xcb_glx_get_string(c,
+                                                  contextTag,
+                                                  name),
+                               NULL);
+
+   /* The spec doesn't mention this, but the Xorg server replies with
+    * a string already terminated with '\0'. */
+   uint32_t len = xcb_glx_get_string_string_length(reply);
+   char* buf = Xmalloc(len);
+   memcpy(buf, xcb_glx_get_string_string(reply), len);
+   free(reply);
+
+   return buf;
+}
+
+#else
+
 /**
  * GLX protocol structure for the ficticious "GXLGenericGetString" request.
  * 
@@ -39,12 +103,13 @@
  * an identical binary layout.  The only difference between them is the
  * meaning of the \c for_whom field and the value of the \c glxCode.
  */
-typedef struct GLXGenericGetString {
-    CARD8 reqType;
-    CARD8 glxCode;
-    CARD16 length B16;
-    CARD32 for_whom B32;
-    CARD32 name B32;
+typedef struct GLXGenericGetString
+{
+   CARD8 reqType;
+   CARD8 glxCode;
+   CARD16 length B16;
+   CARD32 for_whom B32;
+   CARD32 name B32;
 } xGLXGenericGetStringReq;
 
 /* These defines are only needed to make the GetReq macro happy.
@@ -53,50 +118,74 @@ typedef struct GLXGenericGetString {
 #define X_GLXGenericGetString 0
 
 /**
- * Query the Server GLX string and cache it in the display private.
+ * Query the Server GLX string.
  * This routine will allocate the necessay space for the string.
  */
-char *
-__glXGetStringFromServer( Display * dpy, int opcode, CARD32 glxCode,
-			  CARD32 for_whom, CARD32 name )
+static char *
+__glXGetStringFromServer(Display * dpy, int opcode, CARD32 glxCode,
+                         CARD32 for_whom, CARD32 name)
 {
-    xGLXGenericGetStringReq *req;
-    xGLXSingleReply reply;
-    int length;
-    int numbytes;
-    char * buf;
+   xGLXGenericGetStringReq *req;
+   xGLXSingleReply reply;
+   int length;
+   int numbytes;
+   char *buf;
 
 
-    LockDisplay( dpy );
+   LockDisplay(dpy);
 
 
-    /* All of the GLX protocol requests for getting a string from the server
-     * look the same.  The exact meaning of the for_whom field is usually
-     * either the screen number (for glXQueryServerString) or the context tag
-     * (for GLXSingle).
-     */
+   /* All of the GLX protocol requests for getting a string from the server
+    * look the same.  The exact meaning of the for_whom field is usually
+    * either the screen number (for glXQueryServerString) or the context tag
+    * (for GLXSingle).
+    */
 
-    GetReq( GLXGenericGetString, req );
-    req->reqType = opcode;
-    req->glxCode = glxCode;
-    req->for_whom = for_whom;
-    req->name = name;
+   GetReq(GLXGenericGetString, req);
+   req->reqType = opcode;
+   req->glxCode = glxCode;
+   req->for_whom = for_whom;
+   req->name = name;
 
-    _XReply( dpy, (xReply *) & reply, 0, False );
+   _XReply(dpy, (xReply *) & reply, 0, False);
 
-    length = reply.length * 4;
-    numbytes = reply.size;
+   length = reply.length * 4;
+   numbytes = reply.size;
 
-    buf = (char *) Xmalloc( numbytes );
-    if ( buf != NULL ) {
-        _XRead( dpy, buf, numbytes );
-	length -= numbytes;
-    }
+   buf = (char *) Xmalloc(numbytes);
+   if (buf != NULL) {
+      _XRead(dpy, buf, numbytes);
+      length -= numbytes;
+   }
 
-    _XEatData( dpy, length );
+   _XEatData(dpy, length);
 
-    UnlockDisplay( dpy );
-    SyncHandle();
+   UnlockDisplay(dpy);
+   SyncHandle();
 
-    return buf;
+   return buf;
+}
+
+char *
+__glXQueryServerString(Display* dpy,
+                       int opcode,
+                       CARD32 screen,
+                       CARD32 name)
+{
+   return __glXGetStringFromServer(dpy, opcode,
+                                   X_GLXQueryServerString,
+                                   screen, name);
 }
+
+char *
+__glXGetString(Display* dpy,
+               int opcode,
+               CARD32 contextTag,
+               CARD32 name)
+{
+   return __glXGetStringFromServer(dpy, opcode, X_GLsop_GetString,
+                                   contextTag, name);
+}
+
+#endif /* USE_XCB */
+
diff --git a/src/glx/x11/glx_texture_compression.c b/src/glx/x11/glx_texture_compression.c
deleted file mode 100644
index 56768580177..00000000000
--- a/src/glx/x11/glx_texture_compression.c
+++ /dev/null
@@ -1,347 +0,0 @@
-/*
- * (C) Copyright IBM Corporation 2004
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/**
- * \file glx_texture_compression.c
- * Contains the routines required to implement GLX protocol for
- * ARB_texture_compression and related extensions.
- *
- * \sa http://oss.sgi.com/projects/ogl-sample/registry/ARB/texture_compression.txt
- *
- * \author Ian Romanick <idr@us.ibm.com>
- */
-
-#include "packrender.h"
-#include "packsingle.h"
-#include "indirect.h"
-
-#include <assert.h>
-
-
-void
-__indirect_glGetCompressedTexImageARB( GLenum target, GLint level, 
-				    GLvoid * img )
-{
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXGetTexImageReply reply;
-    size_t image_bytes;
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN( X_GLsop_GetCompressedTexImage, 8 );
-    __GLX_SINGLE_PUT_LONG( 0, target );
-    __GLX_SINGLE_PUT_LONG( 4, level );
-    __GLX_SINGLE_READ_XREPLY();
-
-    image_bytes = reply.width;
-    assert( image_bytes <= ((4 * reply.length) - 0) );
-    assert( image_bytes >= ((4 * reply.length) - 3) );
-    
-    if ( image_bytes != 0 ) {
-	_XRead( dpy, (char *) img, image_bytes );
-	if ( image_bytes < (4 * reply.length) ) {
-	    _XEatData( dpy, (4 * reply.length) - image_bytes );
-	}
-    }
-
-    __GLX_SINGLE_END();
-}
-
-
-/**
- * Internal function used for \c glCompressedTexImage1D and
- * \c glCompressedTexImage2D.
- */
-static void
-CompressedTexImage1D2D( GLenum target, GLint level,
-			GLenum internal_format, 
-			GLsizei width, GLsizei height,
-			GLint border, GLsizei image_size,
-			const GLvoid *data, CARD32 rop )
-{
-    __GLX_DECLARE_VARIABLES();
-
-    __GLX_LOAD_VARIABLES();
-    if ( gc->currentDpy == NULL ) {
-	return;
-    }
-
-    if ( (target == GL_PROXY_TEXTURE_1D)
-	 || (target == GL_PROXY_TEXTURE_2D)
-	 || (target == GL_PROXY_TEXTURE_CUBE_MAP) ) {
-	compsize = 0;
-    }
-    else {
-	compsize = image_size;
-    }
-
-    cmdlen = __GLX_PAD( __GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE 
-			+ compsize );
-    if ( cmdlen <= gc->maxSmallRenderCommandSize ) {
-	__GLX_BEGIN_VARIABLE( rop, cmdlen );
-	__GLX_PUT_LONG( 4, target );
-	__GLX_PUT_LONG( 8, level );
-	__GLX_PUT_LONG( 12, internal_format );
-	__GLX_PUT_LONG( 16, width );
-	__GLX_PUT_LONG( 20, height );
-	__GLX_PUT_LONG( 24, border );
-	__GLX_PUT_LONG( 28, image_size );
-	if ( compsize != 0 ) {
-	    __GLX_PUT_CHAR_ARRAY( __GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE,
-				  data, image_size );
-	}
-	__GLX_END( cmdlen );
-    }
-    else {
-	assert( compsize != 0 );
-
-	__GLX_BEGIN_VARIABLE_LARGE( rop, cmdlen + 4 );
-	__GLX_PUT_LONG( 8, target );
-	__GLX_PUT_LONG( 12, level );
-	__GLX_PUT_LONG( 16, internal_format );
-	__GLX_PUT_LONG( 20, width );
-	__GLX_PUT_LONG( 24, height );
-	__GLX_PUT_LONG( 28, border );
-	__GLX_PUT_LONG( 32, image_size );
-	__glXSendLargeCommand( gc, gc->pc,
-			       __GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE + 4,
-			       data, image_size );
-    }
-}
-
-
-/**
- * Internal function used for \c glCompressedTexSubImage1D and
- * \c glCompressedTexSubImage2D.
- */
-static void
-CompressedTexSubImage1D2D( GLenum target, GLint level,
-			   GLsizei xoffset, GLsizei yoffset,
-			   GLsizei width, GLsizei height,
-			   GLenum format, GLsizei image_size,
-			   const GLvoid *data, CARD32 rop )
-{
-    __GLX_DECLARE_VARIABLES();
-
-    __GLX_LOAD_VARIABLES();
-    if ( gc->currentDpy == NULL ) {
-	return;
-    }
-
-    if ( target == GL_PROXY_TEXTURE_3D ) {
-	compsize = 0;
-    }
-    else {
-	compsize = image_size;
-    }
-
-    cmdlen = __GLX_PAD( __GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE
-			+ compsize );
-    if ( cmdlen <= gc->maxSmallRenderCommandSize ) {
-	__GLX_BEGIN_VARIABLE( rop, cmdlen );
-	__GLX_PUT_LONG( 4, target );
-	__GLX_PUT_LONG( 8, level );
-	__GLX_PUT_LONG( 12, xoffset );
-	__GLX_PUT_LONG( 16, yoffset );
-	__GLX_PUT_LONG( 20, width );
-	__GLX_PUT_LONG( 24, height );
-	__GLX_PUT_LONG( 28, format );
-	__GLX_PUT_LONG( 32, image_size );
-	if ( compsize != 0 ) {
-	    __GLX_PUT_CHAR_ARRAY( __GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE, 
-				  data, image_size );
-	}
-	__GLX_END( cmdlen );
-    }
-    else {
-	assert( compsize != 0 );
-
-	__GLX_BEGIN_VARIABLE_LARGE( rop, cmdlen + 4 );
-	__GLX_PUT_LONG( 8, target );
-	__GLX_PUT_LONG( 12, level );
-	__GLX_PUT_LONG( 16, xoffset );
-	__GLX_PUT_LONG( 20, yoffset );
-	__GLX_PUT_LONG( 24, width );
-	__GLX_PUT_LONG( 28, height );
-	__GLX_PUT_LONG( 32, format );
-	__GLX_PUT_LONG( 36, image_size );
-	__glXSendLargeCommand( gc, gc->pc,
-			       __GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE + 4,
-			       data, image_size );
-    }
-}
-
-
-void
-__indirect_glCompressedTexImage1DARB( GLenum target, GLint level,
-				   GLenum internal_format, GLsizei width,
-				   GLint border, GLsizei image_size,
-				   const GLvoid *data )
-{
-    CompressedTexImage1D2D( target, level, internal_format, width, 0,
-			    border, image_size, data,
-			    X_GLrop_CompressedTexImage1D );
-}
-
-
-void
-__indirect_glCompressedTexImage2DARB( GLenum target, GLint level,
-				   GLenum internal_format,
-				   GLsizei width, GLsizei height,
-				   GLint border, GLsizei image_size,
-				   const GLvoid *data )
-{
-    CompressedTexImage1D2D( target, level, internal_format, width, height,
-			    border, image_size, data,
-			    X_GLrop_CompressedTexImage2D );
-}
-
-
-void
-__indirect_glCompressedTexImage3DARB( GLenum target, GLint level,
-				   GLenum internal_format,
-				   GLsizei width, GLsizei height, GLsizei depth,
-				   GLint border, GLsizei image_size,
-				   const GLvoid *data )
-{
-    __GLX_DECLARE_VARIABLES();
-    
-    __GLX_LOAD_VARIABLES();
-    if ( gc->currentDpy == NULL ) {
-	return;
-    }
-
-    cmdlen = __GLX_PAD( __GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE
-			+ image_size );
-    if ( cmdlen <= gc->maxSmallRenderCommandSize ) {
-	__GLX_BEGIN_VARIABLE( X_GLrop_CompressedTexImage3D, cmdlen );
-	__GLX_PUT_LONG( 4, target );
-	__GLX_PUT_LONG( 8, level );
-	__GLX_PUT_LONG( 12, internal_format );
-	__GLX_PUT_LONG( 16, width );
-	__GLX_PUT_LONG( 20, height );
-	__GLX_PUT_LONG( 24, depth );
-	__GLX_PUT_LONG( 28, border );
-	__GLX_PUT_LONG( 32, image_size );
-	if ( image_size != 0 ) {
-	    __GLX_PUT_CHAR_ARRAY( __GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE,
-				  data, image_size );
-	}
-	__GLX_END( cmdlen );
-    }
-    else {
-	__GLX_BEGIN_VARIABLE_LARGE( X_GLrop_CompressedTexImage3D,
-				    cmdlen + 4 );
-	__GLX_PUT_LONG( 8, target );
-	__GLX_PUT_LONG( 12, level );
-	__GLX_PUT_LONG( 16, internal_format );
-	__GLX_PUT_LONG( 20, width );
-	__GLX_PUT_LONG( 24, height );
-	__GLX_PUT_LONG( 28, depth );
-	__GLX_PUT_LONG( 32, border );
-	__GLX_PUT_LONG( 36, image_size );
-	__glXSendLargeCommand( gc, gc->pc,
-			       __GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE + 4,
-			       data, image_size );
-    }
-}
-
-
-void
-__indirect_glCompressedTexSubImage1DARB( GLenum target, GLint level,
-				      GLint xoffset,
-				      GLsizei width,
-				      GLenum format, GLsizei image_size,
-				      const GLvoid *data )
-{
-    CompressedTexSubImage1D2D( target, level, xoffset, 0, width, 0,
-			       format, image_size, data,
-			       X_GLrop_CompressedTexSubImage1D );
-}
-
-
-void
-__indirect_glCompressedTexSubImage2DARB( GLenum target, GLint level,
-				      GLint xoffset, GLint yoffset,
-				      GLsizei width, GLsizei height,
-				      GLenum format, GLsizei image_size,
-				      const GLvoid *data )
-{
-    CompressedTexSubImage1D2D( target, level, xoffset, yoffset, width, height,
-			       format, image_size, data,
-			       X_GLrop_CompressedTexSubImage2D );
-}
-
-
-void
-__indirect_glCompressedTexSubImage3DARB( GLenum target, GLint level,
-				      GLint xoffset, GLint yoffset, GLint zoffset,
-				      GLsizei width, GLsizei height, GLsizei depth,
-				      GLenum format, GLsizei image_size,
-				      const GLvoid *data )
-{
-    __GLX_DECLARE_VARIABLES();
-    
-    __GLX_LOAD_VARIABLES();
-    if ( gc->currentDpy == NULL ) {
-	return;
-    }
-
-    cmdlen = __GLX_PAD( __GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE
-			+ image_size );
-    if ( cmdlen <= gc->maxSmallRenderCommandSize ) {
-	__GLX_BEGIN_VARIABLE( X_GLrop_CompressedTexSubImage3D, cmdlen );
-	__GLX_PUT_LONG( 4, target );
-	__GLX_PUT_LONG( 8, level );
-	__GLX_PUT_LONG( 12, xoffset );
-	__GLX_PUT_LONG( 16, yoffset );
-	__GLX_PUT_LONG( 20, zoffset );
-	__GLX_PUT_LONG( 24, width );
-	__GLX_PUT_LONG( 28, height );
-	__GLX_PUT_LONG( 32, depth );
-	__GLX_PUT_LONG( 36, format );
-	__GLX_PUT_LONG( 40, image_size );
-	if ( image_size != 0 ) {
-	    __GLX_PUT_CHAR_ARRAY( __GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE,
-				  data, image_size );
-	}
-	__GLX_END( cmdlen );
-    }
-    else {
-	__GLX_BEGIN_VARIABLE_LARGE( X_GLrop_CompressedTexSubImage3D,
-				    cmdlen + 4 );
-	__GLX_PUT_LONG( 8, target );
-	__GLX_PUT_LONG( 12, level );
-	__GLX_PUT_LONG( 16, xoffset );
-	__GLX_PUT_LONG( 20, yoffset );
-	__GLX_PUT_LONG( 24, zoffset );
-	__GLX_PUT_LONG( 28, width );
-	__GLX_PUT_LONG( 32, height );
-	__GLX_PUT_LONG( 36, depth );
-	__GLX_PUT_LONG( 40, format );
-	__GLX_PUT_LONG( 44, image_size );
-	__glXSendLargeCommand( gc, gc->pc,
-			       __GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE + 4,
-			       data, image_size );
-    }
-}
diff --git a/src/glx/x11/glxclient.h b/src/glx/x11/glxclient.h
index f9d076f56f8..467b0ebb7f9 100644
--- a/src/glx/x11/glxclient.h
+++ b/src/glx/x11/glxclient.h
@@ -1,36 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /**
  * \file glxclient.h
@@ -57,20 +53,41 @@
 #include "GL/glxint.h"
 #include "GL/glxproto.h"
 #include "GL/internal/glcore.h"
-#include "glapitable.h"
-#include "glxextensions.h"
+#include "glapi/glapitable.h"
+#include "glxhash.h"
 #if defined( USE_XTHREADS )
 # include <X11/Xthreads.h>
 #elif defined( PTHREADS )
 # include <pthread.h>
 #endif
 
+#include "glxextensions.h"
+
+
+/* If we build the library with gcc's -fvisibility=hidden flag, we'll
+ * use the PUBLIC macro to mark functions that are to be exported.
+ *
+ * We also need to define a USED attribute, so the optimizer doesn't 
+ * inline a static function that we later use in an alias. - ajax
+ */
+#if defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 303
+#  define PUBLIC __attribute__((visibility("default")))
+#  define USED __attribute__((used))
+#else
+#  define PUBLIC
+#  define USED
+#endif
+
+
+
 #define GLX_MAJOR_VERSION	1	/* current version numbers */
 #define GLX_MINOR_VERSION	4
 
 #define __GLX_MAX_TEXTURE_UNITS 32
 
+typedef struct __GLXscreenConfigsRec __GLXscreenConfigs;
 typedef struct __GLXcontextRec __GLXcontext;
+typedef struct __GLXdrawableRec __GLXdrawable;
 typedef struct __GLXdisplayPrivateRec __GLXdisplayPrivate;
 typedef struct _glapi_table __GLapi;
 
@@ -78,6 +95,9 @@ typedef struct _glapi_table __GLapi;
 
 #ifdef GLX_DIRECT_RENDERING
 
+#define containerOf(ptr, type, member)			\
+    (type *)( (char *)ptr - offsetof(type,member) )
+
 #include <GL/internal/dri_interface.h>
 
 
@@ -85,43 +105,69 @@ typedef struct _glapi_table __GLapi;
  * Display dependent methods.  This structure is initialized during the
  * \c driCreateDisplay call.
  */
-struct __DRIdisplayRec {
+typedef struct __GLXDRIdisplayRec __GLXDRIdisplay;
+typedef struct __GLXDRIscreenRec __GLXDRIscreen;
+typedef struct __GLXDRIdrawableRec __GLXDRIdrawable;
+typedef struct __GLXDRIcontextRec __GLXDRIcontext;
+
+#include "glxextensions.h"
+
+struct __GLXDRIdisplayRec {
     /**
      * Method to destroy the private DRI display data.
      */
-    void (*destroyDisplay)(Display *dpy, void *displayPrivate);
+    void (*destroyDisplay)(__GLXDRIdisplay *display);
 
-    /**
-     * Opaque pointer to private per display direct rendering data.
-     * \c NULL if direct rendering is not supported on this display.
-     */
-    struct __DRIdisplayPrivateRec *private;
+    __GLXDRIscreen *(*createScreen)(__GLXscreenConfigs *psc, int screen,
+				    __GLXdisplayPrivate *priv);
+};
 
-    /**
-     * Array of pointers to methods to create and initialize the private DRI
-     * screen data.
-     */
-    PFNCREATENEWSCREENFUNC * createNewScreen;
+struct __GLXDRIscreenRec {
+
+    void (*destroyScreen)(__GLXscreenConfigs *psc);
+
+    __GLXDRIcontext *(*createContext)(__GLXscreenConfigs *psc,
+				      const __GLcontextModes *mode,
+				      GLXContext gc,
+				      GLXContext shareList, int renderType);
+	
+    __GLXDRIdrawable *(*createDrawable)(__GLXscreenConfigs *psc,
+					XID drawable,
+					GLXDrawable glxDrawable,
+					const __GLcontextModes *modes);
+
+    void (*swapBuffers)(__GLXDRIdrawable *pdraw);
+    void (*copySubBuffer)(__GLXDRIdrawable *pdraw,
+			  int x, int y, int width, int height);
+};
+
+struct __GLXDRIcontextRec {
+    void (*destroyContext)(__GLXDRIcontext *context, __GLXscreenConfigs *psc,
+			   Display *dpy);
+    Bool (*bindContext)(__GLXDRIcontext *context,
+			__GLXDRIdrawable *pdraw,
+			__GLXDRIdrawable *pread);
+    
+    void (*unbindContext)(__GLXDRIcontext *context);
 };
 
+struct __GLXDRIdrawableRec {
+    void (*destroyDrawable)(__GLXDRIdrawable *drawable);
 
-/*
-** We keep a linked list of these structures, one per DRI device driver.
-*/
-struct __DRIdriverRec {
-   const char *name;
-   void *handle;
-   PFNCREATENEWSCREENFUNC createNewScreenFunc;
-   struct __DRIdriverRec *next;
+    XID xDrawable;
+    XID drawable;
+    __GLXscreenConfigs *psc;
+    GLenum textureTarget;
+    __DRIdrawable *driDrawable;
 };
 
 /*
 ** Function to create and DRI display data and initialize the display
 ** dependent methods.
 */
-extern void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp);
-
-extern  __DRIdriver *driGetDriver(Display *dpy, int scrNum);
+extern __GLXDRIdisplay *driswCreateDisplay(Display *dpy);
+extern __GLXDRIdisplay *driCreateDisplay(Display *dpy);
+extern __GLXDRIdisplay *dri2CreateDisplay(Display *dpy);
 
 extern void DRI_glXUseXFont( Font font, int first, int count, int listbase );
 
@@ -133,8 +179,6 @@ extern const char *glXGetScreenDriver (Display *dpy, int scrNum);
 
 extern const char *glXGetDriverConfig (const char *driverName);
 
-extern Bool __glXWindowExists(Display *dpy, GLXDrawable draw);
-
 #endif
 
 /************************************************************************/
@@ -226,18 +270,10 @@ struct __GLXcontextRec {
     XID share_xid;
 
     /**
-     * Visual id.
-     * 
-     * \deprecated
-     * This filed has been largely been replaced by the \c mode field, but
-     * the work is not quite done.
-     */
-    VisualID vid;
-
-    /**
      * Screen number.
      */
     GLint screen;
+    __GLXscreenConfigs *psc;
 
     /**
      * \c GL_TRUE if the context was created with ImportContext, which
@@ -343,24 +379,15 @@ struct __GLXcontextRec {
      */
     GLint majorOpcode;
 
-#ifdef GLX_DIRECT_RENDERING
     /**
-     * Per context direct rendering interface functions and data.
+     * Pointer to the mode used to create this context.
      */
-    __DRIcontext driContext;
+    const __GLcontextModes * mode;
+
+#ifdef GLX_DIRECT_RENDERING
+    __GLXDRIcontext *driContext;
+    __DRIcontext *__driContext;
 #endif
-    
-    /**
-     * \c GLXFBConfigID used to create this context.  May be \c None.  This
-     * field has been replaced by the \c mode field.
-     *
-     * \since Internal API version 20030317.
-     *
-     * \deprecated
-     * This filed has been largely been replaced by the \c mode field, but
-     * the work is not quite done.
-     */
-    GLXFBConfigID  fbconfigID;
 
     /**
      * The current read-drawable for this context.  Will be None if this
@@ -438,7 +465,7 @@ extern void __glFreeAttributeState(__GLXcontext *);
  * One of these records exists per screen of the display.  It contains
  * a pointer to the config data for that screen (if the screen supports GL).
  */
-typedef struct __GLXscreenConfigsRec {
+struct __GLXscreenConfigsRec {
     /**
      * GLX extension string reported by the X-server.
      */
@@ -454,13 +481,48 @@ typedef struct __GLXscreenConfigsRec {
     /**
      * Per screen direct rendering interface functions and data.
      */
-    __DRIscreen driScreen;
+    __DRIscreen *__driScreen;
+    const __DRIcoreExtension *core;
+    const __DRIlegacyExtension *legacy;
+    const __DRIswrastExtension *swrast;
+    const __DRIdri2Extension *dri2;
+    __glxHashTable *drawHash;
+    Display *dpy;
+    int scr, fd;
+    void *driver;
+
+    __GLXDRIscreen *driScreen;
+
+#ifdef __DRI_COPY_SUB_BUFFER
+    const __DRIcopySubBufferExtension *driCopySubBuffer;
+#endif
+
+#ifdef __DRI_SWAP_CONTROL
+    const __DRIswapControlExtension *swapControl;
+#endif
+
+#ifdef __DRI_ALLOCATE
+    const __DRIallocateExtension *allocate;
+#endif
+
+#ifdef __DRI_FRAME_TRACKING
+    const __DRIframeTrackingExtension *frameTracking;
+#endif
+
+#ifdef __DRI_MEDIA_STREAM_COUNTER
+    const __DRImediaStreamCounterExtension *msc;
+#endif
+
+#ifdef __DRI_TEX_BUFFER
+    const __DRItexBufferExtension *texBuffer;
+#endif
+
 #endif
 
     /**
-     * Linked list of configurations for this screen.
+     * Linked list of glx visuals and  fbconfigs for this screen.
      */
-    __GLcontextModes *configs;
+    __GLcontextModes *visuals, *configs;
 
     /**
      * Per-screen dynamic GLX extension tracking.  The \c direct_support
@@ -474,7 +536,7 @@ typedef struct __GLXscreenConfigsRec {
     GLboolean ext_list_first_time;
     /*@}*/
 
-} __GLXscreenConfigs;
+};
 
 /**
  * Per display private data.  One of these records exists for each display
@@ -523,11 +585,12 @@ struct __GLXdisplayPrivateRec {
     /**
      * Per display direct rendering interface functions and data.
      */
-    __DRIdisplay driDisplay;
+    __GLXDRIdisplay *driswDisplay;
+    __GLXDRIdisplay *driDisplay;
+    __GLXDRIdisplay *dri2Display;
 #endif
 };
 
-void __glXFreeContext(__GLXcontext*);
 
 extern GLubyte *__glXFlushRenderBuffer(__GLXcontext*, GLubyte*);
 
@@ -573,6 +636,10 @@ extern __GLXcontext *__glXcurrentContext;
 
 #endif /* defined( USE_XTHREADS ) || defined( PTHREADS ) */
 
+extern void __glXSetCurrentContextNull(void);
+
+extern void __glXFreeContext(__GLXcontext*);
+
 
 /*
 ** Global lock for all threads in this address space using the GLX
@@ -647,9 +714,10 @@ extern void __glEmptyImage(__GLXcontext*, GLint, GLint, GLint, GLint, GLenum,
 
 
 /*
-** Allocate and Initialize Vertex Array client state 
+** Allocate and Initialize Vertex Array client state, and free.
 */
-extern void __glXInitVertexArrayState(__GLXcontext*);
+extern void __glXInitVertexArrayState(__GLXcontext *);
+extern void __glXFreeVertexArrayState(__GLXcontext *);
 
 /*
 ** Inform the Server of the major and minor numbers and of the client
@@ -673,8 +741,10 @@ extern void _XSend(Display*, const void*, long);
 extern void __glXInitializeVisualConfigFromTags( __GLcontextModes *config,
     int count, const INT32 *bp, Bool tagged_only, Bool fbconfig_style_tags );
 
-extern char * __glXGetStringFromServer( Display * dpy, int opcode,
-    CARD32 glxCode, CARD32 for_whom, CARD32 name );
+extern char * __glXQueryServerString(Display* dpy, int opcode,
+                                     CARD32 screen, CARD32 name);
+extern char * __glXGetString(Display* dpy, int opcode,
+                             CARD32 screen, CARD32 name);
 
 extern char *__glXstrdup(const char *str);
 
@@ -682,13 +752,16 @@ extern char *__glXstrdup(const char *str);
 extern const char __glXGLClientVersion[];
 extern const char __glXGLClientExtensions[];
 
-/* Determine the internal API version */
-extern int __glXGetInternalVersion(void);
-
 /* Get the unadjusted system time */
 extern int __glXGetUST( int64_t * ust );
 
-extern Bool __glXGetMscRateOML(Display * dpy, GLXDrawable drawable,
-    int32_t * numerator, int32_t * denominator);
+extern GLboolean __glXGetMscRateOML(Display * dpy, GLXDrawable drawable,
+				    int32_t * numerator, int32_t * denominator);
+
+#ifdef GLX_DIRECT_RENDERING
+GLboolean
+__driGetMscRateOML(__DRIdrawable *draw,
+		   int32_t *numerator, int32_t *denominator, void *private);
+#endif
 
 #endif /* !__GLX_client_h__ */
diff --git a/src/glx/x11/glxcmds.c b/src/glx/x11/glxcmds.c
index 80281896f69..1d373ed29d8 100644
--- a/src/glx/x11/glxcmds.c
+++ b/src/glx/x11/glxcmds.c
@@ -1,100 +1,147 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /**
  * \file glxcmds.c
  * Client-side GLX interface.
  */
 
-#include <inttypes.h>
 #include "glxclient.h"
-#include <X11/extensions/extutil.h>
-#include <X11/extensions/Xext.h>
-#include <assert.h>
-#include <string.h>
 #include "glapi.h"
+#include "glxextensions.h"
+#include "glcontextmodes.h"
+
 #ifdef GLX_DIRECT_RENDERING
-#include "indirect_init.h"
+#include <sys/time.h>
 #include <X11/extensions/xf86vmode.h>
 #include "xf86dri.h"
 #endif
-#include "glxextensions.h"
-#include "glcontextmodes.h"
-#include "glheader.h"
-#include <sys/time.h>
+
+#if defined(USE_XCB)
+#include <X11/Xlib-xcb.h>
+#include <xcb/xcb.h>
+#include <xcb/glx.h>
+#endif
 
 static const char __glXGLXClientVendorName[] = "SGI";
 static const char __glXGLXClientVersion[] = "1.4";
 
 
 /****************************************************************************/
+
+#ifdef GLX_DIRECT_RENDERING
+
+static Bool windowExistsFlag;
+static int windowExistsErrorHandler(Display *dpy, XErrorEvent *xerr)
+{
+    if (xerr->error_code == BadWindow) {
+	windowExistsFlag = GL_FALSE;
+    }
+    return 0;
+}
+
+/**
+ * Find drawables in the local hash that have been destroyed on the
+ * server.
+ * 
+ * \param dpy    Display to destroy drawables for
+ * \param screen Screen number to destroy drawables for
+ */
+static void GarbageCollectDRIDrawables(Display *dpy, __GLXscreenConfigs *sc)
+{
+    XID draw;
+    __GLXDRIdrawable *pdraw;
+    XWindowAttributes xwa;
+    int (*oldXErrorHandler)(Display *, XErrorEvent *);
+
+    /* Set no-op error handler so Xlib doesn't bail out if the windows
+     * has alreay been destroyed on the server. */
+    XSync(dpy, GL_FALSE);
+    oldXErrorHandler = XSetErrorHandler(windowExistsErrorHandler);
+
+    if (__glxHashFirst(sc->drawHash, &draw, (void *)&pdraw) == 1) {
+	do {
+	    windowExistsFlag = GL_TRUE;
+	    XGetWindowAttributes(dpy, draw, &xwa); /* dummy request */
+	    if (!windowExistsFlag) {
+		/* Destroy the local drawable data, if the drawable no
+		   longer exists in the Xserver */
+		(*pdraw->destroyDrawable)(pdraw);
+                __glxHashDelete(sc->drawHash, draw);
+	    }
+	} while (__glxHashNext(sc->drawHash, &draw, (void *)&pdraw) == 1);
+    }
+
+    XSync(dpy, GL_FALSE);
+    XSetErrorHandler(oldXErrorHandler);
+}
+
+extern __GLXDRIdrawable *
+GetGLXDRIDrawable(Display *dpy, GLXDrawable drawable, int * const scrn_num);
+
 /**
  * Get the __DRIdrawable for the drawable associated with a GLXContext
  * 
  * \param dpy       The display associated with \c drawable.
  * \param drawable  GLXDrawable whose __DRIdrawable part is to be retrieved.
+ * \param scrn_num  If non-NULL, the drawables screen is stored there
  * \returns  A pointer to the context's __DRIdrawable on success, or NULL if
  *           the drawable is not associated with a direct-rendering context.
  */
-
-#ifdef GLX_DIRECT_RENDERING
-static __DRIdrawable *
-GetDRIDrawable( Display *dpy, GLXDrawable drawable, int * const scrn_num )
+_X_HIDDEN __GLXDRIdrawable *
+GetGLXDRIDrawable(Display *dpy, GLXDrawable drawable, int * const scrn_num)
 {
-    __GLXdisplayPrivate * const priv = __glXInitialize(dpy);
-
-    if ( (priv != NULL) && (priv->driDisplay.private != NULL) ) {
-	const unsigned  screen_count = ScreenCount(dpy);
-	unsigned   i;
-
-	for ( i = 0 ; i < screen_count ; i++ ) {
-	    __DRIscreen * const psc = &priv->screenConfigs[i].driScreen;
-	    __DRIdrawable * const pdraw = (psc->private != NULL)
-	       ? (*psc->getDrawable)(dpy, drawable, psc->private) : NULL;
+    __GLXdisplayPrivate *priv = __glXInitialize(dpy);
+    __GLXDRIdrawable *pdraw;
+    const unsigned  screen_count = ScreenCount(dpy);
+    unsigned   i;
+    __GLXscreenConfigs *psc;
 
-	    if ( pdraw != NULL ) {
-		if ( scrn_num != NULL ) {
-		    *scrn_num = i;
-		}
-		return pdraw;
-	    }
+    if (priv == NULL)
+	return NULL;
+    
+    for (i = 0; i < screen_count; i++) {
+	psc = &priv->screenConfigs[i];
+	if (psc->drawHash == NULL)
+	    continue;
+
+	if (__glxHashLookup(psc->drawHash, drawable, (void *) &pdraw) == 0) {
+	    if (scrn_num != NULL)
+		*scrn_num = i;
+	    return pdraw;
 	}
     }
 
     return NULL;
 }
+
 #endif
 
 
@@ -264,9 +311,9 @@ GLXContext AllocateGLXContext( Display *dpy )
     */
     gc->fastImageUnpack = GL_FALSE;
     gc->fillImage = __glFillImage;
-    gc->isDirect = GL_FALSE;
     gc->pc = gc->buf;
     gc->bufEnd = gc->buf + bufSize;
+    gc->isDirect = GL_FALSE;
     if (__glXDebug) {
 	/*
 	** Set limit register so that there will be one command per packet
@@ -312,6 +359,10 @@ CreateContext(Display *dpy, XVisualInfo *vis,
 	      Bool use_glx_1_3, int renderType)
 {
     GLXContext gc;
+#ifdef GLX_DIRECT_RENDERING
+    int screen = (fbconfig == NULL) ? vis->screen : fbconfig->screen;
+    __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
+#endif
 
     if ( dpy == NULL )
        return NULL;
@@ -325,41 +376,36 @@ CreateContext(Display *dpy, XVisualInfo *vis,
 	    return NULL;
 
 #ifdef GLX_DIRECT_RENDERING
-	if (allowDirect) {
-	    int screen = (fbconfig == NULL) ? vis->screen : fbconfig->screen;
-	    __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
+	if (allowDirect && psc->driScreen) {
 	    const __GLcontextModes * mode;
 
-	    /* The value of fbconfig cannot change because it is tested
-	     * later in the function.
-	     */
-	    if ( fbconfig == NULL ) {
-		/* FIXME: Is it possible for the __GLcontextModes structure
-		 * FIXME: to not be found?
-		 */
-		mode = _gl_context_modes_find_visual( psc->configs,
-						      vis->visualid );
-		assert( mode != NULL );
-		assert( mode->screen == screen );
+	    if (fbconfig == NULL) {
+		mode = _gl_context_modes_find_visual(psc->visuals, vis->visualid);
+		if (mode == NULL) {
+		   xError error;
+
+		   error.errorCode = BadValue;
+		   error.resourceID = vis->visualid;
+		   error.sequenceNumber = dpy->request;
+		   error.type = X_Error;
+		   error.majorCode = gc->majorOpcode;
+		   error.minorCode = X_GLXCreateContext;
+		   _XError(dpy, &error);
+		   return None;
+		}
 	    }
 	    else {
 		mode = fbconfig;
 	    }
 
-	    if (psc && psc->driScreen.private) {
-		void * const shared = (shareList != NULL)
-		    ? shareList->driContext.private : NULL;
-		gc->driContext.private = 
-		  (*psc->driScreen.createNewContext)( dpy, mode, renderType,
-						      shared,
-						      &gc->driContext );
-		if (gc->driContext.private) {
-		    gc->isDirect = GL_TRUE;
-		    gc->screen = mode->screen;
-		    gc->vid = mode->visualID;
-		    gc->fbconfigID = mode->fbconfigID;
-		    gc->driContext.mode = mode;
-		}
+	    gc->driContext = psc->driScreen->createContext(psc, mode, gc,
+							   shareList,
+							   renderType);
+	    if (gc->driContext != NULL) {
+		gc->screen = mode->screen;
+		gc->psc = psc;
+		gc->mode = mode;
+		gc->isDirect = GL_TRUE;
 	    }
 	}
 #endif
@@ -376,7 +422,11 @@ CreateContext(Display *dpy, XVisualInfo *vis,
 	    req->visual = vis->visualid;
 	    req->screen = vis->screen;
 	    req->shareList = shareList ? shareList->xid : None;
-	    req->isDirect = gc->isDirect;
+#ifdef GLX_DIRECT_RENDERING
+	    req->isDirect = gc->driContext != NULL;
+#else
+	    req->isDirect = 0;
+#endif
 	}
 	else if ( use_glx_1_3 ) {
 	    xGLXCreateNewContextReq *req;
@@ -390,7 +440,11 @@ CreateContext(Display *dpy, XVisualInfo *vis,
 	    req->screen = fbconfig->screen;
 	    req->renderType = renderType;
 	    req->shareList = shareList ? shareList->xid : None;
-	    req->isDirect = gc->isDirect;
+#ifdef GLX_DIRECT_RENDERING
+	    req->isDirect = gc->driContext != NULL;
+#else
+	    req->isDirect = 0;
+#endif
 	}
 	else {
 	    xGLXVendorPrivateWithReplyReq *vpreq;
@@ -408,7 +462,11 @@ CreateContext(Display *dpy, XVisualInfo *vis,
 	    req->screen = fbconfig->screen;
 	    req->renderType = renderType;
 	    req->shareList = shareList ? shareList->xid : None;
-	    req->isDirect = gc->isDirect;
+#ifdef GLX_DIRECT_RENDERING
+	    req->isDirect = gc->driContext != NULL;
+#else
+	    req->isDirect = 0;
+#endif
 	}
 
 	UnlockDisplay(dpy);
@@ -430,7 +488,7 @@ PUBLIC GLXContext glXCreateContext(Display *dpy, XVisualInfo *vis,
 			False, 0);
 }
 
-void __glXFreeContext(__GLXcontext *gc)
+_X_HIDDEN void __glXFreeContext(__GLXcontext *gc)
 {
     if (gc->vendor) XFree((char *) gc->vendor);
     if (gc->renderer) XFree((char *) gc->renderer);
@@ -466,15 +524,15 @@ DestroyContext(Display *dpy, GLXContext gc)
 
 #ifdef GLX_DIRECT_RENDERING
     /* Destroy the direct rendering context */
-    if (gc->isDirect) {
-	if (gc->driContext.private) {
-	    (*gc->driContext.destroyContext)(dpy, gc->screen,
-					     gc->driContext.private);
-	    gc->driContext.private = NULL;
-	}
+    if (gc->driContext) {
+	(*gc->driContext->destroyContext)(gc->driContext, gc->psc, dpy);
+	gc->driContext = NULL;
+	GarbageCollectDRIDrawables(dpy, gc->psc);
     }
 #endif
 
+    __glXFreeVertexArrayState(gc);
+
     if (gc->currentDpy) {
 	/* Have to free later cuz it's in use now */
 	__glXUnlock();
@@ -552,7 +610,7 @@ PUBLIC void glXWaitGL(void)
     __glXFlushRenderBuffer(gc, gc->pc);
 
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect) {
+    if (gc->driContext) {
 /* This bit of ugliness unwraps the glFinish function */
 #ifdef glFinish
 #undef glFinish
@@ -588,7 +646,7 @@ PUBLIC void glXWaitX(void)
     __glXFlushRenderBuffer(gc, gc->pc);
 
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect) {
+    if (gc->driContext) {
 	XSync(dpy, False);
 	return;
     }
@@ -618,7 +676,7 @@ PUBLIC void glXUseXFont(Font font, int first, int count, int listBase)
     (void) __glXFlushRenderBuffer(gc, gc->pc);
 
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect) {
+    if (gc->driContext) {
       DRI_glXUseXFont(font, first, count, listBase);
       return;
     }
@@ -658,7 +716,7 @@ PUBLIC void glXCopyContext(Display *dpy, GLXContext source,
     }
 
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect) {
+    if (gc->driContext) {
 	/* NOT_DONE: This does not work yet */
     }
 #endif
@@ -697,8 +755,10 @@ PUBLIC void glXCopyContext(Display *dpy, GLXContext source,
  */
 static Bool __glXIsDirect(Display *dpy, GLXContextID contextID)
 {
+#if !defined(USE_XCB)
     xGLXIsDirectReq *req;
     xGLXIsDirectReply reply;
+#endif
     CARD8 opcode;
 
     opcode = __glXSetupForCommand(dpy);
@@ -706,6 +766,18 @@ static Bool __glXIsDirect(Display *dpy, GLXContextID contextID)
 	return GL_FALSE;
     }
 
+#ifdef USE_XCB
+   xcb_connection_t* c = XGetXCBConnection(dpy);
+   xcb_glx_is_direct_reply_t* reply =
+      xcb_glx_is_direct_reply(c,
+                              xcb_glx_is_direct(c, contextID),
+                              NULL);
+
+   const Bool is_direct = reply->is_direct ? True : False;
+   free(reply);
+
+   return is_direct;
+#else
     /* Send the glXIsDirect request */
     LockDisplay(dpy);
     GetReq(GLXIsDirect,req);
@@ -717,6 +789,7 @@ static Bool __glXIsDirect(Display *dpy, GLXContextID contextID)
     SyncHandle();
 
     return reply.isDirect;
+#endif /* USE_XCB */
 }
 
 /**
@@ -730,7 +803,7 @@ PUBLIC Bool glXIsDirect(Display *dpy, GLXContext gc)
     if (!gc) {
 	return GL_FALSE;
 #ifdef GLX_DIRECT_RENDERING
-    } else if (gc->isDirect) {
+    } else if (gc->driContext) {
 	return GL_TRUE;
 #endif
     }
@@ -788,15 +861,15 @@ PUBLIC void glXDestroyGLXPixmap(Display *dpy, GLXPixmap glxpixmap)
 
 PUBLIC void glXSwapBuffers(Display *dpy, GLXDrawable drawable)
 {
-    xGLXSwapBuffersReq *req;
     GLXContext gc;
     GLXContextTag tag;
     CARD8 opcode;
 #ifdef GLX_DIRECT_RENDERING
-    __DRIdrawable *pdraw = GetDRIDrawable( dpy, drawable, NULL );
+    __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, NULL);
 
-    if ( pdraw != NULL ) {
-	(*pdraw->swapBuffers)(dpy, pdraw->private);
+    if (pdraw != NULL) {
+	glFlush();	    
+	(*pdraw->psc->driScreen->swapBuffers)(pdraw);
 	return;
     }
 #endif
@@ -818,6 +891,13 @@ PUBLIC void glXSwapBuffers(Display *dpy, GLXDrawable drawable)
 	tag = 0;
     }
 
+#ifdef USE_XCB
+    xcb_connection_t* c = XGetXCBConnection(dpy);
+    xcb_glx_swap_buffers(c, tag, drawable);
+    xcb_flush(c);
+#else
+    xGLXSwapBuffersReq *req;
+
     /* Send the glXSwapBuffers request */
     LockDisplay(dpy);
     GetReq(GLXSwapBuffers,req);
@@ -828,6 +908,7 @@ PUBLIC void glXSwapBuffers(Display *dpy, GLXDrawable drawable)
     UnlockDisplay(dpy);
     SyncHandle();
     XFlush(dpy);
+#endif /* USE_XCB */
 }
 
 
@@ -840,12 +921,12 @@ PUBLIC int glXGetConfig(Display *dpy, XVisualInfo *vis, int attribute,
 {
     __GLXdisplayPrivate *priv;
     __GLXscreenConfigs *psc;
+    __GLcontextModes *modes;
     int   status;
 
     status = GetGLXPrivScreenConfig( dpy, vis->screen, & priv, & psc );
     if ( status == Success ) {
-	const __GLcontextModes * const modes = _gl_context_modes_find_visual(
-					     psc->configs, vis->visualid );
+	modes = _gl_context_modes_find_visual(psc->visuals, vis->visualid);
 
 	/* Lookup attribute after first finding a match on the visual */
 	if ( modes != NULL ) {
@@ -1223,7 +1304,7 @@ PUBLIC XVisualInfo *glXChooseVisual(Display *dpy, int screen, int *attribList)
     ** Compute a score for those that do
     ** Remember which visual, if any, got the highest score
     */
-    for ( modes = psc->configs ; modes != NULL ; modes = modes->next ) {
+    for ( modes = psc->visuals ; modes != NULL ; modes = modes->next ) {
 	if ( fbconfigs_compatible( & test_config, modes )
 	     && ((best_config == NULL)
 		 || (fbconfig_compare( (const __GLcontextModes * const * const)&modes, &best_config ) < 0)) ) {
@@ -1261,14 +1342,13 @@ PUBLIC const char *glXQueryExtensionsString( Display *dpy, int screen )
 
     if (!psc->effectiveGLXexts) {
         if (!psc->serverGLXexts) {
-	    psc->serverGLXexts = __glXGetStringFromServer(dpy, priv->majorOpcode,
-					  	   X_GLXQueryServerString,
-					  	   screen, GLX_EXTENSIONS);
+           psc->serverGLXexts =
+              __glXQueryServerString(dpy, priv->majorOpcode, screen, GLX_EXTENSIONS);
 	}
 
 	__glXCalculateUsableExtensions(psc,
 #ifdef GLX_DIRECT_RENDERING
-				       (psc->driScreen.private != NULL),
+				       (psc->driScreen != NULL),
 #else
 				       GL_FALSE,
 #endif
@@ -1318,8 +1398,7 @@ PUBLIC const char *glXQueryServerString( Display *dpy, int screen, int name )
     }
 
     if ( *str == NULL ) {
-	*str = __glXGetStringFromServer(dpy, priv->majorOpcode,
-					X_GLXQueryServerString, screen, name);
+       *str = __glXQueryServerString(dpy, priv->majorOpcode, screen, name);
     }
     
     return *str;
@@ -1327,9 +1406,18 @@ PUBLIC const char *glXQueryServerString( Display *dpy, int screen, int name )
 
 void __glXClientInfo (  Display *dpy, int opcode  )
 {
-    xGLXClientInfoReq *req;
-    int size;
     char * ext_str = __glXGetClientGLExtensionString();
+    int size = strlen( ext_str ) + 1;
+
+#ifdef USE_XCB
+   xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_glx_client_info(c,
+                       GLX_MAJOR_VERSION,
+                       GLX_MINOR_VERSION,
+                       size,
+                       (const uint8_t *)ext_str);
+#else
+    xGLXClientInfoReq *req;
 
     /* Send the glXClientInfo request */
     LockDisplay(dpy);
@@ -1339,14 +1427,14 @@ void __glXClientInfo (  Display *dpy, int opcode  )
     req->major = GLX_MAJOR_VERSION;
     req->minor = GLX_MINOR_VERSION;
 
-    size = strlen( ext_str ) + 1;
     req->length += (size + 3) >> 2;
     req->numbytes = size;
     Data(dpy, ext_str, size);
 
     UnlockDisplay(dpy);
     SyncHandle();
-    
+#endif /* USE_XCB */
+
     Xfree( ext_str );
 }
 
@@ -1447,13 +1535,15 @@ static int __glXQueryContextInfo(Display *dpy, GLXContext ctx)
 		    ctx->share_xid = *pProp++;
 		    break;
 		case GLX_VISUAL_ID_EXT:
-		    ctx->vid = *pProp++;
+		    ctx->mode =
+			_gl_context_modes_find_visual(ctx->psc->visuals, *pProp++);
 		    break;
 		case GLX_SCREEN:
 		    ctx->screen = *pProp++;
 		    break;
 		case GLX_FBCONFIG_ID:
-		    ctx->fbconfigID = *pProp++;
+		    ctx->mode =
+			_gl_context_modes_find_fbconfig(ctx->psc->configs, *pProp++);
 		    break;
 		case GLX_RENDER_TYPE:
 		    ctx->renderType = *pProp++;
@@ -1478,7 +1568,11 @@ glXQueryContext(Display *dpy, GLXContext ctx, int attribute, int *value)
     int retVal;
 
     /* get the information from the server if we don't have it already */
-    if (!ctx->isDirect && (ctx->vid == None)) {
+#ifdef GLX_DIRECT_RENDERING
+    if (!ctx->driContext && (ctx->mode == NULL)) {
+#else
+    if (ctx->mode == NULL) {
+#endif
 	retVal = __glXQueryContextInfo(dpy, ctx);
 	if (Success != retVal) return retVal;
     }
@@ -1487,13 +1581,13 @@ glXQueryContext(Display *dpy, GLXContext ctx, int attribute, int *value)
 	*value = (int)(ctx->share_xid);
 	break;
     case GLX_VISUAL_ID_EXT:
-	*value = (int)(ctx->vid);
+	*value = ctx->mode ? ctx->mode->visualID : None;
 	break;
     case GLX_SCREEN:
 	*value = (int)(ctx->screen);
 	break;
     case GLX_FBCONFIG_ID:
-	*value = (int)(ctx->fbconfigID);
+	*value = ctx->mode ? ctx->mode->fbconfigID : None;
 	break;
     case GLX_RENDER_TYPE:
 	*value = (int)(ctx->renderType);
@@ -1591,6 +1685,7 @@ PUBLIC GLXFBConfig *glXGetFBConfigs(Display *dpy, int screen, int *nelements)
     __GLcontextModes ** config = NULL;
     int   i;
 
+    *nelements = 0;
     if ( (priv->screenConfigs != NULL)
 	 && (screen >= 0) && (screen <= ScreenCount(dpy))
 	 && (priv->screenConfigs[screen].configs != NULL)
@@ -1615,8 +1710,10 @@ PUBLIC GLXFBConfig *glXGetFBConfigs(Display *dpy, int screen, int *nelements)
 	    for ( modes = priv->screenConfigs[screen].configs
 		  ; modes != NULL
 		  ; modes = modes->next ) {
-		config[i] = modes;
-		i++;
+		if ( modes->fbconfigID != GLX_DONT_CARE ) {
+		    config[i] = modes;
+		    i++;
+		}
 	    }
 	}
     }
@@ -1668,16 +1765,15 @@ static int __glXSwapIntervalSGI(int interval)
       return GLX_BAD_VALUE;
    }
 
-#ifdef GLX_DIRECT_RENDERING
-   if ( gc->isDirect ) {
+#ifdef __DRI_SWAP_CONTROL
+   if (gc->driContext) {
        __GLXscreenConfigs * const psc = GetGLXScreenConfigs( gc->currentDpy,
 							     gc->screen );
-       __DRIdrawable * const pdraw = GetDRIDrawable( gc->currentDpy,
-						     gc->currentDrawable,
-						     NULL );
-       if ( __glXExtensionBitIsEnabled( psc, SGI_swap_control_bit )
-	    && (pdraw != NULL) ) {
-	   pdraw->swap_interval = interval;
+       __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(gc->currentDpy,
+						   gc->currentDrawable,
+						   NULL);
+       if (psc->swapControl != NULL && pdraw != NULL) {
+	   psc->swapControl->setSwapInterval(pdraw->driDrawable, interval);
 	   return 0;
        }
        else {
@@ -1715,25 +1811,22 @@ static int __glXSwapIntervalSGI(int interval)
 */
 static int __glXSwapIntervalMESA(unsigned int interval)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_SWAP_CONTROL
    GLXContext gc = __glXGetCurrentContext();
 
    if ( interval < 0 ) {
       return GLX_BAD_VALUE;
    }
 
-   if ( (gc != NULL) && gc->isDirect ) {
+   if (gc != NULL && gc->driContext) {
       __GLXscreenConfigs * const psc = GetGLXScreenConfigs( gc->currentDpy,
 							    gc->screen );
       
-      if ( (psc != NULL) && (psc->driScreen.private != NULL)
-	   && __glXExtensionBitIsEnabled( psc, MESA_swap_control_bit ) ) {
-	 __DRIdrawable * const pdraw = 
-	     (*psc->driScreen.getDrawable)(gc->currentDpy,
-					   gc->currentDrawable,
-					   psc->driScreen.private);
-	 if ( pdraw != NULL ) {
-	    pdraw->swap_interval = interval;
+      if ( (psc != NULL) && (psc->driScreen != NULL) ) {
+	 __GLXDRIdrawable *pdraw = 
+	     GetGLXDRIDrawable(gc->currentDpy, gc->currentDrawable, NULL);
+	 if (psc->swapControl != NULL && pdraw != NULL) {
+	    psc->swapControl->setSwapInterval(pdraw->driDrawable, interval);
 	    return 0;
 	 }
       }
@@ -1748,21 +1841,18 @@ static int __glXSwapIntervalMESA(unsigned int interval)
 
 static int __glXGetSwapIntervalMESA(void)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_SWAP_CONTROL
    GLXContext gc = __glXGetCurrentContext();
 
-   if ( (gc != NULL) && gc->isDirect ) {
+   if (gc != NULL && gc->driContext) {
       __GLXscreenConfigs * const psc = GetGLXScreenConfigs( gc->currentDpy,
 							    gc->screen );
       
-      if ( (psc != NULL) && (psc->driScreen.private != NULL)
-	   && __glXExtensionBitIsEnabled( psc, MESA_swap_control_bit ) ) {
-	 __DRIdrawable * const pdraw = 
-	     (*psc->driScreen.getDrawable)(gc->currentDpy,
-					   gc->currentDrawable,
-					   psc->driScreen.private);
-	 if ( pdraw != NULL ) {
-	    return pdraw->swap_interval;
+      if ( (psc != NULL) && (psc->driScreen != NULL) ) {
+	 __GLXDRIdrawable *pdraw = 
+	     GetGLXDRIDrawable(gc->currentDpy, gc->currentDrawable, NULL);
+	 if (psc->swapControl != NULL && pdraw != NULL) {
+	    return psc->swapControl->getSwapInterval(pdraw->driDrawable);
 	 }
       }
    }
@@ -1779,15 +1869,13 @@ static int __glXGetSwapIntervalMESA(void)
 static GLint __glXBeginFrameTrackingMESA(Display *dpy, GLXDrawable drawable)
 {
    int   status = GLX_BAD_CONTEXT;
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_FRAME_TRACKING
    int screen;
-   __DRIdrawable * const pdraw = GetDRIDrawable(dpy, drawable, & screen);
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
 
-   if ( (pdraw != NULL) && (pdraw->frameTracking != NULL)
-	&& __glXExtensionBitIsEnabled( psc, MESA_swap_frame_usage_bit ) ) {
-      status = pdraw->frameTracking( dpy, pdraw->private, GL_TRUE );
-   }
+   if (pdraw != NULL && psc->frameTracking != NULL)
+       status = psc->frameTracking->frameTracking(pdraw->driDrawable, GL_TRUE);
 #else
    (void) dpy;
    (void) drawable;
@@ -1799,15 +1887,14 @@ static GLint __glXBeginFrameTrackingMESA(Display *dpy, GLXDrawable drawable)
 static GLint __glXEndFrameTrackingMESA(Display *dpy, GLXDrawable drawable)
 {
    int   status = GLX_BAD_CONTEXT;
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_FRAME_TRACKING
    int screen;
-   __DRIdrawable * const pdraw = GetDRIDrawable(dpy, drawable, & screen);
-   __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, & screen);
+   __GLXscreenConfigs *psc = GetGLXScreenConfigs(dpy, screen);
 
-   if ( (pdraw != NULL) && (pdraw->frameTracking != NULL)
-	&& __glXExtensionBitIsEnabled( psc, MESA_swap_frame_usage_bit ) ) {
-      status = pdraw->frameTracking( dpy, pdraw->private, GL_FALSE );
-   }
+   if (pdraw != NULL && psc->frameTracking != NULL)
+       status = psc->frameTracking->frameTracking(pdraw->driDrawable,
+						  GL_FALSE);
 #else
    (void) dpy;
    (void) drawable;
@@ -1820,19 +1907,20 @@ static GLint __glXGetFrameUsageMESA(Display *dpy, GLXDrawable drawable,
 				    GLfloat *usage)
 {
    int   status = GLX_BAD_CONTEXT;
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_FRAME_TRACKING
    int screen;
-   __DRIdrawable * const pdraw = GetDRIDrawable(dpy, drawable, & screen);
+   __GLXDRIdrawable * const pdraw = GetGLXDRIDrawable(dpy, drawable, & screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
 
-   if ( (pdraw != NULL ) && (pdraw->queryFrameTracking != NULL)
-	&& __glXExtensionBitIsEnabled( psc, MESA_swap_frame_usage_bit ) ) {
-      int64_t sbc, missedFrames;
-      float   lastMissedUsage;
+   if (pdraw != NULL && psc->frameTracking != NULL) {
+       int64_t sbc, missedFrames;
+       float   lastMissedUsage;
 
-      status = pdraw->queryFrameTracking( dpy, pdraw->private, &sbc,
-					  &missedFrames, &lastMissedUsage,
-					  usage );
+       status = psc->frameTracking->queryFrameTracking(pdraw->driDrawable,
+						       &sbc,
+						       &missedFrames,
+						       &lastMissedUsage,
+						       usage);
    }
 #else
    (void) dpy;
@@ -1848,18 +1936,17 @@ static GLint __glXQueryFrameTrackingMESA(Display *dpy, GLXDrawable drawable,
 					 GLfloat *lastMissedUsage)
 {
    int   status = GLX_BAD_CONTEXT;
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_FRAME_TRACKING
    int screen;
-   __DRIdrawable * const pdraw = GetDRIDrawable(dpy, drawable, & screen);
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, & screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
 
-   if ( (pdraw != NULL ) && (pdraw->queryFrameTracking != NULL)
-	&& __glXExtensionBitIsEnabled( psc, MESA_swap_frame_usage_bit ) ) {
+   if (pdraw != NULL && psc->frameTracking != NULL) {
       float   usage;
 
-      status = pdraw->queryFrameTracking( dpy, pdraw->private, sbc,
-					  missedFrames, lastMissedUsage,
-					  & usage );
+      status = psc->frameTracking->queryFrameTracking(pdraw->driDrawable,
+						      sbc, missedFrames,
+						      lastMissedUsage, &usage);
    }
 #else
    (void) dpy;
@@ -1881,21 +1968,24 @@ static int __glXGetVideoSyncSGI(unsigned int *count)
     * FIXME: there should be a GLX encoding for this call.  I can find no
     * FIXME: documentation for the GLX encoding.
     */
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_MEDIA_STREAM_COUNTER
    GLXContext gc = __glXGetCurrentContext();
 
 
-   if ( (gc != NULL) && gc->isDirect ) {
+   if (gc != NULL && gc->driContext) {
       __GLXscreenConfigs * const psc = GetGLXScreenConfigs( gc->currentDpy,
 							    gc->screen );
-      if ( __glXExtensionBitIsEnabled( psc, SGI_video_sync_bit )
-	   && psc->driScreen.private && psc->driScreen.getMSC) {
-	 int       ret;
-	 int64_t   temp;
+      if ( psc->msc && psc->driScreen ) {
+          __GLXDRIdrawable *pdraw = 
+              GetGLXDRIDrawable(gc->currentDpy, gc->currentDrawable, NULL);
+	  int64_t temp; 
+	  int ret;
+ 
+	  ret = (*psc->msc->getDrawableMSC)(psc->__driScreen,
+					    pdraw->driDrawable, &temp);
+	  *count = (unsigned) temp;
 
-	 ret = psc->driScreen.getMSC( psc->driScreen.private, & temp );
-	 *count = (unsigned) temp;
-	 return (ret == 0) ? 0 : GLX_BAD_CONTEXT;
+	  return (ret == 0) ? 0 : GLX_BAD_CONTEXT;
       }
    }
 #else
@@ -1906,32 +1996,26 @@ static int __glXGetVideoSyncSGI(unsigned int *count)
 
 static int __glXWaitVideoSyncSGI(int divisor, int remainder, unsigned int *count)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_MEDIA_STREAM_COUNTER
    GLXContext gc = __glXGetCurrentContext();
 
    if ( divisor <= 0 || remainder < 0 )
      return GLX_BAD_VALUE;
 
-   if ( (gc != NULL) && gc->isDirect ) {
+   if (gc != NULL && gc->driContext) {
       __GLXscreenConfigs * const psc = GetGLXScreenConfigs( gc->currentDpy,
 							    gc->screen );
-      if ( __glXExtensionBitIsEnabled( psc, SGI_video_sync_bit )
-	   && psc->driScreen.private ) {
-	 __DRIdrawable * const pdraw = 
-	     (*psc->driScreen.getDrawable)(gc->currentDpy,
-					   gc->currentDrawable,
-					   psc->driScreen.private);
-	 if ( (pdraw != NULL) && (pdraw->waitForMSC != NULL) ) {
-	    int       ret;
-	    int64_t   msc;
-	    int64_t   sbc;
-
-	    ret = (*pdraw->waitForMSC)( gc->currentDpy, pdraw->private,
-					0, divisor, remainder,
-					& msc, & sbc );
-	    *count = (unsigned) msc;
-	    return (ret == 0) ? 0 : GLX_BAD_CONTEXT;
-	 }
+      if (psc->msc != NULL && psc->driScreen ) {
+	 __GLXDRIdrawable *pdraw = 
+	     GetGLXDRIDrawable(gc->currentDpy, gc->currentDrawable, NULL);
+	 int       ret;
+	 int64_t   msc;
+	 int64_t   sbc;
+
+	 ret = (*psc->msc->waitForMSC)(pdraw->driDrawable, 0,
+				       divisor, remainder, &msc, &sbc);
+	 *count = (unsigned) msc;
+	 return (ret == 0) ? 0 : GLX_BAD_CONTEXT;
       }
    }
 #else
@@ -2083,20 +2167,19 @@ static Bool __glXQueryMaxSwapBarriersSGIX(Display *dpy, int screen, int *max)
 static Bool __glXGetSyncValuesOML(Display *dpy, GLXDrawable drawable,
 				  int64_t *ust, int64_t *msc, int64_t *sbc)
 {
-#ifdef GLX_DIRECT_RENDERING
+#if defined(__DRI_SWAP_BUFFER_COUNTER) && defined(__DRI_MEDIA_STREAM_COUNTER)
     __GLXdisplayPrivate * const priv = __glXInitialize(dpy);
 
     if ( priv != NULL ) {
 	int   i;
-	__DRIdrawable * const pdraw = GetDRIDrawable( dpy, drawable, & i );
+	__GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &i);
 	__GLXscreenConfigs * const psc = &priv->screenConfigs[i];
 
 	assert( (pdraw == NULL) || (i != -1) );
-	return ( (pdraw && pdraw->getSBC && psc->driScreen.getMSC)
-		 && __glXExtensionBitIsEnabled( psc, OML_sync_control_bit )
-		 && ((*psc->driScreen.getMSC)( psc->driScreen.private, msc ) == 0)
-		 && ((*pdraw->getSBC)( dpy, psc->driScreen.private, sbc ) == 0)
-		 && (__glXGetUST( ust ) == 0) );
+	return ( (pdraw && psc->sbc && psc->msc)
+		 && ((*psc->msc->getMSC)(psc->driScreen, msc) == 0)
+		 && ((*psc->sbc->getSBC)(pdraw->driDrawable, sbc) == 0)
+		 && (__glXGetUST(ust) == 0) );
     }
 #else
    (void) dpy;
@@ -2108,6 +2191,68 @@ static Bool __glXGetSyncValuesOML(Display *dpy, GLXDrawable drawable,
    return False;
 }
 
+#ifdef GLX_DIRECT_RENDERING
+_X_HIDDEN GLboolean
+__driGetMscRateOML(__DRIdrawable *draw,
+		   int32_t *numerator, int32_t *denominator, void *private)
+{
+#ifdef XF86VIDMODE
+    __GLXscreenConfigs *psc;
+    XF86VidModeModeLine   mode_line;
+    int   dot_clock;
+    int   i;
+    __GLXDRIdrawable *glxDraw = private;
+
+    psc = glxDraw->psc;
+    if (XF86VidModeQueryVersion(psc->dpy, &i, &i) &&
+	XF86VidModeGetModeLine(psc->dpy, psc->scr, &dot_clock, &mode_line) ) {
+	unsigned   n = dot_clock * 1000;
+	unsigned   d = mode_line.vtotal * mode_line.htotal;
+	
+# define V_INTERLACE 0x010
+# define V_DBLSCAN   0x020
+
+	if (mode_line.flags & V_INTERLACE)
+	    n *= 2;
+	else if (mode_line.flags & V_DBLSCAN)
+	    d *= 2;
+
+	/* The OML_sync_control spec requires that if the refresh rate is a
+	 * whole number, that the returned numerator be equal to the refresh
+	 * rate and the denominator be 1.
+	 */
+
+	if (n % d == 0) {
+	    n /= d;
+	    d = 1;
+	}
+	else {
+	    static const unsigned f[] = { 13, 11, 7, 5, 3, 2, 0 };
+
+	    /* This is a poor man's way to reduce a fraction.  It's far from
+	     * perfect, but it will work well enough for this situation.
+	     */
+
+	    for (i = 0; f[i] != 0; i++) {
+		while (n % f[i] == 0 && d % f[i] == 0) {
+		    d /= f[i];
+		    n /= f[i];
+		}
+	    }
+	}
+
+	*numerator = n;
+	*denominator = d;
+
+	return True;
+    }
+    else
+	return False;
+#else
+    return False;
+#endif
+}
+#endif
 
 /**
  * Determine the refresh rate of the specified drawable and display.
@@ -2125,70 +2270,17 @@ static Bool __glXGetSyncValuesOML(Display *dpy, GLXDrawable drawable,
  *       when GLX_OML_sync_control appears in the client extension string.
  */
 
-Bool __glXGetMscRateOML(Display * dpy, GLXDrawable drawable,
-			int32_t * numerator, int32_t * denominator)
+_X_HIDDEN GLboolean __glXGetMscRateOML(Display * dpy, GLXDrawable drawable,
+				       int32_t * numerator,
+				       int32_t * denominator)
 {
 #if defined( GLX_DIRECT_RENDERING ) && defined( XF86VIDMODE )
-   __GLXdisplayPrivate * const priv = __glXInitialize(dpy);
-
+    __GLXDRIdrawable *draw = GetGLXDRIDrawable(dpy, drawable, NULL);
 
-   if ( priv != NULL ) {
-      XF86VidModeModeLine   mode_line;
-      int   dot_clock;
-      int   screen_num;
-      int   i;
-
-
-      GetDRIDrawable( dpy, drawable, & screen_num );
-      if ( (screen_num != -1)
-	   && XF86VidModeQueryVersion( dpy, & i, & i )
-	   && XF86VidModeGetModeLine( dpy, screen_num, & dot_clock,
-				      & mode_line ) ) {
-	 unsigned   n = dot_clock * 1000;
-	 unsigned   d = mode_line.vtotal * mode_line.htotal;
-
-# define V_INTERLACE 0x010
-# define V_DBLSCAN   0x020
-
-	 if ( (mode_line.flags & V_INTERLACE) ) {
-	    n *= 2;
-	 }
-	 else if ( (mode_line.flags & V_DBLSCAN) ) {
-	    d *= 2;
-	 }
-
-	 /* The OML_sync_control spec requires that if the refresh rate is a
-	  * whole number, that the returned numerator be equal to the refresh
-	  * rate and the denominator be 1.
-	  */
-
-	 if ( (n % d) == 0 ) {
-	    n /= d;
-	    d = 1;
-	 }
-	 else {
-	    static const unsigned f[] = { 13, 11, 7, 5, 3, 2, 0 };
-
-
-	    /* This is a poor man's way to reduce a fraction.  It's far from
-	     * perfect, but it will work well enough for this situation.
-	     */
-
-	    for ( i = 0 ; f[i] != 0 ; i++ ) {
-	       while ( ((n % f[i]) == 0) && ((d % f[i]) == 0) ) {
-		  d /= f[i];
-		  n /= f[i];
-	       }
-	    }
-	 }
-
-	 *numerator = n;
-	 *denominator = d;
+    if (draw == NULL)
+	return False;
 
-	 (void) drawable;
-	 return True;
-      }
-   }
+    return __driGetMscRateOML(draw->driDrawable, numerator, denominator, draw);
 #else
    (void) dpy;
    (void) drawable;
@@ -2203,9 +2295,9 @@ static int64_t __glXSwapBuffersMscOML(Display *dpy, GLXDrawable drawable,
 				      int64_t target_msc, int64_t divisor,
 				      int64_t remainder)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_SWAP_BUFFER_COUNTER
    int screen;
-   __DRIdrawable *pdraw = GetDRIDrawable( dpy, drawable, & screen );
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, screen );
 
    /* The OML_sync_control spec says these should "generate a GLX_BAD_VALUE
@@ -2218,11 +2310,10 @@ static int64_t __glXSwapBuffersMscOML(Display *dpy, GLXDrawable drawable,
    if ( divisor > 0 && remainder >= divisor )
       return -1;
 
-   if ( (pdraw != NULL) && (pdraw->swapBuffersMSC != NULL)
-       && __glXExtensionBitIsEnabled( psc, OML_sync_control_bit ) ) {
-      return (*pdraw->swapBuffersMSC)(dpy, pdraw->private, target_msc,
-				      divisor, remainder);
-   }
+   if (pdraw != NULL && psc->counters != NULL)
+      return (*psc->sbc->swapBuffersMSC)(pdraw->driDrawable, target_msc,
+					 divisor, remainder);
+
 #else
    (void) dpy;
    (void) drawable;
@@ -2239,9 +2330,9 @@ static Bool __glXWaitForMscOML(Display * dpy, GLXDrawable drawable,
 			       int64_t remainder, int64_t *ust,
 			       int64_t *msc, int64_t *sbc)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_MEDIA_STREAM_COUNTER
    int screen;
-   __DRIdrawable *pdraw = GetDRIDrawable( dpy, drawable, & screen );
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, screen );
    int  ret;
 
@@ -2253,10 +2344,9 @@ static Bool __glXWaitForMscOML(Display * dpy, GLXDrawable drawable,
    if ( divisor > 0 && remainder >= divisor )
       return False;
 
-   if ( (pdraw != NULL) && (pdraw->waitForMSC != NULL)
-	&& __glXExtensionBitIsEnabled( psc, OML_sync_control_bit ) ) {
-      ret = (*pdraw->waitForMSC)( dpy, pdraw->private, target_msc,
-				  divisor, remainder, msc, sbc );
+   if (pdraw != NULL && psc->msc != NULL) {
+      ret = (*psc->msc->waitForMSC)(pdraw->driDrawable, target_msc,
+				    divisor, remainder, msc, sbc);
 
       /* __glXGetUST returns zero on success and non-zero on failure.
        * This function returns True on success and False on failure.
@@ -2281,9 +2371,9 @@ static Bool __glXWaitForSbcOML(Display * dpy, GLXDrawable drawable,
 			       int64_t target_sbc, int64_t *ust,
 			       int64_t *msc, int64_t *sbc )
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_SWAP_BUFFER_COUNTER
    int screen;
-   __DRIdrawable *pdraw = GetDRIDrawable( dpy, drawable, & screen );
+   __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, screen );
    int  ret;
 
@@ -2293,9 +2383,8 @@ static Bool __glXWaitForSbcOML(Display * dpy, GLXDrawable drawable,
    if ( target_sbc < 0 )
       return False;
 
-   if ( (pdraw != NULL) && (pdraw->waitForSBC != NULL)
-	&& __glXExtensionBitIsEnabled( psc, OML_sync_control_bit )) {
-      ret = (*pdraw->waitForSBC)( dpy, pdraw->private, target_sbc, msc, sbc );
+   if (pdraw != NULL && psc->sbc != NULL) {
+      ret = (*psc->sbc->waitForSBC)(pdraw->driDrawable, target_sbc, msc, sbc);
 
       /* __glXGetUST returns zero on success and non-zero on failure.
        * This function returns True on success and False on failure.
@@ -2323,16 +2412,13 @@ PUBLIC void *glXAllocateMemoryMESA(Display *dpy, int scrn,
 				   size_t size, float readFreq,
 				   float writeFreq, float priority)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_ALLOCATE
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, scrn );
 
-   if ( __glXExtensionBitIsEnabled( psc, MESA_allocate_memory_bit ) ) {
-      if (psc && psc->driScreen.private && psc->driScreen.allocateMemory) {
-	 return (*psc->driScreen.allocateMemory)( dpy, scrn, size,
-						  readFreq, writeFreq,
-						  priority );
-      }
-   }
+   if (psc && psc->allocate)
+       return (*psc->allocate->allocateMemory)(psc->__driScreen, size,
+					       readFreq, writeFreq, priority);
+
 #else
    (void) dpy;
    (void) scrn;
@@ -2348,14 +2434,12 @@ PUBLIC void *glXAllocateMemoryMESA(Display *dpy, int scrn,
 
 PUBLIC void glXFreeMemoryMESA(Display *dpy, int scrn, void *pointer)
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_ALLOCATE
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, scrn );
 
-   if ( __glXExtensionBitIsEnabled( psc, MESA_allocate_memory_bit ) ) {
-      if (psc && psc->driScreen.private && psc->driScreen.freeMemory) {
-	 (*psc->driScreen.freeMemory)( dpy, scrn, pointer );
-      }
-   }
+   if (psc && psc->allocate)
+	 (*psc->allocate->freeMemory)(psc->__driScreen, pointer);
+
 #else
    (void) dpy;
    (void) scrn;
@@ -2367,14 +2451,12 @@ PUBLIC void glXFreeMemoryMESA(Display *dpy, int scrn, void *pointer)
 PUBLIC GLuint glXGetMemoryOffsetMESA( Display *dpy, int scrn,
 				      const void *pointer )
 {
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_ALLOCATE
    __GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, scrn );
 
-   if ( __glXExtensionBitIsEnabled( psc, MESA_allocate_memory_bit ) ) {
-      if (psc && psc->driScreen.private && psc->driScreen.memoryOffset) {
-	 return (*psc->driScreen.memoryOffset)( dpy, scrn, pointer );
-      }
-   }
+   if (psc && psc->allocate)
+       return (*psc->allocate->memoryOffset)(psc->__driScreen, pointer);
+
 #else
    (void) dpy;
    (void) scrn;
@@ -2447,13 +2529,14 @@ static void __glXCopySubBufferMESA(Display *dpy, GLXDrawable drawable,
     INT32 *x_ptr, *y_ptr, *w_ptr, *h_ptr;
     CARD8 opcode;
 
-#ifdef GLX_DIRECT_RENDERING
+#ifdef __DRI_COPY_SUB_BUFFER
     int screen;
-    __DRIdrawable *pdraw = GetDRIDrawable( dpy, drawable, & screen );
+    __GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, &screen);
     if ( pdraw != NULL ) {
-	__GLXscreenConfigs * const psc = GetGLXScreenConfigs( dpy, screen );
-	if ( __glXExtensionBitIsEnabled( psc, MESA_copy_sub_buffer_bit ) ) {
-	    (*pdraw->copySubBuffer)(dpy, pdraw->private, x, y, width, height);
+	__GLXscreenConfigs * const psc = GetGLXScreenConfigs(dpy, screen);
+	if (psc->driScreen->copySubBuffer != NULL) {
+	    glFlush();	    
+	    (*psc->driScreen->copySubBuffer)(pdraw, x, y, width, height);
 	}
 
 	return;
@@ -2529,8 +2612,16 @@ static void __glXBindTexImageEXT(Display *dpy,
     }
  
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect)
+    if (gc->driContext) {
+	__GLXDRIdrawable *pdraw = GetGLXDRIDrawable(dpy, drawable, NULL);
+
+	if (pdraw != NULL)
+	    (*pdraw->psc->texBuffer->setTexBuffer)(gc->__driContext,
+						   pdraw->textureTarget,
+						   pdraw->driDrawable);
+
 	return;
+    }
 #endif
 
     opcode = __glXSetupForCommand(dpy);
@@ -2581,7 +2672,7 @@ static void __glXReleaseTexImageEXT(Display *dpy,
 	return;
 
 #ifdef GLX_DIRECT_RENDERING
-    if (gc->isDirect)
+    if (gc->driContext)
 	return;
 #endif
 
@@ -2613,7 +2704,7 @@ static void __glXReleaseTexImageEXT(Display *dpy,
  * 
  * \sa strdup
  */
-char *
+_X_HIDDEN char *
 __glXstrdup(const char *str)
 {
    char *copy;
@@ -2841,98 +2932,6 @@ PUBLIC void (*glXGetProcAddress(const GLubyte *procName))( void )
 
 #ifdef GLX_DIRECT_RENDERING
 /**
- * Retrieves the verion of the internal libGL API in YYYYMMDD format.  This
- * might be used by the DRI drivers to determine how new libGL is at runtime.
- * Drivers should not call this function directly.  They should instead use
- * \c glXGetProcAddress to obtain a pointer to the function.
- * 
- * \returns An 8-digit decimal number representing the internal libGL API in
- *          YYYYMMDD format.
- * 
- * \sa glXGetProcAddress, PFNGLXGETINTERNALVERSIONPROC
- *
- * \since Internal API version 20021121.
- */
-int __glXGetInternalVersion(void)
-{
-    /* History:
-     * 20021121 - Initial version
-     * 20021128 - Added __glXWindowExists() function
-     * 20021207 - Added support for dynamic GLX extensions,
-     *            GLX_SGI_swap_control, GLX_SGI_video_sync,
-     *            GLX_OML_sync_control, and GLX_MESA_swap_control.
-     *            Never officially released.  Do NOT test against
-     *            this version.  Use 20030317 instead.
-     * 20030317 - Added support GLX_SGIX_fbconfig,
-     *            GLX_MESA_swap_frame_usage, GLX_OML_swap_method,
-     *            GLX_{ARB,SGIS}_multisample, and
-     *            GLX_SGIX_visual_select_group.
-     * 20030606 - Added support for GLX_SGI_make_current_read.
-     * 20030813 - Made support for dynamic extensions multi-head aware.
-     * 20030818 - Added support for GLX_MESA_allocate_memory in place of the
-     *            deprecated GLX_NV_vertex_array_range & GLX_MESA_agp_offset
-     *            interfaces.
-     * 20031201 - Added support for the first round of DRI interface changes.
-     *            Do NOT test against this version!  It has binary
-     *            compatibility bugs, use 20040317 instead.
-     * 20040317 - Added the 'mode' field to __DRIcontextRec.
-     * 20040415 - Added support for bindContext3 and unbindContext3.
-     * 20040602 - Add __glXGetDrawableInfo.  I though that was there
-     *            months ago. :(
-     * 20050727 - Gut all the old interfaces.  This breaks compatability with
-     *            any DRI driver built to any previous version.
-     * 20060314 - Added support for GLX_MESA_copy_sub_buffer.
-     * 20070105 - Added support for damage reporting.
-     */
-    return 20070105;
-}
-
-
-
-static Bool windowExistsFlag;
-
-static int windowExistsErrorHandler(Display *dpy, XErrorEvent *xerr)
-{
-    if (xerr->error_code == BadWindow) {
-        windowExistsFlag = GL_FALSE;
-    }
-    return 0;
-}
-
-/**
- * Determine if a window associated with a \c GLXDrawable exists on the
- * X-server.  This function is not used internally by libGL.  It is provided
- * as a utility function for DRI drivers.
- * Drivers should not call this function directly.  They should instead use
- * \c glXGetProcAddress to obtain a pointer to the function.
- *
- * \param dpy  Display associated with the drawable to be queried.
- * \param draw \c GLXDrawable to test.
- * 
- * \returns \c GL_TRUE if a window exists that is associated with \c draw,
- *          otherwise \c GL_FALSE is returned.
- * 
- * \warning This function is not currently thread-safe.
- *
- * \sa glXGetProcAddress
- *
- * \since Internal API version 20021128.
- */
-Bool __glXWindowExists(Display *dpy, GLXDrawable draw)
-{
-    XWindowAttributes xwa;
-    int (*oldXErrorHandler)(Display *, XErrorEvent *);
-
-    XSync(dpy, GL_FALSE);
-    windowExistsFlag = GL_TRUE;
-    oldXErrorHandler = XSetErrorHandler(windowExistsErrorHandler);
-    XGetWindowAttributes(dpy, draw, &xwa); /* dummy request */
-    XSetErrorHandler(oldXErrorHandler);
-    return windowExistsFlag;
-}
-
-
-/**
  * Get the unadjusted system time (UST).  Currently, the UST is measured in
  * microseconds since Epoc.  The actual resolution of the UST may vary from
  * system to system, and the units may vary from release to release.
@@ -2946,7 +2945,7 @@ Bool __glXWindowExists(Display *dpy, GLXDrawable draw)
  *
  * \since Internal API version 20030317.
  */
-int __glXGetUST( int64_t * ust )
+_X_HIDDEN int __glXGetUST( int64_t * ust )
 {
     struct timeval  tv;
     
diff --git a/src/glx/x11/glxcurrent.c b/src/glx/x11/glxcurrent.c
new file mode 100644
index 00000000000..f3eb045d9f9
--- /dev/null
+++ b/src/glx/x11/glxcurrent.c
@@ -0,0 +1,531 @@
+/*
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
+
+/**
+ * \file glxcurrent.c
+ * Client-side GLX interface for current context management.
+ */
+
+#include "glxclient.h"
+#include "glapi.h"
+#include "indirect_init.h"
+
+#ifdef GLX_DIRECT_RENDERING
+#include "xf86dri.h"
+#endif
+
+/*
+** We setup some dummy structures here so that the API can be used
+** even if no context is current.
+*/
+
+static GLubyte dummyBuffer[__GLX_BUFFER_LIMIT_SIZE];
+
+/*
+** Dummy context used by small commands when there is no current context.
+** All the
+** gl and glx entry points are designed to operate as nop's when using
+** the dummy context structure.
+*/
+static __GLXcontext dummyContext = {
+   &dummyBuffer[0],
+   &dummyBuffer[0],
+   &dummyBuffer[0],
+   &dummyBuffer[__GLX_BUFFER_LIMIT_SIZE],
+   sizeof(dummyBuffer),
+};
+
+
+/*
+** All indirect rendering contexts will share the same indirect dispatch table.
+*/
+static __GLapi *IndirectAPI = NULL;
+
+
+/*
+ * Current context management and locking
+ */
+
+#if defined( USE_XTHREADS )
+
+/* thread safe */
+static GLboolean TSDinitialized = GL_FALSE;
+static xthread_key_t ContextTSD;
+
+_X_HIDDEN __GLXcontext *
+__glXGetCurrentContext(void)
+{
+   if (!TSDinitialized) {
+      xthread_key_create(&ContextTSD, NULL);
+      TSDinitialized = GL_TRUE;
+      return &dummyContext;
+   }
+   else {
+      void *p;
+      xthread_get_specific(ContextTSD, &p);
+      if (!p)
+         return &dummyContext;
+      else
+         return (__GLXcontext *) p;
+   }
+}
+
+_X_HIDDEN void
+__glXSetCurrentContext(__GLXcontext * c)
+{
+   if (!TSDinitialized) {
+      xthread_key_create(&ContextTSD, NULL);
+      TSDinitialized = GL_TRUE;
+   }
+   xthread_set_specific(ContextTSD, c);
+}
+
+
+/* Used by the __glXLock() and __glXUnlock() macros */
+_X_HIDDEN xmutex_rec __glXmutex;
+
+#elif defined( PTHREADS )
+
+_X_HIDDEN pthread_mutex_t __glXmutex = PTHREAD_MUTEX_INITIALIZER;
+
+# if defined( GLX_USE_TLS )
+
+/**
+ * Per-thread GLX context pointer.
+ * 
+ * \c __glXSetCurrentContext is written is such a way that this pointer can
+ * \b never be \c NULL.  This is important!  Because of this
+ * \c __glXGetCurrentContext can be implemented as trivial macro.
+ */
+__thread void *__glX_tls_Context __attribute__ ((tls_model("initial-exec")))
+   = &dummyContext;
+
+_X_HIDDEN void
+__glXSetCurrentContext(__GLXcontext * c)
+{
+   __glX_tls_Context = (c != NULL) ? c : &dummyContext;
+}
+
+# else
+
+static pthread_once_t once_control = PTHREAD_ONCE_INIT;
+
+/**
+ * Per-thread data key.
+ * 
+ * Once \c init_thread_data has been called, the per-thread data key will
+ * take a value of \c NULL.  As each new thread is created the default
+ * value, in that thread, will be \c NULL.
+ */
+static pthread_key_t ContextTSD;
+
+/**
+ * Initialize the per-thread data key.
+ * 
+ * This function is called \b exactly once per-process (not per-thread!) to
+ * initialize the per-thread data key.  This is ideally done using the
+ * \c pthread_once mechanism.
+ */
+static void
+init_thread_data(void)
+{
+   if (pthread_key_create(&ContextTSD, NULL) != 0) {
+      perror("pthread_key_create");
+      exit(-1);
+   }
+}
+
+_X_HIDDEN void
+__glXSetCurrentContext(__GLXcontext * c)
+{
+   pthread_once(&once_control, init_thread_data);
+   pthread_setspecific(ContextTSD, c);
+}
+
+_X_HIDDEN __GLXcontext *
+__glXGetCurrentContext(void)
+{
+   void *v;
+
+   pthread_once(&once_control, init_thread_data);
+
+   v = pthread_getspecific(ContextTSD);
+   return (v == NULL) ? &dummyContext : (__GLXcontext *) v;
+}
+
+# endif /* defined( GLX_USE_TLS ) */
+
+#elif defined( THREADS )
+
+#error Unknown threading method specified.
+
+#else
+
+/* not thread safe */
+_X_HIDDEN __GLXcontext *__glXcurrentContext = &dummyContext;
+
+#endif
+
+
+_X_HIDDEN void
+__glXSetCurrentContextNull(void)
+{
+   __glXSetCurrentContext(&dummyContext);
+#ifdef GLX_DIRECT_RENDERING
+   _glapi_set_dispatch(NULL);   /* no-op functions */
+#endif
+}
+
+
+/************************************************************************/
+
+PUBLIC GLXContext
+glXGetCurrentContext(void)
+{
+   GLXContext cx = __glXGetCurrentContext();
+
+   if (cx == &dummyContext) {
+      return NULL;
+   }
+   else {
+      return cx;
+   }
+}
+
+PUBLIC GLXDrawable
+glXGetCurrentDrawable(void)
+{
+   GLXContext gc = __glXGetCurrentContext();
+   return gc->currentDrawable;
+}
+
+
+/************************************************************************/
+
+/**
+ * Sends a GLX protocol message to the specified display to make the context
+ * and the drawables current.
+ *
+ * \param dpy     Display to send the message to.
+ * \param opcode  Major opcode value for the display.
+ * \param gc_id   Context tag for the context to be made current.
+ * \param draw    Drawable ID for the "draw" drawable.
+ * \param read    Drawable ID for the "read" drawable.
+ * \param reply   Space to store the X-server's reply.
+ *
+ * \warning
+ * This function assumes that \c dpy is locked with \c LockDisplay on entry.
+ */
+static Bool
+SendMakeCurrentRequest(Display * dpy, CARD8 opcode,
+                       GLXContextID gc_id, GLXContextTag gc_tag,
+                       GLXDrawable draw, GLXDrawable read,
+                       xGLXMakeCurrentReply * reply)
+{
+   Bool ret;
+
+
+   LockDisplay(dpy);
+
+   if (draw == read) {
+      xGLXMakeCurrentReq *req;
+
+      GetReq(GLXMakeCurrent, req);
+      req->reqType = opcode;
+      req->glxCode = X_GLXMakeCurrent;
+      req->drawable = draw;
+      req->context = gc_id;
+      req->oldContextTag = gc_tag;
+   }
+   else {
+      __GLXdisplayPrivate *priv = __glXInitialize(dpy);
+
+      /* If the server can support the GLX 1.3 version, we should
+       * perfer that.  Not only that, some servers support GLX 1.3 but
+       * not the SGI extension.
+       */
+
+      if ((priv->majorVersion > 1) || (priv->minorVersion >= 3)) {
+         xGLXMakeContextCurrentReq *req;
+
+         GetReq(GLXMakeContextCurrent, req);
+         req->reqType = opcode;
+         req->glxCode = X_GLXMakeContextCurrent;
+         req->drawable = draw;
+         req->readdrawable = read;
+         req->context = gc_id;
+         req->oldContextTag = gc_tag;
+      }
+      else {
+         xGLXVendorPrivateWithReplyReq *vpreq;
+         xGLXMakeCurrentReadSGIReq *req;
+
+         GetReqExtra(GLXVendorPrivateWithReply,
+                     sz_xGLXMakeCurrentReadSGIReq -
+                     sz_xGLXVendorPrivateWithReplyReq, vpreq);
+         req = (xGLXMakeCurrentReadSGIReq *) vpreq;
+         req->reqType = opcode;
+         req->glxCode = X_GLXVendorPrivateWithReply;
+         req->vendorCode = X_GLXvop_MakeCurrentReadSGI;
+         req->drawable = draw;
+         req->readable = read;
+         req->context = gc_id;
+         req->oldContextTag = gc_tag;
+      }
+   }
+
+   ret = _XReply(dpy, (xReply *) reply, 0, False);
+
+   UnlockDisplay(dpy);
+   SyncHandle();
+
+   return ret;
+}
+
+
+#ifdef GLX_DIRECT_RENDERING
+static __GLXDRIdrawable *
+FetchDRIDrawable(Display * dpy,
+                 GLXDrawable glxDrawable, GLXContext gc, Bool pre13)
+{
+   __GLXdisplayPrivate *const priv = __glXInitialize(dpy);
+   __GLXDRIdrawable *pdraw;
+   __GLXscreenConfigs *psc;
+   XID drawable;
+
+   if (priv == NULL)
+      return NULL;
+
+   psc = &priv->screenConfigs[gc->screen];
+   if (psc->drawHash == NULL)
+      return NULL;
+
+   if (__glxHashLookup(psc->drawHash, glxDrawable, (void *) &pdraw) == 0)
+      return pdraw;
+
+   /* If this is glXMakeCurrent (pre GLX 1.3) we allow creating the
+    * GLX drawable on the fly.  Otherwise we pass None as the X
+    * drawable */
+   if (pre13)
+      drawable = glxDrawable;
+   else
+      drawable = None;
+
+   pdraw = psc->driScreen->createDrawable(psc, drawable,
+                                          glxDrawable, gc->mode);
+   if (__glxHashInsert(psc->drawHash, glxDrawable, pdraw)) {
+      (*pdraw->destroyDrawable) (pdraw);
+      return NULL;
+   }
+
+   return pdraw;
+}
+#endif /* GLX_DIRECT_RENDERING */
+
+
+/**
+ * Make a particular context current.
+ * 
+ * \note This is in this file so that it can access dummyContext.
+ */
+static Bool
+MakeContextCurrent(Display * dpy, GLXDrawable draw,
+                   GLXDrawable read, GLXContext gc, Bool pre13)
+{
+   xGLXMakeCurrentReply reply;
+   const GLXContext oldGC = __glXGetCurrentContext();
+   const CARD8 opcode = __glXSetupForCommand(dpy);
+   const CARD8 oldOpcode = ((gc == oldGC) || (oldGC == &dummyContext))
+      ? opcode : __glXSetupForCommand(oldGC->currentDpy);
+   Bool bindReturnValue;
+
+
+   if (!opcode || !oldOpcode) {
+      return GL_FALSE;
+   }
+
+   /* Make sure that the new context has a nonzero ID.  In the request,
+    * a zero context ID is used only to mean that we bind to no current
+    * context.
+    */
+   if ((gc != NULL) && (gc->xid == None)) {
+      return GL_FALSE;
+   }
+
+   _glapi_check_multithread();
+
+#ifdef GLX_DIRECT_RENDERING
+   /* Bind the direct rendering context to the drawable */
+   if (gc && gc->driContext) {
+      __GLXDRIdrawable *pdraw = FetchDRIDrawable(dpy, draw, gc, pre13);
+      __GLXDRIdrawable *pread = FetchDRIDrawable(dpy, read, gc, pre13);
+
+      bindReturnValue =
+         (gc->driContext->bindContext) (gc->driContext, pdraw, pread);
+   }
+   else
+#endif
+   {
+      /* Send a glXMakeCurrent request to bind the new context. */
+      bindReturnValue =
+         SendMakeCurrentRequest(dpy, opcode, gc ? gc->xid : None,
+                                ((dpy != oldGC->currentDpy)
+                                 || oldGC->isDirect)
+                                ? None : oldGC->currentContextTag, draw, read,
+                                &reply);
+   }
+
+
+   if (!bindReturnValue) {
+      return False;
+   }
+
+#ifdef GLX_DIRECT_RENDERING
+   if ((dpy != oldGC->currentDpy || (gc && gc->driContext)) &&
+       !oldGC->isDirect && oldGC != &dummyContext) {
+#else
+   if ((dpy != oldGC->currentDpy) && oldGC != &dummyContext) {
+#endif
+      xGLXMakeCurrentReply dummy_reply;
+
+      /* We are either switching from one dpy to another and have to
+       * send a request to the previous dpy to unbind the previous
+       * context, or we are switching away from a indirect context to
+       * a direct context and have to send a request to the dpy to
+       * unbind the previous context.
+       */
+      (void) SendMakeCurrentRequest(oldGC->currentDpy, oldOpcode, None,
+                                    oldGC->currentContextTag, None, None,
+                                    &dummy_reply);
+   }
+#ifdef GLX_DIRECT_RENDERING
+   else if (oldGC->driContext) {
+      oldGC->driContext->unbindContext(oldGC->driContext);
+   }
+#endif
+
+
+   /* Update our notion of what is current */
+   __glXLock();
+   if (gc == oldGC) {
+      /* Even though the contexts are the same the drawable might have
+       * changed.  Note that gc cannot be the dummy, and that oldGC
+       * cannot be NULL, therefore if they are the same, gc is not
+       * NULL and not the dummy.
+       */
+      gc->currentDrawable = draw;
+      gc->currentReadable = read;
+   }
+   else {
+      if (oldGC != &dummyContext) {
+         /* Old current context is no longer current to anybody */
+         oldGC->currentDpy = 0;
+         oldGC->currentDrawable = None;
+         oldGC->currentReadable = None;
+         oldGC->currentContextTag = 0;
+
+         if (oldGC->xid == None) {
+            /* We are switching away from a context that was
+             * previously destroyed, so we need to free the memory
+             * for the old handle.
+             */
+#ifdef GLX_DIRECT_RENDERING
+            /* Destroy the old direct rendering context */
+            if (oldGC->driContext) {
+               oldGC->driContext->destroyContext(oldGC->driContext,
+                                                 oldGC->psc,
+                                                 oldGC->createDpy);
+               oldGC->driContext = NULL;
+            }
+#endif
+            __glXFreeContext(oldGC);
+         }
+      }
+      if (gc) {
+         __glXSetCurrentContext(gc);
+
+         gc->currentDpy = dpy;
+         gc->currentDrawable = draw;
+         gc->currentReadable = read;
+
+#ifdef GLX_DIRECT_RENDERING
+         if (!gc->driContext) {
+#endif
+            if (!IndirectAPI)
+               IndirectAPI = __glXNewIndirectAPI();
+            _glapi_set_dispatch(IndirectAPI);
+
+#ifdef GLX_USE_APPLEGL
+            do {
+               extern void XAppleDRIUseIndirectDispatch(void);
+               XAppleDRIUseIndirectDispatch();
+            } while (0);
+#endif
+
+            __GLXattribute *state =
+               (__GLXattribute *) (gc->client_state_private);
+
+            gc->currentContextTag = reply.contextTag;
+            if (state->array_state == NULL) {
+               (void) glGetString(GL_EXTENSIONS);
+               (void) glGetString(GL_VERSION);
+               __glXInitVertexArrayState(gc);
+            }
+#ifdef GLX_DIRECT_RENDERING
+         }
+         else {
+            gc->currentContextTag = -1;
+         }
+#endif
+      }
+      else {
+         __glXSetCurrentContextNull();
+      }
+   }
+   __glXUnlock();
+   return GL_TRUE;
+}
+
+
+PUBLIC Bool
+glXMakeCurrent(Display * dpy, GLXDrawable draw, GLXContext gc)
+{
+   return MakeContextCurrent(dpy, draw, draw, gc, True);
+}
+
+PUBLIC
+GLX_ALIAS(Bool, glXMakeCurrentReadSGI,
+          (Display * dpy, GLXDrawable d, GLXDrawable r, GLXContext ctx),
+          (dpy, d, r, ctx, False), MakeContextCurrent)
+
+     PUBLIC GLX_ALIAS(Bool, glXMakeContextCurrent,
+                      (Display * dpy, GLXDrawable d, GLXDrawable r,
+                       GLXContext ctx), (dpy, d, r, ctx, False),
+                      MakeContextCurrent)
diff --git a/src/glx/x11/glxext.c b/src/glx/x11/glxext.c
index 6403cbd56df..be6edf9b193 100644
--- a/src/glx/x11/glxext.c
+++ b/src/glx/x11/glxext.c
@@ -1,38 +1,32 @@
-
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /**
  * \file glxext.c
@@ -41,28 +35,15 @@
  * Direct rendering support added by Precision Insight, Inc.
  *
  * \author Kevin E. Martin <kevin@precisioninsight.com>
- */     
+ */
 
+#include <assert.h>
 #include "glxclient.h"
-#include <stdio.h>
 #include <X11/extensions/Xext.h>
 #include <X11/extensions/extutil.h>
-#include <X11/extensions/Xfixes.h>
-#include <X11/extensions/Xdamage.h>
-#include <assert.h>
-#include "indirect_init.h"
 #include "glapi.h"
 #include "glxextensions.h"
 #include "glcontextmodes.h"
-#include "glheader.h"
-
-#ifdef GLX_DIRECT_RENDERING
-#include <inttypes.h>
-#include <sys/mman.h>
-#include "xf86dri.h"
-#include "sarea.h"
-#include "dri_glx.h"
-#endif
 
 #ifdef USE_XCB
 #include <X11/Xlib-xcb.h>
@@ -70,210 +51,29 @@
 #include <xcb/glx.h>
 #endif
 
-#include <assert.h>
 
 #ifdef DEBUG
-void __glXDumpDrawBuffer(__GLXcontext *ctx);
+void __glXDumpDrawBuffer(__GLXcontext * ctx);
 #endif
 
 #ifdef USE_SPARC_ASM
-/*
- * This is where our dispatch table's bounds are.
- * And the static mesa_init is taken directly from
- * Mesa's 'sparc.c' initializer.
- *
- * We need something like this here, because this version
- * of openGL/glx never initializes a Mesa context, and so
- * the address of the dispatch table pointer never gets stuffed
- * into the dispatch jump table otherwise.
- *
- * It matters only on SPARC, and only if you are using assembler
- * code instead of C-code indirect dispatch.
- *
- * -- FEM, 04.xii.03
- */
-extern unsigned int _mesa_sparc_glapi_begin;
-extern unsigned int _mesa_sparc_glapi_end;
-extern void __glapi_sparc_icache_flush(unsigned int *);
 static void _glx_mesa_init_sparc_glapi_relocs(void);
 static int _mesa_sparc_needs_init = 1;
-#define INIT_MESA_SPARC { \
-    if(_mesa_sparc_needs_init) { \
+#define INIT_MESA_SPARC do {               \
+   if (_mesa_sparc_needs_init) {           \
       _glx_mesa_init_sparc_glapi_relocs(); \
-      _mesa_sparc_needs_init = 0; \
-  } \
-}
+      _mesa_sparc_needs_init = 0;          \
+   }                                       \
+  } while(0)
 #else
-#define INIT_MESA_SPARC
+#define INIT_MESA_SPARC do { } while(0)
 #endif
 
-#ifdef GLX_DIRECT_RENDERING
-static __DRIscreen *__glXFindDRIScreen(__DRInativeDisplay *dpy, int scrn);
-#endif /* GLX_DIRECT_RENDERING */
-
-static Bool MakeContextCurrent(Display *dpy, GLXDrawable draw,
-    GLXDrawable read, GLXContext gc);
-
-/*
-** We setup some dummy structures here so that the API can be used
-** even if no context is current.
-*/
-
-static GLubyte dummyBuffer[__GLX_BUFFER_LIMIT_SIZE];
-
-/*
-** Dummy context used by small commands when there is no current context.
-** All the
-** gl and glx entry points are designed to operate as nop's when using
-** the dummy context structure.
-*/
-static __GLXcontext dummyContext = {
-    &dummyBuffer[0],
-    &dummyBuffer[0],
-    &dummyBuffer[0],
-    &dummyBuffer[__GLX_BUFFER_LIMIT_SIZE],
-    sizeof(dummyBuffer),
-};
-
-
-/*
-** All indirect rendering contexts will share the same indirect dispatch table.
-*/
-static __GLapi *IndirectAPI = NULL;
-
-
-/*
- * Current context management and locking
- */
-
-#if defined( USE_XTHREADS )
-
-/* thread safe */
-static GLboolean TSDinitialized = GL_FALSE;
-static xthread_key_t ContextTSD;
-
-__GLXcontext *__glXGetCurrentContext(void)
-{
-   if (!TSDinitialized) {
-      xthread_key_create(&ContextTSD, NULL);
-      TSDinitialized = GL_TRUE;
-      return &dummyContext;
-   }
-   else {
-      void *p;
-      xthread_get_specific(ContextTSD, &p);
-      if (!p)
-         return &dummyContext;
-      else
-         return (__GLXcontext *) p;
-   }
-}
-
-void __glXSetCurrentContext(__GLXcontext *c)
-{
-   if (!TSDinitialized) {
-      xthread_key_create(&ContextTSD, NULL);
-      TSDinitialized = GL_TRUE;
-   }
-   xthread_set_specific(ContextTSD, c);
-}
-
-
-/* Used by the __glXLock() and __glXUnlock() macros */
-xmutex_rec __glXmutex;
-
-#elif defined( PTHREADS )
-
-pthread_mutex_t __glXmutex = PTHREAD_MUTEX_INITIALIZER;
-
-# if defined( GLX_USE_TLS )
-
-/**
- * Per-thread GLX context pointer.
- * 
- * \c __glXSetCurrentContext is written is such a way that this pointer can
- * \b never be \c NULL.  This is important!  Because of this
- * \c __glXGetCurrentContext can be implemented as trivial macro.
- */
-__thread void * __glX_tls_Context __attribute__((tls_model("initial-exec")))
-    = &dummyContext;
-
-void __glXSetCurrentContext( __GLXcontext * c )
-{
-    __glX_tls_Context = (c != NULL) ? c : &dummyContext;
-}
-
-# else
-
-static pthread_once_t once_control = PTHREAD_ONCE_INIT;
-
-/**
- * Per-thread data key.
- * 
- * Once \c init_thread_data has been called, the per-thread data key will
- * take a value of \c NULL.  As each new thread is created the default
- * value, in that thread, will be \c NULL.
- */
-static pthread_key_t ContextTSD;
-
-/**
- * Initialize the per-thread data key.
- * 
- * This function is called \b exactly once per-process (not per-thread!) to
- * initialize the per-thread data key.  This is ideally done using the
- * \c pthread_once mechanism.
- */
-static void init_thread_data( void )
-{
-    if ( pthread_key_create( & ContextTSD, NULL ) != 0 ) {
-	perror( "pthread_key_create" );
-	exit( -1 );
-    }
-}
-
-void __glXSetCurrentContext( __GLXcontext * c )
-{
-    pthread_once( & once_control, init_thread_data );
-    pthread_setspecific( ContextTSD, c );
-}
-
-__GLXcontext * __glXGetCurrentContext( void )
-{
-    void * v;
-
-    pthread_once( & once_control, init_thread_data );
-
-    v = pthread_getspecific( ContextTSD );
-    return (v == NULL) ? & dummyContext : (__GLXcontext *) v;
-}
-
-# endif /* defined( GLX_USE_TLS ) */
-
-#elif defined( THREADS )
-
-#error Unknown threading method specified.
-
-#else
-
-/* not thread safe */
-__GLXcontext *__glXcurrentContext = &dummyContext;
-
-#endif
-
-
 /*
 ** You can set this cell to 1 to force the gl drawing stuff to be
 ** one command per packet
 */
-int __glXDebug = 0;
-
-/*
-** forward prototype declarations
-*/
-int __glXCloseDisplay(Display *dpy, XExtCodes *codes);
-
-
-/************************************************************************/
+_X_HIDDEN int __glXDebug = 0;
 
 /* Extension required boiler plate */
 
@@ -281,128 +81,136 @@ static char *__glXExtensionName = GLX_EXTENSION_NAME;
 XExtensionInfo *__glXExtensionInfo = NULL;
 
 static /* const */ char *error_list[] = {
-    "GLXBadContext",
-    "GLXBadContextState",
-    "GLXBadDrawable",
-    "GLXBadPixmap",
-    "GLXBadContextTag",
-    "GLXBadCurrentWindow",
-    "GLXBadRenderRequest",
-    "GLXBadLargeRequest",
-    "GLXUnsupportedPrivateRequest",
-    "GLXBadFBConfig",
-    "GLXBadPbuffer",
-    "GLXBadCurrentDrawable",
-    "GLXBadWindow",
+   "GLXBadContext",
+   "GLXBadContextState",
+   "GLXBadDrawable",
+   "GLXBadPixmap",
+   "GLXBadContextTag",
+   "GLXBadCurrentWindow",
+   "GLXBadRenderRequest",
+   "GLXBadLargeRequest",
+   "GLXUnsupportedPrivateRequest",
+   "GLXBadFBConfig",
+   "GLXBadPbuffer",
+   "GLXBadCurrentDrawable",
+   "GLXBadWindow",
 };
 
-int __glXCloseDisplay(Display *dpy, XExtCodes *codes)
+static int
+__glXCloseDisplay(Display * dpy, XExtCodes * codes)
 {
-  GLXContext gc;
+   GLXContext gc;
 
-  gc = __glXGetCurrentContext();
-  if (dpy == gc->currentDpy) {
-    __glXSetCurrentContext(&dummyContext);
-#ifdef GLX_DIRECT_RENDERING
-    _glapi_set_dispatch(NULL);  /* no-op functions */
-#endif
-    __glXFreeContext(gc);
-  }
+   gc = __glXGetCurrentContext();
+   if (dpy == gc->currentDpy) {
+      __glXSetCurrentContextNull();
+      __glXFreeContext(gc);
+   }
 
-  return XextRemoveDisplay(__glXExtensionInfo, dpy);
+   return XextRemoveDisplay(__glXExtensionInfo, dpy);
 }
 
 
-static XEXT_GENERATE_ERROR_STRING(__glXErrorString, __glXExtensionName,
-				  __GLX_NUMBER_ERRORS, error_list)
-
-static /* const */ XExtensionHooks __glXExtensionHooks = {
-    NULL,				/* create_gc */
-    NULL,				/* copy_gc */
-    NULL,				/* flush_gc */
-    NULL,				/* free_gc */
-    NULL,				/* create_font */
-    NULL,				/* free_font */
-    __glXCloseDisplay,			/* close_display */
-    NULL,				/* wire_to_event */
-    NULL,				/* event_to_wire */
-    NULL,				/* error */
-    __glXErrorString,			/* error_string */
-};
+static
+XEXT_GENERATE_ERROR_STRING(__glXErrorString, __glXExtensionName,
+                           __GLX_NUMBER_ERRORS, error_list)
+
+     static /* const */ XExtensionHooks __glXExtensionHooks = {
+        NULL,                   /* create_gc */
+        NULL,                   /* copy_gc */
+        NULL,                   /* flush_gc */
+        NULL,                   /* free_gc */
+        NULL,                   /* create_font */
+        NULL,                   /* free_font */
+        __glXCloseDisplay,      /* close_display */
+        NULL,                   /* wire_to_event */
+        NULL,                   /* event_to_wire */
+        NULL,                   /* error */
+        __glXErrorString,       /* error_string */
+     };
 
 static
 XEXT_GENERATE_FIND_DISPLAY(__glXFindDisplay, __glXExtensionInfo,
-			   __glXExtensionName, &__glXExtensionHooks,
-			   __GLX_NUMBER_EVENTS, NULL)
+                           __glXExtensionName, &__glXExtensionHooks,
+                           __GLX_NUMBER_EVENTS, NULL)
 
 /************************************************************************/
-
 /*
 ** Free the per screen configs data as well as the array of
 ** __glXScreenConfigs.
 */
-static void FreeScreenConfigs(__GLXdisplayPrivate *priv)
+static void
+FreeScreenConfigs(__GLXdisplayPrivate * priv)
 {
-    __GLXscreenConfigs *psc;
-    GLint i, screens;
-
-    /* Free screen configuration information */
-    psc = priv->screenConfigs;
-    screens = ScreenCount(priv->dpy);
-    for (i = 0; i < screens; i++, psc++) {
-	if (psc->configs) {
-	    _gl_context_modes_destroy( psc->configs );
-	    if(psc->effectiveGLXexts)
-		Xfree(psc->effectiveGLXexts);
-
-	    psc->configs = NULL;	/* NOTE: just for paranoia */
-	}
-	Xfree((char*) psc->serverGLXexts);
+   __GLXscreenConfigs *psc;
+   GLint i, screens;
+
+   /* Free screen configuration information */
+   psc = priv->screenConfigs;
+   screens = ScreenCount(priv->dpy);
+   for (i = 0; i < screens; i++, psc++) {
+      if (psc->configs) {
+         _gl_context_modes_destroy(psc->configs);
+         if (psc->effectiveGLXexts)
+            Xfree(psc->effectiveGLXexts);
+         psc->configs = NULL;   /* NOTE: just for paranoia */
+      }
+      if (psc->visuals) {
+         _gl_context_modes_destroy(psc->visuals);
+         psc->visuals = NULL;   /* NOTE: just for paranoia */
+      }
+      Xfree((char *) psc->serverGLXexts);
 
 #ifdef GLX_DIRECT_RENDERING
-	/* Free the direct rendering per screen data */
-	if (psc->driScreen.private)
-	    (*psc->driScreen.destroyScreen)(priv->dpy, i,
-					    psc->driScreen.private);
-	psc->driScreen.private = NULL;
+      if (psc->driScreen) {
+         psc->driScreen->destroyScreen(psc);
+         __glxHashDestroy(psc->drawHash);
+         XFree(psc->driScreen);
+         psc->driScreen = NULL;
+      }
 #endif
-    }
-    XFree((char*) priv->screenConfigs);
+   }
+   XFree((char *) priv->screenConfigs);
+   priv->screenConfigs = NULL;
 }
 
 /*
 ** Release the private memory referred to in a display private
 ** structure.  The caller will free the extension structure.
 */
-static int __glXFreeDisplayPrivate(XExtData *extension)
+static int
+__glXFreeDisplayPrivate(XExtData * extension)
 {
-    __GLXdisplayPrivate *priv;
-
-    priv = (__GLXdisplayPrivate*) extension->private_data;
-    FreeScreenConfigs(priv);
-    if(priv->serverGLXvendor) {
-	Xfree((char*)priv->serverGLXvendor);
-	priv->serverGLXvendor = 0x0; /* to protect against double free's */
-    }
-    if(priv->serverGLXversion) {
-	Xfree((char*)priv->serverGLXversion);
-	priv->serverGLXversion = 0x0; /* to protect against double free's */
-    }
+   __GLXdisplayPrivate *priv;
+
+   priv = (__GLXdisplayPrivate *) extension->private_data;
+   FreeScreenConfigs(priv);
+   if (priv->serverGLXvendor) {
+      Xfree((char *) priv->serverGLXvendor);
+      priv->serverGLXvendor = 0x0;      /* to protect against double free's */
+   }
+   if (priv->serverGLXversion) {
+      Xfree((char *) priv->serverGLXversion);
+      priv->serverGLXversion = 0x0;     /* to protect against double free's */
+   }
 
 #ifdef GLX_DIRECT_RENDERING
-    /* Free the direct rendering per display data */
-    if (priv->driDisplay.private)
-	(*priv->driDisplay.destroyDisplay)(priv->dpy,
-					   priv->driDisplay.private);
-    priv->driDisplay.private = NULL;
-    if (priv->driDisplay.createNewScreen) {
-        Xfree(priv->driDisplay.createNewScreen); /* free array of ptrs */
-        priv->driDisplay.createNewScreen = NULL;
-    }
+   /* Free the direct rendering per display data */
+   if (priv->driswDisplay)
+      (*priv->driswDisplay->destroyDisplay) (priv->driswDisplay);
+   priv->driswDisplay = NULL;
+
+   if (priv->driDisplay)
+      (*priv->driDisplay->destroyDisplay) (priv->driDisplay);
+   priv->driDisplay = NULL;
+
+   if (priv->dri2Display)
+      (*priv->dri2Display->destroyDisplay) (priv->dri2Display);
+   priv->dri2Display = NULL;
 #endif
 
-    Xfree((char*) priv);
-    return 0;
+   Xfree((char *) priv);
+   return 0;
 }
 
 /************************************************************************/
@@ -411,72 +219,93 @@ static int __glXFreeDisplayPrivate(XExtData *extension)
 ** Query the version of the GLX extension.  This procedure works even if
 ** the client extension is not completely set up.
 */
-static Bool QueryVersion(Display *dpy, int opcode, int *major, int *minor)
+static Bool
+QueryVersion(Display * dpy, int opcode, int *major, int *minor)
 {
-    xGLXQueryVersionReq *req;
-    xGLXQueryVersionReply reply;
-
-    /* Send the glXQueryVersion request */
-    LockDisplay(dpy);
-    GetReq(GLXQueryVersion,req);
-    req->reqType = opcode;
-    req->glxCode = X_GLXQueryVersion;
-    req->majorVersion = GLX_MAJOR_VERSION;
-    req->minorVersion = GLX_MINOR_VERSION;
-    _XReply(dpy, (xReply*) &reply, 0, False);
-    UnlockDisplay(dpy);
-    SyncHandle();
-
-    if (reply.majorVersion != GLX_MAJOR_VERSION) {
-	/*
-	** The server does not support the same major release as this
-	** client.
-	*/
-	return GL_FALSE;
-    }
-    *major = reply.majorVersion;
-    *minor = min(reply.minorVersion, GLX_MINOR_VERSION);
-    return GL_TRUE;
+#ifdef USE_XCB
+   xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_glx_query_version_reply_t* reply =
+      xcb_glx_query_version_reply(c,
+                                  xcb_glx_query_version(c,
+                                                        GLX_MAJOR_VERSION,
+                                                        GLX_MINOR_VERSION),
+                                  NULL);
+
+   if(reply->major_version != GLX_MAJOR_VERSION)
+   {
+      free(reply);
+      return GL_FALSE;
+   }
+   *major = reply->major_version;
+   *minor = min(reply->minor_version, GLX_MINOR_VERSION);
+   free(reply);
+   return GL_TRUE;
+#else
+   xGLXQueryVersionReq *req;
+   xGLXQueryVersionReply reply;
+
+   /* Send the glXQueryVersion request */
+   LockDisplay(dpy);
+   GetReq(GLXQueryVersion, req);
+   req->reqType = opcode;
+   req->glxCode = X_GLXQueryVersion;
+   req->majorVersion = GLX_MAJOR_VERSION;
+   req->minorVersion = GLX_MINOR_VERSION;
+   _XReply(dpy, (xReply *) & reply, 0, False);
+   UnlockDisplay(dpy);
+   SyncHandle();
+
+   if (reply.majorVersion != GLX_MAJOR_VERSION) {
+      /*
+       ** The server does not support the same major release as this
+       ** client.
+       */
+      return GL_FALSE;
+   }
+   *major = reply.majorVersion;
+   *minor = min(reply.minorVersion, GLX_MINOR_VERSION);
+   return GL_TRUE;
+#endif /* USE_XCB */
 }
 
 
-void 
-__glXInitializeVisualConfigFromTags( __GLcontextModes *config, int count, 
-				     const INT32 *bp, Bool tagged_only,
-				     Bool fbconfig_style_tags )
+_X_HIDDEN void
+__glXInitializeVisualConfigFromTags(__GLcontextModes * config, int count,
+                                    const INT32 * bp, Bool tagged_only,
+                                    Bool fbconfig_style_tags)
 {
-    int i;
+   int i;
 
-    if (!tagged_only) {
-	/* Copy in the first set of properties */
-	config->visualID = *bp++;
+   if (!tagged_only) {
+      /* Copy in the first set of properties */
+      config->visualID = *bp++;
 
-	config->visualType = _gl_convert_from_x_visual_type( *bp++ );
+      config->visualType = _gl_convert_from_x_visual_type(*bp++);
 
-	config->rgbMode = *bp++;
+      config->rgbMode = *bp++;
 
-	config->redBits = *bp++;
-	config->greenBits = *bp++;
-	config->blueBits = *bp++;
-	config->alphaBits = *bp++;
-	config->accumRedBits = *bp++;
-	config->accumGreenBits = *bp++;
-	config->accumBlueBits = *bp++;
-	config->accumAlphaBits = *bp++;
+      config->redBits = *bp++;
+      config->greenBits = *bp++;
+      config->blueBits = *bp++;
+      config->alphaBits = *bp++;
+      config->accumRedBits = *bp++;
+      config->accumGreenBits = *bp++;
+      config->accumBlueBits = *bp++;
+      config->accumAlphaBits = *bp++;
 
-	config->doubleBufferMode = *bp++;
-	config->stereoMode = *bp++;
+      config->doubleBufferMode = *bp++;
+      config->stereoMode = *bp++;
 
-	config->rgbBits = *bp++;
-	config->depthBits = *bp++;
-	config->stencilBits = *bp++;
-	config->numAuxBuffers = *bp++;
-	config->level = *bp++;
+      config->rgbBits = *bp++;
+      config->depthBits = *bp++;
+      config->stencilBits = *bp++;
+      config->numAuxBuffers = *bp++;
+      config->level = *bp++;
 
-	count -= __GLX_MIN_CONFIG_PROPS;
-    }
+      count -= __GLX_MIN_CONFIG_PROPS;
+   }
 
-    /*
+   /*
     ** Additional properties may be in a list at the end
     ** of the reply.  They are in pairs of property type
     ** and property value.
@@ -485,858 +314,494 @@ __glXInitializeVisualConfigFromTags( __GLcontextModes *config, int count,
 #define FETCH_OR_SET(tag) \
     config-> tag = ( fbconfig_style_tags ) ? *bp++ : 1
 
-    for (i = 0; i < count; i += 2 ) {
-	switch(*bp++) {
-	  case GLX_RGBA:
-	    FETCH_OR_SET( rgbMode );
-	    break;
-	  case GLX_BUFFER_SIZE:
-	    config->rgbBits = *bp++;
-	    break;
-	  case GLX_LEVEL:
-	    config->level = *bp++;
-	    break;
-	  case GLX_DOUBLEBUFFER:
-	    FETCH_OR_SET( doubleBufferMode );
-	    break;
-	  case GLX_STEREO:
-	    FETCH_OR_SET( stereoMode );
-	    break;
-	  case GLX_AUX_BUFFERS:
-	    config->numAuxBuffers = *bp++;
-	    break;
-	  case GLX_RED_SIZE:
-	    config->redBits = *bp++;
-	    break;
-	  case GLX_GREEN_SIZE:
-	    config->greenBits = *bp++;
-	    break;
-	  case GLX_BLUE_SIZE:
-	    config->blueBits = *bp++;
-	    break;
-	  case GLX_ALPHA_SIZE:
-	    config->alphaBits = *bp++;
-	    break;
-	  case GLX_DEPTH_SIZE:
-	    config->depthBits = *bp++;
-	    break;
-	  case GLX_STENCIL_SIZE:
-	    config->stencilBits = *bp++;
-	    break;
-	  case GLX_ACCUM_RED_SIZE:
-	    config->accumRedBits = *bp++;
-	    break;
-	  case GLX_ACCUM_GREEN_SIZE:
-	    config->accumGreenBits = *bp++;
-	    break;
-	  case GLX_ACCUM_BLUE_SIZE:
-	    config->accumBlueBits = *bp++;
-	    break;
-	  case GLX_ACCUM_ALPHA_SIZE:
-	    config->accumAlphaBits = *bp++;
-	    break;
-	  case GLX_VISUAL_CAVEAT_EXT:
-	    config->visualRating = *bp++;    
-	    break;
-	  case GLX_X_VISUAL_TYPE:
-	    config->visualType = *bp++;
-	    break;
-	  case GLX_TRANSPARENT_TYPE:
-	    config->transparentPixel = *bp++;    
-	    break;
-	  case GLX_TRANSPARENT_INDEX_VALUE:
-	    config->transparentIndex = *bp++;    
-	    break;
-	  case GLX_TRANSPARENT_RED_VALUE:
-	    config->transparentRed = *bp++;    
-	    break;
-	  case GLX_TRANSPARENT_GREEN_VALUE:
-	    config->transparentGreen = *bp++;    
-	    break;
-	  case GLX_TRANSPARENT_BLUE_VALUE:
-	    config->transparentBlue = *bp++;    
-	    break;
-	  case GLX_TRANSPARENT_ALPHA_VALUE:
-	    config->transparentAlpha = *bp++;    
-	    break;
-	  case GLX_VISUAL_ID:
-	    config->visualID = *bp++;
-	    break;
-	  case GLX_DRAWABLE_TYPE:
-	    config->drawableType = *bp++;
-	    break;
-	  case GLX_RENDER_TYPE:
-	    config->renderType = *bp++;
-	    break;
-	  case GLX_X_RENDERABLE:
-	    config->xRenderable = *bp++;
-	    break;
-	  case GLX_FBCONFIG_ID:
-	    config->fbconfigID = *bp++;
-	    break;
-	  case GLX_MAX_PBUFFER_WIDTH:
-	    config->maxPbufferWidth = *bp++;
-	    break;
-	  case GLX_MAX_PBUFFER_HEIGHT:
-	    config->maxPbufferHeight = *bp++;
-	    break;
-	  case GLX_MAX_PBUFFER_PIXELS:
-	    config->maxPbufferPixels = *bp++;
-	    break;
-	  case GLX_OPTIMAL_PBUFFER_WIDTH_SGIX:
-	    config->optimalPbufferWidth = *bp++;
-	    break;
-	  case GLX_OPTIMAL_PBUFFER_HEIGHT_SGIX:
-	    config->optimalPbufferHeight = *bp++;
-	    break;
-	  case GLX_VISUAL_SELECT_GROUP_SGIX:
-	    config->visualSelectGroup = *bp++;
-	    break;
-	  case GLX_SWAP_METHOD_OML:
-	    config->swapMethod = *bp++;
-	    break;
-	  case GLX_SAMPLE_BUFFERS_SGIS:
-	    config->sampleBuffers = *bp++;
-	    break;
-	  case GLX_SAMPLES_SGIS:
-	    config->samples = *bp++;
-	    break;
-	case GLX_BIND_TO_TEXTURE_RGB_EXT:
-	    config->bindToTextureRgb = *bp++;
-	    break;
-	case GLX_BIND_TO_TEXTURE_RGBA_EXT:
-	    config->bindToTextureRgba = *bp++;
-	    break;
-	case GLX_BIND_TO_MIPMAP_TEXTURE_EXT:
-	    config->bindToMipmapTexture = *bp++;
-	    break;
-	case GLX_BIND_TO_TEXTURE_TARGETS_EXT:
-	    config->bindToTextureTargets = *bp++;
-	    break;
-	case GLX_Y_INVERTED_EXT:
-	    config->yInverted = *bp++;
-	    break;
-	  case None:
-	    i = count;
-	    break;
-	  default:
-	    break;
-	}
-    }
-
-    config->renderType = (config->rgbMode) ? GLX_RGBA_BIT : GLX_COLOR_INDEX_BIT;
-
-    config->haveAccumBuffer = ((config->accumRedBits +
-			       config->accumGreenBits +
-			       config->accumBlueBits +
-			       config->accumAlphaBits) > 0);
-    config->haveDepthBuffer = (config->depthBits > 0);
-    config->haveStencilBuffer = (config->stencilBits > 0);
-}
+   for (i = 0; i < count; i += 2) {
+      switch (*bp++) {
+      case GLX_RGBA:
+         FETCH_OR_SET(rgbMode);
+         break;
+      case GLX_BUFFER_SIZE:
+         config->rgbBits = *bp++;
+         break;
+      case GLX_LEVEL:
+         config->level = *bp++;
+         break;
+      case GLX_DOUBLEBUFFER:
+         FETCH_OR_SET(doubleBufferMode);
+         break;
+      case GLX_STEREO:
+         FETCH_OR_SET(stereoMode);
+         break;
+      case GLX_AUX_BUFFERS:
+         config->numAuxBuffers = *bp++;
+         break;
+      case GLX_RED_SIZE:
+         config->redBits = *bp++;
+         break;
+      case GLX_GREEN_SIZE:
+         config->greenBits = *bp++;
+         break;
+      case GLX_BLUE_SIZE:
+         config->blueBits = *bp++;
+         break;
+      case GLX_ALPHA_SIZE:
+         config->alphaBits = *bp++;
+         break;
+      case GLX_DEPTH_SIZE:
+         config->depthBits = *bp++;
+         break;
+      case GLX_STENCIL_SIZE:
+         config->stencilBits = *bp++;
+         break;
+      case GLX_ACCUM_RED_SIZE:
+         config->accumRedBits = *bp++;
+         break;
+      case GLX_ACCUM_GREEN_SIZE:
+         config->accumGreenBits = *bp++;
+         break;
+      case GLX_ACCUM_BLUE_SIZE:
+         config->accumBlueBits = *bp++;
+         break;
+      case GLX_ACCUM_ALPHA_SIZE:
+         config->accumAlphaBits = *bp++;
+         break;
+      case GLX_VISUAL_CAVEAT_EXT:
+         config->visualRating = *bp++;
+         break;
+      case GLX_X_VISUAL_TYPE:
+         config->visualType = *bp++;
+         break;
+      case GLX_TRANSPARENT_TYPE:
+         config->transparentPixel = *bp++;
+         break;
+      case GLX_TRANSPARENT_INDEX_VALUE:
+         config->transparentIndex = *bp++;
+         break;
+      case GLX_TRANSPARENT_RED_VALUE:
+         config->transparentRed = *bp++;
+         break;
+      case GLX_TRANSPARENT_GREEN_VALUE:
+         config->transparentGreen = *bp++;
+         break;
+      case GLX_TRANSPARENT_BLUE_VALUE:
+         config->transparentBlue = *bp++;
+         break;
+      case GLX_TRANSPARENT_ALPHA_VALUE:
+         config->transparentAlpha = *bp++;
+         break;
+      case GLX_VISUAL_ID:
+         config->visualID = *bp++;
+         break;
+      case GLX_DRAWABLE_TYPE:
+         config->drawableType = *bp++;
+         break;
+      case GLX_RENDER_TYPE:
+         config->renderType = *bp++;
+         break;
+      case GLX_X_RENDERABLE:
+         config->xRenderable = *bp++;
+         break;
+      case GLX_FBCONFIG_ID:
+         config->fbconfigID = *bp++;
+         break;
+      case GLX_MAX_PBUFFER_WIDTH:
+         config->maxPbufferWidth = *bp++;
+         break;
+      case GLX_MAX_PBUFFER_HEIGHT:
+         config->maxPbufferHeight = *bp++;
+         break;
+      case GLX_MAX_PBUFFER_PIXELS:
+         config->maxPbufferPixels = *bp++;
+         break;
+      case GLX_OPTIMAL_PBUFFER_WIDTH_SGIX:
+         config->optimalPbufferWidth = *bp++;
+         break;
+      case GLX_OPTIMAL_PBUFFER_HEIGHT_SGIX:
+         config->optimalPbufferHeight = *bp++;
+         break;
+      case GLX_VISUAL_SELECT_GROUP_SGIX:
+         config->visualSelectGroup = *bp++;
+         break;
+      case GLX_SWAP_METHOD_OML:
+         config->swapMethod = *bp++;
+         break;
+      case GLX_SAMPLE_BUFFERS_SGIS:
+         config->sampleBuffers = *bp++;
+         break;
+      case GLX_SAMPLES_SGIS:
+         config->samples = *bp++;
+         break;
+      case GLX_BIND_TO_TEXTURE_RGB_EXT:
+         config->bindToTextureRgb = *bp++;
+         break;
+      case GLX_BIND_TO_TEXTURE_RGBA_EXT:
+         config->bindToTextureRgba = *bp++;
+         break;
+      case GLX_BIND_TO_MIPMAP_TEXTURE_EXT:
+         config->bindToMipmapTexture = *bp++;
+         break;
+      case GLX_BIND_TO_TEXTURE_TARGETS_EXT:
+         config->bindToTextureTargets = *bp++;
+         break;
+      case GLX_Y_INVERTED_EXT:
+         config->yInverted = *bp++;
+         break;
+      case None:
+         i = count;
+         break;
+      default:
+         break;
+      }
+   }
 
+   config->renderType =
+      (config->rgbMode) ? GLX_RGBA_BIT : GLX_COLOR_INDEX_BIT;
 
-#ifdef GLX_DIRECT_RENDERING
-static unsigned
-filter_modes( __GLcontextModes ** server_modes,
-	      const __GLcontextModes * driver_modes )
-{
-    __GLcontextModes * m;
-    __GLcontextModes ** prev_next;
-    const __GLcontextModes * check;
-    unsigned modes_count = 0;
-
-    if ( driver_modes == NULL ) {
-	fprintf(stderr, "libGL warning: 3D driver returned no fbconfigs.\n");
-	return 0;
-    }
-
-    /* For each mode in server_modes, check to see if a matching mode exists
-     * in driver_modes.  If not, then the mode is not available.
-     */
-
-    prev_next = server_modes;
-    for ( m = *prev_next ; m != NULL ; m = *prev_next ) {
-	GLboolean do_delete = GL_TRUE;
-
-	for ( check = driver_modes ; check != NULL ; check = check->next ) {
-	    if ( _gl_context_modes_are_same( m, check ) ) {
-		do_delete = GL_FALSE;
-		break;
-	    }
-	}
-
-	/* The 3D has to support all the modes that match the GLX visuals
-	 * sent from the X server.
-	 */
-	if ( do_delete && (m->visualID != 0) ) {
-	    do_delete = GL_FALSE;
-
-            if (getenv("LIBGL_DEBUG")) {
-               fprintf(stderr, "libGL warning: 3D driver claims to not support "
-                       "visual 0x%02x\n", m->visualID);
-            }
-	}
-
-	if ( do_delete ) {
-	    *prev_next = m->next;
-
-	    m->next = NULL;
-	    _gl_context_modes_destroy( m );
-	}
-	else {
-	    modes_count++;
-	    prev_next = & m->next;
-	}
-    }
-
-    return modes_count;
+   config->haveAccumBuffer = ((config->accumRedBits +
+                               config->accumGreenBits +
+                               config->accumBlueBits +
+                               config->accumAlphaBits) > 0);
+   config->haveDepthBuffer = (config->depthBits > 0);
+   config->haveStencilBuffer = (config->stencilBits > 0);
 }
 
-
-/**
- * Implement \c __DRIinterfaceMethods::getProcAddress.
- */
-static __DRIfuncPtr get_proc_address( const char * proc_name )
+static __GLcontextModes *
+createConfigsFromProperties(Display * dpy, int nvisuals, int nprops,
+                            int screen, GLboolean tagged_only)
 {
-    if (strcmp( proc_name, "glxEnableExtension" ) == 0) {
-	return (__DRIfuncPtr) __glXScrEnableExtension;
-    }
-    
-    return NULL;
-}
+   INT32 buf[__GLX_TOTAL_CONFIG], *props;
+   unsigned prop_size;
+   __GLcontextModes *modes, *m;
+   int i;
+
+   if (nprops == 0)
+      return NULL;
+
+   /* FIXME: Is the __GLX_MIN_CONFIG_PROPS test correct for FBconfigs? */
+
+   /* Check number of properties */
+   if (nprops < __GLX_MIN_CONFIG_PROPS || nprops > __GLX_MAX_CONFIG_PROPS)
+      return NULL;
+
+   /* Allocate memory for our config structure */
+   modes = _gl_context_modes_create(nvisuals, sizeof(__GLcontextModes));
+   if (!modes)
+      return NULL;
+
+   prop_size = nprops * __GLX_SIZE_INT32;
+   if (prop_size <= sizeof(buf))
+      props = buf;
+   else
+      props = Xmalloc(prop_size);
+
+   /* Read each config structure and convert it into our format */
+   m = modes;
+   for (i = 0; i < nvisuals; i++) {
+      _XRead(dpy, (char *) props, prop_size);
+      /* Older X servers don't send this so we default it here. */
+      m->drawableType = GLX_WINDOW_BIT;
+      __glXInitializeVisualConfigFromTags(m, nprops, props,
+                                          tagged_only, GL_TRUE);
+      m->screen = screen;
+      m = m->next;
+   }
 
-#ifdef XDAMAGE_1_1_INTERFACE
-static GLboolean has_damage_post(__DRInativeDisplay *dpy)
-{
-    static GLboolean inited = GL_FALSE;
-    static GLboolean has_damage;
-
-    if (!inited) {
-	int major, minor;
-
-	if (XDamageQueryVersion(dpy, &major, &minor) &&
-	    major == 1 && minor >= 1)
-	{
-	    has_damage = GL_TRUE;
-	} else {
-	    has_damage = GL_FALSE;
-	}
-	inited = GL_TRUE;
-    }
-
-    return has_damage;
+   if (props != buf)
+      Xfree(props);
+
+   return modes;
 }
-#endif /* XDAMAGE_1_1_INTERFACE */
 
-static void __glXReportDamage(__DRInativeDisplay *dpy, int screen,
-			      __DRIid drawable,
-			      int x, int y,
-			      drm_clip_rect_t *rects, int num_rects,
-			      GLboolean front_buffer)
+static GLboolean
+getVisualConfigs(Display * dpy, __GLXdisplayPrivate * priv, int screen)
 {
-#ifdef XDAMAGE_1_1_INTERFACE
-    XRectangle *xrects;
-    XserverRegion region;
-    int i;
-    int x_off, y_off;
-
-    if (!has_damage_post(dpy))
-	return;
-
-    if (front_buffer) {
-	x_off = x;
-	y_off = y;
-	drawable = RootWindow(dpy, screen);
-    } else{
-	x_off = 0;
-	y_off = 0;
-    }
-
-    xrects = malloc(sizeof(XRectangle) * num_rects);
-    if (xrects == NULL)
-	return;
-
-    for (i = 0; i < num_rects; i++) {
-	xrects[i].x = rects[i].x1 + x_off;
-	xrects[i].y = rects[i].y1 + y_off;
-	xrects[i].width = rects[i].x2 - rects[i].x1;
-	xrects[i].height = rects[i].y2 - rects[i].y1;
-    }
-    region = XFixesCreateRegion(dpy, xrects, num_rects);
-    free(xrects);
-    XDamageAdd(dpy, drawable, region);
-    XFixesDestroyRegion(dpy, region);
-#endif
+   xGLXGetVisualConfigsReq *req;
+   __GLXscreenConfigs *psc;
+   xGLXGetVisualConfigsReply reply;
+
+   LockDisplay(dpy);
+
+   psc = priv->screenConfigs + screen;
+   psc->visuals = NULL;
+   GetReq(GLXGetVisualConfigs, req);
+   req->reqType = priv->majorOpcode;
+   req->glxCode = X_GLXGetVisualConfigs;
+   req->screen = screen;
+
+   if (!_XReply(dpy, (xReply *) & reply, 0, False))
+      goto out;
+
+   psc->visuals = createConfigsFromProperties(dpy,
+                                              reply.numVisuals,
+                                              reply.numProps,
+                                              screen, GL_FALSE);
+
+ out:
+   UnlockDisplay(dpy);
+   return psc->visuals != NULL;
 }
 
-/**
- * Table of functions exported by the loader to the driver.
- */
-static const __DRIinterfaceMethods interface_methods = {
-    get_proc_address,
-
-    _gl_context_modes_create,
-    _gl_context_modes_destroy,
-      
-    __glXFindDRIScreen,
-    __glXWindowExists,
-      
-    XF86DRICreateContextWithConfig,
-    XF86DRIDestroyContext,
-
-    XF86DRICreateDrawable,
-    XF86DRIDestroyDrawable,
-    XF86DRIGetDrawableInfo,
-
-    __glXGetUST,
-    __glXGetMscRateOML,
-
-    __glXReportDamage,
-};
+static GLboolean
+getFBConfigs(Display * dpy, __GLXdisplayPrivate * priv, int screen)
+{
+   xGLXGetFBConfigsReq *fb_req;
+   xGLXGetFBConfigsSGIXReq *sgi_req;
+   xGLXVendorPrivateWithReplyReq *vpreq;
+   xGLXGetFBConfigsReply reply;
+   __GLXscreenConfigs *psc;
+
+   psc = priv->screenConfigs + screen;
+   psc->serverGLXexts = __glXQueryServerString(dpy, priv->majorOpcode, screen, GLX_EXTENSIONS);
+
+   LockDisplay(dpy);
+
+   psc->configs = NULL;
+   if (atof(priv->serverGLXversion) >= 1.3) {
+      GetReq(GLXGetFBConfigs, fb_req);
+      fb_req->reqType = priv->majorOpcode;
+      fb_req->glxCode = X_GLXGetFBConfigs;
+      fb_req->screen = screen;
+   }
+   else if (strstr(psc->serverGLXexts, "GLX_SGIX_fbconfig") != NULL) {
+      GetReqExtra(GLXVendorPrivateWithReply,
+                  sz_xGLXGetFBConfigsSGIXReq +
+                  sz_xGLXVendorPrivateWithReplyReq, vpreq);
+      sgi_req = (xGLXGetFBConfigsSGIXReq *) vpreq;
+      sgi_req->reqType = priv->majorOpcode;
+      sgi_req->glxCode = X_GLXVendorPrivateWithReply;
+      sgi_req->vendorCode = X_GLXvop_GetFBConfigsSGIX;
+      sgi_req->screen = screen;
+   }
+   else
+      goto out;
 
+   if (!_XReply(dpy, (xReply *) & reply, 0, False))
+      goto out;
 
+   psc->configs = createConfigsFromProperties(dpy,
+                                              reply.numFBConfigs,
+                                              reply.numAttribs * 2,
+                                              screen, GL_TRUE);
 
-/**
- * Perform the required libGL-side initialization and call the client-side
- * driver's \c __driCreateNewScreen function.
- * 
- * \param dpy    Display pointer.
- * \param scrn   Screen number on the display.
- * \param psc    DRI screen information.
- * \param driDpy DRI display information.
- * \param createNewScreen  Pointer to the client-side driver's
- *               \c __driCreateNewScreen function.
- * \returns A pointer to the \c __DRIscreenPrivate structure returned by
- *          the client-side driver on success, or \c NULL on failure.
- * 
- * \todo This function needs to be modified to remove context-modes from the
- *       list stored in the \c __GLXscreenConfigsRec to match the list
- *       returned by the client-side driver.
- */
-static void *
-CallCreateNewScreen(Display *dpy, int scrn, __DRIscreen *psc,
-		    __DRIdisplay * driDpy,
-		    PFNCREATENEWSCREENFUNC createNewScreen)
-{
-    __DRIscreenPrivate *psp = NULL;
-#ifndef GLX_USE_APPLEGL
-    drm_handle_t hSAREA;
-    drmAddress pSAREA = MAP_FAILED;
-    char *BusID;
-    __DRIversion   ddx_version;
-    __DRIversion   dri_version;
-    __DRIversion   drm_version;
-    __DRIframebuffer  framebuffer;
-    int   fd = -1;
-    int   status;
-    const char * err_msg;
-    const char * err_extra;
-    int api_ver = __glXGetInternalVersion();
-
-
-    dri_version.major = driDpy->private->driMajor;
-    dri_version.minor = driDpy->private->driMinor;
-    dri_version.patch = driDpy->private->driPatch;
-
-
-    err_msg = "XF86DRIOpenConnection";
-    err_extra = NULL;
-
-    framebuffer.base = MAP_FAILED;
-    framebuffer.dev_priv = NULL;
-
-    if (XF86DRIOpenConnection(dpy, scrn, &hSAREA, &BusID)) {
-        int newlyopened;
-	fd = drmOpenOnce(NULL,BusID, &newlyopened);
-	Xfree(BusID); /* No longer needed */
-
-	err_msg = "open DRM";
-	err_extra = strerror( -fd );
-
-	if (fd >= 0) {
-	    drm_magic_t magic;
-
-	    err_msg = "drmGetMagic";
-	    err_extra = NULL;
-
-	    if (!drmGetMagic(fd, &magic)) {
-		drmVersionPtr version = drmGetVersion(fd);
-		if (version) {
-		    drm_version.major = version->version_major;
-		    drm_version.minor = version->version_minor;
-		    drm_version.patch = version->version_patchlevel;
-		    drmFreeVersion(version);
-		}
-		else {
-		    drm_version.major = -1;
-		    drm_version.minor = -1;
-		    drm_version.patch = -1;
-		}
-
-		err_msg = "XF86DRIAuthConnection";
-		if (!newlyopened || XF86DRIAuthConnection(dpy, scrn, magic)) {
-		    char *driverName;
-
-		    /*
-		     * Get device name (like "tdfx") and the ddx version
-		     * numbers.  We'll check the version in each DRI driver's
-		     * "createNewScreen" function.
-		     */
-		    err_msg = "XF86DRIGetClientDriverName";
-		    if (XF86DRIGetClientDriverName(dpy, scrn,
-						   &ddx_version.major,
-						   &ddx_version.minor,
-						   &ddx_version.patch,
-						   &driverName)) {
-			drm_handle_t  hFB;
-			int        junk;
-
-			/* No longer needed. */
-			Xfree( driverName );
-
-
-			/*
-			 * Get device-specific info.  pDevPriv will point to a struct
-			 * (such as DRIRADEONRec in xfree86/driver/ati/radeon_dri.h)
-			 * that has information about the screen size, depth, pitch,
-			 * ancilliary buffers, DRM mmap handles, etc.
-			 */
-			err_msg = "XF86DRIGetDeviceInfo";
-			if (XF86DRIGetDeviceInfo(dpy, scrn,
-						 &hFB,
-						 &junk,
-						 &framebuffer.size,
-						 &framebuffer.stride,
-						 &framebuffer.dev_priv_size,
-						 &framebuffer.dev_priv)) {
-			    framebuffer.width = DisplayWidth(dpy, scrn);
-			    framebuffer.height = DisplayHeight(dpy, scrn);
-
-			    /*
-			     * Map the framebuffer region.
-			     */
-			    status = drmMap(fd, hFB, framebuffer.size, 
-					    (drmAddressPtr)&framebuffer.base);
-
-			    err_msg = "drmMap of framebuffer";
-			    err_extra = strerror( -status );
-
-			    if ( status == 0 ) {
-				/*
-				 * Map the SAREA region.  Further mmap regions
-				 * may be setup in each DRI driver's
-				 * "createNewScreen" function.
-				 */
-				status = drmMap(fd, hSAREA, SAREA_MAX, 
-						&pSAREA);
-
-				err_msg = "drmMap of sarea";
-				err_extra = strerror( -status );
-
-				if ( status == 0 ) {
-				    __GLcontextModes * driver_modes = NULL;
-				    __GLXscreenConfigs *configs = psc->screenConfigs;
-
-				    err_msg = "InitDriver";
-				    err_extra = NULL;
-				    psp = (*createNewScreen)(dpy, scrn,
-							     psc,
-							     configs->configs,
-							     & ddx_version,
-							     & dri_version,
-							     & drm_version,
-							     & framebuffer,
-							     pSAREA,
-							     fd,
-							     api_ver,
-							     & interface_methods,
-							     & driver_modes );
-
-				    filter_modes( & configs->configs,
-						  driver_modes );
-				    _gl_context_modes_destroy( driver_modes );
-				}
-			    }
-			}
-		    }
-		}
-	    }
-	}
-    }
-
-    if ( psp == NULL ) {
-	if ( pSAREA != MAP_FAILED ) {
-	    (void)drmUnmap(pSAREA, SAREA_MAX);
-	}
-
-	if ( framebuffer.base != MAP_FAILED ) {
-	    (void)drmUnmap((drmAddress)framebuffer.base, framebuffer.size);
-	}
-
-	if ( framebuffer.dev_priv != NULL ) {
-	    Xfree(framebuffer.dev_priv);
-	}
-
-	if ( fd >= 0 ) {
-	    (void)drmCloseOnce(fd);
-	}
-
-	(void)XF86DRICloseConnection(dpy, scrn);
-
-	if ( err_extra != NULL ) {
-	    fprintf(stderr, "libGL error: %s failed (%s)\n", err_msg,
-		    err_extra);
-	}
-	else {
-	    fprintf(stderr, "libGL error: %s failed\n", err_msg );
-	}
-
-        fprintf(stderr, "libGL error: reverting to (slow) indirect rendering\n");
-    }
-#endif /* !GLX_USE_APPLEGL */
-
-    return psp;
+ out:
+   UnlockDisplay(dpy);
+   return psc->configs != NULL;
 }
-#endif /* GLX_DIRECT_RENDERING */
-
 
 /*
 ** Allocate the memory for the per screen configs for each screen.
 ** If that works then fetch the per screen configs data.
 */
-static Bool AllocAndFetchScreenConfigs(Display *dpy, __GLXdisplayPrivate *priv)
+static Bool
+AllocAndFetchScreenConfigs(Display * dpy, __GLXdisplayPrivate * priv)
 {
-    xGLXGetVisualConfigsReq *req;
-    xGLXGetFBConfigsReq *fb_req;
-    xGLXVendorPrivateWithReplyReq *vpreq;
-    xGLXGetFBConfigsSGIXReq *sgi_req;
-    xGLXGetVisualConfigsReply reply;
-    __GLXscreenConfigs *psc;
-    __GLcontextModes *config;
-    GLint i, j, nprops, screens;
-    INT32 buf[__GLX_TOTAL_CONFIG], *props;
-    unsigned supported_request = 0;
-    unsigned prop_size;
-
-    /*
+   __GLXscreenConfigs *psc;
+   GLint i, screens;
+
+   /*
     ** First allocate memory for the array of per screen configs.
     */
-    screens = ScreenCount(dpy);
-    psc = (__GLXscreenConfigs*) Xmalloc(screens * sizeof(__GLXscreenConfigs));
-    if (!psc) {
-	return GL_FALSE;
-    }
-    memset(psc, 0, screens * sizeof(__GLXscreenConfigs));
-    priv->screenConfigs = psc;
-    
-    priv->serverGLXversion = __glXGetStringFromServer(dpy, priv->majorOpcode,
-					 X_GLXQueryServerString,
-					 0, GLX_VERSION);
-    if ( priv->serverGLXversion == NULL ) {
-	FreeScreenConfigs(priv);
-	return GL_FALSE;
-    }
-
-    if ( atof( priv->serverGLXversion ) >= 1.3 ) {
-	supported_request = 1;
-    }
-
-    /*
-    ** Now fetch each screens configs structures.  If a screen supports
-    ** GL (by returning a numVisuals > 0) then allocate memory for our
-    ** config structure and then fill it in.
-    */
-    for (i = 0; i < screens; i++, psc++) {
-	if ( supported_request != 1 ) {
-	    psc->serverGLXexts = __glXGetStringFromServer(dpy, priv->majorOpcode,
-							  X_GLXQueryServerString,
-							  i, GLX_EXTENSIONS);
-	    if ( strstr( psc->serverGLXexts, "GLX_SGIX_fbconfig" ) != NULL ) {
-		supported_request = 2;
-	    }
-	    else {
-		supported_request = 3;
-	    }
-	}
-
-
-	LockDisplay(dpy);
-	switch( supported_request ) {
-	    case 1:
-	    GetReq(GLXGetFBConfigs,fb_req);
-	    fb_req->reqType = priv->majorOpcode;
-	    fb_req->glxCode = X_GLXGetFBConfigs;
-	    fb_req->screen = i;
-	    break;
-	   
-	    case 2:
-	    GetReqExtra(GLXVendorPrivateWithReply,
-			sz_xGLXGetFBConfigsSGIXReq-sz_xGLXVendorPrivateWithReplyReq,vpreq);
-	    sgi_req = (xGLXGetFBConfigsSGIXReq *) vpreq;
-	    sgi_req->reqType = priv->majorOpcode;
-	    sgi_req->glxCode = X_GLXVendorPrivateWithReply;
-	    sgi_req->vendorCode = X_GLXvop_GetFBConfigsSGIX;
-	    sgi_req->screen = i;
-	    break;
-
-	    case 3:
-	    GetReq(GLXGetVisualConfigs,req);
-	    req->reqType = priv->majorOpcode;
-	    req->glxCode = X_GLXGetVisualConfigs;
-	    req->screen = i;
-	    break;
- 	}
-
-	if (!_XReply(dpy, (xReply*) &reply, 0, False)) {
-	    /* Something is busted. Punt. */
-	    UnlockDisplay(dpy);
-	    SyncHandle();
-	    FreeScreenConfigs(priv);
-	    return GL_FALSE;
-	}
-
-	if (!reply.numVisuals) {
-	    /* This screen does not support GL rendering */
-	    UnlockDisplay(dpy);
-	    continue;
-	}
-
-	/* FIXME: Is the __GLX_MIN_CONFIG_PROPS test correct for
-	 * FIXME: FBconfigs? 
-	 */
-	/* Check number of properties */
-	nprops = reply.numProps;
-	if ((nprops < __GLX_MIN_CONFIG_PROPS) ||
-	    (nprops > __GLX_MAX_CONFIG_PROPS)) {
-	    /* Huh?  Not in protocol defined limits.  Punt */
-	    UnlockDisplay(dpy);
-	    SyncHandle();
-	    FreeScreenConfigs(priv);
-	    return GL_FALSE;
-	}
-
-	/* Allocate memory for our config structure */
-	psc->configs = _gl_context_modes_create(reply.numVisuals,
-						sizeof(__GLcontextModes));
-	if (!psc->configs) {
-	    UnlockDisplay(dpy);
-	    SyncHandle();
-	    FreeScreenConfigs(priv);
-	    return GL_FALSE;
-	}
-
-	/* Allocate memory for the properties, if needed */
-	if ( supported_request != 3 ) {
-	    nprops *= 2;
-	}
-
-	prop_size = nprops * __GLX_SIZE_INT32;
-
-	if (prop_size <= sizeof(buf)) {
- 	    props = buf;
- 	} else {
-	    props = (INT32 *) Xmalloc(prop_size);
- 	} 
-
-	/* Read each config structure and convert it into our format */
-        config = psc->configs;
-	for (j = 0; j < reply.numVisuals; j++) {
-	    assert( config != NULL );
-	    _XRead(dpy, (char *)props, prop_size);
-
-	    if ( supported_request != 3 ) {
-		config->rgbMode = GL_TRUE;
-		config->drawableType = GLX_WINDOW_BIT;
-	    }
-	    else {
-		config->drawableType = GLX_WINDOW_BIT | GLX_PIXMAP_BIT;
-	    }
-
-	    __glXInitializeVisualConfigFromTags( config, nprops, props,
-						 (supported_request != 3),
-						 GL_TRUE );
-	    if ( config->fbconfigID == GLX_DONT_CARE ) {
-		config->fbconfigID = config->visualID;
-	    }
-	    config->screen = i;
-	    config = config->next;
-	}
-	if (props != buf) {
-	    Xfree((char *)props);
-	}
-	UnlockDisplay(dpy);
+   screens = ScreenCount(dpy);
+   psc = (__GLXscreenConfigs *) Xmalloc(screens * sizeof(__GLXscreenConfigs));
+   if (!psc) {
+      return GL_FALSE;
+   }
+   memset(psc, 0, screens * sizeof(__GLXscreenConfigs));
+   priv->screenConfigs = psc;
+
+   priv->serverGLXversion = __glXQueryServerString(dpy, priv->majorOpcode, 0, GLX_VERSION);
+   if (priv->serverGLXversion == NULL) {
+      FreeScreenConfigs(priv);
+      return GL_FALSE;
+   }
+
+   for (i = 0; i < screens; i++, psc++) {
+      getVisualConfigs(dpy, priv, i);
+      getFBConfigs(dpy, priv, i);
 
 #ifdef GLX_DIRECT_RENDERING
-        /* Initialize per screen dynamic client GLX extensions */
-	psc->ext_list_first_time = GL_TRUE;
-	/* Initialize the direct rendering per screen data and functions */
-	if (priv->driDisplay.private != NULL) {
-	    /* FIXME: Should it be some sort of an error if createNewScreen[i]
-	     * FIXME: is NULL?
-	     */
-	    if (priv->driDisplay.createNewScreen &&
-		priv->driDisplay.createNewScreen[i]) {
-
-		psc->driScreen.screenConfigs = (void *)psc;
-		psc->driScreen.private =
-		    CallCreateNewScreen(dpy, i, & psc->driScreen,
-					& priv->driDisplay,
-					priv->driDisplay.createNewScreen[i] );
-	    }
-	}
+      psc->scr = i;
+      psc->dpy = dpy;
+      psc->drawHash = __glxHashCreate();
+      if (psc->drawHash == NULL)
+         continue;
+
+      if (priv->dri2Display)
+         psc->driScreen = (*priv->dri2Display->createScreen) (psc, i, priv);
+
+      if (psc->driScreen == NULL && priv->driDisplay)
+         psc->driScreen = (*priv->driDisplay->createScreen) (psc, i, priv);
+
+      if (psc->driScreen == NULL && priv->driswDisplay)
+         psc->driScreen = (*priv->driswDisplay->createScreen) (psc, i, priv);
+
+      if (psc->driScreen == NULL) {
+         __glxHashDestroy(psc->drawHash);
+         psc->drawHash = NULL;
+      }
 #endif
-    }
-    SyncHandle();
-    return GL_TRUE;
+   }
+   SyncHandle();
+   return GL_TRUE;
 }
 
 /*
 ** Initialize the client side extension code.
 */
-__GLXdisplayPrivate *__glXInitialize(Display* dpy)
+_X_HIDDEN __GLXdisplayPrivate *
+__glXInitialize(Display * dpy)
 {
-    XExtDisplayInfo *info = __glXFindDisplay(dpy);
-    XExtData **privList, *private, *found;
-    __GLXdisplayPrivate *dpyPriv;
-    XEDataObject dataObj;
-    int major, minor;
+   XExtDisplayInfo *info = __glXFindDisplay(dpy);
+   XExtData **privList, *private, *found;
+   __GLXdisplayPrivate *dpyPriv;
+   XEDataObject dataObj;
+   int major, minor;
+#ifdef GLX_DIRECT_RENDERING
+   Bool glx_direct, glx_accel;
+#endif
 
 #if defined(USE_XTHREADS)
-    {
-        static int firstCall = 1;
-        if (firstCall) {
-            /* initialize the GLX mutexes */
-            xmutex_init(&__glXmutex);
-            firstCall = 0;
-        }
-    }
+   {
+      static int firstCall = 1;
+      if (firstCall) {
+         /* initialize the GLX mutexes */
+         xmutex_init(&__glXmutex);
+         firstCall = 0;
+      }
+   }
 #endif
 
-    INIT_MESA_SPARC
-    /* The one and only long long lock */
-    __glXLock();
-
-    if (!XextHasExtension(info)) {
-	/* No GLX extension supported by this server. Oh well. */
-	__glXUnlock();
-	XMissingExtension(dpy, __glXExtensionName);
-	return 0;
-    }
-
-    /* See if a display private already exists.  If so, return it */
-    dataObj.display = dpy;
-    privList = XEHeadOfExtensionList(dataObj);
-    found = XFindOnExtensionList(privList, info->codes->extension);
-    if (found) {
-	__glXUnlock();
-	return (__GLXdisplayPrivate *) found->private_data;
-    }
-
-    /* See if the versions are compatible */
-    if (!QueryVersion(dpy, info->codes->major_opcode, &major, &minor)) {
-	/* The client and server do not agree on versions.  Punt. */
-	__glXUnlock();
-	return 0;
-    }
-
-    /*
+   INIT_MESA_SPARC;
+   /* The one and only long long lock */
+   __glXLock();
+
+   if (!XextHasExtension(info)) {
+      /* No GLX extension supported by this server. Oh well. */
+      __glXUnlock();
+      XMissingExtension(dpy, __glXExtensionName);
+      return 0;
+   }
+
+   /* See if a display private already exists.  If so, return it */
+   dataObj.display = dpy;
+   privList = XEHeadOfExtensionList(dataObj);
+   found = XFindOnExtensionList(privList, info->codes->extension);
+   if (found) {
+      __glXUnlock();
+      return (__GLXdisplayPrivate *) found->private_data;
+   }
+
+   /* See if the versions are compatible */
+   if (!QueryVersion(dpy, info->codes->major_opcode, &major, &minor)) {
+      /* The client and server do not agree on versions.  Punt. */
+      __glXUnlock();
+      return 0;
+   }
+
+   /*
     ** Allocate memory for all the pieces needed for this buffer.
     */
-    private = (XExtData *) Xmalloc(sizeof(XExtData));
-    if (!private) {
-	__glXUnlock();
-	return 0;
-    }
-    dpyPriv = (__GLXdisplayPrivate *) Xcalloc(1, sizeof(__GLXdisplayPrivate));
-    if (!dpyPriv) {
-	__glXUnlock();
-	Xfree((char*) private);
-	return 0;
-    }
-
-    /*
+   private = (XExtData *) Xmalloc(sizeof(XExtData));
+   if (!private) {
+      __glXUnlock();
+      return 0;
+   }
+   dpyPriv = (__GLXdisplayPrivate *) Xcalloc(1, sizeof(__GLXdisplayPrivate));
+   if (!dpyPriv) {
+      __glXUnlock();
+      Xfree((char *) private);
+      return 0;
+   }
+
+   /*
     ** Init the display private and then read in the screen config
     ** structures from the server.
     */
-    dpyPriv->majorOpcode = info->codes->major_opcode;
-    dpyPriv->majorVersion = major;
-    dpyPriv->minorVersion = minor;
-    dpyPriv->dpy = dpy;
+   dpyPriv->majorOpcode = info->codes->major_opcode;
+   dpyPriv->majorVersion = major;
+   dpyPriv->minorVersion = minor;
+   dpyPriv->dpy = dpy;
 
-    dpyPriv->serverGLXvendor = 0x0; 
-    dpyPriv->serverGLXversion = 0x0;
+   dpyPriv->serverGLXvendor = 0x0;
+   dpyPriv->serverGLXversion = 0x0;
 
 #ifdef GLX_DIRECT_RENDERING
-    /*
+   glx_direct = (getenv("LIBGL_ALWAYS_INDIRECT") == NULL);
+   glx_accel = (getenv("LIBGL_ALWAYS_SOFTWARE") == NULL);
+
+   /*
     ** Initialize the direct rendering per display data and functions.
     ** Note: This _must_ be done before calling any other DRI routines
     ** (e.g., those called in AllocAndFetchScreenConfigs).
     */
-    if (getenv("LIBGL_ALWAYS_INDIRECT") == NULL) {
-        dpyPriv->driDisplay.private =
-            driCreateDisplay(dpy, &dpyPriv->driDisplay);
-    }
+   if (glx_direct && glx_accel) {
+      dpyPriv->dri2Display = dri2CreateDisplay(dpy);
+      dpyPriv->driDisplay = driCreateDisplay(dpy);
+   }
+   if (glx_direct)
+      dpyPriv->driswDisplay = driswCreateDisplay(dpy);
 #endif
 
-    if (!AllocAndFetchScreenConfigs(dpy, dpyPriv)) {
-	__glXUnlock();
-	Xfree((char*) dpyPriv);
-	Xfree((char*) private);
-	return 0;
-    }
+   if (!AllocAndFetchScreenConfigs(dpy, dpyPriv)) {
+      __glXUnlock();
+      Xfree((char *) dpyPriv);
+      Xfree((char *) private);
+      return 0;
+   }
 
-    /*
+   /*
     ** Fill in the private structure.  This is the actual structure that
     ** hangs off of the Display structure.  Our private structure is
     ** referred to by this structure.  Got that?
     */
-    private->number = info->codes->extension;
-    private->next = 0;
-    private->free_private = __glXFreeDisplayPrivate;
-    private->private_data = (char *) dpyPriv;
-    XAddToExtensionList(privList, private);
-
-    if (dpyPriv->majorVersion == 1 && dpyPriv->minorVersion >= 1) {
-        __glXClientInfo(dpy, dpyPriv->majorOpcode);
-    }
-    __glXUnlock();
-
-    return dpyPriv;
+   private->number = info->codes->extension;
+   private->next = 0;
+   private->free_private = __glXFreeDisplayPrivate;
+   private->private_data = (char *) dpyPriv;
+   XAddToExtensionList(privList, private);
+
+   if (dpyPriv->majorVersion == 1 && dpyPriv->minorVersion >= 1) {
+      __glXClientInfo(dpy, dpyPriv->majorOpcode);
+   }
+   __glXUnlock();
+
+   return dpyPriv;
 }
 
 /*
 ** Setup for sending a GLX command on dpy.  Make sure the extension is
 ** initialized.  Try to avoid calling __glXInitialize as its kinda slow.
 */
-CARD8 __glXSetupForCommand(Display *dpy)
+_X_HIDDEN CARD8
+__glXSetupForCommand(Display * dpy)
 {
-    GLXContext gc;
-    __GLXdisplayPrivate *priv;
-
-    /* If this thread has a current context, flush its rendering commands */
-    gc = __glXGetCurrentContext();
-    if (gc->currentDpy) {
-	/* Flush rendering buffer of the current context, if any */
-	(void) __glXFlushRenderBuffer(gc, gc->pc);
-
-	if (gc->currentDpy == dpy) {
-	    /* Use opcode from gc because its right */
-	    INIT_MESA_SPARC
-	    return gc->majorOpcode;
-	} else {
-	    /*
-	    ** Have to get info about argument dpy because it might be to
-	    ** a different server
-	    */
-	}
-    }
-
-    /* Forced to lookup extension via the slow initialize route */
-    priv = __glXInitialize(dpy);
-    if (!priv) {
-	return 0;
-    }
-    return priv->majorOpcode;
+   GLXContext gc;
+   __GLXdisplayPrivate *priv;
+
+   /* If this thread has a current context, flush its rendering commands */
+   gc = __glXGetCurrentContext();
+   if (gc->currentDpy) {
+      /* Flush rendering buffer of the current context, if any */
+      (void) __glXFlushRenderBuffer(gc, gc->pc);
+
+      if (gc->currentDpy == dpy) {
+         /* Use opcode from gc because its right */
+         INIT_MESA_SPARC;
+         return gc->majorOpcode;
+      }
+      else {
+         /*
+          ** Have to get info about argument dpy because it might be to
+          ** a different server
+          */
+      }
+   }
+
+   /* Forced to lookup extension via the slow initialize route */
+   priv = __glXInitialize(dpy);
+   if (!priv) {
+      return 0;
+   }
+   return priv->majorOpcode;
 }
 
 /**
@@ -1349,36 +814,38 @@ CARD8 __glXSetupForCommand(Display *dpy)
  * Modify this function to use \c ctx->pc instead of the explicit
  * \c pc parameter.
  */
-GLubyte *__glXFlushRenderBuffer(__GLXcontext *ctx, GLubyte *pc)
+_X_HIDDEN GLubyte *
+__glXFlushRenderBuffer(__GLXcontext * ctx, GLubyte * pc)
 {
-    Display * const dpy = ctx->currentDpy;
+   Display *const dpy = ctx->currentDpy;
 #ifdef USE_XCB
-    xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_connection_t *c = XGetXCBConnection(dpy);
 #else
-    xGLXRenderReq *req;
+   xGLXRenderReq *req;
 #endif /* USE_XCB */
-    const GLint size = pc - ctx->buf;
+   const GLint size = pc - ctx->buf;
 
-    if ( (dpy != NULL) && (size > 0) ) {
+   if ((dpy != NULL) && (size > 0)) {
 #ifdef USE_XCB
-	xcb_glx_render(c, ctx->currentContextTag, size, (char *)ctx->buf);
+      xcb_glx_render(c, ctx->currentContextTag, size,
+                     (const uint8_t *) ctx->buf);
 #else
-	/* Send the entire buffer as an X request */
-	LockDisplay(dpy);
-	GetReq(GLXRender,req); 
-	req->reqType = ctx->majorOpcode;
-	req->glxCode = X_GLXRender; 
-	req->contextTag = ctx->currentContextTag;
-	req->length += (size + 3) >> 2;
-	_XSend(dpy, (char *)ctx->buf, size);
-	UnlockDisplay(dpy);
-	SyncHandle();
+      /* Send the entire buffer as an X request */
+      LockDisplay(dpy);
+      GetReq(GLXRender, req);
+      req->reqType = ctx->majorOpcode;
+      req->glxCode = X_GLXRender;
+      req->contextTag = ctx->currentContextTag;
+      req->length += (size + 3) >> 2;
+      _XSend(dpy, (char *) ctx->buf, size);
+      UnlockDisplay(dpy);
+      SyncHandle();
 #endif
-    }
+   }
 
-    /* Reset pointer and return it */
-    ctx->pc = ctx->buf;
-    return ctx->pc;
+   /* Reset pointer and return it */
+   ctx->pc = ctx->buf;
+   return ctx->pc;
 }
 
 
@@ -1398,35 +865,36 @@ GLubyte *__glXFlushRenderBuffer(__GLXcontext *ctx, GLubyte *pc)
  * \param data           Command data.
  * \param dataLen        Size, in bytes, of the command data.
  */
-void __glXSendLargeChunk(__GLXcontext *gc, GLint requestNumber, 
-			 GLint totalRequests,
-			 const GLvoid * data, GLint dataLen)
+_X_HIDDEN void
+__glXSendLargeChunk(__GLXcontext * gc, GLint requestNumber,
+                    GLint totalRequests, const GLvoid * data, GLint dataLen)
 {
-    Display *dpy = gc->currentDpy;
+   Display *dpy = gc->currentDpy;
 #ifdef USE_XCB
-    xcb_connection_t *c = XGetXCBConnection(dpy);
-    xcb_glx_render_large(c, gc->currentContextTag, requestNumber, totalRequests, dataLen, data);
+   xcb_connection_t *c = XGetXCBConnection(dpy);
+   xcb_glx_render_large(c, gc->currentContextTag, requestNumber,
+                        totalRequests, dataLen, data);
 #else
-    xGLXRenderLargeReq *req;
-    
-    if ( requestNumber == 1 ) {
-	LockDisplay(dpy);
-    }
-
-    GetReq(GLXRenderLarge,req); 
-    req->reqType = gc->majorOpcode;
-    req->glxCode = X_GLXRenderLarge; 
-    req->contextTag = gc->currentContextTag;
-    req->length += (dataLen + 3) >> 2;
-    req->requestNumber = requestNumber;
-    req->requestTotal = totalRequests;
-    req->dataBytes = dataLen;
-    Data(dpy, data, dataLen);
-
-    if ( requestNumber == totalRequests ) {
-	UnlockDisplay(dpy);
-	SyncHandle();
-    }
+   xGLXRenderLargeReq *req;
+
+   if (requestNumber == 1) {
+      LockDisplay(dpy);
+   }
+
+   GetReq(GLXRenderLarge, req);
+   req->reqType = gc->majorOpcode;
+   req->glxCode = X_GLXRenderLarge;
+   req->contextTag = gc->currentContextTag;
+   req->length += (dataLen + 3) >> 2;
+   req->requestNumber = requestNumber;
+   req->requestTotal = totalRequests;
+   req->dataBytes = dataLen;
+   Data(dpy, data, dataLen);
+
+   if (requestNumber == totalRequests) {
+      UnlockDisplay(dpy);
+      SyncHandle();
+   }
 #endif /* USE_XCB */
 }
 
@@ -1446,422 +914,118 @@ void __glXSendLargeChunk(__GLXcontext *gc, GLint requestNumber,
  * \param data       Command data.
  * \param dataLen    Size, in bytes, of the command data.
  */
-void __glXSendLargeCommand(__GLXcontext *ctx,
-			   const GLvoid *header, GLint headerLen,
-			   const GLvoid *data, GLint dataLen)
+_X_HIDDEN void
+__glXSendLargeCommand(__GLXcontext * ctx,
+                      const GLvoid * header, GLint headerLen,
+                      const GLvoid * data, GLint dataLen)
 {
-    GLint maxSize;
-    GLint totalRequests, requestNumber;
+   GLint maxSize;
+   GLint totalRequests, requestNumber;
 
-    /*
+   /*
     ** Calculate the maximum amount of data can be stuffed into a single
     ** packet.  sz_xGLXRenderReq is added because bufSize is the maximum
     ** packet size minus sz_xGLXRenderReq.
     */
-    maxSize = (ctx->bufSize + sz_xGLXRenderReq) - sz_xGLXRenderLargeReq;
-    totalRequests = 1 + (dataLen / maxSize);
-    if (dataLen % maxSize) totalRequests++;
+   maxSize = (ctx->bufSize + sz_xGLXRenderReq) - sz_xGLXRenderLargeReq;
+   totalRequests = 1 + (dataLen / maxSize);
+   if (dataLen % maxSize)
+      totalRequests++;
 
-    /*
+   /*
     ** Send all of the command, except the large array, as one request.
     */
-    assert( headerLen <= maxSize );
-    __glXSendLargeChunk(ctx, 1, totalRequests, header, headerLen);
+   assert(headerLen <= maxSize);
+   __glXSendLargeChunk(ctx, 1, totalRequests, header, headerLen);
 
-    /*
+   /*
     ** Send enough requests until the whole array is sent.
     */
-    for ( requestNumber = 2 ; requestNumber <= (totalRequests - 1) ; requestNumber++ ) {
-	__glXSendLargeChunk(ctx, requestNumber, totalRequests, data, maxSize);
-	data = (const GLvoid *) (((const GLubyte *) data) + maxSize);
-	dataLen -= maxSize;
-	assert( dataLen > 0 );
-    }
-
-    assert( dataLen <= maxSize );
-    __glXSendLargeChunk(ctx, requestNumber, totalRequests, data, dataLen);
-}
-
-/************************************************************************/
-
-PUBLIC GLXContext glXGetCurrentContext(void)
-{
-    GLXContext cx = __glXGetCurrentContext();
-    
-    if (cx == &dummyContext) {
-	return NULL;
-    } else {
-	return cx;
-    }
-}
+   for (requestNumber = 2; requestNumber <= (totalRequests - 1);
+        requestNumber++) {
+      __glXSendLargeChunk(ctx, requestNumber, totalRequests, data, maxSize);
+      data = (const GLvoid *) (((const GLubyte *) data) + maxSize);
+      dataLen -= maxSize;
+      assert(dataLen > 0);
+   }
 
-PUBLIC GLXDrawable glXGetCurrentDrawable(void)
-{
-    GLXContext gc = __glXGetCurrentContext();
-    return gc->currentDrawable;
+   assert(dataLen <= maxSize);
+   __glXSendLargeChunk(ctx, requestNumber, totalRequests, data, dataLen);
 }
 
-
 /************************************************************************/
 
-#ifdef GLX_DIRECT_RENDERING
-/* Return the DRI per screen structure */
-__DRIscreen *__glXFindDRIScreen(__DRInativeDisplay *dpy, int scrn)
-{
-    __DRIscreen *pDRIScreen = NULL;
-    XExtDisplayInfo *info = __glXFindDisplay(dpy);
-    XExtData **privList, *found;
-    __GLXdisplayPrivate *dpyPriv;
-    XEDataObject dataObj;
-
-    __glXLock();
-    dataObj.display = dpy;
-    privList = XEHeadOfExtensionList(dataObj);
-    found = XFindOnExtensionList(privList, info->codes->extension);
-    __glXUnlock();
-
-    if (found) {
-	dpyPriv = (__GLXdisplayPrivate *)found->private_data;
-	pDRIScreen = &dpyPriv->screenConfigs[scrn].driScreen;
-    }
-
-    return pDRIScreen;
-}
-#endif
-
-/************************************************************************/
-
-static Bool SendMakeCurrentRequest( Display *dpy, CARD8 opcode,
-    GLXContextID gc, GLXContextTag old_gc, GLXDrawable draw, GLXDrawable read,
-    xGLXMakeCurrentReply * reply );
-
-/**
- * Sends a GLX protocol message to the specified display to make the context
- * and the drawables current.
- *
- * \param dpy     Display to send the message to.
- * \param opcode  Major opcode value for the display.
- * \param gc_id   Context tag for the context to be made current.
- * \param draw    Drawable ID for the "draw" drawable.
- * \param read    Drawable ID for the "read" drawable.
- * \param reply   Space to store the X-server's reply.
- *
- * \warning
- * This function assumes that \c dpy is locked with \c LockDisplay on entry.
- */
-static Bool SendMakeCurrentRequest(Display *dpy, CARD8 opcode,
-				   GLXContextID gc_id, GLXContextTag gc_tag,
-				   GLXDrawable draw, GLXDrawable read,
-				   xGLXMakeCurrentReply *reply)
-{
-    Bool ret;
-
-
-    LockDisplay(dpy);
-
-    if (draw == read) {
-	xGLXMakeCurrentReq *req;
-
-	GetReq(GLXMakeCurrent,req);
-	req->reqType = opcode;
-	req->glxCode = X_GLXMakeCurrent;
-	req->drawable = draw;
-	req->context = gc_id;
-	req->oldContextTag = gc_tag;
-    }
-    else {
-	__GLXdisplayPrivate *priv = __glXInitialize(dpy);
-
-	/* If the server can support the GLX 1.3 version, we should
-	 * perfer that.  Not only that, some servers support GLX 1.3 but
-	 * not the SGI extension.
-	 */
-
-	if ((priv->majorVersion > 1) || (priv->minorVersion >= 3)) {
-	    xGLXMakeContextCurrentReq *req;
-
-	    GetReq(GLXMakeContextCurrent,req);
-	    req->reqType = opcode;
-	    req->glxCode = X_GLXMakeContextCurrent;
-	    req->drawable = draw;
-	    req->readdrawable = read;
-	    req->context = gc_id;
-	    req->oldContextTag = gc_tag;
-	}
-	else {
-	    xGLXVendorPrivateWithReplyReq *vpreq;
-	    xGLXMakeCurrentReadSGIReq *req;
-
-	    GetReqExtra(GLXVendorPrivateWithReply,
-			sz_xGLXMakeCurrentReadSGIReq-sz_xGLXVendorPrivateWithReplyReq,vpreq);
-	    req = (xGLXMakeCurrentReadSGIReq *)vpreq;
-	    req->reqType = opcode;
-	    req->glxCode = X_GLXVendorPrivateWithReply;
-	    req->vendorCode = X_GLXvop_MakeCurrentReadSGI;
-	    req->drawable = draw;
-	    req->readable = read;
-	    req->context = gc_id;
-	    req->oldContextTag = gc_tag;
-	}
-    }
-
-    ret = _XReply(dpy, (xReply*) reply, 0, False);
-
-    UnlockDisplay(dpy);
-    SyncHandle();
-
-    return ret;
-}
-
-
-#ifdef GLX_DIRECT_RENDERING
-static Bool BindContextWrapper( Display *dpy, GLXContext gc,
-				GLXDrawable draw, GLXDrawable read )
-{
-    return (*gc->driContext.bindContext)(dpy, gc->screen, draw, read,
-					 & gc->driContext);
-}
-
-
-static Bool UnbindContextWrapper( GLXContext gc )
-{
-    return (*gc->driContext.unbindContext)(gc->currentDpy, gc->screen, 
-					   gc->currentDrawable,
-					   gc->currentReadable,
-					   & gc->driContext );
-}
-#endif /* GLX_DIRECT_RENDERING */
-
-
-/**
- * Make a particular context current.
- * 
- * \note This is in this file so that it can access dummyContext.
- */
-USED static Bool MakeContextCurrent(Display *dpy, GLXDrawable draw,
-                                    GLXDrawable read, GLXContext gc)
-{
-    xGLXMakeCurrentReply reply;
-    const GLXContext oldGC = __glXGetCurrentContext();
-    const CARD8 opcode = __glXSetupForCommand(dpy);
-    const CARD8 oldOpcode = ((gc == oldGC) || (oldGC == &dummyContext))
-      ? opcode : __glXSetupForCommand(oldGC->currentDpy);
-    Bool bindReturnValue;
-
-
-    if (!opcode || !oldOpcode) {
-	return GL_FALSE;
-    }
-
-    /* Make sure that the new context has a nonzero ID.  In the request,
-     * a zero context ID is used only to mean that we bind to no current
-     * context.
-     */
-    if ((gc != NULL) && (gc->xid == None)) {
-	return GL_FALSE;
-    }
-
-#ifndef GLX_DIRECT_RENDERING
-    if (gc && gc->isDirect) {
-	return GL_FALSE;
-    }
-#endif
-
-    _glapi_check_multithread();
-
-#ifdef GLX_DIRECT_RENDERING
-    /* Bind the direct rendering context to the drawable */
-    if (gc && gc->isDirect) {
-	bindReturnValue = (gc->driContext.private) 
-	  ? BindContextWrapper(dpy, gc, draw, read)
-	  : False;
-    } else
-#endif
-    {
-	/* Send a glXMakeCurrent request to bind the new context. */
-	bindReturnValue = 
-	  SendMakeCurrentRequest(dpy, opcode, gc ? gc->xid : None,
-				 ((dpy != oldGC->currentDpy) || oldGC->isDirect)
-				 ? None : oldGC->currentContextTag,
-				 draw, read, &reply);
-    }
-
-
-    if (!bindReturnValue) {
-	return False;
-    }
-
-    if ((dpy != oldGC->currentDpy || (gc && gc->isDirect)) &&
-	!oldGC->isDirect && oldGC != &dummyContext) {
-	xGLXMakeCurrentReply dummy_reply;
-
-	/* We are either switching from one dpy to another and have to
-	 * send a request to the previous dpy to unbind the previous
-	 * context, or we are switching away from a indirect context to
-	 * a direct context and have to send a request to the dpy to
-	 * unbind the previous context.
-	 */
-	(void) SendMakeCurrentRequest(oldGC->currentDpy, oldOpcode, None,
-				      oldGC->currentContextTag, None, None,
-				      & dummy_reply);
-    }
-#ifdef GLX_DIRECT_RENDERING
-    else if (oldGC->isDirect && oldGC->driContext.private) {
-	(void) UnbindContextWrapper(oldGC);
-    }
-#endif
-
-
-    /* Update our notion of what is current */
-    __glXLock();
-    if (gc == oldGC) {
-	/* Even though the contexts are the same the drawable might have
-	 * changed.  Note that gc cannot be the dummy, and that oldGC
-	 * cannot be NULL, therefore if they are the same, gc is not
-	 * NULL and not the dummy.
-	 */
-	gc->currentDrawable = draw;
-	gc->currentReadable = read;
-    } else {
-	if (oldGC != &dummyContext) {
-	    /* Old current context is no longer current to anybody */
-	    oldGC->currentDpy = 0;
-	    oldGC->currentDrawable = None;
-	    oldGC->currentReadable = None;
-	    oldGC->currentContextTag = 0;
-
-	    if (oldGC->xid == None) {
-		/* We are switching away from a context that was
-		 * previously destroyed, so we need to free the memory
-		 * for the old handle.
-		 */
-#ifdef GLX_DIRECT_RENDERING
-		/* Destroy the old direct rendering context */
-		if (oldGC->isDirect) {
-		    if (oldGC->driContext.private) {
-			(*oldGC->driContext.destroyContext)
-			    (dpy, oldGC->screen, oldGC->driContext.private);
-			oldGC->driContext.private = NULL;
-		    }
-		}
-#endif
-		__glXFreeContext(oldGC);
-	    }
-	}
-	if (gc) {
-	    __glXSetCurrentContext(gc);
-
-	    gc->currentDpy = dpy;
-	    gc->currentDrawable = draw;
-	    gc->currentReadable = read;
-
-            if (!gc->isDirect) {
-               if (!IndirectAPI)
-                  IndirectAPI = __glXNewIndirectAPI();
-               _glapi_set_dispatch(IndirectAPI);
-
-#ifdef GLX_USE_APPLEGL
-               do {
-                   extern void XAppleDRIUseIndirectDispatch(void);
-                   XAppleDRIUseIndirectDispatch();
-               } while (0);
-#endif
-
-		__GLXattribute *state = 
-		  (__GLXattribute *)(gc->client_state_private);
-
-		gc->currentContextTag = reply.contextTag;
-		if (state->array_state == NULL) {
-		    (void) glGetString(GL_EXTENSIONS);
-		    (void) glGetString(GL_VERSION);
-		    __glXInitVertexArrayState(gc);
-		}
-	    }
-	    else {
-		gc->currentContextTag = -1;
-	    }
-	} else {
-	    __glXSetCurrentContext(&dummyContext);
-#ifdef GLX_DIRECT_RENDERING
-            _glapi_set_dispatch(NULL);  /* no-op functions */
-#endif
-	}
-    }
-    __glXUnlock();
-    return GL_TRUE;
-}
-
-
-PUBLIC Bool glXMakeCurrent(Display *dpy, GLXDrawable draw, GLXContext gc)
-{
-    return MakeContextCurrent( dpy, draw, draw, gc );
-}
-
-PUBLIC GLX_ALIAS(Bool, glXMakeCurrentReadSGI,
-	  (Display *dpy, GLXDrawable d, GLXDrawable r, GLXContext ctx),
-	  (dpy, d, r, ctx), MakeContextCurrent)
-
-PUBLIC GLX_ALIAS(Bool, glXMakeContextCurrent,
-	  (Display *dpy, GLXDrawable d, GLXDrawable r, GLXContext ctx),
-	  (dpy, d, r, ctx), MakeContextCurrent)
-
-
 #ifdef DEBUG
-void __glXDumpDrawBuffer(__GLXcontext *ctx)
+_X_HIDDEN void
+__glXDumpDrawBuffer(__GLXcontext * ctx)
 {
-    GLubyte *p = ctx->buf;
-    GLubyte *end = ctx->pc;
-    GLushort opcode, length;
-
-    while (p < end) {
-	/* Fetch opcode */
-	opcode = *((GLushort*) p);
-	length = *((GLushort*) (p + 2));
-	printf("%2x: %5d: ", opcode, length);
-	length -= 4;
-	p += 4;
-	while (length > 0) {
-	    printf("%08x ", *((unsigned *) p));
-	    p += 4;
-	    length -= 4;
-	}
-	printf("\n");
-    }	    
+   GLubyte *p = ctx->buf;
+   GLubyte *end = ctx->pc;
+   GLushort opcode, length;
+
+   while (p < end) {
+      /* Fetch opcode */
+      opcode = *((GLushort *) p);
+      length = *((GLushort *) (p + 2));
+      printf("%2x: %5d: ", opcode, length);
+      length -= 4;
+      p += 4;
+      while (length > 0) {
+         printf("%08x ", *((unsigned *) p));
+         p += 4;
+         length -= 4;
+      }
+      printf("\n");
+   }
 }
 #endif
 
 #ifdef  USE_SPARC_ASM
 /*
- * Used only when we are sparc, using sparc assembler.
+ * This is where our dispatch table's bounds are.
+ * And the static mesa_init is taken directly from
+ * Mesa's 'sparc.c' initializer.
+ *
+ * We need something like this here, because this version
+ * of openGL/glx never initializes a Mesa context, and so
+ * the address of the dispatch table pointer never gets stuffed
+ * into the dispatch jump table otherwise.
+ *
+ * It matters only on SPARC, and only if you are using assembler
+ * code instead of C-code indirect dispatch.
  *
+ * -- FEM, 04.xii.03
  */
+extern unsigned int _mesa_sparc_glapi_begin;
+extern unsigned int _mesa_sparc_glapi_end;
+extern void __glapi_sparc_icache_flush(unsigned int *);
 
 static void
 _glx_mesa_init_sparc_glapi_relocs(void)
 {
-	unsigned int *insn_ptr, *end_ptr;
-	unsigned long disp_addr;
-
-	insn_ptr = &_mesa_sparc_glapi_begin;
-	end_ptr = &_mesa_sparc_glapi_end;
-	disp_addr = (unsigned long) &_glapi_Dispatch;
-
-	/*
-         * Verbatim from Mesa sparc.c.  It's needed because there doesn't
-         * seem to be a better way to do this:
-         *
-         * UNCONDITIONAL_JUMP ( (*_glapi_Dispatch) + entry_offset )
-         *
-         * This code is patching in the ADDRESS of the pointer to the
-         * dispatch table.  Hence, it must be called exactly once, because
-         * that address is not going to change.
-         *
-         * What it points to can change, but Mesa (and hence, we) assume
-         * that there is only one pointer.
-         *
-	 */
-	while (insn_ptr < end_ptr) {
-#if ( defined(__sparc_v9__) && ( !defined(__linux__) || defined(__linux_64__) ) )	
+   unsigned int *insn_ptr, *end_ptr;
+   unsigned long disp_addr;
+
+   insn_ptr = &_mesa_sparc_glapi_begin;
+   end_ptr = &_mesa_sparc_glapi_end;
+   disp_addr = (unsigned long) &_glapi_Dispatch;
+
+   /*
+    * Verbatim from Mesa sparc.c.  It's needed because there doesn't
+    * seem to be a better way to do this:
+    *
+    * UNCONDITIONAL_JUMP ( (*_glapi_Dispatch) + entry_offset )
+    *
+    * This code is patching in the ADDRESS of the pointer to the
+    * dispatch table.  Hence, it must be called exactly once, because
+    * that address is not going to change.
+    *
+    * What it points to can change, but Mesa (and hence, we) assume
+    * that there is only one pointer.
+    *
+    */
+   while (insn_ptr < end_ptr) {
+#if ( defined(__sparc_v9__) && ( !defined(__linux__) || defined(__linux_64__) ) )
 /*
 	This code patches for 64-bit addresses.  This had better
 	not happen for Sparc/Linux, no matter what architecture we
@@ -1870,20 +1034,19 @@ _glx_mesa_init_sparc_glapi_relocs(void)
         The 'defined(__linux_64__)' is used here as a placeholder for
         when we do do 64-bit usermode on sparc linux.
 	*/
-		insn_ptr[0] |= (disp_addr >> (32 + 10));
-		insn_ptr[1] |= ((disp_addr & 0xffffffff) >> 10);
-		__glapi_sparc_icache_flush(&insn_ptr[0]);
-		insn_ptr[2] |= ((disp_addr >> 32) & ((1 << 10) - 1));
-		insn_ptr[3] |= (disp_addr & ((1 << 10) - 1));
-		__glapi_sparc_icache_flush(&insn_ptr[2]);
-		insn_ptr += 11;
+      insn_ptr[0] |= (disp_addr >> (32 + 10));
+      insn_ptr[1] |= ((disp_addr & 0xffffffff) >> 10);
+      __glapi_sparc_icache_flush(&insn_ptr[0]);
+      insn_ptr[2] |= ((disp_addr >> 32) & ((1 << 10) - 1));
+      insn_ptr[3] |= (disp_addr & ((1 << 10) - 1));
+      __glapi_sparc_icache_flush(&insn_ptr[2]);
+      insn_ptr += 11;
 #else
-		insn_ptr[0] |= (disp_addr >> 10);
-		insn_ptr[1] |= (disp_addr & ((1 << 10) - 1));
-		__glapi_sparc_icache_flush(&insn_ptr[0]);
-		insn_ptr += 5;
+      insn_ptr[0] |= (disp_addr >> 10);
+      insn_ptr[1] |= (disp_addr & ((1 << 10) - 1));
+      __glapi_sparc_icache_flush(&insn_ptr[0]);
+      insn_ptr += 5;
 #endif
-	}
+   }
 }
-#endif  /* sparc ASM in use */
-
+#endif /* sparc ASM in use */
diff --git a/src/glx/x11/glxextensions.c b/src/glx/x11/glxextensions.c
index 1d99b61db08..d061fa93336 100644
--- a/src/glx/x11/glxextensions.c
+++ b/src/glx/x11/glxextensions.c
@@ -34,7 +34,7 @@
 #include <string.h>
 #include "glapi.h"
 #include "glxextensions.h"
-#include "simple_list.h"
+
 
 #define SET_BIT(m,b)   (m[ (b) / 8 ] |=  (1U << ((b) % 8)))
 #define CLR_BIT(m,b)   (m[ (b) / 8 ] &= ~(1U << ((b) % 8)))
@@ -48,27 +48,29 @@
 #define EXT_ENABLED(bit,supported) (IS_SET( supported, bit ))
 
 
-struct extension_info {
-   const char * const name;
-   unsigned   name_len;
+struct extension_info
+{
+   const char *const name;
+   unsigned name_len;
 
-   unsigned char  bit;
+   unsigned char bit;
 
    /* This is the lowest version of GLX that "requires" this extension.
     * For example, GLX 1.3 requires SGIX_fbconfig, SGIX_pbuffer, and
     * SGI_make_current_read.  If the extension is not required by any known
     * version of GLX, use 0, 0.
     */
-   unsigned char  version_major;
-   unsigned char  version_minor;
-   unsigned char  client_support;
-   unsigned char  direct_support;
-   unsigned char  client_only;       /** Is the extension client-side only? */
-   unsigned char  direct_only;       /** Is the extension for direct
+   unsigned char version_major;
+   unsigned char version_minor;
+   unsigned char client_support;
+   unsigned char direct_support;
+   unsigned char client_only;        /** Is the extension client-side only? */
+   unsigned char direct_only;        /** Is the extension for direct
 				      * contexts only?
 				      */
 };
 
+/* *INDENT-OFF* */
 static const struct extension_info known_glx_extensions[] = {
    { GLX(ARB_get_proc_address),        VER(1,4), Y, N, Y, N },
    { GLX(ARB_multisample),             VER(1,4), Y, Y, N, N },
@@ -245,14 +247,15 @@ static const struct extension_info known_gl_extensions[] = {
    { GL(SUN_slice_accum),                VER(0,0), Y, N, N, N },
    { NULL }
 };
+/* *INDENT-ON* */
 
 
 /* global bit-fields of available extensions and their characteristics */
 static unsigned char client_glx_support[8];
 static unsigned char client_glx_only[8];
 static unsigned char direct_glx_only[8];
-static unsigned char client_gl_support[ __GL_EXT_BYTES ];
-static unsigned char client_gl_only[ __GL_EXT_BYTES ];
+static unsigned char client_gl_support[__GL_EXT_BYTES];
+static unsigned char client_gl_only[__GL_EXT_BYTES];
 
 /**
  * Bits representing the set of extensions that are enabled by default in all
@@ -267,12 +270,13 @@ static const unsigned gl_major = 1;
 static const unsigned gl_minor = 4;
 
 /* client extensions string */
-static const char * __glXGLXClientExtensions = NULL;
+static const char *__glXGLXClientExtensions = NULL;
 
-static void __glXExtensionsCtr( void );
-static void __glXExtensionsCtrScreen( __GLXscreenConfigs *psc );
-static void __glXProcessServerString( const struct extension_info * ext,
-    const char * server_string, unsigned char * server_support );
+static void __glXExtensionsCtr(void);
+static void __glXExtensionsCtrScreen(__GLXscreenConfigs * psc);
+static void __glXProcessServerString(const struct extension_info *ext,
+                                     const char *server_string,
+                                     unsigned char *server_support);
 
 /**
  * Set the state of a GLX extension.
@@ -283,24 +287,24 @@ static void __glXProcessServerString( const struct extension_info * ext,
  * \param supported Table in which the state of the extension is to be set.
  */
 static void
-set_glx_extension( const struct extension_info * ext,
-		   const char * name, unsigned name_len, GLboolean state,
-		   unsigned char * supported )
+set_glx_extension(const struct extension_info *ext,
+                  const char *name, unsigned name_len, GLboolean state,
+                  unsigned char *supported)
 {
-   unsigned   i;
+   unsigned i;
 
 
-   for ( i = 0 ; ext[i].name != NULL ; i++ ) {
-      if ( (name_len == ext[i].name_len)
-	   && (strncmp( ext[i].name, name, name_len ) == 0) ) {
-	 if ( state ) {
-	    SET_BIT( supported, ext[i].bit );
-	 }
-	 else {
-	    CLR_BIT( supported, ext[i].bit );
-	 }
+   for (i = 0; ext[i].name != NULL; i++) {
+      if ((name_len == ext[i].name_len)
+          && (strncmp(ext[i].name, name, name_len) == 0)) {
+         if (state) {
+            SET_BIT(supported, ext[i].bit);
+         }
+         else {
+            CLR_BIT(supported, ext[i].bit);
+         }
 
-	 return;
+         return;
       }
    }
 }
@@ -321,121 +325,106 @@ set_glx_extension( const struct extension_info * ext,
  * the data pointed by \c server_support must be preinitialized to zero.
  */
 static void
-__glXProcessServerString( const struct extension_info * ext,
-			  const char * server_string,
-			  unsigned char * server_support )
+__glXProcessServerString(const struct extension_info *ext,
+                         const char *server_string,
+                         unsigned char *server_support)
 {
-   unsigned  base;
-   unsigned  len;
+   unsigned base;
+   unsigned len;
 
-   for ( base = 0 ; server_string[ base ] != NUL ; /* empty */ ) {
+   for (base = 0; server_string[base] != NUL; /* empty */ ) {
       /* Determine the length of the next extension name.
        */
-      for ( len = 0
-	    ; (server_string[ base + len ] != SEPARATOR)
-	    && (server_string[ base + len ] != NUL)
-	    ; len++ ) {
-	 /* empty */
+      for (len = 0; (server_string[base + len] != SEPARATOR)
+           && (server_string[base + len] != NUL);
+           len++) {
+         /* empty */
       }
 
       /* Set the bit for the extension in the server_support table.
        */
-      set_glx_extension( ext, & server_string[ base ], len, GL_TRUE,
-			 server_support );
+      set_glx_extension(ext, &server_string[base], len, GL_TRUE,
+                        server_support);
+
 
- 
       /* Advance to the next extension string.  This means that we skip
        * over the previous string and any trialing white-space.
        */
-      for ( base += len ;
-	    (server_string[ base ] == SEPARATOR)
-	    && (server_string[ base ] != NUL)
-	    ; base++ ) {
-	 /* empty */
+      for (base += len; (server_string[base] == SEPARATOR)
+           && (server_string[base] != NUL);
+           base++) {
+         /* empty */
       }
    }
 }
 
-
-/**
- * Enable a named GLX extension on a given screen.
- * Drivers should not call this function directly.  They should instead use
- * \c glXGetProcAddress to obtain a pointer to the function.
- *
- * \param psc   Pointer to GLX per-screen record.
- * \param name  Name of the extension to enable.
- *
- * \sa glXGetProcAddress
- *
- * \since Internal API version 20030813.
- */
 void
-__glXScrEnableExtension( __GLXscreenConfigs *psc, const char * name )
+__glXEnableDirectExtension(__GLXscreenConfigs * psc, const char *name)
 {
    __glXExtensionsCtr();
    __glXExtensionsCtrScreen(psc);
-   set_glx_extension( known_glx_extensions, name, strlen( name ), GL_TRUE,
-		      psc->direct_support );
-}
 
+   set_glx_extension(known_glx_extensions,
+                     name, strlen(name), GL_TRUE, psc->direct_support);
+}
 
 /**
  * Initialize global extension support tables.
  */
 
 static void
-__glXExtensionsCtr( void )
+__glXExtensionsCtr(void)
 {
-   unsigned   i;
+   unsigned i;
    static GLboolean ext_list_first_time = GL_TRUE;
 
 
-   if ( ext_list_first_time ) {
+   if (ext_list_first_time) {
       ext_list_first_time = GL_FALSE;
 
-      (void) memset( client_glx_support, 0, sizeof( client_glx_support ) );
-      (void) memset( direct_glx_support, 0, sizeof( direct_glx_support ) );
-      (void) memset( client_glx_only,    0, sizeof( client_glx_only ) );
-      (void) memset( direct_glx_only,    0, sizeof( direct_glx_only ) );
+      (void) memset(client_glx_support, 0, sizeof(client_glx_support));
+      (void) memset(direct_glx_support, 0, sizeof(direct_glx_support));
+      (void) memset(client_glx_only, 0, sizeof(client_glx_only));
+      (void) memset(direct_glx_only, 0, sizeof(direct_glx_only));
 
-      (void) memset( client_gl_support,  0, sizeof( client_gl_support ) );
-      (void) memset( client_gl_only,     0, sizeof( client_gl_only ) );
+      (void) memset(client_gl_support, 0, sizeof(client_gl_support));
+      (void) memset(client_gl_only, 0, sizeof(client_gl_only));
 
-      for ( i = 0 ; known_glx_extensions[i].name != NULL ; i++ ) {
-	 const unsigned  bit = known_glx_extensions[i].bit;
+      for (i = 0; known_glx_extensions[i].name != NULL; i++) {
+         const unsigned bit = known_glx_extensions[i].bit;
 
-	 if ( known_glx_extensions[i].client_support ) {
-	    SET_BIT( client_glx_support, bit );
-	 }
+         if (known_glx_extensions[i].client_support) {
+            SET_BIT(client_glx_support, bit);
+         }
 
-	 if ( known_glx_extensions[i].direct_support ) {
-	    SET_BIT( direct_glx_support, bit );
-	 }
+         if (known_glx_extensions[i].direct_support) {
+            SET_BIT(direct_glx_support, bit);
+         }
 
-	 if ( known_glx_extensions[i].client_only ) {
-	    SET_BIT( client_glx_only, bit );
-	 }
+         if (known_glx_extensions[i].client_only) {
+            SET_BIT(client_glx_only, bit);
+         }
 
-	 if ( known_glx_extensions[i].direct_only ) {
-	    SET_BIT( direct_glx_only, bit );
-	 }
+         if (known_glx_extensions[i].direct_only) {
+            SET_BIT(direct_glx_only, bit);
+         }
       }
 
-      for ( i = 0 ; known_gl_extensions[i].name != NULL ; i++ ) {
-	 const unsigned  bit = known_gl_extensions[i].bit;
+      for (i = 0; known_gl_extensions[i].name != NULL; i++) {
+         const unsigned bit = known_gl_extensions[i].bit;
 
-	 if ( known_gl_extensions[i].client_support ) {
-	    SET_BIT( client_gl_support, bit );
-	 }
+         if (known_gl_extensions[i].client_support) {
+            SET_BIT(client_gl_support, bit);
+         }
 
-	 if ( known_gl_extensions[i].client_only ) {
-	    SET_BIT( client_gl_only, bit );
-	 }
+         if (known_gl_extensions[i].client_only) {
+            SET_BIT(client_gl_only, bit);
+         }
       }
-      
+
 #if 0
-      fprintf( stderr, "[%s:%u] Maximum client library version: %u.%u\n",
-	       __func__, __LINE__, gl_major, gl_minor );
+      fprintf(stderr, "[%s:%u] Maximum client library version: %u.%u\n",
+              __func__, __LINE__, gl_major, gl_minor);
 #endif
    }
 }
@@ -448,13 +437,13 @@ __glXExtensionsCtr( void )
  */
 
 static void
-__glXExtensionsCtrScreen( __GLXscreenConfigs *psc )
+__glXExtensionsCtrScreen(__GLXscreenConfigs * psc)
 {
-    if (psc->ext_list_first_time) {
-	psc->ext_list_first_time = GL_FALSE;
-	(void) memcpy( psc->direct_support, direct_glx_support,
-		       sizeof( direct_glx_support ) );
-    }
+   if (psc->ext_list_first_time) {
+      psc->ext_list_first_time = GL_FALSE;
+      (void) memcpy(psc->direct_support, direct_glx_support,
+                    sizeof(direct_glx_support));
+   }
 }
 
 
@@ -468,14 +457,14 @@ __glXExtensionsCtrScreen( __GLXscreenConfigs *psc )
  *          \c NULL, then \c GL_FALSE is returned.
  */
 GLboolean
-__glXExtensionBitIsEnabled( __GLXscreenConfigs *psc, unsigned bit )
+__glXExtensionBitIsEnabled(__GLXscreenConfigs * psc, unsigned bit)
 {
    GLboolean enabled = GL_FALSE;
 
-   if ( psc != NULL ) {
+   if (psc != NULL) {
       __glXExtensionsCtr();
-      __glXExtensionsCtrScreen( psc );
-      enabled = EXT_ENABLED( bit, psc->direct_support );
+      __glXExtensionsCtrScreen(psc);
+      enabled = EXT_ENABLED(bit, psc->direct_support);
    }
 
    return enabled;
@@ -487,12 +476,12 @@ __glXExtensionBitIsEnabled( __GLXscreenConfigs *psc, unsigned bit )
  *
  */
 GLboolean
-__glExtensionBitIsEnabled( const __GLXcontext * gc, unsigned bit )
+__glExtensionBitIsEnabled(const __GLXcontext * gc, unsigned bit)
 {
    GLboolean enabled = GL_FALSE;
 
-   if ( gc != NULL ) {
-      enabled = EXT_ENABLED( bit, gc->gl_extension_bits );
+   if (gc != NULL) {
+      enabled = EXT_ENABLED(bit, gc->gl_extension_bits);
    }
 
    return enabled;
@@ -504,34 +493,34 @@ __glExtensionBitIsEnabled( const __GLXcontext * gc, unsigned bit )
  * Convert a bit-field to a string of supported extensions.
  */
 static char *
-__glXGetStringFromTable( const struct extension_info * ext,
-			 const unsigned char * supported )
+__glXGetStringFromTable(const struct extension_info *ext,
+                        const unsigned char *supported)
 {
-   unsigned   i;
-   unsigned   ext_str_len;
-   char * ext_str;
-   char * point;
+   unsigned i;
+   unsigned ext_str_len;
+   char *ext_str;
+   char *point;
 
 
    ext_str_len = 0;
-   for ( i = 0 ; ext[i].name != NULL ; i++ ) {
-      if ( EXT_ENABLED( ext[i].bit, supported ) ) {
-	 ext_str_len += ext[i].name_len + 1;
+   for (i = 0; ext[i].name != NULL; i++) {
+      if (EXT_ENABLED(ext[i].bit, supported)) {
+         ext_str_len += ext[i].name_len + 1;
       }
    }
 
-   ext_str = Xmalloc( ext_str_len + 1 );
-   if ( ext_str != NULL ) {
+   ext_str = Xmalloc(ext_str_len + 1);
+   if (ext_str != NULL) {
       point = ext_str;
 
-      for ( i = 0 ; ext[i].name != NULL ; i++ ) {
-	 if ( EXT_ENABLED( ext[i].bit, supported ) ) {
-	    (void) memcpy( point, ext[i].name, ext[i].name_len );
-	    point += ext[i].name_len;
+      for (i = 0; ext[i].name != NULL; i++) {
+         if (EXT_ENABLED(ext[i].bit, supported)) {
+            (void) memcpy(point, ext[i].name, ext[i].name_len);
+            point += ext[i].name_len;
 
-	    *point = ' ';
-	    point++;
-	 }
+            *point = ' ';
+            point++;
+         }
       }
 
       *point = '\0';
@@ -545,12 +534,12 @@ __glXGetStringFromTable( const struct extension_info * ext,
  * Get the string of client library supported extensions.
  */
 const char *
-__glXGetClientExtensions( void )
+__glXGetClientExtensions(void)
 {
-   if ( __glXGLXClientExtensions == NULL ) {
+   if (__glXGLXClientExtensions == NULL) {
       __glXExtensionsCtr();
-      __glXGLXClientExtensions = __glXGetStringFromTable( known_glx_extensions,
-							  client_glx_support );
+      __glXGLXClientExtensions = __glXGetStringFromTable(known_glx_extensions,
+                                                         client_glx_support);
    }
 
    return __glXGLXClientExtensions;
@@ -568,20 +557,20 @@ __glXGetClientExtensions( void )
  */
 
 void
-__glXCalculateUsableExtensions( __GLXscreenConfigs *psc,
-				GLboolean display_is_direct_capable,
-				int minor_version )
+__glXCalculateUsableExtensions(__GLXscreenConfigs * psc,
+                               GLboolean display_is_direct_capable,
+                               int minor_version)
 {
    unsigned char server_support[8];
    unsigned char usable[8];
-   unsigned      i;
+   unsigned i;
 
    __glXExtensionsCtr();
-   __glXExtensionsCtrScreen( psc );
+   __glXExtensionsCtrScreen(psc);
 
-   (void) memset( server_support, 0, sizeof( server_support ) );
-   __glXProcessServerString( known_glx_extensions,
-			     psc->serverGLXexts, server_support );
+   (void) memset(server_support, 0, sizeof(server_support));
+   __glXProcessServerString(known_glx_extensions,
+                            psc->serverGLXexts, server_support);
 
 
    /* This is a hack.  Some servers support GLX 1.3 but don't export
@@ -590,20 +579,20 @@ __glXCalculateUsableExtensions( __GLXscreenConfigs *psc,
     * "emulated" as well.
     */
 
-   if ( minor_version >= 3 ) {
-      SET_BIT( server_support, EXT_visual_info_bit );
-      SET_BIT( server_support, EXT_visual_rating_bit );
-      SET_BIT( server_support, SGI_make_current_read_bit );
-      SET_BIT( server_support, SGIX_fbconfig_bit );
-      SET_BIT( server_support, SGIX_pbuffer_bit );
-       
+   if (minor_version >= 3) {
+      SET_BIT(server_support, EXT_visual_info_bit);
+      SET_BIT(server_support, EXT_visual_rating_bit);
+      SET_BIT(server_support, SGI_make_current_read_bit);
+      SET_BIT(server_support, SGIX_fbconfig_bit);
+      SET_BIT(server_support, SGIX_pbuffer_bit);
+
       /* This one is a little iffy.  GLX 1.3 doesn't incorporate all of this
        * extension.  However, the only part that is not strictly client-side
        * is shared.  That's the glXQueryContext / glXQueryContextInfoEXT
        * function.
        */
 
-      SET_BIT( server_support, EXT_import_context_bit );
+      SET_BIT(server_support, EXT_import_context_bit);
    }
 
 
@@ -617,22 +606,24 @@ __glXCalculateUsableExtensions( __GLXscreenConfigs *psc,
     * support it.
     */
 
-   if ( display_is_direct_capable ) {
-      for ( i = 0 ; i < 8 ; i++ ) {
-	 usable[i] = (client_glx_support[i] & client_glx_only[i])
-	     | (client_glx_support[i] & psc->direct_support[i] & server_support[i])
-	     | (client_glx_support[i] & psc->direct_support[i] & direct_glx_only[i]);
+   if (display_is_direct_capable) {
+      for (i = 0; i < 8; i++) {
+         usable[i] = (client_glx_support[i] & client_glx_only[i])
+            | (client_glx_support[i] & psc->
+               direct_support[i] & server_support[i])
+            | (client_glx_support[i] & psc->
+               direct_support[i] & direct_glx_only[i]);
       }
    }
    else {
-      for ( i = 0 ; i < 8 ; i++ ) {
-	 usable[i] = (client_glx_support[i] & client_glx_only[i])
-	     | (client_glx_support[i] & server_support[i]);
+      for (i = 0; i < 8; i++) {
+         usable[i] = (client_glx_support[i] & client_glx_only[i])
+            | (client_glx_support[i] & server_support[i]);
       }
    }
 
-   psc->effectiveGLXexts = __glXGetStringFromTable( known_glx_extensions,
-						    usable );
+   psc->effectiveGLXexts = __glXGetStringFromTable(known_glx_extensions,
+                                                   usable);
 }
 
 
@@ -647,32 +638,33 @@ __glXCalculateUsableExtensions( __GLXscreenConfigs *psc,
  */
 
 void
-__glXCalculateUsableGLExtensions( __GLXcontext * gc,
-				  const char * server_string,
-				  int major_version, int minor_version )
+__glXCalculateUsableGLExtensions(__GLXcontext * gc,
+                                 const char *server_string,
+                                 int major_version, int minor_version)
 {
-   unsigned char server_support[ __GL_EXT_BYTES ];
-   unsigned char usable[ __GL_EXT_BYTES ];
-   unsigned      i;
+   unsigned char server_support[__GL_EXT_BYTES];
+   unsigned char usable[__GL_EXT_BYTES];
+   unsigned i;
 
 
    __glXExtensionsCtr();
 
-   (void) memset( server_support, 0, sizeof( server_support ) );
-   __glXProcessServerString( known_gl_extensions, server_string,
-			     server_support );
+   (void) memset(server_support, 0, sizeof(server_support));
+   __glXProcessServerString(known_gl_extensions, server_string,
+                            server_support);
 
 
    /* Handle lazy servers that don't export all the extensions strings that
     * are part of the GL core version that they support.
     */
 
-   for ( i = 0 ; i < __GL_EXT_BYTES ; i++ ) {
-      if ( (known_gl_extensions[i].version_major != 0)
-	   && ((major_version > known_gl_extensions[i].version_major)
-	       || ((major_version == known_gl_extensions[i].version_major)
-		   && (minor_version >= known_gl_extensions[i].version_minor))) ) {
-	 SET_BIT( server_support, known_gl_extensions[i].bit );
+   for (i = 0; i < __GL_EXT_BYTES; i++) {
+      if ((known_gl_extensions[i].version_major != 0)
+          && ((major_version > known_gl_extensions[i].version_major)
+              || ((major_version == known_gl_extensions[i].version_major)
+                  && (minor_version >=
+                      known_gl_extensions[i].version_minor)))) {
+         SET_BIT(server_support, known_gl_extensions[i].bit);
       }
    }
 
@@ -682,14 +674,14 @@ __glXCalculateUsableGLExtensions( __GLXcontext * gc,
     * and it only needs client-side support.
     */
 
-   for ( i = 0 ; i < __GL_EXT_BYTES ; i++ ) {
+   for (i = 0; i < __GL_EXT_BYTES; i++) {
       usable[i] = (client_gl_support[i] & client_gl_only[i])
-	| (client_gl_support[i] & server_support[i]);
+         | (client_gl_support[i] & server_support[i]);
    }
 
-   gc->extensions = (unsigned char *) 
-     __glXGetStringFromTable( known_gl_extensions, usable );
-   (void) memcpy( gc->gl_extension_bits, usable, sizeof( usable ) );
+   gc->extensions = (unsigned char *)
+      __glXGetStringFromTable(known_gl_extensions, usable);
+   (void) memcpy(gc->gl_extension_bits, usable, sizeof(usable));
 }
 
 
@@ -698,11 +690,11 @@ __glXCalculateUsableGLExtensions( __GLXcontext * gc,
  * rendering.
  */
 void
-__glXGetGLVersion( int * major_version, int * minor_version )
-{		   
-    __glXExtensionsCtr();
-    *major_version = gl_major;
-    *minor_version = gl_minor;
+__glXGetGLVersion(int *major_version, int *minor_version)
+{
+   __glXExtensionsCtr();
+   *major_version = gl_major;
+   *minor_version = gl_minor;
 }
 
 
@@ -712,8 +704,8 @@ __glXGetGLVersion( int * major_version, int * minor_version )
  * supported by the client to the server.
  */
 char *
-__glXGetClientGLExtensionString( void )
+__glXGetClientGLExtensionString(void)
 {
-    __glXExtensionsCtr();
-    return __glXGetStringFromTable( known_gl_extensions, client_gl_support );
+   __glXExtensionsCtr();
+   return __glXGetStringFromTable(known_gl_extensions, client_gl_support);
 }
diff --git a/src/glx/x11/glxextensions.h b/src/glx/x11/glxextensions.h
index a4241b6b7f9..9f1c697487d 100644
--- a/src/glx/x11/glxextensions.h
+++ b/src/glx/x11/glxextensions.h
@@ -31,7 +31,8 @@
 #ifndef GLX_GLXEXTENSIONS_H
 #define GLX_GLXEXTENSIONS_H
 
-enum {
+enum
+{
    ARB_get_proc_address_bit = 0,
    ARB_multisample_bit,
    ARB_render_texture_bit,
@@ -40,7 +41,7 @@ enum {
    EXT_visual_rating_bit,
    EXT_import_context_bit,
    MESA_agp_offset_bit,
-   MESA_allocate_memory_bit, /* Replaces MESA_agp_offset & NV_vertex_array_range */
+   MESA_allocate_memory_bit,    /* Replaces MESA_agp_offset & NV_vertex_array_range */
    MESA_copy_sub_buffer_bit,
    MESA_depth_float_bit,
    MESA_pixmap_colormap_bit,
@@ -67,7 +68,8 @@ enum {
    EXT_texture_from_pixmap_bit
 };
 
-enum {
+enum
+{
    GL_ARB_depth_texture_bit = 0,
    GL_ARB_draw_buffers_bit,
    GL_ARB_fragment_program_bit,
@@ -230,26 +232,34 @@ enum {
 struct __GLXscreenConfigsRec;
 struct __GLXcontextRec;
 
-extern GLboolean __glXExtensionBitIsEnabled( struct __GLXscreenConfigsRec *psc, unsigned bit );
-extern const char * __glXGetClientExtensions( void );
-extern void __glXCalculateUsableExtensions( struct __GLXscreenConfigsRec *psc,
-    GLboolean display_is_direct_capable, int server_minor_version );
-extern void __glXScrEnableExtension( struct __GLXscreenConfigsRec *psc, const char * name );
-extern void __glXCalculateUsableGLExtensions( struct __GLXcontextRec * gc,
-    const char * server_string, int major_version, int minor_version );
-extern void __glXGetGLVersion( int * major_version, int * minor_version );
-extern char * __glXGetClientGLExtensionString( void );
+extern GLboolean __glXExtensionBitIsEnabled(struct __GLXscreenConfigsRec *psc,
+                                            unsigned bit);
+extern const char *__glXGetClientExtensions(void);
+extern void __glXCalculateUsableExtensions(struct __GLXscreenConfigsRec *psc,
+                                           GLboolean
+                                           display_is_direct_capable,
+                                           int server_minor_version);
 
-extern GLboolean __glExtensionBitIsEnabled( const struct __GLXcontextRec * gc, 
-    unsigned bit );
+extern void __glXCalculateUsableGLExtensions(struct __GLXcontextRec *gc,
+                                             const char *server_string,
+                                             int major_version,
+                                             int minor_version);
+extern void __glXGetGLVersion(int *major_version, int *minor_version);
+extern char *__glXGetClientGLExtensionString(void);
 
+extern GLboolean __glExtensionBitIsEnabled(const struct __GLXcontextRec *gc,
+                                           unsigned bit);
+
+extern void
+__glXEnableDirectExtension(struct __GLXscreenConfigsRec *psc,
+                           const char *name);
 
 /* Source-level backwards compatibility with old drivers. They won't
  * find the respective functions, though. 
  */
-typedef void (* PFNGLXENABLEEXTENSIONPROC) ( const char * name,
-    GLboolean force_client );
-typedef void (* PFNGLXDISABLEEXTENSIONPROC) ( const char * name );
+typedef void (*PFNGLXENABLEEXTENSIONPROC) (const char *name,
+                                           GLboolean force_client);
+typedef void (*PFNGLXDISABLEEXTENSIONPROC) (const char *name);
 
 /* GLX_ALIAS should be used for functions with a non-void return type.
    GLX_ALIAS_VOID is for functions with a void return type. */
@@ -259,17 +269,17 @@ typedef void (* PFNGLXDISABLEEXTENSIONPROC) ( const char * name );
 #else
 # if defined(__GNUC__) && !defined(GLX_ALIAS_UNSUPPORTED)
 #  define GLX_ALIAS(return_type, real_func, proto_args, args, aliased_func) \
-	return_type  real_func  proto_args \
-	__attribute__ ((alias( # aliased_func ) ));
+   return_type  real_func  proto_args                                   \
+   __attribute__ ((alias( # aliased_func ) ));
 #  define GLX_ALIAS_VOID(real_func, proto_args, args, aliased_func) \
-	GLX_ALIAS(void, real_func, proto_args, args, aliased_func)
+   GLX_ALIAS(void, real_func, proto_args, args, aliased_func)
 # else
 #  define GLX_ALIAS(return_type, real_func, proto_args, args, aliased_func) \
-	return_type  real_func  proto_args \
-	{ return aliased_func args ; }
+   return_type  real_func  proto_args                                   \
+   { return aliased_func args ; }
 #  define GLX_ALIAS_VOID(real_func, proto_args, args, aliased_func) \
-	void  real_func  proto_args \
-	{ aliased_func args ; }
+   void  real_func  proto_args                                      \
+   { aliased_func args ; }
 # endif /* __GNUC__ */
 #endif /* GLX_NO_STATIC_EXTENSION_FUNCTIONS */
 
diff --git a/src/glx/x11/glxhash.c b/src/glx/x11/glxhash.c
new file mode 100644
index 00000000000..74cd4f344df
--- /dev/null
+++ b/src/glx/x11/glxhash.c
@@ -0,0 +1,461 @@
+/* glxhash.c -- Small hash table support for integer -> integer mapping
+ * Taken from libdrm.
+ *
+ * Created: Sun Apr 18 09:35:45 1999 by faith@precisioninsight.com
+ *
+ * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Rickard E. (Rik) Faith <faith@valinux.com>
+ *
+ * DESCRIPTION
+ *
+ * This file contains a straightforward implementation of a fixed-sized
+ * hash table using self-organizing linked lists [Knuth73, pp. 398-399] for
+ * collision resolution.  There are two potentially interesting things
+ * about this implementation:
+ *
+ * 1) The table is power-of-two sized.  Prime sized tables are more
+ * traditional, but do not have a significant advantage over power-of-two
+ * sized table, especially when double hashing is not used for collision
+ * resolution.
+ *
+ * 2) The hash computation uses a table of random integers [Hanson97,
+ * pp. 39-41].
+ *
+ * FUTURE ENHANCEMENTS
+ *
+ * With a table size of 512, the current implementation is sufficient for a
+ * few hundred keys.  Since this is well above the expected size of the
+ * tables for which this implementation was designed, the implementation of
+ * dynamic hash tables was postponed until the need arises.  A common (and
+ * naive) approach to dynamic hash table implementation simply creates a
+ * new hash table when necessary, rehashes all the data into the new table,
+ * and destroys the old table.  The approach in [Larson88] is superior in
+ * two ways: 1) only a portion of the table is expanded when needed,
+ * distributing the expansion cost over several insertions, and 2) portions
+ * of the table can be locked, enabling a scalable thread-safe
+ * implementation.
+ *
+ * REFERENCES
+ *
+ * [Hanson97] David R. Hanson.  C Interfaces and Implementations:
+ * Techniques for Creating Reusable Software.  Reading, Massachusetts:
+ * Addison-Wesley, 1997.
+ *
+ * [Knuth73] Donald E. Knuth. The Art of Computer Programming.  Volume 3:
+ * Sorting and Searching.  Reading, Massachusetts: Addison-Wesley, 1973.
+ *
+ * [Larson88] Per-Ake Larson. "Dynamic Hash Tables".  CACM 31(4), April
+ * 1988, pp. 446-457.
+ *
+ */
+
+#include "glxhash.h"
+#include <X11/Xfuncproto.h>
+
+#define HASH_MAIN 0
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#define HASH_MAGIC 0xdeadbeef
+#define HASH_DEBUG 0
+#define HASH_SIZE  512          /* Good for about 100 entries */
+                                /* If you change this value, you probably
+                                   have to change the HashHash hashing
+                                   function! */
+
+#define HASH_ALLOC malloc
+#define HASH_FREE  free
+#define HASH_RANDOM_DECL
+#define HASH_RANDOM_INIT(seed)  srandom(seed)
+#define HASH_RANDOM             random()
+#define HASH_RANDOM_DESTROY
+
+typedef struct __glxHashBucket
+{
+   unsigned long key;
+   void *value;
+   struct __glxHashBucket *next;
+} __glxHashBucket, *__glxHashBucketPtr;
+
+typedef struct __glxHashTable *__glxHashTablePtr;
+struct __glxHashTable
+{
+   unsigned long magic;
+   unsigned long hits;          /* At top of linked list */
+   unsigned long partials;      /* Not at top of linked list */
+   unsigned long misses;        /* Not in table */
+   __glxHashBucketPtr buckets[HASH_SIZE];
+   int p0;
+   __glxHashBucketPtr p1;
+};
+
+static unsigned long
+HashHash(unsigned long key)
+{
+   unsigned long hash = 0;
+   unsigned long tmp = key;
+   static int init = 0;
+   static unsigned long scatter[256];
+   int i;
+
+   if (!init) {
+      HASH_RANDOM_DECL;
+      HASH_RANDOM_INIT(37);
+      for (i = 0; i < 256; i++)
+         scatter[i] = HASH_RANDOM;
+      HASH_RANDOM_DESTROY;
+      ++init;
+   }
+
+   while (tmp) {
+      hash = (hash << 1) + scatter[tmp & 0xff];
+      tmp >>= 8;
+   }
+
+   hash %= HASH_SIZE;
+#if HASH_DEBUG
+   printf("Hash(%d) = %d\n", key, hash);
+#endif
+   return hash;
+}
+
+_X_HIDDEN __glxHashTable *
+__glxHashCreate(void)
+{
+   __glxHashTablePtr table;
+   int i;
+
+   table = HASH_ALLOC(sizeof(*table));
+   if (!table)
+      return NULL;
+   table->magic = HASH_MAGIC;
+   table->hits = 0;
+   table->partials = 0;
+   table->misses = 0;
+
+   for (i = 0; i < HASH_SIZE; i++)
+      table->buckets[i] = NULL;
+   return table;
+}
+
+_X_HIDDEN int
+__glxHashDestroy(__glxHashTable * t)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+   __glxHashBucketPtr bucket;
+   __glxHashBucketPtr next;
+   int i;
+
+   if (table->magic != HASH_MAGIC)
+      return -1;                /* Bad magic */
+
+   for (i = 0; i < HASH_SIZE; i++) {
+      for (bucket = table->buckets[i]; bucket;) {
+         next = bucket->next;
+         HASH_FREE(bucket);
+         bucket = next;
+      }
+   }
+   HASH_FREE(table);
+   return 0;
+}
+
+/* Find the bucket and organize the list so that this bucket is at the
+   top. */
+
+static __glxHashBucketPtr
+HashFind(__glxHashTablePtr table, unsigned long key, unsigned long *h)
+{
+   unsigned long hash = HashHash(key);
+   __glxHashBucketPtr prev = NULL;
+   __glxHashBucketPtr bucket;
+
+   if (h)
+      *h = hash;
+
+   for (bucket = table->buckets[hash]; bucket; bucket = bucket->next) {
+      if (bucket->key == key) {
+         if (prev) {
+            /* Organize */
+            prev->next = bucket->next;
+            bucket->next = table->buckets[hash];
+            table->buckets[hash] = bucket;
+            ++table->partials;
+         }
+         else {
+            ++table->hits;
+         }
+         return bucket;
+      }
+      prev = bucket;
+   }
+   ++table->misses;
+   return NULL;
+}
+
+_X_HIDDEN int
+__glxHashLookup(__glxHashTable * t, unsigned long key, void **value)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+   __glxHashBucketPtr bucket;
+
+   if (!table || table->magic != HASH_MAGIC)
+      return -1;                /* Bad magic */
+
+   bucket = HashFind(table, key, NULL);
+   if (!bucket)
+      return 1;                 /* Not found */
+   *value = bucket->value;
+   return 0;                    /* Found */
+}
+
+_X_HIDDEN int
+__glxHashInsert(__glxHashTable * t, unsigned long key, void *value)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+   __glxHashBucketPtr bucket;
+   unsigned long hash;
+
+   if (table->magic != HASH_MAGIC)
+      return -1;                /* Bad magic */
+
+   if (HashFind(table, key, &hash))
+      return 1;                 /* Already in table */
+
+   bucket = HASH_ALLOC(sizeof(*bucket));
+   if (!bucket)
+      return -1;                /* Error */
+   bucket->key = key;
+   bucket->value = value;
+   bucket->next = table->buckets[hash];
+   table->buckets[hash] = bucket;
+#if HASH_DEBUG
+   printf("Inserted %d at %d/%p\n", key, hash, bucket);
+#endif
+   return 0;                    /* Added to table */
+}
+
+_X_HIDDEN int
+__glxHashDelete(__glxHashTable * t, unsigned long key)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+   unsigned long hash;
+   __glxHashBucketPtr bucket;
+
+   if (table->magic != HASH_MAGIC)
+      return -1;                /* Bad magic */
+
+   bucket = HashFind(table, key, &hash);
+
+   if (!bucket)
+      return 1;                 /* Not found */
+
+   table->buckets[hash] = bucket->next;
+   HASH_FREE(bucket);
+   return 0;
+}
+
+_X_HIDDEN int
+__glxHashNext(__glxHashTable * t, unsigned long *key, void **value)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+
+   while (table->p0 < HASH_SIZE) {
+      if (table->p1) {
+         *key = table->p1->key;
+         *value = table->p1->value;
+         table->p1 = table->p1->next;
+         return 1;
+      }
+      table->p1 = table->buckets[table->p0];
+      ++table->p0;
+   }
+   return 0;
+}
+
+_X_HIDDEN int
+__glxHashFirst(__glxHashTable * t, unsigned long *key, void **value)
+{
+   __glxHashTablePtr table = (__glxHashTablePtr) t;
+
+   if (table->magic != HASH_MAGIC)
+      return -1;                /* Bad magic */
+
+   table->p0 = 0;
+   table->p1 = table->buckets[0];
+   return __glxHashNext(table, key, value);
+}
+
+#if HASH_MAIN
+#define DIST_LIMIT 10
+static int dist[DIST_LIMIT];
+
+static void
+clear_dist(void)
+{
+   int i;
+
+   for (i = 0; i < DIST_LIMIT; i++)
+      dist[i] = 0;
+}
+
+static int
+count_entries(__glxHashBucketPtr bucket)
+{
+   int count = 0;
+
+   for (; bucket; bucket = bucket->next)
+      ++count;
+   return count;
+}
+
+static void
+update_dist(int count)
+{
+   if (count >= DIST_LIMIT)
+      ++dist[DIST_LIMIT - 1];
+   else
+      ++dist[count];
+}
+
+static void
+compute_dist(__glxHashTablePtr table)
+{
+   int i;
+   __glxHashBucketPtr bucket;
+
+   printf("Hits = %ld, partials = %ld, misses = %ld\n",
+          table->hits, table->partials, table->misses);
+   clear_dist();
+   for (i = 0; i < HASH_SIZE; i++) {
+      bucket = table->buckets[i];
+      update_dist(count_entries(bucket));
+   }
+   for (i = 0; i < DIST_LIMIT; i++) {
+      if (i != DIST_LIMIT - 1)
+         printf("%5d %10d\n", i, dist[i]);
+      else
+         printf("other %10d\n", dist[i]);
+   }
+}
+
+static void
+check_table(__glxHashTablePtr table, unsigned long key, unsigned long value)
+{
+   unsigned long retval = 0;
+   int retcode = __glxHashLookup(table, key, &retval);
+
+   switch (retcode) {
+   case -1:
+      printf("Bad magic = 0x%08lx:"
+             " key = %lu, expected = %lu, returned = %lu\n",
+             table->magic, key, value, retval);
+      break;
+   case 1:
+      printf("Not found: key = %lu, expected = %lu returned = %lu\n",
+             key, value, retval);
+      break;
+   case 0:
+      if (value != retval)
+         printf("Bad value: key = %lu, expected = %lu, returned = %lu\n",
+                key, value, retval);
+      break;
+   default:
+      printf("Bad retcode = %d: key = %lu, expected = %lu, returned = %lu\n",
+             retcode, key, value, retval);
+      break;
+   }
+}
+
+int
+main(void)
+{
+   __glxHashTablePtr table;
+   int i;
+
+   printf("\n***** 256 consecutive integers ****\n");
+   table = __glxHashCreate();
+   for (i = 0; i < 256; i++)
+      __glxHashInsert(table, i, i);
+   for (i = 0; i < 256; i++)
+      check_table(table, i, i);
+   for (i = 256; i >= 0; i--)
+      check_table(table, i, i);
+   compute_dist(table);
+   __glxHashDestroy(table);
+
+   printf("\n***** 1024 consecutive integers ****\n");
+   table = __glxHashCreate();
+   for (i = 0; i < 1024; i++)
+      __glxHashInsert(table, i, i);
+   for (i = 0; i < 1024; i++)
+      check_table(table, i, i);
+   for (i = 1024; i >= 0; i--)
+      check_table(table, i, i);
+   compute_dist(table);
+   __glxHashDestroy(table);
+
+   printf("\n***** 1024 consecutive page addresses (4k pages) ****\n");
+   table = __glxHashCreate();
+   for (i = 0; i < 1024; i++)
+      __glxHashInsert(table, i * 4096, i);
+   for (i = 0; i < 1024; i++)
+      check_table(table, i * 4096, i);
+   for (i = 1024; i >= 0; i--)
+      check_table(table, i * 4096, i);
+   compute_dist(table);
+   __glxHashDestroy(table);
+
+   printf("\n***** 1024 random integers ****\n");
+   table = __glxHashCreate();
+   srandom(0xbeefbeef);
+   for (i = 0; i < 1024; i++)
+      __glxHashInsert(table, random(), i);
+   srandom(0xbeefbeef);
+   for (i = 0; i < 1024; i++)
+      check_table(table, random(), i);
+   srandom(0xbeefbeef);
+   for (i = 0; i < 1024; i++)
+      check_table(table, random(), i);
+   compute_dist(table);
+   __glxHashDestroy(table);
+
+   printf("\n***** 5000 random integers ****\n");
+   table = __glxHashCreate();
+   srandom(0xbeefbeef);
+   for (i = 0; i < 5000; i++)
+      __glxHashInsert(table, random(), i);
+   srandom(0xbeefbeef);
+   for (i = 0; i < 5000; i++)
+      check_table(table, random(), i);
+   srandom(0xbeefbeef);
+   for (i = 0; i < 5000; i++)
+      check_table(table, random(), i);
+   compute_dist(table);
+   __glxHashDestroy(table);
+
+   return 0;
+}
+#endif
diff --git a/src/glx/x11/glxhash.h b/src/glx/x11/glxhash.h
new file mode 100644
index 00000000000..710712dd28a
--- /dev/null
+++ b/src/glx/x11/glxhash.h
@@ -0,0 +1,20 @@
+#ifndef _GLX_HASH_H_
+#define _GLX_HASH_H_
+
+
+typedef struct __glxHashTable __glxHashTable;
+
+/* Hash table routines */
+extern __glxHashTable *__glxHashCreate(void);
+extern int __glxHashDestroy(__glxHashTable * t);
+extern int __glxHashLookup(__glxHashTable * t, unsigned long key,
+                           void **value);
+extern int __glxHashInsert(__glxHashTable * t, unsigned long key,
+                           void *value);
+extern int __glxHashDelete(__glxHashTable * t, unsigned long key);
+extern int __glxHashFirst(__glxHashTable * t, unsigned long *key,
+                          void **value);
+extern int __glxHashNext(__glxHashTable * t, unsigned long *key,
+                         void **value);
+
+#endif /* _GLX_HASH_H_ */
diff --git a/src/glx/x11/indirect.c b/src/glx/x11/indirect.c
index fbb2a91956c..fb87abc0a5a 100644
--- a/src/glx/x11/indirect.c
+++ b/src/glx/x11/indirect.c
@@ -300,7 +300,9 @@ __indirect_glNewList(GLuint list, GLenum mode)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -324,7 +326,9 @@ __indirect_glEndList(void)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 0;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -360,6 +364,10 @@ __indirect_glCallLists(GLsizei n, GLenum type, const GLvoid * lists)
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint compsize = __glCallLists_size(type);
     const GLuint cmdlen = 12 + __GLX_PAD((compsize * n));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -393,7 +401,9 @@ __indirect_glDeleteLists(GLuint list, GLsizei range)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -418,7 +428,9 @@ __indirect_glGenLists(GLsizei range)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     GLuint retval = (GLuint) 0;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3582,6 +3594,10 @@ __indirect_glPixelMapfv(GLenum map, GLsizei mapsize, const GLfloat * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((mapsize * 4));
+    if (mapsize < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((mapsize >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -3615,6 +3631,10 @@ __indirect_glPixelMapuiv(GLenum map, GLsizei mapsize, const GLuint * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((mapsize * 4));
+    if (mapsize < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((mapsize >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -3648,6 +3668,10 @@ __indirect_glPixelMapusv(GLenum map, GLsizei mapsize, const GLushort * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((mapsize * 2));
+    if (mapsize < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((mapsize >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -3716,7 +3740,9 @@ __indirect_glReadPixels(GLint x, GLint y, GLsizei width, GLsizei height,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 28;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3808,7 +3834,9 @@ __indirect_glGetClipPlane(GLenum plane, GLdouble * equation)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3841,7 +3869,9 @@ __indirect_glGetLightfv(GLenum light, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3879,7 +3909,9 @@ __indirect_glGetLightiv(GLenum light, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3917,7 +3949,9 @@ __indirect_glGetMapdv(GLenum target, GLenum query, GLdouble * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3953,7 +3987,9 @@ __indirect_glGetMapfv(GLenum target, GLenum query, GLfloat * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -3989,7 +4025,9 @@ __indirect_glGetMapiv(GLenum target, GLenum query, GLint * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4025,7 +4063,9 @@ __indirect_glGetMaterialfv(GLenum face, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4063,7 +4103,9 @@ __indirect_glGetMaterialiv(GLenum face, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4101,7 +4143,9 @@ __indirect_glGetPixelMapfv(GLenum map, GLfloat * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4137,7 +4181,9 @@ __indirect_glGetPixelMapuiv(GLenum map, GLuint * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4174,7 +4220,9 @@ __indirect_glGetPixelMapusv(GLenum map, GLushort * values)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4210,9 +4258,10 @@ void
 __indirect_glGetPolygonStipple(GLubyte *mask)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
-    const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4247,7 +4296,9 @@ __indirect_glGetTexEnvfv(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4285,7 +4336,9 @@ __indirect_glGetTexEnviv(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4323,7 +4376,9 @@ __indirect_glGetTexGendv(GLenum coord, GLenum pname, GLdouble * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4361,7 +4416,9 @@ __indirect_glGetTexGenfv(GLenum coord, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4399,7 +4456,9 @@ __indirect_glGetTexGeniv(GLenum coord, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4439,7 +4498,9 @@ __indirect_glGetTexImage(GLenum target, GLint level, GLenum format,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 20;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4483,7 +4544,9 @@ __indirect_glGetTexParameterfv(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4522,7 +4585,9 @@ __indirect_glGetTexParameteriv(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4562,7 +4627,9 @@ __indirect_glGetTexLevelParameterfv(GLenum target, GLint level, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 12;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4603,7 +4670,9 @@ __indirect_glGetTexLevelParameteriv(GLenum target, GLint level, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 12;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -4644,7 +4713,9 @@ __indirect_glIsList(GLuint list)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     GLboolean retval = (GLboolean) 0;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5004,72 +5075,6 @@ __indirect_glPolygonOffset(GLfloat factor, GLfloat units)
     }
 }
 
-#define X_GLsop_AreTexturesResident 143
-GLboolean
-__indirect_glAreTexturesResident(GLsizei n, const GLuint * textures,
-                                 GLboolean * residences)
-{
-    __GLXcontext *const gc = __glXGetCurrentContext();
-    Display *const dpy = gc->currentDpy;
-    GLboolean retval = (GLboolean) 0;
-    const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
-    if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
-#ifdef USE_XCB
-        xcb_connection_t *c = XGetXCBConnection(dpy);
-        (void) __glXFlushRenderBuffer(gc, gc->pc);
-        xcb_glx_are_textures_resident_reply_t *reply =
-            xcb_glx_are_textures_resident_reply(c,
-                                                xcb_glx_are_textures_resident
-                                                (c, gc->currentContextTag, n,
-                                                 textures), NULL);
-        (void) memcpy(residences, xcb_glx_are_textures_resident_data(reply),
-                      xcb_glx_are_textures_resident_data_length(reply) *
-                      sizeof(GLboolean));
-        retval = reply->ret_val;
-        free(reply);
-#else
-        GLubyte const *pc =
-            __glXSetupSingleRequest(gc, X_GLsop_AreTexturesResident, cmdlen);
-        (void) memcpy((void *) (pc + 0), (void *) (&n), 4);
-        (void) memcpy((void *) (pc + 4), (void *) (textures), (n * 4));
-        retval = (GLboolean) __glXReadReply(dpy, 1, residences, GL_TRUE);
-        UnlockDisplay(dpy);
-        SyncHandle();
-#endif /* USE_XCB */
-    }
-    return retval;
-}
-
-#define X_GLvop_AreTexturesResidentEXT 11
-GLboolean
-glAreTexturesResidentEXT(GLsizei n, const GLuint * textures,
-                         GLboolean * residences)
-{
-    __GLXcontext *const gc = __glXGetCurrentContext();
-
-    if (gc->isDirect) {
-        return CALL_AreTexturesResident(GET_DISPATCH(),
-                                        (n, textures, residences));
-    } else {
-        __GLXcontext *const gc = __glXGetCurrentContext();
-        Display *const dpy = gc->currentDpy;
-        GLboolean retval = (GLboolean) 0;
-        const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
-        if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
-            GLubyte const *pc =
-                __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
-                                        X_GLvop_AreTexturesResidentEXT,
-                                        cmdlen);
-            (void) memcpy((void *) (pc + 0), (void *) (&n), 4);
-            (void) memcpy((void *) (pc + 4), (void *) (textures), (n * 4));
-            retval = (GLboolean) __glXReadReply(dpy, 1, residences, GL_TRUE);
-            UnlockDisplay(dpy);
-            SyncHandle();
-        }
-        return retval;
-    }
-}
-
 #define X_GLrop_CopyTexImage1D 4119
 void
 __indirect_glCopyTexImage1D(GLenum target, GLint level, GLenum internalformat,
@@ -5163,7 +5168,13 @@ __indirect_glDeleteTextures(GLsizei n, const GLuint * textures)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+#endif
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5187,12 +5198,16 @@ glDeleteTexturesEXT(GLsizei n, const GLuint * textures)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_DeleteTextures(GET_DISPATCH(), (n, textures));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
         Display *const dpy = gc->currentDpy;
         const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+        if (n < 0) {
+            __glXSetError(gc, GL_INVALID_VALUE);
+            return;
+        }
         if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
             GLubyte const *pc =
                 __glXSetupVendorRequest(gc, X_GLXVendorPrivate,
@@ -5212,7 +5227,13 @@ __indirect_glGenTextures(GLsizei n, GLuint * textures)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5245,12 +5266,16 @@ glGenTexturesEXT(GLsizei n, GLuint * textures)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GenTextures(GET_DISPATCH(), (n, textures));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
         Display *const dpy = gc->currentDpy;
         const GLuint cmdlen = 4;
+        if (n < 0) {
+            __glXSetError(gc, GL_INVALID_VALUE);
+            return;
+        }
         if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
             GLubyte const *pc =
                 __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -5271,7 +5296,9 @@ __indirect_glIsTexture(GLuint texture)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     GLboolean retval = (GLboolean) 0;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5301,7 +5328,7 @@ glIsTextureEXT(GLuint texture)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         return CALL_IsTexture(GET_DISPATCH(), (texture));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -5328,6 +5355,10 @@ __indirect_glPrioritizeTextures(GLsizei n, const GLuint * textures,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 8 + __GLX_PAD((n * 4)) + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_PrioritizeTextures, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&n), 4);
@@ -5366,8 +5397,7 @@ __glx_TexSubImage_1D2D(unsigned opcode, unsigned dim, GLenum target,
             (void) memcpy((void *) (gc->pc + 44), (void *) (&height), 4);
             (void) memcpy((void *) (gc->pc + 48), (void *) (&format), 4);
             (void) memcpy((void *) (gc->pc + 52), (void *) (&type), 4);
-            (void) memcpy((void *) (gc->pc + 56),
-                          (void *) ((pixels == NULL) ? one : zero), 4);
+            (void) memset((void *) (gc->pc + 56), 0, 4);
             if (compsize > 0) {
                 (*gc->fillImage) (gc, dim, width, height, 1, format, type,
                                   pixels, gc->pc + 60, gc->pc + 4);
@@ -5393,7 +5423,7 @@ __glx_TexSubImage_1D2D(unsigned opcode, unsigned dim, GLenum target,
             (void) memcpy((void *) (pc + 48), (void *) (&height), 4);
             (void) memcpy((void *) (pc + 52), (void *) (&format), 4);
             (void) memcpy((void *) (pc + 56), (void *) (&type), 4);
-            (void) memcpy((void *) (pc + 60), zero, 4);
+            (void) memset((void *) (pc + 60), 0, 4);
             __glXSendLargeImage(gc, compsize, dim, width, height, 1, format,
                                 type, pixels, pc + 64, pc + 8);
         }
@@ -5565,7 +5595,9 @@ __indirect_glGetColorTable(GLenum target, GLenum format, GLenum type,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 16;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5609,7 +5641,7 @@ glGetColorTableEXT(GLenum target, GLenum format, GLenum type, GLvoid * table)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetColorTable(GET_DISPATCH(), (target, format, type, table));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -5641,7 +5673,9 @@ __indirect_glGetColorTableParameterfv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5680,7 +5714,7 @@ glGetColorTableParameterfvEXT(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetColorTableParameterfv(GET_DISPATCH(),
                                       (target, pname, params));
     } else {
@@ -5709,7 +5743,9 @@ __indirect_glGetColorTableParameteriv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -5748,7 +5784,7 @@ glGetColorTableParameterivEXT(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetColorTableParameteriv(GET_DISPATCH(),
                                       (target, pname, params));
     } else {
@@ -6029,7 +6065,9 @@ __indirect_glGetConvolutionFilter(GLenum target, GLenum format, GLenum type,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 16;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6069,7 +6107,7 @@ gl_dispatch_stub_356(GLenum target, GLenum format, GLenum type,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetConvolutionFilter(GET_DISPATCH(),
                                   (target, format, type, image));
     } else {
@@ -6103,7 +6141,9 @@ __indirect_glGetConvolutionParameterfv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6142,7 +6182,7 @@ gl_dispatch_stub_357(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetConvolutionParameterfv(GET_DISPATCH(),
                                        (target, pname, params));
     } else {
@@ -6171,7 +6211,9 @@ __indirect_glGetConvolutionParameteriv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6210,7 +6252,7 @@ gl_dispatch_stub_358(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetConvolutionParameteriv(GET_DISPATCH(),
                                        (target, pname, params));
     } else {
@@ -6240,7 +6282,9 @@ __indirect_glGetHistogram(GLenum target, GLboolean reset, GLenum format,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 16;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6285,7 +6329,7 @@ gl_dispatch_stub_361(GLenum target, GLboolean reset, GLenum format,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetHistogram(GET_DISPATCH(),
                           (target, reset, format, type, values));
     } else {
@@ -6319,7 +6363,9 @@ __indirect_glGetHistogramParameterfv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6357,7 +6403,7 @@ gl_dispatch_stub_362(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetHistogramParameterfv(GET_DISPATCH(), (target, pname, params));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -6385,7 +6431,9 @@ __indirect_glGetHistogramParameteriv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6423,7 +6471,7 @@ gl_dispatch_stub_363(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetHistogramParameteriv(GET_DISPATCH(), (target, pname, params));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -6452,7 +6500,9 @@ __indirect_glGetMinmax(GLenum target, GLboolean reset, GLenum format,
     __GLXcontext *const gc = __glXGetCurrentContext();
     const __GLXattribute *const state = gc->client_state_private;
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 16;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6493,7 +6543,7 @@ gl_dispatch_stub_364(GLenum target, GLboolean reset, GLenum format,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetMinmax(GET_DISPATCH(), (target, reset, format, type, values));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -6526,7 +6576,9 @@ __indirect_glGetMinmaxParameterfv(GLenum target, GLenum pname,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6562,7 +6614,7 @@ gl_dispatch_stub_365(GLenum target, GLenum pname, GLfloat * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetMinmaxParameterfv(GET_DISPATCH(), (target, pname, params));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -6589,7 +6641,9 @@ __indirect_glGetMinmaxParameteriv(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -6625,7 +6679,7 @@ gl_dispatch_stub_366(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
 
-    if (gc->isDirect) {
+    if (gc->driContext) {
         CALL_GetMinmaxParameteriv(GET_DISPATCH(), (target, pname, params));
     } else {
         __GLXcontext *const gc = __glXGetCurrentContext();
@@ -6814,8 +6868,7 @@ __glx_TexSubImage_3D4D(unsigned opcode, unsigned dim, GLenum target,
             (void) memcpy((void *) (gc->pc + 76), (void *) (&extent), 4);
             (void) memcpy((void *) (gc->pc + 80), (void *) (&format), 4);
             (void) memcpy((void *) (gc->pc + 84), (void *) (&type), 4);
-            (void) memcpy((void *) (gc->pc + 88),
-                          (void *) ((pixels == NULL) ? one : zero), 4);
+            (void) memset((void *) (gc->pc + 88), 0, 4);
             if (compsize > 0) {
                 (*gc->fillImage) (gc, dim, width, height, depth, format, type,
                                   pixels, gc->pc + 92, gc->pc + 4);
@@ -6845,7 +6898,7 @@ __glx_TexSubImage_3D4D(unsigned opcode, unsigned dim, GLenum target,
             (void) memcpy((void *) (pc + 80), (void *) (&extent), 4);
             (void) memcpy((void *) (pc + 84), (void *) (&format), 4);
             (void) memcpy((void *) (pc + 88), (void *) (&type), 4);
-            (void) memcpy((void *) (pc + 92), zero, 4);
+            (void) memset((void *) (pc + 92), 0, 4);
             __glXSendLargeImage(gc, compsize, dim, width, height, depth,
                                 format, type, pixels, pc + 96, pc + 8);
         }
@@ -7585,6 +7638,10 @@ __indirect_glProgramStringARB(GLenum target, GLenum format, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 16 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((len >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -8200,7 +8257,13 @@ __indirect_glDeleteQueriesARB(GLsizei n, const GLuint * ids)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+#endif
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8238,7 +8301,13 @@ __indirect_glGenQueriesARB(GLsizei n, GLuint * ids)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8271,7 +8340,9 @@ __indirect_glGetQueryObjectivARB(GLuint id, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8307,7 +8378,9 @@ __indirect_glGetQueryObjectuivARB(GLuint id, GLenum pname, GLuint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8343,7 +8416,9 @@ __indirect_glGetQueryivARB(GLenum target, GLenum pname, GLint * params)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
+#ifndef USE_XCB
     const GLuint cmdlen = 8;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8383,7 +8458,9 @@ __indirect_glIsQueryARB(GLuint id)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     GLboolean retval = (GLboolean) 0;
+#ifndef USE_XCB
     const GLuint cmdlen = 4;
+#endif
     if (__builtin_expect(dpy != NULL, 1)) {
 #ifdef USE_XCB
         xcb_connection_t *c = XGetXCBConnection(dpy);
@@ -8414,6 +8491,10 @@ __indirect_glDrawBuffersARB(GLsizei n, const GLenum * bufs)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 8 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (gc->currentDpy != NULL), 1)) {
         if (cmdlen <= gc->maxSmallRenderCommandSize) {
             if ((gc->pc + cmdlen) > gc->bufEnd) {
@@ -8774,6 +8855,10 @@ __indirect_glAreProgramsResidentNV(GLsizei n, const GLuint * ids,
     Display *const dpy = gc->currentDpy;
     GLboolean retval = (GLboolean) 0;
     const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return 0;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -8809,6 +8894,10 @@ __indirect_glDeleteProgramsNV(GLsizei n, const GLuint * programs)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivate,
@@ -8845,6 +8934,10 @@ __indirect_glGenProgramsNV(GLsizei n, GLuint * programs)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 4;
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -9051,6 +9144,10 @@ __indirect_glLoadProgramNV(GLenum target, GLuint id, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 16 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(len >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_LoadProgramNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4);
@@ -9071,6 +9168,10 @@ __indirect_glProgramParameters4dvNV(GLenum target, GLuint index, GLuint num,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 16 + __GLX_PAD((num * 32));
+    if (num < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(num >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramParameters4dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4);
@@ -9091,6 +9192,10 @@ __indirect_glProgramParameters4fvNV(GLenum target, GLuint index, GLuint num,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 16 + __GLX_PAD((num * 16));
+    if (num < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(num >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramParameters4fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&target), 4);
@@ -9110,6 +9215,10 @@ __indirect_glRequestResidentProgramsNV(GLsizei n, const GLuint * ids)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 8 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_RequestResidentProgramsNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&n), 4);
@@ -9561,6 +9670,10 @@ __indirect_glVertexAttribs1dvNV(GLuint index, GLsizei n, const GLdouble * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 8));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs1dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9579,6 +9692,10 @@ __indirect_glVertexAttribs1fvNV(GLuint index, GLsizei n, const GLfloat * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs1fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9597,6 +9714,10 @@ __indirect_glVertexAttribs1svNV(GLuint index, GLsizei n, const GLshort * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 2));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs1svNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9615,6 +9736,10 @@ __indirect_glVertexAttribs2dvNV(GLuint index, GLsizei n, const GLdouble * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 16));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs2dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9633,6 +9758,10 @@ __indirect_glVertexAttribs2fvNV(GLuint index, GLsizei n, const GLfloat * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 8));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs2fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9651,6 +9780,10 @@ __indirect_glVertexAttribs2svNV(GLuint index, GLsizei n, const GLshort * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs2svNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9669,6 +9802,10 @@ __indirect_glVertexAttribs3dvNV(GLuint index, GLsizei n, const GLdouble * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 24));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs3dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9687,6 +9824,10 @@ __indirect_glVertexAttribs3fvNV(GLuint index, GLsizei n, const GLfloat * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 12));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs3fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9705,6 +9846,10 @@ __indirect_glVertexAttribs3svNV(GLuint index, GLsizei n, const GLshort * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 6));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs3svNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9723,6 +9868,10 @@ __indirect_glVertexAttribs4dvNV(GLuint index, GLsizei n, const GLdouble * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 32));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs4dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9741,6 +9890,10 @@ __indirect_glVertexAttribs4fvNV(GLuint index, GLsizei n, const GLfloat * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 16));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs4fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9759,6 +9912,10 @@ __indirect_glVertexAttribs4svNV(GLuint index, GLsizei n, const GLshort * v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 8));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs4svNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9777,6 +9934,10 @@ __indirect_glVertexAttribs4ubvNV(GLuint index, GLsizei n, const GLubyte *v)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 12 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_VertexAttribs4ubvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&index), 4);
@@ -9843,6 +10004,10 @@ __indirect_glGetProgramNamedParameterdvNV(GLuint id, GLsizei len,
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 8 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((len >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -9867,6 +10032,10 @@ __indirect_glGetProgramNamedParameterfvNV(GLuint id, GLsizei len,
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 8 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((len >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -9890,6 +10059,10 @@ __indirect_glProgramNamedParameter4dNV(GLuint id, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 44 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(len >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramNamedParameter4dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&x), 8);
@@ -9914,6 +10087,10 @@ __indirect_glProgramNamedParameter4dvNV(GLuint id, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 44 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(len >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramNamedParameter4dvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (v), 32);
@@ -9935,6 +10112,10 @@ __indirect_glProgramNamedParameter4fNV(GLuint id, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 28 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(len >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramNamedParameter4fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&id), 4);
@@ -9959,6 +10140,10 @@ __indirect_glProgramNamedParameter4fvNV(GLuint id, GLsizei len,
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 28 + __GLX_PAD(len);
+    if (len < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(len >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_ProgramNamedParameter4fvNV, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&id), 4);
@@ -10044,6 +10229,10 @@ __indirect_glDeleteFramebuffersEXT(GLsizei n, const GLuint * framebuffers)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 8 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_DeleteFramebuffersEXT, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&n), 4);
@@ -10062,6 +10251,10 @@ __indirect_glDeleteRenderbuffersEXT(GLsizei n, const GLuint * renderbuffers)
 {
     __GLXcontext *const gc = __glXGetCurrentContext();
     const GLuint cmdlen = 8 + __GLX_PAD((n * 4));
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect(n >= 0, 1)) {
         emit_header(gc->pc, X_GLrop_DeleteRenderbuffersEXT, cmdlen);
         (void) memcpy((void *) (gc->pc + 4), (void *) (&n), 4);
@@ -10161,6 +10354,10 @@ __indirect_glGenFramebuffersEXT(GLsizei n, GLuint * framebuffers)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 4;
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
@@ -10180,6 +10377,10 @@ __indirect_glGenRenderbuffersEXT(GLsizei n, GLuint * renderbuffers)
     __GLXcontext *const gc = __glXGetCurrentContext();
     Display *const dpy = gc->currentDpy;
     const GLuint cmdlen = 4;
+    if (n < 0) {
+        __glXSetError(gc, GL_INVALID_VALUE);
+        return;
+    }
     if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
         GLubyte const *pc =
             __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
diff --git a/src/glx/x11/indirect_texture_compression.c b/src/glx/x11/indirect_texture_compression.c
new file mode 100644
index 00000000000..fa927ebdf68
--- /dev/null
+++ b/src/glx/x11/indirect_texture_compression.c
@@ -0,0 +1,343 @@
+/*
+ * (C) Copyright IBM Corporation 2004
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file glx_texture_compression.c
+ * Contains the routines required to implement GLX protocol for
+ * ARB_texture_compression and related extensions.
+ *
+ * \sa http://oss.sgi.com/projects/ogl-sample/registry/ARB/texture_compression.txt
+ *
+ * \author Ian Romanick <idr@us.ibm.com>
+ */
+
+#include "packrender.h"
+#include "packsingle.h"
+#include "indirect.h"
+
+#include <assert.h>
+
+
+void
+__indirect_glGetCompressedTexImageARB(GLenum target, GLint level,
+                                      GLvoid * img)
+{
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXGetTexImageReply reply;
+   size_t image_bytes;
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetCompressedTexImage, 8);
+   __GLX_SINGLE_PUT_LONG(0, target);
+   __GLX_SINGLE_PUT_LONG(4, level);
+   __GLX_SINGLE_READ_XREPLY();
+
+   image_bytes = reply.width;
+   assert(image_bytes <= ((4 * reply.length) - 0));
+   assert(image_bytes >= ((4 * reply.length) - 3));
+
+   if (image_bytes != 0) {
+      _XRead(dpy, (char *) img, image_bytes);
+      if (image_bytes < (4 * reply.length)) {
+         _XEatData(dpy, (4 * reply.length) - image_bytes);
+      }
+   }
+
+   __GLX_SINGLE_END();
+}
+
+
+/**
+ * Internal function used for \c glCompressedTexImage1D and
+ * \c glCompressedTexImage2D.
+ */
+static void
+CompressedTexImage1D2D(GLenum target, GLint level,
+                       GLenum internal_format,
+                       GLsizei width, GLsizei height,
+                       GLint border, GLsizei image_size,
+                       const GLvoid * data, CARD32 rop)
+{
+   __GLX_DECLARE_VARIABLES();
+
+   __GLX_LOAD_VARIABLES();
+   if (gc->currentDpy == NULL) {
+      return;
+   }
+
+   if ((target == GL_PROXY_TEXTURE_1D)
+       || (target == GL_PROXY_TEXTURE_2D)
+       || (target == GL_PROXY_TEXTURE_CUBE_MAP)) {
+      compsize = 0;
+   }
+   else {
+      compsize = image_size;
+   }
+
+   cmdlen = __GLX_PAD(__GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE + compsize);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      __GLX_BEGIN_VARIABLE(rop, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_LONG(8, level);
+      __GLX_PUT_LONG(12, internal_format);
+      __GLX_PUT_LONG(16, width);
+      __GLX_PUT_LONG(20, height);
+      __GLX_PUT_LONG(24, border);
+      __GLX_PUT_LONG(28, image_size);
+      if (compsize != 0) {
+         __GLX_PUT_CHAR_ARRAY(__GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE,
+                              data, image_size);
+      }
+      __GLX_END(cmdlen);
+   }
+   else {
+      assert(compsize != 0);
+
+      __GLX_BEGIN_VARIABLE_LARGE(rop, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_LONG(12, level);
+      __GLX_PUT_LONG(16, internal_format);
+      __GLX_PUT_LONG(20, width);
+      __GLX_PUT_LONG(24, height);
+      __GLX_PUT_LONG(28, border);
+      __GLX_PUT_LONG(32, image_size);
+      __glXSendLargeCommand(gc, gc->pc,
+                            __GLX_COMPRESSED_TEXIMAGE_CMD_HDR_SIZE + 4,
+                            data, image_size);
+   }
+}
+
+
+/**
+ * Internal function used for \c glCompressedTexSubImage1D and
+ * \c glCompressedTexSubImage2D.
+ */
+static void
+CompressedTexSubImage1D2D(GLenum target, GLint level,
+                          GLsizei xoffset, GLsizei yoffset,
+                          GLsizei width, GLsizei height,
+                          GLenum format, GLsizei image_size,
+                          const GLvoid * data, CARD32 rop)
+{
+   __GLX_DECLARE_VARIABLES();
+
+   __GLX_LOAD_VARIABLES();
+   if (gc->currentDpy == NULL) {
+      return;
+   }
+
+   if (target == GL_PROXY_TEXTURE_3D) {
+      compsize = 0;
+   }
+   else {
+      compsize = image_size;
+   }
+
+   cmdlen = __GLX_PAD(__GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE + compsize);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      __GLX_BEGIN_VARIABLE(rop, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_LONG(8, level);
+      __GLX_PUT_LONG(12, xoffset);
+      __GLX_PUT_LONG(16, yoffset);
+      __GLX_PUT_LONG(20, width);
+      __GLX_PUT_LONG(24, height);
+      __GLX_PUT_LONG(28, format);
+      __GLX_PUT_LONG(32, image_size);
+      if (compsize != 0) {
+         __GLX_PUT_CHAR_ARRAY(__GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE,
+                              data, image_size);
+      }
+      __GLX_END(cmdlen);
+   }
+   else {
+      assert(compsize != 0);
+
+      __GLX_BEGIN_VARIABLE_LARGE(rop, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_LONG(12, level);
+      __GLX_PUT_LONG(16, xoffset);
+      __GLX_PUT_LONG(20, yoffset);
+      __GLX_PUT_LONG(24, width);
+      __GLX_PUT_LONG(28, height);
+      __GLX_PUT_LONG(32, format);
+      __GLX_PUT_LONG(36, image_size);
+      __glXSendLargeCommand(gc, gc->pc,
+                            __GLX_COMPRESSED_TEXSUBIMAGE_CMD_HDR_SIZE + 4,
+                            data, image_size);
+   }
+}
+
+
+void
+__indirect_glCompressedTexImage1DARB(GLenum target, GLint level,
+                                     GLenum internal_format, GLsizei width,
+                                     GLint border, GLsizei image_size,
+                                     const GLvoid * data)
+{
+   CompressedTexImage1D2D(target, level, internal_format, width, 0,
+                          border, image_size, data,
+                          X_GLrop_CompressedTexImage1D);
+}
+
+
+void
+__indirect_glCompressedTexImage2DARB(GLenum target, GLint level,
+                                     GLenum internal_format,
+                                     GLsizei width, GLsizei height,
+                                     GLint border, GLsizei image_size,
+                                     const GLvoid * data)
+{
+   CompressedTexImage1D2D(target, level, internal_format, width, height,
+                          border, image_size, data,
+                          X_GLrop_CompressedTexImage2D);
+}
+
+
+void
+__indirect_glCompressedTexImage3DARB(GLenum target, GLint level,
+                                     GLenum internal_format,
+                                     GLsizei width, GLsizei height,
+                                     GLsizei depth, GLint border,
+                                     GLsizei image_size, const GLvoid * data)
+{
+   __GLX_DECLARE_VARIABLES();
+
+   __GLX_LOAD_VARIABLES();
+   if (gc->currentDpy == NULL) {
+      return;
+   }
+
+   cmdlen = __GLX_PAD(__GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE + image_size);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      __GLX_BEGIN_VARIABLE(X_GLrop_CompressedTexImage3D, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_LONG(8, level);
+      __GLX_PUT_LONG(12, internal_format);
+      __GLX_PUT_LONG(16, width);
+      __GLX_PUT_LONG(20, height);
+      __GLX_PUT_LONG(24, depth);
+      __GLX_PUT_LONG(28, border);
+      __GLX_PUT_LONG(32, image_size);
+      if (image_size != 0) {
+         __GLX_PUT_CHAR_ARRAY(__GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE,
+                              data, image_size);
+      }
+      __GLX_END(cmdlen);
+   }
+   else {
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_CompressedTexImage3D, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_LONG(12, level);
+      __GLX_PUT_LONG(16, internal_format);
+      __GLX_PUT_LONG(20, width);
+      __GLX_PUT_LONG(24, height);
+      __GLX_PUT_LONG(28, depth);
+      __GLX_PUT_LONG(32, border);
+      __GLX_PUT_LONG(36, image_size);
+      __glXSendLargeCommand(gc, gc->pc,
+                            __GLX_COMPRESSED_TEXIMAGE_3D_CMD_HDR_SIZE + 4,
+                            data, image_size);
+   }
+}
+
+
+void
+__indirect_glCompressedTexSubImage1DARB(GLenum target, GLint level,
+                                        GLint xoffset,
+                                        GLsizei width,
+                                        GLenum format, GLsizei image_size,
+                                        const GLvoid * data)
+{
+   CompressedTexSubImage1D2D(target, level, xoffset, 0, width, 0,
+                             format, image_size, data,
+                             X_GLrop_CompressedTexSubImage1D);
+}
+
+
+void
+__indirect_glCompressedTexSubImage2DARB(GLenum target, GLint level,
+                                        GLint xoffset, GLint yoffset,
+                                        GLsizei width, GLsizei height,
+                                        GLenum format, GLsizei image_size,
+                                        const GLvoid * data)
+{
+   CompressedTexSubImage1D2D(target, level, xoffset, yoffset, width, height,
+                             format, image_size, data,
+                             X_GLrop_CompressedTexSubImage2D);
+}
+
+
+void
+__indirect_glCompressedTexSubImage3DARB(GLenum target, GLint level,
+                                        GLint xoffset, GLint yoffset,
+                                        GLint zoffset, GLsizei width,
+                                        GLsizei height, GLsizei depth,
+                                        GLenum format, GLsizei image_size,
+                                        const GLvoid * data)
+{
+   __GLX_DECLARE_VARIABLES();
+
+   __GLX_LOAD_VARIABLES();
+   if (gc->currentDpy == NULL) {
+      return;
+   }
+
+   cmdlen = __GLX_PAD(__GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE
+                      + image_size);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      __GLX_BEGIN_VARIABLE(X_GLrop_CompressedTexSubImage3D, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_LONG(8, level);
+      __GLX_PUT_LONG(12, xoffset);
+      __GLX_PUT_LONG(16, yoffset);
+      __GLX_PUT_LONG(20, zoffset);
+      __GLX_PUT_LONG(24, width);
+      __GLX_PUT_LONG(28, height);
+      __GLX_PUT_LONG(32, depth);
+      __GLX_PUT_LONG(36, format);
+      __GLX_PUT_LONG(40, image_size);
+      if (image_size != 0) {
+         __GLX_PUT_CHAR_ARRAY(__GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE,
+                              data, image_size);
+      }
+      __GLX_END(cmdlen);
+   }
+   else {
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_CompressedTexSubImage3D, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_LONG(12, level);
+      __GLX_PUT_LONG(16, xoffset);
+      __GLX_PUT_LONG(20, yoffset);
+      __GLX_PUT_LONG(24, zoffset);
+      __GLX_PUT_LONG(28, width);
+      __GLX_PUT_LONG(32, height);
+      __GLX_PUT_LONG(36, depth);
+      __GLX_PUT_LONG(40, format);
+      __GLX_PUT_LONG(44, image_size);
+      __glXSendLargeCommand(gc, gc->pc,
+                            __GLX_COMPRESSED_TEXSUBIMAGE_3D_CMD_HDR_SIZE + 4,
+                            data, image_size);
+   }
+}
diff --git a/src/glx/x11/indirect_transpose_matrix.c b/src/glx/x11/indirect_transpose_matrix.c
index 2144410e5ac..618db9f5d4f 100644
--- a/src/glx/x11/indirect_transpose_matrix.c
+++ b/src/glx/x11/indirect_transpose_matrix.c
@@ -25,59 +25,61 @@
 #include <GL/gl.h>
 #include "indirect.h"
 
-static void TransposeMatrixf(const GLfloat s[16], GLfloat d[16])
+static void
+TransposeMatrixf(const GLfloat s[16], GLfloat d[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < 4; j++) {
-            d[i*4+j] = s[j*4+i];
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < 4; j++) {
+         d[i * 4 + j] = s[j * 4 + i];
+      }
+   }
 }
 
-static void TransposeMatrixd(const GLdouble s[16], GLdouble d[16])
+static void
+TransposeMatrixd(const GLdouble s[16], GLdouble d[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < 4; j++) {
-            d[i*4+j] = s[j*4+i];
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < 4; j++) {
+         d[i * 4 + j] = s[j * 4 + i];
+      }
+   }
 }
 
 
 void
-__indirect_glLoadTransposeMatrixdARB( const GLdouble * m )
+__indirect_glLoadTransposeMatrixdARB(const GLdouble * m)
 {
    GLdouble mt[16];
-   
-   TransposeMatrixd( m, mt );
-   __indirect_glLoadMatrixd( mt );
+
+   TransposeMatrixd(m, mt);
+   __indirect_glLoadMatrixd(mt);
 }
 
 void
-__indirect_glLoadTransposeMatrixfARB( const GLfloat * m )
+__indirect_glLoadTransposeMatrixfARB(const GLfloat * m)
 {
    GLfloat mt[16];
 
-   TransposeMatrixf( m, mt );
-   __indirect_glLoadMatrixf( mt );
+   TransposeMatrixf(m, mt);
+   __indirect_glLoadMatrixf(mt);
 }
 
 void
-__indirect_glMultTransposeMatrixdARB( const GLdouble * m )
+__indirect_glMultTransposeMatrixdARB(const GLdouble * m)
 {
    GLdouble mt[16];
-   
-   TransposeMatrixd( m, mt );
-   __indirect_glMultMatrixd( mt );
+
+   TransposeMatrixd(m, mt);
+   __indirect_glMultMatrixd(mt);
 }
 
 void
-__indirect_glMultTransposeMatrixfARB( const GLfloat * m )
+__indirect_glMultTransposeMatrixfARB(const GLfloat * m)
 {
    GLfloat mt[16];
 
-   TransposeMatrixf( m, mt );
-   __indirect_glMultMatrixf( mt );
+   TransposeMatrixf(m, mt);
+   __indirect_glMultMatrixf(mt);
 }
diff --git a/src/glx/x11/indirect_vertex_array.c b/src/glx/x11/indirect_vertex_array.c
index 90ec277c41b..bb5232f6187 100644
--- a/src/glx/x11/indirect_vertex_array.c
+++ b/src/glx/x11/indirect_vertex_array.c
@@ -32,7 +32,7 @@
 #include <GL/glxproto.h>
 #include "glxextensions.h"
 #include "indirect_vertex_array.h"
-#include "indirect_va_private.h"
+#include "indirect_vertex_array_priv.h"
 
 #define __GLX_PAD(n) (((n)+3) & ~3)
 
@@ -62,28 +62,31 @@
  * "vbo", to support multiple texture coordinate arrays, generic attributes,
  * and vertex buffer objects.
  *
- * \author Ian Romanick <idr@us.ibm.com>
+ * \author Ian Romanick <ian.d.romanick@intel.com>
  */
 
-static void emit_DrawArrays_none( GLenum mode, GLint first, GLsizei count );
-static void emit_DrawArrays_old ( GLenum mode, GLint first, GLsizei count );
+static void emit_DrawArrays_none(GLenum mode, GLint first, GLsizei count);
+static void emit_DrawArrays_old(GLenum mode, GLint first, GLsizei count);
 
-static void emit_DrawElements_none( GLenum mode, GLsizei count, GLenum type,
-    const GLvoid *indices );
-static void emit_DrawElements_old ( GLenum mode, GLsizei count, GLenum type,
-    const GLvoid *indices );
+static void emit_DrawElements_none(GLenum mode, GLsizei count, GLenum type,
+                                   const GLvoid * indices);
+static void emit_DrawElements_old(GLenum mode, GLsizei count, GLenum type,
+                                  const GLvoid * indices);
 
 
-static GLubyte * emit_element_none( GLubyte * dst,
-    const struct array_state_vector * arrays, unsigned index );
-static GLubyte * emit_element_old( GLubyte * dst,
-    const struct array_state_vector * arrays, unsigned index );
-static struct array_state * get_array_entry(
-    const struct array_state_vector * arrays, GLenum key, unsigned index );
-static void fill_array_info_cache( struct array_state_vector * arrays );
-static GLboolean validate_mode(__GLXcontext *gc, GLenum mode);
-static GLboolean validate_count(__GLXcontext *gc, GLsizei count);
-static GLboolean validate_type(__GLXcontext *gc, GLenum type);
+static GLubyte *emit_element_none(GLubyte * dst,
+                                  const struct array_state_vector *arrays,
+                                  unsigned index);
+static GLubyte *emit_element_old(GLubyte * dst,
+                                 const struct array_state_vector *arrays,
+                                 unsigned index);
+static struct array_state *get_array_entry(const struct array_state_vector
+                                           *arrays, GLenum key,
+                                           unsigned index);
+static void fill_array_info_cache(struct array_state_vector *arrays);
+static GLboolean validate_mode(__GLXcontext * gc, GLenum mode);
+static GLboolean validate_count(__GLXcontext * gc, GLsizei count);
+static GLboolean validate_type(__GLXcontext * gc, GLenum type);
 
 
 /**
@@ -97,10 +100,34 @@ static GLboolean validate_type(__GLXcontext *gc, GLenum type);
  * \c GL_3_BYTES, or \c GL_4_BYTES.
  */
 const GLuint __glXTypeSize_table[16] = {
-    1, 1, 2, 2, 4, 4, 4, 0, 0, 0, 8, 0, 0, 0, 0, 0
+   1, 1, 2, 2, 4, 4, 4, 0, 0, 0, 8, 0, 0, 0, 0, 0
 };
 
 
+/**
+ * Free the per-context array state that was allocated with
+ * __glXInitVertexArrayState().
+ */
+void
+__glXFreeVertexArrayState(__GLXcontext * gc)
+{
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+
+   if (arrays) {
+      if (arrays->stack) {
+         free(arrays->stack);
+         arrays->stack = NULL;
+      }
+      if (arrays->arrays) {
+         free(arrays->arrays);
+         arrays->arrays = NULL;
+      }
+      free(arrays);
+      state->array_state = NULL;
+   }
+}
+
 
 /**
  * Initialize vertex array state of a GLX context.
@@ -117,154 +144,154 @@ const GLuint __glXTypeSize_table[16] = {
  * Return values from malloc are not properly tested.
  */
 void
-__glXInitVertexArrayState( __GLXcontext * gc )
+__glXInitVertexArrayState(__GLXcontext * gc)
 {
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays;
-
-    unsigned array_count;
-    int texture_units = 1, vertex_program_attribs = 0;
-    unsigned i, j;
-
-    GLboolean got_fog = GL_FALSE;
-    GLboolean got_secondary_color = GL_FALSE;
-
-
-    arrays = calloc( 1, sizeof( struct array_state_vector ) );
-    state->array_state = arrays;
-
-    arrays->old_DrawArrays_possible = !state->NoDrawArraysProtocol;
-    arrays->new_DrawArrays_possible = GL_FALSE;
-    arrays->DrawArrays = NULL;
-
-    arrays->active_texture_unit = 0;
-
-
-    /* Determine how many arrays are actually needed.  Only arrays that
-     * are supported by the server are create.  For example, if the server
-     * supports only 2 texture units, then only 2 texture coordinate arrays
-     * are created.
-     * 
-     * At the very least, GL_VERTEX_ARRAY, GL_NORMAL_ARRAY,
-     * GL_COLOR_ARRAY, GL_INDEX_ARRAY, GL_TEXTURE_COORD_ARRAY, and
-     * GL_EDGE_FLAG_ARRAY are supported.
-     */
-
-    array_count = 5;
-    
-    if ( __glExtensionBitIsEnabled( gc, GL_EXT_fog_coord_bit )
-	 || (gc->server_major > 1) || (gc->server_minor >= 4) ) {
-	got_fog = GL_TRUE;
-	array_count++;
-    }
-
-    if ( __glExtensionBitIsEnabled( gc, GL_EXT_secondary_color_bit )
-	 || (gc->server_major > 1) || (gc->server_minor >= 4) ) {
-	got_secondary_color = GL_TRUE;
-	array_count++;
-    }
-
-    if ( __glExtensionBitIsEnabled( gc, GL_ARB_multitexture_bit )
-	 || (gc->server_major > 1) || (gc->server_minor >= 3) ) {
-	__indirect_glGetIntegerv( GL_MAX_TEXTURE_UNITS, & texture_units );
-    }
-
-    if ( __glExtensionBitIsEnabled( gc, GL_ARB_vertex_program_bit ) ) {
-	__indirect_glGetProgramivARB( GL_VERTEX_PROGRAM_ARB,
-				      GL_MAX_PROGRAM_ATTRIBS_ARB,
-				      & vertex_program_attribs );
-    }
-
-    arrays->num_texture_units = texture_units;
-    arrays->num_vertex_program_attribs = vertex_program_attribs;
-    array_count += texture_units + vertex_program_attribs;
-    arrays->num_arrays = array_count;
-    arrays->arrays = calloc( array_count, sizeof( struct array_state ) );
-
-    arrays->arrays[0].data_type = GL_FLOAT;
-    arrays->arrays[0].count = 3;
-    arrays->arrays[0].key = GL_NORMAL_ARRAY;
-    arrays->arrays[0].normalized = GL_TRUE;
-    arrays->arrays[0].old_DrawArrays_possible = GL_TRUE;
-
-    arrays->arrays[1].data_type = GL_FLOAT;
-    arrays->arrays[1].count = 4;
-    arrays->arrays[1].key = GL_COLOR_ARRAY;
-    arrays->arrays[1].normalized = GL_TRUE;
-    arrays->arrays[1].old_DrawArrays_possible = GL_TRUE;
-
-    arrays->arrays[2].data_type = GL_FLOAT;
-    arrays->arrays[2].count = 1;
-    arrays->arrays[2].key = GL_INDEX_ARRAY;
-    arrays->arrays[2].old_DrawArrays_possible = GL_TRUE;
-
-    arrays->arrays[3].data_type = GL_UNSIGNED_BYTE;
-    arrays->arrays[3].count = 1;
-    arrays->arrays[3].key = GL_EDGE_FLAG_ARRAY;
-    arrays->arrays[3].old_DrawArrays_possible = GL_TRUE;
-
-    for ( i = 0 ; i < texture_units ; i++ ) {
-	arrays->arrays[4 + i].data_type = GL_FLOAT;
-	arrays->arrays[4 + i].count = 4;
-	arrays->arrays[4 + i].key = GL_TEXTURE_COORD_ARRAY;
-
-	arrays->arrays[4 + i].old_DrawArrays_possible = (i == 0);
-	arrays->arrays[4 + i].index = i;
-
-	arrays->arrays[4 + i].header[1] = i + GL_TEXTURE0;
-    }
-    
-    i = 4 + texture_units;
-
-    if ( got_fog ) {
-	arrays->arrays[i].data_type = GL_FLOAT;
-	arrays->arrays[i].count = 1;
-	arrays->arrays[i].key = GL_FOG_COORDINATE_ARRAY;
-	arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
-	i++;
-    }
-
-    if ( got_secondary_color ) {
-	arrays->arrays[i].data_type = GL_FLOAT;
-	arrays->arrays[i].count = 3;
-	arrays->arrays[i].key = GL_SECONDARY_COLOR_ARRAY;
-	arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
-	arrays->arrays[i].normalized = GL_TRUE;
-	i++;
-    }
-
-
-    for ( j = 0 ; j < vertex_program_attribs ; j++ ) {
-	const unsigned idx = (vertex_program_attribs - (j + 1));
-
-
-	arrays->arrays[idx + i].data_type = GL_FLOAT;
-	arrays->arrays[idx + i].count = 4;
-	arrays->arrays[idx + i].key = GL_VERTEX_ATTRIB_ARRAY_POINTER;
-
-	arrays->arrays[idx + i].old_DrawArrays_possible = 0;
-	arrays->arrays[idx + i].index = idx;
-
-	arrays->arrays[idx + i].header[1] = idx;
-    }
-
-    i += vertex_program_attribs;
-
-
-    /* Vertex array *must* be last becuase of the way that
-     * emit_DrawArrays_none works.
-     */
-
-    arrays->arrays[i].data_type = GL_FLOAT;
-    arrays->arrays[i].count = 4;
-    arrays->arrays[i].key = GL_VERTEX_ARRAY;
-    arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
-
-    assert( (i + 1) == arrays->num_arrays );
-
-    arrays->stack_index = 0;
-    arrays->stack = malloc( sizeof( struct array_stack_state )
-			    * arrays->num_arrays );
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays;
+
+   unsigned array_count;
+   int texture_units = 1, vertex_program_attribs = 0;
+   unsigned i, j;
+
+   GLboolean got_fog = GL_FALSE;
+   GLboolean got_secondary_color = GL_FALSE;
+
+
+   arrays = calloc(1, sizeof(struct array_state_vector));
+   state->array_state = arrays;
+
+   arrays->old_DrawArrays_possible = !state->NoDrawArraysProtocol;
+   arrays->new_DrawArrays_possible = GL_FALSE;
+   arrays->DrawArrays = NULL;
+
+   arrays->active_texture_unit = 0;
+
+
+   /* Determine how many arrays are actually needed.  Only arrays that
+    * are supported by the server are create.  For example, if the server
+    * supports only 2 texture units, then only 2 texture coordinate arrays
+    * are created.
+    * 
+    * At the very least, GL_VERTEX_ARRAY, GL_NORMAL_ARRAY,
+    * GL_COLOR_ARRAY, GL_INDEX_ARRAY, GL_TEXTURE_COORD_ARRAY, and
+    * GL_EDGE_FLAG_ARRAY are supported.
+    */
+
+   array_count = 5;
+
+   if (__glExtensionBitIsEnabled(gc, GL_EXT_fog_coord_bit)
+       || (gc->server_major > 1) || (gc->server_minor >= 4)) {
+      got_fog = GL_TRUE;
+      array_count++;
+   }
+
+   if (__glExtensionBitIsEnabled(gc, GL_EXT_secondary_color_bit)
+       || (gc->server_major > 1) || (gc->server_minor >= 4)) {
+      got_secondary_color = GL_TRUE;
+      array_count++;
+   }
+
+   if (__glExtensionBitIsEnabled(gc, GL_ARB_multitexture_bit)
+       || (gc->server_major > 1) || (gc->server_minor >= 3)) {
+      __indirect_glGetIntegerv(GL_MAX_TEXTURE_UNITS, &texture_units);
+   }
+
+   if (__glExtensionBitIsEnabled(gc, GL_ARB_vertex_program_bit)) {
+      __indirect_glGetProgramivARB(GL_VERTEX_PROGRAM_ARB,
+                                   GL_MAX_PROGRAM_ATTRIBS_ARB,
+                                   &vertex_program_attribs);
+   }
+
+   arrays->num_texture_units = texture_units;
+   arrays->num_vertex_program_attribs = vertex_program_attribs;
+   array_count += texture_units + vertex_program_attribs;
+   arrays->num_arrays = array_count;
+   arrays->arrays = calloc(array_count, sizeof(struct array_state));
+
+   arrays->arrays[0].data_type = GL_FLOAT;
+   arrays->arrays[0].count = 3;
+   arrays->arrays[0].key = GL_NORMAL_ARRAY;
+   arrays->arrays[0].normalized = GL_TRUE;
+   arrays->arrays[0].old_DrawArrays_possible = GL_TRUE;
+
+   arrays->arrays[1].data_type = GL_FLOAT;
+   arrays->arrays[1].count = 4;
+   arrays->arrays[1].key = GL_COLOR_ARRAY;
+   arrays->arrays[1].normalized = GL_TRUE;
+   arrays->arrays[1].old_DrawArrays_possible = GL_TRUE;
+
+   arrays->arrays[2].data_type = GL_FLOAT;
+   arrays->arrays[2].count = 1;
+   arrays->arrays[2].key = GL_INDEX_ARRAY;
+   arrays->arrays[2].old_DrawArrays_possible = GL_TRUE;
+
+   arrays->arrays[3].data_type = GL_UNSIGNED_BYTE;
+   arrays->arrays[3].count = 1;
+   arrays->arrays[3].key = GL_EDGE_FLAG_ARRAY;
+   arrays->arrays[3].old_DrawArrays_possible = GL_TRUE;
+
+   for (i = 0; i < texture_units; i++) {
+      arrays->arrays[4 + i].data_type = GL_FLOAT;
+      arrays->arrays[4 + i].count = 4;
+      arrays->arrays[4 + i].key = GL_TEXTURE_COORD_ARRAY;
+
+      arrays->arrays[4 + i].old_DrawArrays_possible = (i == 0);
+      arrays->arrays[4 + i].index = i;
+
+      arrays->arrays[4 + i].header[1] = i + GL_TEXTURE0;
+   }
+
+   i = 4 + texture_units;
+
+   if (got_fog) {
+      arrays->arrays[i].data_type = GL_FLOAT;
+      arrays->arrays[i].count = 1;
+      arrays->arrays[i].key = GL_FOG_COORDINATE_ARRAY;
+      arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
+      i++;
+   }
+
+   if (got_secondary_color) {
+      arrays->arrays[i].data_type = GL_FLOAT;
+      arrays->arrays[i].count = 3;
+      arrays->arrays[i].key = GL_SECONDARY_COLOR_ARRAY;
+      arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
+      arrays->arrays[i].normalized = GL_TRUE;
+      i++;
+   }
+
+
+   for (j = 0; j < vertex_program_attribs; j++) {
+      const unsigned idx = (vertex_program_attribs - (j + 1));
+
+
+      arrays->arrays[idx + i].data_type = GL_FLOAT;
+      arrays->arrays[idx + i].count = 4;
+      arrays->arrays[idx + i].key = GL_VERTEX_ATTRIB_ARRAY_POINTER;
+
+      arrays->arrays[idx + i].old_DrawArrays_possible = 0;
+      arrays->arrays[idx + i].index = idx;
+
+      arrays->arrays[idx + i].header[1] = idx;
+   }
+
+   i += vertex_program_attribs;
+
+
+   /* Vertex array *must* be last becuase of the way that
+    * emit_DrawArrays_none works.
+    */
+
+   arrays->arrays[i].data_type = GL_FLOAT;
+   arrays->arrays[i].count = 4;
+   arrays->arrays[i].key = GL_VERTEX_ARRAY;
+   arrays->arrays[i].old_DrawArrays_possible = GL_TRUE;
+
+   assert((i + 1) == arrays->num_arrays);
+
+   arrays->stack_index = 0;
+   arrays->stack = malloc(sizeof(struct array_stack_state)
+                          * arrays->num_arrays);
 }
 
 
@@ -274,19 +301,19 @@ __glXInitVertexArrayState( __GLXcontext * gc )
  * implement the enabled vertex arrays.
  */
 static size_t
-calculate_single_vertex_size_none( const struct array_state_vector * arrays )
+calculate_single_vertex_size_none(const struct array_state_vector *arrays)
 {
-    size_t single_vertex_size = 0;
-    unsigned   i;
+   size_t single_vertex_size = 0;
+   unsigned i;
 
 
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-    	if ( arrays->arrays[i].enabled ) {
-	    single_vertex_size += ((uint16_t *)arrays->arrays[i].header)[0];
-	}
-    }
-    
-    return single_vertex_size;
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if (arrays->arrays[i].enabled) {
+         single_vertex_size += ((uint16_t *) arrays->arrays[i].header)[0];
+      }
+   }
+
+   return single_vertex_size;
 }
 
 
@@ -294,39 +321,37 @@ calculate_single_vertex_size_none( const struct array_state_vector * arrays )
  * Emit a single element using non-DrawArrays protocol.
  */
 GLubyte *
-emit_element_none( GLubyte * dst,
-		   const struct array_state_vector * arrays,
-		   unsigned index )
+emit_element_none(GLubyte * dst,
+                  const struct array_state_vector * arrays, unsigned index)
 {
-    unsigned i;
+   unsigned i;
 
 
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	if ( arrays->arrays[i].enabled ) {
-	    const size_t offset = index * arrays->arrays[i].true_stride;
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if (arrays->arrays[i].enabled) {
+         const size_t offset = index * arrays->arrays[i].true_stride;
 
-	    /* The generic attributes can have more data than is in the
-	     * elements.  This is because a vertex array can be a 2 element,
-	     * normalized, unsigned short, but the "closest" immediate mode
-	     * protocol is for a 4Nus.  Since the sizes are small, the
-	     * performance impact on modern processors should be negligible.
-	     */
-	    (void) memset( dst, 0,
-			   ((uint16_t *)arrays->arrays[i].header)[0] );
+         /* The generic attributes can have more data than is in the
+          * elements.  This is because a vertex array can be a 2 element,
+          * normalized, unsigned short, but the "closest" immediate mode
+          * protocol is for a 4Nus.  Since the sizes are small, the
+          * performance impact on modern processors should be negligible.
+          */
+         (void) memset(dst, 0, ((uint16_t *) arrays->arrays[i].header)[0]);
 
-	    (void) memcpy( dst, arrays->arrays[i].header, 
-			   arrays->arrays[i].header_size );
+         (void) memcpy(dst, arrays->arrays[i].header,
+                       arrays->arrays[i].header_size);
 
-	    dst += arrays->arrays[i].header_size;
+         dst += arrays->arrays[i].header_size;
 
-	    (void) memcpy( dst, ((GLubyte *) arrays->arrays[i].data) + offset,
-			   arrays->arrays[i].element_size );
+         (void) memcpy(dst, ((GLubyte *) arrays->arrays[i].data) + offset,
+                       arrays->arrays[i].element_size);
 
-	    dst += __GLX_PAD( arrays->arrays[i].element_size );
-	}
-    }
+         dst += __GLX_PAD(arrays->arrays[i].element_size);
+      }
+   }
 
-    return dst;
+   return dst;
 }
 
 
@@ -335,120 +360,119 @@ emit_element_none( GLubyte * dst,
  * EXT_vertex_arrays / OpenGL 1.1.
  */
 GLubyte *
-emit_element_old( GLubyte * dst,
-		  const struct array_state_vector * arrays,
-		  unsigned index )
+emit_element_old(GLubyte * dst,
+                 const struct array_state_vector * arrays, unsigned index)
 {
-    unsigned i;
+   unsigned i;
 
 
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	if ( arrays->arrays[i].enabled ) {
-	    const size_t offset = index * arrays->arrays[i].true_stride;
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if (arrays->arrays[i].enabled) {
+         const size_t offset = index * arrays->arrays[i].true_stride;
 
-	    (void) memcpy( dst, ((GLubyte *) arrays->arrays[i].data) + offset,
-			   arrays->arrays[i].element_size );
+         (void) memcpy(dst, ((GLubyte *) arrays->arrays[i].data) + offset,
+                       arrays->arrays[i].element_size);
 
-	    dst += __GLX_PAD( arrays->arrays[i].element_size );
-	}
-    }
+         dst += __GLX_PAD(arrays->arrays[i].element_size);
+      }
+   }
 
-    return dst;
+   return dst;
 }
 
 
 struct array_state *
-get_array_entry( const struct array_state_vector * arrays,
-		 GLenum key, unsigned index )
+get_array_entry(const struct array_state_vector *arrays,
+                GLenum key, unsigned index)
 {
-    unsigned  i;
-
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	if ( (arrays->arrays[i].key == key)
-	     && (arrays->arrays[i].index == index) ) {
-	    return & arrays->arrays[i];
-	}
-    }
-    
-    return NULL;
+   unsigned i;
+
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if ((arrays->arrays[i].key == key)
+          && (arrays->arrays[i].index == index)) {
+         return &arrays->arrays[i];
+      }
+   }
+
+   return NULL;
 }
 
 
 static GLboolean
-allocate_array_info_cache( struct array_state_vector * arrays,
-			   size_t required_size )
+allocate_array_info_cache(struct array_state_vector *arrays,
+                          size_t required_size)
 {
 #define MAX_HEADER_SIZE 20
-    if ( arrays->array_info_cache_buffer_size < required_size ) {
-	GLubyte * temp = realloc( arrays->array_info_cache_base,
-				  required_size + MAX_HEADER_SIZE );
+   if (arrays->array_info_cache_buffer_size < required_size) {
+      GLubyte *temp = realloc(arrays->array_info_cache_base,
+                              required_size + MAX_HEADER_SIZE);
 
-	if ( temp == NULL ) {
-	    return GL_FALSE;
-	}
+      if (temp == NULL) {
+         return GL_FALSE;
+      }
 
-	arrays->array_info_cache_base = temp;
-	arrays->array_info_cache = temp + MAX_HEADER_SIZE;
-	arrays->array_info_cache_buffer_size = required_size;
-    }
+      arrays->array_info_cache_base = temp;
+      arrays->array_info_cache = temp + MAX_HEADER_SIZE;
+      arrays->array_info_cache_buffer_size = required_size;
+   }
 
-    arrays->array_info_cache_size = required_size;
-    return GL_TRUE;
+   arrays->array_info_cache_size = required_size;
+   return GL_TRUE;
 }
 
 
 /**
  */
 void
-fill_array_info_cache( struct array_state_vector * arrays )
+fill_array_info_cache(struct array_state_vector *arrays)
 {
-    GLboolean old_DrawArrays_possible;
-    unsigned  i;
-
-
-    /* Determine how many arrays are enabled.
-     */
-
-    arrays->enabled_client_array_count = 0;
-    old_DrawArrays_possible = arrays->old_DrawArrays_possible;
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	if ( arrays->arrays[i].enabled ) {
-	    arrays->enabled_client_array_count++;
-	    old_DrawArrays_possible &= arrays->arrays[i].old_DrawArrays_possible;
-	}
-    }
-    
-    if ( arrays->new_DrawArrays_possible ) {
-	assert( ! arrays->new_DrawArrays_possible );
-    }
-    else if ( old_DrawArrays_possible ) {
-	const size_t required_size = arrays->enabled_client_array_count * 12;
-	uint32_t * info;
-
-
-	if ( ! allocate_array_info_cache( arrays, required_size ) ) {
-	    return;
-	}
-
-
-	info = (uint32_t *) arrays->array_info_cache;
-	for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	    if ( arrays->arrays[i].enabled ) {
-		*(info++) = arrays->arrays[i].data_type;
-		*(info++) = arrays->arrays[i].count;
-		*(info++) = arrays->arrays[i].key;
-	    }
-	}
-
-	arrays->DrawArrays = emit_DrawArrays_old;
-	arrays->DrawElements = emit_DrawElements_old;
-    }
-    else {
-	arrays->DrawArrays = emit_DrawArrays_none;
-	arrays->DrawElements = emit_DrawElements_none;
-    }
-
-    arrays->array_info_cache_valid = GL_TRUE;
+   GLboolean old_DrawArrays_possible;
+   unsigned i;
+
+
+   /* Determine how many arrays are enabled.
+    */
+
+   arrays->enabled_client_array_count = 0;
+   old_DrawArrays_possible = arrays->old_DrawArrays_possible;
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if (arrays->arrays[i].enabled) {
+         arrays->enabled_client_array_count++;
+         old_DrawArrays_possible &= arrays->arrays[i].old_DrawArrays_possible;
+      }
+   }
+
+   if (arrays->new_DrawArrays_possible) {
+      assert(!arrays->new_DrawArrays_possible);
+   }
+   else if (old_DrawArrays_possible) {
+      const size_t required_size = arrays->enabled_client_array_count * 12;
+      uint32_t *info;
+
+
+      if (!allocate_array_info_cache(arrays, required_size)) {
+         return;
+      }
+
+
+      info = (uint32_t *) arrays->array_info_cache;
+      for (i = 0; i < arrays->num_arrays; i++) {
+         if (arrays->arrays[i].enabled) {
+            *(info++) = arrays->arrays[i].data_type;
+            *(info++) = arrays->arrays[i].count;
+            *(info++) = arrays->arrays[i].key;
+         }
+      }
+
+      arrays->DrawArrays = emit_DrawArrays_old;
+      arrays->DrawElements = emit_DrawElements_old;
+   }
+   else {
+      arrays->DrawArrays = emit_DrawArrays_none;
+      arrays->DrawElements = emit_DrawElements_none;
+   }
+
+   arrays->array_info_cache_valid = GL_TRUE;
 }
 
 
@@ -460,48 +484,48 @@ fill_array_info_cache( struct array_state_vector * arrays )
  * vertex state is enabled that is not compatible with that protocol.
  */
 void
-emit_DrawArrays_none( GLenum mode, GLint first, GLsizei count )
+emit_DrawArrays_none(GLenum mode, GLint first, GLsizei count)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
 
-    size_t single_vertex_size;
-    GLubyte * pc;
-    unsigned  i;
-    static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
-    static const uint16_t end_cmd[2]   = { 4, X_GLrop_End };
+   size_t single_vertex_size;
+   GLubyte *pc;
+   unsigned i;
+   static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
+   static const uint16_t end_cmd[2] = { 4, X_GLrop_End };
 
 
-    single_vertex_size = calculate_single_vertex_size_none( arrays );
+   single_vertex_size = calculate_single_vertex_size_none(arrays);
 
-    pc = gc->pc;
+   pc = gc->pc;
 
-    (void) memcpy( pc, begin_cmd, 4 );
-    *(int *)(pc +  4) = mode;
+   (void) memcpy(pc, begin_cmd, 4);
+   *(int *) (pc + 4) = mode;
 
-    pc += 8;
+   pc += 8;
 
-    for ( i = 0 ; i < count ; i++ ) {
-	if ( (pc + single_vertex_size) >= gc->bufEnd ) {
-	    pc = __glXFlushRenderBuffer(gc, gc->pc);
-	}
+   for (i = 0; i < count; i++) {
+      if ((pc + single_vertex_size) >= gc->bufEnd) {
+         pc = __glXFlushRenderBuffer(gc, pc);
+      }
 
-	pc = emit_element_none( pc, arrays, first + i );
-    }
+      pc = emit_element_none(pc, arrays, first + i);
+   }
 
-    if ( (pc + 4) >= gc->bufEnd ) {
-	pc = __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   if ((pc + 4) >= gc->bufEnd) {
+      pc = __glXFlushRenderBuffer(gc, pc);
+   }
 
-    (void) memcpy( pc, end_cmd, 4 );
-    pc += 4;
+   (void) memcpy(pc, end_cmd, 4);
+   pc += 4;
 
-    gc->pc = pc;
-    if ( gc->pc > gc->limit ) {
-	(void) __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   gc->pc = pc;
+   if (gc->pc > gc->limit) {
+      (void) __glXFlushRenderBuffer(gc, gc->pc);
+   }
 }
 
 
@@ -524,321 +548,321 @@ emit_DrawArrays_none( GLenum mode, GLint first, GLsizei count )
  * A pointer to the buffer for array data.
  */
 static GLubyte *
-emit_DrawArrays_header_old( __GLXcontext * gc,
-			    struct array_state_vector * arrays,
-			    size_t * elements_per_request,
-			    size_t * total_requests,
-			    GLenum mode, GLsizei count )
+emit_DrawArrays_header_old(__GLXcontext * gc,
+                           struct array_state_vector *arrays,
+                           size_t * elements_per_request,
+                           unsigned int *total_requests,
+                           GLenum mode, GLsizei count)
 {
-    size_t command_size;
-    size_t single_vertex_size;
-    const unsigned header_size = 16;
-    unsigned  i;
-    GLubyte * pc;
-
-
-    /* Determine the size of the whole command.  This includes the header,
-     * the ARRAY_INFO data and the array data.  Once this size is calculated,
-     * it will be known whether a Render or RenderLarge command is needed.
-     */
-
-    single_vertex_size = 0;
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	if ( arrays->arrays[i].enabled ) {
-	    single_vertex_size += __GLX_PAD( arrays->arrays[i].element_size );
-	}
-    }
-
-    command_size = arrays->array_info_cache_size + header_size 
+   size_t command_size;
+   size_t single_vertex_size;
+   const unsigned header_size = 16;
+   unsigned i;
+   GLubyte *pc;
+
+
+   /* Determine the size of the whole command.  This includes the header,
+    * the ARRAY_INFO data and the array data.  Once this size is calculated,
+    * it will be known whether a Render or RenderLarge command is needed.
+    */
+
+   single_vertex_size = 0;
+   for (i = 0; i < arrays->num_arrays; i++) {
+      if (arrays->arrays[i].enabled) {
+         single_vertex_size += __GLX_PAD(arrays->arrays[i].element_size);
+      }
+   }
+
+   command_size = arrays->array_info_cache_size + header_size
       + (single_vertex_size * count);
 
 
-    /* Write the header for either a Render command or a RenderLarge
-     * command.  After the header is written, write the ARRAY_INFO data.
-     */
+   /* Write the header for either a Render command or a RenderLarge
+    * command.  After the header is written, write the ARRAY_INFO data.
+    */
+
+   if (command_size > gc->maxSmallRenderCommandSize) {
+      /* maxSize is the maximum amount of data can be stuffed into a single
+       * packet.  sz_xGLXRenderReq is added because bufSize is the maximum
+       * packet size minus sz_xGLXRenderReq.
+       */
+      const size_t maxSize = (gc->bufSize + sz_xGLXRenderReq)
+         - sz_xGLXRenderLargeReq;
+      unsigned vertex_requests;
 
-    if ( command_size > gc->maxSmallRenderCommandSize ) {
-	/* maxSize is the maximum amount of data can be stuffed into a single
-	 * packet.  sz_xGLXRenderReq is added because bufSize is the maximum
-	 * packet size minus sz_xGLXRenderReq.
-	 */
-	const size_t maxSize = (gc->bufSize + sz_xGLXRenderReq)
-	  - sz_xGLXRenderLargeReq;
-	unsigned vertex_requests;
 
+      /* Calculate the number of data packets that will be required to send
+       * the whole command.  To do this, the number of verticies that
+       * will fit in a single buffer must be calculated.
+       * 
+       * The important value here is elements_per_request.  This is the
+       * number of complete array elements that will fit in a single
+       * buffer.  There may be some wasted space at the end of the buffer,
+       * but splitting elements across buffer boundries would be painful.
+       */
 
-	/* Calculate the number of data packets that will be required to send
-	 * the whole command.  To do this, the number of verticies that
-	 * will fit in a single buffer must be calculated.
-	 * 
-	 * The important value here is elements_per_request.  This is the
-	 * number of complete array elements that will fit in a single
-	 * buffer.  There may be some wasted space at the end of the buffer,
-	 * but splitting elements across buffer boundries would be painful.
-	 */
+      elements_per_request[0] = maxSize / single_vertex_size;
 
-	elements_per_request[0] = maxSize / single_vertex_size;
+      vertex_requests = (count + elements_per_request[0] - 1)
+         / elements_per_request[0];
 
-	vertex_requests = (count + elements_per_request[0] - 1)
-	  / elements_per_request[0];
-	  
-	*total_requests = vertex_requests + 1;
+      *total_requests = vertex_requests + 1;
 
 
-	__glXFlushRenderBuffer(gc, gc->pc);
+      __glXFlushRenderBuffer(gc, gc->pc);
 
-	command_size += 4;
+      command_size += 4;
 
-	pc = ((GLubyte *) arrays->array_info_cache) - (header_size + 4);
-	*(uint32_t *)(pc +  0) = command_size;
-	*(uint32_t *)(pc +  4) = X_GLrop_DrawArrays;
-	*(uint32_t *)(pc +  8) = count;
-	*(uint32_t *)(pc + 12) = arrays->enabled_client_array_count;
-	*(uint32_t *)(pc + 16) = mode;
+      pc = ((GLubyte *) arrays->array_info_cache) - (header_size + 4);
+      *(uint32_t *) (pc + 0) = command_size;
+      *(uint32_t *) (pc + 4) = X_GLrop_DrawArrays;
+      *(uint32_t *) (pc + 8) = count;
+      *(uint32_t *) (pc + 12) = arrays->enabled_client_array_count;
+      *(uint32_t *) (pc + 16) = mode;
 
-	__glXSendLargeChunk( gc, 1, *total_requests, pc,
-			     header_size + 4 + arrays->array_info_cache_size );
+      __glXSendLargeChunk(gc, 1, *total_requests, pc,
+                          header_size + 4 + arrays->array_info_cache_size);
 
-	pc = gc->pc;
-    }
-    else {
-	if ( (gc->pc + command_size) >= gc->bufEnd ) {
-	    (void) __glXFlushRenderBuffer(gc, gc->pc);
-	}
+      pc = gc->pc;
+   }
+   else {
+      if ((gc->pc + command_size) >= gc->bufEnd) {
+         (void) __glXFlushRenderBuffer(gc, gc->pc);
+      }
 
-	pc = gc->pc;
-	*(uint16_t *)(pc +  0) = command_size;
-	*(uint16_t *)(pc +  2) = X_GLrop_DrawArrays;
-	*(uint32_t *)(pc +  4) = count;
-	*(uint32_t *)(pc +  8) = arrays->enabled_client_array_count;
-	*(uint32_t *)(pc + 12) = mode;
+      pc = gc->pc;
+      *(uint16_t *) (pc + 0) = command_size;
+      *(uint16_t *) (pc + 2) = X_GLrop_DrawArrays;
+      *(uint32_t *) (pc + 4) = count;
+      *(uint32_t *) (pc + 8) = arrays->enabled_client_array_count;
+      *(uint32_t *) (pc + 12) = mode;
 
-	pc += header_size;
+      pc += header_size;
 
-	(void) memcpy( pc, arrays->array_info_cache,
-		       arrays->array_info_cache_size );
-	pc += arrays->array_info_cache_size;
+      (void) memcpy(pc, arrays->array_info_cache,
+                    arrays->array_info_cache_size);
+      pc += arrays->array_info_cache_size;
 
-	*elements_per_request = count;
-	*total_requests = 0;
-    }
+      *elements_per_request = count;
+      *total_requests = 0;
+   }
 
 
-    return pc;
+   return pc;
 }
 
 
 /**
  */
 void
-emit_DrawArrays_old( GLenum mode, GLint first, GLsizei count )
+emit_DrawArrays_old(GLenum mode, GLint first, GLsizei count)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
 
-    GLubyte * pc;
-    size_t elements_per_request;
-    size_t total_requests = 0;
-    unsigned i;
-    size_t total_sent = 0;
+   GLubyte *pc;
+   size_t elements_per_request;
+   unsigned total_requests = 0;
+   unsigned i;
+   size_t total_sent = 0;
 
 
-    pc = emit_DrawArrays_header_old( gc, arrays, & elements_per_request,
-				     & total_requests, mode, count);
+   pc = emit_DrawArrays_header_old(gc, arrays, &elements_per_request,
+                                   &total_requests, mode, count);
 
-    
-    /* Write the arrays.
-     */
 
-    if ( total_requests == 0 ) {
-	assert( elements_per_request >= count );
+   /* Write the arrays.
+    */
 
-	for ( i = 0 ; i < count ; i++ ) {
-	    pc = emit_element_old( pc, arrays, i + first );
-	}
+   if (total_requests == 0) {
+      assert(elements_per_request >= count);
 
-	assert( pc <= gc->bufEnd );
+      for (i = 0; i < count; i++) {
+         pc = emit_element_old(pc, arrays, i + first);
+      }
 
-	gc->pc = pc;
-	if ( gc->pc > gc->limit ) {
-	    (void) __glXFlushRenderBuffer(gc, gc->pc);
-	}
-    }
-    else {
-	unsigned req;
+      assert(pc <= gc->bufEnd);
 
+      gc->pc = pc;
+      if (gc->pc > gc->limit) {
+         (void) __glXFlushRenderBuffer(gc, gc->pc);
+      }
+   }
+   else {
+      unsigned req;
 
-	for ( req = 2 ; req <= total_requests ; req++ ) {
-	    if ( count < elements_per_request ) {
-		elements_per_request = count;
-	    }
 
-	    pc = gc->pc;
-	    for ( i = 0 ; i < elements_per_request ; i++ ) {
-		pc = emit_element_old( pc, arrays, i + first );
-	    }
+      for (req = 2; req <= total_requests; req++) {
+         if (count < elements_per_request) {
+            elements_per_request = count;
+         }
 
-	    first += elements_per_request;
+         pc = gc->pc;
+         for (i = 0; i < elements_per_request; i++) {
+            pc = emit_element_old(pc, arrays, i + first);
+         }
 
-	    total_sent += (size_t) (pc - gc->pc);
-	    __glXSendLargeChunk( gc, req, total_requests, gc->pc,
-				 pc - gc->pc );
+         first += elements_per_request;
 
-	    count -= elements_per_request;
-	}
-    }
+         total_sent += (size_t) (pc - gc->pc);
+         __glXSendLargeChunk(gc, req, total_requests, gc->pc, pc - gc->pc);
+
+         count -= elements_per_request;
+      }
+   }
 }
 
 
 void
-emit_DrawElements_none( GLenum mode, GLsizei count, GLenum type,
-			const GLvoid *indices )
+emit_DrawElements_none(GLenum mode, GLsizei count, GLenum type,
+                       const GLvoid * indices)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
-    static const uint16_t end_cmd[2]   = { 4, X_GLrop_End };
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   static const uint16_t begin_cmd[2] = { 8, X_GLrop_Begin };
+   static const uint16_t end_cmd[2] = { 4, X_GLrop_End };
 
-    GLubyte * pc;
-    size_t single_vertex_size;
-    unsigned  i;
+   GLubyte *pc;
+   size_t single_vertex_size;
+   unsigned i;
 
 
-    single_vertex_size = calculate_single_vertex_size_none( arrays );
+   single_vertex_size = calculate_single_vertex_size_none(arrays);
 
 
-    if ( (gc->pc + single_vertex_size) >= gc->bufEnd ) {
-	gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   if ((gc->pc + single_vertex_size) >= gc->bufEnd) {
+      gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
+   }
 
-    pc = gc->pc;
+   pc = gc->pc;
 
-    (void) memcpy( pc, begin_cmd, 4 );
-    *(int *)(pc +  4) = mode;
+   (void) memcpy(pc, begin_cmd, 4);
+   *(int *) (pc + 4) = mode;
 
-    pc += 8;
+   pc += 8;
 
-    for ( i = 0 ; i < count ; i++ ) {
-	unsigned  index = 0;
+   for (i = 0; i < count; i++) {
+      unsigned index = 0;
 
-	if ( (pc + single_vertex_size) >= gc->bufEnd ) {
-	    pc = __glXFlushRenderBuffer(gc, gc->pc);
-	}
+      if ((pc + single_vertex_size) >= gc->bufEnd) {
+         pc = __glXFlushRenderBuffer(gc, pc);
+      }
 
-	switch( type ) {
-	case GL_UNSIGNED_INT:
-	    index = (unsigned) (((GLuint *) indices)[i]);
-	    break;
-	case GL_UNSIGNED_SHORT:
-	    index = (unsigned) (((GLushort *) indices)[i]);
-	    break;
-	case GL_UNSIGNED_BYTE:
-	    index = (unsigned) (((GLubyte *) indices)[i]);
-	    break;
-	}
-	pc = emit_element_none( pc, arrays, index );
-    }
+      switch (type) {
+      case GL_UNSIGNED_INT:
+         index = (unsigned) (((GLuint *) indices)[i]);
+         break;
+      case GL_UNSIGNED_SHORT:
+         index = (unsigned) (((GLushort *) indices)[i]);
+         break;
+      case GL_UNSIGNED_BYTE:
+         index = (unsigned) (((GLubyte *) indices)[i]);
+         break;
+      }
+      pc = emit_element_none(pc, arrays, index);
+   }
 
-    if ( (pc + 4) >= gc->bufEnd ) {
-	pc = __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   if ((pc + 4) >= gc->bufEnd) {
+      pc = __glXFlushRenderBuffer(gc, pc);
+   }
 
-    (void) memcpy( pc, end_cmd, 4 );
-    pc += 4;
+   (void) memcpy(pc, end_cmd, 4);
+   pc += 4;
 
-    gc->pc = pc;
-    if ( gc->pc > gc->limit ) {
-	(void) __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   gc->pc = pc;
+   if (gc->pc > gc->limit) {
+      (void) __glXFlushRenderBuffer(gc, gc->pc);
+   }
 }
 
 
 /**
  */
 void
-emit_DrawElements_old( GLenum mode, GLsizei count, GLenum type,
-		       const GLvoid *indices )
+emit_DrawElements_old(GLenum mode, GLsizei count, GLenum type,
+                      const GLvoid * indices)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-
-    GLubyte * pc;
-    size_t elements_per_request;
-    size_t total_requests = 0;
-    unsigned i;
-    unsigned req;
-
-
-    pc = emit_DrawArrays_header_old( gc, arrays, & elements_per_request,
-				     & total_requests, mode, count);
-
-    
-    /* Write the arrays.
-     */
-
-    req = 2;
-    while ( count > 0 ) {
-	if ( count < elements_per_request ) {
-	    elements_per_request = count;
-	}
-
-	switch( type ) {
-	case GL_UNSIGNED_INT: {
-	    const GLuint   * ui_ptr = (const GLuint   *) indices;
-
-	    for ( i = 0 ; i < elements_per_request ; i++ ) {
-		const GLint index = (GLint) *(ui_ptr++);
-		pc = emit_element_old( pc, arrays, index );
-	    }
-	    break;
-	}
-	case GL_UNSIGNED_SHORT: {
-	    const GLushort * us_ptr = (const GLushort *) indices;
-
-	    for ( i = 0 ; i < elements_per_request ; i++ ) {
-		const GLint index = (GLint) *(us_ptr++);
-		pc = emit_element_old( pc, arrays, index );
-	    }
-	    break;
-	}
-	case GL_UNSIGNED_BYTE: {
-	    const GLubyte  * ub_ptr = (const GLubyte  *) indices;
-
-	    for ( i = 0 ; i < elements_per_request ; i++ ) {
-		const GLint index = (GLint) *(ub_ptr++);
-		pc = emit_element_old( pc, arrays, index );
-	    }
-	    break;
-	}
-	}
-
-	if ( total_requests != 0 ) {
-	    __glXSendLargeChunk( gc, req, total_requests, gc->pc,
-				 pc - gc->pc );
-	    pc = gc->pc;
-	    req++;
-	}
-
-	count -= elements_per_request;
-    }
-
-
-    assert( (total_requests == 0) || ((req - 1) == total_requests) );
-
-    if ( total_requests == 0 ) {
-	assert( pc <= gc->bufEnd );
-
-	gc->pc = pc;
-	if ( gc->pc > gc->limit ) {
-	    (void) __glXFlushRenderBuffer(gc, gc->pc);
-	}
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+
+   GLubyte *pc;
+   size_t elements_per_request;
+   unsigned total_requests = 0;
+   unsigned i;
+   unsigned req;
+   unsigned req_element = 0;
+
+
+   pc = emit_DrawArrays_header_old(gc, arrays, &elements_per_request,
+                                   &total_requests, mode, count);
+
+
+   /* Write the arrays.
+    */
+
+   req = 2;
+   while (count > 0) {
+      if (count < elements_per_request) {
+         elements_per_request = count;
+      }
+
+      switch (type) {
+      case GL_UNSIGNED_INT:{
+            const GLuint *ui_ptr = (const GLuint *) indices + req_element;
+
+            for (i = 0; i < elements_per_request; i++) {
+               const GLint index = (GLint) * (ui_ptr++);
+               pc = emit_element_old(pc, arrays, index);
+            }
+            break;
+         }
+      case GL_UNSIGNED_SHORT:{
+            const GLushort *us_ptr = (const GLushort *) indices + req_element;
+
+            for (i = 0; i < elements_per_request; i++) {
+               const GLint index = (GLint) * (us_ptr++);
+               pc = emit_element_old(pc, arrays, index);
+            }
+            break;
+         }
+      case GL_UNSIGNED_BYTE:{
+            const GLubyte *ub_ptr = (const GLubyte *) indices + req_element;
+
+            for (i = 0; i < elements_per_request; i++) {
+               const GLint index = (GLint) * (ub_ptr++);
+               pc = emit_element_old(pc, arrays, index);
+            }
+            break;
+         }
+      }
+
+      if (total_requests != 0) {
+         __glXSendLargeChunk(gc, req, total_requests, gc->pc, pc - gc->pc);
+         pc = gc->pc;
+         req++;
+      }
+
+      count -= elements_per_request;
+      req_element += elements_per_request;
+   }
+
+
+   assert((total_requests == 0) || ((req - 1) == total_requests));
+
+   if (total_requests == 0) {
+      assert(pc <= gc->bufEnd);
+
+      gc->pc = pc;
+      if (gc->pc > gc->limit) {
+         (void) __glXFlushRenderBuffer(gc, gc->pc);
+      }
+   }
 }
 
 
@@ -850,26 +874,26 @@ emit_DrawElements_old( GLenum mode, GLsizei count, GLenum type,
  * \c GL_TRUE if the argument is valid, \c GL_FALSE if is not.
  */
 static GLboolean
-validate_mode(__GLXcontext *gc, GLenum mode)
+validate_mode(__GLXcontext * gc, GLenum mode)
 {
-    switch(mode) {
-      case GL_POINTS:
-      case GL_LINE_STRIP:
-      case GL_LINE_LOOP:
-      case GL_LINES:
-      case GL_TRIANGLE_STRIP:
-      case GL_TRIANGLE_FAN:
-      case GL_TRIANGLES:
-      case GL_QUAD_STRIP:
-      case GL_QUADS:
-      case GL_POLYGON:
-	break;
-      default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return GL_FALSE;
-    }
-    
-    return GL_TRUE;
+   switch (mode) {
+   case GL_POINTS:
+   case GL_LINE_STRIP:
+   case GL_LINE_LOOP:
+   case GL_LINES:
+   case GL_TRIANGLE_STRIP:
+   case GL_TRIANGLE_FAN:
+   case GL_TRIANGLES:
+   case GL_QUAD_STRIP:
+   case GL_QUADS:
+   case GL_POLYGON:
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
 }
 
 
@@ -883,13 +907,13 @@ validate_mode(__GLXcontext *gc, GLenum mode)
  * \c GL_TRUE if the argument is valid, \c GL_FALSE if it is not.
  */
 static GLboolean
-validate_count(__GLXcontext *gc, GLsizei count)
+validate_count(__GLXcontext * gc, GLsizei count)
 {
-    if (count < 0) {
-	__glXSetError(gc, GL_INVALID_VALUE);
-    }
+   if (count < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+   }
 
-    return (count > 0);
+   return (count > 0);
 }
 
 
@@ -901,154 +925,161 @@ validate_count(__GLXcontext *gc, GLsizei count)
  * \returns
  * \c GL_TRUE if the argument is valid, \c GL_FALSE if it is not.
  */
-static GLboolean validate_type(__GLXcontext *gc, GLenum type)
+static GLboolean
+validate_type(__GLXcontext * gc, GLenum type)
 {
-    switch( type ) {
-    case GL_UNSIGNED_INT:
-    case GL_UNSIGNED_SHORT:
-    case GL_UNSIGNED_BYTE:
-	return GL_TRUE;
-     default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return GL_FALSE;
-    }
+   switch (type) {
+   case GL_UNSIGNED_INT:
+   case GL_UNSIGNED_SHORT:
+   case GL_UNSIGNED_BYTE:
+      return GL_TRUE;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return GL_FALSE;
+   }
 }
 
 
-void __indirect_glDrawArrays(GLenum mode, GLint first, GLsizei count)
+void
+__indirect_glDrawArrays(GLenum mode, GLint first, GLsizei count)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    
-
-    if ( validate_mode(gc, mode) && validate_count(gc, count) ) {
-	if ( ! arrays->array_info_cache_valid ) {
-	    fill_array_info_cache( arrays );
-	}
-
-	arrays->DrawArrays(mode, first, count);
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+
+
+   if (validate_mode(gc, mode) && validate_count(gc, count)) {
+      if (!arrays->array_info_cache_valid) {
+         fill_array_info_cache(arrays);
+      }
+
+      arrays->DrawArrays(mode, first, count);
+   }
 }
 
 
-void __indirect_glArrayElement(GLint index)
+void
+__indirect_glArrayElement(GLint index)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
 
-    size_t single_vertex_size;
+   size_t single_vertex_size;
 
 
-    single_vertex_size = calculate_single_vertex_size_none( arrays );
+   single_vertex_size = calculate_single_vertex_size_none(arrays);
 
-    if ( (gc->pc + single_vertex_size) >= gc->bufEnd ) {
-	gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   if ((gc->pc + single_vertex_size) >= gc->bufEnd) {
+      gc->pc = __glXFlushRenderBuffer(gc, gc->pc);
+   }
 
-    gc->pc = emit_element_none( gc->pc, arrays, index );
+   gc->pc = emit_element_none(gc->pc, arrays, index);
 
-    if ( gc->pc > gc->limit ) {
-	(void) __glXFlushRenderBuffer(gc, gc->pc);
-    }
+   if (gc->pc > gc->limit) {
+      (void) __glXFlushRenderBuffer(gc, gc->pc);
+   }
 }
 
 
-void __indirect_glDrawElements(GLenum mode, GLsizei count, GLenum type,
-		    const GLvoid *indices)
+void
+__indirect_glDrawElements(GLenum mode, GLsizei count, GLenum type,
+                          const GLvoid * indices)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
 
 
-    if ( validate_mode(gc, mode) && validate_count(gc, count)
-	 && validate_type(gc, type) ) {
-	if ( ! arrays->array_info_cache_valid ) {
-	    fill_array_info_cache( arrays );
-	}
+   if (validate_mode(gc, mode) && validate_count(gc, count)
+       && validate_type(gc, type)) {
+      if (!arrays->array_info_cache_valid) {
+         fill_array_info_cache(arrays);
+      }
 
-	arrays->DrawElements(mode, count, type, indices);
-    }
+      arrays->DrawElements(mode, count, type, indices);
+   }
 }
 
 
-void __indirect_glDrawRangeElements(GLenum mode, GLuint start, GLuint end,
-				    GLsizei count, GLenum type,
-				    const GLvoid *indices)
+void
+__indirect_glDrawRangeElements(GLenum mode, GLuint start, GLuint end,
+                               GLsizei count, GLenum type,
+                               const GLvoid * indices)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
 
 
-    if ( validate_mode(gc, mode) && validate_count(gc, count)
-	 && validate_type(gc, type) ) {
-	if (end < start) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
+   if (validate_mode(gc, mode) && validate_count(gc, count)
+       && validate_type(gc, type)) {
+      if (end < start) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
 
-	if ( ! arrays->array_info_cache_valid ) {
-	    fill_array_info_cache( arrays );
-	}
+      if (!arrays->array_info_cache_valid) {
+         fill_array_info_cache(arrays);
+      }
 
-	arrays->DrawElements(mode, count, type, indices);
-    }
+      arrays->DrawElements(mode, count, type, indices);
+   }
 }
 
 
-void __indirect_glMultiDrawArraysEXT(GLenum mode, GLint *first, GLsizei *count,
-		       GLsizei primcount)
+void
+__indirect_glMultiDrawArraysEXT(GLenum mode, GLint * first, GLsizei * count,
+                                GLsizei primcount)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    GLsizei  i;
-
-
-    if ( validate_mode(gc, mode) ) {
-	if ( ! arrays->array_info_cache_valid ) {
-	    fill_array_info_cache( arrays );
-	}
-
-	for ( i = 0 ; i < primcount ; i++ ) {
-	    if ( validate_count( gc, count[i] ) ) {
-		arrays->DrawArrays(mode, first[i], count[i]);
-	    }
-	}
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   GLsizei i;
+
+
+   if (validate_mode(gc, mode)) {
+      if (!arrays->array_info_cache_valid) {
+         fill_array_info_cache(arrays);
+      }
+
+      for (i = 0; i < primcount; i++) {
+         if (validate_count(gc, count[i])) {
+            arrays->DrawArrays(mode, first[i], count[i]);
+         }
+      }
+   }
 }
 
 
-void __indirect_glMultiDrawElementsEXT(GLenum mode, const GLsizei *count,
-			 GLenum type, const GLvoid ** indices,
-			 GLsizei primcount)
+void
+__indirect_glMultiDrawElementsEXT(GLenum mode, const GLsizei * count,
+                                  GLenum type, const GLvoid ** indices,
+                                  GLsizei primcount)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    const __GLXattribute * state = 
-       (const __GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    GLsizei  i;
-
-
-    if ( validate_mode(gc, mode) && validate_type(gc, type) ) {
-	if ( ! arrays->array_info_cache_valid ) {
-	    fill_array_info_cache( arrays );
-	}
-
-	for ( i = 0 ; i < primcount ; i++ ) {
-	    if ( validate_count( gc, count[i] ) ) {
-		arrays->DrawElements(mode, count[i], type, indices[i]);
-	    }
-	}
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   const __GLXattribute *state =
+      (const __GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   GLsizei i;
+
+
+   if (validate_mode(gc, mode) && validate_type(gc, type)) {
+      if (!arrays->array_info_cache_valid) {
+         fill_array_info_cache(arrays);
+      }
+
+      for (i = 0; i < primcount; i++) {
+         if (validate_count(gc, count[i])) {
+            arrays->DrawElements(mode, count[i], type, indices[i]);
+         }
+      }
+   }
 }
 
 
@@ -1070,475 +1101,577 @@ void __indirect_glMultiDrawElementsEXT(GLenum mode, const GLsizei *count,
     } while(0)
 
 
-void __indirect_glVertexPointer( GLint size, GLenum type, GLsizei stride,
-				 const GLvoid * pointer )
+void
+__indirect_glVertexPointer(GLint size, GLenum type, GLsizei stride,
+                           const GLvoid * pointer)
 {
-    static const uint16_t short_ops[5] = {
-	0, 0, X_GLrop_Vertex2sv, X_GLrop_Vertex3sv, X_GLrop_Vertex4sv
-    };
-    static const uint16_t int_ops[5] = {
-	0, 0, X_GLrop_Vertex2iv, X_GLrop_Vertex3iv, X_GLrop_Vertex4iv
-    };
-    static const uint16_t float_ops[5] = {
-	0, 0, X_GLrop_Vertex2fv, X_GLrop_Vertex3fv, X_GLrop_Vertex4fv
-    };
-    static const uint16_t double_ops[5] = {
-	0, 0, X_GLrop_Vertex2dv, X_GLrop_Vertex3dv, X_GLrop_Vertex4dv
-    };
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (size < 2 || size > 4 || stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-    switch ( type ) {
-    case GL_SHORT:	opcode = short_ops[size]; break;
-    case GL_INT:	opcode = int_ops[size]; break;
-    case GL_FLOAT:	opcode = float_ops[size]; break;
-    case GL_DOUBLE:	opcode = double_ops[size]; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_VERTEX_ARRAY, 0 );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_FALSE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   static const uint16_t short_ops[5] = {
+      0, 0, X_GLrop_Vertex2sv, X_GLrop_Vertex3sv, X_GLrop_Vertex4sv
+   };
+   static const uint16_t int_ops[5] = {
+      0, 0, X_GLrop_Vertex2iv, X_GLrop_Vertex3iv, X_GLrop_Vertex4iv
+   };
+   static const uint16_t float_ops[5] = {
+      0, 0, X_GLrop_Vertex2fv, X_GLrop_Vertex3fv, X_GLrop_Vertex4fv
+   };
+   static const uint16_t double_ops[5] = {
+      0, 0, X_GLrop_Vertex2dv, X_GLrop_Vertex3dv, X_GLrop_Vertex4dv
+   };
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (size < 2 || size > 4 || stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_SHORT:
+      opcode = short_ops[size];
+      break;
+   case GL_INT:
+      opcode = int_ops[size];
+      break;
+   case GL_FLOAT:
+      opcode = float_ops[size];
+      break;
+   case GL_DOUBLE:
+      opcode = double_ops[size];
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_VERTEX_ARRAY, 0);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, size, GL_FALSE, 4,
+                          opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glNormalPointer( GLenum type, GLsizei stride,
-				 const GLvoid * pointer )
+void
+__indirect_glNormalPointer(GLenum type, GLsizei stride,
+                           const GLvoid * pointer)
 {
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-
-    switch ( type ) {
-    case GL_BYTE:	opcode = X_GLrop_Normal3bv; break;
-    case GL_SHORT:	opcode = X_GLrop_Normal3sv; break;
-    case GL_INT:	opcode = X_GLrop_Normal3iv; break;
-    case GL_FLOAT:	opcode = X_GLrop_Normal3fv; break;
-    case GL_DOUBLE:	opcode = X_GLrop_Normal3dv; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_NORMAL_ARRAY, 0 );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 3, GL_TRUE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_BYTE:
+      opcode = X_GLrop_Normal3bv;
+      break;
+   case GL_SHORT:
+      opcode = X_GLrop_Normal3sv;
+      break;
+   case GL_INT:
+      opcode = X_GLrop_Normal3iv;
+      break;
+   case GL_FLOAT:
+      opcode = X_GLrop_Normal3fv;
+      break;
+   case GL_DOUBLE:
+      opcode = X_GLrop_Normal3dv;
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_NORMAL_ARRAY, 0);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, 3, GL_TRUE, 4, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glColorPointer( GLint size, GLenum type, GLsizei stride,
-				const GLvoid * pointer )
+void
+__indirect_glColorPointer(GLint size, GLenum type, GLsizei stride,
+                          const GLvoid * pointer)
 {
-    static const uint16_t byte_ops[5] = {
-	0, 0, 0, X_GLrop_Color3bv, X_GLrop_Color4bv
-    };
-    static const uint16_t ubyte_ops[5] = {
-	0, 0, 0, X_GLrop_Color3ubv, X_GLrop_Color4ubv
-    };
-    static const uint16_t short_ops[5] = {
-	0, 0, 0, X_GLrop_Color3sv, X_GLrop_Color4sv
-    };
-    static const uint16_t ushort_ops[5] = {
-	0, 0, 0, X_GLrop_Color3usv, X_GLrop_Color4usv
-    };
-    static const uint16_t int_ops[5] = {
-	0, 0, 0, X_GLrop_Color3iv, X_GLrop_Color4iv
-    };
-    static const uint16_t uint_ops[5] = {
-	0, 0, 0, X_GLrop_Color3uiv, X_GLrop_Color4uiv
-    };
-    static const uint16_t float_ops[5] = {
-	0, 0, 0, X_GLrop_Color3fv, X_GLrop_Color4fv
-    };
-    static const uint16_t double_ops[5] = {
-	0, 0, 0, X_GLrop_Color3dv, X_GLrop_Color4dv
-    };
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (size < 3 || size > 4 || stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-    switch ( type ) {
-    case GL_BYTE:		opcode = byte_ops[size]; break;
-    case GL_UNSIGNED_BYTE:	opcode = ubyte_ops[size]; break;
-    case GL_SHORT:		opcode = short_ops[size]; break;
-    case GL_UNSIGNED_SHORT:	opcode = ushort_ops[size]; break;
-    case GL_INT:		opcode = int_ops[size]; break;
-    case GL_UNSIGNED_INT:	opcode = uint_ops[size]; break;
-    case GL_FLOAT:		opcode = float_ops[size]; break;
-    case GL_DOUBLE:		opcode = double_ops[size]; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_COLOR_ARRAY, 0 );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_TRUE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   static const uint16_t byte_ops[5] = {
+      0, 0, 0, X_GLrop_Color3bv, X_GLrop_Color4bv
+   };
+   static const uint16_t ubyte_ops[5] = {
+      0, 0, 0, X_GLrop_Color3ubv, X_GLrop_Color4ubv
+   };
+   static const uint16_t short_ops[5] = {
+      0, 0, 0, X_GLrop_Color3sv, X_GLrop_Color4sv
+   };
+   static const uint16_t ushort_ops[5] = {
+      0, 0, 0, X_GLrop_Color3usv, X_GLrop_Color4usv
+   };
+   static const uint16_t int_ops[5] = {
+      0, 0, 0, X_GLrop_Color3iv, X_GLrop_Color4iv
+   };
+   static const uint16_t uint_ops[5] = {
+      0, 0, 0, X_GLrop_Color3uiv, X_GLrop_Color4uiv
+   };
+   static const uint16_t float_ops[5] = {
+      0, 0, 0, X_GLrop_Color3fv, X_GLrop_Color4fv
+   };
+   static const uint16_t double_ops[5] = {
+      0, 0, 0, X_GLrop_Color3dv, X_GLrop_Color4dv
+   };
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (size < 3 || size > 4 || stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_BYTE:
+      opcode = byte_ops[size];
+      break;
+   case GL_UNSIGNED_BYTE:
+      opcode = ubyte_ops[size];
+      break;
+   case GL_SHORT:
+      opcode = short_ops[size];
+      break;
+   case GL_UNSIGNED_SHORT:
+      opcode = ushort_ops[size];
+      break;
+   case GL_INT:
+      opcode = int_ops[size];
+      break;
+   case GL_UNSIGNED_INT:
+      opcode = uint_ops[size];
+      break;
+   case GL_FLOAT:
+      opcode = float_ops[size];
+      break;
+   case GL_DOUBLE:
+      opcode = double_ops[size];
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_COLOR_ARRAY, 0);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, size, GL_TRUE, 4, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glIndexPointer( GLenum type, GLsizei stride,
-				const GLvoid * pointer )
+void
+__indirect_glIndexPointer(GLenum type, GLsizei stride, const GLvoid * pointer)
 {
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-    switch ( type ) {
-    case GL_UNSIGNED_BYTE:	opcode = X_GLrop_Indexubv; break;
-    case GL_SHORT:		opcode = X_GLrop_Indexsv; break;
-    case GL_INT:		opcode = X_GLrop_Indexiv; break;
-    case GL_FLOAT:		opcode = X_GLrop_Indexfv; break;
-    case GL_DOUBLE:		opcode = X_GLrop_Indexdv; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_INDEX_ARRAY, 0 );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 1, GL_FALSE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_UNSIGNED_BYTE:
+      opcode = X_GLrop_Indexubv;
+      break;
+   case GL_SHORT:
+      opcode = X_GLrop_Indexsv;
+      break;
+   case GL_INT:
+      opcode = X_GLrop_Indexiv;
+      break;
+   case GL_FLOAT:
+      opcode = X_GLrop_Indexfv;
+      break;
+   case GL_DOUBLE:
+      opcode = X_GLrop_Indexdv;
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_INDEX_ARRAY, 0);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, 1, GL_FALSE, 4, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glEdgeFlagPointer( GLsizei stride, const GLvoid * pointer )
+void
+__indirect_glEdgeFlagPointer(GLsizei stride, const GLvoid * pointer)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-
-    a = get_array_entry( arrays, GL_EDGE_FLAG_ARRAY, 0 );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, GL_UNSIGNED_BYTE, stride, 1, GL_FALSE,
-			    4, X_GLrop_EdgeFlagv );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+
+   a = get_array_entry(arrays, GL_EDGE_FLAG_ARRAY, 0);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, GL_UNSIGNED_BYTE, stride, 1, GL_FALSE,
+                          4, X_GLrop_EdgeFlagv);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glTexCoordPointer( GLint size, GLenum type, GLsizei stride,
-				   const GLvoid * pointer )
+void
+__indirect_glTexCoordPointer(GLint size, GLenum type, GLsizei stride,
+                             const GLvoid * pointer)
 {
-    static const uint16_t short_ops[5] = {
-	0, X_GLrop_TexCoord1sv, X_GLrop_TexCoord2sv, X_GLrop_TexCoord3sv, X_GLrop_TexCoord4sv
-    };
-    static const uint16_t int_ops[5] = {
-	0, X_GLrop_TexCoord1iv, X_GLrop_TexCoord2iv, X_GLrop_TexCoord3iv, X_GLrop_TexCoord4iv
-    };
-    static const uint16_t float_ops[5] = {
-	0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2fv, X_GLrop_TexCoord3fv, X_GLrop_TexCoord4fv
-    };
-    static const uint16_t double_ops[5] = {
-	0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2dv, X_GLrop_TexCoord3dv, X_GLrop_TexCoord4dv
-    };
-
-    static const uint16_t mshort_ops[5] = {
-	0, X_GLrop_MultiTexCoord1svARB, X_GLrop_MultiTexCoord2svARB, X_GLrop_MultiTexCoord3svARB, X_GLrop_MultiTexCoord4svARB
-    };
-    static const uint16_t mint_ops[5] = {
-	0, X_GLrop_MultiTexCoord1ivARB, X_GLrop_MultiTexCoord2ivARB, X_GLrop_MultiTexCoord3ivARB, X_GLrop_MultiTexCoord4ivARB
-    };
-    static const uint16_t mfloat_ops[5] = {
-	0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2fvARB, X_GLrop_MultiTexCoord3fvARB, X_GLrop_MultiTexCoord4fvARB
-    };
-    static const uint16_t mdouble_ops[5] = {
-	0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2dvARB, X_GLrop_MultiTexCoord3dvARB, X_GLrop_MultiTexCoord4dvARB
-    };
-
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-    unsigned header_size;
-    unsigned index;
-
-
-    if (size < 1 || size > 4 || stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-    index = arrays->active_texture_unit;
-    if ( index == 0 ) {
-	switch ( type ) {
-	case GL_SHORT:		opcode = short_ops[size]; break;
-	case GL_INT:		opcode = int_ops[size]; break;
-	case GL_FLOAT:		opcode = float_ops[size]; break;
-	case GL_DOUBLE:		opcode = double_ops[size]; break;
-	default:
-	    __glXSetError(gc, GL_INVALID_ENUM);
-	    return;
-	}
-
-	header_size = 4;
-    }
-    else {
-	switch ( type ) {
-	case GL_SHORT:		opcode = mshort_ops[size]; break;
-	case GL_INT:		opcode = mint_ops[size]; break;
-	case GL_FLOAT:		opcode = mfloat_ops[size]; break;
-	case GL_DOUBLE:		opcode = mdouble_ops[size]; break;
-	default:
-	    __glXSetError(gc, GL_INVALID_ENUM);
-	    return;
-	}
-
-	header_size = 8;
-    }
-
-    a = get_array_entry( arrays, GL_TEXTURE_COORD_ARRAY, index );
-    assert( a != NULL );
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_FALSE,
-			    header_size, opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   static const uint16_t short_ops[5] = {
+      0, X_GLrop_TexCoord1sv, X_GLrop_TexCoord2sv, X_GLrop_TexCoord3sv,
+         X_GLrop_TexCoord4sv
+   };
+   static const uint16_t int_ops[5] = {
+      0, X_GLrop_TexCoord1iv, X_GLrop_TexCoord2iv, X_GLrop_TexCoord3iv,
+         X_GLrop_TexCoord4iv
+   };
+   static const uint16_t float_ops[5] = {
+      0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2fv, X_GLrop_TexCoord3fv,
+         X_GLrop_TexCoord4fv
+   };
+   static const uint16_t double_ops[5] = {
+      0, X_GLrop_TexCoord1dv, X_GLrop_TexCoord2dv, X_GLrop_TexCoord3dv,
+         X_GLrop_TexCoord4dv
+   };
+
+   static const uint16_t mshort_ops[5] = {
+      0, X_GLrop_MultiTexCoord1svARB, X_GLrop_MultiTexCoord2svARB,
+         X_GLrop_MultiTexCoord3svARB, X_GLrop_MultiTexCoord4svARB
+   };
+   static const uint16_t mint_ops[5] = {
+      0, X_GLrop_MultiTexCoord1ivARB, X_GLrop_MultiTexCoord2ivARB,
+         X_GLrop_MultiTexCoord3ivARB, X_GLrop_MultiTexCoord4ivARB
+   };
+   static const uint16_t mfloat_ops[5] = {
+      0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2fvARB,
+         X_GLrop_MultiTexCoord3fvARB, X_GLrop_MultiTexCoord4fvARB
+   };
+   static const uint16_t mdouble_ops[5] = {
+      0, X_GLrop_MultiTexCoord1dvARB, X_GLrop_MultiTexCoord2dvARB,
+         X_GLrop_MultiTexCoord3dvARB, X_GLrop_MultiTexCoord4dvARB
+   };
+
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+   unsigned header_size;
+   unsigned index;
+
+
+   if (size < 1 || size > 4 || stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   index = arrays->active_texture_unit;
+   if (index == 0) {
+      switch (type) {
+      case GL_SHORT:
+         opcode = short_ops[size];
+         break;
+      case GL_INT:
+         opcode = int_ops[size];
+         break;
+      case GL_FLOAT:
+         opcode = float_ops[size];
+         break;
+      case GL_DOUBLE:
+         opcode = double_ops[size];
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_ENUM);
+         return;
+      }
+
+      header_size = 4;
+   }
+   else {
+      switch (type) {
+      case GL_SHORT:
+         opcode = mshort_ops[size];
+         break;
+      case GL_INT:
+         opcode = mint_ops[size];
+         break;
+      case GL_FLOAT:
+         opcode = mfloat_ops[size];
+         break;
+      case GL_DOUBLE:
+         opcode = mdouble_ops[size];
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_ENUM);
+         return;
+      }
+
+      header_size = 8;
+   }
+
+   a = get_array_entry(arrays, GL_TEXTURE_COORD_ARRAY, index);
+   assert(a != NULL);
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, size, GL_FALSE,
+                          header_size, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glSecondaryColorPointerEXT( GLint size, GLenum type, GLsizei stride,
-				const GLvoid * pointer )
+void
+__indirect_glSecondaryColorPointerEXT(GLint size, GLenum type, GLsizei stride,
+                                      const GLvoid * pointer)
 {
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (size != 3 || stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-    
-    switch ( type ) {
-    case GL_BYTE:		opcode = 4126; break;
-    case GL_UNSIGNED_BYTE:	opcode = 4131; break;
-    case GL_SHORT:		opcode = 4127; break;
-    case GL_UNSIGNED_SHORT:	opcode = 4132; break;
-    case GL_INT:		opcode = 4128; break;
-    case GL_UNSIGNED_INT:	opcode = 4133; break;
-    case GL_FLOAT:		opcode = 4129; break;
-    case GL_DOUBLE:		opcode = 4130; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_SECONDARY_COLOR_ARRAY, 0 );
-    if ( a == NULL ) {
-        __glXSetError(gc, GL_INVALID_OPERATION);
-        return;
-    }
-
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, GL_TRUE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (size != 3 || stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_BYTE:
+      opcode = 4126;
+      break;
+   case GL_UNSIGNED_BYTE:
+      opcode = 4131;
+      break;
+   case GL_SHORT:
+      opcode = 4127;
+      break;
+   case GL_UNSIGNED_SHORT:
+      opcode = 4132;
+      break;
+   case GL_INT:
+      opcode = 4128;
+      break;
+   case GL_UNSIGNED_INT:
+      opcode = 4133;
+      break;
+   case GL_FLOAT:
+      opcode = 4129;
+      break;
+   case GL_DOUBLE:
+      opcode = 4130;
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_SECONDARY_COLOR_ARRAY, 0);
+   if (a == NULL) {
+      __glXSetError(gc, GL_INVALID_OPERATION);
+      return;
+   }
+
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, size, GL_TRUE, 4, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glFogCoordPointerEXT( GLenum type, GLsizei stride,
-				const GLvoid * pointer )
+void
+__indirect_glFogCoordPointerEXT(GLenum type, GLsizei stride,
+                                const GLvoid * pointer)
 {
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-
-
-    if (stride < 0) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-
-    switch ( type ) {
-    case GL_FLOAT:		opcode = 4124; break;
-    case GL_DOUBLE:		opcode = 4125; break;
-    default:
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    a = get_array_entry( arrays, GL_FOG_COORD_ARRAY, 0 );
-    if ( a == NULL ) {
-        __glXSetError(gc, GL_INVALID_OPERATION);
-        return;
-    }
-
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, 1, GL_FALSE, 4,
-			    opcode );
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
+
+   if (stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   switch (type) {
+   case GL_FLOAT:
+      opcode = 4124;
+      break;
+   case GL_DOUBLE:
+      opcode = 4125;
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   a = get_array_entry(arrays, GL_FOG_COORD_ARRAY, 0);
+   if (a == NULL) {
+      __glXSetError(gc, GL_INVALID_OPERATION);
+      return;
+   }
+
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, 1, GL_FALSE, 4, opcode);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
-void __indirect_glVertexAttribPointerARB(GLuint index, GLint size,
-					 GLenum type, GLboolean normalized,
-					 GLsizei stride,
-					 const GLvoid * pointer)
+void
+__indirect_glVertexAttribPointerARB(GLuint index, GLint size,
+                                    GLenum type, GLboolean normalized,
+                                    GLsizei stride, const GLvoid * pointer)
 {
-    static const uint16_t short_ops[5]  = { 0, 4189, 4190, 4191, 4192 };
-    static const uint16_t float_ops[5]  = { 0, 4193, 4194, 4195, 4196 };
-    static const uint16_t double_ops[5] = { 0, 4197, 4198, 4199, 4200 };
-
-    uint16_t opcode;
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-    unsigned true_immediate_count;
-    unsigned true_immediate_size;
-
-
-    if ( (size < 1) || (size > 4) || (stride < 0) 
-	 || (index > arrays->num_vertex_program_attribs) ){
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-
-    if ( normalized && (type != GL_FLOAT) && (type != GL_DOUBLE)) {
-	switch( type ) {
-	case GL_BYTE:           opcode = X_GLrop_VertexAttrib4NbvARB;  break;
-	case GL_UNSIGNED_BYTE:  opcode = X_GLrop_VertexAttrib4NubvARB; break;
-	case GL_SHORT:          opcode = X_GLrop_VertexAttrib4NsvARB;  break;
-	case GL_UNSIGNED_SHORT: opcode = X_GLrop_VertexAttrib4NusvARB; break;
-	case GL_INT:            opcode = X_GLrop_VertexAttrib4NivARB;  break;
-	case GL_UNSIGNED_INT:   opcode = X_GLrop_VertexAttrib4NuivARB; break;
-	default:
-	    __glXSetError(gc, GL_INVALID_ENUM);
-	    return;
-	}
-	
-	true_immediate_count = 4;
-    }
-    else {
-	true_immediate_count = size;
-
-	switch( type ) {
-	case GL_BYTE:
-	    opcode = X_GLrop_VertexAttrib4bvARB;
-	    true_immediate_count = 4;
-	    break;
-	case GL_UNSIGNED_BYTE:
-	    opcode = X_GLrop_VertexAttrib4ubvARB;
-	    true_immediate_count = 4;
-	    break;
-	case GL_SHORT:
-	    opcode = short_ops[size];
-	    break;
-	case GL_UNSIGNED_SHORT:
-	    opcode = X_GLrop_VertexAttrib4usvARB;
-	    true_immediate_count = 4;
-	    break;
-	case GL_INT:
-            opcode = X_GLrop_VertexAttrib4ivARB;
-	    true_immediate_count = 4;
-	    break;
-	case GL_UNSIGNED_INT:
-	    opcode = X_GLrop_VertexAttrib4uivARB;
-	    true_immediate_count = 4;
-	    break;
-	case GL_FLOAT:
-	    opcode = float_ops[size];
-	    break;
-	case GL_DOUBLE:
-	    opcode = double_ops[size];
-	    break;
-	default:
-	    __glXSetError(gc, GL_INVALID_ENUM);
-	    return;
-	}
-    }
-
-    a = get_array_entry( arrays, GL_VERTEX_ATTRIB_ARRAY_POINTER, index );
-    if ( a == NULL ) {
-        __glXSetError(gc, GL_INVALID_OPERATION);
-        return;
-    }
-
-    COMMON_ARRAY_DATA_INIT( a, pointer, type, stride, size, normalized, 8,
-			    opcode );
-
-    true_immediate_size = __glXTypeSize(type) * true_immediate_count;
-    ((uint16_t *) (a)->header)[0] = __GLX_PAD(a->header_size 
-					      + true_immediate_size);
-
-    if ( a->enabled ) {
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   static const uint16_t short_ops[5] = { 0, 4189, 4190, 4191, 4192 };
+   static const uint16_t float_ops[5] = { 0, 4193, 4194, 4195, 4196 };
+   static const uint16_t double_ops[5] = { 0, 4197, 4198, 4199, 4200 };
+
+   uint16_t opcode;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+   unsigned true_immediate_count;
+   unsigned true_immediate_size;
+
+
+   if ((size < 1) || (size > 4) || (stride < 0)
+       || (index > arrays->num_vertex_program_attribs)) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+   if (normalized && (type != GL_FLOAT) && (type != GL_DOUBLE)) {
+      switch (type) {
+      case GL_BYTE:
+         opcode = X_GLrop_VertexAttrib4NbvARB;
+         break;
+      case GL_UNSIGNED_BYTE:
+         opcode = X_GLrop_VertexAttrib4NubvARB;
+         break;
+      case GL_SHORT:
+         opcode = X_GLrop_VertexAttrib4NsvARB;
+         break;
+      case GL_UNSIGNED_SHORT:
+         opcode = X_GLrop_VertexAttrib4NusvARB;
+         break;
+      case GL_INT:
+         opcode = X_GLrop_VertexAttrib4NivARB;
+         break;
+      case GL_UNSIGNED_INT:
+         opcode = X_GLrop_VertexAttrib4NuivARB;
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_ENUM);
+         return;
+      }
+
+      true_immediate_count = 4;
+   }
+   else {
+      true_immediate_count = size;
+
+      switch (type) {
+      case GL_BYTE:
+         opcode = X_GLrop_VertexAttrib4bvARB;
+         true_immediate_count = 4;
+         break;
+      case GL_UNSIGNED_BYTE:
+         opcode = X_GLrop_VertexAttrib4ubvARB;
+         true_immediate_count = 4;
+         break;
+      case GL_SHORT:
+         opcode = short_ops[size];
+         break;
+      case GL_UNSIGNED_SHORT:
+         opcode = X_GLrop_VertexAttrib4usvARB;
+         true_immediate_count = 4;
+         break;
+      case GL_INT:
+         opcode = X_GLrop_VertexAttrib4ivARB;
+         true_immediate_count = 4;
+         break;
+      case GL_UNSIGNED_INT:
+         opcode = X_GLrop_VertexAttrib4uivARB;
+         true_immediate_count = 4;
+         break;
+      case GL_FLOAT:
+         opcode = float_ops[size];
+         break;
+      case GL_DOUBLE:
+         opcode = double_ops[size];
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_ENUM);
+         return;
+      }
+   }
+
+   a = get_array_entry(arrays, GL_VERTEX_ATTRIB_ARRAY_POINTER, index);
+   if (a == NULL) {
+      __glXSetError(gc, GL_INVALID_OPERATION);
+      return;
+   }
+
+   COMMON_ARRAY_DATA_INIT(a, pointer, type, stride, size, normalized, 8,
+                          opcode);
+
+   true_immediate_size = __glXTypeSize(type) * true_immediate_count;
+   ((uint16_t *) (a)->header)[0] = __GLX_PAD(a->header_size
+                                             + true_immediate_size);
+
+   if (a->enabled) {
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 }
 
 
@@ -1551,295 +1684,302 @@ void __indirect_glVertexAttribPointerARB(GLuint index, GLint size,
  * client just sends all the data to the server and lets the server deal
  * with it.
  */
-void __indirect_glVertexAttribPointerNV( GLuint index, GLint size,
-					 GLenum type, GLsizei stride,
-					 const GLvoid * pointer)
+void
+__indirect_glVertexAttribPointerNV(GLuint index, GLint size,
+                                   GLenum type, GLsizei stride,
+                                   const GLvoid * pointer)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    GLboolean normalized = GL_FALSE;
-
-
-    switch( type ) {
-    case GL_UNSIGNED_BYTE:
-	if ( size != 4 ) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	normalized = GL_TRUE;
-
-    case GL_SHORT:
-    case GL_FLOAT:
-    case GL_DOUBLE:
-	__indirect_glVertexAttribPointerARB(index, size, type,
-					    normalized,
-					    stride, pointer);
-	return;
-    default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    }
+   __GLXcontext *gc = __glXGetCurrentContext();
+   GLboolean normalized = GL_FALSE;
+
+
+   switch (type) {
+   case GL_UNSIGNED_BYTE:
+      if (size != 4) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      normalized = GL_TRUE;
+
+   case GL_SHORT:
+   case GL_FLOAT:
+   case GL_DOUBLE:
+      __indirect_glVertexAttribPointerARB(index, size, type,
+                                          normalized, stride, pointer);
+      return;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
 }
 
 
-void __indirect_glClientActiveTextureARB(GLenum texture)
+void
+__indirect_glClientActiveTextureARB(GLenum texture)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    __GLXattribute * const state = (__GLXattribute *)(gc->client_state_private);
-    struct array_state_vector * const arrays = state->array_state;
-    const GLint unit = (GLint) texture - GL_TEXTURE0;
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   __GLXattribute *const state =
+      (__GLXattribute *) (gc->client_state_private);
+   struct array_state_vector *const arrays = state->array_state;
+   const GLint unit = (GLint) texture - GL_TEXTURE0;
 
 
-    if ( (unit < 0) || (unit >= arrays->num_texture_units) ) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    }
+   if ((unit < 0) || (unit >= arrays->num_texture_units)) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
 
-    arrays->active_texture_unit = unit;
+   arrays->active_texture_unit = unit;
 }
 
 
 /**
+ * Modify the enable state for the selected array
  */
 GLboolean
-__glXSetArrayEnable( __GLXattribute * state,
-		     GLenum key, unsigned index, GLboolean enable )
+__glXSetArrayEnable(__GLXattribute * state, GLenum key, unsigned index,
+                    GLboolean enable)
 {
-    struct array_state_vector * arrays = state->array_state;
-    struct array_state * a;
-    
+   struct array_state_vector *arrays = state->array_state;
+   struct array_state *a;
+
 
-    if ( key == GL_TEXTURE_COORD_ARRAY ) {
-	index = arrays->active_texture_unit;
-    }
+   /* Texture coordinate arrays have an implict index set when the
+    * application calls glClientActiveTexture.
+    */
+   if (key == GL_TEXTURE_COORD_ARRAY) {
+      index = arrays->active_texture_unit;
+   }
 
-    a = get_array_entry( arrays, key, index );
+   a = get_array_entry(arrays, key, index);
 
-    if ( (a != NULL) && (a->enabled != enable) ) {
-	a->enabled = enable;
-	arrays->array_info_cache_valid = GL_FALSE;
-    }
+   if ((a != NULL) && (a->enabled != enable)) {
+      a->enabled = enable;
+      arrays->array_info_cache_valid = GL_FALSE;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 void
-__glXArrayDisableAll( __GLXattribute * state )
+__glXArrayDisableAll(__GLXattribute * state)
 {
-    struct array_state_vector * arrays = state->array_state;
-    unsigned  i;
+   struct array_state_vector *arrays = state->array_state;
+   unsigned i;
 
 
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	arrays->arrays[i].enabled = GL_FALSE;
-    }
+   for (i = 0; i < arrays->num_arrays; i++) {
+      arrays->arrays[i].enabled = GL_FALSE;
+   }
 
-    arrays->array_info_cache_valid = GL_FALSE;
+   arrays->array_info_cache_valid = GL_FALSE;
 }
 
 
 /**
  */
 GLboolean
-__glXGetArrayEnable( const __GLXattribute * const state,
-		     GLenum key, unsigned index, GLintptr * dest )
+__glXGetArrayEnable(const __GLXattribute * const state,
+                    GLenum key, unsigned index, GLintptr * dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
-    if ( a != NULL ) {
-	*dest = (GLintptr) a->enabled;
-    }
+   if (a != NULL) {
+      *dest = (GLintptr) a->enabled;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLboolean
-__glXGetArrayType( const __GLXattribute * const state,
-		     GLenum key, unsigned index, GLintptr * dest )
+__glXGetArrayType(const __GLXattribute * const state,
+                  GLenum key, unsigned index, GLintptr * dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
-    if ( a != NULL ) {
-	*dest = (GLintptr) a->data_type;
-    }
+   if (a != NULL) {
+      *dest = (GLintptr) a->data_type;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLboolean
-__glXGetArraySize( const __GLXattribute * const state,
-		     GLenum key, unsigned index, GLintptr * dest )
+__glXGetArraySize(const __GLXattribute * const state,
+                  GLenum key, unsigned index, GLintptr * dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
-    if ( a != NULL ) {
-	*dest = (GLintptr) a->count;
-    }
+   if (a != NULL) {
+      *dest = (GLintptr) a->count;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLboolean
-__glXGetArrayStride( const __GLXattribute * const state,
-		     GLenum key, unsigned index, GLintptr * dest )
+__glXGetArrayStride(const __GLXattribute * const state,
+                    GLenum key, unsigned index, GLintptr * dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
-    if ( a != NULL ) {
-	*dest = (GLintptr) a->user_stride;
-    }
+   if (a != NULL) {
+      *dest = (GLintptr) a->user_stride;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLboolean
-__glXGetArrayPointer( const __GLXattribute * const state,
-		      GLenum key, unsigned index, void ** dest )
+__glXGetArrayPointer(const __GLXattribute * const state,
+                     GLenum key, unsigned index, void **dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
 
-    if ( a != NULL ) {
-	*dest = (void *) (a->data);
-    }
+   if (a != NULL) {
+      *dest = (void *) (a->data);
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLboolean
-__glXGetArrayNormalized( const __GLXattribute * const state,
-			 GLenum key, unsigned index, GLintptr * dest )
+__glXGetArrayNormalized(const __GLXattribute * const state,
+                        GLenum key, unsigned index, GLintptr * dest)
 {
-    const struct array_state_vector * arrays = state->array_state;
-    const struct array_state * a = get_array_entry( (struct array_state_vector *) arrays,
-						    key, index );
+   const struct array_state_vector *arrays = state->array_state;
+   const struct array_state *a =
+      get_array_entry((struct array_state_vector *) arrays,
+                      key, index);
 
 
-    if ( a != NULL ) {
-	*dest = (GLintptr) a->normalized;
-    }
+   if (a != NULL) {
+      *dest = (GLintptr) a->normalized;
+   }
 
-    return (a != NULL);
+   return (a != NULL);
 }
 
 
 /**
  */
 GLuint
-__glXGetActiveTextureUnit( const __GLXattribute * const state )
+__glXGetActiveTextureUnit(const __GLXattribute * const state)
 {
-    return state->array_state->active_texture_unit;
+   return state->array_state->active_texture_unit;
 }
 
 
 void
-__glXPushArrayState( __GLXattribute * state )
+__glXPushArrayState(__GLXattribute * state)
 {
-    struct array_state_vector * arrays = state->array_state;
-    struct array_stack_state * stack = & arrays->stack[ (arrays->stack_index * arrays->num_arrays)];
-    unsigned  i;
-
-    /* XXX are we pushing _all_ the necessary fields? */
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	stack[i].data = arrays->arrays[i].data;
-	stack[i].data_type = arrays->arrays[i].data_type;
-	stack[i].user_stride = arrays->arrays[i].user_stride;
-	stack[i].count = arrays->arrays[i].count;
-	stack[i].key = arrays->arrays[i].key;
-        stack[i].index = arrays->arrays[i].index;
-	stack[i].enabled = arrays->arrays[i].enabled;
-    }
-
-    arrays->active_texture_unit_stack[ arrays->stack_index ] =
+   struct array_state_vector *arrays = state->array_state;
+   struct array_stack_state *stack =
+      &arrays->stack[(arrays->stack_index * arrays->num_arrays)];
+   unsigned i;
+
+   /* XXX are we pushing _all_ the necessary fields? */
+   for (i = 0; i < arrays->num_arrays; i++) {
+      stack[i].data = arrays->arrays[i].data;
+      stack[i].data_type = arrays->arrays[i].data_type;
+      stack[i].user_stride = arrays->arrays[i].user_stride;
+      stack[i].count = arrays->arrays[i].count;
+      stack[i].key = arrays->arrays[i].key;
+      stack[i].index = arrays->arrays[i].index;
+      stack[i].enabled = arrays->arrays[i].enabled;
+   }
+
+   arrays->active_texture_unit_stack[arrays->stack_index] =
       arrays->active_texture_unit;
 
-    arrays->stack_index++;
+   arrays->stack_index++;
 }
 
 
 void
-__glXPopArrayState( __GLXattribute * state )
+__glXPopArrayState(__GLXattribute * state)
 {
-    struct array_state_vector * arrays = state->array_state;
-    struct array_stack_state * stack;
-    unsigned  i;
-
-
-    arrays->stack_index--;
-    stack = & arrays->stack[ (arrays->stack_index * arrays->num_arrays) ];
-
-    for ( i = 0 ; i < arrays->num_arrays ; i++ ) {
-	switch ( stack[i].key ) {
-	case GL_NORMAL_ARRAY:
-	    __indirect_glNormalPointer( stack[i].data_type,
-					stack[i].user_stride,
-					stack[i].data );
-	    break;
-	case GL_COLOR_ARRAY:
-	    __indirect_glColorPointer( stack[i].count,
-				       stack[i].data_type,
-				       stack[i].user_stride,
-				       stack[i].data );
-	    break;
-	case GL_INDEX_ARRAY:
-	    __indirect_glIndexPointer( stack[i].data_type,
-				       stack[i].user_stride,
-				       stack[i].data );
-	    break;
-	case GL_EDGE_FLAG_ARRAY:
-	    __indirect_glEdgeFlagPointer( stack[i].user_stride,
-					  stack[i].data );
-	    break;
-	case GL_TEXTURE_COORD_ARRAY:
-	    arrays->active_texture_unit = stack[i].index;
-	    __indirect_glTexCoordPointer( stack[i].count,
-					  stack[i].data_type,
-					  stack[i].user_stride,
-					  stack[i].data );
-	    break;
-	case GL_SECONDARY_COLOR_ARRAY:
-	    __indirect_glSecondaryColorPointerEXT( stack[i].count,
-						   stack[i].data_type,
-						   stack[i].user_stride,
-						   stack[i].data );
-	    break;
-	case GL_FOG_COORDINATE_ARRAY:
-	    __indirect_glFogCoordPointerEXT( stack[i].data_type,
-					     stack[i].user_stride,
-					     stack[i].data );
-	    break;
-
-	}
-	
-	__glXSetArrayEnable( state, stack[i].key, stack[i].index,
-			     stack[i].enabled );
-    }
-
-    arrays->active_texture_unit = 
-      arrays->active_texture_unit_stack[ arrays->stack_index ];
+   struct array_state_vector *arrays = state->array_state;
+   struct array_stack_state *stack;
+   unsigned i;
+
+
+   arrays->stack_index--;
+   stack = &arrays->stack[(arrays->stack_index * arrays->num_arrays)];
+
+   for (i = 0; i < arrays->num_arrays; i++) {
+      switch (stack[i].key) {
+      case GL_NORMAL_ARRAY:
+         __indirect_glNormalPointer(stack[i].data_type,
+                                    stack[i].user_stride, stack[i].data);
+         break;
+      case GL_COLOR_ARRAY:
+         __indirect_glColorPointer(stack[i].count,
+                                   stack[i].data_type,
+                                   stack[i].user_stride, stack[i].data);
+         break;
+      case GL_INDEX_ARRAY:
+         __indirect_glIndexPointer(stack[i].data_type,
+                                   stack[i].user_stride, stack[i].data);
+         break;
+      case GL_EDGE_FLAG_ARRAY:
+         __indirect_glEdgeFlagPointer(stack[i].user_stride, stack[i].data);
+         break;
+      case GL_TEXTURE_COORD_ARRAY:
+         arrays->active_texture_unit = stack[i].index;
+         __indirect_glTexCoordPointer(stack[i].count,
+                                      stack[i].data_type,
+                                      stack[i].user_stride, stack[i].data);
+         break;
+      case GL_SECONDARY_COLOR_ARRAY:
+         __indirect_glSecondaryColorPointerEXT(stack[i].count,
+                                               stack[i].data_type,
+                                               stack[i].user_stride,
+                                               stack[i].data);
+         break;
+      case GL_FOG_COORDINATE_ARRAY:
+         __indirect_glFogCoordPointerEXT(stack[i].data_type,
+                                         stack[i].user_stride, stack[i].data);
+         break;
+
+      }
+
+      __glXSetArrayEnable(state, stack[i].key, stack[i].index,
+                          stack[i].enabled);
+   }
+
+   arrays->active_texture_unit =
+      arrays->active_texture_unit_stack[arrays->stack_index];
 }
diff --git a/src/glx/x11/indirect_vertex_array.h b/src/glx/x11/indirect_vertex_array.h
index caab62b672e..37380320edd 100644
--- a/src/glx/x11/indirect_vertex_array.h
+++ b/src/glx/x11/indirect_vertex_array.h
@@ -31,27 +31,34 @@ extern const GLuint __glXTypeSize_table[16];
 #define __glXTypeSize(e) ((((e) & ~0x0f) != 0x1400) \
     ? 0 : __glXTypeSize_table[ (e) & 0x0f ])
 
-extern void __glXArrayDisableAll( __GLXattribute * state );
-
-extern GLboolean __glXSetArrayEnable( __GLXattribute * state,
-    GLenum key, unsigned index, GLboolean enable );
-
-extern GLboolean __glXGetArrayEnable( const __GLXattribute * const state,
-    GLenum key, unsigned index, GLintptr * dest );
-extern GLboolean __glXGetArraySize( const __GLXattribute * const state,
-    GLenum key, unsigned index, GLintptr * dest );
-extern GLboolean __glXGetArrayType( const __GLXattribute * const state,
-    GLenum key, unsigned index, GLintptr * dest );
-extern GLboolean __glXGetArrayStride( const __GLXattribute * const state,
-    GLenum key, unsigned index, GLintptr * dest );
-extern GLboolean __glXGetArrayPointer( const __GLXattribute * const state,
-    GLenum key, unsigned index, void ** dest );
-extern GLboolean __glXGetArrayNormalized( const __GLXattribute * const state,
-    GLenum key, unsigned index, GLintptr * dest );
-
-extern void __glXPushArrayState( __GLXattribute * state );
-extern void __glXPopArrayState( __GLXattribute * state );
-
-extern GLuint __glXGetActiveTextureUnit( const __GLXattribute * const state );
+extern void __glXArrayDisableAll(__GLXattribute * state);
+
+extern GLboolean __glXSetArrayEnable(__GLXattribute * state,
+                                     GLenum key, unsigned index,
+                                     GLboolean enable);
+
+extern GLboolean __glXGetArrayEnable(const __GLXattribute * const state,
+                                     GLenum key, unsigned index,
+                                     GLintptr * dest);
+extern GLboolean __glXGetArraySize(const __GLXattribute * const state,
+                                   GLenum key, unsigned index,
+                                   GLintptr * dest);
+extern GLboolean __glXGetArrayType(const __GLXattribute * const state,
+                                   GLenum key, unsigned index,
+                                   GLintptr * dest);
+extern GLboolean __glXGetArrayStride(const __GLXattribute * const state,
+                                     GLenum key, unsigned index,
+                                     GLintptr * dest);
+extern GLboolean __glXGetArrayPointer(const __GLXattribute * const state,
+                                      GLenum key, unsigned index,
+                                      void **dest);
+extern GLboolean __glXGetArrayNormalized(const __GLXattribute * const state,
+                                         GLenum key, unsigned index,
+                                         GLintptr * dest);
+
+extern void __glXPushArrayState(__GLXattribute * state);
+extern void __glXPopArrayState(__GLXattribute * state);
+
+extern GLuint __glXGetActiveTextureUnit(const __GLXattribute * const state);
 
 #endif /* INDIRECT_VERTEX_ARRAY_H */
diff --git a/src/glx/x11/indirect_va_private.h b/src/glx/x11/indirect_vertex_array_priv.h
index ab97dc645fb..56dac37c783 100644
--- a/src/glx/x11/indirect_va_private.h
+++ b/src/glx/x11/indirect_vertex_array_priv.h
@@ -42,86 +42,87 @@
 /**
  * State descriptor for a single array of vertex data.
  */
-struct array_state {
+struct array_state
+{
     /**
      * Pointer to the application supplied data.
      */
-    const void * data;
-    
+   const void *data;
+
     /**
      * Enum representing the type of the application supplied data.
      */
-    GLenum data_type;
+   GLenum data_type;
 
     /**
      * Stride value supplied by the application.  This value is not used
      * internally.  It is only kept so that it can be queried by the
      * application using glGet*v.
      */
-    GLsizei user_stride;
+   GLsizei user_stride;
 
     /**
      * Calculated size, in bytes, of a single element in the array.  This
      * is calculated based on \c count and the size of the data type
      * represented by \c data_type.
      */
-    GLsizei element_size;
+   GLsizei element_size;
 
     /**
      * Actual byte-stride from one element to the next.  This value will
      * be equal to either \c user_stride or \c element_stride.
      */
-    GLsizei true_stride;
+   GLsizei true_stride;
 
     /**
      * Number of data values in each element.
      */
-    GLint count;
+   GLint count;
 
     /**
      * "Normalized" data is on the range [0,1] (unsigned) or [-1,1] (signed).
      * This is used for mapping integral types to floating point types.
      */
-    GLboolean normalized;
+   GLboolean normalized;
 
     /**
      * Pre-calculated GLX protocol command header.
      */
-    uint32_t header[2];
-    
+   uint32_t header[2];
+
     /**
      * Size of the header data.  For simple data, like glColorPointerfv,
      * this is 4.  For complex data that requires either a count (e.g.,
      * glWeightfvARB), an index (e.g., glVertexAttrib1fvARB), or a
      * selector enum (e.g., glMultiTexCoord2fv) this is 8.
      */
-    unsigned header_size;
-    
+   unsigned header_size;
+
     /**
      * Set to \c GL_TRUE if this array is enabled.  Otherwise, it is set
      * to \c GL_FALSE.
      */
-    GLboolean  enabled;
+   GLboolean enabled;
 
     /**
      * For multi-arrayed data (e.g., texture coordinates, generic vertex
      * program attributes, etc.), this specifies which array this is.
      */
-    unsigned index;
-    
+   unsigned index;
+
     /**
      * Per-array-type key.  For most arrays, this will be the GL enum for
      * that array (e.g., GL_VERTEX_ARRAY for vertex data, GL_NORMAL_ARRAY
      * for normal data, GL_TEXTURE_COORD_ARRAY for texture coordinate data,
      * etc.).
      */
-    GLenum key;
+   GLenum key;
 
     /**
      * If this array can be used with the "classic" \c glDrawArrays protocol,
      * this is set to \c GL_TRUE.  Otherwise, it is set to \c GL_FALSE.
      */
-    GLboolean old_DrawArrays_possible;
+   GLboolean old_DrawArrays_possible;
 };
 
 
@@ -129,28 +130,29 @@ struct array_state {
  * Array state that is pushed / poped by \c glPushClientAttrib and
  * \c glPopClientAttrib.
  */
-struct array_stack_state {
+struct array_stack_state
+{
     /**
      * Pointer to the application supplied data.
      */
-    const void * data;
-    
+   const void *data;
+
     /**
      * Enum representing the type of the application supplied data.
      */
-    GLenum data_type;
+   GLenum data_type;
 
     /**
      * Stride value supplied by the application.  This value is not used
      * internally.  It is only kept so that it can be queried by the
      * application using glGet*v.
      */
-    GLsizei user_stride;
+   GLsizei user_stride;
 
     /**
      * Number of data values in each element.
      */
-    GLint count;
+   GLint count;
 
     /**
      * Per-array-type key.  For most arrays, this will be the GL enum for
@@ -158,30 +160,31 @@ struct array_stack_state {
      * for normal data, GL_TEXTURE_COORD_ARRAY for texture coordinate data,
      * etc.).
      */
-    GLenum key;
+   GLenum key;
 
     /**
      * For multi-arrayed data (e.g., texture coordinates, generic vertex
      * program attributes, etc.), this specifies which array this is.
      */
-    unsigned index;
+   unsigned index;
 
     /**
      * Set to \c GL_TRUE if this array is enabled.  Otherwise, it is set
      * to \c GL_FALSE.
      */
-    GLboolean  enabled;
+   GLboolean enabled;
 };
 
 
 /**
  * Collection of all the vertex array state.
  */
-struct array_state_vector {
+struct array_state_vector
+{
     /**
      * Number of arrays tracked by \c ::arrays.
      */
-    size_t num_arrays;
+   size_t num_arrays;
 
     /**
      * Array of vertex array state.  This array contains all of the valid
@@ -190,13 +193,13 @@ struct array_state_vector {
      * EXT_fog_coord, there won't be a GL_FOG_COORD_ARRAY entry in this
      * array.
      */
-    struct array_state * arrays;
+   struct array_state *arrays;
 
     /**
      * Number of currently enabled client-side arrays.  The value of this 
      * field is only valid if \c array_info_cache_valid is true.
      */
-    size_t enabled_client_array_count;
+   size_t enabled_client_array_count;
 
     /**
      * \name ARRAY_INFO cache.
@@ -214,12 +217,12 @@ struct array_state_vector {
      * \c array_info_cache_buffer_size.  \c array_info_cache_base stores a
      * pointer to the true start of the buffer (i.e., what malloc returned).
      */
-    /*@{*/
-    size_t array_info_cache_size;
-    size_t array_info_cache_buffer_size;
-    void * array_info_cache;
-    void * array_info_cache_base;
-    /*@}*/
+   /*@{ */
+   size_t array_info_cache_size;
+   size_t array_info_cache_buffer_size;
+   void *array_info_cache;
+   void *array_info_cache_base;
+   /*@} */
 
 
     /**
@@ -228,7 +231,7 @@ struct array_state_vector {
      * modifying the array settings for an enabled array and enabling /
      * disabling an array.
      */
-    GLboolean array_info_cache_valid;
+   GLboolean array_info_cache_valid;
 
     /**
      * Is it possible to use the GL 1.1 / EXT_vertex_arrays protocol?  Use
@@ -241,7 +244,7 @@ struct array_state_vector {
      * opcodes for \c glDrawArrays.  For servers that advertise one or the
      * other, there should be a way to select which opcode to use.
      */
-    GLboolean old_DrawArrays_possible;
+   GLboolean old_DrawArrays_possible;
 
     /**
      * Is it possible to use the new GL X.X / ARB_vertex_buffer_object
@@ -251,15 +254,15 @@ struct array_state_vector {
      * This protocol has not yet been defined by the ARB, but is currently a
      * work in progress.  This field is a place-holder.
      */
-    GLboolean new_DrawArrays_possible;
+   GLboolean new_DrawArrays_possible;
 
     /**
      * Active texture unit set by \c glClientActiveTexture.
      * 
      * \sa __glXGetActiveTextureUnit
      */
-    unsigned active_texture_unit;
-    
+   unsigned active_texture_unit;
+
     /**
      * Number of supported texture units.  Even if ARB_multitexture /
      * GL 1.3 are not supported, this will be at least 1.  When multitexture
@@ -271,7 +274,7 @@ struct array_state_vector {
      * instead (if GL 2.0 / ARB_fragment_shader / ARB_fragment_program /
      * NV_fragment_program are supported).
      */
-    unsigned num_texture_units;
+   unsigned num_texture_units;
 
     /**
      * Number of generic vertex program attribs.  If GL_ARB_vertex_program
@@ -279,7 +282,7 @@ struct array_state_vector {
      * queries by calling \c glGetProgramiv with \c GL_VERTEX_PROGRAM_ARB
      * and \c GL_MAX_PROGRAM_ATTRIBS_ARB.
      */
-    unsigned num_vertex_program_attribs;
+   unsigned num_vertex_program_attribs;
 
     /**
      * \n Methods for implementing various GL functions.
@@ -294,15 +297,15 @@ struct array_state_vector {
      * \todo
      * Write code to plug these functions directly into the dispatch table.
      */
-    /*@{*/
-    void (*DrawArrays)( GLenum, GLint, GLsizei );
-    void (*DrawElements)( GLenum mode, GLsizei count, GLenum type,
-			  const GLvoid *indices );
-    /*@}*/
-
-    struct array_stack_state * stack;
-    unsigned active_texture_unit_stack[ __GL_CLIENT_ATTRIB_STACK_DEPTH ];
-    unsigned stack_index;
+   /*@{ */
+   void (*DrawArrays) (GLenum, GLint, GLsizei);
+   void (*DrawElements) (GLenum mode, GLsizei count, GLenum type,
+                         const GLvoid * indices);
+   /*@} */
+
+   struct array_stack_state *stack;
+   unsigned active_texture_unit_stack[__GL_CLIENT_ATTRIB_STACK_DEPTH];
+   unsigned stack_index;
 };
 
 #endif /* _INDIRECT_VA_PRIVATE_ */
diff --git a/src/glx/x11/indirect_vertex_program.c b/src/glx/x11/indirect_vertex_program.c
index 5383dcc9810..865a4b1ba14 100644
--- a/src/glx/x11/indirect_vertex_program.c
+++ b/src/glx/x11/indirect_vertex_program.c
@@ -31,96 +31,104 @@
 #include <GL/glxproto.h>
 
 static void
-do_vertex_attrib_enable( GLuint index, GLboolean val )
+do_vertex_attrib_enable(GLuint index, GLboolean val)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
 
-    if ( ! __glXSetArrayEnable( state, GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB,
-				index, val ) ) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-    }
+   if (!__glXSetArrayEnable(state, GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB,
+                            index, val)) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+   }
 }
 
 
-void __indirect_glEnableVertexAttribArrayARB( GLuint index )
+void
+__indirect_glEnableVertexAttribArrayARB(GLuint index)
 {
-    do_vertex_attrib_enable( index, GL_TRUE );
+   do_vertex_attrib_enable(index, GL_TRUE);
 }
 
 
-void __indirect_glDisableVertexAttribArrayARB( GLuint index )
+void
+__indirect_glDisableVertexAttribArrayARB(GLuint index)
 {
-    do_vertex_attrib_enable( index, GL_FALSE );
+   do_vertex_attrib_enable(index, GL_FALSE);
 }
 
 
 static void
-get_parameter( unsigned opcode, unsigned size, GLenum target, GLuint index,
-	       void * params )
+get_parameter(unsigned opcode, unsigned size, GLenum target, GLuint index,
+              void *params)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    Display * const dpy = gc->currentDpy;
-    const GLuint cmdlen = 12;
-
-    if (__builtin_expect(dpy != NULL, 1)) {
-        GLubyte const * pc = __glXSetupVendorRequest(gc,
-						     X_GLXVendorPrivateWithReply,
-						     opcode, cmdlen);
-
-	*((GLenum *)(pc +  0)) = target;
-	*((GLuint *)(pc +  4)) = index;
-	*((GLuint *)(pc +  8)) = 0;
-
-        (void) __glXReadReply(dpy, size, params, GL_FALSE);
-        UnlockDisplay(dpy); SyncHandle();
-    }
-    return;
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   Display *const dpy = gc->currentDpy;
+   const GLuint cmdlen = 12;
+
+   if (__builtin_expect(dpy != NULL, 1)) {
+      GLubyte const *pc = __glXSetupVendorRequest(gc,
+                                                  X_GLXVendorPrivateWithReply,
+                                                  opcode, cmdlen);
+
+      *((GLenum *) (pc + 0)) = target;
+      *((GLuint *) (pc + 4)) = index;
+      *((GLuint *) (pc + 8)) = 0;
+
+      (void) __glXReadReply(dpy, size, params, GL_FALSE);
+      UnlockDisplay(dpy);
+      SyncHandle();
+   }
+   return;
 }
 
 
-void __indirect_glGetProgramEnvParameterfvARB( GLenum target, GLuint index,
-					       GLfloat * params )
+void
+__indirect_glGetProgramEnvParameterfvARB(GLenum target, GLuint index,
+                                         GLfloat * params)
 {
-    get_parameter( 1296, 4, target, index, params );
+   get_parameter(1296, 4, target, index, params);
 }
 
 
-void __indirect_glGetProgramEnvParameterdvARB( GLenum target, GLuint index,
-					       GLdouble * params )
+void
+__indirect_glGetProgramEnvParameterdvARB(GLenum target, GLuint index,
+                                         GLdouble * params)
 {
-    get_parameter( 1297, 8, target, index, params );
+   get_parameter(1297, 8, target, index, params);
 }
 
 
-void __indirect_glGetProgramLocalParameterfvARB( GLenum target, GLuint index,
-						 GLfloat * params )
+void
+__indirect_glGetProgramLocalParameterfvARB(GLenum target, GLuint index,
+                                           GLfloat * params)
 {
-    get_parameter( 1305, 4, target, index, params );
+   get_parameter(1305, 4, target, index, params);
 }
 
 
-void __indirect_glGetProgramLocalParameterdvARB( GLenum target, GLuint index,
-						 GLdouble * params )
+void
+__indirect_glGetProgramLocalParameterdvARB(GLenum target, GLuint index,
+                                           GLdouble * params)
 {
-    get_parameter( 1306, 8, target, index, params );
+   get_parameter(1306, 8, target, index, params);
 }
 
 
-void __indirect_glGetVertexAttribPointervNV( GLuint index, GLenum pname,
-					     GLvoid ** pointer )
+void
+__indirect_glGetVertexAttribPointervNV(GLuint index, GLenum pname,
+                                       GLvoid ** pointer)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
- 
-    if ( pname != GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB ) {
-	__glXSetError( gc, GL_INVALID_ENUM );
-    }
-    
-    if ( ! __glXGetArrayPointer( state, GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB,
-				 index, pointer ) ) {
-	__glXSetError( gc, GL_INVALID_VALUE );
-    }
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+
+   if (pname != GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+   }
+
+   if (!__glXGetArrayPointer(state, GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB,
+                             index, pointer)) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+   }
 }
 
 
@@ -131,143 +139,151 @@ void __indirect_glGetVertexAttribPointervNV( GLuint index, GLenum pname,
  * On success \c GL_TRUE is returned.  Otherwise, \c GL_FALSE is returned.
  */
 static GLboolean
-get_attrib_array_data( __GLXattribute * state, GLuint index, GLenum cap,
-		       GLintptr * data )
+get_attrib_array_data(__GLXattribute * state, GLuint index, GLenum cap,
+                      GLintptr * data)
 {
-    GLboolean retval = GL_FALSE;
-    const GLenum attrib = GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB;
+   GLboolean retval = GL_FALSE;
+   const GLenum attrib = GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB;
+
+   switch (cap) {
+   case GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB:
+      retval = __glXGetArrayEnable(state, attrib, index, data);
+      break;
 
-    switch( cap ) {
-    case GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB:
-	retval = __glXGetArrayEnable( state, attrib, index, data );
-	break;
+   case GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB:
+      retval = __glXGetArraySize(state, attrib, index, data);
+      break;
 
-    case GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB:
-	retval = __glXGetArraySize( state, attrib, index, data );
-	break;
+   case GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB:
+      retval = __glXGetArrayStride(state, attrib, index, data);
+      break;
 
-    case GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB:
-	retval = __glXGetArrayStride( state, attrib, index, data );
-	break;
+   case GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB:
+      retval = __glXGetArrayType(state, attrib, index, data);
+      break;
 
-    case GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB:
-	retval = __glXGetArrayType( state, attrib, index, data );
-	break;
+   case GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB:
+      retval = __glXGetArrayNormalized(state, attrib, index, data);
+      break;
+   }
 
-    case GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB:
-	retval = __glXGetArrayNormalized( state, attrib, index, data );
-	break;
-    }
 
-    
-    return retval;
+   return retval;
 }
 
 
-static void get_vertex_attrib( __GLXcontext * gc, unsigned vop,
-			       GLuint index, GLenum pname,
-			       xReply * reply )
+static void
+get_vertex_attrib(__GLXcontext * gc, unsigned vop,
+                  GLuint index, GLenum pname, xReply * reply)
 {
-    Display * const dpy = gc->currentDpy;
-    GLubyte * const pc = __glXSetupVendorRequest(gc,
-						 X_GLXVendorPrivateWithReply,
-						 vop, 8);
-						     
-    *((uint32_t *)(pc +  0)) = index;
-    *((uint32_t *)(pc +  4)) = pname;
-
-    (void) _XReply( dpy, reply, 0, False );
+   Display *const dpy = gc->currentDpy;
+   GLubyte *const pc = __glXSetupVendorRequest(gc,
+                                               X_GLXVendorPrivateWithReply,
+                                               vop, 8);
+
+   *((uint32_t *) (pc + 0)) = index;
+   *((uint32_t *) (pc + 4)) = pname;
+
+   (void) _XReply(dpy, reply, 0, False);
 }
 
 
-void __indirect_glGetVertexAttribivARB( GLuint index, GLenum pname,
-					GLint * params )
+void
+__indirect_glGetVertexAttribivARB(GLuint index, GLenum pname, GLint * params)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    Display * const dpy = gc->currentDpy;
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    xGLXSingleReply reply;
-
-
-    get_vertex_attrib( gc, 1303, index, pname, (xReply *) & reply );
-
-    if ( reply.size != 0 ) {
-	if ( ! get_attrib_array_data( state, index, pname, params ) ) {
-	    if (reply.size == 1) {
-		*params = (GLint) reply.pad3;
-	    } 
-	    else {
-		_XRead(dpy, (void *) params, 4 * reply.size);
-	    }
-	}
-    }
-
-    UnlockDisplay(dpy);
-    SyncHandle();
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   Display *const dpy = gc->currentDpy;
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   xGLXSingleReply reply;
+
+
+   get_vertex_attrib(gc, 1303, index, pname, (xReply *) & reply);
+
+   if (reply.size != 0) {
+      GLintptr data;
+
+
+      if (get_attrib_array_data(state, index, pname, &data)) {
+         *params = (GLint) data;
+      }
+      else {
+         if (reply.size == 1) {
+            *params = (GLint) reply.pad3;
+         }
+         else {
+            _XRead(dpy, (void *) params, 4 * reply.size);
+         }
+      }
+   }
+
+   UnlockDisplay(dpy);
+   SyncHandle();
 }
 
 
-void __indirect_glGetVertexAttribfvARB( GLuint index, GLenum pname,
-					GLfloat * params )
+void
+__indirect_glGetVertexAttribfvARB(GLuint index, GLenum pname,
+                                  GLfloat * params)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    Display * const dpy = gc->currentDpy;
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    xGLXSingleReply reply;
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   Display *const dpy = gc->currentDpy;
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   xGLXSingleReply reply;
 
 
-    get_vertex_attrib( gc, 1302, index, pname, (xReply *) & reply );
+   get_vertex_attrib(gc, 1302, index, pname, (xReply *) & reply);
 
-    if ( reply.size != 0 ) {
-	GLintptr data;
+   if (reply.size != 0) {
+      GLintptr data;
 
 
-	if ( get_attrib_array_data( state, index, pname, & data ) ) {
-	    *params = (GLfloat) data;
-	}
-	else {
-	    if (reply.size == 1) {
-		(void) memcpy( params, & reply.pad3, sizeof( GLfloat ) );
-	    } 
-	    else {
-		_XRead(dpy, (void *) params, 4 * reply.size);
-	    }
-	}
-    }
+      if (get_attrib_array_data(state, index, pname, &data)) {
+         *params = (GLfloat) data;
+      }
+      else {
+         if (reply.size == 1) {
+            (void) memcpy(params, &reply.pad3, sizeof(GLfloat));
+         }
+         else {
+            _XRead(dpy, (void *) params, 4 * reply.size);
+         }
+      }
+   }
 
-    UnlockDisplay(dpy);
-    SyncHandle();
+   UnlockDisplay(dpy);
+   SyncHandle();
 }
 
 
-void __indirect_glGetVertexAttribdvARB( GLuint index, GLenum pname,
-					GLdouble * params )
+void
+__indirect_glGetVertexAttribdvARB(GLuint index, GLenum pname,
+                                  GLdouble * params)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-    Display * const dpy = gc->currentDpy;
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    xGLXSingleReply reply;
+   __GLXcontext *const gc = __glXGetCurrentContext();
+   Display *const dpy = gc->currentDpy;
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   xGLXSingleReply reply;
 
 
-    get_vertex_attrib( gc, 1301, index, pname, (xReply *) & reply );
+   get_vertex_attrib(gc, 1301, index, pname, (xReply *) & reply);
 
-    if ( reply.size != 0 ) {
-	GLintptr data;
+   if (reply.size != 0) {
+      GLintptr data;
 
 
-	if ( get_attrib_array_data( state, index, pname, & data ) ) {
-	    *params = (GLdouble) data;
-	}
-	else {
-	    if (reply.size == 1) {
-		(void) memcpy( params, & reply.pad3, sizeof( GLdouble ) );
-	    } 
-	    else {
-		_XRead(dpy, (void *) params, 8 * reply.size);
-	    }
-	}
-    }
+      if (get_attrib_array_data(state, index, pname, &data)) {
+         *params = (GLdouble) data;
+      }
+      else {
+         if (reply.size == 1) {
+            (void) memcpy(params, &reply.pad3, sizeof(GLdouble));
+         }
+         else {
+            _XRead(dpy, (void *) params, 8 * reply.size);
+         }
+      }
+   }
 
-    UnlockDisplay(dpy);
-    SyncHandle();
+   UnlockDisplay(dpy);
+   SyncHandle();
 }
diff --git a/src/glx/x11/indirect_window_pos.c b/src/glx/x11/indirect_window_pos.c
index 533f8ef1a40..e97be3506d1 100644
--- a/src/glx/x11/indirect_window_pos.c
+++ b/src/glx/x11/indirect_window_pos.c
@@ -27,72 +27,86 @@
 #include <GL/gl.h>
 #include "indirect.h"
 
-void __indirect_glWindowPos2dMESA(GLdouble x, GLdouble y)
+void
+__indirect_glWindowPos2dMESA(GLdouble x, GLdouble y)
 {
-	__indirect_glWindowPos3fMESA(x, y, 0.0);
+   __indirect_glWindowPos3fMESA(x, y, 0.0);
 }
 
-void __indirect_glWindowPos2iMESA(GLint x, GLint y)
+void
+__indirect_glWindowPos2iMESA(GLint x, GLint y)
 {
-	__indirect_glWindowPos3fMESA(x, y, 0.0);
+   __indirect_glWindowPos3fMESA(x, y, 0.0);
 }
 
-void __indirect_glWindowPos2fMESA(GLfloat x, GLfloat y)
+void
+__indirect_glWindowPos2fMESA(GLfloat x, GLfloat y)
 {
-	__indirect_glWindowPos3fMESA(x, y, 0.0);
+   __indirect_glWindowPos3fMESA(x, y, 0.0);
 }
 
-void __indirect_glWindowPos2sMESA(GLshort x, GLshort y)
+void
+__indirect_glWindowPos2sMESA(GLshort x, GLshort y)
 {
-	__indirect_glWindowPos3fMESA(x, y, 0.0);
+   __indirect_glWindowPos3fMESA(x, y, 0.0);
 }
 
-void __indirect_glWindowPos2dvMESA(const GLdouble * p)
+void
+__indirect_glWindowPos2dvMESA(const GLdouble * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
+   __indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
 }
 
-void __indirect_glWindowPos2fvMESA(const GLfloat * p)
+void
+__indirect_glWindowPos2fvMESA(const GLfloat * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
+   __indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
 }
 
-void __indirect_glWindowPos2ivMESA(const GLint * p)
+void
+__indirect_glWindowPos2ivMESA(const GLint * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
+   __indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
 }
 
-void __indirect_glWindowPos2svMESA(const GLshort * p)
+void
+__indirect_glWindowPos2svMESA(const GLshort * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
+   __indirect_glWindowPos3fMESA(p[0], p[1], 0.0);
 }
 
-void __indirect_glWindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z)
+void
+__indirect_glWindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z)
 {
-	__indirect_glWindowPos3fMESA(x, y, z);
+   __indirect_glWindowPos3fMESA(x, y, z);
 }
 
-void __indirect_glWindowPos3iMESA(GLint x, GLint y, GLint z)
+void
+__indirect_glWindowPos3iMESA(GLint x, GLint y, GLint z)
 {
-	__indirect_glWindowPos3fMESA(x, y, z);
+   __indirect_glWindowPos3fMESA(x, y, z);
 }
 
-void __indirect_glWindowPos3sMESA(GLshort x, GLshort y, GLshort z)
+void
+__indirect_glWindowPos3sMESA(GLshort x, GLshort y, GLshort z)
 {
-	__indirect_glWindowPos3fMESA(x, y, z);
+   __indirect_glWindowPos3fMESA(x, y, z);
 }
 
-void __indirect_glWindowPos3dvMESA(const GLdouble * p)
+void
+__indirect_glWindowPos3dvMESA(const GLdouble * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
+   __indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
 }
 
-void __indirect_glWindowPos3ivMESA(const GLint * p)
+void
+__indirect_glWindowPos3ivMESA(const GLint * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
+   __indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
 }
 
-void __indirect_glWindowPos3svMESA(const GLshort * p)
+void
+__indirect_glWindowPos3svMESA(const GLshort * p)
 {
-	__indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
+   __indirect_glWindowPos3fMESA(p[0], p[1], p[2]);
 }
diff --git a/src/glx/x11/packrender.h b/src/glx/x11/packrender.h
index 8e3119d1b23..30f6d44bbd2 100644
--- a/src/glx/x11/packrender.h
+++ b/src/glx/x11/packrender.h
@@ -2,39 +2,34 @@
 #define __GLX_packrender_h__
 
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "glxclient.h"
 
@@ -53,23 +48,23 @@
 #define __GLX_PAD(a) (((a)+3) & ~3)
 
 /*
- ** Network size parameters
- */
+** Network size parameters
+*/
 #define sz_double 8
 
 /* Setup for all commands */
-#define __GLX_DECLARE_VARIABLES() \
-    __GLXcontext *gc;		  \
-    GLubyte *pc, *pixelHeaderPC;  \
-    GLuint compsize, cmdlen
-
-#define __GLX_LOAD_VARIABLES()	   \
-    gc = __glXGetCurrentContext(); \
-    pc = gc->pc;                   \
-    /* Muffle compilers */		    \
-    cmdlen = 0;         (void)cmdlen;	    \
-    compsize = 0;       (void)compsize;	    \
-    pixelHeaderPC = 0;  (void)pixelHeaderPC
+#define __GLX_DECLARE_VARIABLES()               \
+   __GLXcontext *gc;                            \
+   GLubyte *pc, *pixelHeaderPC;                 \
+   GLuint compsize, cmdlen
+
+#define __GLX_LOAD_VARIABLES()     \
+   gc = __glXGetCurrentContext();  \
+   pc = gc->pc;                    \
+   /* Muffle compilers */                  \
+   cmdlen = 0;         (void)cmdlen;          \
+   compsize = 0;       (void)compsize;        \
+   pixelHeaderPC = 0;  (void)pixelHeaderPC
 
 /*
 ** Variable sized command support macro.  This macro is used by calls
@@ -78,53 +73,53 @@
 ** If the buffer can't hold the command then it is flushed so that
 ** the command will fit in the next buffer.
 */
-#define __GLX_BEGIN_VARIABLE(opcode,size)	\
-    if (pc + (size) > gc->bufEnd) {		\
-	pc = __glXFlushRenderBuffer(gc, pc);	\
-    }						\
-    __GLX_PUT_SHORT(0,size);			\
-    __GLX_PUT_SHORT(2,opcode)
-
-#define __GLX_BEGIN_VARIABLE_LARGE(opcode,size)	\
-    pc = __glXFlushRenderBuffer(gc, pc);	\
-    __GLX_PUT_LONG(0,size);			\
-    __GLX_PUT_LONG(4,opcode)
-
-#define __GLX_BEGIN_VARIABLE_WITH_PIXEL(opcode,size) \
-    if (pc + (size) > gc->bufEnd) {		     \
-	pc = __glXFlushRenderBuffer(gc, pc);	     \
-    }						     \
-    __GLX_PUT_SHORT(0,size);			     \
-    __GLX_PUT_SHORT(2,opcode);			     \
-    pc += __GLX_RENDER_HDR_SIZE;		     \
-    pixelHeaderPC = pc;				     \
-    pc += __GLX_PIXEL_HDR_SIZE
-
-#define __GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL(opcode,size) \
-    pc = __glXFlushRenderBuffer(gc, pc);		   \
-    __GLX_PUT_LONG(0,size);				   \
-    __GLX_PUT_LONG(4,opcode);				   \
-    pc += __GLX_RENDER_LARGE_HDR_SIZE;			   \
-    pixelHeaderPC = pc;					   \
-    pc += __GLX_PIXEL_HDR_SIZE
-
-#define __GLX_BEGIN_VARIABLE_WITH_PIXEL_3D(opcode,size) \
-    if (pc + (size) > gc->bufEnd) {		        \
-	pc = __glXFlushRenderBuffer(gc, pc); 	        \
-    }						        \
-    __GLX_PUT_SHORT(0,size);			        \
-    __GLX_PUT_SHORT(2,opcode);			        \
-    pc += __GLX_RENDER_HDR_SIZE;		        \
-    pixelHeaderPC = pc;				        \
-    pc += __GLX_PIXEL_3D_HDR_SIZE
-
-#define __GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL_3D(opcode,size) \
-    pc = __glXFlushRenderBuffer(gc, pc);	   	      \
-    __GLX_PUT_LONG(0,size);				      \
-    __GLX_PUT_LONG(4,opcode);				      \
-    pc += __GLX_RENDER_LARGE_HDR_SIZE;			      \
-    pixelHeaderPC = pc;					      \
-    pc += __GLX_PIXEL_3D_HDR_SIZE
+#define __GLX_BEGIN_VARIABLE(opcode,size)       \
+   if (pc + (size) > gc->bufEnd) {              \
+      pc = __glXFlushRenderBuffer(gc, pc);      \
+   }                                            \
+   __GLX_PUT_SHORT(0,size);                     \
+   __GLX_PUT_SHORT(2,opcode)
+
+#define __GLX_BEGIN_VARIABLE_LARGE(opcode,size) \
+   pc = __glXFlushRenderBuffer(gc, pc);         \
+   __GLX_PUT_LONG(0,size);                      \
+   __GLX_PUT_LONG(4,opcode)
+
+#define __GLX_BEGIN_VARIABLE_WITH_PIXEL(opcode,size)  \
+   if (pc + (size) > gc->bufEnd) {                    \
+      pc = __glXFlushRenderBuffer(gc, pc);            \
+   }                                                  \
+   __GLX_PUT_SHORT(0,size);                           \
+   __GLX_PUT_SHORT(2,opcode);                         \
+   pc += __GLX_RENDER_HDR_SIZE;                       \
+   pixelHeaderPC = pc;                                \
+   pc += __GLX_PIXEL_HDR_SIZE
+
+#define __GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL(opcode,size)  \
+   pc = __glXFlushRenderBuffer(gc, pc);                     \
+   __GLX_PUT_LONG(0,size);                                  \
+   __GLX_PUT_LONG(4,opcode);                                \
+   pc += __GLX_RENDER_LARGE_HDR_SIZE;                       \
+   pixelHeaderPC = pc;                                      \
+   pc += __GLX_PIXEL_HDR_SIZE
+
+#define __GLX_BEGIN_VARIABLE_WITH_PIXEL_3D(opcode,size)  \
+   if (pc + (size) > gc->bufEnd) {                       \
+      pc = __glXFlushRenderBuffer(gc, pc);               \
+   }                                                     \
+   __GLX_PUT_SHORT(0,size);                              \
+   __GLX_PUT_SHORT(2,opcode);                            \
+   pc += __GLX_RENDER_HDR_SIZE;                          \
+   pixelHeaderPC = pc;                                   \
+   pc += __GLX_PIXEL_3D_HDR_SIZE
+
+#define __GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL_3D(opcode,size)  \
+   pc = __glXFlushRenderBuffer(gc, pc);                        \
+   __GLX_PUT_LONG(0,size);                                     \
+   __GLX_PUT_LONG(4,opcode);                                   \
+   pc += __GLX_RENDER_LARGE_HDR_SIZE;                          \
+   pixelHeaderPC = pc;                                         \
+   pc += __GLX_PIXEL_3D_HDR_SIZE
 
 /*
 ** Fixed size command support macro.  This macro is used by calls that
@@ -134,8 +129,8 @@
 ** before doing the storage work.
 */
 #define __GLX_BEGIN(opcode,size) \
-    __GLX_PUT_SHORT(0,size);     \
-    __GLX_PUT_SHORT(2,opcode)
+   __GLX_PUT_SHORT(0,size);      \
+   __GLX_PUT_SHORT(2,opcode)
 
 /*
 ** Finish a rendering command by advancing the pc.  If the pc is now past
@@ -145,52 +140,52 @@
 ** rendering buffer is flushed out into the X protocol stream (which may
 ** or may not do I/O).
 */
-#define __GLX_END(size)			       \
-    pc += size;				       \
-    if (pc > gc->limit) {		       \
-	(void) __glXFlushRenderBuffer(gc, pc); \
-    } else {				       \
-	gc->pc = pc;			       \
-    }
+#define __GLX_END(size)           \
+   pc += size;                       \
+   if (pc > gc->limit) {                  \
+      (void) __glXFlushRenderBuffer(gc, pc);    \
+   } else {                                     \
+      gc->pc = pc;                              \
+   }
 
 /* Array copy macros */
-#define __GLX_MEM_COPY(dest,src,bytes) \
-    if (src && dest) \
-    	memcpy(dest, src, bytes)
+#define __GLX_MEM_COPY(dest,src,bytes)          \
+   if (src && dest)                             \
+      memcpy(dest, src, bytes)
 
 /* Single item copy macros */
-#define __GLX_PUT_CHAR(offset,a) \
-    *((INT8 *) (pc + offset)) = a
+#define __GLX_PUT_CHAR(offset,a)                \
+   *((INT8 *) (pc + offset)) = a
 
 #ifndef _CRAY
-#define __GLX_PUT_SHORT(offset,a) \
-    *((INT16 *) (pc + offset)) = a
+#define __GLX_PUT_SHORT(offset,a)               \
+   *((INT16 *) (pc + offset)) = a
 
-#define __GLX_PUT_LONG(offset,a) \
-    *((INT32 *) (pc + offset)) = a
+#define __GLX_PUT_LONG(offset,a)                \
+   *((INT32 *) (pc + offset)) = a
 
-#define __GLX_PUT_FLOAT(offset,a) \
-    *((FLOAT32 *) (pc + offset)) = a
+#define __GLX_PUT_FLOAT(offset,a)               \
+   *((FLOAT32 *) (pc + offset)) = a
 
 #else
-#define __GLX_PUT_SHORT(offset,a) \
- { GLubyte *cp = (pc+offset); \
-   int shift = (64-16) - ((int)(cp) >> (64-6)); \
-   *(int *)cp = (*(int *)cp & ~(0xffff << shift)) | ((a & 0xffff) << shift); }
+#define __GLX_PUT_SHORT(offset,a)               \
+   { GLubyte *cp = (pc+offset);                 \
+      int shift = (64-16) - ((int)(cp) >> (64-6));                      \
+      *(int *)cp = (*(int *)cp & ~(0xffff << shift)) | ((a & 0xffff) << shift); }
 
-#define __GLX_PUT_LONG(offset,a) \
- { GLubyte *cp = (pc+offset); \
-   int shift = (64-32) - ((int)(cp) >> (64-6)); \
-   *(int *)cp = (*(int *)cp & ~(0xffffffff << shift)) | ((a & 0xffffffff) << shift); }
+#define __GLX_PUT_LONG(offset,a)                \
+   { GLubyte *cp = (pc+offset);                 \
+      int shift = (64-32) - ((int)(cp) >> (64-6));                      \
+      *(int *)cp = (*(int *)cp & ~(0xffffffff << shift)) | ((a & 0xffffffff) << shift); }
 
-#define __GLX_PUT_FLOAT(offset,a) \
-    gl_put_float((pc + offset),a)
+#define __GLX_PUT_FLOAT(offset,a)               \
+   gl_put_float((pc + offset),a)
 
-#define __GLX_PUT_DOUBLE(offset,a) \
-    gl_put_double(pc + offset, a)
+#define __GLX_PUT_DOUBLE(offset,a)              \
+   gl_put_double(pc + offset, a)
 
-extern void gl_put_float(/*GLubyte *, struct cray_single*/);
-extern void gl_put_double(/*GLubyte *, struct cray_double*/);
+extern void gl_put_float( /*GLubyte *, struct cray_single */ );
+extern void gl_put_double( /*GLubyte *, struct cray_double */ );
 #endif
 
 #ifndef _CRAY
@@ -200,48 +195,48 @@ extern void gl_put_double(/*GLubyte *, struct cray_double*/);
 ** This can certainly be done better for a particular machine
 ** architecture!
 */
-#define __GLX_PUT_DOUBLE(offset,a) \
-    __GLX_MEM_COPY(pc + offset, &a, 8)
+#define __GLX_PUT_DOUBLE(offset,a)              \
+   __GLX_MEM_COPY(pc + offset, &a, 8)
 #else
-#define __GLX_PUT_DOUBLE(offset,a) \
-    *((FLOAT64 *) (pc + offset)) = a
+#define __GLX_PUT_DOUBLE(offset,a)              \
+   *((FLOAT64 *) (pc + offset)) = a
 #endif
 
 #endif
 
-#define __GLX_PUT_CHAR_ARRAY(offset,a,alen) \
-    __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT8)
+#define __GLX_PUT_CHAR_ARRAY(offset,a,alen)                 \
+   __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT8)
 
 #ifndef _CRAY
-#define __GLX_PUT_SHORT_ARRAY(offset,a,alen) \
-    __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT16)
+#define __GLX_PUT_SHORT_ARRAY(offset,a,alen)                \
+   __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT16)
 
-#define __GLX_PUT_LONG_ARRAY(offset,a,alen) \
-    __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT32)
+#define __GLX_PUT_LONG_ARRAY(offset,a,alen)                 \
+   __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_INT32)
 
-#define __GLX_PUT_FLOAT_ARRAY(offset,a,alen) \
-    __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_FLOAT32)
+#define __GLX_PUT_FLOAT_ARRAY(offset,a,alen)                   \
+   __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_FLOAT32)
 
-#define __GLX_PUT_DOUBLE_ARRAY(offset,a,alen) \
-    __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_FLOAT64)
+#define __GLX_PUT_DOUBLE_ARRAY(offset,a,alen)                  \
+   __GLX_MEM_COPY(pc + offset, a, alen * __GLX_SIZE_FLOAT64)
 
 #else
-#define __GLX_PUT_SHORT_ARRAY(offset,a,alen) \
-    gl_put_short_array((GLubyte *)(pc + offset), a, alen * __GLX_SIZE_INT16)
+#define __GLX_PUT_SHORT_ARRAY(offset,a,alen)                            \
+   gl_put_short_array((GLubyte *)(pc + offset), a, alen * __GLX_SIZE_INT16)
 
-#define __GLX_PUT_LONG_ARRAY(offset,a,alen) \
-    gl_put_long_array((GLubyte *)(pc + offset), (long *)a, alen * __GLX_SIZE_INT32)
+#define __GLX_PUT_LONG_ARRAY(offset,a,alen)                             \
+   gl_put_long_array((GLubyte *)(pc + offset), (long *)a, alen * __GLX_SIZE_INT32)
 
-#define __GLX_PUT_FLOAT_ARRAY(offset,a,alen) \
-    gl_put_float_array((GLubyte *)(pc + offset), (float *)a, alen * __GLX_SIZE_FLOAT32)
+#define __GLX_PUT_FLOAT_ARRAY(offset,a,alen)                            \
+   gl_put_float_array((GLubyte *)(pc + offset), (float *)a, alen * __GLX_SIZE_FLOAT32)
 
-#define __GLX_PUT_DOUBLE_ARRAY(offset,a,alen) \
-    gl_put_double_array((GLubyte *)(pc + offset), (double *)a, alen * __GLX_SIZE_FLOAT64)
+#define __GLX_PUT_DOUBLE_ARRAY(offset,a,alen)                           \
+   gl_put_double_array((GLubyte *)(pc + offset), (double *)a, alen * __GLX_SIZE_FLOAT64)
 
-extern gl_put_short_array (GLubyte *, short *, int);
-extern gl_put_long_array (GLubyte *, long *, int);
-extern gl_put_float_array (GLubyte *, float *, int);
-extern gl_put_double_array (GLubyte *, double *, int);
+extern gl_put_short_array(GLubyte *, short *, int);
+extern gl_put_long_array(GLubyte *, long *, int);
+extern gl_put_float_array(GLubyte *, float *, int);
+extern gl_put_double_array(GLubyte *, double *, int);
 
 #endif /* _CRAY */
 
diff --git a/src/glx/x11/packsingle.h b/src/glx/x11/packsingle.h
index c69c543921b..f33a873f3a7 100644
--- a/src/glx/x11/packsingle.h
+++ b/src/glx/x11/packsingle.h
@@ -2,39 +2,34 @@
 #define __GLX_packsingle_h__
 
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "packrender.h"
 
@@ -53,107 +48,107 @@
 #define X_GLXSingle 0
 
 /* Declare common variables used during a single command */
-#define __GLX_SINGLE_DECLARE_VARIABLES()	 \
-    __GLXcontext *gc = __glXGetCurrentContext(); \
-    GLubyte *pc, *pixelHeaderPC;		 \
-    GLuint compsize, cmdlen;			 \
-    Display *dpy = gc->currentDpy;		 \
-    xGLXSingleReq *req
-
-#define __GLX_SINGLE_LOAD_VARIABLES() \
-    pc = gc->pc;           \
-    /* Muffle compilers */		     \
-    pixelHeaderPC = 0;  (void)pixelHeaderPC; \
-    compsize = 0;       (void)compsize;	     \
-    cmdlen = 0;         (void)cmdlen
+#define __GLX_SINGLE_DECLARE_VARIABLES()         \
+   __GLXcontext *gc = __glXGetCurrentContext();  \
+   GLubyte *pc, *pixelHeaderPC;                  \
+   GLuint compsize, cmdlen;                      \
+   Display *dpy = gc->currentDpy;                \
+   xGLXSingleReq *req
+
+#define __GLX_SINGLE_LOAD_VARIABLES()           \
+   pc = gc->pc;                                 \
+   /* Muffle compilers */                       \
+   pixelHeaderPC = 0;  (void)pixelHeaderPC;     \
+   compsize = 0;       (void)compsize;          \
+   cmdlen = 0;         (void)cmdlen
 
 /* Start a single command */
-#define __GLX_SINGLE_BEGIN(opcode,bytes)	   \
-    if (dpy) {					   \
-	(void) __glXFlushRenderBuffer(gc, pc);	   \
-	LockDisplay(dpy);			   \
-	GetReqExtra(GLXSingle,bytes,req);	   \
-	req->reqType = gc->majorOpcode;		   \
-	req->glxCode = opcode;			   \
-	req->contextTag = gc->currentContextTag;   \
-	pc = ((GLubyte *)(req) + sz_xGLXSingleReq)
+#define __GLX_SINGLE_BEGIN(opcode,bytes)        \
+   if (dpy) {                                   \
+   (void) __glXFlushRenderBuffer(gc, pc);       \
+   LockDisplay(dpy);                            \
+   GetReqExtra(GLXSingle,bytes,req);            \
+   req->reqType = gc->majorOpcode;              \
+   req->glxCode = opcode;                       \
+   req->contextTag = gc->currentContextTag;     \
+   pc = ((GLubyte *)(req) + sz_xGLXSingleReq)
 
 /* End a single command */
-#define __GLX_SINGLE_END()			   \
-	UnlockDisplay(dpy);			   \
-	SyncHandle();				   \
-    }
+#define __GLX_SINGLE_END()       \
+   UnlockDisplay(dpy);           \
+   SyncHandle();                 \
+   }
 
 /* Store data to sending for a single command */
-#define __GLX_SINGLE_PUT_CHAR(offset,a)	\
-    *((INT8 *) (pc + offset)) = a
+#define __GLX_SINGLE_PUT_CHAR(offset,a)         \
+   *((INT8 *) (pc + offset)) = a
 
 #ifndef CRAY
-#define __GLX_SINGLE_PUT_SHORT(offset,a) \
-    *((INT16 *) (pc + offset)) = a
+#define __GLX_SINGLE_PUT_SHORT(offset,a)        \
+   *((INT16 *) (pc + offset)) = a
 
-#define __GLX_SINGLE_PUT_LONG(offset,a)	\
-    *((INT32 *) (pc + offset)) = a
+#define __GLX_SINGLE_PUT_LONG(offset,a)         \
+   *((INT32 *) (pc + offset)) = a
 
-#define __GLX_SINGLE_PUT_FLOAT(offset,a) \
-    *((FLOAT32 *) (pc + offset)) = a
+#define __GLX_SINGLE_PUT_FLOAT(offset,a)        \
+   *((FLOAT32 *) (pc + offset)) = a
 
 #else
-#define __GLX_SINGLE_PUT_SHORT(offset,a) \
-    { GLubyte *cp = (pc+offset); \
-      int shift = (64-16) - ((int)(cp) >> (64-6)); \
+#define __GLX_SINGLE_PUT_SHORT(offset,a)        \
+   { GLubyte *cp = (pc+offset);                    \
+      int shift = (64-16) - ((int)(cp) >> (64-6));                      \
       *(int *)cp = (*(int *)cp & ~(0xffff << shift)) | ((a & 0xffff) << shift); }
 
-#define __GLX_SINGLE_PUT_LONG(offset,a) \
-    { GLubyte *cp = (pc+offset); \
-      int shift = (64-32) - ((int)(cp) >> (64-6)); \
+#define __GLX_SINGLE_PUT_LONG(offset,a)         \
+   { GLubyte *cp = (pc+offset);                    \
+      int shift = (64-32) - ((int)(cp) >> (64-6));                      \
       *(int *)cp = (*(int *)cp & ~(0xffffffff << shift)) | ((a & 0xffffffff) << shift); }
 
-#define __GLX_SINGLE_PUT_FLOAT(offset,a) \
-    gl_put_float(pc + offset, a)
+#define __GLX_SINGLE_PUT_FLOAT(offset,a)        \
+   gl_put_float(pc + offset, a)
 #endif
 
 /* Read support macros */
-#define __GLX_SINGLE_READ_XREPLY()		    \
-    (void) _XReply(dpy, (xReply*) &reply, 0, False)
+#define __GLX_SINGLE_READ_XREPLY()                    \
+   (void) _XReply(dpy, (xReply*) &reply, 0, False)
 
-#define __GLX_SINGLE_GET_RETVAL(a,cast)	\
-    a = (cast) reply.retval
+#define __GLX_SINGLE_GET_RETVAL(a,cast)         \
+   a = (cast) reply.retval
 
-#define __GLX_SINGLE_GET_SIZE(a) \
-    a = (GLint) reply.size
+#define __GLX_SINGLE_GET_SIZE(a)                \
+   a = (GLint) reply.size
 
 #ifndef _CRAY
-#define __GLX_SINGLE_GET_CHAR(p) \
-    *p = *(GLbyte *)&reply.pad3;
+#define __GLX_SINGLE_GET_CHAR(p)                \
+   *p = *(GLbyte *)&reply.pad3;
 
-#define __GLX_SINGLE_GET_SHORT(p) \
-    *p = *(GLshort *)&reply.pad3;
+#define __GLX_SINGLE_GET_SHORT(p)               \
+   *p = *(GLshort *)&reply.pad3;
 
-#define __GLX_SINGLE_GET_LONG(p) \
-    *p = *(GLint *)&reply.pad3;
+#define __GLX_SINGLE_GET_LONG(p)                \
+   *p = *(GLint *)&reply.pad3;
 
-#define __GLX_SINGLE_GET_FLOAT(p) \
-    *p = *(GLfloat *)&reply.pad3;
+#define __GLX_SINGLE_GET_FLOAT(p)               \
+   *p = *(GLfloat *)&reply.pad3;
 
 #else
-#define __GLX_SINGLE_GET_CHAR(p) \
-    *p = reply.pad3 >> 24;
+#define __GLX_SINGLE_GET_CHAR(p)                \
+   *p = reply.pad3 >> 24;
 
-#define __GLX_SINGLE_GET_SHORT(p) \
-    {int t = reply.pad3 >> 16; \
-     *p = (t & 0x8000) ? (t | ~0xffff) : (t & 0xffff);}
+#define __GLX_SINGLE_GET_SHORT(p)               \
+   {int t = reply.pad3 >> 16;                            \
+      *p = (t & 0x8000) ? (t | ~0xffff) : (t & 0xffff);}
 
-#define __GLX_SINGLE_GET_LONG(p) \
-    {int t = reply.pad3; \
-     *p = (t & 0x80000000) ? (t | ~0xffffffff) : (t & 0xffffffff);}
+#define __GLX_SINGLE_GET_LONG(p)                \
+   {int t = reply.pad3;                                              \
+      *p = (t & 0x80000000) ? (t | ~0xffffffff) : (t & 0xffffffff);}
 
 #define PAD3OFFSET 16
-#define __GLX_SINGLE_GET_FLOAT(p) \
-    *p = gl_ntoh_float((GLubyte *)&reply + PAD3OFFSET);
+#define __GLX_SINGLE_GET_FLOAT(p)                        \
+   *p = gl_ntoh_float((GLubyte *)&reply + PAD3OFFSET);
 
-#define __GLX_SINGLE_GET_DOUBLE(p) \
-    *p = gl_ntoh_double((GLubyte *)&reply + PAD3OFFSET);
+#define __GLX_SINGLE_GET_DOUBLE(p)                       \
+   *p = gl_ntoh_double((GLubyte *)&reply + PAD3OFFSET);
 
 extern float gl_ntoh_float(GLubyte *);
 extern float gl_ntoh_double(GLubyte *);
@@ -162,57 +157,57 @@ extern float gl_ntoh_double(GLubyte *);
 #ifndef _CRAY
 
 #ifdef __GLX_ALIGN64
-#define __GLX_SINGLE_GET_DOUBLE(p) \
-    __GLX_MEM_COPY(p, &reply.pad3, 8)
+#define __GLX_SINGLE_GET_DOUBLE(p)              \
+   __GLX_MEM_COPY(p, &reply.pad3, 8)
 #else
-#define __GLX_SINGLE_GET_DOUBLE(p) \
-    *p = *(GLdouble *)&reply.pad3
+#define __GLX_SINGLE_GET_DOUBLE(p)              \
+   *p = *(GLdouble *)&reply.pad3
 #endif
 
 #endif
-	  
+
 /* Get an array of typed data */
-#define __GLX_SINGLE_GET_VOID_ARRAY(a,alen)	\
-{						\
-    GLint slop = alen*__GLX_SIZE_INT8 & 3;	\
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT8);	\
-    if (slop) _XEatData(dpy,4-slop);		\
-}
-
-#define __GLX_SINGLE_GET_CHAR_ARRAY(a,alen)	\
-{						\
-    GLint slop = alen*__GLX_SIZE_INT8 & 3;	\
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT8);	\
-    if (slop) _XEatData(dpy,4-slop);		\
-}
-					
-
-#define __GLX_SINGLE_GET_SHORT_ARRAY(a,alen)	\
-{						\
-    GLint slop = (alen*__GLX_SIZE_INT16) & 3;	\
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT16);\
-    if (slop) _XEatData(dpy,4-slop);		\
-}
-
-#define __GLX_SINGLE_GET_LONG_ARRAY(a,alen) \
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT32);  	
+#define __GLX_SINGLE_GET_VOID_ARRAY(a,alen)     \
+   {                                            \
+      GLint slop = alen*__GLX_SIZE_INT8 & 3;    \
+      _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT8);  \
+      if (slop) _XEatData(dpy,4-slop);             \
+   }
+
+#define __GLX_SINGLE_GET_CHAR_ARRAY(a,alen)     \
+   {                                            \
+      GLint slop = alen*__GLX_SIZE_INT8 & 3;    \
+      _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT8);  \
+      if (slop) _XEatData(dpy,4-slop);             \
+   }
+
+
+#define __GLX_SINGLE_GET_SHORT_ARRAY(a,alen)    \
+   {                                            \
+      GLint slop = (alen*__GLX_SIZE_INT16) & 3;    \
+      _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT16); \
+      if (slop) _XEatData(dpy,4-slop);             \
+   }
+
+#define __GLX_SINGLE_GET_LONG_ARRAY(a,alen)        \
+   _XRead(dpy,(char *)a,alen*__GLX_SIZE_INT32);
 
 #ifndef _CRAY
-#define __GLX_SINGLE_GET_FLOAT_ARRAY(a,alen) \
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_FLOAT32);  	
+#define __GLX_SINGLE_GET_FLOAT_ARRAY(a,alen)       \
+   _XRead(dpy,(char *)a,alen*__GLX_SIZE_FLOAT32);
 
-#define __GLX_SINGLE_GET_DOUBLE_ARRAY(a,alen) \
-    _XRead(dpy,(char *)a,alen*__GLX_SIZE_FLOAT64);  	
+#define __GLX_SINGLE_GET_DOUBLE_ARRAY(a,alen)      \
+   _XRead(dpy,(char *)a,alen*__GLX_SIZE_FLOAT64);
 
 #else
-#define __GLX_SINGLE_GET_FLOAT_ARRAY(a,alen) \
-    gl_get_float_array(dpy,a,alen);
+#define __GLX_SINGLE_GET_FLOAT_ARRAY(a,alen)    \
+   gl_get_float_array(dpy,a,alen);
 
-#define __GLX_SINGLE_GET_DOUBLE_ARRAY(a,alen) \
-    gl_get_double_array(dpy, a, alen);
+#define __GLX_SINGLE_GET_DOUBLE_ARRAY(a,alen)   \
+   gl_get_double_array(dpy, a, alen);
 
-extern void gl_get_float_array(Display *dpy, float *a, int alen);
-extern void gl_get_double_array(Display *dpy, double *a, int alen);
+extern void gl_get_float_array(Display * dpy, float *a, int alen);
+extern void gl_get_double_array(Display * dpy, double *a, int alen);
 #endif
 
 #endif /* !__GLX_packsingle_h__ */
diff --git a/src/glx/x11/pixel.c b/src/glx/x11/pixel.c
index 279555bdfdf..d36ca31e7dd 100644
--- a/src/glx/x11/pixel.c
+++ b/src/glx/x11/pixel.c
@@ -1,81 +1,76 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "packrender.h"
 
 static const GLubyte MsbToLsbTable[256] = {
-    0x00, 0x80, 0x40, 0xc0, 0x20, 0xa0, 0x60, 0xe0,
-    0x10, 0x90, 0x50, 0xd0, 0x30, 0xb0, 0x70, 0xf0,
-    0x08, 0x88, 0x48, 0xc8, 0x28, 0xa8, 0x68, 0xe8,
-    0x18, 0x98, 0x58, 0xd8, 0x38, 0xb8, 0x78, 0xf8,
-    0x04, 0x84, 0x44, 0xc4, 0x24, 0xa4, 0x64, 0xe4,
-    0x14, 0x94, 0x54, 0xd4, 0x34, 0xb4, 0x74, 0xf4,
-    0x0c, 0x8c, 0x4c, 0xcc, 0x2c, 0xac, 0x6c, 0xec,
-    0x1c, 0x9c, 0x5c, 0xdc, 0x3c, 0xbc, 0x7c, 0xfc,
-    0x02, 0x82, 0x42, 0xc2, 0x22, 0xa2, 0x62, 0xe2,
-    0x12, 0x92, 0x52, 0xd2, 0x32, 0xb2, 0x72, 0xf2,
-    0x0a, 0x8a, 0x4a, 0xca, 0x2a, 0xaa, 0x6a, 0xea,
-    0x1a, 0x9a, 0x5a, 0xda, 0x3a, 0xba, 0x7a, 0xfa,
-    0x06, 0x86, 0x46, 0xc6, 0x26, 0xa6, 0x66, 0xe6,
-    0x16, 0x96, 0x56, 0xd6, 0x36, 0xb6, 0x76, 0xf6,
-    0x0e, 0x8e, 0x4e, 0xce, 0x2e, 0xae, 0x6e, 0xee,
-    0x1e, 0x9e, 0x5e, 0xde, 0x3e, 0xbe, 0x7e, 0xfe,
-    0x01, 0x81, 0x41, 0xc1, 0x21, 0xa1, 0x61, 0xe1,
-    0x11, 0x91, 0x51, 0xd1, 0x31, 0xb1, 0x71, 0xf1,
-    0x09, 0x89, 0x49, 0xc9, 0x29, 0xa9, 0x69, 0xe9,
-    0x19, 0x99, 0x59, 0xd9, 0x39, 0xb9, 0x79, 0xf9,
-    0x05, 0x85, 0x45, 0xc5, 0x25, 0xa5, 0x65, 0xe5,
-    0x15, 0x95, 0x55, 0xd5, 0x35, 0xb5, 0x75, 0xf5,
-    0x0d, 0x8d, 0x4d, 0xcd, 0x2d, 0xad, 0x6d, 0xed,
-    0x1d, 0x9d, 0x5d, 0xdd, 0x3d, 0xbd, 0x7d, 0xfd,
-    0x03, 0x83, 0x43, 0xc3, 0x23, 0xa3, 0x63, 0xe3,
-    0x13, 0x93, 0x53, 0xd3, 0x33, 0xb3, 0x73, 0xf3,
-    0x0b, 0x8b, 0x4b, 0xcb, 0x2b, 0xab, 0x6b, 0xeb,
-    0x1b, 0x9b, 0x5b, 0xdb, 0x3b, 0xbb, 0x7b, 0xfb,
-    0x07, 0x87, 0x47, 0xc7, 0x27, 0xa7, 0x67, 0xe7,
-    0x17, 0x97, 0x57, 0xd7, 0x37, 0xb7, 0x77, 0xf7,
-    0x0f, 0x8f, 0x4f, 0xcf, 0x2f, 0xaf, 0x6f, 0xef,
-    0x1f, 0x9f, 0x5f, 0xdf, 0x3f, 0xbf, 0x7f, 0xff,
+   0x00, 0x80, 0x40, 0xc0, 0x20, 0xa0, 0x60, 0xe0,
+   0x10, 0x90, 0x50, 0xd0, 0x30, 0xb0, 0x70, 0xf0,
+   0x08, 0x88, 0x48, 0xc8, 0x28, 0xa8, 0x68, 0xe8,
+   0x18, 0x98, 0x58, 0xd8, 0x38, 0xb8, 0x78, 0xf8,
+   0x04, 0x84, 0x44, 0xc4, 0x24, 0xa4, 0x64, 0xe4,
+   0x14, 0x94, 0x54, 0xd4, 0x34, 0xb4, 0x74, 0xf4,
+   0x0c, 0x8c, 0x4c, 0xcc, 0x2c, 0xac, 0x6c, 0xec,
+   0x1c, 0x9c, 0x5c, 0xdc, 0x3c, 0xbc, 0x7c, 0xfc,
+   0x02, 0x82, 0x42, 0xc2, 0x22, 0xa2, 0x62, 0xe2,
+   0x12, 0x92, 0x52, 0xd2, 0x32, 0xb2, 0x72, 0xf2,
+   0x0a, 0x8a, 0x4a, 0xca, 0x2a, 0xaa, 0x6a, 0xea,
+   0x1a, 0x9a, 0x5a, 0xda, 0x3a, 0xba, 0x7a, 0xfa,
+   0x06, 0x86, 0x46, 0xc6, 0x26, 0xa6, 0x66, 0xe6,
+   0x16, 0x96, 0x56, 0xd6, 0x36, 0xb6, 0x76, 0xf6,
+   0x0e, 0x8e, 0x4e, 0xce, 0x2e, 0xae, 0x6e, 0xee,
+   0x1e, 0x9e, 0x5e, 0xde, 0x3e, 0xbe, 0x7e, 0xfe,
+   0x01, 0x81, 0x41, 0xc1, 0x21, 0xa1, 0x61, 0xe1,
+   0x11, 0x91, 0x51, 0xd1, 0x31, 0xb1, 0x71, 0xf1,
+   0x09, 0x89, 0x49, 0xc9, 0x29, 0xa9, 0x69, 0xe9,
+   0x19, 0x99, 0x59, 0xd9, 0x39, 0xb9, 0x79, 0xf9,
+   0x05, 0x85, 0x45, 0xc5, 0x25, 0xa5, 0x65, 0xe5,
+   0x15, 0x95, 0x55, 0xd5, 0x35, 0xb5, 0x75, 0xf5,
+   0x0d, 0x8d, 0x4d, 0xcd, 0x2d, 0xad, 0x6d, 0xed,
+   0x1d, 0x9d, 0x5d, 0xdd, 0x3d, 0xbd, 0x7d, 0xfd,
+   0x03, 0x83, 0x43, 0xc3, 0x23, 0xa3, 0x63, 0xe3,
+   0x13, 0x93, 0x53, 0xd3, 0x33, 0xb3, 0x73, 0xf3,
+   0x0b, 0x8b, 0x4b, 0xcb, 0x2b, 0xab, 0x6b, 0xeb,
+   0x1b, 0x9b, 0x5b, 0xdb, 0x3b, 0xbb, 0x7b, 0xfb,
+   0x07, 0x87, 0x47, 0xc7, 0x27, 0xa7, 0x67, 0xe7,
+   0x17, 0x97, 0x57, 0xd7, 0x37, 0xb7, 0x77, 0xf7,
+   0x0f, 0x8f, 0x4f, 0xcf, 0x2f, 0xaf, 0x6f, 0xef,
+   0x1f, 0x9f, 0x5f, 0xdf, 0x3f, 0xbf, 0x7f, 0xff,
 };
 
 static const GLubyte LowBitsMask[9] = {
-    0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff,
+   0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff,
 };
 
 static const GLubyte HighBitsMask[9] = {
-    0x00, 0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff,
+   0x00, 0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff,
 };
 
 
@@ -84,76 +79,80 @@ static const GLubyte HighBitsMask[9] = {
 ** data is transfered into the destImage buffer.  Return in modes the
 ** set of pixel modes that are to be done by the server.
 */
-static void FillBitmap(__GLXcontext *gc, GLint width, GLint height,
-		       GLenum format, const GLvoid *userdata,
-		       GLubyte *destImage)
+static void
+FillBitmap(__GLXcontext * gc, GLint width, GLint height,
+           GLenum format, const GLvoid * userdata, GLubyte * destImage)
 {
-    const __GLXattribute * state = gc->client_state_private;
-    GLint rowLength = state->storeUnpack.rowLength;
-    GLint alignment = state->storeUnpack.alignment;
-    GLint skipPixels = state->storeUnpack.skipPixels;
-    GLint skipRows = state->storeUnpack.skipRows;
-    GLint lsbFirst = state->storeUnpack.lsbFirst;
-    GLint elementsLeft, bitOffset, currentByte, nextByte, highBitMask;
-    GLint lowBitMask, i;
-    GLint components, groupsPerRow, rowSize, padding, elementsPerRow;
-    const GLubyte *start, *iter;
+   const __GLXattribute *state = gc->client_state_private;
+   GLint rowLength = state->storeUnpack.rowLength;
+   GLint alignment = state->storeUnpack.alignment;
+   GLint skipPixels = state->storeUnpack.skipPixels;
+   GLint skipRows = state->storeUnpack.skipRows;
+   GLint lsbFirst = state->storeUnpack.lsbFirst;
+   GLint elementsLeft, bitOffset, currentByte, nextByte, highBitMask;
+   GLint lowBitMask, i;
+   GLint components, groupsPerRow, rowSize, padding, elementsPerRow;
+   const GLubyte *start, *iter;
 
-    if (rowLength > 0) {
-	groupsPerRow = rowLength;
-    } else {
-	groupsPerRow = width;
-    }
-    components = __glElementsPerGroup(format,GL_BITMAP);
-    rowSize = (groupsPerRow * components + 7) >> 3;
-    padding = (rowSize % alignment);
-    if (padding) {
-	rowSize += alignment - padding;
-    }
-    start = ((const GLubyte*) userdata) + skipRows * rowSize + 
-	((skipPixels * components) >> 3);
-    bitOffset = (skipPixels * components) & 7;
-    highBitMask = LowBitsMask[8-bitOffset];
-    lowBitMask = HighBitsMask[bitOffset];
-    elementsPerRow = width * components;
-    for (i = 0; i < height; i++) {
-	elementsLeft = elementsPerRow;
-	iter = start;
-	while (elementsLeft) {
-	    /* First retrieve low bits from current byte */
-	    if (lsbFirst) {
-		currentByte = MsbToLsbTable[iter[0]];
-	    } else {
-		currentByte = iter[0];
-	    }
-	    if (bitOffset) {
-		/* Need to read next byte to finish current byte */
-		if (elementsLeft > (8 - bitOffset)) {
-		    if (lsbFirst) {
-			nextByte = MsbToLsbTable[iter[1]];
-		    } else {
-			nextByte = iter[1];
-		    }
-		    currentByte = 
-			((currentByte & highBitMask) << bitOffset) |
-			    ((nextByte & lowBitMask) >> (8 - bitOffset));
-		} else {
-		    currentByte = 
-			((currentByte & highBitMask) << bitOffset);
-		}
-	    } 
-	    if (elementsLeft >= 8) {
-		*destImage = currentByte;
-		elementsLeft -= 8;
-	    } else {
-		*destImage = currentByte & HighBitsMask[elementsLeft];
-		elementsLeft = 0;
-	    }
-	    destImage++;
-	    iter++;
-	}
-	start += rowSize;
-    }
+   if (rowLength > 0) {
+      groupsPerRow = rowLength;
+   }
+   else {
+      groupsPerRow = width;
+   }
+   components = __glElementsPerGroup(format, GL_BITMAP);
+   rowSize = (groupsPerRow * components + 7) >> 3;
+   padding = (rowSize % alignment);
+   if (padding) {
+      rowSize += alignment - padding;
+   }
+   start = ((const GLubyte *) userdata) + skipRows * rowSize +
+      ((skipPixels * components) >> 3);
+   bitOffset = (skipPixels * components) & 7;
+   highBitMask = LowBitsMask[8 - bitOffset];
+   lowBitMask = HighBitsMask[bitOffset];
+   elementsPerRow = width * components;
+   for (i = 0; i < height; i++) {
+      elementsLeft = elementsPerRow;
+      iter = start;
+      while (elementsLeft) {
+         /* First retrieve low bits from current byte */
+         if (lsbFirst) {
+            currentByte = MsbToLsbTable[iter[0]];
+         }
+         else {
+            currentByte = iter[0];
+         }
+         if (bitOffset) {
+            /* Need to read next byte to finish current byte */
+            if (elementsLeft > (8 - bitOffset)) {
+               if (lsbFirst) {
+                  nextByte = MsbToLsbTable[iter[1]];
+               }
+               else {
+                  nextByte = iter[1];
+               }
+               currentByte =
+                  ((currentByte & highBitMask) << bitOffset) |
+                  ((nextByte & lowBitMask) >> (8 - bitOffset));
+            }
+            else {
+               currentByte = ((currentByte & highBitMask) << bitOffset);
+            }
+         }
+         if (elementsLeft >= 8) {
+            *destImage = currentByte;
+            elementsLeft -= 8;
+         }
+         else {
+            *destImage = currentByte & HighBitsMask[elementsLeft];
+            elementsLeft = 0;
+         }
+         destImage++;
+         iter++;
+      }
+      start += rowSize;
+   }
 }
 
 /*
@@ -161,209 +160,224 @@ static void FillBitmap(__GLXcontext *gc, GLint width, GLint height,
 ** The internal packed array format used has LSB_FIRST = FALSE and 
 ** ALIGNMENT = 1.
 */
-void __glFillImage(__GLXcontext *gc, GLint dim, GLint width, GLint height,
-		   GLint depth, GLenum format, GLenum type,
-		   const GLvoid *userdata, GLubyte *newimage, GLubyte *modes)
+void
+__glFillImage(__GLXcontext * gc, GLint dim, GLint width, GLint height,
+              GLint depth, GLenum format, GLenum type,
+              const GLvoid * userdata, GLubyte * newimage, GLubyte * modes)
 {
-    const __GLXattribute * state = gc->client_state_private;
-    GLint rowLength = state->storeUnpack.rowLength;
-    GLint imageHeight = state->storeUnpack.imageHeight;
-    GLint alignment = state->storeUnpack.alignment;
-    GLint skipPixels = state->storeUnpack.skipPixels;
-    GLint skipRows = state->storeUnpack.skipRows;
-    GLint skipImages = state->storeUnpack.skipImages;
-    GLint swapBytes = state->storeUnpack.swapEndian;
-    GLint components, elementSize, rowSize, padding, groupsPerRow, groupSize;
-    GLint elementsPerRow, imageSize, rowsPerImage, h, i, j, k;
-    const GLubyte *start, *iter, *itera, *iterb, *iterc;
-    GLubyte *iter2;
+   const __GLXattribute *state = gc->client_state_private;
+   GLint rowLength = state->storeUnpack.rowLength;
+   GLint imageHeight = state->storeUnpack.imageHeight;
+   GLint alignment = state->storeUnpack.alignment;
+   GLint skipPixels = state->storeUnpack.skipPixels;
+   GLint skipRows = state->storeUnpack.skipRows;
+   GLint skipImages = state->storeUnpack.skipImages;
+   GLint swapBytes = state->storeUnpack.swapEndian;
+   GLint components, elementSize, rowSize, padding, groupsPerRow, groupSize;
+   GLint elementsPerRow, imageSize, rowsPerImage, h, i, j, k;
+   const GLubyte *start, *iter, *itera, *iterb, *iterc;
+   GLubyte *iter2;
 
-    if (type == GL_BITMAP) {
-	FillBitmap(gc, width, height, format, userdata, newimage);
-    } else {
-	components = __glElementsPerGroup(format,type);
-	if (rowLength > 0) {
-	    groupsPerRow = rowLength;
-	} else {
-	    groupsPerRow = width;
-	}
-	if (imageHeight > 0) {
-	    rowsPerImage = imageHeight;
-	} else {
-	    rowsPerImage = height;
-	}
+   if (type == GL_BITMAP) {
+      FillBitmap(gc, width, height, format, userdata, newimage);
+   }
+   else {
+      components = __glElementsPerGroup(format, type);
+      if (rowLength > 0) {
+         groupsPerRow = rowLength;
+      }
+      else {
+         groupsPerRow = width;
+      }
+      if (imageHeight > 0) {
+         rowsPerImage = imageHeight;
+      }
+      else {
+         rowsPerImage = height;
+      }
 
-	elementSize = __glBytesPerElement(type);
-	groupSize = elementSize * components;
-	if (elementSize == 1) swapBytes = 0;
+      elementSize = __glBytesPerElement(type);
+      groupSize = elementSize * components;
+      if (elementSize == 1)
+         swapBytes = 0;
 
-	rowSize = groupsPerRow * groupSize;
-	padding = (rowSize % alignment);
-	if (padding) {
-	    rowSize += alignment - padding;
-	}
-	imageSize = rowSize * rowsPerImage;
-	start = ((const GLubyte*) userdata) + skipImages * imageSize +
-		skipRows * rowSize + skipPixels * groupSize;
-	iter2 = newimage;
-	elementsPerRow = width * components;
+      rowSize = groupsPerRow * groupSize;
+      padding = (rowSize % alignment);
+      if (padding) {
+         rowSize += alignment - padding;
+      }
+      imageSize = rowSize * rowsPerImage;
+      start = ((const GLubyte *) userdata) + skipImages * imageSize +
+         skipRows * rowSize + skipPixels * groupSize;
+      iter2 = newimage;
+      elementsPerRow = width * components;
 
-	if (swapBytes) {
-	    itera = start;
-	    for (h = 0; h < depth; h++) {
-		iterb = itera;
-		for (i = 0; i < height; i++) {
-		    iterc = iterb;
-		    for (j = 0; j < elementsPerRow; j++) {
-			for (k = 1; k <= elementSize; k++) {
-			    iter2[k-1] = iterc[elementSize - k];
-			}
-			iter2 += elementSize;
-			iterc += elementSize;
-		    }
-		    iterb += rowSize;
-		}
-		itera += imageSize;
-	    }
-	} else {
-	    itera = start;
-	    for (h = 0; h < depth; h++) {
-		if (rowSize == elementsPerRow * elementSize) {
-		    /* Ha!  This is mondo easy! */
-		    __GLX_MEM_COPY(iter2, itera,
-				   elementsPerRow * elementSize * height);
-		    iter2 += elementsPerRow * elementSize * height;
-		} else {
-		    iter = itera;
-		    for (i = 0; i < height; i++) {
-			__GLX_MEM_COPY(iter2, iter, elementsPerRow*elementSize);
-			iter2 += elementsPerRow * elementSize;
-			iter += rowSize;
-		    }
-		}
-		itera += imageSize;
-	    }    
-	}
-    }
+      if (swapBytes) {
+         itera = start;
+         for (h = 0; h < depth; h++) {
+            iterb = itera;
+            for (i = 0; i < height; i++) {
+               iterc = iterb;
+               for (j = 0; j < elementsPerRow; j++) {
+                  for (k = 1; k <= elementSize; k++) {
+                     iter2[k - 1] = iterc[elementSize - k];
+                  }
+                  iter2 += elementSize;
+                  iterc += elementSize;
+               }
+               iterb += rowSize;
+            }
+            itera += imageSize;
+         }
+      }
+      else {
+         itera = start;
+         for (h = 0; h < depth; h++) {
+            if (rowSize == elementsPerRow * elementSize) {
+               /* Ha!  This is mondo easy! */
+               __GLX_MEM_COPY(iter2, itera,
+                              elementsPerRow * elementSize * height);
+               iter2 += elementsPerRow * elementSize * height;
+            }
+            else {
+               iter = itera;
+               for (i = 0; i < height; i++) {
+                  __GLX_MEM_COPY(iter2, iter, elementsPerRow * elementSize);
+                  iter2 += elementsPerRow * elementSize;
+                  iter += rowSize;
+               }
+            }
+            itera += imageSize;
+         }
+      }
+   }
 
-    /* Setup store modes that describe what we just did */
-    if (modes) {
-       if ( dim < 3 ) {
-	  (void) memcpy( modes, __glXDefaultPixelStore + 4, 20 );
-       }
-       else {
-	  (void) memcpy( modes, __glXDefaultPixelStore + 0, 36 );
-       }
-    }
+   /* Setup store modes that describe what we just did */
+   if (modes) {
+      if (dim < 3) {
+         (void) memcpy(modes, __glXDefaultPixelStore + 4, 20);
+      }
+      else {
+         (void) memcpy(modes, __glXDefaultPixelStore + 0, 36);
+      }
+   }
 }
 
 /*
 ** Empty a bitmap in LSB_FIRST=GL_FALSE and ALIGNMENT=4 format packing it
 ** into the clients memory using the pixel store PACK modes.
 */
-static void EmptyBitmap(__GLXcontext *gc, GLint width, GLint height,
-			GLenum format, const GLubyte *sourceImage,
-			GLvoid *userdata)
+static void
+EmptyBitmap(__GLXcontext * gc, GLint width, GLint height,
+            GLenum format, const GLubyte * sourceImage, GLvoid * userdata)
 {
-    const __GLXattribute * state = gc->client_state_private;
-    GLint rowLength = state->storePack.rowLength;
-    GLint alignment = state->storePack.alignment;
-    GLint skipPixels = state->storePack.skipPixels;
-    GLint skipRows = state->storePack.skipRows;
-    GLint lsbFirst = state->storePack.lsbFirst;
-    GLint components, groupsPerRow, rowSize, padding, elementsPerRow;
-    GLint sourceRowSize, sourcePadding, sourceSkip;
-    GLubyte *start, *iter;
-    GLint elementsLeft, bitOffset, currentByte, highBitMask, lowBitMask;
-    GLint writeMask, i;
-    GLubyte writeByte;
+   const __GLXattribute *state = gc->client_state_private;
+   GLint rowLength = state->storePack.rowLength;
+   GLint alignment = state->storePack.alignment;
+   GLint skipPixels = state->storePack.skipPixels;
+   GLint skipRows = state->storePack.skipRows;
+   GLint lsbFirst = state->storePack.lsbFirst;
+   GLint components, groupsPerRow, rowSize, padding, elementsPerRow;
+   GLint sourceRowSize, sourcePadding, sourceSkip;
+   GLubyte *start, *iter;
+   GLint elementsLeft, bitOffset, currentByte, highBitMask, lowBitMask;
+   GLint writeMask, i;
+   GLubyte writeByte;
 
-    components = __glElementsPerGroup(format,GL_BITMAP);
-    if (rowLength > 0) {
-	groupsPerRow = rowLength;
-    } else {
-	groupsPerRow = width;
-    }
+   components = __glElementsPerGroup(format, GL_BITMAP);
+   if (rowLength > 0) {
+      groupsPerRow = rowLength;
+   }
+   else {
+      groupsPerRow = width;
+   }
 
-    rowSize = (groupsPerRow * components + 7) >> 3;
-    padding = (rowSize % alignment);
-    if (padding) {
-	rowSize += alignment - padding;
-    }
-    sourceRowSize = (width * components + 7) >> 3;
-    sourcePadding = (sourceRowSize % 4);
-    if (sourcePadding) {
-	sourceSkip = 4 - sourcePadding;
-    } else {
-	sourceSkip = 0;
-    }
-    start = ((GLubyte*) userdata) + skipRows * rowSize + 
-	((skipPixels * components) >> 3);
-    bitOffset = (skipPixels * components) & 7;
-    highBitMask = LowBitsMask[8-bitOffset];
-    lowBitMask = HighBitsMask[bitOffset];
-    elementsPerRow = width * components;
-    for (i = 0; i < height; i++) {
-	elementsLeft = elementsPerRow;
-	iter = start;
-	writeMask = highBitMask;
-	writeByte = 0;
-	while (elementsLeft) {
-	    /* Set up writeMask (to write to current byte) */
-	    if (elementsLeft + bitOffset < 8) {
-		/* Need to trim writeMask */
-		writeMask &= HighBitsMask[bitOffset+elementsLeft];
-	    }
+   rowSize = (groupsPerRow * components + 7) >> 3;
+   padding = (rowSize % alignment);
+   if (padding) {
+      rowSize += alignment - padding;
+   }
+   sourceRowSize = (width * components + 7) >> 3;
+   sourcePadding = (sourceRowSize % 4);
+   if (sourcePadding) {
+      sourceSkip = 4 - sourcePadding;
+   }
+   else {
+      sourceSkip = 0;
+   }
+   start = ((GLubyte *) userdata) + skipRows * rowSize +
+      ((skipPixels * components) >> 3);
+   bitOffset = (skipPixels * components) & 7;
+   highBitMask = LowBitsMask[8 - bitOffset];
+   lowBitMask = HighBitsMask[bitOffset];
+   elementsPerRow = width * components;
+   for (i = 0; i < height; i++) {
+      elementsLeft = elementsPerRow;
+      iter = start;
+      writeMask = highBitMask;
+      writeByte = 0;
+      while (elementsLeft) {
+         /* Set up writeMask (to write to current byte) */
+         if (elementsLeft + bitOffset < 8) {
+            /* Need to trim writeMask */
+            writeMask &= HighBitsMask[bitOffset + elementsLeft];
+         }
 
-	    if (lsbFirst) {
-		currentByte = MsbToLsbTable[iter[0]];
-	    } else {
-		currentByte = iter[0];
-	    }
+         if (lsbFirst) {
+            currentByte = MsbToLsbTable[iter[0]];
+         }
+         else {
+            currentByte = iter[0];
+         }
 
-	    if (bitOffset) {
-		writeByte |= (sourceImage[0] >> bitOffset);
-		currentByte = (currentByte & ~writeMask) | 
-		    (writeByte & writeMask);
-		writeByte = (sourceImage[0] << (8 - bitOffset));
-	    } else {
-		currentByte = (currentByte & ~writeMask) |
-		    (sourceImage[0] & writeMask);
-	    }
+         if (bitOffset) {
+            writeByte |= (sourceImage[0] >> bitOffset);
+            currentByte = (currentByte & ~writeMask) |
+               (writeByte & writeMask);
+            writeByte = (sourceImage[0] << (8 - bitOffset));
+         }
+         else {
+            currentByte = (currentByte & ~writeMask) |
+               (sourceImage[0] & writeMask);
+         }
 
-	    if (lsbFirst) {
-		iter[0] = MsbToLsbTable[currentByte];
-	    } else {
-		iter[0] = currentByte;
-	    }
+         if (lsbFirst) {
+            iter[0] = MsbToLsbTable[currentByte];
+         }
+         else {
+            iter[0] = currentByte;
+         }
 
-	    if (elementsLeft >= 8) {
-		elementsLeft -= 8;
-	    } else {
-		elementsLeft = 0;
-	    }
-	    sourceImage++;
-	    iter++;
-	    writeMask = 0xff;
-	}
-	if (writeByte) {
-	    /* Some data left over that still needs writing */
-	    writeMask &= lowBitMask;
-	    if (lsbFirst) {
-		currentByte = MsbToLsbTable[iter[0]];
-	    } else {
-		currentByte = iter[0];
-	    }
-	    currentByte = (currentByte & ~writeMask) | (writeByte & writeMask);
-	    if (lsbFirst) {
-		iter[0] = MsbToLsbTable[currentByte];
-	    } else {
-		iter[0] = currentByte;
-	    }
-	}
-	start += rowSize;
-	sourceImage += sourceSkip;
-    }
+         if (elementsLeft >= 8) {
+            elementsLeft -= 8;
+         }
+         else {
+            elementsLeft = 0;
+         }
+         sourceImage++;
+         iter++;
+         writeMask = 0xff;
+      }
+      if (writeByte) {
+         /* Some data left over that still needs writing */
+         writeMask &= lowBitMask;
+         if (lsbFirst) {
+            currentByte = MsbToLsbTable[iter[0]];
+         }
+         else {
+            currentByte = iter[0];
+         }
+         currentByte = (currentByte & ~writeMask) | (writeByte & writeMask);
+         if (lsbFirst) {
+            iter[0] = MsbToLsbTable[currentByte];
+         }
+         else {
+            iter[0] = currentByte;
+         }
+      }
+      start += rowSize;
+      sourceImage += sourceSkip;
+   }
 }
 
 /*
@@ -373,70 +387,75 @@ static void EmptyBitmap(__GLXcontext *gc, GLint width, GLint height,
 ** Named __glEmptyImage() because it is the opposite of __glFillImage().
 */
 /* ARGSUSED */
-void __glEmptyImage(__GLXcontext *gc, GLint dim, GLint width, GLint height, 
-		    GLint depth, GLenum format, GLenum type,
-		    const GLubyte *sourceImage, GLvoid *userdata)
+void
+__glEmptyImage(__GLXcontext * gc, GLint dim, GLint width, GLint height,
+               GLint depth, GLenum format, GLenum type,
+               const GLubyte * sourceImage, GLvoid * userdata)
 {
-    const __GLXattribute * state = gc->client_state_private;
-    GLint rowLength = state->storePack.rowLength;
-    GLint imageHeight = state->storePack.imageHeight;
-    GLint alignment = state->storePack.alignment;
-    GLint skipPixels = state->storePack.skipPixels;
-    GLint skipRows = state->storePack.skipRows;
-    GLint skipImages = state->storePack.skipImages;
-    GLint components, elementSize, rowSize, padding, groupsPerRow, groupSize;
-    GLint elementsPerRow, sourceRowSize, sourcePadding, h, i;
-    GLint imageSize, rowsPerImage;
-    GLubyte *start, *iter, *itera;
+   const __GLXattribute *state = gc->client_state_private;
+   GLint rowLength = state->storePack.rowLength;
+   GLint imageHeight = state->storePack.imageHeight;
+   GLint alignment = state->storePack.alignment;
+   GLint skipPixels = state->storePack.skipPixels;
+   GLint skipRows = state->storePack.skipRows;
+   GLint skipImages = state->storePack.skipImages;
+   GLint components, elementSize, rowSize, padding, groupsPerRow, groupSize;
+   GLint elementsPerRow, sourceRowSize, sourcePadding, h, i;
+   GLint imageSize, rowsPerImage;
+   GLubyte *start, *iter, *itera;
 
-    if (type == GL_BITMAP) {
-	EmptyBitmap(gc, width, height, format, sourceImage, userdata);
-    } else {
-	components = __glElementsPerGroup(format,type);
-	if (rowLength > 0) {
-	    groupsPerRow = rowLength;
-	} else {
-	    groupsPerRow = width;
-	}
-	if (imageHeight > 0) {
-	    rowsPerImage = imageHeight;
-	} else {
-	    rowsPerImage = height;
-	}
-	elementSize = __glBytesPerElement(type);
-	groupSize = elementSize * components;
-	rowSize = groupsPerRow * groupSize;
-	padding = (rowSize % alignment);
-	if (padding) {
-	    rowSize += alignment - padding;
-	}
-	sourceRowSize = width * groupSize;
-	sourcePadding = (sourceRowSize % 4);
-	if (sourcePadding) {
-	    sourceRowSize += 4 - sourcePadding;
-	}
-	imageSize = sourceRowSize * rowsPerImage;
-	start = ((GLubyte*) userdata) + skipImages * imageSize +
-		skipRows * rowSize + skipPixels * groupSize;
-	elementsPerRow = width * components;
+   if (type == GL_BITMAP) {
+      EmptyBitmap(gc, width, height, format, sourceImage, userdata);
+   }
+   else {
+      components = __glElementsPerGroup(format, type);
+      if (rowLength > 0) {
+         groupsPerRow = rowLength;
+      }
+      else {
+         groupsPerRow = width;
+      }
+      if (imageHeight > 0) {
+         rowsPerImage = imageHeight;
+      }
+      else {
+         rowsPerImage = height;
+      }
+      elementSize = __glBytesPerElement(type);
+      groupSize = elementSize * components;
+      rowSize = groupsPerRow * groupSize;
+      padding = (rowSize % alignment);
+      if (padding) {
+         rowSize += alignment - padding;
+      }
+      sourceRowSize = width * groupSize;
+      sourcePadding = (sourceRowSize % 4);
+      if (sourcePadding) {
+         sourceRowSize += 4 - sourcePadding;
+      }
+      imageSize = sourceRowSize * rowsPerImage;
+      start = ((GLubyte *) userdata) + skipImages * imageSize +
+         skipRows * rowSize + skipPixels * groupSize;
+      elementsPerRow = width * components;
 
-	itera = start;
-	for (h = 0; h < depth; h++) {
-	    if ((rowSize == sourceRowSize) && (sourcePadding == 0)) {
-		/* Ha!  This is mondo easy! */
-		__GLX_MEM_COPY(itera, sourceImage,
-			       elementsPerRow * elementSize * height);
-		sourceImage += elementsPerRow * elementSize * height;
-	    } else {
-		iter = itera;
-		for (i = 0; i < height; i++) {
-		    __GLX_MEM_COPY(iter, sourceImage,
-				   elementsPerRow * elementSize);
-		    sourceImage += sourceRowSize;
-		    iter += rowSize;
-		}
-	    }
-	    itera += imageSize;
-	}
-    }
+      itera = start;
+      for (h = 0; h < depth; h++) {
+         if ((rowSize == sourceRowSize) && (sourcePadding == 0)) {
+            /* Ha!  This is mondo easy! */
+            __GLX_MEM_COPY(itera, sourceImage,
+                           elementsPerRow * elementSize * height);
+            sourceImage += elementsPerRow * elementSize * height;
+         }
+         else {
+            iter = itera;
+            for (i = 0; i < height; i++) {
+               __GLX_MEM_COPY(iter, sourceImage,
+                              elementsPerRow * elementSize);
+               sourceImage += sourceRowSize;
+               iter += rowSize;
+            }
+         }
+         itera += imageSize;
+      }
+   }
 }
diff --git a/src/glx/x11/pixelstore.c b/src/glx/x11/pixelstore.c
index 6f25ed786ef..0eb31bf1e30 100644
--- a/src/glx/x11/pixelstore.c
+++ b/src/glx/x11/pixelstore.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "glxclient.h"
 #include "indirect.h"
@@ -47,148 +42,157 @@
  * \sa __indirect_glPixelStorei,  __indirect_glPixelStoref
  */
 static void
-send_PixelStore( __GLXcontext * gc, unsigned sop, GLenum pname, 
-		 const void * param )
+send_PixelStore(__GLXcontext * gc, unsigned sop, GLenum pname,
+                const void *param)
 {
-    Display * const dpy = gc->currentDpy;
-    const GLuint cmdlen = 8;
-    if (__builtin_expect(dpy != NULL, 1)) {
-        GLubyte const * pc = __glXSetupSingleRequest(gc, sop, cmdlen);
-        (void) memcpy((void *)(pc + 0), (void *)(&pname), 4);
-        (void) memcpy((void *)(pc + 4), param, 4);
-        UnlockDisplay(dpy); SyncHandle();
-    }
-    return;
+   Display *const dpy = gc->currentDpy;
+   const GLuint cmdlen = 8;
+   if (__builtin_expect(dpy != NULL, 1)) {
+      GLubyte const *pc = __glXSetupSingleRequest(gc, sop, cmdlen);
+      (void) memcpy((void *) (pc + 0), (void *) (&pname), 4);
+      (void) memcpy((void *) (pc + 4), param, 4);
+      UnlockDisplay(dpy);
+      SyncHandle();
+   }
+   return;
 }
 
 /*
 ** Specify parameters that control the storage format of pixel arrays.
 */
-void __indirect_glPixelStoref(GLenum pname, GLfloat param)
+void
+__indirect_glPixelStoref(GLenum pname, GLfloat param)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = gc->client_state_private;
-    Display *dpy = gc->currentDpy;
-    GLuint a;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = gc->client_state_private;
+   Display *dpy = gc->currentDpy;
+   GLuint a;
 
-    if (!dpy) return;
+   if (!dpy)
+      return;
 
-    switch (pname) {
-      case GL_PACK_ROW_LENGTH:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.rowLength = a;
-	break;
-      case GL_PACK_IMAGE_HEIGHT:
-        a = (GLuint) (param + 0.5);
-        if (((GLint) a) < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storePack.imageHeight = a;
-        break;
-      case GL_PACK_SKIP_ROWS:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.skipRows = a;
-	break;
-      case GL_PACK_SKIP_PIXELS:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.skipPixels = a;
-	break;
-      case GL_PACK_SKIP_IMAGES:
-        a = (GLuint) (param + 0.5);
-        if (((GLint) a) < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storePack.skipImages = a;
-        break;
-      case GL_PACK_ALIGNMENT:
-	a = (GLint) (param + 0.5);
-	switch (a) {
-	  case 1: case 2: case 4: case 8:
-	    state->storePack.alignment = a;
-	    break;
-	  default:
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	break;
-      case GL_PACK_SWAP_BYTES:
-	state->storePack.swapEndian = (param != 0);
-	break;
-      case GL_PACK_LSB_FIRST:
-	state->storePack.lsbFirst = (param != 0);
-	break;
+   switch (pname) {
+   case GL_PACK_ROW_LENGTH:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.rowLength = a;
+      break;
+   case GL_PACK_IMAGE_HEIGHT:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.imageHeight = a;
+      break;
+   case GL_PACK_SKIP_ROWS:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipRows = a;
+      break;
+   case GL_PACK_SKIP_PIXELS:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipPixels = a;
+      break;
+   case GL_PACK_SKIP_IMAGES:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipImages = a;
+      break;
+   case GL_PACK_ALIGNMENT:
+      a = (GLint) (param + 0.5);
+      switch (a) {
+      case 1:
+      case 2:
+      case 4:
+      case 8:
+         state->storePack.alignment = a;
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      break;
+   case GL_PACK_SWAP_BYTES:
+      state->storePack.swapEndian = (param != 0);
+      break;
+   case GL_PACK_LSB_FIRST:
+      state->storePack.lsbFirst = (param != 0);
+      break;
 
-      case GL_UNPACK_ROW_LENGTH:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.rowLength = a;
-	break;
-      case GL_UNPACK_IMAGE_HEIGHT:
-        a = (GLuint) (param + 0.5);
-        if (((GLint) a) < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storeUnpack.imageHeight = a;
-        break;
-      case GL_UNPACK_SKIP_ROWS:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.skipRows = a;
-	break;
-      case GL_UNPACK_SKIP_PIXELS:
-	a = (GLuint) (param + 0.5);
-	if (((GLint) a) < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.skipPixels = a;
-	break;
-      case GL_UNPACK_SKIP_IMAGES:
-        a = (GLuint) (param + 0.5);
-        if (((GLint) a) < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storeUnpack.skipImages = a;
-        break;
-      case GL_UNPACK_ALIGNMENT:
-	a = (GLint) (param + 0.5);
-	switch (a) {
-	  case 1: case 2: case 4: case 8:
-	    state->storeUnpack.alignment = a;
-	    break;
-	  default:
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	break;
-      case GL_UNPACK_SWAP_BYTES:
-	state->storeUnpack.swapEndian = (param != 0);
-	break;
-      case GL_UNPACK_LSB_FIRST:
-	state->storeUnpack.lsbFirst = (param != 0);
-	break;
+   case GL_UNPACK_ROW_LENGTH:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.rowLength = a;
+      break;
+   case GL_UNPACK_IMAGE_HEIGHT:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.imageHeight = a;
+      break;
+   case GL_UNPACK_SKIP_ROWS:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipRows = a;
+      break;
+   case GL_UNPACK_SKIP_PIXELS:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipPixels = a;
+      break;
+   case GL_UNPACK_SKIP_IMAGES:
+      a = (GLuint) (param + 0.5);
+      if (((GLint) a) < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipImages = a;
+      break;
+   case GL_UNPACK_ALIGNMENT:
+      a = (GLint) (param + 0.5);
+      switch (a) {
+      case 1:
+      case 2:
+      case 4:
+      case 8:
+         state->storeUnpack.alignment = a;
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      break;
+   case GL_UNPACK_SWAP_BYTES:
+      state->storeUnpack.swapEndian = (param != 0);
+      break;
+   case GL_UNPACK_LSB_FIRST:
+      state->storeUnpack.lsbFirst = (param != 0);
+      break;
 
       /* Group all of the pixel store modes that need to be sent to the
        * server here.  Care must be used to only send modes to the server that
@@ -196,128 +200,136 @@ void __indirect_glPixelStoref(GLenum pname, GLfloat param)
        * server.  GL_PACK_INVERT_MESA is safe in this respect, but other,
        * future modes may not be.
        */
-      case GL_PACK_INVERT_MESA:
-	send_PixelStore( gc, X_GLsop_PixelStoref, pname, & param );
-	break;
+   case GL_PACK_INVERT_MESA:
+      send_PixelStore(gc, X_GLsop_PixelStoref, pname, &param);
+      break;
 
-      default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	break;
-    }
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      break;
+   }
 }
 
-void __indirect_glPixelStorei(GLenum pname, GLint param)
+void
+__indirect_glPixelStorei(GLenum pname, GLint param)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = gc->client_state_private;
-    Display *dpy = gc->currentDpy;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = gc->client_state_private;
+   Display *dpy = gc->currentDpy;
 
-    if (!dpy) return;
+   if (!dpy)
+      return;
 
-    switch (pname) {
-      case GL_PACK_ROW_LENGTH:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.rowLength = param;
-	break;
-      case GL_PACK_IMAGE_HEIGHT:
-        if (param < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storePack.imageHeight = param;
-        break;
-      case GL_PACK_SKIP_ROWS:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.skipRows = param;
-	break;
-      case GL_PACK_SKIP_PIXELS:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storePack.skipPixels = param;
-	break;
-      case GL_PACK_SKIP_IMAGES:
-        if (param < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storePack.skipImages = param;
-        break;
-      case GL_PACK_ALIGNMENT:
-	switch (param) {
-	  case 1: case 2: case 4: case 8:
-	    state->storePack.alignment = param;
-	    break;
-	  default:
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	break;
-      case GL_PACK_SWAP_BYTES:
-	state->storePack.swapEndian = (param != 0);
-	break;
-      case GL_PACK_LSB_FIRST:
-	state->storePack.lsbFirst = (param != 0);
-	break;
+   switch (pname) {
+   case GL_PACK_ROW_LENGTH:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.rowLength = param;
+      break;
+   case GL_PACK_IMAGE_HEIGHT:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.imageHeight = param;
+      break;
+   case GL_PACK_SKIP_ROWS:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipRows = param;
+      break;
+   case GL_PACK_SKIP_PIXELS:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipPixels = param;
+      break;
+   case GL_PACK_SKIP_IMAGES:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storePack.skipImages = param;
+      break;
+   case GL_PACK_ALIGNMENT:
+      switch (param) {
+      case 1:
+      case 2:
+      case 4:
+      case 8:
+         state->storePack.alignment = param;
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      break;
+   case GL_PACK_SWAP_BYTES:
+      state->storePack.swapEndian = (param != 0);
+      break;
+   case GL_PACK_LSB_FIRST:
+      state->storePack.lsbFirst = (param != 0);
+      break;
 
-      case GL_UNPACK_ROW_LENGTH:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.rowLength = param;
-	break;
-      case GL_UNPACK_IMAGE_HEIGHT:
-        if (param < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storeUnpack.imageHeight = param;
-        break;
-      case GL_UNPACK_SKIP_ROWS:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.skipRows = param;
-	break;
-      case GL_UNPACK_SKIP_PIXELS:
-	if (param < 0) {
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	state->storeUnpack.skipPixels = param;
-	break;
-      case GL_UNPACK_SKIP_IMAGES:
-        if (param < 0) {
-            __glXSetError(gc, GL_INVALID_VALUE);
-            return;
-        }
-        state->storeUnpack.skipImages = param;
-        break;
-      case GL_UNPACK_ALIGNMENT:
-	switch (param) {
-	  case 1: case 2: case 4: case 8:
-	    state->storeUnpack.alignment = param;
-	    break;
-	  default:
-	    __glXSetError(gc, GL_INVALID_VALUE);
-	    return;
-	}
-	break;
-      case GL_UNPACK_SWAP_BYTES:
-	state->storeUnpack.swapEndian = (param != 0);
-	break;
-      case GL_UNPACK_LSB_FIRST:
-	state->storeUnpack.lsbFirst = (param != 0);
-	break;
+   case GL_UNPACK_ROW_LENGTH:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.rowLength = param;
+      break;
+   case GL_UNPACK_IMAGE_HEIGHT:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.imageHeight = param;
+      break;
+   case GL_UNPACK_SKIP_ROWS:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipRows = param;
+      break;
+   case GL_UNPACK_SKIP_PIXELS:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipPixels = param;
+      break;
+   case GL_UNPACK_SKIP_IMAGES:
+      if (param < 0) {
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      state->storeUnpack.skipImages = param;
+      break;
+   case GL_UNPACK_ALIGNMENT:
+      switch (param) {
+      case 1:
+      case 2:
+      case 4:
+      case 8:
+         state->storeUnpack.alignment = param;
+         break;
+      default:
+         __glXSetError(gc, GL_INVALID_VALUE);
+         return;
+      }
+      break;
+   case GL_UNPACK_SWAP_BYTES:
+      state->storeUnpack.swapEndian = (param != 0);
+      break;
+   case GL_UNPACK_LSB_FIRST:
+      state->storeUnpack.lsbFirst = (param != 0);
+      break;
 
       /* Group all of the pixel store modes that need to be sent to the
        * server here.  Care must be used to only send modes to the server that
@@ -325,12 +337,12 @@ void __indirect_glPixelStorei(GLenum pname, GLint param)
        * server.  GL_PACK_INVERT_MESA is safe in this respect, but other,
        * future modes may not be.
        */
-      case GL_PACK_INVERT_MESA:
-	send_PixelStore( gc, X_GLsop_PixelStorei, pname, & param );
-	break;
+   case GL_PACK_INVERT_MESA:
+      send_PixelStore(gc, X_GLsop_PixelStorei, pname, &param);
+      break;
 
-      default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	break;
-    }
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      break;
+   }
 }
diff --git a/src/glx/x11/render2.c b/src/glx/x11/render2.c
index b17ad974c85..762950f4ac4 100644
--- a/src/glx/x11/render2.c
+++ b/src/glx/x11/render2.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "packrender.h"
 #include "indirect.h"
@@ -43,320 +38,344 @@
 ** use the pixel header.  See renderpix.c for those routines.
 */
 
-void __indirect_glMap1d(GLenum target, GLdouble u1, GLdouble u2, GLint stride,
-	     GLint order, const GLdouble *pnts)
+void
+__indirect_glMap1d(GLenum target, GLdouble u1, GLdouble u2, GLint stride,
+                   GLint order, const GLdouble * pnts)
 {
-    __GLX_DECLARE_VARIABLES();
-    GLint k;
+   __GLX_DECLARE_VARIABLES();
+   GLint k;
 
-    __GLX_LOAD_VARIABLES();
-    k = __glMap1d_size(target);
-    if (k == 0) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    } else if (stride < k || order <= 0) {
-	__glXSetError(gc, GL_INVALID_VALUE);
-	return;
-    }
-    compsize = k * order * __GLX_SIZE_FLOAT64;
-    cmdlen = 28+compsize;
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   k = __glMap1d_size(target);
+   if (k == 0) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+   else if (stride < k || order <= 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+   compsize = k * order * __GLX_SIZE_FLOAT64;
+   cmdlen = 28 + compsize;
+   if (!gc->currentDpy)
+      return;
 
-    if (cmdlen <= gc->maxSmallRenderCommandSize) {
-	/* Use GLXRender protocol to send small command */
-	__GLX_BEGIN_VARIABLE(X_GLrop_Map1d,cmdlen);
-	__GLX_PUT_DOUBLE(4,u1);
-	__GLX_PUT_DOUBLE(12,u2);
-	__GLX_PUT_LONG(20,target);
-	__GLX_PUT_LONG(24,order);
-	/*
-	** NOTE: the doubles that follow are not aligned because of 3
-	** longs preceeding
-	*/
-	__glFillMap1d(k, order, stride, pnts, (pc+28));
-	__GLX_END(cmdlen);
-    } else {
-	/* Use GLXRenderLarge protocol to send command */
-	__GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map1d,cmdlen+4);
-	__GLX_PUT_DOUBLE(8,u1);
-	__GLX_PUT_DOUBLE(16,u2);
-	__GLX_PUT_LONG(24,target);
-	__GLX_PUT_LONG(28,order);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      /* Use GLXRender protocol to send small command */
+      __GLX_BEGIN_VARIABLE(X_GLrop_Map1d, cmdlen);
+      __GLX_PUT_DOUBLE(4, u1);
+      __GLX_PUT_DOUBLE(12, u2);
+      __GLX_PUT_LONG(20, target);
+      __GLX_PUT_LONG(24, order);
+      /*
+       ** NOTE: the doubles that follow are not aligned because of 3
+       ** longs preceeding
+       */
+      __glFillMap1d(k, order, stride, pnts, (pc + 28));
+      __GLX_END(cmdlen);
+   }
+   else {
+      /* Use GLXRenderLarge protocol to send command */
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map1d, cmdlen + 4);
+      __GLX_PUT_DOUBLE(8, u1);
+      __GLX_PUT_DOUBLE(16, u2);
+      __GLX_PUT_LONG(24, target);
+      __GLX_PUT_LONG(28, order);
 
-	/*
-	** NOTE: the doubles that follow are not aligned because of 3
-	** longs preceeding
-	*/
-	if (stride != k) {
-	    GLubyte *buf;
+      /*
+       ** NOTE: the doubles that follow are not aligned because of 3
+       ** longs preceeding
+       */
+      if (stride != k) {
+         GLubyte *buf;
 
-	    buf = (GLubyte *) Xmalloc(compsize);
-	    if (!buf) {
-		__glXSetError(gc, GL_OUT_OF_MEMORY);
-		return;
-	    }
-	    __glFillMap1d(k, order, stride, pnts, buf);
-	    __glXSendLargeCommand(gc, pc, 32, buf, compsize);
-	    Xfree((char*) buf);
-	} else {
-	    /* Data is already packed.  Just send it out */
-	    __glXSendLargeCommand(gc, pc, 32, pnts, compsize);
-	}
-    }
+         buf = (GLubyte *) Xmalloc(compsize);
+         if (!buf) {
+            __glXSetError(gc, GL_OUT_OF_MEMORY);
+            return;
+         }
+         __glFillMap1d(k, order, stride, pnts, buf);
+         __glXSendLargeCommand(gc, pc, 32, buf, compsize);
+         Xfree((char *) buf);
+      }
+      else {
+         /* Data is already packed.  Just send it out */
+         __glXSendLargeCommand(gc, pc, 32, pnts, compsize);
+      }
+   }
 }
 
-void __indirect_glMap1f(GLenum target, GLfloat u1, GLfloat u2, GLint stride,
-	     GLint order, const GLfloat *pnts)
+void
+__indirect_glMap1f(GLenum target, GLfloat u1, GLfloat u2, GLint stride,
+                   GLint order, const GLfloat * pnts)
 {
-    __GLX_DECLARE_VARIABLES();
-    GLint k;
+   __GLX_DECLARE_VARIABLES();
+   GLint k;
 
-    __GLX_LOAD_VARIABLES();
-    k = __glMap1f_size(target);
-    if (k == 0) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    } else if (stride < k || order <= 0) {
-	__glXSetError(gc, GL_INVALID_VALUE);
-	return;
-    }
-    compsize = k * order * __GLX_SIZE_FLOAT32;
-    cmdlen = 20+compsize;
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   k = __glMap1f_size(target);
+   if (k == 0) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+   else if (stride < k || order <= 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+   compsize = k * order * __GLX_SIZE_FLOAT32;
+   cmdlen = 20 + compsize;
+   if (!gc->currentDpy)
+      return;
 
-    /*
+   /*
     ** The order that arguments are packed is different from the order
     ** for glMap1d.
     */
-    if (cmdlen <= gc->maxSmallRenderCommandSize) {
-	/* Use GLXRender protocol to send small command */
-	__GLX_BEGIN_VARIABLE(X_GLrop_Map1f,cmdlen);
-	__GLX_PUT_LONG(4,target);
-	__GLX_PUT_FLOAT(8,u1);
-	__GLX_PUT_FLOAT(12,u2);
-	__GLX_PUT_LONG(16,order);
-	__glFillMap1f(k, order, stride, pnts, (GLubyte*) (pc+20));
-	__GLX_END(cmdlen);
-    } else {
-	/* Use GLXRenderLarge protocol to send command */
-	__GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map1f,cmdlen+4);
-	__GLX_PUT_LONG(8,target);
-	__GLX_PUT_FLOAT(12,u1);
-	__GLX_PUT_FLOAT(16,u2);
-	__GLX_PUT_LONG(20,order);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      /* Use GLXRender protocol to send small command */
+      __GLX_BEGIN_VARIABLE(X_GLrop_Map1f, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_FLOAT(8, u1);
+      __GLX_PUT_FLOAT(12, u2);
+      __GLX_PUT_LONG(16, order);
+      __glFillMap1f(k, order, stride, pnts, (GLubyte *) (pc + 20));
+      __GLX_END(cmdlen);
+   }
+   else {
+      /* Use GLXRenderLarge protocol to send command */
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map1f, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_FLOAT(12, u1);
+      __GLX_PUT_FLOAT(16, u2);
+      __GLX_PUT_LONG(20, order);
 
-	if (stride != k) {
-	    GLubyte *buf;
+      if (stride != k) {
+         GLubyte *buf;
 
-	    buf = (GLubyte *) Xmalloc(compsize);
-	    if (!buf) {
-		__glXSetError(gc, GL_OUT_OF_MEMORY);
-		return;
-	    }
-	    __glFillMap1f(k, order, stride, pnts, buf);
-	    __glXSendLargeCommand(gc, pc, 24, buf, compsize);
-	    Xfree((char*) buf);
-	} else {
-	    /* Data is already packed.  Just send it out */
-	    __glXSendLargeCommand(gc, pc, 24, pnts, compsize);
-	}
-    }
+         buf = (GLubyte *) Xmalloc(compsize);
+         if (!buf) {
+            __glXSetError(gc, GL_OUT_OF_MEMORY);
+            return;
+         }
+         __glFillMap1f(k, order, stride, pnts, buf);
+         __glXSendLargeCommand(gc, pc, 24, buf, compsize);
+         Xfree((char *) buf);
+      }
+      else {
+         /* Data is already packed.  Just send it out */
+         __glXSendLargeCommand(gc, pc, 24, pnts, compsize);
+      }
+   }
 }
 
-void __indirect_glMap2d(GLenum target, GLdouble u1, GLdouble u2, GLint ustr, GLint uord,
-	     GLdouble v1, GLdouble v2, GLint vstr, GLint vord,
-	     const GLdouble *pnts)
+void
+__indirect_glMap2d(GLenum target, GLdouble u1, GLdouble u2, GLint ustr,
+                   GLint uord, GLdouble v1, GLdouble v2, GLint vstr,
+                   GLint vord, const GLdouble * pnts)
 {
-    __GLX_DECLARE_VARIABLES();
-    GLint k;
+   __GLX_DECLARE_VARIABLES();
+   GLint k;
 
-    __GLX_LOAD_VARIABLES();
-    k = __glMap2d_size(target);
-    if (k == 0) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    } else if (vstr < k || ustr < k || vord <= 0 || uord <= 0) {
-	__glXSetError(gc, GL_INVALID_VALUE);
-	return;
-    }
-    compsize = k * uord * vord * __GLX_SIZE_FLOAT64;
-    cmdlen = 48+compsize; 
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   k = __glMap2d_size(target);
+   if (k == 0) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+   else if (vstr < k || ustr < k || vord <= 0 || uord <= 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+   compsize = k * uord * vord * __GLX_SIZE_FLOAT64;
+   cmdlen = 48 + compsize;
+   if (!gc->currentDpy)
+      return;
 
-    if (cmdlen <= gc->maxSmallRenderCommandSize) {
-	/* Use GLXRender protocol to send small command */
-	__GLX_BEGIN_VARIABLE(X_GLrop_Map2d,cmdlen);
-	__GLX_PUT_DOUBLE(4,u1);
-	__GLX_PUT_DOUBLE(12,u2);
-	__GLX_PUT_DOUBLE(20,v1);
-	__GLX_PUT_DOUBLE(28,v2);
-	__GLX_PUT_LONG(36,target);
-	__GLX_PUT_LONG(40,uord);
-	__GLX_PUT_LONG(44,vord);
-	/*
-	** Pack into a u-major ordering.
-	** NOTE: the doubles that follow are not aligned because of 5
-	** longs preceeding
-	*/
-	__glFillMap2d(k, uord, vord, ustr, vstr, pnts, (GLdouble*) (pc+48));
-	__GLX_END(cmdlen);
-    } else {
-	/* Use GLXRenderLarge protocol to send command */
-	__GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map2d,cmdlen+4);
-	__GLX_PUT_DOUBLE(8,u1);
-	__GLX_PUT_DOUBLE(16,u2);
-	__GLX_PUT_DOUBLE(24,v1);
-	__GLX_PUT_DOUBLE(32,v2);
-	__GLX_PUT_LONG(40,target);
-	__GLX_PUT_LONG(44,uord);
-	__GLX_PUT_LONG(48,vord);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      /* Use GLXRender protocol to send small command */
+      __GLX_BEGIN_VARIABLE(X_GLrop_Map2d, cmdlen);
+      __GLX_PUT_DOUBLE(4, u1);
+      __GLX_PUT_DOUBLE(12, u2);
+      __GLX_PUT_DOUBLE(20, v1);
+      __GLX_PUT_DOUBLE(28, v2);
+      __GLX_PUT_LONG(36, target);
+      __GLX_PUT_LONG(40, uord);
+      __GLX_PUT_LONG(44, vord);
+      /*
+       ** Pack into a u-major ordering.
+       ** NOTE: the doubles that follow are not aligned because of 5
+       ** longs preceeding
+       */
+      __glFillMap2d(k, uord, vord, ustr, vstr, pnts, (GLdouble *) (pc + 48));
+      __GLX_END(cmdlen);
+   }
+   else {
+      /* Use GLXRenderLarge protocol to send command */
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map2d, cmdlen + 4);
+      __GLX_PUT_DOUBLE(8, u1);
+      __GLX_PUT_DOUBLE(16, u2);
+      __GLX_PUT_DOUBLE(24, v1);
+      __GLX_PUT_DOUBLE(32, v2);
+      __GLX_PUT_LONG(40, target);
+      __GLX_PUT_LONG(44, uord);
+      __GLX_PUT_LONG(48, vord);
 
-	/*
-	** NOTE: the doubles that follow are not aligned because of 5
-	** longs preceeding
-	*/
-	if ((vstr != k) || (ustr != k*vord)) {
-	    GLdouble *buf;
+      /*
+       ** NOTE: the doubles that follow are not aligned because of 5
+       ** longs preceeding
+       */
+      if ((vstr != k) || (ustr != k * vord)) {
+         GLdouble *buf;
 
-	    buf = (GLdouble *) Xmalloc(compsize);
-	    if (!buf) {
-		__glXSetError(gc, GL_OUT_OF_MEMORY);
-		return;
-	    }
-	    /*
-	    ** Pack into a u-major ordering.
-	    */
-	    __glFillMap2d(k, uord, vord, ustr, vstr, pnts, buf);
-	    __glXSendLargeCommand(gc, pc, 52, buf, compsize);
-	    Xfree((char*) buf);
-	} else {
-	    /* Data is already packed.  Just send it out */
-	    __glXSendLargeCommand(gc, pc, 52, pnts, compsize);
-	}
-    }
+         buf = (GLdouble *) Xmalloc(compsize);
+         if (!buf) {
+            __glXSetError(gc, GL_OUT_OF_MEMORY);
+            return;
+         }
+         /*
+          ** Pack into a u-major ordering.
+          */
+         __glFillMap2d(k, uord, vord, ustr, vstr, pnts, buf);
+         __glXSendLargeCommand(gc, pc, 52, buf, compsize);
+         Xfree((char *) buf);
+      }
+      else {
+         /* Data is already packed.  Just send it out */
+         __glXSendLargeCommand(gc, pc, 52, pnts, compsize);
+      }
+   }
 }
 
-void __indirect_glMap2f(GLenum target, GLfloat u1, GLfloat u2, GLint ustr, GLint uord,
-	     GLfloat v1, GLfloat v2, GLint vstr, GLint vord,
-	     const GLfloat *pnts)
+void
+__indirect_glMap2f(GLenum target, GLfloat u1, GLfloat u2, GLint ustr,
+                   GLint uord, GLfloat v1, GLfloat v2, GLint vstr, GLint vord,
+                   const GLfloat * pnts)
 {
-    __GLX_DECLARE_VARIABLES();
-    GLint k;
+   __GLX_DECLARE_VARIABLES();
+   GLint k;
 
-    __GLX_LOAD_VARIABLES();
-    k = __glMap2f_size(target);
-    if (k == 0) {
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
-    } else if (vstr < k || ustr < k || vord <= 0 || uord <= 0) {
-	__glXSetError(gc, GL_INVALID_VALUE);
-	return;
-    }
-    compsize = k * uord * vord * __GLX_SIZE_FLOAT32;
-    cmdlen = 32+compsize; 
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   k = __glMap2f_size(target);
+   if (k == 0) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+   else if (vstr < k || ustr < k || vord <= 0 || uord <= 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+   compsize = k * uord * vord * __GLX_SIZE_FLOAT32;
+   cmdlen = 32 + compsize;
+   if (!gc->currentDpy)
+      return;
 
-    /*
+   /*
     ** The order that arguments are packed is different from the order
     ** for glMap2d.
     */
-    if (cmdlen <= gc->maxSmallRenderCommandSize) {
-	/* Use GLXRender protocol to send small command */
-	__GLX_BEGIN_VARIABLE(X_GLrop_Map2f,cmdlen);
-	__GLX_PUT_LONG(4,target);
-	__GLX_PUT_FLOAT(8,u1);
-	__GLX_PUT_FLOAT(12,u2);
-	__GLX_PUT_LONG(16,uord);
-	__GLX_PUT_FLOAT(20,v1);
-	__GLX_PUT_FLOAT(24,v2);
-	__GLX_PUT_LONG(28,vord);
-	/*
-	** Pack into a u-major ordering.
-	*/
-	__glFillMap2f(k, uord, vord, ustr, vstr, pnts, (GLfloat*) (pc+32));
-	__GLX_END(cmdlen);
-    } else {
-	/* Use GLXRenderLarge protocol to send command */
-	__GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map2f,cmdlen+4);
-	__GLX_PUT_LONG(8,target);
-	__GLX_PUT_FLOAT(12,u1);
-	__GLX_PUT_FLOAT(16,u2);
-	__GLX_PUT_LONG(20,uord);
-	__GLX_PUT_FLOAT(24,v1);
-	__GLX_PUT_FLOAT(28,v2);
-	__GLX_PUT_LONG(32,vord);
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      /* Use GLXRender protocol to send small command */
+      __GLX_BEGIN_VARIABLE(X_GLrop_Map2f, cmdlen);
+      __GLX_PUT_LONG(4, target);
+      __GLX_PUT_FLOAT(8, u1);
+      __GLX_PUT_FLOAT(12, u2);
+      __GLX_PUT_LONG(16, uord);
+      __GLX_PUT_FLOAT(20, v1);
+      __GLX_PUT_FLOAT(24, v2);
+      __GLX_PUT_LONG(28, vord);
+      /*
+       ** Pack into a u-major ordering.
+       */
+      __glFillMap2f(k, uord, vord, ustr, vstr, pnts, (GLfloat *) (pc + 32));
+      __GLX_END(cmdlen);
+   }
+   else {
+      /* Use GLXRenderLarge protocol to send command */
+      __GLX_BEGIN_VARIABLE_LARGE(X_GLrop_Map2f, cmdlen + 4);
+      __GLX_PUT_LONG(8, target);
+      __GLX_PUT_FLOAT(12, u1);
+      __GLX_PUT_FLOAT(16, u2);
+      __GLX_PUT_LONG(20, uord);
+      __GLX_PUT_FLOAT(24, v1);
+      __GLX_PUT_FLOAT(28, v2);
+      __GLX_PUT_LONG(32, vord);
 
-	if ((vstr != k) || (ustr != k*vord)) {
-	    GLfloat *buf;
+      if ((vstr != k) || (ustr != k * vord)) {
+         GLfloat *buf;
 
-	    buf = (GLfloat *) Xmalloc(compsize);
-	    if (!buf) {
-		__glXSetError(gc, GL_OUT_OF_MEMORY);
-		return;
-	    }
-	    /*
-	    ** Pack into a u-major ordering.
-	    */
-	    __glFillMap2f(k, uord, vord, ustr, vstr, pnts, buf);
-	    __glXSendLargeCommand(gc, pc, 36, buf, compsize);
-	    Xfree((char*) buf);
-	} else {
-	    /* Data is already packed.  Just send it out */
-	    __glXSendLargeCommand(gc, pc, 36, pnts, compsize);
-	}
-    }
+         buf = (GLfloat *) Xmalloc(compsize);
+         if (!buf) {
+            __glXSetError(gc, GL_OUT_OF_MEMORY);
+            return;
+         }
+         /*
+          ** Pack into a u-major ordering.
+          */
+         __glFillMap2f(k, uord, vord, ustr, vstr, pnts, buf);
+         __glXSendLargeCommand(gc, pc, 36, buf, compsize);
+         Xfree((char *) buf);
+      }
+      else {
+         /* Data is already packed.  Just send it out */
+         __glXSendLargeCommand(gc, pc, 36, pnts, compsize);
+      }
+   }
 }
 
-void __indirect_glEnable(GLenum cap)
+void
+__indirect_glEnable(GLenum cap)
 {
-    __GLX_DECLARE_VARIABLES();
+   __GLX_DECLARE_VARIABLES();
 
-    __GLX_LOAD_VARIABLES();
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   if (!gc->currentDpy)
+      return;
 
-    switch(cap) {
-	case GL_COLOR_ARRAY:
-	case GL_EDGE_FLAG_ARRAY:
-	case GL_INDEX_ARRAY:
-	case GL_NORMAL_ARRAY:
-	case GL_TEXTURE_COORD_ARRAY:
-	case GL_VERTEX_ARRAY:
-	case GL_SECONDARY_COLOR_ARRAY:
-	case GL_FOG_COORD_ARRAY:
-	    __indirect_glEnableClientState(cap);
-	    return;
-	default:
-	    break;
-    }
+   switch (cap) {
+   case GL_COLOR_ARRAY:
+   case GL_EDGE_FLAG_ARRAY:
+   case GL_INDEX_ARRAY:
+   case GL_NORMAL_ARRAY:
+   case GL_TEXTURE_COORD_ARRAY:
+   case GL_VERTEX_ARRAY:
+   case GL_SECONDARY_COLOR_ARRAY:
+   case GL_FOG_COORD_ARRAY:
+      __indirect_glEnableClientState(cap);
+      return;
+   default:
+      break;
+   }
 
-    __GLX_BEGIN(X_GLrop_Enable,8);
-    __GLX_PUT_LONG(4,cap);
-    __GLX_END(8);
+   __GLX_BEGIN(X_GLrop_Enable, 8);
+   __GLX_PUT_LONG(4, cap);
+   __GLX_END(8);
 }
 
-void __indirect_glDisable(GLenum cap)
+void
+__indirect_glDisable(GLenum cap)
 {
-    __GLX_DECLARE_VARIABLES();
+   __GLX_DECLARE_VARIABLES();
 
-    __GLX_LOAD_VARIABLES();
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   if (!gc->currentDpy)
+      return;
 
-    switch(cap) {
-	case GL_COLOR_ARRAY:
-	case GL_EDGE_FLAG_ARRAY:
-	case GL_INDEX_ARRAY:
-	case GL_NORMAL_ARRAY:
-	case GL_TEXTURE_COORD_ARRAY:
-	case GL_VERTEX_ARRAY:
-	case GL_SECONDARY_COLOR_ARRAY:
-	case GL_FOG_COORD_ARRAY:
-	    __indirect_glDisableClientState(cap);
-	    return;
-	default:
-	    break;
-    }
+   switch (cap) {
+   case GL_COLOR_ARRAY:
+   case GL_EDGE_FLAG_ARRAY:
+   case GL_INDEX_ARRAY:
+   case GL_NORMAL_ARRAY:
+   case GL_TEXTURE_COORD_ARRAY:
+   case GL_VERTEX_ARRAY:
+   case GL_SECONDARY_COLOR_ARRAY:
+   case GL_FOG_COORD_ARRAY:
+      __indirect_glDisableClientState(cap);
+      return;
+   default:
+      break;
+   }
 
-    __GLX_BEGIN(X_GLrop_Disable,8);
-    __GLX_PUT_LONG(4,cap);
-    __GLX_END(8);
+   __GLX_BEGIN(X_GLrop_Disable, 8);
+   __GLX_PUT_LONG(4, cap);
+   __GLX_END(8);
 }
diff --git a/src/glx/x11/renderpix.c b/src/glx/x11/renderpix.c
index 41a7a2d762e..9919bbcde3b 100644
--- a/src/glx/x11/renderpix.c
+++ b/src/glx/x11/renderpix.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 /*
  * (C) Copyright IBM Corporation 2005
@@ -87,42 +82,43 @@
  * broken.
  */
 void
-__glXSendLargeImage(__GLXcontext *gc, GLint compsize, GLint dim,
-		    GLint width, GLint height, GLint depth,
-		    GLenum format, GLenum type, const GLvoid *src,
-		    GLubyte *pc, GLubyte *modes)
+__glXSendLargeImage(__GLXcontext * gc, GLint compsize, GLint dim,
+                    GLint width, GLint height, GLint depth,
+                    GLenum format, GLenum type, const GLvoid * src,
+                    GLubyte * pc, GLubyte * modes)
 {
-    if ( !gc->fastImageUnpack || (src == NULL) ) {
-	/* Allocate a temporary holding buffer */
-	GLubyte *buf = (GLubyte *) Xmalloc(compsize);
-	if (!buf) {
-	    __glXSetError(gc, GL_OUT_OF_MEMORY);
-	    return;
-	}
+   if (!gc->fastImageUnpack || (src == NULL)) {
+      /* Allocate a temporary holding buffer */
+      GLubyte *buf = (GLubyte *) Xmalloc(compsize);
+      if (!buf) {
+         __glXSetError(gc, GL_OUT_OF_MEMORY);
+         return;
+      }
 
-	/* Apply pixel store unpack modes to copy data into buf */
-	if ( src != NULL ) {
-	    (*gc->fillImage)(gc, dim, width, height, depth, format, type,
-			     src, buf, modes);
-	}
-	else {
-	    if ( dim < 3 ) {
-		(void) memcpy( modes, __glXDefaultPixelStore + 4, 20 );
-	    }
-	    else {
-		(void) memcpy( modes, __glXDefaultPixelStore + 0, 36 );
-	    }
-	}
+      /* Apply pixel store unpack modes to copy data into buf */
+      if (src != NULL) {
+         (*gc->fillImage) (gc, dim, width, height, depth, format, type,
+                           src, buf, modes);
+      }
+      else {
+         if (dim < 3) {
+            (void) memcpy(modes, __glXDefaultPixelStore + 4, 20);
+         }
+         else {
+            (void) memcpy(modes, __glXDefaultPixelStore + 0, 36);
+         }
+      }
 
-	/* Send large command */
-	__glXSendLargeCommand(gc, gc->pc, pc - gc->pc, buf, compsize);
+      /* Send large command */
+      __glXSendLargeCommand(gc, gc->pc, pc - gc->pc, buf, compsize);
 
-	/* Free buffer */
-	Xfree((char*) buf);
-    } else {
-	/* Just send the data straight as is */
-	__glXSendLargeCommand(gc, gc->pc, pc - gc->pc, pc, compsize);
-    }
+      /* Free buffer */
+      Xfree((char *) buf);
+   }
+   else {
+      /* Just send the data straight as is */
+      __glXSendLargeCommand(gc, gc->pc, pc - gc->pc, pc, compsize);
+   }
 }
 
 /************************************************************************/
@@ -134,81 +130,89 @@ __glXSendLargeImage(__GLXcontext *gc, GLint compsize, GLint dim,
  * The \c fastImageUnpack path, which is thankfully never used, is completely
  * broken.
  */
-void __indirect_glSeparableFilter2D(GLenum target, GLenum internalformat,
-				GLsizei width, GLsizei height, GLenum format,
-				GLenum type, const GLvoid *row,
-				const GLvoid *column)
+void
+__indirect_glSeparableFilter2D(GLenum target, GLenum internalformat,
+                               GLsizei width, GLsizei height, GLenum format,
+                               GLenum type, const GLvoid * row,
+                               const GLvoid * column)
 {
-    __GLX_DECLARE_VARIABLES();
-    GLuint compsize2, hdrlen, totalhdrlen, image1len, image2len;
+   __GLX_DECLARE_VARIABLES();
+   GLuint compsize2, hdrlen, totalhdrlen, image1len, image2len;
 
-    __GLX_LOAD_VARIABLES();
-    compsize = __glImageSize(width, 1, 1, format, type, 0);
-    compsize2 = __glImageSize(height, 1, 1, format, type, 0);
-    totalhdrlen = __GLX_PAD(__GLX_CONV_FILT_CMD_HDR_SIZE);
-    hdrlen = __GLX_PAD(__GLX_CONV_FILT_HDR_SIZE);
-    image1len = __GLX_PAD(compsize);
-    image2len = __GLX_PAD(compsize2);
-    cmdlen = totalhdrlen + image1len + image2len;
-    if (!gc->currentDpy) return;
+   __GLX_LOAD_VARIABLES();
+   compsize = __glImageSize(width, 1, 1, format, type, 0);
+   compsize2 = __glImageSize(height, 1, 1, format, type, 0);
+   totalhdrlen = __GLX_PAD(__GLX_CONV_FILT_CMD_HDR_SIZE);
+   hdrlen = __GLX_PAD(__GLX_CONV_FILT_HDR_SIZE);
+   image1len = __GLX_PAD(compsize);
+   image2len = __GLX_PAD(compsize2);
+   cmdlen = totalhdrlen + image1len + image2len;
+   if (!gc->currentDpy)
+      return;
 
-    if (cmdlen <= gc->maxSmallRenderCommandSize) {
-	/* Use GLXRender protocol to send small command */
-	__GLX_BEGIN_VARIABLE_WITH_PIXEL(X_GLrop_SeparableFilter2D, cmdlen);
-	__GLX_PUT_LONG(0,target);
-	__GLX_PUT_LONG(4,internalformat);
-	__GLX_PUT_LONG(8,width);
-	__GLX_PUT_LONG(12,height);
-	__GLX_PUT_LONG(16,format);
-	__GLX_PUT_LONG(20,type);
-	pc += hdrlen;
-	if (compsize > 0) {
-	    (*gc->fillImage)(gc, 1, width, 1, 1, format, type,
-			     row, pc, pixelHeaderPC);
-	    pc += image1len;
-	}
-	if (compsize2 > 0) {
-            (*gc->fillImage)(gc, 1, height, 1, 1, format, type,
-			     column, pc, NULL);
-	    pc += image2len;
-	}
-	if ((compsize == 0) && (compsize2 == 0)) {
-	    /* Setup default store modes */
-	    (void) memcpy( pixelHeaderPC, __glXDefaultPixelStore + 4, 20 );
-	}
-	__GLX_END(0);
-    } else {
-	const GLint bufsize = image1len + image2len;
+   if (cmdlen <= gc->maxSmallRenderCommandSize) {
+      /* Use GLXRender protocol to send small command */
+      __GLX_BEGIN_VARIABLE_WITH_PIXEL(X_GLrop_SeparableFilter2D, cmdlen);
+      __GLX_PUT_LONG(0, target);
+      __GLX_PUT_LONG(4, internalformat);
+      __GLX_PUT_LONG(8, width);
+      __GLX_PUT_LONG(12, height);
+      __GLX_PUT_LONG(16, format);
+      __GLX_PUT_LONG(20, type);
+      pc += hdrlen;
+      if (compsize > 0) {
+         (*gc->fillImage) (gc, 1, width, 1, 1, format, type,
+                           row, pc, pixelHeaderPC);
+         pc += image1len;
+      }
+      if (compsize2 > 0) {
+         (*gc->fillImage) (gc, 1, height, 1, 1, format, type,
+                           column, pc, NULL);
+         pc += image2len;
+      }
+      if ((compsize == 0) && (compsize2 == 0)) {
+         /* Setup default store modes */
+         (void) memcpy(pixelHeaderPC, __glXDefaultPixelStore + 4, 20);
+      }
+      __GLX_END(0);
+   }
+   else {
+      const GLint bufsize = image1len + image2len;
 
-	/* Use GLXRenderLarge protocol to send command */
-	__GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL(X_GLrop_SeparableFilter2D,cmdlen+4);
-	__GLX_PUT_LONG(0,target);
-	__GLX_PUT_LONG(4,internalformat);
-	__GLX_PUT_LONG(8,width);
-	__GLX_PUT_LONG(12,height);
-	__GLX_PUT_LONG(16,format);
-	__GLX_PUT_LONG(20,type);
-	pc += hdrlen;
+      /* Use GLXRenderLarge protocol to send command */
+      __GLX_BEGIN_VARIABLE_LARGE_WITH_PIXEL(X_GLrop_SeparableFilter2D,
+                                            cmdlen + 4);
+      __GLX_PUT_LONG(0, target);
+      __GLX_PUT_LONG(4, internalformat);
+      __GLX_PUT_LONG(8, width);
+      __GLX_PUT_LONG(12, height);
+      __GLX_PUT_LONG(16, format);
+      __GLX_PUT_LONG(20, type);
+      pc += hdrlen;
 
-	if (!gc->fastImageUnpack) {
-	    /* Allocate a temporary holding buffer */
-	    GLubyte *buf = (GLubyte *) Xmalloc(bufsize);
-	    if (!buf) {
-		__glXSetError(gc, GL_OUT_OF_MEMORY);
-		return;
-	    }
-	    (*gc->fillImage)(gc, 1, width, 1, 1, format, type, row, buf, pixelHeaderPC);
+      if (!gc->fastImageUnpack) {
+         /* Allocate a temporary holding buffer */
+         GLubyte *buf = (GLubyte *) Xmalloc(bufsize);
+         if (!buf) {
+            __glXSetError(gc, GL_OUT_OF_MEMORY);
+            return;
+         }
+         (*gc->fillImage) (gc, 1, width, 1, 1, format, type, row, buf,
+                           pixelHeaderPC);
 
-	    (*gc->fillImage)(gc, 1, height, 1, 1, format, type, column,
-				 buf + image1len, pixelHeaderPC);
+         (*gc->fillImage) (gc, 1, height, 1, 1, format, type, column,
+                           buf + image1len, pixelHeaderPC);
 
-	    /* Send large command */
-	    __glXSendLargeCommand(gc, gc->pc, (GLint)(pc - gc->pc), buf, bufsize);
-	    /* Free buffer */
-	    Xfree((char*) buf);
-	} else {
-	    /* Just send the data straight as is */
-	    __glXSendLargeCommand(gc, gc->pc, (GLint)(pc - gc->pc), pc, bufsize);
-	}
-    }
+         /* Send large command */
+         __glXSendLargeCommand(gc, gc->pc, (GLint) (pc - gc->pc), buf,
+                               bufsize);
+         /* Free buffer */
+         Xfree((char *) buf);
+      }
+      else {
+         /* Just send the data straight as is */
+         __glXSendLargeCommand(gc, gc->pc, (GLint) (pc - gc->pc), pc,
+                               bufsize);
+      }
+   }
 }
diff --git a/src/glx/x11/single2.c b/src/glx/x11/single2.c
index 35fe417b62b..0b2d5113cc8 100644
--- a/src/glx/x11/single2.c
+++ b/src/glx/x11/single2.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include <stdio.h>
 #include <assert.h>
@@ -40,57 +35,69 @@
 #include "glxextensions.h"
 #include "indirect.h"
 #include "indirect_vertex_array.h"
+#include "dispatch.h"
+#include "glapi.h"
+#ifdef USE_XCB
+#include <xcb/xcb.h>
+#include <xcb/glx.h>
+#include <X11/Xlib-xcb.h>
+#endif /* USE_XCB */
+
 
 /* Used for GL_ARB_transpose_matrix */
-static void TransposeMatrixf(GLfloat m[16])
+static void
+TransposeMatrixf(GLfloat m[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < i; j++) {
-            GLfloat tmp = m[i*4+j];
-            m[i*4+j] = m[j*4+i];
-            m[j*4+i] = tmp;
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < i; j++) {
+         GLfloat tmp = m[i * 4 + j];
+         m[i * 4 + j] = m[j * 4 + i];
+         m[j * 4 + i] = tmp;
+      }
+   }
 }
 
 /* Used for GL_ARB_transpose_matrix */
-static void TransposeMatrixb(GLboolean m[16])
+static void
+TransposeMatrixb(GLboolean m[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < i; j++) {
-            GLboolean tmp = m[i*4+j];
-            m[i*4+j] = m[j*4+i];
-            m[j*4+i] = tmp;
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < i; j++) {
+         GLboolean tmp = m[i * 4 + j];
+         m[i * 4 + j] = m[j * 4 + i];
+         m[j * 4 + i] = tmp;
+      }
+   }
 }
 
 /* Used for GL_ARB_transpose_matrix */
-static void TransposeMatrixd(GLdouble m[16])
+static void
+TransposeMatrixd(GLdouble m[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < i; j++) {
-            GLdouble tmp = m[i*4+j];
-            m[i*4+j] = m[j*4+i];
-            m[j*4+i] = tmp;
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < i; j++) {
+         GLdouble tmp = m[i * 4 + j];
+         m[i * 4 + j] = m[j * 4 + i];
+         m[j * 4 + i] = tmp;
+      }
+   }
 }
 
 /* Used for GL_ARB_transpose_matrix */
-static void TransposeMatrixi(GLint m[16])
+static void
+TransposeMatrixi(GLint m[16])
 {
-    int i, j;
-    for (i = 0; i < 4; i++) {
-        for (j = 0; j < i; j++) {
-            GLint tmp = m[i*4+j];
-            m[i*4+j] = m[j*4+i];
-            m[j*4+i] = tmp;
-        }
-    }
+   int i, j;
+   for (i = 0; i < 4; i++) {
+      for (j = 0; j < i; j++) {
+         GLint tmp = m[i * 4 + j];
+         m[i * 4 + j] = m[j * 4 + i];
+         m[j * 4 + i] = tmp;
+      }
+   }
 }
 
 
@@ -99,41 +106,42 @@ static void TransposeMatrixi(GLint m[16])
  * that are not transpose-matrix enums are unaffected.
  */
 static GLenum
-RemapTransposeEnum( GLenum e )
+RemapTransposeEnum(GLenum e)
 {
-    switch( e ) {
-    case GL_TRANSPOSE_MODELVIEW_MATRIX:
-    case GL_TRANSPOSE_PROJECTION_MATRIX:
-    case GL_TRANSPOSE_TEXTURE_MATRIX:
-	return e - (GL_TRANSPOSE_MODELVIEW_MATRIX - GL_MODELVIEW_MATRIX);
-    case GL_TRANSPOSE_COLOR_MATRIX:
-	return GL_COLOR_MATRIX;
-    default:
-	return e;
-    };
+   switch (e) {
+   case GL_TRANSPOSE_MODELVIEW_MATRIX:
+   case GL_TRANSPOSE_PROJECTION_MATRIX:
+   case GL_TRANSPOSE_TEXTURE_MATRIX:
+      return e - (GL_TRANSPOSE_MODELVIEW_MATRIX - GL_MODELVIEW_MATRIX);
+   case GL_TRANSPOSE_COLOR_MATRIX:
+      return GL_COLOR_MATRIX;
+   default:
+      return e;
+   };
 }
 
 
-GLenum __indirect_glGetError(void)
+GLenum
+__indirect_glGetError(void)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    GLuint retval = GL_NO_ERROR;
-    xGLXGetErrorReply reply;
-
-    if (gc->error) {
-	/* Use internal error first */
-	retval = gc->error;
-	gc->error = GL_NO_ERROR;
-	return retval;
-    }
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_GetError,0);
-    __GLX_SINGLE_READ_XREPLY();
-    retval = reply.error;
-    __GLX_SINGLE_END();
-
-    return retval;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   GLuint retval = GL_NO_ERROR;
+   xGLXGetErrorReply reply;
+
+   if (gc->error) {
+      /* Use internal error first */
+      retval = gc->error;
+      gc->error = GL_NO_ERROR;
+      return retval;
+   }
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetError, 0);
+   __GLX_SINGLE_READ_XREPLY();
+   retval = reply.error;
+   __GLX_SINGLE_END();
+
+   return retval;
 }
 
 
@@ -144,447 +152,473 @@ GLenum __indirect_glGetError(void)
  * On success \c GL_TRUE is returned.  Otherwise, \c GL_FALSE is returned.
  */
 static GLboolean
-get_client_data( __GLXcontext * gc, GLenum cap, GLintptr * data )
+get_client_data(__GLXcontext * gc, GLenum cap, GLintptr * data)
 {
-    GLboolean retval = GL_TRUE;
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    const GLint tex_unit = __glXGetActiveTextureUnit( state );
-
-
-    switch( cap ) {
-    case GL_VERTEX_ARRAY:
-    case GL_NORMAL_ARRAY:
-    case GL_COLOR_ARRAY:
-    case GL_INDEX_ARRAY:
-    case GL_EDGE_FLAG_ARRAY:
-    case GL_SECONDARY_COLOR_ARRAY:
-    case GL_FOG_COORD_ARRAY:
-	retval = __glXGetArrayEnable( state, cap, 0, data );
-	break;
-
-    case GL_VERTEX_ARRAY_SIZE:
-	retval = __glXGetArraySize( state, GL_VERTEX_ARRAY, 0, data );
-	break;
-    case GL_COLOR_ARRAY_SIZE:
-	retval = __glXGetArraySize( state, GL_COLOR_ARRAY, 0, data );
-	break;
-    case GL_SECONDARY_COLOR_ARRAY_SIZE:
-	retval = __glXGetArraySize( state, GL_SECONDARY_COLOR_ARRAY, 0, data );
-	break;
-
-    case GL_VERTEX_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_VERTEX_ARRAY, 0, data );
-	break;
-    case GL_NORMAL_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_NORMAL_ARRAY, 0, data );
-	break;
-    case GL_INDEX_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_INDEX_ARRAY, 0, data );
-	break;
-    case GL_COLOR_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_COLOR_ARRAY, 0, data );
-	break;
-    case GL_SECONDARY_COLOR_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_SECONDARY_COLOR_ARRAY, 0, data );
-	break;
-    case GL_FOG_COORD_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_FOG_COORD_ARRAY, 0, data );
-	break;
-
-    case GL_VERTEX_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_VERTEX_ARRAY, 0, data );
-	break;
-    case GL_NORMAL_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_NORMAL_ARRAY, 0, data );
-	break;
-    case GL_INDEX_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_INDEX_ARRAY, 0, data );
-	break;
-    case GL_EDGE_FLAG_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_EDGE_FLAG_ARRAY, 0, data );
-	break;
-    case GL_COLOR_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_COLOR_ARRAY, 0, data );
-	break;
-    case GL_SECONDARY_COLOR_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_SECONDARY_COLOR_ARRAY, 0, data );
-	break;
-    case GL_FOG_COORD_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_FOG_COORD_ARRAY, 0, data );
-	break;
-
-    case GL_TEXTURE_COORD_ARRAY:
-	retval = __glXGetArrayEnable( state, GL_TEXTURE_COORD_ARRAY, tex_unit, data );
-	break;
-    case GL_TEXTURE_COORD_ARRAY_SIZE:
-	retval = __glXGetArraySize( state, GL_TEXTURE_COORD_ARRAY, tex_unit, data );
-	break;
-    case GL_TEXTURE_COORD_ARRAY_TYPE:
-	retval = __glXGetArrayType( state, GL_TEXTURE_COORD_ARRAY, tex_unit, data );
-	break;
-    case GL_TEXTURE_COORD_ARRAY_STRIDE:
-	retval = __glXGetArrayStride( state, GL_TEXTURE_COORD_ARRAY, tex_unit, data );
-	break;
-
-    case GL_MAX_ELEMENTS_VERTICES:
-    case GL_MAX_ELEMENTS_INDICES:
-	retval = GL_TRUE;
-	*data = ~0UL;
-	break;
-
-    
-    case GL_PACK_ROW_LENGTH:
-	*data = (GLintptr)state->storePack.rowLength;
-	break;
-    case GL_PACK_IMAGE_HEIGHT:
-	*data = (GLintptr)state->storePack.imageHeight;
-	break;
-    case GL_PACK_SKIP_ROWS:
-	*data = (GLintptr)state->storePack.skipRows;
-	break;
-    case GL_PACK_SKIP_PIXELS:
-	*data = (GLintptr)state->storePack.skipPixels;
-	break;
-    case GL_PACK_SKIP_IMAGES:
-	*data = (GLintptr)state->storePack.skipImages;
-	break;
-    case GL_PACK_ALIGNMENT:
-	*data = (GLintptr)state->storePack.alignment;
-	break;
-    case GL_PACK_SWAP_BYTES:
-	*data = (GLintptr)state->storePack.swapEndian;
-	break;
-    case GL_PACK_LSB_FIRST:
-	*data = (GLintptr)state->storePack.lsbFirst;
-	break;
-    case GL_UNPACK_ROW_LENGTH:
-	*data = (GLintptr)state->storeUnpack.rowLength;
-	break;
-    case GL_UNPACK_IMAGE_HEIGHT:
-	*data = (GLintptr)state->storeUnpack.imageHeight;
-	break;
-    case GL_UNPACK_SKIP_ROWS:
-	*data = (GLintptr)state->storeUnpack.skipRows;
-	break;
-    case GL_UNPACK_SKIP_PIXELS:
-	*data = (GLintptr)state->storeUnpack.skipPixels;
-	break;
-    case GL_UNPACK_SKIP_IMAGES:
-	*data = (GLintptr)state->storeUnpack.skipImages;
-	break;
-    case GL_UNPACK_ALIGNMENT:
-	*data = (GLintptr)state->storeUnpack.alignment;
-	break;
-    case GL_UNPACK_SWAP_BYTES:
-	*data = (GLintptr)state->storeUnpack.swapEndian;
-	break;
-    case GL_UNPACK_LSB_FIRST:
-	*data = (GLintptr)state->storeUnpack.lsbFirst;
-	break;
-    case GL_CLIENT_ATTRIB_STACK_DEPTH:
-        *data = (GLintptr)(gc->attributes.stackPointer - gc->attributes.stack);
-	break;
-    case GL_MAX_CLIENT_ATTRIB_STACK_DEPTH:
-	*data = (GLintptr)__GL_CLIENT_ATTRIB_STACK_DEPTH;
-	break;
-    case GL_CLIENT_ACTIVE_TEXTURE:
-	*data = (GLintptr)(tex_unit + GL_TEXTURE0);
-	break;
-
-    default:
-	retval = GL_FALSE;
-	break;
-    }
-
-
-    return retval;
+   GLboolean retval = GL_TRUE;
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   const GLint tex_unit = __glXGetActiveTextureUnit(state);
+
+
+   switch (cap) {
+   case GL_VERTEX_ARRAY:
+   case GL_NORMAL_ARRAY:
+   case GL_COLOR_ARRAY:
+   case GL_INDEX_ARRAY:
+   case GL_EDGE_FLAG_ARRAY:
+   case GL_SECONDARY_COLOR_ARRAY:
+   case GL_FOG_COORD_ARRAY:
+      retval = __glXGetArrayEnable(state, cap, 0, data);
+      break;
+
+   case GL_VERTEX_ARRAY_SIZE:
+      retval = __glXGetArraySize(state, GL_VERTEX_ARRAY, 0, data);
+      break;
+   case GL_COLOR_ARRAY_SIZE:
+      retval = __glXGetArraySize(state, GL_COLOR_ARRAY, 0, data);
+      break;
+   case GL_SECONDARY_COLOR_ARRAY_SIZE:
+      retval = __glXGetArraySize(state, GL_SECONDARY_COLOR_ARRAY, 0, data);
+      break;
+
+   case GL_VERTEX_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_VERTEX_ARRAY, 0, data);
+      break;
+   case GL_NORMAL_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_NORMAL_ARRAY, 0, data);
+      break;
+   case GL_INDEX_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_INDEX_ARRAY, 0, data);
+      break;
+   case GL_COLOR_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_COLOR_ARRAY, 0, data);
+      break;
+   case GL_SECONDARY_COLOR_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_SECONDARY_COLOR_ARRAY, 0, data);
+      break;
+   case GL_FOG_COORD_ARRAY_TYPE:
+      retval = __glXGetArrayType(state, GL_FOG_COORD_ARRAY, 0, data);
+      break;
+
+   case GL_VERTEX_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_VERTEX_ARRAY, 0, data);
+      break;
+   case GL_NORMAL_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_NORMAL_ARRAY, 0, data);
+      break;
+   case GL_INDEX_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_INDEX_ARRAY, 0, data);
+      break;
+   case GL_EDGE_FLAG_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_EDGE_FLAG_ARRAY, 0, data);
+      break;
+   case GL_COLOR_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_COLOR_ARRAY, 0, data);
+      break;
+   case GL_SECONDARY_COLOR_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_SECONDARY_COLOR_ARRAY, 0, data);
+      break;
+   case GL_FOG_COORD_ARRAY_STRIDE:
+      retval = __glXGetArrayStride(state, GL_FOG_COORD_ARRAY, 0, data);
+      break;
+
+   case GL_TEXTURE_COORD_ARRAY:
+      retval =
+         __glXGetArrayEnable(state, GL_TEXTURE_COORD_ARRAY, tex_unit, data);
+      break;
+   case GL_TEXTURE_COORD_ARRAY_SIZE:
+      retval =
+         __glXGetArraySize(state, GL_TEXTURE_COORD_ARRAY, tex_unit, data);
+      break;
+   case GL_TEXTURE_COORD_ARRAY_TYPE:
+      retval =
+         __glXGetArrayType(state, GL_TEXTURE_COORD_ARRAY, tex_unit, data);
+      break;
+   case GL_TEXTURE_COORD_ARRAY_STRIDE:
+      retval =
+         __glXGetArrayStride(state, GL_TEXTURE_COORD_ARRAY, tex_unit, data);
+      break;
+
+   case GL_MAX_ELEMENTS_VERTICES:
+   case GL_MAX_ELEMENTS_INDICES:
+      retval = GL_TRUE;
+      *data = ~0UL;
+      break;
+
+
+   case GL_PACK_ROW_LENGTH:
+      *data = (GLintptr) state->storePack.rowLength;
+      break;
+   case GL_PACK_IMAGE_HEIGHT:
+      *data = (GLintptr) state->storePack.imageHeight;
+      break;
+   case GL_PACK_SKIP_ROWS:
+      *data = (GLintptr) state->storePack.skipRows;
+      break;
+   case GL_PACK_SKIP_PIXELS:
+      *data = (GLintptr) state->storePack.skipPixels;
+      break;
+   case GL_PACK_SKIP_IMAGES:
+      *data = (GLintptr) state->storePack.skipImages;
+      break;
+   case GL_PACK_ALIGNMENT:
+      *data = (GLintptr) state->storePack.alignment;
+      break;
+   case GL_PACK_SWAP_BYTES:
+      *data = (GLintptr) state->storePack.swapEndian;
+      break;
+   case GL_PACK_LSB_FIRST:
+      *data = (GLintptr) state->storePack.lsbFirst;
+      break;
+   case GL_UNPACK_ROW_LENGTH:
+      *data = (GLintptr) state->storeUnpack.rowLength;
+      break;
+   case GL_UNPACK_IMAGE_HEIGHT:
+      *data = (GLintptr) state->storeUnpack.imageHeight;
+      break;
+   case GL_UNPACK_SKIP_ROWS:
+      *data = (GLintptr) state->storeUnpack.skipRows;
+      break;
+   case GL_UNPACK_SKIP_PIXELS:
+      *data = (GLintptr) state->storeUnpack.skipPixels;
+      break;
+   case GL_UNPACK_SKIP_IMAGES:
+      *data = (GLintptr) state->storeUnpack.skipImages;
+      break;
+   case GL_UNPACK_ALIGNMENT:
+      *data = (GLintptr) state->storeUnpack.alignment;
+      break;
+   case GL_UNPACK_SWAP_BYTES:
+      *data = (GLintptr) state->storeUnpack.swapEndian;
+      break;
+   case GL_UNPACK_LSB_FIRST:
+      *data = (GLintptr) state->storeUnpack.lsbFirst;
+      break;
+   case GL_CLIENT_ATTRIB_STACK_DEPTH:
+      *data = (GLintptr) (gc->attributes.stackPointer - gc->attributes.stack);
+      break;
+   case GL_MAX_CLIENT_ATTRIB_STACK_DEPTH:
+      *data = (GLintptr) __GL_CLIENT_ATTRIB_STACK_DEPTH;
+      break;
+   case GL_CLIENT_ACTIVE_TEXTURE:
+      *data = (GLintptr) (tex_unit + GL_TEXTURE0);
+      break;
+
+   default:
+      retval = GL_FALSE;
+      break;
+   }
+
+
+   return retval;
 }
 
 
-void __indirect_glGetBooleanv(GLenum val, GLboolean *b)
+void
+__indirect_glGetBooleanv(GLenum val, GLboolean * b)
 {
-    const GLenum origVal = val;
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXSingleReply reply;
-
-    val = RemapTransposeEnum( val );
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_GetBooleanv,4);
-    __GLX_SINGLE_PUT_LONG(0,val);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_SIZE(compsize);
-
-    if (compsize == 0) {
-	/*
-	** Error occured; don't modify user's buffer.
-	*/
-    } else {
-	GLintptr data;
-
-	/*
-	** We still needed to send the request to the server in order to
-	** find out whether it was legal to make a query (it's illegal,
-	** for example, to call a query between glBegin() and glEnd()).
-	*/
-
-	if ( get_client_data( gc, val, & data ) ) {
-	    *b = (GLboolean) data;
-	}
-	else {
-	    /*
-	     ** Not a local value, so use what we got from the server.
-	     */
-	    if (compsize == 1) {
-		__GLX_SINGLE_GET_CHAR(b);
-	    } else {
-		__GLX_SINGLE_GET_CHAR_ARRAY(b,compsize);
-		if (val != origVal) {
-		    /* matrix transpose */
-		    TransposeMatrixb(b);
-                }
-	    }
-	}
-    }
-    __GLX_SINGLE_END();
+   const GLenum origVal = val;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXSingleReply reply;
+
+   val = RemapTransposeEnum(val);
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetBooleanv, 4);
+   __GLX_SINGLE_PUT_LONG(0, val);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_SIZE(compsize);
+
+   if (compsize == 0) {
+      /*
+       ** Error occured; don't modify user's buffer.
+       */
+   }
+   else {
+      GLintptr data;
+
+      /*
+       ** We still needed to send the request to the server in order to
+       ** find out whether it was legal to make a query (it's illegal,
+       ** for example, to call a query between glBegin() and glEnd()).
+       */
+
+      if (get_client_data(gc, val, &data)) {
+         *b = (GLboolean) data;
+      }
+      else {
+         /*
+          ** Not a local value, so use what we got from the server.
+          */
+         if (compsize == 1) {
+            __GLX_SINGLE_GET_CHAR(b);
+         }
+         else {
+            __GLX_SINGLE_GET_CHAR_ARRAY(b, compsize);
+            if (val != origVal) {
+               /* matrix transpose */
+               TransposeMatrixb(b);
+            }
+         }
+      }
+   }
+   __GLX_SINGLE_END();
 }
 
-void __indirect_glGetDoublev(GLenum val, GLdouble *d)
+void
+__indirect_glGetDoublev(GLenum val, GLdouble * d)
 {
-    const GLenum origVal = val;
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXSingleReply reply;
-
-    val = RemapTransposeEnum( val );
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_GetDoublev,4);
-    __GLX_SINGLE_PUT_LONG(0,val);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_SIZE(compsize);
-
-    if (compsize == 0) {
-	/*
-	** Error occured; don't modify user's buffer.
-	*/
-    } else {
-	GLintptr data;
-
-	/*
-	** We still needed to send the request to the server in order to
-	** find out whether it was legal to make a query (it's illegal,
-	** for example, to call a query between glBegin() and glEnd()).
-	*/
-
-	if ( get_client_data( gc, val, & data ) ) {
-	    *d = (GLdouble) data;
-	}
-	else {
-	    /*
-	     ** Not a local value, so use what we got from the server.
-	     */
-	    if (compsize == 1) {
-		__GLX_SINGLE_GET_DOUBLE(d);
-	    } else {
-		__GLX_SINGLE_GET_DOUBLE_ARRAY(d,compsize);
-		if (val != origVal) {
-		    /* matrix transpose */
-		    TransposeMatrixd(d);
-		}
-	    }
-	}
-    }
-    __GLX_SINGLE_END();
+   const GLenum origVal = val;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXSingleReply reply;
+
+   val = RemapTransposeEnum(val);
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetDoublev, 4);
+   __GLX_SINGLE_PUT_LONG(0, val);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_SIZE(compsize);
+
+   if (compsize == 0) {
+      /*
+       ** Error occured; don't modify user's buffer.
+       */
+   }
+   else {
+      GLintptr data;
+
+      /*
+       ** We still needed to send the request to the server in order to
+       ** find out whether it was legal to make a query (it's illegal,
+       ** for example, to call a query between glBegin() and glEnd()).
+       */
+
+      if (get_client_data(gc, val, &data)) {
+         *d = (GLdouble) data;
+      }
+      else {
+         /*
+          ** Not a local value, so use what we got from the server.
+          */
+         if (compsize == 1) {
+            __GLX_SINGLE_GET_DOUBLE(d);
+         }
+         else {
+            __GLX_SINGLE_GET_DOUBLE_ARRAY(d, compsize);
+            if (val != origVal) {
+               /* matrix transpose */
+               TransposeMatrixd(d);
+            }
+         }
+      }
+   }
+   __GLX_SINGLE_END();
 }
 
-void __indirect_glGetFloatv(GLenum val, GLfloat *f)
+void
+__indirect_glGetFloatv(GLenum val, GLfloat * f)
 {
-    const GLenum origVal = val;
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXSingleReply reply;
-
-    val = RemapTransposeEnum( val );
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_GetFloatv,4);
-    __GLX_SINGLE_PUT_LONG(0,val);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_SIZE(compsize);
-
-    if (compsize == 0) {
-	/*
-	** Error occured; don't modify user's buffer.
-	*/
-    } else {
-	GLintptr data;
-
-	/*
-	** We still needed to send the request to the server in order to
-	** find out whether it was legal to make a query (it's illegal,
-	** for example, to call a query between glBegin() and glEnd()).
-	*/
-
-	if ( get_client_data( gc, val, & data ) ) {
-	    *f = (GLfloat) data;
-	}
-	else {
-	    /*
-	     ** Not a local value, so use what we got from the server.
-	     */
-	    if (compsize == 1) {
-		__GLX_SINGLE_GET_FLOAT(f);
-	    } else {
-		__GLX_SINGLE_GET_FLOAT_ARRAY(f,compsize);
-		if (val != origVal) {
-		    /* matrix transpose */
-		    TransposeMatrixf(f);
-                }
-	    }
-	}
-    }
-    __GLX_SINGLE_END();
+   const GLenum origVal = val;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXSingleReply reply;
+
+   val = RemapTransposeEnum(val);
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetFloatv, 4);
+   __GLX_SINGLE_PUT_LONG(0, val);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_SIZE(compsize);
+
+   if (compsize == 0) {
+      /*
+       ** Error occured; don't modify user's buffer.
+       */
+   }
+   else {
+      GLintptr data;
+
+      /*
+       ** We still needed to send the request to the server in order to
+       ** find out whether it was legal to make a query (it's illegal,
+       ** for example, to call a query between glBegin() and glEnd()).
+       */
+
+      if (get_client_data(gc, val, &data)) {
+         *f = (GLfloat) data;
+      }
+      else {
+         /*
+          ** Not a local value, so use what we got from the server.
+          */
+         if (compsize == 1) {
+            __GLX_SINGLE_GET_FLOAT(f);
+         }
+         else {
+            __GLX_SINGLE_GET_FLOAT_ARRAY(f, compsize);
+            if (val != origVal) {
+               /* matrix transpose */
+               TransposeMatrixf(f);
+            }
+         }
+      }
+   }
+   __GLX_SINGLE_END();
 }
 
-void __indirect_glGetIntegerv(GLenum val, GLint *i)
+void
+__indirect_glGetIntegerv(GLenum val, GLint * i)
 {
-    const GLenum origVal = val;
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXSingleReply reply;
-
-    val = RemapTransposeEnum( val );
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_GetIntegerv,4);
-    __GLX_SINGLE_PUT_LONG(0,val);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_SIZE(compsize);
-
-    if (compsize == 0) {
-	/*
-	** Error occured; don't modify user's buffer.
-	*/
-    } else {
-	GLintptr data;
-
-	/*
-	** We still needed to send the request to the server in order to
-	** find out whether it was legal to make a query (it's illegal,
-	** for example, to call a query between glBegin() and glEnd()).
-	*/
-
-	if ( get_client_data( gc, val, & data ) ) {
-	    *i = (GLint) data;
-	}
-	else {
-	    /*
-	     ** Not a local value, so use what we got from the server.
-	     */
-	    if (compsize == 1) {
-		__GLX_SINGLE_GET_LONG(i);
-	    } else {
-		__GLX_SINGLE_GET_LONG_ARRAY(i,compsize);
-		if (val != origVal) {
-		    /* matrix transpose */
-		    TransposeMatrixi(i);
-                }
-	    }
-	}
-    }
-    __GLX_SINGLE_END();
+   const GLenum origVal = val;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXSingleReply reply;
+
+   val = RemapTransposeEnum(val);
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_GetIntegerv, 4);
+   __GLX_SINGLE_PUT_LONG(0, val);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_SIZE(compsize);
+
+   if (compsize == 0) {
+      /*
+       ** Error occured; don't modify user's buffer.
+       */
+   }
+   else {
+      GLintptr data;
+
+      /*
+       ** We still needed to send the request to the server in order to
+       ** find out whether it was legal to make a query (it's illegal,
+       ** for example, to call a query between glBegin() and glEnd()).
+       */
+
+      if (get_client_data(gc, val, &data)) {
+         *i = (GLint) data;
+      }
+      else {
+         /*
+          ** Not a local value, so use what we got from the server.
+          */
+         if (compsize == 1) {
+            __GLX_SINGLE_GET_LONG(i);
+         }
+         else {
+            __GLX_SINGLE_GET_LONG_ARRAY(i, compsize);
+            if (val != origVal) {
+               /* matrix transpose */
+               TransposeMatrixi(i);
+            }
+         }
+      }
+   }
+   __GLX_SINGLE_END();
 }
 
 /*
 ** Send all pending commands to server.
 */
-void __indirect_glFlush(void)
+void
+__indirect_glFlush(void)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
+   __GLX_SINGLE_DECLARE_VARIABLES();
 
-    if (!dpy) return;
+   if (!dpy)
+      return;
 
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_Flush,0);
-    __GLX_SINGLE_END();
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_Flush, 0);
+   __GLX_SINGLE_END();
 
-    /* And finally flush the X protocol data */
-    XFlush(dpy);
+   /* And finally flush the X protocol data */
+   XFlush(dpy);
 }
 
-void __indirect_glFeedbackBuffer(GLsizei size, GLenum type, GLfloat *buffer)
+void
+__indirect_glFeedbackBuffer(GLsizei size, GLenum type, GLfloat * buffer)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
+   __GLX_SINGLE_DECLARE_VARIABLES();
 
-    if (!dpy) return;
+   if (!dpy)
+      return;
 
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_FeedbackBuffer,8);
-    __GLX_SINGLE_PUT_LONG(0,size);
-    __GLX_SINGLE_PUT_LONG(4,type);
-    __GLX_SINGLE_END();
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_FeedbackBuffer, 8);
+   __GLX_SINGLE_PUT_LONG(0, size);
+   __GLX_SINGLE_PUT_LONG(4, type);
+   __GLX_SINGLE_END();
 
-    gc->feedbackBuf = buffer;
+   gc->feedbackBuf = buffer;
 }
 
-void __indirect_glSelectBuffer(GLsizei numnames, GLuint *buffer)
+void
+__indirect_glSelectBuffer(GLsizei numnames, GLuint * buffer)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
+   __GLX_SINGLE_DECLARE_VARIABLES();
 
-    if (!dpy) return;
+   if (!dpy)
+      return;
 
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_SelectBuffer,4);
-    __GLX_SINGLE_PUT_LONG(0,numnames);
-    __GLX_SINGLE_END();
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_SelectBuffer, 4);
+   __GLX_SINGLE_PUT_LONG(0, numnames);
+   __GLX_SINGLE_END();
 
-    gc->selectBuf = buffer;
+   gc->selectBuf = buffer;
 }
 
-GLint __indirect_glRenderMode(GLenum mode)
+GLint
+__indirect_glRenderMode(GLenum mode)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    GLint retval = 0;
-    xGLXRenderModeReply reply;
-
-    if (!dpy) return -1;
-
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_RenderMode,4);
-    __GLX_SINGLE_PUT_LONG(0,mode);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_RETVAL(retval,GLint);
-
-    if (reply.newMode != mode) {
-	/*
-	** Switch to new mode did not take effect, therefore an error
-	** occured.  When an error happens the server won't send us any
-	** other data.
-	*/
-    } else {
-	/* Read the feedback or selection data */
-	if (gc->renderMode == GL_FEEDBACK) {
-	    __GLX_SINGLE_GET_SIZE(compsize);
-	    __GLX_SINGLE_GET_FLOAT_ARRAY(gc->feedbackBuf, compsize);
-	} else
-	if (gc->renderMode == GL_SELECT) {
-	    __GLX_SINGLE_GET_SIZE(compsize);
-	    __GLX_SINGLE_GET_LONG_ARRAY(gc->selectBuf, compsize);
-	}
-	gc->renderMode = mode;
-    }
-    __GLX_SINGLE_END();
-
-    return retval;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   GLint retval = 0;
+   xGLXRenderModeReply reply;
+
+   if (!dpy)
+      return -1;
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_RenderMode, 4);
+   __GLX_SINGLE_PUT_LONG(0, mode);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_RETVAL(retval, GLint);
+
+   if (reply.newMode != mode) {
+      /*
+       ** Switch to new mode did not take effect, therefore an error
+       ** occured.  When an error happens the server won't send us any
+       ** other data.
+       */
+   }
+   else {
+      /* Read the feedback or selection data */
+      if (gc->renderMode == GL_FEEDBACK) {
+         __GLX_SINGLE_GET_SIZE(compsize);
+         __GLX_SINGLE_GET_FLOAT_ARRAY(gc->feedbackBuf, compsize);
+      }
+      else if (gc->renderMode == GL_SELECT) {
+         __GLX_SINGLE_GET_SIZE(compsize);
+         __GLX_SINGLE_GET_LONG_ARRAY(gc->selectBuf, compsize);
+      }
+      gc->renderMode = mode;
+   }
+   __GLX_SINGLE_END();
+
+   return retval;
 }
 
-void __indirect_glFinish(void)
+void
+__indirect_glFinish(void)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    xGLXSingleReply reply;
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   xGLXSingleReply reply;
 
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_Finish,0);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_END();
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_Finish, 0);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_END();
 }
 
 
@@ -592,241 +626,346 @@ void __indirect_glFinish(void)
  * Extract the major and minor version numbers from a version string.
  */
 static void
-version_from_string( const char * ver, 
-		     int * major_version, int * minor_version )
+version_from_string(const char *ver, int *major_version, int *minor_version)
 {
-    const char * end;
-    long major;
-    long minor;
-
-    major = strtol( ver, (char **) & end, 10 );
-    minor = strtol( end + 1, NULL, 10 );
-    *major_version = major;
-    *minor_version = minor;
+   const char *end;
+   long major;
+   long minor;
+
+   major = strtol(ver, (char **) &end, 10);
+   minor = strtol(end + 1, NULL, 10);
+   *major_version = major;
+   *minor_version = minor;
 }
 
 
-const GLubyte *__indirect_glGetString(GLenum name)
+const GLubyte *
+__indirect_glGetString(GLenum name)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    Display *dpy = gc->currentDpy;
-    GLubyte *s = NULL;
+   __GLXcontext *gc = __glXGetCurrentContext();
+   Display *dpy = gc->currentDpy;
+   GLubyte *s = NULL;
 
-    if (!dpy) return 0;
+   if (!dpy)
+      return 0;
 
-    /*
+   /*
     ** Return the cached copy if the string has already been fetched
     */
-    switch(name) {
-      case GL_VENDOR:
-	if (gc->vendor) return gc->vendor;
-	break;
-      case GL_RENDERER:
-	if (gc->renderer) return gc->renderer;
-	break;
-      case GL_VERSION:
-	if (gc->version) return gc->version;
-	break;
-      case GL_EXTENSIONS:
-	if (gc->extensions) return gc->extensions;
-	break;
-      default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return 0;
-    }
-
-    /*
+   switch (name) {
+   case GL_VENDOR:
+      if (gc->vendor)
+         return gc->vendor;
+      break;
+   case GL_RENDERER:
+      if (gc->renderer)
+         return gc->renderer;
+      break;
+   case GL_VERSION:
+      if (gc->version)
+         return gc->version;
+      break;
+   case GL_EXTENSIONS:
+      if (gc->extensions)
+         return gc->extensions;
+      break;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return 0;
+   }
+
+   /*
     ** Get requested string from server
     */
 
-    (void) __glXFlushRenderBuffer( gc, gc->pc );
-    s = (GLubyte *) __glXGetStringFromServer( dpy, gc->majorOpcode,
-				  X_GLsop_GetString, gc->currentContextTag,
-				  name );
-    if (!s) {
-	/* Throw data on the floor */
-	__glXSetError(gc, GL_OUT_OF_MEMORY);
-    } else {
-	/*
-	** Update local cache
-	*/
-	switch(name) {
-	case GL_VENDOR:
-	    gc->vendor = s;
-	    break;
-
-	case GL_RENDERER:
-	    gc->renderer = s;
-	    break;
-
-	case GL_VERSION: {
-	    int client_major;
-	    int client_minor;
-
-	    version_from_string( (char *) s, 
-				 & gc->server_major, & gc->server_minor );
-	    __glXGetGLVersion( & client_major, & client_minor );
-
-	    if ( (gc->server_major < client_major)
-		 || ((gc->server_major == client_major) 
-		     && (gc->server_minor <= client_minor)) ) {
-		gc->version = s;
-	    }
-	    else {
-		/* Allow 7 bytes for the client-side GL version.  This allows
-		 * for upto version 999.999.  I'm not holding my breath for
-		 * that one!  The extra 4 is for the ' ()\0' that will be
-		 * added.
-		 */
-		const size_t size = 7 + strlen( (char *) s ) + 4;
-
-		gc->version = Xmalloc( size );
-		if ( gc->version == NULL ) {
-		    /* If we couldn't allocate memory for the new string,
-		     * make a best-effort and just copy the client-side version
-		     * to the string and use that.  It probably doesn't
-		     * matter what is done here.  If there not memory available
-		     * for a short string, the system is probably going to die
-		     * soon anyway.
-		     */
-		    snprintf( (char *) s, strlen( (char *) s ) + 1, "%u.%u",
-			      client_major, client_minor );
-		    gc->version = s;
-		}
-		else {
-		    snprintf( (char *)gc->version, size, "%u.%u (%s)",
-			      client_major, client_minor, s );
-		    Xfree( s );
-		    s = gc->version;
-		}
-	    }
-	    break;
-	}
-
-	case GL_EXTENSIONS: {
-	    int major = 1;
-	    int minor = 0;
-
-	    /* This code is currently disabled.  I was reminded that some
-	     * vendors intentionally exclude some extensions from their
-	     * extension string that are part of the core version they
-	     * advertise.  In particular, on Nvidia drivers this means that
-	     * the functionality is supported by the driver, but is not
-	     * hardware accelerated.  For example, a TNT will show core
-	     * version 1.5, but most of the post-1.2 functionality is a
-	     * software fallback.
-	     * 
-	     * I don't want to break applications that rely on this odd
-	     * behavior.  At the same time, the code is written and tested,
-	     * so I didn't want to throw it away.  Therefore, the code is here
-	     * but disabled.  In the future, we may wish to and an environment
-	     * variable to enable it.
-	     */
-	    
+   (void) __glXFlushRenderBuffer(gc, gc->pc);
+   s = (GLubyte *) __glXGetString(dpy, gc->majorOpcode, gc->currentContextTag, name);
+   if (!s) {
+      /* Throw data on the floor */
+      __glXSetError(gc, GL_OUT_OF_MEMORY);
+   }
+   else {
+      /*
+       ** Update local cache
+       */
+      switch (name) {
+      case GL_VENDOR:
+         gc->vendor = s;
+         break;
+
+      case GL_RENDERER:
+         gc->renderer = s;
+         break;
+
+      case GL_VERSION:{
+            int client_major;
+            int client_minor;
+
+            version_from_string((char *) s,
+                                &gc->server_major, &gc->server_minor);
+            __glXGetGLVersion(&client_major, &client_minor);
+
+            if ((gc->server_major < client_major)
+                || ((gc->server_major == client_major)
+                    && (gc->server_minor <= client_minor))) {
+               gc->version = s;
+            }
+            else {
+               /* Allow 7 bytes for the client-side GL version.  This allows
+                * for upto version 999.999.  I'm not holding my breath for
+                * that one!  The extra 4 is for the ' ()\0' that will be
+                * added.
+                */
+               const size_t size = 7 + strlen((char *) s) + 4;
+
+               gc->version = Xmalloc(size);
+               if (gc->version == NULL) {
+                  /* If we couldn't allocate memory for the new string,
+                   * make a best-effort and just copy the client-side version
+                   * to the string and use that.  It probably doesn't
+                   * matter what is done here.  If there not memory available
+                   * for a short string, the system is probably going to die
+                   * soon anyway.
+                   */
+                  snprintf((char *) s, strlen((char *) s) + 1, "%u.%u",
+                           client_major, client_minor);
+                  gc->version = s;
+               }
+               else {
+                  snprintf((char *) gc->version, size, "%u.%u (%s)",
+                           client_major, client_minor, s);
+                  Xfree(s);
+                  s = gc->version;
+               }
+            }
+            break;
+         }
+
+      case GL_EXTENSIONS:{
+            int major = 1;
+            int minor = 0;
+
+            /* This code is currently disabled.  I was reminded that some
+             * vendors intentionally exclude some extensions from their
+             * extension string that are part of the core version they
+             * advertise.  In particular, on Nvidia drivers this means that
+             * the functionality is supported by the driver, but is not
+             * hardware accelerated.  For example, a TNT will show core
+             * version 1.5, but most of the post-1.2 functionality is a
+             * software fallback.
+             * 
+             * I don't want to break applications that rely on this odd
+             * behavior.  At the same time, the code is written and tested,
+             * so I didn't want to throw it away.  Therefore, the code is here
+             * but disabled.  In the future, we may wish to and an environment
+             * variable to enable it.
+             */
+
 #if 0
-	    /* Call glGetString just to make sure that gc->server_major and
-	     * gc->server_minor are set.  This version may be higher than we
-	     * can completely support, but it may imply support for some
-	     * extensions that we can support.
-	     * 
-	     * For example, at the time of this writing, the client-side
-	     * library only supports upto core GL version 1.2.  However, cubic
-	     * textures, multitexture, multisampling, and some other 1.3
-	     * features are supported.  If the server reports back version
-	     * 1.3, but does not report all of those extensions, we will
-	     * enable them.
-	     */
-	    (void *) glGetString( GL_VERSION );
-	    major = gc->server_major,
-	    minor = gc->server_minor;
+            /* Call glGetString just to make sure that gc->server_major and
+             * gc->server_minor are set.  This version may be higher than we
+             * can completely support, but it may imply support for some
+             * extensions that we can support.
+             * 
+             * For example, at the time of this writing, the client-side
+             * library only supports upto core GL version 1.2.  However, cubic
+             * textures, multitexture, multisampling, and some other 1.3
+             * features are supported.  If the server reports back version
+             * 1.3, but does not report all of those extensions, we will
+             * enable them.
+             */
+            (void *) glGetString(GL_VERSION);
+            major = gc->server_major, minor = gc->server_minor;
 #endif
 
-	    __glXCalculateUsableGLExtensions( gc, (char *) s, major, minor );
-	    XFree( s );
-	    s = gc->extensions;
-	    break;
-	}
-	}
-    }
-    return s;
+            __glXCalculateUsableGLExtensions(gc, (char *) s, major, minor);
+            XFree(s);
+            s = gc->extensions;
+            break;
+         }
+      }
+   }
+   return s;
 }
 
-GLboolean __indirect_glIsEnabled(GLenum cap)
+GLboolean
+__indirect_glIsEnabled(GLenum cap)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    xGLXSingleReply reply;
-    GLboolean retval = 0;
-    GLintptr enable;
-
-    if (!dpy) return 0;
-
-    switch(cap) {
-    case GL_VERTEX_ARRAY:
-    case GL_NORMAL_ARRAY:
-    case GL_COLOR_ARRAY:
-    case GL_INDEX_ARRAY:
-    case GL_EDGE_FLAG_ARRAY:
-    case GL_SECONDARY_COLOR_ARRAY:
-    case GL_FOG_COORD_ARRAY:
-	retval = __glXGetArrayEnable( state, cap, 0, & enable );
-	assert( retval );
-	return (GLboolean) enable;
-	break;
-    case GL_TEXTURE_COORD_ARRAY:
-	retval = __glXGetArrayEnable( state, GL_TEXTURE_COORD_ARRAY,
-				      __glXGetActiveTextureUnit( state ), & enable );
-	assert( retval );
-	return (GLboolean) enable;
-	break;
-    }
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   xGLXSingleReply reply;
+   GLboolean retval = 0;
+   GLintptr enable;
+
+   if (!dpy)
+      return 0;
+
+   switch (cap) {
+   case GL_VERTEX_ARRAY:
+   case GL_NORMAL_ARRAY:
+   case GL_COLOR_ARRAY:
+   case GL_INDEX_ARRAY:
+   case GL_EDGE_FLAG_ARRAY:
+   case GL_SECONDARY_COLOR_ARRAY:
+   case GL_FOG_COORD_ARRAY:
+      retval = __glXGetArrayEnable(state, cap, 0, &enable);
+      assert(retval);
+      return (GLboolean) enable;
+      break;
+   case GL_TEXTURE_COORD_ARRAY:
+      retval = __glXGetArrayEnable(state, GL_TEXTURE_COORD_ARRAY,
+                                   __glXGetActiveTextureUnit(state), &enable);
+      assert(retval);
+      return (GLboolean) enable;
+      break;
+   }
+
+   __GLX_SINGLE_LOAD_VARIABLES();
+   __GLX_SINGLE_BEGIN(X_GLsop_IsEnabled, 4);
+   __GLX_SINGLE_PUT_LONG(0, cap);
+   __GLX_SINGLE_READ_XREPLY();
+   __GLX_SINGLE_GET_RETVAL(retval, GLboolean);
+   __GLX_SINGLE_END();
+   return retval;
+}
 
-    __GLX_SINGLE_LOAD_VARIABLES();
-    __GLX_SINGLE_BEGIN(X_GLsop_IsEnabled,4);
-    __GLX_SINGLE_PUT_LONG(0,cap);
-    __GLX_SINGLE_READ_XREPLY();
-    __GLX_SINGLE_GET_RETVAL(retval, GLboolean);
-    __GLX_SINGLE_END();
-    return retval;
+void
+__indirect_glGetPointerv(GLenum pname, void **params)
+{
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
+   Display *dpy = gc->currentDpy;
+
+   if (!dpy)
+      return;
+
+   switch (pname) {
+   case GL_VERTEX_ARRAY_POINTER:
+   case GL_NORMAL_ARRAY_POINTER:
+   case GL_COLOR_ARRAY_POINTER:
+   case GL_INDEX_ARRAY_POINTER:
+   case GL_EDGE_FLAG_ARRAY_POINTER:
+      __glXGetArrayPointer(state, pname - GL_VERTEX_ARRAY_POINTER
+                           + GL_VERTEX_ARRAY, 0, params);
+      return;
+   case GL_TEXTURE_COORD_ARRAY_POINTER:
+      __glXGetArrayPointer(state, GL_TEXTURE_COORD_ARRAY,
+                           __glXGetActiveTextureUnit(state), params);
+      return;
+   case GL_SECONDARY_COLOR_ARRAY_POINTER:
+   case GL_FOG_COORD_ARRAY_POINTER:
+      __glXGetArrayPointer(state, pname - GL_FOG_COORD_ARRAY_POINTER
+                           + GL_FOG_COORD_ARRAY, 0, params);
+      return;
+   case GL_FEEDBACK_BUFFER_POINTER:
+      *params = (void *) gc->feedbackBuf;
+      return;
+   case GL_SELECTION_BUFFER_POINTER:
+      *params = (void *) gc->selectBuf;
+      return;
+   default:
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
 }
 
-void __indirect_glGetPointerv(GLenum pname, void **params)
+
+
+/**
+ * This was previously auto-generated, but we need to special-case
+ * how we handle writing into the 'residences' buffer when n%4!=0.
+ */
+#define X_GLsop_AreTexturesResident 143
+GLboolean
+__indirect_glAreTexturesResident(GLsizei n, const GLuint * textures,
+                                 GLboolean * residences)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
-    Display *dpy = gc->currentDpy;
-
-    if (!dpy) return;
-
-    switch(pname) {
-    case GL_VERTEX_ARRAY_POINTER:
-    case GL_NORMAL_ARRAY_POINTER:
-    case GL_COLOR_ARRAY_POINTER:
-    case GL_INDEX_ARRAY_POINTER:
-    case GL_EDGE_FLAG_ARRAY_POINTER:
-	__glXGetArrayPointer( state, pname - GL_VERTEX_ARRAY_POINTER 
-			      + GL_VERTEX_ARRAY,
-			      0, params );
-	return;
-    case GL_TEXTURE_COORD_ARRAY_POINTER:
-	__glXGetArrayPointer( state, GL_TEXTURE_COORD_ARRAY, 
-			      __glXGetActiveTextureUnit( state ), params );
-	return;
-    case GL_SECONDARY_COLOR_ARRAY_POINTER:
-    case GL_FOG_COORD_ARRAY_POINTER:
-	__glXGetArrayPointer( state, pname - GL_FOG_COORD_ARRAY_POINTER
-			      + GL_FOG_COORD_ARRAY, 
-			      0, params );
-	return;
-    case GL_FEEDBACK_BUFFER_POINTER:
-	*params = (void *)gc->feedbackBuf;
-	return;
-    case GL_SELECTION_BUFFER_POINTER:
-	*params = (void *)gc->selectBuf;
-	return;
-    default:
-	__glXSetError(gc, GL_INVALID_ENUM);
-	return;
+    __GLXcontext *const gc = __glXGetCurrentContext();
+    Display *const dpy = gc->currentDpy;
+    GLboolean retval = (GLboolean) 0;
+    const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+    if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
+#ifdef USE_XCB
+        xcb_connection_t *c = XGetXCBConnection(dpy);
+        (void) __glXFlushRenderBuffer(gc, gc->pc);
+        xcb_glx_are_textures_resident_reply_t *reply =
+            xcb_glx_are_textures_resident_reply(c,
+                                                xcb_glx_are_textures_resident
+                                                (c, gc->currentContextTag, n,
+                                                 textures), NULL);
+        (void) memcpy(residences, xcb_glx_are_textures_resident_data(reply),
+                      xcb_glx_are_textures_resident_data_length(reply) *
+                      sizeof(GLboolean));
+        retval = reply->ret_val;
+        free(reply);
+#else
+        GLubyte const *pc =
+            __glXSetupSingleRequest(gc, X_GLsop_AreTexturesResident, cmdlen);
+        (void) memcpy((void *) (pc + 0), (void *) (&n), 4);
+        (void) memcpy((void *) (pc + 4), (void *) (textures), (n * 4));
+        if (n & 3) {
+           /* n is not a multiple of four.
+            * When reply_is_always_array is TRUE, __glXReadReply() will
+            * put a multiple of four bytes into the dest buffer.  If the
+            * caller's buffer is not a multiple of four in size, we'll write
+            * out of bounds.  So use a temporary buffer that's a few bytes
+            * larger.
+            */
+           GLboolean *res4 = malloc((n + 3) & ~3);
+           retval = (GLboolean) __glXReadReply(dpy, 1, res4, GL_TRUE);
+           memcpy(residences, res4, n);
+           free(res4);
+        }
+        else {
+           retval = (GLboolean) __glXReadReply(dpy, 1, residences, GL_TRUE);
+        }
+        UnlockDisplay(dpy);
+        SyncHandle();
+#endif /* USE_XCB */
     }
+    return retval;
 }
 
+
+/**
+ * This was previously auto-generated, but we need to special-case
+ * how we handle writing into the 'residences' buffer when n%4!=0.
+ */
+#define X_GLvop_AreTexturesResidentEXT 11
+GLboolean
+glAreTexturesResidentEXT(GLsizei n, const GLuint * textures,
+                         GLboolean * residences)
+{
+    __GLXcontext *const gc = __glXGetCurrentContext();
+
+    if (gc->isDirect) {
+        return CALL_AreTexturesResident(GET_DISPATCH(),
+                                        (n, textures, residences));
+    } else {
+        __GLXcontext *const gc = __glXGetCurrentContext();
+        Display *const dpy = gc->currentDpy;
+        GLboolean retval = (GLboolean) 0;
+        const GLuint cmdlen = 4 + __GLX_PAD((n * 4));
+        if (__builtin_expect((n >= 0) && (dpy != NULL), 1)) {
+            GLubyte const *pc =
+                __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
+                                        X_GLvop_AreTexturesResidentEXT,
+                                        cmdlen);
+            (void) memcpy((void *) (pc + 0), (void *) (&n), 4);
+            (void) memcpy((void *) (pc + 4), (void *) (textures), (n * 4));
+            if (n & 3) {
+               /* see comments in __indirect_glAreTexturesResident() */
+               GLboolean *res4 = malloc((n + 3) & ~3);
+               retval = (GLboolean) __glXReadReply(dpy, 1, res4, GL_TRUE);
+               memcpy(residences, res4, n);
+               free(res4);
+            }
+            else {
+               retval = (GLboolean) __glXReadReply(dpy, 1, residences, GL_TRUE);
+            }
+            UnlockDisplay(dpy);
+            SyncHandle();
+        }
+        return retval;
+    }
+}
diff --git a/src/glx/x11/singlepix.c b/src/glx/x11/singlepix.c
index cd88684f709..fa12ac3bbcb 100644
--- a/src/glx/x11/singlepix.c
+++ b/src/glx/x11/singlepix.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "packsingle.h"
 #include "indirect.h"
@@ -41,151 +36,160 @@
 #include "glapioffsets.h"
 #include <GL/glxproto.h>
 
-void __indirect_glGetSeparableFilter(GLenum target, GLenum format, GLenum type,
-			  GLvoid *row, GLvoid *column, GLvoid *span)
+void
+__indirect_glGetSeparableFilter(GLenum target, GLenum format, GLenum type,
+                                GLvoid * row, GLvoid * column, GLvoid * span)
 {
-    __GLX_SINGLE_DECLARE_VARIABLES();
-    const __GLXattribute * state;
-    xGLXGetSeparableFilterReply reply;
-    GLubyte *rowBuf, *colBuf;
-
-    if (!dpy) return;
-    __GLX_SINGLE_LOAD_VARIABLES();
-    state = gc->client_state_private;
-
-    /* Send request */
-    __GLX_SINGLE_BEGIN(X_GLsop_GetSeparableFilter, __GLX_PAD(13));
-    __GLX_SINGLE_PUT_LONG(0,target);
-    __GLX_SINGLE_PUT_LONG(4,format);
-    __GLX_SINGLE_PUT_LONG(8,type);
-    __GLX_SINGLE_PUT_CHAR(12,state->storePack.swapEndian);
-    __GLX_SINGLE_READ_XREPLY();
-    compsize = reply.length << 2;
-
-    if (compsize != 0) {
-	GLint width, height;
-	GLint widthsize, heightsize;
-
-	width = reply.width;
-	height = reply.height;
-
-	widthsize = __glImageSize(width,1,1,format, type, 0);
-	heightsize = __glImageSize(height,1,1,format, type, 0);
-
-	/* Allocate a holding buffer to transform the data from */
-	rowBuf = (GLubyte*) Xmalloc(widthsize);
-	if (!rowBuf) {
-	    /* Throw data away */
-	    _XEatData(dpy, compsize);
-	    __glXSetError(gc, GL_OUT_OF_MEMORY);
-	    UnlockDisplay(dpy);
-	    SyncHandle();
-	    return;
-	} else {
-	    __GLX_SINGLE_GET_CHAR_ARRAY(((char*)rowBuf),widthsize);
-	    __glEmptyImage(gc, 1, width, 1, 1, format, type, rowBuf, row);
-	    Xfree((char*) rowBuf);
-	}
-	colBuf = (GLubyte*) Xmalloc(heightsize);
-	if (!colBuf) {
-	    /* Throw data away */
-	    _XEatData(dpy, compsize - __GLX_PAD(widthsize));
-	    __glXSetError(gc, GL_OUT_OF_MEMORY);
-	    UnlockDisplay(dpy);
-	    SyncHandle();
-	    return;
-	} else {
-	    __GLX_SINGLE_GET_CHAR_ARRAY(((char*)colBuf),heightsize);
-	    __glEmptyImage(gc, 1, height, 1, 1, format, type, colBuf, column);
-	    Xfree((char*) colBuf);
-	}
-    } else {
-	/*
-	** don't modify user's buffer.
-	*/
-    }
-    __GLX_SINGLE_END();
-    
+   __GLX_SINGLE_DECLARE_VARIABLES();
+   const __GLXattribute *state;
+   xGLXGetSeparableFilterReply reply;
+   GLubyte *rowBuf, *colBuf;
+
+   if (!dpy)
+      return;
+   __GLX_SINGLE_LOAD_VARIABLES();
+   state = gc->client_state_private;
+
+   /* Send request */
+   __GLX_SINGLE_BEGIN(X_GLsop_GetSeparableFilter, __GLX_PAD(13));
+   __GLX_SINGLE_PUT_LONG(0, target);
+   __GLX_SINGLE_PUT_LONG(4, format);
+   __GLX_SINGLE_PUT_LONG(8, type);
+   __GLX_SINGLE_PUT_CHAR(12, state->storePack.swapEndian);
+   __GLX_SINGLE_READ_XREPLY();
+   compsize = reply.length << 2;
+
+   if (compsize != 0) {
+      GLint width, height;
+      GLint widthsize, heightsize;
+
+      width = reply.width;
+      height = reply.height;
+
+      widthsize = __glImageSize(width, 1, 1, format, type, 0);
+      heightsize = __glImageSize(height, 1, 1, format, type, 0);
+
+      /* Allocate a holding buffer to transform the data from */
+      rowBuf = (GLubyte *) Xmalloc(widthsize);
+      if (!rowBuf) {
+         /* Throw data away */
+         _XEatData(dpy, compsize);
+         __glXSetError(gc, GL_OUT_OF_MEMORY);
+         UnlockDisplay(dpy);
+         SyncHandle();
+         return;
+      }
+      else {
+         __GLX_SINGLE_GET_CHAR_ARRAY(((char *) rowBuf), widthsize);
+         __glEmptyImage(gc, 1, width, 1, 1, format, type, rowBuf, row);
+         Xfree((char *) rowBuf);
+      }
+      colBuf = (GLubyte *) Xmalloc(heightsize);
+      if (!colBuf) {
+         /* Throw data away */
+         _XEatData(dpy, compsize - __GLX_PAD(widthsize));
+         __glXSetError(gc, GL_OUT_OF_MEMORY);
+         UnlockDisplay(dpy);
+         SyncHandle();
+         return;
+      }
+      else {
+         __GLX_SINGLE_GET_CHAR_ARRAY(((char *) colBuf), heightsize);
+         __glEmptyImage(gc, 1, height, 1, 1, format, type, colBuf, column);
+         Xfree((char *) colBuf);
+      }
+   }
+   else {
+      /*
+       ** don't modify user's buffer.
+       */
+   }
+   __GLX_SINGLE_END();
+
 }
 
 
 #define CONCAT(a,b) a ## b
 #define NAME(o) CONCAT(gl_dispatch_stub_, o)
 
-void NAME(_gloffset_GetSeparableFilter)(GLenum target, GLenum format, GLenum type,
-					GLvoid *row, GLvoid *column, GLvoid *span)
+void NAME(_gloffset_GetSeparableFilter) (GLenum target, GLenum format,
+                                         GLenum type, GLvoid * row,
+                                         GLvoid * column, GLvoid * span)
 {
-    __GLXcontext * const gc = __glXGetCurrentContext();
-
-    if (gc->isDirect) {
-	CALL_GetSeparableFilter(GET_DISPATCH(),
-				(target, format, type, row, column, span));
-	return;
-    }
-    else {
-        Display *const dpy = gc->currentDpy;
-	const GLuint cmdlen = __GLX_PAD(13);
-
-	if (dpy != NULL) {
-	    const __GLXattribute * const state = gc->client_state_private;
-	    xGLXGetSeparableFilterReply reply;
-	    GLubyte const *pc =
-	      __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
-				      X_GLvop_GetSeparableFilterEXT, cmdlen);
-	    unsigned compsize;
-
-
-	    (void) memcpy((void *) (pc + 0), (void *) (&target), 4);
-	    (void) memcpy((void *) (pc + 4), (void *) (&format), 4);
-	    (void) memcpy((void *) (pc + 8), (void *) (&type), 4);
-	    *(int8_t *) (pc + 12) = state->storePack.swapEndian;
-
-	    (void) _XReply(dpy, (xReply *) & reply, 0, False);
-
-	    compsize = reply.length << 2;
-
-	    if (compsize != 0) {
-		const GLint width = reply.width;
-		const GLint height = reply.height;
-		const GLint widthsize =
-		  __glImageSize(width, 1, 1, format, type, 0);
-		const GLint heightsize =
-		  __glImageSize(height, 1, 1, format, type, 0);
-		GLubyte * const buf =
-		  (GLubyte*) Xmalloc((widthsize > heightsize) ? widthsize : heightsize);
-
-		if (buf == NULL) {
-		    /* Throw data away */
-		    _XEatData(dpy, compsize);
-		    __glXSetError(gc, GL_OUT_OF_MEMORY);
-
-		    UnlockDisplay(dpy);
-		    SyncHandle();
-		    return;
-		} else {
-		    int extra;
-		    
-		    extra = 4 - (widthsize & 3);
-		    _XRead(dpy, (char *)buf, widthsize);
-		    if (extra < 4) {
-			_XEatData(dpy, extra);
-		    }
-
-		    __glEmptyImage(gc, 1, width, 1, 1, format, type, buf,
-				   row);
-
-		    extra = 4 - (heightsize & 3);
-		    _XRead(dpy, (char *)buf, heightsize);
-		    if (extra < 4) {
-			_XEatData(dpy, extra);
-		    }
-
-		    __glEmptyImage(gc, 1, height, 1, 1, format, type, buf,
-				   column);
-
-		    Xfree((char*) buf);
-		}
-	    }
-	}
-    }
+   __GLXcontext *const gc = __glXGetCurrentContext();
+
+#ifdef GLX_DIRECT_RENDERING
+   if (gc->driContext) {
+      CALL_GetSeparableFilter(GET_DISPATCH(),
+                              (target, format, type, row, column, span));
+      return;
+   }
+   else
+#endif
+   {
+      Display *const dpy = gc->currentDpy;
+      const GLuint cmdlen = __GLX_PAD(13);
+
+      if (dpy != NULL) {
+         const __GLXattribute *const state = gc->client_state_private;
+         xGLXGetSeparableFilterReply reply;
+         GLubyte const *pc =
+            __glXSetupVendorRequest(gc, X_GLXVendorPrivateWithReply,
+                                    X_GLvop_GetSeparableFilterEXT, cmdlen);
+         unsigned compsize;
+
+
+         (void) memcpy((void *) (pc + 0), (void *) (&target), 4);
+         (void) memcpy((void *) (pc + 4), (void *) (&format), 4);
+         (void) memcpy((void *) (pc + 8), (void *) (&type), 4);
+         *(int8_t *) (pc + 12) = state->storePack.swapEndian;
+
+         (void) _XReply(dpy, (xReply *) & reply, 0, False);
+
+         compsize = reply.length << 2;
+
+         if (compsize != 0) {
+            const GLint width = reply.width;
+            const GLint height = reply.height;
+            const GLint widthsize =
+               __glImageSize(width, 1, 1, format, type, 0);
+            const GLint heightsize =
+               __glImageSize(height, 1, 1, format, type, 0);
+            GLubyte *const buf =
+               (GLubyte *) Xmalloc((widthsize > heightsize) ? widthsize :
+                                   heightsize);
+
+            if (buf == NULL) {
+               /* Throw data away */
+               _XEatData(dpy, compsize);
+               __glXSetError(gc, GL_OUT_OF_MEMORY);
+
+               UnlockDisplay(dpy);
+               SyncHandle();
+               return;
+            }
+            else {
+               int extra;
+
+               extra = 4 - (widthsize & 3);
+               _XRead(dpy, (char *) buf, widthsize);
+               if (extra < 4) {
+                  _XEatData(dpy, extra);
+               }
+
+               __glEmptyImage(gc, 1, width, 1, 1, format, type, buf, row);
+
+               extra = 4 - (heightsize & 3);
+               _XRead(dpy, (char *) buf, heightsize);
+               if (extra < 4) {
+                  _XEatData(dpy, extra);
+               }
+
+               __glEmptyImage(gc, 1, height, 1, 1, format, type, buf, column);
+
+               Xfree((char *) buf);
+            }
+         }
+      }
+   }
 }
diff --git a/src/glx/x11/vertarr.c b/src/glx/x11/vertarr.c
index d50560ba1a5..398cfb1e796 100644
--- a/src/glx/x11/vertarr.c
+++ b/src/glx/x11/vertarr.c
@@ -1,37 +1,32 @@
 /*
-** License Applicability. Except to the extent portions of this file are
-** made subject to an alternative license as permitted in the SGI Free
-** Software License B, Version 1.1 (the "License"), the contents of this
-** file are subject only to the provisions of the License. You may not use
-** this file except in compliance with the License. You may obtain a copy
-** of the License at Silicon Graphics, Inc., attn: Legal Services, 1600
-** Amphitheatre Parkway, Mountain View, CA 94043-1351, or at:
-** 
-** http://oss.sgi.com/projects/FreeB
-** 
-** Note that, as provided in the License, the Software is distributed on an
-** "AS IS" basis, with ALL EXPRESS AND IMPLIED WARRANTIES AND CONDITIONS
-** DISCLAIMED, INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES AND
-** CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A
-** PARTICULAR PURPOSE, AND NON-INFRINGEMENT.
-** 
-** Original Code. The Original Code is: OpenGL Sample Implementation,
-** Version 1.2.1, released January 26, 2000, developed by Silicon Graphics,
-** Inc. The Original Code is Copyright (c) 1991-2000 Silicon Graphics, Inc.
-** Copyright in any portions created by third parties is as indicated
-** elsewhere herein. All Rights Reserved.
-** 
-** Additional Notice Provisions: The application programming interfaces
-** established by SGI in conjunction with the Original Code are The
-** OpenGL(R) Graphics System: A Specification (Version 1.2.1), released
-** April 1, 1999; The OpenGL(R) Graphics System Utility Library (Version
-** 1.3), released November 4, 1998; and OpenGL(R) Graphics with the X
-** Window System(R) (Version 1.3), released October 19, 1998. This software
-** was created using the OpenGL(R) version 1.2.1 Sample Implementation
-** published by SGI, but has not been independently verified as being
-** compliant with the OpenGL(R) version 1.2.1 Specification.
-**
-*/
+ * SGI FREE SOFTWARE LICENSE B (Version 2.0, Sept. 18, 2008)
+ * Copyright (C) 1991-2000 Silicon Graphics, Inc. All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice including the dates of first publication and
+ * either this permission notice or a reference to
+ * http://oss.sgi.com/projects/FreeB/
+ * shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * SILICON GRAPHICS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF
+ * OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Except as contained in this notice, the name of Silicon Graphics, Inc.
+ * shall not be used in advertising or otherwise to promote the sale, use or
+ * other dealings in this Software without prior written authorization from
+ * Silicon Graphics, Inc.
+ */
 
 #include "glxclient.h"
 #include "indirect.h"
@@ -50,153 +45,168 @@
  * happen.
  */
 /*@{*/
-void __indirect_glColorPointerEXT(GLint size, GLenum type, GLsizei stride,
-			    GLsizei count, const GLvoid * pointer )
+void
+__indirect_glColorPointerEXT(GLint size, GLenum type, GLsizei stride,
+                             GLsizei count, const GLvoid * pointer)
 {
-    (void) count; __indirect_glColorPointer( size, type, stride, pointer );
+   (void) count;
+   __indirect_glColorPointer(size, type, stride, pointer);
 }
 
-void __indirect_glEdgeFlagPointerEXT(GLsizei stride,
-			       GLsizei count, const GLboolean * pointer )
+void
+__indirect_glEdgeFlagPointerEXT(GLsizei stride,
+                                GLsizei count, const GLboolean * pointer)
 {
-    (void) count; __indirect_glEdgeFlagPointer( stride, pointer );
+   (void) count;
+   __indirect_glEdgeFlagPointer(stride, pointer);
 }
 
-void __indirect_glIndexPointerEXT(GLenum type, GLsizei stride,
-			    GLsizei count, const GLvoid * pointer )
+void
+__indirect_glIndexPointerEXT(GLenum type, GLsizei stride,
+                             GLsizei count, const GLvoid * pointer)
 {
-    (void) count; __indirect_glIndexPointer( type, stride, pointer );
+   (void) count;
+   __indirect_glIndexPointer(type, stride, pointer);
 }
 
-void __indirect_glNormalPointerEXT(GLenum type, GLsizei stride, GLsizei count,
-			     const GLvoid * pointer )
+void
+__indirect_glNormalPointerEXT(GLenum type, GLsizei stride, GLsizei count,
+                              const GLvoid * pointer)
 {
-    (void) count; __indirect_glNormalPointer( type, stride, pointer );
+   (void) count;
+   __indirect_glNormalPointer(type, stride, pointer);
 }
 
-void __indirect_glTexCoordPointerEXT(GLint size, GLenum type, GLsizei stride,
-			       GLsizei count, const GLvoid * pointer )
+void
+__indirect_glTexCoordPointerEXT(GLint size, GLenum type, GLsizei stride,
+                                GLsizei count, const GLvoid * pointer)
 {
-    (void) count; __indirect_glTexCoordPointer( size, type, stride, pointer );
+   (void) count;
+   __indirect_glTexCoordPointer(size, type, stride, pointer);
 }
 
-void __indirect_glVertexPointerEXT(GLint size, GLenum type, GLsizei stride,
-			    GLsizei count, const GLvoid * pointer )
+void
+__indirect_glVertexPointerEXT(GLint size, GLenum type, GLsizei stride,
+                              GLsizei count, const GLvoid * pointer)
 {
-    (void) count; __indirect_glVertexPointer( size, type, stride, pointer );
+   (void) count;
+   __indirect_glVertexPointer(size, type, stride, pointer);
 }
+
 /*@}*/
 
 /*****************************************************************************/
 
-void __indirect_glInterleavedArrays(GLenum format, GLsizei stride, const GLvoid *pointer)
+void
+__indirect_glInterleavedArrays(GLenum format, GLsizei stride,
+                               const GLvoid * pointer)
 {
-    __GLXcontext *gc = __glXGetCurrentContext();
-    __GLXattribute * state = (__GLXattribute *)(gc->client_state_private);
+   __GLXcontext *gc = __glXGetCurrentContext();
+   __GLXattribute *state = (__GLXattribute *) (gc->client_state_private);
 
 #define NONE  {0, 0, 0}
 #define F(x)  {GL_FLOAT, x, x * sizeof(GLfloat)}
 #define UB4   {GL_UNSIGNED_BYTE, 4, 4 * sizeof(GLubyte)}
 
-    /* Each row in this array describes the elements of a particular
-     * interleaved array mode.  Each column describes, in the order in which
-     * they appear in the interleaved arrays, one of the four possible types
-     * of vertex data that can appear in an interleaved array.
-     */
-    struct {
+   /* Each row in this array describes the elements of a particular
+    * interleaved array mode.  Each column describes, in the order in which
+    * they appear in the interleaved arrays, one of the four possible types
+    * of vertex data that can appear in an interleaved array.
+    */
+   struct
+   {
         /**
 	 * The enum describing the GL type, as would be passed to the
 	 * appropriate gl*Pointer function.
 	 */
-	GLushort type;
+      GLushort type;
 
         /**
 	 * Number of elements in the subarray, as would be passed (as the
 	 * \c size parameter) to the appropriate gl*Pointer function.
 	 */
-        GLubyte count;
-       
+      GLubyte count;
+
         /**
 	 * True size of a single element in the subarray, as would be passed
 	 * (as the \c stride parameter) to the appropriate gl*Pointer
 	 * function.
 	 */
-        GLubyte size;
-    }
-    static const modes[14][4] = {
+      GLubyte size;
+   }
+   static const modes[14][4] = {
       /* texture color normal vertex */
-	{NONE,   NONE, NONE,  F(2)}, /* GL_V2F */
-	{NONE,   NONE, NONE,  F(3)}, /* GL_V3F */
-	{NONE,   UB4,  NONE,  F(2)}, /* GL_C4UB_V2F */
-	{NONE,   UB4,  NONE,  F(3)}, /* GL_C4UB_V3F */
-	{NONE,   F(3), NONE,  F(3)}, /* GL_C3F_V3F */
-	{NONE,   NONE, F(3),  F(3)}, /* GL_N3F_V3F */
-	{NONE,   F(4), F(3),  F(3)}, /* GL_C4F_N3F_V3F */
-	{F(2),   NONE, NONE,  F(3)}, /* GL_T2F_V3F */
-	{F(4),   NONE, NONE,  F(4)}, /* GL_T4F_V4F */
-	{F(2),   UB4,  NONE,  F(3)}, /* GL_T2F_C4UB_V3F */
-	{F(2),   F(3), NONE,  F(3)}, /* GL_T2F_C3F_V3F */
-	{F(2),   NONE, F(3),  F(3)}, /* GL_T2F_N3F_V3F */
-	{F(2),   F(4), F(3),  F(3)}, /* GL_T2F_C4F_N3F_V3F */
-	{F(4),   F(4), F(3),  F(4)}, /* GL_T4F_C4F_N3F_V4F */
-    };
+      {NONE, NONE, NONE, F(2)}, /* GL_V2F */
+      {NONE, NONE, NONE, F(3)}, /* GL_V3F */
+      {NONE, UB4, NONE, F(2)},  /* GL_C4UB_V2F */
+      {NONE, UB4, NONE, F(3)},  /* GL_C4UB_V3F */
+      {NONE, F(3), NONE, F(3)}, /* GL_C3F_V3F */
+      {NONE, NONE, F(3), F(3)}, /* GL_N3F_V3F */
+      {NONE, F(4), F(3), F(3)}, /* GL_C4F_N3F_V3F */
+      {F(2), NONE, NONE, F(3)}, /* GL_T2F_V3F */
+      {F(4), NONE, NONE, F(4)}, /* GL_T4F_V4F */
+      {F(2), UB4, NONE, F(3)},  /* GL_T2F_C4UB_V3F */
+      {F(2), F(3), NONE, F(3)}, /* GL_T2F_C3F_V3F */
+      {F(2), NONE, F(3), F(3)}, /* GL_T2F_N3F_V3F */
+      {F(2), F(4), F(3), F(3)}, /* GL_T2F_C4F_N3F_V3F */
+      {F(4), F(4), F(3), F(4)}, /* GL_T4F_C4F_N3F_V4F */
+   };
 #undef NONE
 #undef F
 #undef UB4
 
-    GLint trueStride, size;
-    int offsets[4];
-    unsigned i;
-    const int idx = format - GL_V2F;
-
-   
-    /* All valid formats are on the range [GL_V2F, GL_V2F+0x0D].  Since idx
-     * is just the format biased by -GL_V2F, all valid idx values are on the
-     * range [0, 0x0D].
-     */
-    if ( (idx < 0) || (idx > 0x0D) ) {
-        __glXSetError(gc, GL_INVALID_ENUM);
-        return;
-    }
-
-    if ( stride < 0 ) {
-        __glXSetError(gc, GL_INVALID_VALUE);
-        return;
-    }
-
-
-    /* If the 'count' for a subarray is non-zero, then the offset of its
-     * first element is at the currently accumulated 'size'.
-     */
-    size = 0;
-    for ( i = 0 ; i < 4 ; i++ ) {
-        offsets[i] = (modes[idx][i].count != 0) ? size : -1;
-        size += modes[idx][i].size;
-    }
-
-    trueStride = (stride == 0) ? size : stride;
-
-    __glXArrayDisableAll( state );
-
-    if ( offsets[0] >= 0 ) {
-	__indirect_glEnableClientState(GL_TEXTURE_COORD_ARRAY);
-	__indirect_glTexCoordPointer( modes[idx][0].count, GL_FLOAT,
-				      trueStride, 
-				      (const char *) pointer );
-    }
-    if ( offsets[1] >= 0 ) {
-	__indirect_glEnableClientState(GL_COLOR_ARRAY);
-	__indirect_glColorPointer( modes[idx][1].count, modes[idx][1].type,
-				   trueStride, 
-				   (const char *) pointer + offsets[1] );
-    }
-    if ( offsets[2] >= 0 ) {
-	__indirect_glEnableClientState(GL_NORMAL_ARRAY);
-	__indirect_glNormalPointer( GL_FLOAT, trueStride,
-				    (const char *)pointer + offsets[2] );
-    }
-    __indirect_glEnableClientState(GL_VERTEX_ARRAY);
-    __indirect_glVertexPointer( modes[idx][3].count, GL_FLOAT,
-				trueStride, 
-				(const char *)pointer + offsets[3] );
+   GLint trueStride, size;
+   int offsets[4];
+   unsigned i;
+   const int idx = format - GL_V2F;
+
+
+   /* All valid formats are on the range [GL_V2F, GL_V2F+0x0D].  Since idx
+    * is just the format biased by -GL_V2F, all valid idx values are on the
+    * range [0, 0x0D].
+    */
+   if ((idx < 0) || (idx > 0x0D)) {
+      __glXSetError(gc, GL_INVALID_ENUM);
+      return;
+   }
+
+   if (stride < 0) {
+      __glXSetError(gc, GL_INVALID_VALUE);
+      return;
+   }
+
+
+   /* If the 'count' for a subarray is non-zero, then the offset of its
+    * first element is at the currently accumulated 'size'.
+    */
+   size = 0;
+   for (i = 0; i < 4; i++) {
+      offsets[i] = (modes[idx][i].count != 0) ? size : -1;
+      size += modes[idx][i].size;
+   }
+
+   trueStride = (stride == 0) ? size : stride;
+
+   __glXArrayDisableAll(state);
+
+   if (offsets[0] >= 0) {
+      __indirect_glEnableClientState(GL_TEXTURE_COORD_ARRAY);
+      __indirect_glTexCoordPointer(modes[idx][0].count, GL_FLOAT,
+                                   trueStride, (const char *) pointer);
+   }
+   if (offsets[1] >= 0) {
+      __indirect_glEnableClientState(GL_COLOR_ARRAY);
+      __indirect_glColorPointer(modes[idx][1].count, modes[idx][1].type,
+                                trueStride,
+                                (const char *) pointer + offsets[1]);
+   }
+   if (offsets[2] >= 0) {
+      __indirect_glEnableClientState(GL_NORMAL_ARRAY);
+      __indirect_glNormalPointer(GL_FLOAT, trueStride,
+                                 (const char *) pointer + offsets[2]);
+   }
+   __indirect_glEnableClientState(GL_VERTEX_ARRAY);
+   __indirect_glVertexPointer(modes[idx][3].count, GL_FLOAT,
+                              trueStride,
+                              (const char *) pointer + offsets[3]);
 }
diff --git a/src/glx/x11/xf86dri.h b/src/glx/x11/xf86dri.h
index c8c878f1274..f2d0dd54355 100644
--- a/src/glx/x11/xf86dri.h
+++ b/src/glx/x11/xf86dri.h
@@ -41,81 +41,80 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <X11/Xfuncproto.h>
 #include <xf86drm.h>
 
-#define X_XF86DRIQueryVersion			0
-#define X_XF86DRIQueryDirectRenderingCapable	1
-#define X_XF86DRIOpenConnection			2
-#define X_XF86DRICloseConnection		3
-#define X_XF86DRIGetClientDriverName		4
-#define X_XF86DRICreateContext			5
-#define X_XF86DRIDestroyContext			6
-#define X_XF86DRICreateDrawable			7
-#define X_XF86DRIDestroyDrawable		8
-#define X_XF86DRIGetDrawableInfo		9
-#define X_XF86DRIGetDeviceInfo			10
+#define X_XF86DRIQueryVersion                   0
+#define X_XF86DRIQueryDirectRenderingCapable    1
+#define X_XF86DRIOpenConnection                 2
+#define X_XF86DRICloseConnection                3
+#define X_XF86DRIGetClientDriverName            4
+#define X_XF86DRICreateContext                  5
+#define X_XF86DRIDestroyContext                 6
+#define X_XF86DRICreateDrawable                 7
+#define X_XF86DRIDestroyDrawable                8
+#define X_XF86DRIGetDrawableInfo                9
+#define X_XF86DRIGetDeviceInfo                  10
 #define X_XF86DRIAuthConnection                 11
-#define X_XF86DRIOpenFullScreen                 12   /* Deprecated */
-#define X_XF86DRICloseFullScreen                13   /* Deprecated */
+#define X_XF86DRIOpenFullScreen                 12      /* Deprecated */
+#define X_XF86DRICloseFullScreen                13      /* Deprecated */
 
-#define XF86DRINumberEvents		0
+#define XF86DRINumberEvents            0
 
-#define XF86DRIClientNotLocal		0
-#define XF86DRIOperationNotSupported	1
-#define XF86DRINumberErrors		(XF86DRIOperationNotSupported + 1)
+#define XF86DRIClientNotLocal          0
+#define XF86DRIOperationNotSupported   1
+#define XF86DRINumberErrors            (XF86DRIOperationNotSupported + 1)
 
 #ifndef _XF86DRI_SERVER_
 
-#include <GL/internal/dri_interface.h>
-
 _XFUNCPROTOBEGIN
+   Bool XF86DRIQueryExtension(Display * dpy, int *event_base,
+                              int *error_base);
 
-Bool XF86DRIQueryExtension( Display *dpy, int *event_base, int *error_base );
-
-Bool XF86DRIQueryVersion( Display *dpy, int *majorVersion, int *minorVersion,
-    int *patchVersion );
+Bool XF86DRIQueryVersion(Display * dpy, int *majorVersion, int *minorVersion,
+                         int *patchVersion);
 
-Bool XF86DRIQueryDirectRenderingCapable( Display *dpy, int screen,
-    Bool *isCapable );
+Bool XF86DRIQueryDirectRenderingCapable(Display * dpy, int screen,
+                                        Bool * isCapable);
 
-Bool XF86DRIOpenConnection( Display *dpy, int screen, drm_handle_t *hSAREA,
-    char **busIDString );
+Bool XF86DRIOpenConnection(Display * dpy, int screen, drm_handle_t * hSAREA,
+                           char **busIDString);
 
-Bool XF86DRIAuthConnection( Display *dpy, int screen, drm_magic_t magic );
+Bool XF86DRIAuthConnection(Display * dpy, int screen, drm_magic_t magic);
 
-Bool XF86DRICloseConnection( Display *dpy, int screen );
+Bool XF86DRICloseConnection(Display * dpy, int screen);
 
-Bool XF86DRIGetClientDriverName( Display *dpy, int screen,
-    int *ddxDriverMajorVersion, int *ddxDriverMinorVersion,
-    int *ddxDriverPatchVersion, char **clientDriverName );
+Bool XF86DRIGetClientDriverName(Display * dpy, int screen,
+                                int *ddxDriverMajorVersion,
+                                int *ddxDriverMinorVersion,
+                                int *ddxDriverPatchVersion,
+                                char **clientDriverName);
 
-Bool XF86DRICreateContext( Display *dpy, int screen, Visual *visual,
-    XID *ptr_to_returned_context_id, drm_context_t *hHWContext );
+Bool XF86DRICreateContext(Display * dpy, int screen, Visual * visual,
+                          XID * ptr_to_returned_context_id,
+                          drm_context_t * hHWContext);
 
-Bool XF86DRICreateContextWithConfig( Display *dpy, int screen, int configID,
-    XID *ptr_to_returned_context_id, drm_context_t *hHWContext );
+Bool XF86DRICreateContextWithConfig(Display * dpy, int screen, int configID,
+                                    XID * ptr_to_returned_context_id,
+                                    drm_context_t * hHWContext);
 
-extern GLboolean XF86DRIDestroyContext( __DRInativeDisplay *dpy, int screen,
-    __DRIid context_id );
+extern Bool XF86DRIDestroyContext(Display * dpy, int screen, XID context_id);
 
-extern GLboolean XF86DRICreateDrawable( __DRInativeDisplay *dpy, int screen,
-    __DRIid drawable, drm_drawable_t *hHWDrawable );
+extern Bool XF86DRICreateDrawable(Display * dpy, int screen,
+                                  XID drawable, drm_drawable_t * hHWDrawable);
 
-extern GLboolean XF86DRIDestroyDrawable( __DRInativeDisplay *dpy, int screen, 
-    __DRIid drawable);
+extern Bool XF86DRIDestroyDrawable(Display * dpy, int screen, XID drawable);
 
-Bool XF86DRIGetDrawableInfo( Display *dpy, int screen, Drawable drawable,
-    unsigned int *index, unsigned int *stamp, 
-    int *X, int *Y, int *W, int *H,
-    int *numClipRects, drm_clip_rect_t ** pClipRects,
-    int *backX, int *backY,
-    int *numBackClipRects, drm_clip_rect_t **pBackClipRects );
+Bool XF86DRIGetDrawableInfo(Display * dpy, int screen, Drawable drawable,
+                            unsigned int *index, unsigned int *stamp,
+                            int *X, int *Y, int *W, int *H,
+                            int *numClipRects, drm_clip_rect_t ** pClipRects,
+                            int *backX, int *backY,
+                            int *numBackClipRects,
+                            drm_clip_rect_t ** pBackClipRects);
 
-Bool XF86DRIGetDeviceInfo( Display *dpy, int screen,
-    drm_handle_t *hFrameBuffer, int *fbOrigin, int *fbSize,
-    int *fbStride, int *devPrivateSize, void **pDevPrivate );
+Bool XF86DRIGetDeviceInfo(Display * dpy, int screen,
+                          drm_handle_t * hFrameBuffer, int *fbOrigin,
+                          int *fbSize, int *fbStride, int *devPrivateSize,
+                          void **pDevPrivate);
 
 _XFUNCPROTOEND
-
 #endif /* _XF86DRI_SERVER_ */
-
 #endif /* _XF86DRI_H_ */
-
diff --git a/src/glx/x11/xf86dristr.h b/src/glx/x11/xf86dristr.h
index b834bd1a1a0..c8fbe9d4c70 100644
--- a/src/glx/x11/xf86dristr.h
+++ b/src/glx/x11/xf86dristr.h
@@ -54,287 +54,312 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define XF86DRI_MINOR_VERSION	1
 #define XF86DRI_PATCH_VERSION	0
 
-typedef struct _XF86DRIQueryVersion {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIQueryVersion */
-    CARD16	length B16;
+typedef struct _XF86DRIQueryVersion
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIQueryVersion */
+   CARD16 length B16;
 } xXF86DRIQueryVersionReq;
 #define sz_xXF86DRIQueryVersionReq	4
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD16	majorVersion B16;	/* major version of DRI protocol */
-    CARD16	minorVersion B16;	/* minor version of DRI protocol */
-    CARD32	patchVersion B32;       /* patch version of DRI protocol */
-    CARD32	pad3 B32;
-    CARD32	pad4 B32;
-    CARD32	pad5 B32;
-    CARD32	pad6 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD16 majorVersion B16;     /* major version of DRI protocol */
+   CARD16 minorVersion B16;     /* minor version of DRI protocol */
+   CARD32 patchVersion B32;     /* patch version of DRI protocol */
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRIQueryVersionReply;
 #define sz_xXF86DRIQueryVersionReply	32
 
-typedef struct _XF86DRIQueryDirectRenderingCapable {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* X_DRIQueryDirectRenderingCapable */
-    CARD16	length B16;
-    CARD32	screen B32;
+typedef struct _XF86DRIQueryDirectRenderingCapable
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* X_DRIQueryDirectRenderingCapable */
+   CARD16 length B16;
+   CARD32 screen B32;
 } xXF86DRIQueryDirectRenderingCapableReq;
 #define sz_xXF86DRIQueryDirectRenderingCapableReq	8
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    BOOL	isCapable;
-    BOOL	pad2;
-    BOOL	pad3;
-    BOOL	pad4;
-    CARD32	pad5 B32;
-    CARD32	pad6 B32;
-    CARD32	pad7 B32;
-    CARD32	pad8 B32;
-    CARD32	pad9 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   BOOL isCapable;
+   BOOL pad2;
+   BOOL pad3;
+   BOOL pad4;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
+   CARD32 pad7 B32;
+   CARD32 pad8 B32;
+   CARD32 pad9 B32;
 } xXF86DRIQueryDirectRenderingCapableReply;
 #define sz_xXF86DRIQueryDirectRenderingCapableReply	32
 
-typedef struct _XF86DRIOpenConnection {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIOpenConnection */
-    CARD16	length B16;
-    CARD32	screen B32;
+typedef struct _XF86DRIOpenConnection
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIOpenConnection */
+   CARD16 length B16;
+   CARD32 screen B32;
 } xXF86DRIOpenConnectionReq;
 #define sz_xXF86DRIOpenConnectionReq	8
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	hSAREALow B32;
-    CARD32	hSAREAHigh B32;
-    CARD32	busIdStringLength B32;
-    CARD32	pad6 B32;
-    CARD32	pad7 B32;
-    CARD32	pad8 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 hSAREALow B32;
+   CARD32 hSAREAHigh B32;
+   CARD32 busIdStringLength B32;
+   CARD32 pad6 B32;
+   CARD32 pad7 B32;
+   CARD32 pad8 B32;
 } xXF86DRIOpenConnectionReply;
 #define sz_xXF86DRIOpenConnectionReply	32
 
-typedef struct _XF86DRIAuthConnection {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRICloseConnection */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32      magic B32;
+typedef struct _XF86DRIAuthConnection
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRICloseConnection */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 magic B32;
 } xXF86DRIAuthConnectionReq;
 #define sz_xXF86DRIAuthConnectionReq	12
 
-typedef struct {
-    BYTE        type;
-    BOOL        pad1;
-    CARD16      sequenceNumber B16;
-    CARD32      length B32;
-    CARD32      authenticated B32;
-    CARD32      pad2 B32;
-    CARD32      pad3 B32;
-    CARD32      pad4 B32;
-    CARD32      pad5 B32;
-    CARD32      pad6 B32;
+typedef struct
+{
+   BYTE type;
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 authenticated B32;
+   CARD32 pad2 B32;
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRIAuthConnectionReply;
 #define zx_xXF86DRIAuthConnectionReply  32
 
-typedef struct _XF86DRICloseConnection {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRICloseConnection */
-    CARD16	length B16;
-    CARD32	screen B32;
+typedef struct _XF86DRICloseConnection
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRICloseConnection */
+   CARD16 length B16;
+   CARD32 screen B32;
 } xXF86DRICloseConnectionReq;
 #define sz_xXF86DRICloseConnectionReq	8
 
-typedef struct _XF86DRIGetClientDriverName {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIGetClientDriverName */
-    CARD16	length B16;
-    CARD32	screen B32;
+typedef struct _XF86DRIGetClientDriverName
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIGetClientDriverName */
+   CARD16 length B16;
+   CARD32 screen B32;
 } xXF86DRIGetClientDriverNameReq;
 #define sz_xXF86DRIGetClientDriverNameReq	8
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	ddxDriverMajorVersion B32;
-    CARD32	ddxDriverMinorVersion B32;
-    CARD32	ddxDriverPatchVersion B32;
-    CARD32	clientDriverNameLength B32;
-    CARD32	pad5 B32;
-    CARD32	pad6 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 ddxDriverMajorVersion B32;
+   CARD32 ddxDriverMinorVersion B32;
+   CARD32 ddxDriverPatchVersion B32;
+   CARD32 clientDriverNameLength B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRIGetClientDriverNameReply;
 #define sz_xXF86DRIGetClientDriverNameReply	32
 
-typedef struct _XF86DRICreateContext {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRICreateContext */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32	visual B32;
-    CARD32	context B32;
+typedef struct _XF86DRICreateContext
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRICreateContext */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 visual B32;
+   CARD32 context B32;
 } xXF86DRICreateContextReq;
 #define sz_xXF86DRICreateContextReq	16
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	hHWContext B32;
-    CARD32	pad2 B32;
-    CARD32	pad3 B32;
-    CARD32	pad4 B32;
-    CARD32	pad5 B32;
-    CARD32	pad6 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 hHWContext B32;
+   CARD32 pad2 B32;
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRICreateContextReply;
 #define sz_xXF86DRICreateContextReply	32
 
-typedef struct _XF86DRIDestroyContext {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIDestroyContext */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32	context B32;
+typedef struct _XF86DRIDestroyContext
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIDestroyContext */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 context B32;
 } xXF86DRIDestroyContextReq;
 #define sz_xXF86DRIDestroyContextReq	12
 
-typedef struct _XF86DRICreateDrawable {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRICreateDrawable */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32	drawable B32;
+typedef struct _XF86DRICreateDrawable
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRICreateDrawable */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 drawable B32;
 } xXF86DRICreateDrawableReq;
 #define sz_xXF86DRICreateDrawableReq	12
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	hHWDrawable B32;
-    CARD32	pad2 B32;
-    CARD32	pad3 B32;
-    CARD32	pad4 B32;
-    CARD32	pad5 B32;
-    CARD32	pad6 B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 hHWDrawable B32;
+   CARD32 pad2 B32;
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRICreateDrawableReply;
 #define sz_xXF86DRICreateDrawableReply	32
 
-typedef struct _XF86DRIDestroyDrawable {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIDestroyDrawable */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32	drawable B32;
+typedef struct _XF86DRIDestroyDrawable
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIDestroyDrawable */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 drawable B32;
 } xXF86DRIDestroyDrawableReq;
 #define sz_xXF86DRIDestroyDrawableReq	12
 
-typedef struct _XF86DRIGetDrawableInfo {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIGetDrawableInfo */
-    CARD16	length B16;
-    CARD32	screen B32;
-    CARD32	drawable B32;
+typedef struct _XF86DRIGetDrawableInfo
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIGetDrawableInfo */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 drawable B32;
 } xXF86DRIGetDrawableInfoReq;
 #define sz_xXF86DRIGetDrawableInfoReq	12
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	drawableTableIndex B32;
-    CARD32	drawableTableStamp B32;
-    INT16	drawableX B16;
-    INT16	drawableY B16;
-    INT16	drawableWidth B16;
-    INT16	drawableHeight B16;
-    CARD32	numClipRects B32;
-    INT16       backX B16;
-    INT16       backY B16;
-    CARD32      numBackClipRects B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 drawableTableIndex B32;
+   CARD32 drawableTableStamp B32;
+   INT16 drawableX B16;
+   INT16 drawableY B16;
+   INT16 drawableWidth B16;
+   INT16 drawableHeight B16;
+   CARD32 numClipRects B32;
+   INT16 backX B16;
+   INT16 backY B16;
+   CARD32 numBackClipRects B32;
 } xXF86DRIGetDrawableInfoReply;
 
 #define sz_xXF86DRIGetDrawableInfoReply	36
 
 
-typedef struct _XF86DRIGetDeviceInfo {
-    CARD8	reqType;		/* always DRIReqCode */
-    CARD8	driReqType;		/* always X_DRIGetDeviceInfo */
-    CARD16	length B16;
-    CARD32	screen B32;
+typedef struct _XF86DRIGetDeviceInfo
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIGetDeviceInfo */
+   CARD16 length B16;
+   CARD32 screen B32;
 } xXF86DRIGetDeviceInfoReq;
 #define sz_xXF86DRIGetDeviceInfoReq	8
 
-typedef struct {
-    BYTE	type;			/* X_Reply */
-    BOOL	pad1;
-    CARD16	sequenceNumber B16;
-    CARD32	length B32;
-    CARD32	hFrameBufferLow B32;
-    CARD32	hFrameBufferHigh B32;
-    CARD32	framebufferOrigin B32;
-    CARD32	framebufferSize B32;
-    CARD32	framebufferStride B32;
-    CARD32	devPrivateSize B32;
+typedef struct
+{
+   BYTE type;                   /* X_Reply */
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 hFrameBufferLow B32;
+   CARD32 hFrameBufferHigh B32;
+   CARD32 framebufferOrigin B32;
+   CARD32 framebufferSize B32;
+   CARD32 framebufferStride B32;
+   CARD32 devPrivateSize B32;
 } xXF86DRIGetDeviceInfoReply;
 #define sz_xXF86DRIGetDeviceInfoReply	32
 
-typedef struct _XF86DRIOpenFullScreen {
-    CARD8       reqType;	/* always DRIReqCode */
-    CARD8       driReqType;	/* always X_DRIOpenFullScreen */
-    CARD16      length B16;
-    CARD32      screen B32;
-    CARD32      drawable B32;
+typedef struct _XF86DRIOpenFullScreen
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRIOpenFullScreen */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 drawable B32;
 } xXF86DRIOpenFullScreenReq;
 #define sz_xXF86DRIOpenFullScreenReq    12
 
-typedef struct {
-    BYTE        type;
-    BOOL        pad1;
-    CARD16      sequenceNumber B16;
-    CARD32      length B32;
-    CARD32      isFullScreen B32;
-    CARD32      pad2 B32;
-    CARD32      pad3 B32;
-    CARD32      pad4 B32;
-    CARD32      pad5 B32;
-    CARD32      pad6 B32;
+typedef struct
+{
+   BYTE type;
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 isFullScreen B32;
+   CARD32 pad2 B32;
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
 } xXF86DRIOpenFullScreenReply;
 #define sz_xXF86DRIOpenFullScreenReply  32
 
-typedef struct _XF86DRICloseFullScreen {
-    CARD8       reqType;	/* always DRIReqCode */
-    CARD8       driReqType;	/* always X_DRICloseFullScreen */
-    CARD16      length B16;
-    CARD32      screen B32;
-    CARD32      drawable B32;
+typedef struct _XF86DRICloseFullScreen
+{
+   CARD8 reqType;               /* always DRIReqCode */
+   CARD8 driReqType;            /* always X_DRICloseFullScreen */
+   CARD16 length B16;
+   CARD32 screen B32;
+   CARD32 drawable B32;
 } xXF86DRICloseFullScreenReq;
 #define sz_xXF86DRICloseFullScreenReq   12
 
-typedef struct {
-    BYTE        type;
-    BOOL        pad1;
-    CARD16      sequenceNumber B16;
-    CARD32      length B32;
-    CARD32      pad2 B32;
-    CARD32      pad3 B32;
-    CARD32      pad4 B32;
-    CARD32      pad5 B32;
-    CARD32      pad6 B32;
-    CARD32      pad7 B32;
+typedef struct
+{
+   BYTE type;
+   BOOL pad1;
+   CARD16 sequenceNumber B16;
+   CARD32 length B32;
+   CARD32 pad2 B32;
+   CARD32 pad3 B32;
+   CARD32 pad4 B32;
+   CARD32 pad5 B32;
+   CARD32 pad6 B32;
+   CARD32 pad7 B32;
 } xXF86DRICloseFullScreenReply;
 #define sz_xXF86DRICloseFullScreenReply  32
 
diff --git a/src/glx/x11/xfont.c b/src/glx/x11/xfont.c
index f3e3da3e790..797fd7a4900 100644
--- a/src/glx/x11/xfont.c
+++ b/src/glx/x11/xfont.c
@@ -33,6 +33,7 @@
   called by that routine when direct rendering is enabled.
 */
 
+#ifdef GLX_DIRECT_RENDERING
 
 #include "glxclient.h"
 
@@ -47,59 +48,56 @@
 int debug_xfonts = 0;
 
 static void
-dump_char_struct (XCharStruct *ch, char *prefix)
+dump_char_struct(XCharStruct * ch, char *prefix)
 {
-  printf ("%slbearing = %d, rbearing = %d, width = %d\n",
+   printf("%slbearing = %d, rbearing = %d, width = %d\n",
           prefix, ch->lbearing, ch->rbearing, ch->width);
-  printf ("%sascent = %d, descent = %d, attributes = %u\n",
+   printf("%sascent = %d, descent = %d, attributes = %u\n",
           prefix, ch->ascent, ch->descent, (unsigned int) ch->attributes);
 }
 
 static void
-dump_font_struct (XFontStruct *font)
+dump_font_struct(XFontStruct * font)
 {
-  printf ("ascent = %d, descent = %d\n", font->ascent, font->descent);
-  printf ("char_or_byte2 = (%u,%u)\n",
+   printf("ascent = %d, descent = %d\n", font->ascent, font->descent);
+   printf("char_or_byte2 = (%u,%u)\n",
           font->min_char_or_byte2, font->max_char_or_byte2);
-  printf ("byte1 = (%u,%u)\n", font->min_byte1, font->max_byte1);
-  printf ("all_chars_exist = %s\n", font->all_chars_exist ? "True" : 
-"False");
-  printf ("default_char = %c (\\%03o)\n",
-          (char) (isprint (font->default_char) ? font->default_char : ' '),
+   printf("byte1 = (%u,%u)\n", font->min_byte1, font->max_byte1);
+   printf("all_chars_exist = %s\n", font->all_chars_exist ? "True" : "False");
+   printf("default_char = %c (\\%03o)\n",
+          (char) (isprint(font->default_char) ? font->default_char : ' '),
           font->default_char);
-  dump_char_struct (&font->min_bounds, "min> ");
-  dump_char_struct (&font->max_bounds, "max> ");
+   dump_char_struct(&font->min_bounds, "min> ");
+   dump_char_struct(&font->max_bounds, "max> ");
 #if 0
-  for (c = font->min_char_or_byte2; c <= font->max_char_or_byte2; c++)
-    {
+   for (c = font->min_char_or_byte2; c <= font->max_char_or_byte2; c++) {
       char prefix[8];
-      sprintf (prefix, "%d> ", c);
-      dump_char_struct (&font->per_char[c], prefix);
-    }
+      sprintf(prefix, "%d> ", c);
+      dump_char_struct(&font->per_char[c], prefix);
+   }
 #endif
 }
 
 static void
-dump_bitmap (unsigned int width, unsigned int height, GLubyte *bitmap)
+dump_bitmap(unsigned int width, unsigned int height, GLubyte * bitmap)
 {
-  unsigned int x, y;
-
-  printf ("    ");
-  for (x = 0; x < 8*width; x++)
-    printf ("%o", 7 - (x % 8));
-  putchar ('\n');
-  for (y = 0; y < height; y++)
-    {
-      printf ("%3o:", y);
-      for (x = 0; x < 8*width; x++)
-        putchar ((bitmap[width*(height - y - 1) + x/8] & (1 << (7 - (x % 
-8))))
+   unsigned int x, y;
+
+   printf("    ");
+   for (x = 0; x < 8 * width; x++)
+      printf("%o", 7 - (x % 8));
+   putchar('\n');
+   for (y = 0; y < height; y++) {
+      printf("%3o:", y);
+      for (x = 0; x < 8 * width; x++)
+         putchar((bitmap[width * (height - y - 1) + x / 8] & (1 << (7 - (x %
+                                                                         8))))
                  ? '*' : '.');
-      printf ("   ");
+      printf("   ");
       for (x = 0; x < width; x++)
-        printf ("0x%02x, ", bitmap[width*(height - y - 1) + x]);
-      putchar ('\n');
-    }
+         printf("0x%02x, ", bitmap[width * (height - y - 1) + x]);
+      putchar('\n');
+   }
 }
 #endif /* DEBUG */
 
@@ -131,170 +129,172 @@ dump_bitmap (unsigned int width, unsigned int height, GLubyte *bitmap)
  * Generate OpenGL-compatible bitmap.
  */
 static void
-fill_bitmap (Display *dpy, Window win, GC gc,
-             unsigned int width, unsigned int height,
-             int x0, int y0, unsigned int c, GLubyte *bitmap)
+fill_bitmap(Display * dpy, Window win, GC gc,
+            unsigned int width, unsigned int height,
+            int x0, int y0, unsigned int c, GLubyte * bitmap)
 {
-  XImage *image;
-  unsigned int x, y;
-  Pixmap pixmap;
-  XChar2b char2b;
-
-  pixmap = XCreatePixmap (dpy, win, 8*width, height, 1);
-  XSetForeground(dpy, gc, 0);
-  XFillRectangle (dpy, pixmap, gc, 0, 0, 8*width, height);
-  XSetForeground(dpy, gc, 1);
-
-  char2b.byte1 = (c >> 8) & 0xff;
-  char2b.byte2 = (c & 0xff);
-
-  XDrawString16 (dpy, pixmap, gc, x0, y0, &char2b, 1);
-
-  image = XGetImage (dpy, pixmap, 0, 0, 8*width, height, 1, XYPixmap);
-  if (image) {
-    /* Fill the bitmap (X11 and OpenGL are upside down wrt each other).  */
-    for (y = 0; y < height; y++)
-      for (x = 0; x < 8*width; x++)
-        if (XGetPixel (image, x, y))
-          bitmap[width*(height - y - 1) + x/8] |= (1 << (7 - (x % 8)));
-    XDestroyImage (image);
-  }
-
-  XFreePixmap (dpy, pixmap);
+   XImage *image;
+   unsigned int x, y;
+   Pixmap pixmap;
+   XChar2b char2b;
+
+   pixmap = XCreatePixmap(dpy, win, 8 * width, height, 1);
+   XSetForeground(dpy, gc, 0);
+   XFillRectangle(dpy, pixmap, gc, 0, 0, 8 * width, height);
+   XSetForeground(dpy, gc, 1);
+
+   char2b.byte1 = (c >> 8) & 0xff;
+   char2b.byte2 = (c & 0xff);
+
+   XDrawString16(dpy, pixmap, gc, x0, y0, &char2b, 1);
+
+   image = XGetImage(dpy, pixmap, 0, 0, 8 * width, height, 1, XYPixmap);
+   if (image) {
+      /* Fill the bitmap (X11 and OpenGL are upside down wrt each other).  */
+      for (y = 0; y < height; y++)
+         for (x = 0; x < 8 * width; x++)
+            if (XGetPixel(image, x, y))
+               bitmap[width * (height - y - 1) + x / 8] |=
+                  (1 << (7 - (x % 8)));
+      XDestroyImage(image);
+   }
+
+   XFreePixmap(dpy, pixmap);
 }
 
 /*
  * determine if a given glyph is valid and return the
  * corresponding XCharStruct.
  */
-static XCharStruct *isvalid(XFontStruct *fs, int which)
+static XCharStruct *
+isvalid(XFontStruct * fs, int which)
 {
-  unsigned int  rows,pages;
-  int           byte1 = 0, byte2 = 0;
-  int           i,valid = 1;
-
-  rows = fs->max_byte1 - fs->min_byte1 + 1;
-  pages = fs->max_char_or_byte2 - fs->min_char_or_byte2 + 1;
-
-  if (rows == 1) {
-    /* "linear" fonts */
-    if ((fs->min_char_or_byte2 > which) ||
-        (fs->max_char_or_byte2 < which)) valid = 0;
-  } else {
-    /* "matrix" fonts */
-    byte2 = which & 0xff;
-    byte1 = which >> 8;
-    if ((fs->min_char_or_byte2 > byte2) ||
-        (fs->max_char_or_byte2 < byte2) ||
-        (fs->min_byte1 > byte1) ||
-        (fs->max_byte1 < byte1)) valid = 0;
-  }
-
-  if (valid) {
-    if (fs->per_char) {
-      if (rows == 1) {
-        /* "linear" fonts */
-        return(fs->per_char + (which-fs->min_char_or_byte2) );
-      } else {
-        /* "matrix" fonts */
-        i = ((byte1 - fs->min_byte1) * pages) +
-             (byte2 - fs->min_char_or_byte2);
-        return(fs->per_char + i);
+   unsigned int rows, pages;
+   int byte1 = 0, byte2 = 0;
+   int i, valid = 1;
+
+   rows = fs->max_byte1 - fs->min_byte1 + 1;
+   pages = fs->max_char_or_byte2 - fs->min_char_or_byte2 + 1;
+
+   if (rows == 1) {
+      /* "linear" fonts */
+      if ((fs->min_char_or_byte2 > which) || (fs->max_char_or_byte2 < which))
+         valid = 0;
+   }
+   else {
+      /* "matrix" fonts */
+      byte2 = which & 0xff;
+      byte1 = which >> 8;
+      if ((fs->min_char_or_byte2 > byte2) ||
+          (fs->max_char_or_byte2 < byte2) ||
+          (fs->min_byte1 > byte1) || (fs->max_byte1 < byte1))
+         valid = 0;
+   }
+
+   if (valid) {
+      if (fs->per_char) {
+         if (rows == 1) {
+            /* "linear" fonts */
+            return (fs->per_char + (which - fs->min_char_or_byte2));
+         }
+         else {
+            /* "matrix" fonts */
+            i = ((byte1 - fs->min_byte1) * pages) +
+               (byte2 - fs->min_char_or_byte2);
+            return (fs->per_char + i);
+         }
       }
-    } else {
-        return(&fs->min_bounds);
-    }
-  }
-  return(NULL);
+      else {
+         return (&fs->min_bounds);
+      }
+   }
+   return (NULL);
 }
 
-
-void DRI_glXUseXFont( Font font, int first, int count, int listbase )
+_X_HIDDEN void
+DRI_glXUseXFont(Font font, int first, int count, int listbase)
 {
-  GLXContext CC;
-  Display *dpy;
-  Window win;
-  Pixmap pixmap;
-  GC gc;
-  XGCValues values;
-  unsigned long valuemask;
-  XFontStruct *fs;
-
-  GLint swapbytes, lsbfirst, rowlength;
-  GLint skiprows, skippixels, alignment;
-
-  unsigned int max_width, max_height, max_bm_width, max_bm_height;
-  GLubyte *bm;
-
-  int i;
-
-  CC = __glXGetCurrentContext();
-  dpy = CC->currentDpy;
-  win = CC->currentDrawable;
-
-  fs = XQueryFont (dpy, font);
-  if (!fs)
-    {
+   GLXContext CC;
+   Display *dpy;
+   Window win;
+   Pixmap pixmap;
+   GC gc;
+   XGCValues values;
+   unsigned long valuemask;
+   XFontStruct *fs;
+
+   GLint swapbytes, lsbfirst, rowlength;
+   GLint skiprows, skippixels, alignment;
+
+   unsigned int max_width, max_height, max_bm_width, max_bm_height;
+   GLubyte *bm;
+
+   int i;
+
+   CC = __glXGetCurrentContext();
+   dpy = CC->currentDpy;
+   win = CC->currentDrawable;
+
+   fs = XQueryFont(dpy, font);
+   if (!fs) {
       __glXSetError(CC, GL_INVALID_VALUE);
       return;
-    }
-
-  /* Allocate a bitmap that can fit all characters.  */
-  max_width = fs->max_bounds.rbearing - fs->min_bounds.lbearing;
-  max_height = fs->max_bounds.ascent + fs->max_bounds.descent;
-  max_bm_width = (max_width + 7) / 8;
-  max_bm_height = max_height;
-
-  bm = (GLubyte *) Xmalloc((max_bm_width * max_bm_height) * sizeof 
-(GLubyte));
-  if (!bm) {
-      XFreeFontInfo( NULL, fs, 1 );
+   }
+
+   /* Allocate a bitmap that can fit all characters.  */
+   max_width = fs->max_bounds.rbearing - fs->min_bounds.lbearing;
+   max_height = fs->max_bounds.ascent + fs->max_bounds.descent;
+   max_bm_width = (max_width + 7) / 8;
+   max_bm_height = max_height;
+
+   bm = (GLubyte *) Xmalloc((max_bm_width * max_bm_height) * sizeof(GLubyte));
+   if (!bm) {
+      XFreeFontInfo(NULL, fs, 1);
       __glXSetError(CC, GL_OUT_OF_MEMORY);
       return;
-    }
+   }
 
 #if 0
-  /* get the page info */
-  pages = fs->max_char_or_byte2 - fs->min_char_or_byte2 + 1;
-  firstchar = (fs->min_byte1 << 8) + fs->min_char_or_byte2;
-  lastchar = (fs->max_byte1 << 8) + fs->max_char_or_byte2;
-  rows = fs->max_byte1 - fs->min_byte1 + 1;
-  unsigned int first_char, last_char, pages, rows;
+   /* get the page info */
+   pages = fs->max_char_or_byte2 - fs->min_char_or_byte2 + 1;
+   firstchar = (fs->min_byte1 << 8) + fs->min_char_or_byte2;
+   lastchar = (fs->max_byte1 << 8) + fs->max_char_or_byte2;
+   rows = fs->max_byte1 - fs->min_byte1 + 1;
+   unsigned int first_char, last_char, pages, rows;
 #endif
 
-  /* Save the current packing mode for bitmaps.  */
-  glGetIntegerv (GL_UNPACK_SWAP_BYTES, &swapbytes);
-  glGetIntegerv (GL_UNPACK_LSB_FIRST, &lsbfirst);
-  glGetIntegerv (GL_UNPACK_ROW_LENGTH, &rowlength);
-  glGetIntegerv (GL_UNPACK_SKIP_ROWS, &skiprows);
-  glGetIntegerv (GL_UNPACK_SKIP_PIXELS, &skippixels);
-  glGetIntegerv (GL_UNPACK_ALIGNMENT, &alignment);
-
-  /* Enforce a standard packing mode which is compatible with
-     fill_bitmap() from above.  This is actually the default mode,
-     except for the (non)alignment.  */
-  glPixelStorei (GL_UNPACK_SWAP_BYTES, GL_FALSE);
-  glPixelStorei (GL_UNPACK_LSB_FIRST, GL_FALSE);
-  glPixelStorei (GL_UNPACK_ROW_LENGTH, 0);
-  glPixelStorei (GL_UNPACK_SKIP_ROWS, 0);
-  glPixelStorei (GL_UNPACK_SKIP_PIXELS, 0);
-  glPixelStorei (GL_UNPACK_ALIGNMENT, 1);
-
-  pixmap = XCreatePixmap (dpy, win, 10, 10, 1);
-  values.foreground = BlackPixel (dpy, DefaultScreen (dpy));
-  values.background = WhitePixel (dpy, DefaultScreen (dpy));
-  values.font = fs->fid;
-  valuemask = GCForeground | GCBackground | GCFont;
-  gc = XCreateGC (dpy, pixmap, valuemask, &values);
-  XFreePixmap (dpy, pixmap);
+   /* Save the current packing mode for bitmaps.  */
+   glGetIntegerv(GL_UNPACK_SWAP_BYTES, &swapbytes);
+   glGetIntegerv(GL_UNPACK_LSB_FIRST, &lsbfirst);
+   glGetIntegerv(GL_UNPACK_ROW_LENGTH, &rowlength);
+   glGetIntegerv(GL_UNPACK_SKIP_ROWS, &skiprows);
+   glGetIntegerv(GL_UNPACK_SKIP_PIXELS, &skippixels);
+   glGetIntegerv(GL_UNPACK_ALIGNMENT, &alignment);
+
+   /* Enforce a standard packing mode which is compatible with
+      fill_bitmap() from above.  This is actually the default mode,
+      except for the (non)alignment.  */
+   glPixelStorei(GL_UNPACK_SWAP_BYTES, GL_FALSE);
+   glPixelStorei(GL_UNPACK_LSB_FIRST, GL_FALSE);
+   glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
+   glPixelStorei(GL_UNPACK_SKIP_ROWS, 0);
+   glPixelStorei(GL_UNPACK_SKIP_PIXELS, 0);
+   glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
+
+   pixmap = XCreatePixmap(dpy, win, 10, 10, 1);
+   values.foreground = BlackPixel(dpy, DefaultScreen(dpy));
+   values.background = WhitePixel(dpy, DefaultScreen(dpy));
+   values.font = fs->fid;
+   valuemask = GCForeground | GCBackground | GCFont;
+   gc = XCreateGC(dpy, pixmap, valuemask, &values);
+   XFreePixmap(dpy, pixmap);
 
 #ifdef DEBUG
-  if (debug_xfonts)
-    dump_font_struct (fs);
+   if (debug_xfonts)
+      dump_font_struct(fs);
 #endif
 
-  for (i = 0; i < count; i++)
-    {
+   for (i = 0; i < count; i++) {
       unsigned int width, height, bm_width, bm_height;
       GLfloat x0, y0, dx, dy;
       XCharStruct *ch;
@@ -306,17 +306,18 @@ void DRI_glXUseXFont( Font font, int first, int count, int listbase )
       /* check on index validity and get the bounds */
       ch = isvalid(fs, c);
       if (!ch) {
-        ch = &fs->max_bounds;
-        valid = 0;
-      } else {
-        valid = 1;
+         ch = &fs->max_bounds;
+         valid = 0;
+      }
+      else {
+         valid = 1;
       }
 
 #ifdef DEBUG
       if (debug_xfonts) {
-          char s[7];
-          sprintf (s, isprint (c) ? "%c> " : "\\%03o> ", c);
-          dump_char_struct (ch, s);
+         char s[7];
+         sprintf(s, isprint(c) ? "%c> " : "\\%03o> ", c);
+         dump_char_struct(ch, s);
       }
 #endif
 
@@ -324,13 +325,13 @@ void DRI_glXUseXFont( Font font, int first, int count, int listbase )
          straight from the glXUseXFont(3) manpage.  */
       width = ch->rbearing - ch->lbearing;
       height = ch->ascent + ch->descent;
-      x0 = - ch->lbearing;
+      x0 = -ch->lbearing;
       y0 = ch->descent - 1;
       dx = ch->width;
       dy = 0;
 
       /* X11's starting point.  */
-      x = - ch->lbearing;
+      x = -ch->lbearing;
       y = ch->ascent;
 
       /* Round the width to a multiple of eight.  We will use this also
@@ -339,38 +340,38 @@ void DRI_glXUseXFont( Font font, int first, int count, int listbase )
       bm_width = (width + 7) / 8;
       bm_height = height;
 
-      glNewList (list, GL_COMPILE);
-        if (valid && (bm_width > 0) && (bm_height > 0)) {
+      glNewList(list, GL_COMPILE);
+      if (valid && (bm_width > 0) && (bm_height > 0)) {
 
-            memset (bm, '\0', bm_width * bm_height);
-            fill_bitmap (dpy, win, gc, bm_width, bm_height, x, y, c, bm);
+         memset(bm, '\0', bm_width * bm_height);
+         fill_bitmap(dpy, win, gc, bm_width, bm_height, x, y, c, bm);
 
-            glBitmap (width, height, x0, y0, dx, dy, bm);
+         glBitmap(width, height, x0, y0, dx, dy, bm);
 #ifdef DEBUG
-            if (debug_xfonts) {
-                printf ("width/height = %u/%u\n", width, height);
-                printf ("bm_width/bm_height = %u/%u\n", bm_width, 
-bm_height);
-                dump_bitmap (bm_width, bm_height, bm);
-              }
+         if (debug_xfonts) {
+            printf("width/height = %u/%u\n", width, height);
+            printf("bm_width/bm_height = %u/%u\n", bm_width, bm_height);
+            dump_bitmap(bm_width, bm_height, bm);
+         }
 #endif
-          } else {
-            glBitmap (0, 0, 0.0, 0.0, dx, dy, NULL);
-          }
-      glEndList ();
-    }
-
-  Xfree(bm);
-  XFreeFontInfo( NULL, fs, 1 );
-  XFreeGC (dpy, gc);
-
-  /* Restore saved packing modes.  */
-  glPixelStorei(GL_UNPACK_SWAP_BYTES, swapbytes);
-  glPixelStorei(GL_UNPACK_LSB_FIRST, lsbfirst);
-  glPixelStorei(GL_UNPACK_ROW_LENGTH, rowlength);
-  glPixelStorei(GL_UNPACK_SKIP_ROWS, skiprows);
-  glPixelStorei(GL_UNPACK_SKIP_PIXELS, skippixels);
-  glPixelStorei(GL_UNPACK_ALIGNMENT, alignment);
+      }
+      else {
+         glBitmap(0, 0, 0.0, 0.0, dx, dy, NULL);
+      }
+      glEndList();
+   }
+
+   Xfree(bm);
+   XFreeFontInfo(NULL, fs, 1);
+   XFreeGC(dpy, gc);
+
+   /* Restore saved packing modes.  */
+   glPixelStorei(GL_UNPACK_SWAP_BYTES, swapbytes);
+   glPixelStorei(GL_UNPACK_LSB_FIRST, lsbfirst);
+   glPixelStorei(GL_UNPACK_ROW_LENGTH, rowlength);
+   glPixelStorei(GL_UNPACK_SKIP_ROWS, skiprows);
+   glPixelStorei(GL_UNPACK_SKIP_PIXELS, skippixels);
+   glPixelStorei(GL_UNPACK_ALIGNMENT, alignment);
 }
 
-/* The End. */
+#endif
diff --git a/src/libXvMC/Makefile b/src/libXvMC/Makefile
new file mode 100644
index 00000000000..4de26e96202
--- /dev/null
+++ b/src/libXvMC/Makefile
@@ -0,0 +1,73 @@
+TARGET		= libXvMCg3dvl.so
+SONAME		= libXvMCg3dvl.so.1
+GALLIUMDIR	= ../gallium
+
+OBJECTS         = block.o surface.o context.o subpicture.o attributes.o
+
+ifeq (${DRIVER}, softpipe)
+OBJECTS         += ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o
+endif
+
+CFLAGS	+= -g -fPIC -Wall				\
+	   -I${GALLIUMDIR}/state_trackers/g3dvl		\
+	   -I${GALLIUMDIR}/winsys/g3dvl			\
+	   -I${GALLIUMDIR}/include			\
+	   -I${GALLIUMDIR}/auxiliary			\
+	   -I${GALLIUMDIR}/drivers
+
+ifeq (${DRIVER}, softpipe)
+LDFLAGS	+= -L${GALLIUMDIR}/state_trackers/g3dvl		\
+	   -L${GALLIUMDIR}/drivers/softpipe		\
+	   -L${GALLIUMDIR}/auxiliary/tgsi		\
+	   -L${GALLIUMDIR}/auxiliary/draw		\
+	   -L${GALLIUMDIR}/auxiliary/translate		\
+	   -L${GALLIUMDIR}/auxiliary/cso_cache		\
+	   -L${GALLIUMDIR}/auxiliary/util		\
+	   -L${GALLIUMDIR}/auxiliary/rtasm
+else
+LDFLAGS	+= -L${GALLIUMDIR}/state_trackers/g3dvl		\
+   	   -L${GALLIUMDIR}/winsys/g3dvl/nouveau		\
+   	   -L${GALLIUMDIR}/auxiliary/util
+endif
+
+ifeq (${DRIVER}, softpipe)
+LIBS	+= -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lm
+else
+LIBS	+= -lg3dvl -lnouveau_dri -lutil
+endif
+
+#############################################
+
+ifeq (${DRIVER}, softpipe)
+.PHONY	= all clean g3dvl
+else
+.PHONY	= all clean g3dvl nouveau_winsys
+endif
+
+all: ${TARGET}
+
+ifeq (${DRIVER}, softpipe)
+${TARGET}: ${OBJECTS} g3dvl
+	$(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS}
+
+g3dvl:
+	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE}
+
+clean:
+	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean
+	rm -rf ${OBJECTS} ${TARGET}
+else
+${TARGET}: ${OBJECTS} g3dvl nouveau_winsys
+	$(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS}
+
+g3dvl:
+	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE}
+
+nouveau_winsys:
+	cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE}
+
+clean:
+	cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean
+	cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} clean
+	rm -rf ${OBJECTS} ${TARGET}
+endif
diff --git a/src/libXvMC/attributes.c b/src/libXvMC/attributes.c
new file mode 100644
index 00000000000..674524b8b82
--- /dev/null
+++ b/src/libXvMC/attributes.c
@@ -0,0 +1,20 @@
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/Xvlib.h>
+#include <X11/extensions/XvMC.h>
+
+XvAttribute* XvMCQueryAttributes(Display *display, XvMCContext *context, int *number)
+{
+	return NULL;
+}
+
+Status XvMCSetAttribute(Display *display, XvMCContext *context, Atom attribute, int value)
+{
+	return BadImplementation;
+}
+
+Status XvMCGetAttribute(Display *display, XvMCContext *context, Atom attribute, int *value)
+{
+	return BadImplementation;
+}
+
diff --git a/src/libXvMC/block.c b/src/libXvMC/block.c
new file mode 100644
index 00000000000..328b035576b
--- /dev/null
+++ b/src/libXvMC/block.c
@@ -0,0 +1,79 @@
+#include <assert.h>
+#include <stdlib.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMC.h>
+#include <vl_display.h>
+#include <vl_screen.h>
+#include <vl_context.h>
+
+#define BLOCK_SIZE (64 * 2)
+
+Status XvMCCreateBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks)
+{
+	struct vlContext *vl_ctx;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+	if (num_blocks == 0)
+		return BadValue;
+
+	assert(blocks);
+
+	vl_ctx = context->privData;
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
+
+	blocks->context_id = context->context_id;
+	blocks->num_blocks = num_blocks;
+	blocks->blocks = malloc(BLOCK_SIZE * num_blocks);
+	/* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */
+	blocks->privData = display;
+
+	return Success;
+}
+
+Status XvMCDestroyBlocks(Display *display, XvMCBlockArray *blocks)
+{
+	assert(display);
+	assert(blocks);
+	assert(display == blocks->privData);
+	free(blocks->blocks);
+
+	return Success;
+}
+
+Status XvMCCreateMacroBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks)
+{
+	struct vlContext *vl_ctx;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+	if (num_blocks == 0)
+		return BadValue;
+
+	assert(blocks);
+
+	vl_ctx = context->privData;
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
+
+	blocks->context_id = context->context_id;
+	blocks->num_blocks = num_blocks;
+	blocks->macro_blocks = malloc(sizeof(XvMCMacroBlock) * num_blocks);
+	/* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */
+	blocks->privData = display;
+
+	return Success;
+}
+
+Status XvMCDestroyMacroBlocks(Display *display, XvMCMacroBlockArray *blocks)
+{
+	assert(display);
+	assert(blocks);
+	assert(display == blocks->privData);
+	free(blocks->macro_blocks);
+
+	return Success;
+}
diff --git a/src/libXvMC/context.c b/src/libXvMC/context.c
new file mode 100644
index 00000000000..698410d4c2e
--- /dev/null
+++ b/src/libXvMC/context.c
@@ -0,0 +1,201 @@
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMClib.h>
+#include <pipe/p_context.h>
+#include <vl_display.h>
+#include <vl_screen.h>
+#include <vl_context.h>
+#include <vl_winsys.h>
+
+static Status Validate
+(
+	Display *display,
+	XvPortID port,
+	int surface_type_id,
+	unsigned int width,
+	unsigned int height,
+	int flags,
+	int *found_port,
+	int *chroma_format,
+	int *mc_type
+)
+{
+	unsigned int	found_surface = 0;
+	XvAdaptorInfo	*adaptor_info;
+	unsigned int	num_adaptors;
+	int		num_types;
+	unsigned int	max_width, max_height;
+	Status		ret;
+	unsigned int	i, j, k;
+
+	assert(display && chroma_format);
+
+	*found_port = 0;
+
+	ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info);
+	if (ret != Success)
+		return ret;
+
+	/* Scan through all adaptors looking for this port and surface */
+	for (i = 0; i < num_adaptors && !*found_port; ++i)
+	{
+		/* Scan through all ports of this adaptor looking for our port */
+		for (j = 0; j < adaptor_info[i].num_ports && !*found_port; ++j)
+		{
+			/* If this is our port, scan through all its surfaces looking for our surface */
+			if (adaptor_info[i].base_id + j == port)
+			{
+				XvMCSurfaceInfo *surface_info;
+
+				*found_port = 1;
+				surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types);
+
+				if (surface_info)
+				{
+					for (k = 0; k < num_types && !found_surface; ++k)
+					{
+						if (surface_info[k].surface_type_id == surface_type_id)
+						{
+							found_surface = 1;
+							max_width = surface_info[k].max_width;
+							max_height = surface_info[k].max_height;
+							*chroma_format = surface_info[k].chroma_format;
+							*mc_type = surface_info[k].mc_type;
+						}
+					}
+
+					XFree(surface_info);
+				}
+				else
+				{
+					XvFreeAdaptorInfo(adaptor_info);
+					return BadAlloc;
+				}
+			}
+		}
+	}
+
+	XvFreeAdaptorInfo(adaptor_info);
+
+	if (!*found_port)
+		return XvBadPort;
+	if (!found_surface)
+		return BadMatch;
+	if (width > max_width || height > max_height)
+		return BadValue;
+	if (flags != XVMC_DIRECT && flags != 0)
+		return BadValue;
+
+	return Success;
+}
+
+static enum vlProfile ProfileToVL(int xvmc_profile)
+{
+	if (xvmc_profile & XVMC_MPEG_1)
+		assert(0);
+	else if (xvmc_profile & XVMC_MPEG_2)
+		return vlProfileMpeg2Main;
+	else if (xvmc_profile & XVMC_H263)
+		assert(0);
+	else if (xvmc_profile & XVMC_MPEG_4)
+		assert(0);
+	else
+		assert(0);
+
+	return -1;
+}
+
+static enum vlEntryPoint EntryToVL(int xvmc_entry)
+{
+	return xvmc_entry & XVMC_IDCT ? vlEntryPointIDCT : vlEntryPointMC;
+}
+
+static enum vlFormat FormatToVL(int xvmc_format)
+{
+	switch (xvmc_format)
+	{
+		case XVMC_CHROMA_FORMAT_420:
+			return vlFormatYCbCr420;
+		case XVMC_CHROMA_FORMAT_422:
+			return vlFormatYCbCr422;
+		case XVMC_CHROMA_FORMAT_444:
+			return vlFormatYCbCr444;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context)
+{
+	int			found_port;
+	int			chroma_format;
+	int			mc_type;
+	Status			ret;
+	struct vlDisplay	*vl_dpy;
+	struct vlScreen		*vl_scrn;
+	struct vlContext	*vl_ctx;
+	struct pipe_context	*pipe;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+
+	ret = Validate(display, port, surface_type_id, width, height, flags, &found_port, &chroma_format, &mc_type);
+
+	/* XXX: Success and XvBadPort have the same value */
+	if (ret != Success || !found_port)
+		return ret;
+
+	/* XXX: Assumes default screen, should check which screen port is on */
+	pipe = create_pipe_context(display, XDefaultScreen(display));
+
+	assert(pipe);
+
+	vlCreateDisplay(display, &vl_dpy);
+	vlCreateScreen(vl_dpy, XDefaultScreen(display), pipe->screen, &vl_scrn);
+	vlCreateContext
+	(
+		vl_scrn,
+		pipe,
+		width,
+		height,
+		FormatToVL(chroma_format),
+		ProfileToVL(mc_type),
+		EntryToVL(mc_type),
+		&vl_ctx
+	);
+
+	context->context_id = XAllocID(display);
+	context->surface_type_id = surface_type_id;
+	context->width = width;
+	context->height = height;
+	context->flags = flags;
+	context->port = port;
+	context->privData = vl_ctx;
+
+	return Success;
+}
+
+Status XvMCDestroyContext(Display *display, XvMCContext *context)
+{
+	struct vlContext	*vl_ctx;
+	struct pipe_context	*pipe;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+
+	vl_ctx = context->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
+
+	pipe = vlGetPipeContext(vl_ctx);
+	vlDestroyContext(vl_ctx);
+	destroy_pipe_context(pipe);
+
+	return Success;
+}
diff --git a/src/libXvMC/subpicture.c b/src/libXvMC/subpicture.c
new file mode 100644
index 00000000000..c8f70c90d0c
--- /dev/null
+++ b/src/libXvMC/subpicture.c
@@ -0,0 +1,215 @@
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/Xvlib.h>
+#include <X11/extensions/XvMC.h>
+
+Status XvMCCreateSubpicture
+(
+	Display *display,
+	XvMCContext *context,
+	XvMCSubpicture *subpicture,
+	unsigned short width,
+	unsigned short height,
+	int xvimage_id
+)
+{
+	assert(display);
+	
+	if (!context)
+		return XvMCBadContext;
+	
+	assert(subpicture);
+	
+	if (width > 2048 || height > 2048)
+		return BadValue;
+	
+	if (xvimage_id != 123)
+		return BadMatch;
+	
+	subpicture->subpicture_id = XAllocID(display);
+	subpicture->context_id = context->context_id;
+	subpicture->xvimage_id = xvimage_id;
+	subpicture->width = width;
+	subpicture->height = height;
+	subpicture->num_palette_entries = 0;
+	subpicture->entry_bytes = 0;
+	subpicture->component_order[0] = 0;
+	subpicture->component_order[1] = 0;
+	subpicture->component_order[2] = 0;
+	subpicture->component_order[3] = 0;
+	/* TODO: subpicture->privData = ;*/
+	
+	return Success;
+}
+
+Status XvMCClearSubpicture
+(
+	Display *display,
+	XvMCSubpicture *subpicture,
+	short x,
+	short y,
+	unsigned short width,
+	unsigned short height,
+	unsigned int color
+)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	/* TODO: Assert clear rect is within bounds? Or clip? */
+	
+	return Success;
+}
+
+Status XvMCCompositeSubpicture
+(
+	Display *display,
+	XvMCSubpicture *subpicture,
+	XvImage *image,
+	short srcx,
+	short srcy,
+	unsigned short width,
+	unsigned short height,
+	short dstx,
+	short dsty
+)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	assert(image);
+	
+	if (subpicture->xvimage_id != image->id)
+		return BadMatch;
+	
+	/* TODO: Assert rects are within bounds? Or clip? */
+	
+	return Success;
+}
+
+Status XvMCDestroySubpicture(Display *display, XvMCSubpicture *subpicture)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	return BadImplementation;
+}
+
+Status XvMCSetSubpicturePalette(Display *display, XvMCSubpicture *subpicture, unsigned char *palette)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	assert(palette);
+	
+	/* We don't support paletted subpictures */
+	return BadMatch;
+}
+
+Status XvMCBlendSubpicture
+(
+	Display *display,
+	XvMCSurface *target_surface,
+	XvMCSubpicture *subpicture,
+	short subx,
+	short suby,
+	unsigned short subw,
+	unsigned short subh,
+	short surfx,
+	short surfy,
+	unsigned short surfw,
+	unsigned short surfh
+)
+{
+	assert(display);
+	
+	if (!target_surface)
+		return XvMCBadSurface;
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	if (target_surface->context_id != subpicture->context_id)
+		return BadMatch;
+	
+	/* TODO: Assert rects are within bounds? Or clip? */
+	return Success;
+}
+
+Status XvMCBlendSubpicture2
+(
+	Display *display,
+	XvMCSurface *source_surface,
+	XvMCSurface *target_surface,
+	XvMCSubpicture *subpicture,
+	short subx,
+	short suby,
+	unsigned short subw,
+	unsigned short subh,
+	short surfx,
+	short surfy,
+	unsigned short surfw,
+	unsigned short surfh
+)
+{
+	assert(display);
+	
+	if (!source_surface || !target_surface)
+		return XvMCBadSurface;
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	if (source_surface->context_id != subpicture->context_id)
+		return BadMatch;
+		
+	if (source_surface->context_id != subpicture->context_id)
+		return BadMatch;
+	
+	/* TODO: Assert rects are within bounds? Or clip? */
+	return Success;
+}
+
+Status XvMCSyncSubpicture(Display *display, XvMCSubpicture *subpicture)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	return Success;
+}
+
+Status XvMCFlushSubpicture(Display *display, XvMCSubpicture *subpicture)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	return Success;
+}
+
+Status XvMCGetSubpictureStatus(Display *display, XvMCSubpicture *subpicture, int *status)
+{
+	assert(display);
+	
+	if (!subpicture)
+		return XvMCBadSubpicture;
+	
+	assert(status);
+	
+	/* TODO */
+	*status = 0;
+	
+	return Success;
+}
+
diff --git a/src/libXvMC/surface.c b/src/libXvMC/surface.c
new file mode 100644
index 00000000000..6031b39d108
--- /dev/null
+++ b/src/libXvMC/surface.c
@@ -0,0 +1,350 @@
+#include <assert.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMC.h>
+#include <vl_display.h>
+#include <vl_screen.h>
+#include <vl_context.h>
+#include <vl_surface.h>
+#include <vl_types.h>
+
+static enum vlMacroBlockType TypeToVL(int xvmc_mb_type)
+{
+	if (xvmc_mb_type & XVMC_MB_TYPE_INTRA)
+		return vlMacroBlockTypeIntra;
+	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD)
+		return vlMacroBlockTypeFwdPredicted;
+	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD)
+		return vlMacroBlockTypeBkwdPredicted;
+	if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD))
+		return vlMacroBlockTypeBiPredicted;
+
+	assert(0);
+
+	return -1;
+}
+
+static enum vlPictureType PictureToVL(int xvmc_pic)
+{
+	switch (xvmc_pic)
+	{
+		case XVMC_TOP_FIELD:
+			return vlPictureTypeTopField;
+		case XVMC_BOTTOM_FIELD:
+			return vlPictureTypeBottomField;
+		case XVMC_FRAME_PICTURE:
+			return vlPictureTypeFrame;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+static enum vlMotionType MotionToVL(int xvmc_motion_type, int xvmc_dct_type)
+{
+	switch (xvmc_motion_type)
+	{
+		case XVMC_PREDICTION_FRAME:
+			return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ? vlMotionType16x8 : vlMotionTypeFrame;
+		case XVMC_PREDICTION_FIELD:
+			return vlMotionTypeField;
+		case XVMC_PREDICTION_DUAL_PRIME:
+			return vlMotionTypeDualPrime;
+		default:
+			assert(0);
+	}
+
+	return -1;
+}
+
+Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface)
+{
+	struct vlContext *vl_ctx;
+	struct vlSurface *vl_sfc;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+	if (!surface)
+		return XvMCBadSurface;
+
+	vl_ctx = context->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
+
+	vlCreateSurface
+	(
+		vlContextGetScreen(vl_ctx),
+		context->width,
+		context->height,
+		vlGetPictureFormat(vl_ctx),
+		&vl_sfc
+	);
+
+	vlBindToContext(vl_sfc, vl_ctx);
+
+	surface->surface_id = XAllocID(display);
+	surface->context_id = context->context_id;
+	surface->surface_type_id = context->surface_type_id;
+	surface->width = context->width;
+	surface->height = context->height;
+	surface->privData = vl_sfc;
+
+	return Success;
+}
+
+Status XvMCRenderSurface
+(
+	Display *display,
+	XvMCContext *context,
+	unsigned int picture_structure,
+	XvMCSurface *target_surface,
+	XvMCSurface *past_surface,
+	XvMCSurface *future_surface,
+	unsigned int flags,
+	unsigned int num_macroblocks,
+	unsigned int first_macroblock,
+	XvMCMacroBlockArray *macroblocks,
+	XvMCBlockArray *blocks
+)
+{
+	struct vlContext		*vl_ctx;
+	struct vlSurface		*target_vl_surface;
+	struct vlSurface		*past_vl_surface;
+	struct vlSurface		*future_vl_surface;
+	struct vlMpeg2MacroBlockBatch	batch;
+	struct vlMpeg2MacroBlock	vl_macroblocks[num_macroblocks];
+	unsigned int			i;
+
+	assert(display);
+
+	if (!context)
+		return XvMCBadContext;
+	if (!target_surface)
+		return XvMCBadSurface;
+
+	if
+	(
+		picture_structure != XVMC_TOP_FIELD &&
+		picture_structure != XVMC_BOTTOM_FIELD &&
+		picture_structure != XVMC_FRAME_PICTURE
+	)
+		return BadValue;
+	if (future_surface && !past_surface)
+		return BadMatch;
+
+	vl_ctx = context->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx))));
+
+	target_vl_surface = target_surface->privData;
+	past_vl_surface = past_surface ? past_surface->privData : NULL;
+	future_vl_surface = future_surface ? future_surface->privData : NULL;
+
+	assert(context->context_id == target_surface->context_id);
+	assert(!past_surface || context->context_id == past_surface->context_id);
+	assert(!future_surface || context->context_id == future_surface->context_id);
+
+	assert(macroblocks);
+	assert(blocks);
+
+	assert(macroblocks->context_id == context->context_id);
+	assert(blocks->context_id == context->context_id);
+
+	assert(flags == 0 || flags == XVMC_SECOND_FIELD);
+
+	batch.past_surface = past_vl_surface;
+	batch.future_surface = future_vl_surface;
+	batch.picture_type = PictureToVL(picture_structure);
+	batch.field_order = flags & XVMC_SECOND_FIELD ? vlFieldOrderSecond : vlFieldOrderFirst;
+	batch.num_macroblocks = num_macroblocks;
+	batch.macroblocks = vl_macroblocks;
+
+	for (i = 0; i < num_macroblocks; ++i)
+	{
+		unsigned int j = first_macroblock + i;
+
+		unsigned int k, l, m;
+
+		batch.macroblocks[i].mbx = macroblocks->macro_blocks[j].x;
+		batch.macroblocks[i].mby = macroblocks->macro_blocks[j].y;
+		batch.macroblocks[i].mb_type = TypeToVL(macroblocks->macro_blocks[j].macroblock_type);
+		if (batch.macroblocks[i].mb_type != vlMacroBlockTypeIntra)
+			batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type, macroblocks->macro_blocks[j].dct_type);
+		batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type == XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded;
+
+		for (k = 0; k < 2; ++k)
+			for (l = 0; l < 2; ++l)
+				for (m = 0; m < 2; ++m)
+					batch.macroblocks[i].PMV[k][l][m] = macroblocks->macro_blocks[j].PMV[k][l][m];
+
+		batch.macroblocks[i].cbp = macroblocks->macro_blocks[j].coded_block_pattern;
+		batch.macroblocks[i].blocks = blocks->blocks + (macroblocks->macro_blocks[j].index * 64);
+	}
+
+	vlRenderMacroBlocksMpeg2(&batch, target_vl_surface);
+
+	return Success;
+}
+
+Status XvMCFlushSurface(Display *display, XvMCSurface *surface)
+{
+	struct vlSurface *vl_sfc;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	vl_sfc = surface->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
+
+	vlSurfaceFlush(vl_sfc);
+
+	return Success;
+}
+
+Status XvMCSyncSurface(Display *display, XvMCSurface *surface)
+{
+	struct vlSurface *vl_sfc;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	vl_sfc = surface->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
+
+	vlSurfaceSync(vl_sfc);
+
+	return Success;
+}
+
+Status XvMCPutSurface
+(
+	Display *display,
+	XvMCSurface *surface,
+	Drawable drawable,
+	short srcx,
+	short srcy,
+	unsigned short srcw,
+	unsigned short srch,
+	short destx,
+	short desty,
+	unsigned short destw,
+	unsigned short desth,
+	int flags
+)
+{
+	Window			root;
+	int			x, y;
+	unsigned int		width, height;
+	unsigned int		border_width;
+	unsigned int		depth;
+	struct vlSurface	*vl_sfc;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	if (XGetGeometry(display, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable)
+		return BadDrawable;
+
+	assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE);
+
+	/* TODO: Correct for negative srcx,srcy & destx,desty by clipping */
+
+	assert(srcx + srcw - 1 < surface->width);
+	assert(srcy + srch - 1 < surface->height);
+	assert(destx + destw - 1 < width);
+	assert(desty + desth - 1 < height);
+
+	vl_sfc = surface->privData;
+
+	vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, width, height, PictureToVL(flags));
+
+	return Success;
+}
+
+Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status)
+{
+	struct vlSurface	*vl_sfc;
+	enum vlResourceStatus	res_status;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	assert(status);
+
+	vl_sfc = surface->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
+
+	vlSurfaceGetStatus(vl_sfc, &res_status);
+
+	switch (res_status)
+	{
+		case vlResourceStatusFree:
+		{
+			*status = 0;
+			break;
+		}
+		case vlResourceStatusRendering:
+		{
+			*status = XVMC_RENDERING;
+			break;
+		}
+		case vlResourceStatusDisplaying:
+		{
+			*status = XVMC_DISPLAYING;
+			break;
+		}
+		default:
+			assert(0);
+	}
+
+	return Success;
+}
+
+Status XvMCDestroySurface(Display *display, XvMCSurface *surface)
+{
+	struct vlSurface *vl_sfc;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	vl_sfc = surface->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
+
+	vlDestroySurface(vl_sfc);
+
+	return Success;
+}
+
+Status XvMCHideSurface(Display *display, XvMCSurface *surface)
+{
+	struct vlSurface *vl_sfc;
+
+	assert(display);
+
+	if (!surface)
+		return XvMCBadSurface;
+
+	vl_sfc = surface->privData;
+
+	assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc))));
+
+	/* No op, only for overlaid rendering */
+
+	return Success;
+}
diff --git a/src/libXvMC/tests/.gitignore b/src/libXvMC/tests/.gitignore
new file mode 100644
index 00000000000..e1d2f9023df
--- /dev/null
+++ b/src/libXvMC/tests/.gitignore
@@ -0,0 +1,5 @@
+test_context
+test_surface
+test_blocks
+test_rendering
+xvmc_bench
diff --git a/src/libXvMC/tests/Makefile b/src/libXvMC/tests/Makefile
new file mode 100644
index 00000000000..de095161d29
--- /dev/null
+++ b/src/libXvMC/tests/Makefile
@@ -0,0 +1,27 @@
+CFLAGS	+= -g -Wall
+LDFLAGS	+=
+LIBS	+= -lXvMCW -lXvMC -lXv
+
+#############################################
+
+.PHONY	= all clean
+
+all: test_context test_surface test_blocks test_rendering xvmc_bench
+
+test_context: test_context.o testlib.o
+	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
+
+test_surface: test_surface.o testlib.o
+	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
+
+test_blocks: test_blocks.o testlib.o
+	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
+
+test_rendering: test_rendering.o testlib.o
+	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
+
+xvmc_bench: xvmc_bench.o testlib.o
+	$(CC) ${LDFLAGS} -o $@ $^ ${LIBS}
+
+clean:
+	rm -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench
diff --git a/src/libXvMC/tests/test_blocks.c b/src/libXvMC/tests/test_blocks.c
new file mode 100644
index 00000000000..dc80adfa652
--- /dev/null
+++ b/src/libXvMC/tests/test_blocks.c
@@ -0,0 +1,84 @@
+#include <assert.h>
+#include <error.h>
+#include "testlib.h"
+
+int main(int argc, char **argv)
+{
+	const unsigned int	width = 16, height = 16;
+	const unsigned int	min_required_blocks = 1, min_required_macroblocks = 1;
+	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
+
+	Display			*display;
+	XvPortID		port_num;
+	int			surface_type_id;
+	unsigned int		is_overlay, intra_unsigned;
+	int			colorkey;
+	XvMCContext		context;
+	XvMCSurface		surface;
+	XvMCBlockArray		blocks = {0};
+	XvMCMacroBlockArray	macroblocks = {0};
+
+	display = XOpenDisplay(NULL);
+
+	if (!GetPort
+	(
+		display,
+		width,
+		height,
+		XVMC_CHROMA_FORMAT_420,
+    		mc_types,
+    		2,
+    		&port_num,
+    		&surface_type_id,
+    		&is_overlay,
+    		&intra_unsigned
+	))
+	{
+		XCloseDisplay(display);
+		error(1, 0, "Error, unable to find a good port.\n");
+	}
+
+	if (is_overlay)
+	{
+		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
+		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
+	}
+
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success);
+	assert(XvMCCreateSurface(display, &context, &surface) == Success);
+
+	/* Test NULL context */
+	assert(XvMCCreateBlocks(display, NULL, 1, &blocks) == XvMCBadContext);
+	/* Test 0 blocks */
+	assert(XvMCCreateBlocks(display, &context, 0, &blocks) == BadValue);
+	/* Test valid params */
+	assert(XvMCCreateBlocks(display, &context, min_required_blocks, &blocks) == Success);
+	/* Test context id assigned and correct */
+	assert(blocks.context_id == context.context_id);
+	/* Test number of blocks assigned and correct */
+	assert(blocks.num_blocks == min_required_blocks);
+	/* Test block pointer valid */
+	assert(blocks.blocks != NULL);
+	/* Test NULL context */
+	assert(XvMCCreateMacroBlocks(display, NULL, 1, &macroblocks) == XvMCBadContext);
+	/* Test 0 macroblocks */
+	assert(XvMCCreateMacroBlocks(display, &context, 0, &macroblocks) == BadValue);
+	/* Test valid params */
+	assert(XvMCCreateMacroBlocks(display, &context, min_required_macroblocks, &macroblocks) == Success);
+	/* Test context id assigned and correct */
+	assert(macroblocks.context_id == context.context_id);
+	/* Test macroblock pointer valid */
+	assert(macroblocks.macro_blocks != NULL);
+	/* Test valid params */
+	assert(XvMCDestroyMacroBlocks(display, &macroblocks) == Success);
+	/* Test valid params */
+	assert(XvMCDestroyBlocks(display, &blocks) == Success);
+
+	assert(XvMCDestroySurface(display, &surface) == Success);
+	assert(XvMCDestroyContext(display, &context) == Success);
+
+	XvUngrabPort(display, port_num, CurrentTime);
+	XCloseDisplay(display);
+
+	return 0;
+}
diff --git a/src/libXvMC/tests/test_context.c b/src/libXvMC/tests/test_context.c
new file mode 100644
index 00000000000..53f7449cd0b
--- /dev/null
+++ b/src/libXvMC/tests/test_context.c
@@ -0,0 +1,92 @@
+#include <assert.h>
+#include <error.h>
+#include "testlib.h"
+
+int main(int argc, char **argv)
+{
+	const unsigned int	width = 16, height = 16;
+	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
+
+	Display			*display;
+	XvPortID		port_num;
+	int			surface_type_id;
+	unsigned int		is_overlay, intra_unsigned;
+	int			colorkey;
+	XvMCContext		context = {0};
+
+	display = XOpenDisplay(NULL);
+
+	if (!GetPort
+	(
+		display,
+		width,
+		height,
+		XVMC_CHROMA_FORMAT_420,
+    		mc_types,
+    		2,
+    		&port_num,
+    		&surface_type_id,
+    		&is_overlay,
+    		&intra_unsigned
+	))
+	{
+		XCloseDisplay(display);
+		error(1, 0, "Error, unable to find a good port.\n");
+	}
+
+	if (is_overlay)
+	{
+		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
+		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
+	}
+
+	/* Test NULL context */
+	/* XXX: XvMCBadContext not a valid return for XvMCCreateContext in the XvMC API, but openChrome driver returns it */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, NULL) == XvMCBadContext);
+	/* Test invalid port */
+	/* XXX: Success and XvBadPort have the same value, if this call actually gets passed the validation step as of now we'll crash later */
+	assert(XvMCCreateContext(display, -1, surface_type_id, width, height, XVMC_DIRECT, &context) == XvBadPort);
+	/* Test invalid surface */
+	assert(XvMCCreateContext(display, port_num, -1, width, height, XVMC_DIRECT, &context) == BadMatch);
+	/* Test invalid flags */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, -1, &context) == BadValue);
+	/* Test huge width */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, 16384, height, XVMC_DIRECT, &context) == BadValue);
+	/* Test huge height */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, 16384, XVMC_DIRECT, &context) == BadValue);
+	/* Test huge width & height */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, 16384, 16384, XVMC_DIRECT, &context) == BadValue);
+	/* Test valid params */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success);
+	/* Test context id assigned */
+	assert(context.context_id != 0);
+	/* Test surface type id assigned and correct */
+	assert(context.surface_type_id == surface_type_id);
+	/* Test width & height assigned and correct */
+	assert(context.width == width && context.height == height);
+	/* Test port assigned and correct */
+	assert(context.port == port_num);
+	/* Test flags assigned and correct */
+	assert(context.flags == XVMC_DIRECT);
+	/* Test NULL context */
+	assert(XvMCDestroyContext(display, NULL) == XvMCBadContext);
+	/* Test valid params */
+	assert(XvMCDestroyContext(display, &context) == Success);
+	/* Test awkward but valid width */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width + 1, height, XVMC_DIRECT, &context) == Success);
+	assert(context.width >= width + 1);
+	assert(XvMCDestroyContext(display, &context) == Success);
+	/* Test awkward but valid height */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height + 1, XVMC_DIRECT, &context) == Success);
+	assert(context.height >= height + 1);
+	assert(XvMCDestroyContext(display, &context) == Success);
+	/* Test awkward but valid width & height */
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width + 1, height + 1, XVMC_DIRECT, &context) == Success);
+	assert(context.width >= width + 1 && context.height >= height + 1);
+	assert(XvMCDestroyContext(display, &context) == Success);
+
+	XvUngrabPort(display, port_num, CurrentTime);
+	XCloseDisplay(display);
+
+	return 0;
+}
diff --git a/src/libXvMC/tests/test_rendering.c b/src/libXvMC/tests/test_rendering.c
new file mode 100644
index 00000000000..1e7467a3aa6
--- /dev/null
+++ b/src/libXvMC/tests/test_rendering.c
@@ -0,0 +1,287 @@
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+#include <error.h>
+#include "testlib.h"
+
+#define BLOCK_WIDTH			8
+#define BLOCK_HEIGHT			8
+#define BLOCK_SIZE			(BLOCK_WIDTH * BLOCK_HEIGHT)
+#define MACROBLOCK_WIDTH		16
+#define MACROBLOCK_HEIGHT		16
+#define MACROBLOCK_WIDTH_IN_BLOCKS	(MACROBLOCK_WIDTH / BLOCK_WIDTH)
+#define MACROBLOCK_HEIGHT_IN_BLOCKS	(MACROBLOCK_HEIGHT / BLOCK_HEIGHT)
+#define BLOCKS_PER_MACROBLOCK		6
+
+#define INPUT_WIDTH			16
+#define INPUT_HEIGHT			16
+#define INPUT_WIDTH_IN_MACROBLOCKS	(INPUT_WIDTH / MACROBLOCK_WIDTH)
+#define INPUT_HEIGHT_IN_MACROBLOCKS	(INPUT_HEIGHT / MACROBLOCK_HEIGHT)
+#define NUM_MACROBLOCKS			(INPUT_WIDTH_IN_MACROBLOCKS * INPUT_HEIGHT_IN_MACROBLOCKS)
+
+#define DEFAULT_OUTPUT_WIDTH		INPUT_WIDTH
+#define DEFAULT_OUTPUT_HEIGHT		INPUT_HEIGHT
+#define DEFAULT_ACCEPTABLE_ERR		0.01
+
+void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt)
+{
+	int fail = 0;
+	int i;
+
+	*output_width = DEFAULT_OUTPUT_WIDTH;
+	*output_height = DEFAULT_OUTPUT_WIDTH;
+	*acceptable_error = DEFAULT_ACCEPTABLE_ERR;
+	*prompt = 1;
+
+	for (i = 1; i < argc && !fail; ++i)
+	{
+		if (!strcmp(argv[i], "-w"))
+		{
+			if (sscanf(argv[++i], "%u", output_width) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-h"))
+		{
+			if (sscanf(argv[++i], "%u", output_height) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-e"))
+		{
+			if (sscanf(argv[++i], "%lf", acceptable_error) != 1)
+				fail = 1;
+		}
+		else if (strcmp(argv[i], "-n"))
+			*prompt = 0;
+		else
+			fail = 1;
+	}
+
+	if (fail)
+		error
+		(
+			1, 0,
+			"Bad argument.\n"
+			"\n"
+			"Usage: %s [options]\n"
+			"\t-w <width>\tOutput width\n"
+			"\t-h <height>\tOutput height\n"
+			"\t-e <error>\tAcceptable margin of error per pixel, from 0 to 1\n"
+			"\t-n\tDon't prompt for quit\n",
+			argv[0]
+		);
+}
+
+void Gradient(short *block, unsigned int start, unsigned int stop, int horizontal)
+{
+	unsigned int x, y;
+	unsigned int range = stop - start;
+
+	if (horizontal)
+	{
+		for (y = 0; y < BLOCK_HEIGHT; ++y)
+			for (x = 0; x < BLOCK_WIDTH; ++x)
+				block[y * BLOCK_WIDTH + x] = (short)(start + range * (x / (float)(BLOCK_WIDTH - 1)));
+	}
+	else
+	{
+		for (y = 0; y < BLOCK_HEIGHT; ++y)
+			for (x = 0; x < BLOCK_WIDTH; ++x)
+				block[y * BLOCK_WIDTH + x] = (short)(start + range * (y / (float)(BLOCK_HEIGHT - 1)));
+	}
+}
+
+int main(int argc, char **argv)
+{
+	unsigned int		output_width;
+	unsigned int		output_height;
+	double			acceptable_error;
+	int			prompt;
+	Display			*display;
+	Window			root, window;
+	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
+	XvPortID		port_num;
+	int			surface_type_id;
+	unsigned int		is_overlay, intra_unsigned;
+	int			colorkey;
+	XvMCContext		context;
+	XvMCSurface		surface;
+	XvMCBlockArray		block_array;
+	XvMCMacroBlockArray	mb_array;
+	int			mbx, mby, bx, by;
+	XvMCMacroBlock		*mb;
+	short			*blocks;
+	int			quit = 0;
+
+	ParseArgs(argc, argv, &output_width, &output_height, &acceptable_error, &prompt);
+
+	display = XOpenDisplay(NULL);
+
+	if (!GetPort
+	(
+		display,
+		INPUT_WIDTH,
+		INPUT_HEIGHT,
+		XVMC_CHROMA_FORMAT_420,
+    		mc_types,
+    		2,
+    		&port_num,
+    		&surface_type_id,
+    		&is_overlay,
+    		&intra_unsigned
+	))
+	{
+		XCloseDisplay(display);
+		error(1, 0, "Error, unable to find a good port.\n");
+	}
+
+	if (is_overlay)
+	{
+		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
+		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
+	}
+
+	root = XDefaultRootWindow(display);
+	window = XCreateSimpleWindow(display, root, 0, 0, output_width, output_height, 0, 0, colorkey);
+
+	assert(XvMCCreateContext(display, port_num, surface_type_id, INPUT_WIDTH, INPUT_HEIGHT, XVMC_DIRECT, &context) == Success);
+	assert(XvMCCreateSurface(display, &context, &surface) == Success);
+	assert(XvMCCreateBlocks(display, &context, NUM_MACROBLOCKS * BLOCKS_PER_MACROBLOCK, &block_array) == Success);
+	assert(XvMCCreateMacroBlocks(display, &context, NUM_MACROBLOCKS, &mb_array) == Success);
+
+	mb = mb_array.macro_blocks;
+	blocks = block_array.blocks;
+
+	for (mby = 0; mby < INPUT_HEIGHT_IN_MACROBLOCKS; ++mby)
+		for (mbx = 0; mbx < INPUT_WIDTH_IN_MACROBLOCKS; ++mbx)
+		{
+			mb->x = mbx;
+			mb->y = mby;
+			mb->macroblock_type = XVMC_MB_TYPE_INTRA;
+			/*mb->motion_type = ;*/
+			/*mb->motion_vertical_field_select = ;*/
+			mb->dct_type = XVMC_DCT_TYPE_FRAME;
+			/*mb->PMV[0][0][0] = ;
+			mb->PMV[0][0][1] = ;
+			mb->PMV[0][1][0] = ;
+			mb->PMV[0][1][1] = ;
+			mb->PMV[1][0][0] = ;
+			mb->PMV[1][0][1] = ;
+			mb->PMV[1][1][0] = ;
+			mb->PMV[1][1][1] = ;*/
+			mb->index = (mby * INPUT_WIDTH_IN_MACROBLOCKS + mbx) * BLOCKS_PER_MACROBLOCK;
+			mb->coded_block_pattern = 0x3F;
+
+			mb++;
+
+			for (by = 0; by < MACROBLOCK_HEIGHT_IN_BLOCKS; ++by)
+				for (bx = 0; bx < MACROBLOCK_WIDTH_IN_BLOCKS; ++bx)
+				{
+					const int start = 16, stop = 235, range = stop - start;
+
+					Gradient
+					(
+						blocks,
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))),
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))),
+						1
+					);
+
+					blocks += BLOCK_SIZE;
+				}
+
+			for (by = 0; by < MACROBLOCK_HEIGHT_IN_BLOCKS / 2; ++by)
+				for (bx = 0; bx < MACROBLOCK_WIDTH_IN_BLOCKS / 2; ++bx)
+				{
+					const int start = 16, stop = 240, range = stop - start;
+
+					Gradient
+					(
+						blocks,
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))),
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))),
+						1
+					);
+
+					blocks += BLOCK_SIZE;
+
+					Gradient
+					(
+						blocks,
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH) / (float)(INPUT_WIDTH - 1))),
+						(short)(start + range * ((mbx * MACROBLOCK_WIDTH + bx * BLOCK_WIDTH + BLOCK_WIDTH - 1) / (float)(INPUT_WIDTH - 1))),
+						1
+					);
+
+					blocks += BLOCK_SIZE;
+				}
+		}
+
+	XSelectInput(display, window, ExposureMask | KeyPressMask);
+	XMapWindow(display, window);
+	XSync(display, 0);
+
+	/* Test NULL context */
+	assert(XvMCRenderSurface(display, NULL, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == XvMCBadContext);
+	/* Test NULL surface */
+	assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, NULL, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == XvMCBadSurface);
+	/* Test bad picture structure */
+	assert(XvMCRenderSurface(display, &context, 0, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == BadValue);
+	/* Test valid params */
+	assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, NUM_MACROBLOCKS, 0, &mb_array, &block_array) == Success);
+
+	/* Test NULL surface */
+	assert(XvMCPutSurface(display, NULL, window, 0, 0, INPUT_WIDTH, INPUT_HEIGHT, 0, 0, output_width, output_height, XVMC_FRAME_PICTURE) == XvMCBadSurface);
+	/* Test bad window */
+	/* XXX: X halts with a bad drawable for some reason, doesn't return BadDrawable as expected */
+	/*assert(XvMCPutSurface(display, &surface, 0, 0, 0, width, height, 0, 0, width, height, XVMC_FRAME_PICTURE) == BadDrawable);*/
+
+	if (prompt)
+	{
+		puts("Press any button to quit...");
+
+		while (!quit)
+		{
+			if (XPending(display) > 0)
+			{
+				XEvent event;
+
+				XNextEvent(display, &event);
+
+				switch (event.type)
+				{
+					case Expose:
+					{
+						/* Test valid params */
+						assert
+						(
+							XvMCPutSurface
+							(
+								display, &surface, window,
+								0, 0, INPUT_WIDTH, INPUT_HEIGHT,
+								0, 0, output_width, output_height,
+								XVMC_FRAME_PICTURE
+							) == Success
+						);
+						break;
+					}
+					case KeyPress:
+					{
+						quit = 1;
+						break;
+					}
+				}
+			}
+		}
+	}
+
+	assert(XvMCDestroyBlocks(display, &block_array) == Success);
+	assert(XvMCDestroyMacroBlocks(display, &mb_array) == Success);
+	assert(XvMCDestroySurface(display, &surface) == Success);
+	assert(XvMCDestroyContext(display, &context) == Success);
+
+	XvUngrabPort(display, port_num, CurrentTime);
+	XDestroyWindow(display, window);
+	XCloseDisplay(display);
+
+	return 0;
+}
diff --git a/src/libXvMC/tests/test_surface.c b/src/libXvMC/tests/test_surface.c
new file mode 100644
index 00000000000..25ebdcc4fcf
--- /dev/null
+++ b/src/libXvMC/tests/test_surface.c
@@ -0,0 +1,72 @@
+#include <assert.h>
+#include <error.h>
+#include "testlib.h"
+
+int main(int argc, char **argv)
+{
+	const unsigned int	width = 16, height = 16;
+	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
+	
+	Display			*display;
+	XvPortID		port_num;
+	int			surface_type_id;
+	unsigned int		is_overlay, intra_unsigned;
+	int			colorkey;
+	XvMCContext		context;
+	XvMCSurface		surface = {0};
+	
+	display = XOpenDisplay(NULL);
+	
+	if (!GetPort
+	(
+		display,
+		width,
+		height,
+		XVMC_CHROMA_FORMAT_420,
+    		mc_types,
+    		2,
+    		&port_num,
+    		&surface_type_id,
+    		&is_overlay,
+    		&intra_unsigned
+	))
+	{
+		XCloseDisplay(display);
+		error(1, 0, "Error, unable to find a good port.\n");
+	}
+	
+	if (is_overlay)
+	{
+		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
+		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
+	}
+	
+	assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success);
+	
+	/* Test NULL context */
+	assert(XvMCCreateSurface(display, NULL, &surface) == XvMCBadContext);
+	/* Test NULL surface */
+	assert(XvMCCreateSurface(display, &context, NULL) == XvMCBadSurface);
+	/* Test valid params */
+	assert(XvMCCreateSurface(display, &context, &surface) == Success);
+	/* Test surface id assigned */
+	assert(surface.surface_id != 0);
+	/* Test context id assigned and correct */
+	assert(surface.context_id == context.context_id);
+	/* Test surface type id assigned and correct */
+	assert(surface.surface_type_id == surface_type_id);
+	/* Test width & height assigned and correct */
+	assert(surface.width == width && surface.height == height);
+	/* Test valid params */
+	assert(XvMCDestroySurface(display, &surface) == Success);
+	/* Test NULL surface */
+	assert(XvMCDestroySurface(display, NULL) == XvMCBadSurface);
+	
+	assert(XvMCDestroyContext(display, &context) == Success);
+	
+	XvUngrabPort(display, port_num, CurrentTime);
+	XCloseDisplay(display);
+	
+	return 0;
+}
+
diff --git a/src/libXvMC/tests/testlib.c b/src/libXvMC/tests/testlib.c
new file mode 100644
index 00000000000..59a03ca813f
--- /dev/null
+++ b/src/libXvMC/tests/testlib.c
@@ -0,0 +1,119 @@
+#include "testlib.h"
+#include <stdio.h>
+
+/*
+void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line)
+{
+	fputs(doc_string, stderr);
+	if (!pred)
+		fprintf(stderr, " FAIL!\n\t\"%s\" at %s:%u\n", pred_string, file, line);
+	else
+		fputs(" PASS!\n", stderr);
+}
+*/
+
+int GetPort
+(
+	Display *display,
+	unsigned int width,
+	unsigned int height,
+	unsigned int chroma_format,
+	const unsigned int *mc_types,
+	unsigned int num_mc_types,
+	XvPortID *port_id,
+	int *surface_type_id,
+	unsigned int *is_overlay,
+	unsigned int *intra_unsigned
+)
+{
+	unsigned int	found_port = 0;
+	XvAdaptorInfo	*adaptor_info;
+	unsigned int	num_adaptors;
+	int		num_types;
+	int		ev_base, err_base;
+	unsigned int	i, j, k, l;
+
+	if (!XvMCQueryExtension(display, &ev_base, &err_base))
+		return 0;
+	if (XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info) != Success)
+		return 0;
+
+	for (i = 0; i < num_adaptors && !found_port; ++i)
+	{
+		if (adaptor_info[i].type & XvImageMask)
+		{
+			XvMCSurfaceInfo *surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types);
+
+			if (surface_info)
+			{
+				for (j = 0; j < num_types && !found_port; ++j)
+				{
+					if
+					(
+						surface_info[j].chroma_format == chroma_format &&
+						surface_info[j].max_width >= width &&
+						surface_info[j].max_height >= height
+					)
+					{
+						for (k = 0; k < num_mc_types && !found_port; ++k)
+						{
+							if (surface_info[j].mc_type == mc_types[k])
+							{
+								for (l = 0; l < adaptor_info[i].num_ports && !found_port; ++l)
+								{
+									if (XvGrabPort(display, adaptor_info[i].base_id + l, CurrentTime) == Success)
+									{
+										*port_id = adaptor_info[i].base_id + l;
+										*surface_type_id = surface_info[j].surface_type_id;
+										*is_overlay = surface_info[j].flags & XVMC_OVERLAID_SURFACE;
+										*intra_unsigned = surface_info[j].flags & XVMC_INTRA_UNSIGNED;
+										found_port = 1;
+									}
+								}
+							}
+						}
+					}
+				}
+
+				XFree(surface_info);
+			}
+		}
+	}
+
+	XvFreeAdaptorInfo(adaptor_info);
+
+	return found_port;
+}
+
+unsigned int align(unsigned int value, unsigned int alignment)
+{
+	return (value + alignment - 1) & ~(alignment - 1);
+}
+
+/* From the glibc manual */
+int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y)
+{
+	/* Perform the carry for the later subtraction by updating y. */
+	if (x->tv_usec < y->tv_usec)
+	{
+		int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1;
+		y->tv_usec -= 1000000 * nsec;
+		y->tv_sec += nsec;
+	}
+	if (x->tv_usec - y->tv_usec > 1000000)
+	{
+		int nsec = (x->tv_usec - y->tv_usec) / 1000000;
+		y->tv_usec += 1000000 * nsec;
+		y->tv_sec -= nsec;
+	}
+
+	/*
+	 * Compute the time remaining to wait.
+	 * tv_usec is certainly positive.
+	 */
+	result->tv_sec = x->tv_sec - y->tv_sec;
+	result->tv_usec = x->tv_usec - y->tv_usec;
+
+	/* Return 1 if result is negative. */
+	return x->tv_sec < y->tv_sec;
+}
diff --git a/src/libXvMC/tests/testlib.h b/src/libXvMC/tests/testlib.h
new file mode 100644
index 00000000000..af71ad74e17
--- /dev/null
+++ b/src/libXvMC/tests/testlib.h
@@ -0,0 +1,42 @@
+#ifndef testlib_h
+#define testlib_h
+
+/*
+#define TEST(pred, doc)	test(pred, #pred, doc, __FILE__, __LINE__)
+
+void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line);
+*/
+
+#include <sys/time.h>
+#include <X11/Xlib.h>
+#include <X11/extensions/XvMClib.h>
+
+/*
+ * display: IN			A valid X display
+ * width, height: IN		Surface size that the port must display
+ * chroma_format: IN		Chroma format that the port must display
+ * mc_types, num_mc_types: IN	List of MC types that the port must support, first port that matches the first mc_type will be returned
+ * port_id: OUT			Your port's ID
+ * surface_type_id: OUT		Your port's surface ID
+ * is_overlay: OUT		If 1, port uses overlay surfaces, you need to set a colorkey
+ * intra_unsigned: OUT		If 1, port uses unsigned values for intra-coded blocks
+ */
+int GetPort
+(
+	Display *display,
+	unsigned int width,
+	unsigned int height,
+	unsigned int chroma_format,
+	const unsigned int *mc_types,
+	unsigned int num_mc_types,
+	XvPortID *port_id,
+	int *surface_type_id,
+	unsigned int *is_overlay,
+	unsigned int *intra_unsigned
+);
+
+unsigned int align(unsigned int value, unsigned int alignment);
+
+int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y);
+
+#endif
diff --git a/src/libXvMC/tests/xvmc_bench.c b/src/libXvMC/tests/xvmc_bench.c
new file mode 100644
index 00000000000..d5a39ecf17d
--- /dev/null
+++ b/src/libXvMC/tests/xvmc_bench.c
@@ -0,0 +1,271 @@
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+#include <error.h>
+#include <sys/time.h>
+#include "testlib.h"
+
+#define MACROBLOCK_WIDTH		16
+#define MACROBLOCK_HEIGHT		16
+#define BLOCKS_PER_MACROBLOCK		6
+
+#define DEFAULT_INPUT_WIDTH		720
+#define DEFAULT_INPUT_HEIGHT		480
+#define DEFAULT_REPS			100
+
+#define PIPELINE_STEP_MC		1
+#define PIPELINE_STEP_CSC		2
+#define PIPELINE_STEP_SWAP		4
+
+#define MB_TYPE_I			1
+#define MB_TYPE_P			2
+#define MB_TYPE_B			4
+
+struct Config
+{
+	unsigned int input_width;
+	unsigned int input_height;
+	unsigned int output_width;
+	unsigned int output_height;
+	unsigned int pipeline;
+	unsigned int mb_types;
+	unsigned int reps;
+};
+
+void ParseArgs(int argc, char **argv, struct Config *config)
+{
+	int fail = 0;
+	int i;
+
+	config->input_width = DEFAULT_INPUT_WIDTH;
+	config->input_height = DEFAULT_INPUT_HEIGHT;
+	config->output_width = 0;
+	config->output_height = 0;
+	config->pipeline = 0;
+	config->mb_types = 0;
+	config->reps = DEFAULT_REPS;
+
+	for (i = 1; i < argc && !fail; ++i)
+	{
+		if (!strcmp(argv[i], "-iw"))
+		{
+			if (sscanf(argv[++i], "%u", &config->input_width) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-ih"))
+		{
+			if (sscanf(argv[++i], "%u", &config->input_height) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-ow"))
+		{
+			if (sscanf(argv[++i], "%u", &config->output_width) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-oh"))
+		{
+			if (sscanf(argv[++i], "%u", &config->output_height) != 1)
+				fail = 1;
+		}
+		else if (!strcmp(argv[i], "-p"))
+		{
+			char *token = strtok(argv[++i], ",");
+
+			while (token && !fail)
+			{
+				if (!strcmp(token, "mc"))
+					config->pipeline |= PIPELINE_STEP_MC;
+				else if (!strcmp(token, "csc"))
+					config->pipeline |= PIPELINE_STEP_CSC;
+				else if (!strcmp(token, "swp"))
+					config->pipeline |= PIPELINE_STEP_SWAP;
+				else
+					fail = 1;
+
+				if (!fail)
+					token = strtok(NULL, ",");
+			}
+		}
+		else if (!strcmp(argv[i], "-mb"))
+		{
+			char *token = strtok(argv[++i], ",");
+
+			while (token && !fail)
+			{
+				if (strcmp(token, "i"))
+					config->mb_types |= MB_TYPE_I;
+				else if (strcmp(token, "p"))
+					config->mb_types |= MB_TYPE_P;
+				else if (strcmp(token, "b"))
+					config->mb_types |= MB_TYPE_B;
+				else
+					fail = 1;
+
+				if (!fail)
+					token = strtok(NULL, ",");
+			}
+		}
+		else if (!strcmp(argv[i], "-r"))
+		{
+			if (sscanf(argv[++i], "%u", &config->reps) != 1)
+				fail = 1;
+		}
+		else
+			fail = 1;
+	}
+
+	if (fail)
+		error
+		(
+			1, 0,
+			"Bad argument.\n"
+			"\n"
+			"Usage: %s [options]\n"
+			"\t-iw <width>\tInput width\n"
+			"\t-ih <height>\tInput height\n"
+			"\t-ow <width>\tOutput width\n"
+			"\t-oh <height>\tOutput height\n"
+			"\t-p <pipeline>\tPipeline to test\n"
+			"\t-mb <mb type>\tMacroBlock types to use\n"
+			"\t-r <reps>\tRepetitions\n\n"
+			"\tPipeline steps: mc,csc,swap\n"
+			"\tMB types: i,p,b\n",
+			argv[0]
+		);
+
+	if (config->output_width == 0)
+		config->output_width = config->input_width;
+	if (config->output_height == 0)
+		config->output_height = config->input_height;
+	if (!config->pipeline)
+		config->pipeline = PIPELINE_STEP_MC | PIPELINE_STEP_CSC | PIPELINE_STEP_SWAP;
+	if (!config->mb_types)
+		config->mb_types = MB_TYPE_I | MB_TYPE_P | MB_TYPE_B;
+}
+
+int main(int argc, char **argv)
+{
+	struct Config		config;
+	Display			*display;
+	Window			root, window;
+	const unsigned int	mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2};
+	XvPortID		port_num;
+	int			surface_type_id;
+	unsigned int		is_overlay, intra_unsigned;
+	int			colorkey;
+	XvMCContext		context;
+	XvMCSurface		surface;
+	XvMCBlockArray		block_array;
+	XvMCMacroBlockArray	mb_array;
+	unsigned int		mbw, mbh;
+	unsigned int		mbx, mby;
+	unsigned int		reps;
+	struct timeval		start, stop, diff;
+	double			diff_secs;
+
+	ParseArgs(argc, argv, &config);
+
+	mbw = align(config.input_width, MACROBLOCK_WIDTH) / MACROBLOCK_WIDTH;
+	mbh = align(config.input_height, MACROBLOCK_HEIGHT) / MACROBLOCK_HEIGHT;
+
+	display = XOpenDisplay(NULL);
+
+	if (!GetPort
+	(
+		display,
+		config.input_width,
+		config.input_height,
+		XVMC_CHROMA_FORMAT_420,
+    		mc_types,
+    		2,
+    		&port_num,
+    		&surface_type_id,
+    		&is_overlay,
+    		&intra_unsigned
+	))
+	{
+		XCloseDisplay(display);
+		error(1, 0, "Error, unable to find a good port.\n");
+	}
+
+	if (is_overlay)
+	{
+		Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0);
+		XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey);
+	}
+
+	root = XDefaultRootWindow(display);
+	window = XCreateSimpleWindow(display, root, 0, 0, config.output_width, config.output_height, 0, 0, colorkey);
+
+	assert(XvMCCreateContext(display, port_num, surface_type_id, config.input_width, config.input_height, XVMC_DIRECT, &context) == Success);
+	assert(XvMCCreateSurface(display, &context, &surface) == Success);
+	assert(XvMCCreateBlocks(display, &context, mbw * mbh * BLOCKS_PER_MACROBLOCK, &block_array) == Success);
+	assert(XvMCCreateMacroBlocks(display, &context, mbw * mbh, &mb_array) == Success);
+
+	for (mby = 0; mby < mbh; ++mby)
+		for (mbx = 0; mbx < mbw; ++mbx)
+		{
+			mb_array.macro_blocks[mby * mbw + mbx].x = mbx;
+			mb_array.macro_blocks[mby * mbw + mbx].y = mby;
+			mb_array.macro_blocks[mby * mbw + mbx].macroblock_type = XVMC_MB_TYPE_INTRA;
+			/*mb->motion_type = ;*/
+			/*mb->motion_vertical_field_select = ;*/
+			mb_array.macro_blocks[mby * mbw + mbx].dct_type = XVMC_DCT_TYPE_FRAME;
+			/*mb->PMV[0][0][0] = ;
+			mb->PMV[0][0][1] = ;
+			mb->PMV[0][1][0] = ;
+			mb->PMV[0][1][1] = ;
+			mb->PMV[1][0][0] = ;
+			mb->PMV[1][0][1] = ;
+			mb->PMV[1][1][0] = ;
+			mb->PMV[1][1][1] = ;*/
+			mb_array.macro_blocks[mby * mbw + mbx].index = (mby * mbw + mbx) * BLOCKS_PER_MACROBLOCK;
+			mb_array.macro_blocks[mby * mbw + mbx].coded_block_pattern = 0x3F;
+		}
+
+	XSelectInput(display, window, ExposureMask | KeyPressMask);
+	XMapWindow(display, window);
+	XSync(display, 0);
+
+	gettimeofday(&start, NULL);
+
+	for (reps = 0; reps < config.reps; ++reps)
+	{
+		if (config.pipeline & PIPELINE_STEP_MC)
+		{
+			assert(XvMCRenderSurface(display, &context, XVMC_FRAME_PICTURE, &surface, NULL, NULL, 0, mbw * mbh, 0, &mb_array, &block_array) == Success);
+			assert(XvMCFlushSurface(display, &surface) == Success);
+		}
+		if (config.pipeline & PIPELINE_STEP_CSC)
+			assert(XvMCPutSurface(display, &surface, window, 0, 0, config.input_width, config.input_height, 0, 0, config.output_width, config.output_height, XVMC_FRAME_PICTURE) == Success);
+	}
+
+	gettimeofday(&stop, NULL);
+
+	timeval_subtract(&diff, &stop, &start);
+	diff_secs = (double)diff.tv_sec + (double)diff.tv_usec / 1000000.0;
+
+	printf("XvMC Benchmark\n");
+	printf("Input: %u,%u\nOutput: %u,%u\n", config.input_width, config.input_height, config.output_width, config.output_height);
+	printf("Pipeline: ");
+	if (config.pipeline & PIPELINE_STEP_MC)
+		printf("|mc|");
+	if (config.pipeline & PIPELINE_STEP_CSC)
+		printf("|csc|");
+	if (config.pipeline & PIPELINE_STEP_SWAP)
+		printf("|swap|");
+	printf("\n");
+	printf("Reps: %u\n", config.reps);
+	printf("Total time: %.2lf (%.2lf reps / sec)\n", diff_secs, config.reps / diff_secs);
+
+	assert(XvMCDestroyBlocks(display, &block_array) == Success);
+	assert(XvMCDestroyMacroBlocks(display, &mb_array) == Success);
+	assert(XvMCDestroySurface(display, &surface) == Success);
+	assert(XvMCDestroyContext(display, &context) == Success);
+
+	XvUngrabPort(display, port_num, CurrentTime);
+	XDestroyWindow(display, window);
+	XCloseDisplay(display);
+
+	return 0;
+}
diff --git a/src/mesa/Makefile b/src/mesa/Makefile
index cd42cd654b0..97b011a7d45 100644
--- a/src/mesa/Makefile
+++ b/src/mesa/Makefile
@@ -5,7 +5,6 @@ include $(TOP)/configs/current
 
 include sources.mak
 
-
 .SUFFIXES : .cpp
 
 .c.o:
@@ -19,21 +18,25 @@ include sources.mak
 
 
 
-GLAPI_LIB = libglapi.a
+# Default: build dependencies, then asm_subdirs, then convenience
+# libs (.a) and finally the device drivers:
+default: depend asm_subdirs libmesa.a libglapi.a driver_subdirs
 
 
 # Default: build dependencies, then asm_subdirs, then convenience
 # libs (.a) and finally the device drivers:
 default: depend asm_subdirs libmesa.a $(GLAPI_LIB) driver_subdirs
 
-
-
 ######################################################################
 # Helper libraries used by many drivers:
 
 # Make archive of core mesa object files
 libmesa.a: $(MESA_OBJECTS)
-	@ $(TOP)/bin/mklib -o mesa -static $(MESA_OBJECTS)
+	@ $(MKLIB) -o mesa -static $(MESA_OBJECTS)
+
+# Make archive of gl* API dispatcher functions only
+libglapi.a: $(GLAPI_OBJECTS)
+	@ $(MKLIB) -o glapi -static $(GLAPI_OBJECTS)
 
 # Make archive of gl* API dispatcher functions only
 $(GLAPI_LIB): $(GLAPI_OBJECTS)
@@ -43,10 +46,9 @@ $(GLAPI_LIB): $(GLAPI_OBJECTS)
 		$(TOP)/bin/mklib -o glapi -static $(GLAPI_OBJECTS) ; \
 	fi
 
-
 ######################################################################
 # Device drivers
-driver_subdirs: libmesa.a $(GLAPI_LIB)
+driver_subdirs: libmesa.a libglapi.a
 	(cd drivers && $(MAKE))
 
 
@@ -95,30 +97,48 @@ install: default
 	done
 
 
-pcedit = sed \
+gl_pcedit = sed \
 	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
-	-e 's,@LIB_DIR@,$(LIB_DIR),' \
-	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),'
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
+	-e 's,@GL_PC_REQ_PRIV@,$(GL_PC_REQ_PRIV),' \
+	-e 's,@GL_PC_LIB_PRIV@,$(GL_PC_LIB_PRIV),' \
+	-e 's,@GL_PC_CFLAGS@,$(GL_PC_CFLAGS),'
 
 gl.pc: gl.pc.in
-	$(pcedit) $< > $@
+	$(gl_pcedit) $< > $@
+
+osmesa_pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
+	-e 's,@OSMESA_LIB@,$(OSMESA_LIB),' \
+	-e 's,@OSMESA_PC_REQ@,$(OSMESA_PC_REQ),' \
+	-e 's,@OSMESA_PC_LIB_PRIV@,$(OSMESA_PC_LIB_PRIV),'
+
+osmesa.pc: osmesa.pc.in
+	$(osmesa_pcedit) $< > $@
 
 install-headers:
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/include/GL
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL
 	$(INSTALL) -m 644 $(TOP)/include/GL/*.h \
-		$(DESTDIR)$(INSTALL_DIR)/include/GL
+		$(DESTDIR)$(INSTALL_INC_DIR)/GL
 
 install-libgl: default gl.pc install-headers
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)/pkgconfig
-	$(INSTALL) $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)* \
-		$(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) -m 644 gl.pc $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)/pkgconfig
-
-install-osmesa: default
-	$(INSTALL) -d $(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
-	$(INSTALL) $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME)* \
-		$(DESTDIR)$(INSTALL_DIR)/$(LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) $(TOP)/$(LIB_DIR)/$(GL_LIB_GLOB) \
+		$(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 gl.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+
+install-osmesa: default osmesa.pc
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_GLOB) \
+		$(DESTDIR)$(INSTALL_LIB_DIR)
+	$(INSTALL) -m 644 osmesa.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 install-dri: default
 	cd drivers/dri && $(MAKE) install
@@ -133,10 +153,10 @@ tags:
 clean:
 	-rm -f */*.o
 	-rm -f */*/*.o
-	-rm -f depend depend.bak libmesa.a $(GLAPI_LIB)
+	-rm -f depend depend.bak libmesa.a libglapi.a
 	-rm -f drivers/*/*.o
+	-rm -f *.pc
 	-@cd drivers/dri && $(MAKE) clean
-	-@cd drivers/xorg && $(MAKE) clean
 	-@cd drivers/x11 && $(MAKE) clean
 	-@cd drivers/osmesa && $(MAKE) clean
 	-@cd x86 && $(MAKE) clean
diff --git a/src/mesa/Makefile.mgw b/src/mesa/Makefile.mgw
new file mode 100644
index 00000000000..097c390a83e
--- /dev/null
+++ b/src/mesa/Makefile.mgw
@@ -0,0 +1,235 @@
+# Mesa 3-D graphics library
+# Version:  7.0
+# 
+# Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+# 
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+# 
+# The above copyright notice and this permission notice shall be included
+# in all copies or substantial portions of the Software.
+# 
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+# OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+# MinGW core makefile v1.4 for Mesa
+#
+#  Copyright (C) 2002 - Daniel Borca
+#  Email : dborca@users.sourceforge.net
+#  Web   : http://www.geocities.com/dborca
+
+# MinGW core-gl makefile updated for Mesa 7.0
+#
+#  updated : by Heromyth, on 2007-7-21
+#  Email   : zxpmyth@yahoo.com.cn
+#  Bugs    : 1) All the default settings work fine. But the setting X86=1 can't work. 
+#            The others havn't been tested yet.
+#            2) The generated DLLs are *not* compatible with the ones built
+#            with the other compilers like VC8, especially for GLUT. 
+#            3) Although more tests are needed, it can be used individually!
+
+
+#
+#  Available options:
+#
+#    Environment variables:
+#	CFLAGS
+#
+#	GLIDE		path to Glide3 SDK; used with FX.
+#			default = $(TOP)/glide3
+#	FX=1		build for 3dfx Glide3. Note that this disables
+#			compilation of most WMesa code and requires fxMesa.
+#			As a consequence, you'll need the Win32 Glide3
+#			library to build any application.
+#			default = no
+#	ICD=1		build the installable client driver interface
+#			(windows opengl driver interface)
+#			default = no
+#	X86=1		optimize for x86 (if possible, use MMX, SSE, 3DNow).
+#			default = no
+#
+#    Targets:
+#	all:		build GL
+#	clean:		remove object files
+#
+
+
+.PHONY: all clean
+.INTERMEDIATE: x86/gen_matypes.exe
+.SUFFIXES: .rc .res
+
+# Set this to the prefix of your build tools, i.e. mingw32-
+TOOLS_PREFIX = mingw32-
+
+ifeq ($(ICD),1)
+  LIB_NAME = mesa32
+else
+  LIB_NAME = opengl32
+endif
+
+DLL_EXT = .dll
+IMP_EXT = .a
+LIB_PRE = lib
+STRIP = -s
+
+AR = ar
+ARFLAGS = crus
+DLLTOOL = dlltool
+
+TOP = ../..
+GLIDE ?= $(TOP)/glide3
+LIBDIR = $(TOP)/lib
+
+GL_DLL = $(LIB_NAME)$(DLL_EXT)
+GL_IMP = $(LIB_PRE)$(LIB_NAME)$(IMP_EXT)
+
+MESA_LIB = libmesa.a
+CC = $(TOOLS_PREFIX)gcc
+
+LDLIBS = -lgdi32 -luser32 -liberty
+LDFLAGS = $(STRIP) -shared -fPIC -Wl,--kill-at
+
+CFLAGS += -DBUILD_GL32 -D_DLL -DMESA_MINWARN
+CFLAGS += -DNDEBUG -DUSE_EXTERNAL_DXTN_LIB=1
+
+ifeq ($(FX),1)
+  CFLAGS += -I$(GLIDE)/include -DFX
+  LDLIBS += -L$(GLIDE)/lib -lglide3x
+  GL_DEF = drivers/windows/fx/fxopengl.def
+  GL_RES = drivers/windows/fx/fx.rc
+else
+  ifeq ($(ICD),1)
+    CFLAGS += -DUSE_MGL_NAMESPACE
+    GL_DEF = drivers/windows/icd/mesa.def
+  else    
+    GL_DEF = $(LIB_NAME).def
+  endif
+endif
+
+
+
+UNLINK = del $(subst /,\,$(1))
+ifneq ($(wildcard $(addsuffix /rm.exe,$(subst ;, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+ifneq ($(wildcard $(addsuffix /rm,$(subst :, ,$(PATH)))),)
+UNLINK = $(RM) $(1)
+endif
+
+include sources.mak
+
+CFLAGS += $(INCLUDE_DIRS) 
+
+ifeq ($(X86),1)
+CFLAGS += -DUSE_X86_ASM
+CFLAGS += -DUSE_MMX_ASM
+CFLAGS += -DUSE_SSE_ASM
+CFLAGS += -DUSE_3DNOW_ASM
+X86_SOURCES += $(X86_API)
+else
+X86_SOURCES =
+endif
+
+ifeq ($(FX),1)
+DRIVER_SOURCES = \
+	$(GLIDE_DRIVER_SOURCES) \
+	drivers/windows/fx/fxwgl.c
+else
+ifeq ($(ICD),1)
+DRIVER_SOURCES = \
+	drivers/windows/gdi/wmesa.c \
+	drivers/windows/icd/icd.c
+else
+DRIVER_SOURCES = \
+	drivers/windows/gdi/wmesa.c \
+	drivers/windows/gdi/wgl.c
+endif
+endif
+
+SOURCES = $(MESA_SOURCES) $(GLAPI_SOURCES) $(X86_SOURCES) $(DRIVER_SOURCES)
+
+OBJECTS = $(addsuffix .o,$(basename $(SOURCES)))
+
+X86_OBJECTS = $(addsuffix .o,$(basename $(X86_SOURCES)))
+
+RESOURCE = $(GL_RES:.rc=.res)
+
+.c.o:
+	$(CC) -o $@ $(CFLAGS) -c $<
+.s.o:
+	$(CC) -o $@ $(CFLAGS) -x assembler-with-cpp -c $<
+
+.rc.res:
+	windres -o $@ -Irc -Ocoff $<
+
+
+all: $(LIBDIR) $(LIBDIR)/$(GL_DLL) $(LIBDIR)/$(GL_IMP)
+
+$(LIBDIR):
+	mkdir -p $(LIBDIR)
+
+$(LIBDIR)/$(GL_DLL) $(LIBDIR)/$(GL_IMP): $(OBJECTS) $(RESOURCE)
+	$(CC) $(LDFLAGS) -o $(LIBDIR)/$(GL_DLL) $^ $(LDLIBS)
+	$(DLLTOOL) --as=as --dllname $(LIB_NAME) --output-def $(LIBDIR)/$(GL_DEF) $^
+	$(DLLTOOL) --as=as -k --dllname $(LIB_NAME) --output-lib $(LIBDIR)/$(GL_IMP) --def $(LIBDIR)/$(GL_DEF)
+
+
+$(X86_OBJECTS): x86/matypes.h
+
+x86/matypes.h: x86/gen_matypes.exe
+	$(subst /,\,$< > $@)
+
+x86/gen_matypes.exe: x86/gen_matypes.c
+	$(CC) -o $@ $(CFLAGS) -s $<
+
+# [dBorca]
+# glapi_x86.S needs some adjustments
+# in order to generate correct entrypoints
+# Trick: change the following condition to
+# be always false if you need C entrypoints
+# with USE_X86_ASM (useful for trace/debug)
+ifeq (1,1)
+x86/glapi_x86.o: x86/glapi_x86.S
+	$(CC) -o $@ $(CFLAGS) -DSTDCALL_API -c $<
+else
+main/dispatch.o: main/dispatch.c
+	$(CC) -o $@ $(CFLAGS) -UUSE_X86_ASM -c $<
+glapi/glapi.o: glapi/glapi.c
+	$(CC) -o $@ $(CFLAGS) -UUSE_X86_ASM -c $<
+endif
+
+# [dBorca]
+# if we want codegen, we have to stdcall
+tnl/t_vtx_x86_gcc.o: tnl/t_vtx_x86_gcc.S
+	$(CC) -o $@ $(CFLAGS) -DSTDCALL_API -c $<
+
+clean:
+	-$(call UNLINK,glapi/*.o)
+	-$(call UNLINK,main/*.o)
+	-$(call UNLINK,math/*.o)
+	-$(call UNLINK,vbo/*.o)
+	-$(call UNLINK,shader/*.o)
+	-$(call UNLINK,shader/slang/*.o)
+	-$(call UNLINK,shader/grammar/*.o)
+	-$(call UNLINK,sparc/*.o)
+	-$(call UNLINK,ppc/*.o)
+	-$(call UNLINK,swrast/*.o)
+	-$(call UNLINK,swrast_setup/*.o)
+	-$(call UNLINK,tnl/*.o)
+	-$(call UNLINK,x86/*.o)
+	-$(call UNLINK,x86/rtasm/*.o)
+	-$(call UNLINK,x86-64/*.o)
+	-$(call UNLINK,drivers/common/*.o)
+	-$(call UNLINK,drivers/glide/*.o)
+	-$(call UNLINK,drivers/windows/fx/*.o)
+	-$(call UNLINK,drivers/windows/fx/*.res)
+	-$(call UNLINK,drivers/windows/gdi/*.o)
+	-$(call UNLINK,drivers/windows/icd/*.o)
diff --git a/src/mesa/SConscript b/src/mesa/SConscript
index a690d576b15..f4e138c8ab1 100644
--- a/src/mesa/SConscript
+++ b/src/mesa/SConscript
@@ -283,6 +283,10 @@ if env['platform'] != 'winddk':
 			'x86-64/glapi_x86-64.S'
 		]
 	elif gcc and env['machine'] == 'ppc':
+		env.Append(CPPDEFINES = [
+			'USE_PPC_ASM', 
+			'USE_VMX_ASM', 
+		])
 		mesa_sources += [
 			'ppc/common_ppc.c',
 		]
diff --git a/src/mesa/descrip.mms b/src/mesa/descrip.mms
new file mode 100644
index 00000000000..dbfa336a912
--- /dev/null
+++ b/src/mesa/descrip.mms
@@ -0,0 +1,27 @@
+# Makefile for Mesa for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Date last revision : 7 March 2007
+
+all :
+	set default [.main]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.glapi]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.shader]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.drivers.common]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.x11]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.osmesa]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [--.math]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.tnl]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.swrast]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.swrast_setup]
+	$(MMS)$(MMSQUALIFIERS)
+	set default [-.vbo]
+	$(MMS)$(MMSQUALIFIERS)
diff --git a/src/mesa/drivers/allegro/amesa.c b/src/mesa/drivers/allegro/amesa.c
index c12d42148b6..ade62518489 100644
--- a/src/mesa/drivers/allegro/amesa.c
+++ b/src/mesa/drivers/allegro/amesa.c
@@ -21,11 +21,11 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <allegro.h>
-#include "buffers.h"
-#include "context.h"
-#include "imports.h"
-#include "matrix.h"
-#include "mtypes.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/mtypes.h"
 #include "GL/amesa.h"
 
 
diff --git a/src/mesa/drivers/beos/Makefile b/src/mesa/drivers/beos/Makefile
index 342d7ce0243..c79dd24c390 100644
--- a/src/mesa/drivers/beos/Makefile
+++ b/src/mesa/drivers/beos/Makefile
@@ -181,9 +181,10 @@ $(TOP)/$(LIB_DIR):
 	mkdir $(TOP)/$(LIB_DIR)
 
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(OBJECTS) $(MESA_MODULES) $(GLU_MODULES)
-	@$(TOP)/bin/mklib -o $(GL_LIB) -ldflags '$(LDFLAGS)' -install $(TOP)/$(LIB_DIR) \
+	@$(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
-		$(MKLIB_OPTIONS) $(GL_LIB_DEPS) $(OBJECTS) $(MESA_MODULES) $(GLU_MODULES)
+		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) $(GL_LIB_DEPS) \
+		$(OBJECTS) $(MESA_MODULES) $(GLU_MODULES)
 
 # $(GLU_OBJECTS):
 #	cd $(GLU_DIR) && $(MAKE) $< ;
diff --git a/src/mesa/drivers/common/descrip.mms b/src/mesa/drivers/common/descrip.mms
new file mode 100644
index 00000000000..d5bbc69dfd5
--- /dev/null
+++ b/src/mesa/drivers/common/descrip.mms
@@ -0,0 +1,42 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 29 September 2008
+
+.first
+	define gl [----.include.gl]
+	define math [--.math]
+	define tnl [--.tnl]
+	define swrast [--.swrast]
+	define glapi [--.glapi]
+	define shader [--.shader]
+	define main [--.main]
+
+.include [----]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [----.include],[--.main],[--.glapi],[--.shader]
+LIBDIR = [----.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)\
+	/float=ieee/ieee=denorm/warn=disable=(PTRMISMATCH)
+
+SOURCES = driverfuncs.c
+
+OBJECTS =driverfuncs.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+driverfuncs.obj : driverfuncs.c
diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c
index b5b383b4e42..986f751bdc1 100644
--- a/src/mesa/drivers/common/driverfuncs.c
+++ b/src/mesa/drivers/common/driverfuncs.c
@@ -23,35 +23,36 @@
  */
 
 
-#include "glheader.h"
-#include "imports.h"
-#include "buffers.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "mipmap.h"
-#include "program.h"
-#include "prog_execute.h"
-#include "queryobj.h"
-#include "renderbuffer.h"
-#include "texcompress.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texobj.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/arrayobj.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "main/mipmap.h"
+#include "main/queryobj.h"
+#include "main/renderbuffer.h"
+#include "main/texcompress.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "main/texstore.h"
 #if FEATURE_ARB_vertex_buffer_object
-#include "bufferobj.h"
+#include "main/bufferobj.h"
 #endif
 #if FEATURE_EXT_framebuffer_object
-#include "fbobject.h"
-#include "texrender.h"
+#include "main/fbobject.h"
+#include "main/texrender.h"
 #endif
-#include "shader_api.h"
-#include "arrayobj.h"
 
-#include "driverfuncs.h"
+#include "shader/program.h"
+#include "shader/prog_execute.h"
+#include "shader/shader_api.h"
 #include "tnl/tnl.h"
 #include "swrast/swrast.h"
 
+#include "driverfuncs.h"
+
 
 
 /**
@@ -115,6 +116,8 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->DeleteTexture = _mesa_delete_texture_object;
    driver->NewTextureImage = _mesa_new_texture_image;
    driver->FreeTexImageData = _mesa_free_texture_image_data; 
+   driver->MapTexture = NULL;
+   driver->UnmapTexture = NULL;
    driver->TextureMemCpy = _mesa_memcpy; 
    driver->IsTextureResident = NULL;
    driver->PrioritizeTexture = NULL;
diff --git a/src/mesa/drivers/common/sources b/src/mesa/drivers/common/sources
deleted file mode 100644
index 90e29d78d3f..00000000000
--- a/src/mesa/drivers/common/sources
+++ /dev/null
@@ -1,2 +0,0 @@
-MESA_DRIVER_COMMON_SOURCES = \
-driverfuncs.c
diff --git a/src/mesa/drivers/directfb/Makefile b/src/mesa/drivers/directfb/Makefile
index ece0457cffa..94c82a2c9c9 100644
--- a/src/mesa/drivers/directfb/Makefile
+++ b/src/mesa/drivers/directfb/Makefile
@@ -37,7 +37,7 @@ default: directfb-libgl directfbgl_mesa
 
 # XXX this used to be in src/mesa/Makefile and is probably broken now
 directfb-libgl: $(LIBS)
-	@ $(TOP)/bin/mklib -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+	@ $(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
 		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) $(LIBS) \
 		$(GL_LIB_DEPS)
diff --git a/src/mesa/drivers/directfb/idirectfbgl_mesa.c b/src/mesa/drivers/directfb/idirectfbgl_mesa.c
index 694eeb054df..93593403c13 100644
--- a/src/mesa/drivers/directfb/idirectfbgl_mesa.c
+++ b/src/mesa/drivers/directfb/idirectfbgl_mesa.c
@@ -36,16 +36,16 @@
 #include <direct/interface.h>
 
 #undef CLAMP
-#include "glheader.h"
-#include "buffers.h"
-#include "context.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "imports.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
+#include "main/imports.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -102,7 +102,7 @@ typedef struct {
 static pthread_mutex_t global_lock = PTHREAD_MUTEX_INITIALIZER;
 static unsigned int    global_ref  = 0;
 
-static inline int directfbgl_init( void )
+static INLINE int directfbgl_init( void )
 {
      pthread_mutexattr_t attr;
      int                 ret;
@@ -118,7 +118,7 @@ static inline int directfbgl_init( void )
      return ret;
 }
 
-static inline void directfbgl_finish( void )
+static INLINE void directfbgl_finish( void )
 {
      if (--global_ref == 0)
           pthread_mutex_destroy( &global_lock );
@@ -503,10 +503,31 @@ dfbRenderbufferStorage( GLcontext *ctx, struct gl_renderbuffer *render,
           (((S[GCOMP]) & 0xf0)     ) | \
           (((S[BCOMP]) & 0xf0) >> 4) )
 #define FETCH_PIXEL(D, P) \
-   D[RCOMP] = ((*P & 0x0f00) >> 4); \
-   D[GCOMP] = ((*P & 0x00f0)     ); \
-   D[BCOMP] = ((*P & 0x000f) << 4); \
-   D[ACOMP] = ((*P & 0xf000) >> 8)
+   D[RCOMP] = ((*P & 0x0f00) >> 4) | ((*P & 0x0f00) >>  8); \
+   D[GCOMP] = ((*P & 0x00f0)     ) | ((*P & 0x00f0) >>  4); \
+   D[BCOMP] = ((*P & 0x000f) << 4) | ((*P & 0x000f)      ); \
+   D[ACOMP] = ((*P & 0xf000) >> 8) | ((*P & 0xf000) >> 12)
+
+#include "swrast/s_spantemp.h"
+
+/* RGB444 */
+#define NAME(PREFIX) PREFIX##_RGB444
+#define FORMAT GL_RGBA8
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   IDirectFBGL_data *data = (IDirectFBGL_data*) ctx->DriverCtx;
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLushort *P = (GLushort *) (data->video.end - (Y) * data->video.pitch + (X) * 2);
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(P, X, Y, S) \
+   *P = ( (((S[RCOMP]) & 0xf0) << 4) | \
+          (((S[GCOMP]) & 0xf0)     ) | \
+          (((S[BCOMP]) & 0xf0) >> 4) )
+#define FETCH_PIXEL(D, P) \
+   D[RCOMP] = ((*P & 0x0f00) >> 4) | ((*P & 0x0f00) >> 8); \
+   D[GCOMP] = ((*P & 0x00f0)     ) | ((*P & 0x00f0) >> 4); \
+   D[BCOMP] = ((*P & 0x000f) << 4) | ((*P & 0x000f)     ); \
+   D[ACOMP] = 0xff
 
 #include "swrast/s_spantemp.h"
 
@@ -557,13 +578,34 @@ dfbRenderbufferStorage( GLcontext *ctx, struct gl_renderbuffer *render,
           (((S[GCOMP]) & 0xf8) <<  2) | \
           (((S[BCOMP])       ) >>  3) )
 #define FETCH_PIXEL(D, P) \
-   D[RCOMP] = ((*P & 0x7c00) >> 7); \
-   D[GCOMP] = ((*P & 0x03e0) >> 2); \
-   D[BCOMP] = ((*P & 0x001f) << 3); \
+   D[RCOMP] = ((*P & 0x7c00) >> 7) | ((*P & 0x7c00) >> 12); \
+   D[GCOMP] = ((*P & 0x03e0) >> 2) | ((*P & 0x03e0) >>  7); \
+   D[BCOMP] = ((*P & 0x001f) << 3) | ((*P & 0x001f) <<  2); \
    D[ACOMP] = ((*P & 0x8000) ? 0xff : 0)
 
 #include "swrast/s_spantemp.h"
 
+/* RGB555 */
+#define NAME(PREFIX) PREFIX##_RGB555
+#define FORMAT GL_RGBA8
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   IDirectFBGL_data *data = (IDirectFBGL_data*) ctx->DriverCtx;
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLushort *P = (GLushort *) (data->video.end - (Y) * data->video.pitch + (X) * 2);
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(P, X, Y, S) \
+   *P = ( (((S[RCOMP]) & 0xf8) <<  7) | \
+          (((S[GCOMP]) & 0xf8) <<  2) | \
+          (((S[BCOMP])       ) >>  3) )
+#define FETCH_PIXEL(D, P) \
+   D[RCOMP] = ((*P & 0x7c00) >> 7) | ((*P & 0x7c00) >> 12); \
+   D[GCOMP] = ((*P & 0x03e0) >> 2) | ((*P & 0x03e0) >>  7); \
+   D[BCOMP] = ((*P & 0x001f) << 3) | ((*P & 0x001f) <<  2); \
+   D[ACOMP] = 0xff
+
+#include "swrast/s_spantemp.h"
+
 /* RGB16 */
 #define NAME(PREFIX) PREFIX##_RGB16
 #define FORMAT GL_RGBA8
@@ -578,9 +620,9 @@ dfbRenderbufferStorage( GLcontext *ctx, struct gl_renderbuffer *render,
           (((S[GCOMP]) & 0xfc) << 3) | \
           (((S[BCOMP])       ) >> 3) )
 #define FETCH_PIXEL(D, P) \
-   D[RCOMP] = ((*P & 0xf800) >> 8); \
-   D[GCOMP] = ((*P & 0x07e0) >> 3); \
-   D[BCOMP] = ((*P & 0x001f) << 3); \
+   D[RCOMP] = ((*P & 0xf800) >> 8) | ((*P & 0xf800) >> 13); \
+   D[GCOMP] = ((*P & 0x07e0) >> 3) | ((*P & 0x07e0) >>  9); \
+   D[BCOMP] = ((*P & 0x001f) << 3) | ((*P & 0x001f) >>  2); \
    D[ACOMP] = 0xff
 
 #include "swrast/s_spantemp.h"
@@ -706,22 +748,24 @@ directfbgl_init_visual( GLvisual              *visual,
                blueBits  = 2;
                break;
           case DSPF_ARGB4444:
+               alphaBits = 4;
+          case DSPF_RGB444:
                redBits   = 4;
                greenBits = 4;
                blueBits  = 4;
-               alphaBits = 4;
                break;
           case DSPF_ARGB2554:
+               alphaBits = 2;
                redBits   = 5;
                greenBits = 5;
                blueBits  = 4;
-               alphaBits = 2;
                break;
           case DSPF_ARGB1555:
+               alphaBits = 1;
+          case DSPF_RGB555:
                redBits   = 5;
                greenBits = 5;
                blueBits  = 5;
-               alphaBits = 1;
                break;
           case DSPF_RGB16:
                redBits   = 5;
@@ -818,6 +862,15 @@ directfbgl_create_context( GLcontext        *context,
                data->render.PutValues     = put_values_ARGB4444;
                data->render.PutMonoValues = put_mono_values_ARGB4444;
                break;
+          case DSPF_RGB444: 
+               data->render.GetRow        = get_row_RGB444;
+               data->render.GetValues     = get_values_RGB444;
+               data->render.PutRow        = put_row_RGB444;
+               data->render.PutRowRGB     = put_row_rgb_RGB444;
+               data->render.PutMonoRow    = put_mono_row_RGB444;
+               data->render.PutValues     = put_values_RGB444;
+               data->render.PutMonoValues = put_mono_values_RGB444;
+               break;
           case DSPF_ARGB2554: 
                data->render.GetRow        = get_row_ARGB2554;
                data->render.GetValues     = get_values_ARGB2554;
@@ -836,6 +889,15 @@ directfbgl_create_context( GLcontext        *context,
                data->render.PutValues     = put_values_ARGB1555;
                data->render.PutMonoValues = put_mono_values_ARGB1555;
                break;
+          case DSPF_RGB555:
+               data->render.GetRow        = get_row_RGB555;
+               data->render.GetValues     = get_values_RGB555;
+               data->render.PutRow        = put_row_RGB555;
+               data->render.PutRowRGB     = put_row_rgb_RGB555;
+               data->render.PutMonoRow    = put_mono_row_RGB555;
+               data->render.PutValues     = put_values_RGB555;
+               data->render.PutMonoValues = put_mono_values_RGB555;
+               break;
           case DSPF_RGB16:
                data->render.GetRow        = get_row_RGB16;
                data->render.GetValues     = get_values_RGB16;
diff --git a/src/mesa/drivers/dos/dmesa.c b/src/mesa/drivers/dos/dmesa.c
index ee87e638529..003c06a8fff 100644
--- a/src/mesa/drivers/dos/dmesa.c
+++ b/src/mesa/drivers/dos/dmesa.c
@@ -31,9 +31,9 @@
  */
 
 
-#include "context.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "video.h"
 
diff --git a/src/mesa/drivers/dri/Makefile b/src/mesa/drivers/dri/Makefile
index f466ce6c3cc..9e49fb16f53 100644
--- a/src/mesa/drivers/dri/Makefile
+++ b/src/mesa/drivers/dri/Makefile
@@ -20,19 +20,36 @@ subdirs:
 		fi \
 	done
 
+pcedit = sed \
+	-e 's,@INSTALL_DIR@,$(INSTALL_DIR),' \
+	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
+	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
+	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
+	-e 's,@DRI_DRIVER_DIR@,$(DRI_DRIVER_SEARCH_DIR),' \
+	-e 's,@DRI_PC_REQ_PRIV@,$(DRI_PC_REQ_PRIV),'
 
-install:
+dri.pc: dri.pc.in
+	$(pcedit) $< > $@
+
+
+install: dri.pc
 	@for dir in $(DRI_DIRS) ; do \
 		if [ -d $$dir ] ; then \
 			(cd $$dir && $(MAKE) install) || exit 1 ; \
 		fi \
 	done
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_INC_DIR)/GL/internal
+	$(INSTALL) -m 0644 $(TOP)/include/GL/internal/dri_interface.h \
+	  $(DESTDIR)$(INSTALL_INC_DIR)/GL/internal
+	$(INSTALL) -d $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
+	$(INSTALL) -m 0644 dri.pc $(DESTDIR)$(INSTALL_LIB_DIR)/pkgconfig
 
 
 clean:
-	@for dir in $(DRI_DIRS) ; do \
+	-@for dir in $(DRI_DIRS) ; do \
 		if [ -d $$dir ] ; then \
 			(cd $$dir && $(MAKE) clean) ; \
 		fi \
 	done
 	-rm -f common/*.o
+	-rm -f *.pc
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 3e7e527a982..2fa36bab3f9 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -13,11 +13,6 @@ COMMON_SOURCES = $(COMMON_GALLIUM_SOURCES) \
         ../common/texmem.c \
         ../common/drirenderbuffer.c
 
-COMMON_BM_SOURCES = \
-	../common/dri_bufmgr.c \
-	../common/dri_drmpool.c
-
-
 ifeq ($(WINDOW_SYSTEM),dri)
 WINOBJ=
 WINLIB=
@@ -47,15 +42,7 @@ SHARED_INCLUDES = \
 	-I$(TOP)/src/mesa/drivers/dri/common \
 	-Iserver \
 	-I$(TOP)/include \
-	-I$(TOP)/include/GL/internal \
 	-I$(TOP)/src/mesa \
-	-I$(TOP)/src/mesa/main \
-	-I$(TOP)/src/mesa/glapi \
-	-I$(TOP)/src/mesa/math \
-	-I$(TOP)/src/mesa/transform \
-	-I$(TOP)/src/mesa/shader \
-	-I$(TOP)/src/mesa/swrast \
-	-I$(TOP)/src/mesa/swrast_setup \
 	-I$(TOP)/src/egl/main \
 	-I$(TOP)/src/egl/drivers/dri \
 	$(LIBDRM_CFLAGS)
@@ -72,12 +59,12 @@ SHARED_INCLUDES = \
 
 ##### TARGETS #####
 
-default: depend symlinks $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+default: symlinks depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
 
 
 $(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(PIPE_DRIVERS) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
-	$(TOP)/bin/mklib -noprefix -o $@ \
-		$(OBJECTS) $(PIPE_DRIVERS) $(MESA_MODULES)  $(WINOBJ) $(DRI_LIB_DEPS)
+	$(MKLIB) -o $@ -noprefix -linker '$(CC)' -ldflags '$(LDFLAGS)' \
+		$(OBJECTS) $(PIPE_DRIVERS) $(MESA_MODULES) $(WINOBJ) $(DRI_LIB_DEPS)
 
 
 $(TOP)/$(LIB_DIR)/$(LIBNAME): $(LIBNAME)
@@ -88,7 +75,7 @@ depend: $(C_SOURCES) $(ASM_SOURCES) $(SYMLINKS)
 	rm -f depend
 	touch depend
 	$(MKDEP) $(MKDEP_OPTIONS) $(DRIVER_DEFINES) $(INCLUDES) $(C_SOURCES) \
-		$(ASM_SOURCES) 2> /dev/null
+		$(ASM_SOURCES)
 
 
 # Emacs tags
@@ -103,8 +90,8 @@ clean:
 
 
 install: $(LIBNAME)
-	$(INSTALL) -d $(DRI_DRIVER_INSTALL_DIR)
-	$(INSTALL) -m 755 $(LIBNAME) $(DRI_DRIVER_INSTALL_DIR)
+	$(INSTALL) -d $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR)
+	$(INSTALL) -m 755 $(LIBNAME) $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR)
 
 
-include depend
+-include depend
diff --git a/src/mesa/drivers/dri/common/depthtmp.h b/src/mesa/drivers/dri/common/depthtmp.h
index 55199abf6a3..fd2dab3b422 100644
--- a/src/mesa/drivers/dri/common/depthtmp.h
+++ b/src/mesa/drivers/dri/common/depthtmp.h
@@ -29,7 +29,7 @@ static void TAG(WriteDepthSpan)( GLcontext *ctx,
 {
    HW_WRITE_LOCK()
       {
-         const GLuint *depth = (const GLuint *) values;
+         const VALUE_TYPE *depth = (const VALUE_TYPE *) values;
 	 GLint x1;
 	 GLint n1;
 	 LOCAL_DEPTH_VARS;
@@ -134,7 +134,7 @@ static void TAG(WriteDepthPixels)( GLcontext *ctx,
 {
    HW_WRITE_LOCK()
       {
-         const GLuint *depth = (const GLuint *) values;
+         const VALUE_TYPE *depth = (const VALUE_TYPE *) values;
 	 GLuint i;
 	 LOCAL_DEPTH_VARS;
 
@@ -180,7 +180,7 @@ static void TAG(ReadDepthSpan)( GLcontext *ctx,
 {
    HW_READ_LOCK()
       {
-         GLuint *depth = (GLuint *) values;
+         VALUE_TYPE *depth = (VALUE_TYPE *) values;
 	 GLint x1, n1;
 	 LOCAL_DEPTH_VARS;
 
@@ -215,7 +215,7 @@ static void TAG(ReadDepthPixels)( GLcontext *ctx,
 {
    HW_READ_LOCK()
       {
-         GLuint *depth = (GLuint *) values;
+         VALUE_TYPE *depth = (VALUE_TYPE *) values;
 	 GLuint i;
 	 LOCAL_DEPTH_VARS;
 
@@ -267,3 +267,4 @@ static void TAG(InitDepthPointers)(struct gl_renderbuffer *rb)
 #undef READ_DEPTH
 #endif
 #undef TAG
+#undef VALUE_TYPE
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr.c b/src/mesa/drivers/dri/common/dri_bufmgr.c
deleted file mode 100644
index 5747307f3b6..00000000000
--- a/src/mesa/drivers/dri/common/dri_bufmgr.c
+++ /dev/null
@@ -1,535 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellström <thomas-at-tungstengraphics-dot-com>
- *          Keith Whitwell <keithw-at-tungstengraphics-dot-com>
- */
-
-#include <xf86drm.h>
-#include <stdlib.h>
-#include "glthread.h"
-#include "errno.h"
-#include "dri_bufmgr.h"
-#include "string.h"
-#include "imports.h"
-#include "dri_bufpool.h"
-
-_glthread_DECLARE_STATIC_MUTEX(bmMutex);
-
-/*
- * TODO: Introduce fence pools in the same way as 
- * buffer object pools.
- */
-
-
-
-typedef struct _DriFenceObject
-{
-   int fd;
-   _glthread_Mutex mutex;
-   int refCount;
-   const char *name;
-   drmFence fence;
-} DriFenceObject;
-
-typedef struct _DriBufferObject
-{
-   DriBufferPool *pool;
-   _glthread_Mutex mutex;
-   int refCount;
-   const char *name;
-   unsigned flags;
-   unsigned hint;
-   unsigned alignment;
-   void *private;
-   /* user-space buffer: */
-   unsigned userBuffer;
-   void *userData;
-   unsigned userSize;
-} DriBufferObject;
-
-
-void
-bmError(int val, const char *file, const char *function, int line)
-{
-   _mesa_printf("Fatal video memory manager error \"%s\".\n"
-                "Check kernel logs or set the LIBGL_DEBUG\n"
-                "environment variable to \"verbose\" for more info.\n"
-                "Detected in file %s, line %d, function %s.\n",
-                strerror(-val), file, line, function);
-#ifndef NDEBUG
-   abort();
-#else
-   abort();
-#endif
-}
-
-DriFenceObject *
-driFenceBuffers(int fd, char *name, unsigned flags)
-{
-   DriFenceObject *fence = (DriFenceObject *) malloc(sizeof(*fence));
-   int ret;
-
-   if (!fence)
-      BM_CKFATAL(-EINVAL);
-
-   _glthread_LOCK_MUTEX(bmMutex);
-   fence->refCount = 1;
-   fence->name = name;
-   fence->fd = fd;
-   _glthread_INIT_MUTEX(fence->mutex);
-   ret = drmFenceBuffers(fd, flags, &fence->fence);
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   if (ret) {
-      free(fence);
-      BM_CKFATAL(ret);
-   }
-   return fence;
-}
-
-
-unsigned 
-driFenceType(DriFenceObject * fence)
-{
-    unsigned ret;
-
-    _glthread_LOCK_MUTEX(bmMutex);
-    ret = fence->fence.flags;
-    _glthread_UNLOCK_MUTEX(bmMutex);
-    
-    return ret;
-}
-
-
-DriFenceObject *
-driFenceReference(DriFenceObject * fence)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   ++fence->refCount;
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   return fence;
-}
-
-void
-driFenceUnReference(DriFenceObject * fence)
-{
-   if (!fence)
-      return;
-
-   _glthread_LOCK_MUTEX(bmMutex);
-   if (--fence->refCount == 0) {
-      drmFenceDestroy(fence->fd, &fence->fence);
-      free(fence);
-   }
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driFenceFinish(DriFenceObject * fence, unsigned type, int lazy)
-{
-   int ret;
-   unsigned flags = (lazy) ? DRM_FENCE_FLAG_WAIT_LAZY : 0;
-
-   _glthread_LOCK_MUTEX(fence->mutex);
-   ret = drmFenceWait(fence->fd, flags, &fence->fence, type);
-   _glthread_UNLOCK_MUTEX(fence->mutex);
-   BM_CKFATAL(ret);
-}
-
-int
-driFenceSignaled(DriFenceObject * fence, unsigned type)
-{
-   int signaled;
-   int ret;
-
-   if (fence == NULL)
-      return GL_TRUE;
-
-   _glthread_LOCK_MUTEX(fence->mutex);
-   ret = drmFenceSignaled(fence->fd, &fence->fence, type, &signaled);
-   _glthread_UNLOCK_MUTEX(fence->mutex);
-   BM_CKFATAL(ret);
-   return signaled;
-}
-
-
-extern drmBO *
-driBOKernel(struct _DriBufferObject *buf)
-{
-   drmBO *ret;
-
-   assert(buf->private != NULL);
-   ret = buf->pool->kernel(buf->pool, buf->private);
-   if (!ret)
-      BM_CKFATAL(-EINVAL);
-
-   return ret;
-}
-
-void
-driBOWaitIdle(struct _DriBufferObject *buf, int lazy)
-{
-   struct _DriBufferPool *pool;
-   void *priv;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   pool = buf->pool;
-   priv = buf->private;
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   
-   assert(priv != NULL);
-   BM_CKFATAL(buf->pool->waitIdle(pool, priv, lazy));
-}
-
-void *
-driBOMap(struct _DriBufferObject *buf, unsigned flags, unsigned hint)
-{
-   if (buf->userBuffer) {
-      return buf->userData;
-   }
-   else {
-      void *virtual;
-
-      assert(buf->private != NULL);
-
-      _glthread_LOCK_MUTEX(buf->mutex);
-      BM_CKFATAL(buf->pool->map(buf->pool, buf->private, flags, hint, &virtual));
-      _glthread_UNLOCK_MUTEX(buf->mutex);
-      return virtual;
-   }
-}
-
-void
-driBOUnmap(struct _DriBufferObject *buf)
-{
-   if (!buf->userBuffer) {
-      assert(buf->private != NULL);
-
-      buf->pool->unmap(buf->pool, buf->private);
-   }
-}
-
-unsigned long
-driBOOffset(struct _DriBufferObject *buf)
-{
-   unsigned long ret;
-
-   assert(buf->private != NULL);
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   ret = buf->pool->offset(buf->pool, buf->private);
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   return ret;
-}
-
-unsigned
-driBOFlags(struct _DriBufferObject *buf)
-{
-   unsigned ret;
-
-   assert(buf->private != NULL);
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   ret = buf->pool->flags(buf->pool, buf->private);
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   return ret;
-}
-
-struct _DriBufferObject *
-driBOReference(struct _DriBufferObject *buf)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   if (++buf->refCount == 1) {
-      BM_CKFATAL(-EINVAL);
-   }
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   return buf;
-}
-
-void
-driBOUnReference(struct _DriBufferObject *buf)
-{
-   int tmp;
-
-   if (!buf)
-      return;
-
-   _glthread_LOCK_MUTEX(bmMutex);
-   tmp = --buf->refCount;
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   if (!tmp) {
-      if (buf->private)
-         buf->pool->destroy(buf->pool, buf->private);
-      free(buf);
-   }
-}
-
-void
-driBOData(struct _DriBufferObject *buf,
-          unsigned size, const void *data, unsigned flags)
-{
-   void *virtual;
-   int newBuffer;
-   struct _DriBufferPool *pool;
-
-   assert(!buf->userBuffer); /* XXX just do a memcpy? */
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   pool = buf->pool;
-   if (!pool->create) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "driBOData called on invalid buffer\n");
-      BM_CKFATAL(-EINVAL);
-   }
-   newBuffer = !buf->private || (pool->size(pool, buf->private) < size) ||
-      pool->map(pool, buf->private, DRM_BO_FLAG_WRITE,
-                DRM_BO_HINT_DONT_BLOCK, &virtual);
-
-   if (newBuffer) {
-      if (buf->private)
-         pool->destroy(pool, buf->private);
-      if (!flags)
-         flags = buf->flags;
-      buf->private = pool->create(pool, size, flags, DRM_BO_HINT_DONT_FENCE, 
-				  buf->alignment);
-      if (!buf->private)
-         BM_CKFATAL(-ENOMEM);
-      BM_CKFATAL(pool->map(pool, buf->private,
-                           DRM_BO_FLAG_WRITE,
-                           DRM_BO_HINT_DONT_BLOCK, &virtual));
-   }
-
-   if (data != NULL)
-      memcpy(virtual, data, size);
-
-   BM_CKFATAL(pool->unmap(pool, buf->private));
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-void
-driBOSubData(struct _DriBufferObject *buf,
-             unsigned long offset, unsigned long size, const void *data)
-{
-   void *virtual;
-
-   assert(!buf->userBuffer); /* XXX just do a memcpy? */
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (size && data) {
-      BM_CKFATAL(buf->pool->map(buf->pool, buf->private,
-                                DRM_BO_FLAG_WRITE, 0, &virtual));
-      memcpy((unsigned char *) virtual + offset, data, size);
-      BM_CKFATAL(buf->pool->unmap(buf->pool, buf->private));
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-void
-driBOGetSubData(struct _DriBufferObject *buf,
-                unsigned long offset, unsigned long size, void *data)
-{
-   void *virtual;
-
-   assert(!buf->userBuffer); /* XXX just do a memcpy? */
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (size && data) {
-      BM_CKFATAL(buf->pool->map(buf->pool, buf->private,
-                                DRM_BO_FLAG_READ, 0, &virtual));
-      memcpy(data, (unsigned char *) virtual + offset, size);
-      BM_CKFATAL(buf->pool->unmap(buf->pool, buf->private));
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-void
-driBOSetStatic(struct _DriBufferObject *buf,
-               unsigned long offset,
-               unsigned long size, void *virtual, unsigned flags)
-{
-   assert(!buf->userBuffer); /* XXX what to do? */
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (buf->private != NULL) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "Invalid buffer for setStatic\n");
-      BM_CKFATAL(-EINVAL);
-   }
-   if (buf->pool->setstatic == NULL) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "Invalid buffer pool for setStatic\n");
-      BM_CKFATAL(-EINVAL);
-   }
-
-   if (!flags)
-      flags = buf->flags;
-
-   buf->private = buf->pool->setstatic(buf->pool, offset, size,
-                                       virtual, flags);
-   if (!buf->private) {
-      _mesa_error(NULL, GL_OUT_OF_MEMORY,
-                  "Invalid buffer pool for setStatic\n");
-      BM_CKFATAL(-ENOMEM);
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-
-
-void
-driGenBuffers(struct _DriBufferPool *pool,
-              const char *name,
-              unsigned n,
-              struct _DriBufferObject *buffers[],
-              unsigned alignment, unsigned flags, unsigned hint)
-{
-   struct _DriBufferObject *buf;
-   int i;
-
-   flags = (flags) ? flags : DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_MEM_VRAM |
-      DRM_BO_FLAG_MEM_LOCAL | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE;
-
-
-   for (i = 0; i < n; ++i) {
-      buf = (struct _DriBufferObject *) calloc(1, sizeof(*buf));
-      if (!buf)
-         BM_CKFATAL(-ENOMEM);
-
-      _glthread_INIT_MUTEX(buf->mutex);
-      _glthread_LOCK_MUTEX(buf->mutex);
-      _glthread_LOCK_MUTEX(bmMutex);
-      buf->refCount = 1;
-      _glthread_UNLOCK_MUTEX(bmMutex);
-      buf->flags = flags;
-      buf->hint = hint;
-      buf->name = name;
-      buf->alignment = alignment;
-      buf->pool = pool;
-      _glthread_UNLOCK_MUTEX(buf->mutex);
-      buffers[i] = buf;
-   }
-}
-
-void
-driGenUserBuffer(struct _DriBufferPool *pool,
-                 const char *name,
-                 struct _DriBufferObject **buffers,
-                 void *ptr, unsigned bytes)
-{
-   const unsigned alignment = 1, flags = 0, hint = 0;
-
-   driGenBuffers(pool, name, 1, buffers, alignment, flags, hint);
-
-   (*buffers)->userBuffer = 1;
-   (*buffers)->userData = ptr;
-   (*buffers)->userSize = bytes;
-}
-
-
-void
-driDeleteBuffers(unsigned n, struct _DriBufferObject *buffers[])
-{
-   int i;
-
-   for (i = 0; i < n; ++i) {
-      driBOUnReference(buffers[i]);
-   }
-}
-
-
-void
-driInitBufMgr(int fd)
-{
-   ;
-}
-
-
-void
-driBOCreateList(int target, drmBOList * list)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOCreateList(target, list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driBOResetList(drmBOList * list)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOResetList(list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driBOAddListItem(drmBOList * list, struct _DriBufferObject *buf,
-                 unsigned flags, unsigned mask)
-{
-   int newItem;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmAddValidateItem(list, driBOKernel(buf),
-                                 flags, mask, &newItem));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-
-   /*
-    * Tell userspace pools to validate the buffer. This should be a 
-    * noop if the pool is already validated.
-    * FIXME: We should have a list for this as well.
-    */
-
-   if (buf->pool->validate) {
-      BM_CKFATAL(buf->pool->validate(buf->pool, buf->private));
-   }
-
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-void
-driBOFence(struct _DriBufferObject *buf, struct _DriFenceObject *fence)
-{
-   _glthread_LOCK_MUTEX(buf->mutex);
-   BM_CKFATAL(buf->pool->fence(buf->pool, buf->private, fence));
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-
-}
-
-void
-driBOValidateList(int fd, drmBOList * list)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOValidateList(fd, list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driPoolTakeDown(struct _DriBufferPool *pool)
-{
-   pool->takeDown(pool);
-
-}
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr.h b/src/mesa/drivers/dri/common/dri_bufmgr.h
deleted file mode 100644
index ee4ce2cbdef..00000000000
--- a/src/mesa/drivers/dri/common/dri_bufmgr.h
+++ /dev/null
@@ -1,103 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellström <thomas-at-tungstengraphics-dot-com>
- *          Keith Whitwell <keithw-at-tungstengraphics-dot-com>
- */
-
-#ifndef _DRI_BUFMGR_H_
-#define _DRI_BUFMGR_H_
-#include <xf86drm.h>
-
-
-struct _DriFenceObject;
-struct _DriBufferObject;
-struct _DriBufferPool;
-
-extern struct _DriFenceObject *driFenceBuffers(int fd, char *name,
-                                               unsigned flags);
-
-extern struct _DriFenceObject *driFenceReference(struct _DriFenceObject *fence);
-
-extern void driFenceUnReference(struct _DriFenceObject *fence);
-
-extern void
-driFenceFinish(struct _DriFenceObject *fence, unsigned type, int lazy);
-
-extern int driFenceSignaled(struct _DriFenceObject *fence, unsigned type);
-extern unsigned driFenceType(struct _DriFenceObject *fence);
-
-/*
- * Return a pointer to the libdrm buffer object this DriBufferObject
- * uses.
- */
-
-extern drmBO *driBOKernel(struct _DriBufferObject *buf);
-extern void *driBOMap(struct _DriBufferObject *buf, unsigned flags,
-                      unsigned hint);
-extern void driBOUnmap(struct _DriBufferObject *buf);
-extern unsigned long driBOOffset(struct _DriBufferObject *buf);
-extern unsigned driBOFlags(struct _DriBufferObject *buf);
-extern struct _DriBufferObject *driBOReference(struct _DriBufferObject *buf);
-extern void driBOUnReference(struct _DriBufferObject *buf);
-extern void driBOData(struct _DriBufferObject *r_buf,
-                      unsigned size, const void *data, unsigned flags);
-extern void driBOSubData(struct _DriBufferObject *buf,
-                         unsigned long offset, unsigned long size,
-                         const void *data);
-extern void driBOGetSubData(struct _DriBufferObject *buf,
-                            unsigned long offset, unsigned long size,
-                            void *data);
-extern void driGenBuffers(struct _DriBufferPool *pool,
-                          const char *name,
-                          unsigned n,
-                          struct _DriBufferObject *buffers[],
-                          unsigned alignment, unsigned flags, unsigned hint);
-extern void driGenUserBuffer(struct _DriBufferPool *pool,
-                             const char *name,
-                             struct _DriBufferObject *buffers[],
-                             void *ptr, unsigned bytes);
-extern void driDeleteBuffers(unsigned n, struct _DriBufferObject *buffers[]);
-extern void driInitBufMgr(int fd);
-extern void driBOCreateList(int target, drmBOList * list);
-extern void driBOResetList(drmBOList * list);
-extern void driBOAddListItem(drmBOList * list, struct _DriBufferObject *buf,
-                             unsigned flags, unsigned mask);
-extern void driBOValidateList(int fd, drmBOList * list);
-
-extern void driBOFence(struct _DriBufferObject *buf,
-                       struct _DriFenceObject *fence);
-
-extern void driPoolTakeDown(struct _DriBufferPool *pool);
-extern void driBOSetStatic(struct _DriBufferObject *buf,
-                           unsigned long offset,
-                           unsigned long size, void *virtual, unsigned flags);
-extern void driBOWaitIdle(struct _DriBufferObject *buf, int lazy);
-extern void driPoolTakeDown(struct _DriBufferPool *pool);
-
-#endif
diff --git a/src/mesa/drivers/dri/common/dri_bufpool.h b/src/mesa/drivers/dri/common/dri_bufpool.h
deleted file mode 100644
index c6fb2c3ce01..00000000000
--- a/src/mesa/drivers/dri/common/dri_bufpool.h
+++ /dev/null
@@ -1,86 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellström <thomas-at-tungstengraphics-dot-com>
- */
-
-#ifndef _DRI_BUFPOOL_H_
-#define _DRI_BUFPOOL_H_
-
-#include <xf86drm.h>
-struct _DriFenceObject;
-
-typedef struct _DriBufferPool
-{
-   int fd;
-   int (*map) (struct _DriBufferPool * pool, void *private,
-               unsigned flags, int hint, void **virtual);
-   int (*unmap) (struct _DriBufferPool * pool, void *private);
-   int (*destroy) (struct _DriBufferPool * pool, void *private);
-   unsigned long (*offset) (struct _DriBufferPool * pool, void *private);
-   unsigned (*flags) (struct _DriBufferPool * pool, void *private);
-   unsigned long (*size) (struct _DriBufferPool * pool, void *private);
-   void *(*create) (struct _DriBufferPool * pool, unsigned long size,
-                    unsigned flags, unsigned hint, unsigned alignment);
-   int (*fence) (struct _DriBufferPool * pool, void *private,
-                 struct _DriFenceObject * fence);
-   drmBO *(*kernel) (struct _DriBufferPool * pool, void *private);
-   int (*validate) (struct _DriBufferPool * pool, void *private);
-   void *(*setstatic) (struct _DriBufferPool * pool, unsigned long offset,
-                       unsigned long size, void *virtual, unsigned flags);
-   int (*waitIdle) (struct _DriBufferPool *pool, void *private,
-		    int lazy);
-   void (*takeDown) (struct _DriBufferPool * pool);
-   void *data;
-} DriBufferPool;
-
-extern void bmError(int val, const char *file, const char *function,
-                    int line);
-#define BM_CKFATAL(val)					       \
-  do{							       \
-    int tstVal = (val);					       \
-    if (tstVal) 					       \
-      bmError(tstVal, __FILE__, __FUNCTION__, __LINE__);       \
-  } while(0);
-
-
-
-
-
-/*
- * Builtin pools.
- */
-
-/*
- * Kernel buffer objects. Size in multiples of page size. Page size aligned.
- */
-
-extern struct _DriBufferPool *driDRMPoolInit(int fd);
-extern struct _DriBufferPool *driDRMStaticPoolInit(int fd);
-
-#endif
diff --git a/src/mesa/drivers/dri/common/dri_drmpool.c b/src/mesa/drivers/dri/common/dri_drmpool.c
deleted file mode 100644
index 878a148b397..00000000000
--- a/src/mesa/drivers/dri/common/dri_drmpool.c
+++ /dev/null
@@ -1,227 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellström <thomas-at-tungstengraphics-dot-com>
- */
-
-#include <xf86drm.h>
-#include <stdlib.h>
-#include <unistd.h>
-#include "dri_bufpool.h"
-
-/*
- * Buffer pool implementation using DRM buffer objects as DRI buffer objects.
- */
-
-static void *
-pool_create(struct _DriBufferPool *pool,
-            unsigned long size, unsigned flags, unsigned hint,
-            unsigned alignment)
-{
-   drmBO *buf = (drmBO *) malloc(sizeof(*buf));
-   int ret;
-   unsigned pageSize = getpagesize();
-
-   if (!buf)
-      return NULL;
-
-   if ((alignment > pageSize) && (alignment % pageSize)) {
-      return NULL;
-   }
-
-   ret = drmBOCreate(pool->fd, 0, size, alignment / pageSize,
-		     NULL, drm_bo_type_dc,
-                     flags, hint, buf);
-   if (ret) {
-      free(buf);
-      return NULL;
-   }
-
-   return (void *) buf;
-}
-
-static int
-pool_destroy(struct _DriBufferPool *pool, void *private)
-{
-   int ret;
-   drmBO *buf = (drmBO *) private;
-   ret = drmBODestroy(pool->fd, buf);
-   free(buf);
-   return ret;
-}
-
-static int
-pool_map(struct _DriBufferPool *pool, void *private, unsigned flags,
-         int hint, void **virtual)
-{
-   drmBO *buf = (drmBO *) private;
-
-   return drmBOMap(pool->fd, buf, flags, hint, virtual);
-}
-
-static int
-pool_unmap(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return drmBOUnmap(pool->fd, buf);
-}
-
-static unsigned long
-pool_offset(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->offset;
-}
-
-static unsigned
-pool_flags(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->flags;
-}
-
-
-static unsigned long
-pool_size(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->size;
-}
-
-static int
-pool_fence(struct _DriBufferPool *pool, void *private,
-           struct _DriFenceObject *fence)
-{
-   /*
-    * Noop. The kernel handles all fencing.
-    */
-
-   return 0;
-}
-
-static drmBO *
-pool_kernel(struct _DriBufferPool *pool, void *private)
-{
-   return (drmBO *) private;
-}
-
-static int
-pool_waitIdle(struct _DriBufferPool *pool, void *private, int lazy)
-{
-   drmBO *buf = (drmBO *) private;
-   return drmBOWaitIdle(pool->fd, buf, (lazy) ? DRM_BO_HINT_WAIT_LAZY:0);
-}
-
-    
-static void
-pool_takedown(struct _DriBufferPool *pool)
-{
-   free(pool);
-}
-
-
-struct _DriBufferPool *
-driDRMPoolInit(int fd)
-{
-   struct _DriBufferPool *pool;
-
-   pool = (struct _DriBufferPool *) malloc(sizeof(*pool));
-
-   if (!pool)
-      return NULL;
-
-   pool->fd = fd;
-   pool->map = &pool_map;
-   pool->unmap = &pool_unmap;
-   pool->destroy = &pool_destroy;
-   pool->offset = &pool_offset;
-   pool->flags = &pool_flags;
-   pool->size = &pool_size;
-   pool->create = &pool_create;
-   pool->fence = &pool_fence;
-   pool->kernel = &pool_kernel;
-   pool->validate = NULL;
-   pool->setstatic = NULL;
-   pool->waitIdle = &pool_waitIdle;
-   pool->takeDown = &pool_takedown;
-   pool->data = NULL;
-   return pool;
-}
-
-
-static void *
-pool_setstatic(struct _DriBufferPool *pool, unsigned long offset,
-               unsigned long size, void *virtual, unsigned flags)
-{
-   drmBO *buf = (drmBO *) malloc(sizeof(*buf));
-   int ret;
-
-   if (!buf)
-      return NULL;
-
-   ret = drmBOCreate(pool->fd, offset, size, 0, NULL, drm_bo_type_fake,
-                     flags, DRM_BO_HINT_DONT_FENCE, buf);
-
-   if (ret) {
-      free(buf);
-      return NULL;
-   }
-
-   buf->virtual = virtual;
-
-   return (void *) buf;
-}
-
-
-struct _DriBufferPool *
-driDRMStaticPoolInit(int fd)
-{
-   struct _DriBufferPool *pool;
-
-   pool = (struct _DriBufferPool *) malloc(sizeof(*pool));
-
-   if (!pool)
-      return NULL;
-
-   pool->fd = fd;
-   pool->map = &pool_map;
-   pool->unmap = &pool_unmap;
-   pool->destroy = &pool_destroy;
-   pool->offset = &pool_offset;
-   pool->flags = &pool_flags;
-   pool->size = &pool_size;
-   pool->create = NULL;
-   pool->fence = &pool_fence;
-   pool->kernel = &pool_kernel;
-   pool->validate = NULL;
-   pool->setstatic = &pool_setstatic;
-   pool->waitIdle = &pool_waitIdle;
-   pool->takeDown = &pool_takedown;
-   pool->data = NULL;
-   return pool;
-}
diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c
index ac6e3cdda8e..a2316e2662b 100644
--- a/src/mesa/drivers/dri/common/dri_util.c
+++ b/src/mesa/drivers/dri/common/dri_util.c
@@ -1,28 +1,4 @@
-/*
- * Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
+/* $XFree86: xc/lib/GL/dri/dri_util.c,v 1.7 2003/04/28 17:01:25 dawes Exp $ */
 /**
  * \file dri_util.c
  * DRI utility functions.
@@ -50,44 +26,24 @@
 #define MAP_FAILED ((void *)-1)
 #endif
 
-#include "imports.h"
+#include "main/imports.h"
 #define None 0
 
 #include "dri_util.h"
 #include "drm_sarea.h"
+#include "utils.h"
 
 #ifndef GLX_OML_sync_control
-typedef GLboolean ( * PFNGLXGETMSCRATEOMLPROC) (__DRInativeDisplay *dpy, __DRIid drawable, int32_t *numerator, int32_t *denominator);
+typedef GLboolean ( * PFNGLXGETMSCRATEOMLPROC) (__DRIdrawable *drawable, int32_t *numerator, int32_t *denominator);
 #endif
 
-/* This pointer *must* be set by the driver's __driCreateNewScreen funciton!
- */
-const __DRIinterfaceMethods * dri_interface = NULL;
-
-/**
- * This is used in a couple of places that call \c driCreateNewDrawable.
- */
-static const int empty_attribute_list[1] = { None };
-
-
 /**
- * Cached copy of the internal API version used by libGL and the client-side
- * DRI driver.
+ * This is just a token extension used to signal that the driver
+ * supports setting a read drawable.
  */
-static int api_ver = 0;
-
-/* forward declarations */
-static int driQueryFrameTracking( __DRInativeDisplay *dpy, void *priv,
-                                  int64_t *sbc, int64_t *missedFrames,
-                                  float *lastMissedUsage, float *usage );
-
-static void *driCreateNewDrawable(__DRInativeDisplay *dpy,
-                                  const __GLcontextModes *modes,
-                                  __DRIid draw, __DRIdrawable *pdraw,
-                                  int renderType, const int *attrs);
-
-static void driDestroyDrawable(__DRInativeDisplay *dpy, void *drawablePrivate);
-
+const __DRIextension driReadDrawableExtension = {
+    __DRI_READ_DRAWABLE, __DRI_READ_DRAWABLE_VERSION
+};
 
 /**
  * Print message to \c stderr if the \c LIBGL_DEBUG environment variable
@@ -111,64 +67,19 @@ __driUtilMessage(const char *f, ...)
     }
 }
 
-
-/*****************************************************************/
-/** \name Drawable list management */
-/*****************************************************************/
-/*@{*/
-
-static GLboolean __driAddDrawable(void *drawHash, __DRIdrawable *pdraw)
-{
-    __DRIdrawablePrivate *pdp = (__DRIdrawablePrivate *)pdraw->private;
-
-    if (drmHashInsert(drawHash, pdp->draw, pdraw))
-	return GL_FALSE;
-
-    return GL_TRUE;
-}
-
-static __DRIdrawable *__driFindDrawable(void *drawHash, __DRIid draw)
+GLint
+driIntersectArea( drm_clip_rect_t rect1, drm_clip_rect_t rect2 )
 {
-    int retcode;
-    __DRIdrawable *pdraw;
-
-    retcode = drmHashLookup(drawHash, draw, (void *)&pdraw);
-    if (retcode)
-	return NULL;
-
-    return pdraw;
-}
-
+   if (rect2.x1 > rect1.x1) rect1.x1 = rect2.x1;
+   if (rect2.x2 < rect1.x2) rect1.x2 = rect2.x2;
+   if (rect2.y1 > rect1.y1) rect1.y1 = rect2.y1;
+   if (rect2.y2 < rect1.y2) rect1.y2 = rect2.y2;
 
-/**
- * Find drawables in the local hash that have been destroyed on the
- * server.
- * 
- * \param drawHash  Hash-table containing all known drawables.
- */
-static void __driGarbageCollectDrawables(void *drawHash)
-{
-    __DRIid draw;
-    __DRInativeDisplay *dpy;
-    __DRIdrawable *pdraw;
+   if (rect1.x1 > rect1.x2 || rect1.y1 > rect1.y2) return 0;
 
-    if (drmHashFirst(drawHash, &draw, (void *)&pdraw) == 1) {
-	do {
-	    __DRIdrawablePrivate *pdp = (__DRIdrawablePrivate *)pdraw->private;
-	    dpy = pdp->driScreenPriv->display;
-	    if (! (*dri_interface->windowExists)(dpy, draw)) {
-		/* Destroy the local drawable data, if the drawable no
-		   longer exists in the Xserver */
-		(*pdraw->destroyDrawable)(dpy, pdraw->private);
-		_mesa_free(pdraw);
-	    }
-	} while (drmHashNext(drawHash, &draw, (void *)&pdraw) == 1);
-    }
+   return (rect1.x2 - rect1.x1) * (rect1.y2 - rect1.y1);
 }
 
-/*@}*/
-
-
 /*****************************************************************/
 /** \name Context (un)binding functions                          */
 /*****************************************************************/
@@ -177,10 +88,7 @@ static void __driGarbageCollectDrawables(void *drawHash)
 /**
  * Unbind context.
  * 
- * \param dpy the display handle.
- * \param scrn the screen number.
- * \param draw drawable.
- * \param read Current reading drawable.
+ * \param scrn the screen.
  * \param gc context.
  *
  * \return \c GL_TRUE on success, or \c GL_FALSE on failure.
@@ -193,56 +101,27 @@ static void __driGarbageCollectDrawables(void *drawHash)
  * While casting the opaque private pointers associated with the parameters
  * into their respective real types it also assures they are not \c NULL. 
  */
-static GLboolean driUnbindContext(__DRInativeDisplay *dpy, int scrn,
-			      __DRIid draw, __DRIid read,
-			      __DRIcontext *ctx)
+static int driUnbindContext(__DRIcontext *pcp)
 {
-    __DRIscreen *pDRIScreen;
-    __DRIdrawable *pdraw;
-    __DRIdrawable *pread;
-    __DRIcontextPrivate *pcp;
-    __DRIscreenPrivate *psp;
-    __DRIdrawablePrivate *pdp;
-    __DRIdrawablePrivate *prp;
+    __DRIscreen *psp;
+    __DRIdrawable *pdp;
+    __DRIdrawable *prp;
 
     /*
     ** Assume error checking is done properly in glXMakeCurrent before
     ** calling driUnbindContext.
     */
 
-    if (ctx == NULL || draw == None || read == None) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-
-    pDRIScreen = (*dri_interface->getScreen)(dpy, scrn);
-    if ( (pDRIScreen == NULL) || (pDRIScreen->private == NULL) ) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-
-    psp = (__DRIscreenPrivate *)pDRIScreen->private;
-    pcp = (__DRIcontextPrivate *)ctx->private;
-
-    pdraw = __driFindDrawable(psp->drawHash, draw);
-    if (!pdraw) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-    pdp = (__DRIdrawablePrivate *)pdraw->private;
-
-    pread = __driFindDrawable(psp->drawHash, read);
-    if (!pread) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-    prp = (__DRIdrawablePrivate *)pread->private;
+    if (pcp == NULL)
+        return GL_FALSE;
 
+    psp = pcp->driScreenPriv;
+    pdp = pcp->driDrawablePriv;
+    prp = pcp->driReadablePriv;
 
     /* Let driver unbind drawable from context */
     (*psp->DriverAPI.UnbindContext)(pcp);
 
-
     if (pdp->refcount == 0) {
 	/* ERROR!!! */
 	return GL_FALSE;
@@ -259,12 +138,6 @@ static GLboolean driUnbindContext(__DRInativeDisplay *dpy, int scrn,
 	prp->refcount--;
     }
 
-   /* destroy the drawables if they no longer exist on the server */
-   if ((pdp->refcount == 0) || (prp->refcount == 0)) {
-      /* probably shouldn't need the collector here,
-         as we know the affected drawables (or could there be others?) */
-      __driGarbageCollectDrawables(pdp->driScreenPriv->drawHash);
-   }
 
     /* XXX this is disabled so that if we call SwapBuffers on an unbound
      * window we can determine the last context bound to the window and
@@ -279,77 +152,24 @@ static GLboolean driUnbindContext(__DRInativeDisplay *dpy, int scrn,
     return GL_TRUE;
 }
 
-
 /**
  * This function takes both a read buffer and a draw buffer.  This is needed
  * for \c glXMakeCurrentReadSGI or GLX 1.3's \c glXMakeContextCurrent
  * function.
- * 
- * \bug This function calls \c driCreateNewDrawable in two places with the
- *      \c renderType hard-coded to \c GLX_WINDOW_BIT.  Some checking might
- *      be needed in those places when support for pbuffers and / or pixmaps
- *      is added.  Is it safe to assume that the drawable is a window?
  */
-static GLboolean DoBindContext(__DRInativeDisplay *dpy,
-			  __DRIid draw, __DRIid read,
-			  __DRIcontext *ctx, const __GLcontextModes * modes,
-			  __DRIscreenPrivate *psp)
+static int driBindContext(__DRIcontext *pcp,
+			  __DRIdrawable *pdp,
+			  __DRIdrawable *prp)
 {
-    __DRIdrawable *pdraw;
-    __DRIdrawablePrivate *pdp;
-    __DRIdrawable *pread;
-    __DRIdrawablePrivate *prp;
-    __DRIcontextPrivate * const pcp = ctx->private;
-
-
-    /* Find the _DRIdrawable which corresponds to the writing drawable. */
-    pdraw = __driFindDrawable(psp->drawHash, draw);
-    if (!pdraw) {
-	/* Allocate a new drawable */
-	pdraw = (__DRIdrawable *)_mesa_malloc(sizeof(__DRIdrawable));
-	if (!pdraw) {
-	    /* ERROR!!! */
-	    return GL_FALSE;
-	}
-
-	/* Create a new drawable */
-	driCreateNewDrawable(dpy, modes, draw, pdraw, GLX_WINDOW_BIT,
-			     empty_attribute_list);
-	if (!pdraw->private) {
-	    /* ERROR!!! */
-	    _mesa_free(pdraw);
-	    return GL_FALSE;
-	}
+    __DRIscreenPrivate *psp = pcp->driScreenPriv;
 
-    }
-    pdp = (__DRIdrawablePrivate *) pdraw->private;
+    /*
+    ** Assume error checking is done properly in glXMakeCurrent before
+    ** calling driBindContext.
+    */
 
-    /* Find the _DRIdrawable which corresponds to the reading drawable. */
-    if (read == draw) {
-        /* read buffer == draw buffer */
-        prp = pdp;
-    }
-    else {
-        pread = __driFindDrawable(psp->drawHash, read);
-        if (!pread) {
-            /* Allocate a new drawable */
-            pread = (__DRIdrawable *)_mesa_malloc(sizeof(__DRIdrawable));
-            if (!pread) {
-                /* ERROR!!! */
-                return GL_FALSE;
-            }
-
-            /* Create a new drawable */
-	    driCreateNewDrawable(dpy, modes, read, pread, GLX_WINDOW_BIT,
-				 empty_attribute_list);
-            if (!pread->private) {
-                /* ERROR!!! */
-                _mesa_free(pread);
-                return GL_FALSE;
-            }
-        }
-        prp = (__DRIdrawablePrivate *) pread->private;
-    }
+    if (pcp == NULL || pdp == None || prp == None)
+	return GL_FALSE;
 
     /* Bind the drawable to the context */
     pcp->driDrawablePriv = pdp;
@@ -364,16 +184,19 @@ static GLboolean DoBindContext(__DRInativeDisplay *dpy,
     ** Now that we have a context associated with this drawable, we can
     ** initialize the drawable information if has not been done before.
     */
-    if (!pdp->pStamp || *pdp->pStamp != pdp->lastStamp) {
-	DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
-	__driUtilUpdateDrawableInfo(pdp);
-	DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
-    }
 
-    if ((pdp != prp) && (!prp->pStamp || *prp->pStamp != prp->lastStamp)) {
-	DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
-	__driUtilUpdateDrawableInfo(prp);
-	DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
+    if (!psp->dri2.enabled) {
+	if (!pdp->pStamp || *pdp->pStamp != pdp->lastStamp) {
+	    DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
+	    __driUtilUpdateDrawableInfo(pdp);
+	    DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
+	}
+	
+	if ((pdp != prp) && (!prp->pStamp || *prp->pStamp != prp->lastStamp)) {
+	    DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
+	    __driUtilUpdateDrawableInfo(prp);
+	    DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
+	}
     }
 
     /* Call device-specific MakeCurrent */
@@ -382,37 +205,6 @@ static GLboolean DoBindContext(__DRInativeDisplay *dpy,
     return GL_TRUE;
 }
 
-
-/**
- * This function takes both a read buffer and a draw buffer.  This is needed
- * for \c glXMakeCurrentReadSGI or GLX 1.3's \c glXMakeContextCurrent
- * function.
- */
-static GLboolean driBindContext(__DRInativeDisplay *dpy, int scrn,
-                            __DRIid draw, __DRIid read,
-                            __DRIcontext * ctx)
-{
-    __DRIscreen *pDRIScreen;
-
-    /*
-    ** Assume error checking is done properly in glXMakeCurrent before
-    ** calling driBindContext.
-    */
-
-    if (ctx == NULL || draw == None || read == None) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-
-    pDRIScreen = (*dri_interface->getScreen)(dpy, scrn);
-    if ( (pDRIScreen == NULL) || (pDRIScreen->private == NULL) ) {
-	/* ERROR!!! */
-	return GL_FALSE;
-    }
-
-    return DoBindContext( dpy, draw, read, ctx, ctx->mode,
-			  (__DRIscreenPrivate *)pDRIScreen->private );
-}
 /*@}*/
 
 
@@ -436,7 +228,7 @@ static GLboolean driBindContext(__DRInativeDisplay *dpy, int scrn,
 void
 __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
 {
-    __DRIscreenPrivate *psp;
+    __DRIscreenPrivate *psp = pdp->driScreenPriv;
     __DRIcontextPrivate *pcp = pdp->driContextPriv;
     
     if (!pcp 
@@ -447,15 +239,6 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
 	 */
     }
 
-    psp = pdp->driScreenPriv;
-    if (!psp) {
-	/* ERROR!!! */
-       _mesa_problem(NULL, "Warning! Possible infinite loop due to bug "
-		     "in file %s, line %d\n",
-		     __FILE__, __LINE__);
-	return;
-    }
-
     if (pdp->pClipRects) {
 	_mesa_free(pdp->pClipRects); 
 	pdp->pClipRects = NULL;
@@ -468,15 +251,15 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
 
     DRM_SPINUNLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
 
-    if (!__driFindDrawable(psp->drawHash, pdp->draw) ||
-	! (*dri_interface->getDrawableInfo)(pdp->display, pdp->screen, pdp->draw,
+    if (! (*psp->getDrawableInfo->getDrawableInfo)(pdp,
 			  &pdp->index, &pdp->lastStamp,
 			  &pdp->x, &pdp->y, &pdp->w, &pdp->h,
 			  &pdp->numClipRects, &pdp->pClipRects,
 			  &pdp->backX,
 			  &pdp->backY,
 			  &pdp->numBackClipRects,
-			  &pdp->pBackClipRects )) {
+			  &pdp->pBackClipRects,
+			  pdp->loaderPrivate)) {
 	/* Error -- eg the window may have been destroyed.  Keep going
 	 * with no cliprects.
 	 */
@@ -490,7 +273,6 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
        pdp->pStamp = &(psp->pSAREA->drawableTable[pdp->index].stamp);
 
     DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);
-
 }
 
 /*@}*/
@@ -500,10 +282,28 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
 /*****************************************************************/
 /*@{*/
 
+static void driReportDamage(__DRIdrawable *pdp,
+			    struct drm_clip_rect *pClipRects, int numClipRects)
+{
+    __DRIscreen *psp = pdp->driScreenPriv;
+
+    /* Check that we actually have the new damage report method */
+    if (psp->damage) {
+	/* Report the damage.  Currently, all our drivers draw
+	 * directly to the front buffer, so we report the damage there
+	 * rather than to the backing storein (if any).
+	 */
+	(*psp->damage->reportDamage)(pdp,
+				     pdp->x, pdp->y,
+				     pClipRects, numClipRects,
+				     GL_TRUE, pdp->loaderPrivate);
+    }
+}
+
+
 /**
  * Swap buffers.
  *
- * \param dpy the display handle.
  * \param drawablePrivate opaque pointer to the per-drawable private info.
  * 
  * \internal
@@ -511,78 +311,29 @@ __driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
  * 
  * Is called directly from glXSwapBuffers().
  */
-static void driSwapBuffers( __DRInativeDisplay *dpy, void *drawablePrivate )
+static void driSwapBuffers(__DRIdrawable *dPriv)
 {
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate;
-    drm_clip_rect_t rect;
+    __DRIscreen *psp = dPriv->driScreenPriv;
 
-    dPriv->swapBuffers(dPriv);
-
-    /* Check that we actually have the new damage report method */
-    if (api_ver < 20070105 || dri_interface->reportDamage == NULL)
-	return;
-
-    /* Assume it's affecting the whole drawable for now */
-    rect.x1 = 0;
-    rect.y1 = 0;
-    rect.x2 = rect.x1 + dPriv->w;
-    rect.y2 = rect.y1 + dPriv->h;
-
-    /* Report the damage.  Currently, all our drivers draw directly to the
-     * front buffer, so we report the damage there rather than to the backing
-     * store (if any).
-     */
-    (*dri_interface->reportDamage)(dpy, dPriv->screen, dPriv->draw,
-				   dPriv->x, dPriv->y,
-				   &rect, 1, GL_TRUE);
-}
-
-/**
- * Called directly from a number of higher-level GLX functions.
- */
-static int driGetMSC( void *screenPrivate, int64_t *msc )
-{
-    __DRIscreenPrivate *sPriv = (__DRIscreenPrivate *) screenPrivate;
+    psp->DriverAPI.SwapBuffers(dPriv);
 
-    return sPriv->DriverAPI.GetMSC( sPriv, msc );
+    driReportDamage(dPriv, dPriv->pClipRects, dPriv->numClipRects);
 }
 
-/**
- * Called directly from a number of higher-level GLX functions.
- */
-static int driGetSBC( __DRInativeDisplay *dpy, void *drawablePrivate, int64_t *sbc )
+static int driDrawableGetMSC( __DRIscreen *sPriv, __DRIdrawable *dPriv,
+			      int64_t *msc )
 {
-   __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate;
-   __DRIswapInfo  sInfo;
-   int  status;
-
-
-   status = dPriv->driScreenPriv->DriverAPI.GetSwapInfo( dPriv, & sInfo );
-   *sbc = sInfo.swap_count;
-
-   return status;
+    return sPriv->DriverAPI.GetDrawableMSC(sPriv, dPriv, msc);
 }
 
-static int driWaitForSBC( __DRInativeDisplay * dpy, void *drawablePriv,
-			  int64_t target_sbc,
-			  int64_t * msc, int64_t * sbc )
-{
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePriv;
-
-    return dPriv->driScreenPriv->DriverAPI.WaitForSBC( dPriv, target_sbc,
-                                                       msc, sbc );
-}
 
-static int driWaitForMSC( __DRInativeDisplay * dpy, void *drawablePriv,
-			  int64_t target_msc,
-			  int64_t divisor, int64_t remainder,
-			  int64_t * msc, int64_t * sbc )
+static int driWaitForMSC(__DRIdrawable *dPriv, int64_t target_msc,
+			 int64_t divisor, int64_t remainder,
+			 int64_t * msc, int64_t * sbc)
 {
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePriv;
     __DRIswapInfo  sInfo;
     int  status;
 
-
     status = dPriv->driScreenPriv->DriverAPI.WaitForMSC( dPriv, target_msc,
                                                          divisor, remainder,
                                                          msc );
@@ -600,63 +351,72 @@ static int driWaitForMSC( __DRInativeDisplay * dpy, void *drawablePriv,
     return status;
 }
 
-static int64_t driSwapBuffersMSC( __DRInativeDisplay * dpy, void *drawablePriv,
-				  int64_t target_msc,
-				  int64_t divisor, int64_t remainder )
-{
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePriv;
 
-    return dPriv->driScreenPriv->DriverAPI.SwapBuffersMSC( dPriv, target_msc,
-                                                           divisor, 
-                                                           remainder );
-}
+const __DRImediaStreamCounterExtension driMediaStreamCounterExtension = {
+    { __DRI_MEDIA_STREAM_COUNTER, __DRI_MEDIA_STREAM_COUNTER_VERSION },
+    driWaitForMSC,
+    driDrawableGetMSC,
+};
+
 
-static void driCopySubBuffer( __DRInativeDisplay *dpy, void *drawablePrivate,
+static void driCopySubBuffer(__DRIdrawable *dPriv,
 			      int x, int y, int w, int h)
 {
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate;
+    drm_clip_rect_t rect;
+
+    rect.x1 = x;
+    rect.y1 = dPriv->h - y - h;
+    rect.x2 = x + w;
+    rect.y2 = rect.y1 + h;
+    driReportDamage(dPriv, &rect, 1);
+
     dPriv->driScreenPriv->DriverAPI.CopySubBuffer(dPriv, x, y, w, h);
-    (void) dpy;
 }
 
+const __DRIcopySubBufferExtension driCopySubBufferExtension = {
+    { __DRI_COPY_SUB_BUFFER, __DRI_COPY_SUB_BUFFER_VERSION },
+    driCopySubBuffer
+};
+
+static void driSetSwapInterval(__DRIdrawable *dPriv, unsigned int interval)
+{
+    dPriv->swap_interval = interval;
+}
+
+static unsigned int driGetSwapInterval(__DRIdrawable *dPriv)
+{
+    return dPriv->swap_interval;
+}
+
+const __DRIswapControlExtension driSwapControlExtension = {
+    { __DRI_SWAP_CONTROL, __DRI_SWAP_CONTROL_VERSION },
+    driSetSwapInterval,
+    driGetSwapInterval
+};
+
+
 /**
  * This is called via __DRIscreenRec's createNewDrawable pointer.
  */
-static void *driCreateNewDrawable(__DRInativeDisplay *dpy,
-				  const __GLcontextModes *modes,
-				  __DRIid draw,
-				  __DRIdrawable *pdraw,
-				  int renderType,
-				  const int *attrs)
+static __DRIdrawable *
+driCreateNewDrawable(__DRIscreen *psp, const __DRIconfig *config,
+		     drm_drawable_t hwDrawable, int renderType,
+		     const int *attrs, void *data)
 {
-    __DRIscreen * const pDRIScreen = (*dri_interface->getScreen)(dpy, modes->screen);
-    __DRIscreenPrivate *psp;
-    __DRIdrawablePrivate *pdp;
-
-
-    pdraw->private = NULL;
+    __DRIdrawable *pdp;
 
     /* Since pbuffers are not yet supported, no drawable attributes are
      * supported either.
      */
     (void) attrs;
 
-    if ( (pDRIScreen == NULL) || (pDRIScreen->private == NULL) ) {
-	return NULL;
-    }
-
-    pdp = (__DRIdrawablePrivate *)_mesa_malloc(sizeof(__DRIdrawablePrivate));
+    pdp = _mesa_malloc(sizeof *pdp);
     if (!pdp) {
 	return NULL;
     }
 
-    if (!(*dri_interface->createDrawable)(dpy, modes->screen, draw, &pdp->hHWDrawable)) {
-	_mesa_free(pdp);
-	return NULL;
-    }
-
-    pdp->draw = draw;
-    pdp->pdraw = pdraw;
+    pdp->loaderPrivate = data;
+    pdp->hHWDrawable = hwDrawable;
     pdp->refcount = 0;
     pdp->pStamp = NULL;
     pdp->lastStamp = 0;
@@ -669,80 +429,56 @@ static void *driCreateNewDrawable(__DRInativeDisplay *dpy,
     pdp->numBackClipRects = 0;
     pdp->pClipRects = NULL;
     pdp->pBackClipRects = NULL;
-    pdp->display = dpy;
-    pdp->screen = modes->screen;
+    pdp->vblSeq = 0;
+    pdp->vblFlags = 0;
 
-    psp = (__DRIscreenPrivate *)pDRIScreen->private;
     pdp->driScreenPriv = psp;
     pdp->driContextPriv = &psp->dummyContextPriv;
 
-    if (!(*psp->DriverAPI.CreateBuffer)(psp, pdp, modes,
+    if (!(*psp->DriverAPI.CreateBuffer)(psp, pdp, &config->modes,
 					renderType == GLX_PIXMAP_BIT)) {
-       (void)(*dri_interface->destroyDrawable)(dpy, modes->screen, pdp->draw);
        _mesa_free(pdp);
        return NULL;
     }
 
-    pdraw->private = pdp;
-    pdraw->destroyDrawable = driDestroyDrawable;
-    pdraw->swapBuffers = driSwapBuffers;  /* called by glXSwapBuffers() */
-
-    pdraw->getSBC = driGetSBC;
-    pdraw->waitForSBC = driWaitForSBC;
-    pdraw->waitForMSC = driWaitForMSC;
-    pdraw->swapBuffersMSC = driSwapBuffersMSC;
-    pdraw->frameTracking = NULL;
-    pdraw->queryFrameTracking = driQueryFrameTracking;
-
-    if (driCompareGLXAPIVersion (20060314) >= 0)
-	pdraw->copySubBuffer = driCopySubBuffer;
+    pdp->msc_base = 0;
 
     /* This special default value is replaced with the configured
      * default value when the drawable is first bound to a direct
      * rendering context. 
      */
-    pdraw->swap_interval = (unsigned)-1;
-
-    pdp->swapBuffers = psp->DriverAPI.SwapBuffers;
+    pdp->swap_interval = (unsigned)-1;
 
-    /* Add pdraw to drawable list */
-    if (!__driAddDrawable(psp->drawHash, pdraw)) {
-	/* ERROR!!! */
-	(*pdraw->destroyDrawable)(dpy, pdp);
-	_mesa_free(pdp);
-	pdp = NULL;
-	pdraw->private = NULL;
-    }
-
-   return (void *) pdp;
+    return pdp;
 }
 
+
 static __DRIdrawable *
-driGetDrawable(__DRInativeDisplay *dpy, __DRIid draw, void *screenPrivate)
+dri2CreateNewDrawable(__DRIscreen *screen,
+		      const __DRIconfig *config,
+		      void *loaderPrivate)
 {
-    __DRIscreenPrivate *psp = (__DRIscreenPrivate *) screenPrivate;
+    __DRIdrawable *pdraw;
 
-    /*
-    ** Make sure this routine returns NULL if the drawable is not bound
-    ** to a direct rendering context!
-    */
-    return __driFindDrawable(psp->drawHash, draw);
+    pdraw = driCreateNewDrawable(screen, config, 0, 0, NULL, loaderPrivate);
+    if (!pdraw)
+    	return NULL;
+
+    pdraw->pClipRects = _mesa_malloc(sizeof *pdraw->pBackClipRects);
+    pdraw->pBackClipRects = _mesa_malloc(sizeof *pdraw->pBackClipRects);
+
+    return pdraw;
 }
 
+
 static void
-driDestroyDrawable(__DRInativeDisplay *dpy, void *drawablePrivate)
+driDestroyDrawable(__DRIdrawable *pdp)
 {
-    __DRIdrawablePrivate *pdp = (__DRIdrawablePrivate *) drawablePrivate;
     __DRIscreenPrivate *psp;
-    int scrn;
 
     if (pdp) {
 	psp = pdp->driScreenPriv;
-	scrn = psp->myNum;
         (*psp->DriverAPI.DestroyBuffer)(pdp);
-	if ((*dri_interface->windowExists)(dpy, pdp->draw))
-	    (void)(*dri_interface->destroyDrawable)(dpy, scrn, pdp->draw);
-	drmHashDelete(psp->drawHash, pdp->draw);
 	if (pdp->pClipRects) {
 	    _mesa_free(pdp->pClipRects);
 	    pdp->pClipRects = NULL;
@@ -766,23 +502,15 @@ driDestroyDrawable(__DRInativeDisplay *dpy, void *drawablePrivate)
 /**
  * Destroy the per-context private information.
  * 
- * \param dpy the display handle.
- * \param scrn the screen number.
- * \param contextPrivate opaque pointer to the per-drawable private info.
- *
  * \internal
  * This function calls __DriverAPIRec::DestroyContext on \p contextPrivate, calls
  * drmDestroyContext(), and finally frees \p contextPrivate.
  */
 static void
-driDestroyContext(__DRInativeDisplay *dpy, int scrn, void *contextPrivate)
+driDestroyContext(__DRIcontext *pcp)
 {
-    __DRIcontextPrivate  *pcp   = (__DRIcontextPrivate *) contextPrivate;
-
     if (pcp) {
 	(*pcp->driScreenPriv->DriverAPI.DestroyContext)(pcp);
-	__driGarbageCollectDrawables(pcp->driScreenPriv->drawHash);
-	(void) (*dri_interface->destroyContext)(dpy, scrn, pcp->contextID);
 	_mesa_free(pcp);
     }
 }
@@ -791,13 +519,9 @@ driDestroyContext(__DRInativeDisplay *dpy, int scrn, void *contextPrivate)
 /**
  * Create the per-drawable private driver information.
  * 
- * \param dpy           The display handle.
- * \param modes         Mode used to create the new context.
  * \param render_type   Type of rendering target.  \c GLX_RGBA is the only
  *                      type likely to ever be supported for direct-rendering.
- * \param sharedPrivate The shared context dependent methods or \c NULL if
- *                      non-existent.
- * \param pctx          DRI context to receive the context dependent methods.
+ * \param shared        Context with which to share textures, etc. or NULL
  *
  * \returns An opaque pointer to the per-context private information on
  *          success, or \c NULL on failure.
@@ -809,36 +533,18 @@ driDestroyContext(__DRInativeDisplay *dpy, int scrn, void *contextPrivate)
  * context.
  *
  */
-static void *
-driCreateNewContext(__DRInativeDisplay *dpy, const __GLcontextModes *modes,
-		    int render_type, void *sharedPrivate, __DRIcontext *pctx)
+static __DRIcontext *
+driCreateNewContext(__DRIscreen *psp, const __DRIconfig *config,
+		    int render_type, __DRIcontext *shared, 
+		    drm_context_t hwContext, void *data)
 {
-    __DRIscreen *pDRIScreen;
-    __DRIcontextPrivate *pcp;
-    __DRIcontextPrivate *pshare = (__DRIcontextPrivate *) sharedPrivate;
-    __DRIscreenPrivate *psp;
-    void * const shareCtx = (pshare != NULL) ? pshare->driverPrivate : NULL;
-
-    pDRIScreen = (*dri_interface->getScreen)(dpy, modes->screen);
-    if ( (pDRIScreen == NULL) || (pDRIScreen->private == NULL) ) {
-	/* ERROR!!! */
-	return NULL;
-    } 
-
-    psp = (__DRIscreenPrivate *)pDRIScreen->private;
-
-    pcp = (__DRIcontextPrivate *)_mesa_malloc(sizeof(__DRIcontextPrivate));
-    if (!pcp) {
-	return NULL;
-    }
+    __DRIcontext *pcp;
+    void * const shareCtx = (shared != NULL) ? shared->driverPrivate : NULL;
 
-    if (! (*dri_interface->createContext)(dpy, modes->screen, modes->fbconfigID,
-					&pcp->contextID, &pcp->hHWContext)) {
-	_mesa_free(pcp);
+    pcp = _mesa_malloc(sizeof *pcp);
+    if (!pcp)
 	return NULL;
-    }
 
-    pcp->display = dpy;
     pcp->driScreenPriv = psp;
     pcp->driDrawablePriv = NULL;
 
@@ -846,8 +552,7 @@ driCreateNewContext(__DRInativeDisplay *dpy, const __GLcontextModes *modes,
      * context.
      */
 
-    if (!psp->dummyContextPriv.driScreenPriv) {
-        psp->dummyContextPriv.contextID = 0;
+    if (!psp->dri2.enabled && !psp->dummyContextPriv.driScreenPriv) {
         psp->dummyContextPriv.hHWContext = psp->pSAREA->dummy_context;
         psp->dummyContextPriv.driScreenPriv = psp;
         psp->dummyContextPriv.driDrawablePriv = NULL;
@@ -855,21 +560,31 @@ driCreateNewContext(__DRInativeDisplay *dpy, const __GLcontextModes *modes,
 	/* No other fields should be used! */
     }
 
-    pctx->destroyContext = driDestroyContext;
-    pctx->bindContext    = driBindContext;
-    pctx->unbindContext  = driUnbindContext;
+    pcp->hHWContext = hwContext;
 
-    if ( !(*psp->DriverAPI.CreateContext)(modes, pcp, shareCtx) ) {
-        (void) (*dri_interface->destroyContext)(dpy, modes->screen,
-						pcp->contextID);
+    if ( !(*psp->DriverAPI.CreateContext)(&config->modes, pcp, shareCtx) ) {
         _mesa_free(pcp);
         return NULL;
     }
 
-    __driGarbageCollectDrawables(pcp->driScreenPriv->drawHash);
-
     return pcp;
 }
+
+
+static __DRIcontext *
+dri2CreateNewContext(__DRIscreen *screen, const __DRIconfig *config,
+		      __DRIcontext *shared, void *data)
+{
+    return driCreateNewContext(screen, config, 0, shared, 0, data);
+}
+
+
+static int
+driCopyContext(__DRIcontext *dest, __DRIcontext *src, unsigned long mask)
+{
+    return GL_FALSE;
+}
+
 /*@}*/
 
 
@@ -881,18 +596,12 @@ driCreateNewContext(__DRInativeDisplay *dpy, const __GLcontextModes *modes,
 /**
  * Destroy the per-screen private information.
  * 
- * \param dpy the display handle.
- * \param scrn the screen number.
- * \param screenPrivate opaque pointer to the per-screen private information.
- *
  * \internal
  * This function calls __DriverAPIRec::DestroyScreen on \p screenPrivate, calls
  * drmClose(), and finally frees \p screenPrivate.
  */
-static void driDestroyScreen(__DRInativeDisplay *dpy, int scrn, void *screenPrivate)
+static void driDestroyScreen(__DRIscreen *psp)
 {
-    __DRIscreenPrivate *psp = (__DRIscreenPrivate *) screenPrivate;
-
     if (psp) {
 	/* No interaction with the X-server is possible at this point.  This
 	 * routine is called after XCloseDisplay, so there is no protocol
@@ -902,34 +611,44 @@ static void driDestroyScreen(__DRInativeDisplay *dpy, int scrn, void *screenPriv
 	if (psp->DriverAPI.DestroyScreen)
 	    (*psp->DriverAPI.DestroyScreen)(psp);
 
-	(void)drmUnmap((drmAddress)psp->pSAREA, SAREA_MAX);
-	(void)drmUnmap((drmAddress)psp->pFB, psp->fbSize);
-	_mesa_free(psp->pDevPriv);
-	(void)drmCloseOnce(psp->fd);
-	if ( psp->modes != NULL ) {
-	    (*dri_interface->destroyContextModes)( psp->modes );
+	if (!psp->dri2.enabled) {
+	   (void)drmUnmap((drmAddress)psp->pSAREA, SAREA_MAX);
+	   (void)drmUnmap((drmAddress)psp->pFB, psp->fbSize);
+	   (void)drmCloseOnce(psp->fd);
 	}
 
-	assert(psp->drawHash);
-	drmHashDestroy(psp->drawHash);
-
 	_mesa_free(psp);
     }
 }
 
+static void
+setupLoaderExtensions(__DRIscreen *psp,
+		      const __DRIextension **extensions)
+{
+    int i;
+
+    for (i = 0; extensions[i]; i++) {
+	if (strcmp(extensions[i]->name, __DRI_GET_DRAWABLE_INFO) == 0)
+	    psp->getDrawableInfo = (__DRIgetDrawableInfoExtension *) extensions[i];
+	if (strcmp(extensions[i]->name, __DRI_DAMAGE) == 0)
+	    psp->damage = (__DRIdamageExtension *) extensions[i];
+	if (strcmp(extensions[i]->name, __DRI_SYSTEM_TIME) == 0)
+	    psp->systemTime = (__DRIsystemTimeExtension *) extensions[i];
+	if (strcmp(extensions[i]->name, __DRI_DRI2_LOADER) == 0)
+	    psp->dri2.loader = (__DRIdri2LoaderExtension *) extensions[i];
+    }
+}
 
 /**
- * Utility function used to create a new driver-private screen structure.
+ * This is the bootstrap function for the driver.  libGL supplies all of the
+ * requisite information about the system, and the driver initializes itself.
+ * This routine also fills in the linked list pointed to by \c driver_modes
+ * with the \c __GLcontextModes that the driver can support for windows or
+ * pbuffers.
+ *
+ * For legacy DRI.
  * 
- * \param dpy   Display pointer
  * \param scrn  Index of the screen
- * \param psc   DRI screen data (not driver private)
- * \param modes Linked list of known display modes.  This list is, at a
- *              minimum, a list of modes based on the current display mode.
- *              These roughly match the set of available X11 visuals, but it
- *              need not be limited to X11!  The calling libGL should create
- *              a list that will inform the driver of the current display
- *              mode (i.e., color buffer depth, depth buffer depth, etc.).
  * \param ddx_version Version of the 2D DDX.  This may not be meaningful for
  *                    all drivers.
  * \param dri_version Version of the "server-side" DRI.
@@ -938,48 +657,33 @@ static void driDestroyScreen(__DRInativeDisplay *dpy, int scrn, void *screenPriv
  *                     framebuffer.
  * \param pSAREA       Pointer the the SAREA.
  * \param fd           Device handle for the DRM.
- * \param internal_api_version  Version of the internal interface between the
- *                              driver and libGL.
- * \param driverAPI Driver API functions used by other routines in dri_util.c.
+ * \param extensions   ??
+ * \param driver_modes  Returns modes suppoted by the driver
+ * \param loaderPrivate  ??
  * 
- * \note
- * There is no need to check the minimum API version in this function.  Since
- * the \c __driCreateNewScreen function is versioned, it is impossible for a
- * loader that is too old to even load this driver.
+ * \note There is no need to check the minimum API version in this
+ * function.  Since the name of this function is versioned, it is
+ * impossible for a loader that is too old to even load this driver.
  */
-__DRIscreenPrivate *
-__driUtilCreateNewScreen(__DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			 __GLcontextModes * modes,
-			 const __DRIversion * ddx_version,
-			 const __DRIversion * dri_version,
-			 const __DRIversion * drm_version,
-			 const __DRIframebuffer * frame_buffer,
-			 drm_sarea_t *pSAREA,
-			 int fd,
-			 int internal_api_version,
-			 const struct __DriverAPIRec *driverAPI)
+static __DRIscreen *
+driCreateNewScreen(int scrn,
+		   const __DRIversion *ddx_version,
+		   const __DRIversion *dri_version,
+		   const __DRIversion *drm_version,
+		   const __DRIframebuffer *frame_buffer,
+		   drmAddress pSAREA, int fd, 
+		   const __DRIextension **extensions,
+		   const __DRIconfig ***driver_modes,
+		   void *loaderPrivate)
 {
-    __DRIscreenPrivate *psp;
-
-
-    api_ver = internal_api_version;
-
-    psp = (__DRIscreenPrivate *)_mesa_calloc(sizeof(__DRIscreenPrivate));
-    if (!psp) {
-	return NULL;
-    }
+    static const __DRIextension *emptyExtensionList[] = { NULL };
+    __DRIscreen *psp;
 
-    /* Create the hash table */
-    psp->drawHash = drmHashCreate();
-    if ( psp->drawHash == NULL ) {
-	_mesa_free( psp );
+    psp = _mesa_calloc(sizeof *psp);
+    if (!psp)
 	return NULL;
-    }
 
-    psp->display = dpy;
-    psp->myNum = scrn;
-    psp->psc = psc;
-    psp->modes = modes;
+    setupLoaderExtensions(psp, extensions);
 
     /*
     ** NOT_DONE: This is used by the X server to detect when the client
@@ -988,20 +692,12 @@ __driUtilCreateNewScreen(__DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
     */
     psp->drawLockID = 1;
 
-    psp->drmMajor = drm_version->major;
-    psp->drmMinor = drm_version->minor;
-    psp->drmPatch = drm_version->patch;
-    psp->ddxMajor = ddx_version->major;
-    psp->ddxMinor = ddx_version->minor;
-    psp->ddxPatch = ddx_version->patch;
-    psp->driMajor = dri_version->major;
-    psp->driMinor = dri_version->minor;
-    psp->driPatch = dri_version->patch;
-
-    /* install driver's callback functions */
-    memcpy( &psp->DriverAPI, driverAPI, sizeof(struct __DriverAPIRec) );
+    psp->drm_version = *drm_version;
+    psp->ddx_version = *ddx_version;
+    psp->dri_version = *dri_version;
 
     psp->pSAREA = pSAREA;
+    psp->lock = (drmLock *) &psp->pSAREA->lock;
 
     psp->pFB = frame_buffer->base;
     psp->fbSize = frame_buffer->size;
@@ -1012,7 +708,10 @@ __driUtilCreateNewScreen(__DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
     psp->pDevPriv = frame_buffer->dev_priv;
     psp->fbBPP = psp->fbStride * 8 / frame_buffer->width;
 
+    psp->extensions = emptyExtensionList;
     psp->fd = fd;
+    psp->myNum = scrn;
+    psp->dri2.enabled = GL_FALSE;
 
     /*
     ** Do not init dummy context here; actual initialization will be
@@ -1021,63 +720,125 @@ __driUtilCreateNewScreen(__DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
     */
     psp->dummyContextPriv.driScreenPriv = NULL;
 
-    psc->destroyScreen     = driDestroyScreen;
-    psc->createNewDrawable = driCreateNewDrawable;
-    psc->getDrawable       = driGetDrawable;
-    psc->getMSC            = driGetMSC;
-    psc->createNewContext  = driCreateNewContext;
-
-    if (internal_api_version >= 20070121)
-	psc->setTexOffset  = psp->DriverAPI.setTexOffset;
+    psp->DriverAPI = driDriverAPI;
 
-    if ( (psp->DriverAPI.InitDriver != NULL)
-	 && !(*psp->DriverAPI.InitDriver)(psp) ) {
-	_mesa_free( psp );
+    *driver_modes = driDriverAPI.InitScreen(psp);
+    if (*driver_modes == NULL) {
+	_mesa_free(psp);
 	return NULL;
     }
 
-
     return psp;
 }
 
-
 /**
- * Compare the current GLX API version with a driver supplied required version.
- * 
- * The minimum required version is compared with the API version exported by
- * the \c __glXGetInternalVersion function (in libGL.so).
- * 
- * \param   required_version Minimum required internal GLX API version.
- * \return  A tri-value return, as from strcmp is returned.  A value less
- *          than, equal to, or greater than zero will be returned if the
- *          internal GLX API version is less than, equal to, or greater
- *          than \c required_version.
- *
- * \sa __glXGetInternalVersion().
+ * DRI2
  */
-int driCompareGLXAPIVersion( GLint required_version )
+static __DRIscreen *
+dri2CreateNewScreen(int scrn, int fd,
+		    const __DRIextension **extensions,
+		    const __DRIconfig ***driver_configs, void *data)
 {
-   if ( api_ver > required_version ) {
-      return 1;
-   }
-   else if ( api_ver == required_version ) {
-      return 0;
-   }
+    static const __DRIextension *emptyExtensionList[] = { NULL };
+    __DRIscreen *psp;
+    drmVersionPtr version;
+
+    if (driDriverAPI.InitScreen2 == NULL)
+        return NULL;
+
+    psp = _mesa_malloc(sizeof(*psp));
+    if (!psp)
+	return NULL;
+
+    setupLoaderExtensions(psp, extensions);
+
+    version = drmGetVersion(fd);
+    if (version) {
+	psp->drm_version.major = version->version_major;
+	psp->drm_version.minor = version->version_minor;
+	psp->drm_version.patch = version->version_patchlevel;
+	drmFreeVersion(version);
+    }
+
+    psp->extensions = emptyExtensionList;
+    psp->fd = fd;
+    psp->myNum = scrn;
+    psp->dri2.enabled = GL_TRUE;
 
-   return -1;
+    psp->DriverAPI = driDriverAPI;
+    *driver_configs = driDriverAPI.InitScreen2(psp);
+    if (*driver_configs == NULL) {
+	_mesa_free(psp);
+	return NULL;
+    }
+
+    psp->DriverAPI = driDriverAPI;
+
+    return psp;
 }
 
+static const __DRIextension **driGetExtensions(__DRIscreen *psp)
+{
+    return psp->extensions;
+}
+
+/** Core interface */
+const __DRIcoreExtension driCoreExtension = {
+    { __DRI_CORE, __DRI_CORE_VERSION },
+    NULL,
+    driDestroyScreen,
+    driGetExtensions,
+    driGetConfigAttrib,
+    driIndexConfigAttrib,
+    NULL,
+    driDestroyDrawable,
+    driSwapBuffers,
+    NULL,
+    driCopyContext,
+    driDestroyContext,
+    driBindContext,
+    driUnbindContext
+};
+
+/** Legacy DRI interface */
+const __DRIlegacyExtension driLegacyExtension = {
+    { __DRI_LEGACY, __DRI_LEGACY_VERSION },
+    driCreateNewScreen,
+    driCreateNewDrawable,
+    driCreateNewContext,
+};
+
+/** Legacy DRI interface */
+const __DRIdri2Extension driDRI2Extension = {
+    { __DRI_DRI2, __DRI_DRI2_VERSION },
+    dri2CreateNewScreen,
+    dri2CreateNewDrawable,
+    dri2CreateNewContext,
+};
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    &driDRI2Extension.base,
+    NULL
+};
 
 static int
-driQueryFrameTracking( __DRInativeDisplay * dpy, void * priv,
-		       int64_t * sbc, int64_t * missedFrames,
-		       float * lastMissedUsage, float * usage )
+driFrameTracking(__DRIdrawable *drawable, GLboolean enable)
+{
+    return GLX_BAD_CONTEXT;
+}
+
+static int
+driQueryFrameTracking(__DRIdrawable *dpriv,
+		      int64_t * sbc, int64_t * missedFrames,
+		      float * lastMissedUsage, float * usage)
 {
    __DRIswapInfo   sInfo;
    int             status;
    int64_t         ust;
-   __DRIdrawablePrivate * dpriv = (__DRIdrawablePrivate *) priv;
-
+   __DRIscreenPrivate *psp = dpriv->driScreenPriv;
 
    status = dpriv->driScreenPriv->DriverAPI.GetSwapInfo( dpriv, & sInfo );
    if ( status == 0 ) {
@@ -1085,13 +846,18 @@ driQueryFrameTracking( __DRInativeDisplay * dpy, void * priv,
       *missedFrames = sInfo.swap_missed_count;
       *lastMissedUsage = sInfo.swap_missed_usage;
 
-      (*dri_interface->getUST)( & ust );
+      (*psp->systemTime->getUST)( & ust );
       *usage = driCalculateSwapUsage( dpriv, sInfo.swap_ust, ust );
    }
 
    return status;
 }
 
+const __DRIframeTrackingExtension driFrameTrackingExtension = {
+    { __DRI_FRAME_TRACKING, __DRI_FRAME_TRACKING_VERSION },
+    driFrameTracking,
+    driQueryFrameTracking    
+};
 
 /**
  * Calculate amount of swap interval used between GLX buffer swaps.
@@ -1129,11 +895,10 @@ driCalculateSwapUsage( __DRIdrawablePrivate *dPriv, int64_t last_swap_ust,
    int32_t   d;
    int       interval;
    float     usage = 1.0;
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
-
-   if ( (*dri_interface->getMSCRate)( dPriv->display, dPriv->draw, &n, &d ) ) {
-      interval = (dPriv->pdraw->swap_interval != 0)
-	  ? dPriv->pdraw->swap_interval : 1;
+   if ( (*psp->systemTime->getMSCRate)(dPriv, &n, &d, dPriv->loaderPrivate) ) {
+      interval = (dPriv->swap_interval != 0) ? dPriv->swap_interval : 1;
 
 
       /* We want to calculate
diff --git a/src/mesa/drivers/dri/common/dri_util.h b/src/mesa/drivers/dri/common/dri_util.h
index c0cda5d25fc..0feb57b3c6e 100644
--- a/src/mesa/drivers/dri/common/dri_util.h
+++ b/src/mesa/drivers/dri/common/dri_util.h
@@ -48,21 +48,32 @@
 #define _DRI_UTIL_H_
 
 #include <GL/gl.h>
-#include "drm.h"
-#include "drm_sarea.h"
-#include "xf86drm.h"
+#include <drm.h>
+#include <drm_sarea.h>
+#include <xf86drm.h>
+#include "main/glheader.h"
 #include "GL/internal/glcore.h"
 #include "GL/internal/dri_interface.h"
 
 #define GLX_BAD_CONTEXT                    5
 
-typedef struct __DRIdisplayPrivateRec  __DRIdisplayPrivate;
-typedef struct __DRIscreenPrivateRec   __DRIscreenPrivate;
-typedef struct __DRIcontextPrivateRec  __DRIcontextPrivate;
-typedef struct __DRIdrawablePrivateRec __DRIdrawablePrivate;
 typedef struct __DRIswapInfoRec        __DRIswapInfo;
-typedef struct __DRIutilversionRec2    __DRIutilversion2;
 
+/* Typedefs to avoid rewriting the world. */
+typedef struct __DRIscreenRec	__DRIscreenPrivate;
+typedef struct __DRIdrawableRec	__DRIdrawablePrivate;
+typedef struct __DRIcontextRec	__DRIcontextPrivate;
+
+/**
+ * Extensions.
+ */
+extern const __DRIlegacyExtension driLegacyExtension;
+extern const __DRIcoreExtension driCoreExtension;
+extern const __DRIextension driReadDrawableExtension;
+extern const __DRIcopySubBufferExtension driCopySubBufferExtension;
+extern const __DRIswapControlExtension driSwapControlExtension;
+extern const __DRIframeTrackingExtension driFrameTrackingExtension;
+extern const __DRImediaStreamCounterExtension driMediaStreamCounterExtension;
 
 /**
  * Used by DRI_VALIDATE_DRAWABLE_INFO
@@ -78,7 +89,7 @@ typedef struct __DRIutilversionRec2    __DRIutilversion2;
 /**
  * Utility macro to validate the drawable information.
  *
- * See __DRIdrawablePrivate::pStamp and __DRIdrawablePrivate::lastStamp.
+ * See __DRIdrawable::pStamp and __DRIdrawable::lastStamp.
  */
 #define DRI_VALIDATE_DRAWABLE_INFO(psp, pdp)                            \
 do {                                                                    \
@@ -107,94 +118,95 @@ do {                                                                    \
  * this structure.
  */
 struct __DriverAPIRec {
-    /** 
-     * Driver initialization callback
-     */
-    GLboolean (*InitDriver)(__DRIscreenPrivate *driScrnPriv);
-    
+    const __DRIconfig **(*InitScreen) (__DRIscreen * priv);
+
     /**
      * Screen destruction callback
      */
-    void (*DestroyScreen)(__DRIscreenPrivate *driScrnPriv);
+    void (*DestroyScreen)(__DRIscreen *driScrnPriv);
 
     /**
      * Context creation callback
      */	    	    
     GLboolean (*CreateContext)(const __GLcontextModes *glVis,
-                               __DRIcontextPrivate *driContextPriv,
+                               __DRIcontext *driContextPriv,
                                void *sharedContextPrivate);
 
     /**
      * Context destruction callback
      */
-    void (*DestroyContext)(__DRIcontextPrivate *driContextPriv);
+    void (*DestroyContext)(__DRIcontext *driContextPriv);
 
     /**
      * Buffer (drawable) creation callback
      */
-    GLboolean (*CreateBuffer)(__DRIscreenPrivate *driScrnPriv,
-                              __DRIdrawablePrivate *driDrawPriv,
+    GLboolean (*CreateBuffer)(__DRIscreen *driScrnPriv,
+                              __DRIdrawable *driDrawPriv,
                               const __GLcontextModes *glVis,
                               GLboolean pixmapBuffer);
     
     /**
      * Buffer (drawable) destruction callback
      */
-    void (*DestroyBuffer)(__DRIdrawablePrivate *driDrawPriv);
+    void (*DestroyBuffer)(__DRIdrawable *driDrawPriv);
 
     /**
      * Buffer swapping callback 
      */
-    void (*SwapBuffers)(__DRIdrawablePrivate *driDrawPriv);
+    void (*SwapBuffers)(__DRIdrawable *driDrawPriv);
 
     /**
      * Context activation callback
      */
-    GLboolean (*MakeCurrent)(__DRIcontextPrivate *driContextPriv,
-                             __DRIdrawablePrivate *driDrawPriv,
-                             __DRIdrawablePrivate *driReadPriv);
+    GLboolean (*MakeCurrent)(__DRIcontext *driContextPriv,
+                             __DRIdrawable *driDrawPriv,
+                             __DRIdrawable *driReadPriv);
 
     /**
      * Context unbinding callback
      */
-    GLboolean (*UnbindContext)(__DRIcontextPrivate *driContextPriv);
+    GLboolean (*UnbindContext)(__DRIcontext *driContextPriv);
   
     /**
      * Retrieves statistics about buffer swap operations.  Required if
      * GLX_OML_sync_control or GLX_MESA_swap_frame_usage is supported.
      */
-    int (*GetSwapInfo)( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo );
-
+    int (*GetSwapInfo)( __DRIdrawable *dPriv, __DRIswapInfo * sInfo );
 
-    /**
-     * Required if GLX_SGI_video_sync or GLX_OML_sync_control is
-     * supported.
-     */
-    int (*GetMSC)( __DRIscreenPrivate * priv, int64_t * count );
 
     /**
      * These are required if GLX_OML_sync_control is supported.
      */
     /*@{*/
-    int (*WaitForMSC)( __DRIdrawablePrivate *priv, int64_t target_msc, 
+    int (*WaitForMSC)( __DRIdrawable *priv, int64_t target_msc, 
 		       int64_t divisor, int64_t remainder,
 		       int64_t * msc );
-    int (*WaitForSBC)( __DRIdrawablePrivate *priv, int64_t target_sbc,
+    int (*WaitForSBC)( __DRIdrawable *priv, int64_t target_sbc,
 		       int64_t * msc, int64_t * sbc );
 
-    int64_t (*SwapBuffersMSC)( __DRIdrawablePrivate *priv, int64_t target_msc,
+    int64_t (*SwapBuffersMSC)( __DRIdrawable *priv, int64_t target_msc,
 			       int64_t divisor, int64_t remainder );
     /*@}*/
-    void (*CopySubBuffer)(__DRIdrawablePrivate *driDrawPriv,
+    void (*CopySubBuffer)(__DRIdrawable *driDrawPriv,
 			  int x, int y, int w, int h);
 
     /**
-     * See corresponding field in \c __DRIscreenRec.
+     * New version of GetMSC so we can pass drawable data to the low
+     * level DRM driver (e.g. pipe info).  Required if
+     * GLX_SGI_video_sync or GLX_OML_sync_control is supported.
      */
-    void (*setTexOffset)(__DRIcontext *pDRICtx, GLint texname,
-			 unsigned long long offset, GLint depth, GLuint pitch);
+    int (*GetDrawableMSC) ( __DRIscreen * priv,
+			    __DRIdrawable *drawablePrivate,
+			    int64_t *count);
+
+
+
+    /* DRI2 Entry point */
+    const __DRIconfig **(*InitScreen2) (__DRIscreen * priv);
 };
 
+extern const struct __DriverAPIRec driDriverAPI;
+
 
 struct __DRIswapInfoRec {
     /** 
@@ -230,7 +242,7 @@ struct __DRIswapInfoRec {
 /**
  * Per-drawable private DRI driver information.
  */
-struct __DRIdrawablePrivateRec {
+struct __DRIdrawableRec {
     /**
      * Kernel drawable handle
      */
@@ -244,10 +256,10 @@ struct __DRIdrawablePrivateRec {
     void *driverPrivate;
 
     /**
-     * X's drawable ID associated with this private drawable.
+     * Private data from the loader.  We just hold on to it and pass
+     * it back when calling into loader provided functions.
      */
-    __DRIid draw;
-    __DRIdrawable *pdraw;
+    void *loaderPrivate;
 
     /**
      * Reference count for number of context's currently bound to this
@@ -272,7 +284,7 @@ struct __DRIdrawablePrivateRec {
     /**
      * Last value of the stamp.
      *
-     * If this differs from the value stored at __DRIdrawablePrivate::pStamp,
+     * If this differs from the value stored at __DRIdrawable::pStamp,
      * then the drawable information has been modified by the X server, and the
      * drawable information (below) should be retrieved from the X server.
      */
@@ -306,41 +318,52 @@ struct __DRIdrawablePrivateRec {
     /*@}*/
 
     /**
-     * Pointer to context to which this drawable is currently bound.
+     * \name Vertical blank tracking information
+     * Used for waiting on vertical blank events.
      */
-    __DRIcontextPrivate *driContextPriv;
+    /*@{*/
+    unsigned int vblSeq;
+    unsigned int vblFlags;
+    /*@}*/
 
     /**
-     * Pointer to screen on which this drawable was created.
+     * \name Monotonic MSC tracking
+     *
+     * Low level driver is responsible for updating msc_base and
+     * vblSeq values so that higher level code can calculate
+     * a new msc value or msc target for a WaitMSC call.  The new value
+     * will be:
+     *   msc = msc_base + get_vblank_count() - vblank_base;
+     *
+     * And for waiting on a value, core code will use:
+     *   actual_target = target_msc - msc_base + vblank_base;
      */
-    __DRIscreenPrivate *driScreenPriv;
+    /*@{*/
+    int64_t vblank_base;
+    int64_t msc_base;
+    /*@}*/
 
     /**
-     * \name Display and screen information.
-     * 
-     * Basically just need these for when the locking code needs to call
-     * \c __driUtilUpdateDrawableInfo.
+     * Pointer to context to which this drawable is currently bound.
      */
-    /*@{*/
-    __DRInativeDisplay *display;
-    int screen;
-    /*@}*/
+    __DRIcontext *driContextPriv;
 
     /**
-     * Called via glXSwapBuffers().
+     * Pointer to screen on which this drawable was created.
      */
-    void (*swapBuffers)( __DRIdrawablePrivate *dPriv );
+    __DRIscreen *driScreenPriv;
+
+    /**
+     * Controls swap interval as used by GLX_SGI_swap_control and
+     * GLX_MESA_swap_control.
+     */
+    unsigned int swap_interval;
 };
 
 /**
  * Per-context private driver information.
  */
-struct __DRIcontextPrivateRec {
-    /**
-     * Kernel context handle used to access the device lock.
-     */
-    __DRIid contextID;
-
+struct __DRIcontextRec {
     /**
      * Kernel context handle used to access the device lock.
      */
@@ -352,35 +375,30 @@ struct __DRIcontextPrivateRec {
     void *driverPrivate;
 
     /**
-     * This context's display pointer.
+     * Pointer back to the \c __DRIcontext that contains this structure.
      */
-    __DRInativeDisplay *display;
+    __DRIcontext *pctx;
 
     /**
      * Pointer to drawable currently bound to this context for drawing.
      */
-    __DRIdrawablePrivate *driDrawablePriv;
+    __DRIdrawable *driDrawablePriv;
 
     /**
      * Pointer to drawable currently bound to this context for reading.
      */
-    __DRIdrawablePrivate *driReadablePriv;
+    __DRIdrawable *driReadablePriv;
 
     /**
      * Pointer to screen on which this context was created.
      */
-    __DRIscreenPrivate *driScreenPriv;
+    __DRIscreen *driScreenPriv;
 };
 
 /**
  * Per-screen private driver information.
  */
-struct __DRIscreenPrivateRec {
-    /**
-     * Display for this screen
-     */
-    __DRInativeDisplay *display;
-
+struct __DRIscreenRec {
     /**
      * Current screen's number
      */
@@ -391,38 +409,21 @@ struct __DRIscreenPrivateRec {
      */
     struct __DriverAPIRec DriverAPI;
 
+    const __DRIextension **extensions;
     /**
-     * \name DDX version
      * DDX / 2D driver version information.
-     * \todo Replace these fields with a \c __DRIversionRec.
      */
-    /*@{*/
-    int ddxMajor;
-    int ddxMinor;
-    int ddxPatch;
-    /*@}*/
+    __DRIversion ddx_version;
 
     /**
-     * \name DRI version
      * DRI X extension version information.
-     * \todo Replace these fields with a \c __DRIversionRec.
      */
-    /*@{*/
-    int driMajor;
-    int driMinor;
-    int driPatch;
-    /*@}*/
+    __DRIversion dri_version;
 
     /**
-     * \name DRM version
      * DRM (kernel module) version information.
-     * \todo Replace these fields with a \c __DRIversionRec.
      */
-    /*@{*/
-    int drmMajor;
-    int drmMinor;
-    int drmPatch;
-    /*@}*/
+    __DRIversion drm_version;
 
     /**
      * ID used when the client sets the drawable lock.
@@ -485,12 +486,7 @@ struct __DRIscreenPrivateRec {
      * context is created when the first "real" context is created on this
      * screen.
      */
-    __DRIcontextPrivate dummyContextPriv;
-
-    /**
-     * Hash table to hold the drawable information for this screen.
-     */
-    void *drawHash;
+    __DRIcontext dummyContextPriv;
 
     /**
      * Device-dependent private information (not stored in the SAREA).
@@ -500,65 +496,38 @@ struct __DRIscreenPrivateRec {
     void *private;
 
     /**
-     * GLX visuals / FBConfigs for this screen.  These are stored as a
-     * linked list.
-     * 
-     * \note
-     * This field is \b only used in conjunction with the old interfaces.  If
-     * the new interfaces are used, this field will be set to \c NULL and will
-     * not be dereferenced.
-     */
-    __GLcontextModes *modes;
-
-    /**
      * Pointer back to the \c __DRIscreen that contains this structure.
      */
-
     __DRIscreen *psc;
-};
 
+    /* Extensions provided by the loader. */
+    const __DRIgetDrawableInfoExtension *getDrawableInfo;
+    const __DRIsystemTimeExtension *systemTime;
+    const __DRIdamageExtension *damage;
 
-/**
- * Used to store a version which includes a major range instead of a single
- * major version number.
- */
-struct __DRIutilversionRec2 {
-    int    major_min;    /** min allowed Major version number. */
-    int    major_max;    /** max allowed Major version number. */
-    int    minor;        /**< Minor version number. */
-    int    patch;        /**< Patch-level. */
-};
+    struct {
+	/* Flag to indicate that this is a DRI2 screen.  Many of the above
+	 * fields will not be valid or initializaed in that case. */
+	int enabled;
+	__DRIdri2LoaderExtension *loader;
+    } dri2;
 
+    /* The lock actually in use, old sarea or DRI2 */
+    drmLock *lock;
+};
 
 extern void
 __driUtilMessage(const char *f, ...);
 
 
 extern void
-__driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp);
-
-
-extern __DRIscreenPrivate * __driUtilCreateNewScreen( __DRInativeDisplay *dpy,
-    int scrn, __DRIscreen *psc, __GLcontextModes * modes,
-    const __DRIversion * ddx_version, const __DRIversion * dri_version,
-    const __DRIversion * drm_version, const __DRIframebuffer * frame_buffer,
-    drm_sarea_t *pSAREA, int fd, int internal_api_version,
-    const struct __DriverAPIRec *driverAPI );
-
-/* Test the version of the internal GLX API.  Returns a value like strcmp. */
-extern int
-driCompareGLXAPIVersion( GLint required_version );
+__driUtilUpdateDrawableInfo(__DRIdrawable *pdp);
 
 extern float
-driCalculateSwapUsage( __DRIdrawablePrivate *dPriv,
+driCalculateSwapUsage( __DRIdrawable *dPriv,
 		       int64_t last_swap_ust, int64_t current_ust );
 
-/**
- * Pointer to the \c __DRIinterfaceMethods passed to the driver by the loader.
- * 
- * This pointer is set in the driver's \c __driCreateNewScreen function and
- * is defined in dri_util.c.
- */
-extern const __DRIinterfaceMethods * dri_interface;
+extern GLint
+driIntersectArea( drm_clip_rect_t rect1, drm_clip_rect_t rect2 );
 
 #endif /* _DRI_UTIL_H_ */
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.c b/src/mesa/drivers/dri/common/drirenderbuffer.c
index d34da53479e..15af99136cd 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.c
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.c
@@ -1,9 +1,9 @@
 
-#include "mtypes.h"
+#include "main/mtypes.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
+#include "main/imports.h"
 #include "drirenderbuffer.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "imports.h"
 
 
 /**
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.h b/src/mesa/drivers/dri/common/drirenderbuffer.h
index 747f92fcdbe..cf55286b30f 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.h
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.h
@@ -10,7 +10,7 @@
 #ifndef DRIRENDERBUFFER_H
 #define DRIRENDERBUFFER_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "dri_util.h"
 
 
diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h
index 65e96657b8a..b977ebf0153 100644
--- a/src/mesa/drivers/dri/common/extension_helper.h
+++ b/src/mesa/drivers/dri/common/extension_helper.h
@@ -26,7 +26,7 @@
  */
 
 #include "utils.h"
-#include "dispatch.h"
+#include "glapi/dispatch.h"
 
 #ifndef NULL
 # define NULL 0
diff --git a/src/mesa/drivers/dri/common/memops.h b/src/mesa/drivers/dri/common/memops.h
index 4952d788e81..9cd1d8ec3fa 100644
--- a/src/mesa/drivers/dri/common/memops.h
+++ b/src/mesa/drivers/dri/common/memops.h
@@ -4,7 +4,7 @@
 * memset an area in I/O space
 * We need to be careful about this on some archs
 */
-static __inline__ void drimemsetio(void* address, int c, int size)
+static INLINE void drimemsetio(void* address, int c, int size)
 {
 #if defined(__powerpc__) || defined(__ia64__)
      int i;
diff --git a/src/mesa/drivers/dri/common/mmio.h b/src/mesa/drivers/dri/common/mmio.h
index 891056e1705..ce95d8c9075 100644
--- a/src/mesa/drivers/dri/common/mmio.h
+++ b/src/mesa/drivers/dri/common/mmio.h
@@ -33,11 +33,11 @@
 #ifndef MMIO_H
 #define MMIO_H
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 #if defined( __powerpc__ )
 
-static __inline__ uint32_t
+static INLINE uint32_t
 read_MMIO_LE32( volatile void * base, unsigned long offset )
 {
    uint32_t val;
@@ -50,7 +50,7 @@ read_MMIO_LE32( volatile void * base, unsigned long offset )
 
 #else
 
-static __inline__ uint32_t
+static INLINE uint32_t
 read_MMIO_LE32( volatile void * base, unsigned long offset )
 {
    volatile uint32_t * p = (volatile uint32_t *) (((volatile char *) base) + offset);
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h
index 50f3cf55816..f2868cb58a6 100644
--- a/src/mesa/drivers/dri/common/spantmp2.h
+++ b/src/mesa/drivers/dri/common/spantmp2.h
@@ -33,7 +33,7 @@
  * \author Ian Romanick <idr@us.ibm.com>
  */
 
-#include "colormac.h"
+#include "main/colormac.h"
 #include "spantmp_common.h"
 
 #ifndef DBG
@@ -48,36 +48,34 @@
 #define HW_WRITE_CLIPLOOP()	HW_CLIPLOOP()
 #endif
 
-
 #if (SPANTMP_PIXEL_FMT == GL_RGB)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_5_6_5)
 
 /**
  ** GL_RGB, GL_UNSIGNED_SHORT_5_6_5
  **/
 
+#ifndef GET_VALUE
 #ifndef GET_PTR
 #define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
 #endif
 
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
 #define INIT_MONO_PIXEL(p, color) \
   p = PACK_COLOR_565( color[0], color[1], color[2] )
 
 #define WRITE_RGBA( _x, _y, r, g, b, a )				\
-    do {                                                                \
-       GLshort * _p = (GLshort *) GET_PTR(_x, _y);                      \
-       _p[0] = ((((int)r & 0xf8) << 8) | (((int)g & 0xfc) << 3) |	\
-		   (((int)b & 0xf8) >> 3));                             \
-   } while(0)
+   PUT_VALUE(_x, _y, ((((int)r & 0xf8) << 8) |				\
+		      (((int)g & 0xfc) << 3) |				\
+		      (((int)b & 0xf8) >> 3)))				\
 
-#define WRITE_PIXEL( _x, _y, p )					\
-   do {                                                                 \
-      GLushort * _p = (GLushort *) GET_PTR(_x, _y);                     \
-      _p[0] = p;                                                        \
-   } while(0)
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
 
 #define READ_RGBA( rgba, _x, _y )					\
    do {									\
-      GLushort p = *(volatile GLshort *) GET_PTR(_x, _y);               \
+      GLushort p = GET_VALUE(_x, _y);					\
       rgba[0] = ((p >> 8) & 0xf8) * 255 / 0xf8;				\
       rgba[1] = ((p >> 3) & 0xfc) * 255 / 0xfc;				\
       rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8;				\
@@ -90,31 +88,32 @@
  ** GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV
  **/
 
+#ifndef GET_VALUE
 #ifndef GET_PTR
 #define GET_PTR(_x, _y) (     buf + (_x) * 4 + (_y) * pitch)
 #endif
 
+#define GET_VALUE(_x, _y) *(volatile GLuint *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLuint *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
 # define INIT_MONO_PIXEL(p, color)                       \
      p = PACK_COLOR_8888(color[3], color[0], color[1], color[2]) 
 
 # define WRITE_RGBA(_x, _y, r, g, b, a)                                 \
-    do {                                                                \
-       GLuint * _p = (GLuint *) GET_PTR(_x, _y);                        \
-       _p[0] = ((r << 16) | (g << 8) | (b << 0) | (a << 24));           \
-    } while(0)
+   PUT_VALUE(_x, _y, ((r << 16) |					\
+		      (g << 8) |					\
+		      (b << 0) |					\
+		      (a << 24)))
 
-#define WRITE_PIXEL(_x, _y, p)                                          \
-    do {                                                                \
-       GLuint * _p = (GLuint *) GET_PTR(_x, _y);                        \
-       _p[0] = p;                                                       \
-    } while(0)
+#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p)
 
 # if defined( USE_X86_ASM )
 #  define READ_RGBA(rgba, _x, _y)                                       \
     do {                                                                \
-        GLuint p = *(volatile GLuint *) GET_PTR(_x, _y);                \
+       GLuint p = GET_VALUE(_x, _y);					\
        __asm__ __volatile__( "bswap	%0; rorl $8, %0"                \
-				: "=r" (p) : "r" (p) );                 \
+				: "=r" (p) : "0" (p) );                 \
        ((GLuint *)rgba)[0] = p;                                         \
     } while (0)
 # elif defined( MESA_BIG_ENDIAN )
@@ -123,14 +122,14 @@
      */
 #  define READ_RGBA( rgba, _x, _y )				        \
      do {								\
-        GLuint p = *(volatile GLuint *) GET_PTR(_x, _y);                \
+        GLuint p = GET_VALUE(_x, _y);					\
         GLuint t = p;                                                   \
         *((uint32_t *) rgba) = (t >> 24) | (p << 8);                    \
      } while (0)
 # else
 #  define READ_RGBA( rgba, _x, _y )				        \
      do {								\
-        GLuint p = *(volatile GLuint *) GET_PTR(_x, _y);                \
+        GLuint p = GET_VALUE(_x, _y);					\
 	rgba[0] = (p >> 16) & 0xff;					\
 	rgba[1] = (p >>  8) & 0xff;					\
 	rgba[2] = (p >>  0) & 0xff;					\
@@ -389,7 +388,8 @@ static void TAG(ReadRGBASpan)( GLcontext *ctx,
 }
 
 
-#if defined(USE_MMX_ASM) && \
+#if defined(GET_PTR) && \
+   defined(USE_MMX_ASM) && \
    (((SPANTMP_PIXEL_FMT == GL_BGRA) && \
 	(SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)) || \
     ((SPANTMP_PIXEL_FMT == GL_RGB) && \
@@ -440,7 +440,8 @@ static void TAG2(ReadRGBASpan,_MMX)( GLcontext *ctx,
 #endif
 
 
-#if defined(USE_SSE_ASM) && \
+#if defined(GET_PTR) &&	\
+   defined(USE_SSE_ASM) && \
    (SPANTMP_PIXEL_FMT == GL_BGRA) && \
      (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
 static void TAG2(ReadRGBASpan,_SSE2)( GLcontext *ctx,
@@ -474,7 +475,8 @@ static void TAG2(ReadRGBASpan,_SSE2)( GLcontext *ctx,
 }
 #endif
 
-#if defined(USE_SSE_ASM) && \
+#if defined(GET_PTR) &&	\
+   defined(USE_SSE_ASM) && \
    (SPANTMP_PIXEL_FMT == GL_BGRA) && \
      (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
 static void TAG2(ReadRGBASpan,_SSE)( GLcontext *ctx,
@@ -567,6 +569,7 @@ static void TAG(InitPointers)(struct gl_renderbuffer *rb)
    rb->PutMonoValues = TAG(WriteMonoRGBAPixels);
    rb->GetValues = TAG(ReadRGBAPixels);
 
+#if defined(GET_PTR)
 #if defined(USE_SSE_ASM) && \
    (SPANTMP_PIXEL_FMT == GL_BGRA) && \
      (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
@@ -596,6 +599,7 @@ static void TAG(InitPointers)(struct gl_renderbuffer *rb)
    }
    else
 #endif
+#endif /* GET_PTR */
    {
       if (DBG) fprintf( stderr, "Using %s version of GetRow\n", "C" );
       rb->GetRow = TAG(ReadRGBASpan);
@@ -610,6 +614,8 @@ static void TAG(InitPointers)(struct gl_renderbuffer *rb)
 #undef READ_RGBA
 #undef TAG
 #undef TAG2
+#undef GET_VALUE
+#undef PUT_VALUE
 #undef GET_PTR
 #undef SPANTMP_PIXEL_FMT
 #undef SPANTMP_PIXEL_TYPE
diff --git a/src/mesa/drivers/dri/common/texmem.c b/src/mesa/drivers/dri/common/texmem.c
index a81cc2413d1..ff174a251d2 100644
--- a/src/mesa/drivers/dri/common/texmem.c
+++ b/src/mesa/drivers/dri/common/texmem.c
@@ -43,10 +43,10 @@
  */
 
 #include "texmem.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "macros.h"
-#include "texformat.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/texformat.h"
 
 #include <assert.h>
 
@@ -1277,6 +1277,7 @@ driCalculateTextureFirstLastLevel( driTextureObject * t )
       else {
 	 firstLevel = tObj->BaseLevel + (GLint)(tObj->MinLod + 0.5);
 	 firstLevel = MAX2(firstLevel, tObj->BaseLevel);
+	 firstLevel = MIN2(firstLevel, tObj->BaseLevel + baseImage->MaxLog2);
 	 lastLevel = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5);
 	 lastLevel = MAX2(lastLevel, t->tObj->BaseLevel);
 	 lastLevel = MIN2(lastLevel, t->tObj->BaseLevel + baseImage->MaxLog2);
diff --git a/src/mesa/drivers/dri/common/texmem.h b/src/mesa/drivers/dri/common/texmem.h
index ffed7dd66e6..9c065da8b4f 100644
--- a/src/mesa/drivers/dri/common/texmem.h
+++ b/src/mesa/drivers/dri/common/texmem.h
@@ -38,8 +38,8 @@
 #ifndef DRI_TEXMEM_H
 #define DRI_TEXMEM_H
 
-#include "mtypes.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
 #include "xf86drm.h"
 
 struct dri_tex_heap;
diff --git a/src/mesa/drivers/dri/common/utils.c b/src/mesa/drivers/dri/common/utils.c
index 67cffead00f..2a1ded3871d 100644
--- a/src/mesa/drivers/dri/common/utils.c
+++ b/src/mesa/drivers/dri/common/utils.c
@@ -31,10 +31,10 @@
 
 #include <string.h>
 #include <stdlib.h>
-#include "mtypes.h"
-#include "extensions.h"
+#include "main/mtypes.h"
+#include "main/extensions.h"
+#include "glapi/dispatch.h"
 #include "utils.h"
-#include "dispatch.h"
 
 int driDispatchRemapTable[ driDispatchRemapTable_size ];
 
@@ -421,21 +421,6 @@ driCheckDriDdxDrmVersions2(const char * driver_name,
 				drmActual, drmExpected);
 }
 
-
-
-GLint
-driIntersectArea( drm_clip_rect_t rect1, drm_clip_rect_t rect2 )
-{
-   if (rect2.x1 > rect1.x1) rect1.x1 = rect2.x1;
-   if (rect2.x2 < rect1.x2) rect1.x2 = rect2.x2;
-   if (rect2.y1 > rect1.y1) rect1.y1 = rect2.y1;
-   if (rect2.y2 < rect1.y2) rect1.y2 = rect2.y2;
-
-   if (rect1.x1 > rect1.x2 || rect1.y1 > rect1.y2) return 0;
-
-   return (rect1.x2 - rect1.x1) * (rect1.y2 - rect1.y1);
-}
-
 GLboolean driClipRectToFramebuffer( const GLframebuffer *buffer,
 				    GLint *x, GLint *y,
 				    GLsizei *width, GLsizei *height )
@@ -469,8 +454,6 @@ GLboolean driClipRectToFramebuffer( const GLframebuffer *buffer,
    return GL_TRUE;
 }
 
-
-
 /**
  * Creates a set of \c __GLcontextModes that a driver will expose.
  * 
@@ -541,85 +524,98 @@ GLboolean driClipRectToFramebuffer( const GLframebuffer *buffer,
  * \c GL_UNSIGNED_3BYTE_8_8_8, \c GL_4FLOAT_32_32_32_32, 
  * \c GL_4HALF_16_16_16_16, etc.  We can cross that bridge when we come to it.
  */
-GLboolean
-driFillInModes( __GLcontextModes ** ptr_to_modes,
-		GLenum fb_format, GLenum fb_type,
-		const uint8_t * depth_bits, const uint8_t * stencil_bits,
-		unsigned num_depth_stencil_bits,
-		const GLenum * db_modes, unsigned num_db_modes,
-		const u_int8_t * msaa_samples, unsigned num_msaa_modes,
-		int visType )
+__DRIconfig **
+driCreateConfigs(GLenum fb_format, GLenum fb_type,
+		 const uint8_t * depth_bits, const uint8_t * stencil_bits,
+		 unsigned num_depth_stencil_bits,
+		 const GLenum * db_modes, unsigned num_db_modes,
+	 	 const u_int8_t * msaa_samples, unsigned num_msaa_modes)
 {
-   static const uint8_t bits_table[3][4] = {
+   static const uint8_t bits_table[4][4] = {
      /* R  G  B  A */
+      { 3, 3, 2, 0 }, /* Any GL_UNSIGNED_BYTE_3_3_2 */
       { 5, 6, 5, 0 }, /* Any GL_UNSIGNED_SHORT_5_6_5 */
       { 8, 8, 8, 0 }, /* Any RGB with any GL_UNSIGNED_INT_8_8_8_8 */
       { 8, 8, 8, 8 }  /* Any RGBA with any GL_UNSIGNED_INT_8_8_8_8 */
    };
 
-   /* The following arrays are all indexed by the fb_type masked with 0x07.
-    * Given the four supported fb_type values, this results in valid array
-    * indices of 3, 4, 5, and 7.
-    */
-   static const uint32_t masks_table_rgb[8][4] = {
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
+   static const uint32_t masks_table_rgb[6][4] = {
+      { 0x000000E0, 0x0000001C, 0x00000003, 0x00000000 }, /* 3_3_2       */
+      { 0x00000007, 0x00000038, 0x000000C0, 0x00000000 }, /* 2_3_3_REV   */
       { 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 }, /* 5_6_5       */
       { 0x0000001F, 0x000007E0, 0x0000F800, 0x00000000 }, /* 5_6_5_REV   */
       { 0xFF000000, 0x00FF0000, 0x0000FF00, 0x00000000 }, /* 8_8_8_8     */
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
       { 0x000000FF, 0x0000FF00, 0x00FF0000, 0x00000000 }  /* 8_8_8_8_REV */
    };
 
-   static const uint32_t masks_table_rgba[8][4] = {
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
+   static const uint32_t masks_table_rgba[6][4] = {
+      { 0x000000E0, 0x0000001C, 0x00000003, 0x00000000 }, /* 3_3_2       */
+      { 0x00000007, 0x00000038, 0x000000C0, 0x00000000 }, /* 2_3_3_REV   */
       { 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 }, /* 5_6_5       */
       { 0x0000001F, 0x000007E0, 0x0000F800, 0x00000000 }, /* 5_6_5_REV   */
       { 0xFF000000, 0x00FF0000, 0x0000FF00, 0x000000FF }, /* 8_8_8_8     */
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
       { 0x000000FF, 0x0000FF00, 0x00FF0000, 0xFF000000 }, /* 8_8_8_8_REV */
    };
 
-   static const uint32_t masks_table_bgr[8][4] = {
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
+   static const uint32_t masks_table_bgr[6][4] = {
+      { 0x00000007, 0x00000038, 0x000000C0, 0x00000000 }, /* 3_3_2       */
+      { 0x000000E0, 0x0000001C, 0x00000003, 0x00000000 }, /* 2_3_3_REV   */
       { 0x0000001F, 0x000007E0, 0x0000F800, 0x00000000 }, /* 5_6_5       */
       { 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 }, /* 5_6_5_REV   */
       { 0x0000FF00, 0x00FF0000, 0xFF000000, 0x00000000 }, /* 8_8_8_8     */
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
       { 0x00FF0000, 0x0000FF00, 0x000000FF, 0x00000000 }, /* 8_8_8_8_REV */
    };
 
-   static const uint32_t masks_table_bgra[8][4] = {
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
+   static const uint32_t masks_table_bgra[6][4] = {
+      { 0x00000007, 0x00000038, 0x000000C0, 0x00000000 }, /* 3_3_2       */
+      { 0x000000E0, 0x0000001C, 0x00000003, 0x00000000 }, /* 2_3_3_REV   */
       { 0x0000001F, 0x000007E0, 0x0000F800, 0x00000000 }, /* 5_6_5       */
       { 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 }, /* 5_6_5_REV   */
       { 0x0000FF00, 0x00FF0000, 0xFF000000, 0x000000FF }, /* 8_8_8_8     */
-      { 0x00000000, 0x00000000, 0x00000000, 0x00000000 },
       { 0x00FF0000, 0x0000FF00, 0x000000FF, 0xFF000000 }, /* 8_8_8_8_REV */
    };
 
-   static const uint8_t bytes_per_pixel[8] = {
-      0, 0, 0, 2, 2, 4, 0, 4
+   static const uint8_t bytes_per_pixel[6] = {
+      1, /* 3_3_2       */
+      1, /* 2_3_3_REV   */
+      2, /* 5_6_5       */
+      2, /* 5_6_5_REV   */
+      4, /* 8_8_8_8     */
+      4  /* 8_8_8_8_REV */
    };
 
    const uint8_t  * bits;
    const uint32_t * masks;
-   const int index = fb_type & 0x07;
-   __GLcontextModes * modes = *ptr_to_modes;
+   int index;
+   __DRIconfig **configs, **c;
+   __GLcontextModes *modes;
    unsigned i, j, k, h;
-
-
-   if ( bytes_per_pixel[ index ] == 0 ) {
-      fprintf( stderr, "[%s:%u] Framebuffer type 0x%04x has 0 bytes per pixel.\n",
-	       __FUNCTION__, __LINE__, fb_type );
-      return GL_FALSE;
+   unsigned num_modes;
+   unsigned num_accum_bits = 2;
+
+   switch ( fb_type ) {
+      case GL_UNSIGNED_BYTE_3_3_2:
+	 index = 0;
+	 break;
+      case GL_UNSIGNED_BYTE_2_3_3_REV:
+	 index = 1;
+	 break;
+      case GL_UNSIGNED_SHORT_5_6_5:
+	 index = 2;
+	 break;
+      case GL_UNSIGNED_SHORT_5_6_5_REV:
+	 index = 3;
+	 break;
+      case GL_UNSIGNED_INT_8_8_8_8:
+	 index = 4;
+	 break;
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+	 index = 5;
+	 break;
+      default:
+	 fprintf( stderr, "[%s:%u] Unknown framebuffer type 0x%04x.\n",
+               __FUNCTION__, __LINE__, fb_type );
+	 return NULL;
    }
 
 
@@ -631,41 +627,56 @@ driFillInModes( __GLcontextModes ** ptr_to_modes,
 
    switch ( fb_format ) {
       case GL_RGB:
-         bits = (bytes_per_pixel[ index ] == 2)
-	     ? bits_table[0] : bits_table[1];
          masks = masks_table_rgb[ index ];
          break;
 
       case GL_RGBA:
-         bits = (bytes_per_pixel[ index ] == 2)
-	     ? bits_table[0] : bits_table[2];
          masks = masks_table_rgba[ index ];
          break;
 
       case GL_BGR:
-         bits = (bytes_per_pixel[ index ] == 2)
-	     ? bits_table[0] : bits_table[1];
          masks = masks_table_bgr[ index ];
          break;
 
       case GL_BGRA:
-         bits = (bytes_per_pixel[ index ] == 2)
-	     ? bits_table[0] : bits_table[2];
          masks = masks_table_bgra[ index ];
          break;
 
       default:
-         fprintf( stderr, "[%s:%u] Framebuffer format 0x%04x is not GL_RGB, GL_RGBA, GL_BGR, or GL_BGRA.\n",
-	       __FUNCTION__, __LINE__, fb_format );
-         return GL_FALSE;
+         fprintf( stderr, "[%s:%u] Unknown framebuffer format 0x%04x.\n",
+               __FUNCTION__, __LINE__, fb_format );
+         return NULL;
+   }
+
+   switch ( bytes_per_pixel[ index ] ) {
+      case 1:
+	 bits = bits_table[0];
+	 break;
+      case 2:
+	 bits = bits_table[1];
+	 break;
+      default:
+	 bits = ((fb_format == GL_RGB) || (fb_format == GL_BGR))
+	    ? bits_table[2]
+	    : bits_table[3];
+	 break;
    }
 
+   num_modes = num_depth_stencil_bits * num_db_modes * num_accum_bits * num_msaa_modes;
+   configs = _mesa_calloc((num_modes + 1) * sizeof *configs);
+   if (configs == NULL)
+       return NULL;
 
+    c = configs;
     for ( k = 0 ; k < num_depth_stencil_bits ; k++ ) {
 	for ( i = 0 ; i < num_db_modes ; i++ ) {
 	    for ( h = 0 ; h < num_msaa_modes; h++ ) {
-		for ( j = 0 ; j < 2 ; j++ ) {
+	    	for ( j = 0 ; j < num_accum_bits ; j++ ) {
+		    *c = _mesa_malloc (sizeof **c);
+		    modes = &(*c)->modes;
+		    c++;
 
+		    memset(modes, 0, sizeof *modes);
 		    modes->redBits   = bits[0];
 		    modes->greenBits = bits[1];
 		    modes->blueBits  = bits[2];
@@ -675,7 +686,7 @@ driFillInModes( __GLcontextModes ** ptr_to_modes,
 		    modes->blueMask  = masks[2];
 		    modes->alphaMask = masks[3];
 		    modes->rgbBits   = modes->redBits + modes->greenBits
-			+ modes->blueBits + modes->alphaBits;
+		    	+ modes->blueBits + modes->alphaBits;
 
 		    modes->accumRedBits   = 16 * j;
 		    modes->accumGreenBits = 16 * j;
@@ -686,35 +697,190 @@ driFillInModes( __GLcontextModes ** ptr_to_modes,
 		    modes->stencilBits = stencil_bits[k];
 		    modes->depthBits = depth_bits[k];
 
-		    modes->visualType = visType;
+		    modes->transparentPixel = GLX_NONE;
+		    modes->transparentRed = GLX_DONT_CARE;
+		    modes->transparentGreen = GLX_DONT_CARE;
+		    modes->transparentBlue = GLX_DONT_CARE;
+		    modes->transparentAlpha = GLX_DONT_CARE;
+		    modes->transparentIndex = GLX_DONT_CARE;
+		    modes->visualType = GLX_DONT_CARE;
 		    modes->renderType = GLX_RGBA_BIT;
 		    modes->drawableType = GLX_WINDOW_BIT;
 		    modes->rgbMode = GL_TRUE;
 
 		    if ( db_modes[i] == GLX_NONE ) {
-			modes->doubleBufferMode = GL_FALSE;
+		    	modes->doubleBufferMode = GL_FALSE;
 		    }
 		    else {
-			modes->doubleBufferMode = GL_TRUE;
-			modes->swapMethod = db_modes[i];
+		    	modes->doubleBufferMode = GL_TRUE;
+		    	modes->swapMethod = db_modes[i];
 		    }
 
 		    modes->samples = msaa_samples[h];
 		    modes->sampleBuffers = modes->samples ? 1 : 0;
 
+
 		    modes->haveAccumBuffer = ((modes->accumRedBits +
-					       modes->accumGreenBits +
-					       modes->accumBlueBits +
-					       modes->accumAlphaBits) > 0);
+					   modes->accumGreenBits +
+					   modes->accumBlueBits +
+					   modes->accumAlphaBits) > 0);
 		    modes->haveDepthBuffer = (modes->depthBits > 0);
 		    modes->haveStencilBuffer = (modes->stencilBits > 0);
 
-		    modes = modes->next;
+		    modes->bindToTextureRgb = GL_TRUE;
+		    modes->bindToTextureRgba = GL_TRUE;
+		    modes->bindToMipmapTexture = GL_FALSE;
+		    modes->bindToTextureTargets = modes->rgbMode ?
+		    	__DRI_ATTRIB_TEXTURE_1D_BIT |
+		    	__DRI_ATTRIB_TEXTURE_2D_BIT |
+		    	__DRI_ATTRIB_TEXTURE_RECTANGLE_BIT :
+		    	0;
 		}
 	    }
 	}
     }
+    *c = NULL;
+
+    return configs;
+}
+
+const __DRIconfig **driConcatConfigs(__DRIconfig **a, __DRIconfig **b)
+{
+    const __DRIconfig **all;
+    int i, j, index;
+
+    i = 0;
+    while (a[i] != NULL)
+	i++;
+    j = 0;
+    while (b[j] != NULL)
+	j++;
+   
+    all = _mesa_malloc((i + j + 1) * sizeof *all);
+    index = 0;
+    for (i = 0; a[i] != NULL; i++)
+	all[index++] = a[i];
+    for (j = 0; b[j] != NULL; j++)
+	all[index++] = b[j];
+    all[index++] = NULL;
+
+    _mesa_free(a);
+    _mesa_free(b);
+
+    return all;
+}
+
+#define __ATTRIB(attrib, field) \
+    { attrib, offsetof(__GLcontextModes, field) }
+
+static const struct { unsigned int attrib, offset; } attribMap[] = {
+    __ATTRIB(__DRI_ATTRIB_BUFFER_SIZE,			rgbBits),
+    __ATTRIB(__DRI_ATTRIB_LEVEL,			level),
+    __ATTRIB(__DRI_ATTRIB_RED_SIZE,			redBits),
+    __ATTRIB(__DRI_ATTRIB_GREEN_SIZE,			greenBits),
+    __ATTRIB(__DRI_ATTRIB_BLUE_SIZE,			blueBits),
+    __ATTRIB(__DRI_ATTRIB_ALPHA_SIZE,			alphaBits),
+    __ATTRIB(__DRI_ATTRIB_DEPTH_SIZE,			depthBits),
+    __ATTRIB(__DRI_ATTRIB_STENCIL_SIZE,			stencilBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_RED_SIZE,		accumRedBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_GREEN_SIZE,		accumGreenBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_BLUE_SIZE,		accumBlueBits),
+    __ATTRIB(__DRI_ATTRIB_ACCUM_ALPHA_SIZE,		accumAlphaBits),
+    __ATTRIB(__DRI_ATTRIB_SAMPLE_BUFFERS,		sampleBuffers),
+    __ATTRIB(__DRI_ATTRIB_SAMPLES,			samples),
+    __ATTRIB(__DRI_ATTRIB_DOUBLE_BUFFER,		doubleBufferMode),
+    __ATTRIB(__DRI_ATTRIB_STEREO,			stereoMode),
+    __ATTRIB(__DRI_ATTRIB_AUX_BUFFERS,			numAuxBuffers),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_TYPE,		transparentPixel),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_INDEX_VALUE,	transparentPixel),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_RED_VALUE,	transparentRed),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_GREEN_VALUE,	transparentGreen),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_BLUE_VALUE,	transparentBlue),
+    __ATTRIB(__DRI_ATTRIB_TRANSPARENT_ALPHA_VALUE,	transparentAlpha),
+    __ATTRIB(__DRI_ATTRIB_FLOAT_MODE,			floatMode),
+    __ATTRIB(__DRI_ATTRIB_RED_MASK,			redMask),
+    __ATTRIB(__DRI_ATTRIB_GREEN_MASK,			greenMask),
+    __ATTRIB(__DRI_ATTRIB_BLUE_MASK,			blueMask),
+    __ATTRIB(__DRI_ATTRIB_ALPHA_MASK,			alphaMask),
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_WIDTH,		maxPbufferWidth),
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_HEIGHT,		maxPbufferHeight),
+    __ATTRIB(__DRI_ATTRIB_MAX_PBUFFER_PIXELS,		maxPbufferPixels),
+    __ATTRIB(__DRI_ATTRIB_OPTIMAL_PBUFFER_WIDTH,	optimalPbufferWidth),
+    __ATTRIB(__DRI_ATTRIB_OPTIMAL_PBUFFER_HEIGHT,	optimalPbufferHeight),
+    __ATTRIB(__DRI_ATTRIB_SWAP_METHOD,			swapMethod),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_RGB,		bindToTextureRgb),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_RGBA,		bindToTextureRgba),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_MIPMAP_TEXTURE,	bindToMipmapTexture),
+    __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_TARGETS,	bindToTextureTargets),
+    __ATTRIB(__DRI_ATTRIB_YINVERTED,			yInverted),
+
+    /* The struct field doesn't matter here, these are handled by the
+     * switch in driGetConfigAttribIndex.  We need them in the array
+     * so the iterator includes them though.*/
+    __ATTRIB(__DRI_ATTRIB_RENDER_TYPE,			level),
+    __ATTRIB(__DRI_ATTRIB_CONFIG_CAVEAT,		level),
+    __ATTRIB(__DRI_ATTRIB_SWAP_METHOD,			level)
+};
+
+#define ARRAY_SIZE(a) (sizeof (a) / sizeof ((a)[0]))
+
+static int
+driGetConfigAttribIndex(const __DRIconfig *config,
+			unsigned int index, unsigned int *value)
+{
+    switch (attribMap[index].attrib) {
+    case __DRI_ATTRIB_RENDER_TYPE:
+	if (config->modes.rgbMode)
+	    *value = __DRI_ATTRIB_RGBA_BIT;
+	else
+	    *value = __DRI_ATTRIB_COLOR_INDEX_BIT;
+	break;
+    case __DRI_ATTRIB_CONFIG_CAVEAT:
+	if (config->modes.visualRating == GLX_NON_CONFORMANT_CONFIG)
+	    *value = __DRI_ATTRIB_NON_CONFORMANT_CONFIG;
+	else if (config->modes.visualRating == GLX_SLOW_CONFIG)
+	    *value = __DRI_ATTRIB_SLOW_BIT;
+	else
+	    *value = 0;
+	break;
+    case __DRI_ATTRIB_SWAP_METHOD:
+	break;
+
+    case __DRI_ATTRIB_FLOAT_MODE:
+        *value = config->modes.floatMode;
+        break;
+
+    default:
+	*value = *(unsigned int *)
+	    ((char *) &config->modes + attribMap[index].offset);
+	
+	break;
+    }
 
-    *ptr_to_modes = modes;
     return GL_TRUE;
 }
+
+int
+driGetConfigAttrib(const __DRIconfig *config,
+		   unsigned int attrib, unsigned int *value)
+{
+    int i;
+
+    for (i = 0; i < ARRAY_SIZE(attribMap); i++)
+	if (attribMap[i].attrib == attrib)
+	    return driGetConfigAttribIndex(config, i, value);
+
+    return GL_FALSE;
+}
+
+int
+driIndexConfigAttrib(const __DRIconfig *config, int index,
+		     unsigned int *attrib, unsigned int *value)
+{
+    if (index >= 0 && index < ARRAY_SIZE(attribMap)) {
+	*attrib = attribMap[index].attrib;
+	return driGetConfigAttribIndex(config, index, value);
+    }
+
+    return GL_FALSE;
+}
diff --git a/src/mesa/drivers/dri/common/utils.h b/src/mesa/drivers/dri/common/utils.h
index 20940c21b4e..4e27bd21a1f 100644
--- a/src/mesa/drivers/dri/common/utils.h
+++ b/src/mesa/drivers/dri/common/utils.h
@@ -28,8 +28,11 @@
 #ifndef DRI_DEBUG_H
 #define DRI_DEBUG_H
 
-#include "context.h"
-#include "dri_util.h"
+#include <GL/gl.h>
+#include <GL/internal/dri_interface.h>
+#include "main/context.h"
+
+typedef struct __DRIutilversionRec2    __DRIutilversion2;
 
 struct dri_debug_control {
     const char * string;
@@ -83,6 +86,17 @@ struct dri_extension {
     const struct dri_extension_function * functions;
 };
 
+/**
+ * Used to store a version which includes a major range instead of a single
+ * major version number.
+ */
+struct __DRIutilversionRec2 {
+    int    major_min;    /** min allowed Major version number. */
+    int    major_max;    /** max allowed Major version number. */
+    int    minor;        /**< Minor version number. */
+    int    patch;        /**< Patch-level. */
+};
+
 extern unsigned driParseDebugString( const char * debug,
     const struct dri_debug_control * control );
 
@@ -105,17 +119,28 @@ extern GLboolean driCheckDriDdxDrmVersions3(const char * driver_name,
     const __DRIversion * ddxActual, const __DRIutilversion2 * ddxExpected,
     const __DRIversion * drmActual, const __DRIversion * drmExpected);
 
-extern GLint driIntersectArea( drm_clip_rect_t rect1, drm_clip_rect_t rect2 );
-
 extern GLboolean driClipRectToFramebuffer( const GLframebuffer *buffer,
 					   GLint *x, GLint *y,
 					   GLsizei *width, GLsizei *height );
 
-extern GLboolean driFillInModes( __GLcontextModes ** modes,
-    GLenum fb_format, GLenum fb_type,
-    const uint8_t * depth_bits, const uint8_t * stencil_bits,
-    unsigned num_depth_stencil_bits,
-    const GLenum * db_modes, unsigned num_db_modes,
-    const u_int8_t * msaa_samples, unsigned num_msaa_modes, int visType );
+struct __DRIconfigRec {
+    __GLcontextModes modes;
+};
+
+extern __DRIconfig **
+driCreateConfigs(GLenum fb_format, GLenum fb_type,
+		 const uint8_t * depth_bits, const uint8_t * stencil_bits,
+		 unsigned num_depth_stencil_bits,
+		 const GLenum * db_modes, unsigned num_db_modes,
+    		 const uint8_t * msaa_samples, unsigned num_msaa_modes);
+
+const __DRIconfig **driConcatConfigs(__DRIconfig **a, __DRIconfig **b);
+
+int
+driGetConfigAttrib(const __DRIconfig *config,
+		   unsigned int attrib, unsigned int *value);
+int
+driIndexConfigAttrib(const __DRIconfig *config, int index,
+		     unsigned int *attrib, unsigned int *value);
 
 #endif /* DRI_DEBUG_H */
diff --git a/src/mesa/drivers/dri/common/vblank.c b/src/mesa/drivers/dri/common/vblank.c
index 094950d3626..d610253fe6f 100644
--- a/src/mesa/drivers/dri/common/vblank.c
+++ b/src/mesa/drivers/dri/common/vblank.c
@@ -26,14 +26,24 @@
  *    Ian Romanick <idr@us.ibm.com>
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "xf86drm.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/dd.h"
 #include "vblank.h"
 #include "xmlpool.h"
 
+static unsigned int msc_to_vblank(__DRIdrawablePrivate * dPriv, int64_t msc)
+{
+   return (unsigned int)(msc - dPriv->msc_base + dPriv->vblank_base);
+}
+
+static int64_t vblank_to_msc(__DRIdrawablePrivate * dPriv, unsigned int vblank)
+{
+   return (int64_t)(vblank - dPriv->vblank_base + dPriv->msc_base);
+}
+
 
 /****************************************************************************/
 /**
@@ -41,7 +51,7 @@
  *
  * Stores the 64-bit count of vertical refreshes since some (arbitrary)
  * point in time in \c count.  Unless the value wraps around, which it
- * may, it will never decrease.
+ * may, it will never decrease for a given drawable.
  *
  * \warning This function is called from \c glXGetVideoSyncSGI, which expects
  * a \c count of type \c unsigned (32-bit), and \c glXGetSyncValuesOML, which 
@@ -49,11 +59,14 @@
  * currently always returns a \c sequence of type \c unsigned.
  *
  * \param priv   Pointer to the DRI screen private struct.
+ * \param dPriv  Pointer to the DRI drawable private struct
  * \param count  Storage to hold MSC counter.
  * \return       Zero is returned on success.  A negative errno value
  *               is returned on failure.
  */
-int driGetMSC32( __DRIscreenPrivate * priv, int64_t * count )
+int driDrawableGetMSC32( __DRIscreenPrivate * priv,
+			 __DRIdrawablePrivate * dPriv,
+			 int64_t * count)
 {
    drmVBlank vbl;
    int ret;
@@ -62,14 +75,21 @@ int driGetMSC32( __DRIscreenPrivate * priv, int64_t * count )
 
    vbl.request.type = DRM_VBLANK_RELATIVE;
    vbl.request.sequence = 0;
+   if ( dPriv && dPriv->vblFlags & VBLANK_FLAG_SECONDARY )
+      vbl.request.type |= DRM_VBLANK_SECONDARY;
 
    ret = drmWaitVBlank( priv->fd, &vbl );
-   *count = (int64_t)vbl.reply.sequence;
+
+   if (dPriv) {
+      *count = vblank_to_msc(dPriv, vbl.reply.sequence);
+   } else {
+      /* Old driver (no knowledge of drawable MSC callback) */
+      *count = vbl.reply.sequence;
+   }
 
    return ret;
 }
 
-
 /****************************************************************************/
 /**
  * Wait for a specified refresh count.  This implements most of the
@@ -122,7 +142,9 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
           */
          vbl.request.type = dont_wait ? DRM_VBLANK_RELATIVE :
                                         DRM_VBLANK_ABSOLUTE;
-         vbl.request.sequence = next;
+         vbl.request.sequence = next ? msc_to_vblank(priv, next) : 0;
+	 if ( priv->vblFlags & VBLANK_FLAG_SECONDARY )
+	    vbl.request.type |= DRM_VBLANK_SECONDARY;
 
 	 if ( drmWaitVBlank( priv->driScreenPriv->fd, &vbl ) != 0 ) {
 	    /* FIXME: This doesn't seem like the right thing to return here.
@@ -130,8 +152,10 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
 	    return GLX_BAD_CONTEXT;
 	 }
 
+	 *msc = vblank_to_msc(priv, vbl.reply.sequence);
+
          dont_wait = 0;
-         if (target_msc != 0 && vbl.reply.sequence == target)
+         if (target_msc != 0 && *msc == target)
             break;
 
          /* Assuming the wait-done test fails, the next refresh to wait for
@@ -141,9 +165,9 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
           * If this refresh has already happened, we add divisor to obtain 
           * the next refresh after the current one that will satisfy it.
           */
-         r = (vbl.reply.sequence % (unsigned int)divisor);
-         next = (vbl.reply.sequence - r + (unsigned int)remainder);
-         if (next <= vbl.reply.sequence) next += (unsigned int)divisor;
+         r = (*msc % (unsigned int)divisor);
+         next = (*msc - r + (unsigned int)remainder);
+         if (next <= *msc) next += (unsigned int)divisor;
 
       } while ( r != (unsigned int)remainder );
    }
@@ -153,7 +177,10 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
        */
 
       vbl.request.type = DRM_VBLANK_ABSOLUTE;
-      vbl.request.sequence = target_msc;
+      vbl.request.sequence = target_msc ? msc_to_vblank(priv, target_msc) : 0;
+
+      if ( priv->vblFlags & VBLANK_FLAG_SECONDARY )
+	 vbl.request.type |= DRM_VBLANK_SECONDARY;
 
       if ( drmWaitVBlank( priv->driScreenPriv->fd, &vbl ) != 0 ) {
 	 /* FIXME: This doesn't seem like the right thing to return here.
@@ -162,8 +189,8 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
       }
    }
 
-   *msc  = (target_msc & 0xffffffff00000000LL);
-   *msc |= vbl.reply.sequence;
+   *msc = vblank_to_msc(priv, vbl.reply.sequence);
+
    if ( *msc < target_msc ) {
       *msc += 0x0000000100000000LL;
    }
@@ -232,8 +259,8 @@ static int do_wait( drmVBlank * vbl, GLuint * vbl_seq, int fd )
       if ( first_time ) {
 	 fprintf(stderr, 
 		 "%s: drmWaitVBlank returned %d, IRQs don't seem to be"
-		 " working correctly.\nTry running with LIBGL_THROTTLE_REFRESH"
-		 " and LIBL_SYNC_REFRESH unset.\n", __FUNCTION__, ret);
+		 " working correctly.\nTry adjusting the vblank_mode"
+		 " configuration parameter.\n", __FUNCTION__, ret);
 	 first_time = GL_FALSE;
       }
 
@@ -247,20 +274,42 @@ static int do_wait( drmVBlank * vbl, GLuint * vbl_seq, int fd )
 
 /****************************************************************************/
 /**
+ * Returns the default swap interval of the given drawable.
+ */
+
+static unsigned
+driGetDefaultVBlankInterval( const  __DRIdrawablePrivate *priv )
+{
+   if ( (priv->vblFlags & (VBLANK_FLAG_THROTTLE | VBLANK_FLAG_SYNC)) != 0 ) {
+      return 1;
+   }
+   else {
+      return 0;
+   }
+}
+
+
+/****************************************************************************/
+/**
  * Sets the default swap interval when the drawable is first bound to a
  * direct rendering context.
  */
 
-void driDrawableInitVBlank( __DRIdrawablePrivate *priv, GLuint flags,
-			    GLuint *vbl_seq )
+void driDrawableInitVBlank( __DRIdrawablePrivate *priv )
 {
-   if ( priv->pdraw->swap_interval == (unsigned)-1 ) {
+   if ( priv->swap_interval == (unsigned)-1 &&
+	!( priv->vblFlags & VBLANK_FLAG_NO_IRQ ) ) {
       /* Get current vertical blank sequence */
-      drmVBlank vbl = { .request={ .type = DRM_VBLANK_RELATIVE, .sequence = 0 } };
-      do_wait( &vbl, vbl_seq, priv->driScreenPriv->fd );
-
-      priv->pdraw->swap_interval = (flags & (VBLANK_FLAG_THROTTLE |
-					     VBLANK_FLAG_SYNC)) != 0 ? 1 : 0;
+      drmVBlank vbl;
+ 
+      vbl.request.type = DRM_VBLANK_RELATIVE;
+      if ( priv->vblFlags & VBLANK_FLAG_SECONDARY )
+ 	 vbl.request.type |= DRM_VBLANK_SECONDARY;
+      vbl.request.sequence = 0;
+      do_wait( &vbl, &priv->vblSeq, priv->driScreenPriv->fd );
+      priv->vblank_base = priv->vblSeq;
+
+      priv->swap_interval = driGetDefaultVBlankInterval( priv );
    }
 }
 
@@ -271,21 +320,17 @@ void driDrawableInitVBlank( __DRIdrawablePrivate *priv, GLuint flags,
  */
 
 unsigned
-driGetVBlankInterval( const  __DRIdrawablePrivate *priv, GLuint flags )
+driGetVBlankInterval( const  __DRIdrawablePrivate *priv )
 {
-   if ( (flags & VBLANK_FLAG_INTERVAL) != 0 ) {
+   if ( (priv->vblFlags & VBLANK_FLAG_INTERVAL) != 0 ) {
       /* this must have been initialized when the drawable was first bound
        * to a direct rendering context. */
-      assert ( priv->pdraw->swap_interval != (unsigned)-1 );
+      assert ( priv->swap_interval != (unsigned)-1 );
 
-      return priv->pdraw->swap_interval;
-   }
-   else if ( (flags & (VBLANK_FLAG_THROTTLE | VBLANK_FLAG_SYNC)) != 0 ) {
-      return 1;
-   }
-   else {
-      return 0;
+      return priv->swap_interval;
    }
+   else 
+      return driGetDefaultVBlankInterval( priv );
 }
 
 
@@ -295,18 +340,17 @@ driGetVBlankInterval( const  __DRIdrawablePrivate *priv, GLuint flags )
  */
 
 void
-driGetCurrentVBlank( const  __DRIdrawablePrivate *priv, GLuint flags,
-		     GLuint *vbl_seq )
+driGetCurrentVBlank( __DRIdrawablePrivate *priv )
 {
    drmVBlank vbl;
 
    vbl.request.type = DRM_VBLANK_RELATIVE;
-   if ( flags & VBLANK_FLAG_SECONDARY ) {
+   if ( priv->vblFlags & VBLANK_FLAG_SECONDARY ) {
       vbl.request.type |= DRM_VBLANK_SECONDARY;
    }
    vbl.request.sequence = 0;
 
-   (void) do_wait( &vbl, vbl_seq, priv->driScreenPriv->fd );
+   (void) do_wait( &vbl, &priv->vblSeq, priv->driScreenPriv->fd );
 }
 
 
@@ -314,19 +358,15 @@ driGetCurrentVBlank( const  __DRIdrawablePrivate *priv, GLuint flags,
 /**
  * Waits for the vertical blank for use with glXSwapBuffers.
  * 
- * \param vbl_seq  Vertical blank sequence number (MSC) after the last buffer
- *                 swap.  Updated after this wait.
- * \param flags    \c VBLANK_FLAG bits that control how long to wait.
  * \param missed_deadline  Set to \c GL_TRUE if the MSC after waiting is later
- *                 than the "target" based on \c flags.  The idea is that if
- *                 \c missed_deadline is set, then the application is not 
- *                 achieving its desired framerate.
+ *                 than the "target" based on \c priv->vblFlags.  The idea is
+ *                 that if \c missed_deadline is set, then the application is
+ *                 not achieving its desired framerate.
  * \return         Zero on success, -1 on error.
  */
 
 int
-driWaitForVBlank( const  __DRIdrawablePrivate *priv, GLuint * vbl_seq,
-		  GLuint flags, GLboolean * missed_deadline )
+driWaitForVBlank( __DRIdrawablePrivate *priv, GLboolean * missed_deadline )
 {
    drmVBlank vbl;
    unsigned   original_seq;
@@ -335,10 +375,10 @@ driWaitForVBlank( const  __DRIdrawablePrivate *priv, GLuint * vbl_seq,
    unsigned   diff;
 
    *missed_deadline = GL_FALSE;
-   if ( (flags & (VBLANK_FLAG_INTERVAL |
-		  VBLANK_FLAG_THROTTLE |
-		  VBLANK_FLAG_SYNC)) == 0 ||
-	(flags & VBLANK_FLAG_NO_IRQ) != 0 ) {
+   if ( (priv->vblFlags & (VBLANK_FLAG_INTERVAL |
+			   VBLANK_FLAG_THROTTLE |
+			   VBLANK_FLAG_SYNC)) == 0 ||
+	(priv->vblFlags & VBLANK_FLAG_NO_IRQ) != 0 ) {
       return 0;
    }
 
@@ -349,44 +389,45 @@ driWaitForVBlank( const  __DRIdrawablePrivate *priv, GLuint * vbl_seq,
     *
     * VBLANK_FLAG_INTERVAL and VBLANK_FLAG_THROTTLE mean to wait for at
     * least one vertical blank since the last wait.  Since do_wait modifies
-    * vbl_seq, we have to save the original value of vbl_seq for the
+    * priv->vblSeq, we have to save the original value of priv->vblSeq for the
     * VBLANK_FLAG_INTERVAL / VBLANK_FLAG_THROTTLE calculation later.
     */
 
-   original_seq = *vbl_seq;
-   interval = driGetVBlankInterval(priv, flags);
+   original_seq = priv->vblSeq;
+   interval = driGetVBlankInterval(priv);
    deadline = original_seq + interval;
 
    vbl.request.type = DRM_VBLANK_RELATIVE;
-   if ( flags & VBLANK_FLAG_SECONDARY ) {
+   if ( priv->vblFlags & VBLANK_FLAG_SECONDARY ) {
       vbl.request.type |= DRM_VBLANK_SECONDARY;
    }
-   vbl.request.sequence = ((flags & VBLANK_FLAG_SYNC) != 0) ? 1 : 0;
+   vbl.request.sequence = ((priv->vblFlags & VBLANK_FLAG_SYNC) != 0) ? 1 : 0;
 
-   if ( do_wait( & vbl, vbl_seq, priv->driScreenPriv->fd ) != 0 ) {
+   if ( do_wait( & vbl, &priv->vblSeq, priv->driScreenPriv->fd ) != 0 ) {
       return -1;
    }
 
-   diff = *vbl_seq - deadline;
+   diff = priv->vblSeq - deadline;
 
    /* No need to wait again if we've already reached the target */
    if (diff <= (1 << 23)) {
-      *missed_deadline = (flags & VBLANK_FLAG_SYNC) ? (diff > 0) : GL_TRUE;
+      *missed_deadline = (priv->vblFlags & VBLANK_FLAG_SYNC) ? (diff > 0) :
+			 GL_TRUE;
       return 0;
    }
 
    /* Wait until the target vertical blank. */
    vbl.request.type = DRM_VBLANK_ABSOLUTE;
-   if ( flags & VBLANK_FLAG_SECONDARY ) {
+   if ( priv->vblFlags & VBLANK_FLAG_SECONDARY ) {
       vbl.request.type |= DRM_VBLANK_SECONDARY;
    }
    vbl.request.sequence = deadline;
 
-   if ( do_wait( & vbl, vbl_seq, priv->driScreenPriv->fd ) != 0 ) {
+   if ( do_wait( & vbl, &priv->vblSeq, priv->driScreenPriv->fd ) != 0 ) {
       return -1;
    }
 
-   diff = *vbl_seq - deadline;
+   diff = priv->vblSeq - deadline;
    *missed_deadline = diff > 0 && diff <= (1 << 23);
 
    return 0;
diff --git a/src/mesa/drivers/dri/common/vblank.h b/src/mesa/drivers/dri/common/vblank.h
index 52c1933ca5b..8b2c761a116 100644
--- a/src/mesa/drivers/dri/common/vblank.h
+++ b/src/mesa/drivers/dri/common/vblank.h
@@ -29,7 +29,7 @@
 #ifndef DRI_VBLANK_H
 #define DRI_VBLANK_H
 
-#include "context.h"
+#include "main/context.h"
 #include "dri_util.h"
 #include "xmlconfig.h"
 
@@ -45,17 +45,17 @@
 					  */
 
 extern int driGetMSC32( __DRIscreenPrivate * priv, int64_t * count );
+extern int driDrawableGetMSC32( __DRIscreenPrivate * priv,
+				__DRIdrawablePrivate * drawablePrivate,
+				int64_t * count);
 extern int driWaitForMSC32( __DRIdrawablePrivate *priv,
     int64_t target_msc, int64_t divisor, int64_t remainder, int64_t * msc );
 extern GLuint driGetDefaultVBlankFlags( const driOptionCache *optionCache );
-extern void driDrawableInitVBlank ( __DRIdrawablePrivate *priv, GLuint flags,
-				    GLuint *vbl_seq );
-extern unsigned driGetVBlankInterval( const  __DRIdrawablePrivate *priv,
-				      GLuint flags );
-extern void driGetCurrentVBlank( const  __DRIdrawablePrivate *priv,
-				 GLuint flags, GLuint *vbl_seq );
-extern int driWaitForVBlank( const __DRIdrawablePrivate *priv,
-    GLuint * vbl_seq, GLuint flags, GLboolean * missed_deadline );
+extern void driDrawableInitVBlank ( __DRIdrawablePrivate *priv );
+extern unsigned driGetVBlankInterval( const  __DRIdrawablePrivate *priv );
+extern void driGetCurrentVBlank( __DRIdrawablePrivate *priv );
+extern int driWaitForVBlank( __DRIdrawablePrivate *priv,
+			     GLboolean * missed_deadline );
 
 #undef usleep
 #include <unistd.h>  /* for usleep() */
diff --git a/src/mesa/drivers/dri/common/xmlconfig.c b/src/mesa/drivers/dri/common/xmlconfig.c
index b635894fe53..46ba2ffbfed 100644
--- a/src/mesa/drivers/dri/common/xmlconfig.c
+++ b/src/mesa/drivers/dri/common/xmlconfig.c
@@ -27,7 +27,7 @@
  * \author Felix Kuehling
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 #include <string.h>
 #include <assert.h>
@@ -35,7 +35,7 @@
 #include <fcntl.h>
 #include <unistd.h>
 #include <errno.h>
-#include "imports.h"
+#include "main/imports.h"
 #include "dri_util.h"
 #include "xmlconfig.h"
 
@@ -63,6 +63,12 @@ extern char *program_invocation_name, *program_invocation_short_name;
 #elif defined(__NetBSD__) && defined(__NetBSD_Version) && (__NetBSD_Version >= 106000100)
 #    include <stdlib.h>
 #    define GET_PROGRAM_NAME() getprogname()
+#elif defined(__sun)
+/* Solaris has getexecname() which returns the full path - return just
+   the basename to match BSD getprogname() */
+#    include <stdlib.h>
+#    include <libgen.h>
+#    define GET_PROGRAM_NAME() basename(getexecname())
 #endif
 
 #if !defined(GET_PROGRAM_NAME)
@@ -279,7 +285,7 @@ static GLfloat strToF (const XML_Char *string, const XML_Char **tail) {
 /** \brief Parse a value of a given type. */
 static GLboolean parseValue (driOptionValue *v, driOptionType type,
 			     const XML_Char *string) {
-    const XML_Char *tail;
+    const XML_Char *tail = NULL;
   /* skip leading white-space */
     string += strspn (string, " \f\n\r\t\v");
     switch (type) {
@@ -403,40 +409,40 @@ static GLboolean checkValue (const driOptionValue *v, const driOptionInfo *info)
 /** \brief Output a warning message. */
 #define XML_WARNING1(msg) do {\
     __driUtilMessage ("Warning in %s line %d, column %d: "msg, data->name, \
-                      XML_GetCurrentLineNumber(data->parser), \
-                      XML_GetCurrentColumnNumber(data->parser)); \
+                      (int) XML_GetCurrentLineNumber(data->parser), \
+                      (int) XML_GetCurrentColumnNumber(data->parser)); \
 } while (0)
 #define XML_WARNING(msg,args...) do { \
     __driUtilMessage ("Warning in %s line %d, column %d: "msg, data->name, \
-                      XML_GetCurrentLineNumber(data->parser), \
-                      XML_GetCurrentColumnNumber(data->parser), \
+                      (int) XML_GetCurrentLineNumber(data->parser), \
+                      (int) XML_GetCurrentColumnNumber(data->parser), \
                       args); \
 } while (0)
 /** \brief Output an error message. */
 #define XML_ERROR1(msg) do { \
     __driUtilMessage ("Error in %s line %d, column %d: "msg, data->name, \
-                      XML_GetCurrentLineNumber(data->parser), \
-                      XML_GetCurrentColumnNumber(data->parser)); \
+                      (int) XML_GetCurrentLineNumber(data->parser), \
+                      (int) XML_GetCurrentColumnNumber(data->parser)); \
 } while (0)
 #define XML_ERROR(msg,args...) do { \
     __driUtilMessage ("Error in %s line %d, column %d: "msg, data->name, \
-                      XML_GetCurrentLineNumber(data->parser), \
-                      XML_GetCurrentColumnNumber(data->parser), \
+                      (int) XML_GetCurrentLineNumber(data->parser), \
+                      (int) XML_GetCurrentColumnNumber(data->parser), \
                       args); \
 } while (0)
 /** \brief Output a fatal error message and abort. */
 #define XML_FATAL1(msg) do { \
     fprintf (stderr, "Fatal error in %s line %d, column %d: "msg"\n", \
              data->name, \
-             XML_GetCurrentLineNumber(data->parser), \
-             XML_GetCurrentColumnNumber(data->parser)); \
+             (int) XML_GetCurrentLineNumber(data->parser),	\
+             (int) XML_GetCurrentColumnNumber(data->parser)); \
     abort();\
 } while (0)
 #define XML_FATAL(msg,args...) do { \
     fprintf (stderr, "Fatal error in %s line %d, column %d: "msg"\n", \
              data->name, \
-             XML_GetCurrentLineNumber(data->parser), \
-             XML_GetCurrentColumnNumber(data->parser), \
+             (int) XML_GetCurrentLineNumber(data->parser),	\
+             (int) XML_GetCurrentColumnNumber(data->parser),		\
              args); \
     abort();\
 } while (0)
diff --git a/src/mesa/drivers/dri/common/xmlpool/Makefile b/src/mesa/drivers/dri/common/xmlpool/Makefile
index b077809cd18..62ec919ea68 100644
--- a/src/mesa/drivers/dri/common/xmlpool/Makefile
+++ b/src/mesa/drivers/dri/common/xmlpool/Makefile
@@ -57,8 +57,8 @@ all: options.h
 # Only intermediate files are cleaned up. options.h is not deleted because
 # it's in CVS. 
 clean:
-	rm -f $(POT) *~
-	rm -rf $(LANGS)
+	-rm -f $(POT) *~
+	-rm -rf $(LANGS)
 
 # Default target options.h
 options.h: t_options.h mo
diff --git a/src/mesa/drivers/dri/dri.pc.in b/src/mesa/drivers/dri/dri.pc.in
new file mode 100644
index 00000000000..695aa6cfd66
--- /dev/null
+++ b/src/mesa/drivers/dri/dri.pc.in
@@ -0,0 +1,11 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+dridriverdir=@DRI_DRIVER_DIR@
+
+Name: dri
+Description: Direct Rendering Infrastructure
+Version: @VERSION@
+Requires.private: @DRI_PC_REQ_PRIV@
+Cflags: -I${includedir}
diff --git a/src/mesa/drivers/dri/fb/fb_dri.c b/src/mesa/drivers/dri/fb/fb_dri.c
index 523a0c0380f..f1194d7ce82 100644
--- a/src/mesa/drivers/dri/fb/fb_dri.c
+++ b/src/mesa/drivers/dri/fb/fb_dri.c
@@ -47,14 +47,14 @@
 #include "drirenderbuffer.h"
 
 #include "buffers.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 #include "tnl/t_pipeline.h"
 #include "drivers/common/driverfuncs.h"
 
@@ -657,8 +657,9 @@ struct DRIDriverRec __driDriver = {
 };
 
 static __GLcontextModes *
-fbFillInModes( unsigned pixel_bits, unsigned depth_bits,
-                 unsigned stencil_bits, GLboolean have_back_buffer )
+fbFillInModes( __DRIscreenPrivate *psp,
+	       unsigned pixel_bits, unsigned depth_bits,
+	       unsigned stencil_bits, GLboolean have_back_buffer )
 {
    __GLcontextModes * modes;
    __GLcontextModes * m;
@@ -705,7 +706,7 @@ fbFillInModes( unsigned pixel_bits, unsigned depth_bits,
       fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
    }
 
-   modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
+   modes = (*psp->contextModes->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
    m = modes;
    if ( ! driFillInModes( & m, fb_format, fb_type,
           depth_bits_array, stencil_bits_array, depth_buffer_factor,
@@ -776,7 +777,7 @@ void * __driCreateNewScreen( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc
                                          frame_buffer, pSAREA, fd,
                                          internal_api_version, &fbAPI);
           if ( psp != NULL ) {
-	     *driver_modes = fbFillInModes( psp->fbBPP,
+	     *driver_modes = fbFillInModes( psp, psp->fbBPP,
 					    (psp->fbBPP == 16) ? 16 : 24,
 					    (psp->fbBPP == 16) ? 0  : 8,
 					    1);
diff --git a/src/mesa/drivers/dri/fb/fb_egl.c b/src/mesa/drivers/dri/fb/fb_egl.c
index cc6a266df30..35c268441c3 100644
--- a/src/mesa/drivers/dri/fb/fb_egl.c
+++ b/src/mesa/drivers/dri/fb/fb_egl.c
@@ -14,20 +14,20 @@
 
 #include "utils.h"
 #include "buffers.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 #include "tnl/t_pipeline.h"
 #include "drivers/common/driverfuncs.h"
 #include "drirenderbuffer.h"
 
 #include "eglconfig.h"
-#include "eglcontext.h"
+#include "eglmain/context.h"
 #include "egldisplay.h"
 #include "egldriver.h"
 #include "eglglobals.h"
diff --git a/src/mesa/drivers/dri/ffb/ffb_bitmap.c b/src/mesa/drivers/dri/ffb/ffb_bitmap.c
index 1aa66859a6f..f89c0412dfe 100644
--- a/src/mesa/drivers/dri/ffb/ffb_bitmap.c
+++ b/src/mesa/drivers/dri/ffb/ffb_bitmap.c
@@ -30,8 +30,8 @@
 #include "ffb_lock.h"
 #include "ffb_bitmap.h"
 #include "swrast/swrast.h"
-#include "image.h"
-#include "macros.h"
+#include "main/image.h"
+#include "main/macros.h"
 
 /* Compute ceiling of integer quotient of A divided by B: */
 #define CEILING( A, B )  ( (A) % (B) == 0 ? (A)/(B) : (A)/(B)+1 )
diff --git a/src/mesa/drivers/dri/ffb/ffb_clear.c b/src/mesa/drivers/dri/ffb/ffb_clear.c
index 7de05b5cf0c..776fb487f8d 100644
--- a/src/mesa/drivers/dri/ffb/ffb_clear.c
+++ b/src/mesa/drivers/dri/ffb/ffb_clear.c
@@ -25,10 +25,10 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
-#include "extensions.h"
+#include "main/mtypes.h"
+#include "main/extensions.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_depth.h"
@@ -79,7 +79,7 @@ struct ff_fixups {
 /* Compute fixups of non-page aligned areas after a page fill.
  * Return the number of fixups needed.
  */
-static __inline__ int
+static INLINE int
 CreatorComputePageFillFixups(struct ff_fixups *fixups,
 			     int x, int y, int w, int h,
 			     int paligned_x, int paligned_y,
diff --git a/src/mesa/drivers/dri/ffb/ffb_context.h b/src/mesa/drivers/dri/ffb/ffb_context.h
index 0ab75fce47c..77f87d41c30 100644
--- a/src/mesa/drivers/dri/ffb/ffb_context.h
+++ b/src/mesa/drivers/dri/ffb/ffb_context.h
@@ -5,7 +5,7 @@
 #include "dri_util.h"
 #include "drm.h"
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 #include "ffb_xmesa.h"
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_dd.c b/src/mesa/drivers/dri/ffb/ffb_dd.c
index f64a577d1fc..cf83b91f0d4 100644
--- a/src/mesa/drivers/dri/ffb/ffb_dd.c
+++ b/src/mesa/drivers/dri/ffb/ffb_dd.c
@@ -25,8 +25,9 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
+#include "main/extensions.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_depth.h"
@@ -35,7 +36,6 @@
 #include "ffb_tris.h"
 #include "ffb_clear.h"
 #include "ffb_lock.h"
-#include "extensions.h"
 
 #define FFB_DATE	"20021125"
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_dd.h b/src/mesa/drivers/dri/ffb/ffb_dd.h
index e065ebbecd9..1198cda30a1 100644
--- a/src/mesa/drivers/dri/ffb/ffb_dd.h
+++ b/src/mesa/drivers/dri/ffb/ffb_dd.h
@@ -28,7 +28,7 @@
 #ifndef _FFB_DD_H
 #define _FFB_DD_H
 
-#include "context.h"
+#include "main/context.h"
 
 void ffbDDInitDriverFuncs(GLcontext *ctx);
 void ffbDDExtensionsInit(GLcontext *ctx);
diff --git a/src/mesa/drivers/dri/ffb/ffb_depth.c b/src/mesa/drivers/dri/ffb/ffb_depth.c
index cca6212f501..71f204d21e2 100644
--- a/src/mesa/drivers/dri/ffb/ffb_depth.c
+++ b/src/mesa/drivers/dri/ffb/ffb_depth.c
@@ -25,7 +25,7 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
@@ -33,7 +33,6 @@
 #include "ffb_depth.h"
 #include "ffb_lock.h"
 
-#include "swrast/swrast.h"
 
 #undef DEPTH_TRACE
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_lines.c b/src/mesa/drivers/dri/ffb/ffb_lines.c
index 82947014644..19dff509354 100644
--- a/src/mesa/drivers/dri/ffb/ffb_lines.c
+++ b/src/mesa/drivers/dri/ffb/ffb_lines.c
@@ -25,8 +25,9 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
+#include "main/extensions.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_depth.h"
@@ -35,7 +36,6 @@
 #include "ffb_lines.h"
 #include "ffb_tris.h"
 #include "ffb_lock.h"
-#include "extensions.h"
 
 #undef FFB_LINE_TRACE
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_linetmp.h b/src/mesa/drivers/dri/ffb/ffb_linetmp.h
index e9d8260e1aa..10e1375259b 100644
--- a/src/mesa/drivers/dri/ffb/ffb_linetmp.h
+++ b/src/mesa/drivers/dri/ffb/ffb_linetmp.h
@@ -1,5 +1,5 @@
 
-static __inline void TAG(ffb_line)(GLcontext *ctx, ffb_vertex *v0, 
+static INLINE void TAG(ffb_line)(GLcontext *ctx, ffb_vertex *v0, 
 				   ffb_vertex *v1 )
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
diff --git a/src/mesa/drivers/dri/ffb/ffb_points.c b/src/mesa/drivers/dri/ffb/ffb_points.c
index d00255ccee5..9c37a47aeb9 100644
--- a/src/mesa/drivers/dri/ffb/ffb_points.c
+++ b/src/mesa/drivers/dri/ffb/ffb_points.c
@@ -25,7 +25,7 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "ffb_dd.h"
 #include "ffb_context.h"
 #include "ffb_vb.h"
diff --git a/src/mesa/drivers/dri/ffb/ffb_pointtmp.h b/src/mesa/drivers/dri/ffb/ffb_pointtmp.h
index 2c91426b3a2..3003de70c69 100644
--- a/src/mesa/drivers/dri/ffb/ffb_pointtmp.h
+++ b/src/mesa/drivers/dri/ffb/ffb_pointtmp.h
@@ -1,5 +1,5 @@
 
-static __inline void TAG(ffb_draw_point)(GLcontext *ctx, ffb_vertex *tmp )
+static INLINE void TAG(ffb_draw_point)(GLcontext *ctx, ffb_vertex *tmp )
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
 	ffb_fbcPtr ffb = fmesa->regs;
diff --git a/src/mesa/drivers/dri/ffb/ffb_span.c b/src/mesa/drivers/dri/ffb/ffb_span.c
index 59ac4146786..0d3d6040955 100644
--- a/src/mesa/drivers/dri/ffb/ffb_span.c
+++ b/src/mesa/drivers/dri/ffb/ffb_span.c
@@ -25,7 +25,7 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_context.h"
diff --git a/src/mesa/drivers/dri/ffb/ffb_state.c b/src/mesa/drivers/dri/ffb/ffb_state.c
index 880ad8be0a8..ee0fe4e0dbe 100644
--- a/src/mesa/drivers/dri/ffb/ffb_state.c
+++ b/src/mesa/drivers/dri/ffb/ffb_state.c
@@ -25,9 +25,18 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
-#include "colormac.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/mm.h"
+#include "main/extensions.h"
+#include "main/enums.h"
+
+#include "vbo/vbo.h"
+#include "tnl/tnl.h"
+#include "tnl/t_pipeline.h"
+#include "swrast/swrast.h"
+#include "swrast_setup/swrast_setup.h"
+
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_depth.h"
@@ -36,15 +45,6 @@
 #include "ffb_tris.h"
 #include "ffb_state.h"
 #include "ffb_lock.h"
-#include "extensions.h"
-#include "enums.h"
-
-#include "swrast/swrast.h"
-#include "vbo/vbo.h"
-#include "tnl/tnl.h"
-#include "swrast_setup/swrast_setup.h"
-
-#include "tnl/t_pipeline.h"
 
 #undef STATE_TRACE
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_stencil.c b/src/mesa/drivers/dri/ffb/ffb_stencil.c
index d535b1b7781..921a83d2742 100644
--- a/src/mesa/drivers/dri/ffb/ffb_stencil.c
+++ b/src/mesa/drivers/dri/ffb/ffb_stencil.c
@@ -25,7 +25,7 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "ffb_dd.h"
 #include "ffb_span.h"
 #include "ffb_context.h"
diff --git a/src/mesa/drivers/dri/ffb/ffb_tex.c b/src/mesa/drivers/dri/ffb/ffb_tex.c
index 6503b0f4e78..69d30aedbaf 100644
--- a/src/mesa/drivers/dri/ffb/ffb_tex.c
+++ b/src/mesa/drivers/dri/ffb/ffb_tex.c
@@ -25,8 +25,8 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
 #include "ffb_tex.h"
 
 /* No texture unit, all software. */
diff --git a/src/mesa/drivers/dri/ffb/ffb_tris.c b/src/mesa/drivers/dri/ffb/ffb_tris.c
index c2857f61bde..d785c157181 100644
--- a/src/mesa/drivers/dri/ffb/ffb_tris.c
+++ b/src/mesa/drivers/dri/ffb/ffb_tris.c
@@ -25,12 +25,12 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
 #include "swrast/s_context.h"
+#include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_vb.c b/src/mesa/drivers/dri/ffb/ffb_vb.c
index edc9d791243..f9c6fd1f310 100644
--- a/src/mesa/drivers/dri/ffb/ffb_vb.c
+++ b/src/mesa/drivers/dri/ffb/ffb_vb.c
@@ -28,7 +28,7 @@
 #include "ffb_xmesa.h"
 #include "ffb_context.h"
 #include "ffb_vb.h"
-#include "imports.h"
+#include "main/imports.h"
 #include "tnl/t_context.h"
 #include "swrast_setup/swrast_setup.h"
 #include "math/m_translate.h"
diff --git a/src/mesa/drivers/dri/ffb/ffb_vb.h b/src/mesa/drivers/dri/ffb/ffb_vb.h
index af669bce307..238b9940bfa 100644
--- a/src/mesa/drivers/dri/ffb/ffb_vb.h
+++ b/src/mesa/drivers/dri/ffb/ffb_vb.h
@@ -2,8 +2,8 @@
 #ifndef _FFB_VB_H
 #define _FFB_VB_H
 
-#include "mtypes.h"
-#include "macros.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "tnl/t_context.h"
 #include "swrast/swrast.h"
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_vtxfmt.c b/src/mesa/drivers/dri/ffb/ffb_vtxfmt.c
index 8b60f095c96..90f44b0e911 100644
--- a/src/mesa/drivers/dri/ffb/ffb_vtxfmt.c
+++ b/src/mesa/drivers/dri/ffb/ffb_vtxfmt.c
@@ -25,27 +25,23 @@
  *    David S. Miller <davem@redhat.com>
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "api_noop.h"
-#include "context.h"
+#include "main/context.h"
 #include "light.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "simple_list.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
 #include "vtxfmt.h"
 #include "ffb_xmesa.h"
 #include "ffb_context.h"
 #include "ffb_vb.h"
 #include "tnl/tnl.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 
 #include "ffb_vtxfmt.h"
 
-#ifndef __GNUC__
-#define __inline  /**/
-#endif
-
 #define TNL_VERTEX			ffbTnlVertex
 
 #define INTERP_RGBA(t, out, a, b)		\
@@ -60,7 +56,7 @@ do {						\
 
 /* Color functions: */
 
-static __inline void ffb_recalc_base_color(GLcontext *ctx)
+static INLINE void ffb_recalc_base_color(GLcontext *ctx)
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
 	struct gl_light *light;
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.c b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
index a31cd57e882..679f8561d24 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.c
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
@@ -26,12 +26,12 @@
  */
 
 #include "ffb_xmesa.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "matrix.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "main/matrix.h"
+#include "main/renderbuffer.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
 #include "utils.h"
 
 #include "swrast/swrast.h"
@@ -226,7 +226,7 @@ ffbCreateContext(const __GLcontextModes *mesaVis,
 	fmesa->driScreen = sPriv;
 	fmesa->ffb_sarea = FFB_DRISHARE(sPriv->pSAREA);
 
-	/* Register and framebuffer hw pointers. */
+	/* Register and framebuffer pointers. */
 	fmesa->regs = ffbScreen->regs;
 	fmesa->sfb32 = ffbScreen->sfb32;
 
@@ -604,35 +604,18 @@ void ffbXMesaUpdateState(ffbContextPtr fmesa)
 	}
 }
 
-static const struct __DriverAPIRec ffbAPI = {
-   .InitDriver      = ffbInitDriver,
-   .DestroyScreen   = ffbDestroyScreen,
-   .CreateContext   = ffbCreateContext,
-   .DestroyContext  = ffbDestroyContext,
-   .CreateBuffer    = ffbCreateBuffer,
-   .DestroyBuffer   = ffbDestroyBuffer,
-   .SwapBuffers     = ffbSwapBuffers,
-   .MakeCurrent     = ffbMakeCurrent,
-   .UnbindContext   = ffbUnbindContext,
-   .GetSwapInfo     = NULL,
-   .GetMSC          = NULL,
-   .WaitForMSC      = NULL,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL
-};
-
-
-static __GLcontextModes *
-ffbFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
+static const __DRIconfig **
+ffbFillInModes( __DRIscreenPrivate *psp,
+		unsigned pixel_bits, unsigned depth_bits,
+		unsigned stencil_bits, GLboolean have_back_buffer )
 {
-   __GLcontextModes * modes;
-   __GLcontextModes * m;
-   unsigned num_modes;
+   __DRIconfig **configs;
+   __GLcontextModes *m;
    unsigned depth_buffer_factor;
    unsigned back_buffer_factor;
    GLenum fb_format;
    GLenum fb_type;
+   int i;
 
    /* GLX_SWAP_COPY_OML is only supported because the FFB driver doesn't
     * support pageflipping at all.
@@ -644,7 +627,6 @@ ffbFillInModes( unsigned pixel_bits, unsigned depth_bits,
    uint8_t depth_bits_array[3];
    uint8_t stencil_bits_array[3];
 
-
    depth_bits_array[0] = 0;
    depth_bits_array[1] = depth_bits;
    depth_bits_array[2] = depth_bits;
@@ -660,8 +642,6 @@ ffbFillInModes( unsigned pixel_bits, unsigned depth_bits,
    depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
    back_buffer_factor  = (have_back_buffer) ? 3 : 1;
 
-   num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -671,82 +651,68 @@ ffbFillInModes( unsigned pixel_bits, unsigned depth_bits,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-   modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-   m = modes;
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
+   configs = driCreateConfigs(fb_format, fb_type,
+			      depth_bits_array, stencil_bits_array,
+			      depth_buffer_factor, back_buffer_modes,
+			      back_buffer_factor);
+   if (configs == NULL) {
+    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
    }
 
-
    /* Mark the visual as slow if there are "fake" stencil bits.
     */
-   for ( m = modes ; m != NULL ; m = m->next ) {
-      if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	 m->visualRating = GLX_SLOW_CONFIG;
+   for (i = 0; configs[i]; i++) {
+      m = &configs[i]->modes;
+      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+         m->visualRating = GLX_SLOW_CONFIG;
       }
    }
 
-   return modes;
+   return (const __DRIconfig **) configs;
 }
 
 
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+static const __DRIconfig **
+ffbInitScreen(__DRIscreen *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 0, 1, 1 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 0, 0, 1 };
 
-   dri_interface = interface;
-
    if ( ! driCheckDriDdxDrmVersions2( "ffb",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
 
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &ffbAPI);
-   if ( psp != NULL ) {
-      *driver_modes = ffbFillInModes( 32, 16, 0, GL_TRUE );
-   }
+   if (!ffbInitDriver(psp))
+       return NULL;
 
-   return (void *) psp;
+   return ffbFillInModes( psp, 32, 16, 0, GL_TRUE );
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = ffbInitScreen,
+   .DestroyScreen   = ffbDestroyScreen,
+   .CreateContext   = ffbCreateContext,
+   .DestroyContext  = ffbDestroyContext,
+   .CreateBuffer    = ffbCreateBuffer,
+   .DestroyBuffer   = ffbDestroyBuffer,
+   .SwapBuffers     = ffbSwapBuffers,
+   .MakeCurrent     = ffbMakeCurrent,
+   .UnbindContext   = ffbUnbindContext,
+   .GetSwapInfo     = NULL,
+   .GetDrawableMSC  = NULL,
+   .WaitForMSC      = NULL,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL
+};
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.h b/src/mesa/drivers/dri/ffb/ffb_xmesa.h
index bc8cfe9f217..255da4c5f84 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.h
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.h
@@ -4,7 +4,7 @@
 
 #include <sys/time.h>
 #include "dri_util.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "ffb_drishare.h"
 #include "ffb_regs.h"
 #include "ffb_dac.h"
diff --git a/src/mesa/drivers/dri/gamma/gamma_context.c b/src/mesa/drivers/dri/gamma/gamma_context.c
index b1dcbfcdcfa..c91bedce3a8 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.c
+++ b/src/mesa/drivers/dri/gamma/gamma_context.c
@@ -24,7 +24,7 @@
  * 3DLabs Gamma driver.
  *
  */
-#include "gamma_context.h"
+#include "gammacontext.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -35,16 +35,16 @@
 
 #include "drivers/common/driverfuncs.h"
 
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
 #if defined(USE_X86_ASM)
 #include "x86/common_x86_asm.h"
 #endif
-#include "simple_list.h"
-#include "mm.h"
+#include "main/simple_list.h"
+#include "main/mm.h"
 
 
 #include "gamma_vb.h"
diff --git a/src/mesa/drivers/dri/gamma/gamma_context.h b/src/mesa/drivers/dri/gamma/gamma_context.h
index 86af674ae75..a32ccb6007b 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.h
+++ b/src/mesa/drivers/dri/gamma/gamma_context.h
@@ -31,12 +31,12 @@
 #include "drm_sarea.h"
 #include "colormac.h"
 #include "gamma_regs.h"
-#include "gamma_macros.h"
+#include "gamma_main/macros.h"
 #include "gamma_screen.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 #include "glint_dri.h"
-#include "mm.h"
+#include "main/mm.h"
 
 typedef union {
     unsigned int i;
@@ -383,7 +383,7 @@ struct gamma_context {
     int                 TextureCount;
 };
 
-static __inline GLuint gammaPackColor( GLuint cpp,
+static INLINE GLuint gammaPackColor( GLuint cpp,
 					GLubyte r, GLubyte g,
 					GLubyte b, GLubyte a )
 {
diff --git a/src/mesa/drivers/dri/gamma/gamma_dd.c b/src/mesa/drivers/dri/gamma/gamma_dd.c
index 63e3ab8fa5d..7a81ef59937 100644
--- a/src/mesa/drivers/dri/gamma/gamma_dd.c
+++ b/src/mesa/drivers/dri/gamma/gamma_dd.c
@@ -23,14 +23,14 @@
  *
  */
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_vb.h"
 #include "gamma_lock.h"
 #if defined(USE_X86_ASM)
 #include "x86/common_x86_asm.h"
 #endif
 
-#include "context.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
 
 #define GAMMA_DATE	"20021125"
diff --git a/src/mesa/drivers/dri/gamma/gamma_inithw.c b/src/mesa/drivers/dri/gamma/gamma_inithw.c
index 79b54aacb5f..525ad89354b 100644
--- a/src/mesa/drivers/dri/gamma/gamma_inithw.c
+++ b/src/mesa/drivers/dri/gamma/gamma_inithw.c
@@ -24,7 +24,7 @@
  *
  */
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "glint_dri.h"
 
 void gammaInitHW( gammaContextPtr gmesa )
diff --git a/src/mesa/drivers/dri/gamma/gamma_lock.c b/src/mesa/drivers/dri/gamma/gamma_lock.c
index 97eea75541e..8f2d01688c9 100644
--- a/src/mesa/drivers/dri/gamma/gamma_lock.c
+++ b/src/mesa/drivers/dri/gamma/gamma_lock.c
@@ -1,5 +1,5 @@
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_lock.h"
 #include "drirenderbuffer.h"
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_render.c b/src/mesa/drivers/dri/gamma/gamma_render.c
index 9180ef925d0..1b9fd169f46 100644
--- a/src/mesa/drivers/dri/gamma/gamma_render.c
+++ b/src/mesa/drivers/dri/gamma/gamma_render.c
@@ -25,15 +25,15 @@
  *
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_tris.h"
 #include "gamma_vb.h"
 
@@ -129,13 +129,13 @@ static const GLuint hw_prim[GL_POLYGON+1] = {
    B_PrimType_Polygon
 };
 
-static __inline void gammaStartPrimitive( gammaContextPtr gmesa, GLenum prim )
+static INLINE void gammaStartPrimitive( gammaContextPtr gmesa, GLenum prim )
 {
    CHECK_DMA_BUFFER(gmesa, 1);
    WRITE(gmesa->buf, Begin, gmesa->Begin | hw_prim[prim]);
 }
 
-static __inline void gammaEndPrimitive( gammaContextPtr gmesa )
+static INLINE void gammaEndPrimitive( gammaContextPtr gmesa )
 {
    GLcontext *ctx = gmesa->glCtx;
 
@@ -193,7 +193,7 @@ static GLboolean gamma_run_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_screen.c b/src/mesa/drivers/dri/gamma/gamma_screen.c
index 0b91d059e92..f899ebec96d 100644
--- a/src/mesa/drivers/dri/gamma/gamma_screen.c
+++ b/src/mesa/drivers/dri/gamma/gamma_screen.c
@@ -23,11 +23,11 @@
  *
  */
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_vb.h"
 #include "glint_dri.h"
 
-#include "imports.h"
+#include "main/imports.h"
 
 gammaScreenPtr gammaCreateScreen( __DRIscreenPrivate *sPriv )
 {
diff --git a/src/mesa/drivers/dri/gamma/gamma_span.c b/src/mesa/drivers/dri/gamma/gamma_span.c
index ea9a20595c9..cdaaac3f3ac 100644
--- a/src/mesa/drivers/dri/gamma/gamma_span.c
+++ b/src/mesa/drivers/dri/gamma/gamma_span.c
@@ -1,5 +1,5 @@
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_lock.h"
 #include "colormac.h"
 
@@ -111,6 +111,8 @@ do {									\
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLushort *)(buf + (_x)*2 + (_y)*pitch) = d;
 
@@ -124,6 +126,8 @@ do {									\
 #if 0 /* Unused */
 /* 32 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = d;
 
@@ -137,6 +141,8 @@ do {									\
 
 /* 24/8 bit interleaved depth/stencil functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) {			\
    GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
    tmp &= 0xff;						\
diff --git a/src/mesa/drivers/dri/gamma/gamma_state.c b/src/mesa/drivers/dri/gamma/gamma_state.c
index 9e7626a6fd3..59272f9bc91 100644
--- a/src/mesa/drivers/dri/gamma/gamma_state.c
+++ b/src/mesa/drivers/dri/gamma/gamma_state.c
@@ -24,10 +24,10 @@
  * 3DLabs Gamma driver
  */
 
-#include "gamma_context.h"
-#include "gamma_macros.h"
+#include "gammacontext.h"
+#include "gamma_main/macros.h"
 #include "buffers.h"
-#include "macros.h"
+#include "main/macros.h"
 #include "glint_dri.h"
 #include "colormac.h"
 #include "swrast/swrast.h"
diff --git a/src/mesa/drivers/dri/gamma/gamma_tex.c b/src/mesa/drivers/dri/gamma/gamma_tex.c
index 3b4ee4e581c..2ffb790f28a 100644
--- a/src/mesa/drivers/dri/gamma/gamma_tex.c
+++ b/src/mesa/drivers/dri/gamma/gamma_tex.c
@@ -2,19 +2,19 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/texstore.h"
 #include "teximage.h"
-#include "texformat.h"
+#include "main/texformat.h"
 #include "texobj.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
-#include "gamma_context.h"
+#include "main/mm.h"
+#include "gammacontext.h"
 #include "colormac.h"
 
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_texmem.c b/src/mesa/drivers/dri/gamma/gamma_texmem.c
index 94ecb5c2f61..4cb47e179ee 100644
--- a/src/mesa/drivers/dri/gamma/gamma_texmem.c
+++ b/src/mesa/drivers/dri/gamma/gamma_texmem.c
@@ -2,16 +2,16 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "colormac.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "glint_dri.h"
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_lock.h"
 
 void gammaDestroyTexObj(gammaContextPtr gmesa, gammaTextureObjectPtr t)
diff --git a/src/mesa/drivers/dri/gamma/gamma_texstate.c b/src/mesa/drivers/dri/gamma/gamma_texstate.c
index b9bd6d4cee0..b3a318d581b 100644
--- a/src/mesa/drivers/dri/gamma/gamma_texstate.c
+++ b/src/mesa/drivers/dri/gamma/gamma_texstate.c
@@ -2,14 +2,14 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-
-#include "mm.h"
-#include "gamma_context.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+
+#include "main/mm.h"
+#include "gammacontext.h"
 
 static void gammaSetTexImages( gammaContextPtr gmesa, 
 			      struct gl_texture_object *tObj )
diff --git a/src/mesa/drivers/dri/gamma/gamma_tris.c b/src/mesa/drivers/dri/gamma/gamma_tris.c
index 83bf56a141a..2903daf3f12 100644
--- a/src/mesa/drivers/dri/gamma/gamma_tris.c
+++ b/src/mesa/drivers/dri/gamma/gamma_tris.c
@@ -25,19 +25,19 @@
  * 3DLabs Gamma driver.
  */
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_vb.h"
 #include "gamma_tris.h"
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "colormac.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 #include "tnl/t_pipeline.h"
 
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_vb.c b/src/mesa/drivers/dri/gamma/gamma_vb.c
index f23f585fc02..23ca0714c59 100644
--- a/src/mesa/drivers/dri/gamma/gamma_vb.c
+++ b/src/mesa/drivers/dri/gamma/gamma_vb.c
@@ -25,17 +25,17 @@
  * 3DLabs Gamma driver.
  */
  
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
 #include "colormac.h"
 
 #include "swrast_setup/swrast_setup.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 #include "tnl/tnl.h"
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_vb.h"
 #include "gamma_tris.h"
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_vb.h b/src/mesa/drivers/dri/gamma/gamma_vb.h
index feda25c4c6d..8701226f590 100644
--- a/src/mesa/drivers/dri/gamma/gamma_vb.h
+++ b/src/mesa/drivers/dri/gamma/gamma_vb.h
@@ -28,7 +28,7 @@
 #ifndef GAMMAVB_INC
 #define GAMMAVB_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 
 #define _GAMMA_NEW_VERTEX (_NEW_TEXTURE |		\
diff --git a/src/mesa/drivers/dri/gamma/gamma_xmesa.c b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
index 4c0ebe18992..2a28902e1e9 100644
--- a/src/mesa/drivers/dri/gamma/gamma_xmesa.c
+++ b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
@@ -24,10 +24,10 @@
  * 3DLabs Gamma driver
  */
 
-#include "gamma_context.h"
+#include "gammacontext.h"
 #include "gamma_vb.h"
-#include "context.h"
-#include "matrix.h"
+#include "main/context.h"
+#include "main/matrix.h"
 #include "glint_dri.h"
 
 #include "swrast/swrast.h"
@@ -237,7 +237,7 @@ gammaUnbindContext( __DRIcontextPrivate *driContextPriv )
    return GL_TRUE;
 }
 
-static struct __DriverAPIRec gammaAPI = {
+const struct __DriverAPIRec driDriverAPI = {
    gammaInitDriver,
    gammaDestroyScreen,
    gammaCreateContext,
diff --git a/src/mesa/drivers/dri/glcore/glcore_driver.c b/src/mesa/drivers/dri/glcore/glcore_driver.c
deleted file mode 100644
index 25778160419..00000000000
--- a/src/mesa/drivers/dri/glcore/glcore_driver.c
+++ /dev/null
@@ -1,128 +0,0 @@
-/*
- * Copyright 2006 Red Hat, Inc.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * on the rights to use, copy, modify, merge, publish, distribute, sub
- * license, and/or sell copies of the Software, and to permit persons to whom
- * the Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
- * THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
- * IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/*
- * This implements a software-only "DRI" driver.  It doesn't actually speak
- * any DRI protocol or talk to the DRM, it just looks enough like a DRI driver
- * that libglx in the server can load it for software rendering in the
- * unaccelerated case.
- */
-
-static GLboolean
-glcoreInitDriver(__DRIscreenPrivate *driScreenPriv)
-{
-}
-
-static void
-glcoreDestroyScreen(__DRIScreenPrivate *driScreenPriv)
-{
-}
-
-static GLboolean
-glcoreCreateContext(const __GLcontextModes *glVisual,
-                    __DRIcontextPrivate *driContextPriv,
-                    void *shared_context)
-{
-}
-
-static void
-glcoreDestroyContext(__DRIcontextPrivate *driContextPriv)
-{
-}
-
-static GLboolean
-glcoreCreateBuffer(__DRIscreenPrivate *driScreenPriv,
-                   __DRIdrawablePrivate *driDrawablePriv,
-                   const __GLcontextModes *mesaVisual,
-                   GLboolean isPixmap)
-{
-}
-
-static void
-glcoreDestroyBuffer(__DRIdrawablePrivate *driDrawablePriv)
-{
-}
-
-static void
-glcoreSwapBuffers(__DRIdrawablePrivate *driDrawablePriv)
-{
-}
-
-static GLboolean
-glcoreMakeCurrent(__DRIcontextPrivate *driContextPriv,
-                  __DRIdrawablePrivate *driDrawablePriv,
-                  __DRIdrawablePrivate *driReadablePriv)
-{
-}
-
-static GLboolean
-glcoreUnbindContext(__DRIcontextPrivate *driContextPriv)
-{
-}
-
-static struct __DriverAPIRec glcore_api = {
-    .InitDriver     = glcoreInitDriver,
-    .DestroyScreen  = glcoreDestroyScreen,
-    .CreateContext  = glcoreCreateContext,
-    .DestroyContext = glcoreDestroyContext,
-    .CreateBuffer   = glcoreCreateBuffer,
-    .DestroyBuffer  = glcoreDestroyBuffer,
-    .SwapBuffers    = glcoreSwapBuffers,
-    .MakeCurrent    = glcoreMakeCurrent,
-    .UnbindContext  = glcoreUnbindContext,
-};  
-
-static __GLcontextModes *
-glcoreFillInModes(unsigned pixel_bits)
-{
-}
-
-PUBLIC void *
-__driCreateNewScreen_20050727(__DRInativeDisplay *dpy, int scrn,
-                              __DRIscreen *psc, const __GLcontextModes *modes,
-                              const __DRIversion *ddx_version,
-                              const __DRIversion *dri_version,
-                              const __DRIversion *drm_version,
-                              const __DRIframebuffer *fb, drmAddress pSarea,
-                              int fd, int internal_api_version,
-                              const ___DRIinterfaceMethods *interface,
-                              __GLcontextModes **driver_modes)
-{
-    __DRIscreenPrivate *driScreenPriv;
-    glcoreDriverPrivate *glcoreDriverPriv;
-    
-    /* would normally check ddx/dri/drm versions here */
-
-    driScreenPriv = __driUtilCreateNewScreen(dpy, scrn, psc, NULL, ddx_version,
-                                             dri_version, drm_version, fb,
-                                             internal_api_version, &glcore_api);
-    if (!driScreenPriv)
-        return NULL;
-
-    glcoreDriverPriv = driScreenPriv->pDrvPriv;
-
-    *driver_modes = glcoreFillInModes(glcoreDriverPriv->bpp);
-
-    driInitExtensions(NULL, NULL, GL_FALSE);
-
-    return driScreenPriv;
-}
diff --git a/src/mesa/drivers/dri/i810/i810context.c b/src/mesa/drivers/dri/i810/i810context.c
index f90b3682f83..c281a4990e4 100644
--- a/src/mesa/drivers/dri/i810/i810context.c
+++ b/src/mesa/drivers/dri/i810/i810context.c
@@ -32,14 +32,14 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
 
-#include "glheader.h"
-#include "context.h"
-#include "matrix.h"
-#include "simple_list.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "imports.h"
-#include "points.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/matrix.h"
+#include "main/simple_list.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/imports.h"
+#include "main/points.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -423,11 +423,11 @@ void i810XMesaSetBackClipRects( i810ContextPtr imesa )
 static void i810XMesaWindowMoved( i810ContextPtr imesa )
 {
    /* Determine current color drawing buffer */
-   switch (imesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0]) {
-   case BUFFER_BIT_FRONT_LEFT:
+   switch (imesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0]) {
+   case BUFFER_FRONT_LEFT:
       i810XMesaSetFrontClipRects( imesa );
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       i810XMesaSetBackClipRects( imesa );
       break;
    default:
@@ -485,11 +485,11 @@ i810UpdatePageFlipping( i810ContextPtr imesa )
    int front = 0;
 
    /* Determine current color drawing buffer */
-   switch (ctx->DrawBuffer->_ColorDrawBufferMask[0]) {
-   case BUFFER_BIT_FRONT_LEFT:
+   switch (ctx->DrawBuffer->_ColorDrawBufferIndexes[0]) {
+   case BUFFER_FRONT_LEFT:
       front = 1;
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       front = 0;
       break;
    default:
diff --git a/src/mesa/drivers/dri/i810/i810context.h b/src/mesa/drivers/dri/i810/i810context.h
index 47080420595..4b8c71d7c6c 100644
--- a/src/mesa/drivers/dri/i810/i810context.h
+++ b/src/mesa/drivers/dri/i810/i810context.h
@@ -30,8 +30,8 @@ typedef struct i810_context_t *i810ContextPtr;
 typedef struct i810_texture_object_t *i810TextureObjectPtr;
 
 #include "drm.h"
-#include "mtypes.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
 
 #include "i810screen.h"
 #include "i810tex.h"
diff --git a/src/mesa/drivers/dri/i810/i810ioctl.c b/src/mesa/drivers/dri/i810/i810ioctl.c
index 95726fb2525..3df9c2ac478 100644
--- a/src/mesa/drivers/dri/i810/i810ioctl.c
+++ b/src/mesa/drivers/dri/i810/i810ioctl.c
@@ -1,17 +1,17 @@
 
 #include <unistd.h> /* for usleep() */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/dd.h"
 #include "swrast/swrast.h"
-#include "mm.h"
+#include "main/mm.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
 
-#include "i810context.h"
+#include "main/context.h"
 #include "i810ioctl.h"
 #include "i810state.h"
 
diff --git a/src/mesa/drivers/dri/i810/i810ioctl.h b/src/mesa/drivers/dri/i810/i810ioctl.h
index 748d29ae36a..dfd6e210889 100644
--- a/src/mesa/drivers/dri/i810/i810ioctl.h
+++ b/src/mesa/drivers/dri/i810/i810ioctl.h
@@ -33,7 +33,7 @@ do {								\
    }								\
 } while (0)
 
-static __inline GLuint *i810AllocDmaLow( i810ContextPtr imesa, int bytes )
+static INLINE GLuint *i810AllocDmaLow( i810ContextPtr imesa, int bytes )
 {
    if (imesa->vertex_low + bytes > imesa->vertex_high) 
       i810FlushPrimsGetBuffer( imesa );
diff --git a/src/mesa/drivers/dri/i810/i810render.c b/src/mesa/drivers/dri/i810/i810render.c
index a31d54236cb..1d98e006885 100644
--- a/src/mesa/drivers/dri/i810/i810render.c
+++ b/src/mesa/drivers/dri/i810/i810render.c
@@ -31,11 +31,11 @@
  * dma buffers.  Use strip/fan hardware acceleration where possible.
  *
  */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "tnl/t_context.h"
 
@@ -144,7 +144,7 @@ static GLboolean i810_run_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/i810/i810screen.c b/src/mesa/drivers/dri/i810/i810screen.c
index 733c97b2ad4..48603f5d79f 100644
--- a/src/mesa/drivers/dri/i810/i810screen.c
+++ b/src/mesa/drivers/dri/i810/i810screen.c
@@ -32,14 +32,14 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "fbobject.h"
-#include "matrix.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "main/fbobject.h"
+#include "main/matrix.h"
+#include "main/renderbuffer.h"
+#include "main/simple_list.h"
 #include "utils.h"
 
 #include "i810screen.h"
@@ -55,77 +55,13 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 extern const struct dri_extension card_extensions[];
 
-static __GLcontextModes *fill_in_modes( __GLcontextModes *modes,
-				       unsigned pixel_bits,
-				       unsigned depth_bits,
-				       unsigned stencil_bits,
-				       const GLenum * db_modes,
-				       unsigned num_db_modes,
-				       int visType )
-{
-    static const uint8_t bits[1][4] = {
-	{          5,          6,          5,          0 }
-    };
-
-    static const uint32_t masks[1][4] = {
-	{ 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 }
-    };
-
-    unsigned   i;
-    unsigned   j;
-    const unsigned index = 0;
-
-    for ( i = 0 ; i < num_db_modes ; i++ ) {
-	for ( j = 0 ; j < 2 ; j++ ) {
-
-	    modes->redBits   = bits[index][0];
-	    modes->greenBits = bits[index][1];
-	    modes->blueBits  = bits[index][2];
-	    modes->alphaBits = bits[index][3];
-	    modes->redMask   = masks[index][0];
-	    modes->greenMask = masks[index][1];
-	    modes->blueMask  = masks[index][2];
-	    modes->alphaMask = masks[index][3];
-	    modes->rgbBits   = modes->redBits + modes->greenBits
-		+ modes->blueBits + modes->alphaBits;
-
-	    modes->accumRedBits   = 16 * j;
-	    modes->accumGreenBits = 16 * j;
-	    modes->accumBlueBits  = 16 * j;
-	    modes->accumAlphaBits = (masks[index][3] != 0) ? 16 * j : 0;
-	    modes->visualRating = (j == 0) ? GLX_NONE : GLX_SLOW_CONFIG;
-
-	    modes->stencilBits = stencil_bits;
-	    modes->depthBits = depth_bits;
-
-	    modes->visualType = visType;
-	    modes->renderType = GLX_RGBA_BIT;
-	    modes->drawableType = GLX_WINDOW_BIT;
-	    modes->rgbMode = GL_TRUE;
-
-	    if ( db_modes[i] == GLX_NONE ) {
-		modes->doubleBufferMode = GL_FALSE;
-	    }
-	    else {
-		modes->doubleBufferMode = GL_TRUE;
-		modes->swapMethod = db_modes[i];
-	    }
-
-	    modes = modes->next;
-	}
-    }
-
-    return modes;
-
-}
-
-
-static __GLcontextModes *
-i810FillInModes( unsigned pixel_bits, unsigned depth_bits,
+static const __DRIconfig **
+i810FillInModes( __DRIscreenPrivate *psp,
+		 unsigned pixel_bits, unsigned depth_bits,
 		 unsigned stencil_bits, GLboolean have_back_buffer )
-{    __GLcontextModes * modes;
+{
+    __DRIconfig **configs;
     __GLcontextModes * m;
-    unsigned num_modes;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     unsigned i;
@@ -139,50 +75,42 @@ i810FillInModes( unsigned pixel_bits, unsigned depth_bits,
 	GLX_NONE, GLX_SWAP_UNDEFINED_OML /*, GLX_SWAP_COPY_OML */
     };
 
-    int depth_buffer_modes[2][2];
+    uint8_t depth_bits_array[2];
+    uint8_t stencil_bits_array[2];
 
-
-    depth_buffer_modes[0][0] = depth_bits;
-    depth_buffer_modes[1][0] = depth_bits;
+    depth_bits_array[0] = depth_bits;
+    depth_bits_array[1] = depth_bits;
 
     /* Just like with the accumulation buffer, always provide some modes
      * with a stencil buffer.  It will be a sw fallback, but some apps won't
      * care about that.
      */
-    depth_buffer_modes[0][1] = 0;
-    depth_buffer_modes[1][1] = (stencil_bits == 0) ? 8 : stencil_bits;
+    stencil_bits_array[0] = 0;
+    stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    for ( i = 0 ; i < depth_buffer_factor ; i++ ) {
-	m = fill_in_modes( m, pixel_bits,
-			   depth_buffer_modes[i][0], depth_buffer_modes[i][1],
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR );
-    }
-
-    for ( i = 0 ; i < depth_buffer_factor ; i++ ) {
-	m = fill_in_modes( m, pixel_bits,
-			   depth_buffer_modes[i][0], depth_buffer_modes[i][1],
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR );
+    configs = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor,
+			       back_buffer_modes, back_buffer_factor);
+    if (configs == NULL) {
+	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
+		 __func__, __LINE__ );
+	return NULL;
     }
 
     /* Mark the visual as slow if there are "fake" stencil bits.
      */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-	if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
+    for (i = 0; configs[i]; i++) {
+	m = &configs[i]->modes;
+	if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
 	    m->visualRating = GLX_SLOW_CONFIG;
 	}
     }
 
-    return modes;
-
+    return (const __DRIconfig **) configs;
 }
 
      
@@ -218,12 +146,24 @@ static drmBufMapPtr i810_create_empty_buffers(void)
 }
 
 
-static GLboolean
-i810InitDriver(__DRIscreenPrivate *sPriv)
+static const __DRIconfig **
+i810InitScreen(__DRIscreen *sPriv)
 {
+   static const __DRIversion ddx_expected = { 1, 0, 0 };
+   static const __DRIversion dri_expected = { 4, 0, 0 };
+   static const __DRIversion drm_expected = { 1, 2, 0 };
    i810ScreenPrivate *i810Screen;
    I810DRIPtr         gDRIPriv = (I810DRIPtr)sPriv->pDevPriv;
 
+   if ( ! driCheckDriDdxDrmVersions2( "i810",
+				      &sPriv->dri_version, & dri_expected,
+				      &sPriv->ddx_version, & ddx_expected,
+				      &sPriv->drm_version, & drm_expected ) ) {
+      return NULL;
+   }
+
+   driInitExtensions( NULL, card_extensions, GL_TRUE );
+
    if (sPriv->devPrivSize != sizeof(I810DRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(I810DRIRec) does not match passed size from device driver\n");
       return GL_FALSE;
@@ -287,8 +227,8 @@ i810InitDriver(__DRIscreenPrivate *sPriv)
 	      i810Screen->depth.handle,
 	      i810Screen->depth.size,
 	      (drmAddress *)&i810Screen->depth.map) != 0) {
-      FREE(i810Screen);
       drmUnmap(i810Screen->back.map, i810Screen->back.size);
+      FREE(i810Screen);
       sPriv->private = NULL;
       __driUtilMessage("i810InitDriver: drmMap (2) failed");
       return GL_FALSE;
@@ -311,7 +251,7 @@ i810InitDriver(__DRIscreenPrivate *sPriv)
 
    i810Screen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
 
-   return GL_TRUE;
+   return i810FillInModes(sPriv, 16, 16, 0, 1);
 }
 
 static void
@@ -400,9 +340,8 @@ i810DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
     _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
 }
 
-
-static const struct __DriverAPIRec i810API = {
-   .InitDriver      = i810InitDriver,
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = i810InitScreen,
    .DestroyScreen   = i810DestroyScreen,
    .CreateContext   = i810CreateContext,
    .DestroyContext  = i810DestroyContext,
@@ -412,60 +351,8 @@ static const struct __DriverAPIRec i810API = {
    .MakeCurrent     = i810MakeCurrent,
    .UnbindContext   = i810UnbindContext,
    .GetSwapInfo     = NULL,
-   .GetMSC          = NULL,
+   .GetDrawableMSC  = NULL,
    .WaitForMSC      = NULL,
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- *
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd,
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 1, 0, 0 };
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 2, 0 };
-
-   dri_interface = interface;
-
-   if ( ! driCheckDriDdxDrmVersions2( "i810",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &i810API);
-   if ( psp != NULL ) {
-      *driver_modes = i810FillInModes( 16,
-				       16, 0,
-				       1);
-      driInitExtensions( NULL, card_extensions, GL_TRUE );
-   }
-
-   return (void *) psp;
-}
diff --git a/src/mesa/drivers/dri/i810/i810span.c b/src/mesa/drivers/dri/i810/i810span.c
index 2112800eebc..510723f4456 100644
--- a/src/mesa/drivers/dri/i810/i810span.c
+++ b/src/mesa/drivers/dri/i810/i810span.c
@@ -1,14 +1,14 @@
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "swrast/swrast.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
 
 #include "i810span.h"
 #include "i810ioctl.h"
-#include "swrast/swrast.h"
 
 
 #define DBG 0
@@ -67,6 +67,8 @@ do {									\
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d ) \
    *(GLushort *)(buf + (_x)*2 + (_y)*pitch)  = d;
 
diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c
index e203c74f52d..1e7a6cfe471 100644
--- a/src/mesa/drivers/dri/i810/i810state.c
+++ b/src/mesa/drivers/dri/i810/i810state.c
@@ -1,11 +1,16 @@
 
 #include <stdio.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "dd.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/dd.h"
+#include "main/colormac.h"
+#include "swrast/swrast.h"
+#include "tnl/tnl.h"
+#include "tnl/t_pipeline.h"
+#include "vbo/vbo.h"
+#include "swrast_setup/swrast_setup.h"
 
 #include "texmem.h"
 
@@ -19,14 +24,8 @@
 #include "i810tris.h"
 #include "i810ioctl.h"
 
-#include "swrast/swrast.h"
-#include "tnl/tnl.h"
-#include "vbo/vbo.h"
-#include "swrast_setup/swrast_setup.h"
-
-#include "tnl/t_pipeline.h"
 
-static __inline__ GLuint i810PackColor(GLuint format,
+static INLINE GLuint i810PackColor(GLuint format,
 				       GLubyte r, GLubyte g,
 				       GLubyte b, GLubyte a)
 {
@@ -291,18 +290,20 @@ void i810DrawBuffer(GLcontext *ctx, GLenum mode )
    i810ContextPtr imesa = I810_CONTEXT(ctx);
    int front = 0;
   
-   /*
-    * _DrawDestMask is easier to cope with than <mode>.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0]) {
-   case BUFFER_BIT_FRONT_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
+      FALLBACK( imesa, I810_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0]) {
+   case BUFFER_FRONT_LEFT:
      front = 1;
      break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
      front = 0;
      break;
    default:
-      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( imesa, I810_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
diff --git a/src/mesa/drivers/dri/i810/i810tex.c b/src/mesa/drivers/dri/i810/i810tex.c
index 730bc90eafd..ba4e6b5b0b1 100644
--- a/src/mesa/drivers/dri/i810/i810tex.c
+++ b/src/mesa/drivers/dri/i810/i810tex.c
@@ -22,20 +22,21 @@
  *
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texmem.h"
-#include "texobj.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "main/colormac.h"
+#include "main/texobj.h"
+#include "main/mm.h"
 #include "swrast/swrast.h"
-#include "colormac.h"
-#include "texobj.h"
-#include "mm.h"
+
+#include "texmem.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
diff --git a/src/mesa/drivers/dri/i810/i810tex.h b/src/mesa/drivers/dri/i810/i810tex.h
index c6ab4c8e6d1..d9809270308 100644
--- a/src/mesa/drivers/dri/i810/i810tex.h
+++ b/src/mesa/drivers/dri/i810/i810tex.h
@@ -26,8 +26,8 @@
 #ifndef I810TEX_INC
 #define I810TEX_INC
 
-#include "mtypes.h"
-#include "mm.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
 
 #include "i810context.h"
 #include "i810_3d_reg.h"
diff --git a/src/mesa/drivers/dri/i810/i810texmem.c b/src/mesa/drivers/dri/i810/i810texmem.c
index 08900cc67d8..5ad66dbf5cb 100644
--- a/src/mesa/drivers/dri/i810/i810texmem.c
+++ b/src/mesa/drivers/dri/i810/i810texmem.c
@@ -23,18 +23,17 @@
  *
  */
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "colormac.h"
-#include "mm.h"
-#include "texformat.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/mm.h"
+#include "main/texformat.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
-
 #include "i810context.h"
 #include "i810tex.h"
 #include "i810state.h"
@@ -64,7 +63,7 @@ void i810DestroyTexObj(i810ContextPtr imesa, i810TextureObjectPtr t)
 /* From linux kernel i386 header files, copes with odd sizes better
  * than COPY_DWORDS would:
  */
-static __inline__ void * __memcpy(void * to, const void * from, size_t n)
+static INLINE void * __memcpy(void * to, const void * from, size_t n)
 {
 int d0, d1, d2;
 __asm__ __volatile__(
diff --git a/src/mesa/drivers/dri/i810/i810texstate.c b/src/mesa/drivers/dri/i810/i810texstate.c
index 558aef9eee4..0e09f54c41b 100644
--- a/src/mesa/drivers/dri/i810/i810texstate.c
+++ b/src/mesa/drivers/dri/i810/i810texstate.c
@@ -22,14 +22,13 @@
  *
  */
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "texformat.h"
-#include "simple_list.h"
-#include "enums.h"
-
-#include "mm.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/texformat.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/mm.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
diff --git a/src/mesa/drivers/dri/i810/i810tris.c b/src/mesa/drivers/dri/i810/i810tris.c
index 40ab436b951..b508496fb69 100644
--- a/src/mesa/drivers/dri/i810/i810tris.c
+++ b/src/mesa/drivers/dri/i810/i810tris.c
@@ -30,11 +30,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "enums.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/colormac.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -74,7 +74,7 @@ do {									\
 } while (0)
 #endif
 
-static __inline__ void i810_draw_triangle( i810ContextPtr imesa,
+static INLINE void i810_draw_triangle( i810ContextPtr imesa,
 					   i810VertexPtr v0,
 					   i810VertexPtr v1,
 					   i810VertexPtr v2 )
@@ -89,7 +89,7 @@ static __inline__ void i810_draw_triangle( i810ContextPtr imesa,
 }
 
 
-static __inline__ void i810_draw_quad( i810ContextPtr imesa,
+static INLINE void i810_draw_quad( i810ContextPtr imesa,
 				       i810VertexPtr v0,
 				       i810VertexPtr v1,
 				       i810VertexPtr v2,
@@ -108,7 +108,7 @@ static __inline__ void i810_draw_quad( i810ContextPtr imesa,
 }
 
 
-static __inline__ void i810_draw_point( i810ContextPtr imesa,
+static INLINE void i810_draw_point( i810ContextPtr imesa,
 					i810VertexPtr tmp )
 {
    GLfloat sz = 0.5 * CLAMP(imesa->glCtx->Point.Size,
@@ -132,7 +132,7 @@ static __inline__ void i810_draw_point( i810ContextPtr imesa,
 }
 
 
-static __inline__ void i810_draw_line( i810ContextPtr imesa,
+static INLINE void i810_draw_line( i810ContextPtr imesa,
 				       i810VertexPtr v0,
 				       i810VertexPtr v1 )
 {
diff --git a/src/mesa/drivers/dri/i810/i810tris.h b/src/mesa/drivers/dri/i810/i810tris.h
index 3d0dd916ca1..ab026be0a51 100644
--- a/src/mesa/drivers/dri/i810/i810tris.h
+++ b/src/mesa/drivers/dri/i810/i810tris.h
@@ -26,7 +26,7 @@
 #ifndef I810TRIS_INC
 #define I810TRIS_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void i810PrintRenderState( const char *msg, GLuint state );
 extern void i810InitTriFuncs( GLcontext *ctx );
diff --git a/src/mesa/drivers/dri/i810/i810vb.c b/src/mesa/drivers/dri/i810/i810vb.c
index 3439192b0d3..30890dc9b70 100644
--- a/src/mesa/drivers/dri/i810/i810vb.c
+++ b/src/mesa/drivers/dri/i810/i810vb.c
@@ -24,18 +24,17 @@
  */
  
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
-
 #include "i810context.h"
 #include "i810vb.h"
 #include "i810ioctl.h"
diff --git a/src/mesa/drivers/dri/i810/i810vb.h b/src/mesa/drivers/dri/i810/i810vb.h
index 55d0d2409e1..1f704e45695 100644
--- a/src/mesa/drivers/dri/i810/i810vb.h
+++ b/src/mesa/drivers/dri/i810/i810vb.h
@@ -26,7 +26,7 @@
 #ifndef I810VB_INC
 #define I810VB_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 
 #define _I810_NEW_VERTEX (_NEW_TEXTURE |			\
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 805abf75e08..5858e0ee9fd 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -7,16 +7,6 @@ LIBNAME = i915_dri.so
 MINIGLX_SOURCES = server/intel_dri.c
 
 DRIVER_SOURCES = \
-	i915_context.c \
-	i915_debug.c \
-	i915_fragprog.c \
-	i915_metaops.c \
-	i915_program.c \
-	i915_state.c \
-	i915_tex.c \
-	i915_texprog.c \
-	i915_texstate.c \
-	i915_vtbl.c \
 	i830_context.c \
 	i830_metaops.c \
 	i830_state.c \
@@ -24,18 +14,44 @@ DRIVER_SOURCES = \
 	i830_tex.c \
 	i830_texstate.c \
 	i830_vtbl.c \
+	intel_render.c \
+	intel_regions.c \
+	intel_buffer_objects.c \
 	intel_batchbuffer.c \
-	intel_context.c \
-	intel_ioctl.c \
+	intel_mipmap_tree.c \
+	i915_tex_layout.c \
+	intel_tex_layout.c \
+	intel_tex_image.c \
+	intel_tex_subimage.c \
+	intel_tex_copy.c \
+	intel_tex_validate.c \
+	intel_tex_format.c \
+	intel_tex.c \
 	intel_pixel.c \
-	intel_render.c \
-	intel_rotate.c \
+	intel_pixel_bitmap.c \
+	intel_pixel_copy.c \
+	intel_pixel_draw.c \
+	intel_pixel_read.c \
+	intel_buffers.c \
+	intel_blit.c \
+	i915_tex.c \
+	i915_texstate.c \
+	i915_context.c \
+	i915_debug.c \
+	i915_debug_fp.c \
+	i915_fragprog.c \
+	i915_metaops.c \
+	i915_program.c \
+	i915_state.c \
+	i915_vtbl.c \
+	intel_context.c \
+	intel_decode.c \
 	intel_screen.c \
 	intel_span.c \
 	intel_state.c \
-	intel_tex.c \
-	intel_texmem.c \
-	intel_tris.c 
+	intel_tris.c \
+	intel_fbo.c \
+	intel_depthstencil.c
 
 C_SOURCES = \
 	$(COMMON_SOURCES) \
@@ -43,8 +59,16 @@ C_SOURCES = \
 
 ASM_SOURCES = 
 
+DRIVER_DEFINES = -I../intel -I../intel/server -DI915 \
+	$(shell pkg-config libdrm --atleast-version=2.3.1 \
+				&& echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP")
 
+DRI_LIB_DEPS += -ldrm_intel
 
 include ../Makefile.template
 
+intel_decode.o: ../intel/intel_decode.c
+
+intel_tex_layout.o: ../intel/intel_tex_layout.c
+
 symlinks:
diff --git a/src/mesa/drivers/dri/i915/i830_context.c b/src/mesa/drivers/dri/i915/i830_context.c
index 7ca601e1b5c..09b1ec922ff 100644
--- a/src/mesa/drivers/dri/i915/i830_context.c
+++ b/src/mesa/drivers/dri/i915/i830_context.c
@@ -26,99 +26,87 @@
  **************************************************************************/
 
 #include "i830_context.h"
-#include "imports.h"
+#include "main/imports.h"
 #include "texmem.h"
 #include "intel_tex.h"
 #include "tnl/tnl.h"
 #include "tnl/t_vertex.h"
 #include "tnl/t_context.h"
+#include "tnl/t_pipeline.h"
 #include "utils.h"
+#include "intel_span.h"
+#include "intel_pixel.h"
+#include "intel_tris.h"
 
 /***************************************
  * Mesa's Driver Functions
  ***************************************/
 
-static const struct dri_extension i830_extensions[] =
+static void
+i830InitDriverFunctions(struct dd_function_table *functions)
 {
-    { "GL_ARB_texture_env_crossbar",       NULL },
-    { NULL,                                NULL }
-};
-
-
-static void i830InitDriverFunctions( struct dd_function_table *functions )
-{
-   intelInitDriverFunctions( functions );
-   i830InitStateFuncs( functions );
-   i830InitTextureFuncs( functions );
+   intelInitDriverFunctions(functions);
+   i830InitStateFuncs(functions);
+   i830InitTextureFuncs(functions);
 }
 
+extern const struct tnl_pipeline_stage *intel_pipeline[];
 
-GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate)
+GLboolean
+i830CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate)
 {
    struct dd_function_table functions;
-   i830ContextPtr i830 = (i830ContextPtr) CALLOC_STRUCT(i830_context);
-   intelContextPtr intel = &i830->intel;
+   struct i830_context *i830 = CALLOC_STRUCT(i830_context);
+   struct intel_context *intel = &i830->intel;
    GLcontext *ctx = &intel->ctx;
-   GLuint i;
-   if (!i830) return GL_FALSE;
+   if (!i830)
+      return GL_FALSE;
 
-   i830InitVtbl( i830 );
-   i830InitDriverFunctions( &functions );
+   i830InitVtbl(i830);
+   i830InitDriverFunctions(&functions);
 
-   if (!intelInitContext( intel, mesaVis, driContextPriv,
-			  sharedContextPrivate, &functions )) {
+   if (!intelInitContext(intel, mesaVis, driContextPriv,
+                         sharedContextPrivate, &functions)) {
       FREE(i830);
       return GL_FALSE;
    }
 
+   /* Initialize swrast, tnl driver tables: */
+   intelInitSpanFuncs(ctx);
+   intelInitTriFuncs(ctx);
+
+   /* Install the customized pipeline: */
+   _tnl_destroy_pipeline(ctx);
+   _tnl_install_pipeline(ctx, intel_pipeline);
+
+   if (intel->no_rast)
+      FALLBACK(intel, INTEL_FALLBACK_USER, 1);
+
    intel->ctx.Const.MaxTextureUnits = I830_TEX_UNITS;
    intel->ctx.Const.MaxTextureImageUnits = I830_TEX_UNITS;
    intel->ctx.Const.MaxTextureCoordUnits = I830_TEX_UNITS;
 
-   intel->nr_heaps = 1;
-   intel->texture_heaps[0] = 
-      driCreateTextureHeap( 0, intel,
-			    intel->intelScreen->tex.size,
-			    12,
-			    I830_NR_TEX_REGIONS,
-			    intel->sarea->texList,
-			    (unsigned *) & intel->sarea->texAge,
-			    & intel->swapped,
-			    sizeof( struct i830_texture_object ),
-			    (destroy_texture_object_t *)intelDestroyTexObj );
-
-   /* FIXME: driCalculateMaxTextureLevels assumes that mipmaps are tightly
-    * FIXME: packed, but they're not in Intel graphics hardware.
+   /* Advertise the full hardware capabilities.  The new memory
+    * manager should cope much better with overload situations:
     */
-   intel->ctx.Const.MaxTextureUnits = I830_TEX_UNITS;
-   i = driQueryOptioni( &intel->optionCache, "allow_large_textures");
-   driCalculateMaxTextureLevels( intel->texture_heaps,
-				 intel->nr_heaps,
-				 &intel->ctx.Const,
-				 4,
-				 11, /* max 2D texture size is 2048x2048 */
-				 8,  /* max 3D texture size is 256^3 */
-				 10, /* max CUBE texture size is 1024x1024 */
-				 11, /* max RECT. supported */
-				 12,
-				 GL_FALSE,
-				 i );
-
-   _tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-		       18 * sizeof(GLfloat) );
-
-   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
+   ctx->Const.MaxTextureLevels = 12;
+   ctx->Const.Max3DTextureLevels = 9;
+   ctx->Const.MaxCubeTextureLevels = 11;
+   ctx->Const.MaxTextureRectSize = (1 << 11);
+   ctx->Const.MaxTextureUnits = I830_TEX_UNITS;
 
-   driInitExtensions( ctx, i830_extensions, GL_FALSE );
+   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
+                      18 * sizeof(GLfloat));
 
-   i830InitState( i830 );
+   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
 
+   i830InitState(i830);
+   i830InitMetaFuncs(i830);
 
-   _tnl_allow_vertex_fog( ctx, 1 ); 
-   _tnl_allow_pixel_fog( ctx, 0 ); 
+   _tnl_allow_vertex_fog(ctx, 1);
+   _tnl_allow_pixel_fog(ctx, 0);
 
    return GL_TRUE;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_context.h b/src/mesa/drivers/dri/i915/i830_context.h
index bae777dd5a4..1bdb32049d7 100644
--- a/src/mesa/drivers/dri/i915/i830_context.h
+++ b/src/mesa/drivers/dri/i915/i830_context.h
@@ -49,17 +49,21 @@
  */
 #define I830_DESTREG_CBUFADDR0 0
 #define I830_DESTREG_CBUFADDR1 1
-#define I830_DESTREG_CBUFADDR2 2
-#define I830_DESTREG_DBUFADDR0 3
-#define I830_DESTREG_DBUFADDR1 4
-#define I830_DESTREG_DBUFADDR2 5
-#define I830_DESTREG_DV0 6
-#define I830_DESTREG_DV1 7
-#define I830_DESTREG_SENABLE 8
-#define I830_DESTREG_SR0 9
-#define I830_DESTREG_SR1 10
-#define I830_DESTREG_SR2 11
-#define I830_DEST_SETUP_SIZE 12
+#define I830_DESTREG_DBUFADDR0 2
+#define I830_DESTREG_DBUFADDR1 3
+#define I830_DESTREG_DV0 4
+#define I830_DESTREG_DV1 5
+#define I830_DESTREG_SENABLE 6
+#define I830_DESTREG_SR0 7
+#define I830_DESTREG_SR1 8
+#define I830_DESTREG_SR2 9
+#define I830_DESTREG_DRAWRECT0 10
+#define I830_DESTREG_DRAWRECT1 11
+#define I830_DESTREG_DRAWRECT2 12
+#define I830_DESTREG_DRAWRECT3 13
+#define I830_DESTREG_DRAWRECT4 14
+#define I830_DESTREG_DRAWRECT5 15
+#define I830_DEST_SETUP_SIZE 16
 
 #define I830_CTXREG_STATE1		0
 #define I830_CTXREG_STATE2		1
@@ -73,7 +77,7 @@
 #define I830_CTXREG_AA			9
 #define I830_CTXREG_FOGCOLOR		10
 #define I830_CTXREG_BLENDCOLOR0		11
-#define I830_CTXREG_BLENDCOLOR1		12 
+#define I830_CTXREG_BLENDCOLOR1		12
 #define I830_CTXREG_VF			13
 #define I830_CTXREG_VF2			14
 #define I830_CTXREG_MCSB0		15
@@ -84,17 +88,16 @@
 #define I830_STPREG_ST1        1
 #define I830_STP_SETUP_SIZE    2
 
-#define I830_TEXREG_TM0LI      0 /* load immediate 2 texture map n */
-#define I830_TEXREG_TM0S0      1
-#define I830_TEXREG_TM0S1      2
-#define I830_TEXREG_TM0S2      3
-#define I830_TEXREG_TM0S3      4
-#define I830_TEXREG_TM0S4      5
-#define I830_TEXREG_MCS	       6	/* _3DSTATE_MAP_COORD_SETS */
-#define I830_TEXREG_CUBE       7	/* _3DSTATE_MAP_SUBE */
-#define I830_TEX_SETUP_SIZE    8
+#define I830_TEXREG_TM0LI      0        /* load immediate 2 texture map n */
+#define I830_TEXREG_TM0S1      1
+#define I830_TEXREG_TM0S2      2
+#define I830_TEXREG_TM0S3      3
+#define I830_TEXREG_TM0S4      4
+#define I830_TEXREG_MCS	       5        /* _3DSTATE_MAP_COORD_SETS */
+#define I830_TEXREG_CUBE       6        /* _3DSTATE_MAP_SUBE */
+#define I830_TEX_SETUP_SIZE    7
 
-#define I830_TEXBLEND_SIZE	12	/* (4 args + op) * 2 + COLOR_FACTOR */
+#define I830_TEXBLEND_SIZE	12      /* (4 args + op) * 2 + COLOR_FACTOR */
 
 struct i830_texture_object
 {
@@ -104,30 +107,39 @@ struct i830_texture_object
 
 #define I830_TEX_UNITS 4
 
-struct i830_hw_state {
+struct i830_hw_state
+{
    GLuint Ctx[I830_CTX_SETUP_SIZE];
    GLuint Buffer[I830_DEST_SETUP_SIZE];
    GLuint Stipple[I830_STP_SETUP_SIZE];
    GLuint Tex[I830_TEX_UNITS][I830_TEX_SETUP_SIZE];
    GLuint TexBlend[I830_TEX_UNITS][I830_TEXBLEND_SIZE];
    GLuint TexBlendWordsUsed[I830_TEX_UNITS];
-   GLuint emitted;		/* I810_UPLOAD_* */
+
+   struct intel_region *draw_region;
+   struct intel_region *depth_region;
+
+   /* Regions aren't actually that appropriate here as the memory may
+    * be from a PBO or FBO.  Will have to do this for draw and depth for
+    * FBO's...
+    */
+   dri_bo *tex_buffer[I830_TEX_UNITS];
+   GLuint tex_offset[I830_TEX_UNITS];
+
+   GLuint emitted;              /* I810_UPLOAD_* */
    GLuint active;
 };
 
-struct i830_context 
+struct i830_context
 {
    struct intel_context intel;
-   
-   DECLARE_RENDERINPUTS(last_index_bitset);
+
+   GLuint lodbias_tm0s3[MAX_TEXTURE_UNITS];
+     DECLARE_RENDERINPUTS(last_index_bitset);
 
    struct i830_hw_state meta, initial, state, *current;
 };
 
-typedef struct i830_context *i830ContextPtr;
-typedef struct i830_texture_object *i830TextureObjectPtr;
-
-#define I830_CONTEXT(ctx)	((i830ContextPtr)(ctx))
 
 
 
@@ -148,71 +160,56 @@ do {						\
 
 /* i830_vtbl.c
  */
-extern void 
-i830InitVtbl( i830ContextPtr i830 );
+extern void i830InitVtbl(struct i830_context *i830);
 
+extern void
+i830_state_draw_region(struct intel_context *intel,
+                       struct i830_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region);
 /* i830_context.c
  */
-extern GLboolean 
-i830CreateContext( const __GLcontextModes *mesaVis,
-		   __DRIcontextPrivate *driContextPriv,
-		   void *sharedContextPrivate);
+extern GLboolean
+i830CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate);
 
 /* i830_tex.c, i830_texstate.c
  */
-extern void 
-i830UpdateTextureState( intelContextPtr intel );
-
-extern void 
-i830InitTextureFuncs( struct dd_function_table *functions );
+extern void i830UpdateTextureState(struct intel_context *intel);
 
-extern intelTextureObjectPtr
-i830AllocTexObj( struct gl_texture_object *tObj );
+extern void i830InitTextureFuncs(struct dd_function_table *functions);
 
 /* i830_texblend.c
  */
-extern GLuint i830SetTexEnvCombine(i830ContextPtr i830,
-    const struct gl_tex_env_combine_state * combine, GLint blendUnit,
-     GLuint texel_op, GLuint *state, const GLfloat *factor );
+extern GLuint i830SetTexEnvCombine(struct i830_context *i830,
+                                   const struct gl_tex_env_combine_state
+                                   *combine, GLint blendUnit, GLuint texel_op,
+                                   GLuint * state, const GLfloat * factor);
 
-extern void 
-i830EmitTextureBlend( i830ContextPtr i830 );
+extern void i830EmitTextureBlend(struct i830_context *i830);
 
 
 /* i830_state.c
  */
-extern void 
-i830InitStateFuncs( struct dd_function_table *functions );
+extern void i830InitStateFuncs(struct dd_function_table *functions);
 
-extern void 
-i830EmitState( i830ContextPtr i830 );
+extern void i830EmitState(struct i830_context *i830);
 
-extern void 
-i830InitState( i830ContextPtr i830 );
+extern void i830InitState(struct i830_context *i830);
 
 /* i830_metaops.c
  */
-extern GLboolean
-i830TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels );
-
-extern GLboolean
-i830TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels );
-
-extern void 
-i830ClearWithTris( intelContextPtr intel, GLbitfield mask,
-		   GLboolean all, GLint cx, GLint cy, GLint cw, GLint ch);
+extern void i830InitMetaFuncs(struct i830_context *i830);
 
-extern void
-i830RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf);
+/*======================================================================
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct i830_context *
+i830_context(GLcontext * ctx)
+{
+   return (struct i830_context *) ctx;
+}
 
 #endif
-
diff --git a/src/mesa/drivers/dri/i915/i830_metaops.c b/src/mesa/drivers/dri/i915/i830_metaops.c
index c1d7fe349c0..2cce661c868 100644
--- a/src/mesa/drivers/dri/i915/i830_metaops.c
+++ b/src/mesa/drivers/dri/i915/i830_metaops.c
@@ -25,15 +25,15 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "utils.h"
 
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
-#include "intel_ioctl.h"
+#include "intel_regions.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
@@ -41,34 +41,26 @@
 /* A large amount of state doesn't need to be uploaded.
  */
 #define ACTIVE (I830_UPLOAD_INVARIENT |         \
-		I830_UPLOAD_TEXBLEND(0) |	\
-		I830_UPLOAD_STIPPLE |		\
 		I830_UPLOAD_CTX |		\
 		I830_UPLOAD_BUFFERS |		\
-		I830_UPLOAD_TEX(0))		
+		I830_UPLOAD_STIPPLE |		\
+		I830_UPLOAD_TEXBLEND(0) |	\
+		I830_UPLOAD_TEX(0))
 
 
 #define SET_STATE( i830, STATE )		\
 do {						\
-   i830->current->emitted = 0;			\
+   i830->current->emitted &= ~ACTIVE;			\
    i830->current = &i830->STATE;		\
-   i830->current->emitted = 0;			\
+   i830->current->emitted &= ~ACTIVE;			\
 } while (0)
 
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void set_initial_state( i830ContextPtr i830 )
-{
-   memcpy(&i830->meta, &i830->initial, sizeof(i830->meta) );
-   i830->meta.active = ACTIVE;
-   i830->meta.emitted = 0;
-}
-
 
-static void set_no_depth_stencil_write( i830ContextPtr i830 )
+static void
+set_no_stencil_write(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
     */
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
@@ -76,6 +68,13 @@ static void set_no_depth_stencil_write( i830ContextPtr i830 )
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_STENCIL_WRITE;
 
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
+}
+
+static void
+set_no_depth_write(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
     */
@@ -87,35 +86,56 @@ static void set_no_depth_stencil_write( i830ContextPtr i830 )
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
-/* Set stencil unit to replace always with the reference value.
+/* Set depth unit to replace.
  */
-static void set_stencil_replace( i830ContextPtr i830,
-				 GLuint s_mask,
-				 GLuint s_clear)
+static void
+set_depth_replace(struct intel_context *intel)
 {
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
+    * ctx->Driver.DepthMask( ctx, GL_TRUE )
     */
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DEPTH_WRITE;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DEPTH_WRITE;
+
+   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
+    */
+   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
+   i830->meta.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
+                                          DEPTH_TEST_FUNC
+                                          (COMPAREFUNC_ALWAYS));
+
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
+}
+
+
+/* Set stencil unit to replace always with the reference value.
+ */
+static void
+set_stencil_replace(struct intel_context *intel,
+                    GLuint s_mask, GLuint s_clear)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
+    */
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
 
    /* ctx->Driver.StencilMask( ctx, s_mask )
     */
    i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					   STENCIL_WRITE_MASK((s_mask&0xff)));
+                                          STENCIL_WRITE_MASK((s_mask &
+                                                              0xff)));
 
    /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
     */
    i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |= 
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
       (ENABLE_STENCIL_PARMS |
        STENCIL_FAIL_OP(STENCILOP_REPLACE) |
        STENCIL_PASS_DEPTH_FAIL_OP(STENCILOP_REPLACE) |
@@ -125,14 +145,14 @@ static void set_stencil_replace( i830ContextPtr i830,
     */
    i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					   STENCIL_TEST_MASK(0xff));
+                                          STENCIL_TEST_MASK(0xff));
 
    i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-						ENABLE_STENCIL_TEST_FUNC_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |= 
+                                               ENABLE_STENCIL_TEST_FUNC_MASK);
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
       (ENABLE_STENCIL_REF_VALUE |
        ENABLE_STENCIL_TEST_FUNC |
-       STENCIL_REF_VALUE((s_clear&0xff)) |
+       STENCIL_REF_VALUE((s_clear & 0xff)) |
        STENCIL_TEST_FUNC(COMPAREFUNC_ALWAYS));
 
 
@@ -141,38 +161,43 @@ static void set_stencil_replace( i830ContextPtr i830,
 }
 
 
-static void set_color_mask( i830ContextPtr i830, GLboolean state )
+static void
+set_color_mask(struct intel_context *intel, GLboolean state)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
    const GLuint mask = ((1 << WRITEMASK_RED_SHIFT) |
-			(1 << WRITEMASK_GREEN_SHIFT) |
-			(1 << WRITEMASK_BLUE_SHIFT) |
-			(1 << WRITEMASK_ALPHA_SHIFT));
+                        (1 << WRITEMASK_GREEN_SHIFT) |
+                        (1 << WRITEMASK_BLUE_SHIFT) |
+                        (1 << WRITEMASK_ALPHA_SHIFT));
 
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~mask;
 
    if (state) {
-      i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= 
-	 (i830->state.Ctx[I830_CTXREG_ENABLES_2] & mask);
+      i830->meta.Ctx[I830_CTXREG_ENABLES_2] |=
+         (i830->state.Ctx[I830_CTXREG_ENABLES_2] & mask);
    }
-      
+
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
 /* Installs a one-stage passthrough texture blend pipeline.  Is there
  * more that can be done to turn off texturing?
  */
-static void set_no_texture( i830ContextPtr i830 )
+static void
+set_no_texture(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
    static const struct gl_tex_env_combine_state comb = {
       GL_NONE, GL_NONE,
-      { GL_TEXTURE, 0, 0, }, { GL_TEXTURE, 0, 0, },
-      { GL_SRC_COLOR, 0, 0 }, { GL_SRC_ALPHA, 0, 0 },
+      {GL_TEXTURE, 0, 0,}, {GL_TEXTURE, 0, 0,},
+      {GL_SRC_COLOR, 0, 0}, {GL_SRC_ALPHA, 0, 0},
       0, 0, 0, 0
    };
 
    i830->meta.TexBlendWordsUsed[0] =
-     i830SetTexEnvCombine( i830, & comb, 0, TEXBLENDARG_TEXEL0,
-			   i830->meta.TexBlend[0], NULL);
+      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
+                           i830->meta.TexBlend[0], NULL);
 
    i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
    i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
@@ -181,18 +206,22 @@ static void set_no_texture( i830ContextPtr i830 )
 /* Set up a single element blend stage for 'replace' texturing with no
  * funny ops.
  */
-static void enable_texture_blend_replace( i830ContextPtr i830 )
+static void
+set_texture_blend_replace(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
    static const struct gl_tex_env_combine_state comb = {
       GL_REPLACE, GL_REPLACE,
-      { GL_TEXTURE, GL_TEXTURE, GL_TEXTURE }, { GL_TEXTURE, GL_TEXTURE, GL_TEXTURE, },
-      { GL_SRC_COLOR, GL_SRC_COLOR, GL_SRC_COLOR }, { GL_SRC_ALPHA, GL_SRC_ALPHA, GL_SRC_ALPHA },
+      {GL_TEXTURE, GL_TEXTURE, GL_TEXTURE,}, {GL_TEXTURE, GL_TEXTURE,
+                                              GL_TEXTURE,},
+      {GL_SRC_COLOR, GL_SRC_COLOR, GL_SRC_COLOR}, {GL_SRC_ALPHA, GL_SRC_ALPHA,
+                                                   GL_SRC_ALPHA},
       0, 0, 1, 1
    };
 
    i830->meta.TexBlendWordsUsed[0] =
-     i830SetTexEnvCombine( i830, & comb, 0, TEXBLENDARG_TEXEL0,
-			   i830->meta.TexBlend[0], NULL);
+      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
+                           i830->meta.TexBlend[0], NULL);
 
    i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
    i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
@@ -206,717 +235,222 @@ static void enable_texture_blend_replace( i830ContextPtr i830 )
 /* Set up an arbitary piece of memory as a rectangular texture
  * (including the front or back buffer).
  */
-static void set_tex_rect_source( i830ContextPtr i830,
-				 GLuint offset,
-				 GLuint width, 
-				 GLuint height,
-				 GLuint pitch, /* in bytes */
-				 GLuint textureFormat )
+static GLboolean
+set_tex_rect_source(struct intel_context *intel,
+                    dri_bo *buffer,
+                    GLuint offset,
+                    GLuint pitch, GLuint height, GLenum format, GLenum type)
 {
-   GLint numLevels = 1;
+   struct i830_context *i830 = i830_context(&intel->ctx);
    GLuint *setup = i830->meta.Tex[0];
+   GLint numLevels = 1;
+   GLuint textureFormat;
+   GLuint cpp;
 
-/*    fprintf(stderr, "%s: offset: %x w: %d h: %d pitch %d format %x\n", */
-/* 	   __FUNCTION__, offset, width, height, pitch, textureFormat ); */
+   /* A full implementation of this would do the upload through
+    * glTexImage2d, and get all the conversion operations at that
+    * point.  We are restricted, but still at least have access to the
+    * fragment program swizzle.
+    */
+   switch (format) {
+   case GL_BGRA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGBA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_BGR:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5_REV:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGB:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
 
-   setup[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 | 
-			       (LOAD_TEXTURE_MAP0 << 0) | 4);
-   setup[I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE | offset);
+   default:
+      return GL_FALSE;
+   }
+
+   i830->meta.tex_buffer[0] = buffer;
+   i830->meta.tex_offset[0] = offset;
+
+   setup[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
+                               (LOAD_TEXTURE_MAP0 << 0) | 4);
    setup[I830_TEXREG_TM0S1] = (((height - 1) << TM0S1_HEIGHT_SHIFT) |
-			       ((width - 1) << TM0S1_WIDTH_SHIFT) |
-			       textureFormat);
-   setup[I830_TEXREG_TM0S2] = ((((pitch / 4) - 1) << TM0S2_PITCH_SHIFT));   
-   setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAX_MIP_MASK;
-   setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_MIP_MASK;
-   setup[I830_TEXREG_TM0S3] |= ((numLevels - 1)*4) << TM0S3_MIN_MIP_SHIFT;
+                               ((pitch - 1) << TM0S1_WIDTH_SHIFT) |
+                               textureFormat);
+   setup[I830_TEXREG_TM0S2] =
+      (((((pitch * cpp) / 4) -
+         1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
+
+   setup[I830_TEXREG_TM0S3] =
+      ((((numLevels -
+          1) *
+         4) << TM0S3_MIN_MIP_SHIFT) | (FILTER_NEAREST <<
+                                       TM0S3_MIN_FILTER_SHIFT) |
+       (MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT) | (FILTER_NEAREST <<
+                                                     TM0S3_MAG_FILTER_SHIFT));
+
+   setup[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(0));
 
    setup[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-			     MAP_UNIT(0) |
-			     ENABLE_TEXCOORD_PARAMS |
-			     TEXCOORDS_ARE_IN_TEXELUNITS |
-			     TEXCOORDTYPE_CARTESIAN |
-			     ENABLE_ADDR_V_CNTL |
-			     TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
-			     ENABLE_ADDR_U_CNTL |
-			     TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
+                             MAP_UNIT(0) |
+                             ENABLE_TEXCOORD_PARAMS |
+                             TEXCOORDS_ARE_IN_TEXELUNITS |
+                             TEXCOORDTYPE_CARTESIAN |
+                             ENABLE_ADDR_V_CNTL |
+                             TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
+                             ENABLE_ADDR_U_CNTL |
+                             TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
 
    i830->meta.emitted &= ~I830_UPLOAD_TEX(0);
+   return GL_TRUE;
 }
 
 
-/* Select between front and back draw buffers.
- */
-static void set_draw_region( i830ContextPtr i830,
-			      const intelRegion *region )
-{
-   i830->meta.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->meta.Buffer[I830_DESTREG_CBUFADDR2] = region->offset;
-   i830->meta.emitted &= ~I830_UPLOAD_BUFFERS;
-}
-
-/* Setup an arbitary draw format, useful for targeting
- * texture or agp memory.
- */
-#if 0
-static void set_draw_format( i830ContextPtr i830,
-			     GLuint format,
-			     GLuint depth_format)
-{
-   i830->meta.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					  DSTORG_VERT_BIAS(0x8) | /* .5 */
-					  format |
-					  DEPTH_IS_Z |
-					  depth_format);
-}
-#endif
-
-
-static void set_vertex_format( i830ContextPtr i830 )
+static void
+set_vertex_format(struct intel_context *intel)
 {
-   i830->meta.Ctx[I830_CTXREG_VF] =  (_3DSTATE_VFT0_CMD |
-				      VFT0_TEX_COUNT(1) |
-				      VFT0_DIFFUSE |
-				      VFT0_SPEC |
-				      VFT0_XYZW);
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830->meta.Ctx[I830_CTXREG_VF] = (_3DSTATE_VFT0_CMD |
+                                     VFT0_TEX_COUNT(1) |
+                                     VFT0_DIFFUSE | VFT0_XYZ);
    i830->meta.Ctx[I830_CTXREG_VF2] = (_3DSTATE_VFT1_CMD |
-				      VFT1_TEX0_FMT(TEXCOORDFMT_2D) |
-				      VFT1_TEX1_FMT(TEXCOORDFMT_2D) | 
-				      VFT1_TEX2_FMT(TEXCOORDFMT_2D) |
-				      VFT1_TEX3_FMT(TEXCOORDFMT_2D));
+                                      VFT1_TEX0_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX1_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX2_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX3_FMT(TEXCOORDFMT_2D));
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
 
-static void draw_quad(i830ContextPtr i830, 
-		      GLfloat x0, GLfloat x1,
-		      GLfloat y0, GLfloat y1, 
-		      GLubyte red, GLubyte green,
-		      GLubyte blue, GLubyte alpha,
-		      GLfloat s0, GLfloat s1,
-		      GLfloat t0, GLfloat t1 )
-{
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i830->intel, 
-						PRIM3D_TRIFAN, 
-						4*vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i;
-
-   
-/*    fprintf(stderr, "%s: %f,%f-%f,%f 0x%x%x%x%x %f,%f-%f,%f\n", */
-/* 	   __FUNCTION__, */
-/* 	   x0,y0,x1,y1,red,green,blue,alpha,s0,t0,s1,t1); */
-
-
-   /* initial vertex, left bottom */
-   tmp.v.x = x0;
-   tmp.v.y = y0;
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0;
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-   tmp.v.u0 = s0;
-   tmp.v.v0 = t0;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right bottom */
-   vb += 8;
-   tmp.v.x = x1;
-   tmp.v.u0 = s1;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right top */
-   vb += 8;
-   tmp.v.y = y1;
-   tmp.v.v0 = t1;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* left top */
-   vb += 8;
-   tmp.v.x = x0;
-   tmp.v.u0 = s0;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-/*    fprintf(stderr, "%s: DV1: %x\n",  */
-/* 	   __FUNCTION__, i830->meta.Buffer[I830_DESTREG_DV1]); */
-}
-
-static void draw_poly(i830ContextPtr i830, 
-		      GLubyte red, GLubyte green, GLubyte blue, GLubyte alpha,
-                      GLuint numVerts,
-                      GLfloat verts[][2],
-                      GLfloat texcoords[][2])
+static void
+meta_import_pixel_state(struct intel_context *intel)
 {
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i830->intel, 
-						PRIM3D_TRIFAN, 
-						numVerts * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i, k;
-
-   /* initial constant vertex fields */
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-
-   for (k = 0; k < numVerts; k++) {
-      tmp.v.x = verts[k][0];
-      tmp.v.y = verts[k][1];
-      tmp.v.u0 = texcoords[k][0];
-      tmp.v.v0 = texcoords[k][1];
-
-      for (i = 0 ; i < vertex_size ; i++)
-         vb[i] = tmp.ui[i];
-
-      vb += vertex_size;
-   }
-}
-
-void 
-i830ClearWithTris(intelContextPtr intel, GLbitfield mask,
-		  GLboolean allFoo,
-		  GLint cxFoo, GLint cyFoo, GLint cwFoo, GLint chFoo)
-{
-   i830ContextPtr i830 = I830_CONTEXT( intel );
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   intelScreenPrivate *screen = intel->intelScreen;
-   int x0, y0, x1, y1;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-
-   INTEL_FIREVERTICES(intel);
-   SET_STATE( i830, meta );
-   set_initial_state( i830 );
-/*    set_no_texture( i830 ); */
-   set_vertex_format( i830 ); 
-
-   LOCK_HARDWARE(intel);
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   if(!all) {
-      x0 = cx;
-      y0 = cy;
-      x1 = x0 + cw;
-      y1 = y0 + ch;
-   } else {
-      x0 = 0;
-      y0 = 0;
-      x1 = x0 + dPriv->w;
-      y1 = y0 + dPriv->h;
-   }
-
-   /* Don't do any clipping to screen - these are window coordinates.
-    * The active cliprects will be applied as for any other geometry.
-    */
-
-   if(mask & BUFFER_BIT_FRONT_LEFT) {
-      set_no_depth_stencil_write( i830 );
-      set_color_mask( i830, GL_TRUE );
-      set_draw_region( i830, &screen->front );
-      draw_quad(i830, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if(mask & BUFFER_BIT_BACK_LEFT) {
-      set_no_depth_stencil_write( i830 );
-      set_color_mask( i830, GL_TRUE );
-      set_draw_region( i830, &screen->back );
-
-      draw_quad(i830, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if(mask & BUFFER_BIT_STENCIL) {
-      set_stencil_replace( i830, 
-			   intel->ctx.Stencil.WriteMask[0], 
-			   intel->ctx.Stencil.Clear);
-
-      set_color_mask( i830, GL_FALSE );
-      set_draw_region( i830, &screen->front );
-      draw_quad( i830, x0, x1, y0, y1, 0, 0, 0, 0, 0, 0, 0, 0 );
-   }
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   i830->meta.Ctx[I830_CTXREG_STATE1] = i830->state.Ctx[I830_CTXREG_STATE1];
+   i830->meta.Ctx[I830_CTXREG_STATE2] = i830->state.Ctx[I830_CTXREG_STATE2];
+   i830->meta.Ctx[I830_CTXREG_STATE3] = i830->state.Ctx[I830_CTXREG_STATE3];
+   i830->meta.Ctx[I830_CTXREG_STATE4] = i830->state.Ctx[I830_CTXREG_STATE4];
+   i830->meta.Ctx[I830_CTXREG_STATE5] = i830->state.Ctx[I830_CTXREG_STATE5];
+   i830->meta.Ctx[I830_CTXREG_IALPHAB] = i830->state.Ctx[I830_CTXREG_IALPHAB];
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] =
+      i830->state.Ctx[I830_CTXREG_STENCILTST];
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] =
+      i830->state.Ctx[I830_CTXREG_ENABLES_1];
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] =
+      i830->state.Ctx[I830_CTXREG_ENABLES_2];
+   i830->meta.Ctx[I830_CTXREG_AA] = i830->state.Ctx[I830_CTXREG_AA];
+   i830->meta.Ctx[I830_CTXREG_FOGCOLOR] =
+      i830->state.Ctx[I830_CTXREG_FOGCOLOR];
+   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR0] =
+      i830->state.Ctx[I830_CTXREG_BLENDCOLOR0];
+   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR1] =
+      i830->state.Ctx[I830_CTXREG_BLENDCOLOR1];
+   i830->meta.Ctx[I830_CTXREG_MCSB0] = i830->state.Ctx[I830_CTXREG_MCSB0];
+   i830->meta.Ctx[I830_CTXREG_MCSB1] = i830->state.Ctx[I830_CTXREG_MCSB1];
+
+
+   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~CULLMODE_MASK;
+   i830->meta.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
 
-   UNLOCK_HARDWARE(intel);
 
-   INTEL_FIREVERTICES(intel);
-   SET_STATE( i830, state );
+   i830->meta.Buffer[I830_DESTREG_SENABLE] =
+      i830->state.Buffer[I830_DESTREG_SENABLE];
+   i830->meta.Buffer[I830_DESTREG_SR1] = i830->state.Buffer[I830_DESTREG_SR1];
+   i830->meta.Buffer[I830_DESTREG_SR2] = i830->state.Buffer[I830_DESTREG_SR2];
+   i830->meta.emitted &= ~I830_UPLOAD_BUFFERS;
 }
 
 
-#if 0
 
-GLboolean
-i830TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels )
+/* Select between front and back draw buffers.
+ */
+static void
+meta_draw_region(struct intel_context *intel,
+                 struct intel_region *color_region,
+                 struct intel_region *depth_region)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   intelScreenPrivate *screen = i830->intel.intelScreen;
-   GLint pitch = pack->RowLength ? pack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = i830->intel.driDrawable;
-   int textureFormat;
-   GLenum glTextureFormat;
-   int src_offset = i830->meta.Buffer[I830_DESTREG_CBUFADDR2];
-   int destOffset = intelAgpOffsetFromVirtual( &i830->intel, pixels);
-   int destFormat, depthFormat, destPitch;
-   drm_clip_rect_t tmp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-
-   if (	ctx->_ImageTransferState ||
-	pack->SwapBytes ||
-	pack->LsbFirst ||
-	!pack->Invert) {
-      fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   switch (screen->fbFormat) {
-   case DV_PF_565:
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      glTextureFormat = GL_RGB;
-      break;
-   case DV_PF_555:
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      glTextureFormat = GL_RGBA;
-      break;
-   case DV_PF_8888:
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      glTextureFormat = GL_RGBA;
-      break;
-   default:
-      fprintf(stderr, "%s: textureFormat failed %x\n", __FUNCTION__,
-	      screen->fbFormat);
-      return GL_FALSE;
-   }
-
-
-   switch (type) {
-   case GL_UNSIGNED_SHORT_5_6_5: 
-      if (format != GL_RGB) return GL_FALSE;
-      destFormat = COLR_BUF_RGB565; 
-      depthFormat = DEPTH_FRMT_16_FIXED;
-      destPitch = pitch * 2;
-      break;
-   case GL_UNSIGNED_INT_8_8_8_8_REV: 
-      if (format != GL_BGRA) return GL_FALSE;
-      destFormat = COLR_BUF_ARGB8888; 
-      depthFormat = DEPTH_FRMT_24_FIXED_8_OTHER;
-      destPitch = pitch * 4;
-      break;
-   default:
-      fprintf(stderr, "%s: destFormat failed %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(type));
-      return GL_FALSE;
-   }
-
-   destFormat |= (0x02<<24);
-
-/*    fprintf(stderr, "type: %s destFormat: %x\n", */
-/* 	   _mesa_lookup_enum_by_nr(type), */
-/* 	   destFormat); */
-
-   intelFlush( ctx );
-
-   SET_STATE( i830, meta );
-   set_initial_state( i830 );
-   set_no_depth_stencil_write( i830 );
-
-   LOCK_HARDWARE( intel );
-   {
-      intelWaitForIdle( intel ); /* required by GL */
-
-      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-#if 0
-      /* FIXME -- Just emit the correct state
-       */
-      if (i830SetParam(i830->driFd, I830_SETPARAM_CBUFFER_PITCH, 
-		      destPitch) != 0) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: setparam failed\n", __FUNCTION__);
-	 return GL_FALSE;
-      }
-#endif
-
-
-      y = dPriv->h - y - height;
-      x += dPriv->x;
-      y += dPriv->y;
-
-
-      /* Set the frontbuffer up as a large rectangular texture.
-       */
-      set_tex_rect_source( i830, 
-			   src_offset, 
-			   screen->width, 
-			   screen->height, 
-			   screen->front.pitch, 
-			   textureFormat ); 
-   
-   
-      enable_texture_blend_replace( i830 ); 
-
-
-      /* Set the 3d engine to draw into the agp memory
-       */
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
-      set_draw_region( i830, destOffset ); 
-      set_draw_format( i830, destFormat, depthFormat );  
-
-
-      /* Draw a single quad, no cliprects:
-       */
-      i830->intel.numClipRects = 1;
-      i830->intel.pClipRects = &tmp;
-      i830->intel.pClipRects[0].x1 = 0;
-      i830->intel.pClipRects[0].y1 = 0;
-      i830->intel.pClipRects[0].x2 = width;
-      i830->intel.pClipRects[0].y2 = height;
-
-      draw_quad( i830, 
-		 0, width, 0, height, 
-		 0, 255, 0, 0, 
-		 x, x+width, y, y+height );
-
-      intelWindowMoved( intel );
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( ctx ); /* required by GL */
-
-   SET_STATE( i830, state );
-   return GL_TRUE;
+   i830_state_draw_region(intel, &i830->meta, color_region, depth_region);
 }
 
 
-GLboolean
-i830TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels )
+/* Operations where the 3D engine is decoupled temporarily from the
+ * current GL state and used for other purposes than simply rendering
+ * incoming triangles.
+ */
+static void
+install_meta_state(struct intel_context *intel)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLint pitch = unpack->RowLength ? unpack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   int textureFormat;
-   GLenum glTextureFormat;
-   int dst_offset = i830->meta.Buffer[I830_DESTREG_CBUFADDR2];
-   int src_offset = intelAgpOffsetFromVirtual( intel, pixels );
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Todo -- upload images that aren't in agp space, then texture
-    * from them.  
-    */
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   memcpy(&i830->meta, &i830->initial, sizeof(i830->meta));
 
-   if ( !intelIsAgpMemory( intel, pixels, pitch*height ) ) {
-      fprintf(stderr, "%s: intelIsAgpMemory failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Todo -- don't want to clobber all the drawing state like we do
-    * for readpixels -- most of this state can be handled just fine.
-    */
-   if (	ctx->_ImageTransferState ||
-	unpack->SwapBytes ||
-	unpack->LsbFirst ||
-	ctx->Color.AlphaEnabled || 
-	ctx->Depth.Test ||
-	ctx->Fog.Enabled ||
-	ctx->Scissor.Enabled ||
-	ctx->Stencil.Enabled ||
-	!ctx->Color.ColorMask[0] ||
-	!ctx->Color.ColorMask[1] ||
-	!ctx->Color.ColorMask[2] ||
-	!ctx->Color.ColorMask[3] ||
-	ctx->Color.ColorLogicOpEnabled ||
-	ctx->Texture._EnabledUnits ||
-	ctx->Depth.OcclusionTest) {
-      fprintf(stderr, "%s: other tests failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Todo -- remove these restrictions:
-    */
-   if (ctx->Pixel.ZoomX != 1.0F ||
-       ctx->Pixel.ZoomY != -1.0F)
-      return GL_FALSE;
-
-
-
-   switch (type) {
-   case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-      if (format != GL_BGRA) return GL_FALSE;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      glTextureFormat = GL_RGBA;
-      break;
-   case GL_UNSIGNED_SHORT_5_6_5: 
-      if (format != GL_RGB) return GL_FALSE;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      glTextureFormat = GL_RGB;
-      break;
-   case GL_UNSIGNED_SHORT_8_8_MESA: 
-      if (format != GL_YCBCR_MESA) return GL_FALSE;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY 
-/*  		       | TM0S1_COLORSPACE_CONVERSION */
-	 );
-      glTextureFormat = GL_YCBCR_MESA;
-      break;
-   case GL_UNSIGNED_SHORT_8_8_REV_MESA: 
-      if (format != GL_YCBCR_MESA) return GL_FALSE;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL 
-/* 		       | TM0S1_COLORSPACE_CONVERSION */
-	 );
-      glTextureFormat = GL_YCBCR_MESA;
-      break;
-   case GL_UNSIGNED_INT_8_8_8_8_REV: 
-      if (format != GL_BGRA) return GL_FALSE;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      glTextureFormat = GL_RGBA;
-      break;
-   default:
-      fprintf(stderr, "%s: destFormat failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   intelFlush( ctx );
-
-   SET_STATE( i830, meta );
-
-   LOCK_HARDWARE( intel );
-   {
-      intelWaitForIdle( intel ); /* required by GL */
-
-      y -= height;			/* cope with pixel zoom */
-
-      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-
-      y = dPriv->h - y - height;
-
-      set_initial_state( i830 );
-
-      /* Set the pixel image up as a rectangular texture.
-       */
-      set_tex_rect_source( i830, 
-			   src_offset, 
-			   width, 
-			   height, 
-			   pitch, /* XXXX!!!! -- /2 sometimes */
-			   textureFormat ); 
-   
-   
-      enable_texture_blend_replace( i830 ); 
-
-   
-      /* Draw to the current draw buffer:
-       */
-      set_draw_offset( i830, dst_offset );
-
-      /* Draw a quad, use regular cliprects
-       */
-/*       fprintf(stderr, "x: %d y: %d width %d height %d\n", x, y, width, height); */
+   i830->meta.active = ACTIVE;
+   i830->meta.emitted = 0;
 
-      draw_quad( i830, 
-		 x, x+width, y, y+height,
-		 0, 255, 0, 0, 
-		 0, width, 0, height );
+   SET_STATE(i830, meta);
+   set_vertex_format(intel);
+   set_no_texture(intel);
+}
 
-      intelWindowMoved( intel );
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( ctx ); /* required by GL */
-   
+static void
+leave_meta_state(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   intel_region_release(&i830->meta.draw_region);
+   intel_region_release(&i830->meta.depth_region);
+/*    intel_region_release(intel, &i830->meta.tex_region[0]); */
    SET_STATE(i830, state);
-
-   return GL_TRUE;
 }
 
-#endif
 
-/**
- * Copy the window contents named by dPriv to the rotated (or reflected)
- * color buffer.
- * srcBuf is BUFFER_BIT_FRONT_LEFT or BUFFER_BIT_BACK_LEFT to indicate the source.
- */
+
 void
-i830RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf)
+i830InitMetaFuncs(struct i830_context *i830)
 {
-   i830ContextPtr i830 = I830_CONTEXT( intel );
-   intelScreenPrivate *screen = intel->intelScreen;
-   const GLuint cpp = screen->cpp;
-   drm_clip_rect_t fullRect;
-   GLuint textureFormat, srcOffset, srcPitch;
-   const drm_clip_rect_t *clipRects;
-   int numClipRects;
-   int i;
-
-   int xOrig, yOrig;
-   int origNumClipRects;
-   drm_clip_rect_t *origRects;
-
-   /*
-    * set up hardware state
-    */
-   intelFlush( &intel->ctx );
-
-   SET_STATE( i830, meta ); 
-   set_initial_state( i830 ); 
-   set_no_texture( i830 ); 
-   set_vertex_format( i830 ); 
-   set_no_depth_stencil_write( i830 );
-   set_color_mask( i830, GL_FALSE );
-
-   LOCK_HARDWARE(intel);
-
-   /* save current drawing origin and cliprects (restored at end) */
-   xOrig = intel->drawX;
-   yOrig = intel->drawY;
-   origNumClipRects = intel->numClipRects;
-   origRects = intel->pClipRects;
-
-   if (!intel->numClipRects)
-      goto done;
-
-   /*
-    * set drawing origin, cliprects for full-screen access to rotated screen
-    */
-   fullRect.x1 = 0;
-   fullRect.y1 = 0;
-   fullRect.x2 = screen->rotatedWidth;
-   fullRect.y2 = screen->rotatedHeight;
-   intel->drawX = 0;
-   intel->drawY = 0;
-   intel->numClipRects = 1;
-   intel->pClipRects = &fullRect;
-
-   set_draw_region( i830, &screen->rotated );
-
-   if (cpp == 4)
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   else
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-
-   if (srcBuf == BUFFER_BIT_FRONT_LEFT) {
-      srcPitch = screen->front.pitch;   /* in bytes */
-      srcOffset = screen->front.offset; /* bytes */
-      clipRects = dPriv->pClipRects;
-      numClipRects = dPriv->numClipRects;
-   }
-   else {
-      srcPitch = screen->back.pitch;   /* in bytes */
-      srcOffset = screen->back.offset; /* bytes */
-      clipRects = dPriv->pBackClipRects;
-      numClipRects = dPriv->numBackClipRects;
-   }
-
-   /* set the whole screen up as a texture to avoid alignment issues */
-   set_tex_rect_source(i830,
-                       srcOffset,
-                       screen->width,
-                       screen->height,
-                       srcPitch,
-                       textureFormat);
-
-   enable_texture_blend_replace(i830);
-
-   /*
-    * loop over the source window's cliprects
-    */
-   for (i = 0; i < numClipRects; i++) {
-      int srcX0 = clipRects[i].x1;
-      int srcY0 = clipRects[i].y1;
-      int srcX1 = clipRects[i].x2;
-      int srcY1 = clipRects[i].y2;
-      GLfloat verts[4][2], tex[4][2];
-      int j;
-
-      /* build vertices for four corners of clip rect */
-      verts[0][0] = srcX0;  verts[0][1] = srcY0;
-      verts[1][0] = srcX1;  verts[1][1] = srcY0;
-      verts[2][0] = srcX1;  verts[2][1] = srcY1;
-      verts[3][0] = srcX0;  verts[3][1] = srcY1;
-
-      /* .. and texcoords */
-      tex[0][0] = srcX0;  tex[0][1] = srcY0;
-      tex[1][0] = srcX1;  tex[1][1] = srcY0;
-      tex[2][0] = srcX1;  tex[2][1] = srcY1;
-      tex[3][0] = srcX0;  tex[3][1] = srcY1;
-
-      /* transform coords to rotated screen coords */
-
-      for (j = 0; j < 4; j++) {
-         matrix23TransformCoordf(&screen->rotMatrix,
-                                 &verts[j][0], &verts[j][1]);
-      }
-
-      /* draw polygon to map source image to dest region */
-      draw_poly(i830, 255, 255, 255, 255, 4, verts, tex);
-
-   } /* cliprect loop */
-
-   intelFlushBatchLocked( intel, GL_FALSE, GL_FALSE, GL_FALSE );
-
- done:
-   /* restore original drawing origin and cliprects */
-   intel->drawX = xOrig;
-   intel->drawY = yOrig;
-   intel->numClipRects = origNumClipRects;
-   intel->pClipRects = origRects;
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i830, state );
+   i830->intel.vtbl.install_meta_state = install_meta_state;
+   i830->intel.vtbl.leave_meta_state = leave_meta_state;
+   i830->intel.vtbl.meta_no_depth_write = set_no_depth_write;
+   i830->intel.vtbl.meta_no_stencil_write = set_no_stencil_write;
+   i830->intel.vtbl.meta_stencil_replace = set_stencil_replace;
+   i830->intel.vtbl.meta_depth_replace = set_depth_replace;
+   i830->intel.vtbl.meta_color_mask = set_color_mask;
+   i830->intel.vtbl.meta_no_texture = set_no_texture;
+   i830->intel.vtbl.meta_texture_blend_replace = set_texture_blend_replace;
+   i830->intel.vtbl.meta_tex_rect_source = set_tex_rect_source;
+   i830->intel.vtbl.meta_draw_region = meta_draw_region;
+   i830->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_reg.h b/src/mesa/drivers/dri/i915/i830_reg.h
index 98cee2f214c..d210c2d08e4 100644
--- a/src/mesa/drivers/dri/i915/i830_reg.h
+++ b/src/mesa/drivers/dri/i915/i830_reg.h
@@ -407,7 +407,7 @@
 #define LOGICOP_SET			0xf
 #define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
 #define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		((x)<<8)
+#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
 #define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
 #define ENABLE_STENCIL_WRITE_MASK	(1<<16)
 #define STENCIL_WRITE_MASK(x)		((x)&0xff)
@@ -494,10 +494,6 @@
 #define VFT1_TEX0_FMT(x)	(x)
 #define VFT1_TEX0_MASK          3
 #define VFT1_TEX1_SHIFT         2
-#define TEXCOORDFMT_2D		0
-#define TEXCOORDFMT_3D		1
-#define TEXCOORDFMT_4D		2
-#define TEXCOORDFMT_1D		3
 
 /*New stuff picked up along the way */
 
@@ -554,8 +550,8 @@
 #define    MAPSURF_4BIT_INDEXED		   (7<<6)
 #define TM0S1_MT_FORMAT_MASK         (0x7 << 3)
 #define TM0S1_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3) /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_IDX_RGB565	           (0<<3) /* SURFACE_8BIT_INDEXED */
+#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
+#define    MT_8BIT_IDX_RGB565	           (0<<3)       /* SURFACE_8BIT_INDEXED */
 #define    MT_8BIT_IDX_ARGB1555	           (1<<3)
 #define    MT_8BIT_IDX_ARGB4444	           (2<<3)
 #define    MT_8BIT_IDX_AY88		   (3<<3)
@@ -563,9 +559,9 @@
 #define    MT_8BIT_IDX_BUMP_88DVDU 	   (5<<3)
 #define    MT_8BIT_IDX_BUMP_655LDVDU	   (6<<3)
 #define    MT_8BIT_IDX_ARGB8888	           (7<<3)
-#define    MT_8BIT_I8		           (0<<3) /* SURFACE_8BIT */
+#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
 #define    MT_8BIT_L8		           (1<<3)
-#define    MT_16BIT_RGB565 		   (0<<3) /* SURFACE_16BIT */
+#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
 #define    MT_16BIT_ARGB1555		   (1<<3)
 #define    MT_16BIT_ARGB4444		   (2<<3)
 #define    MT_16BIT_AY88		   (3<<3)
@@ -573,16 +569,17 @@
 #define    MT_16BIT_BUMP_88DVDU	           (5<<3)
 #define    MT_16BIT_BUMP_655LDVDU	   (6<<3)
 #define    MT_16BIT_DIB_RGB565_8888	   (7<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3) /* SURFACE_32BIT */
+#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
 #define    MT_32BIT_ABGR8888		   (1<<3)
+#define    MT_32BIT_XRGB8888		   (2<<3)       /* XXX: Guess from i915_reg.h */
 #define    MT_32BIT_BUMP_XLDVDU_8888	   (6<<3)
 #define    MT_32BIT_DIB_8888		   (7<<3)
-#define    MT_411_YUV411		   (0<<3) /* SURFACE_411 */
-#define    MT_422_YCRCB_SWAPY	           (0<<3) /* SURFACE_422 */
+#define    MT_411_YUV411		   (0<<3)       /* SURFACE_411 */
+#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
 #define    MT_422_YCRCB_NORMAL	           (1<<3)
 #define    MT_422_YCRCB_SWAPUV	           (2<<3)
 #define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3) /* SURFACE_COMPRESSED */
+#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
 #define    MT_COMPRESS_DXT2_3	           (1<<3)
 #define    MT_COMPRESS_DXT4_5	           (2<<3)
 #define    MT_COMPRESS_FXT1		   (3<<3)
@@ -634,8 +631,4 @@
 #define ENABLE_TEX_STREAM_MAP_IDX	(1<<3)
 #define TEX_STREAM_MAP_IDX(x)		(x)
 
-
-#define MI_FLUSH           ((0<<29)|(4<<23))
-#define FLUSH_MAP_CACHE    (1<<0)
-
 #endif
diff --git a/src/mesa/drivers/dri/i915/i830_state.c b/src/mesa/drivers/dri/i915/i830_state.c
index f7980201f9a..d9cad0c4bf8 100644
--- a/src/mesa/drivers/dri/i915/i830_state.c
+++ b/src/mesa/drivers/dri/i915/i830_state.c
@@ -26,11 +26,11 @@
  **************************************************************************/
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/dd.h"
 
 #include "texmem.h"
 
@@ -38,149 +38,151 @@
 
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
+#include "intel_fbo.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
 
+#define FILE_DEBUG_FLAG DEBUG_STATE
+
 static void
-i830StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func, GLint ref,
+i830StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
                         GLuint mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
 
    mask = mask & 0xff;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(func), ref, mask);
+   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(func), ref, mask);
 
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					STENCIL_TEST_MASK(mask));
+                                           STENCIL_TEST_MASK(mask));
    i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-					     ENABLE_STENCIL_TEST_FUNC_MASK);
+                                                ENABLE_STENCIL_TEST_FUNC_MASK);
    i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_REF_VALUE |
-					    ENABLE_STENCIL_TEST_FUNC |
-					    STENCIL_REF_VALUE(ref) |
-					    STENCIL_TEST_FUNC(test));
+                                               ENABLE_STENCIL_TEST_FUNC |
+                                               STENCIL_REF_VALUE(ref) |
+                                               STENCIL_TEST_FUNC(test));
 }
 
 static void
-i830StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+i830StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : mask 0x%x\n", __FUNCTION__, mask);
+   struct i830_context *i830 = i830_context(ctx);
 
+   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
+   
    mask = mask & 0xff;
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(mask));
+                                           STENCIL_WRITE_MASK(mask));
 }
 
 static void
-i830StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
+i830StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
                       GLenum zpass)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int fop, dfop, dpop;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(fail),
-	      _mesa_lookup_enum_by_nr(zfail),
-	      _mesa_lookup_enum_by_nr(zpass));
+   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(fail),
+       _mesa_lookup_enum_by_nr(zfail), 
+       _mesa_lookup_enum_by_nr(zpass));
 
-   fop = 0; dfop = 0; dpop = 0;
+   fop = 0;
+   dfop = 0;
+   dpop = 0;
 
-   switch(fail) {
-   case GL_KEEP: 
-      fop = STENCILOP_KEEP; 
+   switch (fail) {
+   case GL_KEEP:
+      fop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      fop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      fop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      fop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      fop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       fop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       fop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      fop = STENCILOP_INCR; 
+      fop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      fop = STENCILOP_DECR; 
+      fop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      fop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      fop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
-   switch(zfail) {
-   case GL_KEEP: 
-      dfop = STENCILOP_KEEP; 
+   switch (zfail) {
+   case GL_KEEP:
+      dfop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      dfop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      dfop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      dfop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      dfop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       dfop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       dfop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      dfop = STENCILOP_INCR; 
+      dfop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      dfop = STENCILOP_DECR; 
+      dfop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      dfop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      dfop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
-   switch(zpass) {
-   case GL_KEEP: 
-      dpop = STENCILOP_KEEP; 
+   switch (zpass) {
+   case GL_KEEP:
+      dpop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      dpop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      dpop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      dpop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      dpop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       dpop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       dpop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      dpop = STENCILOP_INCR; 
+      dpop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      dpop = STENCILOP_DECR; 
+      dpop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      dpop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      dpop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
 
@@ -188,27 +190,30 @@ i830StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
    i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_PARMS |
-					    STENCIL_FAIL_OP(fop) |
-					    STENCIL_PASS_DEPTH_FAIL_OP(dfop) |
-					    STENCIL_PASS_DEPTH_PASS_OP(dpop));
+                                               STENCIL_FAIL_OP(fop) |
+                                               STENCIL_PASS_DEPTH_FAIL_OP
+                                               (dfop) |
+                                               STENCIL_PASS_DEPTH_PASS_OP
+                                               (dpop));
 }
 
-static void i830AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+static void
+i830AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
    GLubyte refByte;
    GLuint refInt;
 
    UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
-   refInt = (GLuint)refByte;
+   refInt = (GLuint) refByte;
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE2] &= ~ALPHA_TEST_REF_MASK;
    i830->state.Ctx[I830_CTXREG_STATE2] |= (ENABLE_ALPHA_TEST_FUNC |
-					ENABLE_ALPHA_REF_VALUE |
-					ALPHA_TEST_FUNC(test) |
-					ALPHA_REF_VALUE(refInt));
+                                           ENABLE_ALPHA_REF_VALUE |
+                                           ALPHA_TEST_FUNC(test) |
+                                           ALPHA_REF_VALUE(refInt));
 }
 
 /**
@@ -221,45 +226,49 @@ static void i830AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
  * This function is substantially different from the old i830-specific driver.
  * I'm not sure which is correct.
  */
-static void i830EvalLogicOpBlendState(GLcontext *ctx)
+static void
+i830EvalLogicOpBlendState(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
    if (RGBA_LOGICOP_ENABLED(ctx)) {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-					      ENABLE_LOGIC_OP);
-   } else if (ctx->Color.BlendEnabled) {
+                                                 ENABLE_LOGIC_OP);
+   }
+   else if (ctx->Color.BlendEnabled) {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (ENABLE_COLOR_BLEND |
-					      DISABLE_LOGIC_OP);
-   } else {
+                                                 DISABLE_LOGIC_OP);
+   }
+   else {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-					      DISABLE_LOGIC_OP);
+                                                 DISABLE_LOGIC_OP);
    }
 }
 
-static void i830BlendColor(GLcontext *ctx, const GLfloat color[4])
+static void
+i830BlendColor(GLcontext * ctx, const GLfloat color[4])
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    GLubyte r, g, b, a;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] = (a<<24) | (r<<16) | (g<<8) | b;
+   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] =
+      (a << 24) | (r << 16) | (g << 8) | b;
 }
 
 /**
@@ -268,9 +277,10 @@ static void i830BlendColor(GLcontext *ctx, const GLfloat color[4])
  * function because some blend equations (i.e., \c GL_MIN and \c GL_MAX)
  * change the interpretation of the blend function.
  */
-static void i830_set_blend_state( GLcontext * ctx )
+static void
+i830_set_blend_state(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int funcA;
    int funcRGB;
    int eqnA;
@@ -279,71 +289,72 @@ static void i830_set_blend_state( GLcontext * ctx )
    int s1;
 
 
-   funcRGB = SRC_BLND_FACT( intel_translate_blend_factor( ctx->Color.BlendSrcRGB ) )
-       | DST_BLND_FACT( intel_translate_blend_factor( ctx->Color.BlendDstRGB ) );
+   funcRGB =
+      SRC_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcRGB))
+      | DST_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstRGB));
 
-   switch(ctx->Color.BlendEquationRGB) {
+   switch (ctx->Color.BlendEquationRGB) {
    case GL_FUNC_ADD:
-      eqnRGB = BLENDFUNC_ADD; 
+      eqnRGB = BLENDFUNC_ADD;
       break;
    case GL_MIN:
       eqnRGB = BLENDFUNC_MIN;
       funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_MAX: 
+   case GL_MAX:
       eqnRGB = BLENDFUNC_MAX;
       funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_FUNC_SUBTRACT: 
-      eqnRGB = BLENDFUNC_SUB; 
+   case GL_FUNC_SUBTRACT:
+      eqnRGB = BLENDFUNC_SUB;
       break;
    case GL_FUNC_REVERSE_SUBTRACT:
-      eqnRGB = BLENDFUNC_RVRSE_SUB; 
+      eqnRGB = BLENDFUNC_RVRSE_SUB;
       break;
    default:
-      fprintf( stderr, "[%s:%u] Invalid RGB blend equation (0x%04x).\n",
-	       __FUNCTION__, __LINE__, ctx->Color.BlendEquationRGB );
+      fprintf(stderr, "[%s:%u] Invalid RGB blend equation (0x%04x).\n",
+              __FUNCTION__, __LINE__, ctx->Color.BlendEquationRGB);
       return;
    }
 
 
-   funcA = SRC_ABLEND_FACT( intel_translate_blend_factor( ctx->Color.BlendSrcA ) )
-       | DST_ABLEND_FACT( intel_translate_blend_factor( ctx->Color.BlendDstA ) );
+   funcA = SRC_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcA))
+      | DST_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstA));
 
-   switch(ctx->Color.BlendEquationA) {
+   switch (ctx->Color.BlendEquationA) {
    case GL_FUNC_ADD:
-      eqnA = BLENDFUNC_ADD; 
+      eqnA = BLENDFUNC_ADD;
       break;
-   case GL_MIN: 
+   case GL_MIN:
       eqnA = BLENDFUNC_MIN;
       funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_MAX: 
+   case GL_MAX:
       eqnA = BLENDFUNC_MAX;
       funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_FUNC_SUBTRACT: 
-      eqnA = BLENDFUNC_SUB; 
+   case GL_FUNC_SUBTRACT:
+      eqnA = BLENDFUNC_SUB;
       break;
    case GL_FUNC_REVERSE_SUBTRACT:
-      eqnA = BLENDFUNC_RVRSE_SUB; 
+      eqnA = BLENDFUNC_RVRSE_SUB;
       break;
    default:
-      fprintf( stderr, "[%s:%u] Invalid alpha blend equation (0x%04x).\n",
-	       __FUNCTION__, __LINE__, ctx->Color.BlendEquationA );
+      fprintf(stderr, "[%s:%u] Invalid alpha blend equation (0x%04x).\n",
+              __FUNCTION__, __LINE__, ctx->Color.BlendEquationA);
       return;
    }
 
    iab = eqnA | funcA
-       | _3DSTATE_INDPT_ALPHA_BLEND_CMD
-       | ENABLE_SRC_ABLEND_FACTOR | ENABLE_DST_ABLEND_FACTOR
-       | ENABLE_ALPHA_BLENDFUNC;
+      | _3DSTATE_INDPT_ALPHA_BLEND_CMD
+      | ENABLE_SRC_ABLEND_FACTOR | ENABLE_DST_ABLEND_FACTOR
+      | ENABLE_ALPHA_BLENDFUNC;
    s1 = eqnRGB | funcRGB
-       | _3DSTATE_MODES_1_CMD
-       | ENABLE_SRC_BLND_FACTOR | ENABLE_DST_BLND_FACTOR
-       | ENABLE_COLR_BLND_FUNC;
+      | _3DSTATE_MODES_1_CMD
+      | ENABLE_SRC_BLND_FACTOR | ENABLE_DST_BLND_FACTOR
+      | ENABLE_COLR_BLND_FUNC;
 
-   if ( (eqnA | funcA) != (eqnRGB | funcRGB) )
+   if ((eqnA | funcA) != (eqnRGB | funcRGB))
       iab |= ENABLE_INDPT_ALPHA_BLEND;
    else
       iab |= DISABLE_INDPT_ALPHA_BLEND;
@@ -363,70 +374,68 @@ static void i830_set_blend_state( GLcontext * ctx )
    i830EvalLogicOpBlendState(ctx);
 
    if (0) {
-      fprintf(stderr, "[%s:%u] STATE1: 0x%08x IALPHAB: 0x%08x blend is %sabled\n",
-	      __FUNCTION__, __LINE__,
-	      i830->state.Ctx[I830_CTXREG_STATE1],
-	      i830->state.Ctx[I830_CTXREG_IALPHAB],
-	      (ctx->Color.BlendEnabled) ? "en" : "dis");
+      fprintf(stderr,
+              "[%s:%u] STATE1: 0x%08x IALPHAB: 0x%08x blend is %sabled\n",
+              __FUNCTION__, __LINE__, i830->state.Ctx[I830_CTXREG_STATE1],
+              i830->state.Ctx[I830_CTXREG_IALPHAB],
+              (ctx->Color.BlendEnabled) ? "en" : "dis");
    }
 }
 
 
-static void i830BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB,
-				      GLenum modeA) 
+static void
+i830BlendEquationSeparate(GLcontext * ctx, GLenum modeRGB, GLenum modeA)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s -> %s, %s\n", __FUNCTION__,
-	     _mesa_lookup_enum_by_nr(modeRGB),
-	     _mesa_lookup_enum_by_nr(modeA));
+   DBG("%s -> %s, %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(modeRGB),
+       _mesa_lookup_enum_by_nr(modeA));
 
    (void) modeRGB;
    (void) modeA;
-   i830_set_blend_state( ctx );
+   i830_set_blend_state(ctx);
 }
 
 
-static void i830BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, 
-				  GLenum dfactorRGB, GLenum sfactorA,
-				  GLenum dfactorA )
+static void
+i830BlendFuncSeparate(GLcontext * ctx, GLenum sfactorRGB,
+                      GLenum dfactorRGB, GLenum sfactorA, GLenum dfactorA)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s -> RGB(%s, %s) A(%s, %s)\n", __FUNCTION__,
-	     _mesa_lookup_enum_by_nr(sfactorRGB),
-	     _mesa_lookup_enum_by_nr(dfactorRGB),
-	     _mesa_lookup_enum_by_nr(sfactorA),
-	     _mesa_lookup_enum_by_nr(dfactorA));
+   DBG("%s -> RGB(%s, %s) A(%s, %s)\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(sfactorRGB),
+       _mesa_lookup_enum_by_nr(dfactorRGB),
+       _mesa_lookup_enum_by_nr(sfactorA),
+       _mesa_lookup_enum_by_nr(dfactorA));
 
    (void) sfactorRGB;
    (void) dfactorRGB;
    (void) sfactorA;
    (void) dfactorA;
-   i830_set_blend_state( ctx );
+   i830_set_blend_state(ctx);
 }
 
 
 
-static void i830DepthFunc(GLcontext *ctx, GLenum func)
+static void
+i830DepthFunc(GLcontext * ctx, GLenum func)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
    i830->state.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
-				       DEPTH_TEST_FUNC(test));
+                                           DEPTH_TEST_FUNC(test));
 }
 
-static void i830DepthMask(GLcontext *ctx, GLboolean flag)
+static void
+i830DepthMask(GLcontext * ctx, GLboolean flag)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s flag (%d)\n", __FUNCTION__, flag);
+   struct i830_context *i830 = i830_context(ctx);
 
+   DBG("%s flag (%d)\n", __FUNCTION__, flag);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
    i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
@@ -443,14 +452,15 @@ static void i830DepthMask(GLcontext *ctx, GLboolean flag)
  * The i830 supports a 4x4 stipple natively, GL wants 32x32.
  * Fortunately stipple is usually a repeating pattern.
  */
-static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
+static void
+i830PolygonStipple(GLcontext * ctx, const GLubyte * mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    const GLubyte *m = mask;
    GLubyte p[4];
-   int i,j,k;
+   int i, j, k;
    int active = (ctx->Polygon.StippleFlag &&
-		 i830->intel.reduced_primitive == GL_TRIANGLES);
+                 i830->intel.reduced_primitive == GL_TRIANGLES);
    GLuint newMask;
 
    if (active) {
@@ -458,23 +468,26 @@ static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
       i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
    }
 
-   p[0] = mask[12] & 0xf; p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf; p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf; p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf; p[3] |= p[3] << 4;
-
-   for (k = 0 ; k < 8 ; k++)
-      for (j = 3 ; j >= 0; j--)
-	 for (i = 0 ; i < 4 ; i++, m++)
-	    if (*m != p[j]) {
-	       i830->intel.hw_stipple = 0;
-	       return;
-	    }
+   p[0] = mask[12] & 0xf;
+   p[0] |= p[0] << 4;
+   p[1] = mask[8] & 0xf;
+   p[1] |= p[1] << 4;
+   p[2] = mask[4] & 0xf;
+   p[2] |= p[2] << 4;
+   p[3] = mask[0] & 0xf;
+   p[3] |= p[3] << 4;
+
+   for (k = 0; k < 8; k++)
+      for (j = 3; j >= 0; j--)
+         for (i = 0; i < 4; i++, m++)
+            if (*m != p[j]) {
+               i830->intel.hw_stipple = 0;
+               return;
+            }
 
    newMask = (((p[0] & 0xf) << 0) |
-	      ((p[1] & 0xf) << 4) |
-	      ((p[2] & 0xf) << 8) |
-	      ((p[3] & 0xf) << 12));
+              ((p[1] & 0xf) << 4) |
+              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
 
 
    if (newMask == 0xffff || newMask == 0x0) {
@@ -495,49 +508,54 @@ static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
 /* =============================================================
  * Hardware clipping
  */
-static void i830Scissor(GLcontext *ctx, GLint x, GLint y, 
-			GLsizei w, GLsizei h)
+static void
+i830Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   intelScreenPrivate *screen = i830->intel.intelScreen;
+   struct i830_context *i830 = i830_context(ctx);
    int x1, y1, x2, y2;
 
-   if (!i830->intel.driDrawable)
+   if (!ctx->DrawBuffer)
       return;
 
-   x1 = x;
-   y1 = i830->intel.driDrawable->h - (y + h);
-   x2 = x + w - 1;
-   y2 = y1 + h - 1;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "[%s] x(%d) y(%d) w(%d) h(%d)\n", __FUNCTION__,
-	      x, y, w, h);
+   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
 
-   if (x1 < 0) x1 = 0;
-   if (y1 < 0) y1 = 0;
-   if (x2 < 0) x2 = 0;
-   if (y2 < 0) y2 = 0;
-
-   if (x2 >= screen->width) x2 = screen->width-1;
-   if (y2 >= screen->height) y2 = screen->height-1;
-   if (x1 >= screen->width) x1 = screen->width-1;
-   if (y1 >= screen->height) y1 = screen->height-1;
+   if (ctx->DrawBuffer->Name == 0) {
+      x1 = x;
+      y1 = ctx->DrawBuffer->Height - (y + h);
+      x2 = x + w - 1;
+      y2 = y1 + h - 1;
+      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   else {
+      /* FBO - not inverted
+       */
+      x1 = x;
+      y1 = y;
+      x2 = x + w - 1;
+      y2 = y + h - 1;
+      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
 
+   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
+   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
+   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
+   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
+   
+   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
 
    I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
    i830->state.Buffer[I830_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
    i830->state.Buffer[I830_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
 }
 
-static void i830LogicOp(GLcontext *ctx, GLenum opcode)
+static void
+i830LogicOp(GLcontext * ctx, GLenum opcode)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   int tmp = intel_translate_logic_op( opcode );
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i830_context *i830 = i830_context(ctx);
+   int tmp = intel_translate_logic_op(opcode);
 
+   DBG("%s\n", __FUNCTION__);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~LOGICOP_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
@@ -545,14 +563,14 @@ static void i830LogicOp(GLcontext *ctx, GLenum opcode)
 
 
 
-static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
+static void
+i830CullFaceFrontFace(GLcontext * ctx, GLenum unused)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    GLuint mode;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (!ctx->Polygon.CullFlag) {
       mode = CULLMODE_NONE;
    }
@@ -560,9 +578,9 @@ static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
       mode = CULLMODE_CW;
 
       if (ctx->Polygon.CullFaceMode == GL_FRONT)
-	 mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
+         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
       if (ctx->Polygon.FrontFace != GL_CCW)
-	 mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
+         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
    }
    else {
       mode = CULLMODE_BOTH;
@@ -573,18 +591,18 @@ static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    i830->state.Ctx[I830_CTXREG_STATE3] |= ENABLE_CULL_MODE | mode;
 }
 
-static void i830LineWidth( GLcontext *ctx, GLfloat widthf )
+static void
+i830LineWidth(GLcontext * ctx, GLfloat widthf)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
    int width;
    int state5;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   width = (int)(widthf * 2);
-   CLAMP_SELF(width, 1, 15);
+   DBG("%s\n", __FUNCTION__);
    
+   width = (int) (widthf * 2);
+   CLAMP_SELF(width, 1, 15);
+
    state5 = i830->state.Ctx[I830_CTXREG_STATE5] & ~FIXED_LINE_WIDTH_MASK;
    state5 |= (ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(width));
 
@@ -594,19 +612,19 @@ static void i830LineWidth( GLcontext *ctx, GLfloat widthf )
    }
 }
 
-static void i830PointSize(GLcontext *ctx, GLfloat size)
+static void
+i830PointSize(GLcontext * ctx, GLfloat size)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLint point_size = (int)size;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i830_context *i830 = i830_context(ctx);
+   GLint point_size = (int) size;
 
+   DBG("%s\n", __FUNCTION__);
+   
    CLAMP_SELF(point_size, 1, 256);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE5] &= ~FIXED_POINT_WIDTH_MASK;
    i830->state.Ctx[I830_CTXREG_STATE5] |= (ENABLE_FIXED_POINT_WIDTH |
-				       FIXED_POINT_WIDTH(point_size));
+                                           FIXED_POINT_WIDTH(point_size));
 }
 
 
@@ -614,23 +632,21 @@ static void i830PointSize(GLcontext *ctx, GLfloat size)
  * Color masks
  */
 
-static void i830ColorMask(GLcontext *ctx,
-			  GLboolean r, GLboolean g,
-			  GLboolean b, GLboolean a)
+static void
+i830ColorMask(GLcontext * ctx,
+              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
    GLuint tmp = 0;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
+   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
 
    tmp = ((i830->state.Ctx[I830_CTXREG_ENABLES_2] & ~WRITEMASK_MASK) |
-	  ENABLE_COLOR_MASK |
-	  ENABLE_COLOR_WRITE |
-	  ((!r) << WRITEMASK_RED_SHIFT) |
-	  ((!g) << WRITEMASK_GREEN_SHIFT) |
-	  ((!b) << WRITEMASK_BLUE_SHIFT) |
-	  ((!a) << WRITEMASK_ALPHA_SHIFT));
+          ENABLE_COLOR_MASK |
+          ENABLE_COLOR_WRITE |
+          ((!r) << WRITEMASK_RED_SHIFT) |
+          ((!g) << WRITEMASK_GREEN_SHIFT) |
+          ((!b) << WRITEMASK_BLUE_SHIFT) | ((!a) << WRITEMASK_ALPHA_SHIFT));
 
    if (tmp != i830->state.Ctx[I830_CTXREG_ENABLES_2]) {
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
@@ -638,9 +654,10 @@ static void i830ColorMask(GLcontext *ctx,
    }
 }
 
-static void update_specular( GLcontext *ctx )
+static void
+update_specular(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_SPEC_ADD_MASK;
@@ -651,22 +668,22 @@ static void update_specular( GLcontext *ctx )
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_SPEC_ADD;
 }
 
-static void i830LightModelfv(GLcontext *ctx, GLenum pname, 
-			     const GLfloat *param)
+static void
+i830LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular( ctx );
+      update_specular(ctx);
    }
 }
 
 /* In Mesa 3.5 we can reliably do native flatshading.
  */
-static void i830ShadeModel(GLcontext *ctx, GLenum mode)
+static void
+i830ShadeModel(GLcontext * ctx, GLenum mode)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
 
@@ -675,58 +692,62 @@ static void i830ShadeModel(GLcontext *ctx, GLenum mode)
    i830->state.Ctx[I830_CTXREG_STATE3] &= ~SHADE_MODE_MASK;
 
    if (mode == GL_FLAT) {
-     i830->state.Ctx[I830_CTXREG_STATE3] |= (ALPHA_SHADE_MODE(SHADE_MODE_FLAT) |
-					  FOG_SHADE_MODE(SHADE_MODE_FLAT) |
-					  SPEC_SHADE_MODE(SHADE_MODE_FLAT) |
-					  COLOR_SHADE_MODE(SHADE_MODE_FLAT));
-   } else {
-     i830->state.Ctx[I830_CTXREG_STATE3] |= (ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  COLOR_SHADE_MODE(SHADE_MODE_LINEAR));
+      i830->state.Ctx[I830_CTXREG_STATE3] |=
+         (ALPHA_SHADE_MODE(SHADE_MODE_FLAT) | FOG_SHADE_MODE(SHADE_MODE_FLAT)
+          | SPEC_SHADE_MODE(SHADE_MODE_FLAT) |
+          COLOR_SHADE_MODE(SHADE_MODE_FLAT));
+   }
+   else {
+      i830->state.Ctx[I830_CTXREG_STATE3] |=
+         (ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
+          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
+          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
+          COLOR_SHADE_MODE(SHADE_MODE_LINEAR));
    }
 }
 
 /* =============================================================
  * Fog
  */
-static void i830Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
+static void
+i830Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i830_context *i830 = i830_context(ctx);
 
-   if (pname == GL_FOG_COLOR) {      
-      GLuint color = (((GLubyte)(ctx->Fog.Color[0]*255.0F) << 16) |
-		      ((GLubyte)(ctx->Fog.Color[1]*255.0F) << 8) |
-		      ((GLubyte)(ctx->Fog.Color[2]*255.0F) << 0));
+   DBG("%s\n", __FUNCTION__);
+   
+   if (pname == GL_FOG_COLOR) {
+      GLuint color = (((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
+                      ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
+                      ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
 
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_FOGCOLOR] = (_3DSTATE_FOG_COLOR_CMD | color);
+      i830->state.Ctx[I830_CTXREG_FOGCOLOR] =
+         (_3DSTATE_FOG_COLOR_CMD | color);
    }
 }
 
 /* =============================================================
  */
 
-static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+static void
+i830Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
 
-   switch(cap) {
+   switch (cap) {
    case GL_LIGHTING:
    case GL_COLOR_SUM:
-      update_specular( ctx );
+      update_specular(ctx);
       break;
 
    case GL_ALPHA_TEST:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_ALPHA_TEST_MASK;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_ALPHA_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_ALPHA_TEST;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_ALPHA_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_ALPHA_TEST;
 
       break;
 
@@ -739,18 +760,18 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 
       /* Logicop doesn't seem to work at 16bpp:
        */
-      if (i830->intel.intelScreen->cpp == 2)
-	 FALLBACK( &i830->intel, I830_FALLBACK_LOGICOP, state );
+      if (i830->intel.ctx.Visual.rgbBits == 16)
+         FALLBACK(&i830->intel, I830_FALLBACK_LOGICOP, state);
       break;
- 
+
    case GL_DITHER:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DITHER;
 
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DITHER;
+         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DITHER;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DITHER;
+         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DITHER;
       break;
 
    case GL_DEPTH_TEST:
@@ -758,46 +779,44 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
 
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
 
       /* Also turn off depth writes when GL_DEPTH_TEST is disabled:
        */
-      i830DepthMask( ctx, ctx->Depth.Mask );
+      i830DepthMask(ctx, ctx->Depth.Mask);
       break;
 
    case GL_SCISSOR_TEST:
       I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
-      
+
       if (state)
-	 i830->state.Buffer[I830_DESTREG_SENABLE] = 
-	    (_3DSTATE_SCISSOR_ENABLE_CMD |
-	     ENABLE_SCISSOR_RECT);
+         i830->state.Buffer[I830_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
       else
-	 i830->state.Buffer[I830_DESTREG_SENABLE] = 
-	    (_3DSTATE_SCISSOR_ENABLE_CMD |
-	     DISABLE_SCISSOR_RECT);
+         i830->state.Buffer[I830_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
 
       break;
 
    case GL_LINE_SMOOTH:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      
+
       i830->state.Ctx[I830_CTXREG_AA] &= ~AA_LINE_ENABLE;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_ENABLE;
+         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_ENABLE;
       else
-	 i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_DISABLE;
+         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_DISABLE;
       break;
 
    case GL_FOG:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_FOG_MASK;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_FOG;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_FOG;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_FOG;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_FOG;
       break;
 
    case GL_CULL_FACE:
@@ -808,20 +827,32 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       break;
 
    case GL_STENCIL_TEST:
-      if (i830->intel.hw_stencil) {
-	 I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-	 if (state) {
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-	 } else {
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_STENCIL_WRITE;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_STENCIL_WRITE;
-	 }
-      } else {
-	 FALLBACK( &i830->intel, I830_FALLBACK_STENCIL, state );
+      {
+         GLboolean hw_stencil = GL_FALSE;
+         if (ctx->DrawBuffer) {
+            struct intel_renderbuffer *irbStencil
+               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
+            hw_stencil = (irbStencil && irbStencil->region);
+         }
+         if (hw_stencil) {
+            I830_STATECHANGE(i830, I830_UPLOAD_CTX);
+
+            if (state) {
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
+            }
+            else {
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] &=
+                  ~ENABLE_STENCIL_WRITE;
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] |=
+                  DISABLE_STENCIL_WRITE;
+            }
+         }
+         else {
+            FALLBACK(&i830->intel, I830_FALLBACK_STENCIL, state);
+         }
       }
       break;
 
@@ -830,13 +861,12 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
        * I'll do more testing later to find out exactly which hardware
        * supports it.  Disabled for now.
        */
-      if (i830->intel.hw_stipple && 
-	  i830->intel.reduced_primitive == GL_TRIANGLES)
-      {
-	 I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-	 i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
-	 if (state)
-	    i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
+      if (i830->intel.hw_stipple &&
+          i830->intel.reduced_primitive == GL_TRIANGLES) {
+         I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
+         i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
+         if (state)
+            i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
       }
       break;
 
@@ -846,206 +876,172 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 }
 
 
-static void i830_init_packets( i830ContextPtr i830 )
+static void
+i830_init_packets(struct i830_context *i830)
 {
-   intelScreenPrivate *screen = i830->intel.intelScreen;
-
    /* Zero all state */
    memset(&i830->state, 0, sizeof(i830->state));
 
    /* Set default blend state */
    i830->state.TexBlend[0][0] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-				  TEXPIPE_COLOR |
-				  ENABLE_TEXOUTPUT_WRT_SEL |
-				  TEXOP_OUTPUT_CURRENT |
-				  DISABLE_TEX_CNTRL_STAGE |
-				  TEXOP_SCALE_1X |
-				  TEXOP_MODIFY_PARMS |
-				  TEXOP_LAST_STAGE |
-				  TEXBLENDOP_ARG1);
+                                 TEXPIPE_COLOR |
+                                 ENABLE_TEXOUTPUT_WRT_SEL |
+                                 TEXOP_OUTPUT_CURRENT |
+                                 DISABLE_TEX_CNTRL_STAGE |
+                                 TEXOP_SCALE_1X |
+                                 TEXOP_MODIFY_PARMS |
+                                 TEXOP_LAST_STAGE | TEXBLENDOP_ARG1);
    i830->state.TexBlend[0][1] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-				  TEXPIPE_ALPHA |
-				  ENABLE_TEXOUTPUT_WRT_SEL |
-				  TEXOP_OUTPUT_CURRENT |
-				  TEXOP_SCALE_1X |
-				  TEXOP_MODIFY_PARMS |
-				  TEXBLENDOP_ARG1);
+                                 TEXPIPE_ALPHA |
+                                 ENABLE_TEXOUTPUT_WRT_SEL |
+                                 TEXOP_OUTPUT_CURRENT |
+                                 TEXOP_SCALE_1X |
+                                 TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    i830->state.TexBlend[0][2] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-				  TEXPIPE_COLOR |
-				  TEXBLEND_ARG1 |
-				  TEXBLENDARG_MODIFY_PARMS |
-				  TEXBLENDARG_DIFFUSE);
+                                 TEXPIPE_COLOR |
+                                 TEXBLEND_ARG1 |
+                                 TEXBLENDARG_MODIFY_PARMS |
+                                 TEXBLENDARG_DIFFUSE);
    i830->state.TexBlend[0][3] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-				  TEXPIPE_ALPHA |
-				  TEXBLEND_ARG1 |
-				  TEXBLENDARG_MODIFY_PARMS |
-				  TEXBLENDARG_DIFFUSE);
+                                 TEXPIPE_ALPHA |
+                                 TEXBLEND_ARG1 |
+                                 TEXBLENDARG_MODIFY_PARMS |
+                                 TEXBLENDARG_DIFFUSE);
 
    i830->state.TexBlendWordsUsed[0] = 4;
 
 
-   i830->state.Ctx[I830_CTXREG_VF] =  0;
+   i830->state.Ctx[I830_CTXREG_VF] = 0;
    i830->state.Ctx[I830_CTXREG_VF2] = 0;
 
    i830->state.Ctx[I830_CTXREG_AA] = (_3DSTATE_AA_CMD |
-				      AA_LINE_ECAAR_WIDTH_ENABLE |
-				      AA_LINE_ECAAR_WIDTH_1_0 |
-				      AA_LINE_REGION_WIDTH_ENABLE |
-				      AA_LINE_REGION_WIDTH_1_0 | 
-				      AA_LINE_DISABLE);
+                                      AA_LINE_ECAAR_WIDTH_ENABLE |
+                                      AA_LINE_ECAAR_WIDTH_1_0 |
+                                      AA_LINE_REGION_WIDTH_ENABLE |
+                                      AA_LINE_REGION_WIDTH_1_0 |
+                                      AA_LINE_DISABLE);
 
    i830->state.Ctx[I830_CTXREG_ENABLES_1] = (_3DSTATE_ENABLES_1_CMD |
-					     DISABLE_LOGIC_OP |
-					     DISABLE_STENCIL_TEST |
-					     DISABLE_DEPTH_BIAS |
-					     DISABLE_SPEC_ADD |
-					     DISABLE_FOG |
-					     DISABLE_ALPHA_TEST |
-					     DISABLE_COLOR_BLEND |
-					     DISABLE_DEPTH_TEST);
-
+                                             DISABLE_LOGIC_OP |
+                                             DISABLE_STENCIL_TEST |
+                                             DISABLE_DEPTH_BIAS |
+                                             DISABLE_SPEC_ADD |
+                                             DISABLE_FOG |
+                                             DISABLE_ALPHA_TEST |
+                                             DISABLE_COLOR_BLEND |
+                                             DISABLE_DEPTH_TEST);
+
+#if 000                         /* XXX all the stencil enable state is set in i830Enable(), right? */
    if (i830->intel.hw_stencil) {
       i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-						ENABLE_STENCIL_WRITE |
-						ENABLE_TEX_CACHE |
-						ENABLE_DITHER |
-						ENABLE_COLOR_MASK |
-						/* set no color comps disabled */
-						ENABLE_COLOR_WRITE |
-						ENABLE_DEPTH_WRITE);
-   } else {
+                                                ENABLE_STENCIL_WRITE |
+                                                ENABLE_TEX_CACHE |
+                                                ENABLE_DITHER |
+                                                ENABLE_COLOR_MASK |
+                                                /* set no color comps disabled */
+                                                ENABLE_COLOR_WRITE |
+                                                ENABLE_DEPTH_WRITE);
+   }
+   else
+#endif
+   {
       i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-						DISABLE_STENCIL_WRITE |
-						ENABLE_TEX_CACHE |
-						ENABLE_DITHER |
-						ENABLE_COLOR_MASK |
-						/* set no color comps disabled */
-						ENABLE_COLOR_WRITE |
-						ENABLE_DEPTH_WRITE);
+                                                DISABLE_STENCIL_WRITE |
+                                                ENABLE_TEX_CACHE |
+                                                ENABLE_DITHER |
+                                                ENABLE_COLOR_MASK |
+                                                /* set no color comps disabled */
+                                                ENABLE_COLOR_WRITE |
+                                                ENABLE_DEPTH_WRITE);
    }
 
    i830->state.Ctx[I830_CTXREG_STATE1] = (_3DSTATE_MODES_1_CMD |
-					  ENABLE_COLR_BLND_FUNC |
-					  BLENDFUNC_ADD |
-					  ENABLE_SRC_BLND_FACTOR |
-					  SRC_BLND_FACT(BLENDFACT_ONE) | 
-					  ENABLE_DST_BLND_FACTOR |
-					  DST_BLND_FACT(BLENDFACT_ZERO) );
+                                          ENABLE_COLR_BLND_FUNC |
+                                          BLENDFUNC_ADD |
+                                          ENABLE_SRC_BLND_FACTOR |
+                                          SRC_BLND_FACT(BLENDFACT_ONE) |
+                                          ENABLE_DST_BLND_FACTOR |
+                                          DST_BLND_FACT(BLENDFACT_ZERO));
 
    i830->state.Ctx[I830_CTXREG_STATE2] = (_3DSTATE_MODES_2_CMD |
-					  ENABLE_GLOBAL_DEPTH_BIAS | 
-					  GLOBAL_DEPTH_BIAS(0) |
-					  ENABLE_ALPHA_TEST_FUNC | 
-					  ALPHA_TEST_FUNC(COMPAREFUNC_ALWAYS) |
-					  ALPHA_REF_VALUE(0) );
+                                          ENABLE_GLOBAL_DEPTH_BIAS |
+                                          GLOBAL_DEPTH_BIAS(0) |
+                                          ENABLE_ALPHA_TEST_FUNC |
+                                          ALPHA_TEST_FUNC(COMPAREFUNC_ALWAYS)
+                                          | ALPHA_REF_VALUE(0));
 
    i830->state.Ctx[I830_CTXREG_STATE3] = (_3DSTATE_MODES_3_CMD |
-					  ENABLE_DEPTH_TEST_FUNC |
-					  DEPTH_TEST_FUNC(COMPAREFUNC_LESS) |
-					  ENABLE_ALPHA_SHADE_MODE |
-					  ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_FOG_SHADE_MODE |
-					  FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_SPEC_SHADE_MODE |
-					  SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_COLOR_SHADE_MODE |
-					  COLOR_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_CULL_MODE |
-					  CULLMODE_NONE);
+                                          ENABLE_DEPTH_TEST_FUNC |
+                                          DEPTH_TEST_FUNC(COMPAREFUNC_LESS) |
+                                          ENABLE_ALPHA_SHADE_MODE |
+                                          ALPHA_SHADE_MODE(SHADE_MODE_LINEAR)
+                                          | ENABLE_FOG_SHADE_MODE |
+                                          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
+                                          ENABLE_SPEC_SHADE_MODE |
+                                          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
+                                          ENABLE_COLOR_SHADE_MODE |
+                                          COLOR_SHADE_MODE(SHADE_MODE_LINEAR)
+                                          | ENABLE_CULL_MODE | CULLMODE_NONE);
 
    i830->state.Ctx[I830_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-					  ENABLE_LOGIC_OP_FUNC |
-					  LOGIC_OP_FUNC(LOGICOP_COPY) |
-					  ENABLE_STENCIL_TEST_MASK |
-					  STENCIL_TEST_MASK(0xff) |
-					  ENABLE_STENCIL_WRITE_MASK |
-					  STENCIL_WRITE_MASK(0xff));
+                                          ENABLE_LOGIC_OP_FUNC |
+                                          LOGIC_OP_FUNC(LOGICOP_COPY) |
+                                          ENABLE_STENCIL_TEST_MASK |
+                                          STENCIL_TEST_MASK(0xff) |
+                                          ENABLE_STENCIL_WRITE_MASK |
+                                          STENCIL_WRITE_MASK(0xff));
 
    i830->state.Ctx[I830_CTXREG_STENCILTST] = (_3DSTATE_STENCIL_TEST_CMD |
-					      ENABLE_STENCIL_PARMS |
-					      STENCIL_FAIL_OP(STENCILOP_KEEP) |
-					      STENCIL_PASS_DEPTH_FAIL_OP(STENCILOP_KEEP) |
-					      STENCIL_PASS_DEPTH_PASS_OP(STENCILOP_KEEP) |
-					      ENABLE_STENCIL_TEST_FUNC |
-					      STENCIL_TEST_FUNC(COMPAREFUNC_ALWAYS) |
-					      ENABLE_STENCIL_REF_VALUE |
-					      STENCIL_REF_VALUE(0) );
-
-   i830->state.Ctx[I830_CTXREG_STATE5] = (_3DSTATE_MODES_5_CMD |
-					  FLUSH_TEXTURE_CACHE |
-					  ENABLE_SPRITE_POINT_TEX |
-					  SPRITE_POINT_TEX_OFF |
-					  ENABLE_FIXED_LINE_WIDTH |
-					  FIXED_LINE_WIDTH(0x2) | /* 1.0 */
-					  ENABLE_FIXED_POINT_WIDTH |
-					  FIXED_POINT_WIDTH(1) );
+                                              ENABLE_STENCIL_PARMS |
+                                              STENCIL_FAIL_OP(STENCILOP_KEEP)
+                                              |
+                                              STENCIL_PASS_DEPTH_FAIL_OP
+                                              (STENCILOP_KEEP) |
+                                              STENCIL_PASS_DEPTH_PASS_OP
+                                              (STENCILOP_KEEP) |
+                                              ENABLE_STENCIL_TEST_FUNC |
+                                              STENCIL_TEST_FUNC
+                                              (COMPAREFUNC_ALWAYS) |
+                                              ENABLE_STENCIL_REF_VALUE |
+                                              STENCIL_REF_VALUE(0));
+
+   i830->state.Ctx[I830_CTXREG_STATE5] = (_3DSTATE_MODES_5_CMD | FLUSH_TEXTURE_CACHE | ENABLE_SPRITE_POINT_TEX | SPRITE_POINT_TEX_OFF | ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(0x2) |       /* 1.0 */
+                                          ENABLE_FIXED_POINT_WIDTH |
+                                          FIXED_POINT_WIDTH(1));
 
    i830->state.Ctx[I830_CTXREG_IALPHAB] = (_3DSTATE_INDPT_ALPHA_BLEND_CMD |
-					   DISABLE_INDPT_ALPHA_BLEND |
-					   ENABLE_ALPHA_BLENDFUNC |
-					   ABLENDFUNC_ADD);
+                                           DISABLE_INDPT_ALPHA_BLEND |
+                                           ENABLE_ALPHA_BLENDFUNC |
+                                           ABLENDFUNC_ADD);
 
    i830->state.Ctx[I830_CTXREG_FOGCOLOR] = (_3DSTATE_FOG_COLOR_CMD |
-					    FOG_COLOR_RED(0) |
-					    FOG_COLOR_GREEN(0) |
-					    FOG_COLOR_BLUE(0));
+                                            FOG_COLOR_RED(0) |
+                                            FOG_COLOR_GREEN(0) |
+                                            FOG_COLOR_BLUE(0));
 
    i830->state.Ctx[I830_CTXREG_BLENDCOLOR0] = _3DSTATE_CONST_BLEND_COLOR_CMD;
    i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] = 0;
 
    i830->state.Ctx[I830_CTXREG_MCSB0] = _3DSTATE_MAP_COORD_SETBIND_CMD;
    i830->state.Ctx[I830_CTXREG_MCSB1] = (TEXBIND_SET3(TEXCOORDSRC_VTXSET_3) |
-					 TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |
-					 TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |
-					 TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));
-					 
-
-   i830->state.Stipple[I830_STPREG_ST0] = _3DSTATE_STIPPLE;
-
-   i830->state.Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] = 
-      (BUF_3D_ID_COLOR_BACK | 
-       BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
+                                         TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |
+                                         TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |
+                                         TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));
 
 
-   i830->state.Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] = 
-      (BUF_3D_ID_DEPTH |
-       BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_DBUFADDR2] = screen->depth.offset;
-
+   i830->state.Stipple[I830_STPREG_ST0] = _3DSTATE_STIPPLE;
 
    i830->state.Buffer[I830_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
-
-   switch (screen->fbFormat) {
-   case DV_PF_555:
-   case DV_PF_565:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					      DSTORG_VERT_BIAS(0x8) | /* .5 */
-					      screen->fbFormat |
-					      DEPTH_IS_Z |
-					      DEPTH_FRMT_16_FIXED);
-      break;
-   case DV_PF_8888:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					      DSTORG_VERT_BIAS(0x8) | /* .5 */
-					      screen->fbFormat |
-					      DEPTH_IS_Z |
-					      DEPTH_FRMT_24_FIXED_8_OTHER);
-      break;
-   }
-
    i830->state.Buffer[I830_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-					       DISABLE_SCISSOR_RECT);
+                                               DISABLE_SCISSOR_RECT);
    i830->state.Buffer[I830_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
    i830->state.Buffer[I830_DESTREG_SR1] = 0;
    i830->state.Buffer[I830_DESTREG_SR2] = 0;
 }
 
 
-void i830InitStateFuncs( struct dd_function_table *functions )
+void
+i830InitStateFuncs(struct dd_function_table *functions)
 {
    functions->AlphaFunc = i830AlphaFunc;
    functions->BlendColor = i830BlendColor;
@@ -1070,20 +1066,21 @@ void i830InitStateFuncs( struct dd_function_table *functions )
    functions->StencilOpSeparate = i830StencilOpSeparate;
 }
 
-void i830InitState( i830ContextPtr i830 )
+void
+i830InitState(struct i830_context *i830)
 {
    GLcontext *ctx = &i830->intel.ctx;
 
-   i830_init_packets( i830 );
+   i830_init_packets(i830);
 
    _mesa_init_driver_state(ctx);
 
-   memcpy( &i830->initial, &i830->state, sizeof(i830->state) );
+   memcpy(&i830->initial, &i830->state, sizeof(i830->state));
 
    i830->current = &i830->state;
    i830->state.emitted = 0;
-   i830->state.active = (I830_UPLOAD_TEXBLEND(0) |
-			 I830_UPLOAD_STIPPLE |
-			 I830_UPLOAD_CTX |
-			 I830_UPLOAD_BUFFERS);
+   i830->state.active = (I830_UPLOAD_INVARIENT |
+                         I830_UPLOAD_TEXBLEND(0) |
+                         I830_UPLOAD_STIPPLE |
+                         I830_UPLOAD_CTX | I830_UPLOAD_BUFFERS);
 }
diff --git a/src/mesa/drivers/dri/i915/i830_tex.c b/src/mesa/drivers/dri/i915/i830_tex.c
index 3c4aedb35cb..34ac42a78e0 100644
--- a/src/mesa/drivers/dri/i915/i830_tex.c
+++ b/src/mesa/drivers/dri/i915/i830_tex.c
@@ -25,281 +25,31 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "texmem.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/mm.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
-
-#include "intel_ioctl.h"
+#include "texmem.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
 
 
 
-
-/**
- * Set the texture wrap modes.
- * 
- * The i830M (and related graphics cores) do not support GL_CLAMP.  The Intel
- * drivers for "other operating systems" implement GL_CLAMP as
- * GL_CLAMP_TO_EDGE, so the same is done here.
- * 
- * \param t Texture object whose wrap modes are to be set
- * \param swrap Wrap mode for the \a s texture coordinate
- * \param twrap Wrap mode for the \a t texture coordinate
- */
-static void i830SetTexWrapping(i830TextureObjectPtr tex,
-			       GLenum swrap, 
-			       GLenum twrap)
-{
-   tex->Setup[I830_TEXREG_MCS] &= ~(TEXCOORD_ADDR_U_MASK|TEXCOORD_ADDR_V_MASK);
-
-   switch( swrap ) {
-   case GL_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP);
-      break;
-   case GL_CLAMP:
-   case GL_CLAMP_TO_EDGE:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_CLAMP);
-      break;
-   case GL_CLAMP_TO_BORDER:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_CLAMP_BORDER);
-      break;
-   case GL_MIRRORED_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_MIRROR);
-      break;
-   default:
-      break;
-   }
-
-   switch( twrap ) {
-   case GL_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP);
-      break;
-   case GL_CLAMP:
-   case GL_CLAMP_TO_EDGE:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_CLAMP);
-      break;
-   case GL_CLAMP_TO_BORDER:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_CLAMP_BORDER);
-      break;
-   case GL_MIRRORED_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |=
-			TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_MIRROR);
-      break;
-   default:
-      break;
-   }
-}
-
-
-/**
- * Set the texture magnification and minification modes.
- * 
- * \param t Texture whose filter modes are to be set
- * \param minf Texture minification mode
- * \param magf Texture magnification mode
- * \param bias LOD bias for this texture unit.
- */
-
-static void i830SetTexFilter( i830TextureObjectPtr t, GLenum minf, GLenum magf,
-			      GLfloat maxanisotropy )
-{
-   int minFilt = 0, mipFilt = 0, magFilt = 0;
-
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if ( maxanisotropy > 1.0 ) {
-      minFilt = FILTER_ANISOTROPIC;
-      magFilt = FILTER_ANISOTROPIC;
-   }
-   else {
-      switch (minf) {
-      case GL_NEAREST:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_NONE;
-	 break;
-      case GL_LINEAR:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_NONE;
-	 break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_NEAREST;
-	 break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_NEAREST;
-	 break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_LINEAR;
-	 break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }
-
-      switch (magf) {
-      case GL_NEAREST:
-	 magFilt = FILTER_NEAREST;
-	 break;
-      case GL_LINEAR:
-	 magFilt = FILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }  
-   }
-
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIP_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAG_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] |= ((minFilt << TM0S3_MIN_FILTER_SHIFT) |
-				   (mipFilt << TM0S3_MIP_FILTER_SHIFT) |
-				   (magFilt << TM0S3_MAG_FILTER_SHIFT));
-}
-
-static void i830SetTexBorderColor(i830TextureObjectPtr t, GLubyte color[4])
+static void
+i830TexEnv(GLcontext * ctx, GLenum target,
+           GLenum pname, const GLfloat * param)
 {
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-    t->Setup[I830_TEXREG_TM0S4] = 
-        INTEL_PACKCOLOR8888(color[0],color[1],color[2],color[3]);
-}
-
-
-/**
- * Allocate space for and load the mesa images into the texture memory block.
- * This will happen before drawing with a new texture, or drawing with a
- * texture after it was swapped out or teximaged again.
- */
-
-intelTextureObjectPtr i830AllocTexObj( struct gl_texture_object *texObj )
-{
-   i830TextureObjectPtr t = CALLOC_STRUCT( i830_texture_object );
-   if ( !t ) 
-      return NULL;
-
-   texObj->DriverData = t;
-   t->intel.base.tObj = texObj;
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-   make_empty_list( &t->intel.base );
-
-   t->Setup[I830_TEXREG_TM0LI] = 0; /* not used */
-   t->Setup[I830_TEXREG_TM0S0] = 0;
-   t->Setup[I830_TEXREG_TM0S1] = 0;
-   t->Setup[I830_TEXREG_TM0S2] = 0;
-   t->Setup[I830_TEXREG_TM0S3] = 0;
-   t->Setup[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-				MAP_UNIT(0) |
-				ENABLE_TEXCOORD_PARAMS |
-				TEXCOORDS_ARE_NORMAL |
-				TEXCOORDTYPE_CARTESIAN |
-				ENABLE_ADDR_V_CNTL |
-				TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
-				ENABLE_ADDR_U_CNTL |
-				TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
-
-   
-   i830SetTexWrapping( t, texObj->WrapS, texObj->WrapT );
-   i830SetTexFilter( t, texObj->MinFilter, texObj->MagFilter, 
-		     texObj->MaxAnisotropy );
-   i830SetTexBorderColor( t, texObj->_BorderChan );
-
-   return &t->intel;
-}
-
-
-static void i830TexParameter( GLcontext *ctx, GLenum target,
-			      struct gl_texture_object *tObj,
-			      GLenum pname, const GLfloat *params )
-{
-   i830TextureObjectPtr t = (i830TextureObjectPtr) tObj->DriverData;
-   if (!t)
-      return;
 
    switch (pname) {
-   case GL_TEXTURE_MIN_FILTER:
-   case GL_TEXTURE_MAG_FILTER:
-   case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-      i830SetTexFilter( t, tObj->MinFilter, tObj->MagFilter,
-			tObj->MaxAnisotropy);
-      break;
-
-   case GL_TEXTURE_WRAP_S:
-   case GL_TEXTURE_WRAP_T:
-      i830SetTexWrapping( t, tObj->WrapS, tObj->WrapT );
-      break;
-  
-   case GL_TEXTURE_BORDER_COLOR:
-      i830SetTexBorderColor( t, tObj->_BorderChan );
-      break;
-
-   case GL_TEXTURE_BASE_LEVEL:
-   case GL_TEXTURE_MAX_LEVEL:
-   case GL_TEXTURE_MIN_LOD:
-   case GL_TEXTURE_MAX_LOD:
-      /* The i830 and its successors can do a lot of this without
-       * reloading the textures.  A project for someone?
-       */
-      intelFlush( ctx );
-      driSwapOutTextureObject( (driTextureObject *) t );
-      break;
-
-   default:
-      return;
-   }
-
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-}
-
-
-static void i830TexEnv( GLcontext *ctx, GLenum target, 
-			GLenum pname, const GLfloat *param )
-{
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
-   GLuint unit = ctx->Texture.CurrentUnit;
-
-   switch (pname) {
-   case GL_TEXTURE_ENV_COLOR: 
-#if 0
-   {
-      GLubyte r, g, b, a;
-      GLuint col;
-      
-      UNCLAMPED_FLOAT_TO_UBYTE(r, param[RCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(g, param[GCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(b, param[BCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(a, param[ACOMP]);
-
-      col = ((a << 24) | (r << 16) | (g << 8) | b);
-
-      if (col != i830->state.TexEnv[unit][I830_TEXENVREG_COL1]) {
-	 I830_STATECHANGE(i830, I830_UPLOAD_TEXENV);
-	 i830->state.TexEnv[unit][I830_TEXENVREG_COL1] = col;
-      }
-
-      break;
-   }
-#endif
+   case GL_TEXTURE_ENV_COLOR:
    case GL_TEXTURE_ENV_MODE:
    case GL_COMBINE_RGB:
    case GL_COMBINE_ALPHA:
@@ -319,38 +69,32 @@ static void i830TexEnv( GLcontext *ctx, GLenum target,
    case GL_ALPHA_SCALE:
       break;
 
-   case GL_TEXTURE_LOD_BIAS: {
-      int b = (int) ((*param) * 16.0);
-      if (b > 63) b = 63;
-      if (b < -64) b = -64;
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_TM0S3] &= ~TM0S3_LOD_BIAS_MASK;
-      i830->state.Tex[unit][I830_TEXREG_TM0S3] |= 
-	 ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK);
-      break;
-   }
+   case GL_TEXTURE_LOD_BIAS:{
+         struct i830_context *i830 = i830_context(ctx);
+         GLuint unit = ctx->Texture.CurrentUnit;
+         int b = (int) ((*param) * 16.0);
+         if (b > 63)
+            b = 63;
+         if (b < -64)
+            b = -64;
+         I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
+         i830->lodbias_tm0s3[unit] =
+            ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK);
+         break;
+      }
 
    default:
       break;
    }
 }
 
-static void i830BindTexture( GLcontext *ctx, GLenum target,
-			    struct gl_texture_object *texObj )
-{
-   i830TextureObjectPtr tex;
-   
-   if (!texObj->DriverData)
-      i830AllocTexObj( texObj );
-   
-   tex = (i830TextureObjectPtr)texObj->DriverData;
-}
 
 
 
-void i830InitTextureFuncs( struct dd_function_table *functions )
+void
+i830InitTextureFuncs(struct dd_function_table *functions)
 {
-   functions->BindTexture 		= i830BindTexture;
-   functions->TexEnv                    = i830TexEnv;
-   functions->TexParameter              = i830TexParameter;
+/*
+   functions->TexEnv = i830TexEnv;
+*/
 }
diff --git a/src/mesa/drivers/dri/i915/i830_texblend.c b/src/mesa/drivers/dri/i915/i830_texblend.c
index 49e0347643c..09f7f37e765 100644
--- a/src/mesa/drivers/dri/i915/i830_texblend.c
+++ b/src/mesa/drivers/dri/i915/i830_texblend.c
@@ -25,18 +25,16 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texformat.h"
-#include "texstore.h"
-
-#include "mm.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/mm.h"
 
 #include "intel_screen.h"
-#include "intel_ioctl.h"
 #include "intel_tex.h"
 
 #include "i830_context.h"
@@ -46,46 +44,42 @@
 /* ================================================================
  * Texture combine functions
  */
-static GLuint pass_through( GLuint *state, GLuint blendUnit )
+static GLuint
+pass_through(GLuint * state, GLuint blendUnit)
 {
    state[0] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-	       TEXPIPE_COLOR |
-	       ENABLE_TEXOUTPUT_WRT_SEL |
-	       TEXOP_OUTPUT_CURRENT |
-	       DISABLE_TEX_CNTRL_STAGE |
-	       TEXOP_SCALE_1X |
-	       TEXOP_MODIFY_PARMS |
-	       TEXBLENDOP_ARG1);
+               TEXPIPE_COLOR |
+               ENABLE_TEXOUTPUT_WRT_SEL |
+               TEXOP_OUTPUT_CURRENT |
+               DISABLE_TEX_CNTRL_STAGE |
+               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    state[1] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-	       TEXPIPE_ALPHA |
-	       ENABLE_TEXOUTPUT_WRT_SEL |
-	       TEXOP_OUTPUT_CURRENT |
-	       TEXOP_SCALE_1X |
-	       TEXOP_MODIFY_PARMS |
-	       TEXBLENDOP_ARG1);
+               TEXPIPE_ALPHA |
+               ENABLE_TEXOUTPUT_WRT_SEL |
+               TEXOP_OUTPUT_CURRENT |
+               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    state[2] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-	       TEXPIPE_COLOR |
-	       TEXBLEND_ARG1 |
-	       TEXBLENDARG_MODIFY_PARMS |
-	       TEXBLENDARG_CURRENT);
+               TEXPIPE_COLOR |
+               TEXBLEND_ARG1 |
+               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
    state[3] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-	       TEXPIPE_ALPHA |
-	       TEXBLEND_ARG1 |
-	       TEXBLENDARG_MODIFY_PARMS |
-	       TEXBLENDARG_CURRENT);
+               TEXPIPE_ALPHA |
+               TEXBLEND_ARG1 |
+               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
 
    return 4;
 }
 
-static GLuint emit_factor( GLuint blendUnit, GLuint *state, GLuint count, 
-			   const GLfloat *factor )
+static GLuint
+emit_factor(GLuint blendUnit, GLuint * state, GLuint count,
+            const GLfloat * factor)
 {
    GLubyte r, g, b, a;
    GLuint col;
-      
+
    if (0)
       fprintf(stderr, "emit constant %d: %.2f %.2f %.2f %.2f\n",
-	  blendUnit, factor[0], factor[1], factor[2], factor[3]);
+              blendUnit, factor[0], factor[1], factor[2], factor[3]);
 
    UNCLAMPED_FLOAT_TO_UBYTE(r, factor[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, factor[1]);
@@ -94,21 +88,27 @@ static GLuint emit_factor( GLuint blendUnit, GLuint *state, GLuint count,
 
    col = ((a << 24) | (r << 16) | (g << 8) | b);
 
-   state[count++] = _3DSTATE_COLOR_FACTOR_N_CMD(blendUnit); 
+   state[count++] = _3DSTATE_COLOR_FACTOR_N_CMD(blendUnit);
    state[count++] = col;
 
    return count;
 }
 
 
-static __inline__ GLuint GetTexelOp(GLint unit)
+static INLINE GLuint
+GetTexelOp(GLint unit)
 {
-   switch(unit) {
-   case 0: return TEXBLENDARG_TEXEL0;
-   case 1: return TEXBLENDARG_TEXEL1;
-   case 2: return TEXBLENDARG_TEXEL2;
-   case 3: return TEXBLENDARG_TEXEL3;
-   default: return TEXBLENDARG_TEXEL0;
+   switch (unit) {
+   case 0:
+      return TEXBLENDARG_TEXEL0;
+   case 1:
+      return TEXBLENDARG_TEXEL1;
+   case 2:
+      return TEXBLENDARG_TEXEL2;
+   case 3:
+      return TEXBLENDARG_TEXEL3;
+   default:
+      return TEXBLENDARG_TEXEL0;
    }
 }
 
@@ -132,12 +132,10 @@ static __inline__ GLuint GetTexelOp(GLint unit)
  * partial support for the extension?
  */
 GLuint
-i830SetTexEnvCombine(i830ContextPtr i830,
-		     const struct gl_tex_env_combine_state * combine,
-		     GLint blendUnit,
-		     GLuint texel_op,
-		     GLuint *state,
-		     const GLfloat *factor )
+i830SetTexEnvCombine(struct i830_context * i830,
+                     const struct gl_tex_env_combine_state * combine,
+                     GLint blendUnit,
+                     GLuint texel_op, GLuint * state, const GLfloat * factor)
 {
    const GLuint numColorArgs = combine->_NumArgsRGB;
    const GLuint numAlphaArgs = combine->_NumArgsA;
@@ -162,7 +160,7 @@ i830SetTexEnvCombine(i830ContextPtr i830,
       TEXPIPE_ALPHA | TEXBLEND_ARG0 | TEXBLENDARG_MODIFY_PARMS,
    };
 
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
+   if (INTEL_DEBUG & DEBUG_TEXTURE)
       fprintf(stderr, "%s\n", __FUNCTION__);
 
 
@@ -188,23 +186,23 @@ i830SetTexEnvCombine(i830ContextPtr i830,
    }
 
 
-   switch(combine->ModeRGB) {
-   case GL_REPLACE: 
+   switch (combine->ModeRGB) {
+   case GL_REPLACE:
       blendop = TEXBLENDOP_ARG1;
       break;
-   case GL_MODULATE: 
+   case GL_MODULATE:
       blendop = TEXBLENDOP_MODULATE;
       break;
-   case GL_ADD: 
+   case GL_ADD:
       blendop = TEXBLENDOP_ADD;
       break;
    case GL_ADD_SIGNED:
-      blendop = TEXBLENDOP_ADDSIGNED; 
+      blendop = TEXBLENDOP_ADDSIGNED;
       break;
    case GL_INTERPOLATE:
-      blendop = TEXBLENDOP_BLEND; 
+      blendop = TEXBLENDOP_BLEND;
       break;
-   case GL_SUBTRACT: 
+   case GL_SUBTRACT:
       blendop = TEXBLENDOP_SUBTRACT;
       break;
    case GL_DOT3_RGB_EXT:
@@ -215,55 +213,54 @@ i830SetTexEnvCombine(i830ContextPtr i830,
    case GL_DOT3_RGBA:
       blendop = TEXBLENDOP_DOT3;
       break;
-   default: 
-      return pass_through( state, blendUnit );
+   default:
+      return pass_through(state, blendUnit);
    }
 
    blendop |= (rgb_shift << TEXOP_SCALE_SHIFT);
 
 
    /* Handle RGB args */
-   for(i = 0; i < 3; i++) {
-      switch(combine->SourceRGB[i]) {
-      case GL_TEXTURE: 
-	 args_RGB[i] = texel_op;
-	 break;
+   for (i = 0; i < 3; i++) {
+      switch (combine->SourceRGB[i]) {
+      case GL_TEXTURE:
+         args_RGB[i] = texel_op;
+         break;
       case GL_TEXTURE0:
       case GL_TEXTURE1:
       case GL_TEXTURE2:
       case GL_TEXTURE3:
-	 args_RGB[i] = GetTexelOp( combine->SourceRGB[i] - GL_TEXTURE0 );
-	 break;
+         args_RGB[i] = GetTexelOp(combine->SourceRGB[i] - GL_TEXTURE0);
+         break;
       case GL_CONSTANT:
-	 args_RGB[i] = TEXBLENDARG_FACTOR_N; 
-	 need_factor = 1;
-	 break;
+         args_RGB[i] = TEXBLENDARG_FACTOR_N;
+         need_factor = 1;
+         break;
       case GL_PRIMARY_COLOR:
-	 args_RGB[i] = TEXBLENDARG_DIFFUSE;
-	 break;
+         args_RGB[i] = TEXBLENDARG_DIFFUSE;
+         break;
       case GL_PREVIOUS:
-	 args_RGB[i] = TEXBLENDARG_CURRENT; 
-	 break;
-      default: 
-	 return pass_through( state, blendUnit );
+         args_RGB[i] = TEXBLENDARG_CURRENT;
+         break;
+      default:
+         return pass_through(state, blendUnit);
       }
 
-      switch(combine->OperandRGB[i]) {
-      case GL_SRC_COLOR: 
-	 args_RGB[i] |= 0;
-	 break;
-      case GL_ONE_MINUS_SRC_COLOR: 
-	 args_RGB[i] |= TEXBLENDARG_INV_ARG;
-	 break;
-      case GL_SRC_ALPHA: 
-	 args_RGB[i] |= TEXBLENDARG_REPLICATE_ALPHA;
-	 break;
-      case GL_ONE_MINUS_SRC_ALPHA: 
-	 args_RGB[i] |= (TEXBLENDARG_REPLICATE_ALPHA | 
-			 TEXBLENDARG_INV_ARG);
-	 break;
-      default: 
-	 return pass_through( state, blendUnit );
+      switch (combine->OperandRGB[i]) {
+      case GL_SRC_COLOR:
+         args_RGB[i] |= 0;
+         break;
+      case GL_ONE_MINUS_SRC_COLOR:
+         args_RGB[i] |= TEXBLENDARG_INV_ARG;
+         break;
+      case GL_SRC_ALPHA:
+         args_RGB[i] |= TEXBLENDARG_REPLICATE_ALPHA;
+         break;
+      case GL_ONE_MINUS_SRC_ALPHA:
+         args_RGB[i] |= (TEXBLENDARG_REPLICATE_ALPHA | TEXBLENDARG_INV_ARG);
+         break;
+      default:
+         return pass_through(state, blendUnit);
       }
    }
 
@@ -275,76 +272,76 @@ i830SetTexEnvCombine(i830ContextPtr i830,
     * Note - the global factor is set up with alpha == .5, so 
     * the alpha part of the DOT4 calculation should be zero.
     */
-   if ( combine->ModeRGB == GL_DOT3_RGBA_EXT || 
-	combine->ModeRGB == GL_DOT3_RGBA ) {
+   if (combine->ModeRGB == GL_DOT3_RGBA_EXT ||
+       combine->ModeRGB == GL_DOT3_RGBA) {
       ablendop = TEXBLENDOP_DOT4;
-      args_A[0] = TEXBLENDARG_FACTOR; /* the global factor */
+      args_A[0] = TEXBLENDARG_FACTOR;   /* the global factor */
       args_A[1] = TEXBLENDARG_FACTOR;
       args_A[2] = TEXBLENDARG_FACTOR;
    }
    else {
-      switch(combine->ModeA) {
-      case GL_REPLACE: 
-	 ablendop = TEXBLENDOP_ARG1;
-	 break;
-      case GL_MODULATE: 
-	 ablendop = TEXBLENDOP_MODULATE;
-	 break;
-      case GL_ADD: 
-	 ablendop = TEXBLENDOP_ADD;
-	 break;
+      switch (combine->ModeA) {
+      case GL_REPLACE:
+         ablendop = TEXBLENDOP_ARG1;
+         break;
+      case GL_MODULATE:
+         ablendop = TEXBLENDOP_MODULATE;
+         break;
+      case GL_ADD:
+         ablendop = TEXBLENDOP_ADD;
+         break;
       case GL_ADD_SIGNED:
-	 ablendop = TEXBLENDOP_ADDSIGNED; 
-	 break;
+         ablendop = TEXBLENDOP_ADDSIGNED;
+         break;
       case GL_INTERPOLATE:
-	 ablendop = TEXBLENDOP_BLEND; 
-	 break;
-      case GL_SUBTRACT: 
-	 ablendop = TEXBLENDOP_SUBTRACT;
-	 break;
+         ablendop = TEXBLENDOP_BLEND;
+         break;
+      case GL_SUBTRACT:
+         ablendop = TEXBLENDOP_SUBTRACT;
+         break;
       default:
-	 return pass_through( state, blendUnit );
+         return pass_through(state, blendUnit);
       }
 
 
       ablendop |= (alpha_shift << TEXOP_SCALE_SHIFT);
 
       /* Handle A args */
-      for(i = 0; i < 3; i++) {
-	 switch(combine->SourceA[i]) {
-	 case GL_TEXTURE: 
-	    args_A[i] = texel_op;
-	    break;
-	 case GL_TEXTURE0:
-	 case GL_TEXTURE1:
-	 case GL_TEXTURE2:
-	 case GL_TEXTURE3:
-	    args_A[i] = GetTexelOp( combine->SourceA[i] - GL_TEXTURE0 );
-	    break;
-	 case GL_CONSTANT:
-	    args_A[i] = TEXBLENDARG_FACTOR_N; 
-	    need_factor = 1;
-	    break;
-	 case GL_PRIMARY_COLOR:
-	    args_A[i] = TEXBLENDARG_DIFFUSE; 
-	    break;
-	 case GL_PREVIOUS:
-	    args_A[i] = TEXBLENDARG_CURRENT; 
-	    break;
-	 default: 
-	    return pass_through( state, blendUnit );
-	 }
-
-	 switch(combine->OperandA[i]) {
-	 case GL_SRC_ALPHA: 
-	    args_A[i] |= 0;
-	    break;
-	 case GL_ONE_MINUS_SRC_ALPHA: 
-	    args_A[i] |= TEXBLENDARG_INV_ARG;
-	    break;
-	 default: 
-	    return pass_through( state, blendUnit );
-	 }
+      for (i = 0; i < 3; i++) {
+         switch (combine->SourceA[i]) {
+         case GL_TEXTURE:
+            args_A[i] = texel_op;
+            break;
+         case GL_TEXTURE0:
+         case GL_TEXTURE1:
+         case GL_TEXTURE2:
+         case GL_TEXTURE3:
+            args_A[i] = GetTexelOp(combine->SourceA[i] - GL_TEXTURE0);
+            break;
+         case GL_CONSTANT:
+            args_A[i] = TEXBLENDARG_FACTOR_N;
+            need_factor = 1;
+            break;
+         case GL_PRIMARY_COLOR:
+            args_A[i] = TEXBLENDARG_DIFFUSE;
+            break;
+         case GL_PREVIOUS:
+            args_A[i] = TEXBLENDARG_CURRENT;
+            break;
+         default:
+            return pass_through(state, blendUnit);
+         }
+
+         switch (combine->OperandA[i]) {
+         case GL_SRC_ALPHA:
+            args_A[i] |= 0;
+            break;
+         case GL_ONE_MINUS_SRC_ALPHA:
+            args_A[i] |= TEXBLENDARG_INV_ARG;
+            break;
+         default:
+            return pass_through(state, blendUnit);
+         }
       }
    }
 
@@ -363,86 +360,86 @@ i830SetTexEnvCombine(i830ContextPtr i830,
 
    used = 0;
    state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-		    TEXPIPE_COLOR |
-		    ENABLE_TEXOUTPUT_WRT_SEL |
-		    TEXOP_OUTPUT_CURRENT |
-		    DISABLE_TEX_CNTRL_STAGE |
-		    TEXOP_MODIFY_PARMS |
-		    blendop);
+                    TEXPIPE_COLOR |
+                    ENABLE_TEXOUTPUT_WRT_SEL |
+                    TEXOP_OUTPUT_CURRENT |
+                    DISABLE_TEX_CNTRL_STAGE | TEXOP_MODIFY_PARMS | blendop);
    state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-		    TEXPIPE_ALPHA |
-		    ENABLE_TEXOUTPUT_WRT_SEL |
-		    TEXOP_OUTPUT_CURRENT |
-		    TEXOP_MODIFY_PARMS |
-		    ablendop);
+                    TEXPIPE_ALPHA |
+                    ENABLE_TEXOUTPUT_WRT_SEL |
+                    TEXOP_OUTPUT_CURRENT | TEXOP_MODIFY_PARMS | ablendop);
 
-   for ( i = 0 ; i < numColorArgs ; i++ ) {
+   for (i = 0; i < numColorArgs; i++) {
       state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-		       tex_blend_rgb[i] | args_RGB[i]);
+                       tex_blend_rgb[i] | args_RGB[i]);
    }
 
-   for ( i = 0 ; i < numAlphaArgs ; i++ ) {
+   for (i = 0; i < numAlphaArgs; i++) {
       state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-		       tex_blend_a[i] | args_A[i]);
+                       tex_blend_a[i] | args_A[i]);
    }
 
 
-   if (need_factor) 
-      return emit_factor( blendUnit, state, used, factor );
-   else 
+   if (need_factor)
+      return emit_factor(blendUnit, state, used, factor);
+   else
       return used;
 }
 
 
-static void emit_texblend( i830ContextPtr i830, GLuint unit, GLuint blendUnit,
-			   GLboolean last_stage )
+static void
+emit_texblend(struct i830_context *i830, GLuint unit, GLuint blendUnit,
+              GLboolean last_stage)
 {
    struct gl_texture_unit *texUnit = &i830->intel.ctx.Texture.Unit[unit];
    GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
 
 
-   if (0) fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
+   if (0)
+      fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
 
    /* Update i830->state.TexBlend
-    */ 
-   tmp_sz = i830SetTexEnvCombine(i830, texUnit->_CurrentCombine, blendUnit, 
-				 GetTexelOp(unit), tmp,
-				 texUnit->EnvColor );
+    */
+   tmp_sz = i830SetTexEnvCombine(i830, texUnit->_CurrentCombine, blendUnit,
+                                 GetTexelOp(unit), tmp, texUnit->EnvColor);
 
-   if (last_stage) 
+   if (last_stage)
       tmp[0] |= TEXOP_LAST_STAGE;
 
    if (tmp_sz != i830->state.TexBlendWordsUsed[blendUnit] ||
-       memcmp( tmp, i830->state.TexBlend[blendUnit], tmp_sz * sizeof(GLuint))) {
-      
-      I830_STATECHANGE( i830, I830_UPLOAD_TEXBLEND(blendUnit) );
-      memcpy( i830->state.TexBlend[blendUnit], tmp, tmp_sz * sizeof(GLuint));
+       memcmp(tmp, i830->state.TexBlend[blendUnit],
+              tmp_sz * sizeof(GLuint))) {
+
+      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(blendUnit));
+      memcpy(i830->state.TexBlend[blendUnit], tmp, tmp_sz * sizeof(GLuint));
       i830->state.TexBlendWordsUsed[blendUnit] = tmp_sz;
    }
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(blendUnit), GL_TRUE);
 }
 
-static void emit_passthrough( i830ContextPtr i830 )
+static void
+emit_passthrough(struct i830_context *i830)
 {
    GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
    GLuint unit = 0;
 
-   tmp_sz = pass_through( tmp, unit );
+   tmp_sz = pass_through(tmp, unit);
    tmp[0] |= TEXOP_LAST_STAGE;
 
    if (tmp_sz != i830->state.TexBlendWordsUsed[unit] ||
-       memcmp( tmp, i830->state.TexBlend[unit], tmp_sz * sizeof(GLuint))) {
-      
-      I830_STATECHANGE( i830, I830_UPLOAD_TEXBLEND(unit) );
-      memcpy( i830->state.TexBlend[unit], tmp, tmp_sz * sizeof(GLuint));
+       memcmp(tmp, i830->state.TexBlend[unit], tmp_sz * sizeof(GLuint))) {
+
+      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(unit));
+      memcpy(i830->state.TexBlend[unit], tmp, tmp_sz * sizeof(GLuint));
       i830->state.TexBlendWordsUsed[unit] = tmp_sz;
    }
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(unit), GL_TRUE);
 }
 
-void i830EmitTextureBlend( i830ContextPtr i830 )
+void
+i830EmitTextureBlend(struct i830_context *i830)
 {
    GLcontext *ctx = &i830->intel.ctx;
    GLuint unit, last_stage = 0, blendunit = 0;
@@ -450,16 +447,15 @@ void i830EmitTextureBlend( i830ContextPtr i830 )
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND_ALL, GL_FALSE);
 
    if (ctx->Texture._EnabledUnits) {
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) 
-	    last_stage = unit;
+      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
+         if (ctx->Texture.Unit[unit]._ReallyEnabled)
+            last_stage = unit;
 
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) 
-	    emit_texblend( i830, unit, blendunit++, last_stage == unit );
+      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
+         if (ctx->Texture.Unit[unit]._ReallyEnabled)
+            emit_texblend(i830, unit, blendunit++, last_stage == unit);
    }
    else {
-      emit_passthrough( i830 );
+      emit_passthrough(i830);
    }
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_texstate.c b/src/mesa/drivers/dri/i915/i830_texstate.c
index ba972dac8f1..c718bb0055d 100644
--- a/src/mesa/drivers/dri/i915/i830_texstate.c
+++ b/src/mesa/drivers/dri/i915/i830_texstate.c
@@ -25,459 +25,326 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texformat.h"
-#include "texstore.h"
-
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_ioctl.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/texformat.h"
+
+#include "intel_mipmap_tree.h"
 #include "intel_tex.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
 
-static const GLint initial_offsets[6][2] = { {0,0},
-				       {0,2},
-				       {1,0},
-				       {1,2},
-				       {1,1},
-				       {1,3} };
-
-static const GLint step_offsets[6][2] = { {0,2},
-				    {0,2},
-				    {-1,2},
-				    {-1,2},
-				    {-1,1},
-				    {-1,1} };
 
-#define I830_TEX_UNIT_ENABLED(unit)		(1<<unit)
 
-static GLboolean i830SetTexImages( i830ContextPtr i830, 
-				  struct gl_texture_object *tObj )
+static GLuint
+translate_texture_format(GLuint mesa_format)
 {
-   GLuint total_height, pitch, i, textureFormat;
-   i830TextureObjectPtr t = (i830TextureObjectPtr) tObj->DriverData;
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   GLint firstLevel, lastLevel, numLevels;
-
-   switch( baseImage->TexFormat->MesaFormat ) {
+   switch (mesa_format) {
    case MESA_FORMAT_L8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_L8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_L8;
    case MESA_FORMAT_I8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8;
    case MESA_FORMAT_A8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8; /* Kludge -- check with conform, glean */
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8; /* Kludge! */
    case MESA_FORMAT_AL88:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_AY88;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_AY88;
    case MESA_FORMAT_RGB565:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_RGB565;
    case MESA_FORMAT_ARGB1555:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
    case MESA_FORMAT_ARGB4444:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB4444;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      t->intel.texelBytes = 4;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      break;
-
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
    case MESA_FORMAT_YCBCR_REV:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL | 
-		       TM0S1_COLORSPACE_CONVERSION);
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY | /* ??? */
-		       TM0S1_COLORSPACE_CONVERSION);
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
    case MESA_FORMAT_RGB_FXT1:
    case MESA_FORMAT_RGBA_FXT1:
-     t->intel.texelBytes = 2;
-     textureFormat = MAPSURF_COMPRESSED | MT_COMPRESS_FXT1;
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
    case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-     /* 
-      * DXTn pitches are Width/4 * blocksize in bytes 
-      * for DXT1: blocksize=8 so Width/4*8 = Width * 2 
-      * for DXT3/5: blocksize=16 so Width/4*16 = Width * 4
-      */
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-     break;
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
    case MESA_FORMAT_RGBA_DXT3:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-     break;
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
    case MESA_FORMAT_RGBA_DXT5:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
    default:
-      fprintf(stderr, "%s: bad image format\n", __FUNCTION__);
+      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
       abort();
+      return 0;
    }
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    * This depends on the base image size, GL_TEXTURE_MIN_LOD,
-    * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
-    * Yes, this looks overly complicated, but it's all needed.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
+}
 
 
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
 
 
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      pitch = dim * t->intel.texelBytes;
-      pitch *= 2;		/* double pitch for cube layouts */
-      pitch = (pitch + 3) & ~3;
-      
-      total_height = dim * 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    if (!t->intel.image[face][i].image) {
-	       fprintf(stderr, "no image %d %d\n", face, i);
-	       break;		/* can't happen */
-	    }
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    x += step_offsets[face][0] * d;
-	    y += step_offsets[face][1] * d;
-	 }
-      }
-      break;
-   }
+/* The i915 (and related graphics cores) do not support GL_CLAMP.  The
+ * Intel drivers for "other operating systems" implement GL_CLAMP as
+ * GL_CLAMP_TO_EDGE, so the same is done here.
+ */
+static GLuint
+translate_wrap_mode(GLenum wrap)
+{
+   switch (wrap) {
+   case GL_REPEAT:
+      return TEXCOORDMODE_WRAP;
+   case GL_CLAMP:
+   case GL_CLAMP_TO_EDGE:
+      return TEXCOORDMODE_CLAMP;        /* not really correct */
+   case GL_CLAMP_TO_BORDER:
+      return TEXCOORDMODE_CLAMP_BORDER;
+   case GL_MIRRORED_REPEAT:
+      return TEXCOORDMODE_MIRROR;
    default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      for ( total_height = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 if (t->intel.image[0][i].image->IsCompressed)
-	 {
-	   if (t->intel.image[0][i].image->Height > 4)
-	     total_height += t->intel.image[0][i].image->Height/4;
-	   else
-	     total_height += 1;
-	 }
-	 else
-	   total_height += MAX2(2, t->intel.image[0][i].image->Height);
-      }
-      break;
+      return TEXCOORDMODE_WRAP;
    }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = total_height*pitch;
-   t->intel.max_level = i-1;
-   t->Setup[I830_TEXREG_TM0S1] = 
-      (((tObj->Image[0][firstLevel]->Height - 1) << TM0S1_HEIGHT_SHIFT) |
-       ((tObj->Image[0][firstLevel]->Width - 1) << TM0S1_WIDTH_SHIFT) |
-       textureFormat);
-   t->Setup[I830_TEXREG_TM0S2] = 
-      (((pitch / 4) - 1) << TM0S2_PITCH_SHIFT) |
-      TM0S2_CUBE_FACE_ENA_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAX_MIP_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_MIP_MASK;
-   t->Setup[I830_TEXREG_TM0S3] |= ((numLevels - 1)*4) << TM0S3_MIN_MIP_SHIFT;
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-
-   return intelUploadTexImages( &i830->intel, &t->intel, 0 );
 }
 
 
-static void i830_import_tex_unit( i830ContextPtr i830, 
-			   i830TextureObjectPtr t,
-			   GLuint unit )
+/* Recalculate all state from scratch.  Perhaps not the most
+ * efficient, but this has gotten complex enough that we need
+ * something which is understandable and reliable.
+ */
+static GLboolean
+i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 {
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
-      fprintf(stderr, "%s unit(%d)\n", __FUNCTION__, unit);
-   
-   if (i830->intel.CurrentTexObj[unit]) 
-      i830->intel.CurrentTexObj[unit]->base.bound &= ~(1U << unit);
-
-   i830->intel.CurrentTexObj[unit] = (intelTextureObjectPtr)t;
-   t->intel.base.bound |= (1 << unit);
-
-   I830_STATECHANGE( i830, I830_UPLOAD_TEX(unit) );
-
-   i830->state.Tex[unit][I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 | 
-					       (LOAD_TEXTURE_MAP0 << unit) | 4);
-   i830->state.Tex[unit][I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE |
-					       t->intel.TextureOffset);
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S1] = t->Setup[I830_TEXREG_TM0S1];
-   i830->state.Tex[unit][I830_TEXREG_TM0S2] = t->Setup[I830_TEXREG_TM0S2];
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S3] &= TM0S3_LOD_BIAS_MASK;
-   i830->state.Tex[unit][I830_TEXREG_TM0S3] |= (t->Setup[I830_TEXREG_TM0S3] &
-						~TM0S3_LOD_BIAS_MASK);
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S4] = t->Setup[I830_TEXREG_TM0S4];
-   i830->state.Tex[unit][I830_TEXREG_MCS] = (t->Setup[I830_TEXREG_MCS] & 
-					     ~MAP_UNIT_MASK);   
-   i830->state.Tex[unit][I830_TEXREG_CUBE] = t->Setup[I830_TEXREG_CUBE];
-   i830->state.Tex[unit][I830_TEXREG_MCS] |= MAP_UNIT(unit);
-
-   t->intel.dirty &= ~I830_UPLOAD_TEX(unit);
-}
-
+   GLcontext *ctx = &intel->ctx;
+   struct i830_context *i830 = i830_context(ctx);
+   struct gl_texture_unit *tUnit = &ctx->Texture.Unit[unit];
+   struct gl_texture_object *tObj = tUnit->_Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   struct gl_texture_image *firstImage;
+   GLuint *state = i830->state.Tex[unit], format, pitch;
+   GLint lodbias;
 
+   memset(state, 0, sizeof(state));
 
-static GLboolean enable_tex_common( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i830TextureObjectPtr t = (i830TextureObjectPtr)tObj->DriverData;
+   /*We need to refcount these. */
 
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
+   if (i830->state.tex_buffer[unit] != NULL) {
+       dri_bo_unreference(i830->state.tex_buffer[unit]);
+       i830->state.tex_buffer[unit] = NULL;
+   }
 
-   /* Fallback if there's a texture border */
-   if ( tObj->Image[0][tObj->BaseLevel]->Border > 0 ) {
-      fprintf(stderr, "Texture border\n");
+   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
       return GL_FALSE;
-   }
 
-   /* Upload teximages (not pipelined)
+   /* Get first image here, since intelObj->firstLevel will get set in
+    * the intel_finalize_mipmap_tree() call above.
     */
-   if (t->intel.base.dirty_images[0]) {
-      if (!i830SetTexImages( i830, tObj )) {
-	 return GL_FALSE;
+   firstImage = tObj->Image[0][intelObj->firstLevel];
+
+   if (intelObj->imageOverride) {
+      i830->state.tex_buffer[unit] = NULL;
+      i830->state.tex_offset[unit] = intelObj->textureOffset;
+
+      switch (intelObj->depthOverride) {
+      case 32:
+	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
+	 break;
+      case 24:
+      default:
+	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
+	 break;
+      case 16:
+	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
+	 break;
       }
-   }
-
-   /* Update state if this is a different texture object to last
-    * time.
-    */
-   if (i830->intel.CurrentTexObj[unit] != &t->intel || 
-       (t->intel.dirty & I830_UPLOAD_TEX(unit))) {
-      i830_import_tex_unit( i830, t, unit);
-   }
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
-
-   return GL_TRUE;
-}
-
-static GLboolean enable_tex_rect( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
 
-   mcs &= ~TEXCOORDS_ARE_NORMAL;
-   mcs |= TEXCOORDS_ARE_IN_TEXELUNITS;
+      pitch = intelObj->pitchOverride;
+   } else {
+      dri_bo_reference(intelObj->mt->region->buffer);
+      i830->state.tex_buffer[unit] = intelObj->mt->region->buffer;
+      i830->state.tex_offset[unit] = intel_miptree_image_offset(intelObj->mt,
+								0, intelObj->
+								firstLevel);
 
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (0 != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = 0;
+      format = translate_texture_format(firstImage->TexFormat->MesaFormat);
+      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
    }
 
-   return GL_TRUE;
-}
+   state[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
+                               (LOAD_TEXTURE_MAP0 << unit) | 4);
 
+/*    state[I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE | */
+/* 			       t->intel.TextureOffset); */
 
-static GLboolean enable_tex_2d( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
 
-   mcs &= ~TEXCOORDS_ARE_IN_TEXELUNITS;
-   mcs |= TEXCOORDS_ARE_NORMAL;
+   state[I830_TEXREG_TM0S1] =
+      (((firstImage->Height - 1) << TM0S1_HEIGHT_SHIFT) |
+       ((firstImage->Width - 1) << TM0S1_WIDTH_SHIFT) | format);
+
+   state[I830_TEXREG_TM0S2] =
+      ((((pitch / 4) - 1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
 
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (0 != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = 0;
+   {
+      if (tObj->Target == GL_TEXTURE_CUBE_MAP)
+         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit) |
+                                    CUBE_NEGX_ENABLE |
+                                    CUBE_POSX_ENABLE |
+                                    CUBE_NEGY_ENABLE |
+                                    CUBE_POSY_ENABLE |
+                                    CUBE_NEGZ_ENABLE | CUBE_POSZ_ENABLE);
+      else
+         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit));
    }
 
-   return GL_TRUE;
-}
 
- 
-static GLboolean enable_tex_cube( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i830TextureObjectPtr t = (i830TextureObjectPtr)tObj->DriverData;
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
-   const GLuint cube = CUBE_NEGX_ENABLE | CUBE_POSX_ENABLE
-     | CUBE_NEGY_ENABLE | CUBE_POSY_ENABLE
-     | CUBE_NEGZ_ENABLE | CUBE_POSZ_ENABLE;
-   GLuint face;
-
-   mcs &= ~TEXCOORDS_ARE_IN_TEXELUNITS;
-   mcs |= TEXCOORDS_ARE_NORMAL;
-
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (cube != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = cube;
-   }
 
-   /* Upload teximages (not pipelined)
-    */
-   if ( t->intel.base.dirty_images[0] || t->intel.base.dirty_images[1] ||
-        t->intel.base.dirty_images[2] || t->intel.base.dirty_images[3] ||
-        t->intel.base.dirty_images[4] || t->intel.base.dirty_images[5] ) {
-      i830SetTexImages( i830, tObj );
-   }
 
-   /* upload (per face) */
-   for (face = 0; face < 6; face++) {
-      if (t->intel.base.dirty_images[face]) {
-	 if (!intelUploadTexImages( &i830->intel, &t->intel, face )) {
-	    return GL_FALSE;
-	 }
+   {
+      GLuint minFilt, mipFilt, magFilt;
+
+      switch (tObj->MinFilter) {
+      case GL_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_NEAREST_MIPMAP_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_LINEAR_MIPMAP_NEAREST:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_NEAREST_MIPMAP_LINEAR:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      case GL_LINEAR_MIPMAP_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      default:
+         return GL_FALSE;
       }
-   }
-
 
-   return GL_TRUE;
-}
+      if (tObj->MaxAnisotropy > 1.0) {
+         minFilt = FILTER_ANISOTROPIC;
+         magFilt = FILTER_ANISOTROPIC;
+      }
+      else {
+         switch (tObj->MagFilter) {
+         case GL_NEAREST:
+            magFilt = FILTER_NEAREST;
+            break;
+         case GL_LINEAR:
+            magFilt = FILTER_LINEAR;
+            break;
+         default:
+            return GL_FALSE;
+         }
+      }
 
+      lodbias = (int) ((tUnit->LodBias + tObj->LodBias) * 16.0);
+      if (lodbias < -64)
+          lodbias = -64;
+      if (lodbias > 63)
+          lodbias = 63;
+      
+      state[I830_TEXREG_TM0S3] = ((lodbias << TM0S3_LOD_BIAS_SHIFT) & 
+                                  TM0S3_LOD_BIAS_MASK);
+#if 0
+      /* YUV conversion:
+       */
+      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
+          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
+         state[I830_TEXREG_TM0S3] |= SS2_COLORSPACE_CONVERSION;
+#endif
+
+      state[I830_TEXREG_TM0S3] |= ((intelObj->lastLevel -
+                                    intelObj->firstLevel) *
+                                   4) << TM0S3_MIN_MIP_SHIFT;
+
+      state[I830_TEXREG_TM0S3] |= ((minFilt << TM0S3_MIN_FILTER_SHIFT) |
+                                   (mipFilt << TM0S3_MIP_FILTER_SHIFT) |
+                                   (magFilt << TM0S3_MAG_FILTER_SHIFT));
+   }
 
-static GLboolean disable_tex( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   {
+      GLenum ws = tObj->WrapS;
+      GLenum wt = tObj->WrapT;
 
-   /* This is happening too often.  I need to conditionally send diffuse
-    * state to the card.  Perhaps a diffuse dirty flag of some kind.
-    * Will need to change this logic if more than 2 texture units are
-    * used.  We need to only do this up to the last unit enabled, or unit
-    * one if nothing is enabled.
-    */
 
-   if ( i830->intel.CurrentTexObj[unit] != NULL ) {
-      /* The old texture is no longer bound to this texture unit.
-       * Mark it as such.
+      /* 3D textures not available on i830
        */
-
-      i830->intel.CurrentTexObj[unit]->base.bound &= ~(1U << 0);
-      i830->intel.CurrentTexObj[unit] = NULL;
+      if (tObj->Target == GL_TEXTURE_3D)
+         return GL_FALSE;
+
+      state[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
+                                MAP_UNIT(unit) |
+                                ENABLE_TEXCOORD_PARAMS |
+                                ss3 |
+                                ENABLE_ADDR_V_CNTL |
+                                TEXCOORD_ADDR_V_MODE(translate_wrap_mode(wt))
+                                | ENABLE_ADDR_U_CNTL |
+                                TEXCOORD_ADDR_U_MODE(translate_wrap_mode
+                                                     (ws)));
    }
 
-   return GL_TRUE;
-}
 
-static GLboolean i830UpdateTexUnit( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   state[I830_TEXREG_TM0S4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
+                                                  tObj->_BorderChan[1],
+                                                  tObj->_BorderChan[2],
+                                                  tObj->_BorderChan[3]);
 
-   if (texUnit->_ReallyEnabled &&
-       INTEL_CONTEXT(ctx)->intelScreen->tex.size < 2048 * 1024)
-      return GL_FALSE;
 
-   switch(texUnit->_ReallyEnabled) {
-   case TEXTURE_1D_BIT:
-   case TEXTURE_2D_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_2d( ctx, unit ));
-   case TEXTURE_RECT_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_rect( ctx, unit ));
-   case TEXTURE_CUBE_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_cube( ctx, unit ));
-   case 0:
-      return disable_tex( ctx, unit );
-   default:
-      return GL_FALSE;
-   }
+   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
+   /* memcmp was already disabled, but definitely won't work as the
+    * region might now change and that wouldn't be detected:
+    */
+   I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
+   return GL_TRUE;
 }
 
 
-void i830UpdateTextureState( intelContextPtr intel )
-{
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   GLcontext *ctx = &intel->ctx;
-   GLboolean ok;
-
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
 
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX_ALL, GL_FALSE);
 
-   ok = (i830UpdateTexUnit( ctx, 0 ) &&
-	 i830UpdateTexUnit( ctx, 1 ) &&
-	 i830UpdateTexUnit( ctx, 2 ) &&
-	 i830UpdateTexUnit( ctx, 3 ));
+void
+i830UpdateTextureState(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLboolean ok = GL_TRUE;
+   GLuint i;
+
+   for (i = 0; i < I830_TEX_UNITS && ok; i++) {
+      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
+      case TEXTURE_1D_BIT:
+      case TEXTURE_2D_BIT:
+      case TEXTURE_CUBE_BIT:
+         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_NORMAL);
+         break;
+      case TEXTURE_RECT_BIT:
+         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_IN_TEXELUNITS);
+         break;
+      case 0:{
+	 struct i830_context *i830 = i830_context(&intel->ctx);
+         if (i830->state.active & I830_UPLOAD_TEX(i)) 
+            I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(i), GL_FALSE);
+
+	 if (i830->state.tex_buffer[i] != NULL) {
+	    dri_bo_unreference(i830->state.tex_buffer[i]);
+	    i830->state.tex_buffer[i] = NULL;
+	 }
+         break;
+      }
+      case TEXTURE_3D_BIT:
+      default:
+         ok = GL_FALSE;
+         break;
+      }
+   }
 
-   FALLBACK( intel, I830_FALLBACK_TEXTURE, !ok );
+   FALLBACK(intel, I830_FALLBACK_TEXTURE, !ok);
 
    if (ok)
-      i830EmitTextureBlend( i830 );
+      i830EmitTextureBlend(i830);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index d40cf705a35..3b3ff2bceda 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -25,17 +25,20 @@
  * 
  **************************************************************************/
 
+#include "glapi/glapi.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
-
 #include "intel_batchbuffer.h"
-
+#include "intel_regions.h"
+#include "intel_tris.h"
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
 
-static GLboolean i830_check_vertex_size( intelContextPtr intel,
-					 GLuint expected );
+#define FILE_DEBUG_FLAG DEBUG_STATE
+
+static GLboolean i830_check_vertex_size(struct intel_context *intel,
+                                        GLuint expected);
 
 #define SZ_TO_HW(sz)  ((sz-2)&0x3)
 #define EMIT_SZ(sz)   (EMIT_1F + (sz) - 1)
@@ -59,10 +62,16 @@ do {									\
 #define VRTX_TEX_SET_FMT(n, x)          ((x)<<((n)*2))
 #define TEXBIND_SET(n, x) 		((x)<<((n)*4))
 
-static void i830_render_start( intelContextPtr intel )
+static void
+i830_render_prevalidate(struct intel_context *intel)
+{
+}
+
+static void
+i830_render_start(struct intel_context *intel)
 {
    GLcontext *ctx = &intel->ctx;
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
    DECLARE_RENDERINPUTS(index_bitset);
@@ -70,7 +79,7 @@ static void i830_render_start( intelContextPtr intel )
    GLuint v2 = _3DSTATE_VFT1_CMD;
    GLuint mcsb1 = 0;
 
-   RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
+   RENDERINPUTS_COPY(index_bitset, tnl->render_inputs_bitset);
 
    /* Important:
     */
@@ -80,196 +89,215 @@ static void i830_render_start( intelContextPtr intel )
    /* EMIT_ATTR's must be in order as they tell t_vertex.c how to
     * build up a hardware vertex.
     */
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, VFT0_XYZW );
+   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, VFT0_XYZW);
       intel->coloroffset = 4;
    }
    else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, VFT0_XYZ );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, VFT0_XYZ);
       intel->coloroffset = 3;
    }
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POINTSIZE )) {
-      EMIT_ATTR( _TNL_ATTRIB_POINTSIZE, EMIT_1F, VFT0_POINT_WIDTH );
+   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_POINTSIZE)) {
+      EMIT_ATTR(_TNL_ATTRIB_POINTSIZE, EMIT_1F, VFT0_POINT_WIDTH);
    }
 
-   EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, VFT0_DIFFUSE );
-      
+   EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, VFT0_DIFFUSE);
+
    intel->specoffset = 0;
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 ) ||
-       RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG )) {
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 )) {
+   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1) ||
+       RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG)) {
+      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1)) {
          intel->specoffset = intel->coloroffset + 1;
-         EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, VFT0_SPEC );
+         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, VFT0_SPEC);
       }
       else
-         EMIT_PAD( 3 );
+         EMIT_PAD(3);
 
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG ))
-         EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1UB_1F, VFT0_SPEC );
+      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG))
+         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, VFT0_SPEC);
       else
-         EMIT_PAD( 1 );
+         EMIT_PAD(1);
    }
 
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
+   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
       int i, count = 0;
 
       for (i = 0; i < I830_TEX_UNITS; i++) {
-         if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
+         if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_TEX(i))) {
             GLuint sz = VB->TexCoordPtr[i]->size;
             GLuint emit;
-            GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] & 
+            GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] &
                           ~TEXCOORDTYPE_MASK);
 
-	    switch (sz) {
-	    case 1: 
-	    case 2: 
-	       emit = EMIT_2F; 
-	       sz = 2; 
-	       mcs |= TEXCOORDTYPE_CARTESIAN; 
-	       break;
-	    case 3:
-	       emit = EMIT_3F; 
-	       sz = 3;
-	       mcs |= TEXCOORDTYPE_VECTOR;
-	       break;
-	    case 4: 
-	       emit = EMIT_3F_XYW; 
-	       sz = 3;     
-	       mcs |= TEXCOORDTYPE_HOMOGENEOUS;
-	       break;
-	    default: 
-	       continue;
-	    };
-	      
-
-	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, emit, 0 );	       
-	    v2 |= VRTX_TEX_SET_FMT(count, SZ_TO_HW(sz));
-	    mcsb1 |= (count+8)<<(i*4);
-
-	    if (mcs != i830->state.Tex[i][I830_TEXREG_MCS]) {
-	       I830_STATECHANGE(i830, I830_UPLOAD_TEX(i));
-	       i830->state.Tex[i][I830_TEXREG_MCS] = mcs;
-	    }
-
-	    count++;
-	 }
+            switch (sz) {
+            case 1:
+            case 2:
+               emit = EMIT_2F;
+               sz = 2;
+               mcs |= TEXCOORDTYPE_CARTESIAN;
+               break;
+            case 3:
+               emit = EMIT_3F;
+               sz = 3;
+               mcs |= TEXCOORDTYPE_VECTOR;
+               break;
+            case 4:
+               emit = EMIT_3F_XYW;
+               sz = 3;
+               mcs |= TEXCOORDTYPE_HOMOGENEOUS;
+               break;
+            default:
+               continue;
+            };
+
+
+            EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, emit, 0);
+            v2 |= VRTX_TEX_SET_FMT(count, SZ_TO_HW(sz));
+            mcsb1 |= (count + 8) << (i * 4);
+
+            if (mcs != i830->state.Tex[i][I830_TEXREG_MCS]) {
+               I830_STATECHANGE(i830, I830_UPLOAD_TEX(i));
+               i830->state.Tex[i][I830_TEXREG_MCS] = mcs;
+            }
+
+            count++;
+         }
       }
 
       v0 |= VFT0_TEX_COUNT(count);
    }
-   
+
    /* Only need to change the vertex emit code if there has been a
     * statechange to a new hardware vertex format:
     */
    if (v0 != i830->state.Ctx[I830_CTXREG_VF] ||
        v2 != i830->state.Ctx[I830_CTXREG_VF2] ||
        mcsb1 != i830->state.Ctx[I830_CTXREG_MCSB1] ||
-       !RENDERINPUTS_EQUAL( index_bitset, i830->last_index_bitset )) {
-    
-      I830_STATECHANGE( i830, I830_UPLOAD_CTX );
+       !RENDERINPUTS_EQUAL(index_bitset, i830->last_index_bitset)) {
+      int k;
+
+      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
       /* Must do this *after* statechange, so as not to affect
        * buffered vertices reliant on the old state:
        */
-      intel->vertex_size = 
-	 _tnl_install_attrs( ctx, 
-			     intel->vertex_attrs, 
-			     intel->vertex_attr_count,
-			     intel->ViewportMatrix.m, 0 );
+      intel->vertex_size =
+         _tnl_install_attrs(ctx,
+                            intel->vertex_attrs,
+                            intel->vertex_attr_count,
+                            intel->ViewportMatrix.m, 0);
 
       intel->vertex_size >>= 2;
 
       i830->state.Ctx[I830_CTXREG_VF] = v0;
       i830->state.Ctx[I830_CTXREG_VF2] = v2;
       i830->state.Ctx[I830_CTXREG_MCSB1] = mcsb1;
-      RENDERINPUTS_COPY( i830->last_index_bitset, index_bitset );
+      RENDERINPUTS_COPY(i830->last_index_bitset, index_bitset);
 
-      assert(i830_check_vertex_size( intel, intel->vertex_size ));
+      k = i830_check_vertex_size(intel, intel->vertex_size);
+      assert(k);
    }
 }
 
-static void i830_reduced_primitive_state( intelContextPtr intel,
-					  GLenum rprim )
+static void
+i830_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
 {
-    i830ContextPtr i830 = I830_CONTEXT(intel);
-    GLuint st1 = i830->state.Stipple[I830_STPREG_ST1];
-
-    st1 &= ~ST1_ENABLE;
-
-    switch (rprim) {
-    case GL_TRIANGLES:
-       if (intel->ctx.Polygon.StippleFlag &&
-	   intel->hw_stipple)
-	  st1 |= ST1_ENABLE;
-       break;
-    case GL_LINES:
-    case GL_POINTS:
-    default:
-       break;
-    }
-
-    i830->intel.reduced_primitive = rprim;
-
-    if (st1 != i830->state.Stipple[I830_STPREG_ST1]) {
-       I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-       i830->state.Stipple[I830_STPREG_ST1] = st1;
-    }
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLuint st1 = i830->state.Stipple[I830_STPREG_ST1];
+
+   st1 &= ~ST1_ENABLE;
+
+   switch (rprim) {
+   case GL_TRIANGLES:
+      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
+         st1 |= ST1_ENABLE;
+      break;
+   case GL_LINES:
+   case GL_POINTS:
+   default:
+      break;
+   }
+
+   i830->intel.reduced_primitive = rprim;
+
+   if (st1 != i830->state.Stipple[I830_STPREG_ST1]) {
+      INTEL_FIREVERTICES(intel);
+
+      I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
+      i830->state.Stipple[I830_STPREG_ST1] = st1;
+   }
 }
 
 /* Pull apart the vertex format registers and figure out how large a
  * vertex is supposed to be. 
  */
-static GLboolean i830_check_vertex_size( intelContextPtr intel,
-					 GLuint expected )
+static GLboolean
+i830_check_vertex_size(struct intel_context *intel, GLuint expected)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(&intel->ctx);
    int vft0 = i830->current->Ctx[I830_CTXREG_VF];
    int vft1 = i830->current->Ctx[I830_CTXREG_VF2];
    int nrtex = (vft0 & VFT0_TEX_COUNT_MASK) >> VFT0_TEX_COUNT_SHIFT;
    int i, sz = 0;
 
    switch (vft0 & VFT0_XYZW_MASK) {
-   case VFT0_XY: sz = 2; break;
-   case VFT0_XYZ: sz = 3; break;
-   case VFT0_XYW: sz = 3; break;
-   case VFT0_XYZW: sz = 4; break;
-   default: 
+   case VFT0_XY:
+      sz = 2;
+      break;
+   case VFT0_XYZ:
+      sz = 3;
+      break;
+   case VFT0_XYW:
+      sz = 3;
+      break;
+   case VFT0_XYZW:
+      sz = 4;
+      break;
+   default:
       fprintf(stderr, "no xyzw specified\n");
       return 0;
    }
 
-   if (vft0 & VFT0_SPEC) sz++;
-   if (vft0 & VFT0_DIFFUSE) sz++;
-   if (vft0 & VFT0_DEPTH_OFFSET) sz++;
-   if (vft0 & VFT0_POINT_WIDTH) sz++;
-	
-   for (i = 0 ; i < nrtex ; i++) { 
+   if (vft0 & VFT0_SPEC)
+      sz++;
+   if (vft0 & VFT0_DIFFUSE)
+      sz++;
+   if (vft0 & VFT0_DEPTH_OFFSET)
+      sz++;
+   if (vft0 & VFT0_POINT_WIDTH)
+      sz++;
+
+   for (i = 0; i < nrtex; i++) {
       switch (vft1 & VFT1_TEX0_MASK) {
-      case TEXCOORDFMT_2D: sz += 2; break;
-      case TEXCOORDFMT_3D: sz += 3; break;
-      case TEXCOORDFMT_4D: sz += 4; break;
-      case TEXCOORDFMT_1D: sz += 1; break;
+      case TEXCOORDFMT_2D:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_3D:
+         sz += 3;
+         break;
+      case TEXCOORDFMT_4D:
+         sz += 4;
+         break;
+      case TEXCOORDFMT_1D:
+         sz += 1;
+         break;
       }
       vft1 >>= VFT1_TEX1_SHIFT;
    }
-	
-   if (sz != expected) 
+
+   if (sz != expected)
       fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-   
+
    return sz == expected;
 }
 
-static void i830_emit_invarient_state( intelContextPtr intel )
+static void
+i830_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH( 40 );
-
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(0));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(1));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(2));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(3));
+   BEGIN_BATCH(40, IGNORE_CLIPRECTS);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -282,37 +310,35 @@ static void i830_emit_invarient_state( intelContextPtr intel )
 
    OUT_BATCH(_3DSTATE_FOG_MODE_CMD);
    OUT_BATCH(FOGFUNC_ENABLE |
-	     FOG_LINEAR_CONST | 
-	     FOGSRC_INDEX_Z | 
-	     ENABLE_FOG_DENSITY);
+             FOG_LINEAR_CONST | FOGSRC_INDEX_Z | ENABLE_FOG_DENSITY);
    OUT_BATCH(0);
    OUT_BATCH(0);
 
 
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(0) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(0) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(0));
+             MAP_UNIT(0) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(0) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(0));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(1) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(1) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(1));
+             MAP_UNIT(1) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(1) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(1));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(2) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(2) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(2));
+             MAP_UNIT(2) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(2) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(2));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(3) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(3) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(3));
+             MAP_UNIT(3) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(3) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(3));
 
    OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
    OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(0));
@@ -324,21 +350,13 @@ static void i830_emit_invarient_state( intelContextPtr intel )
    OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(3));
 
    OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-	     ENABLE_POINT_RASTER_RULE |
-	     OGL_POINT_RASTER_RULE |
-	     ENABLE_LINE_STRIP_PROVOKE_VRTX |
-	     ENABLE_TRI_FAN_PROVOKE_VRTX |
-	     ENABLE_TRI_STRIP_PROVOKE_VRTX |
-	     LINE_STRIP_PROVOKE_VRTX(1) |
-	     TRI_FAN_PROVOKE_VRTX(2) | 
-	     TRI_STRIP_PROVOKE_VRTX(2));
-
-   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | 
-	     DISABLE_SCISSOR_RECT);
-
-   OUT_BATCH(_3DSTATE_SCISSOR_RECT_0_CMD);
-   OUT_BATCH(0);
-   OUT_BATCH(0);
+             ENABLE_POINT_RASTER_RULE |
+             OGL_POINT_RASTER_RULE |
+             ENABLE_LINE_STRIP_PROVOKE_VRTX |
+             ENABLE_TRI_FAN_PROVOKE_VRTX |
+             ENABLE_TRI_STRIP_PROVOKE_VRTX |
+             LINE_STRIP_PROVOKE_VRTX(1) |
+             TRI_FAN_PROVOKE_VRTX(2) | TRI_STRIP_PROVOKE_VRTX(2));
 
    OUT_BATCH(_3DSTATE_VERTEX_TRANSFORM);
    OUT_BATCH(DISABLE_VIEWPORT_TRANSFORM | DISABLE_PERSPECTIVE_DIVIDE);
@@ -349,45 +367,46 @@ static void i830_emit_invarient_state( intelContextPtr intel )
 
 
    OUT_BATCH(_3DSTATE_COLOR_FACTOR_CMD);
-   OUT_BATCH(0x80808080);	/* .5 required in alpha for GL_DOT3_RGBA_EXT */
+   OUT_BATCH(0x80808080);       /* .5 required in alpha for GL_DOT3_RGBA_EXT */
 
    ADVANCE_BATCH();
 }
 
 
 #define emit( intel, state, size )			\
-do {							\
-   int k;						\
-   BEGIN_BATCH( size / sizeof(GLuint));			\
-   for (k = 0 ; k < size / sizeof(GLuint) ; k++)	\
-      OUT_BATCH(state[k]);				\
-   ADVANCE_BATCH();					\
-} while (0);
-
-static GLuint get_state_size( struct i830_hw_state *state )
+   intel_batchbuffer_data(intel->batch, state, size, IGNORE_CLIPRECTS )
+
+static GLuint
+get_dirty(struct i830_hw_state *state)
+{
+   return state->active & ~state->emitted;
+}
+
+static GLuint
+get_state_size(struct i830_hw_state *state)
 {
-   GLuint dirty = state->active & ~state->emitted;
+   GLuint dirty = get_dirty(state);
    GLuint sz = 0;
    GLuint i;
 
    if (dirty & I830_UPLOAD_INVARIENT)
       sz += 40 * sizeof(int);
 
-   if (dirty & I830_UPLOAD_CTX) 
+   if (dirty & I830_UPLOAD_CTX)
       sz += sizeof(state->Ctx);
 
-   if (dirty & I830_UPLOAD_BUFFERS) 
+   if (dirty & I830_UPLOAD_BUFFERS)
       sz += sizeof(state->Buffer);
 
-   if (dirty & I830_UPLOAD_STIPPLE) 
+   if (dirty & I830_UPLOAD_STIPPLE)
       sz += sizeof(state->Stipple);
 
    for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i)))  
-	 sz += sizeof(state->Tex[i]); 
+      if ((dirty & I830_UPLOAD_TEX(i)))
+         sz += sizeof(state->Tex[i]);
 
-      if (dirty & I830_UPLOAD_TEXBLEND(i)) 
-	 sz += state->TexBlendWordsUsed[i] * 4;
+      if (dirty & I830_UPLOAD_TEXBLEND(i))
+         sz += state->TexBlendWordsUsed[i] * 4;
    }
 
    return sz;
@@ -396,139 +415,349 @@ static GLuint get_state_size( struct i830_hw_state *state )
 
 /* Push the state into the sarea and/or texture memory.
  */
-static void i830_emit_state( intelContextPtr intel )
+static void
+i830_emit_state(struct intel_context *intel)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(&intel->ctx);
    struct i830_hw_state *state = i830->current;
-   int i;
-   GLuint dirty = state->active & ~state->emitted;
-   GLuint counter = intel->batch.counter;
+   int i, count;
+   GLuint dirty;
+   GET_CURRENT_CONTEXT(ctx);
    BATCH_LOCALS;
+   dri_bo *aper_array[3 + I830_TEX_UNITS];
+   int aper_count;
+
+   /* We don't hold the lock at this point, so want to make sure that
+    * there won't be a buffer wrap between the state emits and the primitive
+    * emit header.
+    *
+    * It might be better to talk about explicit places where
+    * scheduling is allowed, rather than assume that it is whenever a
+    * batchbuffer fills up.
+    *
+    * Set the space as LOOP_CLIPRECTS now, since that's what our primitives
+    * will be emitted under.
+    */
+   intel_batchbuffer_require_space(intel->batch,
+				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE,
+				   LOOP_CLIPRECTS);
+   count = 0;
+ again:
+   aper_count = 0;
+   dirty = get_dirty(state);
+
+   aper_array[aper_count++] = intel->batch->buf;
+   if (dirty & I830_UPLOAD_BUFFERS) {
+      aper_array[aper_count++] = state->draw_region->buffer;
+      if (state->depth_region)
+         aper_array[aper_count++] = state->depth_region->buffer;
+   }
 
-   if (intel->batch.space < get_state_size(state)) {
-      intelFlushBatch(intel, GL_TRUE);
-      dirty = state->active & ~state->emitted;
-      counter = intel->batch.counter;
+   for (i = 0; i < I830_TEX_UNITS; i++)
+     if (dirty & I830_UPLOAD_TEX(i)) {
+	if (state->tex_buffer[i]) {
+	   aper_array[aper_count++] = state->tex_buffer[i];
+	}
+     }
+
+   if (dri_bufmgr_check_aperture_space(aper_array, aper_count)) {
+       if (count == 0) {
+	   count++;
+	   intel_batchbuffer_flush(intel->batch);
+	   goto again;
+       } else {
+	   _mesa_error(ctx, GL_OUT_OF_MEMORY, "i830 emit state");
+	   assert(0);
+       }
    }
 
+
+   /* Do this here as we may have flushed the batchbuffer above,
+    * causing more state to be dirty!
+    */
+   dirty = get_dirty(state);
+   state->emitted |= dirty;
+   assert(get_dirty(state) == 0);
+
    if (dirty & I830_UPLOAD_INVARIENT) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_INVARIENT:\n"); 
-      i830_emit_invarient_state( intel );
+      DBG("I830_UPLOAD_INVARIENT:\n");
+      i830_emit_invarient_state(intel);
    }
 
    if (dirty & I830_UPLOAD_CTX) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_CTX:\n"); 
-      emit( i830, state->Ctx, sizeof(state->Ctx) );
+      DBG("I830_UPLOAD_CTX:\n");
+      emit(intel, state->Ctx, sizeof(state->Ctx));
+
    }
 
    if (dirty & I830_UPLOAD_BUFFERS) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_BUFFERS:\n"); 
-      emit( i830, state->Buffer, sizeof(state->Buffer) );
-   }
+      DBG("I830_UPLOAD_BUFFERS:\n");
+      BEGIN_BATCH(I830_DEST_SETUP_SIZE + 2, IGNORE_CLIPRECTS);
+      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR1]);
+      OUT_RELOC(state->draw_region->buffer,
+		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+                state->draw_region->draw_offset);
+
+      if (state->depth_region) {
+         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR0]);
+         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR1]);
+         OUT_RELOC(state->depth_region->buffer,
+		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+                   state->depth_region->draw_offset);
+      }
 
+      OUT_BATCH(state->Buffer[I830_DESTREG_DV0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DV1]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SENABLE]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR1]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR2]);
+
+      if (intel->constant_cliprect) {
+	 assert(state->Buffer[I830_DESTREG_DRAWRECT0] != MI_NOOP);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT0]);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT1]);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT2]);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT3]);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT4]);
+	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT5]);
+      }
+      ADVANCE_BATCH();
+   }
+   
    if (dirty & I830_UPLOAD_STIPPLE) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_STIPPLE:\n"); 
-      emit( i830, state->Stipple, sizeof(state->Stipple) );
+      DBG("I830_UPLOAD_STIPPLE:\n");
+      emit(intel, state->Stipple, sizeof(state->Stipple));
    }
 
    for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i))) { 
- 	 if (VERBOSE) fprintf(stderr, "I830_UPLOAD_TEX(%d):\n", i); 
-	 emit( i830, state->Tex[i], sizeof(state->Tex[i])); 
-      } 
+      if ((dirty & I830_UPLOAD_TEX(i))) {
+         DBG("I830_UPLOAD_TEX(%d):\n", i);
+
+         BEGIN_BATCH(I830_TEX_SETUP_SIZE + 1, IGNORE_CLIPRECTS);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0LI]);
+
+         if (state->tex_buffer[i]) {
+            OUT_RELOC(state->tex_buffer[i],
+		      I915_GEM_DOMAIN_SAMPLER, 0,
+                      state->tex_offset[i] | TM0S0_USE_FENCE);
+         }
+	 else if (state == &i830->meta) {
+	    assert(i == 0);
+	    OUT_BATCH(0);
+	 }
+	 else {
+	    OUT_BATCH(state->tex_offset[i]);
+	 }
+
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S1]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S2]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S3]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S4]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_MCS]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_CUBE]);
+      }
 
       if (dirty & I830_UPLOAD_TEXBLEND(i)) {
-	 if (VERBOSE) fprintf(stderr, "I830_UPLOAD_TEXBLEND(%d):\n", i); 
-	 emit( i830, state->TexBlend[i], 
-	       state->TexBlendWordsUsed[i] * 4 );
+         DBG("I830_UPLOAD_TEXBLEND(%d): %d words\n", i,
+             state->TexBlendWordsUsed[i]);
+         emit(intel, state->TexBlend[i], state->TexBlendWordsUsed[i] * 4);
       }
    }
 
-   state->emitted |= dirty;
-   intel->batch.last_emit_state = counter;
-   assert(counter == intel->batch.counter);
+   intel->batch->dirty_state &= ~dirty;
+   assert(get_dirty(state) == 0);
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
 }
 
-static void i830_destroy_context( intelContextPtr intel )
+static void
+i830_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   intel_region_release(&i830->state.draw_region);
+   intel_region_release(&i830->state.depth_region);
+   intel_region_release(&i830->meta.draw_region);
+   intel_region_release(&i830->meta.depth_region);
+   intel_region_release(&i830->initial.draw_region);
+   intel_region_release(&i830->initial.depth_region);
+
+   for (i = 0; i < I830_TEX_UNITS; i++) {
+      if (i830->state.tex_buffer[i] != NULL) {
+	 dri_bo_unreference(i830->state.tex_buffer[i]);
+	 i830->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
-static void
-i830_set_color_region(intelContextPtr intel, const intelRegion *region)
+
+void
+i830_state_draw_region(struct intel_context *intel,
+		       struct i830_hw_state *state,
+		       struct intel_region *color_region,
+		       struct intel_region *depth_region)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   I830_STATECHANGE( i830, I830_UPLOAD_BUFFERS );
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_CBUFADDR2] = region->offset;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLcontext *ctx = &intel->ctx;
+   GLuint value;
+
+   ASSERT(state == &i830->state || state == &i830->meta);
+
+   if (state->draw_region != color_region) {
+      intel_region_release(&state->draw_region);
+      intel_region_reference(&state->draw_region, color_region);
+   }
+   if (state->depth_region != depth_region) {
+      intel_region_release(&state->depth_region);
+      intel_region_reference(&state->depth_region, depth_region);
+   }
+
+   /*
+    * Set stride/cpp values
+    */
+   if (color_region) {
+      state->Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I830_DESTREG_CBUFADDR1] =
+         (BUF_3D_ID_COLOR_BACK |
+          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   if (depth_region) {
+      state->Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I830_DESTREG_DBUFADDR1] =
+         (BUF_3D_ID_DEPTH |
+          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   /*
+    * Compute/set I830_DESTREG_DV1 value
+    */
+   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
+            DSTORG_VERT_BIAS(0x8) | DEPTH_IS_Z);    /* .5 */
+            
+   if (color_region && color_region->cpp == 4) {
+      value |= DV_PF_8888;
+   }
+   else {
+      value |= DV_PF_565;
+   }
+   if (depth_region && depth_region->cpp == 4) {
+      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
+   }
+   else {
+      value |= DEPTH_FRMT_16_FIXED;
+   }
+   state->Buffer[I830_DESTREG_DV1] = value;
+
+   if (intel->constant_cliprect) {
+      state->Buffer[I830_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
+      state->Buffer[I830_DESTREG_DRAWRECT1] = 0;
+      state->Buffer[I830_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
+      state->Buffer[I830_DESTREG_DRAWRECT3] =
+	 (ctx->DrawBuffer->Width & 0xffff) |
+	 (ctx->DrawBuffer->Height << 16);
+      state->Buffer[I830_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
+      state->Buffer[I830_DESTREG_DRAWRECT5] = 0;
+   } else {
+      state->Buffer[I830_DESTREG_DRAWRECT0] = MI_NOOP;
+      state->Buffer[I830_DESTREG_DRAWRECT1] = MI_NOOP;
+      state->Buffer[I830_DESTREG_DRAWRECT2] = MI_NOOP;
+      state->Buffer[I830_DESTREG_DRAWRECT3] = MI_NOOP;
+      state->Buffer[I830_DESTREG_DRAWRECT4] = MI_NOOP;
+      state->Buffer[I830_DESTREG_DRAWRECT5] = MI_NOOP;
+   }
+
+   I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
+
+
 }
 
 
 static void
-i830_set_z_region(intelContextPtr intel, const intelRegion *region)
+i830_set_draw_region(struct intel_context *intel,
+                     struct intel_region *color_regions[],
+                     struct intel_region *depth_region,
+		     GLuint num_regions)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   I830_STATECHANGE( i830, I830_UPLOAD_BUFFERS );
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_DBUFADDR2] = region->offset;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830_state_draw_region(intel, &i830->state, color_regions[0], depth_region);
 }
 
-
+#if 0
 static void
 i830_update_color_z_regions(intelContextPtr intel,
-                            const intelRegion *colorRegion,
-                            const intelRegion *depthRegion)
+                            const intelRegion * colorRegion,
+                            const intelRegion * depthRegion)
 {
    i830ContextPtr i830 = I830_CONTEXT(intel);
 
    i830->state.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) | BUF_3D_USE_FENCE);
+      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) |
+       BUF_3D_USE_FENCE);
    i830->state.Buffer[I830_DESTREG_CBUFADDR2] = colorRegion->offset;
 
    i830->state.Buffer[I830_DESTREG_DBUFADDR1] =
       (BUF_3D_ID_DEPTH | BUF_3D_PITCH(depthRegion->pitch) | BUF_3D_USE_FENCE);
    i830->state.Buffer[I830_DESTREG_DBUFADDR2] = depthRegion->offset;
 }
+#endif
 
 
 /* This isn't really handled at the moment.
  */
-static void i830_lost_hardware( intelContextPtr intel )
+static void
+i830_new_batch(struct intel_context *intel)
 {
-   I830_CONTEXT(intel)->state.emitted = 0;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830->state.emitted = 0;
+
+   /* Check that we didn't just wrap our batchbuffer at a bad time. */
+   assert(!intel->no_batch_wrap);
 }
 
 
 
-static void i830_emit_flush( intelContextPtr intel )
+static GLuint
+i830_flush_cmd(void)
 {
-   BATCH_LOCALS;
-
-   BEGIN_BATCH(2);
-   OUT_BATCH( MI_FLUSH | FLUSH_MAP_CACHE ); 
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
+   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
 
+static void 
+i830_assert_not_dirty( struct intel_context *intel )
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   struct i830_hw_state *state = i830->current;
+   assert(!get_dirty(state));
+}
 
+static void
+i830_note_unlock( struct intel_context *intel )
+{
+    /* nothing */
+}
 
-void i830InitVtbl( i830ContextPtr i830 )
+void
+i830InitVtbl(struct i830_context *i830)
 {
-   i830->intel.vtbl.alloc_tex_obj = i830AllocTexObj;
    i830->intel.vtbl.check_vertex_size = i830_check_vertex_size;
-   i830->intel.vtbl.clear_with_tris = i830ClearWithTris;
-   i830->intel.vtbl.rotate_window = i830RotateWindow;
    i830->intel.vtbl.destroy = i830_destroy_context;
    i830->intel.vtbl.emit_state = i830_emit_state;
-   i830->intel.vtbl.lost_hardware = i830_lost_hardware;
+   i830->intel.vtbl.new_batch = i830_new_batch;
    i830->intel.vtbl.reduced_primitive_state = i830_reduced_primitive_state;
-   i830->intel.vtbl.set_color_region = i830_set_color_region;
-   i830->intel.vtbl.set_z_region = i830_set_z_region;
-   i830->intel.vtbl.update_color_z_regions = i830_update_color_z_regions;
+   i830->intel.vtbl.set_draw_region = i830_set_draw_region;
    i830->intel.vtbl.update_texture_state = i830UpdateTextureState;
-   i830->intel.vtbl.emit_flush = i830_emit_flush;
+   i830->intel.vtbl.flush_cmd = i830_flush_cmd;
    i830->intel.vtbl.render_start = i830_render_start;
+   i830->intel.vtbl.render_prevalidate = i830_render_prevalidate;
+   i830->intel.vtbl.assert_not_dirty = i830_assert_not_dirty;
+   i830->intel.vtbl.note_unlock = i830_note_unlock; 
+   i830->intel.vtbl.finish_batch = intel_finish_vb;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index 2bc1cae9c31..e0ddc7fd613 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -26,7 +26,7 @@
  **************************************************************************/
 
 #include "i915_context.h"
-#include "imports.h"
+#include "main/imports.h"
 #include "intel_tex.h"
 #include "intel_tris.h"
 #include "tnl/t_context.h"
@@ -36,151 +36,151 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "vbo/vbo.h"
-
 
 #include "utils.h"
 #include "i915_reg.h"
 
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
+#include "intel_tris.h"
+#include "intel_span.h"
+#include "intel_pixel.h"
+
 /***************************************
  * Mesa's Driver Functions
  ***************************************/
 
-static const struct dri_extension i915_extensions[] =
-{
-    { "GL_ARB_depth_texture",              NULL },
-    { "GL_ARB_fragment_program",           NULL },
-    { "GL_ARB_shadow",                     NULL },
-    { "GL_ARB_texture_env_crossbar",       NULL },
-    { "GL_EXT_shadow_funcs",               NULL },
-    /* ARB extn won't work if not enabled */
-    { "GL_SGIX_depth_texture",             NULL },
-    { NULL,                                NULL }
+static const struct dri_extension i915_extensions[] = {
+   {"GL_ARB_depth_texture", NULL},
+   {"GL_ARB_fragment_program", NULL},
+   {"GL_ARB_shadow", NULL},
+   {"GL_ARB_texture_non_power_of_two", NULL},
+   {"GL_EXT_shadow_funcs", NULL},
+   {NULL, NULL}
 };
 
 /* Override intel default.
  */
-static void i915InvalidateState( GLcontext *ctx, GLuint new_state )
+static void
+i915InvalidateState(GLcontext * ctx, GLuint new_state)
 {
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   _tnl_invalidate_vertex_state( ctx, new_state );
-   INTEL_CONTEXT(ctx)->NewGLState |= new_state;
+   _swrast_InvalidateState(ctx, new_state);
+   _swsetup_InvalidateState(ctx, new_state);
+   _vbo_InvalidateState(ctx, new_state);
+   _tnl_InvalidateState(ctx, new_state);
+   _tnl_invalidate_vertex_state(ctx, new_state);
+   intel_context(ctx)->NewGLState |= new_state;
 
    /* Todo: gather state values under which tracked parameters become
     * invalidated, add callbacks for things like
     * ProgramLocalParameters, etc.
     */
    {
-      struct i915_fragment_program *p = 
-	 (struct i915_fragment_program *)ctx->FragmentProgram._Current;
+      struct i915_fragment_program *p =
+         (struct i915_fragment_program *) ctx->FragmentProgram._Current;
       if (p && p->nr_params)
-	 p->params_uptodate = 0;
+         p->params_uptodate = 0;
    }
 
-   if (new_state & (_NEW_FOG|_NEW_HINT|_NEW_PROGRAM))
+   if (new_state & (_NEW_FOG | _NEW_HINT | _NEW_PROGRAM))
       i915_update_fog(ctx);
 }
 
 
-static void i915InitDriverFunctions( struct dd_function_table *functions )
+static void
+i915InitDriverFunctions(struct dd_function_table *functions)
 {
-   intelInitDriverFunctions( functions );
-   i915InitStateFunctions( functions );
-   i915InitTextureFuncs( functions );
-   i915InitFragProgFuncs( functions );
+   intelInitDriverFunctions(functions);
+   i915InitStateFunctions(functions);
+   i915InitTextureFuncs(functions);
+   i915InitFragProgFuncs(functions);
    functions->UpdateState = i915InvalidateState;
 }
 
 
+extern const struct tnl_pipeline_stage *intel_pipeline[];
 
-GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate)
+GLboolean
+i915CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate)
 {
    struct dd_function_table functions;
-   i915ContextPtr i915 = (i915ContextPtr) CALLOC_STRUCT(i915_context);
-   intelContextPtr intel = &i915->intel;
+   struct i915_context *i915 =
+      (struct i915_context *) CALLOC_STRUCT(i915_context);
+   struct intel_context *intel = &i915->intel;
    GLcontext *ctx = &intel->ctx;
-   GLuint i;
 
-   if (!i915) return GL_FALSE;
+   if (!i915)
+      return GL_FALSE;
+
+   if (0)
+      _mesa_printf("\ntexmem-0-3 branch\n\n");
 
-   i915InitVtbl( i915 );
+   i915InitVtbl(i915);
+   i915InitMetaFuncs(i915);
 
-   i915InitDriverFunctions( &functions );
+   i915InitDriverFunctions(&functions);
 
-   if (!intelInitContext( intel, mesaVis, driContextPriv,
-			  sharedContextPrivate, &functions )) {
+   if (!intelInitContext(intel, mesaVis, driContextPriv,
+                         sharedContextPrivate, &functions)) {
       FREE(i915);
       return GL_FALSE;
    }
 
+   /* Initialize swrast, tnl driver tables: */
+   intelInitSpanFuncs(ctx);
+   intelInitTriFuncs(ctx);
+
+   /* Install the customized pipeline: */
+   _tnl_destroy_pipeline(ctx);
+   _tnl_install_pipeline(ctx, intel_pipeline);
+
+   if (intel->no_rast)
+      FALLBACK(intel, INTEL_FALLBACK_USER, 1);
+
    ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS;
 
-   intel->nr_heaps = 1;
-   intel->texture_heaps[0] = 
-      driCreateTextureHeap( 0, intel,
-			    intel->intelScreen->tex.size,
-			    12,
-			    I830_NR_TEX_REGIONS,
-			    intel->sarea->texList,
-			    (unsigned *) & intel->sarea->texAge,
-			    & intel->swapped,
-			    sizeof( struct i915_texture_object ),
-			    (destroy_texture_object_t *)intelDestroyTexObj );
-
-   /* FIXME: driCalculateMaxTextureLevels assumes that mipmaps are
-    * tightly packed, but they're not in Intel graphics
-    * hardware.
+
+   /* Advertise the full hardware capabilities.  The new memory
+    * manager should cope much better with overload situations:
     */
+   ctx->Const.MaxTextureLevels = 12;
+   ctx->Const.Max3DTextureLevels = 9;
+   ctx->Const.MaxCubeTextureLevels = 12;
+   ctx->Const.MaxTextureRectSize = (1 << 11);
    ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
-   i = driQueryOptioni( &intel->optionCache, "allow_large_textures");
-   driCalculateMaxTextureLevels( intel->texture_heaps,
-				 intel->nr_heaps,
-				 &intel->ctx.Const,
-				 4,
-				 11, /* max 2D texture size is 2048x2048 */
-				 8,  /* 3D texture */
-				 11, /* cube texture. */
-				 11, /* rect texture */
-				 12,
-				 GL_FALSE,
-				 i );
 
    /* GL_ARB_fragment_program limits - don't think Mesa actually
     * validates programs against these, and in any case one ARB
     * instruction can translate to more than one HW instruction, so
     * we'll still have to check and fallback each time.
     */
-   
    ctx->Const.FragmentProgram.MaxNativeTemps = I915_MAX_TEMPORARY;
-   ctx->Const.FragmentProgram.MaxNativeAttribs = 11; /* 8 tex, 2 color, fog */
+   ctx->Const.FragmentProgram.MaxNativeAttribs = 11;    /* 8 tex, 2 color, fog */
    ctx->Const.FragmentProgram.MaxNativeParameters = I915_MAX_CONSTANT;
    ctx->Const.FragmentProgram.MaxNativeAluInstructions = I915_MAX_ALU_INSN;
    ctx->Const.FragmentProgram.MaxNativeTexInstructions = I915_MAX_TEX_INSN;
-   ctx->Const.FragmentProgram.MaxNativeInstructions = (I915_MAX_ALU_INSN + 
-						I915_MAX_TEX_INSN);
-   ctx->Const.FragmentProgram.MaxNativeTexIndirections = I915_MAX_TEX_INDIRECT;
+   ctx->Const.FragmentProgram.MaxNativeInstructions = (I915_MAX_ALU_INSN +
+                                                       I915_MAX_TEX_INSN);
+   ctx->Const.FragmentProgram.MaxNativeTexIndirections =
+      I915_MAX_TEX_INDIRECT;
    ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0; /* I don't think we have one */
+
    ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
    ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
 
-
-   driInitExtensions( ctx, i915_extensions, GL_FALSE );
+   driInitExtensions(ctx, i915_extensions, GL_FALSE);
 
 
-   _tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-		       36 * sizeof(GLfloat) );
+   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
+                      36 * sizeof(GLfloat));
 
    intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
 
-   i915InitState( i915 );
+   i915InitState(i915);
 
    return GL_TRUE;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i915_context.h b/src/mesa/drivers/dri/i915/i915_context.h
index ec1550126a6..87bbf5f9271 100644
--- a/src/mesa/drivers/dri/i915/i915_context.h
+++ b/src/mesa/drivers/dri/i915/i915_context.h
@@ -29,6 +29,7 @@
 #define I915CONTEXT_INC
 
 #include "intel_context.h"
+#include "i915_reg.h"
 
 #define I915_FALLBACK_TEXTURE		 0x1000
 #define I915_FALLBACK_COLORMASK		 0x2000
@@ -46,6 +47,7 @@
 #define I915_UPLOAD_CONSTANTS        0x10
 #define I915_UPLOAD_FOG              0x20
 #define I915_UPLOAD_INVARIENT        0x40
+#define I915_UPLOAD_DEFAULTS         0x80
 #define I915_UPLOAD_TEX(i)           (0x00010000<<(i))
 #define I915_UPLOAD_TEX_ALL          (0x00ff0000)
 #define I915_UPLOAD_TEX_0_SHIFT      16
@@ -55,17 +57,21 @@
  */
 #define I915_DESTREG_CBUFADDR0 0
 #define I915_DESTREG_CBUFADDR1 1
-#define I915_DESTREG_CBUFADDR2 2
 #define I915_DESTREG_DBUFADDR0 3
 #define I915_DESTREG_DBUFADDR1 4
-#define I915_DESTREG_DBUFADDR2 5
 #define I915_DESTREG_DV0 6
 #define I915_DESTREG_DV1 7
 #define I915_DESTREG_SENABLE 8
 #define I915_DESTREG_SR0 9
 #define I915_DESTREG_SR1 10
 #define I915_DESTREG_SR2 11
-#define I915_DEST_SETUP_SIZE 12
+#define I915_DESTREG_DRAWRECT0 12
+#define I915_DESTREG_DRAWRECT1 13
+#define I915_DESTREG_DRAWRECT2 14
+#define I915_DESTREG_DRAWRECT3 15
+#define I915_DESTREG_DRAWRECT4 16
+#define I915_DESTREG_DRAWRECT5 17
+#define I915_DEST_SETUP_SIZE 18
 
 #define I915_CTXREG_STATE4		0
 #define I915_CTXREG_LI	        	1
@@ -89,7 +95,6 @@
 #define I915_STPREG_ST1        1
 #define I915_STP_SETUP_SIZE    2
 
-#define I915_TEXREG_MS2        0
 #define I915_TEXREG_MS3        1
 #define I915_TEXREG_MS4        2
 #define I915_TEXREG_SS2        3
@@ -97,24 +102,38 @@
 #define I915_TEXREG_SS4        5
 #define I915_TEX_SETUP_SIZE    6
 
+#define I915_DEFREG_C0    0
+#define I915_DEFREG_C1    1
+#define I915_DEFREG_S0    2
+#define I915_DEFREG_S1    3
+#define I915_DEFREG_Z0    4
+#define I915_DEFREG_Z1    5
+#define I915_DEF_SETUP_SIZE    6
+
+
 #define I915_MAX_CONSTANT      32
 #define I915_CONSTANT_SIZE     (2+(4*I915_MAX_CONSTANT))
 
 
 #define I915_PROGRAM_SIZE      192
 
+#define I915_MAX_INSN          (I915_MAX_TEX_INSN+I915_MAX_ALU_INSN)
 
 /* Hardware version of a parsed fragment program.  "Derived" from the
  * mesa fragment_program struct.
  */
-struct i915_fragment_program {
+struct i915_fragment_program
+{
    struct gl_fragment_program FragProg;
 
    GLboolean translated;
    GLboolean params_uptodate;
    GLboolean on_hardware;
-   GLboolean error;		/* If program is malformed for any reason. */
+   GLboolean error;             /* If program is malformed for any reason. */
 
+   /** Record of which phases R registers were last written in. */
+   GLuint register_phases[16];
+   GLuint indirections;
    GLuint nr_tex_indirect;
    GLuint nr_tex_insn;
    GLuint nr_alu_insn;
@@ -135,52 +154,40 @@ struct i915_fragment_program {
    GLuint constant_flags[I915_MAX_CONSTANT];
    GLuint nr_constants;
 
-   GLuint *csr;			/* Cursor, points into program.
-				 */
+   GLuint *csr;                 /* Cursor, points into program.
+                                 */
 
-   GLuint *decl;		/* Cursor, points into declarations.
-				 */
-   
-   GLuint decl_s;		/* flags for which s regs need to be decl'd */
-   GLuint decl_t;		/* flags for which t regs need to be decl'd */
+   GLuint *decl;                /* Cursor, points into declarations.
+                                 */
 
-   GLuint temp_flag;		/* Tracks temporary regs which are in
-				 * use.
-				 */
+   GLuint decl_s;               /* flags for which s regs need to be decl'd */
+   GLuint decl_t;               /* flags for which t regs need to be decl'd */
 
-   GLuint utemp_flag;		/* Tracks TYPE_U temporary regs which are in
-				 * use.
-				 */
+   GLuint temp_flag;            /* Tracks temporary regs which are in
+                                 * use.
+                                 */
 
+   GLuint utemp_flag;           /* Tracks TYPE_U temporary regs which are in
+                                 * use.
+                                 */
 
 
+   /* Track which R registers are "live" for each instruction.
+    * A register is live between the time it's written to and the last time
+    * it's read. */
+   GLuint usedRegs[I915_MAX_INSN];
+
    /* Helpers for i915_fragprog.c:
     */
    GLuint wpos_tex;
    GLboolean depth_written;
 
-   struct { 
-      GLuint reg;		/* Hardware constant idx */
-      const GLfloat *values; 	/* Pointer to tracked values */
+   struct
+   {
+      GLuint reg;               /* Hardware constant idx */
+      const GLfloat *values;    /* Pointer to tracked values */
    } param[I915_MAX_CONSTANT];
    GLuint nr_params;
-      
-
-
-
-   /* Helpers for i915_texprog.c:
-    */
-   GLuint src_texture;		/* Reg containing sampled texture color,
-				 * else UREG_BAD.
-				 */
-
-   GLuint src_previous;		/* Reg containing color from previous 
-				 * stage.  May need to be decl'd.
-				 */
-
-   GLuint last_tex_stage;	/* Number of last enabled texture unit */
-
-   struct vertex_buffer *VB;
 };
 
 
@@ -188,67 +195,68 @@ struct i915_fragment_program {
 
 
 
-struct i915_texture_object
-{
-   struct intel_texture_object intel;
-   GLenum lastTarget;
-   GLboolean refs_border_color;
-   GLuint Setup[I915_TEX_SETUP_SIZE];
-};
 
 #define I915_TEX_UNITS 8
 
 
-struct i915_hw_state {
+struct i915_hw_state
+{
    GLuint Ctx[I915_CTX_SETUP_SIZE];
    GLuint Buffer[I915_DEST_SETUP_SIZE];
    GLuint Stipple[I915_STP_SETUP_SIZE];
    GLuint Fog[I915_FOG_SETUP_SIZE];
+   GLuint Defaults[I915_DEF_SETUP_SIZE];
    GLuint Tex[I915_TEX_UNITS][I915_TEX_SETUP_SIZE];
    GLuint Constant[I915_CONSTANT_SIZE];
    GLuint ConstantSize;
    GLuint Program[I915_PROGRAM_SIZE];
    GLuint ProgramSize;
-   GLuint active;		/* I915_UPLOAD_* */
-   GLuint emitted;		/* I915_UPLOAD_* */
+
+   /* Region pointers for relocation: 
+    */
+   struct intel_region *draw_region;
+   struct intel_region *depth_region;
+/*    struct intel_region *tex_region[I915_TEX_UNITS]; */
+
+   /* Regions aren't actually that appropriate here as the memory may
+    * be from a PBO or FBO.  Will have to do this for draw and depth for
+    * FBO's...
+    */
+   dri_bo *tex_buffer[I915_TEX_UNITS];
+   GLuint tex_offset[I915_TEX_UNITS];
+
+
+   GLuint active;               /* I915_UPLOAD_* */
+   GLuint emitted;              /* I915_UPLOAD_* */
 };
 
 #define I915_FOG_PIXEL  2
 #define I915_FOG_VERTEX 1
 #define I915_FOG_NONE   0
 
-struct i915_context 
+struct i915_context
 {
    struct intel_context intel;
 
    GLuint last_ReallyEnabled;
    GLuint vertex_fog;
+   GLuint lodbias_ss2[MAX_TEXTURE_UNITS];
+
 
-   struct i915_fragment_program tex_program;
    struct i915_fragment_program *current_program;
 
    struct i915_hw_state meta, initial, state, *current;
 };
 
 
-typedef struct i915_context *i915ContextPtr;
-typedef struct i915_texture_object *i915TextureObjectPtr;
-
-#define I915_CONTEXT(ctx)	((i915ContextPtr)(ctx))
-
-
-
 #define I915_STATECHANGE(i915, flag)					\
 do {									\
-   if (0) fprintf(stderr, "I915_STATECHANGE %x in %s\n", flag, __FUNCTION__);	\
    INTEL_FIREVERTICES( &(i915)->intel );					\
    (i915)->state.emitted &= ~(flag);					\
 } while (0)
 
 #define I915_ACTIVESTATE(i915, flag, mode)			\
 do {								\
-   if (0) fprintf(stderr, "I915_ACTIVESTATE %x %d in %s\n",	\
-		  flag, mode, __FUNCTION__);			\
    INTEL_FIREVERTICES( &(i915)->intel );				\
    if (mode)							\
       (i915)->state.active |= (flag);				\
@@ -260,7 +268,13 @@ do {								\
 /*======================================================================
  * i915_vtbl.c
  */
-extern void i915InitVtbl( i915ContextPtr i915 );
+extern void i915InitVtbl(struct i915_context *i915);
+
+extern void
+i915_state_draw_region(struct intel_context *intel,
+                       struct i915_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region);
 
 
 
@@ -289,70 +303,58 @@ do {									\
 /*======================================================================
  * i915_context.c
  */
-extern GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-				    __DRIcontextPrivate *driContextPriv,
-				    void *sharedContextPrivate);
-
-
-/*======================================================================
- * i915_texprog.c
- */
-extern void i915ValidateTextureProgram( i915ContextPtr i915 );
+extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
 
 
 /*======================================================================
  * i915_debug.c
  */
-extern void i915_disassemble_program( const GLuint *program, GLuint sz );
-extern void i915_print_ureg( const char *msg, GLuint ureg );
+extern void i915_disassemble_program(const GLuint * program, GLuint sz);
+extern void i915_print_ureg(const char *msg, GLuint ureg);
 
 
 /*======================================================================
  * i915_state.c
  */
-extern void i915InitStateFunctions( struct dd_function_table *functions );
-extern void i915InitState( i915ContextPtr i915 );
-extern void i915_update_fog(GLcontext *ctxx);
+extern void i915InitStateFunctions(struct dd_function_table *functions);
+extern void i915InitState(struct i915_context *i915);
+extern void i915_update_fog(GLcontext * ctx);
 
 
 /*======================================================================
  * i915_tex.c
  */
-extern void i915UpdateTextureState( intelContextPtr intel );
-extern void i915InitTextureFuncs( struct dd_function_table *functions );
-extern intelTextureObjectPtr i915AllocTexObj( struct gl_texture_object *texObj );
+extern void i915UpdateTextureState(struct intel_context *intel);
+extern void i915InitTextureFuncs(struct dd_function_table *functions);
 
 /*======================================================================
  * i915_metaops.c
  */
-extern GLboolean
-i915TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels );
-
-extern GLboolean
-i915TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels );
+void i915InitMetaFuncs(struct i915_context *i915);
 
-extern void 
-i915ClearWithTris( intelContextPtr intel, GLbitfield mask,
-		   GLboolean all, GLint cx, GLint cy, GLint cw, GLint ch);
-
-
-extern void
-i915RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf);
 
 /*======================================================================
  * i915_fragprog.c
  */
-extern void i915ValidateFragmentProgram( i915ContextPtr i915 );
-extern void i915InitFragProgFuncs( struct dd_function_table *functions );
-	
-#endif
+extern void i915ValidateFragmentProgram(struct i915_context *i915);
+extern void i915InitFragProgFuncs(struct dd_function_table *functions);
+
+/*======================================================================
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct i915_context *
+i915_context(GLcontext * ctx)
+{
+   return (struct i915_context *) ctx;
+}
+
 
+
+#define I915_CONTEXT(ctx)	i915_context(ctx)
+
+
+
+#endif
diff --git a/src/mesa/drivers/dri/i915/i915_debug.c b/src/mesa/drivers/dri/i915/i915_debug.c
index 054b561605a..f7bb7ea44c9 100644
--- a/src/mesa/drivers/dri/i915/i915_debug.c
+++ b/src/mesa/drivers/dri/i915/i915_debug.c
@@ -25,275 +25,824 @@
  * 
  **************************************************************************/
 
+#include "main/imports.h"
+
 #include "i915_reg.h"
 #include "i915_context.h"
-#include <stdio.h>
-
-
-static const char *opcodes[0x20] = {
-   "NOP",
-   "ADD",
-   "MOV",
-   "MUL",
-   "MAD",
-   "DP2ADD",
-   "DP3",
-   "DP4",
-   "FRC",
-   "RCP",
-   "RSQ",
-   "EXP",
-   "LOG",
-   "CMP",
-   "MIN",
-   "MAX",
-   "FLR",
-   "MOD",
-   "TRC",
-   "SGE",
-   "SLT",
-   "TEXLD",
-   "TEXLDP",
-   "TEXLDB",
-   "TEXKILL",
-   "DCL",
-   "0x1a",
-   "0x1b",
-   "0x1c",
-   "0x1d",
-   "0x1e",
-   "0x1f",
-};
-
-
-static const int args[0x20] = {
-   0,				/* 0 nop */
-   2,				/* 1 add */
-   1,				/* 2 mov */
-   2,				/* 3 m ul */
-   3, 				/* 4 mad */
-   3,				/* 5 dp2add */
-   2,				/* 6 dp3 */
-   2,				/* 7 dp4 */
-   1,				/* 8 frc */
-   1,				/* 9 rcp */
-   1,				/* a rsq */
-   1,				/* b exp */
-   1,				/* c log */
-   3,				/* d cmp */
-   2,				/* e min */
-   2,				/* f max */
-   1,				/* 10 flr */
-   1,				/* 11 mod */
-   1,				/* 12 trc */
-   2,				/* 13 sge */
-   2,				/* 14 slt */
-   1,
-   1,
-   1,
-   1,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-};
-
-
-static const char *regname[0x8] = {
-   "R",
-   "T",
-   "CONST",
-   "S",
-   "OC",
-   "OD",
-   "U",
-   "UNKNOWN",
-};
-
-static void print_reg_type_nr( GLuint type, GLuint nr )
+#include "i915_debug.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
+
+static GLboolean debug( struct debug_stream *stream, const char *name, GLuint len )
 {
-   switch (type) {
-   case REG_TYPE_T:
-      switch (nr) {
-      case T_DIFFUSE: fprintf(stderr, "T_DIFFUSE"); return;
-      case T_SPECULAR: fprintf(stderr, "T_SPECULAR"); return;
-      case T_FOG_W: fprintf(stderr, "T_FOG_W"); return;
-      default: fprintf(stderr, "T_TEX%d", nr); return;
-      }
-   case REG_TYPE_OC:
-      if (nr == 0) {
-	 fprintf(stderr, "oC");
-	 return;
-      }
-      break;
-   case REG_TYPE_OD:
-      if (nr == 0) {
-	 fprintf(stderr, "oD");
-	 return;
-      }
-      break;
-   default:
-      break;
+   GLuint i;
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
    }
 
-   fprintf(stderr, "%s[%d]", regname[type], nr);
-}
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
 
-#define REG_SWIZZLE_MASK 0x7777
-#define REG_NEGATE_MASK 0x8888
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);   
+   PRINTF("\n");
 
-#define REG_SWIZZLE_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
-		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
 
 
-static void print_reg_neg_swizzle( GLuint reg )
+static const char *get_prim_name( GLuint val )
 {
-   int i;
-
-   if ((reg & REG_SWIZZLE_MASK) == REG_SWIZZLE_XYZW &&
-       (reg & REG_NEGATE_MASK) == 0)
-      return;
-
-   fprintf(stderr, ".");
-
-   for (i = 3 ; i >= 0; i--) {
-      if (reg & (1<<((i*4)+3))) 
-	 fprintf(stderr, "-");
-	 
-      switch ((reg>>(i*4)) & 0x7) {
-      case 0: fprintf(stderr, "x"); break;
-      case 1: fprintf(stderr, "y"); break;
-      case 2: fprintf(stderr, "z"); break;
-      case 3: fprintf(stderr, "w"); break;
-      case 4: fprintf(stderr, "0"); break;
-      case 5: fprintf(stderr, "1"); break;
-      default: fprintf(stderr, "?"); break;
-      }
+   switch (val & PRIM3D_MASK) {
+   case PRIM3D_TRILIST: return "TRILIST"; break;
+   case PRIM3D_TRISTRIP: return "TRISTRIP"; break;
+   case PRIM3D_TRISTRIP_RVRSE: return "TRISTRIP_RVRSE"; break;
+   case PRIM3D_TRIFAN: return "TRIFAN"; break;
+   case PRIM3D_POLY: return "POLY"; break;
+   case PRIM3D_LINELIST: return "LINELIST"; break;
+   case PRIM3D_LINESTRIP: return "LINESTRIP"; break;
+   case PRIM3D_RECTLIST: return "RECTLIST"; break;
+   case PRIM3D_POINTLIST: return "POINTLIST"; break;
+   case PRIM3D_DIB: return "DIB"; break;
+   case PRIM3D_CLEAR_RECT: return "CLEAR_RECT"; break;
+   case PRIM3D_ZONE_INIT: return "ZONE_INIT"; break;
+   default: return "????"; break;
    }
 }
 
-
-static void print_src_reg( GLuint dword )
+static GLboolean debug_prim( struct debug_stream *stream, const char *name, 
+			     GLboolean dump_floats,
+			     GLuint len )
 {
-   GLuint nr = (dword >> A2_SRC2_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A2_SRC2_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr( type, nr );
-   print_reg_neg_swizzle( dword );
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i;
+   
+
+
+   PRINTF("%s %s (%d dwords):\n", name, prim, len);
+   PRINTF("\t0x%08x\n",  ptr[0]);   
+   for (i = 1; i < len; i++) {
+      if (dump_floats)
+	 PRINTF("\t0x%08x // %f\n",  ptr[i], *(GLfloat *)&ptr[i]);   
+      else
+	 PRINTF("\t0x%08x\n",  ptr[i]);   
+   }
+
+      
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
 }
+   
+
 
-void i915_print_ureg( const char *msg, GLuint ureg )
+
+static GLboolean debug_program( struct debug_stream *stream, const char *name, GLuint len )
 {
-   fprintf(stderr, "%s: ", msg);
-   print_src_reg( ureg >> 8 );
-   fprintf(stderr, "\n");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
+   }
+
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   i915_disassemble_program( ptr, len );
+
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
 }
 
-static void print_dest_reg( GLuint dword )
+
+static GLboolean debug_chain( struct debug_stream *stream, const char *name, GLuint len )
 {
-   GLuint nr = (dword >> A0_DEST_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A0_DEST_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr( type, nr );
-   if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
-      return;
-   fprintf(stderr, ".");
-   if (dword & A0_DEST_CHANNEL_X) fprintf(stderr, "x");
-   if (dword & A0_DEST_CHANNEL_Y) fprintf(stderr, "y");
-   if (dword & A0_DEST_CHANNEL_Z) fprintf(stderr, "z");
-   if (dword & A0_DEST_CHANNEL_W) fprintf(stderr, "w");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint old_offset = stream->offset + len * sizeof(GLuint);
+   GLuint i;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+
+   stream->offset = ptr[1] & ~0x3;
+   
+   if (stream->offset < old_offset)
+      PRINTF("\n... skipping backwards from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+   else
+      PRINTF("\n... skipping from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+
+
+   return GL_TRUE;
 }
 
 
-#define GET_SRC0_REG(r0, r1) ((r0<<14)|(r1>>A1_SRC0_CHANNEL_W_SHIFT))
-#define GET_SRC1_REG(r0, r1) ((r0<<8)|(r1>>A2_SRC1_CHANNEL_W_SHIFT))
-#define GET_SRC2_REG(r)      (r)
+static GLboolean debug_variable_length_prim( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i, len;
+
+   GLushort *idx = (GLushort *)(ptr+1);
+   for (i = 0; idx[i] != 0xffff; i++)
+      ;
+
+   len = 1+(i+2)/2;
+
+   PRINTF("3DPRIM, %s variable length %d indicies (%d dwords):\n", prim, i, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
+}
 
 
-static void print_arith_op( GLuint opcode, const GLuint *program )
+#define BITS( dw, hi, lo, ... )				\
+do {							\
+   unsigned himask = 0xffffffffU >> (31 - (hi));		\
+   PRINTF("\t\t ");				\
+   PRINTF(__VA_ARGS__);			\
+   PRINTF(": 0x%x\n", ((dw) & himask) >> (lo));	\
+} while (0)
+
+#define MBZ( dw, hi, lo) do {							\
+   unsigned x = (dw) >> (lo);				\
+   unsigned lomask = (1 << (lo)) - 1;			\
+   unsigned himask;					\
+   himask = (1UL << (hi)) - 1;				\
+   assert ((x & himask & ~lomask) == 0);	\
+} while (0)
+
+#define FLAG( dw, bit, ... )			\
+do {							\
+   if (((dw) >> (bit)) & 1) {				\
+      PRINTF("\t\t ");				\
+      PRINTF(__VA_ARGS__);			\
+      PRINTF("\n");				\
+   }							\
+} while (0)
+
+static GLboolean debug_load_immediate( struct debug_stream *stream,
+				       const char *name,
+				       GLuint len )
 {
-   if (opcode != A0_NOP) {
-      print_dest_reg(program[0]);
-      if (program[0] & A0_DEST_SATURATE)
-	 fprintf(stderr, " = SATURATE ");
-      else
-	 fprintf(stderr, " = ");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 4) & 0xff;
+   GLuint j = 0;
+   
+   PRINTF("%s (%d dwords, flags: %x):\n", name, len, bits);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   if (bits & (1<<0)) {
+      PRINTF("\t  LIS0: 0x%08x\n", ptr[j]);
+      PRINTF("\t vb address: 0x%08x\n", (ptr[j] & ~0x3));
+      BITS(ptr[j], 0, 0, "vb invalidate disable");
+      j++;
+   }
+   if (bits & (1<<1)) {
+      PRINTF("\t  LIS1: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 29, 24, "vb dword width");
+      BITS(ptr[j], 21, 16, "vb dword pitch");
+      BITS(ptr[j], 15, 0, "vb max index");
+      j++;
    }
+   if (bits & (1<<2)) {
+      int i;
+      PRINTF("\t  LIS2: 0x%08x\n", ptr[j]);
+      for (i = 0; i < 8; i++) {
+	 unsigned tc = (ptr[j] >> (i * 4)) & 0xf;
+	 if (tc != 0xf)
+	    BITS(tc, 3, 0, "tex coord %d", i);
+      }
+      j++;
+   }
+   if (bits & (1<<3)) {
+      PRINTF("\t  LIS3: 0x%08x\n", ptr[j]);
+      j++;
+   }
+   if (bits & (1<<4)) {
+      PRINTF("\t  LIS4: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 23, "point width");
+      BITS(ptr[j], 22, 19, "line width");
+      FLAG(ptr[j], 18, "alpha flatshade");
+      FLAG(ptr[j], 17, "fog flatshade");
+      FLAG(ptr[j], 16, "spec flatshade");
+      FLAG(ptr[j], 15, "rgb flatshade");
+      BITS(ptr[j], 14, 13, "cull mode");
+      FLAG(ptr[j], 12, "vfmt: point width");
+      FLAG(ptr[j], 11, "vfmt: specular/fog");
+      FLAG(ptr[j], 10, "vfmt: rgba");
+      FLAG(ptr[j], 9, "vfmt: depth offset");
+      BITS(ptr[j], 8, 6, "vfmt: position (2==xyzw)");
+      FLAG(ptr[j], 5, "force dflt diffuse");
+      FLAG(ptr[j], 4, "force dflt specular");
+      FLAG(ptr[j], 3, "local depth offset enable");
+      FLAG(ptr[j], 2, "vfmt: fp32 fog coord");
+      FLAG(ptr[j], 1, "sprite point");
+      FLAG(ptr[j], 0, "antialiasing");
+      j++;
+   }
+   if (bits & (1<<5)) {
+      PRINTF("\t  LIS5: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 28, "rgba write disables");
+      FLAG(ptr[j], 27,     "force dflt point width");
+      FLAG(ptr[j], 26,     "last pixel enable");
+      FLAG(ptr[j], 25,     "global z offset enable");
+      FLAG(ptr[j], 24,     "fog enable");
+      BITS(ptr[j], 23, 16, "stencil ref");
+      BITS(ptr[j], 15, 13, "stencil test");
+      BITS(ptr[j], 12, 10, "stencil fail op");
+      BITS(ptr[j], 9, 7,   "stencil pass z fail op");
+      BITS(ptr[j], 6, 4,   "stencil pass z pass op");
+      FLAG(ptr[j], 3,      "stencil write enable");
+      FLAG(ptr[j], 2,      "stencil test enable");
+      FLAG(ptr[j], 1,      "color dither enable");
+      FLAG(ptr[j], 0,      "logiop enable");
+      j++;
+   }
+   if (bits & (1<<6)) {
+      PRINTF("\t  LIS6: 0x%08x\n", ptr[j]);
+      FLAG(ptr[j], 31,      "alpha test enable");
+      BITS(ptr[j], 30, 28,  "alpha func");
+      BITS(ptr[j], 27, 20,  "alpha ref");
+      FLAG(ptr[j], 19,      "depth test enable");
+      BITS(ptr[j], 18, 16,  "depth func");
+      FLAG(ptr[j], 15,      "blend enable");
+      BITS(ptr[j], 14, 12,  "blend func");
+      BITS(ptr[j], 11, 8,   "blend src factor");
+      BITS(ptr[j], 7,  4,   "blend dst factor");
+      FLAG(ptr[j], 3,       "depth write enable");
+      FLAG(ptr[j], 2,       "color write enable");
+      BITS(ptr[j], 1,  0,   "provoking vertex"); 
+      j++;
+   }
+
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   PRINTF("\n");
 
-   print_src_reg(GET_SRC0_REG(program[0], program[1]));
-   if (args[opcode] == 1) {
-      fprintf(stderr, "\n");
-      return;
+   assert(j == len);
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 
+
+
+static GLboolean debug_load_indirect( struct debug_stream *stream,
+				      const char *name,
+				      GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 8) & 0x3f;
+   GLuint i, j = 0;
+   
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   for (i = 0; i < 6; i++) {
+      if (bits & (1<<i)) {
+	 switch (1<<(8+i)) {
+	 case LI0_STATE_STATIC_INDIRECT:
+	    PRINTF("        STATIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_DYNAMIC_INDIRECT:
+	    PRINTF("       DYNAMIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    break;
+	 case LI0_STATE_SAMPLER:
+	    PRINTF("       SAMPLER: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_MAP:
+	    PRINTF("           MAP: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_PROGRAM:
+	    PRINTF("       PROGRAM: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_CONSTANTS:
+	    PRINTF("     CONSTANTS: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 default:
+	    assert(0);
+	    break;
+	 }
+      }
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC1_REG(program[1], program[2]));
-   if (args[opcode] == 2) { 
-      fprintf(stderr, "\n");
-      return;
+   if (bits == 0) {
+      PRINTF("\t  DUMMY: 0x%08x\n", ptr[j++]);
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC2_REG(program[2]));
-   fprintf(stderr, "\n");
-   return;
+   PRINTF("\n");
+
+
+   assert(j == len);
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 	
+static void BR13( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   FLAG(val, 30, "clipping enable");
+   BITS(val, 25, 24, "color depth (3==32bpp)");
+   BITS(val, 23, 16, "raster op");
+   BITS(val, 15, 0,  "dest pitch");
 }
 
 
-static void print_tex_op( GLuint opcode, const GLuint *program )
+static void BR2223( struct debug_stream *stream,
+		    GLuint val22, GLuint val23 )
 {
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, " = ");
+   union { GLuint val; short field[2]; } BR22, BR23;
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   BR22.val = val22;
+   BR23.val = val23;
 
-   fprintf(stderr, "S[%d],", 
-	   program[0] & T0_SAMPLER_NR_MASK);
+   PRINTF("\t0x%08x\n",  val22);
+   BITS(val22, 31, 16, "dest y1");
+   BITS(val22, 15, 0,  "dest x1");
 
-   print_reg_type_nr( (program[1]>>T1_ADDRESS_REG_TYPE_SHIFT) & REG_TYPE_MASK,
-		      (program[1]>>T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK );
-   fprintf(stderr, "\n");
+   PRINTF("\t0x%08x\n",  val23);
+   BITS(val23, 31, 16, "dest y2");
+   BITS(val23, 15, 0,  "dest x2");
+
+   /* The blit engine may produce unexpected results when these aren't met */
+   assert(BR22.field[0] < BR23.field[0]);
+   assert(BR22.field[1] < BR23.field[1]);
 }
 
-static void print_dcl_op( GLuint opcode, const GLuint *program )
+static void BR09( struct debug_stream *stream,
+		  GLuint val )
 {
-   fprintf(stderr, "%s ", opcodes[opcode]);
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, "\n");
+   PRINTF("\t0x%08x -- dest address\n",  val);
 }
 
+static void BR26( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "src y1");
+   BITS(val, 15, 0,  "src x1");
+}
+
+static void BR11( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 15, 0,  "src pitch");
+}
 
-void i915_disassemble_program( const GLuint *program, GLuint sz )
+static void BR12( struct debug_stream *stream,
+		  GLuint val )
 {
-   GLuint size = program[0] & 0x1ff;
-   GLint i;
+   PRINTF("\t0x%08x -- src address\n",  val);
+}
+
+static void BR16( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- color\n",  val);
+}
+   
+static GLboolean debug_copy_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   BR13(stream, ptr[j++]);
+   BR2223(stream, ptr[j], ptr[j+1]);
+   j += 2;
+   BR09(stream, ptr[j++]);
+   BR26(stream, ptr[j++]);
+   BR11(stream, ptr[j++]);
+   BR12(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_color_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   BR13(stream, ptr[j++]);
+   BR2223(stream, ptr[j], ptr[j+1]);
+   j += 2;
+   BR09(stream, ptr[j++]);
+   BR16(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_modes4( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j]);
+   BITS(ptr[j], 21, 18, "logicop func");
+   FLAG(ptr[j], 17, "stencil test mask modify-enable");
+   FLAG(ptr[j], 16, "stencil write mask modify-enable");
+   BITS(ptr[j], 15, 8, "stencil test mask");
+   BITS(ptr[j], 7, 0,  "stencil write mask");
+   j++;
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_map_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "map mask");
+      j++;
+   }
+
+   while (j < len) {
+      {
+	 PRINTF("\t  TMn.0: 0x%08x\n", ptr[j]);
+	 PRINTF("\t map address: 0x%08x\n", (ptr[j] & ~0x3));
+	 FLAG(ptr[j], 1, "vertical line stride");
+	 FLAG(ptr[j], 0, "vertical line stride offset");
+	 j++;
+      }
+
+      {
+	 PRINTF("\t  TMn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "height");
+	 BITS(ptr[j], 20, 10, "width");
+	 BITS(ptr[j], 9, 7, "surface format");
+	 BITS(ptr[j], 6, 3, "texel format");
+	 FLAG(ptr[j], 2, "use fence regs");
+	 FLAG(ptr[j], 1, "tiled surface");
+	 FLAG(ptr[j], 0, "tile walk ymajor");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TMn.2: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "dword pitch");
+	 BITS(ptr[j], 20, 15, "cube face enables");
+	 BITS(ptr[j], 14, 9, "max lod");
+	 FLAG(ptr[j], 8,     "mip layout right");
+	 BITS(ptr[j], 7, 0, "depth");
+	 j++;
+      }
+   }
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_sampler_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "sampler mask");
+      j++;
+   }
+
+   while (j < len) {
+      {
+	 PRINTF("\t  TSn.0: 0x%08x\n", ptr[j]);
+	 FLAG(ptr[j], 31, "reverse gamma");
+	 FLAG(ptr[j], 30, "planar to packed");
+	 FLAG(ptr[j], 29, "yuv->rgb");
+	 BITS(ptr[j], 28, 27, "chromakey index");
+	 BITS(ptr[j], 26, 22, "base mip level");
+	 BITS(ptr[j], 21, 20, "mip mode filter");
+	 BITS(ptr[j], 19, 17, "mag mode filter");
+	 BITS(ptr[j], 16, 14, "min mode filter");
+	 BITS(ptr[j], 13, 5,  "lod bias (s4.4)");
+	 FLAG(ptr[j], 4,      "shadow enable");
+	 FLAG(ptr[j], 3,      "max-aniso-4");
+	 BITS(ptr[j], 2, 0,   "shadow func");
+	 j++;
+      }
+
+      {
+	 PRINTF("\t  TSn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 24, "min lod");
+	 MBZ( ptr[j], 23, 18 );
+	 FLAG(ptr[j], 17,     "kill pixel enable");
+	 FLAG(ptr[j], 16,     "keyed tex filter mode");
+	 FLAG(ptr[j], 15,     "chromakey enable");
+	 BITS(ptr[j], 14, 12, "tcx wrap mode");
+	 BITS(ptr[j], 11, 9,  "tcy wrap mode");
+	 BITS(ptr[j], 8,  6,  "tcz wrap mode");
+	 FLAG(ptr[j], 5,      "normalized coords");
+	 BITS(ptr[j], 4,  1,  "map (surface) index");
+	 FLAG(ptr[j], 0,      "EAST deinterlacer enable");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TSn.2: 0x%08x  (default color)\n", ptr[j]);
+	 j++;
+      }
+   }
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_dest_vars( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      FLAG(ptr[j], 31,     "early classic ztest");
+      FLAG(ptr[j], 30,     "opengl tex default color");
+      FLAG(ptr[j], 29,     "bypass iz");
+      FLAG(ptr[j], 28,     "lod preclamp");
+      BITS(ptr[j], 27, 26, "dither pattern");
+      FLAG(ptr[j], 25,     "linear gamma blend");
+      FLAG(ptr[j], 24,     "debug dither");
+      BITS(ptr[j], 23, 20, "dstorg x");
+      BITS(ptr[j], 19, 16, "dstorg y");
+      MBZ (ptr[j], 15, 15 );
+      BITS(ptr[j], 14, 12, "422 write select");
+      BITS(ptr[j], 11, 8,  "cbuf format");
+      BITS(ptr[j], 3, 2,   "zbuf format");
+      FLAG(ptr[j], 1,      "vert line stride");
+      FLAG(ptr[j], 1,      "vert line stride offset");
+      j++;
+   }
    
-   fprintf(stderr, "BEGIN\n");
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
 
-   if (size+2 != sz) {
-      fprintf(stderr, "%s: program size mismatch %d/%d\n", __FUNCTION__,
-	      size+2, sz);
-      exit(1);
+static GLboolean debug_buf_info( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 28, 28, "aux buffer id");
+      BITS(ptr[j], 27, 24, "buffer id (7=depth, 3=back)");
+      FLAG(ptr[j], 23,     "use fence regs");
+      FLAG(ptr[j], 22,     "tiled surface");
+      FLAG(ptr[j], 21,     "tile walk ymajor");
+      MBZ (ptr[j], 20, 14);
+      BITS(ptr[j], 13, 2,  "dword pitch");
+      MBZ (ptr[j], 2,  0);
+      j++;
    }
+   
+   PRINTF("\t0x%08x -- buffer base address\n",  ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
 
-   program ++;
-   for (i = 1 ; i < sz ; i+=3, program+=3) {
-      GLuint opcode = program[0] & (0x1f<<24);
-
-      if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
-	 print_arith_op(opcode >> 24, program);
-      else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
-	 print_tex_op(opcode >> 24, program);
-      else if (opcode == D0_DCL)
-	 print_dcl_op(opcode >> 24, program);
-      else 
-	 fprintf(stderr, "Unknown opcode 0x%x\n", opcode);
+static GLboolean i915_debug_packet( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint cmd = *ptr;
+   
+   switch (((cmd >> 29) & 0x7)) {
+   case 0x0:
+      switch ((cmd >> 23) & 0x3f) {
+      case 0x0:
+	 return debug(stream, "MI_NOOP", 1);
+      case 0x3:
+	 return debug(stream, "MI_WAIT_FOR_EVENT", 1);
+      case 0x4:
+	 return debug(stream, "MI_FLUSH", 1);
+      case 0xA:
+	 debug(stream, "MI_BATCH_BUFFER_END", 1);
+	 return GL_FALSE;
+      case 0x22:
+	 return debug(stream, "MI_LOAD_REGISTER_IMM", 3);
+      case 0x31:
+	 return debug_chain(stream, "MI_BATCH_BUFFER_START", 2);
+      default:
+	 break;
+      }
+      break;
+   case 0x1:
+      break;
+   case 0x2:
+      switch ((cmd >> 22) & 0xff) {	 
+      case 0x50:
+	 return debug_color_blit(stream, "XY_COLOR_BLT", (cmd & 0xff) + 2);
+      case 0x53:
+	 return debug_copy_blit(stream, "XY_SRC_COPY_BLT", (cmd & 0xff) + 2);
+      default:
+	 return debug(stream, "blit command", (cmd & 0xff) + 2);
+      }
+      break;
+   case 0x3:
+      switch ((cmd >> 24) & 0x1f) {	 
+      case 0x6:
+	 return debug(stream, "3DSTATE_ANTI_ALIASING", 1);
+      case 0x7:
+	 return debug(stream, "3DSTATE_RASTERIZATION_RULES", 1);
+      case 0x8:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_OPS", 2);
+      case 0x9:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_MASKS", 1);
+      case 0xb:
+	 return debug(stream, "3DSTATE_INDEPENDENT_ALPHA_BLEND", 1);
+      case 0xc:
+	 return debug(stream, "3DSTATE_MODES5", 1);	 
+      case 0xd:
+	 return debug_modes4(stream, "3DSTATE_MODES4", 1);
+      case 0x15:
+	 return debug(stream, "3DSTATE_FOG_COLOR", 1);
+      case 0x16:
+	 return debug(stream, "3DSTATE_COORD_SET_BINDINGS", 1);
+      case 0x1c:
+	 /* 3DState16NP */
+	 switch((cmd >> 19) & 0x1f) {
+	 case 0x10:
+	    return debug(stream, "3DSTATE_SCISSOR_ENABLE", 1);
+	 case 0x11:
+	    return debug(stream, "3DSTATE_DEPTH_SUBRECTANGLE_DISABLE", 1);
+	 default:
+	    break;
+	 }
+	 break;
+      case 0x1d:
+	 /* 3DStateMW */
+	 switch ((cmd >> 16) & 0xff) {
+	 case 0x0:
+	    return debug_map_state(stream, "3DSTATE_MAP_STATE", (cmd & 0x1f) + 2);
+	 case 0x1:
+	    return debug_sampler_state(stream, "3DSTATE_SAMPLER_STATE", (cmd & 0x1f) + 2);
+	 case 0x4:
+	    return debug_load_immediate(stream, "3DSTATE_LOAD_STATE_IMMEDIATE", (cmd & 0xf) + 2);
+	 case 0x5:
+	    return debug_program(stream, "3DSTATE_PIXEL_SHADER_PROGRAM", (cmd & 0x1ff) + 2);
+	 case 0x6:
+	    return debug(stream, "3DSTATE_PIXEL_SHADER_CONSTANTS", (cmd & 0xff) + 2);
+	 case 0x7:
+	    return debug_load_indirect(stream, "3DSTATE_LOAD_INDIRECT", (cmd & 0xff) + 2);
+	 case 0x80:
+	    return debug(stream, "3DSTATE_DRAWING_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x81:
+	    return debug(stream, "3DSTATE_SCISSOR_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x83:
+	    return debug(stream, "3DSTATE_SPAN_STIPPLE", (cmd & 0xffff) + 2);
+	 case 0x85:
+	    return debug_dest_vars(stream, "3DSTATE_DEST_BUFFER_VARS", (cmd & 0xffff) + 2);
+	 case 0x88:
+	    return debug(stream, "3DSTATE_CONSTANT_BLEND_COLOR", (cmd & 0xffff) + 2);
+	 case 0x89:
+	    return debug(stream, "3DSTATE_FOG_MODE", (cmd & 0xffff) + 2);
+	 case 0x8e:
+	    return debug_buf_info(stream, "3DSTATE_BUFFER_INFO", (cmd & 0xffff) + 2);
+	 case 0x97:
+	    return debug(stream, "3DSTATE_DEPTH_OFFSET_SCALE", (cmd & 0xffff) + 2);
+	 case 0x98:
+	    return debug(stream, "3DSTATE_DEFAULT_Z", (cmd & 0xffff) + 2);
+	 case 0x99:
+	    return debug(stream, "3DSTATE_DEFAULT_DIFFUSE", (cmd & 0xffff) + 2);
+	 case 0x9a:
+	    return debug(stream, "3DSTATE_DEFAULT_SPECULAR", (cmd & 0xffff) + 2);
+	 case 0x9c:
+	    return debug(stream, "3DSTATE_CLEAR_PARAMETERS", (cmd & 0xffff) + 2);
+	 default:
+	    assert(0);
+	    return 0;
+	 }
+	 break;
+      case 0x1e:
+	 if (cmd & (1 << 23))
+	    return debug(stream, "???", (cmd & 0xffff) + 1);
+	 else
+	    return debug(stream, "", 1);
+	 break;
+      case 0x1f:
+	 if ((cmd & (1 << 23)) == 0)	
+	    return debug_prim(stream, "3DPRIM (inline)", 1, (cmd & 0x1ffff) + 2);
+	 else if (cmd & (1 << 17)) 
+	 {
+	    if ((cmd & 0xffff) == 0)
+	       return debug_variable_length_prim(stream);
+	    else
+	       return debug_prim(stream, "3DPRIM (indexed)", 0, (((cmd & 0xffff) + 1) / 2) + 1);
+	 }
+	 else
+	    return debug_prim(stream, "3DPRIM  (indirect sequential)", 0, 2); 
+	 break;
+      default:
+	 return debug(stream, "", 0);
+      }
+   default:
+      assert(0);
+      return 0;
+   }
+
+   assert(0);
+   return 0;
+}
+
+
+
+void
+i915_dump_batchbuffer( GLuint *start,
+		       GLuint *end )
+{
+   struct debug_stream stream;
+   GLuint bytes = (end - start) * 4;
+   GLboolean done = GL_FALSE;
+
+   PRINTF("\n\nBATCH: (%d)\n", bytes / 4);
+
+   stream.offset = 0;
+   stream.ptr = (char *)start;
+   stream.print_addresses = 0;
+
+   while (!done &&
+	  stream.offset < bytes &&
+	  stream.offset >= 0)
+   {
+      if (!i915_debug_packet( &stream ))
+	 break;
+
+      assert(stream.offset <= bytes &&
+	     stream.offset >= 0);
    }
 
-   fprintf(stderr, "END\n\n");
+   PRINTF("END-BATCH\n\n\n");
 }
+
+
diff --git a/src/mesa/drivers/dri/i915/intel_tex.h b/src/mesa/drivers/dri/i915/i915_debug.h
index 9b7e5502326..0643a8c631c 100644
--- a/src/mesa/drivers/dri/i915/intel_tex.h
+++ b/src/mesa/drivers/dri/i915/i915_debug.h
@@ -1,8 +1,8 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -25,21 +25,31 @@
  * 
  **************************************************************************/
 
-#ifndef INTELTEX_INC
-#define INTELTEX_INC
+/* Authors:  Keith Whitwell <keith@tungstengraphics.com>
+ */
+
+#ifndef I915_DEBUG_H
+#define I915_DEBUG_H
+
+struct i915_context;
+
+struct debug_stream 
+{
+   unsigned offset;		/* current gtt offset */
+   char *ptr;		/* pointer to gtt offset zero */
+   char *end;		/* pointer to gtt offset zero */
+   unsigned print_addresses;
+};
+
+
 
-#include "mtypes.h"
-#include "intel_context.h"
-#include "texmem.h"
+extern void i915_disassemble_program(const unsigned *program, unsigned sz);
+extern void i915_print_ureg(const char *msg, unsigned ureg);
 
 
-void intelInitTextureFuncs( struct dd_function_table *functions );
+void
+i915_dump_batchbuffer( unsigned *start,
+		       unsigned *end );
 
-void intelDestroyTexObj( intelContextPtr intel, intelTextureObjectPtr t );
-int intelUploadTexImages( intelContextPtr intel, intelTextureObjectPtr t,
-			  GLuint face );
 
-GLboolean 
-intel_driReinitTextureHeap( driTexHeap *heap,
-			    unsigned size );
 #endif
diff --git a/src/mesa/drivers/dri/i915/i915_debug_fp.c b/src/mesa/drivers/dri/i915/i915_debug_fp.c
new file mode 100644
index 00000000000..84347a01efe
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/i915_debug_fp.c
@@ -0,0 +1,333 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include <stdio.h>
+
+#include "i915_reg.h"
+#include "i915_debug.h"
+#include "main/imports.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_print.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
+
+static const char *opcodes[0x20] = {
+   "NOP",
+   "ADD",
+   "MOV",
+   "MUL",
+   "MAD",
+   "DP2ADD",
+   "DP3",
+   "DP4",
+   "FRC",
+   "RCP",
+   "RSQ",
+   "EXP",
+   "LOG",
+   "CMP",
+   "MIN",
+   "MAX",
+   "FLR",
+   "MOD",
+   "TRC",
+   "SGE",
+   "SLT",
+   "TEXLD",
+   "TEXLDP",
+   "TEXLDB",
+   "TEXKILL",
+   "DCL",
+   "0x1a",
+   "0x1b",
+   "0x1c",
+   "0x1d",
+   "0x1e",
+   "0x1f",
+};
+
+
+static const int args[0x20] = {
+   0,                           /* 0 nop */
+   2,                           /* 1 add */
+   1,                           /* 2 mov */
+   2,                           /* 3 m ul */
+   3,                           /* 4 mad */
+   3,                           /* 5 dp2add */
+   2,                           /* 6 dp3 */
+   2,                           /* 7 dp4 */
+   1,                           /* 8 frc */
+   1,                           /* 9 rcp */
+   1,                           /* a rsq */
+   1,                           /* b exp */
+   1,                           /* c log */
+   3,                           /* d cmp */
+   2,                           /* e min */
+   2,                           /* f max */
+   1,                           /* 10 flr */
+   1,                           /* 11 mod */
+   1,                           /* 12 trc */
+   2,                           /* 13 sge */
+   2,                           /* 14 slt */
+   1,
+   1,
+   1,
+   1,
+   0,
+   0,
+   0,
+   0,
+   0,
+   0,
+   0,
+};
+
+
+static const char *regname[0x8] = {
+   "R",
+   "T",
+   "CONST",
+   "S",
+   "OC",
+   "OD",
+   "U",
+   "UNKNOWN",
+};
+
+static void
+print_reg_type_nr(GLuint type, GLuint nr)
+{
+   switch (type) {
+   case REG_TYPE_T:
+      switch (nr) {
+      case T_DIFFUSE:
+         PRINTF("T_DIFFUSE");
+         return;
+      case T_SPECULAR:
+         PRINTF("T_SPECULAR");
+         return;
+      case T_FOG_W:
+         PRINTF("T_FOG_W");
+         return;
+      default:
+         PRINTF("T_TEX%d", nr);
+         return;
+      }
+   case REG_TYPE_OC:
+      if (nr == 0) {
+         PRINTF("oC");
+         return;
+      }
+      break;
+   case REG_TYPE_OD:
+      if (nr == 0) {
+         PRINTF("oD");
+         return;
+      }
+      break;
+   default:
+      break;
+   }
+
+   PRINTF("%s[%d]", regname[type], nr);
+}
+
+#define REG_SWIZZLE_MASK 0x7777
+#define REG_NEGATE_MASK 0x8888
+
+#define REG_SWIZZLE_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
+		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
+		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
+		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
+
+
+static void
+print_reg_neg_swizzle(GLuint reg)
+{
+   int i;
+
+   if ((reg & REG_SWIZZLE_MASK) == REG_SWIZZLE_XYZW &&
+       (reg & REG_NEGATE_MASK) == 0)
+      return;
+
+   PRINTF(".");
+
+   for (i = 3; i >= 0; i--) {
+      if (reg & (1 << ((i * 4) + 3)))
+         PRINTF("-");
+
+      switch ((reg >> (i * 4)) & 0x7) {
+      case 0:
+         PRINTF("x");
+         break;
+      case 1:
+         PRINTF("y");
+         break;
+      case 2:
+         PRINTF("z");
+         break;
+      case 3:
+         PRINTF("w");
+         break;
+      case 4:
+         PRINTF("0");
+         break;
+      case 5:
+         PRINTF("1");
+         break;
+      default:
+         PRINTF("?");
+         break;
+      }
+   }
+}
+
+
+static void
+print_src_reg(GLuint dword)
+{
+   GLuint nr = (dword >> A2_SRC2_NR_SHIFT) & REG_NR_MASK;
+   GLuint type = (dword >> A2_SRC2_TYPE_SHIFT) & REG_TYPE_MASK;
+   print_reg_type_nr(type, nr);
+   print_reg_neg_swizzle(dword);
+}
+
+
+static void
+print_dest_reg(GLuint dword)
+{
+   GLuint nr = (dword >> A0_DEST_NR_SHIFT) & REG_NR_MASK;
+   GLuint type = (dword >> A0_DEST_TYPE_SHIFT) & REG_TYPE_MASK;
+   print_reg_type_nr(type, nr);
+   if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
+      return;
+   PRINTF(".");
+   if (dword & A0_DEST_CHANNEL_X)
+      PRINTF("x");
+   if (dword & A0_DEST_CHANNEL_Y)
+      PRINTF("y");
+   if (dword & A0_DEST_CHANNEL_Z)
+      PRINTF("z");
+   if (dword & A0_DEST_CHANNEL_W)
+      PRINTF("w");
+}
+
+
+#define GET_SRC0_REG(r0, r1) ((r0<<14)|(r1>>A1_SRC0_CHANNEL_W_SHIFT))
+#define GET_SRC1_REG(r0, r1) ((r0<<8)|(r1>>A2_SRC1_CHANNEL_W_SHIFT))
+#define GET_SRC2_REG(r)      (r)
+
+
+static void
+print_arith_op(GLuint opcode, const GLuint * program)
+{
+   if (opcode != A0_NOP) {
+      print_dest_reg(program[0]);
+      if (program[0] & A0_DEST_SATURATE)
+         PRINTF(" = SATURATE ");
+      else
+         PRINTF(" = ");
+   }
+
+   PRINTF("%s ", opcodes[opcode]);
+
+   print_src_reg(GET_SRC0_REG(program[0], program[1]));
+   if (args[opcode] == 1) {
+      PRINTF("\n");
+      return;
+   }
+
+   PRINTF(", ");
+   print_src_reg(GET_SRC1_REG(program[1], program[2]));
+   if (args[opcode] == 2) {
+      PRINTF("\n");
+      return;
+   }
+
+   PRINTF(", ");
+   print_src_reg(GET_SRC2_REG(program[2]));
+   PRINTF("\n");
+   return;
+}
+
+
+static void
+print_tex_op(GLuint opcode, const GLuint * program)
+{
+   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
+   PRINTF(" = ");
+
+   PRINTF("%s ", opcodes[opcode]);
+
+   PRINTF("S[%d],", program[0] & T0_SAMPLER_NR_MASK);
+
+   print_reg_type_nr((program[1] >> T1_ADDRESS_REG_TYPE_SHIFT) &
+                     REG_TYPE_MASK,
+                     (program[1] >> T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK);
+   PRINTF("\n");
+}
+
+static void
+print_dcl_op(GLuint opcode, const GLuint * program)
+{
+   PRINTF("%s ", opcodes[opcode]);
+   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
+   PRINTF("\n");
+}
+
+
+void
+i915_disassemble_program(const GLuint * program, GLuint sz)
+{
+   GLuint size = program[0] & 0x1ff;
+   GLint i;
+
+   PRINTF("\t\tBEGIN\n");
+
+   assert(size + 2 == sz);
+
+   program++;
+   for (i = 1; i < sz; i += 3, program += 3) {
+      GLuint opcode = program[0] & (0x1f << 24);
+
+      PRINTF("\t\t");
+
+      if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
+         print_arith_op(opcode >> 24, program);
+      else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
+         print_tex_op(opcode >> 24, program);
+      else if (opcode == D0_DCL)
+         print_dcl_op(opcode >> 24, program);
+      else
+         PRINTF("Unknown opcode 0x%x\n", opcode);
+   }
+
+   PRINTF("\t\tEND\n\n");
+}
+
+
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 702b8788282..8bd761ec6a1 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -25,44 +25,60 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/program.h"
+#include "shader/programopt.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
+
 #include "intel_batchbuffer.h"
 
 #include "i915_reg.h"
 #include "i915_context.h"
 #include "i915_program.h"
 
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "program.h"
-#include "programopt.h"
-
-
+static const GLfloat sin_quad_constants[2][4] = {
+   {
+      2.0,
+      -1.0,
+      .5,
+      .75
+   },
+   {
+      4.0,
+      -4.0,
+      1.0 / (2.0 * M_PI),
+      .2225
+   }
+};
 
-/* 1, -1/3!, 1/5!, -1/7! */
-static const GLfloat sin_constants[4] = {  1.0, 
-					   -1.0/(3*2*1),
-					   1.0/(5*4*3*2*1),
-					   -1.0/(7*6*5*4*3*2*1) };
+static const GLfloat sin_constants[4] = { 1.0,
+   -1.0 / (3 * 2 * 1),
+   1.0 / (5 * 4 * 3 * 2 * 1),
+   -1.0 / (7 * 6 * 5 * 4 * 3 * 2 * 1)
+};
 
 /* 1, -1/2!, 1/4!, -1/6! */
-static const GLfloat cos_constants[4] = {  1.0, 
-					   -1.0/(2*1),
-					   1.0/(4*3*2*1),
-					   -1.0/(6*5*4*3*2*1) };
+static const GLfloat cos_constants[4] = { 1.0,
+   -1.0 / (2 * 1),
+   1.0 / (4 * 3 * 2 * 1),
+   -1.0 / (6 * 5 * 4 * 3 * 2 * 1)
+};
 
 /**
  * Retrieve a ureg for the given source register.  Will emit
  * constants, apply swizzling and negation as needed.
  */
-static GLuint src_vector( struct i915_fragment_program *p,
-			  const struct prog_src_register *source,
-			  const struct gl_fragment_program *program )
+static GLuint
+src_vector(struct i915_fragment_program *p,
+           const struct prog_src_register *source,
+           const struct gl_fragment_program *program)
 {
    GLuint src;
 
@@ -70,136 +86,152 @@ static GLuint src_vector( struct i915_fragment_program *p,
 
       /* Registers:
        */
-      case PROGRAM_TEMPORARY:
-	 if (source->Index >= I915_MAX_TEMPORARY) {
-	    i915_program_error( p, "Exceeded max temporary reg" );
-	    return 0;
-	 }
-	 src = UREG( REG_TYPE_R, source->Index );
+   case PROGRAM_TEMPORARY:
+      if (source->Index >= I915_MAX_TEMPORARY) {
+         i915_program_error(p, "Exceeded max temporary reg");
+         return 0;
+      }
+      src = UREG(REG_TYPE_R, source->Index);
+      break;
+   case PROGRAM_INPUT:
+      switch (source->Index) {
+      case FRAG_ATTRIB_WPOS:
+         src = i915_emit_decl(p, REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL);
          break;
-      case PROGRAM_INPUT:
-	 switch (source->Index) {
-	 case FRAG_ATTRIB_WPOS:
-	    src = i915_emit_decl( p,  REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL ); 
-	    break;
-	 case FRAG_ATTRIB_COL0:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL ); 
-	    break;
-	 case FRAG_ATTRIB_COL1:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_SPECULAR, D0_CHANNEL_XYZ ); 
-	    src = swizzle( src, X, Y, Z, ONE );
-	    break;
-	 case FRAG_ATTRIB_FOGC:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_FOG_W, D0_CHANNEL_W ); 
-	    src = swizzle( src, W, W, W, W );
-	    break;
-	 case FRAG_ATTRIB_TEX0:
-	 case FRAG_ATTRIB_TEX1:
-	 case FRAG_ATTRIB_TEX2:
-	 case FRAG_ATTRIB_TEX3:
-	 case FRAG_ATTRIB_TEX4:
-	 case FRAG_ATTRIB_TEX5:
-	 case FRAG_ATTRIB_TEX6:
-	 case FRAG_ATTRIB_TEX7:
-	    src = i915_emit_decl( p,  REG_TYPE_T, 
-				 T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
-				 D0_CHANNEL_ALL ); 
-	    break;
-
-	 default:
-	    i915_program_error( p, "Bad source->Index" ); 
-	    return 0;
-	 }
+      case FRAG_ATTRIB_COL0:
+         src = i915_emit_decl(p, REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL);
+         break;
+      case FRAG_ATTRIB_COL1:
+         src = i915_emit_decl(p, REG_TYPE_T, T_SPECULAR, D0_CHANNEL_XYZ);
+         src = swizzle(src, X, Y, Z, ONE);
+         break;
+      case FRAG_ATTRIB_FOGC:
+         src = i915_emit_decl(p, REG_TYPE_T, T_FOG_W, D0_CHANNEL_W);
+         src = swizzle(src, W, ZERO, ZERO, ONE);
+         break;
+      case FRAG_ATTRIB_TEX0:
+      case FRAG_ATTRIB_TEX1:
+      case FRAG_ATTRIB_TEX2:
+      case FRAG_ATTRIB_TEX3:
+      case FRAG_ATTRIB_TEX4:
+      case FRAG_ATTRIB_TEX5:
+      case FRAG_ATTRIB_TEX6:
+      case FRAG_ATTRIB_TEX7:
+         src = i915_emit_decl(p, REG_TYPE_T,
+                              T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
+                              D0_CHANNEL_ALL);
          break;
-
-	 /* Various paramters and env values.  All emitted to
-	  * hardware as program constants.
-	  */
-      case PROGRAM_LOCAL_PARAM:
-         src = i915_emit_param4fv( 
-	    p, program->Base.LocalParams[source->Index]);
-	 break;
-
-      case PROGRAM_ENV_PARAM:
-         src = i915_emit_param4fv( 
-	    p, p->ctx->FragmentProgram.Parameters[source->Index]);
-	 break;
-
-      case PROGRAM_CONSTANT:
-      case PROGRAM_STATE_VAR:
-      case PROGRAM_NAMED_PARAM:
-         src = i915_emit_param4fv( 
-	    p, program->Base.Parameters->ParameterValues[source->Index] );
-	 break;
 
       default:
-	 i915_program_error( p, "Bad source->File" ); 
-	 return 0;
+         i915_program_error(p, "Bad source->Index");
+         return 0;
+      }
+      break;
+
+      /* Various paramters and env values.  All emitted to
+       * hardware as program constants.
+       */
+   case PROGRAM_LOCAL_PARAM:
+      src = i915_emit_param4fv(p, program->Base.LocalParams[source->Index]);
+      break;
+
+   case PROGRAM_ENV_PARAM:
+      src =
+         i915_emit_param4fv(p,
+                            p->ctx->FragmentProgram.Parameters[source->
+                                                               Index]);
+      break;
+
+   case PROGRAM_CONSTANT:
+   case PROGRAM_STATE_VAR:
+   case PROGRAM_NAMED_PARAM:
+      src =
+         i915_emit_param4fv(p,
+                            program->Base.Parameters->ParameterValues[source->
+                                                                      Index]);
+      break;
+
+   default:
+      i915_program_error(p, "Bad source->File");
+      return 0;
    }
 
-   src = swizzle(src, 
-		 GET_SWZ(source->Swizzle, 0),
-		 GET_SWZ(source->Swizzle, 1),
-		 GET_SWZ(source->Swizzle, 2),
-		 GET_SWZ(source->Swizzle, 3));
+   src = swizzle(src,
+                 GET_SWZ(source->Swizzle, 0),
+                 GET_SWZ(source->Swizzle, 1),
+                 GET_SWZ(source->Swizzle, 2), GET_SWZ(source->Swizzle, 3));
 
    if (source->NegateBase)
-      src = negate( src, 
-		    GET_BIT(source->NegateBase, 0),
-		    GET_BIT(source->NegateBase, 1),
-		    GET_BIT(source->NegateBase, 2),
-		    GET_BIT(source->NegateBase, 3));
+      src = negate(src,
+                   GET_BIT(source->NegateBase, 0),
+                   GET_BIT(source->NegateBase, 1),
+                   GET_BIT(source->NegateBase, 2),
+                   GET_BIT(source->NegateBase, 3));
 
    return src;
 }
 
 
-static GLuint get_result_vector( struct i915_fragment_program *p,
-				 const struct prog_instruction *inst )
+static GLuint
+get_result_vector(struct i915_fragment_program *p,
+                  const struct prog_instruction *inst)
 {
    switch (inst->DstReg.File) {
    case PROGRAM_OUTPUT:
       switch (inst->DstReg.Index) {
-      case FRAG_RESULT_COLR: 
-	 return UREG(REG_TYPE_OC, 0);
-      case FRAG_RESULT_DEPR: 
-	 p->depth_written = 1;
-	 return UREG(REG_TYPE_OD, 0);
-      default: 
-	 i915_program_error( p, "Bad inst->DstReg.Index" ); 
-	 return 0;
+      case FRAG_RESULT_COLR:
+         return UREG(REG_TYPE_OC, 0);
+      case FRAG_RESULT_DEPR:
+         p->depth_written = 1;
+         return UREG(REG_TYPE_OD, 0);
+      default:
+         i915_program_error(p, "Bad inst->DstReg.Index");
+         return 0;
       }
    case PROGRAM_TEMPORARY:
       return UREG(REG_TYPE_R, inst->DstReg.Index);
    default:
-      i915_program_error( p, "Bad inst->DstReg.File" ); 
+      i915_program_error(p, "Bad inst->DstReg.File");
       return 0;
    }
 }
-   
-static GLuint get_result_flags( const struct prog_instruction *inst )
+
+static GLuint
+get_result_flags(const struct prog_instruction *inst)
 {
    GLuint flags = 0;
 
-   if (inst->SaturateMode == SATURATE_ZERO_ONE) flags |= A0_DEST_SATURATE;
-   if (inst->DstReg.WriteMask & WRITEMASK_X) flags |= A0_DEST_CHANNEL_X;
-   if (inst->DstReg.WriteMask & WRITEMASK_Y) flags |= A0_DEST_CHANNEL_Y;
-   if (inst->DstReg.WriteMask & WRITEMASK_Z) flags |= A0_DEST_CHANNEL_Z;
-   if (inst->DstReg.WriteMask & WRITEMASK_W) flags |= A0_DEST_CHANNEL_W;
+   if (inst->SaturateMode == SATURATE_ZERO_ONE)
+      flags |= A0_DEST_SATURATE;
+   if (inst->DstReg.WriteMask & WRITEMASK_X)
+      flags |= A0_DEST_CHANNEL_X;
+   if (inst->DstReg.WriteMask & WRITEMASK_Y)
+      flags |= A0_DEST_CHANNEL_Y;
+   if (inst->DstReg.WriteMask & WRITEMASK_Z)
+      flags |= A0_DEST_CHANNEL_Z;
+   if (inst->DstReg.WriteMask & WRITEMASK_W)
+      flags |= A0_DEST_CHANNEL_W;
 
    return flags;
 }
 
-static GLuint translate_tex_src_target( struct i915_fragment_program *p,
-				     GLubyte bit )
+static GLuint
+translate_tex_src_target(struct i915_fragment_program *p, GLubyte bit)
 {
    switch (bit) {
-   case TEXTURE_1D_INDEX:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_2D_INDEX:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_RECT_INDEX: return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_3D_INDEX:   return D0_SAMPLE_TYPE_VOLUME;
-   case TEXTURE_CUBE_INDEX: return D0_SAMPLE_TYPE_CUBE;
-   default: i915_program_error(p, "TexSrcBit"); return 0;
+   case TEXTURE_1D_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_2D_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_RECT_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_3D_INDEX:
+      return D0_SAMPLE_TYPE_VOLUME;
+   case TEXTURE_CUBE_INDEX:
+      return D0_SAMPLE_TYPE_CUBE;
+   default:
+      i915_program_error(p, "TexSrcBit");
+      return 0;
    }
 }
 
@@ -211,7 +243,7 @@ do {								\
    GLuint coord = src_vector( p, &inst->SrcReg[0], program);	\
    /* Texel lookup */						\
 								\
-   i915_emit_texld( p,						\
+   i915_emit_texld( p, get_live_regs(p, inst),						\
 	       get_result_vector( p, inst ),			\
 	       get_result_flags( inst ),			\
 	       sampler,						\
@@ -234,6 +266,43 @@ do {									\
 #define EMIT_2ARG_ARITH( OP ) EMIT_ARITH( OP, 2 )
 #define EMIT_3ARG_ARITH( OP ) EMIT_ARITH( OP, 3 )
 
+/* 
+ * TODO: consider moving this into core 
+ */
+static void calc_live_regs( struct i915_fragment_program *p )
+{
+    const struct gl_fragment_program *program = p->ctx->FragmentProgram._Current;
+    GLuint regsUsed = 0xffff0000;
+    GLint i;
+   
+    for (i = program->Base.NumInstructions - 1; i >= 0; i--) {
+        struct prog_instruction *inst = &program->Base.Instructions[i];
+        int opArgs = _mesa_num_inst_src_regs(inst->Opcode);
+        int a;
+
+        /* Register is written to: unmark as live for this and preceeding ops */ 
+        if (inst->DstReg.File == PROGRAM_TEMPORARY)
+            regsUsed &= ~(1 << inst->DstReg.Index);
+
+        for (a = 0; a < opArgs; a++) {
+            /* Register is read from: mark as live for this and preceeding ops */ 
+            if (inst->SrcReg[a].File == PROGRAM_TEMPORARY)
+                regsUsed |= 1 << inst->SrcReg[a].Index;
+        }
+
+        p->usedRegs[i] = regsUsed;
+    }
+}
+
+static GLuint get_live_regs( struct i915_fragment_program *p, 
+                             const struct prog_instruction *inst )
+{
+    const struct gl_fragment_program *program = p->ctx->FragmentProgram._Current;
+    GLuint nr = inst - program->Base.Instructions;
+
+    return p->usedRegs[nr];
+}
+ 
 
 /* Possible concerns:
  *
@@ -246,9 +315,11 @@ do {									\
  * can lead to confusion -- hopefully we cope with it ok now.
  *
  */
-static void upload_program( struct i915_fragment_program *p )
+static void
+upload_program(struct i915_fragment_program *p)
 {
-   const struct gl_fragment_program *program = p->ctx->FragmentProgram._Current;
+   const struct gl_fragment_program *program =
+      p->ctx->FragmentProgram._Current;
    const struct prog_instruction *inst = program->Base.Instructions;
 
 /*    _mesa_debug_fp_inst(program->Base.NumInstructions, inst); */
@@ -258,510 +329,551 @@ static void upload_program( struct i915_fragment_program *p )
     * this being uploaded to hardware.
     */
    if (inst[0].Opcode == OPCODE_END) {
-      GLuint tmp = i915_get_utemp( p );
-      i915_emit_arith( p,
-		      A0_MOV,
-		      UREG(REG_TYPE_OC, 0), 
-		      A0_DEST_CHANNEL_ALL, 0,
-		      swizzle(tmp,ONE,ZERO,ONE,ONE), 0, 0);
+      GLuint tmp = i915_get_utemp(p);
+      i915_emit_arith(p,
+                      A0_MOV,
+                      UREG(REG_TYPE_OC, 0),
+                      A0_DEST_CHANNEL_ALL, 0,
+                      swizzle(tmp, ONE, ZERO, ONE, ONE), 0, 0);
       return;
    }
 
+   if (program->Base.NumInstructions > I915_MAX_INSN) {
+       i915_program_error( p, "Exceeded max instructions" );
+       return;
+    }
+
+   /* Not always needed:
+    */
+   calc_live_regs(p);
+
    while (1) {
       GLuint src0, src1, src2, flags;
-      GLuint tmp = 0;
+      GLuint tmp = 0, consts0 = 0, consts1 = 0;
 
       switch (inst->Opcode) {
-      case OPCODE_ABS: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 i915_emit_arith( p, 
-			 A0_MAX,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, negate(src0, 1,1,1,1), 0);
-	 break;
-
-      case OPCODE_ADD: 
-	 EMIT_2ARG_ARITH( A0_ADD );
-	 break;
-
-      case OPCODE_CMP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 src2 = src_vector( p, &inst->SrcReg[2], program);
-	 i915_emit_arith( p, 
-			 A0_CMP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, src2, src1);	/* NOTE: order of src2, src1 */
-	 break;
+      case OPCODE_ABS:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         i915_emit_arith(p,
+                         A0_MAX,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         src0, negate(src0, 1, 1, 1, 1), 0);
+         break;
 
-      case OPCODE_COS:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
+      case OPCODE_ADD:
+         EMIT_2ARG_ARITH(A0_ADD);
+         break;
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
-			 0);
+      case OPCODE_CMP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         src2 = src_vector(p, &inst->SrcReg[2], program);
+         i915_emit_arith(p, A0_CMP, get_result_vector(p, inst), get_result_flags(inst), 0, src0, src2, src1);   /* NOTE: order of src2, src1 */
+         break;
 
-	 i915_emit_arith( p, 
-			 A0_MOD,
+      case OPCODE_COS:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+	 consts0 = i915_emit_const4fv(p, sin_quad_constants[0]);
+	 consts1 = i915_emit_const4fv(p, sin_quad_constants[1]);
+
+	 /* Reduce range from repeating about [-pi,pi] to [-1,1] */
+         i915_emit_arith(p,
+                         A0_MAD,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         src0,
+			 swizzle(consts1, Z, ZERO, ZERO, ZERO), /* 1/(2pi) */
+			 swizzle(consts0, W, ZERO, ZERO, ZERO)); /* .75 */
+
+         i915_emit_arith(p, A0_FRC, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
+
+	 i915_emit_arith(p,
+			 A0_MAD,
 			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 0, 0 );
+			 tmp,
+			 swizzle(consts0, X, ZERO, ZERO, ZERO), /* 2 */
+			 swizzle(consts0, Y, ZERO, ZERO, ZERO)); /* -1 */
 
-	 /* By choosing different taylor constants, could get rid of this mul:
+	 /* Compute COS with the same calculation used for SIN, but a
+	  * different source range has been mapped to [-1,1] this time.
 	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
+
+	 /* tmp.y = abs(tmp.x); {x, abs(x), 0, 0} */
+	 i915_emit_arith(p,
+                         A0_MAX,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0),
 			 0);
 
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, 1
-	  * t0 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-	  * result = DP4 t0, cos_constants
-	  */
-	 i915_emit_arith( p, 
+	 /* tmp.y = tmp.y * tmp.x; {x, x * abs(x), 0, 0} */
+	 i915_emit_arith(p,
 			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(tmp, X,X,ONE,ONE), 
-			 swizzle(tmp, X,ONE,ONE,ONE), 0);
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 tmp,
+			 0);
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XYZ, 0,
-			 swizzle(tmp, X,Y,X,ONE), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
+	 /* tmp.x = tmp.xy DP sin_quad_constants[2].xy */
+         i915_emit_arith(p,
+                         A0_DP3,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+			 tmp,
+                         swizzle(consts1, X, Y, ZERO, ZERO),
+			 0);
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XYZ, 0,
-			 swizzle(tmp, X,X,Z,ONE), 
-			 swizzle(tmp, Z,ONE,ONE,ONE), 0);
-	    
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(tmp, ONE,Z,Y,X),
-			 i915_emit_const4fv( p, cos_constants ), 0);
-
-	 break;
-
-      case OPCODE_DP3: 
-	 EMIT_2ARG_ARITH( A0_DP3 );
-	 break;
-
-      case OPCODE_DP4: 
-	 EMIT_2ARG_ARITH( A0_DP4 );
-	 break;
-
-      case OPCODE_DPH:  
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0, X,Y,Z,ONE), src1, 0);
-	 break;
-
-      case OPCODE_DST: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 /* result[0] = 1    * 1;
-	  * result[1] = a[1] * b[1];
-	  * result[2] = a[2] * 1;
-	  * result[3] = 1    * b[3];
+	 /* tmp.x now contains a first approximation (y).  Now, weight it
+	  * against tmp.y**2 to get closer.
 	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0, ONE, Y, Z,   ONE), 
-			 swizzle(src1, ONE, Y, ONE, W  ),
+	 i915_emit_arith(p,
+                         A0_MAX,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0),
 			 0);
-	 break;
 
-      case OPCODE_EX2: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
+	 /* tmp.y = tmp.x * tmp.y - tmp.x; {y, y * abs(y) - y, 0, 0} */
+	 i915_emit_arith(p,
+			 A0_MAD,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 swizzle(tmp, ZERO, Y, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0));
+
+	 /* result = .2225 * tmp.y + tmp.x =.2225(y * abs(y) - y) + y= */
+	 i915_emit_arith(p,
+			 A0_MAD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+			 swizzle(consts1, W, W, W, W),
+			 swizzle(tmp, Y, Y, Y, Y),
+			 swizzle(tmp, X, X, X, X));
+         break;
+
+      case OPCODE_DP3:
+         EMIT_2ARG_ARITH(A0_DP3);
+         break;
+
+      case OPCODE_DP4:
+         EMIT_2ARG_ARITH(A0_DP4);
+         break;
+
+      case OPCODE_DPH:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
 
-	 i915_emit_arith( p, 
-			 A0_EXP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
+         i915_emit_arith(p,
+                         A0_DP4,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, Y, Z, ONE), src1, 0);
+         break;
+
+      case OPCODE_DST:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+
+         /* result[0] = 1    * 1;
+          * result[1] = a[1] * b[1];
+          * result[2] = a[2] * 1;
+          * result[3] = 1    * b[3];
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, ONE, Y, Z, ONE),
+                         swizzle(src1, ONE, Y, ONE, W), 0);
+         break;
 
-      case OPCODE_FLR: 
-	 EMIT_1ARG_ARITH( A0_FLR );
-	 break;
+      case OPCODE_EX2:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
 
-      case OPCODE_FRC: 
-	 EMIT_1ARG_ARITH( A0_FRC );
-	 break;
+         i915_emit_arith(p,
+                         A0_EXP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_FLR:
+         EMIT_1ARG_ARITH(A0_FLR);
+         break;
+
+      case OPCODE_FRC:
+         EMIT_1ARG_ARITH(A0_FRC);
+         break;
 
       case OPCODE_KIL:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 i915_emit_texld( p,
-			 tmp, A0_DEST_CHANNEL_ALL, /* use a dummy dest reg */
-			 0,
-			 src0,
-			 T0_TEXKILL );
-	 break;
-
-      case OPCODE_LG2: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-
-	 i915_emit_arith( p, 
-			 A0_LOG,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-
-      case OPCODE_LIT: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 /* tmp = max( a.xyzw, a.00zw )
-	  * XXX: Clamp tmp.w to -128..128
-	  * tmp.y = log(tmp.y)
-	  * tmp.y = tmp.w * tmp.y
-	  * tmp.y = exp(tmp.y)
-	  * result = cmp (a.11-x1, a.1x01, a.1xy1 )
-	  */
-	 i915_emit_arith( p, A0_MAX, tmp, A0_DEST_CHANNEL_ALL, 0, 
-			 src0, swizzle(src0, ZERO, ZERO, Z, W), 0 );
-
-	 i915_emit_arith( p, A0_LOG, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, Y, Y, Y, Y), 0, 0 );
-
-	 i915_emit_arith( p, A0_MUL, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, ZERO, Y, ZERO, ZERO), 
-			 swizzle(tmp, ZERO, W, ZERO, ZERO), 0 );
-
-	 i915_emit_arith( p, A0_EXP, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, Y, Y, Y, Y), 0, 0 );
-
-	 i915_emit_arith( p, A0_CMP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 negate(swizzle(tmp, ONE, ONE, X, ONE),0,0,1,0),
-			 swizzle(tmp, ONE, X, ZERO, ONE),
-			 swizzle(tmp, ONE, X, Y, ONE));
-		     
-	 break;
-
-      case OPCODE_LRP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 src2 = src_vector( p, &inst->SrcReg[2], program);
-	 flags = get_result_flags( inst );
-	 tmp = i915_get_utemp( p );
-
-	 /* b*a + c*(1-a)
-	  *
-	  * b*a + c - ca 
-	  *
-	  * tmp = b*a + c, 
-	  * result = (-c)*a + tmp 
-	  */
-	 i915_emit_arith( p, A0_MAD, tmp, 
-			 flags & A0_DEST_CHANNEL_ALL, 0,
-			 src1, src0, src2 );
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_texld(p, get_live_regs(p, inst),
+                         tmp, A0_DEST_CHANNEL_ALL,   /* use a dummy dest reg */
+                         0, src0, T0_TEXKILL);
+         break;
+
+      case OPCODE_LG2:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_LOG,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_LIT:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         /* tmp = max( a.xyzw, a.00zw )
+          * XXX: Clamp tmp.w to -128..128
+          * tmp.y = log(tmp.y)
+          * tmp.y = tmp.w * tmp.y
+          * tmp.y = exp(tmp.y)
+          * result = cmp (a.11-x1, a.1x01, a.1xy1 )
+          */
+         i915_emit_arith(p, A0_MAX, tmp, A0_DEST_CHANNEL_ALL, 0,
+                         src0, swizzle(src0, ZERO, ZERO, Z, W), 0);
+
+         i915_emit_arith(p, A0_LOG, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
+
+         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, ZERO, Y, ZERO, ZERO),
+                         swizzle(tmp, ZERO, W, ZERO, ZERO), 0);
+
+         i915_emit_arith(p, A0_EXP, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
+
+         i915_emit_arith(p, A0_CMP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         negate(swizzle(tmp, ONE, ONE, X, ONE), 0, 0, 1, 0),
+                         swizzle(tmp, ONE, X, ZERO, ONE),
+                         swizzle(tmp, ONE, X, Y, ONE));
 
-	 i915_emit_arith( p, A0_MAD, 
-			 get_result_vector( p, inst ), 
-			 flags, 0, 
-			 negate(src2, 1,1,1,1), src0, tmp );
-	 break;
+         break;
+
+      case OPCODE_LRP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         src2 = src_vector(p, &inst->SrcReg[2], program);
+         flags = get_result_flags(inst);
+         tmp = i915_get_utemp(p);
+
+         /* b*a + c*(1-a)
+          *
+          * b*a + c - ca 
+          *
+          * tmp = b*a + c, 
+          * result = (-c)*a + tmp 
+          */
+         i915_emit_arith(p, A0_MAD, tmp,
+                         flags & A0_DEST_CHANNEL_ALL, 0, src1, src0, src2);
+
+         i915_emit_arith(p, A0_MAD,
+                         get_result_vector(p, inst),
+                         flags, 0, negate(src2, 1, 1, 1, 1), src0, tmp);
+         break;
 
       case OPCODE_MAD:
-	 EMIT_3ARG_ARITH( A0_MAD );
-	 break;
+         EMIT_3ARG_ARITH(A0_MAD);
+         break;
 
       case OPCODE_MAX:
-	 EMIT_2ARG_ARITH( A0_MAX );
-	 break;
-
-      case OPCODE_MIN: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 flags = get_result_flags( inst );
-
-	 i915_emit_arith( p, 
-			 A0_MAX,
-			 tmp, flags & A0_DEST_CHANNEL_ALL, 0,
-			 negate(src0,1,1,1,1), 
-			 negate(src1,1,1,1,1), 0);
-
-	 i915_emit_arith( p,
-			 A0_MOV,
-			 get_result_vector( p, inst ), 
-			 flags, 0,
-			 negate(tmp, 1,1,1,1), 0, 0);
-	 break;
-
-      case OPCODE_MOV: 
-	 EMIT_1ARG_ARITH( A0_MOV );
-	 break;
-
-      case OPCODE_MUL: 
-	 EMIT_2ARG_ARITH( A0_MUL );
-	 break;
-
-      case OPCODE_POW: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 flags = get_result_flags( inst );
-
-	 /* XXX: masking on intermediate values, here and elsewhere.
-	  */
-	 i915_emit_arith( p, 
-			 A0_LOG,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
+         EMIT_2ARG_ARITH(A0_MAX);
+         break;
 
-	 i915_emit_arith( p,
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, src1, 0);
+      case OPCODE_MIN:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+         flags = get_result_flags(inst);
+
+         i915_emit_arith(p,
+                         A0_MAX,
+                         tmp, flags & A0_DEST_CHANNEL_ALL, 0,
+                         negate(src0, 1, 1, 1, 1),
+                         negate(src1, 1, 1, 1, 1), 0);
+
+         i915_emit_arith(p,
+                         A0_MOV,
+                         get_result_vector(p, inst),
+                         flags, 0, negate(tmp, 1, 1, 1, 1), 0, 0);
+         break;
 
+      case OPCODE_MOV:
+         EMIT_1ARG_ARITH(A0_MOV);
+         break;
 
-	 i915_emit_arith( p,
-			 A0_EXP,
-			 get_result_vector( p, inst ), 
-			 flags, 0,
-			 swizzle(tmp,X,X,X,X), 0, 0);
+      case OPCODE_MUL:
+         EMIT_2ARG_ARITH(A0_MUL);
+         break;
 
-	 break;
+      case OPCODE_POW:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+         flags = get_result_flags(inst);
 
-      case OPCODE_RCP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
+         /* XXX: masking on intermediate values, here and elsewhere.
+          */
+         i915_emit_arith(p,
+                         A0_LOG,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
 
-	 i915_emit_arith( p, 
-			 A0_RCP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
+         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_X, 0, tmp, src1, 0);
 
-      case OPCODE_RSQ: 
 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
+         i915_emit_arith(p,
+                         A0_EXP,
+                         get_result_vector(p, inst),
+                         flags, 0, swizzle(tmp, X, X, X, X), 0, 0);
+
+         break;
+
+      case OPCODE_RCP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_RCP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_RSQ:
+
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_RSQ,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
 
-	 i915_emit_arith( p, 
-			 A0_RSQ,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-	 
       case OPCODE_SCS:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-	  * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-	  * scs.x = DP4 t1, sin_constants
-	  * t1 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-	  * scs.y = DP4 t1, cos_constants
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(src0, X,X,ONE,ONE), 
-			 swizzle(src0, X,ONE,ONE,ONE), 0);
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         /* 
+          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
+          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
+          * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
+          * scs.x = DP4 t1, sin_constants
+          * t1 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
+          * scs.y = DP4 t1, cos_constants
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XY, 0,
+                         swizzle(src0, X, X, ONE, ONE),
+                         swizzle(src0, X, ONE, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(tmp, X, Y, X, Y),
+                         swizzle(tmp, X, X, ONE, ONE), 0);
+
+         if (inst->DstReg.WriteMask & WRITEMASK_Y) {
+            GLuint tmp1;
+
+            if (inst->DstReg.WriteMask & WRITEMASK_X)
+               tmp1 = i915_get_utemp(p);
+            else
+               tmp1 = tmp;
+
+            i915_emit_arith(p,
+                            A0_MUL,
+                            tmp1, A0_DEST_CHANNEL_ALL, 0,
+                            swizzle(tmp, X, Y, Y, W),
+                            swizzle(tmp, X, Z, ONE, ONE), 0);
+
+            i915_emit_arith(p,
+                            A0_DP4,
+                            get_result_vector(p, inst),
+                            A0_DEST_CHANNEL_Y, 0,
+                            swizzle(tmp1, W, Z, Y, X),
+                            i915_emit_const4fv(p, sin_constants), 0);
+         }
+
+         if (inst->DstReg.WriteMask & WRITEMASK_X) {
+            i915_emit_arith(p,
+                            A0_MUL,
+                            tmp, A0_DEST_CHANNEL_XYZ, 0,
+                            swizzle(tmp, X, X, Z, ONE),
+                            swizzle(tmp, Z, ONE, ONE, ONE), 0);
+
+            i915_emit_arith(p,
+                            A0_DP4,
+                            get_result_vector(p, inst),
+                            A0_DEST_CHANNEL_X, 0,
+                            swizzle(tmp, ONE, Z, Y, X),
+                            i915_emit_const4fv(p, cos_constants), 0);
+         }
+         break;
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,X,Y), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
-
-	 if (inst->DstReg.WriteMask & WRITEMASK_Y) {
-	    GLuint tmp1;
-	    
-	    if (inst->DstReg.WriteMask & WRITEMASK_X)
-	       tmp1 = i915_get_utemp( p );
-	    else
-	       tmp1 = tmp;
-
-	    i915_emit_arith( p, 
-			    A0_MUL,
-			    tmp1, A0_DEST_CHANNEL_ALL, 0,
-			    swizzle(tmp, X,Y,Y,W), 
-			    swizzle(tmp, X,Z,ONE,ONE), 0);
-	    
-	    i915_emit_arith( p, 
-			    A0_DP4,
-			    get_result_vector( p, inst ), 
-			    A0_DEST_CHANNEL_Y, 0,
-			    swizzle(tmp1, W,Z,Y,X),
-			    i915_emit_const4fv( p, sin_constants ), 0);
-	 }
-
-	 if (inst->DstReg.WriteMask & WRITEMASK_X) {
-	    i915_emit_arith( p, 
-			    A0_MUL,
-			    tmp, A0_DEST_CHANNEL_XYZ, 0,
-			    swizzle(tmp, X,X,Z,ONE), 
-			    swizzle(tmp, Z,ONE,ONE,ONE), 0);
-	    
-	    i915_emit_arith( p, 
-			    A0_DP4,
-			    get_result_vector( p, inst ), 
-			    A0_DEST_CHANNEL_X, 0,
-			    swizzle(tmp, ONE,Z,Y,X), 
-			    i915_emit_const4fv( p, cos_constants ), 0);
-	 }
-	 break;
-
-      case OPCODE_SGE: 
-	 EMIT_2ARG_ARITH( A0_SGE );
-	 break;
+      case OPCODE_SGE:
+         EMIT_2ARG_ARITH(A0_SGE);
+         break;
 
       case OPCODE_SIN:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+	 consts0 = i915_emit_const4fv(p, sin_quad_constants[0]);
+	 consts1 = i915_emit_const4fv(p, sin_quad_constants[1]);
+
+	 /* Reduce range from repeating about [-pi,pi] to [-1,1] */
+         i915_emit_arith(p,
+                         A0_MAD,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         src0,
+			 swizzle(consts1, Z, ZERO, ZERO, ZERO), /* 1/(2pi) */
+			 swizzle(consts0, Z, ZERO, ZERO, ZERO)); /* .5 */
+
+         i915_emit_arith(p, A0_FRC, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
+
+	 i915_emit_arith(p,
+			 A0_MAD,
+			 tmp, A0_DEST_CHANNEL_X, 0,
+			 tmp,
+			 swizzle(consts0, X, ZERO, ZERO, ZERO), /* 2 */
+			 swizzle(consts0, Y, ZERO, ZERO, ZERO)); /* -1 */
 
-	 i915_emit_arith( p, 
+	 /* Compute sin using a quadratic and quartic.  It gives continuity
+	  * that repeating the Taylor series lacks every 2*pi, and has
+	  * reduced error.
+	  *
+	  * The idea was described at:
+	  * http://www.devmaster.net/forums/showthread.php?t=5784
+	  */
+
+	 /* tmp.y = abs(tmp.x); {x, abs(x), 0, 0} */
+	 i915_emit_arith(p,
+                         A0_MAX,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0),
+			 0);
+
+	 /* tmp.y = tmp.y * tmp.x; {x, x * abs(x), 0, 0} */
+	 i915_emit_arith(p,
 			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 tmp,
 			 0);
 
-	 i915_emit_arith( p, 
-			 A0_MOD,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 0, 0 );
+	 /* tmp.x = tmp.xy DP sin_quad_constants[2].xy */
+         i915_emit_arith(p,
+                         A0_DP3,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+			 tmp,
+                         swizzle(consts1, X, Y, ZERO, ZERO),
+			 0);
 
-	 /* By choosing different taylor constants, could get rid of this mul:
+	 /* tmp.x now contains a first approximation (y).  Now, weight it
+	  * against tmp.y**2 to get closer.
 	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
+	 i915_emit_arith(p,
+                         A0_MAX,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0),
 			 0);
 
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-	  * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-	  * result = DP4 t1.wzyx, sin_constants
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(tmp, X,X,ONE,ONE), 
-			 swizzle(tmp, X,ONE,ONE,ONE), 0);
+	 /* tmp.y = tmp.x * tmp.y - tmp.x; {y, y * abs(y) - y, 0, 0} */
+	 i915_emit_arith(p,
+			 A0_MAD,
+			 tmp, A0_DEST_CHANNEL_Y, 0,
+			 swizzle(tmp, ZERO, X, ZERO, ZERO),
+			 swizzle(tmp, ZERO, Y, ZERO, ZERO),
+			 negate(swizzle(tmp, ZERO, X, ZERO, ZERO), 0, 1, 0, 0));
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,X,Y), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
+	 /* result = .2225 * tmp.y + tmp.x =.2225(y * abs(y) - y) + y= */
+	 i915_emit_arith(p,
+			 A0_MAD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+			 swizzle(consts1, W, W, W, W),
+			 swizzle(tmp, Y, Y, Y, Y),
+			 swizzle(tmp, X, X, X, X));
 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,Y,W), 
-			 swizzle(tmp, X,Z,ONE,ONE), 0);
-	    
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(tmp, W, Z, Y, X ),
-			 i915_emit_const4fv( p, sin_constants ), 0);
-	 break;
-
-      case OPCODE_SLT: 
-	 EMIT_2ARG_ARITH( A0_SLT );
-	 break;
-
-      case OPCODE_SUB: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 i915_emit_arith( p, 
-			 A0_ADD,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, negate(src1, 1,1,1,1), 0);
-	 break;
-
-      case OPCODE_SWZ: 
-	 EMIT_1ARG_ARITH( A0_MOV ); /* extended swizzle handled natively */
-	 break;
-
-      case OPCODE_TEX: 
-	 EMIT_TEX( T0_TEXLD );
-	 break;
+         break;
+
+      case OPCODE_SLT:
+         EMIT_2ARG_ARITH(A0_SLT);
+         break;
+
+      case OPCODE_SUB:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+
+         i915_emit_arith(p,
+                         A0_ADD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         src0, negate(src1, 1, 1, 1, 1), 0);
+         break;
+
+      case OPCODE_SWZ:
+         EMIT_1ARG_ARITH(A0_MOV);       /* extended swizzle handled natively */
+         break;
+
+      case OPCODE_TEX:
+         EMIT_TEX(T0_TEXLD);
+         break;
 
       case OPCODE_TXB:
-	 EMIT_TEX( T0_TEXLDB );
-	 break;
+         EMIT_TEX(T0_TEXLDB);
+         break;
 
       case OPCODE_TXP:
-	 EMIT_TEX( T0_TEXLDP );
-	 break;
+         EMIT_TEX(T0_TEXLDP);
+         break;
 
       case OPCODE_XPD:
-	 /* Cross product:
-	  *      result.x = src0.y * src1.z - src0.z * src1.y;
-	  *      result.y = src0.z * src1.x - src0.x * src1.z;
-	  *      result.z = src0.x * src1.y - src0.y * src1.x;
-	  *      result.w = undef;
-	  */
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(src0,Z,X,Y,ONE), 
-			 swizzle(src1,Y,Z,X,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MAD,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,Y,Z,X,ONE), 
-			 swizzle(src1,Z,X,Y,ONE), 
-			 negate(tmp,1,1,1,0));
-	 break;
+         /* Cross product:
+          *      result.x = src0.y * src1.z - src0.z * src1.y;
+          *      result.y = src0.z * src1.x - src0.x * src1.z;
+          *      result.z = src0.x * src1.y - src0.y * src1.x;
+          *      result.w = undef;
+          */
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(src0, Z, X, Y, ONE),
+                         swizzle(src1, Y, Z, X, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MAD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, Y, Z, X, ONE),
+                         swizzle(src1, Z, X, Y, ONE),
+                         negate(tmp, 1, 1, 1, 0));
+         break;
 
       case OPCODE_END:
-	 return;
-	 
+         return;
+
       default:
-	 i915_program_error( p, "bad opcode" );
-	 return;
+         i915_program_error(p, "bad opcode");
+         return;
       }
 
       inst++;
-      i915_release_utemps( p ); 
+      i915_release_utemps(p);
    }
 }
 
@@ -769,21 +881,22 @@ static void upload_program( struct i915_fragment_program *p )
  * emit, just move the value into its correct position at the end of
  * the program:
  */
-static void fixup_depth_write( struct i915_fragment_program *p )
+static void
+fixup_depth_write(struct i915_fragment_program *p)
 {
    if (p->depth_written) {
       GLuint depth = UREG(REG_TYPE_OD, 0);
 
-      i915_emit_arith( p, 
-		      A0_MOV,
-		      depth, A0_DEST_CHANNEL_W, 0,
-		      swizzle(depth,X,Y,Z,Z), 
-		      0, 0);
+      i915_emit_arith(p,
+                      A0_MOV,
+                      depth, A0_DEST_CHANNEL_W, 0,
+                      swizzle(depth, X, Y, Z, Z), 0, 0);
    }
 }
 
 
-static void check_wpos( struct i915_fragment_program *p )
+static void
+check_wpos(struct i915_fragment_program *p)
 {
    GLuint inputs = p->FragProg.Base.InputsRead;
    GLint i;
@@ -791,12 +904,12 @@ static void check_wpos( struct i915_fragment_program *p )
    p->wpos_tex = -1;
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
-      if (inputs & FRAG_BIT_TEX(i)) 
-	 continue;
+      if (inputs & FRAG_BIT_TEX(i))
+         continue;
       else if (inputs & FRAG_BIT_WPOS) {
-	 p->wpos_tex = i;
-	 inputs &= ~FRAG_BIT_WPOS;
-      }   
+         p->wpos_tex = i;
+         inputs &= ~FRAG_BIT_WPOS;
+      }
    }
 
    if (inputs & FRAG_BIT_WPOS) {
@@ -805,53 +918,54 @@ static void check_wpos( struct i915_fragment_program *p )
 }
 
 
-static void translate_program( struct i915_fragment_program *p )
+static void
+translate_program(struct i915_fragment_program *p)
 {
-   i915ContextPtr i915 = I915_CONTEXT(p->ctx);
-   
-   i915_init_program( i915, p );
-   check_wpos( p ); 
-   upload_program( p );
-   fixup_depth_write( p );
-   i915_fini_program( p ); 
-   
+   struct i915_context *i915 = I915_CONTEXT(p->ctx);
+
+   i915_init_program(i915, p);
+   check_wpos(p);
+   upload_program(p);
+   fixup_depth_write(p);
+   i915_fini_program(p);
+
    p->translated = 1;
 }
 
 
-static void track_params( struct i915_fragment_program *p )
+static void
+track_params(struct i915_fragment_program *p)
 {
    GLint i;
 
    if (p->nr_params)
-      _mesa_load_state_parameters(p->ctx, p->FragProg.Base.Parameters); 
+      _mesa_load_state_parameters(p->ctx, p->FragProg.Base.Parameters);
 
    for (i = 0; i < p->nr_params; i++) {
       GLint reg = p->param[i].reg;
-      COPY_4V( p->constant[reg], p->param[i].values );
+      COPY_4V(p->constant[reg], p->param[i].values);
    }
-   
+
    p->params_uptodate = 1;
-   p->on_hardware = 0;		/* overkill */
+   p->on_hardware = 0;          /* overkill */
 }
 
 
-static void i915BindProgram( GLcontext *ctx,
-			    GLenum target, 
-			    struct gl_program *prog )
+static void
+i915BindProgram(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      i915ContextPtr i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_context *i915 = I915_CONTEXT(ctx);
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
+
+      if (i915->current_program == p)
+         return;
 
-      if (i915->current_program == p) 
-	 return;
-      
       if (i915->current_program) {
-	 i915->current_program->on_hardware = 0;
-	 i915->current_program->params_uptodate = 0;
+         i915->current_program->on_hardware = 0;
+         i915->current_program->params_uptodate = 0;
       }
-      
+
       i915->current_program = p;
 
       assert(p->on_hardware == 0);
@@ -860,71 +974,70 @@ static void i915BindProgram( GLcontext *ctx,
    }
 }
 
-static struct gl_program *i915NewProgram( GLcontext *ctx,
-				      GLenum target, 
-				      GLuint id )
+static struct gl_program *
+i915NewProgram(GLcontext * ctx, GLenum target, GLuint id)
 {
    switch (target) {
    case GL_VERTEX_PROGRAM_ARB:
-      return _mesa_init_vertex_program( ctx, CALLOC_STRUCT(gl_vertex_program),
-					target, id );
-
-   case GL_FRAGMENT_PROGRAM_ARB: {
-      struct i915_fragment_program *prog = CALLOC_STRUCT(i915_fragment_program);
-      if (prog) {
-	 i915_init_program( I915_CONTEXT(ctx), prog );
-
-	 return _mesa_init_fragment_program( ctx, &prog->FragProg,
-					     target, id );
+      return _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
+                                       target, id);
+
+   case GL_FRAGMENT_PROGRAM_ARB:{
+         struct i915_fragment_program *prog =
+            CALLOC_STRUCT(i915_fragment_program);
+         if (prog) {
+            i915_init_program(I915_CONTEXT(ctx), prog);
+
+            return _mesa_init_fragment_program(ctx, &prog->FragProg,
+                                               target, id);
+         }
+         else
+            return NULL;
       }
-      else
-	 return NULL;
-   }
 
    default:
       /* Just fallback:
        */
-      return _mesa_new_program( ctx, target, id );
+      return _mesa_new_program(ctx, target, id);
    }
 }
 
-static void i915DeleteProgram( GLcontext *ctx,
-			      struct gl_program *prog )
+static void
+i915DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 {
    if (prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
-      i915ContextPtr i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
-      
-      if (i915->current_program == p) 
-	 i915->current_program = 0;
+      struct i915_context *i915 = I915_CONTEXT(ctx);
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
+
+      if (i915->current_program == p)
+         i915->current_program = 0;
    }
 
-   _mesa_delete_program( ctx, prog );
+   _mesa_delete_program(ctx, prog);
 }
 
 
-static GLboolean i915IsProgramNative( GLcontext *ctx,
-				     GLenum target, 
-				     struct gl_program *prog )
+static GLboolean
+i915IsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
 
       if (!p->translated)
-	 translate_program( p );
-      
+         translate_program(p);
+
       return !p->error;
    }
    else
       return GL_TRUE;
 }
 
-static void i915ProgramStringNotify( GLcontext *ctx,
-				    GLenum target,
-				    struct gl_program *prog )
+static void
+i915ProgramStringNotify(GLcontext * ctx,
+                        GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
       p->translated = 0;
 
       /* Hack: make sure fog is correctly enabled according to this
@@ -941,28 +1054,28 @@ static void i915ProgramStringNotify( GLcontext *ctx,
 }
 
 
-void i915ValidateFragmentProgram( i915ContextPtr i915 )
+void
+i915ValidateFragmentProgram(struct i915_context *i915)
 {
    GLcontext *ctx = &i915->intel.ctx;
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
 
-   struct i915_fragment_program *p = 
-      (struct i915_fragment_program *)ctx->FragmentProgram._Current;
+   struct i915_fragment_program *p =
+      (struct i915_fragment_program *) ctx->FragmentProgram._Current;
 
    const GLuint inputsRead = p->FragProg.Base.InputsRead;
    GLuint s4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_VFMT_MASK;
    GLuint s2 = S2_TEXCOORD_NONE;
    int i, offset = 0;
 
-   if (i915->current_program != p) 
-   {
+   if (i915->current_program != p) {
       if (i915->current_program) {
-	 i915->current_program->on_hardware = 0;
-	 i915->current_program->params_uptodate = 0;
+         i915->current_program->on_hardware = 0;
+         i915->current_program->params_uptodate = 0;
       }
-      
+
       i915->current_program = p;
    }
 
@@ -971,8 +1084,8 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
     */
    VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
 
-   if (!p->translated) 
-      translate_program( p );
+   if (!p->translated)
+      translate_program(p);
 
    intel->vertex_attr_count = 0;
    intel->wpos_offset = 0;
@@ -981,31 +1094,31 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
    intel->specoffset = 0;
 
    if (inputsRead & FRAG_BITS_TEX_ANY) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16);
    }
    else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12 );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12);
    }
 
    if (inputsRead & FRAG_BIT_COL0) {
       intel->coloroffset = offset / 4;
-      EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4 );
+      EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4);
    }
-   
-   if ((inputsRead & (FRAG_BIT_COL1|FRAG_BIT_FOGC)) || 
+
+   if ((inputsRead & (FRAG_BIT_COL1 | FRAG_BIT_FOGC)) ||
        i915->vertex_fog != I915_FOG_NONE) {
 
       if (inputsRead & FRAG_BIT_COL1) {
-	 intel->specoffset = offset / 4;
-	 EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3 );
+         intel->specoffset = offset / 4;
+         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3);
       }
       else
-	 EMIT_PAD(3);
+         EMIT_PAD(3);
 
-      if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE) 
-	 EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1 );
+      if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE)
+         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1);
       else
-	 EMIT_PAD( 1 );
+         EMIT_PAD(1);
    }
 
    /* XXX this was disabled, but enabling this code helped fix the Glean
@@ -1013,63 +1126,66 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
     */
 #if 1
    if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE) {
-      EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1F, S4_VFMT_FOG_PARAM, 4 );
+      EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1F, S4_VFMT_FOG_PARAM, 4);
    }
 #endif
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
       if (inputsRead & FRAG_BIT_TEX(i)) {
-	 int sz = VB->TexCoordPtr[i]->size;
-	    
-	 s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	 s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
+         int sz = VB->TexCoordPtr[i]->size;
 
-	 EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_SZ(sz), 0, sz * 4 );
+         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
+         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
+
+         EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, EMIT_SZ(sz), 0, sz * 4);
       }
       else if (i == p->wpos_tex) {
-	
-	 /* If WPOS is required, duplicate the XYZ position data in an
-	  * unused texture coordinate:
-	  */
-	 s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	 s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(3));
 
-	 intel->wpos_offset = offset;
-	 intel->wpos_size = 3 * sizeof(GLuint);
+         /* If WPOS is required, duplicate the XYZ position data in an
+          * unused texture coordinate:
+          */
+         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
+         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(3));
+
+         intel->wpos_offset = offset;
+         intel->wpos_size = 3 * sizeof(GLuint);
 
-	 EMIT_PAD( intel->wpos_size );
-      }   
+         EMIT_PAD(intel->wpos_size);
+      }
    }
 
    if (s2 != i915->state.Ctx[I915_CTXREG_LIS2] ||
        s4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-    
-      I915_STATECHANGE( i915, I915_UPLOAD_CTX );
+      int k;
+
+      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
       /* Must do this *after* statechange, so as not to affect
        * buffered vertices reliant on the old state:
        */
-      intel->vertex_size = _tnl_install_attrs( &intel->ctx, 
-					       intel->vertex_attrs, 
-					       intel->vertex_attr_count,
-					       intel->ViewportMatrix.m, 0 ); 
+      intel->vertex_size = _tnl_install_attrs(&intel->ctx,
+                                              intel->vertex_attrs,
+                                              intel->vertex_attr_count,
+                                              intel->ViewportMatrix.m, 0);
 
       intel->vertex_size >>= 2;
 
       i915->state.Ctx[I915_CTXREG_LIS2] = s2;
       i915->state.Ctx[I915_CTXREG_LIS4] = s4;
 
-      assert(intel->vtbl.check_vertex_size( intel, intel->vertex_size ));
+      k = intel->vtbl.check_vertex_size(intel, intel->vertex_size);
+      assert(k);
    }
 
-   if (!p->params_uptodate) 
-      track_params( p );
+   if (!p->params_uptodate)
+      track_params(p);
 
-   if (!p->on_hardware) 
-      i915_upload_program( i915, p );
+   if (!p->on_hardware)
+      i915_upload_program(i915, p);
 }
 
-void i915InitFragProgFuncs( struct dd_function_table *functions )
+void
+i915InitFragProgFuncs(struct dd_function_table *functions)
 {
    functions->BindProgram = i915BindProgram;
    functions->NewProgram = i915NewProgram;
diff --git a/src/mesa/drivers/dri/i915/i915_metaops.c b/src/mesa/drivers/dri/i915/i915_metaops.c
index 1be7ac4c485..90a78c6082b 100644
--- a/src/mesa/drivers/dri/i915/i915_metaops.c
+++ b/src/mesa/drivers/dri/i915/i915_metaops.c
@@ -25,137 +25,177 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "utils.h"
 
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
-#include "intel_ioctl.h"
-#include "intel_rotate.h"
+#include "intel_regions.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-/* A large amount of state doesn't need to be uploaded.
+/* We touch almost everything:
  */
-#define ACTIVE (I915_UPLOAD_INVARIENT |         \
-		I915_UPLOAD_PROGRAM | 		\
-		I915_UPLOAD_STIPPLE |		\
+#define ACTIVE (I915_UPLOAD_INVARIENT | 	\
 		I915_UPLOAD_CTX |		\
 		I915_UPLOAD_BUFFERS |		\
-		I915_UPLOAD_TEX(0))		
+		I915_UPLOAD_STIPPLE |		\
+                I915_UPLOAD_PROGRAM | 		\
+                I915_UPLOAD_FOG | 		\
+		I915_UPLOAD_TEX(0))
 
-#define SET_STATE( i915, STATE )			\
+#define SET_STATE( i915, STATE )		\
 do {						\
    i915->current->emitted &= ~ACTIVE;		\
-   i915->current = &i915->STATE;			\
+   i915->current = &i915->STATE;		\
    i915->current->emitted &= ~ACTIVE;		\
 } while (0)
 
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void set_initial_state( i915ContextPtr i915 )
-{
-   memcpy(&i915->meta, &i915->initial, sizeof(i915->meta) );
-   i915->meta.active = ACTIVE;
-   i915->meta.emitted = 0;
-}
 
-
-static void set_no_depth_stencil_write( i915ContextPtr i915 )
+static void
+meta_no_stencil_write(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
     */
-   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE | 
-				       S5_STENCIL_WRITE_ENABLE);
+   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
+                                         S5_STENCIL_WRITE_ENABLE);
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+static void
+meta_no_depth_write(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
     */
    i915->meta.Ctx[I915_CTXREG_LIS6] &= ~(S6_DEPTH_TEST_ENABLE |
-				       S6_DEPTH_WRITE_ENABLE);
+                                         S6_DEPTH_WRITE_ENABLE);
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
+static void
+meta_depth_replace(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
+   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_TRUE )
+    * ctx->Driver.DepthMask( ctx, GL_TRUE )
+    */
+   i915->meta.Ctx[I915_CTXREG_LIS6] |= (S6_DEPTH_TEST_ENABLE |
+                                        S6_DEPTH_WRITE_ENABLE);
+
+   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
+    */
+   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
+   i915->meta.Ctx[I915_CTXREG_LIS6] |=
+      COMPAREFUNC_ALWAYS << S6_DEPTH_TEST_FUNC_SHIFT;
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+
 /* Set stencil unit to replace always with the reference value.
  */
-static void set_stencil_replace( i915ContextPtr i915,
-				 GLuint s_mask,
-				 GLuint s_clear)
+static void
+meta_stencil_replace(struct intel_context *intel,
+                     GLuint s_mask, GLuint s_clear)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    GLuint op = STENCILOP_REPLACE;
    GLuint func = COMPAREFUNC_ALWAYS;
 
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
     */
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE | 
-				      S5_STENCIL_WRITE_ENABLE);
-
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~(S6_DEPTH_TEST_ENABLE |
-				       S6_DEPTH_WRITE_ENABLE);
-
+   i915->meta.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
+                                        S5_STENCIL_WRITE_ENABLE);
 
    /* ctx->Driver.StencilMask( ctx, s_mask )
     */
    i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
 
    i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(s_mask));
-
+                                          STENCIL_WRITE_MASK(s_mask));
 
    /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
     */
    i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-				       S5_STENCIL_PASS_Z_FAIL_MASK |
-				       S5_STENCIL_PASS_Z_PASS_MASK);
+                                         S5_STENCIL_PASS_Z_FAIL_MASK |
+                                         S5_STENCIL_PASS_Z_PASS_MASK);
 
    i915->meta.Ctx[I915_CTXREG_LIS5] |= ((op << S5_STENCIL_FAIL_SHIFT) |
-				      (op << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-				      (op << S5_STENCIL_PASS_Z_PASS_SHIFT));
+                                        (op << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
+                                        (op << S5_STENCIL_PASS_Z_PASS_SHIFT));
 
 
    /* ctx->Driver.StencilFunc( ctx, GL_ALWAYS, s_ref, ~0 )
     */
    i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					STENCIL_TEST_MASK(0xff));
+                                          STENCIL_TEST_MASK(0xff));
 
    i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-				       S5_STENCIL_TEST_FUNC_MASK);
-					
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((s_clear << S5_STENCIL_REF_SHIFT) |  
-				      (func << S5_STENCIL_TEST_FUNC_SHIFT)); 
+                                         S5_STENCIL_TEST_FUNC_MASK);
+
+   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((s_clear << S5_STENCIL_REF_SHIFT) |
+                                        (func << S5_STENCIL_TEST_FUNC_SHIFT));
 
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
 
-static void set_color_mask( i915ContextPtr i915, GLboolean state )
+static void
+meta_color_mask(struct intel_context *intel, GLboolean state)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    const GLuint mask = (S5_WRITEDISABLE_RED |
-			S5_WRITEDISABLE_GREEN |
-			S5_WRITEDISABLE_BLUE |
-			S5_WRITEDISABLE_ALPHA);
+                        S5_WRITEDISABLE_GREEN |
+                        S5_WRITEDISABLE_BLUE | S5_WRITEDISABLE_ALPHA);
 
    /* Copy colormask state from "regular" hw context.
     */
    if (state) {
       i915->meta.Ctx[I915_CTXREG_LIS5] &= ~mask;
-      i915->meta.Ctx[I915_CTXREG_LIS5] |= 
-	 (i915->state.Ctx[I915_CTXREG_LIS5] & mask);
+      i915->meta.Ctx[I915_CTXREG_LIS5] |=
+         (i915->state.Ctx[I915_CTXREG_LIS5] & mask);
    }
-   else 
+   else
       i915->meta.Ctx[I915_CTXREG_LIS5] |= mask;
-      
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+
+
+static void
+meta_import_pixel_state(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   memcpy(i915->meta.Fog, i915->state.Fog, I915_FOG_SETUP_SIZE * 4);
+
+   i915->meta.Ctx[I915_CTXREG_LIS5] = i915->state.Ctx[I915_CTXREG_LIS5];
+   i915->meta.Ctx[I915_CTXREG_LIS6] = i915->state.Ctx[I915_CTXREG_LIS6];
+   i915->meta.Ctx[I915_CTXREG_STATE4] = i915->state.Ctx[I915_CTXREG_STATE4];
+   i915->meta.Ctx[I915_CTXREG_BLENDCOLOR1] =
+      i915->state.Ctx[I915_CTXREG_BLENDCOLOR1];
+   i915->meta.Ctx[I915_CTXREG_IAB] = i915->state.Ctx[I915_CTXREG_IAB];
+
+   i915->meta.Buffer[I915_DESTREG_SENABLE] =
+      i915->state.Buffer[I915_DESTREG_SENABLE];
+   i915->meta.Buffer[I915_DESTREG_SR1] = i915->state.Buffer[I915_DESTREG_SR1];
+   i915->meta.Buffer[I915_DESTREG_SR2] = i915->state.Buffer[I915_DESTREG_SR2];
+
+   i915->meta.emitted &= ~I915_UPLOAD_FOG;
+   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
@@ -212,69 +252,64 @@ static void set_color_mask( i915ContextPtr i915, GLboolean state )
 
 
 
-static void set_no_texture( i915ContextPtr i915 )
+static void
+meta_no_texture(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    static const GLuint prog[] = {
       _3DSTATE_PIXEL_SHADER_PROGRAM,
 
       /* Declare incoming diffuse color:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_T_DIFFUSE ) |
-       D0_CHANNEL_ALL),
+      (D0_DCL | D0_DECL_REG(REG_T_DIFFUSE) | D0_CHANNEL_ALL),
       D1_MBZ,
       D2_MBZ,
 
       /* output-color = mov(t_diffuse)
        */
       (A0_MOV |
-       A0_DEST_REG( REG_OC ) |
-       A0_DEST_CHANNEL_ALL |
-       A0_SRC0_REG( REG_T_DIFFUSE )),
+       A0_DEST_REG(REG_OC) |
+       A0_DEST_CHANNEL_ALL | A0_SRC0_REG(REG_T_DIFFUSE)),
       (A1_SRC0_XYZW),
       0,
    };
 
-   
-   memcpy( i915->meta.Program, prog, sizeof(prog) );
+
+   memcpy(i915->meta.Program, prog, sizeof(prog));
    i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
    i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
    i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
 }
 
-
-static void enable_texture_blend_replace( i915ContextPtr i915 )
+static void
+meta_texture_blend_replace(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    static const GLuint prog[] = {
       _3DSTATE_PIXEL_SHADER_PROGRAM,
 
       /* Declare the sampler:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_S(0) ) |
-       D0_SAMPLE_TYPE_2D |
-       D0_CHANNEL_NONE),
+      (D0_DCL | D0_DECL_REG(REG_S(0)) | D0_SAMPLE_TYPE_2D | D0_CHANNEL_NONE),
       D1_MBZ,
       D2_MBZ,
 
       /* Declare the interpolated texture coordinate:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_T_TEX(0) ) |
-       D0_CHANNEL_ALL),
+      (D0_DCL | D0_DECL_REG(REG_T_TEX(0)) | D0_CHANNEL_ALL),
       D1_MBZ,
       D2_MBZ,
 
       /* output-color = texld(sample0, texcoord0) 
        */
-      (T0_TEXLD | 
-       T0_DEST_REG( REG_OC ) |
-       T0_SAMPLER( 0 )),
+      (T0_TEXLD | T0_DEST_REG(REG_OC) | T0_SAMPLER(0)),
       T1_ADDRESS_REG(REG_TYPE_T, 0),
       T2_MBZ
    };
 
-   memcpy( i915->meta.Program, prog, sizeof(prog) );
+   memcpy(i915->meta.Program, prog, sizeof(prog));
    i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
    i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
    i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
@@ -287,425 +322,186 @@ static void enable_texture_blend_replace( i915ContextPtr i915 )
 /* Set up an arbitary piece of memory as a rectangular texture
  * (including the front or back buffer).
  */
-static void set_tex_rect_source( i915ContextPtr i915,
-				 GLuint offset,
-				 GLuint width, 
-				 GLuint height,
-				 GLuint pitch, /* in bytes! */
-				 GLuint textureFormat )
+static GLboolean
+meta_tex_rect_source(struct intel_context *intel,
+                     dri_bo *buffer,
+                     GLuint offset,
+                     GLuint pitch, GLuint height, GLenum format, GLenum type)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    GLuint unit = 0;
    GLint numLevels = 1;
    GLuint *state = i915->meta.Tex[0];
+   GLuint textureFormat;
+   GLuint cpp;
 
-#if 0
-   printf("TexRect source offset 0x%x  pitch %d\n", offset, pitch);
-#endif
+   /* A full implementation of this would do the upload through
+    * glTexImage2d, and get all the conversion operations at that
+    * point.  We are restricted, but still at least have access to the
+    * fragment program swizzle.
+    */
+   switch (format) {
+   case GL_BGRA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGBA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_BGR:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5_REV:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGB:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
 
-/*    fprintf(stderr, "%s: offset: %x w: %d h: %d pitch %d format %x\n", */
-/* 	   __FUNCTION__, offset, width, height, pitch, textureFormat ); */
+   default:
+      return GL_FALSE;
+   }
+
+
+   if ((pitch * cpp) & 3) {
+      _mesa_printf("%s: texture is not dword pitch\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+/*    intel_region_release(&i915->meta.tex_region[0]); */
+/*    intel_region_reference(&i915->meta.tex_region[0], region); */
+   i915->meta.tex_buffer[0] = buffer;
+   i915->meta.tex_offset[0] = offset;
 
-   state[I915_TEXREG_MS2] = offset;
    state[I915_TEXREG_MS3] = (((height - 1) << MS3_HEIGHT_SHIFT) |
-			    ((width - 1) << MS3_WIDTH_SHIFT) |
-			    textureFormat |
-			    MS3_USE_FENCE_REGS);
+                             ((pitch - 1) << MS3_WIDTH_SHIFT) |
+                             textureFormat | MS3_USE_FENCE_REGS);
 
-   state[I915_TEXREG_MS4] = ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | 
-			    ((((numLevels-1) * 4)) << MS4_MAX_LOD_SHIFT));
+   state[I915_TEXREG_MS4] = (((((pitch * cpp) / 4) - 1) << MS4_PITCH_SHIFT) |
+                             MS4_CUBE_FACE_ENA_MASK |
+                             ((((numLevels - 1) * 4)) << MS4_MAX_LOD_SHIFT));
 
    state[I915_TEXREG_SS2] = ((FILTER_NEAREST << SS2_MIN_FILTER_SHIFT) |
-			    (MIPFILTER_NONE << SS2_MIP_FILTER_SHIFT) |
-			    (FILTER_NEAREST << SS2_MAG_FILTER_SHIFT));
+                             (MIPFILTER_NONE << SS2_MIP_FILTER_SHIFT) |
+                             (FILTER_NEAREST << SS2_MAG_FILTER_SHIFT));
+
    state[I915_TEXREG_SS3] = ((TEXCOORDMODE_WRAP << SS3_TCX_ADDR_MODE_SHIFT) |
-			    (TEXCOORDMODE_WRAP << SS3_TCY_ADDR_MODE_SHIFT) |
-			    (TEXCOORDMODE_WRAP << SS3_TCZ_ADDR_MODE_SHIFT) |
-			    (unit<<SS3_TEXTUREMAP_INDEX_SHIFT));
+                             (TEXCOORDMODE_WRAP << SS3_TCY_ADDR_MODE_SHIFT) |
+                             (TEXCOORDMODE_WRAP << SS3_TCZ_ADDR_MODE_SHIFT) |
+                             (unit << SS3_TEXTUREMAP_INDEX_SHIFT));
 
    state[I915_TEXREG_SS4] = 0;
 
    i915->meta.emitted &= ~I915_UPLOAD_TEX(0);
+   return GL_TRUE;
 }
 
 
-/* Select between front and back draw buffers.
+/**
+ * Set the color and depth drawing region for meta ops.
  */
-static void set_draw_region( i915ContextPtr i915, const intelRegion *region )
+static void
+meta_draw_region(struct intel_context *intel,
+                 struct intel_region *color_region,
+                 struct intel_region *depth_region)
 {
-#if 0
-   printf("Rotate into region: offset 0x%x  pitch %d\n",
-          region->offset, region->pitch);
-#endif
-   i915->meta.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->meta.Buffer[I915_DESTREG_CBUFADDR2] = region->offset;
-   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   i915_state_draw_region(intel, &i915->meta, color_region, depth_region);
 }
 
 
-#if 0
-/* Setup an arbitary draw format, useful for targeting texture or agp
- * memory.
- */
-static void set_draw_format( i915ContextPtr i915,
-			     GLuint format,
-			     GLuint depth_format)
+static void
+set_vertex_format(struct intel_context *intel)
 {
-   i915->meta.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					DSTORG_VERT_BIAS(0x8) | /* .5 */
-					format |
-					LOD_PRECLAMP_OGL |
-					TEX_DEFAULT_COLOR_OGL |
-					depth_format);
-
-   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
-/*    fprintf(stderr, "%s: DV1: %x\n",  */
-/* 	   __FUNCTION__, i915->meta.Buffer[I915_DESTREG_DV1]); */
-}
-#endif
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
-static void set_vertex_format( i915ContextPtr i915 )
-{
-   i915->meta.Ctx[I915_CTXREG_LIS2] = 
+   i915->meta.Ctx[I915_CTXREG_LIS2] =
       (S2_TEXCOORD_FMT(0, TEXCOORDFMT_2D) |
-       S2_TEXCOORD_FMT(1, TEXCOORDFMT_NOT_PRESENT) | 
+       S2_TEXCOORD_FMT(1, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(2, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(3, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(4, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(5, TEXCOORDFMT_NOT_PRESENT) | 
+       S2_TEXCOORD_FMT(5, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(6, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(7, TEXCOORDFMT_NOT_PRESENT));
 
    i915->meta.Ctx[I915_CTXREG_LIS4] &= ~S4_VFMT_MASK;
 
-   i915->meta.Ctx[I915_CTXREG_LIS4] |= 
-      (S4_VFMT_COLOR |
-       S4_VFMT_SPEC_FOG |
-       S4_VFMT_XYZW);
+   i915->meta.Ctx[I915_CTXREG_LIS4] |= (S4_VFMT_COLOR | S4_VFMT_XYZ);
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
-
 }
 
 
-static void draw_quad(i915ContextPtr i915, 
-		      GLfloat x0, GLfloat x1,
-		      GLfloat y0, GLfloat y1, 
-		      GLubyte red, GLubyte green,
-		      GLubyte blue, GLubyte alpha,
-		      GLfloat s0, GLfloat s1,
-		      GLfloat t0, GLfloat t1 )
-{
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i915->intel, 
-						PRIM3D_TRIFAN, 
-						4 * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i;
-
-   if (0)
-      fprintf(stderr, "%s: %f,%f-%f,%f 0x%x%x%x%x %f,%f-%f,%f\n",
-	      __FUNCTION__,
-	      x0,y0,x1,y1,red,green,blue,alpha,s0,t0,s1,t1);
-
-
-   /* initial vertex, left bottom */
-   tmp.v.x = x0;
-   tmp.v.y = y0;
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-   tmp.v.u0 = s0;
-   tmp.v.v0 = t0;
-
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right bottom */
-   vb += vertex_size;
-   tmp.v.x = x1;
-   tmp.v.u0 = s1;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right top */
-   vb += vertex_size;
-   tmp.v.y = y1;
-   tmp.v.v0 = t1;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* left top */
-   vb += vertex_size;
-   tmp.v.x = x0;
-   tmp.v.u0 = s0;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-}
-
 
-static void draw_poly(i915ContextPtr i915, 
-		      GLubyte red, GLubyte green, GLubyte blue, GLubyte alpha,
-                      GLuint numVerts,
-                      /*const*/ GLfloat verts[][2],
-                      /*const*/ GLfloat texcoords[][2])
+/* Operations where the 3D engine is decoupled temporarily from the
+ * current GL state and used for other purposes than simply rendering
+ * incoming triangles.
+ */
+static void
+install_meta_state(struct intel_context *intel)
 {
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i915->intel, 
-						PRIM3D_TRIFAN, 
-						numVerts * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i, k;
-
-   /* initial constant vertex fields */
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-
-   for (k = 0; k < numVerts; k++) {
-      tmp.v.x = verts[k][0];
-      tmp.v.y = verts[k][1];
-      tmp.v.u0 = texcoords[k][0];
-      tmp.v.v0 = texcoords[k][1];
-
-      for (i = 0 ; i < vertex_size ; i++)
-         vb[i] = tmp.ui[i];
-
-      vb += vertex_size;
-   }
-}
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   memcpy(&i915->meta, &i915->initial, sizeof(i915->meta));
+   i915->meta.active = ACTIVE;
+   i915->meta.emitted = 0;
 
+   SET_STATE(i915, meta);
+   set_vertex_format(intel);
+   meta_no_texture(intel);
+}
 
-void 
-i915ClearWithTris(intelContextPtr intel, GLbitfield buffers,
-		  GLboolean allFoo,
-		  GLint cxFoo, GLint cyFoo, GLint cwFoo, GLint chFoo)
+static void
+leave_meta_state(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT( intel );
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   intelScreenPrivate *screen = intel->intelScreen;
-   int x0, y0, x1, y1;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-
-   SET_STATE( i915, meta ); 
-   set_initial_state( i915 ); 
-   set_no_texture( i915 ); 
-   set_vertex_format( i915 ); 
-
-   LOCK_HARDWARE(intel);
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   if (!all) {
-      x0 = cx;
-      y0 = cy;
-      x1 = x0 + cw;
-      y1 = y0 + ch;
-   } else {
-      x0 = 0;
-      y0 = 0;
-      x1 = x0 + dPriv->w;
-      y1 = y0 + dPriv->h;
-   }
-
-   /* Don't do any clipping to screen - these are window coordinates.
-    * The active cliprects will be applied as for any other geometry.
-    */
-
-   if (buffers & BUFFER_BIT_FRONT_LEFT) { 
-      set_no_depth_stencil_write( i915 );
-      set_color_mask( i915, GL_TRUE );
-      set_draw_region( i915, &screen->front );
-
-      draw_quad(i915, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green, 
- 		intel->clear_blue, intel->clear_alpha, 
-		0, 0, 0, 0);
-   }
-
-   if (buffers & BUFFER_BIT_BACK_LEFT) {
-      set_no_depth_stencil_write( i915 );
-      set_color_mask( i915, GL_TRUE );
-      set_draw_region( i915, &screen->back );
-
-      draw_quad(i915, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if (buffers & BUFFER_BIT_STENCIL) {
-      set_stencil_replace( i915, 
-			   intel->ctx.Stencil.WriteMask[0], 
-			   intel->ctx.Stencil.Clear);
-      
-      set_color_mask( i915, GL_FALSE );
-      set_draw_region( i915, &screen->front ); /* could be either? */
-
-      draw_quad( i915, x0, x1, y0, y1, 0, 0, 0, 0, 0, 0, 0, 0 );
-   }
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i915, state );
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   intel_region_release(&i915->meta.draw_region);
+   intel_region_release(&i915->meta.depth_region);
+/*    intel_region_release(&i915->meta.tex_region[0]); */
+   SET_STATE(i915, state);
 }
 
 
-/**
- * Copy the window contents named by dPriv to the rotated (or reflected)
- * color buffer.
- * srcBuf is BUFFER_BIT_FRONT_LEFT or BUFFER_BIT_BACK_LEFT to indicate the source.
- */
+
 void
-i915RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf)
+i915InitMetaFuncs(struct i915_context *i915)
 {
-   i915ContextPtr i915 = I915_CONTEXT( intel );
-   intelScreenPrivate *screen = intel->intelScreen;
-   const GLuint cpp = screen->cpp;
-   drm_clip_rect_t fullRect;
-   GLuint textureFormat, srcOffset, srcPitch;
-   const drm_clip_rect_t *clipRects;
-   int numClipRects;
-   int i;
-
-   int xOrig, yOrig;
-   int origNumClipRects;
-   drm_clip_rect_t *origRects;
-
-   /*
-    * set up hardware state
-    */
-   intelFlush( &intel->ctx );
-
-   SET_STATE( i915, meta ); 
-   set_initial_state( i915 ); 
-   set_no_texture( i915 ); 
-   set_vertex_format( i915 ); 
-   set_no_depth_stencil_write( i915 );
-   set_color_mask( i915, GL_TRUE );
-
-   LOCK_HARDWARE(intel);
-
-   /* save current drawing origin and cliprects (restored at end) */
-   xOrig = intel->drawX;
-   yOrig = intel->drawY;
-   origNumClipRects = intel->numClipRects;
-   origRects = intel->pClipRects;
-
-   if (!intel->numClipRects)
-      goto done;
-
-   /*
-    * set drawing origin, cliprects for full-screen access to rotated screen
-    */
-   fullRect.x1 = 0;
-   fullRect.y1 = 0;
-   fullRect.x2 = screen->rotatedWidth;
-   fullRect.y2 = screen->rotatedHeight;
-   intel->drawX = 0;
-   intel->drawY = 0;
-   intel->numClipRects = 1;
-   intel->pClipRects = &fullRect;
-
-   set_draw_region( i915, &screen->rotated );
-
-   if (cpp == 4)
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   else
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-
-   if (srcBuf == BUFFER_BIT_FRONT_LEFT) {
-      srcPitch = screen->front.pitch;   /* in bytes */
-      srcOffset = screen->front.offset; /* bytes */
-      clipRects = dPriv->pClipRects;
-      numClipRects = dPriv->numClipRects;
-   }
-   else {
-      srcPitch = screen->back.pitch;   /* in bytes */
-      srcOffset = screen->back.offset; /* bytes */
-      clipRects = dPriv->pBackClipRects;
-      numClipRects = dPriv->numBackClipRects;
-   }
-
-   /* set the whole screen up as a texture to avoid alignment issues */
-   set_tex_rect_source(i915,
-                       srcOffset,
-                       screen->width,
-		       screen->height,
-                       srcPitch,
-                       textureFormat);
-
-   enable_texture_blend_replace(i915);
-
-   /*
-    * loop over the source window's cliprects
-    */
-   for (i = 0; i < numClipRects; i++) {
-      int srcX0 = clipRects[i].x1;
-      int srcY0 = clipRects[i].y1;
-      int srcX1 = clipRects[i].x2;
-      int srcY1 = clipRects[i].y2;
-      GLfloat verts[4][2], tex[4][2];
-      int j;
-
-      /* build vertices for four corners of clip rect */
-      verts[0][0] = srcX0;  verts[0][1] = srcY0;
-      verts[1][0] = srcX1;  verts[1][1] = srcY0;
-      verts[2][0] = srcX1;  verts[2][1] = srcY1;
-      verts[3][0] = srcX0;  verts[3][1] = srcY1;
-
-      /* .. and texcoords */
-      tex[0][0] = srcX0;  tex[0][1] = srcY0;
-      tex[1][0] = srcX1;  tex[1][1] = srcY0;
-      tex[2][0] = srcX1;  tex[2][1] = srcY1;
-      tex[3][0] = srcX0;  tex[3][1] = srcY1;
-
-      /* transform coords to rotated screen coords */
-      for (j = 0; j < 4; j++) {
-         matrix23TransformCoordf(&screen->rotMatrix,
-                                 &verts[j][0], &verts[j][1]);
-      }
-
-      /* draw polygon to map source image to dest region */
-      draw_poly(i915, 255, 255, 255, 255, 4, verts, tex);
-
-   } /* cliprect loop */
-
-   intelFlushBatchLocked( intel, GL_FALSE, GL_FALSE, GL_FALSE );
-
- done:
-   /* restore original drawing origin and cliprects */
-   intel->drawX = xOrig;
-   intel->drawY = yOrig;
-   intel->numClipRects = origNumClipRects;
-   intel->pClipRects = origRects;
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i915, state );
+   i915->intel.vtbl.install_meta_state = install_meta_state;
+   i915->intel.vtbl.leave_meta_state = leave_meta_state;
+   i915->intel.vtbl.meta_no_depth_write = meta_no_depth_write;
+   i915->intel.vtbl.meta_no_stencil_write = meta_no_stencil_write;
+   i915->intel.vtbl.meta_stencil_replace = meta_stencil_replace;
+   i915->intel.vtbl.meta_depth_replace = meta_depth_replace;
+   i915->intel.vtbl.meta_color_mask = meta_color_mask;
+   i915->intel.vtbl.meta_no_texture = meta_no_texture;
+   i915->intel.vtbl.meta_texture_blend_replace = meta_texture_blend_replace;
+   i915->intel.vtbl.meta_tex_rect_source = meta_tex_rect_source;
+   i915->intel.vtbl.meta_draw_region = meta_draw_region;
+   i915->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i915_program.c b/src/mesa/drivers/dri/i915/i915_program.c
index 68491124449..e87700f8e0a 100644
--- a/src/mesa/drivers/dri/i915/i915_program.c
+++ b/src/mesa/drivers/dri/i915/i915_program.c
@@ -27,9 +27,9 @@
 
 #include <strings.h>
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "tnl/t_context.h"
 #include "intel_batchbuffer.h"
@@ -72,58 +72,62 @@
 
 #define I915_CONSTFLAG_PARAM 0x1f
 
-GLuint i915_get_temp( struct i915_fragment_program *p )
+GLuint
+i915_get_temp(struct i915_fragment_program *p)
 {
-   int bit = ffs( ~p->temp_flag );
+   int bit = ffs(~p->temp_flag);
    if (!bit) {
       fprintf(stderr, "%s: out of temporaries\n", __FILE__);
       exit(1);
    }
 
-   p->temp_flag |= 1<<(bit-1);
-   return UREG(REG_TYPE_R, (bit-1));
+   p->temp_flag |= 1 << (bit - 1);
+   return UREG(REG_TYPE_R, (bit - 1));
 }
 
 
-GLuint i915_get_utemp( struct i915_fragment_program *p )
+GLuint
+i915_get_utemp(struct i915_fragment_program * p)
 {
-   int bit = ffs( ~p->utemp_flag );
+   int bit = ffs(~p->utemp_flag);
    if (!bit) {
       fprintf(stderr, "%s: out of temporaries\n", __FILE__);
       exit(1);
    }
 
-   p->utemp_flag |= 1<<(bit-1);
-   return UREG(REG_TYPE_U, (bit-1));
+   p->utemp_flag |= 1 << (bit - 1);
+   return UREG(REG_TYPE_U, (bit - 1));
 }
 
-void i915_release_utemps( struct i915_fragment_program *p )
+void
+i915_release_utemps(struct i915_fragment_program *p)
 {
    p->utemp_flag = ~0x7;
 }
 
 
-GLuint i915_emit_decl( struct i915_fragment_program *p,
-		      GLuint type, GLuint nr, GLuint d0_flags )
+GLuint
+i915_emit_decl(struct i915_fragment_program *p,
+               GLuint type, GLuint nr, GLuint d0_flags)
 {
    GLuint reg = UREG(type, nr);
 
    if (type == REG_TYPE_T) {
-      if (p->decl_t & (1<<nr))
-	 return reg;
+      if (p->decl_t & (1 << nr))
+         return reg;
 
-      p->decl_t |= (1<<nr);
+      p->decl_t |= (1 << nr);
    }
    else if (type == REG_TYPE_S) {
-      if (p->decl_s & (1<<nr))
-	 return reg;
+      if (p->decl_s & (1 << nr))
+         return reg;
 
-      p->decl_s |= (1<<nr);
+      p->decl_s |= (1 << nr);
    }
-   else 
+   else
       return reg;
 
-   *(p->decl++) = (D0_DCL | D0_DEST( reg ) | d0_flags);
+   *(p->decl++) = (D0_DCL | D0_DEST(reg) | d0_flags);
    *(p->decl++) = D1_MBZ;
    *(p->decl++) = D2_MBZ;
 
@@ -131,24 +135,26 @@ GLuint i915_emit_decl( struct i915_fragment_program *p,
    return reg;
 }
 
-GLuint i915_emit_arith( struct i915_fragment_program *p,
-		       GLuint op,
-		       GLuint dest,
-		       GLuint mask,
-		       GLuint saturate,
-		       GLuint src0,
-		       GLuint src1,
-		       GLuint src2 )
+GLuint
+i915_emit_arith(struct i915_fragment_program * p,
+                GLuint op,
+                GLuint dest,
+                GLuint mask,
+                GLuint saturate, GLuint src0, GLuint src1, GLuint src2)
 {
    GLuint c[3];
    GLuint nr_const = 0;
 
    assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
-   assert(dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
+   dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest));
+   assert(dest);
 
-   if (GET_UREG_TYPE(src0) == REG_TYPE_CONST) c[nr_const++] = 0;
-   if (GET_UREG_TYPE(src1) == REG_TYPE_CONST) c[nr_const++] = 1;
-   if (GET_UREG_TYPE(src2) == REG_TYPE_CONST) c[nr_const++] = 2;
+   if (GET_UREG_TYPE(src0) == REG_TYPE_CONST)
+      c[nr_const++] = 0;
+   if (GET_UREG_TYPE(src1) == REG_TYPE_CONST)
+      c[nr_const++] = 1;
+   if (GET_UREG_TYPE(src2) == REG_TYPE_CONST)
+      c[nr_const++] = 2;
 
    /* Recursively call this function to MOV additional const values
     * into temporary registers.  Use utemp registers for this -
@@ -164,67 +170,105 @@ GLuint i915_emit_arith( struct i915_fragment_program *p,
       old_utemp_flag = p->utemp_flag;
 
       first = GET_UREG_NR(s[c[0]]);
-      for (i = 1 ; i < nr_const ; i++) {
-	 if (GET_UREG_NR(s[c[i]]) != first) {
-	    GLuint tmp = i915_get_utemp(p);
-
-	    i915_emit_arith( p, A0_MOV, tmp, A0_DEST_CHANNEL_ALL, 0,
-			    s[c[i]], 0, 0 );
-	    s[c[i]] = tmp;
-	 }
+      for (i = 1; i < nr_const; i++) {
+         if (GET_UREG_NR(s[c[i]]) != first) {
+            GLuint tmp = i915_get_utemp(p);
+
+            i915_emit_arith(p, A0_MOV, tmp, A0_DEST_CHANNEL_ALL, 0,
+                            s[c[i]], 0, 0);
+            s[c[i]] = tmp;
+         }
       }
 
       src0 = s[0];
       src1 = s[1];
       src2 = s[2];
-      p->utemp_flag = old_utemp_flag; /* restore */
+      p->utemp_flag = old_utemp_flag;   /* restore */
    }
 
-   *(p->csr++) = (op | 
-		  A0_DEST( dest ) |
-		  mask | 
-		  saturate |
-		  A0_SRC0( src0 ));
-   *(p->csr++) = (A1_SRC0( src0 ) |
-		  A1_SRC1( src1 ));
-   *(p->csr++) = (A2_SRC1( src1 ) |
-		  A2_SRC2( src2 ));
+   *(p->csr++) = (op | A0_DEST(dest) | mask | saturate | A0_SRC0(src0));
+   *(p->csr++) = (A1_SRC0(src0) | A1_SRC1(src1));
+   *(p->csr++) = (A2_SRC1(src1) | A2_SRC2(src2));
+
+   if (GET_UREG_TYPE(dest) == REG_TYPE_R)
+      p->register_phases[GET_UREG_NR(dest)] = p->nr_tex_indirect;
 
    p->nr_alu_insn++;
    return dest;
 }
 
+static GLuint get_free_rreg (struct i915_fragment_program *p, 
+                             GLuint live_regs)
+{
+    int bit = ffs(~live_regs);
+    if (!bit) {
+        i915_program_error(p, "Can't find free R reg");
+        return UREG_BAD;
+    }
+    return UREG(REG_TYPE_R, bit - 1);
+}
+
 GLuint i915_emit_texld( struct i915_fragment_program *p,
+			GLuint live_regs,               
 			GLuint dest,
 			GLuint destmask,
 			GLuint sampler,
 			GLuint coord,
 			GLuint op )
 {
-   if (coord != UREG(GET_UREG_TYPE(coord), GET_UREG_NR(coord))) {
-      /* No real way to work around this in the general case - need to
-       * allocate and declare a new temporary register (a utemp won't
-       * do).  Will fallback for now.
-       */
-      i915_program_error(p, "Can't (yet) swizzle TEX arguments");
-      return 0;
-   }
-
-   /* Don't worry about saturate as we only support  
+    if (coord != UREG(GET_UREG_TYPE(coord), GET_UREG_NR(coord))) {
+        /* With the help of the "needed registers" table created earlier, pick
+         * a register we can MOV the swizzled TC to (since TEX doesn't support
+         * swizzled sources) */
+        GLuint swizCoord = get_free_rreg(p, live_regs);
+        if (swizCoord == UREG_BAD) 
+            return 0;
+
+        i915_emit_arith( p, A0_MOV, swizCoord, A0_DEST_CHANNEL_ALL, 0, coord, 0, 0 );
+        coord = swizCoord;
+    }
+
+   /* Don't worry about saturate as we only support texture formats
+    * that are always in the 0..1 range.
     */
    if (destmask != A0_DEST_CHANNEL_ALL) {
       GLuint tmp = i915_get_utemp(p);
-      i915_emit_texld( p, tmp, A0_DEST_CHANNEL_ALL, sampler, coord, op );
+      i915_emit_texld( p, 0, tmp, A0_DEST_CHANNEL_ALL, sampler, coord, op );
       i915_emit_arith( p, A0_MOV, dest, destmask, 0, tmp, 0, 0 );
       return dest;
    }
    else {
       assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
       assert(dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
+      /* Can't use unsaved temps for coords, as the phase boundary would result
+       * in the contents becoming undefined.
+       */
+      assert(GET_UREG_TYPE(coord) != REG_TYPE_U);
+
+      if ((GET_UREG_TYPE(coord) != REG_TYPE_R) &&
+          (GET_UREG_TYPE(coord) != REG_TYPE_OC) &&
+          (GET_UREG_TYPE(coord) != REG_TYPE_OD) &&
+          (GET_UREG_TYPE(coord) != REG_TYPE_T)) {
+          GLuint  tmpCoord = get_free_rreg(p, live_regs);
+          
+          if (tmpCoord == UREG_BAD) 
+              return 0;
+
+          i915_emit_arith(p, A0_MOV, tmpCoord, A0_DEST_CHANNEL_ALL, 0, coord, 0, 0);
+          coord = tmpCoord;
+      }
 
-      if (GET_UREG_TYPE(coord) != REG_TYPE_T) {
+      /* Output register being oC or oD defines a phase boundary */
+      if (GET_UREG_TYPE(dest) == REG_TYPE_OC ||
+	  GET_UREG_TYPE(dest) == REG_TYPE_OD)
+	 p->nr_tex_indirect++;
+
+      /* Reading from an r# register whose contents depend on output of the
+       * current phase defines a phase boundary.
+       */
+      if (GET_UREG_TYPE(coord) == REG_TYPE_R &&
+	  p->register_phases[GET_UREG_NR(coord)] == p->nr_tex_indirect)
 	 p->nr_tex_indirect++;
-      }
 
       *(p->csr++) = (op | 
 		     T0_DEST( dest ) |
@@ -233,30 +277,37 @@ GLuint i915_emit_texld( struct i915_fragment_program *p,
       *(p->csr++) = T1_ADDRESS_REG( coord );
       *(p->csr++) = T2_MBZ;
 
+      if (GET_UREG_TYPE(dest) == REG_TYPE_R)
+	 p->register_phases[GET_UREG_NR(dest)] = p->nr_tex_indirect;
+
       p->nr_tex_insn++;
       return dest;
    }
 }
 
 
-GLuint i915_emit_const1f( struct i915_fragment_program *p, GLfloat c0 )
+GLuint
+i915_emit_const1f(struct i915_fragment_program * p, GLfloat c0)
 {
    GLint reg, idx;
 
-   if (c0 == 0.0) return swizzle(UREG(REG_TYPE_R, 0), ZERO, ZERO, ZERO, ZERO);
-   if (c0 == 1.0) return swizzle(UREG(REG_TYPE_R, 0), ONE,  ONE,  ONE,  ONE );
+   if (c0 == 0.0)
+      return swizzle(UREG(REG_TYPE_R, 0), ZERO, ZERO, ZERO, ZERO);
+   if (c0 == 1.0)
+      return swizzle(UREG(REG_TYPE_R, 0), ONE, ONE, ONE, ONE);
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-	 continue;
+         continue;
       for (idx = 0; idx < 4; idx++) {
-	 if (!(p->constant_flags[reg] & (1<<idx)) ||
-	     p->constant[reg][idx] == c0) {
-	    p->constant[reg][idx] = c0;
-	    p->constant_flags[reg] |= 1<<idx;
-	    if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	    return swizzle(UREG(REG_TYPE_CONST, reg),idx,ZERO,ZERO,ONE);
-	 }
+         if (!(p->constant_flags[reg] & (1 << idx)) ||
+             p->constant[reg][idx] == c0) {
+            p->constant[reg][idx] = c0;
+            p->constant_flags[reg] |= 1 << idx;
+            if (reg + 1 > p->nr_constants)
+               p->nr_constants = reg + 1;
+            return swizzle(UREG(REG_TYPE_CONST, reg), idx, ZERO, ZERO, ONE);
+         }
       }
    }
 
@@ -265,29 +316,35 @@ GLuint i915_emit_const1f( struct i915_fragment_program *p, GLfloat c0 )
    return 0;
 }
 
-GLuint i915_emit_const2f( struct i915_fragment_program *p, 
-			 GLfloat c0, GLfloat c1 )
+GLuint
+i915_emit_const2f(struct i915_fragment_program * p, GLfloat c0, GLfloat c1)
 {
    GLint reg, idx;
 
-   if (c0 == 0.0) return swizzle(i915_emit_const1f(p, c1), ZERO, X, Z, W);
-   if (c0 == 1.0) return swizzle(i915_emit_const1f(p, c1), ONE,  X, Z, W); 
+   if (c0 == 0.0)
+      return swizzle(i915_emit_const1f(p, c1), ZERO, X, Z, W);
+   if (c0 == 1.0)
+      return swizzle(i915_emit_const1f(p, c1), ONE, X, Z, W);
 
-   if (c1 == 0.0) return swizzle(i915_emit_const1f(p, c0), X, ZERO, Z, W);
-   if (c1 == 1.0) return swizzle(i915_emit_const1f(p, c0), X, ONE,  Z, W);
+   if (c1 == 0.0)
+      return swizzle(i915_emit_const1f(p, c0), X, ZERO, Z, W);
+   if (c1 == 1.0)
+      return swizzle(i915_emit_const1f(p, c0), X, ONE, Z, W);
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0xf ||
-	  p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-	 continue;
+          p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
+         continue;
       for (idx = 0; idx < 3; idx++) {
-	 if (!(p->constant_flags[reg] & (3<<idx))) {
-	    p->constant[reg][idx] = c0;
-	    p->constant[reg][idx+1] = c1;
-	    p->constant_flags[reg] |= 3<<idx;
-	    if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	    return swizzle(UREG(REG_TYPE_CONST, reg),idx,idx+1,ZERO,ONE);
-	 }
+         if (!(p->constant_flags[reg] & (3 << idx))) {
+            p->constant[reg][idx] = c0;
+            p->constant[reg][idx + 1] = c1;
+            p->constant_flags[reg] |= 3 << idx;
+            if (reg + 1 > p->nr_constants)
+               p->nr_constants = reg + 1;
+            return swizzle(UREG(REG_TYPE_CONST, reg), idx, idx + 1, ZERO,
+                           ONE);
+         }
       }
    }
 
@@ -298,27 +355,28 @@ GLuint i915_emit_const2f( struct i915_fragment_program *p,
 
 
 
-GLuint i915_emit_const4f( struct i915_fragment_program *p, 
-			 GLfloat c0, GLfloat c1, GLfloat c2, GLfloat c3 )
+GLuint
+i915_emit_const4f(struct i915_fragment_program * p,
+                  GLfloat c0, GLfloat c1, GLfloat c2, GLfloat c3)
 {
    GLint reg;
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0xf &&
-	  p->constant[reg][0] == c0 &&
-	  p->constant[reg][1] == c1 &&
-	  p->constant[reg][2] == c2 &&
-	  p->constant[reg][3] == c3) {
-	 return UREG(REG_TYPE_CONST, reg);
+          p->constant[reg][0] == c0 &&
+          p->constant[reg][1] == c1 &&
+          p->constant[reg][2] == c2 && p->constant[reg][3] == c3) {
+         return UREG(REG_TYPE_CONST, reg);
       }
       else if (p->constant_flags[reg] == 0) {
-	 p->constant[reg][0] = c0;
-	 p->constant[reg][1] = c1;
-	 p->constant[reg][2] = c2;
-	 p->constant[reg][3] = c3;
-	 p->constant_flags[reg] = 0xf;
-	 if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	 return UREG(REG_TYPE_CONST, reg);
+         p->constant[reg][0] = c0;
+         p->constant[reg][1] = c1;
+         p->constant[reg][2] = c2;
+         p->constant[reg][3] = c3;
+         p->constant_flags[reg] = 0xf;
+         if (reg + 1 > p->nr_constants)
+            p->nr_constants = reg + 1;
+         return UREG(REG_TYPE_CONST, reg);
       }
    }
 
@@ -328,34 +386,36 @@ GLuint i915_emit_const4f( struct i915_fragment_program *p,
 }
 
 
-GLuint i915_emit_const4fv( struct i915_fragment_program *p, const GLfloat *c )
+GLuint
+i915_emit_const4fv(struct i915_fragment_program * p, const GLfloat * c)
 {
-   return i915_emit_const4f( p, c[0], c[1], c[2], c[3] );
+   return i915_emit_const4f(p, c[0], c[1], c[2], c[3]);
 }
 
 
-GLuint i915_emit_param4fv( struct i915_fragment_program *p, 
-			  const GLfloat *values )
+GLuint
+i915_emit_param4fv(struct i915_fragment_program * p, const GLfloat * values)
 {
    GLint reg, i;
 
    for (i = 0; i < p->nr_params; i++) {
       if (p->param[i].values == values)
-	 return UREG(REG_TYPE_CONST, p->param[i].reg);
+         return UREG(REG_TYPE_CONST, p->param[i].reg);
    }
 
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0) {
-	 p->constant_flags[reg] = I915_CONSTFLAG_PARAM;
-	 i = p->nr_params++;
+         p->constant_flags[reg] = I915_CONSTFLAG_PARAM;
+         i = p->nr_params++;
 
-	 p->param[i].values = values;
-	 p->param[i].reg = reg;
-	 p->params_uptodate = 0;
+         p->param[i].values = values;
+         p->param[i].reg = reg;
+         p->params_uptodate = 0;
 
-	 if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	 return UREG(REG_TYPE_CONST, reg);
+         if (reg + 1 > p->nr_constants)
+            p->nr_constants = reg + 1;
+         return UREG(REG_TYPE_CONST, reg);
       }
    }
 
@@ -366,30 +426,32 @@ GLuint i915_emit_param4fv( struct i915_fragment_program *p,
 
 
 
-
-void i915_program_error( struct i915_fragment_program *p, const char *msg )
+void
+i915_program_error(struct i915_fragment_program *p, const char *msg)
 {
    _mesa_problem(NULL, "i915_program_error: %s", msg);
    p->error = 1;
 }
 
-void i915_init_program( i915ContextPtr i915, struct i915_fragment_program *p )
+
+void
+i915_init_program(struct i915_context *i915, struct i915_fragment_program *p)
 {
    GLcontext *ctx = &i915->intel.ctx;
-   TNLcontext *tnl = TNL_CONTEXT( ctx );
-   
+
    p->translated = 0;
    p->params_uptodate = 0;
    p->on_hardware = 0;
    p->error = 0;
 
-   p->nr_tex_indirect = 1;	/* correct? */
+   memset(&p->register_phases, 0, sizeof(p->register_phases));
+   p->nr_tex_indirect = 1;
    p->nr_tex_insn = 0;
    p->nr_alu_insn = 0;
    p->nr_decl_insn = 0;
 
-   p->ctx = ctx;  
-   memset( p->constant_flags, 0, sizeof(p->constant_flags) );
+   p->ctx = ctx;
+   memset(p->constant_flags, 0, sizeof(p->constant_flags));
 
    p->nr_constants = 0;
    p->csr = p->program;
@@ -402,21 +464,17 @@ void i915_init_program( i915ContextPtr i915, struct i915_fragment_program *p )
    p->depth_written = 0;
    p->nr_params = 0;
 
-   p->src_texture = UREG_BAD;
-   p->src_previous = UREG(REG_TYPE_T, T_DIFFUSE);
-   p->last_tex_stage = 0;
-   p->VB = &tnl->vb;
-
    *(p->decl++) = _3DSTATE_PIXEL_SHADER_PROGRAM;
 }
 
 
-void i915_fini_program( struct i915_fragment_program *p )
+void
+i915_fini_program(struct i915_fragment_program *p)
 {
    GLuint program_size = p->csr - p->program;
    GLuint decl_size = p->decl - p->declarations;
-   
-   if (p->nr_tex_indirect > I915_MAX_TEX_INDIRECT) 
+
+   if (p->nr_tex_indirect > I915_MAX_TEX_INDIRECT)
       i915_program_error(p, "Exceeded max nr indirect texture lookups");
 
    if (p->nr_tex_insn > I915_MAX_TEX_INSN)
@@ -446,22 +504,24 @@ void i915_fini_program( struct i915_fragment_program *p )
    p->declarations[0] |= program_size + decl_size - 2;
 }
 
-void i915_upload_program( i915ContextPtr i915, struct i915_fragment_program *p )
+void
+i915_upload_program(struct i915_context *i915,
+                    struct i915_fragment_program *p)
 {
    GLuint program_size = p->csr - p->program;
    GLuint decl_size = p->decl - p->declarations;
 
-   FALLBACK( &i915->intel, I915_FALLBACK_PROGRAM, p->error );
+   FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, p->error);
 
    /* Could just go straight to the batchbuffer from here:
     */
    if (i915->state.ProgramSize != (program_size + decl_size) ||
-       memcmp(i915->state.Program + decl_size, p->program, 
-	      program_size*sizeof(int)) != 0) {
-      I915_STATECHANGE( i915, I915_UPLOAD_PROGRAM );
-      memcpy(i915->state.Program, p->declarations, decl_size*sizeof(int));
+       memcmp(i915->state.Program + decl_size, p->program,
+              program_size * sizeof(int)) != 0) {
+      I915_STATECHANGE(i915, I915_UPLOAD_PROGRAM);
+      memcpy(i915->state.Program, p->declarations, decl_size * sizeof(int));
       memcpy(i915->state.Program + decl_size, p->program,
-	     program_size*sizeof(int));
+             program_size * sizeof(int));
       i915->state.ProgramSize = decl_size + program_size;
    }
 
@@ -470,30 +530,28 @@ void i915_upload_program( i915ContextPtr i915, struct i915_fragment_program *p )
     */
    if (p->nr_constants) {
       GLuint nr = p->nr_constants;
-      
-      I915_ACTIVESTATE( i915, I915_UPLOAD_CONSTANTS, 1 );
-      I915_STATECHANGE( i915, I915_UPLOAD_CONSTANTS );
+
+      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 1);
+      I915_STATECHANGE(i915, I915_UPLOAD_CONSTANTS);
 
       i915->state.Constant[0] = _3DSTATE_PIXEL_SHADER_CONSTANTS | ((nr) * 4);
-      i915->state.Constant[1] = (1<<(nr-1)) | ((1<<(nr-1))-1);
-      
-      memcpy(&i915->state.Constant[2], p->constant, 4*sizeof(int)*(nr));
+      i915->state.Constant[1] = (1 << (nr - 1)) | ((1 << (nr - 1)) - 1);
+
+      memcpy(&i915->state.Constant[2], p->constant, 4 * sizeof(int) * (nr));
       i915->state.ConstantSize = 2 + (nr) * 4;
 
       if (0) {
-	 GLuint i;
-	 for (i = 0; i < nr; i++) {
-	    fprintf(stderr, "const[%d]: %f %f %f %f\n", i, 
-		    p->constant[i][0],
-		    p->constant[i][1],
-		    p->constant[i][2],
-		    p->constant[i][3]);
-	 }
+         GLuint i;
+         for (i = 0; i < nr; i++) {
+            fprintf(stderr, "const[%d]: %f %f %f %f\n", i,
+                    p->constant[i][0],
+                    p->constant[i][1], p->constant[i][2], p->constant[i][3]);
+         }
       }
    }
    else {
-      I915_ACTIVESTATE( i915, I915_UPLOAD_CONSTANTS, 0 );
-   }  
+      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 0);
+   }
 
    p->on_hardware = 1;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_program.h b/src/mesa/drivers/dri/i915/i915_program.h
index 8891a177855..14a3f08801f 100644
--- a/src/mesa/drivers/dri/i915/i915_program.h
+++ b/src/mesa/drivers/dri/i915/i915_program.h
@@ -48,11 +48,11 @@
 #define UREG_CHANNEL_W_NEGATE_SHIFT   11
 #define UREG_CHANNEL_W_SHIFT          8
 #define UREG_CHANNEL_ZERO_NEGATE_MBZ  5
-#define UREG_CHANNEL_ZERO_SHIFT       4      
+#define UREG_CHANNEL_ZERO_SHIFT       4
 #define UREG_CHANNEL_ONE_NEGATE_MBZ   1
-#define UREG_CHANNEL_ONE_SHIFT        0      
+#define UREG_CHANNEL_ONE_SHIFT        0
 
-#define UREG_BAD          0xffffffff /* not a valid ureg */
+#define UREG_BAD          0xffffffff    /* not a valid ureg */
 
 #define X    SRC_X
 #define Y    SRC_Y
@@ -84,78 +84,76 @@
 
 /* One neat thing about the UREG representation:  
  */
-static __inline int swizzle( int reg, int x, int y, int z, int w )
+static INLINE int
+swizzle(int reg, int x, int y, int z, int w)
 {
    return ((reg & ~UREG_XYZW_CHANNEL_MASK) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, x ), 0 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, y ), 1 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, z ), 2 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, w ), 3 ));
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, x), 0) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, y), 1) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, z), 2) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, w), 3));
 }
 
 /* Another neat thing about the UREG representation:  
  */
-static __inline int negate( int reg, int x, int y, int z, int w )
+static INLINE int
+negate(int reg, int x, int y, int z, int w)
 {
-   return reg ^ (((x&1)<<UREG_CHANNEL_X_NEGATE_SHIFT)|
-		 ((y&1)<<UREG_CHANNEL_Y_NEGATE_SHIFT)|
-		 ((z&1)<<UREG_CHANNEL_Z_NEGATE_SHIFT)|
-		 ((w&1)<<UREG_CHANNEL_W_NEGATE_SHIFT));
+   return reg ^ (((x & 1) << UREG_CHANNEL_X_NEGATE_SHIFT) |
+                 ((y & 1) << UREG_CHANNEL_Y_NEGATE_SHIFT) |
+                 ((z & 1) << UREG_CHANNEL_Z_NEGATE_SHIFT) |
+                 ((w & 1) << UREG_CHANNEL_W_NEGATE_SHIFT));
 }
 
 
-extern GLuint i915_get_temp( struct i915_fragment_program *p );
-extern GLuint i915_get_utemp( struct i915_fragment_program *p );
-extern void i915_release_utemps( struct i915_fragment_program *p );
+extern GLuint i915_get_temp(struct i915_fragment_program *p);
+extern GLuint i915_get_utemp(struct i915_fragment_program *p);
+extern void i915_release_utemps(struct i915_fragment_program *p);
 
 
-extern GLuint i915_emit_texld( struct i915_fragment_program *p,
-			      GLuint dest,
-			      GLuint destmask,
-			      GLuint sampler,
-			      GLuint coord,
-			      GLuint op );
+extern GLuint i915_emit_texld(struct i915_fragment_program *p,
+                              GLuint live_regs,
+                              GLuint dest,
+                              GLuint destmask,
+                              GLuint sampler, GLuint coord, GLuint op);
 
-extern GLuint i915_emit_arith( struct i915_fragment_program *p,
-			      GLuint op,
-			      GLuint dest,
-			      GLuint mask,
-			      GLuint saturate,
-			      GLuint src0,
-			      GLuint src1,
-			      GLuint src2 );
+extern GLuint i915_emit_arith(struct i915_fragment_program *p,
+                              GLuint op,
+                              GLuint dest,
+                              GLuint mask,
+                              GLuint saturate,
+                              GLuint src0, GLuint src1, GLuint src2);
 
-extern GLuint i915_emit_decl( struct i915_fragment_program *p,
-			     GLuint type, GLuint nr, GLuint d0_flags );
+extern GLuint i915_emit_decl(struct i915_fragment_program *p,
+                             GLuint type, GLuint nr, GLuint d0_flags);
 
 
-extern GLuint i915_emit_const1f( struct i915_fragment_program *p, 
-				GLfloat c0 );
+extern GLuint i915_emit_const1f(struct i915_fragment_program *p, GLfloat c0);
 
-extern GLuint i915_emit_const2f( struct i915_fragment_program *p, 
-				GLfloat c0, GLfloat c1 );
+extern GLuint i915_emit_const2f(struct i915_fragment_program *p,
+                                GLfloat c0, GLfloat c1);
 
-extern GLuint i915_emit_const4fv( struct i915_fragment_program *p,
-				 const GLfloat *c );
+extern GLuint i915_emit_const4fv(struct i915_fragment_program *p,
+                                 const GLfloat * c);
 
-extern GLuint i915_emit_const4f( struct i915_fragment_program *p, 
-				GLfloat c0, GLfloat c1, 
-				GLfloat c2, GLfloat c3 );
+extern GLuint i915_emit_const4f(struct i915_fragment_program *p,
+                                GLfloat c0, GLfloat c1,
+                                GLfloat c2, GLfloat c3);
 
 
-extern GLuint i915_emit_param4fv( struct i915_fragment_program *p, 
-				 const GLfloat *values );
+extern GLuint i915_emit_param4fv(struct i915_fragment_program *p,
+                                 const GLfloat * values);
 
-extern void i915_program_error( struct i915_fragment_program *p,
-                                const char *msg );
+extern void i915_program_error(struct i915_fragment_program *p,
+                               const char *msg);
 
-extern void i915_init_program( i915ContextPtr i915,
-			      struct i915_fragment_program *p );
+extern void i915_init_program(struct i915_context *i915,
+                              struct i915_fragment_program *p);
 
-extern void i915_upload_program( i915ContextPtr i915, 
-				struct i915_fragment_program *p );
+extern void i915_upload_program(struct i915_context *i915,
+                                struct i915_fragment_program *p);
 
-extern void i915_fini_program( struct i915_fragment_program *p );
+extern void i915_fini_program(struct i915_fragment_program *p);
 
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_reg.h b/src/mesa/drivers/dri/i915/i915_reg.h
index 694cd4c8c3c..8891e11c6fd 100644
--- a/src/mesa/drivers/dri/i915/i915_reg.h
+++ b/src/mesa/drivers/dri/i915/i915_reg.h
@@ -34,8 +34,6 @@
 
 #define I915_SET_FIELD( var, mask, value ) (var &= ~(mask), var |= value)
 
-#define CMD_3D (0x3<<29)
-
 #define PRIM3D_INLINE		(CMD_3D | (0x1f<<24))
 #define PRIM3D_TRILIST		(0x0<<18)
 #define PRIM3D_TRISTRIP 	(0x1<<18)
@@ -112,6 +110,20 @@
 /* 3DSTATE_CHROMA_KEY */
 
 /* 3DSTATE_CLEAR_PARAMETERS, p150 */
+/* 
+ * Sets the color, depth and stencil clear values used by the
+ * CLEAR_RECT and ZONE_INIT primitive types, respectively.  These
+ * primitives set override most 3d state and only take a minimal x/y
+ * vertex.  The color/z/stencil information is supplied here and
+ * therefore cannot vary per vertex.
+ */
+#define _3DSTATE_CLEAR_PARAMETERS	(CMD_3D | (0x1d<<24) | (0x9c<<16) | 5)
+/* Dword 1 */
+#define CLEARPARAM_CLEAR_RECT		(1 << 16)
+#define CLEARPARAM_ZONE_INIT		(0 << 16)
+#define CLEARPARAM_WRITE_COLOR		(1 << 2)
+#define CLEARPARAM_WRITE_DEPTH		(1 << 1)
+#define CLEARPARAM_WRITE_STENCIL	(1 << 0)
 
 /* 3DSTATE_CONSTANT_BLEND_COLOR, p153 */
 #define _3DSTATE_CONST_BLEND_COLOR_CMD	(CMD_3D | (0x1d<<24) | (0x88<<16))
@@ -313,120 +325,21 @@
 #define SCISSOR_RECT_0_YMAX(x)		((x)<<16)
 #define SCISSOR_RECT_0_XMAX(x)		(x)
 
-/* p189 */
-#define _3DSTATE_LOAD_STATE_IMMEDIATE_1   ((0x3<<29)|(0x1d<<24)|(0x04<<16))
-#define I1_LOAD_S(n)                      (1<<(4+n))
-
-#define S0_VB_OFFSET_MASK              0xffffffc
-#define S0_AUTO_CACHE_INV_DISABLE      (1<<0)
-
-#define S1_VERTEX_WIDTH_SHIFT          24
-#define S1_VERTEX_WIDTH_MASK           (0x3f<<24)
-#define S1_VERTEX_PITCH_SHIFT          16
-#define S1_VERTEX_PITCH_MASK           (0x3f<<16)
-
-#define TEXCOORDFMT_2D                 0x0
-#define TEXCOORDFMT_3D                 0x1
-#define TEXCOORDFMT_4D                 0x2
-#define TEXCOORDFMT_1D                 0x3
-#define TEXCOORDFMT_2D_16              0x4
-#define TEXCOORDFMT_4D_16              0x5
-#define TEXCOORDFMT_NOT_PRESENT        0xf
-#define S2_TEXCOORD_FMT0_MASK            0xf
-#define S2_TEXCOORD_FMT1_SHIFT           4
-#define S2_TEXCOORD_FMT(unit, type)    ((type)<<(unit*4))
-#define S2_TEXCOORD_NONE               (~0)
-
-/* S3 not interesting */
-
-#define S4_POINT_WIDTH_SHIFT           23
-#define S4_POINT_WIDTH_MASK            (0x1ff<<23)
-#define S4_LINE_WIDTH_SHIFT            19
-#define S4_LINE_WIDTH_ONE              (0x2<<19)
-#define S4_LINE_WIDTH_MASK             (0xf<<19)
-#define S4_FLATSHADE_ALPHA             (1<<18)
-#define S4_FLATSHADE_FOG               (1<<17)
-#define S4_FLATSHADE_SPECULAR          (1<<16)
-#define S4_FLATSHADE_COLOR             (1<<15)
-#define S4_CULLMODE_BOTH	       (0<<13)
-#define S4_CULLMODE_NONE	       (1<<13)
-#define S4_CULLMODE_CW		       (2<<13)
-#define S4_CULLMODE_CCW		       (3<<13)
-#define S4_CULLMODE_MASK	       (3<<13)
-#define S4_VFMT_POINT_WIDTH            (1<<12)
-#define S4_VFMT_SPEC_FOG               (1<<11)
-#define S4_VFMT_COLOR                  (1<<10)
-#define S4_VFMT_DEPTH_OFFSET           (1<<9)
-#define S4_VFMT_XYZ     	       (1<<6)
-#define S4_VFMT_XYZW     	       (2<<6)
-#define S4_VFMT_XY     		       (3<<6)
-#define S4_VFMT_XYW     	       (4<<6)
-#define S4_VFMT_XYZW_MASK              (7<<6)
-#define S4_FORCE_DEFAULT_DIFFUSE       (1<<5)
-#define S4_FORCE_DEFAULT_SPECULAR      (1<<4)
-#define S4_LOCAL_DEPTH_OFFSET_ENABLE   (1<<3)
-#define S4_VFMT_FOG_PARAM              (1<<2)
-#define S4_SPRITE_POINT_ENABLE         (1<<1)
-#define S4_LINE_ANTIALIAS_ENABLE       (1<<0)
-
-#define S4_VFMT_MASK (S4_VFMT_POINT_WIDTH   | 	\
-		      S4_VFMT_SPEC_FOG      |	\
-		      S4_VFMT_COLOR         |	\
-		      S4_VFMT_DEPTH_OFFSET  |	\
-		      S4_VFMT_XYZW_MASK     |	\
-		      S4_VFMT_FOG_PARAM)
-
-
-#define S5_WRITEDISABLE_ALPHA          (1<<31)
-#define S5_WRITEDISABLE_RED            (1<<30)
-#define S5_WRITEDISABLE_GREEN          (1<<29)
-#define S5_WRITEDISABLE_BLUE           (1<<28)
-#define S5_WRITEDISABLE_MASK           (0xf<<28)
-#define S5_FORCE_DEFAULT_POINT_SIZE    (1<<27)
-#define S5_LAST_PIXEL_ENABLE           (1<<26)
-#define S5_GLOBAL_DEPTH_OFFSET_ENABLE  (1<<25)
-#define S5_FOG_ENABLE                  (1<<24)
-#define S5_STENCIL_REF_SHIFT           16
-#define S5_STENCIL_REF_MASK            (0xff<<16)
-#define S5_STENCIL_TEST_FUNC_SHIFT     13
-#define S5_STENCIL_TEST_FUNC_MASK      (0x7<<13)
-#define S5_STENCIL_FAIL_SHIFT          10
-#define S5_STENCIL_FAIL_MASK           (0x7<<10)
-#define S5_STENCIL_PASS_Z_FAIL_SHIFT   7
-#define S5_STENCIL_PASS_Z_FAIL_MASK    (0x7<<7)
-#define S5_STENCIL_PASS_Z_PASS_SHIFT   4
-#define S5_STENCIL_PASS_Z_PASS_MASK    (0x7<<4)
-#define S5_STENCIL_WRITE_ENABLE        (1<<3)
-#define S5_STENCIL_TEST_ENABLE         (1<<2)
-#define S5_COLOR_DITHER_ENABLE         (1<<1)
-#define S5_LOGICOP_ENABLE              (1<<0)
-
-
-#define S6_ALPHA_TEST_ENABLE           (1<<31)
-#define S6_ALPHA_TEST_FUNC_SHIFT       28
-#define S6_ALPHA_TEST_FUNC_MASK        (0x7<<28)
-#define S6_ALPHA_REF_SHIFT             20
-#define S6_ALPHA_REF_MASK              (0xff<<20)
-#define S6_DEPTH_TEST_ENABLE           (1<<19)
-#define S6_DEPTH_TEST_FUNC_SHIFT       16
-#define S6_DEPTH_TEST_FUNC_MASK        (0x7<<16)
-#define S6_CBUF_BLEND_ENABLE           (1<<15)
-#define S6_CBUF_BLEND_FUNC_SHIFT       12
-#define S6_CBUF_BLEND_FUNC_MASK        (0x7<<12)
-#define S6_CBUF_SRC_BLEND_FACT_SHIFT   8
-#define S6_CBUF_SRC_BLEND_FACT_MASK    (0xf<<8)
-#define S6_CBUF_DST_BLEND_FACT_SHIFT   4
-#define S6_CBUF_DST_BLEND_FACT_MASK    (0xf<<4)
-#define S6_DEPTH_WRITE_ENABLE          (1<<3)
-#define S6_COLOR_WRITE_ENABLE          (1<<2)
-#define S6_TRISTRIP_PV_SHIFT           0
-#define S6_TRISTRIP_PV_MASK            (0x3<<0)
-
-#define S7_DEPTH_OFFSET_CONST_MASK     ~0
+/* Helper macros for blend factors
+ */
+#define DST_BLND_FACT(f) ((f)<<S6_CBUF_DST_BLEND_FACT_SHIFT)
+#define SRC_BLND_FACT(f) ((f)<<S6_CBUF_SRC_BLEND_FACT_SHIFT)
+#define DST_ABLND_FACT(f) ((f)<<IAB_DST_FACTOR_SHIFT)
+#define SRC_ABLND_FACT(f) ((f)<<IAB_SRC_FACTOR_SHIFT)
+
+
+
 
 /* 3DSTATE_MAP_DEINTERLACER_PARAMETERS */
-/* 3DSTATE_MAP_PALETTE_LOAD_32, p206 */
 
+/* 3DSTATE_MAP_PALETTE_LOAD_32, p206 */
+#define _3DSTATE_MAP_PALETTE_LOAD_32    (CMD_3D|(0x1d<<24)|(0x8f<<16))
+/* subsequent dwords up to length (max 16) are ARGB8888 color values */
 
 /* _3DSTATE_MODES_4, p218 */
 #define _3DSTATE_MODES_4_CMD		(CMD_3D|(0x0d<<24))
@@ -435,7 +348,7 @@
 #define LOGICOP_MASK			(0xf<<18)
 #define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
 #define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		((x)<<8)
+#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
 #define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
 #define ENABLE_STENCIL_WRITE_MASK	(1<<16)
 #define STENCIL_WRITE_MASK(x)		((x)&0xff)
@@ -458,7 +371,7 @@
 
 
 #define I915_MAX_TEX_INDIRECT 4
-#define I915_MAX_TEX_INSN     32     
+#define I915_MAX_TEX_INSN     32
 #define I915_MAX_ALU_INSN     64
 #define I915_MAX_DECL_INSN    27
 #define I915_MAX_TEMPORARY    16
@@ -470,33 +383,33 @@
  */
 #define _3DSTATE_PIXEL_SHADER_PROGRAM    (CMD_3D|(0x1d<<24)|(0x5<<16))
 
-#define REG_TYPE_R                 0 /* temporary regs, no need to
-				      * dcl, must be written before
-				      * read -- Preserved between
-				      * phases. 
-				      */
-#define REG_TYPE_T                 1 /* Interpolated values, must be
-				      * dcl'ed before use.
-				      *
-				      * 0..7: texture coord,
-				      * 8: diffuse spec,
-				      * 9: specular color,
-				      * 10: fog parameter in w.
-				      */
-#define REG_TYPE_CONST             2 /* Restriction: only one const
-				      * can be referenced per
-				      * instruction, though it may be
-				      * selected for multiple inputs.
-				      * Constants not initialized
-				      * default to zero.
-				      */
-#define REG_TYPE_S                 3 /* sampler */
-#define REG_TYPE_OC                4 /* output color (rgba) */
-#define REG_TYPE_OD                5 /* output depth (w), xyz are
-				      * temporaries.  If not written,
-				      * interpolated depth is used?
-				      */
-#define REG_TYPE_U                 6 /* unpreserved temporaries */
+#define REG_TYPE_R                 0    /* temporary regs, no need to
+                                         * dcl, must be written before
+                                         * read -- Preserved between
+                                         * phases. 
+                                         */
+#define REG_TYPE_T                 1    /* Interpolated values, must be
+                                         * dcl'ed before use.
+                                         *
+                                         * 0..7: texture coord,
+                                         * 8: diffuse spec,
+                                         * 9: specular color,
+                                         * 10: fog parameter in w.
+                                         */
+#define REG_TYPE_CONST             2    /* Restriction: only one const
+                                         * can be referenced per
+                                         * instruction, though it may be
+                                         * selected for multiple inputs.
+                                         * Constants not initialized
+                                         * default to zero.
+                                         */
+#define REG_TYPE_S                 3    /* sampler */
+#define REG_TYPE_OC                4    /* output color (rgba) */
+#define REG_TYPE_OD                5    /* output depth (w), xyz are
+                                         * temporaries.  If not written,
+                                         * interpolated depth is used?
+                                         */
+#define REG_TYPE_U                 6    /* unpreserved temporaries */
 #define REG_TYPE_MASK              0x7
 #define REG_NR_MASK                0xf
 
@@ -513,34 +426,34 @@
 #define T_TEX7     7
 #define T_DIFFUSE  8
 #define T_SPECULAR 9
-#define T_FOG_W    10		/* interpolated fog is in W coord */
+#define T_FOG_W    10           /* interpolated fog is in W coord */
 
 /* Arithmetic instructions */
 
 /* .replicate_swizzle == selection and replication of a particular
  * scalar channel, ie., .xxxx, .yyyy, .zzzz or .wwww 
  */
-#define A0_NOP    (0x0<<24)		/* no operation */
-#define A0_ADD    (0x1<<24)		/* dst = src0 + src1 */
-#define A0_MOV    (0x2<<24)		/* dst = src0 */
-#define A0_MUL    (0x3<<24)		/* dst = src0 * src1 */
-#define A0_MAD    (0x4<<24)		/* dst = src0 * src1 + src2 */
-#define A0_DP2ADD (0x5<<24)		/* dst.xyzw = src0.xy dot src1.xy + src2.replicate_swizzle */
-#define A0_DP3    (0x6<<24)		/* dst.xyzw = src0.xyz dot src1.xyz */
-#define A0_DP4    (0x7<<24)		/* dst.xyzw = src0.xyzw dot src1.xyzw */
-#define A0_FRC    (0x8<<24)		/* dst = src0 - floor(src0) */
-#define A0_RCP    (0x9<<24)		/* dst.xyzw = 1/(src0.replicate_swizzle) */
-#define A0_RSQ    (0xa<<24)		/* dst.xyzw = 1/(sqrt(abs(src0.replicate_swizzle))) */
-#define A0_EXP    (0xb<<24)		/* dst.xyzw = exp2(src0.replicate_swizzle) */
-#define A0_LOG    (0xc<<24)		/* dst.xyzw = log2(abs(src0.replicate_swizzle)) */
-#define A0_CMP    (0xd<<24)		/* dst = (src0 >= 0.0) ? src1 : src2 */
-#define A0_MIN    (0xe<<24)		/* dst = (src0 < src1) ? src0 : src1 */
-#define A0_MAX    (0xf<<24)		/* dst = (src0 >= src1) ? src0 : src1 */
-#define A0_FLR    (0x10<<24)		/* dst = floor(src0) */
-#define A0_MOD    (0x11<<24)		/* dst = src0 fmod 1.0 */
-#define A0_TRC    (0x12<<24)		/* dst = int(src0) */
-#define A0_SGE    (0x13<<24)		/* dst = src0 >= src1 ? 1.0 : 0.0 */
-#define A0_SLT    (0x14<<24)		/* dst = src0 < src1 ? 1.0 : 0.0 */
+#define A0_NOP    (0x0<<24)     /* no operation */
+#define A0_ADD    (0x1<<24)     /* dst = src0 + src1 */
+#define A0_MOV    (0x2<<24)     /* dst = src0 */
+#define A0_MUL    (0x3<<24)     /* dst = src0 * src1 */
+#define A0_MAD    (0x4<<24)     /* dst = src0 * src1 + src2 */
+#define A0_DP2ADD (0x5<<24)     /* dst.xyzw = src0.xy dot src1.xy + src2.replicate_swizzle */
+#define A0_DP3    (0x6<<24)     /* dst.xyzw = src0.xyz dot src1.xyz */
+#define A0_DP4    (0x7<<24)     /* dst.xyzw = src0.xyzw dot src1.xyzw */
+#define A0_FRC    (0x8<<24)     /* dst = src0 - floor(src0) */
+#define A0_RCP    (0x9<<24)     /* dst.xyzw = 1/(src0.replicate_swizzle) */
+#define A0_RSQ    (0xa<<24)     /* dst.xyzw = 1/(sqrt(abs(src0.replicate_swizzle))) */
+#define A0_EXP    (0xb<<24)     /* dst.xyzw = exp2(src0.replicate_swizzle) */
+#define A0_LOG    (0xc<<24)     /* dst.xyzw = log2(abs(src0.replicate_swizzle)) */
+#define A0_CMP    (0xd<<24)     /* dst = (src0 >= 0.0) ? src1 : src2 */
+#define A0_MIN    (0xe<<24)     /* dst = (src0 < src1) ? src0 : src1 */
+#define A0_MAX    (0xf<<24)     /* dst = (src0 >= src1) ? src0 : src1 */
+#define A0_FLR    (0x10<<24)    /* dst = floor(src0) */
+#define A0_MOD    (0x11<<24)    /* dst = src0 fmod 1.0 */
+#define A0_TRC    (0x12<<24)    /* dst = int(src0) */
+#define A0_SGE    (0x13<<24)    /* dst = src0 >= src1 ? 1.0 : 0.0 */
+#define A0_SLT    (0x14<<24)    /* dst = src0 < src1 ? 1.0 : 0.0 */
 #define A0_DEST_SATURATE                 (1<<22)
 #define A0_DEST_TYPE_SHIFT                19
 /* Allow: R, OC, OD, U */
@@ -599,23 +512,23 @@
 
 
 /* Texture instructions */
-#define T0_TEXLD     (0x15<<24)	/* Sample texture using predeclared
-				 * sampler and address, and output
-				 * filtered texel data to destination
-				 * register */
-#define T0_TEXLDP    (0x16<<24)	/* Same as texld but performs a
-				 * perspective divide of the texture
-				 * coordinate .xyz values by .w before
-				 * sampling. */
-#define T0_TEXLDB    (0x17<<24)	/* Same as texld but biases the
-				 * computed LOD by w.  Only S4.6 two's
-				 * comp is used.  This implies that a
-				 * float to fixed conversion is
-				 * done. */
-#define T0_TEXKILL   (0x18<<24)	/* Does not perform a sampling
-				 * operation.  Simply kills the pixel
-				 * if any channel of the address
-				 * register is < 0.0. */
+#define T0_TEXLD     (0x15<<24) /* Sample texture using predeclared
+                                 * sampler and address, and output
+                                 * filtered texel data to destination
+                                 * register */
+#define T0_TEXLDP    (0x16<<24) /* Same as texld but performs a
+                                 * perspective divide of the texture
+                                 * coordinate .xyz values by .w before
+                                 * sampling. */
+#define T0_TEXLDB    (0x17<<24) /* Same as texld but biases the
+                                 * computed LOD by w.  Only S4.6 two's
+                                 * comp is used.  This implies that a
+                                 * float to fixed conversion is
+                                 * done. */
+#define T0_TEXKILL   (0x18<<24) /* Does not perform a sampling
+                                 * operation.  Simply kills the pixel
+                                 * if any channel of the address
+                                 * register is < 0.0. */
 #define T0_DEST_TYPE_SHIFT                19
 /* Allow: R, OC, OD, U */
 /* Note: U (unpreserved) regs do not retain their values between
@@ -627,18 +540,18 @@
  */
 #define T0_DEST_NR_SHIFT                 14
 /* Allow R: 0..15, OC,OD: 0..0, U: 0..2 */
-#define T0_SAMPLER_NR_SHIFT              0 /* This field ignored for TEXKILL */
+#define T0_SAMPLER_NR_SHIFT              0      /* This field ignored for TEXKILL */
 #define T0_SAMPLER_NR_MASK               (0xf<<0)
 
-#define T1_ADDRESS_REG_TYPE_SHIFT        24 /* Reg to use as texture coord */
+#define T1_ADDRESS_REG_TYPE_SHIFT        24     /* Reg to use as texture coord */
 /* Allow R, T, OC, OD -- R, OC, OD are 'dependent' reads, new program phase */
 #define T1_ADDRESS_REG_NR_SHIFT          17
 #define T2_MBZ                           0
 
 /* Declaration instructions */
-#define D0_DCL       (0x19<<24)	/* Declare a t (interpolated attrib)
-				 * register or an s (sampler)
-				 * register. */
+#define D0_DCL       (0x19<<24) /* Declare a t (interpolated attrib)
+                                 * register or an s (sampler)
+                                 * register. */
 #define D0_SAMPLE_TYPE_SHIFT              22
 #define D0_SAMPLE_TYPE_2D                 (0x0<<22)
 #define D0_SAMPLE_TYPE_CUBE               (0x1<<22)
@@ -695,12 +608,12 @@
 #define    MAPSURF_4BIT_INDEXED		   (7<<7)
 #define MS3_MT_FORMAT_MASK         (0x7 << 3)
 #define MS3_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3) /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_I8		           (0<<3) /* SURFACE_8BIT */
+#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
+#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
 #define    MT_8BIT_L8		           (1<<3)
 #define    MT_8BIT_A8		           (4<<3)
 #define    MT_8BIT_MONO8	           (5<<3)
-#define    MT_16BIT_RGB565 		   (0<<3) /* SURFACE_16BIT */
+#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
 #define    MT_16BIT_ARGB1555		   (1<<3)
 #define    MT_16BIT_ARGB4444		   (2<<3)
 #define    MT_16BIT_AY88		   (3<<3)
@@ -709,7 +622,7 @@
 #define    MT_16BIT_I16	                   (7<<3)
 #define    MT_16BIT_L16	                   (8<<3)
 #define    MT_16BIT_A16	                   (9<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3) /* SURFACE_32BIT */
+#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
 #define    MT_32BIT_ABGR8888		   (1<<3)
 #define    MT_32BIT_XRGB8888		   (2<<3)
 #define    MT_32BIT_XBGR8888		   (3<<3)
@@ -725,11 +638,11 @@
 #define    MT_32BIT_xI824	           (0xD<<3)
 #define    MT_32BIT_xA824	           (0xE<<3)
 #define    MT_32BIT_xL824	           (0xF<<3)
-#define    MT_422_YCRCB_SWAPY	           (0<<3) /* SURFACE_422 */
+#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
 #define    MT_422_YCRCB_NORMAL	           (1<<3)
 #define    MT_422_YCRCB_SWAPUV	           (2<<3)
 #define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3) /* SURFACE_COMPRESSED */
+#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
 #define    MT_COMPRESS_DXT2_3	           (1<<3)
 #define    MT_COMPRESS_DXT4_5	           (2<<3)
 #define    MT_COMPRESS_FXT1		   (3<<3)
@@ -751,7 +664,7 @@
 #define MS4_MIP_LAYOUT_LEGACY           (0<<8)
 #define MS4_MIP_LAYOUT_BELOW_LPT        (0<<8)
 #define MS4_MIP_LAYOUT_RIGHT_LPT        (1<<8)
-#define MS4_VOLUME_DEPTH_SHIFT          0    
+#define MS4_VOLUME_DEPTH_SHIFT          0
 #define MS4_VOLUME_DEPTH_MASK           (0xff<<0)
 
 /* p244 */
@@ -779,7 +692,7 @@
 #define   FILTER_4X4_1    	3
 #define   FILTER_4X4_2    	4
 #define   FILTER_4X4_FLAT 	5
-#define   FILTER_6X5_MONO   	6 /* XXX - check */
+#define   FILTER_6X5_MONO   	6       /* XXX - check */
 #define SS2_MIN_FILTER_SHIFT          14
 #define SS2_MIN_FILTER_MASK           (0x7<<14)
 #define SS2_LOD_BIAS_SHIFT            5
@@ -826,10 +739,8 @@
 #define ST1_ENABLE               (1<<16)
 #define ST1_MASK                 (0xffff)
 
-
-#define MI_FLUSH           ((0<<29)|(4<<23))
-#define FLUSH_MAP_CACHE    (1<<0)
-#define FLUSH_RENDER_CACHE (1<<1)
-
+#define _3DSTATE_DEFAULT_Z          ((0x3<<29)|(0x1d<<24)|(0x98<<16))
+#define _3DSTATE_DEFAULT_DIFFUSE    ((0x3<<29)|(0x1d<<24)|(0x99<<16))
+#define _3DSTATE_DEFAULT_SPECULAR   ((0x3<<29)|(0x1d<<24)|(0x9a<<16))
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c
index 1c4ec747558..9d04358e4f3 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -26,11 +26,11 @@
  **************************************************************************/
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/dd.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 
@@ -38,97 +38,99 @@
 
 #include "drivers/common/driverfuncs.h"
 
+#include "intel_fbo.h"
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-
+#define FILE_DEBUG_FLAG DEBUG_STATE
 
 static void
-i915StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func, GLint ref,
+i915StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
                         GLuint mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
 
    mask = mask & 0xff;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(func), ref, mask);
+   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(func), ref, mask);
 
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					 STENCIL_TEST_MASK(mask));
+                                           STENCIL_TEST_MASK(mask));
 
    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-					S5_STENCIL_TEST_FUNC_MASK);
-					
-   i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) |  
- 				       (test << S5_STENCIL_TEST_FUNC_SHIFT)); 
+                                          S5_STENCIL_TEST_FUNC_MASK);
+
+   i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) |
+                                         (test <<
+                                          S5_STENCIL_TEST_FUNC_SHIFT));
 }
 
 static void
-i915StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+i915StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : mask 0x%x\n", __FUNCTION__, mask);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
+   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
+   
    mask = mask & 0xff;
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(mask));
+                                           STENCIL_WRITE_MASK(mask));
 }
 
 
 static void
-i915StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
+i915StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
                       GLenum zpass)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int fop = intel_translate_stencil_op(fail); 
-   int dfop = intel_translate_stencil_op(zfail); 
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int fop = intel_translate_stencil_op(fail);
+   int dfop = intel_translate_stencil_op(zfail);
    int dpop = intel_translate_stencil_op(zpass);
 
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(fail),
-	      _mesa_lookup_enum_by_nr(zfail),
-	      _mesa_lookup_enum_by_nr(zpass));
+   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(fail),
+       _mesa_lookup_enum_by_nr(zfail), _mesa_lookup_enum_by_nr(zpass));
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-					S5_STENCIL_PASS_Z_FAIL_MASK |
-					S5_STENCIL_PASS_Z_PASS_MASK);
+                                          S5_STENCIL_PASS_Z_FAIL_MASK |
+                                          S5_STENCIL_PASS_Z_PASS_MASK);
 
    i915->state.Ctx[I915_CTXREG_LIS5] |= ((fop << S5_STENCIL_FAIL_SHIFT) |
-				       (dfop << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-				       (dpop << S5_STENCIL_PASS_Z_PASS_SHIFT));
+                                         (dfop <<
+                                          S5_STENCIL_PASS_Z_FAIL_SHIFT) |
+                                         (dpop <<
+                                          S5_STENCIL_PASS_Z_PASS_SHIFT));
 }
 
-static void i915AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+static void
+i915AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
    GLubyte refByte;
 
    UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_LIS6] &= ~(S6_ALPHA_TEST_FUNC_MASK |
-					S6_ALPHA_REF_MASK);
+                                          S6_ALPHA_REF_MASK);
    i915->state.Ctx[I915_CTXREG_LIS6] |= ((test << S6_ALPHA_TEST_FUNC_SHIFT) |
-				       (((GLuint)refByte) << S6_ALPHA_REF_SHIFT));
+                                         (((GLuint) refByte) <<
+                                          S6_ALPHA_REF_SHIFT));
 }
 
 /* This function makes sure that the proper enables are
@@ -137,41 +139,45 @@ static void i915AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
  * could change the LogicOp or Independant Alpha Blend without subsequent
  * calls to glEnable.
  */
-static void i915EvalLogicOpBlendState(GLcontext *ctx)
+static void
+i915EvalLogicOpBlendState(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (RGBA_LOGICOP_ENABLED(ctx)) {
       i915->state.Ctx[I915_CTXREG_LIS5] |= S5_LOGICOP_ENABLE;
       i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
-   } else {
+   }
+   else {
       i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_LOGICOP_ENABLE;
 
       if (ctx->Color.BlendEnabled) {
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_CBUF_BLEND_ENABLE;
-      } else {
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_CBUF_BLEND_ENABLE;
+      }
+      else {
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
       }
    }
 }
 
-static void i915BlendColor(GLcontext *ctx, const GLfloat color[4])
+static void
+i915BlendColor(GLcontext * ctx, const GLfloat color[4])
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLubyte r, g, b, a;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = (a<<24) | (r<<16) | (g<<8) | b;
+   i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] =
+      (a << 24) | (r << 16) | (g << 8) | b;
 }
 
 
@@ -182,31 +188,37 @@ static void i915BlendColor(GLcontext *ctx, const GLfloat color[4])
 
 
 
-static GLuint translate_blend_equation( GLenum mode )
+static GLuint
+translate_blend_equation(GLenum mode)
 {
    switch (mode) {
-   case GL_FUNC_ADD: return BLENDFUNC_ADD; 
-   case GL_MIN: return BLENDFUNC_MIN; 
-   case GL_MAX: return BLENDFUNC_MAX; 
-   case GL_FUNC_SUBTRACT: return BLENDFUNC_SUBTRACT; 
-   case GL_FUNC_REVERSE_SUBTRACT: return BLENDFUNC_REVERSE_SUBTRACT; 
-   default: return 0;
+   case GL_FUNC_ADD:
+      return BLENDFUNC_ADD;
+   case GL_MIN:
+      return BLENDFUNC_MIN;
+   case GL_MAX:
+      return BLENDFUNC_MAX;
+   case GL_FUNC_SUBTRACT:
+      return BLENDFUNC_SUBTRACT;
+   case GL_FUNC_REVERSE_SUBTRACT:
+      return BLENDFUNC_REVERSE_SUBTRACT;
+   default:
+      return 0;
    }
 }
 
-static void i915UpdateBlendState( GLcontext *ctx )
+static void
+i915UpdateBlendState(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   GLuint iab = (i915->state.Ctx[I915_CTXREG_IAB] & 
-		 ~(IAB_SRC_FACTOR_MASK |
-		   IAB_DST_FACTOR_MASK |
-		   (BLENDFUNC_MASK << IAB_FUNC_SHIFT) |
-		   IAB_ENABLE));
-
-   GLuint lis6 = (i915->state.Ctx[I915_CTXREG_LIS6] & 
-		  ~(S6_CBUF_SRC_BLEND_FACT_MASK |
-		    S6_CBUF_DST_BLEND_FACT_MASK |
-		    S6_CBUF_BLEND_FUNC_MASK));
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   GLuint iab = (i915->state.Ctx[I915_CTXREG_IAB] &
+                 ~(IAB_SRC_FACTOR_MASK |
+                   IAB_DST_FACTOR_MASK |
+                   (BLENDFUNC_MASK << IAB_FUNC_SHIFT) | IAB_ENABLE));
+
+   GLuint lis6 = (i915->state.Ctx[I915_CTXREG_LIS6] &
+                  ~(S6_CBUF_SRC_BLEND_FACT_MASK |
+                    S6_CBUF_DST_BLEND_FACT_MASK | S6_CBUF_BLEND_FUNC_MASK));
 
    GLuint eqRGB = ctx->Color.BlendEquationRGB;
    GLuint eqA = ctx->Color.BlendEquationA;
@@ -223,15 +235,15 @@ static void i915UpdateBlendState( GLcontext *ctx )
       srcA = dstA = GL_ONE;
    }
 
-   lis6 |= SRC_BLND_FACT(intel_translate_blend_factor(srcRGB)); 
-   lis6 |= DST_BLND_FACT(intel_translate_blend_factor(dstRGB)); 
-   lis6 |= translate_blend_equation( eqRGB ) << S6_CBUF_BLEND_FUNC_SHIFT;
+   lis6 |= SRC_BLND_FACT(intel_translate_blend_factor(srcRGB));
+   lis6 |= DST_BLND_FACT(intel_translate_blend_factor(dstRGB));
+   lis6 |= translate_blend_equation(eqRGB) << S6_CBUF_BLEND_FUNC_SHIFT;
 
-   iab |= SRC_ABLND_FACT(intel_translate_blend_factor(srcA)); 
-   iab |= DST_ABLND_FACT(intel_translate_blend_factor(dstA)); 
-   iab |= translate_blend_equation( eqA ) << IAB_FUNC_SHIFT;
+   iab |= SRC_ABLND_FACT(intel_translate_blend_factor(srcA));
+   iab |= DST_ABLND_FACT(intel_translate_blend_factor(dstA));
+   iab |= translate_blend_equation(eqA) << IAB_FUNC_SHIFT;
 
-   if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB) 
+   if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB)
       iab |= IAB_ENABLE;
 
    if (iab != i915->state.Ctx[I915_CTXREG_IAB] ||
@@ -246,41 +258,41 @@ static void i915UpdateBlendState( GLcontext *ctx )
 }
 
 
-static void i915BlendFuncSeparate(GLcontext *ctx, GLenum srcRGB, 
-				 GLenum dstRGB, GLenum srcA,
-				 GLenum dstA )
-{  
-   i915UpdateBlendState( ctx );
+static void
+i915BlendFuncSeparate(GLcontext * ctx, GLenum srcRGB,
+                      GLenum dstRGB, GLenum srcA, GLenum dstA)
+{
+   i915UpdateBlendState(ctx);
 }
 
 
-static void i915BlendEquationSeparate(GLcontext *ctx, GLenum eqRGB,
-				     GLenum eqA) 
+static void
+i915BlendEquationSeparate(GLcontext * ctx, GLenum eqRGB, GLenum eqA)
 {
-   i915UpdateBlendState( ctx );
+   i915UpdateBlendState(ctx);
 }
 
 
-static void i915DepthFunc(GLcontext *ctx, GLenum func)
+static void
+i915DepthFunc(GLcontext * ctx, GLenum func)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
 
+   DBG("%s\n", __FUNCTION__);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
    i915->state.Ctx[I915_CTXREG_LIS6] |= test << S6_DEPTH_TEST_FUNC_SHIFT;
 }
 
-static void i915DepthMask(GLcontext *ctx, GLboolean flag)
+static void
+i915DepthMask(GLcontext * ctx, GLboolean flag)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s flag (%d)\n", __FUNCTION__, flag);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
+   DBG("%s flag (%d)\n", __FUNCTION__, flag);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (flag && ctx->Depth.Test)
@@ -295,14 +307,15 @@ static void i915DepthMask(GLcontext *ctx, GLboolean flag)
  * The i915 supports a 4x4 stipple natively, GL wants 32x32.
  * Fortunately stipple is usually a repeating pattern.
  */
-static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
+static void
+i915PolygonStipple(GLcontext * ctx, const GLubyte * mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   const GLubyte *m = mask;
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   const GLubyte *m;
    GLubyte p[4];
-   int i,j,k;
+   int i, j, k;
    int active = (ctx->Polygon.StippleFlag &&
-		 i915->intel.reduced_primitive == GL_TRIANGLES);
+                 i915->intel.reduced_primitive == GL_TRIANGLES);
    GLuint newMask;
 
    if (active) {
@@ -310,23 +323,32 @@ static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
       i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
    }
 
-   p[0] = mask[12] & 0xf; p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf; p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf; p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf; p[3] |= p[3] << 4;
-
-   for (k = 0 ; k < 8 ; k++)
-      for (j = 3 ; j >= 0; j--)
-	 for (i = 0 ; i < 4 ; i++, m++)
-	    if (*m != p[j]) {
-	       i915->intel.hw_stipple = 0;
-	       return;
-	    }
+   /* Use the already unpacked stipple data from the context rather than the
+    * uninterpreted mask passed in.
+    */
+   mask = (const GLubyte *)ctx->PolygonStipple;
+   m = mask;
+
+   p[0] = mask[12] & 0xf;
+   p[0] |= p[0] << 4;
+   p[1] = mask[8] & 0xf;
+   p[1] |= p[1] << 4;
+   p[2] = mask[4] & 0xf;
+   p[2] |= p[2] << 4;
+   p[3] = mask[0] & 0xf;
+   p[3] |= p[3] << 4;
+
+   for (k = 0; k < 8; k++)
+      for (j = 3; j >= 0; j--)
+         for (i = 0; i < 4; i++, m++)
+            if (*m != p[j]) {
+               i915->intel.hw_stipple = 0;
+               return;
+            }
 
    newMask = (((p[0] & 0xf) << 0) |
-	      ((p[1] & 0xf) << 4) |
-	      ((p[2] & 0xf) << 8) |
-	      ((p[3] & 0xf) << 12));
+              ((p[1] & 0xf) << 4) |
+              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
 
 
    if (newMask == 0xffff || newMask == 0x0) {
@@ -347,49 +369,54 @@ static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
 /* =============================================================
  * Hardware clipping
  */
-static void i915Scissor(GLcontext *ctx, GLint x, GLint y, 
-			GLsizei w, GLsizei h)
+static void
+i915Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   intelScreenPrivate *screen = i915->intel.intelScreen;
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int x1, y1, x2, y2;
 
-   if (!i915->intel.driDrawable)
+   if (!ctx->DrawBuffer)
       return;
 
-   x1 = x;
-   y1 = i915->intel.driDrawable->h - (y + h);
-   x2 = x + w - 1;
-   y2 = y1 + h - 1;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "[%s] x(%d) y(%d) w(%d) h(%d)\n", __FUNCTION__,
-	      x, y, w, h);
-
-   if (x1 < 0) x1 = 0;
-   if (y1 < 0) y1 = 0;
-   if (x2 < 0) x2 = 0;
-   if (y2 < 0) y2 = 0;
-
-   if (x2 >= screen->width) x2 = screen->width-1;
-   if (y2 >= screen->height) y2 = screen->height-1;
-   if (x1 >= screen->width) x1 = screen->width-1;
-   if (y1 >= screen->height) y1 = screen->height-1;
+   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
 
+   if (ctx->DrawBuffer->Name == 0) {
+      x1 = x;
+      y1 = ctx->DrawBuffer->Height - (y + h);
+      x2 = x + w - 1;
+      y2 = y1 + h - 1;
+      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   else {
+      /* FBO - not inverted
+       */
+      x1 = x;
+      y1 = y;
+      x2 = x + w - 1;
+      y2 = y + h - 1;
+      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   
+   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
+   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
+   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
+   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
+   
+   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
 
    I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
    i915->state.Buffer[I915_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
    i915->state.Buffer[I915_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
 }
 
-static void i915LogicOp(GLcontext *ctx, GLenum opcode)
+static void
+i915LogicOp(GLcontext * ctx, GLenum opcode)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int tmp = intel_translate_logic_op(opcode);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~LOGICOP_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
@@ -397,13 +424,14 @@ static void i915LogicOp(GLcontext *ctx, GLenum opcode)
 
 
 
-static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
+static void
+i915CullFaceFrontFace(GLcontext * ctx, GLenum unused)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLuint mode;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   DBG("%s %d\n", __FUNCTION__,
+       ctx->DrawBuffer ? ctx->DrawBuffer->Name : 0);
 
    if (!ctx->Polygon.CullFlag) {
       mode = S4_CULLMODE_NONE;
@@ -411,10 +439,12 @@ static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    else if (ctx->Polygon.CullFaceMode != GL_FRONT_AND_BACK) {
       mode = S4_CULLMODE_CW;
 
+      if (ctx->DrawBuffer && ctx->DrawBuffer->Name != 0)
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
       if (ctx->Polygon.CullFaceMode == GL_FRONT)
-	 mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
       if (ctx->Polygon.FrontFace != GL_CCW)
-	 mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
    }
    else {
       mode = S4_CULLMODE_BOTH;
@@ -425,16 +455,16 @@ static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    i915->state.Ctx[I915_CTXREG_LIS4] |= mode;
 }
 
-static void i915LineWidth( GLcontext *ctx, GLfloat widthf )
+static void
+i915LineWidth(GLcontext * ctx, GLfloat widthf)
 {
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_LINE_WIDTH_MASK;
    int width;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   width = (int)(widthf * 2);
+   DBG("%s\n", __FUNCTION__);
+   
+   width = (int) (widthf * 2);
    CLAMP_SELF(width, 1, 0xf);
    lis4 |= width << S4_LINE_WIDTH_SHIFT;
 
@@ -444,15 +474,15 @@ static void i915LineWidth( GLcontext *ctx, GLfloat widthf )
    }
 }
 
-static void i915PointSize(GLcontext *ctx, GLfloat size)
+static void
+i915PointSize(GLcontext * ctx, GLfloat size)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_POINT_WIDTH_MASK;
-   GLint point_size = (int)size;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s\n", __FUNCTION__);
+   GLint point_size = (int) size;
 
+   DBG("%s\n", __FUNCTION__);
+   
    CLAMP_SELF(point_size, 1, 255);
    lis4 |= point_size << S4_POINT_WIDTH_SHIFT;
 
@@ -467,20 +497,24 @@ static void i915PointSize(GLcontext *ctx, GLfloat size)
  * Color masks
  */
 
-static void i915ColorMask(GLcontext *ctx,
-			 GLboolean r, GLboolean g,
-			 GLboolean b, GLboolean a)
+static void
+i915ColorMask(GLcontext * ctx,
+              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
 {
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLuint tmp = i915->state.Ctx[I915_CTXREG_LIS5] & ~S5_WRITEDISABLE_MASK;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
+   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b,
+       a);
 
-   if (!r) tmp |= S5_WRITEDISABLE_RED;
-   if (!g) tmp |= S5_WRITEDISABLE_GREEN;
-   if (!b) tmp |= S5_WRITEDISABLE_BLUE;
-   if (!a) tmp |= S5_WRITEDISABLE_ALPHA;
+   if (!r)
+      tmp |= S5_WRITEDISABLE_RED;
+   if (!g)
+      tmp |= S5_WRITEDISABLE_GREEN;
+   if (!b)
+      tmp |= S5_WRITEDISABLE_BLUE;
+   if (!a)
+      tmp |= S5_WRITEDISABLE_ALPHA;
 
    if (tmp != i915->state.Ctx[I915_CTXREG_LIS5]) {
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
@@ -488,54 +522,55 @@ static void i915ColorMask(GLcontext *ctx,
    }
 }
 
-static void update_specular( GLcontext *ctx )
+static void
+update_specular(GLcontext * ctx)
 {
    /* A hack to trigger the rebuild of the fragment program.
     */
-   INTEL_CONTEXT(ctx)->NewGLState |= _NEW_TEXTURE;
-   I915_CONTEXT(ctx)->tex_program.translated = 0; 
+   intel_context(ctx)->NewGLState |= _NEW_TEXTURE;
 }
 
-static void i915LightModelfv(GLcontext *ctx, GLenum pname, 
-			     const GLfloat *param)
+static void
+i915LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular( ctx );
+      update_specular(ctx);
    }
 }
 
-static void i915ShadeModel(GLcontext *ctx, GLenum mode)
+static void
+i915ShadeModel(GLcontext * ctx, GLenum mode)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (mode == GL_SMOOTH) {
-     i915->state.Ctx[I915_CTXREG_LIS4] &= ~(S4_FLATSHADE_ALPHA | 
-					  S4_FLATSHADE_COLOR | 
-					  S4_FLATSHADE_SPECULAR);
-   } else {
-     i915->state.Ctx[I915_CTXREG_LIS4] |= (S4_FLATSHADE_ALPHA | 
-					 S4_FLATSHADE_COLOR | 
-					 S4_FLATSHADE_SPECULAR);
+      i915->state.Ctx[I915_CTXREG_LIS4] &= ~(S4_FLATSHADE_ALPHA |
+                                             S4_FLATSHADE_COLOR |
+                                             S4_FLATSHADE_SPECULAR);
+   }
+   else {
+      i915->state.Ctx[I915_CTXREG_LIS4] |= (S4_FLATSHADE_ALPHA |
+                                            S4_FLATSHADE_COLOR |
+                                            S4_FLATSHADE_SPECULAR);
    }
 }
 
 /* =============================================================
  * Fog
  */
-void i915_update_fog( GLcontext *ctx )
+void
+i915_update_fog(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLenum mode;
    GLboolean enabled;
    GLboolean try_pixel_fog;
-   
+
    if (ctx->FragmentProgram._Active) {
       /* Pull in static fog state from program */
-      
       mode = ctx->FragmentProgram._Current->FogOption;
       enabled = (mode != GL_NONE);
       try_pixel_fog = 0;
@@ -546,7 +581,7 @@ void i915_update_fog( GLcontext *ctx )
 #if 0
       /* XXX - DISABLED -- Need ortho fallback */
       try_pixel_fog = (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT
-                       &&ctx->Hint.Fog == GL_NICEST);
+                       && ctx->Hint.Fog == GL_NICEST);
 #else
       try_pixel_fog = 0;
 #endif
@@ -559,48 +594,49 @@ void i915_update_fog( GLcontext *ctx )
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
       i915->vertex_fog = I915_FOG_PIXEL;
-	 
+
       switch (mode) {
       case GL_LINEAR:
-	 if (ctx->Fog.End <= ctx->Fog.Start) {
-	    /* XXX - this won't work with fragment programs.  Need to
-	     * either fallback or append fog instructions to end of
-	     * program in the case of linear fog.
-	     */
-	    i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
-	    i915->vertex_fog = I915_FOG_VERTEX;
-	 }
-	 else {
+         if (ctx->Fog.End <= ctx->Fog.Start) {
+            /* XXX - this won't work with fragment programs.  Need to
+             * either fallback or append fog instructions to end of
+             * program in the case of linear fog.
+             */
+            printf("vertex fog!\n");
+            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
+            i915->vertex_fog = I915_FOG_VERTEX;
+         }
+         else {
             GLfloat c2 = 1.0 / (ctx->Fog.End - ctx->Fog.Start);
             GLfloat c1 = ctx->Fog.End * c2;
 
-	    i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_C1_MASK;
-	    i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_LINEAR;
-	    i915->state.Fog[I915_FOGREG_MODE1] |= 
-	       ((GLuint)(c1 * FMC1_C1_ONE)) & FMC1_C1_MASK;
-
-	    if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-	       i915->state.Fog[I915_FOGREG_MODE2]
-                  = (GLuint)(c2 * FMC2_C2_ONE);
-	    }
-	    else {
-	       fi_type fi;
-	       fi.f = c2; 
-	       i915->state.Fog[I915_FOGREG_MODE2] = fi.i; 
-	    }
-	 }
-	 break;
+            i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_C1_MASK;
+            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_LINEAR;
+            i915->state.Fog[I915_FOGREG_MODE1] |=
+               ((GLuint) (c1 * FMC1_C1_ONE)) & FMC1_C1_MASK;
+
+            if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
+               i915->state.Fog[I915_FOGREG_MODE2]
+                  = (GLuint) (c2 * FMC2_C2_ONE);
+            }
+            else {
+               fi_type fi;
+               fi.f = c2;
+               i915->state.Fog[I915_FOGREG_MODE2] = fi.i;
+            }
+         }
+         break;
       case GL_EXP:
-	 i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP;
-	 break;
+         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP;
+         break;
       case GL_EXP2:
-	 i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP2;
-	 break;
+         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP2;
+         break;
       default:
-	 break;
+         break;
       }
    }
-   else /* if (i915->vertex_fog != I915_FOG_VERTEX) */ {      
+   else { /* if (i915->vertex_fog != I915_FOG_VERTEX) */
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
       i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
@@ -622,38 +658,38 @@ void i915_update_fog( GLcontext *ctx )
 }
 
 static void
-i915Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
+i915Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
    switch (pname) {
-   case GL_FOG_COORDINATE_SOURCE_EXT: 
+   case GL_FOG_COORDINATE_SOURCE_EXT:
    case GL_FOG_MODE:
    case GL_FOG_START:
-   case GL_FOG_END: 
+   case GL_FOG_END:
       break;
 
    case GL_FOG_DENSITY:
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
 
       if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-	 i915->state.Fog[I915_FOGREG_MODE3]
-            = (GLuint)(ctx->Fog.Density * FMC3_D_ONE);
+         i915->state.Fog[I915_FOGREG_MODE3] =
+            (GLuint) (ctx->Fog.Density * FMC3_D_ONE);
       }
       else {
-	 union { float f; int i; } fi;
-	 fi.f = ctx->Fog.Density; 
-	 i915->state.Fog[I915_FOGREG_MODE3] = fi.i; 
+         fi_type fi;
+         fi.f = ctx->Fog.Density;
+         i915->state.Fog[I915_FOGREG_MODE3] = fi.i;
       }
       break;
 
-   case GL_FOG_COLOR: 
+   case GL_FOG_COLOR:
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_COLOR] = 
-	 (_3DSTATE_FOG_COLOR_CMD | 
-	  ((GLubyte)(ctx->Fog.Color[0]*255.0F) << 16) |
-	  ((GLubyte)(ctx->Fog.Color[1]*255.0F) << 8) |
-	  ((GLubyte)(ctx->Fog.Color[2]*255.0F) << 0));
+      i915->state.Fog[I915_FOGREG_COLOR] =
+         (_3DSTATE_FOG_COLOR_CMD |
+          ((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
+          ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
+          ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
       break;
 
    default:
@@ -661,7 +697,8 @@ i915Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
    }
 }
 
-static void i915Hint(GLcontext *ctx, GLenum target, GLenum state)
+static void
+i915Hint(GLcontext * ctx, GLenum target, GLenum state)
 {
    switch (target) {
    case GL_FOG_HINT:
@@ -674,25 +711,26 @@ static void i915Hint(GLcontext *ctx, GLenum target, GLenum state)
 /* =============================================================
  */
 
-static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+static void
+i915Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
-   switch(cap) {
+   switch (cap) {
    case GL_TEXTURE_2D:
       break;
 
    case GL_LIGHTING:
    case GL_COLOR_SUM:
-      update_specular( ctx );
+      update_specular(ctx);
       break;
 
    case GL_ALPHA_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_ALPHA_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_ALPHA_TEST_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_ALPHA_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_ALPHA_TEST_ENABLE;
       break;
 
    case GL_BLEND:
@@ -704,8 +742,8 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 
       /* Logicop doesn't seem to work at 16bpp:
        */
-      if (i915->intel.intelScreen->cpp == 2)
-	 FALLBACK( &i915->intel, I915_FALLBACK_LOGICOP, state );
+      if (ctx->Visual.rgbBits == 16)
+         FALLBACK(&i915->intel, I915_FALLBACK_LOGICOP, state);
       break;
 
    case GL_FRAGMENT_PROGRAM_ARB:
@@ -714,37 +752,37 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
    case GL_DITHER:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_COLOR_DITHER_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_COLOR_DITHER_ENABLE;
       break;
 
    case GL_DEPTH_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_TEST_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_ENABLE;
 
-      i915DepthMask( ctx, ctx->Depth.Mask );
+      i915DepthMask(ctx, ctx->Depth.Mask);
       break;
 
    case GL_SCISSOR_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
       if (state)
-	 i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						   ENABLE_SCISSOR_RECT);
+         i915->state.Buffer[I915_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
       else
-	 i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						   DISABLE_SCISSOR_RECT);
+         i915->state.Buffer[I915_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
       break;
 
    case GL_LINE_SMOOTH:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS4] |= S4_LINE_ANTIALIAS_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS4] |= S4_LINE_ANTIALIAS_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_LINE_ANTIALIAS_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_LINE_ANTIALIAS_ENABLE;
       break;
 
    case GL_FOG:
@@ -755,16 +793,25 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       break;
 
    case GL_STENCIL_TEST:
-      if (i915->intel.hw_stencil) {
-	 I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-	 if (state)
-	    i915->state.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
-						S5_STENCIL_WRITE_ENABLE);
-	 else
-	    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE | 
-						 S5_STENCIL_WRITE_ENABLE);
-      } else {
-	 FALLBACK( &i915->intel, I915_FALLBACK_STENCIL, state );
+      {
+         GLboolean hw_stencil = GL_FALSE;
+         if (ctx->DrawBuffer) {
+            struct intel_renderbuffer *irbStencil
+               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
+            hw_stencil = (irbStencil && irbStencil->region);
+         }
+         if (hw_stencil) {
+            I915_STATECHANGE(i915, I915_UPLOAD_CTX);
+            if (state)
+               i915->state.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
+                                                     S5_STENCIL_WRITE_ENABLE);
+            else
+               i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
+                                                      S5_STENCIL_WRITE_ENABLE);
+         }
+         else {
+            FALLBACK(&i915->intel, I915_FALLBACK_STENCIL, state);
+         }
       }
       break;
 
@@ -773,23 +820,20 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
        * I'll do more testing later to find out exactly which hardware
        * supports it.  Disabled for now.
        */
-      if (i915->intel.hw_stipple && 
-	  i915->intel.reduced_primitive == GL_TRIANGLES)
-      {
-	 I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-	 if (state)
-	    i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
-	 else
-	    i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
+      if (i915->intel.hw_stipple &&
+          i915->intel.reduced_primitive == GL_TRIANGLES) {
+         I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
+         if (state)
+            i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
+         else
+            i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
       }
       break;
 
    case GL_POLYGON_SMOOTH:
-      FALLBACK( &i915->intel, I915_FALLBACK_POLYGON_SMOOTH, state );
       break;
 
    case GL_POINT_SMOOTH:
-      FALLBACK( &i915->intel, I915_FALLBACK_POINT_SMOOTH, state );
       break;
 
    default:
@@ -798,10 +842,9 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 }
 
 
-static void i915_init_packets( i915ContextPtr i915 )
+static void
+i915_init_packets(struct i915_context *i915)
 {
-   intelScreenPrivate *screen = i915->intel.intelScreen;
-
    /* Zero all state */
    memset(&i915->state, 0, sizeof(i915->state));
 
@@ -811,39 +854,35 @@ static void i915_init_packets( i915ContextPtr i915 )
       /* Probably don't want to upload all this stuff every time one 
        * piece changes.
        */
-      i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 | 
-				       I1_LOAD_S(2) |
-				       I1_LOAD_S(4) |
-				       I1_LOAD_S(5) |
-				       I1_LOAD_S(6) | 
-				       (3));
+      i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
+                                         I1_LOAD_S(2) |
+                                         I1_LOAD_S(4) |
+                                         I1_LOAD_S(5) | I1_LOAD_S(6) | (3));
       i915->state.Ctx[I915_CTXREG_LIS2] = 0;
       i915->state.Ctx[I915_CTXREG_LIS4] = 0;
       i915->state.Ctx[I915_CTXREG_LIS5] = 0;
 
-      if (screen->cpp == 2)
-	 i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
+      if (i915->intel.ctx.Visual.rgbBits == 16)
+         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
 
 
       i915->state.Ctx[I915_CTXREG_LIS6] = (S6_COLOR_WRITE_ENABLE |
-					 (2 << S6_TRISTRIP_PV_SHIFT));
+                                           (2 << S6_TRISTRIP_PV_SHIFT));
 
       i915->state.Ctx[I915_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-					   ENABLE_LOGIC_OP_FUNC |
-					   LOGIC_OP_FUNC(LOGICOP_COPY) |
-					   ENABLE_STENCIL_TEST_MASK |
-					   STENCIL_TEST_MASK(0xff) |
-					   ENABLE_STENCIL_WRITE_MASK |
-					   STENCIL_WRITE_MASK(0xff));
-
-
-      i915->state.Ctx[I915_CTXREG_IAB] = (_3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD |
-					IAB_MODIFY_ENABLE |
-					IAB_MODIFY_FUNC |
-					IAB_MODIFY_SRC_FACTOR |
-					IAB_MODIFY_DST_FACTOR);
-
-      i915->state.Ctx[I915_CTXREG_BLENDCOLOR0] = _3DSTATE_CONST_BLEND_COLOR_CMD;
+                                             ENABLE_LOGIC_OP_FUNC |
+                                             LOGIC_OP_FUNC(LOGICOP_COPY) |
+                                             ENABLE_STENCIL_TEST_MASK |
+                                             STENCIL_TEST_MASK(0xff) |
+                                             ENABLE_STENCIL_WRITE_MASK |
+                                             STENCIL_WRITE_MASK(0xff));
+
+      i915->state.Ctx[I915_CTXREG_IAB] =
+         (_3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD | IAB_MODIFY_ENABLE |
+          IAB_MODIFY_FUNC | IAB_MODIFY_SRC_FACTOR | IAB_MODIFY_DST_FACTOR);
+
+      i915->state.Ctx[I915_CTXREG_BLENDCOLOR0] =
+         _3DSTATE_CONST_BLEND_COLOR_CMD;
       i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = 0;
 
    }
@@ -858,79 +897,50 @@ static void i915_init_packets( i915ContextPtr i915 )
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE0] = _3DSTATE_FOG_MODE_CMD;
       i915->state.Fog[I915_FOGREG_MODE1] = (FMC1_FOGFUNC_MODIFY_ENABLE |
-					  FMC1_FOGFUNC_VERTEX |
-					  FMC1_FOGINDEX_MODIFY_ENABLE |
-					  FMC1_FOGINDEX_W |
-					  FMC1_C1_C2_MODIFY_ENABLE |
-					  FMC1_DENSITY_MODIFY_ENABLE);
+                                            FMC1_FOGFUNC_VERTEX |
+                                            FMC1_FOGINDEX_MODIFY_ENABLE |
+                                            FMC1_FOGINDEX_W |
+                                            FMC1_C1_C2_MODIFY_ENABLE |
+                                            FMC1_DENSITY_MODIFY_ENABLE);
       i915->state.Fog[I915_FOGREG_COLOR] = _3DSTATE_FOG_COLOR_CMD;
    }
 
-
    {
-      I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
-      /* color buffer offset/stride */
-      i915->state.Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      i915->state.Buffer[I915_DESTREG_CBUFADDR1] = 
-	 (BUF_3D_ID_COLOR_BACK | 
-	  BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
-	  BUF_3D_USE_FENCE);
-      /*i915->state.Buffer[I915_DESTREG_CBUFADDR2] is the offset */
-
-
-      /* depth/Z buffer offset/stride */
-      i915->state.Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      i915->state.Buffer[I915_DESTREG_DBUFADDR1] = 
-	 (BUF_3D_ID_DEPTH |
-	  BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
-	  BUF_3D_USE_FENCE);
-      i915->state.Buffer[I915_DESTREG_DBUFADDR2] = screen->depth.offset;
-
-
       i915->state.Buffer[I915_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
 
-      /* color/depth pixel format */
-      switch (screen->fbFormat) {
-      case DV_PF_555:
-      case DV_PF_565:
-	 i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					       DSTORG_VERT_BIAS(0x8) | /* .5 */
-					       LOD_PRECLAMP_OGL |
-					       TEX_DEFAULT_COLOR_OGL |
-					       DITHER_FULL_ALWAYS |
-					       screen->fbFormat |
-					       DEPTH_FRMT_16_FIXED);
-	 break;
-      case DV_PF_8888:
-	 i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					       DSTORG_VERT_BIAS(0x8) | /* .5 */
-					       LOD_PRECLAMP_OGL |
-					       TEX_DEFAULT_COLOR_OGL |
-					       screen->fbFormat |
-					       DEPTH_FRMT_24_FIXED_8_OTHER);
-	 break;
-      }
-
       /* scissor */
-      i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						DISABLE_SCISSOR_RECT);
+      i915->state.Buffer[I915_DESTREG_SENABLE] =
+         (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
       i915->state.Buffer[I915_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
       i915->state.Buffer[I915_DESTREG_SR1] = 0;
       i915->state.Buffer[I915_DESTREG_SR2] = 0;
    }
 
 
+#if 0
+   {
+      I915_STATECHANGE(i915, I915_UPLOAD_DEFAULTS);
+      i915->state.Default[I915_DEFREG_C0] = _3DSTATE_DEFAULT_DIFFUSE;
+      i915->state.Default[I915_DEFREG_C1] = 0;
+      i915->state.Default[I915_DEFREG_S0] = _3DSTATE_DEFAULT_SPECULAR;
+      i915->state.Default[I915_DEFREG_S1] = 0;
+      i915->state.Default[I915_DEFREG_Z0] = _3DSTATE_DEFAULT_Z;
+      i915->state.Default[I915_DEFREG_Z1] = 0;
+   }
+#endif
+
+
    /* These will be emitted every at the head of every buffer, unless
     * we get hardware contexts working.
     */
-   i915->state.active = (I915_UPLOAD_PROGRAM | 
-			 I915_UPLOAD_STIPPLE | 
-			 I915_UPLOAD_CTX | 
-			 I915_UPLOAD_BUFFERS | 
-			 I915_UPLOAD_INVARIENT);
+   i915->state.active = (I915_UPLOAD_PROGRAM |
+                         I915_UPLOAD_STIPPLE |
+                         I915_UPLOAD_CTX |
+                         I915_UPLOAD_BUFFERS | I915_UPLOAD_INVARIENT);
 }
 
-void i915InitStateFunctions( struct dd_function_table *functions )
+void
+i915InitStateFunctions(struct dd_function_table *functions)
 {
    functions->AlphaFunc = i915AlphaFunc;
    functions->BlendColor = i915BlendColor;
@@ -957,14 +967,15 @@ void i915InitStateFunctions( struct dd_function_table *functions )
 }
 
 
-void i915InitState( i915ContextPtr i915 )
+void
+i915InitState(struct i915_context *i915)
 {
    GLcontext *ctx = &i915->intel.ctx;
 
-   i915_init_packets( i915 );
+   i915_init_packets(i915);
 
    _mesa_init_driver_state(ctx);
 
-   memcpy( &i915->initial, &i915->state, sizeof(i915->state) );
+   memcpy(&i915->initial, &i915->state, sizeof(i915->state));
    i915->current = &i915->state;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_tex.c b/src/mesa/drivers/dri/i915/i915_tex.c
index d9609d31933..e38d8fe79d1 100644
--- a/src/mesa/drivers/dri/i915/i915_tex.c
+++ b/src/mesa/drivers/dri/i915/i915_tex.c
@@ -25,129 +25,43 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "texmem.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/mm.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
-
-#include "intel_ioctl.h"
+#include "texmem.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
 
 
-
-
-
-/**
- * Allocate space for and load the mesa images into the texture memory block.
- * This will happen before drawing with a new texture, or drawing with a
- * texture after it was swapped out or teximaged again.
- */
-
-intelTextureObjectPtr i915AllocTexObj( struct gl_texture_object *texObj )
-{
-   i915TextureObjectPtr t = CALLOC_STRUCT( i915_texture_object );
-   if ( !t ) 
-      return NULL;
-
-   texObj->DriverData = t;
-   t->intel.base.tObj = texObj;
-   t->intel.dirty = I915_UPLOAD_TEX_ALL;
-   make_empty_list( &t->intel.base );
-   return &t->intel;
-}
-
-
-static void i915TexParameter( GLcontext *ctx, GLenum target,
-			     struct gl_texture_object *tObj,
-			     GLenum pname, const GLfloat *params )
-{
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
- 
-   switch (pname) {
-   case GL_TEXTURE_MIN_FILTER:
-   case GL_TEXTURE_MAG_FILTER:
-   case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-   case GL_TEXTURE_WRAP_S:
-   case GL_TEXTURE_WRAP_T:
-   case GL_TEXTURE_WRAP_R:
-   case GL_TEXTURE_BORDER_COLOR:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   case GL_TEXTURE_COMPARE_MODE:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-   case GL_TEXTURE_COMPARE_FUNC:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   case GL_TEXTURE_BASE_LEVEL:
-   case GL_TEXTURE_MAX_LEVEL:
-   case GL_TEXTURE_MIN_LOD:
-   case GL_TEXTURE_MAX_LOD:
-      /* The i915 and its successors can do a lot of this without
-       * reloading the textures.  A project for someone?
-       */
-      intelFlush( ctx );
-      driSwapOutTextureObject( (driTextureObject *) t );
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   default:
-      return;
-   }
-}
-
-
-static void i915TexEnv( GLcontext *ctx, GLenum target, 
-			GLenum pname, const GLfloat *param )
+static void
+i915TexEnv(GLcontext * ctx, GLenum target,
+           GLenum pname, const GLfloat * param)
 {
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
-   GLuint unit = ctx->Texture.CurrentUnit;
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
    switch (pname) {
-   case GL_TEXTURE_ENV_COLOR: 	/* Should be a tracked param */
-   case GL_TEXTURE_ENV_MODE:
-   case GL_COMBINE_RGB:
-   case GL_COMBINE_ALPHA:
-   case GL_SOURCE0_RGB:
-   case GL_SOURCE1_RGB:
-   case GL_SOURCE2_RGB:
-   case GL_SOURCE0_ALPHA:
-   case GL_SOURCE1_ALPHA:
-   case GL_SOURCE2_ALPHA:
-   case GL_OPERAND0_RGB:
-   case GL_OPERAND1_RGB:
-   case GL_OPERAND2_RGB:
-   case GL_OPERAND0_ALPHA:
-   case GL_OPERAND1_ALPHA:
-   case GL_OPERAND2_ALPHA:
-   case GL_RGB_SCALE:
-   case GL_ALPHA_SCALE:
-      i915->tex_program.translated = 0; 
-      break;
-
-   case GL_TEXTURE_LOD_BIAS: {
-      int b = (int) ((*param) * 16.0);
-      if (b > 255) b = 255;
-      if (b < -256) b = -256;
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS2] &= ~SS2_LOD_BIAS_MASK;
-      i915->state.Tex[unit][I915_TEXREG_SS2] |= 
-	 ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK);
-      break;
-   }
+   case GL_TEXTURE_LOD_BIAS:{
+         GLuint unit = ctx->Texture.CurrentUnit;
+         GLint b = (int) ((*param) * 16.0);
+         if (b > 255)
+            b = 255;
+         if (b < -256)
+            b = -256;
+         I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
+         i915->lodbias_ss2[unit] =
+            ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK);
+         break;
+      }
 
    default:
       break;
@@ -155,33 +69,10 @@ static void i915TexEnv( GLcontext *ctx, GLenum target,
 }
 
 
-static void i915BindTexture( GLcontext *ctx, GLenum target,
-			    struct gl_texture_object *texObj )
-{
-   i915TextureObjectPtr tex;
-   
-   if (!texObj->DriverData)
-      i915AllocTexObj( texObj );
-   
-   tex = (i915TextureObjectPtr)texObj->DriverData;
-
-   if (tex->lastTarget != texObj->Target) {
-      tex->intel.dirty = I915_UPLOAD_TEX_ALL;
-      tex->lastTarget = texObj->Target;
-   }
-
-   /* Need this if image format changes between bound textures.
-    * Could try and shortcircuit by checking for differences in
-    * state between incoming and outgoing textures:
-    */
-   I915_CONTEXT(ctx)->tex_program.translated = 0; 
-}
-
-
-
-void i915InitTextureFuncs( struct dd_function_table *functions )
+void
+i915InitTextureFuncs(struct dd_function_table *functions)
 {
-   functions->BindTexture = i915BindTexture;
+/*
    functions->TexEnv = i915TexEnv;
-   functions->TexParameter = i915TexParameter;
+*/
 }
diff --git a/src/mesa/drivers/dri/i915/i915_tex_layout.c b/src/mesa/drivers/dri/i915/i915_tex_layout.c
new file mode 100644
index 00000000000..d44a2f47b37
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/i915_tex_layout.c
@@ -0,0 +1,477 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/** @file i915_tex_layout.c
+ * Code to layout images in a mipmap tree for i830M-GM915 and G945 and beyond.
+ */
+
+#include "intel_mipmap_tree.h"
+#include "intel_tex_layout.h"
+#include "main/macros.h"
+#include "intel_context.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+static GLint initial_offsets[6][2] = {
+   [FACE_POS_X] = {0, 0},
+   [FACE_POS_Y] = {1, 0},
+   [FACE_POS_Z] = {1, 1},
+   [FACE_NEG_X] = {0, 2},
+   [FACE_NEG_Y] = {1, 2},
+   [FACE_NEG_Z] = {1, 3},
+};
+
+
+static GLint step_offsets[6][2] = {
+   [FACE_POS_X] = {0, 2},
+   [FACE_POS_Y] = {-1, 2},
+   [FACE_POS_Z] = {-1, 1},
+   [FACE_NEG_X] = {0, 2},
+   [FACE_NEG_Y] = {-1, 2},
+   [FACE_NEG_Z] = {-1, 1},
+};
+
+/**
+ * Cube texture map layout for i830M-GM915.
+ *
+ * Hardware layout looks like:
+ *
+ * +-------+-------+
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |  +x   |  +y   |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * +---+---+-------+
+ * |   |   |       |
+ * | +x| +y|       |
+ * |   |   |       |
+ * |   |   |       |
+ * +-+-+---+  +z   |
+ * | | |   |       |
+ * +-+-+ +z|       |
+ *   | |   |       |
+ * +-+-+---+-------+
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |  -x   |  -y   |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * +---+---+-------+
+ * |   |   |       |
+ * | -x| -y|       |
+ * |   |   |       |
+ * |   |   |       |
+ * +-+-+---+  -z   |
+ * | | |   |       |
+ * +-+-+ -z|       |
+ *   | |   |       |
+ *   +-+---+-------+
+ *
+ */
+static void
+i915_miptree_layout_cube(struct intel_context *intel,
+			 struct intel_mipmap_tree * mt)
+{
+   const GLuint dim = mt->width0;
+   GLuint face;
+   GLuint lvlWidth = mt->width0, lvlHeight = mt->height0;
+   GLint level;
+
+   assert(lvlWidth == lvlHeight); /* cubemap images are square */
+
+   /* double pitch for cube layouts */
+   mt->pitch = intel_miptree_pitch_align (intel, mt, dim * 2);
+   mt->total_height = dim * 4;
+
+   for (level = mt->first_level; level <= mt->last_level; level++) {
+      intel_miptree_set_level_info(mt, level, 6,
+				   0, 0,
+				   /*OLD: mt->pitch, mt->total_height,*/
+				   lvlWidth, lvlHeight,
+				   1);
+      lvlWidth /= 2;
+      lvlHeight /= 2;
+   }
+
+   for (face = 0; face < 6; face++) {
+      GLuint x = initial_offsets[face][0] * dim;
+      GLuint y = initial_offsets[face][1] * dim;
+      GLuint d = dim;
+
+      for (level = mt->first_level; level <= mt->last_level; level++) {
+	 intel_miptree_set_image_offset(mt, level, face, x, y);
+
+	 if (d == 0)
+	    _mesa_printf("cube mipmap %d/%d (%d..%d) is 0x0\n",
+			 face, level, mt->first_level, mt->last_level);
+
+	 d >>= 1;
+	 x += step_offsets[face][0] * d;
+	 y += step_offsets[face][1] * d;
+      }
+   }
+}
+
+static void
+i915_miptree_layout_3d(struct intel_context *intel,
+		       struct intel_mipmap_tree * mt)
+{
+   GLuint width = mt->width0;
+   GLuint height = mt->height0;
+   GLuint depth = mt->depth0;
+   GLuint stack_height = 0;
+   GLint level;
+
+   /* Calculate the size of a single slice. */
+   mt->pitch = intel_miptree_pitch_align (intel, mt, mt->width0);
+
+   /* XXX: hardware expects/requires 9 levels at minimum. */
+   for (level = mt->first_level; level <= MAX2(8, mt->last_level); level++) {
+      intel_miptree_set_level_info(mt, level, depth, 0, mt->total_height,
+				   width, height, depth);
+
+      stack_height += MAX2(2, height);
+
+      width = minify(width);
+      height = minify(height);
+      depth = minify(depth);
+   }
+
+   /* Fixup depth image_offsets: */
+   depth = mt->depth0;
+   for (level = mt->first_level; level <= mt->last_level; level++) {
+      GLuint i;
+      for (i = 0; i < depth; i++) {
+	 intel_miptree_set_image_offset(mt, level, i,
+					0, i * stack_height);
+      }
+
+      depth = minify(depth);
+   }
+
+   /* Multiply slice size by texture depth for total size.  It's
+    * remarkable how wasteful of memory the i915 texture layouts
+    * are.  They are largely fixed in the i945.
+    */
+   mt->total_height = stack_height * mt->depth0;
+}
+
+static void
+i915_miptree_layout_2d(struct intel_context *intel,
+		       struct intel_mipmap_tree * mt)
+{
+   GLuint width = mt->width0;
+   GLuint height = mt->height0;
+   GLuint img_height;
+   GLint level;
+
+   mt->pitch = intel_miptree_pitch_align (intel, mt, mt->width0);
+   mt->total_height = 0;
+
+   for (level = mt->first_level; level <= mt->last_level; level++) {
+      intel_miptree_set_level_info(mt, level, 1,
+				   0, mt->total_height,
+				   width, height, 1);
+
+      if (mt->compressed)
+	 img_height = MAX2(1, height / 4);
+      else
+	 img_height = (MAX2(2, height) + 1) & ~1;
+
+      mt->total_height += img_height;
+
+      width = minify(width);
+      height = minify(height);
+   }
+}
+
+GLboolean
+i915_miptree_layout(struct intel_context *intel, struct intel_mipmap_tree * mt)
+{
+   switch (mt->target) {
+   case GL_TEXTURE_CUBE_MAP:
+      i915_miptree_layout_cube(intel, mt);
+      break;
+   case GL_TEXTURE_3D:
+      i915_miptree_layout_3d(intel, mt);
+      break;
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_RECTANGLE_ARB:
+      i915_miptree_layout_2d(intel, mt);
+      break;
+   default:
+      _mesa_problem(NULL, "Unexpected tex target in i915_miptree_layout()");
+      break;
+   }
+
+   DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__,
+       mt->pitch,
+       mt->total_height, mt->cpp, mt->pitch * mt->total_height * mt->cpp);
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Cube texture map layout for GM945 and later.
+ *
+ * The hardware layout looks like the 830-915 layout, except for the small
+ * sizes.  A zoomed in view of the layout for 945 is:
+ *
+ * +-------+-------+
+ * |  8x8  |  8x8  |
+ * |       |       |
+ * |       |       |
+ * |  +x   |  +y   |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * +---+---+-------+
+ * |4x4|   |  8x8  |
+ * | +x|   |       |
+ * |   |   |       |
+ * |   |   |       |
+ * +---+   |  +z   |
+ * |4x4|   |       |
+ * | +y|   |       |
+ * |   |   |       |
+ * +---+   +-------+
+ *
+ * ...
+ *
+ * +-------+-------+
+ * |  8x8  |  8x8  |
+ * |       |       |
+ * |       |       |
+ * |  -x   |  -y   |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * |       |       |
+ * +---+---+-------+
+ * |4x4|   |  8x8  |
+ * | -x|   |       |
+ * |   |   |       |
+ * |   |   |       |
+ * +---+   |  -z   |
+ * |4x4|   |       |
+ * | -y|   |       |
+ * |   |   |       |
+ * +---+   +---+---+---+---+---+---+---+---+---+
+ * |4x4|   |4x4|   |2x2|   |2x2|   |2x2|   |2x2|
+ * | +z|   | -z|   | +x|   | +y|   | +z|   | -x| ...
+ * |   |   |   |   |   |   |   |   |   |   |   |
+ * +---+   +---+   +---+   +---+   +---+   +---+
+ *
+ * The bottom row continues with the remaining 2x2 then the 1x1 mip contents
+ * in order, with each of them aligned to a 4x4 block boundary.  Thus, for
+ * 32x32 cube maps and smaller, the bottom row layout is going to dictate the
+ * pitch of the tree.  For a tree with 4x4 images, the pitch is at least
+ * 14 * 8 = 112 texels, for 2x2 it is at least 12 * 8 texels, and for 1x1
+ * it is 6 * 8 texels.
+ */
+
+static void
+i945_miptree_layout_cube(struct intel_context *intel,
+			 struct intel_mipmap_tree * mt)
+{
+   const GLuint dim = mt->width0;
+   GLuint face;
+   GLuint lvlWidth = mt->width0, lvlHeight = mt->height0;
+   GLint level;
+
+   assert(lvlWidth == lvlHeight); /* cubemap images are square */
+
+   /* Depending on the size of the largest images, pitch can be
+    * determined either by the old-style packing of cubemap faces,
+    * or the final row of 4x4, 2x2 and 1x1 faces below this.
+    */
+   if (dim > 32)
+      mt->pitch = intel_miptree_pitch_align (intel, mt, dim * 2);
+   else
+      mt->pitch = intel_miptree_pitch_align (intel, mt, 14 * 8);
+
+   if (dim >= 4)
+      mt->total_height = dim * 4 + 4;
+   else
+      mt->total_height = 4;
+
+   /* Set all the levels to effectively occupy the whole rectangular region. */
+   for (level = mt->first_level; level <= mt->last_level; level++) {
+      intel_miptree_set_level_info(mt, level, 6,
+				   0, 0,
+				   lvlWidth, lvlHeight, 1);
+      lvlWidth /= 2;
+      lvlHeight /= 2;
+   }
+
+   for (face = 0; face < 6; face++) {
+      GLuint x = initial_offsets[face][0] * dim;
+      GLuint y = initial_offsets[face][1] * dim;
+      GLuint d = dim;
+
+      if (dim == 4 && face >= 4) {
+	 y = mt->total_height - 4;
+	 x = (face - 4) * 8;
+      } else if (dim < 4 && (face > 0 || mt->first_level > 0)) {
+	 y = mt->total_height - 4;
+	 x = face * 8;
+      }
+
+      for (level = mt->first_level; level <= mt->last_level; level++) {
+	 intel_miptree_set_image_offset(mt, level, face, x, y);
+
+	 d >>= 1;
+
+	 switch (d) {
+	 case 4:
+	    switch (face) {
+	    case FACE_POS_X:
+	    case FACE_NEG_X:
+	       x += step_offsets[face][0] * d;
+	       y += step_offsets[face][1] * d;
+	       break;
+	    case FACE_POS_Y:
+	    case FACE_NEG_Y:
+	       y += 12;
+	       x -= 8;
+	       break;
+	    case FACE_POS_Z:
+	    case FACE_NEG_Z:
+	       y = mt->total_height - 4;
+	       x = (face - 4) * 8;
+	       break;
+	    }
+
+	 case 2:
+	    y = mt->total_height - 4;
+	    x = 16 + face * 8;
+	    break;
+
+	 case 1:
+	    x += 48;
+	    break;
+
+	 default:
+	    x += step_offsets[face][0] * d;
+	    y += step_offsets[face][1] * d;
+	    break;
+	 }
+      }
+   }
+}
+
+static void
+i945_miptree_layout_3d(struct intel_context *intel,
+		       struct intel_mipmap_tree * mt)
+{
+   GLuint width = mt->width0;
+   GLuint height = mt->height0;
+   GLuint depth = mt->depth0;
+   GLuint pack_x_pitch, pack_x_nr;
+   GLuint pack_y_pitch;
+   GLuint level;
+
+   mt->pitch = intel_miptree_pitch_align (intel, mt, mt->width0);
+   mt->total_height = 0;
+
+   pack_y_pitch = MAX2(mt->height0, 2);
+   pack_x_pitch = mt->pitch;
+   pack_x_nr = 1;
+
+   for (level = mt->first_level; level <= mt->last_level; level++) {
+      GLint x = 0;
+      GLint y = 0;
+      GLint q, j;
+
+      intel_miptree_set_level_info(mt, level, depth,
+				   0, mt->total_height,
+				   width, height, depth);
+
+      for (q = 0; q < depth;) {
+	 for (j = 0; j < pack_x_nr && q < depth; j++, q++) {
+	    intel_miptree_set_image_offset(mt, level, q, x, y);
+	    x += pack_x_pitch;
+	 }
+
+	 x = 0;
+	 y += pack_y_pitch;
+      }
+
+      mt->total_height += y;
+
+      if (pack_x_pitch > 4) {
+	 pack_x_pitch >>= 1;
+	 pack_x_nr <<= 1;
+	 assert(pack_x_pitch * pack_x_nr <= mt->pitch);
+      }
+
+      if (pack_y_pitch > 2) {
+	 pack_y_pitch >>= 1;
+      }
+
+      width = minify(width);
+      height = minify(height);
+      depth = minify(depth);
+   }
+}
+
+GLboolean
+i945_miptree_layout(struct intel_context *intel, struct intel_mipmap_tree * mt)
+{
+   switch (mt->target) {
+   case GL_TEXTURE_CUBE_MAP:
+      i945_miptree_layout_cube(intel, mt);
+      break;
+   case GL_TEXTURE_3D:
+      i945_miptree_layout_3d(intel, mt);
+      break;
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_RECTANGLE_ARB:
+      i945_miptree_layout_2d(intel, mt);
+      break;
+   default:
+      _mesa_problem(NULL, "Unexpected tex target in i945_miptree_layout()");
+      break;
+   }
+
+   DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__,
+       mt->pitch,
+       mt->total_height, mt->cpp, mt->pitch * mt->total_height * mt->cpp);
+
+   return GL_TRUE;
+}
diff --git a/src/mesa/drivers/dri/i915/i915_texprog.c b/src/mesa/drivers/dri/i915/i915_texprog.c
deleted file mode 100644
index f6a8b0205a6..00000000000
--- a/src/mesa/drivers/dri/i915/i915_texprog.c
+++ /dev/null
@@ -1,676 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include <strings.h>
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "tnl/t_context.h"
-#include "intel_batchbuffer.h"
-
-#include "i915_reg.h"
-#include "i915_context.h"
-#include "i915_program.h"
-
-static GLuint translate_tex_src_bit( struct i915_fragment_program *p,
-				     GLubyte bit )
-{
-   switch (bit) {
-   case TEXTURE_1D_BIT:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_2D_BIT:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_RECT_BIT: return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_3D_BIT:   return D0_SAMPLE_TYPE_VOLUME;
-   case TEXTURE_CUBE_BIT: return D0_SAMPLE_TYPE_CUBE;
-   default: i915_program_error(p, "TexSrcBit"); return 0;
-   }
-}
-
-static GLuint get_source( struct i915_fragment_program *p, 
-			  GLenum src, GLuint unit )
-{
-   switch (src) {
-   case GL_TEXTURE: 
-      if (p->src_texture == UREG_BAD) {
-
-	 /* TODO: Use D0_CHANNEL_XY where possible.
-	  */
-	 GLuint dim = translate_tex_src_bit( p, p->ctx->Texture.Unit[unit]._ReallyEnabled);
-	 GLuint sampler = i915_emit_decl(p, REG_TYPE_S, unit, dim);
-	 GLuint texcoord = i915_emit_decl(p, REG_TYPE_T, unit, D0_CHANNEL_ALL);
-	 GLuint tmp = i915_get_temp( p );
-	 GLuint op = T0_TEXLD;
-
-	 if (p->VB->TexCoordPtr[unit]->size == 4)
-	    op = T0_TEXLDP;
-
-	 p->src_texture = i915_emit_texld( p, tmp, A0_DEST_CHANNEL_ALL, 
-					  sampler, texcoord, op );
-      }
-
-      return p->src_texture;
-
-      /* Crossbar: */
-   case GL_TEXTURE0:
-   case GL_TEXTURE1:
-   case GL_TEXTURE2:
-   case GL_TEXTURE3:
-   case GL_TEXTURE4:
-   case GL_TEXTURE5:
-   case GL_TEXTURE6:
-   case GL_TEXTURE7: {
-      return UREG_BAD;
-   }
-
-   case GL_CONSTANT:
-      return i915_emit_const4fv( p, p->ctx->Texture.Unit[unit].EnvColor );
-   case GL_PRIMARY_COLOR:
-      return i915_emit_decl(p, REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL);
-   case GL_PREVIOUS:
-   default: 
-      i915_emit_decl(p, 
-		GET_UREG_TYPE(p->src_previous),
-		GET_UREG_NR(p->src_previous), D0_CHANNEL_ALL); 
-      return p->src_previous;
-   }
-}
-			
-
-static GLuint emit_combine_source( struct i915_fragment_program *p, 
-				   GLuint mask,
-				   GLuint unit,
-				   GLenum source, 
-				   GLenum operand )
-{
-   GLuint arg, src;
-
-   src = get_source(p, source, unit);
-
-   switch (operand) {
-   case GL_ONE_MINUS_SRC_COLOR: 
-      /* Get unused tmp,
-       * Emit tmp = 1.0 + arg.-x-y-z-w
-       */
-      arg = i915_get_temp( p );
-      return i915_emit_arith( p, A0_ADD, arg, mask, 0,
-		  swizzle(src, ONE, ONE, ONE, ONE ),
-		  negate(src, 1,1,1,1), 0);
-
-   case GL_SRC_ALPHA: 
-      if (mask == A0_DEST_CHANNEL_W)
-	 return src;
-      else
-	 return swizzle( src, W, W, W, W );
-   case GL_ONE_MINUS_SRC_ALPHA: 
-      /* Get unused tmp,
-       * Emit tmp = 1.0 + arg.-w-w-w-w
-       */
-      arg = i915_get_temp( p );
-      return i915_emit_arith( p, A0_ADD, arg, mask, 0,
-			 swizzle(src, ONE, ONE, ONE, ONE ),
-			 negate( swizzle(src,W,W,W,W), 1,1,1,1), 0);
-   case GL_SRC_COLOR: 
-   default:
-      return src;
-   }
-}
-
-
-
-static int nr_args( GLenum mode )
-{
-   switch (mode) {
-   case GL_REPLACE: return 1; 
-   case GL_MODULATE: return 2;
-   case GL_ADD: return 2;
-   case GL_ADD_SIGNED: return 2;
-   case GL_INTERPOLATE:	return 3;
-   case GL_SUBTRACT: return 2;
-   case GL_DOT3_RGB_EXT: return 2;
-   case GL_DOT3_RGBA_EXT: return 2;
-   case GL_DOT3_RGB: return 2;
-   case GL_DOT3_RGBA: return 2;
-   default: return 0;
-   }
-}
-
-
-static GLboolean args_match( struct gl_texture_unit *texUnit )
-{
-   int i, nr = nr_args(texUnit->Combine.ModeRGB);
-
-   for (i = 0 ; i < nr ; i++) {
-      if (texUnit->Combine.SourceA[i] != texUnit->Combine.SourceRGB[i]) 
-	 return GL_FALSE;
-
-      switch(texUnit->Combine.OperandA[i]) {
-      case GL_SRC_ALPHA: 
-	 switch(texUnit->Combine.OperandRGB[i]) {
-	 case GL_SRC_COLOR: 
-	 case GL_SRC_ALPHA: 
-	    break;
-	 default:
-	    return GL_FALSE;
-	 }
-	 break;
-      case GL_ONE_MINUS_SRC_ALPHA: 
-	 switch(texUnit->Combine.OperandRGB[i]) {
-	 case GL_ONE_MINUS_SRC_COLOR: 
-	 case GL_ONE_MINUS_SRC_ALPHA: 
-	    break;
-	 default:
-	    return GL_FALSE;
-	 }
-	 break;
-      default: 
-	 return GL_FALSE;	/* impossible */
-      }
-   }
-
-   return GL_TRUE;
-}
-
-
-static GLuint emit_combine( struct i915_fragment_program *p,
-			    GLuint dest,
-			    GLuint mask,
-			    GLuint saturate,
-			    GLuint unit,
-			    GLenum mode,
-			    const GLenum *source,
-			    const GLenum *operand)
-{
-   int tmp, src[3], nr = nr_args(mode);
-   int i;
-
-   for (i = 0; i < nr; i++)
-      src[i] = emit_combine_source( p, mask, unit, source[i], operand[i] );
-
-   switch (mode) {
-   case GL_REPLACE: 
-      if (mask == A0_DEST_CHANNEL_ALL && !saturate)
-	 return src[0];
-      else
-	 return i915_emit_arith( p, A0_MOV, dest, mask, saturate, src[0], 0, 0 );
-   case GL_MODULATE: 
-      return i915_emit_arith( p, A0_MUL, dest, mask, saturate,
-			     src[0], src[1], 0 );
-   case GL_ADD: 
-      return i915_emit_arith( p, A0_ADD, dest, mask, saturate, 
-			     src[0], src[1], 0 );
-   case GL_ADD_SIGNED:
-      /* tmp = arg0 + arg1
-       * result = tmp + -.5
-       */
-      tmp = i915_emit_const1f(p, .5);
-      tmp = negate(swizzle(tmp,X,X,X,X),1,1,1,1);
-      i915_emit_arith( p, A0_ADD, dest, mask, 0, src[0], src[1], 0 );
-      i915_emit_arith( p, A0_ADD, dest, mask, saturate, dest, tmp, 0 );
-      return dest;
-   case GL_INTERPOLATE:		/* TWO INSTRUCTIONS */
-      /* Arg0 * (Arg2) + Arg1 * (1-Arg2)
-       *
-       * Arg0*Arg2 + Arg1 - Arg1Arg2 
-       *
-       * tmp = Arg0*Arg2 + Arg1, 
-       * result = (-Arg1)Arg2 + tmp 
-       */
-      tmp = i915_get_temp( p );
-      i915_emit_arith( p, A0_MAD, tmp, mask, 0, src[0], src[2], src[1] );
-      i915_emit_arith( p, A0_MAD, dest, mask, saturate, 
-		      negate(src[1], 1,1,1,1), src[2], tmp );
-      return dest;
-   case GL_SUBTRACT: 
-      /* negate src[1] */
-      return i915_emit_arith( p, A0_ADD, dest, mask, saturate, src[0],
-			 negate(src[1],1,1,1,1), 0 );
-
-   case GL_DOT3_RGBA:
-   case GL_DOT3_RGBA_EXT: 
-   case GL_DOT3_RGB_EXT:
-   case GL_DOT3_RGB: {
-      GLuint tmp0 = i915_get_temp( p );
-      GLuint tmp1 = i915_get_temp( p );
-      GLuint neg1 = negate(swizzle(i915_emit_const1f(p, 1),X,X,X,X), 1,1,1,1);
-      GLuint two = swizzle(i915_emit_const1f(p, 2),X,X,X,X);
-      i915_emit_arith( p, A0_MAD, tmp0, A0_DEST_CHANNEL_ALL, 0, 
-		      two, src[0], neg1);
-      if (src[0] == src[1])
-	 tmp1 = tmp0;
-      else
-	 i915_emit_arith( p, A0_MAD, tmp1, A0_DEST_CHANNEL_ALL, 0, 
-			 two, src[1], neg1);
-      i915_emit_arith( p, A0_DP3, dest, mask, saturate, tmp0, tmp1, 0);
-      return dest;
-   }
-
-   default: 
-      return src[0];
-   }
-}
-
-static GLuint get_dest( struct i915_fragment_program *p, int unit )
-{
-   if (p->ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)
-      return i915_get_temp( p );
-   else if (unit != p->last_tex_stage)
-      return i915_get_temp( p );
-   else
-      return UREG(REG_TYPE_OC, 0);
-}
-      
-
-
-static GLuint emit_texenv( struct i915_fragment_program *p, int unit )
-{
-   struct gl_texture_unit *texUnit = &p->ctx->Texture.Unit[unit];
-   GLenum envMode = texUnit->EnvMode;
-   struct gl_texture_object *tObj = texUnit->_Current;
-   GLenum format = tObj->Image[0][tObj->BaseLevel]->_BaseFormat;
-   GLuint saturate = unit < p->last_tex_stage ? A0_DEST_SATURATE : 0;
-
-   switch(envMode) {
-   case GL_BLEND: {
-      const int cf = get_source(p, GL_PREVIOUS, unit);
-      const int cc = get_source(p, GL_CONSTANT, unit);
-      const int cs = get_source(p, GL_TEXTURE, unit);
-      const int out = get_dest(p, unit);
-
-      if (format == GL_INTENSITY) {
-	 /* cv = cf(1 - cs) + cc.cs
-	  * cv = cf - cf.cs + cc.cs
-	  */
-	 /* u[2] = MAD( -cf * cs + cf )
-	  * cv   = MAD( cc * cs + u[2] )
-	  */
-	 
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0, 
-			 negate(cf,1,1,1,1), cs, cf );
-
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cc, cs, out );
-
-	 return out;
-      } else {
-	 /* cv = cf(1 - cs) + cc.cs
-	  * cv = cf - cf.cs + cc.cs
-	  * av =      af.as
-	  */
-	 /* u[2] = MAD( cf.-x-y-zw * cs.xyzw + cf.xyz0 )
-	  * oC   = MAD( cc.xyz0 * cs.xyz0 + u[2].xyzw )
-	  */
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0,
-			 negate(cf,1,1,1,0),  
-			 cs,
-			 swizzle(cf,X,Y,Z,ZERO) );
-
-
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cc,X,Y,Z,ZERO),  
-			 swizzle(cs,X,Y,Z,ZERO),
-			 out );
-
-	 return out;
-      }
-   }
-
-   case GL_DECAL: {
-      if (format == GL_RGB ||
-	  format == GL_RGBA) {
-	 int cf = get_source( p, GL_PREVIOUS, unit );
-	 int cs = get_source( p, GL_TEXTURE, unit );
-	 int out = get_dest(p, unit);
-	 
-	 /* cv = cf(1-as) + cs.as
-	  * cv = cf.(-as) + cf + cs.as
-	  * av = af
-	  */ 
-	 
-	 /* u[2] = mad( cf.xyzw * cs.-w-w-w1 + cf.xyz0 )
-	  * oc = mad( cs.xyz0 * cs.www0 + u[2].xyzw )
-	  */
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0,
-			 cf,  
-			 negate(swizzle(cs,W,W,W,ONE),1,1,1,0),
-			 swizzle(cf,X,Y,Z,ZERO) );
-	 
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cs,X,Y,Z,ZERO),  
-			 swizzle(cs,W,W,W,ZERO),
-			 out );
-	 return out;
-      }
-      else {
-	 return get_source( p, GL_PREVIOUS, unit );
-      }
-   }
-
-   case GL_REPLACE: {
-      const int cs = get_source( p, GL_TEXTURE, unit );	/* saturated */
-      switch (format) {
-      case GL_ALPHA: {
-	 const int cf = get_source( p, GL_PREVIOUS, unit ); /* saturated */
-	 i915_emit_arith( p, A0_MOV, cs, A0_DEST_CHANNEL_XYZ, 0, cf, 0, 0 );
-	 return cs;
-      }
-      case GL_RGB:
-      case GL_LUMINANCE: {
-	 const int cf = get_source( p, GL_PREVIOUS, unit ); /* saturated */
-	 i915_emit_arith( p, A0_MOV, cs, A0_DEST_CHANNEL_W, 0, cf, 0, 0 );
-	 return cs;
-      }
-      default:
-	 return cs;
-      }
-   }
-
-   case GL_MODULATE: {
-      const int cf = get_source( p, GL_PREVIOUS, unit );
-      const int cs = get_source( p, GL_TEXTURE, unit );
-      const int out = get_dest(p, unit);
-      switch (format) {
-      case GL_ALPHA: 
-	 i915_emit_arith( p, A0_MUL, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cs, ONE, ONE, ONE, W), cf, 0 );
-	 break;
-      default:
-	 i915_emit_arith( p, A0_MUL, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cs, cf, 0 );
-	 break;
-      }
-      return out;
-   }
-   case GL_ADD: {
-      int cf = get_source( p, GL_PREVIOUS, unit );
-      int cs = get_source( p, GL_TEXTURE, unit );
-      const int out = get_dest( p, unit );
-
-      if (format == GL_INTENSITY) {
-	 /* output-color.rgba = add( incoming, u[1] )
-	  */
-	 i915_emit_arith( p, A0_ADD, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cs, cf, 0 );
-	 return out;
-      }
-      else {
-	 /* cv.xyz = cf.xyz + cs.xyz
-	  * cv.w   = cf.w * cs.w
-	  *
-	  * cv.xyzw = MAD( cf.111w * cs.xyzw + cf.xyz0 )
-	  */
- 	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cf,ONE,ONE,ONE,W), 
-			 cs,  
-			 swizzle(cf,X,Y,Z,ZERO) ); 
-	 return out;
-      }
-      break;
-   }
-   case GL_COMBINE: {
-      GLuint rgb_shift, alpha_shift, out, shift;
-      GLuint dest = get_dest(p, unit);
-
-      /* The EXT version of the DOT3 extension does not support the
-       * scale factor, but the ARB version (and the version in OpenGL
-       * 1.3) does.
-       */
-      switch (texUnit->Combine.ModeRGB) {
-      case GL_DOT3_RGB_EXT:
-	 alpha_shift = texUnit->Combine.ScaleShiftA;
-	 rgb_shift = 0;
-	 break;
-
-      case GL_DOT3_RGBA_EXT:
-	 alpha_shift = 0;
-	 rgb_shift = 0;
-	 break;
-
-      default:
-	 rgb_shift = texUnit->Combine.ScaleShiftRGB;
-	 alpha_shift = texUnit->Combine.ScaleShiftA;
-	 break;
-      }
-
-
-      /* Emit the RGB and A combine ops
-       */
-      if (texUnit->Combine.ModeRGB == texUnit->Combine.ModeA && 
-	  args_match( texUnit )) {
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_ALL, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-      }
-      else if (texUnit->Combine.ModeRGB == GL_DOT3_RGBA_EXT ||
-	       texUnit->Combine.ModeRGB == GL_DOT3_RGBA) {
-
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_ALL, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-      }
-      else {
-	 /* Need to do something to stop from re-emitting identical
-	  * argument calculations here:
-	  */
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_XYZ, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_W, saturate,
-			     unit,
-			     texUnit->Combine.ModeA,
-			     texUnit->Combine.SourceA,
-			     texUnit->Combine.OperandA );
-      }
-
-      /* Deal with the final shift:
-       */
-      if (alpha_shift || rgb_shift) {
-	 if (rgb_shift == alpha_shift) {
-	    shift = i915_emit_const1f(p, 1<<rgb_shift);
-	    shift = swizzle(shift,X,X,X,X);
-	 }
-	 else {
-	    shift = i915_emit_const2f(p, 1<<rgb_shift, 1<<alpha_shift);
-	    shift = swizzle(shift,X,X,X,Y);
-	 }
-	 return i915_emit_arith( p, A0_MUL, dest, A0_DEST_CHANNEL_ALL, 
-				saturate, out, shift, 0 );
-      }
-
-      return out;
-   }
-
-   default:
-      return get_source(p, GL_PREVIOUS, 0);
-   }
-}
-
-static void emit_program_fini( struct i915_fragment_program *p )
-{
-   int cf = get_source( p, GL_PREVIOUS, 0 );
-   int out = UREG( REG_TYPE_OC, 0 );
-
-   if (p->ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) {
-      /* Emit specular add.
-       */
-      GLuint s = i915_emit_decl(p, REG_TYPE_T, T_SPECULAR, D0_CHANNEL_ALL);
-      i915_emit_arith( p, A0_ADD, out, A0_DEST_CHANNEL_ALL, 0, cf, 
-		  swizzle(s, X,Y,Z,ZERO), 0 );
-   }
-   else if (cf != out) {
-      /* Will wind up in here if no texture enabled or a couple of
-       * other scenarios (GL_REPLACE for instance).
-       */
-      i915_emit_arith( p, A0_MOV, out, A0_DEST_CHANNEL_ALL, 0, cf, 0, 0 );
-   }
-}
-
-
-static void i915EmitTextureProgram( i915ContextPtr i915 )
-{
-   GLcontext *ctx = &i915->intel.ctx;
-   struct i915_fragment_program *p = &i915->tex_program;
-   GLuint unit;
-
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
-
-   i915_init_program( i915, p );
-
-   if (ctx->Texture._EnabledUnits) {
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-	    p->last_tex_stage = unit;
-	 }
-
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-	    p->src_previous = emit_texenv( p, unit );
-	    p->src_texture = UREG_BAD;
-	    p->temp_flag = 0xffff000;
-	    p->temp_flag |= 1 << GET_UREG_NR(p->src_previous);
-	 }
-   }
-
-   emit_program_fini( p );
-
-   i915_fini_program( p );
-   i915_upload_program( i915, p );
-
-   p->translated = 1;
-}
-
-
-void i915ValidateTextureProgram( i915ContextPtr i915 )
-{
-   intelContextPtr intel = &i915->intel;
-   GLcontext *ctx = &intel->ctx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-   DECLARE_RENDERINPUTS(index_bitset);
-   int i, offset;
-   GLuint s4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_VFMT_MASK;
-   GLuint s2 = S2_TEXCOORD_NONE;
-
-   RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
-
-   /* Important:
-    */
-   VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
-   intel->vertex_attr_count = 0;
-   intel->coloroffset = 0;
-   intel->specoffset = 0;
-   offset = 0;
-
-   if (i915->current_program) {
-      i915->current_program->on_hardware = 0;
-      i915->current_program->params_uptodate = 0;
-   }
-
-   if (i915->vertex_fog == I915_FOG_PIXEL) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
-      RENDERINPUTS_CLEAR( index_bitset, _TNL_ATTRIB_FOG );
-   }
-   else if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
-   }
-   else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12 );
-   }
-
-   /* How undefined is undefined? */
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POINTSIZE )) {
-      EMIT_ATTR( _TNL_ATTRIB_POINTSIZE, EMIT_1F, S4_VFMT_POINT_WIDTH, 4 );
-   }
-      
-   intel->coloroffset = offset / 4;
-   EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4 );
-            
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 ) ||
-       RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG )) {
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 )) {
-	 intel->specoffset = offset / 4;
-	 EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3 );
-      } else 
-	 EMIT_PAD( 3 );
-      
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG ))
-	 EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1 );
-      else
-	 EMIT_PAD( 1 );
-   }
-
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      for (i = 0; i < 8; i++) {
-	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    int sz = VB->TexCoordPtr[i]->size;
-	    
-	    s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	    s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
-
-	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_SZ(sz), 0, sz * 4 );
-	 }
-      }
-   }
-
-   /* Only need to change the vertex emit code if there has been a
-    * statechange to a new hardware vertex format:
-    */
-   if (s2 != i915->state.Ctx[I915_CTXREG_LIS2] ||
-       s4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-    
-      I915_STATECHANGE( i915, I915_UPLOAD_CTX );
-
-      i915->tex_program.translated = 0;
-
-      /* Must do this *after* statechange, so as not to affect
-       * buffered vertices reliant on the old state:
-       */
-      intel->vertex_size = _tnl_install_attrs( ctx, 
-					       intel->vertex_attrs, 
-					       intel->vertex_attr_count,
-					       intel->ViewportMatrix.m, 0 ); 
-
-      intel->vertex_size >>= 2;
-
-      i915->state.Ctx[I915_CTXREG_LIS2] = s2;
-      i915->state.Ctx[I915_CTXREG_LIS4] = s4;
-
-      assert(intel->vtbl.check_vertex_size( intel, intel->vertex_size ));
-   }
-
-   if (!i915->tex_program.translated ||
-       i915->last_ReallyEnabled != ctx->Texture._EnabledUnits) {
-      i915EmitTextureProgram( i915 );      
-      i915->last_ReallyEnabled = ctx->Texture._EnabledUnits;
-   }
-}
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index a19d4b65840..d53e2cbd5aa 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -25,902 +25,368 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texformat.h"
-#include "texstore.h"
-
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_ioctl.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/texformat.h"
+
+#include "intel_mipmap_tree.h"
 #include "intel_tex.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-static GLint initial_offsets[6][2] = { {0,0},
-				       {0,2},
-				       {1,0},
-				       {1,2},
-				       {1,1},
-				       {1,3} };
-
-
-static GLint step_offsets[6][2] = { {0,2},
-				    {0,2},
-				    {-1,2},
-				    {-1,2},
-				    {-1,1},
-				    {-1,1} };
-
-
-#define I915_TEX_UNIT_ENABLED(unit)		(1<<unit)
-
-static void i915LayoutTextureImages( i915ContextPtr i915,
-				     struct gl_texture_object *tObj )
-{
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   GLint firstLevel, lastLevel, numLevels;
-   GLint i, total_height, pitch;
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
-
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
-
-
-
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      pitch = dim * t->intel.texelBytes;
-      pitch *= 2;		/* double pitch for cube layouts */
-      pitch = (pitch + 3) & ~3;
-      
-      total_height = dim * 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    if (!t->intel.image[face][i].image) {
-	       fprintf(stderr, "no image %d %d\n", face, i);
-	       break;		/* can't happen */
-	    }
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    x += step_offsets[face][0] * d;
-	    y += step_offsets[face][1] * d;
-	 }
-      }
-      break;
-   }
-   case GL_TEXTURE_3D: {
-      GLuint virtual_height;
-      GLuint tmp_numLevels = numLevels;
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      /* Calculate the size of a single slice.  Hardware demands a
-       * minimum of 8 mipmaps, some of which might ultimately not be
-       * used:
-       */
-      if (tmp_numLevels < 9)
-	 tmp_numLevels = 9;
-
-      virtual_height = tObj->Image[0][firstLevel]->Height;
-
-      for ( total_height = i = 0 ; i < tmp_numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (t->intel.image[0][i].image) {
-	    t->intel.image[0][i].offset = total_height * pitch;
-	    t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 }
-
-	 total_height += MAX2(2, virtual_height);
-	 virtual_height >>= 1;
-      }
-
-      t->intel.depth_pitch = total_height * pitch;
-
-      /* Multiply slice size by texture depth for total size.  It's
-       * remarkable how wasteful of memory all the i8x0 texture
-       * layouts are.
-       */
-      total_height *= t->intel.image[0][0].image->Depth;
-      break;
-   }
-   default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      for ( total_height = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 if (t->intel.image[0][i].image->IsCompressed) {
-	    total_height += (t->intel.image[0][i].image->Height + 3) / 4;
-	 }
-	 else
-	   total_height += MAX2(2, t->intel.image[0][i].image->Height);
-      }
-      break;
-   }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = total_height*pitch;
-   t->intel.max_level = numLevels-1;
-}
-
-
-static void i945LayoutTextureImages( i915ContextPtr i915,
-				    struct gl_texture_object *tObj )
-{
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   GLint firstLevel, lastLevel, numLevels;
-   GLint i, total_height, pitch, sz, max_offset = 0, offset;
-
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
-
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
-
-
-
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      /* Depending on the size of the largest images, pitch can be
-       * determined either by the old-style packing of cubemap faces,
-       * or the final row of 4x4, 2x2 and 1x1 faces below this. 
-       */
-      if (dim > 32) {
-	 pitch = dim * t->intel.texelBytes;
-	 pitch *= 2;		/* double pitch for cube layouts */
-	 pitch = (pitch + 3) & ~3;
-      }
-      else {
-	 pitch = 14 * 8 * t->intel.texelBytes; /* determined by row of
-						* little maps at
-						* bottom */
-      }
-      
-      total_height = dim * 4 + 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 if (dim == 4 && face >= 4) {
-	    y = total_height - 4;
-	    x = (face - 4) * 8;
-	 }
-	 else if (dim < 4) {
-	    y = total_height - 4;
-	    x = face * 8;
-	 }
-
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-
-
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    assert(t->intel.image[face][i].image);
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    
-	    switch (d) {
-	    case 4:
-	       switch (face) {
-	       case FACE_POS_X:
-	       case FACE_NEG_X:
-		  x += step_offsets[face][0] * d;
-		  y += step_offsets[face][1] * d;
-		  break;
-	       case FACE_POS_Y:
-	       case FACE_NEG_Y:
-		  y += 12;
-		  x -= 8;
-		  break;
-	       case FACE_POS_Z:
-	       case FACE_NEG_Z:
-		  y = total_height - 4;
-		  x = (face - 4) * 8;
-		  break;
-	       }
-
-	    case 2:
-	       y = total_height - 4;
-	       x = 16 + face * 8;
-	       break;
-
-	    case 1:
-	       x += 48;
-	       break;
-	       
-	    default:
-	       x += step_offsets[face][0] * d;
-	       y += step_offsets[face][1] * d;
-	       break;
-	    }
-	 }
-      }
-      max_offset = total_height * pitch;
-      break;
-   }
-   case GL_TEXTURE_3D: {
-      GLuint depth_packing = 0, depth_pack_pitch;
-      GLuint tmp_numLevels = numLevels;
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      depth_pack_pitch = pitch;
-      
-      t->intel.base.dirty_images[0] = ~0;
-
-
-      for ( total_height = i = 0 ; i < tmp_numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 
-
-
-	 total_height += MAX2(2, t->intel.image[0][i].image->Height) * 
-	    MAX2((t->intel.image[0][i].image->Depth >> depth_packing), 1);
-
-	 /* When alignment dominates, can't increase depth packing?
-	  * Or does pitch grow???  What are the alignment constraints,
-	  * anyway?
-	  */
-	 if (depth_pack_pitch > 4) {
-	    depth_packing++;
-	    depth_pack_pitch <<= 2;
-	 }
-      }
-
-      max_offset = total_height * pitch;
-      break;
-   }
-   default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-      max_offset = 0;
-
-      for ( offset = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = offset;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-
-	 if (t->intel.image[0][i].image->IsCompressed)
-	    sz = MAX2(1, t->intel.image[0][i].image->Height/4) * pitch;
-	 else
-	    sz = MAX2(2, t->intel.image[0][i].image->Height) * pitch;
-	 
-	 /* Because the images are packed better, the final offset
-	  * might not be the maximal one:
-	  */
-	 max_offset = MAX2(max_offset, offset + sz);
-
-	 /* LPT change: step right after second mipmap.
-	  */
-	 if (i == 1) 
-	    offset += pitch / 2;
-	 else 
-	    offset += sz;
 
-      }
-      break;
-   }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = max_offset;
-   t->intel.max_level = numLevels-1;
-}
-
-
-
-
-static void i915SetTexImages( i915ContextPtr i915, 
-			     struct gl_texture_object *tObj )
+static GLuint
+translate_texture_format(GLuint mesa_format, GLenum DepthMode)
 {
-   GLuint textureFormat;
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   GLint ss2 = 0;
-
-   switch( baseImage->TexFormat->MesaFormat ) {
+   switch (mesa_format) {
    case MESA_FORMAT_L8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_L8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_L8;
    case MESA_FORMAT_I8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8;
    case MESA_FORMAT_A8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_A8; 
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_A8;
    case MESA_FORMAT_AL88:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_AY88;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_AY88;
    case MESA_FORMAT_RGB565:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_RGB565;
    case MESA_FORMAT_ARGB1555:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
    case MESA_FORMAT_ARGB4444:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB4444;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      t->intel.texelBytes = 4;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      break;
-
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
    case MESA_FORMAT_YCBCR_REV:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL);
-      ss2 |= SS2_COLORSPACE_CONVERSION;
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY);
-      ss2 |= SS2_COLORSPACE_CONVERSION;
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
    case MESA_FORMAT_RGB_FXT1:
    case MESA_FORMAT_RGBA_FXT1:
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
    case MESA_FORMAT_Z16:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_16BIT | MT_16BIT_L16);
-      break;
-
+      if (DepthMode == GL_ALPHA)
+          return (MAPSURF_16BIT | MT_16BIT_A16);
+      else if (DepthMode == GL_INTENSITY)
+          return (MAPSURF_16BIT | MT_16BIT_I16);
+      else
+          return (MAPSURF_16BIT | MT_16BIT_L16);
    case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-     /* 
-      * DXTn pitches are Width/4 * blocksize in bytes 
-      * for DXT1: blocksize=8 so Width/4*8 = Width * 2 
-      * for DXT3/5: blocksize=16 so Width/4*16 = Width * 4
-      */
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
    case MESA_FORMAT_RGBA_DXT3:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
    case MESA_FORMAT_RGBA_DXT5:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-     break;
-
-#if 0
-   case MESA_FORMAT_Z24_S8:
-      t->intel.texelBytes = 4;
-      textureFormat = (MAPSURF_32BIT | MT_32BIT_xL824);
-      break;
-#endif
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
+   case MESA_FORMAT_S8_Z24:
+      return (MAPSURF_32BIT | MT_32BIT_xI824);
    default:
-      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__,
-	      baseImage->TexFormat->MesaFormat);
+      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
       abort();
+      return 0;
    }
+}
 
-   switch (i915->intel.intelScreen->deviceID) {
-   case PCI_CHIP_I945_G:
-   case PCI_CHIP_I945_GM:
-   case PCI_CHIP_I945_GME:
-   case PCI_CHIP_G33_G:
-   case PCI_CHIP_Q33_G:
-   case PCI_CHIP_Q35_G:
-       i945LayoutTextureImages( i915, tObj );
-       break;
-   default:
-       i915LayoutTextureImages( i915, tObj );
-       break;
-   }
-
-   t->Setup[I915_TEXREG_MS3] = 
-      (((tObj->Image[0][t->intel.base.firstLevel]->Height - 1) << MS3_HEIGHT_SHIFT) |
-       ((tObj->Image[0][t->intel.base.firstLevel]->Width - 1) << MS3_WIDTH_SHIFT) |
-       textureFormat |
-       MS3_USE_FENCE_REGS);
-
-   t->Setup[I915_TEXREG_MS4] = 
-      ((((t->intel.Pitch / 4) - 1) << MS4_PITCH_SHIFT) | 
-       MS4_CUBE_FACE_ENA_MASK |
-       (((t->intel.max_level * 4)) << MS4_MAX_LOD_SHIFT) |
-       ((tObj->Image[0][t->intel.base.firstLevel]->Depth - 1) << MS4_VOLUME_DEPTH_SHIFT));
-
-   t->Setup[I915_TEXREG_SS2] &= ~(SS2_COLORSPACE_CONVERSION);
-   t->Setup[I915_TEXREG_SS2] |= ss2;
-
-   t->intel.dirty = I915_UPLOAD_TEX_ALL;
 
-}
 
 
 /* The i915 (and related graphics cores) do not support GL_CLAMP.  The
  * Intel drivers for "other operating systems" implement GL_CLAMP as
  * GL_CLAMP_TO_EDGE, so the same is done here.
  */
-static GLuint translate_wrap_mode( GLenum wrap )
+static GLuint
+translate_wrap_mode(GLenum wrap)
 {
-   switch( wrap ) {
-   case GL_REPEAT: return TEXCOORDMODE_WRAP;
-   case GL_CLAMP:  return TEXCOORDMODE_CLAMP_EDGE; /* not quite correct */
-   case GL_CLAMP_TO_EDGE: return TEXCOORDMODE_CLAMP_EDGE;
-   case GL_CLAMP_TO_BORDER: return TEXCOORDMODE_CLAMP_BORDER;
-   case GL_MIRRORED_REPEAT: return TEXCOORDMODE_MIRROR;
-   default: return TEXCOORDMODE_WRAP;
-   }
-}
-
-
-/**
- */
-static void i915ImportTexObjState( struct gl_texture_object *texObj )
-{   
-   i915TextureObjectPtr t = (i915TextureObjectPtr)texObj->DriverData;
-   int minFilt = 0, mipFilt = 0, magFilt = 0, shadow = 0;
-
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   switch (texObj->MinFilter) {
-   case GL_NEAREST:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_NONE;
-      break;
-   case GL_LINEAR:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_NONE;
-      break;
-   case GL_NEAREST_MIPMAP_NEAREST:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_NEAREST;
-      break;
-   case GL_LINEAR_MIPMAP_NEAREST:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_NEAREST;
-      break;
-   case GL_NEAREST_MIPMAP_LINEAR:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_LINEAR;
-      break;
-   case GL_LINEAR_MIPMAP_LINEAR:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_LINEAR;
-      break;
+   switch (wrap) {
+   case GL_REPEAT:
+      return TEXCOORDMODE_WRAP;
+   case GL_CLAMP:
+      return TEXCOORDMODE_CLAMP_EDGE;   /* not quite correct */
+   case GL_CLAMP_TO_EDGE:
+      return TEXCOORDMODE_CLAMP_EDGE;
+   case GL_CLAMP_TO_BORDER:
+      return TEXCOORDMODE_CLAMP_BORDER;
+   case GL_MIRRORED_REPEAT:
+      return TEXCOORDMODE_MIRROR;
    default:
-      break;
-   }
-
-   if ( texObj->MaxAnisotropy > 1.0 ) {
-      minFilt = FILTER_ANISOTROPIC; 
-      magFilt = FILTER_ANISOTROPIC;
-   }
-   else {
-      switch (texObj->MagFilter) {
-      case GL_NEAREST:
-	 magFilt = FILTER_NEAREST;
-	 break;
-      case GL_LINEAR:
-	 magFilt = FILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }  
-   }
-
-   if (texObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB && 
-       texObj->Target != GL_TEXTURE_3D) {
-
-      shadow = SS2_SHADOW_ENABLE;
-      shadow |= intel_translate_compare_func( texObj->CompareFunc );
-      
-      minFilt = FILTER_4X4_FLAT;
-      magFilt = FILTER_4X4_FLAT;
-   }
-
-
-   t->Setup[I915_TEXREG_SS2] &= ~(SS2_MIN_FILTER_MASK |
-				 SS2_MIP_FILTER_MASK |
-				 SS2_MAG_FILTER_MASK |
-				 SS2_SHADOW_ENABLE |
-				 SS2_SHADOW_FUNC_MASK);
-   t->Setup[I915_TEXREG_SS2] |= ((minFilt << SS2_MIN_FILTER_SHIFT) |
-				(mipFilt << SS2_MIP_FILTER_SHIFT) |
-				(magFilt << SS2_MAG_FILTER_SHIFT) |
-				shadow);
-
-   {
-      GLuint ss3 = t->Setup[I915_TEXREG_SS3] & ~(SS3_TCX_ADDR_MODE_MASK |
-						SS3_TCY_ADDR_MODE_MASK |
-						SS3_TCZ_ADDR_MODE_MASK);
-      GLenum ws = texObj->WrapS;
-      GLenum wt = texObj->WrapT;
-      GLenum wr = texObj->WrapR;
-      
-      t->refs_border_color = 0;
-
-      if (texObj->Target == GL_TEXTURE_3D &&
-	  (texObj->MinFilter != GL_NEAREST ||
-	   texObj->MagFilter != GL_NEAREST)) {
-	 
-	 /* Try to mimic GL_CLAMP functionality a little better -
-	  * switch to CLAMP_TO_BORDER whenever a non-NEAREST filter is
-	  * in use.  Only do this for 3D textures at the moment --
-	  * doing it universally would fix the conform texbc.c
-	  * failure, though.
-	  */
-	 if (ws == GL_CLAMP) ws = GL_CLAMP_TO_BORDER;
-	 if (wt == GL_CLAMP) wt = GL_CLAMP_TO_BORDER;
-	 if (wr == GL_CLAMP) wr = GL_CLAMP_TO_BORDER;
-
-	 /* 3D textures don't seem to respect the border color.
-	  * Fallback if there's ever a danger that they might refer to
-	  * it.
-	  */
-	 if (ws == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-	 if (wt == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-	 if (wr == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-      }
-
-      ss3 |= translate_wrap_mode(ws) << SS3_TCX_ADDR_MODE_SHIFT;
-      ss3 |= translate_wrap_mode(wt) << SS3_TCY_ADDR_MODE_SHIFT;
-      ss3 |= translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT;
-   
-      if (ss3 != t->Setup[I915_TEXREG_SS3]) {
-	 t->intel.dirty = I915_UPLOAD_TEX_ALL;
-	 t->Setup[I915_TEXREG_SS3] = ss3;
-      }
-   }
-
-   {   
-      const GLubyte *color = texObj->_BorderChan;
-
-      t->Setup[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(color[0],color[1],
-						     color[2],color[3]);
+      return TEXCOORDMODE_WRAP;
    }
 }
 
 
 
-static void i915_import_tex_unit( i915ContextPtr i915, 
-				 i915TextureObjectPtr t,
-				 GLuint unit )
+/* Recalculate all state from scratch.  Perhaps not the most
+ * efficient, but this has gotten complex enough that we need
+ * something which is understandable and reliable.
+ */
+static GLboolean
+i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 {
-   GLuint state[I915_TEX_SETUP_SIZE];
+   GLcontext *ctx = &intel->ctx;
+   struct i915_context *i915 = i915_context(ctx);
+   struct gl_texture_unit *tUnit = &ctx->Texture.Unit[unit];
+   struct gl_texture_object *tObj = tUnit->_Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   struct gl_texture_image *firstImage;
+   GLuint *state = i915->state.Tex[unit], format, pitch;
+   GLint lodbias;
 
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
-      fprintf(stderr, "%s unit(%d)\n", __FUNCTION__, unit);
-   
-   if (i915->intel.CurrentTexObj[unit]) 
-      i915->intel.CurrentTexObj[unit]->base.bound &= ~(1U << unit);
+   memset(state, 0, sizeof(state));
 
-   i915->intel.CurrentTexObj[unit] = (intelTextureObjectPtr)t;
-   t->intel.base.bound |= (1 << unit);
+   /*We need to refcount these. */
 
-   if (t->intel.dirty & I915_UPLOAD_TEX(unit)) {
-      i915ImportTexObjState( t->intel.base.tObj );
-      t->intel.dirty &= ~I915_UPLOAD_TEX(unit);
+   if (i915->state.tex_buffer[unit] != NULL) {
+       dri_bo_unreference(i915->state.tex_buffer[unit]);
+       i915->state.tex_buffer[unit] = NULL;
    }
 
-   state[I915_TEXREG_MS2] = t->intel.TextureOffset;
-   state[I915_TEXREG_MS3] = t->Setup[I915_TEXREG_MS3];
-   state[I915_TEXREG_MS4] = t->Setup[I915_TEXREG_MS4];
-
-   state[I915_TEXREG_SS2] = (i915->state.Tex[unit][I915_TEXREG_SS2] &
-			    SS2_LOD_BIAS_MASK);
-   state[I915_TEXREG_SS2] |= (t->Setup[I915_TEXREG_SS2] & ~SS2_LOD_BIAS_MASK);
-
-   state[I915_TEXREG_SS3] = (i915->state.Tex[unit][I915_TEXREG_SS3] &
-			    SS3_NORMALIZED_COORDS);
-   state[I915_TEXREG_SS3] |= (t->Setup[I915_TEXREG_SS3] &
-			     ~(SS3_NORMALIZED_COORDS|
-			       SS3_TEXTUREMAP_INDEX_MASK));
+   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
+      return GL_FALSE;
 
-   state[I915_TEXREG_SS3] |= (unit<<SS3_TEXTUREMAP_INDEX_SHIFT);
+   /* Get first image here, since intelObj->firstLevel will get set in
+    * the intel_finalize_mipmap_tree() call above.
+    */
+   firstImage = tObj->Image[0][intelObj->firstLevel];
 
-   state[I915_TEXREG_SS4] = t->Setup[I915_TEXREG_SS4];
+   if (intelObj->imageOverride) {
+      i915->state.tex_buffer[unit] = NULL;
+      i915->state.tex_offset[unit] = intelObj->textureOffset;
 
+      switch (intelObj->depthOverride) {
+      case 32:
+	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
+	 break;
+      case 24:
+      default:
+	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
+	 break;
+      case 16:
+	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
+	 break;
+      }
 
-   if (memcmp(state, i915->state.Tex[unit], sizeof(state)) != 0) {
-      I915_STATECHANGE( i915, I915_UPLOAD_TEX(unit) );
-      memcpy(i915->state.Tex[unit], state, sizeof(state));
+      pitch = intelObj->pitchOverride;
+   } else {
+      dri_bo_reference(intelObj->mt->region->buffer);
+      i915->state.tex_buffer[unit] = intelObj->mt->region->buffer;
+      i915->state.tex_offset[unit] =  intel_miptree_image_offset(intelObj->mt,
+								 0, intelObj->
+								 firstLevel);
+
+      format = translate_texture_format(firstImage->TexFormat->MesaFormat, 
+		tObj->DepthMode);
+      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
    }
-}
-
 
+   state[I915_TEXREG_MS3] =
+      (((firstImage->Height - 1) << MS3_HEIGHT_SHIFT) |
+       ((firstImage->Width - 1) << MS3_WIDTH_SHIFT) | format |
+       MS3_USE_FENCE_REGS);
 
-static GLboolean enable_tex_common( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-
-   if (0) fprintf(stderr, "%s %d\n", __FUNCTION__, unit);
+   state[I915_TEXREG_MS4] =
+     ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | MS4_CUBE_FACE_ENA_MASK |
+       ((((intelObj->lastLevel - intelObj->firstLevel) * 4)) <<
+	MS4_MAX_LOD_SHIFT) | ((firstImage->Depth - 1) <<
+			      MS4_VOLUME_DEPTH_SHIFT));
 
-   if (!(i915->state.active & I915_UPLOAD_TEX(unit))) {
-      I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_TRUE);
-   }
 
-   /* Fallback if there's a texture border */
-   if ( tObj->Image[0][tObj->BaseLevel]->Border > 0 ) {
-      return GL_FALSE;
-   }
+   {
+      GLuint minFilt, mipFilt, magFilt;
 
+      switch (tObj->MinFilter) {
+      case GL_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_NEAREST_MIPMAP_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_LINEAR_MIPMAP_NEAREST:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_NEAREST_MIPMAP_LINEAR:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      case GL_LINEAR_MIPMAP_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      default:
+         return GL_FALSE;
+      }
 
-   /* Update state if this is a different texture object to last
-    * time.
-    */
-   if (i915->intel.CurrentTexObj[unit] != &t->intel || 
-       (t->intel.dirty & I915_UPLOAD_TEX(unit))) {
-      i915_import_tex_unit( i915, t, unit);
-      i915->tex_program.translated = 0;
-   }
+      if (tObj->MaxAnisotropy > 1.0) {
+         minFilt = FILTER_ANISOTROPIC;
+         magFilt = FILTER_ANISOTROPIC;
+      }
+      else {
+         switch (tObj->MagFilter) {
+         case GL_NEAREST:
+            magFilt = FILTER_NEAREST;
+            break;
+         case GL_LINEAR:
+            magFilt = FILTER_LINEAR;
+            break;
+         default:
+            return GL_FALSE;
+         }
+      }
 
-   return GL_TRUE;
-}
+      lodbias = (int) ((tUnit->LodBias + tObj->LodBias) * 16.0);
+      if (lodbias < -256)
+          lodbias = -256;
+      if (lodbias > 255)
+          lodbias = 255;
+      state[I915_TEXREG_SS2] = ((lodbias << SS2_LOD_BIAS_SHIFT) & 
+                                SS2_LOD_BIAS_MASK);
 
-static GLboolean enable_tex_rect( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
+      /* YUV conversion:
+       */
+      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
+          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
+         state[I915_TEXREG_SS2] |= SS2_COLORSPACE_CONVERSION;
 
-   ss3 &= ~SS3_NORMALIZED_COORDS;
+      /* Shadow:
+       */
+      if (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB &&
+          tObj->Target != GL_TEXTURE_3D) {
+         if (tObj->Target == GL_TEXTURE_1D) 
+            return GL_FALSE;
 
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
-   }
+         state[I915_TEXREG_SS2] |=
+            (SS2_SHADOW_ENABLE |
+             intel_translate_shadow_compare_func(tObj->CompareFunc));
 
-   /* Upload teximages (not pipelined)
-    */
-   if (t->intel.base.dirty_images[0]) {
-      i915SetTexImages( i915, tObj );
-      if (!intelUploadTexImages( &i915->intel, &t->intel, 0 )) {
-	 return GL_FALSE;
+         minFilt = FILTER_4X4_FLAT;
+         magFilt = FILTER_4X4_FLAT;
       }
-   }
-
-   return GL_TRUE;
-}
 
+      state[I915_TEXREG_SS2] |= ((minFilt << SS2_MIN_FILTER_SHIFT) |
+                                 (mipFilt << SS2_MIP_FILTER_SHIFT) |
+                                 (magFilt << SS2_MAG_FILTER_SHIFT));
+   }
 
-static GLboolean enable_tex_2d( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
+   {
+      GLenum ws = tObj->WrapS;
+      GLenum wt = tObj->WrapT;
+      GLenum wr = tObj->WrapR;
 
-   ss3 |= SS3_NORMALIZED_COORDS;
 
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
-   }
+      /* 3D textures don't seem to respect the border color.
+       * Fallback if there's ever a danger that they might refer to
+       * it.  
+       * 
+       * Effectively this means fallback on 3D clamp or
+       * clamp_to_border.
+       */
+      if (tObj->Target == GL_TEXTURE_3D &&
+          (tObj->MinFilter != GL_NEAREST ||
+           tObj->MagFilter != GL_NEAREST) &&
+          (ws == GL_CLAMP ||
+           wt == GL_CLAMP ||
+           wr == GL_CLAMP ||
+           ws == GL_CLAMP_TO_BORDER ||
+           wt == GL_CLAMP_TO_BORDER || wr == GL_CLAMP_TO_BORDER))
+         return GL_FALSE;
+
+      /* Only support TEXCOORDMODE_CLAMP_EDGE and TEXCOORDMODE_CUBE (not 
+       * used) when using cube map texture coordinates
+       */
+      if (tObj->Target == GL_TEXTURE_CUBE_MAP_ARB &&
+          (((ws != GL_CLAMP) && (ws != GL_CLAMP_TO_EDGE)) ||
+           ((wr != GL_CLAMP) && (wr != GL_CLAMP_TO_EDGE))))
+          return GL_FALSE;
 
-   /* Upload teximages (not pipelined)
-    */
-   if (t->intel.base.dirty_images[0]) {
-      i915SetTexImages( i915, tObj );
-      if (!intelUploadTexImages( &i915->intel, &t->intel, 0 )) {
-	 return GL_FALSE;
-      }
-   }
+      state[I915_TEXREG_SS3] = ss3;     /* SS3_NORMALIZED_COORDS */
 
-   return GL_TRUE;
-}
+      state[I915_TEXREG_SS3] |=
+         ((translate_wrap_mode(ws) << SS3_TCX_ADDR_MODE_SHIFT) |
+          (translate_wrap_mode(wt) << SS3_TCY_ADDR_MODE_SHIFT) |
+          (translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT));
 
-static GLboolean enable_tex_cube( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
-   GLuint face;
-
-   ss3 |= SS3_NORMALIZED_COORDS;
-
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
+      state[I915_TEXREG_SS3] |= (unit << SS3_TEXTUREMAP_INDEX_SHIFT);
    }
 
-   /* Upload teximages (not pipelined)
-    */
-   if ( t->intel.base.dirty_images[0] || t->intel.base.dirty_images[1] ||
-        t->intel.base.dirty_images[2] || t->intel.base.dirty_images[3] ||
-        t->intel.base.dirty_images[4] || t->intel.base.dirty_images[5] ) {
-      i915SetTexImages( i915, tObj );
-   }
 
-   /* upload (per face) */
-   for (face = 0; face < 6; face++) {
-      if (t->intel.base.dirty_images[face]) {
-	 if (!intelUploadTexImages( &i915->intel, &t->intel, face )) {
-	    return GL_FALSE;
-	 }
-      }
+   if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
+      /* GL specs that border color for depth textures is taken from the
+       * R channel, while the hardware uses A.  Spam R into all the channels
+       * for safety.
+       */
+      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
+						   tObj->_BorderChan[0],
+						   tObj->_BorderChan[0],
+						   tObj->_BorderChan[0]);
+   } else {
+      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
+						   tObj->_BorderChan[1],
+						   tObj->_BorderChan[2],
+						   tObj->_BorderChan[3]);
    }
 
 
-   return GL_TRUE;
-}
-
-static GLboolean enable_tex_3d( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-
-   /* 3D textures on I915 seem to get bogus border colors, hence this
-    * fallback:
+   I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_TRUE);
+   /* memcmp was already disabled, but definitely won't work as the
+    * region might now change and that wouldn't be detected:
     */
-   if (t->refs_border_color)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
+   I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
 
- 
-static GLboolean disable_tex( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (i915->state.active & I915_UPLOAD_TEX(unit)) {
-      I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_FALSE);
-   }
 
-   /* The old texture is no longer bound to this texture unit.
-    * Mark it as such.
-    */
-   if ( i915->intel.CurrentTexObj[unit] != NULL ) {
-      i915->intel.CurrentTexObj[unit]->base.bound &= ~(1U << 0);
-      i915->intel.CurrentTexObj[unit] = NULL;
-   }
+#if 0
+   DBG(TEXTURE, "state[I915_TEXREG_SS2] = 0x%x\n", state[I915_TEXREG_SS2]);
+   DBG(TEXTURE, "state[I915_TEXREG_SS3] = 0x%x\n", state[I915_TEXREG_SS3]);
+   DBG(TEXTURE, "state[I915_TEXREG_SS4] = 0x%x\n", state[I915_TEXREG_SS4]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS2] = 0x%x\n", state[I915_TEXREG_MS2]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS3] = 0x%x\n", state[I915_TEXREG_MS3]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS4] = 0x%x\n", state[I915_TEXREG_MS4]);
+#endif
 
    return GL_TRUE;
 }
 
-static GLboolean i915UpdateTexUnit( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-
-   if (texUnit->_ReallyEnabled &&
-       INTEL_CONTEXT(ctx)->intelScreen->tex.size < 2048 * 1024)
-      return GL_FALSE;
 
-   switch (texUnit->_ReallyEnabled) {
-   case TEXTURE_1D_BIT:
-   case TEXTURE_2D_BIT:
-      return (enable_tex_2d( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_RECT_BIT:
-      return (enable_tex_rect( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_CUBE_BIT:
-      return (enable_tex_cube( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_3D_BIT:
-       return (enable_tex_2d( ctx, unit ) && 
-	       enable_tex_common( ctx, unit ) &&
-	       enable_tex_3d( ctx, unit)); 
-   case 0:
-      return disable_tex( ctx, unit );
-   default:
-      return GL_FALSE;
-   }
-}
 
 
-void i915UpdateTextureState( intelContextPtr intel )
+void
+i915UpdateTextureState(struct intel_context *intel)
 {
-   GLcontext *ctx = &intel->ctx;
    GLboolean ok = GL_TRUE;
    GLuint i;
 
-   for (i = 0 ; i < I915_TEX_UNITS && ok ; i++) {
-      ok = i915UpdateTexUnit( ctx, i );
+   for (i = 0; i < I915_TEX_UNITS && ok; i++) {
+      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
+      case TEXTURE_1D_BIT:
+      case TEXTURE_2D_BIT:
+      case TEXTURE_CUBE_BIT:
+      case TEXTURE_3D_BIT:
+         ok = i915_update_tex_unit(intel, i, SS3_NORMALIZED_COORDS);
+         break;
+      case TEXTURE_RECT_BIT:
+         ok = i915_update_tex_unit(intel, i, 0);
+         break;
+      case 0:{
+            struct i915_context *i915 = i915_context(&intel->ctx);
+            if (i915->state.active & I915_UPLOAD_TEX(i))
+               I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(i), GL_FALSE);
+
+	    if (i915->state.tex_buffer[i] != NULL) {
+	       dri_bo_unreference(i915->state.tex_buffer[i]);
+	       i915->state.tex_buffer[i] = NULL;
+	    }
+
+            break;
+         }
+      default:
+         ok = GL_FALSE;
+         break;
+      }
    }
 
-   FALLBACK( intel, I915_FALLBACK_TEXTURE, !ok );
+   FALLBACK(intel, I915_FALLBACK_TEXTURE, !ok);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index cc8a605e505..e79c955d64d 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -27,125 +27,158 @@
 
 
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
 
 #include "intel_batchbuffer.h"
+#include "intel_tex.h"
+#include "intel_regions.h"
+#include "intel_tris.h"
 
 #include "i915_reg.h"
 #include "i915_context.h"
 
-static void i915_render_start( intelContextPtr intel )
+#include "glapi/glapi.h"
+
+static void
+i915_render_prevalidate(struct intel_context *intel)
 {
-   GLcontext *ctx = &intel->ctx;
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
-   if (ctx->FragmentProgram._Active) 
-      i915ValidateFragmentProgram( i915 );
-   else {
-      assert(!ctx->FragmentProgram._MaintainTexEnvProgram);
-      i915ValidateTextureProgram( i915 );
-   }
+   if (!intel->Fallback)
+       i915ValidateFragmentProgram(i915);
+}
+
+static void
+i915_render_start(struct intel_context *intel)
+{
 }
 
 
-static void i915_reduced_primitive_state( intelContextPtr intel,
-					  GLenum rprim )
+static void
+i915_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
 {
-    i915ContextPtr i915 = I915_CONTEXT(intel);
-    GLuint st1 = i915->state.Stipple[I915_STPREG_ST1];
-
-    st1 &= ~ST1_ENABLE;
-
-    switch (rprim) {
-    case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
-    case GL_TRIANGLES:
-       if (intel->ctx.Polygon.StippleFlag &&
-	   intel->hw_stipple)
-	  st1 |= ST1_ENABLE;
-       break;
-    case GL_LINES:
-    case GL_POINTS:
-    default:
-       break;
-    }
-
-    i915->intel.reduced_primitive = rprim;
-
-    if (st1 != i915->state.Stipple[I915_STPREG_ST1]) {
-       I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-       i915->state.Stipple[I915_STPREG_ST1] = st1;
-    }
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   GLuint st1 = i915->state.Stipple[I915_STPREG_ST1];
+
+   st1 &= ~ST1_ENABLE;
+
+   switch (rprim) {
+   case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
+   case GL_TRIANGLES:
+      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
+         st1 |= ST1_ENABLE;
+      break;
+   case GL_LINES:
+   case GL_POINTS:
+   default:
+      break;
+   }
+
+   i915->intel.reduced_primitive = rprim;
+
+   if (st1 != i915->state.Stipple[I915_STPREG_ST1]) {
+      INTEL_FIREVERTICES(intel);
+
+      I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
+      i915->state.Stipple[I915_STPREG_ST1] = st1;
+   }
 }
 
 
 /* Pull apart the vertex format registers and figure out how large a
  * vertex is supposed to be. 
  */
-static GLboolean i915_check_vertex_size( intelContextPtr intel,
-					 GLuint expected )
+static GLboolean
+i915_check_vertex_size(struct intel_context *intel, GLuint expected)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
    int lis2 = i915->current->Ctx[I915_CTXREG_LIS2];
    int lis4 = i915->current->Ctx[I915_CTXREG_LIS4];
    int i, sz = 0;
 
    switch (lis4 & S4_VFMT_XYZW_MASK) {
-   case S4_VFMT_XY: sz = 2; break;
-   case S4_VFMT_XYZ: sz = 3; break;
-   case S4_VFMT_XYW: sz = 3; break;
-   case S4_VFMT_XYZW: sz = 4; break;
-   default: 
+   case S4_VFMT_XY:
+      sz = 2;
+      break;
+   case S4_VFMT_XYZ:
+      sz = 3;
+      break;
+   case S4_VFMT_XYW:
+      sz = 3;
+      break;
+   case S4_VFMT_XYZW:
+      sz = 4;
+      break;
+   default:
       fprintf(stderr, "no xyzw specified\n");
       return 0;
    }
 
-   if (lis4 & S4_VFMT_SPEC_FOG) sz++;
-   if (lis4 & S4_VFMT_COLOR) sz++;
-   if (lis4 & S4_VFMT_DEPTH_OFFSET) sz++;
-   if (lis4 & S4_VFMT_POINT_WIDTH) sz++;
-   if (lis4 & S4_VFMT_FOG_PARAM) sz++;
-	
-   for (i = 0 ; i < 8 ; i++) { 
+   if (lis4 & S4_VFMT_SPEC_FOG)
+      sz++;
+   if (lis4 & S4_VFMT_COLOR)
+      sz++;
+   if (lis4 & S4_VFMT_DEPTH_OFFSET)
+      sz++;
+   if (lis4 & S4_VFMT_POINT_WIDTH)
+      sz++;
+   if (lis4 & S4_VFMT_FOG_PARAM)
+      sz++;
+
+   for (i = 0; i < 8; i++) {
       switch (lis2 & S2_TEXCOORD_FMT0_MASK) {
-      case TEXCOORDFMT_2D: sz += 2; break;
-      case TEXCOORDFMT_3D: sz += 3; break;
-      case TEXCOORDFMT_4D: sz += 4; break;
-      case TEXCOORDFMT_1D: sz += 1; break;
-      case TEXCOORDFMT_2D_16: sz += 1; break;
-      case TEXCOORDFMT_4D_16: sz += 2; break;
-      case TEXCOORDFMT_NOT_PRESENT: break;
+      case TEXCOORDFMT_2D:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_3D:
+         sz += 3;
+         break;
+      case TEXCOORDFMT_4D:
+         sz += 4;
+         break;
+      case TEXCOORDFMT_1D:
+         sz += 1;
+         break;
+      case TEXCOORDFMT_2D_16:
+         sz += 1;
+         break;
+      case TEXCOORDFMT_4D_16:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_NOT_PRESENT:
+         break;
       default:
-	 fprintf(stderr, "bad texcoord fmt %d\n", i);
-	 return GL_FALSE;
+         fprintf(stderr, "bad texcoord fmt %d\n", i);
+         return GL_FALSE;
       }
       lis2 >>= S2_TEXCOORD_FMT1_SHIFT;
    }
-	
-   if (sz != expected) 
+
+   if (sz != expected)
       fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-   
+
    return sz == expected;
 }
 
 
-static void i915_emit_invarient_state( intelContextPtr intel )
+static void
+i915_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH( 20 );
+   BEGIN_BATCH(200, IGNORE_CLIPRECTS);
 
    OUT_BATCH(_3DSTATE_AA_CMD |
-	     AA_LINE_ECAAR_WIDTH_ENABLE |
-	     AA_LINE_ECAAR_WIDTH_1_0 |
-	     AA_LINE_REGION_WIDTH_ENABLE |
-	     AA_LINE_REGION_WIDTH_1_0);
+             AA_LINE_ECAAR_WIDTH_ENABLE |
+             AA_LINE_ECAAR_WIDTH_1_0 |
+             AA_LINE_REGION_WIDTH_ENABLE | AA_LINE_REGION_WIDTH_1_0);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -158,35 +191,27 @@ static void i915_emit_invarient_state( intelContextPtr intel )
 
    /* Don't support texture crossbar yet */
    OUT_BATCH(_3DSTATE_COORD_SET_BINDINGS |
-	     CSB_TCB(0, 0) |
-	     CSB_TCB(1, 1) |
-	     CSB_TCB(2, 2) |
-	     CSB_TCB(3, 3) |
-	     CSB_TCB(4, 4) |
-	     CSB_TCB(5, 5) |
-	     CSB_TCB(6, 6) |
-	     CSB_TCB(7, 7));
+             CSB_TCB(0, 0) |
+             CSB_TCB(1, 1) |
+             CSB_TCB(2, 2) |
+             CSB_TCB(3, 3) |
+             CSB_TCB(4, 4) | CSB_TCB(5, 5) | CSB_TCB(6, 6) | CSB_TCB(7, 7));
 
    OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-	     ENABLE_POINT_RASTER_RULE |
-	     OGL_POINT_RASTER_RULE |
-	     ENABLE_LINE_STRIP_PROVOKE_VRTX |
-	     ENABLE_TRI_FAN_PROVOKE_VRTX |
-	     LINE_STRIP_PROVOKE_VRTX(1) |
-	     TRI_FAN_PROVOKE_VRTX(2) | 
-	     ENABLE_TEXKILL_3D_4D |
-	     TEXKILL_4D);
+             ENABLE_POINT_RASTER_RULE |
+             OGL_POINT_RASTER_RULE |
+             ENABLE_LINE_STRIP_PROVOKE_VRTX |
+             ENABLE_TRI_FAN_PROVOKE_VRTX |
+             LINE_STRIP_PROVOKE_VRTX(1) |
+             TRI_FAN_PROVOKE_VRTX(2) | ENABLE_TEXKILL_3D_4D | TEXKILL_4D);
 
    /* Need to initialize this to zero.
     */
-   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | 
-	     I1_LOAD_S(3) |
-	     (0));
+   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (0));
    OUT_BATCH(0);
- 
+
    /* XXX: Use this */
-   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | 
-	     DISABLE_SCISSOR_RECT);
+   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
 
    OUT_BATCH(_3DSTATE_SCISSOR_RECT_0_CMD);
    OUT_BATCH(0);
@@ -194,29 +219,23 @@ static void i915_emit_invarient_state( intelContextPtr intel )
 
    OUT_BATCH(_3DSTATE_DEPTH_SUBRECT_DISABLE);
 
-   OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0); /* disable indirect state */
+   OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0);       /* disable indirect state */
    OUT_BATCH(0);
 
 
    /* Don't support twosided stencil yet */
-   OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS |
-	     BFO_ENABLE_STENCIL_TWO_SIDE |
-	     0 );
-   
+   OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS | BFO_ENABLE_STENCIL_TWO_SIDE | 0);
+   OUT_BATCH(0);
+
    ADVANCE_BATCH();
 }
 
 
-#define emit( intel, state, size )			\
-do {							\
-   int k;						\
-   BEGIN_BATCH( (size) / sizeof(GLuint));		\
-   for (k = 0 ; k < (size) / sizeof(GLuint) ; k++)	\
-      OUT_BATCH((state)[k]);				\
-   ADVANCE_BATCH();					\
-} while (0);
+#define emit(intel, state, size )		     \
+   intel_batchbuffer_data(intel->batch, state, size, IGNORE_CLIPRECTS )
 
-static GLuint get_dirty( struct i915_hw_state *state )
+static GLuint
+get_dirty(struct i915_hw_state *state)
 {
    GLuint dirty;
 
@@ -227,94 +246,183 @@ static GLuint get_dirty( struct i915_hw_state *state )
    if (dirty & I915_UPLOAD_TEX_ALL)
       state->emitted &= ~I915_UPLOAD_TEX_ALL;
    dirty = state->active & ~state->emitted;
-
    return dirty;
 }
 
 
-static GLuint get_state_size( struct i915_hw_state *state )
+static GLuint
+get_state_size(struct i915_hw_state *state)
 {
    GLuint dirty = get_dirty(state);
    GLuint i;
    GLuint sz = 0;
 
    if (dirty & I915_UPLOAD_INVARIENT)
-      sz += 20 * sizeof(int);
+      sz += 30 * 4;
 
    if (dirty & I915_UPLOAD_CTX)
       sz += sizeof(state->Ctx);
 
-   if (dirty & I915_UPLOAD_BUFFERS) 
+   if (dirty & I915_UPLOAD_BUFFERS)
       sz += sizeof(state->Buffer);
 
    if (dirty & I915_UPLOAD_STIPPLE)
       sz += sizeof(state->Stipple);
 
-   if (dirty & I915_UPLOAD_FOG) 
+   if (dirty & I915_UPLOAD_FOG)
       sz += sizeof(state->Fog);
 
    if (dirty & I915_UPLOAD_TEX_ALL) {
       int nr = 0;
-      for (i = 0; i < I915_TEX_UNITS; i++) 
-	 if (dirty & I915_UPLOAD_TEX(i)) 
-	    nr++;
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i))
+            nr++;
 
-      sz += (2+nr*3) * sizeof(GLuint) * 2;
+      sz += (2 + nr * 3) * sizeof(GLuint) * 2;
    }
 
-   if (dirty & I915_UPLOAD_CONSTANTS) 
+   if (dirty & I915_UPLOAD_CONSTANTS)
       sz += state->ConstantSize * sizeof(GLuint);
 
-   if (dirty & I915_UPLOAD_PROGRAM) 
+   if (dirty & I915_UPLOAD_PROGRAM)
       sz += state->ProgramSize * sizeof(GLuint);
 
    return sz;
 }
 
-
 /* Push the state into the sarea and/or texture memory.
  */
-static void i915_emit_state( intelContextPtr intel )
+static void
+i915_emit_state(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
    struct i915_hw_state *state = i915->current;
-   int i;
-   GLuint dirty = get_dirty(state);
-   GLuint counter = intel->batch.counter;
+   int i, count, aper_count;
+   GLuint dirty;
+   dri_bo *aper_array[3 + I915_TEX_UNITS];
+   GET_CURRENT_CONTEXT(ctx);
    BATCH_LOCALS;
 
-   if (intel->batch.space < get_state_size(state)) {
-      intelFlushBatch(intel, GL_TRUE);
-      dirty = get_dirty(state);
-      counter = intel->batch.counter;
+   /* We don't hold the lock at this point, so want to make sure that
+    * there won't be a buffer wrap between the state emits and the primitive
+    * emit header.
+    *
+    * It might be better to talk about explicit places where
+    * scheduling is allowed, rather than assume that it is whenever a
+    * batchbuffer fills up.
+    *
+    * Set the space as LOOP_CLIPRECTS now, since that's what our primitives
+    * will be emitted under.
+    */
+   intel_batchbuffer_require_space(intel->batch,
+				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE,
+				   LOOP_CLIPRECTS);
+   count = 0;
+ again:
+   aper_count = 0;
+   dirty = get_dirty(state);
+
+   aper_array[aper_count++] = intel->batch->buf;
+   if (dirty & I915_UPLOAD_BUFFERS) {
+      aper_array[aper_count++] = state->draw_region->buffer;
+      if (state->depth_region)
+	 aper_array[aper_count++] = state->depth_region->buffer;
+   }
+
+   if (dirty & I915_UPLOAD_TEX_ALL) {
+      for (i = 0; i < I915_TEX_UNITS; i++) {
+	 if (dirty & I915_UPLOAD_TEX(i)) {
+	    if (state->tex_buffer[i]) {
+	       aper_array[aper_count++] = state->tex_buffer[i];
+	    }
+	 }
+      }
+   }
+
+   if (dri_bufmgr_check_aperture_space(aper_array, aper_count)) {
+       if (count == 0) {
+	   count++;
+	   intel_batchbuffer_flush(intel->batch);
+	   goto again;
+       } else {
+	   _mesa_error(ctx, GL_OUT_OF_MEMORY, "i915 emit state");
+	   assert(0);
+       }
    }
 
-   if (VERBOSE) 
+   /* work out list of buffers to emit */
+   
+   /* Do this here as we may have flushed the batchbuffer above,
+    * causing more state to be dirty!
+    */
+   dirty = get_dirty(state);
+   state->emitted |= dirty;
+   assert(get_dirty(state) == 0);
+
+   if (INTEL_DEBUG & DEBUG_STATE)
       fprintf(stderr, "%s dirty: %x\n", __FUNCTION__, dirty);
 
    if (dirty & I915_UPLOAD_INVARIENT) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_INVARIENT:\n"); 
-      i915_emit_invarient_state( intel );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_INVARIENT:\n");
+      i915_emit_invarient_state(intel);
    }
 
    if (dirty & I915_UPLOAD_CTX) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_CTX:\n"); 
-      emit( i915, state->Ctx, sizeof(state->Ctx) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_CTX:\n");
+
+      emit(intel, state->Ctx, sizeof(state->Ctx));
    }
 
    if (dirty & I915_UPLOAD_BUFFERS) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_BUFFERS:\n"); 
-      emit( i915, state->Buffer, sizeof(state->Buffer) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_BUFFERS:\n");
+      BEGIN_BATCH(I915_DEST_SETUP_SIZE + 2, IGNORE_CLIPRECTS);
+      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR1]);
+      OUT_RELOC(state->draw_region->buffer,
+		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+                state->draw_region->draw_offset);
+
+      if (state->depth_region) {
+         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR0]);
+         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR1]);
+         OUT_RELOC(state->depth_region->buffer,
+		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+                   state->depth_region->draw_offset);
+      }
+
+      OUT_BATCH(state->Buffer[I915_DESTREG_DV0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DV1]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SENABLE]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR1]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR2]);
+
+      if (intel->constant_cliprect) {
+	 assert(state->Buffer[I915_DESTREG_DRAWRECT0] != MI_NOOP);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT0]);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT1]);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT2]);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT3]);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT4]);
+	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT5]);
+      }
+
+      ADVANCE_BATCH();
    }
 
    if (dirty & I915_UPLOAD_STIPPLE) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_STIPPLE:\n"); 
-      emit( i915, state->Stipple, sizeof(state->Stipple) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_STIPPLE:\n");
+      emit(intel, state->Stipple, sizeof(state->Stipple));
    }
 
    if (dirty & I915_UPLOAD_FOG) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_FOG:\n"); 
-      emit( i915, state->Fog, sizeof(state->Fog) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_FOG:\n");
+      emit(intel, state->Fog, sizeof(state->Fog));
    }
 
    /* Combine all the dirty texture state into a single command to
@@ -323,141 +431,245 @@ static void i915_emit_state( intelContextPtr intel )
    if (dirty & I915_UPLOAD_TEX_ALL) {
       int nr = 0;
 
-      for (i = 0; i < I915_TEX_UNITS; i++) 
-	 if (dirty & I915_UPLOAD_TEX(i)) 
-	    nr++;
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i))
+            nr++;
 
-      BEGIN_BATCH(2+nr*3);
-      OUT_BATCH(_3DSTATE_MAP_STATE | (3*nr));
+      BEGIN_BATCH(2 + nr * 3, IGNORE_CLIPRECTS);
+      OUT_BATCH(_3DSTATE_MAP_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0 ; i < I915_TEX_UNITS ; i++)
-	 if (dirty & I915_UPLOAD_TEX(i)) {
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS2]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS3]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS4]);
-	 }
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i)) {
+
+            if (state->tex_buffer[i]) {
+               OUT_RELOC(state->tex_buffer[i],
+			 I915_GEM_DOMAIN_SAMPLER, 0,
+                         state->tex_offset[i]);
+            }
+            else if (state == &i915->meta) {
+               assert(i == 0);
+               OUT_BATCH(0);
+            }
+            else {
+               OUT_BATCH(state->tex_offset[i]);
+            }
+
+            OUT_BATCH(state->Tex[i][I915_TEXREG_MS3]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_MS4]);
+         }
       ADVANCE_BATCH();
 
-      BEGIN_BATCH(2+nr*3);
-      OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3*nr));
+      BEGIN_BATCH(2 + nr * 3, IGNORE_CLIPRECTS);
+      OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0 ; i < I915_TEX_UNITS ; i++)
-	 if (dirty & I915_UPLOAD_TEX(i)) {
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS2]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS3]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS4]);
-	 }
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i)) {
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS2]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS3]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS4]);
+         }
       ADVANCE_BATCH();
    }
 
    if (dirty & I915_UPLOAD_CONSTANTS) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_CONSTANTS:\n"); 
-      emit( i915, state->Constant, state->ConstantSize * sizeof(GLuint) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_CONSTANTS:\n");
+      emit(intel, state->Constant, state->ConstantSize * sizeof(GLuint));
    }
 
    if (dirty & I915_UPLOAD_PROGRAM) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_PROGRAM:\n"); 
+      if (state->ProgramSize) {
+         if (INTEL_DEBUG & DEBUG_STATE)
+            fprintf(stderr, "I915_UPLOAD_PROGRAM:\n");
+
+         assert((state->Program[0] & 0x1ff) + 2 == state->ProgramSize);
 
-      assert((state->Program[0] & 0x1ff)+2 == state->ProgramSize);
-      
-      emit( i915, state->Program, state->ProgramSize * sizeof(GLuint) );
-      if (VERBOSE)
-	 i915_disassemble_program( state->Program, state->ProgramSize );
+         emit(intel, state->Program, state->ProgramSize * sizeof(GLuint));
+         if (INTEL_DEBUG & DEBUG_STATE)
+            i915_disassemble_program(state->Program, state->ProgramSize);
+      }
    }
 
-   state->emitted |= dirty;
-   intel->batch.last_emit_state = counter;
-   assert(counter == intel->batch.counter);
+   intel->batch->dirty_state &= ~dirty;
+   assert(get_dirty(state) == 0);
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
 }
 
-static void i915_destroy_context( intelContextPtr intel )
+static void
+i915_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
+   intel_region_release(&i915->state.draw_region);
+   intel_region_release(&i915->state.depth_region);
+   intel_region_release(&i915->meta.draw_region);
+   intel_region_release(&i915->meta.depth_region);
+   intel_region_release(&i915->initial.draw_region);
+   intel_region_release(&i915->initial.depth_region);
+
+   for (i = 0; i < I915_TEX_UNITS; i++) {
+      if (i915->state.tex_buffer[i] != NULL) {
+	 dri_bo_unreference(i915->state.tex_buffer[i]);
+	 i915->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
 
 /**
- * Set the color buffer drawing region.
+ * Set the drawing regions for the color and depth/stencil buffers.
+ * This involves setting the pitch, cpp and buffer ID/location.
+ * Also set pixel format for color and Z rendering
+ * Used for setting both regular and meta state.
  */
-static void
-i915_set_color_region( intelContextPtr intel, const intelRegion *region)
+void
+i915_state_draw_region(struct intel_context *intel,
+                       struct i915_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
-   I915_STATECHANGE( i915, I915_UPLOAD_BUFFERS );
-   i915->state.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_CBUFADDR2] = region->offset;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   GLcontext *ctx = &intel->ctx;
+   GLuint value;
+
+   ASSERT(state == &i915->state || state == &i915->meta);
+
+   if (state->draw_region != color_region) {
+      intel_region_release(&state->draw_region);
+      intel_region_reference(&state->draw_region, color_region);
+   }
+   if (state->depth_region != depth_region) {
+      intel_region_release(&state->depth_region);
+      intel_region_reference(&state->depth_region, depth_region);
+   }
+
+   /*
+    * Set stride/cpp values
+    */
+   if (color_region) {
+      state->Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I915_DESTREG_CBUFADDR1] =
+         (BUF_3D_ID_COLOR_BACK |
+          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   if (depth_region) {
+      state->Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I915_DESTREG_DBUFADDR1] =
+         (BUF_3D_ID_DEPTH |
+          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   /*
+    * Compute/set I915_DESTREG_DV1 value
+    */
+   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
+            DSTORG_VERT_BIAS(0x8) |     /* .5 */
+            LOD_PRECLAMP_OGL | TEX_DEFAULT_COLOR_OGL);
+   if (color_region && color_region->cpp == 4) {
+      value |= DV_PF_8888;
+   }
+   else {
+      value |= (DITHER_FULL_ALWAYS | DV_PF_565);
+   }
+   if (depth_region && depth_region->cpp == 4) {
+      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
+   }
+   else {
+      value |= DEPTH_FRMT_16_FIXED;
+   }
+   state->Buffer[I915_DESTREG_DV1] = value;
+
+   if (intel->constant_cliprect) {
+      state->Buffer[I915_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
+      state->Buffer[I915_DESTREG_DRAWRECT1] = 0;
+      state->Buffer[I915_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
+      state->Buffer[I915_DESTREG_DRAWRECT3] =
+	 (ctx->DrawBuffer->Width & 0xffff) |
+	 (ctx->DrawBuffer->Height << 16);
+      state->Buffer[I915_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
+      state->Buffer[I915_DESTREG_DRAWRECT5] = 0;
+   } else {
+      state->Buffer[I915_DESTREG_DRAWRECT0] = MI_NOOP;
+      state->Buffer[I915_DESTREG_DRAWRECT1] = MI_NOOP;
+      state->Buffer[I915_DESTREG_DRAWRECT2] = MI_NOOP;
+      state->Buffer[I915_DESTREG_DRAWRECT3] = MI_NOOP;
+      state->Buffer[I915_DESTREG_DRAWRECT4] = MI_NOOP;
+      state->Buffer[I915_DESTREG_DRAWRECT5] = MI_NOOP;
+   }
+
+   I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
 }
 
 
-/**
- * specify the z-buffer/stencil region
- */
 static void
-i915_set_z_region( intelContextPtr intel, const intelRegion *region)
+i915_set_draw_region(struct intel_context *intel,
+                     struct intel_region *color_regions[],
+                     struct intel_region *depth_region,
+		     GLuint num_regions)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
-   I915_STATECHANGE( i915, I915_UPLOAD_BUFFERS );
-   i915->state.Buffer[I915_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_DBUFADDR2] = region->offset;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   i915_state_draw_region(intel, &i915->state, color_regions[0], depth_region);
 }
 
 
-/**
- * Set both the color and Z/stencil drawing regions.
- * Similar to two previous functions, but don't use I915_STATECHANGE()
- */
+
 static void
-i915_update_color_z_regions(intelContextPtr intel,
-                            const intelRegion *colorRegion,
-                            const intelRegion *depthRegion)
+i915_new_batch(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
-   i915->state.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_CBUFADDR2] = colorRegion->offset;
+   /* Mark all state as needing to be emitted when starting a new batchbuffer.
+    * Using hardware contexts would be an alternative, but they have some
+    * difficulties associated with them (physical address requirements).
+    */
+   i915->state.emitted = 0;
 
-   i915->state.Buffer[I915_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH |
-       BUF_3D_PITCH(depthRegion->pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_DBUFADDR2] = depthRegion->offset;
+   /* Check that we didn't just wrap our batchbuffer at a bad time. */
+   assert(!intel->no_batch_wrap);
 }
 
-
-static void i915_lost_hardware( intelContextPtr intel )
+static GLuint
+i915_flush_cmd(void)
 {
-   I915_CONTEXT(intel)->state.emitted = 0;
+   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
-static void i915_emit_flush( intelContextPtr intel )
+static void 
+i915_assert_not_dirty( struct intel_context *intel )
 {
-   BATCH_LOCALS;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   struct i915_hw_state *state = i915->current;
+   GLuint dirty = get_dirty(state);
+   assert(!dirty);
+}
 
-   BEGIN_BATCH(2);
-   OUT_BATCH( MI_FLUSH | FLUSH_MAP_CACHE | FLUSH_RENDER_CACHE ); 
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
+static void
+i915_note_unlock( struct intel_context *intel )
+{
+    /* nothing */
 }
 
 
-void i915InitVtbl( i915ContextPtr i915 )
+void
+i915InitVtbl(struct i915_context *i915)
 {
-   i915->intel.vtbl.alloc_tex_obj = i915AllocTexObj;
    i915->intel.vtbl.check_vertex_size = i915_check_vertex_size;
-   i915->intel.vtbl.clear_with_tris = i915ClearWithTris;
-   i915->intel.vtbl.rotate_window = i915RotateWindow;
    i915->intel.vtbl.destroy = i915_destroy_context;
    i915->intel.vtbl.emit_state = i915_emit_state;
-   i915->intel.vtbl.lost_hardware = i915_lost_hardware;
+   i915->intel.vtbl.new_batch = i915_new_batch;
    i915->intel.vtbl.reduced_primitive_state = i915_reduced_primitive_state;
    i915->intel.vtbl.render_start = i915_render_start;
-   i915->intel.vtbl.set_color_region = i915_set_color_region;
-   i915->intel.vtbl.set_z_region = i915_set_z_region;
-   i915->intel.vtbl.update_color_z_regions = i915_update_color_z_regions;
+   i915->intel.vtbl.render_prevalidate = i915_render_prevalidate;
+   i915->intel.vtbl.set_draw_region = i915_set_draw_region;
    i915->intel.vtbl.update_texture_state = i915UpdateTextureState;
-   i915->intel.vtbl.emit_flush = i915_emit_flush;
+   i915->intel.vtbl.flush_cmd = i915_flush_cmd;
+   i915->intel.vtbl.assert_not_dirty = i915_assert_not_dirty;
+   i915->intel.vtbl.note_unlock = i915_note_unlock; 
+   i915->intel.vtbl.finish_batch = intel_finish_vb;
 }
-
diff --git a/src/mesa/drivers/dri/i915/intel_batchbuffer.c b/src/mesa/drivers/dri/i915/intel_batchbuffer.c
index 803b41b2567..d38cdf31cc6 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i915/intel_batchbuffer.c
@@ -1,829 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <errno.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "enums.h"
-#include "vblank.h"
-
-#include "intel_reg.h"
-#include "intel_batchbuffer.h"
-#include "intel_context.h"
-
-
-
-
-/* ================================================================
- * Performance monitoring functions
- */
-
-static void intel_fill_box( intelContextPtr intel,
-			    GLshort x, GLshort y,
-			    GLshort w, GLshort h,
-			    GLubyte r, GLubyte g, GLubyte b )
-{
-   x += intel->drawX;
-   y += intel->drawY;
-
-   if (x >= 0 && y >= 0 &&
-       x+w < intel->intelScreen->width &&
-       y+h < intel->intelScreen->height)
-      intelEmitFillBlitLocked( intel, 
-			       intel->intelScreen->cpp,
-			       intel->intelScreen->back.pitch,
-			       intel->intelScreen->back.offset,
-			       x, y, w, h,
-			       INTEL_PACKCOLOR(intel->intelScreen->fbFormat,
-					       r,g,b,0xff));
-}
-
-static void intel_draw_performance_boxes( intelContextPtr intel )
-{
-   /* Purple box for page flipping
-    */
-   if ( intel->perf_boxes & I830_BOX_FLIP ) 
-      intel_fill_box( intel, 4, 4, 8, 8, 255, 0, 255 );
-
-   /* Red box if we have to wait for idle at any point
-    */
-   if ( intel->perf_boxes & I830_BOX_WAIT ) 
-      intel_fill_box( intel, 16, 4, 8, 8, 255, 0, 0 );
-
-   /* Blue box: lost context?
-    */
-   if ( intel->perf_boxes & I830_BOX_LOST_CONTEXT ) 
-      intel_fill_box( intel, 28, 4, 8, 8, 0, 0, 255 );
-
-   /* Yellow box for texture swaps
-    */
-   if ( intel->perf_boxes & I830_BOX_TEXTURE_LOAD ) 
-      intel_fill_box( intel, 40, 4, 8, 8, 255, 255, 0 );
-
-   /* Green box if hardware never idles (as far as we can tell)
-    */
-   if ( !(intel->perf_boxes & I830_BOX_RING_EMPTY) ) 
-      intel_fill_box( intel, 64, 4, 8, 8, 0, 255, 0 );
-
-
-   /* Draw bars indicating number of buffers allocated 
-    * (not a great measure, easily confused)
-    */
-#if 0
-   if (intel->dma_used) {
-      int bar = intel->dma_used / 10240;
-      if (bar > 100) bar = 100;
-      if (bar < 1) bar = 1;
-      intel_fill_box( intel, 4, 16, bar, 4, 196, 128, 128 );
-      intel->dma_used = 0;
-   }
-#endif
-
-   intel->perf_boxes = 0;
-}
-
-
-
-
-
-
-static int bad_prim_vertex_nr( int primitive, int nr )
-{
-   switch (primitive & PRIM3D_MASK) {
-   case PRIM3D_POINTLIST:
-      return nr < 1;
-   case PRIM3D_LINELIST:
-      return (nr & 1) || nr == 0;
-   case PRIM3D_LINESTRIP:
-      return nr < 2;
-   case PRIM3D_TRILIST:
-   case PRIM3D_RECTLIST:
-      return nr % 3 || nr == 0;
-   case PRIM3D_POLY:
-   case PRIM3D_TRIFAN:
-   case PRIM3D_TRISTRIP:
-   case PRIM3D_TRISTRIP_RVRSE:
-      return nr < 3;
-   default:
-      return 1;
-   }	
-}
-
-static void intel_flush_inline_primitive( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   GLuint used = intel->batch.ptr - intel->prim.start_ptr;
-   GLuint vertcount;
-
-   assert(intel->prim.primitive != ~0);
-
-   if (1) {
-      /* Check vertex size against the vertex we're specifying to
-       * hardware.  If it's wrong, ditch the primitive.
-       */ 
-      if (!intel->vtbl.check_vertex_size( intel, intel->vertex_size )) 
-	 goto do_discard;
-
-      vertcount = (used - 4)/ (intel->vertex_size * 4);
-
-      if (!vertcount)
-	 goto do_discard;
-      
-      if (vertcount * intel->vertex_size * 4 != used - 4) {
-	 fprintf(stderr, "vertex size confusion %d %d\n", used, 
-		 intel->vertex_size * vertcount * 4);
-	 goto do_discard;
-      }
-
-      if (bad_prim_vertex_nr( intel->prim.primitive, vertcount )) {
-	 fprintf(stderr, "bad_prim_vertex_nr %x %d\n", intel->prim.primitive,
-		 vertcount);
-	 goto do_discard;
-      }
-   }
-
-   if (used < 8)
-      goto do_discard;
-
-   *(int *)intel->prim.start_ptr = (_3DPRIMITIVE | 
-				    intel->prim.primitive |
-				    (used/4-2));
-
-   goto finished;
-   
- do_discard:
-   intel->batch.ptr -= used;
-   intel->batch.space += used;
-   assert(intel->batch.space >= 0);
-
- finished:
-   intel->prim.primitive = ~0;
-   intel->prim.start_ptr = 0;
-   intel->prim.flush = 0;
-}
-
-
-/* Emit a primitive referencing vertices in a vertex buffer.
- */
-void intelStartInlinePrimitive( intelContextPtr intel, GLuint prim )
-{
-   BATCH_LOCALS;
-
-   if (0)
-      fprintf(stderr, "%s %x\n", __FUNCTION__, prim);
-
-
-   /* Finish any in-progress primitive:
-    */
-   INTEL_FIREVERTICES( intel );
-   
-   /* Emit outstanding state:
-    */
-   intel->vtbl.emit_state( intel );
-   
-   /* Make sure there is some space in this buffer:
-    */
-   if (intel->vertex_size * 10 * sizeof(GLuint) >= intel->batch.space) {
-      intelFlushBatch(intel, GL_TRUE); 
-      intel->vtbl.emit_state( intel );
-   }
-
-#if 1
-   if (((unsigned long)intel->batch.ptr) & 0x4) {
-      BEGIN_BATCH(1);
-      OUT_BATCH(0);
-      ADVANCE_BATCH();
-   }
-#endif
-
-   /* Emit a slot which will be filled with the inline primitive
-    * command later.
-    */
-   BEGIN_BATCH(2);
-   OUT_BATCH( 0 );
-
-   intel->prim.start_ptr = batch_ptr;
-   intel->prim.primitive = prim;
-   intel->prim.flush = intel_flush_inline_primitive;
-   intel->batch.contains_geometry = 1;
-
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
-}
-
-
-void intelRestartInlinePrimitive( intelContextPtr intel )
-{
-   GLuint prim = intel->prim.primitive;
-
-   intel_flush_inline_primitive( &intel->ctx );
-   if (1) intelFlushBatch(intel, GL_TRUE); /* GL_TRUE - is critical */
-   intelStartInlinePrimitive( intel, prim );
-}
-
-
-
-void intelWrapInlinePrimitive( intelContextPtr intel )
-{
-   GLuint prim = intel->prim.primitive;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-   intel_flush_inline_primitive( &intel->ctx );
-   intelFlushBatch(intel, GL_TRUE);
-   intelStartInlinePrimitive( intel, prim );
-}
-
-
-/* Emit a primitive with space for inline vertices.
- */
-GLuint *intelEmitInlinePrimitiveLocked(intelContextPtr intel, 
-				       int primitive,
-				       int dwords,
-				       int vertex_size )
-{
-   GLuint *tmp = 0;
-   BATCH_LOCALS;
-
-   if (0)
-      fprintf(stderr, "%s 0x%x %d\n", __FUNCTION__, primitive, dwords);
-
-   /* Emit outstanding state:
-    */
-   intel->vtbl.emit_state( intel );
-
-   if ((1+dwords)*4 >= intel->batch.space) {
-      intelFlushBatch(intel, GL_TRUE); 
-      intel->vtbl.emit_state( intel );
-   }
-
-
-   if (1) {
-      int used = dwords * 4;
-      int vertcount;
-
-      /* Check vertex size against the vertex we're specifying to
-       * hardware.  If it's wrong, ditch the primitive.
-       */ 
-      if (!intel->vtbl.check_vertex_size( intel, vertex_size )) 
-	 goto do_discard;
-
-      vertcount = dwords / vertex_size;
-      
-      if (dwords % vertex_size) {
-	 fprintf(stderr, "did not request a whole number of vertices\n");
-	 goto do_discard;
-      }
-
-      if (bad_prim_vertex_nr( primitive, vertcount )) {
-	 fprintf(stderr, "bad_prim_vertex_nr %x %d\n", primitive, vertcount);
-	 goto do_discard;
-      }
-
-      if (used < 8)
-	 goto do_discard;
-   }
-
-   /* Emit 3D_PRIMITIVE commands:
-    */
-   BEGIN_BATCH(1 + dwords);
-   OUT_BATCH( _3DPRIMITIVE | 
-	      primitive |
-	      (dwords-1) );
-
-   tmp = (GLuint *)batch_ptr;
-   batch_ptr += dwords * 4;
-
-   ADVANCE_BATCH();
-
-   intel->batch.contains_geometry = 1;
-
- do_discard:
-   return tmp;
-}
-
-
-static void intelWaitForFrameCompletion( intelContextPtr intel )
-{
-  drm_i915_sarea_t *sarea = (drm_i915_sarea_t *)intel->sarea;
-
-   if (intel->do_irqs) {
-      if (intelGetLastFrame(intel) < sarea->last_dispatch) {
-	 if (!intel->irqsEmitted) {
-	    while (intelGetLastFrame (intel) < sarea->last_dispatch)
-	       ;
-	 }
-	 else {
-	    intelWaitIrq( intel, intel->alloc.irq_emitted );	
-	 }
-	 intel->irqsEmitted = 10;
-      }
-
-      if (intel->irqsEmitted) {
-	 LOCK_HARDWARE( intel ); 
-	 intelEmitIrqLocked( intel );
-	 intel->irqsEmitted--;
-	 UNLOCK_HARDWARE( intel ); 
-      }
-   } 
-   else {
-      while (intelGetLastFrame (intel) < sarea->last_dispatch) {
-	 if (intel->do_usleeps) 
-	    DO_USLEEP( 1 );
-      }
-   }
-}
-
-/*
- * Copy the back buffer to the front buffer. 
- */
-void intelCopyBuffer( const __DRIdrawablePrivate *dPriv,
-		      const drm_clip_rect_t	 *rect)
-{
-   intelContextPtr intel;
-   const intelScreenPrivate *intelScreen;
-   GLboolean   missed_target;
-   int64_t ust;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-
-   intelFlush( &intel->ctx );
-   
-   intelScreen = intel->intelScreen;
-
-   if (!rect && !intel->swap_scheduled && intelScreen->drmMinor >= 6 &&
-       !(intel->vblank_flags & VBLANK_FLAG_NO_IRQ) &&
-       intelScreen->current_rotation == 0) {
-      unsigned int interval = driGetVBlankInterval(dPriv, intel->vblank_flags);
-      unsigned int target;
-      drm_i915_vblank_swap_t swap;
-
-      swap.drawable = dPriv->hHWDrawable;
-      swap.seqtype = DRM_VBLANK_ABSOLUTE;
-      target = swap.sequence = intel->vbl_seq + interval;
-
-      if (intel->vblank_flags & VBLANK_FLAG_SYNC) {
-	 swap.seqtype |= DRM_VBLANK_NEXTONMISS;
-      } else if (interval == 0) {
-	 goto noschedule;
-      }
-
-      if ( intel->vblank_flags & VBLANK_FLAG_SECONDARY ) {
-	 swap.seqtype |= DRM_VBLANK_SECONDARY;
-      }
-
-      if (!drmCommandWriteRead(intel->driFd, DRM_I915_VBLANK_SWAP, &swap,
-                              sizeof(swap))) {
-        intel->swap_scheduled = 1;
-        intel->vbl_seq = swap.sequence;
-        swap.sequence -= target;
-        missed_target = swap.sequence > 0 && swap.sequence <= (1 << 23);
-      }
-   } else {
-      intel->swap_scheduled = 0;
-   }
-noschedule:
-
-   if (!intel->swap_scheduled) {
-      intelWaitForFrameCompletion( intel );
-      LOCK_HARDWARE( intel );
-
-      if (!rect)
-      {
-	 UNLOCK_HARDWARE( intel );
-	 driWaitForVBlank( dPriv, &intel->vbl_seq, intel->vblank_flags, & missed_target );
-	 LOCK_HARDWARE( intel );
-      }
-      {
-	 const intelScreenPrivate *intelScreen = intel->intelScreen;
-	 const __DRIdrawablePrivate *dPriv = intel->driDrawable;
-	 const int nbox = dPriv->numClipRects;
-	 const drm_clip_rect_t *pbox = dPriv->pClipRects;
-	 drm_clip_rect_t box;
-	 const int cpp = intelScreen->cpp;
-	 const int pitch = intelScreen->front.pitch; /* in bytes */
-	 int i;
-	 GLuint CMD, BR13;
-	 BATCH_LOCALS;
-
-	 switch(cpp) {
-	 case 2: 
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24);
-	    CMD = XY_SRC_COPY_BLT_CMD;
-	    break;
-	 case 4:
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24) | (1<<25);
-	    CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-		   XY_SRC_COPY_BLT_WRITE_RGB);
-	    break;
-	 default:
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24);
-	    CMD = XY_SRC_COPY_BLT_CMD;
-	    break;
-	 }
-   
-	 if (0) 
-	    intel_draw_performance_boxes( intel );
-
-	 for (i = 0 ; i < nbox; i++, pbox++) 
-	 {
-	    if (pbox->x1 > pbox->x2 ||
-		pbox->y1 > pbox->y2 ||
-		pbox->x2 > intelScreen->width ||
-		pbox->y2 > intelScreen->height) {
-	       _mesa_warning(&intel->ctx, "Bad cliprect in intelCopyBuffer()");
-	       continue;
-	    }
-
-	    box = *pbox;
-
-	    if (rect)
-	    {
-	       if (rect->x1 > box.x1)
-		  box.x1 = rect->x1;
-	       if (rect->y1 > box.y1)
-		  box.y1 = rect->y1;
-	       if (rect->x2 < box.x2)
-		  box.x2 = rect->x2;
-	       if (rect->y2 < box.y2)
-		  box.y2 = rect->y2;
-
-	       if (box.x1 > box.x2 || box.y1 > box.y2)
-		  continue;
-	    }
-
-	    BEGIN_BATCH( 8);
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (box.y1 << 16) | box.x1 );
-	    OUT_BATCH( (box.y2 << 16) | box.x2 );
-
-	    if (intel->sarea->pf_current_page == 0) 
-	       OUT_BATCH( intelScreen->front.offset );
-	    else
-	       OUT_BATCH( intelScreen->back.offset );			
-
-	    OUT_BATCH( (box.y1 << 16) | box.x1 );
-	    OUT_BATCH( BR13 & 0xffff );
-
-	    if (intel->sarea->pf_current_page == 0) 
-	       OUT_BATCH( intelScreen->back.offset );			
-	    else
-	       OUT_BATCH( intelScreen->front.offset );
-
-	    ADVANCE_BATCH();
-	 }
-      }
-      intelFlushBatchLocked( intel, GL_TRUE, GL_TRUE, GL_TRUE );
-      UNLOCK_HARDWARE( intel );
-   }
-
-   if (!rect)
-   {
-       intel->swap_count++;
-       (*dri_interface->getUST)(&ust);
-       if (missed_target) {
-	   intel->swap_missed_count++;
-	   intel->swap_missed_ust = ust -  intel->swap_ust;
-       }
-   
-       intel->swap_ust = ust;
-   }
-}
-
-
-
-
-void intelEmitFillBlitLocked( intelContextPtr intel,
-			      GLuint cpp,
-			      GLshort dst_pitch,  /* in bytes */
-			      GLuint dst_offset,
-			      GLshort x, GLshort y, 
-			      GLshort w, GLshort h,
-			      GLuint color )
-{
-   GLuint BR13, CMD;
-   BATCH_LOCALS;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = dst_pitch | (0xF0 << 16) | (1<<24);
-      CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = dst_pitch | (0xF0 << 16) | (1<<24) | (1<<25);
-      CMD = (XY_COLOR_BLT_CMD | XY_COLOR_BLT_WRITE_ALPHA |
-	     XY_COLOR_BLT_WRITE_RGB);
-      break;
-   default:
-      return;
-   }
-
-   BEGIN_BATCH( 6);
-   OUT_BATCH( CMD );
-   OUT_BATCH( BR13 );
-   OUT_BATCH( (y << 16) | x );
-   OUT_BATCH( ((y+h) << 16) | (x+w) );
-   OUT_BATCH( dst_offset );
-   OUT_BATCH( color );
-   ADVANCE_BATCH();
-}
-
-
-/* Copy BitBlt
- */
-void intelEmitCopyBlitLocked( intelContextPtr intel,
-			      GLuint cpp,
-			      GLshort src_pitch,
-			      GLuint  src_offset,
-			      GLshort dst_pitch,
-			      GLuint  dst_offset,
-			      GLshort src_x, GLshort src_y,
-			      GLshort dst_x, GLshort dst_y,
-			      GLshort w, GLshort h )
-{
-   GLuint CMD, BR13;
-   int dst_y2 = dst_y + h;
-   int dst_x2 = dst_x + w;
-   BATCH_LOCALS;
-
-   src_pitch *= cpp;
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = dst_pitch | (0xCC << 16) | (1<<24);
-      CMD = XY_SRC_COPY_BLT_CMD;
-      break;
-   case 4:
-      BR13 = dst_pitch | (0xCC << 16) | (1<<24) | (1<<25);
-      CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-	     XY_SRC_COPY_BLT_WRITE_RGB);
-      break;
-   default:
-      return;
-   }
-
-   if (dst_y2 < dst_y ||
-       dst_x2 < dst_x) {
-      return;
-   }
-
-   BEGIN_BATCH( 12);
-   OUT_BATCH( CMD );
-   OUT_BATCH( BR13 );
-   OUT_BATCH( (dst_y << 16) | dst_x );
-   OUT_BATCH( (dst_y2 << 16) | dst_x2 );
-   OUT_BATCH( dst_offset );	
-   OUT_BATCH( (src_y << 16) | src_x );
-   OUT_BATCH( src_pitch );
-   OUT_BATCH( src_offset ); 
-   ADVANCE_BATCH();
-}
-
-
-
-void intelClearWithBlit(GLcontext *ctx, GLbitfield buffers, GLboolean allFoo,
-                        GLint cx1Foo, GLint cy1Foo, GLint cwFoo, GLint chFoo)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   GLuint clear_depth, clear_color;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-   GLint pitch;
-   GLint cpp = intelScreen->cpp;
-   GLint i;
-   GLuint BR13, CMD, D_CMD;
-   BATCH_LOCALS;
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   pitch = intelScreen->front.pitch;
-
-   clear_color = intel->ClearColor;
-   clear_depth = 0;
-
-   if (buffers & BUFFER_BIT_DEPTH) {
-      clear_depth = (GLuint)(ctx->Depth.Clear * intel->ClearDepth);
-   }
-
-   if (buffers & BUFFER_BIT_STENCIL) {
-      clear_depth |= (ctx->Stencil.Clear & 0xff) << 24;
-   }
-
-   switch(cpp) {
-   case 2: 
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24);
-      D_CMD = CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24) | (1<<25);
-      CMD = (XY_COLOR_BLT_CMD |
-	     XY_COLOR_BLT_WRITE_ALPHA | 
-	     XY_COLOR_BLT_WRITE_RGB);
-      D_CMD = XY_COLOR_BLT_CMD;
-      if (buffers & BUFFER_BIT_DEPTH) D_CMD |= XY_COLOR_BLT_WRITE_RGB;
-      if (buffers & BUFFER_BIT_STENCIL) D_CMD |= XY_COLOR_BLT_WRITE_ALPHA;
-      break;
-   default:
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24);
-      D_CMD = CMD = XY_COLOR_BLT_CMD;
-      break;
-   }
-
-   {
-      /* flip top to bottom */
-      cy = intel->driDrawable->h - cy - ch;
-      cx = cx + intel->drawX;
-      cy += intel->drawY;
-
-      /* adjust for page flipping */
-      if ( intel->sarea->pf_current_page == 1 ) {
-	 GLuint tmp = buffers;
-
-	 buffers &= ~(BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT);
-	 if ( tmp & BUFFER_BIT_FRONT_LEFT ) buffers |= BUFFER_BIT_BACK_LEFT;
-	 if ( tmp & BUFFER_BIT_BACK_LEFT )  buffers |= BUFFER_BIT_FRONT_LEFT;
-      }
-
-      for (i = 0 ; i < intel->numClipRects ; i++) 
-      { 	 
-	 drm_clip_rect_t *box = &intel->pClipRects[i];	 
-	 drm_clip_rect_t b;
-
-	 if (!all) {
-	    GLint x = box->x1;
-	    GLint y = box->y1;
-	    GLint w = box->x2 - x;
-	    GLint h = box->y2 - y;
-
-	    if (x < cx) w -= cx - x, x = cx; 
-	    if (y < cy) h -= cy - y, y = cy;
-	    if (x + w > cx + cw) w = cx + cw - x;
-	    if (y + h > cy + ch) h = cy + ch - y;
-	    if (w <= 0) continue;
-	    if (h <= 0) continue;
-
-	    b.x1 = x;
-	    b.y1 = y;
-	    b.x2 = x + w;
-	    b.y2 = y + h;      
-	 } else {
-	    b = *box;
-	 }
-
-
-	 if (b.x1 > b.x2 ||
-	     b.y1 > b.y2 ||
-	     b.x2 > intelScreen->width ||
-	     b.y2 > intelScreen->height)
-	    continue;
-
-	 if ( buffers & BUFFER_BIT_FRONT_LEFT ) {	    
-	    BEGIN_BATCH( 6);	    
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->front.offset );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
-
-	 if ( buffers & BUFFER_BIT_BACK_LEFT ) {
-	    BEGIN_BATCH( 6); 
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->back.offset );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
-
-	 if ( buffers & (BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH) ) {
-	    BEGIN_BATCH( 6);
-	    OUT_BATCH( D_CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->depth.offset );
-	    OUT_BATCH( clear_depth );
-	    ADVANCE_BATCH();
-	 }      
-      }
-   }
-   intelFlushBatchLocked( intel, GL_TRUE, GL_FALSE, GL_TRUE );
-   UNLOCK_HARDWARE( intel );
-}
-
-
-
-
-void intelDestroyBatchBuffer( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-
-   if (intel->alloc.offset) {
-      intelFreeAGP( intel, intel->alloc.ptr );
-      intel->alloc.ptr = NULL;
-      intel->alloc.offset = 0;
-   }
-   else if (intel->alloc.ptr) {
-      free(intel->alloc.ptr);
-      intel->alloc.ptr = NULL;
-   }
-
-   memset(&intel->batch, 0, sizeof(intel->batch));
-}
-
-
-void intelInitBatchBuffer( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-
-   /* This path isn't really safe with rotate:
-    */
-   if (getenv("INTEL_BATCH") && intel->intelScreen->allow_batchbuffer) {      
-      switch (intel->intelScreen->deviceID) {
-      case PCI_CHIP_I865_G:
-	 /* HW bug?  Seems to crash if batchbuffer crosses 4k boundary.
-	  */
-	 intel->alloc.size = 8 * 1024; 
-	 break;
-      default:
-	 /* This is the smallest amount of memory the kernel deals with.
-	  * We'd ideally like to make this smaller.
-	  */
-	 intel->alloc.size = 1 << intel->intelScreen->logTextureGranularity;
-	 break;
-      }
-
-      intel->alloc.ptr = intelAllocateAGP( intel, intel->alloc.size );
-      if (intel->alloc.ptr)
-	 intel->alloc.offset = 
-	    intelAgpOffsetFromVirtual( intel, intel->alloc.ptr );
-      else
-         intel->alloc.offset = 0; /* OK? */
-   }
-
-   /* The default is now to use a local buffer and pass that to the
-    * kernel.  This is also a fallback if allocation fails on the
-    * above path:
-    */
-   if (!intel->alloc.ptr) {
-      intel->alloc.size = 8 * 1024;
-      intel->alloc.ptr = malloc( intel->alloc.size );
-      intel->alloc.offset = 0;
-   }
-
-   assert(intel->alloc.ptr);
-}
+../intel/intel_batchbuffer.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_batchbuffer.h b/src/mesa/drivers/dri/i915/intel_batchbuffer.h
deleted file mode 100644
index 577d07137ff..00000000000
--- a/src/mesa/drivers/dri/i915/intel_batchbuffer.h
+++ /dev/null
@@ -1,126 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_BATCHBUFFER_H
-#define INTEL_BATCHBUFFER_H
-
-#include "intel_context.h"
-#include "intel_ioctl.h"
-
-
-#define BATCH_LOCALS	GLubyte *batch_ptr;
-
-/* #define VERBOSE 0 */
-#ifndef VERBOSE
-extern int VERBOSE;
-#endif
-
-
-#define BEGIN_BATCH(n)							\
-do {									\
-   if (VERBOSE) fprintf(stderr, 					\
-			"BEGIN_BATCH(%ld) in %s, %d dwords free\n",	\
-			((unsigned long)n), __FUNCTION__,		\
-			intel->batch.space/4);				\
-   if (intel->batch.space < (n)*4)					\
-      intelFlushBatch(intel, GL_TRUE);					\
-   if (intel->batch.space == intel->batch.size)	intel->batch.func = __FUNCTION__;			\
-   batch_ptr = intel->batch.ptr;					\
-} while (0)
-
-#define OUT_BATCH(n)					\
-do {							\
-   *(GLuint *)batch_ptr = (n);				\
-   if (VERBOSE) fprintf(stderr, " -- %08x at %s/%d\n", (n), __FILE__, __LINE__);	\
-   batch_ptr += 4;					\
-} while (0)
-
-#define ADVANCE_BATCH()						\
-do {								\
-   if (VERBOSE) fprintf(stderr, "ADVANCE_BATCH()\n");		\
-   intel->batch.space -= (batch_ptr - intel->batch.ptr);	\
-   intel->batch.ptr = batch_ptr;				\
-   assert(intel->batch.space >= 0);				\
-} while(0)
-
-extern void intelInitBatchBuffer( GLcontext *ctx );
-extern void intelDestroyBatchBuffer( GLcontext *ctx );
-
-extern void intelStartInlinePrimitive( intelContextPtr intel, GLuint prim );
-extern void intelWrapInlinePrimitive( intelContextPtr intel );
-extern void intelRestartInlinePrimitive( intelContextPtr intel );
-extern GLuint *intelEmitInlinePrimitiveLocked(intelContextPtr intel, 
-					      int primitive, int dwords,
-					      int vertex_size);
-extern void intelCopyBuffer( const __DRIdrawablePrivate *dpriv,
-			     const drm_clip_rect_t	*rect);
-extern void intelClearWithBlit(GLcontext *ctx, GLbitfield mask, GLboolean all,
-			     GLint cx1, GLint cy1, GLint cw, GLint ch);
-
-extern void intelEmitCopyBlitLocked( intelContextPtr intel,
-				     GLuint cpp,
-				     GLshort src_pitch,
-				     GLuint  src_offset,
-				     GLshort dst_pitch,
-				     GLuint  dst_offset,
-				     GLshort srcx, GLshort srcy,
-				     GLshort dstx, GLshort dsty,
-				     GLshort w, GLshort h );
-
-extern void intelEmitFillBlitLocked( intelContextPtr intel,
-				     GLuint cpp,
-				     GLshort dst_pitch,
-				     GLuint dst_offset,
-				     GLshort x, GLshort y, 
-				     GLshort w, GLshort h,
-				     GLuint color );
-
-
-
-
-static __inline GLuint *intelExtendInlinePrimitive( intelContextPtr intel, 
-						GLuint dwords )
-{
-   GLuint sz = dwords * sizeof(GLuint);
-   GLuint *ptr;
-
-   if (intel->batch.space < sz) {
-      intelWrapInlinePrimitive( intel );
-/*       assert(intel->batch.space >= sz); */
-   }
-
-/*    assert(intel->prim.primitive != ~0); */
-   ptr = (GLuint *)intel->batch.ptr;
-   intel->batch.ptr += sz;
-   intel->batch.space -= sz;
-
-   return ptr;
-}
-
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915/intel_blit.c b/src/mesa/drivers/dri/i915/intel_blit.c
new file mode 120000
index 00000000000..dd6c8d17c28
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_blit.c
@@ -0,0 +1 @@
+../intel/intel_blit.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_buffer_objects.c b/src/mesa/drivers/dri/i915/intel_buffer_objects.c
new file mode 120000
index 00000000000..e06dd3c8d3c
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_buffer_objects.c
@@ -0,0 +1 @@
+../intel/intel_buffer_objects.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_buffers.c b/src/mesa/drivers/dri/i915/intel_buffers.c
new file mode 120000
index 00000000000..c86daa49f47
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_buffers.c
@@ -0,0 +1 @@
+../intel/intel_buffers.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_context.c b/src/mesa/drivers/dri/i915/intel_context.c
index 11c23f24a1b..27a1cbb255e 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_context.c
+++ b/src/mesa/drivers/dri/i915/intel_context.c
@@ -1,776 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "matrix.h"
-#include "simple_list.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "imports.h"
-#include "points.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "vbo/vbo.h"
-
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vertex.h"
-
-#include "drivers/common/driverfuncs.h"
-
-#include "intel_screen.h"
-
-#include "i830_dri.h"
-#include "i830_common.h"
-
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_tris.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-
-#include "vblank.h"
-#include "utils.h"
-#include "xmlpool.h" /* for symbolic values of enum-type options */
-#ifndef INTEL_DEBUG
-int INTEL_DEBUG = (0);
-#endif
-
-#define need_GL_ARB_multisample
-#define need_GL_ARB_point_parameters
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
-#define need_GL_ARB_vertex_program
-#define need_GL_ARB_window_pos
-#define need_GL_EXT_blend_color
-#define need_GL_EXT_blend_equation_separate
-#define need_GL_EXT_blend_func_separate
-#define need_GL_EXT_blend_minmax
-#define need_GL_EXT_cull_vertex
-#define need_GL_EXT_fog_coord
-#define need_GL_EXT_multi_draw_arrays
-#define need_GL_EXT_secondary_color
-#define need_GL_NV_vertex_program
-#include "extension_helper.h"
-
-#ifndef VERBOSE
-int VERBOSE = 0;
-#endif
-
-#if DEBUG_LOCKING
-char *prevLockFile;
-int prevLockLine;
-#endif
-
-/***************************************
- * Mesa's Driver Functions
- ***************************************/
-
-#define DRIVER_DATE "20061017"
-
-const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
-{
-   const char * chipset;
-   static char buffer[128];
-
-   switch (name) {
-   case GL_VENDOR:
-      return (GLubyte *)"Tungsten Graphics, Inc";
-      break;
-      
-   case GL_RENDERER:
-      switch (INTEL_CONTEXT(ctx)->intelScreen->deviceID) {
-      case PCI_CHIP_845_G:
-	 chipset = "Intel(R) 845G"; break;
-      case PCI_CHIP_I830_M:
-	 chipset = "Intel(R) 830M"; break;
-      case PCI_CHIP_I855_GM:
-	 chipset = "Intel(R) 852GM/855GM"; break;
-      case PCI_CHIP_I865_G:
-	 chipset = "Intel(R) 865G"; break;
-      case PCI_CHIP_I915_G:
-	 chipset = "Intel(R) 915G"; break;
-      case PCI_CHIP_I915_GM:
-	 chipset = "Intel(R) 915GM"; break;
-      case PCI_CHIP_I945_G:
-	 chipset = "Intel(R) 945G"; break;
-      case PCI_CHIP_I945_GM:
-	 chipset = "Intel(R) 945GM"; break;
-      case PCI_CHIP_I945_GME:
-	 chipset = "Intel(R) 945GME"; break;
-      case PCI_CHIP_G33_G:
-	 chipset = "Intel(R) G33"; break;
-      case PCI_CHIP_Q35_G:
-	 chipset = "Intel(R) Q35"; break;
-      case PCI_CHIP_Q33_G:
-	 chipset = "Intel(R) Q33"; break;
-      default:
-	 chipset = "Unknown Intel Chipset"; break;
-      }
-
-      (void) driGetRendererString( buffer, chipset, DRIVER_DATE, 0 );
-      return (GLubyte *) buffer;
-
-   default:
-      return NULL;
-   }
-}
-
-
-/**
- * Extension strings exported by the intel driver.
- *
- * \note
- * It appears that ARB_texture_env_crossbar has "disappeared" compared to the
- * old i830-specific driver.
- */
-const struct dri_extension card_extensions[] =
-{
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
-    { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
-    { "GL_ARB_texture_border_clamp",       NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
-    { "GL_ARB_texture_cube_map",           NULL },
-    { "GL_ARB_texture_env_add",            NULL },
-    { "GL_ARB_texture_env_combine",        NULL },
-    { "GL_ARB_texture_env_dot3",           NULL },
-    { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_texture_rectangle",          NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
-    { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
-    { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
-    { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
-    { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
-    { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
-    { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
-    { "GL_EXT_blend_subtract",             NULL },
-    { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
-    { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
-    { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
-    { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
-    { "GL_EXT_stencil_wrap",               NULL },
-    { "GL_EXT_texture_edge_clamp",         NULL },
-    { "GL_EXT_texture_env_combine",        NULL },
-    { "GL_EXT_texture_env_dot3",           NULL },
-    { "GL_EXT_texture_filter_anisotropic", NULL },
-    { "GL_EXT_texture_lod_bias",           NULL },
-    { "GL_3DFX_texture_compression_FXT1",  NULL },
-    { "GL_APPLE_client_storage",           NULL },
-    { "GL_MESA_pack_invert",               NULL },
-    { "GL_MESA_ycbcr_texture",             NULL },
-    { "GL_NV_blend_square",                NULL },
-    { "GL_NV_vertex_program",              GL_NV_vertex_program_functions },
-    { "GL_NV_vertex_program1_1",           NULL },
-    { "GL_SGIS_generate_mipmap",           NULL },
-    { NULL,                                NULL }
-};
-
-extern const struct tnl_pipeline_stage _intel_render_stage;
-
-static const struct tnl_pipeline_stage *intel_pipeline[] = {
-   &_tnl_vertex_transform_stage,
-   &_tnl_vertex_cull_stage,
-   &_tnl_normal_transform_stage,
-   &_tnl_lighting_stage,
-   &_tnl_fog_coordinate_stage,
-   &_tnl_texgen_stage,
-   &_tnl_texture_transform_stage,
-   &_tnl_point_attenuation_stage,
-   &_tnl_vertex_program_stage,
-#if 1
-   &_intel_render_stage,     /* ADD: unclipped rastersetup-to-dma */
-#endif
-   &_tnl_render_stage,
-   0,
-};
-
-
-static const struct dri_debug_control debug_control[] =
-{
-    { "fall",  DEBUG_FALLBACKS },
-    { "tex",   DEBUG_TEXTURE },
-    { "ioctl", DEBUG_IOCTL },
-    { "prim",  DEBUG_PRIMS },
-    { "vert",  DEBUG_VERTS },
-    { "state", DEBUG_STATE },
-    { "verb",  DEBUG_VERBOSE },
-    { "dri",   DEBUG_DRI },
-    { "dma",   DEBUG_DMA },
-    { "san",   DEBUG_SANITY },
-    { "sync",  DEBUG_SYNC },
-    { "sleep", DEBUG_SLEEP },
-    { "pix",   DEBUG_PIXEL },
-    { NULL,    0 }
-};
-
-
-static void intelInvalidateState( GLcontext *ctx, GLuint new_state )
-{
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   _tnl_invalidate_vertex_state( ctx, new_state );
-   INTEL_CONTEXT(ctx)->NewGLState |= new_state;
-}
-
-
-void intelInitDriverFunctions( struct dd_function_table *functions )
-{
-   _mesa_init_driver_functions( functions );
-
-   functions->Clear = intelClear;
-   functions->Flush = intelglFlush;
-   functions->Finish = intelFinish;
-   functions->GetString = intelGetString;
-   functions->UpdateState = intelInvalidateState;
-
-   intelInitTextureFuncs( functions );
-   intelInitPixelFuncs( functions );
-   intelInitStateFuncs( functions );
-}
-
-static void intel_emit_invarient_state( GLcontext *ctx )
-{
-}
-
-
-
-GLboolean intelInitContext( intelContextPtr intel,
-			    const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate,
-			    struct dd_function_table *functions )
-{
-   GLcontext *ctx = &intel->ctx;
-   GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   drmI830Sarea *saPriv = (drmI830Sarea *)
-      (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
-   int fthrottle_mode;
-
-   if (!_mesa_initialize_context(&intel->ctx,
-				 mesaVis, shareCtx, 
-				 functions,
-				 (void*) intel))
-      return GL_FALSE;
-
-   driContextPriv->driverPrivate = intel;
-   intel->intelScreen = intelScreen;
-   intel->driScreen = sPriv;
-   intel->sarea = saPriv;
-
-
-   (void) memset( intel->texture_heaps, 0, sizeof( intel->texture_heaps ) );
-   make_empty_list( & intel->swapped );
-
-   driParseConfigFiles (&intel->optionCache, &intelScreen->optionCache,
-			intel->driScreen->myNum, "i915");
-
-   ctx->Const.MaxTextureMaxAnisotropy = 2.0;
-
-   ctx->Const.MinLineWidth = 1.0;
-   ctx->Const.MinLineWidthAA = 1.0;
-   ctx->Const.MaxLineWidth = 3.0;
-   ctx->Const.MaxLineWidthAA = 3.0;
-   ctx->Const.LineWidthGranularity = 1.0;
-
-   ctx->Const.MinPointSize = 1.0;
-   ctx->Const.MinPointSizeAA = 1.0;
-   ctx->Const.MaxPointSize = 255.0;
-   ctx->Const.MaxPointSizeAA = 3.0;
-   ctx->Const.PointSizeGranularity = 1.0;
-
-   /* reinitialize the context point state.
-    * It depend on constants in __GLcontextRec::Const
-    */
-   _mesa_init_point(ctx);
-
-   /* Initialize the software rasterizer and helper modules. */
-   _swrast_CreateContext( ctx );
-   _vbo_CreateContext( ctx );
-   _tnl_CreateContext( ctx );
-   _swsetup_CreateContext( ctx );
-
-   /* Install the customized pipeline: */
-   _tnl_destroy_pipeline( ctx );
-   _tnl_install_pipeline( ctx, intel_pipeline );
-
-   /* Configure swrast to match hardware characteristics: */
-   _swrast_allow_pixel_fog( ctx, GL_FALSE );
-   _swrast_allow_vertex_fog( ctx, GL_TRUE );
-
-   /* Dri stuff */
-   intel->hHWContext = driContextPriv->hHWContext;
-   intel->driFd = sPriv->fd;
-   intel->driHwLock = (drmLock *) &sPriv->pSAREA->lock;
-
-   intel->hw_stencil = mesaVis->stencilBits && mesaVis->depthBits == 24;
-   intel->hw_stipple = 1;
-
-   switch(mesaVis->depthBits) {
-   case 0:			/* what to do in this case? */
-   case 16:
-      intel->depth_scale = 1.0/0xffff;
-      intel->polygon_offset_scale = 1.0/0xffff;
-      intel->depth_clear_mask = ~0;
-      intel->ClearDepth = 0xffff;
-      break;
-   case 24:
-      intel->depth_scale = 1.0/0xffffff;
-      intel->polygon_offset_scale = 2.0/0xffffff; /* req'd to pass glean */
-      intel->depth_clear_mask = 0x00ffffff;
-      intel->stencil_clear_mask = 0xff000000;
-      intel->ClearDepth = 0x00ffffff;
-      break;
-   default:
-      assert(0); 
-      break;
-   }
-
-   /* Initialize swrast, tnl driver tables: */
-   intelInitSpanFuncs( ctx );
-   intelInitTriFuncs( ctx );
-
-
-   intel->RenderIndex = ~0;
-
-   fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
-   intel->iw.irq_seq = -1;
-   intel->irqsEmitted = 0;
-
-   intel->do_irqs = (intel->intelScreen->irq_active &&
-		     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
-
-   intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
-
-   intel->vblank_flags = (intel->intelScreen->irq_active != 0)
-       ? driGetDefaultVBlankFlags(&intel->optionCache) : VBLANK_FLAG_NO_IRQ;
-
-   (*dri_interface->getUST)(&intel->swap_ust);
-   _math_matrix_ctr (&intel->ViewportMatrix);
-
-   driInitExtensions( ctx, card_extensions, GL_TRUE );
-
-   if (intel->ctx.Mesa_DXTn) {
-     _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
-     _mesa_enable_extension( ctx, "GL_S3_s3tc" );
-   }
-   else if (driQueryOptionb (&intel->optionCache, "force_s3tc_enable")) {
-     _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
-   }
-
-/*    driInitTextureObjects( ctx, & intel->swapped, */
-/* 			  DRI_TEXMGR_DO_TEXTURE_1D | */
-/* 			  DRI_TEXMGR_DO_TEXTURE_2D |  */
-/* 			  DRI_TEXMGR_DO_TEXTURE_RECT ); */
-
-
-   intelInitBatchBuffer(&intel->ctx);
-   intel->prim.flush = intel_emit_invarient_state;
-   intel->prim.primitive = ~0;
-
-
-#if DO_DEBUG
-   INTEL_DEBUG  = driParseDebugString( getenv( "INTEL_DEBUG" ),
-				       debug_control );
-   INTEL_DEBUG |= driParseDebugString( getenv( "INTEL_DEBUG" ),
-				       debug_control );
-#endif
-
-#ifndef VERBOSE
-   if (getenv("INTEL_VERBOSE"))
-      VERBOSE=1;
-#endif
-
-   if (getenv("INTEL_NO_RAST") || 
-       getenv("INTEL_NO_RAST")) {
-      fprintf(stderr, "disabling 3D rasterization\n");
-      FALLBACK(intel, INTEL_FALLBACK_USER, 1); 
-   }
-
-   return GL_TRUE;
-}
-
-void intelDestroyContext(__DRIcontextPrivate *driContextPriv)
-{
-   intelContextPtr intel = (intelContextPtr) driContextPriv->driverPrivate;
-
-   assert(intel); /* should never be null */
-   if (intel) {
-      GLboolean   release_texture_heaps;
-
-      INTEL_FIREVERTICES( intel );
-
-      intel->vtbl.destroy( intel );
-
-      release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
-      _swsetup_DestroyContext (&intel->ctx);
-      _tnl_DestroyContext (&intel->ctx);
-      _vbo_DestroyContext (&intel->ctx);
-
-      _swrast_DestroyContext (&intel->ctx);
-      intel->Fallback = 0;	/* don't call _swrast_Flush later */
-
-      intelDestroyBatchBuffer(&intel->ctx);
-      
-
-      if ( release_texture_heaps ) {
-         /* This share group is about to go away, free our private
-          * texture object data.
-          */
-         int i;
-
-         for ( i = 0 ; i < intel->nr_heaps ; i++ ) {
-	    driDestroyTextureHeap( intel->texture_heaps[ i ] );
-	    intel->texture_heaps[ i ] = NULL;
-         }
-
-	 assert( is_empty_list( & intel->swapped ) );
-      }
-
-      /* free the Mesa context */
-      _mesa_destroy_context(&intel->ctx);
-   }
-}
-
-void intelSetFrontClipRects( intelContextPtr intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!dPriv) return;
-
-   intel->numClipRects = dPriv->numClipRects;
-   intel->pClipRects = dPriv->pClipRects;
-   intel->drawX = dPriv->x;
-   intel->drawY = dPriv->y;
-}
-
-
-void intelSetBackClipRects( intelContextPtr intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!dPriv) return;
-
-   if (intel->sarea->pf_enabled == 0 && dPriv->numBackClipRects == 0) {
-      intel->numClipRects = dPriv->numClipRects;
-      intel->pClipRects = dPriv->pClipRects;
-      intel->drawX = dPriv->x;
-      intel->drawY = dPriv->y;
-   } else {
-      intel->numClipRects = dPriv->numBackClipRects;
-      intel->pClipRects = dPriv->pBackClipRects;
-      intel->drawX = dPriv->backX;
-      intel->drawY = dPriv->backY;
-      
-      if (dPriv->numBackClipRects == 1 &&
-	  dPriv->x == dPriv->backX &&
-	  dPriv->y == dPriv->backY) {
-      
-	 /* Repeat the calculation of the back cliprect dimensions here
-	  * as early versions of dri.a in the Xserver are incorrect.  Try
-	  * very hard not to restrict future versions of dri.a which
-	  * might eg. allocate truly private back buffers.
-	  */
-	 int x1, y1;
-	 int x2, y2;
-	 
-	 x1 = dPriv->x;
-	 y1 = dPriv->y;      
-	 x2 = dPriv->x + dPriv->w;
-	 y2 = dPriv->y + dPriv->h;
-	 
-	 if (x1 < 0) x1 = 0;
-	 if (y1 < 0) y1 = 0;
-	 if (x2 > intel->intelScreen->width) x2 = intel->intelScreen->width;
-	 if (y2 > intel->intelScreen->height) y2 = intel->intelScreen->height;
-
-	 if (x1 == dPriv->pBackClipRects[0].x1 &&
-	     y1 == dPriv->pBackClipRects[0].y1) {
-
-	    dPriv->pBackClipRects[0].x2 = x2;
-	    dPriv->pBackClipRects[0].y2 = y2;
-	 }
-      }
-   }
-}
-
-
-void intelWindowMoved( intelContextPtr intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   GLframebuffer *drawFb = (GLframebuffer *) dPriv->driverPrivate;
-
-   if (!intel->ctx.DrawBuffer) {
-      intelSetFrontClipRects( intel );
-   }
-   else {
-      driUpdateFramebufferSize(&intel->ctx, dPriv);
-      switch (drawFb->_ColorDrawBufferMask[0]) {
-      case BUFFER_BIT_FRONT_LEFT:
-	 intelSetFrontClipRects( intel );
-	 break;
-      case BUFFER_BIT_BACK_LEFT:
-	 intelSetBackClipRects( intel );
-	 break;
-      default:
-	 /* glDrawBuffer(GL_NONE or GL_FRONT_AND_BACK): software fallback */
-	 intelSetFrontClipRects( intel );
-      }
-   }
-
-   if (drawFb->Width != dPriv->w || drawFb->Height != dPriv->h) {
-      /* update Mesa's notion of framebuffer/window size */
-      _mesa_resize_framebuffer(&intel->ctx, drawFb, dPriv->w, dPriv->h);
-      drawFb->Initialized = GL_TRUE; /* XXX remove someday */
-   }
-
-   /* Set state we know depends on drawable parameters:
-    */
-   {
-      GLcontext *ctx = &intel->ctx;
-
-      if (intel->intelScreen->driScrnPriv->ddxMinor >= 7) {
-	 drmI830Sarea *sarea = intel->sarea;
-	 drm_clip_rect_t drw_rect = { .x1 = dPriv->x, .x2 = dPriv->x + dPriv->w,
-				      .y1 = dPriv->y, .y2 = dPriv->y + dPriv->h };
-	 drm_clip_rect_t pipeA_rect = { .x1 = sarea->pipeA_x,
-					.x2 = sarea->pipeA_x + sarea->pipeA_w,
-					.y1 = sarea->pipeA_y,
-					.y2 = sarea->pipeA_y + sarea->pipeA_h };
-	 drm_clip_rect_t pipeB_rect = { .x1 = sarea->pipeB_x,
-					.x2 = sarea->pipeB_x + sarea->pipeB_w,
-					.y1 = sarea->pipeB_y,
-					.y2 = sarea->pipeB_y + sarea->pipeB_h };
-	 GLint areaA = driIntersectArea( drw_rect, pipeA_rect );
-	 GLint areaB = driIntersectArea( drw_rect, pipeB_rect );
-	 GLuint flags = intel->vblank_flags;
-
-	 if (areaB > areaA || (areaA == areaB && areaB > 0)) {
-	    flags = intel->vblank_flags | VBLANK_FLAG_SECONDARY;
-	 } else {
-	    flags = intel->vblank_flags & ~VBLANK_FLAG_SECONDARY;
-	 }
-
-	 if (flags != intel->vblank_flags) {
-	    intel->vblank_flags = flags;
-	    driGetCurrentVBlank(dPriv, intel->vblank_flags, &intel->vbl_seq);
-	 }
-      } else {
-	 intel->vblank_flags &= ~VBLANK_FLAG_SECONDARY;
-      }
-
-      ctx->Driver.Scissor( ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			   ctx->Scissor.Width, ctx->Scissor.Height );
-      
-      ctx->Driver.DepthRange( ctx, 
-			      ctx->Viewport.Near,
-			      ctx->Viewport.Far );
-   }
-}
-
-GLboolean intelUnbindContext(__DRIcontextPrivate *driContextPriv)
-{
-   return GL_TRUE;
-}
-
-GLboolean intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-			  __DRIdrawablePrivate *driDrawPriv,
-			  __DRIdrawablePrivate *driReadPriv)
-{
-
-   if (driContextPriv) {
-      intelContextPtr intel = (intelContextPtr) driContextPriv->driverPrivate;
-
-      if ( intel->driDrawable != driDrawPriv ) {
-	 /* Shouldn't the readbuffer be stored also? */
-	 driDrawableInitVBlank( driDrawPriv, intel->vblank_flags,
-				&intel->vbl_seq );
-
-	 intel->driDrawable = driDrawPriv;
-	 intelWindowMoved( intel );
-      }
-
-      _mesa_make_current(&intel->ctx,
-			 (GLframebuffer *) driDrawPriv->driverPrivate,
-			 (GLframebuffer *) driReadPriv->driverPrivate);
-
-      intel->ctx.Driver.DrawBuffer( &intel->ctx, intel->ctx.Color.DrawBuffer[0] );
-   } else {
-      _mesa_make_current(NULL, NULL, NULL);
-   }
-
-   return GL_TRUE;
-}
-
-/**
- * Use the information in the sarea to update the screen parameters
- * related to screen rotation.
- */
-static void
-intelUpdateScreenRotation(intelContextPtr intel,
-                          __DRIscreenPrivate *sPriv,
-                          drmI830Sarea *sarea)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   intelRegion *colorBuf;
-
-   intelUnmapScreenRegions(intelScreen);
-
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   /* update the current hw offsets for the color and depth buffers */
-   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT)
-      colorBuf = &intelScreen->back;
-   else
-      colorBuf = &intelScreen->front;
-   intel->vtbl.update_color_z_regions(intel, colorBuf, &intelScreen->depth);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr, "ERROR Remapping screen regions!!!\n");
-   }
-}
-
-void intelGetLock( intelContextPtr intel, GLuint flags )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   __DRIscreenPrivate *sPriv = intel->driScreen;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   drmI830Sarea * sarea = intel->sarea;
-   unsigned   i;
-
-   drmGetLock(intel->driFd, intel->hHWContext, flags);
-
-   /* If the window moved, may need to set a new cliprect now.
-    *
-    * NOTE: This releases and regains the hw lock, so all state
-    * checking must be done *after* this call:
-    */
-   if (dPriv)
-      DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
-
-   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
-      intelWindowMoved( intel );
-      intel->lastStamp = dPriv->lastStamp;
-   }
-
-   /* If we lost context, need to dump all registers to hardware.
-    * Note that we don't care about 2d contexts, even if they perform
-    * accelerated commands, so the DRI locking in the X server is even
-    * more broken than usual.
-    */
-
-   if (sarea->width != intelScreen->width ||
-       sarea->height != intelScreen->height ||
-       sarea->rotation != intelScreen->current_rotation) {
-      intelUpdateScreenRotation(intel, sPriv, sarea);
-
-      /* This will drop the outstanding batchbuffer on the floor */
-      intel->batch.ptr -= (intel->batch.size - intel->batch.space);
-      intel->batch.space = intel->batch.size;
-      /* lose all primitives */
-      intel->prim.primitive = ~0;
-      intel->prim.start_ptr = 0;
-      intel->prim.flush = 0;
-      intel->vtbl.lost_hardware( intel ); 
-
-      intel->lastStamp = 0; /* force window update */
-
-      /* Release batch buffer
-       */
-      intelDestroyBatchBuffer(&intel->ctx);
-      intelInitBatchBuffer(&intel->ctx);
-      intel->prim.flush = intel_emit_invarient_state;
-
-      /* Still need to reset the global LRU?
-       */
-      intel_driReinitTextureHeap( intel->texture_heaps[0], intel->intelScreen->tex.size );
-   }
-
-   /* Shared texture managment - if another client has played with
-    * texture space, figure out which if any of our textures have been
-    * ejected, and update our global LRU.
-    */
-   for ( i = 0 ; i < intel->nr_heaps ; i++ ) {
-      DRI_AGE_TEXTURES( intel->texture_heaps[ i ] );
-   }
-}
-
-
-void intelSwapBuffers( __DRIdrawablePrivate *dPriv )
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      intelContextPtr intel;
-      GLcontext *ctx;
-      intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-      ctx = &intel->ctx;
-      if (ctx->Visual.doubleBufferMode) {
-         intelScreenPrivate *screen = intel->intelScreen;
-	 _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-	 if ( 0 /*intel->doPageFlip*/ ) { /* doPageFlip is never set !!! */
-	    intelPageFlip( dPriv );
-	 } else {
-	     intelCopyBuffer( dPriv, NULL );
-	 }
-         if (screen->current_rotation != 0) {
-            intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
-         }
-      }
-   } else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-void intelCopySubBuffer( __DRIdrawablePrivate *dPriv,
-			 int x, int y, int w, int h )
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      intelContextPtr intel;
-      GLcontext *ctx;
-      intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-      ctx = &intel->ctx;
-      if (ctx->Visual.doubleBufferMode) {
-	 drm_clip_rect_t rect;
-	 rect.x1 = x + dPriv->x;
-	 rect.y1 = (dPriv->h - y - h) + dPriv->y;
-	 rect.x2 = rect.x1 + w;
-	 rect.y2 = rect.y1 + h;
-	 _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-	 intelCopyBuffer( dPriv, &rect );
-      }
-   } else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
+../intel/intel_context.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_context.h b/src/mesa/drivers/dri/i915/intel_context.h
deleted file mode 100644
index 3b50107d73f..00000000000
--- a/src/mesa/drivers/dri/i915/intel_context.h
+++ /dev/null
@@ -1,563 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTELCONTEXT_INC
-#define INTELCONTEXT_INC
-
-
-
-#include "mtypes.h"
-#include "drm.h"
-#include "mm.h"
-#include "texmem.h"
-#include "vblank.h"
-
-#include "intel_screen.h"
-#include "i915_drm.h"
-#include "i830_common.h"
-#include "tnl/t_vertex.h"
-
-#define TAG(x) intel##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-#define DV_PF_555  (1<<8)
-#define DV_PF_565  (2<<8)
-#define DV_PF_8888 (3<<8)
-
-#define INTEL_CONTEXT(ctx)	((intelContextPtr)(ctx))
-
-typedef struct intel_context intelContext;
-typedef struct intel_context *intelContextPtr;
-typedef struct intel_texture_object *intelTextureObjectPtr;
-
-typedef void (*intel_tri_func)(intelContextPtr, intelVertex *, intelVertex *,
-							  intelVertex *);
-typedef void (*intel_line_func)(intelContextPtr, intelVertex *, intelVertex *);
-typedef void (*intel_point_func)(intelContextPtr, intelVertex *);
-
-#define INTEL_FALLBACK_DRAW_BUFFER	 0x1
-#define INTEL_FALLBACK_READ_BUFFER	 0x2
-#define INTEL_FALLBACK_USER		 0x4
-#define INTEL_FALLBACK_NO_BATCHBUFFER	 0x8
-#define INTEL_FALLBACK_NO_TEXMEM	 0x10
-#define INTEL_FALLBACK_RENDERMODE	 0x20
-
-extern void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode );
-#define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
-
-
-#define INTEL_TEX_MAXLEVELS 10
-
-
-struct intel_texture_object
-{
-   driTextureObject    base;	/* the parent class */
-
-   GLuint texelBytes;
-   GLuint age;
-   GLuint Pitch;
-   GLuint Height;
-   GLuint TextureOffset;
-   GLubyte *BufAddr;   
-
-   GLuint min_level;
-   GLuint max_level;
-   GLuint depth_pitch;
-
-   struct {
-      const struct gl_texture_image *image;
-      GLuint offset;       /* into BufAddr */
-      GLuint height;
-      GLuint internalFormat;
-   } image[6][INTEL_TEX_MAXLEVELS];
-
-   GLuint dirty;
-   GLuint firstLevel,lastLevel;
-};
-
-
-struct intel_context
-{
-   GLcontext ctx;		/* the parent class */
-
-   struct {
-      void (*destroy)( intelContextPtr intel ); 
-      void (*emit_state)( intelContextPtr intel );
-      void (*lost_hardware)( intelContextPtr intel );
-      void (*update_texture_state)( intelContextPtr intel );
-
-      void (*render_start)( intelContextPtr intel );
-      void (*set_color_region)( intelContextPtr intel, const intelRegion *reg );
-      void (*set_z_region)( intelContextPtr intel, const intelRegion *reg );
-      void (*update_color_z_regions)(intelContextPtr intel,
-                                     const intelRegion *colorRegion,
-                                     const intelRegion *depthRegion);
-      void (*emit_flush)( intelContextPtr intel );
-      void (*reduced_primitive_state)( intelContextPtr intel, GLenum rprim );
-
-      GLboolean (*check_vertex_size)( intelContextPtr intel, GLuint expected );
-
-      void (*clear_with_tris)( intelContextPtr intel, GLbitfield mask,
-			       GLboolean all, 
-			       GLint cx, GLint cy, GLint cw, GLint ch);
-
-      void (*rotate_window)( intelContextPtr intel,
-                             __DRIdrawablePrivate *dPriv, GLuint srcBuf);
-
-      intelTextureObjectPtr (*alloc_tex_obj)( struct gl_texture_object *tObj );
-
-   } vtbl;
-
-   GLint refcount;   
-   GLuint Fallback;
-   GLuint NewGLState;
-   
-   struct {
-      GLuint start_offset;
-      GLint size;
-      GLint space;
-      GLubyte *ptr;
-      GLuint counter;
-      GLuint last_emit_state;
-      GLboolean contains_geometry;
-      const char *func;
-      GLuint last_swap;
-   } batch;
-      
-   struct {
-      void *ptr;
-      GLint size;
-      GLuint offset;
-      GLuint active_buf;
-      GLuint irq_emitted;
-   } alloc;
-
-   struct {
-      GLuint primitive;
-      GLubyte *start_ptr;      
-      void (*flush)( GLcontext * );
-   } prim;
-
-   GLboolean locked;
-
-   GLubyte clear_red;
-   GLubyte clear_green;
-   GLubyte clear_blue;
-   GLubyte clear_alpha;
-   GLuint ClearColor;
-   GLuint ClearDepth;
-
-   GLuint coloroffset;
-   GLuint specoffset;
-
-   /* Support for duplicating XYZW as WPOS parameter (crutch for I915).
-    */
-   GLuint wpos_offset;
-   GLuint wpos_size;
-
-   struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
-   GLuint vertex_attr_count;
-
-   GLfloat depth_scale;
-   GLfloat polygon_offset_scale; /* dependent on depth_scale, bpp */
-   GLuint depth_clear_mask;
-   GLuint stencil_clear_mask;
-
-   GLboolean hw_stencil;
-   GLboolean hw_stipple;
-   
-   /* Texture object bookkeeping
-    */
-   GLuint                nr_heaps;
-   driTexHeap          * texture_heaps[1];
-   driTextureObject      swapped;
-   GLuint                lastStamp;
-
-   struct intel_texture_object *CurrentTexObj[MAX_TEXTURE_UNITS];
-
-   /* State for intelvb.c and inteltris.c.
-    */
-   GLuint RenderIndex;
-   GLmatrix ViewportMatrix;
-   GLenum render_primitive;
-   GLenum reduced_primitive;
-   GLuint vertex_size;
-   unsigned char *verts;	   /* points to tnl->clipspace.vertex_buf */
-
-
-   /* Fallback rasterization functions 
-    */
-   intel_point_func draw_point;
-   intel_line_func draw_line;
-   intel_tri_func draw_tri;
-
-   /* Drawing buffer state
-    */
-   intelRegion *drawRegion;  /* current drawing buffer */
-   intelRegion *readRegion;  /* current reading buffer */
-
-   int drawX;			/* origin of drawable in draw buffer */
-   int drawY;
-   GLuint numClipRects;		/* cliprects for that buffer */
-   drm_clip_rect_t *pClipRects;
-
-   int dirtyAge;
-   int perf_boxes;
-
-   GLuint do_usleeps;
-   int do_irqs;
-   GLuint irqsEmitted;
-   drm_i915_irq_wait_t iw;
-
-   GLboolean scissor;
-   drm_clip_rect_t draw_rect;
-   drm_clip_rect_t scissor_rect;
-
-   drm_context_t hHWContext;
-   drmLock *driHwLock;
-   int driFd;
-
-   __DRIdrawablePrivate *driDrawable;
-   __DRIscreenPrivate *driScreen;
-   intelScreenPrivate *intelScreen; 
-   drmI830Sarea *sarea; 
-
-   /**
-    * Configuration cache
-    */
-   driOptionCache optionCache;
-
-   /* VBI
-    */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
-   int64_t swap_ust;
-   int64_t swap_missed_ust;
-
-   GLuint swap_count;
-   GLuint swap_missed_count;
-
-   GLuint swap_scheduled;
-};
-
-
-#define DEBUG_LOCKING	1
-
-#if DEBUG_LOCKING
-extern char *prevLockFile;
-extern int prevLockLine;
-
-#define DEBUG_LOCK()							\
-   do {									\
-      prevLockFile = (__FILE__);					\
-      prevLockLine = (__LINE__);					\
-   } while (0)
-
-#define DEBUG_RESET()							\
-   do {									\
-      prevLockFile = 0;							\
-      prevLockLine = 0;							\
-   } while (0)
-
-/* Slightly less broken way of detecting recursive locking in a
- * threaded environment.  The right way to do this would be to make
- * prevLockFile, prevLockLine thread-local.
- *
- * This technique instead checks to see if the same context is
- * requesting the lock twice -- this will not catch application
- * breakages where the same context is active in two different threads
- * at once, but it will catch driver breakages (recursive locking) in
- * threaded apps.
- */
-#define DEBUG_CHECK_LOCK()						\
-   do {									\
-      if ( *((volatile int *)intel->driHwLock) == 			\
-	   (DRM_LOCK_HELD | intel->hHWContext) ) {			\
-	 fprintf( stderr,						\
-		  "LOCK SET!\n\tPrevious %s:%d\n\tCurrent: %s:%d\n",	\
-		  prevLockFile, prevLockLine, __FILE__, __LINE__ );	\
-	 abort();							\
-      }									\
-   } while (0)
-
-#else
-
-#define DEBUG_LOCK()
-#define DEBUG_RESET()
-#define DEBUG_CHECK_LOCK()
-
-#endif
-
-
-
-
-/* Lock the hardware and validate our state.  
- */
-#define LOCK_HARDWARE( intel )				\
-do {							\
-    char __ret=0;					\
-    DEBUG_CHECK_LOCK();					\
-    assert(!(intel)->locked);				\
-    if ((intel)->swap_scheduled) {			\
-        drmVBlank vbl;					\
-        vbl.request.type = DRM_VBLANK_ABSOLUTE;		\
-        if ((intel)->vblank_flags &			\
-            VBLANK_FLAG_SECONDARY) {			\
-            vbl.request.type |= DRM_VBLANK_SECONDARY;	\
-        }						\
-        vbl.request.sequence = (intel)->vbl_seq;	\
-        drmWaitVBlank((intel)->driFd, &vbl);		\
-        (intel)->swap_scheduled = 0;			\
-    }							\
-    DRM_CAS((intel)->driHwLock, (intel)->hHWContext,	\
-        (DRM_LOCK_HELD|(intel)->hHWContext), __ret);	\
-    if (__ret)						\
-        intelGetLock( (intel), 0 );			\
-      DEBUG_LOCK();					\
-    (intel)->locked = 1;				\
-}while (0)
- 
-  
-  /* Unlock the hardware using the global current context 
-   */
-#define UNLOCK_HARDWARE(intel)						\
-do {									\
-   intel->locked = 0;							\
-   if (0) { 								\
-      intel->perf_boxes |= intel->sarea->perf_boxes;  			\
-      intel->sarea->perf_boxes = 0;					\
-   }									\
-   DRM_UNLOCK((intel)->driFd, (intel)->driHwLock, (intel)->hHWContext);	\
-   DEBUG_RESET();							\
-} while (0)
-
-
-#define SUBPIXEL_X 0.125
-#define SUBPIXEL_Y 0.125
-
-#define INTEL_FIREVERTICES(intel)		\
-do {						\
-   if ((intel)->prim.flush)			\
-      (intel)->prim.flush(&(intel)->ctx);		\
-} while (0)
-
-/* ================================================================
- * Color packing:
- */
-
-#define INTEL_PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define INTEL_PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define INTEL_PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define INTEL_PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
-(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
- (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
-  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
-   0)))
-
-
-
-/* ================================================================
- * From linux kernel i386 header files, copes with odd sizes better
- * than COPY_DWORDS would:
- */
-#if defined(i386) || defined(__i386__)
-static __inline__ void * __memcpy(void * to, const void * from, size_t n)
-{
-   int d0, d1, d2;
-   __asm__ __volatile__(
-      "rep ; movsl\n\t"
-      "testb $2,%b4\n\t"
-      "je 1f\n\t"
-      "movsw\n"
-      "1:\ttestb $1,%b4\n\t"
-      "je 2f\n\t"
-      "movsb\n"
-      "2:"
-      : "=&c" (d0), "=&D" (d1), "=&S" (d2)
-      :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
-      : "memory");
-   return (to);
-}
-#else
-#define __memcpy(a,b,c) memcpy(a,b,c)
-#endif
-
-
-
-/* ================================================================
- * Debugging:
- */
-#define DO_DEBUG		1
-#if DO_DEBUG
-extern int INTEL_DEBUG;
-#else
-#define INTEL_DEBUG		0
-#endif
-
-#define DEBUG_TEXTURE	0x1
-#define DEBUG_STATE	0x2
-#define DEBUG_IOCTL	0x4
-#define DEBUG_PRIMS	0x8
-#define DEBUG_VERTS	0x10
-#define DEBUG_FALLBACKS	0x20
-#define DEBUG_VERBOSE	0x40
-#define DEBUG_DRI       0x80
-#define DEBUG_DMA       0x100
-#define DEBUG_SANITY    0x200
-#define DEBUG_SYNC      0x400
-#define DEBUG_SLEEP     0x800
-#define DEBUG_PIXEL     0x1000
-
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-#define PCI_CHIP_I855_GM		0x3582
-#define PCI_CHIP_I865_G			0x2572
-#define PCI_CHIP_I915_G			0x2582
-#define PCI_CHIP_I915_GM		0x2592
-#define PCI_CHIP_I945_G			0x2772
-#define PCI_CHIP_I945_GM		0x27A2
-#define PCI_CHIP_I945_GME		0x27AE
-#define PCI_CHIP_G33_G			0x29C2
-#define PCI_CHIP_Q35_G			0x29B2
-#define PCI_CHIP_Q33_G			0x29D2
-
-
-/* ================================================================
- * intel_context.c:
- */
-
-extern void intelInitDriverFunctions( struct dd_function_table *functions );
-
-extern GLboolean intelInitContext( intelContextPtr intel, 
-				   const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate,
-				   struct dd_function_table *functions );
-
-extern void intelGetLock(intelContextPtr intel, GLuint flags);
-extern void intelSetBackClipRects(intelContextPtr intel);
-extern void intelSetFrontClipRects(intelContextPtr intel);
-extern void intelWindowMoved( intelContextPtr intel );
-
-extern const GLubyte *intelGetString( GLcontext *ctx, GLenum name );
-
-
-/* ================================================================
- * intel_state.c:
- */
-extern void intelInitStateFuncs( struct dd_function_table *functions );
-
-#define COMPAREFUNC_ALWAYS		0
-#define COMPAREFUNC_NEVER		0x1
-#define COMPAREFUNC_LESS		0x2
-#define COMPAREFUNC_EQUAL		0x3
-#define COMPAREFUNC_LEQUAL		0x4
-#define COMPAREFUNC_GREATER		0x5
-#define COMPAREFUNC_NOTEQUAL		0x6
-#define COMPAREFUNC_GEQUAL		0x7
-
-#define STENCILOP_KEEP			0
-#define STENCILOP_ZERO			0x1
-#define STENCILOP_REPLACE		0x2
-#define STENCILOP_INCRSAT		0x3
-#define STENCILOP_DECRSAT		0x4
-#define STENCILOP_INCR			0x5
-#define STENCILOP_DECR			0x6
-#define STENCILOP_INVERT		0x7
-
-#define LOGICOP_CLEAR			0
-#define LOGICOP_NOR			0x1
-#define LOGICOP_AND_INV 		0x2
-#define LOGICOP_COPY_INV		0x3
-#define LOGICOP_AND_RVRSE		0x4
-#define LOGICOP_INV			0x5
-#define LOGICOP_XOR			0x6
-#define LOGICOP_NAND			0x7
-#define LOGICOP_AND			0x8
-#define LOGICOP_EQUIV			0x9
-#define LOGICOP_NOOP			0xa
-#define LOGICOP_OR_INV			0xb
-#define LOGICOP_COPY			0xc
-#define LOGICOP_OR_RVRSE		0xd
-#define LOGICOP_OR			0xe
-#define LOGICOP_SET			0xf
-
-#define BLENDFACT_ZERO			0x01
-#define BLENDFACT_ONE			0x02
-#define BLENDFACT_SRC_COLR		0x03
-#define BLENDFACT_INV_SRC_COLR 		0x04
-#define BLENDFACT_SRC_ALPHA		0x05
-#define BLENDFACT_INV_SRC_ALPHA 	0x06
-#define BLENDFACT_DST_ALPHA		0x07
-#define BLENDFACT_INV_DST_ALPHA 	0x08
-#define BLENDFACT_DST_COLR		0x09
-#define BLENDFACT_INV_DST_COLR		0x0a
-#define BLENDFACT_SRC_ALPHA_SATURATE	0x0b
-#define BLENDFACT_CONST_COLOR		0x0c
-#define BLENDFACT_INV_CONST_COLOR	0x0d
-#define BLENDFACT_CONST_ALPHA		0x0e
-#define BLENDFACT_INV_CONST_ALPHA	0x0f
-#define BLENDFACT_MASK          	0x0f
-
-
-extern int intel_translate_compare_func( GLenum func );
-extern int intel_translate_stencil_op( GLenum op );
-extern int intel_translate_blend_factor( GLenum factor );
-extern int intel_translate_logic_op( GLenum opcode );
-
-
-/* ================================================================
- * intel_ioctl.c:
- */
-extern void intel_dump_batchbuffer( long offset,
-				    int *ptr,
-				    int count );
-
-
-/* ================================================================
- * intel_pixel.c:
- */	
-extern void intelInitPixelFuncs( struct dd_function_table *functions );
-
-
-
-#endif
-
diff --git a/src/mesa/drivers/dri/i915/intel_decode.c b/src/mesa/drivers/dri/i915/intel_decode.c
new file mode 120000
index 00000000000..f671b6cbb13
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_decode.c
@@ -0,0 +1 @@
+../intel/intel_decode.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_depthstencil.c b/src/mesa/drivers/dri/i915/intel_depthstencil.c
new file mode 120000
index 00000000000..4ac4ae690a3
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_depthstencil.c
@@ -0,0 +1 @@
+../intel/intel_depthstencil.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_fbo.c b/src/mesa/drivers/dri/i915/intel_fbo.c
new file mode 120000
index 00000000000..a19f86dcc57
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_fbo.c
@@ -0,0 +1 @@
+../intel/intel_fbo.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_ioctl.c b/src/mesa/drivers/dri/i915/intel_ioctl.c
deleted file mode 100644
index aa7c6d106ce..00000000000
--- a/src/mesa/drivers/dri/i915/intel_ioctl.c
+++ /dev/null
@@ -1,659 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <unistd.h>
-#include <errno.h>
-#include <sched.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "swrast/swrast.h"
-
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "drm.h"
-
-uint32_t intelGetLastFrame (intelContextPtr intel) 
-{
-   int ret;
-   uint32_t frame;
-   drm_i915_getparam_t gp;
-   
-   gp.param = I915_PARAM_LAST_DISPATCH;
-   gp.value = (int *)&frame;
-   ret = drmCommandWriteRead( intel->driFd, DRM_I915_GETPARAM,
-			      &gp, sizeof(gp) );
-   return frame;
-}
-
-int intelEmitIrqLocked( intelContextPtr intel )
-{
-   drmI830IrqEmit ie;
-   int ret, seq;
-      
-   assert(((*(int *)intel->driHwLock) & ~DRM_LOCK_CONT) == 
-	  (DRM_LOCK_HELD|intel->hHWContext));
-
-   ie.irq_seq = &seq;
-	 
-   ret = drmCommandWriteRead( intel->driFd, DRM_I830_IRQ_EMIT, 
-			      &ie, sizeof(ie) );
-   if ( ret ) {
-      fprintf( stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret );
-      exit(1);
-   }
-   
-   if (0)
-      fprintf(stderr, "%s -->  %d\n", __FUNCTION__, seq );
-
-   return seq;
-}
-
-void intelWaitIrq( intelContextPtr intel, int seq )
-{
-   int ret;
-      
-   if (0)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, seq );
-
-   intel->iw.irq_seq = seq;
-	 
-   do {
-     ret = drmCommandWrite( intel->driFd, DRM_I830_IRQ_WAIT, &intel->iw, sizeof(intel->iw) );
-   } while (ret == -EAGAIN || ret == -EINTR);
-
-   if ( ret ) {
-      fprintf( stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret );
-      if (0)
-	 intel_dump_batchbuffer( intel->alloc.offset,
-				 intel->alloc.ptr,
-				 intel->alloc.size );
-      exit(1);
-   }
-}
-
-
-
-static void age_intel( intelContextPtr intel, int age )
-{
-   GLuint i;
-
-   for (i = 0 ; i < MAX_TEXTURE_UNITS ; i++)
-      if (intel->CurrentTexObj[i]) 
-	 intel->CurrentTexObj[i]->age = age;
-}
-
-void intel_dump_batchbuffer( long offset,
-			     int *ptr,
-			     int count )
-{
-   int i;
-   fprintf(stderr, "\n\n\nSTART BATCH (%d dwords):\n", count);
-   for (i = 0; i < count/4; i += 4) 
-      fprintf(stderr, "\t0x%x: 0x%08x 0x%08x 0x%08x 0x%08x\n", 
-	      (unsigned int)offset + i*4, ptr[i], ptr[i+1], ptr[i+2], ptr[i+3]);
-   fprintf(stderr, "END BATCH\n\n\n");
-}
-
-void intelRefillBatchLocked( intelContextPtr intel, GLboolean allow_unlock )
-{
-   GLuint last_irq = intel->alloc.irq_emitted;
-   GLuint half = intel->alloc.size / 2;
-   GLuint buf = (intel->alloc.active_buf ^= 1);
-
-   intel->alloc.irq_emitted = intelEmitIrqLocked( intel );
-
-   if (last_irq) {
-      if (allow_unlock) UNLOCK_HARDWARE( intel ); 
-      intelWaitIrq( intel, last_irq );
-      if (allow_unlock) LOCK_HARDWARE( intel ); 
-   }
-
-   if (0)
-      fprintf(stderr, "%s: now using half %d\n", __FUNCTION__, buf);
-
-   intel->batch.start_offset = intel->alloc.offset + buf * half;
-   intel->batch.ptr = (unsigned char *)intel->alloc.ptr + buf * half;
-   intel->batch.size = half - 8;
-   intel->batch.space = half - 8;
-   assert(intel->batch.space >= 0);
-}
-
-#define MI_BATCH_BUFFER_END 	(0xA<<23)
-
-
-void intelFlushBatchLocked( intelContextPtr intel, 
-			    GLboolean ignore_cliprects,
-			    GLboolean refill,
-			    GLboolean allow_unlock)
-{
-   drmI830BatchBuffer batch;
-
-   assert(intel->locked);
-
-   if (0)
-      fprintf(stderr, "%s used %d of %d offset %x..%x refill %d (started in %s)\n",
-	      __FUNCTION__, 
-	      (intel->batch.size - intel->batch.space), 
-	      intel->batch.size,
-	      intel->batch.start_offset,
-	      intel->batch.start_offset + 
-	      (intel->batch.size - intel->batch.space), 
-	      refill,
-	      intel->batch.func);
-
-   /* Throw away non-effective packets.  Won't work once we have
-    * hardware contexts which would preserve statechanges beyond a
-    * single buffer.
-    */
-   if (intel->numClipRects == 0 && !ignore_cliprects) {
-      
-      /* Without this yeild, an application with no cliprects can hog
-       * the hardware.  Without unlocking, the effect is much worse -
-       * effectively a lock-out of other contexts.
-       */
-      if (allow_unlock) {
-	 UNLOCK_HARDWARE( intel );
-	 sched_yield();
-	 LOCK_HARDWARE( intel );
-      }
-
-      /* Note that any state thought to have been emitted actually
-       * hasn't:
-       */
-      intel->batch.ptr -= (intel->batch.size - intel->batch.space);
-      intel->batch.space = intel->batch.size;
-      intel->vtbl.lost_hardware( intel ); 
-   }
-
-   if (intel->batch.space != intel->batch.size) {
-
-      if (intel->sarea->ctxOwner != intel->hHWContext) {
-	 intel->perf_boxes |= I830_BOX_LOST_CONTEXT;
-	 intel->sarea->ctxOwner = intel->hHWContext;
-      }
-
-      batch.start = intel->batch.start_offset;
-      batch.used = intel->batch.size - intel->batch.space;
-      batch.cliprects = intel->pClipRects;
-      batch.num_cliprects = ignore_cliprects ? 0 : intel->numClipRects;
-      batch.DR1 = 0;
-      batch.DR4 = ((((GLuint)intel->drawX) & 0xffff) | 
-		   (((GLuint)intel->drawY) << 16));
-      
-      if (intel->alloc.offset) {
-	 if ((batch.used & 0x4) == 0) {
-	    ((int *)intel->batch.ptr)[0] = 0;
-	    ((int *)intel->batch.ptr)[1] = MI_BATCH_BUFFER_END;
-	    batch.used += 0x8;
-	    intel->batch.ptr += 0x8;
-	 }
-	 else {
-	    ((int *)intel->batch.ptr)[0] = MI_BATCH_BUFFER_END;
-	    batch.used += 0x4;
-	    intel->batch.ptr += 0x4;
-	 }      
-      }
-
-      if (0)
- 	 intel_dump_batchbuffer( batch.start,
-				 (int *)(intel->batch.ptr - batch.used),
-				 batch.used );
-
-      intel->batch.start_offset += batch.used;
-      intel->batch.size -= batch.used;
-
-      if (intel->batch.size < 8) {
-	 refill = GL_TRUE;
-	 intel->batch.space = intel->batch.size = 0;
-      }
-      else {
-	 intel->batch.size -= 8;
-	 intel->batch.space = intel->batch.size;
-      }
-
-
-      assert(intel->batch.space >= 0);
-      assert(batch.start >= intel->alloc.offset);
-      assert(batch.start < intel->alloc.offset + intel->alloc.size);
-      assert(batch.start + batch.used > intel->alloc.offset);
-      assert(batch.start + batch.used <= 
-	     intel->alloc.offset + intel->alloc.size);
-
-
-      if (intel->alloc.offset) {
-	 if (drmCommandWrite (intel->driFd, DRM_I830_BATCHBUFFER, &batch, 
-			      sizeof(batch))) {
-	    fprintf(stderr, "DRM_I830_BATCHBUFFER: %d\n",  -errno);
-	    UNLOCK_HARDWARE(intel);
-	    exit(1);
-	 }
-      } else {
-	 drmI830CmdBuffer cmd;
-	 cmd.buf = (char *)intel->alloc.ptr + batch.start;
-	 cmd.sz = batch.used;
-	 cmd.DR1 = batch.DR1;
-	 cmd.DR4 = batch.DR4;
-	 cmd.num_cliprects = batch.num_cliprects;
-	 cmd.cliprects = batch.cliprects;
-	 
-	 if (drmCommandWrite (intel->driFd, DRM_I830_CMDBUFFER, &cmd, 
-			      sizeof(cmd))) {
-	    fprintf(stderr, "DRM_I830_CMDBUFFER: %d\n",  -errno);
-	    UNLOCK_HARDWARE(intel);
-	    exit(1);
-	 }
-      }	 
-
-      
-      age_intel(intel, intel->sarea->last_enqueue);
-
-      /* FIXME: use hardware contexts to avoid 'losing' hardware after
-       * each buffer flush.
-       */
-      if (intel->batch.contains_geometry) 
-	 assert(intel->batch.last_emit_state == intel->batch.counter);
-
-      intel->batch.counter++;
-      intel->batch.contains_geometry = 0;
-      intel->batch.func = 0;
-      intel->vtbl.lost_hardware( intel );
-   }
-
-   if (refill)
-      intelRefillBatchLocked( intel, allow_unlock );
-}
-
-void intelFlushBatch( intelContextPtr intel, GLboolean refill )
-{
-   if (intel->locked) {
-      intelFlushBatchLocked( intel, GL_FALSE, refill, GL_FALSE );
-   } 
-   else {
-      LOCK_HARDWARE(intel);
-      intelFlushBatchLocked( intel, GL_FALSE, refill, GL_TRUE );
-      UNLOCK_HARDWARE(intel);
-   }
-}
-
-
-void intelWaitForIdle( intelContextPtr intel )
-{   
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   intel->vtbl.emit_flush( intel );
-   intelFlushBatch( intel, GL_TRUE );
-
-   /* Use an irq to wait for dma idle -- Need to track lost contexts
-    * to shortcircuit consecutive calls to this function:
-    */
-   intelWaitIrq( intel, intel->alloc.irq_emitted );
-   intel->alloc.irq_emitted = 0;
-}
-
-
-/**
- * Check if we need to rotate/warp the front color buffer to the
- * rotated screen.  We generally need to do this when we get a glFlush
- * or glFinish after drawing to the front color buffer.
- */
-static void
-intelCheckFrontRotate(GLcontext *ctx)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
-      intelScreenPrivate *screen = intel->intelScreen;
-      if (screen->current_rotation != 0) {
-         __DRIdrawablePrivate *dPriv = intel->driDrawable;
-         intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
-      }
-   }
-}
-
-
-/**
- * NOT directly called via glFlush.
- */
-void intelFlush( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-
-   if (intel->Fallback)
-      _swrast_flush( ctx );
-
-   INTEL_FIREVERTICES( intel );
-
-   if (intel->batch.size != intel->batch.space)
-      intelFlushBatch( intel, GL_FALSE );
-}
-
-
-/**
- * Called via glFlush.
- */
-void intelglFlush( GLcontext *ctx )
-{
-   intelFlush(ctx);
-   intelCheckFrontRotate(ctx);
-}
-
-
-void intelFinish( GLcontext *ctx  ) 
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   intelFlush( ctx );
-   intelWaitForIdle( intel );
-   intelCheckFrontRotate(ctx);
-}
-
-
-void intelClear(GLcontext *ctx, GLbitfield mask)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   const GLuint colorMask = *((GLuint *) &ctx->Color.ColorMask);
-   GLbitfield tri_mask = 0;
-   GLbitfield blit_mask = 0;
-   GLbitfield swrast_mask = 0;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Take care of cliprects, which are handled differently for
-    * clears, etc.
-    */
-   intelFlush( &intel->ctx );
-
-   if (mask & BUFFER_BIT_FRONT_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_FRONT_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_FRONT_LEFT;
-      }
-   }
-
-   if (mask & BUFFER_BIT_BACK_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_BACK_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_BACK_LEFT;
-      }
-   }
-
-   if (mask & BUFFER_BIT_DEPTH) {
-      blit_mask |= BUFFER_BIT_DEPTH;
-   }
-
-   if (mask & BUFFER_BIT_STENCIL) {
-      if (!intel->hw_stencil) {
-	 swrast_mask |= BUFFER_BIT_STENCIL;
-      }
-      else if ((ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
-	 tri_mask |= BUFFER_BIT_STENCIL;
-      } 
-      else {
-	 blit_mask |= BUFFER_BIT_STENCIL;
-      }
-   }
-
-   swrast_mask |= (mask & BUFFER_BIT_ACCUM);
-
-   if (blit_mask) 
-      intelClearWithBlit( ctx, blit_mask, 0, 0, 0, 0, 0);
-
-   if (tri_mask) 
-      intel->vtbl.clear_with_tris( intel, tri_mask, 0, 0, 0, 0, 0);
-
-   if (swrast_mask)
-      _swrast_Clear( ctx, swrast_mask );
-}
-
-
-void
-intelRotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                  GLuint srcBuffer)
-{
-   if (intel->vtbl.rotate_window) {
-      intel->vtbl.rotate_window(intel, dPriv, srcBuffer);
-   }
-}
-
-
-void *intelAllocateAGP( intelContextPtr intel, GLsizei size )
-{
-   int region_offset;
-   drmI830MemAlloc alloc;
-   int ret;
-
-   if (0)
-      fprintf(stderr, "%s: %d bytes\n", __FUNCTION__, size);
-
-   alloc.region = I830_MEM_REGION_AGP;
-   alloc.alignment = 0;
-   alloc.size = size;
-   alloc.region_offset = &region_offset;
-
-   LOCK_HARDWARE(intel);
-
-   /* Make sure the global heap is initialized
-    */
-   if (intel->texture_heaps[0])
-      driAgeTextures( intel->texture_heaps[0] );
-
-
-   ret = drmCommandWriteRead( intel->driFd,
-			      DRM_I830_ALLOC,
-			      &alloc, sizeof(alloc));
-   
-   if (ret) {
-      fprintf(stderr, "%s: DRM_I830_ALLOC ret %d\n", __FUNCTION__, ret);
-      UNLOCK_HARDWARE(intel);
-      return NULL;
-   }
-   
-   if (0)
-      fprintf(stderr, "%s: allocated %d bytes\n", __FUNCTION__, size);
-
-   /* Need to propogate this information (agp memory in use) to our
-    * local texture lru.  The kernel has already updated the global
-    * lru.  An alternative would have been to allocate memory the
-    * usual way and then notify the kernel to pin the allocation.
-    */
-   if (intel->texture_heaps[0])
-      driAgeTextures( intel->texture_heaps[0] );
-
-   UNLOCK_HARDWARE(intel);   
-
-   return (void *)((char *)intel->intelScreen->tex.map + region_offset);
-}
-
-void intelFreeAGP( intelContextPtr intel, void *pointer )
-{
-   int region_offset;
-   drmI830MemFree memfree;
-   int ret;
-
-   region_offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-
-   if (region_offset < 0 || 
-       region_offset > intel->intelScreen->tex.size) {
-      fprintf(stderr, "offset %d outside range 0..%d\n", region_offset,
-	      intel->intelScreen->tex.size);
-      return;
-   }
-
-   memfree.region = I830_MEM_REGION_AGP;
-   memfree.region_offset = region_offset;
-   
-   ret = drmCommandWrite( intel->driFd,
-			  DRM_I830_FREE,
-			  &memfree, sizeof(memfree));
-   
-   if (ret) 
-      fprintf(stderr, "%s: DRM_I830_FREE ret %d\n", __FUNCTION__, ret);
-}
-
-/* This version of AllocateMemoryMESA allocates only agp memory, and
- * only does so after the point at which the driver has been
- * initialized.
- *
- * Theoretically a valid context isn't required.  However, in this
- * implementation, it is, as I'm using the hardware lock to protect
- * the kernel data structures, and the current context to get the
- * device fd.
- */
-void *intelAllocateMemoryMESA(__DRInativeDisplay *dpy, int scrn,
-			      GLsizei size, GLfloat readfreq,
-			      GLfloat writefreq, GLfloat priority)
-{
-   GET_CURRENT_CONTEXT(ctx);
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s sz %d %f/%f/%f\n", __FUNCTION__, size, readfreq, 
-	      writefreq, priority);
-
-   if (getenv("INTEL_NO_ALLOC"))
-      return NULL;
-   
-   if (!ctx || INTEL_CONTEXT(ctx) == 0) 
-      return NULL;
-   
-   return intelAllocateAGP( INTEL_CONTEXT(ctx), size );
-}
-
-
-/* Called via glXFreeMemoryMESA() */
-void intelFreeMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLvoid *pointer)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   if (INTEL_DEBUG & DEBUG_IOCTL) 
-      fprintf(stderr, "%s %p\n", __FUNCTION__, pointer);
-
-   if (!ctx || INTEL_CONTEXT(ctx) == 0) {
-      fprintf(stderr, "%s: no context\n", __FUNCTION__);
-      return;
-   }
-
-   intelFreeAGP( INTEL_CONTEXT(ctx), pointer );
-}
-
-/* Called via glXGetMemoryOffsetMESA() 
- *
- * Returns offset of pointer from the start of agp aperture.
- */
-GLuint intelGetMemoryOffsetMESA(__DRInativeDisplay *dpy, int scrn, 
-				const GLvoid *pointer)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   intelContextPtr intel;
-
-   if (!ctx || !(intel = INTEL_CONTEXT(ctx)) ) {
-      fprintf(stderr, "%s: no context\n", __FUNCTION__);
-      return ~0;
-   }
-
-   if (!intelIsAgpMemory( intel, pointer, 0 ))
-      return ~0;
-
-   return intelAgpOffsetFromVirtual( intel, pointer );
-}
-
-
-GLboolean intelIsAgpMemory( intelContextPtr intel, const GLvoid *pointer,
-			   GLint size )
-{
-   int offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-   int valid = (size >= 0 &&
-		offset >= 0 &&
-		offset + size < intel->intelScreen->tex.size);
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "intelIsAgpMemory( %p ) : %d\n", pointer, valid );
-   
-   return valid;
-}
-
-
-GLuint intelAgpOffsetFromVirtual( intelContextPtr intel, const GLvoid *pointer )
-{
-   int offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-
-   if (offset < 0 || offset > intel->intelScreen->tex.size)
-      return ~0;
-   else
-      return intel->intelScreen->tex.offset + offset;
-}
-
-
-
-
-
-/* Flip the front & back buffes
- */
-void intelPageFlip( const __DRIdrawablePrivate *dPriv )
-{
-#if 0
-   intelContextPtr intel;
-   int tmp, ret;
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-
-   if (dPriv->pClipRects) {
-      *(drm_clip_rect_t *)intel->sarea->boxes = dPriv->pClipRects[0];
-      intel->sarea->nbox = 1;
-   }
-
-   ret = drmCommandNone(intel->driFd, DRM_I830_FLIP); 
-   if (ret) {
-      fprintf(stderr, "%s: %d\n", __FUNCTION__, ret);
-      UNLOCK_HARDWARE( intel );
-      exit(1);
-   }
-
-   tmp = intel->sarea->last_enqueue;
-   intelRefillBatchLocked( intel );
-   UNLOCK_HARDWARE( intel );
-
-
-   intelSetDrawBuffer( &intel->ctx, intel->ctx.Color.DriverDrawBuffer );
-#endif
-}
diff --git a/src/mesa/drivers/dri/i915/intel_ioctl.h b/src/mesa/drivers/dri/i915/intel_ioctl.h
deleted file mode 100644
index 8d79e9d73b4..00000000000
--- a/src/mesa/drivers/dri/i915/intel_ioctl.h
+++ /dev/null
@@ -1,72 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_IOCTL_H
-#define INTEL_IOCTL_H
-
-#include "intel_context.h"
-
-extern void intelWaitAgeLocked( intelContextPtr intel, int age, GLboolean unlock );
-
-extern void intelClear(GLcontext *ctx, GLbitfield mask);
-
-extern void intelPageFlip( const __DRIdrawablePrivate *dpriv );
-
-extern void intelRotateWindow(intelContextPtr intel,
-                              __DRIdrawablePrivate *dPriv, GLuint srcBuffer);
-
-extern void intelWaitForIdle( intelContextPtr intel );
-extern void intelFlushBatch( intelContextPtr intel, GLboolean refill );
-extern void intelFlushBatchLocked( intelContextPtr intel,
-				   GLboolean ignore_cliprects,
-				   GLboolean refill,
-				   GLboolean allow_unlock);
-extern void intelRefillBatchLocked( intelContextPtr intel, GLboolean allow_unlock );
-extern void intelFinish( GLcontext *ctx );
-extern void intelFlush( GLcontext *ctx );
-extern void intelglFlush( GLcontext *ctx );
-
-extern void *intelAllocateAGP( intelContextPtr intel, GLsizei size );
-extern void intelFreeAGP( intelContextPtr intel, void *pointer );
-
-extern void *intelAllocateMemoryMESA( __DRInativeDisplay *dpy, int scrn, 
-				      GLsizei size, GLfloat readfreq,
-				      GLfloat writefreq, GLfloat priority );
-
-extern void intelFreeMemoryMESA( __DRInativeDisplay *dpy, int scrn, 
-				 GLvoid *pointer );
-
-extern GLuint intelGetMemoryOffsetMESA( __DRInativeDisplay *dpy, int scrn, const GLvoid *pointer );
-extern GLboolean intelIsAgpMemory( intelContextPtr intel, const GLvoid *pointer,
-				  GLint size );
-
-extern GLuint intelAgpOffsetFromVirtual( intelContextPtr intel, const GLvoid *p );
-
-extern void intelWaitIrq( intelContextPtr intel, int seq );
-extern uint32_t intelGetLastFrame (intelContextPtr intel);
-extern int intelEmitIrqLocked( intelContextPtr intel );
-#endif
diff --git a/src/mesa/drivers/dri/i915/intel_mipmap_tree.c b/src/mesa/drivers/dri/i915/intel_mipmap_tree.c
new file mode 120000
index 00000000000..242fed0b6ae
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_mipmap_tree.c
@@ -0,0 +1 @@
+../intel/intel_mipmap_tree.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_pixel.c b/src/mesa/drivers/dri/i915/intel_pixel.c
index c3030d42b04..d733c5e8745 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_pixel.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel.c
@@ -1,524 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "swrast/swrast.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-
-
-
-static GLboolean
-check_color( const GLcontext *ctx, GLenum type, GLenum format,
-	     const struct gl_pixelstore_attrib *packing,
-	     const void *pixels, GLint sz, GLint pitch )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLuint cpp = intel->intelScreen->cpp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (	(pitch & 63) ||
-	ctx->_ImageTransferState ||
-	packing->SwapBytes ||
-	packing->LsbFirst) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: failed 1\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   if ( type == GL_UNSIGNED_INT_8_8_8_8_REV && 
-	cpp == 4 && 
-	format == GL_BGRA ) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: passed 2\n", __FUNCTION__);
-      return GL_TRUE;
-   }
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s: failed\n", __FUNCTION__);
-
-   return GL_FALSE;
-}
-
-static GLboolean
-check_color_per_fragment_ops( const GLcontext *ctx )
-{
-   int result;
-   result = (!(     ctx->Color.AlphaEnabled || 
-		    ctx->Depth.Test ||
-		    ctx->Fog.Enabled ||
-		    ctx->Scissor.Enabled ||
-		    ctx->Stencil.Enabled ||
-		    !ctx->Color.ColorMask[0] ||
-		    !ctx->Color.ColorMask[1] ||
-		    !ctx->Color.ColorMask[2] ||
-		    !ctx->Color.ColorMask[3] ||
-		    ctx->Color.ColorLogicOpEnabled ||
-		    ctx->Texture._EnabledUnits
-           ) &&
-	   ctx->Current.RasterPosValid);
-   
-   return result;
-}
-
-
-/**
- * Clip the given rectangle against the buffer's bounds (including scissor).
- * \param size returns the 
- * \return GL_TRUE if any pixels remain, GL_FALSE if totally clipped.
- *
- * XXX Replace this with _mesa_clip_drawpixels() and _mesa_clip_readpixels()
- * from Mesa 6.4.  We shouldn't apply scissor for ReadPixels.
- */
-static GLboolean
-clip_pixelrect( const GLcontext *ctx,
-		const GLframebuffer *buffer,
-		GLint *x, GLint *y,
-		GLsizei *width, GLsizei *height)
-{
-   /* left clipping */
-   if (*x < buffer->_Xmin) {
-      *width -= (buffer->_Xmin - *x);
-      *x = buffer->_Xmin;
-   }
-
-   /* right clipping */
-   if (*x + *width > buffer->_Xmax)
-      *width -= (*x + *width - buffer->_Xmax - 1);
-
-   if (*width <= 0)
-      return GL_FALSE;
-
-   /* bottom clipping */
-   if (*y < buffer->_Ymin) {
-      *height -= (buffer->_Ymin - *y);
-      *y = buffer->_Ymin;
-   }
-
-   /* top clipping */
-   if (*y + *height > buffer->_Ymax)
-      *height -= (*y + *height - buffer->_Ymax - 1);
-
-   if (*height <= 0)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-/**
- * Compute intersection of a clipping rectangle and pixel rectangle,
- * returning results in x/y/w/hOut vars.
- * \return GL_TRUE if there's intersection, GL_FALSE if disjoint.
- */
-static INLINE GLboolean
-intersect_region(const drm_clip_rect_t *box,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLint *xOut, GLint *yOut, GLint *wOut, GLint *hOut)
-{
-   GLint bx = box->x1;
-   GLint by = box->y1;
-   GLint bw = box->x2 - bx;
-   GLint bh = box->y2 - by;
-
-   if (bx < x) bw -= x - bx, bx = x;
-   if (by < y) bh -= y - by, by = y;
-   if (bx + bw > x + width) bw = x + width - bx;
-   if (by + bh > y + height) bh = y + height - by;
-
-   *xOut = bx;
-   *yOut = by;
-   *wOut = bw;
-   *hOut = bh;
-
-   if (bw <= 0) return GL_FALSE;
-   if (bh <= 0) return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-
-static GLboolean
-intelTryReadPixels( GLcontext *ctx,
-		  GLint x, GLint y, GLsizei width, GLsizei height,
-		  GLenum format, GLenum type,
-		  const struct gl_pixelstore_attrib *pack,
-		  GLvoid *pixels )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLint size = 0; /* not really used */
-   GLint pitch = pack->RowLength ? pack->RowLength : width;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Only accelerate reading to agp buffers.
-    */
-   if ( !intelIsAgpMemory(intel, pixels, 
-			pitch * height * intel->intelScreen->cpp ) ) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: dest not agp\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Need GL_PACK_INVERT_MESA to cope with upsidedown results from
-    * blitter:
-    */
-   if (!pack->Invert) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: MESA_PACK_INVERT not set\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   if (!check_color(ctx, type, format, pack, pixels, size, pitch))
-      return GL_FALSE;
-
-   switch ( intel->intelScreen->cpp ) {
-   case 4:
-      break;
-   default:
-      return GL_FALSE;
-   }
-
-
-   /* Although the blits go on the command buffer, need to do this and
-    * fire with lock held to guarentee cliprects and drawing offset are
-    * correct.
-    *
-    * This is an unusual situation however, as the code which flushes
-    * a full command buffer expects to be called unlocked.  As a
-    * workaround, immediately flush the buffer on aquiring the lock.
-    */
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-   {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      int nbox = dPriv->numClipRects;
-      int src_offset = intel->readRegion->offset;
-      int src_pitch = intel->intelScreen->front.pitch;
-      int dst_offset = intelAgpOffsetFromVirtual( intel, pixels);
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      int i;
-
-      assert(dst_offset != ~0);  /* should have been caught above */
-
-      if (!clip_pixelrect(ctx, ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s totally clipped -- nothing to do\n",
-		    __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-      /* convert to screen coords (y=0=top) */
-      y = dPriv->h - y - height;
-      x += dPriv->x;
-      y += dPriv->y;
-
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "readpixel blit src_pitch %d dst_pitch %d\n",
-		 src_pitch, pitch);
-
-      /* We don't really have to do window clipping for readpixels.
-       * The OpenGL spec says that pixels read from outside the
-       * visible window region (pixel ownership) have undefined value.
-       */
-      for (i = 0 ; i < nbox ; i++)
-      {
-         GLint bx, by, bw, bh;
-         if (intersect_region(box+i, x, y, width, height,
-                              &bx, &by, &bw, &bh)) {
-            intelEmitCopyBlitLocked( intel,
-                                     intel->intelScreen->cpp,
-                                     src_pitch, src_offset,
-                                     pitch, dst_offset,
-                                     bx, by,
-                                     bx - x, by - y,
-                                     bw, bh );
-         }
-      }
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( &intel->ctx );
-
-   return GL_TRUE;
-}
-
-static void
-intelReadPixels( GLcontext *ctx,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLenum format, GLenum type,
-		 const struct gl_pixelstore_attrib *pack,
-		 GLvoid *pixels )
-{
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (!intelTryReadPixels( ctx, x, y, width, height, format, type, pack, 
-                            pixels))
-      _swrast_ReadPixels( ctx, x, y, width, height, format, type, pack, 
-			  pixels);
-}
-
-
-
-
-static void do_draw_pix( GLcontext *ctx,
-			 GLint x, GLint y, GLsizei width, GLsizei height,
-			 GLint pitch,
-			 const void *pixels,
-			 GLuint dest )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   drm_clip_rect_t *box = dPriv->pClipRects;
-   int nbox = dPriv->numClipRects;
-   int i;
-   int src_offset = intelAgpOffsetFromVirtual( intel, pixels);
-   int src_pitch = pitch;
-
-   assert(src_offset != ~0);  /* should be caught earlier */
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-   if (ctx->DrawBuffer)
-   {
-      y -= height;			/* cope with pixel zoom */
-   
-      if (!clip_pixelrect(ctx, ctx->DrawBuffer,
-			  &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 return;
-      }
-
-      y = dPriv->h - y - height; 	/* convert from gl to hardware coords */
-      x += dPriv->x;
-      y += dPriv->y;
-
-      for (i = 0 ; i < nbox ; i++ )
-      {
-	 GLint bx, by, bw, bh;
-	 if (intersect_region(box + i, x, y, width, height,
-			      &bx, &by, &bw, &bh)) {
-            intelEmitCopyBlitLocked( intel,
-                                     intel->intelScreen->cpp,
-                                     src_pitch, src_offset,
-                                     intel->intelScreen->front.pitch,
-                                     intel->drawRegion->offset,
-                                     bx - x, by - y,
-                                     bx, by,
-                                     bw, bh );
-         }
-      }
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( &intel->ctx );
-}
-
-
-
-static GLboolean
-intelTryDrawPixels( GLcontext *ctx,
-		  GLint x, GLint y, GLsizei width, GLsizei height,
-		  GLenum format, GLenum type,
-		  const struct gl_pixelstore_attrib *unpack,
-		  const GLvoid *pixels )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLint pitch = unpack->RowLength ? unpack->RowLength : width;
-   GLuint dest;
-   GLuint cpp = intel->intelScreen->cpp;
-   GLint size = width * pitch * cpp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   switch (format) {
-   case GL_RGB:
-   case GL_RGBA:
-   case GL_BGRA:
-      dest = intel->drawRegion->offset;
-
-      /* Planemask doesn't have full support in blits.
-       */
-      if (!ctx->Color.ColorMask[RCOMP] ||
-	  !ctx->Color.ColorMask[GCOMP] ||
-	  !ctx->Color.ColorMask[BCOMP] ||
-	  !ctx->Color.ColorMask[ACOMP]) {
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s: planemask\n", __FUNCTION__);
-	 return GL_FALSE;	
-      }
-
-      /* Can't do conversions on agp reads/draws. 
-       */
-      if ( !intelIsAgpMemory( intel, pixels, size ) ) {
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s: not agp memory\n", __FUNCTION__);
-	 return GL_FALSE;
-      }
-
-      if (!check_color(ctx, type, format, unpack, pixels, size, pitch)) {
-	 return GL_FALSE;
-      }
-      if (!check_color_per_fragment_ops(ctx)) {
-	 return GL_FALSE;
-      }
-
-      if (ctx->Pixel.ZoomX != 1.0F ||
-	  ctx->Pixel.ZoomY != -1.0F)
-	 return GL_FALSE;
-      break;
-
-   default:
-      return GL_FALSE;
-   }
-
-   if ( intelIsAgpMemory(intel, pixels, size) )
-   {
-      do_draw_pix( ctx, x, y, width, height, pitch, pixels, dest );
-      return GL_TRUE;
-   }
-   else if (0)
-   {
-      /* Pixels is in regular memory -- get dma buffers and perform
-       * upload through them.  No point doing this for regular uploads
-       * but once we remove some of the restrictions above (colormask,
-       * pixelformat conversion, zoom?, etc), this could be a win.
-       */
-   }
-   else
-      return GL_FALSE;
-
-   return GL_FALSE;
-}
-
-static void
-intelDrawPixels( GLcontext *ctx,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLenum format, GLenum type,
-		 const struct gl_pixelstore_attrib *unpack,
-		 const GLvoid *pixels )
-{
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (intelTryDrawPixels( ctx, x, y, width, height, format, type,
-                           unpack, pixels ))
-      return;
-
-   if (ctx->FragmentProgram._Current == ctx->FragmentProgram._TexEnvProgram) {
-      /*
-       * We don't want the i915 texenv program to be applied to DrawPixels.
-       * This is really just a performance optimization (mesa will other-
-       * wise happily run the fragment program on each pixel in the image).
-       */
-      struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
-   /* can't just set current frag prog to 0 here as on buffer resize
-      we'll get new state checks which will segfault. Remains a hack. */
-      ctx->FragmentProgram._Current = NULL;
-      ctx->FragmentProgram._UseTexEnvProgram = GL_FALSE;
-      ctx->FragmentProgram._Active = GL_FALSE;
-      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-                          unpack, pixels );
-      ctx->FragmentProgram._Current = fpSave;
-      ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
-      ctx->FragmentProgram._Active = GL_TRUE;
-   }
-   else {
-      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-                          unpack, pixels );
-   }
-}
-
-
-
-
-/**
- * Implement glCopyPixels for the front color buffer (or back buffer Pixmap)
- * for the color buffer.  Don't support zooming, pixel transfer, etc.
- * We do support copying from one window to another, ala glXMakeCurrentRead.
- */
-static void
-intelCopyPixels( GLcontext *ctx,
-		 GLint srcx, GLint srcy, GLsizei width, GLsizei height,
-		 GLint destx, GLint desty, GLenum type )
-{
-#if 0
-   const XMesaContext xmesa = XMESA_CONTEXT(ctx);
-   const SWcontext *swrast = SWRAST_CONTEXT( ctx );
-   XMesaDisplay *dpy = xmesa->xm_visual->display;
-   const XMesaDrawable drawBuffer = xmesa->xm_draw_buffer->buffer;
-   const XMesaDrawable readBuffer = xmesa->xm_read_buffer->buffer;
-   const XMesaGC gc = xmesa->xm_draw_buffer->gc;
-
-   ASSERT(dpy);
-   ASSERT(gc);
-
-   if (drawBuffer &&  /* buffer != 0 means it's a Window or Pixmap */
-       readBuffer &&
-       type == GL_COLOR &&
-       (swrast->_RasterMask & ~CLIP_BIT) == 0 && /* no blend, z-test, etc */
-       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
-       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
-       ctx->Pixel.ZoomY == 1.0) {
-      /* Note: we don't do any special clipping work here.  We could,
-       * but X will do it for us.
-       */
-      srcy = FLIP(xmesa->xm_read_buffer, srcy) - height + 1;
-      desty = FLIP(xmesa->xm_draw_buffer, desty) - height + 1;
-      XCopyArea(dpy, readBuffer, drawBuffer, gc,
-                srcx, srcy, width, height, destx, desty);
-   }
-#else
-   _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type );
-#endif
-}
-
-
-
-
-void intelInitPixelFuncs( struct dd_function_table *functions )
-{
-   functions->CopyPixels = intelCopyPixels;
-   if (!getenv("INTEL_NO_BLITS")) {
-      functions->ReadPixels = intelReadPixels;  
-      functions->DrawPixels = intelDrawPixels; 
-   }
-}
+../intel/intel_pixel.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
new file mode 120000
index 00000000000..9085c7b0397
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
@@ -0,0 +1 @@
+../intel/intel_pixel_bitmap.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_pixel_copy.c b/src/mesa/drivers/dri/i915/intel_pixel_copy.c
new file mode 120000
index 00000000000..ee433605904
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_pixel_copy.c
@@ -0,0 +1 @@
+../intel/intel_pixel_copy.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_pixel_draw.c b/src/mesa/drivers/dri/i915/intel_pixel_draw.c
new file mode 120000
index 00000000000..8431a24edfc
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_pixel_draw.c
@@ -0,0 +1 @@
+../intel/intel_pixel_draw.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_pixel_read.c b/src/mesa/drivers/dri/i915/intel_pixel_read.c
new file mode 100644
index 00000000000..56087aacd4e
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_pixel_read.c
@@ -0,0 +1,306 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/image.h"
+#include "main/bufferobj.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_regions.h"
+#include "intel_pixel.h"
+#include "intel_buffer_objects.h"
+
+/* For many applications, the new ability to pull the source buffers
+ * back out of the GTT and then do the packing/conversion operations
+ * in software will be as much of an improvement as trying to get the
+ * blitter and/or texture engine to do the work. 
+ *
+ * This step is gated on private backbuffers.
+ * 
+ * Obviously the frontbuffer can't be pulled back, so that is either
+ * an argument for blit/texture readpixels, or for blitting to a
+ * temporary and then pulling that back.
+ *
+ * When the destination is a pbo, however, it's not clear if it is
+ * ever going to be pulled to main memory (though the access param
+ * will be a good hint).  So it sounds like we do want to be able to
+ * choose between blit/texture implementation on the gpu and pullback
+ * and cpu-based copying.
+ *
+ * Unless you can magically turn client memory into a PBO for the
+ * duration of this call, there will be a cpu-based copying step in
+ * any case.
+ */
+
+
+static GLboolean
+do_texture_readpixels(GLcontext * ctx,
+                      GLint x, GLint y, GLsizei width, GLsizei height,
+                      GLenum format, GLenum type,
+                      const struct gl_pixelstore_attrib *pack,
+                      struct intel_region *dest_region)
+{
+#if 0
+   struct intel_context *intel = intel_context(ctx);
+   intelScreenPrivate *screen = intel->intelScreen;
+   GLint pitch = pack->RowLength ? pack->RowLength : width;
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   int textureFormat;
+   GLenum glTextureFormat;
+   int destFormat, depthFormat, destPitch;
+   drm_clip_rect_t tmp;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+
+   if (ctx->_ImageTransferState ||
+       pack->SwapBytes || pack->LsbFirst || !pack->Invert) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   intel->vtbl.meta_texrect_source(intel, intel_readbuf_region(intel));
+
+   if (!intel->vtbl.meta_render_dest(intel, dest_region, type, format)) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         fprintf(stderr, "%s: couldn't set dest %s/%s\n",
+                 __FUNCTION__,
+                 _mesa_lookup_enum_by_nr(type),
+                 _mesa_lookup_enum_by_nr(format));
+      return GL_FALSE;
+   }
+
+   LOCK_HARDWARE(intel);
+
+   if (intel->driDrawable->numClipRects) {
+      intel->vtbl.install_meta_state(intel);
+      intel->vtbl.meta_no_depth_write(intel);
+      intel->vtbl.meta_no_stencil_write(intel);
+
+      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
+         UNLOCK_HARDWARE(intel);
+         SET_STATE(i830, state);
+         if (INTEL_DEBUG & DEBUG_PIXEL)
+            fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
+         return GL_TRUE;
+      }
+
+      y = dPriv->h - y - height;
+      x += dPriv->x;
+      y += dPriv->y;
+
+
+      /* Set the frontbuffer up as a large rectangular texture.
+       */
+      intel->vtbl.meta_tex_rect_source(intel, src_region, textureFormat);
+
+
+      intel->vtbl.meta_texture_blend_replace(i830, glTextureFormat);
+
+
+      /* Set the 3d engine to draw into the destination region:
+       */
+
+      intel->vtbl.meta_draw_region(intel, dest_region);
+      intel->vtbl.meta_draw_format(intel, destFormat, depthFormat);     /* ?? */
+
+
+      /* Draw a single quad, no cliprects:
+       */
+      intel->vtbl.meta_disable_cliprects(intel);
+
+      intel->vtbl.draw_quad(intel,
+                            0, width, 0, height,
+                            0x00ff00ff, x, x + width, y, y + height);
+
+      intel->vtbl.leave_meta_state(intel);
+   }
+   UNLOCK_HARDWARE(intel);
+
+   intel_region_wait_fence(ctx, dest_region);   /* required by GL */
+   return GL_TRUE;
+#endif
+
+   return GL_FALSE;
+}
+
+
+
+
+static GLboolean
+do_blit_readpixels(GLcontext * ctx,
+                   GLint x, GLint y, GLsizei width, GLsizei height,
+                   GLenum format, GLenum type,
+                   const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *src = intel_readbuf_region(intel);
+   struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj);
+   GLuint dst_offset;
+   GLuint rowLength;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s\n", __FUNCTION__);
+
+   if (!src)
+      return GL_FALSE;
+
+   if (dst) {
+      /* XXX This validation should be done by core mesa:
+       */
+      if (!_mesa_validate_pbo_access(2, pack, width, height, 1,
+                                     format, type, pixels)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels");
+         return GL_TRUE;
+      }
+   }
+   else {
+      /* PBO only for now:
+       */
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s - not PBO\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+
+   if (ctx->_ImageTransferState ||
+       !intel_check_blit_format(src, format, type)) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s - bad format for blit\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s: bad packing params\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   if (pack->RowLength > 0)
+      rowLength = pack->RowLength;
+   else
+      rowLength = width;
+
+   if (pack->Invert) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+   else {
+      rowLength = -rowLength;
+   }
+
+   /* XXX 64-bit cast? */
+   dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height,
+                                             format, type, 0, 0, 0);
+
+
+   /* Although the blits go on the command buffer, need to do this and
+    * fire with lock held to guarentee cliprects are correct.
+    */
+   intelFlush(&intel->ctx);
+   LOCK_HARDWARE(intel);
+
+   if (intel->driDrawable->numClipRects) {
+      GLboolean all = (width * height * src->cpp == dst->Base.Size &&
+                       x == 0 && dst_offset == 0);
+
+      dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
+						  all ? INTEL_WRITE_FULL :
+						  INTEL_WRITE_PART);
+      __DRIdrawablePrivate *dPriv = intel->driDrawable;
+      int nbox = dPriv->numClipRects;
+      drm_clip_rect_t *box = dPriv->pClipRects;
+      drm_clip_rect_t rect;
+      drm_clip_rect_t src_rect;
+      int i;
+
+      src_rect.x1 = dPriv->x + x;
+      src_rect.y1 = dPriv->y + dPriv->h - (y + height);
+      src_rect.x2 = src_rect.x1 + width;
+      src_rect.y2 = src_rect.y1 + height;
+
+
+
+      for (i = 0; i < nbox; i++) {
+         if (!intel_intersect_cliprects(&rect, &src_rect, &box[i]))
+            continue;
+
+         intelEmitCopyBlit(intel,
+                           src->cpp,
+                           src->pitch, src->buffer, 0, src->tiling,
+                           rowLength, dst_buffer, dst_offset, GL_FALSE,
+                           rect.x1,
+                           rect.y1,
+                           rect.x1 - src_rect.x1,
+                           rect.y2 - src_rect.y2,
+                           rect.x2 - rect.x1, rect.y2 - rect.y1,
+			   GL_COPY);
+      }
+   }
+   UNLOCK_HARDWARE(intel);
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s - DONE\n", __FUNCTION__);
+
+   return GL_TRUE;
+}
+
+void
+intelReadPixels(GLcontext * ctx,
+                GLint x, GLint y, GLsizei width, GLsizei height,
+                GLenum format, GLenum type,
+                const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
+{
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   intelFlush(ctx);
+
+   if (do_blit_readpixels
+       (ctx, x, y, width, height, format, type, pack, pixels))
+      return;
+
+   if (do_texture_readpixels
+       (ctx, x, y, width, height, format, type, pack, pixels))
+      return;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
+
+   _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels);
+}
diff --git a/src/mesa/drivers/dri/i915/intel_reg.h b/src/mesa/drivers/dri/i915/intel_reg.h
deleted file mode 100644
index 1ec153266c7..00000000000
--- a/src/mesa/drivers/dri/i915/intel_reg.h
+++ /dev/null
@@ -1,84 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef _INTEL_REG_H_
-#define _INTEL_REG_H_
-
-
-
-#define CMD_3D (0x3<<29)
-
-
-#define _3DPRIMITIVE         ((0x3<<29)|(0x1f<<24))
-#define PRIM_INDIRECT            (1<<23)
-#define PRIM_INLINE              (0<<23)
-#define PRIM_INDIRECT_SEQUENTIAL (0<<17)
-#define PRIM_INDIRECT_ELTS       (1<<17)
-
-#define PRIM3D_TRILIST		(0x0<<18)
-#define PRIM3D_TRISTRIP 	(0x1<<18)
-#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
-#define PRIM3D_TRIFAN		(0x3<<18)
-#define PRIM3D_POLY		(0x4<<18)
-#define PRIM3D_LINELIST 	(0x5<<18)
-#define PRIM3D_LINESTRIP	(0x6<<18)
-#define PRIM3D_RECTLIST 	(0x7<<18)
-#define PRIM3D_POINTLIST	(0x8<<18)
-#define PRIM3D_DIB		(0x9<<18)
-#define PRIM3D_MASK		(0x1f<<18)
-
-#define I915PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define I915PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define I915PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define I915PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-
-
-#define BR00_BITBLT_CLIENT   0x40000000
-#define BR00_OP_COLOR_BLT    0x10000000
-#define BR00_OP_SRC_COPY_BLT 0x10C00000
-#define BR13_SOLID_PATTERN   0x80000000
-
-#define XY_COLOR_BLT_CMD		((2<<29)|(0x50<<22)|0x4)
-#define XY_COLOR_BLT_WRITE_ALPHA	(1<<21)
-#define XY_COLOR_BLT_WRITE_RGB		(1<<20)
-
-#define XY_SRC_COPY_BLT_CMD             ((2<<29)|(0x53<<22)|6)
-#define XY_SRC_COPY_BLT_WRITE_ALPHA     (1<<21)
-#define XY_SRC_COPY_BLT_WRITE_RGB       (1<<20)
-
-#endif
diff --git a/src/mesa/drivers/dri/i915/intel_regions.c b/src/mesa/drivers/dri/i915/intel_regions.c
new file mode 120000
index 00000000000..89b2f15c10f
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_regions.c
@@ -0,0 +1 @@
+../intel/intel_regions.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index d9438ba0fd8..410052b3c2b 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -30,15 +30,16 @@
  * dma buffers.  Use strip/fan hardware acceleration where possible.
  *
  */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
 
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
+#include "tnl/t_pipeline.h"
 
 #include "intel_screen.h"
 #include "intel_context.h"
@@ -51,14 +52,14 @@
  * dma buffers.  Use strip/fan hardware primitives where possible.
  * Try to simulate missing primitives with indexed vertices.
  */
-#define HAVE_POINTS      0  /* Has it, but can't use because subpixel has to
-			     * be adjusted for points on the INTEL/I845G
-			     */
+#define HAVE_POINTS      0      /* Has it, but can't use because subpixel has to
+                                 * be adjusted for points on the INTEL/I845G
+                                 */
 #define HAVE_LINES       1
 #define HAVE_LINE_STRIPS 1
 #define HAVE_TRIANGLES   1
 #define HAVE_TRI_STRIPS  1
-#define HAVE_TRI_STRIP_1 0  /* has it, template can't use it yet */
+#define HAVE_TRI_STRIP_1 0      /* has it, template can't use it yet */
 #define HAVE_TRI_FANS    1
 #define HAVE_POLYGONS    1
 #define HAVE_QUADS       0
@@ -66,7 +67,7 @@
 
 #define HAVE_ELTS        0
 
-static GLuint hw_prim[GL_POLYGON+1] = {
+static uint32_t hw_prim[GL_POLYGON + 1] = {
    0,
    PRIM3D_LINELIST,
    PRIM3D_LINESTRIP,
@@ -79,7 +80,7 @@ static GLuint hw_prim[GL_POLYGON+1] = {
    PRIM3D_POLY
 };
 
-static const GLenum reduced_prim[GL_POLYGON+1] = {  
+static const GLenum reduced_prim[GL_POLYGON + 1] = {
    GL_POINTS,
    GL_LINES,
    GL_LINES,
@@ -92,58 +93,79 @@ static const GLenum reduced_prim[GL_POLYGON+1] = {
    GL_TRIANGLES
 };
 
-static const int scale_prim[GL_POLYGON+1] = {  
-   0,				/* fallback case */
+static const int scale_prim[GL_POLYGON + 1] = {
+   0,                           /* fallback case */
    1,
    2,
    2,
    1,
    3,
    3,
-   0,				/* fallback case */
-   0,				/* fallback case */
+   0,                           /* fallback case */
+   0,                           /* fallback case */
    3
 };
 
 
-static void intelDmaPrimitive( intelContextPtr intel, GLenum prim )
+static void
+intelDmaPrimitive(struct intel_context *intel, GLenum prim)
 {
-   if (0) fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
+   if (0)
+      fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
    INTEL_FIREVERTICES(intel);
-   intel->vtbl.reduced_primitive_state( intel, reduced_prim[prim] );
-   intelStartInlinePrimitive( intel, hw_prim[prim] );
+   intel->vtbl.reduced_primitive_state(intel, reduced_prim[prim]);
+   intel_set_prim(intel, hw_prim[prim]);
+}
+
+static inline GLuint intel_get_vb_max(struct intel_context *intel)
+{
+   GLuint ret;
+
+   if (intel->intelScreen->no_vbo)
+      ret = intel->batch->size - 1500;
+   else
+      ret = INTEL_VB_SIZE;
+   ret /= (intel->vertex_size * 4);
+   return ret;
 }
 
+static inline GLuint intel_get_current_max(struct intel_context *intel)
+{
 
-#define LOCAL_VARS intelContextPtr intel = INTEL_CONTEXT(ctx)
+   if (intel->intelScreen->no_vbo)
+      return intel_get_vb_max(intel);
+   else
+      return (INTEL_VB_SIZE - intel->prim.current_offset) / (intel->vertex_size * 4);
+}
+
+#define LOCAL_VARS struct intel_context *intel = intel_context(ctx)
 #define INIT( prim ) 				\
 do {						\
    intelDmaPrimitive( intel, prim );		\
 } while (0)
-#define FLUSH()  INTEL_FIREVERTICES( intel )
 
-#define GET_SUBSEQUENT_VB_MAX_VERTS() \
-  (((intel->alloc.size / 2) - 1500) / (intel->vertex_size*4))
-#define GET_CURRENT_VB_MAX_VERTS() GET_SUBSEQUENT_VB_MAX_VERTS()
+#define FLUSH() INTEL_FIREVERTICES(intel)
+
+#define GET_SUBSEQUENT_VB_MAX_VERTS() intel_get_vb_max(intel)
+#define GET_CURRENT_VB_MAX_VERTS() intel_get_current_max(intel)
+
+#define ALLOC_VERTS(nr) intel_get_prim_space(intel, nr)
 
-#define ALLOC_VERTS( nr ) \
-   intelExtendInlinePrimitive( intel, (nr) * intel->vertex_size )
-  
 #define EMIT_VERTS( ctx, j, nr, buf ) \
-  _tnl_emit_vertices_to_buffer(ctx, j, (j)+(nr), buf )  
+  _tnl_emit_vertices_to_buffer(ctx, j, (j)+(nr), buf )
 
 #define TAG(x) intel_##x
 #include "tnl_dd/t_dd_dmatmp.h"
-  
-  
+
+
 /**********************************************************************/
 /*                          Render pipeline stage                     */
 /**********************************************************************/
 
 /* Heuristic to choose between the two render paths:  
  */
-static GLboolean choose_render( intelContextPtr intel,
-				struct vertex_buffer *VB )
+static GLboolean
+choose_render(struct intel_context *intel, struct vertex_buffer *VB)
 {
    int vertsz = intel->vertex_size;
    int cost_render = 0;
@@ -153,20 +175,20 @@ static GLboolean choose_render( intelContextPtr intel,
    int nr_rverts = 0;
    int rprim = intel->reduced_primitive;
    int i = 0;
-   
-   for (i = 0 ; i < VB->PrimitiveCount ; i++) {
+
+   for (i = 0; i < VB->PrimitiveCount; i++) {
       GLuint prim = VB->Primitive[i].mode;
       GLuint length = VB->Primitive[i].count;
 
       if (!length)
-	 continue;
+         continue;
 
       nr_prims++;
       nr_rverts += length * scale_prim[prim & PRIM_MODE_MASK];
 
       if (reduced_prim[prim & PRIM_MODE_MASK] != rprim) {
-	 nr_rprims++;
-	 rprim = reduced_prim[prim & PRIM_MODE_MASK];
+         nr_rprims++;
+         rprim = reduced_prim[prim & PRIM_MODE_MASK];
       }
    }
 
@@ -177,64 +199,82 @@ static GLboolean choose_render( intelContextPtr intel,
 
    /* One point for every 1024 dwords (4k) of dma:
     */
-   cost_render += (vertsz * i) / 1024; 
-   cost_fallback += (vertsz * nr_rverts) / 1024; 
+   cost_render += (vertsz * i) / 1024;
+   cost_fallback += (vertsz * nr_rverts) / 1024;
 
    if (0)
       fprintf(stderr, "cost render: %d fallback: %d\n",
-	      cost_render, cost_fallback);
+              cost_render, cost_fallback);
 
-   if (cost_render > cost_fallback) 
+   if (cost_render > cost_fallback)
       return GL_FALSE;
 
    return GL_TRUE;
 }
 
 
-static GLboolean intel_run_render( GLcontext *ctx, 
-				 struct tnl_pipeline_stage *stage )
+static GLboolean
+intel_run_render(GLcontext * ctx, struct tnl_pipeline_stage *stage)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
    GLuint i;
 
+   intel->vtbl.render_prevalidate( intel );
+
    /* Don't handle clipping or indexed vertices.
     */
-   if (intel->RenderIndex != 0 || 
-       !intel_validate_render( ctx, VB ) || 
-       !choose_render( intel, VB )) {
+   if (intel->RenderIndex != 0 ||
+       !intel_validate_render(ctx, VB) || !choose_render(intel, VB)) {
       return GL_TRUE;
    }
 
    tnl->clipspace.new_inputs |= VERT_BIT_POS;
 
-   tnl->Driver.Render.Start( ctx );
-   
-   for (i = 0 ; i < VB->PrimitiveCount ; i++)
-   {
-      GLuint prim = VB->Primitive[i].mode;
+   tnl->Driver.Render.Start(ctx);
+
+   for (i = 0; i < VB->PrimitiveCount; i++) {
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
       if (!length)
-	 continue;
+         continue;
 
-      intel_render_tab_verts[prim & PRIM_MODE_MASK]( ctx, start, start + length,
-						     prim );
+      intel_render_tab_verts[prim & PRIM_MODE_MASK] (ctx, start,
+                                                     start + length, prim);
    }
-      
-   tnl->Driver.Render.Finish( ctx );
 
-   return GL_FALSE;     /* finished the pipe */
+   tnl->Driver.Render.Finish(ctx);
+
+   INTEL_FIREVERTICES(intel);
+
+   return GL_FALSE;             /* finished the pipe */
 }
 
-const struct tnl_pipeline_stage _intel_render_stage =
-{
+static const struct tnl_pipeline_stage _intel_render_stage = {
    "intel render",
    NULL,
    NULL,
    NULL,
    NULL,
-   intel_run_render	/* run */
+   intel_run_render             /* run */
+};
+
+const struct tnl_pipeline_stage *intel_pipeline[] = {
+   &_tnl_vertex_transform_stage,
+   &_tnl_vertex_cull_stage,
+   &_tnl_normal_transform_stage,
+   &_tnl_lighting_stage,
+   &_tnl_fog_coordinate_stage,
+   &_tnl_texgen_stage,
+   &_tnl_texture_transform_stage,
+   &_tnl_point_attenuation_stage,
+   &_tnl_vertex_program_stage,
+#if 1
+   &_intel_render_stage,        /* ADD: unclipped rastersetup-to-dma */
+#endif
+   &_tnl_render_stage,
+   0,
 };
diff --git a/src/mesa/drivers/dri/i915/intel_rotate.c b/src/mesa/drivers/dri/i915/intel_rotate.c
deleted file mode 100644
index a77640ee54d..00000000000
--- a/src/mesa/drivers/dri/i915/intel_rotate.c
+++ /dev/null
@@ -1,221 +0,0 @@
-
-/**
- * Routines for simple 2D->2D transformations for rotated, flipped screens.
- *
- * XXX This code is not intel-specific.  Move it into a common/utility
- * someday.
- */
-
-#include "intel_rotate.h"
-
-#define MIN2(A, B)   ( ((A) < (B)) ? (A) : (B) )
-
-#define ABS(A)  ( ((A) < 0) ? -(A) : (A) )
-
-
-void
-matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02,
-            int m10, int m11, int m12)
-{
-   m->m00 = m00;   m->m01 = m01;   m->m02 = m02;
-   m->m10 = m10;   m->m11 = m11;   m->m12 = m12;
-}
-
-
-/*
- * Transform (x,y) coordinate by the given matrix.
- */
-void
-matrix23TransformCoordf(const struct matrix23 *m, float *x, float *y)
-{
-   const float x0 = *x;
-   const float y0 = *y;
-
-   *x = m->m00 * x0 + m->m01 * y0 + m->m02;
-   *y = m->m10 * x0 + m->m11 * y0 + m->m12;
-}
-
-
-void
-matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y)
-{
-   const int x0 = *x;
-   const int y0 = *y;
-
-   *x = m->m00 * x0 + m->m01 * y0 + m->m02;
-   *y = m->m10 * x0 + m->m11 * y0 + m->m12;
-}
-
-
-/*
- * Transform a width and height by the given matrix.
- * XXX this could be optimized quite a bit.
- */
-void
-matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist)
-{
-   int x0 = 0, y0 = 0;
-   int x1 = *xDist, y1 = 0;
-   int x2 = 0, y2 = *yDist;
-   matrix23TransformCoordi(m, &x0, &y0);
-   matrix23TransformCoordi(m, &x1, &y1);
-   matrix23TransformCoordi(m, &x2, &y2);
-
-   *xDist = (x1 - x0) + (x2 - x0);
-   *yDist = (y1 - y0) + (y2 - y0);
-
-   if (*xDist < 0)
-       *xDist = -*xDist;
-   if (*yDist < 0)
-       *yDist = -*yDist;
-}
-
-
-/**
- * Transform the rect defined by (x, y, w, h) by m.
- */
-void
-matrix23TransformRect(const struct matrix23 *m, int *x, int *y, int *w, int *h)
-{
-   int x0 = *x, y0 = *y;
-   int x1 = *x + *w, y1 = *y;
-   int x2 = *x + *w, y2 = *y + *h;
-   int x3 = *x, y3 = *y + *h;
-   matrix23TransformCoordi(m, &x0, &y0);
-   matrix23TransformCoordi(m, &x1, &y1);
-   matrix23TransformCoordi(m, &x2, &y2);
-   matrix23TransformCoordi(m, &x3, &y3);
-   *w = ABS(x1 - x0) + ABS(x2 - x1);
-   /**w = ABS(*w);*/
-   *h = ABS(y1 - y0) + ABS(y2 - y1);
-   /**h = ABS(*h);*/
-   *x = MIN2(x0, x1);
-   *x = MIN2(*x, x2);
-   *y = MIN2(y0, y1);
-   *y = MIN2(*y, y2);
-}
-
-
-/*
- * Make rotation matrix for width X height screen.
- */
-void
-matrix23Rotate(struct matrix23 *m, int width, int height, int angle)
-{
-   switch (angle) {
-   case 0:
-      matrix23Set(m, 1, 0, 0, 0, 1, 0);
-      break;
-   case 90:
-      matrix23Set(m, 0, 1, 0,  -1, 0, width);
-      break;
-   case 180:
-      matrix23Set(m, -1, 0, width,  0, -1, height);
-      break;
-   case 270:
-      matrix23Set(m, 0, -1, height,  1, 0, 0);
-      break;
-   default:
-      /*abort()*/;
-   }
-}
-
-
-/*
- * Make flip/reflection matrix for width X height screen.
- */
-void
-matrix23Flip(struct matrix23 *m, int width, int height, int xflip, int yflip)
-{
-   if (xflip) {
-      m->m00 = -1;  m->m01 = 0;   m->m02 = width - 1;
-   }
-   else {
-      m->m00 = 1;   m->m01 = 0;   m->m02 = 0;
-   }
-   if (yflip) {
-      m->m10 = 0;   m->m11 = -1;  m->m12 = height - 1;
-   }
-   else {
-      m->m10 = 0;   m->m11 = 1;   m->m12 = 0;
-   }
-}
-
-
-/*
- * result = a * b
- */
-void
-matrix23Multiply(struct matrix23 *result,
-                 const struct matrix23 *a, const struct matrix23 *b)
-{
-   result->m00 = a->m00 * b->m00 + a->m01 * b->m10;
-   result->m01 = a->m00 * b->m01 + a->m01 * b->m11;
-   result->m02 = a->m00 * b->m02 + a->m01 * b->m12 + a->m02;
-
-   result->m10 = a->m10 * b->m00 + a->m11 * b->m10;
-   result->m11 = a->m10 * b->m01 + a->m11 * b->m11;
-   result->m12 = a->m10 * b->m02 + a->m11 * b->m12 + a->m12;
-}
-
-
-#if 000
-
-#include <stdio.h>
-
-int
-main(int argc, char *argv[])
-{
-   int width = 500, height = 400;
-   int rot;
-   int fx = 0, fy = 0;  /* flip x and/or y ? */
-   int coords[4][2];
-
-   /* four corner coords to test with */
-   coords[0][0] = 0;  coords[0][1] = 0;
-   coords[1][0] = width-1;  coords[1][1] = 0;
-   coords[2][0] = width-1;  coords[2][1] = height-1;
-   coords[3][0] = 0;  coords[3][1] = height-1;
-
-
-   for (rot = 0; rot < 360; rot += 90) {
-      struct matrix23 rotate, flip, m;
-      int i;
-
-      printf("Rot %d, xFlip %d, yFlip %d:\n", rot, fx, fy);
-
-      /* make transformation matrix 'm' */
-      matrix23Rotate(&rotate, width, height, rot);
-      matrix23Flip(&flip, width, height, fx, fy);
-      matrix23Multiply(&m, &rotate, &flip);
-
-      /* xform four coords */
-      for (i = 0; i < 4; i++) {
-         int x = coords[i][0];
-         int y = coords[i][1];
-         matrix23TransformCoordi(&m, &x, &y);
-         printf("  %d, %d  -> %d %d\n", coords[i][0], coords[i][1], x, y);
-      }
-
-      /* xform width, height */
-      {
-         int x = width;
-         int y = height;
-         matrix23TransformDistance(&m, &x, &y);
-         printf("  %d x %d -> %d x %d\n", width, height, x, y);
-      }
-
-      /* xform rect */
-      {
-         int x = 50, y = 10, w = 200, h = 100;
-         matrix23TransformRect(&m, &x, &y, &w, &h);
-         printf("  %d,%d %d x %d -> %d, %d %d x %d\n", 50, 10, 200, 100,
-                x, y, w, h);
-      }
-
-   }
-
-   return 0;
-}
-#endif
diff --git a/src/mesa/drivers/dri/i915/intel_rotate.h b/src/mesa/drivers/dri/i915/intel_rotate.h
deleted file mode 100644
index 0da45d20ce5..00000000000
--- a/src/mesa/drivers/dri/i915/intel_rotate.h
+++ /dev/null
@@ -1,41 +0,0 @@
-#ifndef INTEL_ROTATE_H
-#define INTEL_ROTATE_H 1
-
-struct matrix23
-{
-   int m00, m01, m02;
-   int m10, m11, m12;
-};
-
-
-
-extern void
-matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02,
-            int m10, int m11, int m12);
-
-extern void
-matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y);
-
-extern void
-matrix23TransformCoordf(const struct matrix23 *m, float *x, float *y);
-
-extern void
-matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist);
-
-extern void
-matrix23TransformRect(const struct matrix23 *m,
-                      int *x, int *y, int *w, int *h);
-
-extern void
-matrix23Rotate(struct matrix23 *m, int width, int height, int angle);
-
-extern void
-matrix23Flip(struct matrix23 *m, int width, int height, int xflip, int yflip);
-
-extern void
-matrix23Multiply(struct matrix23 *result,
-                 const struct matrix23 *a, const struct matrix23 *b);
-
-
-#endif /* INTEL_ROTATE_H */
diff --git a/src/mesa/drivers/dri/i915/intel_screen.c b/src/mesa/drivers/dri/i915/intel_screen.c
index 7e7935510a0..f2db48272b9 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_screen.c
+++ b/src/mesa/drivers/dri/i915/intel_screen.c
@@ -1,690 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "matrix.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
-#include "utils.h"
-#include "vblank.h"
-#include "xmlpool.h"
-
-
-#include "intel_screen.h"
-
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_tris.h"
-#include "intel_ioctl.h"
-
-#include "i830_dri.h"
-
-PUBLIC const char __driConfigOptions[] =
-DRI_CONF_BEGIN
-    DRI_CONF_SECTION_PERFORMANCE
-       DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS) 
-       DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
-    DRI_CONF_SECTION_END
-    DRI_CONF_SECTION_QUALITY
-       DRI_CONF_FORCE_S3TC_ENABLE(false)
-       DRI_CONF_ALLOW_LARGE_TEXTURES(1)
-      DRI_CONF_SECTION_END
-DRI_CONF_END;
-const GLuint __driNConfigOptions = 4;
-
-#ifdef USE_NEW_INTERFACE
-static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
-#endif /*USE_NEW_INTERFACE*/
-
-extern const struct dri_extension card_extensions[];
-
-/**
- * Map all the memory regions described by the screen.
- * \return GL_TRUE if success, GL_FALSE if error.
- */
-GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-
-   if (intelScreen->front.handle) {
-      if (drmMap(sPriv->fd,
-                 intelScreen->front.handle,
-                 intelScreen->front.size,
-                 (drmAddress *)&intelScreen->front.map) != 0) {
-         _mesa_problem(NULL, "drmMap(frontbuffer) failed!");
-         return GL_FALSE;
-      }
-   }
-   else {
-      _mesa_warning(NULL, "no front buffer handle in intelMapScreenRegions!");
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->back.handle,
-              intelScreen->back.size,
-              (drmAddress *)&intelScreen->back.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->depth.handle,
-              intelScreen->depth.size,
-              (drmAddress *)&intelScreen->depth.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->tex.handle,
-              intelScreen->tex.size,
-              (drmAddress *)&intelScreen->tex.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (0)
-      printf("Mappings:  front: %p  back: %p  depth: %p  tex: %p\n",
-          intelScreen->front.map,
-          intelScreen->back.map,
-          intelScreen->depth.map,
-          intelScreen->tex.map);
-   return GL_TRUE;
-}
-
-
-void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen)
-{
-#define REALLY_UNMAP 1
-   if (intelScreen->front.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->front.map, intelScreen->front.size) != 0)
-         printf("drmUnmap front failed!\n");
-#endif
-      intelScreen->front.map = NULL;
-   }
-   if (intelScreen->back.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->back.map, intelScreen->back.size) != 0)
-         printf("drmUnmap back failed!\n");
-#endif
-      intelScreen->back.map = NULL;
-   }
-   if (intelScreen->depth.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->depth.map, intelScreen->depth.size);
-      intelScreen->depth.map = NULL;
-#endif
-   }
-   if (intelScreen->tex.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->tex.map, intelScreen->tex.size);
-      intelScreen->tex.map = NULL;
-#endif
-   }
-}
-
-
-static void
-intelPrintDRIInfo(intelScreenPrivate *intelScreen,
-                  __DRIscreenPrivate *sPriv,
-                  I830DRIPtr gDRIPriv)
-{
-   fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->front.size, intelScreen->front.offset,
-           intelScreen->front.pitch);
-   fprintf(stderr, "*** Back size:    0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->back.size, intelScreen->back.offset,
-           intelScreen->back.pitch);
-   fprintf(stderr, "*** Depth size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->depth.size, intelScreen->depth.offset,
-           intelScreen->depth.pitch);
-   fprintf(stderr, "*** Rotated size: 0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->rotated.size, intelScreen->rotated.offset,
-           intelScreen->rotated.pitch);
-   fprintf(stderr, "*** Texture size: 0x%x  offset: 0x%x\n",
-           intelScreen->tex.size, intelScreen->tex.offset);
-   fprintf(stderr, "*** Memory : 0x%x\n", gDRIPriv->mem);
-}
-
-
-static void
-intelPrintSAREA(const drmI830Sarea *sarea)
-{
-   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width, sarea->height);
-   fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
-   fprintf(stderr,
-           "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->front_offset, sarea->front_size,
-           (unsigned) sarea->front_handle);
-   fprintf(stderr,
-           "SAREA: back  offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->back_offset, sarea->back_size,
-           (unsigned) sarea->back_handle);
-   fprintf(stderr, "SAREA: depth offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->depth_offset, sarea->depth_size,
-           (unsigned) sarea->depth_handle);
-   fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->tex_offset, sarea->tex_size,
-           (unsigned) sarea->tex_handle);
-   fprintf(stderr, "SAREA: rotation: %d\n", sarea->rotation);
-   fprintf(stderr,
-           "SAREA: rotated offset: 0x%08x  size: 0x%x\n",
-           sarea->rotated_offset, sarea->rotated_size);
-   fprintf(stderr, "SAREA: rotated pitch: %d\n", sarea->rotated_pitch);
-}
-
-
-/**
- * A number of the screen parameters are obtained/computed from
- * information in the SAREA.  This function updates those parameters.
- */
-void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           drmI830Sarea *sarea)
-{
-   intelScreen->width = sarea->width;
-   intelScreen->height = sarea->height;
-
-   intelScreen->front.offset = sarea->front_offset;
-   intelScreen->front.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->front.handle = sarea->front_handle;
-   intelScreen->front.size = sarea->front_size;
-
-   intelScreen->back.offset = sarea->back_offset;
-   intelScreen->back.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->back.handle = sarea->back_handle;
-   intelScreen->back.size = sarea->back_size;
-			 
-   intelScreen->depth.offset = sarea->depth_offset;
-   intelScreen->depth.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->depth.handle = sarea->depth_handle;
-   intelScreen->depth.size = sarea->depth_size;
-
-   intelScreen->tex.offset = sarea->tex_offset;
-   intelScreen->logTextureGranularity = sarea->log_tex_granularity;
-   intelScreen->tex.handle = sarea->tex_handle;
-   intelScreen->tex.size = sarea->tex_size;
-
-   intelScreen->rotated.offset = sarea->rotated_offset;
-   intelScreen->rotated.pitch = sarea->rotated_pitch * intelScreen->cpp;
-   intelScreen->rotated.size = sarea->rotated_size;
-   intelScreen->current_rotation = sarea->rotation;
-   matrix23Rotate(&intelScreen->rotMatrix,
-                  sarea->width, sarea->height, sarea->rotation);
-   intelScreen->rotatedWidth = sarea->virtualX;
-   intelScreen->rotatedHeight = sarea->virtualY;
-
-   if (0)
-      intelPrintSAREA(sarea);
-}
-
-
-static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen;
-   I830DRIPtr         gDRIPriv = (I830DRIPtr)sPriv->pDevPriv;
-   drmI830Sarea *sarea;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
-
-   if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
-      fprintf(stderr,"\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
-      return GL_FALSE;
-   }
-
-   /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *)CALLOC(sizeof(intelScreenPrivate));
-   if (!intelScreen) {
-      fprintf(stderr,"\nERROR!  Allocating private area failed\n");
-      return GL_FALSE;
-   }
-   /* parse information in __driConfigOptions */
-   driParseOptionInfo (&intelScreen->optionCache,
-		       __driConfigOptions, __driNConfigOptions);
-
-   intelScreen->driScrnPriv = sPriv;
-   sPriv->private = (void *)intelScreen;
-   intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
-   sarea = (drmI830Sarea *)
-         (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
-
-   intelScreen->deviceID = gDRIPriv->deviceID;
-   intelScreen->mem = gDRIPriv->mem;
-   intelScreen->cpp = gDRIPriv->cpp;
-
-   switch (gDRIPriv->bitsPerPixel) {
-   case 15: intelScreen->fbFormat = DV_PF_555; break;
-   case 16: intelScreen->fbFormat = DV_PF_565; break;
-   case 32: intelScreen->fbFormat = DV_PF_8888; break;
-   }
-			 
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   if (0)
-      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr,"\nERROR!  mapping regions\n");
-      _mesa_free(intelScreen);
-      sPriv->private = NULL;
-      return GL_FALSE;
-   }
-
-   intelScreen->drmMinor = sPriv->drmMinor;
-
-   /* Determine if IRQs are active? */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_IRQ_ACTIVE;
-      gp.value = &intelScreen->irq_active;
-
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
-      if (ret) {
-	 fprintf(stderr, "drmI830GetParam: %d\n", ret);
-	 return GL_FALSE;
-      }
-   }
-
-   /* Determine if batchbuffers are allowed */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_ALLOW_BATCHBUFFER;
-      gp.value = &intelScreen->allow_batchbuffer;
-
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
-      if (ret) {
-	 fprintf(stderr, "drmI830GetParam: (%d) %d\n", gp.param, ret);
-	 return GL_FALSE;
-      }
-   }
-
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
-      (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
-      (*glx_enable_extension)( psc, "GLX_MESA_allocate_memory" );
-      (*glx_enable_extension)( psc, "GLX_MESA_copy_sub_buffer" );
-   }
-   
-   sPriv->psc->allocateMemory = (void *) intelAllocateMemoryMESA;
-   sPriv->psc->freeMemory     = (void *) intelFreeMemoryMESA;
-   sPriv->psc->memoryOffset   = (void *) intelGetMemoryOffsetMESA;
-
-   return GL_TRUE;
-}
-		
-		
-static void intelDestroyScreen(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-
-   intelUnmapScreenRegions(intelScreen);
-
-   driDestroyOptionInfo (&intelScreen->optionCache);
-
-   FREE(intelScreen);
-   sPriv->private = NULL;
-}
-
-
-static GLboolean intelCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-				    __DRIdrawablePrivate *driDrawPriv,
-				    const __GLcontextModes *mesaVis,
-				    GLboolean isPixmap )
-{
-   intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
-
-   if (isPixmap) {
-      return GL_FALSE; /* not implemented */
-   } else {
-      GLboolean swStencil = (mesaVis->stencilBits > 0 && 
-			     mesaVis->depthBits != 24);
-
-      struct gl_framebuffer *fb = _mesa_create_framebuffer(mesaVis);
-
-      {
-         driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->front.map,
-                                 screen->cpp,
-                                 screen->front.offset, screen->front.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(frontRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
-      }
-
-      if (mesaVis->doubleBufferMode) {
-         driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->back.map,
-                                 screen->cpp,
-                                 screen->back.offset, screen->back.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(backRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
-      }
-
-      if (mesaVis->depthBits == 16) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-      else if (mesaVis->depthBits == 24) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-
-      if (mesaVis->stencilBits > 0 && !swStencil) {
-         driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(stencilRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
-      }
-
-      _mesa_add_soft_renderbuffers(fb,
-                                   GL_FALSE, /* color */
-                                   GL_FALSE, /* depth */
-                                   swStencil,
-                                   mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* alpha */
-                                   GL_FALSE /* aux */);
-      driDrawPriv->driverPrivate = (void *) fb;
-
-      return (driDrawPriv->driverPrivate != NULL);
-   }
-}
-
-static void intelDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
-{
-   _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
-}
-
-
-/**
- * Get information about previous buffer swaps.
- */
-static int
-intelGetSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
-{
-   intelContextPtr intel;
-
-   if ( (dPriv == NULL) || (dPriv->driContextPriv == NULL)
-	|| (dPriv->driContextPriv->driverPrivate == NULL)
-	|| (sInfo == NULL) ) {
-      return -1;
-   }
-
-   intel = dPriv->driContextPriv->driverPrivate;
-   sInfo->swap_count = intel->swap_count;
-   sInfo->swap_ust = intel->swap_ust;
-   sInfo->swap_missed_count = intel->swap_missed_count;
-
-   sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
-       ? driCalculateSwapUsage( dPriv, 0, intel->swap_missed_ust )
-       : 0.0;
-
-   return 0;
-}
-
-
-/* There are probably better ways to do this, such as an
- * init-designated function to register chipids and createcontext
- * functions.
- */
-extern GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
-				    __DRIcontextPrivate *driContextPriv,
-				    void *sharedContextPrivate);
-
-extern GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate);
-
-
-
-
-static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate)
-{
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-
-   switch (intelScreen->deviceID) {
-   case PCI_CHIP_845_G:
-   case PCI_CHIP_I830_M:
-   case PCI_CHIP_I855_GM:
-   case PCI_CHIP_I865_G:
-      return i830CreateContext( mesaVis, driContextPriv, 
-				sharedContextPrivate );
-
-   case PCI_CHIP_I915_G:
-   case PCI_CHIP_I915_GM:
-   case PCI_CHIP_I945_G:
-   case PCI_CHIP_I945_GM:
-   case PCI_CHIP_I945_GME:
-   case PCI_CHIP_G33_G:
-   case PCI_CHIP_Q35_G:
-   case PCI_CHIP_Q33_G:
-      return i915CreateContext( mesaVis, driContextPriv, 
-			       sharedContextPrivate );
- 
-   default:
-      fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
-      return GL_FALSE;
-   }
-}
-
-
-static const struct __DriverAPIRec intelAPI = {
-   .InitDriver      = intelInitDriver,
-   .DestroyScreen   = intelDestroyScreen,
-   .CreateContext   = intelCreateContext,
-   .DestroyContext  = intelDestroyContext,
-   .CreateBuffer    = intelCreateBuffer,
-   .DestroyBuffer   = intelDestroyBuffer,
-   .SwapBuffers     = intelSwapBuffers,
-   .MakeCurrent     = intelMakeCurrent,
-   .UnbindContext   = intelUnbindContext,
-   .GetSwapInfo     = intelGetSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = intelCopySubBuffer
-};
-
-
-static __GLcontextModes *
-intelFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
-{
-   __GLcontextModes * modes;
-   __GLcontextModes * m;
-   unsigned num_modes;
-   unsigned depth_buffer_factor;
-   unsigned back_buffer_factor;
-   GLenum fb_format;
-   GLenum fb_type;
-
-   /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
-    * support pageflipping at all.
-    */
-   static const GLenum back_buffer_modes[] = {
-      GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
-   };
-
-   uint8_t depth_bits_array[3];
-   uint8_t stencil_bits_array[3];
-
-
-   depth_bits_array[0] = 0;
-   depth_bits_array[1] = depth_bits;
-   depth_bits_array[2] = depth_bits;
-
-   /* Just like with the accumulation buffer, always provide some modes
-    * with a stencil buffer.  It will be a sw fallback, but some apps won't
-    * care about that.
-    */
-   stencil_bits_array[0] = 0;
-   stencil_bits_array[1] = 0;
-   stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
-   back_buffer_factor  = (have_back_buffer) ? 3 : 1;
-
-   num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
-    if ( pixel_bits == 16 ) {
-        fb_format = GL_RGB;
-        fb_type = GL_UNSIGNED_SHORT_5_6_5;
-    }
-    else {
-        fb_format = GL_BGRA;
-        fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
-    }
-
-   modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-   m = modes;
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-
-   /* Mark the visual as slow if there are "fake" stencil bits.
-    */
-   for ( m = modes ; m != NULL ; m = m->next ) {
-      if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	 m->visualRating = GLX_SLOW_CONFIG;
-      }
-   }
-
-   return modes;
-}
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 1, 5, 0 };
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 4, 0 };
-
-   dri_interface = interface;
-
-   if ( ! driCheckDriDdxDrmVersions2( "i915",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &intelAPI);
-   if ( psp != NULL ) {
-      I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-      *driver_modes = intelFillInModes( dri_priv->cpp * 8,
-					(dri_priv->cpp == 2) ? 16 : 24,
-					(dri_priv->cpp == 2) ? 0  : 8,
-					1 );
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
-
-   return (void *) psp;
-}
+../intel/intel_screen.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_screen.h b/src/mesa/drivers/dri/i915/intel_screen.h
deleted file mode 100644
index 24cfd9bf8b2..00000000000
--- a/src/mesa/drivers/dri/i915/intel_screen.h
+++ /dev/null
@@ -1,112 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef _INTEL_INIT_H_
-#define _INTEL_INIT_H_
-
-#include <sys/time.h>
-#include "xmlconfig.h"
-#include "dri_util.h"
-#include "intel_rotate.h"
-#include "i830_common.h"
-
-
-/* This roughly corresponds to a gl_renderbuffer (Mesa 6.4) */
-typedef struct {
-   drm_handle_t handle;
-   drmSize size;        /* region size in bytes */
-   char *map;           /* memory map */
-   int offset;          /* from start of video mem, in bytes */
-   int pitch;           /* row stride, in bytes */
-} intelRegion;
-
-typedef struct 
-{
-   intelRegion front;
-   intelRegion back;
-   intelRegion rotated;
-   intelRegion depth;
-   intelRegion tex;
-   
-   int deviceID;
-   int width;
-   int height;
-   int mem;         /* unused */
-   
-   int cpp;         /* for front and back buffers */
-   int fbFormat;
-
-   int logTextureGranularity;
-   
-   __DRIscreenPrivate *driScrnPriv;
-   unsigned int sarea_priv_offset;
-
-   int drmMinor;
-
-   int irq_active;
-   int allow_batchbuffer;
-
-   struct matrix23 rotMatrix;
-
-   int current_rotation;  /* 0, 90, 180 or 270 */
-   int rotatedWidth, rotatedHeight;
-
-   /**
-   * Configuration cache with default values for all contexts
-   */
-   driOptionCache optionCache;
-} intelScreenPrivate;
-
-
-extern GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv);
-
-extern void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen);
-
-extern void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           drmI830Sarea *sarea);
-
-extern void
-intelDestroyContext(__DRIcontextPrivate *driContextPriv);
-
-extern GLboolean
-intelUnbindContext(__DRIcontextPrivate *driContextPriv);
-
-extern GLboolean
-intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv);
-
-extern void
-intelSwapBuffers(__DRIdrawablePrivate *dPriv);
-
-extern void
-intelCopySubBuffer( __DRIdrawablePrivate *dPriv, int x, int y, int w, int h );
-
-#endif
diff --git a/src/mesa/drivers/dri/i915/intel_span.c b/src/mesa/drivers/dri/i915/intel_span.c
index c3ffc4b2ac0..05e5e8e5833 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_span.c
+++ b/src/mesa/drivers/dri/i915/intel_span.c
@@ -1,258 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
-
-#include "intel_screen.h"
-
-#include "intel_span.h"
-#include "intel_ioctl.h"
-#include "swrast/swrast.h"
-
-
-#define DBG 0
-
-#define LOCAL_VARS						\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLushort p;							\
-   (void) buf; (void) p
-
-#define LOCAL_DEPTH_VARS					\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch
-
-#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS 
-
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR565(color[0],color[1],color[2])
-
-#define Y_FLIP(_y) (height - _y - 1)
-
-#define HW_LOCK()
-
-#define HW_UNLOCK()
-
-/* 16 bit, 565 rgb color spanline and pixel functions
- */
-#define WRITE_RGBA( _x, _y, r, g, b, a )				\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = ( (((int)r & 0xf8) << 8) |	\
-		                             (((int)g & 0xfc) << 3) |	\
-		                             (((int)b & 0xf8) >> 3))
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch) = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (((p >> 11) & 0x1f) * 255) / 31;			\
-   rgba[1] = (((p >>  5) & 0x3f) * 255) / 63;			\
-   rgba[2] = (((p >>  0) & 0x1f) * 255) / 31;			\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_565
-#include "spantmp.h"
-
-/* 15 bit, 555 rgb color spanline and pixel functions
- */
-#define WRITE_RGBA( _x, _y, r, g, b, a )			\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = (((r & 0xf8) << 7) |	\
-		                            ((g & 0xf8) << 3) |	\
-                         		    ((b & 0xf8) >> 3))
-
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (p >> 7) & 0xf8;					\
-   rgba[1] = (p >> 3) & 0xf8;					\
-   rgba[2] = (p << 3) & 0xf8;					\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_555
-#include "spantmp.h"
-
-/* 16 bit depthbuffer functions.
- */
-#define WRITE_DEPTH( _x, _y, d ) \
-   *(GLushort *)(buf + (_x)*2 + (_y)*pitch)  = d;
-
-#define READ_DEPTH( d, _x, _y )	\
-   d = *(GLushort *)(buf + (_x)*2 + (_y)*pitch);	 
-
-
-#define TAG(x) intel##x##_z16
-#include "depthtmp.h"
-
-
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *)drb->Base.Data +				\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLuint p;							\
-   (void) buf; (void) p
-
-#undef INIT_MONO_PIXEL
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR8888(color[0],color[1],color[2],color[3])
-
-/* 32 bit, 8888 argb color spanline and pixel functions
- */
-#define WRITE_RGBA(_x, _y, r, g, b, a)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = ((r << 16) |	\
-					  (g << 8)  |	\
-					  (b << 0)  |	\
-					  (a << 24) )
-
-#define WRITE_PIXEL(_x, _y, p)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = p
-
-
-#define READ_RGBA(rgba, _x, _y)					\
-    do {							\
-	GLuint p = *(GLuint *)(buf + _x*4 + _y*pitch);		\
-	rgba[0] = (p >> 16) & 0xff;				\
-	rgba[1] = (p >> 8)  & 0xff;				\
-	rgba[2] = (p >> 0)  & 0xff;				\
-	rgba[3] = (p >> 24) & 0xff;				\
-    } while (0)
-
-#define TAG(x) intel##x##_8888
-#include "spantmp.h"
-
-
-/* 24/8 bit interleaved depth/stencil functions
- */
-#define WRITE_DEPTH( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xff000000;					\
-   tmp |= (d) & 0xffffff;				\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
-}
-
-#define READ_DEPTH( d, _x, _y )		\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) & 0xffffff;
-
-
-#define TAG(x) intel##x##_z24_s8
-#include "depthtmp.h"
-
-#define WRITE_STENCIL( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xffffff;					\
-   tmp |= ((d)<<24);					\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
-}
-
-#define READ_STENCIL( d, _x, _y )			\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) >> 24;
-
-#define TAG(x) intel##x##_z24_s8
-#include "stenciltmp.h"
-
-
-/* Move locking out to get reasonable span performance.
- */
-void intelSpanRenderStart( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-
-   intelFlush(&intel->ctx);
-   LOCK_HARDWARE(intel);
-   intelWaitForIdle(intel);
-}
-
-void intelSpanRenderFinish( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   _swrast_flush( ctx );
-   UNLOCK_HARDWARE( intel );
-}
-
-void intelInitSpanFuncs( GLcontext *ctx )
-{
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart = intelSpanRenderStart;
-   swdd->SpanRenderFinish = intelSpanRenderFinish; 
-}
-
-
-/**
- * Plug in the Get/Put routines for the given driRenderbuffer.
- */
-void
-intelSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
-{
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 5 && vis->blueBits == 5) {
-         intelInitPointers_555(&drb->Base);
-      }
-      else if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         intelInitPointers_565(&drb->Base);
-      }
-      else {
-         assert(vis->redBits == 8);
-         assert(vis->greenBits == 8);
-         assert(vis->blueBits == 8);
-         intelInitPointers_8888(&drb->Base);
-      }
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
-      intelInitDepthPointers_z16(&drb->Base);
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
-      intelInitDepthPointers_z24_s8(&drb->Base);
-   }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
-      intelInitStencilPointers_z24_s8(&drb->Base);
-   }
-}
+../intel/intel_span.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_state.c b/src/mesa/drivers/dri/i915/intel_state.c
index e5988a5ed6c..4aa43e5f3ab 100644
--- a/src/mesa/drivers/dri/i915/intel_state.c
+++ b/src/mesa/drivers/dri/i915/intel_state.c
@@ -26,256 +26,272 @@
  **************************************************************************/
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/dd.h"
 
 #include "intel_screen.h"
 #include "intel_context.h"
+#include "intel_fbo.h"
+#include "intel_regions.h"
 #include "swrast/swrast.h"
 
-int intel_translate_compare_func( GLenum func )
+int 
+intel_translate_shadow_compare_func( GLenum func )
 {
    switch(func) {
    case GL_NEVER: 
-      return COMPAREFUNC_NEVER; 
+       return COMPAREFUNC_ALWAYS; 
    case GL_LESS: 
-      return COMPAREFUNC_LESS; 
+       return COMPAREFUNC_LEQUAL; 
    case GL_LEQUAL: 
-      return COMPAREFUNC_LEQUAL; 
+       return COMPAREFUNC_LESS;
    case GL_GREATER: 
-      return COMPAREFUNC_GREATER; 
+       return COMPAREFUNC_GEQUAL; 
    case GL_GEQUAL: 
-      return COMPAREFUNC_GEQUAL; 
+      return COMPAREFUNC_GREATER; 
    case GL_NOTEQUAL: 
-      return COMPAREFUNC_NOTEQUAL; 
-   case GL_EQUAL: 
       return COMPAREFUNC_EQUAL; 
+   case GL_EQUAL: 
+      return COMPAREFUNC_NOTEQUAL; 
    case GL_ALWAYS: 
-      return COMPAREFUNC_ALWAYS; 
+       return COMPAREFUNC_NEVER; 
+   }
+
+   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
+   return COMPAREFUNC_NEVER; 
+}
+
+int
+intel_translate_compare_func(GLenum func)
+{
+   switch (func) {
+   case GL_NEVER:
+      return COMPAREFUNC_NEVER;
+   case GL_LESS:
+      return COMPAREFUNC_LESS;
+   case GL_LEQUAL:
+      return COMPAREFUNC_LEQUAL;
+   case GL_GREATER:
+      return COMPAREFUNC_GREATER;
+   case GL_GEQUAL:
+      return COMPAREFUNC_GEQUAL;
+   case GL_NOTEQUAL:
+      return COMPAREFUNC_NOTEQUAL;
+   case GL_EQUAL:
+      return COMPAREFUNC_EQUAL;
+   case GL_ALWAYS:
+      return COMPAREFUNC_ALWAYS;
    }
 
    fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_ALWAYS; 
+   return COMPAREFUNC_ALWAYS;
 }
 
-int intel_translate_stencil_op( GLenum op )
+int
+intel_translate_stencil_op(GLenum op)
 {
-   switch(op) {
-   case GL_KEEP: 
-      return STENCILOP_KEEP; 
-   case GL_ZERO: 
-      return STENCILOP_ZERO; 
-   case GL_REPLACE: 
-      return STENCILOP_REPLACE; 
-   case GL_INCR: 
+   switch (op) {
+   case GL_KEEP:
+      return STENCILOP_KEEP;
+   case GL_ZERO:
+      return STENCILOP_ZERO;
+   case GL_REPLACE:
+      return STENCILOP_REPLACE;
+   case GL_INCR:
       return STENCILOP_INCRSAT;
-   case GL_DECR: 
+   case GL_DECR:
       return STENCILOP_DECRSAT;
    case GL_INCR_WRAP:
-      return STENCILOP_INCR; 
+      return STENCILOP_INCR;
    case GL_DECR_WRAP:
-      return STENCILOP_DECR; 
-   case GL_INVERT: 
-      return STENCILOP_INVERT; 
-   default: 
+      return STENCILOP_DECR;
+   case GL_INVERT:
+      return STENCILOP_INVERT;
+   default:
       return STENCILOP_ZERO;
    }
 }
 
-int intel_translate_blend_factor( GLenum factor )
+int
+intel_translate_blend_factor(GLenum factor)
 {
-   switch(factor) {
-   case GL_ZERO: 
-      return BLENDFACT_ZERO; 
-   case GL_SRC_ALPHA: 
-      return BLENDFACT_SRC_ALPHA; 
-   case GL_ONE: 
-      return BLENDFACT_ONE; 
-   case GL_SRC_COLOR: 
-      return BLENDFACT_SRC_COLR; 
-   case GL_ONE_MINUS_SRC_COLOR: 
-      return BLENDFACT_INV_SRC_COLR; 
-   case GL_DST_COLOR: 
-      return BLENDFACT_DST_COLR; 
-   case GL_ONE_MINUS_DST_COLOR: 
-      return BLENDFACT_INV_DST_COLR; 
+   switch (factor) {
+   case GL_ZERO:
+      return BLENDFACT_ZERO;
+   case GL_SRC_ALPHA:
+      return BLENDFACT_SRC_ALPHA;
+   case GL_ONE:
+      return BLENDFACT_ONE;
+   case GL_SRC_COLOR:
+      return BLENDFACT_SRC_COLR;
+   case GL_ONE_MINUS_SRC_COLOR:
+      return BLENDFACT_INV_SRC_COLR;
+   case GL_DST_COLOR:
+      return BLENDFACT_DST_COLR;
+   case GL_ONE_MINUS_DST_COLOR:
+      return BLENDFACT_INV_DST_COLR;
    case GL_ONE_MINUS_SRC_ALPHA:
-      return BLENDFACT_INV_SRC_ALPHA; 
-   case GL_DST_ALPHA: 
-      return BLENDFACT_DST_ALPHA; 
+      return BLENDFACT_INV_SRC_ALPHA;
+   case GL_DST_ALPHA:
+      return BLENDFACT_DST_ALPHA;
    case GL_ONE_MINUS_DST_ALPHA:
-      return BLENDFACT_INV_DST_ALPHA; 
-   case GL_SRC_ALPHA_SATURATE: 
+      return BLENDFACT_INV_DST_ALPHA;
+   case GL_SRC_ALPHA_SATURATE:
       return BLENDFACT_SRC_ALPHA_SATURATE;
    case GL_CONSTANT_COLOR:
-      return BLENDFACT_CONST_COLOR; 
+      return BLENDFACT_CONST_COLOR;
    case GL_ONE_MINUS_CONSTANT_COLOR:
       return BLENDFACT_INV_CONST_COLOR;
    case GL_CONSTANT_ALPHA:
-      return BLENDFACT_CONST_ALPHA; 
+      return BLENDFACT_CONST_ALPHA;
    case GL_ONE_MINUS_CONSTANT_ALPHA:
       return BLENDFACT_INV_CONST_ALPHA;
    }
-   
+
    fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, factor);
    return BLENDFACT_ZERO;
 }
 
-int intel_translate_logic_op( GLenum opcode )
+int
+intel_translate_logic_op(GLenum opcode)
 {
-   switch(opcode) {
-   case GL_CLEAR: 
-      return LOGICOP_CLEAR; 
-   case GL_AND: 
-      return LOGICOP_AND; 
-   case GL_AND_REVERSE: 
-      return LOGICOP_AND_RVRSE; 
-   case GL_COPY: 
-      return LOGICOP_COPY; 
-   case GL_COPY_INVERTED: 
-      return LOGICOP_COPY_INV; 
-   case GL_AND_INVERTED: 
-      return LOGICOP_AND_INV; 
-   case GL_NOOP: 
-      return LOGICOP_NOOP; 
-   case GL_XOR: 
-      return LOGICOP_XOR; 
-   case GL_OR: 
-      return LOGICOP_OR; 
-   case GL_OR_INVERTED: 
-      return LOGICOP_OR_INV; 
-   case GL_NOR: 
-      return LOGICOP_NOR; 
-   case GL_EQUIV: 
-      return LOGICOP_EQUIV; 
-   case GL_INVERT: 
-      return LOGICOP_INV; 
-   case GL_OR_REVERSE: 
-      return LOGICOP_OR_RVRSE; 
-   case GL_NAND: 
-      return LOGICOP_NAND; 
-   case GL_SET: 
-      return LOGICOP_SET; 
-   default:
+   switch (opcode) {
+   case GL_CLEAR:
+      return LOGICOP_CLEAR;
+   case GL_AND:
+      return LOGICOP_AND;
+   case GL_AND_REVERSE:
+      return LOGICOP_AND_RVRSE;
+   case GL_COPY:
+      return LOGICOP_COPY;
+   case GL_COPY_INVERTED:
+      return LOGICOP_COPY_INV;
+   case GL_AND_INVERTED:
+      return LOGICOP_AND_INV;
+   case GL_NOOP:
+      return LOGICOP_NOOP;
+   case GL_XOR:
+      return LOGICOP_XOR;
+   case GL_OR:
+      return LOGICOP_OR;
+   case GL_OR_INVERTED:
+      return LOGICOP_OR_INV;
+   case GL_NOR:
+      return LOGICOP_NOR;
+   case GL_EQUIV:
+      return LOGICOP_EQUIV;
+   case GL_INVERT:
+      return LOGICOP_INV;
+   case GL_OR_REVERSE:
+      return LOGICOP_OR_RVRSE;
+   case GL_NAND:
+      return LOGICOP_NAND;
+   case GL_SET:
       return LOGICOP_SET;
-   }
-}
-
-static void intelDrawBuffer(GLcontext *ctx, GLenum mode )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   int front = 0;
- 
-   if (!ctx->DrawBuffer)
-      return;
-
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-      front = 1;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
-   case BUFFER_BIT_BACK_LEFT:
-      front = 0;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
    default:
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_TRUE );
-      return;
-   }
-
-   if ( intel->sarea->pf_current_page == 1 ) 
-      front ^= 1;
-   
-   intelSetFrontClipRects( intel );
-
-   if (front) {
-      intel->drawRegion = &intel->intelScreen->front;
-      intel->readRegion = &intel->intelScreen->front;
-   } else {
-      intel->drawRegion = &intel->intelScreen->back;
-      intel->readRegion = &intel->intelScreen->back;
+      return LOGICOP_SET;
    }
-
-   intel->vtbl.set_color_region( intel, intel->drawRegion );
-}
-
-static void intelReadBuffer( GLcontext *ctx, GLenum mode )
-{
-   /* nothing, until we implement h/w glRead/CopyPixels or CopyTexImage */
 }
 
 
-static void intelClearColor(GLcontext *ctx, const GLfloat color[4])
+static void
+intelClearColor(GLcontext * ctx, const GLfloat color[4])
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   intelScreenPrivate *screen = intel->intelScreen;
+   struct intel_context *intel = intel_context(ctx);
+   GLubyte clear[4];
 
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_red, color[0]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_green, color[1]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_blue, color[2]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_alpha, color[3]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
 
-   intel->ClearColor = INTEL_PACKCOLOR(screen->fbFormat,
-				       intel->clear_red, 
-				       intel->clear_green, 
-				       intel->clear_blue, 
-				       intel->clear_alpha);
+   /* compute both 32 and 16-bit clear values */
+   intel->ClearColor8888 = INTEL_PACKCOLOR8888(clear[0], clear[1],
+                                               clear[2], clear[3]);
+   intel->ClearColor565 = INTEL_PACKCOLOR565(clear[0], clear[1], clear[2]);
 }
 
 
-static void intelCalcViewport( GLcontext *ctx )
+/**
+ * Update the viewport transformation matrix.  Depends on:
+ *  - viewport pos/size
+ *  - depthrange
+ *  - window pos/size or FBO size
+ */
+static void
+intelCalcViewport(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    const GLfloat *v = ctx->Viewport._WindowMap.m;
+   const GLfloat depthScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
    GLfloat *m = intel->ViewportMatrix.m;
-   GLint h = 0;
+   GLfloat yScale, yBias;
 
-   if (intel->driDrawable) 
-      h = intel->driDrawable->h + SUBPIXEL_Y;
+   if (ctx->DrawBuffer->Name) {
+      /* User created FBO */
+      struct intel_renderbuffer *irb
+         = intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
+      if (irb && !irb->RenderToTexture) {
+         /* y=0=top */
+         yScale = -1.0;
+         yBias = irb->Base.Height;
+      }
+      else {
+         /* y=0=bottom */
+         yScale = 1.0;
+         yBias = 0.0;
+      }
+   }
+   else {
+      /* window buffer, y=0=top */
+      yScale = -1.0;
+      yBias = (intel->driDrawable) ? intel->driDrawable->h : 0.0F;
+   }
 
-   /* See also intel_translate_vertex.  SUBPIXEL adjustments can be done
-    * via state vars, too.
-    */
-   m[MAT_SX] =   v[MAT_SX];
-   m[MAT_TX] =   v[MAT_TX] + SUBPIXEL_X;
-   m[MAT_SY] = - v[MAT_SY];
-   m[MAT_TY] = - v[MAT_TY] + h;
-   m[MAT_SZ] =   v[MAT_SZ] * intel->depth_scale;
-   m[MAT_TZ] =   v[MAT_TZ] * intel->depth_scale;
+   m[MAT_SX] = v[MAT_SX];
+   m[MAT_TX] = v[MAT_TX];
+
+   m[MAT_SY] = v[MAT_SY] * yScale;
+   m[MAT_TY] = v[MAT_TY] * yScale + yBias;
+
+   m[MAT_SZ] = v[MAT_SZ] * depthScale;
+   m[MAT_TZ] = v[MAT_TZ] * depthScale;
 }
 
-static void intelViewport( GLcontext *ctx,
-			  GLint x, GLint y,
-			  GLsizei width, GLsizei height )
+static void
+intelViewport(GLcontext * ctx,
+              GLint x, GLint y, GLsizei width, GLsizei height)
 {
-   intelCalcViewport( ctx );
+   intelCalcViewport(ctx);
+
+   intel_viewport(ctx, x, y, width, height);
 }
 
-static void intelDepthRange( GLcontext *ctx,
-			    GLclampd nearval, GLclampd farval )
+static void
+intelDepthRange(GLcontext * ctx, GLclampd nearval, GLclampd farval)
 {
-   intelCalcViewport( ctx );
+   intelCalcViewport(ctx);
 }
 
 /* Fallback to swrast for select and feedback.
  */
-static void intelRenderMode( GLcontext *ctx, GLenum mode )
+static void
+intelRenderMode(GLcontext * ctx, GLenum mode)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   FALLBACK( intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER) );
+   struct intel_context *intel = intel_context(ctx);
+   FALLBACK(intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER));
 }
 
 
-void intelInitStateFuncs( struct dd_function_table *functions )
+void
+intelInitStateFuncs(struct dd_function_table *functions)
 {
-   functions->DrawBuffer = intelDrawBuffer;
-   functions->ReadBuffer = intelReadBuffer;
    functions->RenderMode = intelRenderMode;
    functions->Viewport = intelViewport;
    functions->DepthRange = intelDepthRange;
    functions->ClearColor = intelClearColor;
 }
-
diff --git a/src/mesa/drivers/dri/i915/intel_structs.h b/src/mesa/drivers/dri/i915/intel_structs.h
new file mode 100644
index 00000000000..522e3bd92c2
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_structs.h
@@ -0,0 +1,132 @@
+#ifndef INTEL_STRUCTS_H
+#define INTEL_STRUCTS_H
+
+struct br0 {
+   GLuint length:8;
+   GLuint pad0:3;
+   GLuint dst_tiled:1;
+   GLuint pad1:8;
+   GLuint write_rgb:1;
+   GLuint write_alpha:1;
+   GLuint opcode:7;
+   GLuint client:3;
+};
+
+   
+struct br13 {
+   GLint dest_pitch:16;
+   GLuint rop:8;
+   GLuint color_depth:2;
+   GLuint pad1:3;
+   GLuint mono_source_transparency:1;
+   GLuint clipping_enable:1;
+   GLuint pad0:1;
+};
+
+
+
+/* This is an attempt to move some of the 2D interaction in this
+ * driver to using structs for packets rather than a bunch of #defines
+ * and dwords.
+ */
+struct xy_color_blit {
+   struct br0 br0;
+   struct br13 br13;
+
+   struct {
+      GLuint dest_x1:16;
+      GLuint dest_y1:16;
+   } dw2;
+
+   struct {
+      GLuint dest_x2:16;
+      GLuint dest_y2:16;
+   } dw3;
+   
+   GLuint dest_base_addr;
+   GLuint color;
+};
+
+struct xy_src_copy_blit {
+   struct br0 br0;
+   struct br13 br13;
+
+   struct {
+      GLuint dest_x1:16;
+      GLuint dest_y1:16;
+   } dw2;
+
+   struct {
+      GLuint dest_x2:16;
+      GLuint dest_y2:16;
+   } dw3;
+   
+   GLuint dest_base_addr;
+
+   struct {
+      GLuint src_x1:16;
+      GLuint src_y1:16;
+   } dw5;
+
+   struct {
+      GLint src_pitch:16;
+      GLuint pad:16;
+   } dw6;
+   
+   GLuint src_base_addr;
+};
+
+struct xy_setup_blit {
+   struct br0 br0;
+   struct br13 br13;
+
+   struct {
+      GLuint clip_x1:16;
+      GLuint clip_y1:16;
+   } dw2;
+
+   struct {
+      GLuint clip_x2:16;
+      GLuint clip_y2:16;
+   } dw3;
+      
+   GLuint dest_base_addr;
+   GLuint background_color;
+   GLuint foreground_color;
+   GLuint pattern_base_addr;
+};
+
+
+struct xy_text_immediate_blit {
+   struct {
+      GLuint length:8;
+      GLuint pad2:3;
+      GLuint dst_tiled:1;
+      GLuint pad1:4;
+      GLuint byte_packed:1;
+      GLuint pad0:5;
+      GLuint opcode:7;
+      GLuint client:3;
+   } dw0;
+
+   struct {
+      GLuint dest_x1:16;
+      GLuint dest_y1:16;
+   } dw1;
+
+   struct {
+      GLuint dest_x2:16;
+      GLuint dest_y2:16;
+   } dw2;   
+
+   /* Src bitmap data follows as inline dwords.
+    */
+};
+
+
+#define CLIENT_2D 0x2
+#define OPCODE_XY_SETUP_BLT 0x1
+#define OPCODE_XY_COLOR_BLT 0x50
+#define OPCODE_XY_TEXT_IMMEDIATE_BLT 0x31
+
+#endif
diff --git a/src/mesa/drivers/dri/i915/intel_tex.c b/src/mesa/drivers/dri/i915/intel_tex.c
index 5bd280652af..d77ce749a3e 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_tex.c
+++ b/src/mesa/drivers/dri/i915/intel_tex.c
@@ -1,877 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texmem.h"
-#include "texobj.h"
-#include "swrast/swrast.h"
-
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-#include "intel_context.h"
-#include "intel_tex.h"
-#include "intel_ioctl.h"
-
-
-
-static GLboolean
-intelValidateClientStorage( intelContextPtr intel, GLenum target,
-			    GLint internalFormat,
-			    GLint srcWidth, GLint srcHeight, 
-			    GLenum format, GLenum type,  const void *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage)
-
-{
-   GLcontext *ctx = &intel->ctx;
-   int texelBytes;
-
-   if (0)
-      fprintf(stderr, "intformat %s format %s type %s\n",
-	      _mesa_lookup_enum_by_nr( internalFormat ),
-	      _mesa_lookup_enum_by_nr( format ),
-	      _mesa_lookup_enum_by_nr( type ));
-
-   if (!ctx->Unpack.ClientStorage)
-      return 0;
-
-   if (ctx->_ImageTransferState ||
-       texImage->IsCompressed ||
-       texObj->GenerateMipmap)
-      return 0;
-
-
-   /* This list is incomplete
-    */
-   switch ( internalFormat ) {
-   case GL_RGBA:
-      if ( format == GL_BGRA && type == GL_UNSIGNED_INT_8_8_8_8_REV ) {
-	 texImage->TexFormat = &_mesa_texformat_argb8888;
-	 texelBytes = 4;
-      }
-      else
-	 return 0;
-      break;
-
-   case GL_RGB:
-      if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
-	 texImage->TexFormat = &_mesa_texformat_rgb565;
-	 texelBytes = 2;
-      }
-      else
-	 return 0;
-      break;
-
-   case GL_YCBCR_MESA:
-      if ( format == GL_YCBCR_MESA && 
-	   type == GL_UNSIGNED_SHORT_8_8_REV_APPLE ) {
-	 texImage->TexFormat = &_mesa_texformat_ycbcr_rev;
-	 texelBytes = 2;
-      }
-      else if ( format == GL_YCBCR_MESA && 
-		(type == GL_UNSIGNED_SHORT_8_8_APPLE || 
-		 type == GL_UNSIGNED_BYTE)) {
-	 texImage->TexFormat = &_mesa_texformat_ycbcr;
-	 texelBytes = 2;
-      }
-      else
-	 return 0;
-      break;
-      
-	 
-   default:
-      return 0;
-   }
-
-   /* Could deal with these packing issues, but currently don't:
-    */
-   if (packing->SkipPixels || 
-       packing->SkipRows || 
-       packing->SwapBytes ||
-       packing->LsbFirst) {
-      return 0;
-   }
-
-   {      
-      GLint srcRowStride = _mesa_image_row_stride(packing, srcWidth,
-						  format, type);
-
-      
-      if (0)
-	 fprintf(stderr, "%s: srcRowStride %d/%x\n", 
-		 __FUNCTION__, srcRowStride, srcRowStride);
-
-      /* Could check this later in upload, pitch restrictions could be
-       * relaxed, but would need to store the image pitch somewhere,
-       * as packing details might change before image is uploaded:
-       */
-      if (!intelIsAgpMemory( intel, pixels, srcHeight * srcRowStride ) ||
-	  (srcRowStride & 63))
-	 return 0;
-
-
-      /* Have validated that _mesa_transfer_teximage would be a straight
-       * memcpy at this point.  NOTE: future calls to TexSubImage will
-       * overwrite the client data.  This is explicitly mentioned in the
-       * extension spec.
-       */
-      texImage->Data = (void *)pixels;
-      texImage->IsClientData = GL_TRUE;
-      texImage->RowStride = srcRowStride / texelBytes;
-      return 1;
-   }
-}
-
- 
-
-static void intelTexImage1D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-
-   texImage->IsClientData = GL_FALSE;
-
-   _mesa_store_teximage1d( ctx, target, level, internalFormat,
-			   width, border, format, type,
-			   pixels, packing, texObj, texImage );
-
-   t->dirty_images[0] |= (1 << level);
-}
-
-static void intelTexSubImage1D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset,
-				GLsizei width,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-
-   _mesa_store_texsubimage1d(ctx, target, level, xoffset, width, 
-			     format, type, pixels, packing, texObj,
-			     texImage);
-}
-
-
-/* Handles 2D, CUBE, RECT:
- */
-static void intelTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint height, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
-
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   if (intelValidateClientStorage( INTEL_CONTEXT(ctx), target, 
-				   internalFormat, 
-				   width, height, 
-				   format, type, pixels, 
-				   packing, texObj, texImage)) {
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "%s: Using client storage\n", __FUNCTION__); 
-   }
-   else {
-      _mesa_store_teximage2d( ctx, target, level, internalFormat,
-			      width, height, border, format, type,
-			      pixels, packing, texObj, texImage );
-
-      t->dirty_images[face] |= (1 << level);
-   }
-}
-
-static void intelTexSubImage2D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset, GLint yoffset,
-			       GLsizei width, GLsizei height,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
-
-   if (texImage->IsClientData &&
-       (char *)pixels == (char *)texImage->Data + 
-       ((xoffset + yoffset * texImage->RowStride) * 
-	texImage->TexFormat->TexelBytes)) {
-
-      /* Notification only - no upload required */
-   }
-   else {
-      assert( t ); /* this _should_ be true */
-      intelFlush( ctx );
-      driSwapOutTextureObject( t );
-
-      _mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width, 
-				height, format, type, pixels, packing, texObj,
-				texImage);
-
-      t->dirty_images[face] |= (1 << level);
-   }
-}
-
-static void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-                              GLint internalFormat,
-                              GLint width, GLint height, GLint border,
-                              GLsizei imageSize, const GLvoid *data,
-                              struct gl_texture_object *texObj,
-                              struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
-
-   assert(t);
-   intelFlush( ctx );
-   
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-     fprintf(stderr, "%s: Using normal storage\n", __FUNCTION__); 
-   
-   _mesa_store_compressed_teximage2d(ctx, target, level, internalFormat, width,
-				     height, border, imageSize, data, texObj, texImage);
-   
-   t->dirty_images[face] |= (1 << level);
-}
-
-
-static void intelCompressedTexSubImage2D( GLcontext *ctx, GLenum target, GLint level,
-                                 GLint xoffset, GLint yoffset,
-                                 GLsizei width, GLsizei height,
-                                 GLenum format,
-                                 GLsizei imageSize, const GLvoid *data,
-                                 struct gl_texture_object *texObj,
-                                 struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
-
-   assert( t ); /* this _should_ be true */
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-   
-   _mesa_store_compressed_texsubimage2d(ctx, target, level, xoffset, yoffset, width,
-					height, format, imageSize, data, texObj, texImage);
-   
-   t->dirty_images[face] |= (1 << level);
-}
-
-
-static void intelTexImage3D( GLcontext *ctx, GLenum target, GLint level,
-                            GLint internalFormat,
-                            GLint width, GLint height, GLint depth,
-                            GLint border,
-                            GLenum format, GLenum type, const GLvoid *pixels,
-                            const struct gl_pixelstore_attrib *packing,
-                            struct gl_texture_object *texObj,
-                            struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert(t);
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   _mesa_store_teximage3d(ctx, target, level, internalFormat,
-			  width, height, depth, border,
-			  format, type, pixels,
-			  &ctx->Unpack, texObj, texImage);
-   
-   t->dirty_images[0] |= (1 << level);
-}
-
-
-static void
-intelTexSubImage3D( GLcontext *ctx, GLenum target, GLint level,
-                   GLint xoffset, GLint yoffset, GLint zoffset,
-                   GLsizei width, GLsizei height, GLsizei depth,
-                   GLenum format, GLenum type,
-                   const GLvoid *pixels,
-                   const struct gl_pixelstore_attrib *packing,
-                   struct gl_texture_object *texObj,
-                   struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert( t ); /* this _should_ be true */
-   driSwapOutTextureObject( t );
-
-   _mesa_store_texsubimage3d(ctx, target, level, xoffset, yoffset, zoffset,
-                             width, height, depth,
-                             format, type, pixels, packing, texObj, texImage);
-
-   t->dirty_images[0] |= (1 << level);
-}
-
-
-
-
-static void intelDeleteTexture( GLcontext *ctx, struct gl_texture_object *tObj )
-{
-   driTextureObject * t = (driTextureObject *) tObj->DriverData;
-
-   if ( t != NULL ) {
-      intelFlush( ctx );
-      driDestroyTextureObject( t );
-   }
-   
-   /* Free mipmap images and the texture object itself */
-   _mesa_delete_texture_object(ctx, tObj);
-}
-
-
-static const struct gl_texture_format *
-intelChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
-			 GLenum format, GLenum type )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   const GLboolean do32bpt = ( intel->intelScreen->cpp == 4 &&
-			       intel->intelScreen->tex.size > 4*1024*1024);
-
-   switch ( internalFormat ) {
-   case 4:
-   case GL_RGBA:
-   case GL_COMPRESSED_RGBA:
-      if ( format == GL_BGRA ) {
-	 if ( type == GL_UNSIGNED_INT_8_8_8_8_REV ) {
-	    return &_mesa_texformat_argb8888;
-	 }
-         else if ( type == GL_UNSIGNED_SHORT_4_4_4_4_REV ) {
-            return &_mesa_texformat_argb4444;
-	 }
-         else if ( type == GL_UNSIGNED_SHORT_1_5_5_5_REV ) {
-	    return &_mesa_texformat_argb1555;
-	 }
-      }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
-
-   case 3:
-   case GL_RGB:
-   case GL_COMPRESSED_RGB:
-      if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
-	 return &_mesa_texformat_rgb565;
-      }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
-
-   case GL_RGBA8:
-   case GL_RGB10_A2:
-   case GL_RGBA12:
-   case GL_RGBA16:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
-
-   case GL_RGBA4:
-   case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
-
-   case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
-
-   case GL_RGB8:
-   case GL_RGB10:
-   case GL_RGB12:
-   case GL_RGB16:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
-
-   case GL_RGB5:
-   case GL_RGB4:
-   case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
-
-   case GL_ALPHA:
-   case GL_ALPHA4:
-   case GL_ALPHA8:
-   case GL_ALPHA12:
-   case GL_ALPHA16:
-   case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
-
-   case 1:
-   case GL_LUMINANCE:
-   case GL_LUMINANCE4:
-   case GL_LUMINANCE8:
-   case GL_LUMINANCE12:
-   case GL_LUMINANCE16:
-   case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
-
-   case 2:
-   case GL_LUMINANCE_ALPHA:
-   case GL_LUMINANCE4_ALPHA4:
-   case GL_LUMINANCE6_ALPHA2:
-   case GL_LUMINANCE8_ALPHA8:
-   case GL_LUMINANCE12_ALPHA4:
-   case GL_LUMINANCE12_ALPHA12:
-   case GL_LUMINANCE16_ALPHA16:
-   case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
-
-   case GL_INTENSITY:
-   case GL_INTENSITY4:
-   case GL_INTENSITY8:
-   case GL_INTENSITY12:
-   case GL_INTENSITY16:
-   case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
-
-   case GL_YCBCR_MESA:
-      if (type == GL_UNSIGNED_SHORT_8_8_MESA ||
-	  type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
-      else
-         return &_mesa_texformat_ycbcr_rev;
-
-   case GL_COMPRESSED_RGB_FXT1_3DFX:
-     return &_mesa_texformat_rgb_fxt1;
-   case GL_COMPRESSED_RGBA_FXT1_3DFX:
-     return &_mesa_texformat_rgba_fxt1;
-
-   case GL_RGB_S3TC:
-   case GL_RGB4_S3TC:
-   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgb_dxt1;
-
-   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgba_dxt1;
-
-   case GL_RGBA_S3TC:
-   case GL_RGBA4_S3TC:
-   case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-     return &_mesa_texformat_rgba_dxt3;
-
-   case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
-
-   case GL_DEPTH_COMPONENT:
-   case GL_DEPTH_COMPONENT16:
-   case GL_DEPTH_COMPONENT24:
-   case GL_DEPTH_COMPONENT32:
-      return &_mesa_texformat_z16;
-
-   default:
-      fprintf(stderr, "unexpected texture format %s in %s\n", 
-	      _mesa_lookup_enum_by_nr(internalFormat),
-	      __FUNCTION__);
-      return NULL;
-   }
-
-   return NULL; /* never get here */
-}
-
-
-
-void intelDestroyTexObj(intelContextPtr intel, intelTextureObjectPtr t)
-{
-   unsigned   i;
-
-   if ( intel == NULL ) 
-      return;
-
-   if ( t->age > intel->dirtyAge )
-      intel->dirtyAge = t->age;
-
-   for ( i = 0 ; i < MAX_TEXTURE_UNITS ; i++ ) {
-      if ( t == intel->CurrentTexObj[ i ] ) 
-	 intel->CurrentTexObj[ i ] = NULL;
-   }
-}
-
-
-
-/* Upload an image from mesa's internal copy.  Image may be 1D, 2D or
- * 3D.  Cubemaps are expanded elsewhere.
- */
-static void intelUploadTexImage( intelContextPtr intel,
-				 intelTextureObjectPtr t,
-				 const struct gl_texture_image *image,
-				 const GLuint offset )
-{
-
-   if (!image || !image->Data) 
-      return;
-
-   if (image->Depth == 1 && image->IsClientData) {
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "Blit uploading\n");
-
-      /* Do it with a blit.
-       */
-      intelEmitCopyBlitLocked( intel,
-			       image->TexFormat->TexelBytes,
-			       image->RowStride, /* ? */
-			       intelGetMemoryOffsetMESA( NULL, 0, image->Data ),
-			       t->Pitch / image->TexFormat->TexelBytes,
-			       intelGetMemoryOffsetMESA( NULL, 0, t->BufAddr + offset ),
-			       0, 0,
-			       0, 0,
-			       image->Width,
-			       image->Height);
-   }
-   else if (image->IsCompressed) {
-      GLuint row_len = 0;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint j;
-
-      /* must always copy whole blocks (8/16 bytes) */
-      switch (image->InternalFormat) {
-	case GL_COMPRESSED_RGB_FXT1_3DFX:
-	case GL_COMPRESSED_RGBA_FXT1_3DFX:
-	case GL_RGB_S3TC:
-	case GL_RGB4_S3TC:
-	case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-	case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-	  row_len = (image->Width * 2 + 7) & ~7;
-	  break;
-	case GL_RGBA_S3TC:
-	case GL_RGBA4_S3TC:
-	case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-	case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-	  row_len = (image->Width * 4 + 15) & ~15;
-	  break;
-	default:
-	  fprintf(stderr,"Internal Compressed format not supported %d\n", image->InternalFormat);
-	  break;
-      }
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len) {
-	 for (j = 0 ; j < (image->Height + 3)/4 ; j++, dst += (t->Pitch)) {
-	   __memcpy(dst, src, row_len );
-	   src += row_len;
-	 }
-      }
-   }
-   /* Time for another vtbl entry:
-    */
-   else if (intel->intelScreen->deviceID == PCI_CHIP_I945_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_I945_GM ||
-            intel->intelScreen->deviceID == PCI_CHIP_I945_GME ||
-            intel->intelScreen->deviceID == PCI_CHIP_G33_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_Q33_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_Q35_G) {
-      GLuint row_len = image->Width * image->TexFormat->TexelBytes;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint d, j;
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len == t->Pitch) {
-	 memcpy( dst, src, row_len * image->Height * image->Depth );
-      }
-      else { 
-	 GLuint x = 0, y = 0;
-
-	 for (d = 0 ; d < image->Depth ; d++) {
-	    GLubyte *dst0 = dst + x + y * t->Pitch;
-
-	    for (j = 0 ; j < image->Height ; j++) {
-	       __memcpy(dst0, src, row_len );
-	       src += row_len;
-	       dst0 += t->Pitch;
-	    }
-
-	    x += MIN2(4, row_len); /* Guess: 4 byte minimum alignment */
-	    if (x > t->Pitch) {
-	       x = 0;
-	       y += image->Height;
-	    }
-	 }
-      }
-
-   }
-   else {
-      GLuint row_len = image->Width * image->TexFormat->TexelBytes;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint d, j;
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len == t->Pitch) {
-	 for (d = 0; d < image->Depth; d++) {
-	    memcpy( dst, src, t->Pitch * image->Height );
-	    dst += t->depth_pitch;
-	    src += row_len * image->Height;
-	 }
-      }
-      else { 
-	 for (d = 0 ; d < image->Depth ; d++) {
-	    for (j = 0 ; j < image->Height ; j++) {
-	       __memcpy(dst, src, row_len );
-	       src += row_len;
-	       dst += t->Pitch;
-	    }
-
-	    dst += t->depth_pitch - (t->Pitch * image->Height);
-	 }
-      }
-   }
-}
-
-
-
-int intelUploadTexImages( intelContextPtr intel, 
-			  intelTextureObjectPtr t,
-			  GLuint face)
-{
-   const int numLevels = t->base.lastLevel - t->base.firstLevel + 1;
-   const struct gl_texture_image *firstImage = t->image[face][t->base.firstLevel].image;
-   int pitch = firstImage->RowStride * firstImage->TexFormat->TexelBytes;
-
-   /* Can we texture out of the existing client data? */
-   if ( numLevels == 1 &&
-	firstImage->IsClientData &&
-	(pitch & 3) == 0) {
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "AGP texturing from client memory\n");
-
-      t->TextureOffset = intelAgpOffsetFromVirtual( intel, firstImage->Data );
-      t->BufAddr = 0;
-      t->dirty = ~0;
-      return GL_TRUE;
-   }
-   else {
-      if (INTEL_DEBUG & DEBUG_TEXTURE) 
-	 fprintf(stderr, "Uploading client data to agp\n");
-
-      INTEL_FIREVERTICES( intel );
-      LOCK_HARDWARE( intel );
-
-      if ( t->base.memBlock == NULL ) {
-	 int heap;
-
-	 heap = driAllocateTexture( intel->texture_heaps, intel->nr_heaps,
-				    (driTextureObject *) t );
-	 if ( heap == -1 ) {
-	    UNLOCK_HARDWARE( intel );
-	    return GL_FALSE;
-	 }
-
-	 /* Set the base offset of the texture image */
-	 t->BufAddr = (GLubyte *) (intel->intelScreen->tex.map + 
-				   t->base.memBlock->ofs);
-	 t->TextureOffset = intel->intelScreen->tex.offset + t->base.memBlock->ofs;
-	 t->dirty = ~0;
-      }
-
-
-      /* Let the world know we've used this memory recently.
-       */
-      driUpdateTextureLRU( (driTextureObject *) t );
-
-
-      /* Upload any images that are new */
-      if (t->base.dirty_images[face]) {
-	 int i;
-
- 	 intelWaitForIdle( intel );
-	    
-	 for (i = 0 ; i < numLevels ; i++) { 
-	    int level = i + t->base.firstLevel;
-
-	    if (t->base.dirty_images[face] & (1<<level)) {
-
-	       const struct gl_texture_image *image = t->image[face][i].image;
-	       GLuint offset = t->image[face][i].offset;
-
-     	       if (INTEL_DEBUG & DEBUG_TEXTURE)
-	          fprintf(stderr, "upload level %d, offset %x\n", 
-			  level, offset);
-
-	       intelUploadTexImage( intel, t, image, offset );
-	    }
-	 }
-	 t->base.dirty_images[face] = 0;
-	 intel->perf_boxes |= I830_BOX_TEXTURE_LOAD;
-      }
-      
-      UNLOCK_HARDWARE( intel );
-      return GL_TRUE;
-   }
-}
-
-/**
- * Allocate a new texture object.
- * Called via ctx->Driver.NewTextureObject.
- * Note: this function will be called during context creation to
- * allocate the default texture objects.
- * Note: we could use containment here to 'derive' the driver-specific
- * texture object from the core mesa gl_texture_object.  Not done at this time.
- */
-static struct gl_texture_object *
-intelNewTextureObject( GLcontext *ctx, GLuint name, GLenum target )
-{
-   struct gl_texture_object *obj = _mesa_new_texture_object(ctx, name, target);
-   INTEL_CONTEXT(ctx)->vtbl.alloc_tex_obj( obj );
-   return obj;
-}
-
-
-void intelInitTextureFuncs( struct dd_function_table *functions )
-{
-   functions->NewTextureObject          = intelNewTextureObject;
-   functions->ChooseTextureFormat       = intelChooseTextureFormat;
-   functions->TexImage1D                = intelTexImage1D;
-   functions->TexImage2D                = intelTexImage2D;
-   functions->TexImage3D                = intelTexImage3D;
-   functions->TexSubImage1D             = intelTexSubImage1D;
-   functions->TexSubImage2D             = intelTexSubImage2D;
-   functions->TexSubImage3D             = intelTexSubImage3D;
-   functions->CopyTexImage1D            = _swrast_copy_teximage1d;
-   functions->CopyTexImage2D            = _swrast_copy_teximage2d;
-   functions->CopyTexSubImage1D         = _swrast_copy_texsubimage1d;
-   functions->CopyTexSubImage2D         = _swrast_copy_texsubimage2d;
-   functions->CopyTexSubImage3D         = _swrast_copy_texsubimage3d;
-   functions->DeleteTexture             = intelDeleteTexture;
-   functions->UpdateTexturePalette      = NULL;
-   functions->IsTextureResident         = driIsTextureResident;
-   functions->TestProxyTexImage         = _mesa_test_proxy_teximage;
-   functions->DeleteTexture             = intelDeleteTexture;
-   functions->CompressedTexImage2D      = intelCompressedTexImage2D;
-   functions->CompressedTexSubImage2D   = intelCompressedTexSubImage2D;
-}
+../intel/intel_tex.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_copy.c b/src/mesa/drivers/dri/i915/intel_tex_copy.c
new file mode 120000
index 00000000000..87196c5d1ed
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_copy.c
@@ -0,0 +1 @@
+../intel/intel_tex_copy.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_format.c b/src/mesa/drivers/dri/i915/intel_tex_format.c
new file mode 120000
index 00000000000..3415f754705
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_format.c
@@ -0,0 +1 @@
+../intel/intel_tex_format.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_image.c b/src/mesa/drivers/dri/i915/intel_tex_image.c
new file mode 120000
index 00000000000..567abe4974e
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_image.c
@@ -0,0 +1 @@
+../intel/intel_tex_image.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_layout.c b/src/mesa/drivers/dri/i915/intel_tex_layout.c
new file mode 120000
index 00000000000..fe61b441945
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_layout.c
@@ -0,0 +1 @@
+../intel/intel_tex_layout.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_subimage.c b/src/mesa/drivers/dri/i915/intel_tex_subimage.c
new file mode 120000
index 00000000000..b3a8a3d7ca7
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_subimage.c
@@ -0,0 +1 @@
+../intel/intel_tex_subimage.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tex_validate.c b/src/mesa/drivers/dri/i915/intel_tex_validate.c
new file mode 120000
index 00000000000..41a75674c27
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_tex_validate.c
@@ -0,0 +1 @@
+../intel/intel_tex_validate.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_texmem.c b/src/mesa/drivers/dri/i915/intel_texmem.c
deleted file mode 100644
index 09beec95b8c..00000000000
--- a/src/mesa/drivers/dri/i915/intel_texmem.c
+++ /dev/null
@@ -1,72 +0,0 @@
-#include "texmem.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "macros.h"
-
-#include "intel_tex.h"
-
-static GLuint
-driLog2( GLuint n )
-{
-   GLuint log2;
-
-   for ( log2 = 1 ; n > 1 ; log2++ ) {
-      n >>= 1;
-   }
-
-   return log2;
-}
-
-static void calculate_heap_size( driTexHeap * heap, unsigned size, 
-				 unsigned nr_regions, unsigned alignmentShift )
-{
-   unsigned     l;
-
-   l = driLog2( (size - 1) / nr_regions );
-   if ( l < alignmentShift )
-   {
-      l = alignmentShift;
-   }
-
-   heap->logGranularity = l;
-   heap->size = size & ~((1L << l) - 1);
-}
-
-
-GLboolean 
-intel_driReinitTextureHeap( driTexHeap *heap,
-			    unsigned size )
-{
-   driTextureObject *t, *tmp;
-
-   /* Kick out everything:
-    */
-   foreach_s ( t, tmp, & heap->texture_objects ) {
-      if ( t->tObj != NULL ) {
-	 driSwapOutTextureObject( t );
-      }
-      else {
-	 driDestroyTextureObject( t );
-      }
-   }
-   
-   /* Destroy the memory manager:
-    */
-   mmDestroy( heap->memory_heap );
-      
-   /* Recreate the memory manager:
-    */
-   calculate_heap_size(heap, size, heap->nrRegions, heap->alignmentShift);
-   heap->memory_heap = mmInit( 0, heap->size );
-   if ( heap->memory_heap == NULL ) {
-      fprintf(stderr, "driReinitTextureHeap: couldn't recreate memory heap\n");
-      FREE( heap );
-      return GL_FALSE;
-   }
-
-   make_empty_list( & heap->texture_objects );
-
-   return GL_TRUE;
-}
-
-
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index b2787ee60ac..c4708dc7abc 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -25,11 +25,19 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
+/** @file intel_tris.c
+ *
+ * This file contains functions for managing the vertex buffer and emitting
+ * primitives into it.
+ */
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/texobj.h"
+#include "main/state.h"
+#include "main/dd.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -38,19 +46,288 @@
 #include "tnl/t_vertex.h"
 
 #include "intel_screen.h"
+#include "intel_context.h"
 #include "intel_tris.h"
 #include "intel_batchbuffer.h"
+#include "intel_buffers.h"
 #include "intel_reg.h"
 #include "intel_span.h"
+#include "intel_tex.h"
+#include "intel_chipset.h"
+#include "i830_context.h"
+#include "i830_reg.h"
+
+static void intelRenderPrimitive(GLcontext * ctx, GLenum prim);
+static void intelRasterPrimitive(GLcontext * ctx, GLenum rprim,
+                                 GLuint hwprim);
+
+static void
+intel_flush_inline_primitive(struct intel_context *intel)
+{
+   GLuint used = intel->batch->ptr - intel->prim.start_ptr;
+
+   assert(intel->prim.primitive != ~0);
+
+/*    _mesa_printf("/\n"); */
+
+   if (used < 8)
+      goto do_discard;
+
+   *(int *) intel->prim.start_ptr = (_3DPRIMITIVE |
+                                     intel->prim.primitive | (used / 4 - 2));
+
+   goto finished;
+
+ do_discard:
+   intel->batch->ptr -= used;
+
+ finished:
+   intel->prim.primitive = ~0;
+   intel->prim.start_ptr = 0;
+   intel->prim.flush = 0;
+}
+
+static void intel_start_inline(struct intel_context *intel, uint32_t prim)
+{
+   BATCH_LOCALS;
+   uint32_t batch_flags = LOOP_CLIPRECTS;
+
+   intel_wait_flips(intel);
+   intel->vtbl.emit_state(intel);
+
+   intel->no_batch_wrap = GL_TRUE;
+
+   /*_mesa_printf("%s *", __progname);*/
+
+   /* Emit a slot which will be filled with the inline primitive
+    * command later.
+    */
+   BEGIN_BATCH(2, batch_flags);
+   OUT_BATCH(0);
+
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
+
+   intel->prim.start_ptr = intel->batch->ptr;
+   intel->prim.primitive = prim;
+   intel->prim.flush = intel_flush_inline_primitive;
+
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+   intel->no_batch_wrap = GL_FALSE;
+/*    _mesa_printf(">"); */
+}
+
+static void intel_wrap_inline(struct intel_context *intel)
+{
+   GLuint prim = intel->prim.primitive;
+
+   intel_flush_inline_primitive(intel);
+   intel_batchbuffer_flush(intel->batch);
+   intel_start_inline(intel, prim);  /* ??? */
+}
+
+static GLuint *intel_extend_inline(struct intel_context *intel, GLuint dwords)
+{
+   GLuint sz = dwords * sizeof(GLuint);
+   GLuint *ptr;
+
+   assert(intel->prim.flush == intel_flush_inline_primitive);
+
+   if (intel_batchbuffer_space(intel->batch) < sz)
+      intel_wrap_inline(intel);
+
+/*    _mesa_printf("."); */
+
+   intel->vtbl.assert_not_dirty(intel);
+
+   ptr = (GLuint *) intel->batch->ptr;
+   intel->batch->ptr += sz;
+
+   return ptr;
+}
+
+/** Sets the primitive type for a primitive sequence, flushing as needed. */
+void intel_set_prim(struct intel_context *intel, uint32_t prim)
+{
+   /* if we have no VBOs */
 
-/* XXX we shouldn't include these headers in this file, but we need them
- * for fallbackStrings, below.
+   if (intel->intelScreen->no_vbo) {
+      intel_start_inline(intel, prim);
+      return;
+   }
+   if (prim != intel->prim.primitive) {
+      INTEL_FIREVERTICES(intel);
+      intel->prim.primitive = prim;
+   }
+}
+
+/** Returns mapped VB space for the given number of vertices */
+uint32_t *intel_get_prim_space(struct intel_context *intel, unsigned int count)
+{
+   uint32_t *addr;
+
+   if (intel->intelScreen->no_vbo) {
+      return intel_extend_inline(intel, count * intel->vertex_size);
+   }
+
+   /* Check for space in the existing VB */
+   if (intel->prim.vb_bo == NULL ||
+       (intel->prim.current_offset +
+	count * intel->vertex_size * 4) > INTEL_VB_SIZE ||
+       (intel->prim.count + count) >= (1 << 16)) {
+      /* Flush existing prim if any */
+      INTEL_FIREVERTICES(intel);
+
+      intel_finish_vb(intel);
+
+      /* Start a new VB */
+      if (intel->prim.vb == NULL)
+	 intel->prim.vb = malloc(INTEL_VB_SIZE);
+      intel->prim.vb_bo = dri_bo_alloc(intel->bufmgr, "vb",
+				       INTEL_VB_SIZE, 4);
+      intel->prim.start_offset = 0;
+      intel->prim.current_offset = 0;
+   }
+
+   intel->prim.flush = intel_flush_prim;
+
+   addr = (uint32_t *)(intel->prim.vb + intel->prim.current_offset);
+   intel->prim.current_offset += intel->vertex_size * 4 * count;
+   intel->prim.count += count;
+
+   return addr;
+}
+
+/** Dispatches the accumulated primitive to the batchbuffer. */
+void intel_flush_prim(struct intel_context *intel)
+{
+   BATCH_LOCALS;
+   dri_bo *aper_array[2];
+   dri_bo *vb_bo;
+   unsigned int offset, count;
+
+   /* Must be called after an intel_start_prim. */
+   assert(intel->prim.primitive != ~0);
+
+   if (intel->prim.count == 0)
+      return;
+
+   /* Clear the current prims out of the context state so that a batch flush
+    * flush triggered by wait_flips or emit_state doesn't loop back to
+    * flush_prim again.
+    */
+   vb_bo = intel->prim.vb_bo;
+   dri_bo_reference(vb_bo);
+   count = intel->prim.count;
+   intel->prim.count = 0;
+   offset = intel->prim.start_offset;
+   intel->prim.start_offset = intel->prim.current_offset;
+   if (!IS_9XX(intel->intelScreen->deviceID))
+      intel->prim.start_offset = ALIGN(intel->prim.start_offset, 128);
+   intel->prim.flush = NULL;
+
+   intel_wait_flips(intel);
+
+   intel->vtbl.emit_state(intel);
+
+   aper_array[0] = intel->batch->buf;
+   aper_array[1] = vb_bo;
+   if (dri_bufmgr_check_aperture_space(aper_array, 2)) {
+      intel_batchbuffer_flush(intel->batch);
+      intel->vtbl.emit_state(intel);
+   }
+
+   /* Ensure that we don't start a new batch for the following emit, which
+    * depends on the state just emitted. emit_state should be making sure we
+    * have the space for this.
+    */
+   intel->no_batch_wrap = GL_TRUE;
+
+   /* Check that we actually emitted the state into this batch, using the
+    * UPLOAD_CTX bit as the signal.
+    */
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
+
+#if 0
+   printf("emitting %d..%d=%d vertices size %d\n", offset,
+	  intel->prim.current_offset, count,
+	  intel->vertex_size * 4);
+#endif
+
+   if (IS_9XX(intel->intelScreen->deviceID)) {
+      BEGIN_BATCH(5, LOOP_CLIPRECTS);
+      OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
+		I1_LOAD_S(0) | I1_LOAD_S(1) | 1);
+      assert((offset & !S0_VB_OFFSET_MASK) == 0);
+      OUT_RELOC(vb_bo, I915_GEM_DOMAIN_VERTEX, 0, offset);
+      OUT_BATCH((intel->vertex_size << S1_VERTEX_WIDTH_SHIFT) |
+		(intel->vertex_size << S1_VERTEX_PITCH_SHIFT));
+
+      OUT_BATCH(_3DPRIMITIVE |
+		PRIM_INDIRECT |
+		PRIM_INDIRECT_SEQUENTIAL |
+		intel->prim.primitive |
+		count);
+      OUT_BATCH(0); /* Beginning vertex index */
+      ADVANCE_BATCH();
+   } else {
+      struct i830_context *i830 = i830_context(&intel->ctx);
+
+      BEGIN_BATCH(5, LOOP_CLIPRECTS);
+      OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
+		I1_LOAD_S(0) | I1_LOAD_S(2) | 1);
+      /* S0 */
+      assert((offset & !S0_VB_OFFSET_MASK_830) == 0);
+      OUT_RELOC(vb_bo, I915_GEM_DOMAIN_VERTEX, 0,
+		offset | (intel->vertex_size << S0_VB_PITCH_SHIFT_830) |
+		S0_VB_ENABLE_830);
+      /* S2
+       * This is somewhat unfortunate -- VB width is tied up with
+       * vertex format data that we've already uploaded through
+       * _3DSTATE_VFT[01]_CMD.  We may want to replace emits of VFT state with
+       * STATE_IMMEDIATE_1 like this to avoid duplication.
+       */
+      OUT_BATCH((i830->state.Ctx[I830_CTXREG_VF] & VFT0_TEX_COUNT_MASK) >>
+		VFT0_TEX_COUNT_SHIFT << S2_TEX_COUNT_SHIFT_830 |
+		(i830->state.Ctx[I830_CTXREG_VF2] << 16) |
+		intel->vertex_size << S2_VERTEX_0_WIDTH_SHIFT_830);
+
+      OUT_BATCH(_3DPRIMITIVE |
+		PRIM_INDIRECT |
+		PRIM_INDIRECT_SEQUENTIAL |
+		intel->prim.primitive |
+		count);
+      OUT_BATCH(0); /* Beginning vertex index */
+      ADVANCE_BATCH();
+   }
+
+   intel->no_batch_wrap = GL_FALSE;
+
+   dri_bo_unreference(vb_bo);
+}
+
+/**
+ * Uploads the locally-accumulated VB into the buffer object.
+ *
+ * This avoids us thrashing the cachelines in and out as the buffer gets
+ * filled, dispatched, then reused as the hardware completes rendering from it,
+ * and also lets us clflush less if we dispatch with a partially-filled VB.
+ *
+ * This is called normally from get_space when we're finishing a BO, but also
+ * at batch flush time so that we don't try accessing the contents of a
+ * just-dispatched buffer.
  */
-#include "i830_context.h"
-#include "i915_context.h"
+void intel_finish_vb(struct intel_context *intel)
+{
+   if (intel->prim.vb_bo == NULL)
+      return;
 
-static void intelRenderPrimitive( GLcontext *ctx, GLenum prim );
-static void intelRasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
+   dri_bo_subdata(intel->prim.vb_bo, 0, intel->prim.start_offset,
+		  intel->prim.vb);
+   dri_bo_unreference(intel->prim.vb_bo);
+   intel->prim.vb_bo = NULL;
+}
 
 /***********************************************************************
  *                    Emit primitives as inline vertices               *
@@ -69,75 +346,81 @@ do {								\
 #else
 #define COPY_DWORDS( j, vb, vertsize, v )	\
 do {						\
-   if (0) fprintf(stderr, "\n");	\
    for ( j = 0 ; j < vertsize ; j++ ) {		\
-      if (0) fprintf(stderr, "   -- v(%d): %x/%f\n",j,	\
-	      ((GLuint *)v)[j],			\
-	      ((GLfloat *)v)[j]);		\
       vb[j] = ((GLuint *)v)[j];			\
    }						\
    vb += vertsize;				\
 } while (0)
 #endif
 
-static void __inline__ intel_draw_quad( intelContextPtr intel,
-					intelVertexPtr v0,
-					intelVertexPtr v1,
-					intelVertexPtr v2,
-					intelVertexPtr v3 )
+static void
+intel_draw_quad(struct intel_context *intel,
+                intelVertexPtr v0,
+                intelVertexPtr v1, intelVertexPtr v2, intelVertexPtr v3)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 6 * vertsize );
+   GLuint *vb = intel_get_prim_space(intel, 6);
    int j;
 
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v3 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v2 );
-   COPY_DWORDS( j, vb, vertsize, v3 );
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
+
+   /* If smooth shading, draw like a trifan which gives better
+    * rasterization.  Otherwise draw as two triangles with provoking
+    * vertex in third position as required for flat shading.
+    */
+   if (intel->ctx.Light.ShadeModel == GL_FLAT) {
+      COPY_DWORDS(j, vb, vertsize, v3);
+      COPY_DWORDS(j, vb, vertsize, v1);
+   }
+   else {
+      COPY_DWORDS(j, vb, vertsize, v2);
+      COPY_DWORDS(j, vb, vertsize, v0);
+   }
+
+   COPY_DWORDS(j, vb, vertsize, v2);
+   COPY_DWORDS(j, vb, vertsize, v3);
 }
 
-static void __inline__ intel_draw_triangle( intelContextPtr intel,
-					    intelVertexPtr v0,
-					    intelVertexPtr v1,
-					    intelVertexPtr v2 )
+static void
+intel_draw_triangle(struct intel_context *intel,
+                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 3 * vertsize );
+   GLuint *vb = intel_get_prim_space(intel, 3);
    int j;
-   
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v2 );
+
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
+   COPY_DWORDS(j, vb, vertsize, v2);
 }
 
 
-static __inline__ void intel_draw_line( intelContextPtr intel,
-					intelVertexPtr v0,
-					intelVertexPtr v1 )
+static void
+intel_draw_line(struct intel_context *intel,
+                intelVertexPtr v0, intelVertexPtr v1)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 2 * vertsize );
+   GLuint *vb = intel_get_prim_space(intel, 2);
    int j;
 
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
 }
 
 
-static __inline__ void intel_draw_point( intelContextPtr intel,
-					 intelVertexPtr v0 )
+static void
+intel_draw_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, vertsize );
+   GLuint *vb = intel_get_prim_space(intel, 1);
    int j;
 
    /* Adjust for sub pixel position -- still required for conform. */
-   *(float *)&vb[0] = v0->v.x - 0.125;
-   *(float *)&vb[1] = v0->v.y - 0.125;
-   for (j = 2 ; j < vertsize ; j++)
-     vb[j] = v0->ui[j];
+   *(float *) &vb[0] = v0->v.x;
+   *(float *) &vb[1] = v0->v.y;
+   for (j = 2; j < vertsize; j++)
+      vb[j] = v0->ui[j];
 }
 
 
@@ -146,13 +429,17 @@ static __inline__ void intel_draw_point( intelContextPtr intel,
  *                Fixup for ARB_point_parameters                       *
  ***********************************************************************/
 
-static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
+/* Currently not working - VERT_ATTRIB_POINTSIZE isn't correctly
+ * represented in the fragment program InputsRead field.
+ */
+static void
+intel_atten_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLcontext *ctx = &intel->ctx;
    GLfloat psz[4], col[4], restore_psz, restore_alpha;
 
-   _tnl_get_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz );
-   _tnl_get_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col );
+   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
 
    restore_psz = psz[0];
    restore_alpha = col[3];
@@ -170,19 +457,19 @@ static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
       psz[0] = 1.0;
 
    if (restore_psz != psz[0] || restore_alpha != col[3]) {
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col);
-   
-      intel_draw_point( intel, v0 );
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
+
+      intel_draw_point(intel, v0);
 
       psz[0] = restore_psz;
       col[3] = restore_alpha;
 
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
    }
    else
-      intel_draw_point( intel, v0 );
+      intel_draw_point(intel, v0);
 }
 
 
@@ -195,45 +482,59 @@ static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
 
 
 
-static void intel_wpos_triangle( intelContextPtr intel,
-				 intelVertexPtr v0,
-				 intelVertexPtr v1,
-				 intelVertexPtr v2 )
+static void
+intel_wpos_triangle(struct intel_context *intel,
+                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
-   
-   __memcpy( ((char *)v0) + offset, v0, size );
-   __memcpy( ((char *)v1) + offset, v1, size );
-   __memcpy( ((char *)v2) + offset, v2, size );
+   GLfloat *v0_wpos = (GLfloat *)((char *)v0 + offset);
+   GLfloat *v1_wpos = (GLfloat *)((char *)v1 + offset);
+   GLfloat *v2_wpos = (GLfloat *)((char *)v2 + offset);
 
-   intel_draw_triangle( intel, v0, v1, v2 );
+   __memcpy(v0_wpos, v0, size);
+   __memcpy(v1_wpos, v1, size);
+   __memcpy(v2_wpos, v2, size);
+
+   v0_wpos[1] = -v0_wpos[1] + intel->driDrawable->h;
+   v1_wpos[1] = -v1_wpos[1] + intel->driDrawable->h;
+   v2_wpos[1] = -v2_wpos[1] + intel->driDrawable->h;
+
+
+   intel_draw_triangle(intel, v0, v1, v2);
 }
 
 
-static void intel_wpos_line( intelContextPtr intel,
-			     intelVertexPtr v0,
-			     intelVertexPtr v1 )
+static void
+intel_wpos_line(struct intel_context *intel,
+                intelVertexPtr v0, intelVertexPtr v1)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
+   GLfloat *v0_wpos = (GLfloat *)((char *)v0 + offset);
+   GLfloat *v1_wpos = (GLfloat *)((char *)v1 + offset);
+
+   __memcpy(v0_wpos, v0, size);
+   __memcpy(v1_wpos, v1, size);
 
-   __memcpy( ((char *)v0) + offset, v0, size );
-   __memcpy( ((char *)v1) + offset, v1, size );
+   v0_wpos[1] = -v0_wpos[1] + intel->driDrawable->h;
+   v1_wpos[1] = -v1_wpos[1] + intel->driDrawable->h;
 
-   intel_draw_line( intel, v0, v1 );
+   intel_draw_line(intel, v0, v1);
 }
 
 
-static void intel_wpos_point( intelContextPtr intel,
-			      intelVertexPtr v0 )
+static void
+intel_wpos_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
+   GLfloat *v0_wpos = (GLfloat *)((char *)v0 + offset);
 
-   __memcpy( ((char *)v0) + offset, v0, size );
+   __memcpy(v0_wpos, v0, size);
+   v0_wpos[1] = -v0_wpos[1] + intel->driDrawable->h;
 
-   intel_draw_point( intel, v0 );
+   intel_draw_point(intel, v0);
 }
 
 
@@ -290,11 +591,12 @@ do { 						\
 #define INTEL_MAX_TRIFUNC	0x10
 
 
-static struct {
-   tnl_points_func	        points;
-   tnl_line_func		line;
-   tnl_triangle_func	triangle;
-   tnl_quad_func		quad;
+static struct
+{
+   tnl_points_func points;
+   tnl_line_func line;
+   tnl_triangle_func triangle;
+   tnl_quad_func quad;
 } rast_tab[INTEL_MAX_TRIFUNC];
 
 
@@ -355,10 +657,10 @@ do {							\
 #define VERT_RESTORE_SPEC( idx ) if (specoffset) v[idx]->ui[specoffset] = spec[idx]
 
 #define LOCAL_VARS(n)							\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);				\
-   GLuint color[n], spec[n];						\
-   GLuint coloroffset = intel->coloroffset;		\
-   GLboolean specoffset = intel->specoffset;			\
+   struct intel_context *intel = intel_context(ctx);			\
+   GLuint color[n] = { 0, }, spec[n] = { 0, };				\
+   GLuint coloroffset = intel->coloroffset;				\
+   GLboolean specoffset = intel->specoffset;				\
    (void) color; (void) spec; (void) coloroffset; (void) specoffset;
 
 
@@ -366,7 +668,7 @@ do {							\
  *                Helpers for rendering unfilled primitives            *
  ***********************************************************************/
 
-static const GLuint hw_prim[GL_POLYGON+1] = {
+static const GLuint hw_prim[GL_POLYGON + 1] = {
    PRIM3D_POINTLIST,
    PRIM3D_LINELIST,
    PRIM3D_LINELIST,
@@ -456,7 +758,8 @@ static const GLuint hw_prim[GL_POLYGON+1] = {
 #include "tnl_dd/t_dd_tritmp.h"
 
 
-static void init_rast_tab( void )
+static void
+init_rast_tab(void)
 {
    init();
    init_offset();
@@ -487,10 +790,8 @@ static void init_rast_tab( void )
  * primitives.
  */
 static void
-intel_fallback_tri( intelContextPtr intel,
-		   intelVertex *v0,
-		   intelVertex *v1,
-		   intelVertex *v2 )
+intel_fallback_tri(struct intel_context *intel,
+                   intelVertex * v0, intelVertex * v1, intelVertex * v2)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[3];
@@ -498,19 +799,20 @@ intel_fallback_tri( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   _swsetup_Translate( ctx, v1, &v[1] );
-   _swsetup_Translate( ctx, v2, &v[2] );
-   intelSpanRenderStart( ctx );
-   _swrast_Triangle( ctx, &v[0], &v[1], &v[2] );
-   intelSpanRenderFinish( ctx );
+   INTEL_FIREVERTICES(intel);
+
+   _swsetup_Translate(ctx, v0, &v[0]);
+   _swsetup_Translate(ctx, v1, &v[1]);
+   _swsetup_Translate(ctx, v2, &v[2]);
+   intelSpanRenderStart(ctx);
+   _swrast_Triangle(ctx, &v[0], &v[1], &v[2]);
+   intelSpanRenderFinish(ctx);
 }
 
 
 static void
-intel_fallback_line( intelContextPtr intel,
-		    intelVertex *v0,
-		    intelVertex *v1 )
+intel_fallback_line(struct intel_context *intel,
+                    intelVertex * v0, intelVertex * v1)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[2];
@@ -518,17 +820,18 @@ intel_fallback_line( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   _swsetup_Translate( ctx, v1, &v[1] );
-   intelSpanRenderStart( ctx );
-   _swrast_Line( ctx, &v[0], &v[1] );
-   intelSpanRenderFinish( ctx );
-}
+   INTEL_FIREVERTICES(intel);
 
+   _swsetup_Translate(ctx, v0, &v[0]);
+   _swsetup_Translate(ctx, v1, &v[1]);
+   intelSpanRenderStart(ctx);
+   _swrast_Line(ctx, &v[0], &v[1]);
+   intelSpanRenderFinish(ctx);
+}
 
 static void
-intel_fallback_point( intelContextPtr intel,
-		     intelVertex *v0 )
+intel_fallback_point(struct intel_context *intel,
+		     intelVertex * v0)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[1];
@@ -536,12 +839,13 @@ intel_fallback_point( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   intelSpanRenderStart( ctx );
-   _swrast_Point( ctx, &v[0] );
-   intelSpanRenderFinish( ctx );
-}
+   INTEL_FIREVERTICES(intel);
 
+   _swsetup_Translate(ctx, v0, &v[0]);
+   intelSpanRenderStart(ctx);
+   _swrast_Point(ctx, &v[0]);
+   intelSpanRenderFinish(ctx);
+}
 
 
 /**********************************************************************/
@@ -558,7 +862,7 @@ intel_fallback_point( intelContextPtr intel,
 #define INIT(x) intelRenderPrimitive( ctx, x )
 #undef LOCAL_VARS
 #define LOCAL_VARS						\
-    intelContextPtr intel = INTEL_CONTEXT(ctx);			\
+    struct intel_context *intel = intel_context(ctx);			\
     GLubyte *vertptr = (GLubyte *)intel->verts;			\
     const GLuint vertsize = intel->vertex_size;       	\
     const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;	\
@@ -581,10 +885,10 @@ intel_fallback_point( intelContextPtr intel,
 
 
 
-static void intelRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
-				   GLuint n )
+static void
+intelRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    GLuint prim = intel->render_primitive;
@@ -593,39 +897,40 @@ static void intelRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
     */
    {
       GLuint *tmp = VB->Elts;
-      VB->Elts = (GLuint *)elts;
-      tnl->Driver.Render.PrimTabElts[GL_POLYGON]( ctx, 0, n, 
-						  PRIM_BEGIN|PRIM_END );
+      VB->Elts = (GLuint *) elts;
+      tnl->Driver.Render.PrimTabElts[GL_POLYGON] (ctx, 0, n,
+                                                  PRIM_BEGIN | PRIM_END);
       VB->Elts = tmp;
    }
 
    /* Restore the render primitive
     */
    if (prim != GL_POLYGON)
-      tnl->Driver.Render.PrimitiveNotify( ctx, prim );
+      tnl->Driver.Render.PrimitiveNotify(ctx, prim);
 }
 
-static void intelRenderClippedLine( GLcontext *ctx, GLuint ii, GLuint jj )
+static void
+intelRenderClippedLine(GLcontext * ctx, GLuint ii, GLuint jj)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
 
-   tnl->Driver.Render.Line( ctx, ii, jj );
+   tnl->Driver.Render.Line(ctx, ii, jj);
 }
 
-static void intelFastRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
-				       GLuint n )
+static void
+intelFastRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
 {
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
+   struct intel_context *intel = intel_context(ctx);
    const GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, (n-2) * 3 * vertsize );
-   GLubyte *vertptr = (GLubyte *)intel->verts;
-   const GLuint *start = (const GLuint *)V(elts[0]);
-   int i,j;
-
-   for (i = 2 ; i < n ; i++) {
-      COPY_DWORDS( j, vb, vertsize, V(elts[i-1]) );
-      COPY_DWORDS( j, vb, vertsize, V(elts[i]) );
-      COPY_DWORDS( j, vb, vertsize, start );
+   GLuint *vb = intel_get_prim_space(intel, (n - 2) * 3);
+   GLubyte *vertptr = (GLubyte *) intel->verts;
+   const GLuint *start = (const GLuint *) V(elts[0]);
+   int i, j;
+
+   for (i = 2; i < n; i++) {
+      COPY_DWORDS(j, vb, vertsize, V(elts[i - 1]));
+      COPY_DWORDS(j, vb, vertsize, V(elts[i]));
+      COPY_DWORDS(j, vb, vertsize, start);
    }
 }
 
@@ -636,68 +941,75 @@ static void intelFastRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
 
 
 
-#define POINT_FALLBACK (0)
-#define LINE_FALLBACK (DD_LINE_STIPPLE)
-#define TRI_FALLBACK (0)
-#define ANY_FALLBACK_FLAGS (POINT_FALLBACK|LINE_FALLBACK|TRI_FALLBACK|\
-                            DD_TRI_STIPPLE|DD_POINT_ATTEN)
-#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE|DD_TRI_OFFSET|DD_TRI_UNFILLED)
+#define ANY_FALLBACK_FLAGS (DD_LINE_STIPPLE | DD_TRI_STIPPLE | DD_POINT_ATTEN | DD_POINT_SMOOTH | DD_TRI_SMOOTH)
+#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE | DD_TRI_OFFSET | DD_TRI_UNFILLED)
 
-void intelChooseRenderState(GLcontext *ctx)
+void
+intelChooseRenderState(GLcontext * ctx)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    GLuint flags = ctx->_TriangleCaps;
    const struct gl_fragment_program *fprog = ctx->FragmentProgram._Current;
    GLboolean have_wpos = (fprog && (fprog->Base.InputsRead & FRAG_BIT_WPOS));
    GLuint index = 0;
 
    if (INTEL_DEBUG & DEBUG_STATE)
-     fprintf(stderr,"\n%s\n",__FUNCTION__);
+      fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   if ((flags & (ANY_FALLBACK_FLAGS|ANY_RASTER_FLAGS)) || have_wpos) {
+   if ((flags & (ANY_FALLBACK_FLAGS | ANY_RASTER_FLAGS)) || have_wpos) {
 
       if (flags & ANY_RASTER_FLAGS) {
-	 if (flags & DD_TRI_LIGHT_TWOSIDE)    index |= INTEL_TWOSIDE_BIT;
-	 if (flags & DD_TRI_OFFSET)	      index |= INTEL_OFFSET_BIT;
-	 if (flags & DD_TRI_UNFILLED)	      index |= INTEL_UNFILLED_BIT;
+         if (flags & DD_TRI_LIGHT_TWOSIDE)
+            index |= INTEL_TWOSIDE_BIT;
+         if (flags & DD_TRI_OFFSET)
+            index |= INTEL_OFFSET_BIT;
+         if (flags & DD_TRI_UNFILLED)
+            index |= INTEL_UNFILLED_BIT;
       }
 
       if (have_wpos) {
-	 intel->draw_point = intel_wpos_point;
-	 intel->draw_line = intel_wpos_line;
-	 intel->draw_tri = intel_wpos_triangle;
+         intel->draw_point = intel_wpos_point;
+         intel->draw_line = intel_wpos_line;
+         intel->draw_tri = intel_wpos_triangle;
 
-	 /* Make sure these get called:
-	  */
-	 index |= INTEL_FALLBACK_BIT;
+         /* Make sure these get called:
+          */
+         index |= INTEL_FALLBACK_BIT;
       }
       else {
-	 intel->draw_point = intel_draw_point;
-	 intel->draw_line = intel_draw_line;
-	 intel->draw_tri = intel_draw_triangle;
+         intel->draw_point = intel_draw_point;
+         intel->draw_line = intel_draw_line;
+         intel->draw_tri = intel_draw_triangle;
       }
 
       /* Hook in fallbacks for specific primitives.
        */
-      if (flags & ANY_FALLBACK_FLAGS)
-      {
-	 if (flags & POINT_FALLBACK)
-	    intel->draw_point = intel_fallback_point;
-
-	 if (flags & LINE_FALLBACK)
-	    intel->draw_line = intel_fallback_line;
-
-	 if (flags & TRI_FALLBACK)
-	    intel->draw_tri = intel_fallback_tri;
-
-	 if ((flags & DD_TRI_STIPPLE) && !intel->hw_stipple) 
-	    intel->draw_tri = intel_fallback_tri;
-
-	 if (flags & DD_POINT_ATTEN)
-	    intel->draw_point = intel_atten_point;
-
-	 index |= INTEL_FALLBACK_BIT;
+      if (flags & ANY_FALLBACK_FLAGS) {
+         if (flags & DD_LINE_STIPPLE)
+            intel->draw_line = intel_fallback_line;
+
+         if ((flags & DD_TRI_STIPPLE) && !intel->hw_stipple)
+            intel->draw_tri = intel_fallback_tri;
+
+         if (flags & DD_TRI_SMOOTH) {
+	    if (intel->strict_conformance)
+	       intel->draw_tri = intel_fallback_tri;
+	 }
+
+         if (flags & DD_POINT_ATTEN) {
+	    if (0)
+	       intel->draw_point = intel_atten_point;
+	    else
+	       intel->draw_point = intel_fallback_point;
+	 }
+
+	 if (flags & DD_POINT_SMOOTH) {
+	    if (intel->strict_conformance)
+	       intel->draw_point = intel_fallback_point;
+	 }
+
+         index |= INTEL_FALLBACK_BIT;
       }
    }
 
@@ -710,20 +1022,21 @@ void intelChooseRenderState(GLcontext *ctx)
       tnl->Driver.Render.Quad = rast_tab[index].quad;
 
       if (index == 0) {
-	 tnl->Driver.Render.PrimTabVerts = intel_render_tab_verts;
-	 tnl->Driver.Render.PrimTabElts = intel_render_tab_elts;
-	 tnl->Driver.Render.ClippedLine = line; /* from tritmp.h */
-	 tnl->Driver.Render.ClippedPolygon = intelFastRenderClippedPoly;
-      } else {
-	 tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-	 tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
-	 tnl->Driver.Render.ClippedLine = intelRenderClippedLine;
-	 tnl->Driver.Render.ClippedPolygon = intelRenderClippedPoly;
+         tnl->Driver.Render.PrimTabVerts = intel_render_tab_verts;
+         tnl->Driver.Render.PrimTabElts = intel_render_tab_elts;
+         tnl->Driver.Render.ClippedLine = line; /* from tritmp.h */
+         tnl->Driver.Render.ClippedPolygon = intelFastRenderClippedPoly;
+      }
+      else {
+         tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
+         tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
+         tnl->Driver.Render.ClippedLine = intelRenderClippedLine;
+         tnl->Driver.Render.ClippedPolygon = intelRenderClippedPoly;
       }
    }
 }
 
-static const GLenum reduced_prim[GL_POLYGON+1] = {
+static const GLenum reduced_prim[GL_POLYGON + 1] = {
    GL_POINTS,
    GL_LINES,
    GL_LINES,
@@ -744,35 +1057,52 @@ static const GLenum reduced_prim[GL_POLYGON+1] = {
 
 
 
-static void intelRunPipeline( GLcontext *ctx )
+static void
+intelRunPipeline(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
+
+   _mesa_lock_context_textures(ctx);
+   
+   if (ctx->NewState)
+      _mesa_update_state_locked(ctx);
 
    if (intel->NewGLState) {
       if (intel->NewGLState & _NEW_TEXTURE) {
-	 intel->vtbl.update_texture_state( intel ); 
+         intel->vtbl.update_texture_state(intel);
       }
 
       if (!intel->Fallback) {
-	 if (intel->NewGLState & _INTEL_NEW_RENDERSTATE)
-	    intelChooseRenderState( ctx );
+         if (intel->NewGLState & _INTEL_NEW_RENDERSTATE)
+            intelChooseRenderState(ctx);
       }
 
       intel->NewGLState = 0;
    }
 
-   _tnl_run_pipeline( ctx );
+   _tnl_run_pipeline(ctx);
+
+   _mesa_unlock_context_textures(ctx);
 }
 
-static void intelRenderStart( GLcontext *ctx )
+static void
+intelRenderStart(GLcontext * ctx)
 {
-   INTEL_CONTEXT(ctx)->vtbl.render_start( INTEL_CONTEXT(ctx) );
+   struct intel_context *intel = intel_context(ctx);
+
+   intel->vtbl.render_start(intel_context(ctx));
+   intel->vtbl.emit_state(intel);
 }
 
-static void intelRenderFinish( GLcontext *ctx )
+static void
+intelRenderFinish(GLcontext * ctx)
 {
-   if (INTEL_CONTEXT(ctx)->RenderIndex & INTEL_FALLBACK_BIT)
-      _swrast_flush( ctx );
+   struct intel_context *intel = intel_context(ctx);
+
+   if (intel->RenderIndex & INTEL_FALLBACK_BIT)
+      _swrast_flush(ctx);
+
+   INTEL_FIREVERTICES(intel);
 }
 
 
@@ -781,28 +1111,33 @@ static void intelRenderFinish( GLcontext *ctx )
  /* System to flush dma and emit state changes based on the rasterized
   * primitive.
   */
-static void intelRasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim )
+static void
+intelRasterPrimitive(GLcontext * ctx, GLenum rprim, GLuint hwprim)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
 
    if (0)
-      fprintf(stderr, "%s %s %x\n", __FUNCTION__, 
-	      _mesa_lookup_enum_by_nr(rprim), hwprim);
+      fprintf(stderr, "%s %s %x\n", __FUNCTION__,
+              _mesa_lookup_enum_by_nr(rprim), hwprim);
+
+   intel->vtbl.reduced_primitive_state(intel, rprim);
 
-   intel->vtbl.reduced_primitive_state( intel, rprim );
-    
    /* Start a new primitive.  Arrange to have it flushed later on.
     */
-   if (hwprim != intel->prim.primitive) 
-      intelStartInlinePrimitive( intel, hwprim );
+   if (hwprim != intel->prim.primitive) {
+      INTEL_FIREVERTICES(intel);
+
+      intel_set_prim(intel, hwprim);
+   }
 }
 
 
-/* 
- */
-static void intelRenderPrimitive( GLcontext *ctx, GLenum prim )
+ /* 
+  */
+static void
+intelRenderPrimitive(GLcontext * ctx, GLenum prim)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
 
    if (0)
       fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
@@ -817,63 +1152,54 @@ static void intelRenderPrimitive( GLcontext *ctx, GLenum prim )
     * lower level functions in that case, potentially pingponging the
     * state:
     */
-   if (reduced_prim[prim] == GL_TRIANGLES && 
+   if (reduced_prim[prim] == GL_TRIANGLES &&
        (ctx->_TriangleCaps & DD_TRI_UNFILLED))
       return;
 
    /* Set some primitive-dependent state and Start? a new primitive.
     */
-   intelRasterPrimitive( ctx, reduced_prim[prim], hw_prim[prim] );
+   intelRasterPrimitive(ctx, reduced_prim[prim], hw_prim[prim]);
 }
 
 
-/**********************************************************************/
-/*           Transition to/from hardware rasterization.               */
-/**********************************************************************/
-
-static struct {
-   GLuint bit;
-   const char *str;
-} fallbackStrings[] = {
-   { INTEL_FALLBACK_DRAW_BUFFER, "Draw buffer" },
-   { INTEL_FALLBACK_READ_BUFFER, "Read buffer" },
-   { INTEL_FALLBACK_USER, "User" },
-   { INTEL_FALLBACK_NO_BATCHBUFFER, "No Batchbuffer" },
-   { INTEL_FALLBACK_NO_TEXMEM, "No Texmem" },
-   { INTEL_FALLBACK_RENDERMODE, "Rendermode" },
-
-   { I830_FALLBACK_TEXTURE, "i830 texture" },
-   { I830_FALLBACK_COLORMASK, "i830 colormask" },
-   { I830_FALLBACK_STENCIL, "i830 stencil" },
-   { I830_FALLBACK_STIPPLE, "i830 stipple" },
-   { I830_FALLBACK_LOGICOP, "i830 logicop" },
-
-   { I915_FALLBACK_TEXTURE, "i915 texture" },
-   { I915_FALLBACK_COLORMASK, "i915 colormask" },
-   { I915_FALLBACK_STENCIL, "i915 stencil" },
-   { I915_FALLBACK_STIPPLE, "i915 stipple" },
-   { I915_FALLBACK_PROGRAM, "i915 program" },
-   { I915_FALLBACK_LOGICOP, "i915 logicop" },
-   { I915_FALLBACK_POLYGON_SMOOTH, "i915 polygon smooth" },
-   { I915_FALLBACK_POINT_SMOOTH, "i915 point smooth" },
-
-   { 0, NULL }
+ /**********************************************************************/
+ /*           Transition to/from hardware rasterization.               */
+ /**********************************************************************/
+
+static char *fallbackStrings[] = {
+   [0] = "Draw buffer",
+   [1] = "Read buffer",
+   [2] = "Depth buffer",
+   [3] = "Stencil buffer",
+   [4] = "User disable",
+   [5] = "Render mode",
+
+   [12] = "Texture",
+   [13] = "Color mask",
+   [14] = "Stencil",
+   [15] = "Stipple",
+   [16] = "Program",
+   [17] = "Logic op",
+   [18] = "Smooth polygon",
+   [19] = "Smooth point",
 };
 
 
-static const char *
+static char *
 getFallbackString(GLuint bit)
 {
-   int i;
-   for (i = 0; fallbackStrings[i].bit; i++) {
-      if (fallbackStrings[i].bit == bit)
-         return fallbackStrings[i].str;
+   int i = 0;
+   while (bit > 1) {
+      i++;
+      bit >>= 1;
    }
-   return "unknown fallback bit";
+   return fallbackStrings[i];
 }
 
 
-void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
+
+void
+intelFallback(struct intel_context *intel, GLuint bit, GLboolean mode)
 {
    GLcontext *ctx = &intel->ctx;
    TNLcontext *tnl = TNL_CONTEXT(ctx);
@@ -883,20 +1209,19 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
       intel->Fallback |= bit;
       if (oldfallback == 0) {
          intelFlush(ctx);
-         if (INTEL_DEBUG & DEBUG_FALLBACKS) 
-            fprintf(stderr, "ENTER FALLBACK 0x%x: %s\n",
+         if (INTEL_DEBUG & DEBUG_FALLBACKS)
+            fprintf(stderr, "ENTER FALLBACK %x: %s\n",
                     bit, getFallbackString(bit));
-         _swsetup_Wakeup( ctx );
+         _swsetup_Wakeup(ctx);
          intel->RenderIndex = ~0;
       }
    }
    else {
       intel->Fallback &= ~bit;
       if (oldfallback == bit) {
-         _swrast_flush( ctx );
-         if (INTEL_DEBUG & DEBUG_FALLBACKS) 
-            fprintf(stderr, "LEAVE FALLBACK 0x%x: %s\n",
-                    bit, getFallbackString(bit));
+         _swrast_flush(ctx);
+         if (INTEL_DEBUG & DEBUG_FALLBACKS)
+            fprintf(stderr, "LEAVE FALLBACK %s\n", getFallbackString(bit));
          tnl->Driver.Render.Start = intelRenderStart;
          tnl->Driver.Render.PrimitiveNotify = intelRenderPrimitive;
          tnl->Driver.Render.Finish = intelRenderFinish;
@@ -904,18 +1229,99 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
          tnl->Driver.Render.CopyPV = _tnl_copy_pv;
          tnl->Driver.Render.Interp = _tnl_interp;
 
-         _tnl_invalidate_vertex_state( ctx, ~0 );
-         _tnl_invalidate_vertices( ctx, ~0 );
-         _tnl_install_attrs( ctx, 
-                             intel->vertex_attrs, 
-                             intel->vertex_attr_count,
-                             intel->ViewportMatrix.m, 0 ); 
+         _tnl_invalidate_vertex_state(ctx, ~0);
+         _tnl_invalidate_vertices(ctx, ~0);
+         _tnl_install_attrs(ctx,
+                            intel->vertex_attrs,
+                            intel->vertex_attr_count,
+                            intel->ViewportMatrix.m, 0);
 
          intel->NewGLState |= _INTEL_NEW_RENDERSTATE;
       }
    }
 }
 
+union fi
+{
+   GLfloat f;
+   GLint i;
+};
+
+
+/**********************************************************************/
+/*             Used only with the metaops callbacks.                  */
+/**********************************************************************/
+static void
+intel_meta_draw_poly(struct intel_context *intel,
+                     GLuint n,
+                     GLfloat xy[][2],
+                     GLfloat z, GLuint color, GLfloat tex[][2])
+{
+   union fi *vb;
+   GLint i;
+   GLboolean was_locked = intel->locked;
+   unsigned int saved_vertex_size = intel->vertex_size;
+
+   if (!was_locked)
+       LOCK_HARDWARE(intel);
+
+   intel->vertex_size = 6;
+
+   /* All 3d primitives should be emitted with LOOP_CLIPRECTS,
+    * otherwise the drawing origin (DR4) might not be set correctly.
+    */
+   intel_set_prim(intel, PRIM3D_TRIFAN);
+   vb = (union fi *) intel_get_prim_space(intel, n);
+
+   for (i = 0; i < n; i++) {
+      vb[0].f = xy[i][0];
+      vb[1].f = xy[i][1];
+      vb[2].f = z;
+      vb[3].i = color;
+      vb[4].f = tex[i][0];
+      vb[5].f = tex[i][1];
+      vb += 6;
+   }
+
+   INTEL_FIREVERTICES(intel);
+
+   intel->vertex_size = saved_vertex_size;
+
+   if (!was_locked)
+       UNLOCK_HARDWARE(intel);
+}
+
+static void
+intel_meta_draw_quad(struct intel_context *intel,
+                     GLfloat x0, GLfloat x1,
+                     GLfloat y0, GLfloat y1,
+                     GLfloat z,
+                     GLuint color,
+                     GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1)
+{
+   GLfloat xy[4][2];
+   GLfloat tex[4][2];
+
+   xy[0][0] = x0;
+   xy[0][1] = y0;
+   xy[1][0] = x1;
+   xy[1][1] = y0;
+   xy[2][0] = x1;
+   xy[2][1] = y1;
+   xy[3][0] = x0;
+   xy[3][1] = y1;
+
+   tex[0][0] = s0;
+   tex[0][1] = t0;
+   tex[1][0] = s1;
+   tex[1][1] = t0;
+   tex[2][0] = s1;
+   tex[2][1] = t1;
+   tex[3][0] = s0;
+   tex[3][1] = t1;
+
+   intel_meta_draw_poly(intel, 4, xy, z, color, tex);
+}
 
 
 
@@ -924,8 +1330,10 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
 /**********************************************************************/
 
 
-void intelInitTriFuncs( GLcontext *ctx )
+void
+intelInitTriFuncs(GLcontext * ctx)
 {
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    static int firsttime = 1;
 
@@ -942,4 +1350,6 @@ void intelInitTriFuncs( GLcontext *ctx )
    tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
    tnl->Driver.Render.CopyPV = _tnl_copy_pv;
    tnl->Driver.Render.Interp = _tnl_interp;
+
+   intel->vtbl.meta_draw_quad = intel_meta_draw_quad;
 }
diff --git a/src/mesa/drivers/dri/i915/intel_tris.h b/src/mesa/drivers/dri/i915/intel_tris.h
index d7e382fdb3e..55b60a47f91 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.h
+++ b/src/mesa/drivers/dri/i915/intel_tris.h
@@ -28,7 +28,11 @@
 #ifndef INTELTRIS_INC
 #define INTELTRIS_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
+
+#define INTEL_VB_SIZE		(32 * 1024)
+/** 3 dwords of state_immediate and 2 of 3dprim, in intel_flush_prim */
+#define INTEL_PRIM_EMIT_SIZE	(5 * 4)
 
 #define _INTEL_NEW_RENDERSTATE (_DD_NEW_LINE_STIPPLE |		\
 			       _DD_NEW_TRI_UNFILLED |		\
@@ -38,9 +42,13 @@
 			       _NEW_PROGRAM |		\
 			       _NEW_POLYGONSTIPPLE)
 
-extern void intelInitTriFuncs( GLcontext *ctx );
+extern void intelInitTriFuncs(GLcontext * ctx);
+
+extern void intelChooseRenderState(GLcontext * ctx);
 
-extern void intelPrintRenderState( const char *msg, GLuint state );
-extern void intelChooseRenderState( GLcontext *ctx );
+void intel_set_prim(struct intel_context *intel, uint32_t prim);
+GLuint *intel_get_prim_space(struct intel_context *intel, unsigned int count);
+void intel_flush_prim(struct intel_context *intel);
+void intel_finish_vb(struct intel_context *intel);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/server/i830_common.h b/src/mesa/drivers/dri/i915/server/i830_common.h
deleted file mode 100644
index 2b0fee82a88..00000000000
--- a/src/mesa/drivers/dri/i915/server/i830_common.h
+++ /dev/null
@@ -1,211 +0,0 @@
-/**************************************************************************
-
-Copyright 2001 VA Linux Systems Inc., Fremont, California.
-Copyright 2002 Tungsten Graphics Inc., Cedar Park, Texas.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ATI, VA LINUX SYSTEMS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef _I830_COMMON_H_
-#define _I830_COMMON_H_
-
-
-#define I830_NR_TEX_REGIONS 255	/* maximum due to use of chars for next/prev */
-#define I830_LOG_MIN_TEX_REGION_SIZE 14
-
-
-/* Driver specific DRM command indices
- * NOTE: these are not OS specific, but they are driver specific
- */
-#define DRM_I830_INIT                     0x00
-#define DRM_I830_FLUSH                    0x01
-#define DRM_I830_FLIP                     0x02
-#define DRM_I830_BATCHBUFFER              0x03
-#define DRM_I830_IRQ_EMIT                 0x04
-#define DRM_I830_IRQ_WAIT                 0x05
-#define DRM_I830_GETPARAM                 0x06
-#define DRM_I830_SETPARAM                 0x07
-#define DRM_I830_ALLOC                    0x08
-#define DRM_I830_FREE                     0x09
-#define DRM_I830_INIT_HEAP                0x0a
-#define DRM_I830_CMDBUFFER                0x0b
-#define DRM_I830_DESTROY_HEAP             0x0c
-
-typedef struct {
-   enum {
-      I830_INIT_DMA = 0x01,
-      I830_CLEANUP_DMA = 0x02,
-      I830_RESUME_DMA = 0x03
-   } func;
-   unsigned int mmio_offset;
-   int sarea_priv_offset;
-   unsigned int ring_start;
-   unsigned int ring_end;
-   unsigned int ring_size;
-   unsigned int front_offset;
-   unsigned int back_offset;
-   unsigned int depth_offset;
-   unsigned int w;
-   unsigned int h;
-   unsigned int pitch;
-   unsigned int pitch_bits;
-   unsigned int back_pitch;
-   unsigned int depth_pitch;
-   unsigned int cpp;
-   unsigned int chipset;
-} drmI830Init;
-
-typedef struct {
-	drmTextureRegion texList[I830_NR_TEX_REGIONS+1];
-        int last_upload;	/* last time texture was uploaded */
-        int last_enqueue;	/* last time a buffer was enqueued */
-	int last_dispatch;	/* age of the most recently dispatched buffer */
-	int ctxOwner;		/* last context to upload state */
-	int texAge;
-        int pf_enabled;		/* is pageflipping allowed? */
-        int pf_active;               
-        int pf_current_page;	/* which buffer is being displayed? */
-        int perf_boxes;	        /* performance boxes to be displayed */   
-	int width, height;      /* screen size in pixels */
-
-	drm_handle_t front_handle;
-	int front_offset;
-	int front_size;
-
-	drm_handle_t back_handle;
-	int back_offset;
-	int back_size;
-
-	drm_handle_t depth_handle;
-	int depth_offset;
-	int depth_size;
-
-	drm_handle_t tex_handle;
-	int tex_offset;
-	int tex_size;
-	int log_tex_granularity;
-	int pitch;
-	int rotation;           /* 0, 90, 180 or 270 */
-	int rotated_offset;
-	int rotated_size;
-	int rotated_pitch;
-	int virtualX, virtualY;
-
-	unsigned int front_tiled;
-	unsigned int back_tiled;
-	unsigned int depth_tiled;
-	unsigned int rotated_tiled;
-	unsigned int rotated2_tiled;
-
-	int pipeA_x;
-	int pipeA_y;
-	int pipeA_w;
-	int pipeA_h;
-	int pipeB_x;
-	int pipeB_y;
-	int pipeB_w;
-	int pipeB_h;
-} drmI830Sarea;
-
-/* Flags for perf_boxes
- */
-#define I830_BOX_RING_EMPTY    0x1 /* populated by kernel */
-#define I830_BOX_FLIP          0x2 /* populated by kernel */
-#define I830_BOX_WAIT          0x4 /* populated by kernel & client */
-#define I830_BOX_TEXTURE_LOAD  0x8 /* populated by kernel */
-#define I830_BOX_LOST_CONTEXT  0x10 /* populated by client */
-
-
-typedef struct {
-   	int start;		/* agp offset */
-	int used;		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830BatchBuffer;
-
-typedef struct {
-   	char *buf;		/* agp offset */
-	int sz; 		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830CmdBuffer;
- 
-typedef struct {
-	int *irq_seq;
-} drmI830IrqEmit;
-
-typedef struct {
-	int irq_seq;
-} drmI830IrqWait;
-
-typedef struct {
-	int param;
-	int *value;
-} drmI830GetParam;
-
-#define I830_PARAM_IRQ_ACTIVE     1
-#define I830_PARAM_ALLOW_BATCHBUFFER   2 
-
-typedef struct {
-	int param;
-	int value;
-} drmI830SetParam;
-
-#define I830_SETPARAM_USE_MI_BATCHBUFFER_START  1
-#define I830_SETPARAM_TEX_LRU_LOG_GRANULARITY   2
-#define I830_SETPARAM_ALLOW_BATCHBUFFER         3
-
-
-/* A memory manager for regions of shared memory:
- */
-#define I830_MEM_REGION_AGP 1
-
-typedef struct {
-	int region;
-	int alignment;
-	int size;
-	int *region_offset;	/* offset from start of fb or agp */
-} drmI830MemAlloc;
-
-typedef struct {
-	int region;
-	int region_offset;
-} drmI830MemFree;
-
-typedef struct {
-	int region;
-	int size;
-	int start;	
-} drmI830MemInitHeap;
-
-typedef struct {
-	int region;
-} drmI830MemDestroyHeap;
-
-
-#endif /* _I830_DRM_H_ */
diff --git a/src/mesa/drivers/dri/i915/server/i830_dri.h b/src/mesa/drivers/dri/i915/server/i830_dri.h
deleted file mode 100644
index 313eb759b0d..00000000000
--- a/src/mesa/drivers/dri/i915/server/i830_dri.h
+++ /dev/null
@@ -1,72 +0,0 @@
-
-#ifndef _I830_DRI_H
-#define _I830_DRI_H
-
-#include "xf86drm.h"
-#include "i830_common.h"
-
-#define I830_MAX_DRAWABLES 256
-
-#define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 3
-#define I830_PATCHLEVEL 0
-
-#define I830_REG_SIZE 0x80000
-
-typedef struct _I830DRIRec {
-   drm_handle_t regs;
-   drmSize regsSize;
-
-   drmSize backbufferSize;
-   drm_handle_t backbuffer;
-
-   drmSize depthbufferSize;
-   drm_handle_t depthbuffer;
-
-   drmSize rotatedSize;
-   drm_handle_t rotatedbuffer;
-
-   drm_handle_t textures;
-   int textureSize;
-
-   drm_handle_t agp_buffers;
-   drmSize agp_buf_size;
-
-   int deviceID;
-   int width;
-   int height;
-   int mem;
-   int cpp;
-   int bitsPerPixel;
-
-   int fbOffset;
-   int fbStride;
-
-   int backOffset;
-   int backPitch;
-
-   int depthOffset;
-   int depthPitch;
-
-   int rotatedOffset;
-   int rotatedPitch;
-
-   int logTextureGranularity;
-   int textureOffset;
-
-   int irq;
-   int sarea_priv_offset;
-} I830DRIRec, *I830DRIPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830ConfigPrivRec, *I830ConfigPrivPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830DRIContextRec, *I830DRIContextPtr;
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915/server/intel.h b/src/mesa/drivers/dri/i915/server/intel.h
deleted file mode 100644
index d7858a20c8d..00000000000
--- a/src/mesa/drivers/dri/i915/server/intel.h
+++ /dev/null
@@ -1,328 +0,0 @@
-#ifndef _INTEL_H_
-#define _INTEL_H_
-
-#include "xf86drm.h"		/* drm_handle_t, etc */
-
-/* Intel */
-#ifndef PCI_CHIP_I810
-#define PCI_CHIP_I810              0x7121
-#define PCI_CHIP_I810_DC100        0x7123
-#define PCI_CHIP_I810_E            0x7125
-#define PCI_CHIP_I815              0x1132
-#define PCI_CHIP_I810_BRIDGE       0x7120
-#define PCI_CHIP_I810_DC100_BRIDGE 0x7122
-#define PCI_CHIP_I810_E_BRIDGE     0x7124
-#define PCI_CHIP_I815_BRIDGE       0x1130
-#endif
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-
-#ifndef PCI_CHIP_I855_GM
-#define PCI_CHIP_I855_GM	   0x3582
-#define PCI_CHIP_I855_GM_BRIDGE	   0x3580
-#endif
-
-#ifndef PCI_CHIP_I865_G
-#define PCI_CHIP_I865_G		   0x2572
-#define PCI_CHIP_I865_G_BRIDGE	   0x2570
-#endif
-
-#ifndef PCI_CHIP_I915_G
-#define PCI_CHIP_I915_G		   0x2582
-#define PCI_CHIP_I915_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I915_GM
-#define PCI_CHIP_I915_GM	   0x2592
-#define PCI_CHIP_I915_GM_BRIDGE	   0x2590
-#endif
-
-#ifndef PCI_CHIP_E7221_G
-#define PCI_CHIP_E7221_G	   0x258A
-/* Same as I915_G_BRIDGE */
-#define PCI_CHIP_E7221_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I945_G
-#define PCI_CHIP_I945_G        0x2772
-#define PCI_CHIP_I945_G_BRIDGE 0x2770
-#endif
-
-#ifndef PCI_CHIP_I945_GM
-#define PCI_CHIP_I945_GM        0x27A2
-#define PCI_CHIP_I945_GM_BRIDGE 0x27A0
-#endif
-
-#define IS_I810(pI810) (pI810->Chipset == PCI_CHIP_I810 ||	\
-			pI810->Chipset == PCI_CHIP_I810_DC100 || \
-			pI810->Chipset == PCI_CHIP_I810_E)
-#define IS_I815(pI810) (pI810->Chipset == PCI_CHIP_I815)
-#define IS_I830(pI810) (pI810->Chipset == PCI_CHIP_I830_M)
-#define IS_845G(pI810) (pI810->Chipset == PCI_CHIP_845_G)
-#define IS_I85X(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM)
-#define IS_I852(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I852_GM || pI810->variant == I852_GME))
-#define IS_I855(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I855_GM || pI810->variant == I855_GME))
-#define IS_I865G(pI810) (pI810->Chipset == PCI_CHIP_I865_G)
-
-#define IS_I915G(pI810) (pI810->Chipset == PCI_CHIP_I915_G || pI810->Chipset == PCI_CHIP_E7221_G)
-#define IS_I915GM(pI810) (pI810->Chipset == PCI_CHIP_I915_GM)
-#define IS_I945G(pI810) (pI810->Chipset == PCI_CHIP_I945_G)
-#define IS_I945GM(pI810) (pI810->Chipset == PCI_CHIP_I945_GM)
-#define IS_I9XX(pI810) (IS_I915G(pI810) || IS_I915GM(pI810) || IS_I945G(pI810) || IS_I945GM(pI810))
-
-#define IS_MOBILE(pI810) (IS_I830(pI810) || IS_I85X(pI810) || IS_I915GM(pI810) || IS_I945GM(pI810))
-
-#define I830_GMCH_CTRL		0x52
-
-
-#define I830_GMCH_GMS_MASK			0x70
-#define I830_GMCH_GMS_DISABLED		0x00
-#define I830_GMCH_GMS_LOCAL			0x10
-#define I830_GMCH_GMS_STOLEN_512	0x20
-#define I830_GMCH_GMS_STOLEN_1024	0x30
-#define I830_GMCH_GMS_STOLEN_8192	0x40
-
-#define I855_GMCH_GMS_MASK			(0x7 << 4)
-#define I855_GMCH_GMS_DISABLED			0x00
-#define I855_GMCH_GMS_STOLEN_1M			(0x1 << 4)
-#define I855_GMCH_GMS_STOLEN_4M			(0x2 << 4)
-#define I855_GMCH_GMS_STOLEN_8M			(0x3 << 4)
-#define I855_GMCH_GMS_STOLEN_16M		(0x4 << 4)
-#define I855_GMCH_GMS_STOLEN_32M		(0x5 << 4)
-#define I915G_GMCH_GMS_STOLEN_48M		(0x6 << 4)
-#define I915G_GMCH_GMS_STOLEN_64M		(0x7 << 4)
-
-typedef unsigned char Bool;
-#define TRUE 1
-#define FALSE 0
-
-#define PIPE_NONE	0<<0
-#define PIPE_CRT	1<<0
-#define PIPE_TV		1<<1
-#define PIPE_DFP	1<<2
-#define PIPE_LFP	1<<3
-#define PIPE_CRT2	1<<4
-#define PIPE_TV2	1<<5
-#define PIPE_DFP2	1<<6
-#define PIPE_LFP2	1<<7
-
-typedef struct _I830MemPool *I830MemPoolPtr;
-typedef struct _I830MemRange *I830MemRangePtr;
-typedef struct _I830MemRange {
-   long Start;
-   long End;
-   long Size;
-   unsigned long Physical;
-   unsigned long Offset;		/* Offset of AGP-allocated portion */
-   unsigned long Alignment;
-   drm_handle_t Key;
-   unsigned long Pitch; // add pitch
-   I830MemPoolPtr Pool;
-} I830MemRange;
-
-typedef struct _I830MemPool {
-   I830MemRange Total;
-   I830MemRange Free;
-   I830MemRange Fixed;
-   I830MemRange Allocated;
-} I830MemPool;
-
-typedef struct {
-   int tail_mask;
-   I830MemRange mem;
-   unsigned char *virtual_start;
-   int head;
-   int tail;
-   int space;
-} I830RingBuffer;
-
-typedef struct _I830Rec {
-   unsigned char *MMIOBase;
-   unsigned char *FbBase;
-   int cpp;
-
-   unsigned int bios_version;
-
-   /* These are set in PreInit and never changed. */
-   long FbMapSize;
-   long TotalVideoRam;
-   I830MemRange StolenMemory;		/* pre-allocated memory */
-   long BIOSMemorySize;			/* min stolen pool size */
-   int BIOSMemSizeLoc;
-
-   /* These change according to what has been allocated. */
-   long FreeMemory;
-   I830MemRange MemoryAperture;
-   I830MemPool StolenPool;
-   long allocatedMemory;
-
-   /* Regions allocated either from the above pools, or from agpgart. */
-   /* for single and dual head configurations */
-   I830MemRange FrontBuffer;
-   I830MemRange FrontBuffer2;
-   I830MemRange Scratch;
-   I830MemRange Scratch2;
-
-   I830RingBuffer *LpRing;
-
-   I830MemRange BackBuffer;
-   I830MemRange DepthBuffer;
-   I830MemRange TexMem;
-   int TexGranularity;
-   I830MemRange ContextMem;
-   int drmMinor;
-   Bool have3DWindows;
-
-   Bool NeedRingBufferLow;
-   Bool allowPageFlip;
-   Bool disableTiling;
-
-   int Chipset;
-   unsigned long LinearAddr;
-   unsigned long MMIOAddr;
-
-   drmSize           registerSize;     /**< \brief MMIO register map size */
-   drm_handle_t         registerHandle;   /**< \brief MMIO register map handle */
-  //   IOADDRESS ioBase;
-   int               irq;              /**< \brief IRQ number */
-   int GttBound;
-
-   drm_handle_t ring_map;
-   unsigned int Fence[8];
-
-} I830Rec;
-
-/*
- * 12288 is set as the maximum, chosen because it is enough for
- * 1920x1440@32bpp with a 2048 pixel line pitch with some to spare.
- */
-#define I830_MAXIMUM_VBIOS_MEM		12288
-#define I830_DEFAULT_VIDEOMEM_2D	(MB(32) / 1024)
-#define I830_DEFAULT_VIDEOMEM_3D	(MB(64) / 1024)
-
-/* Flags for memory allocation function */
-#define FROM_ANYWHERE			0x00000000
-#define FROM_POOL_ONLY			0x00000001
-#define FROM_NEW_ONLY			0x00000002
-#define FROM_MASK			0x0000000f
-
-#define ALLOCATE_AT_TOP			0x00000010
-#define ALLOCATE_AT_BOTTOM		0x00000020
-#define FORCE_GAPS			0x00000040
-
-#define NEED_PHYSICAL_ADDR		0x00000100
-#define ALIGN_BOTH_ENDS			0x00000200
-#define FORCE_LOW			0x00000400
-
-#define ALLOC_NO_TILING			0x00001000
-#define ALLOC_INITIAL			0x00002000
-
-#define ALLOCATE_DRY_RUN		0x80000000
-
-/* Chipset registers for VIDEO BIOS memory RW access */
-#define _855_DRAM_RW_CONTROL 0x58
-#define _845_DRAM_RW_CONTROL 0x90
-#define DRAM_WRITE    0x33330000
-
-#define KB(x) ((x) * 1024)
-#define MB(x) ((x) * KB(1024))
-
-#define GTT_PAGE_SIZE			KB(4)
-#define ROUND_TO(x, y)			(((x) + (y) - 1) / (y) * (y))
-#define ROUND_DOWN_TO(x, y)		((x) / (y) * (y))
-#define ROUND_TO_PAGE(x)		ROUND_TO((x), GTT_PAGE_SIZE)
-#define ROUND_TO_MB(x)			ROUND_TO((x), MB(1))
-#define PRIMARY_RINGBUFFER_SIZE		KB(128)
-
-
-/* Ring buffer registers, p277, overview p19
- */
-#define LP_RING     0x2030
-#define HP_RING     0x2040
-
-#define RING_TAIL      0x00
-#define TAIL_ADDR           0x000FFFF8
-#define I830_TAIL_MASK	    0x001FFFF8
-
-#define RING_HEAD      0x04
-#define HEAD_WRAP_COUNT     0xFFE00000
-#define HEAD_WRAP_ONE       0x00200000
-#define HEAD_ADDR           0x001FFFFC
-#define I830_HEAD_MASK      0x001FFFFC
-
-#define RING_START     0x08
-#define START_ADDR          0x03FFFFF8
-#define I830_RING_START_MASK	0xFFFFF000
-
-#define RING_LEN       0x0C
-#define RING_NR_PAGES       0x001FF000 
-#define I830_RING_NR_PAGES	0x001FF000
-#define RING_REPORT_MASK    0x00000006
-#define RING_REPORT_64K     0x00000002
-#define RING_REPORT_128K    0x00000004
-#define RING_NO_REPORT      0x00000000
-#define RING_VALID_MASK     0x00000001
-#define RING_VALID          0x00000001
-#define RING_INVALID        0x00000000
-
-
-/* Fence/Tiling ranges [0..7]
- */
-#define FENCE            0x2000
-#define FENCE_NR         8
-
-#define I915G_FENCE_START_MASK	0x0ff00000
-
-#define I830_FENCE_START_MASK	0x07f80000
-
-#define FENCE_START_MASK    0x03F80000
-#define FENCE_X_MAJOR       0x00000000
-#define FENCE_Y_MAJOR       0x00001000
-#define FENCE_SIZE_MASK     0x00000700
-#define FENCE_SIZE_512K     0x00000000
-#define FENCE_SIZE_1M       0x00000100
-#define FENCE_SIZE_2M       0x00000200
-#define FENCE_SIZE_4M       0x00000300
-#define FENCE_SIZE_8M       0x00000400
-#define FENCE_SIZE_16M      0x00000500
-#define FENCE_SIZE_32M      0x00000600
-#define FENCE_SIZE_64M	    0x00000700
-#define I915G_FENCE_SIZE_1M       0x00000000
-#define I915G_FENCE_SIZE_2M       0x00000100
-#define I915G_FENCE_SIZE_4M       0x00000200
-#define I915G_FENCE_SIZE_8M       0x00000300
-#define I915G_FENCE_SIZE_16M      0x00000400
-#define I915G_FENCE_SIZE_32M      0x00000500
-#define I915G_FENCE_SIZE_64M	0x00000600
-#define I915G_FENCE_SIZE_128M	0x00000700
-#define FENCE_PITCH_1       0x00000000
-#define FENCE_PITCH_2       0x00000010
-#define FENCE_PITCH_4       0x00000020
-#define FENCE_PITCH_8       0x00000030
-#define FENCE_PITCH_16      0x00000040
-#define FENCE_PITCH_32      0x00000050
-#define FENCE_PITCH_64	    0x00000060
-#define FENCE_VALID         0x00000001
-
-#include <mmio.h>
-
-#  define MMIO_IN8(base, offset) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset))
-#  define MMIO_IN32(base, offset) \
-	read_MMIO_LE32(base, offset)
-#  define MMIO_OUT8(base, offset, val) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset)) = (val)
-#  define MMIO_OUT32(base, offset, val) \
-	*(volatile unsigned int *)(void *)(((unsigned char*)(base)) + (offset)) = CPU_TO_LE32(val)
-
-
-				/* Memory mapped register access macros */
-#define INREG8(addr)        MMIO_IN8(MMIO, addr)
-#define INREG(addr)         MMIO_IN32(MMIO, addr)
-#define OUTREG8(addr, val)  MMIO_OUT8(MMIO, addr, val)
-#define OUTREG(addr, val)   MMIO_OUT32(MMIO, addr, val)
-
-#define DSPABASE		0x70184
-
-#endif
diff --git a/src/mesa/drivers/dri/i915/server/intel_dri.c b/src/mesa/drivers/dri/i915/server/intel_dri.c
index b6946b75d20..effdd26448a 100644..120000
--- a/src/mesa/drivers/dri/i915/server/intel_dri.c
+++ b/src/mesa/drivers/dri/i915/server/intel_dri.c
@@ -1,1283 +1 @@
-/**
- * \file server/intel_dri.c
- * \brief File to perform the device-specific initialization tasks typically
- * done in the X server.
- *
- * Here they are converted to run in the client (or perhaps a standalone
- * process), and to work with the frame buffer device rather than the X
- * server infrastructure.
- * 
- * Copyright (C) 2006 Dave Airlie (airlied@linux.ie)
-
- Permission is hereby granted, free of charge, to any person obtaining a
- copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sub license, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial portions
- of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR
- ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-*/
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <errno.h>
-#include <unistd.h>
-
-#include "driver.h"
-#include "drm.h"
-
-#include "intel.h"
-#include "i830_dri.h"
-
-#include "memops.h"
-#include "pciaccess.h"
-
-static size_t drm_page_size;
-static int nextTile = 0;
-#define xf86DrvMsg(...) do {} while(0)
-
-static const int pitches[] = {
-  128 * 8,
-  128 * 16,
-  128 * 32,
-  128 * 64,
-  0
-};
-
-static Bool I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea);
-
-static unsigned long
-GetBestTileAlignment(unsigned long size)
-{
-   unsigned long i;
-
-   for (i = KB(512); i < size; i <<= 1)
-      ;
-
-   if (i > MB(64))
-      i = MB(64);
-
-   return i;
-}
-
-static void SetFenceRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-  for (i = 0; i < 8; i++) {
-    OUTREG(FENCE + i * 4, pI830->Fence[i]);
-    //    if (I810_DEBUG & DEBUG_VERBOSE_VGA)
-    fprintf(stderr,"Fence Register : %x\n", pI830->Fence[i]);
-  }
-}
-
-/* Tiled memory is good... really, really good...
- *
- * Need to make it less likely that we miss out on this - probably
- * need to move the frontbuffer away from the 'guarenteed' alignment
- * of the first memory segment, or perhaps allocate a discontigous
- * framebuffer to get more alignment 'sweet spots'.
- */
-static void
-SetFence(const DRIDriverContext *ctx, I830Rec *pI830,
-	 int nr, unsigned int start, unsigned int pitch,
-         unsigned int size)
-{
-   unsigned int val;
-   unsigned int fence_mask = 0;
-   unsigned int fence_pitch;
-
-   if (nr < 0 || nr > 7) {
-      fprintf(stderr,
-		 "SetFence: fence %d out of range\n",nr);
-      return;
-   }
-
-   pI830->Fence[nr] = 0;
-
-   if (IS_I9XX(pI830))
-   	fence_mask = ~I915G_FENCE_START_MASK;
-   else
-   	fence_mask = ~I830_FENCE_START_MASK;
-
-   if (start & fence_mask) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not %s aligned\n",
-		 nr, start, (IS_I9XX(pI830)) ? "1MB" : "512k");
-      return;
-   }
-
-   if (start % size) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not size (%dk) aligned\n",
-		 nr, start, size / 1024);
-      return;
-   }
-
-   if (pitch & 127) {
-      fprintf(stderr,
-		 "SetFence: %d: pitch (%d) not a multiple of 128 bytes\n",
-		 nr, pitch);
-      return;
-   }
-
-   val = (start | FENCE_X_MAJOR | FENCE_VALID);
-
-   if (IS_I9XX(pI830)) {
-   	switch (size) {
-	   case MB(1):
-      		val |= I915G_FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= I915G_FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= I915G_FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= I915G_FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= I915G_FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= I915G_FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= I915G_FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-    } else {
-   	switch (size) {
-	   case KB(512):
-      		val |= FENCE_SIZE_512K;
-      		break;
-	   case MB(1):
-      		val |= FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-   }
-
-   if (IS_I9XX(pI830))
-	fence_pitch = pitch / 512;
-   else
-	fence_pitch = pitch / 128;
-
-   switch (fence_pitch) {
-   case 1:
-      val |= FENCE_PITCH_1;
-      break;
-   case 2:
-      val |= FENCE_PITCH_2;
-      break;
-   case 4:
-      val |= FENCE_PITCH_4;
-      break;
-   case 8:
-      val |= FENCE_PITCH_8;
-      break;
-   case 16:
-      val |= FENCE_PITCH_16;
-      break;
-   case 32:
-      val |= FENCE_PITCH_32;
-      break;
-   case 64:
-      val |= FENCE_PITCH_64;
-      break;
-   default:
-      fprintf(stderr,
-		 "SetFence: %d: illegal pitch (%d)\n", nr, pitch);
-      return;
-   }
-
-   pI830->Fence[nr] = val;
-}
-
-static Bool
-MakeTiles(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *pMem)
-{
-   int pitch, ntiles, i;
-
-   pitch = pMem->Pitch * ctx->cpp;
-   /*
-    * Simply try to break the region up into at most four pieces of size
-    * equal to the alignment.
-    */
-   ntiles = ROUND_TO(pMem->Size, pMem->Alignment) / pMem->Alignment;
-   if (ntiles >= 4) {
-      return FALSE;
-   }
-
-   for (i = 0; i < ntiles; i++, nextTile++) {
-     SetFence(ctx, pI830, nextTile, pMem->Start + i * pMem->Alignment,
-	       pitch, pMem->Alignment);
-   }
-   return TRUE;
-}
-
-static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-
-  /* Clear out */
-  for (i = 0; i < 8; i++)
-    pI830->Fence[i] = 0;
-  
-  nextTile = 0;
-
-  if (pI830->BackBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->BackBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the back buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the back buffer.\n");
-      pI830->allowPageFlip = FALSE;
-    }
-  }
-  
-  if (pI830->DepthBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->DepthBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the depth buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the depth buffer.\n");
-    }
-  }
-
-  return;
-}
-
-static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  struct pci_device host_bridge;
-  uint32_t gmch_ctrl;
-  int memsize = 0;
-  int range;
-
-  memset(&host_bridge, 0, sizeof(host_bridge));
-
-  pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
-  
-  /* We need to reduce the stolen size, by the GTT and the popup.
-   * The GTT varying according the the FbMapSize and the popup is 4KB */
-  range = (ctx->shared.fbSize / (1024*1024)) + 4;
-
-   if (IS_I85X(pI830) || IS_I865G(pI830) || IS_I9XX(pI830)) {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I855_GMCH_GMS_STOLEN_1M:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_4M:
-	 memsize = MB(4) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_8M:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_16M:
-	 memsize = MB(16) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_32M:
-	 memsize = MB(32) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_48M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(48) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_64M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(64) - KB(range);
-	 break;
-      }
-   } else {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I830_GMCH_GMS_STOLEN_512:
-	 memsize = KB(512) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_1024:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_8192:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I830_GMCH_GMS_LOCAL:
-	 memsize = 0;
-	 xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
-		    "Local memory found, but won't be used.\n");
-	 break;
-      }
-   }
-   if (memsize > 0) {
-     fprintf(stderr,
-		 "detected %d kB stolen memory.\n", memsize / 1024);
-   } else {
-     fprintf(stderr,
-		 "no video memory detected.\n");
-   }
-   return memsize;
-}
-
-static int AgpInit(const DRIDriverContext *ctx, I830Rec *info)
-{
-  unsigned long mode = 0x4;
-
-  if (drmAgpAcquire(ctx->drmFD) < 0) {
-    fprintf(stderr, "[gart] AGP not available\n");
-    return 0;
-  }
-  
-  if (drmAgpEnable(ctx->drmFD, mode) < 0) {
-    fprintf(stderr, "[gart] AGP not enabled\n");
-    drmAgpRelease(ctx->drmFD);
-    return 0;
-  }
-  else
-    fprintf(stderr, "[gart] AGP enabled at %dx\n", ctx->agpmode);
-
-  return 1;
-}
-
-/*
- * Allocate memory from the given pool.  Grow the pool if needed and if
- * possible.
- */
-static unsigned long
-AllocFromPool(const DRIDriverContext *ctx, I830Rec *pI830, 
-	      I830MemRange *result, I830MemPool *pool,
-	      long size, unsigned long alignment, int flags)
-{
-   long needed, start, end;
-
-   if (!result || !pool || !size)
-      return 0;
-
-   /* Calculate how much space is needed. */
-   if (alignment <= GTT_PAGE_SIZE)
-      needed = size;
-   else {
-	 start = ROUND_TO(pool->Free.Start, alignment);
-	 end = ROUND_TO(start + size, alignment);
-	 needed = end - pool->Free.Start;
-   }
-   if (needed > pool->Free.Size) {
-     return 0;
-   }
-
-   result->Start = ROUND_TO(pool->Free.Start, alignment);
-   pool->Free.Start += needed;
-   result->End = pool->Free.Start;
-
-   pool->Free.Size = pool->Free.End - pool->Free.Start;
-   result->Size = result->End - result->Start;
-   result->Pool = pool;
-   result->Alignment = alignment;
-   return needed;
-}
-
-static unsigned long AllocFromAGP(const DRIDriverContext *ctx, I830Rec *pI830, long size, unsigned long alignment, I830MemRange  *result)
-{
-   unsigned long start, end;
-   unsigned long newApStart, newApEnd;
-   int ret;
-   if (!result || !size)
-      return 0;
-   
-   if (!alignment)
-     alignment = 4;
-
-   start = ROUND_TO(pI830->MemoryAperture.Start, alignment);
-   end = ROUND_TO(start + size, alignment);
-   newApStart = end;
-   newApEnd = pI830->MemoryAperture.End;
-
-   ret=drmAgpAlloc(ctx->drmFD, size, 0, &(result->Physical), (drm_handle_t *)&(result->Key));
-   
-   if (ret)
-   {
-     fprintf(stderr,"drmAgpAlloc failed %d\n", ret);
-     return 0;
-   }
-   pI830->allocatedMemory += size;
-   pI830->MemoryAperture.Start = newApStart;
-   pI830->MemoryAperture.End = newApEnd;
-   pI830->MemoryAperture.Size = newApEnd - newApStart;
-   //   pI830->FreeMemory -= size;
-   result->Start = start;
-   result->End = start + size;
-   result->Size = size;
-   result->Offset = start;
-   result->Alignment = alignment;
-   result->Pool = NULL;
-  
-   return size;
-}
-
-unsigned long
-I830AllocVidMem(const DRIDriverContext *ctx, I830Rec *pI830,
-                I830MemRange *result, I830MemPool *pool, long size,
-                unsigned long alignment, int flags)
-{
-   unsigned long ret;
-
-   if (!result)
-      return 0;
-
-   /* Make sure these are initialised. */
-   result->Size = 0;
-   result->Key = -1;
-
-   if (!size) {
-      return 0;
-   }
-
-   if (pool->Free.Size < size) {
-      ret = AllocFromAGP(ctx, pI830, size, alignment, result);
-   }
-   else {
-      ret = AllocFromPool(ctx, pI830, result, pool, size, alignment, flags);
-      if (ret == 0)
-         ret = AllocFromAGP(ctx, pI830, size, alignment, result);
-   }
-   return ret;
-}
-
-static Bool BindAgpRange(const DRIDriverContext *ctx, I830MemRange *mem)
-{
-  if (!mem)
-    return FALSE;
-  
-  if (mem->Key == -1)
-    return TRUE;
-
-  return !drmAgpBind(ctx->drmFD, mem->Key, mem->Offset);
-}
-
-/* simple memory allocation routines needed */
-/* put ring buffer in low memory */
-/* need to allocate front, back, depth buffers aligned correctly,
-   allocate ring buffer, 
-*/
-
-/* */
-static Bool
-I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned long size, ret;
-  unsigned long lines, lineSize, align;
-
-  /* allocate ring buffer */
-  memset(pI830->LpRing, 0, sizeof(I830RingBuffer));
-  pI830->LpRing->mem.Key = -1;
-
-  size = PRIMARY_RINGBUFFER_SIZE;
-  
-  ret = I830AllocVidMem(ctx, pI830, &pI830->LpRing->mem, &pI830->StolenPool, size, 0x1000, 0);
-  
-  if (ret != size)
-  {
-    fprintf(stderr,"unable to allocate ring buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  pI830->LpRing->tail_mask = pI830->LpRing->mem.Size - 1;
-
-  
-  /* allocate front buffer */
-  memset(&(pI830->FrontBuffer), 0, sizeof(pI830->FrontBuffer));
-  pI830->FrontBuffer.Key = -1;
-  pI830->FrontBuffer.Pitch = ctx->shared.virtualWidth;
-
-  align = KB(512);  
-
-  lineSize = ctx->shared.virtualWidth * ctx->cpp;
-  lines = (ctx->shared.virtualHeight + 15) / 16 * 16;
-  size = lineSize * lines;
-  size = ROUND_TO_PAGE(size);
-
-  align = GetBestTileAlignment(size);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->FrontBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate front buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->BackBuffer), 0, sizeof(pI830->BackBuffer));
-  pI830->BackBuffer.Key = -1;
-  pI830->BackBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->BackBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate back buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->DepthBuffer), 0, sizeof(pI830->DepthBuffer));
-  pI830->DepthBuffer.Key = -1;
-  pI830->DepthBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->DepthBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate depth buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->ContextMem), 0, sizeof(pI830->ContextMem));
-  pI830->ContextMem.Key = -1;
-  size = KB(32);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->ContextMem, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
-  pI830->TexMem.Key = -1;
-
-  size = 32768 * 1024;
-  ret = AllocFromAGP(ctx, pI830, size, align, &pI830->TexMem);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
-    return FALSE;
-  }
-
-  return TRUE;
-}
-
-static Bool
-I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  if (!BindAgpRange(ctx, &pI830->LpRing->mem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->FrontBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->BackBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->DepthBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->ContextMem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->TexMem))
-    return FALSE;
-
-  return TRUE;
-}
-
-static Bool
-I830CleanupDma(const DRIDriverContext *ctx)
-{
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_CLEANUP_DMA;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-     fprintf(stderr, "I830 Dma Cleanup Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830InitDma(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-   I830RingBuffer *ring = pI830->LpRing;
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_INIT_DMA;
-
-   info.ring_start = ring->mem.Start + pI830->LinearAddr;
-   info.ring_end = ring->mem.End + pI830->LinearAddr;
-   info.ring_size = ring->mem.Size;
-
-   info.mmio_offset = (unsigned int)ctx->MMIOStart;
-
-   info.sarea_priv_offset = sizeof(drm_sarea_t);
-
-   info.front_offset = pI830->FrontBuffer.Start;
-   info.back_offset = pI830->BackBuffer.Start;
-   info.depth_offset = pI830->DepthBuffer.Start;
-   info.w = ctx->shared.virtualWidth;
-   info.h = ctx->shared.virtualHeight;
-   info.pitch = ctx->shared.virtualWidth;
-   info.back_pitch = pI830->BackBuffer.Pitch;
-   info.depth_pitch = pI830->DepthBuffer.Pitch;
-   info.cpp = ctx->cpp;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-      fprintf(stderr,
-		 "I830 Dma Initialization Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static int I830CheckDRMVersion( const DRIDriverContext *ctx,
-				  I830Rec *pI830 )
-{
-   drmVersionPtr  version;
-
-   version = drmGetVersion(ctx->drmFD);
-
-   if (version) {
-     int req_minor, req_patch;
-
-     req_minor = 4;
-     req_patch = 0;	
-
-     if (version->version_major != 1 ||
-	 version->version_minor < req_minor ||
-	 (version->version_minor == req_minor && 
-	  version->version_patchlevel < req_patch)) {
-       /* Incompatible drm version */
-       fprintf(stderr,
-	       "[dri] I830DRIScreenInit failed because of a version "
-	       "mismatch.\n"
-	       "[dri] i915.o kernel module version is %d.%d.%d "
-	       "but version 1.%d.%d or newer is needed.\n"
-	       "[dri] Disabling DRI.\n",
-	       version->version_major,
-	       version->version_minor,
-	       version->version_patchlevel,
-	       req_minor,
-	       req_patch);
-       drmFreeVersion(version);
-       return 0;
-     }
-     
-     pI830->drmMinor = version->version_minor;
-     drmFreeVersion(version);
-   }
-   return 1;
-}
-
-static void
-I830SetRingRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned int itemp;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-   OUTREG(LP_RING + RING_LEN, 0);
-   OUTREG(LP_RING + RING_TAIL, 0);
-   OUTREG(LP_RING + RING_HEAD, 0);
-
-   if ((long)(pI830->LpRing->mem.Start & I830_RING_START_MASK) !=
-       pI830->LpRing->mem.Start) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer start (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Start, I830_RING_START_MASK);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = pI830->LpRing->mem.Start & I830_RING_START_MASK;
-   OUTREG(LP_RING + RING_START, itemp);
-
-   if (((pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES) !=
-       pI830->LpRing->mem.Size - 4096) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer size - 4096 (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Size - 4096,
-		 I830_RING_NR_PAGES);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = (pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES;
-   itemp |= (RING_NO_REPORT | RING_VALID);
-   OUTREG(LP_RING + RING_LEN, itemp);
-
-   pI830->LpRing->head = INREG(LP_RING + RING_HEAD) & I830_HEAD_MASK;
-   pI830->LpRing->tail = INREG(LP_RING + RING_TAIL);
-   pI830->LpRing->space = pI830->LpRing->head - (pI830->LpRing->tail + 8);
-   if (pI830->LpRing->space < 0)
-      pI830->LpRing->space += pI830->LpRing->mem.Size;
-
-   SetFenceRegs(ctx, pI830);
-   
-   /* RESET THE DISPLAY PIPE TO POINT TO THE FRONTBUFFER - hacky
-      hacky hacky */
-   OUTREG(DSPABASE, pI830->FrontBuffer.Start + pI830->LinearAddr);
-
-}
-
-static Bool
-I830SetParam(const DRIDriverContext *ctx, int param, int value)
-{
-   drmI830SetParam sp;
-
-   memset(&sp, 0, sizeof(sp));
-   sp.param = param;
-   sp.value = value;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_SETPARAM, &sp, sizeof(sp))) {
-      fprintf(stderr, "I830 SetParam Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   fprintf(stderr,
-              "[drm] Mapping front buffer\n");
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->front_offset + pI830->LinearAddr),
-                 sarea->front_size,
-                 DRM_FRAME_BUFFER,  /*DRM_AGP,*/
-                 0,
-                 &sarea->front_handle) < 0) {
-     fprintf(stderr,
-	     "[drm] drmAddMap(front_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   ctx->shared.hFrameBuffer = sarea->front_handle;
-   ctx->shared.fbSize = sarea->front_size;
-   fprintf(stderr, "[drm] Front Buffer = 0x%08x\n",
-	   sarea->front_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->back_offset),
-                 sarea->back_size, DRM_AGP, 0,
-                 &sarea->back_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(back_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Back Buffer = 0x%08x\n",
-              sarea->back_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)sarea->depth_offset,
-                 sarea->depth_size, DRM_AGP, 0,
-                 &sarea->depth_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(depth_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
-              sarea->depth_handle);
-
-   if (drmAddMap(ctx->drmFD,
-		 (drm_handle_t)sarea->tex_offset,
-		 sarea->tex_size, DRM_AGP, 0,
-		 &sarea->tex_handle) < 0) {
-      fprintf(stderr,
-		 "[drm] drmAddMap(tex_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] textures = 0x%08x\n",
-	      sarea->tex_handle);
-
-   return TRUE;
-}
-
-
-static void
-I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-#if 1
-   if (sarea->front_handle) {
-      drmRmMap(ctx->drmFD, sarea->front_handle);
-      sarea->front_handle = 0;
-   }
-#endif
-   if (sarea->back_handle) {
-      drmRmMap(ctx->drmFD, sarea->back_handle);
-      sarea->back_handle = 0;
-   }
-   if (sarea->depth_handle) {
-      drmRmMap(ctx->drmFD, sarea->depth_handle);
-      sarea->depth_handle = 0;
-   }
-   if (sarea->tex_handle) {
-      drmRmMap(ctx->drmFD, sarea->tex_handle);
-      sarea->tex_handle = 0;
-   }
-}
-
-static void
-I830InitTextureHeap(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   /* Start up the simple memory manager for agp space */
-   drmI830MemInitHeap drmHeap;
-   drmHeap.region = I830_MEM_REGION_AGP;
-   drmHeap.start  = 0;
-   drmHeap.size   = sarea->tex_size;
-      
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT_HEAP,
-			  &drmHeap, sizeof(drmHeap))) {
-      fprintf(stderr,
-		    "[drm] Failed to initialized agp heap manager\n");
-   } else {
-      fprintf(stderr,
-		    "[drm] Initialized kernel agp heap manager, %d\n",
-		    sarea->tex_size);
-
-      I830SetParam(ctx, I830_SETPARAM_TEX_LRU_LOG_GRANULARITY, 
-		      sarea->log_tex_granularity);
-   }
-}
-
-static Bool
-I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  if (drmAddMap(ctx->drmFD,
-		(drm_handle_t)pI830->LpRing->mem.Start,
-		pI830->LpRing->mem.Size, DRM_AGP, 0,
-		&pI830->ring_map) < 0) {
-    fprintf(stderr,
-	    "[drm] drmAddMap(ring_map) failed. Disabling DRI\n");
-    return FALSE;
-  }
-  fprintf(stderr, "[drm] ring buffer = 0x%08x\n",
-	  pI830->ring_map);
-
-  if (I830InitDma(ctx, pI830) == FALSE) {
-    return FALSE;
-  }
-  
-   /* init to zero to be safe */
-
-  I830DRIMapScreenRegions(ctx, pI830, sarea);
-  I830InitTextureHeap(ctx, pI830, sarea);
-
-   if (ctx->pciDevice != PCI_CHIP_845_G &&
-       ctx->pciDevice != PCI_CHIP_I830_M) {
-      I830SetParam(ctx, I830_SETPARAM_USE_MI_BATCHBUFFER_START, 1 );
-   }
-
-   /* Okay now initialize the dma engine */
-   {
-      pI830->irq = drmGetInterruptFromBusID(ctx->drmFD,
-					    ctx->pciBus,
-					    ctx->pciDevice,
-					    ctx->pciFunc);
-
-      if (drmCtlInstHandler(ctx->drmFD, pI830->irq)) {
-	 fprintf(stderr,
-		    "[drm] failure adding irq handler\n");
-	 pI830->irq = 0;
-	 return FALSE;
-      }
-      else
-	 fprintf(stderr,
-		    "[drm] dma control initialized, using IRQ %d\n",
-		    pI830->irq);
-   }
-
-   fprintf(stderr, "[dri] visual configs initialized\n");
-
-   return TRUE;
-}
-
-static Bool
-I830ClearScreen(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  /* need to drmMap front and back buffers and zero them */
-  drmAddress map_addr;
-  int ret;
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->front_handle,
-	       sarea->front_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map front buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->front_size);
-  drmUnmap(map_addr, sarea->front_size);
-
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->back_handle,
-	       sarea->back_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map back buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->back_size);
-  drmUnmap(map_addr, sarea->back_size);
-
-  return TRUE;
-}
-
-static Bool
-I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
-		  
-{
-   I830DRIPtr pI830DRI;
-   drmI830Sarea *pSAREAPriv;
-   int err;
-      
-   drm_page_size = getpagesize();   
-
-   pI830->registerSize = ctx->MMIOSize;
-   /* This is a hack for now.  We have to have more than a 4k page here
-    * because of the size of the state.  However, the state should be
-    * in a per-context mapping.  This will be added in the Mesa 3.5 port
-    * of the I830 driver.
-    */
-   ctx->shared.SAREASize = SAREA_MAX;
-
-   /* Note that drmOpen will try to load the kernel module, if needed. */
-   ctx->drmFD = drmOpen("i915", NULL );
-   if (ctx->drmFD < 0) {
-      fprintf(stderr, "[drm] drmOpen failed\n");
-      return 0;
-   }
-
-   if ((err = drmSetBusid(ctx->drmFD, ctx->pciBusID)) < 0) {
-      fprintf(stderr, "[drm] drmSetBusid failed (%d, %s), %s\n",
-	      ctx->drmFD, ctx->pciBusID, strerror(-err));
-      return 0;
-   }
-
-   if (drmAddMap( ctx->drmFD,
-		  0,
-		  ctx->shared.SAREASize,
-		  DRM_SHM,
-		  DRM_CONTAINS_LOCK,
-		  &ctx->shared.hSAREA) < 0)
-   {
-     fprintf(stderr, "[drm] drmAddMap failed\n");
-     return 0;
-   }
-
-   fprintf(stderr, "[drm] added %d byte SAREA at 0x%08x\n",
-	   ctx->shared.SAREASize, ctx->shared.hSAREA);
-   
-   if (drmMap( ctx->drmFD,
-	       ctx->shared.hSAREA,
-	       ctx->shared.SAREASize,
-	       (drmAddressPtr)(&ctx->pSAREA)) < 0)
-   {
-      fprintf(stderr, "[drm] drmMap failed\n");
-      return 0;
-   
-   }
-   
-   memset(ctx->pSAREA, 0, ctx->shared.SAREASize);
-   fprintf(stderr, "[drm] mapped SAREA 0x%08x to %p, size %d\n",
-	   ctx->shared.hSAREA, ctx->pSAREA, ctx->shared.SAREASize);
-   
-
-   if (drmAddMap(ctx->drmFD, 
-		 ctx->MMIOStart,
-		 ctx->MMIOSize,
-		 DRM_REGISTERS, 
-		 DRM_READ_ONLY, 
-		 &pI830->registerHandle) < 0) {
-      fprintf(stderr, "[drm] drmAddMap mmio failed\n");	
-      return 0;
-   }
-   fprintf(stderr,
-	   "[drm] register handle = 0x%08x\n", pI830->registerHandle);
-
-
-   if (!I830CheckDRMVersion(ctx, pI830)) {
-     return FALSE;
-   }
-
-   /* Create a 'server' context so we can grab the lock for
-    * initialization ioctls.
-    */
-   if ((err = drmCreateContext(ctx->drmFD, &ctx->serverContext)) != 0) {
-      fprintf(stderr, "%s: drmCreateContext failed %d\n", __FUNCTION__, err);
-      return 0;
-   }
-
-   DRM_LOCK(ctx->drmFD, ctx->pSAREA, ctx->serverContext, 0); 
-
-   /* Initialize the SAREA private data structure */
-   pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				 sizeof(drm_sarea_t));
-   memset(pSAREAPriv, 0, sizeof(*pSAREAPriv));
-
-   pI830->StolenMemory.Size = I830DetectMemory(ctx, pI830);
-   pI830->StolenMemory.Start = 0;
-   pI830->StolenMemory.End = pI830->StolenMemory.Size;
-
-   pI830->MemoryAperture.Start = pI830->StolenMemory.End;
-   pI830->MemoryAperture.End = KB(40000);
-   pI830->MemoryAperture.Size = pI830->MemoryAperture.End - pI830->MemoryAperture.Start;
-
-   pI830->StolenPool.Fixed = pI830->StolenMemory;
-   pI830->StolenPool.Total = pI830->StolenMemory;
-   pI830->StolenPool.Free = pI830->StolenPool.Total;
-   pI830->FreeMemory = pI830->StolenPool.Total.Size;
-
-   if (!AgpInit(ctx, pI830))
-     return FALSE;
-
-   if (I830AllocateMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   if (I830BindMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
-   pSAREAPriv->front_size = pI830->FrontBuffer.Size;
-   pSAREAPriv->width = ctx->shared.virtualWidth;
-   pSAREAPriv->height = ctx->shared.virtualHeight;
-   pSAREAPriv->pitch = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualX = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualY = ctx->shared.virtualHeight;
-   pSAREAPriv->back_offset = pI830->BackBuffer.Start;
-   pSAREAPriv->back_size = pI830->BackBuffer.Size;
-   pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
-   pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
-   pSAREAPriv->tex_offset = pI830->TexMem.Start;
-   pSAREAPriv->tex_size = pI830->TexMem.Size;
-   pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
-
-   ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
-   ctx->driverClientMsgSize = sizeof(I830DRIRec);
-   pI830DRI = (I830DRIPtr)ctx->driverClientMsg;
-   pI830DRI->deviceID = pI830->Chipset;
-   pI830DRI->regsSize = I830_REG_SIZE;
-   pI830DRI->width = ctx->shared.virtualWidth;
-   pI830DRI->height = ctx->shared.virtualHeight;
-   pI830DRI->mem = ctx->shared.fbSize;
-   pI830DRI->cpp = ctx->cpp;
-   pI830DRI->backOffset = pI830->BackBuffer.Start;
-   pI830DRI->backPitch = pI830->BackBuffer.Pitch; 
-
-   pI830DRI->depthOffset = pI830->DepthBuffer.Start;
-   pI830DRI->depthPitch = pI830->DepthBuffer.Pitch; 
-
-   pI830DRI->fbOffset = pI830->FrontBuffer.Start;
-   pI830DRI->fbStride = pI830->FrontBuffer.Pitch;
-
-   pI830DRI->bitsPerPixel = ctx->bpp;
-   pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
-   
-   err = I830DRIDoMappings(ctx, pI830, pSAREAPriv);
-   if (err == FALSE)
-       return FALSE;
-
-   I830SetupMemoryTiling(ctx, pI830);
-
-   /* Quick hack to clear the front & back buffers.  Could also use
-    * the clear ioctl to do this, but would need to setup hw state
-    * first.
-    */
-   I830ClearScreen(ctx, pI830, pSAREAPriv);
-
-   I830SetRingRegs(ctx, pI830);
-
-   return TRUE;
-}
-
-
-/**
- * \brief Validate the fbdev mode.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Saves some registers and returns 1.
- *
- * \sa radeonValidateMode().
- */
-static int i830ValidateMode( const DRIDriverContext *ctx )
-{
-  return 1;
-}
-
-/**
- * \brief Examine mode returned by fbdev.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Restores registers that fbdev has clobbered and returns 1.
- *
- * \sa i810ValidateMode().
- */
-static int i830PostValidateMode( const DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = ctx->driverPrivate;
-
-  I830SetRingRegs(ctx, pI830);
-  return 1;
-}
-
-
-/**
- * \brief Initialize the framebuffer device mode
- *
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Fills in \p info with some default values and some information from \p ctx
- * and then calls I810ScreenInit() for the screen initialization.
- * 
- * Before exiting clears the framebuffer memory accessing it directly.
- */
-static int i830InitFBDev( DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = calloc(1, sizeof(I830Rec));
-  int i;
-
-   {
-      int  dummy = ctx->shared.virtualWidth;
-
-      switch (ctx->bpp / 8) {
-      case 1: dummy = (ctx->shared.virtualWidth + 127) & ~127; break;
-      case 2: dummy = (ctx->shared.virtualWidth +  31) &  ~31; break;
-      case 3:
-      case 4: dummy = (ctx->shared.virtualWidth +  15) &  ~15; break;
-      }
-
-      ctx->shared.virtualWidth = dummy;
-      ctx->shared.Width = ctx->shared.virtualWidth;
-   }
-
-
-   for (i = 0; pitches[i] != 0; i++) {
-     if (pitches[i] >= ctx->shared.virtualWidth) {
-       ctx->shared.virtualWidth = pitches[i];
-       break;
-     }
-   }
-
-   ctx->driverPrivate = (void *)pI830;
-   
-   pI830->LpRing = calloc(1, sizeof(I830RingBuffer));
-   pI830->Chipset = ctx->chipset;
-   pI830->LinearAddr = ctx->FBStart;
-
-   if (!I830ScreenInit( ctx, pI830 ))
-      return 0;
-
-   
-   return 1;
-}
-
-
-/**
- * \brief The screen is being closed, so clean up any state and free any
- * resources used by the DRI.
- *
- * \param ctx display handle.
- *
- * Unmaps the SAREA, closes the DRM device file descriptor and frees the driver
- * private data.
- */
-static void i830HaltFBDev( DRIDriverContext *ctx )
-{
-  drmI830Sarea *pSAREAPriv;
-  I830Rec *pI830 = ctx->driverPrivate;
-
-   if (pI830->irq) {
-       drmCtlUninstHandler(ctx->drmFD);
-       pI830->irq = 0;   }
-
-   I830CleanupDma(ctx);
-
-  pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				sizeof(drm_sarea_t));
-
-  I830DRIUnmapScreenRegions(ctx, pI830, pSAREAPriv);
-  drmUnmap( ctx->pSAREA, ctx->shared.SAREASize );
-  drmClose(ctx->drmFD);
-  
-  if (ctx->driverPrivate) {
-    free(ctx->driverPrivate);
-    ctx->driverPrivate = 0;
-  }
-}
-
-
-extern void i810NotifyFocus( int );
-
-/**
- * \brief Exported driver interface for Mini GLX.
- *
- * \sa DRIDriverRec.
- */
-const struct DRIDriverRec __driDriver = {
-   i830ValidateMode,
-   i830PostValidateMode,
-   i830InitFBDev,
-   i830HaltFBDev,
-   NULL,//I830EngineShutdown,
-   NULL, //I830EngineRestore,  
-#ifndef _EMBEDDED
-   0,
-#else
-   i810NotifyFocus, 
-#endif
-};
+../../intel/server/intel_dri.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 9e4ff112dc3..005460f3547 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -5,25 +5,30 @@ include $(TOP)/configs/current
 LIBNAME = i965_dri.so
 
 DRIVER_SOURCES = \
-	bufmgr_fake.c \
 	intel_batchbuffer.c \
 	intel_blit.c \
 	intel_buffer_objects.c \
 	intel_buffers.c \
 	intel_context.c \
-	intel_ioctl.c \
+	intel_decode.c \
+	intel_depthstencil.c \
+	intel_fbo.c \
 	intel_mipmap_tree.c \
 	intel_regions.c \
 	intel_screen.c \
 	intel_span.c \
-	intel_pixel_copy.c \
+	intel_pixel.c \
 	intel_pixel_bitmap.c \
+	intel_pixel_copy.c \
+	intel_pixel_draw.c \
 	intel_state.c \
 	intel_tex.c \
+	intel_tex_copy.c \
+	intel_tex_format.c \
+	intel_tex_image.c \
 	intel_tex_layout.c \
+	intel_tex_subimage.c \
 	intel_tex_validate.c \
-	brw_aub.c \
-	brw_aub_playback.c \
 	brw_cc.c \
 	brw_clip.c \
 	brw_clip_line.c \
@@ -44,16 +49,16 @@ DRIVER_SOURCES = \
 	brw_gs.c \
 	brw_gs_emit.c \
 	brw_gs_state.c \
-	brw_hal.c \
 	brw_metaops.c \
 	brw_misc_state.c \
 	brw_program.c \
+	brw_queryobj.c \
 	brw_sf.c \
 	brw_sf_emit.c \
 	brw_sf_state.c \
 	brw_state_batch.c \
 	brw_state_cache.c \
-	brw_state_pool.c \
+	brw_state_dump.c \
 	brw_state_upload.c \
 	brw_tex.c \
 	brw_tex_layout.c \
@@ -70,6 +75,7 @@ DRIVER_SOURCES = \
 	brw_wm_emit.c \
 	brw_wm_fp.c \
 	brw_wm_iz.c \
+	brw_wm_glsl.c \
 	brw_wm_pass0.c \
 	brw_wm_pass1.c \
 	brw_wm_pass2.c \
@@ -84,10 +90,12 @@ C_SOURCES = \
 
 ASM_SOURCES = 
 
-DRIVER_DEFINES = -I../intel
+DRIVER_DEFINES = -I../intel -I../intel/server
 
-include ../Makefile.template
+DRI_LIB_DEPS += -ldrm_intel
 
-intel_tex_layout.o: ../intel/intel_tex_layout.c
+include ../Makefile.template
 
 symlinks:
+intel_decode.o: ../intel/intel_decode.c
+intel_tex_layout.o: ../intel/intel_tex_layout.c
diff --git a/src/mesa/drivers/dri/i965/brw_aub.c b/src/mesa/drivers/dri/i965/brw_aub.c
deleted file mode 100644
index f851a5b7955..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub.c
+++ /dev/null
@@ -1,353 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#include "brw_context.h"
-#include "brw_aub.h"
-#include "intel_regions.h"
-#include <stdio.h>
-
-extern char *__progname;
-
-
-/* Registers to control page table
- */
-#define PGETBL_CTL       0x2020
-#define PGETBL_ENABLED   0x1
-
-#define NR_GTT_ENTRIES  65536	/* 256 mb */
-
-#define FAIL										\
-do {											\
-   fprintf(stderr, "failed to write aub data at %s/%d\n", __FUNCTION__, __LINE__);	\
-   exit(1);										\
-} while (0)
-
-
-/* Emit the headers at the top of each aubfile.  Initialize the GTT.
- */
-static void init_aubfile( FILE *aub_file )
-{   
-   struct aub_file_header fh;
-   struct aub_block_header bh;
-   unsigned int data;
-
-   static int nr;
-   
-   nr++;
-
-   /* Emit the aub header:
-    */
-   memset(&fh, 0, sizeof(fh));
-
-   fh.instruction_type = AUB_FILE_HEADER;
-   fh.minor = 0x0;
-   fh.major = 0x7;
-   memcpy(fh.application, __progname, sizeof(fh.application));
-   fh.day = (nr>>24) & 0xff;
-   fh.month = 0x0;
-   fh.year = 0x0;
-   fh.timezone = 0x0;
-   fh.second = nr & 0xff;
-   fh.minute = (nr>>8) & 0xff;
-   fh.hour = (nr>>16) & 0xff;
-   fh.comment_length = 0x0;   
-
-   if (fwrite(&fh, sizeof(fh), 1, aub_file) < 0) 
-      FAIL;
-         
-   /* Setup the GTT starting at main memory address zero (!):
-    */
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_MMI0_WRITE32;
-   bh.type = 0x0;
-   bh.address_space = ADDR_GTT;	/* ??? */
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address = PGETBL_CTL;
-   bh.length = 0x4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   data = 0x0 | PGETBL_ENABLED;
-
-   if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-      FAIL;
-}
-
-
-static void init_aub_gtt( struct brw_context *brw,
-			  GLuint start_offset, 
-			  GLuint size )
-{
-   FILE *aub_file = brw->intel.aub_file;
-   struct aub_block_header bh;
-   unsigned int i;
-
-   assert(start_offset + size < NR_GTT_ENTRIES * 4096);
-
-
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = 0x0;
-   bh.address_space = ADDR_MAIN;
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address =  start_offset / 4096 * 4;
-   bh.length = size / 4096 * 4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   for (i = 0; i < size / 4096; i++) {
-      GLuint data = brw->next_free_page | 1;
-
-      brw->next_free_page += 4096;
-
-      if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-	 FAIL;
-   }
-
-}
-
-static void write_block_header( FILE *aub_file,
-				struct aub_block_header *bh,
-				const GLuint *data,
-				GLuint sz )
-{
-   sz = (sz + 3) & ~3;
-
-   if (fwrite(bh, sizeof(*bh), 1, aub_file) < 0) 
-      FAIL;
-
-   if (fwrite(data, sz, 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-static void write_dump_bmp( FILE *aub_file,
-			    struct aub_dump_bmp *db )
-{
-   if (fwrite(db, sizeof(*db), 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-
-static void brw_aub_gtt_data( struct intel_context *intel,
-			      GLuint offset,
-			      const void *data,
-			      GLuint sz,
-			      GLuint type,
-			      GLuint state_type )
-{
-   struct aub_block_header bh;
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-
-   if (type == DW_GENERAL_STATE) {
-      bh.general_state_type = state_type;
-      bh.surface_state_type = 0;
-   }
-   else {
-      bh.general_state_type = 0;
-      bh.surface_state_type = state_type;
-   }
-
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(intel->aub_file, &bh, data, sz);
-}
-
-
-
-static void brw_aub_gtt_cmds( struct intel_context *intel,
-			      GLuint offset,
-			      const void *data,
-			      GLuint sz )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   struct aub_block_header bh;   
-   GLuint type = CW_PRIMARY_RING_A;
-   
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_COMMAND_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-   bh.general_state_type = 0;
-   bh.surface_state_type = 0;
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(brw->intel.aub_file, &bh, data, sz);
-}
-
-static void brw_aub_dump_bmp( struct intel_context *intel,
-			      GLuint buffer )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   intelScreenPrivate *intelScreen = brw->intel.intelScreen;
-   struct aub_dump_bmp db;
-   GLuint format;
-
-   if (intelScreen->cpp == 4)
-      format = 0x7;
-   else
-      format = 0x3;
-
-
-   if (buffer == 0) {
-      db.instruction_type = AUB_DUMP_BMP;
-      db.xmin = 0;
-      db.ymin = 0;
-      db.format = format;
-      db.bpp = intelScreen->cpp * 8;
-      db.pitch = intelScreen->front.pitch / intelScreen->cpp;
-      db.xsize = intelScreen->width;
-      db.ysize = intelScreen->height;
-      db.addr = intelScreen->front.offset;
-      db.unknown = 0x0;		/* 4: xmajor tiled, 0: not tiled */
-
-      write_dump_bmp(brw->intel.aub_file, &db);
-   }
-   else {
-      db.instruction_type = AUB_DUMP_BMP;
-      db.xmin = 0;
-      db.ymin = 0;
-      db.format = format;
-      db.bpp = intel->back_region->cpp * 8;
-      db.pitch = intel->back_region->pitch;
-      db.xsize = intel->back_region->pitch;
-      db.ysize = intel->back_region->height;
-      db.addr = intelScreen->back.offset;
-      db.unknown = intel->back_region->tiled ? 0x4 : 0x0;
-
-      write_dump_bmp(brw->intel.aub_file, &db);
-   }
-}
-
-/* Attempt to prevent monster aubfiles by closing and reopening when
- * the state pools wrap.
- */
-static void brw_aub_wrap( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);   
-   if (intel->aub_file) {
-      brw_aub_destroy(brw);
-      brw_aub_init(brw);
-   }
-   brw->wrap = 1;		/* ??? */
-}
-
-
-int brw_aub_init( struct brw_context *brw )
-{
-   struct intel_context *intel = &brw->intel;
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   char filename[80];
-   int val;
-   static int i = 0;
-
-   i++;
-
-   if (_mesa_getenv("INTEL_REPLAY"))
-      return 0;
-
-   if (_mesa_getenv("INTEL_AUBFILE")) {
-      val = snprintf(filename, sizeof(filename), "%s%d.aub", _mesa_getenv("INTEL_AUBFILE"), i%4);
-      _mesa_printf("--> Aub file: %s\n", filename);
-      brw->intel.aub_file = fopen(filename, "w");
-   }
-   else if (_mesa_getenv("INTEL_AUB")) {
-      val = snprintf(filename, sizeof(filename), "%s.aub", __progname);
-      if (val < 0 || val > sizeof(filename)) 
-	 strcpy(filename, "default.aub");   
-   
-      _mesa_printf("--> Aub file: %s\n", filename);
-      brw->intel.aub_file = fopen(filename, "w");
-   }
-   else {
-      return 0;
-   }
-
-   if (!brw->intel.aub_file) {
-      _mesa_printf("couldn't open aubfile\n");
-      exit(1);
-   }
-
-   brw->intel.vtbl.aub_commands = brw_aub_gtt_cmds;
-   brw->intel.vtbl.aub_dump_bmp = brw_aub_dump_bmp;
-   brw->intel.vtbl.aub_gtt_data = brw_aub_gtt_data;
-   brw->intel.vtbl.aub_wrap = brw_aub_wrap;
-   
-   init_aubfile(brw->intel.aub_file);
-
-   /* The GTT is located starting address zero in main memory.  Pages
-    * to populate the gtt start after this point.
-    */
-   brw->next_free_page = (NR_GTT_ENTRIES * 4 + 4095) & ~4095;
-
-   /* More or less correspond with all the agp regions mapped by the
-    * driver:
-    */
-   init_aub_gtt(brw, 0, 4096*4); /* so new fulsim doesn't crash */
-   init_aub_gtt(brw, intelScreen->front.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->back.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->depth.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->tex.offset, intelScreen->tex.size);
-
-   return 0;
-}
-
-void brw_aub_destroy( struct brw_context *brw )
-{
-   if (brw->intel.aub_file) {
-      fclose(brw->intel.aub_file);
-      brw->intel.aub_file = NULL;
-   }
-}
diff --git a/src/mesa/drivers/dri/i965/brw_aub.h b/src/mesa/drivers/dri/i965/brw_aub.h
deleted file mode 100644
index 198e36dc3c0..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub.h
+++ /dev/null
@@ -1,172 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-
-#ifndef BRW_AUB_H
-#define BRW_AUB_H
-
-struct aub_file_header {
-   unsigned int instruction_type;
-   unsigned int pad0:16;
-   unsigned int minor:8;
-   unsigned int major:8;
-   unsigned char application[8*4];
-   unsigned int day:8;
-   unsigned int month:8;
-   unsigned int year:16;
-   unsigned int timezone:8;
-   unsigned int second:8;
-   unsigned int minute:8;
-   unsigned int hour:8;
-   unsigned int comment_length:16;   
-   unsigned int pad1:16;
-};
-
-struct aub_block_header {
-   unsigned int instruction_type;
-   unsigned int operation:8;
-   unsigned int type:8;
-   unsigned int address_space:8;
-   unsigned int pad0:8;
-   unsigned int general_state_type:8;
-   unsigned int surface_state_type:8;
-   unsigned int pad1:16;
-   unsigned int address;
-   unsigned int length;
-};
-
-struct aub_dump_bmp {
-   unsigned int instruction_type;
-   unsigned int xmin:16;
-   unsigned int ymin:16;
-   unsigned int pitch:16;
-   unsigned int bpp:8;
-   unsigned int format:8;
-   unsigned int xsize:16;
-   unsigned int ysize:16;
-   unsigned int addr;
-   unsigned int unknown;
-};
-
-enum bh_operation {
-   BH_COMMENT,
-   BH_DATA_WRITE,
-   BH_COMMAND_WRITE,
-   BH_MMI0_WRITE32,
-   BH_END_SCENE,
-   BH_CONFIG_MEMORY_MAP,
-   BH_MAX_OPERATION
-};
-
-enum command_write_type {
-   CW_HWB_RING = 1,
-   CW_PRIMARY_RING_A,
-   CW_PRIMARY_RING_B,		/* XXX - disagreement with listaub! */
-   CW_PRIMARY_RING_C,
-   CW_MAX_TYPE
-};
-
-enum data_write_type {
-   DW_NOTYPE,
-   DW_BATCH_BUFFER,
-   DW_BIN_BUFFER,
-   DW_BIN_POINTER_LIST,
-   DW_SLOW_STATE_BUFFER,
-   DW_VERTEX_BUFFER,
-   DW_2D_MAP,
-   DW_CUBE_MAP,
-   DW_INDIRECT_STATE_BUFFER,
-   DW_VOLUME_MAP,
-   DW_1D_MAP,
-   DW_CONSTANT_BUFFER,
-   DW_CONSTANT_URB_ENTRY,
-   DW_INDEX_BUFFER,
-   DW_GENERAL_STATE,
-   DW_SURFACE_STATE,
-   DW_MEDIA_OBJECT_INDIRECT_DATA,
-   DW_MAX_TYPE
-};
-
-enum data_write_general_state_type {
-   DWGS_NOTYPE,
-   DWGS_VERTEX_SHADER_STATE,
-   DWGS_GEOMETRY_SHADER_STATE ,
-   DWGS_CLIPPER_STATE,
-   DWGS_STRIPS_FANS_STATE,
-   DWGS_WINDOWER_IZ_STATE,
-   DWGS_COLOR_CALC_STATE,
-   DWGS_CLIPPER_VIEWPORT_STATE,	/* was 0x7 */
-   DWGS_STRIPS_FANS_VIEWPORT_STATE,
-   DWGS_COLOR_CALC_VIEWPORT_STATE, /* was 0x9 */
-   DWGS_SAMPLER_STATE,
-   DWGS_KERNEL_INSTRUCTIONS,
-   DWGS_SCRATCH_SPACE,
-   DWGS_SAMPLER_DEFAULT_COLOR,
-   DWGS_INTERFACE_DESCRIPTOR,
-   DWGS_VLD_STATE,
-   DWGS_VFE_STATE,
-   DWGS_MAX_TYPE
-};
-
-enum data_write_surface_state_type {
-   DWSS_NOTYPE,
-   DWSS_BINDING_TABLE_STATE,
-   DWSS_SURFACE_STATE,
-   DWSS_MAX_TYPE
-};
-
-enum memory_map_type {
-   MM_DEFAULT,
-   MM_DYNAMIC,
-   MM_MAX_TYPE
-};
-
-enum address_space {
-   ADDR_GTT,
-   ADDR_LOCAL,
-   ADDR_MAIN,
-   ADDR_MAX
-};
-
-
-#define AUB_FILE_HEADER 0xe085000b
-#define AUB_BLOCK_HEADER 0xe0c10003
-#define AUB_DUMP_BMP 0xe09e0004
-
-struct brw_context;
-struct intel_context;
-
-int brw_aub_init( struct brw_context *brw );
-void brw_aub_destroy( struct brw_context *brw );
-
-int brw_playback_aubfile(struct brw_context *brw,
-			 const char *filename);
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_aub_playback.c b/src/mesa/drivers/dri/i965/brw_aub_playback.c
deleted file mode 100644
index 2433d50c116..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub_playback.c
+++ /dev/null
@@ -1,443 +0,0 @@
-
-#include <stdio.h>
-#include <sys/mman.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <unistd.h>
-#include <fcntl.h>
-
-#include "brw_aub.h"
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "intel_ioctl.h"
-#include "bufmgr.h"
-
-struct aub_state {
-   struct intel_context *intel;
-   const char *map;
-   unsigned int csr;
-   unsigned int sz;
-};
-
-
-static int gobble( struct aub_state *s, int size )
-{
-   if (s->csr + size > s->sz) {
-      _mesa_printf("EOF in %s\n", __FUNCTION__);
-      return 1;
-   }
-
-   s->csr += size;
-   return 0;
-}
-
-static void flush_and_fence( struct aub_state *s )
-{
-   struct intel_context *intel = s->intel;
-   GLuint buf[2];
-
-   buf[0] = intel->vtbl.flush_cmd();
-   buf[1] = 0;
-
-   intel_cmd_ioctl(intel, (char *)&buf, sizeof(buf));
-      
-   intelWaitIrq( intel, intelEmitIrqLocked( intel ));
-}
-
-static void flush_cmds( struct aub_state *s,
-			const void *data,
-			int len )
-{
-   DBG("%s %d\n", __FUNCTION__, len);
-
-   if (len & 0x4) {
-      unsigned int *tmp = malloc(len + 4);
-      DBG("padding to octword\n");
-      memcpy(tmp, data, len);
-      tmp[len/4] = MI_NOOP;
-      flush_cmds(s, tmp, len+4);
-      free(tmp);
-      return;
-   }
-
-   /* For ring data, just send off immediately via an ioctl.
-    * This differs slightly from how the stream was executed
-    * initially as this would have been a batchbuffer.
-    */
-   intel_cmd_ioctl(s->intel, (void *)data, len);
-
-   if (1)
-      flush_and_fence(s);
-}
-
-static const char *pstrings[] = {
-   "none",
-   "POINTLIST",
-   "LINELIST",
-   "LINESTRIP",
-   "TRILIST",
-   "TRISTRIP",
-   "TRIFAN",
-   "QUADLIST",
-   "QUADSTRIP",
-   "LINELIST_ADJ",
-   "LINESTRIP_ADJ",
-   "TRILIST_ADJ",
-   "TRISTRIP_ADJ",
-   "TRISTRIP_REVERSE",
-   "POLYGON",
-   "RECTLIST",
-   "LINELOOP",
-   "POINTLIST_BF",
-   "LINESTRIP_CONT",
-   "LINESTRIP_BF",
-   "LINESTRIP_CONT_BF",
-   "TRIFAN_NOSTIPPLE",
-};
-
-static void do_3d_prim( struct aub_state *s,
-			const void *data,
-			int len )
-{
-   struct brw_3d_primitive prim;
-   const struct brw_3d_primitive *orig = data;
-   int i;
-
-   assert(len == sizeof(prim));
-   memcpy(&prim, data, sizeof(prim));
-
-#define START 0
-#define BLOCK (12*28)
-
-   if (orig->verts_per_instance < BLOCK)
-      flush_cmds(s, &prim, sizeof(prim));
-   else {
-      for (i = START; i + BLOCK < orig->verts_per_instance; i += BLOCK/2) {
-	 prim.start_vert_location = i;
-	 prim.verts_per_instance = BLOCK;
-	 _mesa_printf("%sprim %d/%s verts %d..%d (of %d)\n", 
-		      prim.header.indexed ? "INDEXED " : "",
-		      prim.header.topology, pstrings[prim.header.topology%16],
-		      prim.start_vert_location, 
-		      prim.start_vert_location + prim.verts_per_instance,
-		      orig->verts_per_instance);
-	 flush_cmds(s, &prim, sizeof(prim));
-      }
-   }
-}
-
-
-
-static struct {
-   int cmd;
-   const char *name;
-   int has_length;
-} cmd_info[] = {
-   { 0, "NOOP", 0 },
-   { 0x5410, "XY_COLOR_BLT_RGB", 1 },
-   { 0x5430, "XY_COLOR_BLT_RGBA", 1 },
-   { 0x54d0, "XY_SRC_COPY_BLT_RGB", 1 },
-   { 0x54f0, "XY_SRC_COPY_BLT_RGBA", 1 },
-   { CMD_URB_FENCE, "URB_FENCE",  1 },
-   { CMD_CONST_BUFFER_STATE, "CONST_BUFFER_STATE",  1 },
-   { CMD_CONST_BUFFER, "CONST_BUFFER",  1 },
-   { CMD_STATE_BASE_ADDRESS, "STATE_BASE_ADDRESS",  1 },
-   { CMD_STATE_INSN_POINTER, "STATE_INSN_POINTER",  1 },
-   { CMD_PIPELINE_SELECT, "PIPELINE_SELECT", 0, },
-   { CMD_PIPELINED_STATE_POINTERS, "PIPELINED_STATE_POINTERS", 1 },
-   { CMD_BINDING_TABLE_PTRS, "BINDING_TABLE_PTRS", 1 },
-   { CMD_VERTEX_BUFFER, "VERTEX_BUFFER", 1 },
-   { CMD_VERTEX_ELEMENT, "VERTEX_ELEMENT", 1 },
-   { CMD_INDEX_BUFFER, "INDEX_BUFFER", 1 },
-   { CMD_VF_STATISTICS, "VF_STATISTICS", 0 },
-   { CMD_DRAW_RECT, "DRAW_RECT", 1 },
-   { CMD_BLEND_CONSTANT_COLOR, "BLEND_CONSTANT_COLOR", 1 },
-   { CMD_CHROMA_KEY, "CHROMA_KEY", 1 },
-   { CMD_DEPTH_BUFFER, "DEPTH_BUFFER", 1 },
-   { CMD_POLY_STIPPLE_OFFSET, "POLY_STIPPLE_OFFSET", 1 },
-   { CMD_POLY_STIPPLE_PATTERN, "POLY_STIPPLE_PATTERN", 1 },
-   { CMD_LINE_STIPPLE_PATTERN, "LINE_STIPPLE_PATTERN", 1 },
-   { CMD_GLOBAL_DEPTH_OFFSET_CLAMP, "GLOBAL_DEPTH_OFFSET_CLAMP", 1 },
-   { CMD_PIPE_CONTROL, "PIPE_CONTROL", 1 },
-   { CMD_MI_FLUSH, "MI_FLUSH", 0 },
-   { CMD_3D_PRIM, "3D_PRIM", 1 },
-};
-
-#define NR_CMDS (sizeof(cmd_info)/sizeof(cmd_info[0]))
-
-
-static int find_command( unsigned int cmd )
-{
-   int i;
-
-   for (i = 0; i < NR_CMDS; i++) 
-      if (cmd == cmd_info[i].cmd) 
-	 return i;
-
-   return -1;
-}
-
-
-
-static int parse_commands( struct aub_state *s,
-			   const unsigned int *data,
-			   int len )
-{
-   while (len) {
-      int cmd = data[0] >> 16;
-      int dwords;
-      int i;
-
-      i = find_command(cmd);
-
-      if (i < 0) {
-	 _mesa_printf("couldn't find info for cmd %x\n", cmd);
-	 return 1;
-      }
-
-      if (cmd_info[i].has_length)
-	 dwords = (data[0] & 0xff) + 2;
-      else
-	 dwords = 1;
-
-      _mesa_printf("%s (%d dwords) 0x%x\n", cmd_info[i].name, dwords, data[0]);
-
-      if (len < dwords * 4) {
-	 _mesa_printf("EOF in %s (%d bytes)\n", __FUNCTION__, len);
-	 return 1;
-      }
-
-
-      if (0 && cmd == CMD_3D_PRIM)
-	 do_3d_prim(s, data, dwords * 4);
-      else
-	 flush_cmds(s, data, dwords * 4);
-
-      data += dwords;
-      len -= dwords * 4;
-   }
-
-   return 0;
-}
-
-
-
-static void parse_data_write( struct aub_state *s,
-			     const struct aub_block_header *bh,
-			     void *dest,
-			     const unsigned int *data,
-			     int len )
-{
-   switch (bh->type) {
-   case DW_GENERAL_STATE:
-      switch (bh->general_state_type) {
-      case DWGS_VERTEX_SHADER_STATE: {
-	 struct brw_vs_unit_state vs;
-	 assert(len == sizeof(vs));
-
-	 _mesa_printf("DWGS_VERTEX_SHADER_STATE\n");
-	 memcpy(&vs, data, sizeof(vs));
-
-/* 	 vs.vs6.vert_cache_disable = 1;  */
-/*  	 vs.thread4.max_threads = 4;  */
-
-	 memcpy(dest, &vs, sizeof(vs));
-	 return;
-      }
-      case DWGS_CLIPPER_STATE: {
-	 struct brw_clip_unit_state clip;
-	 assert(len == sizeof(clip));
-
-	 _mesa_printf("DWGS_CLIPPER_STATE\n");
-	 memcpy(&clip, data, sizeof(clip));
-
-/* 	 clip.thread4.max_threads = 0; */
-/*   	 clip.clip5.clip_mode = BRW_CLIPMODE_REJECT_ALL;   */
-
-	 memcpy(dest, &clip, sizeof(clip));
-	 return;
-      }
-
-      case DWGS_NOTYPE:
-      case DWGS_GEOMETRY_SHADER_STATE:
-      case DWGS_STRIPS_FANS_STATE:
-	 break;
-
-      case DWGS_WINDOWER_IZ_STATE: {
-	    struct brw_wm_unit_state wm;
-	    assert(len == sizeof(wm));
-
-	    _mesa_printf("DWGS_WINDOWER_IZ_STATE\n");
-	    memcpy(&wm, data, sizeof(wm));
-
-/* 	    wm.wm5.max_threads = 10; */
-
-	    memcpy(dest, &wm, sizeof(wm));
-	    return;
-	 }
-
-      case DWGS_COLOR_CALC_STATE:
-      case DWGS_CLIPPER_VIEWPORT_STATE:
-      case DWGS_STRIPS_FANS_VIEWPORT_STATE:
-      case DWGS_COLOR_CALC_VIEWPORT_STATE:
-      case DWGS_SAMPLER_STATE:
-      case DWGS_KERNEL_INSTRUCTIONS:
-      case DWGS_SCRATCH_SPACE:
-      case DWGS_SAMPLER_DEFAULT_COLOR:
-      case DWGS_INTERFACE_DESCRIPTOR:
-      case DWGS_VLD_STATE:
-      case DWGS_VFE_STATE:
-      default:
-	 break;
-      }
-      break;
-   case DW_SURFACE_STATE:
-      break;
-   case DW_1D_MAP:
-   case DW_2D_MAP:
-   case DW_CUBE_MAP:
-   case DW_VOLUME_MAP:
-   case DW_CONSTANT_BUFFER:
-   case DW_CONSTANT_URB_ENTRY:
-   case DW_VERTEX_BUFFER:
-   case DW_INDEX_BUFFER:
-   default:
-      break;
-   }
-
-   memcpy(dest, data, len);
-}
-
-
-/* In order to work, the memory layout has to be the same as the X
- * server which created the aubfile.
- */
-static int parse_block_header( struct aub_state *s )
-{
-   struct aub_block_header *bh = (struct aub_block_header *)(s->map + s->csr);
-   void *data = (void *)(bh + 1);
-   unsigned int len = (bh->length + 3) & ~3;
-
-   _mesa_printf("block header at 0x%x\n", s->csr);
-
-   if (s->csr + len + sizeof(*bh) > s->sz) {
-      _mesa_printf("EOF in data in %s\n", __FUNCTION__);
-      return 1;
-   }
-
-   if (bh->address_space == ADDR_GTT) {
-
-      switch (bh->operation)
-      {
-      case BH_DATA_WRITE: {
-	 void *dest = bmFindVirtual( s->intel, bh->address, len );
-	 if (dest == NULL) {
-	    _mesa_printf("Couldn't find virtual address for offset %x\n", bh->address);
-	    return 1;
-	 }
-
-#if 1
-	 parse_data_write(s, bh, dest, data, len);
-#else
-	 memcpy(dest, data, len);
-#endif
-	 break;
-      }
-      case BH_COMMAND_WRITE:
-#if 0
-	 intel_cmd_ioctl(s->intel, (void *)data, len);
-#else
-	 if (parse_commands(s, data, len) != 0)
-	    _mesa_printf("parse_commands failed\n");
-#endif
-	 break;
-      default:
-	 break;
-      }
-   }
-
-   s->csr += sizeof(*bh) + len;
-   return 0;
-}
-
-
-#define AUB_FILE_HEADER 0xe085000b
-#define AUB_BLOCK_HEADER 0xe0c10003
-#define AUB_DUMP_BMP 0xe09e0004
-
-int brw_playback_aubfile(struct brw_context *brw,
-			 const char *filename)
-{
-   struct intel_context *intel = &brw->intel;
-   struct aub_state state;
-   struct stat sb;
-   int fd;
-   int retval = 0;
-
-   state.intel = intel;
-
-   fd = open(filename, O_RDONLY, 0);
-   if (fd < 0) {
-      _mesa_printf("couldn't open aubfile: %s\n", filename);
-      return 1;
-   }
-
-   if (fstat(fd, &sb) != 0) {
-      _mesa_printf("couldn't open %s\n", filename);
-      return 1;
-   }
-
-   state.csr = 0;
-   state.sz = sb.st_size;
-   state.map = mmap(NULL, sb.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
-   
-   if (state.map == NULL) {
-      _mesa_printf("couldn't mmap %s\n", filename);
-      return 1;
-   }
-
-   LOCK_HARDWARE(intel); 
-   {
-      /* Make sure we don't confuse anything that might happen to be
-       * going on with the hardware:
-       */
-/*       bmEvictAll(intel); */
-/*       intel->vtbl.lost_hardware(intel); */
-      
-
-      /* Replay the aubfile item by item: 
-       */
-      while (retval == 0 && 
-	     state.csr != state.sz) {
-	 unsigned int insn = *(unsigned int *)(state.map + state.csr);
-
-	 switch (insn) {
-	 case AUB_FILE_HEADER:
-	    retval = gobble(&state, sizeof(struct aub_file_header));
-	    break;
-	 
-	 case AUB_BLOCK_HEADER:   
-	    retval = parse_block_header(&state);
-	    break;
-	 
-	 case AUB_DUMP_BMP:
-	    retval = gobble(&state, sizeof(struct aub_dump_bmp));
-	    break;
-	 
-	 default:
-	    _mesa_printf("unknown instruction %x\n", insn);
-	    retval = 1;
-	    break;
-	 }
-      }
-   }
-   UNLOCK_HARDWARE(intel);
-   return retval;
-}
-
-
-
-
-
-
-		  
diff --git a/src/mesa/drivers/dri/i965/brw_cc.c b/src/mesa/drivers/dri/i965/brw_cc.c
index 8a1d1527db3..fa8121e02d3 100644
--- a/src/mesa/drivers/dri/i965/brw_cc.c
+++ b/src/mesa/drivers/dri/i965/brw_cc.c
@@ -34,10 +34,10 @@
 #include "brw_state.h"
 #include "brw_defines.h"
 #include "brw_util.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
-static void upload_cc_vp( struct brw_context *brw )
+static void prepare_cc_vp( struct brw_context *brw )
 {
    struct brw_cc_viewport ccv;
 
@@ -46,7 +46,8 @@ static void upload_cc_vp( struct brw_context *brw )
    ccv.min_depth = 0.0;
    ccv.max_depth = 1.0;
 
-   brw->cc.vp_gs_offset = brw_cache_data( &brw->cache[BRW_CC_VP], &ccv );
+   dri_bo_unreference(brw->cc.vp_bo);
+   brw->cc.vp_bo = brw_cache_data( &brw->cache, BRW_CC_VP, &ccv, NULL, 0 );
 }
 
 const struct brw_tracked_state brw_cc_vp = {
@@ -55,57 +56,148 @@ const struct brw_tracked_state brw_cc_vp = {
       .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
-   .update = upload_cc_vp
+   .prepare = prepare_cc_vp
 };
 
+struct brw_cc_unit_key {
+   GLboolean stencil, stencil_two_side, color_blend, alpha_enabled;
 
-static void upload_cc_unit( struct brw_context *brw )
+   GLenum stencil_func[2], stencil_fail_op[2];
+   GLenum stencil_pass_depth_fail_op[2], stencil_pass_depth_pass_op[2];
+   GLubyte stencil_ref[2], stencil_write_mask[2], stencil_test_mask[2];
+   GLenum logic_op;
+
+   GLenum blend_eq_rgb, blend_eq_a;
+   GLenum blend_src_rgb, blend_src_a;
+   GLenum blend_dst_rgb, blend_dst_a;
+
+   GLenum alpha_func;
+   GLclampf alpha_ref;
+
+   GLboolean dither;
+
+   GLboolean depth_test, depth_write;
+   GLenum depth_func;
+};
+
+static void
+cc_unit_populate_key(struct brw_context *brw, struct brw_cc_unit_key *key)
+{
+   struct gl_stencil_attrib *stencil = brw->attribs.Stencil;
+
+   memset(key, 0, sizeof(*key));
+
+   key->stencil = stencil->Enabled;
+   key->stencil_two_side = stencil->_TestTwoSide;
+
+   if (key->stencil) {
+      key->stencil_func[0] = stencil->Function[0];
+      key->stencil_fail_op[0] = stencil->FailFunc[0];
+      key->stencil_pass_depth_fail_op[0] = stencil->ZFailFunc[0];
+      key->stencil_pass_depth_pass_op[0] = stencil->ZPassFunc[0];
+      key->stencil_ref[0] = stencil->Ref[0];
+      key->stencil_write_mask[0] = stencil->WriteMask[0];
+      key->stencil_test_mask[0] = stencil->ValueMask[0];
+   }
+   if (key->stencil_two_side) {
+      key->stencil_func[1] = stencil->Function[1];
+      key->stencil_fail_op[1] = stencil->FailFunc[1];
+      key->stencil_pass_depth_fail_op[1] = stencil->ZFailFunc[1];
+      key->stencil_pass_depth_pass_op[1] = stencil->ZPassFunc[1];
+      key->stencil_ref[1] = stencil->Ref[1];
+      key->stencil_write_mask[1] = stencil->WriteMask[1];
+      key->stencil_test_mask[1] = stencil->ValueMask[1];
+   }
+
+   if (brw->attribs.Color->_LogicOpEnabled)
+      key->logic_op = brw->attribs.Color->LogicOp;
+   else
+      key->logic_op = GL_COPY;
+
+   key->color_blend = brw->attribs.Color->BlendEnabled;
+   if (key->color_blend) {
+      key->blend_eq_rgb = brw->attribs.Color->BlendEquationRGB;
+      key->blend_eq_a = brw->attribs.Color->BlendEquationA;
+      key->blend_src_rgb = brw->attribs.Color->BlendSrcRGB;
+      key->blend_dst_rgb = brw->attribs.Color->BlendDstRGB;
+      key->blend_src_a = brw->attribs.Color->BlendSrcA;
+      key->blend_dst_a = brw->attribs.Color->BlendDstA;
+   }
+
+   key->alpha_enabled = brw->attribs.Color->AlphaEnabled;
+   if (key->alpha_enabled) {
+      key->alpha_func = brw->attribs.Color->AlphaFunc;
+      key->alpha_ref = brw->attribs.Color->AlphaRef;
+   }
+
+   key->dither = brw->attribs.Color->DitherFlag;
+
+   key->depth_test = brw->attribs.Depth->Test;
+   if (key->depth_test) {
+      key->depth_func = brw->attribs.Depth->Func;
+      key->depth_write = brw->attribs.Depth->Mask;
+   }
+}
+
+/**
+ * Creates the state cache entry for the given CC unit key.
+ */
+static dri_bo *
+cc_unit_create_from_key(struct brw_context *brw, struct brw_cc_unit_key *key)
 {
    struct brw_cc_unit_state cc;
-   
+   dri_bo *bo;
+
    memset(&cc, 0, sizeof(cc));
 
    /* _NEW_STENCIL */
-   if (brw->attribs.Stencil->Enabled) {
-      cc.cc0.stencil_enable = brw->attribs.Stencil->Enabled;
-      cc.cc0.stencil_func = intel_translate_compare_func(brw->attribs.Stencil->Function[0]);
-      cc.cc0.stencil_fail_op = intel_translate_stencil_op(brw->attribs.Stencil->FailFunc[0]);
-      cc.cc0.stencil_pass_depth_fail_op = intel_translate_stencil_op(brw->attribs.Stencil->ZFailFunc[0]);
-      cc.cc0.stencil_pass_depth_pass_op = intel_translate_stencil_op(brw->attribs.Stencil->ZPassFunc[0]);
-      cc.cc1.stencil_ref = brw->attribs.Stencil->Ref[0];
-      cc.cc1.stencil_write_mask = brw->attribs.Stencil->WriteMask[0];
-      cc.cc1.stencil_test_mask = brw->attribs.Stencil->ValueMask[0];
-
-      if (brw->attribs.Stencil->TestTwoSide) {
-	 cc.cc0.bf_stencil_enable = brw->attribs.Stencil->TestTwoSide;
-	 cc.cc0.bf_stencil_func = intel_translate_compare_func(brw->attribs.Stencil->Function[1]);
-	 cc.cc0.bf_stencil_fail_op = intel_translate_stencil_op(brw->attribs.Stencil->FailFunc[1]);
-	 cc.cc0.bf_stencil_pass_depth_fail_op = intel_translate_stencil_op(brw->attribs.Stencil->ZFailFunc[1]);
-	 cc.cc0.bf_stencil_pass_depth_pass_op = intel_translate_stencil_op(brw->attribs.Stencil->ZPassFunc[1]);
-	 cc.cc1.bf_stencil_ref = brw->attribs.Stencil->Ref[1];
-	 cc.cc2.bf_stencil_write_mask = brw->attribs.Stencil->WriteMask[1];
-	 cc.cc2.bf_stencil_test_mask = brw->attribs.Stencil->ValueMask[1];
+   if (key->stencil) {
+      cc.cc0.stencil_enable = 1;
+      cc.cc0.stencil_func =
+	 intel_translate_compare_func(key->stencil_func[0]);
+      cc.cc0.stencil_fail_op =
+	 intel_translate_stencil_op(key->stencil_fail_op[0]);
+      cc.cc0.stencil_pass_depth_fail_op =
+	 intel_translate_stencil_op(key->stencil_pass_depth_fail_op[0]);
+      cc.cc0.stencil_pass_depth_pass_op =
+	 intel_translate_stencil_op(key->stencil_pass_depth_pass_op[0]);
+      cc.cc1.stencil_ref = key->stencil_ref[0];
+      cc.cc1.stencil_write_mask = key->stencil_write_mask[0];
+      cc.cc1.stencil_test_mask = key->stencil_test_mask[0];
+
+      if (key->stencil_two_side) {
+	 cc.cc0.bf_stencil_enable = 1;
+	 cc.cc0.bf_stencil_func =
+	    intel_translate_compare_func(key->stencil_func[1]);
+	 cc.cc0.bf_stencil_fail_op =
+	    intel_translate_stencil_op(key->stencil_fail_op[1]);
+	 cc.cc0.bf_stencil_pass_depth_fail_op =
+	    intel_translate_stencil_op(key->stencil_pass_depth_fail_op[1]);
+	 cc.cc0.bf_stencil_pass_depth_pass_op =
+	    intel_translate_stencil_op(key->stencil_pass_depth_pass_op[1]);
+	 cc.cc1.bf_stencil_ref = key->stencil_ref[1];
+	 cc.cc2.bf_stencil_write_mask = key->stencil_write_mask[1];
+	 cc.cc2.bf_stencil_test_mask = key->stencil_test_mask[1];
       }
 
       /* Not really sure about this:
        */
-      if (brw->attribs.Stencil->WriteMask[0] ||
-	  (brw->attribs.Stencil->TestTwoSide && brw->attribs.Stencil->WriteMask[1]))
+      if (key->stencil_write_mask[0] ||
+	  (key->stencil_two_side && key->stencil_write_mask[1]))
 	 cc.cc0.stencil_write_enable = 1;
    }
 
    /* _NEW_COLOR */
-   if (brw->attribs.Color->_LogicOpEnabled) {
+   if (key->logic_op != GL_COPY) {
       cc.cc2.logicop_enable = 1;
-      cc.cc5.logicop_func = intel_translate_logic_op( brw->attribs.Color->LogicOp );
-   }
-   else if (brw->attribs.Color->BlendEnabled) {
-      GLenum eqRGB = brw->attribs.Color->BlendEquationRGB;
-      GLenum eqA = brw->attribs.Color->BlendEquationA;
-      GLenum srcRGB = brw->attribs.Color->BlendSrcRGB;
-      GLenum dstRGB = brw->attribs.Color->BlendDstRGB;
-      GLenum srcA = brw->attribs.Color->BlendSrcA;
-      GLenum dstA = brw->attribs.Color->BlendDstA;
+      cc.cc5.logicop_func = intel_translate_logic_op(key->logic_op);
+   } else if (key->color_blend) {
+      GLenum eqRGB = key->blend_eq_rgb;
+      GLenum eqA = key->blend_eq_a;
+      GLenum srcRGB = key->blend_src_rgb;
+      GLenum dstRGB = key->blend_dst_rgb;
+      GLenum srcA = key->blend_src_a;
+      GLenum dstA = key->blend_dst_a;
 
       if (eqRGB == GL_MIN || eqRGB == GL_MAX) {
 	 srcRGB = dstRGB = GL_ONE;
@@ -115,49 +207,78 @@ static void upload_cc_unit( struct brw_context *brw )
 	 srcA = dstA = GL_ONE;
       }
 
-      cc.cc6.dest_blend_factor = brw_translate_blend_factor(dstRGB); 
-      cc.cc6.src_blend_factor = brw_translate_blend_factor(srcRGB); 
-      cc.cc6.blend_function = brw_translate_blend_equation( eqRGB );
+      cc.cc6.dest_blend_factor = brw_translate_blend_factor(dstRGB);
+      cc.cc6.src_blend_factor = brw_translate_blend_factor(srcRGB);
+      cc.cc6.blend_function = brw_translate_blend_equation(eqRGB);
 
-      cc.cc5.ia_dest_blend_factor = brw_translate_blend_factor(dstA); 
-      cc.cc5.ia_src_blend_factor = brw_translate_blend_factor(srcA); 
-      cc.cc5.ia_blend_function = brw_translate_blend_equation( eqA );
+      cc.cc5.ia_dest_blend_factor = brw_translate_blend_factor(dstA);
+      cc.cc5.ia_src_blend_factor = brw_translate_blend_factor(srcA);
+      cc.cc5.ia_blend_function = brw_translate_blend_equation(eqA);
 
       cc.cc3.blend_enable = 1;
-      cc.cc3.ia_blend_enable = (srcA != srcRGB || 
-				dstA != dstRGB || 
+      cc.cc3.ia_blend_enable = (srcA != srcRGB ||
+				dstA != dstRGB ||
 				eqA != eqRGB);
    }
 
-   if (brw->attribs.Color->AlphaEnabled) {
+   if (key->alpha_enabled) {
       cc.cc3.alpha_test = 1;
-      cc.cc3.alpha_test_func = intel_translate_compare_func(brw->attribs.Color->AlphaFunc);
-
-      UNCLAMPED_FLOAT_TO_UBYTE(cc.cc7.alpha_ref.ub[0], brw->attribs.Color->AlphaRef);
-
+      cc.cc3.alpha_test_func = intel_translate_compare_func(key->alpha_func);
       cc.cc3.alpha_test_format = BRW_ALPHATEST_FORMAT_UNORM8;
+
+      UNCLAMPED_FLOAT_TO_UBYTE(cc.cc7.alpha_ref.ub[0], key->alpha_ref);
    }
 
-   if (brw->attribs.Color->DitherFlag) {
+   if (key->dither) {
       cc.cc5.dither_enable = 1;
-      cc.cc6.y_dither_offset = 0; 
-      cc.cc6.x_dither_offset = 0;     
+      cc.cc6.y_dither_offset = 0;
+      cc.cc6.x_dither_offset = 0;
    }
 
    /* _NEW_DEPTH */
-   if (brw->attribs.Depth->Test) {
-      cc.cc2.depth_test = brw->attribs.Depth->Test;
-      cc.cc2.depth_test_function = intel_translate_compare_func(brw->attribs.Depth->Func);
-      cc.cc2.depth_write_enable = brw->attribs.Depth->Mask;
+   if (key->depth_test) {
+      cc.cc2.depth_test = 1;
+      cc.cc2.depth_test_function = intel_translate_compare_func(key->depth_func);
+      cc.cc2.depth_write_enable = key->depth_write;
    }
- 
+
    /* CACHE_NEW_CC_VP */
-   cc.cc4.cc_viewport_state_offset =  brw->cc.vp_gs_offset >> 5;
- 
+   cc.cc4.cc_viewport_state_offset = brw->cc.vp_bo->offset >> 5; /* reloc */
+
    if (INTEL_DEBUG & DEBUG_STATS)
-      cc.cc5.statistics_enable = 1; 
+      cc.cc5.statistics_enable = 1;
+
+   bo = brw_upload_cache(&brw->cache, BRW_CC_UNIT,
+			 key, sizeof(*key),
+			 &brw->cc.vp_bo, 1,
+			 &cc, sizeof(cc),
+			 NULL, NULL);
+
+   /* Emit CC viewport relocation */
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION,
+		     0,
+		     0,
+		     offsetof(struct brw_cc_unit_state, cc4),
+		     brw->cc.vp_bo);
+
+   return bo;
+}
+
+static void prepare_cc_unit( struct brw_context *brw )
+{
+   struct brw_cc_unit_key key;
+
+   cc_unit_populate_key(brw, &key);
+
+   dri_bo_unreference(brw->cc.state_bo);
+   brw->cc.state_bo = brw_search_cache(&brw->cache, BRW_CC_UNIT,
+				       &key, sizeof(key),
+				       &brw->cc.vp_bo, 1,
+				       NULL);
 
-   brw->cc.state_gs_offset = brw_cache_data( &brw->cache[BRW_CC_UNIT], &cc );
+   if (brw->cc.state_bo == NULL)
+      brw->cc.state_bo = cc_unit_create_from_key(brw, &key);
 }
 
 const struct brw_tracked_state brw_cc_unit = {
@@ -166,7 +287,7 @@ const struct brw_tracked_state brw_cc_unit = {
       .brw = 0,
       .cache = CACHE_NEW_CC_VP
    },
-   .update = upload_cc_unit
+   .prepare = prepare_cc_unit,
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_clip.c b/src/mesa/drivers/dri/i965/brw_clip.c
index 3bec153075a..38d8b704d7e 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.c
+++ b/src/mesa/drivers/dri/i965/brw_clip.c
@@ -29,9 +29,9 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "intel_batchbuffer.h"
 
@@ -60,7 +60,7 @@ static void compile_clip_prog( struct brw_context *brw,
    
    /* Begin the compilation:
     */
-   brw_init_compile(&c.func);
+   brw_init_compile(brw, &c.func);
 
    c.func.single_program_flow = 1;
 
@@ -119,31 +119,19 @@ static void compile_clip_prog( struct brw_context *brw,
 
    /* Upload
     */
-   brw->clip.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_CLIP_PROG],
-						&c.key,
-						sizeof(c.key),
-						program,
-						program_size,
-						&c.prog_data,
-						&brw->clip.prog_data );
+   dri_bo_unreference(brw->clip.prog_bo);
+   brw->clip.prog_bo = brw_upload_cache( &brw->cache,
+					 BRW_CLIP_PROG,
+					 &c.key, sizeof(c.key),
+					 NULL, 0,
+					 program, program_size,
+					 &c.prog_data,
+					 &brw->clip.prog_data );
 }
 
-
-static GLboolean search_cache( struct brw_context *brw, 
-			       struct brw_clip_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_CLIP_PROG], 
-			   key, sizeof(*key),
-			   &brw->clip.prog_data,
-			   &brw->clip.prog_gs_offset);
-}
-
-
-
-
 /* Calculate interpolants for triangle and line rasterization.
  */
-static void upload_clip_prog( struct brw_context *brw )
+static void upload_clip_prog(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct brw_clip_prog_key key;
@@ -180,12 +168,10 @@ static void upload_clip_prog( struct brw_context *brw )
 	       offset_front = 0;
 	       break;
 	    case GL_LINE:
-	       key.do_unfilled = 1;
 	       fill_front = CLIP_LINE;
 	       offset_front = brw->attribs.Polygon->OffsetLine;
 	       break;
 	    case GL_POINT:
-	       key.do_unfilled = 1;
 	       fill_front = CLIP_POINT;
 	       offset_front = brw->attribs.Polygon->OffsetPoint;
 	       break;
@@ -200,22 +186,23 @@ static void upload_clip_prog( struct brw_context *brw )
 	       offset_back = 0;
 	       break;
 	    case GL_LINE:
-	       key.do_unfilled = 1;
 	       fill_back = CLIP_LINE;
 	       offset_back = brw->attribs.Polygon->OffsetLine;
 	       break;
 	    case GL_POINT:
-	       key.do_unfilled = 1;
 	       fill_back = CLIP_POINT;
 	       offset_back = brw->attribs.Polygon->OffsetPoint;
 	       break;
 	    }
 	 }
 
-	 /* Most cases the fixed function units will handle.  Cases where
-	  * one or more polygon faces are unfilled will require help:
-	  */
-	 if (key.do_unfilled) {
+	 if (brw->attribs.Polygon->BackMode != GL_FILL ||
+	     brw->attribs.Polygon->FrontMode != GL_FILL) {
+	    key.do_unfilled = 1;
+
+	    /* Most cases the fixed function units will handle.  Cases where
+	     * one or more polygon faces are unfilled will require help:
+	     */
 	    key.clip_mode = BRW_CLIPMODE_CLIP_NON_REJECTED;
 
 	    if (offset_back || offset_front) {
@@ -248,7 +235,12 @@ static void upload_clip_prog( struct brw_context *brw )
       }
    }
 
-   if (!search_cache(brw, &key))
+   dri_bo_unreference(brw->clip.prog_bo);
+   brw->clip.prog_bo = brw_search_cache(&brw->cache, BRW_CLIP_PROG,
+					&key, sizeof(key),
+					NULL, 0,
+					&brw->clip.prog_data);
+   if (brw->clip.prog_bo == NULL)
       compile_clip_prog( brw, &key );
 }
 
@@ -262,5 +254,5 @@ const struct brw_tracked_state brw_clip_prog = {
       .brw   = (BRW_NEW_REDUCED_PRIMITIVE),
       .cache = CACHE_NEW_VS_PROG
    },
-   .update = upload_clip_prog
+   .prepare = upload_clip_prog
 };
diff --git a/src/mesa/drivers/dri/i965/brw_clip.h b/src/mesa/drivers/dri/i965/brw_clip.h
index 49b2770a514..e06747864b5 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.h
+++ b/src/mesa/drivers/dri/i965/brw_clip.h
@@ -42,7 +42,7 @@
  * up polygon offset and flatshading at this point:
  */
 struct brw_clip_prog_key {
-   GLuint attrs:16;		
+   GLuint attrs:32;		
    GLuint primitive:4;
    GLuint nr_userclip:3;
    GLuint do_flat_shading:1;
@@ -51,7 +51,7 @@ struct brw_clip_prog_key {
    GLuint fill_ccw:2;		/* includes cull information */
    GLuint offset_cw:1;
    GLuint offset_ccw:1;
-   GLuint pad0:1;
+   GLuint pad0:17;
 
    GLuint copy_bfc_cw:1;
    GLuint copy_bfc_ccw:1;
@@ -167,4 +167,9 @@ void brw_clip_copy_colors( struct brw_clip_compile *c,
 
 void brw_clip_init_clipmask( struct brw_clip_compile *c );
 
+struct brw_reg get_tmp( struct brw_clip_compile *c );
+
+void brw_clip_project_position(struct brw_clip_compile *c,
+             struct brw_reg pos );
+
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_clip_line.c b/src/mesa/drivers/dri/i965/brw_clip_line.c
index 83182270eac..c45d48dff8e 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_line.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_line.c
@@ -29,11 +29,11 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
+
 #include "intel_batchbuffer.h"
 
 #include "brw_defines.h"
@@ -130,6 +130,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
    struct brw_instruction *plane_loop;
    struct brw_instruction *plane_active;
    struct brw_instruction *is_negative;
+   struct brw_instruction *is_neg2;
    struct brw_instruction *not_culled;
    struct brw_reg v1_null_ud = retype(vec1(brw_null_reg()), BRW_REGISTER_TYPE_UD);
 
@@ -146,6 +147,16 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
    brw_clip_init_planes(c);
    brw_clip_init_clipmask(c);
 
+   /* -ve rhw workaround */
+   if (!BRW_IS_G4X(p->brw)) {
+      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+      brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2),
+              brw_imm_ud(1<<20));
+      brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(0x3f));
+   }
+
+   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
    plane_loop = brw_DO(p, BRW_EXECUTE_1);
    {
       /* if (planemask & 1)
@@ -183,13 +194,20 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
 	    /* Coming back in.  We know that both cannot be negative
 	     * because the line would have been culled in that case.
 	     */
-	    brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
-	    brw_math_invert(p, c->reg.t, c->reg.t);
-	    brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
 
-	    brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
-	    brw_MOV(p, c->reg.t0, c->reg.t);
-	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	    /* If both are positive, do nothing */
+             brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.dp0, brw_imm_f(0.0));
+             is_neg2 = brw_IF(p, BRW_EXECUTE_1);
+             {
+		brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
+		brw_math_invert(p, c->reg.t, c->reg.t);
+		brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
+
+		brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
+		brw_MOV(p, c->reg.t0, c->reg.t);
+		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	     }
+	     brw_ENDIF(p, is_neg2);
 	 }
 	 brw_ENDIF(p, is_negative);	 
       }
diff --git a/src/mesa/drivers/dri/i965/brw_clip_point.c b/src/mesa/drivers/dri/i965/brw_clip_point.c
index 2346980a562..d17b199b898 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_point.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_point.c
@@ -29,11 +29,11 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
+
 #include "intel_batchbuffer.h"
 
 #include "brw_defines.h"
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index 1e6d6fa1762..9b0d7eab7bf 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -32,55 +32,132 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "macros.h"
+#include "main/macros.h"
 
+struct brw_clip_unit_key {
+   unsigned int total_grf;
+   unsigned int urb_entry_read_length;
+   unsigned int curb_entry_read_length;
+   unsigned int clip_mode;
 
+   unsigned int curbe_offset;
 
-static void upload_clip_unit( struct brw_context *brw )
-{
-   struct brw_clip_unit_state clip;
+   unsigned int nr_urb_entries, urb_size;
+};
 
-   memset(&clip, 0, sizeof(clip));
+static void
+clip_unit_populate_key(struct brw_context *brw, struct brw_clip_unit_key *key)
+{
+   memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_CLIP_PROG */
-   clip.thread0.grf_reg_count = ((brw->clip.prog_data->total_grf-1) & ~15) / 16;
-   clip.thread0.kernel_start_pointer = brw->clip.prog_gs_offset >> 6;
-   clip.thread3.urb_entry_read_length = brw->clip.prog_data->urb_read_length;
-   clip.thread3.const_urb_entry_read_length = brw->clip.prog_data->curb_read_length;
-   clip.clip5.clip_mode = brw->clip.prog_data->clip_mode;
+   key->total_grf = brw->clip.prog_data->total_grf;
+   key->urb_entry_read_length = brw->clip.prog_data->urb_read_length;
+   key->curb_entry_read_length = brw->clip.prog_data->curb_read_length;
+   key->clip_mode = brw->clip.prog_data->clip_mode;
 
    /* BRW_NEW_CURBE_OFFSETS */
-   clip.thread3.const_urb_entry_read_offset = brw->curbe.clip_start * 2;
+   key->curbe_offset = brw->curbe.clip_start;
 
    /* BRW_NEW_URB_FENCE */
-   clip.thread4.nr_urb_entries = brw->urb.nr_clip_entries; 
-   clip.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-   clip.thread4.max_threads = 0; /* Hmm, maybe the max is 1 or 2 threads */
+   key->nr_urb_entries = brw->urb.nr_clip_entries;
+   key->urb_size = brw->urb.vsize;
+}
 
-   if (INTEL_DEBUG & DEBUG_STATS)
-      clip.thread4.stats_enable = 1; 
+static dri_bo *
+clip_unit_create_from_key(struct brw_context *brw,
+			  struct brw_clip_unit_key *key)
+{
+   struct brw_clip_unit_state clip;
+   dri_bo *bo;
+
+   memset(&clip, 0, sizeof(clip));
+
+   clip.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
+   /* reloc */
+   clip.thread0.kernel_start_pointer = brw->clip.prog_bo->offset >> 6;
 
-   /* CONSTANT */
    clip.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
    clip.thread1.single_program_flow = 1;
+
+   clip.thread3.urb_entry_read_length = key->urb_entry_read_length;
+   clip.thread3.const_urb_entry_read_length = key->curb_entry_read_length;
+   clip.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;
    clip.thread3.dispatch_grf_start_reg = 1;
    clip.thread3.urb_entry_read_offset = 0;
+
+   clip.thread4.nr_urb_entries = key->nr_urb_entries;
+   clip.thread4.urb_entry_allocation_size = key->urb_size - 1;
+   /* If we have enough clip URB entries to run two threads, do so.
+    */
+   if (key->nr_urb_entries >= 10) {
+      /* Half of the URB entries go to each thread, and it has to be an
+       * even number.
+       */
+      assert(key->nr_urb_entries % 2 == 0);
+      clip.thread4.max_threads = 2 - 1;
+   } else {
+      assert(key->nr_urb_entries >= 5);
+      clip.thread4.max_threads = 1 - 1;
+   }
+
+   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
+      clip.thread4.max_threads = 0;
+
+   if (INTEL_DEBUG & DEBUG_STATS)
+      clip.thread4.stats_enable = 1;
+
    clip.clip5.userclip_enable_flags = 0x7f;
    clip.clip5.userclip_must_clip = 1;
    clip.clip5.guard_band_enable = 0;
    clip.clip5.viewport_z_clip_enable = 1;
    clip.clip5.viewport_xy_clip_enable = 1;
    clip.clip5.vertex_position_space = BRW_CLIP_NDCSPACE;
-   clip.clip5.api_mode = BRW_CLIP_API_OGL;   
+   clip.clip5.api_mode = BRW_CLIP_API_OGL;
+   clip.clip5.clip_mode = key->clip_mode;
+
+   if (BRW_IS_G4X(brw))
+      clip.clip5.negative_w_clip_test = 1;
+
    clip.clip6.clipper_viewport_state_ptr = 0;
    clip.viewport_xmin = -1;
    clip.viewport_xmax = 1;
    clip.viewport_ymin = -1;
    clip.viewport_ymax = 1;
 
-   brw->clip.state_gs_offset = brw_cache_data( &brw->cache[BRW_CLIP_UNIT], &clip );
+   bo = brw_upload_cache(&brw->cache, BRW_CLIP_UNIT,
+			 key, sizeof(*key),
+			 &brw->clip.prog_bo, 1,
+			 &clip, sizeof(clip),
+			 NULL, NULL);
+
+   /* Emit clip program relocation */
+   assert(brw->clip.prog_bo);
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION,
+		     0,
+		     clip.thread0.grf_reg_count << 1,
+		     offsetof(struct brw_clip_unit_state, thread0),
+		     brw->clip.prog_bo);
+
+   return bo;
 }
 
+static void upload_clip_unit( struct brw_context *brw )
+{
+   struct brw_clip_unit_key key;
+
+   clip_unit_populate_key(brw, &key);
+
+   dri_bo_unreference(brw->clip.state_bo);
+   brw->clip.state_bo = brw_search_cache(&brw->cache, BRW_CLIP_UNIT,
+					 &key, sizeof(key),
+					 &brw->clip.prog_bo, 1,
+					 NULL);
+   if (brw->clip.state_bo == NULL) {
+      brw->clip.state_bo = clip_unit_create_from_key(brw, &key);
+   }
+}
 
 const struct brw_tracked_state brw_clip_unit = {
    .dirty = {
@@ -89,5 +166,5 @@ const struct brw_tracked_state brw_clip_unit = {
 		BRW_NEW_URB_FENCE),
       .cache = CACHE_NEW_CLIP_PROG
    },
-   .update = upload_clip_unit
+   .prepare = upload_clip_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c b/src/mesa/drivers/dri/i965/brw_clip_tri.c
index f62b02cedfd..1dbba37fe7e 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_tri.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c
@@ -29,11 +29,11 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
+
 #include "intel_batchbuffer.h"
 
 #include "brw_defines.h"
@@ -42,6 +42,10 @@
 #include "brw_util.h"
 #include "brw_clip.h"
 
+static void release_tmps( struct brw_clip_compile *c )
+{
+   c->last_tmp = c->first_tmp;
+}
 
 
 void brw_clip_tri_alloc_regs( struct brw_clip_compile *c, 
@@ -78,7 +82,7 @@ void brw_clip_tri_alloc_regs( struct brw_clip_compile *c,
    }
 
    c->reg.t          = brw_vec1_grf(i, 0);
-   c->reg.loopcount  = retype(brw_vec1_grf(i, 1), BRW_REGISTER_TYPE_UD);
+   c->reg.loopcount  = retype(brw_vec1_grf(i, 1), BRW_REGISTER_TYPE_D);
    c->reg.nr_verts   = retype(brw_vec1_grf(i, 2), BRW_REGISTER_TYPE_UD);
    c->reg.planemask  = retype(brw_vec1_grf(i, 3), BRW_REGISTER_TYPE_UD);
    c->reg.plane_equation = brw_vec4_grf(i, 4);
@@ -435,15 +439,103 @@ static void maybe_do_clip_tri( struct brw_clip_compile *c )
    brw_ENDIF(p, do_clip);
 }
 
-
+static void brw_clip_test( struct brw_clip_compile *c )
+{
+    struct brw_reg t = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t2 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t3 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+
+    struct brw_reg v0 = get_tmp(c);
+    struct brw_reg v1 = get_tmp(c);
+    struct brw_reg v2 = get_tmp(c);
+
+    struct brw_indirect vt0 = brw_indirect(0, 0);
+    struct brw_indirect vt1 = brw_indirect(1, 0);
+    struct brw_indirect vt2 = brw_indirect(2, 0);
+
+    struct brw_compile *p = &c->func;
+
+    brw_MOV(p, get_addr_reg(vt0), brw_address(c->reg.vertex[0]));
+    brw_MOV(p, get_addr_reg(vt1), brw_address(c->reg.vertex[1]));
+    brw_MOV(p, get_addr_reg(vt2), brw_address(c->reg.vertex[2]));
+    brw_MOV(p, v0, deref_4f(vt0, c->offset[VERT_RESULT_HPOS]));
+    brw_MOV(p, v1, deref_4f(vt1, c->offset[VERT_RESULT_HPOS]));
+    brw_MOV(p, v2, deref_4f(vt2, c->offset[VERT_RESULT_HPOS]));
+
+    /* test nearz, xmin, ymin plane */
+    brw_CMP(p, t1, BRW_CONDITIONAL_LE, negate(v0), get_element(v0, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t2, BRW_CONDITIONAL_LE, negate(v1), get_element(v1, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t3, BRW_CONDITIONAL_LE, negate(v2), get_element(v2, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_XOR(p, t, t1, t2);
+    brw_XOR(p, t1, t2, t3);
+    brw_OR(p, t, t, t1);
+
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 0), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<5)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 1), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<3)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 2), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<1)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    /* test farz, xmax, ymax plane */
+    brw_CMP(p, t1, BRW_CONDITIONAL_L, v0, get_element(v0, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t2, BRW_CONDITIONAL_L, v1, get_element(v1, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t3, BRW_CONDITIONAL_L, v2, get_element(v2, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    brw_XOR(p, t, t1, t2);
+    brw_XOR(p, t1, t2, t3);
+    brw_OR(p, t, t, t1);
+
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 0), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<4)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 1), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<2)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 2), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<0)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    release_tmps(c);
+}
 
 
 void brw_emit_tri_clip( struct brw_clip_compile *c )
 {
+   struct brw_instruction *neg_rhw;
+   struct brw_compile *p = &c->func;
    brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6);
    brw_clip_tri_init_vertices(c);
    brw_clip_init_clipmask(c);
 
+   /* if -ve rhw workaround bit is set, 
+      do cliptest */
+   if (!BRW_IS_G4X(p->brw)) {
+      brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+      brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2), 
+              brw_imm_ud(1<<20));
+      neg_rhw = brw_IF(p, BRW_EXECUTE_1); 
+      {
+         brw_clip_test(c);
+      }
+      brw_ENDIF(p, neg_rhw);
+   }
    /* Can't push into do_clip_tri because with polygon (or quad)
     * flatshading, need to apply the flatshade here because we don't
     * respect the PV when converting to trifan for emit:
@@ -462,6 +554,3 @@ void brw_emit_tri_clip( struct brw_clip_compile *c )
     */
    brw_clip_kill_thread(c);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_clip_unfilled.c b/src/mesa/drivers/dri/i965/brw_clip_unfilled.c
index 918e0001870..d7ca517927b 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_unfilled.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_unfilled.c
@@ -29,11 +29,11 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
+
 #include "intel_batchbuffer.h"
 
 #include "brw_defines.h"
@@ -58,10 +58,30 @@ static void compute_tri_direction( struct brw_clip_compile *c )
    struct brw_reg v2 = byte_offset(c->reg.vertex[2], c->offset[VERT_RESULT_HPOS]); 
 
 
+   struct brw_reg v0n = get_tmp(c);
+   struct brw_reg v1n = get_tmp(c);
+   struct brw_reg v2n = get_tmp(c);
+
+   /* Convert to NDC.
+    * NOTE: We can't modify the original vertex coordinates,
+    * as it may impact further operations.
+    * So, we have to keep normalized coordinates in temp registers.
+    *
+    * TBD-KC
+    * Try to optimize unnecessary MOV's.
+    */
+   brw_MOV(p, v0n, v0);
+   brw_MOV(p, v1n, v1);
+   brw_MOV(p, v2n, v2);
+
+   brw_clip_project_position(c, v0n);
+   brw_clip_project_position(c, v1n);
+   brw_clip_project_position(c, v2n);
+
    /* Calculate the vectors of two edges of the triangle:
     */
-   brw_ADD(p, e, v0, negate(v2)); 
-   brw_ADD(p, f, v1, negate(v2)); 
+   brw_ADD(p, e, v0n, negate(v2n)); 
+   brw_ADD(p, f, v1n, negate(v2n)); 
 
    /* Take their crossproduct:
     */
@@ -220,8 +240,8 @@ static void apply_one_offset( struct brw_clip_compile *c,
 			  struct brw_indirect vert )
 {
    struct brw_compile *p = &c->func;
-   struct brw_reg pos = deref_4f(vert, c->offset[VERT_RESULT_HPOS]);
-   struct brw_reg z = get_element(pos, 2);
+   struct brw_reg z = deref_1f(vert, c->header_position_offset +
+			       2 * type_sz(BRW_REGISTER_TYPE_F));
 
    brw_ADD(p, z, z, vec1(c->reg.offset));
 }
diff --git a/src/mesa/drivers/dri/i965/brw_clip_util.c b/src/mesa/drivers/dri/i965/brw_clip_util.c
index 19bef19801a..9d3b0be694a 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_util.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_util.c
@@ -30,11 +30,11 @@
   */
 
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
+
 #include "intel_batchbuffer.h"
 
 #include "brw_defines.h"
@@ -46,8 +46,7 @@
 
 
 
-
-static struct brw_reg get_tmp( struct brw_clip_compile *c )
+struct brw_reg get_tmp( struct brw_clip_compile *c )
 {
    struct brw_reg tmp = brw_vec4_grf(c->last_tmp, 0);
 
@@ -90,7 +89,7 @@ void brw_clip_init_planes( struct brw_clip_compile *c )
 
 /* Project 'pos' to screen space (or back again), overwrite with results:
  */
-static void brw_clip_project_position(struct brw_clip_compile *c, struct brw_reg pos )
+void brw_clip_project_position(struct brw_clip_compile *c, struct brw_reg pos )
 {
    struct brw_compile *p = &c->func;
 
@@ -262,7 +261,7 @@ void brw_clip_kill_thread(struct brw_clip_compile *c)
 		 c->reg.R0,
 		 0,		/* allocate */
 		 0,		/* used */
-		 0, 		/* msg len */
+		 1, 		/* msg len */
 		 0, 		/* response len */
 		 1, 		/* eot */
 		 1,		/* writes complete */
@@ -272,6 +271,7 @@ void brw_clip_kill_thread(struct brw_clip_compile *c)
 
 
 
+
 struct brw_reg brw_clip_plane0_address( struct brw_clip_compile *c )
 {
    return brw_address(c->reg.fixed_planes);
@@ -327,8 +327,7 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c )
    
    /* Shift so that lowest outcode bit is rightmost: 
     */
-   brw_MOV(p, c->reg.planemask, incoming);
-   brw_SHR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(26));
+   brw_SHR(p, c->reg.planemask, incoming, brw_imm_ud(26));
 
    if (c->key.nr_userclip) {
       struct brw_reg tmp = retype(vec1(get_tmp(c)), BRW_REGISTER_TYPE_UD);
@@ -342,13 +341,5 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c )
       
       release_tmp(c, tmp);
    }
-
-   /* Test for -ve rhw workaround 
-    */
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-   brw_AND(p, vec1(brw_null_reg()), incoming, brw_imm_ud(1<<20));
-   brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(0x3f));
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index 397a9bd3f5c..1d6ac2cea68 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -30,42 +30,49 @@
   */
 
 
+#include "main/imports.h"
+#include "main/api_noop.h"
+#include "main/vtxfmt.h"
+#include "main/simple_list.h"
+#include "shader/shader_api.h"
+
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_defines.h"
 #include "brw_draw.h"
+#include "brw_state.h"
 #include "brw_vs.h"
-#include "imports.h"
 #include "intel_tex.h"
 #include "intel_blit.h"
 #include "intel_batchbuffer.h"
+#include "intel_pixel.h"
+#include "intel_span.h"
+#include "tnl/t_pipeline.h"
 
 #include "utils.h"
-#include "api_noop.h"
-#include "vtxfmt.h"
+
 
 /***************************************
  * Mesa's Driver Functions
  ***************************************/
 
-static const struct dri_extension brw_extensions[] =
+static void brwUseProgram(GLcontext *ctx, GLuint program)
 {
-    { "GL_ARB_depth_texture",              NULL },
-    { "GL_ARB_fragment_program",           NULL },
-    { "GL_ARB_shadow",                     NULL },
-    { "GL_EXT_shadow_funcs",               NULL },
-    /* ARB extn won't work if not enabled */
-    { "GL_SGIX_depth_texture",             NULL },
-    { "GL_ARB_texture_env_crossbar",       NULL },
-    { NULL,                                NULL }
-};
-
+   _mesa_use_program(ctx, program);
+}
 
+static void brwInitProgFuncs( struct dd_function_table *functions )
+{
+   functions->UseProgram = brwUseProgram;
+}
 static void brwInitDriverFunctions( struct dd_function_table *functions )
 {
    intelInitDriverFunctions( functions );
-   brwInitTextureFuncs( functions );
+
    brwInitFragProgFuncs( functions );
+   brwInitProgFuncs( functions );
+   brw_init_queryobj_functions(functions);
+
+   functions->Viewport = intel_viewport;
 }
 
 
@@ -116,10 +123,15 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
       return GL_FALSE;
    }
 
+   /* Initialize swrast, tnl driver tables: */
+   intelInitSpanFuncs(ctx);
+
+   TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
+
    ctx->Const.MaxTextureUnits = BRW_MAX_TEX_UNIT;
    ctx->Const.MaxTextureImageUnits = BRW_MAX_TEX_UNIT;
    ctx->Const.MaxTextureCoordUnits = BRW_MAX_TEX_UNIT;
-
+   ctx->Const.MaxVertexTextureImageUnits = 0; /* no vertex shader textures */
 
    /* Advertise the full hardware capabilities.  The new memory
     * manager should cope much better with overload situations:
@@ -128,15 +140,9 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
    ctx->Const.Max3DTextureLevels = 9;
    ctx->Const.MaxCubeTextureLevels = 12;
    ctx->Const.MaxTextureRectSize = (1<<11);
-   ctx->Const.MaxTextureUnits = BRW_MAX_TEX_UNIT;
    
 /*    ctx->Const.MaxNativeVertexProgramTemps = 32; */
 
-
-   driInitExtensions( ctx, brw_extensions, GL_FALSE );
-
-   brw_aub_init( brw );
-
    brw_init_attribs( brw );
    brw_init_metaops( brw );
    brw_init_state( brw );
@@ -144,25 +150,14 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
    brw->state.dirty.mesa = ~0;
    brw->state.dirty.brw = ~0;
 
-   memset(&brw->wm.bind, ~0, sizeof(brw->wm.bind));
-
    brw->emit_state_always = 0;
 
-   ctx->FragmentProgram._MaintainTexEnvProgram = 1;
+   ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
+   ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
 
-   brw_draw_init( brw );
-
-   brw_ProgramCacheInit( ctx );
-
-   brw_FrameBufferTexInit( brw );
+   make_empty_list(&brw->query.active_head);
 
-   {
-      const char *filename = getenv("INTEL_REPLAY");
-      if (filename) {
-	 brw_playback_aubfile(brw, filename);
-	 exit(0);
-      }
-   }
+   brw_draw_init( brw );
 
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 08fdc545205..77980109cdf 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -35,7 +35,7 @@
 
 #include "intel_context.h"
 #include "brw_structs.h"
-#include "imports.h"
+#include "main/imports.h"
 
 
 /* Glossary:
@@ -130,24 +130,34 @@ struct brw_context;
 #define BRW_NEW_CONTEXT                 0x80
 #define BRW_NEW_WM_INPUT_DIMENSIONS     0x100
 #define BRW_NEW_INPUT_VARYING           0x200
-#define BRW_NEW_TNL_PROGRAM             0x400
 #define BRW_NEW_PSP                     0x800
 #define BRW_NEW_METAOPS                 0x1000
 #define BRW_NEW_FENCE                   0x2000
-#define BRW_NEW_LOCK                    0x4000
-
-
+#define BRW_NEW_INDICES			0x4000
+#define BRW_NEW_VERTICES		0x8000
+/**
+ * Used for any batch entry with a relocated pointer that will be used
+ * by any 3D rendering.
+ */
+#define BRW_NEW_BATCH			0x10000
+/** brw->depth_region updated */
+#define BRW_NEW_DEPTH_BUFFER		0x20000
+#define BRW_NEW_NR_SURFACES		0x40000
 
 struct brw_state_flags {
+   /** State update flags signalled by mesa internals */
    GLuint mesa;
-   GLuint cache;
+   /**
+    * State update flags signalled as the result of brw_tracked_state updates
+    */
    GLuint brw;
+   /** State update flags signalled by brw_state_cache.c searches */
+   GLuint cache;
 };
 
 struct brw_vertex_program {
    struct gl_vertex_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
@@ -155,7 +165,6 @@ struct brw_vertex_program {
 struct brw_fragment_program {
    struct gl_fragment_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
@@ -230,32 +239,46 @@ struct brw_vs_ouput_sizes {
 
 
 #define BRW_MAX_TEX_UNIT 8
-#define BRW_WM_MAX_SURF BRW_MAX_TEX_UNIT + 1
+#define BRW_WM_MAX_SURF BRW_MAX_TEX_UNIT + MAX_DRAW_BUFFERS
 
-/* Create a fixed sized struct for caching binding tables:
- */
-struct brw_surface_binding_table {
-   GLuint surf_ss_offset[BRW_WM_MAX_SURF];
-};
-
-
-struct brw_cache;
-
-struct brw_mem_pool {
-   struct buffer *buffer;
-
-   GLuint size;
-   GLuint offset;		/* offset of first free byte */
+enum brw_cache_id {
+   BRW_CC_VP,
+   BRW_CC_UNIT,
+   BRW_WM_PROG,
+   BRW_SAMPLER_DEFAULT_COLOR,
+   BRW_SAMPLER,
+   BRW_WM_UNIT,
+   BRW_SF_PROG,
+   BRW_SF_VP,
+   BRW_SF_UNIT,
+   BRW_VS_UNIT,
+   BRW_VS_PROG,
+   BRW_GS_UNIT,
+   BRW_GS_PROG,
+   BRW_CLIP_VP,
+   BRW_CLIP_UNIT,
+   BRW_CLIP_PROG,
+   BRW_SS_SURFACE,
+   BRW_SS_SURF_BIND,
 
-   struct brw_context *brw;
+   BRW_MAX_CACHE
 };
 
 struct brw_cache_item {
+   /**
+    * Effectively part of the key, cache_id identifies what kind of state
+    * buffer is involved, and also which brw->state.dirty.cache flag should
+    * be set when this cache item is chosen.
+    */
+   enum brw_cache_id cache_id;
+   /** 32-bit hash of the key data */
    GLuint hash;
    GLuint key_size;		/* for variable-sized keys */
    const void *key;
+   dri_bo **reloc_bufs;
+   GLuint nr_reloc_bufs;
 
-   GLuint offset;		/* offset within pool's buffer */
+   dri_bo *bo;
    GLuint data_size;
 
    struct brw_cache_item *next;
@@ -264,23 +287,19 @@ struct brw_cache_item {
 
 
 struct brw_cache {
-   GLuint id;
-
-   const char *name;
-
    struct brw_context *brw;
-   struct brw_mem_pool *pool;
 
    struct brw_cache_item **items;
    GLuint size, n_items;
-   
-   GLuint key_size;		/* for fixed-size keys */
-   GLuint aux_size;
 
-   GLuint aub_type;
-   GLuint aub_sub_type;
-   
-   GLuint last_addr;			/* offset of active item */
+   GLuint key_size[BRW_MAX_CACHE];		/* for fixed-size keys */
+   GLuint aux_size[BRW_MAX_CACHE];
+   char *name[BRW_MAX_CACHE];
+
+   /* Record of the last BOs chosen for each cache_id.  Used to set
+    * brw->state.dirty.cache when a new cache item is chosen.
+    */
+   dri_bo *last_bo[BRW_MAX_CACHE];
 };
 
 
@@ -312,34 +331,8 @@ struct brw_state_pointers {
  */
 struct brw_tracked_state {
    struct brw_state_flags dirty;
-   void (*update)( struct brw_context *brw );
-};
-
-
-enum brw_cache_id {
-   BRW_CC_VP,
-   BRW_CC_UNIT,
-   BRW_WM_PROG,
-   BRW_SAMPLER_DEFAULT_COLOR,
-   BRW_SAMPLER,
-   BRW_WM_UNIT,
-   BRW_SF_PROG,
-   BRW_SF_VP,
-   BRW_SF_UNIT,
-   BRW_VS_UNIT,
-   BRW_VS_PROG,
-   BRW_GS_UNIT,
-   BRW_GS_PROG,
-   BRW_CLIP_VP,
-   BRW_CLIP_UNIT,
-   BRW_CLIP_PROG,
-
-   /* These two are in the SS pool:
-    */
-   BRW_SS_SURFACE,
-   BRW_SS_SURF_BIND,
-
-   BRW_MAX_CACHE
+   void (*prepare)( struct brw_context *brw );
+   void (*emit)( struct brw_context *brw );
 };
 
 /* Flags for brw->state.cache.
@@ -363,16 +356,6 @@ enum brw_cache_id {
 #define CACHE_NEW_SURFACE                (1<<BRW_SS_SURFACE)
 #define CACHE_NEW_SURF_BIND              (1<<BRW_SS_SURF_BIND)
 
-
-
-
-enum brw_mempool_id {
-   BRW_GS_POOL,
-   BRW_SS_POOL,
-   BRW_MAX_POOL
-};
-
-
 struct brw_cached_batch_item {
    struct header *header;
    GLuint sz;
@@ -389,12 +372,16 @@ struct brw_cached_batch_item {
 struct brw_vertex_element {
    const struct gl_client_array *glarray;
 
-   struct brw_vertex_element_state *vep;
-
-   GLuint index;
+   /** Size of a complete element */
    GLuint element_size;
+   /** Number of uploaded elements for this input. */
    GLuint count;
-   GLuint vbo_rebase_offset;
+   /** Byte stride between elements in the uploaded array */
+   GLuint stride;
+   /** Offset of the first element within the buffer object */
+   unsigned int offset;
+   /** Buffer object containing the uploaded vertex data */
+   dri_bo *bo;
 };
 
 
@@ -421,7 +408,22 @@ struct brw_tnl_cache {
    GLuint size, n_items;
 };
 
+struct brw_query_object {
+   struct gl_query_object Base;
+
+   /** Doubly linked list of active query objects in the context. */
+   struct brw_query_object *prev, *next;
+
+   /** Last query BO associated with this query. */
+   dri_bo *bo;
+   /** First index in bo with query data for this object. */
+   int first_index;
+   /** Last index in bo with query data for this object. */
+   int last_index;
 
+   /* Total count of pixels from previous BOs */
+   unsigned int count;
+};
 
 struct brw_context 
 {
@@ -429,51 +431,67 @@ struct brw_context
    GLuint primitive;
 
    GLboolean emit_state_always;
-   GLboolean wrap;
    GLboolean tmp_fallback;
+   GLboolean no_batch_wrap;
 
    struct {
       struct brw_state_flags dirty;
       struct brw_tracked_state **atoms;
       GLuint nr_atoms;
 
-
-      struct intel_region *draw_region;
+      GLuint nr_draw_regions;
+      struct intel_region *draw_regions[MAX_DRAW_BUFFERS];
       struct intel_region *depth_region;
+
+      /**
+       * List of buffers accumulated in brw_validate_state to receive
+       * dri_bo_check_aperture treatment before exec, so we can know if we
+       * should flush the batch and try again before emitting primitives.
+       *
+       * This can be a fixed number as we only have a limited number of
+       * objects referenced from the batchbuffer in a primitive emit,
+       * consisting of the vertex buffers, pipelined state pointers,
+       * the CURBE, the depth buffer, and a query BO.
+       */
+      dri_bo *validated_bos[VERT_ATTRIB_MAX + 16];
+      int validated_bo_count;
    } state;
 
    struct brw_state_pointers attribs;
-   struct brw_mem_pool pool[BRW_MAX_POOL];
-   struct brw_cache cache[BRW_MAX_CACHE];
+   struct brw_cache cache;
    struct brw_cached_batch_item *cached_batch_items;
 
    struct {
-
-      /* Arrays with buffer objects to copy non-bufferobj arrays into
-       * for upload:
-       */
-      struct gl_client_array vbo_array[VERT_ATTRIB_MAX];
-
       struct brw_vertex_element inputs[VERT_ATTRIB_MAX];
 
 #define BRW_NR_UPLOAD_BUFS 17
 #define BRW_UPLOAD_INIT_SIZE (128*1024)
 
       struct {
-	 struct gl_buffer_object *vbo[BRW_NR_UPLOAD_BUFS];
-	 GLuint buf;
+	 dri_bo *bo;
 	 GLuint offset;
-	 GLuint size;
-	 GLuint wrap;
       } upload;
 
       /* Summary of size and varying of active arrays, so we can check
        * for changes to this state:
        */
       struct brw_vertex_info info;
+      unsigned int min_index, max_index;
    } vb;
 
    struct {
+      /**
+       * Index buffer for this draw_prims call.
+       *
+       * Updates are signaled by BRW_NEW_INDICES.
+       */
+      const struct _mesa_index_buffer *ib;
+
+      dri_bo *bo;
+      unsigned int offset;
+   } ib;
+
+   struct {
       /* Will be allocated on demand if needed.   
        */
       struct brw_state_pointers attribs;
@@ -483,18 +501,17 @@ struct brw_context
       struct gl_buffer_object *vbo;
 
       struct intel_region *saved_draw_region;
+      GLuint saved_nr_draw_regions;
       struct intel_region *saved_depth_region;
 
-      GLuint restore_draw_mask;
+      GLuint restore_draw_buffers[MAX_DRAW_BUFFERS];
+      GLuint restore_num_draw_buffers;
+
       struct gl_fragment_program *restore_fp;
       
       GLboolean active;
    } metaops;
 
-   /* Track fixed function t&l in a vertex program:
-    */
-   struct gl_vertex_program *tnl_program;
-   struct brw_tnl_cache tnl_program_cache;
 
    /* Active vertex program: 
     */
@@ -552,42 +569,51 @@ struct brw_context
        */
       struct brw_tracked_state tracked_state;
 
-      GLuint gs_offset;
+      dri_bo *curbe_bo;
+      /** Offset within curbe_bo of space for current curbe entry */
+      GLuint curbe_offset;
+      /** Offset within curbe_bo of space for next curbe entry */
+      GLuint curbe_next_offset;
 
       GLfloat *last_buf;
       GLuint last_bufsz;
+      /**
+       *  Whether we should create a new bo instead of reusing the old one
+       * (if we just dispatch the batch pointing at the old one.
+       */
+      GLboolean need_new_bo;
    } curbe;
 
    struct {
       struct brw_vs_prog_data *prog_data;
 
-      GLuint prog_gs_offset;
-      GLuint state_gs_offset;	
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
    } vs;
 
    struct {
       struct brw_gs_prog_data *prog_data;
 
       GLboolean prog_active;
-      GLuint prog_gs_offset;
-      GLuint state_gs_offset;	
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
    } gs;
 
    struct {
       struct brw_clip_prog_data *prog_data;
 
-      GLuint prog_gs_offset;
-      GLuint vp_gs_offset;
-      GLuint state_gs_offset;	
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
+      dri_bo *vp_bo;
    } clip;
 
 
    struct {
       struct brw_sf_prog_data *prog_data;
 
-      GLuint prog_gs_offset;
-      GLuint vp_gs_offset;
-      GLuint state_gs_offset;
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
+      dri_bo *vp_bo;
    } sf;
 
    struct {
@@ -598,36 +624,39 @@ struct brw_context
        */
       GLuint input_size_masks[4];
 
-
-      /* State structs
-       */
-      struct brw_sampler_default_color sdc[BRW_MAX_TEX_UNIT];
-      struct brw_sampler_state sampler[BRW_MAX_TEX_UNIT];
+      /** Array of surface default colors (texture border color) */
+      dri_bo *sdc_bo[BRW_MAX_TEX_UNIT];
 
       GLuint render_surf;
       GLuint nr_surfaces;      
 
       GLuint max_threads;
-      struct buffer *scratch_buffer;
-      GLuint scratch_buffer_size;
+      dri_bo *scratch_buffer;
 
       GLuint sampler_count;
-      GLuint sampler_gs_offset;
+      dri_bo *sampler_bo;
 
-      struct brw_surface_binding_table bind;
-      GLuint bind_ss_offset;
+      /** Binding table of pointers to surf_bo entries */
+      dri_bo *bind_bo;
+      dri_bo *surf_bo[BRW_WM_MAX_SURF];
 
-      GLuint prog_gs_offset;
-      GLuint state_gs_offset;
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
    } wm;
 
 
    struct {
-      GLuint vp_gs_offset;
-      GLuint state_gs_offset;
+      dri_bo *prog_bo;
+      dri_bo *state_bo;
+      dri_bo *vp_bo;
    } cc;
 
-   
+   struct {
+      struct brw_query_object active_head;
+      dri_bo *bo;
+      int index;
+      GLboolean active;
+   } query;
    /* Used to give every program string a unique id
     */
    GLuint program_id;
@@ -652,24 +681,27 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 			    __DRIcontextPrivate *driContextPriv,
 			    void *sharedContextPrivate);
 
-
-
 /*======================================================================
- * brw_state.c
+ * brw_queryobj.c
  */
-void brw_validate_state( struct brw_context *brw );
-void brw_init_state( struct brw_context *brw );
-void brw_destroy_state( struct brw_context *brw );
-
+void brw_init_queryobj_functions(struct dd_function_table *functions);
+void brw_prepare_query_begin(struct brw_context *brw);
+void brw_emit_query_begin(struct brw_context *brw);
+void brw_emit_query_end(struct brw_context *brw);
 
+/*======================================================================
+ * brw_state_dump.c
+ */
+void brw_debug_batch(struct intel_context *intel);
 
 /*======================================================================
  * brw_tex.c
  */
 void brwUpdateTextureState( struct intel_context *intel );
-void brwInitTextureFuncs( struct dd_function_table *functions );
-void brw_FrameBufferTexInit( struct brw_context *brw );
+void brw_FrameBufferTexInit( struct brw_context *brw,
+			     struct intel_region *region );
 void brw_FrameBufferTexDestroy( struct brw_context *brw );
+void brw_validate_textures( struct brw_context *brw );
 
 /*======================================================================
  * brw_metaops.c
@@ -696,11 +728,13 @@ void brw_upload_constant_buffer_state(struct brw_context *brw);
  * Inline conversion functions.  These are better-typed than the
  * macros used previously:
  */
-static inline struct brw_context *
+static INLINE struct brw_context *
 brw_context( GLcontext *ctx )
 {
    return (struct brw_context *)ctx;
 }
 
+#define DO_SETUP_BITS ((1<<(FRAG_ATTRIB_MAX)) - 1)
+
 #endif
 
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index 3f0aaa1f86d..fbf473abf66 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -31,10 +31,10 @@
 
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
 #include "intel_batchbuffer.h"
@@ -42,7 +42,6 @@
 #include "brw_defines.h"
 #include "brw_state.h"
 #include "brw_util.h"
-#include "brw_aub.h"
 
 
 /* Partition the CURBE between the various users of constant values:
@@ -90,7 +89,7 @@ static void calculate_curbe_offsets( struct brw_context *brw )
     */
    if (nr_fp_regs > brw->curbe.wm_size ||
        nr_vp_regs > brw->curbe.vs_size ||
-       nr_clip_regs > brw->curbe.clip_size ||
+       nr_clip_regs != brw->curbe.clip_size ||
        (total_regs < brw->curbe.total_size / 4 &&
 	brw->curbe.total_size > 16)) {
 
@@ -127,7 +126,7 @@ const struct brw_tracked_state brw_curbe_offsets = {
       .brw  = BRW_NEW_VERTEX_PROGRAM,
       .cache = CACHE_NEW_WM_PROG
    },
-   .update = calculate_curbe_offsets
+   .prepare = calculate_curbe_offsets
 };
 
 
@@ -156,19 +155,7 @@ void brw_upload_constant_buffer_state(struct brw_context *brw)
 
    assert(brw->urb.nr_cs_entries);
    BRW_CACHED_BATCH_STRUCT(brw, &cbs);
-}      
-
-#if 0
-const struct brw_tracked_state brw_constant_buffer_state = {
-   .dirty = {
-      .mesa = 0,
-      .brw = BRW_NEW_URB_FENCE,
-      .cache = 0
-   },
-   .update = brw_upload_constant_buffer_state
-};
-#endif
-
+}
 
 static GLfloat fixed_plane[6][4] = {
    { 0,    0,   -1, 1 },
@@ -183,12 +170,11 @@ static GLfloat fixed_plane[6][4] = {
  * cache mechanism, but maybe would benefit from a comparison against
  * the current uploaded set of constants.
  */
-static void upload_constant_buffer(struct brw_context *brw)
+static void prepare_constant_buffer(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct brw_vertex_program *vp = (struct brw_vertex_program *)brw->vertex_program;
    struct brw_fragment_program *fp = (struct brw_fragment_program *)brw->fragment_program;
-   struct brw_mem_pool *pool = &brw->pool[BRW_GS_POOL];
    GLuint sz = brw->curbe.total_size;
    GLuint bufsz = sz * 16 * sizeof(GLfloat);
    GLfloat *buf;
@@ -198,24 +184,17 @@ static void upload_constant_buffer(struct brw_context *brw)
     * function will also be called whenever fp or vp changes.
     */
    brw->curbe.tracked_state.dirty.mesa = (_NEW_TRANSFORM|_NEW_PROJECTION);
-   brw->curbe.tracked_state.dirty.mesa |= vp->param_state;
-   brw->curbe.tracked_state.dirty.mesa |= fp->param_state;
+   brw->curbe.tracked_state.dirty.mesa |= vp->program.Base.Parameters->StateFlags;
+   brw->curbe.tracked_state.dirty.mesa |= fp->program.Base.Parameters->StateFlags;
 
    if (sz == 0) {
-      struct brw_constant_buffer cb;
-      cb.header.opcode = CMD_CONST_BUFFER;
-      cb.header.length = sizeof(cb)/4 - 2;
-      cb.header.valid = 0;
-      cb.bits0.buffer_length = 0;
-      cb.bits0.buffer_address = 0;
-      BRW_BATCH_STRUCT(brw, &cb);
 
       if (brw->curbe.last_buf) {
 	 free(brw->curbe.last_buf);
 	 brw->curbe.last_buf = NULL;
 	 brw->curbe.last_bufsz  = 0;
       }
-       
+
       return;
    }
 
@@ -290,11 +269,11 @@ static void upload_constant_buffer(struct brw_context *brw)
 		   brw->curbe.last_buf ? memcmp(buf, brw->curbe.last_buf, bufsz) : -1);
    }
 
-   if (brw->curbe.last_buf &&
+   if (brw->curbe.curbe_bo != NULL &&
+       brw->curbe.last_buf &&
        bufsz == brw->curbe.last_bufsz &&
        memcmp(buf, brw->curbe.last_buf, bufsz) == 0) {
       free(buf);
-/*       return; */
    } 
    else {
       if (brw->curbe.last_buf)
@@ -302,61 +281,66 @@ static void upload_constant_buffer(struct brw_context *brw)
       brw->curbe.last_buf = buf;
       brw->curbe.last_bufsz = bufsz;
 
-      
-      if (!brw_pool_alloc(pool, 
-			  bufsz,
-			  6,
-			  &brw->curbe.gs_offset)) {
-	 _mesa_printf("out of GS memory for curbe\n");
-	 assert(0);
-	 return;
+      if (brw->curbe.curbe_bo != NULL &&
+	  (brw->curbe.need_new_bo ||
+	   brw->curbe.curbe_next_offset + bufsz > brw->curbe.curbe_bo->size))
+      {
+	 dri_bo_unreference(brw->curbe.curbe_bo);
+	 brw->curbe.curbe_bo = NULL;
+      }
+
+      if (brw->curbe.curbe_bo == NULL) {
+	 /* Allocate a single page for CURBE entries for this batchbuffer.
+	  * They're generally around 64b.
+	  */
+	 brw->curbe.curbe_bo = dri_bo_alloc(brw->intel.bufmgr, "CURBE",
+					    4096, 1 << 6);
+	 brw->curbe.curbe_next_offset = 0;
       }
-            
+
+      brw->curbe.curbe_offset = brw->curbe.curbe_next_offset;
+      brw->curbe.curbe_next_offset += bufsz;
+      brw->curbe.curbe_next_offset = ALIGN(brw->curbe.curbe_next_offset, 64);
 
       /* Copy data to the buffer:
        */
-      bmBufferSubDataAUB(&brw->intel,
-			 pool->buffer,
-			 brw->curbe.gs_offset, 
-			 bufsz, 
-			 buf,
-			 DW_CONSTANT_BUFFER,
-			 0);
+      dri_bo_subdata(brw->curbe.curbe_bo, brw->curbe.curbe_offset, bufsz, buf);
    }
 
-   /* TODO: only emit the constant_buffer packet when necessary, ie:
-      - contents have changed
-      - offset has changed
-      - hw requirements due to other packets emitted.
-   */
-   {
-      struct brw_constant_buffer cb;
-      
-      memset(&cb, 0, sizeof(cb));
-
-      cb.header.opcode = CMD_CONST_BUFFER;
-      cb.header.length = sizeof(cb)/4 - 2;
-      cb.header.valid = 1;
-      cb.bits0.buffer_length = sz - 1;
-      cb.bits0.buffer_address = brw->curbe.gs_offset >> 6;
-      
-      /* Because this provokes an action (ie copy the constants into the
-       * URB), it shouldn't be shortcircuited if identical to the
-       * previous time - because eg. the urb destination may have
-       * changed, or the urb contents different to last time.  
-       *
-       * Note that the data referred to is actually copied internally,
-       * not just used in place according to passed pointer.
-       *
-       * It appears that the CS unit takes care of using each available
-       * URB entry (Const URB Entry == CURBE) in turn, and issuing
-       * flushes as necessary when doublebuffering of CURBEs isn't
-       * possible.
-       */
-/*       intel_batchbuffer_align(brw->intel.batch, 64, sizeof(cb)); */
-      BRW_BATCH_STRUCT(brw, &cb);
-/*       intel_batchbuffer_align(brw->intel.batch, 64, 0); */
+   brw_add_validated_bo(brw, brw->curbe.curbe_bo);
+
+   /* Because this provokes an action (ie copy the constants into the
+    * URB), it shouldn't be shortcircuited if identical to the
+    * previous time - because eg. the urb destination may have
+    * changed, or the urb contents different to last time.
+    *
+    * Note that the data referred to is actually copied internally,
+    * not just used in place according to passed pointer.
+    *
+    * It appears that the CS unit takes care of using each available
+    * URB entry (Const URB Entry == CURBE) in turn, and issuing
+    * flushes as necessary when doublebuffering of CURBEs isn't
+    * possible.
+    */
+}
+
+
+static void emit_constant_buffer(struct brw_context *brw)
+{
+   struct intel_context *intel = &brw->intel;
+   GLuint sz = brw->curbe.total_size;
+
+   BEGIN_BATCH(2, IGNORE_CLIPRECTS);
+   if (sz == 0) {
+      OUT_BATCH((CMD_CONST_BUFFER << 16) | (2 - 2));
+      OUT_BATCH(0);
+   } else {
+      OUT_BATCH((CMD_CONST_BUFFER << 16) | (1 << 8) | (2 - 2));
+      OUT_RELOC(brw->curbe.curbe_bo,
+		I915_GEM_DOMAIN_INSTRUCTION, 0,
+		(sz - 1) + brw->curbe.curbe_offset);
    }
+   ADVANCE_BATCH();
 }
 
 /* This tracked state is unique in that the state it monitors varies
@@ -372,9 +356,11 @@ const struct brw_tracked_state brw_constant_buffer = {
 	       BRW_NEW_VERTEX_PROGRAM |
 	       BRW_NEW_URB_FENCE | /* Implicit - hardware requires this, not used above */
 	       BRW_NEW_PSP | /* Implicit - hardware requires this, not used above */
-	       BRW_NEW_CURBE_OFFSETS),
+	       BRW_NEW_CURBE_OFFSETS |
+	       BRW_NEW_BATCH),
       .cache = (CACHE_NEW_WM_PROG) 
    },
-   .update = upload_constant_buffer
+   .prepare = prepare_constant_buffer,
+   .emit = emit_constant_buffer,
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index e8f878a7018..39c32255f8b 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -33,69 +33,6 @@
 #ifndef BRW_DEFINES_H
 #define BRW_DEFINES_H
 
-/*
- */
-#define MI_NOOP                              0x00
-#define MI_USER_INTERRUPT                    0x02
-#define MI_WAIT_FOR_EVENT                    0x03
-#define MI_FLUSH                             0x04
-#define MI_REPORT_HEAD                       0x07
-#define MI_ARB_ON_OFF                        0x08
-#define MI_BATCH_BUFFER_END                  0x0A
-#define MI_OVERLAY_FLIP                      0x11
-#define MI_LOAD_SCAN_LINES_INCL              0x12
-#define MI_LOAD_SCAN_LINES_EXCL              0x13
-#define MI_DISPLAY_BUFFER_INFO               0x14
-#define MI_SET_CONTEXT                       0x18
-#define MI_STORE_DATA_IMM                    0x20
-#define MI_STORE_DATA_INDEX                  0x21
-#define MI_LOAD_REGISTER_IMM                 0x22
-#define MI_STORE_REGISTER_MEM                0x24
-#define MI_BATCH_BUFFER_START                0x31
-
-#define MI_SYNCHRONOUS_FLIP                  0x0 
-#define MI_ASYNCHRONOUS_FLIP                 0x1
-
-#define MI_BUFFER_SECURE                     0x0 
-#define MI_BUFFER_NONSECURE                  0x1
-
-#define MI_ARBITRATE_AT_CHAIN_POINTS         0x0 
-#define MI_ARBITRATE_BETWEEN_INSTS           0x1
-#define MI_NO_ARBITRATION                    0x3 
-
-#define MI_CONDITION_CODE_WAIT_DISABLED      0x0
-#define MI_CONDITION_CODE_WAIT_0             0x1
-#define MI_CONDITION_CODE_WAIT_1             0x2
-#define MI_CONDITION_CODE_WAIT_2             0x3
-#define MI_CONDITION_CODE_WAIT_3             0x4
-#define MI_CONDITION_CODE_WAIT_4             0x5
-
-#define MI_DISPLAY_PIPE_A                    0x0
-#define MI_DISPLAY_PIPE_B                    0x1
-
-#define MI_DISPLAY_PLANE_A                   0x0 
-#define MI_DISPLAY_PLANE_B                   0x1
-#define MI_DISPLAY_PLANE_C                   0x2
-
-#define MI_STANDARD_FLIP                                 0x0
-#define MI_ENQUEUE_FLIP_PERFORM_BASE_FRAME_NUMBER_LOAD   0x1
-#define MI_ENQUEUE_FLIP_TARGET_FRAME_NUMBER_RELATIVE     0x2
-#define MI_ENQUEUE_FLIP_ABSOLUTE_TARGET_FRAME_NUMBER     0x3
-
-#define MI_PHYSICAL_ADDRESS                  0x0
-#define MI_VIRTUAL_ADDRESS                   0x1
-
-#define MI_BUFFER_MEMORY_MAIN                0x0 
-#define MI_BUFFER_MEMORY_GTT                 0x2
-#define MI_BUFFER_MEMORY_PER_PROCESS_GTT     0x3 
-
-#define MI_FLIP_CONTINUE                     0x0
-#define MI_FLIP_ON                           0x1
-#define MI_FLIP_OFF                          0x2
-
-#define MI_UNTRUSTED_REGISTER_SPACE          0x0
-#define MI_TRUSTED_REGISTER_SPACE            0x1
-
 /* 3D state:
  */
 #define _3DOP_3DSTATE_PIPELINED       0x0
@@ -119,7 +56,6 @@
 #define _3DSTATE_LINE_STIPPLE                 0x08
 #define _3DSTATE_GLOBAL_DEPTH_OFFSET_CLAMP    0x09
 #define _3DCONTROL    0x00
-#define _3DPRIMITIVE  0x00
 
 #define PIPE_CONTROL_NOWRITE          0x00
 #define PIPE_CONTROL_WRITEIMMEDIATE   0x01
@@ -240,6 +176,8 @@
 #define BRW_FRONTWINDING_CW      0
 #define BRW_FRONTWINDING_CCW     1
 
+#define BRW_SPRITE_POINT_ENABLE  16
+
 #define BRW_INDEX_BYTE     0
 #define BRW_INDEX_WORD     1
 #define BRW_INDEX_DWORD    2
@@ -485,20 +423,6 @@
 #define BRW_VERTEX_SUBPIXEL_PRECISION_8BITS  0
 #define BRW_VERTEX_SUBPIXEL_PRECISION_4BITS  1
 
-#define BRW_VERTEXBUFFER_ACCESS_VERTEXDATA     0
-#define BRW_VERTEXBUFFER_ACCESS_INSTANCEDATA   1
-
-#define BRW_VFCOMPONENT_NOSTORE      0
-#define BRW_VFCOMPONENT_STORE_SRC    1
-#define BRW_VFCOMPONENT_STORE_0      2
-#define BRW_VFCOMPONENT_STORE_1_FLT  3
-#define BRW_VFCOMPONENT_STORE_1_INT  4
-#define BRW_VFCOMPONENT_STORE_VID    5
-#define BRW_VFCOMPONENT_STORE_IID    6
-#define BRW_VFCOMPONENT_STORE_PID    7
-
-
-
 /* Execution Unit (EU) defines
  */
 
@@ -815,14 +739,40 @@
 
 #define CMD_STATE_BASE_ADDRESS        0x6101
 #define CMD_STATE_INSN_POINTER        0x6102
-#define CMD_PIPELINE_SELECT           0x6104
+#define CMD_PIPELINE_SELECT_965       0x6104
+#define CMD_PIPELINE_SELECT_GM45      0x6904
 
 #define CMD_PIPELINED_STATE_POINTERS  0x7800
 #define CMD_BINDING_TABLE_PTRS        0x7801
+
 #define CMD_VERTEX_BUFFER             0x7808
+# define BRW_VB0_INDEX_SHIFT		27
+# define BRW_VB0_ACCESS_VERTEXDATA	(0 << 26)
+# define BRW_VB0_ACCESS_INSTANCEDATA	(1 << 26)
+# define BRW_VB0_PITCH_SHIFT		0
+
 #define CMD_VERTEX_ELEMENT            0x7809
+# define BRW_VE0_INDEX_SHIFT		27
+# define BRW_VE0_FORMAT_SHIFT		16
+# define BRW_VE0_VALID			(1 << 26)
+# define BRW_VE0_SRC_OFFSET_SHIFT	0
+# define BRW_VE1_COMPONENT_NOSTORE	0
+# define BRW_VE1_COMPONENT_STORE_SRC	1
+# define BRW_VE1_COMPONENT_STORE_0	2
+# define BRW_VE1_COMPONENT_STORE_1_FLT	3
+# define BRW_VE1_COMPONENT_STORE_1_INT	4
+# define BRW_VE1_COMPONENT_STORE_VID	5
+# define BRW_VE1_COMPONENT_STORE_IID	6
+# define BRW_VE1_COMPONENT_STORE_PID	7
+# define BRW_VE1_COMPONENT_0_SHIFT	28
+# define BRW_VE1_COMPONENT_1_SHIFT	24
+# define BRW_VE1_COMPONENT_2_SHIFT	20
+# define BRW_VE1_COMPONENT_3_SHIFT	16
+# define BRW_VE1_DST_OFFSET_SHIFT	0
+
 #define CMD_INDEX_BUFFER              0x780a
-#define CMD_VF_STATISTICS             0x780b
+#define CMD_VF_STATISTICS_965         0x780b
+#define CMD_VF_STATISTICS_GM45        0x680b
 
 #define CMD_DRAW_RECT                 0x7900
 #define CMD_BLEND_CONSTANT_COLOR      0x7901
@@ -832,6 +782,7 @@
 #define CMD_POLY_STIPPLE_PATTERN      0x7907
 #define CMD_LINE_STIPPLE_PATTERN      0x7908
 #define CMD_GLOBAL_DEPTH_OFFSET_CLAMP 0x7909
+#define CMD_AA_LINE_PARAMETERS        0x790a
 
 #define CMD_PIPE_CONTROL              0x7a00
 
@@ -845,6 +796,11 @@
 #define R02_PRIM_END    0x1
 #define R02_PRIM_START  0x2
 
+#include "intel_chipset.h"
 
+#define BRW_IS_G4X(brw)         (IS_G4X((brw)->intel.intelScreen->deviceID))
+#define CMD_PIPELINE_SELECT(brw)        (BRW_IS_G4X(brw) ? CMD_PIPELINE_SELECT_GM45 : CMD_PIPELINE_SELECT_965)
+#define CMD_VF_STATISTICS(brw)          (BRW_IS_G4X(brw) ? CMD_VF_STATISTICS_GM45 : CMD_VF_STATISTICS_965)
+#define URB_SIZES(brw)                  (BRW_IS_G4X(brw) ? 384 : 256)  /* 512 bit units */
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 0c64d7e756d..c3a26fc82eb 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -27,20 +27,18 @@
 
 #include <stdlib.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "state.h"
-#include "api_validate.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/state.h"
+#include "main/api_validate.h"
+#include "main/enums.h"
 
 #include "brw_draw.h"
 #include "brw_defines.h"
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_state.h"
 #include "brw_fallback.h"
 
-#include "intel_ioctl.h"
 #include "intel_batchbuffer.h"
 #include "intel_buffer_objects.h"
 
@@ -49,9 +47,9 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 
+#define FILE_DEBUG_FLAG DEBUG_BATCH
 
-
-static GLuint hw_prim[GL_POLYGON+1] = {
+static GLuint prim_to_hw_prim[GL_POLYGON+1] = {
    _3DPRIM_POINTLIST,
    _3DPRIM_LINELIST,
    _3DPRIM_LINELOOP,
@@ -105,11 +103,9 @@ static GLuint brw_set_prim(struct brw_context *brw, GLenum prim)
 	 brw->intel.reduced_primitive = reduced_prim[prim];
 	 brw->state.dirty.brw |= BRW_NEW_REDUCED_PRIMITIVE;
       }
-
-      brw_validate_state(brw);
    }
 
-   return hw_prim[prim];
+   return prim_to_hw_prim[prim];
 }
 
 
@@ -124,28 +120,9 @@ static GLuint trim(GLenum prim, GLuint length)
 }
 
 
-static void brw_emit_cliprect( struct brw_context *brw, 
-			       const drm_clip_rect_t *rect )
-{
-   struct brw_drawrect bdr;
-
-   bdr.header.opcode = CMD_DRAW_RECT;
-   bdr.header.length = sizeof(bdr)/4 - 2;
-   bdr.xmin = rect->x1;
-   bdr.xmax = rect->x2 - 1;
-   bdr.ymin = rect->y1;
-   bdr.ymax = rect->y2 - 1;
-   bdr.xorg = brw->intel.drawX;
-   bdr.yorg = brw->intel.drawY;
-
-   intel_batchbuffer_data( brw->intel.batch, &bdr, sizeof(bdr), 
-			   INTEL_BATCH_NO_CLIPRECTS);
-}
-
-
-static void brw_emit_prim( struct brw_context *brw, 
-			   const struct _mesa_prim *prim )
-
+static void brw_emit_prim(struct brw_context *brw,
+			  const struct _mesa_prim *prim,
+			  uint32_t hw_prim)
 {
    struct brw_3d_primitive prim_packet;
 
@@ -156,7 +133,7 @@ static void brw_emit_prim( struct brw_context *brw,
    prim_packet.header.opcode = CMD_3D_PRIM;
    prim_packet.header.length = sizeof(prim_packet)/4 - 2;
    prim_packet.header.pad = 0;
-   prim_packet.header.topology = brw_set_prim(brw, prim->mode);
+   prim_packet.header.topology = hw_prim;
    prim_packet.header.indexed = prim->indexed;
 
    prim_packet.verts_per_instance = trim(prim->mode, prim->count);
@@ -165,20 +142,25 @@ static void brw_emit_prim( struct brw_context *brw,
    prim_packet.start_instance_location = 0;
    prim_packet.base_vert_location = 0;
 
+   /* Can't wrap here, since we rely on the validated state. */
+   brw->no_batch_wrap = GL_TRUE;
    if (prim_packet.verts_per_instance) {
-      intel_batchbuffer_data( brw->intel.batch, &prim_packet, sizeof(prim_packet), 
-			      INTEL_BATCH_NO_CLIPRECTS);
+      intel_batchbuffer_data( brw->intel.batch, &prim_packet,
+			      sizeof(prim_packet), LOOP_CLIPRECTS);
    }
+   brw->no_batch_wrap = GL_FALSE;
 }
 
 static void brw_merge_inputs( struct brw_context *brw,
 		       const struct gl_client_array *arrays[])
 {
-   struct brw_vertex_element *inputs = brw->vb.inputs;
    struct brw_vertex_info old = brw->vb.info;
    GLuint i;
 
-   memset(inputs, 0, sizeof(*inputs));
+   for (i = 0; i < VERT_ATTRIB_MAX; i++)
+      dri_bo_unreference(brw->vb.inputs[i].bo);
+
+   memset(&brw->vb.inputs, 0, sizeof(brw->vb.inputs));
    memset(&brw->vb.info, 0, sizeof(brw->vb.info));
 
    for (i = 0; i < VERT_ATTRIB_MAX; i++) {
@@ -189,7 +171,8 @@ static void brw_merge_inputs( struct brw_context *brw,
 	 if (arrays[i]->StrideB != 0)
 	    brw->vb.info.varying |= 1 << i;
 
-	 brw->vb.info.sizes[i/16] |= (inputs[i].glarray->Size - 1) << ((i%16) * 2);
+	 brw->vb.info.sizes[i/16] |= (brw->vb.inputs[i].glarray->Size - 1) <<
+	    ((i%16) * 2);
       }
    }
 
@@ -271,15 +254,29 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    struct intel_context *intel = intel_context(ctx);
    struct brw_context *brw = brw_context(ctx);
    GLboolean retval = GL_FALSE;
-   GLuint i, j;
+   GLboolean warn = GL_FALSE;
+   GLboolean first_time = GL_TRUE;
+   GLuint i;
 
    if (ctx->NewState)
       _mesa_update_state( ctx );
 
+   if (check_fallbacks(brw, prim, nr_prims))
+      return GL_FALSE;
+
+   brw_validate_textures( brw );
+
    /* Bind all inputs, derive varying and size information:
     */
    brw_merge_inputs( brw, arrays );
-      
+
+   brw->ib.ib = ib;
+   brw->state.dirty.brw |= BRW_NEW_INDICES;
+
+   brw->vb.min_index = min_index;
+   brw->vb.max_index = max_index;
+   brw->state.dirty.brw |= BRW_NEW_VERTICES;
+
    /* Have to validate state quite late.  Will rebuild tnl_program,
     * which depends on varying information.  
     * 
@@ -289,95 +286,73 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
 
    LOCK_HARDWARE(intel);
 
-   if (brw->intel.numClipRects == 0) {
-      assert(intel->batch->ptr == intel->batch->map + intel->batch->offset);
+   if (!intel->constant_cliprect && intel->driDrawable->numClipRects == 0) {
       UNLOCK_HARDWARE(intel);
       return GL_TRUE;
    }
 
-   {
-      /* Set the first primitive early, ahead of validate_state:
-       */
-      brw_set_prim(brw, prim[0].mode);
+   for (i = 0; i < nr_prims; i++) {
+      uint32_t hw_prim;
 
-      /* XXX:  Need to separate validate and upload of state.  
+      /* Flush the batch if it's approaching full, so that we don't wrap while
+       * we've got validated state that needs to be in the same batch as the
+       * primitives.  This fraction is just a guess (minimal full state plus
+       * a primitive is around 512 bytes), and would be better if we had
+       * an upper bound of how much we might emit in a single
+       * brw_try_draw_prims().
        */
-      brw_validate_state( brw );
+      intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
+				      LOOP_CLIPRECTS);
 
-      /* Various fallback checks:
-       */
-      if (brw->intel.Fallback) 
-	 goto out;
+      hw_prim = brw_set_prim(brw, prim[i].mode);
 
-      if (check_fallbacks( brw, prim, nr_prims ))
-	 goto out;
-	  
-      /* Upload index, vertex data: 
-       */
-      if (ib)
-	 brw_upload_indices( brw, ib );
+      if (first_time || (brw->state.dirty.brw & BRW_NEW_PRIMITIVE)) {
+	 first_time = GL_FALSE;
 
-      if (!brw_upload_vertices( brw, min_index, max_index)) {
-	 goto out;
-      }
+	 /* Various fallback checks:  */
+	 if (brw->intel.Fallback)
+	    goto out;
 
-      /* For single cliprect, state is already emitted: 
-       */
-      if (brw->intel.numClipRects == 1) {
-	 for (i = 0; i < nr_prims; i++) {
-	    brw_emit_prim(brw, &prim[i]);   
-	 }
-      }
-      else {
-	 /* Otherwise, explicitly do the cliprects at this point:
-	  */
-	 for (j = 0; j < brw->intel.numClipRects; j++) {
-	    brw_emit_cliprect(brw, &brw->intel.pClipRects[j]);
+	 brw_validate_state(brw);
 
-	    /* Emit prims to batchbuffer: 
+	 /* Check that we can fit our state in with our existing batchbuffer, or
+	  * flush otherwise.
+	  */
+	 if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+					     brw->state.validated_bo_count)) {
+	    static GLboolean warned;
+	    intel_batchbuffer_flush(intel->batch);
+
+	    /* Validate the state after we flushed the batch (which would have
+	     * changed the set of dirty state).  If we still fail to
+	     * check_aperture, warn of what's happening, but attempt to continue
+	     * on since it may succeed anyway, and the user would probably rather
+	     * see a failure and a warning than a fallback.
 	     */
-	    for (i = 0; i < nr_prims; i++) {
-	       brw_emit_prim(brw, &prim[i]);   
+	    brw_validate_state(brw);
+	    if (!warned &&
+		dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+						brw->state.validated_bo_count)) {
+	       warn = GL_TRUE;
+	       warned = GL_TRUE;
 	    }
 	 }
-      }
-      
-      intel->need_flush = GL_TRUE;
-      retval = GL_TRUE;
-   }
-
- out:
 
-   /* Currently have to do this to synchronize with the map/unmap of
-    * the vertex buffer in brw_exec_api.c.  Not sure if there is any
-    * way around this, as not every flush is due to a buffer filling
-    * up.
-    */
-   if (!intel_batchbuffer_flush( brw->intel.batch )) {
-      DBG("%s intel_batchbuffer_flush failed\n", __FUNCTION__);
-      retval = GL_FALSE;
-   }
+	 brw_upload_state(brw);
+      }
 
-   if (retval && intel->thrashing) {
-      bmSetFence(intel);
-   }
+      brw_emit_prim(brw, &prim[i], hw_prim);
 
-   /* Free any old data so it doesn't clog up texture memory - we
-    * won't be referencing it again.
-    */
-   while (brw->vb.upload.wrap != brw->vb.upload.buf) {
-      ctx->Driver.BufferData(ctx,
-			     GL_ARRAY_BUFFER_ARB,
-			     BRW_UPLOAD_INIT_SIZE,
-			     NULL,
-			     GL_DYNAMIC_DRAW_ARB,
-			     brw->vb.upload.vbo[brw->vb.upload.wrap]);
-      brw->vb.upload.wrap++;
-      brw->vb.upload.wrap %= BRW_NR_UPLOAD_BUFS;
+      retval = GL_TRUE;
    }
 
+ out:
    UNLOCK_HARDWARE(intel);
 
+   if (warn)
+      fprintf(stderr, "i965: Single primitive emit potentially exceeded "
+	      "available aperture space\n");
+
    if (!retval)
       DBG("%s failed\n", __FUNCTION__);
 
@@ -420,7 +395,6 @@ void brw_draw_prims( GLcontext *ctx,
 		     GLuint min_index,
 		     GLuint max_index )
 {
-   struct intel_context *intel = intel_context(ctx);
    GLboolean retval;
 
    /* Decide if we want to rebase.  If so we end up recursing once
@@ -435,25 +409,10 @@ void brw_draw_prims( GLcontext *ctx,
       return;
    }
 
-
    /* Make a first attempt at drawing:
     */
    retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
 
-   
-   /* This looks like out-of-memory but potentially we have
-    * situation where there is enough memory but it has become
-    * fragmented.  Clear out all heaps and start from scratch by
-    * faking a contended lock event:  (done elsewhere)
-    */
-   if (!retval && !intel->Fallback && bmError(intel)) {
-      DBG("retrying\n");
-      /* Then try a second time only to upload textures and draw the
-       * primitives:
-       */
-      retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
-   }
-
    /* Otherwise, we really are out of memory.  Pass the drawing
     * command to the software tnl module and which will in turn call
     * swrast to do the drawing.
@@ -463,56 +422,32 @@ void brw_draw_prims( GLcontext *ctx,
       _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
    }
 
-   if (intel->aub_file && (INTEL_DEBUG & DEBUG_SYNC)) {
-      intelFinish( &intel->ctx );
-      intel->aub_wrap = 1;
-   }
-}
-
-
-static void brw_invalidate_vbo_cb( struct intel_context *intel, void *ptr )
-{
-   /* nothing to do, we don't rely on the contents being preserved */
 }
 
-
 void brw_draw_init( struct brw_context *brw )
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct vbo_context *vbo = vbo_context(ctx);
-   GLuint i;
-   
+
    /* Register our drawing function: 
     */
    vbo->draw_prims = brw_draw_prims;
+}
 
-   brw->vb.upload.size = BRW_UPLOAD_INIT_SIZE;
+void brw_draw_destroy( struct brw_context *brw )
+{
+   int i;
 
-   for (i = 0; i < BRW_NR_UPLOAD_BUFS; i++) {
-      brw->vb.upload.vbo[i] = ctx->Driver.NewBufferObject(ctx, 1, GL_ARRAY_BUFFER_ARB);
-      
-      /* NOTE:  These are set to no-backing-store.
-       */
-      bmBufferSetInvalidateCB(&brw->intel,
-			      intel_bufferobj_buffer(intel_buffer_object(brw->vb.upload.vbo[i])),
-			      brw_invalidate_vbo_cb,
-			      &brw->intel,
-			      GL_TRUE);
+   if (brw->vb.upload.bo != NULL) {
+      dri_bo_unreference(brw->vb.upload.bo);
+      brw->vb.upload.bo = NULL;
    }
 
-   ctx->Driver.BufferData( ctx, 
-			   GL_ARRAY_BUFFER_ARB, 
-			   BRW_UPLOAD_INIT_SIZE,
-			   NULL,
-			   GL_DYNAMIC_DRAW_ARB,
-			   brw->vb.upload.vbo[0] );
-}
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
+      dri_bo_unreference(brw->vb.inputs[i].bo);
+      brw->vb.inputs[i].bo = NULL;
+   }
 
-void brw_draw_destroy( struct brw_context *brw )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-   GLuint i;
-   
-   for (i = 0; i < BRW_NR_UPLOAD_BUFS; i++)
-      ctx->Driver.DeleteBuffer(ctx, brw->vb.upload.vbo[i]);
+   dri_bo_unreference(brw->ib.bo);
+   brw->ib.bo = NULL;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_draw.h b/src/mesa/drivers/dri/i965/brw_draw.h
index 0f7b7383102..9aebbdb1b86 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.h
+++ b/src/mesa/drivers/dri/i965/brw_draw.h
@@ -28,7 +28,7 @@
 #ifndef BRW_DRAW_H
 #define BRW_DRAW_H
 
-#include "mtypes.h"		/* for GLcontext... */
+#include "main/mtypes.h"		/* for GLcontext... */
 #include "vbo/vbo.h"
 
 struct brw_context;
@@ -50,16 +50,4 @@ void brw_draw_destroy( struct brw_context *brw );
 void brw_init_current_values(GLcontext *ctx,
 			     struct gl_client_array *arrays);
 
-
-/* brw_draw_upload.c
- */
-void brw_upload_indices( struct brw_context *brw,
-			 const struct _mesa_index_buffer *index_buffer);
-
-GLboolean brw_upload_vertices( struct brw_context *brw,
-			       GLuint min_index,
-			       GLuint max_index );
-
-
-
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 6150cac4aa3..73d6dea01ee 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -27,52 +27,21 @@
 
 #include <stdlib.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "state.h"
-#include "api_validate.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/state.h"
+#include "main/api_validate.h"
+#include "main/enums.h"
 
 #include "brw_draw.h"
 #include "brw_defines.h"
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_state.h"
 #include "brw_fallback.h"
 
-#include "intel_ioctl.h"
 #include "intel_batchbuffer.h"
 #include "intel_buffer_objects.h"
-
-
-struct brw_array_state {
-   union header_union header;
-
-   struct {
-      union {
-	 struct {
-	    GLuint pitch:11; 
-	    GLuint pad:15;
-	    GLuint access_type:1; 
-	    GLuint vb_index:5; 
-	 } bits;
-	 GLuint dword;
-      } vb0;
-   
-      struct buffer *buffer;
-      GLuint offset;
-
-      GLuint max_index;   
-      GLuint instance_data_step_rate;
-
-   } vb[BRW_VBP_MAX];
-};
-
-
-static struct buffer *array_buffer( const struct gl_client_array *array )
-{
-   return intel_bufferobj_buffer(intel_buffer_object(array->BufferObj));
-}
+#include "intel_tex.h"
 
 static GLuint double_types[5] = {
    0,
@@ -247,194 +216,176 @@ static GLuint get_index_type(GLenum type)
    }
 }
 
-static void copy_strided_array( GLubyte *dest, 
-				const GLubyte *src, 
-				GLuint size, 
-				GLuint stride,
-				GLuint count )
-{
-   if (size == stride) 
-      do_memcpy(dest, src, count * size);
-   else {
-      GLuint i,j;
-   
-      for (i = 0; i < count; i++) {
-	 for (j = 0; j < size; j++)
-	    *dest++ = *src++;
-	 src += (stride - size);
-      }
-   }
-}
-
 static void wrap_buffers( struct brw_context *brw,
 			  GLuint size )
 {
-   GLcontext *ctx = &brw->intel.ctx;
-
    if (size < BRW_UPLOAD_INIT_SIZE)
       size = BRW_UPLOAD_INIT_SIZE;
 
-   brw->vb.upload.buf++;
-   brw->vb.upload.buf %= BRW_NR_UPLOAD_BUFS;
    brw->vb.upload.offset = 0;
 
-   ctx->Driver.BufferData(ctx,
-			  GL_ARRAY_BUFFER_ARB,
-			  size,
-			  NULL,
-			  GL_DYNAMIC_DRAW_ARB,
-			  brw->vb.upload.vbo[brw->vb.upload.buf]);
+   if (brw->vb.upload.bo != NULL)
+      dri_bo_unreference(brw->vb.upload.bo);
+   brw->vb.upload.bo = dri_bo_alloc(brw->intel.bufmgr, "temporary VBO",
+				    size, 1);
+
+   /* Set the internal VBO\ to no-backing-store.  We only use them as a
+    * temporary within a brw_try_draw_prims while the lock is held.
+    */
+   /* DON'T DO THIS AS IF WE HAVE TO RE-ORG MEMORY WE NEED SOMEWHERE WITH
+      FAKE TO PUSH THIS STUFF */
+//   if (!brw->intel.ttm)
+//      dri_bo_fake_disable_backing_store(brw->vb.upload.bo, NULL, NULL);
 }
 
 static void get_space( struct brw_context *brw,
 		       GLuint size,
-		       struct gl_buffer_object **vbo_return,
+		       dri_bo **bo_return,
 		       GLuint *offset_return )
 {
-   size = (size + 63) & ~63;
-   
-   if (brw->vb.upload.offset + size > BRW_UPLOAD_INIT_SIZE)
+   size = ALIGN(size, 64);
+
+   if (brw->vb.upload.bo == NULL ||
+       brw->vb.upload.offset + size > brw->vb.upload.bo->size) {
       wrap_buffers(brw, size);
+   }
 
-   *vbo_return = brw->vb.upload.vbo[brw->vb.upload.buf];
+   assert(*bo_return == NULL);
+   dri_bo_reference(brw->vb.upload.bo);
+   *bo_return = brw->vb.upload.bo;
    *offset_return = brw->vb.upload.offset;
-
    brw->vb.upload.offset += size;
 }
 
-
-
-static struct gl_client_array *
+static void
 copy_array_to_vbo_array( struct brw_context *brw,
-			 GLuint i,
-			 const struct gl_client_array *array,
-			 GLuint element_size,
-			 GLuint count)
+			 struct brw_vertex_element *element,
+			 GLuint dst_stride)
 {
-   GLcontext *ctx = &brw->intel.ctx;
-   struct gl_client_array *vbo_array = &brw->vb.vbo_array[i];
-   GLuint size = count * element_size;
-   struct gl_buffer_object *vbo;
-   GLuint offset;
-   GLuint new_stride;
+   GLuint size = element->count * dst_stride;
 
-   get_space(brw, size, &vbo, &offset);
+   get_space(brw, size, &element->bo, &element->offset);
 
-   if (array->StrideB == 0) {
-      assert(count == 1);
-      new_stride = 0;
+   if (element->glarray->StrideB == 0) {
+      assert(element->count == 1);
+      element->stride = 0;
+   } else {
+      element->stride = dst_stride;
    }
-   else 
-      new_stride = element_size;
-
-   vbo_array->Size = array->Size;
-   vbo_array->Type = array->Type;
-   vbo_array->Stride = new_stride;
-   vbo_array->StrideB = new_stride;   
-   vbo_array->Ptr = (const void *)offset;
-   vbo_array->Enabled = 1;
-   vbo_array->Normalized = array->Normalized;
-   vbo_array->_MaxElement = array->_MaxElement;	/* ? */
-   vbo_array->BufferObj = vbo;
-
-   {
-      GLubyte *map = ctx->Driver.MapBuffer(ctx,
-					   GL_ARRAY_BUFFER_ARB,
-					   GL_DYNAMIC_DRAW_ARB,
-					   vbo);
-   
-      map += offset;
 
-      copy_strided_array( map, 
-			  array->Ptr,
-			  element_size,
-			  array->StrideB,
-			  count);
+   if (dst_stride == element->glarray->StrideB) {
+      dri_bo_subdata(element->bo,
+		     element->offset,
+		     size,
+		     element->glarray->Ptr);
+   } else {
+      void *data;
+      char *dest;
+      const char *src = element->glarray->Ptr;
+      int i;
+
+      data = _mesa_malloc(dst_stride * element->count);
+      dest = data;
+      for (i = 0; i < element->count; i++) {
+	 memcpy(dest, src, dst_stride);
+	 src += element->glarray->StrideB;
+	 dest += dst_stride;
+      }
 
-      ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER_ARB, vbo_array->BufferObj);
+      dri_bo_subdata(element->bo,
+		     element->offset,
+		     size,
+		     data);
+      _mesa_free(data);
    }
-
-   return vbo_array;
-}
-
-
-
-static struct gl_client_array *
-interleaved_vbo_array( struct brw_context *brw,
-		       GLuint i,
-		       const struct gl_client_array *uploaded_array,
-		       const struct gl_client_array *array,
-		       const char *ptr)
-{
-   struct gl_client_array *vbo_array = &brw->vb.vbo_array[i];
-
-   vbo_array->Size = array->Size;
-   vbo_array->Type = array->Type;
-   vbo_array->Stride = array->Stride;
-   vbo_array->StrideB = array->StrideB;   
-   vbo_array->Ptr = (const void *)((const char *)uploaded_array->Ptr + 
-				   ((const char *)array->Ptr - ptr));
-   vbo_array->Enabled = 1;
-   vbo_array->Normalized = array->Normalized;
-   vbo_array->_MaxElement = array->_MaxElement;	
-   vbo_array->BufferObj = uploaded_array->BufferObj;
-
-   return vbo_array;
 }
 
-
-GLboolean brw_upload_vertices( struct brw_context *brw,
-			       GLuint min_index,
-			       GLuint max_index )
+static void brw_prepare_vertices(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = intel_context(ctx);
    GLuint tmp = brw->vs.prog_data->inputs_read; 
-   struct brw_vertex_element_packet vep;
-   struct brw_array_state vbp;
    GLuint i;
-   const void *ptr = NULL;
+   const unsigned char *ptr = NULL;
    GLuint interleave = 0;
+   unsigned int min_index = brw->vb.min_index;
+   unsigned int max_index = brw->vb.max_index;
 
    struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
    GLuint nr_enabled = 0;
 
    struct brw_vertex_element *upload[VERT_ATTRIB_MAX];
    GLuint nr_uploads = 0;
-   
-
-   memset(&vbp, 0, sizeof(vbp));
-   memset(&vep, 0, sizeof(vep));
 
    /* First build an array of pointers to ve's in vb.inputs_read
     */
    if (0)
       _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);
-   
+
+   /* Accumulate the list of enabled arrays. */
    while (tmp) {
       GLuint i = _mesa_ffsll(tmp)-1;
       struct brw_vertex_element *input = &brw->vb.inputs[i];
 
       tmp &= ~(1<<i);
       enabled[nr_enabled++] = input;
+   }
+
+   /* XXX: In the rare cases where this happens we fallback all
+    * the way to software rasterization, although a tnl fallback
+    * would be sufficient.  I don't know of *any* real world
+    * cases with > 17 vertex attributes enabled, so it probably
+    * isn't an issue at this point.
+    */
+   if (nr_enabled >= BRW_VEP_MAX) {
+      intel->Fallback = 1;
+      return;
+   }
+
+   for (i = 0; i < nr_enabled; i++) {
+      struct brw_vertex_element *input = enabled[i];
 
-      input->index = i;
       input->element_size = get_size(input->glarray->Type) * input->glarray->Size;
       input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1;
 
-      if (!input->glarray->BufferObj->Name) {
+      if (input->glarray->BufferObj->Name != 0) {
+	 struct intel_buffer_object *intel_buffer =
+	    intel_buffer_object(input->glarray->BufferObj);
+
+	 /* Named buffer object: Just reference its contents directly. */
+	 dri_bo_unreference(input->bo);
+	 input->bo = intel_bufferobj_buffer(intel, intel_buffer,
+					    INTEL_READ);
+	 dri_bo_reference(input->bo);
+	 input->offset = (unsigned long)input->glarray->Ptr;
+	 input->stride = input->glarray->StrideB;
+      } else {
+	 if (input->bo != NULL) {
+	    /* Already-uploaded vertex data is present from a previous
+	     * prepare_vertices, but we had to re-validate state due to
+	     * check_aperture failing and a new batch being produced.
+	     */
+	    continue;
+	 }
+
+	 /* Queue the buffer object up to be uploaded in the next pass,
+	  * when we've decided if we're doing interleaved or not.
+	  */
 	 if (i == 0) {
 	    /* Position array not properly enabled:
 	     */
-	    if (input->glarray->StrideB == 0)
-	       return GL_FALSE;
+            if (input->glarray->StrideB == 0) {
+               intel->Fallback = 1;
+               return;
+            }
 
 	    interleave = input->glarray->StrideB;
 	    ptr = input->glarray->Ptr;
 	 }
 	 else if (interleave != input->glarray->StrideB ||
-		  (const char *)input->glarray->Ptr - (const char *)ptr < 0 ||
-		  (const char *)input->glarray->Ptr - (const char *)ptr > interleave) {
+		  (const unsigned char *)input->glarray->Ptr - ptr < 0 ||
+		  (const unsigned char *)input->glarray->Ptr - ptr > interleave)
+	 {
 	    interleave = 0;
 	 }
 
@@ -451,131 +402,140 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
       }
    }
 
-   /* Upload interleaved arrays if all uploads are interleaved
-    */
-   if (nr_uploads > 1 && 
-       interleave && 
-       interleave <= 256) {
-      struct brw_vertex_element *input0 = upload[0];
-
-      input0->glarray = copy_array_to_vbo_array(brw, 0,
-						input0->glarray, 
-						interleave,
-						input0->count);
+   /* Handle any arrays to be uploaded. */
+   if (nr_uploads > 1 && interleave && interleave <= 256) {
+      /* All uploads are interleaved, so upload the arrays together as
+       * interleaved.  First, upload the contents and set up upload[0].
+       */
+      copy_array_to_vbo_array(brw, upload[0], interleave);
 
       for (i = 1; i < nr_uploads; i++) {
-	 upload[i]->glarray = interleaved_vbo_array(brw,
-						    i,
-						    input0->glarray,
-						    upload[i]->glarray,
-						    ptr);
+	 /* Then, just point upload[i] at upload[0]'s buffer. */
+	 upload[i]->stride = interleave;
+	 upload[i]->offset = upload[0]->offset +
+	    ((const unsigned char *)upload[i]->glarray->Ptr - ptr);
+	 upload[i]->bo = upload[0]->bo;
+	 dri_bo_reference(upload[i]->bo);
       }
    }
    else {
+      /* Upload non-interleaved arrays */
       for (i = 0; i < nr_uploads; i++) {
-	 struct brw_vertex_element *input = upload[i];
-
-	 input->glarray = copy_array_to_vbo_array(brw, i, 
-						  input->glarray,
-						  input->element_size,
-						  input->count);
-
+          copy_array_to_vbo_array(brw, upload[i], upload[i]->element_size);
       }
    }
 
-   /* XXX: In the rare cases where this happens we fallback all
-    * the way to software rasterization, although a tnl fallback
-    * would be sufficient.  I don't know of *any* real world
-    * cases with > 17 vertex attributes enabled, so it probably
-    * isn't an issue at this point.
-    */
-   if (nr_enabled >= BRW_VEP_MAX)
-	 return GL_FALSE;
+   brw_prepare_query_begin(brw);
 
-   /* This still defines a hardware VB for each input, even if they
-    * are interleaved or from the same VBO.  TBD if this makes a
-    * performance difference.
-    */
    for (i = 0; i < nr_enabled; i++) {
       struct brw_vertex_element *input = enabled[i];
 
-      input->vep = &vep.ve[i];
-      input->vep->ve0.src_format = get_surface_type(input->glarray->Type, 
-						    input->glarray->Size,
-						    input->glarray->Normalized);
-      input->vep->ve0.valid = 1;
-      input->vep->ve1.dst_offset = (i) * 4;
-      input->vep->ve1.vfcomponent3 = BRW_VFCOMPONENT_STORE_SRC;
-      input->vep->ve1.vfcomponent2 = BRW_VFCOMPONENT_STORE_SRC;
-      input->vep->ve1.vfcomponent1 = BRW_VFCOMPONENT_STORE_SRC;
-      input->vep->ve1.vfcomponent0 = BRW_VFCOMPONENT_STORE_SRC;
+      brw_add_validated_bo(brw, input->bo);
+   }
+}
 
-      switch (input->glarray->Size) {
-      case 0: input->vep->ve1.vfcomponent0 = BRW_VFCOMPONENT_STORE_0;
-      case 1: input->vep->ve1.vfcomponent1 = BRW_VFCOMPONENT_STORE_0;
-      case 2: input->vep->ve1.vfcomponent2 = BRW_VFCOMPONENT_STORE_0;
-      case 3: input->vep->ve1.vfcomponent3 = BRW_VFCOMPONENT_STORE_1_FLT;
-	 break;
-      }
+static void brw_emit_vertices(struct brw_context *brw)
+{
+   GLcontext *ctx = &brw->intel.ctx;
+   struct intel_context *intel = intel_context(ctx);
+   GLuint tmp = brw->vs.prog_data->inputs_read;
+   struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
+   GLuint i;
+   GLuint nr_enabled = 0;
 
-      input->vep->ve0.vertex_buffer_index = i;
-      input->vep->ve0.src_offset = 0;
+  /* Accumulate the list of enabled arrays. */
+   while (tmp) {
+      i = _mesa_ffsll(tmp)-1;
+      struct brw_vertex_element *input = &brw->vb.inputs[i];
 
-      vbp.vb[i].vb0.bits.pitch = input->glarray->StrideB;
-      vbp.vb[i].vb0.bits.pad = 0;
-      vbp.vb[i].vb0.bits.access_type = BRW_VERTEXBUFFER_ACCESS_VERTEXDATA;
-      vbp.vb[i].vb0.bits.vb_index = i;
-      vbp.vb[i].offset = (GLuint)input->glarray->Ptr;
-      vbp.vb[i].buffer = array_buffer(input->glarray);
-      vbp.vb[i].max_index = max_index;
+      tmp &= ~(1<<i);
+      enabled[nr_enabled++] = input;
    }
 
+   brw_emit_query_begin(brw);
 
-
-   /* Now emit VB and VEP state packets:
+   /* Now emit VB and VEP state packets.
+    *
+    * This still defines a hardware VB for each input, even if they
+    * are interleaved or from the same VBO.  TBD if this makes a
+    * performance difference.
     */
-   vbp.header.bits.length = (1 + nr_enabled * 4) - 2;
-   vbp.header.bits.opcode = CMD_VERTEX_BUFFER;
+   BEGIN_BATCH(1 + nr_enabled * 4, IGNORE_CLIPRECTS);
+   OUT_BATCH((CMD_VERTEX_BUFFER << 16) |
+	     ((1 + nr_enabled * 4) - 2));
 
-   BEGIN_BATCH(vbp.header.bits.length+2, 0);
-   OUT_BATCH( vbp.header.dword );
-   
    for (i = 0; i < nr_enabled; i++) {
-      OUT_BATCH( vbp.vb[i].vb0.dword );
-      OUT_BATCH( bmBufferOffset(&brw->intel, vbp.vb[i].buffer) + vbp.vb[i].offset);
-      OUT_BATCH( vbp.vb[i].max_index );
-      OUT_BATCH( vbp.vb[i].instance_data_step_rate );
+      struct brw_vertex_element *input = enabled[i];
+
+      OUT_BATCH((i << BRW_VB0_INDEX_SHIFT) |
+		BRW_VB0_ACCESS_VERTEXDATA |
+		(input->stride << BRW_VB0_PITCH_SHIFT));
+      OUT_RELOC(input->bo,
+		I915_GEM_DOMAIN_VERTEX, 0,
+		input->offset);
+      OUT_BATCH(brw->vb.max_index);
+      OUT_BATCH(0); /* Instance data step rate */
    }
    ADVANCE_BATCH();
 
-   vep.header.length = (1 + nr_enabled * sizeof(vep.ve[0])/4) - 2;
-   vep.header.opcode = CMD_VERTEX_ELEMENT;
-   brw_cached_batch_struct(brw, &vep, 4 + nr_enabled * sizeof(vep.ve[0]));
-
-   return GL_TRUE;
-}
+   BEGIN_BATCH(1 + nr_enabled * 2, IGNORE_CLIPRECTS);
+   OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + nr_enabled * 2) - 2));
+   for (i = 0; i < nr_enabled; i++) {
+      struct brw_vertex_element *input = enabled[i];
+      uint32_t format = get_surface_type(input->glarray->Type,
+					 input->glarray->Size,
+					 input->glarray->Normalized);
+      uint32_t comp0 = BRW_VE1_COMPONENT_STORE_SRC;
+      uint32_t comp1 = BRW_VE1_COMPONENT_STORE_SRC;
+      uint32_t comp2 = BRW_VE1_COMPONENT_STORE_SRC;
+      uint32_t comp3 = BRW_VE1_COMPONENT_STORE_SRC;
 
+      switch (input->glarray->Size) {
+      case 0: comp0 = BRW_VE1_COMPONENT_STORE_0;
+      case 1: comp1 = BRW_VE1_COMPONENT_STORE_0;
+      case 2: comp2 = BRW_VE1_COMPONENT_STORE_0;
+      case 3: comp3 = BRW_VE1_COMPONENT_STORE_1_FLT;
+	 break;
+      }
 
-static GLuint element_size( GLenum type )
-{
-   switch(type) {
-   case GL_UNSIGNED_INT: return 4;
-   case GL_UNSIGNED_SHORT: return 2;
-   case GL_UNSIGNED_BYTE: return 1;
-   default: assert(0); return 0;
+      OUT_BATCH((i << BRW_VE0_INDEX_SHIFT) |
+		BRW_VE0_VALID |
+		(format << BRW_VE0_FORMAT_SHIFT) |
+		(0 << BRW_VE0_SRC_OFFSET_SHIFT));
+      OUT_BATCH((comp0 << BRW_VE1_COMPONENT_0_SHIFT) |
+		(comp1 << BRW_VE1_COMPONENT_1_SHIFT) |
+		(comp2 << BRW_VE1_COMPONENT_2_SHIFT) |
+		(comp3 << BRW_VE1_COMPONENT_3_SHIFT) |
+		((i * 4) << BRW_VE1_DST_OFFSET_SHIFT));
    }
+   ADVANCE_BATCH();
 }
 
+const struct brw_tracked_state brw_vertices = {
+   .dirty = {
+      .mesa = 0,
+      .brw = BRW_NEW_BATCH | BRW_NEW_VERTICES,
+      .cache = 0,
+   },
+   .prepare = brw_prepare_vertices,
+   .emit = brw_emit_vertices,
+};
 
-
-void brw_upload_indices( struct brw_context *brw,
-			 const struct _mesa_index_buffer *index_buffer )
+static void brw_prepare_indices(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = &brw->intel;
-   GLuint ib_size = get_size(index_buffer->type) * index_buffer->count;
-   struct gl_buffer_object *bufferobj = index_buffer->obj;
-   GLuint offset = (GLuint)index_buffer->ptr;
+   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
+   GLuint ib_size;
+   dri_bo *bo = NULL;
+   struct gl_buffer_object *bufferobj;
+   GLuint offset;
+
+   if (index_buffer == NULL)
+      return;
+
+   ib_size = get_size(index_buffer->type) * index_buffer->count;
+   bufferobj = index_buffer->obj;;
 
    /* Turn into a proper VBO:
     */
@@ -583,23 +543,58 @@ void brw_upload_indices( struct brw_context *brw,
      
       /* Get new bufferobj, offset:
        */
-      get_space(brw, ib_size, &bufferobj, &offset);
+      get_space(brw, ib_size, &bo, &offset);
 
       /* Straight upload
        */
-      ctx->Driver.BufferSubData( ctx,
-				 GL_ELEMENT_ARRAY_BUFFER_ARB,
-				 offset, 
-				 ib_size,
-				 index_buffer->ptr,
-				 bufferobj);
+      dri_bo_subdata(bo, offset, ib_size, index_buffer->ptr);
+   } else {
+      offset = (GLuint)index_buffer->ptr;
+
+      /* If the index buffer isn't aligned to its element size, we have to
+       * rebase it into a temporary.
+       */
+       if ((get_size(index_buffer->type) - 1) & offset) {
+           GLubyte *map = ctx->Driver.MapBuffer(ctx,
+                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                                GL_DYNAMIC_DRAW_ARB,
+                                                bufferobj);
+           map += offset;
+
+	   get_space(brw, ib_size, &bo, &offset);
+
+	   dri_bo_subdata(bo, offset, ib_size, map);
+
+           ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, bufferobj);
+       } else {
+	  bo = intel_bufferobj_buffer(intel, intel_buffer_object(bufferobj),
+				      INTEL_READ);
+	  dri_bo_reference(bo);
+       }
    }
 
+   dri_bo_unreference(brw->ib.bo);
+   brw->ib.bo = bo;
+   brw->ib.offset = offset;
+
+   brw_add_validated_bo(brw, brw->ib.bo);
+}
+
+static void brw_emit_indices(struct brw_context *brw)
+{
+   struct intel_context *intel = &brw->intel;
+   const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
+   GLuint ib_size;
+
+   if (index_buffer == NULL)
+      return;
+
+   ib_size = get_size(index_buffer->type) * index_buffer->count;
+
    /* Emit the indexbuffer packet:
     */
    {
       struct brw_indexbuffer ib;
-      struct buffer *buffer = intel_bufferobj_buffer(intel_buffer_object(bufferobj));
 
       memset(&ib, 0, sizeof(ib));
    
@@ -609,11 +604,25 @@ void brw_upload_indices( struct brw_context *brw,
       ib.header.bits.cut_index_enable = 0;
    
 
-      BEGIN_BATCH(4, 0);
+      BEGIN_BATCH(4, IGNORE_CLIPRECTS);
       OUT_BATCH( ib.header.dword );
-      OUT_BATCH( bmBufferOffset(intel, buffer) + offset );
-      OUT_BATCH( bmBufferOffset(intel, buffer) + offset + ib_size );
+      OUT_RELOC(brw->ib.bo,
+		I915_GEM_DOMAIN_VERTEX, 0,
+		brw->ib.offset);
+      OUT_RELOC(brw->ib.bo,
+		I915_GEM_DOMAIN_VERTEX, 0,
+		brw->ib.offset + ib_size);
       OUT_BATCH( 0 );
       ADVANCE_BATCH();
    }
 }
+
+const struct brw_tracked_state brw_indices = {
+   .dirty = {
+      .mesa = 0,
+      .brw = BRW_NEW_BATCH | BRW_NEW_INDICES,
+      .cache = 0,
+   },
+   .prepare = brw_prepare_indices,
+   .emit = brw_emit_indices,
+};
diff --git a/src/mesa/drivers/dri/i965/brw_eu.c b/src/mesa/drivers/dri/i965/brw_eu.c
index d1244befd78..b3ae4eef334 100644
--- a/src/mesa/drivers/dri/i965/brw_eu.c
+++ b/src/mesa/drivers/dri/i965/brw_eu.c
@@ -101,8 +101,9 @@ void brw_pop_insn_state( struct brw_compile *p )
 
 /***********************************************************************
  */
-void brw_init_compile( struct brw_compile *p )
+void brw_init_compile( struct brw_context *brw, struct brw_compile *p )
 {
+   p->brw = brw;
    p->nr_insn = 0;
    p->current = p->stack;
    memset(p->current, 0, sizeof(p->current[0]));
diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h
index 52f89d577ca..49b422ee2ff 100644
--- a/src/mesa/drivers/dri/i965/brw_eu.h
+++ b/src/mesa/drivers/dri/i965/brw_eu.h
@@ -65,7 +65,7 @@ struct brw_reg
    GLuint abs:1;		/* source only */
    GLuint vstride:4;		/* source only */
    GLuint width:3;		/* src only, align1 only */
-   GLuint hstride:2;   		/* src only, align1 only */
+   GLuint hstride:2;   		/* align1 only */
    GLuint address_mode:1;	/* relative addressing, hopefully! */
    GLuint pad0:1;
 
@@ -105,11 +105,12 @@ struct brw_compile {
 
    GLuint flag_value;
    GLboolean single_program_flow;
+   struct brw_context *brw;
 };
 
 
 
-static __inline int type_sz( GLuint type )
+static INLINE int type_sz( GLuint type )
 {
    switch( type ) {
    case BRW_REGISTER_TYPE_UD:
@@ -128,7 +129,7 @@ static __inline int type_sz( GLuint type )
    }
 }
 
-static __inline struct brw_reg brw_reg( GLuint file,
+static INLINE struct brw_reg brw_reg( GLuint file,
 					GLuint nr,
 					GLuint subnr,
 					GLuint type,
@@ -165,7 +166,7 @@ static __inline struct brw_reg brw_reg( GLuint file,
    return reg;
 }
 
-static __inline struct brw_reg brw_vec16_reg( GLuint file,
+static INLINE struct brw_reg brw_vec16_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
 {
@@ -180,7 +181,7 @@ static __inline struct brw_reg brw_vec16_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
-static __inline struct brw_reg brw_vec8_reg( GLuint file,
+static INLINE struct brw_reg brw_vec8_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
 {
@@ -196,7 +197,7 @@ static __inline struct brw_reg brw_vec8_reg( GLuint file,
 }
 
 
-static __inline struct brw_reg brw_vec4_reg( GLuint file,
+static INLINE struct brw_reg brw_vec4_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
 {
@@ -212,7 +213,7 @@ static __inline struct brw_reg brw_vec4_reg( GLuint file,
 }
 
 
-static __inline struct brw_reg brw_vec2_reg( GLuint file,
+static INLINE struct brw_reg brw_vec2_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
 {
@@ -227,7 +228,7 @@ static __inline struct brw_reg brw_vec2_reg( GLuint file,
 		  WRITEMASK_XY);
 }
 
-static __inline struct brw_reg brw_vec1_reg( GLuint file,
+static INLINE struct brw_reg brw_vec1_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
 {
@@ -243,14 +244,14 @@ static __inline struct brw_reg brw_vec1_reg( GLuint file,
 }
 
 
-static __inline struct brw_reg retype( struct brw_reg reg,
+static INLINE struct brw_reg retype( struct brw_reg reg,
 				       GLuint type )
 {
    reg.type = type;
    return reg;
 }
 
-static __inline struct brw_reg suboffset( struct brw_reg reg,
+static INLINE struct brw_reg suboffset( struct brw_reg reg,
 					  GLuint delta )
 {   
    reg.subnr += delta * type_sz(reg.type);
@@ -258,7 +259,7 @@ static __inline struct brw_reg suboffset( struct brw_reg reg,
 }
 
 
-static __inline struct brw_reg offset( struct brw_reg reg,
+static INLINE struct brw_reg offset( struct brw_reg reg,
 				       GLuint delta )
 {
    reg.nr += delta;
@@ -266,7 +267,7 @@ static __inline struct brw_reg offset( struct brw_reg reg,
 }
 
 
-static __inline struct brw_reg byte_offset( struct brw_reg reg,
+static INLINE struct brw_reg byte_offset( struct brw_reg reg,
 					    GLuint bytes )
 {
    GLuint newoffset = reg.nr * REG_SIZE + reg.subnr + bytes;
@@ -276,28 +277,28 @@ static __inline struct brw_reg byte_offset( struct brw_reg reg,
 }
    
 
-static __inline struct brw_reg brw_uw16_reg( GLuint file,
+static INLINE struct brw_reg brw_uw16_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
 {
    return suboffset(retype(brw_vec16_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
-static __inline struct brw_reg brw_uw8_reg( GLuint file,
+static INLINE struct brw_reg brw_uw8_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
 {
    return suboffset(retype(brw_vec8_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
-static __inline struct brw_reg brw_uw1_reg( GLuint file,
+static INLINE struct brw_reg brw_uw1_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
 {
    return suboffset(retype(brw_vec1_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
-static __inline struct brw_reg brw_imm_reg( GLuint type )
+static INLINE struct brw_reg brw_imm_reg( GLuint type )
 {
    return brw_reg( BRW_IMMEDIATE_VALUE,
 		   0,
@@ -310,38 +311,38 @@ static __inline struct brw_reg brw_imm_reg( GLuint type )
 		   0);      
 }
 
-static __inline struct brw_reg brw_imm_f( GLfloat f )
+static INLINE struct brw_reg brw_imm_f( GLfloat f )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_F);
    imm.dw1.f = f;
    return imm;
 }
 
-static __inline struct brw_reg brw_imm_d( GLint d )
+static INLINE struct brw_reg brw_imm_d( GLint d )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_D);
    imm.dw1.d = d;
    return imm;
 }
 
-static __inline struct brw_reg brw_imm_ud( GLuint ud )
+static INLINE struct brw_reg brw_imm_ud( GLuint ud )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UD);
    imm.dw1.ud = ud;
    return imm;
 }
 
-static __inline struct brw_reg brw_imm_uw( GLushort uw )
+static INLINE struct brw_reg brw_imm_uw( GLushort uw )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UW);
-   imm.dw1.ud = uw;
+   imm.dw1.ud = uw | (uw << 16);
    return imm;
 }
 
-static __inline struct brw_reg brw_imm_w( GLshort w )
+static INLINE struct brw_reg brw_imm_w( GLshort w )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_W);
-   imm.dw1.d = w;
+   imm.dw1.d = w | (w << 16);
    return imm;
 }
 
@@ -351,7 +352,7 @@ static __inline struct brw_reg brw_imm_w( GLshort w )
 
 /* Vector of eight signed half-byte values: 
  */
-static __inline struct brw_reg brw_imm_v( GLuint v )
+static INLINE struct brw_reg brw_imm_v( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_V);
    imm.vstride = BRW_VERTICAL_STRIDE_0;
@@ -363,7 +364,7 @@ static __inline struct brw_reg brw_imm_v( GLuint v )
 
 /* Vector of four 8-bit float values:
  */
-static __inline struct brw_reg brw_imm_vf( GLuint v )
+static INLINE struct brw_reg brw_imm_vf( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_VF);
    imm.vstride = BRW_VERTICAL_STRIDE_0;
@@ -377,7 +378,7 @@ static __inline struct brw_reg brw_imm_vf( GLuint v )
 #define VF_ONE  0x30
 #define VF_NEG  (1<<7)
 
-static __inline struct brw_reg brw_imm_vf4( GLuint v0, 
+static INLINE struct brw_reg brw_imm_vf4( GLuint v0, 
 					    GLuint v1, 
 					    GLuint v2,
 					    GLuint v3)
@@ -394,51 +395,57 @@ static __inline struct brw_reg brw_imm_vf4( GLuint v0,
 }
 
 
-static __inline struct brw_reg brw_address( struct brw_reg reg )
+static INLINE struct brw_reg brw_address( struct brw_reg reg )
 {
    return brw_imm_uw(reg.nr * REG_SIZE + reg.subnr);
 }
 
 
-static __inline struct brw_reg brw_vec1_grf( GLuint nr,
+static INLINE struct brw_reg brw_vec1_grf( GLuint nr,
 					       GLuint subnr )
 {
    return brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static __inline struct brw_reg brw_vec8_grf( GLuint nr,
+static INLINE struct brw_reg brw_vec8_grf( GLuint nr,
 					     GLuint subnr )
 {
    return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static __inline struct brw_reg brw_vec4_grf( GLuint nr,
+static INLINE struct brw_reg brw_vec4_grf( GLuint nr,
 					     GLuint subnr )
 {
    return brw_vec4_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
 
-static __inline struct brw_reg brw_vec2_grf( GLuint nr,
+static INLINE struct brw_reg brw_vec2_grf( GLuint nr,
 					     GLuint subnr )
 {
    return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static __inline struct brw_reg brw_uw8_grf( GLuint nr,
+static INLINE struct brw_reg brw_uw8_grf( GLuint nr,
 					    GLuint subnr )
 {
    return brw_uw8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static __inline struct brw_reg brw_null_reg( void )
+static INLINE struct brw_reg brw_uw16_grf( GLuint nr,
+					    GLuint subnr )
+{
+   return brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
+}
+
+static INLINE struct brw_reg brw_null_reg( void )
 {
    return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
 		       BRW_ARF_NULL, 
 		       0);
 }
 
-static __inline struct brw_reg brw_address_reg( GLuint subnr )
+static INLINE struct brw_reg brw_address_reg( GLuint subnr )
 {
    return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
 		      BRW_ARF_ADDRESS, 
@@ -449,7 +456,7 @@ static __inline struct brw_reg brw_address_reg( GLuint subnr )
  * aren't xyzw.  This goes against the convention for other scalar
  * regs:
  */
-static __inline struct brw_reg brw_ip_reg( void )
+static INLINE struct brw_reg brw_ip_reg( void )
 {
    return brw_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
 		  BRW_ARF_IP, 
@@ -462,7 +469,7 @@ static __inline struct brw_reg brw_ip_reg( void )
 		  WRITEMASK_XYZW); /* NOTE! */
 }
 
-static __inline struct brw_reg brw_acc_reg( void )
+static INLINE struct brw_reg brw_acc_reg( void )
 {
    return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
 		       BRW_ARF_ACCUMULATOR, 
@@ -470,7 +477,7 @@ static __inline struct brw_reg brw_acc_reg( void )
 }
 
 
-static __inline struct brw_reg brw_flag_reg( void )
+static INLINE struct brw_reg brw_flag_reg( void )
 {
    return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE,
 		      BRW_ARF_FLAG,
@@ -478,14 +485,14 @@ static __inline struct brw_reg brw_flag_reg( void )
 }
 
 
-static __inline struct brw_reg brw_mask_reg( GLuint subnr )
+static INLINE struct brw_reg brw_mask_reg( GLuint subnr )
 {
    return brw_uw1_reg(BRW_ARCHITECTURE_REGISTER_FILE,
 		      BRW_ARF_MASK,
 		      subnr);
 }
 
-static __inline struct brw_reg brw_message_reg( GLuint nr )
+static INLINE struct brw_reg brw_message_reg( GLuint nr )
 {
    return brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE,
 		       nr,
@@ -498,7 +505,7 @@ static __inline struct brw_reg brw_message_reg( GLuint nr )
 /* This is almost always called with a numeric constant argument, so
  * make things easy to evaluate at compile time:
  */
-static __inline GLuint cvt( GLuint val )
+static INLINE GLuint cvt( GLuint val )
 {
    switch (val) {
    case 0: return 0;
@@ -512,7 +519,7 @@ static __inline GLuint cvt( GLuint val )
    return 0;
 }
 
-static __inline struct brw_reg stride( struct brw_reg reg,
+static INLINE struct brw_reg stride( struct brw_reg reg,
 				       GLuint vstride,
 				       GLuint width,
 				       GLuint hstride )
@@ -524,43 +531,43 @@ static __inline struct brw_reg stride( struct brw_reg reg,
    return reg;
 }
 
-static __inline struct brw_reg vec16( struct brw_reg reg )
+static INLINE struct brw_reg vec16( struct brw_reg reg )
 {
    return stride(reg, 16,16,1);
 }
 
-static __inline struct brw_reg vec8( struct brw_reg reg )
+static INLINE struct brw_reg vec8( struct brw_reg reg )
 {
    return stride(reg, 8,8,1);
 }
 
-static __inline struct brw_reg vec4( struct brw_reg reg )
+static INLINE struct brw_reg vec4( struct brw_reg reg )
 {
    return stride(reg, 4,4,1);
 }
 
-static __inline struct brw_reg vec2( struct brw_reg reg )
+static INLINE struct brw_reg vec2( struct brw_reg reg )
 {
    return stride(reg, 2,2,1);
 }
 
-static __inline struct brw_reg vec1( struct brw_reg reg )
+static INLINE struct brw_reg vec1( struct brw_reg reg )
 {
    return stride(reg, 0,1,0);
 }
 
-static __inline struct brw_reg get_element( struct brw_reg reg, GLuint elt )
+static INLINE struct brw_reg get_element( struct brw_reg reg, GLuint elt )
 {
    return vec1(suboffset(reg, elt));
 }
 
-static __inline struct brw_reg get_element_ud( struct brw_reg reg, GLuint elt )
+static INLINE struct brw_reg get_element_ud( struct brw_reg reg, GLuint elt )
 {
    return vec1(suboffset(retype(reg, BRW_REGISTER_TYPE_UD), elt));
 }
 
 
-static __inline struct brw_reg brw_swizzle( struct brw_reg reg,
+static INLINE struct brw_reg brw_swizzle( struct brw_reg reg,
 					    GLuint x,
 					    GLuint y, 
 					    GLuint z,
@@ -574,33 +581,33 @@ static __inline struct brw_reg brw_swizzle( struct brw_reg reg,
 }
 
 
-static __inline struct brw_reg brw_swizzle1( struct brw_reg reg,
+static INLINE struct brw_reg brw_swizzle1( struct brw_reg reg,
 					     GLuint x )
 {
    return brw_swizzle(reg, x, x, x, x);
 }
 
-static __inline struct brw_reg brw_writemask( struct brw_reg reg,
+static INLINE struct brw_reg brw_writemask( struct brw_reg reg,
 					      GLuint mask )
 {
    reg.dw1.bits.writemask &= mask;
    return reg;
 }
 
-static __inline struct brw_reg brw_set_writemask( struct brw_reg reg,
+static INLINE struct brw_reg brw_set_writemask( struct brw_reg reg,
 						  GLuint mask )
 {
    reg.dw1.bits.writemask = mask;
    return reg;
 }
 
-static __inline struct brw_reg negate( struct brw_reg reg )
+static INLINE struct brw_reg negate( struct brw_reg reg )
 {
    reg.negate ^= 1;
    return reg;
 }
 
-static __inline struct brw_reg brw_abs( struct brw_reg reg )
+static INLINE struct brw_reg brw_abs( struct brw_reg reg )
 {
    reg.abs = 1;
    return reg;
@@ -608,7 +615,7 @@ static __inline struct brw_reg brw_abs( struct brw_reg reg )
 
 /***********************************************************************
  */
-static __inline struct brw_reg brw_vec4_indirect( GLuint subnr,
+static INLINE struct brw_reg brw_vec4_indirect( GLuint subnr,
 						  GLint offset )
 {
    struct brw_reg reg =  brw_vec4_grf(0, 0);
@@ -618,7 +625,7 @@ static __inline struct brw_reg brw_vec4_indirect( GLuint subnr,
    return reg;
 }
 
-static __inline struct brw_reg brw_vec1_indirect( GLuint subnr,
+static INLINE struct brw_reg brw_vec1_indirect( GLuint subnr,
 						  GLint offset )
 {
    struct brw_reg reg =  brw_vec1_grf(0, 0);
@@ -628,38 +635,48 @@ static __inline struct brw_reg brw_vec1_indirect( GLuint subnr,
    return reg;
 }
 
-static __inline struct brw_reg deref_4f(struct brw_indirect ptr, GLint offset)
+static INLINE struct brw_reg deref_4f(struct brw_indirect ptr, GLint offset)
 {
    return brw_vec4_indirect(ptr.addr_subnr, ptr.addr_offset + offset);
 }
 
-static __inline struct brw_reg deref_1f(struct brw_indirect ptr, GLint offset)
+static INLINE struct brw_reg deref_1f(struct brw_indirect ptr, GLint offset)
 {
    return brw_vec1_indirect(ptr.addr_subnr, ptr.addr_offset + offset);
 }
 
-static __inline struct brw_reg deref_4b(struct brw_indirect ptr, GLint offset)
+static INLINE struct brw_reg deref_4b(struct brw_indirect ptr, GLint offset)
 {
    return retype(deref_4f(ptr, offset), BRW_REGISTER_TYPE_B);
 }
 
-static __inline struct brw_reg deref_1uw(struct brw_indirect ptr, GLint offset)
+static INLINE struct brw_reg deref_1uw(struct brw_indirect ptr, GLint offset)
 {
    return retype(deref_1f(ptr, offset), BRW_REGISTER_TYPE_UW);
 }
 
-static __inline struct brw_reg get_addr_reg(struct brw_indirect ptr)
+static INLINE struct brw_reg deref_1d(struct brw_indirect ptr, GLint offset)
+{
+   return retype(deref_1f(ptr, offset), BRW_REGISTER_TYPE_D);
+}
+
+static INLINE struct brw_reg deref_1ud(struct brw_indirect ptr, GLint offset)
+{
+   return retype(deref_1f(ptr, offset), BRW_REGISTER_TYPE_UD);
+}
+
+static INLINE struct brw_reg get_addr_reg(struct brw_indirect ptr)
 {
    return brw_address_reg(ptr.addr_subnr);
 }
 
-static __inline struct brw_indirect brw_indirect_offset( struct brw_indirect ptr, GLint offset )
+static INLINE struct brw_indirect brw_indirect_offset( struct brw_indirect ptr, GLint offset )
 {
    ptr.addr_offset += offset;
    return ptr;
 }
 
-static __inline struct brw_indirect brw_indirect( GLuint addr_subnr, GLint offset )
+static INLINE struct brw_indirect brw_indirect( GLuint addr_subnr, GLint offset )
 {
    struct brw_indirect ptr;
    ptr.addr_subnr = addr_subnr;
@@ -668,7 +685,10 @@ static __inline struct brw_indirect brw_indirect( GLuint addr_subnr, GLint offse
    return ptr;
 }
 
-
+static INLINE struct brw_instruction *current_insn( struct brw_compile *p)
+{
+	return &p->store[p->nr_insn];
+}
 
 void brw_pop_insn_state( struct brw_compile *p );
 void brw_push_insn_state( struct brw_compile *p );
@@ -680,7 +700,7 @@ void brw_set_predicate_control_flag_value( struct brw_compile *p, GLuint value )
 void brw_set_predicate_control( struct brw_compile *p, GLuint pc );
 void brw_set_conditionalmod( struct brw_compile *p, GLuint conditional );
 
-void brw_init_compile( struct brw_compile *p );
+void brw_init_compile( struct brw_context *, struct brw_compile *p );
 const GLuint *brw_get_program( struct brw_compile *p, GLuint *sz );
 
 
@@ -808,9 +828,11 @@ void brw_ENDIF(struct brw_compile *p,
 struct brw_instruction *brw_DO(struct brw_compile *p,
 			       GLuint execute_size);
 
-void brw_WHILE(struct brw_compile *p, 
+struct brw_instruction *brw_WHILE(struct brw_compile *p, 
 	       struct brw_instruction *patch_insn);
 
+struct brw_instruction *brw_BREAK(struct brw_compile *p);
+struct brw_instruction *brw_CONT(struct brw_compile *p);
 /* Forward jumps:
  */
 void brw_land_fwd_jump(struct brw_compile *p, 
@@ -860,5 +882,6 @@ void brw_math_invert( struct brw_compile *p,
 		      struct brw_reg dst,
 		      struct brw_reg src);
 
-
+void brw_set_src1( struct brw_instruction *insn,
+                          struct brw_reg reg );
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_eu_debug.c b/src/mesa/drivers/dri/i965/brw_eu_debug.c
index 2dff1ad2244..91dbbd5af62 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_debug.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_debug.c
@@ -30,9 +30,9 @@
   */
     
 
-#include "mtypes.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
 #include "brw_eu.h"
-#include "imports.h"
 
 void brw_print_reg( struct brw_reg hwreg )
 {
diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c
index 9992b47d8ae..ce4cf46cfa6 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c
@@ -64,7 +64,9 @@ static void brw_set_dest( struct brw_instruction *insn,
 
       if (insn->header.access_mode == BRW_ALIGN_1) {
 	 insn->bits1.da1.dest_subreg_nr = dest.subnr;
-	 insn->bits1.da1.dest_horiz_stride = BRW_HORIZONTAL_STRIDE_1;
+	 if (dest.hstride == BRW_HORIZONTAL_STRIDE_0)
+	    dest.hstride = BRW_HORIZONTAL_STRIDE_1;
+	 insn->bits1.da1.dest_horiz_stride = dest.hstride;
       }
       else {
 	 insn->bits1.da16.dest_subreg_nr = dest.subnr / 16;
@@ -78,7 +80,9 @@ static void brw_set_dest( struct brw_instruction *insn,
        */
       if (insn->header.access_mode == BRW_ALIGN_1) {
 	 insn->bits1.ia1.dest_indirect_offset = dest.dw1.bits.indirect_offset;
-	 insn->bits1.ia1.dest_horiz_stride = BRW_HORIZONTAL_STRIDE_1;
+	 if (dest.hstride == BRW_HORIZONTAL_STRIDE_0)
+	    dest.hstride = BRW_HORIZONTAL_STRIDE_1;
+	 insn->bits1.ia1.dest_horiz_stride = dest.hstride;
       }
       else {
 	 insn->bits1.ia16.dest_indirect_offset = dest.dw1.bits.indirect_offset;
@@ -164,7 +168,7 @@ static void brw_set_src0( struct brw_instruction *insn,
 }
 
 
-static void brw_set_src1( struct brw_instruction *insn,
+void brw_set_src1( struct brw_instruction *insn,
 			  struct brw_reg reg )
 {
    assert(reg.file != BRW_MESSAGE_REGISTER_FILE);
@@ -186,7 +190,7 @@ static void brw_set_src1( struct brw_instruction *insn,
        * in the future:
        */
       assert (reg.address_mode == BRW_ADDRESS_DIRECT);
-      assert (reg.file == BRW_GENERAL_REGISTER_FILE);
+      //assert (reg.file == BRW_GENERAL_REGISTER_FILE);
 
       if (insn->header.access_mode == BRW_ALIGN_1) {
 	 insn->bits3.da1.src1_subreg_nr = reg.subnr;
@@ -318,7 +322,8 @@ static void brw_set_dp_read_message( struct brw_instruction *insn,
    insn->bits3.dp_read.end_of_thread = end_of_thread;
 }
 
-static void brw_set_sampler_message( struct brw_instruction *insn,
+static void brw_set_sampler_message(struct brw_context *brw,
+                 struct brw_instruction *insn,
 				     GLuint binding_table_index,
 				     GLuint sampler,
 				     GLuint msg_type,
@@ -328,14 +333,24 @@ static void brw_set_sampler_message( struct brw_instruction *insn,
 {
    brw_set_src1(insn, brw_imm_d(0));
 
-   insn->bits3.sampler.binding_table_index = binding_table_index;
-   insn->bits3.sampler.sampler = sampler;
-   insn->bits3.sampler.msg_type = msg_type;
-   insn->bits3.sampler.return_format = BRW_SAMPLER_RETURN_FORMAT_FLOAT32;
-   insn->bits3.sampler.response_length = response_length;
-   insn->bits3.sampler.msg_length = msg_length;
-   insn->bits3.sampler.end_of_thread = eot;
-   insn->bits3.sampler.msg_target = BRW_MESSAGE_TARGET_SAMPLER;
+   if (BRW_IS_G4X(brw)) {
+      insn->bits3.sampler_g4x.binding_table_index = binding_table_index;
+      insn->bits3.sampler_g4x.sampler = sampler;
+      insn->bits3.sampler_g4x.msg_type = msg_type;
+      insn->bits3.sampler_g4x.response_length = response_length;
+      insn->bits3.sampler_g4x.msg_length = msg_length;
+      insn->bits3.sampler_g4x.end_of_thread = eot;
+      insn->bits3.sampler_g4x.msg_target = BRW_MESSAGE_TARGET_SAMPLER;
+   } else {
+      insn->bits3.sampler.binding_table_index = binding_table_index;
+      insn->bits3.sampler.sampler = sampler;
+      insn->bits3.sampler.msg_type = msg_type;
+      insn->bits3.sampler.return_format = BRW_SAMPLER_RETURN_FORMAT_FLOAT32;
+      insn->bits3.sampler.response_length = response_length;
+      insn->bits3.sampler.msg_length = msg_length;
+      insn->bits3.sampler.end_of_thread = eot;
+      insn->bits3.sampler.msg_target = BRW_MESSAGE_TARGET_SAMPLER;
+   }
 }
 
 
@@ -502,6 +517,8 @@ struct brw_instruction *brw_IF(struct brw_compile *p, GLuint execute_size)
    insn->header.compression_control = BRW_COMPRESSION_NONE;
    insn->header.predicate_control = BRW_PREDICATE_NORMAL;
    insn->header.mask_control = BRW_MASK_ENABLE;
+   if (!p->single_program_flow)
+       insn->header.thread_control = BRW_THREAD_SWITCH;
 
    p->current->header.predicate_control = BRW_PREDICATE_NONE;
 
@@ -527,6 +544,8 @@ struct brw_instruction *brw_ELSE(struct brw_compile *p,
    insn->header.compression_control = BRW_COMPRESSION_NONE;
    insn->header.execution_size = if_insn->header.execution_size;
    insn->header.mask_control = BRW_MASK_ENABLE;
+   if (!p->single_program_flow)
+       insn->header.thread_control = BRW_THREAD_SWITCH;
 
    /* Patch the if instruction to point at this instruction.
     */
@@ -568,6 +587,7 @@ void brw_ENDIF(struct brw_compile *p,
       insn->header.compression_control = BRW_COMPRESSION_NONE;
       insn->header.execution_size = patch_insn->header.execution_size;
       insn->header.mask_control = BRW_MASK_ENABLE;
+      insn->header.thread_control = BRW_THREAD_SWITCH;
 
       assert(patch_insn->bits3.if_else.jump_count == 0);
 
@@ -597,6 +617,34 @@ void brw_ENDIF(struct brw_compile *p,
    }
 }
 
+struct brw_instruction *brw_BREAK(struct brw_compile *p)
+{
+   struct brw_instruction *insn;
+   insn = next_insn(p, BRW_OPCODE_BREAK);
+   brw_set_dest(insn, brw_ip_reg());
+   brw_set_src0(insn, brw_ip_reg());
+   brw_set_src1(insn, brw_imm_d(0x0));
+   insn->header.compression_control = BRW_COMPRESSION_NONE;
+   insn->header.execution_size = BRW_EXECUTE_8;
+   /* insn->header.mask_control = BRW_MASK_DISABLE; */
+   insn->bits3.if_else.pad0 = 0;
+   return insn;
+}
+
+struct brw_instruction *brw_CONT(struct brw_compile *p)
+{
+   struct brw_instruction *insn;
+   insn = next_insn(p, BRW_OPCODE_CONTINUE);
+   brw_set_dest(insn, brw_ip_reg());
+   brw_set_src0(insn, brw_ip_reg());
+   brw_set_src1(insn, brw_imm_d(0x0));
+   insn->header.compression_control = BRW_COMPRESSION_NONE;
+   insn->header.execution_size = BRW_EXECUTE_8;
+   /* insn->header.mask_control = BRW_MASK_DISABLE; */
+   insn->bits3.if_else.pad0 = 0;
+   return insn;
+}
+
 /* DO/WHILE loop:
  */
 struct brw_instruction *brw_DO(struct brw_compile *p, GLuint execute_size)
@@ -608,13 +656,15 @@ struct brw_instruction *brw_DO(struct brw_compile *p, GLuint execute_size)
 
       /* Override the defaults for this instruction:
        */
-      brw_set_dest(insn, retype(brw_vec1_grf(0,0), BRW_REGISTER_TYPE_UD));
-      brw_set_src0(insn, retype(brw_vec1_grf(0,0), BRW_REGISTER_TYPE_UD));
-      brw_set_src1(insn, retype(brw_vec1_grf(0,0), BRW_REGISTER_TYPE_UD));
+      brw_set_dest(insn, brw_null_reg());
+      brw_set_src0(insn, brw_null_reg());
+      brw_set_src1(insn, brw_null_reg());
 
       insn->header.compression_control = BRW_COMPRESSION_NONE;
       insn->header.execution_size = execute_size;
+      insn->header.predicate_control = BRW_PREDICATE_NONE;
       /* insn->header.mask_control = BRW_MASK_ENABLE; */
+      /* insn->header.mask_control = BRW_MASK_DISABLE; */
 
       return insn;
    }
@@ -622,7 +672,7 @@ struct brw_instruction *brw_DO(struct brw_compile *p, GLuint execute_size)
 
 
 
-void brw_WHILE(struct brw_compile *p, 
+struct brw_instruction *brw_WHILE(struct brw_compile *p, 
 	       struct brw_instruction *do_insn)
 {
    struct brw_instruction *insn;
@@ -646,14 +696,16 @@ void brw_WHILE(struct brw_compile *p,
       insn->header.execution_size = do_insn->header.execution_size;
 
       assert(do_insn->header.opcode == BRW_OPCODE_DO);
-      insn->bits3.if_else.jump_count = do_insn - insn;
+      insn->bits3.if_else.jump_count = do_insn - insn + 1;
       insn->bits3.if_else.pop_count = 0;
       insn->bits3.if_else.pad0 = 0;
    }
 
 /*    insn->header.mask_control = BRW_MASK_ENABLE; */
 
+   /* insn->header.mask_control = BRW_MASK_DISABLE; */
    p->current->header.predicate_control = BRW_PREDICATE_NONE;   
+   return insn;
 }
 
 
@@ -985,7 +1037,7 @@ void brw_SAMPLE(struct brw_compile *p,
 
       brw_set_dest(insn, dest);
       brw_set_src0(insn, src0);
-      brw_set_sampler_message(insn,
+      brw_set_sampler_message(p->brw, insn,
 			      binding_table_index,
 			      sampler,
 			      msg_type,
diff --git a/src/mesa/drivers/dri/i965/brw_exec_generic.c b/src/mesa/drivers/dri/i965/brw_exec_generic.c
deleted file mode 100644
index 11d1ef76f59..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec_generic.c
+++ /dev/null
@@ -1,530 +0,0 @@
-/**************************************************************************
-
-Copyright 2004 Tungsten Graphics Inc., Cedar Park, Texas.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ATI, TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <keith@tungstengraphics.com>
- */
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "vtxfmt.h"
-#include "dlist.h"
-#include "state.h"
-#include "light.h"
-#include "api_arrayelt.h"
-#include "api_noop.h"
-
-#include "brw_exec.h"
-
-
-/* Versions of all the entrypoints for situations where codegen isn't
- * available.  
- *
- * Note: Only one size for each attribute may be active at once.
- * Eg. if Color3f is installed/active, then Color4f may not be, even
- * if the vertex actually contains 4 color coordinates.  This is
- * because the 3f version won't otherwise set color[3] to 1.0 -- this
- * is the job of the chooser function when switching between Color4f
- * and Color3f.
- */
-#define ATTRFV( ATTR, N )				\
-static void attrib_##ATTR##_##N( const GLfloat *v )	\
-{							\
-   GET_CURRENT_CONTEXT( ctx );				\
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;			\
-							\
-   if ((ATTR) == 0) {					\
-      GLuint i;						\
-							\
-      if (N>0) exec->vtx.vbptr[0] = v[0];		\
-      if (N>1) exec->vtx.vbptr[1] = v[1];		\
-      if (N>2) exec->vtx.vbptr[2] = v[2];		\
-      if (N>3) exec->vtx.vbptr[3] = v[3];		\
-							\
-      for (i = N; i < exec->vtx.vertex_size; i++)	\
-	 exec->vtx.vbptr[i] = exec->vtx.vertex[i];	\
-							\
-      exec->vtx.vbptr += exec->vtx.vertex_size;		\
-      exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES; \
-							\
-      if (++exec->vtx.vert_count >= exec->vtx.max_vert)	\
-	 brw_exec_vtx_wrap( exec );		\
-   }							\
-   else {						\
-      GLfloat *dest = exec->vtx.attrptr[ATTR];		\
-      if (N>0) dest[0] = v[0];				\
-      if (N>1) dest[1] = v[1];				\
-      if (N>2) dest[2] = v[2];				\
-      if (N>3) dest[3] = v[3];				\
-   }							\
-}
-
-#define INIT(TAB, ATTR)						\
-   TAB[ATTR][0] = attrib_##ATTR##_1;				\
-   TAB[ATTR][1] = attrib_##ATTR##_2;				\
-   TAB[ATTR][2] = attrib_##ATTR##_3;				\
-   TAB[ATTR][3] = attrib_##ATTR##_4;
-
-
-#define ATTRS( ATTRIB )				\
-   ATTRFV( ATTRIB, 1 )				\
-   ATTRFV( ATTRIB, 2 )				\
-   ATTRFV( ATTRIB, 3 )				\
-   ATTRFV( ATTRIB, 4 )			
-
-ATTRS( 0 )
-ATTRS( 1 )
-ATTRS( 2 )
-ATTRS( 3 )
-ATTRS( 4 )
-ATTRS( 5 )
-ATTRS( 6 )
-ATTRS( 7 )
-ATTRS( 8 )
-ATTRS( 9 )
-ATTRS( 10 )
-ATTRS( 11 )
-ATTRS( 12 )
-ATTRS( 13 )
-ATTRS( 14 )
-ATTRS( 15 )
-
-void brw_exec_generic_attr_table_init( brw_attrfv_func (*tab)[4] )
-{
-   INIT( tab, 0 );
-   INIT( tab, 1 );
-   INIT( tab, 2 );
-   INIT( tab, 3 );
-   INIT( tab, 4 );
-   INIT( tab, 5 );
-   INIT( tab, 6 );
-   INIT( tab, 7 );
-   INIT( tab, 8 );
-   INIT( tab, 9 );
-   INIT( tab, 10 );
-   INIT( tab, 11 );
-   INIT( tab, 12 );
-   INIT( tab, 13 );
-   INIT( tab, 14 );
-   INIT( tab, 15 );
-}
-
-/* These can be made efficient with codegen.  Further, by adding more
- * logic to do_choose(), the double-dispatch for legacy entrypoints
- * like glVertex3f() can be removed.
- */
-#define DISPATCH_ATTRFV( ATTR, COUNT, P )	\
-do {						\
-   GET_CURRENT_CONTEXT( ctx ); 			\
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec; 		\
-   exec->vtx.tabfv[ATTR][COUNT-1]( P );		\
-} while (0)
-
-#define DISPATCH_ATTR1FV( ATTR, V ) DISPATCH_ATTRFV( ATTR, 1, V )
-#define DISPATCH_ATTR2FV( ATTR, V ) DISPATCH_ATTRFV( ATTR, 2, V )
-#define DISPATCH_ATTR3FV( ATTR, V ) DISPATCH_ATTRFV( ATTR, 3, V )
-#define DISPATCH_ATTR4FV( ATTR, V ) DISPATCH_ATTRFV( ATTR, 4, V )
-
-#define DISPATCH_ATTR1F( ATTR, S ) DISPATCH_ATTRFV( ATTR, 1, &(S) )
-
-#define DISPATCH_ATTR2F( ATTR, S,T ) 		\
-do { 						\
-   GLfloat v[2]; 				\
-   v[0] = S; v[1] = T;				\
-   DISPATCH_ATTR2FV( ATTR, v );			\
-} while (0)
-#define DISPATCH_ATTR3F( ATTR, S,T,R ) 		\
-do { 						\
-   GLfloat v[3]; 				\
-   v[0] = S; v[1] = T; v[2] = R;		\
-   DISPATCH_ATTR3FV( ATTR, v );			\
-} while (0)
-#define DISPATCH_ATTR4F( ATTR, S,T,R,Q )	\
-do { 						\
-   GLfloat v[4]; 				\
-   v[0] = S; v[1] = T; v[2] = R; v[3] = Q;	\
-   DISPATCH_ATTR4FV( ATTR, v );			\
-} while (0)
-
-
-static void GLAPIENTRY brw_Vertex2f( GLfloat x, GLfloat y )
-{
-   DISPATCH_ATTR2F( BRW_ATTRIB_POS, x, y );
-}
-
-static void GLAPIENTRY brw_Vertex2fv( const GLfloat *v )
-{
-   DISPATCH_ATTR2FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY brw_Vertex3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   DISPATCH_ATTR3F( BRW_ATTRIB_POS, x, y, z );
-}
-
-static void GLAPIENTRY brw_Vertex3fv( const GLfloat *v )
-{
-   DISPATCH_ATTR3FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY brw_Vertex4f( GLfloat x, GLfloat y, GLfloat z, 
-				      GLfloat w )
-{
-   DISPATCH_ATTR4F( BRW_ATTRIB_POS, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_Vertex4fv( const GLfloat *v )
-{
-   DISPATCH_ATTR4FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY brw_TexCoord1f( GLfloat x )
-{
-   DISPATCH_ATTR1F( BRW_ATTRIB_TEX0, x );
-}
-
-static void GLAPIENTRY brw_TexCoord1fv( const GLfloat *v )
-{
-   DISPATCH_ATTR1FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY brw_TexCoord2f( GLfloat x, GLfloat y )
-{
-   DISPATCH_ATTR2F( BRW_ATTRIB_TEX0, x, y );
-}
-
-static void GLAPIENTRY brw_TexCoord2fv( const GLfloat *v )
-{
-   DISPATCH_ATTR2FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY brw_TexCoord3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   DISPATCH_ATTR3F( BRW_ATTRIB_TEX0, x, y, z );
-}
-
-static void GLAPIENTRY brw_TexCoord3fv( const GLfloat *v )
-{
-   DISPATCH_ATTR3FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY brw_TexCoord4f( GLfloat x, GLfloat y, GLfloat z,
-					GLfloat w )
-{
-   DISPATCH_ATTR4F( BRW_ATTRIB_TEX0, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_TexCoord4fv( const GLfloat *v )
-{
-   DISPATCH_ATTR4FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY brw_Normal3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   DISPATCH_ATTR3F( BRW_ATTRIB_NORMAL, x, y, z );
-}
-
-static void GLAPIENTRY brw_Normal3fv( const GLfloat *v )
-{
-   DISPATCH_ATTR3FV( BRW_ATTRIB_NORMAL, v );
-}
-
-static void GLAPIENTRY brw_FogCoordfEXT( GLfloat x )
-{
-   DISPATCH_ATTR1F( BRW_ATTRIB_FOG, x );
-}
-
-static void GLAPIENTRY brw_FogCoordfvEXT( const GLfloat *v )
-{
-   DISPATCH_ATTR1FV( BRW_ATTRIB_FOG, v );
-}
-
-static void GLAPIENTRY brw_Color3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   DISPATCH_ATTR3F( BRW_ATTRIB_COLOR0, x, y, z );
-}
-
-static void GLAPIENTRY brw_Color3fv( const GLfloat *v )
-{
-   DISPATCH_ATTR3FV( BRW_ATTRIB_COLOR0, v );
-}
-
-static void GLAPIENTRY brw_Color4f( GLfloat x, GLfloat y, GLfloat z, 
-				     GLfloat w )
-{
-   DISPATCH_ATTR4F( BRW_ATTRIB_COLOR0, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_Color4fv( const GLfloat *v )
-{
-   DISPATCH_ATTR4FV( BRW_ATTRIB_COLOR0, v );
-}
-
-static void GLAPIENTRY brw_SecondaryColor3fEXT( GLfloat x, GLfloat y, 
-						 GLfloat z )
-{
-   DISPATCH_ATTR3F( BRW_ATTRIB_COLOR1, x, y, z );
-}
-
-static void GLAPIENTRY brw_SecondaryColor3fvEXT( const GLfloat *v )
-{
-   DISPATCH_ATTR3FV( BRW_ATTRIB_COLOR1, v );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord1f( GLenum target, GLfloat x  )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR1F( attr, x );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord1fv( GLenum target,
-					      const GLfloat *v )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR1FV( attr, v );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord2f( GLenum target, GLfloat x, 
-					     GLfloat y )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR2F( attr, x, y );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord2fv( GLenum target, 
-					      const GLfloat *v )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR2FV( attr, v );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord3f( GLenum target, GLfloat x, 
-					     GLfloat y, GLfloat z)
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR3F( attr, x, y, z );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord3fv( GLenum target, 
-					      const GLfloat *v )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR3FV( attr, v );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord4f( GLenum target, GLfloat x, 
-					     GLfloat y, GLfloat z,
-					     GLfloat w )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR4F( attr, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_MultiTexCoord4fv( GLenum target, 
-					      const GLfloat *v )
-{
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   DISPATCH_ATTR4FV( attr, v );
-}
-
-
-static void GLAPIENTRY brw_VertexAttrib1fNV( GLuint index, GLfloat x )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR1F( index, x );
-}
-
-static void GLAPIENTRY brw_VertexAttrib1fvNV( GLuint index, 
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR1FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib2fNV( GLuint index, GLfloat x, 
-					      GLfloat y )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR2F( index, x, y );
-}
-
-static void GLAPIENTRY brw_VertexAttrib2fvNV( GLuint index,
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR2FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib3fNV( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR3F( index, x, y, z );
-}
-
-static void GLAPIENTRY brw_VertexAttrib3fvNV( GLuint index,
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR3FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib4fNV( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z,
-					      GLfloat w )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR4F( index, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_VertexAttrib4fvNV( GLuint index, 
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR4FV( index, v );
-}
-
-
-/*
- * XXX adjust index
- */
-
-static void GLAPIENTRY brw_VertexAttrib1fARB( GLuint index, GLfloat x )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR1F( index, x );
-}
-
-static void GLAPIENTRY brw_VertexAttrib1fvARB( GLuint index, 
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR1FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib2fARB( GLuint index, GLfloat x, 
-					      GLfloat y )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR2F( index, x, y );
-}
-
-static void GLAPIENTRY brw_VertexAttrib2fvARB( GLuint index,
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR2FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib3fARB( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR3F( index, x, y, z );
-}
-
-static void GLAPIENTRY brw_VertexAttrib3fvARB( GLuint index,
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR3FV( index, v );
-}
-
-static void GLAPIENTRY brw_VertexAttrib4fARB( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z,
-					      GLfloat w )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR4F( index, x, y, z, w );
-}
-
-static void GLAPIENTRY brw_VertexAttrib4fvARB( GLuint index, 
-					       const GLfloat *v )
-{
-   if (index >= BRW_ATTRIB_FIRST_MATERIAL) index = ERROR_ATTRIB;
-   DISPATCH_ATTR4FV( index, v );
-}
-
-
-/* Install the generic versions of the 2nd level dispatch
- * functions.  Some of these have a codegen alternative.
- */
-void brw_exec_vtx_generic_init( struct brw_exec_context *exec )
-{
-   GLvertexformat *vfmt = &exec->vtxfmt;
-
-   vfmt->Color3f = brw_Color3f;
-   vfmt->Color3fv = brw_Color3fv;
-   vfmt->Color4f = brw_Color4f;
-   vfmt->Color4fv = brw_Color4fv;
-   vfmt->FogCoordfEXT = brw_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = brw_FogCoordfvEXT;
-   vfmt->MultiTexCoord1fARB = brw_MultiTexCoord1f;
-   vfmt->MultiTexCoord1fvARB = brw_MultiTexCoord1fv;
-   vfmt->MultiTexCoord2fARB = brw_MultiTexCoord2f;
-   vfmt->MultiTexCoord2fvARB = brw_MultiTexCoord2fv;
-   vfmt->MultiTexCoord3fARB = brw_MultiTexCoord3f;
-   vfmt->MultiTexCoord3fvARB = brw_MultiTexCoord3fv;
-   vfmt->MultiTexCoord4fARB = brw_MultiTexCoord4f;
-   vfmt->MultiTexCoord4fvARB = brw_MultiTexCoord4fv;
-   vfmt->Normal3f = brw_Normal3f;
-   vfmt->Normal3fv = brw_Normal3fv;
-   vfmt->SecondaryColor3fEXT = brw_SecondaryColor3fEXT;
-   vfmt->SecondaryColor3fvEXT = brw_SecondaryColor3fvEXT;
-   vfmt->TexCoord1f = brw_TexCoord1f;
-   vfmt->TexCoord1fv = brw_TexCoord1fv;
-   vfmt->TexCoord2f = brw_TexCoord2f;
-   vfmt->TexCoord2fv = brw_TexCoord2fv;
-   vfmt->TexCoord3f = brw_TexCoord3f;
-   vfmt->TexCoord3fv = brw_TexCoord3fv;
-   vfmt->TexCoord4f = brw_TexCoord4f;
-   vfmt->TexCoord4fv = brw_TexCoord4fv;
-   vfmt->Vertex2f = brw_Vertex2f;
-   vfmt->Vertex2fv = brw_Vertex2fv;
-   vfmt->Vertex3f = brw_Vertex3f;
-   vfmt->Vertex3fv = brw_Vertex3fv;
-   vfmt->Vertex4f = brw_Vertex4f;
-   vfmt->Vertex4fv = brw_Vertex4fv;
-   vfmt->VertexAttrib1fNV = brw_VertexAttrib1fNV;
-   vfmt->VertexAttrib1fvNV = brw_VertexAttrib1fvNV;
-   vfmt->VertexAttrib2fNV = brw_VertexAttrib2fNV;
-   vfmt->VertexAttrib2fvNV = brw_VertexAttrib2fvNV;
-   vfmt->VertexAttrib3fNV = brw_VertexAttrib3fNV;
-   vfmt->VertexAttrib3fvNV = brw_VertexAttrib3fvNV;
-   vfmt->VertexAttrib4fNV = brw_VertexAttrib4fNV;
-   vfmt->VertexAttrib4fvNV = brw_VertexAttrib4fvNV;
-   vfmt->VertexAttrib1fARB = brw_VertexAttrib1fARB;
-   vfmt->VertexAttrib1fvARB = brw_VertexAttrib1fvARB;
-   vfmt->VertexAttrib2fARB = brw_VertexAttrib2fARB;
-   vfmt->VertexAttrib2fvARB = brw_VertexAttrib2fvARB;
-   vfmt->VertexAttrib3fARB = brw_VertexAttrib3fARB;
-   vfmt->VertexAttrib3fvARB = brw_VertexAttrib3fvARB;
-   vfmt->VertexAttrib4fARB = brw_VertexAttrib4fARB;
-   vfmt->VertexAttrib4fvARB = brw_VertexAttrib4fvARB;
-}
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.c b/src/mesa/drivers/dri/i965/brw_fallback.c
index 86464b2ec5f..4ea660a51a3 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.c
+++ b/src/mesa/drivers/dri/i965/brw_fallback.c
@@ -25,52 +25,47 @@
  * 
  **************************************************************************/
 
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/swrast.h"
 #include "tnl/tnl.h"
-#include "context.h"
 #include "brw_context.h"
 #include "brw_fallback.h"
 
-#include "glheader.h"
-#include "enums.h"
-#include "glapi.h"
-#include "imports.h"
-#include "macros.h"
-#include "mtypes.h"
-
-
-
-
-
+#include "glapi/glapi.h"
 
+#define FILE_DEBUG_FLAG DEBUG_FALLBACKS
 
 static GLboolean do_check_fallback(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    GLuint i;
-   
+
    /* BRW_NEW_METAOPS
     */
    if (brw->metaops.active)
       return GL_FALSE;
 
-   if (brw->intel.no_rast)
-      return GL_TRUE;
-   
-   /* _NEW_BUFFERS
-    */
-   if (ctx->DrawBuffer->_ColorDrawBufferMask[0] != BUFFER_BIT_FRONT_LEFT &&
-       ctx->DrawBuffer->_ColorDrawBufferMask[0] != BUFFER_BIT_BACK_LEFT)
+   if (brw->intel.no_rast) {
+      DBG("FALLBACK: rasterization disabled\n");
       return GL_TRUE;
+   }
 
    /* _NEW_RENDERMODE
     *
     * XXX: need to save/restore RenderMode in metaops state, or
     * somehow move to a new attribs pointer:
     */
-   if (ctx->RenderMode != GL_RENDER)
+   if (ctx->RenderMode != GL_RENDER) {
+      DBG("FALLBACK: render mode\n");
       return GL_TRUE;
+   }
 
    /* _NEW_TEXTURE:
     */
@@ -79,8 +74,10 @@ static GLboolean do_check_fallback(struct brw_context *brw)
       if (texUnit->_ReallyEnabled) {
 	 struct intel_texture_object *intelObj = intel_texture_object(texUnit->_Current);
 	 struct gl_texture_image *texImage = intelObj->base.Image[0][intelObj->firstLevel];
-	 if (texImage->Border)
+	 if (texImage->Border) {
+	    DBG("FALLBACK: texture border\n");
 	    return GL_TRUE;
+	 }
       }
    }
    
@@ -88,6 +85,7 @@ static GLboolean do_check_fallback(struct brw_context *brw)
     */
    if (brw->attribs.Stencil->Enabled && 
        !brw->intel.hw_stencil) {
+      DBG("FALLBACK: stencil\n");
       return GL_TRUE;
    }
 
@@ -106,7 +104,7 @@ const struct brw_tracked_state brw_check_fallback = {
       .brw  = BRW_NEW_METAOPS,
       .cache = 0
    },
-   .update = check_fallback
+   .prepare = check_fallback
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.h b/src/mesa/drivers/dri/i965/brw_fallback.h
index 684a46cd170..50dcdacd17a 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.h
+++ b/src/mesa/drivers/dri/i965/brw_fallback.h
@@ -28,7 +28,7 @@
 #ifndef BRW_FALLBACK_H
 #define BRW_FALLBACK_H
 
-#include "mtypes.h"		/* for GLcontext... */
+#include "main/mtypes.h"		/* for GLcontext... */
 
 struct brw_context;
 struct vbo_prim;
diff --git a/src/mesa/drivers/dri/i965/brw_gs.c b/src/mesa/drivers/dri/i965/brw_gs.c
index 73263a5fff4..a8b74a0afe6 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.c
+++ b/src/mesa/drivers/dri/i965/brw_gs.c
@@ -29,9 +29,9 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
       
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "intel_batchbuffer.h"
 
@@ -65,7 +65,7 @@ static void compile_gs_prog( struct brw_context *brw,
    
    /* Begin the compilation:
     */
-   brw_init_compile(&c.func);
+   brw_init_compile(brw, &c.func);
 
    c.func.single_program_flow = 1;
 
@@ -119,26 +119,15 @@ static void compile_gs_prog( struct brw_context *brw,
 
    /* Upload
     */
-   brw->gs.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_GS_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->gs.prog_data );
+   dri_bo_unreference(brw->gs.prog_bo);
+   brw->gs.prog_bo = brw_upload_cache( &brw->cache, BRW_GS_PROG,
+				       &c.key, sizeof(c.key),
+				       NULL, 0,
+				       program, program_size,
+				       &c.prog_data,
+				       &brw->gs.prog_data );
 }
 
-
-static GLboolean search_cache( struct brw_context *brw, 
-			       struct brw_gs_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_GS_PROG], 
-			   key, sizeof(*key),
-			   &brw->gs.prog_data,
-			   &brw->gs.prog_gs_offset);
-}
-
-
 static const GLenum gs_prim[GL_POLYGON+1] = {  
    GL_POINTS,
    GL_LINES,
@@ -173,10 +162,9 @@ static void populate_key( struct brw_context *brw,
 
 /* Calculate interpolants for triangle and line rasterization.
  */
-static void upload_gs_prog( struct brw_context *brw )
+static void prepare_gs_prog(struct brw_context *brw)
 {
    struct brw_gs_prog_key key;
-
    /* Populate the key:
     */
    populate_key(brw, &key);
@@ -187,7 +175,12 @@ static void upload_gs_prog( struct brw_context *brw )
    }
 
    if (brw->gs.prog_active) {
-      if (!search_cache(brw, &key))
+      dri_bo_unreference(brw->gs.prog_bo);
+      brw->gs.prog_bo = brw_search_cache(&brw->cache, BRW_GS_PROG,
+					 &key, sizeof(key),
+					 NULL, 0,
+					 &brw->gs.prog_data);
+      if (brw->gs.prog_bo == NULL)
 	 compile_gs_prog( brw, &key );
    }
 }
@@ -199,5 +192,5 @@ const struct brw_tracked_state brw_gs_prog = {
       .brw   = BRW_NEW_PRIMITIVE,
       .cache = CACHE_NEW_VS_PROG
    },
-   .update = upload_gs_prog
+   .prepare = prepare_gs_prog
 };
diff --git a/src/mesa/drivers/dri/i965/brw_gs.h b/src/mesa/drivers/dri/i965/brw_gs.h
index 29a4e80ce1b..18a4537c323 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.h
+++ b/src/mesa/drivers/dri/i965/brw_gs.h
@@ -40,11 +40,11 @@
 #define MAX_GS_VERTS (4)	     
 
 struct brw_gs_prog_key {
+   GLuint attrs:32;
    GLuint primitive:4;
-   GLuint attrs:16;		
    GLuint hint_gs_always:1;
    GLuint need_gs_prog:1;
-   GLuint pad:10;
+   GLuint pad:26;
 };
 
 struct brw_gs_compile {
diff --git a/src/mesa/drivers/dri/i965/brw_gs_emit.c b/src/mesa/drivers/dri/i965/brw_gs_emit.c
index 9abb94d82ed..22e0d25c2e7 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_emit.c
@@ -30,9 +30,9 @@
   */
  
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "shader/program.h"
 #include "intel_batchbuffer.h"
diff --git a/src/mesa/drivers/dri/i965/brw_gs_state.c b/src/mesa/drivers/dri/i965/brw_gs_state.c
index 5826c01d4f9..27023cf0344 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_state.c
@@ -34,49 +34,103 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "macros.h"
+#include "main/macros.h"
 
+struct brw_gs_unit_key {
+   unsigned int total_grf;
+   unsigned int urb_entry_read_length;
 
+   unsigned int curbe_offset;
 
-static void upload_gs_unit( struct brw_context *brw )
-{
-   struct brw_gs_unit_state gs;
+   unsigned int nr_urb_entries, urb_size;
+   GLboolean prog_active;
+};
 
-   memset(&gs, 0, sizeof(gs));
+static void
+gs_unit_populate_key(struct brw_context *brw, struct brw_gs_unit_key *key)
+{
+   memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_GS_PROG */
-   if (brw->gs.prog_active) {
-      gs.thread0.grf_reg_count = ((brw->gs.prog_data->total_grf-1) & ~15) / 16;
-      gs.thread0.kernel_start_pointer = brw->gs.prog_gs_offset >> 6;
-      gs.thread3.urb_entry_read_length = brw->gs.prog_data->urb_read_length;
-   }
-   else {
-      gs.thread0.grf_reg_count = 0;
-      gs.thread0.kernel_start_pointer = 0;
-      gs.thread3.urb_entry_read_length = 1;
+   key->prog_active = brw->gs.prog_active;
+   if (key->prog_active) {
+      key->total_grf = brw->gs.prog_data->total_grf;
+      key->urb_entry_read_length = brw->gs.prog_data->urb_read_length;
+   } else {
+      key->total_grf = 1;
+      key->urb_entry_read_length = 1;
    }
 
+   /* BRW_NEW_CURBE_OFFSETS */
+   key->curbe_offset = brw->curbe.clip_start;
+
    /* BRW_NEW_URB_FENCE */
-   gs.thread4.nr_urb_entries = brw->urb.nr_gs_entries; 
-   gs.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
+   key->nr_urb_entries = brw->urb.nr_gs_entries;
+   key->urb_size = brw->urb.vsize;
+}
 
-   gs.thread4.max_threads = 0; /* Hardware requirement */
+static dri_bo *
+gs_unit_create_from_key(struct brw_context *brw, struct brw_gs_unit_key *key)
+{
+   struct brw_gs_unit_state gs;
+   dri_bo *bo;
 
-   if (INTEL_DEBUG & DEBUG_STATS)
-      gs.thread4.stats_enable = 1; 
+   memset(&gs, 0, sizeof(gs));
+
+   gs.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
+   if (key->prog_active) /* reloc */
+      gs.thread0.kernel_start_pointer = brw->gs.prog_bo->offset >> 6;
 
-   /* CONSTANT */
    gs.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
    gs.thread1.single_program_flow = 1;
+
    gs.thread3.dispatch_grf_start_reg = 1;
    gs.thread3.const_urb_entry_read_offset = 0;
    gs.thread3.const_urb_entry_read_length = 0;
    gs.thread3.urb_entry_read_offset = 0;
-   
+   gs.thread3.urb_entry_read_length = key->urb_entry_read_length;
+
+   gs.thread4.nr_urb_entries = key->nr_urb_entries;
+   gs.thread4.urb_entry_allocation_size = key->urb_size - 1;
+
+   gs.thread4.max_threads = 0; /* Hardware requirement */
+
+   if (INTEL_DEBUG & DEBUG_STATS)
+      gs.thread4.stats_enable = 1;
+
+   bo = brw_upload_cache(&brw->cache, BRW_GS_UNIT,
+			 key, sizeof(*key),
+			 &brw->gs.prog_bo, 1,
+			 &gs, sizeof(gs),
+			 NULL, NULL);
 
-   brw->gs.state_gs_offset = brw_cache_data( &brw->cache[BRW_GS_UNIT], &gs );
+   if (key->prog_active) {
+      /* Emit GS program relocation */
+      dri_bo_emit_reloc(bo,
+			I915_GEM_DOMAIN_INSTRUCTION, 0,
+			gs.thread0.grf_reg_count << 1,
+			offsetof(struct brw_gs_unit_state, thread0),
+			brw->gs.prog_bo);
+   }
+
+   return bo;
 }
 
+static void prepare_gs_unit(struct brw_context *brw)
+{
+   struct brw_gs_unit_key key;
+
+   gs_unit_populate_key(brw, &key);
+
+   dri_bo_unreference(brw->gs.state_bo);
+   brw->gs.state_bo = brw_search_cache(&brw->cache, BRW_GS_UNIT,
+				       &key, sizeof(key),
+				       &brw->gs.prog_bo, 1,
+				       NULL);
+   if (brw->gs.state_bo == NULL) {
+      brw->gs.state_bo = gs_unit_create_from_key(brw, &key);
+   }
+}
 
 const struct brw_tracked_state brw_gs_unit = {
    .dirty = {
@@ -85,5 +139,5 @@ const struct brw_tracked_state brw_gs_unit = {
 		BRW_NEW_URB_FENCE),
       .cache = CACHE_NEW_GS_PROG
    },
-   .update = upload_gs_unit
+   .prepare = prepare_gs_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_hal.c b/src/mesa/drivers/dri/i965/brw_hal.c
deleted file mode 100644
index 3126102749b..00000000000
--- a/src/mesa/drivers/dri/i965/brw_hal.c
+++ /dev/null
@@ -1,52 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
-
-#include "intel_batchbuffer.h"
-#include "brw_context.h"
-#include "brw_state.h"
-#include "brw_defines.h"
-#include "brw_hal.h"
-#include <dlfcn.h>
-
-static void *brw_hal_lib;
-static GLboolean brw_hal_tried;
-
-void *
-brw_hal_find_symbol (char *symbol)
-{
-    if (!brw_hal_tried)
-    {
-	char *brw_hal_name = getenv ("INTEL_HAL");
-    
-	if (!brw_hal_name)
-	    brw_hal_name = "/usr/lib/xorg/modules/drivers/intel_hal.so";
-
-	brw_hal_lib = dlopen (brw_hal_name, RTLD_LAZY|RTLD_LOCAL);
-	brw_hal_tried = 1;
-    }
-    if (!brw_hal_lib)
-	return NULL;
-    return dlsym (brw_hal_lib, symbol);
-}
diff --git a/src/mesa/drivers/dri/i965/brw_hal.h b/src/mesa/drivers/dri/i965/brw_hal.h
deleted file mode 100644
index cd86e395b59..00000000000
--- a/src/mesa/drivers/dri/i965/brw_hal.h
+++ /dev/null
@@ -1,27 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
-
-void *
-brw_hal_find_symbol (char *symbol);
diff --git a/src/mesa/drivers/dri/i965/brw_metaops.c b/src/mesa/drivers/dri/i965/brw_metaops.c
index 6e030f191ef..41bfa2e2567 100644
--- a/src/mesa/drivers/dri/i965/brw_metaops.c
+++ b/src/mesa/drivers/dri/i965/brw_metaops.c
@@ -32,15 +32,16 @@
  
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
 
 #include "shader/arbprogparse.h"
 
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
 #include "intel_regions.h"
+#include "intel_buffers.h"
 
 #include "brw_context.h"
 #include "brw_defines.h"
@@ -156,7 +157,7 @@ static const char *fp_tex_prog =
  *   FragmentProgram->_Current
  *   VertexProgram->_Enabled
  *   brw->vertex_program
- *   DrawBuffer->_ColorDrawBufferMask[0]
+ *   DrawBuffer->_ColorDrawBufferIndexes[0]
  * 
  *
  * More if drawpixels-through-texture is added.  
@@ -195,7 +196,7 @@ static void init_metaops_state( struct brw_context *brw )
 				  vp_prog, strlen(vp_prog),
 				  brw->metaops.vp);
 
-   brw->metaops.attribs.VertexProgram->Current = brw->metaops.vp;
+   brw->metaops.attribs.VertexProgram->_Current = brw->metaops.vp;
    brw->metaops.attribs.VertexProgram->_Enabled = GL_TRUE;
 
    brw->metaops.attribs.FragmentProgram->_Current = brw->metaops.fp;
@@ -361,13 +362,21 @@ static void meta_draw_region( struct intel_context *intel,
    struct brw_context *brw = brw_context(&intel->ctx);
 
    if (!brw->metaops.saved_draw_region) {
-      brw->metaops.saved_draw_region = brw->state.draw_region;
+      brw->metaops.saved_draw_region = brw->state.draw_regions[0];
+      brw->metaops.saved_nr_draw_regions = brw->state.nr_draw_regions;
       brw->metaops.saved_depth_region = brw->state.depth_region;
    }
 
-   brw->state.draw_region = draw_region;
+   brw->state.draw_regions[0] = draw_region;
+   brw->state.nr_draw_regions = 1;
    brw->state.depth_region = depth_region;
 
+   if (intel->frame_buffer_texobj != NULL)
+      brw_FrameBufferTexDestroy(brw);
+
+   if (draw_region)
+       brw_FrameBufferTexInit(brw, draw_region);
+
    brw->state.dirty.mesa |= _NEW_BUFFERS;
 }
 
@@ -376,8 +385,7 @@ static void meta_draw_quad(struct intel_context *intel,
 			   GLfloat x0, GLfloat x1,
 			   GLfloat y0, GLfloat y1, 
 			   GLfloat z,
-			   GLubyte red, GLubyte green,
-			   GLubyte blue, GLubyte alpha,
+			   GLuint color,
 			   GLfloat s0, GLfloat s1,
 			   GLfloat t0, GLfloat t1)
 {
@@ -388,7 +396,6 @@ static void meta_draw_quad(struct intel_context *intel,
    struct gl_client_array *attribs[VERT_ATTRIB_MAX];
    struct _mesa_prim prim[1];
    GLfloat pos[4][3];
-   GLubyte color[4];
 
    ctx->Driver.BufferData(ctx,
 			  GL_ARRAY_BUFFER_ARB,
@@ -413,7 +420,6 @@ static void meta_draw_quad(struct intel_context *intel,
    pos[3][1] = y1;
    pos[3][2] = z;
 
-
    ctx->Driver.BufferSubData(ctx,
 			     GL_ARRAY_BUFFER_ARB,
 			     0,
@@ -421,16 +427,15 @@ static void meta_draw_quad(struct intel_context *intel,
 			     pos,
 			     brw->metaops.vbo);
 
-   color[0] = red;
-   color[1] = green;
-   color[2] = blue;
-   color[3] = alpha;
+   /* Convert incoming ARGB to required RGBA */
+   /* Note this color is stored as GL_UNSIGNED_BYTE */
+   color = (color & 0xff00ff00) | (((color >> 16) | (color << 16)) & 0xff00ff);
 
    ctx->Driver.BufferSubData(ctx,
 			     GL_ARRAY_BUFFER_ARB,
 			     sizeof(pos),
 			     sizeof(color),
-			     color,
+			     &color,
 			     brw->metaops.vbo);
 
    /* Ignoring texture coords. 
@@ -486,6 +491,7 @@ static void install_meta_state( struct intel_context *intel )
 {
    GLcontext *ctx = &intel->ctx;
    struct brw_context *brw = brw_context(ctx);
+   GLuint i;
 
    if (!brw->metaops.vbo) {
       init_metaops_state(brw);
@@ -495,12 +501,18 @@ static void install_meta_state( struct intel_context *intel )
    
    meta_no_texture(&brw->intel);
    meta_flat_shade(&brw->intel);
-   brw->metaops.restore_draw_mask = ctx->DrawBuffer->_ColorDrawBufferMask[0];
+   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
+      brw->metaops.restore_draw_buffers[i]
+         = ctx->DrawBuffer->_ColorDrawBufferIndexes[i];
+   }
+   brw->metaops.restore_num_draw_buffers = ctx->DrawBuffer->_NumColorDrawBuffers;
+
    brw->metaops.restore_fp = ctx->FragmentProgram.Current;
 
    /* This works without adjusting refcounts.  Fix later? 
     */
-   brw->metaops.saved_draw_region = brw->state.draw_region;
+   brw->metaops.saved_draw_region = brw->state.draw_regions[0];
+   brw->metaops.saved_nr_draw_regions = brw->state.nr_draw_regions;
    brw->metaops.saved_depth_region = brw->state.depth_region;
    brw->metaops.active = 1;
    
@@ -511,13 +523,20 @@ static void leave_meta_state( struct intel_context *intel )
 {
    GLcontext *ctx = &intel->ctx;
    struct brw_context *brw = brw_context(ctx);
+   GLuint i;
 
    restore_attribs(brw);
 
-   ctx->DrawBuffer->_ColorDrawBufferMask[0] = brw->metaops.restore_draw_mask;
+   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
+      ctx->DrawBuffer->_ColorDrawBufferIndexes[i]
+         = brw->metaops.restore_draw_buffers[i];
+   }
+   ctx->DrawBuffer->_NumColorDrawBuffers = brw->metaops.restore_num_draw_buffers;
+
    ctx->FragmentProgram.Current = brw->metaops.restore_fp;
 
-   brw->state.draw_region = brw->metaops.saved_draw_region;
+   brw->state.draw_regions[0] = brw->metaops.saved_draw_region;
+   brw->state.nr_draw_regions = brw->metaops.saved_nr_draw_regions;
    brw->state.depth_region = brw->metaops.saved_depth_region;
    brw->metaops.saved_draw_region = NULL;
    brw->metaops.saved_depth_region = NULL;
diff --git a/src/mesa/drivers/dri/i965/brw_misc_state.c b/src/mesa/drivers/dri/i965/brw_misc_state.c
index d5779680ffc..627705fa9ba 100644
--- a/src/mesa/drivers/dri/i965/brw_misc_state.c
+++ b/src/mesa/drivers/dri/i965/brw_misc_state.c
@@ -68,148 +68,114 @@ const struct brw_tracked_state brw_blend_constant_color = {
       .brw = 0,
       .cache = 0
    },
-   .update = upload_blend_constant_color
+   .emit = upload_blend_constant_color
 };
 
-/***********************************************************************
- * Drawing rectangle -- Need for AUB file only.
- */
+/* Constant single cliprect for framebuffer object or DRI2 drawing */
 static void upload_drawing_rect(struct brw_context *brw)
 {
    struct intel_context *intel = &brw->intel;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   struct brw_drawrect bdr;
-   int x1, y1;
-   int x2, y2;
+   GLcontext *ctx = &intel->ctx;
 
-   /* If there is a single cliprect, set it here.  Otherwise iterate
-    * over them in brw_draw_prim().
-    */
-   if (brw->intel.numClipRects > 1) 
-      return; 
- 
-   x1 = brw->intel.pClipRects[0].x1;
-   y1 = brw->intel.pClipRects[0].y1;
-   x2 = brw->intel.pClipRects[0].x2;
-   y2 = brw->intel.pClipRects[0].y2;
-	 
-   if (x1 < 0) x1 = 0;
-   if (y1 < 0) y1 = 0;
-   if (x2 > intel->intelScreen->width) x2 = intel->intelScreen->width;
-   if (y2 > intel->intelScreen->height) y2 = intel->intelScreen->height;
-
-   memset(&bdr, 0, sizeof(bdr));
-   bdr.header.opcode = CMD_DRAW_RECT;
-   bdr.header.length = sizeof(bdr)/4 - 2;
-   bdr.xmin = x1;
-   bdr.ymin = y1;
-   bdr.xmax = x2;
-   bdr.ymax = y2;
-   bdr.xorg = dPriv->x;
-   bdr.yorg = dPriv->y;
-
-   /* Can't use BRW_CACHED_BATCH_STRUCT because this is also emitted
-    * uncached in brw_draw.c:
-    */
-   BRW_BATCH_STRUCT(brw, &bdr);
+   if (!intel->constant_cliprect)
+      return;
+
+   BEGIN_BATCH(4, NO_LOOP_CLIPRECTS);
+   OUT_BATCH(_3DSTATE_DRAWRECT_INFO_I965);
+   OUT_BATCH(0); /* xmin, ymin */
+   OUT_BATCH(((ctx->DrawBuffer->Width - 1) & 0xffff) |
+	    ((ctx->DrawBuffer->Height - 1) << 16));
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
 }
 
 const struct brw_tracked_state brw_drawing_rect = {
    .dirty = {
-      .mesa = _NEW_WINDOW_POS,
+      .mesa = _NEW_BUFFERS,
       .brw = 0,
       .cache = 0
    },
-   .update = upload_drawing_rect
+   .emit = upload_drawing_rect
 };
 
-/***********************************************************************
- * Binding table pointers
- */
+static void prepare_binding_table_pointers(struct brw_context *brw)
+{
+   brw_add_validated_bo(brw, brw->wm.bind_bo);
+}
 
+/**
+ * Upload the binding table pointers, which point each stage's array of surface
+ * state pointers.
+ *
+ * The binding table pointers are relative to the surface state base address,
+ * which is 0.
+ */
 static void upload_binding_table_pointers(struct brw_context *brw)
 {
-   struct brw_binding_table_pointers btp;
-   memset(&btp, 0, sizeof(btp));
-
-   /* The binding table has been emitted to the SS pool already, so we
-    * know what its offset is.  When the batch buffer is fired, the
-    * binding table and surface structs will get fixed up to point to
-    * where the textures actually landed, but that won't change the
-    * value of the offsets here:
-    */
-   btp.header.opcode = CMD_BINDING_TABLE_PTRS;
-   btp.header.length = sizeof(btp)/4 - 2;
-   btp.vs = 0;
-   btp.gs = 0;
-   btp.clp = 0;
-   btp.sf = 0;
-   btp.wm = brw->wm.bind_ss_offset;
-
-   BRW_CACHED_BATCH_STRUCT(brw, &btp);
+   struct intel_context *intel = &brw->intel;
+
+   BEGIN_BATCH(6, IGNORE_CLIPRECTS);
+   OUT_BATCH(CMD_BINDING_TABLE_PTRS << 16 | (6 - 2));
+   OUT_BATCH(0); /* vs */
+   OUT_BATCH(0); /* gs */
+   OUT_BATCH(0); /* clip */
+   OUT_BATCH(0); /* sf */
+   OUT_RELOC(brw->wm.bind_bo,
+	     I915_GEM_DOMAIN_SAMPLER, 0,
+	     0);
+   ADVANCE_BATCH();
 }
 
 const struct brw_tracked_state brw_binding_table_pointers = {
    .dirty = {
       .mesa = 0,
-      .brw = 0,
-      .cache = CACHE_NEW_SURF_BIND 
+      .brw = BRW_NEW_BATCH,
+      .cache = CACHE_NEW_SURF_BIND,
    },
-   .update = upload_binding_table_pointers
+   .prepare = prepare_binding_table_pointers,
+   .emit = upload_binding_table_pointers,
 };
 
 
-/***********************************************************************
- * Pipelined state pointers.  This is the key state packet from which
- * the hardware chases pointers to all the uploaded state in VRAM.
+/**
+ * Upload pointers to the per-stage state.
+ *
+ * The state pointers in this packet are all relative to the general state
+ * base address set by CMD_STATE_BASE_ADDRESS, which is 0.
  */
-   
 static void upload_pipelined_state_pointers(struct brw_context *brw )
 {
-   struct brw_pipelined_state_pointers psp;
-   memset(&psp, 0, sizeof(psp));
-
-   psp.header.opcode = CMD_PIPELINED_STATE_POINTERS;
-   psp.header.length = sizeof(psp)/4 - 2;
-
-   psp.vs.offset = brw->vs.state_gs_offset >> 5;
-   psp.sf.offset = brw->sf.state_gs_offset >> 5;
-   psp.wm.offset = brw->wm.state_gs_offset >> 5;
-   psp.cc.offset = brw->cc.state_gs_offset >> 5;
-
-   /* GS gets turned on and off regularly.  Need to re-emit URB fence
-    * after this occurs.  
-    */
-   if (brw->gs.prog_active) {
-      psp.gs.offset = brw->gs.state_gs_offset >> 5;
-      psp.gs.enable = 1;
-   }
+   struct intel_context *intel = &brw->intel;
 
-   if (!brw->metaops.active) {
-      psp.clp.offset = brw->clip.state_gs_offset >> 5;
-      psp.clp.enable = 1;
-   }
+   BEGIN_BATCH(7, IGNORE_CLIPRECTS);
+   OUT_BATCH(CMD_PIPELINED_STATE_POINTERS << 16 | (7 - 2));
+   OUT_RELOC(brw->vs.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
+   if (brw->gs.prog_active)
+      OUT_RELOC(brw->gs.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 1);
+   else
+      OUT_BATCH(0);
+   if (!brw->metaops.active)
+      OUT_RELOC(brw->clip.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 1);
+   else
+      OUT_BATCH(0);
+   OUT_RELOC(brw->sf.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
+   OUT_RELOC(brw->wm.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
+   OUT_RELOC(brw->cc.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
+   ADVANCE_BATCH();
+
+   brw->state.dirty.brw |= BRW_NEW_PSP;
+}
 
 
-   if (BRW_CACHED_BATCH_STRUCT(brw, &psp))
-      brw->state.dirty.brw |= BRW_NEW_PSP;
+static void prepare_psp_urb_cbs(struct brw_context *brw)
+{
+   brw_add_validated_bo(brw, brw->vs.state_bo);
+   brw_add_validated_bo(brw, brw->gs.state_bo);
+   brw_add_validated_bo(brw, brw->clip.state_bo);
+   brw_add_validated_bo(brw, brw->wm.state_bo);
+   brw_add_validated_bo(brw, brw->cc.state_bo);
 }
 
-const struct brw_tracked_state brw_pipelined_state_pointers = {
-   .dirty = {
-      .mesa = 0,
-      .brw = BRW_NEW_METAOPS,
-      .cache = (CACHE_NEW_VS_UNIT | 
-		CACHE_NEW_GS_UNIT | 
-		CACHE_NEW_GS_PROG | 
-		CACHE_NEW_CLIP_UNIT | 
-		CACHE_NEW_SF_UNIT | 
-		CACHE_NEW_WM_UNIT | 
-		CACHE_NEW_CC_UNIT)
-   },
-   .update = upload_pipelined_state_pointers
-};
-
 static void upload_psp_urb_cbs(struct brw_context *brw )
 {
    upload_pipelined_state_pointers(brw);
@@ -217,11 +183,10 @@ static void upload_psp_urb_cbs(struct brw_context *brw )
    brw_upload_constant_buffer_state(brw);
 }
 
-
 const struct brw_tracked_state brw_psp_urb_cbs = {
    .dirty = {
       .mesa = 0,
-      .brw = BRW_NEW_URB_FENCE | BRW_NEW_METAOPS,
+      .brw = BRW_NEW_URB_FENCE | BRW_NEW_METAOPS | BRW_NEW_BATCH,
       .cache = (CACHE_NEW_VS_UNIT | 
 		CACHE_NEW_GS_UNIT | 
 		CACHE_NEW_GS_PROG | 
@@ -230,74 +195,85 @@ const struct brw_tracked_state brw_psp_urb_cbs = {
 		CACHE_NEW_WM_UNIT | 
 		CACHE_NEW_CC_UNIT)
    },
-   .update = upload_psp_urb_cbs
+   .prepare = prepare_psp_urb_cbs,
+   .emit = upload_psp_urb_cbs,
 };
 
+static void prepare_depthbuffer(struct brw_context *brw)
+{
+   struct intel_region *region = brw->state.depth_region;
 
+   if (region != NULL)
+      brw_add_validated_bo(brw, region->buffer);
+}
 
-
-/***********************************************************************
- * Depthbuffer - currently constant, but rotation would change that.
- */
-
-static void upload_depthbuffer(struct brw_context *brw)
+static void emit_depthbuffer(struct brw_context *brw)
 {
-   /* 0x79050003  Depth Buffer */
    struct intel_context *intel = &brw->intel;
    struct intel_region *region = brw->state.depth_region;
-   struct brw_depthbuffer bd;
-   memset(&bd, 0, sizeof(bd));
-
-   bd.header.bits.opcode = CMD_DEPTH_BUFFER;
-   bd.header.bits.length = sizeof(bd)/4-2;
-   bd.dword1.bits.pitch = (region->pitch * region->cpp) - 1;
-   
-   switch (region->cpp) {
-   case 2:
-      bd.dword1.bits.format = BRW_DEPTHFORMAT_D16_UNORM;
-      break;
-   case 4:
-      if (intel->depth_buffer_is_float)
-	 bd.dword1.bits.format = BRW_DEPTHFORMAT_D32_FLOAT;
-      else
-	 bd.dword1.bits.format = BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
-      break;
-   default:
-      assert(0);
-      return;
+   unsigned int len = BRW_IS_G4X(brw) ? 6 : 5;
+
+   if (region == NULL) {
+      BEGIN_BATCH(len, IGNORE_CLIPRECTS);
+      OUT_BATCH(CMD_DEPTH_BUFFER << 16 | (len - 2));
+      OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
+		(BRW_SURFACE_NULL << 29));
+      OUT_BATCH(0);
+      OUT_BATCH(0);
+      OUT_BATCH(0);
+
+      if (BRW_IS_G4X(brw))
+         OUT_BATCH(0);
+
+      ADVANCE_BATCH();
+   } else {
+      unsigned int format;
+
+      switch (region->cpp) {
+      case 2:
+	 format = BRW_DEPTHFORMAT_D16_UNORM;
+	 break;
+      case 4:
+	 if (intel->depth_buffer_is_float)
+	    format = BRW_DEPTHFORMAT_D32_FLOAT;
+	 else
+	    format = BRW_DEPTHFORMAT_D24_UNORM_S8_UINT;
+	 break;
+      default:
+	 assert(0);
+	 return;
+      }
+
+      BEGIN_BATCH(len, IGNORE_CLIPRECTS);
+      OUT_BATCH(CMD_DEPTH_BUFFER << 16 | (len - 2));
+      OUT_BATCH(((region->pitch * region->cpp) - 1) |
+		(format << 18) |
+		(BRW_TILEWALK_YMAJOR << 26) |
+		((region->tiling != I915_TILING_NONE) << 27) |
+		(BRW_SURFACE_2D << 29));
+      OUT_RELOC(region->buffer,
+		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+		0);
+      OUT_BATCH((BRW_SURFACE_MIPMAPLAYOUT_BELOW << 1) |
+		((region->pitch - 1) << 6) |
+		((region->height - 1) << 19));
+      OUT_BATCH(0);
+
+      if (BRW_IS_G4X(brw))
+         OUT_BATCH(0);
+
+      ADVANCE_BATCH();
    }
-
-   bd.dword1.bits.depth_offset_disable = 0; /* coordinate offset */
-
-   /* The depthbuffer can only use YMAJOR tiling...  This is a bit of
-    * a shame as it clashes with the 2d blitter which only supports
-    * XMAJOR tiling...  
-    */
-   bd.dword1.bits.tile_walk = BRW_TILEWALK_YMAJOR;
-   bd.dword1.bits.tiled_surface = intel->depth_region->tiled;
-   bd.dword1.bits.surface_type = BRW_SURFACE_2D;
-
-   /* BRW_NEW_LOCK */
-   bd.dword2_base_addr = bmBufferOffset(intel, region->buffer);    
-
-   bd.dword3.bits.mipmap_layout = BRW_SURFACE_MIPMAPLAYOUT_BELOW;
-   bd.dword3.bits.lod = 0;
-   bd.dword3.bits.width = region->pitch - 1; /* XXX: width ? */
-   bd.dword3.bits.height = region->height - 1;
-
-   bd.dword4.bits.min_array_element = 0;
-   bd.dword4.bits.depth = 0;
-      
-   BRW_CACHED_BATCH_STRUCT(brw, &bd);
 }
 
 const struct brw_tracked_state brw_depthbuffer = {
    .dirty = {
       .mesa = 0,
-      .brw = BRW_NEW_CONTEXT | BRW_NEW_LOCK,
-      .cache = 0
+      .brw = BRW_NEW_DEPTH_BUFFER | BRW_NEW_BATCH,
+      .cache = 0,
    },
-   .update = upload_depthbuffer
+   .prepare = prepare_depthbuffer,
+   .emit = emit_depthbuffer,
 };
 
 
@@ -327,7 +303,7 @@ const struct brw_tracked_state brw_polygon_stipple = {
       .brw = 0,
       .cache = 0
    },
-   .update = upload_polygon_stipple
+   .emit = upload_polygon_stipple
 };
 
 
@@ -350,13 +326,42 @@ static void upload_polygon_stipple_offset(struct brw_context *brw)
    BRW_CACHED_BATCH_STRUCT(brw, &bpso);
 }
 
+#define _NEW_WINDOW_POS 0x40000000
+
 const struct brw_tracked_state brw_polygon_stipple_offset = {
    .dirty = {
       .mesa = _NEW_WINDOW_POS,
       .brw = 0,
       .cache = 0
    },
-   .update = upload_polygon_stipple_offset
+   .emit = upload_polygon_stipple_offset
+};
+
+/**********************************************************************
+ * AA Line parameters
+ */
+static void upload_aa_line_parameters(struct brw_context *brw)
+{
+   struct brw_aa_line_parameters balp;
+   
+   if (!BRW_IS_G4X(brw))
+      return;
+
+   /* use legacy aa line coverage computation */
+   memset(&balp, 0, sizeof(balp));
+   balp.header.opcode = CMD_AA_LINE_PARAMETERS;
+   balp.header.length = sizeof(balp) / 4 - 2;
+   
+   BRW_CACHED_BATCH_STRUCT(brw, &balp);
+}
+
+const struct brw_tracked_state brw_aa_line_parameters = {
+   .dirty = {
+      .mesa = 0,
+      .brw = BRW_NEW_CONTEXT,
+      .cache = 0
+   },
+   .emit = upload_aa_line_parameters
 };
 
 /***********************************************************************
@@ -391,41 +396,7 @@ const struct brw_tracked_state brw_line_stipple = {
       .brw = 0,
       .cache = 0
    },
-   .update = upload_line_stipple
-};
-
-
-
-/***********************************************************************
- * Misc constant state packets
- */
-
-static void upload_pipe_control(struct brw_context *brw)
-{
-   struct brw_pipe_control pc;
-
-   return;
-
-   memset(&pc, 0, sizeof(pc));
-
-   pc.header.opcode = CMD_PIPE_CONTROL;
-   pc.header.length = sizeof(pc)/4 - 2;
-   pc.header.post_sync_operation = PIPE_CONTROL_NOWRITE;
-
-   pc.header.instruction_state_cache_flush_enable = 1;
-
-   pc.bits1.dest_addr_type = PIPE_CONTROL_GTTWRITE_GLOBAL;
-
-   BRW_BATCH_STRUCT(brw, &pc);
-}
-
-const struct brw_tracked_state brw_pipe_control = {
-   .dirty = {
-      .mesa = 0,
-      .brw = BRW_NEW_CONTEXT,
-      .cache = 0
-   },
-   .update = upload_pipe_control
+   .emit = upload_line_stipple
 };
 
 
@@ -441,7 +412,7 @@ static void upload_invarient_state( struct brw_context *brw )
       struct brw_pipeline_select ps;
 
       memset(&ps, 0, sizeof(ps));
-      ps.header.opcode = CMD_PIPELINE_SELECT;
+      ps.header.opcode = CMD_PIPELINE_SELECT(brw);
       ps.header.pipeline_select = 0;
       BRW_BATCH_STRUCT(brw, &ps);
    }
@@ -477,7 +448,7 @@ static void upload_invarient_state( struct brw_context *brw )
       struct brw_vf_statistics vfs;
       memset(&vfs, 0, sizeof(vfs));
 
-      vfs.opcode = CMD_VF_STATISTICS;
+      vfs.opcode = CMD_VF_STATISTICS(brw);
       if (INTEL_DEBUG & DEBUG_STATS)
 	 vfs.statistics_enable = 1; 
 
@@ -491,43 +462,39 @@ const struct brw_tracked_state brw_invarient_state = {
       .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
-   .update = upload_invarient_state
+   .emit = upload_invarient_state
 };
 
-
-/* State pool addresses:
+/**
+ * Define the base addresses which some state is referenced from.
+ *
+ * This allows us to avoid having to emit relocations in many places for
+ * cached state, and instead emit pointers inside of large, mostly-static
+ * state pools.  This comes at the expense of memory, and more expensive cache
+ * misses.
  */
 static void upload_state_base_address( struct brw_context *brw )
 {
    struct intel_context *intel = &brw->intel;
-   struct brw_state_base_address sba;
-      
-   memset(&sba, 0, sizeof(sba));
-
-   sba.header.opcode = CMD_STATE_BASE_ADDRESS;
-   sba.header.length = 0x4;
-
-   /* BRW_NEW_LOCK */
-   sba.bits0.general_state_address = bmBufferOffset(intel, brw->pool[BRW_GS_POOL].buffer) >> 5;
-   sba.bits0.modify_enable = 1;
 
-   /* BRW_NEW_LOCK */
-   sba.bits1.surface_state_address = bmBufferOffset(intel, brw->pool[BRW_SS_POOL].buffer) >> 5;
-   sba.bits1.modify_enable = 1;
-
-   sba.bits2.modify_enable = 1;
-   sba.bits3.modify_enable = 1;
-   sba.bits4.modify_enable = 1;
-
-   BRW_CACHED_BATCH_STRUCT(brw, &sba);
+   /* Output the structure (brw_state_base_address) directly to the
+    * batchbuffer, so we can emit relocations inline.
+    */
+   BEGIN_BATCH(6, IGNORE_CLIPRECTS);
+   OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (6 - 2));
+   OUT_BATCH(1); /* General state base address */
+   OUT_BATCH(1); /* Surface state base address */
+   OUT_BATCH(1); /* Indirect object base address */
+   OUT_BATCH(1); /* General state upper bound */
+   OUT_BATCH(1); /* Indirect object upper bound */
+   ADVANCE_BATCH();
 }
 
-
 const struct brw_tracked_state brw_state_base_address = {
    .dirty = {
       .mesa = 0,
-      .brw = BRW_NEW_CONTEXT | BRW_NEW_LOCK,
-      .cache = 0
+      .brw = BRW_NEW_CONTEXT,
+      .cache = 0,
    },
-   .update = upload_state_base_address
+   .emit = upload_state_base_address
 };
diff --git a/src/mesa/drivers/dri/i965/brw_program.c b/src/mesa/drivers/dri/i965/brw_program.c
index 752fe49bcbf..a18dee85e80 100644
--- a/src/mesa/drivers/dri/i965/brw_program.c
+++ b/src/mesa/drivers/dri/i965/brw_program.c
@@ -29,15 +29,15 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
   
+#include "main/imports.h"
+#include "main/enums.h"
 #include "shader/prog_parameter.h"
-#include "brw_context.h"
-#include "brw_aub.h"
-#include "brw_util.h"
-#include "program.h"
-#include "imports.h"
-#include "enums.h"
+#include "shader/program.h"
+#include "shader/programopt.h"
 #include "tnl/tnl.h"
 
+#include "brw_context.h"
+#include "brw_util.h"
 
 static void brwBindProgram( GLcontext *ctx,
 			    GLenum target, 
@@ -117,7 +117,6 @@ static void brwProgramStringNotify( GLcontext *ctx,
       if (p == fp)
 	 brw->state.dirty.brw |= BRW_NEW_FRAGMENT_PROGRAM;
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
    }
    else if (target == GL_VERTEX_PROGRAM_ARB) {
       struct brw_context *brw = brw_context(ctx);
@@ -125,8 +124,10 @@ static void brwProgramStringNotify( GLcontext *ctx,
       struct brw_vertex_program *vp = (struct brw_vertex_program *)brw->vertex_program;
       if (p == vp)
 	 brw->state.dirty.brw |= BRW_NEW_VERTEX_PROGRAM;
+      if (p->program.IsPositionInvariant) {
+	 _mesa_insert_mvp_code(ctx, &p->program);
+      }
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
 
       /* Also tell tnl about it:
        */
diff --git a/src/mesa/drivers/dri/i965/brw_queryobj.c b/src/mesa/drivers/dri/i965/brw_queryobj.c
new file mode 100644
index 00000000000..cb9169e2eef
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/brw_queryobj.c
@@ -0,0 +1,259 @@
+/*
+ * Copyright © 2008 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+/** @file support for ARB_query_object
+ *
+ * ARB_query_object is implemented by using the PIPE_CONTROL command to stall
+ * execution on the completion of previous depth tests, and write the
+ * current PS_DEPTH_COUNT to a buffer object.
+ *
+ * We use before and after counts when drawing during a query so that
+ * we don't pick up other clients' query data in ours.  To reduce overhead,
+ * a single BO is used to record the query data for all active queries at
+ * once.  This also gives us a simple bound on how much batchbuffer space is
+ * required for handling queries, so that we can be sure that we won't
+ * have to emit a batchbuffer without getting the ending PS_DEPTH_COUNT.
+ */
+#include "main/simple_list.h"
+#include "main/imports.h"
+
+#include "brw_context.h"
+#include "brw_state.h"
+#include "intel_batchbuffer.h"
+#include "intel_reg.h"
+
+/** Waits on the query object's BO and totals the results for this query */
+static void
+brw_queryobj_get_results(struct brw_query_object *query)
+{
+   int i;
+   uint64_t *results;
+
+   if (query->bo == NULL)
+      return;
+
+   /* Map and count the pixels from the current query BO */
+   dri_bo_map(query->bo, GL_FALSE);
+   results = query->bo->virtual;
+   for (i = query->first_index; i <= query->last_index; i++) {
+      query->Base.Result += results[i * 2 + 1] - results[i * 2];
+   }
+   dri_bo_unmap(query->bo);
+
+   dri_bo_unreference(query->bo);
+   query->bo = NULL;
+}
+
+static struct gl_query_object *
+brw_new_query_object(GLcontext *ctx, GLuint id)
+{
+   struct brw_query_object *query;
+
+   query = _mesa_calloc(sizeof(struct brw_query_object));
+
+   query->Base.Id = id;
+   query->Base.Result = 0;
+   query->Base.Active = GL_FALSE;
+   query->Base.Ready = GL_TRUE;
+
+   return &query->Base;
+}
+
+static void
+brw_delete_query(GLcontext *ctx, struct gl_query_object *q)
+{
+   struct brw_query_object *query = (struct brw_query_object *)q;
+
+   dri_bo_unreference(query->bo);
+   _mesa_free(query);
+}
+
+static void
+brw_begin_query(GLcontext *ctx, struct gl_query_object *q)
+{
+   struct brw_context *brw = brw_context(ctx);
+   struct intel_context *intel = intel_context(ctx);
+   struct brw_query_object *query = (struct brw_query_object *)q;
+
+   /* Reset our driver's tracking of query state. */
+   dri_bo_unreference(query->bo);
+   query->bo = NULL;
+   query->first_index = -1;
+   query->last_index = -1;
+
+   insert_at_head(&brw->query.active_head, query);
+   intel->stats_wm++;
+}
+
+/**
+ * Begin the ARB_occlusion_query query on a query object.
+ */
+static void
+brw_end_query(GLcontext *ctx, struct gl_query_object *q)
+{
+   struct brw_context *brw = brw_context(ctx);
+   struct intel_context *intel = intel_context(ctx);
+   struct brw_query_object *query = (struct brw_query_object *)q;
+
+   /* Flush the batchbuffer in case it has writes to our query BO.
+    * Have later queries write to a new query BO so that further rendering
+    * doesn't delay the collection of our results.
+    */
+   if (query->bo) {
+      brw_emit_query_end(brw);
+      intel_batchbuffer_flush(intel->batch);
+
+      dri_bo_unreference(brw->query.bo);
+      brw->query.bo = NULL;
+   }
+
+   remove_from_list(query);
+
+   intel->stats_wm--;
+}
+
+static void brw_wait_query(GLcontext *ctx, struct gl_query_object *q)
+{
+   struct brw_query_object *query = (struct brw_query_object *)q;
+
+   brw_queryobj_get_results(query);
+   query->Base.Ready = GL_TRUE;
+}
+
+static void brw_check_query(GLcontext *ctx, struct gl_query_object *q)
+{
+   /* XXX: Need to expose dri_bo_is_idle from bufmgr. */
+#if 0
+   struct brw_query_object *query = (struct brw_query_object *)q;
+
+   if (dri_bo_is_idle(query->bo)) {
+      brw_queryobj_get_results(query);
+      query->Base.Ready = GL_TRUE;
+   }
+#else
+   brw_wait_query(ctx, q);
+#endif
+}
+
+/** Called to set up the query BO and account for its aperture space */
+void
+brw_prepare_query_begin(struct brw_context *brw)
+{
+   struct intel_context *intel = &brw->intel;
+
+   /* Skip if we're not doing any queries. */
+   if (is_empty_list(&brw->query.active_head))
+      return;
+
+   /* Get a new query BO if we're going to need it. */
+   if (brw->query.bo == NULL ||
+       brw->query.index * 2 + 1 >= 4096 / sizeof(uint64_t)) {
+      dri_bo_unreference(brw->query.bo);
+      brw->query.bo = NULL;
+
+      brw->query.bo = dri_bo_alloc(intel->bufmgr, "query", 4096, 1);
+      brw->query.index = 0;
+   }
+
+   brw_add_validated_bo(brw, brw->query.bo);
+}
+
+/** Called just before primitive drawing to get a beginning PS_DEPTH_COUNT. */
+void
+brw_emit_query_begin(struct brw_context *brw)
+{
+   struct intel_context *intel = &brw->intel;
+   struct brw_query_object *query;
+
+   /* Skip if we're not doing any queries, or we've emitted the start. */
+   if (brw->query.active || is_empty_list(&brw->query.active_head))
+      return;
+
+   BEGIN_BATCH(4, IGNORE_CLIPRECTS);
+   OUT_BATCH(_3DSTATE_PIPE_CONTROL |
+	     PIPE_CONTROL_DEPTH_STALL |
+	     PIPE_CONTROL_WRITE_DEPTH_COUNT);
+   /* This object could be mapped cacheable, but we don't have an exposed
+    * mechanism to support that.  Since it's going uncached, tell GEM that
+    * we're writing to it.  The usual clflush should be all that's required
+    * to pick up the results.
+    */
+   OUT_RELOC(brw->query.bo,
+	     I915_GEM_DOMAIN_INSTRUCTION, I915_GEM_DOMAIN_INSTRUCTION,
+	     PIPE_CONTROL_GLOBAL_GTT_WRITE |
+	     ((brw->query.index * 2) * sizeof(uint64_t)));
+   OUT_BATCH(0);
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+   foreach(query, &brw->query.active_head) {
+      if (query->bo != brw->query.bo) {
+	 if (query->bo != NULL)
+	    brw_queryobj_get_results(query);
+	 dri_bo_reference(brw->query.bo);
+	 query->bo = brw->query.bo;
+	 query->first_index = brw->query.index;
+      }
+      query->last_index = brw->query.index;
+   }
+   brw->query.active = GL_TRUE;
+}
+
+/** Called at batchbuffer flush to get an ending PS_DEPTH_COUNT */
+void
+brw_emit_query_end(struct brw_context *brw)
+{
+   struct intel_context *intel = &brw->intel;
+
+   if (!brw->query.active)
+      return;
+
+   BEGIN_BATCH(4, IGNORE_CLIPRECTS);
+   OUT_BATCH(_3DSTATE_PIPE_CONTROL |
+	     PIPE_CONTROL_DEPTH_STALL |
+	     PIPE_CONTROL_WRITE_DEPTH_COUNT);
+   OUT_RELOC(brw->query.bo,
+	     I915_GEM_DOMAIN_INSTRUCTION, I915_GEM_DOMAIN_INSTRUCTION,
+	     PIPE_CONTROL_GLOBAL_GTT_WRITE |
+	     ((brw->query.index * 2 + 1) * sizeof(uint64_t)));
+   OUT_BATCH(0);
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+   brw->query.active = GL_FALSE;
+   brw->query.index++;
+}
+
+void brw_init_queryobj_functions(struct dd_function_table *functions)
+{
+   functions->NewQueryObject = brw_new_query_object;
+   functions->DeleteQuery = brw_delete_query;
+   functions->BeginQuery = brw_begin_query;
+   functions->EndQuery = brw_end_query;
+   functions->CheckQuery = brw_check_query;
+   functions->WaitQuery = brw_wait_query;
+}
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index d5175399d6c..9dce6cd8e6c 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -30,9 +30,9 @@
   */
   
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "intel_batchbuffer.h"
 
@@ -43,8 +43,6 @@
 #include "brw_sf.h"
 #include "brw_state.h"
 
-#define DO_SETUP_BITS ((1<<FRAG_ATTRIB_MAX)-1)
-
 static void compile_sf_prog( struct brw_context *brw,
 			     struct brw_sf_prog_key *key )
 {
@@ -57,7 +55,7 @@ static void compile_sf_prog( struct brw_context *brw,
 
    /* Begin the compilation:
     */
-   brw_init_compile(&c.func);
+   brw_init_compile(brw, &c.func);
 
    c.key = *key;
    c.nr_attrs = brw_count_bits(c.key.attrs);
@@ -74,6 +72,11 @@ static void compile_sf_prog( struct brw_context *brw,
       if (c.key.attrs & (1<<i)) {
 	 c.attr_to_idx[i] = idx;
 	 c.idx_to_attr[idx] = i;
+	 if (i >= VERT_RESULT_TEX0 && i <= VERT_RESULT_TEX7) {
+		 c.point_attrs[i].CoordReplace = 
+			brw->attribs.Point->CoordReplace[i - VERT_RESULT_TEX0];
+	 } else
+		 c.point_attrs[i].CoordReplace = GL_FALSE;
 	 idx++;
       }
    
@@ -82,15 +85,18 @@ static void compile_sf_prog( struct brw_context *brw,
    switch (key->primitive) {
    case SF_TRIANGLES:
       c.nr_verts = 3;
-      brw_emit_tri_setup( &c );
+      brw_emit_tri_setup( &c, GL_TRUE );
       break;
    case SF_LINES:
       c.nr_verts = 2;
-      brw_emit_line_setup( &c );
+      brw_emit_line_setup( &c, GL_TRUE );
       break;
    case SF_POINTS:
       c.nr_verts = 1;
-      brw_emit_point_setup( &c );
+      if (key->do_point_sprite)
+	  brw_emit_point_sprite_setup( &c, GL_TRUE );
+      else
+	  brw_emit_point_setup( &c, GL_TRUE );
       break;
    case SF_UNFILLED_TRIS:
       c.nr_verts = 3;
@@ -108,29 +114,18 @@ static void compile_sf_prog( struct brw_context *brw,
 
    /* Upload
     */
-   brw->sf.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_SF_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->sf.prog_data );
+   dri_bo_unreference(brw->sf.prog_bo);
+   brw->sf.prog_bo = brw_upload_cache( &brw->cache, BRW_SF_PROG,
+				       &c.key, sizeof(c.key),
+				       NULL, 0,
+				       program, program_size,
+				       &c.prog_data,
+				       &brw->sf.prog_data );
 }
 
-
-static GLboolean search_cache( struct brw_context *brw, 
-			       struct brw_sf_prog_key *key )
-{
-   return brw_search_cache(&brw->cache[BRW_SF_PROG], 
-			   key, sizeof(*key),
-			   &brw->sf.prog_data,
-			   &brw->sf.prog_gs_offset);
-}
-
-
 /* Calculate interpolants for triangle and line rasterization.
  */
-static void upload_sf_prog( struct brw_context *brw )
+static void upload_sf_prog(struct brw_context *brw)
 {
    struct brw_sf_prog_key key;
 
@@ -162,7 +157,8 @@ static void upload_sf_prog( struct brw_context *brw )
       break;
    }
 
-
+   key.do_point_sprite = brw->attribs.Point->PointSprite;
+   key.SpriteOrigin = brw->attribs.Point->SpriteOrigin;
    /* _NEW_LIGHT */
    key.do_flat_shading = (brw->attribs.Light->ShadeModel == GL_FLAT);
    key.do_twoside_color = (brw->attribs.Light->Enabled && brw->attribs.Light->Model.TwoSide);
@@ -171,18 +167,22 @@ static void upload_sf_prog( struct brw_context *brw )
    if (key.do_twoside_color)
       key.frontface_ccw = (brw->attribs.Polygon->FrontFace == GL_CCW);
 
-
-   if (!search_cache(brw, &key))
+   dri_bo_unreference(brw->sf.prog_bo);
+   brw->sf.prog_bo = brw_search_cache(&brw->cache, BRW_SF_PROG,
+				      &key, sizeof(key),
+				      NULL, 0,
+				      &brw->sf.prog_data);
+   if (brw->sf.prog_bo == NULL)
       compile_sf_prog( brw, &key );
 }
 
 
 const struct brw_tracked_state brw_sf_prog = {
    .dirty = {
-      .mesa  = (_NEW_LIGHT|_NEW_POLYGON),
+      .mesa  = (_NEW_LIGHT|_NEW_POLYGON|_NEW_POINT),
       .brw   = (BRW_NEW_REDUCED_PRIMITIVE),
       .cache = CACHE_NEW_VS_PROG
    },
-   .update = upload_sf_prog
+   .prepare = upload_sf_prog
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index fb72b84ba8a..1c0fb70fe06 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -34,9 +34,9 @@
 #define BRW_SF_H
 
 
+#include "shader/program.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "program.h"
 
 
 #define SF_POINTS    0
@@ -45,14 +45,19 @@
 #define SF_UNFILLED_TRIS   3
 
 struct brw_sf_prog_key {
+   GLuint attrs:32;
    GLuint primitive:2;
    GLuint do_twoside_color:1;
    GLuint do_flat_shading:1;
-   GLuint attrs:16;
    GLuint frontface_ccw:1;
-   GLuint pad:11;
+   GLuint do_point_sprite:1;
+   GLuint pad:10;
+   GLenum SpriteOrigin;
 };
 
+struct brw_sf_point_tex {
+	GLboolean CoordReplace;	
+};
 
 struct brw_sf_compile {
    struct brw_compile func;
@@ -94,12 +99,14 @@ struct brw_sf_compile {
 
    GLubyte attr_to_idx[VERT_RESULT_MAX];   
    GLubyte idx_to_attr[VERT_RESULT_MAX];   
+   struct brw_sf_point_tex point_attrs[VERT_RESULT_MAX];
 };
 
  
-void brw_emit_tri_setup( struct brw_sf_compile *c );
-void brw_emit_line_setup( struct brw_sf_compile *c );
-void brw_emit_point_setup( struct brw_sf_compile *c );
+void brw_emit_tri_setup( struct brw_sf_compile *c, GLboolean allocate );
+void brw_emit_line_setup( struct brw_sf_compile *c, GLboolean allocate );
+void brw_emit_point_setup( struct brw_sf_compile *c, GLboolean allocate );
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c, GLboolean allocate );
 void brw_emit_anyprim_setup( struct brw_sf_compile *c );
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_sf_emit.c b/src/mesa/drivers/dri/i965/brw_sf_emit.c
index cbaf018c44a..ffdb0ae6df8 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_emit.c
@@ -30,9 +30,9 @@
   */
    
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 
 #include "intel_batchbuffer.h"
 
@@ -59,6 +59,35 @@ static GLboolean have_attr(struct brw_sf_compile *c,
    return (c->key.attrs & (1<<attr)) ? 1 : 0;
 }
 
+/**
+ * Sets VERT_RESULT_FOGC.Y  for gl_FrontFacing
+ *
+ * This is currently executed if the fragment program uses VERT_RESULT_FOGC
+ * at all, but this could be eliminated with a scan of the FP contents.
+ */
+static void
+do_front_facing( struct brw_sf_compile *c )
+{
+   struct brw_compile *p = &c->func; 
+   int i;
+
+   if (!have_attr(c, VERT_RESULT_FOGC))
+      return;
+
+   brw_push_insn_state(p);
+   brw_CMP(p, brw_null_reg(), 
+        c->key.frontface_ccw ? BRW_CONDITIONAL_G : BRW_CONDITIONAL_L,
+        c->det, brw_imm_f(0));
+   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+   for (i = 0; i < 3; i++) {
+       struct brw_reg fogc = get_vert_attr(c, c->vert[i],FRAG_ATTRIB_FOGC);
+       brw_MOV(p, get_element(fogc, 1), brw_imm_f(0));
+       brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+       brw_MOV(p, get_element(fogc, 1), brw_imm_f(1));
+       brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+   }
+   brw_pop_insn_state(p);
+}
 
 			 
 /*********************************************************************** 
@@ -343,15 +372,19 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
 
 
 
-void brw_emit_tri_setup( struct brw_sf_compile *c )
+void brw_emit_tri_setup( struct brw_sf_compile *c, GLboolean allocate)
 {
    struct brw_compile *p = &c->func;
    GLuint i;
 
    c->nr_verts = 3;
-   alloc_regs(c);
+
+   if (allocate)
+      alloc_regs(c);
+
    invert_det(c);
    copy_z_inv_w(c);
+   do_front_facing(c);
 
    if (c->key.do_twoside_color) 
       do_twoside_color(c);
@@ -428,14 +461,17 @@ void brw_emit_tri_setup( struct brw_sf_compile *c )
 
 
 
-void brw_emit_line_setup( struct brw_sf_compile *c )
+void brw_emit_line_setup( struct brw_sf_compile *c, GLboolean allocate)
 {
    struct brw_compile *p = &c->func;
    GLuint i;
 
 
    c->nr_verts = 2;
-   alloc_regs(c);
+
+   if (allocate)
+      alloc_regs(c);
+
    invert_det(c);
    copy_z_inv_w(c);
 
@@ -497,17 +533,103 @@ void brw_emit_line_setup( struct brw_sf_compile *c )
    } 
 }
 
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c, GLboolean allocate)
+{
+   struct brw_compile *p = &c->func;
+   GLuint i;
+
+   c->nr_verts = 1;
+
+   if (allocate)
+      alloc_regs(c);
+
+   copy_z_inv_w(c);
+   for (i = 0; i < c->nr_setup_regs; i++)
+   {
+      struct brw_sf_point_tex *tex = &c->point_attrs[c->idx_to_attr[2*i]];
+      struct brw_reg a0 = offset(c->vert[0], i);
+      GLushort pc, pc_persp, pc_linear;
+      GLboolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
+            
+      if (pc_persp)
+      {				
+	  if (!tex->CoordReplace) {
+	      brw_set_predicate_control_flag_value(p, pc_persp);
+	      brw_MUL(p, a0, a0, c->inv_w[0]);
+	  }
+      }
+
+      if (tex->CoordReplace) {
+	  /* Caculate 1.0/PointWidth */
+	  brw_math(&c->func,
+		  c->tmp,
+		  BRW_MATH_FUNCTION_INV,
+		  BRW_MATH_SATURATE_NONE,
+		  0,
+		  c->dx0,
+		  BRW_MATH_DATA_SCALAR,
+		  BRW_MATH_PRECISION_FULL);
+
+	  if (c->key.SpriteOrigin == GL_LOWER_LEFT) {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, negate(c->inv_w[0]));
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  } else {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  }
+      } else {
+	  brw_MOV(p, c->m1Cx, brw_imm_ud(0));
+	  brw_MOV(p, c->m2Cy, brw_imm_ud(0));
+      }
+
+      {
+	 brw_set_predicate_control_flag_value(p, pc); 
+	 if (tex->CoordReplace) {
+	     if (c->key.SpriteOrigin == GL_LOWER_LEFT) {
+		 brw_MUL(p, c->m3C0, c->inv_w[0], brw_imm_f(1.0));
+		 brw_MOV(p, vec1(suboffset(c->m3C0, 0)), brw_imm_f(0.0));
+	     }
+	     else
+		 brw_MOV(p, c->m3C0, brw_imm_f(0.0));
+	 } else {
+	 	brw_MOV(p, c->m3C0, a0); /* constant value */
+	 }
+
+	 /* Copy m0..m3 to URB. 
+	  */
+	 brw_urb_WRITE(p, 
+		       brw_null_reg(),
+		       0,
+		       brw_vec8_grf(0, 0),
+		       0, 	/* allocate */
+		       1,	/* used */
+		       4, 	/* msg len */
+		       0,	/* response len */
+		       last, 	/* eot */
+		       last, 	/* writes complete */
+		       i*4,	/* urb destination offset */
+		       BRW_URB_SWIZZLE_TRANSPOSE);
+      }
+   }
+}
 
 /* Points setup - several simplifications as all attributes are
  * constant across the face of the point (point sprites excluded!)
  */
-void brw_emit_point_setup( struct brw_sf_compile *c )
+void brw_emit_point_setup( struct brw_sf_compile *c, GLboolean allocate)
 {
    struct brw_compile *p = &c->func;
    GLuint i;
 
    c->nr_verts = 1;
-   alloc_regs(c);
+   
+   if (allocate)
+      alloc_regs(c);
+
    copy_z_inv_w(c);
 
    brw_MOV(p, c->m1Cx, brw_imm_ud(0)); /* zero - move out of loop */
@@ -561,10 +683,14 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
    struct brw_compile *p = &c->func;
    struct brw_reg ip = brw_ip_reg();
    struct brw_reg payload_prim = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0);
+   struct brw_reg payload_attr = get_element_ud(brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0), 0); 
    struct brw_reg primmask;
    struct brw_instruction *jmp;
    struct brw_reg v1_null_ud = vec1(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD));
+   
+   GLuint saveflag;
 
+   c->nr_verts = 3;
    alloc_regs(c);
 
    primmask = retype(get_element(c->tmp, 0), BRW_REGISTER_TYPE_UD);
@@ -582,8 +708,15 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
 					       (1<<_3DPRIM_TRIFAN_NOSTIPPLE)));
    jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
    {
-      brw_emit_tri_setup( c );
-      /* note - thread killed in subroutine */
+      saveflag = p->flag_value;
+      brw_push_insn_state(p); 
+      brw_emit_tri_setup( c, GL_FALSE );
+      brw_pop_insn_state(p);
+      p->flag_value = saveflag;
+      /* note - thread killed in subroutine, so must
+       * restore the flag which is changed when building
+       * the subroutine. fix #13240
+       */
    }
    brw_land_fwd_jump(p, jmp);
 
@@ -596,12 +729,28 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
 					       (1<<_3DPRIM_LINESTRIP_CONT_BF)));
    jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
    {
-      brw_emit_line_setup( c );
+      saveflag = p->flag_value;
+      brw_push_insn_state(p); 
+      brw_emit_line_setup( c, GL_FALSE );
+      brw_pop_insn_state(p);
+      p->flag_value = saveflag;
       /* note - thread killed in subroutine */
    }
    brw_land_fwd_jump(p, jmp); 
 
-   brw_emit_point_setup( c );
+   brw_set_conditionalmod(p, BRW_CONDITIONAL_Z);
+   brw_AND(p, v1_null_ud, payload_attr, brw_imm_ud(1<<BRW_SPRITE_POINT_ENABLE));
+   jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
+   {
+      saveflag = p->flag_value;
+      brw_push_insn_state(p); 
+      brw_emit_point_sprite_setup( c, GL_FALSE );
+      brw_pop_insn_state(p);
+      p->flag_value = saveflag;
+   }
+   brw_land_fwd_jump(p, jmp); 
+
+   brw_emit_point_setup( c, GL_FALSE );
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index 9a6e5f5f192..506126fcfb0 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -34,70 +34,70 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "macros.h"
+#include "main/macros.h"
+#include "intel_fbo.h"
 
 static void upload_sf_vp(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
+   const GLfloat depth_scale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
    struct brw_sf_viewport sfv;
+   struct intel_renderbuffer *irb =
+      intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
+   GLfloat y_scale, y_bias;
 
    memset(&sfv, 0, sizeof(sfv));
-   
-   if (brw->intel.driDrawable) 
-   {
-      /* _NEW_VIEWPORT, BRW_NEW_METAOPS */
-
-      if (!brw->metaops.active) {
-	 const GLfloat *v = brw->intel.ctx.Viewport._WindowMap.m;
-	 
-	 sfv.viewport.m00 =   v[MAT_SX];
-	 sfv.viewport.m11 = - v[MAT_SY];
-	 sfv.viewport.m22 =   v[MAT_SZ] * brw->intel.depth_scale;
-	 sfv.viewport.m30 =   v[MAT_TX];
-	 sfv.viewport.m31 = - v[MAT_TY] + brw->intel.driDrawable->h;
-	 sfv.viewport.m32 =   v[MAT_TZ] * brw->intel.depth_scale;
-      }
-      else {
-	 sfv.viewport.m00 =   1;
-	 sfv.viewport.m11 = - 1;
-	 sfv.viewport.m22 =   1;
-	 sfv.viewport.m30 =   0;
-	 sfv.viewport.m31 =   brw->intel.driDrawable->h;
-	 sfv.viewport.m32 =   0;
+
+   if (ctx->DrawBuffer->Name) {
+      /* User-created FBO */
+      if (irb && !irb->RenderToTexture) {
+	 y_scale = -1.0;
+	 y_bias = ctx->DrawBuffer->Height;
+      } else {
+	 y_scale = 1.0;
+	 y_bias = 0;
       }
+   } else {
+      y_scale = -1.0;
+      y_bias = ctx->DrawBuffer->Height;
    }
 
-   /* XXX: what state for this? */
-   if (brw->intel.driDrawable)
-   {
-      intelScreenPrivate *screen = brw->intel.intelScreen;
-      /* _NEW_SCISSOR */
-      GLint x = brw->attribs.Scissor->X;
-      GLint y = brw->attribs.Scissor->Y;
-      GLuint w = brw->attribs.Scissor->Width;
-      GLuint h = brw->attribs.Scissor->Height;
-
-      GLint x1 = x;
-      GLint y1 = brw->intel.driDrawable->h - (y + h);
-      GLint x2 = x + w - 1;
-      GLint y2 = y1 + h - 1;
-
-      if (x1 < 0) x1 = 0;
-      if (y1 < 0) y1 = 0;
-      if (x2 < 0) x2 = 0;
-      if (y2 < 0) y2 = 0;
-
-      if (x2 >= screen->width) x2 = screen->width-1;
-      if (y2 >= screen->height) y2 = screen->height-1;
-      if (x1 >= screen->width) x1 = screen->width-1;
-      if (y1 >= screen->height) y1 = screen->height-1;
-      
-      sfv.scissor.xmin = x1;
-      sfv.scissor.xmax = x2;
-      sfv.scissor.ymin = y1;
-      sfv.scissor.ymax = y2;
+   /* _NEW_VIEWPORT, BRW_NEW_METAOPS */
+
+   if (!brw->metaops.active) {
+      const GLfloat *v = ctx->Viewport._WindowMap.m;
+
+      sfv.viewport.m00 = v[MAT_SX];
+      sfv.viewport.m11 = v[MAT_SY] * y_scale;
+      sfv.viewport.m22 = v[MAT_SZ] * depth_scale;
+      sfv.viewport.m30 = v[MAT_TX];
+      sfv.viewport.m31 = v[MAT_TY] * y_scale + y_bias;
+      sfv.viewport.m32 = v[MAT_TZ] * depth_scale;
+   } else {
+      sfv.viewport.m00 =   1;
+      sfv.viewport.m11 = - 1;
+      sfv.viewport.m22 =   1;
+      sfv.viewport.m30 =   0;
+      sfv.viewport.m31 =   ctx->DrawBuffer->Height;
+      sfv.viewport.m32 =   0;
    }
 
-   brw->sf.vp_gs_offset = brw_cache_data( &brw->cache[BRW_SF_VP], &sfv );
+   /* _NEW_SCISSOR */
+
+   /* The scissor only needs to handle the intersection of drawable and
+    * scissor rect.  Clipping to the boundaries of static shared buffers
+    * for front/back/depth is covered by looping over cliprects in brw_draw.c.
+    *
+    * Note that the hardware's coordinates are inclusive, while Mesa's min is
+    * inclusive but max is exclusive.
+    */
+   sfv.scissor.xmin = ctx->DrawBuffer->_Xmin;
+   sfv.scissor.xmax = ctx->DrawBuffer->_Xmax - 1;
+   sfv.scissor.ymin = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymax;
+   sfv.scissor.ymax = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymin - 1;
+
+   dri_bo_unreference(brw->sf.vp_bo);
+   brw->sf.vp_bo = brw_cache_data( &brw->cache, BRW_SF_VP, &sfv, NULL, 0 );
 }
 
 const struct brw_tracked_state brw_sf_vp = {
@@ -107,87 +107,132 @@ const struct brw_tracked_state brw_sf_vp = {
       .brw   = BRW_NEW_METAOPS,
       .cache = 0
    },
-   .update = upload_sf_vp
+   .prepare = upload_sf_vp
 };
 
+struct brw_sf_unit_key {
+   unsigned int total_grf;
+   unsigned int urb_entry_read_length;
 
+   unsigned int nr_urb_entries, urb_size, sfsize;
 
-static void upload_sf_unit( struct brw_context *brw )
+   GLenum front_face, cull_face;
+   GLboolean scissor, line_smooth, point_sprite, point_attenuated;
+   float line_width;
+   float point_size;
+};
+
+static void
+sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key)
+{
+   memset(key, 0, sizeof(*key));
+
+   /* CACHE_NEW_SF_PROG */
+   key->total_grf = brw->sf.prog_data->total_grf;
+   key->urb_entry_read_length = brw->sf.prog_data->urb_read_length;
+
+   /* BRW_NEW_URB_FENCE */
+   key->nr_urb_entries = brw->urb.nr_sf_entries;
+   key->urb_size = brw->urb.vsize;
+   key->sfsize = brw->urb.sfsize;
+
+   key->scissor = brw->attribs.Scissor->Enabled;
+   key->front_face = brw->attribs.Polygon->FrontFace;
+
+   if (brw->attribs.Polygon->CullFlag)
+      key->cull_face = brw->attribs.Polygon->CullFaceMode;
+   else
+      key->cull_face = GL_NONE;
+
+   key->line_width = brw->attribs.Line->Width;
+   key->line_smooth = brw->attribs.Line->SmoothFlag;
+
+   key->point_sprite = brw->attribs.Point->PointSprite;
+   key->point_size = brw->attribs.Point->Size;
+   key->point_attenuated = brw->attribs.Point->_Attenuated;
+}
+
+static dri_bo *
+sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
+			dri_bo **reloc_bufs)
 {
    struct brw_sf_unit_state sf;
+   dri_bo *bo;
+
    memset(&sf, 0, sizeof(sf));
 
-   /* CACHE_NEW_SF_PROG */
-   sf.thread0.grf_reg_count = ((brw->sf.prog_data->total_grf-1) & ~15) / 16;
-   sf.thread0.kernel_start_pointer = brw->sf.prog_gs_offset >> 6;
-   sf.thread3.urb_entry_read_length = brw->sf.prog_data->urb_read_length;
+   sf.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
+   sf.thread0.kernel_start_pointer = brw->sf.prog_bo->offset >> 6; /* reloc */
 
    sf.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
+
    sf.thread3.dispatch_grf_start_reg = 3;
    sf.thread3.urb_entry_read_offset = 1;
+   sf.thread3.urb_entry_read_length = key->urb_entry_read_length;
 
-   /* BRW_NEW_URB_FENCE */
-   sf.thread4.nr_urb_entries = brw->urb.nr_sf_entries;
-   sf.thread4.urb_entry_allocation_size = brw->urb.sfsize - 1;
-   sf.thread4.max_threads = MIN2(12, brw->urb.nr_sf_entries / 2) - 1;
+   sf.thread4.nr_urb_entries = key->nr_urb_entries;
+   sf.thread4.urb_entry_allocation_size = key->sfsize - 1;
+   /* Each SF thread produces 1 PUE, and there can be up to 24 threads */
+   sf.thread4.max_threads = MIN2(24, key->nr_urb_entries) - 1;
 
    if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
-      sf.thread4.max_threads = 0; 
+      sf.thread4.max_threads = 0;
 
    if (INTEL_DEBUG & DEBUG_STATS)
-      sf.thread4.stats_enable = 1; 
+      sf.thread4.stats_enable = 1;
 
    /* CACHE_NEW_SF_VP */
-   sf.sf5.sf_viewport_state_offset = brw->sf.vp_gs_offset >> 5;
-   
+   sf.sf5.sf_viewport_state_offset = brw->sf.vp_bo->offset >> 5; /* reloc */
+
    sf.sf5.viewport_transform = 1;
-   
+
    /* _NEW_SCISSOR */
-   if (brw->attribs.Scissor->Enabled) 
-      sf.sf6.scissor = 1;  
+   if (key->scissor)
+      sf.sf6.scissor = 1;
 
    /* _NEW_POLYGON */
-   if (brw->attribs.Polygon->FrontFace == GL_CCW)
+   if (key->front_face == GL_CCW)
       sf.sf5.front_winding = BRW_FRONTWINDING_CCW;
    else
       sf.sf5.front_winding = BRW_FRONTWINDING_CW;
 
-   if (brw->attribs.Polygon->CullFlag) {
-      switch (brw->attribs.Polygon->CullFaceMode) {
-      case GL_FRONT:
-	 sf.sf6.cull_mode = BRW_CULLMODE_FRONT;
-	 break;
-      case GL_BACK:
-	 sf.sf6.cull_mode = BRW_CULLMODE_BACK;
-	 break;
-      case GL_FRONT_AND_BACK:
-	 sf.sf6.cull_mode = BRW_CULLMODE_BOTH;
-	 break;
-      default:
-	 assert(0);
-	 break;
-      }
-   }
-   else
+   switch (key->cull_face) {
+   case GL_FRONT:
+      sf.sf6.cull_mode = BRW_CULLMODE_FRONT;
+      break;
+   case GL_BACK:
+      sf.sf6.cull_mode = BRW_CULLMODE_BACK;
+      break;
+   case GL_FRONT_AND_BACK:
+      sf.sf6.cull_mode = BRW_CULLMODE_BOTH;
+      break;
+   case GL_NONE:
       sf.sf6.cull_mode = BRW_CULLMODE_NONE;
-      
+      break;
+   default:
+      assert(0);
+      break;
+   }
 
    /* _NEW_LINE */
    /* XXX use ctx->Const.Min/MaxLineWidth here */
-   sf.sf6.line_width = CLAMP(brw->attribs.Line->Width, 1.0, 5.0) * (1<<1);
+   sf.sf6.line_width = CLAMP(key->line_width, 1.0, 5.0) * (1<<1);
 
    sf.sf6.line_endcap_aa_region_width = 1;
-   if (brw->attribs.Line->SmoothFlag)
+   if (key->line_smooth)
       sf.sf6.aa_enable = 1;
-   else if (sf.sf6.line_width <= 0x2) 
-       sf.sf6.line_width = 0; 
+   else if (sf.sf6.line_width <= 0x2)
+       sf.sf6.line_width = 0;
 
    /* _NEW_POINT */
-   sf.sf6.point_rast_rule = 1;	/* opengl conventions */
+   sf.sf6.point_rast_rule = BRW_RASTRULE_UPPER_RIGHT;	/* opengl conventions */
    /* XXX clamp max depends on AA vs. non-AA */
-   sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 3.0) * (1<<3);
-   sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated;
-      
+
+   sf.sf7.sprite_point = key->point_sprite;
+   sf.sf7.point_size = CLAMP(nearbyint(key->point_size), 1, 255) * (1<<3);
+   sf.sf7.use_point_size_state = !key->point_attenuated;
+   sf.sf7.aa_line_distance_mode = 0;
+
    /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
     */
    sf.sf7.trifan_pv = 2;
@@ -200,9 +245,48 @@ static void upload_sf_unit( struct brw_context *brw )
    sf.sf6.dest_org_vbias = 0x8;
    sf.sf6.dest_org_hbias = 0x8;
 
-   brw->sf.state_gs_offset = brw_cache_data( &brw->cache[BRW_SF_UNIT], &sf );
+   bo = brw_upload_cache(&brw->cache, BRW_SF_UNIT,
+			 key, sizeof(*key),
+			 reloc_bufs, 2,
+			 &sf, sizeof(sf),
+			 NULL, NULL);
+
+   /* Emit SF program relocation */
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION, 0,
+		     sf.thread0.grf_reg_count << 1,
+		     offsetof(struct brw_sf_unit_state, thread0),
+		     brw->sf.prog_bo);
+
+   /* Emit SF viewport relocation */
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION, 0,
+		     sf.sf5.front_winding | (sf.sf5.viewport_transform << 1),
+		     offsetof(struct brw_sf_unit_state, sf5),
+		     brw->sf.vp_bo);
+
+   return bo;
 }
 
+static void upload_sf_unit( struct brw_context *brw )
+{
+   struct brw_sf_unit_key key;
+   dri_bo *reloc_bufs[2];
+
+   sf_unit_populate_key(brw, &key);
+
+   reloc_bufs[0] = brw->sf.prog_bo;
+   reloc_bufs[1] = brw->sf.vp_bo;
+
+   dri_bo_unreference(brw->sf.state_bo);
+   brw->sf.state_bo = brw_search_cache(&brw->cache, BRW_SF_UNIT,
+				       &key, sizeof(key),
+				       reloc_bufs, 2,
+				       NULL);
+   if (brw->sf.state_bo == NULL) {
+      brw->sf.state_bo = sf_unit_create_from_key(brw, &key, reloc_bufs);
+   }
+}
 
 const struct brw_tracked_state brw_sf_unit = {
    .dirty = {
@@ -215,7 +299,5 @@ const struct brw_tracked_state brw_sf_unit = {
       .cache = (CACHE_NEW_SF_VP |
 		CACHE_NEW_SF_PROG)
    },
-   .update = upload_sf_unit
+   .prepare = upload_sf_unit,
 };
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index b4cbdd7a380..bb22c03eeb6 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -35,6 +35,16 @@
 
 #include "brw_context.h"
 
+static inline void
+brw_add_validated_bo(struct brw_context *brw, dri_bo *bo)
+{
+   assert(brw->state.validated_bo_count < ARRAY_SIZE(brw->state.validated_bos));
+
+   if (bo != NULL) {
+      dri_bo_reference(bo);
+      brw->state.validated_bos[brw->state.validated_bo_count++] = bo;
+   }
+};
 
 const struct brw_tracked_state brw_blend_constant_color;
 const struct brw_tracked_state brw_cc_unit;
@@ -48,8 +58,8 @@ const struct brw_tracked_state brw_curbe_offsets;
 const struct brw_tracked_state brw_invarient_state;
 const struct brw_tracked_state brw_gs_prog;
 const struct brw_tracked_state brw_gs_unit;
-const struct brw_tracked_state brw_drawing_rect;
 const struct brw_tracked_state brw_line_stipple;
+const struct brw_tracked_state brw_aa_line_parameters;
 const struct brw_tracked_state brw_pipelined_state_pointers;
 const struct brw_tracked_state brw_binding_table_pointers;
 const struct brw_tracked_state brw_depthbuffer;
@@ -74,73 +84,72 @@ const struct brw_tracked_state brw_wm_unit;
 const struct brw_tracked_state brw_psp_urb_cbs;
 
 const struct brw_tracked_state brw_active_vertprog;
-const struct brw_tracked_state brw_tnl_vertprog;
 const struct brw_tracked_state brw_pipe_control;
 
 const struct brw_tracked_state brw_clear_surface_cache;
 const struct brw_tracked_state brw_clear_batch_cache;
 
+const struct brw_tracked_state brw_drawing_rect;
+const struct brw_tracked_state brw_indices;
+const struct brw_tracked_state brw_vertices;
+
+/***********************************************************************
+ * brw_state.c
+ */
+void brw_validate_state(struct brw_context *brw);
+void brw_upload_state(struct brw_context *brw);
+void brw_init_state(struct brw_context *brw);
+void brw_destroy_state(struct brw_context *brw);
+
 /***********************************************************************
  * brw_state_cache.c
  */
-GLuint brw_cache_data(struct brw_cache *cache,
-		      const void *data );
-
-GLuint brw_cache_data_sz(struct brw_cache *cache,
-			 const void *data,
-			 GLuint data_sz);
-
-GLuint brw_upload_cache( struct brw_cache *cache,
-			 const void *key,
-			 GLuint key_sz,
-			 const void *data,
-			 GLuint data_sz,
-			 const void *aux,
-			 void *aux_return );
-
-GLboolean brw_search_cache( struct brw_cache *cache,
-			    const void *key,
-			    GLuint key_size,
-			    void *aux_return,
-			    GLuint *offset_return);
-
-void brw_init_caches( struct brw_context *brw );
-void brw_destroy_caches( struct brw_context *brw );
+dri_bo *brw_cache_data(struct brw_cache *cache,
+		       enum brw_cache_id cache_id,
+		       const void *data,
+		       dri_bo **reloc_bufs,
+		       GLuint nr_reloc_bufs);
+
+dri_bo *brw_cache_data_sz(struct brw_cache *cache,
+			  enum brw_cache_id cache_id,
+			  const void *data,
+			  GLuint data_size,
+			  dri_bo **reloc_bufs,
+			  GLuint nr_reloc_bufs);
+
+dri_bo *brw_upload_cache( struct brw_cache *cache,
+			  enum brw_cache_id cache_id,
+			  const void *key,
+			  GLuint key_sz,
+			  dri_bo **reloc_bufs,
+			  GLuint nr_reloc_bufs,
+			  const void *data,
+			  GLuint data_sz,
+			  const void *aux,
+			  void *aux_return );
+
+dri_bo *brw_search_cache( struct brw_cache *cache,
+			  enum brw_cache_id cache_id,
+			  const void *key,
+			  GLuint key_size,
+			  dri_bo **reloc_bufs,
+			  GLuint nr_reloc_bufs,
+			  void *aux_return);
+void brw_state_cache_check_size( struct brw_context *brw );
+
+void brw_init_cache( struct brw_context *brw );
+void brw_destroy_cache( struct brw_context *brw );
 
 /***********************************************************************
  * brw_state_batch.c
  */
-#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data( brw->intel.batch, (s), sizeof(*(s)), 0)
+#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data( brw->intel.batch, (s), sizeof(*(s)), IGNORE_CLIPRECTS)
 #define BRW_CACHED_BATCH_STRUCT(brw, s) brw_cached_batch_struct( brw, (s), sizeof(*(s)) )
 
 GLboolean brw_cached_batch_struct( struct brw_context *brw,
 				   const void *data,
 				   GLuint sz );
-
 void brw_destroy_batch_cache( struct brw_context *brw );
-
-
-/***********************************************************************
- * brw_state_pool.c
- */
-void brw_init_pools( struct brw_context *brw );
-void brw_destroy_pools( struct brw_context *brw );
-
-GLboolean brw_pool_alloc( struct brw_mem_pool *pool,
-			  GLuint size,
-			  GLuint alignment,
-			  GLuint *offset_return);
-
-void brw_pool_fence( struct brw_context *brw,
-		     struct brw_mem_pool *pool,
-		     GLuint fence );
-
-
-void brw_pool_check_wrap( struct brw_context *brw,
-			  struct brw_mem_pool *pool );
-
-void brw_clear_all_caches( struct brw_context *brw );
-void brw_invalidate_pools( struct brw_context *brw );
 void brw_clear_batch_cache_flush( struct brw_context *brw );
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c
index 909b0acd121..dc87859f3f5 100644
--- a/src/mesa/drivers/dri/i965/brw_state_batch.c
+++ b/src/mesa/drivers/dri/i965/brw_state_batch.c
@@ -32,9 +32,8 @@
 
 
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "intel_batchbuffer.h"
-#include "imports.h"
+#include "main/imports.h"
 
 
 
@@ -49,7 +48,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw,
    struct header *newheader = (struct header *)data;
 
    if (brw->emit_state_always) {
-      intel_batchbuffer_data(brw->intel.batch, data, sz, 0);
+      intel_batchbuffer_data(brw->intel.batch, data, sz, IGNORE_CLIPRECTS);
       return GL_TRUE;
    }
 
@@ -76,7 +75,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw,
 
  emit:
    memcpy(item->header, newheader, sz);
-   intel_batchbuffer_data(brw->intel.batch, data, sz, 0);
+   intel_batchbuffer_data(brw->intel.batch, data, sz, IGNORE_CLIPRECTS);
    return GL_TRUE;
 }
 
@@ -92,21 +91,12 @@ static void clear_batch_cache( struct brw_context *brw )
    }
 
    brw->cached_batch_items = NULL;
-
-
-   brw_clear_all_caches(brw);
-
-   bmReleaseBuffers(&brw->intel);
-   
-   brw_invalidate_pools(brw);
 }
 
 void brw_clear_batch_cache_flush( struct brw_context *brw )
 {
    clear_batch_cache(brw);
 
-   brw->wrap = 0;
-   
 /*    brw_do_flush(brw, BRW_FLUSH_STATE_CACHE|BRW_FLUSH_READ_CACHE); */
    
    brw->state.dirty.mesa |= ~0;
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index 71c6938f9a3..d5b51664066 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -28,12 +28,37 @@
   * Authors:
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
-      
+
+/** @file brw_state_cache.c
+ *
+ * This file implements a simple static state cache for 965.  The consumers
+ * can query the hash table of state using a cache_id, opaque key data,
+ * and list of buffers that will be used in relocations, and receive the
+ * corresponding state buffer object of state (plus associated auxiliary
+ * data) in return.
+ *
+ * The inner workings are a simple hash table based on a CRC of the key data.
+ * The cache_id and relocation target buffers associated with the state
+ * buffer are included as auxiliary key data, but are not part of the hash
+ * value (this should be fixed, but will likely be fixed instead by making
+ * consumers use structured keys).
+ *
+ * Replacement is not implemented.  Instead, when the cache gets too big, at
+ * a safe point (unlock) we throw out all of the cache data and let it
+ * regenerate for the next rendering operation.
+ *
+ * The reloc_buf pointers need to be included as key data, otherwise the
+ * non-unique values stuffed in the offset in key data through
+ * brw_cache_data() may result in successful probe for state buffers
+ * even when the buffer being referenced doesn't match.  The result would be
+ * that the same state cache entry is used twice for different buffers,
+ * only one of the two buffers referenced gets put into the offset, and the
+ * incorrect program is run for the other instance.
+ */
 
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "intel_batchbuffer.h"
-#include "imports.h"
+#include "main/imports.h"
 
 /* XXX: Fixme - have to include these to get the sizes of the prog_key
  * structs:
@@ -44,16 +69,8 @@
 #include "brw_sf.h"
 #include "brw_gs.h"
 
-
-/***********************************************************************
- * Check cache for uploaded version of struct, else upload new one.
- * Fail when memory is exhausted.
- *
- * XXX: FIXME: Currently search is so slow it would be quicker to
- * regenerate the data every time...
- */
-
-static GLuint hash_key( const void *key, GLuint key_size )
+static GLuint hash_key( const void *key, GLuint key_size,
+			dri_bo **reloc_bufs, GLuint nr_reloc_bufs)
 {
    GLuint *ikey = (GLuint *)key;
    GLuint hash = 0, i;
@@ -62,23 +79,63 @@ static GLuint hash_key( const void *key, GLuint key_size )
 
    /* I'm sure this can be improved on:
     */
-   for (i = 0; i < key_size/4; i++)
+   for (i = 0; i < key_size/4; i++) {
       hash ^= ikey[i];
+      hash = (hash << 5) | (hash >> 27);
+   }
+
+   /* Include the BO pointers as key data as well */
+   ikey = (GLuint *)reloc_bufs;
+   key_size = nr_reloc_bufs * sizeof(dri_bo *);
+   for (i = 0; i < key_size/4; i++) {
+      hash ^= ikey[i];
+      hash = (hash << 5) | (hash >> 27);
+   }
 
    return hash;
 }
 
-static struct brw_cache_item *search_cache( struct brw_cache *cache,
-					     GLuint hash,
-					     const void *key,
-					     GLuint key_size)
+/**
+ * Marks a new buffer as being chosen for the given cache id.
+ */
+static void
+update_cache_last(struct brw_cache *cache, enum brw_cache_id cache_id,
+		  dri_bo *bo)
+{
+   if (bo == cache->last_bo[cache_id])
+      return; /* no change */
+
+   dri_bo_unreference(cache->last_bo[cache_id]);
+   cache->last_bo[cache_id] = bo;
+   dri_bo_reference(cache->last_bo[cache_id]);
+   cache->brw->state.dirty.cache |= 1 << cache_id;
+}
+
+static struct brw_cache_item *
+search_cache(struct brw_cache *cache, enum brw_cache_id cache_id,
+	     GLuint hash, const void *key, GLuint key_size,
+	     dri_bo **reloc_bufs, GLuint nr_reloc_bufs)
 {
    struct brw_cache_item *c;
 
+#if 0
+   int bucketcount = 0;
+
+   for (c = cache->items[hash % cache->size]; c; c = c->next)
+      bucketcount++;
+
+   fprintf(stderr, "bucket %d/%d = %d/%d items\n", hash % cache->size,
+	   cache->size, bucketcount, cache->n_items);
+#endif
+
    for (c = cache->items[hash % cache->size]; c; c = c->next) {
-      if (c->hash == hash && 
+      if (c->cache_id == cache_id &&
+	  c->hash == hash &&
 	  c->key_size == key_size &&
-	  memcmp(c->key, key, key_size) == 0)
+	  memcmp(c->key, key, key_size) == 0 &&
+	  c->nr_reloc_bufs == nr_reloc_bufs &&
+	  memcmp(c->reloc_bufs, reloc_bufs,
+		 nr_reloc_bufs * sizeof(dri_bo *)) == 0)
 	 return c;
    }
 
@@ -93,8 +150,7 @@ static void rehash( struct brw_cache *cache )
    GLuint size, i;
 
    size = cache->size * 3;
-   items = (struct brw_cache_item**) _mesa_malloc(size * sizeof(*items));
-   _mesa_memset(items, 0, size * sizeof(*items));
+   items = (struct brw_cache_item**) _mesa_calloc(size * sizeof(*items));
 
    for (i = 0; i < cache->size; i++)
       for (c = cache->items[i]; c; c = next) {
@@ -108,142 +164,180 @@ static void rehash( struct brw_cache *cache )
    cache->size = size;
 }
 
-
-GLboolean brw_search_cache( struct brw_cache *cache,
-			    const void *key,
-			    GLuint key_size,
-			    void *aux_return,
-			    GLuint *offset_return)
+/**
+ * Returns the buffer object matching cache_id and key, or NULL.
+ */
+dri_bo *brw_search_cache( struct brw_cache *cache,
+			  enum brw_cache_id cache_id,
+			  const void *key,
+			  GLuint key_size,
+			  dri_bo **reloc_bufs, GLuint nr_reloc_bufs,
+			  void *aux_return )
 {
    struct brw_cache_item *item;
-   GLuint addr = 0;
-   GLuint hash = hash_key(key, key_size);
+   GLuint hash = hash_key(key, key_size, reloc_bufs, nr_reloc_bufs);
 
-   item = search_cache(cache, hash, key, key_size);
+   item = search_cache(cache, cache_id, hash, key, key_size,
+		       reloc_bufs, nr_reloc_bufs);
 
-   if (item) {
-      if (aux_return) 
-	 *(void **)aux_return = (void *)((char *)item->key + item->key_size);
-      
-      *offset_return = addr = item->offset;
-   }    
-    
-   if (item == NULL || addr != cache->last_addr) {
-      cache->brw->state.dirty.cache |= 1<<cache->id;
-      cache->last_addr = addr;
-   }
-   
-   return item != NULL;
+   if (item == NULL)
+      return NULL;
+
+   if (aux_return)
+      *(void **)aux_return = (void *)((char *)item->key + item->key_size);
+
+   update_cache_last(cache, cache_id, item->bo);
+
+   dri_bo_reference(item->bo);
+   return item->bo;
 }
 
-GLuint brw_upload_cache( struct brw_cache *cache,
-			 const void *key,
-			 GLuint key_size,
-			 const void *data,
-			 GLuint data_size,
-			 const void *aux,
-			 void *aux_return )
-{   
-   GLuint offset;
+dri_bo *
+brw_upload_cache( struct brw_cache *cache,
+		  enum brw_cache_id cache_id,
+		  const void *key,
+		  GLuint key_size,
+		  dri_bo **reloc_bufs,
+		  GLuint nr_reloc_bufs,
+		  const void *data,
+		  GLuint data_size,
+		  const void *aux,
+		  void *aux_return )
+{
    struct brw_cache_item *item = CALLOC_STRUCT(brw_cache_item);
-   GLuint hash = hash_key(key, key_size);
-   void *tmp = _mesa_malloc(key_size + cache->aux_size);
-   
-   if (!brw_pool_alloc(cache->pool, data_size, 6, &offset)) {
-      /* Should not be possible: 
-       */
-      _mesa_printf("brw_pool_alloc failed\n");
-      exit(1);
-   }
+   GLuint hash = hash_key(key, key_size, reloc_bufs, nr_reloc_bufs);
+   GLuint relocs_size = nr_reloc_bufs * sizeof(dri_bo *);
+   GLuint aux_size = cache->aux_size[cache_id];
+   void *tmp;
+   dri_bo *bo;
+   int i;
+
+   /* Create the buffer object to contain the data */
+   bo = dri_bo_alloc(cache->brw->intel.bufmgr,
+		     cache->name[cache_id], data_size, 1 << 6);
+
+
+   /* Set up the memory containing the key, aux_data, and reloc_bufs */
+   tmp = _mesa_malloc(key_size + aux_size + relocs_size);
 
    memcpy(tmp, key, key_size);
+   memcpy(tmp + key_size, aux, cache->aux_size[cache_id]);
+   memcpy(tmp + key_size + aux_size, reloc_bufs, relocs_size);
+   for (i = 0; i < nr_reloc_bufs; i++) {
+      if (reloc_bufs[i] != NULL)
+	 dri_bo_reference(reloc_bufs[i]);
+   }
 
-   if (cache->aux_size)
-      memcpy(tmp+key_size, aux, cache->aux_size);
-	 
+   item->cache_id = cache_id;
    item->key = tmp;
    item->hash = hash;
    item->key_size = key_size;
-   item->offset = offset;
+   item->reloc_bufs = tmp + key_size + aux_size;
+   item->nr_reloc_bufs = nr_reloc_bufs;
+
+   item->bo = bo;
+   dri_bo_reference(bo);
    item->data_size = data_size;
 
-   if (++cache->n_items > cache->size * 1.5)
+   if (cache->n_items > cache->size * 1.5)
       rehash(cache);
-   
+
    hash %= cache->size;
    item->next = cache->items[hash];
    cache->items[hash] = item;
-      
+   cache->n_items++;
+
    if (aux_return) {
-      assert(cache->aux_size);
+      assert(cache->aux_size[cache_id]);
       *(void **)aux_return = (void *)((char *)item->key + item->key_size);
    }
 
    if (INTEL_DEBUG & DEBUG_STATE)
-      _mesa_printf("upload %s: %d bytes to pool buffer %d offset %x\n",
-		   cache->name,
-		   data_size, 
-		   cache->pool->buffer,
-		   offset);
+      _mesa_printf("upload %s: %d bytes to cache id %d\n",
+		   cache->name[cache_id],
+		   data_size, cache_id);
 
-   /* Copy data to the buffer:
-    */
-   bmBufferSubDataAUB(&cache->brw->intel,
-		      cache->pool->buffer,
-		      offset, 
-		      data_size, 
-		      data,
-		      cache->aub_type,
-		      cache->aub_sub_type);
-   
-
-   cache->brw->state.dirty.cache |= 1<<cache->id;
-   cache->last_addr = offset;
-
-   return offset;
+   /* Copy data to the buffer */
+   dri_bo_subdata(bo, 0, data_size, data);
+
+   update_cache_last(cache, cache_id, bo);
+
+   return bo;
 }
 
 /* This doesn't really work with aux data.  Use search/upload instead
  */
-GLuint brw_cache_data_sz(struct brw_cache *cache,
-			 const void *data,
-			 GLuint data_size)
+dri_bo *
+brw_cache_data_sz(struct brw_cache *cache,
+		  enum brw_cache_id cache_id,
+		  const void *data,
+		  GLuint data_size,
+		  dri_bo **reloc_bufs,
+		  GLuint nr_reloc_bufs)
 {
-   GLuint addr;
+   dri_bo *bo;
+   struct brw_cache_item *item;
+   GLuint hash = hash_key(data, data_size, reloc_bufs, nr_reloc_bufs);
 
-   if (!brw_search_cache(cache, data, data_size, NULL, &addr)) {
-      addr = brw_upload_cache(cache, 
-			      data, data_size, 
-			      data, data_size, 
-			      NULL, NULL);
+   item = search_cache(cache, cache_id, hash, data, data_size,
+		       reloc_bufs, nr_reloc_bufs);
+   if (item) {
+      update_cache_last(cache, cache_id, item->bo);
+      dri_bo_reference(item->bo);
+      return item->bo;
    }
 
-   return addr;
+   bo = brw_upload_cache(cache, cache_id,
+			 data, data_size,
+			 reloc_bufs, nr_reloc_bufs,
+			 data, data_size,
+			 NULL, NULL);
+
+   return bo;
 }
 
-GLuint brw_cache_data(struct brw_cache *cache,
-		      const void *data)
+/**
+ * Wrapper around brw_cache_data_sz using the cache_id's canonical key size.
+ *
+ * If nr_reloc_bufs is nonzero, brw_search_cache()/brw_upload_cache() would be
+ * better to use, as the potentially changing offsets in the data-used-as-key
+ * will result in excessive cache misses.
+ */
+dri_bo *
+brw_cache_data(struct brw_cache *cache,
+	       enum brw_cache_id cache_id,
+	       const void *data,
+	       dri_bo **reloc_bufs,
+	       GLuint nr_reloc_bufs)
 {
-   return brw_cache_data_sz(cache, data, cache->key_size);
+   return brw_cache_data_sz(cache, cache_id, data, cache->key_size[cache_id],
+			    reloc_bufs, nr_reloc_bufs);
 }
 
+enum pool_type {
+   DW_SURFACE_STATE,
+   DW_GENERAL_STATE
+};
+
+static void
+brw_init_cache_id( struct brw_context *brw,
+		const char *name,
+		enum brw_cache_id id,
+		GLuint key_size,
+		GLuint aux_size)
+{
+   struct brw_cache *cache = &brw->cache;
 
+   cache->name[id] = strdup(name);
+   cache->key_size[id] = key_size;
+   cache->aux_size[id] = aux_size;
+}
 
-
-
-static void brw_init_cache( struct brw_context *brw, 
-			    const char *name,
-			    GLuint id,
-			    GLuint key_size,
-			    GLuint aux_size,
-			    GLuint aub_type,
-			    GLuint aub_sub_type )
+void brw_init_cache( struct brw_context *brw )
 {
-   struct brw_cache *cache = &brw->cache[id];
+   struct brw_cache *cache = &brw->cache;
+
    cache->brw = brw;
-   cache->id = id;
-   cache->name = name;
-   cache->items = NULL;
 
    cache->size = 7;
    cache->n_items = 0;
@@ -251,200 +345,133 @@ static void brw_init_cache( struct brw_context *brw,
       _mesa_calloc(cache->size * 
 		   sizeof(struct brw_cache_item));
 
-
-   cache->key_size = key_size;
-   cache->aux_size = aux_size;
-   cache->aub_type = aub_type;
-   cache->aub_sub_type = aub_sub_type;
-   switch (aub_type) {
-   case DW_GENERAL_STATE: cache->pool = &brw->pool[BRW_GS_POOL]; break;
-   case DW_SURFACE_STATE: cache->pool = &brw->pool[BRW_SS_POOL]; break;
-   default: assert(0); break;
-   }
-}
-
-void brw_init_caches( struct brw_context *brw )
-{
-
-   brw_init_cache(brw,
-		  "CC_VP",
-		  BRW_CC_VP,
-		  sizeof(struct brw_cc_viewport),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_COLOR_CALC_VIEWPORT_STATE);
-
-   brw_init_cache(brw,
-		  "CC_UNIT",
-		  BRW_CC_UNIT,
-		  sizeof(struct brw_cc_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_COLOR_CALC_STATE);
-
-   brw_init_cache(brw,
-		  "WM_PROG",
-		  BRW_WM_PROG,
-		  sizeof(struct brw_wm_prog_key),
-		  sizeof(struct brw_wm_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
-
-   brw_init_cache(brw,
-		  "SAMPLER_DEFAULT_COLOR",
-		  BRW_SAMPLER_DEFAULT_COLOR,
-		  sizeof(struct brw_sampler_default_color),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_SAMPLER_DEFAULT_COLOR);
-
-   brw_init_cache(brw,
-		  "SAMPLER",
-		  BRW_SAMPLER,
-		  0,		/* variable key/data size */
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_SAMPLER_STATE);
-
-   brw_init_cache(brw,
-		  "WM_UNIT",
-		  BRW_WM_UNIT,
-		  sizeof(struct brw_wm_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_WINDOWER_IZ_STATE);
-
-   brw_init_cache(brw,
-		  "SF_PROG",
-		  BRW_SF_PROG,
-		  sizeof(struct brw_sf_prog_key),
-		  sizeof(struct brw_sf_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
-
-   brw_init_cache(brw,
-		  "SF_VP",
-		  BRW_SF_VP,
-		  sizeof(struct brw_sf_viewport),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_STRIPS_FANS_VIEWPORT_STATE);
-
-   brw_init_cache(brw,
-		  "SF_UNIT",
-		  BRW_SF_UNIT,
-		  sizeof(struct brw_sf_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_STRIPS_FANS_STATE);
-
-   brw_init_cache(brw,
-		  "VS_UNIT",
-		  BRW_VS_UNIT,
-		  sizeof(struct brw_vs_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_VERTEX_SHADER_STATE);
-
-   brw_init_cache(brw,
-		  "VS_PROG",
-		  BRW_VS_PROG,
-		  sizeof(struct brw_vs_prog_key),
-		  sizeof(struct brw_vs_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
-
-   brw_init_cache(brw,
-		  "CLIP_UNIT",
-		  BRW_CLIP_UNIT,
-		  sizeof(struct brw_clip_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_CLIPPER_STATE);
-
-   brw_init_cache(brw,
-		  "CLIP_PROG",
-		  BRW_CLIP_PROG,
-		  sizeof(struct brw_clip_prog_key),
-		  sizeof(struct brw_clip_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
-
-   brw_init_cache(brw,
-		  "GS_UNIT",
-		  BRW_GS_UNIT,
-		  sizeof(struct brw_gs_unit_state),
-		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_GEOMETRY_SHADER_STATE);
-
-   brw_init_cache(brw,
-		  "GS_PROG",
-		  BRW_GS_PROG,
-		  sizeof(struct brw_gs_prog_key),
-		  sizeof(struct brw_gs_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
-
-   brw_init_cache(brw,
-		  "SS_SURFACE",
-		  BRW_SS_SURFACE,
-		  sizeof(struct brw_surface_state),
-		  0,
-		  DW_SURFACE_STATE,
-		  DWSS_SURFACE_STATE);
-
-   brw_init_cache(brw,
-		  "SS_SURF_BIND",
-		  BRW_SS_SURF_BIND,
-		  sizeof(struct brw_surface_binding_table),
-		  0,
-		  DW_SURFACE_STATE,
-		  DWSS_BINDING_TABLE_STATE);
+   brw_init_cache_id(brw,
+		     "CC_VP",
+		     BRW_CC_VP,
+		     sizeof(struct brw_cc_viewport),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "CC_UNIT",
+		     BRW_CC_UNIT,
+		     sizeof(struct brw_cc_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "WM_PROG",
+		     BRW_WM_PROG,
+		     sizeof(struct brw_wm_prog_key),
+		     sizeof(struct brw_wm_prog_data));
+
+   brw_init_cache_id(brw,
+		     "SAMPLER_DEFAULT_COLOR",
+		     BRW_SAMPLER_DEFAULT_COLOR,
+		     sizeof(struct brw_sampler_default_color),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "SAMPLER",
+		     BRW_SAMPLER,
+		     0,		/* variable key/data size */
+		     0);
+
+   brw_init_cache_id(brw,
+		     "WM_UNIT",
+		     BRW_WM_UNIT,
+		     sizeof(struct brw_wm_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "SF_PROG",
+		     BRW_SF_PROG,
+		     sizeof(struct brw_sf_prog_key),
+		     sizeof(struct brw_sf_prog_data));
+
+   brw_init_cache_id(brw,
+		     "SF_VP",
+		     BRW_SF_VP,
+		     sizeof(struct brw_sf_viewport),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "SF_UNIT",
+		     BRW_SF_UNIT,
+		     sizeof(struct brw_sf_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "VS_UNIT",
+		     BRW_VS_UNIT,
+		     sizeof(struct brw_vs_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "VS_PROG",
+		     BRW_VS_PROG,
+		     sizeof(struct brw_vs_prog_key),
+		     sizeof(struct brw_vs_prog_data));
+
+   brw_init_cache_id(brw,
+		     "CLIP_UNIT",
+		     BRW_CLIP_UNIT,
+		     sizeof(struct brw_clip_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "CLIP_PROG",
+		     BRW_CLIP_PROG,
+		     sizeof(struct brw_clip_prog_key),
+		     sizeof(struct brw_clip_prog_data));
+
+   brw_init_cache_id(brw,
+		     "GS_UNIT",
+		     BRW_GS_UNIT,
+		     sizeof(struct brw_gs_unit_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "GS_PROG",
+		     BRW_GS_PROG,
+		     sizeof(struct brw_gs_prog_key),
+		     sizeof(struct brw_gs_prog_data));
+
+   brw_init_cache_id(brw,
+		     "SS_SURFACE",
+		     BRW_SS_SURFACE,
+		     sizeof(struct brw_surface_state),
+		     0);
+
+   brw_init_cache_id(brw,
+		     "SS_SURF_BIND",
+		     BRW_SS_SURF_BIND,
+		     0,
+		     0);
 }
 
-
-/* When we lose hardware context, need to invalidate the surface cache
- * as these structs must be explicitly re-uploaded.  They are subject
- * to fixup by the memory manager as they contain absolute agp
- * offsets, so we need to ensure there is a fresh version of the
- * struct available to receive the fixup.
- *
- * XXX: Need to ensure that there aren't two versions of a surface or
- * bufferobj with different backing data active in the same buffer at
- * once?  Otherwise the cache could confuse them.  Maybe better not to
- * cache at all?
- * 
- * --> Isn't this the same as saying need to ensure batch is flushed
- *         before new data is uploaded to an existing buffer?  We
- *         already try to make sure of that.
- */
-static void clear_cache( struct brw_cache *cache )
+static void
+brw_clear_cache( struct brw_context *brw )
 {
    struct brw_cache_item *c, *next;
    GLuint i;
 
-   for (i = 0; i < cache->size; i++) {
-      for (c = cache->items[i]; c; c = next) {
+   if (INTEL_DEBUG & DEBUG_STATE)
+      _mesa_printf("%s\n", __FUNCTION__);
+
+   for (i = 0; i < brw->cache.size; i++) {
+      for (c = brw->cache.items[i]; c; c = next) {
+	 int j;
+
 	 next = c->next;
+	 for (j = 0; j < c->nr_reloc_bufs; j++)
+	    dri_bo_unreference(c->reloc_bufs[j]);
+	 dri_bo_unreference(c->bo);
 	 free((void *)c->key);
 	 free(c);
       }
-      cache->items[i] = NULL;
+      brw->cache.items[i] = NULL;
    }
 
-   cache->n_items = 0;
-}
-
-void brw_clear_all_caches( struct brw_context *brw )
-{
-   GLint i;
-
-   if (INTEL_DEBUG & DEBUG_STATE)
-      _mesa_printf("%s\n", __FUNCTION__);
-
-   for (i = 0; i < BRW_MAX_CACHE; i++)
-      clear_cache(&brw->cache[i]);      
+   brw->cache.n_items = 0;
 
    if (brw->curbe.last_buf) {
       _mesa_free(brw->curbe.last_buf);
@@ -456,14 +483,25 @@ void brw_clear_all_caches( struct brw_context *brw )
    brw->state.dirty.cache |= ~0;
 }
 
+void brw_state_cache_check_size( struct brw_context *brw )
+{
+   /* un-tuned guess.  We've got around 20 state objects for a total of around
+    * 32k, so 1000 of them is around 1.5MB.
+    */
+   if (brw->cache.n_items > 1000)
+      brw_clear_cache(brw);
+}
 
-
-
-
-void brw_destroy_caches( struct brw_context *brw )
+void brw_destroy_cache( struct brw_context *brw )
 {
    GLuint i;
 
-   for (i = 0; i < BRW_MAX_CACHE; i++)
-      clear_cache(&brw->cache[i]);      
+   brw_clear_cache(brw);
+   for (i = 0; i < BRW_MAX_CACHE; i++) {
+      dri_bo_unreference(brw->cache.last_bo[i]);
+      free(brw->cache.name[i]);
+   }
+   free(brw->cache.items);
+   brw->cache.items = NULL;
+   brw->cache.size = 0;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_state_dump.c b/src/mesa/drivers/dri/i965/brw_state_dump.c
new file mode 100644
index 00000000000..b28c57c2bcf
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/brw_state_dump.c
@@ -0,0 +1,200 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+#include "main/mtypes.h"
+
+#include "brw_context.h"
+#include "brw_state.h"
+#include "brw_defines.h"
+
+/**
+ * Prints out a header, the contents, and the message associated with
+ * the hardware state data given.
+ *
+ * \param name Name of the state object
+ * \param data Pointer to the base of the state object
+ * \param hw_offset Hardware offset of the base of the state data.
+ * \param index Index of the DWORD being output.
+ */
+static void
+state_out(const char *name, void *data, uint32_t hw_offset, int index,
+	  char *fmt, ...)
+{
+    va_list va;
+
+    fprintf(stderr, "%8s: 0x%08x: 0x%08x: ",
+	    name, hw_offset + index * 4, ((uint32_t *)data)[index]);
+    va_start(va, fmt);
+    vfprintf(stderr, fmt, va);
+    va_end(va);
+}
+
+/** Generic, undecoded state buffer debug printout */
+static void
+state_struct_out(const char *name, dri_bo *buffer, unsigned int state_size)
+{
+   int i;
+
+   if (buffer == NULL)
+      return;
+
+   dri_bo_map(buffer, GL_FALSE);
+   for (i = 0; i < state_size / 4; i++) {
+      state_out(name, buffer->virtual, buffer->offset, i,
+		"dword %d\n", i);
+   }
+   dri_bo_unmap(buffer);
+}
+
+static const char *
+get_965_surfacetype(unsigned int surfacetype)
+{
+    switch (surfacetype) {
+    case 0: return "1D";
+    case 1: return "2D";
+    case 2: return "3D";
+    case 3: return "CUBE";
+    case 4: return "BUFFER";
+    case 7: return "NULL";
+    default: return "unknown";
+    }
+}
+
+static void dump_wm_surface_state(struct brw_context *brw)
+{
+   int i;
+
+   for (i = 0; i < brw->wm.nr_surfaces; i++) {
+      dri_bo *surf_bo = brw->wm.surf_bo[i];
+      unsigned int surfoff;
+      struct brw_surface_state *surf;
+      char name[20];
+
+      if (surf_bo == NULL) {
+	 fprintf(stderr, "WM SS%d: NULL\n", i);
+	 continue;
+      }
+      dri_bo_map(surf_bo, GL_FALSE);
+      surfoff = surf_bo->offset;
+      surf = (struct brw_surface_state *)(surf_bo->virtual);
+
+      sprintf(name, "WM SS%d", i);
+      state_out(name, surf, surfoff, 0, "%s\n",
+		get_965_surfacetype(surf->ss0.surface_type));
+      state_out(name, surf, surfoff, 1, "offset\n");
+      state_out(name, surf, surfoff, 2, "%dx%d size, %d mips\n",
+		surf->ss2.width + 1, surf->ss2.height + 1, surf->ss2.mip_count);
+      state_out(name, surf, surfoff, 3, "pitch %d, %stiled\n",
+		surf->ss3.pitch + 1, surf->ss3.tiled_surface ? "" : "not ");
+      state_out(name, surf, surfoff, 4, "mip base %d\n",
+		surf->ss4.min_lod);
+
+      dri_bo_unmap(surf_bo);
+   }
+}
+
+static void dump_sf_viewport_state(struct brw_context *brw)
+{
+   const char *name = "SF VP";
+   struct brw_sf_viewport *vp;
+   uint32_t vp_off;
+
+   if (brw->sf.vp_bo == NULL)
+      return;
+
+   dri_bo_map(brw->sf.vp_bo, GL_FALSE);
+
+   vp = brw->sf.vp_bo->virtual;
+   vp_off = brw->sf.vp_bo->offset;
+
+   state_out(name, vp, vp_off, 0, "m00 = %f\n", vp->viewport.m00);
+   state_out(name, vp, vp_off, 1, "m11 = %f\n", vp->viewport.m11);
+   state_out(name, vp, vp_off, 2, "m22 = %f\n", vp->viewport.m22);
+   state_out(name, vp, vp_off, 3, "m30 = %f\n", vp->viewport.m30);
+   state_out(name, vp, vp_off, 4, "m31 = %f\n", vp->viewport.m31);
+   state_out(name, vp, vp_off, 5, "m32 = %f\n", vp->viewport.m32);
+
+   state_out(name, vp, vp_off, 6, "top left = %d,%d\n",
+	     vp->scissor.xmin, vp->scissor.ymin);
+   state_out(name, vp, vp_off, 7, "bottom right = %d,%d\n",
+	     vp->scissor.xmax, vp->scissor.ymax);
+
+   dri_bo_unmap(brw->sf.vp_bo);
+}
+
+static void brw_debug_prog(const char *name, dri_bo *prog)
+{
+   unsigned int i;
+   uint32_t *data;
+
+   if (prog == NULL)
+      return;
+
+   dri_bo_map(prog, GL_FALSE);
+
+   data = prog->virtual;
+
+   for (i = 0; i < prog->size / 4 / 4; i++) {
+      fprintf(stderr, "%8s: 0x%08x: 0x%08x 0x%08x 0x%08x 0x%08x\n",
+	      name, (unsigned int)prog->offset + i * 4 * 4,
+	      data[i * 4], data[i * 4 + 1], data[i * 4 + 2], data[i * 4 + 3]);
+   }
+
+   dri_bo_unmap(prog);
+}
+
+
+/**
+ * Print additional debug information associated with the batchbuffer
+ * when DEBUG_BATCH is set.
+ *
+ * For 965, this means mapping the state buffers that would have been referenced
+ * by the batchbuffer and dumping them.
+ *
+ * The buffer offsets printed rely on the buffer containing the last offset
+ * it was validated at.
+ */
+void brw_debug_batch(struct intel_context *intel)
+{
+   struct brw_context *brw = brw_context(&intel->ctx);
+
+   state_struct_out("WM bind", brw->wm.bind_bo, 4 * brw->wm.nr_surfaces);
+   dump_wm_surface_state(brw);
+
+   state_struct_out("VS", brw->vs.state_bo, sizeof(struct brw_vs_unit_state));
+   brw_debug_prog("VS prog", brw->vs.prog_bo);
+
+   state_struct_out("GS", brw->gs.state_bo, sizeof(struct brw_gs_unit_state));
+   brw_debug_prog("GS prog", brw->gs.prog_bo);
+
+   state_struct_out("SF", brw->sf.state_bo, sizeof(struct brw_sf_unit_state));
+   dump_sf_viewport_state(brw);
+   brw_debug_prog("SF prog", brw->sf.prog_bo);
+
+   state_struct_out("WM", brw->wm.state_bo, sizeof(struct brw_wm_unit_state));
+   brw_debug_prog("WM prog", brw->wm.prog_bo);
+}
diff --git a/src/mesa/drivers/dri/i965/brw_state_pool.c b/src/mesa/drivers/dri/i965/brw_state_pool.c
deleted file mode 100644
index b9926f2a5d7..00000000000
--- a/src/mesa/drivers/dri/i965/brw_state_pool.c
+++ /dev/null
@@ -1,154 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <keith@tungstengraphics.com>
-  */
-       
-
-#include "brw_state.h"
-#include "imports.h"
-
-#include "intel_ioctl.h"
-#include "bufmgr.h"
-
-GLboolean brw_pool_alloc( struct brw_mem_pool *pool,
-			  GLuint size,
-			  GLuint align,
-			  GLuint *offset_return)
-{
-   GLuint align_mask = (1<<align)-1;
-   GLuint fixup = ((pool->offset + align_mask) & ~align_mask) - pool->offset;
-
-   size = (size + 3) & ~3;
-
-   if (pool->offset + fixup + size >= pool->size) {
-      _mesa_printf("%s failed\n", __FUNCTION__);
-      assert(0);
-      exit(0);
-   }
-
-   pool->offset += fixup;
-   *offset_return = pool->offset;
-   pool->offset += size; 
-  
-   return GL_TRUE;
-}
-
-static
-void brw_invalidate_pool( struct intel_context *intel,
-			  struct brw_mem_pool *pool )
-{
-   if (INTEL_DEBUG & DEBUG_STATE)
-      _mesa_printf("\n\n\n %s \n\n\n", __FUNCTION__);
-   
-   bmBufferData(intel,
-		pool->buffer,
-		pool->size,
-		NULL,
-		0); 
-
-   pool->offset = 0;
-
-   brw_clear_all_caches(pool->brw);
-}
-
-static void brw_invalidate_pool_cb( struct intel_context *intel, void *ptr )
-{
-   struct brw_mem_pool *pool = (struct brw_mem_pool *) ptr;
-
-   pool->offset = 0;
-   brw_clear_all_caches(pool->brw);
-}
-
-
-
-static void brw_init_pool( struct brw_context *brw,
-			   GLuint pool_id,
-			   GLuint size )
-{
-   struct brw_mem_pool *pool = &brw->pool[pool_id];
-
-   pool->size = size;   
-   pool->brw = brw;
-   
-   bmGenBuffers(&brw->intel, "pool", 1, &pool->buffer, 12);
-
-   /* Also want to say not to wait on fences when data is presented
-    */
-   bmBufferSetInvalidateCB(&brw->intel, pool->buffer, 
-			   brw_invalidate_pool_cb, 
-			   pool,
-			   GL_TRUE);   
-
-   bmBufferData(&brw->intel,
-		pool->buffer,
-		pool->size,
-		NULL,
-		0); 
-
-}
-
-static void brw_destroy_pool( struct brw_context *brw,
-			      GLuint pool_id )
-{
-   struct brw_mem_pool *pool = &brw->pool[pool_id];
-   
-   bmDeleteBuffers(&brw->intel, 1, &pool->buffer);
-}
-
-
-void brw_pool_check_wrap( struct brw_context *brw,
-			  struct brw_mem_pool *pool )
-{
-   if (pool->offset > (pool->size * 3) / 4) {
-      if (brw->intel.aub_file)
-	 brw->intel.aub_wrap = 1;
-      else
-	 brw->state.dirty.brw |= BRW_NEW_CONTEXT;
-   }
-
-}
-
-void brw_init_pools( struct brw_context *brw )
-{
-   brw_init_pool(brw, BRW_GS_POOL, 0x80000);
-   brw_init_pool(brw, BRW_SS_POOL, 0x80000);
-}
-
-void brw_destroy_pools( struct brw_context *brw )
-{
-   brw_destroy_pool(brw, BRW_GS_POOL);
-   brw_destroy_pool(brw, BRW_SS_POOL);
-}
-
-
-void brw_invalidate_pools( struct brw_context *brw )
-{
-   brw_invalidate_pool(&brw->intel, &brw->pool[BRW_GS_POOL]);
-   brw_invalidate_pool(&brw->intel, &brw->pool[BRW_SS_POOL]);
-}
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index 92c07c29624..5124535f2a6 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -33,7 +33,6 @@
 
 #include "brw_context.h"
 #include "brw_state.h"
-#include "bufmgr.h"
 #include "intel_batchbuffer.h"
 
 /* This is used to initialize brw->state.atoms[].  We could use this
@@ -46,8 +45,6 @@ const struct brw_tracked_state *atoms[] =
 {
    &brw_check_fallback,
 
-   &brw_tnl_vertprog,
-   &brw_active_vertprog,
    &brw_wm_input_sizes,
    &brw_vs_prog,
    &brw_gs_prog, 
@@ -80,19 +77,17 @@ const struct brw_tracked_state *atoms[] =
     */
    &brw_invarient_state,
    &brw_state_base_address,
-   &brw_pipe_control,
 
    &brw_binding_table_pointers,
    &brw_blend_constant_color,
 
-   &brw_drawing_rect,
    &brw_depthbuffer,
 
    &brw_polygon_stipple,
    &brw_polygon_stipple_offset,
 
    &brw_line_stipple,
-
+   &brw_aa_line_parameters,
    /* Ordering of the commands below is documented as fixed.  
     */
 #if 0
@@ -103,6 +98,9 @@ const struct brw_tracked_state *atoms[] =
    &brw_psp_urb_cbs,
 #endif
 
+   &brw_drawing_rect,
+   &brw_indices,
+   &brw_vertices,
 
    NULL,			/* brw_constant_buffer */
 };
@@ -112,8 +110,7 @@ void brw_init_state( struct brw_context *brw )
 {
    GLuint i;
 
-   brw_init_pools(brw);
-   brw_init_caches(brw);
+   brw_init_cache(brw);
 
    brw->state.atoms = _mesa_malloc(sizeof(atoms));
    brw->state.nr_atoms = sizeof(atoms)/sizeof(*atoms);
@@ -138,9 +135,8 @@ void brw_destroy_state( struct brw_context *brw )
       brw->state.atoms = NULL;
    }
 
-   brw_destroy_caches(brw);
+   brw_destroy_cache(brw);
    brw_destroy_batch_cache(brw);
-   brw_destroy_pools(brw);   
 }
 
 /***********************************************************************
@@ -172,20 +168,146 @@ static void xor_states( struct brw_state_flags *result,
    result->cache = a->cache ^ b->cache;
 }
 
+static void
+brw_clear_validated_bos(struct brw_context *brw)
+{
+   int i;
+
+   /* Clear the last round of validated bos */
+   for (i = 0; i < brw->state.validated_bo_count; i++) {
+      dri_bo_unreference(brw->state.validated_bos[i]);
+      brw->state.validated_bos[i] = NULL;
+   }
+   brw->state.validated_bo_count = 0;
+}
+
+struct dirty_bit_map {
+   uint32_t bit;
+   char *name;
+   uint32_t count;
+};
+
+#define DEFINE_BIT(name) {name, #name, 0}
+
+static struct dirty_bit_map mesa_bits[] = {
+   DEFINE_BIT(_NEW_MODELVIEW),
+   DEFINE_BIT(_NEW_PROJECTION),
+   DEFINE_BIT(_NEW_TEXTURE_MATRIX),
+   DEFINE_BIT(_NEW_COLOR_MATRIX),
+   DEFINE_BIT(_NEW_ACCUM),
+   DEFINE_BIT(_NEW_COLOR),
+   DEFINE_BIT(_NEW_DEPTH),
+   DEFINE_BIT(_NEW_EVAL),
+   DEFINE_BIT(_NEW_FOG),
+   DEFINE_BIT(_NEW_HINT),
+   DEFINE_BIT(_NEW_LIGHT),
+   DEFINE_BIT(_NEW_LINE),
+   DEFINE_BIT(_NEW_PIXEL),
+   DEFINE_BIT(_NEW_POINT),
+   DEFINE_BIT(_NEW_POLYGON),
+   DEFINE_BIT(_NEW_POLYGONSTIPPLE),
+   DEFINE_BIT(_NEW_SCISSOR),
+   DEFINE_BIT(_NEW_STENCIL),
+   DEFINE_BIT(_NEW_TEXTURE),
+   DEFINE_BIT(_NEW_TRANSFORM),
+   DEFINE_BIT(_NEW_VIEWPORT),
+   DEFINE_BIT(_NEW_PACKUNPACK),
+   DEFINE_BIT(_NEW_ARRAY),
+   DEFINE_BIT(_NEW_RENDERMODE),
+   DEFINE_BIT(_NEW_BUFFERS),
+   DEFINE_BIT(_NEW_MULTISAMPLE),
+   DEFINE_BIT(_NEW_TRACK_MATRIX),
+   DEFINE_BIT(_NEW_PROGRAM),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map brw_bits[] = {
+   DEFINE_BIT(BRW_NEW_URB_FENCE),
+   DEFINE_BIT(BRW_NEW_FRAGMENT_PROGRAM),
+   DEFINE_BIT(BRW_NEW_VERTEX_PROGRAM),
+   DEFINE_BIT(BRW_NEW_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_CURBE_OFFSETS),
+   DEFINE_BIT(BRW_NEW_REDUCED_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_CONTEXT),
+   DEFINE_BIT(BRW_NEW_WM_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_INPUT_VARYING),
+   DEFINE_BIT(BRW_NEW_PSP),
+   DEFINE_BIT(BRW_NEW_METAOPS),
+   DEFINE_BIT(BRW_NEW_FENCE),
+   DEFINE_BIT(BRW_NEW_INDICES),
+   DEFINE_BIT(BRW_NEW_VERTICES),
+   DEFINE_BIT(BRW_NEW_BATCH),
+   DEFINE_BIT(BRW_NEW_DEPTH_BUFFER),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map cache_bits[] = {
+   DEFINE_BIT(CACHE_NEW_CC_VP),
+   DEFINE_BIT(CACHE_NEW_CC_UNIT),
+   DEFINE_BIT(CACHE_NEW_WM_PROG),
+   DEFINE_BIT(CACHE_NEW_SAMPLER_DEFAULT_COLOR),
+   DEFINE_BIT(CACHE_NEW_SAMPLER),
+   DEFINE_BIT(CACHE_NEW_WM_UNIT),
+   DEFINE_BIT(CACHE_NEW_SF_PROG),
+   DEFINE_BIT(CACHE_NEW_SF_VP),
+   DEFINE_BIT(CACHE_NEW_SF_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_PROG),
+   DEFINE_BIT(CACHE_NEW_GS_UNIT),
+   DEFINE_BIT(CACHE_NEW_GS_PROG),
+   DEFINE_BIT(CACHE_NEW_CLIP_VP),
+   DEFINE_BIT(CACHE_NEW_CLIP_UNIT),
+   DEFINE_BIT(CACHE_NEW_CLIP_PROG),
+   DEFINE_BIT(CACHE_NEW_SURFACE),
+   DEFINE_BIT(CACHE_NEW_SURF_BIND),
+   {0, 0, 0}
+};
+
+
+static void
+brw_update_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      if (bit_map[i].bit & bits)
+	 bit_map[i].count++;
+   }
+}
+
+static void
+brw_print_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      fprintf(stderr, "0x%08x: %12d (%s)\n",
+	      bit_map[i].bit, bit_map[i].count, bit_map[i].name);
+   }
+}
 
 /***********************************************************************
  * Emit all state:
  */
 void brw_validate_state( struct brw_context *brw )
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_state_flags *state = &brw->state.dirty;
    GLuint i;
 
+   brw_clear_validated_bos(brw);
+
    state->mesa |= brw->intel.NewGLState;
    brw->intel.NewGLState = 0;
 
-   if (brw->wrap)
-      state->brw |= BRW_NEW_CONTEXT;
+   brw_add_validated_bo(brw, intel->batch->buf);
 
    if (brw->emit_state_always) {
       state->mesa |= ~0;
@@ -201,6 +323,10 @@ void brw_validate_state( struct brw_context *brw )
       brw->state.dirty.brw |= BRW_NEW_FRAGMENT_PROGRAM;
    }
 
+   if (brw->vertex_program != brw->attribs.VertexProgram->_Current) {
+      brw->vertex_program = brw->attribs.VertexProgram->_Current;
+      brw->state.dirty.brw |= BRW_NEW_VERTEX_PROGRAM;
+   }
 
    if (state->mesa == 0 &&
        state->cache == 0 &&
@@ -210,13 +336,31 @@ void brw_validate_state( struct brw_context *brw )
    if (brw->state.dirty.brw & BRW_NEW_CONTEXT)
       brw_clear_batch_cache_flush(brw);
 
+   brw->intel.Fallback = 0;
 
-   /* Make an early reference to the state pools, as we don't cope
-    * well with them being evicted from here down.
-    */
-   (void)bmBufferOffset(&brw->intel, brw->pool[BRW_GS_POOL].buffer);
-   (void)bmBufferOffset(&brw->intel, brw->pool[BRW_SS_POOL].buffer);
-   (void)bmBufferOffset(&brw->intel, brw->intel.batch->buffer);
+   /* do prepare stage for all atoms */
+   for (i = 0; i < Elements(atoms); i++) {
+      const struct brw_tracked_state *atom = brw->state.atoms[i];
+
+      if (brw->intel.Fallback)
+         break;
+
+      if (check_state(state, &atom->dirty)) {
+         if (atom->prepare) {
+            atom->prepare(brw);
+        }
+      }
+   }
+}
+
+
+void brw_upload_state(struct brw_context *brw)
+{
+   struct brw_state_flags *state = &brw->state.dirty;
+   int i;
+   static int dirty_count = 0;
+
+   brw_clear_validated_bos(brw);
 
    if (INTEL_DEBUG) {
       /* Debug version which enforces various sanity checks on the
@@ -234,12 +378,14 @@ void brw_validate_state( struct brw_context *brw )
 	 assert(atom->dirty.mesa ||
 		atom->dirty.brw ||
 		atom->dirty.cache);
-	 assert(atom->update);
+
+	 if (brw->intel.Fallback)
+	    break;
 
 	 if (check_state(state, &atom->dirty)) {
-	    brw->state.atoms[i]->update( brw );
-	    
-/* 	    emit_foo(brw); */
+	    if (atom->emit) {
+	       atom->emit( brw );
+	    }
 	 }
 
 	 accumulate_state(&examined, &atom->dirty);
@@ -255,10 +401,31 @@ void brw_validate_state( struct brw_context *brw )
    }
    else {
       for (i = 0; i < Elements(atoms); i++) {	 
-	 if (check_state(state, &brw->state.atoms[i]->dirty))
-	    brw->state.atoms[i]->update( brw );
+	 const struct brw_tracked_state *atom = brw->state.atoms[i];
+
+	 if (brw->intel.Fallback)
+	    break;
+
+	 if (check_state(state, &atom->dirty)) {
+	    if (atom->emit) {
+	       atom->emit( brw );
+	    }
+	 }
+      }
+   }
+
+   if (INTEL_DEBUG & DEBUG_STATE) {
+      brw_update_dirty_count(mesa_bits, state->mesa);
+      brw_update_dirty_count(brw_bits, state->brw);
+      brw_update_dirty_count(cache_bits, state->cache);
+      if (dirty_count++ % 1000 == 0) {
+	 brw_print_dirty_count(mesa_bits, state->mesa);
+	 brw_print_dirty_count(brw_bits, state->brw);
+	 brw_print_dirty_count(cache_bits, state->cache);
+	 fprintf(stderr, "\n");
       }
    }
 
-   memset(state, 0, sizeof(*state));
+   if (!brw->intel.Fallback)
+      memset(state, 0, sizeof(*state));
 }
diff --git a/src/mesa/drivers/dri/i965/brw_structs.h b/src/mesa/drivers/dri/i965/brw_structs.h
index 10fee944e8d..4e577d0f6a8 100644
--- a/src/mesa/drivers/dri/i965/brw_structs.h
+++ b/src/mesa/drivers/dri/i965/brw_structs.h
@@ -141,7 +141,8 @@ struct brw_depthbuffer
       struct {
 	 GLuint pitch:18; 
 	 GLuint format:3; 
-	 GLuint pad:4;
+	 GLuint pad:2;
+	 GLuint software_tiled_rendering_mode:2;
 	 GLuint depth_offset_disable:1; 
 	 GLuint tile_walk:1; 
 	 GLuint tiled_surface:1; 
@@ -166,14 +167,64 @@ struct brw_depthbuffer
 
    union {
       struct {
-	 GLuint pad:12;
-	 GLuint min_array_element:9; 
+	 GLuint pad:10;
+	 GLuint min_array_element:11; 
 	 GLuint depth:11; 
       } bits;
       GLuint dword;
    } dword4;
 };
 
+struct brw_depthbuffer_g4x
+{
+   union header_union header;
+   
+   union {
+      struct {
+	 GLuint pitch:18; 
+	 GLuint format:3; 
+	 GLuint pad:2;
+	 GLuint software_tiled_rendering_mode:2;
+	 GLuint depth_offset_disable:1; 
+	 GLuint tile_walk:1; 
+	 GLuint tiled_surface:1; 
+	 GLuint pad2:1;
+	 GLuint surface_type:3; 
+      } bits;
+      GLuint dword;
+   } dword1;
+   
+   GLuint dword2_base_addr; 
+ 
+   union {
+      struct {
+	 GLuint pad:1;
+	 GLuint mipmap_layout:1; 
+	 GLuint lod:4; 
+	 GLuint width:13; 
+	 GLuint height:13; 
+      } bits;
+      GLuint dword;
+   } dword3;
+
+   union {
+      struct {
+	 GLuint pad:10;
+	 GLuint min_array_element:11; 
+	 GLuint depth:11; 
+      } bits;
+      GLuint dword;
+   } dword4;
+
+   union {
+      struct {
+         GLuint xoffset:16;
+         GLuint yoffset:16;
+      } bits;
+      GLuint dword;
+   } dword5;   /* NEW in Integrated Graphics Device */
+};
+
 struct brw_drawrect
 {
    struct header header;
@@ -213,6 +264,25 @@ struct brw_indexbuffer
    GLuint buffer_end; 
 };
 
+/* NEW in Integrated Graphics Device */
+struct brw_aa_line_parameters
+{
+   struct header header;
+
+   struct {
+      GLuint aa_coverage_scope:8;
+      GLuint pad0:8;
+      GLuint aa_coverage_bias:8;
+      GLuint pad1:8;
+   } bits0;
+
+   struct {
+      GLuint aa_coverage_endcap_slope:8;
+      GLuint pad0:8;
+      GLuint aa_coverage_endcap_bias:8;
+      GLuint pad1:8;
+   } bits1;
+};
 
 struct brw_line_stipple
 {   
@@ -239,39 +309,39 @@ struct brw_pipelined_state_pointers
    
    struct {
       GLuint pad:5;
-      GLuint offset:27; 
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE */
    } vs;
    
    struct
    {
       GLuint enable:1;
       GLuint pad:4;
-      GLuint offset:27; 
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE */
    } gs;
    
    struct
    {
       GLuint enable:1;
       GLuint pad:4;
-      GLuint offset:27; 
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE */
    } clp;
    
    struct
    {
       GLuint pad:5;
-      GLuint offset:27; 
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE */
    } sf;
 
    struct
    {
       GLuint pad:5;
-      GLuint offset:27; 
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE */
    } wm;
    
    struct
    {
       GLuint pad:5;
-      GLuint offset:27; /* KW: check me! */
+      GLuint offset:27; /* Offset from GENERAL_STATE_BASE. KW: check me! */
    } cc;
 };
 
@@ -315,7 +385,8 @@ struct brw_pipe_control
    {
       GLuint length:8;
       GLuint notify_enable:1;
-      GLuint pad:2;
+      GLuint texture_cache_flush_enable:1;
+      GLuint indirect_state_pointers_disable:1;
       GLuint instruction_state_cache_flush_enable:1;
       GLuint write_cache_flush_enable:1;
       GLuint depth_stall_enable:1;
@@ -473,7 +544,7 @@ struct thread0
    GLuint pad0:1;
    GLuint grf_reg_count:3; 
    GLuint pad1:2;
-   GLuint kernel_start_pointer:26; 
+   GLuint kernel_start_pointer:26; /* Offset from GENERAL_STATE_BASE */
 };
 
 struct thread1
@@ -547,8 +618,8 @@ struct brw_clip_unit_state
       GLuint pad1:1;
       GLuint urb_entry_allocation_size:5; 
       GLuint pad2:1;
-      GLuint max_threads:1; 	/* may be less */
-      GLuint pad3:6;
+      GLuint max_threads:5; 	/* may be less */
+      GLuint pad3:2;
    } thread4;   
       
    struct
@@ -557,7 +628,7 @@ struct brw_clip_unit_state
       GLuint clip_mode:3; 
       GLuint userclip_enable_flags:8; 
       GLuint userclip_must_clip:1; 
-      GLuint pad1:1;
+      GLuint negative_w_clip_test:1;
       GLuint guard_band_enable:1; 
       GLuint viewport_z_clip_enable:1; 
       GLuint viewport_xy_clip_enable:1; 
@@ -637,7 +708,7 @@ struct brw_cc_unit_state
    struct
    {
       GLuint pad0:5; 
-      GLuint cc_viewport_state_offset:27; 
+      GLuint cc_viewport_state_offset:27; /* Offset from GENERAL_STATE_BASE */
    } cc4;
    
    struct
@@ -699,7 +770,7 @@ struct brw_sf_unit_state
       GLuint front_winding:1; 
       GLuint viewport_transform:1; 
       GLuint pad0:3;
-      GLuint sf_viewport_state_offset:27; 
+      GLuint sf_viewport_state_offset:27; /* Offset from GENERAL_STATE_BASE */
    } sf5;
    
    struct
@@ -724,7 +795,8 @@ struct brw_sf_unit_state
       GLuint use_point_size_state:1; 
       GLuint subpixel_precision:1; 
       GLuint sprite_point:1; 
-      GLuint pad0:11;
+      GLuint pad0:10;
+      GLuint aa_line_distance_mode:1;
       GLuint trifan_pv:2; 
       GLuint linestrip_pv:2; 
       GLuint tristrip_pv:2; 
@@ -749,8 +821,8 @@ struct brw_gs_unit_state
       GLuint pad1:1;
       GLuint urb_entry_allocation_size:5; 
       GLuint pad2:1;
-      GLuint max_threads:1; 
-      GLuint pad3:6;
+      GLuint max_threads:5; 
+      GLuint pad3:2;
    } thread4;   
       
    struct
@@ -764,9 +836,14 @@ struct brw_gs_unit_state
    struct
    {
       GLuint max_vp_index:4; 
-      GLuint pad0:26;
-      GLuint reorder_enable:1; 
+      GLuint pad0:12;
+      GLuint svbi_post_inc_value:10;
       GLuint pad1:1;
+      GLuint svbi_post_inc_enable:1;
+      GLuint svbi_payload:1;
+      GLuint discard_adjaceny:1;
+      GLuint reorder_enable:1; 
+      GLuint pad2:1;
    } gs6;
 };
 
@@ -786,8 +863,8 @@ struct brw_vs_unit_state
       GLuint pad1:1;
       GLuint urb_entry_allocation_size:5; 
       GLuint pad2:1;
-      GLuint max_threads:4; 
-      GLuint pad3:3;
+      GLuint max_threads:6; 
+      GLuint pad3:1;
    } thread4;   
 
    struct
@@ -815,7 +892,7 @@ struct brw_wm_unit_state
    
    struct {
       GLuint stats_enable:1; 
-      GLuint pad0:1;
+      GLuint depth_buffer_clear:1;
       GLuint sampler_count:3; 
       GLuint sampler_state_pointer:27; 
    } wm4;
@@ -825,7 +902,9 @@ struct brw_wm_unit_state
       GLuint enable_8_pix:1; 
       GLuint enable_16_pix:1; 
       GLuint enable_32_pix:1; 
-      GLuint pad0:7;
+      GLuint enable_con_32_pix:1;
+      GLuint enable_con_64_pix:1;
+      GLuint pad0:5;
       GLuint legacy_global_depth_bias:1; 
       GLuint line_stipple:1; 
       GLuint depth_offset:1; 
@@ -838,9 +917,8 @@ struct brw_wm_unit_state
       GLuint program_computes_depth:1; 
       GLuint program_uses_killpixel:1; 
       GLuint legacy_line_rast: 1; 
-      GLuint pad1:1; 
-      GLuint max_threads:6; 
-      GLuint pad2:1;
+      GLuint transposed_urb_read_enable:1; 
+      GLuint max_threads:7; 
    } wm5;
    
    GLfloat global_depth_offset_constant;  
@@ -924,6 +1002,7 @@ struct brw_sf_viewport
       GLfloat m32;  
    } viewport;
 
+   /* scissor coordinates are inclusive */
    struct {
       GLshort xmin;
       GLshort ymin;
@@ -978,10 +1057,26 @@ struct brw_surface_state
    } ss3;
    
    struct {
-      GLuint pad:19;
-      GLuint min_array_elt:9; 
+      GLuint multisample_position_palette_index:3;
+      GLuint pad1:1;
+      GLuint num_multisamples:3;
+      GLuint pad0:1;
+      GLuint render_target_view_extent:9;
+      GLuint min_array_elt:11;
       GLuint min_lod:4; 
    } ss4;
+
+   struct {
+      GLuint pad1:16;
+      GLuint llc_mapping:1;
+      GLuint mlc_mapping:1;
+      GLuint gfdt:1;
+      GLuint gfdt_src:1;
+      GLuint y_offset:4;
+      GLuint pad0:1;
+      GLuint x_offset:7;
+   } ss5;   /* NEW in Integrated Graphics Device */
+
 };
 
 
@@ -1301,6 +1396,17 @@ struct brw_instruction
 	 GLuint end_of_thread:1;
       } sampler;
 
+      struct {
+         GLuint binding_table_index:8;
+         GLuint sampler:4;
+         GLuint msg_type:4;
+         GLuint response_length:4;
+         GLuint msg_length:4;
+         GLuint msg_target:4;
+         GLuint pad1:3;
+         GLuint end_of_thread:1;
+      } sampler_g4x;
+
       struct brw_urb_immediate urb;
 
       struct {
diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c
index 9d4b9867d24..0bb6f176a0d 100644
--- a/src/mesa/drivers/dri/i965/brw_tex.c
+++ b/src/mesa/drivers/dri/i965/brw_tex.c
@@ -30,165 +30,30 @@
   */
         
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "teximage.h"
-#include "texstore.h"
-#include "texformat.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
+
 #include "texmem.h"
 
 #include "intel_context.h"
-#include "intel_ioctl.h"
 #include "intel_regions.h"
+#include "intel_tex.h"
 #include "brw_context.h"
 #include "brw_defines.h"
 
 
-
-
-static const struct gl_texture_format *
-brwChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
-			 GLenum srcFormat, GLenum srcType )
-{
-   switch ( internalFormat ) {
-   case 4:
-   case GL_RGBA:
-   case GL_COMPRESSED_RGBA:
-      if (srcFormat == GL_BGRA && srcType == GL_UNSIGNED_SHORT_4_4_4_4_REV)
-	 return &_mesa_texformat_argb4444;
-      else if (srcFormat == GL_BGRA && srcType == GL_UNSIGNED_SHORT_1_5_5_5_REV)
-	 return &_mesa_texformat_argb1555;
-      else if ((srcFormat == GL_RGBA && srcType == GL_UNSIGNED_INT_8_8_8_8_REV) ||
-	       (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE) ||
-	       (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_INT_8_8_8_8)) 
-	 return &_mesa_texformat_rgba8888_rev;
-      else
-	 return &_mesa_texformat_argb8888;
-
-   case GL_RGBA8:
-   case GL_RGB10_A2:
-   case GL_RGBA12:
-   case GL_RGBA16:
-      return &_mesa_texformat_argb8888; 
-
-   case GL_RGB8:
-   case GL_RGB10:
-   case GL_RGB12:
-   case GL_RGB16:
-      /* Broadwater doesn't support RGB888 textures, so these must be
-       * stored as ARGB.
-       */
-      return &_mesa_texformat_argb8888;
-
-   case 3:
-   case GL_COMPRESSED_RGB:
-   case GL_RGB:
-      if (srcFormat == GL_RGB &&
-	  srcType == GL_UNSIGNED_SHORT_5_6_5)
-	 return &_mesa_texformat_rgb565;
-      else
-	 return &_mesa_texformat_argb8888;
-
-
-   case GL_RGB5:
-   case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
-
-   case GL_R3_G3_B2:
-   case GL_RGBA2:
-   case GL_RGBA4:
-   case GL_RGB4:
-      return &_mesa_texformat_argb4444;
-
-   case GL_ALPHA:
-   case GL_ALPHA4:
-   case GL_ALPHA8:
-   case GL_ALPHA12:
-   case GL_ALPHA16:
-   case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
-
-   case 1:
-   case GL_LUMINANCE:
-   case GL_LUMINANCE4:
-   case GL_LUMINANCE8:
-   case GL_LUMINANCE12:
-   case GL_LUMINANCE16:
-   case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
-
-   case 2:
-   case GL_LUMINANCE_ALPHA:
-   case GL_LUMINANCE4_ALPHA4:
-   case GL_LUMINANCE6_ALPHA2:
-   case GL_LUMINANCE8_ALPHA8:
-   case GL_LUMINANCE12_ALPHA4:
-   case GL_LUMINANCE12_ALPHA12:
-   case GL_LUMINANCE16_ALPHA16:
-   case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
-
-   case GL_INTENSITY:
-   case GL_INTENSITY4:
-   case GL_INTENSITY8:
-   case GL_INTENSITY12:
-   case GL_INTENSITY16:
-   case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
-
-   case GL_YCBCR_MESA:
-      if (srcType == GL_UNSIGNED_SHORT_8_8_MESA ||
-	  srcType == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
-      else
-         return &_mesa_texformat_ycbcr_rev;
-
-   case GL_COMPRESSED_RGB_FXT1_3DFX:
-       return &_mesa_texformat_rgb_fxt1;
-   case GL_COMPRESSED_RGBA_FXT1_3DFX:
-       return &_mesa_texformat_rgba_fxt1;
-
-   case GL_RGB_S3TC:
-   case GL_RGB4_S3TC:
-   case GL_RGBA_S3TC:
-   case GL_RGBA4_S3TC:
-   case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-   case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgb_dxt1; /* there is no rgba support? */
-
-   case GL_DEPTH_COMPONENT:
-   case GL_DEPTH_COMPONENT16:
-   case GL_DEPTH_COMPONENT24:
-   case GL_DEPTH_COMPONENT32:
-      return &_mesa_texformat_z16;
-
-   default:
-      fprintf(stderr, "unexpected texture format %s in %s\n", 
-	      _mesa_lookup_enum_by_nr(internalFormat),
-	      __FUNCTION__);
-      return NULL;
-   }
-
-   return NULL; /* never get here */
-}
-
-
-void brwInitTextureFuncs( struct dd_function_table *functions )
-{
-   functions->ChooseTextureFormat = brwChooseTextureFormat;
-}
-
-void brw_FrameBufferTexInit( struct brw_context *brw )
+void brw_FrameBufferTexInit( struct brw_context *brw,
+			     struct intel_region *region )
 {
    struct intel_context *intel = &brw->intel;
    GLcontext *ctx = &intel->ctx;
-   struct intel_region *region = intel->front_region;
    struct gl_texture_object *obj;
    struct gl_texture_image *img;
    
@@ -209,6 +74,26 @@ void brw_FrameBufferTexInit( struct brw_context *brw )
 
 void brw_FrameBufferTexDestroy( struct brw_context *brw )
 {
-   brw->intel.ctx.Driver.DeleteTexture( &brw->intel.ctx,
-					brw->intel.frame_buffer_texobj );
+   if (brw->intel.frame_buffer_texobj != NULL)
+      brw->intel.ctx.Driver.DeleteTexture( &brw->intel.ctx,
+					   brw->intel.frame_buffer_texobj );
+   brw->intel.frame_buffer_texobj = NULL;
+}
+
+/**
+ * Finalizes all textures, completing any rendering that needs to be done
+ * to prepare them.
+ */
+void brw_validate_textures( struct brw_context *brw )
+{
+   struct intel_context *intel = &brw->intel;
+   int i;
+
+   for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
+      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
+
+      if (texUnit->_ReallyEnabled) {
+	 intel_finalize_mipmap_tree(intel, i);
+      }
+   }
 }
diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
index af1ad0f1ef1..51a617fcb40 100644
--- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
+++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
@@ -35,10 +35,12 @@
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex_layout.h"
-#include "macros.h"
+#include "intel_context.h"
+#include "main/macros.h"
 
+#define FILE_DEBUG_FLAG DEBUG_MIPTREE
 
-GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
+GLboolean brw_miptree_layout( struct intel_context *intel, struct intel_mipmap_tree *mt )
 {
    /* XXX: these vary depending on image format: 
     */
@@ -53,11 +55,20 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
       GLuint pack_x_pitch, pack_x_nr;
       GLuint pack_y_pitch;
       GLuint level;
+      GLuint align_h = 2;
+      GLuint align_w = 4;
 
-      mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
       mt->total_height = 0;
+      
+      if (mt->compressed) {
+          align_w = intel_compressed_alignment(mt->internal_format);
+          mt->pitch = ALIGN(width, align_w);
+          pack_y_pitch = (height + 3) / 4;
+      } else {
+          mt->pitch = intel_miptree_pitch_align (intel, mt, mt->width0);
+          pack_y_pitch = ALIGN(mt->height0, align_h);
+      }
 
-      pack_y_pitch = MAX2(mt->height0, 2);
       pack_x_pitch = mt->pitch;
       pack_x_nr = 1;
 
@@ -83,26 +94,36 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 
 
 	 mt->total_height += y;
-
-	 if (pack_x_pitch > 4) {
-	    pack_x_pitch >>= 1;
-	    pack_x_nr <<= 1;
-	    assert(pack_x_pitch * pack_x_nr <= mt->pitch);
-	 }
-
-	 if (pack_y_pitch > 2) {
-	    pack_y_pitch >>= 1;
-	 }
-
 	 width  = minify(width);
 	 height = minify(height);
 	 depth  = minify(depth);
+
+    if (mt->compressed) {
+        pack_y_pitch = (height + 3) / 4;
+        
+        if (pack_x_pitch > ALIGN(width, align_w)) {
+            pack_x_pitch = ALIGN(width, align_w);
+            pack_x_nr <<= 1;
+        }
+    } else {
+        if (pack_x_pitch > 4) {
+            pack_x_pitch >>= 1;
+            pack_x_nr <<= 1;
+            assert(pack_x_pitch * pack_x_nr <= mt->pitch);
+        }
+
+        if (pack_y_pitch > 2) {
+            pack_y_pitch >>= 1;
+            pack_y_pitch = ALIGN(pack_y_pitch, align_h);
+        }
+    }
+
       }
       break;
    }
 
    default:
-      i945_miptree_layout_2d(mt);
+      i945_miptree_layout_2d(intel, mt);
       break;
    }
    DBG("%s: %dx%dx%d - sz 0x%x\n", __FUNCTION__, 
diff --git a/src/mesa/drivers/dri/i965/brw_urb.c b/src/mesa/drivers/dri/i965/brw_urb.c
index 64f5904ac68..7673dd36eb9 100644
--- a/src/mesa/drivers/dri/i965/brw_urb.c
+++ b/src/mesa/drivers/dri/i965/brw_urb.c
@@ -35,7 +35,6 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "brw_hal.h"
 
 #define VS 0
 #define GS 1
@@ -43,7 +42,44 @@
 #define SF 3
 #define CS 4
 
-/* XXX: Are the min_entry_size numbers useful?
+/** @file brw_urb.c
+ *
+ * Manages the division of the URB space between the various fixed-function
+ * units.
+ *
+ * See the Thread Initiation Management section of the GEN4 B-Spec, and
+ * the individual *_STATE structures for restrictions on numbers of
+ * entries and threads.
+ */
+
+/*
+ * Generally, a unit requires a min_nr_entries based on how many entries
+ * it produces before the downstream unit gets unblocked and can use and
+ * dereference some of its handles.
+ *
+ * The SF unit preallocates a PUE at the start of thread dispatch, and only
+ * uses that one.  So it requires one entry per thread.
+ *
+ * For CLIP, the SF unit will hold the previous primitive while the
+ * next is getting assembled, meaning that linestrips require 3 CLIP VUEs
+ * (vertices) to ensure continued processing, trifans require 4, and tristrips
+ * require 5.  There can be 1 or 2 threads, and each has the same requirement.
+ *
+ * GS has the same requirement as CLIP, but it never handles tristrips,
+ * so we can lower the minimum to 4 for the POLYGONs (trifans) it produces.
+ * We only run it single-threaded.
+ *
+ * For VS, the number of entries may be 8, 12, 16, or 32 (or 64 on G4X).
+ * Each thread processes 2 preallocated VUEs (vertices) at a time, and they
+ * get streamed down as soon as threads processing earlier vertices get
+ * theirs accepted.
+ *
+ * Each unit will take the number of URB entries we give it (based on the
+ * entry size calculated in brw_vs_emit.c for VUEs, brw_sf_emit.c for PUEs,
+ * and brw_curbe.c for the CURBEs) and decide its maximum number of
+ * threads it can support based on that. in brw_*_state.c.
+ *
+ * XXX: Are the min_entry_size numbers useful?
  * XXX: Verify min_nr_entries, esp for VS.
  * XXX: Verify SF min_entry_size.
  */
@@ -53,9 +89,9 @@ static const struct {
    GLuint min_entry_size;
    GLuint max_entry_size;
 } limits[CS+1] = {
-   { 8, 32, 1, 5 },			/* vs */
+   { 16, 32, 1, 5 },			/* vs */
    { 4, 8,  1, 5 },			/* gs */
-   { 6, 8,  1, 5 },			/* clp */
+   { 5, 10,  1, 5 },			/* clp */
    { 1, 8,  1, 12 },		        /* sf */
    { 1, 4,  1, 32 }			/* cs */
 };
@@ -69,7 +105,7 @@ static GLboolean check_urb_layout( struct brw_context *brw )
    brw->urb.sf_start = brw->urb.clip_start + brw->urb.nr_clip_entries * brw->urb.vsize;
    brw->urb.cs_start = brw->urb.sf_start + brw->urb.nr_sf_entries * brw->urb.sfsize;
 
-   return brw->urb.cs_start + brw->urb.nr_cs_entries * brw->urb.csize <= 256;
+   return brw->urb.cs_start + brw->urb.nr_cs_entries * brw->urb.csize <= URB_SIZES(brw);
 }
 
 /* Most minimal update, forces re-emit of URB fence packet after GS
@@ -81,20 +117,6 @@ static void recalculate_urb_fence( struct brw_context *brw )
    GLuint vsize = brw->vs.prog_data->urb_entry_size;
    GLuint sfsize = brw->sf.prog_data->urb_entry_size;
 
-   static GLboolean (*hal_recalculate_urb_fence) (struct brw_context *brw);
-   static GLboolean hal_tried;
-
-   if (!hal_tried)
-   {
-      hal_recalculate_urb_fence = brw_hal_find_symbol ("intel_hal_recalculate_urb_fence");
-      hal_tried = 1;
-   }
-   if (hal_recalculate_urb_fence)
-   {
-      if ((*hal_recalculate_urb_fence) (brw))
-	 return;
-   }
-   
    if (csize < limits[CS].min_entry_size)
       csize = limits[CS].min_entry_size;
 
@@ -107,9 +129,9 @@ static void recalculate_urb_fence( struct brw_context *brw )
    if (brw->urb.vsize < vsize ||
        brw->urb.sfsize < sfsize ||
        brw->urb.csize < csize ||
-       (brw->urb.constrained && (brw->urb.vsize > brw->urb.vsize ||
-				 brw->urb.sfsize > brw->urb.sfsize ||
-				 brw->urb.csize > brw->urb.csize))) {
+       (brw->urb.constrained && (brw->urb.vsize > vsize ||
+				 brw->urb.sfsize > sfsize ||
+				 brw->urb.csize > csize))) {
       
 
       brw->urb.csize = csize;
@@ -129,6 +151,10 @@ static void recalculate_urb_fence( struct brw_context *brw )
 	 brw->urb.nr_sf_entries = limits[SF].min_nr_entries;	
 	 brw->urb.nr_cs_entries = limits[CS].min_nr_entries;	
 
+	 /* Mark us as operating with constrained nr_entries, so that next
+	  * time we recalculate we'll resize the fences in the hope of
+	  * escaping constrained mode and getting back to normal performance.
+	  */
 	 brw->urb.constrained = 1;
 	 
 	 if (!check_urb_layout(brw)) {
@@ -153,7 +179,7 @@ static void recalculate_urb_fence( struct brw_context *brw )
 		      brw->urb.clip_start,
 		      brw->urb.sf_start,
 		      brw->urb.cs_start, 
-		      256);
+		      URB_SIZES(brw));
       
       brw->state.dirty.brw |= BRW_NEW_URB_FENCE;
    }
@@ -167,7 +193,7 @@ const struct brw_tracked_state brw_recalculate_urb_fence = {
       .cache = (CACHE_NEW_VS_PROG |
 		CACHE_NEW_SF_PROG)
    },
-   .update = recalculate_urb_fence
+   .prepare = recalculate_urb_fence
 };
 
 
@@ -191,25 +217,13 @@ void brw_upload_urb_fence(struct brw_context *brw)
    /* The ordering below is correct, not the layout in the
     * instruction.
     *
-    * There are 256 urb reg pairs in total.
+    * There are 256/384 urb reg pairs in total.
     */
    uf.bits0.vs_fence  = brw->urb.gs_start;
    uf.bits0.gs_fence  = brw->urb.clip_start; 
    uf.bits0.clp_fence = brw->urb.sf_start; 
    uf.bits1.sf_fence  = brw->urb.cs_start; 
-   uf.bits1.cs_fence  = 256;
+   uf.bits1.cs_fence  = URB_SIZES(brw);
 
    BRW_BATCH_STRUCT(brw, &uf);
 }
-
-
-#if 0
-const struct brw_tracked_state brw_urb_fence = {
-   .dirty = {
-      .mesa = 0,
-      .brw = BRW_NEW_URB_FENCE | BRW_NEW_PSP,
-      .cache = 0
-   },
-   .update = brw_upload_urb_fence
-};
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_util.c b/src/mesa/drivers/dri/i965/brw_util.c
index b6deee23769..ce21aa48695 100644
--- a/src/mesa/drivers/dri/i965/brw_util.c
+++ b/src/mesa/drivers/dri/i965/brw_util.c
@@ -30,7 +30,7 @@
   */
          
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "shader/prog_parameter.h"
 #include "brw_util.h"
 #include "brw_defines.h"
@@ -45,86 +45,6 @@ GLuint brw_count_bits( GLuint val )
 }
 
 
-static GLuint brw_parameter_state_flags(const gl_state_index state[])
-{
-   switch (state[0]) {
-   case STATE_MATERIAL:
-   case STATE_LIGHT:
-   case STATE_LIGHTMODEL_AMBIENT:
-   case STATE_LIGHTMODEL_SCENECOLOR:
-   case STATE_LIGHTPROD:
-      return _NEW_LIGHT;
-
-   case STATE_TEXGEN:
-   case STATE_TEXENV_COLOR:
-      return _NEW_TEXTURE;
-
-   case STATE_FOG_COLOR:
-   case STATE_FOG_PARAMS:
-      return _NEW_FOG;
-
-   case STATE_CLIPPLANE:
-      return _NEW_TRANSFORM;
-
-   case STATE_POINT_SIZE:
-   case STATE_POINT_ATTENUATION:
-      return _NEW_POINT;
-
-   case STATE_MODELVIEW_MATRIX:
-      return _NEW_MODELVIEW;
-   case STATE_PROJECTION_MATRIX:
-      return _NEW_PROJECTION;
-   case STATE_MVP_MATRIX:
-      return _NEW_MODELVIEW | _NEW_PROJECTION;
-   case STATE_TEXTURE_MATRIX:
-      return _NEW_TEXTURE_MATRIX;
-   case STATE_PROGRAM_MATRIX:
-      return _NEW_TRACK_MATRIX;
-
-   case STATE_DEPTH_RANGE:
-      return _NEW_VIEWPORT;
-
-   case STATE_FRAGMENT_PROGRAM:
-   case STATE_VERTEX_PROGRAM:
-      return _NEW_PROGRAM;
-
-   case STATE_INTERNAL:
-      switch (state[1]) {
-      case STATE_NORMAL_SCALE:
-	 return _NEW_MODELVIEW;
-      case STATE_TEXRECT_SCALE:
-	 return _NEW_TEXTURE;
-      default:
-	 assert(0);
-	 return 0;
-      }
-
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-
-GLuint
-brw_parameter_list_state_flags(struct gl_program_parameter_list *paramList)
-{
-   GLuint i;
-   GLuint result = 0;
-
-   if (!paramList)
-      return 0;
-
-   for (i = 0; i < paramList->NumParameters; i++) {
-      if (paramList->Parameters[i].Type == PROGRAM_STATE_VAR) {
-         result |= brw_parameter_state_flags(paramList->Parameters[i].StateIndexes);
-      }
-   }
-
-   return result;
-}
-
-
 GLuint brw_translate_blend_equation( GLenum mode )
 {
    switch (mode) {
diff --git a/src/mesa/drivers/dri/i965/brw_util.h b/src/mesa/drivers/dri/i965/brw_util.h
index bd6cc0a2682..33e7cd87e42 100644
--- a/src/mesa/drivers/dri/i965/brw_util.h
+++ b/src/mesa/drivers/dri/i965/brw_util.h
@@ -33,7 +33,7 @@
 #ifndef BRW_UTIL_H
 #define BRW_UTIL_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern GLuint brw_count_bits( GLuint val );
 extern GLuint brw_parameter_list_state_flags(struct gl_program_parameter_list *paramList);
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index e173f6fce3e..1db7ceebcfb 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -49,7 +49,7 @@ static void do_vs_prog( struct brw_context *brw,
    memset(&c, 0, sizeof(c));
    memcpy(&c.key, key, sizeof(*key));
 
-   brw_init_compile(&c.func);
+   brw_init_compile(brw, &c.func);
    c.vp = vp;
 
    c.prog_data.outputs_written = vp->program.Base.OutputsWritten;
@@ -73,19 +73,17 @@ static void do_vs_prog( struct brw_context *brw,
     */
    program = brw_get_program(&c.func, &program_size);
 
-   /*
-    */
-   brw->vs.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_VS_PROG],
-					      &c.key,
-					      sizeof(c.key),
-					      program,
-					      program_size,
-					      &c.prog_data,
-					      &brw->vs.prog_data);
+   dri_bo_unreference(brw->vs.prog_bo);
+   brw->vs.prog_bo = brw_upload_cache( &brw->cache, BRW_VS_PROG,
+				       &c.key, sizeof(c.key),
+				       NULL, 0,
+				       program, program_size,
+				       &c.prog_data,
+				       &brw->vs.prog_data );
 }
 
 
-static void brw_upload_vs_prog( struct brw_context *brw )
+static void brw_upload_vs_prog(struct brw_context *brw)
 {
    struct brw_vs_prog_key key;
    struct brw_vertex_program *vp = 
@@ -110,13 +108,13 @@ static void brw_upload_vs_prog( struct brw_context *brw )
 
    /* Make an early check for the key.
     */
-   if (brw_search_cache(&brw->cache[BRW_VS_PROG], 
-			&key, sizeof(key),
-			&brw->vs.prog_data,
-			&brw->vs.prog_gs_offset))
-       return;
-
-   do_vs_prog(brw, vp, &key);
+   dri_bo_unreference(brw->vs.prog_bo);
+   brw->vs.prog_bo = brw_search_cache(&brw->cache, BRW_VS_PROG,
+				      &key, sizeof(key),
+				      NULL, 0,
+				      &brw->vs.prog_data);
+   if (brw->vs.prog_bo == NULL)
+      do_vs_prog(brw, vp, &key);
 }
 
 
@@ -128,5 +126,5 @@ const struct brw_tracked_state brw_vs_prog = {
       .brw   = BRW_NEW_VERTEX_PROGRAM | BRW_NEW_METAOPS,
       .cache = 0
    },
-   .update = brw_upload_vs_prog
+   .prepare = brw_upload_vs_prog
 };
diff --git a/src/mesa/drivers/dri/i965/brw_vs.h b/src/mesa/drivers/dri/i965/brw_vs.h
index fdb5785d67d..22388ec99d0 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.h
+++ b/src/mesa/drivers/dri/i965/brw_vs.h
@@ -36,7 +36,7 @@
 
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "program.h"
+#include "shader/program.h"
 
 
 struct brw_vs_prog_key {
@@ -67,6 +67,12 @@ struct brw_vs_compile {
    struct brw_reg r1;
    struct brw_reg regs[PROGRAM_ADDRESS+1][128];
    struct brw_reg tmp;
+   struct brw_reg stack;
+
+   struct {	
+       GLboolean used_in_src;
+       struct brw_reg reg;
+   } output_regs[128];
 
    struct brw_reg userplane[6];
 
@@ -74,8 +80,4 @@ struct brw_vs_compile {
 
 void brw_vs_emit( struct brw_vs_compile *c );
 
-
-void brw_ProgramCacheDestroy( GLcontext *ctx );
-void brw_ProgramCacheInit( GLcontext *ctx );
-
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_vs_constval.c b/src/mesa/drivers/dri/i965/brw_vs_constval.c
index caef042f1cd..6fbac02de63 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_constval.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_constval.c
@@ -30,7 +30,7 @@
   */
             
 
-#include "macros.h"
+#include "main/macros.h"
 #include "brw_context.h"
 #include "brw_vs.h"
 
@@ -218,6 +218,6 @@ const struct brw_tracked_state brw_wm_input_sizes = {
       .brw   = BRW_NEW_VERTEX_PROGRAM | BRW_NEW_INPUT_DIMENSIONS,
       .cache = 0
    },
-   .update = calc_wm_input_sizes
+   .prepare = calc_wm_input_sizes
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 6eb11b19ad2..25b4ee85cb0 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -30,8 +30,8 @@
   */
             
 
-#include "program.h"
-#include "macros.h"
+#include "main/macros.h"
+#include "shader/program.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_print.h"
 #include "brw_context.h"
@@ -134,6 +134,16 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 					     WRITEMASK_X);
       reg++;
    }
+
+   for (i = 0; i < 128; i++) {
+       if (c->output_regs[i].used_in_src) {
+            c->output_regs[i].reg = brw_vec8_grf(reg, 0);
+            reg++;
+        }
+   }
+
+   c->stack =  brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, reg, 0);
+   reg += 2;
  
    
    /* Some opcodes need an internal temporary:
@@ -201,7 +211,7 @@ static void unalias2( struct brw_vs_compile *c,
 				    struct brw_reg,
 				    struct brw_reg ))
 {
-   if ((dst.file == arg0.file && dst.nr == arg0.nr) &&
+   if ((dst.file == arg0.file && dst.nr == arg0.nr) ||
        (dst.file == arg1.file && dst.nr == arg1.nr)) {
       struct brw_compile *p = &c->func;
       struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
@@ -213,57 +223,65 @@ static void unalias2( struct brw_vs_compile *c,
    }
 }
 
+static void emit_sop( struct brw_compile *p,
+                      struct brw_reg dst,
+                      struct brw_reg arg0,
+                      struct brw_reg arg1, 
+		      GLuint cond)
+{
+   brw_MOV(p, dst, brw_imm_f(0.0f));
+   brw_CMP(p, brw_null_reg(), cond, arg0, arg1);
+   brw_MOV(p, dst, brw_imm_f(1.0f));
+   brw_set_predicate_control_flag_value(p, 0xff);
+}
 
+static void emit_seq( struct brw_compile *p,
+                      struct brw_reg dst,
+                      struct brw_reg arg0,
+                      struct brw_reg arg1 )
+{
+   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_EQ);
+}
 
-
+static void emit_sne( struct brw_compile *p,
+                      struct brw_reg dst,
+                      struct brw_reg arg0,
+                      struct brw_reg arg1 )
+{
+   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_NEQ);
+}
 static void emit_slt( struct brw_compile *p, 
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-   /* Could be done with an if/else/endif, but this method uses half
-    * the instructions.  Note that we are careful to reference the
-    * arguments before writing the dest.  That means we emit the
-    * instructions in an odd order and have to play with the flag
-    * values.
-    */
-   brw_push_insn_state(p);
-   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0, arg1);
-
-   /* Write all values to 1:
-    */
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-   brw_MOV(p, dst, brw_imm_f(1.0));
+   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_L);
+}
 
-   /* Where the test succeeded, overwite with zero:
-    */
-   brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-   brw_MOV(p, dst, brw_imm_f(0.0));
-   brw_pop_insn_state(p);
+static void emit_sle( struct brw_compile *p, 
+		      struct brw_reg dst,
+		      struct brw_reg arg0,
+		      struct brw_reg arg1 )
+{
+   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_LE);
 }
 
+static void emit_sgt( struct brw_compile *p, 
+		      struct brw_reg dst,
+		      struct brw_reg arg0,
+		      struct brw_reg arg1 )
+{
+   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_G);
+}
 
 static void emit_sge( struct brw_compile *p, 
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-   brw_push_insn_state(p);
-   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0, arg1);
-
-   /* Write all values to zero:
-    */
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-   brw_MOV(p, dst, brw_imm_f(0));
-
-   /* Where the test succeeded, overwite with 1:
-    */
-   brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
-   brw_MOV(p, dst, brw_imm_f(1.0));
-   brw_pop_insn_state(p);
+  emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_GE);
 }
 
-
 static void emit_max( struct brw_compile *p, 
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
@@ -592,9 +610,13 @@ static struct brw_reg get_reg( struct brw_vs_compile *c,
    case PROGRAM_TEMPORARY:
    case PROGRAM_INPUT:
    case PROGRAM_OUTPUT:
-   case PROGRAM_STATE_VAR:
       assert(c->regs[file][index].nr != 0);
       return c->regs[file][index];
+   case PROGRAM_STATE_VAR:
+   case PROGRAM_CONSTANT:
+   case PROGRAM_UNIFORM:
+      assert(c->regs[PROGRAM_STATE_VAR][index].nr != 0);
+      return c->regs[PROGRAM_STATE_VAR][index];
    case PROGRAM_ADDRESS:
       assert(index == 0);
       return c->regs[file][index];
@@ -668,28 +690,28 @@ static void emit_arl( struct brw_vs_compile *c,
  * account.
  */
 static struct brw_reg get_arg( struct brw_vs_compile *c,
-			       struct prog_src_register src )
+			       struct prog_src_register *src )
 {
    struct brw_reg reg;
 
-   if (src.File == PROGRAM_UNDEFINED)
+   if (src->File == PROGRAM_UNDEFINED)
       return brw_null_reg();
 
-   if (src.RelAddr) 
-      reg = deref(c, c->regs[PROGRAM_STATE_VAR][0], src.Index);
+   if (src->RelAddr) 
+      reg = deref(c, c->regs[PROGRAM_STATE_VAR][0], src->Index);
    else
-      reg = get_reg(c, src.File, src.Index);
+      reg = get_reg(c, src->File, src->Index);
 
    /* Convert 3-bit swizzle to 2-bit.  
     */
-   reg.dw1.bits.swizzle = BRW_SWIZZLE4(GET_SWZ(src.Swizzle, 0),
-				       GET_SWZ(src.Swizzle, 1),
-				       GET_SWZ(src.Swizzle, 2),
-				       GET_SWZ(src.Swizzle, 3));
+   reg.dw1.bits.swizzle = BRW_SWIZZLE4(GET_SWZ(src->Swizzle, 0),
+				       GET_SWZ(src->Swizzle, 1),
+				       GET_SWZ(src->Swizzle, 2),
+				       GET_SWZ(src->Swizzle, 3));
 
    /* Note this is ok for non-swizzle instructions: 
     */
-   reg.negate = src.NegateBase ? 1 : 0;   
+   reg.negate = src->NegateBase ? 1 : 0;   
 
    return reg;
 }
@@ -796,8 +818,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
    }
 
 
-   /* Build ndc coords?   TODO: Shortcircuit when w is known to be one.
-    */
+   /* Build ndc coords */
    if (!c->key.know_w_is_one) {
       ndc = get_tmp(c);
       emit_math1(c, BRW_MATH_FUNCTION_INV, ndc, brw_swizzle1(pos, 3), BRW_MATH_PRECISION_FULL);
@@ -807,12 +828,12 @@ static void emit_vertex_write( struct brw_vs_compile *c)
       ndc = pos;
    }
 
-   /* This includes the workaround for -ve rhw, so is no longer an
-    * optional step:
+   /* Update the header for point size, user clipping flags, and -ve rhw
+    * workaround.
     */
    if ((c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) ||
        c->key.nr_userclip ||
-       !c->key.know_w_is_one)
+       (!BRW_IS_G4X(p->brw) && !c->key.know_w_is_one))
    {
       struct brw_reg header1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
       GLuint i;
@@ -845,7 +866,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
        * Later, clipping will detect ucp[6] and ensure the primitive is
        * clipped against all fixed planes.
        */
-      if (!c->key.know_w_is_one) {
+      if (!BRW_IS_G4X(p->brw) && !c->key.know_w_is_one) {
 	 brw_CMP(p,
 		 vec8(brw_null_reg()),
 		 BRW_CONDITIONAL_L,
@@ -891,17 +912,50 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 
 }
 
-
-
+static void 
+post_vs_emit( struct brw_vs_compile *c, struct brw_instruction *end_inst )
+{
+   GLuint nr_insns = c->vp->program.Base.NumInstructions;
+   GLuint insn, target_insn;
+   struct prog_instruction *inst1, *inst2;
+   struct brw_instruction *brw_inst1, *brw_inst2;
+   int offset;
+   for (insn = 0; insn < nr_insns; insn++) {
+       inst1 = &c->vp->program.Base.Instructions[insn];
+       brw_inst1 = inst1->Data;
+       switch (inst1->Opcode) {
+	   case OPCODE_CAL:
+	   case OPCODE_BRA:
+	       target_insn = inst1->BranchTarget;
+	       inst2 = &c->vp->program.Base.Instructions[target_insn];
+	       brw_inst2 = inst2->Data;
+	       offset = brw_inst2 - brw_inst1;
+	       brw_set_src1(brw_inst1, brw_imm_d(offset*16));
+	       break;
+	   case OPCODE_END:
+	       offset = end_inst - brw_inst1;
+	       brw_set_src1(brw_inst1, brw_imm_d(offset*16));
+	       break;
+	   default:
+	       break;
+       }
+   }
+}
 
 /* Emit the fragment program instructions here.
  */
-void brw_vs_emit( struct brw_vs_compile *c )
+void brw_vs_emit(struct brw_vs_compile *c )
 {
+#define MAX_IFSN 32
    struct brw_compile *p = &c->func;
    GLuint nr_insns = c->vp->program.Base.NumInstructions;
-   GLuint insn;
+   GLuint insn, if_insn = 0;
+   struct brw_instruction *end_inst;
+   struct brw_instruction *if_inst[MAX_IFSN];
+   struct brw_indirect stack_index = brw_indirect(0, 0);   
 
+   GLuint index;
+   GLuint file;
 
    if (INTEL_DEBUG & DEBUG_VS) {
       _mesa_printf("\n\n\nvs-emit:\n");
@@ -912,9 +966,24 @@ void brw_vs_emit( struct brw_vs_compile *c )
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
    brw_set_access_mode(p, BRW_ALIGN_16);
    
+   /* Message registers can't be read, so copy the output into GRF register
+      if they are used in source registers */
+   for (insn = 0; insn < nr_insns; insn++) {
+       GLuint i;
+       struct prog_instruction *inst = &c->vp->program.Base.Instructions[insn];
+       for (i = 0; i < 3; i++) {
+	   struct prog_src_register *src = &inst->SrcReg[i];
+	   GLuint index = src->Index;
+	   GLuint file = src->File;	
+	   if (file == PROGRAM_OUTPUT && index != VERT_RESULT_HPOS)
+	       c->output_regs[index].used_in_src = GL_TRUE;
+       }
+   }
+
    /* Static register allocation
     */
    brw_vs_alloc_regs(c);
+   brw_MOV(p, get_addr_reg(stack_index), brw_address(c->stack));
 
    for (insn = 0; insn < nr_insns; insn++) {
 
@@ -924,17 +993,29 @@ void brw_vs_emit( struct brw_vs_compile *c )
       
       /* Get argument regs.  SWZ is special and does this itself.
        */
+      inst->Data = &p->store[p->nr_insn];
       if (inst->Opcode != OPCODE_SWZ)
-	 for (i = 0; i < 3; i++) 
-	    args[i] = get_arg(c, inst->SrcReg[i]);
+	  for (i = 0; i < 3; i++) {
+	      struct prog_src_register *src = &inst->SrcReg[i];
+	      index = src->Index;
+	      file = src->File;	
+	      if (file == PROGRAM_OUTPUT&&c->output_regs[index].used_in_src)
+		  args[i] = c->output_regs[index].reg;
+	      else
+		  args[i] = get_arg(c, src);
+	  }
 
       /* Get dest regs.  Note that it is possible for a reg to be both
        * dst and arg, given the static allocation of registers.  So
        * care needs to be taken emitting multi-operation instructions.
-       */
-      dst = get_dst(c, inst->DstReg);
+       */ 
+      index = inst->DstReg.Index;
+      file = inst->DstReg.File;
+      if (file == PROGRAM_OUTPUT && c->output_regs[index].used_in_src)
+	  dst = c->output_regs[index].reg;
+      else
+	  dst = get_dst(c, inst->DstReg);
 
-      
       switch (inst->Opcode) {
       case OPCODE_ABS:
 	 brw_MOV(p, dst, brw_abs(args[0]));
@@ -942,6 +1023,9 @@ void brw_vs_emit( struct brw_vs_compile *c )
       case OPCODE_ADD:
 	 brw_ADD(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_COS:
+	 emit_math1(c, BRW_MATH_FUNCTION_COS, dst, args[0], BRW_MATH_PRECISION_FULL);
+	 break;
       case OPCODE_DP3:
 	 brw_DP3(p, dst, args[0], args[1]);
 	 break;
@@ -1003,12 +1087,28 @@ void brw_vs_emit( struct brw_vs_compile *c )
       case OPCODE_RSQ:
 	 emit_math1(c, BRW_MATH_FUNCTION_RSQ, dst, args[0], BRW_MATH_PRECISION_FULL);
 	 break;
+
+      case OPCODE_SEQ:
+         emit_seq(p, dst, args[0], args[1]);
+         break;
+      case OPCODE_SIN:
+	 emit_math1(c, BRW_MATH_FUNCTION_SIN, dst, args[0], BRW_MATH_PRECISION_FULL);
+	 break;
+      case OPCODE_SNE:
+         emit_sne(p, dst, args[0], args[1]);
+         break;
       case OPCODE_SGE:
 	 emit_sge(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_SGT:
+         emit_sgt(p, dst, args[0], args[1]);
+        break;
       case OPCODE_SLT:
 	 emit_slt(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_SLE:
+         emit_sle(p, dst, args[0], args[1]);
+         break;
       case OPCODE_SUB:
 	 brw_ADD(p, dst, args[0], negate(args[1]));
 	 break;
@@ -1021,21 +1121,85 @@ void brw_vs_emit( struct brw_vs_compile *c )
       case OPCODE_XPD:
 	 emit_xpd(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_IF:
+	 assert(if_insn < MAX_IFSN);
+         if_inst[if_insn++] = brw_IF(p, BRW_EXECUTE_8);
+	 break;
+      case OPCODE_ELSE:
+	 if_inst[if_insn-1] = brw_ELSE(p, if_inst[if_insn-1]);
+	 break;
+      case OPCODE_ENDIF:
+         assert(if_insn > 0);
+	 brw_ENDIF(p, if_inst[--if_insn]);
+	 break;			
+      case OPCODE_BRA:
+         brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+         brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
+         brw_set_predicate_control_flag_value(p, 0xff);
+        break;
+      case OPCODE_CAL:
+	 brw_set_access_mode(p, BRW_ALIGN_1);
+	 brw_ADD(p, deref_1d(stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
+	 brw_set_access_mode(p, BRW_ALIGN_16);
+	 brw_ADD(p, get_addr_reg(stack_index),
+			 get_addr_reg(stack_index), brw_imm_d(4));
+	 inst->Data = &p->store[p->nr_insn];
+	 brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
+        break;
+      case OPCODE_RET:
+	 brw_ADD(p, get_addr_reg(stack_index),
+			 get_addr_reg(stack_index), brw_imm_d(-4));
+	 brw_set_access_mode(p, BRW_ALIGN_1);
+         brw_MOV(p, brw_ip_reg(), deref_1d(stack_index, 0));
+	 brw_set_access_mode(p, BRW_ALIGN_16);
       case OPCODE_END:	
+         brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
+        break;
       case OPCODE_PRINT:
+      case OPCODE_BGNSUB:
+      case OPCODE_ENDSUB:
 	 break;
       default:
+	 _mesa_printf("Unsupported opcode %i (%s) in vertex shader\n",
+		      inst->Opcode, inst->Opcode < MAX_OPCODE ?
+				    _mesa_opcode_string(inst->Opcode) :
+				    "unknown");
 	 break;
       }
 
+      if ((inst->DstReg.File == PROGRAM_OUTPUT)
+          && (inst->DstReg.Index != VERT_RESULT_HPOS)
+          && c->output_regs[inst->DstReg.Index].used_in_src) {
+         brw_MOV(p, get_dst(c, inst->DstReg), dst);
+      }
+
+      /* Result color clamping.
+       *
+       * When destination register is an output register and
+       * it's primary/secondary front/back color, we have to clamp
+       * the result to [0,1]. This is done by enabling the
+       * saturation bit for the last instruction.
+       *
+       * We don't use brw_set_saturate() as it modifies
+       * p->current->header.saturate, which affects all the subsequent
+       * instructions. Instead, we directly modify the header
+       * of the last (already stored) instruction.
+       */
+      if (inst->DstReg.File == PROGRAM_OUTPUT) {
+         if ((inst->DstReg.Index == VERT_RESULT_COL0)
+             || (inst->DstReg.Index == VERT_RESULT_COL1)
+             || (inst->DstReg.Index == VERT_RESULT_BFC0)
+             || (inst->DstReg.Index == VERT_RESULT_BFC1)) {
+            p->store[p->nr_insn-1].header.saturate = 1;
+         }
+      }
+
       release_tmps(c);
    }
 
+   end_inst = &p->store[p->nr_insn];
    emit_vertex_write(c);
-
+   post_vs_emit(c, end_inst);
+   for (insn = 0; insn < nr_insns; insn++)
+       c->vp->program.Base.Instructions[insn].Data = NULL;
 }
-
-
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_vs_state.c b/src/mesa/drivers/dri/i965/brw_vs_state.c
index c225bf8f5c5..942581696d8 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_state.c
@@ -34,62 +34,123 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "macros.h"
+#include "main/macros.h"
 
-static void upload_vs_unit( struct brw_context *brw )
-{
-   struct brw_vs_unit_state vs;
-
-   memset(&vs, 0, sizeof(vs));
+struct brw_vs_unit_key {
+   unsigned int total_grf;
+   unsigned int urb_entry_read_length;
+   unsigned int curb_entry_read_length;
 
-   /* CACHE_NEW_VS_PROG */
-   vs.thread0.kernel_start_pointer = brw->vs.prog_gs_offset >> 6;
-   vs.thread0.grf_reg_count = ((brw->vs.prog_data->total_grf-1) & ~15) / 16;
-   vs.thread3.urb_entry_read_length = brw->vs.prog_data->urb_read_length;
-   vs.thread3.const_urb_entry_read_length = brw->vs.prog_data->curb_read_length;
-   vs.thread3.dispatch_grf_start_reg = 1;
+   unsigned int curbe_offset;
 
+   unsigned int nr_urb_entries, urb_size;
+};
 
-   /* BRW_NEW_URB_FENCE  */
-   vs.thread4.nr_urb_entries = brw->urb.nr_vs_entries; 
-   vs.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-   vs.thread4.max_threads = MIN2(
-      MAX2(0, (brw->urb.nr_vs_entries - 6) / 2 - 1), 
-      15);
-
+static void
+vs_unit_populate_key(struct brw_context *brw, struct brw_vs_unit_key *key)
+{
+   memset(key, 0, sizeof(*key));
 
+   /* CACHE_NEW_VS_PROG */
+   key->total_grf = brw->vs.prog_data->total_grf;
+   key->urb_entry_read_length = brw->vs.prog_data->urb_read_length;
+   key->curb_entry_read_length = brw->vs.prog_data->curb_read_length;
 
-   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
-      vs.thread4.max_threads = 0; 
+   /* BRW_NEW_URB_FENCE */
+   key->nr_urb_entries = brw->urb.nr_vs_entries;
+   key->urb_size = brw->urb.vsize;
 
    /* BRW_NEW_CURBE_OFFSETS, _NEW_TRANSFORM */
    if (brw->attribs.Transform->ClipPlanesEnabled) {
       /* Note that we read in the userclip planes as well, hence
        * clip_start:
        */
-      vs.thread3.const_urb_entry_read_offset = brw->curbe.clip_start * 2;
+      key->curbe_offset = brw->curbe.clip_start;
    }
    else {
-      vs.thread3.const_urb_entry_read_offset = brw->curbe.vs_start * 2;
+      key->curbe_offset = brw->curbe.vs_start;
    }
+}
+
+static dri_bo *
+vs_unit_create_from_key(struct brw_context *brw, struct brw_vs_unit_key *key)
+{
+   struct brw_vs_unit_state vs;
+   dri_bo *bo;
+   int chipset_max_threads;
 
+   memset(&vs, 0, sizeof(vs));
+
+   vs.thread0.kernel_start_pointer = brw->vs.prog_bo->offset >> 6; /* reloc */
+   vs.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
    vs.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
+   /* Choosing multiple program flow means that we may get 2-vertex threads,
+    * which will have the channel mask for dwords 4-7 enabled in the thread,
+    * and those dwords will be written to the second URB handle when we
+    * brw_urb_WRITE() results.
+    */
+   vs.thread1.single_program_flow = 0;
+   vs.thread3.urb_entry_read_length = key->urb_entry_read_length;
+   vs.thread3.const_urb_entry_read_length = key->curb_entry_read_length;
+   vs.thread3.dispatch_grf_start_reg = 1;
    vs.thread3.urb_entry_read_offset = 0;
+   vs.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;
+
+   vs.thread4.nr_urb_entries = key->nr_urb_entries;
+   vs.thread4.urb_entry_allocation_size = key->urb_size - 1;
+
+   if (BRW_IS_G4X(brw))
+      chipset_max_threads = 32;
+   else
+      chipset_max_threads = 16;
+   vs.thread4.max_threads = CLAMP(key->nr_urb_entries / 2,
+				  1, chipset_max_threads) - 1;
+
+   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
+      vs.thread4.max_threads = 0;
 
    /* No samplers for ARB_vp programs:
     */
    vs.vs5.sampler_count = 0;
 
    if (INTEL_DEBUG & DEBUG_STATS)
-      vs.thread4.stats_enable = 1; 
+      vs.thread4.stats_enable = 1;
 
-   /* Vertex program always enabled: 
+   /* Vertex program always enabled:
     */
    vs.vs6.vs_enable = 1;
 
-   brw->vs.state_gs_offset = brw_cache_data( &brw->cache[BRW_VS_UNIT], &vs );
+   bo = brw_upload_cache(&brw->cache, BRW_VS_UNIT,
+			 key, sizeof(*key),
+			 &brw->vs.prog_bo, 1,
+			 &vs, sizeof(vs),
+			 NULL, NULL);
+
+   /* Emit VS program relocation */
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION, 0,
+		     vs.thread0.grf_reg_count << 1,
+		     offsetof(struct brw_vs_unit_state, thread0),
+		     brw->vs.prog_bo);
+
+   return bo;
 }
 
+static void prepare_vs_unit(struct brw_context *brw)
+{
+   struct brw_vs_unit_key key;
+
+   vs_unit_populate_key(brw, &key);
+
+   dri_bo_unreference(brw->vs.state_bo);
+   brw->vs.state_bo = brw_search_cache(&brw->cache, BRW_VS_UNIT,
+				       &key, sizeof(key),
+				       &brw->vs.prog_bo, 1,
+				       NULL);
+   if (brw->vs.state_bo == NULL) {
+      brw->vs.state_bo = vs_unit_create_from_key(brw, &key);
+   }
+}
 
 const struct brw_tracked_state brw_vs_unit = {
    .dirty = {
@@ -98,5 +159,5 @@ const struct brw_tracked_state brw_vs_unit = {
 		BRW_NEW_URB_FENCE),
       .cache = CACHE_NEW_VS_PROG
    },
-   .update = upload_vs_unit
+   .prepare = prepare_vs_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_vs_tnl.c b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
index b69be350a92..eacc289f1f1 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_tnl.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
@@ -30,1620 +30,18 @@
  */
 
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "shader/prog_parameter.h"
-#include "shader/prog_print.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "brw_vs.h"
 #include "brw_state.h"
 
 
-struct state_key {
-   unsigned light_global_enabled:1;
-   unsigned light_local_viewer:1;
-   unsigned light_twoside:1;
-   unsigned light_color_material:1;
-   unsigned light_color_material_mask:12;
-   unsigned light_material_mask:12;
-   unsigned normalize:1;
-   unsigned rescale_normals:1;
-   unsigned fog_source_is_depth:1;
-   unsigned tnl_do_vertex_fog:1;
-   unsigned separate_specular:1;
-   unsigned fog_option:2;
-   unsigned point_attenuated:1;
-   unsigned texture_enabled_global:1;
-   unsigned fragprog_inputs_read:12;
-
-   struct {
-      unsigned light_enabled:1;
-      unsigned light_eyepos3_is_zero:1;
-      unsigned light_spotcutoff_is_180:1;
-      unsigned light_attenuated:1;      
-      unsigned texunit_really_enabled:1;
-      unsigned texmat_enabled:1;
-      unsigned texgen_enabled:4;
-      unsigned texgen_mode0:4;
-      unsigned texgen_mode1:4;
-      unsigned texgen_mode2:4;
-      unsigned texgen_mode3:4;
-   } unit[8];
-};
-
-
-
-#define FOG_NONE   0
-#define FOG_LINEAR 1
-#define FOG_EXP    2
-#define FOG_EXP2   3
-
-static GLuint translate_fog_mode( GLenum mode )
-{
-   switch (mode) {
-   case GL_LINEAR: return FOG_LINEAR;
-   case GL_EXP: return FOG_EXP;
-   case GL_EXP2: return FOG_EXP2;
-   default: return FOG_NONE;
-   }
-}
-
-#define TXG_NONE           0
-#define TXG_OBJ_LINEAR     1
-#define TXG_EYE_LINEAR     2
-#define TXG_SPHERE_MAP     3
-#define TXG_REFLECTION_MAP 4
-#define TXG_NORMAL_MAP     5
-
-static GLuint translate_texgen( GLboolean enabled, GLenum mode )
-{
-   if (!enabled)
-      return TXG_NONE;
-
-   switch (mode) {
-   case GL_OBJECT_LINEAR: return TXG_OBJ_LINEAR;
-   case GL_EYE_LINEAR: return TXG_EYE_LINEAR;
-   case GL_SPHERE_MAP: return TXG_SPHERE_MAP;
-   case GL_REFLECTION_MAP_NV: return TXG_REFLECTION_MAP;
-   case GL_NORMAL_MAP_NV: return TXG_NORMAL_MAP;
-   default: return TXG_NONE;
-   }
-}
-
-static void make_state_key( GLcontext *ctx, struct state_key *key )
-{
-   struct brw_context *brw = brw_context(ctx);
-   const struct gl_fragment_program *fp = brw->fragment_program;
-   GLuint i;
-
-   /* This now relies on texenvprogram.c being active:
-    */
-   assert(fp);
-
-   memset(key, 0, sizeof(*key));
-
-   /* BRW_NEW_FRAGMENT_PROGRAM */
-   key->fragprog_inputs_read = fp->Base.InputsRead;
-
-   /* _NEW_LIGHT */
-   key->separate_specular = (brw->attribs.Light->Model.ColorControl ==
-			     GL_SEPARATE_SPECULAR_COLOR);
-
-   /* _NEW_LIGHT */
-   if (brw->attribs.Light->Enabled) {
-      key->light_global_enabled = 1;
-
-      if (brw->attribs.Light->Model.LocalViewer)
-	 key->light_local_viewer = 1;
-
-      if (brw->attribs.Light->Model.TwoSide)
-	 key->light_twoside = 1;
-
-      if (brw->attribs.Light->ColorMaterialEnabled) {
-	 key->light_color_material = 1;
-	 key->light_color_material_mask = brw->attribs.Light->ColorMaterialBitmask;
-      }
-
-      /* BRW_NEW_INPUT_VARYING */
-
-      /* For these programs, material values are stuffed into the
-       * generic slots:
-       */
-      for (i = 0 ; i < MAT_ATTRIB_MAX ; i++) 
-	 if (brw->vb.info.varying & (1<<(VERT_ATTRIB_GENERIC0 + i))) 
-	    key->light_material_mask |= 1<<i;
-
-      for (i = 0; i < MAX_LIGHTS; i++) {
-	 struct gl_light *light = &brw->attribs.Light->Light[i];
-
-	 if (light->Enabled) {
-	    key->unit[i].light_enabled = 1;
-
-	    if (light->EyePosition[3] == 0.0)
-	       key->unit[i].light_eyepos3_is_zero = 1;
-	    
-	    if (light->SpotCutoff == 180.0)
-	       key->unit[i].light_spotcutoff_is_180 = 1;
-
-	    if (light->ConstantAttenuation != 1.0 ||
-		light->LinearAttenuation != 0.0 ||
-		light->QuadraticAttenuation != 0.0)
-	       key->unit[i].light_attenuated = 1;
-	 }
-      }
-   }
-
-   /* _NEW_TRANSFORM */
-   if (brw->attribs.Transform->Normalize)
-      key->normalize = 1;
-
-   if (brw->attribs.Transform->RescaleNormals)
-      key->rescale_normals = 1;
-
-   /* BRW_NEW_FRAGMENT_PROGRAM */
-   key->fog_option = translate_fog_mode(fp->FogOption);
-   if (key->fog_option)
-      key->fragprog_inputs_read |= FRAG_BIT_FOGC;
-   
-   /* _NEW_FOG */
-   if (brw->attribs.Fog->FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT)
-      key->fog_source_is_depth = 1;
-   
-   /* _NEW_HINT, ??? */
-   if (1)
-      key->tnl_do_vertex_fog = 1;
-
-   /* _NEW_POINT */
-   if (brw->attribs.Point->_Attenuated)
-      key->point_attenuated = 1;
-
-   /* _NEW_TEXTURE */
-   if (brw->attribs.Texture->_TexGenEnabled ||
-       brw->attribs.Texture->_TexMatEnabled ||
-       brw->attribs.Texture->_EnabledUnits)
-      key->texture_enabled_global = 1;
-      
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
-      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
-
-      if (texUnit->_ReallyEnabled)
- 	 key->unit[i].texunit_really_enabled = 1;
-
-      if (brw->attribs.Texture->_TexMatEnabled & ENABLE_TEXMAT(i))      
-	 key->unit[i].texmat_enabled = 1;
-      
-      if (texUnit->TexGenEnabled) {
-	 key->unit[i].texgen_enabled = 1;
-      
-	 key->unit[i].texgen_mode0 = 
-	    translate_texgen( texUnit->TexGenEnabled & (1<<0),
-			      texUnit->GenModeS );
-	 key->unit[i].texgen_mode1 = 
-	    translate_texgen( texUnit->TexGenEnabled & (1<<1),
-			      texUnit->GenModeT );
-	 key->unit[i].texgen_mode2 = 
-	    translate_texgen( texUnit->TexGenEnabled & (1<<2),
-			      texUnit->GenModeR );
-	 key->unit[i].texgen_mode3 = 
-	    translate_texgen( texUnit->TexGenEnabled & (1<<3),
-			      texUnit->GenModeQ );
-      }
-   }
-}
-
-
-   
-/* Very useful debugging tool - produces annotated listing of
- * generated program with line/function references for each
- * instruction back into this file:
- */
-#define DISASSEM 0
-
-/* Should be tunable by the driver - do we want to do matrix
- * multiplications with DP4's or with MUL/MAD's?  SSE works better
- * with the latter, drivers may differ.
- */
-#define PREFER_DP4 1
-
-
-/* Use uregs to represent registers internally, translate to Mesa's
- * expected formats on emit.  
- *
- * NOTE: These are passed by value extensively in this file rather
- * than as usual by pointer reference.  If this disturbs you, try
- * remembering they are just 32bits in size.
- *
- * GCC is smart enough to deal with these dword-sized structures in
- * much the same way as if I had defined them as dwords and was using
- * macros to access and set the fields.  This is much nicer and easier
- * to evolve.
- */
-struct ureg {
-   GLuint file:4;
-   GLint idx:8;      /* relative addressing may be negative */
-   GLuint negate:1;
-   GLuint swz:12;
-   GLuint pad:7;
-};
-
-
-struct tnl_program {
-   const struct state_key *state;
-   struct gl_vertex_program *program;
-   
-   GLuint nr_instructions;
-   GLuint temp_in_use;
-   GLuint temp_reserved;
-   
-   struct ureg eye_position;
-   struct ureg eye_position_normalized;
-   struct ureg eye_normal;
-   struct ureg identity;
-
-   GLuint materials;
-   GLuint color_materials;
-};
-
-
-const static struct ureg undef = { 
-   PROGRAM_UNDEFINED,
-   ~0,
-   0,
-   0,
-   0
-};
-
-/* Local shorthand:
- */
-#define X    SWIZZLE_X
-#define Y    SWIZZLE_Y
-#define Z    SWIZZLE_Z
-#define W    SWIZZLE_W
-
-
-/* Construct a ureg:
- */
-static struct ureg make_ureg(GLuint file, GLint idx)
-{
-   struct ureg reg;
-   reg.file = file;
-   reg.idx = idx;
-   reg.negate = 0;
-   reg.swz = SWIZZLE_NOOP;
-   reg.pad = 0;
-   return reg;
-}
-
-
-
-static struct ureg ureg_negate( struct ureg reg )
-{
-   reg.negate ^= 1;
-   return reg;
-} 
-
-
-static struct ureg swizzle( struct ureg reg, int x, int y, int z, int w )
-{
-   reg.swz = MAKE_SWIZZLE4(GET_SWZ(reg.swz, x),
-			   GET_SWZ(reg.swz, y),
-			   GET_SWZ(reg.swz, z),
-			   GET_SWZ(reg.swz, w));
-
-   return reg;
-}
-
-static struct ureg swizzle1( struct ureg reg, int x )
-{
-   return swizzle(reg, x, x, x, x);
-}
-
-static struct ureg get_temp( struct tnl_program *p )
-{
-   int bit = ffs( ~p->temp_in_use );
-   if (!bit) {
-      fprintf(stderr, "%s: out of temporaries\n", __FILE__);
-      assert(0);
-   }
-
-   if (bit > p->program->Base.NumTemporaries)
-      p->program->Base.NumTemporaries = bit;
-
-   p->temp_in_use |= 1<<(bit-1);
-   return make_ureg(PROGRAM_TEMPORARY, bit-1);
-}
-
-static struct ureg reserve_temp( struct tnl_program *p )
-{
-   struct ureg temp = get_temp( p );
-   p->temp_reserved |= 1<<temp.idx;
-   return temp;
-}
-
-static void release_temp( struct tnl_program *p, struct ureg reg )
-{
-   if (reg.file == PROGRAM_TEMPORARY) {
-      p->temp_in_use &= ~(1<<reg.idx);
-      p->temp_in_use |= p->temp_reserved; /* can't release reserved temps */
-   }
-}
-
-static void release_temps( struct tnl_program *p )
-{
-   p->temp_in_use = p->temp_reserved;
-}
-
-
-
-static struct ureg register_input( struct tnl_program *p, GLuint input )
-{
-   assert(input < 32);
-
-   p->program->Base.InputsRead |= (1<<input);
-   return make_ureg(PROGRAM_INPUT, input);
-}
-
-static struct ureg register_output( struct tnl_program *p, GLuint output )
-{
-   p->program->Base.OutputsWritten |= (1<<output);
-   return make_ureg(PROGRAM_OUTPUT, output);
-}
-
-static struct ureg register_const4f( struct tnl_program *p, 
-			      GLfloat s0,
-			      GLfloat s1,
-			      GLfloat s2,
-			      GLfloat s3)
-{
-   GLfloat values[4];
-   GLint idx;
-   GLuint swizzle;
-   values[0] = s0;
-   values[1] = s1;
-   values[2] = s2;
-   values[3] = s3;
-   idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
-                                     &swizzle);
-   /* XXX what about swizzle? */
-   return make_ureg(PROGRAM_STATE_VAR, idx);
-}
-
-#define register_const1f(p, s0)         register_const4f(p, s0, 0, 0, 1)
-#define register_scalar_const(p, s0)    register_const4f(p, s0, s0, s0, s0)
-#define register_const2f(p, s0, s1)     register_const4f(p, s0, s1, 0, 1)
-#define register_const3f(p, s0, s1, s2) register_const4f(p, s0, s1, s2, 1)
-
-static GLboolean is_undef( struct ureg reg )
-{
-   return reg.file == PROGRAM_UNDEFINED;
-}
-
-static struct ureg get_identity_param( struct tnl_program *p )
-{
-   if (is_undef(p->identity)) 
-      p->identity = register_const4f(p, 0,0,0,1);
-
-   return p->identity;
-}
-
-static struct ureg register_param5( struct tnl_program *p, 
-                                    GLint s0,
-                                    GLint s1,
-                                    GLint s2,
-                                    GLint s3,
-                                    GLint s4)
-{
-   gl_state_index tokens[STATE_LENGTH];
-   GLint idx;
-   tokens[0] = s0;
-   tokens[1] = s1;
-   tokens[2] = s2;
-   tokens[3] = s3;
-   tokens[4] = s4;
-   idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
-   return make_ureg(PROGRAM_STATE_VAR, idx);
-}
-
-
-#define register_param1(p,s0)          register_param5(p,s0,0,0,0,0)
-#define register_param2(p,s0,s1)       register_param5(p,s0,s1,0,0,0)
-#define register_param3(p,s0,s1,s2)    register_param5(p,s0,s1,s2,0,0)
-#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
-
-
-static void register_matrix_param5( struct tnl_program *p,
-				    GLint s0, /* matrix name */
-				    GLint s1, /* texture matrix number */
-				    GLint s2, /* first row */
-				    GLint s3, /* last row */
-				    GLint s4, /* modifier */
-				    struct ureg *matrix )
-{
-   GLint i;
-
-   /* This is a bit sad as the support is there to pull the whole
-    * matrix out in one go:
-    */
-   for (i = 0; i <= s3 - s2; i++) 
-      matrix[i] = register_param5( p, s0, s1, i, i, s4 );
-}
-
-
-static void emit_arg( struct prog_src_register *src,
-		      struct ureg reg )
-{
-   src->File = reg.file;
-   src->Index = reg.idx;
-   src->Swizzle = reg.swz;
-   src->RelAddr = 0;
-   src->NegateBase = reg.negate;
-   src->Abs = 0;
-   src->NegateAbs = 0;
-}
-
-static void emit_dst( struct prog_dst_register *dst,
-		      struct ureg reg, GLuint mask )
-{
-   dst->File = reg.file;
-   dst->Index = reg.idx;
-   /* allow zero as a shorthand for xyzw */
-   dst->WriteMask = mask ? mask : WRITEMASK_XYZW; 
-   dst->CondMask = 0;
-   dst->CondSwizzle = 0;
-   dst->CondSrc = 0;
-   dst->pad = 0;
-}
-
-static void debug_insn( struct prog_instruction *inst, const char *fn,
-			GLuint line )
-{
-   if (DISASSEM) {
-      static const char *last_fn;
-   
-      if (fn != last_fn) {
-	 last_fn = fn;
-	 _mesa_printf("%s:\n", fn);
-      }
-	 
-      _mesa_printf("%d:\t", line);
-      _mesa_print_instruction(inst);
-   }
-}
-
-
-static void emit_op3fn(struct tnl_program *p,
-		       GLuint op,
-		       struct ureg dest,
-		       GLuint mask,
-		       struct ureg src0,
-		       struct ureg src1,
-		       struct ureg src2,
-		       const char *fn,
-		       GLuint line)
-{
-   GLuint nr = p->program->Base.NumInstructions++;
-      
-   if (nr >= p->nr_instructions) {
-      p->program->Base.Instructions = 
-	 _mesa_realloc(p->program->Base.Instructions,
-		       sizeof(struct prog_instruction) * p->nr_instructions,
-		       sizeof(struct prog_instruction) * (p->nr_instructions *= 2));
-   }
-
-   {      
-      struct prog_instruction *inst = &p->program->Base.Instructions[nr];
-      memset(inst, 0, sizeof(*inst));
-      inst->Opcode = op; 
-      inst->StringPos = 0;
-      inst->Data = 0;
-   
-      emit_arg( &inst->SrcReg[0], src0 );
-      emit_arg( &inst->SrcReg[1], src1 );
-      emit_arg( &inst->SrcReg[2], src2 );   
-
-      emit_dst( &inst->DstReg, dest, mask );
-
-      debug_insn(inst, fn, line);
-   }
-}
-
-   
-
-#define emit_op3(p, op, dst, mask, src0, src1, src2) \
-   emit_op3fn(p, op, dst, mask, src0, src1, src2, __FUNCTION__, __LINE__)
-
-#define emit_op2(p, op, dst, mask, src0, src1) \
-    emit_op3fn(p, op, dst, mask, src0, src1, undef, __FUNCTION__, __LINE__)
-
-#define emit_op1(p, op, dst, mask, src0) \
-    emit_op3fn(p, op, dst, mask, src0, undef, undef, __FUNCTION__, __LINE__)
-
-
-static struct ureg make_temp( struct tnl_program *p, struct ureg reg )
-{
-   if (reg.file == PROGRAM_TEMPORARY && 
-       !(p->temp_reserved & (1<<reg.idx)))
-      return reg;
-   else {
-      struct ureg temp = get_temp(p);
-      emit_op1(p, OPCODE_MOV, temp, 0, reg);
-      return temp;
-   }
-}
-
-
-/* Currently no tracking performed of input/output/register size or
- * active elements.  Could be used to reduce these operations, as
- * could the matrix type.
- */
-static void emit_matrix_transform_vec4( struct tnl_program *p,
-					struct ureg dest,
-					const struct ureg *mat,
-					struct ureg src)
-{
-   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_X, src, mat[0]);
-   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_Y, src, mat[1]);
-   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_Z, src, mat[2]);
-   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_W, src, mat[3]);
-}
-
-/* This version is much easier to implement if writemasks are not
- * supported natively on the target or (like SSE), the target doesn't
- * have a clean/obvious dotproduct implementation.
- */
-static void emit_transpose_matrix_transform_vec4( struct tnl_program *p,
-						  struct ureg dest,
-						  const struct ureg *mat,
-						  struct ureg src)
-{
-   struct ureg tmp;
-
-   if (dest.file != PROGRAM_TEMPORARY)
-      tmp = get_temp(p);
-   else
-      tmp = dest;
-
-   emit_op2(p, OPCODE_MUL, tmp, 0, swizzle1(src,X), mat[0]);
-   emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Y), mat[1], tmp);
-   emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Z), mat[2], tmp);
-   emit_op3(p, OPCODE_MAD, dest, 0, swizzle1(src,W), mat[3], tmp);
-
-   if (dest.file != PROGRAM_TEMPORARY)
-      release_temp(p, tmp);
-}
-
-static void emit_matrix_transform_vec3( struct tnl_program *p,
-					struct ureg dest,
-					const struct ureg *mat,
-					struct ureg src)
-{
-   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_X, src, mat[0]);
-   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_Y, src, mat[1]);
-   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_Z, src, mat[2]);
-}
-
-
-static void emit_normalize_vec3( struct tnl_program *p,
-				 struct ureg dest,
-				 struct ureg src )
-{
-   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_W, src, src);
-   emit_op1(p, OPCODE_RSQ, dest, WRITEMASK_W, swizzle1(dest,W));
-   emit_op2(p, OPCODE_MUL, dest, WRITEMASK_XYZ, src, swizzle1(dest,W));
-}
-
-static void emit_passthrough( struct tnl_program *p, 
-			      GLuint input,
-			      GLuint output )
-{
-   struct ureg out = register_output(p, output);
-   emit_op1(p, OPCODE_MOV, out, 0, register_input(p, input)); 
-}
-
-static struct ureg get_eye_position( struct tnl_program *p )
-{
-   if (is_undef(p->eye_position)) {
-      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
-      struct ureg modelview[4];
-
-      p->eye_position = reserve_temp(p);
-
-      if (PREFER_DP4) {
-	 register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3, 
-				 0, modelview );
-
-	 emit_matrix_transform_vec4(p, p->eye_position, modelview, pos);
-      }
-      else {
-	 register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3, 
-				 STATE_MATRIX_TRANSPOSE, modelview );
-
-	 emit_transpose_matrix_transform_vec4(p, p->eye_position, modelview, pos);
-      }
-   }
-   
-   return p->eye_position;
-}
-
-
-#if 0
-static struct ureg get_eye_z( struct tnl_program *p )
-{
-   if (!is_undef(p->eye_position)) {
-      return swizzle1(p->eye_position, Z);
-   }
-   else if (!is_undef(p->eye_z)) {
-      struct ureg pos = register_input( p, BRW_ATTRIB_POS ); 
-      struct ureg modelview2;
-
-      p->eye_z = reserve_temp(p);
-
-      register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 2, 1, 
-			      STATE_MATRIX, &modelview2 );
-
-      emit_matrix_transform_vec4(p, p->eye_position, modelview, pos);
-      emit_op2(p, OPCODE_DP4, p->eye_z, WRITEMASK_Z, pos, modelview2);
-   }
-   
-   return swizzle1(p->eye_z, Z)
-}
-#endif
-
-
-
-static struct ureg get_eye_position_normalized( struct tnl_program *p )
-{
-   if (is_undef(p->eye_position_normalized)) {
-      struct ureg eye = get_eye_position(p);
-      p->eye_position_normalized = reserve_temp(p);
-      emit_normalize_vec3(p, p->eye_position_normalized, eye);
-   }
-   
-   return p->eye_position_normalized;
-}
-
-
-static struct ureg get_eye_normal( struct tnl_program *p )
-{
-   if (is_undef(p->eye_normal)) {
-      struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
-      struct ureg mvinv[3];
-
-      register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
-			      STATE_MATRIX_INVTRANS, mvinv );
-
-      p->eye_normal = reserve_temp(p);
-
-      /* Transform to eye space:
-       */
-      emit_matrix_transform_vec3( p, p->eye_normal, mvinv, normal );
-
-      /* Normalize/Rescale:
-       */
-      if (p->state->normalize) {
-	 emit_normalize_vec3( p, p->eye_normal, p->eye_normal );
-      }
-      else if (p->state->rescale_normals) {
-	 struct ureg rescale = register_param2(p, STATE_INTERNAL,
-					       STATE_NORMAL_SCALE);
-
-	 emit_op2( p, OPCODE_MUL, p->eye_normal, 0, p->eye_normal, 
-		   swizzle1(rescale, X));
-      }
-   }
-
-   return p->eye_normal;
-}
-
-
-
-static void build_hpos( struct tnl_program *p )
-{
-   struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
-   struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
-   struct ureg mvp[4];
-
-   if (PREFER_DP4) {
-      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
-			      0, mvp );
-      emit_matrix_transform_vec4( p, hpos, mvp, pos );
-   }
-   else {
-      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
-			      STATE_MATRIX_TRANSPOSE, mvp );
-      emit_transpose_matrix_transform_vec4( p, hpos, mvp, pos );
-   }
-}
-
-
-static GLuint material_attrib( GLuint side, GLuint property )
-{
-   return (property - STATE_AMBIENT) * 2 + side;
-}
-
-/* Get a bitmask of which material values vary on a per-vertex basis.
- */
-static void set_material_flags( struct tnl_program *p )
-{
-   p->color_materials = 0;
-   p->materials = 0;
-
-   if (p->state->light_color_material) {
-      p->materials = 
-	 p->color_materials = p->state->light_color_material_mask;
-   }
-
-   p->materials |= p->state->light_material_mask;
-}
-
-
-static struct ureg get_material( struct tnl_program *p, GLuint side, 
-				 GLuint property )
-{
-   GLuint attrib = material_attrib(side, property);
-
-   if (p->color_materials & (1<<attrib))
-      return register_input(p, VERT_ATTRIB_COLOR0);
-   else if (p->materials & (1<<attrib)) 
-      return register_input( p, attrib + _TNL_ATTRIB_MAT_FRONT_AMBIENT );
-   else
-      return register_param3( p, STATE_MATERIAL, side, property );
-}
-
-#define SCENE_COLOR_BITS(side) ((MAT_BIT_FRONT_EMISSION | \
-				 MAT_BIT_FRONT_AMBIENT | \
-				 MAT_BIT_FRONT_DIFFUSE) << (side))
-
-/* Either return a precalculated constant value or emit code to
- * calculate these values dynamically in the case where material calls
- * are present between begin/end pairs.
- *
- * Probably want to shift this to the program compilation phase - if
- * we always emitted the calculation here, a smart compiler could
- * detect that it was constant (given a certain set of inputs), and
- * lift it out of the main loop.  That way the programs created here
- * would be independent of the vertex_buffer details.
- */
-static struct ureg get_scenecolor( struct tnl_program *p, GLuint side )
-{
-   if (p->materials & SCENE_COLOR_BITS(side)) {
-      struct ureg lm_ambient = register_param1(p, STATE_LIGHTMODEL_AMBIENT);
-      struct ureg material_emission = get_material(p, side, STATE_EMISSION);
-      struct ureg material_ambient = get_material(p, side, STATE_AMBIENT);
-      struct ureg material_diffuse = get_material(p, side, STATE_DIFFUSE);
-      struct ureg tmp = make_temp(p, material_diffuse);
-      emit_op3(p, OPCODE_MAD, tmp,  WRITEMASK_XYZ, lm_ambient, 
-	       material_ambient, material_emission);
-      return tmp;
-   }
-   else
-      return register_param2( p, STATE_LIGHTMODEL_SCENECOLOR, side );
-}
-
-
-static struct ureg get_lightprod( struct tnl_program *p, GLuint light, 
-				  GLuint side, GLuint property )
-{
-   GLuint attrib = material_attrib(side, property);
-   if (p->materials & (1<<attrib)) {
-      struct ureg light_value = 
-	 register_param3(p, STATE_LIGHT, light, property);
-      struct ureg material_value = get_material(p, side, property);
-      struct ureg tmp = get_temp(p);
-      emit_op2(p, OPCODE_MUL, tmp,  0, light_value, material_value);
-      return tmp;
-   }
-   else
-      return register_param4(p, STATE_LIGHTPROD, light, side, property);
-}
-
-static struct ureg calculate_light_attenuation( struct tnl_program *p,
-						GLuint i, 
-						struct ureg VPpli,
-						struct ureg dist )
-{
-   struct ureg attenuation = register_param3(p, STATE_LIGHT, i,
-					     STATE_ATTENUATION);
-   struct ureg att = get_temp(p);
-
-   /* Calculate spot attenuation:
-    */
-   if (!p->state->unit[i].light_spotcutoff_is_180) {
-      struct ureg spot_dir_norm = register_param3(p, STATE_INTERNAL,
-						  STATE_SPOT_DIR_NORMALIZED, i);
-      struct ureg spot = get_temp(p);
-      struct ureg slt = get_temp(p);
-
-      emit_op2(p, OPCODE_DP3, spot, 0, ureg_negate(VPpli), spot_dir_norm);
-      emit_op2(p, OPCODE_SLT, slt, 0, swizzle1(spot_dir_norm,W), spot);
-      emit_op2(p, OPCODE_POW, spot, 0, spot, swizzle1(attenuation, W));
-      emit_op2(p, OPCODE_MUL, att, 0, slt, spot);
-
-      release_temp(p, spot);
-      release_temp(p, slt);
-   }
-
-   /* Calculate distance attenuation:
-    */
-   if (p->state->unit[i].light_attenuated) {
-
-      /* 1/d,d,d,1/d */
-      emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist); 
-      /* 1,d,d*d,1/d */
-      emit_op2(p, OPCODE_MUL, dist, WRITEMASK_XZ, dist, swizzle1(dist,Y)); 
-      /* 1/dist-atten */
-      emit_op2(p, OPCODE_DP3, dist, 0, attenuation, dist); 
-
-      if (!p->state->unit[i].light_spotcutoff_is_180) {
-	 /* dist-atten */
-	 emit_op1(p, OPCODE_RCP, dist, 0, dist); 
-	 /* spot-atten * dist-atten */
-	 emit_op2(p, OPCODE_MUL, att, 0, dist, att);	
-      } else {
-	 /* dist-atten */
-	 emit_op1(p, OPCODE_RCP, att, 0, dist); 
-      }
-   }
-
-   return att;
-}
-						
-
-
-
-
-/* Need to add some addtional parameters to allow lighting in object
- * space - STATE_SPOT_DIRECTION and STATE_HALF_VECTOR implicitly assume eye
- * space lighting.
- */
-static void build_lighting( struct tnl_program *p )
-{
-   const GLboolean twoside = p->state->light_twoside;
-   const GLboolean separate = p->state->separate_specular;
-   GLuint nr_lights = 0, count = 0;
-   struct ureg normal = get_eye_normal(p);
-   struct ureg lit = get_temp(p);
-   struct ureg dots = get_temp(p);
-   struct ureg _col0 = undef, _col1 = undef;
-   struct ureg _bfc0 = undef, _bfc1 = undef;
-   GLuint i;
-
-   for (i = 0; i < MAX_LIGHTS; i++) 
-      if (p->state->unit[i].light_enabled)
-	 nr_lights++;
-   
-   set_material_flags(p);
-
-   {
-      struct ureg shininess = get_material(p, 0, STATE_SHININESS);
-      emit_op1(p, OPCODE_MOV, dots,  WRITEMASK_W, swizzle1(shininess,X));
-      release_temp(p, shininess);
-
-      _col0 = make_temp(p, get_scenecolor(p, 0));
-      if (separate)
-	 _col1 = make_temp(p, get_identity_param(p));
-      else
-	 _col1 = _col0;
-
-   }
-
-   if (twoside) {
-      struct ureg shininess = get_material(p, 1, STATE_SHININESS);
-      emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
-	       ureg_negate(swizzle1(shininess,X)));
-      release_temp(p, shininess);
-
-      _bfc0 = make_temp(p, get_scenecolor(p, 1));
-      if (separate)
-	 _bfc1 = make_temp(p, get_identity_param(p));
-      else
-	 _bfc1 = _bfc0;
-   }
-
-
-   /* If no lights, still need to emit the scenecolor.
-    */
-   /* KW: changed to do this always - v1.17 "Fix lighting alpha result"? 
-    */
-   if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-   {
-      struct ureg res0 = register_output( p, VERT_RESULT_COL0 );
-      emit_op1(p, OPCODE_MOV, res0, 0, _col0);
-
-      if (twoside) {
-	 struct ureg res0 = register_output( p, VERT_RESULT_BFC0 );
-	 emit_op1(p, OPCODE_MOV, res0, 0, _bfc0);
-      }
-   }
-
-   if (separate && (p->state->fragprog_inputs_read & FRAG_BIT_COL1)) {
-
-      struct ureg res1 = register_output( p, VERT_RESULT_COL1 );
-      emit_op1(p, OPCODE_MOV, res1, 0, _col1);
-      
-      if (twoside) {
-	 struct ureg res1 = register_output( p, VERT_RESULT_BFC1 );
-	 emit_op1(p, OPCODE_MOV, res1, 0, _bfc1);
-      }
-   }
-      
-   if (nr_lights == 0) {
-      release_temps(p);
-      return;
-   }
-
-
-   for (i = 0; i < MAX_LIGHTS; i++) {
-      if (p->state->unit[i].light_enabled) {
-	 struct ureg half = undef;
-	 struct ureg att = undef, VPpli = undef;
-	  
-	 count++;
-
-	 if (p->state->unit[i].light_eyepos3_is_zero) {
-	    /* Can used precomputed constants in this case.
-	     * Attenuation never applies to infinite lights.
-	     */
-	    VPpli = register_param3(p, STATE_LIGHT, i, 
-				    STATE_POSITION_NORMALIZED); 
-            if (p->state->light_local_viewer) {
-                struct ureg eye_hat = get_eye_position_normalized(p);
-                half = get_temp(p);
-                emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
-                emit_normalize_vec3(p, half, half);
-            } else {
-                half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
-            }
-	 } 
-	 else {
-	    struct ureg Ppli = register_param3(p, STATE_LIGHT, i, 
-					       STATE_POSITION); 
-	    struct ureg V = get_eye_position(p);
-	    struct ureg dist = get_temp(p);
-
-	    VPpli = get_temp(p); 
-	    half = get_temp(p);
- 
-	    /* Calulate VPpli vector
-	     */
-	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); 
-
-	    /* Normalize VPpli.  The dist value also used in
-	     * attenuation below.
-	     */
-	    emit_op2(p, OPCODE_DP3, dist, 0, VPpli, VPpli);
-	    emit_op1(p, OPCODE_RSQ, dist, 0, dist);
-	    emit_op2(p, OPCODE_MUL, VPpli, 0, VPpli, dist);
-
-
-	    /* Calculate  attenuation:
-	     */ 
-	    if (!p->state->unit[i].light_spotcutoff_is_180 ||
-		p->state->unit[i].light_attenuated) {
-	       att = calculate_light_attenuation(p, i, VPpli, dist);
-	    }
-	 
-      
-	    /* Calculate viewer direction, or use infinite viewer:
-	     */
-	    if (p->state->light_local_viewer) {
-	       struct ureg eye_hat = get_eye_position_normalized(p);
-	       emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
-	    }
-	    else {
-	       struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z); 
-	       emit_op2(p, OPCODE_ADD, half, 0, VPpli, z_dir);
-	    }
-
-	    emit_normalize_vec3(p, half, half);
-
-	    release_temp(p, dist);
-	 }
-
-	 /* Calculate dot products:
-	  */
-	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
-	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
-
-	
-	 /* Front face lighting:
-	  */
-	 {
-	    struct ureg ambient = get_lightprod(p, i, 0, STATE_AMBIENT);
-	    struct ureg diffuse = get_lightprod(p, i, 0, STATE_DIFFUSE);
-	    struct ureg specular = get_lightprod(p, i, 0, STATE_SPECULAR);
-	    struct ureg res0, res1;
-	    GLuint mask0, mask1;
-
-	    emit_op1(p, OPCODE_LIT, lit, 0, dots);
-   
-	    if (!is_undef(att)) 
-	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
-
-
-	    mask0 = 0;
-	    mask1 = 0;
-	    res0 = _col0;
-	    res1 = _col1;
-	    
-	    if (count == nr_lights) {
-	       if (separate) {
-		  mask0 = WRITEMASK_XYZ;
-		  mask1 = WRITEMASK_XYZ;
-
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-		     res0 = register_output( p, VERT_RESULT_COL0 );
-
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL1)
-		     res1 = register_output( p, VERT_RESULT_COL1 );
-	       }
-	       else {
-		  mask1 = WRITEMASK_XYZ;
-
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-		     res1 = register_output( p, VERT_RESULT_COL0 );
-	       }
-	    } 
-
-	    emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
-	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _col0);
-	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _col1);
-      
-	    release_temp(p, ambient);
-	    release_temp(p, diffuse);
-	    release_temp(p, specular);
-	 }
-
-	 /* Back face lighting:
-	  */
-	 if (twoside) {
-	    struct ureg ambient = get_lightprod(p, i, 1, STATE_AMBIENT);
-	    struct ureg diffuse = get_lightprod(p, i, 1, STATE_DIFFUSE);
-	    struct ureg specular = get_lightprod(p, i, 1, STATE_SPECULAR);
-	    struct ureg res0, res1;
-	    GLuint mask0, mask1;
-	       
-	    emit_op1(p, OPCODE_LIT, lit, 0, ureg_negate(swizzle(dots,X,Y,W,Z)));
-
-	    if (!is_undef(att)) 
-	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
-
-	    mask0 = 0;
-	    mask1 = 0;
-	    res0 = _bfc0;
-	    res1 = _bfc1;
-
-	    if (count == nr_lights) {
-	       if (separate) {
-		  mask0 = WRITEMASK_XYZ;
-		  mask1 = WRITEMASK_XYZ;
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-		     res0 = register_output( p, VERT_RESULT_BFC0 );
-
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL1)
-		     res1 = register_output( p, VERT_RESULT_BFC1 );
-	       }
-	       else {
-		  mask1 = WRITEMASK_XYZ;
-
-		  if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-		     res1 = register_output( p, VERT_RESULT_BFC0 );
-	       }
-	    }
-
-	    emit_op3(p, OPCODE_MAD, _bfc0, 0, swizzle1(lit,X), ambient, _bfc0);
-	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
-	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
-
-	    release_temp(p, ambient);
-	    release_temp(p, diffuse);
-	    release_temp(p, specular);
-	 }
-
-	 release_temp(p, half);
-	 release_temp(p, VPpli);
-	 release_temp(p, att);
-      }
-   }
-
-   release_temps( p );
-}
-
-
-static void build_fog( struct tnl_program *p )
-{
-   struct ureg fog = register_output(p, VERT_RESULT_FOGC);
-   struct ureg input;
-   GLuint useabs = p->state->fog_source_is_depth && p->state->fog_option &&
-		   (p->state->fog_option != FOG_EXP2);
-
-   if (p->state->fog_source_is_depth) {
-      input = swizzle1(get_eye_position(p), Z);
-   }
-   else {
-      input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
-   }
-
-   if (p->state->fog_option &&
-       p->state->tnl_do_vertex_fog) {
-      struct ureg params = register_param2(p, STATE_INTERNAL,
-					   STATE_FOG_PARAMS_OPTIMIZED);
-      struct ureg tmp = get_temp(p);
-      struct ureg id = get_identity_param(p);
-
-      emit_op1(p, OPCODE_MOV, fog, 0, id);
-
-      if (useabs) {
-	 emit_op1(p, OPCODE_ABS, tmp, 0, input);
-      }
-
-      switch (p->state->fog_option) {
-      case FOG_LINEAR: {
-	 emit_op3(p, OPCODE_MAD, tmp, 0, useabs ? tmp : input,
-			swizzle1(params,X), swizzle1(params,Y));
-	 emit_op2(p, OPCODE_MAX, tmp, 0, tmp, swizzle1(id,X)); /* saturate */
-	 emit_op2(p, OPCODE_MIN, fog, WRITEMASK_X, tmp, swizzle1(id,W));
-	 break;
-      }
-      case FOG_EXP:
-	 emit_op2(p, OPCODE_MUL, tmp, 0, useabs ? tmp : input,
-			swizzle1(params,Z));
-	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, ureg_negate(tmp));
-	 break;
-      case FOG_EXP2:
-	 emit_op2(p, OPCODE_MUL, tmp, 0, input, swizzle1(params,W));
-	 emit_op2(p, OPCODE_MUL, tmp, 0, tmp, tmp);
-	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, ureg_negate(tmp));
-	 break;
-      }
-
-      release_temp(p, tmp);
-   }
-   else {
-      /* results = incoming fog coords (compute fog per-fragment later) 
-       *
-       * KW:  Is it really necessary to do anything in this case?
-       */
-      emit_op1(p, useabs ? OPCODE_ABS : OPCODE_MOV, fog, 0, input);
-   }
-}
- 
-static void build_reflect_texgen( struct tnl_program *p,
-				  struct ureg dest,
-				  GLuint writemask )
-{
-   struct ureg normal = get_eye_normal(p);
-   struct ureg eye_hat = get_eye_position_normalized(p);
-   struct ureg tmp = get_temp(p);
-
-   /* n.u */
-   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
-   /* 2n.u */
-   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
-   /* (-2n.u)n + u */
-   emit_op3(p, OPCODE_MAD, dest, writemask, ureg_negate(tmp), normal, eye_hat);
-
-   release_temp(p, tmp);
-}
-
-static void build_sphere_texgen( struct tnl_program *p,
-				 struct ureg dest,
-				 GLuint writemask )
-{
-   struct ureg normal = get_eye_normal(p);
-   struct ureg eye_hat = get_eye_position_normalized(p);
-   struct ureg tmp = get_temp(p);
-   struct ureg half = register_scalar_const(p, .5);
-   struct ureg r = get_temp(p);
-   struct ureg inv_m = get_temp(p);
-   struct ureg id = get_identity_param(p);
-
-   /* Could share the above calculations, but it would be
-    * a fairly odd state for someone to set (both sphere and
-    * reflection active for different texture coordinate
-    * components.  Of course - if two texture units enable
-    * reflect and/or sphere, things start to tilt in favour
-    * of seperating this out:
-    */
-
-   /* n.u */
-   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
-   /* 2n.u */
-   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
-   /* (-2n.u)n + u */
-   emit_op3(p, OPCODE_MAD, r, 0, ureg_negate(tmp), normal, eye_hat); 
-   /* r + 0,0,1 */
-   emit_op2(p, OPCODE_ADD, tmp, 0, r, swizzle(id,X,Y,W,Z)); 
-   /* rx^2 + ry^2 + (rz+1)^2 */
-   emit_op2(p, OPCODE_DP3, tmp, 0, tmp, tmp); 
-   /* 2/m */
-   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp); 
-   /* 1/m */
-   emit_op2(p, OPCODE_MUL, inv_m, 0, tmp, half); 
-   /* r/m + 1/2 */
-   emit_op3(p, OPCODE_MAD, dest, writemask, r, inv_m, half); 
-	       
-   release_temp(p, tmp);
-   release_temp(p, r);
-   release_temp(p, inv_m);
-}
-
-
-static void build_texture_transform( struct tnl_program *p )
-{
-   GLuint i, j;
-
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
-
-      if (!(p->state->fragprog_inputs_read & (FRAG_BIT_TEX0<<i)))
-	 continue;
-							     
-      if (p->state->unit[i].texgen_enabled || 
-	  p->state->unit[i].texmat_enabled) {
-	 
-	 GLuint texmat_enabled = p->state->unit[i].texmat_enabled;
-	 struct ureg out = register_output(p, VERT_RESULT_TEX0 + i);
-	 struct ureg out_texgen = undef;
-
-	 if (p->state->unit[i].texgen_enabled) {
-	    GLuint copy_mask = 0;
-	    GLuint sphere_mask = 0;
-	    GLuint reflect_mask = 0;
-	    GLuint normal_mask = 0;
-	    GLuint modes[4];
-	 
-	    if (texmat_enabled) 
-	       out_texgen = get_temp(p);
-	    else
-	       out_texgen = out;
-
-	    modes[0] = p->state->unit[i].texgen_mode0;
-	    modes[1] = p->state->unit[i].texgen_mode1;
-	    modes[2] = p->state->unit[i].texgen_mode2;
-	    modes[3] = p->state->unit[i].texgen_mode3;
-
-	    for (j = 0; j < 4; j++) {
-	       switch (modes[j]) {
-	       case TXG_OBJ_LINEAR: {
-		  struct ureg obj = register_input(p, VERT_ATTRIB_POS);
-		  struct ureg plane = 
-		     register_param3(p, STATE_TEXGEN, i,
-				     STATE_TEXGEN_OBJECT_S + j);
-
-		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
-			   obj, plane );
-		  break;
-	       }
-	       case TXG_EYE_LINEAR: {
-		  struct ureg eye = get_eye_position(p);
-		  struct ureg plane = 
-		     register_param3(p, STATE_TEXGEN, i, 
-				     STATE_TEXGEN_EYE_S + j);
-
-		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
-			   eye, plane );
-		  break;
-	       }
-	       case TXG_SPHERE_MAP: 
-		  sphere_mask |= WRITEMASK_X << j;
-		  break;
-	       case TXG_REFLECTION_MAP:
-		  reflect_mask |= WRITEMASK_X << j;
-		  break;
-	       case TXG_NORMAL_MAP: 
-		  normal_mask |= WRITEMASK_X << j;
-		  break;
-	       case TXG_NONE:
-		  copy_mask |= WRITEMASK_X << j;
-	       }
-
-	    }
-
-	 
-	    if (sphere_mask) {
-	       build_sphere_texgen(p, out_texgen, sphere_mask);
-	    }
-
-	    if (reflect_mask) {
-	       build_reflect_texgen(p, out_texgen, reflect_mask);
-	    }
-
-	    if (normal_mask) {
-	       struct ureg normal = get_eye_normal(p);
-	       emit_op1(p, OPCODE_MOV, out_texgen, normal_mask, normal );
-	    }
-
-	    if (copy_mask) {
-	       struct ureg in = register_input(p, VERT_ATTRIB_TEX0+i);
-	       emit_op1(p, OPCODE_MOV, out_texgen, copy_mask, in );
-	    }
-	 }
-
-	 if (texmat_enabled) {
-	    struct ureg texmat[4];
-	    struct ureg in = (!is_undef(out_texgen) ? 
-			      out_texgen : 
-			      register_input(p, VERT_ATTRIB_TEX0+i));
-	    if (PREFER_DP4) {
-	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
-				       0, texmat );
-	       emit_matrix_transform_vec4( p, out, texmat, in );
-	    }
-	    else {
-	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
-				       STATE_MATRIX_TRANSPOSE, texmat );
-	       emit_transpose_matrix_transform_vec4( p, out, texmat, in );
-	    }
-	 }
-
-	 release_temps(p);
-      } 
-      else {
-	 emit_passthrough(p, VERT_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
-      }
-   }
-}
-
-
-/* Seems like it could be tighter:
- */
-static void build_pointsize( struct tnl_program *p )
-{
-   struct ureg eye = get_eye_position(p);
-   struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
-   struct ureg state_attenuation = register_param1(p, STATE_POINT_ATTENUATION);
-   struct ureg out = register_output(p, VERT_RESULT_PSIZ);
-   struct ureg ut = get_temp(p);
-
-   /* 1, Z, Z * Z, 1 */      
-   emit_op1(p, OPCODE_MOV, ut, WRITEMASK_XW, swizzle1(get_identity_param(p), W));
-   emit_op1(p, OPCODE_ABS, ut, WRITEMASK_YZ, swizzle1(eye, Z));
-   emit_op2(p, OPCODE_MUL, ut, WRITEMASK_Z, ut, ut);
-
-
-   /* p1 +  p2 * dist + p3 * dist * dist, 0 */
-   emit_op2(p, OPCODE_DP3, ut, WRITEMASK_X, ut, state_attenuation);
-
-   /* 1 / sqrt(factor) */
-   emit_op1(p, OPCODE_RSQ, ut, WRITEMASK_X, ut ); 
-
-   /* ut = pointSize / factor */
-   emit_op2(p, OPCODE_MUL, ut, WRITEMASK_X, ut, state_size); 
-
-   /* Clamp to min/max - state_size.[yz]
-    */
-   emit_op2(p, OPCODE_MAX, ut, WRITEMASK_X, ut, swizzle1(state_size, Y)); 
-   emit_op2(p, OPCODE_MIN, out, 0, swizzle1(ut, X), swizzle1(state_size, Z)); 
-   
-   release_temp(p, ut);
-}
-
-static void build_tnl_program( struct tnl_program *p )
-{  
-   /* Emit the program, starting with modelviewproject:
-    */
-   build_hpos(p);
-
-   /* Lighting calculations:
-    */
-   if (p->state->fragprog_inputs_read & (FRAG_BIT_COL0|FRAG_BIT_COL1)) {
-      if (p->state->light_global_enabled)
-	 build_lighting(p);
-      else {
-	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-	    emit_passthrough(p, VERT_ATTRIB_COLOR0, VERT_RESULT_COL0);
-
-	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL1)
-	    emit_passthrough(p, VERT_ATTRIB_COLOR1, VERT_RESULT_COL1);
-      }
-   }
-
-   if ((p->state->fragprog_inputs_read & FRAG_BIT_FOGC) ||
-       p->state->fog_option != FOG_NONE)
-      build_fog(p);
-
-   if (p->state->fragprog_inputs_read & FRAG_BITS_TEX_ANY)
-      build_texture_transform(p);
-
-   if (p->state->point_attenuated)
-      build_pointsize(p);
-
-   /* Finish up:
-    */
-   emit_op1(p, OPCODE_END, undef, 0, undef);
-
-   /* Disassemble:
-    */
-   if (DISASSEM) {
-      _mesa_printf ("\n");
-   }
-}
-
-
-static void build_new_tnl_program( const struct state_key *key,
-				   struct gl_vertex_program *program,
-				   GLuint max_temps)
-{
-   struct tnl_program p;
-
-   _mesa_memset(&p, 0, sizeof(p));
-   p.state = key;
-   p.program = program;
-   p.eye_position = undef;
-   p.eye_position_normalized = undef;
-   p.eye_normal = undef;
-   p.identity = undef;
-   p.temp_in_use = 0;
-   p.nr_instructions = 16;
-   
-   if (max_temps >= sizeof(int) * 8)
-      p.temp_reserved = 0;
-   else
-      p.temp_reserved = ~((1<<max_temps)-1);
-
-   p.program->Base.Instructions = 
-      _mesa_malloc(sizeof(struct prog_instruction) * p.nr_instructions);
-   p.program->Base.String = 0;
-   p.program->Base.NumInstructions =
-   p.program->Base.NumTemporaries =
-   p.program->Base.NumParameters =
-   p.program->Base.NumAttributes = p.program->Base.NumAddressRegs = 0;
-   p.program->Base.Parameters = _mesa_new_parameter_list();
-   p.program->Base.InputsRead = 0;
-   p.program->Base.OutputsWritten = 0;
-
-   build_tnl_program( &p );
-}
-
-static void *search_cache( struct brw_tnl_cache *cache,
-			   GLuint hash,
-			   const void *key,
-			   GLuint keysize)
-{
-   struct brw_tnl_cache_item *c;
-
-   for (c = cache->items[hash % cache->size]; c; c = c->next) {
-      if (c->hash == hash && memcmp(c->key, key, keysize) == 0)
-	 return c->data;
-   }
-
-   return NULL;
-}
-
-static void rehash( struct brw_tnl_cache *cache )
-{
-   struct brw_tnl_cache_item **items;
-   struct brw_tnl_cache_item *c, *next;
-   GLuint size, i;
-
-   size = cache->size * 3;
-   items = (struct brw_tnl_cache_item**) _mesa_malloc(size * sizeof(*items));
-   _mesa_memset(items, 0, size * sizeof(*items));
-
-   for (i = 0; i < cache->size; i++)
-      for (c = cache->items[i]; c; c = next) {
-	 next = c->next;
-	 c->next = items[c->hash % size];
-	 items[c->hash % size] = c;
-      }
-
-   FREE(cache->items);
-   cache->items = items;
-   cache->size = size;
-}
-
-static void cache_item( struct brw_tnl_cache *cache,
-			GLuint hash,
-			const struct state_key *key,
-			void *data )
-{
-   struct brw_tnl_cache_item *c = MALLOC(sizeof(*c));
-   c->hash = hash;
-
-   c->key = malloc(sizeof(*key));
-   memcpy(c->key, key, sizeof(*key));
-
-   c->data = data;
-
-   if (++cache->n_items > cache->size * 1.5)
-      rehash(cache);
-
-   c->next = cache->items[hash % cache->size];
-   cache->items[hash % cache->size] = c;
-}
-
-
-static GLuint hash_key( struct state_key *key )
-{
-   GLuint *ikey = (GLuint *)key;
-   GLuint hash = 0, i;
-
-   /* I'm sure this can be improved on, but speed is important:
-    */
-   for (i = 0; i < sizeof(*key)/sizeof(GLuint); i++)
-      hash += ikey[i];
-
-   return hash;
-}
-
-static void update_tnl_program( struct brw_context *brw )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-   struct state_key key;
-   GLuint hash;
-   struct gl_vertex_program *old = brw->tnl_program;
-
-   /* _NEW_PROGRAM */
-   if (brw->attribs.VertexProgram->_Enabled) 
-      return;
-      
-   /* Grab all the relevent state and put it in a single structure:
-    */
-   make_state_key(ctx, &key);
-   hash = hash_key(&key);
-
-   /* Look for an already-prepared program for this state:
-    */
-   brw->tnl_program = (struct gl_vertex_program *)
-      search_cache( &brw->tnl_program_cache, hash, &key, sizeof(key) );
-   
-   /* OK, we'll have to build a new one:
-    */
-   if (!brw->tnl_program) {
-      brw->tnl_program = (struct gl_vertex_program *)
-	 ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); 
-
-      build_new_tnl_program( &key, brw->tnl_program, 
-/* 			     ctx->Const.MaxVertexProgramTemps  */
-			     32
-	 );
-
-      if (ctx->Driver.ProgramStringNotify)
-	 ctx->Driver.ProgramStringNotify( ctx, GL_VERTEX_PROGRAM_ARB, 
-					  &brw->tnl_program->Base );
-
-      cache_item( &brw->tnl_program_cache, 
-		  hash, &key, brw->tnl_program );
-   }
-
-   if (old != brw->tnl_program)
-      brw->state.dirty.brw |= BRW_NEW_TNL_PROGRAM;
-}
-
-/* Note: See brw_draw.c - the vertex program must not rely on
- * brw->primitive or brw->reduced_prim.
- */
-const struct brw_tracked_state brw_tnl_vertprog = {
-   .dirty = {
-      .mesa = (_NEW_PROGRAM | 
-	       _NEW_LIGHT | 
-	       _NEW_TRANSFORM | 
-	       _NEW_FOG | 
-	       _NEW_HINT | 
-	       _NEW_POINT | 
-	       _NEW_TEXTURE),
-      .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
-	      BRW_NEW_INPUT_VARYING),
-      .cache = 0
-   },
-   .update = update_tnl_program
-};
-
-
-
-
-static void update_active_vertprog( struct brw_context *brw )
+static void prepare_active_vertprog( struct brw_context *brw )
 {
    const struct gl_vertex_program *prev = brw->vertex_program;
 
-   /* NEW_PROGRAM */
-   if (brw->attribs.VertexProgram->_Enabled) {
-      brw->vertex_program = brw->attribs.VertexProgram->Current;
-   }
-   else {
-      /* BRW_NEW_TNL_PROGRAM */
-      brw->vertex_program = brw->tnl_program;
-   }
+   brw->vertex_program = brw->attribs.VertexProgram->_Current;
 
    if (brw->vertex_program != prev) 
       brw->state.dirty.brw |= BRW_NEW_VERTEX_PROGRAM;
@@ -1654,37 +52,8 @@ static void update_active_vertprog( struct brw_context *brw )
 const struct brw_tracked_state brw_active_vertprog = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
-      .brw = BRW_NEW_TNL_PROGRAM,
+      .brw = 0,
       .cache = 0
    },
-   .update = update_active_vertprog
+   .prepare = prepare_active_vertprog
 };
-
-
-void brw_ProgramCacheInit( GLcontext *ctx )
-{
-   struct brw_context *brw = brw_context(ctx);
-
-   brw->tnl_program_cache.size = 17;
-   brw->tnl_program_cache.n_items = 0;
-   brw->tnl_program_cache.items = (struct brw_tnl_cache_item **)
-      _mesa_calloc(brw->tnl_program_cache.size * 
-		   sizeof(struct brw_tnl_cache_item));
-}
-
-void brw_ProgramCacheDestroy( GLcontext *ctx )
-{
-   struct brw_context *brw = brw_context(ctx);
-   struct brw_tnl_cache_item *c, *next;
-   GLuint i;
-
-   for (i = 0; i < brw->tnl_program_cache.size; i++)
-      for (c = brw->tnl_program_cache.items[i]; c; c = next) {
-	 next = c->next;
-	 FREE(c->key);
-	 FREE(c->data);
-	 FREE(c);
-      }
-
-   FREE(brw->tnl_program_cache.items);
-}
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 786f30e641e..f7293ef4233 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -32,11 +32,11 @@
 
 
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "intel_batchbuffer.h" 
 #include "intel_regions.h" 
@@ -47,64 +47,117 @@
 
 #include "brw_draw.h"
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
+#include <stdarg.h>
 
-
+static void
+dri_bo_release(dri_bo **bo)
+{
+   dri_bo_unreference(*bo);
+   *bo = NULL;
+}
 
 /* called from intelDestroyContext()
  */
 static void brw_destroy_context( struct intel_context *intel )
 {
-   GLcontext *ctx = &intel->ctx;
    struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw_aub_destroy(brw);
+   int i;
 
    brw_destroy_metaops(brw);
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
 
-   brw_ProgramCacheDestroy( ctx );
    brw_FrameBufferTexDestroy( brw );
+
+   for (i = 0; i < brw->state.nr_draw_regions; i++)
+       intel_region_release(&brw->state.draw_regions[i]);
+   brw->state.nr_draw_regions = 0;
+   intel_region_release(&brw->state.depth_region);
+
+   dri_bo_release(&brw->curbe.curbe_bo);
+   dri_bo_release(&brw->vs.prog_bo);
+   dri_bo_release(&brw->vs.state_bo);
+   dri_bo_release(&brw->gs.prog_bo);
+   dri_bo_release(&brw->gs.state_bo);
+   dri_bo_release(&brw->clip.prog_bo);
+   dri_bo_release(&brw->clip.state_bo);
+   dri_bo_release(&brw->clip.vp_bo);
+   dri_bo_release(&brw->sf.prog_bo);
+   dri_bo_release(&brw->sf.state_bo);
+   dri_bo_release(&brw->sf.vp_bo);
+   for (i = 0; i < BRW_MAX_TEX_UNIT; i++)
+      dri_bo_release(&brw->wm.sdc_bo[i]);
+   dri_bo_release(&brw->wm.bind_bo);
+   for (i = 0; i < BRW_WM_MAX_SURF; i++)
+      dri_bo_release(&brw->wm.surf_bo[i]);
+   dri_bo_release(&brw->wm.prog_bo);
+   dri_bo_release(&brw->wm.state_bo);
+   dri_bo_release(&brw->cc.prog_bo);
+   dri_bo_release(&brw->cc.state_bo);
+   dri_bo_release(&brw->cc.vp_bo);
 }
 
 /* called from intelDrawBuffer()
  */
 static void brw_set_draw_region( struct intel_context *intel, 
-				  struct intel_region *draw_region,
-				  struct intel_region *depth_region)
+				  struct intel_region *draw_regions[],
+				  struct intel_region *depth_region,
+				GLuint num_regions)
 {
    struct brw_context *brw = brw_context(&intel->ctx);
-
-   intel_region_release(intel, &brw->state.draw_region);
-   intel_region_release(intel, &brw->state.depth_region);
-   intel_region_reference(&brw->state.draw_region, draw_region);
+   int i;
+   if (brw->state.depth_region != depth_region)
+      brw->state.dirty.brw |= BRW_NEW_DEPTH_BUFFER;
+   for (i = 0; i < brw->state.nr_draw_regions; i++)
+       intel_region_release(&brw->state.draw_regions[i]);
+   intel_region_release(&brw->state.depth_region);
+   for (i = 0; i < num_regions; i++)
+       intel_region_reference(&brw->state.draw_regions[i], draw_regions[i]);
    intel_region_reference(&brw->state.depth_region, depth_region);
+   brw->state.nr_draw_regions = num_regions;
 }
 
+/* called from intel_batchbuffer_flush and children before sending a
+ * batchbuffer off.
+ */
+static void brw_finish_batch(struct intel_context *intel)
+{
+   struct brw_context *brw = brw_context(&intel->ctx);
+
+   brw_emit_query_end(brw);
+}
 
 /* called from intelFlushBatchLocked
  */
-static void brw_lost_hardware( struct intel_context *intel )
+static void brw_new_batch( struct intel_context *intel )
 {
    struct brw_context *brw = brw_context(&intel->ctx);
 
-   /* Note that we effectively lose the context after this.
-    * 
-    * Setting this flag provokes a state buffer wrap and also flushes
-    * the hardware caches.
-    */
-   brw->state.dirty.brw |= BRW_NEW_CONTEXT;
+   /* Check that we didn't just wrap our batchbuffer at a bad time. */
+   assert(!brw->no_batch_wrap);
+
+   brw->curbe.need_new_bo = GL_TRUE;
 
-   /* Which means there shouldn't be any commands already queued:
+   /* Mark all context state as needing to be re-emitted.
+    * This is probably not as severe as on 915, since almost all of our state
+    * is just in referenced buffers.
     */
-   assert(intel->batch->ptr == intel->batch->map + intel->batch->offset);
+   brw->state.dirty.brw |= BRW_NEW_CONTEXT;
 
    brw->state.dirty.mesa |= ~0;
    brw->state.dirty.brw |= ~0;
    brw->state.dirty.cache |= ~0;
+
+   /* Move to the end of the current upload buffer so that we'll force choosing
+    * a new buffer next time.
+    */
+   if (brw->vb.upload.bo != NULL) {
+      dri_bo_unreference(brw->vb.upload.bo);
+      brw->vb.upload.bo = NULL;
+      brw->vb.upload.offset = 0;
+   }
 }
 
 static void brw_note_fence( struct intel_context *intel, 
@@ -115,12 +168,9 @@ static void brw_note_fence( struct intel_context *intel,
  
 static void brw_note_unlock( struct intel_context *intel )
 {
-  struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw_pool_check_wrap(brw, &brw->pool[BRW_GS_POOL]);
-   brw_pool_check_wrap(brw, &brw->pool[BRW_SS_POOL]);
+   struct brw_context *brw = brw_context(&intel->ctx);
 
-   brw_context(&intel->ctx)->state.dirty.brw |= BRW_NEW_LOCK;
+   brw_state_cache_check_size(brw);
 }
 
 
@@ -156,9 +206,6 @@ static GLuint brw_flush_cmd( void )
    return *(GLuint *)&flush;
 }
 
-
-
-
 static void brw_invalidate_state( struct intel_context *intel, GLuint new_state )
 {
    /* nothing */
@@ -176,10 +223,12 @@ void brwInitVtbl( struct brw_context *brw )
    brw->intel.vtbl.invalidate_state = brw_invalidate_state; 
    brw->intel.vtbl.note_fence = brw_note_fence; 
    brw->intel.vtbl.note_unlock = brw_note_unlock; 
-   brw->intel.vtbl.lost_hardware = brw_lost_hardware;
+   brw->intel.vtbl.new_batch = brw_new_batch;
+   brw->intel.vtbl.finish_batch = brw_finish_batch;
    brw->intel.vtbl.destroy = brw_destroy_context;
    brw->intel.vtbl.set_draw_region = brw_set_draw_region;
    brw->intel.vtbl.flush_cmd = brw_flush_cmd;
    brw->intel.vtbl.emit_flush = brw_emit_flush;
+   brw->intel.vtbl.debug_batch = brw_debug_batch;
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 1497dc79687..bad76793af7 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -29,12 +29,11 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
              
-
+#include "main/texformat.h"
 #include "brw_context.h"
 #include "brw_util.h"
 #include "brw_wm.h"
 #include "brw_state.h"
-#include "brw_hal.h"
 
 
 GLuint brw_wm_nr_args( GLuint opcode )
@@ -66,7 +65,11 @@ GLuint brw_wm_nr_args( GLuint opcode )
    case OPCODE_POW:
    case OPCODE_SUB:
    case OPCODE_SGE:
+   case OPCODE_SGT:
+   case OPCODE_SLE:
    case OPCODE_SLT:
+   case OPCODE_SEQ:
+   case OPCODE_SNE:
    case OPCODE_ADD:
    case OPCODE_MAX:
    case OPCODE_MIN:
@@ -116,20 +119,6 @@ GLuint brw_wm_is_scalar_result( GLuint opcode )
 }
 
 
-static void brw_wm_pass_hal (struct brw_wm_compile *c)
-{
-   static void (*hal_wm_pass) (struct brw_wm_compile *c);
-   static GLboolean hal_tried;
-   
-   if (!hal_tried)
-   {
-      hal_wm_pass = brw_hal_find_symbol ("intel_hal_wm_pass");
-      hal_tried = 1;
-   }
-   if (hal_wm_pass)
-      (*hal_wm_pass) (c);
-}
-
 static void do_wm_prog( struct brw_context *brw,
 			struct brw_fragment_program *fp, 
 			struct brw_wm_prog_key *key)
@@ -150,60 +139,53 @@ static void do_wm_prog( struct brw_context *brw,
    c->fp = fp;
    c->env_param = brw->intel.ctx.FragmentProgram.Parameters;
 
-
-   /* Augment fragment program.  Add instructions for pre- and
-    * post-fragment-program tasks such as interpolation and fogging.
-    */
-   brw_wm_pass_fp(c);
-   
-   /* Translate to intermediate representation.  Build register usage
-    * chains.
-    */
-   brw_wm_pass0(c);
-
-   /* Dead code removal.
-    */
-   brw_wm_pass1(c);
-
-   /* Hal optimization
-    */
-   brw_wm_pass_hal (c);
-   
-   /* Register allocation.
-    */
-   c->grf_limit = BRW_WM_MAX_GRF/2;
-
-   /* This is where we start emitting gen4 code:
-    */
-   brw_init_compile(&c->func);    
-
-   brw_wm_pass2(c);
-
-   c->prog_data.total_grf = c->max_wm_grf;
-   if (c->last_scratch) {
-      c->prog_data.total_scratch =
-	 c->last_scratch + 0x40;
+    brw_init_compile(brw, &c->func);
+   if (brw_wm_is_glsl(&c->fp->program)) {
+       brw_wm_glsl_emit(brw, c);
    } else {
-      c->prog_data.total_scratch = 0;
+       /* Augment fragment program.  Add instructions for pre- and
+	* post-fragment-program tasks such as interpolation and fogging.
+	*/
+       brw_wm_pass_fp(c);
+
+       /* Translate to intermediate representation.  Build register usage
+	* chains.
+	*/
+       brw_wm_pass0(c);
+
+       /* Dead code removal.
+	*/
+       brw_wm_pass1(c);
+
+       /* Register allocation.
+	*/
+       c->grf_limit = BRW_WM_MAX_GRF/2;
+
+       brw_wm_pass2(c);
+
+       c->prog_data.total_grf = c->max_wm_grf;
+       if (c->last_scratch) {
+	   c->prog_data.total_scratch =
+	       c->last_scratch + 0x40;
+       } else {
+	   c->prog_data.total_scratch = 0;
+       }
+
+       /* Emit GEN4 code.
+	*/
+       brw_wm_emit(c);
    }
-
-   /* Emit GEN4 code.
-    */
-   brw_wm_emit(c);
-
    /* get the program
     */
    program = brw_get_program(&c->func, &program_size);
 
-   /*
-    */
-   brw->wm.prog_gs_offset = brw_upload_cache( &brw->cache[BRW_WM_PROG],
-					      &c->key,
-					      sizeof(c->key),
-					      program,
-					      program_size,
-					      &c->prog_data,
-					      &brw->wm.prog_data );
+   dri_bo_unreference(brw->wm.prog_bo);
+   brw->wm.prog_bo = brw_upload_cache( &brw->cache, BRW_WM_PROG,
+				       &c->key, sizeof(c->key),
+				       NULL, 0,
+				       program, program_size,
+				       &c->prog_data,
+				       &brw->wm.prog_data );
 }
 
 
@@ -243,16 +225,11 @@ static void brw_wm_populate_key( struct brw_context *brw,
       lookup |= IZ_STENCIL_TEST_ENABLE_BIT;
 
       if (brw->attribs.Stencil->WriteMask[0] ||
-	  (brw->attribs.Stencil->TestTwoSide && brw->attribs.Stencil->WriteMask[1]))
+	  (brw->attribs.Stencil->_TestTwoSide &&
+	   brw->attribs.Stencil->WriteMask[1]))
 	 lookup |= IZ_STENCIL_WRITE_ENABLE_BIT;
    }
 
-   /* XXX: when should this be disabled?
-    */
-   if (1)
-      lookup |= IZ_EARLY_DEPTH_TEST_BIT;
-
-   
    line_aa = AA_NEVER;
 
    /* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
@@ -285,7 +262,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
 
 
    /* BRW_NEW_WM_INPUT_DIMENSIONS */
-   key->projtex_mask = brw->wm.input_size_masks[4-1]; 
+   key->projtex_mask = brw->wm.input_size_masks[4-1] >> (FRAG_ATTRIB_TEX0 - FRAG_ATTRIB_WPOS); 
 
    /* _NEW_LIGHT */
    key->flat_shade = (brw->attribs.Light->ShadeModel == GL_FLAT);
@@ -296,17 +273,41 @@ static void brw_wm_populate_key( struct brw_context *brw,
       const struct gl_texture_object *t = unit->_Current;
 
       if (unit->_ReallyEnabled) {
-
-	 if (t->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB &&
-	     t->Image[0][t->BaseLevel]->_BaseFormat == GL_DEPTH_COMPONENT) {
-	    key->shadowtex_mask |= 1<<i;
-	 }
-
-	 if (t->Image[0][t->BaseLevel]->InternalFormat == GL_YCBCR_MESA)
+	 if (t->Image[0][t->BaseLevel]->InternalFormat == GL_YCBCR_MESA) {
 	    key->yuvtex_mask |= 1<<i;
+	    if (t->Image[0][t->BaseLevel]->TexFormat->MesaFormat == 
+		    MESA_FORMAT_YCBCR)
+		key->yuvtex_swap_mask |= 1<< i;
+	 }
       }
    }
-	  
+
+   /* Shadow */
+   key->shadowtex_mask = fp->program.Base.ShadowSamplers;
+
+   /* _NEW_BUFFERS */
+   /*
+    * Include the draw buffer origin and height so that we can calculate
+    * fragment position values relative to the bottom left of the drawable,
+    * from the incoming screen origin relative position we get as part of our
+    * payload.
+    *
+    * We could avoid recompiling by including this as a constant referenced by
+    * our program, but if we were to do that it would also be nice to handle
+    * getting that constant updated at batchbuffer submit time (when we
+    * hold the lock and know where the buffer really is) rather than at emit
+    * time when we don't hold the lock and are just guessing.  We could also
+    * just avoid using this as key data if the program doesn't use
+    * fragment.position.
+    *
+    * This pretty much becomes moot with DRI2 and redirected buffers anyway,
+    * as our origins will always be zero then.
+    */
+   if (brw->intel.driDrawable != NULL) {
+      key->origin_x = brw->intel.driDrawable->x;
+      key->origin_y = brw->intel.driDrawable->y;
+      key->drawable_height = brw->intel.driDrawable->h;
+   }
 
    /* Extra info:
     */
@@ -315,7 +316,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
 }
 
 
-static void brw_upload_wm_prog( struct brw_context *brw )
+static void brw_prepare_wm_prog(struct brw_context *brw)
 {
    struct brw_wm_prog_key key;
    struct brw_fragment_program *fp = (struct brw_fragment_program *)
@@ -325,13 +326,13 @@ static void brw_upload_wm_prog( struct brw_context *brw )
 
    /* Make an early check for the key.
     */
-   if (brw_search_cache(&brw->cache[BRW_WM_PROG], 
-			&key, sizeof(key),
-			&brw->wm.prog_data,
-			&brw->wm.prog_gs_offset))
-      return;
-
-   do_wm_prog(brw, fp, &key);
+   dri_bo_unreference(brw->wm.prog_bo);
+   brw->wm.prog_bo = brw_search_cache(&brw->cache, BRW_WM_PROG,
+				      &key, sizeof(key),
+				      NULL, 0,
+				      &brw->wm.prog_data);
+   if (brw->wm.prog_bo == NULL)
+      do_wm_prog(brw, fp, &key);
 }
 
 
@@ -345,12 +346,13 @@ const struct brw_tracked_state brw_wm_prog = {
 		_NEW_POLYGON |
 		_NEW_LINE |
 		_NEW_LIGHT |
+		_NEW_BUFFERS |
 		_NEW_TEXTURE),
       .brw   = (BRW_NEW_FRAGMENT_PROGRAM |
 		BRW_NEW_WM_INPUT_DIMENSIONS |
 		BRW_NEW_REDUCED_PRIMITIVE),
       .cache = 0
    },
-   .update = brw_upload_wm_prog
+   .prepare = brw_prepare_wm_prog
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index f5fddfdb68a..ded079695ee 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -34,9 +34,9 @@
 #define BRW_WM_H
 
 
+#include "shader/prog_instruction.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "prog_instruction.h"
 
 /* A big lookup table is used to figure out which and how many
  * additional regs will inserted before the main payload in the WM
@@ -49,8 +49,7 @@
 #define IZ_DEPTH_TEST_ENABLE_BIT    0x8
 #define IZ_STENCIL_WRITE_ENABLE_BIT 0x10
 #define IZ_STENCIL_TEST_ENABLE_BIT  0x20
-#define IZ_EARLY_DEPTH_TEST_BIT     0x40
-#define IZ_BIT_MAX                  0x80
+#define IZ_BIT_MAX                  0x40
 
 #define AA_NEVER     0
 #define AA_SOMETIMES 1
@@ -69,9 +68,12 @@ struct brw_wm_prog_key {
    GLuint runtime_check_aads_emit:1;
    
    GLuint yuvtex_mask:8;
-   GLuint pad1:24;
+   GLuint yuvtex_swap_mask:8;	/* UV swaped */
+   GLuint pad1:16;
 
    GLuint program_string_id:32;
+   GLuint origin_x, origin_y;
+   GLuint drawable_height;
 };
 
 
@@ -140,6 +142,8 @@ struct brw_wm_instruction {
    GLuint writemask:4;
    GLuint tex_unit:4;   /* texture unit for TEX, TXD, TXP instructions */
    GLuint tex_idx:3;    /* TEXTURE_1D,2D,3D,CUBE,RECT_INDEX source target */
+   GLuint eot:1;    	/* End of thread indicator for FB_WRITE*/
+   GLuint target:10;    /* target binding table index for FB_WRITE*/
 };
 
 
@@ -152,6 +156,7 @@ struct brw_wm_instruction {
 #define BRW_WM_MAX_PARAM 256
 #define BRW_WM_MAX_CONST 256
 #define BRW_WM_MAX_KILLS MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS
+#define BRW_WM_MAX_SUBROUTINE 16
 
 
 
@@ -194,6 +199,8 @@ struct brw_wm_compile {
    GLuint nr_fp_insns;
    GLuint fp_temp;
    GLuint fp_interp_emitted;
+   GLuint fp_fragcolor_emitted;
+   GLuint fp_deriv_emitted;
 
    struct prog_src_register pixel_xy;
    struct prog_src_register delta_xy;
@@ -231,6 +238,18 @@ struct brw_wm_compile {
    GLuint grf_limit;
    GLuint max_wm_grf;
    GLuint last_scratch;
+
+   struct {
+	GLboolean inited;
+	struct brw_reg reg;
+   } wm_regs[PROGRAM_PAYLOAD+1][256][4];
+   struct brw_reg stack;
+   struct brw_reg emit_mask_reg;
+   GLuint reg_index;
+   GLuint tmp_regs[BRW_WM_MAX_GRF];
+   GLuint tmp_index;
+   GLuint tmp_max;
+   GLuint subroutines[BRW_WM_MAX_SUBROUTINE];
 };
 
 
@@ -259,4 +278,6 @@ void brw_wm_lookup_iz( GLuint line_aa,
 		       GLuint lookup,
 		       struct brw_wm_prog_key *key );
 
+GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp);
+void brw_wm_glsl_emit(struct brw_context *brw, struct brw_wm_compile *c);
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index fd605159727..58c78c4b2c5 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -30,7 +30,7 @@
   */
                
 
-#include "macros.h"
+#include "main/macros.h"
 #include "brw_context.h"
 #include "brw_wm.h"
 
@@ -39,7 +39,7 @@
 /* Not quite sure how correct this is - need to understand horiz
  * vs. vertical strides a little better.
  */
-static __inline struct brw_reg sechalf( struct brw_reg reg )
+static INLINE struct brw_reg sechalf( struct brw_reg reg )
 {
    if (reg.vstride)
       reg.nr++;
@@ -122,26 +122,30 @@ static void emit_delta_xy(struct brw_compile *p,
    }
 }
 
-static void emit_wpos_xy(struct brw_compile *p,
-			   const struct brw_reg *dst,
-			   GLuint mask,
-			   const struct brw_reg *arg0)
+static void emit_wpos_xy(struct brw_wm_compile *c,
+			 const struct brw_reg *dst,
+			 GLuint mask,
+			 const struct brw_reg *arg0)
 {
-   /* Calc delta X,Y by subtracting origin in r1 from the pixel
-    * centers.
+   struct brw_compile *p = &c->func;
+
+   /* Calculate the pixel offset from window bottom left into destination
+    * X and Y channels.
     */
    if (mask & WRITEMASK_X) {
-      brw_MOV(p,
+      /* X' = X - origin */
+      brw_ADD(p,
 	      dst[0],
-	      retype(arg0[0], BRW_REGISTER_TYPE_UW));
+	      retype(arg0[0], BRW_REGISTER_TYPE_W),
+	      brw_imm_d(0 - c->key.origin_x));
    }
 
    if (mask & WRITEMASK_Y) {
-      /* TODO -- window_height - Y */
-      brw_MOV(p,
+      /* Y' = height - (Y - origin_y) = height + origin_y - Y */
+      brw_ADD(p,
 	      dst[1],
-	      negate(retype(arg0[1], BRW_REGISTER_TYPE_UW)));
-
+	      negate(retype(arg0[1], BRW_REGISTER_TYPE_W)),
+	      brw_imm_d(c->key.origin_y + c->key.drawable_height - 1));
    }
 }
 
@@ -219,6 +223,10 @@ static void emit_pinterp( struct brw_compile *p,
       if (mask & (1<<i)) {
 	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
 	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
+      }
+   }
+   for(i = 0; i < 4; i++ ) {
+      if (mask & (1<<i)) {
 	 brw_MUL(p, dst[i], dst[i], w[3]);
       }
    }
@@ -229,20 +237,20 @@ static void emit_cinterp( struct brw_compile *p,
 			 GLuint mask,
 			 const struct brw_reg *arg0 )
 {
-   struct brw_reg interp[4];
-   GLuint nr = arg0[0].nr;
-   GLuint i;
-
-   interp[0] = brw_vec1_grf(nr, 0);
-   interp[1] = brw_vec1_grf(nr, 4);
-   interp[2] = brw_vec1_grf(nr+1, 0);
-   interp[3] = brw_vec1_grf(nr+1, 4);
-
-   for(i = 0; i < 4; i++ ) {
-      if (mask & (1<<i)) {
-	 brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
-      }
-   }
+	struct brw_reg interp[4];
+	GLuint nr = arg0[0].nr;
+	GLuint i;
+
+	interp[0] = brw_vec1_grf(nr, 0);
+	interp[1] = brw_vec1_grf(nr, 4);
+	interp[2] = brw_vec1_grf(nr+1, 0);
+	interp[3] = brw_vec1_grf(nr+1, 4);
+
+	for(i = 0; i < 4; i++ ) {
+		if (mask & (1<<i)) {
+			brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
+		}
+	}
 }
 
 
@@ -343,11 +351,10 @@ static void emit_lrp( struct brw_compile *p,
       }
    }
 }
-
-
-static void emit_slt( struct brw_compile *p, 
+static void emit_sop( struct brw_compile *p, 
 		      const struct brw_reg *dst,
 		      GLuint mask,
+		      GLuint cond,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
@@ -356,34 +363,66 @@ static void emit_slt( struct brw_compile *p,
    for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {	
 	 brw_MOV(p, dst[i], brw_imm_f(0));
-	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], arg1[i]);
+	 brw_CMP(p, brw_null_reg(), cond, arg0[i], arg1[i]);
 	 brw_MOV(p, dst[i], brw_imm_f(1.0));
 	 brw_set_predicate_control_flag_value(p, 0xff);
       }
    }
 }
 
-/* Isn't this just the same as the above with the args swapped?
- */
-static void emit_sge( struct brw_compile *p, 
+static void emit_slt( struct brw_compile *p, 
 		      const struct brw_reg *dst,
 		      GLuint mask,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-   GLuint i;
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
+}
 
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {	
-	 brw_MOV(p, dst[i], brw_imm_f(0));
-	 brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0[i], arg1[i]);
-	 brw_MOV(p, dst[i], brw_imm_f(1.0));
-	 brw_set_predicate_control_flag_value(p, 0xff);
-      }
-   }
+static void emit_sle( struct brw_compile *p, 
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1 )
+{
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
 }
 
+static void emit_sgt( struct brw_compile *p, 
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1 )
+{
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
+}
 
+static void emit_sge( struct brw_compile *p, 
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1 )
+{
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
+}
+
+static void emit_seq( struct brw_compile *p, 
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1 )
+{
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
+}
+
+static void emit_sne( struct brw_compile *p, 
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1 )
+{
+	 emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
+}
 
 static void emit_cmp( struct brw_compile *p, 
 		      const struct brw_reg *dst,
@@ -465,6 +504,9 @@ static void emit_dp3( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code*/
+
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
    brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
@@ -482,6 +524,9 @@ static void emit_dp4( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code*/
+
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
    brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
@@ -500,6 +545,9 @@ static void emit_dph( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code*/
+
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
    brw_MUL(p, brw_null_reg(), arg0[0], arg1[0]);
@@ -543,8 +591,11 @@ static void emit_math1( struct brw_compile *p,
 			GLuint mask,
 			const struct brw_reg *arg0 )
 {
-   assert((mask & WRITEMASK_XYZW) == WRITEMASK_X ||
-	  function == BRW_MATH_FUNCTION_SINCOS);
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code*/
+
+   //assert((mask & WRITEMASK_XYZW) == WRITEMASK_X ||
+   //	  function == BRW_MATH_FUNCTION_SINCOS);
    
    brw_MOV(p, brw_message_reg(2), arg0[0]);
 
@@ -567,6 +618,9 @@ static void emit_math2( struct brw_compile *p,
 			const struct brw_reg *arg0,
 			const struct brw_reg *arg1)
 {
+   if (!(mask & WRITEMASK_XYZW))
+      return; /* Do not emit dead code*/
+
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
    brw_push_insn_state(p);
@@ -661,7 +715,7 @@ static void emit_tex( struct brw_wm_compile *c,
 	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
 	      1,
 	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
-	      inst->tex_unit + 1, /* surface */
+	      inst->tex_unit + MAX_DRAW_BUFFERS, /* surface */
 	      inst->tex_unit,	  /* sampler */
 	      inst->writemask,
 	      (shadow ? 
@@ -670,7 +724,6 @@ static void emit_tex( struct brw_wm_compile *c,
 	      responseLength,
 	      msgLength,
 	      0);	
-
 }
 
 
@@ -712,7 +765,7 @@ static void emit_txb( struct brw_wm_compile *c,
 	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
 	      1,
 	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
-	      inst->tex_unit + 1, /* surface */
+	      inst->tex_unit + MAX_DRAW_BUFFERS, /* surface */
 	      inst->tex_unit,	  /* sampler */
 	      inst->writemask,
 	      BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS,
@@ -777,6 +830,7 @@ static void emit_kil( struct brw_wm_compile *c,
       brw_push_insn_state(p);
       brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0[i], brw_imm_f(0));   
       brw_set_predicate_control_flag_value(p, 0xff);
+      brw_set_compression_control(p, BRW_COMPRESSION_NONE);
       brw_AND(p, r0uw, brw_flag_reg(), r0uw);
       brw_pop_insn_state(p);
    }
@@ -784,7 +838,9 @@ static void emit_kil( struct brw_wm_compile *c,
 
 static void fire_fb_write( struct brw_wm_compile *c,
 			   GLuint base_reg,
-			   GLuint nr )
+			   GLuint nr,
+			   GLuint target,
+			   GLuint eot )
 {
    struct brw_compile *p = &c->func;
    
@@ -807,10 +863,10 @@ static void fire_fb_write( struct brw_wm_compile *c,
 		retype(vec16(brw_null_reg()), BRW_REGISTER_TYPE_UW),
 		base_reg,
 		retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW),
-		0,		/* render surface always 0 */
+		target,		
 		nr,
 		0, 
-		1);
+		eot);
 }
 
 static void emit_aa( struct brw_wm_compile *c,
@@ -835,7 +891,9 @@ static void emit_aa( struct brw_wm_compile *c,
 static void emit_fb_write( struct brw_wm_compile *c,
 			   struct brw_reg *arg0,
 			   struct brw_reg *arg1,
-			   struct brw_reg *arg2)
+			   struct brw_reg *arg2,
+			   GLuint target,
+			   GLuint eot)
 {
    struct brw_compile *p = &c->func;
    GLuint nr = 2;
@@ -890,15 +948,16 @@ static void emit_fb_write( struct brw_wm_compile *c,
       GLuint off = c->key.dest_depth_reg % 2;
 
       if (off != 0) {
-	 brw_push_insn_state(p);
-	 brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-	 brw_MOV(p, brw_message_reg(nr), arg1[comp]);
-	 /* 2nd half? */
-	 brw_MOV(p, brw_message_reg(nr+1), offset(arg1[comp],1));
-	 brw_pop_insn_state(p);
+         brw_push_insn_state(p);
+         brw_set_compression_control(p, BRW_COMPRESSION_NONE);
+
+         brw_MOV(p, brw_message_reg(nr), offset(arg1[comp],1));
+         /* 2nd half? */
+         brw_MOV(p, brw_message_reg(nr+1), arg1[comp+1]);
+         brw_pop_insn_state(p);
       }
       else {
-	 brw_MOV(p, brw_message_reg(nr), arg1[comp]);
+         brw_MOV(p, brw_message_reg(nr), arg1[comp]);
       }
       nr += 2;
    }
@@ -908,7 +967,7 @@ static void emit_fb_write( struct brw_wm_compile *c,
       if (c->key.aa_dest_stencil_reg)
 	 emit_aa(c, arg1, 2);
 
-      fire_fb_write(c, 0, nr);
+      fire_fb_write(c, 0, nr, target, eot);
    }
    else {
       struct brw_reg v1_null_ud = vec1(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD));
@@ -925,14 +984,14 @@ static void emit_fb_write( struct brw_wm_compile *c,
       jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
       {
 	 emit_aa(c, arg1, 2);
-	 fire_fb_write(c, 0, nr);
+	 fire_fb_write(c, 0, nr, target, eot);
 	 /* note - thread killed in subroutine */
       }
       brw_land_fwd_jump(p, jmp);
 
       /* ELSE: Shuffle up one register to fill in the hole left for AA:
        */
-      fire_fb_write(c, 1, nr-1);
+      fire_fb_write(c, 1, nr-1, target, eot);
    }
 }
 
@@ -1080,7 +1139,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 break;
 
       case WM_WPOSXY:
-	 emit_wpos_xy(p, dst, dst_flags, args[0]);
+	 emit_wpos_xy(c, dst, dst_flags, args[0]);
 	 break;
 
       case WM_PIXELW:
@@ -1100,7 +1159,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 break;
 
       case WM_FB_WRITE:
-	 emit_fb_write(c, args[0], args[1], args[2]);
+	 emit_fb_write(c, args[0], args[1], args[2], inst->target, inst->eot);
 	 break;
 
 	 /* Straightforward arithmetic:
@@ -1208,9 +1267,21 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_slt(p, dst, dst_flags, args[0], args[1]);
 	 break;
 
+      case OPCODE_SLE:
+	 emit_sle(p, dst, dst_flags, args[0], args[1]);
+	break;
+      case OPCODE_SGT:
+	 emit_sgt(p, dst, dst_flags, args[0], args[1]);
+	break;
       case OPCODE_SGE:
 	 emit_sge(p, dst, dst_flags, args[0], args[1]);
 	 break;
+      case OPCODE_SEQ:
+	 emit_seq(p, dst, dst_flags, args[0], args[1]);
+	break;
+      case OPCODE_SNE:
+	 emit_sne(p, dst, dst_flags, args[0], args[1]);
+	break;
 
       case OPCODE_LIT:
 	 emit_lit(p, dst, dst_flags, args[0]);
@@ -1231,7 +1302,10 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 break;
 
       default:
-	 assert(0);
+	 _mesa_printf("Unsupported opcode %i (%s) in fragment shader\n",
+		      inst->opcode, inst->opcode < MAX_OPCODE ?
+				    _mesa_opcode_string(inst->opcode) :
+				    "unknown");
       }
       
       for (i = 0; i < 4; i++)
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index ff97d87dc45..7f7b957cbe8 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -30,9 +30,9 @@
   */
                
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "brw_context.h"
 #include "brw_wm.h"
 #include "brw_util.h"
@@ -144,7 +144,7 @@ static struct prog_dst_register dst_undef( void )
 
 static struct prog_dst_register get_temp( struct brw_wm_compile *c )
 {
-   int bit = ffs( ~c->fp_temp );
+   int bit = _mesa_ffs( ~c->fp_temp );
 
    if (!bit) {
       _mesa_printf("%s: out of temporaries\n", __FILE__);
@@ -158,7 +158,7 @@ static struct prog_dst_register get_temp( struct brw_wm_compile *c )
 
 static void release_temp( struct brw_wm_compile *c, struct prog_dst_register temp )
 {
-   c->fp_temp &= ~1<<(temp.Index + 1 - FIRST_INTERNAL_TEMP);
+   c->fp_temp &= ~(1 << (temp.Index - FIRST_INTERNAL_TEMP));
 }
 
 
@@ -176,6 +176,7 @@ static struct prog_instruction *emit_insn(struct brw_wm_compile *c,
 {
    struct prog_instruction *inst = get_fp_inst(c);
    *inst = *inst0;
+   inst->Data = (void *)inst0;
    return inst;
 }
 
@@ -201,7 +202,6 @@ static struct prog_instruction * emit_op(struct brw_wm_compile *c,
    inst->SrcReg[0] = src0;
    inst->SrcReg[1] = src1;
    inst->SrcReg[2] = src2;
-   
    return inst;
 }
    
@@ -361,6 +361,37 @@ static void emit_interp( struct brw_wm_compile *c,
    c->fp_interp_emitted |= 1<<idx;
 }
 
+static void emit_ddx( struct brw_wm_compile *c,
+        const struct prog_instruction *inst )
+{
+    GLuint idx = inst->SrcReg[0].Index;
+    struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
+
+    c->fp_deriv_emitted |= 1<<idx;
+    emit_op(c,
+            OPCODE_DDX,
+            inst->DstReg,
+            0, 0, 0,
+            interp,
+            get_pixel_w(c),
+            src_undef());
+}
+
+static void emit_ddy( struct brw_wm_compile *c,
+        const struct prog_instruction *inst )
+{
+    GLuint idx = inst->SrcReg[0].Index;
+    struct prog_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
+
+    c->fp_deriv_emitted |= 1<<idx;
+    emit_op(c,
+            OPCODE_DDY,
+            inst->DstReg,
+            0, 0, 0,
+            interp,
+            get_pixel_w(c),
+            src_undef());
+}
 
 /***********************************************************************
  * Hacks to extend the program parameter and constant lists.
@@ -395,10 +426,6 @@ static struct prog_src_register search_or_add_param5(struct brw_wm_compile *c,
 
    idx = _mesa_add_state_reference( paramList, tokens );
 
-   /* Recalculate state dependency: 
-    */
-   c->fp->param_state = paramList->StateFlags;
-
    return src_reg(PROGRAM_STATE_VAR, idx);
 }
 
@@ -433,7 +460,7 @@ static struct prog_src_register search_or_add_const4f( struct brw_wm_compile *c,
    }
    
    idx = _mesa_add_unnamed_constant( paramList, values, 4, &swizzle );
-   /* XXX what about swizzle? */
+   assert(swizzle == SWIZZLE_NOOP); /* Need to handle swizzle in reg setup */
    return src_reg(PROGRAM_STATE_VAR, idx);
 }
 
@@ -463,17 +490,20 @@ static void precalc_dst( struct brw_wm_compile *c,
 
 
    if (dst.WriteMask & WRITEMASK_XZ) {
+      struct prog_instruction *swz;
       GLuint z = GET_SWZ(src0.Swizzle, Z);
 
       /* dst.xz = swz src0.1zzz
        */
-      emit_op(c,
-	      OPCODE_SWZ,
-	      dst_mask(dst, WRITEMASK_XZ),
-	      inst->SaturateMode, 0, 0,
-	      src_swizzle(src0, SWIZZLE_ONE, z, z, z),
-	      src_undef(),
-	      src_undef());
+      swz = emit_op(c,
+		    OPCODE_SWZ,
+		    dst_mask(dst, WRITEMASK_XZ),
+		    inst->SaturateMode, 0, 0,
+		    src_swizzle(src0, SWIZZLE_ONE, z, z, z),
+		    src_undef(),
+		    src_undef());
+      /* Avoid letting negation flag of src0 affect our 1 constant. */
+      swz->SrcReg[0].NegateBase &= ~NEGATE_X;
    }
    if (dst.WriteMask & WRITEMASK_W) {
       /* dst.w = mov src1.w
@@ -496,15 +526,19 @@ static void precalc_lit( struct brw_wm_compile *c,
    struct prog_dst_register dst = inst->DstReg;
    
    if (dst.WriteMask & WRITEMASK_XW) {
+      struct prog_instruction *swz;
+
       /* dst.xw = swz src0.1111
        */
-      emit_op(c,
-	      OPCODE_SWZ,
-	      dst_mask(dst, WRITEMASK_XW),
-	      0, 0, 0,
-	      src_swizzle1(src0, SWIZZLE_ONE),
-	      src_undef(),
-	      src_undef());
+      swz = emit_op(c,
+		    OPCODE_SWZ,
+		    dst_mask(dst, WRITEMASK_XW),
+		    0, 0, 0,
+		    src_swizzle1(src0, SWIZZLE_ONE),
+		    src_undef(),
+		    src_undef());
+      /* Avoid letting the negation flag of src0 affect our 1 constant. */
+      swz->SrcReg[0].NegateBase = 0;
    }
 
 
@@ -524,13 +558,64 @@ static void precalc_tex( struct brw_wm_compile *c,
 {
    struct prog_src_register coord;
    struct prog_dst_register tmpcoord;
-
-   if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
+   GLuint unit = c->fp->program.Base.SamplerUnits[inst->TexSrcUnit];
+
+   if (inst->TexSrcTarget == TEXTURE_CUBE_INDEX) {
+       struct prog_instruction *out;
+       struct prog_dst_register tmp0 = get_temp(c);
+       struct prog_src_register tmp0src = src_reg_from_dst(tmp0);
+       struct prog_dst_register tmp1 = get_temp(c);
+       struct prog_src_register tmp1src = src_reg_from_dst(tmp1);
+       struct prog_src_register src0 = inst->SrcReg[0];
+
+       tmpcoord = get_temp(c);
+       coord = src_reg_from_dst(tmpcoord);
+
+       out = emit_op(c, OPCODE_MOV,
+                     tmpcoord,
+                     0, 0, 0,
+                     src0,
+                     src_undef(),
+                     src_undef());
+       out->SrcReg[0].NegateBase = 0;
+       out->SrcReg[0].Abs = 1;
+
+       emit_op(c, OPCODE_MAX,
+               tmp0,
+               0, 0, 0,
+               src_swizzle1(coord, X),
+               src_swizzle1(coord, Y),
+               src_undef());
+
+       emit_op(c, OPCODE_MAX,
+               tmp1,
+               0, 0, 0,
+               tmp0src,
+               src_swizzle1(coord, Z),
+               src_undef());
+
+       emit_op(c, OPCODE_RCP,
+               tmp0,
+               0, 0, 0,
+               tmp1src,
+               src_undef(),
+               src_undef());
+
+       emit_op(c, OPCODE_MUL,
+               tmpcoord,
+               0, 0, 0,
+               src0,
+               tmp0src,
+               src_undef());
+
+       release_temp(c, tmp0);
+       release_temp(c, tmp1);
+   } else if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
       struct prog_src_register scale = 
 	 search_or_add_param5( c, 
 			       STATE_INTERNAL, 
 			       STATE_TEXRECT_SCALE,
-			       inst->TexSrcUnit,
+			       unit,
 			       0,0 );
 
       tmpcoord = get_temp(c);
@@ -556,29 +641,33 @@ static void precalc_tex( struct brw_wm_compile *c,
     * conversion requires allocating a temporary variable which we
     * don't have the facility to do that late in the compilation.
     */
-   if (!(c->key.yuvtex_mask & (1<<inst->TexSrcUnit))) {
+   if (!(c->key.yuvtex_mask & (1<<unit))) {
       emit_op(c, 
 	      OPCODE_TEX,
 	      inst->DstReg,
 	      inst->SaturateMode,
-	      inst->TexSrcUnit,
+	      unit,
 	      inst->TexSrcTarget,
 	      coord,
 	      src_undef(),
 	      src_undef());
    }
    else {
+       GLboolean  swap_uv = c->key.yuvtex_swap_mask & (1<<unit);
+
       /* 
 	 CONST C0 = { -.5, -.0625,  -.5, 1.164 }
 	 CONST C1 = { 1.596, -0.813, 2.018, -.391 }
 	 UYV     = TEX ...
 	 UYV.xyz = ADD UYV,     C0
 	 UYV.y   = MUL UYV.y,   C0.w
-	 RGB.xyz = MAD UYV.xxz, C1,   UYV.y
+ 	 if (UV swaped)
+	    RGB.xyz = MAD UYV.zzx, C1,   UYV.y
+	 else
+	    RGB.xyz = MAD UYV.xxz, C1,   UYV.y 
 	 RGB.y   = MAD UYV.z,   C1.w, RGB.y
       */
       struct prog_dst_register dst = inst->DstReg;
-      struct prog_src_register src0 = inst->SrcReg[0];
       struct prog_dst_register tmp = get_temp(c);
       struct prog_src_register tmpsrc = src_reg_from_dst(tmp);
       struct prog_src_register C0 = search_or_add_const4f( c,  -.5, -.0625, -.5, 1.164 );
@@ -590,9 +679,9 @@ static void precalc_tex( struct brw_wm_compile *c,
 	      OPCODE_TEX,
 	      tmp,
 	      inst->SaturateMode,
-	      inst->TexSrcUnit,
+	      unit,
 	      inst->TexSrcTarget,
-	      src0,
+	      coord,
 	      src_undef(),
 	      src_undef());
 
@@ -608,6 +697,7 @@ static void precalc_tex( struct brw_wm_compile *c,
 
       /* YUV.y   = MUL YUV.y, C0.w
        */
+
       emit_op(c,
 	      OPCODE_MUL,
 	      dst_mask(tmp, WRITEMASK_Y),
@@ -616,13 +706,18 @@ static void precalc_tex( struct brw_wm_compile *c,
 	      src_swizzle1(C0, W),
 	      src_undef());
 
-      /* RGB.xyz = MAD YUV.xxz, C1, YUV.y
+      /* 
+       * if (UV swaped)
+       *     RGB.xyz = MAD YUV.zzx, C1, YUV.y
+       * else
+       *     RGB.xyz = MAD YUV.xxz, C1, YUV.y
        */
+
       emit_op(c,
 	      OPCODE_MAD,
 	      dst_mask(dst, WRITEMASK_XYZ),
 	      0, 0, 0,
-	      src_swizzle(tmpsrc, X,X,Z,Z),
+	      swap_uv?src_swizzle(tmpsrc, Z,Z,X,X):src_swizzle(tmpsrc, X,X,Z,Z),
 	      C1,
 	      src_swizzle1(tmpsrc, Y));
 
@@ -639,7 +734,8 @@ static void precalc_tex( struct brw_wm_compile *c,
       release_temp(c, tmp);
    }
 
-   if (inst->TexSrcTarget == GL_TEXTURE_RECTANGLE_NV) 
+   if ((inst->TexSrcTarget == TEXTURE_RECT_INDEX) ||
+       (inst->TexSrcTarget == TEXTURE_CUBE_INDEX))
       release_temp(c, tmpcoord);
 }
 
@@ -660,7 +756,7 @@ static GLboolean projtex( struct brw_wm_compile *c,
       return 0;  /* ut2004 gun rendering !?! */
    else if (src.File == PROGRAM_INPUT && 
 	    GET_SWZ(src.Swizzle, W) == W &&
-	    (c->key.projtex_mask & (1<<src.Index)) == 0)
+           (c->key.projtex_mask & (1<<(src.Index + FRAG_ATTRIB_WPOS - FRAG_ATTRIB_TEX0))) == 0)
       return 0;
    else
       return 1;
@@ -770,14 +866,34 @@ static void emit_fb_write( struct brw_wm_compile *c )
    struct prog_src_register outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
    struct prog_src_register payload_r0_depth = src_reg(PROGRAM_PAYLOAD, PAYLOAD_DEPTH);
    struct prog_src_register outdepth = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DEPR);
+   GLuint i;
 
-   emit_op(c,
-	   WM_FB_WRITE,
-	   dst_mask(dst_undef(),0),
-	   0, 0, 0,
-	   outcolor,
-	   payload_r0_depth,
-	   outdepth);
+   struct prog_instruction *inst, *last_inst;
+   struct brw_context *brw = c->func.brw;
+
+   /* inst->Sampler is not used by backend, 
+      use it for fb write target and eot */
+
+   if (brw->state.nr_draw_regions > 1) {
+       for (i = 0 ; i < brw->state.nr_draw_regions; i++) {
+	   outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0 + i);
+	   last_inst = inst = emit_op(c,
+		   WM_FB_WRITE, dst_mask(dst_undef(),0), 0, 0, 0,
+		   outcolor, payload_r0_depth, outdepth);
+	   inst->Sampler = (i<<1);
+	   if (c->fp_fragcolor_emitted) {
+	       outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
+	       last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
+		       0, 0, 0, outcolor, payload_r0_depth, outdepth);
+	       inst->Sampler = (i<<1);
+	   }
+       }
+       last_inst->Sampler |= 1; //eot
+   }else {
+       inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
+	       0, 0, 0, outcolor, payload_r0_depth, outdepth);
+       inst->Sampler = 1|(0<<1);
+   }
 }
 
 
@@ -803,7 +919,15 @@ static void validate_src_regs( struct brw_wm_compile *c,
    }
 }
 	 
-
+static void validate_dst_regs( struct brw_wm_compile *c,
+			       const struct prog_instruction *inst )
+{
+   if (inst->DstReg.File == PROGRAM_OUTPUT) {
+       GLuint idx = inst->DstReg.Index;
+       if (idx == FRAG_RESULT_COLR)
+	   c->fp_fragcolor_emitted = 1;
+   }
+}
 
 static void print_insns( const struct prog_instruction *insn,
 			 GLuint nr )
@@ -848,12 +972,16 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
 
    for (insn = 0; insn < fp->program.Base.NumInstructions; insn++) {
       const struct prog_instruction *inst = &fp->program.Base.Instructions[insn];
+      validate_src_regs(c, inst);
+      validate_dst_regs(c, inst);
+   }
+   for (insn = 0; insn < fp->program.Base.NumInstructions; insn++) {
+      const struct prog_instruction *inst = &fp->program.Base.Instructions[insn];
       struct prog_instruction *out;
 
       /* Check for INPUT values, emit INTERP instructions where
        * necessary:
        */
-      validate_src_regs(c, inst);
 
 
       switch (inst->Opcode) {
@@ -889,11 +1017,20 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
       case OPCODE_LIT:
 	 precalc_lit(c, inst);
 	 break;
-     
+
+      case OPCODE_TEX:
+	 precalc_tex(c, inst);
+	 break;
+
       case OPCODE_TXP:
 	 precalc_txp(c, inst);
 	 break;
 
+      case OPCODE_TXB:
+	 out = emit_insn(c, inst);
+	 out->TexSrcUnit = fp->program.Base.SamplerUnits[inst->TexSrcUnit];
+	 break;
+
       case OPCODE_XPD: 
 	 out = emit_insn(c, inst);
 	 /* This should probably be done in the parser. 
@@ -907,8 +1044,16 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
 	  */
 	 out->DstReg.WriteMask = 0;
 	 break;
-
+      case OPCODE_DDX:
+	 emit_ddx(c, inst);
+	 break;
+      case OPCODE_DDY:
+         emit_ddy(c, inst);
+	break;
       case OPCODE_END:
+	 emit_fog(c);
+	 emit_fb_write(c);
+	 break;
       case OPCODE_PRINT:
 	 break;
 	 
@@ -917,15 +1062,11 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
 	 break;
       }
    }
-   
-   emit_fog(c);
-   emit_fb_write(c);
-
 
    if (INTEL_DEBUG & DEBUG_WM) {
-      _mesa_printf("\n\n\npass_fp:\n");
-      print_insns( c->prog_instructions, c->nr_fp_insns );
-      _mesa_printf("\n");
+	   _mesa_printf("\n\n\npass_fp:\n");
+	   print_insns( c->prog_instructions, c->nr_fp_insns );
+	   _mesa_printf("\n");
    }
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
new file mode 100644
index 00000000000..baecfdcb799
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -0,0 +1,2519 @@
+#include "main/macros.h"
+#include "shader/prog_parameter.h"
+#include "brw_context.h"
+#include "brw_eu.h"
+#include "brw_wm.h"
+
+enum _subroutine {
+    SUB_NOISE1, SUB_NOISE2, SUB_NOISE3, SUB_NOISE4
+};
+
+/* Only guess, need a flag in gl_fragment_program later */
+GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp)
+{
+    int i;
+    for (i = 0; i < fp->Base.NumInstructions; i++) {
+	struct prog_instruction *inst = &fp->Base.Instructions[i];
+	switch (inst->Opcode) {
+	    case OPCODE_IF:
+	    case OPCODE_TRUNC:
+	    case OPCODE_ENDIF:
+	    case OPCODE_CAL:
+	    case OPCODE_BRK:
+	    case OPCODE_RET:
+	    case OPCODE_DDX:
+	    case OPCODE_DDY:
+	    case OPCODE_NOISE1:
+	    case OPCODE_NOISE2:
+	    case OPCODE_NOISE3:
+	    case OPCODE_NOISE4:
+	    case OPCODE_BGNLOOP:
+		return GL_TRUE; 
+	    default:
+		break;
+	}
+    }
+    return GL_FALSE; 
+}
+
+static void set_reg(struct brw_wm_compile *c, int file, int index, 
+	int component, struct brw_reg reg)
+{
+    c->wm_regs[file][index][component].reg = reg;
+    c->wm_regs[file][index][component].inited = GL_TRUE;
+}
+
+static int get_scalar_dst_index(struct prog_instruction *inst)
+{
+    int i;
+    for (i = 0; i < 4; i++)
+	if (inst->DstReg.WriteMask & (1<<i))
+	    break;
+    return i;
+}
+
+static struct brw_reg alloc_tmp(struct brw_wm_compile *c)
+{
+    struct brw_reg reg;
+    if(c->tmp_index == c->tmp_max)
+	c->tmp_regs[ c->tmp_max++ ] = c->reg_index++;
+    
+    reg = brw_vec8_grf(c->tmp_regs[ c->tmp_index++ ], 0);
+    return reg;
+}
+
+static int mark_tmps(struct brw_wm_compile *c)
+{
+    return c->tmp_index;
+}
+
+static struct brw_reg lookup_tmp( struct brw_wm_compile *c, int index )
+{
+    return brw_vec8_grf( c->tmp_regs[ index ], 0 );
+}
+
+static void release_tmps(struct brw_wm_compile *c, int mark)
+{
+    c->tmp_index = mark;
+}
+
+static struct brw_reg 
+get_reg(struct brw_wm_compile *c, int file, int index, int component, int nr, GLuint neg, GLuint abs)
+{
+    struct brw_reg reg;
+    switch (file) {
+	case PROGRAM_STATE_VAR:
+	case PROGRAM_CONSTANT:
+	case PROGRAM_UNIFORM:
+	    file = PROGRAM_STATE_VAR;
+	    break;
+	case PROGRAM_UNDEFINED:
+	    return brw_null_reg();	
+	default:
+	    break;
+    }
+
+    if(c->wm_regs[file][index][component].inited)
+	reg = c->wm_regs[file][index][component].reg;
+    else 
+	reg = brw_vec8_grf(c->reg_index, 0);
+
+    if(!c->wm_regs[file][index][component].inited) {
+	set_reg(c, file, index, component, reg);
+	c->reg_index++;
+    }
+
+    if (neg & (1<< component)) {
+	reg = negate(reg);
+    }
+    if (abs)
+	reg = brw_abs(reg);
+    return reg;
+}
+
+static void prealloc_reg(struct brw_wm_compile *c)
+{
+    int i, j;
+    struct brw_reg reg;
+    int nr_interp_regs = 0;
+    GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted | c->fp_deriv_emitted;
+
+    for (i = 0; i < 4; i++) {
+	reg = (i < c->key.nr_depth_regs) 
+	    ? brw_vec8_grf(i*2, 0) : brw_vec8_grf(0, 0);
+	set_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH, i, reg);
+    }
+    c->reg_index += 2*c->key.nr_depth_regs;
+    {
+	int nr_params = c->fp->program.Base.Parameters->NumParameters;
+	struct gl_program_parameter_list *plist = 
+	    c->fp->program.Base.Parameters;
+	int index = 0;
+	c->prog_data.nr_params = 4*nr_params;
+	for (i = 0; i < nr_params; i++) {
+	    for (j = 0; j < 4; j++, index++) {
+		reg = brw_vec1_grf(c->reg_index + index/8, 
+			index%8);
+		c->prog_data.param[index] = 
+		    &plist->ParameterValues[i][j];
+		set_reg(c, PROGRAM_STATE_VAR, i, j, reg);
+	    }
+	}
+	c->nr_creg = 2*((4*nr_params+15)/16);
+	c->reg_index += c->nr_creg;
+    }
+    for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
+	if (inputs & (1<<i)) {
+	    nr_interp_regs++;
+	    reg = brw_vec8_grf(c->reg_index, 0);
+	    for (j = 0; j < 4; j++)
+		set_reg(c, PROGRAM_PAYLOAD, i, j, reg);
+	    c->reg_index += 2;
+
+	}
+    }
+    c->prog_data.first_curbe_grf = c->key.nr_depth_regs * 2;
+    c->prog_data.urb_read_length = nr_interp_regs * 2;
+    c->prog_data.curb_read_length = c->nr_creg;
+    c->emit_mask_reg = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
+    c->reg_index++;
+    c->stack =  brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
+    c->reg_index += 2;
+}
+
+static struct brw_reg get_dst_reg(struct brw_wm_compile *c, 
+	struct prog_instruction *inst, int component, int nr)
+{
+    return get_reg(c, inst->DstReg.File, inst->DstReg.Index, component, nr,
+	    0, 0);
+}
+
+static struct brw_reg get_src_reg(struct brw_wm_compile *c, 
+	struct prog_src_register *src, int index, int nr)
+{
+    int component = GET_SWZ(src->Swizzle, index);
+    return get_reg(c, src->File, src->Index, component, nr, 
+	    src->NegateBase, src->Abs);
+}
+
+/* Subroutines are minimal support for resusable instruction sequences.
+   They are implemented as simply as possible to minimise overhead: there
+   is no explicit support for communication between the caller and callee
+   other than saving the return address in a temporary register, nor is
+   there any automatic local storage.  This implies that great care is
+   required before attempting reentrancy or any kind of nested
+   subroutine invocations. */
+static void invoke_subroutine( struct brw_wm_compile *c,
+			       enum _subroutine subroutine,
+			       void (*emit)( struct brw_wm_compile * ) )
+{
+    struct brw_compile *p = &c->func;
+
+    assert( subroutine < BRW_WM_MAX_SUBROUTINE );
+    
+    if( c->subroutines[ subroutine ] ) {
+	/* subroutine previously emitted: reuse existing instructions */
+
+	int mark = mark_tmps( c );
+	struct brw_reg return_address = retype( alloc_tmp( c ),
+						BRW_REGISTER_TYPE_UD );
+	int here = p->nr_insn;
+	
+	brw_push_insn_state(p);
+	brw_set_mask_control(p, BRW_MASK_DISABLE);
+	brw_ADD( p, return_address, brw_ip_reg(), brw_imm_ud( 2 << 4 ) );
+
+	brw_ADD( p, brw_ip_reg(), brw_ip_reg(),
+		 brw_imm_d( ( c->subroutines[ subroutine ] -
+			      here - 1 ) << 4 ) );
+	brw_pop_insn_state(p);
+
+	release_tmps( c, mark );
+    } else {
+	/* previously unused subroutine: emit, and mark for later reuse */
+	
+	int mark = mark_tmps( c );
+	struct brw_reg return_address = retype( alloc_tmp( c ),
+						BRW_REGISTER_TYPE_UD );
+	struct brw_instruction *calc;
+	int base = p->nr_insn;
+	
+	brw_push_insn_state(p);
+	brw_set_mask_control(p, BRW_MASK_DISABLE);
+	calc = brw_ADD( p, return_address, brw_ip_reg(), brw_imm_ud( 0 ) );
+	brw_pop_insn_state(p);
+	
+	c->subroutines[ subroutine ] = p->nr_insn;
+
+	emit( c );
+	
+	brw_push_insn_state(p);
+	brw_set_mask_control(p, BRW_MASK_DISABLE);
+	brw_MOV( p, brw_ip_reg(), return_address );
+	brw_pop_insn_state(p);
+
+	brw_set_src1( calc, brw_imm_ud( ( p->nr_insn - base ) << 4 ) );
+	
+	release_tmps( c, mark );
+    }
+}
+
+static void emit_abs( struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    int i;
+    struct brw_compile *p = &c->func;
+    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+    for (i = 0; i < 4; i++) {
+	if (inst->DstReg.WriteMask & (1<<i)) {
+	    struct brw_reg src, dst;
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    brw_MOV(p, dst, brw_abs(src));
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_trunc( struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    int i;
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    struct brw_reg src, dst;
+	    dst = get_dst_reg(c, inst, i, 1) ;
+	    src = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    brw_RNDD(p, dst, src);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_mov( struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    int i;
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    struct brw_reg src, dst;
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    brw_MOV(p, dst, src);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_pixel_xy(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_reg r1 = brw_vec1_grf(1, 0);
+    struct brw_reg r1_uw = retype(r1, BRW_REGISTER_TYPE_UW);
+
+    struct brw_reg dst0, dst1;
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+
+    dst0 = get_dst_reg(c, inst, 0, 1);
+    dst1 = get_dst_reg(c, inst, 1, 1);
+    /* Calculate pixel centers by adding 1 or 0 to each of the
+     * micro-tile coordinates passed in r1.
+     */
+    if (mask & WRITEMASK_X) {
+	brw_ADD(p,
+		vec8(retype(dst0, BRW_REGISTER_TYPE_UW)),
+		stride(suboffset(r1_uw, 4), 2, 4, 0),
+		brw_imm_v(0x10101010));
+    }
+
+    if (mask & WRITEMASK_Y) {
+	brw_ADD(p,
+		vec8(retype(dst1, BRW_REGISTER_TYPE_UW)),
+		stride(suboffset(r1_uw, 5), 2, 4, 0),
+		brw_imm_v(0x11001100));
+    }
+
+}
+
+static void emit_delta_xy(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_reg r1 = brw_vec1_grf(1, 0);
+    struct brw_reg dst0, dst1, src0, src1;
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+
+    dst0 = get_dst_reg(c, inst, 0, 1);
+    dst1 = get_dst_reg(c, inst, 1, 1);
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    src1 = get_src_reg(c, &inst->SrcReg[0], 1, 1);
+    /* Calc delta X,Y by subtracting origin in r1 from the pixel
+     * centers.
+     */
+    if (mask & WRITEMASK_X) {
+	brw_ADD(p,
+		dst0,
+		retype(src0, BRW_REGISTER_TYPE_UW),
+		negate(r1));
+    }
+
+    if (mask & WRITEMASK_Y) {
+	brw_ADD(p,
+		dst1,
+		retype(src1, BRW_REGISTER_TYPE_UW),
+		negate(suboffset(r1,1)));
+
+    }
+
+}
+
+
+static void fire_fb_write( struct brw_wm_compile *c,
+                           GLuint base_reg,
+                           GLuint nr,
+                           GLuint target,
+                           GLuint eot)
+{
+    struct brw_compile *p = &c->func;
+    /* Pass through control information:
+     */
+    /*  mov (8) m1.0<1>:ud   r1.0<8;8,1>:ud   { Align1 NoMask } */
+    {
+	brw_push_insn_state(p);
+	brw_set_mask_control(p, BRW_MASK_DISABLE); /* ? */
+	brw_MOV(p,
+		brw_message_reg(base_reg + 1),
+		brw_vec8_grf(1, 0));
+	brw_pop_insn_state(p);
+    }
+    /* Send framebuffer write message: */
+    brw_fb_WRITE(p,
+	    retype(vec8(brw_null_reg()), BRW_REGISTER_TYPE_UW),
+	    base_reg,
+	    retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW),
+	    target,              
+	    nr,
+	    0,
+	    eot);
+}
+
+static void emit_fb_write(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    int nr = 2;
+    int channel;
+    GLuint target, eot;
+    struct brw_reg src0;
+
+    /* Reserve a space for AA - may not be needed:
+     */
+    if (c->key.aa_dest_stencil_reg)
+	nr += 1;
+    {
+	brw_push_insn_state(p);
+	for (channel = 0; channel < 4; channel++) {
+	    src0 = get_src_reg(c,  &inst->SrcReg[0], channel, 1);
+	    /*  mov (8) m2.0<1>:ud   r28.0<8;8,1>:ud  { Align1 } */
+	    /*  mov (8) m6.0<1>:ud   r29.0<8;8,1>:ud  { Align1 SecHalf } */
+	    brw_MOV(p, brw_message_reg(nr + channel), src0);
+	}
+	/* skip over the regs populated above: */
+	nr += 8;
+	brw_pop_insn_state(p);
+    }
+
+   if (c->key.source_depth_to_render_target)
+   {
+      if (c->key.computes_depth) {
+         src0 = get_src_reg(c, &inst->SrcReg[2], 2, 1);
+         brw_MOV(p, brw_message_reg(nr), src0);
+      } else {
+         src0 = get_src_reg(c, &inst->SrcReg[1], 1, 1);
+         brw_MOV(p, brw_message_reg(nr), src0);
+      }
+
+      nr += 2;
+   }
+    target = inst->Sampler >> 1;
+    eot = inst->Sampler & 1;
+    fire_fb_write(c, 0, nr, target, eot);
+}
+
+static void emit_pixel_w( struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    if (mask & WRITEMASK_W) {
+	struct brw_reg dst, src0, delta0, delta1;
+	struct brw_reg interp3;
+
+	dst = get_dst_reg(c, inst, 3, 1);
+	src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+	delta0 = get_src_reg(c, &inst->SrcReg[1], 0, 1);
+	delta1 = get_src_reg(c, &inst->SrcReg[1], 1, 1);
+
+	interp3 = brw_vec1_grf(src0.nr+1, 4);
+	/* Calc 1/w - just linterp wpos[3] optimized by putting the
+	 * result straight into a message reg.
+	 */
+	brw_LINE(p, brw_null_reg(), interp3, delta0);
+	brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), delta1);
+
+	/* Calc w */
+	brw_math_16( p, dst,
+		BRW_MATH_FUNCTION_INV,
+		BRW_MATH_SATURATE_NONE,
+		2, brw_null_reg(),
+		BRW_MATH_PRECISION_FULL);
+    }
+}
+
+static void emit_linterp(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg interp[4];
+    struct brw_reg dst, delta0, delta1;
+    struct brw_reg src0;
+
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    delta0 = get_src_reg(c, &inst->SrcReg[1], 0, 1);
+    delta1 = get_src_reg(c, &inst->SrcReg[1], 1, 1);
+    GLuint nr = src0.nr;
+    int i;
+
+    interp[0] = brw_vec1_grf(nr, 0);
+    interp[1] = brw_vec1_grf(nr, 4);
+    interp[2] = brw_vec1_grf(nr+1, 0);
+    interp[3] = brw_vec1_grf(nr+1, 4);
+
+    for(i = 0; i < 4; i++ ) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_LINE(p, brw_null_reg(), interp[i], delta0);
+	    brw_MAC(p, dst, suboffset(interp[i],1), delta1);
+	}
+    }
+}
+
+static void emit_cinterp(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+
+    struct brw_reg interp[4];
+    struct brw_reg dst, src0;
+
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    GLuint nr = src0.nr;
+    int i;
+
+    interp[0] = brw_vec1_grf(nr, 0);
+    interp[1] = brw_vec1_grf(nr, 4);
+    interp[2] = brw_vec1_grf(nr+1, 0);
+    interp[3] = brw_vec1_grf(nr+1, 4);
+
+    for(i = 0; i < 4; i++ ) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV(p, dst, suboffset(interp[i],3));
+	}
+    }
+}
+
+static void emit_pinterp(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+
+    struct brw_reg interp[4];
+    struct brw_reg dst, delta0, delta1;
+    struct brw_reg src0, w;
+
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    delta0 = get_src_reg(c, &inst->SrcReg[1], 0, 1);
+    delta1 = get_src_reg(c, &inst->SrcReg[1], 1, 1);
+    w = get_src_reg(c, &inst->SrcReg[2], 3, 1);
+    GLuint nr = src0.nr;
+    int i;
+
+    interp[0] = brw_vec1_grf(nr, 0);
+    interp[1] = brw_vec1_grf(nr, 4);
+    interp[2] = brw_vec1_grf(nr+1, 0);
+    interp[3] = brw_vec1_grf(nr+1, 4);
+
+    for(i = 0; i < 4; i++ ) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_LINE(p, brw_null_reg(), interp[i], delta0);
+	    brw_MAC(p, dst, suboffset(interp[i],1), 
+		    delta1);
+	    brw_MUL(p, dst, dst, w);
+	}
+    }
+}
+
+static void emit_xpd(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    int i;
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    for (i = 0; i < 4; i++) {
+	GLuint i2 = (i+2)%3;
+	GLuint i1 = (i+1)%3;
+	if (mask & (1<<i)) {
+	    struct brw_reg src0, src1, dst;
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = negate(get_src_reg(c, &inst->SrcReg[0], i2, 1));
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i1, 1);
+	    brw_MUL(p, brw_null_reg(), src0, src1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i1, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i2, 1);
+	    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+	    brw_MAC(p, dst, src0, src1);
+	    brw_set_saturate(p, 0);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_dp3(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_reg src0[3], src1[3], dst;
+    int i;
+    struct brw_compile *p = &c->func;
+    for (i = 0; i < 3; i++) {
+	src0[i] = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	src1[i] = get_src_reg(c, &inst->SrcReg[1], i, 1);
+    }
+
+    dst = get_dst_reg(c, inst, get_scalar_dst_index(inst), 1);
+    brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
+    brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    brw_MAC(p, dst, src0[2], src1[2]);
+    brw_set_saturate(p, 0);
+}
+
+static void emit_dp4(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_reg src0[4], src1[4], dst;
+    int i;
+    struct brw_compile *p = &c->func;
+    for (i = 0; i < 4; i++) {
+	src0[i] = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	src1[i] = get_src_reg(c, &inst->SrcReg[1], i, 1);
+    }
+    dst = get_dst_reg(c, inst, get_scalar_dst_index(inst), 1);
+    brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
+    brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
+    brw_MAC(p, brw_null_reg(), src0[2], src1[2]);
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    brw_MAC(p, dst, src0[3], src1[3]);
+    brw_set_saturate(p, 0);
+}
+
+static void emit_dph(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_reg src0[4], src1[4], dst;
+    int i;
+    struct brw_compile *p = &c->func;
+    for (i = 0; i < 4; i++) {
+	src0[i] = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	src1[i] = get_src_reg(c, &inst->SrcReg[1], i, 1);
+    }
+    dst = get_dst_reg(c, inst, get_scalar_dst_index(inst), 1);
+    brw_MUL(p, brw_null_reg(), src0[0], src1[0]);
+    brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
+    brw_MAC(p, dst, src0[2], src1[2]);
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    brw_ADD(p, dst, src0[3], src1[3]);
+    brw_set_saturate(p, 0);
+}
+
+static void emit_math1(struct brw_wm_compile *c,
+		struct prog_instruction *inst, GLuint func)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, dst;
+
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    dst = get_dst_reg(c, inst, get_scalar_dst_index(inst), 1);
+    brw_MOV(p, brw_message_reg(2), src0);
+    brw_math(p,
+	    dst,
+	    func,
+	    (inst->SaturateMode != SATURATE_OFF) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
+	    2,
+	    brw_null_reg(),
+	    BRW_MATH_DATA_VECTOR,
+	    BRW_MATH_PRECISION_FULL);
+}
+
+static void emit_rcp(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_INV);
+}
+
+static void emit_rsq(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_RSQ);
+}
+
+static void emit_sin(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_SIN);
+}
+
+static void emit_cos(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_COS);
+}
+
+static void emit_ex2(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_EXP);
+}
+
+static void emit_lg2(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_math1(c, inst, BRW_MATH_FUNCTION_LOG);
+}
+
+static void emit_add(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_ADD(p, dst, src0, src1);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_sub(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_ADD(p, dst, src0, negate(src1));
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_mul(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_MUL(p, dst, src0, src1);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_frc(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    brw_FRC(p, dst, src0);
+	}
+    }
+    if (inst->SaturateMode != SATURATE_OFF)
+	brw_set_saturate(p, 0);
+}
+
+static void emit_flr(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    brw_RNDD(p, dst, src0);
+	}
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_max(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg src0, src1, dst;
+    int i;
+    brw_push_insn_state(p);
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_MOV(p, dst, src0);
+	    brw_set_saturate(p, 0);
+
+	    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, src0, src1);
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+	    brw_MOV(p, dst, src1);
+	    brw_set_saturate(p, 0);
+	    brw_set_predicate_control_flag_value(p, 0xff);
+	}
+    }
+    brw_pop_insn_state(p);
+}
+
+static void emit_min(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg src0, src1, dst;
+    int i;
+    brw_push_insn_state(p);
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_MOV(p, dst, src0);
+	    brw_set_saturate(p, 0);
+
+	    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, src1, src0);
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+	    brw_MOV(p, dst, src1);
+	    brw_set_saturate(p, 0);
+	    brw_set_predicate_control_flag_value(p, 0xff);
+	}
+    }
+    brw_pop_insn_state(p);
+}
+
+static void emit_pow(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg dst, src0, src1;
+    dst = get_dst_reg(c, inst, get_scalar_dst_index(inst), 1);
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    src1 = get_src_reg(c, &inst->SrcReg[1], 0, 1);
+
+    brw_MOV(p, brw_message_reg(2), src0);
+    brw_MOV(p, brw_message_reg(3), src1);
+
+    brw_math(p,
+	    dst,
+	    BRW_MATH_FUNCTION_POW,
+	    (inst->SaturateMode != SATURATE_OFF) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
+	    2,
+	    brw_null_reg(),
+	    BRW_MATH_DATA_VECTOR,
+	    BRW_MATH_PRECISION_FULL);
+}
+
+static void emit_lrp(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg dst, tmp1, tmp2, src0, src1, src2;
+    int i;
+    int mark = mark_tmps(c);
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+
+	    if (src1.nr == dst.nr) {
+		tmp1 = alloc_tmp(c);
+		brw_MOV(p, tmp1, src1);
+	    } else
+		tmp1 = src1;
+
+	    src2 = get_src_reg(c, &inst->SrcReg[2], i, 1);
+	    if (src2.nr == dst.nr) {
+		tmp2 = alloc_tmp(c);
+		brw_MOV(p, tmp2, src2);
+	    } else
+		tmp2 = src2;
+
+	    brw_ADD(p, dst, negate(src0), brw_imm_f(1.0));
+	    brw_MUL(p, brw_null_reg(), dst, tmp2);
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_MAC(p, dst, src0, tmp1);
+	    brw_set_saturate(p, 0);
+	}
+	release_tmps(c, mark);
+    }
+}
+
+static void emit_kil(struct brw_wm_compile *c)
+{
+	struct brw_compile *p = &c->func;
+	struct brw_reg depth = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
+	brw_push_insn_state(p);
+	brw_set_mask_control(p, BRW_MASK_DISABLE);
+	brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
+	brw_AND(p, depth, c->emit_mask_reg, depth);
+	brw_pop_insn_state(p);
+}
+
+static void emit_mad(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg dst, src0, src1, src2;
+    int i;
+
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    src2 = get_src_reg(c, &inst->SrcReg[2], i, 1);
+	    brw_MUL(p, dst, src0, src1);
+
+	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
+	    brw_ADD(p, dst, dst, src2);
+	    brw_set_saturate(p, 0);
+	}
+    }
+}
+
+static void emit_sop(struct brw_wm_compile *c,
+		struct prog_instruction *inst, GLuint cond)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg dst, src0, src1;
+    int i;
+
+    for (i = 0; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    src0 = get_src_reg(c, &inst->SrcReg[0], i, 1);
+	    src1 = get_src_reg(c, &inst->SrcReg[1], i, 1);
+	    brw_push_insn_state(p);
+	    brw_CMP(p, brw_null_reg(), cond, src0, src1);
+	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	    brw_MOV(p, dst, brw_imm_f(0.0));
+	    brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+	    brw_MOV(p, dst, brw_imm_f(1.0));
+	    brw_pop_insn_state(p);
+	}
+    }
+}
+
+static void emit_slt(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_L);
+}
+
+static void emit_sle(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_LE);
+}
+
+static void emit_sgt(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_G);
+}
+
+static void emit_sge(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_GE);
+}
+
+static void emit_seq(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_EQ);
+}
+
+static void emit_sne(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    emit_sop(c, inst, BRW_CONDITIONAL_NEQ);
+}
+
+static void emit_ddx(struct brw_wm_compile *c,
+                struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg interp[4];
+    struct brw_reg dst;
+    struct brw_reg src0, w;
+    GLuint nr, i;
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    w = get_src_reg(c, &inst->SrcReg[1], 3, 1);
+    nr = src0.nr;
+    interp[0] = brw_vec1_grf(nr, 0);
+    interp[1] = brw_vec1_grf(nr, 4);
+    interp[2] = brw_vec1_grf(nr+1, 0);
+    interp[3] = brw_vec1_grf(nr+1, 4);
+    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+    for(i = 0; i < 4; i++ ) {
+        if (mask & (1<<i)) {
+            dst = get_dst_reg(c, inst, i, 1);
+            brw_MOV(p, dst, interp[i]);
+            brw_MUL(p, dst, dst, w);
+        }
+    }
+    brw_set_saturate(p, 0);
+}
+
+static void emit_ddy(struct brw_wm_compile *c,
+                struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg interp[4];
+    struct brw_reg dst;
+    struct brw_reg src0, w;
+    GLuint nr, i;
+
+    src0 = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    nr = src0.nr;
+    w = get_src_reg(c, &inst->SrcReg[1], 3, 1);
+    interp[0] = brw_vec1_grf(nr, 0);
+    interp[1] = brw_vec1_grf(nr, 4);
+    interp[2] = brw_vec1_grf(nr+1, 0);
+    interp[3] = brw_vec1_grf(nr+1, 4);
+    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
+    for(i = 0; i < 4; i++ ) {
+        if (mask & (1<<i)) {
+            dst = get_dst_reg(c, inst, i, 1);
+            brw_MOV(p, dst, suboffset(interp[i], 1));
+            brw_MUL(p, dst, dst, w);
+        }
+    }
+    brw_set_saturate(p, 0);
+}
+
+static __inline struct brw_reg high_words( struct brw_reg reg )
+{
+    return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_W ), 1 ),
+		   0, 8, 2 );
+}
+
+static __inline struct brw_reg low_words( struct brw_reg reg )
+{
+    return stride( retype( reg, BRW_REGISTER_TYPE_W ), 0, 8, 2 );
+}
+
+static __inline struct brw_reg even_bytes( struct brw_reg reg )
+{
+    return stride( retype( reg, BRW_REGISTER_TYPE_B ), 0, 16, 2 );
+}
+
+static __inline struct brw_reg odd_bytes( struct brw_reg reg )
+{
+    return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_B ), 1 ),
+		   0, 16, 2 );
+}
+
+/* One-, two- and three-dimensional Perlin noise, similar to the description
+   in _Improving Noise_, Ken Perlin, Computer Graphics vol. 35 no. 3. */
+static void noise1_sub( struct brw_wm_compile *c ) {
+
+    struct brw_compile *p = &c->func;
+    struct brw_reg param,
+	x0, x1, /* gradients at each end */       
+	t, tmp[ 2 ], /* float temporaries */
+	itmp[ 5 ]; /* unsigned integer temporaries (aliases of floats above) */
+    int i;
+    int mark = mark_tmps( c );
+
+    x0 = alloc_tmp( c );
+    x1 = alloc_tmp( c );
+    t = alloc_tmp( c );
+    tmp[ 0 ] = alloc_tmp( c );
+    tmp[ 1 ] = alloc_tmp( c );
+    itmp[ 0 ] = retype( tmp[ 0 ], BRW_REGISTER_TYPE_UD );
+    itmp[ 1 ] = retype( tmp[ 1 ], BRW_REGISTER_TYPE_UD );
+    itmp[ 2 ] = retype( x0, BRW_REGISTER_TYPE_UD );
+    itmp[ 3 ] = retype( x1, BRW_REGISTER_TYPE_UD );
+    itmp[ 4 ] = retype( t, BRW_REGISTER_TYPE_UD );
+    
+    param = lookup_tmp( c, mark - 2 );
+
+    brw_set_access_mode( p, BRW_ALIGN_1 );
+
+    brw_MOV( p, itmp[ 2 ], brw_imm_ud( 0xBA97 ) ); /* constant used later */
+
+    /* Arrange the two end coordinates into scalars (itmp0/itmp1) to
+       be hashed.  Also compute the remainder (offset within the unit
+       length), interleaved to reduce register dependency penalties. */
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param );
+    brw_FRC( p, param, param );
+    brw_ADD( p, itmp[ 1 ], itmp[ 0 ], brw_imm_ud( 1 ) );
+    brw_MOV( p, itmp[ 3 ], brw_imm_ud( 0x79D9 ) ); /* constant used later */
+    brw_MOV( p, itmp[ 4 ], brw_imm_ud( 0xD5B1 ) ); /* constant used later */
+
+    /* We're now ready to perform the hashing.  The two hashes are
+       interleaved for performance.  The hash function used is
+       designed to rapidly achieve avalanche and require only 32x16
+       bit multiplication, and 16-bit swizzles (which we get for
+       free).  We can't use immediate operands in the multiplies,
+       because immediates are permitted only in src1 and the 16-bit
+       factor is permitted only in src0. */
+    for( i = 0; i < 2; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 2 ], itmp[ i ] );
+    for( i = 0; i < 2; i++ )
+       brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		high_words( itmp[ i ] ) );
+    for( i = 0; i < 2; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 3 ], itmp[ i ] );
+    for( i = 0; i < 2; i++ )
+       brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		high_words( itmp[ i ] ) );
+    for( i = 0; i < 2; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 4 ], itmp[ i ] );
+    for( i = 0; i < 2; i++ )
+       brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		high_words( itmp[ i ] ) );
+
+    /* Now we want to initialise the two gradients based on the
+       hashes.  Format conversion from signed integer to float leaves
+       everything scaled too high by a factor of pow( 2, 31 ), but
+       we correct for that right at the end. */
+    brw_ADD( p, t, param, brw_imm_f( -1.0 ) );
+    brw_MOV( p, x0, retype( tmp[ 0 ], BRW_REGISTER_TYPE_D ) );
+    brw_MOV( p, x1, retype( tmp[ 1 ], BRW_REGISTER_TYPE_D ) );
+
+    brw_MUL( p, x0, x0, param );
+    brw_MUL( p, x1, x1, t );
+    
+    /* We interpolate between the gradients using the polynomial
+       6t^5 - 15t^4 + 10t^3 (Perlin). */
+    brw_MUL( p, tmp[ 0 ], param, brw_imm_f( 6.0 ) );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], brw_imm_f( -15.0 ) );
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], brw_imm_f( 10.0 ) );
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param );
+    brw_ADD( p, x1, x1, negate( x0 ) ); /* unrelated work to fill the
+					   pipeline */
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param );
+    brw_MUL( p, param, tmp[ 0 ], param );
+    brw_MUL( p, x1, x1, param );
+    brw_ADD( p, x0, x0, x1 );    
+    /* scale by pow( 2, -30 ), to compensate for the format conversion
+       above and an extra factor of 2 so that a single gradient covers
+       the [-1,1] range */
+    brw_MUL( p, param, x0, brw_imm_f( 0.000000000931322574615478515625 ) );
+
+    release_tmps( c, mark );
+}
+
+static void emit_noise1( struct brw_wm_compile *c,
+			 struct prog_instruction *inst )
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src, param, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    int mark = mark_tmps( c );
+
+    assert( mark == 0 );
+    
+    src = get_src_reg( c, inst->SrcReg, 0, 1 );
+
+    param = alloc_tmp( c );
+
+    brw_MOV( p, param, src );
+
+    invoke_subroutine( c, SUB_NOISE1, noise1_sub );
+    
+    /* Fill in the result: */
+    brw_set_saturate( p, inst->SaturateMode == SATURATE_ZERO_ONE );
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV( p, dst, param );
+	}
+    }
+    if( inst->SaturateMode == SATURATE_ZERO_ONE )
+	brw_set_saturate( p, 0 );
+    
+    release_tmps( c, mark );
+}
+    
+static void noise2_sub( struct brw_wm_compile *c ) {
+
+    struct brw_compile *p = &c->func;
+    struct brw_reg param0, param1,
+	x0y0, x0y1, x1y0, x1y1, /* gradients at each corner */       
+	t, tmp[ 4 ], /* float temporaries */
+	itmp[ 7 ]; /* unsigned integer temporaries (aliases of floats above) */
+    int i;
+    int mark = mark_tmps( c );
+
+    x0y0 = alloc_tmp( c );
+    x0y1 = alloc_tmp( c );
+    x1y0 = alloc_tmp( c );
+    x1y1 = alloc_tmp( c );
+    t = alloc_tmp( c );
+    for( i = 0; i < 4; i++ ) {
+	tmp[ i ] = alloc_tmp( c );
+	itmp[ i ] = retype( tmp[ i ], BRW_REGISTER_TYPE_UD );
+    }
+    itmp[ 4 ] = retype( x0y0, BRW_REGISTER_TYPE_UD );
+    itmp[ 5 ] = retype( x0y1, BRW_REGISTER_TYPE_UD );
+    itmp[ 6 ] = retype( x1y0, BRW_REGISTER_TYPE_UD );
+    
+    param0 = lookup_tmp( c, mark - 3 );
+    param1 = lookup_tmp( c, mark - 2 );
+
+    brw_set_access_mode( p, BRW_ALIGN_1 );
+    
+    /* Arrange the four corner coordinates into scalars (itmp0..itmp3) to
+       be hashed.  Also compute the remainders (offsets within the unit
+       square), interleaved to reduce register dependency penalties. */
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param0 );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param1 );
+    brw_FRC( p, param0, param0 );
+    brw_FRC( p, param1, param1 );
+    brw_MOV( p, itmp[ 4 ], brw_imm_ud( 0xBA97 ) ); /* constant used later */
+    brw_ADD( p, high_words( itmp[ 0 ] ), high_words( itmp[ 0 ] ),
+	     low_words( itmp[ 1 ] ) );
+    brw_MOV( p, itmp[ 5 ], brw_imm_ud( 0x79D9 ) ); /* constant used later */
+    brw_MOV( p, itmp[ 6 ], brw_imm_ud( 0xD5B1 ) ); /* constant used later */
+    brw_ADD( p, itmp[ 1 ], itmp[ 0 ], brw_imm_ud( 0x10000 ) );
+    brw_ADD( p, itmp[ 2 ], itmp[ 0 ], brw_imm_ud( 0x1 ) );
+    brw_ADD( p, itmp[ 3 ], itmp[ 0 ], brw_imm_ud( 0x10001 ) );
+
+    /* We're now ready to perform the hashing.  The four hashes are
+       interleaved for performance.  The hash function used is
+       designed to rapidly achieve avalanche and require only 32x16
+       bit multiplication, and 16-bit swizzles (which we get for
+       free).  We can't use immediate operands in the multiplies,
+       because immediates are permitted only in src1 and the 16-bit
+       factor is permitted only in src0. */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 4 ], itmp[ i ] );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		 high_words( itmp[ i ] ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 5 ], itmp[ i ] );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		 high_words( itmp[ i ] ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, itmp[ i ], itmp[ 6 ], itmp[ i ] );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, low_words( itmp[ i ] ), low_words( itmp[ i ] ),
+		 high_words( itmp[ i ] ) );
+
+    /* Now we want to initialise the four gradients based on the
+       hashes.  Format conversion from signed integer to float leaves
+       everything scaled too high by a factor of pow( 2, 15 ), but
+       we correct for that right at the end. */
+    brw_ADD( p, t, param0, brw_imm_f( -1.0 ) );
+    brw_MOV( p, x0y0, low_words( tmp[ 0 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 1 ] ) );
+    brw_MOV( p, x1y0, low_words( tmp[ 2 ] ) );
+    brw_MOV( p, x1y1, low_words( tmp[ 3 ] ) );
+    
+    brw_MOV( p, tmp[ 0 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 1 ], high_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 2 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 3 ], high_words( tmp[ 3 ] ) );
+    
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param1, brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param0 );
+    brw_MUL( p, x0y1, x0y1, param0 );
+
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param1 );
+    brw_MUL( p, tmp[ 2 ], tmp[ 2 ], param1 );
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], t );
+    brw_MUL( p, tmp[ 3 ], tmp[ 3 ], t );
+
+    brw_ADD( p, x0y0, x0y0, tmp[ 0 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 2 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 1 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 3 ] );
+    
+    /* We interpolate between the gradients using the polynomial
+       6t^5 - 15t^4 + 10t^3 (Perlin). */
+    brw_MUL( p, tmp[ 0 ], param0, brw_imm_f( 6.0 ) );
+    brw_MUL( p, tmp[ 1 ], param1, brw_imm_f( 6.0 ) );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], brw_imm_f( -15.0 ) );
+    brw_ADD( p, tmp[ 1 ], tmp[ 1 ], brw_imm_f( -15.0 ) );
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param0 );
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], param1 );
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) ); /* unrelated work to fill the
+						 pipeline */
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], brw_imm_f( 10.0 ) );
+    brw_ADD( p, tmp[ 1 ], tmp[ 1 ], brw_imm_f( 10.0 ) );
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param0 );
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], param1 );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) ); /* unrelated work to fill the
+						 pipeline */
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], param0 );
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], param1 );
+    brw_MUL( p, param0, tmp[ 0 ], param0 );
+    brw_MUL( p, param1, tmp[ 1 ], param1 );
+    
+    /* Here we interpolate in the y dimension... */
+    brw_MUL( p, x0y1, x0y1, param1 );
+    brw_MUL( p, x1y1, x1y1, param1 );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  There are horrible register dependencies here,
+       but we have nothing else to do. */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, param0 );
+    brw_ADD( p, x0y0, x0y0, x1y0 );
+    
+    /* scale by pow( 2, -15 ), as described above */
+    brw_MUL( p, param0, x0y0, brw_imm_f( 0.000030517578125 ) );
+
+    release_tmps( c, mark );
+}
+
+static void emit_noise2( struct brw_wm_compile *c,
+			 struct prog_instruction *inst )
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, param0, param1, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    int mark = mark_tmps( c );
+
+    assert( mark == 0 );
+    
+    src0 = get_src_reg( c, inst->SrcReg, 0, 1 );
+    src1 = get_src_reg( c, inst->SrcReg, 1, 1 );
+
+    param0 = alloc_tmp( c );
+    param1 = alloc_tmp( c );
+
+    brw_MOV( p, param0, src0 );
+    brw_MOV( p, param1, src1 );
+
+    invoke_subroutine( c, SUB_NOISE2, noise2_sub );
+    
+    /* Fill in the result: */
+    brw_set_saturate( p, inst->SaturateMode == SATURATE_ZERO_ONE );
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV( p, dst, param0 );
+	}
+    }
+    if( inst->SaturateMode == SATURATE_ZERO_ONE )
+	brw_set_saturate( p, 0 );
+    
+    release_tmps( c, mark );
+}
+
+/* The three-dimensional case is much like the one- and two- versions above,
+   but since the number of corners is rapidly growing we now pack 16 16-bit
+   hashes into each register to extract more parallelism from the EUs. */
+static void noise3_sub( struct brw_wm_compile *c ) {
+
+    struct brw_compile *p = &c->func;
+    struct brw_reg param0, param1, param2,
+	x0y0, x0y1, x1y0, x1y1, /* gradients at four of the corners */
+	xi, yi, zi, /* interpolation coefficients */
+	t, tmp[ 8 ], /* float temporaries */
+	itmp[ 8 ], /* unsigned integer temporaries (aliases of floats above) */
+	wtmp[ 8 ]; /* 16-way unsigned word temporaries (aliases of above) */
+    int i;
+    int mark = mark_tmps( c );
+
+    x0y0 = alloc_tmp( c );
+    x0y1 = alloc_tmp( c );
+    x1y0 = alloc_tmp( c );
+    x1y1 = alloc_tmp( c );
+    xi = alloc_tmp( c );
+    yi = alloc_tmp( c );
+    zi = alloc_tmp( c );
+    t = alloc_tmp( c );
+    for( i = 0; i < 8; i++ ) {
+	tmp[ i ] = alloc_tmp( c );
+	itmp[ i ] = retype( tmp[ i ], BRW_REGISTER_TYPE_UD );
+	wtmp[ i ] = brw_uw16_grf( tmp[ i ].nr, 0 );
+    }
+    
+    param0 = lookup_tmp( c, mark - 4 );
+    param1 = lookup_tmp( c, mark - 3 );
+    param2 = lookup_tmp( c, mark - 2 );
+
+    brw_set_access_mode( p, BRW_ALIGN_1 );
+    
+    /* Arrange the eight corner coordinates into scalars (itmp0..itmp3) to
+       be hashed.  Also compute the remainders (offsets within the unit
+       cube), interleaved to reduce register dependency penalties. */
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param0 );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param1 );
+    brw_RNDD( p, retype( itmp[ 2 ], BRW_REGISTER_TYPE_D ), param2 );
+    brw_FRC( p, param0, param0 );
+    brw_FRC( p, param1, param1 );
+    brw_FRC( p, param2, param2 );
+    /* Since we now have only 16 bits of precision in the hash, we must
+       be more careful about thorough mixing to maintain entropy as we
+       squash the input vector into a small scalar. */
+    brw_MUL( p, brw_null_reg(), low_words( itmp[ 0 ] ), brw_imm_uw( 0xBC8F ) );
+    brw_MAC( p, brw_null_reg(), low_words( itmp[ 1 ] ), brw_imm_uw( 0xD0BD ) );
+    brw_MAC( p, low_words( itmp[ 0 ] ), low_words( itmp[ 2 ] ),
+	     brw_imm_uw( 0x9B93 ) );
+    brw_ADD( p, high_words( itmp[ 0 ] ), low_words( itmp[ 0 ] ),
+	     brw_imm_uw( 0xBC8F ) );
+
+    /* Temporarily disable the execution mask while we work with ExecSize=16
+       channels (the mask is set for ExecSize=8 and is probably incorrect).
+       Although this might cause execution of unwanted channels, the code
+       writes only to temporary registers and has no side effects, so
+       disabling the mask is harmless. */
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_ADD( p, wtmp[ 1 ], wtmp[ 0 ], brw_imm_uw( 0xD0BD ) );
+    brw_ADD( p, wtmp[ 2 ], wtmp[ 0 ], brw_imm_uw( 0x9B93 ) );
+    brw_ADD( p, wtmp[ 3 ], wtmp[ 1 ], brw_imm_uw( 0x9B93 ) );
+
+    /* We're now ready to perform the hashing.  The eight hashes are
+       interleaved for performance.  The hash function used is
+       designed to rapidly achieve avalanche and require only 16x16
+       bit multiplication, and 8-bit swizzles (which we get for
+       free). */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0x28D9 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0xC6D5 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    brw_pop_insn_state( p );
+
+    /* Now we want to initialise the four rear gradients based on the
+       hashes.  Format conversion from signed integer to float leaves
+       everything scaled too high by a factor of pow( 2, 15 ), but
+       we correct for that right at the end. */
+    /* x component */
+    brw_ADD( p, t, param0, brw_imm_f( -1.0 ) );
+    brw_MOV( p, x0y0, low_words( tmp[ 0 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 1 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 0 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 1 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 5 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 5 ) );
+    brw_pop_insn_state( p );
+    
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param1, brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param0 );
+    brw_MUL( p, x0y1, x0y1, param0 );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 5 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 5 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param0, brw_imm_f( -1.0 ) );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param1 );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param1 );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param2 );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], param2 );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param2 );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], param2 );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    
+    /* We interpolate between the gradients using the polynomial
+       6t^5 - 15t^4 + 10t^3 (Perlin). */
+    brw_MUL( p, xi, param0, brw_imm_f( 6.0 ) );
+    brw_MUL( p, yi, param1, brw_imm_f( 6.0 ) );
+    brw_MUL( p, zi, param2, brw_imm_f( 6.0 ) );
+    brw_ADD( p, xi, xi, brw_imm_f( -15.0 ) );
+    brw_ADD( p, yi, yi, brw_imm_f( -15.0 ) );
+    brw_ADD( p, zi, zi, brw_imm_f( -15.0 ) );
+    brw_MUL( p, xi, xi, param0 );
+    brw_MUL( p, yi, yi, param1 );
+    brw_MUL( p, zi, zi, param2 );
+    brw_ADD( p, xi, xi, brw_imm_f( 10.0 ) );
+    brw_ADD( p, yi, yi, brw_imm_f( 10.0 ) );
+    brw_ADD( p, zi, zi, brw_imm_f( 10.0 ) );
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) ); /* unrelated work */
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) ); /* unrelated work */
+    brw_MUL( p, xi, xi, param0 );
+    brw_MUL( p, yi, yi, param1 );
+    brw_MUL( p, zi, zi, param2 );
+    brw_MUL( p, xi, xi, param0 );
+    brw_MUL( p, yi, yi, param1 );
+    brw_MUL( p, zi, zi, param2 );
+    brw_MUL( p, xi, xi, param0 );
+    brw_MUL( p, yi, yi, param1 );
+    brw_MUL( p, zi, zi, param2 );
+    
+    /* Here we interpolate in the y dimension... */
+    brw_MUL( p, x0y1, x0y1, yi );
+    brw_MUL( p, x1y1, x1y1, yi );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  Leave the result in tmp[ 0 ] (see below)... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, xi );
+    brw_ADD( p, tmp[ 0 ], x0y0, x1y0 );
+
+    /* Now do the same thing for the front four gradients... */
+    /* x component */
+    brw_MOV( p, x0y0, low_words( tmp[ 2 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 3 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 2 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 3 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 5 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 5 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param1, brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param0 );
+    brw_MUL( p, x0y1, x0y1, param0 );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 5 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 5 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param2, brw_imm_f( -1.0 ) );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param1 );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param1 );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    
+    /* The interpolation coefficients are still around from last time, so
+       again interpolate in the y dimension... */
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) );
+    brw_MUL( p, x0y1, x0y1, yi );
+    brw_MUL( p, x1y1, x1y1, yi );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  The rear face is in tmp[ 0 ] (see above), so this
+       time put the front face in tmp[ 1 ] and we're nearly there... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, xi );
+    brw_ADD( p, tmp[ 1 ], x0y0, x1y0 );
+
+    /* The final interpolation, in the z dimension: */
+    brw_ADD( p, tmp[ 1 ], tmp[ 1 ], negate( tmp[ 0 ] ) );    
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], zi );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], tmp[ 1 ] );
+    
+    /* scale by pow( 2, -15 ), as described above */
+    brw_MUL( p, param0, tmp[ 0 ], brw_imm_f( 0.000030517578125 ) );
+
+    release_tmps( c, mark );
+}
+
+static void emit_noise3( struct brw_wm_compile *c,
+			 struct prog_instruction *inst )
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, src2, param0, param1, param2, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    int mark = mark_tmps( c );
+
+    assert( mark == 0 );
+    
+    src0 = get_src_reg( c, inst->SrcReg, 0, 1 );
+    src1 = get_src_reg( c, inst->SrcReg, 1, 1 );
+    src2 = get_src_reg( c, inst->SrcReg, 2, 1 );
+
+    param0 = alloc_tmp( c );
+    param1 = alloc_tmp( c );
+    param2 = alloc_tmp( c );
+
+    brw_MOV( p, param0, src0 );
+    brw_MOV( p, param1, src1 );
+    brw_MOV( p, param2, src2 );
+
+    invoke_subroutine( c, SUB_NOISE3, noise3_sub );
+    
+    /* Fill in the result: */
+    brw_set_saturate( p, inst->SaturateMode == SATURATE_ZERO_ONE );
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV( p, dst, param0 );
+	}
+    }
+    if( inst->SaturateMode == SATURATE_ZERO_ONE )
+	brw_set_saturate( p, 0 );
+    
+    release_tmps( c, mark );
+}
+    
+/* For the four-dimensional case, the little micro-optimisation benefits
+   we obtain by unrolling all the loops aren't worth the massive bloat it
+   now causes.  Instead, we loop twice around performing a similar operation
+   to noise3, once for the w=0 cube and once for the w=1, with a bit more
+   code to glue it all together. */
+static void noise4_sub( struct brw_wm_compile *c ) {
+
+    struct brw_compile *p = &c->func;
+    struct brw_reg param[ 4 ],
+	x0y0, x0y1, x1y0, x1y1, /* gradients at four of the corners */
+	w0, /* noise for the w=0 cube */
+	floors[ 2 ], /* integer coordinates of base corner of hypercube */
+	interp[ 4 ], /* interpolation coefficients */
+	t, tmp[ 8 ], /* float temporaries */
+	itmp[ 8 ], /* unsigned integer temporaries (aliases of floats above) */
+	wtmp[ 8 ]; /* 16-way unsigned word temporaries (aliases of above) */
+    int i, j;
+    int mark = mark_tmps( c );
+    GLuint loop, origin;
+    
+    x0y0 = alloc_tmp( c );
+    x0y1 = alloc_tmp( c );
+    x1y0 = alloc_tmp( c );
+    x1y1 = alloc_tmp( c );
+    t = alloc_tmp( c );
+    w0 = alloc_tmp( c );    
+    floors[ 0 ] = retype( alloc_tmp( c ), BRW_REGISTER_TYPE_UD );
+    floors[ 1 ] = retype( alloc_tmp( c ), BRW_REGISTER_TYPE_UD );
+
+    for( i = 0; i < 4; i++ ) {
+	param[ i ] = lookup_tmp( c, mark - 5 + i );
+	interp[ i ] = alloc_tmp( c );
+    }
+    
+    for( i = 0; i < 8; i++ ) {
+	tmp[ i ] = alloc_tmp( c );
+	itmp[ i ] = retype( tmp[ i ], BRW_REGISTER_TYPE_UD );
+	wtmp[ i ] = brw_uw16_grf( tmp[ i ].nr, 0 );
+    }
+
+    brw_set_access_mode( p, BRW_ALIGN_1 );
+
+    /* We only want 16 bits of precision from the integral part of each
+       co-ordinate, but unfortunately the RNDD semantics would saturate
+       at 16 bits if we performed the operation directly to a 16-bit
+       destination.  Therefore, we round to 32-bit temporaries where
+       appropriate, and then store only the lower 16 bits. */
+    brw_RNDD( p, retype( floors[ 0 ], BRW_REGISTER_TYPE_D ), param[ 0 ] );
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param[ 1 ] );
+    brw_RNDD( p, retype( floors[ 1 ], BRW_REGISTER_TYPE_D ), param[ 2 ] );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param[ 3 ] );
+    brw_MOV( p, high_words( floors[ 0 ] ), low_words( itmp[ 0 ] ) );
+    brw_MOV( p, high_words( floors[ 1 ] ), low_words( itmp[ 1 ] ) );
+
+    /* Modify the flag register here, because the side effect is useful
+       later (see below).  We know for certain that all flags will be
+       cleared, since the FRC instruction cannot possibly generate
+       negative results.  Even for exceptional inputs (infinities, denormals,
+       NaNs), the architecture guarantees that the L conditional is false. */
+    brw_set_conditionalmod( p, BRW_CONDITIONAL_L );
+    brw_FRC( p, param[ 0 ], param[ 0 ] );
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    for( i = 1; i < 4; i++ )	
+	brw_FRC( p, param[ i ], param[ i ] );
+    
+    /* Calculate the interpolation coefficients (6t^5 - 15t^4 + 10t^3) first
+       of all. */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, interp[ i ], param[ i ], brw_imm_f( 6.0 ) );
+    for( i = 0; i < 4; i++ )
+	brw_ADD( p, interp[ i ], interp[ i ], brw_imm_f( -15.0 ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, interp[ i ], interp[ i ], param[ i ] );
+    for( i = 0; i < 4; i++ )
+	brw_ADD( p, interp[ i ], interp[ i ], brw_imm_f( 10.0 ) );
+    for( j = 0; j < 3; j++ )
+	for( i = 0; i < 4; i++ )
+	    brw_MUL( p, interp[ i ], interp[ i ], param[ i ] );
+
+    /* Mark the current address, as it will be a jump destination.  The
+       following code will be executed twice: first, with the flag
+       register clear indicating the w=0 case, and second with flags
+       set for w=1. */
+    loop = p->nr_insn;
+    
+    /* Arrange the eight corner coordinates into scalars (itmp0..itmp3) to
+       be hashed.  Since we have only 16 bits of precision in the hash, we
+       must be careful about thorough mixing to maintain entropy as we
+       squash the input vector into a small scalar. */
+    brw_MUL( p, brw_null_reg(), low_words( floors[ 0 ] ),
+	     brw_imm_uw( 0xBC8F ) );
+    brw_MAC( p, brw_null_reg(), high_words( floors[ 0 ] ),
+	     brw_imm_uw( 0xD0BD ) );
+    brw_MAC( p, brw_null_reg(), low_words( floors[ 1 ] ),
+	     brw_imm_uw( 0x9B93 ) );
+    brw_MAC( p, low_words( itmp[ 0 ] ), high_words( floors[ 1 ] ),
+	     brw_imm_uw( 0xA359 ) );
+    brw_ADD( p, high_words( itmp[ 0 ] ), low_words( itmp[ 0 ] ),
+	     brw_imm_uw( 0xBC8F ) );
+
+    /* Temporarily disable the execution mask while we work with ExecSize=16
+       channels (the mask is set for ExecSize=8 and is probably incorrect).
+       Although this might cause execution of unwanted channels, the code
+       writes only to temporary registers and has no side effects, so
+       disabling the mask is harmless. */
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_ADD( p, wtmp[ 1 ], wtmp[ 0 ], brw_imm_uw( 0xD0BD ) );
+    brw_ADD( p, wtmp[ 2 ], wtmp[ 0 ], brw_imm_uw( 0x9B93 ) );
+    brw_ADD( p, wtmp[ 3 ], wtmp[ 1 ], brw_imm_uw( 0x9B93 ) );
+
+    /* We're now ready to perform the hashing.  The eight hashes are
+       interleaved for performance.  The hash function used is
+       designed to rapidly achieve avalanche and require only 16x16
+       bit multiplication, and 8-bit swizzles (which we get for
+       free). */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0x28D9 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0xC6D5 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    brw_pop_insn_state( p );
+
+    /* Now we want to initialise the four rear gradients based on the
+       hashes.  Format conversion from signed integer to float leaves
+       everything scaled too high by a factor of pow( 2, 15 ), but
+       we correct for that right at the end. */
+    /* x component */
+    brw_ADD( p, t, param[ 0 ], brw_imm_f( -1.0 ) );
+    brw_MOV( p, x0y0, low_words( tmp[ 0 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 1 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 0 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 1 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+    
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param[ 1 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param[ 0 ] );
+    brw_MUL( p, x0y1, x0y1, param[ 0 ] );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );    
+    /* prepare t for the w component (used below): w the first time through
+       the loop; w - 1 the second time) */
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_ADD( p, t, param[ 3 ], brw_imm_f( -1.0 ) );
+    p->current->header.predicate_inverse = 1;
+    brw_MOV( p, t, param[ 3 ] );
+    p->current->header.predicate_inverse = 0;
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 1 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 1 ] );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], param[ 2 ] );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* w component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param[ 0 ], brw_imm_f( -1.0 ) );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* Here we interpolate in the y dimension... */
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) );
+    brw_MUL( p, x0y1, x0y1, interp[ 1 ] );
+    brw_MUL( p, x1y1, x1y1, interp[ 1 ] );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  Leave the result in tmp[ 0 ] (see below)... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, interp[ 0 ] );
+    brw_ADD( p, tmp[ 0 ], x0y0, x1y0 );
+
+    /* Now do the same thing for the front four gradients... */
+    /* x component */
+    brw_MOV( p, x0y0, low_words( tmp[ 2 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 3 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 2 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 3 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param[ 1 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param[ 0 ] );
+    brw_MUL( p, x0y1, x0y1, param[ 0 ] );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param[ 2 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 1 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 1 ] );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    /* prepare t for the w component (used below): w the first time through
+       the loop; w - 1 the second time) */
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_ADD( p, t, param[ 3 ], brw_imm_f( -1.0 ) );
+    p->current->header.predicate_inverse = 1;
+    brw_MOV( p, t, param[ 3 ] );
+    p->current->header.predicate_inverse = 0;
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* w component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* Interpolate in the y dimension: */
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) );
+    brw_MUL( p, x0y1, x0y1, interp[ 1 ] );
+    brw_MUL( p, x1y1, x1y1, interp[ 1 ] );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  The rear face is in tmp[ 0 ] (see above), so this
+       time put the front face in tmp[ 1 ] and we're nearly there... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, interp[ 0 ] );
+    brw_ADD( p, tmp[ 1 ], x0y0, x1y0 );
+
+    /* Another interpolation, in the z dimension: */
+    brw_ADD( p, tmp[ 1 ], tmp[ 1 ], negate( tmp[ 0 ] ) );    
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], interp[ 2 ] );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], tmp[ 1 ] );
+
+    /* Exit the loop if we've computed both cubes... */
+    origin = p->nr_insn;
+    brw_push_insn_state( p );
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_ADD( p, brw_ip_reg(), brw_ip_reg(), brw_imm_d( 0 ) );
+    brw_pop_insn_state( p );
+
+    /* Save the result for the w=0 case, and increment the w coordinate: */
+    brw_MOV( p, w0, tmp[ 0 ] );
+    brw_ADD( p, high_words( floors[ 1 ] ), high_words( floors[ 1 ] ),
+	     brw_imm_uw( 1 ) );
+
+    /* Loop around for the other cube.  Explicitly set the flag register
+       (unfortunately we must spend an extra instruction to do this: we
+       can't rely on a side effect of the previous MOV or ADD because
+       conditional modifiers which are normally true might be false in
+       exceptional circumstances, e.g. given a NaN input; the add to
+       brw_ip_reg() is not suitable because the IP is not an 8-vector). */
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_MOV( p, brw_flag_reg(), brw_imm_uw( 0xFF ) );
+    brw_ADD( p, brw_ip_reg(), brw_ip_reg(),
+	     brw_imm_d( ( loop - p->nr_insn ) << 4 ) );
+    brw_pop_insn_state( p );
+
+    /* Patch the previous conditional branch now that we know the
+       destination address. */
+    brw_set_src1( p->store + origin,
+		  brw_imm_d( ( p->nr_insn - origin ) << 4 ) );
+
+    /* The very last interpolation. */
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], negate( w0 ) );    
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], interp[ 3 ] );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], w0 );
+
+    /* scale by pow( 2, -15 ), as described above */
+    brw_MUL( p, param[ 0 ], tmp[ 0 ], brw_imm_f( 0.000030517578125 ) );
+
+    release_tmps( c, mark );
+}
+
+static void emit_noise4( struct brw_wm_compile *c,
+			 struct prog_instruction *inst )
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, src2, src3, param0, param1, param2, param3, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    int mark = mark_tmps( c );
+
+    assert( mark == 0 );
+    
+    src0 = get_src_reg( c, inst->SrcReg, 0, 1 );
+    src1 = get_src_reg( c, inst->SrcReg, 1, 1 );
+    src2 = get_src_reg( c, inst->SrcReg, 2, 1 );
+    src3 = get_src_reg( c, inst->SrcReg, 3, 1 );
+
+    param0 = alloc_tmp( c );
+    param1 = alloc_tmp( c );
+    param2 = alloc_tmp( c );
+    param3 = alloc_tmp( c );
+
+    brw_MOV( p, param0, src0 );
+    brw_MOV( p, param1, src1 );
+    brw_MOV( p, param2, src2 );
+    brw_MOV( p, param3, src3 );
+
+    invoke_subroutine( c, SUB_NOISE4, noise4_sub );
+    
+    /* Fill in the result: */
+    brw_set_saturate( p, inst->SaturateMode == SATURATE_ZERO_ONE );
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV( p, dst, param0 );
+	}
+    }
+    if( inst->SaturateMode == SATURATE_ZERO_ONE )
+	brw_set_saturate( p, 0 );
+    
+    release_tmps( c, mark );
+}
+    
+static void emit_wpos_xy(struct brw_wm_compile *c,
+                struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    GLuint mask = inst->DstReg.WriteMask;
+    struct brw_reg src0[2], dst[2];
+
+    dst[0] = get_dst_reg(c, inst, 0, 1);
+    dst[1] = get_dst_reg(c, inst, 1, 1);
+
+    src0[0] = get_src_reg(c, &inst->SrcReg[0], 0, 1);
+    src0[1] = get_src_reg(c, &inst->SrcReg[0], 1, 1);
+
+    /* Calculate the pixel offset from window bottom left into destination
+     * X and Y channels.
+     */
+    if (mask & WRITEMASK_X) {
+	/* X' = X - origin_x */
+	brw_ADD(p,
+		dst[0],
+		retype(src0[0], BRW_REGISTER_TYPE_W),
+		brw_imm_d(0 - c->key.origin_x));
+    }
+
+    if (mask & WRITEMASK_Y) {
+	/* Y' = height - (Y - origin_y) = height + origin_y - Y */
+	brw_ADD(p,
+		dst[1],
+		negate(retype(src0[1], BRW_REGISTER_TYPE_W)),
+		brw_imm_d(c->key.origin_y + c->key.drawable_height - 1));
+    }
+}
+
+/* TODO
+   BIAS on SIMD8 not workind yet...
+ */	
+static void emit_txb(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg dst[4], src[4], payload_reg;
+    GLuint unit = c->fp->program.Base.SamplerUnits[inst->TexSrcUnit];
+
+    GLuint i;
+    payload_reg = get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH, 0, 1, 0, 0);
+    for (i = 0; i < 4; i++) 
+	dst[i] = get_dst_reg(c, inst, i, 1);
+    for (i = 0; i < 4; i++)
+	src[i] = get_src_reg(c, &inst->SrcReg[0], i, 1);
+
+    switch (inst->TexSrcTarget) {
+	case TEXTURE_1D_INDEX:
+	    brw_MOV(p, brw_message_reg(2), src[0]);
+	    brw_MOV(p, brw_message_reg(3), brw_imm_f(0));
+	    brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
+	    break;
+	case TEXTURE_2D_INDEX:
+	case TEXTURE_RECT_INDEX:
+	    brw_MOV(p, brw_message_reg(2), src[0]);
+	    brw_MOV(p, brw_message_reg(3), src[1]);
+	    brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
+	    break;
+	default:
+	    brw_MOV(p, brw_message_reg(2), src[0]);
+	    brw_MOV(p, brw_message_reg(3), src[1]);
+	    brw_MOV(p, brw_message_reg(4), src[2]);
+	    break;
+    }
+    brw_MOV(p, brw_message_reg(5), src[3]);
+    brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
+    brw_SAMPLE(p,
+	    retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW),
+	    1,
+	    retype(payload_reg, BRW_REGISTER_TYPE_UW),
+	    unit + MAX_DRAW_BUFFERS, /* surface */
+	    unit,     /* sampler */
+	    inst->DstReg.WriteMask,
+	    BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS,
+	    4,
+	    4,
+	    0);
+}
+
+static void emit_tex(struct brw_wm_compile *c,
+		struct prog_instruction *inst)
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg dst[4], src[4], payload_reg;
+    GLuint unit = c->fp->program.Base.SamplerUnits[inst->TexSrcUnit];
+
+    GLuint msg_len;
+    GLuint i, nr;
+    GLuint emit;
+    GLboolean shadow = (c->key.shadowtex_mask & (1<<unit)) ? 1 : 0;
+
+    payload_reg = get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH, 0, 1, 0, 0);
+
+    for (i = 0; i < 4; i++) 
+	dst[i] = get_dst_reg(c, inst, i, 1);
+    for (i = 0; i < 4; i++)
+	src[i] = get_src_reg(c, &inst->SrcReg[0], i, 1);
+
+
+    switch (inst->TexSrcTarget) {
+	case TEXTURE_1D_INDEX:
+	    emit = WRITEMASK_X;
+	    nr = 1;
+	    break;
+	case TEXTURE_2D_INDEX:
+	case TEXTURE_RECT_INDEX:
+	    emit = WRITEMASK_XY;
+	    nr = 2;
+	    break;
+	default:
+	    emit = WRITEMASK_XYZ;
+	    nr = 3;
+	    break;
+    }
+    msg_len = 1;
+
+    for (i = 0; i < nr; i++) {
+	static const GLuint swz[4] = {0,1,2,2};
+	if (emit & (1<<i))
+	    brw_MOV(p, brw_message_reg(msg_len+1), src[swz[i]]);
+	else
+	    brw_MOV(p, brw_message_reg(msg_len+1), brw_imm_f(0));
+	msg_len += 1;
+    }
+
+    if (shadow) {
+	brw_MOV(p, brw_message_reg(5), brw_imm_f(0));
+	brw_MOV(p, brw_message_reg(6), src[2]);
+    }
+
+    brw_SAMPLE(p,
+	    retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW),
+	    1,
+	    retype(payload_reg, BRW_REGISTER_TYPE_UW),
+	    unit + MAX_DRAW_BUFFERS, /* surface */
+	    unit,     /* sampler */
+	    inst->DstReg.WriteMask,
+	    BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE,
+	    4,
+	    shadow ? 6 : 4,
+	    0);
+
+    if (shadow)
+	brw_MOV(p, dst[3], brw_imm_f(1.0));
+}
+
+static void post_wm_emit( struct brw_wm_compile *c )
+{
+    GLuint nr_insns = c->fp->program.Base.NumInstructions;
+    GLuint insn, target_insn;
+    struct prog_instruction *inst1, *inst2;
+    struct brw_instruction *brw_inst1, *brw_inst2;
+    int offset;
+    for (insn = 0; insn < nr_insns; insn++) {
+	inst1 = &c->fp->program.Base.Instructions[insn];
+	brw_inst1 = inst1->Data;
+	switch (inst1->Opcode) {
+	    case OPCODE_CAL:
+		target_insn = inst1->BranchTarget;
+		inst2 = &c->fp->program.Base.Instructions[target_insn];
+		brw_inst2 = inst2->Data;
+		offset = brw_inst2 - brw_inst1;
+		brw_set_src1(brw_inst1, brw_imm_d(offset*16));
+		break;
+	    default:
+		break;
+	}
+    }
+}
+
+static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
+{
+#define MAX_IFSN 32
+#define MAX_LOOP_DEPTH 32
+    struct brw_instruction *if_inst[MAX_IFSN], *loop_inst[MAX_LOOP_DEPTH];
+    struct brw_instruction *inst0, *inst1;
+    int i, if_insn = 0, loop_insn = 0;
+    struct brw_compile *p = &c->func;
+    struct brw_indirect stack_index = brw_indirect(0, 0);
+
+    c->reg_index = 0;
+    prealloc_reg(c);
+    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
+    brw_MOV(p, get_addr_reg(stack_index), brw_address(c->stack));
+
+    for (i = 0; i < c->nr_fp_insns; i++) {
+	struct prog_instruction *inst = &c->prog_instructions[i];
+	struct prog_instruction *orig_inst;
+
+	if ((orig_inst = inst->Data) != 0)
+	    orig_inst->Data = current_insn(p);
+
+	if (inst->CondUpdate)
+	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+	else
+	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
+
+	switch (inst->Opcode) {
+	    case WM_PIXELXY:
+		emit_pixel_xy(c, inst);
+		break;
+	    case WM_DELTAXY: 
+		emit_delta_xy(c, inst);
+		break;
+	    case WM_PIXELW:
+		emit_pixel_w(c, inst);
+		break;	
+	    case WM_LINTERP:
+		emit_linterp(c, inst);
+		break;
+	    case WM_PINTERP:
+		emit_pinterp(c, inst);
+		break;
+	    case WM_CINTERP:
+		emit_cinterp(c, inst);
+		break;
+	    case WM_WPOSXY:
+		emit_wpos_xy(c, inst);
+		break;
+	    case WM_FB_WRITE:
+		emit_fb_write(c, inst);
+		break;
+	    case OPCODE_ABS:
+		emit_abs(c, inst);
+		break;
+	    case OPCODE_ADD:
+		emit_add(c, inst);
+		break;
+	    case OPCODE_SUB:
+		emit_sub(c, inst);
+		break;
+	    case OPCODE_FRC:
+		emit_frc(c, inst);
+		break;
+	    case OPCODE_FLR:
+		emit_flr(c, inst);
+		break;
+	    case OPCODE_LRP:
+		emit_lrp(c, inst);
+		break;
+	    case OPCODE_TRUNC:
+		emit_trunc(c, inst);
+		break;
+	    case OPCODE_MOV:
+		emit_mov(c, inst);
+		break;
+	    case OPCODE_DP3:
+		emit_dp3(c, inst);
+		break;
+	    case OPCODE_DP4:
+		emit_dp4(c, inst);
+		break;
+	    case OPCODE_XPD:
+		emit_xpd(c, inst);
+		break;
+	    case OPCODE_DPH:
+		emit_dph(c, inst);
+		break;
+	    case OPCODE_RCP:
+		emit_rcp(c, inst);
+		break;
+	    case OPCODE_RSQ:
+		emit_rsq(c, inst);
+		break;
+	    case OPCODE_SIN:
+		emit_sin(c, inst);
+		break;
+	    case OPCODE_COS:
+		emit_cos(c, inst);
+		break;
+	    case OPCODE_EX2:
+		emit_ex2(c, inst);
+		break;
+	    case OPCODE_LG2:
+		emit_lg2(c, inst);
+		break;
+	    case OPCODE_MAX:	
+		emit_max(c, inst);
+		break;
+	    case OPCODE_MIN:	
+		emit_min(c, inst);
+		break;
+	    case OPCODE_DDX:
+		emit_ddx(c, inst);
+		break;
+	    case OPCODE_DDY:
+                emit_ddy(c, inst);
+                break;
+	    case OPCODE_SLT:
+		emit_slt(c, inst);
+		break;
+	    case OPCODE_SLE:
+		emit_sle(c, inst);
+		break;
+	    case OPCODE_SGT:
+		emit_sgt(c, inst);
+		break;
+	    case OPCODE_SGE:
+		emit_sge(c, inst);
+		break;
+	    case OPCODE_SEQ:
+		emit_seq(c, inst);
+		break;
+	    case OPCODE_SNE:
+		emit_sne(c, inst);
+		break;
+	    case OPCODE_MUL:
+		emit_mul(c, inst);
+		break;
+	    case OPCODE_POW:
+		emit_pow(c, inst);
+		break;
+	    case OPCODE_MAD:
+		emit_mad(c, inst);
+		break;
+	    case OPCODE_NOISE1:
+		emit_noise1(c, inst);
+		break;
+	    case OPCODE_NOISE2:
+		emit_noise2(c, inst);
+		break;
+	    case OPCODE_NOISE3:
+		emit_noise3(c, inst);
+		break;
+	    case OPCODE_NOISE4:
+		emit_noise4(c, inst);
+		break;
+	    case OPCODE_TEX:
+		emit_tex(c, inst);
+		break;
+	    case OPCODE_TXB:
+		emit_txb(c, inst);
+		break;
+	    case OPCODE_KIL_NV:
+		emit_kil(c);
+		break;
+	    case OPCODE_IF:
+		assert(if_insn < MAX_IFSN);
+		if_inst[if_insn++] = brw_IF(p, BRW_EXECUTE_8);
+		break;
+	    case OPCODE_ELSE:
+		if_inst[if_insn-1]  = brw_ELSE(p, if_inst[if_insn-1]);
+		break;
+	    case OPCODE_ENDIF:
+		assert(if_insn > 0);
+		brw_ENDIF(p, if_inst[--if_insn]);
+		break;
+	    case OPCODE_BGNSUB:
+	    case OPCODE_ENDSUB:
+		break;
+	    case OPCODE_CAL: 
+		brw_push_insn_state(p);
+		brw_set_mask_control(p, BRW_MASK_DISABLE);
+                brw_set_access_mode(p, BRW_ALIGN_1);
+                brw_ADD(p, deref_1ud(stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
+                brw_set_access_mode(p, BRW_ALIGN_16);
+                brw_ADD(p, get_addr_reg(stack_index),
+                         get_addr_reg(stack_index), brw_imm_d(4));
+                orig_inst = inst->Data;
+                orig_inst->Data = &p->store[p->nr_insn];
+                brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
+                brw_pop_insn_state(p);
+		break;
+
+	    case OPCODE_RET:
+		brw_push_insn_state(p);
+		brw_set_mask_control(p, BRW_MASK_DISABLE);
+                brw_ADD(p, get_addr_reg(stack_index),
+                        get_addr_reg(stack_index), brw_imm_d(-4));
+                brw_set_access_mode(p, BRW_ALIGN_1);
+                brw_MOV(p, brw_ip_reg(), deref_1ud(stack_index, 0));
+                brw_set_access_mode(p, BRW_ALIGN_16);
+		brw_pop_insn_state(p);
+
+		break;
+	    case OPCODE_BGNLOOP:
+		loop_inst[loop_insn++] = brw_DO(p, BRW_EXECUTE_8);
+		break;
+	    case OPCODE_BRK:
+		brw_BREAK(p);
+		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+		break;
+	    case OPCODE_CONT:
+		brw_CONT(p);
+		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+		break;
+	    case OPCODE_ENDLOOP: 
+		loop_insn--;
+		inst0 = inst1 = brw_WHILE(p, loop_inst[loop_insn]);
+		/* patch all the BREAK instructions from
+		   last BEGINLOOP */
+		while (inst0 > loop_inst[loop_insn]) {
+		    inst0--;
+		    if (inst0->header.opcode == BRW_OPCODE_BREAK) {
+			inst0->bits3.if_else.jump_count = inst1 - inst0 + 1;
+			inst0->bits3.if_else.pop_count = 0;
+		    } else if (inst0->header.opcode == BRW_OPCODE_CONTINUE) {
+                        inst0->bits3.if_else.jump_count = inst1 - inst0;
+                        inst0->bits3.if_else.pop_count = 0;
+                    }
+		}
+		break;
+	    default:
+		_mesa_printf("unsupported IR in fragment shader %d\n",
+			inst->Opcode);
+	}
+	if (inst->CondUpdate)
+	    brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
+	else
+	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    }
+    post_wm_emit(c);
+    for (i = 0; i < c->fp->program.Base.NumInstructions; i++)
+	c->fp->program.Base.Instructions[i].Data = NULL;
+}
+
+void brw_wm_glsl_emit(struct brw_context *brw, struct brw_wm_compile *c)
+{
+    brw_wm_pass_fp(c);
+    brw_wm_emit_glsl(brw, c);
+    c->prog_data.total_grf = c->reg_index;
+    c->prog_data.total_scratch = 0;
+}
diff --git a/src/mesa/drivers/dri/i965/brw_wm_iz.c b/src/mesa/drivers/dri/i965/brw_wm_iz.c
index ec2b976faa7..bd60ac9b315 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_iz.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_iz.c
@@ -30,7 +30,7 @@
   */
                 
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "brw_wm.h"
 
 
@@ -52,70 +52,6 @@ const struct {
 {
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
index 00f6f6b9a4f..205a7160d39 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
@@ -168,6 +168,7 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
       case PROGRAM_PAYLOAD:
       case PROGRAM_TEMPORARY:
       case PROGRAM_OUTPUT:
+      case PROGRAM_VARYING:
 	 break;
 
       case PROGRAM_LOCAL_PARAM:
@@ -179,6 +180,8 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
 	 break;
 
       case PROGRAM_STATE_VAR:
+      case PROGRAM_UNIFORM:
+      case PROGRAM_CONSTANT:
       case PROGRAM_NAMED_PARAM: {
 	 struct gl_program_parameter_list *plist = c->fp->program.Base.Parameters;
 	 
@@ -197,6 +200,7 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
 	    break;
 	    
 	 case PROGRAM_STATE_VAR:
+	 case PROGRAM_UNIFORM:
 	    /* These may change from run to run:
 	     */
 	    ref = get_param_ref(c, &plist->ParameterValues[idx][component] );
@@ -344,6 +348,8 @@ static struct brw_wm_instruction *translate_insn( struct brw_wm_compile *c,
    out->saturate = (inst->SaturateMode != SATURATE_OFF);
    out->tex_unit = inst->TexSrcUnit;
    out->tex_idx = inst->TexSrcTarget;
+   out->eot = inst->Sampler & 1;
+   out->target = inst->Sampler>>1;
 
    /* Args:
     */
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
index d668def7007..f6f3a38e9e0 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
@@ -150,12 +150,17 @@ void brw_wm_pass1( struct brw_wm_compile *c )
       case OPCODE_FLR:
       case OPCODE_FRC:
       case OPCODE_MOV:
+      case OPCODE_SWZ:
 	 read0 = writemask;
 	 break;
 
       case OPCODE_SUB:
       case OPCODE_SLT:
+      case OPCODE_SLE:
       case OPCODE_SGE:
+      case OPCODE_SGT:
+      case OPCODE_SEQ:
+      case OPCODE_SNE:
       case OPCODE_ADD:
       case OPCODE_MAX:
       case OPCODE_MIN:
@@ -253,11 +258,9 @@ void brw_wm_pass1( struct brw_wm_compile *c )
 	 read0 = WRITEMASK_XYW;
 	 break;
 
-      case OPCODE_SWZ:
       case OPCODE_DST:
       case OPCODE_TXP:
       default:
-	 assert(0);
 	 break;
       }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass2.c b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
index a1edbd6168d..6fca9ad220a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass2.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
@@ -69,7 +69,8 @@ static void prealloc_reg(struct brw_wm_compile *c,
  */
 static void init_registers( struct brw_wm_compile *c )
 {
-   GLuint inputs = FRAG_BIT_WPOS | c->fp_interp_emitted;
+   struct brw_context *brw = c->func.brw;
+   GLuint inputs = (brw->vs.prog_data->outputs_written & DO_SETUP_BITS);
    GLuint nr_interp_regs = 0;
    GLuint i = 0;
    GLuint j;
@@ -85,8 +86,15 @@ static void init_registers( struct brw_wm_compile *c )
 
    for (j = 0; j < FRAG_ATTRIB_MAX; j++) 
       if (inputs & (1<<j)) {
+	 /* index for vs output and ps input are not the same 
+	    in shader varying */
+	 GLuint index;
+	 if (j > FRAG_ATTRIB_VAR0)
+	     index = j - (VERT_RESULT_VAR0 - FRAG_ATTRIB_VAR0);
+	 else
+	     index = j;
 	 nr_interp_regs++;
-	 prealloc_reg(c, &c->payload.input_interp[j], i++);
+	 prealloc_reg(c, &c->payload.input_interp[index], i++);
       }
 
    assert(nr_interp_regs >= 1);
@@ -328,7 +336,7 @@ void brw_wm_pass2( struct brw_wm_compile *c )
    c->state = PASS2_DONE;
 
    if (INTEL_DEBUG & DEBUG_WM) {
-      brw_wm_print_program(c, "pass2/done");
+       brw_wm_print_program(c, "pass2/done");
    }
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
index 93d4cfc3a5f..f12ef47a7d7 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
@@ -34,7 +34,7 @@
 #include "brw_state.h"
 #include "brw_defines.h"
 
-#include "macros.h"
+#include "main/macros.h"
 
 
 
@@ -54,7 +54,7 @@ static GLuint translate_wrap_mode( GLenum wrap )
    case GL_REPEAT: 
       return BRW_TEXCOORDMODE_WRAP;
    case GL_CLAMP:  
-      return BRW_TEXCOORDMODE_CLAMP_BORDER; /* conform likes it this way */
+      return BRW_TEXCOORDMODE_CLAMP;
    case GL_CLAMP_TO_EDGE: 
       return BRW_TEXCOORDMODE_CLAMP; /* conform likes it this way */
    case GL_CLAMP_TO_BORDER: 
@@ -79,27 +79,43 @@ static GLint S_FIXED(GLfloat value, GLuint frac_bits)
 }
 
 
-static GLuint upload_default_color( struct brw_context *brw,
-				    const GLfloat *color )
+static dri_bo *upload_default_color( struct brw_context *brw,
+				     const GLfloat *color )
 {
    struct brw_sampler_default_color sdc;
 
    COPY_4V(sdc.color, color); 
    
-   return brw_cache_data( &brw->cache[BRW_SAMPLER_DEFAULT_COLOR], &sdc );
+   return brw_cache_data( &brw->cache, BRW_SAMPLER_DEFAULT_COLOR, &sdc,
+			  NULL, 0 );
 }
 
 
-/*
+struct wm_sampler_key {
+   int sampler_count;
+
+   struct wm_sampler_entry {
+      GLenum wrap_r, wrap_s, wrap_t;
+      float maxlod, minlod;
+      float lod_bias;
+      float max_aniso;
+      GLenum minfilter, magfilter;
+      GLenum comparemode, comparefunc;
+      dri_bo *sdc_bo;
+   } sampler[BRW_MAX_TEX_UNIT];
+};
+
+/**
+ * Sets the sampler state for a single unit based off of the sampler key
+ * entry.
  */
-static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
-				      struct gl_texture_object *texObj,
-				      GLuint sdc_gs_offset,
-				      struct brw_sampler_state *sampler)
-{   
+static void brw_update_sampler_state(struct wm_sampler_entry *key,
+				     dri_bo *sdc_bo,
+				     struct brw_sampler_state *sampler)
+{
    _mesa_memset(sampler, 0, sizeof(*sampler));
 
-   switch (texObj->MinFilter) {
+   switch (key->minfilter) {
    case GL_NEAREST:
       sampler->ss0.min_filter = BRW_MAPFILTER_NEAREST;
       sampler->ss0.mip_filter = BRW_MIPFILTER_NONE;
@@ -130,17 +146,17 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
 
    /* Set Anisotropy: 
     */
-   if ( texObj->MaxAnisotropy > 1.0 ) {
+   if (key->max_aniso > 1.0) {
       sampler->ss0.min_filter = BRW_MAPFILTER_ANISOTROPIC; 
       sampler->ss0.mag_filter = BRW_MAPFILTER_ANISOTROPIC;
 
-      if (texObj->MaxAnisotropy > 2.0) {
-	 sampler->ss3.max_aniso = MAX2((texObj->MaxAnisotropy - 2) / 2,
+      if (key->max_aniso > 2.0) {
+	 sampler->ss3.max_aniso = MAX2((key->max_aniso - 2) / 2,
 				       BRW_ANISORATIO_16);
       }
    }
    else {
-      switch (texObj->MagFilter) {
+      switch (key->magfilter) {
       case GL_NEAREST:
 	 sampler->ss0.mag_filter = BRW_MAPFILTER_NEAREST;
 	 break;
@@ -152,9 +168,9 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
       }  
    }
 
-   sampler->ss1.r_wrap_mode = translate_wrap_mode(texObj->WrapR);
-   sampler->ss1.s_wrap_mode = translate_wrap_mode(texObj->WrapS);
-   sampler->ss1.t_wrap_mode = translate_wrap_mode(texObj->WrapT);
+   sampler->ss1.r_wrap_mode = translate_wrap_mode(key->wrap_r);
+   sampler->ss1.s_wrap_mode = translate_wrap_mode(key->wrap_s);
+   sampler->ss1.t_wrap_mode = translate_wrap_mode(key->wrap_t);
 
    /* Fulsim complains if I don't do this.  Hardware doesn't mind:
     */
@@ -168,17 +184,18 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
 
    /* Set shadow function: 
     */
-   if (texObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) {
+   if (key->comparemode == GL_COMPARE_R_TO_TEXTURE_ARB) {
       /* Shadowing is "enabled" by emitting a particular sampler
        * message (sample_c).  So need to recompile WM program when
        * shadow comparison is enabled on each/any texture unit.
        */
-      sampler->ss0.shadow_function = intel_translate_compare_func(texObj->CompareFunc);
+      sampler->ss0.shadow_function =
+	 intel_translate_shadow_compare_func(key->comparefunc);
    }
 
    /* Set LOD bias: 
     */
-   sampler->ss0.lod_bias = S_FIXED(texUnit->LodBias + texObj->LodBias, 6);
+   sampler->ss0.lod_bias = S_FIXED(CLAMP(key->lod_bias, -16, 15), 6);
 
    sampler->ss0.lod_preclamp = 1; /* OpenGL mode */
    sampler->ss0.default_color_mode = 0; /* OpenGL/DX10 mode */
@@ -192,13 +209,64 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
     */
    sampler->ss0.base_level = U_FIXED(0, 1);
 
-   sampler->ss1.max_lod = U_FIXED(MAX2(texObj->MaxLod, 0), 6);
-   sampler->ss1.min_lod = U_FIXED(MAX2(texObj->MinLod, 0), 6);
+   sampler->ss1.max_lod = U_FIXED(MIN2(MAX2(key->maxlod, 0), 13), 6);
+   sampler->ss1.min_lod = U_FIXED(MIN2(MAX2(key->minlod, 0), 13), 6);
    
-   sampler->ss2.default_color_pointer = sdc_gs_offset >> 5;
+   sampler->ss2.default_color_pointer = sdc_bo->offset >> 5; /* reloc */
 }
 
+/** Sets up the cache key for sampler state for all texture units */
+static void
+brw_wm_sampler_populate_key(struct brw_context *brw,
+			    struct wm_sampler_key *key)
+{
+   int unit;
 
+   memset(key, 0, sizeof(*key));
+
+   for (unit = 0; unit < BRW_MAX_TEX_UNIT; unit++) {
+      if (brw->attribs.Texture->Unit[unit]._ReallyEnabled) {
+	 struct wm_sampler_entry *entry = &key->sampler[unit];
+	 struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[unit];
+	 struct gl_texture_object *texObj = texUnit->_Current;
+	 struct intel_texture_object *intelObj = intel_texture_object(texObj);
+	 struct gl_texture_image *firstImage =
+	    texObj->Image[0][intelObj->firstLevel];
+
+	 entry->wrap_r = texObj->WrapR;
+	 entry->wrap_s = texObj->WrapS;
+	 entry->wrap_t = texObj->WrapT;
+
+	 entry->maxlod = texObj->MaxLod;
+	 entry->minlod = texObj->MinLod;
+	 entry->lod_bias = texUnit->LodBias + texObj->LodBias;
+	 entry->max_aniso = texObj->MaxAnisotropy;
+	 entry->minfilter = texObj->MinFilter;
+	 entry->magfilter = texObj->MagFilter;
+	 entry->comparemode = texObj->CompareMode;
+    entry->comparefunc = texObj->CompareFunc;
+
+	 dri_bo_unreference(brw->wm.sdc_bo[unit]);
+	 if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
+	    float bordercolor[4] = {
+	       texObj->BorderColor[0],
+	       texObj->BorderColor[0],
+	       texObj->BorderColor[0],
+	       texObj->BorderColor[0]
+	    };
+	    /* GL specs that border color for depth textures is taken from the
+	     * R channel, while the hardware uses A.  Spam R into all the
+	     * channels for safety.
+	     */
+	    brw->wm.sdc_bo[unit] = upload_default_color(brw, bordercolor);
+	 } else {
+	    brw->wm.sdc_bo[unit] = upload_default_color(brw,
+							texObj->BorderColor);
+	 }
+	 key->sampler_count = unit + 1;
+      }
+   }
+}
 
 /* All samplers must be uploaded in a single contiguous array, which
  * complicates various things.  However, this is still too confusing -
@@ -206,40 +274,61 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
  */
 static void upload_wm_samplers( struct brw_context *brw )
 {
-   GLuint unit;
-   GLuint sampler_count = 0;
-
-   /* _NEW_TEXTURE */
-   for (unit = 0; unit < BRW_MAX_TEX_UNIT; unit++) {
-      if (brw->attribs.Texture->Unit[unit]._ReallyEnabled) {	 
-	 struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[unit];
-	 struct gl_texture_object *texObj = texUnit->_Current;
+   struct wm_sampler_key key;
+   int i;
 
-	 GLuint sdc_gs_offset = upload_default_color(brw, texObj->BorderColor);
+   brw_wm_sampler_populate_key(brw, &key);
 
-	 brw_update_sampler_state(texUnit,
-				  texObj, 
-				  sdc_gs_offset,
-				  &brw->wm.sampler[unit]);
-
-	 sampler_count = unit + 1;
-      }
-   }
-   
-   if (brw->wm.sampler_count != sampler_count) {
-      brw->wm.sampler_count = sampler_count;
+   if (brw->wm.sampler_count != key.sampler_count) {
+      brw->wm.sampler_count = key.sampler_count;
       brw->state.dirty.cache |= CACHE_NEW_SAMPLER;
    }
 
-   brw->wm.sampler_gs_offset = 0;
+   dri_bo_unreference(brw->wm.sampler_bo);
+   brw->wm.sampler_bo = NULL;
+   if (brw->wm.sampler_count == 0)
+      return;
 
-   if (brw->wm.sampler_count) 
-      brw->wm.sampler_gs_offset = 
-	 brw_cache_data_sz(&brw->cache[BRW_SAMPLER],
-			   brw->wm.sampler,
-			   sizeof(struct brw_sampler_state) * brw->wm.sampler_count);
-}
+   brw->wm.sampler_bo = brw_search_cache(&brw->cache, BRW_SAMPLER,
+					 &key, sizeof(key),
+					 brw->wm.sdc_bo, key.sampler_count,
+					 NULL);
 
+   /* If we didnt find it in the cache, compute the state and put it in the
+    * cache.
+    */
+   if (brw->wm.sampler_bo == NULL) {
+      struct brw_sampler_state sampler[BRW_MAX_TEX_UNIT];
+
+      memset(sampler, 0, sizeof(sampler));
+      for (i = 0; i < key.sampler_count; i++) {
+	 if (brw->wm.sdc_bo[i] == NULL)
+	    continue;
+
+	 brw_update_sampler_state(&key.sampler[i], brw->wm.sdc_bo[i],
+				  &sampler[i]);
+      }
+
+      brw->wm.sampler_bo = brw_upload_cache(&brw->cache, BRW_SAMPLER,
+					    &key, sizeof(key),
+					    brw->wm.sdc_bo, key.sampler_count,
+					    &sampler, sizeof(sampler),
+					    NULL, NULL);
+
+      /* Emit SDC relocations */
+      for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
+	 if (!brw->attribs.Texture->Unit[i]._ReallyEnabled)
+	    continue;
+
+	 dri_bo_emit_reloc(brw->wm.sampler_bo,
+			   I915_GEM_DOMAIN_SAMPLER, 0,
+			   0,
+			   i * sizeof(struct brw_sampler_state) +
+			   offsetof(struct brw_sampler_state, ss2),
+			   brw->wm.sdc_bo[i]);
+      }
+   }
+}
 
 const struct brw_tracked_state brw_wm_samplers = {
    .dirty = {
@@ -247,7 +336,7 @@ const struct brw_tracked_state brw_wm_samplers = {
       .brw = 0,
       .cache = 0
    },
-   .update = upload_wm_samplers
+   .prepare = upload_wm_samplers,
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index ff5cb31bdd1..5302405eda1 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -34,109 +34,140 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "brw_defines.h"
-#include "bufmgr.h"
+#include "brw_wm.h"
 
 /***********************************************************************
  * WM unit - fragment programs and rasterization
  */
 
-static void invalidate_scratch_cb( struct intel_context *intel,
-				   void *unused )
-{
-   /* nothing */
-}
+struct brw_wm_unit_key {
+   unsigned int total_grf, total_scratch;
+   unsigned int urb_entry_read_length;
+   unsigned int curb_entry_read_length;
+   unsigned int dispatch_grf_start_reg;
+
+   unsigned int curbe_offset;
+   unsigned int urb_size;
+
+   unsigned int max_threads;
 
+   unsigned int nr_surfaces, sampler_count;
+   GLboolean uses_depth, computes_depth, uses_kill, is_glsl;
+   GLboolean polygon_stipple, stats_wm, line_stipple, offset_enable;
+   GLfloat offset_units, offset_factor;
+};
 
-static void upload_wm_unit(struct brw_context *brw )
+static void
+wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
 {
+   const struct gl_fragment_program *fp = brw->fragment_program;
    struct intel_context *intel = &brw->intel;
-   struct brw_wm_unit_state wm;
-   GLuint max_threads;
+
+   memset(key, 0, sizeof(*key));
 
    if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
-      max_threads = 0; 
-   else
-      max_threads = 31;
+      key->max_threads = 1;
+   else {
+      /* WM maximum threads is number of EUs times number of threads per EU. */
+      if (BRW_IS_G4X(brw))
+	 key->max_threads = 10 * 5;
+      else
+	 key->max_threads = 8 * 4;
+   }
 
+   /* CACHE_NEW_WM_PROG */
+   key->total_grf = brw->wm.prog_data->total_grf;
+   key->urb_entry_read_length = brw->wm.prog_data->urb_read_length;
+   key->curb_entry_read_length = brw->wm.prog_data->curb_read_length;
+   key->dispatch_grf_start_reg = brw->wm.prog_data->first_curbe_grf;
+   key->total_scratch = ALIGN(brw->wm.prog_data->total_scratch, 1024);
 
-   memset(&wm, 0, sizeof(wm));
+   /* BRW_NEW_URB_FENCE */
+   key->urb_size = brw->urb.vsize;
 
-   /* CACHE_NEW_WM_PROG */
-   wm.thread0.grf_reg_count = ((brw->wm.prog_data->total_grf-1) & ~15) / 16;
-   wm.thread0.kernel_start_pointer = brw->wm.prog_gs_offset >> 6;
-   wm.thread3.dispatch_grf_start_reg = brw->wm.prog_data->first_curbe_grf;
-   wm.thread3.urb_entry_read_length = brw->wm.prog_data->urb_read_length;
-   wm.thread3.const_urb_entry_read_length = brw->wm.prog_data->curb_read_length;
-
-   wm.wm5.max_threads = max_threads;      
-
-   if (brw->wm.prog_data->total_scratch) {
-      GLuint per_thread = (brw->wm.prog_data->total_scratch + 1023) / 1024;
-      GLuint total = per_thread * (max_threads + 1);
-
-      /* Scratch space -- just have to make sure there is sufficient
-       * allocated for the active program and current number of threads.
-       */      
-
-      if (!brw->wm.scratch_buffer) {
-	 bmGenBuffers(intel, "wm scratch", 1, &brw->wm.scratch_buffer, 12);
-	 bmBufferSetInvalidateCB(intel,
-				 brw->wm.scratch_buffer,
-				 invalidate_scratch_cb,
-				 NULL,
-				 GL_FALSE);
-      }
+   /* BRW_NEW_CURBE_OFFSETS */
+   key->curbe_offset = brw->curbe.wm_start;
 
-      if (total > brw->wm.scratch_buffer_size) {
-	 brw->wm.scratch_buffer_size = total;
-	 bmBufferData(intel,
-		      brw->wm.scratch_buffer,
-		      brw->wm.scratch_buffer_size,
-		      NULL,
-		      0);
-      }
-		   
-      assert(per_thread <= 12 * 1024);
-      wm.thread2.per_thread_scratch_space = (per_thread / 1024) - 1;
+   /* BRW_NEW_NR_SURFACEs */
+   key->nr_surfaces = brw->wm.nr_surfaces;
 
-      /* XXX: could make this dynamic as this is so rarely active:
-       */
-      /* BRW_NEW_LOCK */
-      wm.thread2.scratch_space_base_pointer = 
-	 bmBufferOffset(intel, brw->wm.scratch_buffer) >> 10;
-   }
+   /* CACHE_NEW_SAMPLER */
+   key->sampler_count = brw->wm.sampler_count;
+
+   /* _NEW_POLYGONSTIPPLE */
+   key->polygon_stipple = brw->attribs.Polygon->StippleFlag;
 
-   /* CACHE_NEW_SURFACE */
-   wm.thread1.binding_table_entry_count = brw->wm.nr_surfaces;
+   /* BRW_NEW_FRAGMENT_PROGRAM */
+   key->uses_depth = (fp->Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) != 0;
 
-   /* BRW_NEW_CURBE_OFFSETS */
-   wm.thread3.const_urb_entry_read_offset = brw->curbe.wm_start * 2;
+   /* as far as we can tell */
+   key->computes_depth =
+      (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) != 0;
 
-   wm.thread3.urb_entry_read_offset = 0;
+   /* _NEW_COLOR */
+   key->uses_kill = fp->UsesKill || brw->attribs.Color->AlphaEnabled;
+   key->is_glsl = brw_wm_is_glsl(fp);
+
+   /* XXX: This needs a flag to indicate when it changes. */
+   key->stats_wm = intel->stats_wm;
+
+   /* _NEW_LINE */
+   key->line_stipple = brw->attribs.Line->StippleFlag;
+
+   /* _NEW_POLYGON */
+   key->offset_enable = brw->attribs.Polygon->OffsetFill;
+   key->offset_units = brw->attribs.Polygon->OffsetUnits;
+   key->offset_factor = brw->attribs.Polygon->OffsetFactor;
+}
+
+static dri_bo *
+wm_unit_create_from_key(struct brw_context *brw, struct brw_wm_unit_key *key,
+			dri_bo **reloc_bufs)
+{
+   struct brw_wm_unit_state wm;
+   dri_bo *bo;
+
+   memset(&wm, 0, sizeof(wm));
+
+   wm.thread0.grf_reg_count = ALIGN(key->total_grf, 16) / 16 - 1;
+   wm.thread0.kernel_start_pointer = brw->wm.prog_bo->offset >> 6; /* reloc */
    wm.thread1.depth_coef_urb_read_offset = 1;
    wm.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
+   wm.thread1.binding_table_entry_count = key->nr_surfaces;
+
+   if (key->total_scratch != 0) {
+      wm.thread2.scratch_space_base_pointer =
+	 brw->wm.scratch_buffer->offset >> 10; /* reloc */
+      wm.thread2.per_thread_scratch_space = key->total_scratch / 1024 - 1;
+   } else {
+      wm.thread2.scratch_space_base_pointer = 0;
+      wm.thread2.per_thread_scratch_space = 0;
+   }
 
-   /* CACHE_NEW_SAMPLER */
-   wm.wm4.sampler_count = brw->wm.sampler_count;
-   wm.wm4.sampler_state_pointer = brw->wm.sampler_gs_offset >> 5;
+   wm.thread3.dispatch_grf_start_reg = key->dispatch_grf_start_reg;
+   wm.thread3.urb_entry_read_length = key->urb_entry_read_length;
+   wm.thread3.const_urb_entry_read_length = key->curb_entry_read_length;
+   wm.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;
+   wm.thread3.urb_entry_read_offset = 0;
 
-   /* BRW_NEW_FRAGMENT_PROGRAM */
-   {
-      const struct gl_fragment_program *fp = brw->fragment_program; 
-
-      if (fp->Base.InputsRead & (1<<FRAG_ATTRIB_WPOS)) 
-	 wm.wm5.program_uses_depth = 1; /* as far as we can tell */
-   
-      if (fp->Base.OutputsWritten & (1<<FRAG_RESULT_DEPR)) 
-	 wm.wm5.program_computes_depth = 1;
-   
-      /* _NEW_COLOR */
-      if (fp->UsesKill || 
-	  brw->attribs.Color->AlphaEnabled) 
-	 wm.wm5.program_uses_killpixel = 1; 
+   wm.wm4.sampler_count = (key->sampler_count + 1) / 4;
+   if (brw->wm.sampler_bo != NULL) {
+      /* reloc */
+      wm.wm4.sampler_state_pointer = brw->wm.sampler_bo->offset >> 5;
+   } else {
+      wm.wm4.sampler_state_pointer = 0;
    }
 
-   wm.wm5.enable_16_pix = 1;
+   wm.wm5.program_uses_depth = key->uses_depth;
+   wm.wm5.program_computes_depth = key->computes_depth;
+   wm.wm5.program_uses_killpixel = key->uses_kill;
+
+   if (key->is_glsl)
+      wm.wm5.enable_8_pix = 1;
+   else
+      wm.wm5.enable_16_pix = 1;
+
+   wm.wm5.max_threads = key->max_threads - 1;
    wm.wm5.thread_dispatch_enable = 1;	/* AKA: color_write */
    wm.wm5.legacy_line_rast = 0;
    wm.wm5.legacy_global_depth_bias = 0;
@@ -144,34 +175,101 @@ static void upload_wm_unit(struct brw_context *brw )
    wm.wm5.line_aa_region_width = 0;
    wm.wm5.line_endcap_aa_region_width = 1;
 
-   /* _NEW_POLYGONSTIPPLE */
-   if (brw->attribs.Polygon->StippleFlag) 
-      wm.wm5.polygon_stipple = 1;
+   wm.wm5.polygon_stipple = key->polygon_stipple;
 
-   /* _NEW_POLYGON */
-   if (brw->attribs.Polygon->OffsetFill) {
+   if (key->offset_enable) {
       wm.wm5.depth_offset = 1;
       /* Something wierd going on with legacy_global_depth_bias,
        * offset_constant, scaling and MRD.  This value passes glean
        * but gives some odd results elsewere (eg. the
        * quad-offset-units test).
        */
-      wm.global_depth_offset_constant = brw->attribs.Polygon->OffsetUnits * 2;
+      wm.global_depth_offset_constant = key->offset_units * 2;
 
       /* This is the only value that passes glean:
        */
-      wm.global_depth_offset_scale = brw->attribs.Polygon->OffsetFactor;
+      wm.global_depth_offset_scale = key->offset_factor;
    }
 
-   /* _NEW_LINE */
-   if (brw->attribs.Line->StippleFlag) {
-      wm.wm5.line_stipple = 1;
-   }
+   wm.wm5.line_stipple = key->line_stipple;
 
-   if (INTEL_DEBUG & DEBUG_STATS || intel->stats_wm)
+   if (INTEL_DEBUG & DEBUG_STATS || key->stats_wm)
       wm.wm4.stats_enable = 1;
 
-   brw->wm.state_gs_offset = brw_cache_data( &brw->cache[BRW_WM_UNIT], &wm );
+   bo = brw_upload_cache(&brw->cache, BRW_WM_UNIT,
+			 key, sizeof(*key),
+			 reloc_bufs, 3,
+			 &wm, sizeof(wm),
+			 NULL, NULL);
+
+   /* Emit WM program relocation */
+   dri_bo_emit_reloc(bo,
+		     I915_GEM_DOMAIN_INSTRUCTION, 0,
+		     wm.thread0.grf_reg_count << 1,
+		     offsetof(struct brw_wm_unit_state, thread0),
+		     brw->wm.prog_bo);
+
+   /* Emit scratch space relocation */
+   if (key->total_scratch != 0) {
+      dri_bo_emit_reloc(bo,
+			0, 0,
+			wm.thread2.per_thread_scratch_space,
+			offsetof(struct brw_wm_unit_state, thread2),
+			brw->wm.scratch_buffer);
+   }
+
+   /* Emit sampler state relocation */
+   if (key->sampler_count != 0) {
+      dri_bo_emit_reloc(bo,
+			I915_GEM_DOMAIN_INSTRUCTION, 0,
+			wm.wm4.stats_enable | (wm.wm4.sampler_count << 2),
+			offsetof(struct brw_wm_unit_state, wm4),
+			brw->wm.sampler_bo);
+   }
+
+   return bo;
+}
+
+
+static void upload_wm_unit( struct brw_context *brw )
+{
+   struct intel_context *intel = &brw->intel;
+   struct brw_wm_unit_key key;
+   dri_bo *reloc_bufs[3];
+   wm_unit_populate_key(brw, &key);
+
+   /* Allocate the necessary scratch space if we haven't already.  Don't
+    * bother reducing the allocation later, since we use scratch so
+    * rarely.
+    */
+   assert(key.total_scratch <= 12 * 1024);
+   if (key.total_scratch) {
+      GLuint total = key.total_scratch * key.max_threads;
+
+      if (brw->wm.scratch_buffer && total > brw->wm.scratch_buffer->size) {
+	 dri_bo_unreference(brw->wm.scratch_buffer);
+	 brw->wm.scratch_buffer = NULL;
+      }
+      if (brw->wm.scratch_buffer == NULL) {
+	 brw->wm.scratch_buffer = dri_bo_alloc(intel->bufmgr,
+					       "wm scratch",
+					       total,
+					       4096);
+      }
+   }
+
+   reloc_bufs[0] = brw->wm.prog_bo;
+   reloc_bufs[1] = brw->wm.scratch_buffer;
+   reloc_bufs[2] = brw->wm.sampler_bo;
+
+   dri_bo_unreference(brw->wm.state_bo);
+   brw->wm.state_bo = brw_search_cache(&brw->cache, BRW_WM_UNIT,
+				       &key, sizeof(key),
+				       reloc_bufs, 3,
+				       NULL);
+   if (brw->wm.state_bo == NULL) {
+      brw->wm.state_bo = wm_unit_create_from_key(brw, &key, reloc_bufs);
+   }
 }
 
 const struct brw_tracked_state brw_wm_unit = {
@@ -183,12 +281,11 @@ const struct brw_tracked_state brw_wm_unit = {
 
       .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
 	      BRW_NEW_CURBE_OFFSETS |
-	      BRW_NEW_LOCK),
+	      BRW_NEW_NR_SURFACES),
 
-      .cache = (CACHE_NEW_SURFACE | 
-		CACHE_NEW_WM_PROG | 
+      .cache = (CACHE_NEW_WM_PROG |
 		CACHE_NEW_SAMPLER)
    },
-   .update = upload_wm_unit
+   .prepare = upload_wm_unit,
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index d24c618a668..63e14cc3900 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -30,9 +30,9 @@
   */
                    
 
-#include "mtypes.h"
-#include "texformat.h"
-#include "texstore.h"
+#include "main/mtypes.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
 
 #include "intel_mipmap_tree.h"
 #include "intel_batchbuffer.h"
@@ -69,7 +69,7 @@ static GLuint translate_tex_target( GLenum target )
 }
 
 
-static GLuint translate_tex_format( GLuint mesa_format )
+static GLuint translate_tex_format( GLuint mesa_format, GLenum depth_mode )
 {
    switch( mesa_format ) {
    case MESA_FORMAT_L8:
@@ -114,11 +114,32 @@ static GLuint translate_tex_format( GLuint mesa_format )
       return BRW_SURFACEFORMAT_FXT1;
 
    case MESA_FORMAT_Z16:
-      return BRW_SURFACEFORMAT_L16_UNORM;
+      if (depth_mode == GL_INTENSITY) 
+	  return BRW_SURFACEFORMAT_I16_UNORM;
+      else if (depth_mode == GL_ALPHA)
+	  return BRW_SURFACEFORMAT_A16_UNORM;
+      else
+	  return BRW_SURFACEFORMAT_L16_UNORM;
 
-   case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-      return BRW_SURFACEFORMAT_DXT1_RGB;
+       return BRW_SURFACEFORMAT_DXT1_RGB;
+
+   case MESA_FORMAT_RGBA_DXT1:
+       return BRW_SURFACEFORMAT_BC1_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT3:
+       return BRW_SURFACEFORMAT_BC2_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT5:
+       return BRW_SURFACEFORMAT_BC3_UNORM;
+
+   case MESA_FORMAT_SRGBA8:
+      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB;
+   case MESA_FORMAT_SRGB_DXT1:
+      return BRW_SURFACEFORMAT_BC1_UNORM_SRGB;
+
+   case MESA_FORMAT_S8_Z24:
+      return BRW_SURFACEFORMAT_I24X8_UNORM;
 
    default:
       assert(0);
@@ -126,142 +147,347 @@ static GLuint translate_tex_format( GLuint mesa_format )
    }
 }
 
-static
-void brw_update_texture_surface( GLcontext *ctx, 
-				 GLuint unit,
-				 struct brw_surface_state *surf )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct brw_context *brw = brw_context(ctx);
-   struct gl_texture_object *tObj = brw->attribs.Texture->Unit[unit]._Current;
-   struct intel_texture_object *intelObj = intel_texture_object(tObj);
-   struct gl_texture_image *firstImage = tObj->Image[0][intelObj->firstLevel];
+struct brw_wm_surface_key {
+   GLenum target, depthmode;
+   dri_bo *bo;
+   GLint format;
+   GLint first_level, last_level;
+   GLint width, height, depth;
+   GLint pitch, cpp;
+   uint32_t tiling;
+   GLuint offset;
+};
 
-   memset(surf, 0, sizeof(*surf));
+static void
+brw_set_surface_tiling(struct brw_surface_state *surf, uint32_t tiling)
+{
+   switch (tiling) {
+   case I915_TILING_NONE:
+      surf->ss3.tiled_surface = 0;
+      surf->ss3.tile_walk = 0;
+      break;
+   case I915_TILING_X:
+      surf->ss3.tiled_surface = 1;
+      surf->ss3.tile_walk = BRW_TILEWALK_XMAJOR;
+      break;
+   case I915_TILING_Y:
+      surf->ss3.tiled_surface = 1;
+      surf->ss3.tile_walk = BRW_TILEWALK_YMAJOR;
+      break;
+   }
+}
 
-   surf->ss0.mipmap_layout_mode = BRW_SURFACE_MIPMAPLAYOUT_BELOW;   
-   surf->ss0.surface_type = translate_tex_target(tObj->Target);
-   surf->ss0.surface_format = translate_tex_format(firstImage->TexFormat->MesaFormat);
+static dri_bo *
+brw_create_texture_surface( struct brw_context *brw,
+			    struct brw_wm_surface_key *key )
+{
+   struct brw_surface_state surf;
+   dri_bo *bo;
+
+   memset(&surf, 0, sizeof(surf));
+
+   surf.ss0.mipmap_layout_mode = BRW_SURFACE_MIPMAPLAYOUT_BELOW;
+   surf.ss0.surface_type = translate_tex_target(key->target);
+
+   if (key->bo) 
+      surf.ss0.surface_format = translate_tex_format(key->format, key->depthmode);
+   else {
+     switch(key->depth) {
+     case 32: surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM; break;
+     default:
+     case 24: surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8X8_UNORM; break;
+     case 16: surf.ss0.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM; break;
+     }
+   }
 
    /* This is ok for all textures with channel width 8bit or less:
     */
-/*    surf->ss0.data_return_format = BRW_SURFACERETURNFORMAT_S1; */
-
-   /* BRW_NEW_LOCK */
-   surf->ss1.base_addr = bmBufferOffset(intel,
-					intelObj->mt->region->buffer);
-
-   surf->ss2.mip_count = intelObj->lastLevel - intelObj->firstLevel;
-   surf->ss2.width = firstImage->Width - 1;
-   surf->ss2.height = firstImage->Height - 1;
-
-   surf->ss3.tile_walk = BRW_TILEWALK_XMAJOR;
-   surf->ss3.tiled_surface = intelObj->mt->region->tiled; /* always zero */
-   surf->ss3.pitch = (intelObj->mt->pitch * intelObj->mt->cpp) - 1;
-   surf->ss3.depth = firstImage->Depth - 1;
-
-   surf->ss4.min_lod = 0;
+/*    surf.ss0.data_return_format = BRW_SURFACERETURNFORMAT_S1; */
+   if (key->bo)
+     surf.ss1.base_addr = key->bo->offset; /* reloc */
+   else
+     surf.ss1.base_addr = key->offset;
+
+   surf.ss2.mip_count = key->last_level - key->first_level;
+   surf.ss2.width = key->width - 1;
+   surf.ss2.height = key->height - 1;
+   brw_set_surface_tiling(&surf, key->tiling);
+   surf.ss3.pitch = (key->pitch * key->cpp) - 1;
+   surf.ss3.depth = key->depth - 1;
+
+   surf.ss4.min_lod = 0;
  
-   if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
-      surf->ss0.cube_pos_x = 1;
-      surf->ss0.cube_pos_y = 1;
-      surf->ss0.cube_pos_z = 1;
-      surf->ss0.cube_neg_x = 1;
-      surf->ss0.cube_neg_y = 1;
-      surf->ss0.cube_neg_z = 1;
+   if (key->target == GL_TEXTURE_CUBE_MAP) {
+      surf.ss0.cube_pos_x = 1;
+      surf.ss0.cube_pos_y = 1;
+      surf.ss0.cube_pos_z = 1;
+      surf.ss0.cube_neg_x = 1;
+      surf.ss0.cube_neg_y = 1;
+      surf.ss0.cube_neg_z = 1;
    }
-}
-
 
+   bo = brw_upload_cache(&brw->cache, BRW_SS_SURFACE,
+			 key, sizeof(*key),
+			 &key->bo, key->bo ? 1 : 0,
+			 &surf, sizeof(surf),
+			 NULL, NULL);
+
+   if (key->bo) {
+      /* Emit relocation to surface contents */
+      dri_bo_emit_reloc(bo,
+			I915_GEM_DOMAIN_SAMPLER, 0,
+			0,
+			offsetof(struct brw_surface_state, ss1),
+			key->bo);
+   }
+   return bo;
+}
 
-#define OFFSET(TYPE, FIELD) ( (GLuint)&(((TYPE *)0)->FIELD) )
+static void
+brw_update_texture_surface( GLcontext *ctx, GLuint unit )
+{
+   struct brw_context *brw = brw_context(ctx);
+   struct gl_texture_object *tObj = brw->attribs.Texture->Unit[unit]._Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   struct gl_texture_image *firstImage = tObj->Image[0][intelObj->firstLevel];
+   struct brw_wm_surface_key key;
+
+   memset(&key, 0, sizeof(key));
+
+   if (intelObj->imageOverride) {
+      key.pitch = intelObj->pitchOverride / intelObj->mt->cpp;
+      key.depth = intelObj->depthOverride;
+      key.bo = NULL;
+      key.offset = intelObj->textureOffset;
+   } else {
+      key.format = firstImage->TexFormat->MesaFormat;
+      key.pitch = intelObj->mt->pitch;
+      key.depth = firstImage->Depth;
+      key.bo = intelObj->mt->region->buffer;
+      key.offset = 0;
+   }
 
+   key.target = tObj->Target;
+   key.depthmode = tObj->DepthMode;
+   key.first_level = intelObj->firstLevel;
+   key.last_level = intelObj->lastLevel;
+   key.width = firstImage->Width;
+   key.height = firstImage->Height;
+   key.cpp = intelObj->mt->cpp;
+   key.tiling = intelObj->mt->region->tiling;
+
+   dri_bo_unreference(brw->wm.surf_bo[unit + MAX_DRAW_BUFFERS]);
+   brw->wm.surf_bo[unit + MAX_DRAW_BUFFERS] = brw_search_cache(&brw->cache, BRW_SS_SURFACE,
+							       &key, sizeof(key),
+							       &key.bo, key.bo ? 1 : 0,
+							       NULL);
+   if (brw->wm.surf_bo[unit + MAX_DRAW_BUFFERS] == NULL) {
+      brw->wm.surf_bo[unit + MAX_DRAW_BUFFERS] = brw_create_texture_surface(brw, &key);
+   }
+}
 
-static void upload_wm_surfaces(struct brw_context *brw )
+/**
+ * Sets up a surface state structure to point at the given region.
+ * While it is only used for the front/back buffer currently, it should be
+ * usable for further buffers when doing ARB_draw_buffer support.
+ */
+static void
+brw_update_region_surface(struct brw_context *brw, struct intel_region *region,
+			  unsigned int unit, GLboolean cached)
 {
-   GLcontext *ctx = &brw->intel.ctx;
-   struct intel_context *intel = &brw->intel;
-   struct brw_surface_binding_table bind;
-   GLuint i;
-
-   memcpy(&bind, &brw->wm.bind, sizeof(bind));
-      
-   {
+   dri_bo *region_bo = NULL;
+   struct {
+      unsigned int surface_type;
+      unsigned int surface_format;
+      unsigned int width, height, cpp;
+      GLubyte color_mask[4];
+      GLboolean color_blend;
+      uint32_t tiling;
+   } key;
+
+   memset(&key, 0, sizeof(key));
+
+   if (region != NULL) {
+      region_bo = region->buffer;
+
+      key.surface_type = BRW_SURFACE_2D;
+      if (region->cpp == 4)
+	 key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+      else
+	 key.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM;
+      key.tiling = region->tiling;
+      key.width = region->pitch; /* XXX: not really! */
+      key.height = region->height;
+      key.cpp = region->cpp;
+   } else {
+      key.surface_type = BRW_SURFACE_NULL;
+      key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+      key.tiling = 0;
+      key.width = 1;
+      key.height = 1;
+      key.cpp = 4;
+   }
+   memcpy(key.color_mask, brw->attribs.Color->ColorMask,
+	  sizeof(key.color_mask));
+   key.color_blend = (!brw->attribs.Color->_LogicOpEnabled &&
+		      brw->attribs.Color->BlendEnabled);
+
+   dri_bo_unreference(brw->wm.surf_bo[unit]);
+   brw->wm.surf_bo[unit] = NULL;
+   if (cached) 
+       brw->wm.surf_bo[unit] = brw_search_cache(&brw->cache, BRW_SS_SURFACE,
+	       &key, sizeof(key),
+	       &region_bo, 1,
+	       NULL);
+
+   if (brw->wm.surf_bo[unit] == NULL) {
       struct brw_surface_state surf;
-      struct intel_region *region = brw->state.draw_region;
 
       memset(&surf, 0, sizeof(surf));
 
-      if (region->cpp == 4)
-	 surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
-      else 
-	 surf.ss0.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM;
+      surf.ss0.surface_format = key.surface_format;
+      surf.ss0.surface_type = key.surface_type;
+      if (region_bo != NULL)
+	 surf.ss1.base_addr = region_bo->offset; /* reloc */
 
-      surf.ss0.surface_type = BRW_SURFACE_2D;
+      surf.ss2.width = key.width - 1;
+      surf.ss2.height = key.height - 1;
+      brw_set_surface_tiling(&surf, key.tiling);
+      surf.ss3.pitch = (key.width * key.cpp) - 1;
 
       /* _NEW_COLOR */
-      surf.ss0.color_blend = (!brw->attribs.Color->_LogicOpEnabled &&
-			      brw->attribs.Color->BlendEnabled);
+      surf.ss0.color_blend = key.color_blend;
+      surf.ss0.writedisable_red =   !key.color_mask[0];
+      surf.ss0.writedisable_green = !key.color_mask[1];
+      surf.ss0.writedisable_blue =  !key.color_mask[2];
+      surf.ss0.writedisable_alpha = !key.color_mask[3];
+
+      /* Key size will never match key size for textures, so we're safe. */
+      brw->wm.surf_bo[unit] = brw_upload_cache(&brw->cache, BRW_SS_SURFACE,
+					      &key, sizeof(key),
+					       &region_bo, 1,
+					       &surf, sizeof(surf),
+					       NULL, NULL);
+      if (region_bo != NULL) {
+	 /* We might sample from it, and we might render to it, so flag
+	  * them both.  We might be able to figure out from other state
+	  * a more restrictive relocation to emit.
+	  */
+	 dri_bo_emit_reloc(brw->wm.surf_bo[unit],
+			   I915_GEM_DOMAIN_RENDER |
+			   I915_GEM_DOMAIN_SAMPLER,
+			   I915_GEM_DOMAIN_RENDER,
+			   0,
+			   offsetof(struct brw_surface_state, ss1),
+			   region_bo);
+      }
+   }
+}
 
 
-      surf.ss0.writedisable_red =   !brw->attribs.Color->ColorMask[0];
-      surf.ss0.writedisable_green = !brw->attribs.Color->ColorMask[1];
-      surf.ss0.writedisable_blue =  !brw->attribs.Color->ColorMask[2];
-      surf.ss0.writedisable_alpha = !brw->attribs.Color->ColorMask[3];
+/**
+ * Constructs the binding table for the WM surface state, which maps unit
+ * numbers to surface state objects.
+ */
+static dri_bo *
+brw_wm_get_binding_table(struct brw_context *brw)
+{
+   dri_bo *bind_bo;
+
+   bind_bo = brw_search_cache(&brw->cache, BRW_SS_SURF_BIND,
+			      NULL, 0,
+			      brw->wm.surf_bo, brw->wm.nr_surfaces,
+			      NULL);
+
+   if (bind_bo == NULL) {
+      GLuint data_size = brw->wm.nr_surfaces * sizeof(GLuint);
+      uint32_t *data = malloc(data_size);
+      int i;
+
+      for (i = 0; i < brw->wm.nr_surfaces; i++)
+         if (brw->wm.surf_bo[i])
+            data[i] = brw->wm.surf_bo[i]->offset;
+         else
+            data[i] = 0;
+
+      bind_bo = brw_upload_cache( &brw->cache, BRW_SS_SURF_BIND,
+				  NULL, 0,
+				  brw->wm.surf_bo, brw->wm.nr_surfaces,
+				  data, data_size,
+				  NULL, NULL);
+
+      /* Emit binding table relocations to surface state */
+      for (i = 0; i < BRW_WM_MAX_SURF; i++) {
+	 if (brw->wm.surf_bo[i] != NULL) {
+	    dri_bo_emit_reloc(bind_bo,
+			      I915_GEM_DOMAIN_INSTRUCTION, 0,
+			      0,
+			      i * sizeof(GLuint),
+			      brw->wm.surf_bo[i]);
+	 }
+      }
 
-      /* BRW_NEW_LOCK */
-      surf.ss1.base_addr = bmBufferOffset(&brw->intel, region->buffer);
+      free(data);
+   }
 
+   return bind_bo;
+}
 
-      surf.ss2.width = region->pitch - 1; /* XXX: not really! */
-      surf.ss2.height = region->height - 1;
-      surf.ss3.tile_walk = BRW_TILEWALK_XMAJOR;
-      surf.ss3.tiled_surface = region->tiled;
-      surf.ss3.pitch = (region->pitch * region->cpp) - 1;
+static void prepare_wm_surfaces(struct brw_context *brw )
+{
+   GLcontext *ctx = &brw->intel.ctx;
+   struct intel_context *intel = &brw->intel;
+   GLuint i;
+   int old_nr_surfaces;
 
-      brw->wm.bind.surf_ss_offset[0] = brw_cache_data( &brw->cache[BRW_SS_SURFACE], &surf );
-      brw->wm.nr_surfaces = 1;
+   if (brw->state.nr_draw_regions  > 1) {
+      for (i = 0; i < brw->state.nr_draw_regions; i++) {
+         brw_update_region_surface(brw, brw->state.draw_regions[i], i,
+				   GL_FALSE);
+      }
+   }else {
+      brw_update_region_surface(brw, brw->state.draw_regions[0], 0, GL_TRUE);
    }
 
+   old_nr_surfaces = brw->wm.nr_surfaces;
+   brw->wm.nr_surfaces = MAX_DRAW_BUFFERS;
 
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
       struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
 
-      /* _NEW_TEXTURE, BRW_NEW_TEXDATA 
-       */
-      if (texUnit->_ReallyEnabled &&
-	  intel_finalize_mipmap_tree(intel,texUnit->_Current)) {
-
-	 struct brw_surface_state surf;
-
-	 brw_update_texture_surface(ctx, i, &surf);
-
-	 brw->wm.bind.surf_ss_offset[i+1] = brw_cache_data( &brw->cache[BRW_SS_SURFACE], &surf );
-	 brw->wm.nr_surfaces = i+2;
-      }
-      else if( texUnit->_ReallyEnabled &&
-	       texUnit->_Current == intel->frame_buffer_texobj )
-      {
-	 brw->wm.bind.surf_ss_offset[i+1] = brw->wm.bind.surf_ss_offset[0];
-	 brw->wm.nr_surfaces = i+2;
-      }    
-      else {
-	 brw->wm.bind.surf_ss_offset[i+1] = 0;
+      /* _NEW_TEXTURE, BRW_NEW_TEXDATA */
+      if(texUnit->_ReallyEnabled) {
+         if (texUnit->_Current == intel->frame_buffer_texobj) {
+            dri_bo_unreference(brw->wm.surf_bo[i+MAX_DRAW_BUFFERS]);
+            brw->wm.surf_bo[i+MAX_DRAW_BUFFERS] = brw->wm.surf_bo[0];
+            dri_bo_reference(brw->wm.surf_bo[i+MAX_DRAW_BUFFERS]);
+            brw->wm.nr_surfaces = i + MAX_DRAW_BUFFERS + 1;
+         } else {
+            brw_update_texture_surface(ctx, i);
+            brw->wm.nr_surfaces = i + MAX_DRAW_BUFFERS + 1;
+         }
+      } else {
+         dri_bo_unreference(brw->wm.surf_bo[i+MAX_DRAW_BUFFERS]);
+         brw->wm.surf_bo[i+MAX_DRAW_BUFFERS] = NULL;
       }
+
    }
 
-   brw->wm.bind_ss_offset = brw_cache_data( &brw->cache[BRW_SS_SURF_BIND],
-					    &brw->wm.bind );
+   dri_bo_unreference(brw->wm.bind_bo);
+   brw->wm.bind_bo = brw_wm_get_binding_table(brw);
+
+   if (brw->wm.nr_surfaces != old_nr_surfaces)
+      brw->state.dirty.brw |= BRW_NEW_NR_SURFACES;
 }
 
+
 const struct brw_tracked_state brw_wm_surfaces = {
    .dirty = {
       .mesa = _NEW_COLOR | _NEW_TEXTURE | _NEW_BUFFERS,
-      .brw = (BRW_NEW_CONTEXT | 
-	      BRW_NEW_LOCK),	/* required for bmBufferOffset */
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
-   .update = upload_wm_surfaces
+   .prepare = prepare_wm_surfaces,
 };
 
 
diff --git a/src/mesa/drivers/dri/i965/bufmgr.h b/src/mesa/drivers/dri/i965/bufmgr.h
deleted file mode 100644
index e748c0d6d0b..00000000000
--- a/src/mesa/drivers/dri/i965/bufmgr.h
+++ /dev/null
@@ -1,215 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef BUFMGR_H
-#define BUFMGR_H
-
-#include "intel_context.h"
-
-
-/* The buffer manager context.  Opaque.
- */
-struct bufmgr;
-struct buffer;
-
-
-struct bufmgr *bm_fake_intel_Attach( struct intel_context *intel ); 
-
-/* Flags for validate and other calls.  If both NO_UPLOAD and NO_EVICT
- * are specified, ValidateBuffers is essentially a query.
- */
-#define BM_MEM_LOCAL   0x1
-#define BM_MEM_AGP     0x2
-#define BM_MEM_VRAM    0x4	/* not yet used */
-#define BM_WRITE       0x8	/* not yet used */
-#define BM_READ        0x10	/* not yet used */
-#define BM_NO_UPLOAD   0x20
-#define BM_NO_EVICT    0x40
-#define BM_NO_MOVE     0x80	/* not yet used */
-#define BM_NO_ALLOC    0x100	/* legacy "fixed" buffers only */
-#define BM_CLIENT      0x200	/* for map - pointer will be accessed
-				 * without dri lock */
-
-#define BM_MEM_MASK (BM_MEM_LOCAL|BM_MEM_AGP|BM_MEM_VRAM)
-
-
-
-
-/* Create a pool of a given memory type, from a certain offset and a
- * certain size.  
- *
- * Also passed in is a virtual pointer to the start of the pool.  This
- * is useful in the faked-out version in i915 so that MapBuffer can
- * return a pointer to a buffer residing in AGP space.  
- *
- * Flags passed into a pool are inherited by all buffers allocated in
- * that pool.  So pools representing the static front,back,depth
- * buffer allocations should have MEM_AGP|NO_UPLOAD|NO_EVICT|NO_MOVE to match
- * the behaviour of the legacy allocations.
- *
- * Returns -1 for failure, pool number for success.
- */
-int bmInitPool( struct intel_context *, 
-		unsigned long low_offset,
-		void *low_virtual,
-		unsigned long size,
-		unsigned flags);
-
-
-/* Stick closely to ARB_vbo semantics - they're well defined and
- * understood, and drivers can just pass the calls through without too
- * much thunking.
- */
-void bmGenBuffers(struct intel_context *, const char *, unsigned n, struct buffer **buffers,
-		  int align );
-void bmDeleteBuffers(struct intel_context *, unsigned n, struct buffer **buffers);
-
-
-/* Hook to inform faked buffer manager about fixed-position
- * front,depth,back buffers.  These may move to a fully memory-managed
- * scheme, or they may continue to be managed as is.
- */
-struct buffer *bmGenBufferStatic(struct intel_context *,
-				 unsigned pool);
-
-/* On evict, buffer manager will call invalidate_cb() to note that the
- * buffer needs to be reloaded.
- *
- * Buffer is uploaded by calling bmMapBuffer() and copying data into
- * the returned pointer.
- *
- * This is basically a big hack to get some more performance by
- * turning off backing store for buffers where we either have it
- * already (textures) or don't need it (batch buffers, temporary
- * vbo's).
- */
-void bmBufferSetInvalidateCB(struct intel_context *,
-			     struct buffer *buf,
-			     void (*invalidate_cb)( struct intel_context *, void *ptr ),
-			     void *ptr,
-			     GLboolean dont_fence_subdata);
-
-
-/* The driver has more intimate knowledge of the hardare than a GL
- * client would, so flags here is more proscriptive than the usage
- * values in the ARB_vbo interface:
- */
-int bmBufferData(struct intel_context *, 
-		  struct buffer *buf, 
-		  unsigned size, 
-		  const void *data, 
-		  unsigned flags );
-
-int bmBufferSubData(struct intel_context *, 
-		     struct buffer *buf, 
-		     unsigned offset, 
-		     unsigned size, 
-		     const void *data );
-
-
-int bmBufferDataAUB(struct intel_context *, 
-		     struct buffer *buf, 
-		     unsigned size, 
-		     const void *data, 
-		     unsigned flags,
-		     unsigned aubtype,
-		     unsigned aubsubtype );
-
-int bmBufferSubDataAUB(struct intel_context *, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			const void *data,
-			unsigned aubtype,
-			unsigned aubsubtype );
-
-
-/* In this version, taking the offset will provoke an upload on
- * buffers not already resident in AGP:
- */
-unsigned bmBufferOffset(struct intel_context *, 
-			struct buffer *buf);
-
-
-/* Extract data from the buffer:
- */
-void bmBufferGetSubData(struct intel_context *, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			void *data );
-
-void *bmMapBuffer( struct intel_context *,
-		   struct buffer *buf, 
-		   unsigned access );
-
-void bmUnmapBuffer( struct intel_context *,
-		    struct buffer *buf );
-
-void bmUnmapBufferAUB( struct intel_context *,
-		       struct buffer *buf,
-		       unsigned aubtype,
-		       unsigned aubsubtype );
-
-
-/* Pertains to all buffers who's offset has been taken since the last
- * fence or release.
- */
-int bmValidateBuffers( struct intel_context * );
-void bmReleaseBuffers( struct intel_context * );
-
-GLuint bmCtxId( struct intel_context *intel );
-
-
-GLboolean bmError( struct intel_context * );
-void bmEvictAll( struct intel_context * );
-
-void *bmFindVirtual( struct intel_context *intel,
-		     unsigned int offset,
-		     size_t sz );
-
-/* This functionality is used by the buffer manager, not really sure
- * if we need to be exposing it in this way, probably libdrm will
- * offer equivalent calls.
- *
- * For now they can stay, but will likely change/move before final:
- */
-unsigned bmSetFence( struct intel_context * );
-unsigned bmSetFenceLock( struct intel_context * );
-unsigned bmLockAndFence( struct intel_context *intel );
-int bmTestFence( struct intel_context *, unsigned fence );
-void bmFinishFence( struct intel_context *, unsigned fence );
-void bmFinishFenceLock( struct intel_context *, unsigned fence );
-
-void bm_fake_NotifyContendedLockTake( struct intel_context * );
-
-extern int INTEL_DEBUG;
-#define DEBUG_BUFMGR 0x10000000
-
-#define DBG(...)  do { if (INTEL_DEBUG & DEBUG_BUFMGR) _mesa_printf(__VA_ARGS__); } while(0)
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/bufmgr_fake.c b/src/mesa/drivers/dri/i965/bufmgr_fake.c
deleted file mode 100644
index 24ee11edd8c..00000000000
--- a/src/mesa/drivers/dri/i965/bufmgr_fake.c
+++ /dev/null
@@ -1,1460 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/* Originally a fake version of the buffer manager so that we can
- * prototype the changes in a driver fairly quickly, has been fleshed
- * out to a fully functional interim solution.
- *
- * Basically wraps the old style memory management in the new
- * programming interface, but is more expressive and avoids many of
- * the bugs in the old texture manager.
- */
-#include "bufmgr.h"
-
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-
-#include "simple_list.h"
-#include "mm.h"
-#include "imports.h"
-
-#define BM_POOL_MAX 8
-
-/* Internal flags:
- */
-#define BM_NO_BACKING_STORE   0x2000
-#define BM_NO_FENCE_SUBDATA   0x4000
-
-
-static int check_fenced( struct intel_context *intel );
-
-static int nr_attach = 0;
-
-/* Wrapper around mm.c's mem_block, which understands that you must
- * wait for fences to expire before memory can be freed.  This is
- * specific to our use of memcpy for uploads - an upload that was
- * processed through the command queue wouldn't need to care about
- * fences.
- */
-struct block {
-   struct block *next, *prev;
-   struct pool *pool;		/* BM_MEM_AGP */
-   struct mem_block *mem;	/* BM_MEM_AGP */
-
-   unsigned referenced:1;
-   unsigned on_hardware:1;
-   unsigned fenced:1;	
-   
-
-   unsigned fence;		/* BM_MEM_AGP, Split to read_fence, write_fence */
-
-   struct buffer *buf;
-   void *virtual;
-};
-
-
-struct buffer {
-   unsigned id;			/* debug only */
-   const char *name;
-   unsigned size;
-
-   unsigned mapped:1;		
-   unsigned dirty:1;		
-   unsigned aub_dirty:1;	
-   unsigned alignment:13;
-   unsigned flags:16;
-
-   struct block *block;
-   void *backing_store;
-   void (*invalidate_cb)( struct intel_context *, void * );
-   void *invalidate_ptr;
-};
-
-struct pool {
-   unsigned size;
-   unsigned low_offset;
-   struct buffer *static_buffer;
-   unsigned flags;
-   struct mem_block *heap;
-   void *virtual;
-   struct block lru;		/* only allocated, non-fence-pending blocks here */
-};
-
-struct bufmgr {
-   _glthread_Mutex mutex;	/**< for thread safety */
-   struct pool pool[BM_POOL_MAX];
-   unsigned nr_pools;
-
-   unsigned buf_nr;		/* for generating ids */
-
-   struct block referenced;	/* after bmBufferOffset */
-   struct block on_hardware;	/* after bmValidateBuffers */
-   struct block fenced;		/* after bmFenceBuffers (mi_flush, emit irq, write dword) */
-                                /* then to pool->lru or free() */
-
-   unsigned ctxId;
-   unsigned last_fence;
-   unsigned free_on_hardware;
-
-   unsigned fail:1;
-   unsigned need_fence:1;
-};
-
-#define MAXFENCE 0x7fffffff
-
-static GLboolean FENCE_LTE( unsigned a, unsigned b )
-{
-   if (a == b)
-      return GL_TRUE;
-
-   if (a < b && b - a < (1<<24))
-      return GL_TRUE;
-
-   if (a > b && MAXFENCE - a + b < (1<<24))
-      return GL_TRUE;
-
-   return GL_FALSE;
-}
-
-int bmTestFence( struct intel_context *intel, unsigned fence )
-{
-   /* Slight problem with wrap-around:
-    */
-   return fence == 0 || FENCE_LTE(fence, intel->sarea->last_dispatch);
-}
-
-#define LOCK(bm) \
-  int dolock = nr_attach > 1; \
-  if (dolock) _glthread_LOCK_MUTEX(bm->mutex)
-
-#define UNLOCK(bm) \
-  if (dolock) _glthread_UNLOCK_MUTEX(bm->mutex)
-
-
-
-static GLboolean alloc_from_pool( struct intel_context *intel,				
-				  unsigned pool_nr,
-				  struct buffer *buf )
-{
-   struct bufmgr *bm = intel->bm;
-   struct pool *pool = &bm->pool[pool_nr];
-   struct block *block = (struct block *)calloc(sizeof *block, 1);
-   GLuint sz, align = (1<<buf->alignment);
-
-   if (!block)
-      return GL_FALSE;
-
-   sz = (buf->size + align-1) & ~(align-1);
-
-   block->mem = mmAllocMem(pool->heap, 
-			   sz, 
-			   buf->alignment, 0);
-   if (!block->mem) {
-      free(block);
-      return GL_FALSE;
-   }
-
-   make_empty_list(block);
-
-   /* Insert at head or at tail???   
-    */
-   insert_at_tail(&pool->lru, block);
-
-   block->pool = pool;
-   block->virtual = pool->virtual + block->mem->ofs;
-   block->buf = buf;
-
-   buf->block = block;
-
-   return GL_TRUE;
-}
-
-
-
-
-
-
-
-
-/* Release the card storage associated with buf:
- */
-static void free_block( struct intel_context *intel, struct block *block )
-{
-   DBG("free block %p\n", block);
-
-   if (!block) 
-      return;
-
-   check_fenced(intel);
-
-   if (block->referenced) {
-      _mesa_printf("tried to free block on referenced list\n");
-      assert(0);
-   }
-   else if (block->on_hardware) {
-      block->buf = NULL;
-      intel->bm->free_on_hardware += block->mem->size;
-   }
-   else if (block->fenced) {
-      block->buf = NULL;
-   }
-   else {
-      DBG("    - free immediately\n");
-      remove_from_list(block);
-
-      mmFreeMem(block->mem);
-      free(block);
-   }
-}
-
-
-static void alloc_backing_store( struct intel_context *intel, struct buffer *buf )
-{
-   assert(!buf->backing_store);
-   assert(!(buf->flags & (BM_NO_EVICT|BM_NO_BACKING_STORE)));
-
-   buf->backing_store = ALIGN_MALLOC(buf->size, 64);
-}
-
-static void free_backing_store( struct intel_context *intel, struct buffer *buf )
-{
-   assert(!(buf->flags & (BM_NO_EVICT|BM_NO_BACKING_STORE)));
-	  
-   if (buf->backing_store) {
-      ALIGN_FREE(buf->backing_store);
-      buf->backing_store = NULL;
-   }
-}
-
-
-
-
-
-
-static void set_dirty( struct intel_context *intel,
-			      struct buffer *buf )
-{
-   if (buf->flags & BM_NO_BACKING_STORE)
-      buf->invalidate_cb(intel, buf->invalidate_ptr);
-
-   assert(!(buf->flags & BM_NO_EVICT));
-
-   DBG("set_dirty - buf %d\n", buf->id);
-   buf->dirty = 1;
-}
-
-
-static int evict_lru( struct intel_context *intel, GLuint max_fence, GLuint *pool )
-{
-   struct bufmgr *bm = intel->bm;
-   struct block *block, *tmp;
-   int i;
-
-   DBG("%s\n", __FUNCTION__);
-
-   for (i = 0; i < bm->nr_pools; i++) {
-      if (!(bm->pool[i].flags & BM_NO_EVICT)) {
-	 foreach_s(block, tmp, &bm->pool[i].lru) {
-
-	    if (block->buf &&
-		(block->buf->flags & BM_NO_FENCE_SUBDATA))
-	       continue;
-
-	    if (block->fence && max_fence &&
-		!FENCE_LTE(block->fence, max_fence))
-	       return 0;
-
-	    set_dirty(intel, block->buf);
-	    block->buf->block = NULL;
-
-	    free_block(intel, block);
-	    *pool = i;
-	    return 1;
-	 }
-      }
-   }
-
-
-   return 0;
-}
-
-
-#define foreach_s_rev(ptr, t, list)   \
-        for(ptr=(list)->prev,t=(ptr)->prev; list != ptr; ptr=t, t=(t)->prev)
-
-static int evict_mru( struct intel_context *intel, GLuint *pool )
-{
-   struct bufmgr *bm = intel->bm;
-   struct block *block, *tmp;
-   int i;
-
-   DBG("%s\n", __FUNCTION__);
-
-   for (i = 0; i < bm->nr_pools; i++) {
-      if (!(bm->pool[i].flags & BM_NO_EVICT)) {
-	 foreach_s_rev(block, tmp, &bm->pool[i].lru) {
-
-	    if (block->buf &&
-		(block->buf->flags & BM_NO_FENCE_SUBDATA))
-	       continue;
-
-	    set_dirty(intel, block->buf);
-	    block->buf->block = NULL;
-
-	    free_block(intel, block);
-	    *pool = i;
-	    return 1;
-	 }
-      }
-   }
-
-
-   return 0;
-}
-
-
-static int check_fenced( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-   struct block *block, *tmp;
-   int ret = 0;
-
-   foreach_s(block, tmp, &bm->fenced ) {
-      assert(block->fenced);
-
-      if (bmTestFence(intel, block->fence)) {
-
-	 block->fenced = 0;
-
-	 if (!block->buf) {
-	    DBG("delayed free: offset %x sz %x\n", block->mem->ofs, block->mem->size);
-	    remove_from_list(block);
-	    mmFreeMem(block->mem);
-	    free(block);
-	 }
-	 else {
-	    DBG("return to lru: offset %x sz %x\n", block->mem->ofs, block->mem->size);
-	    move_to_tail(&block->pool->lru, block);
-	 }
-
-	 ret = 1;
-      }
-      else {
-	 /* Blocks are ordered by fence, so if one fails, all from
-	  * here will fail also:
-	  */
-	 break;
-      }
-   }
-
-   /* Also check the referenced list: 
-    */
-   foreach_s(block, tmp, &bm->referenced ) {
-      if (block->fenced &&
-	  bmTestFence(intel, block->fence)) {
-	 block->fenced = 0;
-      }
-   }
-
-   
-   DBG("%s: %d\n", __FUNCTION__, ret);
-   return ret;
-}
-
-
-
-static void fence_blocks( struct intel_context *intel,
-			  unsigned fence )
-{
-   struct bufmgr *bm = intel->bm;
-   struct block *block, *tmp;
-
-   foreach_s (block, tmp, &bm->on_hardware) {
-      DBG("Fence block %p (sz 0x%x buf %p) with fence %d\n", block, 
-	  block->mem->size, block->buf, fence);
-      block->fence = fence;
-
-      block->on_hardware = 0;
-      block->fenced = 1;
-
-      /* Move to tail of pending list here
-       */
-      move_to_tail(&bm->fenced, block);
-   }
-
-   /* Also check the referenced list:
-    */  
-   foreach_s (block, tmp, &bm->referenced) {
-      if (block->on_hardware) {
-	 DBG("Fence block %p (sz 0x%x buf %p) with fence %d\n", block, 
-	     block->mem->size, block->buf, fence);
-	 
-	 block->fence = fence;
-	 block->on_hardware = 0;
-	 block->fenced = 1;
-      }
-   }
-
-
-   bm->last_fence = fence;
-   assert(is_empty_list(&bm->on_hardware));
-}
-
-
-
-
-static GLboolean alloc_block( struct intel_context *intel,
-			      struct buffer *buf )
-{
-   struct bufmgr *bm = intel->bm;
-   int i;
-
-   assert(intel->locked);
-
-   DBG("%s 0x%x bytes (%s)\n", __FUNCTION__, buf->size, buf->name);
-
-   for (i = 0; i < bm->nr_pools; i++) {
-      if (!(bm->pool[i].flags & BM_NO_ALLOC) &&
-	  alloc_from_pool(intel, i, buf)) {
-
-	 DBG("%s --> 0x%x (sz %x)\n", __FUNCTION__, 
-	     buf->block->mem->ofs, buf->block->mem->size);
-	 
-	 return GL_TRUE;
-      }
-   }
-
-   DBG("%s --> fail\n", __FUNCTION__);
-   return GL_FALSE;   
-}
-
-
-static GLboolean evict_and_alloc_block( struct intel_context *intel,
-					struct buffer *buf )
-{
-   GLuint pool;
-   struct bufmgr *bm = intel->bm;
-
-   assert(buf->block == NULL);
-
-   /* Put a cap on the amount of free memory we'll allow to accumulate
-    * before emitting a fence.
-    */
-   if (bm->free_on_hardware > 1 * 1024 * 1024) {
-      DBG("fence for free space: %x\n", bm->free_on_hardware);
-      bmSetFence(intel);
-   }
-
-   /* Search for already free memory:
-    */
-   if (alloc_block(intel, buf))
-      return GL_TRUE;
-
-   /* Look for memory that may have become free: 
-    */
-   if (check_fenced(intel) &&
-       alloc_block(intel, buf))
-      return GL_TRUE;
-
-   /* Look for memory blocks not used for >1 frame:
-    */
-   while (evict_lru(intel, intel->second_last_swap_fence, &pool))
-      if (alloc_from_pool(intel, pool, buf))
-	 return GL_TRUE;
-
-   /* If we're not thrashing, allow lru eviction to dig deeper into
-    * recently used textures.  We'll probably be thrashing soon:
-    */
-   if (!intel->thrashing) {
-      while (evict_lru(intel, 0, &pool))
-	 if (alloc_from_pool(intel, pool, buf))
-	    return GL_TRUE;
-   }
-
-   /* Keep thrashing counter alive?
-    */
-   if (intel->thrashing)
-      intel->thrashing = 20;
-
-   /* Wait on any already pending fences - here we are waiting for any
-    * freed memory that has been submitted to hardware and fenced to
-    * become available:
-    */
-   while (!is_empty_list(&bm->fenced)) {
-      GLuint fence = bm->fenced.next->fence;
-      bmFinishFence(intel, fence);
-
-      if (alloc_block(intel, buf))
-	 return GL_TRUE;
-   }
-
-
-   /* 
-    */
-   if (!is_empty_list(&bm->on_hardware)) {
-      bmSetFence(intel);
-
-      while (!is_empty_list(&bm->fenced)) {
-	 GLuint fence = bm->fenced.next->fence;
-	 bmFinishFence(intel, fence);
-      }
-
-      if (!intel->thrashing) {	 
-	 DBG("thrashing\n");
-      }
-      intel->thrashing = 20; 
-
-      if (alloc_block(intel, buf))
-	 return GL_TRUE;
-   }
-
-   while (evict_mru(intel, &pool))
-      if (alloc_from_pool(intel, pool, buf))
-	 return GL_TRUE;
-
-   DBG("%s 0x%x bytes failed\n", __FUNCTION__, buf->size);
-
-   assert(is_empty_list(&bm->on_hardware));
-   assert(is_empty_list(&bm->fenced));
-
-   return GL_FALSE;
-}
-
-
-
-
-
-
-
-
-
-
-/***********************************************************************
- * Public functions
- */
-
-
-/* The initialization functions are skewed in the fake implementation.
- * This call would be to attach to an existing manager, rather than to
- * create a local one.
- */
-struct bufmgr *bm_fake_intel_Attach( struct intel_context *intel )
-{
-   _glthread_DECLARE_STATIC_MUTEX(initMutex);   
-   static struct bufmgr bm;
-   
-   /* This function needs a mutex of its own...
-    */
-   _glthread_LOCK_MUTEX(initMutex);
-
-   if (nr_attach == 0) {
-      _glthread_INIT_MUTEX(bm.mutex);
-
-      make_empty_list(&bm.referenced);
-      make_empty_list(&bm.fenced);
-      make_empty_list(&bm.on_hardware);
-      
-      /* The context id of any of the share group.  This won't be used
-       * in communication with the kernel, so it doesn't matter if
-       * this context is eventually deleted.
-       */
-      bm.ctxId = intel->hHWContext;
-   }
-
-   nr_attach++;
-
-   _glthread_UNLOCK_MUTEX(initMutex);
-
-   return &bm;
-}
-
-
-
-/* The virtual pointer would go away in a true implementation.
- */
-int bmInitPool( struct intel_context *intel, 
-		unsigned long low_offset,
-		void *low_virtual,
-		unsigned long size,
-		unsigned flags)
-{
-   struct bufmgr *bm = intel->bm;
-   int retval = 0;
-
-   LOCK(bm);
-   {
-      GLuint i;
-
-      for (i = 0; i < bm->nr_pools; i++) {
-	 if (bm->pool[i].low_offset == low_offset &&
-	     bm->pool[i].size == size) {
-	    retval = i;
-	    goto out;
-	 }
-      }
-
-
-      if (bm->nr_pools >= BM_POOL_MAX)
-	 retval = -1;
-      else {
-	 i = bm->nr_pools++;
-   
-	 DBG("bmInitPool %d low_offset %x sz %x\n",
-	     i, low_offset, size);
-   
-	 bm->pool[i].low_offset = low_offset;
-	 bm->pool[i].size = size;
-	 bm->pool[i].heap = mmInit( low_offset, size );
-	 bm->pool[i].virtual = low_virtual - low_offset;
-	 bm->pool[i].flags = flags;
-   
-	 make_empty_list(&bm->pool[i].lru);
-	 
-	 retval = i;
-      }
-   }
- out:
-   UNLOCK(bm);
-   return retval;
-}
-
-static struct buffer *do_GenBuffer(struct intel_context *intel, const char *name, int align)
-{
-   struct bufmgr *bm = intel->bm;
-   struct buffer *buf = calloc(sizeof(*buf), 1);
-
-   buf->id = ++bm->buf_nr;
-   buf->name = name;
-   buf->alignment = align;	
-   buf->flags = BM_MEM_AGP|BM_MEM_VRAM|BM_MEM_LOCAL;
-
-   return buf;
-}
-
-
-void *bmFindVirtual( struct intel_context *intel,
-		     unsigned int offset,
-		     size_t sz )
-{
-   struct bufmgr *bm = intel->bm;
-   int i;
-
-   for (i = 0; i < bm->nr_pools; i++)
-      if (offset >= bm->pool[i].low_offset &&
-	  offset + sz <= bm->pool[i].low_offset + bm->pool[i].size)
-	 return bm->pool[i].virtual + offset;
-
-   return NULL;
-}
- 
-
-void bmGenBuffers(struct intel_context *intel, 
-		  const char *name, unsigned n, 
-		  struct buffer **buffers,
-		  int align )
-{
-   struct bufmgr *bm = intel->bm;
-   LOCK(bm);
-   {
-      int i;
-
-      for (i = 0; i < n; i++)
-	 buffers[i] = do_GenBuffer(intel, name, align);
-   }
-   UNLOCK(bm);
-}
-
-
-void bmDeleteBuffers(struct intel_context *intel, unsigned n, struct buffer **buffers)
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      unsigned i;
-   
-      for (i = 0; i < n; i++) {
-	 struct buffer *buf = buffers[i];
-
-	 if (buf && buf->block)
-	    free_block(intel, buf->block);
-
-	 if (buf) 
-	    free(buf);	 
-      }
-   }
-   UNLOCK(bm);
-}
-
-
-
-
-/* Hook to inform faked buffer manager about fixed-position
- * front,depth,back buffers.  These may move to a fully memory-managed
- * scheme, or they may continue to be managed as is.  It will probably
- * be useful to pass a fixed offset here one day.
- */
-struct buffer *bmGenBufferStatic(struct intel_context *intel,
-				 unsigned pool )
-{
-   struct bufmgr *bm = intel->bm;
-   struct buffer *buf;
-   LOCK(bm);
-   {
-      assert(bm->pool[pool].flags & BM_NO_EVICT);
-      assert(bm->pool[pool].flags & BM_NO_MOVE);
-
-      if (bm->pool[pool].static_buffer)
-	 buf = bm->pool[pool].static_buffer;
-      else {
-	 buf = do_GenBuffer(intel, "static", 12);
-   
-	 bm->pool[pool].static_buffer = buf;
-	 assert(!buf->block);
-
-	 buf->size = bm->pool[pool].size;
-	 buf->flags = bm->pool[pool].flags;
-	 buf->alignment = 12;
-	 
-	 if (!alloc_from_pool(intel, pool, buf))
-	    assert(0);
-      }
-   }
-   UNLOCK(bm);
-   return buf;
-}
-
-
-static void wait_quiescent(struct intel_context *intel,
-			   struct block *block)
-{
-   if (block->on_hardware) {
-      assert(intel->bm->need_fence);
-      bmSetFence(intel);
-      assert(!block->on_hardware);
-   }
-
-
-   if (block->fenced) {
-      bmFinishFence(intel, block->fence);
-   }
-
-   assert(!block->on_hardware);
-   assert(!block->fenced);
-}
-
-
-
-/* If buffer size changes, free and reallocate.  Otherwise update in
- * place.
- */
-int bmBufferData(struct intel_context *intel, 
-		 struct buffer *buf, 
-		 unsigned size, 
-		 const void *data, 
-		 unsigned flags )
-{
-   struct bufmgr *bm = intel->bm;
-   int retval = 0;
-
-   LOCK(bm);
-   {
-      DBG("bmBufferData %d sz 0x%x data: %p\n", buf->id, size, data);
-
-      assert(!buf->mapped);
-
-      if (buf->block) {
-	 struct block *block = buf->block;
-
-	 /* Optimistic check to see if we can reuse the block -- not
-	  * required for correctness:
-	  */
-	 if (block->fenced)
-	    check_fenced(intel);
-
-	 if (block->on_hardware ||
-	     block->fenced ||
-	     (buf->size && buf->size != size) || 
-	     (data == NULL)) {
-
-	    assert(!block->referenced);
-
-	    free_block(intel, block);
-	    buf->block = NULL;
-	    buf->dirty = 1;
-	 }
-      }
-
-      buf->size = size;
-      if (buf->block) {
-	 assert (buf->block->mem->size >= size);
-      }
-
-      if (buf->flags & (BM_NO_BACKING_STORE|BM_NO_EVICT)) {
-
-	 assert(intel->locked || data == NULL);
-
-	 if (data != NULL) {
-	    if (!buf->block && !evict_and_alloc_block(intel, buf)) {
-	       bm->fail = 1;
-	       retval = -1;
-	       goto out;
-	    }
-
-	    wait_quiescent(intel, buf->block);
-
-	    DBG("bmBufferData %d offset 0x%x sz 0x%x\n", 
-		buf->id, buf->block->mem->ofs, size);
-
-	    assert(buf->block->virtual == buf->block->pool->virtual + buf->block->mem->ofs);
-
-	    do_memcpy(buf->block->virtual, data, size);
-	 }
-	 buf->dirty = 0;
-      }
-      else {
-	       DBG("%s - set buf %d dirty\n", __FUNCTION__, buf->id);
-	 set_dirty(intel, buf);
-	 free_backing_store(intel, buf);
-   
-	 if (data != NULL) {      
-	    alloc_backing_store(intel, buf);
-	    do_memcpy(buf->backing_store, data, size);
-	 }
-      }
-   }
- out:
-   UNLOCK(bm);
-   return retval;
-}
-
-
-/* Update the buffer in place, in whatever space it is currently resident:
- */
-int bmBufferSubData(struct intel_context *intel, 
-		     struct buffer *buf, 
-		     unsigned offset, 
-		     unsigned size, 
-		     const void *data )
-{
-   struct bufmgr *bm = intel->bm;
-   int retval = 0;
-
-   if (size == 0) 
-      return 0;
-
-   LOCK(bm); 
-   {
-      DBG("bmBufferSubdata %d offset 0x%x sz 0x%x\n", buf->id, offset, size);
-      
-      assert(offset+size <= buf->size);
-
-      if (buf->flags & (BM_NO_EVICT|BM_NO_BACKING_STORE)) {
-
-	 assert(intel->locked);
-
-	 if (!buf->block && !evict_and_alloc_block(intel, buf)) {
-	    bm->fail = 1;
-	    retval = -1;
-	    goto out;
-	 }
-	 
-	 if (!(buf->flags & BM_NO_FENCE_SUBDATA))
-	    wait_quiescent(intel, buf->block);
-
-	 buf->dirty = 0;
-
-	 do_memcpy(buf->block->virtual + offset, data, size);
-      }
-      else {
-	 DBG("%s - set buf %d dirty\n", __FUNCTION__, buf->id);
-	 set_dirty(intel, buf);
-
-	 if (buf->backing_store == NULL)
-	    alloc_backing_store(intel, buf);
-
-	 do_memcpy(buf->backing_store + offset, data, size); 
-      }
-   }
- out:
-   UNLOCK(bm);
-   return retval;
-}
-
-
-
-int bmBufferDataAUB(struct intel_context *intel, 
-		     struct buffer *buf, 
-		     unsigned size, 
-		     const void *data, 
-		     unsigned flags,
-		     unsigned aubtype,
-		     unsigned aubsubtype )
-{
-   int retval = bmBufferData(intel, buf, size, data, flags);
-   
-
-   /* This only works because in this version of the buffer manager we
-    * allocate all buffers statically in agp space and so can emit the
-    * uploads to the aub file with the correct offsets as they happen.
-    */
-   if (retval == 0 && data && intel->aub_file) {
-
-      if (buf->block && !buf->dirty) {
-	 intel->vtbl.aub_gtt_data(intel,
-				      buf->block->mem->ofs,
-				      buf->block->virtual,
-				      size,
-				      aubtype,
-				      aubsubtype);
-	 buf->aub_dirty = 0;
-      }
-   }
-   
-   return retval;
-}
-		       
-
-int bmBufferSubDataAUB(struct intel_context *intel, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			const void *data,
-			unsigned aubtype,
-			unsigned aubsubtype )
-{
-   int retval = bmBufferSubData(intel, buf, offset, size, data);
-   
-
-   /* This only works because in this version of the buffer manager we
-    * allocate all buffers statically in agp space and so can emit the
-    * uploads to the aub file with the correct offsets as they happen.
-    */
-   if (intel->aub_file) {
-      if (retval == 0 && buf->block && !buf->dirty)
-	 intel->vtbl.aub_gtt_data(intel,
-				      buf->block->mem->ofs + offset,
-				      ((const char *)buf->block->virtual) + offset,
-				      size,
-				      aubtype,
-				      aubsubtype);
-   }
-
-   return retval;
-}
-
-void bmUnmapBufferAUB( struct intel_context *intel, 
-		       struct buffer *buf,
-		       unsigned aubtype,
-		       unsigned aubsubtype )
-{
-   bmUnmapBuffer(intel, buf);
-
-   if (intel->aub_file) {
-      /* Hack - exclude the framebuffer mappings.  If you removed
-       * this, you'd get very big aubfiles, but you *would* be able to
-       * see fallback rendering.
-       */
-      if (buf->block  && !buf->dirty && buf->block->pool == &intel->bm->pool[0]) {
-	 buf->aub_dirty = 1;
-      }
-   }
-}
-
-unsigned bmBufferOffset(struct intel_context *intel, 
-			struct buffer *buf)
-{
-   struct bufmgr *bm = intel->bm;
-   unsigned retval = 0;
-
-   LOCK(bm);
-   {
-      assert(intel->locked);
-
-      if (!buf->block &&
-	  !evict_and_alloc_block(intel, buf)) {
-	 bm->fail = 1;
-	 retval = ~0;
-      }
-      else {
-	 assert(buf->block);
-	 assert(buf->block->buf == buf);
-
-	 DBG("Add buf %d (block %p, dirty %d) to referenced list\n", buf->id, buf->block,
-	     buf->dirty);
-
-	 move_to_tail(&bm->referenced, buf->block);
-	 buf->block->referenced = 1;
-
-	 retval = buf->block->mem->ofs;
-      }
-   }
-   UNLOCK(bm);
-
-   return retval;
-}
-
-
-
-/* Extract data from the buffer:
- */
-void bmBufferGetSubData(struct intel_context *intel, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			void *data )
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      DBG("bmBufferSubdata %d offset 0x%x sz 0x%x\n", buf->id, offset, size);
-
-      if (buf->flags & (BM_NO_EVICT|BM_NO_BACKING_STORE)) {
-	 if (buf->block && size) {
-	    wait_quiescent(intel, buf->block);
-	    do_memcpy(data, buf->block->virtual + offset, size); 
-	 }
-      }
-      else {
-	 if (buf->backing_store && size) {
-	    do_memcpy(data, buf->backing_store + offset, size); 
-	 }
-      }
-   }
-   UNLOCK(bm);
-}
-
-
-/* Return a pointer to whatever space the buffer is currently resident in:
- */
-void *bmMapBuffer( struct intel_context *intel,
-		   struct buffer *buf, 
-		   unsigned flags )
-{
-   struct bufmgr *bm = intel->bm;
-   void *retval = NULL;
-
-   LOCK(bm);
-   {
-      DBG("bmMapBuffer %d\n", buf->id);
-
-      if (buf->mapped) {
-	 _mesa_printf("%s: already mapped\n", __FUNCTION__);
-	 retval = NULL;
-      }
-      else if (buf->flags & (BM_NO_BACKING_STORE|BM_NO_EVICT)) {
-
-	 assert(intel->locked);
-
-	 if (!buf->block && !evict_and_alloc_block(intel, buf)) {
-	    DBG("%s: alloc failed\n", __FUNCTION__);
-	    bm->fail = 1;
-	    retval = NULL;
-	 }
-	 else {
-	    assert(buf->block);
-	    buf->dirty = 0;
-
-	    if (!(buf->flags & BM_NO_FENCE_SUBDATA)) 
-	       wait_quiescent(intel, buf->block);
-
-	    buf->mapped = 1;
-	    retval = buf->block->virtual;
-	 }
-      }
-      else {
-	 DBG("%s - set buf %d dirty\n", __FUNCTION__, buf->id);
-	 set_dirty(intel, buf);
-
-	 if (buf->backing_store == 0)
-	    alloc_backing_store(intel, buf);
-
-	 buf->mapped = 1;
-	 retval = buf->backing_store;
-      }
-   }
-   UNLOCK(bm);
-   return retval;
-}
-
-void bmUnmapBuffer( struct intel_context *intel, struct buffer *buf )
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      DBG("bmUnmapBuffer %d\n", buf->id);
-      buf->mapped = 0;
-   }
-   UNLOCK(bm);
-}
-
-
-
-
-/* This is the big hack that turns on BM_NO_BACKING_STORE.  Basically
- * says that an external party will maintain the backing store, eg
- * Mesa's local copy of texture data.
- */
-void bmBufferSetInvalidateCB(struct intel_context *intel,
-			     struct buffer *buf,
-			     void (*invalidate_cb)( struct intel_context *, void *ptr ),
-			     void *ptr,
-			     GLboolean dont_fence_subdata)
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      if (buf->backing_store)
-	 free_backing_store(intel, buf);
-
-      buf->flags |= BM_NO_BACKING_STORE;
-      
-      if (dont_fence_subdata)
-	 buf->flags |= BM_NO_FENCE_SUBDATA;
-
-      DBG("bmBufferSetInvalidateCB set buf %d dirty\n", buf->id);
-      buf->dirty = 1;
-      buf->invalidate_cb = invalidate_cb;
-      buf->invalidate_ptr = ptr;
-
-      /* Note that it is invalid right from the start.  Also note
-       * invalidate_cb is called with the bufmgr locked, so cannot
-       * itself make bufmgr calls.
-       */
-      invalidate_cb( intel, ptr );
-   }
-   UNLOCK(bm);
-}
-
-
-
-
-
-
-
-/* This is only protected against thread interactions by the DRI lock
- * and the policy of ensuring that all dma is flushed prior to
- * releasing that lock.  Otherwise you might have two threads building
- * up a list of buffers to validate at once.
- */
-int bmValidateBuffers( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-   int retval = 0;
-
-   LOCK(bm);
-   {
-      DBG("%s fail %d\n", __FUNCTION__, bm->fail);
-      assert(intel->locked);
-
-      if (!bm->fail) {
-	 struct block *block, *tmp;
-
-	 foreach_s(block, tmp, &bm->referenced) {
-	    struct buffer *buf = block->buf;
-
-	    DBG("Validate buf %d / block %p / dirty %d\n", buf->id, block, buf->dirty);
-
-	    /* Upload the buffer contents if necessary:
-	     */
-	    if (buf->dirty) {
-	       DBG("Upload dirty buf %d (%s) sz %d offset 0x%x\n", buf->id, 
-		   buf->name, buf->size, block->mem->ofs);
-
-	       assert(!(buf->flags & (BM_NO_BACKING_STORE|BM_NO_EVICT)));
-
-	       wait_quiescent(intel, buf->block);
-
-	       do_memcpy(buf->block->virtual,
-			 buf->backing_store, 
-			 buf->size);
-
-	       if (intel->aub_file) {
-		  intel->vtbl.aub_gtt_data(intel,
-					       buf->block->mem->ofs,
-					       buf->backing_store,
-					       buf->size,
-					       0,
-					       0);
-	       }
-
-	       buf->dirty = 0;
-	       buf->aub_dirty = 0;
-	    }
-	    else if (buf->aub_dirty) {
-	       intel->vtbl.aub_gtt_data(intel,
-					    buf->block->mem->ofs,
-					    buf->block->virtual,
-					    buf->size,
-					    0,
-					    0);
-	       buf->aub_dirty = 0;
-	    }
-
-	    block->referenced = 0;
-	    block->on_hardware = 1;
-	    move_to_tail(&bm->on_hardware, block);
-	 }
-
-	 bm->need_fence = 1;
-      }
-
-      retval = bm->fail ? -1 : 0;
-   }
-   UNLOCK(bm);
-
-
-   if (retval != 0)
-      DBG("%s failed\n", __FUNCTION__);
-
-   return retval;
-}
-
-
-
-
-void bmReleaseBuffers( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      struct block *block, *tmp;
-
-      foreach_s (block, tmp, &bm->referenced) {
-
-	 DBG("remove block %p from referenced list\n", block);
-
-	 if (block->on_hardware) {
-	    /* Return to the on-hardware list.
-	     */
-	    move_to_tail(&bm->on_hardware, block);	    
-	 }
-	 else if (block->fenced) {
-	    struct block *s;
-
-	    /* Hmm - have to scan the fenced list to insert the
-	     * buffers in order.  This is O(nm), but rare and the
-	     * numbers are low.
-	     */
-	    foreach (s, &bm->fenced) {
-	       if (FENCE_LTE(block->fence, s->fence))
-		  break;
-	    }
-	    
-	    move_to_tail(s, block);
-	 }
-	 else {			
-	    /* Return to the lru list:
-	     */
-	    move_to_tail(&block->pool->lru, block);
-	 }
-
-	 block->referenced = 0;
-      }
-   }
-   UNLOCK(bm);
-}
-
-
-/* This functionality is used by the buffer manager, not really sure
- * if we need to be exposing it in this way, probably libdrm will
- * offer equivalent calls.
- *
- * For now they can stay, but will likely change/move before final:
- */
-unsigned bmSetFence( struct intel_context *intel )
-{
-   assert(intel->locked);
-
-   /* Emit MI_FLUSH here:
-    */
-   if (intel->bm->need_fence) {
-
-      /* Emit a flush without using a batchbuffer.  Can't rely on the
-       * batchbuffer at this level really.  Would really prefer that
-       * the IRQ ioctly emitted the flush at the same time.
-       */
-      GLuint dword[2];
-      dword[0] = intel->vtbl.flush_cmd();
-      dword[1] = 0;
-      intel_cmd_ioctl(intel, (char *)&dword, sizeof(dword));
-      
-      intel->bm->last_fence = intelEmitIrqLocked( intel );
-      
-      fence_blocks(intel, intel->bm->last_fence);
-
-      intel->vtbl.note_fence(intel, intel->bm->last_fence);
-      intel->bm->need_fence = 0;
-
-      if (intel->thrashing) {
-	 intel->thrashing--;
-	 if (!intel->thrashing)
-	    DBG("not thrashing\n");
-      }
-      
-      intel->bm->free_on_hardware = 0;
-   }
-   
-   return intel->bm->last_fence;
-}
-
-unsigned bmSetFenceLock( struct intel_context *intel )
-{
-  unsigned last;
-  LOCK(intel->bm);
-  last = bmSetFence(intel);
-  UNLOCK(intel->bm);
-  return last;
-}
-unsigned bmLockAndFence( struct intel_context *intel )
-{
-   if (intel->bm->need_fence) {
-      LOCK_HARDWARE(intel);
-      LOCK(intel->bm);
-      bmSetFence(intel);
-      UNLOCK(intel->bm);
-      UNLOCK_HARDWARE(intel);
-   }
-
-   return intel->bm->last_fence;
-}
-
-
-void bmFinishFence( struct intel_context *intel, unsigned fence )
-{
-   if (!bmTestFence(intel, fence)) {
-      DBG("...wait on fence %d\n", fence);
-      intelWaitIrq( intel, fence );
-   }
-   assert(bmTestFence(intel, fence));
-   check_fenced(intel);
-}
-
-void bmFinishFenceLock( struct intel_context *intel, unsigned fence )
-{
-   LOCK(intel->bm);
-   bmFinishFence(intel, fence);
-   UNLOCK(intel->bm);
-}
-
-
-/* Specifically ignore texture memory sharing.
- *  -- just evict everything
- *  -- and wait for idle
- */
-void bm_fake_NotifyContendedLockTake( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      struct block *block, *tmp;
-      GLuint i;
-
-      assert(is_empty_list(&bm->referenced));
-
-      bm->need_fence = 1;
-      bm->fail = 0;
-      bmFinishFence(intel, bmSetFence(intel));
-
-      assert(is_empty_list(&bm->fenced));
-      assert(is_empty_list(&bm->on_hardware));
-
-      for (i = 0; i < bm->nr_pools; i++) {
-	 if (!(bm->pool[i].flags & BM_NO_EVICT)) {
-	    foreach_s(block, tmp, &bm->pool[i].lru) {
-	       assert(bmTestFence(intel, block->fence));
-	       set_dirty(intel, block->buf);
-	    }
-	 }
-      }
-   }
-   UNLOCK(bm);
-}
-
-
-
-void bmEvictAll( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-
-   LOCK(bm);
-   {
-      struct block *block, *tmp;
-      GLuint i;
-
-      DBG("%s\n", __FUNCTION__);
-
-      assert(is_empty_list(&bm->referenced));
-
-      bm->need_fence = 1;
-      bm->fail = 0;
-      bmFinishFence(intel, bmSetFence(intel));
-
-      assert(is_empty_list(&bm->fenced));
-      assert(is_empty_list(&bm->on_hardware));
-
-      for (i = 0; i < bm->nr_pools; i++) {
-	 if (!(bm->pool[i].flags & BM_NO_EVICT)) {
-	    foreach_s(block, tmp, &bm->pool[i].lru) {
-	       assert(bmTestFence(intel, block->fence));
-	       set_dirty(intel, block->buf);
-	       block->buf->block = NULL;
-
-	       free_block(intel, block);
-	    }
-	 }
-      }
-   }
-   UNLOCK(bm);
-}
-
-
-GLboolean bmError( struct intel_context *intel )
-{
-   struct bufmgr *bm = intel->bm;
-   GLboolean retval;
-
-   LOCK(bm);
-   {
-      retval = bm->fail;
-   }
-   UNLOCK(bm);
-
-   return retval;
-}
-
-
-GLuint bmCtxId( struct intel_context *intel )
-{
-   return intel->bm->ctxId;
-}
diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.c b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
index 64885ed9b4b..d38cdf31cc6 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
@@ -1,243 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "imports.h"
-#include "intel_batchbuffer.h"
-#include "intel_ioctl.h"
-#include "bufmgr.h"
-
-
-static void intel_batchbuffer_reset( struct intel_batchbuffer *batch )
-{
-   assert(batch->map == NULL);
-
-   batch->offset = (unsigned long)batch->ptr;
-   batch->offset = (batch->offset + 63) & ~63;
-   batch->ptr = (unsigned char *) batch->offset;
-
-   if (BATCH_SZ - batch->offset < BATCH_REFILL) {
-      bmBufferData(batch->intel, 
-		   batch->buffer,
-		   BATCH_SZ, 
-		   NULL, 
-		   0); 
-      batch->offset = 0;
-      batch->ptr = NULL;
-   }
-		
-   batch->flags = 0;
-}
-
-static void intel_batchbuffer_reset_cb( struct intel_context *intel,
-					void *ptr )
-{
-   struct intel_batchbuffer *batch = (struct intel_batchbuffer *)ptr;
-   assert(batch->map == NULL);
-   batch->flags = 0;
-   batch->offset = 0;
-   batch->ptr = NULL;
-}
-
-GLubyte *intel_batchbuffer_map( struct intel_batchbuffer *batch )
-{
-   if (!batch->map) {
-      batch->map = bmMapBuffer(batch->intel, batch->buffer, 
-			       BM_MEM_AGP|BM_MEM_LOCAL|BM_CLIENT|BM_WRITE);
-      batch->ptr += (unsigned long)batch->map;
-   }
-
-   return batch->map;
-}
-
-void intel_batchbuffer_unmap( struct intel_batchbuffer *batch )
-{
-   if (batch->map) {
-      batch->ptr -= (unsigned long)batch->map;
-      batch->map = NULL;
-      bmUnmapBuffer(batch->intel, batch->buffer);
-   }
-}
-
-
-
-/*======================================================================
- * Public functions
- */
-struct intel_batchbuffer *intel_batchbuffer_alloc( struct intel_context *intel )
-{
-   struct intel_batchbuffer *batch = calloc(sizeof(*batch), 1);
-
-   batch->intel = intel;
-
-   bmGenBuffers(intel, "batch", 1, &batch->buffer, 12);
-
-   bmBufferSetInvalidateCB(intel, batch->buffer,
-			   intel_batchbuffer_reset_cb,
-			   batch,
-			   GL_TRUE);
-
-   bmBufferData(batch->intel,
-		batch->buffer,
-		BATCH_SZ,
-		NULL,
-		0);
-
-
-   return batch;
-}
-
-void intel_batchbuffer_free( struct intel_batchbuffer *batch )
-{
-   if (batch->map) 
-      bmUnmapBuffer(batch->intel, batch->buffer);
-   
-   bmDeleteBuffers(batch->intel, 1, &batch->buffer);
-   free(batch);
-}
-
-
-#define MI_BATCH_BUFFER_END 	(0xA<<23)
-
-
-GLboolean intel_batchbuffer_flush( struct intel_batchbuffer *batch )
-{
-   struct intel_context *intel = batch->intel;
-   GLuint used = batch->ptr - (batch->map + batch->offset);
-   GLuint offset;
-   GLint retval = GL_TRUE;
-
-   assert(intel->locked);
-
-   if (used == 0) {
-      bmReleaseBuffers( batch->intel );
-      return GL_TRUE;
-   }
-
-   /* Add the MI_BATCH_BUFFER_END.  Always add an MI_FLUSH - this is a
-    * performance drain that we would like to avoid.
-    */
-   if (used & 4) {
-      ((int *)batch->ptr)[0] = MI_BATCH_BUFFER_END;
-      batch->ptr += 4;
-      used += 4;
-   }
-   else {
-      ((int *)batch->ptr)[0] = 0;
-      ((int *)batch->ptr)[1] = MI_BATCH_BUFFER_END;
-
-      batch->ptr += 8;
-      used += 8;
-   }
-
-   intel_batchbuffer_unmap(batch);
-
-   /* Get the batch buffer offset: Must call bmBufferOffset() before
-    * bmValidateBuffers(), otherwise the buffer won't be on the inuse
-    * list.
-    */
-   offset = bmBufferOffset(batch->intel, batch->buffer);
-
-   if (bmValidateBuffers( batch->intel ) != 0) {
-      assert(intel->locked);
-      bmReleaseBuffers( batch->intel );
-      retval = GL_FALSE;
-      goto out;
-   }
-
-
-   if (intel->aub_file) {
-      /* Send buffered commands to aubfile as a single packet. 
-       */
-      intel_batchbuffer_map(batch);
-      ((int *)batch->ptr)[-1] = intel->vtbl.flush_cmd();
-      intel->vtbl.aub_commands(intel,
-			       offset, /* Fulsim wierdness - don't adjust */
-			       batch->map + batch->offset,
-			       used);
-      ((int *)batch->ptr)[-1] = MI_BATCH_BUFFER_END;
-      intel_batchbuffer_unmap(batch);
-   }
-
-
-   /* Fire the batch buffer, which was uploaded above:
-    */
-   intel_batch_ioctl(batch->intel, 
-		     offset + batch->offset,
-		     used);
-
-   if (intel->aub_file && 
-       intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT)
-      intel->vtbl.aub_dump_bmp( intel, 0 );
-
-   /* Reset the buffer:
-    */
- out:
-   intel_batchbuffer_reset( batch );
-   intel_batchbuffer_map( batch );
-
-   if (!retval)
-      DBG("%s failed\n", __FUNCTION__);
-
-   return retval;
-}
-
-
-
-
-
-
-
-void intel_batchbuffer_align( struct intel_batchbuffer *batch,
-			      GLuint align,
-			      GLuint sz )
-{
-   unsigned long ptr = (unsigned long) batch->ptr;
-   unsigned long aptr = (ptr + align) & ~((unsigned long)align-1);
-   GLuint fixup = aptr - ptr;
-
-   if (intel_batchbuffer_space(batch) < fixup + sz)
-      intel_batchbuffer_flush(batch);
-   else {
-      memset(batch->ptr, 0, fixup);      
-      batch->ptr += fixup;
-   }
-}
-
-
-
-
-void intel_batchbuffer_data(struct intel_batchbuffer *batch,
-			    const void *data,
-			    GLuint bytes,
-			    GLuint flags)
-{
-   assert((bytes & 3) == 0);
-   intel_batchbuffer_require_space(batch, bytes, flags);
-   __memcpy(batch->ptr, data, bytes);
-   batch->ptr += bytes;
-}
-
+../intel/intel_batchbuffer.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.h b/src/mesa/drivers/dri/i965/intel_batchbuffer.h
deleted file mode 100644
index 25e0a65e99f..00000000000
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.h
+++ /dev/null
@@ -1,127 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_BATCHBUFFER_H
-#define INTEL_BATCHBUFFER_H
-
-#include "mtypes.h"
-#include "bufmgr.h"
-
-struct intel_context;
-
-#define BATCH_SZ (16 * 1024)
-#define BATCH_REFILL 4096
-#define BATCH_RESERVED 16
-
-#define INTEL_BATCH_NO_CLIPRECTS 0x1
-#define INTEL_BATCH_CLIPRECTS    0x2
-
-struct intel_batchbuffer {
-   struct intel_context *intel;
-
-   struct buffer *buffer;
-
-   GLuint flags;
-   unsigned long offset;
-
-   GLubyte *map;
-   GLubyte *ptr; 
-};
-
-struct intel_batchbuffer *intel_batchbuffer_alloc( struct intel_context *intel );
-
-void intel_batchbuffer_free( struct intel_batchbuffer *batch );
-
-
-GLboolean intel_batchbuffer_flush( struct intel_batchbuffer *batch );
-
-void intel_batchbuffer_unmap( struct intel_batchbuffer *batch );
-GLubyte *intel_batchbuffer_map( struct intel_batchbuffer *batch );
-
-
-/* Unlike bmBufferData, this currently requires the buffer be mapped.
- * Consider it a convenience function wrapping multple
- * intel_buffer_dword() calls.
- */
-void intel_batchbuffer_data(struct intel_batchbuffer *batch,
-			    const void *data,
-			    GLuint bytes,
-			    GLuint flags);
-
-void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
-				   GLuint bytes);
-
-
-/* Inline functions - might actually be better off with these
- * non-inlined.  Certainly better off switching all command packets to
- * be passed as structs rather than dwords, but that's a little bit of
- * work...
- */
-static inline GLuint 
-intel_batchbuffer_space( struct intel_batchbuffer *batch )
-{
-   return (BATCH_SZ - BATCH_RESERVED) - (batch->ptr - (batch->map + batch->offset));
-}
-
-
-static inline void 
-intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch,
-			     GLuint dword)
-{
-   assert(batch->map);
-   assert(intel_batchbuffer_space(batch) >= 4);
-   *(GLuint *)(batch->ptr) = dword;
-   batch->ptr += 4;
-}
-
-static inline void 
-intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
-				GLuint sz,
-				GLuint flags)
-{
-   assert(sz < BATCH_SZ - 8);
-   if (intel_batchbuffer_space(batch) < sz ||
-       (batch->flags != 0 && flags != 0 && batch->flags != flags))
-      intel_batchbuffer_flush(batch);
-   
-   batch->flags |= flags;
-}
-
-void intel_batchbuffer_align( struct intel_batchbuffer *batch,
-			      GLuint align,
-			      GLuint sz );
-
-
-/* Here are the crusty old macros, to be removed:
- */
-#define BATCH_LOCALS 
-#define BEGIN_BATCH(n, flags) intel_batchbuffer_require_space(intel->batch, n*4, flags)
-#define OUT_BATCH(d)  intel_batchbuffer_emit_dword(intel->batch, d)
-#define ADVANCE_BATCH() do { } while(0)
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/intel_blit.c b/src/mesa/drivers/dri/i965/intel_blit.c
index f88cbb2328d..dd6c8d17c28 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_blit.c
+++ b/src/mesa/drivers/dri/i965/intel_blit.c
@@ -1,617 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <errno.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "enums.h"
-#include "vblank.h"
-
-#include "intel_reg.h"
-#include "intel_batchbuffer.h"
-#include "intel_context.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "intel_structs.h"
-
-#include "bufmgr.h"
-
-
-
-
-/*
- * Copy the back buffer to the front buffer. 
- */
-void intelCopyBuffer( const __DRIdrawablePrivate *dPriv,
-		      const drm_clip_rect_t *rect ) 
-{
-   struct intel_context *intel;
-   GLboolean   missed_target;
-   int64_t ust;
-
-   DBG("%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-   intelFlush( &intel->ctx );
-
-
-   bmFinishFenceLock(intel, intel->last_swap_fence);
-
-   /* The LOCK_HARDWARE is required for the cliprects.  Buffer offsets
-    * should work regardless.
-    */
-   LOCK_HARDWARE( intel );
-
-   if (!rect)
-   {
-       UNLOCK_HARDWARE( intel );
-       driWaitForVBlank( dPriv, &intel->vbl_seq, intel->vblank_flags, & missed_target );
-       LOCK_HARDWARE( intel );
-   }
-
-   {
-      intelScreenPrivate *intelScreen = intel->intelScreen;
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      int nbox = dPriv->numClipRects;
-      drm_clip_rect_t *pbox = dPriv->pClipRects;
-      int cpp = intelScreen->cpp;
-      struct intel_region *src, *dst;
-      int BR13, CMD;
-      int i;
-      int src_pitch, dst_pitch;
-
-      if (intel->sarea->pf_current_page == 0) {
-	 dst = intel->front_region;
-	 src = intel->back_region;
-      }
-      else {
-	 assert(0);
-	 src = intel->front_region;
-	 dst = intel->back_region;
-      }
-
-      src_pitch = src->pitch * src->cpp;
-      dst_pitch = dst->pitch * dst->cpp;
-
-      if (cpp == 2) {
-	 BR13 = (0xCC << 16) | (1<<24);
-	 CMD = XY_SRC_COPY_BLT_CMD;
-      } 
-      else {
-	 BR13 = (0xCC << 16) | (1<<24) | (1<<25);
-	 CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-		XY_SRC_COPY_BLT_WRITE_RGB);
-      }
-
-      if (src->tiled) {
-	 CMD |= XY_SRC_TILED;
-	 src_pitch /= 4;
-      }
-      
-      if (dst->tiled) {
-	 CMD |= XY_DST_TILED;
- 	 dst_pitch /= 4;
-      }
-  
-      for (i = 0 ; i < nbox; i++, pbox++) 
-      {
-	 drm_clip_rect_t tmp = *pbox;
-
-	 if (rect) {
-	    if (!intel_intersect_cliprects(&tmp, &tmp, rect))
-	       continue;
-	 }
-
-
-	 if (tmp.x1 > tmp.x2 ||
-	     tmp.y1 > tmp.y2 ||
-	     tmp.x2 > intelScreen->width ||
-	     tmp.y2 > intelScreen->height)
-	    continue;
- 
-	 BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS);
-	 OUT_BATCH( CMD );
-	 OUT_BATCH( dst_pitch | BR13 );
-	 OUT_BATCH( (tmp.y1 << 16) | tmp.x1 );
-	 OUT_BATCH( (tmp.y2 << 16) | tmp.x2 );
-	 OUT_BATCH( bmBufferOffset(intel, dst->buffer) );
-	 OUT_BATCH( (tmp.y1 << 16) | tmp.x1 );
-	 OUT_BATCH( src_pitch );
-	 OUT_BATCH( bmBufferOffset(intel, src->buffer) ); 
-	 ADVANCE_BATCH();
-      }
-   }
-
-   intel_batchbuffer_flush( intel->batch );
-   intel->second_last_swap_fence = intel->last_swap_fence;
-   intel->last_swap_fence = bmSetFenceLock( intel );
-   UNLOCK_HARDWARE( intel );
-
-   if (!rect)
-   {
-       intel->swap_count++;
-       (*dri_interface->getUST)(&ust);
-       if (missed_target) {
-	   intel->swap_missed_count++;
-	   intel->swap_missed_ust = ust -  intel->swap_ust;
-       }
-   
-       intel->swap_ust = ust;
-   }
-
-}
-
-
-
-
-void intelEmitFillBlit( struct intel_context *intel,
-			GLuint cpp,
-			GLshort dst_pitch,
-			struct buffer *dst_buffer,
-			GLuint dst_offset,
-			GLboolean dst_tiled,
-			GLshort x, GLshort y, 
-			GLshort w, GLshort h,
-			GLuint color )
-{
-   GLuint BR13, CMD;
-   BATCH_LOCALS;
-
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = (0xF0 << 16) | (1<<24);
-      CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (0xF0 << 16) | (1<<24) | (1<<25);
-      CMD = (XY_COLOR_BLT_CMD | XY_COLOR_BLT_WRITE_ALPHA |
-	     XY_COLOR_BLT_WRITE_RGB);
-      break;
-   default:
-      return;
-   }
-
-   if (dst_tiled) {
-      CMD |= XY_DST_TILED;
-      dst_pitch /= 4;
-   }
-
-   BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS);
-   OUT_BATCH( CMD );
-   OUT_BATCH( dst_pitch | BR13 );
-   OUT_BATCH( (y << 16) | x );
-   OUT_BATCH( ((y+h) << 16) | (x+w) );
-   OUT_BATCH( bmBufferOffset(intel, dst_buffer) + dst_offset );
-   OUT_BATCH( color );
-   ADVANCE_BATCH();
-}
-
-static GLuint translate_raster_op(GLenum logicop)
-{
-   switch(logicop) {
-   case GL_CLEAR: return 0x00;
-   case GL_AND: return 0x88;
-   case GL_AND_REVERSE: return 0x44;
-   case GL_COPY: return 0xCC;
-   case GL_AND_INVERTED: return 0x22;
-   case GL_NOOP: return 0xAA;
-   case GL_XOR: return 0x66;
-   case GL_OR: return 0xEE;
-   case GL_NOR: return 0x11;
-   case GL_EQUIV: return 0x99;
-   case GL_INVERT: return 0x55;
-   case GL_OR_REVERSE: return 0xDD;
-   case GL_COPY_INVERTED: return 0x33;
-   case GL_OR_INVERTED: return 0xBB;
-   case GL_NAND: return 0x77;
-   case GL_SET: return 0xFF;
-   default: return 0;
-   }
-}
-
-
-/* Copy BitBlt
- */
-void intelEmitCopyBlit( struct intel_context *intel,
-			GLuint cpp,
-			GLshort src_pitch,
-			struct buffer *src_buffer,
-			GLuint  src_offset,
-			GLboolean src_tiled,
-			GLshort dst_pitch,
-			struct buffer *dst_buffer,
-			GLuint  dst_offset,
-			GLboolean dst_tiled,
-			GLshort src_x, GLshort src_y,
-			GLshort dst_x, GLshort dst_y,
-			GLshort w, GLshort h,
-			GLenum logic_op )
-{
-   GLuint CMD, BR13;
-   int dst_y2 = dst_y + h;
-   int dst_x2 = dst_x + w;
-   BATCH_LOCALS;
-
-
-   DBG("%s src:buf(%d)/%d %d,%d dst:buf(%d)/%d %d,%d sz:%dx%d op:%d\n",
-       __FUNCTION__,
-       src_buffer, src_pitch, src_x, src_y,
-       dst_buffer, dst_pitch, dst_x, dst_y,
-       w,h,logic_op);
-
-   assert( logic_op - GL_CLEAR >= 0 );
-   assert( logic_op - GL_CLEAR < 0x10 );
-      
-   src_pitch *= cpp;
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = (translate_raster_op(logic_op) << 16) | (1<<24);
-      CMD = XY_SRC_COPY_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (translate_raster_op(logic_op) << 16) | (1<<24) |
-	  (1<<25);
-      CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-	     XY_SRC_COPY_BLT_WRITE_RGB);
-      break;
-   default:
-      return;
-   }
-
-   if (src_tiled) {
-      CMD |= XY_SRC_TILED;
-      src_pitch /= 4;
-   }
-   
-   if (dst_tiled) {
-      CMD |= XY_DST_TILED;
-      dst_pitch /= 4;
-   }
-
-   if (dst_y2 < dst_y ||
-       dst_x2 < dst_x) {
-      return;
-   }
-
-   dst_pitch &= 0xffff;
-   src_pitch &= 0xffff;
-
-   /* Initial y values don't seem to work with negative pitches.  If
-    * we adjust the offsets manually (below), it seems to work fine.
-    *
-    * On the other hand, if we always adjust, the hardware doesn't
-    * know which blit directions to use, so overlapping copypixels get
-    * the wrong result.
-    */
-   if (dst_pitch > 0 && src_pitch > 0) {
-      BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS);
-      OUT_BATCH( CMD );
-      OUT_BATCH( dst_pitch | BR13 );
-      OUT_BATCH( (dst_y << 16) | dst_x );
-      OUT_BATCH( (dst_y2 << 16) | dst_x2 );
-      OUT_BATCH( bmBufferOffset(intel, dst_buffer) + dst_offset );	
-      OUT_BATCH( (src_y << 16) | src_x );
-      OUT_BATCH( src_pitch );
-      OUT_BATCH( bmBufferOffset(intel, src_buffer) + src_offset ); 
-      ADVANCE_BATCH();
-   }
-   else {
-      BEGIN_BATCH(8, INTEL_BATCH_NO_CLIPRECTS);
-      OUT_BATCH( CMD );
-      OUT_BATCH( (dst_pitch & 0xffff) | BR13 );
-      OUT_BATCH( (0 << 16) | dst_x );
-      OUT_BATCH( (h << 16) | dst_x2 );
-      OUT_BATCH( bmBufferOffset(intel, dst_buffer) + dst_offset + dst_y * dst_pitch );	
-      OUT_BATCH( (0 << 16) | src_x );
-      OUT_BATCH( (src_pitch & 0xffff) );
-      OUT_BATCH( bmBufferOffset(intel, src_buffer) + src_offset + src_y * src_pitch ); 
-      ADVANCE_BATCH();
-   }
-}
-
-
-
-void intelClearWithBlit(GLcontext *ctx, GLbitfield flags)
-{
-   struct intel_context *intel = intel_context( ctx );
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   GLuint clear_depth, clear_color;
-   GLint cx, cy, cw, ch;
-   GLint cpp = intelScreen->cpp;
-   GLboolean all;
-   GLint i;
-   struct intel_region *front = intel->front_region;
-   struct intel_region *back = intel->back_region;
-   struct intel_region *depth = intel->depth_region;
-   GLuint BR13, FRONT_CMD, BACK_CMD, DEPTH_CMD;
-   GLuint front_pitch;
-   GLuint back_pitch;
-   GLuint depth_pitch;
-   BATCH_LOCALS;
-
-   
-   clear_color = intel->ClearColor;
-   clear_depth = 0;
-
-   if (flags & BUFFER_BIT_DEPTH) {
-      clear_depth = (GLuint)(ctx->Depth.Clear * intel->ClearDepth);
-   }
-
-   if (flags & BUFFER_BIT_STENCIL) {
-      clear_depth |= (ctx->Stencil.Clear & 0xff) << 24;
-   }
-
-   switch(cpp) {
-   case 2: 
-      BR13 = (0xF0 << 16) | (1<<24);
-      BACK_CMD  = FRONT_CMD = XY_COLOR_BLT_CMD;
-      DEPTH_CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (0xF0 << 16) | (1<<24) | (1<<25);
-      BACK_CMD = FRONT_CMD = (XY_COLOR_BLT_CMD |
-			      XY_COLOR_BLT_WRITE_ALPHA | 
-			      XY_COLOR_BLT_WRITE_RGB);
-      DEPTH_CMD = XY_COLOR_BLT_CMD;
-      if (flags & BUFFER_BIT_DEPTH) DEPTH_CMD |= XY_COLOR_BLT_WRITE_RGB;
-      if (flags & BUFFER_BIT_STENCIL) DEPTH_CMD |= XY_COLOR_BLT_WRITE_ALPHA;
-      break;
-   default:
-      return;
-   }
-
-
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-   {
-      /* get clear bounds after locking */
-      cx = ctx->DrawBuffer->_Xmin;
-      cy = ctx->DrawBuffer->_Ymin;
-      ch = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-      cw = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-      all = (cw == ctx->DrawBuffer->Width && ch == ctx->DrawBuffer->Height);
-
-      /* flip top to bottom */
-      cy = intel->driDrawable->h - cy - ch;
-      cx = cx + intel->drawX;
-      cy += intel->drawY;
-
-      /* adjust for page flipping */
-      if ( intel->sarea->pf_current_page == 0 ) {
-	 front = intel->front_region;
-	 back = intel->back_region;
-      } 
-      else {
-	 back = intel->front_region;
-	 front = intel->back_region;
-      }
-      
-      front_pitch = front->pitch * front->cpp;
-      back_pitch = back->pitch * back->cpp;
-      depth_pitch = depth->pitch * depth->cpp;
-      
-      if (front->tiled) {
-	 FRONT_CMD |= XY_DST_TILED;
-	 front_pitch /= 4;
-      }
-
-      if (back->tiled) {
-	 BACK_CMD |= XY_DST_TILED;
-	 back_pitch /= 4;
-      }
-
-      if (depth->tiled) {
-	 DEPTH_CMD |= XY_DST_TILED;
-	 depth_pitch /= 4;
-      }
-
-      for (i = 0 ; i < intel->numClipRects ; i++) 
-      { 	 
-	 drm_clip_rect_t *box = &intel->pClipRects[i];	 
-	 drm_clip_rect_t b;
-
-	 if (!all) {
-	    GLint x = box->x1;
-	    GLint y = box->y1;
-	    GLint w = box->x2 - x;
-	    GLint h = box->y2 - y;
-
-	    if (x < cx) w -= cx - x, x = cx; 
-	    if (y < cy) h -= cy - y, y = cy;
-	    if (x + w > cx + cw) w = cx + cw - x;
-	    if (y + h > cy + ch) h = cy + ch - y;
-	    if (w <= 0) continue;
-	    if (h <= 0) continue;
-
-	    b.x1 = x;
-	    b.y1 = y;
-	    b.x2 = x + w;
-	    b.y2 = y + h;      
-	 } else {
-	    b = *box;
-	 }
-
-
-	 if (b.x1 > b.x2 ||
-	     b.y1 > b.y2 ||
-	     b.x2 > intelScreen->width ||
-	     b.y2 > intelScreen->height)
-	    continue;
-
-	 if ( flags & BUFFER_BIT_FRONT_LEFT ) {	    
-	    BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS);
-	    OUT_BATCH( FRONT_CMD );
-	    OUT_BATCH( front_pitch | BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( bmBufferOffset(intel, front->buffer) );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
-
-	 if ( flags & BUFFER_BIT_BACK_LEFT ) {
-	    BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS); 
-	    OUT_BATCH( BACK_CMD );
-	    OUT_BATCH( back_pitch | BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( bmBufferOffset(intel, back->buffer) );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
-
-	 if ( flags & (BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH) ) {
-	    BEGIN_BATCH(6, INTEL_BATCH_NO_CLIPRECTS);
-	    OUT_BATCH( DEPTH_CMD );
-	    OUT_BATCH( depth_pitch | BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( bmBufferOffset(intel, depth->buffer) );
-	    OUT_BATCH( clear_depth );
-	    ADVANCE_BATCH();
-	 }      
-      }
-   }
-   intel_batchbuffer_flush( intel->batch );
-   UNLOCK_HARDWARE( intel );
-}
-
-
-
-#define BR13_565  0x1
-#define BR13_8888 0x3
-
-
-void
-intelEmitImmediateColorExpandBlit(struct intel_context *intel,
-				  GLuint cpp,
-				  GLubyte *src_bits, GLuint src_size,
-				  GLuint fg_color,
-				  GLshort dst_pitch,
-				  struct buffer *dst_buffer,
-				  GLuint dst_offset,
-				  GLboolean dst_tiled,
-				  GLshort x, GLshort y, 
-				  GLshort w, GLshort h,
-				  GLenum logic_op)
-{
-   struct xy_setup_blit setup;
-   struct xy_text_immediate_blit text;
-   int dwords = ((src_size + 7) & ~7) / 4;
-
-   assert( logic_op - GL_CLEAR >= 0 );
-   assert( logic_op - GL_CLEAR < 0x10 );
-
-   if (w < 0 || h < 0) 
-      return;
-
-   dst_pitch *= cpp;
-
-   if (dst_tiled) 
-      dst_pitch /= 4;
-
-   DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d, %d bytes %d dwords\n",
-       __FUNCTION__,
-       dst_buffer, dst_pitch, dst_offset, x, y, w, h, src_size, dwords);
-
-   memset(&setup, 0, sizeof(setup));
-   
-   setup.br0.client = CLIENT_2D;
-   setup.br0.opcode = OPCODE_XY_SETUP_BLT;
-   setup.br0.write_alpha = (cpp == 4);
-   setup.br0.write_rgb = (cpp == 4);
-   setup.br0.dst_tiled = dst_tiled;
-   setup.br0.length = (sizeof(setup) / sizeof(int)) - 2;
-      
-   setup.br13.dest_pitch = dst_pitch;
-   setup.br13.rop = translate_raster_op(logic_op);
-   setup.br13.color_depth = (cpp == 4) ? BR13_8888 : BR13_565;
-   setup.br13.clipping_enable = 0;
-   setup.br13.mono_source_transparency = 1;
-
-   setup.dw2.clip_y1 = 0;
-   setup.dw2.clip_x1 = 0;
-   setup.dw3.clip_y2 = 100;
-   setup.dw3.clip_x2 = 100;
-
-   setup.dest_base_addr = bmBufferOffset(intel, dst_buffer) + dst_offset;
-   setup.background_color = 0;
-   setup.foreground_color = fg_color;
-   setup.pattern_base_addr = 0;
-
-   memset(&text, 0, sizeof(text));
-   text.dw0.client = CLIENT_2D;
-   text.dw0.opcode = OPCODE_XY_TEXT_IMMEDIATE_BLT;
-   text.dw0.pad0 = 0;
-   text.dw0.byte_packed = 1;	/* ?maybe? */
-   text.dw0.pad1 = 0;
-   text.dw0.dst_tiled = dst_tiled;
-   text.dw0.pad2 = 0;
-   text.dw0.length = (sizeof(text)/sizeof(int)) - 2 + dwords;
-   text.dw1.dest_y1 = y;	/* duplicates info in setup blit */
-   text.dw1.dest_x1 = x;
-   text.dw2.dest_y2 = y + h;
-   text.dw2.dest_x2 = x + w;
-
-   intel_batchbuffer_require_space( intel->batch,
-				    sizeof(setup) + 
-				    sizeof(text) + 
-				    dwords,
-				    INTEL_BATCH_NO_CLIPRECTS );
-
-   intel_batchbuffer_data( intel->batch,
-			   &setup,
-			   sizeof(setup),
-			   INTEL_BATCH_NO_CLIPRECTS );
-
-   intel_batchbuffer_data( intel->batch,
-			   &text,
-			   sizeof(text),
-			   INTEL_BATCH_NO_CLIPRECTS );
-
-   intel_batchbuffer_data( intel->batch,
-			   src_bits,
-			   dwords * 4,
-			   INTEL_BATCH_NO_CLIPRECTS );
-}
-
+../intel/intel_blit.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_buffer_objects.c b/src/mesa/drivers/dri/i965/intel_buffer_objects.c
index 015e433fd7a..e06dd3c8d3c 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/i965/intel_buffer_objects.c
@@ -1,207 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "imports.h"
-#include "mtypes.h"
-#include "bufferobj.h"
-
-#include "intel_context.h"
-#include "intel_buffer_objects.h"
-#include "bufmgr.h"
-
-
-/**
- * There is some duplication between mesa's bufferobjects and our
- * bufmgr buffers.  Both have an integer handle and a hashtable to
- * lookup an opaque structure.  It would be nice if the handles and
- * internal structure where somehow shared.
- */
-static struct gl_buffer_object *intel_bufferobj_alloc( GLcontext *ctx, 
-						       GLuint name, 
-						       GLenum target )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *obj = MALLOC_STRUCT(intel_buffer_object);
-
-   _mesa_initialize_buffer_object(&obj->Base, name, target);
-
-   /* XXX:  We generate our own handle, which is different to 'name' above.
-    */
-   bmGenBuffers(intel, "bufferobj", 1, &obj->buffer, 6);
-   assert(obj->buffer);
-
-   return &obj->Base;
-}
-
-
-/**
- * Deallocate/free a vertex/pixel buffer object.
- * Called via glDeleteBuffersARB().
- */
-static void intel_bufferobj_free( GLcontext *ctx, 
-				  struct gl_buffer_object *obj )
-{ 
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   assert(intel_obj);
-
-   if (intel_obj->buffer) 
-      bmDeleteBuffers( intel, 1, &intel_obj->buffer );
-  
-   _mesa_free(intel_obj);
-}
-
-
-
-/**
- * Allocate space for and store data in a buffer object.  Any data that was
- * previously stored in the buffer object is lost.  If data is NULL,
- * memory will be allocated, but no copy will occur.
- * Called via glBufferDataARB().
- */
-static void intel_bufferobj_data( GLcontext *ctx, 
-				  GLenum target, 
-				  GLsizeiptrARB size,
-				  const GLvoid *data, 
-				  GLenum usage,
-				  struct gl_buffer_object *obj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   /* XXX: do something useful with 'usage' (eg. populate flags
-    * argument below)
-    */
-   assert(intel_obj);
-
-   obj->Size = size;
-   obj->Usage = usage;
-
-   bmBufferDataAUB(intel, intel_obj->buffer, size, data, 0,
-		   0, 0);
-}
-
-
-/**
- * Replace data in a subrange of buffer object.  If the data range
- * specified by size + offset extends beyond the end of the buffer or
- * if data is NULL, no copy is performed.
- * Called via glBufferSubDataARB().
- */
-static void intel_bufferobj_subdata( GLcontext *ctx, 
-				     GLenum target, 
-				     GLintptrARB offset,
-				     GLsizeiptrARB size, 
-				     const GLvoid * data,
-				     struct gl_buffer_object * obj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   assert(intel_obj);
-   bmBufferSubDataAUB(intel, intel_obj->buffer, offset, size, data, 0, 0);
-}
-
-
-/**
- * Called via glGetBufferSubDataARB().
- */
-static void intel_bufferobj_get_subdata( GLcontext *ctx, 
-					 GLenum target, 
-					 GLintptrARB offset,
-					 GLsizeiptrARB size, 
-					 GLvoid * data,
-					 struct gl_buffer_object * obj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   assert(intel_obj);
-   bmBufferGetSubData(intel, intel_obj->buffer, offset, size, data);
-}
-
-
-
-/**
- * Called via glMapBufferARB().
- */
-static void *intel_bufferobj_map( GLcontext *ctx, 
-				  GLenum target, 
-				  GLenum access,
-				  struct gl_buffer_object *obj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   /* XXX: Translate access to flags arg below:
-    */
-   assert(intel_obj);
-   assert(intel_obj->buffer);
-   obj->Pointer = bmMapBuffer(intel, intel_obj->buffer, 0);
-   return obj->Pointer;
-}
-
-
-/**
- * Called via glMapBufferARB().
- */
-static GLboolean intel_bufferobj_unmap( GLcontext *ctx,
-					GLenum target,
-					struct gl_buffer_object *obj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
-
-   assert(intel_obj);
-   assert(intel_obj->buffer);
-   assert(obj->Pointer);
-   bmUnmapBufferAUB(intel, intel_obj->buffer, 0, 0);
-   obj->Pointer = NULL;
-   return GL_TRUE;
-}
-
-struct buffer *intel_bufferobj_buffer( const struct intel_buffer_object *intel_obj )
-{
-   assert(intel_obj->Base.Name);
-   assert(intel_obj->buffer);
-   return intel_obj->buffer;
-}  
-
-void intel_bufferobj_init( struct intel_context *intel )
-{
-   GLcontext *ctx = &intel->ctx;
-
-   ctx->Driver.NewBufferObject = intel_bufferobj_alloc;
-   ctx->Driver.DeleteBuffer = intel_bufferobj_free;
-   ctx->Driver.BufferData = intel_bufferobj_data;
-   ctx->Driver.BufferSubData = intel_bufferobj_subdata;
-   ctx->Driver.GetBufferSubData = intel_bufferobj_get_subdata;
-   ctx->Driver.MapBuffer = intel_bufferobj_map;
-   ctx->Driver.UnmapBuffer = intel_bufferobj_unmap;
-}
+../intel/intel_buffer_objects.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_buffers.c b/src/mesa/drivers/dri/i965/intel_buffers.c
index d155c039d77..c86daa49f47 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_buffers.c
+++ b/src/mesa/drivers/dri/i965/intel_buffers.c
@@ -1,547 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "intel_batchbuffer.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "macros.h"
-#include "swrast/swrast.h"
-
-GLboolean intel_intersect_cliprects( drm_clip_rect_t *dst,
-				     const drm_clip_rect_t *a,
-				     const drm_clip_rect_t *b )
-{
-   dst->x1 = MAX2(a->x1, b->x1);
-   dst->x2 = MIN2(a->x2, b->x2);
-   dst->y1 = MAX2(a->y1, b->y1);
-   dst->y2 = MIN2(a->y2, b->y2);
-
-   return (dst->x1 <= dst->x2 &&
-	   dst->y1 <= dst->y2);
-}
-
-struct intel_region *intel_drawbuf_region( struct intel_context *intel )
-{
-   switch (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0]) {
-   case BUFFER_BIT_FRONT_LEFT:
-      return intel->front_region;
-   case BUFFER_BIT_BACK_LEFT:
-      return intel->back_region;
-   default:
-      /* Not necessary to fallback - could handle either NONE or
-       * FRONT_AND_BACK cases below.
-       */
-      return NULL;		
-   }
-}
-
-struct intel_region *intel_readbuf_region( struct intel_context *intel )
-{
-   GLcontext *ctx = &intel->ctx;
-
-   /* This will have to change to support EXT_fbo's, but is correct
-    * for now:
-    */
-   switch (ctx->ReadBuffer->_ColorReadBufferIndex) {
-   case BUFFER_FRONT_LEFT:
-      return intel->front_region;
-   case BUFFER_BACK_LEFT:
-      return intel->back_region;
-   default:
-      assert(0);
-      return NULL;
-   }
-}
-
-
-
-static void intelBufferSize(GLframebuffer *buffer,
-			    GLuint *width, 
-			    GLuint *height)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   struct intel_context *intel = intel_context(ctx);
-   /* Need to lock to make sure the driDrawable is uptodate.  This
-    * information is used to resize Mesa's software buffers, so it has
-    * to be correct.
-    */
-   LOCK_HARDWARE(intel);
-   if (intel->driDrawable) {
-      *width = intel->driDrawable->w;
-      *height = intel->driDrawable->h;
-   }
-   else {
-      *width = 0;
-      *height = 0;
-   }
-   UNLOCK_HARDWARE(intel);
-}
-
-
-static void intelSetFrontClipRects( struct intel_context *intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!dPriv) return;
-
-   intel->numClipRects = dPriv->numClipRects;
-   intel->pClipRects = dPriv->pClipRects;
-   intel->drawX = dPriv->x;
-   intel->drawY = dPriv->y;
-}
-
-
-static void intelSetBackClipRects( struct intel_context *intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!dPriv) return;
-
-   if (intel->sarea->pf_enabled == 0 && dPriv->numBackClipRects == 0) {
-      intel->numClipRects = dPriv->numClipRects;
-      intel->pClipRects = dPriv->pClipRects;
-      intel->drawX = dPriv->x;
-      intel->drawY = dPriv->y;
-   } else {
-      intel->numClipRects = dPriv->numBackClipRects;
-      intel->pClipRects = dPriv->pBackClipRects;
-      intel->drawX = dPriv->backX;
-      intel->drawY = dPriv->backY;
-      
-      if (dPriv->numBackClipRects == 1 &&
-	  dPriv->x == dPriv->backX &&
-	  dPriv->y == dPriv->backY) {
-      
-	 /* Repeat the calculation of the back cliprect dimensions here
-	  * as early versions of dri.a in the Xserver are incorrect.  Try
-	  * very hard not to restrict future versions of dri.a which
-	  * might eg. allocate truly private back buffers.
-	  */
-	 int x1, y1;
-	 int x2, y2;
-	 
-	 x1 = dPriv->x;
-	 y1 = dPriv->y;      
-	 x2 = dPriv->x + dPriv->w;
-	 y2 = dPriv->y + dPriv->h;
-	 
-	 if (x1 < 0) x1 = 0;
-	 if (y1 < 0) y1 = 0;
-	 if (x2 > intel->intelScreen->width) x2 = intel->intelScreen->width;
-	 if (y2 > intel->intelScreen->height) y2 = intel->intelScreen->height;
-
-	 if (x1 == dPriv->pBackClipRects[0].x1 &&
-	     y1 == dPriv->pBackClipRects[0].y1) {
-
-	    dPriv->pBackClipRects[0].x2 = x2;
-	    dPriv->pBackClipRects[0].y2 = y2;
-	 }
-      }
-   }
-}
-
-
-void intelWindowMoved( struct intel_context *intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!intel->ctx.DrawBuffer) {
-      intelSetFrontClipRects( intel );
-   }
-   else {
-      switch (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0]) {
-      case BUFFER_BIT_FRONT_LEFT:
-	 intelSetFrontClipRects( intel );
-	 break;
-      case BUFFER_BIT_BACK_LEFT:
-	 intelSetBackClipRects( intel );
-	 break;
-      default:
-	 /* glDrawBuffer(GL_NONE or GL_FRONT_AND_BACK): software fallback */
-	 intelSetFrontClipRects( intel );
-      }
-   }
-
-   _mesa_resize_framebuffer(&intel->ctx,
-			    (GLframebuffer*)dPriv->driverPrivate,
-			    dPriv->w, dPriv->h);
-
-   /* Set state we know depends on drawable parameters:
-    */
-   {
-      GLcontext *ctx = &intel->ctx;
-
-      if (ctx->Driver.Scissor)
-	 ctx->Driver.Scissor( ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			      ctx->Scissor.Width, ctx->Scissor.Height );
-      
-      if (ctx->Driver.DepthRange)
-	 ctx->Driver.DepthRange( ctx, 
-				 ctx->Viewport.Near,
-				 ctx->Viewport.Far );
-
-      intel->NewGLState |= _NEW_SCISSOR;
-   }
-
-   /* This works because the lock is always grabbed before emitting
-    * commands and commands are always flushed prior to releasing
-    * the lock.
-    */
-   intel->NewGLState |= _NEW_WINDOW_POS; 
-}
-
-
-
-/* A true meta version of this would be very simple and additionally
- * machine independent.  Maybe we'll get there one day.
- */
-static void intelClearWithTris(struct intel_context *intel, 
-			       GLbitfield mask)
-{
-   GLcontext *ctx = &intel->ctx;
-   drm_clip_rect_t clear;
-   GLint cx, cy, cw, ch;
-
-   if (INTEL_DEBUG & DEBUG_DRI)
-      _mesa_printf("%s %x\n", __FUNCTION__, mask);
-
-   {
-
-      intel->vtbl.install_meta_state(intel);
-
-      /* Get clear bounds after locking */
-      cx = ctx->DrawBuffer->_Xmin;
-      cy = ctx->DrawBuffer->_Ymin;
-      cw = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-      ch = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-
-      clear.x1 = cx;
-      clear.y1 = cy;
-      clear.x2 = cx + cw;
-      clear.y2 = cy + ch;
-
-      /* Back and stencil cliprects are the same.  Try and do both
-       * buffers at once:
-       */
-      if (mask & (BUFFER_BIT_BACK_LEFT|BUFFER_BIT_STENCIL|BUFFER_BIT_DEPTH)) { 
-	 intel->vtbl.meta_draw_region(intel, 
-				      intel->back_region,
-				      intel->depth_region );
-
-	 if (mask & BUFFER_BIT_BACK_LEFT)
-	    intel->vtbl.meta_color_mask(intel, GL_TRUE );
-	 else
-	    intel->vtbl.meta_color_mask(intel, GL_FALSE );
-
-	 if (mask & BUFFER_BIT_STENCIL) 
-	    intel->vtbl.meta_stencil_replace( intel, 
-					      intel->ctx.Stencil.WriteMask[0], 
-					      intel->ctx.Stencil.Clear);
-	 else
-	    intel->vtbl.meta_no_stencil_write(intel);
-
-	 if (mask & BUFFER_BIT_DEPTH) 
-	    intel->vtbl.meta_depth_replace( intel );
-	 else
-	    intel->vtbl.meta_no_depth_write(intel);
-      
-	 /* XXX: Using INTEL_BATCH_NO_CLIPRECTS here is dangerous as the
-	  * drawing origin may not be correctly emitted.
-	  */
-	 intel->vtbl.meta_draw_quad(intel, 
-				    clear.x1, clear.x2, 
-				    clear.y1, clear.y2, 
-				    intel->ctx.Depth.Clear,
-				    intel->clear_chan[0], 
-				    intel->clear_chan[1], 
-				    intel->clear_chan[2], 
-				    intel->clear_chan[3], 
-				    0, 0, 0, 0);
-      }
-
-      /* Front may have different cliprects: 
-       */
-      if (mask & BUFFER_BIT_FRONT_LEFT) {
-	 intel->vtbl.meta_no_depth_write(intel);
-	 intel->vtbl.meta_no_stencil_write(intel);
-	 intel->vtbl.meta_color_mask(intel, GL_TRUE );
-	 intel->vtbl.meta_draw_region(intel, 
-				      intel->front_region,
-				      intel->depth_region);
-
-	 /* XXX: Using INTEL_BATCH_NO_CLIPRECTS here is dangerous as the
-	  * drawing origin may not be correctly emitted.
-	  */
-	 intel->vtbl.meta_draw_quad(intel, 
-				    clear.x1, clear.x2, 
-				    clear.y1, clear.y2, 
-				    0,
-				    intel->clear_chan[0], 
-				    intel->clear_chan[1], 
-				    intel->clear_chan[2], 
-				    intel->clear_chan[3], 
-				    0, 0, 0, 0);
-      }
-
-      intel->vtbl.leave_meta_state( intel );
-   }
-}
-
-
-
-
-
-static void intelClear(GLcontext *ctx, GLbitfield mask)
-{
-   struct intel_context *intel = intel_context( ctx );
-   const GLuint colorMask = *((GLuint *) &ctx->Color.ColorMask);
-   GLbitfield tri_mask = 0;
-   GLbitfield blit_mask = 0;
-   GLbitfield swrast_mask = 0;
-
-   if (INTEL_DEBUG & DEBUG_DRI)
-      fprintf(stderr, "%s %x\n", __FUNCTION__, mask);
-
-
-   if (mask & BUFFER_BIT_FRONT_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_FRONT_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_FRONT_LEFT;
-      }
-   }
-
-   if (mask & BUFFER_BIT_BACK_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_BACK_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_BACK_LEFT;
-      }
-   }
-
-
-   if (mask & BUFFER_BIT_STENCIL) {
-      if (!intel->hw_stencil) {
-	 swrast_mask |= BUFFER_BIT_STENCIL;
-      }
-      else if ((ctx->Stencil.WriteMask[0] & 0xff) != 0xff ||
-	       intel->depth_region->tiled) {
-	 tri_mask |= BUFFER_BIT_STENCIL;
-      } 
-      else {
-	 blit_mask |= BUFFER_BIT_STENCIL;
-      }
-   }
-
-   /* Do depth with stencil if possible to avoid 2nd pass over the
-    * same buffer.
-    */
-   if (mask & BUFFER_BIT_DEPTH) {
-      if ((tri_mask & BUFFER_BIT_STENCIL) ||
-	  intel->depth_region->tiled)
-	 tri_mask |= BUFFER_BIT_DEPTH;
-      else 
-	 blit_mask |= BUFFER_BIT_DEPTH;
-   }
-
-   swrast_mask |= (mask & BUFFER_BIT_ACCUM);
-
-   intelFlush( ctx );
-
-   if (blit_mask)
-      intelClearWithBlit( ctx, blit_mask );
-
-   if (tri_mask) 
-      intelClearWithTris( intel, tri_mask );
-
-   if (swrast_mask)
-      _swrast_Clear( ctx, swrast_mask );
-}
-
-
-
-
-
-
-
-/* Flip the front & back buffers
- */
-static void intelPageFlip( const __DRIdrawablePrivate *dPriv )
-{
-#if 0
-   struct intel_context *intel;
-   int tmp, ret;
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-
-   if (dPriv->pClipRects) {
-      *(drm_clip_rect_t *)intel->sarea->boxes = dPriv->pClipRects[0];
-      intel->sarea->nbox = 1;
-   }
-
-   ret = drmCommandNone(intel->driFd, DRM_I830_FLIP); 
-   if (ret) {
-      fprintf(stderr, "%s: %d\n", __FUNCTION__, ret);
-      UNLOCK_HARDWARE( intel );
-      exit(1);
-   }
-
-   tmp = intel->sarea->last_enqueue;
-   intelRefillBatchLocked( intel );
-   UNLOCK_HARDWARE( intel );
-
-
-   intelSetDrawBuffer( &intel->ctx, intel->ctx.Color.DriverDrawBuffer );
-#endif
-}
-
-
-void intelSwapBuffers( __DRIdrawablePrivate *dPriv )
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      struct intel_context *intel;
-      GLcontext *ctx;
-      intel = (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-      ctx = &intel->ctx;
-      if (ctx->Visual.doubleBufferMode) {
-	 _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-	 if ( 0 /*intel->doPageFlip*/ ) { /* doPageFlip is never set !!! */
-	    intelPageFlip( dPriv );
-	 } else {
-	    intelCopyBuffer( dPriv, NULL );
-	 }
-	 if (intel->aub_file) {
-	    intelFlush(ctx);
-	    intel->vtbl.aub_dump_bmp( intel, 1 );
-
-	    intel->aub_wrap = 1;
-	 }
-      }
-   } else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-void intelCopySubBuffer( __DRIdrawablePrivate *dPriv,
-			 int x, int y, int w, int h )
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      struct intel_context *intel = dPriv->driContextPriv->driverPrivate;
-      GLcontext *ctx = &intel->ctx;
-
-      if (ctx->Visual.doubleBufferMode) {
-	 drm_clip_rect_t rect;
-	 rect.x1 = x + dPriv->x;
-	 rect.y1 = (dPriv->h - y - h) + dPriv->y;
-	 rect.x2 = rect.x1 + w;
-	 rect.y2 = rect.y1 + h;
-	 _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-	 intelCopyBuffer( dPriv, &rect );
-      }
-   } else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-
-static void intelDrawBuffer(GLcontext *ctx, GLenum mode )
-{
-   struct intel_context *intel = intel_context(ctx);
-   int front = 0;
- 
-   if (!ctx->DrawBuffer)
-      return;
-
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-      front = 1;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
-   case BUFFER_BIT_BACK_LEFT:
-      front = 0;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
-   default:
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_TRUE );
-      return;
-   }
-
-   if ( intel->sarea->pf_current_page == 1 ) 
-      front ^= 1;
-   
-   intelSetFrontClipRects( intel );
-
-
-   if (front) {
-      if (intel->draw_region != intel->front_region) {
-	 intel_region_release(intel, &intel->draw_region);
-	 intel_region_reference(&intel->draw_region, intel->front_region);
-      }
-   } else {
-      if (intel->draw_region != intel->back_region) {
-	 intel_region_release(intel, &intel->draw_region);
-	 intel_region_reference(&intel->draw_region, intel->back_region);
-      }
-   }
-
-   intel->vtbl.set_draw_region( intel, 
-				intel->draw_region,
-				intel->depth_region);
-}
-
-static void intelReadBuffer( GLcontext *ctx, GLenum mode )
-{
-   /* nothing, until we implement h/w glRead/CopyPixels or CopyTexImage */
-}
-
-
-
-void intelInitBufferFuncs( struct dd_function_table *functions )
-{
-   functions->Clear = intelClear;
-   functions->GetBufferSize = intelBufferSize;
-   functions->DrawBuffer = intelDrawBuffer;
-   functions->ReadBuffer = intelReadBuffer;
-}
+../intel/intel_buffers.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_context.c b/src/mesa/drivers/dri/i965/intel_context.c
index 4f51fefe0f6..27a1cbb255e 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_context.c
+++ b/src/mesa/drivers/dri/i965/intel_context.c
@@ -1,712 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "matrix.h"
-#include "simple_list.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "imports.h"
-#include "points.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "vbo/vbo.h"
-
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vertex.h"
-
-#include "drivers/common/driverfuncs.h"
-
-#include "intel_screen.h"
-
-#include "i830_dri.h"
-#include "i830_common.h"
-
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "intel_buffer_objects.h"
-
-#include "bufmgr.h"
-
-#include "utils.h"
-#include "vblank.h"
-#ifndef INTEL_DEBUG
-int INTEL_DEBUG = (0);
-#endif
-
-#define need_GL_ARB_multisample
-#define need_GL_ARB_point_parameters
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
-#define need_GL_ARB_vertex_program
-#define need_GL_ARB_window_pos
-#define need_GL_ARB_occlusion_query
-#define need_GL_EXT_blend_color
-#define need_GL_EXT_blend_equation_separate
-#define need_GL_EXT_blend_func_separate
-#define need_GL_EXT_blend_minmax
-#define need_GL_EXT_cull_vertex
-#define need_GL_EXT_fog_coord
-#define need_GL_EXT_multi_draw_arrays
-#define need_GL_EXT_secondary_color
-#include "extension_helper.h"
-
-#ifndef VERBOSE
-int VERBOSE = 0;
-#endif
-
-/***************************************
- * Mesa's Driver Functions
- ***************************************/
-
-#define DRIVER_VERSION                     "4.1.3002"
-
-static const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
-{
-   const char * chipset;
-   static char buffer[128];
-
-   switch (name) {
-   case GL_VENDOR:
-      return (GLubyte *)"Tungsten Graphics, Inc";
-      break;
-      
-   case GL_RENDERER:
-      switch (intel_context(ctx)->intelScreen->deviceID) {
-      case PCI_CHIP_I965_Q:
-	 chipset = "Intel(R) 965Q";
-         break;
-      case PCI_CHIP_I965_G:
-      case PCI_CHIP_I965_G_1:
-	 chipset = "Intel(R) 965G";
-         break;
-      case PCI_CHIP_I946_GZ:
-	 chipset = "Intel(R) 946GZ";
-         break;
-      case PCI_CHIP_I965_GM:
-	 chipset = "Intel(R) 965GM";
-         break;
-      case PCI_CHIP_I965_GME:
-	 chipset = "Intel(R) 965GME/GLE";
-         break;
-      default:
-	 chipset = "Unknown Intel Chipset";
-      }
-
-      (void) driGetRendererString( buffer, chipset, DRIVER_VERSION, 0 );
-      return (GLubyte *) buffer;
-
-   default:
-      return NULL;
-   }
-}
-
-
-/**
- * Extension strings exported by the intel driver.
- *
- * \note
- * It appears that ARB_texture_env_crossbar has "disappeared" compared to the
- * old i830-specific driver.
- */
-const struct dri_extension card_extensions[] =
-{
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
-    { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
-    { "GL_ARB_texture_border_clamp",       NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
-    { "GL_ARB_texture_cube_map",           NULL },
-    { "GL_ARB_texture_env_add",            NULL },
-    { "GL_ARB_texture_env_combine",        NULL },
-    { "GL_ARB_texture_env_dot3",           NULL },
-    { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_texture_non_power_of_two",   NULL },
-    { "GL_ARB_texture_rectangle",          NULL },
-    { "GL_NV_texture_rectangle",           NULL },
-    { "GL_EXT_texture_rectangle",          NULL },
-    { "GL_ARB_texture_rectangle",          NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
-    { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
-    { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
-    { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
-    { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
-    { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
-    { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
-    { "GL_EXT_blend_logic_op",             NULL },
-    { "GL_EXT_blend_subtract",             NULL },
-    { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
-    { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
-    { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
-    { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
-    { "GL_EXT_stencil_wrap",               NULL },
-    { "GL_EXT_texture_edge_clamp",         NULL },
-    { "GL_EXT_texture_env_combine",        NULL },
-    { "GL_EXT_texture_env_dot3",           NULL },
-    { "GL_EXT_texture_filter_anisotropic", NULL },
-    { "GL_EXT_texture_lod_bias",           NULL },
-    { "GL_3DFX_texture_compression_FXT1",  NULL },
-    { "GL_APPLE_client_storage",           NULL },
-    { "GL_MESA_pack_invert",               NULL },
-    { "GL_MESA_ycbcr_texture",             NULL },
-    { "GL_NV_blend_square",                NULL },
-    { "GL_SGIS_generate_mipmap",           NULL },
-    { NULL,                                NULL }
-};
-
-const struct dri_extension arb_oc_extension = 
-    { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions};
-
-void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
-{	     
-	struct intel_context *intel = ctx?intel_context(ctx):NULL;
-	driInitExtensions(ctx, card_extensions, enable_imaging);
-	if (!ctx || intel->intelScreen->drmMinor >= 8)
-		driInitSingleExtension (ctx, &arb_oc_extension);
-}
-
-static const struct dri_debug_control debug_control[] =
-{
-    { "fall",  DEBUG_FALLBACKS },
-    { "tex",   DEBUG_TEXTURE },
-    { "ioctl", DEBUG_IOCTL },
-    { "prim",  DEBUG_PRIMS },
-    { "vert",  DEBUG_VERTS },
-    { "state", DEBUG_STATE },
-    { "verb",  DEBUG_VERBOSE },
-    { "dri",   DEBUG_DRI },
-    { "dma",   DEBUG_DMA },
-    { "san",   DEBUG_SANITY },
-    { "sync",  DEBUG_SYNC },
-    { "sleep", DEBUG_SLEEP },
-    { "pix",   DEBUG_PIXEL },
-    { "buf",   DEBUG_BUFMGR },
-    { "stats", DEBUG_STATS },
-    { "tile",  DEBUG_TILE },
-    { "sing",  DEBUG_SINGLE_THREAD },
-    { "thre",  DEBUG_SINGLE_THREAD },
-    { "wm",    DEBUG_WM },
-    { "vs",    DEBUG_VS },
-    { NULL,    0 }
-};
-
-
-static void intelInvalidateState( GLcontext *ctx, GLuint new_state )
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   _tnl_invalidate_vertex_state( ctx, new_state );
-   
-   intel->NewGLState |= new_state;
-
-   if (intel->vtbl.invalidate_state)
-      intel->vtbl.invalidate_state( intel, new_state );
-}
-
-
-void intelFlush( GLcontext *ctx )
-{
-   struct intel_context *intel = intel_context( ctx );
-
-   bmLockAndFence(intel);
-}
-
-void intelFinish( GLcontext *ctx ) 
-{
-   struct intel_context *intel = intel_context( ctx );
-
-   bmFinishFence(intel, bmLockAndFence(intel));
-}
-
-static void
-intelBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	struct intel_context *intel = intel_context( ctx );
-	drmI830MMIO io = {
-		.read_write = MMIO_READ,
-		.reg = MMIO_REGS_PS_DEPTH_COUNT,
-		.data = &q->Result 
-	};
-	intel->stats_wm++;
-	intelFinish(&intel->ctx);
-	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
-}
-
-static void
-intelEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	struct intel_context *intel = intel_context( ctx );
-	GLuint64EXT tmp;	
-	drmI830MMIO io = {
-		.read_write = MMIO_READ,
-		.reg = MMIO_REGS_PS_DEPTH_COUNT,
-		.data = &tmp
-	};
-	intelFinish(&intel->ctx);
-	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
-	q->Result = tmp - q->Result;
-	q->Ready = GL_TRUE;
-	intel->stats_wm--;
-}
-
-
-void intelInitDriverFunctions( struct dd_function_table *functions )
-{
-   _mesa_init_driver_functions( functions );
-
-   functions->Flush = intelFlush;
-   functions->Finish = intelFinish;
-   functions->GetString = intelGetString;
-   functions->UpdateState = intelInvalidateState;
-   functions->BeginQuery = intelBeginQuery;
-   functions->EndQuery = intelEndQuery;
-
-   /* CopyPixels can be accelerated even with the current memory
-    * manager:
-    */
-   if (!getenv("INTEL_NO_BLIT")) {
-      functions->CopyPixels = intelCopyPixels;
-      functions->Bitmap = intelBitmap;
-   }
-
-   intelInitTextureFuncs( functions );
-   intelInitStateFuncs( functions );
-   intelInitBufferFuncs( functions );
-}
-
-
-
-GLboolean intelInitContext( struct intel_context *intel,
-			    const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate,
-			    struct dd_function_table *functions )
-{
-   GLcontext *ctx = &intel->ctx;
-   GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   volatile drmI830Sarea *saPriv = (volatile drmI830Sarea *)
-      (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
-
-   if (!_mesa_initialize_context(&intel->ctx,
-				 mesaVis, shareCtx, 
-				 functions,
-				 (void*) intel)) {
-      _mesa_printf("%s: failed to init mesa context\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   driContextPriv->driverPrivate = intel;
-   intel->intelScreen = intelScreen;
-   intel->driScreen = sPriv;
-   intel->sarea = saPriv;
-
-   driParseConfigFiles (&intel->optionCache, &intelScreen->optionCache,
-		   intel->driScreen->myNum, "i965");
-
-   intel->vblank_flags = (intel->intelScreen->irq_active != 0)
-	   ? driGetDefaultVBlankFlags(&intel->optionCache) : VBLANK_FLAG_NO_IRQ;
-
-   ctx->Const.MaxTextureMaxAnisotropy = 2.0;
-
-   if (getenv("INTEL_STRICT_CONFORMANCE")) {
-      intel->strict_conformance = 1;
-   }
-
-   if (intel->strict_conformance) {
-      ctx->Const.MinLineWidth = 1.0;
-      ctx->Const.MinLineWidthAA = 1.0;
-      ctx->Const.MaxLineWidth = 1.0;
-      ctx->Const.MaxLineWidthAA = 1.0;
-      ctx->Const.LineWidthGranularity = 1.0;
-   }
-   else {
-      ctx->Const.MinLineWidth = 1.0;
-      ctx->Const.MinLineWidthAA = 1.0;
-      ctx->Const.MaxLineWidth = 5.0;
-      ctx->Const.MaxLineWidthAA = 5.0;
-      ctx->Const.LineWidthGranularity = 0.5;
-   }
-
-   ctx->Const.MinPointSize = 1.0;
-   ctx->Const.MinPointSizeAA = 1.0;
-   ctx->Const.MaxPointSize = 255.0;
-   ctx->Const.MaxPointSizeAA = 3.0;
-   ctx->Const.PointSizeGranularity = 1.0;
-
-   /* reinitialize the context point state.
-    * It depend on constants in __GLcontextRec::Const
-    */
-   _mesa_init_point(ctx);
-
-   /* Initialize the software rasterizer and helper modules. */
-   _swrast_CreateContext( ctx );
-   _vbo_CreateContext( ctx );
-   _tnl_CreateContext( ctx );
-   _swsetup_CreateContext( ctx );
-
-   TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
-
-   /* Configure swrast to match hardware characteristics: */
-   _swrast_allow_pixel_fog( ctx, GL_FALSE );
-   _swrast_allow_vertex_fog( ctx, GL_TRUE );
-
-   /* Dri stuff */
-   intel->hHWContext = driContextPriv->hHWContext;
-   intel->driFd = sPriv->fd;
-   intel->driHwLock = (drmLock *) &sPriv->pSAREA->lock;
-
-   intel->hw_stencil = mesaVis->stencilBits && mesaVis->depthBits == 24;
-   intel->hw_stipple = 1;
-
-   switch(mesaVis->depthBits) {
-   case 0:			/* what to do in this case? */
-   case 16:
-      intel->depth_scale = 1.0/0xffff;
-      intel->polygon_offset_scale = 1.0/0xffff;
-      intel->depth_clear_mask = ~0;
-      intel->ClearDepth = 0xffff;
-      break;
-   case 24:
-      intel->depth_scale = 1.0/0xffffff;
-      intel->polygon_offset_scale = 2.0/0xffffff; /* req'd to pass glean */
-      intel->depth_clear_mask = 0x00ffffff;
-      intel->stencil_clear_mask = 0xff000000;
-      intel->ClearDepth = 0x00ffffff;
-      break;
-   default:
-      assert(0); 
-      break;
-   }
-
-   /* Initialize swrast, tnl driver tables: */
-   intelInitSpanFuncs( ctx );
-
-   intel->no_hw = getenv("INTEL_NO_HW") != NULL;
-
-   if (!intel->intelScreen->irq_active) {
-      _mesa_printf("IRQs not active.  Exiting\n");
-      exit(1);
-   }
-   intelInitExtensions(ctx, GL_TRUE); 
-
-   INTEL_DEBUG  = driParseDebugString( getenv( "INTEL_DEBUG" ),
-				       debug_control );
-
-
-   /* Buffer manager: 
-    */
-   intel->bm = bm_fake_intel_Attach( intel );
-
-
-   bmInitPool(intel,
-	      intel->intelScreen->tex.offset, /* low offset */
-	      intel->intelScreen->tex.map, /* low virtual */
-	      intel->intelScreen->tex.size,
-	      BM_MEM_AGP);
-
-   /* These are still static, but create regions for them.  
-    */
-   intel->front_region = 
-      intel_region_create_static(intel,
-				 BM_MEM_AGP,
-				 intelScreen->front.offset,
-				 intelScreen->front.map,
-				 intelScreen->cpp,
-				 intelScreen->front.pitch / intelScreen->cpp,
-				 intelScreen->height,
-				 intelScreen->front.size,
-				 intelScreen->front.tiled != 0);
-
-   intel->back_region = 
-      intel_region_create_static(intel,
-				 BM_MEM_AGP,
-				 intelScreen->back.offset,
-				 intelScreen->back.map,
-				 intelScreen->cpp,
-				 intelScreen->back.pitch / intelScreen->cpp,
-				 intelScreen->height,
-				 intelScreen->back.size,
-                                 intelScreen->back.tiled != 0);
-
-   /* Still assuming front.cpp == depth.cpp
-    *
-    * XXX: Setting tiling to false because Depth tiling only supports
-    * YMAJOR but the blitter only supports XMAJOR tiling.  Have to
-    * resolve later.
-    */
-   intel->depth_region = 
-      intel_region_create_static(intel,
-				 BM_MEM_AGP,
-				 intelScreen->depth.offset,
-				 intelScreen->depth.map,
-				 intelScreen->cpp,
-				 intelScreen->depth.pitch / intelScreen->cpp,
-				 intelScreen->height,
-				 intelScreen->depth.size,
-                                 intelScreen->depth.tiled != 0);
-   
-   intel_bufferobj_init( intel );
-   intel->batch = intel_batchbuffer_alloc( intel );
-
-   if (intel->ctx.Mesa_DXTn) {
-      _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
-      _mesa_enable_extension( ctx, "GL_S3_s3tc" );
-   }
-   else if (driQueryOptionb (&intelScreen->optionCache, "force_s3tc_enable")) {
-      _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
-   }
-
-/*    driInitTextureObjects( ctx, & intel->swapped, */
-/* 			  DRI_TEXMGR_DO_TEXTURE_1D | */
-/* 			  DRI_TEXMGR_DO_TEXTURE_2D |  */
-/* 			  DRI_TEXMGR_DO_TEXTURE_RECT ); */
-
-
-   if (getenv("INTEL_NO_RAST")) {
-      fprintf(stderr, "disabling 3D rasterization\n");
-      intel->no_rast = 1;
-   }
-
-
-   return GL_TRUE;
-}
-
-void intelDestroyContext(__DRIcontextPrivate *driContextPriv)
-{
-   struct intel_context *intel = (struct intel_context *) driContextPriv->driverPrivate;
-
-   assert(intel); /* should never be null */
-   if (intel) {
-      GLboolean   release_texture_heaps;
-
-
-      intel->vtbl.destroy( intel );
-
-      release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
-      _swsetup_DestroyContext (&intel->ctx);
-      _tnl_DestroyContext (&intel->ctx);
-      _vbo_DestroyContext (&intel->ctx);
-
-      _swrast_DestroyContext (&intel->ctx);
-      intel->Fallback = 0;	/* don't call _swrast_Flush later */
-      intel_batchbuffer_free(intel->batch);
-      intel->batch = NULL;
-      
-
-      if ( release_texture_heaps ) {
-         /* This share group is about to go away, free our private
-          * texture object data.
-          */
-
-	 /* XXX: destroy the shared bufmgr struct here?
-	  */
-      }
-
-      /* Free the regions created to describe front/back/depth
-       * buffers:
-       */
-#if 0
-      intel_region_release(intel, &intel->front_region);
-      intel_region_release(intel, &intel->back_region);
-      intel_region_release(intel, &intel->depth_region);
-      intel_region_release(intel, &intel->draw_region);
-#endif
-
-      /* free the Mesa context */
-      _mesa_destroy_context(&intel->ctx);
-   }
-
-   driContextPriv->driverPrivate = NULL;
-}
-
-GLboolean intelUnbindContext(__DRIcontextPrivate *driContextPriv)
-{
-   return GL_TRUE;
-}
-
-GLboolean intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-			  __DRIdrawablePrivate *driDrawPriv,
-			  __DRIdrawablePrivate *driReadPriv)
-{
-
-   if (driContextPriv) {
-      struct intel_context *intel = (struct intel_context *) driContextPriv->driverPrivate;
-
-      if ( intel->driDrawable != driDrawPriv ) {
-	 /* Shouldn't the readbuffer be stored also? */
-	 driDrawableInitVBlank( driDrawPriv, intel->vblank_flags,
-		      &intel->vbl_seq );
-
-	 intel->driDrawable = driDrawPriv;
-	 intelWindowMoved( intel );
-      }
-
-      _mesa_make_current(&intel->ctx,
-			 (GLframebuffer *) driDrawPriv->driverPrivate,
-			 (GLframebuffer *) driReadPriv->driverPrivate);
-
-      intel->ctx.Driver.DrawBuffer( &intel->ctx, intel->ctx.Color.DrawBuffer[0] );
-   } else {
-      _mesa_make_current(NULL, NULL, NULL);
-   }
-
-   return GL_TRUE;
-}
-
-
-static void intelContendedLock( struct intel_context *intel, GLuint flags )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   __DRIscreenPrivate *sPriv = intel->driScreen;
-   volatile drmI830Sarea * sarea = intel->sarea;
-   int me = intel->hHWContext;
-   int my_bufmgr = bmCtxId(intel);
-
-   drmGetLock(intel->driFd, intel->hHWContext, flags);
-
-   /* If the window moved, may need to set a new cliprect now.
-    *
-    * NOTE: This releases and regains the hw lock, so all state
-    * checking must be done *after* this call:
-    */
-   if (dPriv)
-      DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
-
-
-   intel->locked = 1;
-   intel->need_flush = 1;
-
-   /* Lost context?
-    */
-   if (sarea->ctxOwner != me) {
-      DBG("Lost Context: sarea->ctxOwner %x me %x\n", sarea->ctxOwner, me);
-      sarea->ctxOwner = me;
-      intel->vtbl.lost_hardware( intel );
-   }
-
-   /* As above, but don't evict the texture data on transitions
-    * between contexts which all share a local buffer manager.
-    */
-   if (sarea->texAge != my_bufmgr) {
-      DBG("Lost Textures: sarea->texAge %x my_bufmgr %x\n", sarea->ctxOwner, my_bufmgr);
-      sarea->texAge = my_bufmgr;
-      bm_fake_NotifyContendedLockTake( intel ); 
-   }
-
-   /* Drawable changed?
-    */
-   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
-      intelWindowMoved( intel );
-      intel->lastStamp = dPriv->lastStamp;
-   }
-}
-
-_glthread_DECLARE_STATIC_MUTEX(lockMutex);
-
-/* Lock the hardware and validate our state.  
- */
-void LOCK_HARDWARE( struct intel_context *intel )
-{
-    char __ret=0;
-
-    _glthread_LOCK_MUTEX(lockMutex);
-    assert(!intel->locked);
-
-
-    DRM_CAS(intel->driHwLock, intel->hHWContext,
-	    (DRM_LOCK_HELD|intel->hHWContext), __ret);
-    if (__ret)
-        intelContendedLock( intel, 0 );
-
-   intel->locked = 1;
-
-   if (intel->aub_wrap) {
-      bm_fake_NotifyContendedLockTake( intel ); 
-      intel->vtbl.lost_hardware( intel );
-      intel->vtbl.aub_wrap(intel);
-      intel->aub_wrap = 0;
-   }
-
-   if (bmError(intel)) {
-      bmEvictAll(intel);
-      intel->vtbl.lost_hardware( intel );
-   }
-
-   /* Make sure nothing has been emitted prior to getting the lock: 
-    */
-   assert(intel->batch->map == 0);
-
-   /* XXX: postpone, may not be needed:
-    */
-   if (!intel_batchbuffer_map(intel->batch)) {
-      bmEvictAll(intel);
-      intel->vtbl.lost_hardware( intel );
-
-      /* This could only fail if the batchbuffer was greater in size
-       * than the available texture memory:
-       */
-      if (!intel_batchbuffer_map(intel->batch)) {
-	 _mesa_printf("double failure to map batchbuffer\n");
-	 assert(0);
-      }
-   }
-}
- 
-  
-/* Unlock the hardware using the global current context 
- */
-void UNLOCK_HARDWARE( struct intel_context *intel )
-{
-   /* Make sure everything has been released: 
-    */
-   assert(intel->batch->ptr == intel->batch->map + intel->batch->offset);
-
-   intel_batchbuffer_unmap(intel->batch);
-   intel->vtbl.note_unlock( intel );
-   intel->locked = 0;
-
-
-
-   DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
-   _glthread_UNLOCK_MUTEX(lockMutex); 
-}
-
-
+../intel/intel_context.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_context.h b/src/mesa/drivers/dri/i965/intel_context.h
deleted file mode 100644
index 406f8483dce..00000000000
--- a/src/mesa/drivers/dri/i965/intel_context.h
+++ /dev/null
@@ -1,528 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTELCONTEXT_INC
-#define INTELCONTEXT_INC
-
-
-
-#include "mtypes.h"
-#include "drm.h"
-#include "texmem.h"
-
-#include "intel_screen.h"
-#include "i830_common.h"
-#include "tnl/t_vertex.h"
-
-#define TAG(x) intel##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-#define DV_PF_555  (1<<8)
-#define DV_PF_565  (2<<8)
-#define DV_PF_8888 (3<<8)
-
-struct intel_region;
-struct intel_context;
-
-typedef void (*intel_tri_func)(struct intel_context *, intelVertex *, intelVertex *,
-							  intelVertex *);
-typedef void (*intel_line_func)(struct intel_context *, intelVertex *, intelVertex *);
-typedef void (*intel_point_func)(struct intel_context *, intelVertex *);
-
-#define INTEL_FALLBACK_DRAW_BUFFER	 0x1
-#define INTEL_FALLBACK_READ_BUFFER	 0x2
-#define INTEL_FALLBACK_USER		 0x4
-#define INTEL_FALLBACK_RENDERMODE	 0x8
-#define INTEL_FALLBACK_TEXTURE   	 0x10
-
-extern void intelFallback( struct intel_context *intel, GLuint bit, GLboolean mode );
-#define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
-
-
-
-struct intel_texture_object
-{
-   struct gl_texture_object base; /* The "parent" object */
-
-   /* The mipmap tree must include at least these levels once
-    * validated:
-    */
-   GLuint firstLevel;
-   GLuint lastLevel;
-
-   GLuint dirty_images[6];
-   GLuint dirty;
-
-   /* On validation any active images held in main memory or in other
-    * regions will be copied to this region and the old storage freed.
-    */
-   struct intel_mipmap_tree *mt;
-};
-
-
-
-struct intel_context
-{
-   GLcontext ctx;		/* the parent class */
-
-   struct {
-      void (*destroy)( struct intel_context *intel ); 
-      void (*emit_state)( struct intel_context *intel );
-      void (*emit_invarient_state)( struct intel_context *intel );
-      void (*lost_hardware)( struct intel_context *intel );
-      void (*note_fence)( struct intel_context *intel, GLuint fence );
-      void (*note_unlock)( struct intel_context *intel );
-      void (*update_texture_state)( struct intel_context *intel );
-
-      void (*render_start)( struct intel_context *intel );
-      void (*set_draw_region)( struct intel_context *intel, 
-			       struct intel_region *draw_region,
-			       struct intel_region *depth_region );
-
-      GLuint (*flush_cmd)( void );
-
-      void (*emit_flush)( struct intel_context *intel,
-			  GLuint unused );
-
-      void (*aub_commands)( struct intel_context *intel, 
-			    GLuint offset,
-			    const void *buf,
-			    GLuint sz );
-      void (*aub_dump_bmp)( struct intel_context *intel, GLuint buffer );
-      void (*aub_wrap)( struct intel_context *intel );
-      void (*aub_gtt_data)( struct intel_context *intel, 
-			    GLuint offset,
-			    const void *src,
-			    GLuint size,
-			    GLuint aubtype, 
-			    GLuint aubsubtype);
-
-
-      void (*reduced_primitive_state)( struct intel_context *intel, GLenum rprim );
-
-      GLboolean (*check_vertex_size)( struct intel_context *intel, GLuint expected );
-
-      void (*invalidate_state)( struct intel_context *intel, GLuint new_state );
-
-      /* Metaops: 
-       */
-      void (*install_meta_state)( struct intel_context *intel );
-      void (*leave_meta_state)( struct intel_context *intel );
-
-      void (*meta_draw_region)( struct intel_context *intel,
-				struct intel_region *draw_region,
-				struct intel_region *depth_region );
-
-      void (*meta_color_mask)( struct intel_context *intel,
-			       GLboolean );
-      
-      void (*meta_stencil_replace)( struct intel_context *intel,
-				    GLuint mask,
-				    GLuint clear );
-
-      void (*meta_depth_replace)( struct intel_context *intel );
-
-      void (*meta_texture_blend_replace) (struct intel_context * intel);
-      
-      void (*meta_no_stencil_write)( struct intel_context *intel );
-      void (*meta_no_depth_write)( struct intel_context *intel );
-      void (*meta_no_texture)( struct intel_context *intel );
-      void (*meta_import_pixel_state) (struct intel_context * intel);
-      void (*meta_frame_buffer_texture)( struct intel_context *intel,
-					 GLint xoff, GLint yoff );
-
-      void (*meta_draw_quad)(struct intel_context *intel, 
-			     GLfloat x0, GLfloat x1,
-			     GLfloat y0, GLfloat y1, 
-			     GLfloat z,
-			     GLubyte red, GLubyte green,
-			     GLubyte blue, GLubyte alpha,
-			     GLfloat s0, GLfloat s1,
-			     GLfloat t0, GLfloat t1);
-
-
-
-   } vtbl;
-
-   GLint refcount;   
-   GLuint Fallback;
-   GLuint NewGLState;
-   
-   GLuint last_swap_fence;
-   GLuint second_last_swap_fence;
-   
-   GLboolean aub_wrap;
-   GLuint stats_wm;
-
-   struct intel_batchbuffer *batch;
-
-   GLubyte clear_chan[4];
-   GLuint ClearColor;
-   GLuint ClearDepth;
-
-   GLfloat depth_scale;
-   GLfloat polygon_offset_scale; /* dependent on depth_scale, bpp */
-   GLuint depth_clear_mask;
-   GLuint stencil_clear_mask;
-
-   GLboolean hw_stencil;
-   GLboolean hw_stipple;
-   GLboolean depth_buffer_is_float;
-   GLboolean no_hw;
-   GLboolean no_rast;
-   GLboolean thrashing;
-   GLboolean locked;
-   GLboolean strict_conformance;
-   GLboolean need_flush;
-
-
-   
-   /* AGP memory buffer manager:
-    */
-   struct bufmgr *bm;
-
-
-   /* State for intelvb.c and inteltris.c.
-    */
-   GLenum render_primitive;
-   GLenum reduced_primitive;
-
-   struct intel_region *front_region;
-   struct intel_region *back_region;
-   struct intel_region *draw_region;
-   struct intel_region *depth_region;
-
-   /* These refer to the current draw (front vs. back) buffer:
-    */
-   int drawX;			/* origin of drawable in draw buffer */
-   int drawY;
-   GLuint numClipRects;		/* cliprects for that buffer */
-   drm_clip_rect_t *pClipRects;
-   struct gl_texture_object *frame_buffer_texobj;
-
-   GLboolean scissor;
-   drm_clip_rect_t draw_rect;
-   drm_clip_rect_t scissor_rect;
-
-   drm_context_t hHWContext;
-   drmLock *driHwLock;
-   int driFd;
-
-   __DRIdrawablePrivate *driDrawable;
-   __DRIscreenPrivate *driScreen;
-   intelScreenPrivate *intelScreen; 
-   volatile drmI830Sarea *sarea; 
-   
-   FILE *aub_file;
-
-   GLuint lastStamp;
-
-   /**
-    * Configuration cache
-    */
-   driOptionCache optionCache;
-
-   /* VBI
-    */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
-   int64_t swap_ust;
-   int64_t swap_missed_ust;
-
-   GLuint swap_count;
-   GLuint swap_missed_count;
-};
-
-/* These are functions now:
- */
-void LOCK_HARDWARE( struct intel_context *intel );
-void UNLOCK_HARDWARE( struct intel_context *intel );
-
-
-#define SUBPIXEL_X 0.125
-#define SUBPIXEL_Y 0.125
-
-/* ================================================================
- * Color packing:
- */
-
-#define INTEL_PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define INTEL_PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define INTEL_PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define INTEL_PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
-(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
- (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
-  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
-   0)))
-
-
-
-/* ================================================================
- * From linux kernel i386 header files, copes with odd sizes better
- * than COPY_DWORDS would:
- */
-#if defined(i386) || defined(__i386__)
-static inline void * __memcpy(void * to, const void * from, size_t n)
-{
-   int d0, d1, d2;
-   __asm__ __volatile__(
-      "rep ; movsl\n\t"
-      "testb $2,%b4\n\t"
-      "je 1f\n\t"
-      "movsw\n"
-      "1:\ttestb $1,%b4\n\t"
-      "je 2f\n\t"
-      "movsb\n"
-      "2:"
-      : "=&c" (d0), "=&D" (d1), "=&S" (d2)
-      :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
-      : "memory");
-   return (to);
-}
-#else
-#define __memcpy(a,b,c) memcpy(a,b,c)
-#endif
-
-
-/* The system memcpy (at least on ubuntu 5.10) has problems copying
- * to agp (writecombined) memory from a source which isn't 64-byte
- * aligned - there is a 4x performance falloff.
- *
- * The x86 __memcpy is immune to this but is slightly slower
- * (10%-ish) than the system memcpy.
- *
- * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
- * isn't much faster than x86_memcpy for agp copies.
- * 
- * TODO: switch dynamically.
- */
-static inline void *do_memcpy( void *dest, const void *src, size_t n )
-{
-   if ( (((unsigned long)src) & 63) ||
-	(((unsigned long)dest) & 63)) {
-      return  __memcpy(dest, src, n);	
-   }
-   else
-      return memcpy(dest, src, n);
-}
-
-
-
-
-
-/* ================================================================
- * Debugging:
- */
-extern int INTEL_DEBUG;
-
-#define DEBUG_TEXTURE	0x1
-#define DEBUG_STATE	0x2
-#define DEBUG_IOCTL	0x4
-#define DEBUG_PRIMS	0x8
-#define DEBUG_VERTS	0x10
-#define DEBUG_FALLBACKS	0x20
-#define DEBUG_VERBOSE	0x40
-#define DEBUG_DRI       0x80
-#define DEBUG_DMA       0x100
-#define DEBUG_SANITY    0x200
-#define DEBUG_SYNC      0x400
-#define DEBUG_SLEEP     0x800
-#define DEBUG_PIXEL     0x1000
-#define DEBUG_STATS     0x2000
-#define DEBUG_TILE      0x4000
-#define DEBUG_SINGLE_THREAD   0x8000
-#define DEBUG_WM        0x10000
-#define DEBUG_URB       0x20000
-#define DEBUG_VS        0x40000
-
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-#define PCI_CHIP_I855_GM		0x3582
-#define PCI_CHIP_I865_G			0x2572
-#define PCI_CHIP_I915_G			0x2582
-#define PCI_CHIP_I915_GM		0x2592
-#define PCI_CHIP_I945_G			0x2772
-#define PCI_CHIP_I965_G			0x29A2
-#define PCI_CHIP_I965_Q			0x2992
-#define PCI_CHIP_I965_G_1		0x2982
-#define PCI_CHIP_I946_GZ		0x2972
-#define PCI_CHIP_I965_GM                0x2A02
-#define PCI_CHIP_I965_GME               0x2A12
-
-
-/* ================================================================
- * intel_context.c:
- */
-
-extern GLboolean intelInitContext( struct intel_context *intel, 
-				   const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate,
-				   struct dd_function_table *functions );
-
-extern void intelGetLock(struct intel_context *intel, GLuint flags);
-
-extern void intelFinish( GLcontext *ctx );
-extern void intelFlush( GLcontext *ctx );
-
-extern void intelInitDriverFunctions( struct dd_function_table *functions );
-
-
-/* ================================================================
- * intel_state.c:
- */
-extern void intelInitStateFuncs( struct dd_function_table *functions );
-
-#define COMPAREFUNC_ALWAYS		0
-#define COMPAREFUNC_NEVER		0x1
-#define COMPAREFUNC_LESS		0x2
-#define COMPAREFUNC_EQUAL		0x3
-#define COMPAREFUNC_LEQUAL		0x4
-#define COMPAREFUNC_GREATER		0x5
-#define COMPAREFUNC_NOTEQUAL		0x6
-#define COMPAREFUNC_GEQUAL		0x7
-
-#define STENCILOP_KEEP			0
-#define STENCILOP_ZERO			0x1
-#define STENCILOP_REPLACE		0x2
-#define STENCILOP_INCRSAT		0x3
-#define STENCILOP_DECRSAT		0x4
-#define STENCILOP_INCR			0x5
-#define STENCILOP_DECR			0x6
-#define STENCILOP_INVERT		0x7
-
-#define LOGICOP_CLEAR			0
-#define LOGICOP_NOR			0x1
-#define LOGICOP_AND_INV 		0x2
-#define LOGICOP_COPY_INV		0x3
-#define LOGICOP_AND_RVRSE		0x4
-#define LOGICOP_INV			0x5
-#define LOGICOP_XOR			0x6
-#define LOGICOP_NAND			0x7
-#define LOGICOP_AND			0x8
-#define LOGICOP_EQUIV			0x9
-#define LOGICOP_NOOP			0xa
-#define LOGICOP_OR_INV			0xb
-#define LOGICOP_COPY			0xc
-#define LOGICOP_OR_RVRSE		0xd
-#define LOGICOP_OR			0xe
-#define LOGICOP_SET			0xf
-
-#define BLENDFACT_ZERO			0x01
-#define BLENDFACT_ONE			0x02
-#define BLENDFACT_SRC_COLR		0x03
-#define BLENDFACT_INV_SRC_COLR 		0x04
-#define BLENDFACT_SRC_ALPHA		0x05
-#define BLENDFACT_INV_SRC_ALPHA 	0x06
-#define BLENDFACT_DST_ALPHA		0x07
-#define BLENDFACT_INV_DST_ALPHA 	0x08
-#define BLENDFACT_DST_COLR		0x09
-#define BLENDFACT_INV_DST_COLR		0x0a
-#define BLENDFACT_SRC_ALPHA_SATURATE	0x0b
-#define BLENDFACT_CONST_COLOR		0x0c
-#define BLENDFACT_INV_CONST_COLOR	0x0d
-#define BLENDFACT_CONST_ALPHA		0x0e
-#define BLENDFACT_INV_CONST_ALPHA	0x0f
-#define BLENDFACT_MASK          	0x0f
-
-
-extern int intel_translate_compare_func( GLenum func );
-extern int intel_translate_stencil_op( GLenum op );
-extern int intel_translate_blend_factor( GLenum factor );
-extern int intel_translate_logic_op( GLenum opcode );
-
-
-/* ================================================================
- * intel_buffers.c:
- */
-void intelInitBufferFuncs( struct dd_function_table *functions );
-
-struct intel_region *intel_readbuf_region( struct intel_context *intel );
-struct intel_region *intel_drawbuf_region( struct intel_context *intel );
-
-extern void intelWindowMoved( struct intel_context *intel );
-
-extern GLboolean intel_intersect_cliprects( drm_clip_rect_t *dest,
-					    const drm_clip_rect_t *a,
-					    const drm_clip_rect_t *b );
-
-
-/* ================================================================
- * intel_pixel_copy.c:
- */
-void intelCopyPixels(GLcontext * ctx,
-                     GLint srcx, GLint srcy,
-                     GLsizei width, GLsizei height,
-                     GLint destx, GLint desty, GLenum type);
-
-GLboolean intel_check_blit_fragment_ops(GLcontext * ctx);
-
-void intelBitmap(GLcontext * ctx,
-		 GLint x, GLint y,
-		 GLsizei width, GLsizei height,
-		 const struct gl_pixelstore_attrib *unpack,
-		 const GLubyte * pixels);
-
-void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging);
-#define _NEW_WINDOW_POS 0x40000000
-
-
-/*======================================================================
- * Inline conversion functions.  
- * These are better-typed than the macros used previously:
- */
-static inline struct intel_context *intel_context( GLcontext *ctx )
-{
-   return (struct intel_context *)ctx;
-}
-
-static inline struct intel_texture_object *intel_texture_object( struct gl_texture_object *obj )
-{
-   return (struct intel_texture_object *)obj;
-}
-
-static inline struct intel_texture_image *intel_texture_image( struct gl_texture_image *img )
-{
-   return (struct intel_texture_image *)img;
-}
-
-#endif
-
diff --git a/src/mesa/drivers/dri/i965/intel_decode.c b/src/mesa/drivers/dri/i965/intel_decode.c
new file mode 120000
index 00000000000..f671b6cbb13
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_decode.c
@@ -0,0 +1 @@
+../intel/intel_decode.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_depthstencil.c b/src/mesa/drivers/dri/i965/intel_depthstencil.c
new file mode 120000
index 00000000000..4ac4ae690a3
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_depthstencil.c
@@ -0,0 +1 @@
+../intel/intel_depthstencil.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_fbo.c b/src/mesa/drivers/dri/i965/intel_fbo.c
new file mode 120000
index 00000000000..a19f86dcc57
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_fbo.c
@@ -0,0 +1 @@
+../intel/intel_fbo.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_ioctl.c b/src/mesa/drivers/dri/i965/intel_ioctl.c
deleted file mode 100644
index 0a8e976f706..00000000000
--- a/src/mesa/drivers/dri/i965/intel_ioctl.c
+++ /dev/null
@@ -1,205 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <unistd.h>
-#include <errno.h>
-#include <sched.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "swrast/swrast.h"
-
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "drm.h"
-#include "bufmgr.h"
-
-static int intelWaitIdleLocked( struct intel_context *intel )
-{
-   static int in_wait_idle = 0;
-   unsigned int fence;
-
-   if (!in_wait_idle) {
-      if (INTEL_DEBUG & DEBUG_SYNC) {
-	 fprintf(stderr, "waiting for idle\n");
-      }
-
-      in_wait_idle = 1;
-      fence = bmSetFence(intel);
-      intelWaitIrq(intel, fence);
-      in_wait_idle = 0;
-
-      return bmTestFence(intel, fence);
-   } else {
-      return 1;
-   }
-}
-
-int intelEmitIrqLocked( struct intel_context *intel )
-{
-   int seq = 1;
-
-   if (!intel->no_hw) {
-      drmI830IrqEmit ie;
-      int ret;
-      
-      assert(((*(int *)intel->driHwLock) & ~DRM_LOCK_CONT) == 
-	     (DRM_LOCK_HELD|intel->hHWContext));
-
-      ie.irq_seq = &seq;
-
-      ret = drmCommandWriteRead( intel->driFd, DRM_I830_IRQ_EMIT, 
-				 &ie, sizeof(ie) );
-      if ( ret ) {
-	 fprintf( stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret );
-	 exit(1);
-      }   
-
-      if (0)
-	 fprintf(stderr, "%s -->  %d\n", __FUNCTION__, seq );
-   }
-
-   return seq;
-}
-
-void intelWaitIrq( struct intel_context *intel, int seq )
-{
-   if (!intel->no_hw) {
-      drmI830IrqWait iw;
-      int ret, lastdispatch;
-      
-      if (0)
-	 fprintf(stderr, "%s %d\n", __FUNCTION__, seq );
-
-      iw.irq_seq = seq;
-	
-      do {
-	 lastdispatch = intel->sarea->last_dispatch;
-	 ret = drmCommandWrite( intel->driFd, DRM_I830_IRQ_WAIT, &iw, sizeof(iw) );
-
-	 /* This seems quite often to return before it should!?! 
-	  */
-      } while (ret == -EAGAIN || ret == -EINTR || (ret == -EBUSY && lastdispatch != intel->sarea->last_dispatch) || (ret == 0 && seq > intel->sarea->last_dispatch)
-	       || (ret == 0 && intel->sarea->last_dispatch - seq >= (1 << 24)));
-      
-
-      if ( ret ) {
-	 fprintf( stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret );
-
-	 if (intel->aub_file) {
-	    intel->vtbl.aub_dump_bmp( intel, intel->ctx.Visual.doubleBufferMode ? 1 : 0 );
-	 }
-
-	 exit(1);
-      }
-   }
-}
-
-
-void intel_batch_ioctl( struct intel_context *intel, 
-			GLuint start_offset,
-			GLuint used)
-{
-   drmI830BatchBuffer batch;
-
-   assert(intel->locked);
-   assert(used);
-
-   if (0)
-      fprintf(stderr, "%s used %d offset %x..%x\n",
-	      __FUNCTION__, 
-	      used, 
-	      start_offset,
-	      start_offset + used);
-
-   batch.start = start_offset;
-   batch.used = used;
-   batch.cliprects = NULL;
-   batch.num_cliprects = 0;
-   batch.DR1 = 0;
-   batch.DR4 = 0;
-      
-   if (INTEL_DEBUG & DEBUG_DMA)
-      fprintf(stderr, "%s: 0x%x..0x%x\n",
-	      __FUNCTION__, 
-	      batch.start, 
-	      batch.start + batch.used * 4);
-
-   if (!intel->no_hw) {
-      if (drmCommandWrite (intel->driFd, DRM_I830_BATCHBUFFER, &batch, 
-			   sizeof(batch))) {
-	 fprintf(stderr, "DRM_I830_BATCHBUFFER: %d\n",  -errno);
-	 UNLOCK_HARDWARE(intel);
-	 exit(1);
-      }
-
-      if (INTEL_DEBUG & DEBUG_SYNC) {
-	intelWaitIdleLocked(intel);
-      }
-   }
-}
-
-void intel_cmd_ioctl( struct intel_context *intel, 
-		      char *buf,
-		      GLuint used)
-{
-   drmI830CmdBuffer cmd;
-
-   assert(intel->locked);
-   assert(used);
-
-   cmd.buf = buf;
-   cmd.sz = used;
-   cmd.cliprects = intel->pClipRects;
-   cmd.num_cliprects = 0;
-   cmd.DR1 = 0;
-   cmd.DR4 = 0;
-      
-   if (INTEL_DEBUG & DEBUG_DMA)
-      fprintf(stderr, "%s: 0x%x..0x%x\n",
-	      __FUNCTION__, 
-	      0, 
-	      0 + cmd.sz);
-
-   if (!intel->no_hw) {
-      if (drmCommandWrite (intel->driFd, DRM_I830_CMDBUFFER, &cmd, 
-			   sizeof(cmd))) {
-	 fprintf(stderr, "DRM_I830_CMDBUFFER: %d\n",  -errno);
-	 UNLOCK_HARDWARE(intel);
-	 exit(1);
-      }
-
-      if (INTEL_DEBUG & DEBUG_SYNC) {
-	intelWaitIdleLocked(intel);
-      }
-   }
-}
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
index 8486086b274..242fed0b6ae 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
@@ -1,247 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "intel_context.h"
-#include "intel_mipmap_tree.h"
-#include "intel_regions.h"
-#include "bufmgr.h"
-#include "enums.h"
-#include "imports.h"
-
-static GLenum target_to_target( GLenum target )
-{
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
-      return GL_TEXTURE_CUBE_MAP_ARB;
-   default:
-      return target;
-   }
-}
-
-struct intel_mipmap_tree *intel_miptree_create( struct intel_context *intel,
-						GLenum target,
-						GLenum internal_format,
-						GLuint first_level,
-						GLuint last_level,
-						GLuint width0,
-						GLuint height0,
-						GLuint depth0,
-						GLuint cpp,
-						GLboolean compressed)
-{
-   GLboolean ok;
-   struct intel_mipmap_tree *mt = calloc(sizeof(*mt), 1);
-
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-      _mesa_printf("%s target %s format %s level %d..%d\n", __FUNCTION__,
-		   _mesa_lookup_enum_by_nr(target),
-		   _mesa_lookup_enum_by_nr(internal_format),
-		   first_level,
-		   last_level);
-
-   mt->target = target_to_target(target);
-   mt->internal_format = internal_format;
-   mt->first_level = first_level;
-   mt->last_level = last_level;
-   mt->width0 = width0;
-   mt->height0 = height0;
-   mt->depth0 = depth0;
-   mt->cpp = compressed ? 2 : cpp;
-   mt->compressed = compressed;
-
-   switch (intel->intelScreen->deviceID) {
-#if 0
-   case PCI_CHIP_I945_G:
-      ok = i945_miptree_layout( mt );
-      break;
-   case PCI_CHIP_I915_G:
-   case PCI_CHIP_I915_GM:
-      ok = i915_miptree_layout( mt );
-      break;
-#endif
-   default:
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 _mesa_printf("assuming BRW texture layouts\n");
-      ok = brw_miptree_layout( mt );
-      break;
-   }
-
-   if (ok)
-      mt->region = intel_region_alloc( intel, 
-				       mt->cpp,
-				       mt->pitch, 
-				       mt->total_height );
-
-   if (!mt->region) {
-      free(mt);
-      return NULL;
-   }
-
-   return mt;
-}
-
-
-
-void intel_miptree_destroy( struct intel_context *intel,
-			    struct intel_mipmap_tree *mt )
-{
-   if (mt) {
-      GLuint i;
-
-      intel_region_release(intel, &(mt->region));
-
-      for (i = 0; i < MAX_TEXTURE_LEVELS; i++)
-	 if (mt->level[i].image_offset)
-	    free(mt->level[i].image_offset);
-
-      free(mt);
-   }
-}
-
-
-
-
-void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
-				  GLuint level,
-				  GLuint nr_images,
-				  GLuint x, GLuint y,
-				  GLuint w, GLuint h, GLuint d)
-{
-   mt->level[level].width = w;
-   mt->level[level].height = h;
-   mt->level[level].depth = d;
-   mt->level[level].level_offset = (x + y * mt->pitch) * mt->cpp;
-   mt->level[level].nr_images = nr_images;
-
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-      _mesa_printf("%s level %d img size: %d,%d level_offset 0x%x\n", __FUNCTION__, level, w, h, 
-		   mt->level[level].level_offset);
-
-   /* Not sure when this would happen, but anyway: 
-    */
-   if (mt->level[level].image_offset) {
-      free(mt->level[level].image_offset);
-      mt->level[level].image_offset = NULL;
-   }
-
-   if (nr_images > 1) {
-      mt->level[level].image_offset = malloc(nr_images * sizeof(GLuint));
-      mt->level[level].image_offset[0] = 0;
-   }
-}
-
-
-
-void intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
-				    GLuint level,
-				    GLuint img,
-				    GLuint x, GLuint y)
-{
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-      _mesa_printf("%s level %d img %d pos %d,%d\n", __FUNCTION__, level, img, x, y);
-
-   if (img == 0)
-      assert(x == 0 && y == 0);
-
-   if (img > 0)
-      mt->level[level].image_offset[img] = (x + y * mt->pitch) * mt->cpp;
-}
-
-
-/* Although we use the image_offset[] array to store relative offsets
- * to cube faces, Mesa doesn't know anything about this and expects
- * each cube face to be treated as a separate image.
- *
- * These functions present that view to mesa:
- */
-const GLuint *intel_miptree_depth_offsets(struct intel_mipmap_tree *mt,
-					  GLuint level)
-{
-   static const GLuint zero = 0;
-
-   if (mt->target != GL_TEXTURE_3D ||
-       mt->level[level].nr_images == 1)
-      return &zero;
-   else
-      return mt->level[level].image_offset;
-}
-
-
-GLuint intel_miptree_image_offset(struct intel_mipmap_tree *mt,
-				  GLuint face,
-				  GLuint level)
-{
-   if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
-      return (mt->level[level].level_offset +
-	      mt->level[level].image_offset[face]);
-   else
-      return mt->level[level].level_offset;
-}
-
-
-
-
-
-
-/* Upload data for a particular image.
- */
-GLboolean intel_miptree_image_data(struct intel_context *intel, 
-				   struct intel_mipmap_tree *dst,
-				   GLuint face,
-				   GLuint level,
-				   const void *src, 
-				   GLuint src_row_pitch,
-				   GLuint src_image_pitch)
-{
-   GLuint depth = dst->level[level].depth;
-   GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
-   const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
-   GLuint i;
-
-   DBG("%s\n", __FUNCTION__);
-   for (i = 0; i < depth; i++) {
-      if (!intel_region_data(intel,
-			     dst->region, 
-			     dst_offset + dst_depth_offset[i],
-			     0,
-			     0,
-			     src,
-			     src_row_pitch,
-			     0, 0,	/* source x,y */
-			     dst->level[level].width,
-			     dst->level[level].height))
-	 return GL_FALSE;
-      src += src_image_pitch;
-   }
-   return GL_TRUE;
-}
-
+../intel/intel_mipmap_tree.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.h b/src/mesa/drivers/dri/i965/intel_mipmap_tree.h
deleted file mode 100644
index dbd7167b778..00000000000
--- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.h
+++ /dev/null
@@ -1,166 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_MIPMAP_TREE_H
-#define INTEL_MIPMAP_TREE_H
-
-#include "intel_regions.h"
-
-/* A layer on top of the intel_regions code which adds:
- *
- * - Code to size and layout a region to hold a set of mipmaps.
- * - Query to determine if a new image fits in an existing tree.
- *
- * The fixed mipmap layout of intel hardware where one offset
- * specifies the position of all images in a mipmap hierachy
- * complicates the implementation of GL texture image commands,
- * compared to hardware where each image is specified with an
- * independent offset.
- *
- * In an ideal world, each texture object would be associated with a
- * single bufmgr buffer or 2d intel_region, and all the images within
- * the texture object would slot into the tree as they arrive.  The
- * reality can be a little messier, as images can arrive from the user
- * with sizes that don't fit in the existing tree, or in an order
- * where the tree layout cannot be guessed immediately.  
- * 
- * This structure encodes an idealized mipmap tree.  The GL image
- * commands build these where possible, otherwise store the images in
- * temporary system buffers.
- */
-
-
-struct intel_mipmap_level {
-   GLuint level_offset;
-   GLuint width;
-   GLuint height;
-   GLuint depth;
-   GLuint nr_images;
-
-   /* Explicitly store the offset of each image for each cube face or
-    * depth value.  Pretty much have to accept that hardware formats
-    * are going to be so diverse that there is no unified way to
-    * compute the offsets of depth/cube images within a mipmap level,
-    * so have to store them as a lookup table:
-    */
-   GLuint *image_offset;
-};
-
-struct intel_mipmap_tree {
-   /* Effectively the key:
-    */
-   GLenum target;
-   GLenum internal_format;
-
-   GLuint first_level;
-   GLuint last_level;
-
-   GLuint width0, height0, depth0;
-   GLuint cpp;
-   GLboolean compressed;
-
-   /* Derived from the above:
-    */   
-   GLuint pitch;
-   GLuint depth_pitch;		/* per-image on i945? */
-   GLuint total_height;
-   
-   /* Includes image offset tables:
-    */
-   struct intel_mipmap_level level[MAX_TEXTURE_LEVELS];
-
-   /* The data is held here:
-    */
-   struct intel_region *region;
-
-   /* These are also refcounted:
-    */
-   GLuint refcount;
-};
-
-
-
-struct intel_mipmap_tree *intel_miptree_create( struct intel_context *intel,
-						GLenum target,
-						GLenum internal_format,
-						GLuint first_level,
-						GLuint last_level,
-						GLuint width0,
-						GLuint height0,
-						GLuint depth0,
-						GLuint cpp,
-						GLboolean compressed);
-
-void intel_miptree_destroy( struct intel_context *intel,
-			    struct intel_mipmap_tree *mt );
-
-
-/* Return the linear offset of an image relative to the start of the
- * tree:
- */
-GLuint intel_miptree_image_offset( struct intel_mipmap_tree *mt,
-				   GLuint face,
-				   GLuint level );
-
-/* Return pointers to each 2d slice within an image.  Indexed by depth
- * value.
- */
-const GLuint *intel_miptree_depth_offsets(struct intel_mipmap_tree *mt,
-					  GLuint level);
-
-
-void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
-				  GLuint level,
-				  GLuint nr_images,
-				  GLuint x, GLuint y,
-				  GLuint w, GLuint h, GLuint d);
-
-void intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
-				    GLuint level,
-				    GLuint img,
-				    GLuint x, GLuint y);
-
-
-/* Upload an image into a tree
- */
-GLboolean intel_miptree_image_data(struct intel_context *intel, 
-				   struct intel_mipmap_tree *dst,
-				   GLuint face,
-				   GLuint level,
-				   const void *src, 
-				   GLuint src_row_pitch,
-				   GLuint src_image_pitch);
-
-/* i915_mipmap_tree.c:
- */
-GLboolean i915_miptree_layout( struct intel_mipmap_tree *mt );
-GLboolean i945_miptree_layout( struct intel_mipmap_tree *mt );
-GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt );
-
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/intel_pixel.c b/src/mesa/drivers/dri/i965/intel_pixel.c
new file mode 120000
index 00000000000..d733c5e8745
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_pixel.c
@@ -0,0 +1 @@
+../intel/intel_pixel.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
index 421fcc5e511..9085c7b0397 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
@@ -1,353 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portionsalloc
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "enums.h"
-#include "image.h"
-#include "colormac.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "bufferobj.h"
-#include "swrast/swrast.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "intel_buffer_objects.h"
-
-
-
-#define FILE_DEBUG_FLAG DEBUG_PIXEL
-
-
-/* Unlike the other intel_pixel_* functions, the expectation here is
- * that the incoming data is not in a PBO.  With the XY_TEXT blit
- * method, there's no benefit haveing it in a PBO, but we could
- * implement a path based on XY_MONO_SRC_COPY_BLIT which might benefit
- * PBO bitmaps.  I think they are probably pretty rare though - I
- * wonder if Xgl uses them?
- */
-static const GLubyte *map_pbo( GLcontext *ctx,
-			       GLsizei width, GLsizei height,
-			       const struct gl_pixelstore_attrib *unpack,
-			       const GLubyte *bitmap )
-{
-   GLubyte *buf;
-
-   if (!_mesa_validate_pbo_access(2, unpack, width, height, 1,
-				  GL_COLOR_INDEX, GL_BITMAP,
-				  (GLvoid *) bitmap)) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,"glBitmap(invalid PBO access)");
-      return NULL;
-   }
-
-   buf = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
-					   GL_READ_ONLY_ARB,
-					   unpack->BufferObj);
-   if (!buf) {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "glBitmap(PBO is mapped)");
-      return NULL;
-   }
-
-   return ADD_POINTERS(buf, bitmap);
-}
-
-static GLboolean test_bit( const GLubyte *src,
-			    GLuint bit )
-{
-   return (src[bit/8] & (1<<(bit % 8))) ? 1 : 0;
-}
-
-static void set_bit( GLubyte *dest,
-			  GLuint bit )
-{
-   dest[bit/8] |= 1 << (bit % 8);
-}
-
-static int align(int x, int align)
-{
-   return (x + align - 1) & ~(align - 1);
-}
-
-/* Extract a rectangle's worth of data from the bitmap.  Called
- * per-cliprect.
- */
-static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
-			      const struct gl_pixelstore_attrib *unpack,
-			      const GLubyte *bitmap,
-			      GLuint x, GLuint y, 
-			      GLuint w, GLuint h,
-			      GLubyte *dest,
-			      GLuint row_align,
-			      GLboolean invert)
-{
-   GLuint src_offset = (x + unpack->SkipPixels) & 0x7;
-   GLuint mask = unpack->LsbFirst ? 0 : 7;
-   GLuint bit = 0;
-   GLint row, col;
-   GLint first, last;
-   GLint incr;
-   GLuint count = 0;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("%s %d,%d %dx%d bitmap %dx%d skip %d src_offset %d mask %d\n",
-		   __FUNCTION__, x,y,w,h,width,height,unpack->SkipPixels, src_offset, mask);
-
-   if (invert) {
-      first = h-1;
-      last = 0;
-      incr = -1;
-   }
-   else {
-      first = 0;
-      last = h-1;
-      incr = 1;
-   }
-
-   /* Require that dest be pre-zero'd.
-    */
-   for (row = first; row != (last+incr); row += incr) {
-      const GLubyte *rowsrc = _mesa_image_address2d(unpack, bitmap, 
-						    width, height, 
-						    GL_COLOR_INDEX, GL_BITMAP, 
-						    y + row, x);
-
-      for (col = 0; col < w; col++, bit++) {
-	 if (test_bit(rowsrc, (col + src_offset) ^ mask)) {
-	    set_bit(dest, bit ^ 7);
-	    count++;
-	 }
-      }
-
-      if (row_align)
-	 bit = (bit + row_align - 1) & ~(row_align - 1);
-   }
-
-   return count;
-}
-
-
-
-
-/*
- * Render a bitmap.
- */
-static GLboolean
-do_blit_bitmap( GLcontext *ctx, 
-		GLint dstx, GLint dsty,
-		GLsizei width, GLsizei height,
-		const struct gl_pixelstore_attrib *unpack,
-		const GLubyte *bitmap )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_region *dst = intel_drawbuf_region(intel);
-   
-   union {
-      GLuint ui;
-      GLubyte ub[4];
-   } color;
-
-
-   if (unpack->BufferObj->Name) {
-      bitmap = map_pbo(ctx, width, height, unpack, bitmap);
-      if (bitmap == NULL)
-	 return GL_TRUE;	/* even though this is an error, we're done */
-   }
-
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[0], ctx->Current.RasterColor[2]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[1], ctx->Current.RasterColor[1]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[2], ctx->Current.RasterColor[0]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[3], ctx->Current.RasterColor[3]);
-
-   /* Does zoom apply to bitmaps?
-    */
-   if (!intel_check_blit_fragment_ops(ctx) ||
-       ctx->Pixel.ZoomX != 1.0F || 
-       ctx->Pixel.ZoomY != 1.0F)
-      return GL_FALSE;
-
-   LOCK_HARDWARE(intel);
-
-   if (intel->driDrawable->numClipRects) {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      drm_clip_rect_t dest_rect;
-      GLint nbox = dPriv->numClipRects;
-      GLint srcx = 0, srcy = 0;
-      GLint orig_screen_x1, orig_screen_y2;
-      GLuint i;
-
-
-      orig_screen_x1 = dPriv->x + dstx;
-      orig_screen_y2 = dPriv->y + (dPriv->h - dsty);
-
-      /* Do scissoring in GL coordinates:
-       */
-      if (ctx->Scissor.Enabled)
-      {
-	 GLint x = ctx->Scissor.X;
-	 GLint y = ctx->Scissor.Y;
-	 GLuint w = ctx->Scissor.Width;
-	 GLuint h = ctx->Scissor.Height;
-
-         if (!_mesa_clip_to_region(x, y, x+w-1, y+h-1, &dstx, &dsty, &width, &height))
-            goto out;
-      }
-
-      /* Convert from GL to hardware coordinates:
-       */
-      dsty = dPriv->y + (dPriv->h - dsty - height);  
-      dstx = dPriv->x + dstx;
-
-      dest_rect.x1 = dstx;
-      dest_rect.y1 = dsty;
-      dest_rect.x2 = dstx + width;
-      dest_rect.y2 = dsty + height;
-
-      for (i = 0; i < nbox; i++) {
-         drm_clip_rect_t rect;
-	 int box_w, box_h;
-	 GLint px, py;
-	 GLuint stipple[32];  
-
-         if (!intel_intersect_cliprects(&rect, &dest_rect, &box[i]))
-            continue;
-
-	 /* Now go back to GL coordinates to figure out what subset of
-	  * the bitmap we are uploading for this cliprect:
-	  */
-	 box_w = rect.x2 - rect.x1;
-	 box_h = rect.y2 - rect.y1;
-	 srcx = rect.x1 - orig_screen_x1;
-	 srcy = orig_screen_y2 - rect.y2;
-
-
-#define DY 32
-#define DX 32
-
-	 /* Then, finally, chop it all into chunks that can be
-	  * digested by hardware:
-	  */
-	 for (py = 0; py < box_h; py += DY) { 
-	    for (px = 0; px < box_w; px += DX) { 
-	       int h = MIN2(DY, box_h - py);
-	       int w = MIN2(DX, box_w - px); 
-	       GLuint sz = align(align(w,8) * h, 64)/8;
-	       GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
-		  ctx->Color.LogicOp : GL_COPY;
-
-	       assert(sz <= sizeof(stipple));
-	       memset(stipple, 0, sz);
-
-	       /* May need to adjust this when padding has been introduced in
-		* sz above:
-		*/
-	       if (get_bitmap_rect(width, height, unpack, 
-				   bitmap,
-				   srcx + px, srcy + py, w, h,
-				   (GLubyte *)stipple,
-				   8,
-				   GL_TRUE) == 0)
-		  continue;
-
-	       /* 
-		*/
-	       intelEmitImmediateColorExpandBlit( intel,
-						  dst->cpp,
-						  (GLubyte *)stipple, 
-						  sz,
-						  color.ui,
-						  dst->pitch,
-						  dst->buffer,
-						  0,
-						  dst->tiled,
-						  rect.x1 + px,
-						  rect.y2 - (py + h),
-						  w, h,
-						  logic_op);
-	    } 
-	 } 
-      }
-      intel->need_flush = GL_TRUE;
-   out:
-      intel_batchbuffer_flush(intel->batch);
-   }
-   UNLOCK_HARDWARE(intel);
-
-
-   if (unpack->BufferObj->Name) {
-      /* done with PBO so unmap it now */
-      ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
-                              unpack->BufferObj);
-   }
-
-   return GL_TRUE;
-}
-
-
-
-
-
-/* There are a large number of possible ways to implement bitmap on
- * this hardware, most of them have some sort of drawback.  Here are a
- * few that spring to mind:
- * 
- * Blit:
- *    - XY_MONO_SRC_BLT_CMD
- *         - use XY_SETUP_CLIP_BLT for cliprect clipping.
- *    - XY_TEXT_BLT
- *    - XY_TEXT_IMMEDIATE_BLT
- *         - blit per cliprect, subject to maximum immediate data size.
- *    - XY_COLOR_BLT 
- *         - per pixel or run of pixels
- *    - XY_PIXEL_BLT
- *         - good for sparse bitmaps
- *
- * 3D engine:
- *    - Point per pixel
- *    - Translate bitmap to an alpha texture and render as a quad
- *    - Chop bitmap up into 32x32 squares and render w/polygon stipple.
- */
-void
-intelBitmap(GLcontext * ctx,
-	    GLint x, GLint y,
-	    GLsizei width, GLsizei height,
-	    const struct gl_pixelstore_attrib *unpack,
-	    const GLubyte * pixels)
-{
-   if (do_blit_bitmap(ctx, x, y, width, height,
-                          unpack, pixels))
-      return;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
-
-   _swrast_Bitmap(ctx, x, y, width, height, unpack, pixels);
-}
+../intel/intel_pixel_bitmap.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_copy.c b/src/mesa/drivers/dri/i965/intel_pixel_copy.c
index 58dc49505fe..ee433605904 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_copy.c
@@ -1,343 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "enums.h"
-#include "image.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "state.h"
-#include "swrast/swrast.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-
-
-static struct intel_region *
-copypix_src_region(struct intel_context *intel, GLenum type)
-{
-   switch (type) {
-   case GL_COLOR:
-      return intel_readbuf_region(intel);
-   case GL_DEPTH:
-      /* Don't think this is really possible execpt at 16bpp, when we have no stencil.
-       */
-      if (intel->depth_region && intel->depth_region->cpp == 2)
-         return intel->depth_region;
-   case GL_STENCIL:
-      /* Don't think this is really possible. 
-       */
-      break;
-   case GL_DEPTH_STENCIL_EXT:
-      /* Does it matter whether it is stencil/depth or depth/stencil?
-       */
-      return intel->depth_region;
-   default:
-      break;
-   }
-
-   return NULL;
-}
-
-
-
-
-/**
- * Check if any fragment operations are in effect which might effect
- * glDraw/CopyPixels.
- */
-GLboolean
-intel_check_blit_fragment_ops(GLcontext * ctx)
-{
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   return !(ctx->_ImageTransferState ||
-	    ctx->RenderMode != GL_RENDER ||
-            ctx->Color.AlphaEnabled ||
-            ctx->Depth.Test ||
-            ctx->Fog.Enabled ||
-            ctx->Stencil.Enabled ||
-            !ctx->Color.ColorMask[0] ||
-            !ctx->Color.ColorMask[1] ||
-            !ctx->Color.ColorMask[2] ||
-            !ctx->Color.ColorMask[3] ||	/* can do this! */
-            ctx->Texture._EnabledUnits ||
-	    ctx->FragmentProgram._Enabled ||
-	    ctx->Color.BlendEnabled);
-}
-
-/* Doesn't work for overlapping regions.  Could do a double copy or
- * just fallback.
- */
-static GLboolean
-do_texture_copypixels(GLcontext * ctx,
-                      GLint srcx, GLint srcy,
-                      GLsizei width, GLsizei height,
-                      GLint dstx, GLint dsty, GLenum type)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_region *dst = intel_drawbuf_region(intel);
-   struct intel_region *src = copypix_src_region(intel, type);
-   GLenum src_format;
-   GLenum src_type;
-
-   DBG("%s %d,%d %dx%d --> %d,%d\n", __FUNCTION__, 
-       srcx, srcy, width, height, dstx, dsty);
-
-   if (!src || !dst || type != GL_COLOR ||
-       ctx->_ImageTransferState ||
-       ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F ||
-       ctx->RenderMode != GL_RENDER ||
-       ctx->Texture._EnabledUnits ||
-       ctx->FragmentProgram._Enabled ||
-       src != dst )
-       return GL_FALSE;
-   
-   /* Can't handle overlapping regions.  Don't have sufficient control
-    * over rasterization to pull it off in-place.  Punt on these for
-    * now.
-    * 
-    * XXX: do a copy to a temporary. 
-    */
-   if (src->buffer == dst->buffer) {
-      drm_clip_rect_t srcbox;
-      drm_clip_rect_t dstbox;
-      drm_clip_rect_t tmp;
-
-      srcbox.x1 = srcx;
-      srcbox.y1 = srcy;
-      srcbox.x2 = srcx + width - 1;
-      srcbox.y2 = srcy + height - 1;
-
-      dstbox.x1 = dstx;
-      dstbox.y1 = dsty;
-      dstbox.x2 = dstx + width - 1;
-      dstbox.y2 = dsty + height - 1;
-
-      DBG("src %d,%d %d,%d\n", srcbox.x1, srcbox.y1, srcbox.x2, srcbox.y2);
-      DBG("dst %d,%d %d,%d (%dx%d) (%f,%f)\n", dstbox.x1, dstbox.y1, dstbox.x2, dstbox.y2,
-	  width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY);
-
-      if (intel_intersect_cliprects(&tmp, &srcbox, &dstbox)) {
-         DBG("%s: regions overlap\n", __FUNCTION__);
-         return GL_FALSE;
-      }
-   }
-
-   intelFlush(&intel->ctx);
-
-   intel->vtbl.install_meta_state(intel);
-
-   /* Is this true?  Also will need to turn depth testing on according
-    * to state:
-    */
-   intel->vtbl.meta_no_stencil_write(intel);
-   intel->vtbl.meta_no_depth_write(intel);
-
-   /* Set the 3d engine to draw into the destination region:
-    */
-   intel->vtbl.meta_draw_region(intel, dst, intel->depth_region);
-
-   intel->vtbl.meta_import_pixel_state(intel);
-
-   if (src->cpp == 2) {
-      src_format = GL_RGB;
-      src_type = GL_UNSIGNED_SHORT_5_6_5;
-   }
-   else {
-      src_format = GL_BGRA;
-      src_type = GL_UNSIGNED_BYTE;
-   }
-
-   /* Set the frontbuffer up as a large rectangular texture.
-    */
-   intel->vtbl.meta_frame_buffer_texture( intel, srcx - dstx, srcy - dsty );
-
-   intel->vtbl.meta_texture_blend_replace(intel);
-   
-   if (intel->driDrawable->numClipRects)
-      intel->vtbl.meta_draw_quad( intel,
-				  dstx, dstx + width,
-				  dsty, dsty + height,
-				  ctx->Current.RasterPos[ 2 ],
-				  0, 0, 0, 0, 0.0, 0.0, 0.0, 0.0 );
-   
-   intel->vtbl.leave_meta_state( intel );
-   
-   DBG("%s: success\n", __FUNCTION__);
-   return GL_TRUE;
-}
-
-/**
- * CopyPixels with the blitter.  Don't support zooming, pixel transfer, etc.
- */
-static GLboolean
-do_blit_copypixels(GLcontext * ctx,
-                   GLint srcx, GLint srcy,
-                   GLsizei width, GLsizei height,
-                   GLint dstx, GLint dsty, GLenum type)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_region *dst = intel_drawbuf_region(intel);
-   struct intel_region *src = copypix_src_region(intel, type);
-
-   /* Copypixels can be more than a straight copy.  Ensure all the
-    * extra operations are disabled:
-    */
-   if (!intel_check_blit_fragment_ops(ctx) ||
-       ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F)
-      return GL_FALSE;
-
-   if (!src || !dst)
-      return GL_FALSE;
-
-
-
-   intelFlush(&intel->ctx);
-
-/*    intel->vtbl.render_start(intel); */
-/*    intel->vtbl.emit_state(intel); */
-
-   LOCK_HARDWARE(intel);
-
-   if (intel->driDrawable->numClipRects) {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      drm_clip_rect_t dest_rect;
-      GLint nbox = dPriv->numClipRects;
-      GLint delta_x = 0;
-      GLint delta_y = 0;
-      GLuint i;
-
-      /* Do scissoring in GL coordinates:
-       */
-      if (ctx->Scissor.Enabled)
-      {
-	 GLint x = ctx->Scissor.X;
-	 GLint y = ctx->Scissor.Y;
-	 GLuint w = ctx->Scissor.Width;
-	 GLuint h = ctx->Scissor.Height;
-	 GLint dx = dstx - srcx;
-         GLint dy = dsty - srcy;
-
-         if (!_mesa_clip_to_region(x, y, x+w-1, y+h-1, &dstx, &dsty, &width, &height))
-            goto out;
-	 
-         srcx = dstx - dx;
-         srcy = dsty - dy;
-      }
-
-      /* Convert from GL to hardware coordinates:
-       */
-      dsty = dPriv->h - dsty - height;  
-      srcy = dPriv->h - srcy - height;  
-      dstx += dPriv->x;
-      dsty += dPriv->y;
-      srcx += dPriv->x;
-      srcy += dPriv->y;
-
-      /* Clip against the source region.  This is the only source
-       * clipping we do.  Dst is clipped with cliprects below.
-       */
-      {
-         delta_x = srcx - dstx;
-         delta_y = srcy - dsty;
-
-         if (!_mesa_clip_to_region(0, 0, src->pitch, src->height,
-                                   &srcx, &srcy, &width, &height))
-            goto out;
-
-         dstx = srcx - delta_x;
-         dsty = srcy - delta_y;
-      }
-
-      dest_rect.x1 = dstx;
-      dest_rect.y1 = dsty;
-      dest_rect.x2 = dstx + width;
-      dest_rect.y2 = dsty + height;
-
-/*       intel->vtbl.emit_flush(intel, 0); */
-
-      /* Could do slightly more clipping: Eg, take the intersection of
-       * the existing set of cliprects and those cliprects translated
-       * by delta_x, delta_y:
-       * 
-       * This code will not overwrite other windows, but will
-       * introduce garbage when copying from obscured window regions.
-       */
-      for (i = 0; i < nbox; i++) {
-         drm_clip_rect_t rect;
-
-         if (!intel_intersect_cliprects(&rect, &dest_rect, &box[i]))
-            continue;
-
-
-         intelEmitCopyBlit(intel, 
-			   dst->cpp, 
-			   src->pitch, src->buffer, 0, src->tiled,
-			   dst->pitch, dst->buffer, 0, dst->tiled,
-			   rect.x1 + delta_x, 
-			   rect.y1 + delta_y,       /* srcx, srcy */
-                           rect.x1, rect.y1,    /* dstx, dsty */
-                           rect.x2 - rect.x1, rect.y2 - rect.y1,
-			   ctx->Color.ColorLogicOpEnabled ?
-			   ctx->Color.LogicOp : GL_COPY);
-      }
-
-      intel->need_flush = GL_TRUE;
-   out:
-      intel_batchbuffer_flush(intel->batch);
-   }
-   UNLOCK_HARDWARE(intel);
-   return GL_TRUE;
-}
-
-void
-intelCopyPixels(GLcontext * ctx,
-                GLint srcx, GLint srcy,
-                GLsizei width, GLsizei height,
-                GLint destx, GLint desty, GLenum type)
-{
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (do_blit_copypixels(ctx, srcx, srcy, width, height, destx, desty, type))
-      return;
-
-   if (do_texture_copypixels(ctx, srcx, srcy, width, height, destx, desty, type))
-      return;
-   
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("fallback to _swrast_CopyPixels\n");
-
-   _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type);
-}
+../intel/intel_pixel_copy.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_draw.c b/src/mesa/drivers/dri/i965/intel_pixel_draw.c
new file mode 120000
index 00000000000..8431a24edfc
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_pixel_draw.c
@@ -0,0 +1 @@
+../intel/intel_pixel_draw.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_reg.h b/src/mesa/drivers/dri/i965/intel_reg.h
deleted file mode 100644
index 3c448b3559a..00000000000
--- a/src/mesa/drivers/dri/i965/intel_reg.h
+++ /dev/null
@@ -1,91 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef _INTEL_REG_H_
-#define _INTEL_REG_H_
-
-
-
-#define CMD_3D (0x3<<29)
-
-
-#define _3DPRIMITIVE         ((0x3<<29)|(0x1f<<24))
-#define PRIM_INDIRECT            (1<<23)
-#define PRIM_INLINE              (0<<23)
-#define PRIM_INDIRECT_SEQUENTIAL (0<<17)
-#define PRIM_INDIRECT_ELTS       (1<<17)
-
-#define PRIM3D_TRILIST		(0x0<<18)
-#define PRIM3D_TRISTRIP 	(0x1<<18)
-#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
-#define PRIM3D_TRIFAN		(0x3<<18)
-#define PRIM3D_POLY		(0x4<<18)
-#define PRIM3D_LINELIST 	(0x5<<18)
-#define PRIM3D_LINESTRIP	(0x6<<18)
-#define PRIM3D_RECTLIST 	(0x7<<18)
-#define PRIM3D_POINTLIST	(0x8<<18)
-#define PRIM3D_DIB		(0x9<<18)
-#define PRIM3D_MASK		(0x1f<<18)
-
-#define I915PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define I915PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define I915PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define I915PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-
-
-#define BR00_BITBLT_CLIENT   0x40000000
-#define BR00_OP_COLOR_BLT    0x10000000
-#define BR00_OP_SRC_COPY_BLT 0x10C00000
-#define BR13_SOLID_PATTERN   0x80000000
-
-#define XY_COLOR_BLT_CMD		((2<<29)|(0x50<<22)|0x4)
-#define XY_COLOR_BLT_WRITE_ALPHA	(1<<21)
-#define XY_COLOR_BLT_WRITE_RGB		(1<<20)
-
-#define XY_SRC_COPY_BLT_CMD             ((2<<29)|(0x53<<22)|6)
-#define XY_SRC_COPY_BLT_WRITE_ALPHA     (1<<21)
-#define XY_SRC_COPY_BLT_WRITE_RGB       (1<<20)
-
-#define XY_SRC_TILED  (1<<15)
-#define XY_DST_TILED  (1<<11)
-
-#define FENCE_LINEAR 0
-#define FENCE_XMAJOR 1
-#define FENCE_YMAJOR 2
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/intel_regions.c b/src/mesa/drivers/dri/i965/intel_regions.c
index 835ecdd7257..89b2f15c10f 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_regions.c
+++ b/src/mesa/drivers/dri/i965/intel_regions.c
@@ -1,295 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/* Provide additional functionality on top of bufmgr buffers:
- *   - 2d semantics and blit operations
- *   - refcounting of buffers for multiple images in a buffer.
- *   - refcounting of buffer mappings.
- *   - some logic for moving the buffers to the best memory pools for
- *     given operations.
- *
- * Most of this is to make it easier to implement the fixed-layout
- * mipmap tree required by intel hardware in the face of GL's
- * programming interface where each image can be specifed in random
- * order and it isn't clear what layout the tree should have until the
- * last moment.
- */
-
-#include "intel_context.h"
-#include "intel_regions.h"
-#include "intel_blit.h"
-#include "bufmgr.h"
-#include "imports.h"
-
-/* XXX: Thread safety?
- */
-GLubyte *intel_region_map(struct intel_context *intel, struct intel_region *region)
-{
-   DBG("%s\n", __FUNCTION__);
-   if (!region->map_refcount++) {
-      region->map = bmMapBuffer(intel, region->buffer, 0);
-      if (!region->map)
-	 region->map_refcount--;
-   }
-
-   return region->map;
-}
-
-void intel_region_unmap(struct intel_context *intel, 
-			struct intel_region *region)
-{
-   DBG("%s\n", __FUNCTION__);
-   if (!--region->map_refcount) {
-      bmUnmapBufferAUB(intel, region->buffer, 0, 0);
-      region->map = NULL;
-   }
-}
-
-struct intel_region *intel_region_alloc( struct intel_context *intel, 
-					 GLuint cpp,
-					 GLuint pitch, 
-					 GLuint height )
-{
-   struct intel_region *region = calloc(sizeof(*region), 1);
-
-   DBG("%s %dx%dx%d == 0x%x bytes\n", __FUNCTION__,
-       cpp, pitch, height, cpp*pitch*height);
-
-   region->cpp = cpp;
-   region->pitch = pitch;
-   region->height = height; 	/* needed? */
-   region->refcount = 1;
-
-   bmGenBuffers(intel, "tex", 1, &region->buffer, 6);
-   bmBufferData(intel, region->buffer, pitch * cpp * height, NULL, 0);
-
-   return region;
-}
-
-void intel_region_reference( struct intel_region **dst,
-			     struct intel_region *src)
-{
-   src->refcount++;
-   assert(*dst == NULL);
-   *dst = src;
-}
-
-void intel_region_release( struct intel_context *intel,
-			   struct intel_region **region )
-{
-   if (!*region)
-      return;
-
-   DBG("%s %d\n", __FUNCTION__, (*region)->refcount-1);
-   
-   if (--(*region)->refcount == 0) {
-      assert((*region)->map_refcount == 0);
-      bmDeleteBuffers(intel, 1, &(*region)->buffer);
-      free(*region);
-   }
-   *region = NULL;
-}
-
-
-struct intel_region *intel_region_create_static( struct intel_context *intel, 
-						 GLuint mem_type,
-						 GLuint offset,
-						 void *virtual,
-						 GLuint cpp,
-						 GLuint pitch, 
-						 GLuint height,
-						 GLuint size,
-						 GLboolean tiled )
-{
-   struct intel_region *region = calloc(sizeof(*region), 1);
-   GLint pool;
-
-   DBG("%s\n", __FUNCTION__);
-
-   region->cpp = cpp;
-   region->pitch = pitch;
-   region->height = height; 	/* needed? */
-   region->refcount = 1;
-   region->tiled = tiled;
-
-   /* Recipe for creating a static buffer - create a static pool with
-    * the right offset and size, generate a buffer and use a special
-    * call to bind it to all of the memory in that pool.
-    */
-   pool = bmInitPool(intel, offset, virtual, size, 
-		     (BM_MEM_AGP |
-		      BM_NO_UPLOAD | 
-		      BM_NO_EVICT | 
-		      BM_NO_MOVE));
-   if (pool < 0) {
-      _mesa_printf("bmInitPool failed for static region\n");
-      exit(1);
-   }
-
-   region->buffer = bmGenBufferStatic(intel, pool);
-
-   return region;
-}
-
-
-
-
-void _mesa_copy_rect( GLubyte *dst,
-		      GLuint cpp,
-		      GLuint dst_pitch,
-		      GLuint dst_x, 
-		      GLuint dst_y,
-		      GLuint width,
-		      GLuint height,
-		      const GLubyte *src,
-		      GLuint src_pitch,
-		      GLuint src_x,
-		      GLuint src_y )
-{
-   GLuint i;
-
-   dst_pitch *= cpp;
-   src_pitch *= cpp;
-   dst += dst_x * cpp;
-   src += src_x * cpp;
-   dst += dst_y * dst_pitch;
-   src += src_y * dst_pitch;
-   width *= cpp;
-
-   if (width == dst_pitch && 
-       width == src_pitch)
-      do_memcpy(dst, src, height * width);
-   else {
-      for (i = 0; i < height; i++) {
-	 do_memcpy(dst, src, width);
-	 dst += dst_pitch;
-	 src += src_pitch;
-      }
-   }
-}
-
-
-/* Upload data to a rectangular sub-region.  Lots of choices how to do this:
- *
- * - memcpy by span to current destination
- * - upload data as new buffer and blit
- *
- * Currently always memcpy.
- */
-GLboolean intel_region_data(struct intel_context *intel, 
-			    struct intel_region *dst,
-			    GLuint dst_offset,
-			    GLuint dstx, GLuint dsty,
-			    const void *src, GLuint src_pitch,
-			    GLuint srcx, GLuint srcy,
-			    GLuint width, GLuint height)
-{
-   DBG("%s\n", __FUNCTION__);
-
-   if (width == dst->pitch && 
-       width == src_pitch &&
-       dst_offset == 0 &&
-       height == dst->height &&
-       srcx == 0 &&
-       srcy == 0) 
-   {
-      return (bmBufferDataAUB(intel,
-			      dst->buffer,
-			      dst->cpp * width * dst->height,
-			      src, 0, 0, 0) == 0);
-   }
-   else {
-      GLubyte *map = intel_region_map(intel, dst);
-
-      if (map) {
-	 assert (dst_offset + dstx + width + 
-		 (dsty + height - 1) * dst->pitch * dst->cpp <= 
-		 dst->pitch * dst->cpp * dst->height);
-	 
-	 _mesa_copy_rect(map + dst_offset,
-			 dst->cpp,
-			 dst->pitch,
-			 dstx, dsty,
-			 width, height,
-			 src,
-			 src_pitch,
-			 srcx, srcy);      
-	 
-	 intel_region_unmap(intel, dst);
-	 return GL_TRUE;
-      }
-      else 
-	 return GL_FALSE;
-   }
-}
-			  
-/* Copy rectangular sub-regions. Need better logic about when to
- * push buffers into AGP - will currently do so whenever possible.
- */
-void intel_region_copy( struct intel_context *intel,
-			struct intel_region *dst,
-			GLuint dst_offset,
-			GLuint dstx, GLuint dsty,
-			struct intel_region *src,
-			GLuint src_offset,
-			GLuint srcx, GLuint srcy,
-			GLuint width, GLuint height )
-{
-   DBG("%s\n", __FUNCTION__);
-
-   assert(src->cpp == dst->cpp);
-
-   intelEmitCopyBlit(intel,
-		     dst->cpp,
-		     src->pitch, src->buffer, src_offset, src->tiled,
-		     dst->pitch, dst->buffer, dst_offset, dst->tiled,
-		     srcx, srcy,
-		     dstx, dsty,
-		     width, height,
-		     GL_COPY );
-}
-
-/* Fill a rectangular sub-region.  Need better logic about when to
- * push buffers into AGP - will currently do so whenever possible.
- */
-void intel_region_fill( struct intel_context *intel,
-			struct intel_region *dst,
-			GLuint dst_offset,
-			GLuint dstx, GLuint dsty,
-			GLuint width, GLuint height,
-			GLuint color )
-{
-   DBG("%s\n", __FUNCTION__);
-   
-   intelEmitFillBlit(intel,
-		     dst->cpp,
-		     dst->pitch, dst->buffer, dst_offset, dst->tiled,
-		     dstx, dsty,
-		     width, height,
-		     color );
-}
-
+../intel/intel_regions.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_regions.h b/src/mesa/drivers/dri/i965/intel_regions.h
deleted file mode 100644
index d2235f1275b..00000000000
--- a/src/mesa/drivers/dri/i965/intel_regions.h
+++ /dev/null
@@ -1,140 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_REGIONS_H
-#define INTEL_REGIONS_H
-
-#include "mtypes.h"
-#include "bufmgr.h"		/* for DBG! */
-struct intel_context;
-
-/* A layer on top of the bufmgr buffers that adds a few useful things:
- *
- * - Refcounting for local buffer references.
- * - Refcounting for buffer maps
- * - Buffer dimensions - pitch and height.
- * - Blitter commands for copying 2D regions between buffers.
- */
-struct intel_region {
-   struct buffer *buffer;
-   GLuint refcount;
-   GLuint cpp;
-   GLuint pitch;
-   GLuint height;
-   GLboolean tiled;
-   GLubyte *map;
-   GLuint map_refcount;
-};
-
-/* Allocate a refcounted region.  Pointers to regions should only be
- * copied by calling intel_reference_region().
- *
- * No support for dynamically allocating tiled regions at this point.
- */
-struct intel_region *intel_region_alloc( struct intel_context *intel,
-					 GLuint cpp,
-					 GLuint pitch, 
-					 GLuint height );
-
-void intel_region_reference( struct intel_region **dst, 
-			     struct intel_region *src );
-
-void intel_region_release(struct intel_context *intel,
-			  struct intel_region **ib );
-
-/* Static regions may be tiled.  The assumption is that the X server
- * has set up fence registers to define tiled zones in agp and these
- * buffers are within those zones.  Tiling regions without fence
- * registers is more work.
- */
-struct intel_region *intel_region_create_static( struct intel_context *intel,
-						 GLuint mem_type,
-						 GLuint offset,
-						 void *virtual,
-						 GLuint cpp,
-						 GLuint pitch,
-						 GLuint height,
-						 GLuint size,
-						 GLboolean tiled );
-
-/* Map/unmap regions.  This is refcounted also: 
- */
-GLubyte *intel_region_map(struct intel_context *intel, 
-		       struct intel_region *ib);
-
-void intel_region_unmap(struct intel_context *intel,
-			struct intel_region *ib);
-
-
-/* Upload data to a rectangular sub-region
- */
-GLboolean intel_region_data(struct intel_context *intel, 
-			    struct intel_region *dest,
-			    GLuint dest_offset,
-			    GLuint destx, GLuint desty,
-			    const void *src, GLuint src_stride,
-			    GLuint srcx, GLuint srcy,
-			    GLuint width, GLuint height);
-			  
-/* Copy rectangular sub-regions
- */
-void intel_region_copy( struct intel_context *intel,
-			struct intel_region *dest,
-			GLuint dest_offset,
-			GLuint destx, GLuint desty,
-			struct intel_region *src,
-			GLuint src_offset,
-			GLuint srcx, GLuint srcy,
-			GLuint width, GLuint height );
-
-/* Fill a rectangular sub-region
- */
-void intel_region_fill( struct intel_context *intel,
-			struct intel_region *dest,
-			GLuint dest_offset,
-			GLuint destx, GLuint desty,
-			GLuint width, GLuint height,
-			GLuint color );
-
-
-/***********************************************************************
- * Misc utilities: move to somewhere generic
- */
-void _mesa_copy_rect( GLubyte *dst,
-		      GLuint cpp,
-		      GLuint dst_pitch,
-		      GLuint dst_x, 
-		      GLuint dst_y,
-		      GLuint width,
-		      GLuint height,
-		      const GLubyte *src,
-		      GLuint src_pitch,
-		      GLuint src_x,
-		      GLuint src_y );
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/intel_screen.c b/src/mesa/drivers/dri/i965/intel_screen.c
index bc3dd30b7ed..f2db48272b9 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_screen.c
+++ b/src/mesa/drivers/dri/i965/intel_screen.c
@@ -1,701 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "matrix.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
-#include "utils.h"
-#include "vblank.h"
-#include "xmlpool.h"
-
-
-#include "intel_screen.h"
-
-#include "intel_context.h"
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_ioctl.h"
-
-#include "i830_dri.h"
-
-PUBLIC const char __driConfigOptions[] =
-DRI_CONF_BEGIN
-    DRI_CONF_SECTION_PERFORMANCE
-       DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS) 
-       DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
-    DRI_CONF_SECTION_END
-    DRI_CONF_SECTION_QUALITY
-       DRI_CONF_FORCE_S3TC_ENABLE(false)
-       DRI_CONF_ALLOW_LARGE_TEXTURES(1)
-      DRI_CONF_SECTION_END
-DRI_CONF_END;
-const GLuint __driNConfigOptions = 4;
-
-#ifdef USE_NEW_INTERFACE
-static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
-#endif /*USE_NEW_INTERFACE*/
-
-/**
- * Map all the memory regions described by the screen.
- * \return GL_TRUE if success, GL_FALSE if error.
- */
-GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-
-   if (intelScreen->front.handle) {
-      if (drmMap(sPriv->fd,
-                 intelScreen->front.handle,
-                 intelScreen->front.size,
-                 (drmAddress *)&intelScreen->front.map) != 0) {
-         _mesa_problem(NULL, "drmMap(frontbuffer) failed!");
-         return GL_FALSE;
-      }
-   } else {
-      /* Use the old static allocation method if the server isn't setting up
-       * a movable handle for us.  Add in the front buffer offset from
-       * framebuffer start, as our span routines (unlike other drivers) expect
-       * the renderbuffer address to point to the beginning of the
-       * renderbuffer.
-       */
-      intelScreen->front.map = (char *)sPriv->pFB;
-      if (intelScreen->front.map == NULL) {
-	 fprintf(stderr, "Failed to find framebuffer mapping\n");
-	 return GL_FALSE;
-      }
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->back.handle,
-              intelScreen->back.size,
-              (drmAddress *)&intelScreen->back.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->depth.handle,
-              intelScreen->depth.size,
-              (drmAddress *)&intelScreen->depth.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (drmMap(sPriv->fd,
-              intelScreen->tex.handle,
-              intelScreen->tex.size,
-              (drmAddress *)&intelScreen->tex.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (0)
-      printf("Mappings:  front: %p  back: %p  depth: %p  tex: %p\n",
-          intelScreen->front.map,
-          intelScreen->back.map,
-          intelScreen->depth.map,
-          intelScreen->tex.map);
-   return GL_TRUE;
-}
-
-
-void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen)
-{
-#define REALLY_UNMAP 1
-   /* If front.handle is present, we're doing the dynamic front buffer mapping,
-    * but if we've fallen back to static allocation then we shouldn't try to
-    * unmap here.
-    */
-   if (intelScreen->front.handle) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->front.map, intelScreen->front.size) != 0)
-         printf("drmUnmap front failed!\n");
-#endif
-      intelScreen->front.map = NULL;
-   }
-   if (intelScreen->back.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->back.map, intelScreen->back.size) != 0)
-         printf("drmUnmap back failed!\n");
-#endif
-      intelScreen->back.map = NULL;
-   }
-   if (intelScreen->depth.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->depth.map, intelScreen->depth.size);
-      intelScreen->depth.map = NULL;
-#endif
-   }
-   if (intelScreen->tex.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->tex.map, intelScreen->tex.size);
-      intelScreen->tex.map = NULL;
-#endif
-   }
-}
-
-
-static void
-intelPrintDRIInfo(intelScreenPrivate *intelScreen,
-                  __DRIscreenPrivate *sPriv,
-                  I830DRIPtr gDRIPriv)
-{
-   fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->front.size, intelScreen->front.offset,
-           intelScreen->front.pitch);
-   fprintf(stderr, "*** Back size:    0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->back.size, intelScreen->back.offset,
-           intelScreen->back.pitch);
-   fprintf(stderr, "*** Depth size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->depth.size, intelScreen->depth.offset,
-           intelScreen->depth.pitch);
-   fprintf(stderr, "*** Rotated size: 0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->rotated.size, intelScreen->rotated.offset,
-           intelScreen->rotated.pitch);
-   fprintf(stderr, "*** Texture size: 0x%x  offset: 0x%x\n",
-           intelScreen->tex.size, intelScreen->tex.offset);
-   fprintf(stderr, "*** Memory : 0x%x\n", gDRIPriv->mem);
-}
-
-
-static void
-intelPrintSAREA(volatile drmI830Sarea *sarea)
-{
-   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width, sarea->height);
-   fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
-   fprintf(stderr,
-           "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->front_offset, sarea->front_size,
-           (unsigned) sarea->front_handle);
-   fprintf(stderr,
-           "SAREA: back  offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->back_offset, sarea->back_size,
-           (unsigned) sarea->back_handle);
-   fprintf(stderr, "SAREA: depth offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->depth_offset, sarea->depth_size,
-           (unsigned) sarea->depth_handle);
-   fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->tex_offset, sarea->tex_size,
-           (unsigned) sarea->tex_handle);
-   fprintf(stderr, "SAREA: rotation: %d\n", sarea->rotation);
-   fprintf(stderr,
-           "SAREA: rotated offset: 0x%08x  size: 0x%x\n",
-           sarea->rotated_offset, sarea->rotated_size);
-   fprintf(stderr, "SAREA: rotated pitch: %d\n", sarea->rotated_pitch);
-}
-
-
-/**
- * A number of the screen parameters are obtained/computed from
- * information in the SAREA.  This function updates those parameters.
- */
-void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           volatile drmI830Sarea *sarea)
-{
-   intelScreen->width = sarea->width;
-   intelScreen->height = sarea->height;
-
-   intelScreen->front.offset = sarea->front_offset;
-   intelScreen->front.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->front.handle = sarea->front_handle;
-   intelScreen->front.size = sarea->front_size;
-   intelScreen->front.tiled = sarea->front_tiled;
-
-   intelScreen->back.offset = sarea->back_offset;
-   intelScreen->back.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->back.handle = sarea->back_handle;
-   intelScreen->back.size = sarea->back_size;
-   intelScreen->back.tiled = sarea->back_tiled;
-
-   intelScreen->depth.offset = sarea->depth_offset;
-   intelScreen->depth.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->depth.handle = sarea->depth_handle;
-   intelScreen->depth.size = sarea->depth_size;
-   intelScreen->depth.tiled = sarea->depth_tiled;
-
-   intelScreen->tex.offset = sarea->tex_offset;
-   intelScreen->logTextureGranularity = sarea->log_tex_granularity;
-   intelScreen->tex.handle = sarea->tex_handle;
-   intelScreen->tex.size = sarea->tex_size;
-
-   intelScreen->rotated.offset = sarea->rotated_offset;
-   intelScreen->rotated.pitch = sarea->rotated_pitch * intelScreen->cpp;
-   intelScreen->rotated.size = sarea->rotated_size;
-   intelScreen->rotated.tiled = sarea->rotated_tiled;
-   intelScreen->current_rotation = sarea->rotation;
-#if 0
-   matrix23Rotate(&intelScreen->rotMatrix,
-                  sarea->width, sarea->height, sarea->rotation);
-#endif
-   intelScreen->rotatedWidth = sarea->virtualX;
-   intelScreen->rotatedHeight = sarea->virtualY;
-
-   if (0)
-      intelPrintSAREA(sarea);
-}
-
-
-static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen;
-   I830DRIPtr         gDRIPriv = (I830DRIPtr)sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
-   volatile drmI830Sarea *sarea;
-
-   if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
-      fprintf(stderr,"\nERROR!  sizeof(I830DRIRec) (%ld) does not match passed size from device driver (%d)\n", (unsigned long)sizeof(I830DRIRec), sPriv->devPrivSize);
-      return GL_FALSE;
-   }
-
-   /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *)CALLOC(sizeof(intelScreenPrivate));
-   if (!intelScreen) {
-      fprintf(stderr,"\nERROR!  Allocating private area failed\n");
-      return GL_FALSE;
-   }
-   /* parse information in __driConfigOptions */
-   driParseOptionInfo (&intelScreen->optionCache,
-		       __driConfigOptions, __driNConfigOptions);
-
-   intelScreen->driScrnPriv = sPriv;
-   sPriv->private = (void *)intelScreen;
-   intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
-   sarea = (volatile drmI830Sarea *)
-         (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
-
-   intelScreen->deviceID = gDRIPriv->deviceID;
-   intelScreen->mem = gDRIPriv->mem;
-   intelScreen->cpp = gDRIPriv->cpp;
-
-   switch (gDRIPriv->bitsPerPixel) {
-   case 15: intelScreen->fbFormat = DV_PF_555; break;
-   case 16: intelScreen->fbFormat = DV_PF_565; break;
-   case 32: intelScreen->fbFormat = DV_PF_8888; break;
-   }
-			 
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   if (0)
-      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr,"\nERROR!  mapping regions\n");
-      _mesa_free(intelScreen);
-      sPriv->private = NULL;
-      return GL_FALSE;
-   }
-
-   intelScreen->drmMinor = sPriv->drmMinor;
-
-   /* Determine if IRQs are active? */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_IRQ_ACTIVE;
-      gp.value = &intelScreen->irq_active;
-
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
-      if (ret) {
-	 fprintf(stderr, "drmI830GetParam: %d\n", ret);
-	 return GL_FALSE;
-      }
-   }
-
-   /* Determine if batchbuffers are allowed */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_ALLOW_BATCHBUFFER;
-      gp.value = &intelScreen->allow_batchbuffer;
-
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
-      if (ret) {
-	 fprintf(stderr, "drmI830GetParam: (%d) %d\n", gp.param, ret);
-	 return GL_FALSE;
-      }
-   }
-
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
-      (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
-      (*glx_enable_extension)( psc, "GLX_MESA_copy_sub_buffer" );
-   }
-   
-   return GL_TRUE;
-}
-
-
-static void intelDestroyScreen(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-
-   intelUnmapScreenRegions(intelScreen);
-   FREE(intelScreen);
-   sPriv->private = NULL;
-}
-
-static GLboolean intelCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-				    __DRIdrawablePrivate *driDrawPriv,
-				    const __GLcontextModes *mesaVis,
-				    GLboolean isPixmap )
-{
-   intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
-
-   if (isPixmap) {
-      return GL_FALSE; /* not implemented */
-   } else {
-      GLboolean swStencil = (mesaVis->stencilBits > 0 && 
-			     mesaVis->depthBits != 24);
-
-      struct gl_framebuffer *fb = _mesa_create_framebuffer(mesaVis);
-
-      {
-         driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->front.map,
-                                 screen->cpp,
-                                 screen->front.offset, screen->front.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(frontRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
-      }
-
-      if (mesaVis->doubleBufferMode) {
-         driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->back.map,
-                                 screen->cpp,
-                                 screen->back.offset, screen->back.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(backRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
-      }
-
-      if (mesaVis->depthBits == 16) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-      else if (mesaVis->depthBits == 24) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
-      }
-
-      if (mesaVis->stencilBits > 0 && !swStencil) {
-         driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(stencilRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
-      }
-
-      _mesa_add_soft_renderbuffers(fb,
-                                   GL_FALSE, /* color */
-                                   GL_FALSE, /* depth */
-                                   swStencil,
-                                   mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* alpha */
-                                   GL_FALSE /* aux */);
-      driDrawPriv->driverPrivate = (void *) fb;
-
-      return (driDrawPriv->driverPrivate != NULL);
-   }
-}
-
-static void intelDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
-{
-   _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
-}
-
-
-/**
- * Get information about previous buffer swaps.
- */
-static int
-intelGetSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
-{
-   struct intel_context *intel;
-
-   if ( (dPriv == NULL) || (dPriv->driContextPriv == NULL)
-	|| (dPriv->driContextPriv->driverPrivate == NULL)
-	|| (sInfo == NULL) ) {
-      return -1;
-   }
-
-   intel = dPriv->driContextPriv->driverPrivate;
-   sInfo->swap_count = intel->swap_count;
-   sInfo->swap_ust = intel->swap_ust;
-   sInfo->swap_missed_count = intel->swap_missed_count;
-
-   sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
-       ? driCalculateSwapUsage( dPriv, 0, intel->swap_missed_ust )
-       : 0.0;
-
-   return 0;
-}
-
-
-/* There are probably better ways to do this, such as an
- * init-designated function to register chipids and createcontext
- * functions.
- */
-extern GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
-				    __DRIcontextPrivate *driContextPriv,
-				    void *sharedContextPrivate);
-
-extern GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate);
-
-extern GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate);
-
-
-
-
-static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate)
-{
-#if 0
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   switch (intelScreen->deviceID) {
-   case PCI_CHIP_845_G:
-   case PCI_CHIP_I830_M:
-   case PCI_CHIP_I855_GM:
-   case PCI_CHIP_I865_G:
-      return i830CreateContext( mesaVis, driContextPriv, 
-				sharedContextPrivate );
-
-   case PCI_CHIP_I915_G:
-   case PCI_CHIP_I915_GM:
-   case PCI_CHIP_I945_G:
-   case PCI_CHIP_I945_GM:
-      return i915CreateContext( mesaVis, driContextPriv, 
-			       sharedContextPrivate );
- 
-   default:
-      fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
-      return GL_FALSE;
-   }
-#else
-   return brwCreateContext( mesaVis, driContextPriv, 
-			    sharedContextPrivate );
-#endif
-}
-
-
-static const struct __DriverAPIRec intelAPI = {
-   .InitDriver      = intelInitDriver,
-   .DestroyScreen   = intelDestroyScreen,
-   .CreateContext   = intelCreateContext,
-   .DestroyContext  = intelDestroyContext,
-   .CreateBuffer    = intelCreateBuffer,
-   .DestroyBuffer   = intelDestroyBuffer,
-   .SwapBuffers     = intelSwapBuffers,
-   .MakeCurrent     = intelMakeCurrent,
-   .UnbindContext   = intelUnbindContext,
-   .GetSwapInfo     = intelGetSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = intelCopySubBuffer
-};
-
-
-static __GLcontextModes *
-intelFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
-{
-   __GLcontextModes * modes;
-   __GLcontextModes * m;
-   unsigned num_modes;
-   unsigned depth_buffer_factor;
-   unsigned back_buffer_factor;
-   GLenum fb_format;
-   GLenum fb_type;
-
-   /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
-    * support pageflipping at all.
-    */
-   static const GLenum back_buffer_modes[] = {
-      GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
-   };
-
-   uint8_t depth_bits_array[3];
-   uint8_t stencil_bits_array[3];
-
-
-   depth_bits_array[0] = 0;
-   depth_bits_array[1] = depth_bits;
-   depth_bits_array[2] = depth_bits;
-
-   /* Just like with the accumulation buffer, always provide some modes
-    * with a stencil buffer.  It will be a sw fallback, but some apps won't
-    * care about that.
-    */
-   stencil_bits_array[0] = 0;
-   stencil_bits_array[1] = 0;
-   stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
-   back_buffer_factor  = (have_back_buffer) ? 3 : 1;
-
-   num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
-    if ( pixel_bits == 16 ) {
-        fb_format = GL_RGB;
-        fb_type = GL_UNSIGNED_SHORT_5_6_5;
-    }
-    else {
-        fb_format = GL_BGRA;
-        fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
-    }
-
-   modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-   m = modes;
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-
-   /* Mark the visual as slow if there are "fake" stencil bits.
-    */
-   for ( m = modes ; m != NULL ; m = m->next ) {
-      if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	 m->visualRating = GLX_SLOW_CONFIG;
-      }
-   }
-
-   return modes;
-}
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 1, 6, 0 };
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 3, 0 };
-
-   dri_interface = interface;
-
-   if ( ! driCheckDriDdxDrmVersions2( "i915",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &intelAPI);
-   if ( psp != NULL ) {
-      I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-      *driver_modes = intelFillInModes( dri_priv->cpp * 8,
-					(dri_priv->cpp == 2) ? 16 : 24,
-					(dri_priv->cpp == 2) ? 0  : 8,
-					GL_TRUE );
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      intelInitExtensions(NULL, GL_FALSE);
-   }
-
-   return (void *) psp;
-}
+../intel/intel_screen.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_span.c b/src/mesa/drivers/dri/i965/intel_span.c
index 60fbeccdc54..05e5e8e5833 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_span.c
+++ b/src/mesa/drivers/dri/i965/intel_span.c
@@ -1,283 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
-
-#include "intel_screen.h"
-#include "intel_regions.h"
-#include "intel_span.h"
-#include "intel_ioctl.h"
-#include "intel_tex.h"
-#include "intel_batchbuffer.h"
-#include "swrast/swrast.h"
-
-#undef DBG
-#define DBG 0
-
-#define LOCAL_VARS						\
-   struct intel_context *intel = intel_context(ctx);                    \
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLushort p;							\
-   (void) buf; (void) p
-
-#define LOCAL_DEPTH_VARS					\
-   struct intel_context *intel = intel_context(ctx);                    \
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch
-
-#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS 
-
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR565(color[0],color[1],color[2])
-
-#define Y_FLIP(_y) (height - _y - 1)
-
-#define HW_LOCK()
-
-#define HW_UNLOCK()
-
-/* 16 bit, 565 rgb color spanline and pixel functions
- */
-#define WRITE_RGBA( _x, _y, r, g, b, a )				\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = ( (((int)r & 0xf8) << 8) |	\
-		                             (((int)g & 0xfc) << 3) |	\
-		                             (((int)b & 0xf8) >> 3))
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch) = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (((p >> 11) & 0x1f) * 255) / 31;			\
-   rgba[1] = (((p >>  5) & 0x3f) * 255) / 63;			\
-   rgba[2] = (((p >>  0) & 0x1f) * 255) / 31;			\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_565
-#include "spantmp.h"
-
-/* 15 bit, 555 rgb color spanline and pixel functions
- */
-#define WRITE_RGBA( _x, _y, r, g, b, a )			\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = (((r & 0xf8) << 7) |	\
-		                            ((g & 0xf8) << 3) |	\
-                         		    ((b & 0xf8) >> 3))
-
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (p >> 7) & 0xf8;					\
-   rgba[1] = (p >> 3) & 0xf8;					\
-   rgba[2] = (p << 3) & 0xf8;					\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_555
-#include "spantmp.h"
-
-/* 16 bit depthbuffer functions.
- */
-#define WRITE_DEPTH( _x, _y, d ) \
-   *(GLushort *)(buf + (_x)*2 + (_y)*pitch)  = d;
-
-#define READ_DEPTH( d, _x, _y )	\
-   d = *(GLushort *)(buf + (_x)*2 + (_y)*pitch);	 
-
-
-#define TAG(x) intel##x##_z16
-#include "depthtmp.h"
-
-
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-   struct intel_context *intel = intel_context(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *)drb->Base.Data +				\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLuint p;							\
-   (void) buf; (void) p
-
-#undef INIT_MONO_PIXEL
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR8888(color[0],color[1],color[2],color[3])
-
-/* 32 bit, 8888 argb color spanline and pixel functions
- */
-#define WRITE_RGBA(_x, _y, r, g, b, a)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = ((r << 16) |	\
-					  (g << 8)  |	\
-					  (b << 0)  |	\
-					  (a << 24) )
-
-#define WRITE_PIXEL(_x, _y, p)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = p
-
-
-#define READ_RGBA(rgba, _x, _y)					\
-    do {							\
-	GLuint p = *(GLuint *)(buf + _x*4 + _y*pitch);		\
-	rgba[0] = (p >> 16) & 0xff;				\
-	rgba[1] = (p >> 8)  & 0xff;				\
-	rgba[2] = (p >> 0)  & 0xff;				\
-	rgba[3] = (p >> 24) & 0xff;				\
-    } while (0)
-
-#define TAG(x) intel##x##_8888
-#include "spantmp.h"
-
-
-/* 24/8 bit interleaved depth/stencil functions
- */
-#define WRITE_DEPTH( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xff000000;					\
-   tmp |= (d) & 0xffffff;				\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
-}
-
-#define READ_DEPTH( d, _x, _y )		\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) & 0xffffff;
-
-
-#define TAG(x) intel##x##_z24_s8
-#include "depthtmp.h"
-
-#define WRITE_STENCIL( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xffffff;					\
-   tmp |= ((d)<<24);					\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
-}
-
-#define READ_STENCIL( d, _x, _y )			\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) >> 24;
-
-#define TAG(x) intel##x##_z24_s8
-#include "stenciltmp.h"
-
-
-/* Move locking out to get reasonable span performance.
- */
-void intelSpanRenderStart( GLcontext *ctx )
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   if (intel->need_flush) {
-      LOCK_HARDWARE(intel);
-      intel->vtbl.emit_flush(intel, 0);
-      intel_batchbuffer_flush(intel->batch);
-      intel->need_flush = 0;
-      UNLOCK_HARDWARE(intel);
-      intelFinish(&intel->ctx);
-   }
-
-
-   LOCK_HARDWARE(intel);
-
-   /* Just map the framebuffer and all textures.  Bufmgr code will
-    * take care of waiting on the necessary fences:
-    */
-   intel_region_map(intel, intel->front_region);
-   intel_region_map(intel, intel->back_region);
-   intel_region_map(intel, intel->depth_region);
-}
-
-void intelSpanRenderFinish( GLcontext *ctx )
-{
-   struct intel_context *intel = intel_context( ctx );
-
-   _swrast_flush( ctx );
-
-   /* Now unmap the framebuffer:
-    */
-   intel_region_unmap(intel, intel->front_region);
-   intel_region_unmap(intel, intel->back_region);
-   intel_region_unmap(intel, intel->depth_region);
-
-   UNLOCK_HARDWARE( intel );
-}
-
-void intelInitSpanFuncs( GLcontext *ctx )
-{
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart = intelSpanRenderStart;
-   swdd->SpanRenderFinish = intelSpanRenderFinish; 
-}
-
-
-/**
- * Plug in the Get/Put routines for the given driRenderbuffer.
- */
-void
-intelSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
-{
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 5 && vis->blueBits == 5) {
-         intelInitPointers_555(&drb->Base);
-      }
-      else if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         intelInitPointers_565(&drb->Base);
-      }
-      else {
-         assert(vis->redBits == 8);
-         assert(vis->greenBits == 8);
-         assert(vis->blueBits == 8);
-         intelInitPointers_8888(&drb->Base);
-      }
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
-      intelInitDepthPointers_z16(&drb->Base);
-   }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
-      intelInitDepthPointers_z24_s8(&drb->Base);
-   }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
-      intelInitStencilPointers_z24_s8(&drb->Base);
-   }
-}
+../intel/intel_span.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_state.c b/src/mesa/drivers/dri/i965/intel_state.c
index 2e442db6198..67ef5f78c1a 100644
--- a/src/mesa/drivers/dri/i965/intel_state.c
+++ b/src/mesa/drivers/dri/i965/intel_state.c
@@ -26,18 +26,43 @@
  **************************************************************************/
 
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "colormac.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/dd.h"
 
 #include "intel_screen.h"
 #include "intel_context.h"
 #include "intel_regions.h"
 #include "swrast/swrast.h"
 
+int intel_translate_shadow_compare_func( GLenum func )
+{
+   switch(func) {
+   case GL_NEVER: 
+       return COMPAREFUNC_ALWAYS; 
+   case GL_LESS: 
+       return COMPAREFUNC_LEQUAL; 
+   case GL_LEQUAL: 
+       return COMPAREFUNC_LESS;
+   case GL_GREATER: 
+       return COMPAREFUNC_GEQUAL; 
+   case GL_GEQUAL: 
+      return COMPAREFUNC_GREATER; 
+   case GL_NOTEQUAL: 
+      return COMPAREFUNC_EQUAL; 
+   case GL_EQUAL: 
+      return COMPAREFUNC_NOTEQUAL; 
+   case GL_ALWAYS: 
+       return COMPAREFUNC_NEVER; 
+   }
+
+   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
+   return COMPAREFUNC_NEVER; 
+}
+
 int intel_translate_compare_func( GLenum func )
 {
    switch(func) {
@@ -170,15 +195,16 @@ int intel_translate_logic_op( GLenum opcode )
 static void intelClearColor(GLcontext *ctx, const GLfloat color[4])
 {
    struct intel_context *intel = intel_context(ctx);
-   intelScreenPrivate *screen = intel->intelScreen;
 
    UNCLAMPED_FLOAT_TO_RGBA_CHAN(intel->clear_chan, color);
 
-   intel->ClearColor = INTEL_PACKCOLOR(screen->fbFormat,
-				       intel->clear_chan[0], 
-				       intel->clear_chan[1], 
-				       intel->clear_chan[2], 
-				       intel->clear_chan[3]);
+   intel->ClearColor8888 = INTEL_PACKCOLOR8888(intel->clear_chan[0],
+					       intel->clear_chan[1],
+					       intel->clear_chan[2],
+					       intel->clear_chan[3]);
+   intel->ClearColor565 = INTEL_PACKCOLOR565(intel->clear_chan[0],
+					     intel->clear_chan[1],
+					     intel->clear_chan[2]);
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/intel_tex.c b/src/mesa/drivers/dri/i965/intel_tex.c
index 4523969bfa5..d77ce749a3e 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_tex.c
+++ b/src/mesa/drivers/dri/i965/intel_tex.c
@@ -1,315 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "mtypes.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texobj.h"
-#include "swrast/swrast.h"
-
-
-#include "intel_context.h"
-#include "intel_tex.h"
-#include "intel_mipmap_tree.h"
-
-
-static GLuint target_to_face( GLenum target )
-{
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
-      return ((GLuint) target - 
-	      (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X);
-   default:
-      return 0;
-   }
-}
-
-static void intelTexImage1D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   _mesa_store_teximage1d( ctx, target, level, internalFormat,
-			   width, border, format, type,
-			   pixels, packing, texObj, texImage );
-
-   intelObj->dirty_images[0] |= (1 << level);
-   intelObj->dirty |= 1;
-}
-
-static void intelTexSubImage1D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset,
-				GLsizei width,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   _mesa_store_texsubimage1d(ctx, target, level, xoffset, width, 
-			     format, type, pixels, packing, texObj,
-			     texImage);
-
-   intelObj->dirty_images[0] |= (1 << level);
-   intelObj->dirty |= 1;
-}
-
-
-/* Handles 2D, CUBE, RECT:
- */
-static void intelTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint height, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   GLuint face = target_to_face(target);
-
-   _mesa_store_teximage2d( ctx, target, level, internalFormat,
-			   width, height, border, format, type,
-			   pixels, packing, texObj, texImage );
-
-   intelObj->dirty_images[face] |= (1 << level);
-   intelObj->dirty |= 1 << face;
-}
-
-static void intelTexSubImage2D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset, GLint yoffset,
-			       GLsizei width, GLsizei height,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   GLuint face = target_to_face(target);
-
-   _mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width, 
-			     height, format, type, pixels, packing, texObj,
-			     texImage);
-
-   intelObj->dirty_images[face] |= (1 << level);
-   intelObj->dirty |= 1 << face;
-}
-
-static void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-                              GLint internalFormat,
-                              GLint width, GLint height, GLint border,
-                              GLsizei imageSize, const GLvoid *data,
-                              struct gl_texture_object *texObj,
-                              struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   GLuint face = target_to_face(target);
-
-   _mesa_store_compressed_teximage2d(ctx, target, level, internalFormat, width,
-				     height, border, imageSize, data, texObj, texImage);
-   
-   intelObj->dirty_images[face] |= (1 << level);
-   intelObj->dirty |= 1 << face;
-}
-
-
-static void intelCompressedTexSubImage2D( GLcontext *ctx, GLenum target, GLint level,
-                                 GLint xoffset, GLint yoffset,
-                                 GLsizei width, GLsizei height,
-                                 GLenum format,
-                                 GLsizei imageSize, const GLvoid *data,
-                                 struct gl_texture_object *texObj,
-                                 struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   GLuint face = target_to_face(target);
-
-   _mesa_store_compressed_texsubimage2d(ctx, target, level, xoffset, yoffset, width,
-					height, format, imageSize, data, texObj, texImage);
-   
-   intelObj->dirty_images[face] |= (1 << level);
-   intelObj->dirty |= 1 << face;
-}
-
-
-static void intelTexImage3D( GLcontext *ctx, GLenum target, GLint level,
-                            GLint internalFormat,
-                            GLint width, GLint height, GLint depth,
-                            GLint border,
-                            GLenum format, GLenum type, const GLvoid *pixels,
-                            const struct gl_pixelstore_attrib *packing,
-                            struct gl_texture_object *texObj,
-                            struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   _mesa_store_teximage3d(ctx, target, level, internalFormat,
-			  width, height, depth, border,
-			  format, type, pixels,
-			  &ctx->Unpack, texObj, texImage);
-   
-   intelObj->dirty_images[0] |= (1 << level);
-   intelObj->dirty |= 1 << 0;
-}
-
-
-static void
-intelTexSubImage3D( GLcontext *ctx, GLenum target, GLint level,
-                   GLint xoffset, GLint yoffset, GLint zoffset,
-                   GLsizei width, GLsizei height, GLsizei depth,
-                   GLenum format, GLenum type,
-                   const GLvoid *pixels,
-                   const struct gl_pixelstore_attrib *packing,
-                   struct gl_texture_object *texObj,
-                   struct gl_texture_image *texImage )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   _mesa_store_texsubimage3d(ctx, target, level, xoffset, yoffset, zoffset,
-                             width, height, depth,
-                             format, type, pixels, packing, texObj, texImage);
-
-   intelObj->dirty_images[0] |= (1 << level);
-   intelObj->dirty |= 1 << 0;
-}
-
-
-
-
-static struct gl_texture_object *intelNewTextureObject( GLcontext *ctx, 
-							GLuint name, 
-							GLenum target )
-{
-   struct intel_texture_object *obj = CALLOC_STRUCT(intel_texture_object);
-
-   _mesa_initialize_texture_object(&obj->base, name, target);
-
-   return &obj->base;
-}
-
-static GLboolean intelIsTextureResident(GLcontext *ctx,
-                                      struct gl_texture_object *texObj)
-{
-#if 0
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-   
-   return 
-      intelObj->mt && 
-      intelObj->mt->region && 
-      intel_is_region_resident(intel, intelObj->mt->region);
-#endif
-   return 1;
-}
-
-
-
-static void intelTexParameter( GLcontext *ctx, 
-			       GLenum target,
-			       struct gl_texture_object *texObj,
-			       GLenum pname, 
-			       const GLfloat *params )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
- 
-   switch (pname) {
-      /* Anything which can affect the calculation of firstLevel and
-       * lastLevel, as changes to these may invalidate the miptree.
-       */
-   case GL_TEXTURE_MIN_FILTER:
-   case GL_TEXTURE_MAG_FILTER:
-   case GL_TEXTURE_BASE_LEVEL:
-   case GL_TEXTURE_MAX_LEVEL:
-   case GL_TEXTURE_MIN_LOD:
-   case GL_TEXTURE_MAX_LOD:
-      intelObj->dirty |= 1;
-      break;
-
-   default:
-      break;
-   }
-}
-
-
-static void
-intel_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj )
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   if (intelObj->mt)
-      intel_miptree_destroy(intel, intelObj->mt);
-
-   _mesa_delete_texture_object( ctx, texObj );
-}
-
-void intelInitTextureFuncs( struct dd_function_table *functions )
-{
-   functions->NewTextureObject          = intelNewTextureObject;
-   functions->TexImage1D                = intelTexImage1D;
-   functions->TexImage2D                = intelTexImage2D;
-   functions->TexImage3D                = intelTexImage3D;
-   functions->TexSubImage1D             = intelTexSubImage1D;
-   functions->TexSubImage2D             = intelTexSubImage2D;
-   functions->TexSubImage3D             = intelTexSubImage3D;
-   functions->CopyTexImage1D            = _swrast_copy_teximage1d;
-   functions->CopyTexImage2D            = _swrast_copy_teximage2d;
-   functions->CopyTexSubImage1D         = _swrast_copy_texsubimage1d;
-   functions->CopyTexSubImage2D         = _swrast_copy_texsubimage2d;
-   functions->CopyTexSubImage3D         = _swrast_copy_texsubimage3d;
-   functions->DeleteTexture             = intel_delete_texture_object;
-   functions->UpdateTexturePalette      = NULL;
-   functions->IsTextureResident = intelIsTextureResident;
-   functions->TestProxyTexImage         = _mesa_test_proxy_teximage;
-   functions->CompressedTexImage2D      = intelCompressedTexImage2D;
-   functions->CompressedTexSubImage2D   = intelCompressedTexSubImage2D;
-   functions->TexParameter              = intelTexParameter;
-}
-
-
-
-
-
+../intel/intel_tex.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_tex_copy.c b/src/mesa/drivers/dri/i965/intel_tex_copy.c
new file mode 120000
index 00000000000..87196c5d1ed
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_tex_copy.c
@@ -0,0 +1 @@
+../intel/intel_tex_copy.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_tex_format.c b/src/mesa/drivers/dri/i965/intel_tex_format.c
new file mode 120000
index 00000000000..3415f754705
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_tex_format.c
@@ -0,0 +1 @@
+../intel/intel_tex_format.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_tex_image.c b/src/mesa/drivers/dri/i965/intel_tex_image.c
new file mode 120000
index 00000000000..567abe4974e
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_tex_image.c
@@ -0,0 +1 @@
+../intel/intel_tex_image.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_tex_subimage.c b/src/mesa/drivers/dri/i965/intel_tex_subimage.c
new file mode 120000
index 00000000000..b3a8a3d7ca7
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_tex_subimage.c
@@ -0,0 +1 @@
+../intel/intel_tex_subimage.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_tex_validate.c b/src/mesa/drivers/dri/i965/intel_tex_validate.c
index 44ee94614d5..41a75674c27 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i965/intel_tex_validate.c
@@ -1,256 +1 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "mtypes.h"
-#include "macros.h"
-
-#include "intel_context.h"
-#include "intel_mipmap_tree.h"
-#include "intel_tex.h"
-#include "bufmgr.h"
-
-/**
- * Compute which mipmap levels that really need to be sent to the hardware.
- * This depends on the base image size, GL_TEXTURE_MIN_LOD,
- * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
- */
-static void intel_calculate_first_last_level( struct intel_texture_object *intelObj )
-{
-   struct gl_texture_object *tObj = &intelObj->base;
-   const struct gl_texture_image * const baseImage =
-       tObj->Image[0][tObj->BaseLevel];
-
-   /* These must be signed values.  MinLod and MaxLod can be negative numbers,
-    * and having firstLevel and lastLevel as signed prevents the need for
-    * extra sign checks.
-    */
-   int   firstLevel;
-   int   lastLevel;
-
-   /* Yes, this looks overly complicated, but it's all needed.
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_1D:
-   case GL_TEXTURE_2D:
-   case GL_TEXTURE_3D:
-   case GL_TEXTURE_CUBE_MAP:
-      if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) {
-         /* GL_NEAREST and GL_LINEAR only care about GL_TEXTURE_BASE_LEVEL.
-          */
-         firstLevel = lastLevel = tObj->BaseLevel;
-      }
-      else {
-	 /* Currently not taking min/max lod into account here, those
-	  * values are programmed as sampler state elsewhere and we
-	  * upload the same mipmap levels regardless.  Not sure if
-	  * this makes sense as it means it isn't possible for the app
-	  * to use min/max lod to reduce texture memory pressure:
-	  */
-	 firstLevel = tObj->BaseLevel;
-	 lastLevel = MIN2(tObj->BaseLevel + baseImage->MaxLog2, 
-			  tObj->MaxLevel);
-	 lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
-      }
-      break;
-   case GL_TEXTURE_RECTANGLE_NV:
-   case GL_TEXTURE_4D_SGIS:
-      firstLevel = lastLevel = 0;
-      break;
-   default:
-      return;
-   }
-
-   /* save these values */
-   intelObj->firstLevel = firstLevel;
-   intelObj->lastLevel = lastLevel;
-}
-
-static GLboolean copy_image_data_to_tree( struct intel_context *intel,
-					  struct intel_texture_object *intelObj,
-					  struct gl_texture_image *texImage,
-					  GLuint face,
-					  GLuint level)
-{
-   return intel_miptree_image_data(intel,
-				   intelObj->mt,
-				   face,
-				   level,
-				   texImage->Data,
-				   texImage->RowStride,
-				   (texImage->RowStride * 
-				    texImage->Height * 
-				    texImage->TexFormat->TexelBytes));
-}
-
-static void intel_texture_invalidate( struct intel_texture_object *intelObj )
-{
-   GLint nr_faces, face;
-   intelObj->dirty = ~0;
-
-   nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-   for (face = 0; face < nr_faces; face++) 
-      intelObj->dirty_images[face] = ~0;
-}
-
-static void intel_texture_invalidate_cb( struct intel_context *intel,
-					 void *ptr )
-{
-   intel_texture_invalidate( (struct intel_texture_object *) ptr );
-}
-
-
-/*  
- */
-GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
-				   struct gl_texture_object *tObj )
-{
-   struct intel_texture_object *intelObj = intel_texture_object(tObj);
-   GLuint face, i;
-   GLuint nr_faces = 0;
-   struct gl_texture_image *firstImage;
-
-   if( tObj == intel->frame_buffer_texobj )
-      return GL_FALSE;
-   
-   /* We know/require this is true by now: 
-    */
-   assert(intelObj->base._Complete);
-
-   /* What levels must the tree include at a minimum?
-    */
-   if (intelObj->dirty) {
-      intel_calculate_first_last_level( intelObj );
-/*       intel_miptree_destroy(intel, intelObj->mt); */
-/*       intelObj->mt = NULL; */
-   }
-
-   firstImage = intelObj->base.Image[0][intelObj->firstLevel];
-
-   /* Fallback case:
-    */
-   if (firstImage->Border) {
-      if (intelObj->mt) {
-	 intel_miptree_destroy(intel, intelObj->mt);
-	 intelObj->mt = NULL;
-	 /* Set all images dirty:
-	  */
-	 intel_texture_invalidate(intelObj);
-      }
-      return GL_FALSE;
-   }
-
-
-
-   /* Check tree can hold all active levels.  Check tree matches
-    * target, imageFormat, etc.
-    */
-   if (intelObj->mt &&
-       (intelObj->mt->target != intelObj->base.Target ||
-	intelObj->mt->internal_format != firstImage->InternalFormat ||
-	intelObj->mt->first_level != intelObj->firstLevel ||
-	intelObj->mt->last_level != intelObj->lastLevel ||
-	intelObj->mt->width0 != firstImage->Width ||
-	intelObj->mt->height0 != firstImage->Height ||
-	intelObj->mt->depth0 != firstImage->Depth ||
-	intelObj->mt->cpp != firstImage->TexFormat->TexelBytes ||
-	intelObj->mt->compressed != firstImage->IsCompressed)) 
-   {
-      intel_miptree_destroy(intel, intelObj->mt);
-      intelObj->mt = NULL;
-      
-      /* Set all images dirty:
-       */
-      intel_texture_invalidate(intelObj);
-   }
-      
-
-   /* May need to create a new tree:
-    */
-   if (!intelObj->mt) {
-      intelObj->mt = intel_miptree_create(intel,
-					  intelObj->base.Target,
-					  firstImage->InternalFormat,
-					  intelObj->firstLevel,
-					  intelObj->lastLevel,
-					  firstImage->Width,
-					  firstImage->Height,
-					  firstImage->Depth,
-					  firstImage->TexFormat->TexelBytes,
-					  firstImage->IsCompressed);
-
-      /* Tell the buffer manager that we will manage the backing
-       * store, but we still want it to do fencing for us.
-       */
-      bmBufferSetInvalidateCB(intel, 
-			      intelObj->mt->region->buffer,
-			      intel_texture_invalidate_cb,
-			      intelObj,
-			      GL_FALSE);
-   }
-
-   /* Pull in any images not in the object's tree:
-    */
-   if (intelObj->dirty) {
-      nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-      for (face = 0; face < nr_faces; face++) {
-	 if (intelObj->dirty_images[face]) {
-	    for (i = intelObj->firstLevel; i <= intelObj->lastLevel; i++) {
-	       struct gl_texture_image *texImage = intelObj->base.Image[face][i];
-
-	       /* Need to import images in main memory or held in other trees.
-		*/
-	       if (intelObj->dirty_images[face] & (1<<i) &&
-		   texImage) {
-
-		  if (INTEL_DEBUG & DEBUG_TEXTURE)
-		     _mesa_printf("copy data from image %d (%p) into object miptree\n",
-				  i,
-				  texImage->Data);
-
-		  if (!copy_image_data_to_tree(intel,
-					       intelObj,
-					       texImage,
-					       face,
-					       i))
-		     return GL_FALSE;
-
-	       }
-	    }
-	 }
-      }
-
-      /* Only clear the dirty flags if everything went ok:
-       */
-      for (face = 0; face < nr_faces; face++) {
-	 intelObj->dirty_images[face] = 0;
-      }
-
-      intelObj->dirty = 0;
-   }
-
-   return GL_TRUE;
-}
+../intel/intel_tex_validate.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/server/i830_common.h b/src/mesa/drivers/dri/i965/server/i830_common.h
deleted file mode 100644
index 49eb145f8bf..00000000000
--- a/src/mesa/drivers/dri/i965/server/i830_common.h
+++ /dev/null
@@ -1,221 +0,0 @@
-/**************************************************************************
-
-Copyright 2001 VA Linux Systems Inc., Fremont, California.
-Copyright 2002 Tungsten Graphics Inc., Cedar Park, Texas.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ATI, VA LINUX SYSTEMS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef _I830_COMMON_H_
-#define _I830_COMMON_H_
-
-
-#define I830_NR_TEX_REGIONS 255	/* maximum due to use of chars for next/prev */
-#define I830_LOG_MIN_TEX_REGION_SIZE 14
-
-
-/* Driver specific DRM command indices
- * NOTE: these are not OS specific, but they are driver specific
- */
-#define DRM_I830_INIT                     0x00
-#define DRM_I830_FLUSH                    0x01
-#define DRM_I830_FLIP                     0x02
-#define DRM_I830_BATCHBUFFER              0x03
-#define DRM_I830_IRQ_EMIT                 0x04
-#define DRM_I830_IRQ_WAIT                 0x05
-#define DRM_I830_GETPARAM                 0x06
-#define DRM_I830_SETPARAM                 0x07
-#define DRM_I830_ALLOC                    0x08
-#define DRM_I830_FREE                     0x09
-#define DRM_I830_INIT_HEAP                0x0a
-#define DRM_I830_CMDBUFFER                0x0b
-#define DRM_I830_DESTROY_HEAP             0x0c
-#define DRM_I830_MMIO		       	  0x10
-
-typedef struct {
-   enum {
-      I830_INIT_DMA = 0x01,
-      I830_CLEANUP_DMA = 0x02,
-      I830_RESUME_DMA = 0x03
-   } func;
-   unsigned int mmio_offset;
-   int sarea_priv_offset;
-   unsigned int ring_start;
-   unsigned int ring_end;
-   unsigned int ring_size;
-   unsigned int front_offset;
-   unsigned int back_offset;
-   unsigned int depth_offset;
-   unsigned int w;
-   unsigned int h;
-   unsigned int pitch;
-   unsigned int pitch_bits;
-   unsigned int back_pitch;
-   unsigned int depth_pitch;
-   unsigned int cpp;
-   unsigned int chipset;
-} drmI830Init;
-
-typedef struct {
-	drmTextureRegion texList[I830_NR_TEX_REGIONS+1];
-        int last_upload;	/* last time texture was uploaded */
-        int last_enqueue;	/* last time a buffer was enqueued */
-        volatile int last_dispatch;	/* age of the most recently dispatched buffer */
-	int ctxOwner;		/* last context to upload state */
-	int texAge;
-        int pf_enabled;		/* is pageflipping allowed? */
-        int pf_active;               
-        int pf_current_page;	/* which buffer is being displayed? */
-        int perf_boxes;	        /* performance boxes to be displayed */   
-	int width, height;      /* screen size in pixels */
-
-	drm_handle_t front_handle;
-	int front_offset;
-	int front_size;
-
-	drm_handle_t back_handle;
-	int back_offset;
-	int back_size;
-
-	drm_handle_t depth_handle;
-	int depth_offset;
-	int depth_size;
-
-	drm_handle_t tex_handle;
-	int tex_offset;
-	int tex_size;
-	int log_tex_granularity;
-	int pitch;
-	int rotation;           /* 0, 90, 180 or 270 */
-	int rotated_offset;
-	int rotated_size;
-	int rotated_pitch;
-	int virtualX, virtualY;
-
-	unsigned int front_tiled;
-        unsigned int back_tiled;
-        unsigned int depth_tiled;
-        unsigned int rotated_tiled;
-        unsigned int rotated2_tiled;
-} drmI830Sarea;
-
-/* Flags for perf_boxes
- */
-#define I830_BOX_RING_EMPTY    0x1 /* populated by kernel */
-#define I830_BOX_FLIP          0x2 /* populated by kernel */
-#define I830_BOX_WAIT          0x4 /* populated by kernel & client */
-#define I830_BOX_TEXTURE_LOAD  0x8 /* populated by kernel */
-#define I830_BOX_LOST_CONTEXT  0x10 /* populated by client */
-
-
-typedef struct {
-   	int start;		/* agp offset */
-	int used;		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830BatchBuffer;
-
-typedef struct {
-   	char *buf;		/* agp offset */
-	int sz; 		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830CmdBuffer;
- 
-typedef struct {
-	int *irq_seq;
-} drmI830IrqEmit;
-
-typedef struct {
-	int irq_seq;
-} drmI830IrqWait;
-
-typedef struct {
-	int param;
-	int *value;
-} drmI830GetParam;
-
-#define I830_PARAM_IRQ_ACTIVE     1
-#define I830_PARAM_ALLOW_BATCHBUFFER   2 
-
-typedef struct {
-	int param;
-	int value;
-} drmI830SetParam;
-
-#define I830_SETPARAM_USE_MI_BATCHBUFFER_START  1
-#define I830_SETPARAM_TEX_LRU_LOG_GRANULARITY   2
-#define I830_SETPARAM_ALLOW_BATCHBUFFER         3
-
-
-/* A memory manager for regions of shared memory:
- */
-#define I830_MEM_REGION_AGP 1
-
-typedef struct {
-	int region;
-	int alignment;
-	int size;
-	int *region_offset;	/* offset from start of fb or agp */
-} drmI830MemAlloc;
-
-typedef struct {
-	int region;
-	int region_offset;
-} drmI830MemFree;
-
-typedef struct {
-	int region;
-	int size;
-	int start;	
-} drmI830MemInitHeap;
-
-typedef struct {
-	int region;
-} drmI830MemDestroyHeap;
-
-#define MMIO_READ  0
-#define MMIO_WRITE 1
-
-#define MMIO_REGS_IA_PRIMATIVES_COUNT           0
-#define MMIO_REGS_IA_VERTICES_COUNT             1
-#define MMIO_REGS_VS_INVOCATION_COUNT           2
-#define MMIO_REGS_GS_PRIMITIVES_COUNT           3
-#define MMIO_REGS_GS_INVOCATION_COUNT           4
-#define MMIO_REGS_CL_PRIMITIVES_COUNT           5
-#define MMIO_REGS_CL_INVOCATION_COUNT           6
-#define MMIO_REGS_PS_INVOCATION_COUNT           7
-#define MMIO_REGS_PS_DEPTH_COUNT                8
-
-typedef struct {
-        unsigned int read_write:1;
-        unsigned int reg:31;
-        void __user *data;
-} drmI830MMIO;
-
-#endif /* _I830_DRM_H_ */
diff --git a/src/mesa/drivers/dri/i965/server/intel_dri.c b/src/mesa/drivers/dri/i965/server/intel_dri.c
index 169fdbece30..effdd26448a 100644..120000
--- a/src/mesa/drivers/dri/i965/server/intel_dri.c
+++ b/src/mesa/drivers/dri/i965/server/intel_dri.c
@@ -1,1282 +1 @@
-/**
- * \file server/intel_dri.c
- * \brief File to perform the device-specific initialization tasks typically
- * done in the X server.
- *
- * Here they are converted to run in the client (or perhaps a standalone
- * process), and to work with the frame buffer device rather than the X
- * server infrastructure.
- * 
- * Copyright (C) 2006 Dave Airlie (airlied@linux.ie)
-
- Permission is hereby granted, free of charge, to any person obtaining a
- copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sub license, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial portions
- of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR
- ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-*/
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <errno.h>
-#include <unistd.h>
-
-#include "driver.h"
-#include "drm.h"
-
-#include "intel.h"
-#include "i830_dri.h"
-
-#include "memops.h"
-#include "pciaccess.h"
-
-static size_t drm_page_size;
-static int nextTile = 0;
-#define xf86DrvMsg(...) do {} while(0)
-
-static const int pitches[] = {
-  128 * 8,
-  128 * 16,
-  128 * 32,
-  128 * 64,
-  0
-};
-
-static Bool I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea);
-
-static unsigned long
-GetBestTileAlignment(unsigned long size)
-{
-   unsigned long i;
-
-   for (i = KB(512); i < size; i <<= 1)
-      ;
-
-   if (i > MB(64))
-      i = MB(64);
-
-   return i;
-}
-
-static void SetFenceRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-  for (i = 0; i < 8; i++) {
-    OUTREG(FENCE + i * 4, pI830->Fence[i]);
-    //    if (I810_DEBUG & DEBUG_VERBOSE_VGA)
-    fprintf(stderr,"Fence Register : %x\n", pI830->Fence[i]);
-  }
-}
-
-/* Tiled memory is good... really, really good...
- *
- * Need to make it less likely that we miss out on this - probably
- * need to move the frontbuffer away from the 'guarenteed' alignment
- * of the first memory segment, or perhaps allocate a discontigous
- * framebuffer to get more alignment 'sweet spots'.
- */
-static void
-SetFence(const DRIDriverContext *ctx, I830Rec *pI830,
-	 int nr, unsigned int start, unsigned int pitch,
-         unsigned int size)
-{
-   unsigned int val;
-   unsigned int fence_mask = 0;
-   unsigned int fence_pitch;
-
-   if (nr < 0 || nr > 7) {
-      fprintf(stderr,
-		 "SetFence: fence %d out of range\n",nr);
-      return;
-   }
-
-   pI830->Fence[nr] = 0;
-
-   if (IS_I9XX(pI830))
-   	fence_mask = ~I915G_FENCE_START_MASK;
-   else
-   	fence_mask = ~I830_FENCE_START_MASK;
-
-   if (start & fence_mask) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not %s aligned\n",
-		 nr, start, (IS_I9XX(pI830)) ? "1MB" : "512k");
-      return;
-   }
-
-   if (start % size) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not size (%dk) aligned\n",
-		 nr, start, size / 1024);
-      return;
-   }
-
-   if (pitch & 127) {
-      fprintf(stderr,
-		 "SetFence: %d: pitch (%d) not a multiple of 128 bytes\n",
-		 nr, pitch);
-      return;
-   }
-
-   val = (start | FENCE_X_MAJOR | FENCE_VALID);
-
-   if (IS_I9XX(pI830)) {
-   	switch (size) {
-	   case MB(1):
-      		val |= I915G_FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= I915G_FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= I915G_FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= I915G_FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= I915G_FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= I915G_FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= I915G_FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-    } else {
-   	switch (size) {
-	   case KB(512):
-      		val |= FENCE_SIZE_512K;
-      		break;
-	   case MB(1):
-      		val |= FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-   }
-
-   if (IS_I9XX(pI830))
-	fence_pitch = pitch / 512;
-   else
-	fence_pitch = pitch / 128;
-
-   switch (fence_pitch) {
-   case 1:
-      val |= FENCE_PITCH_1;
-      break;
-   case 2:
-      val |= FENCE_PITCH_2;
-      break;
-   case 4:
-      val |= FENCE_PITCH_4;
-      break;
-   case 8:
-      val |= FENCE_PITCH_8;
-      break;
-   case 16:
-      val |= FENCE_PITCH_16;
-      break;
-   case 32:
-      val |= FENCE_PITCH_32;
-      break;
-   case 64:
-      val |= FENCE_PITCH_64;
-      break;
-   default:
-      fprintf(stderr,
-		 "SetFence: %d: illegal pitch (%d)\n", nr, pitch);
-      return;
-   }
-
-   pI830->Fence[nr] = val;
-}
-
-static Bool
-MakeTiles(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *pMem)
-{
-   int pitch, ntiles, i;
-
-   pitch = pMem->Pitch * ctx->cpp;
-   /*
-    * Simply try to break the region up into at most four pieces of size
-    * equal to the alignment.
-    */
-   ntiles = ROUND_TO(pMem->Size, pMem->Alignment) / pMem->Alignment;
-   if (ntiles >= 4) {
-      return FALSE;
-   }
-
-   for (i = 0; i < ntiles; i++, nextTile++) {
-     SetFence(ctx, pI830, nextTile, pMem->Start + i * pMem->Alignment,
-	       pitch, pMem->Alignment);
-   }
-   return TRUE;
-}
-
-static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-
-  /* Clear out */
-  for (i = 0; i < 8; i++)
-    pI830->Fence[i] = 0;
-  
-  nextTile = 0;
-
-  if (pI830->BackBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->BackBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the back buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the back buffer.\n");
-      pI830->allowPageFlip = FALSE;
-    }
-  }
-  
-  if (pI830->DepthBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->DepthBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the depth buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the depth buffer.\n");
-    }
-  }
-
-  return;
-}
-
-static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  struct pci_device host_bridge;
-  uint32_t gmch_ctrl;
-  int memsize = 0;
-  int range;
-
-  memset(&host_bridge, 0, sizeof(host_bridge));
-
-  pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
-  
-  /* We need to reduce the stolen size, by the GTT and the popup.
-   * The GTT varying according the the FbMapSize and the popup is 4KB */
-  range = (ctx->shared.fbSize / (1024*1024)) + 4;
-
-   if (IS_I85X(pI830) || IS_I865G(pI830) || IS_I9XX(pI830)) {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I855_GMCH_GMS_STOLEN_1M:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_4M:
-	 memsize = MB(4) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_8M:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_16M:
-	 memsize = MB(16) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_32M:
-	 memsize = MB(32) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_48M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(48) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_64M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(64) - KB(range);
-	 break;
-      }
-   } else {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I830_GMCH_GMS_STOLEN_512:
-	 memsize = KB(512) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_1024:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_8192:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I830_GMCH_GMS_LOCAL:
-	 memsize = 0;
-	 xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
-		    "Local memory found, but won't be used.\n");
-	 break;
-      }
-   }
-   if (memsize > 0) {
-     fprintf(stderr,
-		 "detected %d kB stolen memory.\n", memsize / 1024);
-   } else {
-     fprintf(stderr,
-		 "no video memory detected.\n");
-   }
-   return memsize;
-}
-
-static int AgpInit(const DRIDriverContext *ctx, I830Rec *info)
-{
-  unsigned long mode = 0x4;
-
-  if (drmAgpAcquire(ctx->drmFD) < 0) {
-    fprintf(stderr, "[gart] AGP not available\n");
-    return 0;
-  }
-  
-  if (drmAgpEnable(ctx->drmFD, mode) < 0) {
-    fprintf(stderr, "[gart] AGP not enabled\n");
-    drmAgpRelease(ctx->drmFD);
-    return 0;
-  }
-  else
-    fprintf(stderr, "[gart] AGP enabled at %dx\n", ctx->agpmode);
-
-  return 1;
-}
-
-/*
- * Allocate memory from the given pool.  Grow the pool if needed and if
- * possible.
- */
-static unsigned long
-AllocFromPool(const DRIDriverContext *ctx, I830Rec *pI830, 
-	      I830MemRange *result, I830MemPool *pool,
-	      long size, unsigned long alignment, int flags)
-{
-   long needed, start, end;
-
-   if (!result || !pool || !size)
-      return 0;
-
-   /* Calculate how much space is needed. */
-   if (alignment <= GTT_PAGE_SIZE)
-      needed = size;
-   else {
-	 start = ROUND_TO(pool->Free.Start, alignment);
-	 end = ROUND_TO(start + size, alignment);
-	 needed = end - pool->Free.Start;
-   }
-   if (needed > pool->Free.Size) {
-     return 0;
-   }
-
-   result->Start = ROUND_TO(pool->Free.Start, alignment);
-   pool->Free.Start += needed;
-   result->End = pool->Free.Start;
-
-   pool->Free.Size = pool->Free.End - pool->Free.Start;
-   result->Size = result->End - result->Start;
-   result->Pool = pool;
-   result->Alignment = alignment;
-   return needed;
-}
-
-static unsigned long AllocFromAGP(const DRIDriverContext *ctx, I830Rec *pI830, long size, unsigned long alignment, I830MemRange  *result)
-{
-   unsigned long start, end;
-   unsigned long newApStart, newApEnd;
-   int ret;
-   if (!result || !size)
-      return 0;
-   
-   if (!alignment)
-     alignment = 4;
-
-   start = ROUND_TO(pI830->MemoryAperture.Start, alignment);
-   end = ROUND_TO(start + size, alignment);
-   newApStart = end;
-   newApEnd = pI830->MemoryAperture.End;
-
-   ret=drmAgpAlloc(ctx->drmFD, size, 0, &(result->Physical), (drm_handle_t *)&(result->Key));
-   
-   if (ret)
-   {
-     fprintf(stderr,"drmAgpAlloc failed %d\n", ret);
-     return 0;
-   }
-   pI830->allocatedMemory += size;
-   pI830->MemoryAperture.Start = newApStart;
-   pI830->MemoryAperture.End = newApEnd;
-   pI830->MemoryAperture.Size = newApEnd - newApStart;
-   //   pI830->FreeMemory -= size;
-   result->Start = start;
-   result->End = start + size;
-   result->Size = size;
-   result->Offset = start;
-   result->Alignment = alignment;
-   result->Pool = NULL;
-  
-   return size;
-}
-
-unsigned long
-I830AllocVidMem(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *result, I830MemPool *pool, long size, unsigned long alignment, int flags)
-{
-  int ret;
-
-  if (!result)
-    return 0;
-
-   /* Make sure these are initialised. */
-   result->Size = 0;
-   result->Key = -1;
-
-   if (!size) {
-      return 0;
-   }
-
-   if (pool->Free.Size < size)
-     return AllocFromAGP(ctx, pI830, size, alignment, result);
-   else
-   {
-     ret = AllocFromPool(ctx, pI830, result, pool, size, alignment, flags);
-
-     if (ret==0)
-       return AllocFromAGP(ctx, pI830, size, alignment, result);
-     return ret;
-   }
-}
-
-static Bool BindAgpRange(const DRIDriverContext *ctx, I830MemRange *mem)
-{
-  if (!mem)
-    return FALSE;
-  
-  if (mem->Key == -1)
-    return TRUE;
-
-  return !drmAgpBind(ctx->drmFD, mem->Key, mem->Offset);
-}
-
-/* simple memory allocation routines needed */
-/* put ring buffer in low memory */
-/* need to allocate front, back, depth buffers aligned correctly,
-   allocate ring buffer, 
-*/
-
-/* */
-static Bool
-I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned long size, ret;
-  unsigned long lines, lineSize, align;
-
-  /* allocate ring buffer */
-  memset(pI830->LpRing, 0, sizeof(I830RingBuffer));
-  pI830->LpRing->mem.Key = -1;
-
-  size = PRIMARY_RINGBUFFER_SIZE;
-  
-  ret = I830AllocVidMem(ctx, pI830, &pI830->LpRing->mem, &pI830->StolenPool, size, 0x1000, 0);
-  
-  if (ret != size)
-  {
-    fprintf(stderr,"unable to allocate ring buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  pI830->LpRing->tail_mask = pI830->LpRing->mem.Size - 1;
-
-  
-  /* allocate front buffer */
-  memset(&(pI830->FrontBuffer), 0, sizeof(pI830->FrontBuffer));
-  pI830->FrontBuffer.Key = -1;
-  pI830->FrontBuffer.Pitch = ctx->shared.virtualWidth;
-
-  align = KB(512);  
-
-  lineSize = ctx->shared.virtualWidth * ctx->cpp;
-  lines = (ctx->shared.virtualHeight + 15) / 16 * 16;
-  size = lineSize * lines;
-  size = ROUND_TO_PAGE(size);
-
-  align = GetBestTileAlignment(size);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->FrontBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate front buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->BackBuffer), 0, sizeof(pI830->BackBuffer));
-  pI830->BackBuffer.Key = -1;
-  pI830->BackBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->BackBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate back buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->DepthBuffer), 0, sizeof(pI830->DepthBuffer));
-  pI830->DepthBuffer.Key = -1;
-  pI830->DepthBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->DepthBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate depth buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->ContextMem), 0, sizeof(pI830->ContextMem));
-  pI830->ContextMem.Key = -1;
-  size = KB(32);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->ContextMem, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
-  pI830->TexMem.Key = -1;
-
-  size = 32768 * 1024;
-  ret = AllocFromAGP(ctx, pI830, size, align, &pI830->TexMem);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
-    return FALSE;
-  }
-
-  return TRUE;
-}
-
-static Bool
-I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  if (!BindAgpRange(ctx, &pI830->LpRing->mem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->FrontBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->BackBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->DepthBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->ContextMem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->TexMem))
-    return FALSE;
-
-  return TRUE;
-}
-
-static Bool
-I830CleanupDma(const DRIDriverContext *ctx)
-{
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_CLEANUP_DMA;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-     fprintf(stderr, "I830 Dma Cleanup Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830InitDma(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-   I830RingBuffer *ring = pI830->LpRing;
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_INIT_DMA;
-
-   info.ring_start = ring->mem.Start + pI830->LinearAddr;
-   info.ring_end = ring->mem.End + pI830->LinearAddr;
-   info.ring_size = ring->mem.Size;
-
-   info.mmio_offset = (unsigned int)ctx->MMIOStart;
-
-   info.sarea_priv_offset = sizeof(drm_sarea_t);
-
-   info.front_offset = pI830->FrontBuffer.Start;
-   info.back_offset = pI830->BackBuffer.Start;
-   info.depth_offset = pI830->DepthBuffer.Start;
-   info.w = ctx->shared.virtualWidth;
-   info.h = ctx->shared.virtualHeight;
-   info.pitch = ctx->shared.virtualWidth;
-   info.back_pitch = pI830->BackBuffer.Pitch;
-   info.depth_pitch = pI830->DepthBuffer.Pitch;
-   info.cpp = ctx->cpp;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-      fprintf(stderr,
-		 "I830 Dma Initialization Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static int I830CheckDRMVersion( const DRIDriverContext *ctx,
-				  I830Rec *pI830 )
-{
-   drmVersionPtr  version;
-
-   version = drmGetVersion(ctx->drmFD);
-
-   if (version) {
-     int req_minor, req_patch;
-
-     req_minor = 4;
-     req_patch = 0;	
-
-     if (version->version_major != 1 ||
-	 version->version_minor < req_minor ||
-	 (version->version_minor == req_minor && 
-	  version->version_patchlevel < req_patch)) {
-       /* Incompatible drm version */
-       fprintf(stderr,
-	       "[dri] I830DRIScreenInit failed because of a version "
-	       "mismatch.\n"
-	       "[dri] i915.o kernel module version is %d.%d.%d "
-	       "but version 1.%d.%d or newer is needed.\n"
-	       "[dri] Disabling DRI.\n",
-	       version->version_major,
-	       version->version_minor,
-	       version->version_patchlevel,
-	       req_minor,
-	       req_patch);
-       drmFreeVersion(version);
-       return 0;
-     }
-     
-     pI830->drmMinor = version->version_minor;
-     drmFreeVersion(version);
-   }
-   return 1;
-}
-
-static void
-I830SetRingRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned int itemp;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-   OUTREG(LP_RING + RING_LEN, 0);
-   OUTREG(LP_RING + RING_TAIL, 0);
-   OUTREG(LP_RING + RING_HEAD, 0);
-
-   if ((long)(pI830->LpRing->mem.Start & I830_RING_START_MASK) !=
-       pI830->LpRing->mem.Start) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer start (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Start, I830_RING_START_MASK);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = pI830->LpRing->mem.Start & I830_RING_START_MASK;
-   OUTREG(LP_RING + RING_START, itemp);
-
-   if (((pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES) !=
-       pI830->LpRing->mem.Size - 4096) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer size - 4096 (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Size - 4096,
-		 I830_RING_NR_PAGES);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = (pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES;
-   itemp |= (RING_NO_REPORT | RING_VALID);
-   OUTREG(LP_RING + RING_LEN, itemp);
-
-   pI830->LpRing->head = INREG(LP_RING + RING_HEAD) & I830_HEAD_MASK;
-   pI830->LpRing->tail = INREG(LP_RING + RING_TAIL);
-   pI830->LpRing->space = pI830->LpRing->head - (pI830->LpRing->tail + 8);
-   if (pI830->LpRing->space < 0)
-      pI830->LpRing->space += pI830->LpRing->mem.Size;
-
-   SetFenceRegs(ctx, pI830);
-   
-   /* RESET THE DISPLAY PIPE TO POINT TO THE FRONTBUFFER - hacky
-      hacky hacky */
-   OUTREG(DSPABASE, pI830->FrontBuffer.Start + pI830->LinearAddr);
-
-}
-
-static Bool
-I830SetParam(const DRIDriverContext *ctx, int param, int value)
-{
-   drmI830SetParam sp;
-
-   memset(&sp, 0, sizeof(sp));
-   sp.param = param;
-   sp.value = value;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_SETPARAM, &sp, sizeof(sp))) {
-      fprintf(stderr, "I830 SetParam Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   fprintf(stderr,
-              "[drm] Mapping front buffer\n");
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->front_offset + pI830->LinearAddr),
-                 sarea->front_size,
-                 DRM_FRAME_BUFFER,  /*DRM_AGP,*/
-                 0,
-                 &sarea->front_handle) < 0) {
-     fprintf(stderr,
-	     "[drm] drmAddMap(front_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   ctx->shared.hFrameBuffer = sarea->front_handle;
-   ctx->shared.fbSize = sarea->front_size;
-   fprintf(stderr, "[drm] Front Buffer = 0x%08x\n",
-	   sarea->front_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->back_offset),
-                 sarea->back_size, DRM_AGP, 0,
-                 &sarea->back_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(back_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Back Buffer = 0x%08x\n",
-              sarea->back_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)sarea->depth_offset,
-                 sarea->depth_size, DRM_AGP, 0,
-                 &sarea->depth_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(depth_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
-              sarea->depth_handle);
-
-   if (drmAddMap(ctx->drmFD,
-		 (drm_handle_t)sarea->tex_offset,
-		 sarea->tex_size, DRM_AGP, 0,
-		 &sarea->tex_handle) < 0) {
-      fprintf(stderr,
-		 "[drm] drmAddMap(tex_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] textures = 0x%08x\n",
-	      sarea->tex_handle);
-
-   return TRUE;
-}
-
-
-static void
-I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-#if 1
-   if (sarea->front_handle) {
-      drmRmMap(ctx->drmFD, sarea->front_handle);
-      sarea->front_handle = 0;
-   }
-#endif
-   if (sarea->back_handle) {
-      drmRmMap(ctx->drmFD, sarea->back_handle);
-      sarea->back_handle = 0;
-   }
-   if (sarea->depth_handle) {
-      drmRmMap(ctx->drmFD, sarea->depth_handle);
-      sarea->depth_handle = 0;
-   }
-   if (sarea->tex_handle) {
-      drmRmMap(ctx->drmFD, sarea->tex_handle);
-      sarea->tex_handle = 0;
-   }
-}
-
-static void
-I830InitTextureHeap(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   /* Start up the simple memory manager for agp space */
-   drmI830MemInitHeap drmHeap;
-   drmHeap.region = I830_MEM_REGION_AGP;
-   drmHeap.start  = 0;
-   drmHeap.size   = sarea->tex_size;
-      
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT_HEAP,
-			  &drmHeap, sizeof(drmHeap))) {
-      fprintf(stderr,
-		    "[drm] Failed to initialized agp heap manager\n");
-   } else {
-      fprintf(stderr,
-		    "[drm] Initialized kernel agp heap manager, %d\n",
-		    sarea->tex_size);
-
-      I830SetParam(ctx, I830_SETPARAM_TEX_LRU_LOG_GRANULARITY, 
-		      sarea->log_tex_granularity);
-   }
-}
-
-static Bool
-I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  if (drmAddMap(ctx->drmFD,
-		(drm_handle_t)pI830->LpRing->mem.Start,
-		pI830->LpRing->mem.Size, DRM_AGP, 0,
-		&pI830->ring_map) < 0) {
-    fprintf(stderr,
-	    "[drm] drmAddMap(ring_map) failed. Disabling DRI\n");
-    return FALSE;
-  }
-  fprintf(stderr, "[drm] ring buffer = 0x%08x\n",
-	  pI830->ring_map);
-
-  if (I830InitDma(ctx, pI830) == FALSE) {
-    return FALSE;
-  }
-  
-   /* init to zero to be safe */
-
-  I830DRIMapScreenRegions(ctx, pI830, sarea);
-  I830InitTextureHeap(ctx, pI830, sarea);
-
-   if (ctx->pciDevice != PCI_CHIP_845_G &&
-       ctx->pciDevice != PCI_CHIP_I830_M) {
-      I830SetParam(ctx, I830_SETPARAM_USE_MI_BATCHBUFFER_START, 1 );
-   }
-
-   /* Okay now initialize the dma engine */
-   {
-      pI830->irq = drmGetInterruptFromBusID(ctx->drmFD,
-					    ctx->pciBus,
-					    ctx->pciDevice,
-					    ctx->pciFunc);
-
-      if (drmCtlInstHandler(ctx->drmFD, pI830->irq)) {
-	 fprintf(stderr,
-		    "[drm] failure adding irq handler\n");
-	 pI830->irq = 0;
-	 return FALSE;
-      }
-      else
-	 fprintf(stderr,
-		    "[drm] dma control initialized, using IRQ %d\n",
-		    pI830->irq);
-   }
-
-   fprintf(stderr, "[dri] visual configs initialized\n");
-
-   return TRUE;
-}
-
-static Bool
-I830ClearScreen(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  /* need to drmMap front and back buffers and zero them */
-  drmAddress map_addr;
-  int ret;
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->front_handle,
-	       sarea->front_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map front buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->front_size);
-  drmUnmap(map_addr, sarea->front_size);
-
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->back_handle,
-	       sarea->back_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map back buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->back_size);
-  drmUnmap(map_addr, sarea->back_size);
-
-  return TRUE;
-}
-
-static Bool
-I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
-		  
-{
-   I830DRIPtr pI830DRI;
-   drmI830Sarea *pSAREAPriv;
-   int err;
-      
-   drm_page_size = getpagesize();   
-
-   pI830->registerSize = ctx->MMIOSize;
-   /* This is a hack for now.  We have to have more than a 4k page here
-    * because of the size of the state.  However, the state should be
-    * in a per-context mapping.  This will be added in the Mesa 3.5 port
-    * of the I830 driver.
-    */
-   ctx->shared.SAREASize = SAREA_MAX;
-
-   /* Note that drmOpen will try to load the kernel module, if needed. */
-   ctx->drmFD = drmOpen("i915", NULL );
-   if (ctx->drmFD < 0) {
-      fprintf(stderr, "[drm] drmOpen failed\n");
-      return 0;
-   }
-
-   if ((err = drmSetBusid(ctx->drmFD, ctx->pciBusID)) < 0) {
-      fprintf(stderr, "[drm] drmSetBusid failed (%d, %s), %s\n",
-	      ctx->drmFD, ctx->pciBusID, strerror(-err));
-      return 0;
-   }
-
-   if (drmAddMap( ctx->drmFD,
-		  0,
-		  ctx->shared.SAREASize,
-		  DRM_SHM,
-		  DRM_CONTAINS_LOCK,
-		  &ctx->shared.hSAREA) < 0)
-   {
-     fprintf(stderr, "[drm] drmAddMap failed\n");
-     return 0;
-   }
-
-   fprintf(stderr, "[drm] added %d byte SAREA at 0x%08x\n",
-	   ctx->shared.SAREASize, ctx->shared.hSAREA);
-   
-   if (drmMap( ctx->drmFD,
-	       ctx->shared.hSAREA,
-	       ctx->shared.SAREASize,
-	       (drmAddressPtr)(&ctx->pSAREA)) < 0)
-   {
-      fprintf(stderr, "[drm] drmMap failed\n");
-      return 0;
-   
-   }
-   
-   memset(ctx->pSAREA, 0, ctx->shared.SAREASize);
-   fprintf(stderr, "[drm] mapped SAREA 0x%08x to %p, size %d\n",
-	   ctx->shared.hSAREA, ctx->pSAREA, ctx->shared.SAREASize);
-   
-
-   if (drmAddMap(ctx->drmFD, 
-		 ctx->MMIOStart,
-		 ctx->MMIOSize,
-		 DRM_REGISTERS, 
-		 DRM_READ_ONLY, 
-		 &pI830->registerHandle) < 0) {
-      fprintf(stderr, "[drm] drmAddMap mmio failed\n");	
-      return 0;
-   }
-   fprintf(stderr,
-	   "[drm] register handle = 0x%08x\n", pI830->registerHandle);
-
-
-   if (!I830CheckDRMVersion(ctx, pI830)) {
-     return FALSE;
-   }
-
-   /* Create a 'server' context so we can grab the lock for
-    * initialization ioctls.
-    */
-   if ((err = drmCreateContext(ctx->drmFD, &ctx->serverContext)) != 0) {
-      fprintf(stderr, "%s: drmCreateContext failed %d\n", __FUNCTION__, err);
-      return 0;
-   }
-
-   DRM_LOCK(ctx->drmFD, ctx->pSAREA, ctx->serverContext, 0); 
-
-   /* Initialize the SAREA private data structure */
-   pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				 sizeof(drm_sarea_t));
-   memset(pSAREAPriv, 0, sizeof(*pSAREAPriv));
-
-   pI830->StolenMemory.Size = I830DetectMemory(ctx, pI830);
-   pI830->StolenMemory.Start = 0;
-   pI830->StolenMemory.End = pI830->StolenMemory.Size;
-
-   pI830->MemoryAperture.Start = pI830->StolenMemory.End;
-   pI830->MemoryAperture.End = KB(40000);
-   pI830->MemoryAperture.Size = pI830->MemoryAperture.End - pI830->MemoryAperture.Start;
-
-   pI830->StolenPool.Fixed = pI830->StolenMemory;
-   pI830->StolenPool.Total = pI830->StolenMemory;
-   pI830->StolenPool.Free = pI830->StolenPool.Total;
-   pI830->FreeMemory = pI830->StolenPool.Total.Size;
-
-   if (!AgpInit(ctx, pI830))
-     return FALSE;
-
-   if (I830AllocateMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   if (I830BindMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
-   pSAREAPriv->front_size = pI830->FrontBuffer.Size;
-   pSAREAPriv->width = ctx->shared.virtualWidth;
-   pSAREAPriv->height = ctx->shared.virtualHeight;
-   pSAREAPriv->pitch = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualX = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualY = ctx->shared.virtualHeight;
-   pSAREAPriv->back_offset = pI830->BackBuffer.Start;
-   pSAREAPriv->back_size = pI830->BackBuffer.Size;
-   pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
-   pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
-   pSAREAPriv->tex_offset = pI830->TexMem.Start;
-   pSAREAPriv->tex_size = pI830->TexMem.Size;
-   pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
-
-   ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
-   ctx->driverClientMsgSize = sizeof(I830DRIRec);
-   pI830DRI = (I830DRIPtr)ctx->driverClientMsg;
-   pI830DRI->deviceID = pI830->Chipset;
-   pI830DRI->regsSize = I830_REG_SIZE;
-   pI830DRI->width = ctx->shared.virtualWidth;
-   pI830DRI->height = ctx->shared.virtualHeight;
-   pI830DRI->mem = ctx->shared.fbSize;
-   pI830DRI->cpp = ctx->cpp;
-   pI830DRI->backOffset = pI830->BackBuffer.Start;
-   pI830DRI->backPitch = pI830->BackBuffer.Pitch; 
-
-   pI830DRI->depthOffset = pI830->DepthBuffer.Start;
-   pI830DRI->depthPitch = pI830->DepthBuffer.Pitch; 
-
-   pI830DRI->fbOffset = pI830->FrontBuffer.Start;
-   pI830DRI->fbStride = pI830->FrontBuffer.Pitch;
-
-   pI830DRI->bitsPerPixel = ctx->bpp;
-   pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
-   
-   err = I830DRIDoMappings(ctx, pI830, pSAREAPriv);
-   if (err == FALSE)
-       return FALSE;
-
-   I830SetupMemoryTiling(ctx, pI830);
-
-   /* Quick hack to clear the front & back buffers.  Could also use
-    * the clear ioctl to do this, but would need to setup hw state
-    * first.
-    */
-   I830ClearScreen(ctx, pI830, pSAREAPriv);
-
-   I830SetRingRegs(ctx, pI830);
-
-   return TRUE;
-}
-
-
-/**
- * \brief Validate the fbdev mode.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Saves some registers and returns 1.
- *
- * \sa radeonValidateMode().
- */
-static int i830ValidateMode( const DRIDriverContext *ctx )
-{
-  return 1;
-}
-
-/**
- * \brief Examine mode returned by fbdev.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Restores registers that fbdev has clobbered and returns 1.
- *
- * \sa i810ValidateMode().
- */
-static int i830PostValidateMode( const DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = ctx->driverPrivate;
-
-  I830SetRingRegs(ctx, pI830);
-  return 1;
-}
-
-
-/**
- * \brief Initialize the framebuffer device mode
- *
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Fills in \p info with some default values and some information from \p ctx
- * and then calls I810ScreenInit() for the screen initialization.
- * 
- * Before exiting clears the framebuffer memory accessing it directly.
- */
-static int i830InitFBDev( DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = calloc(1, sizeof(I830Rec));
-  int i;
-
-   {
-      int  dummy = ctx->shared.virtualWidth;
-
-      switch (ctx->bpp / 8) {
-      case 1: dummy = (ctx->shared.virtualWidth + 127) & ~127; break;
-      case 2: dummy = (ctx->shared.virtualWidth +  31) &  ~31; break;
-      case 3:
-      case 4: dummy = (ctx->shared.virtualWidth +  15) &  ~15; break;
-      }
-
-      ctx->shared.virtualWidth = dummy;
-      ctx->shared.Width = ctx->shared.virtualWidth;
-   }
-
-
-   for (i = 0; pitches[i] != 0; i++) {
-     if (pitches[i] >= ctx->shared.virtualWidth) {
-       ctx->shared.virtualWidth = pitches[i];
-       break;
-     }
-   }
-
-   ctx->driverPrivate = (void *)pI830;
-   
-   pI830->LpRing = calloc(1, sizeof(I830RingBuffer));
-   pI830->Chipset = ctx->chipset;
-   pI830->LinearAddr = ctx->FBStart;
-
-   if (!I830ScreenInit( ctx, pI830 ))
-      return 0;
-
-   
-   return 1;
-}
-
-
-/**
- * \brief The screen is being closed, so clean up any state and free any
- * resources used by the DRI.
- *
- * \param ctx display handle.
- *
- * Unmaps the SAREA, closes the DRM device file descriptor and frees the driver
- * private data.
- */
-static void i830HaltFBDev( DRIDriverContext *ctx )
-{
-  drmI830Sarea *pSAREAPriv;
-  I830Rec *pI830 = ctx->driverPrivate;
-
-   if (pI830->irq) {
-       drmCtlUninstHandler(ctx->drmFD);
-       pI830->irq = 0;   }
-
-   I830CleanupDma(ctx);
-
-  pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				sizeof(drm_sarea_t));
-
-  I830DRIUnmapScreenRegions(ctx, pI830, pSAREAPriv);
-  drmUnmap( ctx->pSAREA, ctx->shared.SAREASize );
-  drmClose(ctx->drmFD);
-  
-  if (ctx->driverPrivate) {
-    free(ctx->driverPrivate);
-    ctx->driverPrivate = 0;
-  }
-}
-
-
-extern void i810NotifyFocus( int );
-
-/**
- * \brief Exported driver interface for Mini GLX.
- *
- * \sa DRIDriverRec.
- */
-const struct DRIDriverRec __driDriver = {
-   i830ValidateMode,
-   i830PostValidateMode,
-   i830InitFBDev,
-   i830HaltFBDev,
-   NULL,//I830EngineShutdown,
-   NULL, //I830EngineRestore,  
-#ifndef _EMBEDDED
-   0,
-#else
-   i810NotifyFocus, 
-#endif
-};
+../../intel/server/intel_dri.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
new file mode 100644
index 00000000000..9d9937289a8
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
@@ -0,0 +1,301 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_decode.h"
+#include "intel_reg.h"
+#include "intel_bufmgr.h"
+#include "intel_buffers.h"
+
+/* Relocations in kernel space:
+ *    - pass dma buffer seperately
+ *    - memory manager knows how to patch
+ *    - pass list of dependent buffers
+ *    - pass relocation list
+ *
+ * Either:
+ *    - get back an offset for buffer to fire
+ *    - memory manager knows how to fire buffer
+ *
+ * Really want the buffer to be AGP and pinned.
+ *
+ */
+
+/* Cliprect fence: The highest fence protecting a dma buffer
+ * containing explicit cliprect information.  Like the old drawable
+ * lock but irq-driven.  X server must wait for this fence to expire
+ * before changing cliprects [and then doing sw rendering?].  For
+ * other dma buffers, the scheduler will grab current cliprect info
+ * and mix into buffer.  X server must hold the lock while changing
+ * cliprects???  Make per-drawable.  Need cliprects in shared memory
+ * -- beats storing them with every cmd buffer in the queue.
+ *
+ * ==> X server must wait for this fence to expire before touching the
+ * framebuffer with new cliprects.
+ *
+ * ==> Cliprect-dependent buffers associated with a
+ * cliprect-timestamp.  All of the buffers associated with a timestamp
+ * must go to hardware before any buffer with a newer timestamp.
+ *
+ * ==> Dma should be queued per-drawable for correct X/GL
+ * synchronization.  Or can fences be used for this?
+ *
+ * Applies to: Blit operations, metaops, X server operations -- X
+ * server automatically waits on its own dma to complete before
+ * modifying cliprects ???
+ */
+
+void
+intel_batchbuffer_reset(struct intel_batchbuffer *batch)
+{
+   struct intel_context *intel = batch->intel;
+
+   if (batch->buf != NULL) {
+      dri_bo_unreference(batch->buf);
+      batch->buf = NULL;
+   }
+
+   if (!batch->buffer && intel->ttm == GL_TRUE)
+      batch->buffer = malloc (intel->maxBatchSize);
+
+   batch->buf = dri_bo_alloc(intel->bufmgr, "batchbuffer",
+			     intel->maxBatchSize, 4096);
+   if (batch->buffer)
+      batch->map = batch->buffer;
+   else {
+      dri_bo_map(batch->buf, GL_TRUE);
+      batch->map = batch->buf->virtual;
+   }
+   batch->size = intel->maxBatchSize;
+   batch->ptr = batch->map;
+   batch->dirty_state = ~0;
+   batch->cliprect_mode = IGNORE_CLIPRECTS;
+}
+
+struct intel_batchbuffer *
+intel_batchbuffer_alloc(struct intel_context *intel)
+{
+   struct intel_batchbuffer *batch = calloc(sizeof(*batch), 1);
+
+   batch->intel = intel;
+   intel_batchbuffer_reset(batch);
+
+   return batch;
+}
+
+void
+intel_batchbuffer_free(struct intel_batchbuffer *batch)
+{
+   if (batch->buffer)
+      free (batch->buffer);
+   else {
+      if (batch->map) {
+	 dri_bo_unmap(batch->buf);
+	 batch->map = NULL;
+      }
+   }
+   dri_bo_unreference(batch->buf);
+   batch->buf = NULL;
+   free(batch);
+}
+
+
+
+/* TODO: Push this whole function into bufmgr.
+ */
+static void
+do_flush_locked(struct intel_batchbuffer *batch,
+		GLuint used, GLboolean allow_unlock)
+{
+   struct intel_context *intel = batch->intel;
+   int ret = 0;
+   unsigned int num_cliprects = 0;
+   struct drm_clip_rect *cliprects = NULL;
+   int x_off = 0, y_off = 0;
+
+   if (batch->buffer)
+      dri_bo_subdata (batch->buf, 0, used, batch->buffer);
+   else
+      dri_bo_unmap(batch->buf);
+
+   batch->map = NULL;
+   batch->ptr = NULL;
+
+
+   if (batch->cliprect_mode == LOOP_CLIPRECTS) {
+      intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   }
+   /* Dispatch the batchbuffer, if it has some effect (nonzero cliprects).
+    * Can't short-circuit like this once we have hardware contexts, but we
+    * should always be in DRI2 mode by then anyway.
+    */
+   if ((batch->cliprect_mode != LOOP_CLIPRECTS ||
+	num_cliprects != 0) && !intel->no_hw) {
+      dri_bo_exec(batch->buf, used, cliprects, num_cliprects,
+		  (x_off & 0xffff) | (y_off << 16));
+   }
+
+   if (batch->cliprect_mode == LOOP_CLIPRECTS && num_cliprects == 0) {
+      if (allow_unlock) {
+	 /* If we are not doing any actual user-visible rendering,
+	  * do a sched_yield to keep the app from pegging the cpu while
+	  * achieving nothing.
+	  */
+         UNLOCK_HARDWARE(intel);
+         sched_yield();
+         LOCK_HARDWARE(intel);
+      }
+   }
+
+   if (INTEL_DEBUG & DEBUG_BATCH) {
+      dri_bo_map(batch->buf, GL_FALSE);
+      intel_decode(batch->buf->virtual, used / 4, batch->buf->offset,
+		   intel->intelScreen->deviceID);
+      dri_bo_unmap(batch->buf);
+
+      if (intel->vtbl.debug_batch != NULL)
+	 intel->vtbl.debug_batch(intel);
+   }
+
+   if (ret != 0) {
+      UNLOCK_HARDWARE(intel);
+      exit(1);
+   }
+   intel->vtbl.new_batch(intel);
+}
+
+void
+_intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
+			 int line)
+{
+   struct intel_context *intel = batch->intel;
+   GLuint used = batch->ptr - batch->map;
+   GLboolean was_locked = intel->locked;
+
+   if (used == 0) {
+      batch->cliprect_mode = IGNORE_CLIPRECTS;
+      return;
+   }
+
+   if (INTEL_DEBUG & DEBUG_BATCH)
+      fprintf(stderr, "%s:%d: Batchbuffer flush with %db used\n", file, line,
+	      used);
+
+   /* Emit a flush if the bufmgr doesn't do it for us. */
+   if (!intel->ttm) {
+      *(GLuint *) (batch->ptr) = intel->vtbl.flush_cmd();
+      batch->ptr += 4;
+      used = batch->ptr - batch->map;
+   }
+
+   /* Round batchbuffer usage to 2 DWORDs. */
+
+   if ((used & 4) == 0) {
+      *(GLuint *) (batch->ptr) = 0; /* noop */
+      batch->ptr += 4;
+      used = batch->ptr - batch->map;
+   }
+
+   /* Mark the end of the buffer. */
+   *(GLuint *) (batch->ptr) = MI_BATCH_BUFFER_END; /* noop */
+   batch->ptr += 4;
+   used = batch->ptr - batch->map;
+
+   /* Workaround for recursive batchbuffer flushing: If the window is
+    * moved, we can get into a case where we try to flush during a
+    * flush.  What happens is that when we try to grab the lock for
+    * the first flush, we detect that the window moved which then
+    * causes another flush (from the intel_draw_buffer() call in
+    * intelUpdatePageFlipping()).  To work around this we reset the
+    * batchbuffer tail pointer before trying to get the lock.  This
+    * prevent the nested buffer flush, but a better fix would be to
+    * avoid that in the first place. */
+   batch->ptr = batch->map;
+
+   if (intel->vtbl.finish_batch)
+      intel->vtbl.finish_batch(intel);
+
+   /* TODO: Just pass the relocation list and dma buffer up to the
+    * kernel.
+    */
+   if (!was_locked)
+      LOCK_HARDWARE(intel);
+
+   do_flush_locked(batch, used, GL_FALSE);
+
+   if (!was_locked)
+      UNLOCK_HARDWARE(intel);
+
+   if (INTEL_DEBUG & DEBUG_SYNC) {
+      fprintf(stderr, "waiting for idle\n");
+      dri_bo_map(batch->buf, GL_TRUE);
+      dri_bo_unmap(batch->buf);
+   }
+
+   /* Reset the buffer:
+    */
+   intel_batchbuffer_reset(batch);
+}
+
+
+/*  This is the only way buffers get added to the validate list.
+ */
+GLboolean
+intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
+                             dri_bo *buffer,
+                             uint32_t read_domains, uint32_t write_domain,
+			     uint32_t delta)
+{
+   int ret;
+
+   if (batch->ptr - batch->map > batch->buf->size)
+    _mesa_printf ("bad relocation ptr %p map %p offset %d size %d\n",
+		  batch->ptr, batch->map, batch->ptr - batch->map, batch->buf->size);
+   ret = dri_bo_emit_reloc(batch->buf, read_domains, write_domain,
+			   delta, batch->ptr - batch->map, buffer);
+
+   /*
+    * Using the old buffer offset, write in what the right data would be, in case
+    * the buffer doesn't move and we can short-circuit the relocation processing
+    * in the kernel
+    */
+   intel_batchbuffer_emit_dword (batch, buffer->offset + delta);
+
+   return GL_TRUE;
+}
+
+void
+intel_batchbuffer_data(struct intel_batchbuffer *batch,
+                       const void *data, GLuint bytes,
+		       enum cliprect_mode cliprect_mode)
+{
+   assert((bytes & 3) == 0);
+   intel_batchbuffer_require_space(batch, bytes, cliprect_mode);
+   __memcpy(batch->ptr, data, bytes);
+   batch->ptr += bytes;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
new file mode 100644
index 00000000000..51579df09e7
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -0,0 +1,184 @@
+#ifndef INTEL_BATCHBUFFER_H
+#define INTEL_BATCHBUFFER_H
+
+#include "main/mtypes.h"
+
+#include "intel_context.h"
+#include "intel_bufmgr.h"
+#include "intel_reg.h"
+
+#define BATCH_SZ 16384
+#define BATCH_RESERVED 16
+
+enum cliprect_mode {
+   /**
+    * Batchbuffer contents may be looped over per cliprect, but do not
+    * require it.
+    */
+   IGNORE_CLIPRECTS,
+   /**
+    * Batchbuffer contents require looping over per cliprect at batch submit
+    * time.
+    *
+    * This will be upgraded to NO_LOOP_CLIPRECTS when there's a single
+    * constant cliprect, as in DRI2 or FBO rendering.
+    */
+   LOOP_CLIPRECTS,
+   /**
+    * Batchbuffer contents contain drawing that should not be executed multiple
+    * times.
+    */
+   NO_LOOP_CLIPRECTS,
+   /**
+    * Batchbuffer contents contain drawing that already handles cliprects, such
+    * as 2D drawing to front/back/depth that doesn't respect DRAWING_RECTANGLE.
+    *
+    * Equivalent behavior to NO_LOOP_CLIPRECTS, but may not persist in batch
+    * outside of LOCK/UNLOCK.  This is upgraded to just NO_LOOP_CLIPRECTS when
+    * there's a constant cliprect, as in DRI2 or FBO rendering.
+    */
+   REFERENCES_CLIPRECTS
+};
+
+struct intel_batchbuffer
+{
+   struct intel_context *intel;
+
+   dri_bo *buf;
+
+   GLubyte *buffer;
+
+   GLubyte *map;
+   GLubyte *ptr;
+
+   enum cliprect_mode cliprect_mode;
+
+   GLuint size;
+
+   /** Tracking of BEGIN_BATCH()/OUT_BATCH()/ADVANCE_BATCH() debugging */
+   struct {
+      GLuint total;
+      GLubyte *start_ptr;
+   } emit;
+
+   GLuint dirty_state;
+};
+
+struct intel_batchbuffer *intel_batchbuffer_alloc(struct intel_context
+                                                  *intel);
+
+void intel_batchbuffer_free(struct intel_batchbuffer *batch);
+
+
+void _intel_batchbuffer_flush(struct intel_batchbuffer *batch,
+			      const char *file, int line);
+
+#define intel_batchbuffer_flush(batch) \
+	_intel_batchbuffer_flush(batch, __FILE__, __LINE__)
+
+void intel_batchbuffer_reset(struct intel_batchbuffer *batch);
+
+
+/* Unlike bmBufferData, this currently requires the buffer be mapped.
+ * Consider it a convenience function wrapping multple
+ * intel_buffer_dword() calls.
+ */
+void intel_batchbuffer_data(struct intel_batchbuffer *batch,
+                            const void *data, GLuint bytes,
+			    enum cliprect_mode cliprect_mode);
+
+void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
+                                     GLuint bytes);
+
+GLboolean intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
+                                       dri_bo *buffer,
+				       uint32_t read_domains,
+				       uint32_t write_domain,
+				       uint32_t offset);
+
+/* Inline functions - might actually be better off with these
+ * non-inlined.  Certainly better off switching all command packets to
+ * be passed as structs rather than dwords, but that's a little bit of
+ * work...
+ */
+static INLINE GLint
+intel_batchbuffer_space(struct intel_batchbuffer *batch)
+{
+   return (batch->size - BATCH_RESERVED) - (batch->ptr - batch->map);
+}
+
+
+static INLINE void
+intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch, GLuint dword)
+{
+   assert(batch->map);
+   assert(intel_batchbuffer_space(batch) >= 4);
+   *(GLuint *) (batch->ptr) = dword;
+   batch->ptr += 4;
+}
+
+static INLINE void
+intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
+                                GLuint sz,
+				enum cliprect_mode cliprect_mode)
+{
+   assert(sz < batch->size - 8);
+   if (intel_batchbuffer_space(batch) < sz)
+      intel_batchbuffer_flush(batch);
+
+   if ((cliprect_mode == LOOP_CLIPRECTS ||
+	cliprect_mode == REFERENCES_CLIPRECTS) &&
+       batch->intel->constant_cliprect)
+      cliprect_mode = NO_LOOP_CLIPRECTS;
+
+   if (cliprect_mode != IGNORE_CLIPRECTS) {
+      if (batch->cliprect_mode == IGNORE_CLIPRECTS) {
+	 batch->cliprect_mode = cliprect_mode;
+      } else {
+	 if (batch->cliprect_mode != cliprect_mode) {
+	    intel_batchbuffer_flush(batch);
+	    batch->cliprect_mode = cliprect_mode;
+	 }
+      }
+   }
+}
+
+/* Here are the crusty old macros, to be removed:
+ */
+#define BATCH_LOCALS
+
+#define BEGIN_BATCH(n, cliprect_mode) do {				\
+   intel_batchbuffer_require_space(intel->batch, (n)*4, cliprect_mode); \
+   assert(intel->batch->emit.start_ptr == NULL);			\
+   intel->batch->emit.total = (n) * 4;					\
+   intel->batch->emit.start_ptr = intel->batch->ptr;			\
+} while (0)
+
+#define OUT_BATCH(d) intel_batchbuffer_emit_dword(intel->batch, d)
+
+#define OUT_RELOC(buf, read_domains, write_domain, delta) do {		\
+   assert((delta) >= 0);						\
+   intel_batchbuffer_emit_reloc(intel->batch, buf,			\
+				read_domains, write_domain, delta);	\
+} while (0)
+
+#define ADVANCE_BATCH() do {						\
+   unsigned int _n = intel->batch->ptr - intel->batch->emit.start_ptr;	\
+   assert(intel->batch->emit.start_ptr != NULL);			\
+   if (_n != intel->batch->emit.total) {				\
+      fprintf(stderr, "ADVANCE_BATCH: %d of %d dwords emitted\n",	\
+	      _n, intel->batch->emit.total);				\
+      abort();								\
+   }									\
+   intel->batch->emit.start_ptr = NULL;					\
+} while(0)
+
+
+static INLINE void
+intel_batchbuffer_emit_mi_flush(struct intel_batchbuffer *batch)
+{
+   intel_batchbuffer_require_space(batch, 4, IGNORE_CLIPRECTS);
+   intel_batchbuffer_emit_dword(batch, MI_FLUSH);
+}
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c
new file mode 100644
index 00000000000..2f1639d4a4c
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_blit.c
@@ -0,0 +1,649 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include <stdio.h>
+#include <errno.h>
+
+#include "main/mtypes.h"
+#include "main/context.h"
+#include "main/enums.h"
+
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_context.h"
+#include "intel_fbo.h"
+#include "intel_reg.h"
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
+#include "intel_chipset.h"
+
+#define FILE_DEBUG_FLAG DEBUG_BLIT
+
+/**
+ * Copy the back color buffer to the front color buffer. 
+ * Used for SwapBuffers().
+ */
+void
+intelCopyBuffer(const __DRIdrawablePrivate * dPriv,
+                const drm_clip_rect_t * rect)
+{
+
+   struct intel_context *intel;
+   const intelScreenPrivate *intelScreen;
+
+   DBG("%s\n", __FUNCTION__);
+
+   assert(dPriv);
+
+   intel = intelScreenContext(dPriv->driScreenPriv->private);
+   if (!intel)
+      return;
+
+   intelScreen = intel->intelScreen;
+
+   /* The LOCK_HARDWARE is required for the cliprects.  Buffer offsets
+    * should work regardless.
+    */
+   LOCK_HARDWARE(intel);
+
+   if (dPriv && dPriv->numClipRects) {
+      struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+      struct intel_region *src, *dst;
+      int nbox = dPriv->numClipRects;
+      drm_clip_rect_t *pbox = dPriv->pClipRects;
+      int cpp;
+      int src_pitch, dst_pitch;
+      unsigned short src_x, src_y;
+      int BR13, CMD;
+      int i;
+      dri_bo *aper_array[3];
+
+      src = intel_get_rb_region(&intel_fb->Base, BUFFER_BACK_LEFT);
+      dst = intel_get_rb_region(&intel_fb->Base, BUFFER_FRONT_LEFT);
+
+      src_pitch = src->pitch * src->cpp;
+      dst_pitch = dst->pitch * dst->cpp;
+
+      cpp = src->cpp;
+
+      ASSERT(intel_fb);
+      ASSERT(intel_fb->Base.Name == 0);    /* Not a user-created FBO */
+      ASSERT(src);
+      ASSERT(dst);
+      ASSERT(src->cpp == dst->cpp);
+
+      if (cpp == 2) {
+	 BR13 = (0xCC << 16) | (1 << 24);
+	 CMD = XY_SRC_COPY_BLT_CMD;
+      }
+      else {
+	 BR13 = (0xCC << 16) | (1 << 24) | (1 << 25);
+	 CMD = XY_SRC_COPY_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
+      }
+
+#ifndef I915
+      if (src->tiling != I915_TILING_NONE) {
+	 CMD |= XY_SRC_TILED;
+	 src_pitch /= 4;
+      }
+      if (dst->tiling != I915_TILING_NONE) {
+	 CMD |= XY_DST_TILED;
+	 dst_pitch /= 4;
+      }
+#endif
+      /* do space/cliprects check before going any further */
+      intel_batchbuffer_require_space(intel->batch, 8 * 4,
+				      REFERENCES_CLIPRECTS);
+   again:
+      aper_array[0] = intel->batch->buf;
+      aper_array[1] = dst->buffer;
+      aper_array[2] = src->buffer;
+
+      if (dri_bufmgr_check_aperture_space(aper_array, 3) != 0) {
+	intel_batchbuffer_flush(intel->batch);
+	goto again;
+      }
+
+      for (i = 0; i < nbox; i++, pbox++) {
+	 drm_clip_rect_t box = *pbox;
+
+	 if (rect) {
+	    if (!intel_intersect_cliprects(&box, &box, rect))
+	       continue;
+	 }
+
+	 if (box.x1 >= box.x2 ||
+	     box.y1 >= box.y2)
+	    continue;
+
+	 assert(box.x1 < box.x2);
+	 assert(box.y1 < box.y2);
+	 src_x = box.x1 - dPriv->x + dPriv->backX;
+	 src_y = box.y1 - dPriv->y + dPriv->backY;
+
+	 BEGIN_BATCH(8, REFERENCES_CLIPRECTS);
+	 OUT_BATCH(CMD);
+	 OUT_BATCH(BR13 | dst_pitch);
+	 OUT_BATCH((box.y1 << 16) | box.x1);
+	 OUT_BATCH((box.y2 << 16) | box.x2);
+
+	 OUT_RELOC(dst->buffer,
+		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+		   0);
+	 OUT_BATCH((src_y << 16) | src_x);
+	 OUT_BATCH(src_pitch);
+	 OUT_RELOC(src->buffer,
+		   I915_GEM_DOMAIN_RENDER, 0,
+		   0);
+	 ADVANCE_BATCH();
+      }
+
+      /* Flush the rendering and the batch so that the results all land on the
+       * screen in a timely fashion.
+       */
+      intel_batchbuffer_emit_mi_flush(intel->batch);
+      intel_batchbuffer_flush(intel->batch);
+   }
+
+   UNLOCK_HARDWARE(intel);
+}
+
+
+
+
+void
+intelEmitFillBlit(struct intel_context *intel,
+		  GLuint cpp,
+		  GLshort dst_pitch,
+		  dri_bo *dst_buffer,
+		  GLuint dst_offset,
+		  uint32_t dst_tiling,
+		  GLshort x, GLshort y,
+		  GLshort w, GLshort h,
+		  GLuint color)
+{
+   GLuint BR13, CMD;
+   BATCH_LOCALS;
+
+   dst_pitch *= cpp;
+
+   switch (cpp) {
+   case 1:
+   case 2:
+   case 3:
+      BR13 = (0xF0 << 16) | (1 << 24);
+      CMD = XY_COLOR_BLT_CMD;
+      break;
+   case 4:
+      BR13 = (0xF0 << 16) | (1 << 24) | (1 << 25);
+      CMD = XY_COLOR_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
+      break;
+   default:
+      return;
+   }
+#ifndef I915
+   if (dst_tiling != I915_TILING_NONE) {
+      CMD |= XY_DST_TILED;
+      dst_pitch /= 4;
+   }
+#endif
+
+   DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n",
+       __FUNCTION__, dst_buffer, dst_pitch, dst_offset, x, y, w, h);
+
+   assert(w > 0);
+   assert(h > 0);
+
+   BEGIN_BATCH(6, NO_LOOP_CLIPRECTS);
+   OUT_BATCH(CMD);
+   OUT_BATCH(BR13 | dst_pitch);
+   OUT_BATCH((y << 16) | x);
+   OUT_BATCH(((y + h) << 16) | (x + w));
+   OUT_RELOC(dst_buffer,
+	     I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+	     dst_offset);
+   OUT_BATCH(color);
+   ADVANCE_BATCH();
+}
+
+static GLuint translate_raster_op(GLenum logicop)
+{
+   switch(logicop) {
+   case GL_CLEAR: return 0x00;
+   case GL_AND: return 0x88;
+   case GL_AND_REVERSE: return 0x44;
+   case GL_COPY: return 0xCC;
+   case GL_AND_INVERTED: return 0x22;
+   case GL_NOOP: return 0xAA;
+   case GL_XOR: return 0x66;
+   case GL_OR: return 0xEE;
+   case GL_NOR: return 0x11;
+   case GL_EQUIV: return 0x99;
+   case GL_INVERT: return 0x55;
+   case GL_OR_REVERSE: return 0xDD;
+   case GL_COPY_INVERTED: return 0x33;
+   case GL_OR_INVERTED: return 0xBB;
+   case GL_NAND: return 0x77;
+   case GL_SET: return 0xFF;
+   default: return 0;
+   }
+}
+
+
+/* Copy BitBlt
+ */
+void
+intelEmitCopyBlit(struct intel_context *intel,
+		  GLuint cpp,
+		  GLshort src_pitch,
+		  dri_bo *src_buffer,
+		  GLuint src_offset,
+		  uint32_t src_tiling,
+		  GLshort dst_pitch,
+		  dri_bo *dst_buffer,
+		  GLuint dst_offset,
+		  uint32_t dst_tiling,
+		  GLshort src_x, GLshort src_y,
+		  GLshort dst_x, GLshort dst_y,
+		  GLshort w, GLshort h,
+		  GLenum logic_op)
+{
+   GLuint CMD, BR13, pass = 0;
+   int dst_y2 = dst_y + h;
+   int dst_x2 = dst_x + w;
+   dri_bo *aper_array[3];
+   BATCH_LOCALS;
+
+   /* do space/cliprects check before going any further */
+   do {
+       aper_array[0] = intel->batch->buf;
+       aper_array[1] = dst_buffer;
+       aper_array[2] = src_buffer;
+
+       if (dri_bufmgr_check_aperture_space(aper_array, 3) != 0) {
+           intel_batchbuffer_flush(intel->batch);
+           pass++;
+       } else
+           break;
+   } while (pass < 2);
+
+   if (pass >= 2) {
+       GLboolean locked = GL_FALSE;       
+       if (!intel->locked) {
+           LOCK_HARDWARE(intel);
+           locked = GL_TRUE;
+       }
+
+       dri_bo_map(dst_buffer, GL_TRUE);
+       dri_bo_map(src_buffer, GL_FALSE);
+       _mesa_copy_rect((GLubyte *)dst_buffer->virtual + dst_offset,
+                       cpp,
+                       dst_pitch,
+                       dst_x, dst_y, 
+                       w, h, 
+                       (GLubyte *)src_buffer->virtual + src_offset, 
+                       src_pitch,
+                       src_x, src_y);
+       
+       dri_bo_unmap(src_buffer);
+       dri_bo_unmap(dst_buffer);
+       
+       if (locked)
+           UNLOCK_HARDWARE(intel);
+
+       return;
+   }
+
+   intel_batchbuffer_require_space(intel->batch, 8 * 4, NO_LOOP_CLIPRECTS);
+   DBG("%s src:buf(%p)/%d+%d %d,%d dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n",
+       __FUNCTION__,
+       src_buffer, src_pitch, src_offset, src_x, src_y,
+       dst_buffer, dst_pitch, dst_offset, dst_x, dst_y, w, h);
+
+   src_pitch *= cpp;
+   dst_pitch *= cpp;
+
+   BR13 = translate_raster_op(logic_op) << 16;
+
+   switch (cpp) {
+   case 1:
+   case 2:
+   case 3:
+      BR13 |= (1 << 24);
+      CMD = XY_SRC_COPY_BLT_CMD;
+      break;
+   case 4:
+      BR13 |= (1 << 24) | (1 << 25);
+      CMD = XY_SRC_COPY_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
+      break;
+   default:
+      return;
+   }
+
+#ifndef I915
+   if (dst_tiling != I915_TILING_NONE) {
+      CMD |= XY_DST_TILED;
+      dst_pitch /= 4;
+   }
+   if (src_tiling != I915_TILING_NONE) {
+      CMD |= XY_SRC_TILED;
+      src_pitch /= 4;
+   }
+#endif
+
+   if (dst_y2 <= dst_y || dst_x2 <= dst_x) {
+      return;
+   }
+
+   assert(dst_x < dst_x2);
+   assert(dst_y < dst_y2);
+
+   BEGIN_BATCH(8, NO_LOOP_CLIPRECTS);
+   OUT_BATCH(CMD);
+   OUT_BATCH(BR13 | (uint16_t)dst_pitch);
+   OUT_BATCH((dst_y << 16) | dst_x);
+   OUT_BATCH((dst_y2 << 16) | dst_x2);
+   OUT_RELOC(dst_buffer,
+	     I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+	     dst_offset);
+   OUT_BATCH((src_y << 16) | src_x);
+   OUT_BATCH((uint16_t)src_pitch);
+   OUT_RELOC(src_buffer,
+	     I915_GEM_DOMAIN_RENDER, 0,
+	     src_offset);
+   ADVANCE_BATCH();
+
+   intel_batchbuffer_emit_mi_flush(intel->batch);
+}
+
+
+/**
+ * Use blitting to clear the renderbuffers named by 'flags'.
+ * Note: we can't use the ctx->DrawBuffer->_ColorDrawBufferIndexes field
+ * since that might include software renderbuffers or renderbuffers
+ * which we're clearing with triangles.
+ * \param mask  bitmask of BUFFER_BIT_* values indicating buffers to clear
+ */
+void
+intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   GLuint clear_depth;
+   GLbitfield skipBuffers = 0;
+   unsigned int num_cliprects;
+   struct drm_clip_rect *cliprects;
+   int x_off, y_off;
+   BATCH_LOCALS;
+
+   /*
+    * Compute values for clearing the buffers.
+    */
+   clear_depth = 0;
+   if (mask & BUFFER_BIT_DEPTH) {
+      clear_depth = (GLuint) (fb->_DepthMax * ctx->Depth.Clear);
+   }
+   if (mask & BUFFER_BIT_STENCIL) {
+      clear_depth |= (ctx->Stencil.Clear & 0xff) << 24;
+   }
+
+   /* If clearing both depth and stencil, skip BUFFER_BIT_STENCIL in
+    * the loop below.
+    */
+   if ((mask & BUFFER_BIT_DEPTH) && (mask & BUFFER_BIT_STENCIL)) {
+      skipBuffers = BUFFER_BIT_STENCIL;
+   }
+
+   /* XXX Move this flush/lock into the following conditional? */
+   intelFlush(&intel->ctx);
+   LOCK_HARDWARE(intel);
+
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   if (num_cliprects) {
+      GLint cx, cy, cw, ch;
+      drm_clip_rect_t clear;
+      int i;
+
+      /* Get clear bounds after locking */
+      cx = fb->_Xmin;
+      cy = fb->_Ymin;
+      cw = fb->_Xmax - cx;
+      ch = fb->_Ymax - cy;
+
+      if (fb->Name == 0) {
+         /* clearing a window */
+
+         /* flip top to bottom */
+         clear.x1 = cx + x_off;
+         clear.y1 = intel->driDrawable->y + intel->driDrawable->h - cy - ch;
+         clear.x2 = clear.x1 + cw;
+         clear.y2 = clear.y1 + ch;
+      }
+      else {
+         /* clearing FBO */
+         assert(num_cliprects == 1);
+         assert(cliprects == &intel->fboRect);
+         clear.x1 = cx;
+         clear.y1 = cy;
+         clear.x2 = clear.x1 + cw;
+         clear.y2 = clear.y1 + ch;
+         /* no change to mask */
+      }
+
+      for (i = 0; i < num_cliprects; i++) {
+         const drm_clip_rect_t *box = &cliprects[i];
+         drm_clip_rect_t b;
+         GLuint buf;
+         GLuint clearMask = mask;      /* use copy, since we modify it below */
+         GLboolean all = (cw == fb->Width && ch == fb->Height);
+
+         if (!all) {
+            intel_intersect_cliprects(&b, &clear, box);
+         }
+         else {
+            b = *box;
+         }
+
+         if (b.x1 >= b.x2 || b.y1 >= b.y2)
+            continue;
+
+         if (0)
+            _mesa_printf("clear %d,%d..%d,%d, mask %x\n",
+                         b.x1, b.y1, b.x2, b.y2, mask);
+
+         /* Loop over all renderbuffers */
+         for (buf = 0; buf < BUFFER_COUNT && clearMask; buf++) {
+            const GLbitfield bufBit = 1 << buf;
+            if ((clearMask & bufBit) && !(bufBit & skipBuffers)) {
+               /* OK, clear this renderbuffer */
+               struct intel_region *irb_region =
+		  intel_get_rb_region(fb, buf);
+               dri_bo *write_buffer =
+                  intel_region_buffer(intel, irb_region,
+                                      all ? INTEL_WRITE_FULL :
+                                      INTEL_WRITE_PART);
+
+               GLuint clearVal;
+               GLint pitch, cpp;
+               GLuint BR13, CMD;
+
+               ASSERT(irb_region);
+
+               pitch = irb_region->pitch;
+               cpp = irb_region->cpp;
+
+               DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n",
+                   __FUNCTION__,
+                   irb_region->buffer, (pitch * cpp),
+                   irb_region->draw_offset,
+                   b.x1, b.y1, b.x2 - b.x1, b.y2 - b.y1);
+
+	       BR13 = 0xf0 << 16;
+	       CMD = XY_COLOR_BLT_CMD;
+
+               /* Setup the blit command */
+               if (cpp == 4) {
+                  BR13 |= (1 << 24) | (1 << 25);
+                  if (buf == BUFFER_DEPTH || buf == BUFFER_STENCIL) {
+                     if (clearMask & BUFFER_BIT_DEPTH)
+                        CMD |= XY_BLT_WRITE_RGB;
+                     if (clearMask & BUFFER_BIT_STENCIL)
+                        CMD |= XY_BLT_WRITE_ALPHA;
+                  }
+                  else {
+                     /* clearing RGBA */
+                     CMD |= XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
+                  }
+               }
+               else {
+                  ASSERT(cpp == 2 || cpp == 0);
+                  BR13 |= (1 << 24);
+               }
+
+#ifndef I915
+	       if (irb_region->tiling != I915_TILING_NONE) {
+		  CMD |= XY_DST_TILED;
+		  pitch /= 4;
+	       }
+#endif
+	       BR13 |= (pitch * cpp);
+
+               if (buf == BUFFER_DEPTH || buf == BUFFER_STENCIL) {
+                  clearVal = clear_depth;
+               }
+               else {
+                  clearVal = (cpp == 4)
+                     ? intel->ClearColor8888 : intel->ClearColor565;
+               }
+               /*
+                  _mesa_debug(ctx, "hardware blit clear buf %d rb id %d\n",
+                  buf, irb->Base.Name);
+                */
+	       intel_wait_flips(intel);
+
+               assert(b.x1 < b.x2);
+               assert(b.y1 < b.y2);
+
+               BEGIN_BATCH(6, REFERENCES_CLIPRECTS);
+               OUT_BATCH(CMD);
+               OUT_BATCH(BR13);
+               OUT_BATCH((b.y1 << 16) | b.x1);
+               OUT_BATCH((b.y2 << 16) | b.x2);
+               OUT_RELOC(write_buffer,
+			 I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+                         irb_region->draw_offset);
+               OUT_BATCH(clearVal);
+               ADVANCE_BATCH();
+               clearMask &= ~bufBit;    /* turn off bit, for faster loop exit */
+            }
+         }
+      }
+      intel_batchbuffer_emit_mi_flush(intel->batch);
+   }
+
+   UNLOCK_HARDWARE(intel);
+}
+
+void
+intelEmitImmediateColorExpandBlit(struct intel_context *intel,
+				  GLuint cpp,
+				  GLubyte *src_bits, GLuint src_size,
+				  GLuint fg_color,
+				  GLshort dst_pitch,
+				  dri_bo *dst_buffer,
+				  GLuint dst_offset,
+				  uint32_t dst_tiling,
+				  GLshort x, GLshort y,
+				  GLshort w, GLshort h,
+				  GLenum logic_op)
+{
+   int dwords = ALIGN(src_size, 8) / 4;
+   uint32_t opcode, br13, blit_cmd;
+
+   assert( logic_op - GL_CLEAR >= 0 );
+   assert( logic_op - GL_CLEAR < 0x10 );
+
+   if (w < 0 || h < 0)
+      return;
+
+   dst_pitch *= cpp;
+
+   DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d, %d bytes %d dwords\n",
+       __FUNCTION__,
+       dst_buffer, dst_pitch, dst_offset, x, y, w, h, src_size, dwords);
+
+   intel_batchbuffer_require_space( intel->batch,
+				    (8 * 4) +
+				    (3 * 4) +
+				    dwords * 4,
+				    REFERENCES_CLIPRECTS );
+
+   opcode = XY_SETUP_BLT_CMD;
+   if (cpp == 4)
+      opcode |= XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
+#ifndef I915
+   if (dst_tiling != I915_TILING_NONE) {
+      opcode |= XY_DST_TILED;
+      dst_pitch /= 4;
+   }
+#endif
+
+   br13 = dst_pitch | (translate_raster_op(logic_op) << 16) | (1 << 29);
+   if (cpp == 2)
+      br13 |= BR13_565;
+   else
+      br13 |= BR13_8888;
+
+   blit_cmd = XY_TEXT_IMMEDIATE_BLIT_CMD | XY_TEXT_BYTE_PACKED; /* packing? */
+   if (dst_tiling != I915_TILING_NONE)
+      blit_cmd |= XY_DST_TILED;
+
+   BEGIN_BATCH(8 + 3, REFERENCES_CLIPRECTS);
+   OUT_BATCH(opcode);
+   OUT_BATCH(br13);
+   OUT_BATCH((0 << 16) | 0); /* clip x1, y1 */
+   OUT_BATCH((100 << 16) | 100); /* clip x2, y2 */
+   OUT_RELOC(dst_buffer,
+	     I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+	     dst_offset);
+   OUT_BATCH(0); /* bg */
+   OUT_BATCH(fg_color); /* fg */
+   OUT_BATCH(0); /* pattern base addr */
+
+   OUT_BATCH(blit_cmd | ((3 - 2) + dwords));
+   OUT_BATCH((y << 16) | x);
+   OUT_BATCH(((y + h) << 16) | (x + w));
+   ADVANCE_BATCH();
+
+   intel_batchbuffer_data( intel->batch,
+			   src_bits,
+			   dwords * 4,
+			   REFERENCES_CLIPRECTS );
+
+   intel_batchbuffer_emit_mi_flush(intel->batch);
+}
diff --git a/src/mesa/drivers/dri/i965/intel_blit.h b/src/mesa/drivers/dri/intel/intel_blit.h
index e361545c8fa..52065b13ed7 100644
--- a/src/mesa/drivers/dri/i965/intel_blit.h
+++ b/src/mesa/drivers/dri/intel/intel_blit.h
@@ -29,38 +29,35 @@
 #define INTEL_BLIT_H
 
 #include "intel_context.h"
-#include "intel_ioctl.h"
 
-struct buffer;
+extern void intelCopyBuffer(const __DRIdrawablePrivate * dpriv,
+                            const drm_clip_rect_t * rect);
 
-extern void intelCopyBuffer( const __DRIdrawablePrivate *dpriv,
-			     const drm_clip_rect_t *rect );
-extern void intelClearWithBlit(GLcontext *ctx, GLbitfield mask);
+extern void intelClearWithBlit(GLcontext * ctx, GLbitfield mask);
 
-extern void intelEmitCopyBlit( struct intel_context *intel,
-			       GLuint cpp,
-			       GLshort src_pitch,
-			       struct buffer *src_buffer,
-			       GLuint  src_offset,
-			       GLboolean src_tiled,
-			       GLshort dst_pitch,
-			       struct buffer *dst_buffer,
-			       GLuint  dst_offset,
-			       GLboolean dst_tiled,
-			       GLshort srcx, GLshort srcy,
-			       GLshort dstx, GLshort dsty,
-			       GLshort w, GLshort h,
-			       GLenum logic_op );
+extern void intelEmitCopyBlit(struct intel_context *intel,
+                              GLuint cpp,
+                              GLshort src_pitch,
+                              dri_bo *src_buffer,
+                              GLuint src_offset,
+			      uint32_t src_tiling,
+                              GLshort dst_pitch,
+                              dri_bo *dst_buffer,
+                              GLuint dst_offset,
+			      uint32_t dst_tiling,
+                              GLshort srcx, GLshort srcy,
+                              GLshort dstx, GLshort dsty,
+                              GLshort w, GLshort h,
+			      GLenum logicop );
 
-extern void intelEmitFillBlit( struct intel_context *intel,
-			       GLuint cpp,
-			       GLshort dst_pitch,
-			       struct buffer *dst_buffer,
-			       GLuint dst_offset,
-			       GLboolean dst_tiled,
-			       GLshort x, GLshort y, 
-			       GLshort w, GLshort h,
-			       GLuint color );
+extern void intelEmitFillBlit(struct intel_context *intel,
+                              GLuint cpp,
+                              GLshort dst_pitch,
+                              dri_bo *dst_buffer,
+                              GLuint dst_offset,
+			      uint32_t dst_tiling,
+                              GLshort x, GLshort y,
+                              GLshort w, GLshort h, GLuint color);
 
 void
 intelEmitImmediateColorExpandBlit(struct intel_context *intel,
@@ -68,11 +65,11 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  GLubyte *src_bits, GLuint src_size,
 				  GLuint fg_color,
 				  GLshort dst_pitch,
-				  struct buffer *dst_buffer,
+				  dri_bo *dst_buffer,
 				  GLuint dst_offset,
-				  GLboolean dst_tiled,
-				  GLshort dst_x, GLshort dst_y, 
+				  uint32_t dst_tiling,
+				  GLshort x, GLshort y,
 				  GLshort w, GLshort h,
-				  GLenum logic_op );
+				  GLenum logic_op);
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
new file mode 100644
index 00000000000..60d7bb3770c
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
@@ -0,0 +1,284 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/bufferobj.h"
+
+#include "intel_context.h"
+#include "intel_buffer_objects.h"
+#include "intel_batchbuffer.h"
+#include "intel_regions.h"
+
+static GLboolean intel_bufferobj_unmap(GLcontext * ctx,
+				       GLenum target,
+				       struct gl_buffer_object *obj);
+
+/** Allocates a new dri_bo to store the data for the buffer object. */
+static void
+intel_bufferobj_alloc_buffer(struct intel_context *intel,
+			     struct intel_buffer_object *intel_obj)
+{
+   intel_obj->buffer = dri_bo_alloc(intel->bufmgr, "bufferobj",
+				    intel_obj->Base.Size, 64);
+}
+
+/**
+ * There is some duplication between mesa's bufferobjects and our
+ * bufmgr buffers.  Both have an integer handle and a hashtable to
+ * lookup an opaque structure.  It would be nice if the handles and
+ * internal structure where somehow shared.
+ */
+static struct gl_buffer_object *
+intel_bufferobj_alloc(GLcontext * ctx, GLuint name, GLenum target)
+{
+   struct intel_buffer_object *obj = CALLOC_STRUCT(intel_buffer_object);
+
+   _mesa_initialize_buffer_object(&obj->Base, name, target);
+
+   obj->buffer = NULL;
+
+   return &obj->Base;
+}
+
+/* Break the COW tie to the region.  The region gets to keep the data.
+ */
+void
+intel_bufferobj_release_region(struct intel_context *intel,
+                               struct intel_buffer_object *intel_obj)
+{
+   assert(intel_obj->region->buffer == intel_obj->buffer);
+   intel_obj->region->pbo = NULL;
+   intel_obj->region = NULL;
+
+   dri_bo_unreference(intel_obj->buffer);
+   intel_obj->buffer = NULL;
+}
+
+/* Break the COW tie to the region.  Both the pbo and the region end
+ * up with a copy of the data.
+ */
+void
+intel_bufferobj_cow(struct intel_context *intel,
+                    struct intel_buffer_object *intel_obj)
+{
+   assert(intel_obj->region);
+   intel_region_cow(intel, intel_obj->region);
+}
+
+
+/**
+ * Deallocate/free a vertex/pixel buffer object.
+ * Called via glDeleteBuffersARB().
+ */
+static void
+intel_bufferobj_free(GLcontext * ctx, struct gl_buffer_object *obj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   assert(intel_obj);
+
+   /* Buffer objects are automatically unmapped when deleting according
+    * to the spec.
+    */
+   if (obj->Pointer)
+      intel_bufferobj_unmap(ctx, 0, obj);
+
+   if (intel_obj->region) {
+      intel_bufferobj_release_region(intel, intel_obj);
+   }
+   else if (intel_obj->buffer) {
+      dri_bo_unreference(intel_obj->buffer);
+   }
+
+   _mesa_free(intel_obj);
+}
+
+
+
+/**
+ * Allocate space for and store data in a buffer object.  Any data that was
+ * previously stored in the buffer object is lost.  If data is NULL,
+ * memory will be allocated, but no copy will occur.
+ * Called via glBufferDataARB().
+ */
+static void
+intel_bufferobj_data(GLcontext * ctx,
+                     GLenum target,
+                     GLsizeiptrARB size,
+                     const GLvoid * data,
+                     GLenum usage, struct gl_buffer_object *obj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   intel_obj->Base.Size = size;
+   intel_obj->Base.Usage = usage;
+
+   /* Buffer objects are automatically unmapped when creating new data buffers
+    * according to the spec.
+    */
+   if (obj->Pointer)
+      intel_bufferobj_unmap(ctx, 0, obj);
+
+   if (intel_obj->region)
+      intel_bufferobj_release_region(intel, intel_obj);
+
+   if (intel_obj->buffer != NULL) {
+      dri_bo_unreference(intel_obj->buffer);
+      intel_obj->buffer = NULL;
+   }
+   if (size != 0) {
+      intel_bufferobj_alloc_buffer(intel, intel_obj);
+
+      if (data != NULL)
+	 dri_bo_subdata(intel_obj->buffer, 0, size, data);
+   }
+}
+
+
+/**
+ * Replace data in a subrange of buffer object.  If the data range
+ * specified by size + offset extends beyond the end of the buffer or
+ * if data is NULL, no copy is performed.
+ * Called via glBufferSubDataARB().
+ */
+static void
+intel_bufferobj_subdata(GLcontext * ctx,
+                        GLenum target,
+                        GLintptrARB offset,
+                        GLsizeiptrARB size,
+                        const GLvoid * data, struct gl_buffer_object *obj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   assert(intel_obj);
+
+   if (intel_obj->region)
+      intel_bufferobj_cow(intel, intel_obj);
+
+   dri_bo_subdata(intel_obj->buffer, offset, size, data);
+}
+
+
+/**
+ * Called via glGetBufferSubDataARB().
+ */
+static void
+intel_bufferobj_get_subdata(GLcontext * ctx,
+                            GLenum target,
+                            GLintptrARB offset,
+                            GLsizeiptrARB size,
+                            GLvoid * data, struct gl_buffer_object *obj)
+{
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   assert(intel_obj);
+   dri_bo_get_subdata(intel_obj->buffer, offset, size, data);
+}
+
+
+
+/**
+ * Called via glMapBufferARB().
+ */
+static void *
+intel_bufferobj_map(GLcontext * ctx,
+                    GLenum target,
+                    GLenum access, struct gl_buffer_object *obj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   /* XXX: Translate access to flags arg below:
+    */
+   assert(intel_obj);
+
+   if (intel_obj->region)
+      intel_bufferobj_cow(intel, intel_obj);
+
+   if (intel_obj->buffer == NULL) {
+      obj->Pointer = NULL;
+      return NULL;
+   }
+
+   dri_bo_map(intel_obj->buffer, GL_TRUE);
+   obj->Pointer = intel_obj->buffer->virtual;
+   return obj->Pointer;
+}
+
+
+/**
+ * Called via glMapBufferARB().
+ */
+static GLboolean
+intel_bufferobj_unmap(GLcontext * ctx,
+                      GLenum target, struct gl_buffer_object *obj)
+{
+   struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
+
+   assert(intel_obj);
+   if (intel_obj->buffer != NULL) {
+      assert(obj->Pointer);
+      dri_bo_unmap(intel_obj->buffer);
+      obj->Pointer = NULL;
+   }
+   return GL_TRUE;
+}
+
+dri_bo *
+intel_bufferobj_buffer(struct intel_context *intel,
+                       struct intel_buffer_object *intel_obj, GLuint flag)
+{
+   if (intel_obj->region) {
+      if (flag == INTEL_WRITE_PART)
+         intel_bufferobj_cow(intel, intel_obj);
+      else if (flag == INTEL_WRITE_FULL) {
+         intel_bufferobj_release_region(intel, intel_obj);
+	 intel_bufferobj_alloc_buffer(intel, intel_obj);
+      }
+   }
+
+   return intel_obj->buffer;
+}
+
+void
+intel_bufferobj_init(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   ctx->Driver.NewBufferObject = intel_bufferobj_alloc;
+   ctx->Driver.DeleteBuffer = intel_bufferobj_free;
+   ctx->Driver.BufferData = intel_bufferobj_data;
+   ctx->Driver.BufferSubData = intel_bufferobj_subdata;
+   ctx->Driver.GetBufferSubData = intel_bufferobj_get_subdata;
+   ctx->Driver.MapBuffer = intel_bufferobj_map;
+   ctx->Driver.UnmapBuffer = intel_bufferobj_unmap;
+}
diff --git a/src/mesa/drivers/dri/i965/intel_buffer_objects.h b/src/mesa/drivers/dri/intel/intel_buffer_objects.h
index 4b38803e576..bf6dbd58f27 100644
--- a/src/mesa/drivers/dri/i965/intel_buffer_objects.h
+++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.h
@@ -1,6 +1,6 @@
- /**************************************************************************
+/**************************************************************************
  * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2005 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -28,28 +28,36 @@
 #ifndef INTEL_BUFFEROBJ_H
 #define INTEL_BUFFEROBJ_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 struct intel_context;
+struct intel_region;
 struct gl_buffer_object;
 
 
 /**
  * Intel vertex/pixel buffer object, derived from Mesa's gl_buffer_object.
  */
-struct intel_buffer_object {
+struct intel_buffer_object
+{
    struct gl_buffer_object Base;
-   struct buffer *buffer;   /* the low-level buffer manager's buffer handle */
+   dri_bo *buffer;     /* the low-level buffer manager's buffer handle */
+
+   struct intel_region *region; /* Is there a zero-copy texture
+                                   associated with this (pixel)
+                                   buffer object? */
 };
 
 
 /* Get the bm buffer associated with a GL bufferobject:
  */
-struct buffer *intel_bufferobj_buffer( const struct intel_buffer_object *obj );
+dri_bo *intel_bufferobj_buffer(struct intel_context *intel,
+			       struct intel_buffer_object
+			       *obj, GLuint flag);
 
 /* Hook the bufferobject implementation into mesa: 
  */
-void intel_bufferobj_init( struct intel_context *intel );
+void intel_bufferobj_init(struct intel_context *intel);
 
 
 
@@ -58,13 +66,21 @@ void intel_bufferobj_init( struct intel_context *intel );
  * the Name == 0 test is the only way to identify them and avoid
  * casting them erroneously to our structs.
  */
-static inline struct intel_buffer_object *
-intel_buffer_object( struct gl_buffer_object *obj )
+static INLINE struct intel_buffer_object *
+intel_buffer_object(struct gl_buffer_object *obj)
 {
    if (obj->Name)
-      return (struct intel_buffer_object *)obj;
+      return (struct intel_buffer_object *) obj;
    else
       return NULL;
 }
 
+/* Helpers for zerocopy image uploads.  See also intel_regions.h:
+ */
+void intel_bufferobj_cow(struct intel_context *intel,
+                         struct intel_buffer_object *intel_obj);
+void intel_bufferobj_release_region(struct intel_context *intel,
+                                    struct intel_buffer_object *intel_obj);
+
+
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.c b/src/mesa/drivers/dri/intel/intel_buffers.c
new file mode 100644
index 00000000000..f8f009c6a30
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_buffers.c
@@ -0,0 +1,1024 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_chipset.h"
+#include "intel_depthstencil.h"
+#include "intel_fbo.h"
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
+#include "intel_reg.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "swrast/swrast.h"
+#include "utils.h"
+#include "drirenderbuffer.h"
+#include "vblank.h"
+#include "i915_drm.h"
+
+/* This block can be removed when libdrm >= 2.3.1 is required */
+
+#ifndef DRM_IOCTL_I915_FLIP
+
+#define DRM_VBLANK_FLIP 0x8000000
+
+typedef struct drm_i915_flip {
+   int pipes;
+} drm_i915_flip_t;
+
+#undef DRM_IOCTL_I915_FLIP
+#define DRM_IOCTL_I915_FLIP DRM_IOW(DRM_COMMAND_BASE + DRM_I915_FLIP, \
+				    drm_i915_flip_t)
+
+#endif
+
+#define FILE_DEBUG_FLAG DEBUG_BLIT
+
+/**
+ * XXX move this into a new dri/common/cliprects.c file.
+ */
+GLboolean
+intel_intersect_cliprects(drm_clip_rect_t * dst,
+                          const drm_clip_rect_t * a,
+                          const drm_clip_rect_t * b)
+{
+   GLint bx = b->x1;
+   GLint by = b->y1;
+   GLint bw = b->x2 - bx;
+   GLint bh = b->y2 - by;
+
+   if (bx < a->x1)
+      bw -= a->x1 - bx, bx = a->x1;
+   if (by < a->y1)
+      bh -= a->y1 - by, by = a->y1;
+   if (bx + bw > a->x2)
+      bw = a->x2 - bx;
+   if (by + bh > a->y2)
+      bh = a->y2 - by;
+   if (bw <= 0)
+      return GL_FALSE;
+   if (bh <= 0)
+      return GL_FALSE;
+
+   dst->x1 = bx;
+   dst->y1 = by;
+   dst->x2 = bx + bw;
+   dst->y2 = by + bh;
+
+   return GL_TRUE;
+}
+
+/**
+ * Return pointer to current color drawing region, or NULL.
+ */
+struct intel_region *
+intel_drawbuf_region(struct intel_context *intel)
+{
+   struct intel_renderbuffer *irbColor =
+      intel_renderbuffer(intel->ctx.DrawBuffer->_ColorDrawBuffers[0]);
+   if (irbColor)
+      return irbColor->region;
+   else
+      return NULL;
+}
+
+/**
+ * Return pointer to current color reading region, or NULL.
+ */
+struct intel_region *
+intel_readbuf_region(struct intel_context *intel)
+{
+   struct intel_renderbuffer *irb
+      = intel_renderbuffer(intel->ctx.ReadBuffer->_ColorReadBuffer);
+   if (irb)
+      return irb->region;
+   else
+      return NULL;
+}
+
+void
+intel_get_cliprects(struct intel_context *intel,
+		    struct drm_clip_rect **cliprects,
+		    unsigned int *num_cliprects,
+		    int *x_off, int *y_off)
+{
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+
+   if (intel->constant_cliprect) {
+      /* FBO or DRI2 rendering, which can just use the fb's size. */
+      intel->fboRect.x1 = 0;
+      intel->fboRect.y1 = 0;
+      intel->fboRect.x2 = intel->ctx.DrawBuffer->Width;
+      intel->fboRect.y2 = intel->ctx.DrawBuffer->Height;
+
+      *cliprects = &intel->fboRect;
+      *num_cliprects = 1;
+      *x_off = 0;
+      *y_off = 0;
+   } else if (intel->front_cliprects ||
+       intel_fb->pf_active || dPriv->numBackClipRects == 0) {
+      /* use the front clip rects */
+      *cliprects = dPriv->pClipRects;
+      *num_cliprects = dPriv->numClipRects;
+      *x_off = dPriv->x;
+      *y_off = dPriv->y;
+   }
+   else {
+      /* use the back clip rects */
+      *num_cliprects = dPriv->numBackClipRects;
+      *cliprects = dPriv->pBackClipRects;
+      *x_off = dPriv->backX;
+      *y_off = dPriv->backY;
+   }
+}
+
+static void
+intelUpdatePageFlipping(struct intel_context *intel,
+			GLint areaA, GLint areaB)
+{
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+   GLboolean pf_active;
+   GLint pf_planes;
+
+   /* Update page flipping info */
+   pf_planes = 0;
+
+   if (areaA > 0)
+      pf_planes |= 1;
+
+   if (areaB > 0)
+      pf_planes |= 2;
+
+   intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
+				(intel_fb->pf_planes & 0x2)) & 0x3;
+
+   intel_fb->pf_num_pages = intel->intelScreen->third.handle ? 3 : 2;
+
+   pf_active = pf_planes && (pf_planes & intel->sarea->pf_active) == pf_planes;
+
+   if (INTEL_DEBUG & DEBUG_LOCK)
+      if (pf_active != intel_fb->pf_active)
+	 _mesa_printf("%s - Page flipping %sactive\n", __progname,
+		      pf_active ? "" : "in");
+
+   if (pf_active) {
+      /* Sync pages between planes if flipping on both at the same time */
+      if (pf_planes == 0x3 && pf_planes != intel_fb->pf_planes &&
+	  (intel->sarea->pf_current_page & 0x3) !=
+	  (((intel->sarea->pf_current_page) >> 2) & 0x3)) {
+	 drm_i915_flip_t flip;
+
+	 if (intel_fb->pf_current_page ==
+	     (intel->sarea->pf_current_page & 0x3)) {
+	    /* XXX: This is ugly, but emitting two flips 'in a row' can cause
+	     * lockups for unknown reasons.
+	     */
+	    intel->sarea->pf_current_page =
+	       intel->sarea->pf_current_page & 0x3;
+	    intel->sarea->pf_current_page |=
+	       ((intel_fb->pf_current_page + intel_fb->pf_num_pages - 1) %
+		intel_fb->pf_num_pages) << 2;
+
+	    flip.pipes = 0x2;
+	 } else {
+	    intel->sarea->pf_current_page =
+	       intel->sarea->pf_current_page & (0x3 << 2);
+	    intel->sarea->pf_current_page |=
+	       (intel_fb->pf_current_page + intel_fb->pf_num_pages - 1) %
+	       intel_fb->pf_num_pages;
+
+	    flip.pipes = 0x1;
+	 }
+
+	 drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
+      }
+
+      intel_fb->pf_planes = pf_planes;
+   }
+
+   intel_fb->pf_active = pf_active;
+   intel_flip_renderbuffers(intel_fb);
+   intel_draw_buffer(&intel->ctx, intel->ctx.DrawBuffer);
+}
+
+/**
+ * This will be called whenever the currently bound window is moved/resized.
+ * XXX: actually, it seems to NOT be called when the window is only moved (BP).
+ */
+void
+intelWindowMoved(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+
+   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
+       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
+      volatile struct drm_i915_sarea *sarea = intel->sarea;
+      drm_clip_rect_t drw_rect = { .x1 = dPriv->x, .x2 = dPriv->x + dPriv->w,
+				   .y1 = dPriv->y, .y2 = dPriv->y + dPriv->h };
+      drm_clip_rect_t planeA_rect = { .x1 = sarea->planeA_x, .y1 = sarea->planeA_y,
+				     .x2 = sarea->planeA_x + sarea->planeA_w,
+				     .y2 = sarea->planeA_y + sarea->planeA_h };
+      drm_clip_rect_t planeB_rect = { .x1 = sarea->planeB_x, .y1 = sarea->planeB_y,
+				     .x2 = sarea->planeB_x + sarea->planeB_w,
+				     .y2 = sarea->planeB_y + sarea->planeB_h };
+      GLint areaA = driIntersectArea( drw_rect, planeA_rect );
+      GLint areaB = driIntersectArea( drw_rect, planeB_rect );
+      GLuint flags = dPriv->vblFlags;
+
+      intelUpdatePageFlipping(intel, areaA, areaB);
+
+      /* Update vblank info
+       */
+      if (areaB > areaA || (areaA == areaB && areaB > 0)) {
+	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
+      } else {
+	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
+      }
+
+      /* Check to see if we changed pipes */
+      if (flags != dPriv->vblFlags && dPriv->vblFlags &&
+	  !(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ)) {
+	 int64_t count;
+	 drmVBlank vbl;
+	 int i;
+
+	 /*
+	  * Deal with page flipping
+	  */
+	 vbl.request.type = DRM_VBLANK_ABSOLUTE;
+
+	 if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
+	    vbl.request.type |= DRM_VBLANK_SECONDARY;
+	 }
+
+	 for (i = 0; i < intel_fb->pf_num_pages; i++) {
+	    if (!intel_fb->color_rb[i] ||
+		(intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <=
+		(1<<23))
+	       continue;
+
+	    vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending;
+	    drmWaitVBlank(intel->driFd, &vbl);
+	 }
+
+	 /*
+	  * Update msc_base from old pipe
+	  */
+	 driDrawableGetMSC32(dPriv->driScreenPriv, dPriv, &count);
+	 dPriv->msc_base = count;
+	 /*
+	  * Then get new vblank_base and vblSeq values
+	  */
+	 dPriv->vblFlags = flags;
+	 driGetCurrentVBlank(dPriv);
+	 dPriv->vblank_base = dPriv->vblSeq;
+
+	 intel_fb->vbl_waited = dPriv->vblSeq;
+
+	 for (i = 0; i < intel_fb->pf_num_pages; i++) {
+	    if (intel_fb->color_rb[i])
+	       intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_waited;
+	 }
+      }
+   } else {
+      dPriv->vblFlags &= ~VBLANK_FLAG_SECONDARY;
+   }
+
+   /* Update Mesa's notion of window size */
+   driUpdateFramebufferSize(ctx, dPriv);
+   intel_fb->Base.Initialized = GL_TRUE; /* XXX remove someday */
+
+   /* Update hardware scissor */
+   if (ctx->Driver.Scissor != NULL) {
+      ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			  ctx->Scissor.Width, ctx->Scissor.Height);
+   }
+
+   /* Re-calculate viewport related state */
+   if (ctx->Driver.DepthRange != NULL)
+      ctx->Driver.DepthRange( ctx, ctx->Viewport.Near, ctx->Viewport.Far );
+}
+
+
+
+/* A true meta version of this would be very simple and additionally
+ * machine independent.  Maybe we'll get there one day.
+ */
+static void
+intelClearWithTris(struct intel_context *intel, GLbitfield mask)
+{
+   GLcontext *ctx = &intel->ctx;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   GLuint buf;
+
+   intel->vtbl.install_meta_state(intel);
+
+   /* Back and stencil cliprects are the same.  Try and do both
+    * buffers at once:
+    */
+   if (mask & (BUFFER_BIT_BACK_LEFT | BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH)) {
+      struct intel_region *backRegion =
+	 intel_get_rb_region(fb, BUFFER_BACK_LEFT);
+      struct intel_region *depthRegion =
+	 intel_get_rb_region(fb, BUFFER_DEPTH);
+
+      intel->vtbl.meta_draw_region(intel, backRegion, depthRegion);
+
+      if (mask & BUFFER_BIT_BACK_LEFT)
+	 intel->vtbl.meta_color_mask(intel, GL_TRUE);
+      else
+	 intel->vtbl.meta_color_mask(intel, GL_FALSE);
+
+      if (mask & BUFFER_BIT_STENCIL)
+	 intel->vtbl.meta_stencil_replace(intel,
+					  intel->ctx.Stencil.WriteMask[0],
+					  intel->ctx.Stencil.Clear);
+      else
+	 intel->vtbl.meta_no_stencil_write(intel);
+
+      if (mask & BUFFER_BIT_DEPTH)
+	 intel->vtbl.meta_depth_replace(intel);
+      else
+	 intel->vtbl.meta_no_depth_write(intel);
+
+      intel->vtbl.meta_draw_quad(intel,
+				 fb->_Xmin,
+				 fb->_Xmax,
+				 fb->_Ymin,
+				 fb->_Ymax,
+				 intel->ctx.Depth.Clear,
+				 intel->ClearColor8888,
+				 0, 0, 0, 0);   /* texcoords */
+
+      mask &= ~(BUFFER_BIT_BACK_LEFT | BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH);
+   }
+
+   /* clear the remaining (color) renderbuffers */
+   for (buf = 0; buf < BUFFER_COUNT && mask; buf++) {
+      const GLuint bufBit = 1 << buf;
+      if (mask & bufBit) {
+	 struct intel_renderbuffer *irbColor =
+	    intel_renderbuffer(fb->Attachment[buf].Renderbuffer);
+
+	 ASSERT(irbColor);
+
+	 intel->vtbl.meta_no_depth_write(intel);
+	 intel->vtbl.meta_no_stencil_write(intel);
+	 intel->vtbl.meta_color_mask(intel, GL_TRUE);
+	 intel->vtbl.meta_draw_region(intel, irbColor->region, NULL);
+
+	 intel->vtbl.meta_draw_quad(intel,
+				    fb->_Xmin,
+				    fb->_Xmax,
+				    fb->_Ymin,
+				    fb->_Ymax,
+				    0, intel->ClearColor8888,
+				    0, 0, 0, 0);   /* texcoords */
+
+	 mask &= ~bufBit;
+      }
+   }
+
+   intel->vtbl.leave_meta_state(intel);
+}
+
+static const char *buffer_names[] = {
+   [BUFFER_FRONT_LEFT] = "front",
+   [BUFFER_BACK_LEFT] = "back",
+   [BUFFER_FRONT_RIGHT] = "front right",
+   [BUFFER_BACK_RIGHT] = "back right",
+   [BUFFER_AUX0] = "aux0",
+   [BUFFER_AUX1] = "aux1",
+   [BUFFER_AUX2] = "aux2",
+   [BUFFER_AUX3] = "aux3",
+   [BUFFER_DEPTH] = "depth",
+   [BUFFER_STENCIL] = "stencil",
+   [BUFFER_ACCUM] = "accum",
+   [BUFFER_COLOR0] = "color0",
+   [BUFFER_COLOR1] = "color1",
+   [BUFFER_COLOR2] = "color2",
+   [BUFFER_COLOR3] = "color3",
+   [BUFFER_COLOR4] = "color4",
+   [BUFFER_COLOR5] = "color5",
+   [BUFFER_COLOR6] = "color6",
+   [BUFFER_COLOR7] = "color7",
+};
+
+/**
+ * Called by ctx->Driver.Clear.
+ */
+static void
+intelClear(GLcontext *ctx, GLbitfield mask)
+{
+   struct intel_context *intel = intel_context(ctx);
+   const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
+   GLbitfield tri_mask = 0;
+   GLbitfield blit_mask = 0;
+   GLbitfield swrast_mask = 0;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   GLuint i;
+
+   if (0)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   /* HW color buffers (front, back, aux, generic FBO, etc) */
+   if (colorMask == ~0) {
+      /* clear all R,G,B,A */
+      /* XXX FBO: need to check if colorbuffers are software RBOs! */
+      blit_mask |= (mask & BUFFER_BITS_COLOR);
+   }
+   else {
+      /* glColorMask in effect */
+      tri_mask |= (mask & BUFFER_BITS_COLOR);
+   }
+
+   /* HW stencil */
+   if (mask & BUFFER_BIT_STENCIL) {
+      const struct intel_region *stencilRegion
+         = intel_get_rb_region(fb, BUFFER_STENCIL);
+      if (stencilRegion) {
+         /* have hw stencil */
+         if (IS_965(intel->intelScreen->deviceID) ||
+	     (ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
+	    /* We have to use the 3D engine if we're clearing a partial mask
+	     * of the stencil buffer, or if we're on a 965 which has a tiled
+	     * depth/stencil buffer in a layout we can't blit to.
+	     */
+            tri_mask |= BUFFER_BIT_STENCIL;
+         }
+         else {
+            /* clearing all stencil bits, use blitting */
+            blit_mask |= BUFFER_BIT_STENCIL;
+         }
+      }
+   }
+
+   /* HW depth */
+   if (mask & BUFFER_BIT_DEPTH) {
+      /* clear depth with whatever method is used for stencil (see above) */
+      if (IS_965(intel->intelScreen->deviceID) ||
+	  tri_mask & BUFFER_BIT_STENCIL)
+         tri_mask |= BUFFER_BIT_DEPTH;
+      else
+         blit_mask |= BUFFER_BIT_DEPTH;
+   }
+
+   /* SW fallback clearing */
+   swrast_mask = mask & ~tri_mask & ~blit_mask;
+
+   for (i = 0; i < BUFFER_COUNT; i++) {
+      GLuint bufBit = 1 << i;
+      if ((blit_mask | tri_mask) & bufBit) {
+         if (!fb->Attachment[i].Renderbuffer->ClassID) {
+            blit_mask &= ~bufBit;
+            tri_mask &= ~bufBit;
+            swrast_mask |= bufBit;
+         }
+      }
+   }
+
+   if (blit_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("blit clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (blit_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      intelClearWithBlit(ctx, blit_mask);
+   }
+
+   if (tri_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("tri clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (tri_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      intelClearWithTris(intel, tri_mask);
+   }
+
+   if (swrast_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("swrast clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (swrast_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      _swrast_Clear(ctx, swrast_mask);
+   }
+}
+
+
+/* Emit wait for pending flips */
+void
+intel_wait_flips(struct intel_context *intel)
+{
+   struct intel_framebuffer *intel_fb =
+      (struct intel_framebuffer *) intel->ctx.DrawBuffer;
+   struct intel_renderbuffer *intel_rb =
+      intel_get_renderbuffer(&intel_fb->Base,
+			     intel_fb->Base._ColorDrawBufferIndexes[0] ==
+			     BUFFER_FRONT_LEFT ? BUFFER_FRONT_LEFT :
+			     BUFFER_BACK_LEFT);
+
+   if (intel->intelScreen->driScrnPriv->dri2.enabled)
+      return;
+
+   if (intel_fb->Base.Name == 0 && intel_rb &&
+       intel_rb->pf_pending == intel_fb->pf_seq) {
+      GLint pf_planes = intel_fb->pf_planes;
+      BATCH_LOCALS;
+
+      /* Wait for pending flips to take effect */
+      BEGIN_BATCH(2, NO_LOOP_CLIPRECTS);
+      OUT_BATCH(pf_planes & 0x1 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_A_FLIP)
+		: 0);
+      OUT_BATCH(pf_planes & 0x2 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_B_FLIP)
+		: 0);
+      ADVANCE_BATCH();
+
+      intel_rb->pf_pending--;
+   }
+}
+
+
+/* Flip the front & back buffers
+ */
+static GLboolean
+intelPageFlip(const __DRIdrawablePrivate * dPriv)
+{
+   struct intel_context *intel;
+   int ret;
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+
+   if (INTEL_DEBUG & DEBUG_IOCTL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   assert(dPriv);
+   assert(dPriv->driContextPriv);
+   assert(dPriv->driContextPriv->driverPrivate);
+
+   intel = (struct intel_context *) dPriv->driContextPriv->driverPrivate;
+
+   if (intel->intelScreen->drmMinor < 9)
+      return GL_FALSE;
+
+   intelFlush(&intel->ctx);
+
+   ret = 0;
+
+   LOCK_HARDWARE(intel);
+
+   if (dPriv->numClipRects && intel_fb->pf_active) {
+      drm_i915_flip_t flip;
+
+      flip.pipes = intel_fb->pf_planes;
+
+      ret = drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
+   }
+
+   UNLOCK_HARDWARE(intel);
+
+   if (ret || !intel_fb->pf_active)
+      return GL_FALSE;
+
+   if (!dPriv->numClipRects) {
+      usleep(10000);	/* throttle invisible client 10ms */
+   }
+
+   intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
+				(intel_fb->pf_planes & 0x2)) & 0x3;
+
+   if (dPriv->numClipRects != 0) {
+      intel_get_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT)->pf_pending =
+      intel_get_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT)->pf_pending =
+	 ++intel_fb->pf_seq;
+   }
+
+   intel_flip_renderbuffers(intel_fb);
+   intel_draw_buffer(&intel->ctx, &intel_fb->Base);
+
+   return GL_TRUE;
+}
+
+static GLboolean
+intelScheduleSwap(__DRIdrawablePrivate * dPriv, GLboolean *missed_target)
+{
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+   unsigned int interval;
+   struct intel_context *intel =
+      intelScreenContext(dPriv->driScreenPriv->private);
+   const intelScreenPrivate *intelScreen = intel->intelScreen;
+   unsigned int target;
+   drm_i915_vblank_swap_t swap;
+   GLboolean ret;
+
+   if (!dPriv->vblFlags ||
+       (dPriv->vblFlags & VBLANK_FLAG_NO_IRQ) ||
+       intelScreen->drmMinor < (intel_fb->pf_active ? 9 : 6))
+      return GL_FALSE;
+
+   interval = driGetVBlankInterval(dPriv);
+
+   swap.seqtype = DRM_VBLANK_ABSOLUTE;
+
+   if (dPriv->vblFlags & VBLANK_FLAG_SYNC) {
+      swap.seqtype |= DRM_VBLANK_NEXTONMISS;
+   } else if (interval == 0)
+      return GL_FALSE;
+
+   swap.drawable = dPriv->hHWDrawable;
+   target = swap.sequence = dPriv->vblSeq + interval;
+
+   if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
+      swap.seqtype |= DRM_VBLANK_SECONDARY;
+   }
+
+   LOCK_HARDWARE(intel);
+
+   intel_batchbuffer_flush(intel->batch);
+
+   if ( intel_fb->pf_active ) {
+      swap.seqtype |= DRM_VBLANK_FLIP;
+
+      intel_fb->pf_current_page = (((intel->sarea->pf_current_page >>
+				     (intel_fb->pf_planes & 0x2)) & 0x3) + 1) %
+				  intel_fb->pf_num_pages;
+   }
+
+   if (!drmCommandWriteRead(intel->driFd, DRM_I915_VBLANK_SWAP, &swap,
+			    sizeof(swap))) {
+      dPriv->vblSeq = swap.sequence;
+      swap.sequence -= target;
+      *missed_target = swap.sequence > 0 && swap.sequence <= (1 << 23);
+
+      intel_get_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT)->vbl_pending =
+	 intel_get_renderbuffer(&intel_fb->Base,
+				BUFFER_FRONT_LEFT)->vbl_pending =
+	 dPriv->vblSeq;
+
+      if (swap.seqtype & DRM_VBLANK_FLIP) {
+	 intel_flip_renderbuffers(intel_fb);
+	 intel_draw_buffer(&intel->ctx, intel->ctx.DrawBuffer);
+      }
+
+      ret = GL_TRUE;
+   } else {
+      if (swap.seqtype & DRM_VBLANK_FLIP) {
+	 intel_fb->pf_current_page = ((intel->sarea->pf_current_page >>
+					(intel_fb->pf_planes & 0x2)) & 0x3) %
+				     intel_fb->pf_num_pages;
+      }
+
+      ret = GL_FALSE;
+   }
+
+   UNLOCK_HARDWARE(intel);
+
+   return ret;
+}
+  
+void
+intelSwapBuffers(__DRIdrawablePrivate * dPriv)
+{
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+
+   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
+      GET_CURRENT_CONTEXT(ctx);
+      struct intel_context *intel;
+
+      if (ctx == NULL)
+	 return;
+
+      intel = intel_context(ctx);
+
+      if (ctx->Visual.doubleBufferMode) {
+	 GLboolean missed_target;
+	 struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+	 int64_t ust;
+         
+	 _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
+
+         if (!intelScheduleSwap(dPriv, &missed_target)) {
+	    driWaitForVBlank(dPriv, &missed_target);
+
+	    /*
+	     * Update each buffer's vbl_pending so we don't get too out of
+	     * sync
+	     */
+	    intel_get_renderbuffer(&intel_fb->Base,
+				   BUFFER_BACK_LEFT)->vbl_pending = 
+		    intel_get_renderbuffer(&intel_fb->Base,
+					   BUFFER_FRONT_LEFT)->vbl_pending =
+		    dPriv->vblSeq;
+	    if (!intelPageFlip(dPriv)) {
+	       intelCopyBuffer(dPriv, NULL);
+	    }
+	 }
+
+	 intel_fb->swap_count++;
+	 (*psp->systemTime->getUST) (&ust);
+	 if (missed_target) {
+	    intel_fb->swap_missed_count++;
+	    intel_fb->swap_missed_ust = ust - intel_fb->swap_ust;
+	 }
+
+	 intel_fb->swap_ust = ust;
+      }
+      drmCommandNone(intel->driFd, DRM_I915_GEM_THROTTLE);
+
+   }
+   else {
+      /* XXX this shouldn't be an error but we can't handle it for now */
+      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
+   }
+}
+
+void
+intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h)
+{
+   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
+      struct intel_context *intel =
+         (struct intel_context *) dPriv->driContextPriv->driverPrivate;
+      GLcontext *ctx = &intel->ctx;
+
+      if (ctx->Visual.doubleBufferMode) {
+         drm_clip_rect_t rect;
+         rect.x1 = x + dPriv->x;
+         rect.y1 = (dPriv->h - y - h) + dPriv->y;
+         rect.x2 = rect.x1 + w;
+         rect.y2 = rect.y1 + h;
+         _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
+         intelCopyBuffer(dPriv, &rect);
+      }
+   }
+   else {
+      /* XXX this shouldn't be an error but we can't handle it for now */
+      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
+   }
+}
+
+
+/**
+ * Update the hardware state for drawing into a window or framebuffer object.
+ *
+ * Called by glDrawBuffer, glBindFramebufferEXT, MakeCurrent, and other
+ * places within the driver.
+ *
+ * Basically, this needs to be called any time the current framebuffer
+ * changes, the renderbuffers change, or we need to draw into different
+ * color buffers.
+ */
+void
+intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *colorRegions[MAX_DRAW_BUFFERS], *depthRegion = NULL;
+   struct intel_renderbuffer *irbDepth = NULL, *irbStencil = NULL;
+
+   if (!fb) {
+      /* this can happen during the initial context initialization */
+      return;
+   }
+
+   /* Do this here, note core Mesa, since this function is called from
+    * many places within the driver.
+    */
+   if (ctx->NewState & (_NEW_BUFFERS | _NEW_COLOR | _NEW_PIXEL)) {
+      /* this updates the DrawBuffer->_NumColorDrawBuffers fields, etc */
+      _mesa_update_framebuffer(ctx);
+      /* this updates the DrawBuffer's Width/Height if it's a FBO */
+      _mesa_update_draw_buffer_bounds(ctx);
+   }
+
+   if (fb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+      /* this may occur when we're called by glBindFrameBuffer() during
+       * the process of someone setting up renderbuffers, etc.
+       */
+      /*_mesa_debug(ctx, "DrawBuffer: incomplete user FBO\n");*/
+      return;
+   }
+
+   if (fb->Name)
+      intel_validate_paired_depth_stencil(ctx, fb);
+
+   /*
+    * How many color buffers are we drawing into?
+    */
+   if (fb->_NumColorDrawBuffers == 0) {
+      /* writing to 0  */
+      colorRegions[0] = NULL;
+      intel->constant_cliprect = GL_TRUE;
+   } else if (fb->_NumColorDrawBuffers > 1) {
+       int i;
+       struct intel_renderbuffer *irb;
+
+       for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
+           irb = intel_renderbuffer(fb->_ColorDrawBuffers[i]);
+           colorRegions[i] = irb ? irb->region : NULL;
+       }
+       intel->constant_cliprect = GL_TRUE;
+   }
+   else {
+      /* Get the intel_renderbuffer for the single colorbuffer we're drawing
+       * into, and set up cliprects if it's .
+       */
+      if (fb->Name == 0) {
+	 intel->constant_cliprect = intel->driScreen->dri2.enabled;
+	 /* drawing to window system buffer */
+	 if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
+	    if (!intel->constant_cliprect && !intel->front_cliprects)
+	       intel_batchbuffer_flush(intel->batch);
+	    intel->front_cliprects = GL_TRUE;
+	    colorRegions[0] = intel_get_rb_region(fb, BUFFER_FRONT_LEFT);
+	 }
+	 else {
+	    if (!intel->constant_cliprect && intel->front_cliprects)
+	       intel_batchbuffer_flush(intel->batch);
+	    intel->front_cliprects = GL_FALSE;
+	    colorRegions[0]= intel_get_rb_region(fb, BUFFER_BACK_LEFT);
+	 }
+      }
+      else {
+	 /* drawing to user-created FBO */
+	 struct intel_renderbuffer *irb;
+	 irb = intel_renderbuffer(fb->_ColorDrawBuffers[0]);
+	 colorRegions[0] = (irb && irb->region) ? irb->region : NULL;
+	 intel->constant_cliprect = GL_TRUE;
+      }
+   }
+
+   /* Update culling direction which changes depending on the
+    * orientation of the buffer:
+    */
+   if (ctx->Driver.FrontFace)
+      ctx->Driver.FrontFace(ctx, ctx->Polygon.FrontFace);
+   else
+      ctx->NewState |= _NEW_POLYGON;
+
+   if (!colorRegions[0]) {
+      FALLBACK(intel, INTEL_FALLBACK_DRAW_BUFFER, GL_TRUE);
+   }
+   else {
+      FALLBACK(intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE);
+   }
+
+   /***
+    *** Get depth buffer region and check if we need a software fallback.
+    *** Note that the depth buffer is usually a DEPTH_STENCIL buffer.
+    ***/
+   if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped) {
+      irbDepth = intel_renderbuffer(fb->_DepthBuffer->Wrapped);
+      if (irbDepth && irbDepth->region) {
+         FALLBACK(intel, INTEL_FALLBACK_DEPTH_BUFFER, GL_FALSE);
+         depthRegion = irbDepth->region;
+      }
+      else {
+         FALLBACK(intel, INTEL_FALLBACK_DEPTH_BUFFER, GL_TRUE);
+         depthRegion = NULL;
+      }
+   }
+   else {
+      /* not using depth buffer */
+      FALLBACK(intel, INTEL_FALLBACK_DEPTH_BUFFER, GL_FALSE);
+      depthRegion = NULL;
+   }
+
+   /***
+    *** Stencil buffer
+    *** This can only be hardware accelerated if we're using a
+    *** combined DEPTH_STENCIL buffer (for now anyway).
+    ***/
+   if (fb->_StencilBuffer && fb->_StencilBuffer->Wrapped) {
+      irbStencil = intel_renderbuffer(fb->_StencilBuffer->Wrapped);
+      if (irbStencil && irbStencil->region) {
+         ASSERT(irbStencil->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+         FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
+         /* need to re-compute stencil hw state */
+	 if (ctx->Driver.Enable != NULL)
+	    ctx->Driver.Enable(ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled);
+	 else
+	    ctx->NewState |= _NEW_STENCIL;
+         if (!depthRegion)
+            depthRegion = irbStencil->region;
+      }
+      else {
+         FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_TRUE);
+      }
+   }
+   else {
+      /* XXX FBO: instead of FALSE, pass ctx->Stencil.Enabled ??? */
+      FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
+      /* need to re-compute stencil hw state */
+      if (ctx->Driver.Enable != NULL)
+	 ctx->Driver.Enable(ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled);
+      else
+	 ctx->NewState |= _NEW_STENCIL;
+   }
+
+   /*
+    * Update depth test state
+    */
+   if (ctx->Driver.Enable) {
+      if (ctx->Depth.Test && fb->Visual.depthBits > 0) {
+	 ctx->Driver.Enable(ctx, GL_DEPTH_TEST, GL_TRUE);
+      } else {
+	 ctx->Driver.Enable(ctx, GL_DEPTH_TEST, GL_FALSE);
+      }
+   } else {
+      ctx->NewState |= _NEW_DEPTH;
+   }
+
+   intel->vtbl.set_draw_region(intel, colorRegions, depthRegion, 
+	fb->_NumColorDrawBuffers);
+
+   /* update viewport since it depends on window size */
+   if (ctx->Driver.Viewport) {
+      ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
+			   ctx->Viewport.Width, ctx->Viewport.Height);
+   } else {
+      ctx->NewState |= _NEW_VIEWPORT;
+   }
+
+   /* Set state we know depends on drawable parameters:
+    */
+   if (ctx->Driver.Scissor)
+      ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			  ctx->Scissor.Width, ctx->Scissor.Height);
+   intel->NewGLState |= _NEW_SCISSOR;
+
+   if (ctx->Driver.DepthRange)
+      ctx->Driver.DepthRange(ctx,
+			     ctx->Viewport.Near,
+			     ctx->Viewport.Far);
+}
+
+
+static void
+intelDrawBuffer(GLcontext * ctx, GLenum mode)
+{
+   intel_draw_buffer(ctx, ctx->DrawBuffer);
+}
+
+
+static void
+intelReadBuffer(GLcontext * ctx, GLenum mode)
+{
+   if (ctx->ReadBuffer == ctx->DrawBuffer) {
+      /* This will update FBO completeness status.
+       * A framebuffer will be incomplete if the GL_READ_BUFFER setting
+       * refers to a missing renderbuffer.  Calling glReadBuffer can set
+       * that straight and can make the drawing buffer complete.
+       */
+      intel_draw_buffer(ctx, ctx->DrawBuffer);
+   }
+   /* Generally, functions which read pixels (glReadPixels, glCopyPixels, etc)
+    * reference ctx->ReadBuffer and do appropriate state checks.
+    */
+}
+
+
+void
+intelInitBufferFuncs(struct dd_function_table *functions)
+{
+   functions->Clear = intelClear;
+   functions->DrawBuffer = intelDrawBuffer;
+   functions->ReadBuffer = intelReadBuffer;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.h b/src/mesa/drivers/dri/intel/intel_buffers.h
new file mode 100644
index 00000000000..e5afb37dd1a
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_buffers.h
@@ -0,0 +1,63 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_BUFFERS_H
+#define INTEL_BUFFERS_H
+
+#include "dri_util.h"
+#include "drm.h"
+
+struct intel_context;
+struct intel_framebuffer;
+
+
+extern GLboolean
+intel_intersect_cliprects(drm_clip_rect_t * dest,
+                          const drm_clip_rect_t * a,
+                          const drm_clip_rect_t * b);
+
+extern struct intel_region *intel_readbuf_region(struct intel_context *intel);
+
+extern struct intel_region *intel_drawbuf_region(struct intel_context *intel);
+
+extern void intel_wait_flips(struct intel_context *intel);
+
+extern void intelSwapBuffers(__DRIdrawablePrivate * dPriv);
+
+extern void intelWindowMoved(struct intel_context *intel);
+
+extern void intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb);
+
+extern void intelInitBufferFuncs(struct dd_function_table *functions);
+
+void intel_get_cliprects(struct intel_context *intel,
+			 struct drm_clip_rect **cliprects,
+			 unsigned int *num_cliprects,
+			 int *x_off, int *y_off);
+
+#endif /* INTEL_BUFFERS_H */
diff --git a/src/mesa/drivers/dri/intel/intel_chipset.h b/src/mesa/drivers/dri/intel/intel_chipset.h
new file mode 100644
index 00000000000..d1b4941601e
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_chipset.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+#define PCI_CHIP_I810			0x7121
+#define PCI_CHIP_I810_DC100		0x7123
+#define PCI_CHIP_I810_E			0x7125
+#define PCI_CHIP_I815			0x1132
+
+#define PCI_CHIP_I830_M			0x3577
+#define PCI_CHIP_845_G			0x2562
+#define PCI_CHIP_I855_GM		0x3582
+#define PCI_CHIP_I865_G			0x2572
+
+#define PCI_CHIP_I915_G			0x2582
+#define PCI_CHIP_E7221_G		0x258A
+#define PCI_CHIP_I915_GM		0x2592
+#define PCI_CHIP_I945_G			0x2772
+#define PCI_CHIP_I945_GM		0x27A2
+#define PCI_CHIP_I945_GME		0x27AE
+
+#define PCI_CHIP_Q35_G			0x29B2
+#define PCI_CHIP_G33_G			0x29C2
+#define PCI_CHIP_Q33_G			0x29D2
+
+#define PCI_CHIP_I965_G			0x29A2
+#define PCI_CHIP_I965_Q			0x2992
+#define PCI_CHIP_I965_G_1		0x2982
+#define PCI_CHIP_I946_GZ		0x2972
+#define PCI_CHIP_I965_GM                0x2A02
+#define PCI_CHIP_I965_GME               0x2A12
+
+#define PCI_CHIP_GM45_GM                0x2A42
+
+#define PCI_CHIP_IGD_E_G                0x2E02
+#define PCI_CHIP_Q45_G                  0x2E12
+#define PCI_CHIP_G45_G                  0x2E22
+#define PCI_CHIP_G41_G                  0x2E32
+
+#define IS_MOBILE(devid)	(devid == PCI_CHIP_I855_GM || \
+				 devid == PCI_CHIP_I915_GM || \
+				 devid == PCI_CHIP_I945_GM || \
+				 devid == PCI_CHIP_I945_GME || \
+				 devid == PCI_CHIP_I965_GM || \
+				 devid == PCI_CHIP_I965_GME || \
+				 devid == PCI_CHIP_GM45_GM)
+
+#define IS_G45(devid)           (devid == PCI_CHIP_IGD_E_G || \
+                                 devid == PCI_CHIP_Q45_G || \
+                                 devid == PCI_CHIP_G45_G || \
+                                 devid == PCI_CHIP_G41_G)
+#define IS_GM45(devid)          (devid == PCI_CHIP_GM45_GM)
+#define IS_G4X(devid)		(IS_G45(devid) || IS_GM45(devid))
+
+#define IS_915(devid)		(devid == PCI_CHIP_I915_G || \
+				 devid == PCI_CHIP_E7221_G || \
+				 devid == PCI_CHIP_I915_GM)
+
+#define IS_945(devid)		(devid == PCI_CHIP_I945_G || \
+				 devid == PCI_CHIP_I945_GM || \
+				 devid == PCI_CHIP_I945_GME || \
+				 devid == PCI_CHIP_G33_G || \
+				 devid == PCI_CHIP_Q33_G || \
+				 devid == PCI_CHIP_Q35_G)
+
+#define IS_965(devid)		(devid == PCI_CHIP_I965_G || \
+				 devid == PCI_CHIP_I965_Q || \
+				 devid == PCI_CHIP_I965_G_1 || \
+				 devid == PCI_CHIP_I965_GM || \
+				 devid == PCI_CHIP_I965_GME || \
+				 devid == PCI_CHIP_I946_GZ || \
+				 IS_G4X(devid))
+
+#define IS_9XX(devid)		(IS_915(devid) || \
+				 IS_945(devid) || \
+				 IS_965(devid))
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
new file mode 100644
index 00000000000..6c625b428c0
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -0,0 +1,1022 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/matrix.h"
+#include "main/simple_list.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/imports.h"
+#include "main/points.h"
+
+#include "swrast/swrast.h"
+#include "swrast_setup/swrast_setup.h"
+#include "tnl/tnl.h"
+
+#include "tnl/t_pipeline.h"
+#include "tnl/t_vertex.h"
+
+#include "drivers/common/driverfuncs.h"
+
+#include "intel_screen.h"
+
+#include "i830_dri.h"
+
+#include "intel_chipset.h"
+#include "intel_buffers.h"
+#include "intel_tex.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_pixel.h"
+#include "intel_regions.h"
+#include "intel_buffer_objects.h"
+#include "intel_fbo.h"
+#include "intel_decode.h"
+#include "intel_bufmgr.h"
+
+#include "drirenderbuffer.h"
+#include "vblank.h"
+#include "utils.h"
+#include "xmlpool.h"            /* for symbolic values of enum-type options */
+#ifndef INTEL_DEBUG
+int INTEL_DEBUG = (0);
+#endif
+
+#define need_GL_ARB_multisample
+#define need_GL_ARB_occlusion_query
+#define need_GL_ARB_point_parameters
+#define need_GL_ARB_shader_objects
+#define need_GL_ARB_texture_compression
+#define need_GL_ARB_vertex_buffer_object
+#define need_GL_ARB_vertex_program
+#define need_GL_ARB_vertex_shader
+#define need_GL_ARB_window_pos
+#define need_GL_EXT_blend_color
+#define need_GL_EXT_blend_equation_separate
+#define need_GL_EXT_blend_func_separate
+#define need_GL_EXT_blend_minmax
+#define need_GL_EXT_cull_vertex
+#define need_GL_EXT_fog_coord
+#define need_GL_EXT_framebuffer_object
+#define need_GL_EXT_multi_draw_arrays
+#define need_GL_EXT_point_parameters
+#define need_GL_EXT_secondary_color
+#define need_GL_ATI_separate_stencil
+#define need_GL_NV_point_sprite
+#define need_GL_NV_vertex_program
+#define need_GL_VERSION_2_0
+#define need_GL_VERSION_2_1
+
+#include "extension_helper.h"
+
+#define DRIVER_DATE                     "20080716"
+#define DRIVER_DATE_GEM                 "GEM " DRIVER_DATE
+
+static const GLubyte *
+intelGetString(GLcontext * ctx, GLenum name)
+{
+   const struct intel_context *const intel = intel_context(ctx);
+   const char *chipset;
+   static char buffer[128];
+
+   switch (name) {
+   case GL_VENDOR:
+      return (GLubyte *) "Tungsten Graphics, Inc";
+      break;
+
+   case GL_RENDERER:
+      switch (intel->intelScreen->deviceID) {
+      case PCI_CHIP_845_G:
+         chipset = "Intel(R) 845G";
+         break;
+      case PCI_CHIP_I830_M:
+         chipset = "Intel(R) 830M";
+         break;
+      case PCI_CHIP_I855_GM:
+         chipset = "Intel(R) 852GM/855GM";
+         break;
+      case PCI_CHIP_I865_G:
+         chipset = "Intel(R) 865G";
+         break;
+      case PCI_CHIP_I915_G:
+         chipset = "Intel(R) 915G";
+         break;
+      case PCI_CHIP_E7221_G:
+	 chipset = "Intel (R) E7221G (i915)";
+	 break;
+      case PCI_CHIP_I915_GM:
+         chipset = "Intel(R) 915GM";
+         break;
+      case PCI_CHIP_I945_G:
+         chipset = "Intel(R) 945G";
+         break;
+      case PCI_CHIP_I945_GM:
+         chipset = "Intel(R) 945GM";
+         break;
+      case PCI_CHIP_I945_GME:
+         chipset = "Intel(R) 945GME";
+         break;
+      case PCI_CHIP_G33_G:
+	 chipset = "Intel(R) G33";
+	 break;
+      case PCI_CHIP_Q35_G:
+	 chipset = "Intel(R) Q35";
+	 break;
+      case PCI_CHIP_Q33_G:
+	 chipset = "Intel(R) Q33";
+	 break;
+      case PCI_CHIP_I965_Q:
+	 chipset = "Intel(R) 965Q";
+	 break;
+      case PCI_CHIP_I965_G:
+      case PCI_CHIP_I965_G_1:
+	 chipset = "Intel(R) 965G";
+	 break;
+      case PCI_CHIP_I946_GZ:
+	 chipset = "Intel(R) 946GZ";
+	 break;
+      case PCI_CHIP_I965_GM:
+	 chipset = "Intel(R) 965GM";
+	 break;
+      case PCI_CHIP_I965_GME:
+	 chipset = "Intel(R) 965GME/GLE";
+	 break;
+      case PCI_CHIP_GM45_GM:
+	 chipset = "Mobile Intel® GM45 Express Chipset";
+	 break; 
+      case PCI_CHIP_IGD_E_G:
+	 chipset = "Intel(R) Integrated Graphics Device";
+	 break;
+      case PCI_CHIP_G45_G:
+         chipset = "Intel(R) G45/G43";
+         break;
+      case PCI_CHIP_Q45_G:
+         chipset = "Intel(R) Q45/Q43";
+         break;
+      case PCI_CHIP_G41_G:
+         chipset = "Intel(R) G41";
+         break;
+      default:
+         chipset = "Unknown Intel Chipset";
+         break;
+      }
+
+      (void) driGetRendererString(buffer, chipset, 
+				  (intel->ttm) ? DRIVER_DATE_GEM : DRIVER_DATE,
+				  0);
+      return (GLubyte *) buffer;
+
+   default:
+      return NULL;
+   }
+}
+
+void
+intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
+{
+   struct intel_framebuffer *intel_fb = drawable->driverPrivate;
+   struct intel_renderbuffer *rb;
+   struct intel_region *region, *depth_region;
+   struct intel_context *intel = context->driverPrivate;
+   __DRIbuffer *buffers;
+   __DRIscreen *screen;
+   int i, count;
+   unsigned int attachments[10];
+   uint32_t name;
+   const char *region_name;
+
+   if (INTEL_DEBUG & DEBUG_DRI)
+      fprintf(stderr, "enter %s, drawable %p\n", __func__, drawable);
+
+   screen = intel->intelScreen->driScrnPriv;
+
+   i = 0;
+   if (intel_fb->color_rb[0])
+      attachments[i++] = __DRI_BUFFER_FRONT_LEFT;
+   if (intel_fb->color_rb[1])
+      attachments[i++] = __DRI_BUFFER_BACK_LEFT;
+   if (intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH))
+      attachments[i++] = __DRI_BUFFER_DEPTH;
+   if (intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL))
+      attachments[i++] = __DRI_BUFFER_STENCIL;
+
+   buffers = (*screen->dri2.loader->getBuffers)(drawable,
+						&drawable->w,
+						&drawable->h,
+						attachments, i,
+						&count,
+						drawable->loaderPrivate);
+
+   if (buffers == NULL)
+      return;
+
+   drawable->x = 0;
+   drawable->y = 0;
+   drawable->backX = 0;
+   drawable->backY = 0;
+   drawable->numClipRects = 1;
+   drawable->pClipRects[0].x1 = 0;
+   drawable->pClipRects[0].y1 = 0;
+   drawable->pClipRects[0].x2 = drawable->w;
+   drawable->pClipRects[0].y2 = drawable->h;
+   drawable->numBackClipRects = 1;
+   drawable->pBackClipRects[0].x1 = 0;
+   drawable->pBackClipRects[0].y1 = 0;
+   drawable->pBackClipRects[0].x2 = drawable->w;
+   drawable->pBackClipRects[0].y2 = drawable->h;
+
+   depth_region = NULL;
+   for (i = 0; i < count; i++) {
+       switch (buffers[i].attachment) {
+       case __DRI_BUFFER_FRONT_LEFT:
+	   rb = intel_fb->color_rb[0];
+	   region_name = "dri2 front buffer";
+	   break;
+
+       case __DRI_BUFFER_BACK_LEFT:
+	   rb = intel_fb->color_rb[1];
+	   region_name = "dri2 back buffer";
+	   break;
+
+       case __DRI_BUFFER_DEPTH:
+	   rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
+	   region_name = "dri2 depth buffer";
+	   break;
+
+       case __DRI_BUFFER_STENCIL:
+	   rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
+	   region_name = "dri2 stencil buffer";
+	   break;
+
+       case __DRI_BUFFER_ACCUM:
+       default:
+	   fprintf(stderr,
+		   "unhandled buffer attach event, attacment type %d\n",
+		   buffers[i].attachment);
+	   return;
+       }
+
+       if (rb->region) {
+	  dri_bo_flink(rb->region->buffer, &name);
+	  if (name == buffers[i].name)
+	     continue;
+       }
+
+       if (INTEL_DEBUG & DEBUG_DRI)
+	  fprintf(stderr,
+		  "attaching buffer %d, at %d, cpp %d, pitch %d\n",
+		  buffers[i].name, buffers[i].attachment,
+		  buffers[i].cpp, buffers[i].pitch);
+       
+       if (buffers[i].attachment == __DRI_BUFFER_STENCIL && depth_region) {
+	  if (INTEL_DEBUG & DEBUG_DRI)
+	     fprintf(stderr, "(reusing depth buffer as stencil)\n");
+	  intel_region_reference(&region, depth_region);
+       }
+       else
+          region = intel_region_alloc_for_handle(intel, buffers[i].cpp,
+						 drawable->w,
+						 drawable->h,
+						 buffers[i].pitch / buffers[i].cpp,
+						 buffers[i].name,
+						 region_name);
+
+       if (buffers[i].attachment == __DRI_BUFFER_DEPTH)
+	  depth_region = region;
+
+       intel_renderbuffer_set_region(rb, region);
+       intel_region_release(&region);
+   }
+
+   driUpdateFramebufferSize(&intel->ctx, drawable);
+}
+
+void
+intel_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+{
+    struct intel_context *intel = intel_context(ctx);
+    __DRIcontext *driContext = intel->driContext;
+    void (*old_viewport)(GLcontext *ctx, GLint x, GLint y,
+			 GLsizei w, GLsizei h);
+
+    if (!driContext->driScreenPriv->dri2.enabled)
+	return;
+
+    intel_update_renderbuffers(driContext, driContext->driDrawablePriv);
+    if (driContext->driDrawablePriv != driContext->driReadablePriv)
+	intel_update_renderbuffers(driContext, driContext->driReadablePriv);
+
+    old_viewport = ctx->Driver.Viewport;
+    ctx->Driver.Viewport = NULL;
+    intel->driDrawable = driContext->driDrawablePriv;
+    intelWindowMoved(intel);
+    intel_draw_buffer(ctx, intel->ctx.DrawBuffer);
+    ctx->Driver.Viewport = old_viewport;
+}
+
+/**
+ * Extension strings exported by the intel driver.
+ *
+ * Extensions supported by all chips supported by i830_dri, i915_dri, or
+ * i965_dri.
+ */
+static const struct dri_extension card_extensions[] = {
+   { "GL_ARB_multisample",                GL_ARB_multisample_functions },
+   { "GL_ARB_multitexture",               NULL },
+   { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
+   { "GL_ARB_texture_border_clamp",       NULL },
+   { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
+   { "GL_ARB_texture_cube_map",           NULL },
+   { "GL_ARB_texture_env_add",            NULL },
+   { "GL_ARB_texture_env_combine",        NULL },
+   { "GL_ARB_texture_env_crossbar",       NULL },
+   { "GL_ARB_texture_env_dot3",           NULL },
+   { "GL_ARB_texture_mirrored_repeat",    NULL },
+   { "GL_ARB_texture_rectangle",          NULL },
+   { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
+   { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
+   { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
+   { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
+   { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
+   { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
+   { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
+   { "GL_EXT_blend_logic_op",             NULL },
+   { "GL_EXT_blend_subtract",             NULL },
+   { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
+   { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
+   { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
+   { "GL_EXT_packed_depth_stencil",       NULL },
+   { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
+   { "GL_EXT_stencil_wrap",               NULL },
+   { "GL_EXT_texture_edge_clamp",         NULL },
+   { "GL_EXT_texture_env_combine",        NULL },
+   { "GL_EXT_texture_env_dot3",           NULL },
+   { "GL_EXT_texture_filter_anisotropic", NULL },
+   { "GL_EXT_texture_lod_bias",           NULL },
+   { "GL_3DFX_texture_compression_FXT1",  NULL },
+   { "GL_APPLE_client_storage",           NULL },
+   { "GL_MESA_pack_invert",               NULL },
+   { "GL_MESA_ycbcr_texture",             NULL },
+   { "GL_NV_blend_square",                NULL },
+   { "GL_NV_point_sprite",                GL_NV_point_sprite_functions },
+   { "GL_NV_vertex_program",              GL_NV_vertex_program_functions },
+   { "GL_NV_vertex_program1_1",           NULL },
+   { "GL_SGIS_generate_mipmap",           NULL },
+   { NULL, NULL }
+};
+
+static const struct dri_extension brw_extensions[] = {
+   { "GL_ARB_depth_texture",              NULL },
+   { "GL_ARB_draw_buffers",               NULL },
+   { "GL_ARB_fragment_program",           NULL },
+   { "GL_ARB_fragment_program_shadow",    NULL },
+   { "GL_ARB_fragment_shader",            NULL },
+   { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
+   { "GL_ARB_point_sprite", 		  NULL },
+   { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
+   { "GL_ARB_shading_language_100",       GL_VERSION_2_0_functions },
+#if 0
+   /* Support for GLSL 1.20 is currently broken in core Mesa.
+    */
+   { "GL_ARB_shading_language_120",       GL_VERSION_2_1_functions },
+#endif
+   { "GL_ARB_shadow",                     NULL },
+   { "GL_ARB_texture_non_power_of_two",   NULL },
+   { "GL_ARB_vertex_shader",              GL_ARB_vertex_shader_functions },
+   { "GL_EXT_shadow_funcs",               NULL },
+   { "GL_EXT_texture_sRGB",		  NULL },
+   { "GL_ATI_separate_stencil",           GL_ATI_separate_stencil_functions },
+   { NULL,                                NULL }
+};
+
+static const struct dri_extension arb_oq_extensions[] = {
+   { NULL, NULL }
+};
+
+static const struct dri_extension ttm_extensions[] = {
+   { "GL_ARB_pixel_buffer_object",        NULL },
+   { "GL_EXT_framebuffer_object",         GL_EXT_framebuffer_object_functions },
+   { NULL, NULL }
+};
+
+/**
+ * Initializes potential list of extensions if ctx == NULL, or actually enables
+ * extensions for a context.
+ */
+void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
+{
+   struct intel_context *intel = ctx?intel_context(ctx):NULL;
+
+   /* Disable imaging extension until convolution is working in teximage paths.
+    */
+   enable_imaging = GL_FALSE;
+
+   driInitExtensions(ctx, card_extensions, enable_imaging);
+
+   if (intel == NULL || intel->ttm)
+      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
+
+   if (intel == NULL || IS_965(intel->intelScreen->deviceID))
+      driInitExtensions(ctx, brw_extensions, GL_FALSE);
+}
+
+static const struct dri_debug_control debug_control[] = {
+   { "tex",   DEBUG_TEXTURE},
+   { "state", DEBUG_STATE},
+   { "ioctl", DEBUG_IOCTL},
+   { "blit",  DEBUG_BLIT},
+   { "mip",   DEBUG_MIPTREE},
+   { "fall",  DEBUG_FALLBACKS},
+   { "verb",  DEBUG_VERBOSE},
+   { "bat",   DEBUG_BATCH},
+   { "pix",   DEBUG_PIXEL},
+   { "buf",   DEBUG_BUFMGR},
+   { "reg",   DEBUG_REGION},
+   { "fbo",   DEBUG_FBO},
+   { "lock",  DEBUG_LOCK},
+   { "sync",  DEBUG_SYNC},
+   { "prim",  DEBUG_PRIMS },
+   { "vert",  DEBUG_VERTS },
+   { "dri",   DEBUG_DRI },
+   { "dma",   DEBUG_DMA },
+   { "san",   DEBUG_SANITY },
+   { "sleep", DEBUG_SLEEP },
+   { "stats", DEBUG_STATS },
+   { "tile",  DEBUG_TILE },
+   { "sing",  DEBUG_SINGLE_THREAD },
+   { "thre",  DEBUG_SINGLE_THREAD },
+   { "wm",    DEBUG_WM },
+   { "urb",   DEBUG_URB },
+   { "vs",    DEBUG_VS },
+   { NULL,    0 }
+};
+
+
+static void
+intelInvalidateState(GLcontext * ctx, GLuint new_state)
+{
+    struct intel_context *intel = intel_context(ctx);
+
+   _swrast_InvalidateState(ctx, new_state);
+   _swsetup_InvalidateState(ctx, new_state);
+   _vbo_InvalidateState(ctx, new_state);
+   _tnl_InvalidateState(ctx, new_state);
+   _tnl_invalidate_vertex_state(ctx, new_state);
+
+   intel->NewGLState |= new_state;
+
+   if (intel->vtbl.invalidate_state)
+      intel->vtbl.invalidate_state( intel, new_state );
+}
+
+
+void
+intelFlush(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+
+   if (intel->Fallback)
+      _swrast_flush(ctx);
+
+   if (!IS_965(intel->intelScreen->deviceID))
+      INTEL_FIREVERTICES(intel);
+
+   /* Emit a flush so that any frontbuffer rendering that might have occurred
+    * lands onscreen in a timely manner, even if the X Server doesn't trigger
+    * a flush for us.
+    */
+   intel_batchbuffer_emit_mi_flush(intel->batch);
+
+   if (intel->batch->map != intel->batch->ptr)
+      intel_batchbuffer_flush(intel->batch);
+}
+
+void
+intelFinish(GLcontext * ctx)
+{
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   int i;
+
+   intelFlush(ctx);
+
+   for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
+       struct intel_renderbuffer *irb;
+
+       irb = intel_renderbuffer(fb->_ColorDrawBuffers[i]);
+
+       if (irb->region)
+	  dri_bo_wait_rendering(irb->region->buffer);
+   }
+   if (fb->_DepthBuffer) {
+      /* XXX: Wait on buffer idle */
+   }
+}
+
+void
+intelInitDriverFunctions(struct dd_function_table *functions)
+{
+   _mesa_init_driver_functions(functions);
+
+   functions->Flush = intelFlush;
+   functions->Finish = intelFinish;
+   functions->GetString = intelGetString;
+   functions->UpdateState = intelInvalidateState;
+
+   functions->CopyColorTable = _swrast_CopyColorTable;
+   functions->CopyColorSubTable = _swrast_CopyColorSubTable;
+   functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
+   functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
+
+   intelInitTextureFuncs(functions);
+   intelInitStateFuncs(functions);
+   intelInitBufferFuncs(functions);
+   intelInitPixelFuncs(functions);
+}
+
+
+GLboolean
+intelInitContext(struct intel_context *intel,
+                 const __GLcontextModes * mesaVis,
+                 __DRIcontextPrivate * driContextPriv,
+                 void *sharedContextPrivate,
+                 struct dd_function_table *functions)
+{
+   GLcontext *ctx = &intel->ctx;
+   GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
+   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+   int fthrottle_mode;
+
+   if (!_mesa_initialize_context(&intel->ctx, mesaVis, shareCtx,
+                                 functions, (void *) intel)) {
+      _mesa_printf("%s: failed to init mesa context\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   driContextPriv->driverPrivate = intel;
+   intel->intelScreen = intelScreen;
+   intel->driScreen = sPriv;
+   intel->sarea = intelScreen->sarea;
+   intel->driContext = driContextPriv;
+
+   /* Dri stuff */
+   intel->hHWContext = driContextPriv->hHWContext;
+   intel->driFd = sPriv->fd;
+   intel->driHwLock = sPriv->lock;
+
+   driParseConfigFiles(&intel->optionCache, &intelScreen->optionCache,
+                       intel->driScreen->myNum,
+		       IS_965(intelScreen->deviceID) ? "i965" : "i915");
+   if (intelScreen->deviceID == PCI_CHIP_I865_G)
+      intel->maxBatchSize = 4096;
+   else
+      intel->maxBatchSize = BATCH_SZ;
+
+   intel->bufmgr = intelScreen->bufmgr;
+   intel->ttm = intelScreen->ttm;
+   if (intel->ttm) {
+      int bo_reuse_mode;
+
+      bo_reuse_mode = driQueryOptioni(&intel->optionCache, "bo_reuse");
+      switch (bo_reuse_mode) {
+      case DRI_CONF_BO_REUSE_DISABLED:
+	 break;
+      case DRI_CONF_BO_REUSE_ALL:
+	 intel_bufmgr_gem_enable_reuse(intel->bufmgr);
+	 break;
+      }
+   }
+
+   ctx->Const.MaxTextureMaxAnisotropy = 2.0;
+
+   /* This doesn't yet catch all non-conformant rendering, but it's a
+    * start.
+    */
+   if (getenv("INTEL_STRICT_CONFORMANCE")) {
+      intel->strict_conformance = 1;
+   }
+
+   if (intel->strict_conformance) {
+      ctx->Const.MinLineWidth = 1.0;
+      ctx->Const.MinLineWidthAA = 1.0;
+      ctx->Const.MaxLineWidth = 1.0;
+      ctx->Const.MaxLineWidthAA = 1.0;
+      ctx->Const.LineWidthGranularity = 1.0;
+   }
+   else {
+      ctx->Const.MinLineWidth = 1.0;
+      ctx->Const.MinLineWidthAA = 1.0;
+      ctx->Const.MaxLineWidth = 5.0;
+      ctx->Const.MaxLineWidthAA = 5.0;
+      ctx->Const.LineWidthGranularity = 0.5;
+   }
+
+   ctx->Const.MinPointSize = 1.0;
+   ctx->Const.MinPointSizeAA = 1.0;
+   ctx->Const.MaxPointSize = 255.0;
+   ctx->Const.MaxPointSizeAA = 3.0;
+   ctx->Const.PointSizeGranularity = 1.0;
+
+   /* reinitialize the context point state.
+    * It depend on constants in __GLcontextRec::Const
+    */
+   _mesa_init_point(ctx);
+
+   ctx->Const.MaxColorAttachments = 4;  /* XXX FBO: review this */
+
+   /* Initialize the software rasterizer and helper modules. */
+   _swrast_CreateContext(ctx);
+   _vbo_CreateContext(ctx);
+   _tnl_CreateContext(ctx);
+   _swsetup_CreateContext(ctx);
+ 
+   /* Configure swrast to match hardware characteristics: */
+   _swrast_allow_pixel_fog(ctx, GL_FALSE);
+   _swrast_allow_vertex_fog(ctx, GL_TRUE);
+
+   intel->hw_stencil = mesaVis->stencilBits && mesaVis->depthBits == 24;
+   intel->hw_stipple = 1;
+
+   /* XXX FBO: this doesn't seem to be used anywhere */
+   switch (mesaVis->depthBits) {
+   case 0:                     /* what to do in this case? */
+   case 16:
+      intel->polygon_offset_scale = 1.0;
+      break;
+   case 24:
+      intel->polygon_offset_scale = 2.0;     /* req'd to pass glean */
+      break;
+   default:
+      assert(0);
+      break;
+   }
+
+   if (IS_965(intelScreen->deviceID))
+      intel->polygon_offset_scale /= 0xffff;
+
+   intel->RenderIndex = ~0;
+
+   fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
+   intel->irqsEmitted = 0;
+
+   intel->do_irqs = (intel->intelScreen->irq_active &&
+                     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
+
+   intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
+
+   _math_matrix_ctr(&intel->ViewportMatrix);
+
+   if (IS_965(intelScreen->deviceID) && !intel->intelScreen->irq_active) {
+      _mesa_printf("IRQs not active.  Exiting\n");
+      exit(1);
+   }
+
+   intelInitExtensions(ctx, GL_FALSE);
+
+   INTEL_DEBUG = driParseDebugString(getenv("INTEL_DEBUG"), debug_control);
+   if (INTEL_DEBUG & DEBUG_BUFMGR)
+      dri_bufmgr_set_debug(intel->bufmgr, GL_TRUE);
+
+   if (!sPriv->dri2.enabled)
+      intel_recreate_static_regions(intel);
+
+   intel->batch = intel_batchbuffer_alloc(intel);
+
+   intel_bufferobj_init(intel);
+   intel_fbo_init(intel);
+
+   if (intel->ctx.Mesa_DXTn) {
+      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+      _mesa_enable_extension(ctx, "GL_S3_s3tc");
+   }
+   else if (driQueryOptionb(&intel->optionCache, "force_s3tc_enable")) {
+      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+   }
+
+   intel->prim.primitive = ~0;
+
+   /* Force all software fallbacks */
+   if (driQueryOptionb(&intel->optionCache, "no_rast")) {
+      fprintf(stderr, "disabling 3D rasterization\n");
+      intel->no_rast = 1;
+   }
+
+   /* Disable all hardware rendering (skip emitting batches and fences/waits
+    * to the kernel)
+    */
+   intel->no_hw = getenv("INTEL_NO_HW") != NULL;
+
+   return GL_TRUE;
+}
+
+void
+intelDestroyContext(__DRIcontextPrivate * driContextPriv)
+{
+   struct intel_context *intel =
+      (struct intel_context *) driContextPriv->driverPrivate;
+
+   assert(intel);               /* should never be null */
+   if (intel) {
+      GLboolean release_texture_heaps;
+
+      INTEL_FIREVERTICES(intel);
+
+      intel->vtbl.destroy(intel);
+
+      release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
+      _swsetup_DestroyContext(&intel->ctx);
+      _tnl_DestroyContext(&intel->ctx);
+      _vbo_DestroyContext(&intel->ctx);
+
+      _swrast_DestroyContext(&intel->ctx);
+      intel->Fallback = 0;      /* don't call _swrast_Flush later */
+
+      intel_batchbuffer_free(intel->batch);
+      intel->batch = NULL;
+
+      free(intel->prim.vb);
+      intel->prim.vb = NULL;
+      dri_bo_unreference(intel->prim.vb_bo);
+      intel->prim.vb_bo = NULL;
+
+      if (release_texture_heaps) {
+         /* This share group is about to go away, free our private
+          * texture object data.
+          */
+         if (INTEL_DEBUG & DEBUG_TEXTURE)
+            fprintf(stderr, "do something to free texture heaps\n");
+      }
+
+      intel_region_release(&intel->front_region);
+      intel_region_release(&intel->back_region);
+      intel_region_release(&intel->third_region);
+      intel_region_release(&intel->depth_region);
+
+      driDestroyOptionCache(&intel->optionCache);
+
+      /* free the Mesa context */
+      _mesa_free_context_data(&intel->ctx);
+   }
+}
+
+GLboolean
+intelUnbindContext(__DRIcontextPrivate * driContextPriv)
+{
+   return GL_TRUE;
+}
+
+GLboolean
+intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
+                 __DRIdrawablePrivate * driDrawPriv,
+                 __DRIdrawablePrivate * driReadPriv)
+{
+   __DRIscreenPrivate *psp = driDrawPriv->driScreenPriv;
+
+   if (driContextPriv) {
+      struct intel_context *intel =
+         (struct intel_context *) driContextPriv->driverPrivate;
+      struct intel_framebuffer *intel_fb =
+	 (struct intel_framebuffer *) driDrawPriv->driverPrivate;
+      GLframebuffer *readFb = (GLframebuffer *) driReadPriv->driverPrivate;
+ 
+      if (driContextPriv->driScreenPriv->dri2.enabled) {     
+          intel_update_renderbuffers(driContextPriv, driDrawPriv);
+          if (driDrawPriv != driReadPriv)
+              intel_update_renderbuffers(driContextPriv, driReadPriv);
+      } else {
+          /* XXX FBO temporary fix-ups! */
+          /* if the renderbuffers don't have regions, init them from the context */
+         struct intel_renderbuffer *irbDepth
+            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
+         struct intel_renderbuffer *irbStencil
+            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
+
+         if (intel_fb->color_rb[0]) {
+	    intel_renderbuffer_set_region(intel_fb->color_rb[0],
+					  intel->front_region);
+         }
+         if (intel_fb->color_rb[1]) {
+	    intel_renderbuffer_set_region(intel_fb->color_rb[1],
+					  intel->back_region);
+         }
+#if 0
+         if (intel_fb->color_rb[2]) {
+	    intel_renderbuffer_set_region(intel_fb->color_rb[2],
+					  intel->third_region);
+         }
+#endif
+         if (irbDepth) {
+	    intel_renderbuffer_set_region(irbDepth, intel->depth_region);
+         }
+         if (irbStencil) {
+	    intel_renderbuffer_set_region(irbStencil, intel->depth_region);
+         }
+      }
+
+      /* set GLframebuffer size to match window, if needed */
+      driUpdateFramebufferSize(&intel->ctx, driDrawPriv);
+
+      if (driReadPriv != driDrawPriv) {
+	 driUpdateFramebufferSize(&intel->ctx, driReadPriv);
+      }
+
+      _mesa_make_current(&intel->ctx, &intel_fb->Base, readFb);
+
+      /* The drawbuffer won't always be updated by _mesa_make_current: 
+       */
+      if (intel->ctx.DrawBuffer == &intel_fb->Base) {
+
+	 if (intel->driReadDrawable != driReadPriv)
+	    intel->driReadDrawable = driReadPriv;
+
+	 if (intel->driDrawable != driDrawPriv) {
+	    if (driDrawPriv->swap_interval == (unsigned)-1) {
+	       int i;
+
+	       driDrawPriv->vblFlags = (intel->intelScreen->irq_active != 0)
+		  ? driGetDefaultVBlankFlags(&intel->optionCache)
+		 : VBLANK_FLAG_NO_IRQ;
+
+	       (*psp->systemTime->getUST) (&intel_fb->swap_ust);
+	       driDrawableInitVBlank(driDrawPriv);
+	       intel_fb->vbl_waited = driDrawPriv->vblSeq;
+
+	       for (i = 0; i < (intel->intelScreen->third.handle ? 3 : 2); i++) {
+		  if (intel_fb->color_rb[i])
+		     intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
+	       }
+	    }
+	    intel->driDrawable = driDrawPriv;
+	    intelWindowMoved(intel);
+	 }
+
+	 intel_draw_buffer(&intel->ctx, &intel_fb->Base);
+      }
+   }
+   else {
+      _mesa_make_current(NULL, NULL, NULL);
+   }
+
+   return GL_TRUE;
+}
+
+static void
+intelContendedLock(struct intel_context *intel, GLuint flags)
+{
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   __DRIscreenPrivate *sPriv = intel->driScreen;
+   volatile struct drm_i915_sarea *sarea = intel->sarea;
+   int me = intel->hHWContext;
+
+   drmGetLock(intel->driFd, intel->hHWContext, flags);
+   intel->locked = 1;
+
+   if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - got contended lock\n", __progname);
+
+   /* If the window moved, may need to set a new cliprect now.
+    *
+    * NOTE: This releases and regains the hw lock, so all state
+    * checking must be done *after* this call:
+    */
+   if (dPriv)
+       DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
+
+   if (sarea && sarea->ctxOwner != me) {
+      if (INTEL_DEBUG & DEBUG_BUFMGR) {
+	 fprintf(stderr, "Lost Context: sarea->ctxOwner %x me %x\n",
+		 sarea->ctxOwner, me);
+      }
+      sarea->ctxOwner = me;
+   }
+
+   /* If the last consumer of the texture memory wasn't us, notify the fake
+    * bufmgr and record the new owner.  We should have the memory shared
+    * between contexts of a single fake bufmgr, but this will at least make
+    * things correct for now.
+    */
+   if (!intel->ttm && sarea->texAge != intel->hHWContext) {
+      sarea->texAge = intel->hHWContext;
+      intel_bufmgr_fake_contended_lock_take(intel->bufmgr);
+      if (INTEL_DEBUG & DEBUG_BATCH)
+	 intel_decode_context_reset();
+      if (INTEL_DEBUG & DEBUG_BUFMGR)
+	 fprintf(stderr, "Lost Textures: sarea->texAge %x hw context %x\n",
+		 sarea->ctxOwner, intel->hHWContext);
+   }
+
+   /* Drawable changed?
+    */
+   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
+       intelWindowMoved(intel);
+       intel->lastStamp = dPriv->lastStamp;
+   }
+}
+
+
+_glthread_DECLARE_STATIC_MUTEX(lockMutex);
+
+/* Lock the hardware and validate our state.  
+ */
+void LOCK_HARDWARE( struct intel_context *intel )
+{
+    __DRIdrawable *dPriv = intel->driDrawable;
+    __DRIscreen *sPriv = intel->driScreen;
+    char __ret = 0;
+    struct intel_framebuffer *intel_fb = NULL;
+    struct intel_renderbuffer *intel_rb = NULL;
+
+    _glthread_LOCK_MUTEX(lockMutex);
+    assert(!intel->locked);
+    intel->locked = 1;
+
+    if (intel->driDrawable) {
+       intel_fb = intel->driDrawable->driverPrivate;
+
+       if (intel_fb)
+	  intel_rb =
+	     intel_get_renderbuffer(&intel_fb->Base,
+				    intel_fb->Base._ColorDrawBufferIndexes[0]);
+    }
+
+    if (intel_rb && dPriv->vblFlags &&
+	!(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ) &&
+	(intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
+	drmVBlank vbl;
+
+	vbl.request.type = DRM_VBLANK_ABSOLUTE;
+
+	if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
+	    vbl.request.type |= DRM_VBLANK_SECONDARY;
+	}
+
+	vbl.request.sequence = intel_rb->vbl_pending;
+	drmWaitVBlank(intel->driFd, &vbl);
+	intel_fb->vbl_waited = vbl.reply.sequence;
+    }
+
+    if (!sPriv->dri2.enabled) {
+	DRM_CAS(intel->driHwLock, intel->hHWContext,
+		(DRM_LOCK_HELD|intel->hHWContext), __ret);
+
+	if (__ret)
+	    intelContendedLock( intel, 0 );
+    }
+
+
+    if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - locked\n", __progname);
+}
+
+
+/* Unlock the hardware using the global current context 
+ */
+void UNLOCK_HARDWARE( struct intel_context *intel )
+{
+    __DRIscreen *sPriv = intel->driScreen;
+
+   intel->vtbl.note_unlock( intel );
+   intel->locked = 0;
+
+   if (!sPriv->dri2.enabled)
+      DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
+
+   _glthread_UNLOCK_MUTEX(lockMutex);
+
+   if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - unlocked\n", __progname);
+
+   /**
+    * Nothing should be left in batch outside of LOCK/UNLOCK which references
+    * cliprects.
+    */
+   if (intel->batch->cliprect_mode == REFERENCES_CLIPRECTS)
+      intel_batchbuffer_flush(intel->batch);
+}
+
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
new file mode 100644
index 00000000000..ee43ed7e833
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -0,0 +1,515 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTELCONTEXT_INC
+#define INTELCONTEXT_INC
+
+
+
+#include "main/mtypes.h"
+#include "main/mm.h"
+#include "texmem.h"
+#include "drm.h"
+#include "intel_bufmgr.h"
+
+#include "intel_screen.h"
+#include "intel_tex_obj.h"
+#include "i915_drm.h"
+#include "tnl/t_vertex.h"
+
+#define TAG(x) intel##x
+#include "tnl_dd/t_dd_vertex.h"
+#undef TAG
+
+#define DV_PF_555  (1<<8)
+#define DV_PF_565  (2<<8)
+#define DV_PF_8888 (3<<8)
+
+struct intel_region;
+struct intel_context;
+
+typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
+                                intelVertex *, intelVertex *);
+typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
+                                 intelVertex *);
+typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
+
+#define INTEL_FALLBACK_DRAW_BUFFER	 0x1
+#define INTEL_FALLBACK_READ_BUFFER	 0x2
+#define INTEL_FALLBACK_DEPTH_BUFFER      0x4
+#define INTEL_FALLBACK_STENCIL_BUFFER    0x8
+#define INTEL_FALLBACK_USER		 0x10
+#define INTEL_FALLBACK_RENDERMODE	 0x20
+#define INTEL_FALLBACK_TEXTURE   	 0x40
+
+extern void intelFallback(struct intel_context *intel, GLuint bit,
+                          GLboolean mode);
+#define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
+
+
+#define INTEL_WRITE_PART  0x1
+#define INTEL_WRITE_FULL  0x2
+#define INTEL_READ        0x4
+
+#define INTEL_MAX_FIXUP 64
+
+struct intel_context
+{
+   GLcontext ctx;               /* the parent class */
+
+   struct
+   {
+      void (*destroy) (struct intel_context * intel);
+      void (*emit_state) (struct intel_context * intel);
+      void (*finish_batch) (struct intel_context * intel);
+      void (*new_batch) (struct intel_context * intel);
+      void (*emit_invarient_state) (struct intel_context * intel);
+      void (*note_fence) (struct intel_context *intel, GLuint fence);
+      void (*note_unlock) (struct intel_context *intel);
+      void (*update_texture_state) (struct intel_context * intel);
+
+      void (*render_start) (struct intel_context * intel);
+      void (*render_prevalidate) (struct intel_context * intel);
+      void (*set_draw_region) (struct intel_context * intel,
+                               struct intel_region * draw_regions[],
+                               struct intel_region * depth_region,
+			       GLuint num_regions);
+
+      GLuint (*flush_cmd) (void);
+      void (*emit_flush) (struct intel_context *intel, GLuint unused);
+
+      void (*reduced_primitive_state) (struct intel_context * intel,
+                                       GLenum rprim);
+
+      GLboolean (*check_vertex_size) (struct intel_context * intel,
+				      GLuint expected);
+      void (*invalidate_state) (struct intel_context *intel,
+				GLuint new_state);
+
+
+      /* Metaops: 
+       */
+      void (*install_meta_state) (struct intel_context * intel);
+      void (*leave_meta_state) (struct intel_context * intel);
+
+      void (*meta_draw_region) (struct intel_context * intel,
+                                struct intel_region * draw_region,
+                                struct intel_region * depth_region);
+
+      void (*meta_draw_quad)(struct intel_context *intel,
+			     GLfloat x0, GLfloat x1,
+			     GLfloat y0, GLfloat y1,
+			     GLfloat z,
+			     GLuint color, /* ARGB32 */
+			     GLfloat s0, GLfloat s1,
+			     GLfloat t0, GLfloat t1);
+
+      void (*meta_color_mask) (struct intel_context * intel, GLboolean);
+
+      void (*meta_stencil_replace) (struct intel_context * intel,
+                                    GLuint mask, GLuint clear);
+
+      void (*meta_depth_replace) (struct intel_context * intel);
+
+      void (*meta_texture_blend_replace) (struct intel_context * intel);
+
+      void (*meta_no_stencil_write) (struct intel_context * intel);
+      void (*meta_no_depth_write) (struct intel_context * intel);
+      void (*meta_no_texture) (struct intel_context * intel);
+
+      void (*meta_import_pixel_state) (struct intel_context * intel);
+      void (*meta_frame_buffer_texture) (struct intel_context *intel,
+					 GLint xoff, GLint yoff);
+
+      GLboolean(*meta_tex_rect_source) (struct intel_context * intel,
+					dri_bo * buffer,
+					GLuint offset,
+					GLuint pitch,
+					GLuint height,
+					GLenum format, GLenum type);
+
+      void (*assert_not_dirty) (struct intel_context *intel);
+
+      void (*debug_batch)(struct intel_context *intel);
+   } vtbl;
+
+   GLint refcount;
+   GLuint Fallback;
+   GLuint NewGLState;
+
+   dri_bufmgr *bufmgr;
+   unsigned int maxBatchSize;
+
+   struct intel_region *front_region;
+   struct intel_region *back_region;
+   struct intel_region *third_region;
+   struct intel_region *depth_region;
+
+   /**
+    * This value indicates that the kernel memory manager is being used
+    * instead of the fake client-side memory manager.
+    */
+   GLboolean ttm;
+
+   struct intel_batchbuffer *batch;
+   GLboolean no_batch_wrap;
+   unsigned batch_id;
+
+   struct
+   {
+      GLuint id;
+      uint32_t primitive;	/**< Current hardware primitive type */
+      void (*flush) (struct intel_context *);
+      GLubyte *start_ptr; /**< for i8xx */
+      dri_bo *vb_bo;
+      uint8_t *vb;
+      unsigned int start_offset; /**< Byte offset of primitive sequence */
+      unsigned int current_offset; /**< Byte offset of next vertex */
+      unsigned int count;	/**< Number of vertices in current primitive */
+   } prim;
+
+   GLuint stats_wm;
+   GLboolean locked;
+   char *prevLockFile;
+   int prevLockLine;
+
+   GLubyte clear_chan[4];
+   GLuint ClearColor565;
+   GLuint ClearColor8888;
+
+   /* Offsets of fields within the current vertex:
+    */
+   GLuint coloroffset;
+   GLuint specoffset;
+   GLuint wpos_offset;
+   GLuint wpos_size;
+
+   struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
+   GLuint vertex_attr_count;
+
+   GLfloat polygon_offset_scale;        /* dependent on depth_scale, bpp */
+
+   GLboolean hw_stencil;
+   GLboolean hw_stipple;
+   GLboolean depth_buffer_is_float;
+   GLboolean no_rast;
+   GLboolean strict_conformance;
+
+   /* State for intelvb.c and inteltris.c.
+    */
+   GLuint RenderIndex;
+   GLmatrix ViewportMatrix;
+   GLenum render_primitive;
+   GLenum reduced_primitive;
+   GLuint vertex_size;
+   GLubyte *verts;              /* points to tnl->clipspace.vertex_buf */
+
+   /* Fallback rasterization functions 
+    */
+   intel_point_func draw_point;
+   intel_line_func draw_line;
+   intel_tri_func draw_tri;
+
+   /* These refer to the current drawing buffer:
+    */
+   struct gl_texture_object *frame_buffer_texobj;
+   /**
+    * Set to true if a single constant cliprect should be used in the
+    * batchbuffer.  Otherwise, cliprects must be calculated at batchbuffer
+    * flush time while the lock is held.
+    */
+   GLboolean constant_cliprect;
+   /**
+    * In !constant_cliprect mode, set to true if the front cliprects should be
+    * used instead of back.
+    */
+   GLboolean front_cliprects;
+   drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
+
+   int perf_boxes;
+
+   GLuint do_usleeps;
+   int do_irqs;
+   GLuint irqsEmitted;
+
+   GLboolean scissor;
+   drm_clip_rect_t draw_rect;
+   drm_clip_rect_t scissor_rect;
+
+   drm_context_t hHWContext;
+   drmLock *driHwLock;
+   int driFd;
+
+   __DRIcontextPrivate *driContext;
+   __DRIdrawablePrivate *driDrawable;
+   __DRIdrawablePrivate *driReadDrawable;
+   __DRIscreenPrivate *driScreen;
+   intelScreenPrivate *intelScreen;
+   volatile struct drm_i915_sarea *sarea;
+
+   GLuint lastStamp;
+
+   GLboolean no_hw;
+
+   /**
+    * Configuration cache
+    */
+   driOptionCache optionCache;
+
+   int64_t swap_ust;
+   int64_t swap_missed_ust;
+
+   GLuint swap_count;
+   GLuint swap_missed_count;
+};
+
+/* These are functions now:
+ */
+void LOCK_HARDWARE( struct intel_context *intel );
+void UNLOCK_HARDWARE( struct intel_context *intel );
+
+extern char *__progname;
+
+
+#define SUBPIXEL_X 0.125
+#define SUBPIXEL_Y 0.125
+
+#define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0]))
+#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
+
+#define INTEL_FIREVERTICES(intel)		\
+do {						\
+   if ((intel)->prim.flush)			\
+      (intel)->prim.flush(intel);		\
+} while (0)
+
+/* ================================================================
+ * Color packing:
+ */
+
+#define INTEL_PACKCOLOR4444(r,g,b,a) \
+  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
+
+#define INTEL_PACKCOLOR1555(r,g,b,a) \
+  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
+    ((a) ? 0x8000 : 0))
+
+#define INTEL_PACKCOLOR565(r,g,b) \
+  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
+
+#define INTEL_PACKCOLOR8888(r,g,b,a) \
+  ((a<<24) | (r<<16) | (g<<8) | b)
+
+#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
+(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
+ (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
+  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
+   0)))
+
+/* ================================================================
+ * From linux kernel i386 header files, copes with odd sizes better
+ * than COPY_DWORDS would:
+ * XXX Put this in src/mesa/main/imports.h ???
+ */
+#if defined(i386) || defined(__i386__)
+static INLINE void * __memcpy(void * to, const void * from, size_t n)
+{
+   int d0, d1, d2;
+   __asm__ __volatile__(
+      "rep ; movsl\n\t"
+      "testb $2,%b4\n\t"
+      "je 1f\n\t"
+      "movsw\n"
+      "1:\ttestb $1,%b4\n\t"
+      "je 2f\n\t"
+      "movsb\n"
+      "2:"
+      : "=&c" (d0), "=&D" (d1), "=&S" (d2)
+      :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
+      : "memory");
+   return (to);
+}
+#else
+#define __memcpy(a,b,c) memcpy(a,b,c)
+#endif
+
+
+/* ================================================================
+ * Debugging:
+ */
+extern int INTEL_DEBUG;
+
+#define DEBUG_TEXTURE	0x1
+#define DEBUG_STATE	0x2
+#define DEBUG_IOCTL	0x4
+#define DEBUG_BLIT	0x8
+#define DEBUG_MIPTREE   0x10
+#define DEBUG_FALLBACKS	0x20
+#define DEBUG_VERBOSE	0x40
+#define DEBUG_BATCH     0x80
+#define DEBUG_PIXEL     0x100
+#define DEBUG_BUFMGR    0x200
+#define DEBUG_REGION    0x400
+#define DEBUG_FBO       0x800
+#define DEBUG_LOCK      0x1000
+#define DEBUG_SYNC	0x2000
+#define DEBUG_PRIMS	0x4000
+#define DEBUG_VERTS	0x8000
+#define DEBUG_DRI       0x10000
+#define DEBUG_DMA       0x20000
+#define DEBUG_SANITY    0x40000
+#define DEBUG_SLEEP     0x80000
+#define DEBUG_STATS     0x100000
+#define DEBUG_TILE      0x200000
+#define DEBUG_SINGLE_THREAD   0x400000
+#define DEBUG_WM        0x800000
+#define DEBUG_URB       0x1000000
+#define DEBUG_VS        0x2000000
+
+#define DBG(...) do {						\
+	if (INTEL_DEBUG & FILE_DEBUG_FLAG)			\
+		_mesa_printf(__VA_ARGS__);			\
+} while(0)
+
+#define PCI_CHIP_845_G			0x2562
+#define PCI_CHIP_I830_M			0x3577
+#define PCI_CHIP_I855_GM		0x3582
+#define PCI_CHIP_I865_G			0x2572
+#define PCI_CHIP_I915_G			0x2582
+#define PCI_CHIP_I915_GM		0x2592
+#define PCI_CHIP_I945_G			0x2772
+#define PCI_CHIP_I945_GM		0x27A2
+#define PCI_CHIP_I945_GME		0x27AE
+#define PCI_CHIP_G33_G			0x29C2
+#define PCI_CHIP_Q35_G			0x29B2
+#define PCI_CHIP_Q33_G			0x29D2
+
+
+/* ================================================================
+ * intel_context.c:
+ */
+
+extern GLboolean intelInitContext(struct intel_context *intel,
+                                  const __GLcontextModes * mesaVis,
+                                  __DRIcontextPrivate * driContextPriv,
+                                  void *sharedContextPrivate,
+                                  struct dd_function_table *functions);
+
+extern void intelGetLock(struct intel_context *intel, GLuint flags);
+
+extern void intelFinish(GLcontext * ctx);
+extern void intelFlush(GLcontext * ctx);
+
+extern void intelInitDriverFunctions(struct dd_function_table *functions);
+extern void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging);
+
+
+/* ================================================================
+ * intel_state.c:
+ */
+extern void intelInitStateFuncs(struct dd_function_table *functions);
+
+#define COMPAREFUNC_ALWAYS		0
+#define COMPAREFUNC_NEVER		0x1
+#define COMPAREFUNC_LESS		0x2
+#define COMPAREFUNC_EQUAL		0x3
+#define COMPAREFUNC_LEQUAL		0x4
+#define COMPAREFUNC_GREATER		0x5
+#define COMPAREFUNC_NOTEQUAL		0x6
+#define COMPAREFUNC_GEQUAL		0x7
+
+#define STENCILOP_KEEP			0
+#define STENCILOP_ZERO			0x1
+#define STENCILOP_REPLACE		0x2
+#define STENCILOP_INCRSAT		0x3
+#define STENCILOP_DECRSAT		0x4
+#define STENCILOP_INCR			0x5
+#define STENCILOP_DECR			0x6
+#define STENCILOP_INVERT		0x7
+
+#define LOGICOP_CLEAR			0
+#define LOGICOP_NOR			0x1
+#define LOGICOP_AND_INV 		0x2
+#define LOGICOP_COPY_INV		0x3
+#define LOGICOP_AND_RVRSE		0x4
+#define LOGICOP_INV			0x5
+#define LOGICOP_XOR			0x6
+#define LOGICOP_NAND			0x7
+#define LOGICOP_AND			0x8
+#define LOGICOP_EQUIV			0x9
+#define LOGICOP_NOOP			0xa
+#define LOGICOP_OR_INV			0xb
+#define LOGICOP_COPY			0xc
+#define LOGICOP_OR_RVRSE		0xd
+#define LOGICOP_OR			0xe
+#define LOGICOP_SET			0xf
+
+#define BLENDFACT_ZERO			0x01
+#define BLENDFACT_ONE			0x02
+#define BLENDFACT_SRC_COLR		0x03
+#define BLENDFACT_INV_SRC_COLR 		0x04
+#define BLENDFACT_SRC_ALPHA		0x05
+#define BLENDFACT_INV_SRC_ALPHA 	0x06
+#define BLENDFACT_DST_ALPHA		0x07
+#define BLENDFACT_INV_DST_ALPHA 	0x08
+#define BLENDFACT_DST_COLR		0x09
+#define BLENDFACT_INV_DST_COLR		0x0a
+#define BLENDFACT_SRC_ALPHA_SATURATE	0x0b
+#define BLENDFACT_CONST_COLOR		0x0c
+#define BLENDFACT_INV_CONST_COLOR	0x0d
+#define BLENDFACT_CONST_ALPHA		0x0e
+#define BLENDFACT_INV_CONST_ALPHA	0x0f
+#define BLENDFACT_MASK          	0x0f
+
+enum {
+   DRI_CONF_BO_REUSE_DISABLED,
+   DRI_CONF_BO_REUSE_ALL
+};
+
+extern int intel_translate_shadow_compare_func(GLenum func);
+extern int intel_translate_compare_func(GLenum func);
+extern int intel_translate_stencil_op(GLenum op);
+extern int intel_translate_blend_factor(GLenum factor);
+extern int intel_translate_logic_op(GLenum opcode);
+
+void intel_viewport(GLcontext * ctx, GLint x, GLint y,
+		    GLsizei width, GLsizei height);
+
+void intel_update_renderbuffers(__DRIcontext *context,
+				__DRIdrawable *drawable);
+
+/*======================================================================
+ * Inline conversion functions.  
+ * These are better-typed than the macros used previously:
+ */
+static INLINE struct intel_context *
+intel_context(GLcontext * ctx)
+{
+   return (struct intel_context *) ctx;
+}
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_decode.c b/src/mesa/drivers/dri/intel/intel_decode.c
new file mode 100644
index 00000000000..0b8a287f6fe
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_decode.c
@@ -0,0 +1,1281 @@
+/* -*- c-basic-offset: 4 -*- */
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+/** @file intel_decode.c
+ * This file contains code to print out batchbuffer contents in a
+ * human-readable format.
+ *
+ * The current version only supports i915 packets, and only pretty-prints a
+ * subset of them.  The intention is for it to make just a best attempt to
+ * decode, but never crash in the process.
+ */
+
+#include <stdio.h>
+#include <stdarg.h>
+#include <inttypes.h>
+
+#include "intel_decode.h"
+#include "intel_chipset.h"
+
+#define BUFFER_FAIL(_count, _len, _name) do {			\
+    fprintf(out, "Buffer size too small in %s (%d < %d)\n",	\
+	    (_name), (_count), (_len));				\
+    (*failures)++;						\
+    return count;						\
+} while (0)
+
+static FILE *out;
+static uint32_t saved_s2 = 0, saved_s4 = 0;
+static char saved_s2_set = 0, saved_s4_set = 0;
+
+static float
+int_as_float(uint32_t intval)
+{
+    union intfloat {
+	uint32_t i;
+	float f;
+    } uval;
+
+    uval.i = intval;
+    return uval.f;
+}
+
+static void
+instr_out(uint32_t *data, uint32_t hw_offset, unsigned int index,
+	  char *fmt, ...)
+{
+    va_list va;
+
+    fprintf(out, "0x%08x: 0x%08x:%s ", hw_offset + index * 4, data[index],
+	    index == 0 ? "" : "  ");
+    va_start(va, fmt);
+    vfprintf(out, fmt, va);
+    va_end(va);
+}
+
+
+static int
+decode_mi(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_mi[] = {
+	{ 0x08, 1, 1, "MI_ARB_ON_OFF" },
+	{ 0x0a, 1, 1, "MI_BATCH_BUFFER_END" },
+	{ 0x31, 2, 2, "MI_BATCH_BUFFER_START" },
+	{ 0x14, 3, 3, "MI_DISPLAY_BUFFER_INFO" },
+	{ 0x04, 1, 1, "MI_FLUSH" },
+	{ 0x22, 3, 3, "MI_LOAD_REGISTER_IMM" },
+	{ 0x13, 2, 2, "MI_LOAD_SCAN_LINES_EXCL" },
+	{ 0x12, 2, 2, "MI_LOAD_SCAN_LINES_INCL" },
+	{ 0x00, 1, 1, "MI_NOOP" },
+	{ 0x11, 2, 2, "MI_OVERLAY_FLIP" },
+	{ 0x07, 1, 1, "MI_REPORT_HEAD" },
+	{ 0x18, 2, 2, "MI_SET_CONTEXT" },
+	{ 0x20, 3, 4, "MI_STORE_DATA_IMM" },
+	{ 0x21, 3, 4, "MI_STORE_DATA_INDEX" },
+	{ 0x24, 3, 3, "MI_STORE_REGISTER_MEM" },
+	{ 0x02, 1, 1, "MI_USER_INTERRUPT" },
+	{ 0x03, 1, 1, "MI_WAIT_FOR_EVENT" },
+    };
+
+
+    for (opcode = 0; opcode < sizeof(opcodes_mi) / sizeof(opcodes_mi[0]);
+	 opcode++) {
+	if ((data[0] & 0x1f800000) >> 23 == opcodes_mi[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_mi[opcode].name);
+	    if (opcodes_mi[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 2;
+		if (len < opcodes_mi[opcode].min_len ||
+		    len > opcodes_mi[opcode].max_len)
+		{
+		    fprintf(out, "Bad length in %s\n",
+			    opcodes_mi[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_mi[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "MI UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_2d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode, len;
+    char *format = NULL;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_2d[] = {
+	{ 0x40, 5, 5, "COLOR_BLT" },
+	{ 0x43, 6, 6, "SRC_COPY_BLT" },
+	{ 0x01, 8, 8, "XY_SETUP_BLT" },
+	{ 0x11, 9, 9, "XY_SETUP_MONO_PATTERN_SL_BLT" },
+	{ 0x03, 3, 3, "XY_SETUP_CLIP_BLT" },
+	{ 0x24, 2, 2, "XY_PIXEL_BLT" },
+	{ 0x25, 3, 3, "XY_SCANLINES_BLT" },
+	{ 0x26, 4, 4, "Y_TEXT_BLT" },
+	{ 0x31, 5, 134, "XY_TEXT_IMMEDIATE_BLT" },
+	{ 0x50, 6, 6, "XY_COLOR_BLT" },
+	{ 0x51, 6, 6, "XY_PAT_BLT" },
+	{ 0x76, 8, 8, "XY_PAT_CHROMA_BLT" },
+	{ 0x72, 7, 135, "XY_PAT_BLT_IMMEDIATE" },
+	{ 0x77, 9, 137, "XY_PAT_CHROMA_BLT_IMMEDIATE" },
+	{ 0x52, 9, 9, "XY_MONO_PAT_BLT" },
+	{ 0x59, 7, 7, "XY_MONO_PAT_FIXED_BLT" },
+	{ 0x53, 8, 8, "XY_SRC_COPY_BLT" },
+	{ 0x54, 8, 8, "XY_MONO_SRC_COPY_BLT" },
+	{ 0x71, 9, 137, "XY_MONO_SRC_COPY_IMMEDIATE_BLT" },
+	{ 0x55, 9, 9, "XY_FULL_BLT" },
+	{ 0x55, 9, 137, "XY_FULL_IMMEDIATE_PATTERN_BLT" },
+	{ 0x56, 9, 9, "XY_FULL_MONO_SRC_BLT" },
+	{ 0x75, 10, 138, "XY_FULL_MONO_SRC_IMMEDIATE_PATTERN_BLT" },
+	{ 0x57, 12, 12, "XY_FULL_MONO_PATTERN_BLT" },
+	{ 0x58, 12, 12, "XY_FULL_MONO_PATTERN_MONO_SRC_BLT" },
+    };
+
+    switch ((data[0] & 0x1fc00000) >> 22) {
+    case 0x50:
+	instr_out(data, hw_offset, 0,
+		  "XY_COLOR_BLT (rgb %sabled, alpha %sabled, dst tile %d)\n",
+		  (data[0] & (1 << 20)) ? "en" : "dis",
+		  (data[0] & (1 << 21)) ? "en" : "dis",
+		  (data[0] >> 11) & 1);
+
+	len = (data[0] & 0x000000ff) + 2;
+	if (len != 6)
+	    fprintf(out, "Bad count in XY_COLOR_BLT\n");
+	if (count < 6)
+	    BUFFER_FAIL(count, len, "XY_COLOR_BLT");
+
+	switch ((data[1] >> 24) & 0x3) {
+	case 0:
+	    format="8";
+	    break;
+	case 1:
+	    format="565";
+	    break;
+	case 2:
+	    format="1555";
+	    break;
+	case 3:
+	    format="8888";
+	    break;
+	}
+
+	instr_out(data, hw_offset, 1, "format %s, pitch %d, "
+		  "clipping %sabled\n", format,
+		  (short)(data[1] & 0xffff),
+		  data[1] & (1 << 30) ? "en" : "dis");
+	instr_out(data, hw_offset, 2, "(%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 3, "(%d,%d)\n",
+		  data[3] & 0xffff, data[3] >> 16);
+	instr_out(data, hw_offset, 4, "offset 0x%08x\n", data[4]);
+	instr_out(data, hw_offset, 5, "color\n");
+	return len;
+    case 0x53:
+	instr_out(data, hw_offset, 0,
+		  "XY_SRC_COPY_BLT (rgb %sabled, alpha %sabled, "
+		  "src tile %d, dst tile %d)\n",
+		  (data[0] & (1 << 20)) ? "en" : "dis",
+		  (data[0] & (1 << 21)) ? "en" : "dis",
+		  (data[0] >> 15) & 1,
+		  (data[0] >> 11) & 1);
+
+	len = (data[0] & 0x000000ff) + 2;
+	if (len != 8)
+	    fprintf(out, "Bad count in XY_SRC_COPY_BLT\n");
+	if (count < 8)
+	    BUFFER_FAIL(count, len, "XY_SRC_COPY_BLT");
+
+	switch ((data[1] >> 24) & 0x3) {
+	case 0:
+	    format="8";
+	    break;
+	case 1:
+	    format="565";
+	    break;
+	case 2:
+	    format="1555";
+	    break;
+	case 3:
+	    format="8888";
+	    break;
+	}
+
+	instr_out(data, hw_offset, 1, "format %s, dst pitch %d, "
+		  "clipping %sabled\n", format,
+		  (short)(data[1] & 0xffff),
+		  data[1] & (1 << 30) ? "en" : "dis");
+	instr_out(data, hw_offset, 2, "dst (%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 3, "dst (%d,%d)\n",
+		  data[3] & 0xffff, data[3] >> 16);
+	instr_out(data, hw_offset, 4, "dst offset 0x%08x\n", data[4]);
+	instr_out(data, hw_offset, 5, "src (%d,%d)\n",
+		  data[5] & 0xffff, data[5] >> 16);
+	instr_out(data, hw_offset, 6, "src pitch %d\n",
+		  (short)(data[6] & 0xffff));
+	instr_out(data, hw_offset, 7, "src offset 0x%08x\n", data[7]);
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_2d) / sizeof(opcodes_2d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1fc00000) >> 22 == opcodes_2d[opcode].opcode) {
+	    unsigned int i;
+
+	    len = 1;
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_2d[opcode].name);
+	    if (opcodes_2d[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 2;
+		if (len < opcodes_2d[opcode].min_len ||
+		    len > opcodes_2d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_2d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_2d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "2D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1c(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    switch ((data[0] & 0x00f80000) >> 19) {
+    case 0x11:
+	instr_out(data, hw_offset, 0, "3DSTATE_DEPTH_SUBRECTANGLE_DISALBE\n");
+	return 1;
+    case 0x10:
+	instr_out(data, hw_offset, 0, "3DSTATE_SCISSOR_ENABLE\n");
+	return 1;
+    case 0x01:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_COORD_SET_I830\n");
+	return 1;
+    case 0x0a:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_CUBE_I830\n");
+	return 1;
+    case 0x05:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_TEX_STREAM_I830\n");
+	return 1;
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures, int i830)
+{
+    unsigned int len, i, c, opcode, word, map, sampler, instr;
+
+    struct {
+	uint32_t opcode;
+	int i830_only;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d_1d[] = {
+	{ 0x8e, 0, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x86, 0, 4, 4, "3DSTATE_CHROMA_KEY" },
+	{ 0x9c, 0, 1, 1, "3DSTATE_CLEAR_PARAMETERS" },
+	{ 0x88, 0, 2, 2, "3DSTATE_CONSTANT_BLEND_COLOR" },
+	{ 0x99, 0, 2, 2, "3DSTATE_DEFAULT_DIFFUSE" },
+	{ 0x9a, 0, 2, 2, "3DSTATE_DEFAULT_SPECULAR" },
+	{ 0x98, 0, 2, 2, "3DSTATE_DEFAULT_Z" },
+	{ 0x97, 0, 2, 2, "3DSTATE_DEPTH_OFFSET_SCALE" },
+	{ 0x85, 0, 2, 2, "3DSTATE_DEST_BUFFER_VARIABLES" },
+	{ 0x80, 0, 5, 5, "3DSTATE_DRAWING_RECTANGLE" },
+	{ 0x8e, 0, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x9d, 0, 65, 65, "3DSTATE_FILTER_COEFFICIENTS_4X4" },
+	{ 0x9e, 0, 4, 4, "3DSTATE_MONO_FILTER" },
+	{ 0x89, 0, 4, 4, "3DSTATE_FOG_MODE" },
+	{ 0x8f, 0, 2, 16, "3DSTATE_MAP_PALLETE_LOAD_32" },
+	{ 0x81, 0, 3, 3, "3DSTATE_SCISSOR_RECTANGLE" },
+	{ 0x83, 0, 2, 2, "3DSTATE_SPAN_STIPPLE" },
+	{ 0x8c, 1, 2, 2, "3DSTATE_MAP_COORD_TRANSFORM_I830" },
+	{ 0x8b, 1, 2, 2, "3DSTATE_MAP_VERTEX_TRANSFORM_I830" },
+	{ 0x8d, 1, 3, 3, "3DSTATE_W_STATE_I830" },
+	{ 0x01, 1, 2, 2, "3DSTATE_COLOR_FACTOR_I830" },
+	{ 0x02, 1, 2, 2, "3DSTATE_MAP_COORD_SETBIND_I830" },
+    };
+
+    switch ((data[0] & 0x00ff0000) >> 16) {
+    case 0x07:
+	/* This instruction is unusual.  A 0 length means just 1 DWORD instead of
+	 * 2.  The 0 length is specified in one place to be unsupported, but
+	 * stated to be required in another, and 0 length LOAD_INDIRECTs appear
+	 * to cause no harm at least.
+	 */
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_INDIRECT\n");
+	len = (data[0] & 0x000000ff) + 1;
+	i = 1;
+	if (data[0] & (0x01 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SIS.0\n");
+	    instr_out(data, hw_offset, i++, "SIS.1\n");
+	}
+	if (data[0] & (0x02 << 8)) {
+	    if (i + 1 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "DIS.0\n");
+	}
+	if (data[0] & (0x04 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SSB.0\n");
+	    instr_out(data, hw_offset, i++, "SSB.1\n");
+	}
+	if (data[0] & (0x08 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "MSB.0\n");
+	    instr_out(data, hw_offset, i++, "MSB.1\n");
+	}
+	if (data[0] & (0x10 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSP.0\n");
+	    instr_out(data, hw_offset, i++, "PSP.1\n");
+	}
+	if (data[0] & (0x20 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSC.0\n");
+	    instr_out(data, hw_offset, i++, "PSC.1\n");
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x04:
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_STATE_IMMEDIATE_1\n");
+	len = (data[0] & 0x0000000f) + 2;
+	i = 1;
+	for (word = 0; word <= 7; word++) {
+	    if (data[0] & (1 << (4 + word))) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_LOAD_STATE_IMMEDIATE_1");
+
+		/* save vertex state for decode */
+		if (word == 2) {
+		    saved_s2_set = 1;
+		    saved_s2 = data[i];
+		}
+		if (word == 4) {
+		    saved_s4_set = 1;
+		    saved_s4 = data[i];
+		}
+
+		instr_out(data, hw_offset, i++, "S%d\n", word);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x00:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+
+	i = 1;
+	for (map = 0; map <= 15; map++) {
+	    if (data[1] & (1 << map)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
+		instr_out(data, hw_offset, i++, "map %d MS2\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS3\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS4\n", map);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x06:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_CONSTANTS\n");
+	len = (data[0] & 0x000000ff) + 2;
+
+	i = 1;
+	for (c = 0; c <= 31; c++) {
+	    if (data[1] & (1 << c)) {
+		if (i + 4 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_PIXEL_SHADER_CONSTANTS");
+		instr_out(data, hw_offset, i, "C%d.X = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Y = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Z = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.W = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x05:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	len = (data[0] & 0x000000ff) + 2;
+	if ((len - 1) % 3 != 0 || len > 370) {
+	    fprintf(out, "Bad count in 3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	    (*failures)++;
+	}
+	i = 1;
+	for (instr = 0; instr < (len - 1) / 3; instr++) {
+	    if (i + 3 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+	}
+	return len;
+    case 0x01:
+	if (i830)
+	    break;
+	instr_out(data, hw_offset, 0, "3DSTATE_SAMPLER_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+	i = 1;
+	for (sampler = 0; sampler <= 15; sampler++) {
+	    if (data[1] & (1 << sampler)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_SAMPLER_STATE");
+		instr_out(data, hw_offset, i++, "sampler %d SS2\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS3\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS4\n",
+			  sampler);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_SAMPLER_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d_1d) / sizeof(opcodes_3d_1d[0]);
+	 opcode++)
+    {
+	if (opcodes_3d_1d[opcode].i830_only && !i830)
+	    continue;
+
+	if (((data[0] & 0x00ff0000) >> 16) == opcodes_3d_1d[opcode].opcode) {
+	    len = 1;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d_1d[opcode].name);
+	    if (opcodes_3d_1d[opcode].max_len > 1) {
+		len = (data[0] & 0x0000ffff) + 2;
+		if (len < opcodes_3d_1d[opcode].min_len ||
+		    len > opcodes_3d_1d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n",
+			    opcodes_3d_1d[opcode].name);
+		    (*failures)++;
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len,  opcodes_3d_1d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_primitive(uint32_t *data, int count, uint32_t hw_offset,
+		    int *failures)
+{
+    char immediate = (data[0] & (1 << 23)) == 0;
+    unsigned int len, i;
+    char *primtype;
+
+    switch ((data[0] >> 18) & 0xf) {
+    case 0x0: primtype = "TRILIST"; break;
+    case 0x1: primtype = "TRISTRIP"; break;
+    case 0x2: primtype = "TRISTRIP_REVERSE"; break;
+    case 0x3: primtype = "TRIFAN"; break;
+    case 0x4: primtype = "POLYGON"; break;
+    case 0x5: primtype = "LINELIST"; break;
+    case 0x6: primtype = "LINESTRIP"; break;
+    case 0x7: primtype = "RECTLIST"; break;
+    case 0x8: primtype = "POINTLIST"; break;
+    case 0x9: primtype = "DIB"; break;
+    case 0xa: primtype = "CLEAR_RECT"; break;
+    default: primtype = "unknown"; break;
+    }
+
+    /* XXX: 3DPRIM_DIB not supported */
+    if (immediate) {
+	len = (data[0] & 0x0003ffff) + 2;
+	instr_out(data, hw_offset, 0, "3DPRIMITIVE inline %s\n", primtype);
+	if (count < len)
+	    BUFFER_FAIL(count, len,  "3DPRIMITIVE inline");
+	if (!saved_s2_set || !saved_s4_set) {
+	    fprintf(out, "unknown vertex format\n");
+	    for (i = 1; i < len; i++) {
+		instr_out(data, hw_offset, i,
+			  "           vertex data (%f float)\n",
+			  int_as_float(data[i]));
+	    }
+	} else {
+	    unsigned int vertex = 0;
+	    for (i = 1; i < len;) {
+		unsigned int tc;
+
+#define VERTEX_OUT(fmt, ...) do {					\
+    if (i < len)							\
+	instr_out(data, hw_offset, i, " V%d."fmt"\n", vertex, __VA_ARGS__); \
+    else								\
+	fprintf(out, " missing data in V%d\n", vertex);			\
+    i++;								\
+} while (0)
+
+		VERTEX_OUT("X = %f", int_as_float(data[i]));
+		VERTEX_OUT("Y = %f", int_as_float(data[i]));
+	        switch (saved_s4 >> 6 & 0x7) {
+		case 0x1:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    break;
+		case 0x2:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		case 0x3:
+		    break;
+		case 0x4:
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		default:
+		    fprintf(out, "bad S4 position mask\n");
+		}
+
+		if (saved_s4 & (1 << 10)) {
+		    VERTEX_OUT("color = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 11)) {
+		    VERTEX_OUT("spec = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 12))
+		    VERTEX_OUT("width = 0x%08x)", data[i]);
+
+		for (tc = 0; tc <= 7; tc++) {
+		    switch ((saved_s2 >> (tc * 4)) & 0xf) {
+		    case 0x0:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x1:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x2:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.W = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x3:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x4:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0x5:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			VERTEX_OUT("T%d.ZW = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0xf:
+			break;
+		    default:
+			fprintf(out, "bad S2.T%d format\n", tc);
+		    }
+		}
+		vertex++;
+	    }
+	}
+    } else {
+	/* indirect vertices */
+	len = data[0] & 0x0000ffff; /* index count */
+	if (data[0] & (1 << 17)) {
+	    /* random vertex access */
+	    if (count < (len + 1) / 2 + 1) {
+		BUFFER_FAIL(count, (len + 1) / 2 + 1,
+			    "3DPRIMITIVE random indirect");
+	    }
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE random indirect %s (%d)\n", primtype, len);
+	    if (len == 0) {
+		/* vertex indices continue until 0xffff is found */
+		for (i = 1; i < count; i++) {
+		    if ((data[i] & 0xffff) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: (terminator)\n");
+			return i;
+		    } else if ((data[i] >> 16) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, "
+				  "(terminator)\n",
+				  data[i] & 0xffff);
+			return i;
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+		fprintf(out,
+			"3DPRIMITIVE: no terminator found in index buffer\n");
+		(*failures)++;
+		return count;
+	    } else {
+		/* fixed size vertex index buffer */
+		for (i = 0; i < len; i += 2) {
+		    if (i * 2 == len - 1) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x\n",
+				  data[i] & 0xffff);
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+	    }
+	    return (len + 1) / 2 + 1;
+	} else {
+	    /* sequential vertex access */
+	    if (count < 2)
+		BUFFER_FAIL(count, 2, "3DPRIMITIVE seq indirect");
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE sequential indirect %s, %d starting from "
+		      "%d\n", primtype, len, data[1] & 0xffff);
+	    instr_out(data, hw_offset, 1, "           start\n");
+	    return 2;
+	}
+    }
+
+    return len;
+}
+
+static int
+decode_3d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d[] = {
+	{ 0x06, 1, 1, "3DSTATE_ANTI_ALIASING" },
+	{ 0x08, 1, 1, "3DSTATE_BACKFACE_STENCIL_OPS" },
+	{ 0x09, 1, 1, "3DSTATE_BACKFACE_STENCIL_MASKS" },
+	{ 0x16, 1, 1, "3DSTATE_COORD_SET_BINDINGS" },
+	{ 0x15, 1, 1, "3DSTATE_FOG_COLOR" },
+	{ 0x0b, 1, 1, "3DSTATE_INDEPENDENT_ALPHA_BLEND" },
+	{ 0x0d, 1, 1, "3DSTATE_MODES_4" },
+	{ 0x0c, 1, 1, "3DSTATE_MODES_5" },
+	{ 0x07, 1, 1, "3DSTATE_RASTERIZATION_RULES" },
+    };
+
+    switch ((data[0] & 0x1f000000) >> 24) {
+    case 0x1f:
+	return decode_3d_primitive(data, count, hw_offset, failures);
+    case 0x1d:
+	return decode_3d_1d(data, count, hw_offset, failures, 0);
+    case 0x1c:
+	return decode_3d_1c(data, count, hw_offset, failures);
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1f000000) >> 24 == opcodes_3d[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d[opcode].name);
+	    if (opcodes_3d[opcode].max_len > 1) {
+		len = (data[0] & 0xff) + 2;
+		if (len < opcodes_3d[opcode].min_len ||
+		    len > opcodes_3d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_3d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_3d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static const char *
+get_965_surfacetype(unsigned int surfacetype)
+{
+    switch (surfacetype) {
+    case 0: return "1D";
+    case 1: return "2D";
+    case 2: return "3D";
+    case 3: return "CUBE";
+    case 4: return "BUFFER";
+    case 7: return "NULL";
+    default: return "unknown";
+    }
+}
+
+static const char *
+get_965_depthformat(unsigned int depthformat)
+{
+    switch (depthformat) {
+    case 0: return "s8_z24float";
+    case 1: return "z32float";
+    case 2: return "z24s8";
+    case 5: return "z16";
+    default: return "unknown";
+    }
+}
+
+static const char *
+get_965_element_component(uint32_t data, int component)
+{
+    uint32_t component_control = (data >> (16 + (3 - component) * 4)) & 0x7;
+
+    switch (component_control) {
+    case 0:
+	return "nostore";
+    case 1:
+	switch (component) {
+	case 0: return "X";
+	case 1: return "Y";
+	case 2: return "Z";
+	case 3: return "W";
+	default: return "fail";
+	}
+    case 2:
+	return "0.0";
+    case 3:
+	return "1.0";
+    case 4:
+	return "0x1";
+    case 5:
+	return "VID";
+    default:
+	return "fail";
+    }
+}
+
+static const char *
+get_965_prim_type(uint32_t data)
+{
+    uint32_t primtype = (data >> 10) & 0x1f;
+
+    switch (primtype) {
+    case 0x01: return "point list";
+    case 0x02: return "line list";
+    case 0x03: return "line strip";
+    case 0x04: return "tri list";
+    case 0x05: return "tri strip";
+    case 0x06: return "tri fan";
+    case 0x07: return "quad list";
+    case 0x08: return "quad strip";
+    case 0x09: return "line list adj";
+    case 0x0a: return "line strip adj";
+    case 0x0b: return "tri list adj";
+    case 0x0c: return "tri strip adj";
+    case 0x0d: return "tri strip reverse";
+    case 0x0e: return "polygon";
+    case 0x0f: return "rect list";
+    case 0x10: return "line loop";
+    case 0x11: return "point list bf";
+    case 0x12: return "line strip cont";
+    case 0x13: return "line strip bf";
+    case 0x14: return "line strip cont bf";
+    case 0x15: return "tri fan no stipple";
+    default: return "fail";
+    }
+}
+
+static int
+decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode, len;
+    int i;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d[] = {
+	{ 0x6000, 3, 3, "URB_FENCE" },
+	{ 0x6001, 2, 2, "CS_URB_STATE" },
+	{ 0x6002, 2, 2, "CONSTANT_BUFFER" },
+	{ 0x6101, 6, 6, "STATE_BASE_ADDRESS" },
+	{ 0x6102, 2, 2 , "STATE_SIP" },
+	{ 0x6104, 1, 1, "3DSTATE_PIPELINE_SELECT" },
+	{ 0x680b, 1, 1, "3DSTATE_VF_STATISTICS" },
+	{ 0x6904, 1, 1, "3DSTATE_PIPELINE_SELECT" },
+	{ 0x7800, 7, 7, "3DSTATE_PIPELINED_POINTERS" },
+	{ 0x7801, 6, 6, "3DSTATE_BINDING_TABLE_POINTERS" },
+	{ 0x780b, 1, 1, "3DSTATE_VF_STATISTICS" },
+	{ 0x7808, 5, 257, "3DSTATE_VERTEX_BUFFERS" },
+	{ 0x7809, 3, 256, "3DSTATE_VERTEX_ELEMENTS" },
+	{ 0x780a, 3, 3, "3DSTATE_INDEX_BUFFER" },
+	{ 0x7900, 4, 4, "3DSTATE_DRAWING_RECTANGLE" },
+	{ 0x7901, 5, 5, "3DSTATE_CONSTANT_COLOR" },
+	{ 0x7905, 5, 7, "3DSTATE_DEPTH_BUFFER" },
+	{ 0x7906, 2, 2, "3DSTATE_POLY_STIPPLE_OFFSET" },
+	{ 0x7907, 33, 33, "3DSTATE_POLY_STIPPLE_PATTERN" },
+	{ 0x7908, 3, 3, "3DSTATE_LINE_STIPPLE" },
+	{ 0x7909, 2, 2, "3DSTATE_GLOBAL_DEPTH_OFFSET_CLAMP" },
+	{ 0x790a, 3, 3, "3DSTATE_AA_LINE_PARAMETERS" },
+	{ 0x7b00, 6, 6, "3DPRIMITIVE" },
+    };
+
+    len = (data[0] & 0x0000ffff) + 2;
+
+    switch ((data[0] & 0xffff0000) >> 16) {
+    case 0x6101:
+	if (len != 6)
+	    fprintf(out, "Bad count in STATE_BASE_ADDRESS\n");
+	if (count < 6)
+	    BUFFER_FAIL(count, len, "STATE_BASE_ADDRESS");
+
+	instr_out(data, hw_offset, 0,
+		  "STATE_BASE_ADDRESS\n");
+
+	if (data[1] & 1) {
+	    instr_out(data, hw_offset, 1, "General state at 0x%08x\n",
+		      data[1] & ~1);
+	} else
+	    instr_out(data, hw_offset, 1, "General state not updated\n");
+
+	if (data[2] & 1) {
+	    instr_out(data, hw_offset, 2, "Surface state at 0x%08x\n",
+		      data[2] & ~1);
+	} else
+	    instr_out(data, hw_offset, 2, "Surface state not updated\n");
+
+	if (data[3] & 1) {
+	    instr_out(data, hw_offset, 3, "Indirect state at 0x%08x\n",
+		      data[3] & ~1);
+	} else
+	    instr_out(data, hw_offset, 3, "Indirect state not updated\n");
+
+	if (data[4] & 1) {
+	    instr_out(data, hw_offset, 4, "General state upper bound 0x%08x\n",
+		      data[4] & ~1);
+	} else
+	    instr_out(data, hw_offset, 4, "General state not updated\n");
+
+	if (data[5] & 1) {
+	    instr_out(data, hw_offset, 5, "Indirect state upper bound 0x%08x\n",
+		      data[5] & ~1);
+	} else
+	    instr_out(data, hw_offset, 5, "Indirect state not updated\n");
+
+	return len;
+    case 0x7800:
+	if (len != 7)
+	    fprintf(out, "Bad count in 3DSTATE_PIPELINED_POINTERS\n");
+	if (count < 7)
+	    BUFFER_FAIL(count, len, "3DSTATE_PIPELINED_POINTERS");
+
+	instr_out(data, hw_offset, 0,
+		  "3DSTATE_PIPELINED_POINTERS\n");
+	instr_out(data, hw_offset, 1, "VS state\n");
+	instr_out(data, hw_offset, 2, "GS state\n");
+	instr_out(data, hw_offset, 3, "Clip state\n");
+	instr_out(data, hw_offset, 4, "SF state\n");
+	instr_out(data, hw_offset, 5, "WM state\n");
+	instr_out(data, hw_offset, 6, "CC state\n");
+	return len;
+    case 0x7801:
+	if (len != 6)
+	    fprintf(out, "Bad count in 3DSTATE_BINDING_TABLE_POINTERS\n");
+	if (count < 6)
+	    BUFFER_FAIL(count, len, "3DSTATE_BINDING_TABLE_POINTERS");
+
+	instr_out(data, hw_offset, 0,
+		  "3DSTATE_BINDING_TABLE_POINTERS\n");
+	instr_out(data, hw_offset, 1, "VS binding table\n");
+	instr_out(data, hw_offset, 2, "GS binding table\n");
+	instr_out(data, hw_offset, 3, "Clip binding table\n");
+	instr_out(data, hw_offset, 4, "SF binding table\n");
+	instr_out(data, hw_offset, 5, "WM binding table\n");
+
+	return len;
+
+    case 0x7808:
+	len = (data[0] & 0xff) + 2;
+	if ((len - 1) % 4 != 0)
+	    fprintf(out, "Bad count in 3DSTATE_VERTEX_BUFFERS\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_VERTEX_BUFFERS");
+	instr_out(data, hw_offset, 0, "3DSTATE_VERTEX_BUFFERS\n");
+
+	for (i = 1; i < len;) {
+	    instr_out(data, hw_offset, i, "buffer %d: %s, pitch %db\n",
+		      data[i] >> 27,
+		      data[i] & (1 << 26) ? "random" : "sequential",
+		      data[i] & 0x07ff);
+	    i++;
+	    instr_out(data, hw_offset, i++, "buffer address\n");
+	    instr_out(data, hw_offset, i++, "max index\n");
+	    instr_out(data, hw_offset, i++, "mbz\n");
+	}
+	return len;
+
+    case 0x7809:
+	len = (data[0] & 0xff) + 2;
+	if ((len + 1) % 2 != 0)
+	    fprintf(out, "Bad count in 3DSTATE_VERTEX_ELEMENTS\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_VERTEX_ELEMENTS");
+	instr_out(data, hw_offset, 0, "3DSTATE_VERTEX_ELEMENTS\n");
+
+	for (i = 1; i < len;) {
+	    instr_out(data, hw_offset, i, "buffer %d: %svalid, type 0x%04x, "
+		      "src offset 0x%04xd bytes\n",
+		      data[i] >> 27,
+		      data[i] & (1 << 26) ? "" : "in",
+		      (data[i] >> 16) & 0x1ff,
+		      data[i] & 0x07ff);
+	    i++;
+	    instr_out(data, hw_offset, i, "(%s, %s, %s, %s), "
+		      "dst offset 0x%02x bytes\n",
+		      get_965_element_component(data[i], 0),
+		      get_965_element_component(data[i], 1),
+		      get_965_element_component(data[i], 2),
+		      get_965_element_component(data[i], 3),
+		      (data[i] & 0xff) * 4);
+	    i++;
+	}
+	return len;
+
+    case 0x780a:
+	len = (data[0] & 0xff) + 2;
+	if (len != 3)
+	    fprintf(out, "Bad count in 3DSTATE_INDEX_BUFFER\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_INDEX_BUFFER");
+	instr_out(data, hw_offset, 0, "3DSTATE_INDEX_BUFFER\n");
+	instr_out(data, hw_offset, 1, "beginning buffer address\n");
+	instr_out(data, hw_offset, 2, "ending buffer address\n");
+	return len;
+
+    case 0x7900:
+	if (len != 4)
+	    fprintf(out, "Bad count in 3DSTATE_DRAWING_RECTANGLE\n");
+	if (count < 4)
+	    BUFFER_FAIL(count, len, "3DSTATE_DRAWING_RECTANGLE");
+
+	instr_out(data, hw_offset, 0,
+		  "3DSTATE_DRAWING_RECTANGLE\n");
+	instr_out(data, hw_offset, 1, "top left: %d,%d\n",
+		  data[1] & 0xffff,
+		  (data[1] >> 16) & 0xffff);
+	instr_out(data, hw_offset, 2, "bottom right: %d,%d\n",
+		  data[2] & 0xffff,
+		  (data[2] >> 16) & 0xffff);
+	instr_out(data, hw_offset, 3, "origin: %d,%d\n",
+		  (int)data[3] & 0xffff,
+		  ((int)data[3] >> 16) & 0xffff);
+
+	return len;
+
+    case 0x7905:
+	if (len != 5 && len != 6)
+	    fprintf(out, "Bad count in 3DSTATE_DEPTH_BUFFER\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_DEPTH_BUFFER");
+
+	instr_out(data, hw_offset, 0,
+		  "3DSTATE_DEPTH_BUFFER\n");
+	instr_out(data, hw_offset, 1, "%s, %s, pitch = %d bytes, %stiled\n",
+		  get_965_surfacetype(data[1] >> 29),
+		  get_965_depthformat((data[1] >> 18) & 0x7),
+		  (data[1] & 0x0001ffff) + 1,
+		  data[1] & (1 << 27) ? "" : "not ");
+	instr_out(data, hw_offset, 2, "depth offset\n");
+	instr_out(data, hw_offset, 3, "%dx%d\n",
+		  ((data[3] & 0x0007ffc0) >> 6) + 1,
+		  ((data[3] & 0xfff80000) >> 19) + 1);
+	instr_out(data, hw_offset, 4, "volume depth\n");
+	if (len == 6)
+	    instr_out(data, hw_offset, 5, "\n");
+
+	return len;
+
+    case 0x7b00:
+	len = (data[0] & 0xff) + 2;
+	if (len != 6)
+	    fprintf(out, "Bad count in 3DPRIMITIVE\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DPRIMITIVE");
+
+	instr_out(data, hw_offset, 0,
+		  "3DPRIMITIVE: %s %s\n",
+		  get_965_prim_type(data[0]),
+		  (data[0] & (1 << 15)) ? "random" : "sequential");
+	instr_out(data, hw_offset, 1, "primitive count\n");
+	instr_out(data, hw_offset, 2, "start vertex\n");
+	instr_out(data, hw_offset, 3, "instance count\n");
+	instr_out(data, hw_offset, 4, "start instance\n");
+	instr_out(data, hw_offset, 5, "index bias\n");
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0xffff0000) >> 16 == opcodes_3d[opcode].opcode) {
+	    unsigned int i;
+	    len = 1;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d[opcode].name);
+	    if (opcodes_3d[opcode].max_len > 1) {
+		len = (data[0] & 0xff) + 2;
+		if (len < opcodes_3d[opcode].min_len ||
+		    len > opcodes_3d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_3d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_3d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_i830(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d[] = {
+	{ 0x02, 1, 1, "3DSTATE_MODES_3" },
+	{ 0x03, 1, 1, "3DSTATE_ENABLES_1"},
+	{ 0x04, 1, 1, "3DSTATE_ENABLES_2"},
+	{ 0x05, 1, 1, "3DSTATE_VFT0"},
+	{ 0x06, 1, 1, "3DSTATE_AA"},
+	{ 0x07, 1, 1, "3DSTATE_RASTERIZATION_RULES" },
+	{ 0x08, 1, 1, "3DSTATE_MODES_1" },
+	{ 0x09, 1, 1, "3DSTATE_STENCIL_TEST" },
+	{ 0x0a, 1, 1, "3DSTATE_VFT1"},
+	{ 0x0b, 1, 1, "3DSTATE_INDPT_ALPHA_BLEND" },
+	{ 0x0c, 1, 1, "3DSTATE_MODES_5" },
+	{ 0x0d, 1, 1, "3DSTATE_MAP_BLEND_OP" },
+	{ 0x0e, 1, 1, "3DSTATE_MAP_BLEND_ARG" },
+	{ 0x0f, 1, 1, "3DSTATE_MODES_2" },
+	{ 0x15, 1, 1, "3DSTATE_FOG_COLOR" },
+	{ 0x16, 1, 1, "3DSTATE_MODES_4" },
+    };
+
+    switch ((data[0] & 0x1f000000) >> 24) {
+    case 0x1f:
+	return decode_3d_primitive(data, count, hw_offset, failures);
+    case 0x1d:
+	return decode_3d_1d(data, count, hw_offset, failures, 1);
+    case 0x1c:
+	return decode_3d_1c(data, count, hw_offset, failures);
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1f000000) >> 24 == opcodes_3d[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d[opcode].name);
+	    if (opcodes_3d[opcode].max_len > 1) {
+		len = (data[0] & 0xff) + 2;
+		if (len < opcodes_3d[opcode].min_len ||
+		    len > opcodes_3d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_3d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_3d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+/**
+ * Decodes an i830-i915 batch buffer, writing the output to stdout.
+ *
+ * \param data batch buffer contents
+ * \param count number of DWORDs to decode in the batch buffer
+ * \param hw_offset hardware address for the buffer
+ */
+int
+intel_decode(uint32_t *data, int count, uint32_t hw_offset, uint32_t devid)
+{
+    int index = 0;
+    int failures = 0;
+
+    out = stderr;
+
+    while (index < count) {
+	switch ((data[index] & 0xe0000000) >> 29) {
+	case 0x0:
+	    index += decode_mi(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x2:
+	    index += decode_2d(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x3:
+	    if (IS_965(devid)) {
+		index += decode_3d_965(data + index, count - index,
+				       hw_offset + index * 4, &failures);
+	    } else if (IS_9XX(devid)) {
+		index += decode_3d(data + index, count - index,
+				   hw_offset + index * 4, &failures);
+	    } else {
+		index += decode_3d_i830(data + index, count - index,
+					hw_offset + index * 4, &failures);
+	    }
+	    break;
+	default:
+	    instr_out(data, hw_offset, index, "UNKNOWN\n");
+	    failures++;
+	    index++;
+	    break;
+	}
+	fflush(out);
+    }
+
+    return failures;
+}
+
+void intel_decode_context_reset(void)
+{
+    saved_s2_set = 0;
+    saved_s4_set = 1;
+}
+
diff --git a/src/mesa/drivers/dri/intel/intel_decode.h b/src/mesa/drivers/dri/intel/intel_decode.h
new file mode 100644
index 00000000000..c50644a46b5
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_decode.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <eric@anholt.net>
+ *
+ */
+
+int intel_decode(uint32_t *data, int count, uint32_t hw_offset, uint32_t devid);
+void intel_decode_context_reset(void);
diff --git a/src/mesa/drivers/dri/intel/intel_depthstencil.c b/src/mesa/drivers/dri/intel/intel_depthstencil.c
new file mode 100644
index 00000000000..c2b4d7728b4
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_depthstencil.c
@@ -0,0 +1,252 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/depthstencil.h"
+#include "main/fbobject.h"
+#include "main/framebuffer.h"
+#include "main/hash.h"
+#include "main/mtypes.h"
+#include "main/renderbuffer.h"
+
+#include "intel_context.h"
+#include "intel_fbo.h"
+#include "intel_depthstencil.h"
+#include "intel_regions.h"
+#include "intel_span.h"
+
+/**
+ * The GL_EXT_framebuffer_object allows the user to create their own
+ * framebuffer objects consisting of color renderbuffers (0 or more),
+ * depth renderbuffers (0 or 1) and stencil renderbuffers (0 or 1).
+ *
+ * The spec considers depth and stencil renderbuffers to be totally independent
+ * buffers.  In reality, most graphics hardware today uses a combined
+ * depth+stencil buffer (one 32-bit pixel = 24 bits of Z + 8 bits of stencil).
+ *
+ * This causes difficulty because the user may create some number of depth
+ * renderbuffers and some number of stencil renderbuffers and bind them
+ * together in framebuffers in any combination.
+ *
+ * This code manages all that.
+ *
+ * 1. Depth renderbuffers are always allocated in hardware as 32bpp
+ *    GL_DEPTH24_STENCIL8 buffers.
+ *
+ * 2. Stencil renderbuffers are initially allocated in software as 8bpp
+ *    GL_STENCIL_INDEX8 buffers.
+ *
+ * 3. Depth and Stencil renderbuffers use the PairedStencil and PairedDepth
+ *    fields (respectively) to indicate if the buffer's currently paired
+ *    with another stencil or depth buffer (respectively).
+ *
+ * 4. When a depth and stencil buffer are initially both attached to the
+ *    current framebuffer, we merge the stencil buffer values into the
+ *    depth buffer (really a depth+stencil buffer).  The then hardware uses
+ *    the combined buffer.
+ *
+ * 5. Whenever a depth or stencil buffer is reallocated (with
+ *    glRenderbufferStorage) we undo the pairing and copy the stencil values
+ *    from the combined depth/stencil buffer back to the stencil-only buffer.
+ *
+ * 6. We also undo the pairing when we find a change in buffer bindings.
+ *
+ * 7. If a framebuffer is only using a depth renderbuffer (no stencil), we
+ *    just use the combined depth/stencil buffer and ignore the stencil values.
+ *
+ * 8. If a framebuffer is only using a stencil renderbuffer (no depth) we have
+ *    to promote the 8bpp software stencil buffer to a 32bpp hardware
+ *    depth+stencil buffer.
+ *
+ */
+
+/**
+ * Undo the pairing/interleaving between depth and stencil buffers.
+ * irb should be a depth/stencil or stencil renderbuffer.
+ */
+void
+intel_unpair_depth_stencil(GLcontext *ctx, struct intel_renderbuffer *irb)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct gl_renderbuffer *rb = &irb->Base;
+
+   if (irb->PairedStencil) {
+      /* irb is a depth/stencil buffer */
+      struct gl_renderbuffer *stencilRb;
+      struct intel_renderbuffer *stencilIrb;
+
+      ASSERT(rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+
+      stencilRb = _mesa_lookup_renderbuffer(ctx, irb->PairedStencil);
+      stencilIrb = intel_renderbuffer(stencilRb);
+      if (stencilIrb) {
+         /* need to extract stencil values from the depth buffer */
+	 ASSERT(stencilIrb->PairedDepth == rb->Name);
+	 intel_renderbuffer_map(intel, rb);
+	 intel_renderbuffer_map(intel, stencilRb);
+	 _mesa_extract_stencil(ctx, rb, stencilRb);
+	 intel_renderbuffer_unmap(intel, stencilRb);
+	 intel_renderbuffer_unmap(intel, rb);
+         stencilIrb->PairedDepth = 0;
+      }
+      irb->PairedStencil = 0;
+   }
+   else if (irb->PairedDepth) {
+      /* irb is a stencil buffer */
+      struct gl_renderbuffer *depthRb;
+      struct intel_renderbuffer *depthIrb;
+
+      ASSERT(rb->_ActualFormat == GL_STENCIL_INDEX8_EXT ||
+             rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+
+      depthRb = _mesa_lookup_renderbuffer(ctx, irb->PairedDepth);
+      depthIrb = intel_renderbuffer(depthRb);
+      if (depthIrb) {
+         /* need to extract stencil values from the depth buffer */
+	 ASSERT(depthIrb->PairedStencil == rb->Name);
+	 intel_renderbuffer_map(intel, rb);
+	 intel_renderbuffer_map(intel, depthRb);
+	 _mesa_extract_stencil(ctx, depthRb, rb);
+	 intel_renderbuffer_unmap(intel, depthRb);
+	 intel_renderbuffer_unmap(intel, rb);
+         depthIrb->PairedStencil = 0;
+      }
+      irb->PairedDepth = 0;
+   }
+   else {
+      _mesa_problem(ctx, "Problem in undo_depth_stencil_pairing");
+   }
+
+   ASSERT(irb->PairedStencil == 0);
+   ASSERT(irb->PairedDepth == 0);
+}
+
+
+/**
+ * Examine the depth and stencil renderbuffers which are attached to the
+ * framebuffer.  If both depth and stencil are attached, make sure that the
+ * renderbuffers are 'paired' (combined).  If only depth or only stencil is
+ * attached, undo any previous pairing.
+ *
+ * Must be called if NewState & _NEW_BUFFER (when renderbuffer attachments
+ * change, for example).
+ */
+void
+intel_validate_paired_depth_stencil(GLcontext * ctx,
+                                    struct gl_framebuffer *fb)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_renderbuffer *depthRb, *stencilRb;
+
+   depthRb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
+   stencilRb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
+
+   if (depthRb && stencilRb) {
+      if (depthRb == stencilRb) {
+         /* Using a user-created combined depth/stencil buffer.
+          * Nothing to do.
+          */
+         ASSERT(depthRb->Base._BaseFormat == GL_DEPTH_STENCIL_EXT);
+         ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+      }
+      else {
+         /* Separate depth/stencil buffers, need to interleave now */
+         ASSERT(depthRb->Base._BaseFormat == GL_DEPTH_COMPONENT);
+         ASSERT(stencilRb->Base._BaseFormat == GL_STENCIL_INDEX);
+         /* may need to interleave depth/stencil now */
+         if (depthRb->PairedStencil == stencilRb->Base.Name) {
+            /* OK, the depth and stencil buffers are already interleaved */
+            ASSERT(stencilRb->PairedDepth == depthRb->Base.Name);
+         }
+         else {
+            /* need to setup new pairing/interleaving */
+            if (depthRb->PairedStencil) {
+               intel_unpair_depth_stencil(ctx, depthRb);
+            }
+            if (stencilRb->PairedDepth) {
+               intel_unpair_depth_stencil(ctx, stencilRb);
+            }
+
+            ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+            ASSERT(stencilRb->Base._ActualFormat == GL_STENCIL_INDEX8_EXT ||
+                   stencilRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+
+            /* establish new pairing: interleave stencil into depth buffer */
+	    intel_renderbuffer_map(intel, &depthRb->Base);
+	    intel_renderbuffer_map(intel, &stencilRb->Base);
+            _mesa_insert_stencil(ctx, &depthRb->Base, &stencilRb->Base);
+	    intel_renderbuffer_unmap(intel, &stencilRb->Base);
+	    intel_renderbuffer_unmap(intel, &depthRb->Base);
+            depthRb->PairedStencil = stencilRb->Base.Name;
+            stencilRb->PairedDepth = depthRb->Base.Name;
+         }
+
+      }
+   }
+   else if (depthRb) {
+      /* Depth buffer but no stencil buffer.
+       * We'll use a GL_DEPTH24_STENCIL8 buffer and ignore the stencil bits.
+       */
+      /* can't assert this until storage is allocated:
+         ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+       */
+      /* intel_undo any previous pairing */
+      if (depthRb->PairedStencil) {
+         intel_unpair_depth_stencil(ctx, depthRb);
+      }
+   }
+   else if (stencilRb) {
+      /* Stencil buffer but no depth buffer.
+       * Since h/w doesn't typically support just 8bpp stencil w/out Z,
+       * we'll use a GL_DEPTH24_STENCIL8 buffer and ignore the depth bits.
+       */
+      /* undo any previous pairing */
+      if (stencilRb->PairedDepth) {
+         intel_unpair_depth_stencil(ctx, stencilRb);
+      }
+      if (stencilRb->Base._ActualFormat == GL_STENCIL_INDEX8_EXT) {
+         /* promote buffer to GL_DEPTH24_STENCIL8 for hw rendering */
+         _mesa_promote_stencil(ctx, &stencilRb->Base);
+         ASSERT(stencilRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+      }
+   }
+
+   /* Finally, update the fb->_DepthBuffer and fb->_StencilBuffer fields */
+   _mesa_update_depth_buffer(ctx, fb, BUFFER_DEPTH);
+   if (depthRb && depthRb->PairedStencil)
+      _mesa_update_stencil_buffer(ctx, fb, BUFFER_DEPTH);
+   else
+      _mesa_update_stencil_buffer(ctx, fb, BUFFER_STENCIL);
+
+
+   /* The hardware should use fb->Attachment[BUFFER_DEPTH].Renderbuffer
+    * first, if present, then fb->Attachment[BUFFER_STENCIL].Renderbuffer
+    * if present.
+    */
+}
diff --git a/src/mesa/drivers/dri/intel/intel_depthstencil.h b/src/mesa/drivers/dri/intel/intel_depthstencil.h
new file mode 100644
index 00000000000..740eb0d9895
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_depthstencil.h
@@ -0,0 +1,15 @@
+
+#ifndef INTEL_DEPTH_STENCIL_H
+#define INTEL_DEPTH_STENCIL_H
+
+#include "intel_fbo.h"
+
+extern void
+intel_unpair_depth_stencil(GLcontext * ctx, struct intel_renderbuffer *irb);
+
+extern void
+intel_validate_paired_depth_stencil(GLcontext * ctx,
+                                    struct gl_framebuffer *fb);
+
+
+#endif /* INTEL_DEPTH_STENCIL_H */
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c
new file mode 100644
index 00000000000..7453b96dc5d
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_fbo.c
@@ -0,0 +1,716 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/fbobject.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
+#include "main/context.h"
+#include "main/texformat.h"
+#include "main/texrender.h"
+
+#include "intel_context.h"
+#include "intel_buffers.h"
+#include "intel_depthstencil.h"
+#include "intel_fbo.h"
+#include "intel_mipmap_tree.h"
+#include "intel_regions.h"
+#include "intel_span.h"
+
+
+#define FILE_DEBUG_FLAG DEBUG_FBO
+
+#define INTEL_RB_CLASS 0x12345678
+
+
+/* XXX FBO: move this to intel_context.h (inlined) */
+/**
+ * Return a gl_renderbuffer ptr casted to intel_renderbuffer.
+ * NULL will be returned if the rb isn't really an intel_renderbuffer.
+ * This is determiend by checking the ClassID.
+ */
+struct intel_renderbuffer *
+intel_renderbuffer(struct gl_renderbuffer *rb)
+{
+   struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
+   if (irb && irb->Base.ClassID == INTEL_RB_CLASS) {
+      /*_mesa_warning(NULL, "Returning non-intel Rb\n");*/
+      return irb;
+   }
+   else
+      return NULL;
+}
+
+
+struct intel_renderbuffer *
+intel_get_renderbuffer(struct gl_framebuffer *fb, int attIndex)
+{
+   if (attIndex >= 0)
+      return intel_renderbuffer(fb->Attachment[attIndex].Renderbuffer);
+   else
+      return NULL;
+}
+
+
+void
+intel_flip_renderbuffers(struct intel_framebuffer *intel_fb)
+{
+   int current_page = intel_fb->pf_current_page;
+   int next_page = (current_page + 1) % intel_fb->pf_num_pages;
+   struct gl_renderbuffer *tmp_rb;
+
+   /* Exchange renderbuffers if necessary but make sure their reference counts
+    * are preserved.
+    */
+   if (intel_fb->color_rb[current_page] &&
+       intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer !=
+       &intel_fb->color_rb[current_page]->Base) {
+      tmp_rb = NULL;
+      _mesa_reference_renderbuffer(&tmp_rb,
+	 intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer);
+      tmp_rb = &intel_fb->color_rb[current_page]->Base;
+      _mesa_reference_renderbuffer(
+	 &intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer, tmp_rb);
+      _mesa_reference_renderbuffer(&tmp_rb, NULL);
+   }
+
+   if (intel_fb->color_rb[next_page] &&
+       intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer !=
+       &intel_fb->color_rb[next_page]->Base) {
+      tmp_rb = NULL;
+      _mesa_reference_renderbuffer(&tmp_rb,
+	 intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer);
+      tmp_rb = &intel_fb->color_rb[next_page]->Base;
+      _mesa_reference_renderbuffer(
+	 &intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer, tmp_rb);
+      _mesa_reference_renderbuffer(&tmp_rb, NULL);
+   }
+}
+
+
+struct intel_region *
+intel_get_rb_region(struct gl_framebuffer *fb, GLuint attIndex)
+{
+   struct intel_renderbuffer *irb = intel_get_renderbuffer(fb, attIndex);
+
+   if (irb)
+      return irb->region;
+   else
+      return NULL;
+}
+
+
+
+/**
+ * Create a new framebuffer object.
+ */
+static struct gl_framebuffer *
+intel_new_framebuffer(GLcontext * ctx, GLuint name)
+{
+   /* Only drawable state in intel_framebuffer at this time, just use Mesa's
+    * class
+    */
+   return _mesa_new_framebuffer(ctx, name);
+}
+
+
+static void
+intel_delete_renderbuffer(struct gl_renderbuffer *rb)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);
+
+   ASSERT(irb);
+
+   if (irb->PairedStencil || irb->PairedDepth) {
+      intel_unpair_depth_stencil(ctx, irb);
+   }
+
+   if (irb->span_cache != NULL)
+      _mesa_free(irb->span_cache);
+
+   if (intel && irb->region) {
+      intel_region_release(&irb->region);
+   }
+
+   _mesa_free(irb);
+}
+
+
+
+/**
+ * Return a pointer to a specific pixel in a renderbuffer.
+ */
+static void *
+intel_get_pointer(GLcontext * ctx, struct gl_renderbuffer *rb,
+                  GLint x, GLint y)
+{
+   /* By returning NULL we force all software rendering to go through
+    * the span routines.
+    */
+   return NULL;
+}
+
+
+
+/**
+ * Called via glRenderbufferStorageEXT() to set the format and allocate
+ * storage for a user-created renderbuffer.
+ */
+static GLboolean
+intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
+                                 GLenum internalFormat,
+                                 GLuint width, GLuint height)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);
+   GLboolean softwareBuffer = GL_FALSE;
+   int cpp;
+
+   ASSERT(rb->Name != 0);
+
+   switch (internalFormat) {
+   case GL_R3_G3_B2:
+   case GL_RGB4:
+   case GL_RGB5:
+      rb->_ActualFormat = GL_RGB5;
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->RedBits = 5;
+      rb->GreenBits = 6;
+      rb->BlueBits = 5;
+      cpp = 2;
+      break;
+   case GL_RGB:
+   case GL_RGB8:
+   case GL_RGB10:
+   case GL_RGB12:
+   case GL_RGB16:
+      rb->_ActualFormat = GL_RGB8;
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->RedBits = 8;
+      rb->GreenBits = 8;
+      rb->BlueBits = 8;
+      rb->AlphaBits = 0;
+      cpp = 4;
+      break;
+   case GL_RGBA:
+   case GL_RGBA2:
+   case GL_RGBA4:
+   case GL_RGB5_A1:
+   case GL_RGBA8:
+   case GL_RGB10_A2:
+   case GL_RGBA12:
+   case GL_RGBA16:
+      rb->_ActualFormat = GL_RGBA8;
+      rb->DataType = GL_UNSIGNED_BYTE;
+      rb->RedBits = 8;
+      rb->GreenBits = 8;
+      rb->BlueBits = 8;
+      rb->AlphaBits = 8;
+      cpp = 4;
+      break;
+   case GL_STENCIL_INDEX:
+   case GL_STENCIL_INDEX1_EXT:
+   case GL_STENCIL_INDEX4_EXT:
+   case GL_STENCIL_INDEX8_EXT:
+   case GL_STENCIL_INDEX16_EXT:
+      /* alloc a depth+stencil buffer */
+      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
+      rb->StencilBits = 8;
+      cpp = 4;
+      break;
+   case GL_DEPTH_COMPONENT16:
+      rb->_ActualFormat = GL_DEPTH_COMPONENT16;
+      rb->DataType = GL_UNSIGNED_SHORT;
+      rb->DepthBits = 16;
+      cpp = 2;
+      break;
+   case GL_DEPTH_COMPONENT:
+   case GL_DEPTH_COMPONENT24:
+   case GL_DEPTH_COMPONENT32:
+      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
+      rb->DepthBits = 24;
+      cpp = 4;
+      break;
+   case GL_DEPTH_STENCIL_EXT:
+   case GL_DEPTH24_STENCIL8_EXT:
+      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
+      rb->DepthBits = 24;
+      rb->StencilBits = 8;
+      cpp = 4;
+      break;
+   default:
+      _mesa_problem(ctx,
+                    "Unexpected format in intel_alloc_renderbuffer_storage");
+      return GL_FALSE;
+   }
+
+   intelFlush(ctx);
+
+   /* free old region */
+   if (irb->region) {
+      intel_region_release(&irb->region);
+   }
+
+   /* allocate new memory region/renderbuffer */
+   if (softwareBuffer) {
+      return _mesa_soft_renderbuffer_storage(ctx, rb, internalFormat,
+                                             width, height);
+   }
+   else {
+      /* Choose a pitch to match hardware requirements:
+       */
+      GLuint pitch = ((cpp * width + 63) & ~63) / cpp;
+
+      /* alloc hardware renderbuffer */
+      DBG("Allocating %d x %d Intel RBO (pitch %d)\n", width,
+	  height, pitch);
+
+      irb->region = intel_region_alloc(intel, cpp, width, height, pitch);
+      if (!irb->region)
+         return GL_FALSE;       /* out of memory? */
+
+      ASSERT(irb->region->buffer);
+
+      rb->Width = width;
+      rb->Height = height;
+
+      return GL_TRUE;
+   }
+}
+
+
+
+/**
+ * Called for each hardware renderbuffer when a _window_ is resized.
+ * Just update fields.
+ * Not used for user-created renderbuffers!
+ */
+static GLboolean
+intel_alloc_window_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
+                           GLenum internalFormat, GLuint width, GLuint height)
+{
+   ASSERT(rb->Name == 0);
+   rb->Width = width;
+   rb->Height = height;
+   rb->_ActualFormat = internalFormat;
+
+   return GL_TRUE;
+}
+
+static void
+intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
+		     GLuint width, GLuint height)
+{
+   struct intel_framebuffer *intel_fb = (struct intel_framebuffer*)fb;
+   int i;
+
+   _mesa_resize_framebuffer(ctx, fb, width, height);
+
+   fb->Initialized = GL_TRUE; /* XXX remove someday */
+
+   if (fb->Name != 0) {
+      return;
+   }
+
+   /* Make sure all window system renderbuffers are up to date */
+   for (i = 0; i < 3; i++) {
+      struct gl_renderbuffer *rb = &intel_fb->color_rb[i]->Base;
+
+      /* only resize if size is changing */
+      if (rb && (rb->Width != width || rb->Height != height)) {
+	 rb->AllocStorage(ctx, rb, rb->InternalFormat, width, height);
+      }
+   }
+}
+
+static GLboolean
+intel_nop_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
+                        GLenum internalFormat, GLuint width, GLuint height)
+{
+   _mesa_problem(ctx, "intel_op_alloc_storage should never be called.");
+   return GL_FALSE;
+}
+
+
+void
+intel_renderbuffer_set_region(struct intel_renderbuffer *rb,
+			      struct intel_region *region)
+{
+   struct intel_region *old;
+
+   old = rb->region;
+   rb->region = NULL;
+   intel_region_reference(&rb->region, region);
+   intel_region_release(&old);
+
+   rb->pfPitch = region->pitch;
+}
+
+/**
+ * Create a new intel_renderbuffer which corresponds to an on-screen window,
+ * not a user-created renderbuffer.
+ */
+struct intel_renderbuffer *
+intel_create_renderbuffer(GLenum intFormat)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   struct intel_renderbuffer *irb;
+   const GLuint name = 0;
+
+   irb = CALLOC_STRUCT(intel_renderbuffer);
+   if (!irb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "creating renderbuffer");
+      return NULL;
+   }
+
+   _mesa_init_renderbuffer(&irb->Base, name);
+   irb->Base.ClassID = INTEL_RB_CLASS;
+
+   switch (intFormat) {
+   case GL_RGB5:
+      irb->Base._ActualFormat = GL_RGB5;
+      irb->Base._BaseFormat = GL_RGBA;
+      irb->Base.RedBits = 5;
+      irb->Base.GreenBits = 6;
+      irb->Base.BlueBits = 5;
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
+      break;
+   case GL_RGBA8:
+      irb->Base._ActualFormat = GL_RGBA8;
+      irb->Base._BaseFormat = GL_RGBA;
+      irb->Base.RedBits = 8;
+      irb->Base.GreenBits = 8;
+      irb->Base.BlueBits = 8;
+      irb->Base.AlphaBits = 8;
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
+      break;
+   case GL_STENCIL_INDEX8_EXT:
+      irb->Base._ActualFormat = GL_STENCIL_INDEX8_EXT;
+      irb->Base._BaseFormat = GL_STENCIL_INDEX;
+      irb->Base.StencilBits = 8;
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
+      break;
+   case GL_DEPTH_COMPONENT16:
+      irb->Base._ActualFormat = GL_DEPTH_COMPONENT16;
+      irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
+      irb->Base.DepthBits = 16;
+      irb->Base.DataType = GL_UNSIGNED_SHORT;
+      break;
+   case GL_DEPTH_COMPONENT24:
+      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
+      irb->Base.DepthBits = 24;
+      irb->Base.DataType = GL_UNSIGNED_INT;
+      break;
+   case GL_DEPTH24_STENCIL8_EXT:
+      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT;
+      irb->Base.DepthBits = 24;
+      irb->Base.StencilBits = 8;
+      irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
+      break;
+   default:
+      _mesa_problem(NULL,
+                    "Unexpected intFormat in intel_create_renderbuffer");
+      return NULL;
+   }
+
+   irb->Base.InternalFormat = intFormat;
+
+   /* intel-specific methods */
+   irb->Base.Delete = intel_delete_renderbuffer;
+   irb->Base.AllocStorage = intel_alloc_window_storage;
+   irb->Base.GetPointer = intel_get_pointer;
+
+   return irb;
+}
+
+
+/**
+ * Create a new renderbuffer object.
+ * Typically called via glBindRenderbufferEXT().
+ */
+static struct gl_renderbuffer *
+intel_new_renderbuffer(GLcontext * ctx, GLuint name)
+{
+   /*struct intel_context *intel = intel_context(ctx); */
+   struct intel_renderbuffer *irb;
+
+   irb = CALLOC_STRUCT(intel_renderbuffer);
+   if (!irb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "creating renderbuffer");
+      return NULL;
+   }
+
+   _mesa_init_renderbuffer(&irb->Base, name);
+   irb->Base.ClassID = INTEL_RB_CLASS;
+
+   /* intel-specific methods */
+   irb->Base.Delete = intel_delete_renderbuffer;
+   irb->Base.AllocStorage = intel_alloc_renderbuffer_storage;
+   irb->Base.GetPointer = intel_get_pointer;
+   /* span routines set in alloc_storage function */
+
+   return &irb->Base;
+}
+
+
+/**
+ * Called via glBindFramebufferEXT().
+ */
+static void
+intel_bind_framebuffer(GLcontext * ctx, GLenum target,
+                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
+{
+   if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
+      intel_draw_buffer(ctx, fb);
+      /* Integer depth range depends on depth buffer bits */
+      if (ctx->Driver.DepthRange != NULL)
+	 ctx->Driver.DepthRange(ctx, ctx->Viewport.Near, ctx->Viewport.Far);
+   }
+   else {
+      /* don't need to do anything if target == GL_READ_FRAMEBUFFER_EXT */
+   }
+}
+
+
+/**
+ * Called via glFramebufferRenderbufferEXT().
+ */
+static void
+intel_framebuffer_renderbuffer(GLcontext * ctx,
+                               struct gl_framebuffer *fb,
+                               GLenum attachment, struct gl_renderbuffer *rb)
+{
+   DBG("Intel FramebufferRenderbuffer %u %u\n", fb->Name, rb ? rb->Name : 0);
+
+   intelFlush(ctx);
+
+   _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
+   intel_draw_buffer(ctx, fb);
+}
+
+static GLboolean
+intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb, 
+		     struct gl_texture_image *texImage)
+{
+   if (texImage->TexFormat == &_mesa_texformat_argb8888) {
+      irb->Base._ActualFormat = GL_RGBA8;
+      irb->Base._BaseFormat = GL_RGBA;
+      DBG("Render to RGBA8 texture OK\n");
+   }
+   else if (texImage->TexFormat == &_mesa_texformat_rgb565) {
+      irb->Base._ActualFormat = GL_RGB5;
+      irb->Base._BaseFormat = GL_RGB;
+      DBG("Render to RGB5 texture OK\n");
+   }
+   else if (texImage->TexFormat == &_mesa_texformat_z16) {
+      irb->Base._ActualFormat = GL_DEPTH_COMPONENT16;
+      irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
+      DBG("Render to DEPTH16 texture OK\n");
+   } else if (texImage->TexFormat == &_mesa_texformat_s8_z24) {
+      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT;
+      DBG("Render to DEPTH_STENCIL texture OK\n");
+   }
+   else {
+      DBG("Render to texture BAD FORMAT %d\n",
+	  texImage->TexFormat->MesaFormat);
+      return GL_FALSE;
+   }
+
+   irb->Base.InternalFormat = irb->Base._ActualFormat;
+   irb->Base.Width = texImage->Width;
+   irb->Base.Height = texImage->Height;
+   irb->Base.DataType = GL_UNSIGNED_BYTE;       /* FBO XXX fix */
+   irb->Base.RedBits = texImage->TexFormat->RedBits;
+   irb->Base.GreenBits = texImage->TexFormat->GreenBits;
+   irb->Base.BlueBits = texImage->TexFormat->BlueBits;
+   irb->Base.AlphaBits = texImage->TexFormat->AlphaBits;
+   irb->Base.DepthBits = texImage->TexFormat->DepthBits;
+
+   irb->Base.Delete = intel_delete_renderbuffer;
+   irb->Base.AllocStorage = intel_nop_alloc_storage;
+
+   irb->RenderToTexture = GL_TRUE;
+
+   return GL_TRUE;
+}
+
+/**
+ * When glFramebufferTexture[123]D is called this function sets up the
+ * gl_renderbuffer wrapper around the texture image.
+ * This will have the region info needed for hardware rendering.
+ */
+static struct intel_renderbuffer *
+intel_wrap_texture(GLcontext * ctx, struct gl_texture_image *texImage)
+{
+   const GLuint name = ~0;      /* not significant, but distinct for debugging */
+   struct intel_renderbuffer *irb;
+
+   /* make an intel_renderbuffer to wrap the texture image */
+   irb = CALLOC_STRUCT(intel_renderbuffer);
+   if (!irb) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glFramebufferTexture");
+      return NULL;
+   }
+
+   _mesa_init_renderbuffer(&irb->Base, name);
+   irb->Base.ClassID = INTEL_RB_CLASS;
+
+   if (!intel_update_wrapper(ctx, irb, texImage)) {
+      _mesa_free(irb);
+      return NULL;
+   }
+
+   return irb;
+}
+
+
+/**
+ * Called by glFramebufferTexture[123]DEXT() (and other places) to
+ * prepare for rendering into texture memory.  This might be called
+ * many times to choose different texture levels, cube faces, etc
+ * before intel_finish_render_texture() is ever called.
+ */
+static void
+intel_render_texture(GLcontext * ctx,
+                     struct gl_framebuffer *fb,
+                     struct gl_renderbuffer_attachment *att)
+{
+   struct gl_texture_image *newImage
+      = att->Texture->Image[att->CubeMapFace][att->TextureLevel];
+   struct intel_renderbuffer *irb = intel_renderbuffer(att->Renderbuffer);
+   struct intel_texture_image *intel_image;
+   GLuint imageOffset;
+
+   (void) fb;
+
+   ASSERT(newImage);
+
+   if (newImage->Border != 0) {
+      /* Fallback on drawing to a texture with a border, which won't have a
+       * miptree.
+       */
+       _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
+       _mesa_render_texture(ctx, fb, att);
+       return;
+   } else if (!irb) {
+      irb = intel_wrap_texture(ctx, newImage);
+      if (irb) {
+         /* bind the wrapper to the attachment point */
+         _mesa_reference_renderbuffer(&att->Renderbuffer, &irb->Base);
+      }
+      else {
+         /* fallback to software rendering */
+         _mesa_render_texture(ctx, fb, att);
+         return;
+      }
+   } if (!intel_update_wrapper(ctx, irb, newImage)) {
+       _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
+       _mesa_render_texture(ctx, fb, att);
+       return;
+   }
+
+   DBG("Begin render texture tid %x tex=%u w=%d h=%d refcount=%d\n",
+       _glthread_GetID(),
+       att->Texture->Name, newImage->Width, newImage->Height,
+       irb->Base.RefCount);
+
+   /* point the renderbufer's region to the texture image region */
+   intel_image = intel_texture_image(newImage);
+   if (irb->region != intel_image->mt->region) {
+      if (irb->region)
+	 intel_region_release(&irb->region);
+      intel_region_reference(&irb->region, intel_image->mt->region);
+   }
+
+   /* compute offset of the particular 2D image within the texture region */
+   imageOffset = intel_miptree_image_offset(intel_image->mt,
+                                            att->CubeMapFace,
+                                            att->TextureLevel);
+
+   if (att->Texture->Target == GL_TEXTURE_3D) {
+      const GLuint *offsets = intel_miptree_depth_offsets(intel_image->mt,
+                                                          att->TextureLevel);
+      imageOffset += offsets[att->Zoffset];
+   }
+
+   /* store that offset in the region */
+   intel_image->mt->region->draw_offset = imageOffset;
+
+   /* update drawing region, etc */
+   intel_draw_buffer(ctx, fb);
+}
+
+
+/**
+ * Called by Mesa when rendering to a texture is done.
+ */
+static void
+intel_finish_render_texture(GLcontext * ctx,
+                            struct gl_renderbuffer_attachment *att)
+{
+   struct intel_renderbuffer *irb = intel_renderbuffer(att->Renderbuffer);
+
+   DBG("End render texture (tid %x) tex %u\n", _glthread_GetID(), att->Texture->Name);
+
+   if (irb) {
+      /* just release the region */
+      intel_region_release(&irb->region);
+   }
+   else if (att->Renderbuffer) {
+      /* software fallback */
+      _mesa_finish_render_texture(ctx, att);
+      /* XXX FBO: Need to unmap the buffer (or in intelSpanRenderStart???) */
+   }
+}
+
+
+/**
+ * Do one-time context initializations related to GL_EXT_framebuffer_object.
+ * Hook in device driver functions.
+ */
+void
+intel_fbo_init(struct intel_context *intel)
+{
+   intel->ctx.Driver.NewFramebuffer = intel_new_framebuffer;
+   intel->ctx.Driver.NewRenderbuffer = intel_new_renderbuffer;
+   intel->ctx.Driver.BindFramebuffer = intel_bind_framebuffer;
+   intel->ctx.Driver.FramebufferRenderbuffer = intel_framebuffer_renderbuffer;
+   intel->ctx.Driver.RenderTexture = intel_render_texture;
+   intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture;
+   intel->ctx.Driver.ResizeBuffers = intel_resize_buffers;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.h b/src/mesa/drivers/dri/intel/intel_fbo.h
new file mode 100644
index 00000000000..9d15582d78c
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_fbo.h
@@ -0,0 +1,115 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_FBO_H
+#define INTEL_FBO_H
+
+#include "intel_screen.h"
+
+struct intel_context;
+
+/**
+ * Intel framebuffer, derived from gl_framebuffer.
+ */
+struct intel_framebuffer
+{
+   struct gl_framebuffer Base;
+
+   struct intel_renderbuffer *color_rb[3];
+
+   /* Drawable page flipping state */
+   GLboolean pf_active;
+   GLuint pf_seq;
+   GLint pf_planes;
+   GLint pf_current_page;
+   GLint pf_num_pages;
+
+   /* VBI
+    */
+   GLuint vbl_waited;
+
+   int64_t swap_ust;
+   int64_t swap_missed_ust;
+
+   GLuint swap_count;
+   GLuint swap_missed_count;
+};
+
+
+/**
+ * Intel renderbuffer, derived from gl_renderbuffer.
+ * Note: The PairedDepth and PairedStencil fields use renderbuffer IDs,
+ * not pointers because in some circumstances a deleted renderbuffer could
+ * result in a dangling pointer here.
+ */
+struct intel_renderbuffer
+{
+   struct gl_renderbuffer Base;
+   struct intel_region *region;
+   GLuint pfPitch;              /* possibly paged flipped pitch */
+   GLboolean RenderToTexture;   /* RTT? */
+
+   GLuint PairedDepth;   /**< only used if this is a depth renderbuffer */
+   GLuint PairedStencil; /**< only used if this is a stencil renderbuffer */
+
+   GLuint pf_pending;  /**< sequence number of pending flip */
+
+   GLuint vbl_pending;   /**< vblank sequence number of pending flip */
+
+   uint8_t *span_cache;
+   unsigned long span_cache_offset;
+};
+
+extern struct intel_renderbuffer *intel_renderbuffer(struct gl_renderbuffer
+                                                     *rb);
+
+extern void
+intel_renderbuffer_set_region(struct intel_renderbuffer *irb,
+			      struct intel_region *region);
+
+extern struct intel_renderbuffer *
+intel_create_renderbuffer(GLenum intFormat);
+
+extern void intel_fbo_init(struct intel_context *intel);
+
+
+/* XXX make inline or macro */
+extern struct intel_renderbuffer *intel_get_renderbuffer(struct gl_framebuffer
+                                                         *fb,
+                                                         int attIndex);
+
+extern void intel_flip_renderbuffers(struct intel_framebuffer *intel_fb);
+
+
+/* XXX make inline or macro */
+extern struct intel_region *intel_get_rb_region(struct gl_framebuffer *fb,
+                                                GLuint attIndex);
+
+
+
+
+#endif /* INTEL_FBO_H */
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
new file mode 100644
index 00000000000..bf1c3f03f0e
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
@@ -0,0 +1,492 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "intel_context.h"
+#include "intel_mipmap_tree.h"
+#include "intel_regions.h"
+#include "intel_chipset.h"
+#include "main/enums.h"
+
+#define FILE_DEBUG_FLAG DEBUG_MIPTREE
+
+static GLenum
+target_to_target(GLenum target)
+{
+   switch (target) {
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB:
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
+      return GL_TEXTURE_CUBE_MAP_ARB;
+   default:
+      return target;
+   }
+}
+
+static struct intel_mipmap_tree *
+intel_miptree_create_internal(struct intel_context *intel,
+			      GLenum target,
+			      GLenum internal_format,
+			      GLuint first_level,
+			      GLuint last_level,
+			      GLuint width0,
+			      GLuint height0,
+			      GLuint depth0, GLuint cpp, GLuint compress_byte)
+{
+   GLboolean ok;
+   struct intel_mipmap_tree *mt = calloc(sizeof(*mt), 1);
+
+   DBG("%s target %s format %s level %d..%d\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(target),
+       _mesa_lookup_enum_by_nr(internal_format), first_level, last_level);
+
+   mt->target = target_to_target(target);
+   mt->internal_format = internal_format;
+   mt->first_level = first_level;
+   mt->last_level = last_level;
+   mt->width0 = width0;
+   mt->height0 = height0;
+   mt->depth0 = depth0;
+   mt->cpp = compress_byte ? compress_byte : cpp;
+   mt->compressed = compress_byte ? 1 : 0;
+   mt->refcount = 1; 
+   mt->pitch = 0;
+
+#ifdef I915
+   if (IS_945(intel->intelScreen->deviceID))
+      ok = i945_miptree_layout(intel, mt);
+   else
+      ok = i915_miptree_layout(intel, mt);
+#else
+   ok = brw_miptree_layout(intel, mt);
+#endif
+
+   if (!ok) {
+      free(mt);
+      return NULL;
+   }
+
+   return mt;
+}
+
+struct intel_mipmap_tree *
+intel_miptree_create(struct intel_context *intel,
+		     GLenum target,
+		     GLenum internal_format,
+		     GLuint first_level,
+		     GLuint last_level,
+		     GLuint width0,
+		     GLuint height0,
+		     GLuint depth0, GLuint cpp, GLuint compress_byte)
+{
+   struct intel_mipmap_tree *mt;
+
+   mt = intel_miptree_create_internal(intel, target, internal_format,
+				      first_level, last_level, width0,
+				      height0, depth0, cpp, compress_byte);
+   /*
+    * pitch == 0 || height == 0  indicates the null texture
+    */
+   if (!mt || !mt->pitch || !mt->total_height)
+      return NULL;
+
+   mt->region = intel_region_alloc(intel,
+				   mt->cpp,
+				   mt->pitch,
+				   mt->total_height,
+				   mt->pitch);
+
+   if (!mt->region) {
+       free(mt);
+       return NULL;
+   }
+
+   return mt;
+}
+
+struct intel_mipmap_tree *
+intel_miptree_create_for_region(struct intel_context *intel,
+				GLenum target,
+				GLenum internal_format,
+				GLuint first_level,
+				GLuint last_level,
+				struct intel_region *region,
+				GLuint depth0,
+				GLuint compress_byte)
+{
+   struct intel_mipmap_tree *mt;
+
+   mt = intel_miptree_create_internal(intel, target, internal_format,
+				      first_level, last_level,
+				      region->width, region->height, 1,
+				      region->cpp, compress_byte);
+   if (!mt)
+      return mt;
+#if 0
+   if (mt->pitch != region->pitch) {
+      fprintf(stderr,
+	      "region pitch (%d) doesn't match mipmap tree pitch (%d)\n",
+	      region->pitch, mt->pitch);
+      free(mt);
+      return NULL;
+   }
+#else
+   /* The mipmap tree pitch is aligned to 64 bytes to make sure render
+    * to texture works, but we don't need that for texturing from a
+    * pixmap.  Just override it here. */
+   mt->pitch = region->pitch;
+#endif
+
+   intel_region_reference(&mt->region, region);
+
+   return mt;
+ }
+
+/**
+ * intel_miptree_pitch_align:
+ *
+ * @intel: intel context pointer
+ *
+ * @mt: the miptree to compute pitch alignment for
+ *
+ * @pitch: the natural pitch value
+ *
+ * Given @pitch, compute a larger value which accounts for
+ * any necessary alignment required by the device
+ */
+
+int intel_miptree_pitch_align (struct intel_context *intel,
+			       struct intel_mipmap_tree *mt,
+			       int pitch)
+{
+#ifdef I915
+   GLcontext *ctx = &intel->ctx;
+#endif
+
+   if (!mt->compressed) {
+      int pitch_align;
+
+      if (intel->ttm) {
+	 /* XXX: Align pitch to multiple of 64 bytes for now to allow
+	  * render-to-texture to work in all cases. This should probably be
+	  * replaced at some point by some scheme to only do this when really
+	  * necessary.
+	  */
+	 pitch_align = 64;
+      } else {
+	 pitch_align = 4;
+      }
+
+      pitch = ALIGN(pitch * mt->cpp, pitch_align);
+
+#ifdef I915
+      /* XXX: At least the i915 seems very upset when the pitch is a multiple
+       * of 1024 and sometimes 512 bytes - performance can drop by several
+       * times. Go to the next multiple of the required alignment for now.
+       */
+      if (!(pitch & 511) && 
+	 (pitch + pitch_align) < (1 << ctx->Const.MaxTextureLevels))
+	 pitch += pitch_align;
+#endif
+
+      pitch /= mt->cpp;
+   }
+   return pitch;
+}
+
+void
+intel_miptree_reference(struct intel_mipmap_tree **dst,
+                        struct intel_mipmap_tree *src)
+{
+   src->refcount++;
+   *dst = src;
+   DBG("%s %p refcount now %d\n", __FUNCTION__, src, src->refcount);
+}
+
+void
+intel_miptree_release(struct intel_context *intel,
+                      struct intel_mipmap_tree **mt)
+{
+   if (!*mt)
+      return;
+
+   DBG("%s %p refcount will be %d\n", __FUNCTION__, *mt, (*mt)->refcount - 1);
+   if (--(*mt)->refcount <= 0) {
+      GLuint i;
+
+      DBG("%s deleting %p\n", __FUNCTION__, *mt);
+
+      intel_region_release(&((*mt)->region));
+
+      for (i = 0; i < MAX_TEXTURE_LEVELS; i++)
+         if ((*mt)->level[i].image_offset)
+            free((*mt)->level[i].image_offset);
+
+      free(*mt);
+   }
+   *mt = NULL;
+}
+
+
+
+
+/* Can the image be pulled into a unified mipmap tree.  This mirrors
+ * the completeness test in a lot of ways.
+ *
+ * Not sure whether I want to pass gl_texture_image here.
+ */
+GLboolean
+intel_miptree_match_image(struct intel_mipmap_tree *mt,
+                          struct gl_texture_image *image,
+                          GLuint face, GLuint level)
+{
+   /* Images with borders are never pulled into mipmap trees. 
+    */
+   if (image->Border ||
+       ((image->_BaseFormat == GL_DEPTH_COMPONENT) &&
+        ((image->TexObject->WrapS == GL_CLAMP_TO_BORDER) ||
+         (image->TexObject->WrapT == GL_CLAMP_TO_BORDER)))) 
+      return GL_FALSE;
+
+   if (image->InternalFormat != mt->internal_format ||
+       image->IsCompressed != mt->compressed)
+      return GL_FALSE;
+
+   if (!image->IsCompressed &&
+       !mt->compressed &&
+       image->TexFormat->TexelBytes != mt->cpp)
+      return GL_FALSE;
+
+   /* Test image dimensions against the base level image adjusted for
+    * minification.  This will also catch images not present in the
+    * tree, changed targets, etc.
+    */
+   if (image->Width != mt->level[level].width ||
+       image->Height != mt->level[level].height ||
+       image->Depth != mt->level[level].depth)
+      return GL_FALSE;
+
+   return GL_TRUE;
+}
+
+
+void
+intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
+			     GLuint level,
+			     GLuint nr_images,
+			     GLuint x, GLuint y,
+			     GLuint w, GLuint h, GLuint d)
+{
+   mt->level[level].width = w;
+   mt->level[level].height = h;
+   mt->level[level].depth = d;
+   mt->level[level].level_offset = (x + y * mt->pitch) * mt->cpp;
+   mt->level[level].nr_images = nr_images;
+
+   DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
+       level, w, h, d, x, y, mt->level[level].level_offset);
+
+   /* Not sure when this would happen, but anyway: 
+    */
+   if (mt->level[level].image_offset) {
+      free(mt->level[level].image_offset);
+      mt->level[level].image_offset = NULL;
+   }
+
+   assert(nr_images);
+
+   mt->level[level].image_offset = malloc(nr_images * sizeof(GLuint));
+   mt->level[level].image_offset[0] = 0;
+}
+
+
+
+void
+intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
+			       GLuint level, GLuint img,
+			       GLuint x, GLuint y)
+{
+   if (img == 0 && level == 0)
+      assert(x == 0 && y == 0);
+
+   assert(img < mt->level[level].nr_images);
+
+   mt->level[level].image_offset[img] = (x + y * mt->pitch) * mt->cpp;
+
+   DBG("%s level %d img %d pos %d,%d image_offset %x\n",
+       __FUNCTION__, level, img, x, y, mt->level[level].image_offset[img]);
+}
+
+
+/* Although we use the image_offset[] array to store relative offsets
+ * to cube faces, Mesa doesn't know anything about this and expects
+ * each cube face to be treated as a separate image.
+ *
+ * These functions present that view to mesa:
+ */
+const GLuint *
+intel_miptree_depth_offsets(struct intel_mipmap_tree *mt, GLuint level)
+{
+   static const GLuint zero = 0;
+
+   if (mt->target != GL_TEXTURE_3D || mt->level[level].nr_images == 1)
+      return &zero;
+   else
+      return mt->level[level].image_offset;
+}
+
+
+GLuint
+intel_miptree_image_offset(struct intel_mipmap_tree *mt,
+			   GLuint face, GLuint level)
+{
+   if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
+      return (mt->level[level].level_offset +
+	      mt->level[level].image_offset[face]);
+   else
+      return mt->level[level].level_offset;
+}
+
+
+
+/**
+ * Map a teximage in a mipmap tree.
+ * \param row_stride  returns row stride in bytes
+ * \param image_stride  returns image stride in bytes (for 3D textures).
+ * \param image_offsets pointer to array of pixel offsets from the returned
+ *	  pointer to each depth image
+ * \return address of mapping
+ */
+GLubyte *
+intel_miptree_image_map(struct intel_context * intel,
+                        struct intel_mipmap_tree * mt,
+                        GLuint face,
+                        GLuint level,
+                        GLuint * row_stride, GLuint * image_offsets)
+{
+   DBG("%s \n", __FUNCTION__);
+
+   if (row_stride)
+      *row_stride = mt->pitch * mt->cpp;
+
+   if (mt->target == GL_TEXTURE_3D) {
+      int i;
+
+      for (i = 0; i < mt->level[level].depth; i++)
+	 image_offsets[i] = mt->level[level].image_offset[i] / mt->cpp;
+   } else {
+      assert(mt->level[level].depth == 1);
+      assert(mt->target == GL_TEXTURE_CUBE_MAP ||
+	     mt->level[level].image_offset[0] == 0);
+      image_offsets[0] = 0;
+   }
+
+   return (intel_region_map(intel, mt->region) +
+           intel_miptree_image_offset(mt, face, level));
+}
+
+void
+intel_miptree_image_unmap(struct intel_context *intel,
+                          struct intel_mipmap_tree *mt)
+{
+   DBG("%s\n", __FUNCTION__);
+   intel_region_unmap(intel, mt->region);
+}
+
+
+
+/* Upload data for a particular image.
+ */
+void
+intel_miptree_image_data(struct intel_context *intel,
+			 struct intel_mipmap_tree *dst,
+			 GLuint face,
+			 GLuint level,
+			 void *src,
+			 GLuint src_row_pitch,
+			 GLuint src_image_pitch)
+{
+   GLuint depth = dst->level[level].depth;
+   GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
+   const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
+   GLuint i;
+   GLuint height = 0;
+
+   DBG("%s: %d/%d\n", __FUNCTION__, face, level);
+   for (i = 0; i < depth; i++) {
+      height = dst->level[level].height;
+      if(dst->compressed)
+	 height = (height + 3) / 4;
+      intel_region_data(intel,
+			dst->region,
+			dst_offset + dst_depth_offset[i], /* dst_offset */
+			0, 0,                             /* dstx, dsty */
+			src,
+			src_row_pitch,
+			0, 0,                             /* source x, y */
+			dst->level[level].width, height); /* width, height */
+
+      src += src_image_pitch * dst->cpp;
+   }
+}
+
+extern GLuint intel_compressed_alignment(GLenum);
+/* Copy mipmap image between trees
+ */
+void
+intel_miptree_image_copy(struct intel_context *intel,
+                         struct intel_mipmap_tree *dst,
+                         GLuint face, GLuint level,
+                         struct intel_mipmap_tree *src)
+{
+   GLuint width = src->level[level].width;
+   GLuint height = src->level[level].height;
+   GLuint depth = src->level[level].depth;
+   GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
+   GLuint src_offset = intel_miptree_image_offset(src, face, level);
+   const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
+   const GLuint *src_depth_offset = intel_miptree_depth_offsets(src, level);
+   GLuint i;
+
+   if (dst->compressed) {
+       GLuint alignment = intel_compressed_alignment(dst->internal_format);
+       height = (height + 3) / 4;
+       width = ((width + alignment - 1) & ~(alignment - 1));
+   }
+
+   for (i = 0; i < depth; i++) {
+      intel_region_copy(intel,
+                        dst->region, dst_offset + dst_depth_offset[i],
+                        0,
+                        0,
+                        src->region, src_offset + src_depth_offset[i],
+                        0, 0, width, height);
+   }
+
+}
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.h b/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
new file mode 100644
index 00000000000..c9537dbb9a4
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
@@ -0,0 +1,226 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_MIPMAP_TREE_H
+#define INTEL_MIPMAP_TREE_H
+
+#include "intel_regions.h"
+
+/* A layer on top of the intel_regions code which adds:
+ *
+ * - Code to size and layout a region to hold a set of mipmaps.
+ * - Query to determine if a new image fits in an existing tree.
+ * - More refcounting 
+ *     - maybe able to remove refcounting from intel_region?
+ * - ?
+ *
+ * The fixed mipmap layout of intel hardware where one offset
+ * specifies the position of all images in a mipmap hierachy
+ * complicates the implementation of GL texture image commands,
+ * compared to hardware where each image is specified with an
+ * independent offset.
+ *
+ * In an ideal world, each texture object would be associated with a
+ * single bufmgr buffer or 2d intel_region, and all the images within
+ * the texture object would slot into the tree as they arrive.  The
+ * reality can be a little messier, as images can arrive from the user
+ * with sizes that don't fit in the existing tree, or in an order
+ * where the tree layout cannot be guessed immediately.  
+ * 
+ * This structure encodes an idealized mipmap tree.  The GL image
+ * commands build these where possible, otherwise store the images in
+ * temporary system buffers.
+ */
+
+
+/**
+ * Describes the location of each texture image within a texture region.
+ */
+struct intel_mipmap_level
+{
+   /**
+    * Byte offset to the base of this level.
+    *
+    * This is used for mipmap levels of 1D/2D/3D textures.  However, CUBE
+    * layouts spread images around the whole tree, so the level offset is
+    * always zero in that case.
+    */
+   GLuint level_offset;
+   GLuint width;
+   GLuint height;
+   /** Depth of the mipmap at this level: 1 for 1D/2D/CUBE, n for 3D. */
+   GLuint depth;
+   /** Number of images at this level: 1 for 1D/2D, 6 for CUBE, depth for 3D */
+   GLuint nr_images;
+
+   /**
+    * Byte offset from level_offset to the image for each cube face or depth
+    * level.
+    *
+    * Pretty much have to accept that hardware formats
+    * are going to be so diverse that there is no unified way to
+    * compute the offsets of depth/cube images within a mipmap level,
+    * so have to store them as a lookup table.
+    */
+   GLuint *image_offset;
+};
+
+struct intel_mipmap_tree
+{
+   /* Effectively the key:
+    */
+   GLenum target;
+   GLenum internal_format;
+
+   GLuint first_level;
+   GLuint last_level;
+
+   GLuint width0, height0, depth0; /**< Level zero image dimensions */
+   GLuint cpp;
+   GLboolean compressed;
+
+   /* Derived from the above:
+    */
+   GLuint pitch;
+   GLuint depth_pitch;          /* per-image on i945? */
+   GLuint total_height;
+
+   /* Includes image offset tables:
+    */
+   struct intel_mipmap_level level[MAX_TEXTURE_LEVELS];
+
+   /* The data is held here:
+    */
+   struct intel_region *region;
+
+   /* These are also refcounted:
+    */
+   GLuint refcount;
+};
+
+
+
+struct intel_mipmap_tree *intel_miptree_create(struct intel_context *intel,
+                                               GLenum target,
+                                               GLenum internal_format,
+                                               GLuint first_level,
+                                               GLuint last_level,
+                                               GLuint width0,
+                                               GLuint height0,
+                                               GLuint depth0,
+                                               GLuint cpp,
+                                               GLuint compress_byte);
+
+struct intel_mipmap_tree *
+intel_miptree_create_for_region(struct intel_context *intel,
+				GLenum target,
+				GLenum internal_format,
+				GLuint first_level,
+				GLuint last_level,
+				struct intel_region *region,
+				GLuint depth0,
+				GLuint compress_byte);
+
+int intel_miptree_pitch_align (struct intel_context *intel,
+			       struct intel_mipmap_tree *mt,
+			       int pitch);
+
+void intel_miptree_reference(struct intel_mipmap_tree **dst,
+                             struct intel_mipmap_tree *src);
+
+void intel_miptree_release(struct intel_context *intel,
+                           struct intel_mipmap_tree **mt);
+
+/* Check if an image fits an existing mipmap tree layout
+ */
+GLboolean intel_miptree_match_image(struct intel_mipmap_tree *mt,
+                                    struct gl_texture_image *image,
+                                    GLuint face, GLuint level);
+
+/* Return a pointer to an image within a tree.  Return image stride as
+ * well.
+ */
+GLubyte *intel_miptree_image_map(struct intel_context *intel,
+                                 struct intel_mipmap_tree *mt,
+                                 GLuint face,
+                                 GLuint level,
+                                 GLuint * row_stride, GLuint * image_stride);
+
+void intel_miptree_image_unmap(struct intel_context *intel,
+                               struct intel_mipmap_tree *mt);
+
+
+/* Return the linear offset of an image relative to the start of the
+ * tree:
+ */
+GLuint intel_miptree_image_offset(struct intel_mipmap_tree *mt,
+                                  GLuint face, GLuint level);
+
+/* Return pointers to each 2d slice within an image.  Indexed by depth
+ * value.
+ */
+const GLuint *intel_miptree_depth_offsets(struct intel_mipmap_tree *mt,
+                                          GLuint level);
+
+
+void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
+                                  GLuint level,
+                                  GLuint nr_images,
+                                  GLuint x, GLuint y,
+                                  GLuint w, GLuint h, GLuint d);
+
+void intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
+                                    GLuint level,
+                                    GLuint img, GLuint x, GLuint y);
+
+
+/* Upload an image into a tree
+ */
+void intel_miptree_image_data(struct intel_context *intel,
+                              struct intel_mipmap_tree *dst,
+                              GLuint face,
+                              GLuint level,
+                              void *src,
+                              GLuint src_row_pitch, GLuint src_image_pitch);
+
+/* Copy an image between two trees
+ */
+void intel_miptree_image_copy(struct intel_context *intel,
+                              struct intel_mipmap_tree *dst,
+                              GLuint face, GLuint level,
+                              struct intel_mipmap_tree *src);
+
+/* i915_mipmap_tree.c:
+ */
+GLboolean i915_miptree_layout(struct intel_context *intel,
+			      struct intel_mipmap_tree *mt);
+GLboolean i945_miptree_layout(struct intel_context *intel,
+			      struct intel_mipmap_tree *mt);
+GLboolean brw_miptree_layout(struct intel_context *intel,
+			     struct intel_mipmap_tree *mt);
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
new file mode 100644
index 00000000000..5702ad9bb57
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -0,0 +1,183 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portionsalloc
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/enums.h"
+#include "main/state.h"
+#include "swrast/swrast.h"
+
+#include "intel_context.h"
+#include "intel_pixel.h"
+#include "intel_regions.h"
+
+#define FILE_DEBUG_FLAG DEBUG_PIXEL
+
+static GLenum
+effective_func(GLenum func, GLboolean src_alpha_is_one)
+{
+   if (src_alpha_is_one) {
+      if (func == GL_SRC_ALPHA)
+	 return GL_ONE;
+      if (func == GL_ONE_MINUS_SRC_ALPHA)
+	 return GL_ZERO;
+   }
+
+   return func;
+}
+
+/**
+ * Check if any fragment operations are in effect which might effect
+ * glDraw/CopyPixels.
+ */
+GLboolean
+intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one)
+{
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
+   if (ctx->FragmentProgram._Enabled) {
+      DBG("fallback due to fragment program\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Color.BlendEnabled &&
+       (effective_func(ctx->Color.BlendSrcRGB, src_alpha_is_one) != GL_ONE ||
+	effective_func(ctx->Color.BlendDstRGB, src_alpha_is_one) != GL_ZERO ||
+	ctx->Color.BlendEquationRGB != GL_FUNC_ADD ||
+	effective_func(ctx->Color.BlendSrcA, src_alpha_is_one) != GL_ONE ||
+	effective_func(ctx->Color.BlendDstA, src_alpha_is_one) != GL_ZERO ||
+	ctx->Color.BlendEquationA != GL_FUNC_ADD)) {
+      DBG("fallback due to blend\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Texture._EnabledUnits) {
+      DBG("fallback due to texturing\n");
+      return GL_FALSE;
+   }
+
+   if (!(ctx->Color.ColorMask[0] &&
+	 ctx->Color.ColorMask[1] &&
+	 ctx->Color.ColorMask[2] &&
+	 ctx->Color.ColorMask[3])) {
+      DBG("fallback due to color masking\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Color.AlphaEnabled) {
+      DBG("fallback due to alpha\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Depth.Test) {
+      DBG("fallback due to depth test\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Fog.Enabled) {
+      DBG("fallback due to fog\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->_ImageTransferState) {
+      DBG("fallback due to image transfer\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->Stencil.Enabled) {
+      DBG("fallback due to image stencil\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->RenderMode != GL_RENDER) {
+      DBG("fallback due to render mode\n");
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+
+GLboolean
+intel_check_meta_tex_fragment_ops(GLcontext * ctx)
+{
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
+   /* Some of _ImageTransferState (scale, bias) could be done with
+    * fragment programs on i915.
+    */
+   return !(ctx->_ImageTransferState || ctx->Fog.Enabled ||     /* not done yet */
+            ctx->Texture._EnabledUnits || ctx->FragmentProgram._Enabled);
+}
+
+/* The intel_region struct doesn't really do enough to capture the
+ * format of the pixels in the region.  For now this code assumes that
+ * the region is a display surface and hence is either ARGB8888 or
+ * RGB565.
+ * XXX FBO: If we'd pass in the intel_renderbuffer instead of region, we'd
+ * know the buffer's pixel format.
+ *
+ * \param format  as given to glDraw/ReadPixels
+ * \param type  as given to glDraw/ReadPixels
+ */
+GLboolean
+intel_check_blit_format(struct intel_region * region,
+                        GLenum format, GLenum type)
+{
+   if (region->cpp == 4 &&
+       (type == GL_UNSIGNED_INT_8_8_8_8_REV ||
+        type == GL_UNSIGNED_BYTE) && format == GL_BGRA) {
+      return GL_TRUE;
+   }
+
+   if (region->cpp == 2 &&
+       type == GL_UNSIGNED_SHORT_5_6_5_REV && format == GL_BGR) {
+      return GL_TRUE;
+   }
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s: bad format for blit (cpp %d, type %s format %s)\n",
+              __FUNCTION__, region->cpp,
+              _mesa_lookup_enum_by_nr(type), _mesa_lookup_enum_by_nr(format));
+
+   return GL_FALSE;
+}
+
+
+void
+intelInitPixelFuncs(struct dd_function_table *functions)
+{
+   functions->Accum = _swrast_Accum;
+   if (!getenv("INTEL_NO_BLIT")) {
+      functions->Bitmap = intelBitmap;
+      functions->CopyPixels = intelCopyPixels;
+      functions->DrawPixels = intelDrawPixels;
+#ifdef I915
+      functions->ReadPixels = intelReadPixels;
+#endif
+   }
+}
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.h b/src/mesa/drivers/dri/intel/intel_pixel.h
new file mode 100644
index 00000000000..6fa6effe835
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_pixel.h
@@ -0,0 +1,70 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_PIXEL_H
+#define INTEL_PIXEL_H
+
+#include "main/mtypes.h"
+
+void intelInitPixelFuncs(struct dd_function_table *functions);
+
+GLboolean intel_check_blit_fragment_ops(GLcontext * ctx,
+					GLboolean src_alpha_is_one);
+
+GLboolean intel_check_meta_tex_fragment_ops(GLcontext * ctx);
+
+GLboolean intel_check_blit_format(struct intel_region *region,
+                                  GLenum format, GLenum type);
+
+
+void intelReadPixels(GLcontext * ctx,
+                     GLint x, GLint y,
+                     GLsizei width, GLsizei height,
+                     GLenum format, GLenum type,
+                     const struct gl_pixelstore_attrib *pack,
+                     GLvoid * pixels);
+
+void intelDrawPixels(GLcontext * ctx,
+                     GLint x, GLint y,
+                     GLsizei width, GLsizei height,
+                     GLenum format,
+                     GLenum type,
+                     const struct gl_pixelstore_attrib *unpack,
+                     const GLvoid * pixels);
+
+void intelCopyPixels(GLcontext * ctx,
+                     GLint srcx, GLint srcy,
+                     GLsizei width, GLsizei height,
+                     GLint destx, GLint desty, GLenum type);
+
+void intelBitmap(GLcontext * ctx,
+		 GLint x, GLint y,
+		 GLsizei width, GLsizei height,
+		 const struct gl_pixelstore_attrib *unpack,
+		 const GLubyte * pixels);
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
new file mode 100644
index 00000000000..e3ce1494e5d
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -0,0 +1,358 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portionsalloc
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/colormac.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/bufferobj.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_regions.h"
+#include "intel_buffer_objects.h"
+#include "intel_buffers.h"
+#include "intel_pixel.h"
+#include "intel_reg.h"
+
+
+#define FILE_DEBUG_FLAG DEBUG_PIXEL
+
+
+/* Unlike the other intel_pixel_* functions, the expectation here is
+ * that the incoming data is not in a PBO.  With the XY_TEXT blit
+ * method, there's no benefit haveing it in a PBO, but we could
+ * implement a path based on XY_MONO_SRC_COPY_BLIT which might benefit
+ * PBO bitmaps.  I think they are probably pretty rare though - I
+ * wonder if Xgl uses them?
+ */
+static const GLubyte *map_pbo( GLcontext *ctx,
+			       GLsizei width, GLsizei height,
+			       const struct gl_pixelstore_attrib *unpack,
+			       const GLubyte *bitmap )
+{
+   GLubyte *buf;
+
+   if (!_mesa_validate_pbo_access(2, unpack, width, height, 1,
+				  GL_COLOR_INDEX, GL_BITMAP,
+				  (GLvoid *) bitmap)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,"glBitmap(invalid PBO access)");
+      return NULL;
+   }
+
+   buf = (GLubyte *) ctx->Driver.MapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
+					   GL_READ_ONLY_ARB,
+					   unpack->BufferObj);
+   if (!buf) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glBitmap(PBO is mapped)");
+      return NULL;
+   }
+
+   return ADD_POINTERS(buf, bitmap);
+}
+
+static GLboolean test_bit( const GLubyte *src,
+			    GLuint bit )
+{
+   return (src[bit/8] & (1<<(bit % 8))) ? 1 : 0;
+}
+
+static void set_bit( GLubyte *dest,
+			  GLuint bit )
+{
+   dest[bit/8] |= 1 << (bit % 8);
+}
+
+/* Extract a rectangle's worth of data from the bitmap.  Called
+ * per-cliprect.
+ */
+static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
+			      const struct gl_pixelstore_attrib *unpack,
+			      const GLubyte *bitmap,
+			      GLuint x, GLuint y, 
+			      GLuint w, GLuint h,
+			      GLubyte *dest,
+			      GLuint row_align,
+			      GLboolean invert)
+{
+   GLuint src_offset = (x + unpack->SkipPixels) & 0x7;
+   GLuint mask = unpack->LsbFirst ? 0 : 7;
+   GLuint bit = 0;
+   GLint row, col;
+   GLint first, last;
+   GLint incr;
+   GLuint count = 0;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s %d,%d %dx%d bitmap %dx%d skip %d src_offset %d mask %d\n",
+		   __FUNCTION__, x,y,w,h,width,height,unpack->SkipPixels, src_offset, mask);
+
+   if (invert) {
+      first = h-1;
+      last = 0;
+      incr = -1;
+   }
+   else {
+      first = 0;
+      last = h-1;
+      incr = 1;
+   }
+
+   /* Require that dest be pre-zero'd.
+    */
+   for (row = first; row != (last+incr); row += incr) {
+      const GLubyte *rowsrc = _mesa_image_address2d(unpack, bitmap, 
+						    width, height, 
+						    GL_COLOR_INDEX, GL_BITMAP, 
+						    y + row, x);
+
+      for (col = 0; col < w; col++, bit++) {
+	 if (test_bit(rowsrc, (col + src_offset) ^ mask)) {
+	    set_bit(dest, bit ^ 7);
+	    count++;
+	 }
+      }
+
+      if (row_align)
+	 bit = ALIGN(bit, row_align);
+   }
+
+   return count;
+}
+
+
+
+
+/*
+ * Render a bitmap.
+ */
+static GLboolean
+do_blit_bitmap( GLcontext *ctx, 
+		GLint dstx, GLint dsty,
+		GLsizei width, GLsizei height,
+		const struct gl_pixelstore_attrib *unpack,
+		const GLubyte *bitmap )
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *dst = intel_drawbuf_region(intel);
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   GLfloat tmpColor[4];
+   GLubyte ubcolor[4];
+   GLuint color8888, color565;
+   unsigned int num_cliprects;
+   drm_clip_rect_t *cliprects;
+   int x_off, y_off;
+
+   if (!dst)
+       return GL_FALSE;
+
+   if (unpack->BufferObj->Name) {
+      bitmap = map_pbo(ctx, width, height, unpack, bitmap);
+      if (bitmap == NULL)
+	 return GL_TRUE;	/* even though this is an error, we're done */
+   }
+
+   COPY_4V(tmpColor, ctx->Current.RasterColor);
+
+   if (NEED_SECONDARY_COLOR(ctx)) {
+       ADD_3V(tmpColor, tmpColor, ctx->Current.RasterSecondaryColor);
+   }
+
+   UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[0], tmpColor[0]);
+   UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[1], tmpColor[1]);
+   UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[2], tmpColor[2]);
+   UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[3], tmpColor[3]);
+
+   color8888 = INTEL_PACKCOLOR8888(ubcolor[0], ubcolor[1], ubcolor[2], ubcolor[3]);
+   color565 = INTEL_PACKCOLOR565(ubcolor[0], ubcolor[1], ubcolor[2]);
+
+   if (!intel_check_blit_fragment_ops(ctx, tmpColor[3] == 1.0F))
+      return GL_FALSE;
+
+   LOCK_HARDWARE(intel);
+
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   if (num_cliprects != 0) {
+      drm_clip_rect_t dest_rect;
+      GLint srcx = 0, srcy = 0;
+      GLuint i;
+      GLint orig_dstx = dstx;
+      GLint orig_dsty = dsty;
+
+      /* Clip to buffer bounds and scissor. */
+      if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
+				fb->_Xmax, fb->_Ymax,
+				&dstx, &dsty, &width, &height))
+            goto out;
+
+      /* Convert from GL to hardware coordinates.  Transform original points
+       * along with it so that we can look at cliprects in hw coordinates and
+       * map back to points in the source space.
+       */
+      if (fb->Name == 0) {
+	 /* bitmap to a system framebuffer */
+	 dstx = x_off + dstx;
+	 dsty = y_off + (fb->Height - dsty - height);
+	 orig_dstx = x_off + orig_dstx;
+	 orig_dsty = y_off + (fb->Height - orig_dsty - height);
+      } else {
+	 /* bitmap to a user framebuffer object */
+	 dstx = x_off + dstx;
+	 dsty = y_off + dsty;
+	 orig_dstx = x_off + orig_dstx;
+	 orig_dsty = y_off + orig_dsty;
+      }
+
+      dest_rect.x1 = dstx;
+      dest_rect.y1 = dsty;
+      dest_rect.x2 = dstx + width;
+      dest_rect.y2 = dsty + height;
+
+      for (i = 0; i < num_cliprects; i++) {
+         drm_clip_rect_t rect;
+	 int box_w, box_h;
+	 GLint px, py;
+	 GLuint stipple[32];  
+
+         if (!intel_intersect_cliprects(&rect, &dest_rect, &cliprects[i]))
+            continue;
+
+	 /* Now go back to GL coordinates to figure out what subset of
+	  * the bitmap we are uploading for this cliprect:
+	  */
+	 box_w = rect.x2 - rect.x1;
+	 box_h = rect.y2 - rect.y1;
+	 srcx = rect.x1 - orig_dstx;
+	 srcy = rect.y1 - orig_dsty;
+
+#define DY 32
+#define DX 32
+
+	 /* Then, finally, chop it all into chunks that can be
+	  * digested by hardware:
+	  */
+	 for (py = 0; py < box_h; py += DY) { 
+	    for (px = 0; px < box_w; px += DX) { 
+	       int h = MIN2(DY, box_h - py);
+	       int w = MIN2(DX, box_w - px); 
+	       GLuint sz = ALIGN(ALIGN(w,8) * h, 64)/8;
+	       GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
+		  ctx->Color.LogicOp : GL_COPY;
+
+	       assert(sz <= sizeof(stipple));
+	       memset(stipple, 0, sz);
+
+	       /* May need to adjust this when padding has been introduced in
+		* sz above:
+		*/
+	       if (get_bitmap_rect(width, height, unpack, 
+				   bitmap,
+				   srcx + px, srcy + py, w, h,
+				   (GLubyte *)stipple,
+				   8,
+				   fb->Name == 0 ? GL_TRUE : GL_FALSE) == 0)
+		  continue;
+
+	       /* 
+		*/
+	       intelEmitImmediateColorExpandBlit( intel,
+						  dst->cpp,
+						  (GLubyte *)stipple, 
+						  sz,
+						  (dst->cpp == 2) ? color565 : color8888,
+						  dst->pitch,
+						  dst->buffer,
+						  0,
+						  dst->tiling,
+						  rect.x1 + px,
+						  rect.y2 - (py + h),
+						  w, h,
+						  logic_op);
+	    } 
+	 } 
+      }
+   }
+out:
+   UNLOCK_HARDWARE(intel);
+
+   if (INTEL_DEBUG & DEBUG_SYNC)
+      intel_batchbuffer_flush(intel->batch);
+
+   if (unpack->BufferObj->Name) {
+      /* done with PBO so unmap it now */
+      ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
+                              unpack->BufferObj);
+   }
+
+   return GL_TRUE;
+}
+
+
+
+
+
+/* There are a large number of possible ways to implement bitmap on
+ * this hardware, most of them have some sort of drawback.  Here are a
+ * few that spring to mind:
+ * 
+ * Blit:
+ *    - XY_MONO_SRC_BLT_CMD
+ *         - use XY_SETUP_CLIP_BLT for cliprect clipping.
+ *    - XY_TEXT_BLT
+ *    - XY_TEXT_IMMEDIATE_BLT
+ *         - blit per cliprect, subject to maximum immediate data size.
+ *    - XY_COLOR_BLT 
+ *         - per pixel or run of pixels
+ *    - XY_PIXEL_BLT
+ *         - good for sparse bitmaps
+ *
+ * 3D engine:
+ *    - Point per pixel
+ *    - Translate bitmap to an alpha texture and render as a quad
+ *    - Chop bitmap up into 32x32 squares and render w/polygon stipple.
+ */
+void
+intelBitmap(GLcontext * ctx,
+	    GLint x, GLint y,
+	    GLsizei width, GLsizei height,
+	    const struct gl_pixelstore_attrib *unpack,
+	    const GLubyte * pixels)
+{
+   if (do_blit_bitmap(ctx, x, y, width, height,
+                          unpack, pixels))
+      return;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
+
+   _swrast_Bitmap(ctx, x, y, width, height, unpack, pixels);
+}
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
new file mode 100644
index 00000000000..61d1296c26c
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
@@ -0,0 +1,394 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/state.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_buffers.h"
+#include "intel_blit.h"
+#include "intel_regions.h"
+#include "intel_pixel.h"
+
+#define FILE_DEBUG_FLAG DEBUG_PIXEL
+
+static struct intel_region *
+copypix_src_region(struct intel_context *intel, GLenum type)
+{
+   switch (type) {
+   case GL_COLOR:
+      return intel_readbuf_region(intel);
+   case GL_DEPTH:
+      /* Don't think this is really possible execpt at 16bpp, when we have no stencil.
+       */
+      if (intel->depth_region && intel->depth_region->cpp == 2)
+         return intel->depth_region;
+   case GL_STENCIL:
+      /* Don't think this is really possible. 
+       */
+      break;
+   case GL_DEPTH_STENCIL_EXT:
+      /* Does it matter whether it is stencil/depth or depth/stencil?
+       */
+      return intel->depth_region;
+   default:
+      break;
+   }
+
+   return NULL;
+}
+
+
+/**
+ * Check if any fragment operations are in effect which might effect
+ * glCopyPixels.  Differs from intel_check_blit_fragment_ops in that
+ * we allow Scissor.
+ */
+static GLboolean
+intel_check_copypixel_blit_fragment_ops(GLcontext * ctx)
+{
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
+   /* Could do logicop with the blitter: 
+    */
+   return !(ctx->_ImageTransferState ||
+            ctx->Color.AlphaEnabled ||
+            ctx->Depth.Test ||
+            ctx->Fog.Enabled ||
+            ctx->Stencil.Enabled ||
+            !ctx->Color.ColorMask[0] ||
+            !ctx->Color.ColorMask[1] ||
+            !ctx->Color.ColorMask[2] ||
+            !ctx->Color.ColorMask[3] ||
+            ctx->Texture._EnabledUnits ||
+	    ctx->FragmentProgram._Enabled ||
+	    ctx->Color.BlendEnabled);
+}
+
+#ifdef I915
+/* Doesn't work for overlapping regions.  Could do a double copy or
+ * just fallback.
+ */
+static GLboolean
+do_texture_copypixels(GLcontext * ctx,
+                      GLint srcx, GLint srcy,
+                      GLsizei width, GLsizei height,
+                      GLint dstx, GLint dsty, GLenum type)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *dst = intel_drawbuf_region(intel);
+   struct intel_region *src = copypix_src_region(intel, type);
+   GLenum src_format;
+   GLenum src_type;
+
+   DBG("%s %d,%d %dx%d --> %d,%d\n", __FUNCTION__, 
+       srcx, srcy, width, height, dstx, dsty);
+
+   if (!src || !dst || type != GL_COLOR)
+      return GL_FALSE;
+
+   /* Can't handle overlapping regions.  Don't have sufficient control
+    * over rasterization to pull it off in-place.  Punt on these for
+    * now.
+    * 
+    * XXX: do a copy to a temporary. 
+    */
+   if (src->buffer == dst->buffer) {
+      drm_clip_rect_t srcbox;
+      drm_clip_rect_t dstbox;
+      drm_clip_rect_t tmp;
+
+      srcbox.x1 = srcx;
+      srcbox.y1 = srcy;
+      srcbox.x2 = srcx + width;
+      srcbox.y2 = srcy + height;
+
+      if (ctx->Pixel.ZoomX > 0) {
+	 dstbox.x1 = dstx;
+	 dstbox.x2 = dstx + width * ctx->Pixel.ZoomX;
+      } else {
+	 dstbox.x1 = dstx + width * ctx->Pixel.ZoomX;
+	 dstbox.x2 = dstx;
+      }
+      if (ctx->Pixel.ZoomY > 0) {
+	 dstbox.y1 = dsty;
+	 dstbox.y2 = dsty + height * ctx->Pixel.ZoomY;
+      } else {
+	 dstbox.y1 = dsty + height * ctx->Pixel.ZoomY;
+	 dstbox.y2 = dsty;
+      }
+
+      DBG("src %d,%d %d,%d\n", srcbox.x1, srcbox.y1, srcbox.x2, srcbox.y2);
+      DBG("dst %d,%d %d,%d (%dx%d) (%f,%f)\n", dstbox.x1, dstbox.y1, dstbox.x2, dstbox.y2,
+	  width, height, ctx->Pixel.ZoomX, ctx->Pixel.ZoomY);
+
+      if (intel_intersect_cliprects(&tmp, &srcbox, &dstbox)) {
+         DBG("%s: regions overlap\n", __FUNCTION__);
+         return GL_FALSE;
+      }
+   }
+
+   intelFlush(&intel->ctx);
+
+   intel->vtbl.install_meta_state(intel);
+
+   /* Is this true?  Also will need to turn depth testing on according
+    * to state:
+    */
+   intel->vtbl.meta_no_stencil_write(intel);
+   intel->vtbl.meta_no_depth_write(intel);
+
+   /* Set the 3d engine to draw into the destination region:
+    */
+   intel->vtbl.meta_draw_region(intel, dst, intel->depth_region);
+
+   intel->vtbl.meta_import_pixel_state(intel);
+
+   if (src->cpp == 2) {
+      src_format = GL_RGB;
+      src_type = GL_UNSIGNED_SHORT_5_6_5;
+   }
+   else {
+      src_format = GL_BGRA;
+      src_type = GL_UNSIGNED_BYTE;
+   }
+
+   /* Set the frontbuffer up as a large rectangular texture.
+    */
+   if (!intel->vtbl.meta_tex_rect_source(intel, src->buffer, 0,
+                                         src->pitch,
+                                         src->height, src_format, src_type)) {
+      intel->vtbl.leave_meta_state(intel);
+      return GL_FALSE;
+   }
+
+
+   intel->vtbl.meta_texture_blend_replace(intel);
+
+   LOCK_HARDWARE(intel);
+
+   if (intel->driDrawable->numClipRects) {
+      __DRIdrawablePrivate *dPriv = intel->driDrawable;
+
+
+      srcy = dPriv->h - srcy - height;  /* convert from gl to hardware coords */
+
+      srcx += dPriv->x;
+      srcy += dPriv->y;
+
+      /* Clip against the source region.  This is the only source
+       * clipping we do.  XXX: Just set the texcord wrap mode to clamp
+       * or similar.
+       *
+       */
+      if (0) {
+         GLint orig_x = srcx;
+         GLint orig_y = srcy;
+
+         if (!_mesa_clip_to_region(0, 0, src->pitch, src->height,
+                                   &srcx, &srcy, &width, &height))
+            goto out;
+
+         dstx += srcx - orig_x;
+         dsty += (srcy - orig_y) * ctx->Pixel.ZoomY;
+      }
+
+      /* Just use the regular cliprect mechanism...  Does this need to
+       * even hold the lock???
+       */
+      intel->vtbl.meta_draw_quad(intel,
+				 dstx,
+				 dstx + width * ctx->Pixel.ZoomX,
+				 dPriv->h - (dsty + height * ctx->Pixel.ZoomY),
+				 dPriv->h - (dsty), 0, /* XXX: what z value? */
+				 0x00ff00ff,
+				 srcx, srcx + width, srcy, srcy + height);
+
+    out:
+      intel->vtbl.leave_meta_state(intel);
+      intel_batchbuffer_emit_mi_flush(intel->batch);
+   }
+   UNLOCK_HARDWARE(intel);
+
+   DBG("%s: success\n", __FUNCTION__);
+   return GL_TRUE;
+}
+#endif /* I915 */
+
+
+/**
+ * CopyPixels with the blitter.  Don't support zooming, pixel transfer, etc.
+ */
+static GLboolean
+do_blit_copypixels(GLcontext * ctx,
+                   GLint srcx, GLint srcy,
+                   GLsizei width, GLsizei height,
+                   GLint dstx, GLint dsty, GLenum type)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *dst = intel_drawbuf_region(intel);
+   struct intel_region *src = copypix_src_region(intel, type);
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   struct gl_framebuffer *read_fb = ctx->ReadBuffer;
+   unsigned int num_cliprects;
+   drm_clip_rect_t *cliprects;
+   int x_off, y_off;
+
+   /* Copypixels can be more than a straight copy.  Ensure all the
+    * extra operations are disabled:
+    */
+   if (!intel_check_copypixel_blit_fragment_ops(ctx) ||
+       ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F)
+      return GL_FALSE;
+
+   if (!src || !dst)
+      return GL_FALSE;
+
+
+
+   intelFlush(&intel->ctx);
+
+   LOCK_HARDWARE(intel);
+
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   if (num_cliprects != 0) {
+      GLint delta_x;
+      GLint delta_y;
+      GLint orig_dstx;
+      GLint orig_dsty;
+      GLint orig_srcx;
+      GLint orig_srcy;
+      GLuint i;
+
+      /* XXX: We fail to handle different inversion between read and draw framebuffer. */
+
+      /* Clip to destination buffer. */
+      orig_dstx = dstx;
+      orig_dsty = dsty;
+      if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
+				fb->_Xmax, fb->_Ymax,
+				&dstx, &dsty, &width, &height))
+	 goto out;
+      /* Adjust src coords for our post-clipped destination origin */
+      srcx += dstx - orig_dstx;
+      srcy += dsty - orig_dsty;
+
+      /* Clip to source buffer. */
+      orig_srcx = srcx;
+      orig_srcy = srcy;
+      if (!_mesa_clip_to_region(read_fb->_Xmin, read_fb->_Ymin,
+				read_fb->_Xmax, read_fb->_Ymax,
+				&srcx, &srcy, &width, &height))
+	 goto out;
+      /* Adjust dst coords for our post-clipped source origin */
+      dstx += srcx - orig_srcx;
+      dsty += srcy - orig_srcy;
+
+      /* Convert from GL to hardware coordinates:
+       */
+      if (fb->Name == 0) {
+	 /* copypixels to a system framebuffer */
+	 dstx = x_off + dstx;
+	 dsty = y_off + (fb->Height - dsty - height);
+      } else {
+	 /* copypixels to a user framebuffer object */
+	 dstx = x_off + dstx;
+	 dsty = y_off + dsty;
+      }
+
+      /* Flip source Y if it's a system framebuffer. */
+      if (read_fb->Name == 0) {
+	 srcx = intel->driReadDrawable->x + srcx;
+	 srcy = intel->driReadDrawable->y + (fb->Height - srcy - height);
+      }
+
+      delta_x = srcx - dstx;
+      delta_y = srcy - dsty;
+      /* Could do slightly more clipping: Eg, take the intersection of
+       * the destination cliprects and the read drawable cliprects
+       *
+       * This code will not overwrite other windows, but will
+       * introduce garbage when copying from obscured window regions.
+       */
+      for (i = 0; i < num_cliprects; i++) {
+	 GLint clip_x = dstx;
+	 GLint clip_y = dsty;
+	 GLint clip_w = width;
+	 GLint clip_h = height;
+
+         if (!_mesa_clip_to_region(cliprects[i].x1, cliprects[i].y1,
+				   cliprects[i].x2, cliprects[i].y2,
+				   &clip_x, &clip_y, &clip_w, &clip_h))
+            continue;
+
+         intelEmitCopyBlit(intel, dst->cpp,
+			   src->pitch, src->buffer, 0, src->tiling,
+			   dst->pitch, dst->buffer, 0, dst->tiling,
+			   clip_x + delta_x, clip_y + delta_y, /* srcx, srcy */
+			   clip_x, clip_y, /* dstx, dsty */
+			   clip_w, clip_h,
+			   ctx->Color.ColorLogicOpEnabled ?
+			   ctx->Color.LogicOp : GL_COPY);
+      }
+   }
+out:
+   UNLOCK_HARDWARE(intel);
+
+   DBG("%s: success\n", __FUNCTION__);
+   return GL_TRUE;
+}
+
+
+void
+intelCopyPixels(GLcontext * ctx,
+                GLint srcx, GLint srcy,
+                GLsizei width, GLsizei height,
+                GLint destx, GLint desty, GLenum type)
+{
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   if (do_blit_copypixels(ctx, srcx, srcy, width, height, destx, desty, type))
+      return;
+
+#ifdef I915
+   if (do_texture_copypixels(ctx, srcx, srcy, width, height, destx, desty, type))
+      return;
+#endif
+
+   DBG("fallback to _swrast_CopyPixels\n");
+
+   _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type);
+}
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
new file mode 100644
index 00000000000..8ebbc95a1d0
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
@@ -0,0 +1,430 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portionsalloc
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/bufferobj.h"
+#include "main/teximage.h"
+#include "main/texenv.h"
+#include "main/texobj.h"
+#include "main/texstate.h"
+#include "main/texparam.h"
+#include "main/matrix.h"
+#include "main/varray.h"
+#include "main/attrib.h"
+#include "main/enable.h"
+#include "main/buffers.h"
+#include "main/fbobject.h"
+#include "main/renderbuffer.h"
+#include "main/depth.h"
+#include "main/hash.h"
+#include "main/blend.h"
+#include "glapi/dispatch.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_regions.h"
+#include "intel_pixel.h"
+#include "intel_buffer_objects.h"
+#include "intel_fbo.h"
+
+static GLboolean
+intel_texture_drawpixels(GLcontext * ctx,
+			 GLint x, GLint y,
+			 GLsizei width, GLsizei height,
+			 GLenum format,
+			 GLenum type,
+			 const struct gl_pixelstore_attrib *unpack,
+			 const GLvoid *pixels)
+{
+   GLuint texname;
+   GLfloat vertices[4][4];
+   GLfloat texcoords[4][2];
+
+   /* We're going to mess with texturing with no regard to existing texture
+    * state, so if there is some set up we have to bail.
+    */
+   if (ctx->Texture._EnabledUnits != 0) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels() fallback: texturing enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Can't do textured DrawPixels with a fragment program, unless we were
+    * to generate a new program that sampled our texture and put the results
+    * in the fragment color before the user's program started.
+    */
+   if (ctx->FragmentProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels() fallback: fragment program enabled\n");
+      return GL_FALSE;
+   }
+
+   /* We don't have a way to generate fragments with stencil values which *
+    * will set the resulting stencil value.
+    */
+   if (format == GL_STENCIL_INDEX)
+      return GL_FALSE;
+
+   /* Check that we can load in a texture this big. */
+   if (width > (1 << (ctx->Const.MaxTextureLevels - 1)) ||
+       height > (1 << (ctx->Const.MaxTextureLevels - 1))) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels() fallback: bitmap too large (%dx%d)\n",
+		 width, height);
+      return GL_FALSE;
+   }
+
+   /* To do DEPTH_COMPONENT, we would need to change our setup to not draw to
+    * the color buffer, and sample the texture values into the fragment depth
+    * in a program.
+    */
+   if (format == GL_DEPTH_COMPONENT) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr,
+		 "glDrawPixels() fallback: format == GL_DEPTH_COMPONENT\n");
+      return GL_FALSE;
+   }
+
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+		    GL_CURRENT_BIT);
+   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
+
+   /* XXX: pixel store stuff */
+   _mesa_Disable(GL_POLYGON_STIPPLE);
+
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
+   _mesa_Enable(GL_TEXTURE_2D);
+   _mesa_GenTextures(1, &texname);
+   _mesa_BindTexture(GL_TEXTURE_2D, texname);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
+   /*
+   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_COMBINE_RGB, GL_REPLACE);
+   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_COMBINE_ALPHA, GL_REPLACE);
+   */
+   _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, format,
+		    type, pixels);
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
+
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+
+   /* Create the vertex buffer based on the current raster pos.  The x and y
+    * we're handed are ctx->Current.RasterPos[0,1] rounded to integers.
+    * We also apply the depth.  However, the W component is already multiplied
+    * into ctx->Current.RasterPos[0,1,2] and we can ignore it at this point.
+    */
+   vertices[0][0] = x;
+   vertices[0][1] = y;
+   vertices[0][2] = ctx->Current.RasterPos[2];
+   vertices[0][3] = 1.0;
+   vertices[1][0] = x + width * ctx->Pixel.ZoomX;
+   vertices[1][1] = y;
+   vertices[1][2] = ctx->Current.RasterPos[2];
+   vertices[1][3] = 1.0;
+   vertices[2][0] = x + width * ctx->Pixel.ZoomX;
+   vertices[2][1] = y + height * ctx->Pixel.ZoomY;
+   vertices[2][2] = ctx->Current.RasterPos[2];
+   vertices[2][3] = 1.0;
+   vertices[3][0] = x;
+   vertices[3][1] = y + height * ctx->Pixel.ZoomY;
+   vertices[3][2] = ctx->Current.RasterPos[2];
+   vertices[3][3] = 1.0;
+
+   texcoords[0][0] = 0.0;
+   texcoords[0][1] = 0.0;
+   texcoords[1][0] = 1.0;
+   texcoords[1][1] = 0.0;
+   texcoords[2][0] = 1.0;
+   texcoords[2][1] = 1.0;
+   texcoords[3][0] = 0.0;
+   texcoords[3][1] = 1.0;
+
+   _mesa_VertexPointer(4, GL_FLOAT, 4 * sizeof(GLfloat), &vertices);
+   _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
+   _mesa_Enable(GL_VERTEX_ARRAY);
+   _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
+   CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PopMatrix();
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PopMatrix();
+   _mesa_PopClientAttrib();
+   _mesa_PopAttrib();
+
+   _mesa_DeleteTextures(1, &texname);
+
+   return GL_TRUE;
+}
+
+static GLboolean
+intel_stencil_drawpixels(GLcontext * ctx,
+			 GLint x, GLint y,
+			 GLsizei width, GLsizei height,
+			 GLenum format,
+			 GLenum type,
+			 const struct gl_pixelstore_attrib *unpack,
+			 const GLvoid *pixels)
+{
+   GLuint texname, rb_name, fb_name, old_fb_name;
+   GLfloat vertices[4][2];
+   GLfloat texcoords[4][2];
+   struct intel_renderbuffer *irb;
+   struct intel_renderbuffer *depth_irb;
+   struct gl_renderbuffer *rb;
+   struct gl_pixelstore_attrib old_unpack;
+   GLstencil *stencil_pixels;
+   int row;
+
+   if (format != GL_STENCIL_INDEX)
+      return GL_FALSE;
+
+   /* If there's nothing to write, we're done. */
+   if (ctx->Stencil.WriteMask[0] == 0)
+      return GL_TRUE;
+
+   /* Can't do a per-bit writemask while treating stencil as rgba data. */
+   if ((ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels(STENCIL_INDEX) fallback: "
+		 "stencil mask enabled\n");
+      return GL_FALSE;
+   }
+
+   /* We use FBOs for our wrapping of the depthbuffer into a color
+    * destination.
+    */
+   if (!ctx->Extensions.EXT_framebuffer_object)
+      return GL_FALSE;
+
+   /* We're going to mess with texturing with no regard to existing texture
+    * state, so if there is some set up we have to bail.
+    */
+   if (ctx->Texture._EnabledUnits != 0) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels(STENCIL_INDEX) fallback: "
+		 "texturing enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Can't do textured DrawPixels with a fragment program, unless we were
+    * to generate a new program that sampled our texture and put the results
+    * in the fragment color before the user's program started.
+    */
+   if (ctx->FragmentProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels(STENCIL_INDEX) fallback: "
+		 "fragment program enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Check that we can load in a texture this big. */
+   if (width > (1 << (ctx->Const.MaxTextureLevels - 1)) ||
+       height > (1 << (ctx->Const.MaxTextureLevels - 1))) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glDrawPixels(STENCIL_INDEX) fallback: "
+		 "bitmap too large (%dx%d)\n",
+		 width, height);
+      return GL_FALSE;
+   }
+
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+		    GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
+   old_fb_name = ctx->DrawBuffer->Name;
+
+   _mesa_Disable(GL_POLYGON_STIPPLE);
+   _mesa_Disable(GL_DEPTH_TEST);
+   _mesa_Disable(GL_STENCIL_TEST);
+
+   /* Unpack the supplied stencil values into a ubyte buffer. */
+   assert(sizeof(GLstencil) == sizeof(GLubyte));
+   stencil_pixels = _mesa_malloc(width * height * sizeof(GLstencil));
+   for (row = 0; row < height; row++) {
+      GLvoid *source = _mesa_image_address2d(unpack, pixels,
+					     width, height,
+					     GL_COLOR_INDEX, type,
+					     row, 0);
+      _mesa_unpack_stencil_span(ctx, width, GL_UNSIGNED_BYTE,
+				stencil_pixels +
+				row * width * sizeof(GLstencil),
+				type, source, unpack, ctx->_ImageTransferState);
+   }
+
+   /* Take the current depth/stencil renderbuffer, and make a new one wrapping
+    * it which will be treated as GL_RGBA8 so we can render to it as a color
+    * buffer.
+    */
+   depth_irb = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_DEPTH);
+   irb = intel_create_renderbuffer(GL_RGBA8);
+   rb = &irb->Base;
+   irb->Base.Width = depth_irb->Base.Width;
+   irb->Base.Height = depth_irb->Base.Height;
+   intel_renderbuffer_set_region(irb, depth_irb->region);
+
+   /* Create a name for our renderbuffer, which lets us use other mesa
+    * rb functions for convenience.
+    */
+   _mesa_GenRenderbuffersEXT(1, &rb_name);
+   irb->Base.RefCount++;
+   _mesa_HashInsert(ctx->Shared->RenderBuffers, rb_name, &irb->Base);
+
+   /* Bind the new renderbuffer to the color attachment point. */
+   _mesa_GenFramebuffersEXT(1, &fb_name);
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb_name);
+   _mesa_FramebufferRenderbufferEXT(GL_FRAMEBUFFER_EXT,
+				    GL_COLOR_ATTACHMENT0_EXT,
+				    GL_RENDERBUFFER_EXT,
+				    rb_name);
+   /* Choose to render to the color attachment. */
+   _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT);
+
+   _mesa_DepthMask(GL_FALSE);
+   _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_TRUE);
+
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
+   _mesa_Enable(GL_TEXTURE_2D);
+   _mesa_GenTextures(1, &texname);
+   _mesa_BindTexture(GL_TEXTURE_2D, texname);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
+   old_unpack = ctx->Unpack;
+   ctx->Unpack = ctx->DefaultPacking;
+   _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_INTENSITY, width, height, 0,
+		    GL_RED, GL_UNSIGNED_BYTE, stencil_pixels);
+   ctx->Unpack = old_unpack;
+   _mesa_free(stencil_pixels);
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
+
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+
+   vertices[0][0] = x;
+   vertices[0][1] = y;
+   vertices[1][0] = x + width * ctx->Pixel.ZoomX;
+   vertices[1][1] = y;
+   vertices[2][0] = x + width * ctx->Pixel.ZoomX;
+   vertices[2][1] = y + height * ctx->Pixel.ZoomY;
+   vertices[3][0] = x;
+   vertices[3][1] = y + height * ctx->Pixel.ZoomY;
+
+   texcoords[0][0] = 0.0;
+   texcoords[0][1] = 0.0;
+   texcoords[1][0] = 1.0;
+   texcoords[1][1] = 0.0;
+   texcoords[2][0] = 1.0;
+   texcoords[2][1] = 1.0;
+   texcoords[3][0] = 0.0;
+   texcoords[3][1] = 1.0;
+
+   _mesa_VertexPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &vertices);
+   _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
+   _mesa_Enable(GL_VERTEX_ARRAY);
+   _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
+   CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
+
+   _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, old_fb_name);
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PopMatrix();
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PopMatrix();
+   _mesa_PopClientAttrib();
+   _mesa_PopAttrib();
+
+   _mesa_DeleteTextures(1, &texname);
+   _mesa_DeleteFramebuffersEXT(1, &fb_name);
+   _mesa_DeleteRenderbuffersEXT(1, &rb_name);
+
+   return GL_TRUE;
+}
+
+void
+intelDrawPixels(GLcontext * ctx,
+                GLint x, GLint y,
+                GLsizei width, GLsizei height,
+                GLenum format,
+                GLenum type,
+                const struct gl_pixelstore_attrib *unpack,
+                const GLvoid * pixels)
+{
+   if (intel_texture_drawpixels(ctx, x, y, width, height, format, type,
+				unpack, pixels))
+      return;
+
+   if (intel_stencil_drawpixels(ctx, x, y, width, height, format, type,
+				unpack, pixels))
+      return;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
+
+   if (ctx->FragmentProgram._Current == ctx->FragmentProgram._TexEnvProgram) {
+      /*
+       * We don't want the i915 texenv program to be applied to DrawPixels.
+       * This is really just a performance optimization (mesa will other-
+       * wise happily run the fragment program on each pixel in the image).
+       */
+      struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
+   /* can't just set current frag prog to 0 here as on buffer resize
+      we'll get new state checks which will segfault. Remains a hack. */
+      ctx->FragmentProgram._Current = NULL;
+      ctx->FragmentProgram._UseTexEnvProgram = GL_FALSE;
+      ctx->FragmentProgram._Active = GL_FALSE;
+      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
+                          unpack, pixels );
+      ctx->FragmentProgram._Current = fpSave;
+      ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
+      ctx->FragmentProgram._Active = GL_TRUE;
+      _swrast_InvalidateState(ctx, _NEW_PROGRAM);
+   }
+   else {
+      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
+                          unpack, pixels );
+   }
+}
diff --git a/src/mesa/drivers/dri/intel/intel_reg.h b/src/mesa/drivers/dri/intel/intel_reg.h
new file mode 100644
index 00000000000..57ac8f0cc14
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_reg.h
@@ -0,0 +1,232 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#define CMD_MI				(0x0 << 29)
+#define CMD_2D				(0x2 << 29)
+#define CMD_3D				(0x3 << 29)
+
+#define MI_NOOP				(CMD_MI | 0)
+
+#define MI_BATCH_BUFFER_END		(CMD_MI | 0xA << 23)
+
+#define MI_FLUSH			(CMD_MI | (4 << 23))
+#define FLUSH_MAP_CACHE				(1 << 0)
+#define INHIBIT_FLUSH_RENDER_CACHE		(1 << 2)
+
+/* Stalls command execution waiting for the given events to have occurred. */
+#define MI_WAIT_FOR_EVENT               (CMD_MI | (0x3 << 23))
+#define MI_WAIT_FOR_PLANE_B_FLIP        (1<<6)
+#define MI_WAIT_FOR_PLANE_A_FLIP        (1<<2)
+
+/* p189 */
+#define _3DSTATE_LOAD_STATE_IMMEDIATE_1   (CMD_3D | (0x1d<<24) | (0x04<<16))
+#define I1_LOAD_S(n)                      (1<<(4+n))
+
+#define _3DSTATE_DRAWRECT_INFO		(CMD_3D | (0x1d<<24) | (0x80<<16) | 0x3)
+#define _3DSTATE_DRAWRECT_INFO_I965	(CMD_3D | (3 << 27) | (1 << 24) | 0x2)
+
+/** @{
+ *
+ * PIPE_CONTROL operation, a combination MI_FLUSH and register write with
+ * additional flushing control.
+ */
+#define _3DSTATE_PIPE_CONTROL		(CMD_3D | (3 << 27) | (2 << 24) | 2)
+#define PIPE_CONTROL_NO_WRITE		(0 << 14)
+#define PIPE_CONTROL_WRITE_IMMEDIATE	(1 << 14)
+#define PIPE_CONTROL_WRITE_DEPTH_COUNT	(2 << 14)
+#define PIPE_CONTROL_WRITE_TIMESTAMP	(3 << 14)
+#define PIPE_CONTROL_DEPTH_STALL	(1 << 13)
+#define PIPE_CONTROL_WRITE_FLUSH	(1 << 12)
+#define PIPE_CONTROL_INSTRUCTION_FLUSH	(1 << 11)
+#define PIPE_CONTROL_INTERRUPT_ENABLE	(1 << 8)
+#define PIPE_CONTROL_PPGTT_WRITE	(0 << 2)
+#define PIPE_CONTROL_GLOBAL_GTT_WRITE	(1 << 2)
+
+/** @} */
+
+/** @{
+ * 915 definitions
+ */
+#define S0_VB_OFFSET_MASK		0xffffffc0
+#define S0_AUTO_CACHE_INV_DISABLE	(1<<0)
+/** @} */
+
+/** @{
+ * 830 definitions
+ */
+#define S0_VB_OFFSET_MASK_830		0xffffff80
+#define S0_VB_PITCH_SHIFT_830		1
+#define S0_VB_ENABLE_830		(1<<0)
+/** @} */
+
+#define S1_VERTEX_WIDTH_SHIFT          24
+#define S1_VERTEX_WIDTH_MASK           (0x3f<<24)
+#define S1_VERTEX_PITCH_SHIFT          16
+#define S1_VERTEX_PITCH_MASK           (0x3f<<16)
+
+#define TEXCOORDFMT_2D                 0x0
+#define TEXCOORDFMT_3D                 0x1
+#define TEXCOORDFMT_4D                 0x2
+#define TEXCOORDFMT_1D                 0x3
+#define TEXCOORDFMT_2D_16              0x4
+#define TEXCOORDFMT_4D_16              0x5
+#define TEXCOORDFMT_NOT_PRESENT        0xf
+#define S2_TEXCOORD_FMT0_MASK            0xf
+#define S2_TEXCOORD_FMT1_SHIFT           4
+#define S2_TEXCOORD_FMT(unit, type)    ((type)<<(unit*4))
+#define S2_TEXCOORD_NONE               (~0)
+#define S2_TEX_COUNT_SHIFT_830		12
+#define S2_VERTEX_1_WIDTH_SHIFT_830	0
+#define S2_VERTEX_0_WIDTH_SHIFT_830	6
+/* S3 not interesting */
+
+#define S4_POINT_WIDTH_SHIFT           23
+#define S4_POINT_WIDTH_MASK            (0x1ff<<23)
+#define S4_LINE_WIDTH_SHIFT            19
+#define S4_LINE_WIDTH_ONE              (0x2<<19)
+#define S4_LINE_WIDTH_MASK             (0xf<<19)
+#define S4_FLATSHADE_ALPHA             (1<<18)
+#define S4_FLATSHADE_FOG               (1<<17)
+#define S4_FLATSHADE_SPECULAR          (1<<16)
+#define S4_FLATSHADE_COLOR             (1<<15)
+#define S4_CULLMODE_BOTH	       (0<<13)
+#define S4_CULLMODE_NONE	       (1<<13)
+#define S4_CULLMODE_CW		       (2<<13)
+#define S4_CULLMODE_CCW		       (3<<13)
+#define S4_CULLMODE_MASK	       (3<<13)
+#define S4_VFMT_POINT_WIDTH            (1<<12)
+#define S4_VFMT_SPEC_FOG               (1<<11)
+#define S4_VFMT_COLOR                  (1<<10)
+#define S4_VFMT_DEPTH_OFFSET           (1<<9)
+#define S4_VFMT_XYZ     	       (1<<6)
+#define S4_VFMT_XYZW     	       (2<<6)
+#define S4_VFMT_XY     		       (3<<6)
+#define S4_VFMT_XYW     	       (4<<6)
+#define S4_VFMT_XYZW_MASK              (7<<6)
+#define S4_FORCE_DEFAULT_DIFFUSE       (1<<5)
+#define S4_FORCE_DEFAULT_SPECULAR      (1<<4)
+#define S4_LOCAL_DEPTH_OFFSET_ENABLE   (1<<3)
+#define S4_VFMT_FOG_PARAM              (1<<2)
+#define S4_SPRITE_POINT_ENABLE         (1<<1)
+#define S4_LINE_ANTIALIAS_ENABLE       (1<<0)
+
+#define S4_VFMT_MASK (S4_VFMT_POINT_WIDTH   | 	\
+		      S4_VFMT_SPEC_FOG      |	\
+		      S4_VFMT_COLOR         |	\
+		      S4_VFMT_DEPTH_OFFSET  |	\
+		      S4_VFMT_XYZW_MASK     |	\
+		      S4_VFMT_FOG_PARAM)
+
+
+#define S5_WRITEDISABLE_ALPHA          (1<<31)
+#define S5_WRITEDISABLE_RED            (1<<30)
+#define S5_WRITEDISABLE_GREEN          (1<<29)
+#define S5_WRITEDISABLE_BLUE           (1<<28)
+#define S5_WRITEDISABLE_MASK           (0xf<<28)
+#define S5_FORCE_DEFAULT_POINT_SIZE    (1<<27)
+#define S5_LAST_PIXEL_ENABLE           (1<<26)
+#define S5_GLOBAL_DEPTH_OFFSET_ENABLE  (1<<25)
+#define S5_FOG_ENABLE                  (1<<24)
+#define S5_STENCIL_REF_SHIFT           16
+#define S5_STENCIL_REF_MASK            (0xff<<16)
+#define S5_STENCIL_TEST_FUNC_SHIFT     13
+#define S5_STENCIL_TEST_FUNC_MASK      (0x7<<13)
+#define S5_STENCIL_FAIL_SHIFT          10
+#define S5_STENCIL_FAIL_MASK           (0x7<<10)
+#define S5_STENCIL_PASS_Z_FAIL_SHIFT   7
+#define S5_STENCIL_PASS_Z_FAIL_MASK    (0x7<<7)
+#define S5_STENCIL_PASS_Z_PASS_SHIFT   4
+#define S5_STENCIL_PASS_Z_PASS_MASK    (0x7<<4)
+#define S5_STENCIL_WRITE_ENABLE        (1<<3)
+#define S5_STENCIL_TEST_ENABLE         (1<<2)
+#define S5_COLOR_DITHER_ENABLE         (1<<1)
+#define S5_LOGICOP_ENABLE              (1<<0)
+
+
+#define S6_ALPHA_TEST_ENABLE           (1<<31)
+#define S6_ALPHA_TEST_FUNC_SHIFT       28
+#define S6_ALPHA_TEST_FUNC_MASK        (0x7<<28)
+#define S6_ALPHA_REF_SHIFT             20
+#define S6_ALPHA_REF_MASK              (0xff<<20)
+#define S6_DEPTH_TEST_ENABLE           (1<<19)
+#define S6_DEPTH_TEST_FUNC_SHIFT       16
+#define S6_DEPTH_TEST_FUNC_MASK        (0x7<<16)
+#define S6_CBUF_BLEND_ENABLE           (1<<15)
+#define S6_CBUF_BLEND_FUNC_SHIFT       12
+#define S6_CBUF_BLEND_FUNC_MASK        (0x7<<12)
+#define S6_CBUF_SRC_BLEND_FACT_SHIFT   8
+#define S6_CBUF_SRC_BLEND_FACT_MASK    (0xf<<8)
+#define S6_CBUF_DST_BLEND_FACT_SHIFT   4
+#define S6_CBUF_DST_BLEND_FACT_MASK    (0xf<<4)
+#define S6_DEPTH_WRITE_ENABLE          (1<<3)
+#define S6_COLOR_WRITE_ENABLE          (1<<2)
+#define S6_TRISTRIP_PV_SHIFT           0
+#define S6_TRISTRIP_PV_MASK            (0x3<<0)
+
+#define S7_DEPTH_OFFSET_CONST_MASK     ~0
+
+/* Primitive dispatch on 830-945 */
+#define _3DPRIMITIVE			(CMD_3D | (0x1f << 24))
+#define PRIM_INDIRECT            (1<<23)
+#define PRIM_INLINE              (0<<23)
+#define PRIM_INDIRECT_SEQUENTIAL (0<<17)
+#define PRIM_INDIRECT_ELTS       (1<<17)
+
+#define PRIM3D_TRILIST		(0x0<<18)
+#define PRIM3D_TRISTRIP 	(0x1<<18)
+#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
+#define PRIM3D_TRIFAN		(0x3<<18)
+#define PRIM3D_POLY		(0x4<<18)
+#define PRIM3D_LINELIST 	(0x5<<18)
+#define PRIM3D_LINESTRIP	(0x6<<18)
+#define PRIM3D_RECTLIST 	(0x7<<18)
+#define PRIM3D_POINTLIST	(0x8<<18)
+#define PRIM3D_DIB		(0x9<<18)
+#define PRIM3D_MASK		(0x1f<<18)
+
+#define XY_SETUP_BLT_CMD		(CMD_2D | (0x01 << 22) | 6)
+
+#define XY_COLOR_BLT_CMD		(CMD_2D | (0x50 << 22) | 4)
+
+#define XY_SRC_COPY_BLT_CMD             (CMD_2D | (0x53 << 22) | 6)
+
+#define XY_TEXT_IMMEDIATE_BLIT_CMD	(CMD_2D | (0x31 << 22))
+# define XY_TEXT_BYTE_PACKED		(1 << 16)
+
+/* BR00 */
+#define XY_BLT_WRITE_ALPHA	(1 << 21)
+#define XY_BLT_WRITE_RGB	(1 << 20)
+#define XY_SRC_TILED		(1 << 15)
+#define XY_DST_TILED		(1 << 11)
+
+/* BR13 */
+#define BR13_565		(0x1 << 24)
+#define BR13_8888		(0x3 << 24)
+
+#define FENCE_LINEAR 0
+#define FENCE_XMAJOR 1
+#define FENCE_YMAJOR 2
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
new file mode 100644
index 00000000000..8dbcc3050ee
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -0,0 +1,569 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/* Provide additional functionality on top of bufmgr buffers:
+ *   - 2d semantics and blit operations
+ *   - refcounting of buffers for multiple images in a buffer.
+ *   - refcounting of buffer mappings.
+ *   - some logic for moving the buffers to the best memory pools for
+ *     given operations.
+ *
+ * Most of this is to make it easier to implement the fixed-layout
+ * mipmap tree required by intel hardware in the face of GL's
+ * programming interface where each image can be specifed in random
+ * order and it isn't clear what layout the tree should have until the
+ * last moment.
+ */
+
+#include <sys/ioctl.h>
+#include <errno.h>
+
+#include "intel_context.h"
+#include "intel_regions.h"
+#include "intel_blit.h"
+#include "intel_buffer_objects.h"
+#include "intel_bufmgr.h"
+#include "intel_batchbuffer.h"
+#include "intel_chipset.h"
+
+#define FILE_DEBUG_FLAG DEBUG_REGION
+
+/* XXX: Thread safety?
+ */
+GLubyte *
+intel_region_map(struct intel_context *intel, struct intel_region *region)
+{
+   DBG("%s\n", __FUNCTION__);
+   if (!region->map_refcount++) {
+      if (region->pbo)
+         intel_region_cow(intel, region);
+
+      dri_bo_map(region->buffer, GL_TRUE);
+      region->map = region->buffer->virtual;
+   }
+
+   return region->map;
+}
+
+void
+intel_region_unmap(struct intel_context *intel, struct intel_region *region)
+{
+   DBG("%s\n", __FUNCTION__);
+   if (!--region->map_refcount) {
+      dri_bo_unmap(region->buffer);
+      region->map = NULL;
+   }
+}
+
+static struct intel_region *
+intel_region_alloc_internal(struct intel_context *intel,
+			    GLuint cpp,
+			    GLuint width, GLuint height, GLuint pitch,
+			    dri_bo *buffer)
+{
+   struct intel_region *region;
+
+   DBG("%s\n", __FUNCTION__);
+
+   if (buffer == NULL)
+      return NULL;
+
+   region = calloc(sizeof(*region), 1);
+   region->cpp = cpp;
+   region->width = width;
+   region->height = height;
+   region->pitch = pitch;
+   region->refcount = 1;
+   region->buffer = buffer;
+
+   /* Default to no tiling */
+   region->tiling = I915_TILING_NONE;
+   region->bit_6_swizzle = I915_BIT_6_SWIZZLE_NONE;
+
+   return region;
+}
+
+struct intel_region *
+intel_region_alloc(struct intel_context *intel,
+                   GLuint cpp, GLuint width, GLuint height, GLuint pitch)
+{
+   dri_bo *buffer;
+
+   buffer = dri_bo_alloc(intel->bufmgr, "region",
+			 pitch * cpp * height, 64);
+
+   return intel_region_alloc_internal(intel, cpp, width, height, pitch, buffer);
+}
+
+struct intel_region *
+intel_region_alloc_for_handle(struct intel_context *intel,
+			      GLuint cpp,
+			      GLuint width, GLuint height, GLuint pitch,
+			      GLuint handle, const char *name)
+{
+   struct intel_region *region;
+   dri_bo *buffer;
+   int ret;
+
+   buffer = intel_bo_gem_create_from_name(intel->bufmgr, name, handle);
+
+   region = intel_region_alloc_internal(intel, cpp,
+					width, height, pitch, buffer);
+   if (region == NULL)
+      return region;
+
+   ret = dri_bo_get_tiling(region->buffer, &region->tiling,
+			   &region->bit_6_swizzle);
+   if (ret != 0) {
+      fprintf(stderr, "Couldn't get tiling of buffer %d (%s): %s\n",
+	      handle, name, strerror(-ret));
+      intel_region_release(&region);
+      return NULL;
+   }
+
+   return region;
+}
+
+void
+intel_region_reference(struct intel_region **dst, struct intel_region *src)
+{
+   if (src)
+      DBG("%s %d\n", __FUNCTION__, src->refcount);
+
+   assert(*dst == NULL);
+   if (src) {
+      src->refcount++;
+      *dst = src;
+   }
+}
+
+void
+intel_region_release(struct intel_region **region_handle)
+{
+   struct intel_region *region = *region_handle;
+
+   if (region == NULL)
+      return;
+
+   DBG("%s %d\n", __FUNCTION__, region->refcount - 1);
+
+   ASSERT(region->refcount > 0);
+   region->refcount--;
+
+   if (region->refcount == 0) {
+      assert(region->map_refcount == 0);
+
+      if (region->pbo)
+	 region->pbo->region = NULL;
+      region->pbo = NULL;
+      dri_bo_unreference(region->buffer);
+
+      if (region->classic_map != NULL) {
+	 drmUnmap(region->classic_map,
+			region->pitch * region->cpp * region->height);
+      }
+
+      free(region);
+   }
+   *region_handle = NULL;
+}
+
+/*
+ * XXX Move this into core Mesa?
+ */
+void
+_mesa_copy_rect(GLubyte * dst,
+                GLuint cpp,
+                GLuint dst_pitch,
+                GLuint dst_x,
+                GLuint dst_y,
+                GLuint width,
+                GLuint height,
+                const GLubyte * src,
+                GLuint src_pitch, GLuint src_x, GLuint src_y)
+{
+   GLuint i;
+
+   dst_pitch *= cpp;
+   src_pitch *= cpp;
+   dst += dst_x * cpp;
+   src += src_x * cpp;
+   dst += dst_y * dst_pitch;
+   src += src_y * dst_pitch;
+   width *= cpp;
+
+   if (width == dst_pitch && width == src_pitch)
+      memcpy(dst, src, height * width);
+   else {
+      for (i = 0; i < height; i++) {
+         memcpy(dst, src, width);
+         dst += dst_pitch;
+         src += src_pitch;
+      }
+   }
+}
+
+
+/* Upload data to a rectangular sub-region.  Lots of choices how to do this:
+ *
+ * - memcpy by span to current destination
+ * - upload data as new buffer and blit
+ *
+ * Currently always memcpy.
+ */
+void
+intel_region_data(struct intel_context *intel,
+                  struct intel_region *dst,
+                  GLuint dst_offset,
+                  GLuint dstx, GLuint dsty,
+                  const void *src, GLuint src_pitch,
+                  GLuint srcx, GLuint srcy, GLuint width, GLuint height)
+{
+   GLboolean locked = GL_FALSE;
+
+   DBG("%s\n", __FUNCTION__);
+
+   if (intel == NULL)
+      return;
+
+   if (dst->pbo) {
+      if (dstx == 0 &&
+          dsty == 0 && width == dst->pitch && height == dst->height)
+         intel_region_release_pbo(intel, dst);
+      else
+         intel_region_cow(intel, dst);
+   }
+
+   if (!intel->locked) {
+      LOCK_HARDWARE(intel);
+      locked = GL_TRUE;
+   }
+
+   _mesa_copy_rect(intel_region_map(intel, dst) + dst_offset,
+                   dst->cpp,
+                   dst->pitch,
+                   dstx, dsty, width, height, src, src_pitch, srcx, srcy);
+
+   intel_region_unmap(intel, dst);
+
+   if (locked)
+      UNLOCK_HARDWARE(intel);
+
+}
+
+/* Copy rectangular sub-regions. Need better logic about when to
+ * push buffers into AGP - will currently do so whenever possible.
+ */
+void
+intel_region_copy(struct intel_context *intel,
+                  struct intel_region *dst,
+                  GLuint dst_offset,
+                  GLuint dstx, GLuint dsty,
+                  struct intel_region *src,
+                  GLuint src_offset,
+                  GLuint srcx, GLuint srcy, GLuint width, GLuint height)
+{
+   DBG("%s\n", __FUNCTION__);
+
+   if (intel == NULL)
+      return;
+
+   if (dst->pbo) {
+      if (dstx == 0 &&
+          dsty == 0 && width == dst->pitch && height == dst->height)
+         intel_region_release_pbo(intel, dst);
+      else
+         intel_region_cow(intel, dst);
+   }
+
+   assert(src->cpp == dst->cpp);
+
+   intelEmitCopyBlit(intel,
+                     dst->cpp,
+                     src->pitch, src->buffer, src_offset, src->tiling,
+                     dst->pitch, dst->buffer, dst_offset, dst->tiling,
+                     srcx, srcy, dstx, dsty, width, height,
+		     GL_COPY);
+}
+
+/* Fill a rectangular sub-region.  Need better logic about when to
+ * push buffers into AGP - will currently do so whenever possible.
+ */
+void
+intel_region_fill(struct intel_context *intel,
+                  struct intel_region *dst,
+                  GLuint dst_offset,
+                  GLuint dstx, GLuint dsty,
+                  GLuint width, GLuint height, GLuint color)
+{
+   DBG("%s\n", __FUNCTION__);
+
+   if (intel == NULL)
+      return;   
+
+   if (dst->pbo) {
+      if (dstx == 0 &&
+          dsty == 0 && width == dst->pitch && height == dst->height)
+         intel_region_release_pbo(intel, dst);
+      else
+         intel_region_cow(intel, dst);
+   }
+
+   intelEmitFillBlit(intel,
+                     dst->cpp,
+                     dst->pitch, dst->buffer, dst_offset, dst->tiling,
+                     dstx, dsty, width, height, color);
+}
+
+/* Attach to a pbo, discarding our data.  Effectively zero-copy upload
+ * the pbo's data.
+ */
+void
+intel_region_attach_pbo(struct intel_context *intel,
+                        struct intel_region *region,
+                        struct intel_buffer_object *pbo)
+{
+   if (region->pbo == pbo)
+      return;
+
+   /* If there is already a pbo attached, break the cow tie now.
+    * Don't call intel_region_release_pbo() as that would
+    * unnecessarily allocate a new buffer we would have to immediately
+    * discard.
+    */
+   if (region->pbo) {
+      region->pbo->region = NULL;
+      region->pbo = NULL;
+   }
+
+   if (region->buffer) {
+      dri_bo_unreference(region->buffer);
+      region->buffer = NULL;
+   }
+
+   region->pbo = pbo;
+   region->pbo->region = region;
+   dri_bo_reference(pbo->buffer);
+   region->buffer = pbo->buffer;
+}
+
+
+/* Break the COW tie to the pbo and allocate a new buffer.
+ * The pbo gets to keep the data.
+ */
+void
+intel_region_release_pbo(struct intel_context *intel,
+                         struct intel_region *region)
+{
+   assert(region->buffer == region->pbo->buffer);
+   region->pbo->region = NULL;
+   region->pbo = NULL;
+   dri_bo_unreference(region->buffer);
+   region->buffer = NULL;
+
+   region->buffer = dri_bo_alloc(intel->bufmgr, "region",
+				 region->pitch * region->cpp * region->height,
+				 64);
+}
+
+/* Break the COW tie to the pbo.  Both the pbo and the region end up
+ * with a copy of the data.
+ */
+void
+intel_region_cow(struct intel_context *intel, struct intel_region *region)
+{
+   struct intel_buffer_object *pbo = region->pbo;
+   GLboolean was_locked = intel->locked;
+
+   if (intel == NULL)
+      return;
+
+   intel_region_release_pbo(intel, region);
+
+   assert(region->cpp * region->pitch * region->height == pbo->Base.Size);
+
+   DBG("%s (%d bytes)\n", __FUNCTION__, pbo->Base.Size);
+
+   /* Now blit from the texture buffer to the new buffer: 
+    */
+
+   was_locked = intel->locked;
+   if (!was_locked)
+      LOCK_HARDWARE(intel);
+
+   intelEmitCopyBlit(intel,
+		     region->cpp,
+		     region->pitch, region->buffer, 0, region->tiling,
+		     region->pitch, pbo->buffer, 0, region->tiling,
+		     0, 0, 0, 0,
+		     region->pitch, region->height,
+		     GL_COPY);
+
+   if (!was_locked)
+      UNLOCK_HARDWARE(intel);
+}
+
+dri_bo *
+intel_region_buffer(struct intel_context *intel,
+                    struct intel_region *region, GLuint flag)
+{
+   if (region->pbo) {
+      if (flag == INTEL_WRITE_PART)
+         intel_region_cow(intel, region);
+      else if (flag == INTEL_WRITE_FULL)
+         intel_region_release_pbo(intel, region);
+   }
+
+   return region->buffer;
+}
+
+static struct intel_region *
+intel_recreate_static(struct intel_context *intel,
+		      const char *name,
+		      struct intel_region *region,
+		      intelRegion *region_desc)
+{
+   intelScreenPrivate *intelScreen = intel->intelScreen;
+   int ret;
+
+   if (region == NULL) {
+      region = calloc(sizeof(*region), 1);
+      region->refcount = 1;
+   }
+
+   if (intel->ctx.Visual.rgbBits == 24)
+      region->cpp = 4;
+   else
+      region->cpp = intel->ctx.Visual.rgbBits / 8;
+   region->pitch = intelScreen->pitch;
+   region->height = intelScreen->height;     /* needed? */
+
+   if (region->buffer != NULL) {
+      dri_bo_unreference(region->buffer);
+      region->buffer = NULL;
+   }
+
+   if (intel->ttm) {
+      assert(region_desc->bo_handle != -1);
+      region->buffer = intel_bo_gem_create_from_name(intel->bufmgr,
+						     name,
+						     region_desc->bo_handle);
+
+      ret = dri_bo_get_tiling(region->buffer, &region->tiling,
+			      &region->bit_6_swizzle);
+      if (ret != 0) {
+	 fprintf(stderr, "Couldn't get tiling of buffer %d (%s): %s\n",
+		 region_desc->bo_handle, name, strerror(-ret));
+	 intel_region_release(&region);
+	 return NULL;
+      }
+   } else {
+      if (region->classic_map != NULL) {
+	 drmUnmap(region->classic_map,
+		  region->pitch * region->cpp * region->height);
+	 region->classic_map = NULL;
+      }
+      ret = drmMap(intel->driFd, region_desc->handle,
+		   region->pitch * region->cpp * region->height,
+		   &region->classic_map);
+      if (ret != 0) {
+	 fprintf(stderr, "Failed to drmMap %s buffer\n", name);
+	 free(region);
+	 return NULL;
+      }
+
+      region->buffer = intel_bo_fake_alloc_static(intel->bufmgr,
+						  name,
+						  region_desc->offset,
+						  region->pitch * region->cpp *
+						  region->height,
+						  region->classic_map);
+
+      /* The sarea just gives us a boolean for whether it's tiled or not,
+       * instead of which tiling mode it is.  Guess.
+       */
+      if (region_desc->tiled) {
+	 if (IS_965(intel->intelScreen->deviceID) &&
+	     region_desc == &intelScreen->depth)
+	    region->tiling = I915_TILING_Y;
+	 else
+	    region->tiling = I915_TILING_X;
+      } else {
+	 region->tiling = I915_TILING_NONE;
+      }
+
+      region->bit_6_swizzle = I915_BIT_6_SWIZZLE_NONE;
+   }
+
+   assert(region->buffer != NULL);
+
+   return region;
+}
+
+/**
+ * Create intel_region structs to describe the static front, back, and depth
+ * buffers created by the xserver.
+ *
+ * Although FBO's mean we now no longer use these as render targets in
+ * all circumstances, they won't go away until the back and depth
+ * buffers become private, and the front buffer will remain even then.
+ *
+ * Note that these don't allocate video memory, just describe
+ * allocations alread made by the X server.
+ */
+void
+intel_recreate_static_regions(struct intel_context *intel)
+{
+   intelScreenPrivate *intelScreen = intel->intelScreen;
+
+   intel->front_region =
+      intel_recreate_static(intel, "front",
+			    intel->front_region,
+			    &intelScreen->front);
+
+   intel->back_region =
+      intel_recreate_static(intel, "back",
+			    intel->back_region,
+			    &intelScreen->back);
+
+#ifdef I915
+   if (intelScreen->third.handle) {
+      intel->third_region =
+	 intel_recreate_static(intel, "third",
+			       intel->third_region,
+			       &intelScreen->third);
+   }
+#endif /* I915 */
+
+   /* Still assumes front.cpp == depth.cpp.  We can kill this when we move to
+    * private buffers.
+    */
+   intel->depth_region =
+      intel_recreate_static(intel, "depth",
+			    intel->depth_region,
+			    &intelScreen->depth);
+}
diff --git a/src/mesa/drivers/dri/intel/intel_regions.h b/src/mesa/drivers/dri/intel/intel_regions.h
new file mode 100644
index 00000000000..4b120ba4cee
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_regions.h
@@ -0,0 +1,151 @@
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_REGIONS_H
+#define INTEL_REGIONS_H
+
+/** @file intel_regions.h
+ *
+ * Structure definitions and prototypes for intel_region handling, which is
+ * the basic structure for rectangular collections of pixels stored in a dri_bo.
+ */
+
+#include <xf86drm.h>
+
+#include "main/mtypes.h"
+#include "intel_bufmgr.h"
+
+struct intel_context;
+struct intel_buffer_object;
+
+/**
+ * A layer on top of the bufmgr buffers that adds a few useful things:
+ *
+ * - Refcounting for local buffer references.
+ * - Refcounting for buffer maps
+ * - Buffer dimensions - pitch and height.
+ * - Blitter commands for copying 2D regions between buffers. (really???)
+ */
+struct intel_region
+{
+   dri_bo *buffer;  /**< buffer manager's buffer */
+   GLuint refcount; /**< Reference count for region */
+   GLuint cpp;      /**< bytes per pixel */
+   GLuint width;    /**< in pixels */
+   GLuint height;   /**< in pixels */
+   GLuint pitch;    /**< in pixels */
+   GLubyte *map;    /**< only non-NULL when region is actually mapped */
+   GLuint map_refcount;  /**< Reference count for mapping */
+
+   GLuint draw_offset; /**< Offset of drawing address within the region */
+   uint32_t tiling; /**< Which tiling mode the region is in */
+   uint32_t bit_6_swizzle; /**< GEM flag for address swizzling requirement */
+   drmAddress classic_map; /**< drmMap of the region when not in GEM mode */
+   struct intel_buffer_object *pbo;     /* zero-copy uploads */
+};
+
+
+/* Allocate a refcounted region.  Pointers to regions should only be
+ * copied by calling intel_reference_region().
+ */
+struct intel_region *intel_region_alloc(struct intel_context *intel,
+                                        GLuint cpp, GLuint width,
+                                        GLuint height, GLuint pitch);
+
+struct intel_region *
+intel_region_alloc_for_handle(struct intel_context *intel,
+			      GLuint cpp,
+			      GLuint width, GLuint height, GLuint pitch,
+			      unsigned int handle, const char *name);
+
+void intel_region_reference(struct intel_region **dst,
+                            struct intel_region *src);
+
+void intel_region_release(struct intel_region **ib);
+
+void intel_recreate_static_regions(struct intel_context *intel);
+
+/* Map/unmap regions.  This is refcounted also: 
+ */
+GLubyte *intel_region_map(struct intel_context *intel,
+                          struct intel_region *ib);
+
+void intel_region_unmap(struct intel_context *intel, struct intel_region *ib);
+
+
+/* Upload data to a rectangular sub-region
+ */
+void intel_region_data(struct intel_context *intel,
+                       struct intel_region *dest,
+                       GLuint dest_offset,
+                       GLuint destx, GLuint desty,
+                       const void *src, GLuint src_stride,
+                       GLuint srcx, GLuint srcy, GLuint width, GLuint height);
+
+/* Copy rectangular sub-regions
+ */
+void intel_region_copy(struct intel_context *intel,
+                       struct intel_region *dest,
+                       GLuint dest_offset,
+                       GLuint destx, GLuint desty,
+                       struct intel_region *src,
+                       GLuint src_offset,
+                       GLuint srcx, GLuint srcy, GLuint width, GLuint height);
+
+/* Fill a rectangular sub-region
+ */
+void intel_region_fill(struct intel_context *intel,
+                       struct intel_region *dest,
+                       GLuint dest_offset,
+                       GLuint destx, GLuint desty,
+                       GLuint width, GLuint height, GLuint color);
+
+/* Helpers for zerocopy uploads, particularly texture image uploads:
+ */
+void intel_region_attach_pbo(struct intel_context *intel,
+                             struct intel_region *region,
+                             struct intel_buffer_object *pbo);
+void intel_region_release_pbo(struct intel_context *intel,
+                              struct intel_region *region);
+void intel_region_cow(struct intel_context *intel,
+                      struct intel_region *region);
+
+dri_bo *intel_region_buffer(struct intel_context *intel,
+			    struct intel_region *region,
+			    GLuint flag);
+
+void _mesa_copy_rect(GLubyte * dst,
+                GLuint cpp,
+                GLuint dst_pitch,
+                GLuint dst_x,
+                GLuint dst_y,
+                GLuint width,
+                GLuint height,
+                const GLubyte * src,
+                GLuint src_pitch, GLuint src_x, GLuint src_y);
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
new file mode 100644
index 00000000000..61b55b97b51
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -0,0 +1,750 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "main/matrix.h"
+#include "main/renderbuffer.h"
+#include "main/simple_list.h"
+#include "utils.h"
+#include "vblank.h"
+#include "xmlpool.h"
+
+
+#include "intel_screen.h"
+
+#include "intel_buffers.h"
+#include "intel_tex.h"
+#include "intel_span.h"
+#include "intel_fbo.h"
+#include "intel_chipset.h"
+
+#include "i915_drm.h"
+#include "i830_dri.h"
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
+#include "intel_bufmgr.h"
+
+PUBLIC const char __driConfigOptions[] =
+   DRI_CONF_BEGIN
+   DRI_CONF_SECTION_PERFORMANCE
+      DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS)
+      DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_ALWAYS_SYNC)
+      /* Options correspond to DRI_CONF_BO_REUSE_DISABLED,
+       * DRI_CONF_BO_REUSE_ALL
+       */
+      DRI_CONF_OPT_BEGIN_V(bo_reuse, enum, 1, "0:1")
+	 DRI_CONF_DESC_BEGIN(en, "Buffer object reuse")
+	    DRI_CONF_ENUM(0, "Disable buffer object reuse")
+	    DRI_CONF_ENUM(1, "Enable reuse of all sizes of buffer objects")
+	 DRI_CONF_DESC_END
+      DRI_CONF_OPT_END
+   DRI_CONF_SECTION_END
+   DRI_CONF_SECTION_QUALITY
+      DRI_CONF_FORCE_S3TC_ENABLE(false)
+      DRI_CONF_ALLOW_LARGE_TEXTURES(2)
+   DRI_CONF_SECTION_END
+   DRI_CONF_SECTION_DEBUG
+     DRI_CONF_NO_RAST(false)
+   DRI_CONF_SECTION_END
+DRI_CONF_END;
+
+const GLuint __driNConfigOptions = 6;
+
+#ifdef USE_NEW_INTERFACE
+static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
+#endif /*USE_NEW_INTERFACE */
+
+/**
+ * Map all the memory regions described by the screen.
+ * \return GL_TRUE if success, GL_FALSE if error.
+ */
+GLboolean
+intelMapScreenRegions(__DRIscreenPrivate * sPriv)
+{
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+
+   if (0)
+      _mesa_printf("TEX 0x%08x ", intelScreen->tex.handle);
+   if (intelScreen->tex.size != 0) {
+      if (drmMap(sPriv->fd,
+		 intelScreen->tex.handle,
+		 intelScreen->tex.size,
+		 (drmAddress *) & intelScreen->tex.map) != 0) {
+	 intelUnmapScreenRegions(intelScreen);
+	 return GL_FALSE;
+      }
+   }
+
+   return GL_TRUE;
+}
+
+void
+intelUnmapScreenRegions(intelScreenPrivate * intelScreen)
+{
+   if (intelScreen->tex.map) {
+      drmUnmap(intelScreen->tex.map, intelScreen->tex.size);
+      intelScreen->tex.map = NULL;
+   }
+}
+
+
+static void
+intelPrintDRIInfo(intelScreenPrivate * intelScreen,
+                  __DRIscreenPrivate * sPriv, I830DRIPtr gDRIPriv)
+{
+   fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
+           intelScreen->front.size, intelScreen->front.offset,
+           intelScreen->pitch);
+   fprintf(stderr, "*** Back size:    0x%x  offset: 0x%x  pitch: %d\n",
+           intelScreen->back.size, intelScreen->back.offset,
+           intelScreen->pitch);
+   fprintf(stderr, "*** Depth size:   0x%x  offset: 0x%x  pitch: %d\n",
+           intelScreen->depth.size, intelScreen->depth.offset,
+           intelScreen->pitch);
+   fprintf(stderr, "*** Texture size: 0x%x  offset: 0x%x\n",
+           intelScreen->tex.size, intelScreen->tex.offset);
+   fprintf(stderr, "*** Memory : 0x%x\n", gDRIPriv->mem);
+}
+
+
+static void
+intelPrintSAREA(const struct drm_i915_sarea * sarea)
+{
+   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width,
+           sarea->height);
+   fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
+   fprintf(stderr,
+           "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
+           sarea->front_offset, sarea->front_size,
+           (unsigned) sarea->front_handle, sarea->front_tiled);
+   fprintf(stderr,
+           "SAREA: back  offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
+           sarea->back_offset, sarea->back_size,
+           (unsigned) sarea->back_handle, sarea->back_tiled);
+   fprintf(stderr, "SAREA: depth offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
+           sarea->depth_offset, sarea->depth_size,
+           (unsigned) sarea->depth_handle, sarea->depth_tiled);
+   fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
+           sarea->tex_offset, sarea->tex_size, (unsigned) sarea->tex_handle);
+}
+
+
+/**
+ * A number of the screen parameters are obtained/computed from
+ * information in the SAREA.  This function updates those parameters.
+ */
+void
+intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
+                           struct drm_i915_sarea * sarea)
+{
+   intelScreen->width = sarea->width;
+   intelScreen->height = sarea->height;
+   intelScreen->pitch = sarea->pitch;
+
+   intelScreen->front.offset = sarea->front_offset;
+   intelScreen->front.handle = sarea->front_handle;
+   intelScreen->front.size = sarea->front_size;
+   intelScreen->front.tiled = sarea->front_tiled;
+
+   intelScreen->back.offset = sarea->back_offset;
+   intelScreen->back.handle = sarea->back_handle;
+   intelScreen->back.size = sarea->back_size;
+   intelScreen->back.tiled = sarea->back_tiled;
+
+   if (intelScreen->driScrnPriv->ddx_version.minor >= 8) {
+      intelScreen->third.offset = sarea->third_offset;
+      intelScreen->third.handle = sarea->third_handle;
+      intelScreen->third.size = sarea->third_size;
+      intelScreen->third.tiled = sarea->third_tiled;
+   }
+
+   intelScreen->depth.offset = sarea->depth_offset;
+   intelScreen->depth.handle = sarea->depth_handle;
+   intelScreen->depth.size = sarea->depth_size;
+   intelScreen->depth.tiled = sarea->depth_tiled;
+
+   if (intelScreen->driScrnPriv->ddx_version.minor >= 9) {
+      intelScreen->front.bo_handle = sarea->front_bo_handle;
+      intelScreen->back.bo_handle = sarea->back_bo_handle;
+      intelScreen->third.bo_handle = sarea->third_bo_handle;
+      intelScreen->depth.bo_handle = sarea->depth_bo_handle;
+   } else {
+      intelScreen->front.bo_handle = -1;
+      intelScreen->back.bo_handle = -1;
+      intelScreen->third.bo_handle = -1;
+      intelScreen->depth.bo_handle = -1;
+   }
+
+   intelScreen->tex.offset = sarea->tex_offset;
+   intelScreen->logTextureGranularity = sarea->log_tex_granularity;
+   intelScreen->tex.handle = sarea->tex_handle;
+   intelScreen->tex.size = sarea->tex_size;
+
+   if (0)
+      intelPrintSAREA(sarea);
+}
+
+static const __DRItexOffsetExtension intelTexOffsetExtension = {
+   { __DRI_TEX_OFFSET },
+   intelSetTexOffset,
+};
+
+static const __DRItexBufferExtension intelTexBufferExtension = {
+    { __DRI_TEX_BUFFER, __DRI_TEX_BUFFER_VERSION },
+   intelSetTexBuffer,
+};
+
+static const __DRIextension *intelScreenExtensions[] = {
+    &driReadDrawableExtension,
+    &driCopySubBufferExtension.base,
+    &driSwapControlExtension.base,
+    &driFrameTrackingExtension.base,
+    &driMediaStreamCounterExtension.base,
+    &intelTexOffsetExtension.base,
+    &intelTexBufferExtension.base,
+    NULL
+};
+
+static GLboolean
+intel_get_param(__DRIscreenPrivate *psp, int param, int *value)
+{
+   int ret;
+   struct drm_i915_getparam gp;
+
+   gp.param = param;
+   gp.value = value;
+
+   ret = drmCommandWriteRead(psp->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
+   if (ret) {
+      fprintf(stderr, "drm_i915_getparam: %d\n", ret);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
+{
+   intelScreenPrivate *intelScreen;
+   I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
+   struct drm_i915_sarea *sarea;
+
+   if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
+      fprintf(stderr,
+              "\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
+      return GL_FALSE;
+   }
+
+   /* Allocate the private area */
+   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
+   if (!intelScreen) {
+      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
+      return GL_FALSE;
+   }
+   /* parse information in __driConfigOptions */
+   driParseOptionInfo(&intelScreen->optionCache,
+                      __driConfigOptions, __driNConfigOptions);
+
+   intelScreen->driScrnPriv = sPriv;
+   sPriv->private = (void *) intelScreen;
+   sarea = (struct drm_i915_sarea *)
+      (((GLubyte *) sPriv->pSAREA) + gDRIPriv->sarea_priv_offset);
+   intelScreen->sarea = sarea;
+
+   intelScreen->deviceID = gDRIPriv->deviceID;
+
+   intelUpdateScreenFromSAREA(intelScreen, sarea);
+
+   if (!intelMapScreenRegions(sPriv)) {
+      fprintf(stderr, "\nERROR!  mapping regions\n");
+      _mesa_free(intelScreen);
+      sPriv->private = NULL;
+      return GL_FALSE;
+   }
+
+   if (0)
+      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
+
+   intelScreen->drmMinor = sPriv->drm_version.minor;
+
+   /* Determine if IRQs are active? */
+   if (!intel_get_param(sPriv, I915_PARAM_IRQ_ACTIVE,
+			&intelScreen->irq_active))
+      return GL_FALSE;
+
+   sPriv->extensions = intelScreenExtensions;
+
+   return GL_TRUE;
+}
+
+
+static void
+intelDestroyScreen(__DRIscreenPrivate * sPriv)
+{
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+
+   dri_bufmgr_destroy(intelScreen->bufmgr);
+   intelUnmapScreenRegions(intelScreen);
+
+   FREE(intelScreen);
+   sPriv->private = NULL;
+}
+
+
+/**
+ * This is called when we need to set up GL rendering to a new X window.
+ */
+static GLboolean
+intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
+                  __DRIdrawablePrivate * driDrawPriv,
+                  const __GLcontextModes * mesaVis, GLboolean isPixmap)
+{
+   intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
+
+   if (isPixmap) {
+      return GL_FALSE;          /* not implemented */
+   }
+   else {
+      GLboolean swStencil = (mesaVis->stencilBits > 0 &&
+                             mesaVis->depthBits != 24);
+      GLenum rgbFormat = (mesaVis->redBits == 5 ? GL_RGB5 : GL_RGBA8);
+
+      struct intel_framebuffer *intel_fb = CALLOC_STRUCT(intel_framebuffer);
+
+      if (!intel_fb)
+	 return GL_FALSE;
+
+      _mesa_initialize_framebuffer(&intel_fb->Base, mesaVis);
+
+      /* setup the hardware-based renderbuffers */
+      intel_fb->color_rb[0] = intel_create_renderbuffer(rgbFormat);
+      _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT,
+			     &intel_fb->color_rb[0]->Base);
+
+      if (mesaVis->doubleBufferMode) {
+	 intel_fb->color_rb[1] = intel_create_renderbuffer(rgbFormat);
+
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT,
+				&intel_fb->color_rb[1]->Base);
+
+	 if (screen->third.handle) {
+	    struct gl_renderbuffer *tmp_rb = NULL;
+
+	    intel_fb->color_rb[2] = intel_create_renderbuffer(rgbFormat);
+	    _mesa_reference_renderbuffer(&tmp_rb, &intel_fb->color_rb[2]->Base);
+	 }
+      }
+
+      if (mesaVis->depthBits == 24) {
+	 if (mesaVis->stencilBits == 8) {
+	    /* combined depth/stencil buffer */
+	    struct intel_renderbuffer *depthStencilRb
+	       = intel_create_renderbuffer(GL_DEPTH24_STENCIL8_EXT);
+	    /* note: bind RB to two attachment points */
+	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
+				   &depthStencilRb->Base);
+	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_STENCIL,
+				   &depthStencilRb->Base);
+	 } else {
+	    struct intel_renderbuffer *depthRb
+	       = intel_create_renderbuffer(GL_DEPTH_COMPONENT24);
+	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
+				   &depthRb->Base);
+	 }
+      }
+      else if (mesaVis->depthBits == 16) {
+         /* just 16-bit depth buffer, no hw stencil */
+         struct intel_renderbuffer *depthRb
+	    = intel_create_renderbuffer(GL_DEPTH_COMPONENT16);
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH, &depthRb->Base);
+      }
+
+      /* now add any/all software-based renderbuffers we may need */
+      _mesa_add_soft_renderbuffers(&intel_fb->Base,
+                                   GL_FALSE, /* never sw color */
+                                   GL_FALSE, /* never sw depth */
+                                   swStencil, mesaVis->accumRedBits > 0,
+                                   GL_FALSE, /* never sw alpha */
+                                   GL_FALSE  /* never sw aux */ );
+      driDrawPriv->driverPrivate = (void *) intel_fb;
+
+      return GL_TRUE;
+   }
+}
+
+static void
+intelDestroyBuffer(__DRIdrawablePrivate * driDrawPriv)
+{
+   _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
+}
+
+
+/**
+ * Get information about previous buffer swaps.
+ */
+static int
+intelGetSwapInfo(__DRIdrawablePrivate * dPriv, __DRIswapInfo * sInfo)
+{
+   struct intel_framebuffer *intel_fb;
+
+   if ((dPriv == NULL) || (dPriv->driverPrivate == NULL)
+       || (sInfo == NULL)) {
+      return -1;
+   }
+
+   intel_fb = dPriv->driverPrivate;
+   sInfo->swap_count = intel_fb->swap_count;
+   sInfo->swap_ust = intel_fb->swap_ust;
+   sInfo->swap_missed_count = intel_fb->swap_missed_count;
+
+   sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
+      ? driCalculateSwapUsage(dPriv, 0, intel_fb->swap_missed_ust)
+      : 0.0;
+
+   return 0;
+}
+
+
+/* There are probably better ways to do this, such as an
+ * init-designated function to register chipids and createcontext
+ * functions.
+ */
+extern GLboolean i830CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
+
+extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
+extern GLboolean brwCreateContext(const __GLcontextModes * mesaVis,
+				  __DRIcontextPrivate * driContextPriv,
+				  void *sharedContextPrivate);
+
+static GLboolean
+intelCreateContext(const __GLcontextModes * mesaVis,
+                   __DRIcontextPrivate * driContextPriv,
+                   void *sharedContextPrivate)
+{
+   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+
+#ifdef I915
+   if (IS_9XX(intelScreen->deviceID)) {
+      if (!IS_965(intelScreen->deviceID)) {
+	 return i915CreateContext(mesaVis, driContextPriv,
+				  sharedContextPrivate);
+      }
+   } else {
+      intelScreen->no_vbo = GL_TRUE;
+      return i830CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
+   }
+#else
+   if (IS_965(intelScreen->deviceID))
+      return brwCreateContext(mesaVis, driContextPriv, sharedContextPrivate);
+#endif
+   fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
+   return GL_FALSE;
+}
+
+
+static __DRIconfig **
+intelFillInModes(__DRIscreenPrivate *psp,
+		 unsigned pixel_bits, unsigned depth_bits,
+                 unsigned stencil_bits, GLboolean have_back_buffer)
+{
+   __DRIconfig **configs;
+   __GLcontextModes *m;
+   unsigned depth_buffer_factor;
+   unsigned back_buffer_factor;
+   GLenum fb_format;
+   GLenum fb_type;
+   int i;
+
+   /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
+    * support pageflipping at all.
+    */
+   static const GLenum back_buffer_modes[] = {
+      GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
+   };
+
+   uint8_t depth_bits_array[3];
+   uint8_t stencil_bits_array[3];
+
+   depth_bits_array[0] = 0;
+   depth_bits_array[1] = depth_bits;
+   depth_bits_array[2] = depth_bits;
+
+   /* Just like with the accumulation buffer, always provide some modes
+    * with a stencil buffer.  It will be a sw fallback, but some apps won't
+    * care about that.
+    */
+   stencil_bits_array[0] = 0;
+   stencil_bits_array[1] = 0;
+   if (depth_bits == 24)
+      stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
+
+   stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
+
+   depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
+   back_buffer_factor = (have_back_buffer) ? 3 : 1;
+
+   if (pixel_bits == 16) {
+      fb_format = GL_RGB;
+      fb_type = GL_UNSIGNED_SHORT_5_6_5;
+   }
+   else {
+      fb_format = GL_BGRA;
+      fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+   }
+
+   configs = driCreateConfigs(fb_format, fb_type,
+			      depth_bits_array, stencil_bits_array,
+			      depth_buffer_factor, back_buffer_modes,
+			      back_buffer_factor);
+   if (configs == NULL) {
+    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
+   }
+
+   /* Mark the visual as slow if there are "fake" stencil bits.
+    */
+   for (i = 0; configs[i]; i++) {
+      m = &configs[i]->modes;
+      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+         m->visualRating = GLX_SLOW_CONFIG;
+      }
+   }
+
+   return configs;
+}
+
+static GLboolean
+intel_init_bufmgr(intelScreenPrivate *intelScreen)
+{
+   GLboolean gem_disable = getenv("INTEL_NO_GEM") != NULL;
+   int gem_kernel = 0;
+   GLboolean gem_supported;
+   struct drm_i915_getparam gp;
+   __DRIscreenPrivate *spriv = intelScreen->driScrnPriv;
+
+   intelScreen->no_hw = getenv("INTEL_NO_HW") != NULL;
+
+   gp.param = I915_PARAM_HAS_GEM;
+   gp.value = &gem_kernel;
+
+   (void) drmCommandWriteRead(spriv->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
+
+   /* If we've got a new enough DDX that's initializing GEM and giving us
+    * object handles for the shared buffers, use that.
+    */
+   intelScreen->ttm = GL_FALSE;
+   if (intelScreen->driScrnPriv->dri2.enabled)
+       gem_supported = GL_TRUE;
+   else if (intelScreen->driScrnPriv->ddx_version.minor >= 9 &&
+	    gem_kernel &&
+	    intelScreen->front.bo_handle != -1)
+       gem_supported = GL_TRUE;
+   else
+       gem_supported = GL_FALSE;
+
+   if (!gem_disable && gem_supported) {
+      intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
+      if (intelScreen->bufmgr != NULL)
+	 intelScreen->ttm = GL_TRUE;
+   }
+   /* Otherwise, use the classic buffer manager. */
+   if (intelScreen->bufmgr == NULL) {
+      if (gem_disable) {
+	 fprintf(stderr, "GEM disabled.  Using classic.\n");
+      } else {
+	 fprintf(stderr, "Failed to initialize GEM.  "
+		 "Falling back to classic.\n");
+      }
+
+      if (intelScreen->tex.size == 0) {
+	 fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
+		 __func__, __LINE__);
+	 return GL_FALSE;
+      }
+
+      intelScreen->bufmgr =
+	 intel_bufmgr_fake_init(spriv->fd,
+				intelScreen->tex.offset,
+				intelScreen->tex.map,
+				intelScreen->tex.size,
+				(unsigned int * volatile)
+				&intelScreen->sarea->last_dispatch);
+   }
+
+   /* XXX bufmgr should be per-screen, not per-context */
+   intelScreen->ttm = intelScreen->ttm;
+
+   return GL_TRUE;
+}
+
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * Called when using legacy DRI.
+ * 
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
+ */
+static const __DRIconfig **intelInitScreen(__DRIscreenPrivate *psp)
+{
+   intelScreenPrivate *intelScreen;
+#ifdef I915
+   static const __DRIversion ddx_expected = { 1, 5, 0 };
+#else
+   static const __DRIversion ddx_expected = { 1, 6, 0 };
+#endif
+   static const __DRIversion dri_expected = { 4, 0, 0 };
+   static const __DRIversion drm_expected = { 1, 5, 0 };
+   I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
+
+   if (!driCheckDriDdxDrmVersions2("i915",
+                                   &psp->dri_version, &dri_expected,
+                                   &psp->ddx_version, &ddx_expected,
+                                   &psp->drm_version, &drm_expected)) {
+      return NULL;
+   }
+
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   intelInitExtensions(NULL, GL_TRUE);
+	   
+   if (!intelInitDriver(psp))
+       return NULL;
+
+   psp->extensions = intelScreenExtensions;
+
+   intelScreen = psp->private;
+   if (!intel_init_bufmgr(intelScreen))
+       return GL_FALSE;
+
+   return (const __DRIconfig **)
+       intelFillInModes(psp, dri_priv->cpp * 8,
+			(dri_priv->cpp == 2) ? 16 : 24,
+			(dri_priv->cpp == 2) ? 0  : 8, 1);
+}
+
+struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen)
+{
+  /*
+   * This should probably change to have the screen allocate a dummy
+   * context at screen creation. For now just use the current context.
+   */
+
+  GET_CURRENT_CONTEXT(ctx);
+  if (ctx == NULL) {
+     _mesa_problem(NULL, "No current context in intelScreenContext\n");
+     return NULL;
+  }
+  return intel_context(ctx);
+}
+
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * Called when using DRI2.
+ *
+ * \return the __GLcontextModes supported by this driver
+ */
+static const
+__DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
+{
+   intelScreenPrivate *intelScreen;
+
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   intelInitExtensions(NULL, GL_TRUE);
+
+   /* Allocate the private area */
+   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
+   if (!intelScreen) {
+      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
+      return GL_FALSE;
+   }
+   /* parse information in __driConfigOptions */
+   driParseOptionInfo(&intelScreen->optionCache,
+                      __driConfigOptions, __driNConfigOptions);
+
+   intelScreen->driScrnPriv = psp;
+   psp->private = (void *) intelScreen;
+
+   intelScreen->drmMinor = psp->drm_version.minor;
+
+   /* Determine chipset ID */
+   if (!intel_get_param(psp, I915_PARAM_CHIPSET_ID,
+			&intelScreen->deviceID))
+      return GL_FALSE;
+
+   if (!intel_init_bufmgr(intelScreen))
+       return GL_FALSE;
+
+   intelScreen->irq_active = 1;
+   psp->extensions = intelScreenExtensions;
+
+   return driConcatConfigs(intelFillInModes(psp, 16, 16, 0, 1),
+			   intelFillInModes(psp, 32, 24, 8, 1));
+}
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen		 = intelInitScreen,
+   .DestroyScreen	 = intelDestroyScreen,
+   .CreateContext	 = intelCreateContext,
+   .DestroyContext	 = intelDestroyContext,
+   .CreateBuffer	 = intelCreateBuffer,
+   .DestroyBuffer	 = intelDestroyBuffer,
+   .SwapBuffers		 = intelSwapBuffers,
+   .MakeCurrent		 = intelMakeCurrent,
+   .UnbindContext	 = intelUnbindContext,
+   .GetSwapInfo		 = intelGetSwapInfo,
+   .GetDrawableMSC	 = driDrawableGetMSC32,
+   .WaitForMSC		 = driWaitForMSC32,
+   .CopySubBuffer	 = intelCopySubBuffer,
+
+   .InitScreen2		 = intelInitScreen2,
+};
diff --git a/src/mesa/drivers/dri/i965/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index bf9a716082d..91f0d6d1ae8 100644
--- a/src/mesa/drivers/dri/i965/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -30,85 +30,87 @@
 
 #include <sys/time.h>
 #include "dri_util.h"
+#include "intel_bufmgr.h"
+#include "i915_drm.h"
 #include "xmlconfig.h"
-#include "i830_common.h"
 
 /* XXX: change name or eliminate to avoid conflict with "struct
  * intel_region"!!!
  */
-typedef struct {
+typedef struct
+{
    drm_handle_t handle;
-   drmSize size;        /* region size in bytes */
-   char *map;           /* memory map */
-   int offset;          /* from start of video mem, in bytes */
-   int pitch;           /* row stride, in pixels */
-   unsigned int tiled; 
+   drmSize size;                /* region size in bytes */
+   char *map;                   /* memory map */
+   int offset;                  /* from start of video mem, in bytes */
+   unsigned int bo_handle;	/* buffer object id if available, or -1 */
+   /**
+    * Flags if the region is tiled.
+    *
+    * Not included is Y versus X tiling.
+    */
+   GLboolean tiled;
 } intelRegion;
 
-typedef struct 
+typedef struct
 {
    intelRegion front;
    intelRegion back;
-   intelRegion rotated;
+   intelRegion third;
    intelRegion depth;
    intelRegion tex;
-   
+
    int deviceID;
    int width;
    int height;
-   int mem;         /* unused */
-
-   int cpp;         /* for front and back buffers */
-   int fbFormat;
+   int pitch;                   /* common row stride, in pixels */
 
    int logTextureGranularity;
-   
+
    __DRIscreenPrivate *driScrnPriv;
-   unsigned int sarea_priv_offset;
+
+   volatile struct drm_i915_sarea *sarea;
 
    int drmMinor;
 
    int irq_active;
-   int allow_batchbuffer;
 
-/*    struct matrix23 rotMatrix; */
+   GLboolean no_hw;
 
-   int current_rotation;  /* 0, 90, 180 or 270 */
-   int rotatedWidth, rotatedHeight;
+   GLboolean no_vbo;
+   int ttm;
+   dri_bufmgr *bufmgr;
 
    /**
-    * Configuration cache with default values for all contexts 
-    */
+   * Configuration cache with default values for all contexts
+   */
    driOptionCache optionCache;
 } intelScreenPrivate;
 
 
-extern GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv);
 
-extern void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen);
+extern GLboolean intelMapScreenRegions(__DRIscreenPrivate * sPriv);
 
-extern void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           volatile drmI830Sarea *sarea);
+extern void intelUnmapScreenRegions(intelScreenPrivate * intelScreen);
 
 extern void
-intelDestroyContext(__DRIcontextPrivate *driContextPriv);
+intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
+                           struct drm_i915_sarea * sarea);
 
-extern GLboolean
-intelUnbindContext(__DRIcontextPrivate *driContextPriv);
+extern void intelDestroyContext(__DRIcontextPrivate * driContextPriv);
+
+extern GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv);
 
 extern GLboolean
-intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv);
+intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
+                 __DRIdrawablePrivate * driDrawPriv,
+                 __DRIdrawablePrivate * driReadPriv);
+
+extern void intelSwapBuffers(__DRIdrawablePrivate * dPriv);
 
 extern void
-intelSwapBuffers(__DRIdrawablePrivate *dPriv);
+intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h);
 
-extern void 
-intelCopySubBuffer( __DRIdrawablePrivate *dPriv,
-		    int x, int y, int w, int h );
+extern struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen);
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
new file mode 100644
index 00000000000..8f4e681ffea
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -0,0 +1,791 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+
+#include "intel_buffers.h"
+#include "intel_fbo.h"
+#include "intel_screen.h"
+#include "intel_span.h"
+#include "intel_regions.h"
+#include "intel_tex.h"
+
+#include "swrast/swrast.h"
+
+static void
+intel_set_span_functions(struct intel_context *intel,
+			 struct gl_renderbuffer *rb);
+
+#define SPAN_CACHE_SIZE		4096
+
+static void
+get_span_cache(struct intel_renderbuffer *irb, uint32_t offset)
+{
+   if (irb->span_cache == NULL) {
+      irb->span_cache = _mesa_malloc(SPAN_CACHE_SIZE);
+      irb->span_cache_offset = -1;
+   }
+
+   if ((offset & ~(SPAN_CACHE_SIZE - 1)) != irb->span_cache_offset) {
+      irb->span_cache_offset = offset & ~(SPAN_CACHE_SIZE - 1);
+      dri_bo_get_subdata(irb->region->buffer, irb->span_cache_offset,
+			 SPAN_CACHE_SIZE, irb->span_cache);
+   }
+}
+
+static void
+clear_span_cache(struct intel_renderbuffer *irb)
+{
+   irb->span_cache_offset = -1;
+}
+
+static uint32_t
+pread_32(struct intel_renderbuffer *irb, uint32_t offset)
+{
+   get_span_cache(irb, offset);
+
+   return *(uint32_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1)));
+}
+
+static uint32_t
+pread_xrgb8888(struct intel_renderbuffer *irb, uint32_t offset)
+{
+   get_span_cache(irb, offset);
+
+   return *(uint32_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1))) |
+      0xff000000;
+}
+
+static uint16_t
+pread_16(struct intel_renderbuffer *irb, uint32_t offset)
+{
+   get_span_cache(irb, offset);
+
+   return *(uint16_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1)));
+}
+
+static uint8_t
+pread_8(struct intel_renderbuffer *irb, uint32_t offset)
+{
+   get_span_cache(irb, offset);
+
+   return *(uint8_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1)));
+}
+
+static void
+pwrite_32(struct intel_renderbuffer *irb, uint32_t offset, uint32_t val)
+{
+   clear_span_cache(irb);
+
+   dri_bo_subdata(irb->region->buffer, offset, 4, &val);
+}
+
+static void
+pwrite_xrgb8888(struct intel_renderbuffer *irb, uint32_t offset, uint32_t val)
+{
+   clear_span_cache(irb);
+
+   dri_bo_subdata(irb->region->buffer, offset, 3, &val);
+}
+
+static void
+pwrite_16(struct intel_renderbuffer *irb, uint32_t offset, uint16_t val)
+{
+   clear_span_cache(irb);
+
+   dri_bo_subdata(irb->region->buffer, offset, 2, &val);
+}
+
+static void
+pwrite_8(struct intel_renderbuffer *irb, uint32_t offset, uint8_t val)
+{
+   clear_span_cache(irb);
+
+   dri_bo_subdata(irb->region->buffer, offset, 1, &val);
+}
+
+static uint32_t no_tile_swizzle(struct intel_renderbuffer *irb,
+				int x, int y)
+{
+	return (y * irb->region->pitch + x) * irb->region->cpp;
+}
+
+/*
+ * Deal with tiled surfaces
+ */
+
+static uint32_t x_tile_swizzle(struct intel_renderbuffer *irb,
+			       int x, int y)
+{
+	int	tile_stride;
+	int	xbyte;
+	int	x_tile_off, y_tile_off;
+	int	x_tile_number, y_tile_number;
+	int	tile_off, tile_base;
+	
+	tile_stride = (irb->pfPitch * irb->region->cpp) << 3;
+
+	xbyte = x * irb->region->cpp;
+
+	x_tile_off = xbyte & 0x1ff;
+	y_tile_off = y & 7;
+
+	x_tile_number = xbyte >> 9;
+	y_tile_number = y >> 3;
+
+	tile_off = (y_tile_off << 9) + x_tile_off;
+
+	switch (irb->region->bit_6_swizzle) {
+	case I915_BIT_6_SWIZZLE_NONE:
+	   break;
+	case I915_BIT_6_SWIZZLE_9:
+	   tile_off ^= ((tile_off >> 3) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_10:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_11:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 5) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_10_11:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64) ^
+	      ((tile_off >> 5) & 64);
+	   break;
+	default:
+	   fprintf(stderr, "Unknown tile swizzling mode %d\n",
+		   irb->region->bit_6_swizzle);
+	   exit(1);
+	}
+
+	tile_base = (x_tile_number << 12) + y_tile_number * tile_stride;
+
+#if 0
+	printf("(%d,%d) -> %d + %d = %d (pitch = %d, tstride = %d)\n",
+	       x, y, tile_off, tile_base,
+	       tile_off + tile_base,
+	       irb->pfPitch, tile_stride);
+#endif
+
+	return tile_base + tile_off;
+}
+
+static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
+			       int x, int y)
+{
+	int	tile_stride;
+	int	xbyte;
+	int	x_tile_off, y_tile_off;
+	int	x_tile_number, y_tile_number;
+	int	tile_off, tile_base;
+	
+	tile_stride = (irb->pfPitch * irb->region->cpp) << 5;
+
+	xbyte = x * irb->region->cpp;
+
+	x_tile_off = xbyte & 0x7f;
+	y_tile_off = y & 0x1f;
+
+	x_tile_number = xbyte >> 7;
+	y_tile_number = y >> 5;
+
+	tile_off = ((x_tile_off & ~0xf) << 5) + (y_tile_off << 4) +
+	   (x_tile_off & 0xf);
+
+	switch (irb->region->bit_6_swizzle) {
+	case I915_BIT_6_SWIZZLE_NONE:
+	   break;
+	case I915_BIT_6_SWIZZLE_9:
+	   tile_off ^= ((tile_off >> 3) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_10:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_11:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 5) & 64);
+	   break;
+	case I915_BIT_6_SWIZZLE_9_10_11:
+	   tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64) ^
+	      ((tile_off >> 5) & 64);
+	   break;
+	default:
+	   fprintf(stderr, "Unknown tile swizzling mode %d\n",
+		   irb->region->bit_6_swizzle);
+	   exit(1);
+	}
+
+	tile_base = (x_tile_number << 12) + y_tile_number * tile_stride;
+
+	return tile_base + tile_off;
+}
+
+/*
+  break intelWriteRGBASpan_ARGB8888
+*/
+
+#undef DBG
+#define DBG 0
+
+#define LOCAL_VARS							\
+   struct intel_context *intel = intel_context(ctx);			\
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
+   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
+   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1;	\
+   unsigned int num_cliprects;						\
+   struct drm_clip_rect *cliprects;					\
+   int x_off, y_off;							\
+   GLuint p;								\
+   (void) p;								\
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+
+/* XXX FBO: this is identical to the macro in spantmp2.h except we get
+ * the cliprect info from the context, not the driDrawable.
+ * Move this into spantmp2.h someday.
+ */
+#define HW_CLIPLOOP()							\
+   do {									\
+      int _nc = num_cliprects;						\
+      while ( _nc-- ) {							\
+	 int minx = cliprects[_nc].x1 - x_off;				\
+	 int miny = cliprects[_nc].y1 - y_off;				\
+	 int maxx = cliprects[_nc].x2 - x_off;				\
+	 int maxy = cliprects[_nc].y2 - y_off;
+	
+#if 0
+      }}
+#endif
+
+#define Y_FLIP(_y) ((_y) * yScale + yBias)
+
+/* XXX with GEM, these need to tell the kernel */
+#define HW_LOCK()
+
+#define HW_UNLOCK()
+
+/* Convenience macros to avoid typing the swizzle argument over and over */
+#define NO_TILE(_X, _Y) no_tile_swizzle(irb, (_X) + x_off, (_Y) + y_off)
+#define X_TILE(_X, _Y) x_tile_swizzle(irb, (_X) + x_off, (_Y) + y_off)
+#define Y_TILE(_X, _Y) y_tile_swizzle(irb, (_X) + x_off, (_Y) + y_off)
+
+/* 16 bit, RGB565 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
+
+#define TAG(x)    intel##x##_RGB565
+#define TAG2(x,y) intel##x##_RGB565##y
+#define GET_VALUE(X, Y) pread_16(irb, NO_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_16(irb, NO_TILE(X, Y), V)
+#include "spantmp2.h"
+
+/* 32 bit, ARGB8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel##x##_ARGB8888
+#define TAG2(x,y) intel##x##_ARGB8888##y
+#define GET_VALUE(X, Y) pread_32(irb, NO_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_32(irb, NO_TILE(X, Y), V)
+#include "spantmp2.h"
+
+/* 32 bit, xRGB8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel##x##_xRGB8888
+#define TAG2(x,y) intel##x##_xRGB8888##y
+#define GET_VALUE(X, Y) pread_xrgb8888(irb, NO_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, NO_TILE(X, Y), V)
+#include "spantmp2.h"
+
+/* 16 bit RGB565 color tile spanline and pixel functions
+ */
+
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
+
+#define TAG(x)    intel_XTile_##x##_RGB565
+#define TAG2(x,y) intel_XTile_##x##_RGB565##y
+#define GET_VALUE(X, Y) pread_16(irb, X_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_16(irb, X_TILE(X, Y), V)
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
+
+#define TAG(x)    intel_YTile_##x##_RGB565
+#define TAG2(x,y) intel_YTile_##x##_RGB565##y
+#define GET_VALUE(X, Y) pread_16(irb, Y_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_16(irb, Y_TILE(X, Y), V)
+#include "spantmp2.h"
+
+/* 32 bit ARGB888 color tile spanline and pixel functions
+ */
+
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel_XTile_##x##_ARGB8888
+#define TAG2(x,y) intel_XTile_##x##_ARGB8888##y
+#define GET_VALUE(X, Y) pread_32(irb, X_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_32(irb, X_TILE(X, Y), V)
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel_YTile_##x##_ARGB8888
+#define TAG2(x,y) intel_YTile_##x##_ARGB8888##y
+#define GET_VALUE(X, Y) pread_32(irb, Y_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_32(irb, Y_TILE(X, Y), V)
+#include "spantmp2.h"
+
+/* 32 bit xRGB888 color tile spanline and pixel functions
+ */
+
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel_XTile_##x##_xRGB8888
+#define TAG2(x,y) intel_XTile_##x##_xRGB8888##y
+#define GET_VALUE(X, Y) pread_xrgb8888(irb, X_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, X_TILE(X, Y), V)
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
+
+#define TAG(x)    intel_YTile_##x##_xRGB8888
+#define TAG2(x,y) intel_YTile_##x##_xRGB8888##y
+#define GET_VALUE(X, Y) pread_xrgb8888(irb, Y_TILE(X, Y))
+#define PUT_VALUE(X, Y, V) pwrite_xrgb8888(irb, Y_TILE(X, Y), V)
+#include "spantmp2.h"
+
+#define LOCAL_DEPTH_VARS						\
+   struct intel_context *intel = intel_context(ctx);			\
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
+   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
+   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1; \
+   unsigned int num_cliprects;						\
+   struct drm_clip_rect *cliprects;					\
+   int x_off, y_off;							\
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+
+
+#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS
+
+/**
+ ** 16-bit depthbuffer functions.
+ **/
+#define VALUE_TYPE GLushort
+#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, NO_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = pread_16(irb, NO_TILE(_x, _y))
+#define TAG(x) intel##x##_z16
+#include "depthtmp.h"
+
+
+/**
+ ** 16-bit x tile depthbuffer functions.
+ **/
+#define VALUE_TYPE GLushort
+#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, X_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = pread_16(irb, X_TILE(_x, _y))
+#define TAG(x) intel_XTile_##x##_z16
+#include "depthtmp.h"
+
+/**
+ ** 16-bit y tile depthbuffer functions.
+ **/
+#define VALUE_TYPE GLushort
+#define WRITE_DEPTH(_x, _y, d) pwrite_16(irb, Y_TILE(_x, _y), d)
+#define READ_DEPTH(d, _x, _y) d = pread_16(irb, Y_TILE(_x, _y))
+#define TAG(x) intel_YTile_##x##_z16
+#include "depthtmp.h"
+
+
+/**
+ ** 24/8-bit interleaved depth/stencil functions
+ ** Note: we're actually reading back combined depth+stencil values.
+ ** The wrappers in main/depthstencil.c are used to extract the depth
+ ** and stencil values.
+ **/
+#define VALUE_TYPE GLuint
+
+/* Change ZZZS -> SZZZ */
+#define WRITE_DEPTH(_x, _y, d)					\
+   pwrite_32(irb, NO_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
+
+/* Change SZZZ -> ZZZS */
+#define READ_DEPTH( d, _x, _y ) {				\
+   GLuint tmp = pread_32(irb, NO_TILE(_x, _y));			\
+   d = (tmp << 8) | (tmp >> 24);				\
+}
+
+#define TAG(x) intel##x##_z24_s8
+#include "depthtmp.h"
+
+
+/**
+ ** 24/8-bit x-tile interleaved depth/stencil functions
+ ** Note: we're actually reading back combined depth+stencil values.
+ ** The wrappers in main/depthstencil.c are used to extract the depth
+ ** and stencil values.
+ **/
+#define VALUE_TYPE GLuint
+
+/* Change ZZZS -> SZZZ */
+#define WRITE_DEPTH(_x, _y, d)					\
+   pwrite_32(irb, X_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
+
+/* Change SZZZ -> ZZZS */
+#define READ_DEPTH( d, _x, _y ) {				\
+   GLuint tmp = pread_32(irb, X_TILE(_x, _y));		\
+   d = (tmp << 8) | (tmp >> 24);				\
+}
+
+#define TAG(x) intel_XTile_##x##_z24_s8
+#include "depthtmp.h"
+
+/**
+ ** 24/8-bit y-tile interleaved depth/stencil functions
+ ** Note: we're actually reading back combined depth+stencil values.
+ ** The wrappers in main/depthstencil.c are used to extract the depth
+ ** and stencil values.
+ **/
+#define VALUE_TYPE GLuint
+
+/* Change ZZZS -> SZZZ */
+#define WRITE_DEPTH(_x, _y, d)					\
+   pwrite_32(irb, Y_TILE(_x, _y), ((d) >> 8) | ((d) << 24))
+
+/* Change SZZZ -> ZZZS */
+#define READ_DEPTH( d, _x, _y ) {				\
+   GLuint tmp = pread_32(irb, Y_TILE(_x, _y));			\
+   d = (tmp << 8) | (tmp >> 24);				\
+}
+
+#define TAG(x) intel_YTile_##x##_z24_s8
+#include "depthtmp.h"
+
+
+/**
+ ** 8-bit stencil function (XXX FBO: This is obsolete)
+ **/
+#define WRITE_STENCIL(_x, _y, d) pwrite_8(irb, NO_TILE(_x, _y) + 3, d)
+#define READ_STENCIL(d, _x, _y) d = pread_8(irb, NO_TILE(_x, _y) + 3);
+#define TAG(x) intel##x##_z24_s8
+#include "stenciltmp.h"
+
+/**
+ ** 8-bit x-tile stencil function (XXX FBO: This is obsolete)
+ **/
+#define WRITE_STENCIL(_x, _y, d) pwrite_8(irb, X_TILE(_x, _y) + 3, d)
+#define READ_STENCIL(d, _x, _y) d = pread_8(irb, X_TILE(_x, _y) + 3);
+#define TAG(x) intel_XTile_##x##_z24_s8
+#include "stenciltmp.h"
+
+/**
+ ** 8-bit y-tile stencil function (XXX FBO: This is obsolete)
+ **/
+#define WRITE_STENCIL(_x, _y, d) pwrite_8(irb, Y_TILE(_x, _y) + 3, d)
+#define READ_STENCIL(d, _x, _y) d = pread_8(irb, Y_TILE(_x, _y) + 3)
+#define TAG(x) intel_YTile_##x##_z24_s8
+#include "stenciltmp.h"
+
+void
+intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb)
+{
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);
+
+   if (irb == NULL || irb->region == NULL)
+      return;
+
+   irb->pfPitch = irb->region->pitch;
+
+   intel_set_span_functions(intel, rb);
+}
+
+void
+intel_renderbuffer_unmap(struct intel_context *intel,
+			 struct gl_renderbuffer *rb)
+{
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);
+
+   if (irb == NULL || irb->region == NULL)
+      return;
+
+   clear_span_cache(irb);
+   irb->pfPitch = 0;
+
+   rb->GetRow = NULL;
+   rb->PutRow = NULL;
+}
+
+/**
+ * Map or unmap all the renderbuffers which we may need during
+ * software rendering.
+ * XXX in the future, we could probably convey extra information to
+ * reduce the number of mappings needed.  I.e. if doing a glReadPixels
+ * from the depth buffer, we really only need one mapping.
+ *
+ * XXX Rewrite this function someday.
+ * We can probably just loop over all the renderbuffer attachments,
+ * map/unmap all of them, and not worry about the _ColorDrawBuffers
+ * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields.
+ */
+static void
+intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
+{
+   GLcontext *ctx = &intel->ctx;
+   GLuint i, j;
+
+   /* color draw buffers */
+   for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers; j++) {
+      if (map)
+	 intel_renderbuffer_map(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]);
+      else
+	 intel_renderbuffer_unmap(intel, ctx->DrawBuffer->_ColorDrawBuffers[j]);
+   }
+
+   /* check for render to textures */
+   for (i = 0; i < BUFFER_COUNT; i++) {
+      struct gl_renderbuffer_attachment *att =
+         ctx->DrawBuffer->Attachment + i;
+      struct gl_texture_object *tex = att->Texture;
+      if (tex) {
+         /* render to texture */
+         ASSERT(att->Renderbuffer);
+         if (map)
+            intel_tex_map_images(intel, intel_texture_object(tex));
+         else
+            intel_tex_unmap_images(intel, intel_texture_object(tex));
+      }
+   }
+
+   /* color read buffers */
+   if (map)
+      intel_renderbuffer_map(intel, ctx->ReadBuffer->_ColorReadBuffer);
+   else
+      intel_renderbuffer_unmap(intel, ctx->ReadBuffer->_ColorReadBuffer);
+
+   /* depth buffer (Note wrapper!) */
+   if (ctx->DrawBuffer->_DepthBuffer) {
+      if (map)
+	 intel_renderbuffer_map(intel, ctx->DrawBuffer->_DepthBuffer->Wrapped);
+      else
+	 intel_renderbuffer_unmap(intel,
+				  ctx->DrawBuffer->_DepthBuffer->Wrapped);
+   }
+
+   /* stencil buffer (Note wrapper!) */
+   if (ctx->DrawBuffer->_StencilBuffer) {
+      if (map)
+	 intel_renderbuffer_map(intel,
+				ctx->DrawBuffer->_StencilBuffer->Wrapped);
+      else
+	 intel_renderbuffer_unmap(intel,
+				  ctx->DrawBuffer->_StencilBuffer->Wrapped);
+   }
+}
+
+
+
+/**
+ * Prepare for softare rendering.  Map current read/draw framebuffers'
+ * renderbuffes and all currently bound texture objects.
+ *
+ * Old note: Moved locking out to get reasonable span performance.
+ */
+void
+intelSpanRenderStart(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   GLuint i;
+
+   intelFlush(&intel->ctx);
+   LOCK_HARDWARE(intel);
+
+   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
+      if (ctx->Texture.Unit[i]._ReallyEnabled) {
+         struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
+         intel_tex_map_images(intel, intel_texture_object(texObj));
+      }
+   }
+
+   intel_map_unmap_buffers(intel, GL_TRUE);
+}
+
+/**
+ * Called when done softare rendering.  Unmap the buffers we mapped in
+ * the above function.
+ */
+void
+intelSpanRenderFinish(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   GLuint i;
+
+   _swrast_flush(ctx);
+
+   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
+      if (ctx->Texture.Unit[i]._ReallyEnabled) {
+         struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
+         intel_tex_unmap_images(intel, intel_texture_object(texObj));
+      }
+   }
+
+   intel_map_unmap_buffers(intel, GL_FALSE);
+
+   UNLOCK_HARDWARE(intel);
+}
+
+
+void
+intelInitSpanFuncs(GLcontext * ctx)
+{
+   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
+   swdd->SpanRenderStart = intelSpanRenderStart;
+   swdd->SpanRenderFinish = intelSpanRenderFinish;
+}
+
+
+/**
+ * Plug in appropriate span read/write functions for the given renderbuffer.
+ * These are used for the software fallbacks.
+ */
+static void
+intel_set_span_functions(struct intel_context *intel,
+			 struct gl_renderbuffer *rb)
+{
+   struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
+   uint32_t tiling;
+
+   /* If in GEM mode, we need to do the tile address swizzling ourselves,
+    * instead of the fence registers handling it.
+    */
+   if (intel->ttm)
+      tiling = irb->region->tiling;
+   else
+      tiling = I915_TILING_NONE;
+
+   if (rb->_ActualFormat == GL_RGB5) {
+      /* 565 RGB */
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitPointers_RGB565(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitPointers_RGB565(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitPointers_RGB565(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_RGB8) {
+      /* 8888 RGBx */
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitPointers_xRGB8888(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitPointers_xRGB8888(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitPointers_xRGB8888(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_RGBA8) {
+      /* 8888 RGBA */
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitPointers_ARGB8888(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitPointers_ARGB8888(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitPointers_ARGB8888(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT16) {
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitDepthPointers_z16(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitDepthPointers_z16(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitDepthPointers_z16(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24 ||        /* XXX FBO remove */
+            rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitDepthPointers_z24_s8(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitDepthPointers_z24_s8(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitDepthPointers_z24_s8(rb);
+	 break;
+      }
+   }
+   else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) {
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitStencilPointers_z24_s8(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitStencilPointers_z24_s8(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitStencilPointers_z24_s8(rb);
+	 break;
+      }
+   }
+   else {
+      _mesa_problem(NULL,
+                    "Unexpected _ActualFormat in intelSetSpanFunctions");
+   }
+}
diff --git a/src/mesa/drivers/dri/i915/intel_span.h b/src/mesa/drivers/dri/intel/intel_span.h
index 2d4f8589d0f..acbeb4abe1c 100644
--- a/src/mesa/drivers/dri/i915/intel_span.h
+++ b/src/mesa/drivers/dri/intel/intel_span.h
@@ -28,14 +28,13 @@
 #ifndef _INTEL_SPAN_H
 #define _INTEL_SPAN_H
 
-#include "drirenderbuffer.h"
+extern void intelInitSpanFuncs(GLcontext * ctx);
 
-extern void intelInitSpanFuncs( GLcontext *ctx );
-
-extern void intelSpanRenderFinish( GLcontext *ctx );
-extern void intelSpanRenderStart( GLcontext *ctx );
-
-extern void
-intelSetSpanFunctions(driRenderbuffer *rb, const GLvisual *vis);
+extern void intelSpanRenderFinish(GLcontext * ctx);
+extern void intelSpanRenderStart(GLcontext * ctx);
+void intel_renderbuffer_map(struct intel_context *intel,
+			    struct gl_renderbuffer *rb);
+void intel_renderbuffer_unmap(struct intel_context *intel,
+			      struct gl_renderbuffer *rb);
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_tex.c b/src/mesa/drivers/dri/intel/intel_tex.c
new file mode 100644
index 00000000000..82f8b870095
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex.c
@@ -0,0 +1,249 @@
+#include "swrast/swrast.h"
+#include "main/texobj.h"
+#include "main/teximage.h"
+#include "main/mipmap.h"
+#include "intel_context.h"
+#include "intel_mipmap_tree.h"
+#include "intel_tex.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+static GLboolean
+intelIsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj)
+{
+#if 0
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
+
+   return
+      intelObj->mt &&
+      intelObj->mt->region &&
+      intel_is_region_resident(intel, intelObj->mt->region);
+#endif
+   return 1;
+}
+
+
+
+static struct gl_texture_image *
+intelNewTextureImage(GLcontext * ctx)
+{
+   DBG("%s\n", __FUNCTION__);
+   (void) ctx;
+   return (struct gl_texture_image *) CALLOC_STRUCT(intel_texture_image);
+}
+
+
+static struct gl_texture_object *
+intelNewTextureObject(GLcontext * ctx, GLuint name, GLenum target)
+{
+   struct intel_texture_object *obj = CALLOC_STRUCT(intel_texture_object);
+
+   DBG("%s\n", __FUNCTION__);
+   _mesa_initialize_texture_object(&obj->base, name, target);
+
+   return &obj->base;
+}
+
+static void 
+intelDeleteTextureObject(GLcontext *ctx,
+			 struct gl_texture_object *texObj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
+
+   if (intelObj->mt)
+      intel_miptree_release(intel, &intelObj->mt);
+
+   _mesa_delete_texture_object(ctx, texObj);
+}
+
+
+static void
+intelFreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_image *intelImage = intel_texture_image(texImage);
+
+   DBG("%s\n", __FUNCTION__);
+
+   if (intelImage->mt) {
+      intel_miptree_release(intel, &intelImage->mt);
+   }
+
+   if (texImage->Data) {
+      _mesa_free_texmemory(texImage->Data);
+      texImage->Data = NULL;
+   }
+}
+
+
+/* The system memcpy (at least on ubuntu 5.10) has problems copying
+ * to agp (writecombined) memory from a source which isn't 64-byte
+ * aligned - there is a 4x performance falloff.
+ *
+ * The x86 __memcpy is immune to this but is slightly slower
+ * (10%-ish) than the system memcpy.
+ *
+ * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
+ * isn't much faster than x86_memcpy for agp copies.
+ * 
+ * TODO: switch dynamically.
+ */
+static void *
+do_memcpy(void *dest, const void *src, size_t n)
+{
+   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63)) {
+      return __memcpy(dest, src, n);
+   }
+   else
+      return memcpy(dest, src, n);
+}
+
+
+#if DO_DEBUG && !defined(__ia64__)
+
+#ifndef __x86_64__
+static unsigned
+fastrdtsc(void)
+{
+   unsigned eax;
+   __asm__ volatile ("\t"
+                     "pushl  %%ebx\n\t"
+                     "cpuid\n\t" ".byte 0x0f, 0x31\n\t"
+                     "popl %%ebx\n":"=a" (eax)
+                     :"0"(0)
+                     :"ecx", "edx", "cc");
+
+   return eax;
+}
+#else
+static unsigned
+fastrdtsc(void)
+{
+   unsigned eax;
+   __asm__ volatile ("\t" "cpuid\n\t" ".byte 0x0f, 0x31\n\t":"=a" (eax)
+                     :"0"(0)
+                     :"ecx", "edx", "ebx", "cc");
+
+   return eax;
+}
+#endif
+
+static unsigned
+time_diff(unsigned t, unsigned t2)
+{
+   return ((t < t2) ? t2 - t : 0xFFFFFFFFU - (t - t2 - 1));
+}
+
+
+static void *
+timed_memcpy(void *dest, const void *src, size_t n)
+{
+   void *ret;
+   unsigned t1, t2;
+   double rate;
+
+   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63))
+      _mesa_printf("Warning - non-aligned texture copy!\n");
+
+   t1 = fastrdtsc();
+   ret = do_memcpy(dest, src, n);
+   t2 = fastrdtsc();
+
+   rate = time_diff(t1, t2);
+   rate /= (double) n;
+   _mesa_printf("timed_memcpy: %u %u --> %f clocks/byte\n", t1, t2, rate);
+   return ret;
+}
+#endif /* DO_DEBUG */
+
+/**
+ * Generate new mipmap data from BASE+1 to BASE+p (the minimally-sized mipmap
+ * level).
+ *
+ * The texture object's miptree must be mapped.
+ *
+ * It would be really nice if this was just called by Mesa whenever mipmaps
+ * needed to be regenerated, rather than us having to remember to do so in
+ * each texture image modification path.
+ *
+ * This function should also include an accelerated path.
+ */
+void
+intel_generate_mipmap(GLcontext *ctx, GLenum target,
+                      struct gl_texture_object *texObj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
+   GLuint nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
+   int face, i;
+
+   _mesa_generate_mipmap(ctx, target, texObj);
+
+   /* Update the level information in our private data in the new images, since
+    * it didn't get set as part of a normal TexImage path.
+    */
+   for (face = 0; face < nr_faces; face++) {
+      for (i = texObj->BaseLevel + 1; i < texObj->MaxLevel; i++) {
+         struct intel_texture_image *intelImage;
+
+	 intelImage = intel_texture_image(texObj->Image[face][i]);
+	 if (intelImage == NULL)
+	    break;
+
+	 intelImage->level = i;
+	 intelImage->face = face;
+	 /* Unreference the miptree to signal that the new Data is a bare
+	  * pointer from mesa.
+	  */
+	 intel_miptree_release(intel, &intelImage->mt);
+      }
+   }
+}
+
+static void intelGenerateMipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
+
+   intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel);
+   intel_generate_mipmap(ctx, target, texObj);
+   intel_tex_unmap_level_images(intel, intelObj, texObj->BaseLevel);
+}
+
+void
+intelInitTextureFuncs(struct dd_function_table *functions)
+{
+   functions->ChooseTextureFormat = intelChooseTextureFormat;
+   functions->TexImage1D = intelTexImage1D;
+   functions->TexImage2D = intelTexImage2D;
+   functions->TexImage3D = intelTexImage3D;
+   functions->TexSubImage1D = intelTexSubImage1D;
+   functions->TexSubImage2D = intelTexSubImage2D;
+   functions->TexSubImage3D = intelTexSubImage3D;
+   functions->CopyTexImage1D = intelCopyTexImage1D;
+   functions->CopyTexImage2D = intelCopyTexImage2D;
+   functions->CopyTexSubImage1D = intelCopyTexSubImage1D;
+   functions->CopyTexSubImage2D = intelCopyTexSubImage2D;
+   functions->GetTexImage = intelGetTexImage;
+   functions->GenerateMipmap = intelGenerateMipmap;
+
+   /* compressed texture functions */
+   functions->CompressedTexImage2D = intelCompressedTexImage2D;
+   functions->GetCompressedTexImage = intelGetCompressedTexImage;
+
+   functions->NewTextureObject = intelNewTextureObject;
+   functions->NewTextureImage = intelNewTextureImage;
+   functions->DeleteTexture = intelDeleteTextureObject;
+   functions->FreeTexImageData = intelFreeTextureImageData;
+   functions->UpdateTexturePalette = 0;
+   functions->IsTextureResident = intelIsTextureResident;
+
+#if DO_DEBUG && !defined(__ia64__)
+   if (INTEL_DEBUG & DEBUG_BUFMGR)
+      functions->TextureMemCpy = timed_memcpy;
+   else
+#endif
+      functions->TextureMemCpy = do_memcpy;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h
new file mode 100644
index 00000000000..6219c1c953b
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex.h
@@ -0,0 +1,164 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTELTEX_INC
+#define INTELTEX_INC
+
+#include "main/mtypes.h"
+#include "intel_context.h"
+#include "texmem.h"
+
+
+void intelInitTextureFuncs(struct dd_function_table *functions);
+
+const struct gl_texture_format *intelChooseTextureFormat(GLcontext * ctx,
+                                                         GLint internalFormat,
+                                                         GLenum format,
+                                                         GLenum type);
+
+
+void intelTexImage3D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint height, GLint depth,
+                     GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage3D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset, GLint yoffset, GLint zoffset,
+                        GLsizei width, GLsizei height, GLsizei depth,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelTexImage2D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint height, GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage2D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset, GLint yoffset,
+                        GLsizei width, GLsizei height,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelTexImage1D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage1D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset,
+                        GLsizei width,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
+                         GLenum internalFormat,
+                         GLint x, GLint y, GLsizei width, GLint border);
+
+void intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
+                         GLenum internalFormat,
+                         GLint x, GLint y, GLsizei width, GLsizei height,
+                         GLint border);
+
+void intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
+                            GLint xoffset, GLint x, GLint y, GLsizei width);
+
+void intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
+                            GLint xoffset, GLint yoffset,
+                            GLint x, GLint y, GLsizei width, GLsizei height);
+
+void intelGetTexImage(GLcontext * ctx, GLenum target, GLint level,
+                      GLenum format, GLenum type, GLvoid * pixels,
+                      struct gl_texture_object *texObj,
+                      struct gl_texture_image *texImage);
+
+void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
+				GLint internalFormat,
+				GLint width, GLint height, GLint border,
+				GLsizei imageSize, const GLvoid *data,
+				struct gl_texture_object *texObj,
+				struct gl_texture_image *texImage );
+
+void intelGetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
+				GLvoid *pixels,
+				struct gl_texture_object *texObj,
+				struct gl_texture_image *texImage);
+
+void intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+		       unsigned long long offset, GLint depth, GLuint pitch);
+void intelSetTexBuffer(__DRIcontext *pDRICtx,
+		       GLint target, __DRIdrawable *pDraw);
+
+GLuint intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit);
+
+void intel_tex_map_level_images(struct intel_context *intel,
+				struct intel_texture_object *intelObj,
+				int level);
+
+void intel_tex_unmap_level_images(struct intel_context *intel,
+				  struct intel_texture_object *intelObj,
+				  int level);
+
+void intel_tex_map_images(struct intel_context *intel,
+                          struct intel_texture_object *intelObj);
+
+void intel_tex_unmap_images(struct intel_context *intel,
+                            struct intel_texture_object *intelObj);
+
+int intel_compressed_num_bytes(GLuint mesaFormat);
+
+void intel_generate_mipmap(GLcontext *ctx, GLenum target,
+			   struct gl_texture_object *texObj);
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
new file mode 100644
index 00000000000..b893990d271
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -0,0 +1,306 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/teximage.h"
+#include "main/mipmap.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_buffers.h"
+#include "intel_mipmap_tree.h"
+#include "intel_regions.h"
+#include "intel_fbo.h"
+#include "intel_tex.h"
+#include "intel_blit.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+/**
+ * Get the intel_region which is the source for any glCopyTex[Sub]Image call.
+ *
+ * Do the best we can using the blitter.  A future project is to use
+ * the texture engine and fragment programs for these copies.
+ */
+static const struct intel_region *
+get_teximage_source(struct intel_context *intel, GLenum internalFormat)
+{
+   struct intel_renderbuffer *irb;
+
+   DBG("%s %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(internalFormat));
+
+   switch (internalFormat) {
+   case GL_DEPTH_COMPONENT:
+   case GL_DEPTH_COMPONENT16:
+      irb = intel_get_renderbuffer(intel->ctx.ReadBuffer, BUFFER_DEPTH);
+      if (irb && irb->region && irb->region->cpp == 2)
+         return irb->region;
+      return NULL;
+   case GL_DEPTH24_STENCIL8_EXT:
+   case GL_DEPTH_STENCIL_EXT:
+      irb = intel_get_renderbuffer(intel->ctx.ReadBuffer, BUFFER_DEPTH);
+      if (irb && irb->region && irb->region->cpp == 4)
+         return irb->region;
+      return NULL;
+   case GL_RGBA:
+   case GL_RGBA8:
+      return intel_readbuf_region(intel);
+   case GL_RGB:
+      if (intel->ctx.Visual.rgbBits == 16)
+         return intel_readbuf_region(intel);
+      return NULL;
+   default:
+      return NULL;
+   }
+}
+
+
+static GLboolean
+do_copy_texsubimage(struct intel_context *intel,
+		    GLenum target,
+                    struct intel_texture_image *intelImage,
+                    GLenum internalFormat,
+                    GLint dstx, GLint dsty,
+                    GLint x, GLint y, GLsizei width, GLsizei height)
+{
+   GLcontext *ctx = &intel->ctx;
+   struct gl_texture_object *texObj = intelImage->base.TexObject;
+   const struct intel_region *src =
+      get_teximage_source(intel, internalFormat);
+
+   if (!intelImage->mt || !src) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "%s fail %p %p\n",
+		 __FUNCTION__, intelImage->mt, src);
+      return GL_FALSE;
+   }
+
+   intelFlush(ctx);
+   LOCK_HARDWARE(intel);
+   {
+      GLuint image_offset = intel_miptree_image_offset(intelImage->mt,
+                                                       intelImage->face,
+                                                       intelImage->level);
+      const GLint orig_x = x;
+      const GLint orig_y = y;
+      const struct gl_framebuffer *fb = ctx->DrawBuffer;
+
+      if (_mesa_clip_to_region(fb->_Xmin, fb->_Ymin, fb->_Xmax, fb->_Ymax,
+                               &x, &y, &width, &height)) {
+	 GLshort src_pitch;
+
+         /* Update dst for clipped src.  Need to also clip the source rect.
+          */
+         dstx += x - orig_x;
+         dsty += y - orig_y;
+
+	 /* image_offset may be non-page-aligned, but that's illegal for tiling.
+	  */
+	 assert(intelImage->mt->region->tiling == I915_TILING_NONE);
+
+         if (ctx->ReadBuffer->Name == 0) {
+            /* reading from a window, adjust x, y */
+            __DRIdrawablePrivate *dPriv = intel->driDrawable;
+	    y = dPriv->y + (dPriv->h - (y + height));
+            x += dPriv->x;
+
+	    /* Invert the data coming from the source rectangle due to GL
+	     * and hardware disagreeing on where y=0 is.
+	     *
+	     * It appears that our offsets and pitches get mangled
+	     * appropriately by the hardware, and we don't need to adjust them
+	     * on our own.
+	     */
+	    src_pitch = -src->pitch;
+         }
+         else {
+            /* reading from a FBO, y is already oriented the way we like */
+	    src_pitch = src->pitch;
+         }
+
+         intelEmitCopyBlit(intel,
+                           intelImage->mt->cpp,
+                           src_pitch,
+                           src->buffer,
+                           0,
+			   src->tiling,
+                           intelImage->mt->pitch,
+                           intelImage->mt->region->buffer,
+                           image_offset,
+			   intelImage->mt->region->tiling,
+                           x, y, dstx, dsty, width, height,
+			   GL_COPY);
+      }
+   }
+
+
+   UNLOCK_HARDWARE(intel);
+
+   /* GL_SGIS_generate_mipmap */
+   if (intelImage->level == texObj->BaseLevel && texObj->GenerateMipmap) {
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
+   }
+
+   return GL_TRUE;
+}
+
+
+
+
+
+void
+intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
+                    GLenum internalFormat,
+                    GLint x, GLint y, GLsizei width, GLint border)
+{
+   struct gl_texture_unit *texUnit =
+      &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_object *texObj =
+      _mesa_select_tex_object(ctx, texUnit, target);
+   struct gl_texture_image *texImage =
+      _mesa_select_tex_image(ctx, texObj, target, level);
+
+   if (border)
+      goto fail;
+
+   /* Setup or redefine the texture object, mipmap tree and texture
+    * image.  Don't populate yet.  
+    */
+   ctx->Driver.TexImage1D(ctx, target, level, internalFormat,
+                          width, border,
+                          GL_RGBA, CHAN_TYPE, NULL,
+                          &ctx->DefaultPacking, texObj, texImage);
+
+   if (!do_copy_texsubimage(intel_context(ctx), target,
+                            intel_texture_image(texImage),
+                            internalFormat, 0, 0, x, y, width, 1))
+      goto fail;
+
+   return;
+
+ fail:
+   _swrast_copy_teximage1d(ctx, target, level, internalFormat, x, y,
+                           width, border);
+}
+
+void
+intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
+                    GLenum internalFormat,
+                    GLint x, GLint y, GLsizei width, GLsizei height,
+                    GLint border)
+{
+   struct gl_texture_unit *texUnit =
+      &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_object *texObj =
+      _mesa_select_tex_object(ctx, texUnit, target);
+   struct gl_texture_image *texImage =
+      _mesa_select_tex_image(ctx, texObj, target, level);
+
+   if (border)
+      goto fail;
+
+   /* Setup or redefine the texture object, mipmap tree and texture
+    * image.  Don't populate yet.  
+    */
+   ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
+                          width, height, border,
+                          GL_RGBA, CHAN_TYPE, NULL,
+                          &ctx->DefaultPacking, texObj, texImage);
+
+
+   if (!do_copy_texsubimage(intel_context(ctx), target,
+                            intel_texture_image(texImage),
+                            internalFormat, 0, 0, x, y, width, height))
+      goto fail;
+
+   return;
+
+ fail:
+   _swrast_copy_teximage2d(ctx, target, level, internalFormat, x, y,
+                           width, height, border);
+}
+
+
+void
+intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
+                       GLint xoffset, GLint x, GLint y, GLsizei width)
+{
+   struct gl_texture_unit *texUnit =
+      &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_object *texObj =
+      _mesa_select_tex_object(ctx, texUnit, target);
+   struct gl_texture_image *texImage =
+      _mesa_select_tex_image(ctx, texObj, target, level);
+   GLenum internalFormat = texImage->InternalFormat;
+
+   /* XXX need to check <border> as in above function? */
+
+   /* Need to check texture is compatible with source format. 
+    */
+
+   if (!do_copy_texsubimage(intel_context(ctx), target,
+                            intel_texture_image(texImage),
+                            internalFormat, xoffset, 0, x, y, width, 1)) {
+      _swrast_copy_texsubimage1d(ctx, target, level, xoffset, x, y, width);
+   }
+}
+
+
+
+void
+intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
+                       GLint xoffset, GLint yoffset,
+                       GLint x, GLint y, GLsizei width, GLsizei height)
+{
+   struct gl_texture_unit *texUnit =
+      &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   struct gl_texture_object *texObj =
+      _mesa_select_tex_object(ctx, texUnit, target);
+   struct gl_texture_image *texImage =
+      _mesa_select_tex_image(ctx, texObj, target, level);
+   GLenum internalFormat = texImage->InternalFormat;
+
+
+   /* Need to check texture is compatible with source format. 
+    */
+
+   if (!do_copy_texsubimage(intel_context(ctx), target,
+                            intel_texture_image(texImage),
+                            internalFormat,
+                            xoffset, yoffset, x, y, width, height)) {
+
+      DBG("%s - fallback to swrast\n", __FUNCTION__);
+
+      _swrast_copy_texsubimage2d(ctx, target, level,
+                                 xoffset, yoffset, x, y, width, height);
+   }
+}
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
new file mode 100644
index 00000000000..2be060dd3e3
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -0,0 +1,193 @@
+#include "intel_context.h"
+#include "intel_tex.h"
+#include "main/texformat.h"
+#include "main/enums.h"
+
+/* It works out that this function is fine for all the supported
+ * hardware.  However, there is still a need to map the formats onto
+ * hardware descriptors.
+ */
+/* Note that the i915 can actually support many more formats than
+ * these if we take the step of simply swizzling the colors
+ * immediately after sampling...
+ */
+const struct gl_texture_format *
+intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
+                         GLenum format, GLenum type)
+{
+   struct intel_context *intel = intel_context(ctx);
+   const GLboolean do32bpt = (intel->ctx.Visual.rgbBits == 32);
+
+   switch (internalFormat) {
+   case 4:
+   case GL_RGBA:
+   case GL_COMPRESSED_RGBA:
+      if (format == GL_BGRA) {
+         if (type == GL_UNSIGNED_BYTE || type == GL_UNSIGNED_INT_8_8_8_8_REV) {
+            return &_mesa_texformat_argb8888;
+         }
+         else if (type == GL_UNSIGNED_SHORT_4_4_4_4_REV) {
+            return &_mesa_texformat_argb4444;
+         }
+         else if (type == GL_UNSIGNED_SHORT_1_5_5_5_REV) {
+            return &_mesa_texformat_argb1555;
+         }
+      }
+      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+
+   case 3:
+   case GL_RGB:
+   case GL_COMPRESSED_RGB:
+      if (format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5) {
+         return &_mesa_texformat_rgb565;
+      }
+      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+
+   case GL_RGBA8:
+   case GL_RGB10_A2:
+   case GL_RGBA12:
+   case GL_RGBA16:
+      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+
+   case GL_RGBA4:
+   case GL_RGBA2:
+      return &_mesa_texformat_argb4444;
+
+   case GL_RGB5_A1:
+      return &_mesa_texformat_argb1555;
+
+   case GL_RGB8:
+   case GL_RGB10:
+   case GL_RGB12:
+   case GL_RGB16:
+      return &_mesa_texformat_argb8888;
+
+   case GL_RGB5:
+   case GL_RGB4:
+   case GL_R3_G3_B2:
+      return &_mesa_texformat_rgb565;
+
+   case GL_ALPHA:
+   case GL_ALPHA4:
+   case GL_ALPHA8:
+   case GL_ALPHA12:
+   case GL_ALPHA16:
+   case GL_COMPRESSED_ALPHA:
+      return &_mesa_texformat_a8;
+
+   case 1:
+   case GL_LUMINANCE:
+   case GL_LUMINANCE4:
+   case GL_LUMINANCE8:
+   case GL_LUMINANCE12:
+   case GL_LUMINANCE16:
+   case GL_COMPRESSED_LUMINANCE:
+      return &_mesa_texformat_l8;
+
+   case 2:
+   case GL_LUMINANCE_ALPHA:
+   case GL_LUMINANCE4_ALPHA4:
+   case GL_LUMINANCE6_ALPHA2:
+   case GL_LUMINANCE8_ALPHA8:
+   case GL_LUMINANCE12_ALPHA4:
+   case GL_LUMINANCE12_ALPHA12:
+   case GL_LUMINANCE16_ALPHA16:
+   case GL_COMPRESSED_LUMINANCE_ALPHA:
+      return &_mesa_texformat_al88;
+
+   case GL_INTENSITY:
+   case GL_INTENSITY4:
+   case GL_INTENSITY8:
+   case GL_INTENSITY12:
+   case GL_INTENSITY16:
+   case GL_COMPRESSED_INTENSITY:
+      return &_mesa_texformat_i8;
+
+   case GL_YCBCR_MESA:
+      if (type == GL_UNSIGNED_SHORT_8_8_MESA || type == GL_UNSIGNED_BYTE)
+         return &_mesa_texformat_ycbcr;
+      else
+         return &_mesa_texformat_ycbcr_rev;
+
+   case GL_COMPRESSED_RGB_FXT1_3DFX:
+      return &_mesa_texformat_rgb_fxt1;
+   case GL_COMPRESSED_RGBA_FXT1_3DFX:
+      return &_mesa_texformat_rgba_fxt1;
+
+   case GL_RGB_S3TC:
+   case GL_RGB4_S3TC:
+   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
+      return &_mesa_texformat_rgb_dxt1;
+
+   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
+      return &_mesa_texformat_rgba_dxt1;
+
+   case GL_RGBA_S3TC:
+   case GL_RGBA4_S3TC:
+   case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
+      return &_mesa_texformat_rgba_dxt3;
+
+   case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
+      return &_mesa_texformat_rgba_dxt5;
+
+   case GL_DEPTH_COMPONENT:
+   case GL_DEPTH_COMPONENT16:
+   case GL_DEPTH_COMPONENT24:
+   case GL_DEPTH_COMPONENT32:
+      return &_mesa_texformat_z16;
+
+   case GL_DEPTH_STENCIL_EXT:
+   case GL_DEPTH24_STENCIL8_EXT:
+      return &_mesa_texformat_s8_z24;
+
+#ifndef I915
+   case GL_SRGB_EXT:
+   case GL_SRGB8_EXT:
+   case GL_SRGB_ALPHA_EXT:
+   case GL_SRGB8_ALPHA8_EXT:
+   case GL_SLUMINANCE_EXT:
+   case GL_SLUMINANCE8_EXT:
+   case GL_SLUMINANCE_ALPHA_EXT:
+   case GL_SLUMINANCE8_ALPHA8_EXT:
+   case GL_COMPRESSED_SRGB_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_EXT:
+   case GL_COMPRESSED_SLUMINANCE_EXT:
+   case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
+       return &_mesa_texformat_srgba8;
+   case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
+     return &_mesa_texformat_srgb_dxt1;
+#endif
+
+   default:
+      fprintf(stderr, "unexpected texture format %s in %s\n",
+              _mesa_lookup_enum_by_nr(internalFormat), __FUNCTION__);
+      return NULL;
+   }
+
+   return NULL;                 /* never get here */
+}
+
+int intel_compressed_num_bytes(GLuint mesaFormat)
+{
+   int bytes = 0;
+   switch(mesaFormat) {
+     
+   case MESA_FORMAT_RGB_FXT1:
+   case MESA_FORMAT_RGBA_FXT1:
+   case MESA_FORMAT_RGB_DXT1:
+   case MESA_FORMAT_RGBA_DXT1:
+     bytes = 2;
+     break;
+     
+   case MESA_FORMAT_RGBA_DXT3:
+   case MESA_FORMAT_RGBA_DXT5:
+     bytes = 4;
+   default:
+     break;
+   }
+   
+   return bytes;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c
new file mode 100644
index 00000000000..2ac7dceb0fc
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_image.c
@@ -0,0 +1,778 @@
+
+#include <stdlib.h>
+#include <stdio.h>
+
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/colortab.h"
+#include "main/convolve.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/texcompress.h"
+#include "main/texformat.h"
+#include "main/texobj.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+
+#include "intel_context.h"
+#include "intel_mipmap_tree.h"
+#include "intel_buffer_objects.h"
+#include "intel_batchbuffer.h"
+#include "intel_tex.h"
+#include "intel_blit.h"
+#include "intel_fbo.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+/* Functions to store texture images.  Where possible, mipmap_tree's
+ * will be created or further instantiated with image data, otherwise
+ * images will be stored in malloc'd memory.  A validation step is
+ * required to pull those images into a mipmap tree, or otherwise
+ * decide a fallback is required.
+ */
+
+
+static int
+logbase2(int n)
+{
+   GLint i = 1;
+   GLint log2 = 0;
+
+   while (n > i) {
+      i *= 2;
+      log2++;
+   }
+
+   return log2;
+}
+
+
+/* Otherwise, store it in memory if (Border != 0) or (any dimension ==
+ * 1).
+ *    
+ * Otherwise, if max_level >= level >= min_level, create tree with
+ * space for textures from min_level down to max_level.
+ *
+ * Otherwise, create tree with space for textures from (level
+ * 0)..(1x1).  Consider pruning this tree at a validation if the
+ * saving is worth it.
+ */
+static void
+guess_and_alloc_mipmap_tree(struct intel_context *intel,
+                            struct intel_texture_object *intelObj,
+                            struct intel_texture_image *intelImage)
+{
+   GLuint firstLevel;
+   GLuint lastLevel;
+   GLuint width = intelImage->base.Width;
+   GLuint height = intelImage->base.Height;
+   GLuint depth = intelImage->base.Depth;
+   GLuint l2width, l2height, l2depth;
+   GLuint i, comp_byte = 0;
+
+   DBG("%s\n", __FUNCTION__);
+
+   if (intelImage->base.Border ||
+       ((intelImage->base._BaseFormat == GL_DEPTH_COMPONENT) && 
+        ((intelObj->base.WrapS == GL_CLAMP_TO_BORDER) ||
+         (intelObj->base.WrapT == GL_CLAMP_TO_BORDER))))
+      return;
+
+   if (intelImage->level > intelObj->base.BaseLevel &&
+       (intelImage->base.Width == 1 ||
+        (intelObj->base.Target != GL_TEXTURE_1D &&
+         intelImage->base.Height == 1) ||
+        (intelObj->base.Target == GL_TEXTURE_3D &&
+         intelImage->base.Depth == 1)))
+      return;
+
+   /* If this image disrespects BaseLevel, allocate from level zero.
+    * Usually BaseLevel == 0, so it's unlikely to happen.
+    */
+   if (intelImage->level < intelObj->base.BaseLevel)
+      firstLevel = 0;
+   else
+      firstLevel = intelObj->base.BaseLevel;
+
+
+   /* Figure out image dimensions at start level. 
+    */
+   for (i = intelImage->level; i > firstLevel; i--) {
+      width <<= 1;
+      if (height != 1)
+         height <<= 1;
+      if (depth != 1)
+         depth <<= 1;
+   }
+
+   /* Guess a reasonable value for lastLevel.  This is probably going
+    * to be wrong fairly often and might mean that we have to look at
+    * resizable buffers, or require that buffers implement lazy
+    * pagetable arrangements.
+    */
+   if ((intelObj->base.MinFilter == GL_NEAREST ||
+        intelObj->base.MinFilter == GL_LINEAR) &&
+       intelImage->level == firstLevel) {
+      lastLevel = firstLevel;
+   }
+   else {
+      l2width = logbase2(width);
+      l2height = logbase2(height);
+      l2depth = logbase2(depth);
+      lastLevel = firstLevel + MAX2(MAX2(l2width, l2height), l2depth);
+   }
+
+   assert(!intelObj->mt);
+   if (intelImage->base.IsCompressed)
+      comp_byte = intel_compressed_num_bytes(intelImage->base.TexFormat->MesaFormat);
+   intelObj->mt = intel_miptree_create(intel,
+                                       intelObj->base.Target,
+                                       intelImage->base.InternalFormat,
+                                       firstLevel,
+                                       lastLevel,
+                                       width,
+                                       height,
+                                       depth,
+                                       intelImage->base.TexFormat->TexelBytes,
+                                       comp_byte);
+
+   DBG("%s - success\n", __FUNCTION__);
+}
+
+
+
+
+static GLuint
+target_to_face(GLenum target)
+{
+   switch (target) {
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB:
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB:
+   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
+   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
+      return ((GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X);
+   default:
+      return 0;
+   }
+}
+
+/* There are actually quite a few combinations this will work for,
+ * more than what I've listed here.
+ */
+static GLboolean
+check_pbo_format(GLint internalFormat,
+                 GLenum format, GLenum type,
+                 const struct gl_texture_format *mesa_format)
+{
+   switch (internalFormat) {
+   case 4:
+   case GL_RGBA:
+      return (format == GL_BGRA &&
+              (type == GL_UNSIGNED_BYTE ||
+               type == GL_UNSIGNED_INT_8_8_8_8_REV) &&
+              mesa_format == &_mesa_texformat_argb8888);
+   case 3:
+   case GL_RGB:
+      return (format == GL_RGB &&
+              type == GL_UNSIGNED_SHORT_5_6_5 &&
+              mesa_format == &_mesa_texformat_rgb565);
+   case GL_YCBCR_MESA:
+      return (type == GL_UNSIGNED_SHORT_8_8_MESA || type == GL_UNSIGNED_BYTE);
+   default:
+      return GL_FALSE;
+   }
+}
+
+
+/* XXX: Do this for TexSubImage also:
+ */
+static GLboolean
+try_pbo_upload(struct intel_context *intel,
+               struct intel_texture_image *intelImage,
+               const struct gl_pixelstore_attrib *unpack,
+               GLint internalFormat,
+               GLint width, GLint height,
+               GLenum format, GLenum type, const void *pixels)
+{
+   struct intel_buffer_object *pbo = intel_buffer_object(unpack->BufferObj);
+   GLuint src_offset, src_stride;
+   GLuint dst_offset, dst_stride;
+
+   if (!pbo ||
+       intel->ctx._ImageTransferState ||
+       unpack->SkipPixels || unpack->SkipRows) {
+      _mesa_printf("%s: failure 1\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   src_offset = (GLuint) pixels;
+
+   if (unpack->RowLength > 0)
+      src_stride = unpack->RowLength;
+   else
+      src_stride = width;
+
+   dst_offset = intel_miptree_image_offset(intelImage->mt,
+                                           intelImage->face,
+                                           intelImage->level);
+
+   dst_stride = intelImage->mt->pitch;
+
+   intelFlush(&intel->ctx);
+   LOCK_HARDWARE(intel);
+   {
+      dri_bo *src_buffer = intel_bufferobj_buffer(intel, pbo, INTEL_READ);
+      dri_bo *dst_buffer = intel_region_buffer(intel,
+					       intelImage->mt->region,
+					       INTEL_WRITE_FULL);
+
+
+      intelEmitCopyBlit(intel,
+                        intelImage->mt->cpp,
+                        src_stride, src_buffer, src_offset, GL_FALSE,
+                        dst_stride, dst_buffer, dst_offset, GL_FALSE,
+                        0, 0, 0, 0, width, height,
+			GL_COPY);
+   }
+   UNLOCK_HARDWARE(intel);
+
+   return GL_TRUE;
+}
+
+
+
+static GLboolean
+try_pbo_zcopy(struct intel_context *intel,
+              struct intel_texture_image *intelImage,
+              const struct gl_pixelstore_attrib *unpack,
+              GLint internalFormat,
+              GLint width, GLint height,
+              GLenum format, GLenum type, const void *pixels)
+{
+   struct intel_buffer_object *pbo = intel_buffer_object(unpack->BufferObj);
+   GLuint src_offset, src_stride;
+   GLuint dst_offset, dst_stride;
+
+   if (!pbo ||
+       intel->ctx._ImageTransferState ||
+       unpack->SkipPixels || unpack->SkipRows) {
+      _mesa_printf("%s: failure 1\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   src_offset = (GLuint) pixels;
+
+   if (unpack->RowLength > 0)
+      src_stride = unpack->RowLength;
+   else
+      src_stride = width;
+
+   dst_offset = intel_miptree_image_offset(intelImage->mt,
+                                           intelImage->face,
+                                           intelImage->level);
+
+   dst_stride = intelImage->mt->pitch;
+
+   if (src_stride != dst_stride || dst_offset != 0 || src_offset != 0) {
+      _mesa_printf("%s: failure 2\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   intel_region_attach_pbo(intel, intelImage->mt->region, pbo);
+
+   return GL_TRUE;
+}
+
+
+
+
+
+
+static void
+intelTexImage(GLcontext * ctx,
+              GLint dims,
+              GLenum target, GLint level,
+              GLint internalFormat,
+              GLint width, GLint height, GLint depth,
+              GLint border,
+              GLenum format, GLenum type, const void *pixels,
+              const struct gl_pixelstore_attrib *unpack,
+              struct gl_texture_object *texObj,
+              struct gl_texture_image *texImage, GLsizei imageSize, int compressed)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
+   struct intel_texture_image *intelImage = intel_texture_image(texImage);
+   GLint postConvWidth = width;
+   GLint postConvHeight = height;
+   GLint texelBytes, sizeInBytes;
+   GLuint dstRowStride, srcRowStride = texImage->RowStride;
+
+
+   DBG("%s target %s level %d %dx%dx%d border %d\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(target), level, width, height, depth, border);
+
+   intelFlush(ctx);
+
+   intelImage->face = target_to_face(target);
+   intelImage->level = level;
+
+   if (ctx->_ImageTransferState & IMAGE_CONVOLUTION_BIT) {
+      _mesa_adjust_image_for_convolution(ctx, dims, &postConvWidth,
+                                         &postConvHeight);
+   }
+
+   /* choose the texture format */
+   texImage->TexFormat = intelChooseTextureFormat(ctx, internalFormat,
+                                                  format, type);
+
+   _mesa_set_fetch_functions(texImage, dims);
+
+   if (texImage->TexFormat->TexelBytes == 0) {
+      /* must be a compressed format */
+      texelBytes = 0;
+      texImage->IsCompressed = GL_TRUE;
+      texImage->CompressedSize =
+	 ctx->Driver.CompressedTextureSize(ctx, texImage->Width,
+					   texImage->Height, texImage->Depth,
+					   texImage->TexFormat->MesaFormat);
+   } else {
+      texelBytes = texImage->TexFormat->TexelBytes;
+      
+      /* Minimum pitch of 32 bytes */
+      if (postConvWidth * texelBytes < 32) {
+	 postConvWidth = 32 / texelBytes;
+	 texImage->RowStride = postConvWidth;
+      }
+
+      if (!intelImage->mt) {      
+	  assert(texImage->RowStride == postConvWidth);
+      }
+   }
+
+   /* Release the reference to a potentially orphaned buffer.   
+    * Release any old malloced memory.
+    */
+   if (intelImage->mt) {
+      intel_miptree_release(intel, &intelImage->mt);
+      assert(!texImage->Data);
+   }
+   else if (texImage->Data) {
+      _mesa_free_texmemory(texImage->Data);
+      texImage->Data = NULL;
+   }
+
+   /* If this is the only texture image in the tree, could call
+    * bmBufferData with NULL data to free the old block and avoid
+    * waiting on any outstanding fences.
+    */
+   if (intelObj->mt &&
+       intelObj->mt->first_level == level &&
+       intelObj->mt->last_level == level &&
+       intelObj->mt->target != GL_TEXTURE_CUBE_MAP_ARB &&
+       !intel_miptree_match_image(intelObj->mt, &intelImage->base,
+                                  intelImage->face, intelImage->level)) {
+
+      DBG("release it\n");
+      intel_miptree_release(intel, &intelObj->mt);
+      assert(!intelObj->mt);
+   }
+
+   if (!intelObj->mt) {
+      guess_and_alloc_mipmap_tree(intel, intelObj, intelImage);
+      if (!intelObj->mt) {
+	 DBG("guess_and_alloc_mipmap_tree: failed\n");
+      }
+   }
+
+   assert(!intelImage->mt);
+
+   if (intelObj->mt &&
+       intel_miptree_match_image(intelObj->mt, &intelImage->base,
+                                 intelImage->face, intelImage->level)) {
+
+      intel_miptree_reference(&intelImage->mt, intelObj->mt);
+      assert(intelImage->mt);
+   } else if (intelImage->base.Border == 0) {
+      int comp_byte = 0;
+
+      if (intelImage->base.IsCompressed) {
+	 comp_byte =
+	    intel_compressed_num_bytes(intelImage->base.TexFormat->MesaFormat);
+      }
+
+      /* Didn't fit in the object miptree, but it's suitable for inclusion in
+       * a miptree, so create one just for our level and store it in the image.
+       * It'll get moved into the object miptree at validate time.
+       */
+      intelImage->mt = intel_miptree_create(intel, target, internalFormat,
+					    level, level,
+					    width, height, depth,
+					    intelImage->base.TexFormat->TexelBytes,
+					    comp_byte);
+
+   }
+
+   /* PBO fastpaths:
+    */
+   if (dims <= 2 &&
+       intelImage->mt &&
+       intel_buffer_object(unpack->BufferObj) &&
+       check_pbo_format(internalFormat, format,
+                        type, intelImage->base.TexFormat)) {
+
+      DBG("trying pbo upload\n");
+
+      /* Attempt to texture directly from PBO data (zero copy upload).
+       *
+       * Currently disable as it can lead to worse as well as better
+       * performance (in particular when intel_region_cow() is
+       * required).
+       */
+      if (intelObj->mt == intelImage->mt &&
+          intelObj->mt->first_level == level &&
+          intelObj->mt->last_level == level) {
+
+         if (try_pbo_zcopy(intel, intelImage, unpack,
+                           internalFormat,
+                           width, height, format, type, pixels)) {
+
+            DBG("pbo zcopy upload succeeded\n");
+            return;
+         }
+      }
+
+
+      /* Otherwise, attempt to use the blitter for PBO image uploads.
+       */
+      if (try_pbo_upload(intel, intelImage, unpack,
+                         internalFormat,
+                         width, height, format, type, pixels)) {
+         DBG("pbo upload succeeded\n");
+         return;
+      }
+
+      DBG("pbo upload failed\n");
+   }
+
+
+
+   /* intelCopyTexImage calls this function with pixels == NULL, with
+    * the expectation that the mipmap tree will be set up but nothing
+    * more will be done.  This is where those calls return:
+    */
+   if (compressed) {
+      pixels = _mesa_validate_pbo_compressed_teximage(ctx, imageSize, pixels,
+						      unpack,
+						      "glCompressedTexImage");
+   } else {
+      pixels = _mesa_validate_pbo_teximage(ctx, dims, width, height, 1,
+					   format, type,
+					   pixels, unpack, "glTexImage");
+   }
+
+   LOCK_HARDWARE(intel);
+
+   if (intelImage->mt) {
+      texImage->Data = intel_miptree_image_map(intel,
+                                               intelImage->mt,
+                                               intelImage->face,
+                                               intelImage->level,
+                                               &dstRowStride,
+                                               intelImage->base.ImageOffsets);
+      texImage->RowStride = dstRowStride / intelImage->mt->cpp;
+   }
+   else {
+      /* Allocate regular memory and store the image there temporarily.   */
+      if (texImage->IsCompressed) {
+         sizeInBytes = texImage->CompressedSize;
+         dstRowStride =
+            _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
+         assert(dims != 3);
+      }
+      else {
+         dstRowStride = postConvWidth * texelBytes;
+         sizeInBytes = depth * dstRowStride * postConvHeight;
+      }
+
+      texImage->Data = _mesa_alloc_texmemory(sizeInBytes);
+   }
+
+   DBG("Upload image %dx%dx%d row_len %d "
+       "pitch %d\n",
+       width, height, depth, width * texelBytes, dstRowStride);
+
+   /* Copy data.  Would like to know when it's ok for us to eg. use
+    * the blitter to copy.  Or, use the hardware to do the format
+    * conversion and copy:
+    */
+   if (pixels) {
+       if (compressed) {
+	   if (intelImage->mt) {
+	       struct intel_region *dst = intelImage->mt->region;
+	       _mesa_copy_rect(texImage->Data, dst->cpp, dst->pitch,
+			       0, 0,
+			       intelImage->mt->level[level].width,
+			       intelImage->mt->level[level].height/4,
+			       pixels,
+			       srcRowStride,
+			       0, 0);
+	   } else
+	       memcpy(texImage->Data, pixels, imageSize);
+       } else if (!texImage->TexFormat->StoreImage(ctx, dims, 
+						   texImage->_BaseFormat, 
+						   texImage->TexFormat, 
+						   texImage->Data, 0, 0, 0, /* dstX/Y/Zoffset */
+						   dstRowStride,
+						   texImage->ImageOffsets,
+						   width, height, depth,
+						   format, type, pixels, unpack)) {
+	   _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
+       }
+   }
+
+   /* GL_SGIS_generate_mipmap */
+   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
+      intel_generate_mipmap(ctx, target, texObj);
+   }
+
+   _mesa_unmap_teximage_pbo(ctx, unpack);
+
+   if (intelImage->mt) {
+      intel_miptree_image_unmap(intel, intelImage->mt);
+      texImage->Data = NULL;
+   }
+
+   UNLOCK_HARDWARE(intel);
+}
+
+void
+intelTexImage3D(GLcontext * ctx,
+                GLenum target, GLint level,
+                GLint internalFormat,
+                GLint width, GLint height, GLint depth,
+                GLint border,
+                GLenum format, GLenum type, const void *pixels,
+                const struct gl_pixelstore_attrib *unpack,
+                struct gl_texture_object *texObj,
+                struct gl_texture_image *texImage)
+{
+   intelTexImage(ctx, 3, target, level,
+                 internalFormat, width, height, depth, border,
+                 format, type, pixels, unpack, texObj, texImage, 0, 0);
+}
+
+
+void
+intelTexImage2D(GLcontext * ctx,
+                GLenum target, GLint level,
+                GLint internalFormat,
+                GLint width, GLint height, GLint border,
+                GLenum format, GLenum type, const void *pixels,
+                const struct gl_pixelstore_attrib *unpack,
+                struct gl_texture_object *texObj,
+                struct gl_texture_image *texImage)
+{
+   intelTexImage(ctx, 2, target, level,
+                 internalFormat, width, height, 1, border,
+                 format, type, pixels, unpack, texObj, texImage, 0, 0);
+}
+
+void
+intelTexImage1D(GLcontext * ctx,
+                GLenum target, GLint level,
+                GLint internalFormat,
+                GLint width, GLint border,
+                GLenum format, GLenum type, const void *pixels,
+                const struct gl_pixelstore_attrib *unpack,
+                struct gl_texture_object *texObj,
+                struct gl_texture_image *texImage)
+{
+   intelTexImage(ctx, 1, target, level,
+                 internalFormat, width, 1, 1, border,
+                 format, type, pixels, unpack, texObj, texImage, 0, 0);
+}
+
+void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
+				GLint internalFormat,
+				GLint width, GLint height, GLint border,
+				GLsizei imageSize, const GLvoid *data,
+				struct gl_texture_object *texObj,
+				struct gl_texture_image *texImage )
+{
+   intelTexImage(ctx, 2, target, level,
+		 internalFormat, width, height, 1, border,
+		 0, 0, data, &ctx->Unpack, texObj, texImage, imageSize, 1);
+}
+
+/**
+ * Need to map texture image into memory before copying image data,
+ * then unmap it.
+ */
+static void
+intel_get_tex_image(GLcontext * ctx, GLenum target, GLint level,
+		    GLenum format, GLenum type, GLvoid * pixels,
+		    struct gl_texture_object *texObj,
+		    struct gl_texture_image *texImage, int compressed)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_image *intelImage = intel_texture_image(texImage);
+
+   /* Map */
+   if (intelImage->mt) {
+      /* Image is stored in hardware format in a buffer managed by the
+       * kernel.  Need to explicitly map and unmap it.
+       */
+      intelImage->base.Data =
+         intel_miptree_image_map(intel,
+                                 intelImage->mt,
+                                 intelImage->face,
+                                 intelImage->level,
+                                 &intelImage->base.RowStride,
+                                 intelImage->base.ImageOffsets);
+      intelImage->base.RowStride /= intelImage->mt->cpp;
+   }
+   else {
+      /* Otherwise, the image should actually be stored in
+       * intelImage->base.Data.  This is pretty confusing for
+       * everybody, I'd much prefer to separate the two functions of
+       * texImage->Data - storage for texture images in main memory
+       * and access (ie mappings) of images.  In other words, we'd
+       * create a new texImage->Map field and leave Data simply for
+       * storage.
+       */
+      assert(intelImage->base.Data);
+   }
+
+
+   if (compressed) {
+      _mesa_get_compressed_teximage(ctx, target, level, pixels,
+				    texObj, texImage);
+   } else {
+      _mesa_get_teximage(ctx, target, level, format, type, pixels,
+			 texObj, texImage);
+   }
+     
+
+   /* Unmap */
+   if (intelImage->mt) {
+      intel_miptree_image_unmap(intel, intelImage->mt);
+      intelImage->base.Data = NULL;
+   }
+}
+
+void
+intelGetTexImage(GLcontext * ctx, GLenum target, GLint level,
+                 GLenum format, GLenum type, GLvoid * pixels,
+                 struct gl_texture_object *texObj,
+                 struct gl_texture_image *texImage)
+{
+   intel_get_tex_image(ctx, target, level, format, type, pixels,
+		       texObj, texImage, 0);
+
+
+}
+
+void
+intelGetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
+			   GLvoid *pixels,
+			   struct gl_texture_object *texObj,
+			   struct gl_texture_image *texImage)
+{
+   intel_get_tex_image(ctx, target, level, 0, 0, pixels,
+		       texObj, texImage, 1);
+}
+
+void
+intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+		  unsigned long long offset, GLint depth, GLuint pitch)
+{
+   struct intel_context *intel = pDRICtx->driverPrivate;
+   struct gl_texture_object *tObj = _mesa_lookup_texture(&intel->ctx, texname);
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+
+   if (!intelObj)
+      return;
+
+   if (intelObj->mt)
+      intel_miptree_release(intel, &intelObj->mt);
+
+   intelObj->imageOverride = GL_TRUE;
+   intelObj->depthOverride = depth;
+   intelObj->pitchOverride = pitch;
+
+   if (offset)
+      intelObj->textureOffset = offset;
+}
+
+void
+intelSetTexBuffer(__DRIcontext *pDRICtx, GLint target, __DRIdrawable *dPriv)
+{
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+   struct intel_context *intel = pDRICtx->driverPrivate;
+   struct intel_texture_object *intelObj;
+   struct intel_texture_image *intelImage;
+   struct intel_mipmap_tree *mt;
+   struct intel_renderbuffer *rb;
+   struct gl_texture_unit *texUnit;
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+   int level = 0, type, format, internalFormat;
+
+   texUnit = &intel->ctx.Texture.Unit[intel->ctx.Texture.CurrentUnit];
+   texObj = _mesa_select_tex_object(&intel->ctx, texUnit, target);
+   intelObj = intel_texture_object(texObj);
+
+   if (!intelObj)
+      return;
+
+   intel_update_renderbuffers(pDRICtx, dPriv);
+
+   rb = intel_fb->color_rb[0];
+   /* If the region isn't set, then intel_update_renderbuffers was unable
+    * to get the buffers for the drawable.
+    */
+   if (rb->region == NULL)
+      return;
+
+   type = GL_BGRA;
+   format = GL_UNSIGNED_BYTE;
+   internalFormat = (rb->region->cpp == 3 ? 3 : 4);
+
+   mt = intel_miptree_create_for_region(intel, target,
+					internalFormat,
+					0, 0, rb->region, 1, 0);
+   if (mt == NULL)
+       return;
+
+   _mesa_lock_texture(&intel->ctx, texObj);
+
+   if (intelObj->mt)
+      intel_miptree_release(intel, &intelObj->mt);
+
+   intelObj->mt = mt;
+   texImage = _mesa_get_tex_image(&intel->ctx, texObj, target, level);
+   _mesa_init_teximage_fields(&intel->ctx, target, texImage,
+			      rb->region->width, rb->region->height, 1,
+			      0, internalFormat);
+
+   intelImage = intel_texture_image(texImage);
+   intelImage->face = target_to_face(target);
+   intelImage->level = level;
+   texImage->TexFormat = intelChooseTextureFormat(&intel->ctx, internalFormat,
+                                                  type, format);
+   _mesa_set_fetch_functions(texImage, 2);
+   texImage->RowStride = rb->region->pitch;
+   intel_miptree_reference(&intelImage->mt, intelObj->mt);
+
+   if (!intel_miptree_match_image(intelObj->mt, &intelImage->base,
+				  intelImage->face, intelImage->level)) {
+	   fprintf(stderr, "miptree doesn't match image\n");
+   }
+
+   _mesa_unlock_texture(&intel->ctx, texObj);
+}
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.c b/src/mesa/drivers/dri/intel/intel_tex_layout.c
index fcb5cc39068..e6f9a417790 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.c
@@ -32,15 +32,27 @@
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex_layout.h"
-#include "macros.h"
+#include "intel_context.h"
+#include "main/macros.h"
 
-
-static int align(int value, int alignment)
+GLuint intel_compressed_alignment(GLenum internalFormat)
 {
-   return (value + alignment - 1) & ~(alignment - 1);
+    GLuint alignment = 4;
+
+    switch (internalFormat) {
+    case GL_COMPRESSED_RGB_FXT1_3DFX:
+    case GL_COMPRESSED_RGBA_FXT1_3DFX:
+        alignment = 8;
+        break;
+
+    default:
+        break;
+    }
+
+    return alignment;
 }
 
-void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
+void i945_miptree_layout_2d( struct intel_context *intel, struct intel_mipmap_tree *mt )
 {
    GLint align_h = 2, align_w = 4;
    GLuint level;
@@ -51,23 +63,36 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 
    mt->pitch = mt->width0;
 
+   if (mt->compressed) {
+       align_w = intel_compressed_alignment(mt->internal_format);
+       mt->pitch = ALIGN(mt->width0, align_w);
+   }
+
    /* May need to adjust pitch to accomodate the placement of
     * the 2nd mipmap.  This occurs when the alignment
     * constraints of mipmap placement push the right edge of the
     * 2nd mipmap out past the width of its parent.
     */
    if (mt->first_level != mt->last_level) {
-      GLuint mip1_width = align(minify(mt->width0), align_w)
-			+ minify(minify(mt->width0));
+       GLuint mip1_width;
+
+       if (mt->compressed) {
+           mip1_width = ALIGN(minify(mt->width0), align_w)
+               + ALIGN(minify(minify(mt->width0)), align_w);
+       } else {
+           mip1_width = ALIGN(minify(mt->width0), align_w)
+               + minify(minify(mt->width0));
+       }
 
-      if (mip1_width > mt->width0)
-	 mt->pitch = mip1_width;
+       if (mip1_width > mt->pitch) {
+           mt->pitch = mip1_width;
+       }
    }
 
    /* Pitch must be a whole number of dwords, even though we
     * express it in texels.
     */
-   mt->pitch = align(mt->pitch * mt->cpp, 4) / mt->cpp;
+   mt->pitch = intel_miptree_pitch_align (intel, mt, mt->pitch);
    mt->total_height = 0;
 
    for ( level = mt->first_level ; level <= mt->last_level ; level++ ) {
@@ -79,7 +104,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
       if (mt->compressed)
 	 img_height = MAX2(1, height/4);
       else
-	 img_height = align(height, align_h);
+	 img_height = ALIGN(height, align_h);
 
 
       /* Because the images are packed better, the final offset
@@ -90,7 +115,7 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
       /* Layout_below: step right after second mipmap.
        */
       if (level == mt->first_level + 1) {
-	 x += align(width, align_w);
+	 x += ALIGN(width, align_w);
       }
       else {
 	 y += img_height;
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.h b/src/mesa/drivers/dri/intel/intel_tex_layout.h
index 1e37f8f525f..dbc90e6f9b7 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.h
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.h
@@ -30,7 +30,7 @@
   *   Michel Dänzer <michel@tungstengraphics.com>
   */
 
-#include "macros.h"
+#include "main/macros.h"
 
 
 static GLuint minify( GLuint d )
@@ -38,4 +38,5 @@ static GLuint minify( GLuint d )
    return MAX2(1, d>>1);
 }
 
-extern void i945_miptree_layout_2d( struct intel_mipmap_tree *mt );
+extern void i945_miptree_layout_2d( struct intel_context *intel, struct intel_mipmap_tree *mt );
+extern GLuint intel_compressed_alignment(GLenum);
diff --git a/src/mesa/drivers/dri/intel/intel_tex_obj.h b/src/mesa/drivers/dri/intel/intel_tex_obj.h
new file mode 100644
index 00000000000..5a934615259
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_obj.h
@@ -0,0 +1,83 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef _INTEL_TEX_OBJ_H
+#define _INTEL_TEX_OBJ_H
+
+struct intel_texture_object
+{
+   struct gl_texture_object base;       /* The "parent" object */
+
+   /* The mipmap tree must include at least these levels once
+    * validated:
+    */
+   GLuint firstLevel;
+   GLuint lastLevel;
+
+   /* Offset for firstLevel image:
+    */
+   GLuint textureOffset;
+
+   /* On validation any active images held in main memory or in other
+    * regions will be copied to this region and the old storage freed.
+    */
+   struct intel_mipmap_tree *mt;
+
+   GLboolean imageOverride;
+   GLint depthOverride;
+   GLuint pitchOverride;
+};
+
+struct intel_texture_image
+{
+   struct gl_texture_image base;
+
+   /* These aren't stored in gl_texture_image 
+    */
+   GLuint level;
+   GLuint face;
+
+   /* If intelImage->mt != NULL, image data is stored here.
+    * Else if intelImage->base.Data != NULL, image is stored there.
+    * Else there is no image data.
+    */
+   struct intel_mipmap_tree *mt;
+};
+
+static INLINE struct intel_texture_object *
+intel_texture_object(struct gl_texture_object *obj)
+{
+   return (struct intel_texture_object *) obj;
+}
+
+static INLINE struct intel_texture_image *
+intel_texture_image(struct gl_texture_image *img)
+{
+   return (struct intel_texture_image *) img;
+}
+
+#endif /* _INTEL_TEX_OBJ_H */
diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
new file mode 100644
index 00000000000..b7523618868
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
@@ -0,0 +1,186 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/mtypes.h"
+#include "main/texobj.h"
+#include "main/texstore.h"
+#include "main/texcompress.h"
+#include "main/enums.h"
+
+#include "intel_context.h"
+#include "intel_tex.h"
+#include "intel_mipmap_tree.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+static void
+intelTexSubimage(GLcontext * ctx,
+                 GLint dims,
+                 GLenum target, GLint level,
+                 GLint xoffset, GLint yoffset, GLint zoffset,
+                 GLint width, GLint height, GLint depth,
+                 GLenum format, GLenum type, const void *pixels,
+                 const struct gl_pixelstore_attrib *packing,
+                 struct gl_texture_object *texObj,
+                 struct gl_texture_image *texImage)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_image *intelImage = intel_texture_image(texImage);
+   GLuint dstRowStride = 0;
+   
+   DBG("%s target %s level %d offset %d,%d %dx%d\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(target),
+       level, xoffset, yoffset, width, height);
+
+   intelFlush(ctx);
+
+   pixels =
+      _mesa_validate_pbo_teximage(ctx, dims, width, height, depth, format,
+                                  type, pixels, packing, "glTexSubImage2D");
+   if (!pixels)
+      return;
+
+   LOCK_HARDWARE(intel);
+
+   /* Map buffer if necessary.  Need to lock to prevent other contexts
+    * from uploading the buffer under us.
+    */
+   if (intelImage->mt) 
+      texImage->Data = intel_miptree_image_map(intel,
+                                               intelImage->mt,
+                                               intelImage->face,
+                                               intelImage->level,
+                                               &dstRowStride,
+                                               texImage->ImageOffsets);
+   else {
+      if (texImage->IsCompressed) {
+         dstRowStride =
+            _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
+         assert(dims != 3);
+      }
+      else {
+         dstRowStride = texImage->RowStride * texImage->TexFormat->TexelBytes;
+      }
+   }
+
+   assert(dstRowStride);
+
+   if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat,
+                                        texImage->TexFormat,
+                                        texImage->Data,
+                                        xoffset, yoffset, zoffset,
+                                        dstRowStride,
+                                        texImage->ImageOffsets,
+                                        width, height, depth,
+                                        format, type, pixels, packing)) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage");
+   }
+
+   /* GL_SGIS_generate_mipmap */
+   if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
+      intel_generate_mipmap(ctx, target, texObj);
+   }
+
+   _mesa_unmap_teximage_pbo(ctx, packing);
+
+   if (intelImage->mt) {
+      intel_miptree_image_unmap(intel, intelImage->mt);
+      texImage->Data = NULL;
+   }
+
+   UNLOCK_HARDWARE(intel);
+}
+
+
+
+
+
+void
+intelTexSubImage3D(GLcontext * ctx,
+                   GLenum target,
+                   GLint level,
+                   GLint xoffset, GLint yoffset, GLint zoffset,
+                   GLsizei width, GLsizei height, GLsizei depth,
+                   GLenum format, GLenum type,
+                   const GLvoid * pixels,
+                   const struct gl_pixelstore_attrib *packing,
+                   struct gl_texture_object *texObj,
+                   struct gl_texture_image *texImage)
+{
+
+   intelTexSubimage(ctx, 3,
+                    target, level,
+                    xoffset, yoffset, zoffset,
+                    width, height, depth,
+                    format, type, pixels, packing, texObj, texImage);
+
+}
+
+
+
+void
+intelTexSubImage2D(GLcontext * ctx,
+                   GLenum target,
+                   GLint level,
+                   GLint xoffset, GLint yoffset,
+                   GLsizei width, GLsizei height,
+                   GLenum format, GLenum type,
+                   const GLvoid * pixels,
+                   const struct gl_pixelstore_attrib *packing,
+                   struct gl_texture_object *texObj,
+                   struct gl_texture_image *texImage)
+{
+
+   intelTexSubimage(ctx, 2,
+                    target, level,
+                    xoffset, yoffset, 0,
+                    width, height, 1,
+                    format, type, pixels, packing, texObj, texImage);
+
+}
+
+
+void
+intelTexSubImage1D(GLcontext * ctx,
+                   GLenum target,
+                   GLint level,
+                   GLint xoffset,
+                   GLsizei width,
+                   GLenum format, GLenum type,
+                   const GLvoid * pixels,
+                   const struct gl_pixelstore_attrib *packing,
+                   struct gl_texture_object *texObj,
+                   struct gl_texture_image *texImage)
+{
+   intelTexSubimage(ctx, 1,
+                    target, level,
+                    xoffset, 0, 0,
+                    width, 1, 1,
+                    format, type, pixels, packing, texObj, texImage);
+
+}
diff --git a/src/mesa/drivers/dri/intel/intel_tex_validate.c b/src/mesa/drivers/dri/intel/intel_tex_validate.c
new file mode 100644
index 00000000000..820683d42eb
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_tex_validate.c
@@ -0,0 +1,297 @@
+#include "main/mtypes.h"
+#include "main/macros.h"
+
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_mipmap_tree.h"
+#include "intel_tex.h"
+
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
+
+/**
+ * Compute which mipmap levels that really need to be sent to the hardware.
+ * This depends on the base image size, GL_TEXTURE_MIN_LOD,
+ * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
+ */
+static void
+intel_calculate_first_last_level(struct intel_texture_object *intelObj)
+{
+   struct gl_texture_object *tObj = &intelObj->base;
+   const struct gl_texture_image *const baseImage =
+      tObj->Image[0][tObj->BaseLevel];
+
+   /* These must be signed values.  MinLod and MaxLod can be negative numbers,
+    * and having firstLevel and lastLevel as signed prevents the need for
+    * extra sign checks.
+    */
+   int firstLevel;
+   int lastLevel;
+
+   /* Yes, this looks overly complicated, but it's all needed.
+    */
+   switch (tObj->Target) {
+   case GL_TEXTURE_1D:
+   case GL_TEXTURE_2D:
+   case GL_TEXTURE_3D:
+   case GL_TEXTURE_CUBE_MAP:
+      if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) {
+         /* GL_NEAREST and GL_LINEAR only care about GL_TEXTURE_BASE_LEVEL.
+          */
+         firstLevel = lastLevel = tObj->BaseLevel;
+      }
+      else {
+#ifdef I915
+         firstLevel = tObj->BaseLevel + (GLint) (tObj->MinLod + 0.5);
+         firstLevel = MAX2(firstLevel, tObj->BaseLevel);
+         firstLevel = MIN2(firstLevel, tObj->BaseLevel + baseImage->MaxLog2);
+         lastLevel = tObj->BaseLevel + (GLint) (tObj->MaxLod + 0.5);
+         lastLevel = MAX2(lastLevel, tObj->BaseLevel);
+         lastLevel = MIN2(lastLevel, tObj->BaseLevel + baseImage->MaxLog2);
+         lastLevel = MIN2(lastLevel, tObj->MaxLevel);
+         lastLevel = MAX2(firstLevel, lastLevel);       /* need at least one level */
+#else
+	 /* Currently not taking min/max lod into account here, those
+	  * values are programmed as sampler state elsewhere and we
+	  * upload the same mipmap levels regardless.  Not sure if
+	  * this makes sense as it means it isn't possible for the app
+	  * to use min/max lod to reduce texture memory pressure:
+	  */
+	 firstLevel = tObj->BaseLevel;
+	 lastLevel = MIN2(tObj->BaseLevel + baseImage->MaxLog2,
+			  tObj->MaxLevel);
+	 lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
+#endif
+      }
+      break;
+   case GL_TEXTURE_RECTANGLE_NV:
+   case GL_TEXTURE_4D_SGIS:
+      firstLevel = lastLevel = 0;
+      break;
+   default:
+      return;
+   }
+
+   /* save these values */
+   intelObj->firstLevel = firstLevel;
+   intelObj->lastLevel = lastLevel;
+}
+
+/**
+ * Copies the image's contents at its level into the object's miptree,
+ * and updates the image to point at the object's miptree.
+ */
+static void
+copy_image_data_to_tree(struct intel_context *intel,
+                        struct intel_texture_object *intelObj,
+                        struct intel_texture_image *intelImage)
+{
+   if (intelImage->mt) {
+      /* Copy potentially with the blitter:
+       */
+      intel_miptree_image_copy(intel,
+                               intelObj->mt,
+                               intelImage->face,
+                               intelImage->level, intelImage->mt);
+
+      intel_miptree_release(intel, &intelImage->mt);
+   }
+   else {
+      assert(intelImage->base.Data != NULL);
+
+      /* More straightforward upload.  
+       */
+      intel_miptree_image_data(intel,
+                               intelObj->mt,
+                               intelImage->face,
+                               intelImage->level,
+                               intelImage->base.Data,
+                               intelImage->base.RowStride,
+                               intelImage->base.RowStride *
+                               intelImage->base.Height);
+      _mesa_align_free(intelImage->base.Data);
+      intelImage->base.Data = NULL;
+   }
+
+   intel_miptree_reference(&intelImage->mt, intelObj->mt);
+}
+
+
+/*  
+ */
+GLuint
+intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
+{
+   struct gl_texture_object *tObj = intel->ctx.Texture.Unit[unit]._Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   int comp_byte = 0;
+   int cpp;
+   GLuint face, i;
+   GLuint nr_faces = 0;
+   struct intel_texture_image *firstImage;
+
+   /* We know/require this is true by now: 
+    */
+   assert(intelObj->base._Complete);
+
+   /* What levels must the tree include at a minimum?
+    */
+   intel_calculate_first_last_level(intelObj);
+   firstImage =
+      intel_texture_image(intelObj->base.Image[0][intelObj->firstLevel]);
+
+   /* Fallback case:
+    */
+   if (firstImage->base.Border) {
+      if (intelObj->mt) {
+         intel_miptree_release(intel, &intelObj->mt);
+      }
+      return GL_FALSE;
+   }
+
+
+   /* If both firstImage and intelObj have a tree which can contain
+    * all active images, favour firstImage.  Note that because of the
+    * completeness requirement, we know that the image dimensions
+    * will match.
+    */
+   if (firstImage->mt &&
+       firstImage->mt != intelObj->mt &&
+       firstImage->mt->first_level <= intelObj->firstLevel &&
+       firstImage->mt->last_level >= intelObj->lastLevel) {
+
+      if (intelObj->mt)
+         intel_miptree_release(intel, &intelObj->mt);
+
+      intel_miptree_reference(&intelObj->mt, firstImage->mt);
+   }
+
+   if (firstImage->base.IsCompressed) {
+      comp_byte = intel_compressed_num_bytes(firstImage->base.TexFormat->MesaFormat);
+      cpp = comp_byte;
+   }
+   else cpp = firstImage->base.TexFormat->TexelBytes;
+
+   /* Check tree can hold all active levels.  Check tree matches
+    * target, imageFormat, etc.
+    * 
+    * XXX: For some layouts (eg i945?), the test might have to be
+    * first_level == firstLevel, as the tree isn't valid except at the
+    * original start level.  Hope to get around this by
+    * programming minLod, maxLod, baseLevel into the hardware and
+    * leaving the tree alone.
+    */
+   if (intelObj->mt &&
+       (intelObj->mt->target != intelObj->base.Target ||
+	intelObj->mt->internal_format != firstImage->base.InternalFormat ||
+	intelObj->mt->first_level != intelObj->firstLevel ||
+	intelObj->mt->last_level != intelObj->lastLevel ||
+	intelObj->mt->width0 != firstImage->base.Width ||
+	intelObj->mt->height0 != firstImage->base.Height ||
+	intelObj->mt->depth0 != firstImage->base.Depth ||
+	intelObj->mt->cpp != cpp ||
+	intelObj->mt->compressed != firstImage->base.IsCompressed)) {
+      intel_miptree_release(intel, &intelObj->mt);
+   }
+
+
+   /* May need to create a new tree:
+    */
+   if (!intelObj->mt) {
+      intelObj->mt = intel_miptree_create(intel,
+                                          intelObj->base.Target,
+                                          firstImage->base.InternalFormat,
+                                          intelObj->firstLevel,
+                                          intelObj->lastLevel,
+                                          firstImage->base.Width,
+                                          firstImage->base.Height,
+                                          firstImage->base.Depth,
+                                          cpp,
+                                          comp_byte);
+   }
+
+   /* Pull in any images not in the object's tree:
+    */
+   nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
+   for (face = 0; face < nr_faces; face++) {
+      for (i = intelObj->firstLevel; i <= intelObj->lastLevel; i++) {
+         struct intel_texture_image *intelImage =
+            intel_texture_image(intelObj->base.Image[face][i]);
+
+         /* Need to import images in main memory or held in other trees.
+          */
+         if (intelObj->mt != intelImage->mt) {
+            copy_image_data_to_tree(intel, intelObj, intelImage);
+         }
+      }
+   }
+
+   return GL_TRUE;
+}
+
+void
+intel_tex_map_level_images(struct intel_context *intel,
+			   struct intel_texture_object *intelObj,
+			   int level)
+{
+   GLuint nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
+   GLuint face;
+
+   for (face = 0; face < nr_faces; face++) {
+      struct intel_texture_image *intelImage =
+	 intel_texture_image(intelObj->base.Image[face][level]);
+
+      if (intelImage->mt) {
+	 intelImage->base.Data =
+	    intel_miptree_image_map(intel,
+				    intelImage->mt,
+				    intelImage->face,
+				    intelImage->level,
+				    &intelImage->base.RowStride,
+				    intelImage->base.ImageOffsets);
+	 /* convert stride to texels, not bytes */
+	 intelImage->base.RowStride /= intelImage->mt->cpp;
+	 /* intelImage->base.ImageStride /= intelImage->mt->cpp; */
+      }
+   }
+}
+
+void
+intel_tex_unmap_level_images(struct intel_context *intel,
+			     struct intel_texture_object *intelObj,
+			     int level)
+{
+   GLuint nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
+   GLuint face;
+
+   for (face = 0; face < nr_faces; face++) {
+      struct intel_texture_image *intelImage =
+	 intel_texture_image(intelObj->base.Image[face][level]);
+
+      if (intelImage->mt) {
+	 intel_miptree_image_unmap(intel, intelImage->mt);
+	 intelImage->base.Data = NULL;
+      }
+   }
+}
+
+void
+intel_tex_map_images(struct intel_context *intel,
+                     struct intel_texture_object *intelObj)
+{
+   int i;
+
+   DBG("%s\n", __FUNCTION__);
+
+   for (i = intelObj->firstLevel; i <= intelObj->lastLevel; i++)
+      intel_tex_map_level_images(intel, intelObj, i);
+}
+
+void
+intel_tex_unmap_images(struct intel_context *intel,
+                       struct intel_texture_object *intelObj)
+{
+   int i;
+
+   for (i = intelObj->firstLevel; i <= intelObj->lastLevel; i++)
+      intel_tex_unmap_level_images(intel, intelObj, i);
+}
diff --git a/src/mesa/drivers/dri/i965/server/i830_dri.h b/src/mesa/drivers/dri/intel/server/i830_dri.h
index 68213f69f5d..def049e7a6b 100644
--- a/src/mesa/drivers/dri/i965/server/i830_dri.h
+++ b/src/mesa/drivers/dri/intel/server/i830_dri.h
@@ -1,14 +1,14 @@
+/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_dri.h,v 1.6 2003/09/28 20:15:59 alanh Exp $ */
 
 #ifndef _I830_DRI_H
 #define _I830_DRI_H
 
 #include "xf86drm.h"
-#include "i830_common.h"
 
 #define I830_MAX_DRAWABLES 256
 
 #define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 3
+#define I830_MINOR_VERSION 9
 #define I830_PATCHLEVEL 0
 
 #define I830_REG_SIZE 0x80000
@@ -23,7 +23,7 @@ typedef struct _I830DRIRec {
    drmSize unused3; /* depthbufferSize */
    drm_handle_t unused4; /* depthbuffer */
 
-   drmSize unused5; /* rotatedSize /*/
+   drmSize unused5; /* rotatedSize */
    drm_handle_t unused6; /* rotatedbuffer */
 
    drm_handle_t unused7; /* textures */
diff --git a/src/mesa/drivers/dri/i965/server/intel.h b/src/mesa/drivers/dri/intel/server/intel.h
index d7858a20c8d..6ea72499c1c 100644
--- a/src/mesa/drivers/dri/i965/server/intel.h
+++ b/src/mesa/drivers/dri/intel/server/intel.h
@@ -75,6 +75,9 @@
 
 #define I830_GMCH_CTRL		0x52
 
+#define I830_GMCH_MEM_MASK      0x1
+#define I830_GMCH_MEM_64M       0x1
+#define I830_GMCH_MEM_128M      0
 
 #define I830_GMCH_GMS_MASK			0x70
 #define I830_GMCH_GMS_DISABLED		0x00
@@ -141,7 +144,7 @@ typedef struct _I830Rec {
    unsigned char *MMIOBase;
    unsigned char *FbBase;
    int cpp;
-
+   uint32_t aper_size;
    unsigned int bios_version;
 
    /* These are set in PreInit and never changed. */
diff --git a/src/mesa/drivers/dri/intel/server/intel_dri.c b/src/mesa/drivers/dri/intel/server/intel_dri.c
new file mode 100644
index 00000000000..e49c4214ad4
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/server/intel_dri.c
@@ -0,0 +1,1306 @@
+/**
+ * \file server/intel_dri.c
+ * \brief File to perform the device-specific initialization tasks typically
+ * done in the X server.
+ *
+ * Here they are converted to run in the client (or perhaps a standalone
+ * process), and to work with the frame buffer device rather than the X
+ * server infrastructure.
+ * 
+ * Copyright (C) 2006 Dave Airlie (airlied@linux.ie)
+
+ Permission is hereby granted, free of charge, to any person obtaining a
+ copy of this software and associated documentation files (the
+ "Software"), to deal in the Software without restriction, including
+ without limitation the rights to use, copy, modify, merge, publish,
+ distribute, sub license, and/or sell copies of the Software, and to
+ permit persons to whom the Software is furnished to do so, subject to
+ the following conditions:
+ 
+ The above copyright notice and this permission notice (including the
+ next paragraph) shall be included in all copies or substantial portions
+ of the Software.
+ 
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ IN NO EVENT SHALL THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR
+ ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+*/
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+#include <unistd.h>
+
+#include "driver.h"
+#include "drm.h"
+
+#include "intel.h"
+#include "i830_dri.h"
+
+#include "memops.h"
+#include "pciaccess.h"
+
+static size_t drm_page_size;
+static int nextTile = 0;
+#define xf86DrvMsg(...) do {} while(0)
+
+static const int pitches[] = {
+  128 * 8,
+  128 * 16,
+  128 * 32,
+  128 * 64,
+  0
+};
+
+static Bool I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea);
+
+static unsigned long
+GetBestTileAlignment(unsigned long size)
+{
+   unsigned long i;
+
+   for (i = KB(512); i < size; i <<= 1)
+      ;
+
+   if (i > MB(64))
+      i = MB(64);
+
+   return i;
+}
+
+static void SetFenceRegs(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  int i;
+  unsigned char *MMIO = ctx->MMIOAddress;
+
+  for (i = 0; i < 8; i++) {
+    OUTREG(FENCE + i * 4, pI830->Fence[i]);
+    //    if (I810_DEBUG & DEBUG_VERBOSE_VGA)
+    fprintf(stderr,"Fence Register : %x\n", pI830->Fence[i]);
+  }
+}
+
+/* Tiled memory is good... really, really good...
+ *
+ * Need to make it less likely that we miss out on this - probably
+ * need to move the frontbuffer away from the 'guarenteed' alignment
+ * of the first memory segment, or perhaps allocate a discontigous
+ * framebuffer to get more alignment 'sweet spots'.
+ */
+static void
+SetFence(const DRIDriverContext *ctx, I830Rec *pI830,
+	 int nr, unsigned int start, unsigned int pitch,
+         unsigned int size)
+{
+   unsigned int val;
+   unsigned int fence_mask = 0;
+   unsigned int fence_pitch;
+
+   if (nr < 0 || nr > 7) {
+      fprintf(stderr,
+		 "SetFence: fence %d out of range\n",nr);
+      return;
+   }
+
+   pI830->Fence[nr] = 0;
+
+   if (IS_I9XX(pI830))
+   	fence_mask = ~I915G_FENCE_START_MASK;
+   else
+   	fence_mask = ~I830_FENCE_START_MASK;
+
+   if (start & fence_mask) {
+      fprintf(stderr,
+		 "SetFence: %d: start (0x%08x) is not %s aligned\n",
+		 nr, start, (IS_I9XX(pI830)) ? "1MB" : "512k");
+      return;
+   }
+
+   if (start % size) {
+      fprintf(stderr,
+		 "SetFence: %d: start (0x%08x) is not size (%dk) aligned\n",
+		 nr, start, size / 1024);
+      return;
+   }
+
+   if (pitch & 127) {
+      fprintf(stderr,
+		 "SetFence: %d: pitch (%d) not a multiple of 128 bytes\n",
+		 nr, pitch);
+      return;
+   }
+
+   val = (start | FENCE_X_MAJOR | FENCE_VALID);
+
+   if (IS_I9XX(pI830)) {
+   	switch (size) {
+	   case MB(1):
+      		val |= I915G_FENCE_SIZE_1M;
+      		break;
+   	   case MB(2):
+      		val |= I915G_FENCE_SIZE_2M;
+      		break;
+   	   case MB(4):
+      		val |= I915G_FENCE_SIZE_4M;
+      		break;
+   	   case MB(8):
+      		val |= I915G_FENCE_SIZE_8M;
+      		break;
+   	   case MB(16):
+      		val |= I915G_FENCE_SIZE_16M;
+      		break;
+   	   case MB(32):
+      		val |= I915G_FENCE_SIZE_32M;
+      		break;
+   	   case MB(64):
+      		val |= I915G_FENCE_SIZE_64M;
+      		break;
+   	   default:
+      		fprintf(stderr,
+		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
+      		return;
+   	}
+    } else {
+   	switch (size) {
+	   case KB(512):
+      		val |= FENCE_SIZE_512K;
+      		break;
+	   case MB(1):
+      		val |= FENCE_SIZE_1M;
+      		break;
+   	   case MB(2):
+      		val |= FENCE_SIZE_2M;
+      		break;
+   	   case MB(4):
+      		val |= FENCE_SIZE_4M;
+      		break;
+   	   case MB(8):
+      		val |= FENCE_SIZE_8M;
+      		break;
+   	   case MB(16):
+      		val |= FENCE_SIZE_16M;
+      		break;
+   	   case MB(32):
+      		val |= FENCE_SIZE_32M;
+      		break;
+   	   case MB(64):
+      		val |= FENCE_SIZE_64M;
+      		break;
+   	   default:
+      		fprintf(stderr,
+		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
+      		return;
+   	}
+   }
+
+   if (IS_I9XX(pI830))
+	fence_pitch = pitch / 512;
+   else
+	fence_pitch = pitch / 128;
+
+   switch (fence_pitch) {
+   case 1:
+      val |= FENCE_PITCH_1;
+      break;
+   case 2:
+      val |= FENCE_PITCH_2;
+      break;
+   case 4:
+      val |= FENCE_PITCH_4;
+      break;
+   case 8:
+      val |= FENCE_PITCH_8;
+      break;
+   case 16:
+      val |= FENCE_PITCH_16;
+      break;
+   case 32:
+      val |= FENCE_PITCH_32;
+      break;
+   case 64:
+      val |= FENCE_PITCH_64;
+      break;
+   default:
+      fprintf(stderr,
+		 "SetFence: %d: illegal pitch (%d)\n", nr, pitch);
+      return;
+   }
+
+   pI830->Fence[nr] = val;
+}
+
+static Bool
+MakeTiles(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *pMem)
+{
+   int pitch, ntiles, i;
+
+   pitch = pMem->Pitch * ctx->cpp;
+   /*
+    * Simply try to break the region up into at most four pieces of size
+    * equal to the alignment.
+    */
+   ntiles = ROUND_TO(pMem->Size, pMem->Alignment) / pMem->Alignment;
+   if (ntiles >= 4) {
+      return FALSE;
+   }
+
+   for (i = 0; i < ntiles; i++, nextTile++) {
+     SetFence(ctx, pI830, nextTile, pMem->Start + i * pMem->Alignment,
+	       pitch, pMem->Alignment);
+   }
+   return TRUE;
+}
+
+static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  int i;
+
+  /* Clear out */
+  for (i = 0; i < 8; i++)
+    pI830->Fence[i] = 0;
+  
+  nextTile = 0;
+
+  if (pI830->BackBuffer.Alignment >= KB(512)) {
+    if (MakeTiles(ctx, pI830, &(pI830->BackBuffer))) {
+      fprintf(stderr,
+		 "Activating tiled memory for the back buffer.\n");
+    } else {
+      fprintf(stderr,
+		 "MakeTiles failed for the back buffer.\n");
+      pI830->allowPageFlip = FALSE;
+    }
+  }
+  
+  if (pI830->DepthBuffer.Alignment >= KB(512)) {
+    if (MakeTiles(ctx, pI830, &(pI830->DepthBuffer))) {
+      fprintf(stderr,
+		 "Activating tiled memory for the depth buffer.\n");
+    } else {
+      fprintf(stderr,
+		 "MakeTiles failed for the depth buffer.\n");
+    }
+  }
+
+  return;
+}
+
+static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  struct pci_device host_bridge, ig_dev;
+  uint32_t gmch_ctrl;
+  int memsize = 0;
+  int range;
+  uint32_t aper_size;
+  uint32_t membase2 = 0;
+      
+  memset(&host_bridge, 0, sizeof(host_bridge));
+  memset(&ig_dev, 0, sizeof(ig_dev));
+
+  ig_dev.dev = 2;
+
+  pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
+
+  if (IS_I830(pI830) || IS_845G(pI830)) {
+    if ((gmch_ctrl & I830_GMCH_MEM_MASK) == I830_GMCH_MEM_128M) {
+      aper_size = 0x80000000;
+    } else {
+      aper_size = 0x40000000;
+    }
+  } else {
+    if (IS_I9XX(pI830)) {
+      int ret;
+      ret = pci_device_cfg_read_u32(&ig_dev, &membase2, 0x18);
+      if (membase2 & 0x08000000)
+	aper_size = 0x8000000;
+      else
+	aper_size = 0x10000000;
+
+      fprintf(stderr,"aper size is %08X %08x %d\n", aper_size, membase2, ret);
+    } else
+      aper_size = 0x8000000;
+  }
+
+  pI830->aper_size = aper_size;
+
+
+  /* We need to reduce the stolen size, by the GTT and the popup.
+   * The GTT varying according the the FbMapSize and the popup is 4KB */
+  range = (ctx->shared.fbSize / (1024*1024)) + 4;
+
+   if (IS_I85X(pI830) || IS_I865G(pI830) || IS_I9XX(pI830)) {
+      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
+      case I855_GMCH_GMS_STOLEN_1M:
+	 memsize = MB(1) - KB(range);
+	 break;
+      case I855_GMCH_GMS_STOLEN_4M:
+	 memsize = MB(4) - KB(range);
+	 break;
+      case I855_GMCH_GMS_STOLEN_8M:
+	 memsize = MB(8) - KB(range);
+	 break;
+      case I855_GMCH_GMS_STOLEN_16M:
+	 memsize = MB(16) - KB(range);
+	 break;
+      case I855_GMCH_GMS_STOLEN_32M:
+	 memsize = MB(32) - KB(range);
+	 break;
+      case I915G_GMCH_GMS_STOLEN_48M:
+	 if (IS_I9XX(pI830))
+	    memsize = MB(48) - KB(range);
+	 break;
+      case I915G_GMCH_GMS_STOLEN_64M:
+	 if (IS_I9XX(pI830))
+	    memsize = MB(64) - KB(range);
+	 break;
+      }
+   } else {
+      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
+      case I830_GMCH_GMS_STOLEN_512:
+	 memsize = KB(512) - KB(range);
+	 break;
+      case I830_GMCH_GMS_STOLEN_1024:
+	 memsize = MB(1) - KB(range);
+	 break;
+      case I830_GMCH_GMS_STOLEN_8192:
+	 memsize = MB(8) - KB(range);
+	 break;
+      case I830_GMCH_GMS_LOCAL:
+	 memsize = 0;
+	 xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
+		    "Local memory found, but won't be used.\n");
+	 break;
+      }
+   }
+   if (memsize > 0) {
+     fprintf(stderr,
+		 "detected %d kB stolen memory.\n", memsize / 1024);
+   } else {
+     fprintf(stderr,
+		 "no video memory detected.\n");
+   }
+   return memsize;
+}
+
+static int AgpInit(const DRIDriverContext *ctx, I830Rec *info)
+{
+  unsigned long mode = 0x4;
+
+  if (drmAgpAcquire(ctx->drmFD) < 0) {
+    fprintf(stderr, "[gart] AGP not available\n");
+    return 0;
+  }
+  
+  if (drmAgpEnable(ctx->drmFD, mode) < 0) {
+    fprintf(stderr, "[gart] AGP not enabled\n");
+    drmAgpRelease(ctx->drmFD);
+    return 0;
+  }
+  else
+    fprintf(stderr, "[gart] AGP enabled at %dx\n", ctx->agpmode);
+
+  return 1;
+}
+
+/*
+ * Allocate memory from the given pool.  Grow the pool if needed and if
+ * possible.
+ */
+static unsigned long
+AllocFromPool(const DRIDriverContext *ctx, I830Rec *pI830, 
+	      I830MemRange *result, I830MemPool *pool,
+	      long size, unsigned long alignment, int flags)
+{
+   long needed, start, end;
+
+   if (!result || !pool || !size)
+      return 0;
+
+   /* Calculate how much space is needed. */
+   if (alignment <= GTT_PAGE_SIZE)
+      needed = size;
+   else {
+	 start = ROUND_TO(pool->Free.Start, alignment);
+	 end = ROUND_TO(start + size, alignment);
+	 needed = end - pool->Free.Start;
+   }
+   if (needed > pool->Free.Size) {
+     return 0;
+   }
+
+   result->Start = ROUND_TO(pool->Free.Start, alignment);
+   pool->Free.Start += needed;
+   result->End = pool->Free.Start;
+
+   pool->Free.Size = pool->Free.End - pool->Free.Start;
+   result->Size = result->End - result->Start;
+   result->Pool = pool;
+   result->Alignment = alignment;
+   return needed;
+}
+
+static unsigned long AllocFromAGP(const DRIDriverContext *ctx, I830Rec *pI830, long size, unsigned long alignment, I830MemRange  *result)
+{
+   unsigned long start, end;
+   unsigned long newApStart, newApEnd;
+   int ret;
+   if (!result || !size)
+      return 0;
+   
+   if (!alignment)
+     alignment = 4;
+
+   start = ROUND_TO(pI830->MemoryAperture.Start, alignment);
+   end = ROUND_TO(start + size, alignment);
+   newApStart = end;
+   newApEnd = pI830->MemoryAperture.End;
+
+   ret=drmAgpAlloc(ctx->drmFD, size, 0, &(result->Physical), (drm_handle_t *)&(result->Key));
+   
+   if (ret)
+   {
+     fprintf(stderr,"drmAgpAlloc failed %d\n", ret);
+     return 0;
+   }
+   pI830->allocatedMemory += size;
+   pI830->MemoryAperture.Start = newApStart;
+   pI830->MemoryAperture.End = newApEnd;
+   pI830->MemoryAperture.Size = newApEnd - newApStart;
+   //   pI830->FreeMemory -= size;
+   result->Start = start;
+   result->End = start + size;
+   result->Size = size;
+   result->Offset = start;
+   result->Alignment = alignment;
+   result->Pool = NULL;
+  
+   return size;
+}
+
+unsigned long
+I830AllocVidMem(const DRIDriverContext *ctx, I830Rec *pI830,
+                I830MemRange *result, I830MemPool *pool, long size,
+                unsigned long alignment, int flags)
+{
+   unsigned long ret;
+
+   if (!result)
+      return 0;
+
+   /* Make sure these are initialised. */
+   result->Size = 0;
+   result->Key = -1;
+
+   if (!size) {
+      return 0;
+   }
+
+   if (pool->Free.Size < size) {
+      ret = AllocFromAGP(ctx, pI830, size, alignment, result);
+   }
+   else {
+      ret = AllocFromPool(ctx, pI830, result, pool, size, alignment, flags);
+      if (ret == 0)
+         ret = AllocFromAGP(ctx, pI830, size, alignment, result);
+   }
+   return ret;
+}
+
+static Bool BindAgpRange(const DRIDriverContext *ctx, I830MemRange *mem)
+{
+  if (!mem)
+    return FALSE;
+  
+  if (mem->Key == -1)
+    return TRUE;
+
+  return !drmAgpBind(ctx->drmFD, mem->Key, mem->Offset);
+}
+
+/* simple memory allocation routines needed */
+/* put ring buffer in low memory */
+/* need to allocate front, back, depth buffers aligned correctly,
+   allocate ring buffer, 
+*/
+
+/* */
+static Bool
+I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  unsigned long size, ret;
+  unsigned long lines, lineSize, align;
+
+  /* allocate ring buffer */
+  memset(pI830->LpRing, 0, sizeof(I830RingBuffer));
+  pI830->LpRing->mem.Key = -1;
+
+  size = PRIMARY_RINGBUFFER_SIZE;
+  
+  ret = I830AllocVidMem(ctx, pI830, &pI830->LpRing->mem, &pI830->StolenPool, size, 0x1000, 0);
+  
+  if (ret != size)
+  {
+    fprintf(stderr,"unable to allocate ring buffer %ld\n", ret);
+    return FALSE;
+  }
+
+  pI830->LpRing->tail_mask = pI830->LpRing->mem.Size - 1;
+
+  
+  /* allocate front buffer */
+  memset(&(pI830->FrontBuffer), 0, sizeof(pI830->FrontBuffer));
+  pI830->FrontBuffer.Key = -1;
+  pI830->FrontBuffer.Pitch = ctx->shared.virtualWidth;
+
+  align = KB(512);  
+
+  lineSize = ctx->shared.virtualWidth * ctx->cpp;
+  lines = (ctx->shared.virtualHeight + 15) / 16 * 16;
+  size = lineSize * lines;
+  size = ROUND_TO_PAGE(size);
+
+  align = GetBestTileAlignment(size);
+
+  ret = I830AllocVidMem(ctx, pI830, &pI830->FrontBuffer, &pI830->StolenPool, size, align, 0);
+  if (ret < size)
+  {
+    fprintf(stderr,"unable to allocate front buffer %ld\n", ret);
+    return FALSE;
+  }
+
+  memset(&(pI830->BackBuffer), 0, sizeof(pI830->BackBuffer));
+  pI830->BackBuffer.Key = -1;
+  pI830->BackBuffer.Pitch = ctx->shared.virtualWidth;
+
+  ret = I830AllocVidMem(ctx, pI830, &pI830->BackBuffer, &pI830->StolenPool, size, align, 0);
+  if (ret < size)
+  {
+    fprintf(stderr,"unable to allocate back buffer %ld\n", ret);
+    return FALSE;
+  }
+  
+  memset(&(pI830->DepthBuffer), 0, sizeof(pI830->DepthBuffer));
+  pI830->DepthBuffer.Key = -1;
+  pI830->DepthBuffer.Pitch = ctx->shared.virtualWidth;
+
+  ret = I830AllocVidMem(ctx, pI830, &pI830->DepthBuffer, &pI830->StolenPool, size, align, 0);
+  if (ret < size)
+  {
+    fprintf(stderr,"unable to allocate depth buffer %ld\n", ret);
+    return FALSE;
+  }
+
+  memset(&(pI830->ContextMem), 0, sizeof(pI830->ContextMem));
+  pI830->ContextMem.Key = -1;
+  size = KB(32);
+
+  ret = I830AllocVidMem(ctx, pI830, &pI830->ContextMem, &pI830->StolenPool, size, align, 0);
+  if (ret < size)
+  {
+    fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
+    return FALSE;
+  }
+
+#if 0  
+  memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
+  pI830->TexMem.Key = -1;
+
+  size = 32768 * 1024;
+  ret = AllocFromAGP(ctx, pI830, size, align, &pI830->TexMem);
+  if (ret < size)
+  {
+    fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
+    return FALSE;
+  }
+#endif
+
+  return TRUE;
+}
+
+static Bool
+I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  if (!BindAgpRange(ctx, &pI830->LpRing->mem))
+    return FALSE;
+  if (!BindAgpRange(ctx, &pI830->FrontBuffer))
+    return FALSE;
+  if (!BindAgpRange(ctx, &pI830->BackBuffer))
+    return FALSE;
+  if (!BindAgpRange(ctx, &pI830->DepthBuffer))
+    return FALSE;
+  if (!BindAgpRange(ctx, &pI830->ContextMem))
+    return FALSE;
+#if 0
+  if (!BindAgpRange(ctx, &pI830->TexMem))
+    return FALSE;
+#endif
+  return TRUE;
+}
+
+static void SetupDRIMM(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  unsigned long aperEnd = ROUND_DOWN_TO(pI830->aper_size, GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
+  unsigned long aperStart = ROUND_TO(pI830->aper_size - KB(32768), GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
+
+  fprintf(stderr, "aper size is %08X\n", ctx->shared.fbSize);
+  if (drmMMInit(ctx->drmFD, aperStart, aperEnd - aperStart, DRM_BO_MEM_TT)) {
+      fprintf(stderr,
+	      "DRM MM Initialization Failed\n");
+  } else {
+    fprintf(stderr,
+	    "DRM MM Initialized at offset 0x%lx length %d page\n", aperStart, aperEnd-aperStart);
+  }
+
+}
+
+static Bool
+I830CleanupDma(const DRIDriverContext *ctx)
+{
+   drmI830Init info;
+
+   memset(&info, 0, sizeof(drmI830Init));
+   info.func = I830_CLEANUP_DMA;
+
+   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
+		       &info, sizeof(drmI830Init))) {
+     fprintf(stderr, "I830 Dma Cleanup Failed\n");
+      return FALSE;
+   }
+
+   return TRUE;
+}
+
+static Bool
+I830InitDma(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+   I830RingBuffer *ring = pI830->LpRing;
+   drmI830Init info;
+
+   memset(&info, 0, sizeof(drmI830Init));
+   info.func = I830_INIT_DMA;
+
+   info.ring_start = ring->mem.Start + pI830->LinearAddr;
+   info.ring_end = ring->mem.End + pI830->LinearAddr;
+   info.ring_size = ring->mem.Size;
+
+   info.mmio_offset = (unsigned int)ctx->MMIOStart;
+
+   info.sarea_priv_offset = sizeof(drm_sarea_t);
+
+   info.front_offset = pI830->FrontBuffer.Start;
+   info.back_offset = pI830->BackBuffer.Start;
+   info.depth_offset = pI830->DepthBuffer.Start;
+   info.w = ctx->shared.virtualWidth;
+   info.h = ctx->shared.virtualHeight;
+   info.pitch = ctx->shared.virtualWidth;
+   info.back_pitch = pI830->BackBuffer.Pitch;
+   info.depth_pitch = pI830->DepthBuffer.Pitch;
+   info.cpp = ctx->cpp;
+
+   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
+		       &info, sizeof(drmI830Init))) {
+      fprintf(stderr,
+		 "I830 Dma Initialization Failed\n");
+      return FALSE;
+   }
+
+   return TRUE;
+}
+
+static int I830CheckDRMVersion( const DRIDriverContext *ctx,
+				  I830Rec *pI830 )
+{
+   drmVersionPtr  version;
+
+   version = drmGetVersion(ctx->drmFD);
+
+   if (version) {
+     int req_minor, req_patch;
+
+     req_minor = 4;
+     req_patch = 0;	
+
+     if (version->version_major != 1 ||
+	 version->version_minor < req_minor ||
+	 (version->version_minor == req_minor && 
+	  version->version_patchlevel < req_patch)) {
+       /* Incompatible drm version */
+       fprintf(stderr,
+	       "[dri] I830DRIScreenInit failed because of a version "
+	       "mismatch.\n"
+	       "[dri] i915.o kernel module version is %d.%d.%d "
+	       "but version 1.%d.%d or newer is needed.\n"
+	       "[dri] Disabling DRI.\n",
+	       version->version_major,
+	       version->version_minor,
+	       version->version_patchlevel,
+	       req_minor,
+	       req_patch);
+       drmFreeVersion(version);
+       return 0;
+     }
+     
+     pI830->drmMinor = version->version_minor;
+     drmFreeVersion(version);
+   }
+   return 1;
+}
+
+static void
+I830SetRingRegs(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  unsigned int itemp;
+  unsigned char *MMIO = ctx->MMIOAddress;
+
+   OUTREG(LP_RING + RING_LEN, 0);
+   OUTREG(LP_RING + RING_TAIL, 0);
+   OUTREG(LP_RING + RING_HEAD, 0);
+
+   if ((long)(pI830->LpRing->mem.Start & I830_RING_START_MASK) !=
+       pI830->LpRing->mem.Start) {
+      fprintf(stderr,
+		 "I830SetRingRegs: Ring buffer start (%lx) violates its "
+		 "mask (%x)\n", pI830->LpRing->mem.Start, I830_RING_START_MASK);
+   }
+   /* Don't care about the old value.  Reserved bits must be zero anyway. */
+   itemp = pI830->LpRing->mem.Start & I830_RING_START_MASK;
+   OUTREG(LP_RING + RING_START, itemp);
+
+   if (((pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES) !=
+       pI830->LpRing->mem.Size - 4096) {
+      fprintf(stderr,
+		 "I830SetRingRegs: Ring buffer size - 4096 (%lx) violates its "
+		 "mask (%x)\n", pI830->LpRing->mem.Size - 4096,
+		 I830_RING_NR_PAGES);
+   }
+   /* Don't care about the old value.  Reserved bits must be zero anyway. */
+   itemp = (pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES;
+   itemp |= (RING_NO_REPORT | RING_VALID);
+   OUTREG(LP_RING + RING_LEN, itemp);
+
+   pI830->LpRing->head = INREG(LP_RING + RING_HEAD) & I830_HEAD_MASK;
+   pI830->LpRing->tail = INREG(LP_RING + RING_TAIL);
+   pI830->LpRing->space = pI830->LpRing->head - (pI830->LpRing->tail + 8);
+   if (pI830->LpRing->space < 0)
+      pI830->LpRing->space += pI830->LpRing->mem.Size;
+
+   SetFenceRegs(ctx, pI830);
+   
+   /* RESET THE DISPLAY PIPE TO POINT TO THE FRONTBUFFER - hacky
+      hacky hacky */
+   OUTREG(DSPABASE, pI830->FrontBuffer.Start + pI830->LinearAddr);
+
+}
+
+static Bool
+I830SetParam(const DRIDriverContext *ctx, int param, int value)
+{
+   drmI830SetParam sp;
+
+   memset(&sp, 0, sizeof(sp));
+   sp.param = param;
+   sp.value = value;
+
+   if (drmCommandWrite(ctx->drmFD, DRM_I830_SETPARAM, &sp, sizeof(sp))) {
+      fprintf(stderr, "I830 SetParam Failed\n");
+      return FALSE;
+   }
+
+   return TRUE;
+}
+
+static Bool
+I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
+{
+   fprintf(stderr,
+              "[drm] Mapping front buffer\n");
+
+   if (drmAddMap(ctx->drmFD,
+                 (drm_handle_t)(sarea->front_offset + pI830->LinearAddr),
+                 sarea->front_size,
+                 DRM_FRAME_BUFFER,  /*DRM_AGP,*/
+                 0,
+                 &sarea->front_handle) < 0) {
+     fprintf(stderr,
+	     "[drm] drmAddMap(front_handle) failed. Disabling DRI\n");
+      return FALSE;
+   }
+   ctx->shared.hFrameBuffer = sarea->front_handle;
+   ctx->shared.fbSize = sarea->front_size;
+   fprintf(stderr, "[drm] Front Buffer = 0x%08x\n",
+	   sarea->front_handle);
+
+   if (drmAddMap(ctx->drmFD,
+                 (drm_handle_t)(sarea->back_offset),
+                 sarea->back_size, DRM_AGP, 0,
+                 &sarea->back_handle) < 0) {
+      fprintf(stderr,
+                 "[drm] drmAddMap(back_handle) failed. Disabling DRI\n");
+      return FALSE;
+   }
+   fprintf(stderr, "[drm] Back Buffer = 0x%08x\n",
+              sarea->back_handle);
+
+   if (drmAddMap(ctx->drmFD,
+                 (drm_handle_t)sarea->depth_offset,
+                 sarea->depth_size, DRM_AGP, 0,
+                 &sarea->depth_handle) < 0) {
+      fprintf(stderr,
+                 "[drm] drmAddMap(depth_handle) failed. Disabling DRI\n");
+      return FALSE;
+   }
+   fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
+              sarea->depth_handle);
+
+#if 0
+   if (drmAddMap(ctx->drmFD,
+		 (drm_handle_t)sarea->tex_offset,
+		 sarea->tex_size, DRM_AGP, 0,
+		 &sarea->tex_handle) < 0) {
+      fprintf(stderr,
+		 "[drm] drmAddMap(tex_handle) failed. Disabling DRI\n");
+      return FALSE;
+   }
+   fprintf(stderr, "[drm] textures = 0x%08x\n",
+	      sarea->tex_handle);
+#endif
+   return TRUE;
+}
+
+
+static void
+I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
+{
+#if 1
+   if (sarea->front_handle) {
+      drmRmMap(ctx->drmFD, sarea->front_handle);
+      sarea->front_handle = 0;
+   }
+#endif
+   if (sarea->back_handle) {
+      drmRmMap(ctx->drmFD, sarea->back_handle);
+      sarea->back_handle = 0;
+   }
+   if (sarea->depth_handle) {
+      drmRmMap(ctx->drmFD, sarea->depth_handle);
+      sarea->depth_handle = 0;
+   }
+   if (sarea->tex_handle) {
+      drmRmMap(ctx->drmFD, sarea->tex_handle);
+      sarea->tex_handle = 0;
+   }
+}
+
+static Bool
+I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
+{
+  if (drmAddMap(ctx->drmFD,
+		(drm_handle_t)pI830->LpRing->mem.Start,
+		pI830->LpRing->mem.Size, DRM_AGP, 0,
+		&pI830->ring_map) < 0) {
+    fprintf(stderr,
+	    "[drm] drmAddMap(ring_map) failed. Disabling DRI\n");
+    return FALSE;
+  }
+  fprintf(stderr, "[drm] ring buffer = 0x%08x\n",
+	  pI830->ring_map);
+
+  if (I830InitDma(ctx, pI830) == FALSE) {
+    return FALSE;
+  }
+  
+   /* init to zero to be safe */
+
+  I830DRIMapScreenRegions(ctx, pI830, sarea);
+  SetupDRIMM(ctx, pI830);
+
+   if (ctx->pciDevice != PCI_CHIP_845_G &&
+       ctx->pciDevice != PCI_CHIP_I830_M) {
+      I830SetParam(ctx, I830_SETPARAM_USE_MI_BATCHBUFFER_START, 1 );
+   }
+
+   /* Okay now initialize the dma engine */
+   {
+      pI830->irq = drmGetInterruptFromBusID(ctx->drmFD,
+					    ctx->pciBus,
+					    ctx->pciDevice,
+					    ctx->pciFunc);
+
+      if (drmCtlInstHandler(ctx->drmFD, pI830->irq)) {
+	 fprintf(stderr,
+		    "[drm] failure adding irq handler\n");
+	 pI830->irq = 0;
+	 return FALSE;
+      }
+      else
+	 fprintf(stderr,
+		    "[drm] dma control initialized, using IRQ %d\n",
+		    pI830->irq);
+   }
+
+   fprintf(stderr, "[dri] visual configs initialized\n");
+
+   return TRUE;
+}
+
+static Bool
+I830ClearScreen(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
+{
+  /* need to drmMap front and back buffers and zero them */
+  drmAddress map_addr;
+  int ret;
+
+  ret = drmMap(ctx->drmFD,
+	       sarea->front_handle,
+	       sarea->front_size,
+	       &map_addr);
+
+  if (ret)
+  {
+    fprintf(stderr, "Unable to map front buffer\n");
+    return FALSE;
+  }
+
+  drimemsetio((char *)map_addr,
+	      0,
+	      sarea->front_size);
+  drmUnmap(map_addr, sarea->front_size);
+
+
+  ret = drmMap(ctx->drmFD,
+	       sarea->back_handle,
+	       sarea->back_size,
+	       &map_addr);
+
+  if (ret)
+  {
+    fprintf(stderr, "Unable to map back buffer\n");
+    return FALSE;
+  }
+
+  drimemsetio((char *)map_addr,
+	      0,
+	      sarea->back_size);
+  drmUnmap(map_addr, sarea->back_size);
+
+  return TRUE;
+}
+
+static Bool
+I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
+		  
+{
+   I830DRIPtr pI830DRI;
+   drmI830Sarea *pSAREAPriv;
+   int err;
+      
+   drm_page_size = getpagesize();   
+
+   pI830->registerSize = ctx->MMIOSize;
+   /* This is a hack for now.  We have to have more than a 4k page here
+    * because of the size of the state.  However, the state should be
+    * in a per-context mapping.  This will be added in the Mesa 3.5 port
+    * of the I830 driver.
+    */
+   ctx->shared.SAREASize = SAREA_MAX;
+
+   /* Note that drmOpen will try to load the kernel module, if needed. */
+   ctx->drmFD = drmOpen("i915", NULL );
+   if (ctx->drmFD < 0) {
+      fprintf(stderr, "[drm] drmOpen failed\n");
+      return 0;
+   }
+
+   if ((err = drmSetBusid(ctx->drmFD, ctx->pciBusID)) < 0) {
+      fprintf(stderr, "[drm] drmSetBusid failed (%d, %s), %s\n",
+	      ctx->drmFD, ctx->pciBusID, strerror(-err));
+      return 0;
+   }
+
+   if (drmAddMap( ctx->drmFD,
+		  0,
+		  ctx->shared.SAREASize,
+		  DRM_SHM,
+		  DRM_CONTAINS_LOCK,
+		  &ctx->shared.hSAREA) < 0)
+   {
+     fprintf(stderr, "[drm] drmAddMap failed\n");
+     return 0;
+   }
+
+   fprintf(stderr, "[drm] added %d byte SAREA at 0x%08x\n",
+	   ctx->shared.SAREASize, ctx->shared.hSAREA);
+   
+   if (drmMap( ctx->drmFD,
+	       ctx->shared.hSAREA,
+	       ctx->shared.SAREASize,
+	       (drmAddressPtr)(&ctx->pSAREA)) < 0)
+   {
+      fprintf(stderr, "[drm] drmMap failed\n");
+      return 0;
+   
+   }
+   
+   memset(ctx->pSAREA, 0, ctx->shared.SAREASize);
+   fprintf(stderr, "[drm] mapped SAREA 0x%08x to %p, size %d\n",
+	   ctx->shared.hSAREA, ctx->pSAREA, ctx->shared.SAREASize);
+   
+
+   if (drmAddMap(ctx->drmFD, 
+		 ctx->MMIOStart,
+		 ctx->MMIOSize,
+		 DRM_REGISTERS, 
+		 DRM_READ_ONLY, 
+		 &pI830->registerHandle) < 0) {
+      fprintf(stderr, "[drm] drmAddMap mmio failed\n");	
+      return 0;
+   }
+   fprintf(stderr,
+	   "[drm] register handle = 0x%08x\n", pI830->registerHandle);
+
+
+   if (!I830CheckDRMVersion(ctx, pI830)) {
+     return FALSE;
+   }
+
+   /* Create a 'server' context so we can grab the lock for
+    * initialization ioctls.
+    */
+   if ((err = drmCreateContext(ctx->drmFD, &ctx->serverContext)) != 0) {
+      fprintf(stderr, "%s: drmCreateContext failed %d\n", __FUNCTION__, err);
+      return 0;
+   }
+
+   DRM_LOCK(ctx->drmFD, ctx->pSAREA, ctx->serverContext, 0); 
+
+   /* Initialize the SAREA private data structure */
+   pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
+				 sizeof(drm_sarea_t));
+   memset(pSAREAPriv, 0, sizeof(*pSAREAPriv));
+
+   pI830->StolenMemory.Size = I830DetectMemory(ctx, pI830);
+   pI830->StolenMemory.Start = 0;
+   pI830->StolenMemory.End = pI830->StolenMemory.Size;
+
+   pI830->MemoryAperture.Start = pI830->StolenMemory.End;
+   pI830->MemoryAperture.End = KB(40000);
+   pI830->MemoryAperture.Size = pI830->MemoryAperture.End - pI830->MemoryAperture.Start;
+
+   pI830->StolenPool.Fixed = pI830->StolenMemory;
+   pI830->StolenPool.Total = pI830->StolenMemory;
+   pI830->StolenPool.Free = pI830->StolenPool.Total;
+   pI830->FreeMemory = pI830->StolenPool.Total.Size;
+
+   if (!AgpInit(ctx, pI830))
+     return FALSE;
+
+   if (I830AllocateMemory(ctx, pI830) == FALSE)
+   {
+     return FALSE;
+   }
+
+   if (I830BindMemory(ctx, pI830) == FALSE)
+   {
+     return FALSE;
+   }
+
+   pSAREAPriv->rotated_offset = -1;
+   pSAREAPriv->rotated_size = 0;
+   pSAREAPriv->rotated_pitch = ctx->shared.virtualWidth;
+
+   pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
+   pSAREAPriv->front_size = pI830->FrontBuffer.Size;
+   pSAREAPriv->width = ctx->shared.virtualWidth;
+   pSAREAPriv->height = ctx->shared.virtualHeight;
+   pSAREAPriv->pitch = ctx->shared.virtualWidth;
+   pSAREAPriv->virtualX = ctx->shared.virtualWidth;
+   pSAREAPriv->virtualY = ctx->shared.virtualHeight;
+   pSAREAPriv->back_offset = pI830->BackBuffer.Start;
+   pSAREAPriv->back_size = pI830->BackBuffer.Size;
+   pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
+   pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
+#if 0
+   pSAREAPriv->tex_offset = pI830->TexMem.Start;
+   pSAREAPriv->tex_size = pI830->TexMem.Size;
+#endif
+   pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
+
+   ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
+   ctx->driverClientMsgSize = sizeof(I830DRIRec);
+   pI830DRI = (I830DRIPtr)ctx->driverClientMsg;
+   pI830DRI->deviceID = pI830->Chipset;
+   pI830DRI->regsSize = I830_REG_SIZE;
+   pI830DRI->width = ctx->shared.virtualWidth;
+   pI830DRI->height = ctx->shared.virtualHeight;
+   pI830DRI->mem = ctx->shared.fbSize;
+   pI830DRI->cpp = ctx->cpp;
+
+   pI830DRI->bitsPerPixel = ctx->bpp;
+   pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
+   
+   err = I830DRIDoMappings(ctx, pI830, pSAREAPriv);
+   if (err == FALSE)
+       return FALSE;
+
+   I830SetupMemoryTiling(ctx, pI830);
+
+   /* Quick hack to clear the front & back buffers.  Could also use
+    * the clear ioctl to do this, but would need to setup hw state
+    * first.
+    */
+   I830ClearScreen(ctx, pI830, pSAREAPriv);
+
+   I830SetRingRegs(ctx, pI830);
+
+   return TRUE;
+}
+
+
+/**
+ * \brief Validate the fbdev mode.
+ * 
+ * \param ctx display handle.
+ *
+ * \return one on success, or zero on failure.
+ *
+ * Saves some registers and returns 1.
+ *
+ * \sa radeonValidateMode().
+ */
+static int i830ValidateMode( const DRIDriverContext *ctx )
+{
+  return 1;
+}
+
+/**
+ * \brief Examine mode returned by fbdev.
+ * 
+ * \param ctx display handle.
+ *
+ * \return one on success, or zero on failure.
+ *
+ * Restores registers that fbdev has clobbered and returns 1.
+ *
+ * \sa i810ValidateMode().
+ */
+static int i830PostValidateMode( const DRIDriverContext *ctx )
+{
+  I830Rec *pI830 = ctx->driverPrivate;
+
+  I830SetRingRegs(ctx, pI830);
+  return 1;
+}
+
+
+/**
+ * \brief Initialize the framebuffer device mode
+ *
+ * \param ctx display handle.
+ *
+ * \return one on success, or zero on failure.
+ *
+ * Fills in \p info with some default values and some information from \p ctx
+ * and then calls I810ScreenInit() for the screen initialization.
+ * 
+ * Before exiting clears the framebuffer memory accessing it directly.
+ */
+static int i830InitFBDev( DRIDriverContext *ctx )
+{
+  I830Rec *pI830 = calloc(1, sizeof(I830Rec));
+  int i;
+
+   {
+      int  dummy = ctx->shared.virtualWidth;
+
+      switch (ctx->bpp / 8) {
+      case 1: dummy = (ctx->shared.virtualWidth + 127) & ~127; break;
+      case 2: dummy = (ctx->shared.virtualWidth +  31) &  ~31; break;
+      case 3:
+      case 4: dummy = (ctx->shared.virtualWidth +  15) &  ~15; break;
+      }
+
+      ctx->shared.virtualWidth = dummy;
+      ctx->shared.Width = ctx->shared.virtualWidth;
+   }
+
+
+   for (i = 0; pitches[i] != 0; i++) {
+     if (pitches[i] >= ctx->shared.virtualWidth) {
+       ctx->shared.virtualWidth = pitches[i];
+       break;
+     }
+   }
+
+   ctx->driverPrivate = (void *)pI830;
+   
+   pI830->LpRing = calloc(1, sizeof(I830RingBuffer));
+   pI830->Chipset = ctx->chipset;
+   pI830->LinearAddr = ctx->FBStart;
+
+   if (!I830ScreenInit( ctx, pI830 ))
+      return 0;
+
+   
+   return 1;
+}
+
+
+/**
+ * \brief The screen is being closed, so clean up any state and free any
+ * resources used by the DRI.
+ *
+ * \param ctx display handle.
+ *
+ * Unmaps the SAREA, closes the DRM device file descriptor and frees the driver
+ * private data.
+ */
+static void i830HaltFBDev( DRIDriverContext *ctx )
+{
+  drmI830Sarea *pSAREAPriv;
+  I830Rec *pI830 = ctx->driverPrivate;
+
+   if (pI830->irq) {
+       drmCtlUninstHandler(ctx->drmFD);
+       pI830->irq = 0;   }
+
+   I830CleanupDma(ctx);
+
+  pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
+				sizeof(drm_sarea_t));
+
+  I830DRIUnmapScreenRegions(ctx, pI830, pSAREAPriv);
+  drmUnmap( ctx->pSAREA, ctx->shared.SAREASize );
+  drmClose(ctx->drmFD);
+  
+  if (ctx->driverPrivate) {
+    free(ctx->driverPrivate);
+    ctx->driverPrivate = 0;
+  }
+}
+
+
+extern void i810NotifyFocus( int );
+
+/**
+ * \brief Exported driver interface for Mini GLX.
+ *
+ * \sa DRIDriverRec.
+ */
+const struct DRIDriverRec __driDriver = {
+   i830ValidateMode,
+   i830PostValidateMode,
+   i830InitFBDev,
+   i830HaltFBDev,
+   NULL,//I830EngineShutdown,
+   NULL, //I830EngineRestore,  
+#ifndef _EMBEDDED
+   0,
+#else
+   i810NotifyFocus, 
+#endif
+};
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.c b/src/mesa/drivers/dri/mach64/mach64_context.c
index 7f558e92bc6..99abd209b69 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.c
+++ b/src/mesa/drivers/dri/mach64/mach64_context.c
@@ -29,12 +29,12 @@
  *	Jos�Fonseca <j_r_fonseca@yahoo.co.uk>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -253,9 +253,6 @@ GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
 
    mmesa->do_irqs = (mmesa->mach64Screen->irq && !getenv("MACH64_NO_IRQS"));
 
-   mmesa->vblank_flags = (mmesa->do_irqs)
-      ? driGetDefaultVBlankFlags(&mmesa->optionCache) : VBLANK_FLAG_NO_IRQ;
-
    driContextPriv->driverPrivate = (void *)mmesa;
 
    if (driQueryOptionb(&mmesa->optionCache, "no_rast")) {
@@ -330,10 +327,15 @@ mach64MakeCurrent( __DRIcontextPrivate *driContextPriv,
       }
 
       
-      driDrawableInitVBlank( driDrawPriv, newMach64Ctx->vblank_flags,
-			     &newMach64Ctx->vbl_seq );
-
       if ( newMach64Ctx->driDrawable != driDrawPriv ) {
+	 if (driDrawPriv->swap_interval == (unsigned)-1) {
+	    driDrawPriv->vblFlags = (newMach64Ctx->do_irqs)
+	       ? driGetDefaultVBlankFlags(&newMach64Ctx->optionCache)
+	       : VBLANK_FLAG_NO_IRQ;
+
+	    driDrawableInitVBlank( driDrawPriv );
+	 }
+
 	 newMach64Ctx->driDrawable = driDrawPriv;
 	 mach64CalcViewport( newMach64Ctx->glCtx );
       }
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.h b/src/mesa/drivers/dri/mach64/mach64_context.h
index e925f18c114..55e0618ff80 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.h
+++ b/src/mesa/drivers/dri/mach64/mach64_context.h
@@ -36,7 +36,7 @@
 #include "drm.h"
 #include "mach64_drm.h"
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 #include "mach64_reg.h"
 
@@ -263,8 +263,6 @@ struct mach64_context {
 
    /* VBI
     */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
    GLuint do_irqs;
 
    /* Configuration cache
diff --git a/src/mesa/drivers/dri/mach64/mach64_dd.c b/src/mesa/drivers/dri/mach64/mach64_dd.c
index 7d225ebc888..e400e9a918d 100644
--- a/src/mesa/drivers/dri/mach64/mach64_dd.c
+++ b/src/mesa/drivers/dri/mach64/mach64_dd.c
@@ -35,9 +35,10 @@
 #include "mach64_vb.h"
 #include "mach64_dd.h"
 
-#include "context.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+
 #include "utils.h"
-#include "framebuffer.h"
 
 #define DRIVER_DATE	"20051019"
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_ioctl.c b/src/mesa/drivers/dri/mach64/mach64_ioctl.c
index 6bc2b58ce94..ef5c0625c31 100644
--- a/src/mesa/drivers/dri/mach64/mach64_ioctl.c
+++ b/src/mesa/drivers/dri/mach64/mach64_ioctl.c
@@ -35,8 +35,8 @@
 #include "mach64_ioctl.h"
 #include "mach64_tex.h"
 
-#include "imports.h"
-#include "macros.h"
+#include "main/imports.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 
@@ -279,7 +279,7 @@ static int mach64WaitForFrameCompletion( mach64ContextPtr mmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void mach64CopyBuffer( const __DRIdrawablePrivate *dPriv )
+void mach64CopyBuffer( __DRIdrawablePrivate *dPriv )
 {
    mach64ContextPtr mmesa;
    GLint nbox, i, ret;
@@ -320,7 +320,7 @@ void mach64CopyBuffer( const __DRIdrawablePrivate *dPriv )
 #endif
 
    UNLOCK_HARDWARE( mmesa );
-   driWaitForVBlank( dPriv, &mmesa->vbl_seq, mmesa->vblank_flags, &missed_target );
+   driWaitForVBlank( dPriv, &missed_target );
    LOCK_HARDWARE( mmesa );
 
    /* use front buffer cliprects */
diff --git a/src/mesa/drivers/dri/mach64/mach64_ioctl.h b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
index 2153ab80d78..6ef9bc0bcaf 100644
--- a/src/mesa/drivers/dri/mach64/mach64_ioctl.h
+++ b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
@@ -44,7 +44,7 @@ extern void mach64FlushVerticesLocked( mach64ContextPtr mmesa );
 extern void mach64FlushDMALocked( mach64ContextPtr mmesa );
 extern void mach64UploadHwStateLocked( mach64ContextPtr mmesa );
 
-static __inline void *mach64AllocDmaLow( mach64ContextPtr mmesa, int bytes )
+static INLINE void *mach64AllocDmaLow( mach64ContextPtr mmesa, int bytes )
 {
    CARD32 *head;
 
@@ -60,7 +60,7 @@ static __inline void *mach64AllocDmaLow( mach64ContextPtr mmesa, int bytes )
    return head;
 }
 
-static __inline void *mach64AllocDmaLocked( mach64ContextPtr mmesa, int bytes )
+static INLINE void *mach64AllocDmaLocked( mach64ContextPtr mmesa, int bytes )
 {
    CARD32 *head;
 
@@ -78,7 +78,7 @@ extern void mach64FireBlitLocked( mach64ContextPtr mmesa, void *buffer,
 				  GLint offset, GLint pitch, GLint format,
 				  GLint x, GLint y, GLint width, GLint height );
 
-extern void mach64CopyBuffer( const __DRIdrawablePrivate *dPriv );
+extern void mach64CopyBuffer( __DRIdrawablePrivate *dPriv );
 #if ENABLE_PERF_BOXES
 extern void mach64PerformanceCounters( mach64ContextPtr mmesa );
 extern void mach64PerformanceBoxesLocked( mach64ContextPtr mmesa );
diff --git a/src/mesa/drivers/dri/mach64/mach64_lock.c b/src/mesa/drivers/dri/mach64/mach64_lock.c
index ea605fb0614..d018ba41748 100644
--- a/src/mesa/drivers/dri/mach64/mach64_lock.c
+++ b/src/mesa/drivers/dri/mach64/mach64_lock.c
@@ -70,7 +70,7 @@ void mach64GetLock( mach64ContextPtr mmesa, GLuint flags )
 
    if ( mmesa->lastStamp != dPriv->lastStamp ) {
       mmesa->lastStamp = dPriv->lastStamp;
-      if (mmesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT)
+      if (mmesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT)
          mach64SetCliprects( mmesa->glCtx, GL_BACK_LEFT );
       else
          mach64SetCliprects( mmesa->glCtx, GL_FRONT_LEFT );
diff --git a/src/mesa/drivers/dri/mach64/mach64_screen.c b/src/mesa/drivers/dri/mach64/mach64_screen.c
index 300a3deaaf9..6bfb4c32b15 100644
--- a/src/mesa/drivers/dri/mach64/mach64_screen.c
+++ b/src/mesa/drivers/dri/mach64/mach64_screen.c
@@ -35,10 +35,10 @@
 #include "mach64_vb.h"
 #include "mach64_span.h"
 
-#include "context.h"
-#include "imports.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "utils.h"
 #include "vblank.h"
@@ -69,79 +69,15 @@ static const GLuint __driNConfigOptions = 2;
 
 extern const struct dri_extension card_extensions[];
 
-static __GLcontextModes * fill_in_modes( __GLcontextModes * modes,
-					 unsigned pixel_bits, 
-					 unsigned depth_bits,
-					 unsigned stencil_bits,
-					 const GLenum * db_modes,
-					 unsigned num_db_modes,
-					 int visType )
+static const __DRIconfig **
+mach64FillInModes( __DRIscreenPrivate *psp,
+		   unsigned pixel_bits, unsigned depth_bits,
+		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
-       static const uint8_t bits[2][4] = {
-	{          5,          6,          5,          0 },
-	{          8,          8,          8,          0 }
-    };
-
-    static const uint32_t masks[2][4] = {
-	{ 0x0000F800, 0x000007E0, 0x0000001F, 0x00000000 },
-	{ 0x00FF0000, 0x0000FF00, 0x000000FF, 0x00000000 }
-    };
-
-    unsigned   i;
-    unsigned   j;
-    const unsigned index = ((pixel_bits + 15) / 16) - 1;
-
-    for ( i = 0 ; i < num_db_modes ; i++ ) {
-	for ( j = 0 ; j < 2 ; j++ ) {
-
-	    modes->redBits   = bits[index][0];
-	    modes->greenBits = bits[index][1];
-	    modes->blueBits  = bits[index][2];
-	    modes->alphaBits = bits[index][3];
-	    modes->redMask   = masks[index][0];
-	    modes->greenMask = masks[index][1];
-	    modes->blueMask  = masks[index][2];
-	    modes->alphaMask = masks[index][3];
-	    modes->rgbBits   = modes->redBits + modes->greenBits
-		+ modes->blueBits + modes->alphaBits;
-
-	    modes->accumRedBits   = 16 * j;
-	    modes->accumGreenBits = 16 * j;
-	    modes->accumBlueBits  = 16 * j;
-	    modes->accumAlphaBits = 0;
-	    modes->visualRating = (j == 0) ? GLX_NONE : GLX_SLOW_CONFIG;
-	    modes->drawableType = GLX_WINDOW_BIT | GLX_PIXMAP_BIT;
-	    modes->stencilBits = stencil_bits;
-	    modes->depthBits = depth_bits;
-
-	    modes->visualType = visType;
-	    modes->renderType = GLX_RGBA_BIT;
-	    modes->rgbMode = GL_TRUE;
-
-	    if ( db_modes[i] == GLX_NONE ) {
-
-		modes->doubleBufferMode = GL_FALSE;
-	    }
-	    else {
-		modes->doubleBufferMode = GL_TRUE;
-		modes->swapMethod = db_modes[i];
-	    }
-
-	    modes = modes->next;
-	}
-    }
-    
-    return modes;
-}
-
-
-static __GLcontextModes *
-mach64FillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
-{
-   __GLcontextModes * modes;
+    __DRIconfig **configs;
     __GLcontextModes * m;
-    unsigned num_modes;
+    GLenum fb_format;
+    GLenum fb_type;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     unsigned i;
@@ -155,49 +91,51 @@ mach64FillInModes( unsigned pixel_bits, unsigned depth_bits,
 	GLX_NONE, GLX_SWAP_UNDEFINED_OML /*, GLX_SWAP_COPY_OML */
     };
 
-    int depth_buffer_modes[2][2];
-
+    uint8_t depth_bits_array[2];
+    uint8_t stencil_bits_array[2];
 
-    depth_buffer_modes[0][0] = depth_bits;
-    depth_buffer_modes[1][0] = depth_bits;
+    depth_bits_array[0] = depth_bits;
+    depth_bits_array[1] = depth_bits;
     
     /* Just like with the accumulation buffer, always provide some modes
      * with a stencil buffer.  It will be a sw fallback, but some apps won't
      * care about that.
      */
-    depth_buffer_modes[0][1] = 0;
-    depth_buffer_modes[1][1] = (stencil_bits == 0) ? 8 : stencil_bits;
+    stencil_bits_array[0] = 0;
+    stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    for ( i = 0 ; i < depth_buffer_factor ; i++ ) {
-	m = fill_in_modes( m, pixel_bits, 
-			   depth_buffer_modes[i][0], depth_buffer_modes[i][1],
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR );
+    if (pixel_bits == 16) {
+       fb_format = GL_RGB;
+       fb_type = GL_UNSIGNED_SHORT_5_6_5;
+    }
+    else {
+       fb_format = GL_BGRA;
+       fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    for ( i = 0 ; i < depth_buffer_factor ; i++ ) {
-	m = fill_in_modes( m, pixel_bits, 
-			   depth_buffer_modes[i][0], depth_buffer_modes[i][1],
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR );
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor, back_buffer_modes,
+			       back_buffer_factor);
+    if (configs == NULL) {
+       fprintf(stderr, "[%s:%u] Error creating FBConfig!\n",
+	       __func__, __LINE__);
+       return NULL;
     }
 
     /* Mark the visual as slow if there are "fake" stencil bits.
      */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-       if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ){
-	    m->visualRating = GLX_SLOW_CONFIG;
-	}
+    for (i = 0; configs[i]; i++) {
+       m = &configs[i]->modes;
+       if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+	  m->visualRating = GLX_SLOW_CONFIG;
+       }
     }
 
-    return modes;
+    return (const __DRIconfig **) configs;
 }
 
 
@@ -208,9 +146,7 @@ mach64CreateScreen( __DRIscreenPrivate *sPriv )
 {
    mach64ScreenPtr mach64Screen;
    ATIDRIPtr serverInfo = (ATIDRIPtr)sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
+   int i;
 
    if (sPriv->devPrivSize != sizeof(ATIDRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(ATIDRIRec) does not match passed size from device driver\n");
@@ -319,15 +255,14 @@ mach64CreateScreen( __DRIscreenPrivate *sPriv )
 
    mach64Screen->driScreen = sPriv;
 
-   if ( glx_enable_extension != NULL ) {
-      if ( mach64Screen->irq != 0 ) {
-	 (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-	 (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-	 (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      }
-
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
+   i = 0;
+   mach64Screen->extensions[i++] = &driFrameTrackingExtension.base;
+   if ( mach64Screen->irq != 0 ) {
+      mach64Screen->extensions[i++] = &driSwapControlExtension.base;
+      mach64Screen->extensions[i++] = &driMediaStreamCounterExtension.base;
    }
+   mach64Screen->extensions[i++] = NULL;
+   sPriv->extensions = mach64Screen->extensions;
 
    return mach64Screen;
 }
@@ -475,9 +410,48 @@ mach64InitDriver( __DRIscreenPrivate *driScreen )
    return GL_TRUE;
 }
 
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
+ */
+static const __DRIconfig **
+mach64InitScreen(__DRIscreenPrivate *psp)
+{
+   static const __DRIversion ddx_expected = { 6, 4, 0 };
+   static const __DRIversion dri_expected = { 4, 0, 0 };
+   static const __DRIversion drm_expected = { 2, 0, 0 };
+   ATIDRIPtr dri_priv = (ATIDRIPtr) psp->pDevPriv;
 
-static struct __DriverAPIRec mach64API = {
-   .InitDriver      = mach64InitDriver,
+   if ( ! driCheckDriDdxDrmVersions2( "Mach64",
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) ) {
+      return NULL;
+   }
+   
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+   if (!mach64InitDriver(psp))
+      return NULL;
+
+   return  mach64FillInModes( psp, dri_priv->cpp * 8, 16, 0, 1);
+}
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = mach64InitScreen,
    .DestroyScreen   = mach64DestroyScreen,
    .CreateContext   = mach64CreateContext,
    .DestroyContext  = mach64DestroyContext,
@@ -487,71 +461,9 @@ static struct __DriverAPIRec mach64API = {
    .MakeCurrent     = mach64MakeCurrent,
    .UnbindContext   = mach64UnbindContext,
    .GetSwapInfo     = NULL,
-   .GetMSC          = driGetMSC32,
+   .GetDrawableMSC  = driDrawableGetMSC32,
    .WaitForMSC      = driWaitForMSC32,
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
 
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 6, 4, 0 };
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 2, 0, 0 };
-
-   dri_interface = interface;
-
-   if ( ! driCheckDriDdxDrmVersions2( "Mach64",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &mach64API);
-   if ( psp != NULL ) {
-      ATIDRIPtr dri_priv = (ATIDRIPtr) psp->pDevPriv;
-      *driver_modes = mach64FillInModes( dri_priv->cpp * 8,
-					 16,
-					 0,
-					 1);
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
-
-   return (void *) psp;
-}
diff --git a/src/mesa/drivers/dri/mach64/mach64_screen.h b/src/mesa/drivers/dri/mach64/mach64_screen.h
index 7bf7dc474d1..be5e29a3e58 100644
--- a/src/mesa/drivers/dri/mach64/mach64_screen.h
+++ b/src/mesa/drivers/dri/mach64/mach64_screen.h
@@ -73,6 +73,8 @@ typedef struct {
    __DRIscreenPrivate *driScreen;
 
    driOptionCache optionCache;
+
+   const __DRIextension *extensions[4];
 } mach64ScreenRec, *mach64ScreenPtr;
 
 #endif /* __MACH64_SCREEN_H__ */
diff --git a/src/mesa/drivers/dri/mach64/mach64_span.c b/src/mesa/drivers/dri/mach64/mach64_span.c
index 5c2403f587f..91d46ce32eb 100644
--- a/src/mesa/drivers/dri/mach64/mach64_span.c
+++ b/src/mesa/drivers/dri/mach64/mach64_span.c
@@ -117,6 +117,8 @@
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)(buf + ((_x) + (_y) * drb->pitch) * 2) = d;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_state.c b/src/mesa/drivers/dri/mach64/mach64_state.c
index 9ac51ee5b1d..3a023187ce7 100644
--- a/src/mesa/drivers/dri/mach64/mach64_state.c
+++ b/src/mesa/drivers/dri/mach64/mach64_state.c
@@ -36,9 +36,9 @@
 #include "mach64_vb.h"
 #include "mach64_tex.h"
 
-#include "context.h"
-#include "enums.h"
-#include "colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -726,24 +726,26 @@ static void mach64DDDrawBuffer( GLcontext *ctx, GLenum mode )
 
    FLUSH_BATCH( mmesa );
 
-   /*
-    * _DrawDestMask is easier to cope with than <mode>.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
+      FALLBACK( mmesa, MACH64_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       FALLBACK( mmesa, MACH64_FALLBACK_DRAW_BUFFER, GL_FALSE );
       mach64SetCliprects( ctx, GL_FRONT_LEFT );
       if (MACH64_DEBUG & DEBUG_VERBOSE_MSG)
 	 fprintf(stderr,"%s: BUFFER_BIT_FRONT_LEFT\n", __FUNCTION__);
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       FALLBACK( mmesa, MACH64_FALLBACK_DRAW_BUFFER, GL_FALSE );
       mach64SetCliprects( ctx, GL_BACK_LEFT );
       if (MACH64_DEBUG & DEBUG_VERBOSE_MSG)
 	 fprintf(stderr,"%s: BUFFER_BIT_BACK_LEFT\n", __FUNCTION__);
       break;
    default:
-      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( mmesa, MACH64_FALLBACK_DRAW_BUFFER, GL_TRUE );
       if (MACH64_DEBUG & DEBUG_VERBOSE_MSG)
 	 fprintf(stderr,"%s: fallback (mode=%d)\n", __FUNCTION__, mode);
diff --git a/src/mesa/drivers/dri/mach64/mach64_tex.c b/src/mesa/drivers/dri/mach64/mach64_tex.c
index c42588e064a..1f9d3c57ebe 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tex.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tex.c
@@ -36,15 +36,15 @@
 #include "mach64_tris.h"
 #include "mach64_tex.h"
 
-#include "context.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texobj.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "main/imports.h"
 
 
 static void mach64SetTexWrap( mach64TexObjPtr t,
diff --git a/src/mesa/drivers/dri/mach64/mach64_tex.h b/src/mesa/drivers/dri/mach64/mach64_tex.h
index e67661b9706..8e0b23ed15b 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tex.h
+++ b/src/mesa/drivers/dri/mach64/mach64_tex.h
@@ -72,9 +72,9 @@ extern void mach64InitTextureFuncs( struct dd_function_table *functions );
 #define MACH64PACKCOLOR4444(r, g, b, a)					\
    ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
 
-static __inline__ GLuint mach64PackColor( GLuint cpp,
-					  GLubyte r, GLubyte g,
-					  GLubyte b, GLubyte a )
+static INLINE GLuint mach64PackColor( GLuint cpp,
+                                      GLubyte r, GLubyte g,
+                                      GLubyte b, GLubyte a )
 {
    switch ( cpp ) {
    case 2:
diff --git a/src/mesa/drivers/dri/mach64/mach64_texmem.c b/src/mesa/drivers/dri/mach64/mach64_texmem.c
index d65b2cda6ad..734e547952e 100644
--- a/src/mesa/drivers/dri/mach64/mach64_texmem.c
+++ b/src/mesa/drivers/dri/mach64/mach64_texmem.c
@@ -38,11 +38,11 @@
 #include "mach64_tris.h"
 #include "mach64_tex.h"
 
-#include "context.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "texformat.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
+#include "main/imports.h"
 
 
 /* Destroy hardware state associated with texture `t'.
diff --git a/src/mesa/drivers/dri/mach64/mach64_texstate.c b/src/mesa/drivers/dri/mach64/mach64_texstate.c
index 80c84d67746..fd2369dd882 100644
--- a/src/mesa/drivers/dri/mach64/mach64_texstate.c
+++ b/src/mesa/drivers/dri/mach64/mach64_texstate.c
@@ -29,11 +29,11 @@
  *	José Fonseca <j_r_fonseca@yahoo.co.uk>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
 
 #include "mach64_context.h"
 #include "mach64_ioctl.h"
diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.c b/src/mesa/drivers/dri/mach64/mach64_tris.c
index e4df01106d3..f2e8e2e3ae8 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tris.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tris.c
@@ -29,10 +29,10 @@
  *	José Fonseca <j_r_fonseca@yahoo.co.uk>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -113,7 +113,7 @@ do {									\
 #define COPY_VERTEX_OOA( vb, vertsize, v, n )	DO_COPY_VERTEX( vb, vertsize, v, n, 1 )
 
 
-static __inline void mach64_draw_quad( mach64ContextPtr mmesa,
+static INLINE void mach64_draw_quad( mach64ContextPtr mmesa,
 				       mach64VertexPtr v0,
 				       mach64VertexPtr v1,
 				       mach64VertexPtr v2,
@@ -419,7 +419,7 @@ static __inline void mach64_draw_quad( mach64ContextPtr mmesa,
 #endif
 }
 
-static __inline void mach64_draw_triangle( mach64ContextPtr mmesa,
+static INLINE void mach64_draw_triangle( mach64ContextPtr mmesa,
 					   mach64VertexPtr v0,
 					   mach64VertexPtr v1,
 					   mach64VertexPtr v2 )
@@ -666,7 +666,7 @@ static __inline void mach64_draw_triangle( mach64ContextPtr mmesa,
 #endif
 }
 
-static __inline void mach64_draw_line( mach64ContextPtr mmesa,
+static INLINE void mach64_draw_line( mach64ContextPtr mmesa,
 				     mach64VertexPtr v0,
 				     mach64VertexPtr v1 )
 {
@@ -955,7 +955,7 @@ static __inline void mach64_draw_line( mach64ContextPtr mmesa,
 #endif
 }
 
-static __inline void mach64_draw_point( mach64ContextPtr mmesa,
+static INLINE void mach64_draw_point( mach64ContextPtr mmesa,
 				      mach64VertexPtr v0 )
 {
 #if MACH64_NATIVE_VTXFMT
diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.h b/src/mesa/drivers/dri/mach64/mach64_tris.h
index 4780765a18b..042df42f5bd 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tris.h
+++ b/src/mesa/drivers/dri/mach64/mach64_tris.h
@@ -31,7 +31,7 @@
 #ifndef __MACH64_TRIS_H__
 #define __MACH64_TRIS_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void mach64InitTriFuncs( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_vb.c b/src/mesa/drivers/dri/mach64/mach64_vb.c
index 8aab72a3f3f..e58812e9028 100644
--- a/src/mesa/drivers/dri/mach64/mach64_vb.c
+++ b/src/mesa/drivers/dri/mach64/mach64_vb.c
@@ -29,11 +29,11 @@
  *	José Fonseca <j_r_fonseca@yahoo.co.uk>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
diff --git a/src/mesa/drivers/dri/mach64/mach64_vb.h b/src/mesa/drivers/dri/mach64/mach64_vb.h
index 0d923abce0d..e0b366916b1 100644
--- a/src/mesa/drivers/dri/mach64/mach64_vb.h
+++ b/src/mesa/drivers/dri/mach64/mach64_vb.h
@@ -32,7 +32,7 @@
 #ifndef __MACH64_VB_H__
 #define __MACH64_VB_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "mach64_context.h"
 
diff --git a/src/mesa/drivers/dri/mga/mga_texcombine.c b/src/mesa/drivers/dri/mga/mga_texcombine.c
index bbfa29be5fe..24083d9651b 100644
--- a/src/mesa/drivers/dri/mga/mga_texcombine.c
+++ b/src/mesa/drivers/dri/mga/mga_texcombine.c
@@ -23,7 +23,7 @@
  *    Ville Syrjala <syrjala@sci.fi>
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 #include "mgacontext.h"
 #include "mgatex.h"
diff --git a/src/mesa/drivers/dri/mga/mga_texstate.c b/src/mesa/drivers/dri/mga/mga_texstate.c
index c14ddc95c9a..d4c5b6fd97b 100644
--- a/src/mesa/drivers/dri/mga/mga_texstate.c
+++ b/src/mesa/drivers/dri/mga/mga_texstate.c
@@ -28,20 +28,20 @@
  */
 
 #include <stdlib.h>
-#include "mm.h"
+#include "main/mm.h"
 #include "mgacontext.h"
 #include "mgatex.h"
 #include "mgaregs.h"
 #include "mgatris.h"
 #include "mgaioctl.h"
 
-#include "context.h"
-#include "enums.h"
-#include "macros.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/macros.h"
+#include "main/imports.h"
 
-#include "simple_list.h"
-#include "texformat.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
 
 #define MGA_USE_TABLE_FOR_FORMAT
 #ifdef MGA_USE_TABLE_FOR_FORMAT
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.c b/src/mesa/drivers/dri/mga/mga_xmesa.c
index d0cf06fea5a..86da3a2cacf 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.c
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.c
@@ -34,12 +34,12 @@
 #include "drm.h"
 #include "mga_drm.h"
 #include "mga_xmesa.h"
-#include "context.h"
-#include "matrix.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/context.h"
+#include "main/matrix.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -64,7 +64,7 @@
 #include "utils.h"
 #include "vblank.h"
 
-#include "extensions.h"
+#include "main/extensions.h"
 #include "drirenderbuffer.h"
 
 #include "GL/internal/dri_interface.h"
@@ -74,11 +74,13 @@
 #define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
 #define need_GL_EXT_fog_coord
+#define need_GL_EXT_gpu_program_parameters
 #define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
 #if 0
 #define need_GL_EXT_paletted_texture
 #endif
+#define need_GL_APPLE_vertex_array_object
 #define need_GL_NV_vertex_program
 #include "extension_helper.h"
 
@@ -109,19 +111,18 @@ static const GLuint __driNConfigOptions = 6;
 int MGA_DEBUG = 0;
 #endif
 
-static int getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo );
-
-static __GLcontextModes *
-mgaFillInModes( unsigned pixel_bits, unsigned depth_bits,
+static const __DRIconfig **
+mgaFillInModes( __DRIscreenPrivate *psp,
+		unsigned pixel_bits, unsigned depth_bits,
 		unsigned stencil_bits, GLboolean have_back_buffer )
 {
-    __GLcontextModes * modes;
+    __DRIconfig **configs;
     __GLcontextModes * m;
-    unsigned num_modes;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     GLenum fb_format;
     GLenum fb_type;
+    int i;
 
     /* GLX_SWAP_COPY_OML is only supported because the MGA driver doesn't
      * support pageflipping at all.
@@ -149,8 +150,6 @@ mgaFillInModes( unsigned pixel_bits, unsigned depth_bits,
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -160,46 +159,41 @@ mgaFillInModes( unsigned pixel_bits, unsigned depth_bits,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
-
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR ) ) {
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor,
+			       back_buffer_modes, back_buffer_factor);
+    if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
 	return NULL;
     }
 
-    /* Mark the visual as slow if there are "fake" stencil bits.
-     */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-	if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	    m->visualRating = GLX_SLOW_CONFIG;
-	}
-    }
+   /* Mark the visual as slow if there are "fake" stencil bits.
+    */
+   for (i = 0; configs[i]; i++) {
+      m = &configs[i]->modes;
+      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+         m->visualRating = GLX_SLOW_CONFIG;
+      }
+   }
 
-    return modes;
+   return (const __DRIconfig **) configs;
 }
 
+const __DRIextension *mgaScreenExtensions[] = {
+    &driReadDrawableExtension,
+    &driSwapControlExtension.base,
+    &driFrameTrackingExtension.base,
+    &driMediaStreamCounterExtension.base,
+    NULL
+};
 
 static GLboolean
 mgaInitDriver(__DRIscreenPrivate *sPriv)
 {
    mgaScreenPrivate *mgaScreen;
    MGADRIPtr         serverInfo = (MGADRIPtr)sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-       (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
 
    if (sPriv->devPrivSize != sizeof(MGADRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(MGADRIRec) does not match passed size from device driver\n");
@@ -216,7 +210,7 @@ mgaInitDriver(__DRIscreenPrivate *sPriv)
    mgaScreen->sPriv = sPriv;
    sPriv->private = (void *)mgaScreen;
 
-   if (sPriv->drmMinor >= 1) {
+   if (sPriv->drm_version.minor >= 1) {
       int ret;
       drm_mga_getparam_t gp;
 
@@ -234,13 +228,7 @@ mgaInitDriver(__DRIscreenPrivate *sPriv)
       }
    }
 
-   if ( glx_enable_extension != NULL ) {
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
-      (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
-      (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-   }
+   sPriv->extensions = mgaScreenExtensions;
 
    if (serverInfo->chipset != MGA_CARD_TYPE_G200 &&
        serverInfo->chipset != MGA_CARD_TYPE_G400) {
@@ -273,7 +261,7 @@ mgaInitDriver(__DRIscreenPrivate *sPriv)
     * there is a new, in-kernel mechanism for handling the wait.
     */
 
-   if (mgaScreen->sPriv->drmMinor < 2) {
+   if (mgaScreen->sPriv->drm_version.minor < 2) {
       mgaScreen->mmio.handle = serverInfo->registers.handle;
       mgaScreen->mmio.size = serverInfo->registers.size;
       if ( drmMap( sPriv->fd,
@@ -411,13 +399,15 @@ static const struct dri_extension card_extensions[] =
 #endif
    { "GL_EXT_secondary_color",        GL_EXT_secondary_color_functions },
    { "GL_EXT_stencil_wrap",           NULL },
+   { "GL_APPLE_vertex_array_object",  GL_APPLE_vertex_array_object_functions },
    { "GL_MESA_ycbcr_texture",         NULL },
    { "GL_SGIS_generate_mipmap",       NULL },
    { NULL,                            NULL }
 };
 
-static const struct dri_extension ARB_vp_extension[] = {
+static const struct dri_extension ARB_vp_extensions[] = {
    { "GL_ARB_vertex_program",         GL_ARB_vertex_program_functions },
+   { "GL_EXT_gpu_program_parameters", GL_EXT_gpu_program_parameters_functions },
    { NULL,                            NULL }
 };
 
@@ -622,7 +612,7 @@ mgaCreateContext( const __GLcontextModes *mesaVis,
    }
 
    if ( driQueryOptionb( &mmesa->optionCache, "arb_vertex_program" ) ) {
-      driInitSingleExtension( ctx, ARB_vp_extension );
+      driInitExtensions(ctx, ARB_vp_extensions, GL_FALSE);
    }
    
    if ( driQueryOptionb( &mmesa->optionCache, "nv_vertex_program" ) ) {
@@ -646,10 +636,7 @@ mgaCreateContext( const __GLcontextModes *mesaVis,
 				    debug_control );
 #endif
 
-   mmesa->vblank_flags = (mmesa->mgaScreen->irq == 0)
-       ? VBLANK_FLAG_NO_IRQ : driGetDefaultVBlankFlags(&mmesa->optionCache);
-
-   (*dri_interface->getUST)( & mmesa->swap_ust );
+   (*sPriv->systemTime->getUST)( & mmesa->swap_ust );
 
    if (driQueryOptionb(&mmesa->optionCache, "no_rast")) {
       fprintf(stderr, "disabling 3D acceleration\n");
@@ -878,8 +865,14 @@ mgaMakeCurrent(__DRIcontextPrivate *driContextPriv,
       mgaContextPtr mmesa = (mgaContextPtr) driContextPriv->driverPrivate;
 
       if (mmesa->driDrawable != driDrawPriv) {
-	 driDrawableInitVBlank( driDrawPriv, mmesa->vblank_flags,
-				&mmesa->vbl_seq );
+	 if (driDrawPriv->swap_interval == (unsigned)-1) {
+	    driDrawPriv->vblFlags = (mmesa->mgaScreen->irq == 0)
+	       ? VBLANK_FLAG_NO_IRQ
+	       : driGetDefaultVBlankFlags(&mmesa->optionCache);
+
+	    driDrawableInitVBlank( driDrawPriv );
+	 }
+
 	 mmesa->driDrawable = driDrawPriv;
 	 mmesa->dirty = ~0; 
 	 mmesa->dirty_cliprects = (MGA_FRONT|MGA_BACK); 
@@ -933,88 +926,51 @@ void mgaGetLock( mgaContextPtr mmesa, GLuint flags )
 }
 
 
-static const struct __DriverAPIRec mgaAPI = {
-   .InitDriver      = mgaInitDriver,
-   .DestroyScreen   = mgaDestroyScreen,
-   .CreateContext   = mgaCreateContext,
-   .DestroyContext  = mgaDestroyContext,
-   .CreateBuffer    = mgaCreateBuffer,
-   .DestroyBuffer   = mgaDestroyBuffer,
-   .SwapBuffers     = mgaSwapBuffers,
-   .MakeCurrent     = mgaMakeCurrent,
-   .UnbindContext   = mgaUnbindContext,
-   .GetSwapInfo     = getSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL
-};
-
-
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+static const __DRIconfig **mgaInitScreen(__DRIscreen *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 1, 2, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 3, 0, 0 };
-
-   dri_interface = interface;
+   MGADRIPtr dri_priv = (MGADRIPtr) psp->pDevPriv;
 
    if ( ! driCheckDriDdxDrmVersions2( "MGA",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
 
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &mgaAPI);
-   if ( psp != NULL ) {
-      MGADRIPtr dri_priv = (MGADRIPtr) psp->pDevPriv;
-      *driver_modes = mgaFillInModes( dri_priv->cpp * 8,
-				      (dri_priv->cpp == 2) ? 16 : 24,
-				      (dri_priv->cpp == 2) ? 0  : 8,
-				      (dri_priv->backOffset != dri_priv->depthOffset) );
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-      driInitExtensions( NULL, g400_extensions, GL_FALSE );
-      driInitSingleExtension( NULL, ARB_vp_extension );
-      driInitExtensions( NULL, NV_vp_extensions, GL_FALSE );
 
-   }
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
 
-   return (void *) psp;
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+   driInitExtensions( NULL, g400_extensions, GL_FALSE );
+   driInitExtensions(NULL, ARB_vp_extensions, GL_FALSE);
+   driInitExtensions( NULL, NV_vp_extensions, GL_FALSE );
+
+   if (!mgaInitDriver(psp))
+       return NULL;
+
+   return mgaFillInModes( psp,
+			  dri_priv->cpp * 8,
+			  (dri_priv->cpp == 2) ? 16 : 24,
+			  (dri_priv->cpp == 2) ? 0  : 8,
+			  (dri_priv->backOffset != dri_priv->depthOffset) );
 }
 
 
@@ -1043,3 +999,20 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
 
    return 0;
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = mgaInitScreen,
+   .DestroyScreen   = mgaDestroyScreen,
+   .CreateContext   = mgaCreateContext,
+   .DestroyContext  = mgaDestroyContext,
+   .CreateBuffer    = mgaCreateBuffer,
+   .DestroyBuffer   = mgaDestroyBuffer,
+   .SwapBuffers     = mgaSwapBuffers,
+   .MakeCurrent     = mgaMakeCurrent,
+   .UnbindContext   = mgaUnbindContext,
+   .GetSwapInfo     = getSwapInfo,
+   .GetDrawableMSC  = driDrawableGetMSC32,
+   .WaitForMSC      = driWaitForMSC32,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL
+};
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.h b/src/mesa/drivers/dri/mga/mga_xmesa.h
index 7de7bb06e87..07c22bd5966 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.h
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.h
@@ -31,7 +31,7 @@
 #include <sys/time.h>
 #include "dri_util.h"
 #include "mga_drm.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "mgaregs.h"
 #include "xmlconfig.h"
 
diff --git a/src/mesa/drivers/dri/mga/mgacontext.h b/src/mesa/drivers/dri/mga/mgacontext.h
index 6aa92355b8a..30640a29b35 100644
--- a/src/mesa/drivers/dri/mga/mgacontext.h
+++ b/src/mesa/drivers/dri/mga/mgacontext.h
@@ -32,12 +32,12 @@
 #include "drm.h"
 #include "mga_drm.h"
 #include "dri_util.h"
-#include "mtypes.h"
 #include "xf86drm.h"
-#include "mm.h"
-#include "colormac.h"
+#include "main/mtypes.h"
+#include "main/mm.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 #include "texmem.h"
-#include "macros.h"
 #include "xmlconfig.h"
 
 #define MGA_SET_FIELD(reg,mask,val)  reg = ((reg) & (mask)) | ((val) & ~(mask))
@@ -257,11 +257,6 @@ struct mga_context_t {
    drmBufPtr  vertex_dma_buffer;
    drmBufPtr  iload_buffer;
 
-   /* VBI
-    */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
    int64_t swap_ust;
    int64_t swap_missed_ust;
 
@@ -333,9 +328,9 @@ extern int MGA_DEBUG;
 #define DEBUG_VERBOSE_TEXTURE   0x08
 #define DEBUG_VERBOSE_FALLBACK	0x10
 
-static __inline__ GLuint mgaPackColor(GLuint cpp,
-				      GLubyte r, GLubyte g,
-				      GLubyte b, GLubyte a)
+static INLINE GLuint mgaPackColor(GLuint cpp,
+                                  GLubyte r, GLubyte g,
+                                  GLubyte b, GLubyte a)
 {
    switch (cpp) {
    case 2:
diff --git a/src/mesa/drivers/dri/mga/mgadd.c b/src/mesa/drivers/dri/mga/mgadd.c
index 04336b5ac75..3b1ea22b609 100644
--- a/src/mesa/drivers/dri/mga/mgadd.c
+++ b/src/mesa/drivers/dri/mga/mgadd.c
@@ -26,10 +26,10 @@
  */
 
 
-#include "mtypes.h"
-#include "framebuffer.h"
+#include "main/mtypes.h"
+#include "main/framebuffer.h"
+#include "main/mm.h"
 
-#include "mm.h"
 #include "mgacontext.h"
 #include "mgadd.h"
 #include "mgastate.h"
@@ -40,7 +40,7 @@
 #include "mga_xmesa.h"
 #include "utils.h"
 
-#define DRIVER_DATE	"20061030"
+#define DRIVER_DATE	"20071017"
 
 
 /***************************************
diff --git a/src/mesa/drivers/dri/mga/mgadd.h b/src/mesa/drivers/dri/mga/mgadd.h
index 6830ca67ad2..f92591df459 100644
--- a/src/mesa/drivers/dri/mga/mgadd.h
+++ b/src/mesa/drivers/dri/mga/mgadd.h
@@ -28,7 +28,7 @@
 #ifndef MGADD_INC
 #define MGADD_INC
 
-#include "context.h"
+#include "main/context.h"
 
 extern void mgaInitDriverFuncs( struct dd_function_table *functions );
 
diff --git a/src/mesa/drivers/dri/mga/mgaioctl.c b/src/mesa/drivers/dri/mga/mgaioctl.c
index f8587fc541e..4438bad9209 100644
--- a/src/mesa/drivers/dri/mga/mgaioctl.c
+++ b/src/mesa/drivers/dri/mga/mgaioctl.c
@@ -31,12 +31,12 @@
  */
 
 #include <errno.h>
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/dd.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "drm.h"
 #include "mga_drm.h"
 #include "mgacontext.h"
@@ -55,7 +55,7 @@ mgaSetFence( mgaContextPtr mmesa, uint32_t * fence )
 {
     int ret = ENOSYS;
 
-    if ( mmesa->driScreen->drmMinor >= 2 ) {
+    if ( mmesa->driScreen->drm_version.minor >= 2 ) {
 	ret = drmCommandWriteRead( mmesa->driScreen->fd, DRM_MGA_SET_FENCE,
 				   fence, sizeof( uint32_t ));
 	if (ret) {
@@ -73,7 +73,7 @@ mgaWaitFence( mgaContextPtr mmesa, uint32_t fence, uint32_t * curr_fence )
 {
     int ret = ENOSYS;
 
-    if ( mmesa->driScreen->drmMinor >= 2 ) {
+    if ( mmesa->driScreen->drm_version.minor >= 2 ) {
 	uint32_t temp = fence;
 	
 	ret = drmCommandWriteRead( mmesa->driScreen->fd,
@@ -409,7 +409,7 @@ static void mgaWaitForFrameCompletion( mgaContextPtr mmesa )
 /*
  * Copy the back buffer to the front buffer.
  */
-void mgaCopyBuffer( const __DRIdrawablePrivate *dPriv )
+void mgaCopyBuffer( __DRIdrawablePrivate *dPriv )
 {
    mgaContextPtr mmesa;
    drm_clip_rect_t *pbox;
@@ -417,7 +417,7 @@ void mgaCopyBuffer( const __DRIdrawablePrivate *dPriv )
    GLint ret;
    GLint i;
    GLboolean   missed_target;
-
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
@@ -428,11 +428,10 @@ void mgaCopyBuffer( const __DRIdrawablePrivate *dPriv )
    FLUSH_BATCH( mmesa );
 
    mgaWaitForFrameCompletion( mmesa );
-   driWaitForVBlank( dPriv, & mmesa->vbl_seq, mmesa->vblank_flags,
-		     & missed_target );
+   driWaitForVBlank( dPriv, & missed_target );
    if ( missed_target ) {
       mmesa->swap_missed_count++;
-      (void) (*dri_interface->getUST)( & mmesa->swap_missed_ust );
+      (void) (*psp->systemTime->getUST)( & mmesa->swap_missed_ust );
    }
    LOCK_HARDWARE( mmesa );
 
@@ -470,7 +469,7 @@ void mgaCopyBuffer( const __DRIdrawablePrivate *dPriv )
 
    mmesa->dirty |= MGA_UPLOAD_CLIPRECTS;
    mmesa->swap_count++;
-   (void) (*dri_interface->getUST)( & mmesa->swap_ust );
+   (void) (*psp->systemTime->getUST)( & mmesa->swap_ust );
 }
 
 
diff --git a/src/mesa/drivers/dri/mga/mgaioctl.h b/src/mesa/drivers/dri/mga/mgaioctl.h
index 9aa08c51585..dbc823de802 100644
--- a/src/mesa/drivers/dri/mga/mgaioctl.h
+++ b/src/mesa/drivers/dri/mga/mgaioctl.h
@@ -32,7 +32,7 @@
 #include "mgacontext.h"
 #include "mga_xmesa.h"
 
-void mgaCopyBuffer( const __DRIdrawablePrivate *dPriv );
+void mgaCopyBuffer( __DRIdrawablePrivate *dPriv );
 void mgaWaitForVBlank( mgaContextPtr mmesa );
 
 void mgaGetILoadBufferLocked( mgaContextPtr mmesa );
@@ -61,7 +61,7 @@ void mgaInitIoctlFuncs( struct dd_function_table *functions );
 
 extern drmBufPtr mga_get_buffer_ioctl( mgaContextPtr mmesa );
 
-static __inline
+static INLINE
 GLuint *mgaAllocDmaLow( mgaContextPtr mmesa, int bytes )
 {
    GLuint *head;
diff --git a/src/mesa/drivers/dri/mga/mgapixel.c b/src/mesa/drivers/dri/mga/mgapixel.c
index f309aabbc8e..9f90047ba59 100644
--- a/src/mesa/drivers/dri/mga/mgapixel.c
+++ b/src/mesa/drivers/dri/mga/mgapixel.c
@@ -35,8 +35,8 @@
  * \author Gareth Hughes <gareth@valinux.com>
  */
 
-#include "mtypes.h"
-#include "macros.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "mgadd.h"
 #include "mgacontext.h"
 #include "mgaioctl.h"
@@ -44,7 +44,7 @@
 #include "mgastate.h"
 
 #include "swrast/swrast.h"
-#include "imports.h"
+#include "main/imports.h"
 
 #if 0
 #define IS_AGP_MEM( mmesa, p )						  \
diff --git a/src/mesa/drivers/dri/mga/mgapixel.h b/src/mesa/drivers/dri/mga/mgapixel.h
index b52c8670f3a..f5f300db56d 100644
--- a/src/mesa/drivers/dri/mga/mgapixel.h
+++ b/src/mesa/drivers/dri/mga/mgapixel.h
@@ -28,7 +28,7 @@
 #ifndef MGA_PIXELS_H
 #define MGA_PIXELS_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void mgaDDInitPixelFuncs( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/mga/mgarender.c b/src/mesa/drivers/dri/mga/mgarender.c
index c9e42a8040d..517c3b8f82c 100644
--- a/src/mesa/drivers/dri/mga/mgarender.c
+++ b/src/mesa/drivers/dri/mga/mgarender.c
@@ -38,11 +38,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * dma buffers.  Use strip/fan hardware primitives where possible.
  * Simulate missing primitives with indexed vertices.
  */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "tnl/t_context.h"
 
@@ -143,7 +143,7 @@ static GLboolean mga_run_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/mga/mgaspan.c b/src/mesa/drivers/dri/mga/mgaspan.c
index 05dcbb85263..5b6d323ca98 100644
--- a/src/mesa/drivers/dri/mga/mgaspan.c
+++ b/src/mesa/drivers/dri/mga/mgaspan.c
@@ -25,7 +25,7 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "mgadd.h"
 #include "mgacontext.h"
 #include "mgaspan.h"
@@ -107,6 +107,8 @@
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLushort *)(buf + (_x)*2 + (_y)*pitch) = d;
 
@@ -121,6 +123,8 @@
 
 /* 32 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = d;
 
@@ -134,6 +138,8 @@
 
 /* 24/8 bit interleaved depth/stencil functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) {			\
    GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
    tmp &= 0xff;						\
diff --git a/src/mesa/drivers/dri/mga/mgastate.c b/src/mesa/drivers/dri/mga/mgastate.c
index c20a76f29ef..7c830ec0974 100644
--- a/src/mesa/drivers/dri/mga/mgastate.c
+++ b/src/mesa/drivers/dri/mga/mgastate.c
@@ -26,11 +26,11 @@
  */
 
 
-#include "mtypes.h"
-#include "colormac.h"
-#include "dd.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/dd.h"
+#include "main/mm.h"
 
-#include "mm.h"
 #include "mgacontext.h"
 #include "mgadd.h"
 #include "mgastate.h"
@@ -778,8 +778,6 @@ void mgaUpdateRects( mgaContextPtr mmesa, GLuint buffers )
 {
    __DRIdrawablePrivate *const driDrawable = mmesa->driDrawable;
    __DRIdrawablePrivate *const driReadable = mmesa->driReadable;
-   drm_mga_sarea_t *sarea = mmesa->sarea;
-
 
    mmesa->dirty_cliprects = 0;	
 
@@ -790,9 +788,6 @@ void mgaUpdateRects( mgaContextPtr mmesa, GLuint buffers )
 
    mga_set_cliprects(mmesa);
 
-   sarea->req_drawable = driDrawable->draw;
-   sarea->req_draw_buffer = mmesa->draw_buffer;
-
    mgaUpdateClipping( mmesa->glCtx );
    mgaCalcViewport( mmesa->glCtx );
 }
@@ -804,20 +799,22 @@ static void mgaDDDrawBuffer(GLcontext *ctx, GLenum mode )
 
    FLUSH_BATCH( mmesa );
 
-   /*
-    * _DrawDestMask is easier to cope with than <mode>.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
+      FALLBACK( ctx, MGA_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       mmesa->setup.dstorg = mmesa->mgaScreen->frontOffset;
       mmesa->draw_buffer = MGA_FRONT;
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       mmesa->setup.dstorg = mmesa->mgaScreen->backOffset;
       mmesa->draw_buffer = MGA_BACK;
       break;
    default:
-      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( ctx, MGA_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
diff --git a/src/mesa/drivers/dri/mga/mgatex.c b/src/mesa/drivers/dri/mga/mgatex.c
index 31ea5046dff..2392622b902 100644
--- a/src/mesa/drivers/dri/mga/mgatex.c
+++ b/src/mesa/drivers/dri/mga/mgatex.c
@@ -25,24 +25,24 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mm.h"
+#include "main/glheader.h"
+#include "main/mm.h"
 #include "mgacontext.h"
 #include "mgatex.h"
 #include "mgaregs.h"
 #include "mgatris.h"
 #include "mgaioctl.h"
 
-#include "colormac.h"
-#include "context.h"
-#include "enums.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "macros.h"
-#include "texformat.h"
-#include "texstore.h"
-#include "teximage.h"
-#include "texobj.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
 
 #include "swrast/swrast.h"
 
diff --git a/src/mesa/drivers/dri/mga/mgatexmem.c b/src/mesa/drivers/dri/mga/mgatexmem.c
index 559813f5dec..9a2d62b53b7 100644
--- a/src/mesa/drivers/dri/mga/mgatexmem.c
+++ b/src/mesa/drivers/dri/mga/mgatexmem.c
@@ -25,17 +25,17 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "mgacontext.h"
 #include "mgatex.h"
 #include "mgaregs.h"
 #include "mgaioctl.h"
 #include "mga_xmesa.h"
 
-#include "imports.h"
-#include "simple_list.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
 
 /**
  * Destroy any device-dependent state associated with the texture.  This may
diff --git a/src/mesa/drivers/dri/mga/mgatris.c b/src/mesa/drivers/dri/mga/mgatris.c
index 0c8081cfb96..b93a21c3acf 100644
--- a/src/mesa/drivers/dri/mga/mgatris.c
+++ b/src/mesa/drivers/dri/mga/mgatris.c
@@ -25,15 +25,15 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
+#include "main/mm.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 
-#include "mm.h"
 #include "mgacontext.h"
 #include "mgaioctl.h"
 #include "mgatris.h"
@@ -66,7 +66,7 @@ do {						\
 } while (0)
 #endif
 
-static void __inline__ mga_draw_triangle( mgaContextPtr mmesa,
+static void INLINE mga_draw_triangle( mgaContextPtr mmesa,
 					   mgaVertexPtr v0,
 					   mgaVertexPtr v1,
 					   mgaVertexPtr v2 )
@@ -81,7 +81,7 @@ static void __inline__ mga_draw_triangle( mgaContextPtr mmesa,
 }
 
 
-static void __inline__ mga_draw_quad( mgaContextPtr mmesa,
+static void INLINE mga_draw_quad( mgaContextPtr mmesa,
 				       mgaVertexPtr v0,
 				       mgaVertexPtr v1,
 				       mgaVertexPtr v2,
@@ -100,7 +100,7 @@ static void __inline__ mga_draw_quad( mgaContextPtr mmesa,
 }
 
 
-static __inline__ void mga_draw_point( mgaContextPtr mmesa,
+static INLINE void mga_draw_point( mgaContextPtr mmesa,
 					mgaVertexPtr tmp )
 {
    const GLfloat sz = 0.5 * CLAMP(mmesa->glCtx->Point.Size,
@@ -159,7 +159,7 @@ static __inline__ void mga_draw_point( mgaContextPtr mmesa,
 }
 
 
-static __inline__ void mga_draw_line( mgaContextPtr mmesa,
+static INLINE void mga_draw_line( mgaContextPtr mmesa,
 				      mgaVertexPtr v0,
 				      mgaVertexPtr v1 )
 {
diff --git a/src/mesa/drivers/dri/mga/mgatris.h b/src/mesa/drivers/dri/mga/mgatris.h
index a40fef8307d..43612b80a15 100644
--- a/src/mesa/drivers/dri/mga/mgatris.h
+++ b/src/mesa/drivers/dri/mga/mgatris.h
@@ -28,7 +28,7 @@
 #ifndef MGATRIS_INC
 #define MGATRIS_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void mgaDDInitTriFuncs( GLcontext *ctx );
 extern void mgaChooseRenderState( GLcontext *ctx );
diff --git a/src/mesa/drivers/dri/mga/mgavb.c b/src/mesa/drivers/dri/mga/mgavb.c
index 954fd53ae33..1c635b23a65 100644
--- a/src/mesa/drivers/dri/mga/mgavb.c
+++ b/src/mesa/drivers/dri/mga/mgavb.c
@@ -32,11 +32,11 @@
 #include "mgaioctl.h"
 #include "mga_xmesa.h"
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "tnl/t_context.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/mga/mgavb.h b/src/mesa/drivers/dri/mga/mgavb.h
index f6580e0db9f..8d24ab7b5fe 100644
--- a/src/mesa/drivers/dri/mga/mgavb.h
+++ b/src/mesa/drivers/dri/mga/mgavb.h
@@ -28,7 +28,7 @@
 #ifndef MGAVB_INC
 #define MGAVB_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "mgacontext.h"
 #include "swrast/swrast.h"
 
diff --git a/src/mesa/drivers/dri/nouveau/Makefile b/src/mesa/drivers/dri/nouveau/Makefile
deleted file mode 100644
index 20d2de5eefb..00000000000
--- a/src/mesa/drivers/dri/nouveau/Makefile
+++ /dev/null
@@ -1,53 +0,0 @@
-# src/mesa/drivers/dri/nouveau/Makefile
-
-TOP = ../../../../..
-include $(TOP)/configs/current
-
-LIBNAME = nouveau_dri.so
-
-MINIGLX_SOURCES = 
-
-DRIVER_SOURCES = \
-	nouveau_bufferobj.c      \
-	nouveau_buffers.c        \
-	nouveau_card.c           \
-	nouveau_context.c        \
-	nouveau_driver.c         \
-	nouveau_fifo.c           \
-	nouveau_lock.c           \
-	nouveau_object.c         \
-	nouveau_screen.c         \
-	nouveau_span.c           \
-	nouveau_state.c          \
-	nouveau_state_cache.c    \
-	nouveau_shader.c         \
-	nouveau_shader_0.c       \
-	nouveau_shader_1.c       \
-	nouveau_shader_2.c       \
-	nouveau_tex.c            \
-	nouveau_swtcl.c          \
-	nouveau_sync.c           \
-	nouveau_query.c          \
-	nv04_state.c             \
-	nv04_swtcl.c             \
-	nv10_state.c             \
-	nv10_swtcl.c             \
-	nv20_state.c             \
-	nv20_vertprog.c          \
-	nv30_state.c             \
-	nv30_fragprog.c          \
-	nv30_vertprog.c          \
-	nv40_fragprog.c          \
-	nv40_vertprog.c          \
-	nv50_state.c
-
-C_SOURCES = \
-	$(COMMON_SOURCES) \
-	$(DRIVER_SOURCES) 
-
-ASM_SOURCES = 
-
-
-include ../Makefile.template
-
-symlinks:
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
deleted file mode 100644
index fc14060c049..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
+++ /dev/null
@@ -1,616 +0,0 @@
-#include "bufferobj.h"
-#include "enums.h"
-
-#include "nouveau_bufferobj.h"
-#include "nouveau_buffers.h"
-#include "nouveau_context.h"
-#include "nouveau_drm.h"
-#include "nouveau_object.h"
-#include "nouveau_msg.h"
-
-#define NOUVEAU_MEM_FREE(mem) do {      \
-	nouveau_mem_free(ctx, (mem));   \
-	(mem) = NULL;                   \
-} while(0)
-
-#define DEBUG(fmt,args...) do {                \
-	if (NOUVEAU_DEBUG & DEBUG_BUFFEROBJ) { \
-		fprintf(stderr, "%s: "fmt, __func__, ##args);  \
-	}                                      \
-} while(0)
-
-static GLboolean
-nouveau_bo_download_from_screen(GLcontext *ctx,	GLuint offset, GLuint size,
-						struct gl_buffer_object *bo)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_mem *in_mem;
-
-	DEBUG("bo=%p, offset=%d, size=%d\n", bo, offset, size);
-
-	/* If there's a permanent backing store, blit directly into it */
-	if (nbo->cpu_mem) {
-		if (nbo->cpu_mem != nbo->gpu_mem) {
-			DEBUG("..cpu_mem\n");
-			nouveau_memformat_flat_emit(ctx, nbo->cpu_mem,
-						    nbo->gpu_mem,
-						    offset, offset, size);
-		}
-	} else {
-		DEBUG("..sys_mem\n");
-		in_mem = nouveau_mem_alloc(ctx, NOUVEAU_MEM_AGP, size, 0);
-		if (in_mem) {
-			DEBUG("....via GART\n");
-			/* otherwise, try blitting to faster memory and
-			 * copying from there
-			 */
-			nouveau_memformat_flat_emit(ctx, in_mem, nbo->gpu_mem,
-							 0, offset, size);
-			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
-						       NvSubMemFormat);
-			_mesa_memcpy(nbo->cpu_mem_sys + offset,
-					in_mem->map, size);
-			NOUVEAU_MEM_FREE(in_mem);
-		} else {
-			DEBUG("....direct VRAM copy\n");
-			/* worst case, copy directly from vram */
-			_mesa_memcpy(nbo->cpu_mem_sys + offset,
-				     nbo->gpu_mem + offset,
-				     size);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-static GLboolean
-nouveau_bo_upload_to_screen(GLcontext *ctx, GLuint offset, GLuint size,
-					    struct gl_buffer_object *bo)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_mem *out_mem;
-
-	DEBUG("bo=%p, offset=%d, size=%d\n", bo, offset, size);
-
-	if (nbo->cpu_mem) {
-		if (nbo->cpu_mem != nbo->gpu_mem) {
-			DEBUG("..cpu_mem\n");
-			nouveau_memformat_flat_emit(ctx, nbo->gpu_mem,
-						    nbo->cpu_mem,
-						    offset, offset, size);
-		}
-	} else {
-		out_mem = nouveau_mem_alloc(ctx, NOUVEAU_MEM_AGP |
-						 NOUVEAU_MEM_MAPPED,
-						 size, 0);
-		if (out_mem) {
-			DEBUG("....via GART\n");
-			_mesa_memcpy(out_mem->map,
-					nbo->cpu_mem_sys + offset, size);
-			nouveau_memformat_flat_emit(ctx, nbo->gpu_mem, out_mem,
-						    offset, 0, size);
-			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
-						       NvSubMemFormat);
-			NOUVEAU_MEM_FREE(out_mem);
-		} else {
-			DEBUG("....direct VRAM copy\n");
-			_mesa_memcpy(nbo->gpu_mem->map + offset,
-				     nbo->cpu_mem_sys + offset,
-				     size);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-GLboolean
-nouveau_bo_move_in(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-
-	DEBUG("bo=%p\n", bo);
-
-	if (bo->OnCard)
-		return GL_TRUE;
-	assert(nbo->gpu_mem_flags);
-
-	nbo->gpu_mem = nouveau_mem_alloc(ctx, nbo->gpu_mem_flags |
-					      NOUVEAU_MEM_MAPPED,
-					      bo->Size, 0);
-	assert(nbo->gpu_mem);
-
-	if (nbo->cpu_mem_flags) {
-		if ((nbo->cpu_mem_flags|NOUVEAU_MEM_MAPPED) != nbo->gpu_mem->type) {
-			DEBUG("..need cpu_mem buffer\n");
-
-			nbo->cpu_mem = nouveau_mem_alloc(ctx,
-							 nbo->cpu_mem_flags |
-							 NOUVEAU_MEM_MAPPED,
-							 bo->Size, 0);
-
-			if (nbo->cpu_mem) {
-				DEBUG("....alloc ok, kill sys_mem buffer\n");
-				_mesa_memcpy(nbo->cpu_mem->map,
-					     nbo->cpu_mem_sys, bo->Size);
-				FREE(nbo->cpu_mem_sys);
-			}
-		} else {
-			DEBUG("..cpu direct access to GPU buffer\n");
-			nbo->cpu_mem = nbo->gpu_mem;
-		}
-	}
-	nouveau_bo_upload_to_screen(ctx, 0, bo->Size, bo);
-
-	bo->OnCard = GL_TRUE;
-	return GL_TRUE;
-}
-
-GLboolean
-nouveau_bo_move_out(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	GLuint nr_dirty;
-
-	DEBUG("bo=%p\n", bo);
-	if (!bo->OnCard)
-		return GL_TRUE;
-
-	nr_dirty = nouveau_bo_download_dirty(ctx, bo);
-	if (nbo->cpu_mem) {
-		if (nr_dirty && nbo->cpu_mem != nbo->gpu_mem)
-			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
-						       NvSubMemFormat);
-		DEBUG("..destroy cpu_mem buffer\n");
-		nbo->cpu_mem_sys = malloc(bo->Size);
-		assert(nbo->cpu_mem_sys);
-		_mesa_memcpy(nbo->cpu_mem_sys, nbo->cpu_mem->map, bo->Size);
-		if (nbo->cpu_mem == nbo->gpu_mem)
-			nbo->cpu_mem = NULL;
-		else
-			NOUVEAU_MEM_FREE(nbo->cpu_mem);
-	}
-	NOUVEAU_MEM_FREE(nbo->gpu_mem);
-
-	bo->OnCard = GL_FALSE;
-	return GL_TRUE;
-}
-
-static void
-nouveau_bo_choose_storage_method(GLcontext *ctx, GLenum usage,
-						 struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	GLuint gpu_type = 0;
-	GLuint cpu_type = 0;
-
-	switch (usage) {
-	/* Client source, changes often, used by GL many times */
-	case GL_DYNAMIC_DRAW_ARB:
-		gpu_type = NOUVEAU_MEM_AGP | NOUVEAU_MEM_FB_ACCEPTABLE;
-		cpu_type = NOUVEAU_MEM_AGP;
-		break;
-	/* GL source, changes often, client reads many times */
-	case GL_DYNAMIC_READ_ARB:
-	/* Client source, specified once, used by GL many times */
-	case GL_STATIC_DRAW_ARB:
-	/* GL source, specified once, client reads many times */
-	case GL_STATIC_READ_ARB:
-	/* Client source, specified once, used by GL a few times */
-	case GL_STREAM_DRAW_ARB:
-	/* GL source, specified once, client reads a few times */
-	case GL_STREAM_READ_ARB:
-	/* GL source, changes often, used by GL many times*/
-	case GL_DYNAMIC_COPY_ARB:
-	/* GL source, specified once, used by GL many times */
-	case GL_STATIC_COPY_ARB:
-	/* GL source, specified once, used by GL a few times */
-	case GL_STREAM_COPY_ARB:
-		gpu_type = NOUVEAU_MEM_FB;
-		break;
-	default: 
-		assert(0);
-	}
-
-	nbo->gpu_mem_flags = gpu_type;
-	nbo->cpu_mem_flags = cpu_type;
-	nbo->usage	   = usage;
-}
-
-void
-nouveau_bo_init_storage(GLcontext *ctx,	GLuint valid_gpu_access,
-					GLsizeiptrARB size,
-					const GLvoid *data,
-					GLenum usage,
-					struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-
-	DEBUG("bo=%p\n", bo);
-
-	/* Free up previous buffers if we can't reuse them */
-	if (nbo->usage != usage ||
-			(nbo->gpu_mem && (nbo->gpu_mem->size != size))) {
-		if (nbo->cpu_mem_sys)
-			FREE(nbo->cpu_mem_sys);
-		if (nbo->cpu_mem) {
-			if (nbo->cpu_mem != nbo->gpu_mem)
-				NOUVEAU_MEM_FREE(nbo->cpu_mem);
-			else
-				nbo->cpu_mem = NULL;
-		}
-		if (nbo->gpu_mem)
-			NOUVEAU_MEM_FREE(nbo->gpu_mem);
-
-		bo->OnCard = GL_FALSE;
-		nbo->cpu_mem_sys = calloc(1, size);
-	}
-
-	nouveau_bo_choose_storage_method(ctx, usage, bo);
-	/* Force off flags that may not be ok for a given buffer */
-	nbo->gpu_mem_flags &= valid_gpu_access;
-
-	bo->Usage  = usage;
-	bo->Size   = size;
-
-	if (data) {
-		GLvoid *map = nouveau_bo_map(ctx, GL_WRITE_ONLY_ARB, bo);
-		_mesa_memcpy(map, data, size);
-		nouveau_bo_dirty_all(ctx, GL_FALSE, bo);
-		nouveau_bo_unmap(ctx, bo);
-	}
-}
-
-void *
-nouveau_bo_map(GLcontext *ctx, GLenum access, struct gl_buffer_object *bo)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-
-	DEBUG("bo=%p, access=%s\n", bo, _mesa_lookup_enum_by_nr(access));
-
-	if (bo->OnCard && 
-		(access == GL_READ_ONLY_ARB || access == GL_READ_WRITE_ARB)) {
-		GLuint nr_dirty;
-
-		DEBUG("..on card\n");
-		nr_dirty = nouveau_bo_download_dirty(ctx, bo);
-
-		/* nouveau_bo_download_dirty won't wait unless it needs to
-		 * free a temp buffer, which isn't the case if cpu_mem is
-		 * present.
-		 */
-		if (nr_dirty && nbo->cpu_mem && nbo->cpu_mem != nbo->gpu_mem)
-			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
-						       NvSubMemFormat);
-	}
-
-	if (nbo->cpu_mem) {
-		DEBUG("..access via cpu_mem\n");
-		return nbo->cpu_mem->map;
-	} else {
-		DEBUG("..access via cpu_mem_sys\n");
-		return nbo->cpu_mem_sys;
-	}
-}
-
-void
-nouveau_bo_unmap(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	DEBUG("unmap bo=%p\n", bo);
-}
-
-uint32_t
-nouveau_bo_gpu_ref(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-
-	assert(nbo->mapped == GL_FALSE);
-
-	DEBUG("gpu_ref\n");
-	
-	if (!bo->OnCard) {
-		nouveau_bo_move_in(ctx, bo);
-		bo->OnCard = GL_TRUE;
-	}
-	nouveau_bo_upload_dirty(ctx, bo);
-
-	return nouveau_mem_gpu_offset_get(ctx, nbo->gpu_mem);
-}
-
-void
-nouveau_bo_dirty_linear(GLcontext *ctx, GLboolean on_card,
-			uint32_t offset, uint32_t size,
-			struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_bufferobj_dirty *dirty;
-	uint32_t start = offset;
-	uint32_t end = offset + size;
-	int i;
-
-	if (nbo->cpu_mem == nbo->gpu_mem)
-		return;
-
-	dirty = on_card ? &nbo->gpu_dirty : &nbo->cpu_dirty;
-
-	DEBUG("on_card=%d, offset=%d, size=%d, bo=%p\n",
-			on_card, offset, size, bo);
-
-	for (i=0; i<dirty->nr_dirty; i++) {
-		nouveau_bufferobj_region *r = &dirty->dirty[i];
-
-		/* already dirty */
-		if (start >= r->start && end <= r->end) {
-			DEBUG("..already dirty\n");
-			return;
-		}
-
-		/* add to the end of a region */
-		if (start >= r->start && start <= r->end) {
-			if (end > r->end) {
-				DEBUG("..extend end of region\n");
-				r->end = end;
-				return;
-			}
-		}
-
-		/* add to the start of a region */
-		if (start < r->start && end >= r->end) {
-			DEBUG("..extend start of region\n");
-			r->start = start;
-			/* .. and to the end */
-			if (end > r->end) {
-				DEBUG("....and end\n");
-				r->end = end;
-			}
-			return;
-		}
-	}
-
-	/* new region */
-	DEBUG("..new dirty\n");
-	dirty->nr_dirty++;
-	dirty->dirty = realloc(dirty->dirty,
-			       sizeof(nouveau_bufferobj_region) *
-			       dirty->nr_dirty);
-	dirty->dirty[dirty->nr_dirty - 1].start = start;
-	dirty->dirty[dirty->nr_dirty - 1].end   = end;
-}
-
-void
-nouveau_bo_dirty_all(GLcontext *ctx, GLboolean on_card,
-		     struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_bufferobj_dirty *dirty;
-
-	dirty = on_card ? &nbo->gpu_dirty : &nbo->cpu_dirty;
-	
-	DEBUG("dirty all\n");
-	if (dirty->nr_dirty) {
-		FREE(dirty->dirty);
-		dirty->dirty    = NULL;
-		dirty->nr_dirty = 0;
-	}
-
-	nouveau_bo_dirty_linear(ctx, on_card, 0, bo->Size, bo);
-}
-
-GLuint
-nouveau_bo_upload_dirty(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_bufferobj_dirty *dirty = &nbo->cpu_dirty;
-	GLuint nr_dirty;
-	int i;
-
-	nr_dirty = dirty->nr_dirty;
-	if (!nr_dirty) {
-		DEBUG("clean\n");
-		return nr_dirty;
-	}
-
-	for (i=0; i<nr_dirty; i++) {
-		nouveau_bufferobj_region *r = &dirty->dirty[i];
-
-		DEBUG("dirty %d: o=0x%08x, s=0x%08x\n",
-				i, r->start, r->end - r->start);
-		nouveau_bo_upload_to_screen(ctx,
-					    r->start, r->end - r->start, bo);
-	}
-
-	FREE(dirty->dirty);
-	dirty->dirty    = NULL;
-	dirty->nr_dirty = 0;
-
-	return nr_dirty;
-}
-
-GLuint
-nouveau_bo_download_dirty(GLcontext *ctx, struct gl_buffer_object *bo)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
-	nouveau_bufferobj_dirty *dirty = &nbo->gpu_dirty;
-	GLuint nr_dirty;
-	int i;
-
-	nr_dirty = dirty->nr_dirty;
-	if (nr_dirty) {
-		DEBUG("clean\n");
-		return nr_dirty;
-	}
-	
-	for (i=0; i<nr_dirty; i++) {
-		nouveau_bufferobj_region *r = &dirty->dirty[i];
-
-		DEBUG("dirty %d: o=0x%08x, s=0x%08x\n",
-				i, r->start, r->end - r->start);
-		nouveau_bo_download_from_screen(ctx,
-						r->start,
-						r->end - r->start, bo);
-	}
-
-	FREE(dirty->dirty);
-	dirty->dirty    = NULL;
-	dirty->nr_dirty = 0;
-
-	return nr_dirty;
-}
-
-static void
-nouveauBindBuffer(GLcontext *ctx, GLenum target, struct gl_buffer_object *obj)
-{
-}
-
-static struct gl_buffer_object *
-nouveauNewBufferObject(GLcontext *ctx, GLuint buffer, GLenum target)
-{
-	nouveau_buffer_object *nbo;
-
-	nbo = CALLOC_STRUCT(nouveau_buffer_object_t);
-	if (nbo)
-		_mesa_initialize_buffer_object(&nbo->mesa, buffer, target);
-	DEBUG("bo=%p\n", nbo);
-
-	return nbo ? &nbo->mesa : NULL;
-}
-
-static void
-nouveauDeleteBuffer(GLcontext *ctx, struct gl_buffer_object *obj)
-{
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
-
-	if (nbo->gpu_dirty.nr_dirty)
-		FREE(nbo->gpu_dirty.dirty);
-	if (nbo->cpu_dirty.nr_dirty)
-		FREE(nbo->cpu_dirty.dirty);
-	if (nbo->cpu_mem) nouveau_mem_free(ctx, nbo->cpu_mem);
-	if (nbo->gpu_mem) nouveau_mem_free(ctx, nbo->gpu_mem);
-
-	_mesa_delete_buffer_object(ctx, obj);
-}
-
-static void
-nouveauBufferData(GLcontext *ctx, GLenum target, GLsizeiptrARB size,
-		  const GLvoid *data, GLenum usage,
-		  struct gl_buffer_object *obj)
-{
-	GLuint gpu_flags;
-
-	DEBUG("target=%s, size=%d, data=%p, usage=%s, obj=%p\n",
-			_mesa_lookup_enum_by_nr(target),
-			(GLuint)size, data,
-			_mesa_lookup_enum_by_nr(usage),
-			obj);
-
-	switch (target) {
-	case GL_ELEMENT_ARRAY_BUFFER_ARB:
-		gpu_flags = 0;
-		break;
-	default:
-		gpu_flags = NOUVEAU_BO_VRAM_OK | NOUVEAU_BO_GART_OK;
-		break;
-	}
-	nouveau_bo_init_storage(ctx, gpu_flags, size, data, usage, obj);
-}
-
-static void
-nouveauBufferSubData(GLcontext *ctx, GLenum target, GLintptrARB offset,
-		     GLsizeiptrARB size, const GLvoid *data,
-		     struct gl_buffer_object *obj)
-{
-	GLvoid *out;
-
-	DEBUG("target=%s, offset=0x%x, size=%d, data=%p, obj=%p\n",
-			_mesa_lookup_enum_by_nr(target),
-			(GLuint)offset, (GLuint)size, data, obj);
-
-	out = nouveau_bo_map(ctx, GL_WRITE_ONLY_ARB, obj);
-	_mesa_memcpy(out + offset, data, size);
-	nouveau_bo_dirty_linear(ctx, GL_FALSE, offset, size, obj);
-	nouveau_bo_unmap(ctx, obj);
-}
-
-static void
-nouveauGetBufferSubData(GLcontext *ctx, GLenum target, GLintptrARB offset,
-		     GLsizeiptrARB size, GLvoid *data,
-		     struct gl_buffer_object *obj)
-{
-	const GLvoid *in;
-
-	DEBUG("target=%s, offset=0x%x, size=%d, data=%p, obj=%p\n",
-			_mesa_lookup_enum_by_nr(target),
-			(GLuint)offset, (GLuint)size, data, obj);
-
-	in = nouveau_bo_map(ctx, GL_READ_ONLY_ARB, obj);
-	_mesa_memcpy(data, in + offset, size);
-	nouveau_bo_unmap(ctx, obj);
-}
-
-static void *
-nouveauMapBuffer(GLcontext *ctx, GLenum target, GLenum access,
-		 struct gl_buffer_object *obj)
-{
-	DEBUG("target=%s, access=%s, obj=%p\n",
-			_mesa_lookup_enum_by_nr(target),
-			_mesa_lookup_enum_by_nr(access),
-			obj
-			);
-
-	/* Already mapped.. */
-	if (obj->Pointer)
-		return NULL;
-
-	/* Have to pass READ_WRITE here, nouveau_bo_map will only ensure that
-	 * the cpu_mem buffer is up-to-date if we ask for read access.
-	 *
-	 * However, even if the client only asks for write access, we're still
-	 * forced to reupload the entire buffer.  So, we need the cpu_mem buffer
-	 * to have the correct data all the time.
-	 */
-	obj->Pointer = nouveau_bo_map(ctx, GL_READ_WRITE_ARB, obj);
-
-	/* The GL spec says that a client attempting to write to a bufferobj
-	 * mapped READ_ONLY object may have unpredictable results, possibly
-	 * even program termination.
-	 *
-	 * We're going to use this, and only mark the buffer as dirtied if
-	 * the client asks for write access.
-	 */
-	if (target != GL_READ_ONLY_ARB) {
-		/* We have no way of knowing what was modified by the client,
-		 * so the entire buffer gets dirtied. */
-		nouveau_bo_dirty_all(ctx, GL_FALSE, obj);
-	}
-
-	return obj->Pointer;
-}
-
-static GLboolean
-nouveauUnmapBuffer(GLcontext *ctx, GLenum target, struct gl_buffer_object *obj)
-{
-	DEBUG("target=%s, obj=%p\n", _mesa_lookup_enum_by_nr(target), obj);
-
-	assert(obj->Pointer);
-
-	nouveau_bo_unmap(ctx, obj);
-	obj->Pointer = NULL;
-	return GL_TRUE;
-}
-	  
-void
-nouveauInitBufferObjects(GLcontext *ctx)
-{
-	ctx->Driver.BindBuffer		= nouveauBindBuffer;
-	ctx->Driver.NewBufferObject	= nouveauNewBufferObject;
-	ctx->Driver.DeleteBuffer	= nouveauDeleteBuffer;
-	ctx->Driver.BufferData		= nouveauBufferData;
-	ctx->Driver.BufferSubData	= nouveauBufferSubData;
-	ctx->Driver.GetBufferSubData	= nouveauGetBufferSubData;
-	ctx->Driver.MapBuffer		= nouveauMapBuffer;
-	ctx->Driver.UnmapBuffer		= nouveauUnmapBuffer;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
deleted file mode 100644
index 3439a35e7c8..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
+++ /dev/null
@@ -1,77 +0,0 @@
-#ifndef __NOUVEAU_BUFFEROBJ_H__
-#define __NOUVEAU_BUFFEROBJ_H__
-
-#include "mtypes.h"
-#include "nouveau_buffers.h"
-
-#define NOUVEAU_BO_VRAM_OK (NOUVEAU_MEM_FB | NOUVEAU_MEM_FB_ACCEPTABLE)
-#define NOUVEAU_BO_GART_OK  (NOUVEAU_MEM_AGP | NOUVEAU_MEM_AGP_ACCEPTABLE)
-
-typedef struct nouveau_bufferobj_region_t {
-	uint32_t start;
-	uint32_t end;
-} nouveau_bufferobj_region;
-
-typedef struct nouveau_bufferobj_dirty_t {
-	nouveau_bufferobj_region *dirty;
-	int nr_dirty;
-} nouveau_bufferobj_dirty;
-
-typedef struct nouveau_buffer_object_t {
-	/* Base class, must be first */
-	struct gl_buffer_object mesa;
-
-	GLboolean		mapped;
-	GLenum			usage;
-
-	/* Memory used for GPU access to the buffer*/
-	GLuint			gpu_mem_flags;
-	nouveau_mem *		gpu_mem;
-	nouveau_bufferobj_dirty	gpu_dirty;
-
-	/* Memory used for CPU access to the buffer */
-	GLuint			cpu_mem_flags;
-	nouveau_mem *		cpu_mem;
-	GLvoid *		cpu_mem_sys;
-	nouveau_bufferobj_dirty	cpu_dirty;
-} nouveau_buffer_object;
-
-extern void
-nouveau_bo_init_storage(GLcontext *ctx, GLuint valid_gpu_access,
-			GLsizeiptrARB size, const GLvoid *data, GLenum usage,
-			struct gl_buffer_object *bo);
-
-extern GLboolean
-nouveau_bo_move_in(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern GLboolean
-nouveau_bo_move_out(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern void *
-nouveau_bo_map(GLcontext *ctx, GLenum usage, struct gl_buffer_object *bo);
-
-extern void
-nouveau_bo_unmap(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern uint32_t
-nouveau_bo_gpu_ref(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern void
-nouveau_bo_dirty_linear(GLcontext *ctx, GLboolean on_card,
-			uint32_t offset, uint32_t size,
-			struct gl_buffer_object *bo);
-
-extern void
-nouveau_bo_dirty_all(GLcontext *ctx, GLboolean on_card,
-		     struct gl_buffer_object *bo);
-
-extern GLuint
-nouveau_bo_upload_dirty(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern GLuint
-nouveau_bo_download_dirty(GLcontext *ctx, struct gl_buffer_object *bo);
-
-extern void
-nouveauInitBufferObjects(GLcontext *ctx);
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
deleted file mode 100644
index d498f616c90..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
+++ /dev/null
@@ -1,436 +0,0 @@
-#include "utils.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "fbobject.h"
-
-#include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_msg.h"
-
-#define MAX_MEMFMT_LENGTH 32768
-
-/* Unstrided blit using NV_MEMORY_TO_MEMORY_FORMAT */
-GLboolean
-nouveau_memformat_flat_emit(GLcontext *ctx,
-      			    nouveau_mem *dst, nouveau_mem *src,
-			    GLuint dst_offset, GLuint src_offset,
-			    GLuint size)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   uint32_t src_handle, dst_handle;
-   GLuint count;
-
-   if (src_offset + size > src->size) {
-      MESSAGE("src out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-   if (dst_offset + size > dst->size) {
-      MESSAGE("dst out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-
-   src_handle = (src->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   dst_handle = (dst->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   src_offset += nouveau_mem_gpu_offset_get(ctx, src);
-   dst_offset += nouveau_mem_gpu_offset_get(ctx, dst);
-
-   BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN, 2);
-   OUT_RING       (src_handle);
-   OUT_RING       (dst_handle);
-
-   count = (size / MAX_MEMFMT_LENGTH) + ((size % MAX_MEMFMT_LENGTH) ? 1 : 0);
-
-   while (count--) {
-      GLuint length = (size > MAX_MEMFMT_LENGTH) ? MAX_MEMFMT_LENGTH : size;
-
-      BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
-      OUT_RING       (src_offset);
-      OUT_RING       (dst_offset);
-      OUT_RING       (0); /* pitch in */
-      OUT_RING       (0); /* pitch out */
-      OUT_RING       (length); /* line length */
-      OUT_RING       (1); /* number of lines */
-      OUT_RING       ((1 << 8) /* dst_inc */ | (1 << 0) /* src_inc */);
-      OUT_RING       (0); /* buffer notify? */
-      FIRE_RING();
-
-      src_offset += length;
-      dst_offset += length;
-      size       -= length;
-   }
-
-   return GL_TRUE;
-}
-
-void
-nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   struct drm_nouveau_mem_free memf;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (mem->map)
-      drmUnmap(mem->map, mem->size);
-   memf.flags  = mem->type;
-   memf.offset = mem->offset;
-   drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf, sizeof(memf));
-   FREE(mem);
-}
-
-nouveau_mem *
-nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   struct drm_nouveau_mem_alloc mema;
-   nouveau_mem *mem;
-   int ret;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: requested: type=0x%x, size=0x%x, align=0x%x\n",
-	    __func__, type, (GLuint)size, align);
-   }
-
-   mem = CALLOC(sizeof(nouveau_mem));
-   if (!mem)
-      return NULL;
-
-   mema.flags     = type;
-   mema.size      = mem->size = size;
-   mema.alignment = align;
-   mem->map       = NULL;
-   ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_MEM_ALLOC,
-	 		     &mema, sizeof(mema));
-   if (ret) {
-      FREE(mem);
-      return NULL;
-   }
-   mem->offset = mema.offset;
-   mem->type   = mema.flags;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: actual: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (type & NOUVEAU_MEM_MAPPED)
-      ret = drmMap(nmesa->driFd, mema.map_handle, mem->size, &mem->map);
-   if (ret) {
-      mem->map = NULL;
-      nouveau_mem_free(ctx, mem);
-      mem = NULL;
-   }
-
-   return mem;
-}
-
-uint32_t
-nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   return mem->offset;
-}
-
-static GLboolean
-nouveau_renderbuffer_pixelformat(nouveau_renderbuffer *nrb,
-      				 GLenum internalFormat)
-{
-      nrb->mesa.InternalFormat	= internalFormat;
-
-   /*TODO: We probably want to extend this a bit, and maybe make
-    *      card-specific? 
-    */
-      switch (internalFormat) {
-      case GL_RGBA:
-      case GL_RGBA8:
-	 nrb->mesa._BaseFormat	= GL_RGBA;
-	 nrb->mesa._ActualFormat= GL_RGBA8;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 8;
-	 nrb->mesa.GreenBits	= 8;
-	 nrb->mesa.BlueBits	= 8;
-	 nrb->mesa.AlphaBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_RGB:
-      case GL_RGB5:
-	 nrb->mesa._BaseFormat	= GL_RGB;
-	 nrb->mesa._ActualFormat= GL_RGB5;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 5;
-	 nrb->mesa.GreenBits	= 6;
-	 nrb->mesa.BlueBits	= 5;
-	 nrb->mesa.AlphaBits	= 0;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT16:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH_COMPONENT16;
-	 nrb->mesa.DataType	= GL_UNSIGNED_SHORT;
-	 nrb->mesa.DepthBits	= 16;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT24:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_STENCIL_INDEX8_EXT:
-	 nrb->mesa._BaseFormat	= GL_STENCIL_INDEX;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.StencilBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_DEPTH24_STENCIL8_EXT:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_STENCIL_EXT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->mesa.StencilBits  = 8;
-	 nrb->cpp		= 4;
-	 break;
-      default:
-	 return GL_FALSE;
-	 break;
-      }
-
-      return GL_TRUE;
-}
-
-static GLboolean
-nouveau_renderbuffer_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
-      			     GLenum internalFormat,
-			     GLuint width,
-			     GLuint height)
-{
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
-      fprintf(stderr, "%s: unknown internalFormat\n", __func__);
-      return GL_FALSE;
-   }
-
-   /* If this buffer isn't statically alloc'd, we may need to ask the
-    * drm for more memory */
-   if (!nrb->dPriv && (rb->Width != width || rb->Height != height)) {
-      GLuint pitch;
-
-      /* align pitches to 64 bytes */
-      pitch = ((width * nrb->cpp) + 63) & ~63;
-
-      if (nrb->mem)
-	 nouveau_mem_free(ctx, nrb->mem);
-      nrb->mem = nouveau_mem_alloc(ctx,
-	    			   NOUVEAU_MEM_FB | NOUVEAU_MEM_MAPPED,
-				   pitch*height,
-				   0);
-      if (!nrb->mem)
-	 return GL_FALSE;
-
-      /* update nouveau_renderbuffer info */
-      nrb->offset = nouveau_mem_gpu_offset_get(ctx, nrb->mem);
-      nrb->pitch  = pitch;
-   }
-
-   rb->Width = width;
-   rb->Height = height;
-   rb->InternalFormat = internalFormat;
-   return GL_TRUE;
-}
-
-static void
-nouveau_renderbuffer_delete(struct gl_renderbuffer *rb)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (nrb->mem)
-      nouveau_mem_free(ctx, nrb->mem);
-   FREE(nrb);
-}
-
-nouveau_renderbuffer *
-nouveau_renderbuffer_new(GLenum internalFormat, GLvoid *map,
-      			 GLuint offset,  GLuint pitch,
-			 __DRIdrawablePrivate *dPriv)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, 0);
-
-      nouveau_renderbuffer_pixelformat(nrb, internalFormat);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-
-      nrb->dPriv  = dPriv;
-      nrb->offset = offset;
-      nrb->pitch  = pitch;
-      nrb->map    = map;
-   }
-
-   return nrb;
-}
-
-static void
-nouveau_cliprects_drawable_set(nouveauContextPtr nmesa,
-      			       nouveau_renderbuffer *nrb)
-{
-   __DRIdrawablePrivate *dPriv = nrb->dPriv;
-
-   nmesa->numClipRects	= dPriv->numClipRects;
-   nmesa->pClipRects	= dPriv->pClipRects;
-   nmesa->drawX		= dPriv->x;
-   nmesa->drawY		= dPriv->y;
-   nmesa->drawW		= dPriv->w;
-   nmesa->drawH		= dPriv->h;
-}
-
-static void
-nouveau_cliprects_renderbuffer_set(nouveauContextPtr nmesa,
-      				   nouveau_renderbuffer *nrb)
-{
-   nmesa->numClipRects	= 1;
-   nmesa->pClipRects	= &nmesa->osClipRect;
-   nmesa->osClipRect.x1	= 0;
-   nmesa->osClipRect.y1	= 0;
-   nmesa->osClipRect.x2	= nrb->mesa.Width;
-   nmesa->osClipRect.y2	= nrb->mesa.Height;
-   nmesa->drawX		= 0;
-   nmesa->drawY		= 0;
-   nmesa->drawW		= nrb->mesa.Width;
-   nmesa->drawH		= nrb->mesa.Height;
-}
-
-void
-nouveau_window_moved(GLcontext *ctx)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb;
-
-   nrb = (nouveau_renderbuffer *)ctx->DrawBuffer->_ColorDrawBuffers[0][0];
-   if (!nrb)
-      return;
-
-   if (!nrb->dPriv)
-      nouveau_cliprects_renderbuffer_set(nmesa, nrb);
-   else
-      nouveau_cliprects_drawable_set(nmesa, nrb);
-
-   /* Viewport depends on window size/position, nouveauCalcViewport
-    * will take care of calling the hw-specific WindowMoved
-    */
-   ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
-	 		     ctx->Viewport.Width, ctx->Viewport.Height);
-   /* Scissor depends on window position */
-   ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	 		    ctx->Scissor.Width, ctx->Scissor.Height);
-}
-
-GLboolean
-nouveau_build_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *color[MAX_DRAW_BUFFERS];
-   nouveau_renderbuffer *depth;
-
-   _mesa_update_framebuffer(ctx);
-   _mesa_update_draw_buffer_bounds(ctx);
-
-   color[0] = (nouveau_renderbuffer *)fb->_ColorDrawBuffers[0][0];
-   if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped)
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer->Wrapped;
-   else
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer;
-
-   if (!nmesa->hw_func.BindBuffers(nmesa, 1, color, depth))
-      return GL_FALSE;
-   nouveau_window_moved(ctx);
-
-   return GL_TRUE;
-}
-
-static void
-nouveauDrawBuffer(GLcontext *ctx, GLenum buffer)
-{
-   nouveau_build_framebuffer(ctx, ctx->DrawBuffer);
-}
-
-static struct gl_framebuffer *
-nouveauNewFramebuffer(GLcontext *ctx, GLuint name)
-{
-   return _mesa_new_framebuffer(ctx, name);
-}
-
-static struct gl_renderbuffer *
-nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, name);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-   }
-   return &nrb->mesa;
-}
-
-static void
-nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
-                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
-{
-   if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
-      nouveau_build_framebuffer(ctx, fb);
-   }
-}
-
-static void
-nouveauFramebufferRenderbuffer(GLcontext *ctx,
-      			       struct gl_framebuffer *fb,
-      			       GLenum attachment,
-			       struct gl_renderbuffer *rb)
-{
-   _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
-   nouveau_build_framebuffer(ctx, fb);
-}
-
-static void
-nouveauRenderTexture(GLcontext *ctx,
-      		     struct gl_framebuffer *fb,
-      		     struct gl_renderbuffer_attachment *att)
-{
-}
-
-static void
-nouveauFinishRenderTexture(GLcontext *ctx,
-      			   struct gl_renderbuffer_attachment *att)
-{
-}
-
-void
-nouveauInitBufferFuncs(struct dd_function_table *func)
-{
-   func->DrawBuffer		 = nouveauDrawBuffer;
-
-   func->NewFramebuffer		 = nouveauNewFramebuffer;
-   func->NewRenderbuffer	 = nouveauNewRenderbuffer;
-   func->BindFramebuffer	 = nouveauBindFramebuffer;
-   func->FramebufferRenderbuffer = nouveauFramebufferRenderbuffer;
-   func->RenderTexture		 = nouveauRenderTexture;
-   func->FinishRenderTexture	 = nouveauFinishRenderTexture;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h b/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
deleted file mode 100644
index d86455184c2..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
+++ /dev/null
@@ -1,48 +0,0 @@
-#ifndef __NOUVEAU_BUFFERS_H__
-#define __NOUVEAU_BUFFERS_H__
-
-#include <stdint.h>
-#include "mtypes.h"
-#include "utils.h"
-#include "renderbuffer.h"
-
-typedef struct nouveau_mem_t {
-   int type;
-   uint64_t offset;
-   uint64_t size;
-   void*    map;
-} nouveau_mem;
-
-extern nouveau_mem *nouveau_mem_alloc(GLcontext *ctx, int type,
-      				      GLuint size, GLuint align);
-extern void nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem);
-extern uint32_t nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem);
-
-extern GLboolean nouveau_memformat_flat_emit(GLcontext *ctx,
-      					     nouveau_mem *dst,
-					     nouveau_mem *src,
-					     GLuint dst_offset,
-					     GLuint src_offset,
-					     GLuint size);
-
-typedef struct nouveau_renderbuffer_t {
-   struct gl_renderbuffer mesa; /* must be first! */
-   __DRIdrawablePrivate  *dPriv;
-
-   nouveau_mem *mem;
-   void *	map;
-
-   int		cpp;
-   uint32_t	offset;
-   uint32_t	pitch;
-} nouveau_renderbuffer;
-
-extern nouveau_renderbuffer *nouveau_renderbuffer_new(GLenum internalFormat,
-      GLvoid *map, GLuint offset, GLuint pitch, __DRIdrawablePrivate *dPriv);
-extern void nouveau_window_moved(GLcontext *ctx);
-extern GLboolean nouveau_build_framebuffer(GLcontext *, struct gl_framebuffer *);
-extern nouveau_renderbuffer *nouveau_current_draw_buffer(GLcontext *ctx);
-
-extern void nouveauInitBufferFuncs(struct dd_function_table *func);
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card.c b/src/mesa/drivers/dri/nouveau/nouveau_card.c
deleted file mode 100644
index 91f12f0d704..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_card.c
+++ /dev/null
@@ -1,17 +0,0 @@
-
-#include "nouveau_card.h"
-#include "nouveau_reg.h"
-#include "nouveau_drm.h"
-#include "nouveau_card_list.h"
-
-
-nouveau_card* nouveau_card_lookup(uint32_t device_id)
-{
-	int i;
-	for(i=0;i<sizeof(nouveau_card_list)/sizeof(nouveau_card)-1;i++)
-		if (nouveau_card_list[i].id==(device_id&0xffff))
-			return &(nouveau_card_list[i]);
-	return NULL;
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card.h b/src/mesa/drivers/dri/nouveau/nouveau_card.h
deleted file mode 100644
index 8a4c5f2244e..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_card.h
+++ /dev/null
@@ -1,49 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef __NOUVEAU_CARD_H__
-#define __NOUVEAU_CARD_H__
-
-#include "dri_util.h"
-#include "drm.h"
-#include "nouveau_drm.h"
-
-typedef struct nouveau_card_t {
-	uint16_t id; /* last 4 digits of pci id, last digit is always 0 */
-	char* name; /* the user-friendly card name */
-	uint32_t class_3d; /* the object class this card uses for 3D */
-	uint32_t type; /* the major card family */
-	uint32_t flags;
-}
-nouveau_card;
-
-#define NV_HAS_LMA 0x00000001
-
-extern nouveau_card* nouveau_card_lookup(uint32_t device_id);
-
-#endif
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h b/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
deleted file mode 100644
index 8ec5c4a188a..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
+++ /dev/null
@@ -1,232 +0,0 @@
-static nouveau_card nouveau_card_list[]={
-{0x0008,  "EDGE 3D",                               0,                             NV_03, 0},
-{0x0009,  "EDGE 3D",                               0,                             NV_03, 0},
-{0x0010,  "Mutara V08",                            0,                             NV_03, 0},
-{0x0020,  "RIVA TNT",                              NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x0028,  "RIVA TNT2/TNT2 Pro",                    NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x0029,  "RIVA TNT2 Ultra",                       NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002A,  "Riva TnT2",                             NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002B,  "Riva TnT2",                             NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002C,  "Vanta/Vanta LT",                        NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002D,  "RIVA TNT2 Model 64/Model 64 Pro",       NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002E,  "Vanta",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002F,  "Vanta",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x0040,  "GeForce 6800 Ultra",                    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0041,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0042,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0043,  "NV40.3",                                NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0044,  "GeForce 6800 XT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0045,  "GeForce 6800 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0046,  "GeForce 6800 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0047,  "GeForce 6800 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0048,  "GeForce 6800 XT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0049,  "NV40GL",                                NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x004D,  "Quadro FX 4000",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x004E,  "Quadro FX 4000",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0090,  "GeForce 7800 GTX",                      NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0091,  "GeForce 7800 GTX",                      NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0092,  "GeForce 7800 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0093,  "GeForce 7800 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0098,  "GeForce Go 7800",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0099,  "GE Force Go 7800 GTX",                  NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x009D,  "Quadro FX4500",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00A0,  "Aladdin TNT2",                          NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x00C0,  "GeForce 6800 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00C1,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00C2,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00C3,  "Geforce 6800 XT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00C8,  "GeForce Go 6800",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00C9,  "GeForce Go 6800 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00CC,  "Quadro FX Go1400",                      NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00CD,  "Quadro FX 3450/4000 SDI",               NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00CE,  "Quadro FX 1400",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F0,  "GeForce 6800/GeForce 6800 Ultra",       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F1,  "GeForce 6600/GeForce 6600 GT",          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F2,  "GeForce 6600/GeForce 6600 GT",          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F3,  "GeForce 6200",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F4,  "GeForce 6600 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F5,  "GeForce 7800 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F6,  "GeForce 6600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F8,  "Quadro FX 3400/4400",                   NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00F9,  "GeForce 6800 Ultra/GeForce 6800 GT",    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00FA,  "GeForce PCX 5750",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x00FB,  "GeForce PCX 5900",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x00FC,  "Quadro FX 330/GeForce PCX 5300",        NV30_TCL_PRIMITIVE_3D|0x0600,  NV_30, 0},
-{0x00FD,  "Quadro FX 330/Quadro NVS280",           NV30_TCL_PRIMITIVE_3D|0x0600,  NV_30, 0},
-{0x00FE,  "Quadro FX 1300",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x00FF,  "GeForce PCX 4300",                      NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0100,  "GeForce 256 SDR",                       NV10_TCL_PRIMITIVE_3D,         NV_10, 0},
-{0x0101,  "GeForce 256 DDR",                       NV10_TCL_PRIMITIVE_3D,         NV_10, 0},
-{0x0103,  "Quadro",                                NV10_TCL_PRIMITIVE_3D,         NV_10, 0},
-{0x0110,  "GeForce2 MX/MX 400",                    NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
-{0x0111,  "GeForce2 MX 100 DDR/200 DDR",           NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
-{0x0112,  "GeForce2 Go",                           NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
-{0x0113,  "Quadro2 MXR/EX/Go",                     NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
-{0x0140,  "GeForce 6600 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0141,  "GeForce 6600",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0142,  "GeForce 6600 PCIe",                     NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0144,  "GeForce Go 6600",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0145,  "GeForce 6610 XL",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0146,  "Geforce Go 6600TE/6200TE",              NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0148,  "GeForce Go 6600",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0149,  "GeForce Go 6600 GT",                    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014A,  "Quadro NVS 440",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014C,  "Quadro FX 550",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014D,  "Quadro FX 550",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014E,  "Quadro FX 540",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014F,  "GeForce 6200",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0150,  "GeForce2 GTS/Pro",                      NV11_TCL_PRIMITIVE_3D,         NV_15, 0},
-{0x0151,  "GeForce2 Ti",                           NV11_TCL_PRIMITIVE_3D,         NV_15, 0},
-{0x0152,  "GeForce2 Ultra, Bladerunner",           NV11_TCL_PRIMITIVE_3D,         NV_15, 0},
-{0x0153,  "Quadro2 Pro",                           NV11_TCL_PRIMITIVE_3D,         NV_15, 0},
-{0x0161,  "GeForce 6200 TurboCache(TM)",           NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0162,  "GeForce 6200 SE TurboCache (TM)",       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0163,  "GeForce 6200 LE",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0164,  "GeForce Go 6200",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0165,  "Quadro NVS 285",                        NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0166,  "GeForce Go 6400",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0167,  "GeForce Go 6200 TurboCache",            NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0168,  "GeForce Go 6200 TurboCache",            NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0170,  "GeForce4 MX 460",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0171,  "GeForce4 MX 440",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0172,  "GeForce4 MX 420",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0173,  "GeForce4 MX 440-SE",                    NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0174,  "GeForce4 440 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0175,  "GeForce4 420 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0176,  "GeForce4 420 Go 32M",                   NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0177,  "GeForce4 460 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0178,  "Quadro4 550 XGL",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0179,  "GeForce4 420 Go 32M",                   NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x017A,  "Quadro4 200/400 NVS",                   NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x017B,  "Quadro4 550 XGL",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x017C,  "Quadro4 500 GoGL",                      NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x017D,  "GeForce4 410 Go 16M",                   NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0181,  "GeForce4 MX 440 AGP 8x",                NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0182,  "GeForce4 MX 440SE AGP 8x",              NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0183,  "GeForce4 MX 420 AGP 8x",                NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0185,  "GeForce4 MX 4000 AGP 8x",               NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0186,  "GeForce4 448 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0187,  "GeForce4 488 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0188,  "Quadro4 580 XGL",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x018A,  "Quadro4 NVS AGP 8x",                    NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x018B,  "Quadro4 380 XGL",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x018C,  "Quadro NVS 50 PCI",                     NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x018D,  "GeForce4 448 Go",                       NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0191,  "GeForce 8800 GTX",                      NV30_TCL_PRIMITIVE_3D|0x5000,  NV_50, 0},
-{0x0193,  "GeForce 8800 GTS",                      NV30_TCL_PRIMITIVE_3D|0x5000,  NV_50, 0},
-{0x01A0,  "GeForce2 MX Integrated Graphics",       NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
-{0x01D1,  "GeForce 7300 LE",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01D6,  "GeForce Go 7200",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01D7,  "Quadro NVS 110M / GeForce Go 7300",     NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01D8,  "GeForce Go 7400",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01DA,  "Quadro NVS 110M",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01DF,  "GeForce 7300 GS",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x01F0,  "GeForce4 MX - nForce GPU",              NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0200,  "GeForce3",                              NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
-{0x0201,  "GeForce3 Ti 200",                       NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
-{0x0202,  "GeForce3 Ti 500",                       NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
-{0x0203,  "Quadro DCC",                            NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
-{0x0211,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0212,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0215,  "GeForce 6800 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0218,  "GeForce 6800 XT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0221,  "GeForce 6200",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0240,  "GeForce 6150",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0242,  "GeForce 6100",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0244,  "Geforce 6150 Go",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0250,  "GeForce4 Ti 4600",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0251,  "GeForce4 Ti 4400",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0252,  "GeForce4 Ti",                           NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0253,  "GeForce4 Ti 4200",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0258,  "Quadro4 900 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0259,  "Quadro4 750 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x025B,  "Quadro4 700 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0280,  "GeForce4 Ti 4800",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0281,  "GeForce4 Ti 4200 AGP 8x",               NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0282,  "GeForce4 Ti 4800 SE",                   NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0286,  "GeForce4 Ti 4200 Go AGP 8x",            NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0288,  "Quadro4 980 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0289,  "Quadro4 780 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x028C,  "Quadro4 700 GoGL",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
-{0x0290,  "GeForce 7900 GTX",                      NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0291,  "GeForce 7900 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0292,  "GeForce 7900 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0298,  "GeForce Go 7900 GS",                    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0299,  "GeForce Go 7900 GTX",                   NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029A,  "Quadro FX 2500M",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029B,  "Quadro FX 1500M",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029C,  "Quadro FX 5500",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029D,  "Quadro FX 3500",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029E,  "Quadro FX 1500",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x029F,  "Quadro FX 4500 X2",                     NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x02A0,  "XGPU",                                  NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
-{0x02E1,  "GeForce 7600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0300,  "GeForce FX",                            NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0301,  "GeForce FX 5800 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0302,  "GeForce FX 5800",                       NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0308,  "Quadro FX 2000",                        NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0309,  "Quadro FX 1000",                        NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0311,  "GeForce FX 5600 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0312,  "GeForce FX 5600",                       NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0313,  "NV31",                                  NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0314,  "GeForce FX 5600XT",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0316,  "NV31M",                                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0317,  "NV31M Pro",                             NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031A,  "GeForce FX Go5600",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031B,  "GeForce FX Go5650",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031C,  "NVIDIA Quadro FX Go700",                NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031D,  "NV31GLM",                               NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031E,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x031F,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
-{0x0320,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0321,  "GeForce FX 5200 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0322,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0323,  "GeForce FX 5200LE",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0324,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0325,  "GeForce FX Go5250",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0326,  "GeForce FX 5500",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0327,  "GeForce FX 5100",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0328,  "GeForce FX Go5200 32M/64M",             NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0329,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x032A,  "Quadro NVS 280 PCI",                    NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x032B,  "Quadro FX 500/600 PCI",                 NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x032C,  "GeForce FX Go 5300",                    NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x032D,  "GeForce FX Go5100",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x032F,  "NV34GL",                                NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
-{0x0330,  "GeForce FX 5900 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0331,  "GeForce FX 5900",                       NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0332,  "GeForce FX 5900XT",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0333,  "GeForce FX 5950 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0334,  "GeForce FX 5900ZT",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0338,  "Quadro FX 3000",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x033F,  "Quadro FX 700",                         NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0341,  "GeForce FX 5700 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0342,  "GeForce FX 5700",                       NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0343,  "GeForce FX 5700LE",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0344,  "GeForce FX 5700VE",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0345,  "NV36.5",                                NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0347,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0348,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0349,  "NV36M Pro",                             NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x034B,  "NV36MAP",                               NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x034C,  "Quadro FX Go1000",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x034E,  "Quadro FX 1100",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x034F,  "NV36GL",                                NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0391,  "GeForce 7600 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0392,  "GeForce 7600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0393,  "GeForce 7300 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0398,  "GeForce Go 7600",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x03D0,  "GeForce 6100 nForce 430",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x03D1,  "GeForce 6100 nForce 405",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x03D2,  "GeForce 6100 nForce 400",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x03D5,  "GeForce 6100 nForce 420",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0008,  "NV1",                                   0,                             NV_03, 0},
-{0x0009,  "DAC64",                                 0,                             NV_03, 0},
-{0x0018,  "Riva128",                               0,                             NV_03, 0},
-{0x0019,  "Riva128ZX",                             0,                             NV_03, 0},
-{0x0020,  "TNT",                                   NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x0028,  "TNT2",                                  NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x0029,  "UTNT2",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x002C,  "VTNT2",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-{0x00A0,  "ITNT2",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
-};
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c
deleted file mode 100644
index 44392c0267c..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.c
+++ /dev/null
@@ -1,380 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "glheader.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "framebuffer.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vp_build.h"
-
-#include "drivers/common/driverfuncs.h"
-
-#include "nouveau_context.h"
-#include "nouveau_driver.h"
-//#include "nouveau_state.h"
-#include "nouveau_span.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_tex.h"
-#include "nouveau_msg.h"
-#include "nouveau_reg.h"
-#include "nouveau_lock.h"
-#include "nouveau_query.h"
-#include "nv04_swtcl.h"
-#include "nv10_swtcl.h"
-
-#include "vblank.h"
-#include "utils.h"
-#include "texmem.h"
-#include "xmlpool.h" /* for symbolic values of enum-type options */
-
-#ifndef NOUVEAU_DEBUG
-int NOUVEAU_DEBUG = 0;
-#endif
-
-static const struct dri_debug_control debug_control[] =
-{
-	{ "shaders"   , DEBUG_SHADERS    },
-	{ "mem"       , DEBUG_MEM        },
-	{ "bufferobj" , DEBUG_BUFFEROBJ  },
-	{ NULL        , 0                }
-};
-
-#define need_GL_ARB_vertex_program
-#define need_GL_ARB_occlusion_query
-#include "extension_helper.h"
-
-const struct dri_extension common_extensions[] =
-{
-	{ NULL,    0 }
-};
-
-const struct dri_extension nv10_extensions[] =
-{
-	{ NULL,    0 }
-};
-
-const struct dri_extension nv20_extensions[] =
-{
-	{ NULL,    0 }
-};
-
-const struct dri_extension nv30_extensions[] =
-{
-	{ "GL_ARB_fragment_program",	NULL                            },
-	{ NULL,    0 }
-};
-
-const struct dri_extension nv40_extensions[] =
-{
-   /* ARB_vp can be moved to nv20/30 once the shader backend has been
-    * written for those cards.
-    */
-	{ "GL_ARB_vertex_program",	GL_ARB_vertex_program_functions },
-	{ "GL_ARB_occlusion_query",	GL_ARB_occlusion_query_functions},
-	{ NULL, 0 }
-};
-
-const struct dri_extension nv50_extensions[] =
-{
-	{ NULL,    0 }
-};
-
-/* Create the device specific context.
- */
-GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
-		__DRIcontextPrivate *driContextPriv,
-		void *sharedContextPrivate )
-{
-	GLcontext *ctx, *shareCtx;
-	__DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-	struct dd_function_table functions;
-	nouveauContextPtr nmesa;
-	nouveauScreenPtr screen;
-
-	/* Allocate the context */
-	nmesa = (nouveauContextPtr) CALLOC( sizeof(*nmesa) );
-	if ( !nmesa )
-		return GL_FALSE;
-
-	nmesa->driContext = driContextPriv;
-	nmesa->driScreen = sPriv;
-	nmesa->driDrawable = NULL;
-	nmesa->hHWContext = driContextPriv->hHWContext;
-	nmesa->driHwLock = &sPriv->pSAREA->lock;
-	nmesa->driFd = sPriv->fd;
-
-	nmesa->screen = (nouveauScreenPtr)(sPriv->private);
-	screen=nmesa->screen;
-
-	/* Create the hardware context */
-	if (!nouveauDRMGetParam(nmesa, NOUVEAU_GETPARAM_FB_PHYSICAL,
-		 		&nmesa->vram_phys))
-	   return GL_FALSE;
-	if (!nouveauDRMGetParam(nmesa, NOUVEAU_GETPARAM_FB_SIZE,
-		 		&nmesa->vram_size))
-	   return GL_FALSE;
-	if (!nouveauDRMGetParam(nmesa, NOUVEAU_GETPARAM_AGP_PHYSICAL,
-		 		&nmesa->gart_phys))
-	   return GL_FALSE;
-	if (!nouveauDRMGetParam(nmesa, NOUVEAU_GETPARAM_AGP_SIZE,
-		 		&nmesa->gart_size))
-	   return GL_FALSE;
-	if (!nouveauFifoInit(nmesa))
-	   return GL_FALSE;
-	nouveauObjectInit(nmesa);
-
-
-	/* Init default driver functions then plug in our nouveau-specific functions
-	 * (the texture functions are especially important)
-	 */
-	_mesa_init_driver_functions( &functions );
-	nouveauDriverInitFunctions( &functions );
-	nouveauTexInitFunctions( &functions );
-
-	/* Allocate the Mesa context */
-	if (sharedContextPrivate)
-		shareCtx = ((nouveauContextPtr) sharedContextPrivate)->glCtx;
-	else 
-		shareCtx = NULL;
-	nmesa->glCtx = _mesa_create_context(glVisual, shareCtx,
-			&functions, (void *) nmesa);
-	if (!nmesa->glCtx) {
-		FREE(nmesa);
-		return GL_FALSE;
-	}
-	driContextPriv->driverPrivate = nmesa;
-	ctx = nmesa->glCtx;
-
-	/* Parse configuration files */
-	driParseConfigFiles (&nmesa->optionCache, &screen->optionCache,
-			screen->driScreen->myNum, "nouveau");
-
-	nmesa->sarea = (struct drm_nouveau_sarea *)((char *)sPriv->pSAREA +
-			screen->sarea_priv_offset);
-
-	/* Enable any supported extensions */
-	driInitExtensions(ctx, common_extensions, GL_TRUE);
-	if (nmesa->screen->card->type >= NV_10)
-		driInitExtensions(ctx, nv10_extensions, GL_FALSE);
-	if (nmesa->screen->card->type >= NV_20)
-		driInitExtensions(ctx, nv20_extensions, GL_FALSE);
-	if (nmesa->screen->card->type >= NV_30)
-		driInitExtensions(ctx, nv30_extensions, GL_FALSE);
-	if (nmesa->screen->card->type >= NV_40)
-		driInitExtensions(ctx, nv40_extensions, GL_FALSE);
-	if (nmesa->screen->card->type >= NV_50)
-		driInitExtensions(ctx, nv50_extensions, GL_FALSE);
-
-	nmesa->current_primitive = -1;
-
-	nouveauShaderInitFuncs(ctx);
-	/* Install Mesa's fixed-function texenv shader support */
-	if (nmesa->screen->card->type >= NV_40)
-		ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
-
-	/* Initialize the swrast */
-	_swrast_CreateContext( ctx );
-	_vbo_CreateContext( ctx );
-	_tnl_CreateContext( ctx );
-	_swsetup_CreateContext( ctx );
-
-	_math_matrix_ctr(&nmesa->viewport);
-
-	nouveauDDInitStateFuncs( ctx );
-	nouveauSpanInitFunctions( ctx );
-	nouveauDDInitState( nmesa );
-	switch(nmesa->screen->card->type)
-	{
-		case NV_03:
-			//nv03TriInitFunctions( ctx );
-			break;
-		case NV_04:
-		case NV_05:
-			nv04TriInitFunctions( ctx );
-			break;
-		case NV_10:
-		case NV_11:
-		case NV_17:
-		case NV_20:
-		case NV_30:
-		case NV_40:
-		case NV_44:
-		case NV_50:
-		default:
-			nv10TriInitFunctions( ctx );
-			break;
-	}
-
-	nouveauInitBufferObjects(ctx);
-	if (!nouveauSyncInitFuncs(ctx))
-	   return GL_FALSE;
-	nouveauQueryInitFuncs(ctx);
-	nmesa->hw_func.InitCard(nmesa);
-        nouveauInitState(ctx);
-
-	driContextPriv->driverPrivate = (void *)nmesa;
-
-	NOUVEAU_DEBUG = driParseDebugString( getenv( "NOUVEAU_DEBUG" ),
-			debug_control );
-
-	if (driQueryOptionb(&nmesa->optionCache, "no_rast")) {
-		fprintf(stderr, "disabling 3D acceleration\n");
-		FALLBACK(nmesa, NOUVEAU_FALLBACK_DISABLE, 1);
-	}
-
-	return GL_TRUE;
-}
-
-/* Destroy the device specific context. */
-void nouveauDestroyContext( __DRIcontextPrivate *driContextPriv  )
-{
-	nouveauContextPtr nmesa = (nouveauContextPtr) driContextPriv->driverPrivate;
-
-	assert(nmesa);
-	if ( nmesa ) {
-		/* free the option cache */
-		driDestroyOptionCache (&nmesa->optionCache);
-
-		FREE( nmesa );
-	}
-
-}
-
-
-/* Force the context `c' to be the current context and associate with it
- * buffer `b'.
- */
-GLboolean nouveauMakeCurrent( __DRIcontextPrivate *driContextPriv,
-		__DRIdrawablePrivate *driDrawPriv,
-		__DRIdrawablePrivate *driReadPriv )
-{
-	if ( driContextPriv ) {
-		nouveauContextPtr nmesa = (nouveauContextPtr) driContextPriv->driverPrivate;
-		struct gl_framebuffer *draw_fb =
-			(struct gl_framebuffer*)driDrawPriv->driverPrivate;
-		struct gl_framebuffer *read_fb =
-			(struct gl_framebuffer*)driReadPriv->driverPrivate;
-
-		driDrawableInitVBlank(driDrawPriv, nmesa->vblank_flags, &nmesa->vblank_seq );
-		nmesa->driDrawable = driDrawPriv;
-
-		_mesa_resize_framebuffer(nmesa->glCtx, draw_fb,
-					 driDrawPriv->w, driDrawPriv->h);
-		if (draw_fb != read_fb) {
-			_mesa_resize_framebuffer(nmesa->glCtx, draw_fb,
-						 driReadPriv->w,
-						 driReadPriv->h);
-		}
-		_mesa_make_current(nmesa->glCtx, draw_fb, read_fb);
-
-		nouveau_build_framebuffer(nmesa->glCtx,
-		      			  driDrawPriv->driverPrivate);
-	} else {
-		_mesa_make_current( NULL, NULL, NULL );
-	}
-
-	return GL_TRUE;
-}
-
-
-/* Force the context `c' to be unbound from its buffer.
- */
-GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv )
-{
-	return GL_TRUE;
-}
-
-static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
-				 __DRIdrawablePrivate *dPriv)
-{
-	struct gl_framebuffer *fb;
-	nouveau_renderbuffer *src, *dst;
-	drm_clip_rect_t *box;
-	int nbox, i;
-
-	fb = (struct gl_framebuffer *)dPriv->driverPrivate;
-	dst = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
-	src = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
-
-#ifdef ALLOW_MULTI_SUBCHANNEL
-	LOCK_HARDWARE(nmesa);
-	nbox = dPriv->numClipRects;
-	box  = dPriv->pClipRects;
-
-	if (nbox) {
-		BEGIN_RING_SIZE(NvSubCtxSurf2D,
-				NV10_CONTEXT_SURFACES_2D_FORMAT, 4);
-		if (src->mesa._ActualFormat == GL_RGBA8)
-			OUT_RING       (6); /* X8R8G8B8 */
-		else
-			OUT_RING       (4); /* R5G6B5 */
-		OUT_RING       ((dst->pitch << 16) | src->pitch);
-		OUT_RING       (src->offset);
-		OUT_RING       (dst->offset);
-	}
-
-	for (i=0; i<nbox; i++, box++) {
-		BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_POINT, 3);
-		OUT_RING       (((box->y1 - dPriv->y) << 16) |
-				(box->x1 - dPriv->x));
-		OUT_RING       ((box->y1 << 16) | box->x1);
-		OUT_RING       (((box->y2 - box->y1) << 16) |
-				(box->x2 - box->x1));
-	}
-	FIRE_RING();
-
-	UNLOCK_HARDWARE(nmesa);
-#endif
-}
-
-void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv)
-{
-	if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-		nouveauContextPtr nmesa = dPriv->driContextPriv->driverPrivate;
-
-		if (nmesa->glCtx->Visual.doubleBufferMode) {
-			_mesa_notifySwapBuffers(nmesa->glCtx);
-			nouveauDoSwapBuffers(nmesa, dPriv);
-		}
-
-	}
-}
-
-void nouveauCopySubBuffer(__DRIdrawablePrivate *dPriv,
-			  int x, int y, int w, int h)
-{
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h
deleted file mode 100644
index db4d4cb6b71..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ /dev/null
@@ -1,239 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NOUVEAU_CONTEXT_H__
-#define __NOUVEAU_CONTEXT_H__
-
-#include "dri_util.h"
-#include "drm.h"
-#include "nouveau_drm.h"
-
-#include "mtypes.h"
-#include "tnl/t_vertex.h"
-
-#include "nouveau_screen.h"
-#include "nouveau_state_cache.h"
-#include "nouveau_buffers.h"
-#include "nouveau_shader.h"
-#include "nouveau_sync.h"
-
-#include "xmlconfig.h"
-
-typedef struct nouveau_fifo_t{
-	int channel;
-	uint32_t* buffer;
-	uint32_t* mmio;
-	uint32_t put_base;
-	uint32_t current;
-	uint32_t put;
-	uint32_t free;
-	uint32_t max;
-}
-nouveau_fifo;
-
-#define TAG(x) nouveau##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-/* Subpixel offsets for window coordinates (triangles): */
-#define SUBPIXEL_X  (0.0F)
-#define SUBPIXEL_Y  (0.125F)
-
-struct nouveau_context;
-
-typedef void (*nouveau_tri_func)( struct nouveau_context*, 
-		nouveauVertex *,
-		nouveauVertex *,
-		nouveauVertex * );
-
-typedef void (*nouveau_line_func)( struct nouveau_context*, 
-		nouveauVertex *,
-		nouveauVertex * );
-
-typedef void (*nouveau_point_func)( struct nouveau_context*,
-		nouveauVertex * );
-
-typedef struct nouveau_hw_func_t {
-	/* Initialise any card-specific non-GL related state */
-	GLboolean (*InitCard)(struct nouveau_context *);
-	/* Update buffer offset/pitch/format */
-	GLboolean (*BindBuffers)(struct nouveau_context *, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth);
-	/* Update anything that depends on the window position/size */
-	void      (*WindowMoved)(struct nouveau_context *);
-} nouveau_hw_func;
-
-typedef struct nouveau_context {
-	/* Mesa context */
-	GLcontext *glCtx;
-
-	/* The per-context fifo */
-	nouveau_fifo fifo;
-
-	/* The read-only regs */
-	volatile unsigned char* mmio;
-
-	/* The per-channel notifier block */
-	volatile void *notifier_block;
-
-	/* Physical addresses of AGP/VRAM apertures */
-	uint64_t vram_phys;
-	uint64_t vram_size;
-	uint64_t gart_phys;
-	uint64_t gart_size;
-
-	/* Channel synchronisation */
-	struct drm_nouveau_notifier_alloc *syncNotifier;
-
-	/* ARB_occlusion_query / EXT_timer_query */
-	GLuint		  query_object_max;
-	GLboolean *	  query_alloc;
-	struct drm_nouveau_notifier_alloc *queryNotifier;
-
-	/* Additional hw-specific functions */
-	nouveau_hw_func hw_func;
-
-	/* FIXME : do we want to put all state into a separate struct ? */
-	/* State for tris */
-	GLuint color_offset;
-	GLuint specular_offset;
-
-	/* Vertex state */
-	GLuint vertex_size;
-	GLubyte *verts;
-	struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
-	GLuint vertex_attr_count;
-
-	/* Color buffer clear value */
-	uint32_t clear_color_value;
-
-	/* Depth/stencil clear value */
-	uint32_t clear_value;
-
-	/* Light state */
-	GLboolean lighting_enabled;
-	uint32_t enabled_lights;
-
-	/* Cached state */
-	nouveau_state_cache state_cache;
-
-	/* The drawing fallbacks */
-	GLuint Fallback;
-	nouveau_tri_func draw_tri;
-	nouveau_line_func draw_line;
-	nouveau_point_func draw_point;
-
-	/* Cliprects information */
-	GLuint numClipRects;
-	drm_clip_rect_t *pClipRects;
-	drm_clip_rect_t osClipRect;
-	GLuint drawX, drawY, drawW, drawH;
-
-	/* The rendering context information */
-	GLenum current_primitive; /* the current primitive enum */
-	DECLARE_RENDERINPUTS(render_inputs_bitset); /* the current render inputs */
-
-	/* Shader state */
-	nvsFunc VPfunc;
-	nvsFunc FPfunc;
-	nouveauShader *current_fragprog;
-	nouveauShader *current_vertprog;
-	nouveauShader *passthrough_vp;
-	nouveauShader *passthrough_fp;
-
-	nouveauScreenRec *screen;
-	struct drm_nouveau_sarea *sarea;
-
-	__DRIcontextPrivate  *driContext;    /* DRI context */
-	__DRIscreenPrivate   *driScreen;     /* DRI screen */
-	__DRIdrawablePrivate *driDrawable;   /* DRI drawable bound to this ctx */
-	GLint lastStamp;
-
-	drm_context_t hHWContext;
-	drm_hw_lock_t *driHwLock;
-	int driFd;
-
-	/* Configuration cache */
-	driOptionCache optionCache;
-
-	/* vblank stuff */
-	uint32_t vblank_flags;
-	uint32_t vblank_seq;
-
-	GLuint new_state;
-	GLuint new_render_state;
-	GLuint render_index;
-	GLmatrix viewport;
-	GLfloat depth_scale;
-
-}nouveauContextRec, *nouveauContextPtr;
-
-
-#define NOUVEAU_CONTEXT(ctx)		((nouveauContextPtr)(ctx->DriverCtx))
-
-/* Flags for software fallback cases: */
-#define NOUVEAU_FALLBACK_TEXTURE		0x0001
-#define NOUVEAU_FALLBACK_DRAW_BUFFER		0x0002
-#define NOUVEAU_FALLBACK_READ_BUFFER		0x0004
-#define NOUVEAU_FALLBACK_STENCIL		0x0008
-#define NOUVEAU_FALLBACK_RENDER_MODE		0x0010
-#define NOUVEAU_FALLBACK_LOGICOP		0x0020
-#define NOUVEAU_FALLBACK_SEP_SPECULAR		0x0040
-#define NOUVEAU_FALLBACK_BLEND_EQ		0x0080
-#define NOUVEAU_FALLBACK_BLEND_FUNC		0x0100
-#define NOUVEAU_FALLBACK_PROJTEX		0x0200
-#define NOUVEAU_FALLBACK_DISABLE		0x0400
-
-
-extern GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
-		__DRIcontextPrivate *driContextPriv,
-		void *sharedContextPrivate );
-
-extern void nouveauDestroyContext( __DRIcontextPrivate * );
-
-extern GLboolean nouveauMakeCurrent( __DRIcontextPrivate *driContextPriv,
-		__DRIdrawablePrivate *driDrawPriv,
-		__DRIdrawablePrivate *driReadPriv );
-
-extern GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv );
-
-extern void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv);
-
-extern void nouveauCopySubBuffer(__DRIdrawablePrivate *dPriv,
-				 int x, int y, int w, int h);
-
-/* Debugging utils: */
-extern int NOUVEAU_DEBUG;
-
-#define DEBUG_SHADERS	0x00000001
-#define DEBUG_MEM	0x00000002
-#define DEBUG_BUFFEROBJ 0x00000004
-
-#endif /* __NOUVEAU_CONTEXT_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_ctrlreg.h b/src/mesa/drivers/dri/nouveau/nouveau_ctrlreg.h
deleted file mode 100644
index c9b2d590077..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_ctrlreg.h
+++ /dev/null
@@ -1,44 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin, Sylvain Munaut
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-
-#define NV03_STATUS                                        0x004006b0
-#define NV04_STATUS                                        0x00400700
-
-#define NV03_FIFO_REGS_SIZE                                0x10000
-#    define NV03_FIFO_REGS_DMAPUT                          0x00000040
-#    define NV03_FIFO_REGS_DMAGET                          0x00000044
-
-/* Fifo commands. These are not regs, neither masks */
-#define NV03_FIFO_CMD_JUMP                                 0x20000000
-#define NV03_FIFO_CMD_JUMP_OFFSET_MASK                     0x1ffffffc
-#define NV03_FIFO_CMD_REWIND                               (NV03_FIFO_CMD_JUMP | (0 & NV03_FIFO_CMD_JUMP_OFFSET_MASK))
-
-
-#define NONINC_METHOD                                      0x40000000
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.c b/src/mesa/drivers/dri/nouveau/nouveau_driver.c
deleted file mode 100644
index ddc9535624b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_driver.c
+++ /dev/null
@@ -1,146 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-//#include "nouveau_state.h"
-#include "nouveau_lock.h"
-#include "nouveau_fifo.h"
-#include "nouveau_driver.h"
-#include "swrast/swrast.h"
-
-#include "context.h"
-#include "framebuffer.h"
-
-#include "utils.h"
-
-/* Wrapper for DRM_NOUVEAU_GETPARAM ioctl */
-GLboolean nouveauDRMGetParam(nouveauContextPtr nmesa,
-			     unsigned int      param,
-			     uint64_t*         value)
-{
-	struct drm_nouveau_getparam getp;
-
-	getp.param = param;
-	if (!value || drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_GETPARAM,
-					  &getp, sizeof(getp)))
-		return GL_FALSE;
-	*value = getp.value;
-	return GL_TRUE;
-}
-
-/* Wrapper for DRM_NOUVEAU_GETPARAM ioctl */
-GLboolean nouveauDRMSetParam(nouveauContextPtr nmesa,
-			     unsigned int      param,
-			     uint64_t          value)
-{
-	struct drm_nouveau_setparam setp;
-
-	setp.param = param;
-	setp.value = value;
-	if (drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_SETPARAM, &setp,
-				sizeof(setp)))
-		return GL_FALSE;
-	return GL_TRUE;
-}
-
-/* Return the width and height of the current color buffer */
-static void nouveauGetBufferSize( GLframebuffer *buffer,
-		GLuint *width, GLuint *height )
-{
-	GET_CURRENT_CONTEXT(ctx);
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	LOCK_HARDWARE( nmesa );
-	*width  = nmesa->driDrawable->w;
-	*height = nmesa->driDrawable->h;
-	UNLOCK_HARDWARE( nmesa );
-}
-
-/* glGetString */
-static const GLubyte *nouveauGetString( GLcontext *ctx, GLenum name )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	static char buffer[128];
-	const char * card_name = "Unknown";
-	GLuint agp_mode = 0;
-
-	switch ( name ) {
-		case GL_VENDOR:
-			return (GLubyte *)DRIVER_AUTHOR;
-
-		case GL_RENDERER:
-			card_name=nmesa->screen->card->name;
-
-			switch(nmesa->screen->bus_type)
-			{
-				case NV_PCI:
-				case NV_PCIE:
-				default:
-					agp_mode=0;
-					break;
-				case NV_AGP:
-					agp_mode=nmesa->screen->agp_mode;
-					break;
-			}
-			driGetRendererString( buffer, card_name, DRIVER_DATE,
-					agp_mode );
-			return (GLubyte *)buffer;
-		default:
-			return NULL;
-	}
-}
-
-/* glFlush */
-static void nouveauFlush( GLcontext *ctx )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	FIRE_RING();
-}
-
-/* glFinish */
-static void nouveauFinish( GLcontext *ctx )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveauFlush( ctx );
-	nouveauWaitForIdle( nmesa );
-}
-
-/* glClear */
-static void nouveauClear( GLcontext *ctx, GLbitfield mask )
-{
-	// XXX we really should do something here...
-}
-
-void nouveauDriverInitFunctions( struct dd_function_table *functions )
-{
-	functions->GetBufferSize	= nouveauGetBufferSize;
-	functions->ResizeBuffers	= _mesa_resize_framebuffer;
-	functions->GetString		= nouveauGetString;
-	functions->Flush		= nouveauFlush;
-	functions->Finish		= nouveauFinish;
-	functions->Clear		= nouveauClear;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.h b/src/mesa/drivers/dri/nouveau/nouveau_driver.h
deleted file mode 100644
index 6164012b5b7..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_driver.h
+++ /dev/null
@@ -1,42 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NOUVEAU_DRIVER_H__
-#define __NOUVEAU_DRIVER_H__
-
-#define DRIVER_DATE	"20060219"
-#define DRIVER_AUTHOR	"Stephane Marchesin"
-
-extern void nouveauDriverInitFunctions( struct dd_function_table *functions );
-extern GLboolean nouveauDRMGetParam(nouveauContextPtr nmesa, unsigned int param,
-      				    uint64_t *value);
-extern GLboolean nouveauDRMSetParam(nouveauContextPtr nmesa, unsigned int param,
-      				    uint64_t value);
-
-#endif /* __NOUVEAU_DRIVER_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
deleted file mode 100644
index 5eb53aa46cb..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
+++ /dev/null
@@ -1,152 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#include "vblank.h"
-#include <errno.h>
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
-#include "swrast/swrast.h"
-#include "nouveau_context.h"
-#include "nouveau_msg.h"
-#include "nouveau_fifo.h"
-#include "nouveau_lock.h"
-#include "nouveau_object.h"
-#include "nouveau_sync.h"
-
-#ifdef NOUVEAU_RING_DEBUG
-int nouveau_fifo_remaining=0;
-#endif
-
-
-#define RING_SKIPS 8
-
-void WAIT_RING(nouveauContextPtr nmesa,uint32_t size)
-{
-#ifdef NOUVEAU_RING_DEBUG
-	return;
-#endif
-	uint32_t fifo_get;
-	while(nmesa->fifo.free < size+1) {
-		fifo_get = NV_FIFO_READ_GET();
-
-		if(nmesa->fifo.put >= fifo_get) {
-			nmesa->fifo.free = nmesa->fifo.max - nmesa->fifo.current;
-			if(nmesa->fifo.free < size+1) {
-				OUT_RING(NV03_FIFO_CMD_JUMP | nmesa->fifo.put_base);
-				if(fifo_get <= RING_SKIPS) {
-					if(nmesa->fifo.put <= RING_SKIPS) /* corner case - will be idle */
-						NV_FIFO_WRITE_PUT(RING_SKIPS + 1);
-					do { fifo_get = NV_FIFO_READ_GET(); }
-					while(fifo_get <= RING_SKIPS);
-				}
-				NV_FIFO_WRITE_PUT(RING_SKIPS);
-				nmesa->fifo.current = nmesa->fifo.put = RING_SKIPS;
-				nmesa->fifo.free = fifo_get - (RING_SKIPS + 1);
-			}
-		} else 
-			nmesa->fifo.free = fifo_get - nmesa->fifo.current - 1;
-	}
-}
-
-/* 
- * Wait for the channel to be idle 
- */
-void nouveauWaitForIdleLocked(nouveauContextPtr nmesa)
-{
-	/* Wait for FIFO idle */
-	FIRE_RING();
-	while(RING_AHEAD()>0);
-
-	/* Wait on notifier to indicate all commands in the channel have
-	 * been completed.
-	 */
-	nouveau_notifier_wait_nop(nmesa->glCtx, nmesa->syncNotifier, NvSub3D);
-}
-
-void nouveauWaitForIdle(nouveauContextPtr nmesa)
-{
-	LOCK_HARDWARE(nmesa);
-	nouveauWaitForIdleLocked(nmesa);
-	UNLOCK_HARDWARE(nmesa);
-}
-
-// here we call the fifo initialization ioctl and fill in stuff accordingly
-GLboolean nouveauFifoInit(nouveauContextPtr nmesa)
-{
-	struct drm_nouveau_fifo_alloc fifo_init;
-	int i, ret;
-
-#ifdef NOUVEAU_RING_DEBUG
-	return GL_TRUE;
-#endif
-
-	fifo_init.fb_ctxdma_handle = NvDmaFB;
-	fifo_init.tt_ctxdma_handle = NvDmaTT;
-	ret=drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_FIFO_ALLOC, &fifo_init, sizeof(fifo_init));
-	if (ret) {
-		FATAL("Fifo initialization ioctl failed (returned %d)\n",ret);
-		return GL_FALSE;
-	}
-
-	ret = drmMap(nmesa->driFd, fifo_init.cmdbuf, fifo_init.cmdbuf_size, &nmesa->fifo.buffer);
-	if (ret) {
-		FATAL("Unable to map the fifo (returned %d)\n",ret);
-		return GL_FALSE;
-	}
-
-	ret = drmMap(nmesa->driFd, fifo_init.ctrl, fifo_init.ctrl_size, &nmesa->fifo.mmio);
-	if (ret) {
-		FATAL("Unable to map the control regs (returned %d)\n",ret);
-		return GL_FALSE;
-	}
-
-	ret = drmMap(nmesa->driFd, fifo_init.notifier,
-				   fifo_init.notifier_size,
-				   &nmesa->notifier_block);
-	if (ret) {
-		FATAL("Unable to map the notifier block (returned %d)\n",ret);
-		return GL_FALSE;
-	}
-
-	/* Setup our initial FIFO tracking params */
-	nmesa->fifo.channel  = fifo_init.channel;
-	nmesa->fifo.put_base = fifo_init.put_base;
-	nmesa->fifo.current  = 0;
-	nmesa->fifo.put      = 0;
-	nmesa->fifo.max      = (fifo_init.cmdbuf_size >> 2) - 1;
-	nmesa->fifo.free     = nmesa->fifo.max - nmesa->fifo.current;
-
-	for (i=0; i<RING_SKIPS; i++)
-	   OUT_RING(0);
-	nmesa->fifo.free -= RING_SKIPS;
-
-	MESSAGE("Fifo init ok. Using context %d\n", fifo_init.channel);
-	return GL_TRUE;
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
deleted file mode 100644
index 67f9cd4fc89..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
+++ /dev/null
@@ -1,195 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NOUVEAU_FIFO_H__
-#define __NOUVEAU_FIFO_H__
-
-#include "nouveau_context.h"
-#include "nouveau_ctrlreg.h"
-#include "nouveau_state_cache.h"
-
-//#define NOUVEAU_RING_TRACE
-//#define NOUVEAU_RING_DEBUG
-//#define NOUVEAU_STATE_CACHE_DISABLE
-
-#ifndef NOUVEAU_RING_TRACE
-#define NOUVEAU_RING_TRACE 0
-#else
-#undef NOUVEAU_RING_TRACE
-#define NOUVEAU_RING_TRACE 1
-#endif
-
-#define NV_READ(reg) *(volatile uint32_t *)(nmesa->mmio + (reg))
-
-#define NV_FIFO_READ(reg) *(volatile uint32_t *)(nmesa->fifo.mmio + (reg/4))
-#define NV_FIFO_WRITE(reg,value) *(volatile uint32_t *)(nmesa->fifo.mmio + (reg/4)) = value;
-#define NV_FIFO_READ_GET() ((NV_FIFO_READ(NV03_FIFO_REGS_DMAGET) - nmesa->fifo.put_base) >> 2)
-#define NV_FIFO_WRITE_PUT(val) do { \
-	if (NOUVEAU_RING_TRACE) {\
-		printf("FIRE_RING : 0x%08x\n", nmesa->fifo.current << 2); \
-		fflush(stdout); \
-		sleep(1); \
-	} \
-	NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.put_base); \
-} while(0)
-
-/* 
- * Ring/fifo interface
- *
- * - Begin a ring section with BEGIN_RING_SIZE (if you know the full size in advance)
- * - Output stuff to the ring with either OUT_RINGp (outputs a raw mem chunk), OUT_RING (1 uint32_t) or OUT_RINGf (1 float)
- * - RING_AVAILABLE returns the available fifo (in uint32_ts)
- * - RING_AHEAD returns how much ahead of the last submission point we are
- * - FIRE_RING fires whatever we have that wasn't fired before
- * - WAIT_RING waits for size (in uint32_ts) to be available in the fifo
- */
-
-/* Enable for ring debugging.  Prints out writes to the ring buffer
- * but does not actually write to it.
- */
-#ifdef NOUVEAU_RING_DEBUG
-
-extern int nouveau_fifo_remaining;
-
-#define OUT_RINGp(ptr,sz) do {                                                  \
-uint32_t* p=(uint32_t*)(ptr);							\
-int i; printf("OUT_RINGp: (size 0x%x dwords)\n",sz); for(i=0;i<sz;i++) printf(" 0x%08x   %f\n", *(p+i), *((float*)(p+i))); 	\
-nouveau_fifo_remaining-=sz;							\
-}while(0)
-
-#define OUT_RING(n) do {                                                        \
-    printf("OUT_RINGn: 0x%08x (%s)\n", n, __func__);                            \
-    nouveau_fifo_remaining--;							\
-}while(0)
-
-#define OUT_RINGf(n) do {                                                       \
-    printf("OUT_RINGf: %.04f (%s)\n", n, __func__);                             \
-    nouveau_fifo_remaining--;							\
-}while(0)
-
-#define BEGIN_RING_SIZE(subchannel,tag,size) do {					\
-	if (nouveau_fifo_remaining!=0)							\
-		printf("RING ERROR : remaining %d\n",nouveau_fifo_remaining);		\
-	nouveau_state_cache_flush(nmesa);						\
-	if (nmesa->fifo.free <= (size))							\
-		WAIT_RING(nmesa,(size));						\
-	OUT_RING( ((size)<<18) | ((subchannel) << 13) | (tag));				\
-	nmesa->fifo.free -= ((size) + 1);                                               \
-	nouveau_fifo_remaining=size;							\
-}while(0)
-
-#else
-
-#define OUT_RINGp(ptr,sz) do{							\
-	if (NOUVEAU_RING_TRACE) { \
-		uint32_t* p=(uint32_t*)(ptr);							\
-		int i; printf("OUT_RINGp: (size 0x%x dwords) (%s)\n",sz, __func__); for(i=0;i<sz;i++) printf(" [0x%08x] 0x%08x   %f\n", (nmesa->fifo.current+i) << 2, *(p+i), *((float*)(p+i))); 	\
-	} \
-	memcpy(nmesa->fifo.buffer+nmesa->fifo.current,ptr,(sz)*4);		\
-	nmesa->fifo.current+=(sz);						\
-}while(0)
-
-#define OUT_RING(n) do {							\
-if (NOUVEAU_RING_TRACE) \
-    printf("OUT_RINGn: [0x%08x] 0x%08x (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-nmesa->fifo.buffer[nmesa->fifo.current++]=(n);					\
-}while(0)
-
-#define OUT_RINGf(n) do {							\
-if (NOUVEAU_RING_TRACE) \
-    printf("OUT_RINGf: [0x%08x] %.04f (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-*((float*)(nmesa->fifo.buffer+nmesa->fifo.current++))=(n);			\
-}while(0)
-
-#define BEGIN_RING_SIZE(subchannel,tag,size) do {					\
-	nouveau_state_cache_flush(nmesa);						\
-	if (nmesa->fifo.free <= (size))							\
-		WAIT_RING(nmesa,(size));						\
-	OUT_RING( ((size)<<18) | ((subchannel) << 13) | (tag));				\
-	nmesa->fifo.free -= ((size) + 1);                                               \
-}while(0)
-
-#endif
-
-extern void WAIT_RING(nouveauContextPtr nmesa,uint32_t size);
-extern void nouveau_state_cache_flush(nouveauContextPtr nmesa);
-extern void nouveau_state_cache_init(nouveauContextPtr nmesa);
-
-#ifdef NOUVEAU_STATE_CACHE_DISABLE
-#define BEGIN_RING_CACHE(subc,tag,size) BEGIN_RING_SIZE((subc), (tag), (size))
-#define OUT_RING_CACHE(n) OUT_RING((n))
-#define OUT_RING_CACHEf(n) OUT_RINGf((n))
-#define OUT_RING_CACHEp(ptr, sz) OUT_RINGp((ptr), (sz))
-#else
-#define BEGIN_RING_CACHE(subchannel,tag,size) do {					\
-	nmesa->state_cache.dirty=1;	 						\
-	nmesa->state_cache.current_pos=((tag)/4);					\
-}while(0)
-
-#define OUT_RING_CACHE(n) do {									\
-	if (nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value!=(n))	{	\
-		nmesa->state_cache.atoms[nmesa->state_cache.current_pos].dirty=1; 		\
-		nmesa->state_cache.hdirty[nmesa->state_cache.current_pos/NOUVEAU_STATE_CACHE_HIER_SIZE]=1; 		\
-		nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value=(n);		\
-	}											\
-	nmesa->state_cache.current_pos++;							\
-}while(0)
-
-#define OUT_RING_CACHEf(n) do {									\
-	if ((*(float*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))!=(n)){	\
-		nmesa->state_cache.atoms[nmesa->state_cache.current_pos].dirty=1;	 	\
-		nmesa->state_cache.hdirty[nmesa->state_cache.current_pos/NOUVEAU_STATE_CACHE_HIER_SIZE]=1; 		\
-		(*(float*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))=(n);\
-	}											\
-	nmesa->state_cache.current_pos++;							\
-}while(0)
-
-#define OUT_RING_CACHEp(ptr,sz) do {							\
-uint32_t* p=(uint32_t*)(ptr);								\
-int i; for(i=0;i<sz;i++) OUT_RING_CACHE(*(p+i)); 					\
-}while(0)
-#endif
-
-#define RING_AVAILABLE() (nmesa->fifo.free-1)
-
-#define RING_AHEAD() ((nmesa->fifo.put<=nmesa->fifo.current)?(nmesa->fifo.current-nmesa->fifo.put):nmesa->fifo.max-nmesa->fifo.put+nmesa->fifo.current)
-
-#define FIRE_RING() do {                             \
-	if (nmesa->fifo.current!=nmesa->fifo.put) {  \
-		nmesa->fifo.put=nmesa->fifo.current; \
-		NV_FIFO_WRITE_PUT(nmesa->fifo.put);  \
-	}                                            \
-}while(0)
-
-extern void nouveauWaitForIdle(nouveauContextPtr nmesa);
-extern void nouveauWaitForIdleLocked(nouveauContextPtr nmesa);
-extern GLboolean nouveauFifoInit(nouveauContextPtr nmesa);
-
-#endif /* __NOUVEAU_FIFO_H__ */
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_lock.c b/src/mesa/drivers/dri/nouveau/nouveau_lock.c
deleted file mode 100644
index aa86c9e7838..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_lock.c
+++ /dev/null
@@ -1,81 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#include "nouveau_context.h"
-#include "nouveau_lock.h"
-
-#include "drirenderbuffer.h"
-#include "framebuffer.h"
-
-
-/* Update the hardware state.  This is called if another context has
- * grabbed the hardware lock, which includes the X server.  This
- * function also updates the driver's window state after the X server
- * moves, resizes or restacks a window -- the change will be reflected
- * in the drawable position and clip rects.  Since the X server grabs
- * the hardware lock when it changes the window state, this routine will
- * automatically be called after such a change.
- */
-void nouveauGetLock( nouveauContextPtr nmesa, GLuint flags )
-{
-   __DRIdrawablePrivate *dPriv = nmesa->driDrawable;
-   __DRIscreenPrivate *sPriv = nmesa->driScreen;
-   struct drm_nouveau_sarea *sarea = nmesa->sarea;
-
-   drmGetLock( nmesa->driFd, nmesa->hHWContext, flags );
-
-   /* The window might have moved, so we might need to get new clip
-    * rects.
-    *
-    * NOTE: This releases and regrabs the hw lock to allow the X server
-    * to respond to the DRI protocol request for new drawable info.
-    * Since the hardware state depends on having the latest drawable
-    * clip rects, all state checking must be done _after_ this call.
-    */
-   DRI_VALIDATE_DRAWABLE_INFO( sPriv, dPriv );
-
-   /* If timestamps don't match, the window has been changed */
-   if (nmesa->lastStamp != dPriv->lastStamp) {
-      struct gl_framebuffer *fb = (struct gl_framebuffer *)dPriv->driverPrivate;
-
-      /* _mesa_resize_framebuffer will take care of calling the renderbuffer's
-       * AllocStorage function if we need more memory to hold it */
-      if (fb->Width != dPriv->w || fb->Height != dPriv->h) {
-	 _mesa_resize_framebuffer(nmesa->glCtx, fb, dPriv->w, dPriv->h);
-	 /* resize buffers, will call nouveau_window_moved */
-	 nouveau_build_framebuffer(nmesa->glCtx, fb);
-      } else {
-	 nouveau_window_moved(nmesa->glCtx);
-      }
-
-      nmesa->lastStamp = dPriv->lastStamp;
-   }
-
-   nmesa->numClipRects = dPriv->numClipRects;
-   nmesa->pClipRects = dPriv->pClipRects;
-
-}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_lock.h b/src/mesa/drivers/dri/nouveau/nouveau_lock.h
deleted file mode 100644
index 38bb0014258..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_lock.h
+++ /dev/null
@@ -1,69 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef __NOUVEAU_LOCK_H__
-#define __NOUVEAU_LOCK_H__
-
-#include "nouveau_context.h"
-
-extern void nouveauGetLock( nouveauContextPtr nmesa, GLuint flags );
-
-/*
- * !!! We may want to separate locks from locks with validation.  This
- * could be used to improve performance for those things commands that
- * do not do any drawing !!!
- */
-
-/* Lock the hardware and validate our state.
- */
-#define LOCK_HARDWARE( nmesa )						\
-   do {									\
-      char __ret = 0;							\
-      DEBUG_CHECK_LOCK();						\
-      DRM_CAS( nmesa->driHwLock, nmesa->hHWContext,			\
-	       (DRM_LOCK_HELD | nmesa->hHWContext), __ret );		\
-      if ( __ret )							\
-	 nouveauGetLock( nmesa, 0 );					\
-      DEBUG_LOCK();							\
-   } while (0)
-
-/* Unlock the hardware.
- */
-#define UNLOCK_HARDWARE( nmesa )					\
-   do {									\
-      DRM_UNLOCK( nmesa->driFd,						\
-		  nmesa->driHwLock,					\
-		  nmesa->hHWContext );					\
-      DEBUG_RESET();							\
-   } while (0)
-
-#define DEBUG_LOCK()
-#define DEBUG_RESET()
-#define DEBUG_CHECK_LOCK()
-
-
-#endif /* __NOUVEAU_LOCK_H__ */
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_msg.h b/src/mesa/drivers/dri/nouveau/nouveau_msg.h
deleted file mode 100644
index 5dea2189c72..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_msg.h
+++ /dev/null
@@ -1,67 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-Copyright 2006 Stephane Marchesin. All Rights Reserved
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <keith@tungstengraphics.com>
- *   Nicolai Haehnle <prefect_@gmx.net>
- */
-
-
-#ifndef __NOUVEAU_MSG_H__
-#define __NOUVEAU_MSG_H__
-
-#define WARN_ONCE(a, ...) do {\
-	        static int warn##__LINE__=1;\
-	        if(warn##__LINE__){\
-			                fprintf(stderr, "*********************************WARN_ONCE*********************************\n");\
-			                fprintf(stderr, "File %s function %s line %d\n", __FILE__, __FUNCTION__, __LINE__);\
-			                fprintf(stderr,  a, ## __VA_ARGS__);\
-			                fprintf(stderr, "***************************************************************************\n");\
-			                warn##__LINE__=0;\
-			                } \
-	        }while(0)
-
-#define MESSAGE(a, ...) do{\
-	                fprintf(stderr, "************************************INFO***********************************\n");\
-	                fprintf(stderr, "File %s function %s line %d\n", __FILE__, __FUNCTION__, __LINE__); \
-	                fprintf(stderr,  a, ## __VA_ARGS__);\
-	                fprintf(stderr, "***************************************************************************\n");\
-	        }while(0)
-
-#define FATAL(a, ...) do{\
-	                fprintf(stderr, "***********************************FATAL***********************************\n");\
-	                fprintf(stderr, "File %s function %s line %d\n", __FILE__, __FUNCTION__, __LINE__); \
-	                fprintf(stderr,  a, ## __VA_ARGS__);\
-	                fprintf(stderr, "***************************************************************************\n");\
-	        }while(0)
-
-#endif /* __NOUVEAU_MSG_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.c b/src/mesa/drivers/dri/nouveau/nouveau_object.c
deleted file mode 100644
index a143488e8d5..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.c
+++ /dev/null
@@ -1,67 +0,0 @@
-
-#include "nouveau_fifo.h"
-#include "nouveau_object.h"
-#include "nouveau_reg.h"
-
-
-GLboolean nouveauCreateContextObject(nouveauContextPtr nmesa,
-				     uint32_t handle, int class)
-{
-	struct drm_nouveau_grobj_alloc cto;
-	int ret;
-
-	cto.channel = nmesa->fifo.channel;
-	cto.handle  = handle;
-	cto.class   = class;
-	ret = drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_GROBJ_ALLOC,
-			      &cto, sizeof(cto));
-
-	return ret == 0;
-}
-
-void nouveauObjectOnSubchannel(nouveauContextPtr nmesa, int subchannel, int handle)
-{
-	BEGIN_RING_SIZE(subchannel, 0, 1);
-	OUT_RING(handle);
-}
-
-void nouveauObjectInit(nouveauContextPtr nmesa)
-{
-#ifdef NOUVEAU_RING_DEBUG
-	return;
-#endif
-
-	nouveauCreateContextObject(nmesa, Nv3D, nmesa->screen->card->class_3d);
-	if (nmesa->screen->card->type>=NV_10) {
-		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV10_CONTEXT_SURFACES_2D);
-	} else {
-		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D);
-		nouveauCreateContextObject(nmesa, NvCtxSurf3D, NV04_CONTEXT_SURFACES_3D);
-	}
-	if (nmesa->screen->card->type>=NV_11) {
-		nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT);
-	} else {
-		nouveauCreateContextObject(nmesa, NvImageBlit, NV_IMAGE_BLIT);
-	}
-	nouveauCreateContextObject(nmesa, NvMemFormat, NV_MEMORY_TO_MEMORY_FORMAT);
-
-#ifdef ALLOW_MULTI_SUBCHANNEL
-	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf2D, NvCtxSurf2D);
-	BEGIN_RING_SIZE(NvSubCtxSurf2D, NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY0, 2);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-
-	nouveauObjectOnSubchannel(nmesa, NvSubImageBlit, NvImageBlit);
-	BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_CONTEXT_SURFACES_2D, 1);
-	OUT_RING(NvCtxSurf2D);
-	BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_OPERATION, 1);
-	OUT_RING(3); /* SRCCOPY */
-
-	nouveauObjectOnSubchannel(nmesa, NvSubMemFormat, NvMemFormat);
-#endif
-
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-}
-
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.h b/src/mesa/drivers/dri/nouveau/nouveau_object.h
deleted file mode 100644
index 8c72d014daa..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.h
+++ /dev/null
@@ -1,35 +0,0 @@
-#ifndef __NOUVEAU_OBJECT_H__
-#define __NOUVEAU_OBJECT_H__
-
-#include "nouveau_context.h"
-
-#define ALLOW_MULTI_SUBCHANNEL
-
-void nouveauObjectInit(nouveauContextPtr nmesa);
-
-enum DMAObjects {
-	Nv3D                    = 0x80000019,
-	NvCtxSurf2D		= 0x80000020,
-	NvImageBlit		= 0x80000021,
-	NvMemFormat		= 0x80000022,
-	NvCtxSurf3D		= 0x80000023,
-	NvDmaFB			= 0xD0FB0001,
-	NvDmaTT			= 0xD0AA0001,
-	NvSyncNotify		= 0xD0000001,
-	NvQueryNotify		= 0xD0000002
-};
-
-enum DMASubchannel {
-	NvSubCtxSurf2D	= 0,
-	NvSubImageBlit	= 1,
-	NvSubMemFormat	= 2,
-	NvSubCtxSurf3D	= 3,
-	NvSub3D		= 7,
-};
-
-extern void nouveauObjectOnSubchannel(nouveauContextPtr nmesa, int subchannel, int handle);
-
-extern GLboolean nouveauCreateContextObject(nouveauContextPtr nmesa,
-      					    uint32_t handle, int class);
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.c b/src/mesa/drivers/dri/nouveau/nouveau_query.c
deleted file mode 100644
index 01541400690..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_query.c
+++ /dev/null
@@ -1,199 +0,0 @@
-/*
- * Copyright (C) 2007 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/* GL_ARB_occlusion_query support for NV20/30/40 */
-
-#include "mtypes.h"
-
-#include "nouveau_fifo.h"
-#include "nouveau_msg.h"
-#include "nouveau_object.h"
-#include "nouveau_reg.h"
-#include "nouveau_sync.h"
-#include "nouveau_query.h"
-
-static struct gl_query_object *
-nouveauNewQueryObject(GLcontext *ctx, GLuint id)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_query_object *nq;
-	int i;
-
-	for (i=0; i<nmesa->query_object_max; i++)
-		if (nmesa->query_alloc[i] == GL_FALSE)
-			break;
-	if (i==nmesa->query_object_max)
-		return NULL;
-
-	nq = CALLOC_STRUCT(nouveau_query_object_t);
-	if (nq) {
-		nq->notifier_id	= i;
-
-		nq->mesa.Id     = id;
-		nq->mesa.Result = 0;
-		nq->mesa.Active = GL_FALSE;
-		nq->mesa.Ready  = GL_TRUE;
-	}
-
-	return (struct gl_query_object *)nq;
-}
-
-static void
-nouveauBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_query_object *nq = (nouveau_query_object *)q;
-
-	nouveau_notifier_reset(ctx, nmesa->queryNotifier, nq->notifier_id);
-
-	switch (nmesa->screen->card->type) {
-	case NV_20:
-		BEGIN_RING_CACHE(NvSub3D, 0x17c8, 1);
-		OUT_RING_CACHE  (1);
-		BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
-		OUT_RING_CACHE  (1);
-		break;
-	case NV_30:
-	case NV_40:
-	case NV_44:
-		/* I don't think this is OCC_QUERY enable, but it *is* needed to make
-		 * the SET_OBJECT7 notifier block work with STORE_RESULT.
-		 *
-		 * Also, this appears to reset the pixel pass counter */
-		BEGIN_RING_SIZE(NvSub3D,
-				 NV30_TCL_PRIMITIVE_3D_OCC_QUERY_OR_COLOR_BUFF_ENABLE,
-				 1);
-		OUT_RING        (1);
-		/* Probably OCC_QUERY_ENABLE */
-		BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
-		OUT_RING_CACHE  (1);
-		break;
-	default:
-		WARN_ONCE("no support for this card\n");
-		break;
-	}
-}
-
-static void
-nouveauUpdateQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_query_object *nq = (nouveau_query_object *)q;
-	int status;
-
-	status = nouveau_notifier_status(ctx, nmesa->queryNotifier,
-					 nq->notifier_id);
-
-	q->Ready = (status == NV_NOTIFY_STATE_STATUS_COMPLETED);
-	if (q->Ready)
-		q->Result = nouveau_notifier_return_val(ctx,
-							nmesa->queryNotifier,
-							nq->notifier_id);
-}
-
-static void
-nouveauWaitQueryResult(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_query_object *nq = (nouveau_query_object *)q;
-
-	nouveau_notifier_wait_status(ctx, nmesa->queryNotifier, nq->notifier_id,
-				     NV_NOTIFY_STATE_STATUS_COMPLETED, 0);
-	nouveauUpdateQuery(ctx, target, q);
-}
-
-static void
-nouveauEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
-{
-	nouveau_query_object *nq = (nouveau_query_object *)q;
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	switch (nmesa->screen->card->type) {
-	case NV_20:
-		BEGIN_RING_SIZE(NvSub3D, 0x17d0, 1);
-		OUT_RING       (0x01000000 | nq->notifier_id*32);
-		break;
-	case NV_30:
-	case NV_40:
-	case NV_44:
-		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STORE_RESULT, 1);
-		OUT_RING       (0x01000000 | nq->notifier_id*32);
-		break;
-	default:
-		WARN_ONCE("no support for this card\n");
-		break;
-	}
-	FIRE_RING();
-
-	/*XXX: wait for query to complete, mesa doesn't give the driver
-	 *     an interface to query the status of a query object so 
-	 *     this has to stall the channel.
-	 */
-	nouveauWaitQueryResult(ctx, target, q);
-
-	BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
-	OUT_RING_CACHE  (0);
-}
-
-void
-nouveauQueryInitFuncs(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (nmesa->screen->card->type < NV_20)
-		return;
-
-	nmesa->query_object_max = (0x4000 / 32);
-	nmesa->queryNotifier =
-		nouveau_notifier_new(ctx, NvQueryNotify,
-					  nmesa->query_object_max);
-	nmesa->query_alloc = calloc(nmesa->query_object_max, sizeof(GLboolean));
-
-	switch (nmesa->screen->card->type) {
-	case NV_20:
-		BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
-		OUT_RING_CACHE  (NvQueryNotify);
-		break;
-	case NV_30:
-	case NV_40:
-	case NV_44:
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT7, 1);
-		OUT_RING_CACHE  (NvQueryNotify);
-		break;
-	default:
-		break;
-	};
-
-	ctx->Driver.NewQueryObject	= nouveauNewQueryObject;
-	ctx->Driver.BeginQuery		= nouveauBeginQuery;
-	ctx->Driver.EndQuery		= nouveauEndQuery;
-#if 0
-	ctx->Driver.UpdateQuery		= nouveauUpdateQuery;
-	ctx->Driver.WaitQueryResult	= nouveauWaitQueryResult;
-#endif
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_reg.h b/src/mesa/drivers/dri/nouveau/nouveau_reg.h
deleted file mode 100644
index 8758b538c85..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_reg.h
+++ /dev/null
@@ -1,1505 +0,0 @@
-/*
-   Autogenerated file, do not edit !
-
-**************************************************************************
-
-   Copyright (C) 2006 :
-   Dmitry Baryshkov,
-   Laurent Carlier,
-   Matthieu Castet,
-   Dawid Gajownik,
-   Jeremy Kolb,
-   Stephane Loeuillet,
-   Patrice Mandin,
-   Stephane Marchesin,
-   Serge Martin,
-   Sylvain Munaut,
-   Ben Skeggs,
-   Erik Waling,
-   koala_br,
-   sturmflut.
-
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************
-
-   Created from objects.c rev. 1.398
-*/
-
-#ifndef _NOUVEAU_REG_H
-#define _NOUVEAU_REG_H
-
-/****************************************** 
-Object NV01_CONTEXT_CLIP_RECTANGLE used on: NV03 NV04 NV10 NV15 NV20 NV40 G70
-*/
-#define			NV01_CONTEXT_CLIP_RECTANGLE				0x00000019
-#	define		NV01_CONTEXT_CLIP_RECTANGLE_SET_POINT			0x00000300	/* Parameters: x y */
-#	define		NV01_CONTEXT_CLIP_RECTANGLE_SET_SIZE			0x00000304	/* Parameters: width height */
-
-/****************************************** 
-Object NV_MEMORY_TO_MEMORY_FORMAT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV_MEMORY_TO_MEMORY_FORMAT				0x00000039
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_NOP				0x00000100
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY			0x00000104
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY			0x00000180
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN			0x00000184
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_OUT			0x00000188
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN			0x0000030c
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_OUT			0x00000310
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_PITCH_IN			0x00000314
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT			0x00000318
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN		0x0000031c
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT			0x00000320
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_FORMAT			0x00000324	/* Parameters: src_inc dst_inc */
-#	define		NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY			0x00000328
-
-/****************************************** 
-Object NV03_PRIMITIVE_RASTER_OP used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV03_PRIMITIVE_RASTER_OP				0x00000043
-#	define		NV03_PRIMITIVE_RASTER_OP_NOTIFY				0x00000100
-#	define		NV03_PRIMITIVE_RASTER_OP_DMA_NOTIFY			0x00000180
-#	define		NV03_PRIMITIVE_RASTER_OP_LOGIC_OP			0x00000300	/* Parameters: logic_op */
-
-/****************************************** 
-Object NV04_GDI_RECTANGLE_TEXT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV04_GDI_RECTANGLE_TEXT					0x0000004a
-#	define		NV04_GDI_RECTANGLE_TEXT_SET_DMA_NOTIFY			0x00000180
-#	define		NV04_GDI_RECTANGLE_TEXT_PATTERN				0x00000188
-#	define		NV04_GDI_RECTANGLE_TEXT_ROP5				0x0000018c
-#	define		NV04_GDI_RECTANGLE_TEXT_SURFACE				0x00000198
-#	define		NV04_GDI_RECTANGLE_TEXT_OPERATION			0x000002fc
-#	define		NV04_GDI_RECTANGLE_TEXT_FORMAT				0x00000300
-#	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL			0x000005f4	/* Parameters: left top */
-#	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_BR			0x000005f8	/* Parameters: right bottom */
-#	define		NV04_GDI_RECTANGLE_TEXT_FILL_VALUE			0x000005fc
-#	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_TL			0x00000600	/* Parameters: left top */
-#	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_BR			0x00000604	/* Parameters: right bottom */
-
-/****************************************** 
-Object NV04_SWIZZLED_SURFACE used on: NV04 NV10 NV15 
-*/
-#define			NV04_SWIZZLED_SURFACE					0x00000052
-#	define		NV04_SWIZZLED_SURFACE_DMA_NOTIFY			0x00000180
-#	define		NV04_SWIZZLED_SURFACE_DMA_IMAGE				0x00000184
-#	define		NV04_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2(height) log2(width) color */
-#	define		NV04_SWIZZLED_SURFACE_OFFSET				0x00000304
-
-/****************************************** 
-Object NV04_CONTEXT_SURFACES_3D used on: NV04 
-*/
-#define			NV04_CONTEXT_SURFACES_3D				0x00000053
-#	define		NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY			0x00000180
-#	define		NV04_CONTEXT_SURFACES_3D_DMA_COLOR			0x00000184
-#	define		NV04_CONTEXT_SURFACES_3D_DMA_ZETA			0x00000188
-#	define		NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL		0x000002f8	/* Parameters: x width */
-#	define		NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL			0x000002fc	/* Parameters: y height */
-#	define		NV04_CONTEXT_SURFACES_3D_FORMAT				0x00000300	/* Parameters: color type width height */
-#	define		NV04_CONTEXT_SURFACES_3D_CLIP_SIZE			0x00000304	/* Parameters: width height */
-#	define		NV04_CONTEXT_SURFACES_3D_PITCH				0x00000308	/* Parameters: color zeta */
-#	define		NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR			0x0000030c
-#	define		NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA			0x00000310
-
-/****************************************** 
-Object NV04_DX5_TEXTURED_TRIANGLE used on: NV04 
-*/
-#define			NV04_DX5_TEXTURED_TRIANGLE				0x00000054
-#	define		NV04_DX5_TEXTURED_TRIANGLE_NOP				0x00000100
-#	define		NV04_DX5_TEXTURED_TRIANGLE_NOTIFY			0x00000104
-#	define		NV04_DX5_TEXTURED_TRIANGLE_DMA_NOTIFY			0x00000180
-#	define		NV04_DX5_TEXTURED_TRIANGLE_DMA_1			0x00000184
-#	define		NV04_DX5_TEXTURED_TRIANGLE_DMA_2			0x00000188
-#	define		NV04_DX5_TEXTURED_TRIANGLE_SURFACE			0x0000018c
-#	define		NV04_DX5_TEXTURED_TRIANGLE_COLOR_KEY			0x00000300
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_OFFSET		0x00000304
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT		0x00000308	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER		0x0000030c	/* Parameters: magfilter minfilter lodbias */
-#	define		NV04_DX5_TEXTURED_TRIANGLE_BLEND			0x00000310	/* Parameters: texture benable dst src */
-#	define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL			0x00000314	/* Parameters: alpharef alphafunc alphaenable zenable zwrite zfunc cullmode */
-#	define		NV04_DX5_TEXTURED_TRIANGLE_FOG_COLOR			0x00000318
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_INV_W( d)			(0x0000040c + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_COLOR( d)			(0x00000410 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_SPECULAR( d)			(0x00000414 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_S( d)		(0x00000418 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_T( d)		(0x0000041c + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_DRAW				0x00000600	/* Parameters: v0 v1 v2 v3 v4 v5 */
-
-/****************************************** 
-Object NV04_DX6_MULTITEX_TRIANGLE used on: NV04 NV10 NV15 
-*/
-#define			NV04_DX6_MULTITEX_TRIANGLE				0x00000055
-#	define		NV04_DX6_MULTITEX_TRIANGLE_NOP				0x00000100
-#	define		NV04_DX6_MULTITEX_TRIANGLE_NOTIFY			0x00000104
-#	define		NV04_DX6_MULTITEX_TRIANGLE_DMA_NOTIFY			0x00000180
-#	define		NV04_DX6_MULTITEX_TRIANGLE_DMA_1			0x00000184
-#	define		NV04_DX6_MULTITEX_TRIANGLE_DMA_2			0x00000188
-#	define		NV04_DX6_MULTITEX_TRIANGLE_SURFACE			0x0000018c
-#	define		NV04_DX6_MULTITEX_TRIANGLE_OFFSET0			0x00000308
-#	define		NV04_DX6_MULTITEX_TRIANGLE_OFFSET1			0x0000030c
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0			0x00000310	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1			0x00000314	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0			0x00000318	/* Parameters: magfilter minfilter lodbias */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1			0x0000031c	/* Parameters: magfilter minfilter lodbias */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA		0x00000320
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR		0x00000324
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA		0x0000032c
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR		0x00000330
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR		0x00000334
-#	define		NV04_DX6_MULTITEX_TRIANGLE_BLEND			0x00000338	/* Parameters: benable dst src */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0			0x0000033c	/* Parameters: red_write green_write blue_write alpha_write alpha_write stencil_write alpharef alphafunc alphaenable zenable zwrite zfunc cullmode */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1			0x00000340	/* Parameters: stencil_enable stencil_mask_write stencil_mask_read stencilref stencilfunc */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2			0x00000344	/* Parameters: stencil_fail stencil_zfail stencil_zpass */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FOG_COLOR			0x00000348
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_INV_W( d)			(0x0000040c + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COLOR( d)			(0x00000410 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_SPECULAR( d)			(0x00000414 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_S( d)		(0x00000418 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_T( d)		(0x0000041c + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_S( d)		(0x00000420 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_T( d)		(0x00000424 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_DRAW				0x00000540	/* Parameters: v0 v1 v2 v3 v4 v5 */
-
-/****************************************** 
-Object NV04_COLOR_KEY used on: NV04 NV10 NV15 NV20 NV40 
-*/
-#define			NV04_COLOR_KEY						0x00000057
-#	define		NV04_COLOR_KEY_SET_DMA_NOTIFY				0x00000180
-#	define		NV04_COLOR_KEY_FORMAT					0x00000300
-#	define		NV04_COLOR_KEY_VALUE					0x00000304
-
-/****************************************** 
-Object NV04_SOLID_LINE used on: NV04 
-*/
-#define			NV04_SOLID_LINE						0x0000005c
-#	define		NV04_SOLID_LINE_CLIP_RECTANGLE				0x00000184
-#	define		NV04_SOLID_LINE_PATTERN					0x00000188
-#	define		NV04_SOLID_LINE_ROP					0x0000018c
-#	define		NV04_SOLID_LINE_SURFACE					0x00000198
-#	define		NV04_SOLID_LINE_OPERATION				0x000002fc
-#	define		NV04_SOLID_LINE_COLOR_FORMAT				0x00000300
-#	define		NV04_SOLID_LINE_COLOR_VALUE				0x00000304
-#	define		NV04_SOLID_LINE_START					0x00000400	/* Parameters: x y */
-#	define		NV04_SOLID_LINE_END					0x00000400	/* Parameters: x y */
-
-/****************************************** 
-Object NV04_UNK005E used on: NV04 
-*/
-#define			NV04_UNK005E						0x0000005e
-#	define		NV04_UNK005E_SET_SURFACE				0x00000198
-#	define		NV04_UNK005E_UNK02fc					0x000002fc
-#	define		NV04_UNK005E_UNK0300					0x00000300
-#	define		NV04_UNK005E_COUNTER					0x00000304
-
-/****************************************** 
-Object NV05_SCALED_IMAGE_FROM_MEMORY used on: NV04 
-*/
-#define			NV05_SCALED_IMAGE_FROM_MEMORY				0x00000063
-#	define		NV05_SCALED_IMAGE_FROM_MEMORY_SURFACE			0x00000198
-#	define		NV05_SCALED_IMAGE_FROM_MEMORY_COLOR_CONVERSION		0x000002fc
-#	define		NV05_SCALED_IMAGE_FROM_MEMORY_OPERATION			0x00000304
-
-/****************************************** 
-Object NV04_SCALED_IMAGE_FROM_MEMORY used on: NV04 
-*/
-#define			NV04_SCALED_IMAGE_FROM_MEMORY				0x00000077
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY		0x00000180
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE			0x00000184
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE			0x00000198
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT		0x00000300
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION			0x00000304
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POS			0x00000308	/* Parameters: x y */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE			0x0000030c	/* Parameters: width height */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POS			0x00000310	/* Parameters: x y */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE			0x00000314	/* Parameters: width height */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX			0x00000318	/* Parameters: int frac*0x100000 */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY			0x0000031c	/* Parameters: int frac*0x100000 */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_SIZE			0x00000400	/* Parameters: width height */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT			0x00000404	/* Parameters: pitch origin filter */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_OFFSET			0x00000408
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac*0x10 v_int v_frac*0x10 */
-
-/****************************************** 
-Object NV_IMAGE_FROM_CPU used on: NV04 
-*/
-#define			NV_IMAGE_FROM_CPU					0x00000061
-#	define		NV_IMAGE_FROM_CPU_DMA_NOTIFY				0x00000180
-#	define		NV_IMAGE_FROM_CPU_CLIP_RECTANGLE			0x00000188
-#	define		NV_IMAGE_FROM_CPU_PATTERN				0x0000018c
-#	define		NV_IMAGE_FROM_CPU_ROP					0x00000190
-#	define		NV_IMAGE_FROM_CPU_SURFACE				0x0000019c
-#	define		NV_IMAGE_FROM_CPU_OPERATION				0x000002fc
-#	define		NV_IMAGE_FROM_CPU_FORMAT				0x00000300
-
-/****************************************** 
-Object NV05_IMAGE_FROM_CPU used on: NV04 
-*/
-#define			NV05_IMAGE_FROM_CPU					0x00000065
-#	define		NV05_IMAGE_FROM_CPU_DMA_NOTIFY				0x00000180
-#	define		NV05_IMAGE_FROM_CPU_CLIP_RECTANGLE			0x00000188
-#	define		NV05_IMAGE_FROM_CPU_PATTERN				0x0000018c
-#	define		NV05_IMAGE_FROM_CPU_ROP					0x00000190
-#	define		NV05_IMAGE_FROM_CPU_SURFACE				0x0000019c
-#	define		NV05_IMAGE_FROM_CPU_OPERATION				0x000002fc
-#	define		NV05_IMAGE_FROM_CPU_FORMAT				0x00000300
-#	define		NV05_IMAGE_FROM_CPU_POINT				0x00000304	/* Parameters: x y */
-#	define		NV05_IMAGE_FROM_CPU_SIZE_OUT				0x00000308	/* Parameters: x y */
-#	define		NV05_IMAGE_FROM_CPU_SIZE_IN				0x0000030c	/* Parameters: x y */
-#	define		NV05_IMAGE_FROM_CPU_COLOR( d)				(0x00000400 + d * 0x0004)
-
-/****************************************** 
-Object NV_IMAGE_BLIT used on: NV04 NV10 NV15 NV20 NV40 
-*/
-#define			NV_IMAGE_BLIT						0x0000005f
-#	define		NV_IMAGE_BLIT_DMA_NOTIFY				0x00000180
-#	define		NV_IMAGE_BLIT_COLOR_KEY					0x00000184
-#	define		NV_IMAGE_BLIT_CLIP_RECTANGLE				0x00000188
-#	define		NV_IMAGE_BLIT_PATTERN					0x0000018c
-#	define		NV_IMAGE_BLIT_ROP5					0x00000190
-#	define		NV_IMAGE_BLIT_SURFACE					0x0000019c
-#	define		NV_IMAGE_BLIT_OPERATION					0x000002fc
-#	define		NV_IMAGE_BLIT_POINT_IN					0x00000300	/* Parameters: x y */
-#	define		NV_IMAGE_BLIT_POINT_OUT					0x00000304	/* Parameters: x y */
-#	define		NV_IMAGE_BLIT_SIZE					0x00000308	/* Parameters: width height */
-
-/****************************************** 
-Object NV10_TCL_PRIMITIVE_3D used on: NV10 
-*/
-#define			NV10_TCL_PRIMITIVE_3D					0x00000056
-
-/****************************************** 
-Object NV17_TCL_PRIMITIVE_3D used on: NV15 
-*/
-#define			NV17_TCL_PRIMITIVE_3D					0x00000099
-
-/****************************************** 
-Object NV11_TCL_PRIMITIVE_3D used on: NV15 
-*/
-#define			NV11_TCL_PRIMITIVE_3D					0x00000096
-#	define		NV10_TCL_PRIMITIVE_3D_NOP				0x00000100
-#	define		NV10_TCL_PRIMITIVE_3D_NOTIFY				0x00000104
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_NOTIFY			0x00000180
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY0		0x00000184
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY1		0x00000188
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DISPLAY_LIST			0x0000018c
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY2		0x00000194
-#	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY3		0x00000198
-#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY4		0x000001ac
-#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY5		0x000001b0
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ			0x00000200	/* Parameters: width x */
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT			0x00000204	/* Parameters: height y */
-#	define		NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT			0x00000208	/* Parameters: type color */
-#	define		NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color buffer pitch */
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET			0x00000210
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
-#	define		NV10_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00000218 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00000220 + d * 0x0004)	/* Parameters: wrap_t wrap_s log2(height) log2(width) lod npot format cube_map */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00000228 + d * 0x0004)	/* Parameters: enable anisotropy */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00000230 + d * 0x0004)	/* Parameters: pitch */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00000240 + d * 0x0004)	/* Parameters: width height */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00000248 + d * 0x0004)	/* Parameters: mag_filter min_filter */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00000250 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x000003e0 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x00000540 + y * 0x0010 + x * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000268 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000278 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000280 + d * 0x0004)	/* Parameters: rc1_tx_units_enabled rc1_rc_enabled scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0				0x00000270	/* Parameters: a r g b */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1				0x00000274	/* Parameters: a r g b */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0				0x00000288	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1				0x0000028c	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL			0x00000294	/* Parameters: local_viewer color_control */
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE		0x00000298	/* Parameters: specular diffuse ambient emission */
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_MODE				0x0000029c
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000002a0
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_ENABLE			0x000002a4
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR				0x000002a8	/* Parameters: a b g r */
-#	define		NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE			0x000002bc
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + d * 0x0004)	/* Parameters: x2 x1 */
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + d * 0x0004)	/* Parameters: y2 y1 */
-#	define		NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000300
-#	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000304
-#	define		NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00000308
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x0000030c
-#	define		NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE			0x00000310
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHTING_ENABLE			0x00000314
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE		0x00000318
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE		0x0000031c
-#	define		NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x00000320
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00000324
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE		0x00000328
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE			0x0000032c
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE	0x00000330
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE	0x00000334
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE	0x00000338
-#	define		NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x0000033c
-#	define		NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF			0x00000340
-#	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000344
-#	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000348
-#	define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000034c	/* Parameters: a r g b */
-#	define		NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000350
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x00000354
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000358	/* Parameters: a r g b */
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x0000035c
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_MASK			0x00000360
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC			0x00000364
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF			0x00000368
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK			0x0000036c
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL			0x00000370
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL			0x00000374
-#	define		NV10_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS			0x00000378
-#	define		NV10_TCL_PRIMITIVE_3D_SHADE_MODEL			0x0000037c
-#	define		NV10_TCL_PRIMITIVE_3D_LINE_WIDTH			0x00000380
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x00000384
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x00000388
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x0000038c
-#	define		NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x00000390
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000394
-#	define		NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000398
-#	define		NV10_TCL_PRIMITIVE_3D_CULL_FACE				0x0000039c
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_FACE			0x000003a0
-#	define		NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x000003a4
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R			0x000003a8
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G			0x000003ac
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B			0x000003b0
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A			0x000003b4
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL			0x000003b8	/* Parameters: color_control */
-#	define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x000003bc	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE( d)		(0x000003c0 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE		0x000003e8	/* Parameters: projection modelview0 modelview1 */
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_SIZE			0x000003ec
-#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX( d)		(0x00000400 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX( d)		(0x00000440 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX( d)	(0x00000480 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX( d)	(0x000004c0 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000500 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000600 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000604 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000608 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000060c + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x00000680
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x00000684
-#	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x00000688
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_A	0x000006a0
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_B	0x000006a4
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_C	0x000006a8
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_D	0x000006ac
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_E	0x000006b0
-#	define		NV10_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_F	0x000006b4
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000006c4
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000006c8
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000006cc
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X			0x000006e8
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Y			0x000006ec
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_Z			0x000006f0
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_W			0x000006f4
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x000006f8
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x000006fc
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00000700
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_D			0x00000704
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_E			0x00000708
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x0000070c
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00000710
-#	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00000714
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00000800 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00000804 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00000808 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000080c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00000810 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00000814 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00000818 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000081c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00000820 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00000828 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000082c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00000830 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00000834 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00000838 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000083c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00000840 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00000844 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00000848 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000084c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00000850 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00000854 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00000858 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000085c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00000860 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00000864 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00000868 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000086c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00000870 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00000c00
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00000c04
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00000c08
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X			0x00000c18
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y			0x00000c1c
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z			0x00000c20
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W			0x00000c24
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X			0x00000c30
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00000c34
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00000c38
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000c40	/* Parameters: y x */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000c44	/* Parameters: z */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00000c50
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00000c54
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00000c58
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A			0x00000c5c
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R			0x00000c60
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00000c64
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00000c68
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x00000c6c	/* Parameters: a b g r */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00000c80
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00000c84
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00000c88
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00000c8c	/* Parameters: a b g r */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x00000c90
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x00000c94
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00000c98	/* Parameters: t s */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x00000ca0
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x00000ca4
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x00000ca8
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x00000cac
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x00000cb0	/* Parameters: t s */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x00000cb4	/* Parameters: q r */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x00000cb8
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x00000cbc
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x00000cc0	/* Parameters: t s */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x00000cc8
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x00000ccc
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x00000cd0
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x00000cd4
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x00000cd8	/* Parameters: t s */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x00000cdc	/* Parameters: q r */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00000ce0
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F			0x00000ce4
-#	define		NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE			0x00000cec
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00000d04 + d * 0x0008)	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE		0x00000cf0
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS		0x00000d00
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS		0x00000d04	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL		0x00000d08
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL		0x00000d0c	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2		0x00000d10
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2		0x00000d14	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0		0x00000d18
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0		0x00000d1c	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1		0x00000d20
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1		0x00000d24	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR		0x00000d28
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR		0x00000d2c	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH		0x00000d30
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH		0x00000d34	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG		0x00000d38
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG		0x00000d3c	/* Parameters: stride fields type */
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x00000d40
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x00000d44
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x00000d5c	/* Parameters: pitch */
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000d60
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE		0x00000d68
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_CLEAR_ENABLE		0x00000d6c
-#	define		NV10_TCL_PRIMITIVE_3D_BEGIN_END				0x00000dfc
-#	define		NV10_TCL_PRIMITIVE_3D_INDEX_DATA			0x00000e00	/* Parameters: index1 index0 */
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END		0x000013fc
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS		0x00001400	/* Parameters: count-1 first */
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X			0x00001638
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Y			0x0000163c
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Z			0x00001640
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_W			0x00001644
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE			0x00001658
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA			0x00001800
-
-/****************************************** 
-Object NV10_IMAGE_FROM_CPU used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_IMAGE_FROM_CPU					0x0000008a
-#	define		NV10_IMAGE_FROM_CPU_SET_DMA_NOTIFY			0x00000180
-#	define		NV10_IMAGE_FROM_CPU_SET_CONTEXT_CLIP_RECTANGLE		0x00000188
-#	define		NV10_IMAGE_FROM_CPU_SET_IMAGE_PATTERN			0x0000018c
-#	define		NV10_IMAGE_FROM_CPU_SET_RASTER_OP			0x00000190
-#	define		NV10_IMAGE_FROM_CPU_SET_CONTEXT_SURFACES_2D		0x0000019c
-#	define		NV10_IMAGE_FROM_CPU_OPERATION				0x000002fc
-#	define		NV10_IMAGE_FROM_CPU_FORMAT				0x00000300
-#	define		NV10_IMAGE_FROM_CPU_POINT				0x00000304	/* Parameters: x y */
-#	define		NV10_IMAGE_FROM_CPU_SIZE_OUT				0x00000308	/* Parameters: width height */
-#	define		NV10_IMAGE_FROM_CPU_SIZE_IN				0x0000030c	/* Parameters: width height */
-#	define		NV10_IMAGE_FROM_CPU_HLINE				0x00000400
-
-/****************************************** 
-Object NV10_PRIMITIVE_2D used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_PRIMITIVE_2D					0x0000007b
-#	define		NV10_PRIMITIVE_2D_SET_DMA_NOTIFY			0x00000180
-#	define		NV10_PRIMITIVE_2D_SET_SURFACE				0x00000184
-#	define		NV10_PRIMITIVE_2D_SET_FORMAT				0x00000300
-#	define		NV10_PRIMITIVE_2D_SET_POINT				0x00000304	/* Parameters: x y */
-#	define		NV10_PRIMITIVE_2D_SET_SIZE				0x00000308	/* Parameters: width height */
-#	define		NV10_PRIMITIVE_2D_SET_CLIP_HORIZ			0x0000030c	/* Parameters: width x */
-#	define		NV10_PRIMITIVE_2D_SET_CLIP_VERT				0x00000310	/* Parameters: height y */
-#	define		NV10_PRIMITIVE_2D_SET_DATA( d)				(0x00000400 + d * 0x0004)
-
-/****************************************** 
-Object NV10_IMAGE_BLIT used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_IMAGE_BLIT						0x0000009f
-#	define		NV10_IMAGE_BLIT_NOP					0x00000100
-#	define		NV10_IMAGE_BLIT_NOTIFY					0x00000104
-#	define		NV10_IMAGE_BLIT_SET_DMA_NOTIFY				0x00000180
-#	define		NV10_IMAGE_BLIT_SET_CONTEXT_CLIP_RECTANGLE		0x00000188
-#	define		NV10_IMAGE_BLIT_SET_IMAGE_PATTERN			0x0000018c
-#	define		NV10_IMAGE_BLIT_SET_RASTER_OP				0x00000190
-#	define		NV10_IMAGE_BLIT_SET_CONTEXT_SURFACES_2D			0x0000019c
-#	define		NV10_IMAGE_BLIT_SET_OPERATION				0x000002fc
-#	define		NV10_IMAGE_BLIT_SET_POINT				0x00000300	/* Parameters: x y */
-#	define		NV10_IMAGE_BLIT_SET_PITCH				0x00000304	/* Parameters: skip */
-#	define		NV10_IMAGE_BLIT_SET_SIZE				0x00000308	/* Parameters: width height */
-
-/****************************************** 
-Object NV10_VIDEO_DISPLAY used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_VIDEO_DISPLAY					0x0000007c
-#	define		NV10_VIDEO_DISPLAY_COUNTER				0x00000050
-#	define		NV10_VIDEO_DISPLAY_SET_DMA_FROM_MEMORY			0x00000180
-#	define		NV10_VIDEO_DISPLAY_SET_DMA_IN_MEMORY0			0x00000184
-#	define		NV10_VIDEO_DISPLAY_SET_DMA_IN_MEMORY1			0x00000188
-#	define		NV10_VIDEO_DISPLAY_SET_OBJECT3				0x0000019c
-#	define		NV10_VIDEO_DISPLAY_SIZE					0x000002f8	/* Parameters: height width */
-#	define		NV10_VIDEO_DISPLAY_OFFSET				0x00000300
-
-/****************************************** 
-Object NV10_UNK0072 used on: NV10 NV15 NV20 NV40 G70
-*/
-#define			NV10_UNK0072						0x00000072
-#	define		NV10_UNK0072_COUNTER					0x00000050
-#	define		NV40_UNK0072_SET_OBJECT					0x00000060
-#	define		NV10_UNK0072_SET_DMA_NOTIFY				0x00000180
-
-/****************************************** 
-Object NV10_SCALED_IMAGE_FROM_MEMORY used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_SCALED_IMAGE_FROM_MEMORY				0x00000089
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_COUNTER			0x00000050
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_SET_DMA_IN_MEMORY		0x00000184
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_SET_RASTER_OP		0x0000018c
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_SET_IMAGE_PATTERN		0x00000188
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_SET_SURFACE		0x00000198
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_OPERATION			0x00000304
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_POS			0x00000308	/* Parameters: x y */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE			0x0000030c	/* Parameters: width height */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_POS			0x00000310	/* Parameters: x y */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE			0x00000314	/* Parameters: width height */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_SIZE			0x00000400	/* Parameters: width height */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_FORMAT			0x00000404	/* Parameters: pitch */
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_OFFSET			0x00000408
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac*0x10 v_int v_frac*0x10 */
-
-/****************************************** 
-Object NV10_CONTEXT_SURFACES_2D used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_CONTEXT_SURFACES_2D				0x00000062
-#	define		NV10_CONTEXT_SURFACES_2D_SET_DMA_NOTIFY			0x00000180
-#	define		NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY0		0x00000184
-#	define		NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY1		0x00000188
-#	define		NV10_CONTEXT_SURFACES_2D_FORMAT				0x00000300	/* Parameters: color type width height */
-#	define		NV10_CONTEXT_SURFACES_2D_PITCH				0x00000304	/* Parameters: src dst */
-#	define		NV10_CONTEXT_SURFACES_2D_OFFSET_SRC			0x00000308
-#	define		NV10_CONTEXT_SURFACES_2D_OFFSET_DST			0x0000030c
-
-/****************************************** 
-Object NV04_CONTEXT_SURFACES_2D used on: NV04 NV10 NV15 
-*/
-#define			NV04_CONTEXT_SURFACES_2D				0x00000042
-#	define		NV04_CONTEXT_SURFACES_2D_NOTIFY				0x00000104
-#	define		NV04_CONTEXT_SURFACES_2D_SET_DMA_NOTIFY			0x00000180
-#	define		NV04_CONTEXT_SURFACES_2D_SET_DMA_IMAGE_SRC		0x00000184
-#	define		NV04_CONTEXT_SURFACES_2D_SET_DMA_IMAGE_DST		0x00000188
-#	define		NV04_CONTEXT_SURFACES_2D_FORMAT				0x00000300
-#	define		NV04_CONTEXT_SURFACES_2D_PITCH				0x00000304	/* Parameters: src dst */
-#	define		NV04_CONTEXT_SURFACES_2D_OFFSET_SRC			0x00000308
-#	define		NV04_CONTEXT_SURFACES_2D_OFFSET_DST			0x0000030c
-
-/****************************************** 
-Object NV04_IMAGE_PATTERN used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV04_IMAGE_PATTERN					0x00000044
-#	define		NV04_IMAGE_PATTERN_COLOR_FORMAT				0x00000300
-#	define		NV04_IMAGE_PATTERN_MONO_FORMAT				0x00000304
-#	define		NV04_IMAGE_PATTERN_SELECT				0x0000030c
-#	define		NV04_IMAGE_PATTERN_MONOCHROME_SHAPE			0x00000308
-#	define		NV04_IMAGE_PATTERN_MONOCHROME_COLOR0			0x00000310
-#	define		NV04_IMAGE_PATTERN_MONOCHROME_COLOR1			0x00000314
-#	define		NV04_IMAGE_PATTERN_MONOCHROME_PATTERN0			0x00000318
-#	define		NV04_IMAGE_PATTERN_MONOCHROME_PATTERN1			0x0000031c
-
-/****************************************** 
-Object NV20_SWIZZLED_SURFACE used on: NV20 NV30 NV40 G70
-*/
-#define			NV20_SWIZZLED_SURFACE					0x0000009e
-#	define		NV20_SWIZZLED_SURFACE_SET_OBJECT0			0x00000180
-#	define		NV20_SWIZZLED_SURFACE_SET_OBJECT1			0x00000184
-#	define		NV20_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2(height) log2(width) color */
-#	define		NV20_SWIZZLED_SURFACE_OFFSET				0x00000304
-
-/****************************************** 
-Object NV20_TCL_PRIMITIVE_3D used on: NV20 
-*/
-#define			NV20_TCL_PRIMITIVE_3D					0x00000097
-#	define		NV20_TCL_PRIMITIVE_3D_NOP				0x00000100
-#	define		NV20_TCL_PRIMITIVE_3D_NOTIFY				0x00000104
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT0			0x00000180
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT1			0x00000184
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT2			0x00000188
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT3			0x00000194
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT4			0x00000198
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT5			0x0000019c
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT6			0x000001a0
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT7			0x000001a4
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT8			0x000001a8
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT9			0x000001ac
-#	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT10			0x000001b0
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ			0x00000200	/* Parameters: width x */
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT			0x00000204	/* Parameters: height y */
-#	define		NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT			0x00000208	/* Parameters: type color */
-#	define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color buffer pitch */
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET			0x00000210
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
-#	define		NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x0000022c	/* Parameters: pitch */
-#	define		NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000230
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL			0x00000294
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL		0x00000298	/* Parameters: back_specular back_ambient back_diffuse back_emission front_specular front_ambient front_diffuse front_emission */
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_MODE				0x0000029c
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000002a0
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_ENABLE			0x000002a4
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + d * 0x0004)	/* Parameters: x2 x1 */
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + d * 0x0004)	/* Parameters: y2 y1 */
-#	define		NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000300
-#	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000304
-#	define		NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00000308
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x0000030c
-#	define		NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE			0x00000310
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHTING_ENABLE			0x00000314
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_SMOOTH_ENABLE		0x0000031c
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE		0x00000318
-#	define		NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x00000320
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00000324
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE			0x0000032c
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE	0x00000330
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE	0x00000334
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE	0x00000338
-#	define		NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x0000033c
-#	define		NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF			0x00000340
-#	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000344
-#	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000348
-#	define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000034c	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000350
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x00000354
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000358	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x0000035c
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_MASK			0x00000360
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC			0x00000364
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_REF			0x00000368
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_MASK			0x0000036c
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL			0x00000370
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZFAIL			0x00000374
-#	define		NV20_TCL_PRIMITIVE_3D_STENCIL_OP_ZPASS			0x00000378
-#	define		NV20_TCL_PRIMITIVE_3D_SHADE_MODEL			0x0000037c
-#	define		NV20_TCL_PRIMITIVE_3D_LINE_WIDTH			0x00000380
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x00000384
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x00000388
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x0000038c
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x00000390
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000394
-#	define		NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000398
-#	define		NV20_TCL_PRIMITIVE_3D_CULL_FACE				0x0000039c
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_FACE			0x000003a0
-#	define		NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x000003a4
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R		0x000003a8
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G		0x000003ac
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B		0x000003b0
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A		0x000003b4
-#	define		NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE		0x000003b8
-#	define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x000003bc	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x000003c0 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000420 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_SIZE			0x0000043c
-#	define		NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + y * 0x0010 + x * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000840 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000844 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000848 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000084c + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x000009c0
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x000009c4
-#	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x000009c8
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_A	0x000009e0
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_B	0x000009e4
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_C	0x000009e8
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_D	0x000009ec
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_E	0x000009f0
-#	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_F	0x000009f4
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE			0x00000a1c	/* Parameters: coord_replace r_mode enable */
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_OX			0x00000a20
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_OY			0x00000a24
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_DEPTH_AVG_S		0x00000a28
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_UNKNOWN_A		0x00000a2c
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x00000a30
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x00000a34
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00000a38
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_D			0x00000a3c
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_E			0x00000a40
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x00000a44
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00000a48
-#	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00000a4c
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_PX_DIV2			0x00000af0
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_PY_DIV2			0x00000af4
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_DEPTH_HALF_S		0x00000af8
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_UNKNOWN_B		0x00000afc
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0			0x00000b00
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST1			0x00000b04
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST2			0x00000b08
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST3			0x00000b0c
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_X			0x00000b80
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Y			0x00000b84
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Z			0x00000b88
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_W			0x00000b8c
-#	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID		0x00001ea4
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x00000a10
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x00000a14
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x00000a18
-#	define		NV20_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00001b00 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00001b04 + d * 0x0040)	/* Parameters: log2(height) log2(width) lod format cube_map */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_WRAP(d)			(0x00001b08 + d * 0x0040)	/* Parameters: wrap_s wrap_t wrap_r */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00001b0c + d * 0x0040)	/* Parameters: enable anisotropy */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00001b10 + d * 0x0040)	/* Parameters: pitch */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00001b14 + d * 0x0040)	/* Parameters: mag_filter min_filter */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00001b1c + d * 0x0040)	/* Parameters: width height */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00001b20 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_RC_ENABLE				0x00001e60	/* Parameters: number of rc enabled */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP			0x00001e70	/* Parameters: op0 op1 op2 op3 */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE		0x000017f8	/* Parameters: cull0 cull1 cull2 cull3 */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS		0x00001e78	/* Parameters: prev2 prev3 */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0				0x00001e20	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1				0x00001e24	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0				0x00000288	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1				0x0000028c	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000ac0 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000a60 + d * 0x0004)	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x00000a80 + d * 0x0004)	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000aa0 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00001e40 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000105c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001060 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001064 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_AMBIENT(d)	(0x00000c00 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_DIFFUSE(d)	(0x00000c0c + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_SPECULAR(d)	(0x00000c18 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001068 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000106c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001070 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001040 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001044 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001048 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000104c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001050 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001054 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001058 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_A		0x00001e28
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_B		0x00001e2c
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_C		0x00001e30
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_D		0x00001e34
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_E		0x00001e38
-#	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_F		0x00001e3c
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000147c
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(d)	(0x00001480 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00001500
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00001504
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00001508
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X			0x00001518
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y			0x0000151c
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z			0x00001520
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W			0x00001524
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY			0x00001528	/* Parameters: y x */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW			0x0000152c	/* Parameters: w z */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X			0x00001530
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00001534
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00001538
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00001540	/* Parameters: y x */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00001544	/* Parameters: z */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00001550
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00001554
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00001558
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A			0x0000155c
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R			0x00001560
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00001564
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00001568
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000156c	/* Parameters: a b g r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00001580
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00001584
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00001588
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x0000158c	/* Parameters: a b g r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x00001590
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x00001594
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00001598	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x000015a0
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x000015a4
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x000015a8
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x000015ac
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x000015b0	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x000015b4	/* Parameters: q r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000015b8
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000015bc
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x000015c0	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x000015c8
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x000015cc
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x000015d0
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x000015d4
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x000015d8	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x000015dc	/* Parameters: q r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000015e0
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000015e4
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x000015e8	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S			0x000015f0
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T			0x000015f4
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R			0x000015f8
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q			0x000015fc
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x00001600	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x00001604	/* Parameters: q r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x00001608
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x0000160c
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x00001610	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S			0x00001620
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T			0x00001624
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R			0x00001628
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q			0x0000162c
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x00001630	/* Parameters: t s */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x00001634	/* Parameters: q r */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00001698
-#	define		NV20_TCL_PRIMITIVE_3D_EDGE_FLAG				0x000016bc
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS		0x00001720	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH		0x00001724	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR		0x00001728	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL		0x0000172c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2		0x00001730	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG		0x00001734	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6			0x00001738	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7			0x0000173c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0		0x00001740	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1		0x00001744	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2		0x00001748	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3		0x0000174c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4		0x00001750	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5		0x00001754	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6		0x00001758	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7		0x0000175c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00001760 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS			0x00001760	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH			0x00001764	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR			0x00001768	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL			0x0000176c	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2			0x00001770	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG			0x00001774	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6			0x00001778	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7			0x0000177c	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0			0x00001780	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1			0x00001784	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2			0x00001788	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3			0x0000178c	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4			0x00001790	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5			0x00001794	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6			0x00001798	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7			0x0000179c	/* Parameters: stride fields type */
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000017a0
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000017a4
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000017a8
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A		0x000017ac
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R		0x000017b0
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G		0x000017b4
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B		0x000017b8
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x000017bc
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x000017c0
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE	0x000017c4
-#	define		NV20_TCL_PRIMITIVE_3D_BEGIN_END				0x000017fc
-#	define		NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1			0x00001c30	/* Parameters: x2 x1 */
-#	define		NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1			0x00001c50	/* Parameters: y2 y1 */
-#	define		NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH			0x00001d8c
-#	define		NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB			0x00001d90
-#	define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS		0x00001d94	/* Parameters: clear color a clear color b clear color g clear color r clear depth clear stencil */
-#	define		NV20_TCL_PRIMITIVE_3D_INDEX_DATA			0x00001800	/* Parameters: index1 index0 */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001810	/* Parameters: count_vertices offset_vertices */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001818
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X			0x00001f00
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Y			0x00001f04
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Z			0x00001f08
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_W			0x00001f0c
-
-/****************************************** 
-Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
-*/
-#define			NV30_TCL_PRIMITIVE_3D					0x00000097
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT0			0x00000180
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT1			0x00000184
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT2			0x00000188
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT3			0x0000018c
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT4			0x00000194
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT5			0x00000198
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT6			0x000001a4
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT7			0x000001a8
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT8			0x000001ac
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT9			0x000001b4
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT10			0x000001b8
-#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC0_OBJECT		0x0000019c
-#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC1_OBJECT		0x000001a0
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color0 buffer pitch */
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR0_OFFSET			0x00000210
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR1_OFFSET			0x00000218
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER1_PITCH			0x0000021c	/* Parameters: color1 buffer pitch */
-#	define		NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x0000022c	/* Parameters: pitch */
-#	define		NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000230
-#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000240 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER2_PITCH			0x00000280
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER3_PITCH			0x00000284
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER2_OFFSET			0x00000288
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER3_OFFSET			0x0000028c
-#	define		NV30_TCL_PRIMITIVE_3D_DITHER_ENABLE			0x00000300
-#	define		NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000304
-#	define		NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x00000308
-#	define		NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF			0x0000030c
-#	define		NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000310
-#	define		NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000314
-#	define		NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000318
-#	define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000031c	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000320
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000324	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE		0x00000328
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK			0x0000032c
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC		0x00000330
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF		0x00000334
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK		0x00000338
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL		0x0000033c
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL		0x00000340
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS		0x00000344
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE		0x00000348
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK		0x0000034c
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC		0x00000350
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF		0x00000354
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK		0x00000358
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL		0x0000035c
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL		0x00000360
-#	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS		0x00000364
-#	define		NV30_TCL_PRIMITIVE_3D_SHADE_MODEL			0x00000368
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_ENABLE			0x0000036c
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR				0x00000370
-#	define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123		0x00000370	/* Parameters: buffer3 b buffer3 g buffer3 r buffer3 a buffer2 b buffer2 g buffer2 r buffer2 a buffer1 b buffer1 g buffer1 r buffer1 a */
-#	define		NV30_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x0000037c
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000394
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000398
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R		0x000003a0
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G		0x000003a4
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B		0x000003a8
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A		0x000003b4
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_WIDTH_SMOOTH			0x000003b8
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x000003bc
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x00000400 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + y * 0x0010 + x * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM			0x000008e4
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000008c8
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_MODE				0x000008cc
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x000008d0
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x000008d4
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x000008d8
-#	define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0				0x000008ec	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1				0x000008f0	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0				0x000008f4	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1				0x000008f8	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_ENABLE				0x000008fc	/* Parameters: number of rc enabled */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000900 + d * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000904 + d * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000908 + d * 0x0020)	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x0000090c + d * 0x0020)	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000910 + d * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000914 + d * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0	0x00000200	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1	0x00000204	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0	0x000002c0	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1	0x000002c4	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0			0x00000a00	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1			0x00000a04	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x00000a10
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x00000a14
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x00000a18
-#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS		0x000008c0	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS		0x000008c4	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE			0x00001ee8	/* Parameters: coord_replace r_mode enable */
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_SIZE			0x00001ee0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x00001ec0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x00001ec4
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00001ec8
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_D			0x00001ecc
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_E			0x00001ed0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x00001ed4
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00001ed8
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00001edc
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE		0x00001ee4
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_OX			0x00000a20
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_OY			0x00000a24
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_NPF_DIV2		0x00000a28
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_UNK0_0x0		0x00000a2c
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_PX_DIV2		0x00000a30
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_PY_DIV2		0x00000a34
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_FMN_DIV2		0x00000a38
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_UNK1_0x0		0x00000a3c
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE	0x00000a60
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE	0x00000a64
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE	0x00000a68
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x00000a6c
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x00000a70
-#	define		NV30_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x00000a74
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x00000a78
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x00000a7c
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0			0x00000b80
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST1			0x00000b84
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST2			0x00000b88
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST3			0x00000b8c
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R		0x000017b0
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G		0x000017b4
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B		0x000017b8
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A		0x000017c0
-#	define		NV30_TCL_PRIMITIVE_3D_OCC_QUERY_OR_COLOR_BUFF_ENABLE	0x000017c8
-#	define		NV30_TCL_PRIMITIVE_3D_STORE_RESULT			0x00001800
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000e00 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000e04 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000e08 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x00000e0c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001228 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000122c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001230 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001200 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001204 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001208 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000120c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001210 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001214 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001218 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000121c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001220 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001224 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_A	0x00001400
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_B	0x00001404
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_C	0x00001408
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_D	0x0000140c
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_E	0x00001410
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_F	0x00001414
-#	define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x00001420	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV30_TCL_PRIMITIVE_3D_UNK1D6C_OFFSET			0x00001d6c
-#	define		NV30_TCL_PRIMITIVE_3D_UNK1D70_VALUE			0x00001d70
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE		0x00001db4
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN		0x00001db8	/* Parameters: factor pattern */
-#	define		NV30_TCL_PRIMITIVE_3D_BEGIN_END				0x00001808
-#	define		NV30_TCL_PRIMITIVE_3D_CULL_FACE				0x00001830
-#	define		NV30_TCL_PRIMITIVE_3D_FRONT_FACE			0x00001834
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00001838
-#	define		NV30_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x0000183c
-#	define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH			0x00001d8c
-#	define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB			0x00001d90	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS		0x00001d94
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_A		0x00001e20
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_B		0x00001e24
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_C		0x00001e28
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_D		0x00001e2c
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_E		0x00001e30
-#	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_F		0x00001e34
-#	define		NV30_TCL_PRIMITIVE_3D_DO_VERTICES			0x00001dac
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_FROM_ID			0x00001e9c
-#	define		NV30_TCL_PRIMITIVE_3D_VP_PROGRAM_START_ID		0x00001ea0
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID		0x00001efc
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_X		0x00001f00
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_Y		0x00001f04
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_Z		0x00001f08
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_W		0x00001f0c
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P1_X		0x00001f10
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P1_Y		0x00001f14
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P1_Z		0x00001f18
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P1_W		0x00001f1c
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P2_X		0x00001f20
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P2_Y		0x00001f24
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P2_Z		0x00001f28
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P2_W		0x00001f2c
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_X		0x00001f30
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_Y		0x00001f34
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_Z		0x00001f38
-#	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_W		0x00001f3c
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3X(d)			(0x00001500 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3Y(d)			(0x00001504 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3Z(d)			(0x00001508 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3W(d)			(0x0000150c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4X(d)			(0x00001c00 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4Y(d)			(0x00001c04 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4Z(d)			(0x00001c08 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4W(d)			(0x00001c0c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VB_POINTER_ATTR(d)		(0x00001680 + d * 0x0004)	/* Parameters: source: offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000a90	/* Parameters: y x */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000a94	/* Parameters: z */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x000018c0
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x000018c4
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000018c8
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000018cc
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000018d0
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000018d4
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x000018d8
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x000018dc
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00001920	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x00001924	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x00001928	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x0000192c	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000194c	/* Parameters: a b g r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00001950	/* Parameters: a b g r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x000019c0	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x000019c4	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x000019c8	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x000019cc	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x000019d0	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x000019d4	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x000019d8	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x000019dc	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00001e54
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0			0x00001718
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00001740 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS			0x00001740	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH			0x00001744	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR			0x00001748	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL			0x0000174c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2			0x00001750	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG			0x00001754	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR6			0x00001758	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR7			0x0000175c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0			0x00001760	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1			0x00001764	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2			0x00001768	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3			0x0000176c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4			0x00001770	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5			0x00001774	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6			0x00001778	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7			0x0000177c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000017a0
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000017a4
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000017a8
-#	define		NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM			0x000008e4
-#	define		NV30_TCL_PRIMITIVE_3D_TX_ADDRESS_UNIT(d)		(0x00001a00 + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT(d)			(0x00001a04 + d * 0x0020)	/* Parameters: mipmap type format ncomp cubic */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT(d)			(0x00001a08 + d * 0x0020)	/* Parameters: wrap_s wrap_t wrap_r */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT(d)			(0x00001a0c + d * 0x0020)	/* Parameters: nv40_enable nv30_enable anisotropy */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_SWIZZLE_UNIT(d)		(0x00001a10 + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT(d)			(0x00001a14 + d * 0x0020)	/* Parameters: filter_min filter_mag */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT(d)			(0x00001a18 + d * 0x0020)	/* Parameters: width height */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_UNK07_UNIT(d)			(0x00001a1c + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT(d)			(0x00001840 + d * 0x0004)	/* Parameters: depth NPOT pitch */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001814	/* Parameters: count_vertices offset_vertices */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16			0x0000180c	/* Parameters: 1: 0: */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U32			0x00001810
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001818
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x00000374
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x00000378
-#	define		NV30_TCL_PRIMITIVE_3D_SET_DISPLAY_LIST_MEM_OFFSET	0x0000181c
-#	define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST		0x00001824	/* Parameters: length start offset */
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x00001828
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x0000182c
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000147c
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001480 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_SET_CLIPPING_PLANES		0x00001478
-#	define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG				0x00001ff0	/* Parameters: vertex pos weight normal primary color secondary color fogcoord texture coords 0 texture ccords 1 texture coords 2 texture coords 3 texture coords 4 texture coords 5 texture coords 6 texture coords 7 */
-#	define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG			0x00001ff4	/* Parameters: primary color secondary color backface primary color backface secondary color fogcoord pointsize clip plane 0 clip plane 1 clip plane 2 clip plane 3 clip plane 4 clip plane 5 texture coords 0 texture coords 1 texture coords 2 texture coords 3 texture coords 4 texture coords 5 texture coords 6 texture coords 7 */
-
-/****************************************** 
-Object NV30_CLEAR_BUFFER used on: NV30 NV40 G70
-*/
-#define			NV30_CLEAR_BUFFER					0x00000066
-#	define		NV30_CLEAR_BUFFER_SET_DMA_NOTIFY			0x00000180
-#	define		NV30_CLEAR_BUFFER_SET_IMAGE_PATTERN			0x00000188
-#	define		NV30_CLEAR_BUFFER_SET_RASTER_OP				0x0000018c
-#	define		NV30_CLEAR_BUFFER_SET_CONTEXT_SURFACE_2D		0x00000198
-#	define		NV30_CLEAR_BUFFER_UNK002fc				0x000002fc
-
-/****************************************** 
-Object NV50_TCL_PRIMITIVE_3D used on: 
-*/
-#define			NV50_TCL_PRIMITIVE_3D					0x00000097
-#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_0( d)			(0x00000180 + d * 0x0004)
-#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_1( d)			(0x000001c0 + d * 0x0004)
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00000314
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_X			0x00000380
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_Y			0x00000384
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x000003c0
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x000003c4
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000003c8
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000003cc
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000003d0
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000003d4
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x000003d8
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x000003dc
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00000400
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00000404
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00000408
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X			0x00000420
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00000424
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00000428
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R			0x00000430
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00000434
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00000438
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00000440
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00000444
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00000448
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X			0x00000500
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y			0x00000504
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z			0x00000508
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W			0x0000050c
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00000530
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00000534
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00000538
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A			0x0000053c
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x00000580
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x00000584
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x00000588
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x0000058c
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x00000590
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x00000594
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x00000598
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x0000059c
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S			0x000005a0
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T			0x000005a4
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R			0x000005a8
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q			0x000005ac
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S			0x000005b0
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T			0x000005b4
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R			0x000005b8
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q			0x000005bc
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x000006a0	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x000006a4	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x000006a8	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x000006ac	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY			0x00000700	/* Parameters: y x */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW			0x00000704	/* Parameters: w z */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x00000740	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x00000744	/* Parameters: q r */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x00000748	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x0000074c	/* Parameters: q r */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x00000750	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x00000754	/* Parameters: q r */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x00000758	/* Parameters: t s */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x0000075c	/* Parameters: q r */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000790	/* Parameters: y x */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000794	/* Parameters: z */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000088c	/* Parameters: a b g r */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00000890	/* Parameters: a b g r */
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_X			0x00000a00
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_Y			0x00000a04
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_Z			0x00000a08
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_X			0x00000a0c
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_Y			0x00000a10
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_Z			0x00000a14
-#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000c08
-#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000c0c
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x00000d00 + d * 0x0008)	/* Parameters: x2 x1 */
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x00000d04 + d * 0x0008)	/* Parameters: y2 y1 */
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_FIRST		0x00000d74
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_COUNT		0x00000d78
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_R			0x00000d80
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_G			0x00000d84
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_B			0x00000d88
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_A			0x00000d8c
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH			0x00000d90
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL			0x00000da0
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x00000dac
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x00000db0
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00000db4
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE	0x00000dc0
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE	0x00000dc4
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE	0x00000dc8
-#	define		NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS		0x00000e04	/* Parameters: w x */
-#	define		NV50_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS		0x00000e08	/* Parameters: h y */
-#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID		0x00000f00
-#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_X			0x00000f04
-#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Y			0x00000f08
-#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Z			0x00000f0c
-#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_W			0x00000f10
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF		0x00000f54
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK		0x00000f58
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK		0x00000f5c
-#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x000012cc
-#	define		NV50_TCL_PRIMITIVE_3D_SHADE_MODEL			0x000012d4
-#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x000012e8
-#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x000012ec
-#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x0000130c
-#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF			0x00001310
-#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x00001314
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_R			0x0000131c
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_G			0x00001320
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_B			0x00001324
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_A			0x00001328
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB		0x00001340
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB		0x00001344
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB		0x00001348
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA		0x0000134c
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA		0x00001350
-#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA		0x00001358
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE		0x00001380
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL		0x00001384
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL		0x00001388
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS		0x0000138c
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC		0x00001390
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF		0x00001394
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK			0x00001398
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK		0x0000139c
-#	define		NV50_TCL_PRIMITIVE_3D_LINE_WIDTH			0x000013b0
-#	define		NV50_TCL_PRIMITIVE_3D_POINT_SIZE			0x00001518
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x0000156c
-#	define		NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x00001570
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE		0x00001594
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL		0x00001598
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL		0x0000159c
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS		0x000015a0
-#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC		0x000015a4
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x000015bc
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN			0x000015dc
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_END			0x000015e0
-#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001640
-#	define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE		0x0000166c
-#	define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN		0x00001680	/* Parameters: pattern factor */
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000168c
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001700 + d * 0x0004)
-#	define		NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00001918
-#	define		NV50_TCL_PRIMITIVE_3D_FRONT_FACE			0x0000191c
-#	define		NV50_TCL_PRIMITIVE_3D_CULL_FACE				0x00001920
-#	define		NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE			0x000019c4
-#	define		NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP			0x000019c8
-#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS			0x000019d0	/* Parameters: color stencil depth */
-#	define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK( d)			(0x00001a00 + d * 0x0004)	/* Parameters: a b g r */
-
-/****************************************** 
-Object NV_DMA_FROM_MEMORY used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV_DMA_FROM_MEMORY					0x00000002
-
-/****************************************** 
-Object NV_DMA_TO_MEMORY used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV_DMA_TO_MEMORY					0x00000003
-
-/****************************************** 
-Object NV_DMA_IN_MEMORY used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV_DMA_IN_MEMORY					0x0000003d
-
-/****************************************** 
-Object NvType0046 used on: NV04 
-*/
-#define			NvType0046						0x00000046
-#	define		NvType0046_DMA_NOTIFY					0x00000180
-#	define		NvType0046_DMA_MEM_1					0x00000184
-#	define		NvType0046_DMA_MEM_2					0x00000188
-#	define		NvType0046_DMA_3					0x0000018c
-#	define		NvType0046_DMA_4					0x00000190
-#	define		NvType0046_OBJ_5					0x00000194
-#	define		NvType0046_OBJ_6					0x00000198
-#	define		NvType0046_PITCH1					0x00000304
-#	define		NvType0046_PITCH2					0x0000030c
-#	define		NvType0046_SIZE						0x00000340	/* Parameters: width height */
-#	define		NvType0046_WIDTH					0x00000344	/* Parameters: visible_width blank_width */
-#	define		NvType0046_HSYNC					0x00000348	/* Parameters: hsync_start hsync_len */
-#	define		NvType0046_HEIGHT					0x0000034c	/* Parameters: visible_height blank_height */
-#	define		NvType0046_VSYNC					0x00000350	/* Parameters: vsync_start vsync_len */
-#	define		NvType0046_FULL_SIZE					0x00000354	/* Parameters: full_width full_height */
-#	define		NvType0046_PIXEL_CLK					0x00000358
-#	define		NvType0046_FLAGS					0x0000035c	/* Parameters: doublescan -hsync -vsync depth */
-
-/****************************************** 
-Object NvType0047 used on: NV04 
-*/
-#define			NvType0047						0x00000047
-#	define		NvType0047_DMA_NOTIFY					0x00000180
-#	define		NvType0047_UNK19C					0x0000019c
-#	define		NvType0047_UNK1A0					0x000001a0
-
-/****************************************** 
-Object NvType0049 used on: NV04 
-*/
-#define			NvType0049						0x00000049
-#	define		NvType0049_DMA_NOTIFY					0x00000180
-#	define		NvType0049_DMA_MEM_1					0x00000184
-#	define		NvType0049_DMA_MEM_2					0x00000188
-
-/****************************************** 
-Object NvType004D used on: NV04 
-*/
-#define			NvType004D						0x0000004d
-#	define		NvType004D_DMA_NOTIFY					0x00000180
-#	define		NvType004D_DMA_MEM_1					0x00000184
-#	define		NvType004D_DMA_MEM_2					0x00000188
-#	define		NvType004D_DMA_MEM_3					0x0000018c
-#	define		NvType004D_DMA_MEM_4					0x00000190
-
-#endif /* _NOUVEAU_REG_H */
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.c b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
deleted file mode 100644
index 2cf6f979e40..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_screen.c
+++ /dev/null
@@ -1,382 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-
-#include "nouveau_context.h"
-#include "nouveau_screen.h"
-#include "nouveau_object.h"
-#include "nouveau_span.h"
-
-#include "utils.h"
-#include "context.h"
-#include "vblank.h"
-#include "drirenderbuffer.h"
-
-#include "GL/internal/dri_interface.h"
-
-#include "xmlpool.h"
-
-PUBLIC const char __driConfigOptions[] =
-DRI_CONF_BEGIN
-    DRI_CONF_SECTION_DEBUG
-        DRI_CONF_NO_RAST(false)
-    DRI_CONF_SECTION_END
-DRI_CONF_END;
-static const GLuint __driNConfigOptions = 1;
-
-extern const struct dri_extension common_extensions[];
-extern const struct dri_extension nv10_extensions[];
-extern const struct dri_extension nv20_extensions[];
-extern const struct dri_extension nv30_extensions[];
-extern const struct dri_extension nv40_extensions[];
-extern const struct dri_extension nv50_extensions[];
-
-static nouveauScreenPtr nouveauCreateScreen(__DRIscreenPrivate *sPriv)
-{
-	nouveauScreenPtr screen;
-	NOUVEAUDRIPtr dri_priv=(NOUVEAUDRIPtr)sPriv->pDevPriv;
-
-	/* allocate screen */
-	screen = (nouveauScreenPtr) CALLOC( sizeof(*screen) );
-	if ( !screen ) {         
-		__driUtilMessage("%s: Could not allocate memory for screen structure",__FUNCTION__);
-		return NULL;
-	}
-	
-	screen->card=nouveau_card_lookup(dri_priv->device_id);
-	if (!screen->card) {
-		__driUtilMessage("%s: Unknown card type 0x%04x:0x%04x\n",
-			__func__, dri_priv->device_id >> 16, dri_priv->device_id & 0xFFFF);
-		FREE(screen);
-		return NULL;
-	}
-
-	/* parse information in __driConfigOptions */
-	driParseOptionInfo (&screen->optionCache,__driConfigOptions, __driNConfigOptions);
-
-	screen->fbFormat    = dri_priv->bpp / 8;
-	screen->frontOffset = dri_priv->front_offset;
-	screen->frontPitch  = dri_priv->front_pitch;
-	screen->backOffset  = dri_priv->back_offset;
-	screen->backPitch   = dri_priv->back_pitch;
-	screen->depthOffset = dri_priv->depth_offset;
-	screen->depthPitch  = dri_priv->depth_pitch;
-
-	screen->driScreen = sPriv;
-	return screen;
-}
-
-static void
-nouveauDestroyScreen(__DRIscreenPrivate *sPriv)
-{
-	nouveauScreenPtr screen = (nouveauScreenPtr)sPriv->private;
-
-	if (!screen) return;
-
-	/* free all option information */
-	driDestroyOptionInfo (&screen->optionCache);
-
-	FREE(screen);
-	sPriv->private = NULL;
-}
-
-static GLboolean nouveauInitDriver(__DRIscreenPrivate *sPriv)
-{
-	sPriv->private = (void *) nouveauCreateScreen( sPriv );
-	if ( !sPriv->private ) {
-		nouveauDestroyScreen( sPriv );
-		return GL_FALSE;
-	}
-
-	return GL_TRUE;
-}
-
-/**
- * Create the Mesa framebuffer and renderbuffers for a given window/drawable.
- *
- * \todo This function (and its interface) will need to be updated to support
- * pbuffers.
- */
-static GLboolean
-nouveauCreateBuffer(__DRIscreenPrivate *driScrnPriv,
-                    __DRIdrawablePrivate *driDrawPriv,
-                    const __GLcontextModes *mesaVis,
-                    GLboolean isPixmap)
-{
-	nouveauScreenPtr screen = (nouveauScreenPtr) driScrnPriv->private;
-	nouveau_renderbuffer  *nrb;
-	struct gl_framebuffer *fb;
-	const GLboolean swAccum = mesaVis->accumRedBits > 0;
-	const GLboolean swStencil = mesaVis->stencilBits > 0 && mesaVis->depthBits != 24;
-	GLenum color_format = screen->fbFormat == 4 ? GL_RGBA8 : GL_RGB5;
-
-	if (isPixmap)
-		return GL_FALSE; /* not implemented */
-
-	fb = _mesa_create_framebuffer(mesaVis);
-	if (!fb)
-		return GL_FALSE;
-
-	/* Front buffer */
-	nrb = nouveau_renderbuffer_new(color_format,
-				       driScrnPriv->pFB + screen->frontOffset,
-				       screen->frontOffset,
-				       screen->frontPitch * screen->fbFormat,
-				       driDrawPriv);
-	nouveauSpanSetFunctions(nrb, mesaVis);
-	_mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &nrb->mesa);
-
-	if (0 /* unified buffers if we choose to support them.. */) {
-	} else {
-		if (mesaVis->doubleBufferMode) {
-			nrb = nouveau_renderbuffer_new(color_format, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &nrb->mesa);
-		}
-
-		if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH24_STENCIL8_EXT, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-			_mesa_add_renderbuffer(fb, BUFFER_STENCIL, &nrb->mesa);
-		} else if (mesaVis->depthBits == 24) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT24, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		} else if (mesaVis->depthBits == 16) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT16, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		}
-	}
-
-	_mesa_add_soft_renderbuffers(fb,
-				     GL_FALSE, /* color */
-				     GL_FALSE, /* depth */
-				     swStencil,
-				     swAccum,
-				     GL_FALSE, /* alpha */
-				     GL_FALSE  /* aux */);
-
-	driDrawPriv->driverPrivate = (void *) fb;
-	return (driDrawPriv->driverPrivate != NULL);
-}
-
-
-static void
-nouveauDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
-{
-	_mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
-}
-
-static int
-nouveauGetSwapInfo(__DRIdrawablePrivate *dpriv, __DRIswapInfo *sInfo)
-{
-	return -1;
-}
-
-static const struct __DriverAPIRec nouveauAPI = {
-	.InitDriver      = nouveauInitDriver,
-	.DestroyScreen   = nouveauDestroyScreen,
-	.CreateContext   = nouveauCreateContext,
-	.DestroyContext  = nouveauDestroyContext,
-	.CreateBuffer    = nouveauCreateBuffer,
-	.DestroyBuffer   = nouveauDestroyBuffer,
-	.SwapBuffers     = nouveauSwapBuffers,
-	.MakeCurrent     = nouveauMakeCurrent,
-	.UnbindContext   = nouveauUnbindContext,
-	.GetSwapInfo     = nouveauGetSwapInfo,
-	.GetMSC          = driGetMSC32,
-	.WaitForMSC      = driWaitForMSC32,
-	.WaitForSBC      = NULL,
-	.SwapBuffersMSC  = NULL,
-	.CopySubBuffer   = nouveauCopySubBuffer
-};
-
-
-static __GLcontextModes *
-nouveauFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
-{
-	__GLcontextModes * modes;
-	__GLcontextModes * m;
-	unsigned num_modes;
-	unsigned depth_buffer_factor;
-	unsigned back_buffer_factor;
-	int i;
-
-	static const struct {
-		GLenum format;
-		GLenum type;
-	} fb_format_array[] = {
-		{ GL_RGB , GL_UNSIGNED_SHORT_5_6_5     },
-		{ GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV },
-		{ GL_BGR , GL_UNSIGNED_INT_8_8_8_8_REV },
-	};
-
-	/* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
-	 * support pageflipping at all.
-	 */
-	static const GLenum back_buffer_modes[] = {
-		GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
-	};
-
-	uint8_t depth_bits_array[4]   = { 0, 16, 24, 24 };
-	uint8_t stencil_bits_array[4] = { 0,  0,  0,  8 };
-
-	depth_buffer_factor = 4;
-	back_buffer_factor  = (have_back_buffer) ? 3 : 1;
-
-	num_modes = ((pixel_bits==16) ? 1 : 2) *
-		depth_buffer_factor * back_buffer_factor * 4;
-	modes = (*dri_interface->createContextModes)(num_modes,
-						     sizeof(__GLcontextModes));
-	m = modes;
-
-	for (i=((pixel_bits==16)?0:1);i<((pixel_bits==16)?1:3);i++) {
-		if (!driFillInModes(&m, fb_format_array[i].format,
-					fb_format_array[i].type,
-					depth_bits_array,
-					stencil_bits_array,
-					depth_buffer_factor,
-					back_buffer_modes,
-					back_buffer_factor,
-					GLX_TRUE_COLOR)) {
-		fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-				__func__, __LINE__ );
-		return NULL;
-		}
-
-		if (!driFillInModes(&m, fb_format_array[i].format,
-					fb_format_array[i].type,
-					depth_bits_array,
-					stencil_bits_array,
-					depth_buffer_factor,
-					back_buffer_modes,
-					back_buffer_factor,
-					GLX_DIRECT_COLOR)) {
-		fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-				__func__, __LINE__ );
-		return NULL;
-		}
-	}
-
-	return modes;
-}
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-				     const __GLcontextModes * modes,
-				     const __DRIversion * ddx_version,
-				     const __DRIversion * dri_version,
-				     const __DRIversion * drm_version,
-				     const __DRIframebuffer * frame_buffer,
-				     drmAddress pSAREA, int fd, 
-				     int internal_api_version,
-				     const __DRIinterfaceMethods * interface,
-				     __GLcontextModes ** driver_modes)
-			     
-{
-	__DRIscreenPrivate *psp;
-	static const __DRIversion ddx_expected = { 1, 2, 0 };
-	static const __DRIversion dri_expected = { 4, 0, 0 };
-	static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL };
-#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 9
-#error nouveau_drm.h version doesn't match expected version
-#endif
-	dri_interface = interface;
-
-	if (!driCheckDriDdxDrmVersions2("nouveau",
-					dri_version, & dri_expected,
-					ddx_version, & ddx_expected,
-					drm_version, & drm_expected)) {
-		return NULL;
-	}
-
-	// temporary lock step versioning
-	if (drm_expected.patch!=drm_version->patch) {
-		__driUtilMessage("%s: wrong DRM version, expected %d, got %d\n",
-				__func__,
-				drm_expected.patch, drm_version->patch);
-		return NULL;
-	}
-
-	psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				       ddx_version, dri_version, drm_version,
-				       frame_buffer, pSAREA, fd,
-				       internal_api_version, &nouveauAPI);
-	if ( psp != NULL ) {
-		NOUVEAUDRIPtr dri_priv = (NOUVEAUDRIPtr)psp->pDevPriv;
-
-		*driver_modes = nouveauFillInModes(dri_priv->bpp,
-						   (dri_priv->bpp == 16) ? 16 : 24,
-						   (dri_priv->bpp == 16) ? 0  : 8,
-						   1
-						   );
-
-		/* Calling driInitExtensions here, with a NULL context pointer, does not actually
-		 * enable the extensions.  It just makes sure that all the dispatch offsets for all
-		 * the extensions that *might* be enables are known.  This is needed because the
-		 * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-		 * enable the extensions until we have a context pointer.
-		 * 
-		 * Hello chicken.  Hello egg.  How are you two today?
-		 */
-		driInitExtensions( NULL, common_extensions, GL_FALSE );
-		driInitExtensions( NULL,   nv10_extensions, GL_FALSE );
-		driInitExtensions( NULL,   nv10_extensions, GL_FALSE );
-		driInitExtensions( NULL,   nv30_extensions, GL_FALSE );
-		driInitExtensions( NULL,   nv40_extensions, GL_FALSE );
-		driInitExtensions( NULL,   nv50_extensions, GL_FALSE );
-	}
-
-	return (void *) psp;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.h b/src/mesa/drivers/dri/nouveau/nouveau_screen.h
deleted file mode 100644
index bbe5810128f..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_screen.h
+++ /dev/null
@@ -1,61 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef __NOUVEAU_SCREEN_H__
-#define __NOUVEAU_SCREEN_H__
-
-#include "xmlconfig.h"
-
-#include "nouveau_dri.h"
-#include "nouveau_card.h"
-
-typedef struct {
-	nouveau_card* card;
-	uint32_t bus_type;
-	uint32_t agp_mode;
-
-        GLint fbFormat;
-
-	GLuint frontOffset;
-	GLuint frontPitch;
-	GLuint backOffset;
-	GLuint backPitch;
-
-	GLuint depthOffset;
-	GLuint depthPitch;
-	GLuint spanOffset;
-
-	__DRIscreenPrivate *driScreen;
-	unsigned int sarea_priv_offset;
-
-	/* Configuration cache with default values for all contexts */
-	driOptionCache optionCache;
-
-} nouveauScreenRec, *nouveauScreenPtr;
-
-
-#endif /* __NOUVEAU_SCREEN_H__ */
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.c b/src/mesa/drivers/dri/nouveau/nouveau_shader.c
deleted file mode 100644
index b6837c5de14..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.c
+++ /dev/null
@@ -1,833 +0,0 @@
-/*
- * Copyright (C) 2006 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/*
- * Authors:
- *   Ben Skeggs <darktama@iinet.net.au>
- */
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "extensions.h"
-
-#include "shader/program.h"
-#include "shader/prog_instruction.h"
-/*#include "shader/arbprogparse.h"*/
-#include "tnl/tnl.h"
-
-#include "nouveau_context.h"
-#include "nouveau_shader.h"
-
-/*****************************************************************************
- * Mesa entry points
- */
-static void
-nouveauBindProgram(GLcontext *ctx, GLenum target, struct gl_program *prog)
-{
-   NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
-}
-
-static struct gl_program *
-nouveauNewProgram(GLcontext *ctx, GLenum target, GLuint id)
-{
-   nouveauShader *nvs;
-
-   NVSDBG("target=%s, id=%d\n", _mesa_lookup_enum_by_nr(target), id);
-
-   nvs = CALLOC_STRUCT(_nouveauShader);
-   NVSDBG("prog=%p\n", nvs);
-   switch (target) {
-   case GL_VERTEX_PROGRAM_ARB:
-      return _mesa_init_vertex_program(ctx, &nvs->mesa.vp, target, id);
-   case GL_FRAGMENT_PROGRAM_ARB:
-      return _mesa_init_fragment_program(ctx, &nvs->mesa.fp, target, id);
-   default:
-      _mesa_problem(ctx, "Unsupported shader target");
-      break;
-   }
-
-   FREE(nvs);
-   return NULL;
-}
-
-static void
-nouveauDeleteProgram(GLcontext *ctx, struct gl_program *prog)
-{
-   nouveauShader *nvs = (nouveauShader *)prog;
-
-   NVSDBG("prog=%p\n", prog);
-
-   if (nvs->translated)
-      FREE(nvs->program);
-   _mesa_delete_program(ctx, prog);
-}
-
-static void
-nouveauProgramStringNotify(GLcontext *ctx, GLenum target,
-      			   struct gl_program *prog)
-{
-   nouveauShader *nvs = (nouveauShader *)prog;
-
-   NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
-
-   if (nvs->translated)
-      FREE(nvs->program);
-
-   nvs->error      = GL_FALSE;
-   nvs->translated = GL_FALSE;
-
-   _tnl_program_string(ctx, target, prog);
-}
-
-static GLboolean
-nouveauIsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
-{
-   nouveauShader *nvs = (nouveauShader *)prog;
-
-   NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
-
-   return nvs->translated;
-}
-
-GLboolean
-nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   struct gl_program_parameter_list *plist;
-   int i;
-
-   NVSDBG("prog=%p\n", nvs);
-
-   /* Translate to HW format now if necessary */
-   if (!nvs->translated) {
-      /* Mesa ASM shader -> nouveauShader */
-      if (!nouveau_shader_pass0(ctx, nvs))
-	 return GL_FALSE;
-      /* Basic dead code elimination + register usage info */
-      if (!nouveau_shader_pass1(nvs))
-	 return GL_FALSE;
-      /* nouveauShader -> HW bytecode, HW register alloc */
-      if (!nouveau_shader_pass2(nvs))
-	 return GL_FALSE;
-      assert(nvs->translated);
-      assert(nvs->program);
-   }
-   
-   /* Update state parameters */
-   plist = nvs->mesa.vp.Base.Parameters;
-   _mesa_load_state_parameters(ctx, plist);
-   for (i=0; i<nvs->param_high; i++) {
-      if (!nvs->params[i].in_use)
-	 continue;
-
-      if (!nvs->on_hardware) {
-	 /* if we've been kicked off the hardware there's no guarantee our
-	  * consts are still there.. reupload them all
-	  */
-	 nvs->func->UpdateConst(ctx, nvs, i);
-      } else if (nvs->params[i].source_val) {
-	 /* update any changed state parameters */
-	 if (!TEST_EQ_4V(nvs->params[i].val, nvs->params[i].source_val))
-	    nvs->func->UpdateConst(ctx, nvs, i);
-      }
-   }
-
-   /* Upload program to hardware, this must come after state param update
-    * as >=NV30 fragprogs inline consts into the bytecode.
-    */
-   if (!nvs->on_hardware) {
-      nouveauShader **current;
-
-      if (nvs->mesa.vp.Base.Target == GL_VERTEX_PROGRAM_ARB)
-	 current = &nmesa->current_vertprog;
-      else
-	 current = &nmesa->current_fragprog;
-      if (*current) (*current)->on_hardware = 0;
-
-      nvs->func->UploadToHW(ctx, nvs);
-      nvs->on_hardware = 1;
-
-      *current = nvs;
-   }
-
-   return GL_TRUE;
-}
-
-nouveauShader *
-nvsBuildTextShader(GLcontext *ctx, GLenum target, const char *text)
-{
-   nouveauShader *nvs;
-
-   nvs = CALLOC_STRUCT(_nouveauShader);
-   if (!nvs)
-      return NULL;
-
-   if (target == GL_VERTEX_PROGRAM_ARB) {
-      _mesa_init_vertex_program(ctx, &nvs->mesa.vp, GL_VERTEX_PROGRAM_ARB, 0);
-      _mesa_parse_arb_vertex_program(ctx,
-	    			     GL_VERTEX_PROGRAM_ARB,
-				     text,
-				     strlen(text),
-				     &nvs->mesa.vp);
-   } else if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      _mesa_init_fragment_program(ctx, &nvs->mesa.fp, GL_FRAGMENT_PROGRAM_ARB, 0);
-      _mesa_parse_arb_fragment_program(ctx,
-	    			       GL_FRAGMENT_PROGRAM_ARB,
-				       text,
-				       strlen(text),
-				       &nvs->mesa.fp);
-   }
-
-   nouveau_shader_pass0(ctx, nvs);
-   nouveau_shader_pass1(nvs);
-   nouveau_shader_pass2(nvs);
-
-   return nvs;
-}
-
-static void
-nvsBuildPassthroughVP(GLcontext *ctx)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   const char *vp_text =
-      "!!ARBvp1.0\n"
-      "OPTION ARB_position_invariant;"
-      ""
-      "MOV result.color, vertex.color;\n"
-      "MOV result.texcoord[0], vertex.texcoord[0];\n"
-      "MOV result.texcoord[1], vertex.texcoord[1];\n"
-      "MOV result.texcoord[2], vertex.texcoord[2];\n"
-      "MOV result.texcoord[3], vertex.texcoord[3];\n"
-      "MOV result.texcoord[4], vertex.texcoord[4];\n"
-      "MOV result.texcoord[5], vertex.texcoord[5];\n"
-      "MOV result.texcoord[6], vertex.texcoord[6];\n"
-      "MOV result.texcoord[7], vertex.texcoord[7];\n"
-      "END";
-
-   nmesa->passthrough_vp = nvsBuildTextShader(ctx,
-	 				      GL_VERTEX_PROGRAM_ARB,
-					      vp_text);
-}
-
-static void
-nvsBuildPassthroughFP(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	const char *fp_text =
-		"!!ARBfp1.0\n"
-		"MOV result.color, fragment.color;\n"
-		"END";
-
-	nmesa->passthrough_fp = nvsBuildTextShader(ctx,
-						   GL_FRAGMENT_PROGRAM_ARB,
-						   fp_text);
-}
-
-void
-nouveauShaderInitFuncs(GLcontext * ctx)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   switch (nmesa->screen->card->type) {
-   case NV_20:
-      NV20VPInitShaderFuncs(&nmesa->VPfunc);
-      break;
-   case NV_30:
-      NV30VPInitShaderFuncs(&nmesa->VPfunc);
-      NV30FPInitShaderFuncs(&nmesa->FPfunc);
-      break;
-   case NV_40:
-   case NV_44:
-      NV40VPInitShaderFuncs(&nmesa->VPfunc);
-      NV40FPInitShaderFuncs(&nmesa->FPfunc);
-      break;
-   case NV_50:
-   default:
-      return;
-   }
-
-   /* Build a vertex program that simply passes through all attribs.
-    * Needed to do swtcl on nv40
-    */
-   if (nmesa->screen->card->type >= NV_40)
-      nvsBuildPassthroughVP(ctx);
-
-   /* Needed on NV30, even when using swtcl, if you want to get colours */
-   if (nmesa->screen->card->type >= NV_30)
-      nvsBuildPassthroughFP(ctx);
-
-   ctx->Const.VertexProgram.MaxNativeInstructions    = nmesa->VPfunc.MaxInst;
-   ctx->Const.VertexProgram.MaxNativeAluInstructions = nmesa->VPfunc.MaxInst;
-   ctx->Const.VertexProgram.MaxNativeTexInstructions = nmesa->VPfunc.MaxInst;
-   ctx->Const.VertexProgram.MaxNativeTexIndirections =
-      ctx->Const.VertexProgram.MaxNativeTexInstructions;
-   ctx->Const.VertexProgram.MaxNativeAttribs         = nmesa->VPfunc.MaxAttrib;
-   ctx->Const.VertexProgram.MaxNativeTemps           = nmesa->VPfunc.MaxTemp;
-   ctx->Const.VertexProgram.MaxNativeAddressRegs     = nmesa->VPfunc.MaxAddress;
-   ctx->Const.VertexProgram.MaxNativeParameters      = nmesa->VPfunc.MaxConst;
-
-   if (nmesa->screen->card->type >= NV_30) {
-      ctx->Const.FragmentProgram.MaxNativeInstructions    = nmesa->FPfunc.MaxInst;
-      ctx->Const.FragmentProgram.MaxNativeAluInstructions = nmesa->FPfunc.MaxInst;
-      ctx->Const.FragmentProgram.MaxNativeTexInstructions = nmesa->FPfunc.MaxInst;
-      ctx->Const.FragmentProgram.MaxNativeTexIndirections =
-	 ctx->Const.FragmentProgram.MaxNativeTexInstructions;
-      ctx->Const.FragmentProgram.MaxNativeAttribs         = nmesa->FPfunc.MaxAttrib;
-      ctx->Const.FragmentProgram.MaxNativeTemps           = nmesa->FPfunc.MaxTemp;
-      ctx->Const.FragmentProgram.MaxNativeAddressRegs     = nmesa->FPfunc.MaxAddress;
-      ctx->Const.FragmentProgram.MaxNativeParameters      = nmesa->FPfunc.MaxConst;
-   }
-
-   ctx->Driver.NewProgram		= nouveauNewProgram;
-   ctx->Driver.BindProgram		= nouveauBindProgram;
-   ctx->Driver.DeleteProgram		= nouveauDeleteProgram;
-   ctx->Driver.ProgramStringNotify	= nouveauProgramStringNotify;
-   ctx->Driver.IsProgramNative		= nouveauIsProgramNative;
-}
-
-
-/*****************************************************************************
- * Disassembly support structs
- */
-#define CHECK_RANGE(idx, arr) ((idx)<sizeof(_##arr)/sizeof(const char *)) \
-	? _##arr[(idx)] : #arr"_OOB"
-
-#define NODS      (1<<0)
-#define BRANCH_TR (1<<1)
-#define BRANCH_EL (1<<2)
-#define BRANCH_EN (1<<3)
-#define BRANCH_RE (1<<4)
-#define BRANCH_ALL (BRANCH_TR|BRANCH_EL|BRANCH_EN)
-#define COUNT_INC (1<<4)
-#define COUNT_IND (1<<5)
-#define COUNT_NUM (1<<6)
-#define COUNT_ALL (COUNT_INC|COUNT_IND|COUNT_NUM)
-#define TI_UNIT   (1<<7)
-struct _opcode_info
-{
-   const char *name;
-   int numsrc;
-   int flags;
-};
-
-static struct _opcode_info ops[] = {
-   [NVS_OP_ABS] = {"ABS", 1, 0},
-   [NVS_OP_ADD] = {"ADD", 2, 0},
-   [NVS_OP_ARA] = {"ARA", 1, 0},
-   [NVS_OP_ARL] = {"ARL", 1, 0},
-   [NVS_OP_ARR] = {"ARR", 1, 0},
-   [NVS_OP_BRA] = {"BRA", 0, NODS | BRANCH_TR},
-   [NVS_OP_BRK] = {"BRK", 0, NODS},
-   [NVS_OP_CAL] = {"CAL", 0, NODS | BRANCH_TR},
-   [NVS_OP_CMP] = {"CMP", 2, 0},
-   [NVS_OP_COS] = {"COS", 1, 0},
-   [NVS_OP_DIV] = {"DIV", 2, 0},
-   [NVS_OP_DDX] = {"DDX", 1, 0},
-   [NVS_OP_DDY] = {"DDY", 1, 0},
-   [NVS_OP_DP2] = {"DP2", 2, 0},
-   [NVS_OP_DP2A] = {"DP2A", 3, 0},
-   [NVS_OP_DP3] = {"DP3", 2, 0},
-   [NVS_OP_DP4] = {"DP4", 2, 0},
-   [NVS_OP_DPH] = {"DPH", 2, 0},
-   [NVS_OP_DST] = {"DST", 2, 0},
-   [NVS_OP_EX2] = {"EX2", 1, 0},
-   [NVS_OP_EXP] = {"EXP", 1, 0},
-   [NVS_OP_FLR] = {"FLR", 1, 0},
-   [NVS_OP_FRC] = {"FRC", 1, 0},
-   [NVS_OP_IF] = {"IF", 0, NODS | BRANCH_EL | BRANCH_EN},
-   [NVS_OP_KIL] = {"KIL", 1, 0},
-   [NVS_OP_LG2] = {"LG2", 1, 0},
-   [NVS_OP_LIT] = {"LIT", 1, 0},
-   [NVS_OP_LOG] = {"LOG", 1, 0},
-   [NVS_OP_LOOP] = {"LOOP", 0, NODS | COUNT_ALL | BRANCH_EN},
-   [NVS_OP_LRP] = {"LRP", 3, 0},
-   [NVS_OP_MAD] = {"MAD", 3, 0},
-   [NVS_OP_MAX] = {"MAX", 2, 0},
-   [NVS_OP_MIN] = {"MIN", 2, 0},
-   [NVS_OP_MOV] = {"MOV", 1, 0},
-   [NVS_OP_MUL] = {"MUL", 2, 0},
-   [NVS_OP_NRM] = {"NRM", 1, 0},
-   [NVS_OP_PK2H] = {"PK2H", 1, 0},
-   [NVS_OP_PK2US] = {"PK2US", 1, 0},
-   [NVS_OP_PK4B] = {"PK4B", 1, 0},
-   [NVS_OP_PK4UB] = {"PK4UB", 1, 0},
-   [NVS_OP_POW] = {"POW", 2, 0},
-   [NVS_OP_POPA] = {"POPA", 0, 0},
-   [NVS_OP_PUSHA] = {"PUSHA", 1, NODS},
-   [NVS_OP_RCC] = {"RCC", 1, 0},
-   [NVS_OP_RCP] = {"RCP", 1, 0},
-   [NVS_OP_REP] = {"REP", 0, NODS | BRANCH_EN | COUNT_NUM},
-   [NVS_OP_RET] = {"RET", 0, NODS},
-   [NVS_OP_RFL] = {"RFL", 1, 0},
-   [NVS_OP_RSQ] = {"RSQ", 1, 0},
-   [NVS_OP_SCS] = {"SCS", 1, 0},
-   [NVS_OP_SEQ] = {"SEQ", 2, 0},
-   [NVS_OP_SFL] = {"SFL", 2, 0},
-   [NVS_OP_SGE] = {"SGE", 2, 0},
-   [NVS_OP_SGT] = {"SGT", 2, 0},
-   [NVS_OP_SIN] = {"SIN", 1, 0},
-   [NVS_OP_SLE] = {"SLE", 2, 0},
-   [NVS_OP_SLT] = {"SLT", 2, 0},
-   [NVS_OP_SNE] = {"SNE", 2, 0},
-   [NVS_OP_SSG] = {"SSG", 1, 0},
-   [NVS_OP_STR] = {"STR", 2, 0},
-   [NVS_OP_SUB] = {"SUB", 2, 0},
-   [NVS_OP_TEX] = {"TEX", 1, TI_UNIT},
-   [NVS_OP_TXB] = {"TXB", 1, TI_UNIT},
-   [NVS_OP_TXD] = {"TXD", 3, TI_UNIT},
-   [NVS_OP_TXL] = {"TXL", 1, TI_UNIT},
-   [NVS_OP_TXP] = {"TXP", 1, TI_UNIT},
-   [NVS_OP_UP2H] = {"UP2H", 1, 0},
-   [NVS_OP_UP2US] = {"UP2US", 1, 0},
-   [NVS_OP_UP4B] = {"UP4B", 1, 0},
-   [NVS_OP_UP4UB] = {"UP4UB", 1, 0},
-   [NVS_OP_X2D] = {"X2D", 3, 0},
-   [NVS_OP_XPD] = {"XPD", 2, 0},
-   [NVS_OP_NOP] = {"NOP", 0, NODS},
-};
-
-static struct _opcode_info *
-_get_op_info(int op)
-{
-   if (op >= (sizeof(ops) / sizeof(struct _opcode_info)))
-      return NULL;
-   if (ops[op].name == NULL)
-      return NULL;
-   return &ops[op];
-}
-
-static const char *_SFR_STRING[] = {
-   [NVS_FR_POSITION] = "position",
-   [NVS_FR_WEIGHT] = "weight",
-   [NVS_FR_NORMAL] = "normal",
-   [NVS_FR_COL0] = "color",
-   [NVS_FR_COL1] = "color.secondary",
-   [NVS_FR_BFC0] = "bfc",
-   [NVS_FR_BFC1] = "bfc.secondary",
-   [NVS_FR_FOGCOORD] = "fogcoord",
-   [NVS_FR_POINTSZ] = "pointsize",
-   [NVS_FR_TEXCOORD0] = "texcoord[0]",
-   [NVS_FR_TEXCOORD1] = "texcoord[1]",
-   [NVS_FR_TEXCOORD2] = "texcoord[2]",
-   [NVS_FR_TEXCOORD3] = "texcoord[3]",
-   [NVS_FR_TEXCOORD4] = "texcoord[4]",
-   [NVS_FR_TEXCOORD5] = "texcoord[5]",
-   [NVS_FR_TEXCOORD6] = "texcoord[6]",
-   [NVS_FR_TEXCOORD7] = "texcoord[7]",
-   [NVS_FR_FRAGDATA0] = "data[0]",
-   [NVS_FR_FRAGDATA1] = "data[1]",
-   [NVS_FR_FRAGDATA2] = "data[2]",
-   [NVS_FR_FRAGDATA3] = "data[3]",
-   [NVS_FR_CLIP0] = "clip_plane[0]",
-   [NVS_FR_CLIP1] = "clip_plane[1]",
-   [NVS_FR_CLIP2] = "clip_plane[2]",
-   [NVS_FR_CLIP3] = "clip_plane[3]",
-   [NVS_FR_CLIP4] = "clip_plane[4]",
-   [NVS_FR_CLIP5] = "clip_plane[5]",
-   [NVS_FR_CLIP6] = "clip_plane[6]",
-   [NVS_FR_FACING] = "facing",
-};
-
-#define SFR_STRING(idx) CHECK_RANGE((idx), SFR_STRING)
-
-static const char *_SWZ_STRING[] = {
-   [NVS_SWZ_X] = "x",
-   [NVS_SWZ_Y] = "y",
-   [NVS_SWZ_Z] = "z",
-   [NVS_SWZ_W] = "w"
-};
-
-#define SWZ_STRING(idx) CHECK_RANGE((idx), SWZ_STRING)
-
-static const char *_NVS_PREC_STRING[] = {
-   [NVS_PREC_FLOAT32] = "R",
-   [NVS_PREC_FLOAT16] = "H",
-   [NVS_PREC_FIXED12] = "X",
-   [NVS_PREC_UNKNOWN] = "?"
-};
-
-#define NVS_PREC_STRING(idx) CHECK_RANGE((idx), NVS_PREC_STRING)
-
-static const char *_NVS_COND_STRING[] = {
-   [NVS_COND_FL] = "FL",
-   [NVS_COND_LT] = "LT",
-   [NVS_COND_EQ] = "EQ",
-   [NVS_COND_LE] = "LE",
-   [NVS_COND_GT] = "GT",
-   [NVS_COND_NE] = "NE",
-   [NVS_COND_GE] = "GE",
-   [NVS_COND_TR] = "TR",
-   [NVS_COND_UNKNOWN] = "??"
-};
-
-#define NVS_COND_STRING(idx) CHECK_RANGE((idx), NVS_COND_STRING)
-
-/*****************************************************************************
- * ShaderFragment dumping
- */
-static void
-nvsDumpIndent(int lvl)
-{
-   while (lvl--)
-      printf("  ");
-}
-
-static void
-nvsDumpSwizzle(nvsSwzComp *swz)
-{
-   printf(".%s%s%s%s",
-	  SWZ_STRING(swz[0]),
-	  SWZ_STRING(swz[1]), SWZ_STRING(swz[2]), SWZ_STRING(swz[3])
-      );
-}
-
-static void
-nvsDumpReg(nvsInstruction * inst, nvsRegister * reg)
-{
-   if (reg->negate)
-      printf("-");
-   if (reg->abs)
-      printf("abs(");
-
-   switch (reg->file) {
-   case NVS_FILE_TEMP:
-      printf("R%d", reg->index);
-      nvsDumpSwizzle(reg->swizzle);
-      break;
-   case NVS_FILE_ATTRIB:
-      printf("attrib.%s", SFR_STRING(reg->index));
-      nvsDumpSwizzle(reg->swizzle);
-      break;
-   case NVS_FILE_ADDRESS:
-      printf("A%d", reg->index);
-      break;
-   case NVS_FILE_CONST:
-      if (reg->indexed)
-	 printf("const[A%d.%s + %d]",
-		reg->addr_reg, SWZ_STRING(reg->addr_comp), reg->index);
-      else
-	 printf("const[%d]", reg->index);
-      nvsDumpSwizzle(reg->swizzle);
-      break;
-   default:
-      printf("UNKNOWN_FILE");
-      break;
-   }
-
-   if (reg->abs)
-      printf(")");
-}
-
-static void
-nvsDumpInstruction(nvsInstruction * inst, int slot, int lvl)
-{
-   struct _opcode_info *opr = &ops[inst->op];
-   int i;
-
-   nvsDumpIndent(lvl);
-   printf("%s ", opr->name);
-
-   if (!opr->flags & NODS) {
-      switch (inst->dest.file) {
-      case NVS_FILE_RESULT:
-	 printf("result.%s", SFR_STRING(inst->dest.index));
-	 break;
-      case NVS_FILE_TEMP:
-	 printf("R%d", inst->dest.index);
-	 break;
-      case NVS_FILE_ADDRESS:
-	 printf("A%d", inst->dest.index);
-	 break;
-      default:
-	 printf("UNKNOWN_DST_FILE");
-	 break;
-      }
-
-      if (inst->mask != SMASK_ALL) {
-	 printf(".");
-	 if (inst->mask & SMASK_X)
-	    printf("x");
-	 if (inst->mask & SMASK_Y)
-	    printf("y");
-	 if (inst->mask & SMASK_Z)
-	    printf("z");
-	 if (inst->mask & SMASK_W)
-	    printf("w");
-      }
-
-      if (opr->numsrc)
-	 printf(", ");
-   }
-
-   for (i = 0; i < opr->numsrc; i++) {
-      nvsDumpReg(inst, &inst->src[i]);
-      if (i != opr->numsrc - 1)
-	 printf(", ");
-   }
-   if (opr->flags & TI_UNIT)
-      printf(", texture[%d]", inst->tex_unit);
-
-   printf("\n");
-}
-
-void
-nvsDumpFragmentList(nvsFragmentHeader *f, int lvl)
-{
-   while (f) {
-      switch (f->type) {
-      case NVS_INSTRUCTION:
-	 nvsDumpInstruction((nvsInstruction*)f, 0, lvl);
-	 break;
-      default:
-	 fprintf(stderr, "%s: Only NVS_INSTRUCTION fragments can be in"
-	                 "nvsFragmentList!\n", __func__);
-	 return;
-      }
-      f = f->next;
-   }
-}
-
-/*****************************************************************************
- * HW shader disassembly
- */
-static void
-nvsDisasmHWShaderOp(nvsFunc * shader, int merged)
-{
-   struct _opcode_info *opi;
-   nvsOpcode op;
-   nvsRegFile file;
-   nvsSwzComp swz[4];
-   int i;
-
-   op = shader->GetOpcode(shader, merged);
-   opi = _get_op_info(op);
-   if (!opi) {
-      printf("NO OPINFO!");
-      return;
-   }
-
-   printf("%s", opi->name);
-   if (shader->GetPrecision &&
-       (!(opi->flags & BRANCH_ALL)) && (!(opi->flags * NODS)) &&
-       (op != NVS_OP_NOP))
-      printf("%s", NVS_PREC_STRING(shader->GetPrecision(shader)));
-   if (shader->SupportsConditional && shader->SupportsConditional(shader)) {
-      if (shader->GetConditionUpdate(shader)) {
-	 printf("C%d", shader->GetCondRegID(shader));
-      }
-   }
-   if (shader->GetSaturate && shader->GetSaturate(shader))
-      printf("_SAT");
-
-   if (!(opi->flags & NODS)) {
-      int mask = shader->GetDestMask(shader, merged);
-
-      switch (shader->GetDestFile(shader, merged)) {
-      case NVS_FILE_ADDRESS:
-	 printf(" A%d", shader->GetDestID(shader, merged));
-	 break;
-      case NVS_FILE_TEMP:
-	 printf(" R%d", shader->GetDestID(shader, merged));
-	 break;
-      case NVS_FILE_RESULT:
-	 printf(" result.%s", (SFR_STRING(shader->GetDestID(shader, merged))));
-	 break;
-      default:
-	 printf(" BAD_RESULT_FILE");
-	 break;
-      }
-
-      if (mask != SMASK_ALL) {
-	 printf(".");
-	 if (mask & SMASK_X) printf("x");
-	 if (mask & SMASK_Y) printf("y");
-	 if (mask & SMASK_Z) printf("z");
-	 if (mask & SMASK_W) printf("w");
-      }
-   }
-
-   if (shader->SupportsConditional && shader->SupportsConditional(shader) &&
-       shader->GetConditionTest(shader)) {
-      shader->GetCondRegSwizzle(shader, swz);
-
-      printf(" (%s%d.%s%s%s%s)",
-	    NVS_COND_STRING(shader->GetCondition(shader)),
-	    shader->GetCondRegID(shader),
-	    SWZ_STRING(swz[NVS_SWZ_X]),
-	    SWZ_STRING(swz[NVS_SWZ_Y]),
-	    SWZ_STRING(swz[NVS_SWZ_Z]),
-	    SWZ_STRING(swz[NVS_SWZ_W])
-	    );
-   }
-
-   /* looping */
-   if (opi->flags & COUNT_ALL) {
-      printf(" { ");
-      if (opi->flags & COUNT_NUM) {
-	 printf("%d", shader->GetLoopCount(shader));
-      }
-      if (opi->flags & COUNT_IND) {
-	 printf(", %d", shader->GetLoopInitial(shader));
-      }
-      if (opi->flags & COUNT_INC) {
-	 printf(", %d", shader->GetLoopIncrement(shader));
-      }
-      printf(" }");
-   }
-
-   /* branching */
-   if (opi->flags & BRANCH_TR)
-      printf(" %d", shader->GetBranch(shader));
-   if (opi->flags & BRANCH_EL)
-      printf(" ELSE %d", shader->GetBranchElse(shader));
-   if (opi->flags & BRANCH_EN)
-      printf(" END %d", shader->GetBranchEnd(shader));
-
-   if (!(opi->flags & NODS) && opi->numsrc)
-      printf(",");
-   printf(" ");
-
-   for (i = 0; i < opi->numsrc; i++) {
-      if (shader->GetSourceAbs(shader, merged, i))
-	 printf("abs(");
-      if (shader->GetSourceNegate(shader, merged, i))
-	 printf("-");
-
-      file = shader->GetSourceFile(shader, merged, i);
-      switch (file) {
-      case NVS_FILE_TEMP:
-	 printf("R%d", shader->GetSourceID(shader, merged, i));
-	 break;
-      case NVS_FILE_CONST:
-	 if (shader->GetSourceIndexed(shader, merged, i)) {
-	    printf("c[A%d.%s + 0x%x]",
-		     shader->GetRelAddressRegID(shader),
-		     SWZ_STRING(shader->GetRelAddressSwizzle(shader)),
-		     shader->GetSourceID(shader, merged, i)
-		  );
-	 } else {
-	    float val[4];
-
-	    if (shader->GetSourceConstVal) {
-	       shader->GetSourceConstVal(shader, merged, i, val);
-	       printf("{ %.02f, %.02f, %.02f, %.02f }",
-		     val[0], val[1], val[2], val[3]);
-	    } else {
-	       printf("c[0x%x]", shader->GetSourceID(shader, merged, i));
-	    }
-	 }
-	 break;
-      case NVS_FILE_ATTRIB:
-	 if (shader->GetSourceIndexed(shader, merged, i)) {
-	    printf("attrib[A%d.%s + %d]",
-		     shader->GetRelAddressRegID(shader),
-		     SWZ_STRING(shader->GetRelAddressSwizzle(shader)),
-		     shader->GetSourceID(shader, merged, i)
-		  );
-	 }
-	 else {
-	    printf("attrib.%s",
-		     SFR_STRING(shader->GetSourceID(shader, merged, i))
-		  );
-	 }
-	 break;
-      case NVS_FILE_ADDRESS:
-	 printf("A%d", shader->GetRelAddressRegID(shader));
-	 break;
-      default:
-	 printf("UNKNOWN_SRC_FILE");
-	 break;
-      }
-
-      shader->GetSourceSwizzle(shader, merged, i, swz);
-      if (file != NVS_FILE_ADDRESS &&
-	  (swz[NVS_SWZ_X] != NVS_SWZ_X || swz[NVS_SWZ_Y] != NVS_SWZ_Y ||
-	   swz[NVS_SWZ_Z] != NVS_SWZ_Z || swz[NVS_SWZ_W] != NVS_SWZ_W)) {
-	 printf(".%s%s%s%s", SWZ_STRING(swz[NVS_SWZ_X]),
-	       		     SWZ_STRING(swz[NVS_SWZ_Y]),
-			     SWZ_STRING(swz[NVS_SWZ_Z]),
-			     SWZ_STRING(swz[NVS_SWZ_W]));
-      }
-
-      if (shader->GetSourceAbs(shader, merged, i))
-	 printf(")");
-      if (shader->GetSourceScale) {
-	 int scale = shader->GetSourceScale(shader, merged, i);
-	 if (scale > 1)
-	    printf("{scaled %dx}", scale);
-      }
-      if (i < (opi->numsrc - 1))
-	 printf(", ");
-   }
-
-   if (shader->IsLastInst(shader))
-      printf(" + END");
-}
-
-void
-nvsDisasmHWShader(nvsPtr nvs)
-{
-   nvsFunc *shader = nvs->func;
-   unsigned int iaddr = 0;
-
-   if (!nvs->program) {
-      fprintf(stderr, "No HW program present");
-      return;
-   }
-
-   shader->inst = nvs->program;
-   while (1) {
-      if (shader->inst >= (nvs->program + nvs->program_size)) {
-	 fprintf(stderr, "Reached end of program, but HW inst has no END");
-	 break;
-      }
-
-      printf("\t0x%08x:\n", shader->inst[0]);
-      printf("\t0x%08x:\n", shader->inst[1]);
-      printf("\t0x%08x:\n", shader->inst[2]);
-      printf("\t0x%08x:", shader->inst[3]);
-
-      printf("\n\t\tINST %d.0: ", iaddr);
-      nvsDisasmHWShaderOp(shader, 0);
-      if (shader->HasMergedInst(shader)) {
-	 printf("\n\t\tINST %d.1: ", iaddr);
-	 nvsDisasmHWShaderOp(shader, 1);
-      }
-      printf("\n");
-
-      if (shader->IsLastInst(shader))
-	 break;
-
-      shader->inst += shader->GetOffsetNext(shader);
-      iaddr++;
-   }
-
-   printf("\n");
-}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.h b/src/mesa/drivers/dri/nouveau/nouveau_shader.h
deleted file mode 100644
index 7125a2ae821..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.h
+++ /dev/null
@@ -1,454 +0,0 @@
-#ifndef __SHADER_COMMON_H__
-#define __SHADER_COMMON_H__
-
-#include "mtypes.h"
-#include "bufferobj.h"
-
-#define NVSDBG(fmt, args...) do {                             \
-	if (NOUVEAU_DEBUG & DEBUG_SHADERS) {                  \
-		fprintf(stderr, "%s: "fmt, __func__, ##args); \
-	}                                                     \
-} while(0)
-
-typedef struct _nvsFunc nvsFunc;
-
-#define NVS_MAX_TEMPS   32
-#define NVS_MAX_ATTRIBS 16
-#define NVS_MAX_CONSTS  256
-#define NVS_MAX_ADDRESS 2
-#define NVS_MAX_INSNS   4096
-
-typedef struct _nvs_fragment_header {
-   struct _nvs_fragment_header *parent;
-   struct _nvs_fragment_header *prev;
-   struct _nvs_fragment_header *next;
-   enum {
-      NVS_INSTRUCTION,
-      NVS_BRANCH,
-      NVS_LOOP,
-      NVS_SUBROUTINE
-   } type;
-} nvsFragmentHeader;
-
-typedef union {
-	struct {
-		GLboolean uses_kil;
-		GLuint    num_regs;
-	} NV30FP;
-	struct {
-		uint32_t vp_in_reg;
-		uint32_t vp_out_reg;
-		uint32_t clip_enables;
-	} NV30VP;
-} nvsCardPriv;
-
-typedef struct _nouveauShader {
-   union {
-      struct gl_vertex_program vp;
-      struct gl_fragment_program fp;
-   } mesa;
-   GLcontext *ctx;
-   nvsFunc *func;
-
-   /* State of the final program */
-   GLboolean error;
-   GLboolean translated;
-   GLboolean on_hardware;
-   unsigned int *program;
-   unsigned int program_size;
-   unsigned int program_alloc_size;
-   unsigned int program_start_id;
-   unsigned int program_current;
-   struct gl_buffer_object *program_buffer;
-   int		inst_count;
-
-   nvsCardPriv card_priv;
-   int		vp_attrib_map[NVS_MAX_ATTRIBS];
-
-   struct {
-      GLboolean in_use;
-
-      GLfloat  *source_val;	/* NULL if invariant */
-      float	val[4];
-      /* Hardware-specific tracking, currently only nv30_fragprog
-       * makes use of it.
-       */
-      int	*hw_index;
-      int        hw_index_cnt;
-   } params[NVS_MAX_CONSTS];
-   int param_high;
-
-   /* Pass-private data */
-   void *pass_rec;
-
-   nvsFragmentHeader *program_tree;
-} nouveauShader, *nvsPtr;
-
-typedef enum {
-   NVS_FILE_NONE,
-   NVS_FILE_TEMP,
-   NVS_FILE_ATTRIB,
-   NVS_FILE_CONST,
-   NVS_FILE_RESULT,
-   NVS_FILE_ADDRESS,
-   NVS_FILE_UNKNOWN
-} nvsRegFile;
-
-typedef enum {
-   NVS_OP_UNKNOWN = 0,
-   NVS_OP_NOP,
-   NVS_OP_ABS, NVS_OP_ADD, NVS_OP_ARA, NVS_OP_ARL, NVS_OP_ARR,
-   NVS_OP_BRA, NVS_OP_BRK,
-   NVS_OP_CAL, NVS_OP_CMP, NVS_OP_COS,
-   NVS_OP_DDX, NVS_OP_DDY, NVS_OP_DIV, NVS_OP_DP2, NVS_OP_DP2A, NVS_OP_DP3,
-   NVS_OP_DP4, NVS_OP_DPH, NVS_OP_DST,
-   NVS_OP_EX2, NVS_OP_EXP,
-   NVS_OP_FLR, NVS_OP_FRC,
-   NVS_OP_IF,
-   NVS_OP_KIL,
-   NVS_OP_LG2, NVS_OP_LIT, NVS_OP_LOG, NVS_OP_LOOP, NVS_OP_LRP,
-   NVS_OP_MAD, NVS_OP_MAX, NVS_OP_MIN, NVS_OP_MOV, NVS_OP_MUL,
-   NVS_OP_NRM,
-   NVS_OP_PK2H, NVS_OP_PK2US, NVS_OP_PK4B, NVS_OP_PK4UB, NVS_OP_POW,
-   NVS_OP_POPA, NVS_OP_PUSHA,
-   NVS_OP_RCC, NVS_OP_RCP, NVS_OP_REP, NVS_OP_RET, NVS_OP_RFL, NVS_OP_RSQ,
-   NVS_OP_SCS, NVS_OP_SEQ, NVS_OP_SFL, NVS_OP_SGE, NVS_OP_SGT, NVS_OP_SIN,
-   NVS_OP_SLE, NVS_OP_SLT, NVS_OP_SNE, NVS_OP_SSG, NVS_OP_STR, NVS_OP_SUB,
-   NVS_OP_SWZ,
-   NVS_OP_TEX, NVS_OP_TXB, NVS_OP_TXD, NVS_OP_TXL, NVS_OP_TXP,
-   NVS_OP_UP2H, NVS_OP_UP2US, NVS_OP_UP4B, NVS_OP_UP4UB,
-   NVS_OP_X2D, NVS_OP_XPD,
-   NVS_OP_EMUL
-} nvsOpcode;
-
-typedef enum {
-   NVS_PREC_FLOAT32,
-   NVS_PREC_FLOAT16,
-   NVS_PREC_FIXED12,
-   NVS_PREC_UNKNOWN
-} nvsPrecision;
-
-typedef enum {
-   NVS_SWZ_X = 0,
-   NVS_SWZ_Y = 1,
-   NVS_SWZ_Z = 2,
-   NVS_SWZ_W = 3
-} nvsSwzComp;
-
-typedef enum {
-   NVS_FR_POSITION	= 0,
-   NVS_FR_WEIGHT	= 1,
-   NVS_FR_NORMAL	= 2,
-   NVS_FR_COL0		= 3,
-   NVS_FR_COL1		= 4,
-   NVS_FR_FOGCOORD	= 5,
-   NVS_FR_TEXCOORD0	= 8,
-   NVS_FR_TEXCOORD1	= 9,
-   NVS_FR_TEXCOORD2	= 10,
-   NVS_FR_TEXCOORD3	= 11,
-   NVS_FR_TEXCOORD4	= 12,
-   NVS_FR_TEXCOORD5	= 13,
-   NVS_FR_TEXCOORD6	= 14,
-   NVS_FR_TEXCOORD7	= 15,
-   NVS_FR_BFC0		= 16,
-   NVS_FR_BFC1		= 17,
-   NVS_FR_POINTSZ	= 18,
-   NVS_FR_FRAGDATA0	= 19,
-   NVS_FR_FRAGDATA1	= 20,
-   NVS_FR_FRAGDATA2	= 21,
-   NVS_FR_FRAGDATA3	= 22,
-   NVS_FR_CLIP0		= 23,
-   NVS_FR_CLIP1		= 24,
-   NVS_FR_CLIP2		= 25,
-   NVS_FR_CLIP3		= 26,
-   NVS_FR_CLIP4		= 27,
-   NVS_FR_CLIP5		= 28,
-   NVS_FR_CLIP6		= 29,
-   NVS_FR_FACING	= 30,
-   NVS_FR_UNKNOWN
-} nvsFixedReg;
-
-typedef enum {
-   NVS_COND_FL, NVS_COND_LT, NVS_COND_EQ, NVS_COND_LE, NVS_COND_GT,
-   NVS_COND_NE, NVS_COND_GE, NVS_COND_TR, NVS_COND_UN,
-   NVS_COND_UNKNOWN
-} nvsCond;
-
-typedef struct {
-   nvsRegFile	file;
-   unsigned int		index;
-
-   unsigned int		indexed;
-   unsigned int		addr_reg;
-   nvsSwzComp		addr_comp;
-
-   nvsSwzComp		swizzle[4];
-   int			negate;
-   int			abs;
-} nvsRegister;
-
-static const nvsRegister nvr_unused = {
-   .file	= NVS_FILE_ATTRIB,
-   .index	= 0,
-   .indexed	= 0,
-   .addr_reg	= 0,
-   .addr_comp	= NVS_SWZ_X,
-   .swizzle	= {NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W},
-   .negate	= 0,
-   .abs		= 0,
-};
-
-typedef enum {
-   NVS_TEX_TARGET_1D,
-   NVS_TEX_TARGET_2D,
-   NVS_TEX_TARGET_3D,
-   NVS_TEX_TARGET_CUBE,
-   NVS_TEX_TARGET_RECT,
-   NVS_TEX_TARGET_UNKNOWN = 0
-} nvsTexTarget;
-
-typedef enum {
-	NVS_SCALE_1X	 = 0,
-	NVS_SCALE_2X	 = 1,
-	NVS_SCALE_4X	 = 2,
-	NVS_SCALE_8X	 = 3,
-	NVS_SCALE_INV_2X = 5,
-	NVS_SCALE_INV_4X = 6,
-	NVS_SCALE_INV_8X = 7,
-} nvsScale;
-
-/* Arith/TEX instructions */
-typedef struct nvs_instruction {
-   nvsFragmentHeader header;
-
-   nvsOpcode	op;
-   unsigned int saturate;
-
-   nvsRegister	dest;
-   unsigned int	mask;
-   nvsScale	dest_scale;
-
-   nvsRegister	src[3];
-
-   unsigned int tex_unit;
-   nvsTexTarget tex_target;
-
-   nvsCond	cond;
-   nvsSwzComp	cond_swizzle[4];
-   int		cond_reg;
-   int		cond_test;
-   int		cond_update;
-} nvsInstruction;
-
-/* BRA, CAL, IF */
-typedef struct nvs_branch {
-	nvsFragmentHeader  header;
-
-	nvsOpcode	op;
-
-	nvsCond		cond;
-	nvsSwzComp	cond_swizzle[4];
-	int		cond_test;
-
-	nvsFragmentHeader *target_head;
-	nvsFragmentHeader *target_tail;
-	nvsFragmentHeader *else_head;
-	nvsFragmentHeader *else_tail;
-} nvsBranch;
-
-/* LOOP+ENDLOOP */
-typedef struct {
-	nvsFragmentHeader  header;
-
-	int                count;
-	int                initial;
-	int                increment;
-
-	nvsFragmentHeader *insn_head;
-	nvsFragmentHeader *insn_tail;
-} nvsLoop;
-
-/* label+following instructions */
-typedef struct nvs_subroutine {
-	nvsFragmentHeader  header;
-
-	char *             label;
-	nvsFragmentHeader *insn_head;
-	nvsFragmentHeader *insn_tail;
-} nvsSubroutine;
-
-#define SMASK_X (1<<0)
-#define SMASK_Y (1<<1)
-#define SMASK_Z (1<<2)
-#define SMASK_W (1<<3)
-#define SMASK_ALL (SMASK_X|SMASK_Y|SMASK_Z|SMASK_W)
-
-#define SPOS_ADDRESS 3
-struct _op_xlat {
-	unsigned int	NV;
-	nvsOpcode	SOP;
-	int		srcpos[3];
-};
-#define MOD_OPCODE(t,hw,sop,s0,s1,s2) do { \
-	t[hw].NV = hw; \
-	t[hw].SOP = sop; \
-	t[hw].srcpos[0] = s0; \
-	t[hw].srcpos[1] = s1; \
-	t[hw].srcpos[2] = s2; \
-} while(0)
-
-extern unsigned int NVVP_TX_VOP_COUNT;
-extern unsigned int NVVP_TX_NVS_OP_COUNT;
-extern struct _op_xlat NVVP_TX_VOP[];
-extern struct _op_xlat NVVP_TX_SOP[];
-
-extern unsigned int NVFP_TX_AOP_COUNT;
-extern unsigned int NVFP_TX_BOP_COUNT;
-extern struct _op_xlat NVFP_TX_AOP[];
-extern struct _op_xlat NVFP_TX_BOP[];
-
-extern void NV20VPTXSwizzle(int hwswz, nvsSwzComp *swz);
-extern nvsSwzComp NV20VP_TX_SWIZZLE[4];
-
-#define SCAP_SRC_ABS	(1<<0)
-
-struct _nvsFunc {
-   nvsCardPriv *card_priv;
-
-   unsigned int	MaxInst;
-   unsigned int	MaxAttrib;
-   unsigned int	MaxTemp;
-   unsigned int	MaxAddress;
-   unsigned int	MaxConst;
-   unsigned int	caps;
-
-   unsigned int *inst;
-   void		(*UploadToHW)		(GLcontext *, nouveauShader *);
-   void		(*UpdateConst)		(GLcontext *, nouveauShader *, int);
-
-   struct _op_xlat*(*GetOPTXRec)	(nvsFunc *, int merged);
-   struct _op_xlat*(*GetOPTXFromSOP)	(nvsOpcode, int *id);
-
-   void		(*InitInstruction)	(nvsFunc *);
-   int		(*SupportsOpcode)	(nvsFunc *, nvsOpcode);
-   int		(*SupportsResultScale)	(nvsFunc *, nvsScale);
-   void		(*SetOpcode)		(nvsFunc *, unsigned int opcode,
-	 				 int slot);
-   void		(*SetCCUpdate)		(nvsFunc *);
-   void		(*SetCondition)		(nvsFunc *, int on, nvsCond, int reg,
-	 				 nvsSwzComp *swizzle);
-   void		(*SetResult)		(nvsFunc *, nvsRegister *,
-	 				 unsigned int mask, int slot);
-   void		(*SetResultScale)	(nvsFunc *, nvsScale);
-   void		(*SetSource)		(nvsFunc *, nvsRegister *, int pos);
-   void		(*SetTexImageUnit)	(nvsFunc *, int unit);
-   void		(*SetSaturate)		(nvsFunc *);
-   void		(*SetLastInst)		(nvsFunc *);
-
-   void		(*SetBranchTarget)	(nvsFunc *, int addr);
-   void		(*SetBranchElse)	(nvsFunc *, int addr);
-   void		(*SetBranchEnd)		(nvsFunc *, int addr);
-   void		(*SetLoopParams)	(nvsFunc *, int cnt, int init, int inc);
-
-   int		(*HasMergedInst)	(nvsFunc *);
-   int		(*IsLastInst)		(nvsFunc *);
-   int		(*GetOffsetNext)	(nvsFunc *);
-
-   int		(*GetOpcodeSlot)	(nvsFunc *, int merged);
-   unsigned int	(*GetOpcodeHW)		(nvsFunc *, int slot);
-   nvsOpcode	(*GetOpcode)		(nvsFunc *, int merged);
-
-   nvsPrecision	(*GetPrecision)		(nvsFunc *);
-   int		(*GetSaturate)		(nvsFunc *);
-
-   nvsRegFile	(*GetDestFile)		(nvsFunc *, int merged);
-   unsigned int	(*GetDestID)		(nvsFunc *, int merged);
-   unsigned int	(*GetDestMask)		(nvsFunc *, int merged);
-
-   unsigned int	(*GetSourceHW)		(nvsFunc *, int merged, int pos);
-   nvsRegFile	(*GetSourceFile)	(nvsFunc *, int merged, int pos);
-   int		(*GetSourceID)		(nvsFunc *, int merged, int pos);
-   int		(*GetTexImageUnit)	(nvsFunc *);
-   int		(*GetSourceNegate)	(nvsFunc *, int merged, int pos);
-   int		(*GetSourceAbs)		(nvsFunc *, int merged, int pos);
-   void		(*GetSourceSwizzle)	(nvsFunc *, int merged, int pos,
-	 				 nvsSwzComp *swz);
-   int		(*GetSourceIndexed)	(nvsFunc *, int merged, int pos);
-   void		(*GetSourceConstVal)	(nvsFunc *, int merged, int pos,
-	 				 float *val);
-   int		(*GetSourceScale)	(nvsFunc *, int merged, int pos);
-
-   int		(*GetRelAddressRegID)	(nvsFunc *);
-   nvsSwzComp	(*GetRelAddressSwizzle)	(nvsFunc *);
-
-   int		(*SupportsConditional)	(nvsFunc *);
-   int		(*GetConditionUpdate)	(nvsFunc *);
-   int		(*GetConditionTest)	(nvsFunc *);
-   nvsCond	(*GetCondition)		(nvsFunc *);
-   void		(*GetCondRegSwizzle)	(nvsFunc *, nvsSwzComp *swz);
-   int		(*GetCondRegID)		(nvsFunc *);
-   int		(*GetBranch)		(nvsFunc *);
-   int		(*GetBranchElse)	(nvsFunc *);
-   int		(*GetBranchEnd)		(nvsFunc *);
-
-   int		(*GetLoopCount)		(nvsFunc *);
-   int		(*GetLoopInitial)	(nvsFunc *);
-   int		(*GetLoopIncrement)	(nvsFunc *);
-};
-
-static inline nvsRegister
-nvsNegate(nvsRegister reg)
-{
-   reg.negate = !reg.negate;
-   return reg;
-}
-
-static inline nvsRegister
-nvsAbs(nvsRegister reg)
-{
-   reg.abs = 1;
-   return reg;
-}
-
-static inline nvsRegister
-nvsSwizzle(nvsRegister reg, nvsSwzComp x, nvsSwzComp y,
-      	   nvsSwzComp z, nvsSwzComp w)
-{
-   nvsSwzComp sc[4] = { x, y, z, w };
-   nvsSwzComp oc[4];
-   int i;
-
-   for (i=0;i<4;i++)
-      oc[i] = reg.swizzle[i];
-   for (i=0;i<4;i++)
-      reg.swizzle[i] = oc[sc[i]];
-   return reg;
-}
-
-#define nvsProgramError(nvs,fmt,args...) do {                           \
-	fprintf(stderr, "nvsProgramError (%s): "fmt, __func__, ##args); \
-	(nvs)->error = GL_TRUE;                                         \
-	(nvs)->translated = GL_FALSE;                                   \
-} while(0)
-
-extern GLboolean nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs);
-extern void nvsDisasmHWShader(nvsPtr);
-extern void nvsDumpFragmentList(nvsFragmentHeader *f, int lvl);
-extern nouveauShader *nvsBuildTextShader(GLcontext *ctx, GLenum target,
-      					 const char *text);
-
-extern void NV20VPInitShaderFuncs(nvsFunc *);
-extern void NV30VPInitShaderFuncs(nvsFunc *);
-extern void NV40VPInitShaderFuncs(nvsFunc *);
-
-extern void NV30FPInitShaderFuncs(nvsFunc *);
-extern void NV40FPInitShaderFuncs(nvsFunc *);
-
-extern void nouveauShaderInitFuncs(GLcontext *ctx);
-
-extern GLboolean nouveau_shader_pass0(GLcontext *ctx, nouveauShader *nvs);
-extern GLboolean nouveau_shader_pass1(nvsPtr nvs);
-extern GLboolean nouveau_shader_pass2(nvsPtr nvs);
-
-#endif
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c
deleted file mode 100644
index 8c203cc664b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c
+++ /dev/null
@@ -1,1050 +0,0 @@
-/*
- * Copyright (C) 2006 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/*
- * Authors:
- *   Ben Skeggs <darktama@iinet.net.au>
- */
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "shader/prog_instruction.h"
-#include "shader/prog_parameter.h"
-#include "shader/prog_statevars.h"
-#include "shader/programopt.h"
-
-#include "nouveau_context.h"
-#include "nouveau_shader.h"
-#include "nouveau_msg.h"
-
-static nvsFixedReg _tx_mesa_vp_dst_reg[VERT_RESULT_MAX] = {
-   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7,
-   NVS_FR_POINTSZ, NVS_FR_BFC0, NVS_FR_BFC1, NVS_FR_UNKNOWN /* EDGE */
-};
-
-static nvsFixedReg _tx_mesa_fp_dst_reg[FRAG_RESULT_MAX] = {
-   NVS_FR_FRAGDATA0 /* COLR */, NVS_FR_FRAGDATA0 /* COLH */,
-   NVS_FR_UNKNOWN /* DEPR */
-};
-
-static nvsFixedReg _tx_mesa_fp_src_reg[FRAG_ATTRIB_MAX] = {
-   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7
-};
-
-static nvsSwzComp _tx_mesa_swizzle[4] = {
-   NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W
-};
-
-static nvsOpcode _tx_mesa_opcode[] = {
-   [OPCODE_ABS] = NVS_OP_ABS, [OPCODE_ADD] = NVS_OP_ADD,
-   [OPCODE_ARA] = NVS_OP_ARA, [OPCODE_ARL] = NVS_OP_ARL,
-   [OPCODE_ARL_NV] = NVS_OP_ARL, [OPCODE_ARR] = NVS_OP_ARR,
-   [OPCODE_CMP] = NVS_OP_CMP, [OPCODE_COS] = NVS_OP_COS,
-   [OPCODE_DDX] = NVS_OP_DDX, [OPCODE_DDY] = NVS_OP_DDY,
-   [OPCODE_DP3] = NVS_OP_DP3, [OPCODE_DP4] = NVS_OP_DP4,
-   [OPCODE_DPH] = NVS_OP_DPH, [OPCODE_DST] = NVS_OP_DST,
-   [OPCODE_EX2] = NVS_OP_EX2, [OPCODE_EXP] = NVS_OP_EXP,
-   [OPCODE_FLR] = NVS_OP_FLR, [OPCODE_FRC] = NVS_OP_FRC,
-   [OPCODE_KIL] = NVS_OP_EMUL, [OPCODE_KIL_NV] = NVS_OP_KIL,
-   [OPCODE_LG2] = NVS_OP_LG2, [OPCODE_LIT] = NVS_OP_LIT,
-   [OPCODE_LOG] = NVS_OP_LOG,
-   [OPCODE_LRP] = NVS_OP_LRP,
-   [OPCODE_MAD] = NVS_OP_MAD, [OPCODE_MAX] = NVS_OP_MAX,
-   [OPCODE_MIN] = NVS_OP_MIN, [OPCODE_MOV] = NVS_OP_MOV,
-   [OPCODE_MUL] = NVS_OP_MUL,
-   [OPCODE_PK2H] = NVS_OP_PK2H, [OPCODE_PK2US] = NVS_OP_PK2US,
-   [OPCODE_PK4B] = NVS_OP_PK4B, [OPCODE_PK4UB] = NVS_OP_PK4UB,
-   [OPCODE_POW] = NVS_OP_POW, [OPCODE_POPA] = NVS_OP_POPA,
-   [OPCODE_PUSHA] = NVS_OP_PUSHA,
-   [OPCODE_RCC] = NVS_OP_RCC, [OPCODE_RCP] = NVS_OP_RCP,
-   [OPCODE_RFL] = NVS_OP_RFL, [OPCODE_RSQ] = NVS_OP_RSQ,
-   [OPCODE_SCS] = NVS_OP_SCS, [OPCODE_SEQ] = NVS_OP_SEQ,
-   [OPCODE_SFL] = NVS_OP_SFL, [OPCODE_SGE] = NVS_OP_SGE,
-   [OPCODE_SGT] = NVS_OP_SGT, [OPCODE_SIN] = NVS_OP_SIN,
-   [OPCODE_SLE] = NVS_OP_SLE, [OPCODE_SLT] = NVS_OP_SLT,
-   [OPCODE_SNE] = NVS_OP_SNE, [OPCODE_SSG] = NVS_OP_SSG,
-   [OPCODE_STR] = NVS_OP_STR, [OPCODE_SUB] = NVS_OP_SUB,
-   [OPCODE_SWZ] = NVS_OP_MOV,
-   [OPCODE_TEX] = NVS_OP_TEX, [OPCODE_TXB] = NVS_OP_TXB,
-   [OPCODE_TXD] = NVS_OP_TXD,
-   [OPCODE_TXL] = NVS_OP_TXL, [OPCODE_TXP] = NVS_OP_TXP,
-   [OPCODE_TXP_NV] = NVS_OP_TXP,
-   [OPCODE_UP2H] = NVS_OP_UP2H, [OPCODE_UP2US] = NVS_OP_UP2US,
-   [OPCODE_UP4B] = NVS_OP_UP4B, [OPCODE_UP4UB] = NVS_OP_UP4UB,
-   [OPCODE_X2D] = NVS_OP_X2D,
-   [OPCODE_XPD] = NVS_OP_XPD
-};
-
-static nvsCond _tx_mesa_condmask[] = {
-   NVS_COND_TR, /* workaround mesa not filling a valid value */
-   NVS_COND_GT, NVS_COND_LT, NVS_COND_UN, NVS_COND_GE,
-   NVS_COND_LE, NVS_COND_NE, NVS_COND_NE, NVS_COND_TR, NVS_COND_FL
-};
-
-struct pass0_rec {
-   int nvs_ipos;
-   int next_temp;
-
-   int mesa_const_base;
-   int mesa_const_last;
-
-   int swzconst_done;
-   int swzconst_id;
-   nvsRegister const_half;
-};
-
-#define X NVS_SWZ_X
-#define Y NVS_SWZ_Y
-#define Z NVS_SWZ_Z
-#define W NVS_SWZ_W
-
-#define FILL_CONDITION_FLAGS(fragment) do {                                    \
-	(fragment)->cond =                                                     \
-	pass0_make_condmask(inst->DstReg.CondMask);                            \
-	if ((fragment)->cond != NVS_COND_TR)                                   \
-		(fragment)->cond_test = 1;                                     \
-	(fragment)->cond_reg = inst->CondDst;                                  \
-	pass0_make_swizzle((fragment)->cond_swizzle, inst->DstReg.CondSwizzle);\
-} while(0)
-
-#define ARITH(op,dest,mask,sat,s0,s1,s2) do {               \
-	nvsinst = pass0_emit(nvs, parent, fpos, (op),       \
-		   (dest), (mask), (sat), (s0), (s1), (s2));\
-	FILL_CONDITION_FLAGS(nvsinst);                      \
-} while(0)
-
-#define ARITHu(op,dest,mask,sat,s0,s1,s2) do {              \
-	nvsinst = pass0_emit(nvs, parent, fpos, (op),       \
-		   (dest), (mask), (sat), (s0), (s1), (s2));\
-} while(0)
-
-static void
-pass0_append_fragment(nvsFragmentHeader *parent,
-		      nvsFragmentHeader *fragment,
-		      int pos)
-{
-	nvsFragmentHeader **head, **tail;
-	assert(parent && fragment);
-
-	switch (parent->type) {
-	case NVS_BRANCH:
-		if (pos == 0) {
-			head = &((nvsBranch *)parent)->target_head;
-			tail = &((nvsBranch *)parent)->target_tail;
-		} else {
-			head = &((nvsBranch *)parent)->else_head;
-			tail = &((nvsBranch *)parent)->else_tail;
-		}
-		break;
-	case NVS_LOOP:
-		head = &((nvsLoop *)parent)->insn_head;
-		tail = &((nvsLoop *)parent)->insn_tail;
-		break;
-	case NVS_SUBROUTINE:
-		head = &((nvsSubroutine *)parent)->insn_head;
-		tail = &((nvsSubroutine *)parent)->insn_tail;
-		break;
-	default:
-		assert(0);
-		break;
-	}
-
-	fragment->parent = parent;
-	fragment->prev   = *tail;
-	fragment->next   = NULL;
-	if (!(*head))
-		*head = fragment;
-	else
-		(*tail)->next = fragment;
-	*tail = fragment;
-
-}
-
-static nvsSubroutine *
-pass0_create_subroutine(nouveauShader *nvs, const char *label)
-{
-	nvsSubroutine *sub;
-
-	sub = CALLOC_STRUCT(nvs_subroutine);
-	if (sub) {
-		sub->header.type = NVS_SUBROUTINE;
-		sub->label	 = strdup(label);
-		if (!nvs->program_tree)
-			nvs->program_tree = &sub->header;
-		else
-			pass0_append_fragment(nvs->program_tree,
-					      &sub->header, 0);
-	}
-
-	return sub;
-}
-
-static void
-pass0_make_reg(nouveauShader *nvs, nvsRegister *reg,
-	       nvsRegFile file, unsigned int index)
-{
-	struct pass0_rec *rec = nvs->pass_rec;
-
-	/* defaults */
-	*reg = nvr_unused;
-	/* -1 == quick-and-dirty temp alloc */
-	if (file == NVS_FILE_TEMP && index == -1) {
-		index = rec->next_temp++;
-		assert(index < NVS_MAX_TEMPS);
-	}
-	reg->file   = file;
-	reg->index  = index;
-}
-
-static void
-pass0_make_swizzle(nvsSwzComp *swz, unsigned int mesa)
-{
-	int i;
-
-	for (i=0;i<4;i++)
-		swz[i] = _tx_mesa_swizzle[GET_SWZ(mesa, i)];
-}
-
-static nvsOpcode
-pass0_make_opcode(enum prog_opcode op)
-{
-	if (op > MAX_OPCODE)
-		return NVS_OP_UNKNOWN;
-	return _tx_mesa_opcode[op];
-}
-
-static nvsCond
-pass0_make_condmask(GLuint mesa)
-{
-	if (mesa > COND_FL)
-		return NVS_COND_UNKNOWN;
-	return _tx_mesa_condmask[mesa];
-}
-
-static unsigned int
-pass0_make_mask(GLuint mesa_mask)
-{
-	unsigned int mask = 0;
-
-	if (mesa_mask & WRITEMASK_X) mask |= SMASK_X;
-	if (mesa_mask & WRITEMASK_Y) mask |= SMASK_Y;
-	if (mesa_mask & WRITEMASK_Z) mask |= SMASK_Z;
-	if (mesa_mask & WRITEMASK_W) mask |= SMASK_W;
-
-	return mask;
-}
-
-static GLboolean
-pass0_opcode_is_tex(enum prog_opcode op)
-{
-	switch (op) {
-	case OPCODE_TEX:
-	case OPCODE_TXB:
-	case OPCODE_TXD:
-	case OPCODE_TXL:
-	case OPCODE_TXP:
-		return GL_TRUE;
-	default:
-		break;
-	}
-
-	return GL_FALSE;
-}
-
-static nvsTexTarget
-pass0_make_tex_target(GLuint mesa)
-{
-	switch (mesa) {
-	case TEXTURE_1D_INDEX: return NVS_TEX_TARGET_1D;
-	case TEXTURE_2D_INDEX: return NVS_TEX_TARGET_2D;
-	case TEXTURE_3D_INDEX: return NVS_TEX_TARGET_3D;
-	case TEXTURE_CUBE_INDEX: return NVS_TEX_TARGET_CUBE;
-	case TEXTURE_RECT_INDEX: return NVS_TEX_TARGET_RECT;
-	default:
-		return NVS_TEX_TARGET_UNKNOWN;
-	}
-}
-
-static void
-pass0_make_dst_reg(nvsPtr nvs, nvsRegister *reg,
-      		   struct prog_dst_register *dst)
-{
-	struct gl_program *mesa = (struct gl_program*)&nvs->mesa.vp;
-	nvsFixedReg sfr;
-
-	switch (dst->File) {
-	case PROGRAM_OUTPUT:
-		if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
-			sfr = (dst->Index < VERT_RESULT_MAX) ?
-				_tx_mesa_vp_dst_reg[dst->Index] :
-				NVS_FR_UNKNOWN;
-		} else {
-			sfr = (dst->Index < FRAG_RESULT_MAX) ?
-				_tx_mesa_fp_dst_reg[dst->Index] :
-				NVS_FR_UNKNOWN;
-		}
-		pass0_make_reg(nvs, reg, NVS_FILE_RESULT, sfr);
-		break;
-	case PROGRAM_TEMPORARY:
-		pass0_make_reg(nvs, reg, NVS_FILE_TEMP, dst->Index);
-		break;
-	case PROGRAM_ADDRESS:
-		pass0_make_reg(nvs, reg, NVS_FILE_ADDRESS, dst->Index);
-		break;
-	default:
-		fprintf(stderr, "Unknown dest file %d\n", dst->File);
-		assert(0);
-	}
-}
-
-static void
-pass0_make_src_reg(nvsPtr nvs, nvsRegister *reg, struct prog_src_register *src)
-{
-	struct pass0_rec *rec = nvs->pass_rec;
-	struct gl_program *mesa = (struct gl_program *)&nvs->mesa.vp.Base;
-	int i;
-
-	*reg = nvr_unused;
-
-	switch (src->File) {
-	case PROGRAM_INPUT:
-		reg->file = NVS_FILE_ATTRIB;
-		if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
-			for (i=0; i<NVS_MAX_ATTRIBS; i++) {
-				if (nvs->vp_attrib_map[i] == src->Index) {
-					reg->index = i;
-					break;
-				}
-			}
-			if (i==NVS_MAX_ATTRIBS)
-				reg->index = NVS_FR_UNKNOWN;
-		} else {
-			reg->index = (src->Index < FRAG_ATTRIB_MAX) ?
-				_tx_mesa_fp_src_reg[src->Index] :
-				NVS_FR_UNKNOWN;
-		}
-		break;
-	case PROGRAM_STATE_VAR:
-	case PROGRAM_NAMED_PARAM:
-	case PROGRAM_CONSTANT:
-		reg->file    = NVS_FILE_CONST;
-		reg->index   = src->Index + rec->mesa_const_base;
-		reg->indexed = src->RelAddr;
-		if (reg->indexed) {
-			reg->addr_reg  = 0;
-			reg->addr_comp = NVS_SWZ_X;
-		}
-		break;
-	case PROGRAM_TEMPORARY:
-		reg->file  = NVS_FILE_TEMP;
-		reg->index = src->Index;
-		break;
-	default:
-		fprintf(stderr, "Unknown source type %d\n", src->File);
-		assert(0);
-	}
-
-	/* per-component negate handled elsewhere */
-	reg->negate = src->NegateBase != 0;
-	reg->abs    = src->Abs;
-	pass0_make_swizzle(reg->swizzle, src->Swizzle);
-}
-
-static nvsInstruction *
-pass0_emit(nouveauShader *nvs, nvsFragmentHeader *parent, int fpos,
-	   nvsOpcode op, nvsRegister dst,
-	   unsigned int mask, int saturate,
-	   nvsRegister src0, nvsRegister src1, nvsRegister src2)
-{
-	nvsInstruction *sif;
-
-	sif = CALLOC_STRUCT(nvs_instruction);
-	if (!sif)
-		return NULL;
-
-	/* Seems mesa doesn't explicitly 0 this.. */
-	if (nvs->mesa.vp.Base.Target == GL_VERTEX_PROGRAM_ARB)
-		saturate = 0;
-
-	sif->op		= op;
-	sif->saturate	= saturate;
-	sif->dest	= dst;
-	sif->mask	= mask;
-	sif->dest_scale	= NVS_SCALE_1X;
-	sif->src[0]	= src0;
-	sif->src[1]	= src1;
-	sif->src[2]	= src2;
-	sif->cond	= COND_TR;
-	sif->cond_reg	= 0;
-	sif->cond_test	= 0;
-	sif->cond_update= 0;
-	pass0_make_swizzle(sif->cond_swizzle, SWIZZLE_NOOP);
-	pass0_append_fragment(parent, &sif->header, fpos);
-
-	return sif;
-}
-
-static void
-pass0_fixup_swizzle(nvsPtr nvs, nvsFragmentHeader *parent, int fpos,
-		    struct prog_src_register *src,
-		    unsigned int sm1,
-		    unsigned int sm2)
-{
-	static const float sc[4] = { 1.0, 0.0, -1.0, 0.0 };
-	struct pass0_rec *rec = nvs->pass_rec;
-	int fixup_1, fixup_2;
-	nvsInstruction *nvsinst;
-	nvsRegister sr, dr = nvr_unused;
-	nvsRegister sm1const, sm2const;
-
-	if (!rec->swzconst_done) {
-		struct gl_program *prog = &nvs->mesa.vp.Base;
-                GLuint swizzle;
-		rec->swzconst_id = _mesa_add_unnamed_constant(prog->Parameters,
-							      sc, 4, &swizzle);
-                /* XXX what about swizzle? */
-		rec->swzconst_done = 1;
-		COPY_4V(nvs->params[rec->swzconst_id].val, sc);
-	}
-
-	fixup_1 = (sm1 != MAKE_SWIZZLE4(0,0,0,0) &&
-			sm2 != MAKE_SWIZZLE4(2,2,2,2));
-	fixup_2 = (sm2 != MAKE_SWIZZLE4(2,2,2,2));
-
-	if (src->File != PROGRAM_TEMPORARY && src->File != PROGRAM_INPUT) {
-		/* We can't use more than one const in an instruction,
-		 * so move the const into a temp, and swizzle from there.
-		 *
-		 * TODO: should just emit the swizzled const, instead of
-		 *       swizzling it in the shader.. would need to reswizzle
-		 *       any state params when they change however..
-		 */
-		pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-		pass0_make_src_reg(nvs, &sr, src);
-		ARITHu(NVS_OP_MOV, dr, SMASK_ALL, 0,
-				 sr, nvr_unused, nvr_unused);
-		pass0_make_reg(nvs, &sr, NVS_FILE_TEMP, dr.index);
-	} else {
-		if (fixup_1)
-			src->NegateBase = 0;
-		pass0_make_src_reg(nvs, &sr, src);
-		pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-	}
-
-	pass0_make_reg(nvs, &sm1const, NVS_FILE_CONST, rec->swzconst_id);
-	pass0_make_swizzle(sm1const.swizzle, sm1);
-	if (fixup_1 && fixup_2) {
-		/* Any combination with SWIZZLE_ONE */
-		pass0_make_reg(nvs, &sm2const,
-				    NVS_FILE_CONST, rec->swzconst_id);
-		pass0_make_swizzle(sm2const.swizzle, sm2);
-		ARITHu(NVS_OP_MAD, dr, SMASK_ALL, 0, sr, sm1const, sm2const);
-	} else {
-		/* SWIZZLE_ZERO || arbitrary negate */
-		ARITHu(NVS_OP_MUL, dr, SMASK_ALL, 0, sr, sm1const, nvr_unused);
-	}
-
-	src->File	= PROGRAM_TEMPORARY;
-	src->Index	= dr.index;
-	src->Swizzle	= SWIZZLE_NOOP;
-}
-
-#define SET_SWZ(fs, cp, c) fs = (fs & ~(0x7<<(cp*3))) | (c<<(cp*3))
-static void
-pass0_check_sources(nvsPtr nvs, nvsFragmentHeader *parent, int fpos,
-		    struct prog_instruction *inst)
-{
-	unsigned int insrc = -1, constsrc = -1;
-	int i;
-
-	for (i=0;i<_mesa_num_inst_src_regs(inst->Opcode);i++) {
-		struct prog_src_register *src = &inst->SrcReg[i];
-		unsigned int sm_1 = 0, sm_2 = 0;
-		nvsRegister sr, dr;
-		int do_mov = 0, c;
-
-		/* Build up swizzle masks as if we were going to use
-		 * "MAD new, src, const1, const2" to support arbitrary negation
-		 * and SWIZZLE_ZERO/SWIZZLE_ONE.
-		 */
-		for (c=0;c<4;c++) {
-			if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ZERO) {
-				SET_SWZ(sm_1, c, SWIZZLE_Y); /* 0.0 */
-				SET_SWZ(sm_2, c, SWIZZLE_Y);
-				SET_SWZ(src->Swizzle, c, SWIZZLE_X);
-			} else if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ONE) {
-				SET_SWZ(sm_1, c, SWIZZLE_Y);
-				if (src->NegateBase & (1<<c))
-					SET_SWZ(sm_2, c, SWIZZLE_Z); /* -1.0 */
-				else
-					SET_SWZ(sm_2, c, SWIZZLE_X); /* 1.0 */
-				SET_SWZ(src->Swizzle, c, SWIZZLE_X);
-			} else {
-				if (src->NegateBase & (1<<c))
-					SET_SWZ(sm_1, c, SWIZZLE_Z); /* -[xyzw] */
-				else
-					SET_SWZ(sm_1, c, SWIZZLE_X); /*[xyzw]*/
-				SET_SWZ(sm_2, c, SWIZZLE_Y);
-			}
-		}
-
-		/* Unless we're multiplying by 1.0 or -1.0 on all components,
-		 * and we're adding nothing to any component we have to
-		 * emulate the swizzle.
-		 */
-		if ((sm_1 != MAKE_SWIZZLE4(0,0,0,0) &&
-					sm_1 != MAKE_SWIZZLE4(2,2,2,2)) ||
-				sm_2 != MAKE_SWIZZLE4(1,1,1,1)) {
-			pass0_fixup_swizzle(nvs, parent, fpos, src, sm_1, sm_2);
-			/* The source is definitely in a temp now, so don't
-			 * bother checking  for multiple ATTRIB/CONST regs.
-			 */
-			continue;
-		}
-
-		/* HW can't use more than one ATTRIB or PARAM in a single
-		 * instruction */
-		switch (src->File) {
-		case PROGRAM_INPUT:
-			if (insrc != -1 && insrc != src->Index) 
-				do_mov = 1;
-			else insrc = src->Index;
-			break;
-		case PROGRAM_STATE_VAR:
-			if (constsrc != -1 && constsrc != src->Index)
-				do_mov = 1;
-			else constsrc = src->Index;
-			break;
-		default:
-			break;
-		}
-
-		/* Emit any extra ATTRIB/CONST to a temp, and modify the Mesa
-		 * instruction to point at the temp.
-		 */
-		if (do_mov) {
-			pass0_make_src_reg(nvs, &sr, src);
-			pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-			pass0_emit(nvs, parent, fpos, NVS_OP_MOV,
-					dr, SMASK_ALL, 0,
-					sr, nvr_unused, nvr_unused);
-
-			src->File	= PROGRAM_TEMPORARY;
-			src->Index	= dr.index;
-			src->Swizzle= SWIZZLE_NOOP;
-		}
-	}
-}
-
-static GLboolean
-pass0_emulate_instruction(nouveauShader *nvs,
-			  nvsFragmentHeader *parent, int fpos,
-			  struct prog_instruction *inst)
-{
-	nvsFunc *shader = nvs->func;
-	nvsRegister src[3], dest, temp;
-	nvsInstruction *nvsinst;
-	unsigned int mask = pass0_make_mask(inst->DstReg.WriteMask);
-	int i, sat;
-
-	sat = (inst->SaturateMode == SATURATE_ZERO_ONE);
-
-	/* Build all the "real" regs for the instruction */
-	for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
-		pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
-	if (inst->Opcode != OPCODE_KIL)
-		pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
-
-	switch (inst->Opcode) {
-	case OPCODE_ABS:
-		if (shader->caps & SCAP_SRC_ABS)
-			ARITH(NVS_OP_MOV, dest, mask, sat,
-					nvsAbs(src[0]), nvr_unused, nvr_unused);
-		else
-			ARITH(NVS_OP_MAX, dest, mask, sat,
-					src[0], nvsNegate(src[0]), nvr_unused);
-		break;
-	case OPCODE_CMP:
-		/*XXX: this will clobber CC0... */
-		ARITH (NVS_OP_MOV, dest, mask, sat,
-				src[2], nvr_unused, nvr_unused);
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_MOV, temp, SMASK_ALL, 0,
-				src[0], nvr_unused, nvr_unused);
-		nvsinst->cond_update = 1;
-		nvsinst->cond_reg    = 0;
-		ARITH (NVS_OP_MOV, dest, mask, sat,
-				src[1], nvr_unused, nvr_unused);
-		nvsinst->cond      = COND_LT;
-		nvsinst->cond_reg  = 0;
-		nvsinst->cond_test = 1;
-		break;
-	case OPCODE_DPH:
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_DP3, temp, SMASK_X, 0,
-				 src[0], src[1], nvr_unused);
-		ARITH (NVS_OP_ADD, dest, mask, sat,
-				 nvsSwizzle(temp, X, X, X, X),
-				 nvsSwizzle(src[1], W, W, W, W),
-				 nvr_unused);
-		break;
-	case OPCODE_KIL:
-		/* This is only in ARB shaders, so we don't have to worry
-		 * about clobbering a CC reg as they aren't supported anyway.
-		 *XXX: might have to worry with GLSL however...
-		 */
-		/* MOVC0 temp, src */
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_MOV, temp, SMASK_ALL, 0,
-				 src[0], nvr_unused, nvr_unused);
-		nvsinst->cond_update = 1;
-		nvsinst->cond_reg    = 0;
-		/* KIL_NV (LT0.xyzw) temp */
-		ARITHu(NVS_OP_KIL, nvr_unused, 0, 0,
-				 nvr_unused, nvr_unused, nvr_unused);
-		nvsinst->cond      = COND_LT;
-		nvsinst->cond_reg  = 0;
-		nvsinst->cond_test = 1;
-		pass0_make_swizzle(nvsinst->cond_swizzle, SWIZZLE_NOOP);
-		break;
-	case OPCODE_LRP:
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_MAD, temp, mask, 0,
-				 nvsNegate(src[0]), src[2], src[2]);
-		ARITH (NVS_OP_MAD, dest, mask, sat, src[0], src[1], temp);
-		break;
-	case OPCODE_POW:
-		if (shader->SupportsOpcode(shader, NVS_OP_LG2) &&
-				shader->SupportsOpcode(shader, NVS_OP_EX2)) {
-			pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-			/* LG2 temp.x, src0.c */
-			ARITHu(NVS_OP_LG2, temp, SMASK_X, 0,
-					 nvsSwizzle(src[0], X, X, X, X),
-					 nvr_unused, nvr_unused);
-			/* MUL temp.x, temp.x, src1.c */
-			ARITHu(NVS_OP_MUL, temp, SMASK_X, 0,
-					 nvsSwizzle(temp, X, X, X, X),
-					 nvsSwizzle(src[1], X, X, X, X),
-					 nvr_unused);
-			/* EX2 dest, temp.x */
-			ARITH (NVS_OP_EX2, dest, mask, sat,
-					 nvsSwizzle(temp, X, X, X, X),
-					 nvr_unused, nvr_unused);
-		} else {
-			/* can we use EXP/LOG instead of EX2/LG2?? */
-			fprintf(stderr, "Implement POW for NV20 vtxprog!\n");
-			return GL_FALSE;
-		}
-		break;
-	case OPCODE_RSQ:
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_LG2, temp, SMASK_X, 0,
-				 nvsAbs(nvsSwizzle(src[0], X, X, X, X)),
-				 nvr_unused, nvr_unused);
-		nvsinst->dest_scale = NVS_SCALE_INV_2X;
-		ARITH (NVS_OP_EX2, dest, mask, sat,
-				 nvsNegate(nvsSwizzle(temp, X, X, X, X)),
-				 nvr_unused, nvr_unused);
-		break;
-	case OPCODE_SCS:
-		if (mask & SMASK_X)
-			ARITH(NVS_OP_COS, dest, SMASK_X, sat,
-					nvsSwizzle(src[0], X, X, X, X),
-					nvr_unused, nvr_unused);
-		if (mask & SMASK_Y)
-			ARITH(NVS_OP_SIN, dest, SMASK_Y, sat,
-					nvsSwizzle(src[0], X, X, X, X),
-					nvr_unused, nvr_unused);
-		break;
-	case OPCODE_SUB:
-		ARITH(NVS_OP_ADD, dest, mask, sat,
-				src[0], nvsNegate(src[1]), nvr_unused);
-		break;
-	case OPCODE_XPD:
-		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-		ARITHu(NVS_OP_MUL, temp, SMASK_ALL, 0,
-				 nvsSwizzle(src[0], Z, X, Y, Y),
-				 nvsSwizzle(src[1], Y, Z, X, X),
-				 nvr_unused);
-		ARITH (NVS_OP_MAD, dest, (mask & ~SMASK_W), sat,
-				 nvsSwizzle(src[0], Y, Z, X, X),
-				 nvsSwizzle(src[1], Z, X, Y, Y),
-				 nvsNegate(temp));
-		break;
-	default:
-		WARN_ONCE("hw doesn't support opcode \"%s\","
-			  "and no emulation found\n",
-			  _mesa_opcode_string(inst->Opcode));
-		return GL_FALSE;
-	}
-
-	return GL_TRUE;
-}
-
-static GLboolean
-pass0_translate_arith(nouveauShader *nvs, struct gl_program *prog,
-					  int ipos, int fpos,
-					  nvsFragmentHeader *parent)
-{
-	struct prog_instruction *inst = &prog->Instructions[ipos];
-	nvsFunc *shader = nvs->func;
-	nvsInstruction *nvsinst;
-	GLboolean ret;
-
-	/* Deal with multiple ATTRIB/PARAM in a single instruction */
-	pass0_check_sources(nvs, parent, fpos, inst);
-
-	/* Now it's safe to do the prog_instruction->nvsInstruction
-	 * conversion
-	 */
-	if (shader->SupportsOpcode(shader,
-				   pass0_make_opcode(inst->Opcode))) {
-		nvsRegister src[3], dest;
-		int i;
-
-		for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
-			pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
-		pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
-
-		ARITH(pass0_make_opcode(inst->Opcode), dest,
-				pass0_make_mask(inst->DstReg.WriteMask),
-				(inst->SaturateMode != SATURATE_OFF),
-				src[0], src[1], src[2]);
-		nvsinst->tex_unit   = inst->TexSrcUnit;
-		if (pass0_opcode_is_tex(inst->Opcode))
-			nvsinst->tex_target =
-				pass0_make_tex_target(inst->TexSrcTarget);
-		else
-			nvsinst->tex_target = NVS_TEX_TARGET_UNKNOWN;
-
-		ret = GL_TRUE;
-	} else
-		ret = pass0_emulate_instruction(nvs, parent, fpos, inst);
-
-	return ret;
-}
-
-static GLboolean
-pass0_translate_instructions(nouveauShader *nvs, int ipos, int fpos,
-						 nvsFragmentHeader *parent)
-{
-	struct gl_program *prog = (struct gl_program *)&nvs->mesa.vp;
-
-	while (1) {
-		struct prog_instruction *inst = &prog->Instructions[ipos];
-
-		switch (inst->Opcode) {
-		case OPCODE_END:
-			return GL_TRUE;
-		case OPCODE_BRA:
-		case OPCODE_CAL:
-		case OPCODE_RET:
-		//case OPCODE_LOOP:
-		//case OPCODE_ENDLOOP:
-		//case OPCODE_IF:
-		//case OPCODE_ELSE:
-		//case OPCODE_ENDIF:
-			WARN_ONCE("branch ops unimplemented\n");
-			return GL_FALSE;
-			break;
-		default:
-			if (!pass0_translate_arith(nvs, prog,
-						   ipos, fpos, parent))
-				return GL_FALSE;
-			break;
-		}
-
-		ipos++;
-	}
-
-	return GL_TRUE;
-}
-
-static void
-pass0_build_attrib_map(nouveauShader *nvs, struct gl_vertex_program *vp)
-{
-	GLuint inputs_read = vp->Base.InputsRead;
-	GLuint input_alloc = ~0xFFFF;
-	int i;
-
-	for (i=0; i<NVS_MAX_ATTRIBS; i++)
-		nvs->vp_attrib_map[i] = -1;
-
-	while (inputs_read) {
-		int in = ffs(inputs_read) - 1;
-		int hw;
-		inputs_read &= ~(1<<in);
-
-		if (vp->IsNVProgram) {
-			/* NVvp: must alias */
-			if (in >= VERT_ATTRIB_GENERIC0)
-				hw = in - VERT_ATTRIB_GENERIC0;
-			else
-				hw = in;
-		} else {
-			/* ARBvp: may alias (but we won't)
-			 * GL2.0: must not alias
-			 */
-			if (in >= VERT_ATTRIB_GENERIC0)
-				hw = ffs(~input_alloc) - 1;
-			else 
-				hw = in;
-			input_alloc |= (1<<hw);
-		}
-
-		nvs->vp_attrib_map[hw] = in;
-	}
-
-	if (NOUVEAU_DEBUG & DEBUG_SHADERS) {
-		printf("vtxprog attrib map:\n");
-		for (i=0; i<NVS_MAX_ATTRIBS; i++) {
-			printf(" hw:%d = attrib:%d\n",
-					i, nvs->vp_attrib_map[i]);
-		}
-	}
-}
-
-static void
-pass0_vp_insert_ff_clip_planes(GLcontext *ctx, nouveauShader *nvs)
-{
-	struct gl_program *prog = &nvs->mesa.vp.Base;
-	nvsFragmentHeader *parent = nvs->program_tree;
-	nvsInstruction *nvsinst;
-	GLuint fpos = 0;
-	nvsRegister opos, epos, eqn, mv[4];
-	gl_state_index tokens[STATE_LENGTH]
-           = { STATE_MODELVIEW_MATRIX, 0, 0, 0, 0 };
-	GLint id;
-	int i;
-
-	/* modelview transform */
-	pass0_make_reg(nvs, &opos, NVS_FILE_ATTRIB, NVS_FR_POSITION);
-	pass0_make_reg(nvs, &epos, NVS_FILE_TEMP  , -1);
-	for (i=0; i<4; i++) {
-		tokens[2] = tokens[3] = i;
-		id = _mesa_add_state_reference(prog->Parameters, tokens);
-		pass0_make_reg(nvs, &mv[i], NVS_FILE_CONST, id);
-	}
-	ARITHu(NVS_OP_DP4, epos, SMASK_X, 0, opos, mv[0], nvr_unused);
-	ARITHu(NVS_OP_DP4, epos, SMASK_Y, 0, opos, mv[1], nvr_unused);
-	ARITHu(NVS_OP_DP4, epos, SMASK_Z, 0, opos, mv[2], nvr_unused);
-	ARITHu(NVS_OP_DP4, epos, SMASK_W, 0, opos, mv[3], nvr_unused);
-
-	/* Emit code to emulate fixed-function glClipPlane */
-	for (i=0; i<6; i++) {
-		GLuint clipmask = SMASK_X;
-		nvsRegister clip;
-
-		if (!(ctx->Transform.ClipPlanesEnabled & (1<<i)))
-			continue;
-
-		/* Point a const at a user clipping plane */
-		tokens[0] = STATE_CLIPPLANE;
-		tokens[1] = i;
-		id = _mesa_add_state_reference(prog->Parameters, tokens);
-		pass0_make_reg(nvs, &eqn , NVS_FILE_CONST , id);
-		pass0_make_reg(nvs, &clip, NVS_FILE_RESULT, NVS_FR_CLIP0 + i);
-
-		/*XXX: something else needs to take care of modifying the
-		 *     instructions to write to the correct hw clip register.
-		 */
-		switch (i) {
-		case 0: case 3:	clipmask = SMASK_Y; break;
-		case 1: case 4: clipmask = SMASK_Z; break;
-		case 2: case 5: clipmask = SMASK_W; break;
-		}
-
-		/* Emit transform */
-		ARITHu(NVS_OP_DP4, clip, clipmask, 0, epos, eqn, nvr_unused);
-	}
-}
-
-static void
-pass0_rebase_mesa_consts(nouveauShader *nvs)
-{
-	struct pass0_rec *rec = nvs->pass_rec;
-	struct gl_program *prog = &nvs->mesa.vp.Base;
-	struct prog_instruction *inst = prog->Instructions;
-	int i;
-
-	/*XXX: not a good idea, params->hw_index is malloc'd */
-	memset(nvs->params, 0x00, sizeof(nvs->params));
-
-	/* When doing relative addressing on constants, the hardware needs us
-	 * to fill the "const id" field with a positive value.  Determine the
-	 * most negative index that is used so that all accesses to a
-	 * mesa-provided constant can be rebased to a positive index.
-	 */
-	while (inst->Opcode != OPCODE_END) {
-		for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++) {
-			struct prog_src_register *src = &inst->SrcReg[i];
-
-			switch (src->File) {
-			case PROGRAM_STATE_VAR:
-			case PROGRAM_CONSTANT:
-			case PROGRAM_NAMED_PARAM:
-				if (src->RelAddr && src->Index < 0) {
-					int base = src->Index * -1;
-					if (rec->mesa_const_base < base)
-						rec->mesa_const_base = base;
-				}
-				break;
-			default:
-				break;
-			}
-		}
-
-		inst++;
-	}
-}
-
-static GLboolean
-pass0_resolve_mesa_consts(nouveauShader *nvs)
-{
-	struct pass0_rec *rec = nvs->pass_rec;
-	struct gl_program *prog = &nvs->mesa.vp.Base;
-	struct gl_program_parameter_list *plist = prog->Parameters;
-	int i;
-
-	/* Init all const tracking/alloc info from the parameter list, rather
-	 * than doing it as we translate the program.  Otherwise:
-	 *   1) we can't get at the correct constant info when relative
-	 *      addressing is being used due to src->Index not pointing
-	 *      at the exact const;
-	 *   2) as we add extra consts to the program, mesa will call realloc()
-	 *      and we get invalid pointers to the const data.
-	 */
-	rec->mesa_const_last = plist->NumParameters + rec->mesa_const_base;
-	nvs->param_high = rec->mesa_const_last;
-	for (i=0; i<plist->NumParameters; i++) {
-		int hw = rec->mesa_const_base + i;
-
-		if (hw > NVS_MAX_CONSTS) {
-			nvsProgramError(nvs, "hw = %d > NVS_MAX_CONSTS!\n", hw);
-			return GL_FALSE;
-		}
-
-		switch (plist->Parameters[i].Type) {
-		case PROGRAM_NAMED_PARAM:
-		case PROGRAM_STATE_VAR:
-			nvs->params[hw].in_use     = GL_TRUE;
-			nvs->params[hw].source_val = plist->ParameterValues[i];
-			COPY_4V(nvs->params[hw].val, plist->ParameterValues[i]);
-			break;
-		case PROGRAM_CONSTANT:
-			nvs->params[hw].in_use     = GL_TRUE;
-			nvs->params[hw].source_val = NULL;
-			COPY_4V(nvs->params[hw].val, plist->ParameterValues[i]);
-			break;
-		default:
-			nvsProgramError(nvs, "hit bad type=%d on param %d\n",
-					plist->Parameters[i].Type, i);
-			return GL_FALSE;
-		}
-	}
-
-	return GL_TRUE;
-}
-
-GLboolean
-nouveau_shader_pass0(GLcontext *ctx, nouveauShader *nvs)
-{
-	nouveauContextPtr           nmesa = NOUVEAU_CONTEXT(ctx);
-	struct gl_program          *prog  = (struct gl_program*)nvs;
-	struct gl_vertex_program   *vp    = (struct gl_vertex_program *)prog;
-	struct gl_fragment_program *fp    = (struct gl_fragment_program *)prog;
-	struct pass0_rec *rec;
-	int ret = GL_FALSE;
-
-	NVSDBG("start: nvs=%p\n", nvs);
-
-	/* Previously detected an error, and haven't recieved new program
-	 * string, so fail immediately.
-	 */
-	if (nvs->error) {
-		NVSDBG("failed previous compile attempt, not retrying\n");
-		return GL_FALSE;
-	}
-
-	rec = CALLOC_STRUCT(pass0_rec);
-	if (!rec)
-		return GL_FALSE;
-
-	rec->next_temp = prog->NumTemporaries;
-	nvs->pass_rec  = rec;
-		
-	nvs->program_tree = (nvsFragmentHeader*)
-		pass0_create_subroutine(nvs, "program body");
-	if (!nvs->program_tree) {
-		FREE(rec);
-		return GL_FALSE;
-	}
-
-	switch (prog->Target) {
-	case GL_VERTEX_PROGRAM_ARB:
-		nvs->func = &nmesa->VPfunc;
-
-		if (vp->IsPositionInvariant)
-			_mesa_insert_mvp_code(ctx, vp);
-		pass0_rebase_mesa_consts(nvs);
-
-		if (!prog->String && ctx->Transform.ClipPlanesEnabled)
-			pass0_vp_insert_ff_clip_planes(ctx, nvs);
-
-		pass0_build_attrib_map(nvs, vp);
-		break;
-	case GL_FRAGMENT_PROGRAM_ARB:
-		nvs->func = &nmesa->FPfunc;
-
-		if (fp->FogOption != GL_NONE)
-			_mesa_append_fog_code(ctx, fp);
-		pass0_rebase_mesa_consts(nvs);
-		break;
-	default:
-		fprintf(stderr, "Unknown program type %d", prog->Target);
-		FREE(rec);
-		/* DESTROY TREE!! */
-		return GL_FALSE;
-	}
-	nvs->func->card_priv = &nvs->card_priv;
-
-	ret = pass0_translate_instructions(nvs, 0, 0, nvs->program_tree);
-	if (ret)
-		ret = pass0_resolve_mesa_consts(nvs);	
-	
-	/*XXX: if (!ret) DESTROY TREE!!! */
-
-	FREE(rec);
-	return ret;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c
deleted file mode 100644
index 78c1401f7db..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c
+++ /dev/null
@@ -1,16 +0,0 @@
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "nouveau_context.h"
-#include "nouveau_shader.h"
-
-GLboolean
-nouveau_shader_pass1(nvsPtr nvs)
-{
-   NVSDBG("start: nvs=%p\n", nvs);
-
-   return GL_TRUE;
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c
deleted file mode 100644
index cd27daca88b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c
+++ /dev/null
@@ -1,264 +0,0 @@
-/*
- * Copyright (C) 2006 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/*
- * Authors:
- *   Ben Skeggs <darktama@iinet.net.au>
- */
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "shader/prog_parameter.h"
-#include "shader/prog_print.h"
-
-#include "nouveau_context.h"
-#include "nouveau_shader.h"
-#include "nouveau_msg.h"
-
-struct pass2_rec {
-	/* Map nvsRegister temp ID onto hw temp ID */
-	unsigned int temps[NVS_MAX_TEMPS];
-	/* Track free hw registers */
-	unsigned int hw_temps[NVS_MAX_TEMPS];
-};
-
-static int
-pass2_alloc_hw_temp(nvsPtr nvs)
-{
-	struct pass2_rec *rec = nvs->pass_rec;
-	int i;
-
-	for (i=0; i<nvs->func->MaxTemp; i++) {
-		/* This is a *horrible* hack.. R0 is both temp0 and result.color
-		 * in NV30/40 fragprogs, we can use R0 as a temp before result
-		 * is written however..
-		 */
-		if (nvs->mesa.vp.Base.Target == GL_FRAGMENT_PROGRAM_ARB && i==0)
-			continue;
-		if (rec->hw_temps[i] == 0) {
-			rec->hw_temps[i] = 1;
-			return i;
-		}
-	}
-
-	return -1;
-}
-
-static nvsRegister
-pass2_mangle_reg(nvsPtr nvs, nvsInstruction *inst, nvsRegister reg)
-{
-	struct pass2_rec *rec = nvs->pass_rec;
-
-	if (reg.file == NVS_FILE_TEMP) {
-		if (rec->temps[reg.index] == -1)
-			rec->temps[reg.index] = pass2_alloc_hw_temp(nvs);
-		reg.index = rec->temps[reg.index];
-	}
-
-	return reg;
-}
-
-static void
-pass2_add_instruction(nvsPtr nvs, nvsInstruction *inst, 
-		      struct _op_xlat *op, int slot)
-{
-	nvsSwzComp default_swz[4] = { NVS_SWZ_X, NVS_SWZ_Y,
-				      NVS_SWZ_Z, NVS_SWZ_W };
-	nvsFunc *shader = nvs->func;
-	nvsRegister reg;
-	int i;
-
-	shader->SetOpcode(shader, op->NV, slot);
-	if (inst->saturate	) shader->SetSaturate(shader);
-	if (inst->cond_update	) shader->SetCCUpdate(shader);
-	if (inst->cond_test	) shader->SetCondition(shader, 1, inst->cond,
-						       inst->cond_reg,
-						       inst->cond_swizzle);
-	else			  shader->SetCondition(shader, 0, NVS_COND_TR,
-						       0,
-						       default_swz);
-	switch (inst->op) {
-	case NVS_OP_TEX:
-	case NVS_OP_TXB:
-	case NVS_OP_TXL:
-	case NVS_OP_TXP:
-	case NVS_OP_TXD:
-		shader->SetTexImageUnit(shader, inst->tex_unit);
-		break;
-	default:
-		break;
-	}
-
-	for (i = 0; i < 3; i++) {
-		if (op->srcpos[i] != -1) {
-			reg = pass2_mangle_reg(nvs, inst, inst->src[i]);
-
-			shader->SetSource(shader, &reg, op->srcpos[i]);
-
-			if (reg.file == NVS_FILE_CONST &&
-					shader->GetSourceConstVal) {
-				int idx_slot =
-					nvs->params[reg.index].hw_index_cnt++;
-				nvs->params[reg.index].hw_index = realloc(
-						nvs->params[reg.index].hw_index,
-						sizeof(int) * idx_slot+1);
-				nvs->params[reg.index].hw_index[idx_slot] =
-					nvs->program_current + 4;
-			}
-		}
-	}
-
-	reg = pass2_mangle_reg(nvs, inst, inst->dest);
-	shader->SetResult(shader, &reg, inst->mask, slot);
-
-	if (inst->dest_scale != NVS_SCALE_1X) {
-		shader->SetResultScale(shader, inst->dest_scale);
-	}
-}
-
-static int
-pass2_assemble_instruction(nvsPtr nvs, nvsInstruction *inst, int last)
-{
-	nvsFunc *shader = nvs->func;
-	struct _op_xlat *op;
-	unsigned int hw_inst[8];
-	int slot;
-	int instsz;
-	int i;
-
-	shader->inst = hw_inst;
-
-	/* Assemble this instruction */
-	if (!(op = shader->GetOPTXFromSOP(inst->op, &slot)))
-		return 0;
-	shader->InitInstruction(shader);
-	pass2_add_instruction(nvs, inst, op, slot);
-	if (last)
-		shader->SetLastInst(shader);
-
-	instsz = shader->GetOffsetNext(nvs->func);
-	if (nvs->program_size + instsz >= nvs->program_alloc_size) {
-		nvs->program_alloc_size *= 2;
-		nvs->program = realloc(nvs->program,
-				       nvs->program_alloc_size *
-				       sizeof(uint32_t));
-	}
-
-	for (i=0; i<instsz; i++)
-		nvs->program[nvs->program_current++] = hw_inst[i];
-	nvs->program_size = nvs->program_current;
-	return 1;
-}
-
-static GLboolean
-pass2_translate(nvsPtr nvs, nvsFragmentHeader *f)
-{
-	nvsFunc *shader = nvs->func;
-	GLboolean last;
-
-	while (f) {
-		last = (f == ((nvsSubroutine*)nvs->program_tree)->insn_tail);
-
-		switch (f->type) {
-		case NVS_INSTRUCTION:
-			if (!pass2_assemble_instruction(nvs,
-							(nvsInstruction *)f,
-							last))
-				return GL_FALSE;
-			break;
-		default:
-			WARN_ONCE("Unimplemented fragment type\n");
-			return GL_FALSE;
-		}
-
-		f = f->next;
-	}
-
-	return GL_TRUE;
-}
-
-/* Translate program into hardware format */
-GLboolean
-nouveau_shader_pass2(nvsPtr nvs)
-{
-	struct pass2_rec *rec;
-	int i;
-
-	NVSDBG("start: nvs=%p\n", nvs);
-
-	rec = calloc(1, sizeof(struct pass2_rec));
-	for (i=0; i<NVS_MAX_TEMPS; i++)
-		rec->temps[i] = -1;
-	nvs->pass_rec = rec;
-
-	/* Start off with allocating 4 uint32_t's for each inst, will be grown
-	 * if necessary..
-	 */
-	nvs->program_alloc_size = nvs->mesa.vp.Base.NumInstructions * 4;
-	nvs->program = calloc(nvs->program_alloc_size, sizeof(uint32_t));
-	nvs->program_size    = 0;
-	nvs->program_current = 0;
-
-	if (!pass2_translate(nvs,
-			     ((nvsSubroutine*)nvs->program_tree)->insn_head)) {
-		free(nvs->program);
-		nvs->program = NULL;
-		return GL_FALSE;
-	}
-
-	/* Shrink allocated memory to only what we need */
-	nvs->program = realloc(nvs->program,
-			       nvs->program_size * sizeof(uint32_t));
-	nvs->program_alloc_size = nvs->program_size;
-
-	nvs->translated  = 1;
-	nvs->on_hardware = 0;
-
-	if (NOUVEAU_DEBUG & DEBUG_SHADERS) {
-		fflush(stdout); fflush(stderr);
-		fprintf(stderr, "-----------MESA PROGRAM target=%s, id=0x%x\n",
-				_mesa_lookup_enum_by_nr(
-					nvs->mesa.vp.Base.Target),
-				nvs->mesa.vp.Base.Id);
-		fflush(stdout); fflush(stderr);
-		_mesa_print_program(&nvs->mesa.vp.Base);
-		fflush(stdout); fflush(stderr);
-		fprintf(stderr, "^^^^^^^^^^^^^^^^MESA PROGRAM\n");
-		fflush(stdout); fflush(stderr);
-		fprintf(stderr, "----------------NV PROGRAM\n");
-		fflush(stdout); fflush(stderr);
-		nvsDisasmHWShader(nvs);
-		fflush(stdout); fflush(stderr);
-		fprintf(stderr, "^^^^^^^^^^^^^^^^NV PROGRAM\n");
-		fflush(stdout); fflush(stderr);
-	}
-
-	return GL_TRUE;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.c b/src/mesa/drivers/dri/nouveau/nouveau_span.c
deleted file mode 100644
index 6e3f9fadf4e..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.c
+++ /dev/null
@@ -1,128 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#include "nouveau_context.h"
-#include "nouveau_span.h"
-#include "nouveau_fifo.h"
-#include "nouveau_lock.h"
-
-#include "swrast/swrast.h"
-
-#define HAVE_HW_DEPTH_SPANS	0
-#define HAVE_HW_DEPTH_PIXELS	0
-#define HAVE_HW_STENCIL_SPANS	0
-#define HAVE_HW_STENCIL_PIXELS	0
-
-static char *fake_span[1280*1024*4];
-
-#define HW_CLIPLOOP()							\
-   do {									\
-      int _nc = nmesa->numClipRects;					\
-      while ( _nc-- ) {							\
-	 int minx = nmesa->pClipRects[_nc].x1 - nmesa->drawX;		\
-	 int miny = nmesa->pClipRects[_nc].y1 - nmesa->drawY;		\
-	 int maxx = nmesa->pClipRects[_nc].x2 - nmesa->drawX;		\
-	 int maxy = nmesa->pClipRects[_nc].y2 - nmesa->drawY;
-
-#define LOCAL_VARS							\
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);			\
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer *)rb;		\
-   GLuint height = nrb->mesa.Height;					\
-   GLubyte *map = (GLubyte *)(nrb->map ? nrb->map : nrb->mem->map) +    \
-	 (nmesa->drawY * nrb->pitch) + (nmesa->drawX * nrb->cpp);       \
-   map = fake_span; \
-   GLuint p;								\
-   (void) p;
-
-#define Y_FLIP( _y )            (height - _y - 1)
-
-#define HW_LOCK()
-
-#define HW_UNLOCK()
-
-
-
-/* ================================================================
- * Color buffers
- */
-
-/* RGB565 */ 
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x)    nouveau##x##_RGB565
-#define TAG2(x,y) nouveau##x##_RGB565##y
-#define GET_PTR(X,Y) (map + (Y)*nrb->pitch + (X)*nrb->cpp)
-#include "spantmp2.h"
-
-
-/* ARGB8888 */
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    nouveau##x##_ARGB8888
-#define TAG2(x,y) nouveau##x##_ARGB8888##y
-#define GET_PTR(X,Y) (map + (Y)*nrb->pitch + (X)*nrb->cpp)
-#include "spantmp2.h"
-
-static void
-nouveauSpanRenderStart( GLcontext *ctx )
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   FIRE_RING();
-   LOCK_HARDWARE(nmesa);
-   nouveauWaitForIdleLocked( nmesa );
-}
-
-static void
-nouveauSpanRenderFinish( GLcontext *ctx )
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   _swrast_flush( ctx );
-   nouveauWaitForIdleLocked( nmesa );
-   UNLOCK_HARDWARE( nmesa );
-}
-
-void nouveauSpanInitFunctions( GLcontext *ctx )
-{
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart	= nouveauSpanRenderStart;
-   swdd->SpanRenderFinish	= nouveauSpanRenderFinish;
-}
-
-
-/**
- * Plug in the Get/Put routines for the given driRenderbuffer.
- */
-void
-nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis)
-{
-   if (nrb->mesa._ActualFormat == GL_RGBA8)
-      nouveauInitPointers_ARGB8888(&nrb->mesa);
-   else // if (nrb->mesa._ActualFormat == GL_RGB5)
-      nouveauInitPointers_RGB565(&nrb->mesa);
-}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.h b/src/mesa/drivers/dri/nouveau/nouveau_span.h
deleted file mode 100644
index bc39ecd17b5..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.h
+++ /dev/null
@@ -1,39 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NOUVEAU_SPAN_H__
-#define __NOUVEAU_SPAN_H__
-
-#include "drirenderbuffer.h"
-#include "nouveau_buffers.h"
-
-extern void nouveauSpanInitFunctions( GLcontext *ctx );
-extern void nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis);
-
-#endif /* __NOUVEAU_SPAN_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c
deleted file mode 100644
index f618dcfc99b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.c
+++ /dev/null
@@ -1,347 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Jeremy Kolb
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_state.h"
-#include "nouveau_swtcl.h"
-#include "nouveau_fifo.h"
-
-#include "swrast/swrast.h"
-#include "tnl/tnl.h"
-#include "swrast_setup/swrast_setup.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-static __inline__ GLuint nouveauPackColor(GLuint format,
-				       GLubyte r, GLubyte g,
-				       GLubyte b, GLubyte a)
-{
-   switch (format) {
-   case 2:
-      return PACK_COLOR_565( r, g, b );
-   case 4:
-      return PACK_COLOR_8888( r, g, b, a);
-   default:
-      fprintf(stderr, "unknown format %d\n", (int)format);
-      return 0;
-   }
-}
-
-static void nouveauCalcViewport(GLcontext *ctx)
-{
-    /* Calculate the Viewport Matrix */
-    
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    const GLfloat *v = ctx->Viewport._WindowMap.m;
-    GLfloat *m = nmesa->viewport.m;
-    GLfloat xoffset = nmesa->drawX, yoffset = nmesa->drawY + nmesa->drawH;
-  
-    nmesa->depth_scale = 1.0 / ctx->DrawBuffer->_DepthMaxF;
-
-    m[MAT_SX] =   v[MAT_SX];
-    m[MAT_TX] =   v[MAT_TX] + xoffset + SUBPIXEL_X;
-    m[MAT_SY] = - v[MAT_SY];
-    m[MAT_TY] = (-v[MAT_TY]) + yoffset + SUBPIXEL_Y;
-    m[MAT_SZ] =   v[MAT_SZ] * nmesa->depth_scale;
-    m[MAT_TZ] =   v[MAT_TZ] * nmesa->depth_scale;
-
-    nmesa->hw_func.WindowMoved(nmesa);
-}
-
-static void nouveauViewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-    /* 
-     * Need to send (at least on an nv35 the following:
-     * cons = 4 (this may be bytes per pixel)
-     *
-     * The viewport:
-     * 445   0x0000bee0   {size: 0x0   channel: 0x1   cmd: 0x00009ee0} <-- VIEWPORT_SETUP/HEADER ?
-     * 446   0x00000000   {size: 0x0   channel: 0x0   cmd: 0x00000000} <-- x * cons
-     * 447   0x00000c80   {size: 0x0   channel: 0x0   cmd: 0x00000c80} <-- (height + x) * cons
-     * 448   0x00000000   {size: 0x0   channel: 0x0   cmd: 0x00000000} <-- y * cons
-     * 449   0x00000960   {size: 0x0   channel: 0x0   cmd: 0x00000960} <-- (width + y) * cons
-     * 44a   0x00082a00   {size: 0x2   channel: 0x1   cmd: 0x00000a00} <-- VIEWPORT_DIMS
-     * 44b   0x04000000  <-- (Width_from_glViewport << 16) | x
-     * 44c   0x03000000  <-- (Height_from_glViewport << 16) | (win_height - height - y)
-     *
-     */
-    
-    nouveauCalcViewport(ctx);
-}
-
-static void nouveauDepthRange(GLcontext *ctx, GLclampd near, GLclampd far)
-{
-    nouveauCalcViewport(ctx);
-}
-
-static void nouveauDDUpdateHWState(GLcontext *ctx)
-{
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    int new_state = nmesa->new_state;
-
-    if ( new_state || nmesa->new_render_state & _NEW_TEXTURE )
-    {
-        nmesa->new_state = 0;
-
-        /* Update the various parts of the context's state.
-        */
-        /*
-        if ( new_state & NOUVEAU_NEW_ALPHA )
-            nouveauUpdateAlphaMode( ctx );
-
-        if ( new_state & NOUVEAU_NEW_DEPTH )
-            nouveauUpdateZMode( ctx );
-
-        if ( new_state & NOUVEAU_NEW_FOG )
-            nouveauUpdateFogAttrib( ctx );
-
-        if ( new_state & NOUVEAU_NEW_CLIP )
-            nouveauUpdateClipping( ctx );
-
-        if ( new_state & NOUVEAU_NEW_CULL )
-            nouveauUpdateCull( ctx );
-
-        if ( new_state & NOUVEAU_NEW_MASKS )
-            nouveauUpdateMasks( ctx );
-
-        if ( new_state & NOUVEAU_NEW_WINDOW )
-            nouveauUpdateWindow( ctx );
-
-        if ( nmesa->new_render_state & _NEW_TEXTURE ) {
-            nouveauUpdateTextureState( ctx );
-        }*/
-    }
-}
-
-static void nouveauDDInvalidateState(GLcontext *ctx, GLuint new_state)
-{
-    _swrast_InvalidateState( ctx, new_state );
-    _swsetup_InvalidateState( ctx, new_state );
-    _vbo_InvalidateState( ctx, new_state );
-    _tnl_InvalidateState( ctx, new_state );
-    NOUVEAU_CONTEXT(ctx)->new_render_state |= new_state;
-}
-
-/* Initialize the context's hardware state. */
-void nouveauDDInitState(nouveauContextPtr nmesa)
-{
-    uint32_t type = nmesa->screen->card->type;
-    switch(type)
-    {
-        case NV_03:
-            /* Unimplemented */
-            break;
-        case NV_04:
-        case NV_05:
-            nv04InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
-            break;
-        case NV_10:
-        case NV_11:
-        case NV_17:
-            nv10InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
-            break;
-        case NV_20:
-            nv20InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
-            break;
-        case NV_30:
-        case NV_40:
-        case NV_44:
-            nv30InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
-            break;
-        case NV_50:
-            nv50InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
-            break;
-        default:
-            break;
-    }
-    nouveau_state_cache_init(nmesa);
-}
-
-/* Initialize the driver's state functions */
-void nouveauDDInitStateFuncs(GLcontext *ctx)
-{
-   ctx->Driver.UpdateState		= nouveauDDInvalidateState;
-
-   ctx->Driver.ClearIndex		= NULL;
-   ctx->Driver.ClearColor		= NULL; //nouveauDDClearColor;
-   ctx->Driver.ClearStencil		= NULL; //nouveauDDClearStencil;
-   ctx->Driver.DrawBuffer		= NULL; //nouveauDDDrawBuffer;
-   ctx->Driver.ReadBuffer		= NULL; //nouveauDDReadBuffer;
-
-   ctx->Driver.IndexMask		= NULL;
-   ctx->Driver.ColorMask		= NULL; //nouveauDDColorMask;
-   ctx->Driver.AlphaFunc		= NULL; //nouveauDDAlphaFunc;
-   ctx->Driver.BlendEquationSeparate	= NULL; //nouveauDDBlendEquationSeparate;
-   ctx->Driver.BlendFuncSeparate	= NULL; //nouveauDDBlendFuncSeparate;
-   ctx->Driver.ClearDepth		= NULL; //nouveauDDClearDepth;
-   ctx->Driver.CullFace			= NULL; //nouveauDDCullFace;
-   ctx->Driver.FrontFace		= NULL; //nouveauDDFrontFace;
-   ctx->Driver.DepthFunc		= NULL; //nouveauDDDepthFunc;
-   ctx->Driver.DepthMask		= NULL; //nouveauDDDepthMask;
-   ctx->Driver.Enable			= NULL; //nouveauDDEnable;
-   ctx->Driver.Fogfv			= NULL; //nouveauDDFogfv;
-   ctx->Driver.Hint			= NULL;
-   ctx->Driver.Lightfv			= NULL;
-   ctx->Driver.LightModelfv		= NULL; //nouveauDDLightModelfv;
-   ctx->Driver.LogicOpcode		= NULL; //nouveauDDLogicOpCode;
-   ctx->Driver.PolygonMode		= NULL;
-   ctx->Driver.PolygonStipple		= NULL; //nouveauDDPolygonStipple;
-   ctx->Driver.RenderMode		= NULL; //nouveauDDRenderMode;
-   ctx->Driver.Scissor			= NULL; //nouveauDDScissor;
-   ctx->Driver.ShadeModel		= NULL; //nouveauDDShadeModel;
-   ctx->Driver.StencilFuncSeparate	= NULL; //nouveauDDStencilFuncSeparate;
-   ctx->Driver.StencilMaskSeparate	= NULL; //nouveauDDStencilMaskSeparate;
-   ctx->Driver.StencilOpSeparate	= NULL; //nouveauDDStencilOpSeparate;
-
-   ctx->Driver.DepthRange               = nouveauDepthRange;
-   ctx->Driver.Viewport                 = nouveauViewport;
-
-   /* Pixel path fallbacks.
-    */
-   ctx->Driver.Accum = _swrast_Accum;
-   ctx->Driver.Bitmap = _swrast_Bitmap;
-   ctx->Driver.CopyPixels = _swrast_CopyPixels;
-   ctx->Driver.DrawPixels = _swrast_DrawPixels;
-   ctx->Driver.ReadPixels = _swrast_ReadPixels;
-
-   /* Swrast hooks for imaging extensions:
-    */
-   ctx->Driver.CopyColorTable = _swrast_CopyColorTable;
-   ctx->Driver.CopyColorSubTable = _swrast_CopyColorSubTable;
-   ctx->Driver.CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
-   ctx->Driver.CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
-}
-
-#define STATE_INIT(a) if (ctx->Driver.a) ctx->Driver.a
-
-void nouveauInitState(GLcontext *ctx)
-{
-    /*
-     * Mesa should do this for us:
-     */
-
-    STATE_INIT(AlphaFunc)( ctx, 
-            ctx->Color.AlphaFunc,
-            ctx->Color.AlphaRef);
-
-    STATE_INIT(BlendColor)( ctx,
-            ctx->Color.BlendColor );
-
-    STATE_INIT(BlendEquationSeparate)( ctx, 
-            ctx->Color.BlendEquationRGB,
-            ctx->Color.BlendEquationA);
-
-    STATE_INIT(BlendFuncSeparate)( ctx,
-            ctx->Color.BlendSrcRGB,
-            ctx->Color.BlendDstRGB,
-            ctx->Color.BlendSrcA,
-            ctx->Color.BlendDstA);
-
-    STATE_INIT(ClearColor)( ctx, ctx->Color.ClearColor);
-    STATE_INIT(ClearDepth)( ctx, ctx->Depth.Clear);
-    STATE_INIT(ClearStencil)( ctx, ctx->Stencil.Clear);
-
-    STATE_INIT(ColorMask)( ctx, 
-            ctx->Color.ColorMask[RCOMP],
-            ctx->Color.ColorMask[GCOMP],
-            ctx->Color.ColorMask[BCOMP],
-            ctx->Color.ColorMask[ACOMP]);
-
-    STATE_INIT(CullFace)( ctx, ctx->Polygon.CullFaceMode );
-    STATE_INIT(DepthFunc)( ctx, ctx->Depth.Func );
-    STATE_INIT(DepthMask)( ctx, ctx->Depth.Mask );
-
-    STATE_INIT(Enable)( ctx, GL_ALPHA_TEST, ctx->Color.AlphaEnabled );
-    STATE_INIT(Enable)( ctx, GL_BLEND, ctx->Color.BlendEnabled );
-    STATE_INIT(Enable)( ctx, GL_COLOR_LOGIC_OP, ctx->Color.ColorLogicOpEnabled );
-    STATE_INIT(Enable)( ctx, GL_COLOR_SUM, ctx->Fog.ColorSumEnabled );
-    STATE_INIT(Enable)( ctx, GL_CULL_FACE, ctx->Polygon.CullFlag );
-    STATE_INIT(Enable)( ctx, GL_DEPTH_TEST, ctx->Depth.Test );
-    STATE_INIT(Enable)( ctx, GL_DITHER, ctx->Color.DitherFlag );
-    STATE_INIT(Enable)( ctx, GL_FOG, ctx->Fog.Enabled );
-    STATE_INIT(Enable)( ctx, GL_LIGHTING, ctx->Light.Enabled );
-    STATE_INIT(Enable)( ctx, GL_LINE_SMOOTH, ctx->Line.SmoothFlag );
-    STATE_INIT(Enable)( ctx, GL_LINE_STIPPLE, ctx->Line.StippleFlag );
-    STATE_INIT(Enable)( ctx, GL_POINT_SMOOTH, ctx->Point.SmoothFlag );
-    STATE_INIT(Enable)( ctx, GL_POLYGON_OFFSET_FILL, ctx->Polygon.OffsetFill);
-    STATE_INIT(Enable)( ctx, GL_POLYGON_OFFSET_LINE, ctx->Polygon.OffsetLine);
-    STATE_INIT(Enable)( ctx, GL_POLYGON_OFFSET_POINT, ctx->Polygon.OffsetPoint);
-    STATE_INIT(Enable)( ctx, GL_POLYGON_SMOOTH, ctx->Polygon.SmoothFlag );
-    STATE_INIT(Enable)( ctx, GL_POLYGON_STIPPLE, ctx->Polygon.StippleFlag );
-    STATE_INIT(Enable)( ctx, GL_SCISSOR_TEST, ctx->Scissor.Enabled );
-    STATE_INIT(Enable)( ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled );
-    STATE_INIT(Enable)( ctx, GL_TEXTURE_1D, GL_FALSE );
-    STATE_INIT(Enable)( ctx, GL_TEXTURE_2D, GL_FALSE );
-    STATE_INIT(Enable)( ctx, GL_TEXTURE_RECTANGLE_NV, GL_FALSE );
-    STATE_INIT(Enable)( ctx, GL_TEXTURE_3D, GL_FALSE );
-    STATE_INIT(Enable)( ctx, GL_TEXTURE_CUBE_MAP, GL_FALSE );
-
-    STATE_INIT(Fogfv)( ctx, GL_FOG_COLOR, ctx->Fog.Color );
-    STATE_INIT(Fogfv)( ctx, GL_FOG_MODE, 0 );
-    STATE_INIT(Fogfv)( ctx, GL_FOG_DENSITY, &ctx->Fog.Density );
-    STATE_INIT(Fogfv)( ctx, GL_FOG_START, &ctx->Fog.Start );
-    STATE_INIT(Fogfv)( ctx, GL_FOG_END, &ctx->Fog.End );
-
-    STATE_INIT(FrontFace)( ctx, ctx->Polygon.FrontFace );
-
-    {
-        GLfloat f = (GLfloat)ctx->Light.Model.ColorControl;
-        STATE_INIT(LightModelfv)( ctx, GL_LIGHT_MODEL_COLOR_CONTROL, &f );
-    }
-
-    STATE_INIT(LineStipple)( ctx, ctx->Line.StippleFactor, ctx->Line.StipplePattern );
-    STATE_INIT(LineWidth)( ctx, ctx->Line.Width );
-    STATE_INIT(LogicOpcode)( ctx, ctx->Color.LogicOp );
-    STATE_INIT(PointSize)( ctx, ctx->Point.Size );
-    STATE_INIT(PolygonMode)( ctx, GL_FRONT, ctx->Polygon.FrontMode );
-    STATE_INIT(PolygonMode)( ctx, GL_BACK, ctx->Polygon.BackMode );
-    STATE_INIT(PolygonOffset)( ctx,
-	    ctx->Polygon.OffsetFactor,
-	    ctx->Polygon.OffsetUnits );
-    STATE_INIT(PolygonStipple)( ctx, (const GLubyte *)ctx->PolygonStipple );
-    STATE_INIT(ShadeModel)( ctx, ctx->Light.ShadeModel );
-    STATE_INIT(StencilFuncSeparate)( ctx, GL_FRONT,
-            ctx->Stencil.Function[0],
-            ctx->Stencil.Ref[0],
-            ctx->Stencil.ValueMask[0] );
-    STATE_INIT(StencilFuncSeparate)( ctx, GL_BACK,
-            ctx->Stencil.Function[1],
-            ctx->Stencil.Ref[1],
-            ctx->Stencil.ValueMask[1] );
-    STATE_INIT(StencilMaskSeparate)( ctx, GL_FRONT, ctx->Stencil.WriteMask[0] );
-    STATE_INIT(StencilMaskSeparate)( ctx, GL_BACK, ctx->Stencil.WriteMask[1] );
-    STATE_INIT(StencilOpSeparate)( ctx, GL_FRONT,
-            ctx->Stencil.FailFunc[0],
-            ctx->Stencil.ZFailFunc[0],
-            ctx->Stencil.ZPassFunc[0]);
-    STATE_INIT(StencilOpSeparate)( ctx, GL_BACK,
-            ctx->Stencil.FailFunc[1],
-            ctx->Stencil.ZFailFunc[1],
-            ctx->Stencil.ZPassFunc[1]);
-}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.h b/src/mesa/drivers/dri/nouveau/nouveau_state.h
deleted file mode 100644
index dbac71760b6..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.h
+++ /dev/null
@@ -1,50 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Jeremy Kolb
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#ifndef __NOUVEAU_STATE_H__
-#define __NOUVEAU_STATE_H__
-
-#include "nouveau_context.h"
-
-extern void nouveauDDInitState(nouveauContextPtr nmesa);
-extern void nouveauDDInitStateFuncs(GLcontext *ctx);
-
-extern void nv04InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
-extern void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
-extern void nv20InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
-extern void nv30InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
-extern void nv50InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
-
-extern void nouveauInitState(GLcontext *ctx);
-
-/*
-extern void nouveauDDUpdateState(GLcontext *ctx);
-extern void nouveauDDUpdateHWState(GLcontext *ctx);
-
-extern void nouveauEmitHwStateLocked(nouveauContextPtr nmesa);
-*/
-#endif
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c
deleted file mode 100644
index cb4b9d30270..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c
+++ /dev/null
@@ -1,69 +0,0 @@
-
-#include "nouveau_state_cache.h"
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-
-#define BEGIN_RING_NOFLUSH(subchannel,tag,size) do {					\
-	if (nmesa->fifo.free <= (size))							\
-		WAIT_RING(nmesa,(size));						\
-	OUT_RING( ((size)<<18) | ((subchannel) << 13) | (tag));				\
-	nmesa->fifo.free -= ((size) + 1);                                               \
-}while(0)
-
-// flush all the dirty state
-void nouveau_state_cache_flush(nouveauContextPtr nmesa)
-{
-	int i=0;
-	int run=0;
-
-	// fast-path no state changes
-	if (!nmesa->state_cache.dirty)
-		return;
-	nmesa->state_cache.dirty=0;
-
-	do
-	{
-		// jump to a dirty state
-		while((nmesa->state_cache.hdirty[i/NOUVEAU_STATE_CACHE_HIER_SIZE]==0)&&(i<NOUVEAU_STATE_CACHE_ENTRIES))
-			i=(i&~(NOUVEAU_STATE_CACHE_HIER_SIZE-1))+NOUVEAU_STATE_CACHE_HIER_SIZE;
-		while((nmesa->state_cache.atoms[i].dirty==0)&&(i<NOUVEAU_STATE_CACHE_ENTRIES))
-			i++;
-
-		// figure out a run of dirty values
-		run=0;
-		while((nmesa->state_cache.atoms[i+run].dirty)&&(i+run<NOUVEAU_STATE_CACHE_ENTRIES))
-			run++;
-
-		// output everything as a single run
-		if (run>0) {
-			int j;
-
-			BEGIN_RING_NOFLUSH(NvSub3D, i*4, run);
-			for(j=0;j<run;j++)
-			{
-				OUT_RING(nmesa->state_cache.atoms[i+j].value);
-				nmesa->state_cache.atoms[i+j].dirty=0;
-				if ((i+j)%NOUVEAU_STATE_CACHE_HIER_SIZE==0)
-					nmesa->state_cache.hdirty[(i+j)/NOUVEAU_STATE_CACHE_HIER_SIZE-1]=0;
-			}
-			i+=run;
-		}
-	}
-	while(i<NOUVEAU_STATE_CACHE_ENTRIES);
-	nmesa->state_cache.hdirty[NOUVEAU_STATE_CACHE_HIER_SIZE/NOUVEAU_STATE_CACHE_HIER_SIZE-1]=0;
-}
-
-
-// inits the state cache
-void nouveau_state_cache_init(nouveauContextPtr nmesa)
-{
-	int i;
-	for(i=0;i<NOUVEAU_STATE_CACHE_ENTRIES;i++)
-	{
-		nmesa->state_cache.atoms[i].dirty=0;
-		nmesa->state_cache.atoms[i].value=0xDEADBEEF; // nvidia cards like beef
-	}
-	nmesa->state_cache.dirty=0;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h
deleted file mode 100644
index 5f9d426450b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h
+++ /dev/null
@@ -1,29 +0,0 @@
-
-#ifndef __NOUVEAU_STATE_CACHE_H__
-#define __NOUVEAU_STATE_CACHE_H__
-
-#include "mtypes.h"
-
-#define NOUVEAU_STATE_CACHE_ENTRIES 2048
-// size of a dirty requests block
-// you can play with that and tune the value to increase/decrease performance
-// but keep it a power of 2 !
-#define NOUVEAU_STATE_CACHE_HIER_SIZE  32
-
-typedef struct nouveau_state_atom_t{
-	uint32_t value;
-	uint32_t dirty;
-}nouveau_state_atom;
-
-typedef struct nouveau_state_cache_t{
-	nouveau_state_atom atoms[NOUVEAU_STATE_CACHE_ENTRIES];
-	uint32_t current_pos;
-	// hierarchical dirty flags
-	uint8_t hdirty[NOUVEAU_STATE_CACHE_ENTRIES/NOUVEAU_STATE_CACHE_HIER_SIZE];
-	// master dirty flag
-	uint8_t dirty;
-}nouveau_state_cache;
-
-
-#endif
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c b/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c
deleted file mode 100644
index 8a013bd9995..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c
+++ /dev/null
@@ -1,127 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/* Common software TCL code */
-
-#include "nouveau_context.h"
-#include "nouveau_swtcl.h"
-#include "nv10_swtcl.h"
-#include "nouveau_span.h"
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-#include "tnl/t_pipeline.h"
-
-/* Common tri functions */
-
-/* The fallbacks */
-void nouveau_fallback_tri(struct nouveau_context *nmesa,
-		nouveauVertex *v0,
-		nouveauVertex *v1,
-		nouveauVertex *v2)
-{    
-	GLcontext *ctx = nmesa->glCtx;
-	SWvertex v[3];
-	_swsetup_Translate(ctx, v0, &v[0]);
-	_swsetup_Translate(ctx, v1, &v[1]);
-	_swsetup_Translate(ctx, v2, &v[2]);
-	_swrast_Triangle(ctx, &v[0], &v[1], &v[2]);
-}
-
-
-void nouveau_fallback_line(struct nouveau_context *nmesa,
-		nouveauVertex *v0,
-		nouveauVertex *v1)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	SWvertex v[2];
-	_swsetup_Translate(ctx, v0, &v[0]);
-	_swsetup_Translate(ctx, v1, &v[1]);
-	_swrast_Line(ctx, &v[0], &v[1]);
-}
-
-
-void nouveau_fallback_point(struct nouveau_context *nmesa,
-		nouveauVertex *v0)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	SWvertex v[1];
-	_swsetup_Translate(ctx, v0, &v[0]);
-	_swrast_Point(ctx, &v[0]);
-}
-
-void nouveauFallback(struct nouveau_context *nmesa, GLuint bit, GLboolean mode)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLuint oldfallback = nmesa->Fallback;
-
-	if (mode) {
-		nmesa->Fallback |= bit;
-		if (oldfallback == 0) {
-			if (nmesa->screen->card->type<NV_10) {
-				//nv04FinishPrimitive(nmesa);
-			} else {
-				//nv10FinishPrimitive(nmesa);
-			}
-
-			_swsetup_Wakeup(ctx);
-			nmesa->render_index = ~0;
-		}
-	}
-	else {
-		nmesa->Fallback &= ~bit;
-		if (oldfallback == bit) {
-			_swrast_flush( ctx );
-
-			if (nmesa->screen->card->type<NV_10) {
-				nv04TriInitFunctions(ctx);
-			} else {
-				nv10TriInitFunctions(ctx);
-			}
-
-			_tnl_invalidate_vertex_state( ctx, ~0 );
-			_tnl_invalidate_vertices( ctx, ~0 );
-			_tnl_install_attrs( ctx, 
-					nmesa->vertex_attrs, 
-					nmesa->vertex_attr_count,
-					nmesa->viewport.m, 0 ); 
-		}
-	}    
-}
-
-
-void nouveauRunPipeline( GLcontext *ctx )
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (nmesa->new_state) {
-		nmesa->new_render_state |= nmesa->new_state;
-	}
-
-	_tnl_run_pipeline( ctx );
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.h b/src/mesa/drivers/dri/nouveau/nouveau_swtcl.h
deleted file mode 100644
index ba4d8725a65..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.h
+++ /dev/null
@@ -1,55 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NOUVEAU_SWTCL_H__
-#define __NOUVEAU_SWTCL_H__
-
-#include "nouveau_context.h"
-
-extern void nouveau_fallback_tri(struct nouveau_context *nmesa,
-		nouveauVertex *v0,
-		nouveauVertex *v1,
-		nouveauVertex *v2);
-
-extern void nouveau_fallback_line(struct nouveau_context *nmesa,
-		nouveauVertex *v0,
-		nouveauVertex *v1);
-
-extern void nouveau_fallback_point(struct nouveau_context *nmesa, 
-		nouveauVertex *v0);
-
-extern void nouveauFallback(struct nouveau_context *nmesa, GLuint bit, GLboolean mode);
-
-extern void nouveauRunPipeline( GLcontext *ctx );
-
-extern void nouveauTriInitFunctions( GLcontext *ctx );
-
-
-#endif /* __NOUVEAU_SWTCL_H__ */
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.c b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
deleted file mode 100644
index 8abc847e1e2..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.c
+++ /dev/null
@@ -1,200 +0,0 @@
-/*
- * Copyright (C) 2007 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#include "vblank.h" /* for DO_USLEEP */
-
-#include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_msg.h"
-#include "nouveau_sync.h"
-
-#define NOTIFIER(__v) \
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); \
-	volatile uint32_t *__v = (void*)nmesa->notifier_block + notifier->offset
-
-struct drm_nouveau_notifier_alloc *
-nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	struct drm_nouveau_notifier_alloc *notifier;
-	int ret;
-
-#ifdef NOUVEAU_RING_DEBUG
-	return NULL;
-#endif
-
-	notifier = CALLOC_STRUCT(drm_nouveau_notifier_alloc);
-	if (!notifier)
-		return NULL;
-
-	notifier->channel = nmesa->fifo.channel;
-	notifier->handle  = handle;
-	notifier->count   = count;
-	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_NOTIFIER_ALLOC,
-				  notifier, sizeof(*notifier));
-	if (ret) {
-		MESSAGE("Failed to create notifier 0x%08x: %d\n", handle, ret);
-		FREE(notifier);
-		return NULL;
-	}
-
-	return notifier;
-}
-
-void
-nouveau_notifier_destroy(GLcontext *ctx,
-			 struct drm_nouveau_notifier_alloc *notifier)
-{
-	/*XXX: free notifier object.. */
-	FREE(notifier);
-}
-
-void
-nouveau_notifier_reset(GLcontext *ctx,
-		       struct drm_nouveau_notifier_alloc *notifier,
-		       GLuint id)
-{
-	NOTIFIER(n);
-
-#ifdef NOUVEAU_RING_DEBUG
-	return;
-#endif
-
-	n[NV_NOTIFY_TIME_0      /4] = 0x00000000;
-	n[NV_NOTIFY_TIME_1      /4] = 0x00000000;
-	n[NV_NOTIFY_RETURN_VALUE/4] = 0x00000000;
-	n[NV_NOTIFY_STATE       /4] = (NV_NOTIFY_STATE_STATUS_IN_PROCESS <<
-				       NV_NOTIFY_STATE_STATUS_SHIFT);
-}
-
-GLuint
-nouveau_notifier_status(GLcontext *ctx,
-			struct drm_nouveau_notifier_alloc *notifier,
-			GLuint id)
-{
-	NOTIFIER(n);
-
-	return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT;
-}
-
-GLuint
-nouveau_notifier_return_val(GLcontext *ctx,
-			    struct drm_nouveau_notifier_alloc *notifier,
-			    GLuint id)
-{
-	NOTIFIER(n);
-
-	return n[NV_NOTIFY_RETURN_VALUE/4];
-}
-
-GLboolean
-nouveau_notifier_wait_status(GLcontext *ctx,
-			     struct drm_nouveau_notifier_alloc *notifier,
-			     GLuint id, GLuint status, GLuint timeout)
-{
-	NOTIFIER(n);
-	unsigned int time = 0;
-
-#ifdef NOUVEAU_RING_DEBUG
-	return GL_TRUE;
-#endif
-
-	while (time <= timeout) {
-		if (n[NV_NOTIFY_STATE/4] & NV_NOTIFY_STATE_ERROR_CODE_MASK) {
-			MESSAGE("Notifier returned error: 0x%04x\n",
-					n[NV_NOTIFY_STATE/4] &
-					NV_NOTIFY_STATE_ERROR_CODE_MASK);
-			return GL_FALSE;
-		}
-
-		if (((n[NV_NOTIFY_STATE/4] & NV_NOTIFY_STATE_STATUS_MASK) >>
-				NV_NOTIFY_STATE_STATUS_SHIFT) == status)
-			return GL_TRUE;
-
-		if (timeout) {
-			DO_USLEEP(1);
-			time++;
-		}
-	}
-
-	MESSAGE("Notifier timed out\n");
-	return GL_FALSE;
-}
-
-void
-nouveau_notifier_wait_nop(GLcontext *ctx,
-			  struct drm_nouveau_notifier_alloc *notifier,
-			  GLuint subc)
-{
-	NOTIFIER(n);
-	GLboolean ret;
-
-	nouveau_notifier_reset(ctx, notifier, 0);
-
-	BEGIN_RING_SIZE(subc, NV_NOTIFY, 1);
-	OUT_RING       (NV_NOTIFY_STYLE_WRITE_ONLY);
-	BEGIN_RING_SIZE(subc, NV_NOP, 1);
-	OUT_RING       (0);
-	FIRE_RING();
-
-	ret = nouveau_notifier_wait_status(ctx, notifier, 0,
-					   NV_NOTIFY_STATE_STATUS_COMPLETED,
-					   0 /* no timeout */);
-	if (ret == GL_FALSE) MESSAGE("wait on notifier failed\n");
-}
-
-GLboolean nouveauSyncInitFuncs(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-#ifdef NOUVEAU_RING_DEBUG
-	return GL_TRUE;
-#endif
-
-	nmesa->syncNotifier = nouveau_notifier_new(ctx, NvSyncNotify, 1);
-	if (!nmesa->syncNotifier) {
-		MESSAGE("Failed to create channel sync notifier\n");
-		return GL_FALSE;
-	}
-
-	/* 0x180 is SET_DMA_NOTIFY, should be correct for all supported 3D
-	 * object classes
-	 */
-	BEGIN_RING_CACHE(NvSub3D, 0x180, 1);
-	OUT_RING_CACHE  (NvSyncNotify);
-#ifdef ALLOW_MULTI_SUBCHANNEL
-	BEGIN_RING_SIZE(NvSubMemFormat,
-	      		NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1);
-	OUT_RING       (NvSyncNotify);
-#endif
-
-	return GL_TRUE;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.h b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
deleted file mode 100644
index b76af172762..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.h
+++ /dev/null
@@ -1,71 +0,0 @@
-/*
- * Copyright (C) 2007 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-#ifndef __NOUVEAU_SYNC_H__
-#define __NOUVEAU_SYNC_H__
-
-#include "nouveau_buffers.h"
-
-#define NV_NOTIFIER_SIZE                                                      32
-#define NV_NOTIFY_TIME_0                                              0x00000000
-#define NV_NOTIFY_TIME_1                                              0x00000004
-#define NV_NOTIFY_RETURN_VALUE                                        0x00000008
-#define NV_NOTIFY_STATE                                               0x0000000C
-#define NV_NOTIFY_STATE_STATUS_MASK                                   0xFF000000
-#define NV_NOTIFY_STATE_STATUS_SHIFT                                          24
-#define NV_NOTIFY_STATE_STATUS_COMPLETED                                    0x00
-#define NV_NOTIFY_STATE_STATUS_IN_PROCESS                                   0x01
-#define NV_NOTIFY_STATE_ERROR_CODE_MASK                               0x0000FFFF
-#define NV_NOTIFY_STATE_ERROR_CODE_SHIFT                                       0
-
-/* Methods that (hopefully) all objects have */
-#define NV_NOP                                                        0x00000100
-#define NV_NOTIFY                                                     0x00000104
-#define NV_NOTIFY_STYLE_WRITE_ONLY                                             0
-
-extern struct drm_nouveau_notifier_alloc *
-nouveau_notifier_new(GLcontext *, GLuint handle, GLuint count);
-extern void 
-nouveau_notifier_destroy(GLcontext *, struct drm_nouveau_notifier_alloc *);
-extern void
-nouveau_notifier_reset(GLcontext *, struct drm_nouveau_notifier_alloc *,
-		       GLuint id);
-extern GLuint
-nouveau_notifier_status(GLcontext *, struct drm_nouveau_notifier_alloc *,
-			GLuint id);
-extern GLuint
-nouveau_notifier_return_val(GLcontext *, struct drm_nouveau_notifier_alloc *,
-			    GLuint id);
-extern GLboolean
-nouveau_notifier_wait_status(GLcontext *, struct drm_nouveau_notifier_alloc *,
-			     GLuint id, GLuint status, GLuint timeout);
-extern void
-nouveau_notifier_wait_nop(GLcontext *ctx, struct drm_nouveau_notifier_alloc *,
-			  GLuint subc);
-
-extern GLboolean nouveauSyncInitFuncs(GLcontext *ctx);
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_tex.c b/src/mesa/drivers/dri/nouveau/nouveau_tex.c
deleted file mode 100644
index 0a8d2796695..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_tex.c
+++ /dev/null
@@ -1,49 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#include "nouveau_tex.h"
-
-// XXX needs some love
-void nouveauTexInitFunctions( struct dd_function_table *functions )
-{
-/*
-   functions->TexEnv			= nouveauTexEnv;
-   functions->ChooseTextureFormat	= nouveauChooseTextureFormat;
-   functions->TexImage1D		= nouveauTexImage1D;
-   functions->TexSubImage1D		= nouveauTexSubImage1D;
-   functions->TexImage2D		= nouveauTexImage2D;
-   functions->TexSubImage2D		= nouveauTexSubImage2D;
-   functions->TexParameter		= nouveauTexParameter;
-   functions->BindTexture		= nouveauBindTexture;
-   functions->NewTextureObject		= nouveauNewTextureObject;
-   functions->DeleteTexture		= nouveauDeleteTexture;
-   functions->IsTextureResident		= driIsTextureResident;
-
-   driInitTextureFormats();
-*/
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_tex.h b/src/mesa/drivers/dri/nouveau/nouveau_tex.h
deleted file mode 100644
index 7ac71f8300b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_tex.h
+++ /dev/null
@@ -1,38 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-#ifndef __NOUVEAU_TEX_H__
-#define __NOUVEAU_TEX_H__
-
-#include <errno.h>
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
-
-extern void nouveauTexInitFunctions( struct dd_function_table *functions );
-
-#endif /* __NOUVEAU_TEX_H__ */
diff --git a/src/mesa/drivers/dri/nouveau/nv04_state.c b/src/mesa/drivers/dri/nouveau/nv04_state.c
deleted file mode 100644
index 25df3d2a624..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv04_state.c
+++ /dev/null
@@ -1,540 +0,0 @@
-/**************************************************************************
-
-Copyright 2007 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_msg.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-static uint32_t nv04_compare_func(GLuint f)
-{
-	switch ( f ) {
-		case GL_NEVER:		return 1;
-		case GL_LESS:		return 2;
-		case GL_EQUAL:		return 3;
-		case GL_LEQUAL:		return 4;
-		case GL_GREATER:	return 5;
-		case GL_NOTEQUAL:	return 6;
-		case GL_GEQUAL:		return 7;
-		case GL_ALWAYS:		return 8;
-	}
-	WARN_ONCE("Unable to find the function\n");
-	return 0;
-}
-
-static uint32_t nv04_blend_func(GLuint f)
-{
-	switch ( f ) {
-		case GL_ZERO:			return 0x1;
-		case GL_ONE:			return 0x2;
-		case GL_SRC_COLOR:		return 0x3;
-		case GL_ONE_MINUS_SRC_COLOR:	return 0x4;
-		case GL_SRC_ALPHA:		return 0x5;
-		case GL_ONE_MINUS_SRC_ALPHA:	return 0x6;
-		case GL_DST_ALPHA:		return 0x7;
-		case GL_ONE_MINUS_DST_ALPHA:	return 0x8;
-		case GL_DST_COLOR:		return 0x9;
-		case GL_ONE_MINUS_DST_COLOR:	return 0xA;
-		case GL_SRC_ALPHA_SATURATE:	return 0xB;
-	}
-	WARN_ONCE("Unable to find the function 0x%x\n",f);
-	return 0;
-}
-
-static void nv04_emit_control(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	uint32_t control,cull;
-	GLubyte alpha_ref;
-
-	CLAMPED_FLOAT_TO_UBYTE(alpha_ref, ctx->Color.AlphaRef);
-	control=alpha_ref;
-	control|=(nv04_compare_func(ctx->Color.AlphaFunc)<<8);
-	control|=(ctx->Color.AlphaEnabled<<12);
-	control|=(1<<13);
-	control|=(ctx->Depth.Test<<14);
-	control|=(nv04_compare_func(ctx->Depth.Func)<<16);
-	if ((ctx->Polygon.CullFlag)&&(ctx->Polygon.CullFaceMode!=GL_FRONT_AND_BACK))
-	{
-		if ((ctx->Polygon.FrontFace==GL_CW)&&(ctx->Polygon.CullFaceMode==GL_FRONT))
-			cull=2;
-		if ((ctx->Polygon.FrontFace==GL_CW)&&(ctx->Polygon.CullFaceMode==GL_BACK))
-			cull=3;
-		if ((ctx->Polygon.FrontFace==GL_CCW)&&(ctx->Polygon.CullFaceMode==GL_FRONT))
-			cull=3;
-		if ((ctx->Polygon.FrontFace==GL_CCW)&&(ctx->Polygon.CullFaceMode==GL_BACK))
-			cull=2;
-	}
-	else
-		if (ctx->Polygon.CullFaceMode==GL_FRONT_AND_BACK)
-			cull=0;
-		else
-			cull=1;
-	control|=(cull<<20);
-	control|=(ctx->Color.DitherFlag<<22);
-	if ((ctx->Depth.Test)&&(ctx->Depth.Mask))
-		control|=(1<<24);
-
-	control|=(1<<30); // integer zbuffer format
-
-	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1);
-	OUT_RING_CACHE(control);
-}
-
-static void nv04_emit_blend(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	uint32_t blend;
-
-	blend=0x4; // texture MODULATE_ALPHA
-	blend|=0x20; // alpha is MSB
-	switch(ctx->Light.ShadeModel) {
-		case GL_SMOOTH:blend|=(1<<6);break;
-		case GL_FLAT:  blend|=(2<<6);break;
-		default:break;
-	}
-	if (ctx->Hint.PerspectiveCorrection!=GL_FASTEST)
-		blend|=(1<<8);
-	blend|=(ctx->Fog.Enabled<<16);
-	blend|=(ctx->Color.BlendEnabled<<20);
-	blend|=(nv04_blend_func(ctx->Color.BlendSrcRGB)<<24);
-	blend|=(nv04_blend_func(ctx->Color.BlendDstRGB)<<28);
-
-	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1);
-	OUT_RING_CACHE(blend);
-}
-
-static void nv04_emit_fog_color(GLcontext *ctx)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte c[4];
-	c[0] = FLOAT_TO_UBYTE( ctx->Fog.Color[0] );
-	c[1] = FLOAT_TO_UBYTE( ctx->Fog.Color[1] );
-	c[2] = FLOAT_TO_UBYTE( ctx->Fog.Color[2] );
-	c[3] = FLOAT_TO_UBYTE( ctx->Fog.Color[3] );
-	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_FOG_COLOR, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888_REV(c[0],c[1],c[2],c[3]));
-}
-
-static void nv04AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
-{
-	nv04_emit_control(ctx);
-}
-
-static void nv04BlendColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nv04_emit_blend(ctx);
-}
-
-static void nv04BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
-{
-	nv04_emit_blend(ctx);
-}
-
-
-static void nv04BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
-		GLenum sfactorA, GLenum dfactorA)
-{
-	nv04_emit_blend(ctx);
-}
-
-static void nv04Clear(GLcontext *ctx, GLbitfield mask)
-{
-	/* TODO */
-}
-
-static void nv04ClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-	/* TODO */
-}
-
-static void nv04ClearDepth(GLcontext *ctx, GLclampd d)
-{
-	/* TODO */
-}
-
-static void nv04ClearStencil(GLcontext *ctx, GLint s)
-{
-	/* TODO */
-}
-
-static void nv04ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
-{
-	/* TODO */
-}
-
-static void nv04ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
-		GLboolean bmask, GLboolean amask )
-{
-	/* TODO */
-}
-
-static void nv04ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	/* TODO I need love */
-}
-
-static void nv04CullFace(GLcontext *ctx, GLenum mode)
-{
-	nv04_emit_control(ctx);
-}
-
-static void nv04FrontFace(GLcontext *ctx, GLenum mode)
-{
-	/* TODO */
-}
-
-static void nv04DepthFunc(GLcontext *ctx, GLenum func)
-{
-	nv04_emit_control(ctx);
-}
-
-static void nv04DepthMask(GLcontext *ctx, GLboolean flag)
-{
-	/* TODO */
-}
-
-static void nv04DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
-{
-	/* TODO */
-}
-
-/** Specify the current buffer for writing */
-//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
-/** Specify the buffers for writing for fragment programs*/
-//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
-
-static void nv04Enable(GLcontext *ctx, GLenum cap, GLboolean state)
-{
-	switch(cap)
-	{
-		case GL_ALPHA_TEST:
-			nv04_emit_control(ctx);
-			break;
-//		case GL_AUTO_NORMAL:
-		case GL_BLEND:
-			nv04_emit_blend(ctx);
-			break;
-//		case GL_CLIP_PLANE0:
-//		case GL_CLIP_PLANE1:
-//		case GL_CLIP_PLANE2:
-//		case GL_CLIP_PLANE3:
-//		case GL_CLIP_PLANE4:
-//		case GL_CLIP_PLANE5:
-//		case GL_COLOR_LOGIC_OP:
-//		case GL_COLOR_MATERIAL:
-//		case GL_COLOR_SUM_EXT:
-//		case GL_COLOR_TABLE:
-//		case GL_CONVOLUTION_1D:
-//		case GL_CONVOLUTION_2D:
-		case GL_CULL_FACE:
-			nv04_emit_control(ctx);
-			break;
-		case GL_DEPTH_TEST:
-			nv04_emit_control(ctx);
-			break;
-		case GL_DITHER:
-			nv04_emit_control(ctx);
-			break;
-		case GL_FOG:
-			nv04_emit_blend(ctx);
-			nv04_emit_fog_color(ctx);
-			break;
-//		case GL_HISTOGRAM:
-//		case GL_INDEX_LOGIC_OP:
-//		case GL_LIGHT0:
-//		case GL_LIGHT1:
-//		case GL_LIGHT2:
-//		case GL_LIGHT3:
-//		case GL_LIGHT4:
-//		case GL_LIGHT5:
-//		case GL_LIGHT6:
-//		case GL_LIGHT7:
-//		case GL_LIGHTING:
-//		case GL_LINE_SMOOTH:
-//		case GL_LINE_STIPPLE:
-//		case GL_MAP1_COLOR_4:
-//		case GL_MAP1_INDEX:
-//		case GL_MAP1_NORMAL:
-//		case GL_MAP1_TEXTURE_COORD_1:
-//		case GL_MAP1_TEXTURE_COORD_2:
-//		case GL_MAP1_TEXTURE_COORD_3:
-//		case GL_MAP1_TEXTURE_COORD_4:
-//		case GL_MAP1_VERTEX_3:
-//		case GL_MAP1_VERTEX_4:
-//		case GL_MAP2_COLOR_4:
-//		case GL_MAP2_INDEX:
-//		case GL_MAP2_NORMAL:
-//		case GL_MAP2_TEXTURE_COORD_1:
-//		case GL_MAP2_TEXTURE_COORD_2:
-//		case GL_MAP2_TEXTURE_COORD_3:
-//		case GL_MAP2_TEXTURE_COORD_4:
-//		case GL_MAP2_VERTEX_3:
-//		case GL_MAP2_VERTEX_4:
-//		case GL_MINMAX:
-//		case GL_NORMALIZE:
-//		case GL_POINT_SMOOTH:
-//		case GL_POLYGON_OFFSET_POINT:
-//		case GL_POLYGON_OFFSET_LINE:
-//		case GL_POLYGON_OFFSET_FILL:
-//		case GL_POLYGON_SMOOTH:
-//		case GL_POLYGON_STIPPLE:
-//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
-//		case GL_POST_CONVOLUTION_COLOR_TABLE:
-//		case GL_RESCALE_NORMAL:
-//		case GL_SCISSOR_TEST:
-//		case GL_SEPARABLE_2D:
-//		case GL_STENCIL_TEST:
-//		case GL_TEXTURE_GEN_Q:
-//		case GL_TEXTURE_GEN_R:
-//		case GL_TEXTURE_GEN_S:
-//		case GL_TEXTURE_GEN_T:
-//		case GL_TEXTURE_1D:
-//		case GL_TEXTURE_2D:
-//		case GL_TEXTURE_3D:
-	}
-}
-
-static void nv04Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	nv04_emit_blend(ctx);
-	nv04_emit_fog_color(ctx);
-}
-   
-static void nv04Hint(GLcontext *ctx, GLenum target, GLenum mode)
-{
-	switch(target)
-	{
-		case GL_PERSPECTIVE_CORRECTION_HINT:nv04_emit_blend(ctx);break;
-		default:break;
-	}
-}
-
-static void nv04LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
-{
-	/* TODO not even in your dreams */
-}
-
-static void nv04LineWidth(GLcontext *ctx, GLfloat width)
-{
-	/* TODO */
-}
-
-static void nv04LogicOpcode(GLcontext *ctx, GLenum opcode)
-{
-	/* TODO */
-}
-
-static void nv04PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/* TODO */
-}
-
-static void nv04PointSize(GLcontext *ctx, GLfloat size)
-{
-	/* TODO */
-}
-
-static void nv04PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	/* TODO */
-}
-
-/** Set the scale and units used to calculate depth values */
-static void nv04PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
-{
-	/* TODO */
-}
-
-/** Set the polygon stippling pattern */
-static void nv04PolygonStipple(GLcontext *ctx, const GLubyte *mask )
-{
-	/* TODO */
-}
-
-/* Specifies the current buffer for reading */
-void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
-/** Set rasterization mode */
-void (*RenderMode)(GLcontext *ctx, GLenum mode );
-
-/** Define the scissor box */
-static void nv04Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-	/* TODO */
-}
-
-/** Select flat or smooth shading */
-static void nv04ShadeModel(GLcontext *ctx, GLenum mode)
-{
-	nv04_emit_blend(ctx);
-}
-
-/** OpenGL 2.0 two-sided StencilFunc */
-static void nv04StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
-		GLint ref, GLuint mask)
-{
-	/* TODO */
-}
-
-/** OpenGL 2.0 two-sided StencilMask */
-static void nv04StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
-{
-	/* TODO */
-}
-
-/** OpenGL 2.0 two-sided StencilOp */
-static void nv04StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
-		GLenum zfail, GLenum zpass)
-{
-	/* TODO */
-}
-
-/** Control the generation of texture coordinates */
-void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
-		const GLfloat *params);
-/** Set texture environment parameters */
-void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
-		const GLfloat *param);
-/** Set texture parameters */
-void (*TexParameter)(GLcontext *ctx, GLenum target,
-		struct gl_texture_object *texObj,
-		GLenum pname, const GLfloat *params);
-
-/* Update anything that depends on the window position/size */
-static void nv04WindowMoved(nouveauContextPtr nmesa)
-{
-}
-
-/* Initialise any card-specific non-GL related state */
-static GLboolean nv04InitCard(nouveauContextPtr nmesa)
-{
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf3D, NvCtxSurf3D);
-
-	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY, 3);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_SURFACE, 1);
-	OUT_RING(NvCtxSurf3D);
-	return GL_TRUE;
-}
-
-/* Update buffer offset/pitch/format */
-static GLboolean nv04BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
-{
-	GLuint x, y, w, h;
-	uint32_t depth_pitch=(depth?depth->pitch:0+15)&~15+16;
-	if (depth_pitch<256) depth_pitch=256;
-
-	w = color[0]->mesa.Width;
-	h = color[0]->mesa.Height;
-	x = nmesa->drawX;
-	y = nmesa->drawY;
-
-	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_FORMAT, 1);
-	if (color[0]->mesa._ActualFormat == GL_RGBA8)
-		OUT_RING(0x108/*A8R8G8B8*/);
-	else
-		OUT_RING(0x103/*R5G6B5*/);
-
-	/* FIXME pitches have to be aligned ! */
-	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
-	OUT_RING(color[0]->pitch|(depth_pitch<<16));
-	OUT_RING(color[0]->offset);
-	if (depth) {
-		BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
-		OUT_RING(depth->offset);
-	}
-
-//	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL, 2);
-//	OUT_RING((w<<16)|x);
-//	OUT_RING((h<<16)|y);
-
-
-	/* FIXME not sure... */
-/*	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_CLIP_SIZE, 1);
-	OUT_RING((h<<16)|w);*/
-
-	return GL_TRUE;
-}
-
-void nv04InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	func->AlphaFunc			= nv04AlphaFunc;
-	func->BlendColor		= nv04BlendColor;
-	func->BlendEquationSeparate	= nv04BlendEquationSeparate;
-	func->BlendFuncSeparate		= nv04BlendFuncSeparate;
-	func->Clear			= nv04Clear;
-	func->ClearColor		= nv04ClearColor;
-	func->ClearDepth		= nv04ClearDepth;
-	func->ClearStencil		= nv04ClearStencil;
-	func->ClipPlane			= nv04ClipPlane;
-	func->ColorMask			= nv04ColorMask;
-	func->ColorMaterial		= nv04ColorMaterial;
-	func->CullFace			= nv04CullFace;
-	func->FrontFace			= nv04FrontFace;
-	func->DepthFunc			= nv04DepthFunc;
-	func->DepthMask			= nv04DepthMask;
-	func->DepthRange		= nv04DepthRange;
-	func->Enable			= nv04Enable;
-	func->Fogfv			= nv04Fogfv;
-	func->Hint			= nv04Hint;
-/*	func->Lightfv			= nv04Lightfv;*/
-/*	func->LightModelfv		= nv04LightModelfv; */
-	func->LineStipple		= nv04LineStipple;		/* Not for NV04 */
-	func->LineWidth			= nv04LineWidth;
-	func->LogicOpcode		= nv04LogicOpcode;
-	func->PointParameterfv		= nv04PointParameterfv;
-	func->PointSize			= nv04PointSize;
-	func->PolygonMode		= nv04PolygonMode;
-	func->PolygonOffset		= nv04PolygonOffset;
-	func->PolygonStipple		= nv04PolygonStipple;		/* Not for NV04 */
-/*	func->ReadBuffer		= nv04ReadBuffer;*/
-/*	func->RenderMode		= nv04RenderMode;*/
-	func->Scissor			= nv04Scissor;
-	func->ShadeModel		= nv04ShadeModel;
-	func->StencilFuncSeparate	= nv04StencilFuncSeparate;
-	func->StencilMaskSeparate	= nv04StencilMaskSeparate;
-	func->StencilOpSeparate		= nv04StencilOpSeparate;
-/*	func->TexGen			= nv04TexGen;*/
-/*	func->TexParameter		= nv04TexParameter;*/
-/*	func->TextureMatrix		= nv04TextureMatrix;*/
-
-	nmesa->hw_func.InitCard		= nv04InitCard;
-	nmesa->hw_func.BindBuffers	= nv04BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv04WindowMoved;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv04_swtcl.c b/src/mesa/drivers/dri/nouveau/nv04_swtcl.c
deleted file mode 100644
index cb072e0bdbc..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv04_swtcl.c
+++ /dev/null
@@ -1,619 +0,0 @@
-/*
- * Copyright 2007 Stephane Marchesin. All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sub license,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * VIA, S3 GRAPHICS, AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- */
-
-/* Software TCL for NV04, NV05, NV06 */
-
-#include <stdio.h>
-#include <math.h>
-
-#include "glheader.h"
-#include "context.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
-#include "enums.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-
-#include "nouveau_swtcl.h"
-#include "nv04_swtcl.h"
-#include "nouveau_context.h"
-#include "nouveau_span.h"
-#include "nouveau_reg.h"
-#include "nouveau_tex.h"
-#include "nouveau_fifo.h"
-#include "nouveau_msg.h"
-#include "nouveau_object.h"
-
-static void nv04RasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
-static void nv04RenderPrimitive( GLcontext *ctx, GLenum prim );
-static void nv04ResetLineStipple( GLcontext *ctx );
-
-
-static inline void nv04_2triangles(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2,nouveauVertex* v3,nouveauVertex* v4,nouveauVertex* v5)
-{
-	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xA),49);
-	OUT_RINGp(v0,8);
-	OUT_RINGp(v1,8);
-	OUT_RINGp(v2,8);
-	OUT_RINGp(v3,8);
-	OUT_RINGp(v4,8);
-	OUT_RINGp(v5,8);
-	OUT_RING(0xFEDCBA);
-}
-
-static inline void nv04_1triangle(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2)
-{
-	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xD),25);
-	OUT_RINGp(v0,8);
-	OUT_RINGp(v1,8);
-	OUT_RINGp(v2,8);
-	OUT_RING(0xFED);
-}
-
-static inline void nv04_1quad(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2,nouveauVertex* v3)
-{
-	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xC),33);
-	OUT_RINGp(v0,8);
-	OUT_RINGp(v1,8);
-	OUT_RINGp(v2,8);
-	OUT_RINGp(v3,8);
-	OUT_RING(0xFECEDC);
-}
-
-static inline void nv04_render_points(GLcontext *ctx,GLuint first,GLuint last)
-{
-	WARN_ONCE("Unimplemented\n");
-}
-
-static inline void nv04_render_line(GLcontext *ctx,GLuint v1,GLuint v2)
-{
-	WARN_ONCE("Unimplemented\n");
-}
-
-static inline void nv04_render_triangle(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-
-	nv04_1triangle(nmesa,
-			(nouveauVertex*)(vertptr+v1*vertsize),
-			(nouveauVertex*)(vertptr+v2*vertsize),
-			(nouveauVertex*)(vertptr+v3*vertsize)
-		  );
-}
-
-static inline void nv04_render_quad(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3,GLuint v4)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-
-	nv04_1quad(nmesa,
-			(nouveauVertex*)(vertptr+v1*vertsize),
-			(nouveauVertex*)(vertptr+v2*vertsize),
-			(nouveauVertex*)(vertptr+v3*vertsize),
-			(nouveauVertex*)(vertptr+v4*vertsize)
-		  );
-}
-
-/**********************************************************************/
-/*               Render unclipped begin/end objects                   */
-/**********************************************************************/
-
-static void nv04_render_points_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// erm
-}
-
-static void nv04_render_lines_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// umm
-}
-
-static void nv04_render_line_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// yeah
-}
-
-static void nv04_render_line_loop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// right
-}
-
-static void nv04_render_triangles_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	int i;
-
-	for(i=start;i<count-5;i+=6)
-		nv04_2triangles(nmesa,
-				(nouveauVertex*)(vertptr+(i+0)*vertsize),
-				(nouveauVertex*)(vertptr+(i+1)*vertsize),
-				(nouveauVertex*)(vertptr+(i+2)*vertsize),
-				(nouveauVertex*)(vertptr+(i+3)*vertsize),
-				(nouveauVertex*)(vertptr+(i+4)*vertsize),
-				(nouveauVertex*)(vertptr+(i+5)*vertsize)
-			       );
-	if (i!=count)
-	{
-		nv04_1triangle(nmesa,
-				(nouveauVertex*)(vertptr+(i+0)*vertsize),
-				(nouveauVertex*)(vertptr+(i+1)*vertsize),
-				(nouveauVertex*)(vertptr+(i+2)*vertsize)
-			       );
-		i+=3;
-	}
-	if (i!=count)
-		printf("oops\n");
-}
-
-static void nv04_render_tri_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC};
-	int i,j;
-
-	for(i=start;i<count;i+=14)
-	{
-		int numvert=MIN2(16,count-i);
-		int numtri=numvert-2;
-		if (numvert<3)
-			break;
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),numvert*8);
-		for(j=0;j<numvert;j++)
-			OUT_RINGp((nouveauVertex*)(vertptr+(i+j)*vertsize),8);
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
-		for(j=0;j<numtri/2;j++)
-			OUT_RING(striptbl[j]);
-		if (numtri%2)
-			OUT_RING(striptbl[numtri/2]&0xFFF);
-	}
-}
-
-static void nv04_render_tri_fan_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	uint32_t fantbl[]={0x320210,0x540430,0x760650,0x980870,0xBA0A90,0xDC0CB0,0xFE0ED0};
-	int i,j;
-
-	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),8);
-	OUT_RINGp((nouveauVertex*)(vertptr+start*vertsize),8);
-
-	for(i=start+1;i<count;i+=14)
-	{
-		int numvert=MIN2(15,count-i);
-		int numtri=numvert-1;
-		if (numvert<3)
-			break;
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x1),numvert*8);
-
-		for(j=0;j<numvert;j++)
-			OUT_RINGp((nouveauVertex*)(vertptr+(i+j)*vertsize),8);
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
-		for(j=0;j<numtri/2;j++)
-			OUT_RING(fantbl[j]);
-		if (numtri%2)
-			OUT_RING(fantbl[numtri/2]&0xFFF);
-	}
-}
-
-static void nv04_render_quads_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	int i;
-
-	for(i=start;i<count;i+=4)
-		nv04_1quad(nmesa,
-				(nouveauVertex*)(vertptr+(i+0)*vertsize),
-				(nouveauVertex*)(vertptr+(i+1)*vertsize),
-				(nouveauVertex*)(vertptr+(i+2)*vertsize),
-				(nouveauVertex*)(vertptr+(i+3)*vertsize)
-			       );
-}
-
-static void nv04_render_noop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-}
-
-static void (*nv04_render_tab_verts[GL_POLYGON+2])(GLcontext *,
-							   GLuint,
-							   GLuint,
-							   GLuint) =
-{
-   nv04_render_points_verts,
-   nv04_render_lines_verts,
-   nv04_render_line_loop_verts,
-   nv04_render_line_strip_verts,
-   nv04_render_triangles_verts,
-   nv04_render_tri_strip_verts,
-   nv04_render_tri_fan_verts,
-   nv04_render_quads_verts,
-   nv04_render_tri_strip_verts,  //nv04_render_quad_strip_verts
-   nv04_render_tri_fan_verts,    //nv04_render_poly_verts
-   nv04_render_noop_verts,
-};
-
-
-static void nv04_render_points_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// erm
-}
-
-static void nv04_render_lines_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// umm
-}
-
-static void nv04_render_line_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// yeah
-}
-
-static void nv04_render_line_loop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	// right
-}
-
-static void nv04_render_triangles_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
-	int i;
-
-	for(i=start;i<count-5;i+=6)
-		nv04_2triangles(nmesa,
-				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+2]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+3]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+4]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+5]*vertsize)
-			       );
-	if (i!=count)
-	{
-		nv04_1triangle(nmesa,
-				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+2]*vertsize)
-			       );
-		i+=3;
-	}
-	if (i!=count)
-		printf("oops\n");
-}
-
-static void nv04_render_tri_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC};
-	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
-	int i,j;
-
-	for(i=start;i<count;i+=14)
-	{
-		int numvert=MIN2(16,count-i);
-		int numtri=numvert-2;
-		if (numvert<3)
-			break;
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),numvert*8);
-		for(j=0;j<numvert;j++)
-			OUT_RINGp((nouveauVertex*)(vertptr+elt[i+j]*vertsize),8);
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
-		for(j=0;j<numtri/2;j++)
-			OUT_RING(striptbl[j]);
-		if (numtri%2)
-			OUT_RING(striptbl[numtri/2]&0xFFF);
-	}
-}
-
-static void nv04_render_tri_fan_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	uint32_t fantbl[]={0x320210,0x540430,0x760650,0x980870,0xBA0A90,0xDC0CB0,0xFE0ED0};
-	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
-	int i,j;
-
-	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),8);
-	OUT_RINGp((nouveauVertex*)(vertptr+elt[start]*vertsize),8);
-
-	for(i=start+1;i<count;i+=14)
-	{
-		int numvert=MIN2(15,count-i);
-		int numtri=numvert-2;
-		if (numvert<3)
-			break;
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x1),numvert*8);
-
-		for(j=0;j<numvert;j++)
-			OUT_RINGp((nouveauVertex*)(vertptr+elt[i+j]*vertsize),8);
-
-		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
-		for(j=0;j<numtri/2;j++)
-			OUT_RING(fantbl[j]);
-		if (numtri%2)
-			OUT_RING(fantbl[numtri/2]&0xFFF);
-	}
-}
-
-static void nv04_render_quads_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
-	int i;
-
-	for(i=start;i<count;i+=4)
-		nv04_1quad(nmesa,
-				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+2]*vertsize),
-				(nouveauVertex*)(vertptr+elt[i+3]*vertsize)
-			       );
-}
-
-static void nv04_render_noop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-}
-
-static void (*nv04_render_tab_elts[GL_POLYGON+2])(GLcontext *,
-							   GLuint,
-							   GLuint,
-							   GLuint) =
-{
-   nv04_render_points_elts,
-   nv04_render_lines_elts,
-   nv04_render_line_loop_elts,
-   nv04_render_line_strip_elts,
-   nv04_render_triangles_elts,
-   nv04_render_tri_strip_elts,
-   nv04_render_tri_fan_elts,
-   nv04_render_quads_elts,
-   nv04_render_tri_strip_elts,   // nv04_render_quad_strip_elts,
-   nv04_render_tri_fan_elts,     // nv04_render_poly_elts,
-   nv04_render_noop_elts,
-};
-
-
-/**********************************************************************/
-/*                    Choose render functions                         */
-/**********************************************************************/
-
-
-#define EMIT_ATTR( ATTR, STYLE )					\
-do {									\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].attrib = (ATTR);	\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].format = (STYLE);	\
-   nmesa->vertex_attr_count++;						\
-} while (0)
-
-#define EMIT_PAD( N )							\
-do {									\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].attrib = 0;		\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].format = EMIT_PAD;	\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].offset = (N);		\
-   nmesa->vertex_attr_count++;						\
-} while (0)
-
-static void nv04_render_clipped_line(GLcontext *ctx,GLuint ii,GLuint jj)
-{
-}
-
-static void nv04_render_clipped_poly(GLcontext *ctx,const GLuint *elts,GLuint n)
-{
-}
-
-static void nv04ChooseRenderState(GLcontext *ctx)
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-	tnl->Driver.Render.PrimTabVerts = nv04_render_tab_verts;
-	tnl->Driver.Render.PrimTabElts = nv04_render_tab_elts;
-	tnl->Driver.Render.ClippedLine = nv04_render_clipped_line;
-	tnl->Driver.Render.ClippedPolygon = nv04_render_clipped_poly;
-	tnl->Driver.Render.Points = nv04_render_points;
-	tnl->Driver.Render.Line = nv04_render_line;
-	tnl->Driver.Render.Triangle = nv04_render_triangle;
-	tnl->Driver.Render.Quad = nv04_render_quad;
-}
-
-
-
-static inline void nv04OutputVertexFormat(struct nouveau_context* nmesa)
-{
-	GLcontext* ctx=nmesa->glCtx;
-	DECLARE_RENDERINPUTS(index);
-
-	/*
-	 * Tell t_vertex about the vertex format
-	 */
-	nmesa->vertex_attr_count = 0;
-	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
-
-	// SX SY SZ INVW
-	// FIXME : we use W instead of INVW, but since W=1 it doesn't matter
-	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_POS))
-		EMIT_ATTR(_TNL_ATTRIB_POS,EMIT_4F_VIEWPORT);
-	else
-		EMIT_PAD(4*sizeof(float));
-
-	// COLOR
-	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_COLOR0))
-		EMIT_ATTR(_TNL_ATTRIB_COLOR0,EMIT_4UB_4F_ABGR);
-	else
-		EMIT_PAD(4);
-
-	// SPECULAR
-	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_COLOR1))
-		EMIT_ATTR(_TNL_ATTRIB_COLOR1,EMIT_4UB_4F_ABGR);
-	else
-		EMIT_PAD(4);
-
-	// TEXTURE
-	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_TEX0))
-		EMIT_ATTR(_TNL_ATTRIB_TEX0,EMIT_2F);
-	else
-		EMIT_PAD(2*sizeof(float));
-
-	nmesa->vertex_size=_tnl_install_attrs( ctx,
-			nmesa->vertex_attrs, 
-			nmesa->vertex_attr_count,
-			nmesa->viewport.m, 0 );
-}
-
-
-static void nv04ChooseVertexState( GLcontext *ctx )
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	DECLARE_RENDERINPUTS(index);
-
-	RENDERINPUTS_COPY(index, tnl->render_inputs_bitset);
-	if (!RENDERINPUTS_EQUAL(index, nmesa->render_inputs_bitset))
-	{
-		RENDERINPUTS_COPY(nmesa->render_inputs_bitset, index);
-		nv04OutputVertexFormat(nmesa);
-	}
-}
-
-
-/**********************************************************************/
-/*                 High level hooks for t_vb_render.c                 */
-/**********************************************************************/
-
-
-static void nv04RenderStart(GLcontext *ctx)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (nmesa->new_state) {
-		nmesa->new_render_state |= nmesa->new_state;
-	}
-
-	if (nmesa->new_render_state) {
-		nv04ChooseVertexState(ctx);
-		nv04ChooseRenderState(ctx);
-		nmesa->new_render_state = 0;
-	}
-}
-
-static void nv04RenderFinish(GLcontext *ctx)
-{
-}
-
-
-/* System to flush dma and emit state changes based on the rasterized
- * primitive.
- */
-void nv04RasterPrimitive(GLcontext *ctx,
-		GLenum glprim,
-		GLuint hwprim)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	assert (!nmesa->new_state);
-
-	if (hwprim != nmesa->current_primitive)
-	{
-		nmesa->current_primitive=hwprim;
-		
-	}
-}
-
-static const GLuint hw_prim[GL_POLYGON+1] = {
-	GL_POINTS+1,
-	GL_LINES+1,
-	GL_LINE_STRIP+1,
-	GL_LINE_LOOP+1,
-	GL_TRIANGLES+1,
-	GL_TRIANGLE_STRIP+1,
-	GL_TRIANGLE_FAN+1,
-	GL_QUADS+1,
-	GL_QUAD_STRIP+1,
-	GL_POLYGON+1
-};
-
-/* Callback for mesa:
- */
-static void nv04RenderPrimitive( GLcontext *ctx, GLuint prim )
-{
-	nv04RasterPrimitive( ctx, prim, hw_prim[prim] );
-}
-
-static void nv04ResetLineStipple( GLcontext *ctx )
-{
-	/* FIXME do something here */
-	WARN_ONCE("Unimplemented nv04ResetLineStipple\n");
-}
-
-
-/**********************************************************************/
-/*                            Initialization.                         */
-/**********************************************************************/
-
-void nv04TriInitFunctions(GLcontext *ctx)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-	tnl->Driver.RunPipeline = nouveauRunPipeline;
-	tnl->Driver.Render.Start = nv04RenderStart;
-	tnl->Driver.Render.Finish = nv04RenderFinish;
-	tnl->Driver.Render.PrimitiveNotify = nv04RenderPrimitive;
-	tnl->Driver.Render.ResetLineStipple = nv04ResetLineStipple;
-	tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
-	tnl->Driver.Render.CopyPV = _tnl_copy_pv;
-	tnl->Driver.Render.Interp = _tnl_interp;
-
-	_tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 32 );
-
-	nmesa->verts = (GLubyte *)tnl->clipspace.vertex_buf;
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nv04_swtcl.h b/src/mesa/drivers/dri/nouveau/nv04_swtcl.h
deleted file mode 100644
index 42dde5383eb..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv04_swtcl.h
+++ /dev/null
@@ -1,12 +0,0 @@
-#ifndef __NV04_SWTCL_H__
-#define __NV04_SWTCL_H__
-
-#include "mtypes.h"
-
-extern void nv04Fallback( GLcontext *ctx, GLuint bit, GLboolean mode );
-extern void nv04FinishPrimitive(struct nouveau_context *nmesa);
-extern void nv04TriInitFunctions(GLcontext *ctx);
-#define FALLBACK( nmesa, bit, mode ) nouveauFallback( nmesa->glCtx, bit, mode )
-
-#endif /* __NV04_SWTCL_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state.c b/src/mesa/drivers/dri/nouveau/nv10_state.c
deleted file mode 100644
index 47c4b14ba6b..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv10_state.c
+++ /dev/null
@@ -1,797 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Nouveau
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-static void nv10ViewportScale(nouveauContextPtr nmesa)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-
-	GLfloat max_depth = (ctx->Viewport.Near + ctx->Viewport.Far) * 0.5;
-/*	if (ctx->DrawBuffer) {
-		switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
-			case 16:
-				max_depth *= 32767.0;
-				break;
-			case 24:
-				max_depth *= 16777215.0;
-				break;
-		}
-	} else {*/
-		/* Default to 24 bits range */	
-		max_depth *= 16777215.0;
-/*	}*/
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X, 4);
-	OUT_RING_CACHEf ((((GLfloat) w) * 0.5) - 2048.0);
-	OUT_RING_CACHEf ((((GLfloat) h) * 0.5) - 2048.0);
-	OUT_RING_CACHEf (max_depth);
-	OUT_RING_CACHEf (0.0);
-}
-
-static void nv10AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte ubRef;
-	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC, 2);
-	OUT_RING_CACHE(func);
-	OUT_RING_CACHE(ubRef);
-}
-
-static void nv10BlendColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); 
-	GLubyte cf[4];
-
-	CLAMPED_FLOAT_TO_UBYTE(cf[0], color[0]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[1], color[1]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[2], color[2]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[3], color[3]);
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_COLOR, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888(cf[3], cf[1], cf[2], cf[0]));
-}
-
-static void nv10BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	assert( modeRGB == modeA );
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION, 1);
-	OUT_RING_CACHE(modeRGB);
-}
-
-
-static void nv10BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
-		GLenum sfactorA, GLenum dfactorA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	assert( sfactorRGB == sfactorA );
-	assert( dfactorRGB == dfactorA );
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC, 2);
-	OUT_RING_CACHE(sfactorRGB);
-	OUT_RING_CACHE(dfactorRGB);
-}
-
-static void nv10Clear(GLcontext *ctx, GLbitfield mask)
-{
-	/* TODO */
-}
-
-static void nv10ClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte c[4];
-	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,color);
-	nmesa->clear_color_value = PACK_COLOR_8888(c[3],c[0],c[1],c[2]);
-}
-
-static void nv10ClearDepth(GLcontext *ctx, GLclampd d)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-/*	switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
-		case 16:
-			nmesa->clear_value = (uint32_t)(d*0x7FFF);
-			break;
-		case 24:*/
-			nmesa->clear_value = ((nmesa->clear_value&0x000000FF) |
-				(((uint32_t)(d*0xFFFFFF))<<8));
-/*			break;
-	}*/
-}
-
-static void nv10ClearStencil(GLcontext *ctx, GLint s)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-/*	if (ctx->DrawBuffer->_DepthBuffer->DepthBits == 24) {*/
-		nmesa->clear_value = ((nmesa->clear_value&0xFFFFFF00)|
-			(s&0x000000FF));
-/*	}*/
-}
-
-static void nv10ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane), 4);
-	OUT_RING_CACHEf(equation[0]);
-	OUT_RING_CACHEf(equation[1]);
-	OUT_RING_CACHEf(equation[2]);
-	OUT_RING_CACHEf(equation[3]);
-}
-
-static void nv10ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
-		GLboolean bmask, GLboolean amask )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_MASK, 1);
-	OUT_RING_CACHE(((amask && 0x01) << 24) | ((rmask && 0x01) << 16) | ((gmask && 0x01)<< 8) | ((bmask && 0x01) << 0));
-}
-
-static void nv10ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	/* TODO I need love */
-}
-
-static void nv10CullFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CULL_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv10FrontFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FRONT_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv10DepthFunc(GLcontext *ctx, GLenum func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC, 1);
-	OUT_RING_CACHE(func);
-}
-
-static void nv10DepthMask(GLcontext *ctx, GLboolean flag)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING_CACHE(flag);
-}
-
-static void nv10DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	GLfloat depth_scale = 16777216.0;
-	if (ctx->DrawBuffer->_DepthBuffer->DepthBits == 16) {
-		depth_scale = 32768.0;
-	}
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
-	OUT_RING_CACHEf(nearval * depth_scale);
-	OUT_RING_CACHEf(farval * depth_scale);
-
-	nv10ViewportScale(nmesa);
-}
-
-/** Specify the current buffer for writing */
-//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
-/** Specify the buffers for writing for fragment programs*/
-//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
-
-static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	switch(cap)
-	{
-		case GL_ALPHA_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_AUTO_NORMAL:
-		case GL_BLEND:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_CLIP_PLANE0:
-		case GL_CLIP_PLANE1:
-		case GL_CLIP_PLANE2:
-		case GL_CLIP_PLANE3:
-		case GL_CLIP_PLANE4:
-		case GL_CLIP_PLANE5:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(cap-GL_CLIP_PLANE0), 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_COLOR_LOGIC_OP:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_COLOR_MATERIAL:
-//		case GL_COLOR_SUM_EXT:
-//		case GL_COLOR_TABLE:
-//		case GL_CONVOLUTION_1D:
-//		case GL_CONVOLUTION_2D:
-		case GL_CULL_FACE:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DEPTH_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DITHER:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DITHER_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_FOG:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_HISTOGRAM:
-//		case GL_INDEX_LOGIC_OP:
-		case GL_LIGHT0:
-		case GL_LIGHT1:
-		case GL_LIGHT2:
-		case GL_LIGHT3:
-		case GL_LIGHT4:
-		case GL_LIGHT5:
-		case GL_LIGHT6:
-		case GL_LIGHT7:
-			{
-			uint32_t mask=1<<(2*(cap-GL_LIGHT0));
-			nmesa->enabled_lights=((nmesa->enabled_lights&mask)|(mask*state));
-			if (nmesa->lighting_enabled)
-			{
-				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			}
-			break;
-			}
-		case GL_LIGHTING:
-			nmesa->lighting_enabled=state;
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-			if (nmesa->lighting_enabled)
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			else
-				OUT_RING_CACHE(0x0);
-			break;
-		case GL_LINE_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_LINE_STIPPLE:
-//		case GL_MAP1_COLOR_4:
-//		case GL_MAP1_INDEX:
-//		case GL_MAP1_NORMAL:
-//		case GL_MAP1_TEXTURE_COORD_1:
-//		case GL_MAP1_TEXTURE_COORD_2:
-//		case GL_MAP1_TEXTURE_COORD_3:
-//		case GL_MAP1_TEXTURE_COORD_4:
-//		case GL_MAP1_VERTEX_3:
-//		case GL_MAP1_VERTEX_4:
-//		case GL_MAP2_COLOR_4:
-//		case GL_MAP2_INDEX:
-//		case GL_MAP2_NORMAL:
-//		case GL_MAP2_TEXTURE_COORD_1:
-//		case GL_MAP2_TEXTURE_COORD_2:
-//		case GL_MAP2_TEXTURE_COORD_3:
-//		case GL_MAP2_TEXTURE_COORD_4:
-//		case GL_MAP2_VERTEX_3:
-//		case GL_MAP2_VERTEX_4:
-//		case GL_MINMAX:
-		case GL_NORMALIZE:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POINT_SMOOTH:
-		case GL_POLYGON_OFFSET_POINT:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_LINE:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_FILL:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POLYGON_STIPPLE:
-//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
-//		case GL_POST_CONVOLUTION_COLOR_TABLE:
-//		case GL_RESCALE_NORMAL:
-//		case GL_SCISSOR_TEST:
-//		case GL_SEPARABLE_2D:
-		case GL_STENCIL_TEST:
-			// TODO BACK and FRONT ?
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_TEXTURE_GEN_Q:
-//		case GL_TEXTURE_GEN_R:
-//		case GL_TEXTURE_GEN_S:
-//		case GL_TEXTURE_GEN_T:
-//		case GL_TEXTURE_1D:
-//		case GL_TEXTURE_2D:
-//		case GL_TEXTURE_3D:
-	}
-}
-
-static void nv10Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    switch(pname)
-    {
-        case GL_FOG_MODE:
-            //BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 1);
-            //OUT_RING_CACHE (params);
-            break;
-            /* TODO: unsure about the rest.*/
-        default:
-            break;
-    }
-
-}
-   
-static void nv10Hint(GLcontext *ctx, GLenum target, GLenum mode)
-{
-	/* TODO I need love (fog and line_smooth hints) */
-}
-
-// void (*IndexMask)(GLcontext *ctx, GLuint mask);
-
-enum {
-	SPOTLIGHT_NO_UPDATE,
-	SPOTLIGHT_UPDATE_EXPONENT,
-	SPOTLIGHT_UPDATE_DIRECTION,
-	SPOTLIGHT_UPDATE_ALL
-};
-
-static void nv10Lightfv(GLcontext *ctx, GLenum light, GLenum pname, const GLfloat *params )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLint p = light - GL_LIGHT0;
-	struct gl_light *l = &ctx->Light.Light[p];
-	int spotlight_update = SPOTLIGHT_NO_UPDATE;
-
-	switch(pname)
-	{
-		case GL_AMBIENT:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_DIFFUSE:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPECULAR:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_POSITION:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPOT_DIRECTION:
-			spotlight_update = SPOTLIGHT_UPDATE_DIRECTION;
-			break;
-		case GL_SPOT_EXPONENT:
-			spotlight_update = SPOTLIGHT_UPDATE_EXPONENT;
-			break;
-		case GL_SPOT_CUTOFF:
-			spotlight_update = SPOTLIGHT_UPDATE_ALL;
-			break;
-		case GL_CONSTANT_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_LINEAR_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_QUADRATIC_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		default:
-			break;
-	}
-
-	switch(spotlight_update) {
-		case SPOTLIGHT_UPDATE_DIRECTION:
-			{
-				GLfloat x,y,z;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(p), 3);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_EXPONENT:
-			{
-				GLfloat cc,lc,qc;
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 3);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_ALL:
-			{
-				GLfloat cc,lc,qc, x,y,z, c;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				c = spot_light_coef_a + 1.0;
-				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 7);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-				OUT_RING_CACHEf(c);
-			}
-			break;
-		default:
-			break;
-	}
-}
-
-/** Set the lighting model parameters */
-static void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params);
-
-
-static void nv10LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
-{
-	/* Not for NV10 */
-}
-
-static void nv10LineWidth(GLcontext *ctx, GLfloat width)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LINE_WIDTH, 1);
-	OUT_RING_CACHE(((int) (width * 8.0)) & -4);
-}
-
-static void nv10LogicOpcode(GLcontext *ctx, GLenum opcode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP, 1);
-	OUT_RING_CACHE(opcode);
-}
-
-static void nv10PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/*TODO: not sure what goes here. */
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	
-}
-
-static void nv10PointSize(GLcontext *ctx, GLfloat size)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
-	OUT_RING_CACHE(((int) (size * 8.0)) & -4);
-}
-
-static void nv10PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT, 1);
-		OUT_RING_CACHE(mode);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK, 1);
-		OUT_RING_CACHE(mode);
-	}
-}
-
-/** Set the scale and units used to calculate depth values */
-static void nv10PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 2);
-        OUT_RING_CACHEf(factor);
-        OUT_RING_CACHEf(units);
-}
-
-/** Set the polygon stippling pattern */
-static void nv10PolygonStipple(GLcontext *ctx, const GLubyte *mask )
-{
-	/* Not for NV10 */
-}
-
-/* Specifies the current buffer for reading */
-void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
-/** Set rasterization mode */
-void (*RenderMode)(GLcontext *ctx, GLenum mode );
-
-/** Define the scissor box */
-static void nv10Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-}
-
-/** Select flat or smooth shading */
-static void nv10ShadeModel(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_SHADE_MODEL, 1);
-	OUT_RING_CACHE(mode);
-}
-
-/** OpenGL 2.0 two-sided StencilFunc */
-static void nv10StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
-		GLint ref, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	/* NV10 do not have separate FRONT and BACK stencils */
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC, 3);
-	OUT_RING_CACHE(func);
-	OUT_RING_CACHE(ref);
-	OUT_RING_CACHE(mask);
-}
-
-/** OpenGL 2.0 two-sided StencilMask */
-static void nv10StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	/* NV10 do not have separate FRONT and BACK stencils */
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_MASK, 1);
-	OUT_RING_CACHE(mask);
-}
-
-/** OpenGL 2.0 two-sided StencilOp */
-static void nv10StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
-		GLenum zfail, GLenum zpass)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	/* NV10 do not have separate FRONT and BACK stencils */
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL, 3);
-	OUT_RING_CACHE(fail);
-	OUT_RING_CACHE(zfail);
-	OUT_RING_CACHE(zpass);
-}
-
-/** Control the generation of texture coordinates */
-void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
-		const GLfloat *params);
-/** Set texture environment parameters */
-void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
-		const GLfloat *param);
-/** Set texture parameters */
-void (*TexParameter)(GLcontext *ctx, GLenum target,
-		struct gl_texture_object *texObj,
-		GLenum pname, const GLfloat *params);
-
-static void nv10TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
-        /*XXX: This SHOULD work.*/
-        OUT_RING_CACHEp(mat->m, 16);
-}
-
-/* Update anything that depends on the window position/size */
-static void nv10WindowMoved(nouveauContextPtr nmesa)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLfloat *v = nmesa->viewport.m;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-	GLuint x = ctx->Viewport.X + nmesa->drawX;
-	GLuint y = ctx->Viewport.Y + nmesa->drawY;
-	int i;
-
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
-
-	/* something to do with clears, possibly doesn't belong here */
-	BEGIN_RING_SIZE(NvSub3D, 0x02b4, 1);
-	OUT_RING(0);
-
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
-        OUT_RING_CACHE(((w+x-1) << 16) | x | 0x08000800);
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
-        OUT_RING_CACHE(((h+y-1) << 16) | y | 0x08000800);
-	for (i=1; i<8; i++) {
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 1);
-        	OUT_RING_CACHE(0);
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(i), 1);
-	        OUT_RING_CACHE(0);
-	}
-
-	nv10ViewportScale(nmesa);
-}
-
-/* Initialise any card-specific non-GL related state */
-static GLboolean nv10InitCard(nouveauContextPtr nmesa)
-{
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-
-	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY0, 2);
-	OUT_RING(NvDmaFB);	/* 184 dma_in_memory0 */
-	OUT_RING(NvDmaFB);	/* 188 dma_in_memory1 */
-	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY2, 2);
-	OUT_RING(NvDmaFB);	/* 194 dma_in_memory2 */
-	OUT_RING(NvDmaFB);	/* 198 dma_in_memory3 */
-
-	BEGIN_RING_SIZE(NvSub3D, 0x0290, 1);
-	OUT_RING(0x00100001);
-	BEGIN_RING_SIZE(NvSub3D, 0x03f4, 1);
-	OUT_RING(0);
-
-	if (nmesa->screen->card->type >= NV_11) {
-	        BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
-        	OUT_RING(0);
-	        OUT_RING(1);
-	        OUT_RING(2);
-	}
-
-	return GL_TRUE;
-}
-
-/* Update buffer offset/pitch/format */
-static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
-{
-	GLuint x, y, w, h;
-	GLuint pitch, format, depth_pitch;
-
-	w = color[0]->mesa.Width;
-	h = color[0]->mesa.Height;
-	x = nmesa->drawX;
-	y = nmesa->drawY;
-
-	if (num_color != 1)
-		return GL_FALSE;
-
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 6);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
-	depth_pitch = (depth ? depth->pitch : color[0]->pitch);
-	pitch = (depth_pitch<<16) | color[0]->pitch;
-	format = 0x108;
-	if (color[0]->mesa._ActualFormat != GL_RGBA8) {
-		format = 0x103; /* R5G6B5 color buffer */
-	}
-	OUT_RING_CACHE(format);
-	OUT_RING_CACHE(pitch);
-	OUT_RING_CACHE(color[0]->offset);
-	OUT_RING_CACHE(depth ? depth->offset : color[0]->offset);
-
-	/* Always set to bottom left of buffer */
-	/*BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf ((GLfloat) h);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (0.0);*/
-
-	return GL_TRUE;
-}
-
-void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	func->AlphaFunc			= nv10AlphaFunc;
-	func->BlendColor		= nv10BlendColor;
-	func->BlendEquationSeparate	= nv10BlendEquationSeparate;
-	func->BlendFuncSeparate		= nv10BlendFuncSeparate;
-	func->Clear			= nv10Clear;
-	func->ClearColor		= nv10ClearColor;
-	func->ClearDepth		= nv10ClearDepth;
-	func->ClearStencil		= nv10ClearStencil;
-	func->ClipPlane			= nv10ClipPlane;
-	func->ColorMask			= nv10ColorMask;
-	func->ColorMaterial		= nv10ColorMaterial;
-	func->CullFace			= nv10CullFace;
-	func->FrontFace			= nv10FrontFace;
-	func->DepthFunc			= nv10DepthFunc;
-	func->DepthMask			= nv10DepthMask;
-	func->DepthRange		= nv10DepthRange;
-	func->Enable			= nv10Enable;
-	func->Fogfv			= nv10Fogfv;
-	func->Hint			= nv10Hint;
-	func->Lightfv			= nv10Lightfv;
-/*	func->LightModelfv		= nv10LightModelfv; */
-	func->LineStipple		= nv10LineStipple;		/* Not for NV10 */
-	func->LineWidth			= nv10LineWidth;
-	func->LogicOpcode		= nv10LogicOpcode;
-	func->PointParameterfv		= nv10PointParameterfv;
-	func->PointSize			= nv10PointSize;
-	func->PolygonMode		= nv10PolygonMode;
-	func->PolygonOffset		= nv10PolygonOffset;
-	func->PolygonStipple		= nv10PolygonStipple;		/* Not for NV10 */
-/*	func->ReadBuffer		= nv10ReadBuffer;*/
-/*	func->RenderMode		= nv10RenderMode;*/
-	func->Scissor			= nv10Scissor;
-	func->ShadeModel		= nv10ShadeModel;
-	func->StencilFuncSeparate	= nv10StencilFuncSeparate;
-	func->StencilMaskSeparate	= nv10StencilMaskSeparate;
-	func->StencilOpSeparate		= nv10StencilOpSeparate;
-/*	func->TexGen			= nv10TexGen;*/
-/*	func->TexParameter		= nv10TexParameter;*/
-	func->TextureMatrix		= nv10TextureMatrix;
-
-	nmesa->hw_func.InitCard		= nv10InitCard;
-	nmesa->hw_func.BindBuffers	= nv10BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv10WindowMoved;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
deleted file mode 100644
index 611469b6e41..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ /dev/null
@@ -1,639 +0,0 @@
-/*
- * Copyright 1998-2003 VIA Technologies, Inc. All Rights Reserved.
- * Copyright 2001-2003 S3 Graphics, Inc. All Rights Reserved.
- * Copyright 2006 Stephane Marchesin. All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sub license,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * VIA, S3 GRAPHICS, AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- */
-
-/* Software TCL for NV10, NV20, NV30, NV40, NV50 */
-
-#include <stdio.h>
-#include <math.h>
-
-#include "glheader.h"
-#include "context.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
-#include "enums.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-
-#include "nouveau_swtcl.h"
-#include "nv10_swtcl.h"
-#include "nouveau_context.h"
-#include "nouveau_span.h"
-#include "nouveau_reg.h"
-#include "nouveau_tex.h"
-#include "nouveau_fifo.h"
-#include "nouveau_msg.h"
-#include "nouveau_object.h"
-
-static void nv10RasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
-static void nv10RenderPrimitive( GLcontext *ctx, GLenum prim );
-static void nv10ResetLineStipple( GLcontext *ctx );
-
-
-
-static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t primitive,uint32_t size)
-{
-	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
-		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	else if (nmesa->screen->card->type==NV_20)
-		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	else
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	OUT_RING(primitive);
-
-	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
-		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA|NONINC_METHOD,size);
-	else if (nmesa->screen->card->type==NV_20)
-		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_VERTEX_DATA|NONINC_METHOD,size);
-	else
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_DATA|NONINC_METHOD,size);
-}
-
-inline void nv10FinishPrimitive(struct nouveau_context *nmesa)
-{
-	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
-		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	else if (nmesa->screen->card->type==NV_20)
-		BEGIN_RING_SIZE(NvSub3D,NV20_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	else
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_BEGIN_END,1);
-	OUT_RING(0x0);
-	FIRE_RING();
-}
-
-
-static inline void nv10ExtendPrimitive(struct nouveau_context* nmesa, int size)
-{
-	/* make sure there's enough room. if not, wait */
-	if (RING_AVAILABLE()<size)
-	{
-		WAIT_RING(nmesa,size);
-	}
-}
-
-/**********************************************************************/
-/*               Render unclipped begin/end objects                   */
-/**********************************************************************/
-
-static inline void nv10_render_generic_primitive_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags,GLuint prim)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(count-start)/4;
-
-	nv10ExtendPrimitive(nmesa, size_dword);
-	nv10StartPrimitive(nmesa,prim+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(start*vertsize)),size_dword);
-	nv10FinishPrimitive(nmesa);
-}
-
-static void nv10_render_points_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_POINTS);
-}
-
-static void nv10_render_lines_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_LINES);
-}
-
-static void nv10_render_line_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_LINE_STRIP);
-}
-
-static void nv10_render_line_loop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_LINE_LOOP);
-}
-
-static void nv10_render_triangles_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_TRIANGLES);
-}
-
-static void nv10_render_tri_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_TRIANGLE_STRIP);
-}
-
-static void nv10_render_tri_fan_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_TRIANGLE_FAN);
-}
-
-static void nv10_render_quads_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_QUADS);
-}
-
-static void nv10_render_quad_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_QUAD_STRIP);
-}
-
-static void nv10_render_poly_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_verts(ctx,start,count,flags,GL_POLYGON);
-}
-
-static void nv10_render_noop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-}
-
-static void (*nv10_render_tab_verts[GL_POLYGON+2])(GLcontext *,
-							   GLuint,
-							   GLuint,
-							   GLuint) =
-{
-   nv10_render_points_verts,
-   nv10_render_lines_verts,
-   nv10_render_line_loop_verts,
-   nv10_render_line_strip_verts,
-   nv10_render_triangles_verts,
-   nv10_render_tri_strip_verts,
-   nv10_render_tri_fan_verts,
-   nv10_render_quads_verts,
-   nv10_render_quad_strip_verts,
-   nv10_render_poly_verts,
-   nv10_render_noop_verts,
-};
-
-
-static inline void nv10_render_generic_primitive_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags,GLuint prim)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(count-start)/4;
-	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
-	GLuint j;
-
-	nv10ExtendPrimitive(nmesa, size_dword);
-	nv10StartPrimitive(nmesa,prim+1,size_dword);
-	for (j=start; j<count; j++ ) {
-		OUT_RINGp((nouveauVertex*)(vertptr+(elt[j]*vertsize)),vertsize/4);
-	}
-	nv10FinishPrimitive(nmesa);
-}
-
-static void nv10_render_points_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_POINTS);
-}
-
-static void nv10_render_lines_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_LINES);
-}
-
-static void nv10_render_line_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_LINE_STRIP);
-}
-
-static void nv10_render_line_loop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_LINE_LOOP);
-}
-
-static void nv10_render_triangles_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_TRIANGLES);
-}
-
-static void nv10_render_tri_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_TRIANGLE_STRIP);
-}
-
-static void nv10_render_tri_fan_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_TRIANGLE_FAN);
-}
-
-static void nv10_render_quads_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_QUADS);
-}
-
-static void nv10_render_quad_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_QUAD_STRIP);
-}
-
-static void nv10_render_poly_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-	nv10_render_generic_primitive_elts(ctx,start,count,flags,GL_POLYGON);
-}
-
-static void nv10_render_noop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
-{
-}
-
-static void (*nv10_render_tab_elts[GL_POLYGON+2])(GLcontext *,
-							   GLuint,
-							   GLuint,
-							   GLuint) =
-{
-   nv10_render_points_elts,
-   nv10_render_lines_elts,
-   nv10_render_line_loop_elts,
-   nv10_render_line_strip_elts,
-   nv10_render_triangles_elts,
-   nv10_render_tri_strip_elts,
-   nv10_render_tri_fan_elts,
-   nv10_render_quads_elts,
-   nv10_render_quad_strip_elts,
-   nv10_render_poly_elts,
-   nv10_render_noop_elts,
-};
-
-
-/**********************************************************************/
-/*                    Choose render functions                         */
-/**********************************************************************/
-
-
-#define EMIT_ATTR( ATTR, STYLE )					\
-do {									\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].attrib = (ATTR);	\
-   nmesa->vertex_attrs[nmesa->vertex_attr_count].format = (STYLE);	\
-   nmesa->vertex_attr_count++;						\
-} while (0)
-
-static void nv10_render_clipped_line(GLcontext *ctx,GLuint ii,GLuint jj)
-{
-
-}
-
-static void nv10_render_clipped_poly(GLcontext *ctx,const GLuint *elts,GLuint n)
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *VB = &tnl->vb;
-	GLuint *tmp = VB->Elts;
-	VB->Elts = (GLuint *)elts;
-	nv10_render_generic_primitive_elts( ctx, 0, n, PRIM_BEGIN|PRIM_END,GL_POLYGON );
-	VB->Elts = tmp;
-}
-
-static inline void nv10_render_points(GLcontext *ctx,GLuint first,GLuint last)
-{
-	WARN_ONCE("Unimplemented\n");
-}
-
-static inline void nv10_render_line(GLcontext *ctx,GLuint v1,GLuint v2)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(2)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
-
-	nv10ExtendPrimitive(nmesa, size_dword);
-	nv10StartPrimitive(nmesa,GL_LINES+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
-	nv10FinishPrimitive(nmesa);
-}
-
-static inline void nv10_render_triangle(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(3)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
-
-	nv10ExtendPrimitive(nmesa, size_dword);
-	nv10StartPrimitive(nmesa,GL_TRIANGLES+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
-	nv10FinishPrimitive(nmesa);
-}
-
-static inline void nv10_render_quad(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3,GLuint v4)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
-	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(4)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
-
-	nv10ExtendPrimitive(nmesa, size_dword);
-	nv10StartPrimitive(nmesa,GL_QUADS+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v4*vertsize)),vertsize);
-	nv10FinishPrimitive(nmesa);
-}
-
-
-
-static void nv10ChooseRenderState(GLcontext *ctx)
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	tnl->Driver.Render.PrimTabVerts = nv10_render_tab_verts;
-	tnl->Driver.Render.PrimTabElts = nv10_render_tab_elts;
-	tnl->Driver.Render.ClippedLine = nv10_render_clipped_line;
-	tnl->Driver.Render.ClippedPolygon = nv10_render_clipped_poly;
-	tnl->Driver.Render.Points = nv10_render_points;
-	tnl->Driver.Render.Line = nv10_render_line;
-	tnl->Driver.Render.Triangle = nv10_render_triangle;
-	tnl->Driver.Render.Quad = nv10_render_quad;
-}
-
-
-
-static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
-{
-	GLcontext* ctx=nmesa->glCtx;
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	DECLARE_RENDERINPUTS(index);
-	struct vertex_buffer *VB = &tnl->vb;
-	int attr_size[16];
-	int default_attr_size[8]={3,3,3,4,3,1,4,4};
-	int i;
-	int slots=0;
-	int total_size=0;
-
-	nmesa->vertex_attr_count = 0;
-	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
-
-	/*
-	 * Determine attribute sizes
-	 */
-	for(i=0;i<8;i++)
-	{
-		if (RENDERINPUTS_TEST(index, i))
-			attr_size[i]=default_attr_size[i];
-		else
-			attr_size[i]=0;
-	}
-	for(i=8;i<16;i++)
-	{
-		if (RENDERINPUTS_TEST(index, i))
-			attr_size[i]=VB->TexCoordPtr[i-8]->size;
-		else
-			attr_size[i]=0;
-	}
-
-	/*
-	 * Tell t_vertex about the vertex format
-	 */
-	for(i=0;i<16;i++)
-	{
-		if (RENDERINPUTS_TEST(index, i))
-		{
-			slots=i+1;
-			switch(attr_size[i])
-			{
-				case 1:
-					EMIT_ATTR(i,EMIT_1F);
-					break;
-				case 2:
-					EMIT_ATTR(i,EMIT_2F);
-					break;
-				case 3:
-					EMIT_ATTR(i,EMIT_3F);
-					break;
-				case 4:
-					EMIT_ATTR(i,EMIT_4F);
-					break;
-			}
-			if (i==_TNL_ATTRIB_COLOR0)
-				nmesa->color_offset=total_size;
-			if (i==_TNL_ATTRIB_COLOR1)
-				nmesa->specular_offset=total_size;
-			total_size+=attr_size[i];
-		}
-	}
-
-	nmesa->vertex_size=_tnl_install_attrs( ctx,
-			nmesa->vertex_attrs, 
-			nmesa->vertex_attr_count,
-			NULL, 0 );
-	assert(nmesa->vertex_size==total_size*4);
-
-	/* 
-	 * Tell the hardware about the vertex format
-	 */
-	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) {
-		int size;
-
-#define NV_VERTEX_ATTRIBUTE_TYPE_FLOAT 2
-
-#define NV10_SET_VERTEX_ATTRIB(i,j) \
-	do {	\
-		size = attr_size[j] << 4;	\
-		size |= (attr_size[j]*4) << 8;	\
-		size |= NV_VERTEX_ATTRIBUTE_TYPE_FLOAT;	\
-		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR(i),1);	\
-		OUT_RING_CACHE(size);	\
-	} while (0)
-
-		NV10_SET_VERTEX_ATTRIB(0, _TNL_ATTRIB_POS);
-		NV10_SET_VERTEX_ATTRIB(1, _TNL_ATTRIB_COLOR0);
-		NV10_SET_VERTEX_ATTRIB(2, _TNL_ATTRIB_COLOR1);
-		NV10_SET_VERTEX_ATTRIB(3, _TNL_ATTRIB_TEX0);
-		NV10_SET_VERTEX_ATTRIB(4, _TNL_ATTRIB_TEX1);
-		NV10_SET_VERTEX_ATTRIB(5, _TNL_ATTRIB_NORMAL);
-		NV10_SET_VERTEX_ATTRIB(6, _TNL_ATTRIB_WEIGHT);
-		NV10_SET_VERTEX_ATTRIB(7, _TNL_ATTRIB_FOG);
-
-		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE,1);
-		OUT_RING_CACHE(0);
-	} else if (nmesa->screen->card->type==NV_20) {
-		for(i=0;i<16;i++)
-		{
-			int size=attr_size[i];
-			BEGIN_RING_CACHE(NvSub3D,NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR(i),1);
-			OUT_RING_CACHE(NV_VERTEX_ATTRIBUTE_TYPE_FLOAT|(size*0x10));
-		}
-	} else {
-		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DO_VERTICES, 1);
-		OUT_RING(0);
-		BEGIN_RING_CACHE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS,slots);
-		for(i=0;i<slots;i++)
-		{
-			int size=attr_size[i];
-			OUT_RING_CACHE(NV_VERTEX_ATTRIBUTE_TYPE_FLOAT|(size*0x10));
-		}
-		// FIXME this is probably not needed
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0,1);
-		OUT_RING(0);
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0,1);
-		OUT_RING(0);
-		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0,1);
-		OUT_RING(0);
-	}
-}
-
-
-static void nv10ChooseVertexState( GLcontext *ctx )
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	DECLARE_RENDERINPUTS(index);
-	
-	RENDERINPUTS_COPY(index, tnl->render_inputs_bitset);
-	if (!RENDERINPUTS_EQUAL(index, nmesa->render_inputs_bitset))
-	{
-		RENDERINPUTS_COPY(nmesa->render_inputs_bitset, index);
-		nv10OutputVertexFormat(nmesa);
-	}
-
-	if (nmesa->screen->card->type == NV_30) {
-		nouveauShader *fp;
-		
-		if (ctx->FragmentProgram.Enabled) {
-			fp = (nouveauShader *) ctx->FragmentProgram.Current;
-			nvsUpdateShader(ctx, fp);
-		} else
-			nvsUpdateShader(ctx, nmesa->passthrough_fp);
-	}
-
-	if (nmesa->screen->card->type >= NV_40) {
-		/* Ensure passthrough shader is being used, and mvp matrix
-		 * is up to date
-		 */
-		nvsUpdateShader(ctx, nmesa->passthrough_vp);
-
-		/* Update texenv shader / user fragprog */
-		nvsUpdateShader(ctx, (nouveauShader*)ctx->FragmentProgram._Current);
-	}
-}
-
-
-/**********************************************************************/
-/*                 High level hooks for t_vb_render.c                 */
-/**********************************************************************/
-
-
-static void nv10RenderStart(GLcontext *ctx)
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (nmesa->new_state) {
-		nmesa->new_render_state |= nmesa->new_state;
-	}
-
-	if (nmesa->new_render_state) {
-		nv10ChooseVertexState(ctx);
-		nv10ChooseRenderState(ctx);
-		nmesa->new_render_state = 0;
-	}
-}
-
-static void nv10RenderFinish(GLcontext *ctx)
-{
-}
-
-
-/* System to flush dma and emit state changes based on the rasterized
- * primitive.
- */
-void nv10RasterPrimitive(GLcontext *ctx,
-		GLenum glprim,
-		GLuint hwprim)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-
-	assert (!nmesa->new_state);
-
-	if (hwprim != nmesa->current_primitive)
-	{
-		nmesa->current_primitive=hwprim;
-		
-	}
-}
-
-static const GLuint hw_prim[GL_POLYGON+1] = {
-	GL_POINTS+1,
-	GL_LINES+1,
-	GL_LINE_STRIP+1,
-	GL_LINE_LOOP+1,
-	GL_TRIANGLES+1,
-	GL_TRIANGLE_STRIP+1,
-	GL_TRIANGLE_FAN+1,
-	GL_QUADS+1,
-	GL_QUAD_STRIP+1,
-	GL_POLYGON+1
-};
-
-/* Callback for mesa:
- */
-static void nv10RenderPrimitive( GLcontext *ctx, GLuint prim )
-{
-	nv10RasterPrimitive( ctx, prim, hw_prim[prim] );
-}
-
-static void nv10ResetLineStipple( GLcontext *ctx )
-{
-	/* FIXME do something here */
-	WARN_ONCE("Unimplemented nv10ResetLineStipple\n");
-}
-
-
-/**********************************************************************/
-/*                            Initialization.                         */
-/**********************************************************************/
-
-void nv10TriInitFunctions(GLcontext *ctx)
-{
-	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-	tnl->Driver.RunPipeline = nouveauRunPipeline;
-	tnl->Driver.Render.Start = nv10RenderStart;
-	tnl->Driver.Render.Finish = nv10RenderFinish;
-	tnl->Driver.Render.PrimitiveNotify = nv10RenderPrimitive;
-	tnl->Driver.Render.ResetLineStipple = nv10ResetLineStipple;
-	tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
-	tnl->Driver.Render.CopyPV = _tnl_copy_pv;
-	tnl->Driver.Render.Interp = _tnl_interp;
-
-	_tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-			64 * sizeof(GLfloat) );
-
-	nmesa->verts = (GLubyte *)tnl->clipspace.vertex_buf;
-}
-
-
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.h b/src/mesa/drivers/dri/nouveau/nv10_swtcl.h
deleted file mode 100644
index 7c854addd2d..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Stephane Marchesin
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-
-
-#ifndef __NV10_SWTCL_H__
-#define __NV10_SWTCL_H__
-
-#include "mtypes.h"
-
-extern void nv10Fallback( GLcontext *ctx, GLuint bit, GLboolean mode );
-extern void nv10FinishPrimitive(struct nouveau_context *nmesa);
-extern void nv10TriInitFunctions(GLcontext *ctx);
-#define FALLBACK( nmesa, bit, mode ) nouveauFallback( nmesa->glCtx, bit, mode )
-
-#endif /* __NV10_SWTCL_H__ */
-
diff --git a/src/mesa/drivers/dri/nouveau/nv20_shader.h b/src/mesa/drivers/dri/nouveau/nv20_shader.h
deleted file mode 100644
index 7d2e29db668..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv20_shader.h
+++ /dev/null
@@ -1,121 +0,0 @@
-/* NV20_TCL_PRIMITIVE_3D_0x0B00 */
-#define NV20_VP_INST_0B00	0x00000000 /* always 0? */
-#define NV20_VP_INST0_KNOWN	0
-
-/* NV20_TCL_PRIMITIVE_3D_0x0B04 */
-#define NV20_VP_INST_SCA_OPCODE_SHIFT				25
-#define NV20_VP_INST_SCA_OPCODE_MASK				(0x0F << 25)
-#define NV20_VP_INST_OPCODE_RCP	0x2
-#define NV20_VP_INST_OPCODE_RCC	0x3
-#define NV20_VP_INST_OPCODE_RSQ	0x4
-#define NV20_VP_INST_OPCODE_EXP	0x5
-#define NV20_VP_INST_OPCODE_LOG	0x6
-#define NV20_VP_INST_OPCODE_LIT	0x7
-#define NV20_VP_INST_VEC_OPCODE_SHIFT				21
-#define NV20_VP_INST_VEC_OPCODE_MASK				(0x0F << 21)
-#define NV20_VP_INST_OPCODE_NOP	0x0 /* guess */
-#define NV20_VP_INST_OPCODE_MOV	0x1
-#define NV20_VP_INST_OPCODE_MUL	0x2
-#define NV20_VP_INST_OPCODE_ADD	0x3
-#define NV20_VP_INST_OPCODE_MAD	0x4
-#define NV20_VP_INST_OPCODE_DP3	0x5
-#define NV20_VP_INST_OPCODE_DPH	0x6
-#define NV20_VP_INST_OPCODE_DP4	0x7
-#define NV20_VP_INST_OPCODE_DST	0x8
-#define NV20_VP_INST_OPCODE_MIN	0x9
-#define NV20_VP_INST_OPCODE_MAX	0xA
-#define NV20_VP_INST_OPCODE_SLT	0xB
-#define NV20_VP_INST_OPCODE_SGE	0xC
-#define NV20_VP_INST_OPCODE_ARL	0xD
-#define NV20_VP_INST_CONST_SRC_SHIFT				13
-#define NV20_VP_INST_CONST_SRC_MASK				(0xFF << 13)
-#define NV20_VP_INST_INPUT_SRC_SHIFT				9
-#define NV20_VP_INST_INPUT_SRC_MASK				(0xF << 9) /* guess */
-#define NV20_VP_INST_INPUT_SRC_POS	0
-#define NV20_VP_INST_INPUT_SRC_COL0	3
-#define NV20_VP_INST_INPUT_SRC_COL1	4
-#define NV20_VP_INST_INPUT_SRC_TC(n)	(9+n)
-#define NV20_VP_INST_SRC0H_SHIFT				0
-#define NV20_VP_INST_SRC0H_MASK					(0x1FF << 0)
-#define NV20_VP_INST1_KNOWN	( \
-	NV20_VP_INST_OPCODE_MASK | \
-	NV20_VP_INST_CONST_SRC_MASK | \
-	NV20_VP_INST_INPUT_SRC_MASK | \
-	NV20_VP_INST_SRC0H_MASK \
-	)
-
-/* NV20_TCL_PRIMITIVE_3D_0x0B08 */
-#define NV20_VP_INST_SRC0L_SHIFT				26
-#define NV20_VP_INST_SRC0L_MASK					(0x3F  <<26)
-#define NV20_VP_INST_SRC1_SHIFT					11
-#define NV20_VP_INST_SRC1_MASK					(0x7FFF<<11)
-#define NV20_VP_INST_SRC2H_SHIFT				0
-#define NV20_VP_INST_SRC2H_MASK					(0x7FF << 0)
-
-/* NV20_TCL_PRIMITIVE_3D_0x0B0C */
-#define NV20_VP_INST_SRC2L_SHIFT				28
-#define NV20_VP_INST_SRC2L_MASK					(0x0F  <<28)
-#define NV20_VP_INST_VTEMP_WRITEMASK_SHIFT			24
-#define NV20_VP_INST_VTEMP_WRITEMASK_MASK			(0x0F  <<24)
-#    define NV20_VP_INST_TEMP_WRITEMASK_X	(1<<27)
-#    define NV20_VP_INST_TEMP_WRITEMASK_Y	(1<<26)
-#    define NV20_VP_INST_TEMP_WRITEMASK_Z	(1<<25)
-#    define NV20_VP_INST_TEMP_WRITEMASK_W	(1<<24)
-#define NV20_VP_INST_DEST_TEMP_ID_SHIFT				20
-#define NV20_VP_INST_DEST_TEMP_ID_MASK				(0x0F  <<20)
-#define NV20_VP_INST_STEMP_WRITEMASK_SHIFT			16
-#define NV20_VP_INST_STEMP_WRITEMASK_MASK			(0x0F  <<16)
-#    define NV20_VP_INST_STEMP_WRITEMASK_X	(1<<19)
-#    define NV20_VP_INST_STEMP_WRITEMASK_Y	(1<<18)
-#    define NV20_VP_INST_STEMP_WRITEMASK_Z	(1<<17)
-#    define NV20_VP_INST_STEMP_WRITEMASK_W	(1<<16)
-#define NV20_VP_INST_DEST_WRITEMASK_SHIFT			12
-#define NV20_VP_INST_DEST_WRITEMASK_MASK			(0x0F  <<12)
-#    define NV20_VP_INST_DEST_WRITEMASK_X	(1<<15)
-#    define NV20_VP_INST_DEST_WRITEMASK_Y	(1<<14)
-#    define NV20_VP_INST_DEST_WRITEMASK_Z	(1<<13)
-#    define NV20_VP_INST_DEST_WRITEMASK_W	(1<<12)
-#define NV20_VP_INST_DEST_SHIFT					3
-#define NV20_VP_INST_DEST_MASK					(0xF << 3) /* guess */
-#define NV20_VP_INST_DEST_POS	0
-#define NV20_VP_INST_DEST_COL0	3
-#define NV20_VP_INST_DEST_COL1	4
-#define NV20_VP_INST_DEST_TC(n)	(9+n)
-#define NV20_VP_INST_INDEX_CONST				(1<<1)
-#define NV20_VP_INST3_KNOWN ( \
-	NV20_VP_INST_SRC2L_MASK | \
-	NV20_VP_INST_TEMP_WRITEMASK_MASK | \
-	NV20_VP_INST_DEST_TEMP_ID_MASK | \
-	NV20_VP_INST_STEMP_WRITEMASK_MASK | \
-	NV20_VP_INST_DEST_WRITEMASK_MASK | \
-	NV20_VP_INST_DEST_MASK | \
-	NV20_VP_INST_INDEX_CONST \
-	)
-
-/* Useful to split the source selection regs into their pieces */
-#define NV20_VP_SRC0_HIGH_SHIFT 6
-#define NV20_VP_SRC0_HIGH_MASK  0x00007FC0
-#define NV20_VP_SRC0_LOW_MASK   0x0000003F
-#define NV20_VP_SRC2_HIGH_SHIFT 4
-#define NV20_VP_SRC2_HIGH_MASK  0x00007FF0
-#define NV20_VP_SRC2_LOW_MASK   0x0000000F
-
-#define NV20_VP_SRC_REG_NEGATE					(1<<14)
-#define NV20_VP_SRC_REG_SWZ_X_SHIFT				12
-#define NV20_VP_SRC_REG_SWZ_X_MASK				(0x03  <<12)
-#define NV20_VP_SRC_REG_SWZ_Y_SHIFT				10
-#define NV20_VP_SRC_REG_SWZ_Y_MASK				(0x03  <<10)
-#define NV20_VP_SRC_REG_SWZ_Z_SHIFT				8
-#define NV20_VP_SRC_REG_SWZ_Z_MASK				(0x03  << 8)
-#define NV20_VP_SRC_REG_SWZ_W_SHIFT				6
-#define NV20_VP_SRC_REG_SWZ_W_MASK				(0x03  << 6)
-#define NV20_VP_SRC_REG_SWZ_ALL_SHIFT				6
-#define NV20_VP_SRC_REG_SWZ_ALL_MASK				(0xFF  << 6)
-#define NV20_VP_SRC_REG_TEMP_ID_SHIFT				2
-#define NV20_VP_SRC_REG_TEMP_ID_MASK				(0x0F  << 0)
-#define NV20_VP_SRC_REG_TYPE_SHIFT				0
-#define NV20_VP_SRC_REG_TYPE_MASK				(0x03  << 0)
-#define NV20_VP_SRC_REG_TYPE_TEMP	1
-#define NV20_VP_SRC_REG_TYPE_INPUT	2
-#define NV20_VP_SRC_REG_TYPE_CONST	3	/* guess */
-
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state.c b/src/mesa/drivers/dri/nouveau/nv20_state.c
deleted file mode 100644
index ccf2f6148b4..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv20_state.c
+++ /dev/null
@@ -1,824 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Nouveau
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-static void nv20AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte ubRef;
-	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC, 2);
-	OUT_RING_CACHE(func);
-	OUT_RING_CACHE(ubRef);
-}
-
-static void nv20BlendColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); 
-	GLubyte cf[4];
-
-	CLAMPED_FLOAT_TO_UBYTE(cf[0], color[0]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[1], color[1]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[2], color[2]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[3], color[3]);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_BLEND_COLOR, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888(cf[3], cf[1], cf[2], cf[0]));
-}
-
-static void nv20BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION, 1);
-	OUT_RING_CACHE((modeA<<16) | modeRGB);
-}
-
-
-static void nv20BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
-		GLenum sfactorA, GLenum dfactorA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC, 2);
-	OUT_RING_CACHE((sfactorA<<16) | sfactorRGB);
-	OUT_RING_CACHE((dfactorA<<16) | dfactorRGB);
-}
-
-static void nv20Clear(GLcontext *ctx, GLbitfield mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLuint hw_bufs = 0;
-
-	if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT))
-		hw_bufs |= 0xf0;
-	if (mask & (BUFFER_BIT_DEPTH))
-		hw_bufs |= 0x03;
-
-	if (hw_bufs) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS, 1);
-		OUT_RING_CACHE(hw_bufs);
-	}
-}
-
-static void nv20ClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte c[4];
-	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,color);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888(c[3],c[0],c[1],c[2]));
-}
-
-static void nv20ClearDepth(GLcontext *ctx, GLclampd d)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nmesa->clear_value=((nmesa->clear_value&0x000000FF)|(((uint32_t)(d*0xFFFFFF))<<8));
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH, 1);
-	OUT_RING_CACHE(nmesa->clear_value);
-}
-
-/* we're don't support indexed buffers
-   void (*ClearIndex)(GLcontext *ctx, GLuint index)
- */
-
-static void nv20ClearStencil(GLcontext *ctx, GLint s)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nmesa->clear_value=((nmesa->clear_value&0xFFFFFF00)|(s&0x000000FF));
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH, 1);
-	OUT_RING_CACHE(nmesa->clear_value);
-}
-
-static void nv20ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane), 4);
-	OUT_RING_CACHEf(equation[0]);
-	OUT_RING_CACHEf(equation[1]);
-	OUT_RING_CACHEf(equation[2]);
-	OUT_RING_CACHEf(equation[3]);
-}
-
-static void nv20ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
-		GLboolean bmask, GLboolean amask )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_COLOR_MASK, 1);
-	OUT_RING_CACHE(((amask && 0x01) << 24) | ((rmask && 0x01) << 16) | ((gmask && 0x01)<< 8) | ((bmask && 0x01) << 0));
-}
-
-static void nv20ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	// TODO I need love
-}
-
-static void nv20CullFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CULL_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv20FrontFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_FRONT_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv20DepthFunc(GLcontext *ctx, GLenum func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC, 1);
-	OUT_RING_CACHE(func);
-}
-
-static void nv20DepthMask(GLcontext *ctx, GLboolean flag)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING_CACHE(flag);
-}
-
-static void nv20DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
-	OUT_RING_CACHEf(nearval);
-	OUT_RING_CACHEf(farval);
-}
-
-/** Specify the current buffer for writing */
-//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
-/** Specify the buffers for writing for fragment programs*/
-//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
-
-static void nv20Enable(GLcontext *ctx, GLenum cap, GLboolean state)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	switch(cap)
-	{
-		case GL_ALPHA_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_AUTO_NORMAL:
-		case GL_BLEND:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_CLIP_PLANE0:
-		case GL_CLIP_PLANE1:
-		case GL_CLIP_PLANE2:
-		case GL_CLIP_PLANE3:
-		case GL_CLIP_PLANE4:
-		case GL_CLIP_PLANE5:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(cap-GL_CLIP_PLANE0), 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_COLOR_LOGIC_OP:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_COLOR_MATERIAL:
-//		case GL_COLOR_SUM_EXT:
-//		case GL_COLOR_TABLE:
-//		case GL_CONVOLUTION_1D:
-//		case GL_CONVOLUTION_2D:
-		case GL_CULL_FACE:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DEPTH_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DITHER:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_DITHER_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_FOG:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_FOG_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_HISTOGRAM:
-//		case GL_INDEX_LOGIC_OP:
-		case GL_LIGHT0:
-		case GL_LIGHT1:
-		case GL_LIGHT2:
-		case GL_LIGHT3:
-		case GL_LIGHT4:
-		case GL_LIGHT5:
-		case GL_LIGHT6:
-		case GL_LIGHT7:
-			{
-			uint32_t mask=0x11<<(2*(cap-GL_LIGHT0));
-			nmesa->enabled_lights=((nmesa->enabled_lights&mask)|(mask*state));
-			if (nmesa->lighting_enabled)
-			{
-				BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			}
-			break;
-			}
-		case GL_LIGHTING:
-			nmesa->lighting_enabled=state;
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-			if (nmesa->lighting_enabled)
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			else
-				OUT_RING_CACHE(0x0);
-			break;
-		case GL_LINE_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_LINE_STIPPLE:
-//		case GL_MAP1_COLOR_4:
-//		case GL_MAP1_INDEX:
-//		case GL_MAP1_NORMAL:
-//		case GL_MAP1_TEXTURE_COORD_1:
-//		case GL_MAP1_TEXTURE_COORD_2:
-//		case GL_MAP1_TEXTURE_COORD_3:
-//		case GL_MAP1_TEXTURE_COORD_4:
-//		case GL_MAP1_VERTEX_3:
-//		case GL_MAP1_VERTEX_4:
-//		case GL_MAP2_COLOR_4:
-//		case GL_MAP2_INDEX:
-//		case GL_MAP2_NORMAL:
-//		case GL_MAP2_TEXTURE_COORD_1:
-//		case GL_MAP2_TEXTURE_COORD_2:
-//		case GL_MAP2_TEXTURE_COORD_3:
-//		case GL_MAP2_TEXTURE_COORD_4:
-//		case GL_MAP2_VERTEX_3:
-//		case GL_MAP2_VERTEX_4:
-//		case GL_MINMAX:
-		case GL_NORMALIZE:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POINT_SMOOTH:
-		case GL_POLYGON_OFFSET_POINT:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_LINE:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_FILL:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_STIPPLE:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
-//		case GL_POST_CONVOLUTION_COLOR_TABLE:
-//		case GL_RESCALE_NORMAL:
-		case GL_SCISSOR_TEST:
-			/* No enable bit, nv20Scissor will adjust to max range */
-			ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			      			 ctx->Scissor.Width, ctx->Scissor.Height);
-			break;
-//		case GL_SEPARABLE_2D:
-		case GL_STENCIL_TEST:
-			// TODO BACK and FRONT ?
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_STENCIL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_TEXTURE_GEN_Q:
-//		case GL_TEXTURE_GEN_R:
-//		case GL_TEXTURE_GEN_S:
-//		case GL_TEXTURE_GEN_T:
-//		case GL_TEXTURE_1D:
-//		case GL_TEXTURE_2D:
-//		case GL_TEXTURE_3D:
-	}
-}
-
-static void nv20Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    switch(pname)
-    {
-        case GL_FOG_MODE:
-            BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_FOG_MODE, 1);
-            //OUT_RING_CACHE (params);
-            break;
-            /* TODO: unsure about the rest.*/
-        default:
-            break;
-    }
-
-}
-   
-static void nv20Hint(GLcontext *ctx, GLenum target, GLenum mode)
-{
-	// TODO I need love (fog and line_smooth hints)
-}
-
-// void (*IndexMask)(GLcontext *ctx, GLuint mask);
-
-enum {
-	SPOTLIGHT_NO_UPDATE,
-	SPOTLIGHT_UPDATE_EXPONENT,
-	SPOTLIGHT_UPDATE_DIRECTION,
-	SPOTLIGHT_UPDATE_ALL
-};
-
-static void nv20Lightfv(GLcontext *ctx, GLenum light, GLenum pname, const GLfloat *params )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLint p = light - GL_LIGHT0;
-	struct gl_light *l = &ctx->Light.Light[p];
-	int spotlight_update = SPOTLIGHT_NO_UPDATE;
-
-	/* not sure where the fourth param value goes...*/
-	switch(pname)
-	{
-		case GL_AMBIENT:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_DIFFUSE:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPECULAR:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_POSITION:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPOT_DIRECTION:
-			spotlight_update = SPOTLIGHT_UPDATE_DIRECTION;
-			break;
-		case GL_SPOT_EXPONENT:
-			spotlight_update = SPOTLIGHT_UPDATE_EXPONENT;
-			break;
-		case GL_SPOT_CUTOFF:
-			spotlight_update = SPOTLIGHT_UPDATE_ALL;
-			break;
-		case GL_CONSTANT_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_LINEAR_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_QUADRATIC_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		default:
-			break;
-	}
-
-	switch(spotlight_update) {
-		case SPOTLIGHT_UPDATE_DIRECTION:
-			{
-				GLfloat x,y,z;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(p), 3);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_EXPONENT:
-			{
-				GLfloat cc,lc,qc;
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 3);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_ALL:
-			{
-				GLfloat cc,lc,qc, x,y,z, c;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				c = spot_light_coef_a + 1.0;
-				BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 7);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-				OUT_RING_CACHEf(c);
-			}
-			break;
-		default:
-			break;
-	}
-}
-
-/** Set the lighting model parameters */
-static void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params);
-
-
-static void nv20LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
-{
-/*	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN, 1);
-	OUT_RING_CACHE((pattern << 16) | factor);*/
-}
-
-static void nv20LineWidth(GLcontext *ctx, GLfloat width)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LINE_WIDTH, 1);
-	OUT_RING_CACHEf(width);
-}
-
-static void nv20LogicOpcode(GLcontext *ctx, GLenum opcode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP, 1);
-	OUT_RING_CACHE(opcode);
-}
-
-static void nv20PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/*TODO: not sure what goes here. */
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	
-}
-
-/** Specify the diameter of rasterized points */
-static void nv20PointSize(GLcontext *ctx, GLfloat size)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
-	OUT_RING_CACHEf(size);
-}
-
-/** Select a polygon rasterization mode */
-static void nv20PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT, 1);
-		OUT_RING_CACHE(mode);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK, 1);
-		OUT_RING_CACHE(mode);
-	}
-}
-
-/** Set the scale and units used to calculate depth values */
-static void nv20PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 2);
-        OUT_RING_CACHEf(factor);
-        OUT_RING_CACHEf(units);
-}
-
-/** Set the polygon stippling pattern */
-static void nv20PolygonStipple(GLcontext *ctx, const GLubyte *mask )
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(0), 32);
-        OUT_RING_CACHEp(mask, 32);
-}
-
-/* Specifies the current buffer for reading */
-void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
-/** Set rasterization mode */
-void (*RenderMode)(GLcontext *ctx, GLenum mode );
-
-/** Define the scissor box */
-static void nv20Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	/* There's no scissor enable bit, so adjust the scissor to cover the
-	 * maximum draw buffer bounds
-	 */
-	if (!ctx->Scissor.Enabled) {
-	   x = y = 0;
-	   w = h = 4095;
-	} else {
-	   x += nmesa->drawX;
-	   y += nmesa->drawY;
-	}
-
-        /*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 1);
-        OUT_RING_CACHE((w << 16) | x );
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1, 1);
-        OUT_RING_CACHE((h << 16) | y );*/
-
-}
-
-/** Select flat or smooth shading */
-static void nv20ShadeModel(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SHADE_MODEL, 1);
-	OUT_RING_CACHE(mode);
-}
-
-/** OpenGL 2.0 two-sided StencilFunc */
-static void nv20StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
-		GLint ref, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC, 3);
-	OUT_RING_CACHE(func);
-	OUT_RING_CACHE(ref);
-	OUT_RING_CACHE(mask);
-}
-
-/** OpenGL 2.0 two-sided StencilMask */
-static void nv20StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_STENCIL_MASK, 1);
-	OUT_RING_CACHE(mask);
-}
-
-/** OpenGL 2.0 two-sided StencilOp */
-static void nv20StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
-		GLenum zfail, GLenum zpass)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL, 1);
-	OUT_RING_CACHE(fail);
-	OUT_RING_CACHE(zfail);
-	OUT_RING_CACHE(zpass);
-}
-
-/** Control the generation of texture coordinates */
-void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
-		const GLfloat *params);
-/** Set texture environment parameters */
-void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
-		const GLfloat *param);
-/** Set texture parameters */
-void (*TexParameter)(GLcontext *ctx, GLenum target,
-		struct gl_texture_object *texObj,
-		GLenum pname, const GLfloat *params);
-
-static void nv20TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
-        /*XXX: This SHOULD work.*/
-        OUT_RING_CACHEp(mat->m, 16);
-}
-
-/* Update anything that depends on the window position/size */
-static void nv20WindowMoved(nouveauContextPtr nmesa)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLfloat *v = nmesa->viewport.m;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-	GLuint x = ctx->Viewport.X + nmesa->drawX;
-	GLuint y = ctx->Viewport.Y + nmesa->drawY;
-	int i;
-
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x02b4, 1);
-	OUT_RING(0);
-
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
-        OUT_RING_CACHE((4095 << 16) | 0);
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
-        OUT_RING_CACHE((4095 << 16) | 0);
-	for (i=1; i<8; i++) {
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 1);
-        	OUT_RING_CACHE(0);
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(i), 1);
-	        OUT_RING_CACHE(0);
-	}
-
-	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	      		    ctx->Scissor.Width, ctx->Scissor.Height);
-
-	/* TODO: recalc viewport scale coefs */
-}
-
-/* Initialise any card-specific non-GL related state */
-static GLboolean nv20InitCard(nouveauContextPtr nmesa)
-{
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-
-	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT1, 2);
-	OUT_RING(NvDmaFB);	/* 184 dma_object1 */
-	OUT_RING(NvDmaFB);	/* 188 dma_object2 */
-	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT3, 2);
-	OUT_RING(NvDmaFB);	/* 194 dma_object3 */
-	OUT_RING(NvDmaFB);	/* 198 dma_object4 */
-	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
-	OUT_RING(NvDmaFB);	/* 1a8 dma_object8 */
-
-	BEGIN_RING_SIZE(NvSub3D, 0x17e0, 3);
-	OUT_RINGf(0.0);
-	OUT_RINGf(0.0);
-	OUT_RINGf(1.0);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1e6c, 1);
-	OUT_RING(0x0db6);
-	BEGIN_RING_SIZE(NvSub3D, 0x0290, 1);
-	OUT_RING(0x00100001);
-	BEGIN_RING_SIZE(NvSub3D, 0x09fc, 1);
-	OUT_RING(0);
-	BEGIN_RING_SIZE(NvSub3D, 0x1d80, 1);
-	OUT_RING(1);
-	BEGIN_RING_SIZE(NvSub3D, 0x09f8, 1);
-	OUT_RING(4);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x17ec, 3);
-	OUT_RINGf(0.0);
-	OUT_RINGf(1.0);
-	OUT_RINGf(0.0);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1d88, 1);
-	OUT_RING(3);
-
-	/* FIXME: More dma objects to setup ? */
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1e98, 1);
-	OUT_RING(0);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
-	OUT_RING(0);
-	OUT_RING(1);
-	OUT_RING(2);
-
-	return GL_TRUE;
-}
-
-/* Update buffer offset/pitch/format */
-static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
-{
-	GLuint x, y, w, h;
-	GLuint pitch, format, depth_pitch;
-
-	w = color[0]->mesa.Width;
-	h = color[0]->mesa.Height;
-	x = nmesa->drawX;
-	y = nmesa->drawY;
-
-	if (num_color != 1)
-		return GL_FALSE;
-
-        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 6);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
-	depth_pitch = (depth ? depth->pitch : color[0]->pitch);
-	pitch = (depth_pitch<<16) | color[0]->pitch;
-	format = 0x128;
-	if (color[0]->mesa._ActualFormat != GL_RGBA8) {
-		format = 0x123; /* R5G6B5 color buffer */
-	}
-	OUT_RING_CACHE(format);
-	OUT_RING_CACHE(pitch);
-	OUT_RING_CACHE(color[0]->offset);
-	OUT_RING_CACHE(depth ? depth->offset : color[0]->offset);
-
-	if (depth) {
-		BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH, 2);
-		/* TODO: use a different buffer */
-		OUT_RING(depth->pitch);
-		OUT_RING(depth->offset);
-	}
-
-	/* Always set to bottom left of buffer */
-	/*BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf ((GLfloat) h);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (0.0);*/
-
-	return GL_TRUE;
-}
-
-void nv20InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	func->AlphaFunc			= nv20AlphaFunc;
-	func->BlendColor		= nv20BlendColor;
-	func->BlendEquationSeparate	= nv20BlendEquationSeparate;
-	func->BlendFuncSeparate		= nv20BlendFuncSeparate;
-	func->Clear			= nv20Clear;
-	func->ClearColor		= nv20ClearColor;
-	func->ClearDepth		= nv20ClearDepth;
-	func->ClearStencil		= nv20ClearStencil;
-	func->ClipPlane			= nv20ClipPlane;
-	func->ColorMask			= nv20ColorMask;
-	func->ColorMaterial		= nv20ColorMaterial;
-	func->CullFace			= nv20CullFace;
-	func->FrontFace			= nv20FrontFace;
-	func->DepthFunc			= nv20DepthFunc;
-	func->DepthMask			= nv20DepthMask;
-	func->DepthRange		= nv20DepthRange;
-	func->Enable			= nv20Enable;
-	func->Fogfv			= nv20Fogfv;
-	func->Hint			= nv20Hint;
-	func->Lightfv			= nv20Lightfv;
-/*	func->LightModelfv		= nv20LightModelfv; */
-	func->LineStipple		= nv20LineStipple;
-	func->LineWidth			= nv20LineWidth;
-	func->LogicOpcode		= nv20LogicOpcode;
-	func->PointParameterfv		= nv20PointParameterfv;
-	func->PointSize			= nv20PointSize;
-	func->PolygonMode		= nv20PolygonMode;
-	func->PolygonOffset		= nv20PolygonOffset;
-	func->PolygonStipple		= nv20PolygonStipple;
-/*	func->ReadBuffer		= nv20ReadBuffer;*/
-/*	func->RenderMode		= nv20RenderMode;*/
-	func->Scissor			= nv20Scissor;
-	func->ShadeModel		= nv20ShadeModel;
-	func->StencilFuncSeparate	= nv20StencilFuncSeparate;
-	func->StencilMaskSeparate	= nv20StencilMaskSeparate;
-	func->StencilOpSeparate		= nv20StencilOpSeparate;
-/*	func->TexGen			= nv20TexGen;*/
-/*	func->TexParameter		= nv20TexParameter;*/
-	func->TextureMatrix		= nv20TextureMatrix;
-
-	nmesa->hw_func.InitCard		= nv20InitCard;
-	nmesa->hw_func.BindBuffers	= nv20BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv20WindowMoved;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nv20_vertprog.c b/src/mesa/drivers/dri/nouveau/nv20_vertprog.c
deleted file mode 100644
index 60cfcd70561..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv20_vertprog.c
+++ /dev/null
@@ -1,447 +0,0 @@
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-
-#include "nouveau_shader.h"
-#include "nv20_shader.h"
-
-unsigned int NVVP_TX_VOP_COUNT = 16;
-unsigned int NVVP_TX_NVS_OP_COUNT = 16;
-struct _op_xlat NVVP_TX_VOP[32];
-struct _op_xlat NVVP_TX_SOP[32];
-
-nvsSwzComp NV20VP_TX_SWIZZLE[4] = { NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W };
-
-/*****************************************************************************
- * Support routines
- */
-static void
-NV20VPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   int i;
-
-   /* XXX: missing a way to say what insn we're uploading from, and possible
-    *      the program start position (if NV20 has one) */
-   for (i=0; i<nvs->program_size; i+=4) {
-      BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0, 4);
-      OUT_RING(nvs->program[i + 0]);
-      OUT_RING(nvs->program[i + 1]);
-      OUT_RING(nvs->program[i + 2]);
-      OUT_RING(nvs->program[i + 3]);
-   }
-}
-
-static void
-NV20VPUpdateConst(GLcontext *ctx, nouveauShader *nvs, int id)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   /* Worth checking if the value *actually* changed? Mesa doesn't tell us this
-    * as far as I know..
-    */
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID, 1);
-   OUT_RING (id);
-   BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_X, 4);
-   OUT_RINGf(nvs->params[id].source_val[0]);
-   OUT_RINGf(nvs->params[id].source_val[1]);
-   OUT_RINGf(nvs->params[id].source_val[2]);
-   OUT_RINGf(nvs->params[id].source_val[3]);
-}
-
-/*****************************************************************************
- * Assembly routines
- */
-
-/*****************************************************************************
- * Disassembly routines
- */
-void
-NV20VPTXSwizzle(int hwswz, nvsSwzComp *swz)
-{
-   swz[NVS_SWZ_X] = NV20VP_TX_SWIZZLE[(hwswz & 0xC0) >> 6];
-   swz[NVS_SWZ_Y] = NV20VP_TX_SWIZZLE[(hwswz & 0x30) >> 4];
-   swz[NVS_SWZ_Z] = NV20VP_TX_SWIZZLE[(hwswz & 0x0C) >> 2];
-   swz[NVS_SWZ_W] = NV20VP_TX_SWIZZLE[(hwswz & 0x03) >> 0];
-}
-
-static int
-NV20VPHasMergedInst(nvsFunc * shader)
-{
-   if (shader->GetOpcodeHW(shader, 0) != NV20_VP_INST_OPCODE_NOP &&
-       shader->GetOpcodeHW(shader, 1) != NV20_VP_INST_OPCODE_NOP)
-      printf
-	 ("\n\n*****both opcode fields have values - PLEASE REPORT*****\n");
-   return 0;
-}
-
-static int
-NV20VPIsLastInst(nvsFunc * shader)
-{
-   return ((shader->inst[3] & (1 << 0)) ? 1 : 0);
-}
-
-static int
-NV20VPGetOffsetNext(nvsFunc * shader)
-{
-   return 4;
-}
-
-static struct _op_xlat *
-NV20VPGetOPTXRec(nvsFunc * shader, int merged)
-{
-   struct _op_xlat *opr;
-   int op;
-
-   if (shader->GetOpcodeSlot(shader, merged)) {
-      opr = NVVP_TX_SOP;
-      op = shader->GetOpcodeHW(shader, 1);
-      if (op >= NVVP_TX_NVS_OP_COUNT)
-	 return NULL;
-   }
-   else {
-      opr = NVVP_TX_VOP;
-      op = shader->GetOpcodeHW(shader, 0);
-      if (op >= NVVP_TX_VOP_COUNT)
-	 return NULL;
-   }
-
-   if (opr[op].SOP == NVS_OP_UNKNOWN)
-      return NULL;
-   return &opr[op];
-}
-
-static struct _op_xlat *
-NV20VPGetOPTXFromSOP(nvsOpcode sop, int *id)
-{
-   int i;
-
-   for (i=0;i<NVVP_TX_VOP_COUNT;i++) {
-      if (NVVP_TX_VOP[i].SOP == sop) {
-	 if (id) *id = 0;
-	 return &NVVP_TX_VOP[i];
-      }
-   }
-
-   for (i=0;i<NVVP_TX_NVS_OP_COUNT;i++) {
-      if (NVVP_TX_SOP[i].SOP == sop) {
-	 if (id) *id = 1;
-	 return &NVVP_TX_SOP[i];
-      }
-   }
-
-   return NULL;
-}
-
-static int
-NV20VPGetOpcodeSlot(nvsFunc * shader, int merged)
-{
-   if (shader->HasMergedInst(shader))
-      return merged;
-   if (shader->GetOpcodeHW(shader, 0) == NV20_VP_INST_OPCODE_NOP)
-      return 1;
-   return 0;
-}
-
-static nvsOpcode
-NV20VPGetOpcode(nvsFunc * shader, int merged)
-{
-   struct _op_xlat *opr;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr)
-      return NVS_OP_UNKNOWN;
-
-   return opr->SOP;
-}
-
-static nvsOpcode
-NV20VPGetOpcodeHW(nvsFunc * shader, int slot)
-{
-   if (slot)
-      return (shader->inst[1] & NV20_VP_INST_SCA_OPCODE_MASK)
-	 >> NV20_VP_INST_SCA_OPCODE_SHIFT;
-   return (shader->inst[1] & NV20_VP_INST_VEC_OPCODE_MASK)
-      >> NV20_VP_INST_VEC_OPCODE_SHIFT;
-}
-
-static nvsRegFile
-NV20VPGetDestFile(nvsFunc * shader, int merged)
-{
-   switch (shader->GetOpcode(shader, merged)) {
-   case NVS_OP_ARL:
-      return NVS_FILE_ADDRESS;
-   default:
-      /*FIXME: This probably isn't correct.. */
-      if ((shader->inst[3] & NV20_VP_INST_DEST_WRITEMASK_MASK) == 0)
-	 return NVS_FILE_TEMP;
-      return NVS_FILE_RESULT;
-   }
-}
-
-static unsigned int
-NV20VPGetDestID(nvsFunc * shader, int merged)
-{
-   int id;
-
-   switch (shader->GetDestFile(shader, merged)) {
-   case NVS_FILE_RESULT:
-      id = ((shader->inst[3] & NV20_VP_INST_DEST_MASK)
-	    >> NV20_VP_INST_DEST_SHIFT);
-      switch (id) {
-      case NV20_VP_INST_DEST_POS  : return NVS_FR_POSITION;
-      case NV20_VP_INST_DEST_COL0 : return NVS_FR_COL0;
-      case NV20_VP_INST_DEST_COL1 : return NVS_FR_COL1;
-      case NV20_VP_INST_DEST_TC(0): return NVS_FR_TEXCOORD0;
-      case NV20_VP_INST_DEST_TC(1): return NVS_FR_TEXCOORD1;
-      case NV20_VP_INST_DEST_TC(2): return NVS_FR_TEXCOORD2;
-      case NV20_VP_INST_DEST_TC(3): return NVS_FR_TEXCOORD3;
-      default:
-	 return -1;
-      }
-   case NVS_FILE_ADDRESS:
-      return 0;
-   case NVS_FILE_TEMP:
-      id = ((shader->inst[3] & NV20_VP_INST_DEST_TEMP_ID_MASK)
-	    >> NV20_VP_INST_DEST_TEMP_ID_SHIFT);
-      return id;
-   default:
-      return -1;
-   }
-}
-
-static unsigned int
-NV20VPGetDestMask(nvsFunc * shader, int merged)
-{
-   int hwmask, mask = 0;
-
-   /* Special handling for ARL - hardware only supports a
-    * 1-component address reg
-    */
-   if (shader->GetOpcode(shader, merged) == NVS_OP_ARL)
-      return SMASK_X;
-
-   if (shader->GetDestFile(shader, merged) == NVS_FILE_RESULT)
-      hwmask = (shader->inst[3] & NV20_VP_INST_DEST_WRITEMASK_MASK)
-	 >> NV20_VP_INST_DEST_WRITEMASK_SHIFT;
-   else if (shader->GetOpcodeSlot(shader, merged))
-      hwmask = (shader->inst[3] & NV20_VP_INST_STEMP_WRITEMASK_MASK)
-	 >> NV20_VP_INST_STEMP_WRITEMASK_SHIFT;
-   else
-      hwmask = (shader->inst[3] & NV20_VP_INST_VTEMP_WRITEMASK_MASK)
-	 >> NV20_VP_INST_VTEMP_WRITEMASK_SHIFT;
-
-   if (hwmask & (1 << 3)) mask |= SMASK_X;
-   if (hwmask & (1 << 2)) mask |= SMASK_Y;
-   if (hwmask & (1 << 1)) mask |= SMASK_Z;
-   if (hwmask & (1 << 0)) mask |= SMASK_W;
-
-   return mask;
-}
-
-static unsigned int
-NV20VPGetSourceHW(nvsFunc * shader, int merged, int pos)
-{
-   struct _op_xlat *opr;
-   unsigned int src;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr)
-      return -1;
-
-   switch (opr->srcpos[pos]) {
-   case 0:
-      src = ((shader->inst[1] & NV20_VP_INST_SRC0H_MASK)
-	     >> NV20_VP_INST_SRC0H_SHIFT)
-	 << NV20_VP_SRC0_HIGH_SHIFT;
-      src |= ((shader->inst[2] & NV20_VP_INST_SRC0L_MASK)
-	      >> NV20_VP_INST_SRC0L_SHIFT);
-      break;
-   case 1:
-      src = ((shader->inst[2] & NV20_VP_INST_SRC1_MASK)
-	     >> NV20_VP_INST_SRC1_SHIFT);
-      break;
-   case 2:
-      src = ((shader->inst[2] & NV20_VP_INST_SRC2H_MASK)
-	     >> NV20_VP_INST_SRC2H_SHIFT)
-	 << NV20_VP_SRC2_HIGH_SHIFT;
-      src |= ((shader->inst[3] & NV20_VP_INST_SRC2L_MASK)
-	      >> NV20_VP_INST_SRC2L_SHIFT);
-      break;
-   default:
-      src = -1;
-   }
-
-   return src;
-}
-
-static nvsRegFile
-NV20VPGetSourceFile(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-   struct _op_xlat *opr;
-   int file;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1)
-      return -1;
-
-   switch (opr->srcpos[pos]) {
-   case SPOS_ADDRESS:
-      return NVS_FILE_ADDRESS;
-   default:
-      src = NV20VPGetSourceHW(shader, merged, pos);
-      file = (src & NV20_VP_SRC_REG_TYPE_MASK) >> NV20_VP_SRC_REG_TYPE_SHIFT;
-
-      switch (file) {
-      case NV20_VP_SRC_REG_TYPE_TEMP : return NVS_FILE_TEMP;
-      case NV20_VP_SRC_REG_TYPE_INPUT: return NVS_FILE_ATTRIB;
-      case NV20_VP_SRC_REG_TYPE_CONST: return NVS_FILE_CONST;
-      default:
-	 return NVS_FILE_UNKNOWN;
-      }
-   }
-}
-
-static int
-NV20VPGetSourceID(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_TEMP:
-      src = shader->GetSourceHW(shader, merged, pos);
-      return ((src & NV20_VP_SRC_REG_TEMP_ID_MASK) >>
-	      NV20_VP_SRC_REG_TEMP_ID_SHIFT);
-   case NVS_FILE_CONST:
-      return ((shader->inst[1] & NV20_VP_INST_CONST_SRC_MASK)
-	      >> NV20_VP_INST_CONST_SRC_SHIFT);
-   case NVS_FILE_ATTRIB:
-      src = ((shader->inst[1] & NV20_VP_INST_INPUT_SRC_MASK)
-	     >> NV20_VP_INST_INPUT_SRC_SHIFT);
-      switch (src) {
-      case NV20_VP_INST_INPUT_SRC_POS  : return NVS_FR_POSITION;
-      case NV20_VP_INST_INPUT_SRC_COL0 : return NVS_FR_COL0;
-      case NV20_VP_INST_INPUT_SRC_COL1 : return NVS_FR_COL1;
-      case NV20_VP_INST_INPUT_SRC_TC(0): return NVS_FR_TEXCOORD0;
-      case NV20_VP_INST_INPUT_SRC_TC(1): return NVS_FR_TEXCOORD1;
-      case NV20_VP_INST_INPUT_SRC_TC(2): return NVS_FR_TEXCOORD2;
-      case NV20_VP_INST_INPUT_SRC_TC(3): return NVS_FR_TEXCOORD3;
-      default:
-	 return NVS_FR_UNKNOWN;
-      }
-   default:
-      return -1;
-   }
-}
-
-static int
-NV20VPGetSourceNegate(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-
-   return ((src & NV20_VP_SRC_REG_NEGATE) ? 1 : 0);
-}
-
-static int
-NV20VPGetSourceAbs(nvsFunc * shader, int merged, int pos)
-{
-   /* NV20 can't do ABS on sources?  Appears to be emulated with
-    *      MAX reg, reg, -reg
-    */
-   return 0;
-}
-
-static void
-NV20VPGetSourceSwizzle(nvsFunc * shader, int merged, int pos, nvsSwzComp *swz)
-{
-   unsigned int src;
-   int swzbits;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-   swzbits  =
-      (src & NV20_VP_SRC_REG_SWZ_ALL_MASK) >> NV20_VP_SRC_REG_SWZ_ALL_SHIFT;
-   return NV20VPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV20VPGetSourceIndexed(nvsFunc * shader, int merged, int pos)
-{
-   /* I don't think NV20 can index into attribs, at least no GL
-    * extension is exposed that will allow it.
-    */
-   if (shader->GetSourceFile(shader, merged, pos) != NVS_FILE_CONST)
-      return 0;
-   if (shader->inst[3] & NV20_VP_INST_INDEX_CONST)
-      return 1;
-   return 0;
-}
-
-static int
-NV20VPGetAddressRegID(nvsFunc * shader)
-{
-   /* Only 1 address reg */
-   return 0;
-}
-
-static nvsSwzComp
-NV20VPGetAddressRegSwizzle(nvsFunc * shader)
-{
-   /* Only A0.x available */
-   return NVS_SWZ_X;
-}
-
-void
-NV20VPInitShaderFuncs(nvsFunc * shader)
-{
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_NOP, NVS_OP_NOP, -1, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_MOV, NVS_OP_MOV,  0, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_MUL, NVS_OP_MUL,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_ADD, NVS_OP_ADD,  0,  2, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_MAD, NVS_OP_MAD,  0,  1,  2);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_DP3, NVS_OP_DP3,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_DPH, NVS_OP_DPH,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_DP4, NVS_OP_DP4,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_DST, NVS_OP_DST,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_MIN, NVS_OP_MIN,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_MAX, NVS_OP_MAX,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_SLT, NVS_OP_SLT,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_SGE, NVS_OP_SGE,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV20_VP_INST_OPCODE_ARL, NVS_OP_ARL,  0, -1, -1);
-
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_NOP, NVS_OP_NOP, -1, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_RCP, NVS_OP_RCP,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_RCC, NVS_OP_RCC,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_RSQ, NVS_OP_RSQ,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_EXP, NVS_OP_EXP,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_LOG, NVS_OP_LOG,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV20_VP_INST_OPCODE_LIT, NVS_OP_LIT,  2, -1, -1);
-
-   shader->UploadToHW		= NV20VPUploadToHW;
-   shader->UpdateConst		= NV20VPUpdateConst;
-
-   shader->GetOPTXRec		= NV20VPGetOPTXRec;
-   shader->GetOPTXFromSOP	= NV20VPGetOPTXFromSOP;
-
-   shader->HasMergedInst	= NV20VPHasMergedInst;
-   shader->IsLastInst		= NV20VPIsLastInst;
-   shader->GetOffsetNext	= NV20VPGetOffsetNext;
-   shader->GetOpcodeSlot	= NV20VPGetOpcodeSlot;
-   shader->GetOpcode		= NV20VPGetOpcode;
-   shader->GetOpcodeHW		= NV20VPGetOpcodeHW;
-   shader->GetDestFile		= NV20VPGetDestFile;
-   shader->GetDestID		= NV20VPGetDestID;
-   shader->GetDestMask		= NV20VPGetDestMask;
-   shader->GetSourceHW		= NV20VPGetSourceHW;
-   shader->GetSourceFile	= NV20VPGetSourceFile;
-   shader->GetSourceID		= NV20VPGetSourceID;
-   shader->GetSourceNegate	= NV20VPGetSourceNegate;
-   shader->GetSourceAbs		= NV20VPGetSourceAbs;
-   shader->GetSourceSwizzle	= NV20VPGetSourceSwizzle;
-   shader->GetSourceIndexed	= NV20VPGetSourceIndexed;
-   shader->GetRelAddressRegID	= NV20VPGetAddressRegID;
-   shader->GetRelAddressSwizzle	= NV20VPGetAddressRegSwizzle;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv30_fragprog.c b/src/mesa/drivers/dri/nouveau/nv30_fragprog.c
deleted file mode 100644
index e32452361e8..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv30_fragprog.c
+++ /dev/null
@@ -1,742 +0,0 @@
-#include <stdint.h>
-
-#include "glheader.h"
-#include "macros.h"
-
-#include "nouveau_context.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_drm.h"
-#include "nouveau_shader.h"
-#include "nouveau_object.h"
-#include "nouveau_msg.h"
-#include "nouveau_bufferobj.h"
-#include "nv30_shader.h"
-
-unsigned int NVFP_TX_AOP_COUNT = 64;
-struct _op_xlat NVFP_TX_AOP[64];
-
-/*******************************************************************************
- * Support routines
- */
-
-static void
-NV30FPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nvsCardPriv *priv = &nvs->card_priv;
-   uint32_t offset;
-
-   if (!nvs->program_buffer)
-      nvs->program_buffer = ctx->Driver.NewBufferObject(ctx, 0,
-							GL_ARRAY_BUFFER_ARB);
-
-   /* Should use STATIC_DRAW_ARB if shader doesn't use changable params */
-   nouveau_bo_init_storage(ctx, NOUVEAU_BO_VRAM_OK,
-	 		  nvs->program_size * sizeof(uint32_t),
-			  (const GLvoid *)nvs->program,
-			  GL_DYNAMIC_DRAW_ARB,
-			  nvs->program_buffer);
-
-   offset = nouveau_bo_gpu_ref(ctx, nvs->program_buffer);
-
-   /* Not using state cache here, updated programs at the same address don't
-    * seem to take effect unless the ACTIVE_PROGRAM method is called again.
-    * HW caches the program somewhere?
-    */
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM, 1);
-   OUT_RING       (offset | 1);
-   if (nmesa->screen->card->type == NV_30) {  
-	   BEGIN_RING_SIZE(NvSub3D,
-			   0x1d60 /*NV30_TCL_PRIMITIVE_3D_FP_CONTROL*/, 1);
-	   OUT_RING       ((priv->NV30FP.uses_kil << 7));
-	   BEGIN_RING_SIZE(NvSub3D, 0x1450, 1);
-	   OUT_RING       (priv->NV30FP.num_regs << 16);
-   } else {
-	   BEGIN_RING_SIZE(NvSub3D,
-			   0x1d60 /*NV30_TCL_PRIMITIVE_3D_FP_CONTROL*/, 1);
-	   OUT_RING       ((priv->NV30FP.uses_kil <<  7) |
-			   (priv->NV30FP.num_regs << 24));
-   }
-}
-
-static void
-NV30FPUpdateConst(GLcontext *ctx, nouveauShader *nvs, int id)
-{
-   uint32_t *new     = nvs->params[id].source_val ?
-      (uint32_t*)nvs->params[id].source_val : (uint32_t*)nvs->params[id].val;
-   uint32_t *current;
-   int i;
-
-   for (i=0; i<nvs->params[id].hw_index_cnt; i++) {
-      current = nvs->program + nvs->params[id].hw_index[i];
-      COPY_4V(current, new);
-   }
-   nvs->on_hardware = 0;
-}
-
-/*******************************************************************************
- * Assembly helpers
- */
-static struct _op_xlat *
-NV30FPGetOPTXFromSOP(nvsOpcode op, int *id)
-{
-   int i;
-
-   for (i=0; i<NVFP_TX_AOP_COUNT; i++) {
-      if (NVFP_TX_AOP[i].SOP == op) {
-	 if (id) *id = 0;
-	 return &NVFP_TX_AOP[i];
-      }
-   }
-
-   return NULL;
-}
-
-static int
-NV30FPSupportsOpcode(nvsFunc *shader, nvsOpcode op)
-{
-   if (shader->GetOPTXFromSOP(op, NULL))
-      return 1;
-   return 0;
-}
-
-static void
-NV30FPSetOpcode(nvsFunc *shader, unsigned int opcode, int slot)
-{
-   if (opcode == NV30_FP_OP_OPCODE_KIL)
-      shader->card_priv->NV30FP.uses_kil = GL_TRUE;
-   shader->inst[0] &= ~NV30_FP_OP_OPCODE_MASK;
-   shader->inst[0] |= (opcode << NV30_FP_OP_OPCODE_SHIFT);
-}
-
-static void
-NV30FPSetCCUpdate(nvsFunc *shader)
-{
-   shader->inst[0] |= NV30_FP_OP_COND_WRITE_ENABLE;
-}
-
-static void
-NV30FPSetCondition(nvsFunc *shader, int on, nvsCond cond, int reg,
-      		  nvsSwzComp *swz)
-{
-   nvsSwzComp default_swz[4] = { NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W };
-   unsigned int hwcond;
-
-   /* cond masking is always enabled */
-   if (!on) {
-      cond = NVS_COND_TR;
-      reg  = 0;
-      swz  = default_swz;
-   }
-
-   switch (cond) {
-   case NVS_COND_TR: hwcond = NV30_FP_OP_COND_TR; break;
-   case NVS_COND_FL: hwcond = NV30_FP_OP_COND_FL; break;
-   case NVS_COND_LT: hwcond = NV30_FP_OP_COND_LT; break;
-   case NVS_COND_GT: hwcond = NV30_FP_OP_COND_GT; break;
-   case NVS_COND_LE: hwcond = NV30_FP_OP_COND_LE; break;
-   case NVS_COND_GE: hwcond = NV30_FP_OP_COND_GE; break;
-   case NVS_COND_EQ: hwcond = NV30_FP_OP_COND_EQ; break;
-   case NVS_COND_NE: hwcond = NV30_FP_OP_COND_NE; break;
-   default:
-	WARN_ONCE("unknown fp condmask=%d\n", cond);
-	hwcond = NV30_FP_OP_COND_TR;
-	break;
-   }
-
-   shader->inst[1] &= ~NV30_FP_OP_COND_MASK;
-   shader->inst[1] |= (hwcond << NV30_FP_OP_COND_SHIFT);
-
-   shader->inst[1] &= ~NV30_FP_OP_COND_SWZ_ALL_MASK;
-   shader->inst[1] |= (swz[NVS_SWZ_X] << NV30_FP_OP_COND_SWZ_X_SHIFT);
-   shader->inst[1] |= (swz[NVS_SWZ_Y] << NV30_FP_OP_COND_SWZ_Y_SHIFT);
-   shader->inst[1] |= (swz[NVS_SWZ_Z] << NV30_FP_OP_COND_SWZ_Z_SHIFT);
-   shader->inst[1] |= (swz[NVS_SWZ_W] << NV30_FP_OP_COND_SWZ_W_SHIFT);
-}
-
-static void
-NV30FPSetHighReg(nvsFunc *shader, int id)
-{
-   if (shader->card_priv->NV30FP.num_regs < (id+1)) {
-      if (id == 0)
-	 id = 1; /* necessary? */
-      shader->card_priv->NV30FP.num_regs = (id+1);
-   }
-}
-
-static void
-NV30FPSetResult(nvsFunc *shader, nvsRegister *reg, unsigned int mask, int slot)
-{
-   unsigned int hwreg;
-
-   if (mask & SMASK_X) shader->inst[0] |= NV30_FP_OP_OUT_X;
-   if (mask & SMASK_Y) shader->inst[0] |= NV30_FP_OP_OUT_Y;
-   if (mask & SMASK_Z) shader->inst[0] |= NV30_FP_OP_OUT_Z;
-   if (mask & SMASK_W) shader->inst[0] |= NV30_FP_OP_OUT_W;
-
-   if (reg->file == NVS_FILE_RESULT) {
-      hwreg = 0; /* FIXME: this is only fragment.color */
-      /* This is *not* correct, I have no idea what it is either */
-      shader->inst[0] |= NV30_FP_OP_UNK0_7;
-   } else {
-      shader->inst[0] &= ~NV30_FP_OP_UNK0_7;
-      hwreg = reg->index;
-   }
-   NV30FPSetHighReg(shader, hwreg);
-   shader->inst[0] &= ~NV30_FP_OP_OUT_REG_SHIFT;
-   shader->inst[0] |= (hwreg  << NV30_FP_OP_OUT_REG_SHIFT);
-}
-
-static void
-NV30FPSetSource(nvsFunc *shader, nvsRegister *reg, int pos)
-{
-   unsigned int hwsrc = 0;
-
-   switch (reg->file) {
-   case NVS_FILE_TEMP:
-      hwsrc |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT);
-      hwsrc |= (reg->index << NV30_FP_REG_SRC_SHIFT);
-      NV30FPSetHighReg(shader, reg->index);
-      break;
-   case NVS_FILE_ATTRIB:
-      {
-	 unsigned int hwin;
-
-	 switch (reg->index) {
-	 case NVS_FR_POSITION : hwin = NV30_FP_OP_INPUT_SRC_POSITION; break;
-	 case NVS_FR_COL0     : hwin = NV30_FP_OP_INPUT_SRC_COL0; break;
-	 case NVS_FR_COL1     : hwin = NV30_FP_OP_INPUT_SRC_COL1; break;
-	 case NVS_FR_FOGCOORD : hwin = NV30_FP_OP_INPUT_SRC_FOGC; break;
-	 case NVS_FR_TEXCOORD0: hwin = NV30_FP_OP_INPUT_SRC_TC(0); break;
-	 case NVS_FR_TEXCOORD1: hwin = NV30_FP_OP_INPUT_SRC_TC(1); break;
-	 case NVS_FR_TEXCOORD2: hwin = NV30_FP_OP_INPUT_SRC_TC(2); break;
-	 case NVS_FR_TEXCOORD3: hwin = NV30_FP_OP_INPUT_SRC_TC(3); break;
-	 case NVS_FR_TEXCOORD4: hwin = NV30_FP_OP_INPUT_SRC_TC(4); break;
-	 case NVS_FR_TEXCOORD5: hwin = NV30_FP_OP_INPUT_SRC_TC(5); break;
-	 case NVS_FR_TEXCOORD6: hwin = NV30_FP_OP_INPUT_SRC_TC(6); break;
-	 case NVS_FR_TEXCOORD7: hwin = NV30_FP_OP_INPUT_SRC_TC(7); break;
-	 default:
-		WARN_ONCE("unknown fp input %d\n", reg->index);
-		hwin = NV30_FP_OP_INPUT_SRC_COL0;
-		break;
-	 }
-	 shader->inst[0] &= ~NV30_FP_OP_INPUT_SRC_MASK;
-	 shader->inst[0] |= (hwin << NV30_FP_OP_INPUT_SRC_SHIFT);
-	 hwsrc |= (hwin << NV30_FP_REG_SRC_SHIFT);
-      }
-      hwsrc |= (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT);
-      break;
-   case NVS_FILE_CONST:
-      /* consts are inlined after the inst */
-      hwsrc |= (NV30_FP_REG_TYPE_CONST << NV30_FP_REG_TYPE_SHIFT);
-      break;
-   default:
-      assert(0);
-      break;
-   }
-
-   if (reg->negate)
-      hwsrc |= NV30_FP_REG_NEGATE;
-   if (reg->abs)
-      shader->inst[1] |= (1 << (29+pos));
-   hwsrc |= (reg->swizzle[NVS_SWZ_X] << NV30_FP_REG_SWZ_X_SHIFT);
-   hwsrc |= (reg->swizzle[NVS_SWZ_Y] << NV30_FP_REG_SWZ_Y_SHIFT);
-   hwsrc |= (reg->swizzle[NVS_SWZ_Z] << NV30_FP_REG_SWZ_Z_SHIFT);
-   hwsrc |= (reg->swizzle[NVS_SWZ_W] << NV30_FP_REG_SWZ_W_SHIFT);
-
-   shader->inst[pos+1] &= ~NV30_FP_REG_ALL_MASK;
-   shader->inst[pos+1] |= hwsrc;
-}
-
-static void
-NV30FPSetTexImageUnit(nvsFunc *shader, int unit)
-{
-   shader->inst[0] &= ~NV30_FP_OP_TEX_UNIT_SHIFT;
-   shader->inst[0] |= (unit << NV30_FP_OP_TEX_UNIT_SHIFT);
-}
-
-static void
-NV30FPSetSaturate(nvsFunc *shader)
-{
-   shader->inst[0] |= NV30_FP_OP_OUT_SAT;
-}
-
-static void
-NV30FPInitInstruction(nvsFunc *shader)
-{
-   unsigned int hwsrc;
-
-   shader->inst[0] = 0;
-
-   hwsrc = (NV30_FP_REG_TYPE_INPUT << NV30_FP_REG_TYPE_SHIFT) |
-      	   (NVS_SWZ_X << NV30_FP_REG_SWZ_X_SHIFT) |
-	   (NVS_SWZ_Y << NV30_FP_REG_SWZ_Y_SHIFT) |
-	   (NVS_SWZ_Z << NV30_FP_REG_SWZ_Z_SHIFT) |
-	   (NVS_SWZ_W << NV30_FP_REG_SWZ_W_SHIFT);
-   shader->inst[1] = hwsrc;
-   shader->inst[2] = hwsrc;
-   shader->inst[3] = hwsrc;
-}
-
-static void
-NV30FPSetLastInst(nvsFunc *shader)
-{
-   shader->inst[0] |= 1; 
-}
-
-/*******************************************************************************
- * Disassembly helpers
- */
-static struct _op_xlat *
-NV30FPGetOPTXRec(nvsFunc * shader, int merged)
-{
-   int op;
-
-   op = shader->GetOpcodeHW(shader, 0);
-   if (op > NVFP_TX_AOP_COUNT)
-      return NULL;
-   if (NVFP_TX_AOP[op].SOP == NVS_OP_UNKNOWN)
-      return NULL;
-   return &NVFP_TX_AOP[op];
-}
-
-static int
-NV30FPHasMergedInst(nvsFunc * shader)
-{
-   return 0;
-}
-
-static int
-NV30FPIsLastInst(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV30_FP_OP_PROGRAM_END) ? 1 : 0);
-}
-
-static int
-NV30FPGetOffsetNext(nvsFunc * shader)
-{
-   int i;
-
-   for (i = 0; i < 3; i++)
-      if (shader->GetSourceFile(shader, 0, i) == NVS_FILE_CONST)
-	 return 8;
-   return 4;
-}
-
-static nvsOpcode
-NV30FPGetOpcode(nvsFunc * shader, int merged)
-{
-   struct _op_xlat *opr;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr)
-      return NVS_OP_UNKNOWN;
-
-   return opr->SOP;
-}
-
-static unsigned int
-NV30FPGetOpcodeHW(nvsFunc * shader, int slot)
-{
-   int op;
-
-   op = (shader->inst[0] & NV30_FP_OP_OPCODE_MASK) >> NV30_FP_OP_OPCODE_SHIFT;
-
-   return op;
-}
-
-static nvsRegFile
-NV30FPGetDestFile(nvsFunc * shader, int merged)
-{
-   /* Result regs overlap temporary regs */
-   return NVS_FILE_TEMP;
-}
-
-static unsigned int
-NV30FPGetDestID(nvsFunc * shader, int merged)
-{
-   int id;
-
-   switch (shader->GetDestFile(shader, merged)) {
-   case NVS_FILE_TEMP:
-      id = ((shader->inst[0] & NV30_FP_OP_OUT_REG_MASK)
-	    >> NV30_FP_OP_OUT_REG_SHIFT);
-      return id;
-   default:
-      return -1;
-   }
-}
-
-static unsigned int
-NV30FPGetDestMask(nvsFunc * shader, int merged)
-{
-   unsigned int mask = 0;
-
-   if (shader->inst[0] & NV30_FP_OP_OUT_X) mask |= SMASK_X;
-   if (shader->inst[0] & NV30_FP_OP_OUT_Y) mask |= SMASK_Y;
-   if (shader->inst[0] & NV30_FP_OP_OUT_Z) mask |= SMASK_Z;
-   if (shader->inst[0] & NV30_FP_OP_OUT_W) mask |= SMASK_W;
-
-   return mask;
-}
-
-static unsigned int
-NV30FPGetSourceHW(nvsFunc * shader, int merged, int pos)
-{
-   struct _op_xlat *opr;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1)
-      return -1;
-
-   return shader->inst[opr->srcpos[pos] + 1];
-}
-
-static nvsRegFile
-NV30FPGetSourceFile(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-   struct _op_xlat *opr;
-   int file;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1)
-      return NVS_FILE_UNKNOWN;
-
-   switch (opr->srcpos[pos]) {
-   case SPOS_ADDRESS: return NVS_FILE_ADDRESS;
-   default:
-      src = shader->GetSourceHW(shader, merged, pos);
-      file = (src & NV30_FP_REG_TYPE_MASK) >> NV30_FP_REG_TYPE_SHIFT;
-
-      switch (file) {
-      case NV30_FP_REG_TYPE_TEMP : return NVS_FILE_TEMP;
-      case NV30_FP_REG_TYPE_INPUT: return NVS_FILE_ATTRIB;
-      case NV30_FP_REG_TYPE_CONST: return NVS_FILE_CONST;
-      default:
-	 return NVS_FILE_UNKNOWN;
-      }
-   }
-}
-
-static int
-NV30FPGetSourceID(nvsFunc * shader, int merged, int pos)
-{
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_ATTRIB:
-      switch ((shader->inst[0] & NV30_FP_OP_INPUT_SRC_MASK)
-	      >> NV30_FP_OP_INPUT_SRC_SHIFT) {
-      case NV30_FP_OP_INPUT_SRC_POSITION: return NVS_FR_POSITION;
-      case NV30_FP_OP_INPUT_SRC_COL0    : return NVS_FR_COL0;
-      case NV30_FP_OP_INPUT_SRC_COL1    : return NVS_FR_COL1;
-      case NV30_FP_OP_INPUT_SRC_FOGC    : return NVS_FR_FOGCOORD;
-      case NV30_FP_OP_INPUT_SRC_TC(0)   : return NVS_FR_TEXCOORD0;
-      case NV30_FP_OP_INPUT_SRC_TC(1)   : return NVS_FR_TEXCOORD1;
-      case NV30_FP_OP_INPUT_SRC_TC(2)   : return NVS_FR_TEXCOORD2;
-      case NV30_FP_OP_INPUT_SRC_TC(3)   : return NVS_FR_TEXCOORD3;
-      case NV30_FP_OP_INPUT_SRC_TC(4)   : return NVS_FR_TEXCOORD4;
-      case NV30_FP_OP_INPUT_SRC_TC(5)   : return NVS_FR_TEXCOORD5;
-      case NV30_FP_OP_INPUT_SRC_TC(6)   : return NVS_FR_TEXCOORD6;
-      case NV30_FP_OP_INPUT_SRC_TC(7)   : return NVS_FR_TEXCOORD7;
-      default:
-	 return -1;
-      }
-      break;
-   case NVS_FILE_TEMP:
-      {
-	 unsigned int src;
-
-	 src = shader->GetSourceHW(shader, merged, pos);
-	 return ((src & NV30_FP_REG_SRC_MASK) >> NV30_FP_REG_SRC_SHIFT);
-      }
-   case NVS_FILE_CONST:		/* inlined into fragprog */
-   default:
-      return -1;
-   }
-}
-
-static int
-NV30FPGetTexImageUnit(nvsFunc *shader)
-{
-      return ((shader->inst[0] & NV30_FP_OP_TEX_UNIT_MASK)
-	      >> NV30_FP_OP_TEX_UNIT_SHIFT);
-}
-
-static int
-NV30FPGetSourceNegate(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-
-   if (src == -1)
-      return -1;
-   return ((src & NV30_FP_REG_NEGATE) ? 1 : 0);
-}
-
-static int
-NV30FPGetSourceAbs(nvsFunc * shader, int merged, int pos)
-{
-   struct _op_xlat *opr;
-   static unsigned int abspos[3] = {
-      NV30_FP_OP_OUT_ABS,
-      (1 << 30),		/* guess */
-      (1 << 31)			/* guess */
-   };
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1)
-      return -1;
-
-   return ((shader->inst[1] & abspos[opr->srcpos[pos]]) ? 1 : 0);
-}
-
-nvsSwzComp NV30FP_TX_SWIZZLE[4] = {NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W };
-
-static void
-NV30FPTXSwizzle(int hwswz, nvsSwzComp *swz)
-{
-   swz[NVS_SWZ_W] = NV30FP_TX_SWIZZLE[(hwswz & 0xC0) >> 6];
-   swz[NVS_SWZ_Z] = NV30FP_TX_SWIZZLE[(hwswz & 0x30) >> 4];
-   swz[NVS_SWZ_Y] = NV30FP_TX_SWIZZLE[(hwswz & 0x0C) >> 2];
-   swz[NVS_SWZ_X] = NV30FP_TX_SWIZZLE[(hwswz & 0x03) >> 0];
-}
-
-static void
-NV30FPGetSourceSwizzle(nvsFunc * shader, int merged, int pos, nvsSwzComp *swz)
-{
-   unsigned int src;
-   int swzbits;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-   swzbits = (src & NV30_FP_REG_SWZ_ALL_MASK) >> NV30_FP_REG_SWZ_ALL_SHIFT;
-   NV30FPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV30FPGetSourceIndexed(nvsFunc * shader, int merged, int pos)
-{
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_ATTRIB:
-      return ((shader->inst[3] & NV30_FP_OP_INDEX_INPUT) ? 1 : 0);
-   default:
-      return 0;
-   }
-}
-
-static void
-NV30FPGetSourceConstVal(nvsFunc * shader, int merged, int pos, float *val)
-{
-   val[0] = *(float *) &(shader->inst[4]);
-   val[1] = *(float *) &(shader->inst[5]);
-   val[2] = *(float *) &(shader->inst[6]);
-   val[3] = *(float *) &(shader->inst[7]);
-}
-
-static int
-NV30FPGetSourceScale(nvsFunc * shader, int merged, int pos)
-{
-/*FIXME: is this per-source, only for a specific source, or all sources??*/
-   return (1 << ((shader->inst[2] & NV30_FP_OP_SRC_SCALE_MASK)
-		 >> NV30_FP_OP_SRC_SCALE_SHIFT));
-}
-
-static int
-NV30FPGetAddressRegID(nvsFunc * shader)
-{
-   return 0;
-}
-
-static nvsSwzComp
-NV30FPGetAddressRegSwizzle(nvsFunc * shader)
-{
-   return NVS_SWZ_X;
-}
-
-static int
-NV30FPSupportsConditional(nvsFunc * shader)
-{
-   /*FIXME: Is this true of all ops? */
-   return 1;
-}
-
-static int
-NV30FPGetConditionUpdate(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV30_FP_OP_COND_WRITE_ENABLE) ? 1 : 0);
-}
-
-static int
-NV30FPGetConditionTest(nvsFunc * shader)
-{
-   /*FIXME: always? */
-   return 1;
-}
-
-static nvsCond
-NV30FPGetCondition(nvsFunc * shader)
-{
-   int cond;
-
-   cond = ((shader->inst[1] & NV30_FP_OP_COND_MASK)
-	   >> NV30_FP_OP_COND_SHIFT);
-
-   switch (cond) {
-   case NV30_FP_OP_COND_FL: return NVS_COND_FL;
-   case NV30_FP_OP_COND_LT: return NVS_COND_LT;
-   case NV30_FP_OP_COND_EQ: return NVS_COND_EQ;
-   case NV30_FP_OP_COND_LE: return NVS_COND_LE;
-   case NV30_FP_OP_COND_GT: return NVS_COND_GT;
-   case NV30_FP_OP_COND_NE: return NVS_COND_NE;
-   case NV30_FP_OP_COND_GE: return NVS_COND_GE;
-   case NV30_FP_OP_COND_TR: return NVS_COND_TR;
-   default:
-      return NVS_COND_UNKNOWN;
-   }
-}
-
-static void
-NV30FPGetCondRegSwizzle(nvsFunc * shader, nvsSwzComp *swz)
-{
-   int swzbits;
-
-   swzbits = (shader->inst[1] & NV30_FP_OP_COND_SWZ_ALL_MASK)
-      >> NV30_FP_OP_COND_SWZ_ALL_SHIFT;
-   NV30FPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV30FPGetCondRegID(nvsFunc * shader)
-{
-   return 0;
-}
-
-static nvsPrecision
-NV30FPGetPrecision(nvsFunc * shader)
-{
-   int p;
-
-   p = (shader->inst[0] & NV30_FP_OP_PRECISION_MASK)
-      >> NV30_FP_OP_PRECISION_SHIFT;
-
-   switch (p) {
-   case NV30_FP_PRECISION_FP32: return NVS_PREC_FLOAT32;
-   case NV30_FP_PRECISION_FP16: return NVS_PREC_FLOAT16;
-   case NV30_FP_PRECISION_FX12: return NVS_PREC_FIXED12;
-   default:
-      return NVS_PREC_UNKNOWN;
-   }
-}
-
-static int
-NV30FPGetSaturate(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV30_FP_OP_OUT_SAT) ? 1 : 0);
-}
-
-/*******************************************************************************
- * Init
- */
-void
-NV30FPInitShaderFuncs(nvsFunc * shader)
-{
-   /* These are probably bogus, I made them up... */
-   shader->MaxInst	= 1024;
-   shader->MaxAttrib	= 16;
-   shader->MaxTemp	= 32;
-   shader->MaxAddress	= 1;
-   shader->MaxConst	= 256;
-   shader->caps		= SCAP_SRC_ABS;
-
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_MOV, NVS_OP_MOV, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_MUL, NVS_OP_MUL, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_ADD, NVS_OP_ADD, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_MAD, NVS_OP_MAD, 0, 1, 2);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_DP3, NVS_OP_DP3, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_DP4, NVS_OP_DP4, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_DST, NVS_OP_DST, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_MIN, NVS_OP_MIN, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_MAX, NVS_OP_MAX, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SLT, NVS_OP_SLT, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SGE, NVS_OP_SGE, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_FRC, NVS_OP_FRC, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_FLR, NVS_OP_FLR, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_TEX, NVS_OP_TEX, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_TXD, NVS_OP_TXD, 0, 1, 2);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_TXP, NVS_OP_TXP, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_TXB, NVS_OP_TXB, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SEQ, NVS_OP_SEQ, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SGT, NVS_OP_SGT, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SLE, NVS_OP_SLE, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SNE, NVS_OP_SNE, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_RCP, NVS_OP_RCP, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_LG2, NVS_OP_LG2, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_EX2, NVS_OP_EX2, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_COS, NVS_OP_COS, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_SIN, NVS_OP_SIN, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_NOP, NVS_OP_NOP, -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_DDX, NVS_OP_DDX, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_DDY, NVS_OP_DDY, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_KIL, NVS_OP_KIL, -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_PK4B, NVS_OP_PK4B, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_UP4B, NVS_OP_UP4B, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_PK2H, NVS_OP_PK2H, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_UP2H, NVS_OP_UP2H, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_PK4UB, NVS_OP_PK4UB, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_UP4UB, NVS_OP_UP4UB, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_PK2US, NVS_OP_PK2US, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_UP2US, NVS_OP_UP2US, 0, -1, -1);
-   /*FIXME: Haven't confirmed the source positions for the below opcodes */
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_LIT, NVS_OP_LIT, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_LRP, NVS_OP_LRP, 0, 1, 2);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_POW, NVS_OP_POW, 0, 1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_RSQ, NVS_OP_RSQ, 0, -1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV30_FP_OP_OPCODE_RFL, NVS_OP_RFL, 0, 1, -1);
-
-   shader->GetOPTXRec		= NV30FPGetOPTXRec;
-   shader->GetOPTXFromSOP	= NV30FPGetOPTXFromSOP;
-
-   shader->UploadToHW		= NV30FPUploadToHW;
-   shader->UpdateConst		= NV30FPUpdateConst;
-
-   shader->InitInstruction	= NV30FPInitInstruction;
-   shader->SupportsOpcode	= NV30FPSupportsOpcode;
-   shader->SetOpcode		= NV30FPSetOpcode;
-   shader->SetCCUpdate		= NV30FPSetCCUpdate;
-   shader->SetCondition		= NV30FPSetCondition;
-   shader->SetResult		= NV30FPSetResult;
-   shader->SetSource		= NV30FPSetSource;
-   shader->SetTexImageUnit	= NV30FPSetTexImageUnit;
-   shader->SetSaturate		= NV30FPSetSaturate;
-   shader->SetLastInst		= NV30FPSetLastInst;
-
-   shader->HasMergedInst	= NV30FPHasMergedInst;
-   shader->IsLastInst		= NV30FPIsLastInst;
-   shader->GetOffsetNext	= NV30FPGetOffsetNext;
-   shader->GetOpcode		= NV30FPGetOpcode;
-   shader->GetOpcodeHW		= NV30FPGetOpcodeHW;
-   shader->GetDestFile		= NV30FPGetDestFile;
-   shader->GetDestID		= NV30FPGetDestID;
-   shader->GetDestMask		= NV30FPGetDestMask;
-   shader->GetSourceHW		= NV30FPGetSourceHW;
-   shader->GetSourceFile	= NV30FPGetSourceFile;
-   shader->GetSourceID		= NV30FPGetSourceID;
-   shader->GetTexImageUnit	= NV30FPGetTexImageUnit;
-   shader->GetSourceNegate	= NV30FPGetSourceNegate;
-   shader->GetSourceAbs		= NV30FPGetSourceAbs;
-   shader->GetSourceSwizzle	= NV30FPGetSourceSwizzle;
-   shader->GetSourceIndexed	= NV30FPGetSourceIndexed;
-   shader->GetSourceConstVal	= NV30FPGetSourceConstVal;
-   shader->GetSourceScale	= NV30FPGetSourceScale;
-   shader->GetRelAddressRegID	= NV30FPGetAddressRegID;
-   shader->GetRelAddressSwizzle = NV30FPGetAddressRegSwizzle;
-   shader->GetPrecision		= NV30FPGetPrecision;
-   shader->GetSaturate		= NV30FPGetSaturate;
-   shader->SupportsConditional	= NV30FPSupportsConditional;
-   shader->GetConditionUpdate	= NV30FPGetConditionUpdate;
-   shader->GetConditionTest	= NV30FPGetConditionTest;
-   shader->GetCondition		= NV30FPGetCondition;
-   shader->GetCondRegSwizzle	= NV30FPGetCondRegSwizzle;
-   shader->GetCondRegID		= NV30FPGetCondRegID;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv30_shader.h b/src/mesa/drivers/dri/nouveau/nv30_shader.h
deleted file mode 100644
index 7a027dd4273..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv30_shader.h
+++ /dev/null
@@ -1,379 +0,0 @@
-#ifndef __NV30_SHADER_H__
-#define __NV30_SHADER_H__
-
-/* Vertex programs instruction set
- *
- * 128bit opcodes, split into 4 32-bit ones for ease of use.
- *
- * Non-native instructions
- *	 ABS - MOV + NV40_VP_INST0_DEST_ABS
- *	 POW - EX2 + MUL + LG2
- *	 SUB - ADD, second source negated
- *	 SWZ - MOV
- *	 XPD -  
- *
- * Register access
- *	 - Only one INPUT can be accessed per-instruction (move extras into TEMPs)
- *	 - Only one CONST can be accessed per-instruction (move extras into TEMPs)
- *
- * Relative Addressing
- *	 According to the value returned for MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB
- *	 there are only two address registers available.  The destination in the ARL
- *	 instruction is set to TEMP <n> (The temp isn't actually written).
- *
- *	 When using vanilla ARB_v_p, the proprietary driver will squish both the available
- *	 ADDRESS regs into the first hardware reg in the X and Y components.
- *
- *	 To use an address reg as an index into consts, the CONST_SRC is set to
- *	 (const_base + offset) and INDEX_CONST is set.
- *
- *	 To access the second address reg use ADDR_REG_SELECT_1. A particular component
- *	 of the address regs is selected with ADDR_SWZ.
- *
- *	 Only one address register can be accessed per instruction.
- *
- * Conditional execution (see NV_vertex_program{2,3} for details)
- *	 Conditional execution of an instruction is enabled by setting COND_TEST_ENABLE, and
- *	 selecting the condition which will allow the test to pass with COND_{FL,LT,...}.
- *	 It is possible to swizzle the values in the condition register, which allows for
- *	 testing against an individual component.
- *
- * Branching
- *	 The BRA/CAL instructions seem to follow a slightly different opcode layout.  The
- *	 destination instruction ID (IADDR) overlaps a source field.  Instruction ID's seem to
- *	 be numbered based on the UPLOAD_FROM_ID FIFO command, and is incremented automatically
- *	 on each UPLOAD_INST FIFO command.
- *
- *	 Conditional branching is achieved by using the condition tests described above.
- *	 There doesn't appear to be dedicated looping instructions, but this can be done
- *	 using a temp reg + conditional branching.
- *
- *	 Subroutines may be uploaded before the main program itself, but the first executed
- *	 instruction is determined by the PROGRAM_START_ID FIFO command.
- *
- */
-
-/* DWORD 0 */
-#define NV30_VP_INST_ADDR_REG_SELECT_1				(1 << 24)
-#define NV30_VP_INST_SRC2_ABS			 		(1 << 23) /* guess */
-#define NV30_VP_INST_SRC1_ABS			 		(1 << 22) /* guess */
-#define NV30_VP_INST_SRC0_ABS			 		(1 << 21) /* guess */
-#define NV30_VP_INST_OUT_RESULT					(1 << 20)
-#define NV30_VP_INST_DEST_TEMP_ID_SHIFT				16
-#define NV30_VP_INST_DEST_TEMP_ID_MASK				(0x0F << 16)
-#define NV30_VP_INST_COND_UPDATE_ENABLE				(1<<15)
-#define NV30_VP_INST_COND_TEST_ENABLE				(1<<14)
-#define NV30_VP_INST_COND_SHIFT					11
-#define NV30_VP_INST_COND_MASK					(0x07 << 11)
-#	define NV30_VP_INST_COND_FL	0 /* guess */	
-#	define NV30_VP_INST_COND_LT	1	
-#	define NV30_VP_INST_COND_EQ	2
-#	define NV30_VP_INST_COND_LE	3
-#	define NV30_VP_INST_COND_GT	4
-#	define NV30_VP_INST_COND_NE	5
-#	define NV30_VP_INST_COND_GE	6
-#	define NV30_VP_INST_COND_TR	7 /* guess */
-#define NV30_VP_INST_COND_SWZ_X_SHIFT				9
-#define NV30_VP_INST_COND_SWZ_X_MASK				(0x03 <<  9)
-#define NV30_VP_INST_COND_SWZ_Y_SHIFT				7
-#define NV30_VP_INST_COND_SWZ_Y_MASK				(0x03 <<  7)
-#define NV30_VP_INST_COND_SWZ_Z_SHIFT				5
-#define NV30_VP_INST_COND_SWZ_Z_MASK				(0x03 <<  5)
-#define NV30_VP_INST_COND_SWZ_W_SHIFT				3
-#define NV30_VP_INST_COND_SWZ_W_MASK				(0x03 <<  3)
-#define NV30_VP_INST_COND_SWZ_ALL_SHIFT				3
-#define NV30_VP_INST_COND_SWZ_ALL_MASK				(0xFF <<  3)
-#define NV30_VP_INST_ADDR_SWZ_SHIFT				1
-#define NV30_VP_INST_ADDR_SWZ_MASK				(0x03 <<  1)
-#define NV30_VP_INST_SCA_OPCODEH_SHIFT				0
-#define NV30_VP_INST_SCA_OPCODEH_MASK				(0x01 <<  0)
-
-/* DWORD 1 */
-#define NV30_VP_INST_SCA_OPCODEL_SHIFT				28
-#define NV30_VP_INST_SCA_OPCODEL_MASK				(0x0F << 28)
-#	define NV30_VP_INST_OP_NOP	0x00
-#	define NV30_VP_INST_OP_RCP	0x02
-#	define NV30_VP_INST_OP_RCC	0x03
-#	define NV30_VP_INST_OP_RSQ	0x04
-#	define NV30_VP_INST_OP_EXP	0x05
-#	define NV30_VP_INST_OP_LOG	0x06
-#	define NV30_VP_INST_OP_LIT	0x07
-#	define NV30_VP_INST_OP_BRA	0x09
-#	define NV30_VP_INST_OP_CAL	0x0B
-#	define NV30_VP_INST_OP_RET	0x0C
-#	define NV30_VP_INST_OP_LG2	0x0D
-#	define NV30_VP_INST_OP_EX2	0x0E
-#	define NV30_VP_INST_OP_SIN	0x0F
-#	define NV30_VP_INST_OP_COS	0x10
-#define NV30_VP_INST_VEC_OPCODE_SHIFT				23
-#define NV30_VP_INST_VEC_OPCODE_MASK				(0x1F << 23)
-#	define NV30_VP_INST_OP_NOPV	0x00
-#	define NV30_VP_INST_OP_MOV	0x01
-#	define NV30_VP_INST_OP_MUL	0x02
-#	define NV30_VP_INST_OP_ADD	0x03
-#	define NV30_VP_INST_OP_MAD	0x04
-#	define NV30_VP_INST_OP_DP3	0x05
-#	define NV30_VP_INST_OP_DP4	0x07
-#	define NV30_VP_INST_OP_DPH	0x06
-#	define NV30_VP_INST_OP_DST	0x08
-#	define NV30_VP_INST_OP_MIN	0x09
-#	define NV30_VP_INST_OP_MAX	0x0A
-#	define NV30_VP_INST_OP_SLT	0x0B
-#	define NV30_VP_INST_OP_SGE	0x0C
-#	define NV30_VP_INST_OP_ARL	0x0D
-#	define NV30_VP_INST_OP_FRC	0x0E
-#	define NV30_VP_INST_OP_FLR	0x0F
-#	define NV30_VP_INST_OP_SEQ	0x10
-#	define NV30_VP_INST_OP_SFL	0x11
-#	define NV30_VP_INST_OP_SGT	0x12
-#	define NV30_VP_INST_OP_SLE	0x13
-#	define NV30_VP_INST_OP_SNE	0x14
-#	define NV30_VP_INST_OP_STR	0x15
-#	define NV30_VP_INST_OP_SSG	0x16
-#	define NV30_VP_INST_OP_ARR	0x17
-#	define NV30_VP_INST_OP_ARA	0x18
-#define NV30_VP_INST_CONST_SRC_SHIFT				14
-#define NV30_VP_INST_CONST_SRC_MASK				(0xFF << 14)
-#define NV30_VP_INST_INPUT_SRC_SHIFT				9		/*NV20*/
-#define NV30_VP_INST_INPUT_SRC_MASK				(0x0F <<  9)	/*NV20*/
-#	define NV30_VP_INST_IN_POS	0	  /* These seem to match the bindings specified in */
-#	define NV30_VP_INST_IN_WEIGHT	1	  /* the ARB_v_p spec (2.14.3.1) */
-#	define NV30_VP_INST_IN_NORMAL	2	  
-#	define NV30_VP_INST_IN_COL0	3	  /* Should probably confirm them all though */
-#	define NV30_VP_INST_IN_COL1	4
-#	define NV30_VP_INST_IN_FOGC	5
-#	define NV30_VP_INST_IN_TC0	8
-#	define NV30_VP_INST_IN_TC(n)	(8+n)
-#define NV30_VP_INST_SRC0H_SHIFT				0		/*NV20*/
-#define NV30_VP_INST_SRC0H_MASK					(0x1FF << 0)	/*NV20*/
-
-/* DWORD 2 */
-#define NV30_VP_INST_SRC0L_SHIFT				26		/*NV20*/
-#define NV30_VP_INST_SRC0L_MASK					(0x3F  <<26)	/*NV20*/
-#define NV30_VP_INST_SRC1_SHIFT					11		/*NV20*/
-#define NV30_VP_INST_SRC1_MASK					(0x7FFF<<11)	/*NV20*/
-#define NV30_VP_INST_SRC2H_SHIFT				0		/*NV20*/
-#define NV30_VP_INST_SRC2H_MASK					(0x7FF << 0)	/*NV20*/
-#define NV30_VP_INST_IADDR_SHIFT				2
-#define NV30_VP_INST_IADDR_MASK					(0xFF <<  2) 	/* guess */
-
-/* DWORD 3 */
-#define NV30_VP_INST_SRC2L_SHIFT				28		/*NV20*/
-#define NV30_VP_INST_SRC2L_MASK					(0x0F  <<28)	/*NV20*/
-#define NV30_VP_INST_STEMP_WRITEMASK_SHIFT			24
-#define NV30_VP_INST_STEMP_WRITEMASK_MASK			(0x0F << 24)
-#define NV30_VP_INST_VTEMP_WRITEMASK_SHIFT			20
-#define NV30_VP_INST_VTEMP_WRITEMASK_MASK			(0x0F << 20)
-#define NV30_VP_INST_SDEST_WRITEMASK_SHIFT			16
-#define NV30_VP_INST_SDEST_WRITEMASK_MASK			(0x0F << 16)
-#define NV30_VP_INST_VDEST_WRITEMASK_SHIFT			12		/*NV20*/
-#define NV30_VP_INST_VDEST_WRITEMASK_MASK			(0x0F << 12)	/*NV20*/
-#define NV30_VP_INST_DEST_ID_SHIFT				2
-#define NV30_VP_INST_DEST_ID_MASK				(0x0F <<  2)
-#	define NV30_VP_INST_DEST_POS	0
-#	define NV30_VP_INST_DEST_COL0	3
-#	define NV30_VP_INST_DEST_COL1	4
-#	define NV30_VP_INST_DEST_TC(n)	(8+n)
-
-/* Source-register definition - matches NV20 exactly */
-#define NV30_VP_SRC_REG_NEGATE					(1<<14)
-#define NV30_VP_SRC_REG_SWZ_X_SHIFT				12
-#define NV30_VP_SRC_REG_SWZ_X_MASK				(0x03  <<12)
-#define NV30_VP_SRC_REG_SWZ_Y_SHIFT				10
-#define NV30_VP_SRC_REG_SWZ_Y_MASK				(0x03  <<10)
-#define NV30_VP_SRC_REG_SWZ_Z_SHIFT				8
-#define NV30_VP_SRC_REG_SWZ_Z_MASK				(0x03  << 8)
-#define NV30_VP_SRC_REG_SWZ_W_SHIFT				6
-#define NV30_VP_SRC_REG_SWZ_W_MASK				(0x03  << 6)
-#define NV30_VP_SRC_REG_SWZ_ALL_SHIFT				6
-#define NV30_VP_SRC_REG_SWZ_ALL_MASK				(0xFF  << 6)
-#define NV30_VP_SRC_REG_TEMP_ID_SHIFT				2
-#define NV30_VP_SRC_REG_TEMP_ID_MASK				(0x0F  << 0)
-#define NV30_VP_SRC_REG_TYPE_SHIFT				0
-#define NV30_VP_SRC_REG_TYPE_MASK				(0x03  << 0)
-#define NV30_VP_SRC_REG_TYPE_TEMP	1
-#define NV30_VP_SRC_REG_TYPE_INPUT	2
-#define NV30_VP_SRC_REG_TYPE_CONST	3 /* guess */
-
-/*
- * Each fragment program opcode appears to be comprised of 4 32-bit values.
- *
- *	 0 - Opcode, output reg/mask, ATTRIB source
- *	 1 - Source 0
- *	 2 - Source 1
- *	 3 - Source 2
- *
- * There appears to be no special difference between result regs and temp regs.
- * 		result.color == R0.xyzw
- * 		result.depth == R1.z
- * When the fragprog contains instructions to write depth, NV30_TCL_PRIMITIVE_3D_UNK1D78=0
- * otherwise it is set to 1.
- *
- * Constants are inserted directly after the instruction that uses them.
- * 
- * It appears that it's not possible to use two input registers in one
- * instruction as the input sourcing is done in the instruction dword
- * and not the source selection dwords.  As such instructions such as:
- * 
- *		 ADD result.color, fragment.color, fragment.texcoord[0];
- *
- * must be split into two MOV's and then an ADD (nvidia does this) but
- * I'm not sure why it's not just one MOV and then source the second input
- * in the ADD instruction..
- *
- * Negation of the full source is done with NV30_FP_REG_NEGATE, arbitrary
- * negation requires multiplication with a const.
- *
- * Arbitrary swizzling is supported with the exception of SWIZZLE_ZERO/SWIZZLE_ONE
- * The temp/result regs appear to be initialised to (0.0, 0.0, 0.0, 0.0) as SWIZZLE_ZERO
- * is implemented simply by not writing to the relevant components of the destination.
- *
- * Conditional execution
- *   TODO
- * 
- * Non-native instructions:
- *	 LIT
- *	 LRP - MAD+MAD
- *	 SUB - ADD, negate second source
- *	 RSQ - LG2 + EX2
- *	 POW - LG2 + MUL + EX2
- *	 SCS - COS + SIN
- *	 XPD
- */
-
-//== Opcode / Destination selection ==
-#define NV30_FP_OP_PROGRAM_END					(1 << 0)
-#define NV30_FP_OP_OUT_REG_SHIFT				1
-#define NV30_FP_OP_OUT_REG_MASK					(31 << 1)	/* uncertain */
-/* Needs to be set when writing outputs to get expected result.. */
-#define NV30_FP_OP_UNK0_7					(1 << 7)
-#define NV30_FP_OP_COND_WRITE_ENABLE				(1 << 8)
-#define NV30_FP_OP_OUTMASK_SHIFT				9
-#define NV30_FP_OP_OUTMASK_MASK					(0xF << 9)
-#	define NV30_FP_OP_OUT_X	(1<<9)
-#	define NV30_FP_OP_OUT_Y	(1<<10)
-#	define NV30_FP_OP_OUT_Z	(1<<11)
-#	define NV30_FP_OP_OUT_W	(1<<12)
-/* Uncertain about these, especially the input_src values.. it's possible that
- * they can be dynamically changed.
- */
-#define NV30_FP_OP_INPUT_SRC_SHIFT				13
-#define NV30_FP_OP_INPUT_SRC_MASK				(15 << 13)
-#	define NV30_FP_OP_INPUT_SRC_POSITION	0x0
-#	define NV30_FP_OP_INPUT_SRC_COL0	0x1
-#	define NV30_FP_OP_INPUT_SRC_COL1	0x2
-#	define NV30_FP_OP_INPUT_SRC_FOGC	0x3
-#	define NV30_FP_OP_INPUT_SRC_TC0		0x4
-#	define NV30_FP_OP_INPUT_SRC_TC(n)	(0x4 + n)
-#define NV30_FP_OP_TEX_UNIT_SHIFT				17
-#define NV30_FP_OP_TEX_UNIT_MASK				(0xF << 17) /* guess */
-#define NV30_FP_OP_PRECISION_SHIFT				22
-#define NV30_FP_OP_PRECISION_MASK				(3 << 22)
-#   define NV30_FP_PRECISION_FP32	0
-#   define NV30_FP_PRECISION_FP16	1
-#   define NV30_FP_PRECISION_FX12	2
-#define NV30_FP_OP_OPCODE_SHIFT					24
-#define NV30_FP_OP_OPCODE_MASK					(0x3F << 24)
-#	define NV30_FP_OP_OPCODE_NOP	0x00
-#	define NV30_FP_OP_OPCODE_MOV	0x01
-#	define NV30_FP_OP_OPCODE_MUL	0x02
-#	define NV30_FP_OP_OPCODE_ADD	0x03
-#	define NV30_FP_OP_OPCODE_MAD	0x04
-#	define NV30_FP_OP_OPCODE_DP3	0x05
-#	define NV30_FP_OP_OPCODE_DP4	0x06
-#	define NV30_FP_OP_OPCODE_DST	0x07
-#	define NV30_FP_OP_OPCODE_MIN	0x08
-#	define NV30_FP_OP_OPCODE_MAX	0x09
-#	define NV30_FP_OP_OPCODE_SLT	0x0A
-#	define NV30_FP_OP_OPCODE_SGE	0x0B
-#	define NV30_FP_OP_OPCODE_SLE	0x0C
-#	define NV30_FP_OP_OPCODE_SGT	0x0D
-#	define NV30_FP_OP_OPCODE_SNE	0x0E
-#	define NV30_FP_OP_OPCODE_SEQ	0x0F
-#	define NV30_FP_OP_OPCODE_FRC	0x10
-#	define NV30_FP_OP_OPCODE_FLR	0x11
-#	define NV30_FP_OP_OPCODE_KIL	0x12
-#	define NV30_FP_OP_OPCODE_PK4B   0x13
-#	define NV30_FP_OP_OPCODE_UP4B   0x14
-#	define NV30_FP_OP_OPCODE_DDX	0x15 /* can only write XY */
-#	define NV30_FP_OP_OPCODE_DDY	0x16 /* can only write XY */
-#	define NV30_FP_OP_OPCODE_TEX	0x17
-#	define NV30_FP_OP_OPCODE_TXP	0x18
-#	define NV30_FP_OP_OPCODE_TXD	0x19
-#	define NV30_FP_OP_OPCODE_RCP	0x1A
-#	define NV30_FP_OP_OPCODE_RSQ	0x1B
-#	define NV30_FP_OP_OPCODE_EX2	0x1C
-#	define NV30_FP_OP_OPCODE_LG2	0x1D
-#	define NV30_FP_OP_OPCODE_LIT	0x1E
-#	define NV30_FP_OP_OPCODE_LRP	0x1F
-#	define NV30_FP_OP_OPCODE_COS	0x22
-#	define NV30_FP_OP_OPCODE_SIN	0x23
-#	define NV30_FP_OP_OPCODE_PK2H   0x24
-#	define NV30_FP_OP_OPCODE_UP2H   0x25
-#	define NV30_FP_OP_OPCODE_POW	0x26
-#	define NV30_FP_OP_OPCODE_PK4UB  0x27
-#	define NV30_FP_OP_OPCODE_UP4UB  0x28
-#	define NV30_FP_OP_OPCODE_PK2US  0x29
-#	define NV30_FP_OP_OPCODE_UP2US  0x2A
-#	define NV30_FP_OP_OPCODE_DP2A   0x2E
-#	define NV30_FP_OP_OPCODE_TXB	0x31
-#	define NV30_FP_OP_OPCODE_RFL	0x36
-#define NV30_FP_OP_OUT_SAT					(1 << 31)
-
-/* high order bits of SRC0 */
-#define NV30_FP_OP_OUT_ABS					(1 << 29)
-#define NV30_FP_OP_COND_SWZ_W_SHIFT				27
-#define NV30_FP_OP_COND_SWZ_W_MASK				(3 << 27)
-#define NV30_FP_OP_COND_SWZ_Z_SHIFT				25
-#define NV30_FP_OP_COND_SWZ_Z_MASK				(3 << 25)
-#define NV30_FP_OP_COND_SWZ_Y_SHIFT				23
-#define NV30_FP_OP_COND_SWZ_Y_MASK				(3 << 23)
-#define NV30_FP_OP_COND_SWZ_X_SHIFT				21
-#define NV30_FP_OP_COND_SWZ_X_MASK				(3 << 21)
-#define NV30_FP_OP_COND_SWZ_ALL_SHIFT				21
-#define NV30_FP_OP_COND_SWZ_ALL_MASK				(0xFF << 21)
-#define NV30_FP_OP_COND_SHIFT					18
-#define NV30_FP_OP_COND_MASK					(0x07 << 18)
-#	define NV30_FP_OP_COND_FL	0
-#	define NV30_FP_OP_COND_LT	1
-#	define NV30_FP_OP_COND_EQ	2
-#	define NV30_FP_OP_COND_LE	3
-#	define NV30_FP_OP_COND_GT	4
-#	define NV30_FP_OP_COND_NE	5
-#	define NV30_FP_OP_COND_GE	6
-#	define NV30_FP_OP_COND_TR	7
-
-/* high order bits of SRC1 */
-#define NV30_FP_OP_SRC_SCALE_SHIFT				28
-#define NV30_FP_OP_SRC_SCALE_MASK				(3 << 28)
-
-/* high order bits of SRC2 */
-#define NV30_FP_OP_INDEX_INPUT					(1 << 30)
-
-//== Register selection ==
-#define NV30_FP_REG_ALL_MASK					(0x1FFFF<<0)
-#define NV30_FP_REG_TYPE_SHIFT					0
-#define NV30_FP_REG_TYPE_MASK					(3 << 0)
-#	define NV30_FP_REG_TYPE_TEMP	0
-#	define NV30_FP_REG_TYPE_INPUT	1
-#	define NV30_FP_REG_TYPE_CONST	2
-#define NV30_FP_REG_SRC_SHIFT					2 /* uncertain */
-#define NV30_FP_REG_SRC_MASK					(31 << 2)
-#define NV30_FP_REG_UNK_0					(1 << 8)
-#define NV30_FP_REG_SWZ_ALL_SHIFT				9
-#define NV30_FP_REG_SWZ_ALL_MASK				(255 << 9)
-#define NV30_FP_REG_SWZ_X_SHIFT					9
-#define NV30_FP_REG_SWZ_X_MASK					(3 << 9)
-#define NV30_FP_REG_SWZ_Y_SHIFT					11
-#define NV30_FP_REG_SWZ_Y_MASK					(3 << 11)
-#define NV30_FP_REG_SWZ_Z_SHIFT					13
-#define NV30_FP_REG_SWZ_Z_MASK					(3 << 13)
-#define NV30_FP_REG_SWZ_W_SHIFT					15
-#define NV30_FP_REG_SWZ_W_MASK					(3 << 15)
-#	define NV30_FP_SWIZZLE_X	0
-#	define NV30_FP_SWIZZLE_Y	1
-#	define NV30_FP_SWIZZLE_Z	2
-#	define NV30_FP_SWIZZLE_W	3
-#define NV30_FP_REG_NEGATE					(1 << 17)
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nv30_state.c b/src/mesa/drivers/dri/nouveau/nv30_state.c
deleted file mode 100644
index 9b010954b33..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv30_state.c
+++ /dev/null
@@ -1,1002 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Nouveau
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_state.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-#define NOUVEAU_CARD_USING_SHADERS (nmesa->screen->card->type >= NV_40)
-
-static void nv30AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte ubRef;
-	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
-
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC, 2);
-	OUT_RING_CACHE(func);     /* NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC */
-	OUT_RING_CACHE(ubRef);    /* NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF  */
-}
-
-static void nv30BlendColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); 
-	GLubyte cf[4];
-
-	CLAMPED_FLOAT_TO_UBYTE(cf[0], color[0]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[1], color[1]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[2], color[2]);
-	CLAMPED_FLOAT_TO_UBYTE(cf[3], color[3]);
-
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_BLEND_COLOR, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888(cf[3], cf[1], cf[2], cf[0]));
-}
-
-static void nv30BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_BLEND_EQUATION, 1);
-	OUT_RING_CACHE((modeA<<16) | modeRGB);
-}
-
-
-static void nv30BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
-		GLenum sfactorA, GLenum dfactorA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC, 2);
-	OUT_RING_CACHE((sfactorA<<16) | sfactorRGB);
-	OUT_RING_CACHE((dfactorA<<16) | dfactorRGB);
-}
-
-static void nv30Clear(GLcontext *ctx, GLbitfield mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLuint hw_bufs = 0;
-
-	if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT))
-		hw_bufs |= 0xf0;
-	if (mask & (BUFFER_BIT_DEPTH))
-		hw_bufs |= 0x03;
-
-	if (hw_bufs) {
-		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS, 1);
-		OUT_RING(hw_bufs);
-	}
-}
-
-static void nv30ClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte c[4];
-	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,color);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB, 1);
-	OUT_RING_CACHE(PACK_COLOR_8888(c[3],c[0],c[1],c[2]));
-}
-
-static void nv30ClearDepth(GLcontext *ctx, GLclampd d)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nmesa->clear_value=((nmesa->clear_value&0x000000FF)|(((uint32_t)(d*0xFFFFFF))<<8));
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH, 1);
-	OUT_RING_CACHE(nmesa->clear_value);
-}
-
-/* we're don't support indexed buffers
-   void (*ClearIndex)(GLcontext *ctx, GLuint index)
- */
-
-static void nv30ClearStencil(GLcontext *ctx, GLint s)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nmesa->clear_value=((nmesa->clear_value&0xFFFFFF00)|(s&0x000000FF));
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH, 1);
-	OUT_RING_CACHE(nmesa->clear_value);
-}
-
-static void nv30ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (NOUVEAU_CARD_USING_SHADERS)
-		return;
-
-	plane -= GL_CLIP_PLANE0;
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane), 4);
-	OUT_RING_CACHEf(equation[0]);
-	OUT_RING_CACHEf(equation[1]);
-	OUT_RING_CACHEf(equation[2]);
-	OUT_RING_CACHEf(equation[3]);
-}
-
-static void nv30ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
-		GLboolean bmask, GLboolean amask )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_COLOR_MASK, 1);
-	OUT_RING_CACHE(((amask && 0x01) << 24) | ((rmask && 0x01) << 16) | ((gmask && 0x01)<< 8) | ((bmask && 0x01) << 0));
-}
-
-static void nv30ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	// TODO I need love
-}
-
-static void nv30CullFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CULL_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv30FrontFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FRONT_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv30DepthFunc(GLcontext *ctx, GLenum func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_FUNC, 1);
-	OUT_RING_CACHE(func);
-}
-
-static void nv30DepthMask(GLcontext *ctx, GLboolean flag)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING_CACHE(flag);
-}
-
-static void nv30DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
-	OUT_RING_CACHEf(nearval);
-	OUT_RING_CACHEf(farval);
-}
-
-/** Specify the current buffer for writing */
-//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
-/** Specify the buffers for writing for fragment programs*/
-//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
-
-static void nv30Enable(GLcontext *ctx, GLenum cap, GLboolean state)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	switch(cap)
-	{
-		case GL_ALPHA_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_AUTO_NORMAL:
-		case GL_BLEND:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_CLIP_PLANE0:
-		case GL_CLIP_PLANE1:
-		case GL_CLIP_PLANE2:
-		case GL_CLIP_PLANE3:
-		case GL_CLIP_PLANE4:
-		case GL_CLIP_PLANE5:
-			if (NOUVEAU_CARD_USING_SHADERS) {
-				nouveauShader *nvs = (nouveauShader *)ctx->VertexProgram._Current;
-				if (nvs)
-					nvs->translated = GL_FALSE;
-			} else {
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(cap-GL_CLIP_PLANE0), 1);
-				OUT_RING_CACHE(state);
-			}
-			break;
-		case GL_COLOR_LOGIC_OP:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_COLOR_MATERIAL:
-//		case GL_COLOR_SUM_EXT:
-//		case GL_COLOR_TABLE:
-//		case GL_CONVOLUTION_1D:
-//		case GL_CONVOLUTION_2D:
-		case GL_CULL_FACE:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DEPTH_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DITHER:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DITHER_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_FOG:
-			if (NOUVEAU_CARD_USING_SHADERS)
-				break;
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_HISTOGRAM:
-//		case GL_INDEX_LOGIC_OP:
-		case GL_LIGHT0:
-		case GL_LIGHT1:
-		case GL_LIGHT2:
-		case GL_LIGHT3:
-		case GL_LIGHT4:
-		case GL_LIGHT5:
-		case GL_LIGHT6:
-		case GL_LIGHT7:
-			{
-			uint32_t mask=0x11<<(2*(cap-GL_LIGHT0));
-
-			if (NOUVEAU_CARD_USING_SHADERS)
-				break;
-
-			nmesa->enabled_lights=((nmesa->enabled_lights&mask)|(mask*state));
-			if (nmesa->lighting_enabled)
-			{
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			}
-			break;
-			}
-		case GL_LIGHTING:
-			if (NOUVEAU_CARD_USING_SHADERS)
-				break;
-
-			nmesa->lighting_enabled=state;
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
-			if (nmesa->lighting_enabled)
-				OUT_RING_CACHE(nmesa->enabled_lights);
-			else
-				OUT_RING_CACHE(0x0);
-			break;
-//		case GL_LINE_SMOOTH:
-//		case GL_LINE_STIPPLE:
-//		case GL_MAP1_COLOR_4:
-//		case GL_MAP1_INDEX:
-//		case GL_MAP1_NORMAL:
-//		case GL_MAP1_TEXTURE_COORD_1:
-//		case GL_MAP1_TEXTURE_COORD_2:
-//		case GL_MAP1_TEXTURE_COORD_3:
-//		case GL_MAP1_TEXTURE_COORD_4:
-//		case GL_MAP1_VERTEX_3:
-//		case GL_MAP1_VERTEX_4:
-//		case GL_MAP2_COLOR_4:
-//		case GL_MAP2_INDEX:
-//		case GL_MAP2_NORMAL:
-//		case GL_MAP2_TEXTURE_COORD_1:
-//		case GL_MAP2_TEXTURE_COORD_2:
-//		case GL_MAP2_TEXTURE_COORD_3:
-//		case GL_MAP2_TEXTURE_COORD_4:
-//		case GL_MAP2_VERTEX_3:
-//		case GL_MAP2_VERTEX_4:
-//		case GL_MINMAX:
-		case GL_NORMALIZE:
-			if (nmesa->screen->card->type != NV_44) {
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE, 1);
-				OUT_RING_CACHE(state);
-			}
-			break;
-//		case GL_POINT_SMOOTH:
-		case GL_POLYGON_OFFSET_POINT:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_LINE:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_FILL:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_STIPPLE:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
-//		case GL_POST_CONVOLUTION_COLOR_TABLE:
-//		case GL_RESCALE_NORMAL:
-		case GL_SCISSOR_TEST:
-			/* No enable bit, nv30Scissor will adjust to max range */
-			ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			      			 ctx->Scissor.Width, ctx->Scissor.Height);
-			break;
-//		case GL_SEPARABLE_2D:
-		case GL_STENCIL_TEST:
-			// TODO BACK and FRONT ?
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_TEXTURE_GEN_Q:
-//		case GL_TEXTURE_GEN_R:
-//		case GL_TEXTURE_GEN_S:
-//		case GL_TEXTURE_GEN_T:
-//		case GL_TEXTURE_1D:
-//		case GL_TEXTURE_2D:
-//		case GL_TEXTURE_3D:
-	}
-}
-
-static void nv30Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-    if (NOUVEAU_CARD_USING_SHADERS)
-        return;
-
-    switch(pname)
-    {
-    case GL_FOG_MODE:
-    {
-        int mode = 0;
-        /* The modes are different in GL and the card.  */
-        switch(ctx->Fog.Mode)
-        {
-        case GL_LINEAR:
-            mode = 0x804;
-            break;
-        case GL_EXP:
-            mode = 0x802;
-            break;
-        case GL_EXP2:
-            mode = 0x803;
-            break;
-        }
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_MODE, 1);
-	OUT_RING_CACHE (mode);
-	break;
-    }
-    case GL_FOG_COLOR:
-    {
-	GLubyte c[4];
-	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,params);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_COLOR, 1);
-        /* nvidia ignores the alpha channel */
-	OUT_RING_CACHE(PACK_COLOR_8888_REV(c[0],c[1],c[2],c[3]));
-        break;
-    }
-    case GL_FOG_DENSITY:
-    case GL_FOG_START:
-    case GL_FOG_END:
-    {
-        GLfloat f=0., c=0.;
-        switch(ctx->Fog.Mode)
-        {
-        case GL_LINEAR:
-            f = -1.0/(ctx->Fog.End - ctx->Fog.Start);
-            c = ctx->Fog.Start/(ctx->Fog.End - ctx->Fog.Start) + 2.001953;
-            break;
-        case GL_EXP:
-            f = -0.090168*ctx->Fog.Density;
-            c = 1.5;
-        case GL_EXP2:
-            f = -0.212330*ctx->Fog.Density;
-            c = 1.5;
-        }
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR, 1);
-        OUT_RING_CACHE(f);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT, 1);
-        OUT_RING_CACHE(c);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC, 1);
-        OUT_RING_CACHE(0); /* Is this always the same? */
-        break;
-    }
-//    case GL_FOG_COORD_SRC:
-    default:
-        break;
-    }
-}
-   
-static void nv30Hint(GLcontext *ctx, GLenum target, GLenum mode)
-{
-	// TODO I need love (fog and line_smooth hints)
-}
-
-// void (*IndexMask)(GLcontext *ctx, GLuint mask);
-
-enum {
-	SPOTLIGHT_NO_UPDATE,
-	SPOTLIGHT_UPDATE_EXPONENT,
-	SPOTLIGHT_UPDATE_DIRECTION,
-	SPOTLIGHT_UPDATE_ALL
-};
-
-static void nv30Lightfv(GLcontext *ctx, GLenum light, GLenum pname, const GLfloat *params )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLint p = light - GL_LIGHT0;
-	struct gl_light *l = &ctx->Light.Light[p];
-	int spotlight_update = SPOTLIGHT_NO_UPDATE;
-
-	if (NOUVEAU_CARD_USING_SHADERS)
-	   return;
-
-	/* not sure where the fourth param value goes...*/
-	switch(pname)
-	{
-		case GL_AMBIENT:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_DIFFUSE:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPECULAR:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_POSITION:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(p), 3);
-			OUT_RING_CACHEf(params[0]);
-			OUT_RING_CACHEf(params[1]);
-			OUT_RING_CACHEf(params[2]);
-			break;
-		case GL_SPOT_DIRECTION:
-			spotlight_update = SPOTLIGHT_UPDATE_DIRECTION;
-			break;
-		case GL_SPOT_EXPONENT:
-			spotlight_update = SPOTLIGHT_UPDATE_EXPONENT;
-			break;
-		case GL_SPOT_CUTOFF:
-			spotlight_update = SPOTLIGHT_UPDATE_ALL;
-			break;
-		case GL_CONSTANT_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_LINEAR_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		case GL_QUADRATIC_ATTENUATION:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(p), 1);
-			OUT_RING_CACHEf(*params);
-			break;
-		default:
-			break;
-	}
-
-	switch(spotlight_update) {
-		case SPOTLIGHT_UPDATE_DIRECTION:
-			{
-				GLfloat x,y,z;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(p), 3);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_EXPONENT:
-			{
-				GLfloat cc,lc,qc;
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 3);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-			}
-			break;
-		case SPOTLIGHT_UPDATE_ALL:
-			{
-				GLfloat cc,lc,qc, x,y,z, c;
-				GLfloat spot_light_coef_a = 1.0 / (l->_CosCutoff - 1.0);
-				cc = 1.0;	/* FIXME: These need to be correctly computed */
-				lc = 0.0;
-				qc = 2.0;
-				x = spot_light_coef_a * l->_NormDirection[0];
-				y = spot_light_coef_a * l->_NormDirection[1];
-				z = spot_light_coef_a * l->_NormDirection[2];
-				c = spot_light_coef_a + 1.0;
-				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(p), 7);
-				OUT_RING_CACHEf(cc);
-				OUT_RING_CACHEf(lc);
-				OUT_RING_CACHEf(qc);
-				OUT_RING_CACHEf(x);
-				OUT_RING_CACHEf(y);
-				OUT_RING_CACHEf(z);
-				OUT_RING_CACHEf(c);
-			}
-			break;
-		default:
-			break;
-	}
-}
-
-/** Set the lighting model parameters */
-void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params);
-
-
-static void nv30LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN, 1);
-	OUT_RING_CACHE((pattern << 16) | factor);
-}
-
-static void nv30LineWidth(GLcontext *ctx, GLfloat width)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte ubWidth;
-
-	ubWidth = (GLubyte)(width * 8.0) & 0xFF;
-
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LINE_WIDTH_SMOOTH, 1);
-	OUT_RING_CACHE(ubWidth);
-}
-
-static void nv30LogicOpcode(GLcontext *ctx, GLenum opcode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP, 1);
-	OUT_RING_CACHE(opcode);
-}
-
-static void nv30PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/*TODO: not sure what goes here. */
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	
-}
-
-/** Specify the diameter of rasterized points */
-static void nv30PointSize(GLcontext *ctx, GLfloat size)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
-	OUT_RING_CACHEf(size);
-}
-
-/** Select a polygon rasterization mode */
-static void nv30PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT, 1);
-		OUT_RING_CACHE(mode);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK, 1);
-		OUT_RING_CACHE(mode);
-	}
-}
-
-/** Set the scale and units used to calculate depth values */
-static void nv30PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 2);
-        OUT_RING_CACHEf(factor);
-
-        /* Looks like we always multiply units by 2.0... according to the dumps.*/
-        OUT_RING_CACHEf(units * 2.0);
-}
-
-/** Set the polygon stippling pattern */
-static void nv30PolygonStipple(GLcontext *ctx, const GLubyte *mask )
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(0), 32);
-        OUT_RING_CACHEp(mask, 32);
-}
-
-/* Specifies the current buffer for reading */
-void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
-/** Set rasterization mode */
-void (*RenderMode)(GLcontext *ctx, GLenum mode );
-
-/* Translate GL coords to window coords, clamping w/h to the
- * dimensions of the window.
- */
-static void nv30WindowCoords(nouveauContextPtr nmesa,
-			     GLuint x, GLuint y, GLuint w, GLuint h,
-			     GLuint *wX, GLuint *wY, GLuint *wW, GLuint *wH)
-{
-	if ((x+w) > nmesa->drawW)
-		w = nmesa->drawW - x;
-	(*wX) = x + nmesa->drawX;
-	(*wW) = w;
-
-	if ((y+h) > nmesa->drawH)
-		h = nmesa->drawH - y;
-	(*wY) = (nmesa->drawH - y) - h + nmesa->drawY;
-	(*wH) = h;
-}
-
-/** Define the scissor box */
-static void nv30Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLuint wX, wY, wW, wH;
-
-	/* There's no scissor enable bit, so adjust the scissor to cover the
-	 * maximum draw buffer bounds
-	 */
-	if (!ctx->Scissor.Enabled) {
-	   wX = nmesa->drawX;
-	   wY = nmesa->drawY;
-	   wW = nmesa->drawW;
-	   wH = nmesa->drawH;
-	} else {
-	   nv30WindowCoords(nmesa, x, y, w, h, &wX, &wY, &wW, &wH);
-	}
-
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS, 2);
-        OUT_RING_CACHE  ((wW << 16) | wX);
-        OUT_RING_CACHE  ((wH << 16) | wY);
-}
-
-/** Select flat or smooth shading */
-static void nv30ShadeModel(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SHADE_MODEL, 1);
-	OUT_RING_CACHE(mode);
-}
-
-/** OpenGL 2.0 two-sided StencilFunc */
-static void nv30StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
-		GLint ref, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC, 3);
-		OUT_RING_CACHE(func);
-		OUT_RING_CACHE(ref);
-		OUT_RING_CACHE(mask);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC, 3);
-		OUT_RING_CACHE(func);
-		OUT_RING_CACHE(ref);
-		OUT_RING_CACHE(mask);
-	}
-}
-
-/** OpenGL 2.0 two-sided StencilMask */
-static void nv30StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-}
-
-/** OpenGL 2.0 two-sided StencilOp */
-static void nv30StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
-		GLenum zfail, GLenum zpass)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL, 3);
-		OUT_RING_CACHE(fail);
-		OUT_RING_CACHE(zfail);
-		OUT_RING_CACHE(zpass);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL, 3);
-		OUT_RING_CACHE(fail);
-		OUT_RING_CACHE(zfail);
-		OUT_RING_CACHE(zpass);
-	}
-}
-
-/** Control the generation of texture coordinates */
-void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
-		const GLfloat *params);
-/** Set texture environment parameters */
-void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
-		const GLfloat *param);
-/** Set texture parameters */
-void (*TexParameter)(GLcontext *ctx, GLenum target,
-		struct gl_texture_object *texObj,
-		GLenum pname, const GLfloat *params);
-
-static void nv30TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (!NOUVEAU_CARD_USING_SHADERS) {
-		BEGIN_RING_CACHE(NvSub3D,
-				 NV30_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
-		/*XXX: This SHOULD work.*/
-		OUT_RING_CACHEp(mat->m, 16);
-	}
-}
-
-static void nv30WindowMoved(nouveauContextPtr nmesa)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLfloat *v = nmesa->viewport.m;
-	GLuint wX, wY, wW, wH;
-
-	nv30WindowCoords(nmesa, ctx->Viewport.X, ctx->Viewport.Y, 
-				ctx->Viewport.Width, ctx->Viewport.Height,
-				&wX, &wY, &wW, &wH);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0, 2);
-        OUT_RING_CACHE  ((wW << 16) | wX);
-        OUT_RING_CACHE  ((wH << 16) | wY);
-
-	/* something to do with clears, possibly doesn't belong here */
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0, 2);
-        OUT_RING_CACHE(((nmesa->drawX + nmesa->drawW) << 16) | nmesa->drawX);
-        OUT_RING_CACHE(((nmesa->drawY + nmesa->drawH) << 16) | nmesa->drawY);
-
-	/* viewport transform */
-	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_OX, 8);
-	OUT_RING_CACHEf (v[MAT_TX]);
-	OUT_RING_CACHEf (v[MAT_TY]);
-	OUT_RING_CACHEf (v[MAT_TZ]);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (v[MAT_SX]);
-	OUT_RING_CACHEf (v[MAT_SY]);
-	OUT_RING_CACHEf (v[MAT_SZ]);
-	OUT_RING_CACHEf (0.0);
-
-	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	      		    ctx->Scissor.Width, ctx->Scissor.Height);
-}
-
-static GLboolean nv30InitCard(nouveauContextPtr nmesa)
-{
-	int i;
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT1, 3);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaTT);
-        OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
-	OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT4, 2);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-        BEGIN_RING_SIZE(NvSub3D, 0x1b0, 1); /* SET_OBJECT8B*/
-        OUT_RING(NvDmaFB);
-
-        for(i = 0x2c8; i <= 0x2fc; i += 4)
-        {
-            BEGIN_RING_SIZE(NvSub3D, i, 1);
-            OUT_RING(0x0);
-        }
-
-	BEGIN_RING_SIZE(NvSub3D, 0x0220, 1);
-	OUT_RING(1);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x03b0, 1);
-	OUT_RING(0x00100000);
-	BEGIN_RING_SIZE(NvSub3D, 0x1454, 1);
-	OUT_RING(0);
-	BEGIN_RING_SIZE(NvSub3D, 0x1d80, 1);
-	OUT_RING(3);
-	
-	/* NEW */
-       	BEGIN_RING_SIZE(NvSub3D, 0x1e98, 1);
-        OUT_RING(0);
-        BEGIN_RING_SIZE(NvSub3D, 0x17e0, 3);
-        OUT_RING(0);
-        OUT_RING(0);
-        OUT_RING(0x3f800000);
-        BEGIN_RING_SIZE(NvSub3D, 0x1f80, 16);
-        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
-        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
-        OUT_RING(0x0000ffff);
-        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
-        OUT_RING(0); OUT_RING(0); OUT_RING(0); 
-/*
-        BEGIN_RING_SIZE(NvSub3D, 0x100, 2);
-        OUT_RING(0);
-        OUT_RING(0);
-*/
-        BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
-        OUT_RING(0);
-        OUT_RING(1);
-        OUT_RING(2);
-
-        BEGIN_RING_SIZE(NvSub3D, 0x1d88, 1);
-        OUT_RING(0x00001200);
-
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_RC_ENABLE, 1);
-	OUT_RING       (0);
-
-	return GL_TRUE;
-}
-
-static GLboolean nv40InitCard(nouveauContextPtr nmesa)
-{
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT1, 2);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
-	OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT4, 2);
-	OUT_RING(NvDmaFB);
-	OUT_RING(NvDmaFB);
-	BEGIN_RING_SIZE(NvSub3D, 0x0220, 1);
-	OUT_RING(1);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1ea4, 3);
-	OUT_RING(0x00000010);
-	OUT_RING(0x01000100);
-	OUT_RING(0xff800006);
-	BEGIN_RING_SIZE(NvSub3D, 0x1fc4, 1);
-	OUT_RING(0x06144321);
-	BEGIN_RING_SIZE(NvSub3D, 0x1fc8, 2);
-	OUT_RING(0xedcba987);
-	OUT_RING(0x00000021);
-	BEGIN_RING_SIZE(NvSub3D, 0x1fd0, 1);
-	OUT_RING(0x00171615);
-	BEGIN_RING_SIZE(NvSub3D, 0x1fd4, 1);
-	OUT_RING(0x001b1a19);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1ef8, 1);
-	OUT_RING(0x0020ffff);
-	BEGIN_RING_SIZE(NvSub3D, 0x1d64, 1);
-	OUT_RING(0x00d30000);
-	BEGIN_RING_SIZE(NvSub3D, 0x1e94, 1);
-	OUT_RING(0x00000001);
-
-	return GL_TRUE;
-}
-
-static GLboolean nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
-{
-	GLuint x, y, w, h;
-
-	w = color[0]->mesa.Width;
-	h = color[0]->mesa.Height;
-	x = nmesa->drawX;
-	y = nmesa->drawY;
-
-	if (num_color != 1)
-		return GL_FALSE;
-	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0, 5);
-	OUT_RING        (((w+x)<<16)|x);
-	OUT_RING        (((h+y)<<16)|y);
-	if (color[0]->mesa._ActualFormat == GL_RGBA8)
-		OUT_RING        (0x148);
-	else
-		OUT_RING        (0x143);
-	if (nmesa->screen->card->type >= NV_40)
-		OUT_RING        (color[0]->pitch);
-	else
-		OUT_RING        (color[0]->pitch | (depth ? (depth->pitch << 16): 0));
-	OUT_RING        (color[0]->offset);
-
-	if (depth) {
-		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET, 1);
-		OUT_RING        (depth->offset);
-		if (nmesa->screen->card->type >= NV_40) {
-			BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH, 1);
-			OUT_RING        (depth->pitch);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-void nv30InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	func->AlphaFunc			= nv30AlphaFunc;
-	func->BlendColor		= nv30BlendColor;
-	func->BlendEquationSeparate	= nv30BlendEquationSeparate;
-	func->BlendFuncSeparate		= nv30BlendFuncSeparate;
-	func->Clear			= nv30Clear;
-	func->ClearColor		= nv30ClearColor;
-	func->ClearDepth		= nv30ClearDepth;
-	func->ClearStencil		= nv30ClearStencil;
-	func->ClipPlane			= nv30ClipPlane;
-	func->ColorMask			= nv30ColorMask;
-	func->ColorMaterial		= nv30ColorMaterial;
-	func->CullFace			= nv30CullFace;
-	func->FrontFace			= nv30FrontFace;
-	func->DepthFunc			= nv30DepthFunc;
-	func->DepthMask			= nv30DepthMask;
-	func->DepthRange                = nv30DepthRange;
-	func->Enable			= nv30Enable;
-	func->Fogfv			= nv30Fogfv;
-	func->Hint			= nv30Hint;
-	func->Lightfv			= nv30Lightfv;
-/*	func->LightModelfv		= nv30LightModelfv; */
-	func->LineStipple		= nv30LineStipple;
-	func->LineWidth			= nv30LineWidth;
-	func->LogicOpcode		= nv30LogicOpcode;
-	func->PointParameterfv		= nv30PointParameterfv;
-	func->PointSize			= nv30PointSize;
-	func->PolygonMode		= nv30PolygonMode;
-	func->PolygonOffset		= nv30PolygonOffset;
-	func->PolygonStipple		= nv30PolygonStipple;
-#if 0
-	func->ReadBuffer		= nv30ReadBuffer;
-	func->RenderMode		= nv30RenderMode;
-#endif
-	func->Scissor			= nv30Scissor;
-	func->ShadeModel		= nv30ShadeModel;
-	func->StencilFuncSeparate	= nv30StencilFuncSeparate;
-	func->StencilMaskSeparate	= nv30StencilMaskSeparate;
-	func->StencilOpSeparate		= nv30StencilOpSeparate;
-#if 0
-	func->TexGen			= nv30TexGen;
-	func->TexParameter		= nv30TexParameter;
-#endif
-	func->TextureMatrix		= nv30TextureMatrix;
-
-
-	if (nmesa->screen->card->type >= NV_40)
-	   nmesa->hw_func.InitCard	= nv40InitCard;
-	else 
-	   nmesa->hw_func.InitCard	= nv30InitCard;
-	nmesa->hw_func.BindBuffers	= nv30BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv30WindowMoved;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nv30_vertprog.c b/src/mesa/drivers/dri/nouveau/nv30_vertprog.c
deleted file mode 100644
index d023e8439e3..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv30_vertprog.c
+++ /dev/null
@@ -1,367 +0,0 @@
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-
-#include "nouveau_shader.h"
-#include "nv30_shader.h"
-
-/*****************************************************************************
- * Support routines
- */
-static void
-NV30VPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   int i;
-
-   /* We can do better here and keep more than one VP on the hardware, and
-    * switch between them with PROGRAM_START_ID..
-    */
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_FROM_ID, 1);
-   OUT_RING(0);
-   for (i=0; i<nvs->program_size; i+=4) {
-      BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0, 4);
-      OUT_RING(nvs->program[i + 0]);
-      OUT_RING(nvs->program[i + 1]);
-      OUT_RING(nvs->program[i + 2]);
-      OUT_RING(nvs->program[i + 3]);
-   }
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_PROGRAM_START_ID, 1);
-   OUT_RING(0);
-
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_IN_REG, 2);
-   OUT_RING(nvs->card_priv.NV30VP.vp_in_reg);
-   OUT_RING(nvs->card_priv.NV30VP.vp_out_reg);
-
-   BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_CLIPPING_PLANES, 1);
-   OUT_RING_CACHE  (nvs->card_priv.NV30VP.clip_enables);
-}
-
-static void
-NV30VPUpdateConst(GLcontext *ctx, nouveauShader *nvs, int id)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   GLfloat *val;
-
-   val = nvs->params[id].source_val ?
-      nvs->params[id].source_val : nvs->params[id].val;
-
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID, 5);
-   OUT_RING (id);
-   OUT_RINGp(val, 4);
-}
-
-/*****************************************************************************
- * Assembly routines
- */
-static void
-NV30VPSetBranchTarget(nvsFunc *shader, int addr)
-{
-	shader->inst[2] &= ~NV30_VP_INST_IADDR_MASK;
-	shader->inst[2] |= (addr << NV30_VP_INST_IADDR_SHIFT);
-}
-
-/*****************************************************************************
- * Disassembly routines
- */
-static unsigned int
-NV30VPGetOpcodeHW(nvsFunc * shader, int slot)
-{
-   int op;
-
-   if (slot) {
-      op = (shader->inst[1] & NV30_VP_INST_SCA_OPCODEL_MASK)
-	 >> NV30_VP_INST_SCA_OPCODEL_SHIFT;
-      op |= ((shader->inst[0] & NV30_VP_INST_SCA_OPCODEH_MASK)
-	     >> NV30_VP_INST_SCA_OPCODEH_SHIFT) << 4;
-   }
-   else {
-      op = (shader->inst[1] & NV30_VP_INST_VEC_OPCODE_MASK)
-	 >> NV30_VP_INST_VEC_OPCODE_SHIFT;
-   }
-
-   return op;
-}
-
-static nvsRegFile
-NV30VPGetDestFile(nvsFunc * shader, int merged)
-{
-   switch (shader->GetOpcode(shader, merged)) {
-   case NVS_OP_ARL:
-   case NVS_OP_ARR:
-   case NVS_OP_ARA:
-      return NVS_FILE_ADDRESS;
-   default:
-      /*FIXME: This probably isn't correct.. */
-      if ((shader->inst[3] & NV30_VP_INST_VDEST_WRITEMASK_MASK) != 0)
-	 return NVS_FILE_RESULT;
-      if ((shader->inst[3] & NV30_VP_INST_SDEST_WRITEMASK_MASK) != 0)
-	 return NVS_FILE_RESULT;
-      return NVS_FILE_TEMP;
-   }
-}
-
-static unsigned int
-NV30VPGetDestID(nvsFunc * shader, int merged)
-{
-   int id;
-
-   switch (shader->GetDestFile(shader, merged)) {
-   case NVS_FILE_RESULT:
-      id = ((shader->inst[3] & NV30_VP_INST_DEST_ID_MASK)
-	    >> NV30_VP_INST_DEST_ID_SHIFT);
-      switch (id) {
-      case NV30_VP_INST_DEST_POS  : return NVS_FR_POSITION;
-      case NV30_VP_INST_DEST_COL0 : return NVS_FR_COL0;
-      case NV30_VP_INST_DEST_COL1 : return NVS_FR_COL1;
-      case NV30_VP_INST_DEST_TC(0): return NVS_FR_TEXCOORD0;
-      case NV30_VP_INST_DEST_TC(1): return NVS_FR_TEXCOORD1;
-      case NV30_VP_INST_DEST_TC(2): return NVS_FR_TEXCOORD2;
-      case NV30_VP_INST_DEST_TC(3): return NVS_FR_TEXCOORD3;
-      case NV30_VP_INST_DEST_TC(4): return NVS_FR_TEXCOORD4;
-      case NV30_VP_INST_DEST_TC(5): return NVS_FR_TEXCOORD5;
-      case NV30_VP_INST_DEST_TC(6): return NVS_FR_TEXCOORD6;
-      case NV30_VP_INST_DEST_TC(7): return NVS_FR_TEXCOORD7;
-      default:
-	 return -1;
-      }
-   case NVS_FILE_ADDRESS:
-   case NVS_FILE_TEMP:
-      return (shader->inst[0] & NV30_VP_INST_DEST_TEMP_ID_MASK)
-	 >> NV30_VP_INST_DEST_TEMP_ID_SHIFT;
-   default:
-      return -1;
-   }
-}
-
-static unsigned int
-NV30VPGetDestMask(nvsFunc * shader, int merged)
-{
-   int hwmask, mask = 0;
-
-   if (shader->GetDestFile(shader, merged) == NVS_FILE_RESULT)
-      if (shader->GetOpcodeSlot(shader, merged))
-	 hwmask = (shader->inst[3] & NV30_VP_INST_SDEST_WRITEMASK_MASK)
-	    >> NV30_VP_INST_SDEST_WRITEMASK_SHIFT;
-      else
-	 hwmask = (shader->inst[3] & NV30_VP_INST_VDEST_WRITEMASK_MASK)
-	    >> NV30_VP_INST_VDEST_WRITEMASK_SHIFT;
-   else if (shader->GetOpcodeSlot(shader, merged))
-      hwmask = (shader->inst[3] & NV30_VP_INST_STEMP_WRITEMASK_MASK)
-	 >> NV30_VP_INST_STEMP_WRITEMASK_SHIFT;
-   else
-      hwmask = (shader->inst[3] & NV30_VP_INST_VTEMP_WRITEMASK_MASK)
-	 >> NV30_VP_INST_VTEMP_WRITEMASK_SHIFT;
-
-   if (hwmask & (1 << 3)) mask |= SMASK_X;
-   if (hwmask & (1 << 2)) mask |= SMASK_Y;
-   if (hwmask & (1 << 1)) mask |= SMASK_Z;
-   if (hwmask & (1 << 0)) mask |= SMASK_W;
-
-   return mask;
-}
-
-static int
-NV30VPGetSourceID(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_TEMP:
-      src = shader->GetSourceHW(shader, merged, pos);
-      return ((src & NV30_VP_SRC_REG_TEMP_ID_MASK) >>
-	      NV30_VP_SRC_REG_TEMP_ID_SHIFT);
-   case NVS_FILE_CONST:
-      return ((shader->inst[1] & NV30_VP_INST_CONST_SRC_MASK)
-	      >> NV30_VP_INST_CONST_SRC_SHIFT);
-   case NVS_FILE_ATTRIB:
-      src = ((shader->inst[1] & NV30_VP_INST_INPUT_SRC_MASK)
-	     >> NV30_VP_INST_INPUT_SRC_SHIFT);
-      switch (src) {
-      case NV30_VP_INST_IN_POS  : return NVS_FR_POSITION;
-      case NV30_VP_INST_IN_COL0 : return NVS_FR_COL0;
-      case NV30_VP_INST_IN_COL1 : return NVS_FR_COL1;
-      case NV30_VP_INST_IN_TC(0): return NVS_FR_TEXCOORD0;
-      case NV30_VP_INST_IN_TC(1): return NVS_FR_TEXCOORD1;
-      case NV30_VP_INST_IN_TC(2): return NVS_FR_TEXCOORD2;
-      case NV30_VP_INST_IN_TC(3): return NVS_FR_TEXCOORD3;
-      case NV30_VP_INST_IN_TC(4): return NVS_FR_TEXCOORD4;
-      case NV30_VP_INST_IN_TC(5): return NVS_FR_TEXCOORD5;
-      case NV30_VP_INST_IN_TC(6): return NVS_FR_TEXCOORD6;
-      case NV30_VP_INST_IN_TC(7): return NVS_FR_TEXCOORD7;
-      default:
-	 return NVS_FR_UNKNOWN;
-      }
-   default:
-      return -1;
-   }
-}
-
-static int
-NV30VPGetSourceAbs(nvsFunc * shader, int merged, int pos)
-{
-   struct _op_xlat *opr;
-   static unsigned int abspos[3] = {
-      NV30_VP_INST_SRC0_ABS,
-      NV30_VP_INST_SRC1_ABS,
-      NV30_VP_INST_SRC2_ABS,
-   };
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1 || opr->srcpos[pos] > 2)
-      return 0;
-
-   return ((shader->inst[0] & abspos[opr->srcpos[pos]]) ? 1 : 0);
-}
-
-static int
-NV30VPGetRelAddressRegID(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV30_VP_INST_ADDR_REG_SELECT_1) ? 1 : 0);
-}
-
-static nvsSwzComp
-NV30VPGetRelAddressSwizzle(nvsFunc * shader)
-{
-   nvsSwzComp swz;
-
-   swz = NV20VP_TX_SWIZZLE[(shader->inst[0] & NV30_VP_INST_ADDR_SWZ_MASK)
-			   >> NV30_VP_INST_ADDR_SWZ_SHIFT];
-   return swz;
-}
-
-static int
-NV30VPSupportsConditional(nvsFunc * shader)
-{
-   /*FIXME: Is this true of all ops? */
-   return 1;
-}
-
-static int
-NV30VPGetConditionUpdate(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV30_VP_INST_COND_UPDATE_ENABLE) ? 1 : 0);
-}
-
-static int
-NV30VPGetConditionTest(nvsFunc * shader)
-{
-   int op;
-
-   /* The condition test is unconditionally enabled on some
-    * instructions. ie: the condition test bit does *NOT* have
-    * to be set.
-    *
-    * FIXME: check other relevant ops for this situation.
-    */
-   op = shader->GetOpcodeHW(shader, 1);
-   switch (op) {
-   case NV30_VP_INST_OP_BRA:
-      return 1;
-   default:
-      return ((shader->inst[0] & NV30_VP_INST_COND_TEST_ENABLE) ? 1 : 0);
-   }
-}
-
-static nvsCond
-NV30VPGetCondition(nvsFunc * shader)
-{
-   int cond;
-
-   cond = ((shader->inst[0] & NV30_VP_INST_COND_MASK)
-	   >> NV30_VP_INST_COND_SHIFT);
-
-   switch (cond) {
-   case NV30_VP_INST_COND_FL: return NVS_COND_FL;
-   case NV30_VP_INST_COND_LT: return NVS_COND_LT;
-   case NV30_VP_INST_COND_EQ: return NVS_COND_EQ;
-   case NV30_VP_INST_COND_LE: return NVS_COND_LE;
-   case NV30_VP_INST_COND_GT: return NVS_COND_GT;
-   case NV30_VP_INST_COND_NE: return NVS_COND_NE;
-   case NV30_VP_INST_COND_GE: return NVS_COND_GE;
-   case NV30_VP_INST_COND_TR: return NVS_COND_TR;
-   default:
-      return NVS_COND_UNKNOWN;
-   }
-}
-
-static void
-NV30VPGetCondRegSwizzle(nvsFunc * shader, nvsSwzComp *swz)
-{
-   int swzbits;
-
-   swzbits = (shader->inst[0] & NV30_VP_INST_COND_SWZ_ALL_MASK)
-      >> NV30_VP_INST_COND_SWZ_ALL_SHIFT;
-   NV20VPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV30VPGetCondRegID(nvsFunc * shader)
-{
-   return 0;
-}
-
-
-static int
-NV30VPGetBranch(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV30_VP_INST_IADDR_MASK)
-	   >> NV30_VP_INST_IADDR_SHIFT);
-}
-
-void
-NV30VPInitShaderFuncs(nvsFunc * shader)
-{
-   /* Inherit NV20 code, a lot of it is the same */
-   NV20VPInitShaderFuncs(shader);
-
-   /* Increase max valid opcode ID, and add new instructions */
-   NVVP_TX_VOP_COUNT = NVVP_TX_NVS_OP_COUNT = 32;
-
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_FRC, NVS_OP_FRC,  0, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_FLR, NVS_OP_FLR,  0, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SEQ, NVS_OP_SEQ,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SFL, NVS_OP_SFL,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SGT, NVS_OP_SGT,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SLE, NVS_OP_SLE,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SNE, NVS_OP_SNE,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_STR, NVS_OP_STR,  0,  1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_SSG, NVS_OP_SSG,  0, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_ARR, NVS_OP_ARR,  0, -1, -1);
-   MOD_OPCODE(NVVP_TX_VOP, NV30_VP_INST_OP_ARA, NVS_OP_ARA,  3, -1, -1);
-
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_BRA, NVS_OP_BRA, -1, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_CAL, NVS_OP_CAL, -1, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_RET, NVS_OP_RET, -1, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_LG2, NVS_OP_LG2,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_EX2, NVS_OP_EX2,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_SIN, NVS_OP_SIN,  2, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV30_VP_INST_OP_COS, NVS_OP_COS,  2, -1, -1);
-
-   shader->UploadToHW		= NV30VPUploadToHW;
-   shader->UpdateConst		= NV30VPUpdateConst;
-
-   shader->GetOpcodeHW		= NV30VPGetOpcodeHW;
-
-   shader->GetDestFile		= NV30VPGetDestFile;
-   shader->GetDestID		= NV30VPGetDestID;
-   shader->GetDestMask		= NV30VPGetDestMask;
-
-   shader->GetSourceID		= NV30VPGetSourceID;
-   shader->GetSourceAbs		= NV30VPGetSourceAbs;
-
-   shader->GetRelAddressRegID	= NV30VPGetRelAddressRegID;
-   shader->GetRelAddressSwizzle	= NV30VPGetRelAddressSwizzle;
-
-   shader->SupportsConditional	= NV30VPSupportsConditional;
-   shader->GetConditionUpdate	= NV30VPGetConditionUpdate;
-   shader->GetConditionTest	= NV30VPGetConditionTest;
-   shader->GetCondition		= NV30VPGetCondition;
-   shader->GetCondRegSwizzle	= NV30VPGetCondRegSwizzle;
-   shader->GetCondRegID		= NV30VPGetCondRegID;
-
-   shader->GetBranch		= NV30VPGetBranch;
-   shader->SetBranchTarget	= NV30VPSetBranchTarget;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nv40_fragprog.c b/src/mesa/drivers/dri/nouveau/nv40_fragprog.c
deleted file mode 100644
index 3e4ae0496e4..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv40_fragprog.c
+++ /dev/null
@@ -1,224 +0,0 @@
-#include "nouveau_shader.h"
-#include "nv40_shader.h"
-
-/* branching ops */
-unsigned int NVFP_TX_BOP_COUNT = 5;
-struct _op_xlat NVFP_TX_BOP[64];
-
-
-/*****************************************************************************
- * Assembly routines
- * 	- These extend the NV30 routines, which are almost identical.  NV40
- * 	  just has branching hacked into the instruction set.
- */
-static int
-NV40FPSupportsResultScale(nvsFunc *shader, nvsScale scale)
-{
-	switch (scale) {
-	case NVS_SCALE_1X:
-	case NVS_SCALE_2X:
-	case NVS_SCALE_4X:
-	case NVS_SCALE_8X:
-	case NVS_SCALE_INV_2X:
-	case NVS_SCALE_INV_4X:
-	case NVS_SCALE_INV_8X:
-		return 1;
-	default:
-		return 0;
-	}
-}
-
-static void
-NV40FPSetResultScale(nvsFunc *shader, nvsScale scale)
-{
-	shader->inst[2] &= ~NV40_FP_OP_DST_SCALE_MASK;
-	shader->inst[2] |= ((unsigned int)scale << NV40_FP_OP_DST_SCALE_SHIFT);
-}
-
-static void
-NV40FPSetBranchTarget(nvsFunc *shader, int addr)
-{
-	shader->inst[2] &= ~NV40_FP_OP_IADDR_MASK;
-	shader->inst[2] |= (addr << NV40_FP_OP_IADDR_SHIFT);
-}
-
-static void
-NV40FPSetBranchElse(nvsFunc *shader, int addr)
-{
-	shader->inst[2] &= ~NV40_FP_OP_ELSE_ID_MASK;
-	shader->inst[2] |= (addr << NV40_FP_OP_ELSE_ID_SHIFT);
-}
-
-static void
-NV40FPSetBranchEnd(nvsFunc *shader, int addr)
-{
-	shader->inst[3] &= ~NV40_FP_OP_END_ID_MASK;
-	shader->inst[3] |= (addr << NV40_FP_OP_END_ID_SHIFT);
-}
-
-static void
-NV40FPSetLoopParams(nvsFunc *shader, int count, int initial, int increment)
-{
-	shader->inst[2] &= ~(NV40_FP_OP_LOOP_COUNT_MASK |
-			     NV40_FP_OP_LOOP_INDEX_MASK |
-			     NV40_FP_OP_LOOP_INCR_MASK);
-	shader->inst[2] |= ((count     << NV40_FP_OP_LOOP_COUNT_SHIFT) |
-			    (initial   << NV40_FP_OP_LOOP_INDEX_SHIFT) |
-			    (increment << NV40_FP_OP_LOOP_INCR_SHIFT));
-}
-
-/*****************************************************************************
- * Disassembly routines
- */
-static struct _op_xlat *
-NV40FPGetOPTXRec(nvsFunc * shader, int merged)
-{
-   struct _op_xlat *opr;
-   int op;
-
-   op = shader->GetOpcodeHW(shader, 0);
-   if (shader->inst[2] & NV40_FP_OP_OPCODE_IS_BRANCH) {
-      opr = NVFP_TX_BOP;
-      op &= ~NV40_FP_OP_OPCODE_IS_BRANCH;
-      if (op > NVFP_TX_BOP_COUNT)
-	 return NULL;
-   }
-   else {
-      opr = NVFP_TX_AOP;
-      if (op > NVFP_TX_AOP_COUNT)
-	 return NULL;
-   }
-
-   if (opr[op].SOP == NVS_OP_UNKNOWN)
-      return NULL;
-   return &opr[op];
-}
-
-static int
-NV40FPGetSourceID(nvsFunc * shader, int merged, int pos)
-{
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_ATTRIB:
-      switch ((shader->inst[0] & NV40_FP_OP_INPUT_SRC_MASK)
-	      >> NV40_FP_OP_INPUT_SRC_SHIFT) {
-      case NV40_FP_OP_INPUT_SRC_POSITION: return NVS_FR_POSITION;
-      case NV40_FP_OP_INPUT_SRC_COL0    : return NVS_FR_COL0;
-      case NV40_FP_OP_INPUT_SRC_COL1    : return NVS_FR_COL1;
-      case NV40_FP_OP_INPUT_SRC_FOGC    : return NVS_FR_FOGCOORD;
-      case NV40_FP_OP_INPUT_SRC_TC(0)   : return NVS_FR_TEXCOORD0;
-      case NV40_FP_OP_INPUT_SRC_TC(1)   : return NVS_FR_TEXCOORD1;
-      case NV40_FP_OP_INPUT_SRC_TC(2)   : return NVS_FR_TEXCOORD2;
-      case NV40_FP_OP_INPUT_SRC_TC(3)   : return NVS_FR_TEXCOORD3;
-      case NV40_FP_OP_INPUT_SRC_TC(4)   : return NVS_FR_TEXCOORD4;
-      case NV40_FP_OP_INPUT_SRC_TC(5)   : return NVS_FR_TEXCOORD5;
-      case NV40_FP_OP_INPUT_SRC_TC(6)   : return NVS_FR_TEXCOORD6;
-      case NV40_FP_OP_INPUT_SRC_TC(7)   : return NVS_FR_TEXCOORD7;
-      case NV40_FP_OP_INPUT_SRC_FACING  : return NVS_FR_FACING;
-      default:
-	 return -1;
-      }
-      break;
-   case NVS_FILE_TEMP:
-      {
-	 unsigned int src;
-
-	 src = shader->GetSourceHW(shader, merged, pos);
-	 return ((src & NV40_FP_REG_SRC_MASK) >> NV40_FP_REG_SRC_SHIFT);
-      }
-   case NVS_FILE_CONST:		/* inlined into fragprog */
-   default:
-      return -1;
-   }
-}
-
-static int
-NV40FPGetBranch(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV40_FP_OP_IADDR_MASK)
-	   >> NV40_FP_OP_IADDR_SHIFT);;
-}
-
-static int
-NV40FPGetBranchElse(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV40_FP_OP_ELSE_ID_MASK)
-	   >> NV40_FP_OP_ELSE_ID_SHIFT);
-}
-
-static int
-NV40FPGetBranchEnd(nvsFunc * shader)
-{
-   return ((shader->inst[3] & NV40_FP_OP_END_ID_MASK)
-	   >> NV40_FP_OP_END_ID_SHIFT);
-}
-
-static int
-NV40FPGetLoopCount(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV40_FP_OP_LOOP_COUNT_MASK)
-	   >> NV40_FP_OP_LOOP_COUNT_SHIFT);
-}
-
-static int
-NV40FPGetLoopInitial(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV40_FP_OP_LOOP_INDEX_MASK)
-	   >> NV40_FP_OP_LOOP_INDEX_SHIFT);
-}
-
-static int
-NV40FPGetLoopIncrement(nvsFunc * shader)
-{
-   return ((shader->inst[2] & NV40_FP_OP_LOOP_INCR_MASK)
-	   >> NV40_FP_OP_LOOP_INCR_SHIFT);
-}
-
-void
-NV40FPInitShaderFuncs(nvsFunc * shader)
-{
-   /* Inherit NV30 FP code, it's mostly the same */
-   NV30FPInitShaderFuncs(shader);
-
-   /* Kill off opcodes seen on NV30, but not seen on NV40 - need to find
-    * out if these actually work or not.
-    *
-    * update: either LIT/RSQ don't work on nv40, or I generate bad code for
-    *         them.  haven't tested the others yet
-    */
-   MOD_OPCODE(NVFP_TX_AOP, 0x1B, NVS_OP_UNKNOWN, -1, -1, -1);	/* NV30 RSQ */
-   MOD_OPCODE(NVFP_TX_AOP, 0x1E, NVS_OP_UNKNOWN, -1, -1, -1);	/* NV30 LIT */
-   MOD_OPCODE(NVFP_TX_AOP, 0x1F, NVS_OP_UNKNOWN, -1, -1, -1);	/* NV30 LRP */
-   MOD_OPCODE(NVFP_TX_AOP, 0x26, NVS_OP_UNKNOWN, -1, -1, -1);	/* NV30 POW */
-   MOD_OPCODE(NVFP_TX_AOP, 0x36, NVS_OP_UNKNOWN, -1, -1, -1);	/* NV30 RFL */
-
-   /* Extra opcodes supported on NV40 */
-   MOD_OPCODE(NVFP_TX_AOP, NV40_FP_OP_OPCODE_DIV     , NVS_OP_DIV ,  0,  1, -1);
-   MOD_OPCODE(NVFP_TX_AOP, NV40_FP_OP_OPCODE_DP2A    , NVS_OP_DP2A,  0,  1,  2);
-   MOD_OPCODE(NVFP_TX_AOP, NV40_FP_OP_OPCODE_TXL     , NVS_OP_TXL ,  0, -1, -1);
-
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_BRK , NVS_OP_BRK , -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_CAL , NVS_OP_CAL , -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_IF  , NVS_OP_IF  , -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_LOOP, NVS_OP_LOOP, -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_REP , NVS_OP_REP , -1, -1, -1);
-   MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_RET , NVS_OP_RET , -1, -1, -1);
-
-   shader->SupportsResultScale	= NV40FPSupportsResultScale;
-   shader->SetResultScale	= NV40FPSetResultScale;
-
-   /* fragment.facing */
-   shader->GetSourceID		= NV40FPGetSourceID;
-
-   /* branching */
-   shader->GetOPTXRec		= NV40FPGetOPTXRec;
-   shader->GetBranch		= NV40FPGetBranch;
-   shader->GetBranchElse	= NV40FPGetBranchElse;
-   shader->GetBranchEnd		= NV40FPGetBranchEnd;
-   shader->GetLoopCount		= NV40FPGetLoopCount;
-   shader->GetLoopInitial	= NV40FPGetLoopInitial;
-   shader->GetLoopIncrement	= NV40FPGetLoopIncrement;
-   shader->SetBranchTarget	= NV40FPSetBranchTarget;
-   shader->SetBranchElse	= NV40FPSetBranchElse;
-   shader->SetBranchEnd		= NV40FPSetBranchEnd;
-   shader->SetLoopParams	= NV40FPSetLoopParams;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv40_vertprog.c b/src/mesa/drivers/dri/nouveau/nv40_vertprog.c
deleted file mode 100644
index d054140bcd0..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv40_vertprog.c
+++ /dev/null
@@ -1,778 +0,0 @@
-#include "nouveau_shader.h"
-#include "nouveau_msg.h"
-#include "nv40_shader.h"
-
-/*****************************************************************************
- * Assembly routines
- */
-static int
-NV40VPSupportsOpcode(nvsFunc * shader, nvsOpcode op)
-{
-   if (shader->GetOPTXFromSOP(op, NULL))
-      return 1;
-   return 0;
-}
-
-static void
-NV40VPSetOpcode(nvsFunc *shader, unsigned int opcode, int slot)
-{
-   if (slot) {
-      shader->inst[1] &= ~NV40_VP_INST_SCA_OPCODE_MASK;
-      shader->inst[1] |= (opcode << NV40_VP_INST_SCA_OPCODE_SHIFT);
-   } else {
-      shader->inst[1] &= ~NV40_VP_INST_VEC_OPCODE_MASK;
-      shader->inst[1] |= (opcode << NV40_VP_INST_VEC_OPCODE_SHIFT);
-   }
-}
-
-static void
-NV40VPSetCCUpdate(nvsFunc *shader)
-{
-   shader->inst[0] |= NV40_VP_INST_COND_UPDATE_ENABLE;
-}
-
-static void
-NV40VPSetCondition(nvsFunc *shader, int on, nvsCond cond, int reg,
-      		   nvsSwzComp *swizzle)
-{
-   unsigned int hwcond;
-
-   if (on ) shader->inst[0] |= NV40_VP_INST_COND_TEST_ENABLE;
-   else     shader->inst[0] &= ~NV40_VP_INST_COND_TEST_ENABLE;
-   if (reg) shader->inst[0] |= NV40_VP_INST_COND_REG_SELECT_1;
-   else     shader->inst[0] &= ~NV40_VP_INST_COND_REG_SELECT_1;
-
-   switch (cond) {
-   case NVS_COND_TR: hwcond = NV40_VP_INST_COND_TR; break;
-   case NVS_COND_FL: hwcond = NV40_VP_INST_COND_FL; break;
-   case NVS_COND_LT: hwcond = NV40_VP_INST_COND_LT; break;
-   case NVS_COND_GT: hwcond = NV40_VP_INST_COND_GT; break;
-   case NVS_COND_NE: hwcond = NV40_VP_INST_COND_NE; break;
-   case NVS_COND_EQ: hwcond = NV40_VP_INST_COND_EQ; break;
-   case NVS_COND_GE: hwcond = NV40_VP_INST_COND_GE; break;
-   case NVS_COND_LE: hwcond = NV40_VP_INST_COND_LE; break;
-   default:
-	WARN_ONCE("unknown vp cond %d\n", cond);
-	hwcond = NV40_VP_INST_COND_TR;
-	break;
-   }
-   shader->inst[0] &= ~NV40_VP_INST_COND_MASK;
-   shader->inst[0] |= (hwcond << NV40_VP_INST_COND_SHIFT);
-
-   shader->inst[0] &= ~NV40_VP_INST_COND_SWZ_ALL_MASK;
-   shader->inst[0] |= (swizzle[NVS_SWZ_X] << NV40_VP_INST_COND_SWZ_X_SHIFT);
-   shader->inst[0] |= (swizzle[NVS_SWZ_Y] << NV40_VP_INST_COND_SWZ_Y_SHIFT);
-   shader->inst[0] |= (swizzle[NVS_SWZ_Z] << NV40_VP_INST_COND_SWZ_Z_SHIFT);
-   shader->inst[0] |= (swizzle[NVS_SWZ_W] << NV40_VP_INST_COND_SWZ_W_SHIFT);
-}
-
-/* these just exist here until nouveau_reg.h has them. */
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL0	(1<<0)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1	(1<<1)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0	(1<<2)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1	(1<<3)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC	(1<<4)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ	(1<<5)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0	(1<<6)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1	(1<<7)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2	(1<<8)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3	(1<<9)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4	(1<<10)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5	(1<<11)
-#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0   (1<<14)
-
-static unsigned int
-NV40VPTranslateResultReg(nvsFunc *shader, nvsFixedReg result,
-					  unsigned int *mask_ret)
-{
-	unsigned int *out_reg = &shader->card_priv->NV30VP.vp_out_reg;
-	unsigned int *clip_en = &shader->card_priv->NV30VP.clip_enables;
-
-	*mask_ret = 0xf;
-
-	switch (result) {
-	case NVS_FR_POSITION:
-		/* out_reg POS implied */
-		return NV40_VP_INST_DEST_POS;
-	case NVS_FR_COL0:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL0;
-		return NV40_VP_INST_DEST_COL0;
-	case NVS_FR_COL1:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1;
-		return NV40_VP_INST_DEST_COL1;
-	case NVS_FR_BFC0:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0;
-		return NV40_VP_INST_DEST_BFC0;
-	case NVS_FR_BFC1:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1;
-		return NV40_VP_INST_DEST_BFC1;
-	case NVS_FR_FOGCOORD:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC;
-		*mask_ret = 0x8;
-		return NV40_VP_INST_DEST_FOGC;
-	case NVS_FR_CLIP0:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0;
-		(*clip_en) |= 0x00000002;
-		*mask_ret = 0x4;
-		return NV40_VP_INST_DEST_FOGC;
-	case NVS_FR_CLIP1:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1;
-		(*clip_en) |= 0x00000020;
-		*mask_ret = 0x2;
-		return NV40_VP_INST_DEST_FOGC;
-	case NVS_FR_CLIP2:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2;
-		(*clip_en) |= 0x00000200;
-		*mask_ret = 0x1;
-		return NV40_VP_INST_DEST_FOGC;
-	case NVS_FR_POINTSZ:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ;
-		*mask_ret = 0x8;
-		return NV40_VP_INST_DEST_PSZ;
-	case NVS_FR_CLIP3:
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3;
-		(*clip_en) |= 0x00002000;
-		*mask_ret = 0x4;
-		return NV40_VP_INST_DEST_PSZ;
-	case NVS_FR_CLIP4:
-		(*clip_en) |= 0x00020000;
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4;
-		*mask_ret = 0x2;
-		return NV40_VP_INST_DEST_PSZ;
-	case NVS_FR_CLIP5:
-		(*clip_en) |= 0x00200000;
-		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5;
-		*mask_ret = 0x1;
-		return NV40_VP_INST_DEST_PSZ;
-	case NVS_FR_TEXCOORD0:
-	case NVS_FR_TEXCOORD1:
-	case NVS_FR_TEXCOORD2:
-	case NVS_FR_TEXCOORD3:
-	case NVS_FR_TEXCOORD4:
-	case NVS_FR_TEXCOORD5:
-	case NVS_FR_TEXCOORD6:
-	case NVS_FR_TEXCOORD7:
-	{
-		int unit = result - NVS_FR_TEXCOORD0;
-		(*out_reg) |= (NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0 << unit);
-		return NV40_VP_INST_DEST_TC(unit);
-	}
-	default:
-		WARN_ONCE("unknown vp output %d\n", result);
-		return NV40_VP_INST_DEST_POS;
-	}
-}
-
-static void
-NV40VPSetResult(nvsFunc *shader, nvsRegister * dest, unsigned int mask,
-      		int slot)
-{
-   unsigned int hwmask = 0;
-
-   if (mask & SMASK_X) hwmask |= (1 << 3);
-   if (mask & SMASK_Y) hwmask |= (1 << 2);
-   if (mask & SMASK_Z) hwmask |= (1 << 1);
-   if (mask & SMASK_W) hwmask |= (1 << 0);
-
-   if (dest->file == NVS_FILE_RESULT) {
-      unsigned int valid_mask;
-      int hwidx;
-
-      hwidx = NV40VPTranslateResultReg(shader, dest->index, &valid_mask);
-      if (hwmask & ~valid_mask)
-	      WARN_ONCE("writing invalid components of result reg\n");
-      hwmask &= valid_mask;
-
-      shader->inst[3] &= ~NV40_VP_INST_DEST_MASK;
-      shader->inst[3] |= (hwidx << NV40_VP_INST_DEST_SHIFT);
-
-      if (slot) shader->inst[3] |= NV40_VP_INST_SCA_RESULT;
-      else      shader->inst[0] |= NV40_VP_INST_VEC_RESULT;
-   } else {
-      /* NVS_FILE_TEMP || NVS_FILE_ADDRESS */
-      if (slot) {
-	 shader->inst[3] &= ~NV40_VP_INST_SCA_RESULT;
-	 shader->inst[3] &= ~NV40_VP_INST_SCA_DEST_TEMP_MASK;
-	 shader->inst[3] |= (dest->index << NV40_VP_INST_SCA_DEST_TEMP_SHIFT);
-      } else {
-	 shader->inst[0] &= ~NV40_VP_INST_VEC_RESULT;
-	 shader->inst[0] &= ~(NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20));
-	 shader->inst[0] |= (dest->index << NV40_VP_INST_VEC_DEST_TEMP_SHIFT);
-      }
-   }
-
-   if (slot) {
-      shader->inst[3] &= ~NV40_VP_INST_SCA_WRITEMASK_MASK;
-      shader->inst[3] |= (hwmask << NV40_VP_INST_SCA_WRITEMASK_SHIFT);
-   } else {
-      shader->inst[3] &= ~NV40_VP_INST_VEC_WRITEMASK_MASK;
-      shader->inst[3] |= (hwmask << NV40_VP_INST_VEC_WRITEMASK_SHIFT);
-   }
-}
-
-static void
-NV40VPInsertSource(nvsFunc *shader, unsigned int hw, int pos)
-{
-   switch (pos) {
-   case 0:
-      shader->inst[1] &= ~NV40_VP_INST_SRC0H_MASK;
-      shader->inst[2] &= ~NV40_VP_INST_SRC0L_MASK;
-      shader->inst[1] |= ((hw & NV40_VP_SRC0_HIGH_MASK) >>
-	    NV40_VP_SRC0_HIGH_SHIFT)
-	 << NV40_VP_INST_SRC0H_SHIFT;
-      shader->inst[2] |= (hw & NV40_VP_SRC0_LOW_MASK)
-	 << NV40_VP_INST_SRC0L_SHIFT;
-      break;
-   case 1:
-      shader->inst[2] &= ~NV40_VP_INST_SRC1_MASK;
-      shader->inst[2] |= hw
-	 << NV40_VP_INST_SRC1_SHIFT;
-      break;
-   case 2:
-      shader->inst[2] &= ~NV40_VP_INST_SRC2H_MASK;
-      shader->inst[3] &= ~NV40_VP_INST_SRC2L_MASK;
-      shader->inst[2] |= ((hw & NV40_VP_SRC2_HIGH_MASK) >>
-	    NV40_VP_SRC2_HIGH_SHIFT)
-	 << NV40_VP_INST_SRC2H_SHIFT;
-      shader->inst[3] |= (hw & NV40_VP_SRC2_LOW_MASK)
-	 << NV40_VP_INST_SRC2L_SHIFT;
-      break;
-   default:
-      assert(0);
-      break;
-   }
-}
-
-static void
-NV40VPSetSource(nvsFunc *shader, nvsRegister * src, int pos)
-{
-   unsigned int hw = 0;
-
-   switch (src->file) {
-   case NVS_FILE_ADDRESS:
-      break;
-   case NVS_FILE_ATTRIB:
-      hw |= (NV40_VP_SRC_REG_TYPE_INPUT << NV40_VP_SRC_REG_TYPE_SHIFT);
-
-      shader->inst[1] &= ~NV40_VP_INST_INPUT_SRC_MASK;
-      shader->inst[1] |= (src->index << NV40_VP_INST_INPUT_SRC_SHIFT);
-      shader->card_priv->NV30VP.vp_in_reg |= (1 << src->index);
-      if (src->indexed) {
-	 shader->inst[0] |= NV40_VP_INST_INDEX_INPUT;
-	 if (src->addr_reg)
-	    shader->inst[0] |= NV40_VP_INST_ADDR_REG_SELECT_1;
-	 else
-	    shader->inst[0] &= ~NV40_VP_INST_ADDR_REG_SELECT_1;
-	 shader->inst[0] &= ~NV40_VP_INST_ADDR_SWZ_SHIFT;
-	 shader->inst[0] |= (src->addr_comp << NV40_VP_INST_ADDR_SWZ_SHIFT);
-      } else
-	 shader->inst[0] &= ~NV40_VP_INST_INDEX_INPUT;
-      break;
-   case NVS_FILE_CONST:
-      hw |= (NV40_VP_SRC_REG_TYPE_CONST << NV40_VP_SRC_REG_TYPE_SHIFT);
-
-      shader->inst[1] &= ~NV40_VP_INST_CONST_SRC_MASK;
-      shader->inst[1] |= (src->index << NV40_VP_INST_CONST_SRC_SHIFT);
-      if (src->indexed) {
-	 shader->inst[3] |= NV40_VP_INST_INDEX_CONST;
-	 if (src->addr_reg)
-	    shader->inst[0] |= NV40_VP_INST_ADDR_REG_SELECT_1;
-	 else
-	    shader->inst[0] &= ~NV40_VP_INST_ADDR_REG_SELECT_1;
-	 shader->inst[0] &= ~NV40_VP_INST_ADDR_SWZ_MASK;
-	 shader->inst[0] |= (src->addr_comp << NV40_VP_INST_ADDR_SWZ_SHIFT);
-      } else
-	 shader->inst[3] &= ~NV40_VP_INST_INDEX_CONST;
-      break;
-   case NVS_FILE_TEMP:
-      hw |= (NV40_VP_SRC_REG_TYPE_TEMP << NV40_VP_SRC_REG_TYPE_SHIFT);
-      hw |= (src->index << NV40_VP_SRC_TEMP_SRC_SHIFT);
-      break;
-   default:
-      fprintf(stderr, "unknown source file %d\n", src->file);
-      assert(0);
-      break;
-   }
-
-   if (src->file != NVS_FILE_ADDRESS) {
-      if (src->negate)
-	 hw |= NV40_VP_SRC_NEGATE;
-      if (src->abs)
-	 shader->inst[0] |= (1 << (21 + pos));
-      else
-	 shader->inst[0] &= ~(1 << (21 + pos));
-      hw |= (src->swizzle[0] << NV40_VP_SRC_SWZ_X_SHIFT);
-      hw |= (src->swizzle[1] << NV40_VP_SRC_SWZ_Y_SHIFT);
-      hw |= (src->swizzle[2] << NV40_VP_SRC_SWZ_Z_SHIFT);
-      hw |= (src->swizzle[3] << NV40_VP_SRC_SWZ_W_SHIFT);
-
-      NV40VPInsertSource(shader, hw, pos);
-   }
-}
-
-static void
-NV40VPSetBranchTarget(nvsFunc *shader, int addr)
-{
-	shader->inst[2] &= ~NV40_VP_INST_IADDRH_MASK;
-	shader->inst[2] |= ((addr & 0xf8) >> 3) << NV40_VP_INST_IADDRH_SHIFT;
-	shader->inst[3] &= ~NV40_VP_INST_IADDRL_MASK;
-	shader->inst[3] |= ((addr & 0x07) << NV40_VP_INST_IADDRL_SHIFT);
-}
-
-static void
-NV40VPInitInstruction(nvsFunc *shader)
-{
-   unsigned int hwsrc = 0;
-
-   shader->inst[0] = /*NV40_VP_INST_VEC_RESULT | */
-      		     NV40_VP_INST_VEC_DEST_TEMP_MASK | (1<<20);
-   shader->inst[1] = 0;
-   shader->inst[2] = 0;
-   shader->inst[3] = NV40_VP_INST_SCA_RESULT |
-      		     NV40_VP_INST_SCA_DEST_TEMP_MASK |
-		     NV40_VP_INST_DEST_MASK;
-   
-   hwsrc = (NV40_VP_SRC_REG_TYPE_INPUT << NV40_VP_SRC_REG_TYPE_SHIFT) |
-      	   (NVS_SWZ_X << NV40_VP_SRC_SWZ_X_SHIFT) |
-      	   (NVS_SWZ_Y << NV40_VP_SRC_SWZ_Y_SHIFT) |
-      	   (NVS_SWZ_Z << NV40_VP_SRC_SWZ_Z_SHIFT) |
-      	   (NVS_SWZ_W << NV40_VP_SRC_SWZ_W_SHIFT);
-   NV40VPInsertSource(shader, hwsrc, 0);
-   NV40VPInsertSource(shader, hwsrc, 1);
-   NV40VPInsertSource(shader, hwsrc, 2);
-}
-
-static void
-NV40VPSetLastInst(nvsFunc *shader)
-{
-   shader->inst[3] |= 1;
-}
-
-/*****************************************************************************
- * Disassembly routines
- */
-static int
-NV40VPHasMergedInst(nvsFunc * shader)
-{
-   if (shader->GetOpcodeHW(shader, 0) != NV40_VP_INST_OP_NOP &&
-       shader->GetOpcodeHW(shader, 1) != NV40_VP_INST_OP_NOP)
-      return 1;
-   return 0;
-}
-
-static unsigned int
-NV40VPGetOpcodeHW(nvsFunc * shader, int slot)
-{
-   int op;
-
-   if (slot)
-      op = (shader->inst[1] & NV40_VP_INST_SCA_OPCODE_MASK)
-	 >> NV40_VP_INST_SCA_OPCODE_SHIFT;
-   else
-      op = (shader->inst[1] & NV40_VP_INST_VEC_OPCODE_MASK)
-	 >> NV40_VP_INST_VEC_OPCODE_SHIFT;
-
-   return op;
-}
-
-static nvsRegFile
-NV40VPGetDestFile(nvsFunc * shader, int merged)
-{
-   nvsOpcode op;
-
-   op = shader->GetOpcode(shader, merged);
-   switch (op) {
-   case NVS_OP_ARL:
-   case NVS_OP_ARR:
-   case NVS_OP_ARA:
-   case NVS_OP_POPA:
-      return NVS_FILE_ADDRESS;
-   default:
-      if (shader->GetOpcodeSlot(shader, merged)) {
-	 if (shader->inst[3] & NV40_VP_INST_SCA_RESULT)
-	    return NVS_FILE_RESULT;
-      }
-      else {
-	 if (shader->inst[0] & NV40_VP_INST_VEC_RESULT)
-	    return NVS_FILE_RESULT;
-      }
-      return NVS_FILE_TEMP;
-   }
-
-}
-
-static unsigned int
-NV40VPGetDestID(nvsFunc * shader, int merged)
-{
-   int id;
-
-   switch (shader->GetDestFile(shader, merged)) {
-   case NVS_FILE_RESULT:
-      id = ((shader->inst[3] & NV40_VP_INST_DEST_MASK)
-	    >> NV40_VP_INST_DEST_SHIFT);
-      switch (id) {
-      case NV40_VP_INST_DEST_POS : return NVS_FR_POSITION;
-      case NV40_VP_INST_DEST_COL0: return NVS_FR_COL0;
-      case NV40_VP_INST_DEST_COL1: return NVS_FR_COL1;
-      case NV40_VP_INST_DEST_BFC0: return NVS_FR_BFC0;
-      case NV40_VP_INST_DEST_BFC1: return NVS_FR_BFC1;
-      case NV40_VP_INST_DEST_FOGC: {
-	    int mask = shader->GetDestMask(shader, merged);
-	    switch (mask) {
-	    case SMASK_X: return NVS_FR_FOGCOORD;
-	    case SMASK_Y: return NVS_FR_CLIP0;
-	    case SMASK_Z: return NVS_FR_CLIP1;
-	    case SMASK_W: return NVS_FR_CLIP2;
-	    default:
-	       printf("more than 1 mask component set in FOGC writemask!\n");
-	       return NVS_FR_UNKNOWN;
-	    }
-	 }
-      case NV40_VP_INST_DEST_PSZ:
-	 {
-	    int mask = shader->GetDestMask(shader, merged);
-	    switch (mask) {
-	    case SMASK_X: return NVS_FR_POINTSZ;
-	    case SMASK_Y: return NVS_FR_CLIP3;
-	    case SMASK_Z: return NVS_FR_CLIP4;
-	    case SMASK_W: return NVS_FR_CLIP5;
-	    default:
-	       printf("more than 1 mask component set in PSZ writemask!\n");
-	       return NVS_FR_UNKNOWN;
-	    }
-	 }
-      case NV40_VP_INST_DEST_TC(0): return NVS_FR_TEXCOORD0;
-      case NV40_VP_INST_DEST_TC(1): return NVS_FR_TEXCOORD1;
-      case NV40_VP_INST_DEST_TC(2): return NVS_FR_TEXCOORD2;
-      case NV40_VP_INST_DEST_TC(3): return NVS_FR_TEXCOORD3;
-      case NV40_VP_INST_DEST_TC(4): return NVS_FR_TEXCOORD4;
-      case NV40_VP_INST_DEST_TC(5): return NVS_FR_TEXCOORD5;
-      case NV40_VP_INST_DEST_TC(6): return NVS_FR_TEXCOORD6;
-      case NV40_VP_INST_DEST_TC(7): return NVS_FR_TEXCOORD7;
-      default:
-	 return -1;
-      }
-   case NVS_FILE_ADDRESS:
-      /* Instructions that write address regs are encoded as if
-       * they would write temps.
-       */
-   case NVS_FILE_TEMP:
-      if (shader->GetOpcodeSlot(shader, merged))
-	 id = ((shader->inst[3] & NV40_VP_INST_SCA_DEST_TEMP_MASK)
-	       >> NV40_VP_INST_SCA_DEST_TEMP_SHIFT);
-      else
-	 id = ((shader->inst[0] & NV40_VP_INST_VEC_DEST_TEMP_MASK)
-	       >> NV40_VP_INST_VEC_DEST_TEMP_SHIFT);
-      return id;
-   default:
-      return -1;
-   }
-}
-
-static unsigned int
-NV40VPGetDestMask(nvsFunc * shader, int merged)
-{
-   unsigned int mask = 0;
-
-   if (shader->GetOpcodeSlot(shader, merged)) {
-      if (shader->inst[3] & NV40_VP_INST_SCA_WRITEMASK_X) mask |= SMASK_X;
-      if (shader->inst[3] & NV40_VP_INST_SCA_WRITEMASK_Y) mask |= SMASK_Y;
-      if (shader->inst[3] & NV40_VP_INST_SCA_WRITEMASK_Z) mask |= SMASK_Z;
-      if (shader->inst[3] & NV40_VP_INST_SCA_WRITEMASK_W) mask |= SMASK_W;
-   } else {
-      if (shader->inst[3] & NV40_VP_INST_VEC_WRITEMASK_X) mask |= SMASK_X;
-      if (shader->inst[3] & NV40_VP_INST_VEC_WRITEMASK_Y) mask |= SMASK_Y;
-      if (shader->inst[3] & NV40_VP_INST_VEC_WRITEMASK_Z) mask |= SMASK_Z;
-      if (shader->inst[3] & NV40_VP_INST_VEC_WRITEMASK_W) mask |= SMASK_W;
-   }
-
-   return mask;
-}
-
-static unsigned int
-NV40VPGetSourceHW(nvsFunc * shader, int merged, int pos)
-{
-   struct _op_xlat *opr;
-   unsigned int src;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr)
-      return -1;
-
-   switch (opr->srcpos[pos]) {
-   case 0:
-      src = ((shader->inst[1] & NV40_VP_INST_SRC0H_MASK)
-	     >> NV40_VP_INST_SRC0H_SHIFT)
-	 << NV40_VP_SRC0_HIGH_SHIFT;
-      src |= ((shader->inst[2] & NV40_VP_INST_SRC0L_MASK)
-	      >> NV40_VP_INST_SRC0L_SHIFT);
-      break;
-   case 1:
-      src = ((shader->inst[2] & NV40_VP_INST_SRC1_MASK)
-	     >> NV40_VP_INST_SRC1_SHIFT);
-      break;
-   case 2:
-      src = ((shader->inst[2] & NV40_VP_INST_SRC2H_MASK)
-	     >> NV40_VP_INST_SRC2H_SHIFT)
-	 << NV40_VP_SRC2_HIGH_SHIFT;
-      src |= ((shader->inst[3] & NV40_VP_INST_SRC2L_MASK)
-	      >> NV40_VP_INST_SRC2L_SHIFT);
-      break;
-   default:
-      src = -1;
-   }
-
-   return src;
-}
-
-static nvsRegFile
-NV40VPGetSourceFile(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-   struct _op_xlat *opr;
-   int file;
-
-   opr = shader->GetOPTXRec(shader, merged);
-   if (!opr || opr->srcpos[pos] == -1)
-      return -1;
-
-   switch (opr->srcpos[pos]) {
-   case SPOS_ADDRESS: return NVS_FILE_ADDRESS;
-   default:
-      src = shader->GetSourceHW(shader, merged, pos);
-      file = (src & NV40_VP_SRC_REG_TYPE_MASK) >> NV40_VP_SRC_REG_TYPE_SHIFT;
-
-      switch (file) {
-      case NV40_VP_SRC_REG_TYPE_TEMP : return NVS_FILE_TEMP;
-      case NV40_VP_SRC_REG_TYPE_INPUT: return NVS_FILE_ATTRIB;
-      case NV40_VP_SRC_REG_TYPE_CONST: return NVS_FILE_CONST;
-      default:
-	 return NVS_FILE_UNKNOWN;
-      }
-   }
-}
-
-static int
-NV40VPGetSourceID(nvsFunc * shader, int merged, int pos)
-{
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_ATTRIB:
-      switch ((shader->inst[1] & NV40_VP_INST_INPUT_SRC_MASK)
-	      >> NV40_VP_INST_INPUT_SRC_SHIFT) {
-      case NV40_VP_INST_IN_POS:		return NVS_FR_POSITION;
-      case NV40_VP_INST_IN_WEIGHT:	return NVS_FR_WEIGHT;
-      case NV40_VP_INST_IN_NORMAL:	return NVS_FR_NORMAL;
-      case NV40_VP_INST_IN_COL0:	return NVS_FR_COL0;
-      case NV40_VP_INST_IN_COL1:	return NVS_FR_COL1;
-      case NV40_VP_INST_IN_FOGC:	return NVS_FR_FOGCOORD;
-      case NV40_VP_INST_IN_TC(0):	return NVS_FR_TEXCOORD0;
-      case NV40_VP_INST_IN_TC(1):	return NVS_FR_TEXCOORD1;
-      case NV40_VP_INST_IN_TC(2):	return NVS_FR_TEXCOORD2;
-      case NV40_VP_INST_IN_TC(3):	return NVS_FR_TEXCOORD3;
-      case NV40_VP_INST_IN_TC(4):	return NVS_FR_TEXCOORD4;
-      case NV40_VP_INST_IN_TC(5):	return NVS_FR_TEXCOORD5;
-      case NV40_VP_INST_IN_TC(6):	return NVS_FR_TEXCOORD6;
-      case NV40_VP_INST_IN_TC(7):	return NVS_FR_TEXCOORD7;
-      default:
-	 return -1;
-      }
-      break;
-   case NVS_FILE_CONST:
-      return ((shader->inst[1] & NV40_VP_INST_CONST_SRC_MASK)
-	      >> NV40_VP_INST_CONST_SRC_SHIFT);
-   case NVS_FILE_TEMP:
-      {
-	 unsigned int src;
-
-	 src = shader->GetSourceHW(shader, merged, pos);
-	 return ((src & NV40_VP_SRC_TEMP_SRC_MASK) >>
-		 NV40_VP_SRC_TEMP_SRC_SHIFT);
-      }
-   default:
-      return -1;
-   }
-}
-
-static int
-NV40VPGetSourceNegate(nvsFunc * shader, int merged, int pos)
-{
-   unsigned int src;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-
-   if (src == -1)
-      return -1;
-   return ((src & NV40_VP_SRC_NEGATE) ? 1 : 0);
-}
-
-static void
-NV40VPGetSourceSwizzle(nvsFunc * shader, int merged, int pos, nvsSwzComp *swz)
-{
-   unsigned int src;
-   int swzbits;
-
-   src = shader->GetSourceHW(shader, merged, pos);
-   swzbits = (src & NV40_VP_SRC_SWZ_ALL_MASK) >> NV40_VP_SRC_SWZ_ALL_SHIFT;
-   NV20VPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV40VPGetSourceIndexed(nvsFunc * shader, int merged, int pos)
-{
-   switch (shader->GetSourceFile(shader, merged, pos)) {
-   case NVS_FILE_ATTRIB:
-      return ((shader->inst[0] & NV40_VP_INST_INDEX_INPUT) ? 1 : 0);
-   case NVS_FILE_CONST:
-      return ((shader->inst[3] & NV40_VP_INST_INDEX_CONST) ? 1 : 0);
-   default:
-      return 0;
-   }
-}
-
-static nvsSwzComp
-NV40VPGetAddressRegSwizzle(nvsFunc * shader)
-{
-   nvsSwzComp swz;
-
-   swz = NV20VP_TX_SWIZZLE[(shader->inst[0] & NV40_VP_INST_ADDR_SWZ_MASK)
-			   >> NV40_VP_INST_ADDR_SWZ_SHIFT];
-   return swz;
-}
-
-static int
-NV40VPSupportsConditional(nvsFunc * shader)
-{
-   /*FIXME: Is this true of all ops? */
-   return 1;
-}
-
-static int
-NV40VPGetConditionUpdate(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV40_VP_INST_COND_UPDATE_ENABLE) ? 1 : 0);
-}
-
-static int
-NV40VPGetConditionTest(nvsFunc * shader)
-{
-   int op;
-
-   /* The condition test is unconditionally enabled on some
-    * instructions. ie: the condition test bit does *NOT* have
-    * to be set.
-    *
-    * FIXME: check other relevant ops for this situation.
-    */
-   op = shader->GetOpcodeHW(shader, 1);
-   switch (op) {
-   case NV40_VP_INST_OP_BRA:
-      return 1;
-   default:
-      return ((shader->inst[0] & NV40_VP_INST_COND_TEST_ENABLE) ? 1 : 0);
-   }
-}
-
-static nvsCond
-NV40VPGetCondition(nvsFunc * shader)
-{
-   int cond;
-
-   cond = ((shader->inst[0] & NV40_VP_INST_COND_MASK)
-	   >> NV40_VP_INST_COND_SHIFT);
-
-   switch (cond) {
-   case NV40_VP_INST_COND_FL: return NVS_COND_FL;
-   case NV40_VP_INST_COND_LT: return NVS_COND_LT;
-   case NV40_VP_INST_COND_EQ: return NVS_COND_EQ;
-   case NV40_VP_INST_COND_LE: return NVS_COND_LE;
-   case NV40_VP_INST_COND_GT: return NVS_COND_GT;
-   case NV40_VP_INST_COND_NE: return NVS_COND_NE;
-   case NV40_VP_INST_COND_GE: return NVS_COND_GE;
-   case NV40_VP_INST_COND_TR: return NVS_COND_TR;
-   default:
-      return NVS_COND_UNKNOWN;
-   }
-}
-
-static void
-NV40VPGetCondRegSwizzle(nvsFunc * shader, nvsSwzComp *swz)
-{
-   int swzbits;
-
-   swzbits = (shader->inst[0] & NV40_VP_INST_COND_SWZ_ALL_MASK)
-      >> NV40_VP_INST_COND_SWZ_ALL_SHIFT;
-   NV20VPTXSwizzle(swzbits, swz);
-}
-
-static int
-NV40VPGetCondRegID(nvsFunc * shader)
-{
-   return ((shader->inst[0] & NV40_VP_INST_COND_REG_SELECT_1) ? 1 : 0);
-}
-
-static int
-NV40VPGetBranch(nvsFunc * shader)
-{
-   int addr;
-
-   addr = ((shader->inst[2] & NV40_VP_INST_IADDRH_MASK)
-	   >> NV40_VP_INST_IADDRH_SHIFT) << 3;
-   addr |= ((shader->inst[3] & NV40_VP_INST_IADDRL_MASK)
-	    >> NV40_VP_INST_IADDRL_SHIFT);
-   return addr;
-}
-
-void
-NV40VPInitShaderFuncs(nvsFunc * shader)
-{
-   /* Inherit NV30 VP code, we share some of it */
-   NV30VPInitShaderFuncs(shader);
-
-   /* Limits */
-   shader->MaxInst	= 4096;
-   shader->MaxAttrib	= 16;
-   shader->MaxTemp	= 32;
-   shader->MaxAddress	= 2;
-   shader->MaxConst	= 256;
-   shader->caps		= SCAP_SRC_ABS;
-
-   /* Add extra opcodes for NV40+ */
-//      MOD_OPCODE(NVVP_TX_VOP, NV40_VP_INST_OP_TXWHAT, NVS_OP_TEX  ,  0,  4, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV40_VP_INST_OP_PUSHA, NVS_OP_PUSHA,  3, -1, -1);
-   MOD_OPCODE(NVVP_TX_SOP, NV40_VP_INST_OP_POPA , NVS_OP_POPA , -1, -1, -1);
-
-   shader->InitInstruction	= NV40VPInitInstruction;
-   shader->SupportsOpcode	= NV40VPSupportsOpcode;
-   shader->SetOpcode		= NV40VPSetOpcode;
-   shader->SetCCUpdate		= NV40VPSetCCUpdate;
-   shader->SetCondition		= NV40VPSetCondition;
-   shader->SetResult		= NV40VPSetResult;
-   shader->SetSource		= NV40VPSetSource;
-   shader->SetLastInst		= NV40VPSetLastInst;
-   shader->SetBranchTarget	= NV40VPSetBranchTarget;
-
-   shader->HasMergedInst	= NV40VPHasMergedInst;
-   shader->GetOpcodeHW		= NV40VPGetOpcodeHW;
-
-   shader->GetDestFile		= NV40VPGetDestFile;
-   shader->GetDestID		= NV40VPGetDestID;
-   shader->GetDestMask		= NV40VPGetDestMask;
-
-   shader->GetSourceHW		= NV40VPGetSourceHW;
-   shader->GetSourceFile	= NV40VPGetSourceFile;
-   shader->GetSourceID		= NV40VPGetSourceID;
-   shader->GetSourceNegate	= NV40VPGetSourceNegate;
-   shader->GetSourceSwizzle	= NV40VPGetSourceSwizzle;
-   shader->GetSourceIndexed	= NV40VPGetSourceIndexed;
-
-   shader->GetRelAddressSwizzle	= NV40VPGetAddressRegSwizzle;
-
-   shader->SupportsConditional	= NV40VPSupportsConditional;
-   shader->GetConditionUpdate	= NV40VPGetConditionUpdate;
-   shader->GetConditionTest	= NV40VPGetConditionTest;
-   shader->GetCondition		= NV40VPGetCondition;
-   shader->GetCondRegSwizzle	= NV40VPGetCondRegSwizzle;
-   shader->GetCondRegID		= NV40VPGetCondRegID;
-
-   shader->GetBranch		= NV40VPGetBranch;
-}
diff --git a/src/mesa/drivers/dri/nouveau/nv50_state.c b/src/mesa/drivers/dri/nouveau/nv50_state.c
deleted file mode 100644
index 818e268615c..00000000000
--- a/src/mesa/drivers/dri/nouveau/nv50_state.c
+++ /dev/null
@@ -1,641 +0,0 @@
-/**************************************************************************
-
-Copyright 2006 Nouveau
-All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining a
-copy of this software and associated documentation files (the "Software"),
-to deal in the Software without restriction, including without limitation
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-#include "nouveau_context.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_state.h"
-
-#include "tnl/t_pipeline.h"
-
-#include "mtypes.h"
-#include "colormac.h"
-
-static void nv50AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte ubRef;
-	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF, 2);
-	OUT_RING_CACHE(ubRef);
-	OUT_RING_CACHE(func);
-}
-
-static void nv50BlendColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); 
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_R, 4);
-	OUT_RING_CACHEf(color[0]);
-	OUT_RING_CACHEf(color[1]);
-	OUT_RING_CACHEf(color[2]);
-	OUT_RING_CACHEf(color[3]);
-}
-
-static void nv50BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB, 1);
-	OUT_RING_CACHE(modeRGB);
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA, 1);
-	OUT_RING_CACHE(modeA);
-}
-
-
-static void nv50BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
-		GLenum sfactorA, GLenum dfactorA)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB, 2);
-	OUT_RING_CACHE(sfactorRGB);	/* FIXME, sometimes has |0x4000 */
-	OUT_RING_CACHE(dfactorRGB);	/* FIXME, sometimes has |0x4000 */
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA, 2);
-	OUT_RING_CACHE(sfactorA);	/* FIXME, sometimes has |0x4000 */
-	OUT_RING_CACHE(dfactorA);	/* FIXME, sometimes has |0x4000 */
-}
-
-static void nv50Clear(GLcontext *ctx, GLbitfield mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	GLuint hw_bufs = 0;
-
-	if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT))
-		hw_bufs |= 0x3c;
-	if (mask & (BUFFER_BIT_STENCIL))
-		hw_bufs |= 0x02;
-	if (mask & (BUFFER_BIT_DEPTH))
-		hw_bufs |= 0x01;
-
-	if (hw_bufs) {
-		BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS, 1);
-		OUT_RING(hw_bufs);
-	}
-}
-
-static void nv50ClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_R, 4);
-	OUT_RING_CACHEf(color[0]);
-	OUT_RING_CACHEf(color[1]);
-	OUT_RING_CACHEf(color[2]);
-	OUT_RING_CACHEf(color[3]);
-}
-
-static void nv50ClearDepth(GLcontext *ctx, GLclampd d)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH, 1);
-	OUT_RING_CACHEf(d);
-}
-
-/* we're don't support indexed buffers
-   void (*ClearIndex)(GLcontext *ctx, GLuint index)
- */
-
-static void nv50ClearStencil(GLcontext *ctx, GLint s)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL, 1);
-	OUT_RING_CACHE(s);
-}
-
-static void nv50ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
-{
-	/* Only using shaders */
-}
-
-static void nv50ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
-		GLboolean bmask, GLboolean amask )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	int i;
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_COLOR_MASK(0), 8);
-	for (i=0; i<8; i++) {
-		OUT_RING_CACHE(((amask && 0x01) << 12) | ((bmask && 0x01) << 8) | ((gmask && 0x01)<< 4) | ((rmask && 0x01) << 0));
-	}
-}
-
-static void nv50ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	// TODO I need love
-}
-
-static void nv50CullFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CULL_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv50FrontFace(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_FRONT_FACE, 1);
-	OUT_RING_CACHE(mode);
-}
-
-static void nv50DepthFunc(GLcontext *ctx, GLenum func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC, 1);
-	OUT_RING_CACHE(func);
-}
-
-static void nv50DepthMask(GLcontext *ctx, GLboolean flag)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE, 1);
-	OUT_RING_CACHE(flag);
-}
-
-static void nv50DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
-	OUT_RING_CACHEf(nearval);
-	OUT_RING_CACHEf(farval);
-}
-
-/** Specify the current buffer for writing */
-//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
-/** Specify the buffers for writing for fragment programs*/
-//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
-
-static void nv50Enable(GLcontext *ctx, GLenum cap, GLboolean state)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	switch(cap)
-	{
-		case GL_ALPHA_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_AUTO_NORMAL:
-//		case GL_BLEND:
-//		case GL_CLIP_PLANE0:
-//		case GL_CLIP_PLANE1:
-//		case GL_CLIP_PLANE2:
-//		case GL_CLIP_PLANE3:
-//		case GL_CLIP_PLANE4:
-//		case GL_CLIP_PLANE5:
-		case GL_COLOR_LOGIC_OP:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_COLOR_MATERIAL:
-//		case GL_COLOR_SUM_EXT:
-//		case GL_COLOR_TABLE:
-//		case GL_CONVOLUTION_1D:
-//		case GL_CONVOLUTION_2D:
-		case GL_CULL_FACE:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_DEPTH_TEST:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_DITHER:
-//		case GL_FOG:
-//		case GL_HISTOGRAM:
-//		case GL_INDEX_LOGIC_OP:
-//		case GL_LIGHT0:
-//		case GL_LIGHT1:
-//		case GL_LIGHT2:
-//		case GL_LIGHT3:
-//		case GL_LIGHT4:
-//		case GL_LIGHT5:
-//		case GL_LIGHT6:
-//		case GL_LIGHT7:
-//		case GL_LIGHTING:
-		case GL_LINE_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_LINE_STIPPLE:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_MAP1_COLOR_4:
-//		case GL_MAP1_INDEX:
-//		case GL_MAP1_NORMAL:
-//		case GL_MAP1_TEXTURE_COORD_1:
-//		case GL_MAP1_TEXTURE_COORD_2:
-//		case GL_MAP1_TEXTURE_COORD_3:
-//		case GL_MAP1_TEXTURE_COORD_4:
-//		case GL_MAP1_VERTEX_3:
-//		case GL_MAP1_VERTEX_4:
-//		case GL_MAP2_COLOR_4:
-//		case GL_MAP2_INDEX:
-//		case GL_MAP2_NORMAL:
-//		case GL_MAP2_TEXTURE_COORD_1:
-//		case GL_MAP2_TEXTURE_COORD_2:
-//		case GL_MAP2_TEXTURE_COORD_3:
-//		case GL_MAP2_TEXTURE_COORD_4:
-//		case GL_MAP2_VERTEX_3:
-//		case GL_MAP2_VERTEX_4:
-//		case GL_MINMAX:
-//		case GL_NORMALIZE:
-//		case GL_POINT_SMOOTH:
-		case GL_POLYGON_OFFSET_POINT:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_LINE:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_OFFSET_FILL:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_SMOOTH:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-		case GL_POLYGON_STIPPLE:
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
-//		case GL_POST_CONVOLUTION_COLOR_TABLE:
-//		case GL_RESCALE_NORMAL:
-		case GL_SCISSOR_TEST:
-			/* No enable bit, nv50Scissor will adjust to max range */
-			ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			      			 ctx->Scissor.Width, ctx->Scissor.Height);
-			break;
-//		case GL_SEPARABLE_2D:
-		case GL_STENCIL_TEST:
-			// TODO BACK and FRONT ?
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE, 1);
-			OUT_RING_CACHE(state);
-			break;
-//		case GL_TEXTURE_GEN_Q:
-//		case GL_TEXTURE_GEN_R:
-//		case GL_TEXTURE_GEN_S:
-//		case GL_TEXTURE_GEN_T:
-//		case GL_TEXTURE_1D:
-//		case GL_TEXTURE_2D:
-//		case GL_TEXTURE_3D:
-	}
-}
-
-static void nv50Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/* Only using shaders */
-}
-   
-static void nv50Hint(GLcontext *ctx, GLenum target, GLenum mode)
-{
-	// TODO I need love (fog and line_smooth hints)
-}
-
-// void (*IndexMask)(GLcontext *ctx, GLuint mask);
-
-static void nv50Lightfv(GLcontext *ctx, GLenum light, GLenum pname, const GLfloat *params )
-{
-	/* Only with shaders */
-}
-
-/** Set the lighting model parameters */
-void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params);
-
-
-static void nv50LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN, 1);
-	OUT_RING_CACHE((pattern << 8) | factor);
-}
-
-static void nv50LineWidth(GLcontext *ctx, GLfloat width)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_WIDTH, 1);
-	OUT_RING_CACHEf(width);
-}
-
-static void nv50LogicOpcode(GLcontext *ctx, GLenum opcode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP, 1);
-	OUT_RING_CACHE(opcode);
-}
-
-static void nv50PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
-{
-	/*TODO: not sure what goes here. */
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	
-}
-
-/** Specify the diameter of rasterized points */
-static void nv50PointSize(GLcontext *ctx, GLfloat size)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
-	OUT_RING_CACHEf(size);
-}
-
-/** Select a polygon rasterization mode */
-static void nv50PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT, 1);
-		OUT_RING_CACHE(mode);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK, 1);
-		OUT_RING_CACHE(mode);
-	}
-}
-
-/** Set the scale and units used to calculate depth values */
-static void nv50PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
- 
-        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 1);
-        OUT_RING_CACHEf(factor);
-        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS, 1);
-        OUT_RING_CACHEf(units);
-}
-
-/** Set the polygon stippling pattern */
-static void nv50PolygonStipple(GLcontext *ctx, const GLubyte *mask )
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(0), 32);
-        OUT_RING_CACHEp(mask, 32);
-}
-
-/* Specifies the current buffer for reading */
-void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
-/** Set rasterization mode */
-void (*RenderMode)(GLcontext *ctx, GLenum mode );
-
-/** Define the scissor box */
-static void nv50Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	/* There's no scissor enable bit, so adjust the scissor to cover the
-	 * maximum draw buffer bounds
-	 */
-	if (!ctx->Scissor.Enabled) {
-	   x = y = 0;
-	   w = h = 8191;
-	} else {
-	   x += nmesa->drawX;
-	   y += nmesa->drawY;
-	}
-
-        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS, 2);
-        OUT_RING_CACHE(((w) << 16) | x);
-        OUT_RING_CACHE(((h) << 16) | y);
-}
-
-/** Select flat or smooth shading */
-static void nv50ShadeModel(GLcontext *ctx, GLenum mode)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SHADE_MODEL, 1);
-	OUT_RING_CACHE(mode);
-}
-
-/** OpenGL 2.0 two-sided StencilFunc */
-static void nv50StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
-		GLint ref, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC, 1);
-		OUT_RING_CACHE(func);
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF, 1);
-		OUT_RING_CACHE(ref);
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC, 2);
-		OUT_RING_CACHE(func);
-		OUT_RING_CACHE(ref);
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-}
-
-/** OpenGL 2.0 two-sided StencilMask */
-static void nv50StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK, 1);
-		OUT_RING_CACHE(mask);
-	}
-}
-
-/** OpenGL 2.0 two-sided StencilOp */
-static void nv50StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
-		GLenum zfail, GLenum zpass)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL, 3);
-		OUT_RING_CACHE(fail);
-		OUT_RING_CACHE(zfail);
-		OUT_RING_CACHE(zpass);
-	}
-	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
-		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL, 3);
-		OUT_RING_CACHE(fail);
-		OUT_RING_CACHE(zfail);
-		OUT_RING_CACHE(zpass);
-	}
-}
-
-/** Control the generation of texture coordinates */
-void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
-		const GLfloat *params);
-/** Set texture environment parameters */
-void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
-		const GLfloat *param);
-/** Set texture parameters */
-void (*TexParameter)(GLcontext *ctx, GLenum target,
-		struct gl_texture_object *texObj,
-		GLenum pname, const GLfloat *params);
-
-static void nv50TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
-{
-	/* Only with shaders */
-}
-
-static void nv50WindowMoved(nouveauContextPtr nmesa)
-{
-	GLcontext *ctx = nmesa->glCtx;
-	GLfloat *v = nmesa->viewport.m;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-	GLuint x = ctx->Viewport.X + nmesa->drawX;
-	GLuint y = ctx->Viewport.Y + nmesa->drawY;
-	int i;
-
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 2);
-        OUT_RING_CACHE((8191 << 16) | 0);
-        OUT_RING_CACHE((8191 << 16) | 0);
-	for (i=1; i<8; i++) {
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 2);
-        	OUT_RING_CACHE(0);
-	        OUT_RING_CACHE(0);
-	}
-
-	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	      		    ctx->Scissor.Width, ctx->Scissor.Height);
-}
-
-static GLboolean nv50InitCard(nouveauContextPtr nmesa)
-{
-	int i,j;
-
-	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
-
-	BEGIN_RING_SIZE(NvSub3D, 0x1558, 1);
-	OUT_RING(1);
-
-	BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SET_OBJECT_1(0), 8);
-	for (i=0; i<8; i++) {
-		OUT_RING(NvDmaFB);
-	}
-
-	BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SET_OBJECT_0(0), 12);
-	for (i=0; i<12; i++) {
-		OUT_RING(NvDmaFB);
-	}
-
-	BEGIN_RING_SIZE(NvSub3D, 0x121c, 1);
-	OUT_RING(1);
-
-	for (i=0; i<8; i++) {
-		BEGIN_RING_SIZE(NvSub3D, 0x0200 + (i*0x20), 5);
-		for (j=0; j<5; j++) {
-			OUT_RING(0);
-		}
-	}
-
-	BEGIN_RING_SIZE(NvSub3D, 0x0fe0, 5);
-	OUT_RING(0);
-	OUT_RING(0);
-	OUT_RING(0x16);
-	OUT_RING(0);
-	OUT_RING(0);
-
-	return GL_FALSE;
-}
-
-static GLboolean nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
-      				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
-{
-	return GL_FALSE;
-}
-
-void nv50InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
-{
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-	func->AlphaFunc			= nv50AlphaFunc;
-	func->BlendColor		= nv50BlendColor;
-	func->BlendEquationSeparate	= nv50BlendEquationSeparate;
-	func->BlendFuncSeparate		= nv50BlendFuncSeparate;
-	func->Clear			= nv50Clear;
-	func->ClearColor		= nv50ClearColor;
-	func->ClearDepth		= nv50ClearDepth;
-	func->ClearStencil		= nv50ClearStencil;
-	func->ClipPlane			= nv50ClipPlane;
-	func->ColorMask			= nv50ColorMask;
-	func->ColorMaterial		= nv50ColorMaterial;
-	func->CullFace			= nv50CullFace;
-	func->FrontFace			= nv50FrontFace;
-	func->DepthFunc			= nv50DepthFunc;
-	func->DepthMask			= nv50DepthMask;
-	func->DepthRange		= nv50DepthRange;
-	func->Enable			= nv50Enable;
-	func->Fogfv			= nv50Fogfv;
-	func->Hint			= nv50Hint;
-	func->Lightfv			= nv50Lightfv;
-/*	func->LightModelfv		= nv50LightModelfv;	 */
-	func->LineStipple		= nv50LineStipple;
-	func->LineWidth			= nv50LineWidth;
-	func->LogicOpcode		= nv50LogicOpcode;
-	func->PointParameterfv		= nv50PointParameterfv;
-	func->PointSize			= nv50PointSize;
-	func->PolygonMode		= nv50PolygonMode;
-	func->PolygonOffset		= nv50PolygonOffset;
-	func->PolygonStipple		= nv50PolygonStipple;
-/*	func->ReadBuffer		= nv50ReadBuffer;	*/
-/*	func->RenderMode		= nv50RenderMode;	*/
-	func->Scissor			= nv50Scissor;
-	func->ShadeModel		= nv50ShadeModel;
-	func->StencilFuncSeparate	= nv50StencilFuncSeparate;
-	func->StencilMaskSeparate	= nv50StencilMaskSeparate;
-	func->StencilOpSeparate		= nv50StencilOpSeparate;
-/*	func->TexGen			= nv50TexGen;		*/
-/*	func->TexParameter		= nv50TexParameter;	*/
-	func->TextureMatrix		= nv50TextureMatrix;
-
-	nmesa->hw_func.InitCard		= nv50InitCard;
-	nmesa->hw_func.BindBuffers	= nv50BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv50WindowMoved;
-}
diff --git a/src/mesa/drivers/dri/r128/r128_context.c b/src/mesa/drivers/dri/r128/r128_context.c
index dfc89a2da76..535a98cc01f 100644
--- a/src/mesa/drivers/dri/r128/r128_context.c
+++ b/src/mesa/drivers/dri/r128/r128_context.c
@@ -32,12 +32,12 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -253,7 +253,7 @@ GLboolean r128CreateContext( const __GLcontextModes *glVisual,
    _tnl_allow_vertex_fog( ctx, GL_TRUE );
 
    driInitExtensions( ctx, card_extensions, GL_TRUE );
-   if (sPriv->drmMinor >= 4)
+   if (sPriv->drm_version.minor >= 4)
       _mesa_enable_extension( ctx, "GL_MESA_ycbcr_texture" );
 
    r128InitTriFuncs( ctx );
@@ -261,9 +261,6 @@ GLboolean r128CreateContext( const __GLcontextModes *glVisual,
    r128DDInitSpanFuncs( ctx );
    r128DDInitState( rmesa );
 
-   rmesa->vblank_flags = (rmesa->r128Screen->irq != 0)
-       ? driGetDefaultVBlankFlags(&rmesa->optionCache) : VBLANK_FLAG_NO_IRQ;
-
    driContextPriv->driverPrivate = (void *)rmesa;
 
 #if DO_DEBUG
@@ -346,8 +343,13 @@ r128MakeCurrent( __DRIcontextPrivate *driContextPriv,
 	 newR128Ctx->dirty = R128_UPLOAD_ALL;
       }
 
-      driDrawableInitVBlank( driDrawPriv, newR128Ctx->vblank_flags,
-			     &newR128Ctx->vbl_seq );
+      if (driDrawPriv->swap_interval == (unsigned)-1) {
+	 driDrawPriv->vblFlags = (newR128Ctx->r128Screen->irq != 0)
+	    ? driGetDefaultVBlankFlags(&newR128Ctx->optionCache)
+	    : VBLANK_FLAG_NO_IRQ;
+
+	 driDrawableInitVBlank( driDrawPriv );
+      }
       newR128Ctx->driDrawable = driDrawPriv;
 
       _mesa_make_current( newR128Ctx->glCtx,
diff --git a/src/mesa/drivers/dri/r128/r128_context.h b/src/mesa/drivers/dri/r128/r128_context.h
index 1d10cb0b4fe..0e10209a6ad 100644
--- a/src/mesa/drivers/dri/r128/r128_context.h
+++ b/src/mesa/drivers/dri/r128/r128_context.h
@@ -39,7 +39,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "drm.h"
 #include "r128_drm.h"
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "tnl/t_vertex.h"
 
 #include "r128_reg.h"
@@ -209,11 +209,6 @@ struct r128_context {
    GLuint c_textureBytes;
    GLuint c_vertexBuffers;
 
-   /* VBI
-    */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
    /* Configuration cache
     */
    driOptionCache optionCache;
diff --git a/src/mesa/drivers/dri/r128/r128_dd.c b/src/mesa/drivers/dri/r128/r128_dd.c
index d8e1c70ab77..dfe47f2dd64 100644
--- a/src/mesa/drivers/dri/r128/r128_dd.c
+++ b/src/mesa/drivers/dri/r128/r128_dd.c
@@ -38,8 +38,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_dd.h"
 #include "swrast/swrast.h"
 
-#include "context.h"
-#include "framebuffer.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
 
 #include "utils.h"
 
diff --git a/src/mesa/drivers/dri/r128/r128_ioctl.c b/src/mesa/drivers/dri/r128/r128_ioctl.c
index dcb5d10459e..84ac3d9f797 100644
--- a/src/mesa/drivers/dri/r128/r128_ioctl.c
+++ b/src/mesa/drivers/dri/r128/r128_ioctl.c
@@ -36,8 +36,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_context.h"
 #include "r128_state.h"
 #include "r128_ioctl.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/imports.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 
@@ -248,7 +248,7 @@ static int r128WaitForFrameCompletion( r128ContextPtr rmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void r128CopyBuffer( const __DRIdrawablePrivate *dPriv )
+void r128CopyBuffer( __DRIdrawablePrivate *dPriv )
 {
    r128ContextPtr rmesa;
    GLint nbox, i, ret;
@@ -281,7 +281,7 @@ void r128CopyBuffer( const __DRIdrawablePrivate *dPriv )
    }
 
    UNLOCK_HARDWARE( rmesa );
-   driWaitForVBlank( dPriv, &rmesa->vbl_seq, rmesa->vblank_flags, &missed_target );
+   driWaitForVBlank( dPriv, &missed_target );
    LOCK_HARDWARE( rmesa );
 
    nbox = dPriv->numClipRects;	/* must be in locked region */
@@ -327,7 +327,7 @@ void r128CopyBuffer( const __DRIdrawablePrivate *dPriv )
 #endif
 }
 
-void r128PageFlip( const __DRIdrawablePrivate *dPriv )
+void r128PageFlip( __DRIdrawablePrivate *dPriv )
 {
    r128ContextPtr rmesa;
    GLint ret;
@@ -358,7 +358,7 @@ void r128PageFlip( const __DRIdrawablePrivate *dPriv )
    }
 
    UNLOCK_HARDWARE( rmesa );
-   driWaitForVBlank( dPriv, &rmesa->vbl_seq, rmesa->vblank_flags, &missed_target );
+   driWaitForVBlank( dPriv, &missed_target );
    LOCK_HARDWARE( rmesa );
 
    /* The kernel will have been initialized to perform page flipping
diff --git a/src/mesa/drivers/dri/r128/r128_ioctl.h b/src/mesa/drivers/dri/r128/r128_ioctl.h
index 52ae3a293f8..4b0c9cdc7fe 100644
--- a/src/mesa/drivers/dri/r128/r128_ioctl.h
+++ b/src/mesa/drivers/dri/r128/r128_ioctl.h
@@ -44,7 +44,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 extern drmBufPtr r128GetBufferLocked( r128ContextPtr rmesa );
 extern void r128FlushVerticesLocked( r128ContextPtr rmesa );
 
-static __inline void *r128AllocDmaLow( r128ContextPtr rmesa, int count,
+static INLINE void *r128AllocDmaLow( r128ContextPtr rmesa, int count,
 				       int vert_size )
 {
    uint32_t *head;
@@ -85,8 +85,8 @@ extern void r128ReadDepthSpanLocked( r128ContextPtr rmesa,
 extern void r128ReadDepthPixelsLocked( r128ContextPtr rmesa, GLuint n,
 				       const GLint x[], const GLint y[] );
 
-extern void r128CopyBuffer( const __DRIdrawablePrivate *dPriv );
-extern void r128PageFlip( const __DRIdrawablePrivate *dPriv );
+extern void r128CopyBuffer( __DRIdrawablePrivate *dPriv );
+extern void r128PageFlip( __DRIdrawablePrivate *dPriv );
 void r128WaitForVBlank( r128ContextPtr rmesa );
 
 extern void r128WaitForIdleLocked( r128ContextPtr rmesa );
diff --git a/src/mesa/drivers/dri/r128/r128_lock.c b/src/mesa/drivers/dri/r128/r128_lock.c
index 3478e12ad07..81488a27424 100644
--- a/src/mesa/drivers/dri/r128/r128_lock.c
+++ b/src/mesa/drivers/dri/r128/r128_lock.c
@@ -58,7 +58,7 @@ r128UpdatePageFlipping( r128ContextPtr rmesa )
    rmesa->new_state |= R128_NEW_WINDOW;
 }
 
-/* Update the hardware state.  This is called if another context has
+/* Update the hardware state.  This is called if another main/context.has
  * grabbed the hardware lock, which includes the X server.  This
  * function also updates the driver's window state after the X server
  * moves, resizes or restacks a window -- the change will be reflected
diff --git a/src/mesa/drivers/dri/r128/r128_screen.c b/src/mesa/drivers/dri/r128/r128_screen.c
index 1e9616272f0..cb3a147dbac 100644
--- a/src/mesa/drivers/dri/r128/r128_screen.c
+++ b/src/mesa/drivers/dri/r128/r128_screen.c
@@ -39,10 +39,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_span.h"
 #include "r128_tris.h"
 
-#include "context.h"
-#include "imports.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "utils.h"
 #include "vblank.h"
@@ -97,9 +97,7 @@ r128CreateScreen( __DRIscreenPrivate *sPriv )
 {
    r128ScreenPtr r128Screen;
    R128DRIPtr r128DRIPriv = (R128DRIPtr)sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
+   int i;
 
    if (sPriv->devPrivSize != sizeof(R128DRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(R128DRIRec) does not match passed size from device driver\n");
@@ -120,7 +118,7 @@ r128CreateScreen( __DRIscreenPrivate *sPriv )
    r128Screen->IsPCI = r128DRIPriv->IsPCI;
    r128Screen->sarea_priv_offset = r128DRIPriv->sarea_priv_offset;
    
-   if (sPriv->drmMinor >= 3) {
+   if (sPriv->drm_version.minor >= 3) {
       drm_r128_getparam_t gp;
       int ret;
 
@@ -225,15 +223,14 @@ r128CreateScreen( __DRIscreenPrivate *sPriv )
 
    r128Screen->driScreen = sPriv;
 
-   if ( glx_enable_extension != NULL ) {
-      if ( r128Screen->irq != 0 ) {
-	 (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-	 (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-	 (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      }
-
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
+   i = 0;
+   r128Screen->extensions[i++] = &driFrameTrackingExtension.base;
+   if ( r128Screen->irq != 0 ) {
+       r128Screen->extensions[i++] = &driSwapControlExtension.base;
+       r128Screen->extensions[i++] = &driMediaStreamCounterExtension.base;
    }
+   r128Screen->extensions[i++] = NULL;
+   sPriv->extensions = r128Screen->extensions;
 
    return r128Screen;
 }
@@ -401,37 +398,18 @@ r128InitDriver( __DRIscreenPrivate *sPriv )
    return GL_TRUE;
 }
 
-
-static struct __DriverAPIRec r128API = {
-   .InitDriver      = r128InitDriver,
-   .DestroyScreen   = r128DestroyScreen,
-   .CreateContext   = r128CreateContext,
-   .DestroyContext  = r128DestroyContext,
-   .CreateBuffer    = r128CreateBuffer,
-   .DestroyBuffer   = r128DestroyBuffer,
-   .SwapBuffers     = r128SwapBuffers,
-   .MakeCurrent     = r128MakeCurrent,
-   .UnbindContext   = r128UnbindContext,
-   .GetSwapInfo     = NULL,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL
-
-};
-
-
-static __GLcontextModes *
-r128FillInModes( unsigned pixel_bits, unsigned depth_bits,
+static const __DRIconfig **
+r128FillInModes( __DRIscreenPrivate *psp,
+		 unsigned pixel_bits, unsigned depth_bits,
 		 unsigned stencil_bits, GLboolean have_back_buffer )
 {
-    __GLcontextModes * modes;
+    __DRIconfig **configs;
     __GLcontextModes * m;
-    unsigned num_modes;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     GLenum fb_format;
     GLenum fb_type;
+    int i;
 
     /* Right now GLX_SWAP_COPY_OML isn't supported, but it would be easy
      * enough to add support.  Basically, if a context is created with an
@@ -459,8 +437,6 @@ r128FillInModes( unsigned pixel_bits, unsigned depth_bits,
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -470,97 +446,85 @@ r128FillInModes( unsigned pixel_bits, unsigned depth_bits,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
-
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
+   configs = driCreateConfigs(fb_format, fb_type,
+			      depth_bits_array, stencil_bits_array,
+			      depth_buffer_factor, back_buffer_modes,
+			      back_buffer_factor);
+   if (configs == NULL) {
+    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
+   }
 
-    /* Mark the visual as slow if there are "fake" stencil bits.
-     */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-	if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	    m->visualRating = GLX_SLOW_CONFIG;
-	}
-    }
+   /* Mark the visual as slow if there are "fake" stencil bits.
+    */
+   for (i = 0; configs[i]; i++) {
+      m = &configs[i]->modes;
+      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+         m->visualRating = GLX_SLOW_CONFIG;
+      }
+   }
 
-    return modes;
+   return (const __DRIconfig **) configs;
 }
 
 
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+static const __DRIconfig **
+r128InitScreen(__DRIscreenPrivate *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 4, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 2, 2, 0 };
-
-
-   dri_interface = interface;
+   R128DRIPtr dri_priv = (R128DRIPtr) psp->pDevPriv;
 
    if ( ! driCheckDriDdxDrmVersions2( "Rage128",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
-      
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &r128API);
-   if ( psp != NULL ) {
-      R128DRIPtr dri_priv = (R128DRIPtr) psp->pDevPriv;
-      *driver_modes = r128FillInModes( dri_priv->bpp,
-				       (dri_priv->bpp == 16) ? 16 : 24,
-				       (dri_priv->bpp == 16) ? 0  : 8,
-				       (dri_priv->backOffset != dri_priv->depthOffset) );
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
 
-   return (void *) psp;
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+   if (!r128InitDriver(psp))
+       return NULL;
+
+   return r128FillInModes( psp,
+			   dri_priv->bpp,
+			   (dri_priv->bpp == 16) ? 16 : 24,
+			   (dri_priv->bpp == 16) ? 0  : 8,
+			   (dri_priv->backOffset != dri_priv->depthOffset) );
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = r128InitScreen,
+   .DestroyScreen   = r128DestroyScreen,
+   .CreateContext   = r128CreateContext,
+   .DestroyContext  = r128DestroyContext,
+   .CreateBuffer    = r128CreateBuffer,
+   .DestroyBuffer   = r128DestroyBuffer,
+   .SwapBuffers     = r128SwapBuffers,
+   .MakeCurrent     = r128MakeCurrent,
+   .UnbindContext   = r128UnbindContext,
+   .GetSwapInfo     = NULL,
+   .GetDrawableMSC  = driDrawableGetMSC32,
+   .WaitForMSC      = driWaitForMSC32,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL
+};
diff --git a/src/mesa/drivers/dri/r128/r128_screen.h b/src/mesa/drivers/dri/r128/r128_screen.h
index b31e87661b5..e2fa1677c9f 100644
--- a/src/mesa/drivers/dri/r128/r128_screen.h
+++ b/src/mesa/drivers/dri/r128/r128_screen.h
@@ -77,6 +77,8 @@ typedef struct {
    /* Configuration cache with default values for all contexts */
    driOptionCache optionCache;
 
+   const __DRIextension *extensions[4];
+
 } r128ScreenRec, *r128ScreenPtr;
 
 
diff --git a/src/mesa/drivers/dri/r128/r128_span.c b/src/mesa/drivers/dri/r128/r128_span.c
index c5b6480db9e..dd177e0def5 100644
--- a/src/mesa/drivers/dri/r128/r128_span.c
+++ b/src/mesa/drivers/dri/r128/r128_span.c
@@ -130,6 +130,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* 16-bit depth buffer functions
  */
+#define VALUE_TYPE GLushort
 
 #define WRITE_DEPTH_SPAN()						\
 do {									\
@@ -206,6 +207,8 @@ do {									\
 
 /* 24-bit depth, 8-bit stencil buffer functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH_SPAN()						\
 do {									\
    GLuint buf[n];							\
diff --git a/src/mesa/drivers/dri/r128/r128_state.c b/src/mesa/drivers/dri/r128/r128_state.c
index 58c3a27ee83..451dcd1b553 100644
--- a/src/mesa/drivers/dri/r128/r128_state.c
+++ b/src/mesa/drivers/dri/r128/r128_state.c
@@ -39,9 +39,9 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_tris.h"
 #include "r128_tex.h"
 
-#include "context.h"
-#include "enums.h"
-#include "colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -812,7 +812,7 @@ static void r128UpdateWindow( GLcontext *ctx )
    r128ContextPtr rmesa = R128_CONTEXT(ctx);
    int x = rmesa->driDrawable->x;
    int y = rmesa->driDrawable->y;
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
    driRenderbuffer *drb = (driRenderbuffer *) rb;
 
    rmesa->setup.window_xy_offset = (((y & 0xFFF) << R128_WINDOW_Y_SHIFT) |
@@ -896,18 +896,22 @@ static void r128DDDrawBuffer( GLcontext *ctx, GLenum mode )
 
    FLUSH_BATCH( rmesa );
 
-   /*
-    * _ColorDrawBufferMask is easier to cope with than <mode>.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-   case BUFFER_BIT_BACK_LEFT:
-      FALLBACK( rmesa, R128_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
-   default:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
       /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( rmesa, R128_FALLBACK_DRAW_BUFFER, GL_TRUE );
-      break;
+      return;
+   }
+   else {
+      switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+      case BUFFER_FRONT_LEFT:
+      case BUFFER_BACK_LEFT:
+         FALLBACK( rmesa, R128_FALLBACK_DRAW_BUFFER, GL_FALSE );
+         break;
+      default:
+         /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
+         FALLBACK( rmesa, R128_FALLBACK_DRAW_BUFFER, GL_TRUE );
+         break;
+      }
    }
 
    rmesa->new_state |= R128_NEW_WINDOW;
diff --git a/src/mesa/drivers/dri/r128/r128_tex.c b/src/mesa/drivers/dri/r128/r128_tex.c
index 5712351b037..3fc9c06cfa2 100644
--- a/src/mesa/drivers/dri/r128/r128_tex.c
+++ b/src/mesa/drivers/dri/r128/r128_tex.c
@@ -39,17 +39,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_tex.h"
 #include "r128_texobj.h"
 
-#include "context.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texobj.h"
-#include "imports.h"
-#include "colormac.h"
-#include "texobj.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/texobj.h"
 
 #include "xmlpool.h"
 
diff --git a/src/mesa/drivers/dri/r128/r128_tex.h b/src/mesa/drivers/dri/r128/r128_tex.h
index e863436f4b2..7df8decf76b 100644
--- a/src/mesa/drivers/dri/r128/r128_tex.h
+++ b/src/mesa/drivers/dri/r128/r128_tex.h
@@ -67,9 +67,9 @@ extern void r128InitTextureFuncs( struct dd_function_table *functions );
 #define R128PACKCOLOR4444( r, g, b, a )					\
    ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
 
-static __inline__ uint32_t r128PackColor( GLuint cpp,
-					GLubyte r, GLubyte g,
-					GLubyte b, GLubyte a )
+static INLINE uint32_t r128PackColor( GLuint cpp,
+                                       GLubyte r, GLubyte g,
+                                       GLubyte b, GLubyte a )
 {
     switch ( cpp ) {
     case 2:
diff --git a/src/mesa/drivers/dri/r128/r128_texmem.c b/src/mesa/drivers/dri/r128/r128_texmem.c
index af510a38c95..111fe1fd742 100644
--- a/src/mesa/drivers/dri/r128/r128_texmem.c
+++ b/src/mesa/drivers/dri/r128/r128_texmem.c
@@ -38,11 +38,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r128_tris.h"
 #include "r128_tex.h"
 
-#include "context.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "texformat.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
+#include "main/imports.h"
 
 #define TEX_0	1
 #define TEX_1	2
diff --git a/src/mesa/drivers/dri/r128/r128_texobj.h b/src/mesa/drivers/dri/r128/r128_texobj.h
index 8a0596eea8e..efbbb2df781 100644
--- a/src/mesa/drivers/dri/r128/r128_texobj.h
+++ b/src/mesa/drivers/dri/r128/r128_texobj.h
@@ -35,7 +35,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef _R128_TEXOBJ_H_
 #define _R128_TEXOBJ_H_
 
-#include "mm.h"
+#include "main/mm.h"
 
 /* Individual texture image information.
  */
diff --git a/src/mesa/drivers/dri/r128/r128_texstate.c b/src/mesa/drivers/dri/r128/r128_texstate.c
index 211b9ea2a97..a9c9568003b 100644
--- a/src/mesa/drivers/dri/r128/r128_texstate.c
+++ b/src/mesa/drivers/dri/r128/r128_texstate.c
@@ -32,11 +32,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Brian Paul <brianp@valinux.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
 
 #include "r128_context.h"
 #include "r128_state.h"
diff --git a/src/mesa/drivers/dri/r128/r128_tris.c b/src/mesa/drivers/dri/r128/r128_tris.c
index f2f124360c3..bcc9ffa651e 100644
--- a/src/mesa/drivers/dri/r128/r128_tris.c
+++ b/src/mesa/drivers/dri/r128/r128_tris.c
@@ -33,10 +33,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/r128/r128_tris.h b/src/mesa/drivers/dri/r128/r128_tris.h
index c8f0a4809b9..d90ca31534e 100644
--- a/src/mesa/drivers/dri/r128/r128_tris.h
+++ b/src/mesa/drivers/dri/r128/r128_tris.h
@@ -35,7 +35,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R128_TRIS_H__
 #define __R128_TRIS_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void r128InitTriFuncs( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/r200/r200_cmdbuf.c b/src/mesa/drivers/dri/r200/r200_cmdbuf.c
index c1d51e87001..e1633772a18 100644
--- a/src/mesa/drivers/dri/r200/r200_cmdbuf.c
+++ b/src/mesa/drivers/dri/r200/r200_cmdbuf.c
@@ -31,12 +31,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "macros.h"
-#include "context.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
-#include "simple_list.h"
+#include "main/simple_list.h"
 
 #include "r200_context.h"
 #include "r200_state.h"
diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c
index 5a178442bdb..5531e0a7399 100644
--- a/src/mesa/drivers/dri/r200/r200_context.c
+++ b/src/mesa/drivers/dri/r200/r200_context.c
@@ -32,15 +32,15 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "state.h"
+#include "main/glheader.h"
+#include "main/api_arrayelt.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/state.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -69,6 +69,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_ATI_fragment_shader
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
+#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_blend_equation_separate
 #define need_GL_EXT_blend_func_separate
@@ -132,6 +133,7 @@ const struct dri_extension card_extensions[] =
     { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
     { "GL_EXT_blend_subtract",             NULL },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
+    { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
     { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
     { "GL_EXT_stencil_wrap",               NULL },
     { "GL_EXT_texture_edge_clamp",         NULL },
@@ -277,14 +279,14 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
                                                  "def_max_anisotropy");
 
    if ( driQueryOptionb( &rmesa->optionCache, "hyperz" ) ) {
-      if ( sPriv->drmMinor < 13 )
+      if ( sPriv->drm_version.minor < 13 )
 	 fprintf( stderr, "DRM version 1.%d too old to support HyperZ, "
-			  "disabling.\n",sPriv->drmMinor );
+			  "disabling.\n", sPriv->drm_version.minor );
       else
 	 rmesa->using_hyperz = GL_TRUE;
    }
  
-   if ( sPriv->drmMinor >= 15 )
+   if ( sPriv->drm_version.minor >= 15 )
       rmesa->texmicrotile = GL_TRUE;
 
    /* Init default driver functions then plug in our R200-specific functions
@@ -317,7 +319,7 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
    rmesa->dri.hwContext = driContextPriv->hHWContext;
    rmesa->dri.hwLock = &sPriv->pSAREA->lock;
    rmesa->dri.fd = sPriv->fd;
-   rmesa->dri.drmMinor = sPriv->drmMinor;
+   rmesa->dri.drmMinor = sPriv->drm_version.minor;
 
    rmesa->r200Screen = screen;
    rmesa->sarea = (drm_radeon_sarea_t *)((GLubyte *)sPriv->pSAREA +
@@ -499,13 +501,10 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
 	      fthrottle_mode,
 	      rmesa->r200Screen->irq);
 
-   rmesa->vblank_flags = (rmesa->r200Screen->irq != 0)
-       ? driGetDefaultVBlankFlags(&rmesa->optionCache) : VBLANK_FLAG_NO_IRQ;
-
    rmesa->prefer_gart_client_texturing = 
       (getenv("R200_GART_CLIENT_TEXTURES") != 0);
 
-   (*dri_interface->getUST)( & rmesa->swap_ust );
+   (*sPriv->systemTime->getUST)( & rmesa->swap_ust );
 
 
 #if DO_DEBUG
@@ -666,15 +665,18 @@ r200MakeCurrent( __DRIcontextPrivate *driContextPriv,
       if (R200_DEBUG & DEBUG_DRI)
 	 fprintf(stderr, "%s ctx %p\n", __FUNCTION__, (void *)newCtx->glCtx);
 
-      if ( newCtx->dri.drawable != driDrawPriv ) {
-	 driDrawableInitVBlank( driDrawPriv, newCtx->vblank_flags,
-				&newCtx->vbl_seq );
-      }
-
       newCtx->dri.readable = driReadPriv;
 
       if ( newCtx->dri.drawable != driDrawPriv ||
            newCtx->lastStamp != driDrawPriv->lastStamp ) {
+	 if (driDrawPriv->swap_interval == (unsigned)-1) {
+	    driDrawPriv->vblFlags = (newCtx->r200Screen->irq != 0)
+	       ? driGetDefaultVBlankFlags(&newCtx->optionCache)
+	       : VBLANK_FLAG_NO_IRQ;
+
+	    driDrawableInitVBlank( driDrawPriv );
+	 }
+
 	 newCtx->dri.drawable = driDrawPriv;
 
 	 r200SetCliprects(newCtx);
diff --git a/src/mesa/drivers/dri/r200/r200_context.h b/src/mesa/drivers/dri/r200/r200_context.h
index bec09e8ef6a..14a1dda46ac 100644
--- a/src/mesa/drivers/dri/r200/r200_context.h
+++ b/src/mesa/drivers/dri/r200/r200_context.h
@@ -41,9 +41,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "dri_util.h"
 #include "texmem.h"
 
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
 #include "r200_reg.h"
 #include "r200_vertprog.h"
 
@@ -63,7 +63,7 @@ typedef union { GLfloat f; uint32_t ui32; } float_ui32_type;
 
 #include "r200_lock.h"
 #include "radeon_screen.h"
-#include "mm.h"
+#include "main/mm.h"
 
 /* Flags for software fallback cases */
 /* See correponding strings in r200_swtcl.c */
@@ -179,6 +179,7 @@ struct r200_tex_obj {
 
    drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
 					/* Six, for the cube faces */
+   GLboolean image_override;		/* Image overridden by GLX_EXT_tfp */
 
    GLuint pp_txfilter;		        /* hardware register values */
    GLuint pp_txformat;
@@ -892,11 +893,8 @@ struct r200_context {
    GLuint TexGenCompSel;
    GLmatrix tmpmat;
 
-   /* VBI / buffer swap
+   /* buffer swap
     */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
    int64_t swap_ust;
    int64_t swap_missed_ust;
 
@@ -929,7 +927,7 @@ struct r200_context {
 #define R200_CONTEXT(ctx)		((r200ContextPtr)(ctx->DriverCtx))
 
 
-static __inline GLuint r200PackColor( GLuint cpp,
+static INLINE GLuint r200PackColor( GLuint cpp,
 					GLubyte r, GLubyte g,
 					GLubyte b, GLubyte a )
 {
diff --git a/src/mesa/drivers/dri/r200/r200_fragshader.c b/src/mesa/drivers/dri/r200/r200_fragshader.c
index 5dd3adaef69..d514b28219a 100644
--- a/src/mesa/drivers/dri/r200/r200_fragshader.c
+++ b/src/mesa/drivers/dri/r200/r200_fragshader.c
@@ -24,13 +24,13 @@
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  **************************************************************************/
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
 
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "tnl/t_context.h"
-#include "atifragshader.h"
-#include "program.h"
+#include "shader/atifragshader.h"
+#include "shader/program.h"
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_tex.h"
diff --git a/src/mesa/drivers/dri/r200/r200_ioctl.c b/src/mesa/drivers/dri/r200/r200_ioctl.c
index ed57defb498..0741e57af71 100644
--- a/src/mesa/drivers/dri/r200/r200_ioctl.c
+++ b/src/mesa/drivers/dri/r200/r200_ioctl.c
@@ -35,10 +35,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <sched.h>
 #include <errno.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "macros.h"
-#include "context.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
 
 #include "r200_context.h"
@@ -419,13 +419,14 @@ static void r200WaitForFrameCompletion( r200ContextPtr rmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void r200CopyBuffer( const __DRIdrawablePrivate *dPriv,
+void r200CopyBuffer( __DRIdrawablePrivate *dPriv,
 		      const drm_clip_rect_t	 *rect)
 {
    r200ContextPtr rmesa;
    GLint nbox, i, ret;
    GLboolean   missed_target;
    int64_t ust;
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
@@ -449,7 +450,7 @@ void r200CopyBuffer( const __DRIdrawablePrivate *dPriv,
    if (!rect)
    {
        UNLOCK_HARDWARE( rmesa );
-       driWaitForVBlank( dPriv, & rmesa->vbl_seq, rmesa->vblank_flags, & missed_target );
+       driWaitForVBlank( dPriv, & missed_target );
        LOCK_HARDWARE( rmesa );
    }
 
@@ -476,16 +477,18 @@ void r200CopyBuffer( const __DRIdrawablePrivate *dPriv,
 	     if (rect->y2 < b->y2)
 		 b->y2 = rect->y2;
 
-	     if (b->x1 < b->x2 && b->y1 < b->y2)
-		 b++;
+	     if (b->x1 >= b->x2 || b->y1 >= b->y2)
+		 continue;
 	  }
-	  else
-	      b++;
 
+	  b++;
 	  n++;
       }
       rmesa->sarea->nbox = n;
 
+      if (!n)
+	 continue;
+
       ret = drmCommandNone( rmesa->dri.fd, DRM_RADEON_SWAP );
 
       if ( ret ) {
@@ -501,7 +504,7 @@ void r200CopyBuffer( const __DRIdrawablePrivate *dPriv,
        rmesa->hw.all_dirty = GL_TRUE;
 
        rmesa->swap_count++;
-       (*dri_interface->getUST)( & ust );
+       (*psp->systemTime->getUST)( & ust );
        if ( missed_target ) {
 	   rmesa->swap_missed_count++;
 	   rmesa->swap_missed_ust = ust - rmesa->swap_ust;
@@ -513,11 +516,12 @@ void r200CopyBuffer( const __DRIdrawablePrivate *dPriv,
    }
 }
 
-void r200PageFlip( const __DRIdrawablePrivate *dPriv )
+void r200PageFlip( __DRIdrawablePrivate *dPriv )
 {
    r200ContextPtr rmesa;
    GLint ret;
    GLboolean   missed_target;
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
@@ -553,10 +557,10 @@ void r200PageFlip( const __DRIdrawablePrivate *dPriv )
     */
    r200WaitForFrameCompletion( rmesa );
    UNLOCK_HARDWARE( rmesa );
-   driWaitForVBlank( dPriv, & rmesa->vbl_seq, rmesa->vblank_flags, & missed_target );
+   driWaitForVBlank( dPriv, & missed_target );
    if ( missed_target ) {
       rmesa->swap_missed_count++;
-      (void) (*dri_interface->getUST)( & rmesa->swap_missed_ust );
+      (void) (*psp->systemTime->getUST)( & rmesa->swap_missed_ust );
    }
    LOCK_HARDWARE( rmesa );
 
@@ -570,7 +574,7 @@ void r200PageFlip( const __DRIdrawablePrivate *dPriv )
    }
 
    rmesa->swap_count++;
-   (void) (*dri_interface->getUST)( & rmesa->swap_ust );
+   (void) (*psp->systemTime->getUST)( & rmesa->swap_ust );
 
 #if 000
    if ( rmesa->sarea->pfCurrentPage == 1 ) {
@@ -857,7 +861,7 @@ void r200Finish( GLcontext *ctx )
  * the kernel data structures, and the current context to get the
  * device fd.
  */
-void *r200AllocateMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLsizei size,
+void *r200AllocateMemoryMESA(__DRIscreen *screen, GLsizei size,
 			     GLfloat readfreq, GLfloat writefreq, 
 			     GLfloat priority)
 {
@@ -899,7 +903,7 @@ void *r200AllocateMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLsizei size,
 
 
 /* Called via glXFreeMemoryMESA() */
-void r200FreeMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLvoid *pointer)
+void r200FreeMemoryMESA(__DRIscreen *screen, GLvoid *pointer)
 {
    GET_CURRENT_CONTEXT(ctx);
    r200ContextPtr rmesa;
@@ -936,7 +940,7 @@ void r200FreeMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLvoid *pointer)
 }
 
 /* Called via glXGetMemoryOffsetMESA() */
-GLuint r200GetMemoryOffsetMESA(__DRInativeDisplay *dpy, int scrn, const GLvoid *pointer)
+GLuint r200GetMemoryOffsetMESA(__DRIscreen *screen, const GLvoid *pointer)
 {
    GET_CURRENT_CONTEXT(ctx);
    r200ContextPtr rmesa;
diff --git a/src/mesa/drivers/dri/r200/r200_ioctl.h b/src/mesa/drivers/dri/r200/r200_ioctl.h
index 5ed1555f6a3..f7458e4a0ed 100644
--- a/src/mesa/drivers/dri/r200/r200_ioctl.h
+++ b/src/mesa/drivers/dri/r200/r200_ioctl.h
@@ -35,7 +35,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R200_IOCTL_H__
 #define __R200_IOCTL_H__
 
-#include "simple_list.h"
+#include "main/simple_list.h"
 #include "radeon_dri.h"
 #include "r200_lock.h"
 
@@ -89,19 +89,19 @@ extern void r200ReleaseDmaRegion( r200ContextPtr rmesa,
 				    struct r200_dma_region *region,
 				    const char *caller );
 
-extern void r200CopyBuffer( const __DRIdrawablePrivate *drawable,
+extern void r200CopyBuffer( __DRIdrawablePrivate *drawable,
 			    const drm_clip_rect_t      *rect);
-extern void r200PageFlip( const __DRIdrawablePrivate *drawable );
+extern void r200PageFlip( __DRIdrawablePrivate *drawable );
 extern void r200Flush( GLcontext *ctx );
 extern void r200Finish( GLcontext *ctx );
 extern void r200WaitForIdleLocked( r200ContextPtr rmesa );
 extern void r200WaitForVBlank( r200ContextPtr rmesa );
 extern void r200InitIoctlFuncs( struct dd_function_table *functions );
 
-extern void *r200AllocateMemoryMESA( __DRInativeDisplay *dpy, int scrn, GLsizei size, GLfloat readfreq,
+extern void *r200AllocateMemoryMESA( __DRIscreen *screen, GLsizei size, GLfloat readfreq,
 				   GLfloat writefreq, GLfloat priority );
-extern void r200FreeMemoryMESA( __DRInativeDisplay *dpy, int scrn, GLvoid *pointer );
-extern GLuint r200GetMemoryOffsetMESA( __DRInativeDisplay *dpy, int scrn, const GLvoid *pointer );
+extern void r200FreeMemoryMESA( __DRIscreen *screen, GLvoid *pointer );
+extern GLuint r200GetMemoryOffsetMESA( __DRIscreen *screen, const GLvoid *pointer );
 
 extern GLboolean r200IsGartMemory( r200ContextPtr rmesa, const GLvoid *pointer,
 				   GLint size );
@@ -137,7 +137,7 @@ do {								\
    memcpy( rmesa->hw.ATOM.lastcmd, rmesa->hw.ATOM.cmd,	\
 	   rmesa->hw.ATOM.cmd_size * 4)
 
-static __inline int R200_DB_STATECHANGE( 
+static INLINE int R200_DB_STATECHANGE( 
    r200ContextPtr rmesa,
    struct r200_state_atom *atom )
 {
@@ -183,7 +183,7 @@ do {							\
  * and hang on to the lock until the critical section is finished and we flush
  * the buffer again and unlock.
  */
-static __inline void r200EnsureCmdBufSpace( r200ContextPtr rmesa, int bytes )
+static INLINE void r200EnsureCmdBufSpace( r200ContextPtr rmesa, int bytes )
 {
    if (rmesa->store.cmd_used + bytes > R200_CMD_BUF_SZ)
       r200FlushCmdBuf( rmesa, __FUNCTION__ );
@@ -192,7 +192,7 @@ static __inline void r200EnsureCmdBufSpace( r200ContextPtr rmesa, int bytes )
 
 /* Alloc space in the command buffer
  */
-static __inline char *r200AllocCmdBuf( r200ContextPtr rmesa,
+static INLINE char *r200AllocCmdBuf( r200ContextPtr rmesa,
 					 int bytes, const char *where )
 {
    char * head;
diff --git a/src/mesa/drivers/dri/r200/r200_lock.c b/src/mesa/drivers/dri/r200/r200_lock.c
index f89b526a312..99661a4bfb4 100644
--- a/src/mesa/drivers/dri/r200/r200_lock.c
+++ b/src/mesa/drivers/dri/r200/r200_lock.c
@@ -60,7 +60,7 @@ r200UpdatePageFlipping( r200ContextPtr rmesa )
 
 
 
-/* Update the hardware state.  This is called if another context has
+/* Update the hardware state.  This is called if another main/context.has
  * grabbed the hardware lock, which includes the X server.  This
  * function also updates the driver's window state after the X server
  * moves, resizes or restacks a window -- the change will be reflected
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index 7bc05e2f0bd..8512b9af478 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -32,11 +32,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/imports.h"
+#include "main/macros.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "math/m_translate.h"
diff --git a/src/mesa/drivers/dri/r200/r200_pixel.c b/src/mesa/drivers/dri/r200/r200_pixel.c
index 2f5aab0744b..be68821dc1d 100644
--- a/src/mesa/drivers/dri/r200/r200_pixel.c
+++ b/src/mesa/drivers/dri/r200/r200_pixel.c
@@ -31,10 +31,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 #include "swrast/swrast.h"
 
 #include "r200_context.h"
@@ -294,7 +294,7 @@ static void do_draw_pix( GLcontext *ctx,
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
    __DRIdrawablePrivate *dPriv = rmesa->dri.drawable;
    drm_clip_rect_t *box = dPriv->pClipRects;
-   struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorDrawBuffers[0][0];
+   struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorDrawBuffers[0];
    driRenderbuffer *drb = (driRenderbuffer *) rb;
    int nbox = dPriv->numClipRects;
    int i;
@@ -382,13 +382,13 @@ r200TryDrawPixels( GLcontext *ctx,
    GLint pitch = unpack->RowLength ? unpack->RowLength : width;
    GLuint planemask;
    GLuint cpp = rmesa->r200Screen->cpp;
-   GLint size = width * pitch * cpp;
+   GLint size = height * pitch * cpp;
 
    if (R200_DEBUG & DEBUG_PIXEL)
       fprintf(stderr, "%s\n", __FUNCTION__);
 
    /* check that we're drawing to exactly one color buffer */
-   if (ctx->DrawBuffer->_NumColorDrawBuffers[0] != 1)
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1)
      return GL_FALSE;
 
    switch (format) {
diff --git a/src/mesa/drivers/dri/r200/r200_sanity.c b/src/mesa/drivers/dri/r200/r200_sanity.c
index 00d2f65c998..36530c224e9 100644
--- a/src/mesa/drivers/dri/r200/r200_sanity.c
+++ b/src/mesa/drivers/dri/r200/r200_sanity.c
@@ -34,8 +34,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  
 #include <errno.h> 
 
-#include "glheader.h"
-#include "imports.h"
+#include "main/glheader.h"
+#include "main/imports.h"
 
 #include "r200_context.h"
 #include "r200_ioctl.h"
diff --git a/src/mesa/drivers/dri/r200/r200_span.c b/src/mesa/drivers/dri/r200/r200_span.c
index fe427bdcdec..9783678028a 100644
--- a/src/mesa/drivers/dri/r200/r200_span.c
+++ b/src/mesa/drivers/dri/r200/r200_span.c
@@ -32,10 +32,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
-#include "colormac.h"
 
 #include "r200_context.h"
 #include "r200_ioctl.h"
@@ -172,6 +172,7 @@ r200_mba_z16( driRenderbuffer *drb, GLint x, GLint y )
 
 /* 16-bit depth buffer functions
  */
+#define VALUE_TYPE GLushort
 
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)(buf + r200_mba_z16( drb, _x + xo, _y + yo )) = d;
@@ -185,6 +186,7 @@ r200_mba_z16( driRenderbuffer *drb, GLint x, GLint y )
 
 /* 24 bit depth, 8 bit stencil depthbuffer functions
  */
+#define VALUE_TYPE GLuint
 
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
@@ -255,7 +257,7 @@ static void r200SpanRenderStart( GLcontext *ctx )
    {
       int p;
       driRenderbuffer *drb =
-	 (driRenderbuffer *) ctx->WinSysDrawBuffer->_ColorDrawBuffers[0][0];
+	 (driRenderbuffer *) ctx->WinSysDrawBuffer->_ColorDrawBuffers[0];
       volatile int *buf =
 	 (volatile int *)(rmesa->dri.screen->pFB + drb->offset);
       p = *buf;
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index e9d2d9a3cd1..0eaaaf69ac7 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -33,13 +33,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "api_arrayelt.h"
-#include "enums.h"
-#include "colormac.h"
-#include "light.h"
-#include "framebuffer.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/api_arrayelt.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/light.h"
+#include "main/framebuffer.h"
 
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
@@ -1859,8 +1859,7 @@ void r200SetCliprects( r200ContextPtr rmesa )
    GLframebuffer *const draw_fb = (GLframebuffer*) drawable->driverPrivate;
    GLframebuffer *const read_fb = (GLframebuffer*) readable->driverPrivate;
 
-   if (draw_fb->_ColorDrawBufferMask[0]
-       == BUFFER_BIT_BACK_LEFT) {
+   if (draw_fb->_ColorDrawBufferIndexes[0] == BUFFER_BIT_BACK_LEFT) {
       /* Can't ignore 2d windows if we are page flipping.
        */
       if ( drawable->numBackClipRects == 0 || rmesa->doPageFlip ) {
@@ -1910,17 +1909,18 @@ static void r200DrawBuffer( GLcontext *ctx, GLenum mode )
 
    R200_FIREVERTICES(rmesa);	/* don't pipeline cliprect changes */
 
-   /*
-    * _ColorDrawBufferMask is easier to cope with than <mode>.
-    * Check for software fallback, update cliprects.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-   case BUFFER_BIT_BACK_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      /* 0 (GL_NONE) buffers or multiple color drawing buffers */
+      FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
+   case BUFFER_BACK_LEFT:
       FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
    default:
-      /* 0 (GL_NONE) buffers or multiple color drawing buffers */
       FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
@@ -2445,11 +2445,11 @@ r200UpdateDrawBuffer(GLcontext *ctx)
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    driRenderbuffer *drb;
 
-   if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
+   if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
       /* draw to front */
       drb = (driRenderbuffer *) fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
    }
-   else if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT) {
+   else if (fb->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT) {
       /* draw to back */
       drb = (driRenderbuffer *) fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
    }
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index 0c36cefc161..9e4677eda40 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -31,11 +31,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "enums.h"
-#include "colormac.h"
-#include "api_arrayelt.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/api_arrayelt.h"
 
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index a1ea0198bee..b25f0282445 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -32,13 +32,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "enums.h"
-#include "image.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/imports.h"
+#include "main/macros.h"
 
 #include "swrast/s_context.h"
 #include "swrast/s_fog.h"
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.h b/src/mesa/drivers/dri/r200/r200_swtcl.h
index 7458c549288..8c29fd0c999 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.h
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.h
@@ -34,7 +34,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R200_SWTCL_H__
 #define __R200_SWTCL_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "r200_context.h"
 
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index 2ad35d43906..99aecfe1e90 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -32,12 +32,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "enums.h"
-#include "colormac.h"
-#include "light.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/light.h"
 
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -488,7 +488,7 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/r200/r200_tex.c b/src/mesa/drivers/dri/r200/r200_tex.c
index e7a37dd4c99..5a4db33f441 100644
--- a/src/mesa/drivers/dri/r200/r200_tex.c
+++ b/src/mesa/drivers/dri/r200/r200_tex.c
@@ -31,18 +31,19 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "colormac.h"
-#include "context.h"
-#include "enums.h"
-#include "image.h"
-#include "simple_list.h"
-#include "texformat.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+
 #include "texmem.h"
-#include "teximage.h"
-#include "texobj.h"
 
 #include "r200_context.h"
 #include "r200_state.h"
@@ -102,37 +103,39 @@ static void r200SetTexWrap( r200TexObjPtr t, GLenum swrap, GLenum twrap, GLenum
       _mesa_problem(NULL, "bad S wrap mode in %s", __FUNCTION__);
    }
 
-   switch ( twrap ) {
-   case GL_REPEAT:
-      t->pp_txfilter |= R200_CLAMP_T_WRAP;
-      break;
-   case GL_CLAMP:
-      t->pp_txfilter |= R200_CLAMP_T_CLAMP_GL;
-      is_clamp = GL_TRUE;
-      break;
-   case GL_CLAMP_TO_EDGE:
-      t->pp_txfilter |= R200_CLAMP_T_CLAMP_LAST;
-      break;
-   case GL_CLAMP_TO_BORDER:
-      t->pp_txfilter |= R200_CLAMP_T_CLAMP_GL;
-      is_clamp_to_border = GL_TRUE;
-      break;
-   case GL_MIRRORED_REPEAT:
-      t->pp_txfilter |= R200_CLAMP_T_MIRROR;
-      break;
-   case GL_MIRROR_CLAMP_EXT:
-      t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_GL;
-      is_clamp = GL_TRUE;
-      break;
-   case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-      t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_LAST;
-      break;
-   case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-      t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_GL;
-      is_clamp_to_border = GL_TRUE;
-      break;
-   default:
-      _mesa_problem(NULL, "bad T wrap mode in %s", __FUNCTION__);
+   if (t->base.tObj->Target != GL_TEXTURE_1D) {
+      switch ( twrap ) {
+      case GL_REPEAT:
+         t->pp_txfilter |= R200_CLAMP_T_WRAP;
+         break;
+      case GL_CLAMP:
+         t->pp_txfilter |= R200_CLAMP_T_CLAMP_GL;
+         is_clamp = GL_TRUE;
+         break;
+      case GL_CLAMP_TO_EDGE:
+         t->pp_txfilter |= R200_CLAMP_T_CLAMP_LAST;
+         break;
+      case GL_CLAMP_TO_BORDER:
+         t->pp_txfilter |= R200_CLAMP_T_CLAMP_GL;
+         is_clamp_to_border = GL_TRUE;
+         break;
+      case GL_MIRRORED_REPEAT:
+         t->pp_txfilter |= R200_CLAMP_T_MIRROR;
+         break;
+      case GL_MIRROR_CLAMP_EXT:
+         t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_GL;
+         is_clamp = GL_TRUE;
+         break;
+      case GL_MIRROR_CLAMP_TO_EDGE_EXT:
+         t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_LAST;
+         break;
+      case GL_MIRROR_CLAMP_TO_BORDER_EXT:
+         t->pp_txfilter |= R200_CLAMP_T_MIRROR_CLAMP_GL;
+         is_clamp_to_border = GL_TRUE;
+         break;
+      default:
+         _mesa_problem(NULL, "bad T wrap mode in %s", __FUNCTION__);
+      }
    }
 
    t->pp_txformat_x &= ~R200_CLAMP_Q_MASK;
diff --git a/src/mesa/drivers/dri/r200/r200_tex.h b/src/mesa/drivers/dri/r200/r200_tex.h
index e6c0e00eb07..10ff8e8a660 100644
--- a/src/mesa/drivers/dri/r200/r200_tex.h
+++ b/src/mesa/drivers/dri/r200/r200_tex.h
@@ -35,6 +35,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R200_TEX_H__
 #define __R200_TEX_H__
 
+extern void r200SetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+			     unsigned long long offset, GLint depth,
+			     GLuint pitch);
+
 extern void r200UpdateTextureState( GLcontext *ctx );
 
 extern int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t, GLuint face );
diff --git a/src/mesa/drivers/dri/r200/r200_texmem.c b/src/mesa/drivers/dri/r200/r200_texmem.c
index d926313d576..3b81ac0c802 100644
--- a/src/mesa/drivers/dri/r200/r200_texmem.c
+++ b/src/mesa/drivers/dri/r200/r200_texmem.c
@@ -37,11 +37,11 @@ SOFTWARE.
  
 #include <errno.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "colormac.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_tex.h"
@@ -181,7 +181,8 @@ static void r200UploadRectSubImage( r200ContextPtr rmesa,
       /* In this case, could also use GART texturing.  This is
        * currently disabled, but has been tested & works.
        */
-      t->pp_txoffset = r200GartOffsetFromVirtual( rmesa, texImage->Data );
+      if ( !t->image_override )
+         t->pp_txoffset = r200GartOffsetFromVirtual( rmesa, texImage->Data );
       t->pp_txpitch = texImage->RowStride * texFormat->TexelBytes - 32;
 
       if (R200_DEBUG & DEBUG_TEXTURE)
@@ -467,7 +468,7 @@ int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t, GLuint face )
 	       t->base.firstLevel, t->base.lastLevel );
    }
 
-   if ( !t || t->base.totalSize == 0 )
+   if ( !t || t->base.totalSize == 0 || t->image_override )
       return 0;
 
    if (R200_DEBUG & DEBUG_SYNC) {
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index ae02ec4b638..3f9a2f4ac1b 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -32,12 +32,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
+#include "main/texobj.h"
+#include "main/enums.h"
 
 #include "r200_context.h"
 #include "r200_state.h"
@@ -70,12 +71,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define _INVALID(f) \
     [ MESA_FORMAT_ ## f ] = { 0xffffffff, 0 }
 #define VALID_FORMAT(f) ( ((f) <= MESA_FORMAT_RGBA_DXT5) \
-			     && (tx_table_le[f].format != 0xffffffff) )
+			     && (tx_table_be[f].format != 0xffffffff) )
 
-static const struct {
+struct tx_table {
    GLuint format, filter;
-}
-tx_table_be[] =
+};
+
+static const struct tx_table tx_table_be[] =
 {
    [ MESA_FORMAT_RGBA8888 ] = { R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP, 0 },
    _ALPHA_REV(RGBA8888),
@@ -104,16 +106,13 @@ tx_table_be[] =
    _ALPHA(RGBA_DXT5),
 };
 
-static const struct {
-   GLuint format, filter;
-}
-tx_table_le[] =
+static const struct tx_table tx_table_le[] =
 {
    _ALPHA(RGBA8888),
    [ MESA_FORMAT_RGBA8888_REV ] = { R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP, 0 },
    _ALPHA(ARGB8888),
    _ALPHA_REV(ARGB8888),
-   _INVALID(RGB888),
+   [ MESA_FORMAT_RGB888 ] = { R200_TXFORMAT_ARGB8888, 0 },
    _COLOR(RGB565),
    _COLOR_REV(RGB565),
    _ALPHA(ARGB4444),
@@ -160,30 +159,26 @@ static void r200SetTexImages( r200ContextPtr rmesa,
    GLint i, texelBytes;
    GLint numLevels;
    GLint log2Width, log2Height, log2Depth;
-   const GLuint ui = 1;
-   const GLubyte littleEndian = *((const GLubyte *) &ui);
 
    /* Set the hardware texture format
     */
+   if ( !t->image_override ) {
+      if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
+	 const struct tx_table *table = _mesa_little_endian() ? tx_table_le :
+								tx_table_be;
 
-   t->pp_txformat &= ~(R200_TXFORMAT_FORMAT_MASK |
-		       R200_TXFORMAT_ALPHA_IN_MAP);
-   t->pp_txfilter &= ~R200_YUV_TO_RGB;
+         t->pp_txformat &= ~(R200_TXFORMAT_FORMAT_MASK |
+                             R200_TXFORMAT_ALPHA_IN_MAP);
+         t->pp_txfilter &= ~R200_YUV_TO_RGB;
 
-   if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
-      if (littleEndian) {
-	 t->pp_txformat |= tx_table_le[ baseImage->TexFormat->MesaFormat ].format;
-	 t->pp_txfilter |= tx_table_le[ baseImage->TexFormat->MesaFormat ].filter;
+	 t->pp_txformat |= table[ baseImage->TexFormat->MesaFormat ].format;
+	 t->pp_txfilter |= table[ baseImage->TexFormat->MesaFormat ].filter;
       }
       else {
-	 t->pp_txformat |= tx_table_be[ baseImage->TexFormat->MesaFormat ].format;
-	 t->pp_txfilter |= tx_table_be[ baseImage->TexFormat->MesaFormat ].filter;
+         _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
+         return;
       }
    }
-   else {
-      _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
-      return;
-   }
 
    texelBytes = baseImage->TexFormat->TexelBytes;
 
@@ -380,11 +375,13 @@ static void r200SetTexImages( r200ContextPtr rmesa,
     * requires 64-byte aligned pitches, and we may/may not need the
     * blitter.   NPOT only!
     */
-   if (baseImage->IsCompressed)
-      t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
-   else
-      t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
-   t->pp_txpitch -= 32;
+   if ( !t->image_override ) {
+      if (baseImage->IsCompressed)
+         t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
+      else
+         t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
+      t->pp_txpitch -= 32;
+   }
 
    t->dirty_state = TEX_ALL;
 
@@ -979,6 +976,44 @@ static GLboolean r200UpdateTextureEnv( GLcontext *ctx, int unit, int slot, GLuin
    return GL_TRUE;
 }
 
+void r200SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
+		      unsigned long long offset, GLint depth, GLuint pitch)
+{
+	r200ContextPtr rmesa = pDRICtx->driverPrivate;
+	struct gl_texture_object *tObj =
+	    _mesa_lookup_texture(rmesa->glCtx, texname);
+	r200TexObjPtr t;
+
+	if (!tObj)
+		return;
+
+	t = (r200TexObjPtr) tObj->DriverData;
+
+	t->image_override = GL_TRUE;
+
+	if (!offset)
+		return;
+
+	t->pp_txoffset = offset;
+	t->pp_txpitch = pitch - 32;
+
+	switch (depth) {
+	case 32:
+		t->pp_txformat = tx_table_le[MESA_FORMAT_ARGB8888].format;
+		t->pp_txfilter |= tx_table_le[MESA_FORMAT_ARGB8888].filter;
+		break;
+	case 24:
+	default:
+		t->pp_txformat = tx_table_le[MESA_FORMAT_RGB888].format;
+		t->pp_txfilter |= tx_table_le[MESA_FORMAT_RGB888].filter;
+		break;
+	case 16:
+		t->pp_txformat = tx_table_le[MESA_FORMAT_RGB565].format;
+		t->pp_txfilter |= tx_table_le[MESA_FORMAT_RGB565].filter;
+		break;
+	}
+}
+
 #define REF_COLOR 1
 #define REF_ALPHA 2
 
@@ -1560,7 +1595,7 @@ static GLboolean enable_tex_2d( GLcontext *ctx, int unit )
       R200_FIREVERTICES( rmesa );
       r200SetTexImages( rmesa, tObj );
       r200UploadTexImages( rmesa, (r200TexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock ) 
+      if ( !t->base.memBlock && !t->image_override ) 
 	 return GL_FALSE;
    }
 
@@ -1668,7 +1703,9 @@ static GLboolean enable_tex_rect( GLcontext *ctx, int unit )
       R200_FIREVERTICES( rmesa );
       r200SetTexImages( rmesa, tObj );
       r200UploadTexImages( rmesa, (r200TexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock && !rmesa->prefer_gart_client_texturing ) 
+      if ( !t->base.memBlock &&
+           !t->image_override &&
+           !rmesa->prefer_gart_client_texturing ) 
 	 return GL_FALSE;
    }
 
@@ -1778,6 +1815,12 @@ void r200UpdateTextureState( GLcontext *ctx )
    GLboolean ok;
    GLuint dbg;
 
+   /* NOTE: must not manipulate rmesa->state.texture.unit[].unitneeded or
+      rmesa->state.envneeded before a R200_STATECHANGE (or R200_NEWPRIM) since
+      we use these to determine if we want to emit the corresponding state
+      atoms. */
+   R200_NEWPRIM( rmesa );
+
    if (ctx->ATIFragmentShader._Enabled) {
       GLuint i;
       for (i = 0; i < R200_MAX_TEXTURE_UNITS; i++) {
diff --git a/src/mesa/drivers/dri/r200/r200_vertprog.c b/src/mesa/drivers/dri/r200/r200_vertprog.c
index 6089d617c6b..562992fbb5c 100644
--- a/src/mesa/drivers/dri/r200/r200_vertprog.c
+++ b/src/mesa/drivers/dri/r200/r200_vertprog.c
@@ -30,10 +30,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Aapo Tahkola <aet@rasterburn.org>
  *   Roland Scheidegger <rscheidegger_lists@hispeed.ch>
  */
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "program.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/program.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
@@ -154,7 +154,7 @@ static GLboolean r200VertexProgUpdateParams(GLcontext *ctx, struct r200_vertex_p
    return GL_TRUE;
 }
 
-static __inline unsigned long t_dst_mask(GLuint mask)
+static INLINE unsigned long t_dst_mask(GLuint mask)
 {
    /* WRITEMASK_* is equivalent to VSF_FLAG_* */
    return mask & VSF_FLAG_ALL;
@@ -215,6 +215,7 @@ static unsigned long t_src_class(enum register_file file)
    case PROGRAM_LOCAL_PARAM:
    case PROGRAM_ENV_PARAM:
    case PROGRAM_NAMED_PARAM:
+   case PROGRAM_CONSTANT:
    case PROGRAM_STATE_VAR:
       return VSF_IN_CLASS_PARAM;
    /*
@@ -228,7 +229,7 @@ static unsigned long t_src_class(enum register_file file)
    }
 }
 
-static __inline unsigned long t_swizzle(GLubyte swizzle)
+static INLINE unsigned long t_swizzle(GLubyte swizzle)
 {
 /* this is in fact a NOP as the Mesa SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */
    return swizzle;
@@ -408,6 +409,7 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
    int fog_temp_i = 0;
    int free_inputs;
    int array_count = 0;
+   int u_temp_used;
 
    vp->native = GL_FALSE;
    vp->translated = GL_TRUE;
@@ -744,9 +746,16 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 	 goto next;
 
       case OPCODE_MAD:
+	 /* only 2 read ports into temp memory thus may need the macro op MAD_2
+	    instead (requiring 2 clocks) if all inputs are in temp memory
+	    (and, only if they actually reference 3 distinct temps) */
 	 hw_op=(src[0].File == PROGRAM_TEMPORARY &&
 	    src[1].File == PROGRAM_TEMPORARY &&
-	    src[2].File == PROGRAM_TEMPORARY) ? R200_VPI_OUT_OP_MAD_2 : R200_VPI_OUT_OP_MAD;
+	    src[2].File == PROGRAM_TEMPORARY &&
+	    (((src[0].RelAddr << 8) | src[0].Index) != ((src[1].RelAddr << 8) | src[1].Index)) &&
+	    (((src[0].RelAddr << 8) | src[0].Index) != ((src[2].RelAddr << 8) | src[2].Index)) &&
+	    (((src[1].RelAddr << 8) | src[1].Index) != ((src[2].RelAddr << 8) | src[2].Index))) ?
+	    R200_VPI_OUT_OP_MAD_2 : R200_VPI_OUT_OP_MAD;
 
 	 o_inst->op = MAKE_VSF_OP(hw_op, t_dst(&dst),
 	    t_dst_mask(dst.WriteMask));
@@ -874,8 +883,11 @@ else {
       case OPCODE_XPD:
 	 /* mul r0, r1.yzxw, r2.zxyw
 	    mad r0, -r2.yzxw, r1.zxyw, r0
-	    NOTE: might need MAD_2
 	  */
+	 hw_op=(src[0].File == PROGRAM_TEMPORARY &&
+	    src[1].File == PROGRAM_TEMPORARY &&
+	    (((src[0].RelAddr << 8) | src[0].Index) != ((src[1].RelAddr << 8) | src[1].Index))) ?
+	    R200_VPI_OUT_OP_MAD_2 : R200_VPI_OUT_OP_MAD;
 
 	 o_inst->op = MAKE_VSF_OP(R200_VPI_OUT_OP_MUL,
 	    (u_temp_i << R200_VPI_OUT_REG_INDEX_SHIFT) | R200_VSF_OUT_CLASS_TMP,
@@ -901,7 +913,7 @@ else {
 	 o_inst++;
 	 u_temp_i--;
 
-	 o_inst->op = MAKE_VSF_OP(R200_VPI_OUT_OP_MAD, t_dst(&dst),
+	 o_inst->op = MAKE_VSF_OP(hw_op, t_dst(&dst),
 		t_dst_mask(dst.WriteMask));
 
 	 o_inst->src0 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
@@ -1051,14 +1063,15 @@ else {
          dofogfix = 0;
       }
 
+      u_temp_used = (R200_VSF_MAX_TEMPS - 1) - u_temp_i;
       if (mesa_vp->Base.NumNativeTemporaries <
-	 (mesa_vp->Base.NumTemporaries + (R200_VSF_MAX_TEMPS - 1 - u_temp_i))) {
+	 (mesa_vp->Base.NumTemporaries + u_temp_used)) {
 	 mesa_vp->Base.NumNativeTemporaries =
-	    mesa_vp->Base.NumTemporaries + (R200_VSF_MAX_TEMPS - 1 - u_temp_i);
+	    mesa_vp->Base.NumTemporaries + u_temp_used;
       }
-      if (u_temp_i < mesa_vp->Base.NumTemporaries) {
+      if ((mesa_vp->Base.NumTemporaries + u_temp_used) > R200_VSF_MAX_TEMPS) {
 	 if (R200_DEBUG & DEBUG_FALLBACKS) {
-	    fprintf(stderr, "Ran out of temps, num temps %d, us %d\n", mesa_vp->Base.NumTemporaries, u_temp_i);
+	    fprintf(stderr, "Ran out of temps, num temps %d, us %d\n", mesa_vp->Base.NumTemporaries, u_temp_used);
 	 }
 	 return GL_FALSE;
       }
diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile
index 44248964fdb..6ca934204f3 100644
--- a/src/mesa/drivers/dri/r300/Makefile
+++ b/src/mesa/drivers/dri/r300/Makefile
@@ -28,7 +28,6 @@ DRIVER_SOURCES = \
 		 radeon_span.c \
 		 radeon_state.c \
 		 r300_mem.c \
-		 \
 		 r300_context.c \
 		 r300_ioctl.c \
 		 r300_cmdbuf.c \
@@ -37,8 +36,16 @@ DRIVER_SOURCES = \
 		 r300_texmem.c \
 		 r300_tex.c \
 		 r300_texstate.c \
+		 radeon_program.c \
+		 radeon_program_alu.c \
+		 radeon_program_pair.c \
+		 radeon_nqssadce.c \
 		 r300_vertprog.c \
 		 r300_fragprog.c \
+		 r300_fragprog_swizzle.c \
+		 r300_fragprog_emit.c \
+		 r500_fragprog.c \
+		 r500_fragprog_emit.c \
 		 r300_shader.c \
 		 r300_emit.c \
 		 r300_swtcl.c \
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
index 9eca41fa38c..c9e1dfe9774 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
@@ -33,13 +33,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Nicolai Haehnle <prefect_@gmx.net>
  */
 
-#include "glheader.h"
-#include "state.h"
-#include "imports.h"
-#include "macros.h"
-#include "context.h"
+#include "main/glheader.h"
+#include "main/state.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/context.h"
+#include "main/simple_list.h"
 #include "swrast/swrast.h"
-#include "simple_list.h"
 
 #include "drm.h"
 #include "radeon_drm.h"
@@ -150,7 +150,7 @@ static void r300PrintStateAtom(r300ContextPtr r300, struct r300_state_atom *stat
  * The caller must have ensured that there is enough space in the command
  * buffer.
  */
-static inline void r300EmitAtoms(r300ContextPtr r300, GLboolean dirty)
+static INLINE void r300EmitAtoms(r300ContextPtr r300, GLboolean dirty)
 {
 	struct r300_state_atom *atom;
 	uint32_t *dest;
@@ -164,7 +164,7 @@ static inline void r300EmitAtoms(r300ContextPtr r300, GLboolean dirty)
 	r300->cmdbuf.count_used++;
 
 	/* Emit cache flush */
-	*dest = cmdpacket0(R300_TX_CNTL, 1);
+	*dest = cmdpacket0(R300_TX_INVALTAGS, 1);
 	dest++;
 	r300->cmdbuf.count_used++;
 
@@ -242,6 +242,7 @@ void r300EmitState(r300ContextPtr r300)
 
 #define packet0_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->packet0.count)
 #define vpu_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->vpu.count)
+#define r500fp_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->r500fp.count)
 
 static int check_always(r300ContextPtr r300, struct r300_state_atom *atom)
 {
@@ -262,6 +263,20 @@ static int check_vpu(r300ContextPtr r300, struct r300_state_atom *atom)
 	return cnt ? (cnt * 4) + 1 : 0;
 }
 
+static int check_r500fp(r300ContextPtr r300, struct r300_state_atom *atom)
+{
+	int cnt;
+	cnt = r500fp_count(atom->cmd);
+	return cnt ? (cnt * 6) + 1 : 0;
+}
+
+static int check_r500fp_const(r300ContextPtr r300, struct r300_state_atom *atom)
+{
+	int cnt;
+	cnt = r500fp_count(atom->cmd);
+	return cnt ? (cnt * 4) + 1 : 0;
+}
+
 #define ALLOC_STATE( ATOM, CHK, SZ, IDX )				\
    do {									\
       r300->hw.ATOM.cmd_size = (SZ);					\
@@ -281,10 +296,15 @@ void r300InitCmdBuf(r300ContextPtr r300)
 {
 	int size, mtu;
 	int has_tcl = 1;
+	int is_r500 = 0;
+	int i;
 
 	if (!(r300->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
 		has_tcl = 0;
 
+	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+		is_r500 = 1;
+
 	r300->hw.max_state_size = 2 + 2;	/* reserve extra space for WAIT_IDLE and tex cache flush */
 
 	mtu = r300->radeon.glCtx->Const.MaxTextureUnits;
@@ -299,32 +319,39 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	/* Initialize state atoms */
 	ALLOC_STATE(vpt, always, R300_VPT_CMDSIZE, 0);
 	r300->hw.vpt.cmd[R300_VPT_CMD_0] = cmdpacket0(R300_SE_VPORT_XSCALE, 6);
-	ALLOC_STATE(vap_cntl, always, 2, 0);
-	r300->hw.vap_cntl.cmd[0] = cmdpacket0(R300_VAP_CNTL, 1);
+	ALLOC_STATE(vap_cntl, always, R300_VAP_CNTL_SIZE, 0);
+	r300->hw.vap_cntl.cmd[R300_VAP_CNTL_FLUSH] = cmdpacket0(R300_VAP_PVS_STATE_FLUSH_REG, 1);
+	r300->hw.vap_cntl.cmd[R300_VAP_CNTL_FLUSH_1] = 0;
+	r300->hw.vap_cntl.cmd[R300_VAP_CNTL_CMD] = cmdpacket0(R300_VAP_CNTL, 1);
+	if (is_r500) {
+	    ALLOC_STATE(vap_index_offset, always, 2, 0);
+	    r300->hw.vap_index_offset.cmd[0] = cmdpacket0(R500_VAP_INDEX_OFFSET, 1);
+	    r300->hw.vap_index_offset.cmd[1] = 0;
+	}
 	ALLOC_STATE(vte, always, 3, 0);
 	r300->hw.vte.cmd[0] = cmdpacket0(R300_SE_VTE_CNTL, 2);
-	ALLOC_STATE(unk2134, always, 3, 0);
-	r300->hw.unk2134.cmd[0] = cmdpacket0(0x2134, 2);
+	ALLOC_STATE(vap_vf_max_vtx_indx, always, 3, 0);
+	r300->hw.vap_vf_max_vtx_indx.cmd[0] = cmdpacket0(R300_VAP_VF_MAX_VTX_INDX, 2);
 	ALLOC_STATE(vap_cntl_status, always, 2, 0);
 	r300->hw.vap_cntl_status.cmd[0] = cmdpacket0(R300_VAP_CNTL_STATUS, 1);
 	ALLOC_STATE(vir[0], variable, R300_VIR_CMDSIZE, 0);
 	r300->hw.vir[0].cmd[R300_VIR_CMD_0] =
-	    cmdpacket0(R300_VAP_INPUT_ROUTE_0_0, 1);
+	    cmdpacket0(R300_VAP_PROG_STREAM_CNTL_0, 1);
 	ALLOC_STATE(vir[1], variable, R300_VIR_CMDSIZE, 1);
 	r300->hw.vir[1].cmd[R300_VIR_CMD_0] =
-	    cmdpacket0(R300_VAP_INPUT_ROUTE_1_0, 1);
+	    cmdpacket0(R300_VAP_PROG_STREAM_CNTL_EXT_0, 1);
 	ALLOC_STATE(vic, always, R300_VIC_CMDSIZE, 0);
-	r300->hw.vic.cmd[R300_VIC_CMD_0] = cmdpacket0(R300_VAP_INPUT_CNTL_0, 2);
-	ALLOC_STATE(unk21DC, always, 2, 0);
-	r300->hw.unk21DC.cmd[0] = cmdpacket0(0x21DC, 1);
-	ALLOC_STATE(unk221C, always, 2, 0);
-	r300->hw.unk221C.cmd[0] = cmdpacket0(R300_VAP_UNKNOWN_221C, 1);
-	ALLOC_STATE(vap_clip, always, 5, 0);
-	r300->hw.vap_clip.cmd[0] = cmdpacket0(R300_VAP_CLIP_X_0, 4);
+	r300->hw.vic.cmd[R300_VIC_CMD_0] = cmdpacket0(R300_VAP_VTX_STATE_CNTL, 2);
+	ALLOC_STATE(vap_psc_sgn_norm_cntl, always, 2, 0);
+	r300->hw.vap_psc_sgn_norm_cntl.cmd[0] = cmdpacket0(R300_VAP_PSC_SGN_NORM_CNTL, SGN_NORM_ZERO_CLAMP_MINUS_ONE);
 
 	if (has_tcl) {
-		ALLOC_STATE(unk2288, always, 2, 0);
-		r300->hw.unk2288.cmd[0] = cmdpacket0(R300_VAP_UNKNOWN_2288, 1);
+		ALLOC_STATE(vap_clip_cntl, always, 2, 0);
+		r300->hw.vap_clip_cntl.cmd[0] = cmdpacket0(R300_VAP_CLIP_CNTL, 1);
+		ALLOC_STATE(vap_clip, always, 5, 0);
+		r300->hw.vap_clip.cmd[0] = cmdpacket0(R300_VAP_GB_VERT_CLIP_ADJ, 4);
+		ALLOC_STATE(vap_pvs_vtx_timeout_reg, always, 2, 0);
+		r300->hw.vap_pvs_vtx_timeout_reg.cmd[0] = cmdpacket0(VAP_PVS_VTX_TIMEOUT_REG, 1);
 	}
 
 	ALLOC_STATE(vof, always, R300_VOF_CMDSIZE, 0);
@@ -334,7 +361,7 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	if (has_tcl) {
 		ALLOC_STATE(pvs, always, R300_PVS_CMDSIZE, 0);
 		r300->hw.pvs.cmd[R300_PVS_CMD_0] =
-		    cmdpacket0(R300_VAP_PVS_CNTL_1, 3);
+		    cmdpacket0(R300_VAP_PVS_CODE_CNTL_0, 3);
 	}
 
 	ALLOC_STATE(gb_enable, always, 2, 0);
@@ -343,121 +370,183 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	r300->hw.gb_misc.cmd[0] = cmdpacket0(R300_GB_MSPOS0, 5);
 	ALLOC_STATE(txe, always, R300_TXE_CMDSIZE, 0);
 	r300->hw.txe.cmd[R300_TXE_CMD_0] = cmdpacket0(R300_TX_ENABLE, 1);
-	ALLOC_STATE(unk4200, always, 5, 0);
-	r300->hw.unk4200.cmd[0] = cmdpacket0(0x4200, 4);
-	ALLOC_STATE(unk4214, always, 2, 0);
-	r300->hw.unk4214.cmd[0] = cmdpacket0(0x4214, 1);
+	ALLOC_STATE(ga_point_s0, always, 5, 0);
+	r300->hw.ga_point_s0.cmd[0] = cmdpacket0(R300_GA_POINT_S0, 4);
+	ALLOC_STATE(ga_triangle_stipple, always, 2, 0);
+	r300->hw.ga_triangle_stipple.cmd[0] = cmdpacket0(R300_GA_TRIANGLE_STIPPLE, 1);
 	ALLOC_STATE(ps, always, R300_PS_CMDSIZE, 0);
-	r300->hw.ps.cmd[0] = cmdpacket0(R300_RE_POINTSIZE, 1);
-	ALLOC_STATE(unk4230, always, 4, 0);
-	r300->hw.unk4230.cmd[0] = cmdpacket0(0x4230, 3);
+	r300->hw.ps.cmd[0] = cmdpacket0(R300_GA_POINT_SIZE, 1);
+	ALLOC_STATE(ga_point_minmax, always, 4, 0);
+	r300->hw.ga_point_minmax.cmd[0] = cmdpacket0(R300_GA_POINT_MINMAX, 3);
 	ALLOC_STATE(lcntl, always, 2, 0);
-	r300->hw.lcntl.cmd[0] = cmdpacket0(R300_RE_LINE_CNT, 1);
-	ALLOC_STATE(unk4260, always, 4, 0);
-	r300->hw.unk4260.cmd[0] = cmdpacket0(0x4260, 3);
+	r300->hw.lcntl.cmd[0] = cmdpacket0(R300_GA_LINE_CNTL, 1);
+	ALLOC_STATE(ga_line_stipple, always, 4, 0);
+	r300->hw.ga_line_stipple.cmd[0] = cmdpacket0(R300_GA_LINE_STIPPLE_VALUE, 3);
 	ALLOC_STATE(shade, always, 5, 0);
-	r300->hw.shade.cmd[0] = cmdpacket0(R300_RE_SHADE, 4);
+	r300->hw.shade.cmd[0] = cmdpacket0(R300_GA_ENHANCE, 4);
 	ALLOC_STATE(polygon_mode, always, 4, 0);
-	r300->hw.polygon_mode.cmd[0] = cmdpacket0(R300_RE_POLYGON_MODE, 3);
+	r300->hw.polygon_mode.cmd[0] = cmdpacket0(R300_GA_POLY_MODE, 3);
 	ALLOC_STATE(fogp, always, 3, 0);
-	r300->hw.fogp.cmd[0] = cmdpacket0(R300_RE_FOG_SCALE, 2);
+	r300->hw.fogp.cmd[0] = cmdpacket0(R300_GA_FOG_SCALE, 2);
 	ALLOC_STATE(zbias_cntl, always, 2, 0);
-	r300->hw.zbias_cntl.cmd[0] = cmdpacket0(R300_RE_ZBIAS_CNTL, 1);
+	r300->hw.zbias_cntl.cmd[0] = cmdpacket0(R300_SU_TEX_WRAP, 1);
 	ALLOC_STATE(zbs, always, R300_ZBS_CMDSIZE, 0);
 	r300->hw.zbs.cmd[R300_ZBS_CMD_0] =
-	    cmdpacket0(R300_RE_ZBIAS_T_FACTOR, 4);
+	    cmdpacket0(R300_SU_POLY_OFFSET_FRONT_SCALE, 4);
 	ALLOC_STATE(occlusion_cntl, always, 2, 0);
-	r300->hw.occlusion_cntl.cmd[0] = cmdpacket0(R300_RE_OCCLUSION_CNTL, 1);
+	r300->hw.occlusion_cntl.cmd[0] = cmdpacket0(R300_SU_POLY_OFFSET_ENABLE, 1);
 	ALLOC_STATE(cul, always, R300_CUL_CMDSIZE, 0);
-	r300->hw.cul.cmd[R300_CUL_CMD_0] = cmdpacket0(R300_RE_CULL_CNTL, 1);
-	ALLOC_STATE(unk42C0, always, 3, 0);
-	r300->hw.unk42C0.cmd[0] = cmdpacket0(0x42C0, 2);
+	r300->hw.cul.cmd[R300_CUL_CMD_0] = cmdpacket0(R300_SU_CULL_MODE, 1);
+	ALLOC_STATE(su_depth_scale, always, 3, 0);
+	r300->hw.su_depth_scale.cmd[0] = cmdpacket0(R300_SU_DEPTH_SCALE, 2);
 	ALLOC_STATE(rc, always, R300_RC_CMDSIZE, 0);
-	r300->hw.rc.cmd[R300_RC_CMD_0] = cmdpacket0(R300_RS_CNTL_0, 2);
-	ALLOC_STATE(ri, always, R300_RI_CMDSIZE, 0);
-	r300->hw.ri.cmd[R300_RI_CMD_0] = cmdpacket0(R300_RS_INTERP_0, 8);
-	ALLOC_STATE(rr, variable, R300_RR_CMDSIZE, 0);
-	r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R300_RS_ROUTE_0, 1);
-	ALLOC_STATE(unk43A4, always, 3, 0);
-	r300->hw.unk43A4.cmd[0] = cmdpacket0(0x43A4, 2);
-	ALLOC_STATE(unk43E8, always, 2, 0);
-	r300->hw.unk43E8.cmd[0] = cmdpacket0(0x43E8, 1);
-	ALLOC_STATE(fp, always, R300_FP_CMDSIZE, 0);
-	r300->hw.fp.cmd[R300_FP_CMD_0] = cmdpacket0(R300_PFS_CNTL_0, 3);
-	r300->hw.fp.cmd[R300_FP_CMD_1] = cmdpacket0(R300_PFS_NODE_0, 4);
-	ALLOC_STATE(fpt, variable, R300_FPT_CMDSIZE, 0);
-	r300->hw.fpt.cmd[R300_FPT_CMD_0] = cmdpacket0(R300_PFS_TEXI_0, 0);
-	ALLOC_STATE(unk46A4, always, 6, 0);
-	r300->hw.unk46A4.cmd[0] = cmdpacket0(0x46A4, 5);
-	ALLOC_STATE(fpi[0], variable, R300_FPI_CMDSIZE, 0);
-	r300->hw.fpi[0].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR0_0, 1);
-	ALLOC_STATE(fpi[1], variable, R300_FPI_CMDSIZE, 1);
-	r300->hw.fpi[1].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR1_0, 1);
-	ALLOC_STATE(fpi[2], variable, R300_FPI_CMDSIZE, 2);
-	r300->hw.fpi[2].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR2_0, 1);
-	ALLOC_STATE(fpi[3], variable, R300_FPI_CMDSIZE, 3);
-	r300->hw.fpi[3].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR3_0, 1);
+	r300->hw.rc.cmd[R300_RC_CMD_0] = cmdpacket0(R300_RS_COUNT, 2);
+	if (is_r500) {
+		ALLOC_STATE(ri, always, R500_RI_CMDSIZE, 0);
+		r300->hw.ri.cmd[R300_RI_CMD_0] = cmdpacket0(R500_RS_IP_0, 16);
+		for (i = 0; i < 8; i++) {
+			r300->hw.ri.cmd[R300_RI_CMD_0 + i +1] =
+			  (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_S_SHIFT) |
+                          (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_T_SHIFT) |
+                          (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_R_SHIFT) |
+                          (R500_RS_IP_PTR_K1 << R500_RS_IP_TEX_PTR_Q_SHIFT);
+		}
+		ALLOC_STATE(rr, variable, R300_RR_CMDSIZE, 0);
+		r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R500_RS_INST_0, 1);
+	} else {
+		ALLOC_STATE(ri, always, R300_RI_CMDSIZE, 0);
+		r300->hw.ri.cmd[R300_RI_CMD_0] = cmdpacket0(R300_RS_IP_0, 8);
+		ALLOC_STATE(rr, variable, R300_RR_CMDSIZE, 0);
+		r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R300_RS_INST_0, 1);
+	}
+	ALLOC_STATE(sc_hyperz, always, 3, 0);
+	r300->hw.sc_hyperz.cmd[0] = cmdpacket0(R300_SC_HYPERZ, 2);
+	ALLOC_STATE(sc_screendoor, always, 2, 0);
+	r300->hw.sc_screendoor.cmd[0] = cmdpacket0(R300_SC_SCREENDOOR, 1);
+	ALLOC_STATE(us_out_fmt, always, 6, 0);
+	r300->hw.us_out_fmt.cmd[0] = cmdpacket0(R300_US_OUT_FMT, 5);
+
+	if (is_r500) {
+		ALLOC_STATE(fp, always, R500_FP_CMDSIZE, 0);
+		r300->hw.fp.cmd[R500_FP_CMD_0] = cmdpacket0(R500_US_CONFIG, 2);
+		r300->hw.fp.cmd[R500_FP_CNTL] = R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO;
+		r300->hw.fp.cmd[R500_FP_CMD_1] = cmdpacket0(R500_US_CODE_ADDR, 3);
+		r300->hw.fp.cmd[R500_FP_CMD_2] = cmdpacket0(R500_US_FC_CTRL, 1);
+		r300->hw.fp.cmd[R500_FP_FC_CNTL] = 0; /* FIXME when we add flow control */
+
+		ALLOC_STATE(r500fp, r500fp, R500_FPI_CMDSIZE, 0);
+		r300->hw.r500fp.cmd[R300_FPI_CMD_0] = cmdr500fp(0, 0, 0, 0);
+		ALLOC_STATE(r500fp_const, r500fp_const, R500_FPP_CMDSIZE, 0);
+		r300->hw.r500fp_const.cmd[R300_FPI_CMD_0] = cmdr500fp(0, 0, 1, 0);
+	} else {
+		ALLOC_STATE(fp, always, R300_FP_CMDSIZE, 0);
+		r300->hw.fp.cmd[R300_FP_CMD_0] = cmdpacket0(R300_US_CONFIG, 3);
+		r300->hw.fp.cmd[R300_FP_CMD_1] = cmdpacket0(R300_US_CODE_ADDR_0, 4);
+		ALLOC_STATE(fpt, variable, R300_FPT_CMDSIZE, 0);
+		r300->hw.fpt.cmd[R300_FPT_CMD_0] = cmdpacket0(R300_US_TEX_INST_0, 0);
+
+		ALLOC_STATE(fpi[0], variable, R300_FPI_CMDSIZE, 0);
+		r300->hw.fpi[0].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_RGB_INST_0, 1);
+		ALLOC_STATE(fpi[1], variable, R300_FPI_CMDSIZE, 1);
+		r300->hw.fpi[1].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_RGB_ADDR_0, 1);
+		ALLOC_STATE(fpi[2], variable, R300_FPI_CMDSIZE, 2);
+		r300->hw.fpi[2].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_ALPHA_INST_0, 1);
+		ALLOC_STATE(fpi[3], variable, R300_FPI_CMDSIZE, 3);
+		r300->hw.fpi[3].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_ALPHA_ADDR_0, 1);
+		ALLOC_STATE(fpp, variable, R300_FPP_CMDSIZE, 0);
+		r300->hw.fpp.cmd[R300_FPP_CMD_0] = cmdpacket0(R300_PFS_PARAM_0_X, 0);
+	}
 	ALLOC_STATE(fogs, always, R300_FOGS_CMDSIZE, 0);
-	r300->hw.fogs.cmd[R300_FOGS_CMD_0] = cmdpacket0(R300_RE_FOG_STATE, 1);
+	r300->hw.fogs.cmd[R300_FOGS_CMD_0] = cmdpacket0(R300_FG_FOG_BLEND, 1);
 	ALLOC_STATE(fogc, always, R300_FOGC_CMDSIZE, 0);
-	r300->hw.fogc.cmd[R300_FOGC_CMD_0] = cmdpacket0(R300_FOG_COLOR_R, 3);
+	r300->hw.fogc.cmd[R300_FOGC_CMD_0] = cmdpacket0(R300_FG_FOG_COLOR_R, 3);
 	ALLOC_STATE(at, always, R300_AT_CMDSIZE, 0);
-	r300->hw.at.cmd[R300_AT_CMD_0] = cmdpacket0(R300_PP_ALPHA_TEST, 2);
-	ALLOC_STATE(unk4BD8, always, 2, 0);
-	r300->hw.unk4BD8.cmd[0] = cmdpacket0(0x4BD8, 1);
-	ALLOC_STATE(fpp, variable, R300_FPP_CMDSIZE, 0);
-	r300->hw.fpp.cmd[R300_FPP_CMD_0] = cmdpacket0(R300_PFS_PARAM_0_X, 0);
-	ALLOC_STATE(unk4E00, always, 2, 0);
-	r300->hw.unk4E00.cmd[0] = cmdpacket0(0x4E00, 1);
+	r300->hw.at.cmd[R300_AT_CMD_0] = cmdpacket0(R300_FG_ALPHA_FUNC, 2);
+	ALLOC_STATE(fg_depth_src, always, 2, 0);
+	r300->hw.fg_depth_src.cmd[0] = cmdpacket0(R300_FG_DEPTH_SRC, 1);
+	ALLOC_STATE(rb3d_cctl, always, 2, 0);
+	r300->hw.rb3d_cctl.cmd[0] = cmdpacket0(R300_RB3D_CCTL, 1);
 	ALLOC_STATE(bld, always, R300_BLD_CMDSIZE, 0);
 	r300->hw.bld.cmd[R300_BLD_CMD_0] = cmdpacket0(R300_RB3D_CBLEND, 2);
 	ALLOC_STATE(cmk, always, R300_CMK_CMDSIZE, 0);
-	r300->hw.cmk.cmd[R300_CMK_CMD_0] = cmdpacket0(R300_RB3D_COLORMASK, 1);
-	ALLOC_STATE(blend_color, always, 4, 0);
-	r300->hw.blend_color.cmd[0] = cmdpacket0(R300_RB3D_BLEND_COLOR, 3);
+	r300->hw.cmk.cmd[R300_CMK_CMD_0] = cmdpacket0(RB3D_COLOR_CHANNEL_MASK, 1);
+	if (is_r500) {
+		ALLOC_STATE(blend_color, always, 3, 0);
+		r300->hw.blend_color.cmd[0] = cmdpacket0(R500_RB3D_CONSTANT_COLOR_AR, 2);
+	} else {
+		ALLOC_STATE(blend_color, always, 2, 0);
+		r300->hw.blend_color.cmd[0] = cmdpacket0(R300_RB3D_BLEND_COLOR, 1);
+	}
+	ALLOC_STATE(rop, always, 2, 0);
+	r300->hw.rop.cmd[0] = cmdpacket0(R300_RB3D_ROPCNTL, 1);
 	ALLOC_STATE(cb, always, R300_CB_CMDSIZE, 0);
 	r300->hw.cb.cmd[R300_CB_CMD_0] = cmdpacket0(R300_RB3D_COLOROFFSET0, 1);
 	r300->hw.cb.cmd[R300_CB_CMD_1] = cmdpacket0(R300_RB3D_COLORPITCH0, 1);
-	ALLOC_STATE(unk4E50, always, 10, 0);
-	r300->hw.unk4E50.cmd[0] = cmdpacket0(0x4E50, 9);
-	ALLOC_STATE(unk4E88, always, 2, 0);
-	r300->hw.unk4E88.cmd[0] = cmdpacket0(0x4E88, 1);
-	ALLOC_STATE(unk4EA0, always, 3, 0);
-	r300->hw.unk4EA0.cmd[0] = cmdpacket0(0x4EA0, 2);
+	ALLOC_STATE(rb3d_dither_ctl, always, 10, 0);
+	r300->hw.rb3d_dither_ctl.cmd[0] = cmdpacket0(R300_RB3D_DITHER_CTL, 9);
+	ALLOC_STATE(rb3d_aaresolve_ctl, always, 2, 0);
+	r300->hw.rb3d_aaresolve_ctl.cmd[0] = cmdpacket0(R300_RB3D_AARESOLVE_CTL, 1);
+	ALLOC_STATE(rb3d_discard_src_pixel_lte_threshold, always, 3, 0);
+	r300->hw.rb3d_discard_src_pixel_lte_threshold.cmd[0] = cmdpacket0(R500_RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD, 2);
 	ALLOC_STATE(zs, always, R300_ZS_CMDSIZE, 0);
 	r300->hw.zs.cmd[R300_ZS_CMD_0] =
-	    cmdpacket0(R300_RB3D_ZSTENCIL_CNTL_0, 3);
+	    cmdpacket0(R300_ZB_CNTL, 3);
 	ALLOC_STATE(zstencil_format, always, 5, 0);
 	r300->hw.zstencil_format.cmd[0] =
-	    cmdpacket0(R300_RB3D_ZSTENCIL_FORMAT, 4);
+	    cmdpacket0(R300_ZB_FORMAT, 4);
 	ALLOC_STATE(zb, always, R300_ZB_CMDSIZE, 0);
-	r300->hw.zb.cmd[R300_ZB_CMD_0] = cmdpacket0(R300_RB3D_DEPTHOFFSET, 2);
-	ALLOC_STATE(unk4F28, always, 2, 0);
-	r300->hw.unk4F28.cmd[0] = cmdpacket0(0x4F28, 1);
+	r300->hw.zb.cmd[R300_ZB_CMD_0] = cmdpacket0(R300_ZB_DEPTHOFFSET, 2);
+	ALLOC_STATE(zb_depthclearvalue, always, 2, 0);
+	r300->hw.zb_depthclearvalue.cmd[0] = cmdpacket0(R300_ZB_DEPTHCLEARVALUE, 1);
 	ALLOC_STATE(unk4F30, always, 3, 0);
 	r300->hw.unk4F30.cmd[0] = cmdpacket0(0x4F30, 2);
-	ALLOC_STATE(unk4F44, always, 2, 0);
-	r300->hw.unk4F44.cmd[0] = cmdpacket0(0x4F44, 1);
-	ALLOC_STATE(unk4F54, always, 2, 0);
-	r300->hw.unk4F54.cmd[0] = cmdpacket0(0x4F54, 1);
+	ALLOC_STATE(zb_hiz_offset, always, 2, 0);
+	r300->hw.zb_hiz_offset.cmd[0] = cmdpacket0(R300_ZB_HIZ_OFFSET, 1);
+	ALLOC_STATE(zb_hiz_pitch, always, 2, 0);
+	r300->hw.zb_hiz_pitch.cmd[0] = cmdpacket0(R300_ZB_HIZ_PITCH, 1);
 
 	/* VPU only on TCL */
 	if (has_tcl) {
+   	        int i;
 		ALLOC_STATE(vpi, vpu, R300_VPI_CMDSIZE, 0);
 		r300->hw.vpi.cmd[R300_VPI_CMD_0] =
-		    cmdvpu(R300_PVS_UPLOAD_PROGRAM, 0);
-		ALLOC_STATE(vpp, vpu, R300_VPP_CMDSIZE, 0);
-		r300->hw.vpp.cmd[R300_VPP_CMD_0] =
-		    cmdvpu(R300_PVS_UPLOAD_PARAMETERS, 0);
-		ALLOC_STATE(vps, vpu, R300_VPS_CMDSIZE, 0);
-		r300->hw.vps.cmd[R300_VPS_CMD_0] =
-		    cmdvpu(R300_PVS_UPLOAD_POINTSIZE, 1);
+		    cmdvpu(R300_PVS_CODE_START, 0);
+
+		if (is_r500) {
+		    ALLOC_STATE(vpp, vpu, R300_VPP_CMDSIZE, 0);
+		    r300->hw.vpp.cmd[R300_VPP_CMD_0] =
+			cmdvpu(R500_PVS_CONST_START, 0);
+
+		    ALLOC_STATE(vps, vpu, R300_VPS_CMDSIZE, 0);
+		    r300->hw.vps.cmd[R300_VPS_CMD_0] =
+			cmdvpu(R500_POINT_VPORT_SCALE_OFFSET, 1);
+
+			for (i = 0; i < 6; i++) {
+				ALLOC_STATE(vpucp[i], vpu, R300_VPUCP_CMDSIZE, 0);
+				r300->hw.vpucp[i].cmd[R300_VPUCP_CMD_0] =
+					cmdvpu(R500_PVS_UCP_START + i, 1);
+			}
+		} else {
+		    ALLOC_STATE(vpp, vpu, R300_VPP_CMDSIZE, 0);
+		    r300->hw.vpp.cmd[R300_VPP_CMD_0] =
+			cmdvpu(R300_PVS_CONST_START, 0);
+
+		    ALLOC_STATE(vps, vpu, R300_VPS_CMDSIZE, 0);
+		    r300->hw.vps.cmd[R300_VPS_CMD_0] =
+			cmdvpu(R300_POINT_VPORT_SCALE_OFFSET, 1);
+
+			for (i = 0; i < 6; i++) {
+				ALLOC_STATE(vpucp[i], vpu, R300_VPUCP_CMDSIZE, 0);
+				r300->hw.vpucp[i].cmd[R300_VPUCP_CMD_0] =
+					cmdvpu(R300_PVS_UCP_START + i, 1);
+			}
+		}
 	}
 
 	/* Textures */
 	ALLOC_STATE(tex.filter, variable, mtu + 1, 0);
 	r300->hw.tex.filter.cmd[R300_TEX_CMD_0] =
-	    cmdpacket0(R300_TX_FILTER_0, 0);
+	    cmdpacket0(R300_TX_FILTER0_0, 0);
 
 	ALLOC_STATE(tex.filter_1, variable, mtu + 1, 0);
 	r300->hw.tex.filter_1.cmd[R300_TEX_CMD_0] =
@@ -471,7 +560,7 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	    cmdpacket0(R300_TX_FORMAT_0, 0);
 
 	ALLOC_STATE(tex.pitch, variable, mtu + 1, 0);
-	r300->hw.tex.pitch.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_PITCH_0, 0);
+	r300->hw.tex.pitch.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_FORMAT2_0, 0);
 
 	ALLOC_STATE(tex.offset, variable, mtu + 1, 0);
 	r300->hw.tex.offset.cmd[R300_TEX_CMD_0] =
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.h b/src/mesa/drivers/dri/r300/r300_cmdbuf.h
index acb6e38c6df..a8eaa580bd9 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.h
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.h
@@ -52,7 +52,7 @@ extern void r300DestroyCmdBuf(r300ContextPtr r300);
  *
  * \param dwords The number of dwords we need to be free on the command buffer
  */
-static inline void r300EnsureCmdBufSpace(r300ContextPtr r300,
+static INLINE void r300EnsureCmdBufSpace(r300ContextPtr r300,
 					     int dwords, const char *caller)
 {
 	assert(dwords < r300->cmdbuf.size);
@@ -68,7 +68,7 @@ static inline void r300EnsureCmdBufSpace(r300ContextPtr r300,
  * causes state reemission after a flush. This is necessary to ensure
  * correct hardware state after an unlock.
  */
-static inline uint32_t *r300RawAllocCmdBuf(r300ContextPtr r300,
+static INLINE uint32_t *r300RawAllocCmdBuf(r300ContextPtr r300,
 					       int dwords, const char *caller)
 {
 	uint32_t *ptr;
@@ -80,7 +80,7 @@ static inline uint32_t *r300RawAllocCmdBuf(r300ContextPtr r300,
 	return ptr;
 }
 
-static inline uint32_t *r300AllocCmdBuf(r300ContextPtr r300,
+static INLINE uint32_t *r300AllocCmdBuf(r300ContextPtr r300,
 					    int dwords, const char *caller)
 {
 	uint32_t *ptr;
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 14e0f052fd5..37436275e34 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -35,15 +35,15 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Nicolai Haehnle <prefect_@gmx.net>
  */
 
-#include "glheader.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
-#include "state.h"
-#include "bufferobj.h"
+#include "main/glheader.h"
+#include "main/api_arrayelt.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
+#include "main/state.h"
+#include "main/bufferobj.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -79,11 +79,13 @@ int hw_tcl_on = 1;
 
 #define need_GL_EXT_stencil_two_side
 #define need_GL_ARB_multisample
+#define need_GL_ARB_point_parameters
 #define need_GL_ARB_texture_compression
 #define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
 #define need_GL_EXT_blend_minmax
 //#define need_GL_EXT_fog_coord
+#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_blend_equation_separate
 #define need_GL_EXT_blend_func_separate
@@ -93,8 +95,13 @@ int hw_tcl_on = 1;
 
 const struct dri_extension card_extensions[] = {
   /* *INDENT-OFF* */
+  {"GL_ARB_depth_texture",		NULL},
+  {"GL_ARB_fragment_program",		NULL},
   {"GL_ARB_multisample",		GL_ARB_multisample_functions},
   {"GL_ARB_multitexture",		NULL},
+  {"GL_ARB_point_parameters",		GL_ARB_point_parameters_functions},
+  {"GL_ARB_shadow",			NULL},
+  {"GL_ARB_shadow_ambient",		NULL},
   {"GL_ARB_texture_border_clamp",	NULL},
   {"GL_ARB_texture_compression",	GL_ARB_texture_compression_functions},
   {"GL_ARB_texture_cube_map",		NULL},
@@ -105,14 +112,15 @@ const struct dri_extension card_extensions[] = {
   {"GL_ARB_texture_mirrored_repeat",	NULL},
   {"GL_ARB_vertex_buffer_object",	GL_ARB_vertex_buffer_object_functions},
   {"GL_ARB_vertex_program",		GL_ARB_vertex_program_functions},
-  {"GL_ARB_fragment_program",		NULL},
   {"GL_EXT_blend_equation_separate",	GL_EXT_blend_equation_separate_functions},
   {"GL_EXT_blend_func_separate",	GL_EXT_blend_func_separate_functions},
   {"GL_EXT_blend_minmax",		GL_EXT_blend_minmax_functions},
   {"GL_EXT_blend_subtract",		NULL},
 //  {"GL_EXT_fog_coord",			GL_EXT_fog_coord_functions },
+  {"GL_EXT_multi_draw_arrays",		GL_EXT_multi_draw_arrays_functions},
   {"GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions},
   {"GL_EXT_secondary_color", 		GL_EXT_secondary_color_functions},
+  {"GL_EXT_shadow_funcs",		NULL},
   {"GL_EXT_stencil_two_side",		GL_EXT_stencil_two_side_functions},
   {"GL_EXT_stencil_wrap",		NULL},
   {"GL_EXT_texture_edge_clamp",		NULL},
@@ -273,6 +281,12 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	    MIN2(ctx->Const.MaxTextureImageUnits,
 		 ctx->Const.MaxTextureCoordUnits);
 	ctx->Const.MaxTextureMaxAnisotropy = 16.0;
+	ctx->Const.MaxTextureLodBias = 16.0;
+
+	if (screen->chip_family >= CHIP_FAMILY_RV515) {
+	    ctx->Const.MaxTextureLevels = 13;
+	    ctx->Const.MaxTextureRectSize = 4096;
+	}
 
 	ctx->Const.MinPointSize = 1.0;
 	ctx->Const.MinPointSizeAA = 1.0;
@@ -340,7 +354,7 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	ctx->Const.FragmentProgram.MaxNativeTexIndirections =
 	    PFS_MAX_TEX_INDIRECT;
 	ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0;	/* and these are?? */
-	_tnl_ProgramCacheInit(ctx);
+	ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
 	ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
 
 	driInitExtensions(ctx, card_extensions, GL_TRUE);
@@ -486,7 +500,6 @@ void r300DestroyContext(__DRIcontextPrivate * driContextPriv)
 		release_texture_heaps =
 		    (r300->radeon.glCtx->Shared->RefCount == 1);
 		_swsetup_DestroyContext(r300->radeon.glCtx);
-		_tnl_ProgramCacheDestroy(r300->radeon.glCtx);
 		_tnl_DestroyContext(r300->radeon.glCtx);
 		_vbo_DestroyContext(r300->radeon.glCtx);
 		_swrast_DestroyContext(r300->radeon.glCtx);
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index be6909724a3..c15e9fa3009 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -43,9 +43,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "dri_util.h"
 #include "texmem.h"
 
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
 
 #define USER_BUFFERS
 
@@ -54,7 +54,7 @@ typedef struct r300_context r300ContextRec;
 typedef struct r300_context *r300ContextPtr;
 
 #include "radeon_lock.h"
-#include "mm.h"
+#include "main/mm.h"
 
 /* From http://gcc.gnu.org/onlinedocs/gcc-3.2.3/gcc/Variadic-Macros.html .
    I suppose we could inline this and use macro to fetch out __LINE__ and stuff in case we run into trouble
@@ -73,12 +73,12 @@ typedef struct r300_context *r300ContextPtr;
 	}
 
 #include "r300_vertprog.h"
-#include "r300_fragprog.h"
+#include "r500_fragprog.h"
 
 /**
  * This function takes a float and packs it into a uint32_t
  */
-static inline uint32_t r300PackFloat32(float fl)
+static INLINE uint32_t r300PackFloat32(float fl)
 {
 	union {
 		float fl;
@@ -95,7 +95,7 @@ static inline uint32_t r300PackFloat32(float fl)
  * But it works for most things.  I'll fix it later if someone
  * else with a better clue doesn't
  */
-static inline uint32_t r300PackFloat24(float f)
+static INLINE uint32_t r300PackFloat24(float f)
 {
 	float mantissa;
 	int exponent;
@@ -178,13 +178,6 @@ struct r300_tex_obj {
 	GLuint bufAddr;		/* Offset to start of locally
 				   shared texture block */
 
-	GLuint dirty_state;	/* Flags (1 per texunit) for
-				   whether or not this texobj
-				   has dirty hardware state
-				   (pp_*) that needs to be
-				   brought into the
-				   texunit. */
-
 	drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
 	/* Six, for the cube faces */
 
@@ -330,15 +323,17 @@ struct r300_state_atom {
 #define R300_RI_INTERP_7	8
 #define R300_RI_CMDSIZE		9
 
+#define R500_RI_CMDSIZE	       17
+
 #define R300_RR_CMD_0		0	/* rr is variable size (at least 1) */
-#define R300_RR_ROUTE_0		1
-#define R300_RR_ROUTE_1		2
-#define R300_RR_ROUTE_2		3
-#define R300_RR_ROUTE_3		4
-#define R300_RR_ROUTE_4		5
-#define R300_RR_ROUTE_5		6
-#define R300_RR_ROUTE_6		7
-#define R300_RR_ROUTE_7		8
+#define R300_RR_INST_0		1
+#define R300_RR_INST_1		2
+#define R300_RR_INST_2		3
+#define R300_RR_INST_3		4
+#define R300_RR_INST_4		5
+#define R300_RR_INST_5		6
+#define R300_RR_INST_6		7
+#define R300_RR_INST_7		8
 #define R300_RR_CMDSIZE		9
 
 #define R300_FP_CMD_0		0
@@ -352,6 +347,17 @@ struct r300_state_atom {
 #define R300_FP_NODE3		8
 #define R300_FP_CMDSIZE		9
 
+#define R500_FP_CMD_0           0
+#define R500_FP_CNTL            1
+#define R500_FP_PIXSIZE         2
+#define R500_FP_CMD_1           3
+#define R500_FP_CODE_ADDR       4
+#define R500_FP_CODE_RANGE      5
+#define R500_FP_CODE_OFFSET     6
+#define R500_FP_CMD_2           7
+#define R500_FP_FC_CNTL         8
+#define R500_FP_CMDSIZE         9
+
 #define R300_FPT_CMD_0		0
 #define R300_FPT_INSTR_0	1
 #define R300_FPT_CMDSIZE	65
@@ -359,10 +365,14 @@ struct r300_state_atom {
 #define R300_FPI_CMD_0		0
 #define R300_FPI_INSTR_0	1
 #define R300_FPI_CMDSIZE	65
+/* R500 has space for 512 instructions - 6 dwords per instruction */
+#define R500_FPI_CMDSIZE	(512*6+1)
 
 #define R300_FPP_CMD_0		0
 #define R300_FPP_PARAM_0	1
 #define R300_FPP_CMDSIZE	(32*4+1)
+/* R500 has spcae for 256 constants - 4 dwords per constant */
+#define R500_FPP_CMDSIZE	(256*4+1)
 
 #define R300_FOGS_CMD_0		0
 #define R300_FOGS_STATE		1
@@ -410,6 +420,12 @@ struct r300_state_atom {
 #define R300_ZB_PITCH		2
 #define R300_ZB_CMDSIZE		3
 
+#define R300_VAP_CNTL_FLUSH     0
+#define R300_VAP_CNTL_FLUSH_1   1
+#define R300_VAP_CNTL_CMD       2
+#define R300_VAP_CNTL_INSTR     3
+#define R300_VAP_CNTL_SIZE      4
+
 #define R300_VPI_CMD_0		0
 #define R300_VPI_INSTR_0	1
 #define R300_VPI_CMDSIZE	1025	/* 256 16 byte instructions */
@@ -418,6 +434,13 @@ struct r300_state_atom {
 #define R300_VPP_PARAM_0	1
 #define R300_VPP_CMDSIZE	1025	/* 256 4-component parameters */
 
+#define R300_VPUCP_CMD_0		0
+#define R300_VPUCP_X            1
+#define R300_VPUCP_Y            2
+#define R300_VPUCP_Z            3
+#define R300_VPUCP_W            4
+#define R300_VPUCP_CMDSIZE	5	/* 256 4-component parameters */
+
 #define R300_VPS_CMD_0		0
 #define R300_VPS_ZERO_0		1
 #define R300_VPS_ZERO_1		2
@@ -444,67 +467,72 @@ struct r300_hw_state {
 
 	struct r300_state_atom vpt;	/* viewport (1D98) */
 	struct r300_state_atom vap_cntl;
+        struct r300_state_atom vap_index_offset; /* 0x208c r5xx only */
 	struct r300_state_atom vof;	/* VAP output format register 0x2090 */
 	struct r300_state_atom vte;	/* (20B0) */
-	struct r300_state_atom unk2134;	/* (2134) */
+	struct r300_state_atom vap_vf_max_vtx_indx;	/* Maximum Vertex Indx Clamp (2134) */
 	struct r300_state_atom vap_cntl_status;
 	struct r300_state_atom vir[2];	/* vap input route (2150/21E0) */
 	struct r300_state_atom vic;	/* vap input control (2180) */
-	struct r300_state_atom unk21DC;	/* (21DC) */
-	struct r300_state_atom unk221C;	/* (221C) */
+	struct r300_state_atom vap_psc_sgn_norm_cntl; /* Programmable Stream Control Signed Normalize Control (21DC) */
+	struct r300_state_atom vap_clip_cntl;
 	struct r300_state_atom vap_clip;
-	struct r300_state_atom unk2288;	/* (2288) */
+	struct r300_state_atom vap_pvs_vtx_timeout_reg;	/* Vertex timeout register (2288) */
 	struct r300_state_atom pvs;	/* pvs_cntl (22D0) */
 	struct r300_state_atom gb_enable;	/* (4008) */
 	struct r300_state_atom gb_misc;	/* Multisampling position shifts ? (4010) */
-	struct r300_state_atom unk4200;	/* (4200) */
-	struct r300_state_atom unk4214;	/* (4214) */
+	struct r300_state_atom ga_point_s0;	/* S Texture Coordinate of Vertex 0 for Point texture stuffing (LLC) (4200) */
+	struct r300_state_atom ga_triangle_stipple;	/* (4214) */
 	struct r300_state_atom ps;	/* pointsize (421C) */
-	struct r300_state_atom unk4230;	/* (4230) */
+	struct r300_state_atom ga_point_minmax;	/* (4230) */
 	struct r300_state_atom lcntl;	/* line control */
-	struct r300_state_atom unk4260;	/* (4260) */
+	struct r300_state_atom ga_line_stipple;	/* (4260) */
 	struct r300_state_atom shade;
 	struct r300_state_atom polygon_mode;
 	struct r300_state_atom fogp;	/* fog parameters (4294) */
-	struct r300_state_atom unk429C;	/* (429C) */
+	struct r300_state_atom ga_soft_reset;	/* (429C) */
 	struct r300_state_atom zbias_cntl;
 	struct r300_state_atom zbs;	/* zbias (42A4) */
 	struct r300_state_atom occlusion_cntl;
 	struct r300_state_atom cul;	/* cull cntl (42B8) */
-	struct r300_state_atom unk42C0;	/* (42C0) */
+	struct r300_state_atom su_depth_scale;	/* (42C0) */
 	struct r300_state_atom rc;	/* rs control (4300) */
 	struct r300_state_atom ri;	/* rs interpolators (4310) */
 	struct r300_state_atom rr;	/* rs route (4330) */
-	struct r300_state_atom unk43A4;	/* (43A4) */
-	struct r300_state_atom unk43E8;	/* (43E8) */
+	struct r300_state_atom sc_hyperz;	/* (43A4) */
+	struct r300_state_atom sc_screendoor;	/* (43E8) */
 	struct r300_state_atom fp;	/* fragment program cntl + nodes (4600) */
 	struct r300_state_atom fpt;	/* texi - (4620) */
-	struct r300_state_atom unk46A4;	/* (46A4) */
+	struct r300_state_atom us_out_fmt;	/* (46A4) */
+	struct r300_state_atom r500fp;	/* r500 fp instructions */
+	struct r300_state_atom r500fp_const;	/* r500 fp constants */
 	struct r300_state_atom fpi[4];	/* fp instructions (46C0/47C0/48C0/49C0) */
 	struct r300_state_atom fogs;	/* fog state (4BC0) */
 	struct r300_state_atom fogc;	/* fog color (4BC8) */
 	struct r300_state_atom at;	/* alpha test (4BD4) */
-	struct r300_state_atom unk4BD8;	/* (4BD8) */
+	struct r300_state_atom fg_depth_src;	/* (4BD8) */
 	struct r300_state_atom fpp;	/* 0x4C00 and following */
-	struct r300_state_atom unk4E00;	/* (4E00) */
+	struct r300_state_atom rb3d_cctl;	/* (4E00) */
 	struct r300_state_atom bld;	/* blending (4E04) */
 	struct r300_state_atom cmk;	/* colormask (4E0C) */
 	struct r300_state_atom blend_color;	/* constant blend color */
+	struct r300_state_atom rop;	/* ropcntl */
 	struct r300_state_atom cb;	/* colorbuffer (4E28) */
-	struct r300_state_atom unk4E50;	/* (4E50) */
-	struct r300_state_atom unk4E88;	/* (4E88) */
-	struct r300_state_atom unk4EA0;	/* (4E88) I saw it only written on RV350 hardware..  */
+	struct r300_state_atom rb3d_dither_ctl;	/* (4E50) */
+	struct r300_state_atom rb3d_aaresolve_ctl;	/* (4E88) */
+	struct r300_state_atom rb3d_discard_src_pixel_lte_threshold;	/* (4E88) I saw it only written on RV350 hardware..  */
 	struct r300_state_atom zs;	/* zstencil control (4F00) */
 	struct r300_state_atom zstencil_format;
 	struct r300_state_atom zb;	/* z buffer (4F20) */
-	struct r300_state_atom unk4F28;	/* (4F28) */
+	struct r300_state_atom zb_depthclearvalue;	/* (4F28) */
 	struct r300_state_atom unk4F30;	/* (4F30) */
-	struct r300_state_atom unk4F44;	/* (4F44) */
-	struct r300_state_atom unk4F54;	/* (4F54) */
+	struct r300_state_atom zb_hiz_offset;	/* (4F44) */
+	struct r300_state_atom zb_hiz_pitch;	/* (4F54) */
 
 	struct r300_state_atom vpi;	/* vp instructions */
 	struct r300_state_atom vpp;	/* vp parameters */
 	struct r300_state_atom vps;	/* vertex point size (?) */
+	struct r300_state_atom vpucp[6];	/* vp user clip plane - 6 */
 	/* 8 texture units */
 	/* the state is grouped by function and not by
 	   texture unit. This makes single unit updates
@@ -546,9 +574,7 @@ struct r300_depthbuffer_state {
 };
 
 struct r300_stencilbuffer_state {
-	GLuint clear;
 	GLboolean hw_stencil;
-
 };
 
 /* Vertex shader state */
@@ -596,6 +622,7 @@ extern int hw_tcl_on;
 struct r300_vertex_program_key {
 	GLuint InputsRead;
 	GLuint OutputsWritten;
+	GLuint OutputsAdded;
 };
 
 struct r300_vertex_program {
@@ -627,82 +654,75 @@ struct r300_vertex_program_cont {
 #define PFS_NUM_TEMP_REGS	32
 #define PFS_NUM_CONST_REGS	16
 
-/* Mapping Mesa registers to R300 temporaries */
-struct reg_acc {
-	int reg;		/* Assigned hw temp */
-	unsigned int refcount;	/* Number of uses by mesa program */
-};
+struct r300_pfs_compile_state;
+
 
 /**
- * Describe the current lifetime information for an R300 temporary
+ * Stores state that influences the compilation of a fragment program.
  */
-struct reg_lifetime {
-	/* Index of the first slot where this register is free in the sense
-	   that it can be used as a new destination register.
-	   This is -1 if the register has been assigned to a Mesa register
-	   and the last access to the register has not yet been emitted */
-	int free;
-
-	/* Index of the first slot where this register is currently reserved.
-	   This is used to stop e.g. a scalar operation from being moved
-	   before the allocation time of a register that was first allocated
-	   for a vector operation. */
-	int reserved;
-
-	/* Index of the first slot in which the register can be used as a
-	   source without losing the value that is written by the last
-	   emitted instruction that writes to the register */
-	int vector_valid;
-	int scalar_valid;
-
-	/* Index to the slot where the register was last read.
-	   This is also the first slot in which the register may be written again */
-	int vector_lastread;
-	int scalar_lastread;
+struct r300_fragment_program_external_state {
+	struct {
+		/**
+		 * If the sampler is used as a shadow sampler,
+		 * this field is:
+		 *  0 - GL_LUMINANCE
+		 *  1 - GL_INTENSITY
+		 *  2 - GL_ALPHA
+		 * depending on the depth texture mode.
+		 */
+		GLuint depth_texture_mode : 2;
+
+		/**
+		 * If the sampler is used as a shadow sampler,
+		 * this field is (texture_compare_func - GL_NEVER).
+		 * [e.g. if compare function is GL_LEQUAL, this field is 3]
+		 *
+		 * Otherwise, this field is 0.
+		 */
+		GLuint texture_compare_func : 3;
+	} unit[16];
 };
 
-/**
- * Store usage information about an ALU instruction slot during the
- * compilation of a fragment program.
- */
-#define SLOT_SRC_VECTOR  (1<<0)
-#define SLOT_SRC_SCALAR  (1<<3)
-#define SLOT_SRC_BOTH    (SLOT_SRC_VECTOR | SLOT_SRC_SCALAR)
-#define SLOT_OP_VECTOR   (1<<16)
-#define SLOT_OP_SCALAR   (1<<17)
-#define SLOT_OP_BOTH     (SLOT_OP_VECTOR | SLOT_OP_SCALAR)
-
-struct r300_pfs_compile_slot {
-	/* Bitmask indicating which parts of the slot are used, using SLOT_ constants
-	   defined above */
-	unsigned int used;
-
-	/* Selected sources */
-	int vsrc[3];
-	int ssrc[3];
+
+struct r300_fragment_program_node {
+	int tex_offset; /**< first tex instruction */
+	int tex_end; /**< last tex instruction, relative to tex_offset */
+	int alu_offset; /**< first ALU instruction */
+	int alu_end; /**< last ALU instruction, relative to alu_offset */
+	int flags;
 };
 
 /**
- * Store information during compilation of fragment programs.
+ * Stores an R300 fragment program in its compiled-to-hardware form.
  */
-struct r300_pfs_compile_state {
-	int nrslots;		/* number of ALU slots used so far */
+struct r300_fragment_program_code {
+	struct {
+		int length; /**< total # of texture instructions used */
+		GLuint inst[PFS_MAX_TEX_INST];
+	} tex;
 
-	/* Track which (parts of) slots are already filled with instructions */
-	struct r300_pfs_compile_slot slot[PFS_MAX_ALU_INST];
+	struct {
+		int length; /**< total # of ALU instructions used */
+		struct {
+			GLuint inst0;
+			GLuint inst1;
+			GLuint inst2;
+			GLuint inst3;
+		} inst[PFS_MAX_ALU_INST];
+	} alu;
 
-	/* Track the validity of R300 temporaries */
-	struct reg_lifetime hwtemps[PFS_NUM_TEMP_REGS];
+	struct r300_fragment_program_node node[4];
+	int cur_node;
+	int first_node_has_tex;
 
-	/* Used to map Mesa's inputs/temps onto hardware temps */
-	int temp_in_use;
-	struct reg_acc temps[PFS_NUM_TEMP_REGS];
-	struct reg_acc inputs[32];	/* don't actually need 32... */
+	/**
+	 * Remember which program register a given hardware constant
+	 * belongs to.
+	 */
+	struct prog_src_register constant[PFS_NUM_CONST_REGS];
+	int const_nr;
 
-	/* Track usage of hardware temps, for register allocation,
-	 * indirection detection, etc. */
-	GLuint used_in_node;
-	GLuint dest_in_node;
+	int max_temp_idx;
 };
 
 /**
@@ -712,51 +732,75 @@ struct r300_pfs_compile_state {
 struct r300_fragment_program {
 	struct gl_fragment_program mesa_program;
 
-	GLcontext *ctx;
 	GLboolean translated;
 	GLboolean error;
-	struct r300_pfs_compile_state *cs;
 
-	struct {
-		int length;
-		GLuint inst[PFS_MAX_TEX_INST];
-	} tex;
+	struct r300_fragment_program_external_state state;
+	struct r300_fragment_program_code code;
 
-	struct {
-		struct {
-			GLuint inst0;
-			GLuint inst1;
-			GLuint inst2;
-			GLuint inst3;
-		} inst[PFS_MAX_ALU_INST];
-	} alu;
+	GLboolean WritesDepth;
+	GLuint optimization;
+};
+
+struct r500_pfs_compile_state;
 
+struct r500_fragment_program_external_state {
 	struct {
-		int tex_offset;
-		int tex_end;
-		int alu_offset;
-		int alu_end;
-		int flags;
-	} node[4];
-	int cur_node;
-	int first_node_has_tex;
+		/**
+		 * If the sampler is used as a shadow sampler,
+		 * this field is:
+		 *  0 - GL_LUMINANCE
+		 *  1 - GL_INTENSITY
+		 *  2 - GL_ALPHA
+		 * depending on the depth texture mode.
+		 */
+		GLuint depth_texture_mode : 2;
+
+		/**
+		 * If the sampler is used as a shadow sampler,
+		 * this field is (texture_compare_func - GL_NEVER).
+		 * [e.g. if compare function is GL_LEQUAL, this field is 3]
+		 *
+		 * Otherwise, this field is 0.
+		 */
+		GLuint texture_compare_func : 3;
+	} unit[16];
+};
 
-	int alu_offset;
-	int alu_end;
-	int tex_offset;
-	int tex_end;
-
-	/* Hardware constants.
-	 * Contains a pointer to the value. The destination of the pointer
-	 * is supposed to be updated when GL state changes.
-	 * Typically, this is either a pointer into
-	 * gl_program_parameter_list::ParameterValues, or a pointer to a
-	 * global constant (e.g. for sin/cos-approximation)
+struct r500_fragment_program_code {
+	struct {
+		GLuint inst0;
+		GLuint inst1;
+		GLuint inst2;
+		GLuint inst3;
+		GLuint inst4;
+		GLuint inst5;
+	} inst[512];
+
+	int inst_offset;
+	int inst_end;
+
+	/**
+	 * Remember which program register a given hardware constant
+	 * belongs to.
 	 */
-	const GLfloat *constant[PFS_NUM_CONST_REGS];
+	struct prog_src_register constant[PFS_NUM_CONST_REGS];
 	int const_nr;
 
 	int max_temp_idx;
+};
+
+struct r500_fragment_program {
+	struct gl_fragment_program mesa_program;
+
+	GLcontext *ctx;
+	GLboolean translated;
+	GLboolean error;
+
+	struct r500_fragment_program_external_state state;
+	struct r500_fragment_program_code code;
+
+	GLboolean writes_depth;
 
 	GLuint optimization;
 };
@@ -772,7 +816,6 @@ struct r300_state {
 	struct r300_texture_state texture;
 	int sw_tcl_inputs[VERT_ATTRIB_MAX];
 	struct r300_vertex_shader_state vertex_shader;
-	struct r300_pfs_compile_state pfs_compile;
 	struct r300_dma_region aos[R300_MAX_AOS_ARRAYS];
 	int aos_count;
 
@@ -795,7 +838,7 @@ struct r300_state {
  */
 struct r300_swtcl_info {
    GLuint RenderIndex;
-   
+
    /**
     * Size of a hardware vertex.  This is calculated when \c ::vertex_attrs is
     * installed in the Mesa state vector.
diff --git a/src/mesa/drivers/dri/r300/r300_emit.c b/src/mesa/drivers/dri/r300/r300_emit.c
index 424bf44e595..80bd3389aef 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.c
+++ b/src/mesa/drivers/dri/r300/r300_emit.c
@@ -33,12 +33,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "imports.h"
-#include "macros.h"
-#include "image.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/image.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "math/m_translate.h"
@@ -207,49 +207,57 @@ static void r300EmitVec(GLcontext * ctx, struct r300_dma_region *rvb,
 	}
 }
 
-static GLuint r300VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
+#define DW_SIZE(x) ((inputs[tab[(x)]] << R300_DST_VEC_LOC_SHIFT) |	\
+		    (attribptr[tab[(x)]]->size - 1) << R300_DATA_TYPE_0_SHIFT)
+
+GLuint r300VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
 				 int *inputs, GLint * tab, GLuint nr)
 {
 	GLuint i, dw;
 
 	/* type, inputs, stop bit, size */
-	for (i = 0; i + 1 < nr; i += 2) {
-		dw = R300_INPUT_ROUTE_FLOAT | (inputs[tab[i]] << 8) | (attribptr[tab[i]]->size - 1);
-		dw |= (R300_INPUT_ROUTE_FLOAT | (inputs[tab[i + 1]] << 8) | (attribptr[tab[i + 1]]->size - 1)) << 16;
-		if (i + 2 == nr) {
-			dw |= (R300_VAP_INPUT_ROUTE_END << 16);
+	for (i = 0; i < nr; i += 2) {
+		/* make sure input is valid, would lockup the gpu */
+		assert(inputs[tab[i]] != -1);
+		dw = (R300_SIGNED | DW_SIZE(i));
+		if (i + 1 == nr) {
+			dw |= R300_LAST_VEC << R300_DATA_TYPE_0_SHIFT;
+		} else {
+			assert(inputs[tab[i + 1]] != -1);
+			dw |= (R300_SIGNED |
+			       DW_SIZE(i + 1)) << R300_DATA_TYPE_1_SHIFT;
+			if (i + 2 == nr) {
+				dw |= R300_LAST_VEC << R300_DATA_TYPE_1_SHIFT;
+			}
 		}
 		dst[i >> 1] = dw;
 	}
 
-	if (nr & 1) {
-		dw = R300_INPUT_ROUTE_FLOAT | (inputs[tab[nr - 1]] << 8) | (attribptr[tab[nr - 1]]->size - 1);
-		dw |= R300_VAP_INPUT_ROUTE_END;
-		dst[nr >> 1] = dw;
-	}
-
 	return (nr + 1) >> 1;
 }
 
 static GLuint r300VAPInputRoute1Swizzle(int swizzle[4])
 {
-	return (swizzle[0] << R300_INPUT_ROUTE_X_SHIFT) |
-	    (swizzle[1] << R300_INPUT_ROUTE_Y_SHIFT) |
-	    (swizzle[2] << R300_INPUT_ROUTE_Z_SHIFT) |
-	    (swizzle[3] << R300_INPUT_ROUTE_W_SHIFT);
+	return (swizzle[0] << R300_SWIZZLE_SELECT_X_SHIFT) |
+	    (swizzle[1] << R300_SWIZZLE_SELECT_Y_SHIFT) |
+	    (swizzle[2] << R300_SWIZZLE_SELECT_Z_SHIFT) |
+	    (swizzle[3] << R300_SWIZZLE_SELECT_W_SHIFT);
 }
 
 GLuint r300VAPInputRoute1(uint32_t * dst, int swizzle[][4], GLuint nr)
 {
-	GLuint i;
-
-	for (i = 0; i + 1 < nr; i += 2) {
-		dst[i >> 1] = r300VAPInputRoute1Swizzle(swizzle[i]) | R300_INPUT_ROUTE_ENABLE;
-		dst[i >> 1] |= (r300VAPInputRoute1Swizzle(swizzle[i + 1]) | R300_INPUT_ROUTE_ENABLE) << 16;
-	}
+	GLuint i, dw;
 
-	if (nr & 1) {
-		dst[nr >> 1] = r300VAPInputRoute1Swizzle(swizzle[nr - 1]) | R300_INPUT_ROUTE_ENABLE;
+	for (i = 0; i < nr; i += 2) {
+		dw = (r300VAPInputRoute1Swizzle(swizzle[i]) |
+		      ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y |
+			R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT)) << R300_SWIZZLE0_SHIFT;
+		if (i + 1 < nr) {
+			dw |= (r300VAPInputRoute1Swizzle(swizzle[i + 1]) |
+			       ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y |
+				 R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT)) << R300_SWIZZLE1_SHIFT;
+		}
+		dst[i >> 1] = dw;
 	}
 
 	return (nr + 1) >> 1;
@@ -294,7 +302,7 @@ GLuint r300VAPOutputCntl0(GLcontext * ctx, GLuint OutputsWritten)
 		ret |= R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT;
 
 	if (OutputsWritten & (1 << VERT_RESULT_COL0))
-		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT;
+		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT;
 
 	if (OutputsWritten & (1 << VERT_RESULT_COL1))
 		ret |= R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT;
@@ -536,16 +544,16 @@ void r300ReleaseArrays(GLcontext * ctx)
 
 void r300EmitCacheFlush(r300ContextPtr rmesa)
 {
-        int cmd_reserved = 0;
+	int cmd_reserved = 0;
 	int cmd_written = 0;
 
 	drm_radeon_cmd_header_t *cmd = NULL;
 
 	reg_start(R300_RB3D_DSTCACHE_CTLSTAT, 0);
-	e32(R300_RB3D_DSTCACHE_UNKNOWN_0A);
-
-	reg_start(R300_RB3D_ZCACHE_CTLSTAT, 0);
-	e32(R300_RB3D_ZCACHE_UNKNOWN_03);
-
+	e32(R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_FREE_3D_TAGS |
+	    R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_FLUSH_DIRTY_3D);
 
+	reg_start(R300_ZB_ZCACHE_CTLSTAT, 0);
+	e32(R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_FLUSH_AND_FREE |
+	    R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE);
 }
diff --git a/src/mesa/drivers/dri/r300/r300_emit.h b/src/mesa/drivers/dri/r300/r300_emit.h
index a6d69ec5ff8..89d738339f8 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.h
+++ b/src/mesa/drivers/dri/r300/r300_emit.h
@@ -39,7 +39,7 @@
 #ifndef __R300_EMIT_H__
 #define __R300_EMIT_H__
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "r300_context.h"
 #include "r300_cmdbuf.h"
 #include "radeon_reg.h"
@@ -50,7 +50,7 @@
 #define CP_PACKET3( pkt, n )						\
 	(RADEON_CP_PACKET3 | (pkt) | ((n) << 16))
 
-static inline uint32_t cmdpacket0(int reg, int count)
+static INLINE uint32_t cmdpacket0(int reg, int count)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -62,7 +62,7 @@ static inline uint32_t cmdpacket0(int reg, int count)
 	return cmd.u;
 }
 
-static inline uint32_t cmdvpu(int addr, int count)
+static INLINE uint32_t cmdvpu(int addr, int count)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -74,7 +74,21 @@ static inline uint32_t cmdvpu(int addr, int count)
 	return cmd.u;
 }
 
-static inline uint32_t cmdpacket3(int packet)
+static INLINE uint32_t cmdr500fp(int addr, int count, int type, int clamp)
+{
+	drm_r300_cmd_header_t cmd;
+
+	cmd.r500fp.cmd_type = R300_CMD_R500FP;
+	cmd.r500fp.count = count;
+	cmd.r500fp.adrhi_flags = ((unsigned int)addr & 0x100) >> 8;
+	cmd.r500fp.adrhi_flags |= type ? R500FP_CONSTANT_TYPE : 0;
+	cmd.r500fp.adrhi_flags |= clamp ? R500FP_CONSTANT_CLAMP : 0;
+	cmd.r500fp.adrlo = ((unsigned int)addr & 0x00FF);
+
+	return cmd.u;
+}
+
+static INLINE uint32_t cmdpacket3(int packet)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -84,7 +98,7 @@ static inline uint32_t cmdpacket3(int packet)
 	return cmd.u;
 }
 
-static inline uint32_t cmdcpdelay(unsigned short count)
+static INLINE uint32_t cmdcpdelay(unsigned short count)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -94,7 +108,7 @@ static inline uint32_t cmdcpdelay(unsigned short count)
 	return cmd.u;
 }
 
-static inline uint32_t cmdwait(unsigned char flags)
+static INLINE uint32_t cmdwait(unsigned char flags)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -104,7 +118,7 @@ static inline uint32_t cmdwait(unsigned char flags)
 	return cmd.u;
 }
 
-static inline uint32_t cmdpacify(void)
+static INLINE uint32_t cmdpacify(void)
 {
 	drm_r300_cmd_header_t cmd;
 
@@ -166,6 +180,19 @@ static inline uint32_t cmdpacify(void)
 		cmd[0].i = cmdvpu((dest), _n/4);			\
 	} while (0);
 
+#define r500fp_start_fragment(dest, length)				\
+	do {								\
+		int _n;							\
+		_n = (length);						\
+		cmd = (drm_radeon_cmd_header_t*)			\
+			r300AllocCmdBuf(rmesa,				\
+					(_n+1),				\
+					__FUNCTION__);			\
+		cmd_reserved = _n+1;					\
+		cmd_written =1;						\
+		cmd[0].i = cmdr500fp((dest), _n/6, 0, 0);		\
+	} while (0);
+
 #define start_packet3(packet, count)					\
 	{								\
 		int _n;							\
@@ -191,7 +218,7 @@ static inline uint32_t cmdpacify(void)
 /**
  * Must be sent to switch to 2d commands
  */
-void static inline end_3d(r300ContextPtr rmesa)
+void static INLINE end_3d(r300ContextPtr rmesa)
 {
 	drm_radeon_cmd_header_t *cmd = NULL;
 
@@ -200,7 +227,7 @@ void static inline end_3d(r300ContextPtr rmesa)
 	cmd[0].header.cmd_type = R300_CMD_END3D;
 }
 
-void static inline cp_delay(r300ContextPtr rmesa, unsigned short count)
+void static INLINE cp_delay(r300ContextPtr rmesa, unsigned short count)
 {
 	drm_radeon_cmd_header_t *cmd = NULL;
 
@@ -209,7 +236,7 @@ void static inline cp_delay(r300ContextPtr rmesa, unsigned short count)
 	cmd[0].i = cmdcpdelay(count);
 }
 
-void static inline cp_wait(r300ContextPtr rmesa, unsigned char flags)
+void static INLINE cp_wait(r300ContextPtr rmesa, unsigned char flags)
 {
 	drm_radeon_cmd_header_t *cmd = NULL;
 
@@ -230,6 +257,8 @@ extern int r300NumVerts(r300ContextPtr rmesa, int num_verts, int prim);
 
 extern void r300EmitCacheFlush(r300ContextPtr rmesa);
 
+extern GLuint r300VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
+				 int *inputs, GLint * tab, GLuint nr);
 extern GLuint r300VAPInputRoute1(uint32_t * dst, int swizzle[][4], GLuint nr);
 extern GLuint r300VAPInputCntl0(GLcontext * ctx, GLuint InputsRead);
 extern GLuint r300VAPInputCntl1(GLcontext * ctx, GLuint InputsRead);
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.c b/src/mesa/drivers/dri/r300/r300_fragprog.c
index cce8e685865..4ef7f2bd788 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.c
@@ -28,1974 +28,250 @@
 /**
  * \file
  *
- * \author Ben Skeggs <darktama@iinet.net.au>
+ * Fragment program compiler. Perform transformations on the intermediate
+ * representation until the program is in a form where we can translate
+ * it more or less directly into machine-readable form.
  *
+ * \author Ben Skeggs <darktama@iinet.net.au>
  * \author Jerome Glisse <j.glisse@gmail.com>
- *
- * \todo Depth write, WPOS/FOGC inputs
- *
- * \todo FogOption
- *
- * \todo Verify results of opcodes for accuracy, I've only checked them in
- * specific cases.
  */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_print.h"
 
 #include "r300_context.h"
 #include "r300_fragprog.h"
-#include "r300_reg.h"
+#include "r300_fragprog_swizzle.h"
 #include "r300_state.h"
 
-/*
- * Usefull macros and values
- */
-#define ERROR(fmt, args...) do {			\
-		fprintf(stderr, "%s::%s(): " fmt "\n",	\
-			__FILE__, __FUNCTION__, ##args);	\
-		fp->error = GL_TRUE;			\
-	} while(0)
-
-#define PFS_INVAL 0xFFFFFFFF
-#define COMPILE_STATE struct r300_pfs_compile_state *cs = fp->cs
-
-#define SWIZZLE_XYZ		0
-#define SWIZZLE_XXX		1
-#define SWIZZLE_YYY		2
-#define SWIZZLE_ZZZ		3
-#define SWIZZLE_WWW		4
-#define SWIZZLE_YZX		5
-#define SWIZZLE_ZXY		6
-#define SWIZZLE_WZY		7
-#define SWIZZLE_111		8
-#define SWIZZLE_000		9
-#define SWIZZLE_HHH		10
-
-#define swizzle(r, x, y, z, w) do_swizzle(fp, r,		\
-					  ((SWIZZLE_##x<<0)|	\
-					   (SWIZZLE_##y<<3)|	\
-					   (SWIZZLE_##z<<6)|	\
-					   (SWIZZLE_##w<<9)),	\
-					  0)
-
-#define REG_TYPE_INPUT		0
-#define REG_TYPE_OUTPUT		1
-#define REG_TYPE_TEMP		2
-#define REG_TYPE_CONST		3
-
-#define REG_TYPE_SHIFT		0
-#define REG_INDEX_SHIFT		2
-#define REG_VSWZ_SHIFT		8
-#define REG_SSWZ_SHIFT		13
-#define REG_NEGV_SHIFT		18
-#define REG_NEGS_SHIFT		19
-#define REG_ABS_SHIFT		20
-#define REG_NO_USE_SHIFT	21	// Hack for refcounting
-#define REG_VALID_SHIFT		22	// Does the register contain a defined value?
-#define REG_BUILTIN_SHIFT   23	// Is it a builtin (like all zero/all one)?
-
-#define REG_TYPE_MASK		(0x03 << REG_TYPE_SHIFT)
-#define REG_INDEX_MASK		(0x3F << REG_INDEX_SHIFT)
-#define REG_VSWZ_MASK		(0x1F << REG_VSWZ_SHIFT)
-#define REG_SSWZ_MASK		(0x1F << REG_SSWZ_SHIFT)
-#define REG_NEGV_MASK		(0x01 << REG_NEGV_SHIFT)
-#define REG_NEGS_MASK		(0x01 << REG_NEGS_SHIFT)
-#define REG_ABS_MASK		(0x01 << REG_ABS_SHIFT)
-#define REG_NO_USE_MASK		(0x01 << REG_NO_USE_SHIFT)
-#define REG_VALID_MASK		(0x01 << REG_VALID_SHIFT)
-#define REG_BUILTIN_MASK	(0x01 << REG_BUILTIN_SHIFT)
-
-#define REG(type, index, vswz, sswz, nouse, valid, builtin)	\
-	(((type << REG_TYPE_SHIFT) & REG_TYPE_MASK) |			\
-	 ((index << REG_INDEX_SHIFT) & REG_INDEX_MASK) |		\
-	 ((nouse << REG_NO_USE_SHIFT) & REG_NO_USE_MASK) |		\
-	 ((valid << REG_VALID_SHIFT) & REG_VALID_MASK) |		\
-	 ((builtin << REG_BUILTIN_SHIFT) & REG_BUILTIN_MASK) |	\
-	 ((vswz << REG_VSWZ_SHIFT) & REG_VSWZ_MASK) |			\
-	 ((sswz << REG_SSWZ_SHIFT) & REG_SSWZ_MASK))
-#define REG_GET_TYPE(reg)						\
-	((reg & REG_TYPE_MASK) >> REG_TYPE_SHIFT)
-#define REG_GET_INDEX(reg)						\
-	((reg & REG_INDEX_MASK) >> REG_INDEX_SHIFT)
-#define REG_GET_VSWZ(reg)						\
-	((reg & REG_VSWZ_MASK) >> REG_VSWZ_SHIFT)
-#define REG_GET_SSWZ(reg)						\
-	((reg & REG_SSWZ_MASK) >> REG_SSWZ_SHIFT)
-#define REG_GET_NO_USE(reg)						\
-	((reg & REG_NO_USE_MASK) >> REG_NO_USE_SHIFT)
-#define REG_GET_VALID(reg)						\
-	((reg & REG_VALID_MASK) >> REG_VALID_SHIFT)
-#define REG_GET_BUILTIN(reg)						\
-	((reg & REG_BUILTIN_MASK) >> REG_BUILTIN_SHIFT)
-#define REG_SET_TYPE(reg, type)						\
-	reg = ((reg & ~REG_TYPE_MASK) |					\
-	       ((type << REG_TYPE_SHIFT) & REG_TYPE_MASK))
-#define REG_SET_INDEX(reg, index)					\
-	reg = ((reg & ~REG_INDEX_MASK) |				\
-	       ((index << REG_INDEX_SHIFT) & REG_INDEX_MASK))
-#define REG_SET_VSWZ(reg, vswz)						\
-	reg = ((reg & ~REG_VSWZ_MASK) |					\
-	       ((vswz << REG_VSWZ_SHIFT) & REG_VSWZ_MASK))
-#define REG_SET_SSWZ(reg, sswz)						\
-	reg = ((reg & ~REG_SSWZ_MASK) |					\
-	       ((sswz << REG_SSWZ_SHIFT) & REG_SSWZ_MASK))
-#define REG_SET_NO_USE(reg, nouse)					\
-	reg = ((reg & ~REG_NO_USE_MASK) |				\
-	       ((nouse << REG_NO_USE_SHIFT) & REG_NO_USE_MASK))
-#define REG_SET_VALID(reg, valid)					\
-	reg = ((reg & ~REG_VALID_MASK) |				\
-	       ((valid << REG_VALID_SHIFT) & REG_VALID_MASK))
-#define REG_SET_BUILTIN(reg, builtin)					\
-	reg = ((reg & ~REG_BUILTIN_MASK) |				\
-	       ((builtin << REG_BUILTIN_SHIFT) & REG_BUILTIN_MASK))
-#define REG_ABS(reg)							\
-	reg = (reg | REG_ABS_MASK)
-#define REG_NEGV(reg)							\
-	reg = (reg | REG_NEGV_MASK)
-#define REG_NEGS(reg)							\
-	reg = (reg | REG_NEGS_MASK)
-
-/*
- * Datas structures for fragment program generation
- */
-
-/* description of r300 native hw instructions */
-static const struct {
-	const char *name;
-	int argc;
-	int v_op;
-	int s_op;
-} r300_fpop[] = {
-	/* *INDENT-OFF* */
-	{"MAD", 3, R300_FPI0_OUTC_MAD, R300_FPI2_OUTA_MAD},
-	{"DP3", 2, R300_FPI0_OUTC_DP3, R300_FPI2_OUTA_DP4},
-	{"DP4", 2, R300_FPI0_OUTC_DP4, R300_FPI2_OUTA_DP4},
-	{"MIN", 2, R300_FPI0_OUTC_MIN, R300_FPI2_OUTA_MIN},
-	{"MAX", 2, R300_FPI0_OUTC_MAX, R300_FPI2_OUTA_MAX},
-	{"CMP", 3, R300_FPI0_OUTC_CMP, R300_FPI2_OUTA_CMP},
-	{"FRC", 1, R300_FPI0_OUTC_FRC, R300_FPI2_OUTA_FRC},
-	{"EX2", 1, R300_FPI0_OUTC_REPL_ALPHA, R300_FPI2_OUTA_EX2},
-	{"LG2", 1, R300_FPI0_OUTC_REPL_ALPHA, R300_FPI2_OUTA_LG2},
-	{"RCP", 1, R300_FPI0_OUTC_REPL_ALPHA, R300_FPI2_OUTA_RCP},
-	{"RSQ", 1, R300_FPI0_OUTC_REPL_ALPHA, R300_FPI2_OUTA_RSQ},
-	{"REPL_ALPHA", 1, R300_FPI0_OUTC_REPL_ALPHA, PFS_INVAL},
-	{"CMPH", 3, R300_FPI0_OUTC_CMPH, PFS_INVAL},
-	/* *INDENT-ON* */
-};
-
-/* vector swizzles r300 can support natively, with a couple of
- * cases we handle specially
- *
- * REG_VSWZ/REG_SSWZ is an index into this table
- */
-
-/* mapping from SWIZZLE_* to r300 native values for scalar insns */
-#define SWIZZLE_HALF 6
-
-#define MAKE_SWZ3(x, y, z) (MAKE_SWIZZLE4(SWIZZLE_##x, \
-					  SWIZZLE_##y, \
-					  SWIZZLE_##z, \
-					  SWIZZLE_ZERO))
-/* native swizzles */
-static const struct r300_pfs_swizzle {
-	GLuint hash;		/* swizzle value this matches */
-	GLuint base;		/* base value for hw swizzle */
-	GLuint stride;		/* difference in base between arg0/1/2 */
-	GLuint flags;
-} v_swiz[] = {
-	/* *INDENT-OFF* */
-	{MAKE_SWZ3(X, Y, Z), R300_FPI0_ARGC_SRC0C_XYZ, 4, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(X, X, X), R300_FPI0_ARGC_SRC0C_XXX, 4, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(Y, Y, Y), R300_FPI0_ARGC_SRC0C_YYY, 4, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(Z, Z, Z), R300_FPI0_ARGC_SRC0C_ZZZ, 4, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(W, W, W), R300_FPI0_ARGC_SRC0A, 1, SLOT_SRC_SCALAR},
-	{MAKE_SWZ3(Y, Z, X), R300_FPI0_ARGC_SRC0C_YZX, 1, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(Z, X, Y), R300_FPI0_ARGC_SRC0C_ZXY, 1, SLOT_SRC_VECTOR},
-	{MAKE_SWZ3(W, Z, Y), R300_FPI0_ARGC_SRC0CA_WZY, 1, SLOT_SRC_BOTH},
-	{MAKE_SWZ3(ONE, ONE, ONE), R300_FPI0_ARGC_ONE, 0, 0},
-	{MAKE_SWZ3(ZERO, ZERO, ZERO), R300_FPI0_ARGC_ZERO, 0, 0},
-	{MAKE_SWZ3(HALF, HALF, HALF), R300_FPI0_ARGC_HALF, 0, 0},
-	{PFS_INVAL, 0, 0, 0},
-	/* *INDENT-ON* */
-};
-
-/* used during matching of non-native swizzles */
-#define SWZ_X_MASK (7 << 0)
-#define SWZ_Y_MASK (7 << 3)
-#define SWZ_Z_MASK (7 << 6)
-#define SWZ_W_MASK (7 << 9)
-static const struct {
-	GLuint hash;		/* used to mask matching swizzle components */
-	int mask;		/* actual outmask */
-	int count;		/* count of components matched */
-} s_mask[] = {
-	/* *INDENT-OFF* */
-	{SWZ_X_MASK | SWZ_Y_MASK | SWZ_Z_MASK, 1 | 2 | 4, 3},
-	{SWZ_X_MASK | SWZ_Y_MASK, 1 | 2, 2},
-	{SWZ_X_MASK | SWZ_Z_MASK, 1 | 4, 2},
-	{SWZ_Y_MASK | SWZ_Z_MASK, 2 | 4, 2},
-	{SWZ_X_MASK, 1, 1},
-	{SWZ_Y_MASK, 2, 1},
-	{SWZ_Z_MASK, 4, 1},
-	{PFS_INVAL, PFS_INVAL, PFS_INVAL}
-	/* *INDENT-ON* */
-};
-
-static const struct {
-	int base;		/* hw value of swizzle */
-	int stride;		/* difference between SRC0/1/2 */
-	GLuint flags;
-} s_swiz[] = {
-	/* *INDENT-OFF* */
-	{R300_FPI2_ARGA_SRC0C_X, 3, SLOT_SRC_VECTOR},
-	{R300_FPI2_ARGA_SRC0C_Y, 3, SLOT_SRC_VECTOR},
-	{R300_FPI2_ARGA_SRC0C_Z, 3, SLOT_SRC_VECTOR},
-	{R300_FPI2_ARGA_SRC0A, 1, SLOT_SRC_SCALAR},
-	{R300_FPI2_ARGA_ZERO, 0, 0},
-	{R300_FPI2_ARGA_ONE, 0, 0},
-	{R300_FPI2_ARGA_HALF, 0, 0}
-	/* *INDENT-ON* */
-};
-
-/* boiler-plate reg, for convenience */
-static const GLuint undef = REG(REG_TYPE_TEMP,
-				0,
-				SWIZZLE_XYZ,
-				SWIZZLE_W,
-				GL_FALSE,
-				GL_FALSE,
-				GL_FALSE);
-
-/* constant one source */
-static const GLuint pfs_one = REG(REG_TYPE_CONST,
-				  0,
-				  SWIZZLE_111,
-				  SWIZZLE_ONE,
-				  GL_FALSE,
-				  GL_TRUE,
-				  GL_TRUE);
-
-/* constant half source */
-static const GLuint pfs_half = REG(REG_TYPE_CONST,
-				   0,
-				   SWIZZLE_HHH,
-				   SWIZZLE_HALF,
-				   GL_FALSE,
-				   GL_TRUE,
-				   GL_TRUE);
-
-/* constant zero source */
-static const GLuint pfs_zero = REG(REG_TYPE_CONST,
-				   0,
-				   SWIZZLE_000,
-				   SWIZZLE_ZERO,
-				   GL_FALSE,
-				   GL_TRUE,
-				   GL_TRUE);
+#include "radeon_nqssadce.h"
+#include "radeon_program_alu.h"
 
-/*
- * Common functions prototypes
- */
-static void dump_program(struct r300_fragment_program *fp);
-static void emit_arith(struct r300_fragment_program *fp, int op,
-		       GLuint dest, int mask,
-		       GLuint src0, GLuint src1, GLuint src2, int flags);
 
-/**
- * Get an R300 temporary that can be written to in the given slot.
- */
-static int get_hw_temp(struct r300_fragment_program *fp, int slot)
+static void reset_srcreg(struct prog_src_register* reg)
 {
-	COMPILE_STATE;
-	int r;
-
-	for (r = 0; r < PFS_NUM_TEMP_REGS; ++r) {
-		if (cs->hwtemps[r].free >= 0 && cs->hwtemps[r].free <= slot)
-			break;
-	}
-
-	if (r >= PFS_NUM_TEMP_REGS) {
-		ERROR("Out of hardware temps\n");
-		return 0;
-	}
-	// Reserved is used to avoid the following scenario:
-	//  R300 temporary X is first assigned to Mesa temporary Y during vector ops
-	//  R300 temporary X is then assigned to Mesa temporary Z for further vector ops
-	//  Then scalar ops on Mesa temporary Z are emitted and move back in time
-	//  to overwrite the value of temporary Y.
-	// End scenario.
-	cs->hwtemps[r].reserved = cs->hwtemps[r].free;
-	cs->hwtemps[r].free = -1;
-
-	// Reset to some value that won't mess things up when the user
-	// tries to read from a temporary that hasn't been assigned a value yet.
-	// In the normal case, vector_valid and scalar_valid should be set to
-	// a sane value by the first emit that writes to this temporary.
-	cs->hwtemps[r].vector_valid = 0;
-	cs->hwtemps[r].scalar_valid = 0;
-
-	if (r > fp->max_temp_idx)
-		fp->max_temp_idx = r;
-
-	return r;
+	_mesa_bzero(reg, sizeof(*reg));
+	reg->Swizzle = SWIZZLE_NOOP;
 }
 
-/**
- * Get an R300 temporary that will act as a TEX destination register.
- */
-static int get_hw_temp_tex(struct r300_fragment_program *fp)
+static struct prog_src_register shadow_ambient(struct gl_program *program, int tmu)
 {
-	COMPILE_STATE;
-	int r;
-
-	for (r = 0; r < PFS_NUM_TEMP_REGS; ++r) {
-		if (cs->used_in_node & (1 << r))
-			continue;
-
-		// Note: Be very careful here
-		if (cs->hwtemps[r].free >= 0 && cs->hwtemps[r].free <= 0)
-			break;
-	}
-
-	if (r >= PFS_NUM_TEMP_REGS)
-		return get_hw_temp(fp, 0);	/* Will cause an indirection */
-
-	cs->hwtemps[r].reserved = cs->hwtemps[r].free;
-	cs->hwtemps[r].free = -1;
-
-	// Reset to some value that won't mess things up when the user
-	// tries to read from a temporary that hasn't been assigned a value yet.
-	// In the normal case, vector_valid and scalar_valid should be set to
-	// a sane value by the first emit that writes to this temporary.
-	cs->hwtemps[r].vector_valid = cs->nrslots;
-	cs->hwtemps[r].scalar_valid = cs->nrslots;
-
-	if (r > fp->max_temp_idx)
-		fp->max_temp_idx = r;
-
-	return r;
-}
-
-/**
- * Mark the given hardware register as free.
- */
-static void free_hw_temp(struct r300_fragment_program *fp, int idx)
-{
-	COMPILE_STATE;
-
-	// Be very careful here. Consider sequences like
-	//  MAD r0, r1,r2,r3
-	//  TEX r4, ...
-	// The TEX instruction may be moved in front of the MAD instruction
-	// due to the way nodes work. We don't want to alias r1 and r4 in
-	// this case.
-	// I'm certain the register allocation could be further sanitized,
-	// but it's tricky because of stuff that can happen inside emit_tex
-	// and emit_arith.
-	cs->hwtemps[idx].free = cs->nrslots + 1;
-}
-
-/**
- * Create a new Mesa temporary register.
- */
-static GLuint get_temp_reg(struct r300_fragment_program *fp)
-{
-	COMPILE_STATE;
-	GLuint r = undef;
-	GLuint index;
-
-	index = ffs(~cs->temp_in_use);
-	if (!index) {
-		ERROR("Out of program temps\n");
-		return r;
-	}
-
-	cs->temp_in_use |= (1 << --index);
-	cs->temps[index].refcount = 0xFFFFFFFF;
-	cs->temps[index].reg = -1;
-
-	REG_SET_TYPE(r, REG_TYPE_TEMP);
-	REG_SET_INDEX(r, index);
-	REG_SET_VALID(r, GL_TRUE);
-	return r;
-}
-
-/**
- * Create a new Mesa temporary register that will act as the destination
- * register for a texture read.
- */
-static GLuint get_temp_reg_tex(struct r300_fragment_program *fp)
-{
-	COMPILE_STATE;
-	GLuint r = undef;
-	GLuint index;
-
-	index = ffs(~cs->temp_in_use);
-	if (!index) {
-		ERROR("Out of program temps\n");
-		return r;
-	}
-
-	cs->temp_in_use |= (1 << --index);
-	cs->temps[index].refcount = 0xFFFFFFFF;
-	cs->temps[index].reg = get_hw_temp_tex(fp);
-
-	REG_SET_TYPE(r, REG_TYPE_TEMP);
-	REG_SET_INDEX(r, index);
-	REG_SET_VALID(r, GL_TRUE);
-	return r;
-}
-
-/**
- * Free a Mesa temporary and the associated R300 temporary.
- */
-static void free_temp(struct r300_fragment_program *fp, GLuint r)
-{
-	COMPILE_STATE;
-	GLuint index = REG_GET_INDEX(r);
-
-	if (!(cs->temp_in_use & (1 << index)))
-		return;
+	gl_state_index fail_value_tokens[STATE_LENGTH] = {
+		STATE_INTERNAL, STATE_SHADOW_AMBIENT, 0, 0, 0
+	};
+	struct prog_src_register reg = { 0, };
 
-	if (REG_GET_TYPE(r) == REG_TYPE_TEMP) {
-		free_hw_temp(fp, cs->temps[index].reg);
-		cs->temps[index].reg = -1;
-		cs->temp_in_use &= ~(1 << index);
-	} else if (REG_GET_TYPE(r) == REG_TYPE_INPUT) {
-		free_hw_temp(fp, cs->inputs[index].reg);
-		cs->inputs[index].reg = -1;
-	}
+	fail_value_tokens[2] = tmu;
+	reg.File = PROGRAM_STATE_VAR;
+	reg.Index = _mesa_add_state_reference(program->Parameters, fail_value_tokens);
+	reg.Swizzle = SWIZZLE_WWWW;
+	return reg;
 }
 
 /**
- * Emit a hardware constant/parameter.
+ * Transform TEX, TXP, TXB, and KIL instructions in the following way:
+ *  - premultiply texture coordinates for RECT
+ *  - extract operand swizzles
+ *  - introduce a temporary register when write masks are needed
  *
- * \p cp Stable pointer to an array of 4 floats.
- *  The pointer must be stable in the sense that it remains to be valid
- *  and hold the contents of the constant/parameter throughout the lifetime
- *  of the fragment program (actually, up until the next time the fragment
- *  program is translated).
- */
-static GLuint emit_const4fv(struct r300_fragment_program *fp,
-			    const GLfloat * cp)
-{
-	GLuint reg = undef;
-	int index;
-
-	for (index = 0; index < fp->const_nr; ++index) {
-		if (fp->constant[index] == cp)
-			break;
-	}
-
-	if (index >= fp->const_nr) {
-		if (index >= PFS_NUM_CONST_REGS) {
-			ERROR("Out of hw constants!\n");
-			return reg;
-		}
-
-		fp->const_nr++;
-		fp->constant[index] = cp;
-	}
-
-	REG_SET_TYPE(reg, REG_TYPE_CONST);
-	REG_SET_INDEX(reg, index);
-	REG_SET_VALID(reg, GL_TRUE);
-	return reg;
-}
-
-static inline GLuint negate(GLuint r)
-{
-	REG_NEGS(r);
-	REG_NEGV(r);
-	return r;
-}
-
-/* Hack, to prevent clobbering sources used multiple times when
- * emulating non-native instructions
+ * \todo If/when r5xx uses the radeon_program architecture, this can probably
+ * be reused.
  */
-static inline GLuint keep(GLuint r)
-{
-	REG_SET_NO_USE(r, GL_TRUE);
-	return r;
-}
-
-static inline GLuint absolute(GLuint r)
+static GLboolean transform_TEX(
+	struct radeon_transform_context *t,
+	struct prog_instruction* orig_inst, void* data)
 {
-	REG_ABS(r);
-	return r;
-}
-
-static int swz_native(struct r300_fragment_program *fp,
-		      GLuint src, GLuint * r, GLuint arbneg)
-{
-	/* Native swizzle, handle negation */
-	src = (src & ~REG_NEGS_MASK) | (((arbneg >> 3) & 1) << REG_NEGS_SHIFT);
-
-	if ((arbneg & 0x7) == 0x0) {
-		src = src & ~REG_NEGV_MASK;
-		*r = src;
-	} else if ((arbneg & 0x7) == 0x7) {
-		src |= REG_NEGV_MASK;
-		*r = src;
-	} else {
-		if (!REG_GET_VALID(*r))
-			*r = get_temp_reg(fp);
-		src |= REG_NEGV_MASK;
-		emit_arith(fp,
-			   PFS_OP_MAD,
-			   *r, arbneg & 0x7, keep(src), pfs_one, pfs_zero, 0);
-		src = src & ~REG_NEGV_MASK;
-		emit_arith(fp,
-			   PFS_OP_MAD,
-			   *r,
-			   (arbneg ^ 0x7) | WRITEMASK_W,
-			   src, pfs_one, pfs_zero, 0);
-	}
-
-	return 3;
-}
-
-static int swz_emit_partial(struct r300_fragment_program *fp,
-			    GLuint src,
-			    GLuint * r, int mask, int mc, GLuint arbneg)
-{
-	GLuint tmp;
-	GLuint wmask = 0;
-
-	if (!REG_GET_VALID(*r))
-		*r = get_temp_reg(fp);
-
-	/* A partial match, VSWZ/mask define what parts of the
-	 * desired swizzle we match
-	 */
-	if (mc + s_mask[mask].count == 3) {
-		wmask = WRITEMASK_W;
-		src |= ((arbneg >> 3) & 1) << REG_NEGS_SHIFT;
-	}
-
-	tmp = arbneg & s_mask[mask].mask;
-	if (tmp) {
-		tmp = tmp ^ s_mask[mask].mask;
-		if (tmp) {
-			emit_arith(fp,
-				   PFS_OP_MAD,
-				   *r,
-				   arbneg & s_mask[mask].mask,
-				   keep(src) | REG_NEGV_MASK,
-				   pfs_one, pfs_zero, 0);
-			if (!wmask) {
-				REG_SET_NO_USE(src, GL_TRUE);
-			} else {
-				REG_SET_NO_USE(src, GL_FALSE);
-			}
-			emit_arith(fp,
-				   PFS_OP_MAD,
-				   *r, tmp | wmask, src, pfs_one, pfs_zero, 0);
-		} else {
-			if (!wmask) {
-				REG_SET_NO_USE(src, GL_TRUE);
-			} else {
-				REG_SET_NO_USE(src, GL_FALSE);
-			}
-			emit_arith(fp,
-				   PFS_OP_MAD,
-				   *r,
-				   (arbneg & s_mask[mask].mask) | wmask,
-				   src | REG_NEGV_MASK, pfs_one, pfs_zero, 0);
-		}
-	} else {
-		if (!wmask) {
-			REG_SET_NO_USE(src, GL_TRUE);
-		} else {
-			REG_SET_NO_USE(src, GL_FALSE);
-		}
-		emit_arith(fp, PFS_OP_MAD,
-			   *r,
-			   s_mask[mask].mask | wmask,
-			   src, pfs_one, pfs_zero, 0);
-	}
-
-	return s_mask[mask].count;
-}
-
-static GLuint do_swizzle(struct r300_fragment_program *fp,
-			 GLuint src, GLuint arbswz, GLuint arbneg)
-{
-	GLuint r = undef;
-	GLuint vswz;
-	int c_mask = 0;
-	int v_match = 0;
-
-	/* If swizzling from something without an XYZW native swizzle,
-	 * emit result to a temp, and do new swizzle from the temp.
-	 */
-#if 0
-	if (REG_GET_VSWZ(src) != SWIZZLE_XYZ || REG_GET_SSWZ(src) != SWIZZLE_W) {
-		GLuint temp = get_temp_reg(fp);
-		emit_arith(fp,
-			   PFS_OP_MAD,
-			   temp, WRITEMASK_XYZW, src, pfs_one, pfs_zero, 0);
-		src = temp;
-	}
-#endif
-
-	if (REG_GET_VSWZ(src) != SWIZZLE_XYZ || REG_GET_SSWZ(src) != SWIZZLE_W) {
-		GLuint vsrcswz =
-		    (v_swiz[REG_GET_VSWZ(src)].
-		     hash & (SWZ_X_MASK | SWZ_Y_MASK | SWZ_Z_MASK)) |
-		    REG_GET_SSWZ(src) << 9;
-		GLint i;
-
-		GLuint newswz = 0;
-		GLuint offset;
-		for (i = 0; i < 4; ++i) {
-			offset = GET_SWZ(arbswz, i);
-
-			newswz |=
-			    (offset <= 3) ? GET_SWZ(vsrcswz,
-						    offset) << i *
-			    3 : offset << i * 3;
-		}
-
-		arbswz = newswz & (SWZ_X_MASK | SWZ_Y_MASK | SWZ_Z_MASK);
-		REG_SET_SSWZ(src, GET_SWZ(newswz, 3));
-	} else {
-		/* set scalar swizzling */
-		REG_SET_SSWZ(src, GET_SWZ(arbswz, 3));
-
-	}
-	do {
-		vswz = REG_GET_VSWZ(src);
-		do {
-			int chash;
-
-			REG_SET_VSWZ(src, vswz);
-			chash = v_swiz[REG_GET_VSWZ(src)].hash &
-			    s_mask[c_mask].hash;
-
-			if (chash == (arbswz & s_mask[c_mask].hash)) {
-				if (s_mask[c_mask].count == 3) {
-					v_match += swz_native(fp,
-							      src, &r, arbneg);
-				} else {
-					v_match += swz_emit_partial(fp,
-								    src,
-								    &r,
-								    c_mask,
-								    v_match,
-								    arbneg);
-				}
-
-				if (v_match == 3)
-					return r;
-
-				/* Fill with something invalid.. all 0's was
-				 * wrong before, matched SWIZZLE_X.  So all
-				 * 1's will be okay for now
-				 */
-				arbswz |= (PFS_INVAL & s_mask[c_mask].hash);
-			}
-		} while (v_swiz[++vswz].hash != PFS_INVAL);
-		REG_SET_VSWZ(src, SWIZZLE_XYZ);
-	} while (s_mask[++c_mask].hash != PFS_INVAL);
-
-	ERROR("should NEVER get here\n");
-	return r;
-}
-
-static GLuint t_src(struct r300_fragment_program *fp,
-		    struct prog_src_register fpsrc)
-{
-	GLuint r = undef;
-
-	switch (fpsrc.File) {
-	case PROGRAM_TEMPORARY:
-		REG_SET_INDEX(r, fpsrc.Index);
-		REG_SET_VALID(r, GL_TRUE);
-		REG_SET_TYPE(r, REG_TYPE_TEMP);
-		break;
-	case PROGRAM_INPUT:
-		REG_SET_INDEX(r, fpsrc.Index);
-		REG_SET_VALID(r, GL_TRUE);
-		REG_SET_TYPE(r, REG_TYPE_INPUT);
-		break;
-	case PROGRAM_LOCAL_PARAM:
-		r = emit_const4fv(fp,
-				  fp->mesa_program.Base.LocalParams[fpsrc.
-								    Index]);
-		break;
-	case PROGRAM_ENV_PARAM:
-		r = emit_const4fv(fp,
-				  fp->ctx->FragmentProgram.Parameters[fpsrc.
-								      Index]);
-		break;
-	case PROGRAM_STATE_VAR:
-	case PROGRAM_NAMED_PARAM:
-		r = emit_const4fv(fp,
-				  fp->mesa_program.Base.Parameters->
-				  ParameterValues[fpsrc.Index]);
-		break;
-	default:
-		ERROR("unknown SrcReg->File %x\n", fpsrc.File);
-		return r;
-	}
-
-	/* no point swizzling ONE/ZERO/HALF constants... */
-	if (REG_GET_VSWZ(r) < SWIZZLE_111 || REG_GET_SSWZ(r) < SWIZZLE_ZERO)
-		r = do_swizzle(fp, r, fpsrc.Swizzle, fpsrc.NegateBase);
-	return r;
-}
-
-static GLuint t_scalar_src(struct r300_fragment_program *fp,
-			   struct prog_src_register fpsrc)
-{
-	struct prog_src_register src = fpsrc;
-	int sc = GET_SWZ(fpsrc.Swizzle, 0);	/* X */
-
-	src.Swizzle = ((sc << 0) | (sc << 3) | (sc << 6) | (sc << 9));
-
-	return t_src(fp, src);
-}
-
-static GLuint t_dst(struct r300_fragment_program *fp,
-		    struct prog_dst_register dest)
-{
-	GLuint r = undef;
-
-	switch (dest.File) {
-	case PROGRAM_TEMPORARY:
-		REG_SET_INDEX(r, dest.Index);
-		REG_SET_VALID(r, GL_TRUE);
-		REG_SET_TYPE(r, REG_TYPE_TEMP);
-		return r;
-	case PROGRAM_OUTPUT:
-		REG_SET_TYPE(r, REG_TYPE_OUTPUT);
-		switch (dest.Index) {
-		case FRAG_RESULT_COLR:
-		case FRAG_RESULT_DEPR:
-			REG_SET_INDEX(r, dest.Index);
-			REG_SET_VALID(r, GL_TRUE);
-			return r;
-		default:
-			ERROR("Bad DstReg->Index 0x%x\n", dest.Index);
-			return r;
-		}
-	default:
-		ERROR("Bad DstReg->File 0x%x\n", dest.File);
-		return r;
-	}
-}
-
-static int t_hw_src(struct r300_fragment_program *fp, GLuint src, GLboolean tex)
-{
-	COMPILE_STATE;
-	int idx;
-	int index = REG_GET_INDEX(src);
-
-	switch (REG_GET_TYPE(src)) {
-	case REG_TYPE_TEMP:
-		/* NOTE: if reg==-1 here, a source is being read that
-		 *       hasn't been written to. Undefined results.
-		 */
-		if (cs->temps[index].reg == -1)
-			cs->temps[index].reg = get_hw_temp(fp, cs->nrslots);
-
-		idx = cs->temps[index].reg;
-
-		if (!REG_GET_NO_USE(src) && (--cs->temps[index].refcount == 0))
-			free_temp(fp, src);
-		break;
-	case REG_TYPE_INPUT:
-		idx = cs->inputs[index].reg;
-
-		if (!REG_GET_NO_USE(src) && (--cs->inputs[index].refcount == 0))
-			free_hw_temp(fp, cs->inputs[index].reg);
-		break;
-	case REG_TYPE_CONST:
-		return (index | SRC_CONST);
-	default:
-		ERROR("Invalid type for source reg\n");
-		return (0 | SRC_CONST);
-	}
+	struct r300_fragment_program_compiler *compiler =
+		(struct r300_fragment_program_compiler*)data;
+	struct prog_instruction inst = *orig_inst;
+	struct prog_instruction* tgt;
+	GLboolean destredirect = GL_FALSE;
+
+	if (inst.Opcode != OPCODE_TEX &&
+	    inst.Opcode != OPCODE_TXB &&
+	    inst.Opcode != OPCODE_TXP &&
+	    inst.Opcode != OPCODE_KIL)
+		return GL_FALSE;
 
-	if (!tex)
-		cs->used_in_node |= (1 << idx);
+	if (inst.Opcode != OPCODE_KIL &&
+	    t->Program->ShadowSamplers & (1 << inst.TexSrcUnit)) {
+		GLuint comparefunc = GL_NEVER + compiler->fp->state.unit[inst.TexSrcUnit].texture_compare_func;
 
-	return idx;
-}
+		if (comparefunc == GL_NEVER || comparefunc == GL_ALWAYS) {
+			tgt = radeonAppendInstructions(t->Program, 1);
 
-static int t_hw_dst(struct r300_fragment_program *fp,
-		    GLuint dest, GLboolean tex, int slot)
-{
-	COMPILE_STATE;
-	int idx;
-	GLuint index = REG_GET_INDEX(dest);
-	assert(REG_GET_VALID(dest));
-
-	switch (REG_GET_TYPE(dest)) {
-	case REG_TYPE_TEMP:
-		if (cs->temps[REG_GET_INDEX(dest)].reg == -1) {
-			if (!tex) {
-				cs->temps[index].reg = get_hw_temp(fp, slot);
+			tgt->Opcode = OPCODE_MOV;
+			tgt->DstReg = inst.DstReg;
+			if (comparefunc == GL_ALWAYS) {
+				tgt->SrcReg[0].File = PROGRAM_BUILTIN;
+				tgt->SrcReg[0].Swizzle = SWIZZLE_1111;
 			} else {
-				cs->temps[index].reg = get_hw_temp_tex(fp);
+				tgt->SrcReg[0] = shadow_ambient(t->Program, inst.TexSrcUnit);
 			}
+			return GL_TRUE;
 		}
-		idx = cs->temps[index].reg;
-
-		if (!REG_GET_NO_USE(dest) && (--cs->temps[index].refcount == 0))
-			free_temp(fp, dest);
-
-		cs->dest_in_node |= (1 << idx);
-		cs->used_in_node |= (1 << idx);
-		break;
-	case REG_TYPE_OUTPUT:
-		switch (index) {
-		case FRAG_RESULT_COLR:
-			fp->node[fp->cur_node].flags |=
-			    R300_PFS_NODE_OUTPUT_COLOR;
-			break;
-		case FRAG_RESULT_DEPR:
-			fp->node[fp->cur_node].flags |=
-			    R300_PFS_NODE_OUTPUT_DEPTH;
-			break;
-		}
-		return index;
-		break;
-	default:
-		ERROR("invalid dest reg type %d\n", REG_GET_TYPE(dest));
-		return 0;
-	}
 
-	return idx;
-}
-
-static void emit_nop(struct r300_fragment_program *fp)
-{
-	COMPILE_STATE;
-
-	if (cs->nrslots >= PFS_MAX_ALU_INST) {
-		ERROR("Out of ALU instruction slots\n");
-		return;
+		inst.DstReg.File = PROGRAM_TEMPORARY;
+		inst.DstReg.Index = radeonFindFreeTemporary(t);
+		inst.DstReg.WriteMask = WRITEMASK_XYZW;
 	}
 
-	fp->alu.inst[cs->nrslots].inst0 = NOP_INST0;
-	fp->alu.inst[cs->nrslots].inst1 = NOP_INST1;
-	fp->alu.inst[cs->nrslots].inst2 = NOP_INST2;
-	fp->alu.inst[cs->nrslots].inst3 = NOP_INST3;
-	cs->nrslots++;
-}
-
-static void emit_tex(struct r300_fragment_program *fp,
-		     struct prog_instruction *fpi, int opcode)
-{
-	COMPILE_STATE;
-	GLuint coord = t_src(fp, fpi->SrcReg[0]);
-	GLuint dest = undef, rdest = undef;
-	GLuint din, uin;
-	int unit = fpi->TexSrcUnit;
-	int hwsrc, hwdest;
-	GLuint tempreg = 0;
-
-	uin = cs->used_in_node;
-	din = cs->dest_in_node;
-
-	/* Resolve source/dest to hardware registers */
-	if (opcode != R300_FPITX_OP_KIL) {
-		if (fpi->TexSrcTarget == TEXTURE_RECT_INDEX) {
-			/**
-			 * Hardware uses [0..1]x[0..1] range for rectangle textures
-			 * instead of [0..Width]x[0..Height].
-			 * Add a scaling instruction.
-			 *
-			 * \todo Refactor this once we have proper rewriting/optimization
-			 * support for programs.
-			 */
-			gl_state_index tokens[STATE_LENGTH] = {
-				STATE_INTERNAL, STATE_R300_TEXRECT_FACTOR, 0, 0,
-				0
-			};
-			int factor_index;
-			GLuint factorreg;
-
-			tokens[2] = unit;
-			factor_index =
-			    _mesa_add_state_reference(fp->mesa_program.Base.
-						      Parameters, tokens);
-			factorreg =
-			    emit_const4fv(fp,
-					  fp->mesa_program.Base.Parameters->
-					  ParameterValues[factor_index]);
-			tempreg = keep(get_temp_reg(fp));
-
-			emit_arith(fp, PFS_OP_MAD, tempreg, WRITEMASK_XYZW,
-				   coord, factorreg, pfs_zero, 0);
-
-			/* Ensure correct node indirection */
-			uin = cs->used_in_node;
-			din = cs->dest_in_node;
-
-			hwsrc = t_hw_src(fp, tempreg, GL_TRUE);
-		} else {
-			hwsrc = t_hw_src(fp, coord, GL_TRUE);
-		}
-
-		dest = t_dst(fp, fpi->DstReg);
 
-		/* r300 doesn't seem to be able to do TEX->output reg */
-		if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
-			rdest = dest;
-			dest = get_temp_reg_tex(fp);
-		}
-		hwdest =
-		    t_hw_dst(fp, dest, GL_TRUE,
-			     fp->node[fp->cur_node].alu_offset);
-
-		/* Use a temp that hasn't been used in this node, rather
-		 * than causing an indirection
-		 */
-		if (uin & (1 << hwdest)) {
-			free_hw_temp(fp, hwdest);
-			hwdest = get_hw_temp_tex(fp);
-			cs->temps[REG_GET_INDEX(dest)].reg = hwdest;
-		}
-	} else {
-		hwdest = 0;
-		unit = 0;
-		hwsrc = t_hw_src(fp, coord, GL_TRUE);
-	}
-
-	/* Indirection if source has been written in this node, or if the
-	 * dest has been read/written in this node
+	/* Hardware uses [0..1]x[0..1] range for rectangle textures
+	 * instead of [0..Width]x[0..Height].
+	 * Add a scaling instruction.
 	 */
-	if ((REG_GET_TYPE(coord) != REG_TYPE_CONST &&
-	     (din & (1 << hwsrc))) || (uin & (1 << hwdest))) {
-
-		/* Finish off current node */
-		if (fp->node[fp->cur_node].alu_offset == cs->nrslots)
-			emit_nop(fp);
-
-		fp->node[fp->cur_node].alu_end =
-		    cs->nrslots - fp->node[fp->cur_node].alu_offset - 1;
-		assert(fp->node[fp->cur_node].alu_end >= 0);
-
-		if (++fp->cur_node >= PFS_MAX_TEX_INDIRECT) {
-			ERROR("too many levels of texture indirection\n");
-			return;
-		}
-
-		/* Start new node */
-		fp->node[fp->cur_node].tex_offset = fp->tex.length;
-		fp->node[fp->cur_node].alu_offset = cs->nrslots;
-		fp->node[fp->cur_node].tex_end = -1;
-		fp->node[fp->cur_node].alu_end = -1;
-		fp->node[fp->cur_node].flags = 0;
-		cs->used_in_node = 0;
-		cs->dest_in_node = 0;
-	}
+	if (inst.Opcode != OPCODE_KIL && inst.TexSrcTarget == TEXTURE_RECT_INDEX) {
+		gl_state_index tokens[STATE_LENGTH] = {
+			STATE_INTERNAL, STATE_R300_TEXRECT_FACTOR, 0, 0,
+			0
+		};
 
-	if (fp->cur_node == 0)
-		fp->first_node_has_tex = 1;
+		int tempreg = radeonFindFreeTemporary(t);
+		int factor_index;
 
-	fp->tex.inst[fp->tex.length++] = 0 | (hwsrc << R300_FPITX_SRC_SHIFT)
-	    | (hwdest << R300_FPITX_DST_SHIFT)
-	    | (unit << R300_FPITX_IMAGE_SHIFT)
-	    /* not entirely sure about this */
-	    | (opcode << R300_FPITX_OPCODE_SHIFT);
+		tokens[2] = inst.TexSrcUnit;
+		factor_index = _mesa_add_state_reference(t->Program->Parameters, tokens);
 
-	cs->dest_in_node |= (1 << hwdest);
-	if (REG_GET_TYPE(coord) != REG_TYPE_CONST)
-		cs->used_in_node |= (1 << hwsrc);
+		tgt = radeonAppendInstructions(t->Program, 1);
 
-	fp->node[fp->cur_node].tex_end++;
+		tgt->Opcode = OPCODE_MUL;
+		tgt->DstReg.File = PROGRAM_TEMPORARY;
+		tgt->DstReg.Index = tempreg;
+		tgt->SrcReg[0] = inst.SrcReg[0];
+		tgt->SrcReg[1].File = PROGRAM_STATE_VAR;
+		tgt->SrcReg[1].Index = factor_index;
 
-	/* Copy from temp to output if needed */
-	if (REG_GET_VALID(rdest)) {
-		emit_arith(fp, PFS_OP_MAD, rdest, WRITEMASK_XYZW, dest,
-			   pfs_one, pfs_zero, 0);
-		free_temp(fp, dest);
+		reset_srcreg(&inst.SrcReg[0]);
+		inst.SrcReg[0].File = PROGRAM_TEMPORARY;
+		inst.SrcReg[0].Index = tempreg;
 	}
 
-	/* Free temp register */
-	if (tempreg != 0)
-		free_temp(fp, tempreg);
-}
-
-/**
- * Returns the first slot where we could possibly allow writing to dest,
- * according to register allocation.
- */
-static int get_earliest_allowed_write(struct r300_fragment_program *fp,
-				      GLuint dest, int mask)
-{
-	COMPILE_STATE;
-	int idx;
-	int pos;
-	GLuint index = REG_GET_INDEX(dest);
-	assert(REG_GET_VALID(dest));
-
-	switch (REG_GET_TYPE(dest)) {
-	case REG_TYPE_TEMP:
-		if (cs->temps[index].reg == -1)
-			return 0;
-
-		idx = cs->temps[index].reg;
-		break;
-	case REG_TYPE_OUTPUT:
-		return 0;
-	default:
-		ERROR("invalid dest reg type %d\n", REG_GET_TYPE(dest));
-		return 0;
-	}
-
-	pos = cs->hwtemps[idx].reserved;
-	if (mask & WRITEMASK_XYZ) {
-		if (pos < cs->hwtemps[idx].vector_lastread)
-			pos = cs->hwtemps[idx].vector_lastread;
-	}
-	if (mask & WRITEMASK_W) {
-		if (pos < cs->hwtemps[idx].scalar_lastread)
-			pos = cs->hwtemps[idx].scalar_lastread;
-	}
-
-	return pos;
-}
+	if (inst.Opcode != OPCODE_KIL) {
+		if (inst.DstReg.File != PROGRAM_TEMPORARY ||
+		    inst.DstReg.WriteMask != WRITEMASK_XYZW) {
+			int tempreg = radeonFindFreeTemporary(t);
 
-/**
- * Allocates a slot for an ALU instruction that can consist of
- * a vertex part or a scalar part or both.
- *
- * Sources from src (src[0] to src[argc-1]) are added to the slot in the
- * appropriate position (vector and/or scalar), and their positions are
- * recorded in the srcpos array.
- *
- * This function emits instruction code for the source fetch and the
- * argument selection. It does not emit instruction code for the
- * opcode or the destination selection.
- *
- * @return the index of the slot
- */
-static int find_and_prepare_slot(struct r300_fragment_program *fp,
-				 GLboolean emit_vop,
-				 GLboolean emit_sop,
-				 int argc, GLuint * src, GLuint dest, int mask)
-{
-	COMPILE_STATE;
-	int hwsrc[3];
-	int srcpos[3];
-	unsigned int used;
-	int tempused;
-	int tempvsrc[3];
-	int tempssrc[3];
-	int pos;
-	int regnr;
-	int i, j;
-
-	// Determine instruction slots, whether sources are required on
-	// vector or scalar side, and the smallest slot number where
-	// all source registers are available
-	used = 0;
-	if (emit_vop)
-		used |= SLOT_OP_VECTOR;
-	if (emit_sop)
-		used |= SLOT_OP_SCALAR;
-
-	pos = get_earliest_allowed_write(fp, dest, mask);
-
-	if (fp->node[fp->cur_node].alu_offset > pos)
-		pos = fp->node[fp->cur_node].alu_offset;
-	for (i = 0; i < argc; ++i) {
-		if (!REG_GET_BUILTIN(src[i])) {
-			if (emit_vop)
-				used |= v_swiz[REG_GET_VSWZ(src[i])].flags << i;
-			if (emit_sop)
-				used |= s_swiz[REG_GET_SSWZ(src[i])].flags << i;
-		}
-
-		hwsrc[i] = t_hw_src(fp, src[i], GL_FALSE);	/* Note: sideeffects wrt refcounting! */
-		regnr = hwsrc[i] & 31;
-
-		if (REG_GET_TYPE(src[i]) == REG_TYPE_TEMP) {
-			if (used & (SLOT_SRC_VECTOR << i)) {
-				if (cs->hwtemps[regnr].vector_valid > pos)
-					pos = cs->hwtemps[regnr].vector_valid;
-			}
-			if (used & (SLOT_SRC_SCALAR << i)) {
-				if (cs->hwtemps[regnr].scalar_valid > pos)
-					pos = cs->hwtemps[regnr].scalar_valid;
-			}
+			inst.DstReg.File = PROGRAM_TEMPORARY;
+			inst.DstReg.Index = tempreg;
+			inst.DstReg.WriteMask = WRITEMASK_XYZW;
+			destredirect = GL_TRUE;
 		}
 	}
 
-	// Find a slot that fits
-	for (;; ++pos) {
-		if (cs->slot[pos].used & used & SLOT_OP_BOTH)
-			continue;
-
-		if (pos >= cs->nrslots) {
-			if (cs->nrslots >= PFS_MAX_ALU_INST) {
-				ERROR("Out of ALU instruction slots\n");
-				return -1;
-			}
-
-			fp->alu.inst[pos].inst0 = NOP_INST0;
-			fp->alu.inst[pos].inst1 = NOP_INST1;
-			fp->alu.inst[pos].inst2 = NOP_INST2;
-			fp->alu.inst[pos].inst3 = NOP_INST3;
-
-			cs->nrslots++;
-		}
-		// Note: When we need both parts (vector and scalar) of a source,
-		// we always try to put them into the same position. This makes the
-		// code easier to read, and it is optimal (i.e. one doesn't gain
-		// anything by splitting the parts).
-		// It also avoids headaches with swizzles that access both parts (i.e WXY)
-		tempused = cs->slot[pos].used;
-		for (i = 0; i < 3; ++i) {
-			tempvsrc[i] = cs->slot[pos].vsrc[i];
-			tempssrc[i] = cs->slot[pos].ssrc[i];
-		}
-
-		for (i = 0; i < argc; ++i) {
-			int flags = (used >> i) & SLOT_SRC_BOTH;
-
-			if (!flags) {
-				srcpos[i] = 0;
-				continue;
-			}
-
-			for (j = 0; j < 3; ++j) {
-				if ((tempused >> j) & flags & SLOT_SRC_VECTOR) {
-					if (tempvsrc[j] != hwsrc[i])
-						continue;
-				}
-
-				if ((tempused >> j) & flags & SLOT_SRC_SCALAR) {
-					if (tempssrc[j] != hwsrc[i])
-						continue;
-				}
-
-				break;
-			}
-
-			if (j == 3)
-				break;
-
-			srcpos[i] = j;
-			tempused |= flags << j;
-			if (flags & SLOT_SRC_VECTOR)
-				tempvsrc[j] = hwsrc[i];
-			if (flags & SLOT_SRC_SCALAR)
-				tempssrc[j] = hwsrc[i];
-		}
-
-		if (i == argc)
-			break;
-	}
-
-	// Found a slot, reserve it
-	cs->slot[pos].used = tempused | (used & SLOT_OP_BOTH);
-	for (i = 0; i < 3; ++i) {
-		cs->slot[pos].vsrc[i] = tempvsrc[i];
-		cs->slot[pos].ssrc[i] = tempssrc[i];
-	}
-
-	for (i = 0; i < argc; ++i) {
-		if (REG_GET_TYPE(src[i]) == REG_TYPE_TEMP) {
-			int regnr = hwsrc[i] & 31;
-
-			if (used & (SLOT_SRC_VECTOR << i)) {
-				if (cs->hwtemps[regnr].vector_lastread < pos)
-					cs->hwtemps[regnr].vector_lastread =
-					    pos;
-			}
-			if (used & (SLOT_SRC_SCALAR << i)) {
-				if (cs->hwtemps[regnr].scalar_lastread < pos)
-					cs->hwtemps[regnr].scalar_lastread =
-					    pos;
-			}
-		}
-	}
-
-	// Emit the source fetch code
-	fp->alu.inst[pos].inst1 &= ~R300_FPI1_SRC_MASK;
-	fp->alu.inst[pos].inst1 |=
-	    ((cs->slot[pos].vsrc[0] << R300_FPI1_SRC0C_SHIFT) |
-	     (cs->slot[pos].vsrc[1] << R300_FPI1_SRC1C_SHIFT) |
-	     (cs->slot[pos].vsrc[2] << R300_FPI1_SRC2C_SHIFT));
-
-	fp->alu.inst[pos].inst3 &= ~R300_FPI3_SRC_MASK;
-	fp->alu.inst[pos].inst3 |=
-	    ((cs->slot[pos].ssrc[0] << R300_FPI3_SRC0A_SHIFT) |
-	     (cs->slot[pos].ssrc[1] << R300_FPI3_SRC1A_SHIFT) |
-	     (cs->slot[pos].ssrc[2] << R300_FPI3_SRC2A_SHIFT));
-
-	// Emit the argument selection code
-	if (emit_vop) {
-		int swz[3];
-
-		for (i = 0; i < 3; ++i) {
-			if (i < argc) {
-				swz[i] = (v_swiz[REG_GET_VSWZ(src[i])].base +
-					  (srcpos[i] *
-					   v_swiz[REG_GET_VSWZ(src[i])].
-					   stride)) | ((src[i] & REG_NEGV_MASK)
-						       ? ARG_NEG : 0) | ((src[i]
-									  &
-									  REG_ABS_MASK)
-									 ?
-									 ARG_ABS
-									 : 0);
-			} else {
-				swz[i] = R300_FPI0_ARGC_ZERO;
-			}
-		}
-
-		fp->alu.inst[pos].inst0 &=
-		    ~(R300_FPI0_ARG0C_MASK | R300_FPI0_ARG1C_MASK |
-		      R300_FPI0_ARG2C_MASK);
-		fp->alu.inst[pos].inst0 |=
-		    (swz[0] << R300_FPI0_ARG0C_SHIFT) | (swz[1] <<
-							 R300_FPI0_ARG1C_SHIFT)
-		    | (swz[2] << R300_FPI0_ARG2C_SHIFT);
-	}
-
-	if (emit_sop) {
-		int swz[3];
-
-		for (i = 0; i < 3; ++i) {
-			if (i < argc) {
-				swz[i] = (s_swiz[REG_GET_SSWZ(src[i])].base +
-					  (srcpos[i] *
-					   s_swiz[REG_GET_SSWZ(src[i])].
-					   stride)) | ((src[i] & REG_NEGV_MASK)
-						       ? ARG_NEG : 0) | ((src[i]
-									  &
-									  REG_ABS_MASK)
-									 ?
-									 ARG_ABS
-									 : 0);
-			} else {
-				swz[i] = R300_FPI2_ARGA_ZERO;
-			}
-		}
-
-		fp->alu.inst[pos].inst2 &=
-		    ~(R300_FPI2_ARG0A_MASK | R300_FPI2_ARG1A_MASK |
-		      R300_FPI2_ARG2A_MASK);
-		fp->alu.inst[pos].inst2 |=
-		    (swz[0] << R300_FPI2_ARG0A_SHIFT) | (swz[1] <<
-							 R300_FPI2_ARG1A_SHIFT)
-		    | (swz[2] << R300_FPI2_ARG2A_SHIFT);
-	}
-
-	return pos;
-}
-
-/**
- * Append an ALU instruction to the instruction list.
- */
-static void emit_arith(struct r300_fragment_program *fp,
-		       int op,
-		       GLuint dest,
-		       int mask,
-		       GLuint src0, GLuint src1, GLuint src2, int flags)
-{
-	COMPILE_STATE;
-	GLuint src[3] = { src0, src1, src2 };
-	int hwdest;
-	GLboolean emit_vop, emit_sop;
-	int vop, sop, argc;
-	int pos;
-
-	vop = r300_fpop[op].v_op;
-	sop = r300_fpop[op].s_op;
-	argc = r300_fpop[op].argc;
-
-	if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT &&
-	    REG_GET_INDEX(dest) == FRAG_RESULT_DEPR) {
-		if (mask & WRITEMASK_Z) {
-			mask = WRITEMASK_W;
-		} else {
-			return;
-		}
-	}
-
-	emit_vop = GL_FALSE;
-	emit_sop = GL_FALSE;
-	if ((mask & WRITEMASK_XYZ) || vop == R300_FPI0_OUTC_DP3)
-		emit_vop = GL_TRUE;
-	if ((mask & WRITEMASK_W) || vop == R300_FPI0_OUTC_REPL_ALPHA)
-		emit_sop = GL_TRUE;
-
-	pos =
-	    find_and_prepare_slot(fp, emit_vop, emit_sop, argc, src, dest,
-				  mask);
-	if (pos < 0)
-		return;
-
-	hwdest = t_hw_dst(fp, dest, GL_FALSE, pos);	/* Note: Side effects wrt register allocation */
-
-	if (flags & PFS_FLAG_SAT) {
-		vop |= R300_FPI0_OUTC_SAT;
-		sop |= R300_FPI2_OUTA_SAT;
-	}
-
-	/* Throw the pieces together and get FPI0/1 */
-	if (emit_vop) {
-		fp->alu.inst[pos].inst0 |= vop;
+	tgt = radeonAppendInstructions(t->Program, 1);
+	_mesa_copy_instructions(tgt, &inst, 1);
+
+	if (inst.Opcode != OPCODE_KIL &&
+	    t->Program->ShadowSamplers & (1 << inst.TexSrcUnit)) {
+		GLuint comparefunc = GL_NEVER + compiler->fp->state.unit[inst.TexSrcUnit].texture_compare_func;
+		GLuint depthmode = compiler->fp->state.unit[inst.TexSrcUnit].depth_texture_mode;
+		int rcptemp = radeonFindFreeTemporary(t);
+		int pass, fail;
+
+		tgt = radeonAppendInstructions(t->Program, 3);
+
+		tgt[0].Opcode = OPCODE_RCP;
+		tgt[0].DstReg.File = PROGRAM_TEMPORARY;
+		tgt[0].DstReg.Index = rcptemp;
+		tgt[0].DstReg.WriteMask = WRITEMASK_W;
+		tgt[0].SrcReg[0] = inst.SrcReg[0];
+		tgt[0].SrcReg[0].Swizzle = SWIZZLE_WWWW;
+
+		tgt[1].Opcode = OPCODE_MAD;
+		tgt[1].DstReg = inst.DstReg;
+		tgt[1].DstReg.WriteMask = orig_inst->DstReg.WriteMask;
+		tgt[1].SrcReg[0] = inst.SrcReg[0];
+		tgt[1].SrcReg[0].Swizzle = SWIZZLE_ZZZZ;
+		tgt[1].SrcReg[1].File = PROGRAM_TEMPORARY;
+		tgt[1].SrcReg[1].Index = rcptemp;
+		tgt[1].SrcReg[1].Swizzle = SWIZZLE_WWWW;
+		tgt[1].SrcReg[2].File = PROGRAM_TEMPORARY;
+		tgt[1].SrcReg[2].Index = inst.DstReg.Index;
+		if (depthmode == 0) /* GL_LUMINANCE */
+			tgt[1].SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z);
+		else if (depthmode == 2) /* GL_ALPHA */
+			tgt[1].SrcReg[2].Swizzle = SWIZZLE_WWWW;
+
+		/* Recall that SrcReg[0] is tex, SrcReg[2] is r and:
+		 *   r  < tex  <=>      -tex+r < 0
+		 *   r >= tex  <=> not (-tex+r < 0 */
+		if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
+			tgt[1].SrcReg[2].NegateBase = tgt[0].SrcReg[2].NegateBase ^ NEGATE_XYZW;
+		else
+			tgt[1].SrcReg[0].NegateBase = tgt[0].SrcReg[0].NegateBase ^ NEGATE_XYZW;
 
-		fp->alu.inst[pos].inst1 |= hwdest << R300_FPI1_DSTC_SHIFT;
+		tgt[2].Opcode = OPCODE_CMP;
+		tgt[2].DstReg = orig_inst->DstReg;
+		tgt[2].SrcReg[0].File = PROGRAM_TEMPORARY;
+		tgt[2].SrcReg[0].Index = tgt[1].DstReg.Index;
 
-		if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
-			if (REG_GET_INDEX(dest) == FRAG_RESULT_COLR) {
-				fp->alu.inst[pos].inst1 |=
-				    (mask & WRITEMASK_XYZ) <<
-				    R300_FPI1_DSTC_OUTPUT_MASK_SHIFT;
-			} else
-				assert(0);
+		if (comparefunc == GL_LESS || comparefunc == GL_GREATER) {
+			pass = 1;
+			fail = 2;
 		} else {
-			fp->alu.inst[pos].inst1 |=
-			    (mask & WRITEMASK_XYZ) <<
-			    R300_FPI1_DSTC_REG_MASK_SHIFT;
-
-			cs->hwtemps[hwdest].vector_valid = pos + 1;
+			pass = 2;
+			fail = 1;
 		}
-	}
 
-	/* And now FPI2/3 */
-	if (emit_sop) {
-		fp->alu.inst[pos].inst2 |= sop;
-
-		if (mask & WRITEMASK_W) {
-			if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
-				if (REG_GET_INDEX(dest) == FRAG_RESULT_COLR) {
-					fp->alu.inst[pos].inst3 |=
-					    (hwdest << R300_FPI3_DSTA_SHIFT) |
-					    R300_FPI3_DSTA_OUTPUT;
-				} else if (REG_GET_INDEX(dest) ==
-					   FRAG_RESULT_DEPR) {
-					fp->alu.inst[pos].inst3 |=
-					    R300_FPI3_DSTA_DEPTH;
-				} else
-					assert(0);
-			} else {
-				fp->alu.inst[pos].inst3 |=
-				    (hwdest << R300_FPI3_DSTA_SHIFT) |
-				    R300_FPI3_DSTA_REG;
+		tgt[2].SrcReg[pass].File = PROGRAM_BUILTIN;
+		tgt[2].SrcReg[pass].Swizzle = SWIZZLE_1111;
+		tgt[2].SrcReg[fail] = shadow_ambient(t->Program, inst.TexSrcUnit);
+	} else if (destredirect) {
+		tgt = radeonAppendInstructions(t->Program, 1);
 
-				cs->hwtemps[hwdest].scalar_valid = pos + 1;
-			}
-		}
+		tgt->Opcode = OPCODE_MOV;
+		tgt->DstReg = orig_inst->DstReg;
+		tgt->SrcReg[0].File = PROGRAM_TEMPORARY;
+		tgt->SrcReg[0].Index = inst.DstReg.Index;
 	}
 
-	return;
+	return GL_TRUE;
 }
 
-#if 0
-static GLuint get_attrib(struct r300_fragment_program *fp, GLuint attr)
+
+static void update_params(r300ContextPtr r300, struct r300_fragment_program *fp)
 {
 	struct gl_fragment_program *mp = &fp->mesa_program;
-	GLuint r = undef;
-
-	if (!(mp->Base.InputsRead & (1 << attr))) {
-		ERROR("Attribute %d was not provided!\n", attr);
-		return undef;
-	}
 
-	REG_SET_TYPE(r, REG_TYPE_INPUT);
-	REG_SET_INDEX(r, attr);
-	REG_SET_VALID(r, GL_TRUE);
-	return r;
+	/* Ask Mesa nicely to fill in ParameterValues for us */
+	if (mp->Base.Parameters)
+		_mesa_load_state_parameters(r300->radeon.glCtx, mp->Base.Parameters);
 }
-#endif
-
-static GLfloat SinCosConsts[2][4] = {
-	{
-	 1.273239545,		// 4/PI
-	 -0.405284735,		// -4/(PI*PI)
-	 3.141592654,		// PI
-	 0.2225			// weight
-	 },
-	{
-	 0.75,
-	 0.0,
-	 0.159154943,		// 1/(2*PI)
-	 6.283185307		// 2*PI
-	 }
-};
+
 
 /**
- * Emit a LIT instruction.
- * \p flags may be PFS_FLAG_SAT
+ * Transform the program to support fragment.position.
  *
- * Definition of LIT (from ARB_fragment_program):
- * tmp = VectorLoad(op0);
- * if (tmp.x < 0) tmp.x = 0;
- * if (tmp.y < 0) tmp.y = 0;
- * if (tmp.w < -(128.0-epsilon)) tmp.w = -(128.0-epsilon);
- * else if (tmp.w > 128-epsilon) tmp.w = 128-epsilon;
- * result.x = 1.0;
- * result.y = tmp.x;
- * result.z = (tmp.x > 0) ? RoughApproxPower(tmp.y, tmp.w) : 0.0;
- * result.w = 1.0;
+ * Introduce a small fragment at the start of the program that will be
+ * the only code that directly reads the FRAG_ATTRIB_WPOS input.
+ * All other code pieces that reference that input will be rewritten
+ * to read from a newly allocated temporary.
  *
- * The longest path of computation is the one leading to result.z,
- * consisting of 5 operations. This implementation of LIT takes
- * 5 slots. So unless there's some special undocumented opcode,
- * this implementation is potentially optimal. Unfortunately,
- * emit_arith is a bit too conservative because it doesn't understand
- * partial writes to the vector component.
+ * \todo if/when r5xx supports the radeon_program architecture, this is a
+ * likely candidate for code sharing.
  */
-static const GLfloat LitConst[4] =
-    { 127.999999, 127.999999, 127.999999, -127.999999 };
-
-static void emit_lit(struct r300_fragment_program *fp,
-		     GLuint dest, int mask, GLuint src, int flags)
+static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler)
 {
-	COMPILE_STATE;
-	GLuint cnst;
-	int needTemporary;
-	GLuint temp;
-
-	cnst = emit_const4fv(fp, LitConst);
-
-	needTemporary = 0;
-	if ((mask & WRITEMASK_XYZW) != WRITEMASK_XYZW) {
-		needTemporary = 1;
-	} else if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
-		// LIT is typically followed by DP3/DP4, so there's no point
-		// in creating special code for this case
-		needTemporary = 1;
-	}
-
-	if (needTemporary) {
-		temp = keep(get_temp_reg(fp));
-	} else {
-		temp = keep(dest);
-	}
-
-	// Note: The order of emit_arith inside the slots is relevant,
-	// because emit_arith only looks at scalar vs. vector when resolving
-	// dependencies, and it does not consider individual vector components,
-	// so swizzling between the two parts can create fake dependencies.
-
-	// First slot
-	emit_arith(fp, PFS_OP_MAX, temp, WRITEMASK_XY,
-		   keep(src), pfs_zero, undef, 0);
-	emit_arith(fp, PFS_OP_MAX, temp, WRITEMASK_W, src, cnst, undef, 0);
-
-	// Second slot
-	emit_arith(fp, PFS_OP_MIN, temp, WRITEMASK_Z,
-		   swizzle(temp, W, W, W, W), cnst, undef, 0);
-	emit_arith(fp, PFS_OP_LG2, temp, WRITEMASK_W,
-		   swizzle(temp, Y, Y, Y, Y), undef, undef, 0);
-
-	// Third slot
-	// If desired, we saturate the y result here.
-	// This does not affect the use as a condition variable in the CMP later
-	emit_arith(fp, PFS_OP_MAD, temp, WRITEMASK_W,
-		   temp, swizzle(temp, Z, Z, Z, Z), pfs_zero, 0);
-	emit_arith(fp, PFS_OP_MAD, temp, WRITEMASK_Y,
-		   swizzle(temp, X, X, X, X), pfs_one, pfs_zero, flags);
-
-	// Fourth slot
-	emit_arith(fp, PFS_OP_MAD, temp, WRITEMASK_X,
-		   pfs_one, pfs_one, pfs_zero, 0);
-	emit_arith(fp, PFS_OP_EX2, temp, WRITEMASK_W, temp, undef, undef, 0);
-
-	// Fifth slot
-	emit_arith(fp, PFS_OP_CMP, temp, WRITEMASK_Z,
-		   pfs_zero, swizzle(temp, W, W, W, W),
-		   negate(swizzle(temp, Y, Y, Y, Y)), flags);
-	emit_arith(fp, PFS_OP_MAD, temp, WRITEMASK_W, pfs_one, pfs_one,
-		   pfs_zero, 0);
-
-	if (needTemporary) {
-		emit_arith(fp, PFS_OP_MAD, dest, mask,
-			   temp, pfs_one, pfs_zero, flags);
-		free_temp(fp, temp);
-	} else {
-		// Decrease refcount of the destination
-		t_hw_dst(fp, dest, GL_FALSE, cs->nrslots);
-	}
-}
-
-static GLboolean parse_program(struct r300_fragment_program *fp)
-{
-	struct gl_fragment_program *mp = &fp->mesa_program;
-	const struct prog_instruction *inst = mp->Base.Instructions;
-	struct prog_instruction *fpi;
-	GLuint src[3], dest, temp[2];
-	int flags, mask = 0;
-	int const_sin[2];
-
-	if (!inst || inst[0].Opcode == OPCODE_END) {
-		ERROR("empty program?\n");
-		return GL_FALSE;
-	}
+	GLuint InputsRead = compiler->fp->mesa_program.Base.InputsRead;
 
-	for (fpi = mp->Base.Instructions; fpi->Opcode != OPCODE_END; fpi++) {
-		if (fpi->SaturateMode == SATURATE_ZERO_ONE)
-			flags = PFS_FLAG_SAT;
-		else
-			flags = 0;
-
-		if (fpi->Opcode != OPCODE_KIL) {
-			dest = t_dst(fp, fpi->DstReg);
-			mask = fpi->DstReg.WriteMask;
-		}
-
-		switch (fpi->Opcode) {
-		case OPCODE_ABS:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   absolute(src[0]), pfs_one, pfs_zero, flags);
-			break;
-		case OPCODE_ADD:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], pfs_one, src[1], flags);
-			break;
-		case OPCODE_CMP:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			src[2] = t_src(fp, fpi->SrcReg[2]);
-			/* ARB_f_p - if src0.c < 0.0 ? src1.c : src2.c
-			 *    r300 - if src2.c < 0.0 ? src1.c : src0.c
-			 */
-			emit_arith(fp, PFS_OP_CMP, dest, mask,
-				   src[2], src[1], src[0], flags);
-			break;
-		case OPCODE_COS:
-			/*
-			 * cos using a parabola (see SIN):
-			 * cos(x):
-			 *   x = (x/(2*PI))+0.75
-			 *   x = frac(x)
-			 *   x = (x*2*PI)-PI
-			 *   result = sin(x)
-			 */
-			temp[0] = get_temp_reg(fp);
-			const_sin[0] = emit_const4fv(fp, SinCosConsts[0]);
-			const_sin[1] = emit_const4fv(fp, SinCosConsts[1]);
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-
-			/* add 0.5*PI and do range reduction */
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_X,
-				   swizzle(src[0], X, X, X, X),
-				   swizzle(const_sin[1], Z, Z, Z, Z),
-				   swizzle(const_sin[1], X, X, X, X), 0);
-
-			emit_arith(fp, PFS_OP_FRC, temp[0], WRITEMASK_X,
-				   swizzle(temp[0], X, X, X, X),
-				   undef, undef, 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_Z, swizzle(temp[0], X, X, X, X), swizzle(const_sin[1], W, W, W, W),	//2*PI
-				   negate(swizzle(const_sin[0], Z, Z, Z, Z)),	//-PI
-				   0);
-
-			/* SIN */
-
-			emit_arith(fp, PFS_OP_MAD, temp[0],
-				   WRITEMASK_X | WRITEMASK_Y, swizzle(temp[0],
-								      Z, Z, Z,
-								      Z),
-				   const_sin[0], pfs_zero, 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_X,
-				   swizzle(temp[0], Y, Y, Y, Y),
-				   absolute(swizzle(temp[0], Z, Z, Z, Z)),
-				   swizzle(temp[0], X, X, X, X), 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_Y,
-				   swizzle(temp[0], X, X, X, X),
-				   absolute(swizzle(temp[0], X, X, X, X)),
-				   negate(swizzle(temp[0], X, X, X, X)), 0);
-
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   swizzle(temp[0], Y, Y, Y, Y),
-				   swizzle(const_sin[0], W, W, W, W),
-				   swizzle(temp[0], X, X, X, X), flags);
-
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_DP3:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_DP3, dest, mask,
-				   src[0], src[1], undef, flags);
-			break;
-		case OPCODE_DP4:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_DP4, dest, mask,
-				   src[0], src[1], undef, flags);
-			break;
-		case OPCODE_DPH:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			/* src0.xyz1 -> temp
-			 * DP4 dest, temp, src1
-			 */
-#if 0
-			temp[0] = get_temp_reg(fp);
-			src[0].s_swz = SWIZZLE_ONE;
-			emit_arith(fp, PFS_OP_MAD, temp[0], mask,
-				   src[0], pfs_one, pfs_zero, 0);
-			emit_arith(fp, PFS_OP_DP4, dest, mask,
-				   temp[0], src[1], undef, flags);
-			free_temp(fp, temp[0]);
-#else
-			emit_arith(fp, PFS_OP_DP4, dest, mask,
-				   swizzle(src[0], X, Y, Z, ONE), src[1],
-				   undef, flags);
-#endif
-			break;
-		case OPCODE_DST:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			/* dest.y = src0.y * src1.y */
-			if (mask & WRITEMASK_Y)
-				emit_arith(fp, PFS_OP_MAD, dest, WRITEMASK_Y,
-					   keep(src[0]), keep(src[1]),
-					   pfs_zero, flags);
-			/* dest.z = src0.z */
-			if (mask & WRITEMASK_Z)
-				emit_arith(fp, PFS_OP_MAD, dest, WRITEMASK_Z,
-					   src[0], pfs_one, pfs_zero, flags);
-			/* result.x = 1.0
-			 * result.w = src1.w */
-			if (mask & WRITEMASK_XW) {
-				REG_SET_VSWZ(src[1], SWIZZLE_111);	/*Cheat */
-				emit_arith(fp, PFS_OP_MAD, dest,
-					   mask & WRITEMASK_XW,
-					   src[1], pfs_one, pfs_zero, flags);
-			}
-			break;
-		case OPCODE_EX2:
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_EX2, dest, mask,
-				   src[0], undef, undef, flags);
-			break;
-		case OPCODE_FLR:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			temp[0] = get_temp_reg(fp);
-			/* FRC temp, src0
-			 * MAD dest, src0, 1.0, -temp
-			 */
-			emit_arith(fp, PFS_OP_FRC, temp[0], mask,
-				   keep(src[0]), undef, undef, 0);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], pfs_one, negate(temp[0]), flags);
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_FRC:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_FRC, dest, mask,
-				   src[0], undef, undef, flags);
-			break;
-		case OPCODE_KIL:
-			emit_tex(fp, fpi, R300_FPITX_OP_KIL);
-			break;
-		case OPCODE_LG2:
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_LG2, dest, mask,
-				   src[0], undef, undef, flags);
-			break;
-		case OPCODE_LIT:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			emit_lit(fp, dest, mask, src[0], flags);
-			break;
-		case OPCODE_LRP:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			src[2] = t_src(fp, fpi->SrcReg[2]);
-			/* result = tmp0tmp1 + (1 - tmp0)tmp2
-			 *        = tmp0tmp1 + tmp2 + (-tmp0)tmp2
-			 *     MAD temp, -tmp0, tmp2, tmp2
-			 *     MAD result, tmp0, tmp1, temp
-			 */
-			temp[0] = get_temp_reg(fp);
-			emit_arith(fp, PFS_OP_MAD, temp[0], mask,
-				   negate(keep(src[0])), keep(src[2]), src[2],
-				   0);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], src[1], temp[0], flags);
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_MAD:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			src[2] = t_src(fp, fpi->SrcReg[2]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], src[1], src[2], flags);
-			break;
-		case OPCODE_MAX:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_MAX, dest, mask,
-				   src[0], src[1], undef, flags);
-			break;
-		case OPCODE_MIN:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_MIN, dest, mask,
-				   src[0], src[1], undef, flags);
-			break;
-		case OPCODE_MOV:
-		case OPCODE_SWZ:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], pfs_one, pfs_zero, flags);
-			break;
-		case OPCODE_MUL:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], src[1], pfs_zero, flags);
-			break;
-		case OPCODE_POW:
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-			src[1] = t_scalar_src(fp, fpi->SrcReg[1]);
-			temp[0] = get_temp_reg(fp);
-			emit_arith(fp, PFS_OP_LG2, temp[0], WRITEMASK_W,
-				   src[0], undef, undef, 0);
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_W,
-				   temp[0], src[1], pfs_zero, 0);
-			emit_arith(fp, PFS_OP_EX2, dest, fpi->DstReg.WriteMask,
-				   temp[0], undef, undef, 0);
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_RCP:
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_RCP, dest, mask,
-				   src[0], undef, undef, flags);
-			break;
-		case OPCODE_RSQ:
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-			emit_arith(fp, PFS_OP_RSQ, dest, mask,
-				   absolute(src[0]), pfs_zero, pfs_zero, flags);
-			break;
-		case OPCODE_SCS:
-			/*
-			 * scs using a parabola :
-			 * scs(x):
-			 *   result.x = sin(-abs(x)+0.5*PI)  (cos)
-			 *   result.y = sin(x)               (sin)
-			 *
-			 */
-			temp[0] = get_temp_reg(fp);
-			temp[1] = get_temp_reg(fp);
-			const_sin[0] = emit_const4fv(fp, SinCosConsts[0]);
-			const_sin[1] = emit_const4fv(fp, SinCosConsts[1]);
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-
-			/* x = -abs(x)+0.5*PI */
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_Z, swizzle(const_sin[0], Z, Z, Z, Z),	//PI
-				   pfs_half,
-				   negate(abs
-					  (swizzle(keep(src[0]), X, X, X, X))),
-				   0);
-
-			/* C*x (sin) */
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_W,
-				   swizzle(const_sin[0], Y, Y, Y, Y),
-				   swizzle(keep(src[0]), X, X, X, X),
-				   pfs_zero, 0);
-
-			/* B*x, C*x (cos) */
-			emit_arith(fp, PFS_OP_MAD, temp[0],
-				   WRITEMASK_X | WRITEMASK_Y, swizzle(temp[0],
-								      Z, Z, Z,
-								      Z),
-				   const_sin[0], pfs_zero, 0);
-
-			/* B*x (sin) */
-			emit_arith(fp, PFS_OP_MAD, temp[1], WRITEMASK_W,
-				   swizzle(const_sin[0], X, X, X, X),
-				   keep(src[0]), pfs_zero, 0);
-
-			/* y = B*x + C*x*abs(x) (sin) */
-			emit_arith(fp, PFS_OP_MAD, temp[1], WRITEMASK_Z,
-				   absolute(src[0]),
-				   swizzle(temp[0], W, W, W, W),
-				   swizzle(temp[1], W, W, W, W), 0);
-
-			/* y = B*x + C*x*abs(x) (cos) */
-			emit_arith(fp, PFS_OP_MAD, temp[1], WRITEMASK_W,
-				   swizzle(temp[0], Y, Y, Y, Y),
-				   absolute(swizzle(temp[0], Z, Z, Z, Z)),
-				   swizzle(temp[0], X, X, X, X), 0);
-
-			/* y*abs(y) - y (cos), y*abs(y) - y (sin) */
-			emit_arith(fp, PFS_OP_MAD, temp[0],
-				   WRITEMASK_X | WRITEMASK_Y, swizzle(temp[1],
-								      W, Z, Y,
-								      X),
-				   absolute(swizzle(temp[1], W, Z, Y, X)),
-				   negate(swizzle(temp[1], W, Z, Y, X)), 0);
-
-			/* dest.xy = mad(temp.xy, P, temp2.wz) */
-			emit_arith(fp, PFS_OP_MAD, dest,
-				   mask & (WRITEMASK_X | WRITEMASK_Y), temp[0],
-				   swizzle(const_sin[0], W, W, W, W),
-				   swizzle(temp[1], W, Z, Y, X), flags);
-
-			free_temp(fp, temp[0]);
-			free_temp(fp, temp[1]);
-			break;
-		case OPCODE_SGE:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			temp[0] = get_temp_reg(fp);
-			/* temp = src0 - src1
-			 * dest.c = (temp.c < 0.0) ? 0 : 1
-			 */
-			emit_arith(fp, PFS_OP_MAD, temp[0], mask,
-				   src[0], pfs_one, negate(src[1]), 0);
-			emit_arith(fp, PFS_OP_CMP, dest, mask,
-				   pfs_one, pfs_zero, temp[0], 0);
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_SIN:
-			/*
-			 *  using a parabola:
-			 * sin(x) = 4/pi * x + -4/(pi*pi) * x * abs(x)
-			 * extra precision is obtained by weighting against
-			 * itself squared.
-			 */
-
-			temp[0] = get_temp_reg(fp);
-			const_sin[0] = emit_const4fv(fp, SinCosConsts[0]);
-			const_sin[1] = emit_const4fv(fp, SinCosConsts[1]);
-			src[0] = t_scalar_src(fp, fpi->SrcReg[0]);
-
-			/* do range reduction */
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_X,
-				   swizzle(keep(src[0]), X, X, X, X),
-				   swizzle(const_sin[1], Z, Z, Z, Z),
-				   pfs_half, 0);
-
-			emit_arith(fp, PFS_OP_FRC, temp[0], WRITEMASK_X,
-				   swizzle(temp[0], X, X, X, X),
-				   undef, undef, 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_Z, swizzle(temp[0], X, X, X, X), swizzle(const_sin[1], W, W, W, W),	//2*PI
-				   negate(swizzle(const_sin[0], Z, Z, Z, Z)),	//PI
-				   0);
-
-			/* SIN */
-
-			emit_arith(fp, PFS_OP_MAD, temp[0],
-				   WRITEMASK_X | WRITEMASK_Y, swizzle(temp[0],
-								      Z, Z, Z,
-								      Z),
-				   const_sin[0], pfs_zero, 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_X,
-				   swizzle(temp[0], Y, Y, Y, Y),
-				   absolute(swizzle(temp[0], Z, Z, Z, Z)),
-				   swizzle(temp[0], X, X, X, X), 0);
-
-			emit_arith(fp, PFS_OP_MAD, temp[0], WRITEMASK_Y,
-				   swizzle(temp[0], X, X, X, X),
-				   absolute(swizzle(temp[0], X, X, X, X)),
-				   negate(swizzle(temp[0], X, X, X, X)), 0);
-
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   swizzle(temp[0], Y, Y, Y, Y),
-				   swizzle(const_sin[0], W, W, W, W),
-				   swizzle(temp[0], X, X, X, X), flags);
-
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_SLT:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			temp[0] = get_temp_reg(fp);
-			/* temp = src0 - src1
-			 * dest.c = (temp.c < 0.0) ? 1 : 0
-			 */
-			emit_arith(fp, PFS_OP_MAD, temp[0], mask,
-				   src[0], pfs_one, negate(src[1]), 0);
-			emit_arith(fp, PFS_OP_CMP, dest, mask,
-				   pfs_zero, pfs_one, temp[0], 0);
-			free_temp(fp, temp[0]);
-			break;
-		case OPCODE_SUB:
-			src[0] = t_src(fp, fpi->SrcReg[0]);
-			src[1] = t_src(fp, fpi->SrcReg[1]);
-			emit_arith(fp, PFS_OP_MAD, dest, mask,
-				   src[0], pfs_one, negate(src[1]), flags);
-			break;
-		case OPCODE_TEX:
-			emit_tex(fp, fpi, R300_FPITX_OP_TEX);
-			break;
-		case OPCODE_TXB:
-			emit_tex(fp, fpi, R300_FPITX_OP_TXB);
-			break;
-		case OPCODE_TXP:
-			emit_tex(fp, fpi, R300_FPITX_OP_TXP);
-			break;
-		case OPCODE_XPD:{
-				src[0] = t_src(fp, fpi->SrcReg[0]);
-				src[1] = t_src(fp, fpi->SrcReg[1]);
-				temp[0] = get_temp_reg(fp);
-				/* temp = src0.zxy * src1.yzx */
-				emit_arith(fp, PFS_OP_MAD, temp[0],
-					   WRITEMASK_XYZ, swizzle(keep(src[0]),
-								  Z, X, Y, W),
-					   swizzle(keep(src[1]), Y, Z, X, W),
-					   pfs_zero, 0);
-				/* dest.xyz = src0.yzx * src1.zxy - temp
-				 * dest.w       = undefined
-				 * */
-				emit_arith(fp, PFS_OP_MAD, dest,
-					   mask & WRITEMASK_XYZ, swizzle(src[0],
-									 Y, Z,
-									 X, W),
-					   swizzle(src[1], Z, X, Y, W),
-					   negate(temp[0]), flags);
-				/* cleanup */
-				free_temp(fp, temp[0]);
-				break;
-			}
-		default:
-			ERROR("unknown fpi->Opcode %d\n", fpi->Opcode);
-			break;
-		}
-
-		if (fp->error)
-			return GL_FALSE;
-
-	}
-
-	return GL_TRUE;
-}
+	if (!(InputsRead & FRAG_BIT_WPOS))
+		return;
 
-static void insert_wpos(struct gl_program *prog)
-{
 	static gl_state_index tokens[STATE_LENGTH] = {
 		STATE_INTERNAL, STATE_R300_WINDOW_DIMENSION, 0, 0, 0
 	};
 	struct prog_instruction *fpi;
 	GLuint window_index;
 	int i = 0;
-	GLuint tempregi = prog->NumTemporaries;
-	/* should do something else if no temps left... */
-	prog->NumTemporaries++;
+	GLuint tempregi = _mesa_find_free_register(compiler->program, PROGRAM_TEMPORARY);
 
-	fpi = _mesa_alloc_instructions(prog->NumInstructions + 3);
-	_mesa_init_instructions(fpi, prog->NumInstructions + 3);
+	_mesa_insert_instructions(compiler->program, 0, 3);
+	fpi = compiler->program->Instructions;
 
 	/* perspective divide */
 	fpi[i].Opcode = OPCODE_RCP;
@@ -2027,7 +303,7 @@ static void insert_wpos(struct gl_program *prog)
 	i++;
 
 	/* viewport transformation */
-	window_index = _mesa_add_state_reference(prog->Parameters, tokens);
+	window_index = _mesa_add_state_reference(compiler->program->Parameters, tokens);
 
 	fpi[i].Opcode = OPCODE_MAD;
 
@@ -2052,242 +328,182 @@ static void insert_wpos(struct gl_program *prog)
 	    MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
 	i++;
 
-	_mesa_copy_instructions(&fpi[i], prog->Instructions,
-				prog->NumInstructions);
-
-	free(prog->Instructions);
-
-	prog->Instructions = fpi;
-
-	prog->NumInstructions += i;
-	fpi = &prog->Instructions[prog->NumInstructions - 1];
-
-	assert(fpi->Opcode == OPCODE_END);
-
-	for (fpi = &prog->Instructions[3]; fpi->Opcode != OPCODE_END; fpi++) {
-		for (i = 0; i < 3; i++)
-			if (fpi->SrcReg[i].File == PROGRAM_INPUT &&
-			    fpi->SrcReg[i].Index == FRAG_ATTRIB_WPOS) {
-				fpi->SrcReg[i].File = PROGRAM_TEMPORARY;
-				fpi->SrcReg[i].Index = tempregi;
+	for (; i < compiler->program->NumInstructions; ++i) {
+		int reg;
+		for (reg = 0; reg < 3; reg++) {
+			if (fpi[i].SrcReg[reg].File == PROGRAM_INPUT &&
+			    fpi[i].SrcReg[reg].Index == FRAG_ATTRIB_WPOS) {
+				fpi[i].SrcReg[reg].File = PROGRAM_TEMPORARY;
+				fpi[i].SrcReg[reg].Index = tempregi;
 			}
+		}
 	}
 }
 
-/* - Init structures
- * - Determine what hwregs each input corresponds to
- */
-static void init_program(r300ContextPtr r300, struct r300_fragment_program *fp)
+
+static void nqssadce_init(struct nqssadce_state* s)
 {
-	struct r300_pfs_compile_state *cs = NULL;
-	struct gl_fragment_program *mp = &fp->mesa_program;
-	struct prog_instruction *fpi;
-	GLuint InputsRead = mp->Base.InputsRead;
-	GLuint temps_used = 0;	/* for fp->temps[] */
-	int i, j;
-
-	/* New compile, reset tracking data */
-	fp->optimization =
-	    driQueryOptioni(&r300->radeon.optionCache, "fp_optimization");
-	fp->translated = GL_FALSE;
-	fp->error = GL_FALSE;
-	fp->cs = cs = &(R300_CONTEXT(fp->ctx)->state.pfs_compile);
-	fp->tex.length = 0;
-	fp->cur_node = 0;
-	fp->first_node_has_tex = 0;
-	fp->const_nr = 0;
-	fp->max_temp_idx = 0;
-	fp->node[0].alu_end = -1;
-	fp->node[0].tex_end = -1;
-
-	_mesa_memset(cs, 0, sizeof(*fp->cs));
-	for (i = 0; i < PFS_MAX_ALU_INST; i++) {
-		for (j = 0; j < 3; j++) {
-			cs->slot[i].vsrc[j] = SRC_CONST;
-			cs->slot[i].ssrc[j] = SRC_CONST;
-		}
-	}
+	s->Outputs[FRAG_RESULT_COLR].Sourced = WRITEMASK_XYZW;
+	s->Outputs[FRAG_RESULT_DEPR].Sourced = WRITEMASK_W;
+}
 
-	/* Work out what temps the Mesa inputs correspond to, this must match
-	 * what setup_rs_unit does, which shouldn't be a problem as rs_unit
-	 * configures itself based on the fragprog's InputsRead
-	 *
-	 * NOTE: this depends on get_hw_temp() allocating registers in order,
-	 * starting from register 0.
-	 */
 
-	/* Texcoords come first */
-	for (i = 0; i < fp->ctx->Const.MaxTextureUnits; i++) {
-		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
-			cs->inputs[FRAG_ATTRIB_TEX0 + i].refcount = 0;
-			cs->inputs[FRAG_ATTRIB_TEX0 + i].reg =
-			    get_hw_temp(fp, 0);
-		}
+static GLuint build_dtm(GLuint depthmode)
+{
+	switch(depthmode) {
+	default:
+	case GL_LUMINANCE: return 0;
+	case GL_INTENSITY: return 1;
+	case GL_ALPHA: return 2;
 	}
-	InputsRead &= ~FRAG_BITS_TEX_ANY;
+}
 
-	/* fragment position treated as a texcoord */
-	if (InputsRead & FRAG_BIT_WPOS) {
-		cs->inputs[FRAG_ATTRIB_WPOS].refcount = 0;
-		cs->inputs[FRAG_ATTRIB_WPOS].reg = get_hw_temp(fp, 0);
-		insert_wpos(&mp->Base);
-	}
-	InputsRead &= ~FRAG_BIT_WPOS;
+static GLuint build_func(GLuint comparefunc)
+{
+	return comparefunc - GL_NEVER;
+}
 
-	/* Then primary colour */
-	if (InputsRead & FRAG_BIT_COL0) {
-		cs->inputs[FRAG_ATTRIB_COL0].refcount = 0;
-		cs->inputs[FRAG_ATTRIB_COL0].reg = get_hw_temp(fp, 0);
-	}
-	InputsRead &= ~FRAG_BIT_COL0;
 
-	/* Secondary color */
-	if (InputsRead & FRAG_BIT_COL1) {
-		cs->inputs[FRAG_ATTRIB_COL1].refcount = 0;
-		cs->inputs[FRAG_ATTRIB_COL1].reg = get_hw_temp(fp, 0);
-	}
-	InputsRead &= ~FRAG_BIT_COL1;
-
-	/* Anything else */
-	if (InputsRead) {
-		WARN_ONCE("Don't know how to handle inputs 0x%x\n", InputsRead);
-		/* force read from hwreg 0 for now */
-		for (i = 0; i < 32; i++)
-			if (InputsRead & (1 << i))
-				cs->inputs[i].reg = 0;
-	}
+/**
+ * Collect all external state that is relevant for compiling the given
+ * fragment program.
+ */
+static void build_state(
+	r300ContextPtr r300,
+	struct r300_fragment_program *fp,
+	struct r300_fragment_program_external_state *state)
+{
+	int unit;
 
-	/* Pre-parse the mesa program, grabbing refcounts on input/temp regs.
-	 * That way, we can free up the reg when it's no longer needed
-	 */
-	if (!mp->Base.Instructions) {
-		ERROR("No instructions found in program\n");
-		return;
-	}
+	_mesa_bzero(state, sizeof(*state));
 
-	for (fpi = mp->Base.Instructions; fpi->Opcode != OPCODE_END; fpi++) {
-		int idx;
-
-		for (i = 0; i < 3; i++) {
-			idx = fpi->SrcReg[i].Index;
-			switch (fpi->SrcReg[i].File) {
-			case PROGRAM_TEMPORARY:
-				if (!(temps_used & (1 << idx))) {
-					cs->temps[idx].reg = -1;
-					cs->temps[idx].refcount = 1;
-					temps_used |= (1 << idx);
-				} else
-					cs->temps[idx].refcount++;
-				break;
-			case PROGRAM_INPUT:
-				cs->inputs[idx].refcount++;
-				break;
-			default:
-				break;
-			}
-		}
+	for(unit = 0; unit < 16; ++unit) {
+		if (fp->mesa_program.Base.ShadowSamplers & (1 << unit)) {
+			struct gl_texture_object* tex = r300->radeon.glCtx->Texture.Unit[unit]._Current;
 
-		idx = fpi->DstReg.Index;
-		if (fpi->DstReg.File == PROGRAM_TEMPORARY) {
-			if (!(temps_used & (1 << idx))) {
-				cs->temps[idx].reg = -1;
-				cs->temps[idx].refcount = 1;
-				temps_used |= (1 << idx);
-			} else
-				cs->temps[idx].refcount++;
+			state->unit[unit].depth_texture_mode = build_dtm(tex->DepthMode);
+			state->unit[unit].texture_compare_func = build_func(tex->CompareFunc);
 		}
 	}
-	cs->temp_in_use = temps_used;
 }
 
-static void update_params(struct r300_fragment_program *fp)
-{
-	struct gl_fragment_program *mp = &fp->mesa_program;
-
-	/* Ask Mesa nicely to fill in ParameterValues for us */
-	if (mp->Base.Parameters)
-		_mesa_load_state_parameters(fp->ctx, mp->Base.Parameters);
-}
 
 void r300TranslateFragmentShader(r300ContextPtr r300,
 				 struct r300_fragment_program *fp)
 {
-	struct r300_pfs_compile_state *cs = NULL;
+	struct r300_fragment_program_external_state state;
+
+	build_state(r300, fp, &state);
+	if (_mesa_memcmp(&fp->state, &state, sizeof(state))) {
+		/* TODO: cache compiled programs */
+		fp->translated = GL_FALSE;
+		_mesa_memcpy(&fp->state, &state, sizeof(state));
+	}
 
 	if (!fp->translated) {
+		struct r300_fragment_program_compiler compiler;
+
+		compiler.r300 = r300;
+		compiler.fp = fp;
+		compiler.code = &fp->code;
+		compiler.program = _mesa_clone_program(r300->radeon.glCtx, &fp->mesa_program.Base);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Fragment Program: Initial program:\n");
+			_mesa_print_program(compiler.program);
+		}
 
-		init_program(r300, fp);
-		cs = fp->cs;
+		insert_WPOS_trailer(&compiler);
+
+		struct radeon_program_transformation transformations[] = {
+			{ &transform_TEX, &compiler },
+			{ &radeonTransformALU, 0 },
+			{ &radeonTransformTrigSimple, 0 }
+		};
+		radeonLocalTransform(
+			r300->radeon.glCtx,
+			compiler.program,
+			3, transformations);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Fragment Program: After native rewrite:\n");
+			_mesa_print_program(compiler.program);
+		}
 
-		if (parse_program(fp) == GL_FALSE) {
-			dump_program(fp);
-			return;
+		struct radeon_nqssadce_descr nqssadce = {
+			.Init = &nqssadce_init,
+			.IsNativeSwizzle = &r300FPIsNativeSwizzle,
+			.BuildSwizzle = &r300FPBuildSwizzle,
+			.RewriteDepthOut = GL_TRUE
+		};
+		radeonNqssaDce(r300->radeon.glCtx, compiler.program, &nqssadce);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Compiler: after NqSSA-DCE:\n");
+			_mesa_print_program(compiler.program);
 		}
 
-		/* Finish off */
-		fp->node[fp->cur_node].alu_end =
-		    cs->nrslots - fp->node[fp->cur_node].alu_offset - 1;
-		if (fp->node[fp->cur_node].tex_end < 0)
-			fp->node[fp->cur_node].tex_end = 0;
-		fp->alu_offset = 0;
-		fp->alu_end = cs->nrslots - 1;
-		fp->tex_offset = 0;
-		fp->tex_end = fp->tex.length ? fp->tex.length - 1 : 0;
-		assert(fp->node[fp->cur_node].alu_end >= 0);
-		assert(fp->alu_end >= 0);
-
-		fp->translated = GL_TRUE;
-		if (RADEON_DEBUG & DEBUG_PIXEL)
-			dump_program(fp);
-		r300UpdateStateParameters(fp->ctx, _NEW_PROGRAM);
+		if (!r300FragmentProgramEmit(&compiler))
+			fp->error = GL_TRUE;
+
+		/* Subtle: Rescue any parameters that have been added during transformations */
+		_mesa_free_parameter_list(fp->mesa_program.Base.Parameters);
+		fp->mesa_program.Base.Parameters = compiler.program->Parameters;
+		compiler.program->Parameters = 0;
+
+		_mesa_reference_program(r300->radeon.glCtx, &compiler.program, NULL);
+
+		if (!fp->error)
+			fp->translated = GL_TRUE;
+		if (fp->error || (RADEON_DEBUG & DEBUG_PIXEL))
+			r300FragmentProgramDump(fp, &fp->code);
+		r300UpdateStateParameters(r300->radeon.glCtx, _NEW_PROGRAM);
 	}
 
-	update_params(fp);
+	update_params(r300, fp);
 }
 
 /* just some random things... */
-static void dump_program(struct r300_fragment_program *fp)
+void r300FragmentProgramDump(
+	struct r300_fragment_program *fp,
+	struct r300_fragment_program_code *code)
 {
 	int n, i, j;
 	static int pc = 0;
 
 	fprintf(stderr, "pc=%d*************************************\n", pc++);
 
-	fprintf(stderr, "Mesa program:\n");
-	fprintf(stderr, "-------------\n");
-	_mesa_print_program(&fp->mesa_program.Base);
-	fflush(stdout);
-
 	fprintf(stderr, "Hardware program\n");
 	fprintf(stderr, "----------------\n");
 
-	for (n = 0; n < (fp->cur_node + 1); n++) {
+	for (n = 0; n < (code->cur_node + 1); n++) {
 		fprintf(stderr, "NODE %d: alu_offset: %d, tex_offset: %d, "
-			"alu_end: %d, tex_end: %d\n", n,
-			fp->node[n].alu_offset,
-			fp->node[n].tex_offset,
-			fp->node[n].alu_end, fp->node[n].tex_end);
+			"alu_end: %d, tex_end: %d, flags: %08x\n", n,
+			code->node[n].alu_offset,
+			code->node[n].tex_offset,
+			code->node[n].alu_end, code->node[n].tex_end,
+			code->node[n].flags);
 
-		if (fp->tex.length) {
+		if (n > 0 || code->first_node_has_tex) {
 			fprintf(stderr, "  TEX:\n");
-			for (i = fp->node[n].tex_offset;
-			     i <= fp->node[n].tex_offset + fp->node[n].tex_end;
+			for (i = code->node[n].tex_offset;
+			     i <= code->node[n].tex_offset + code->node[n].tex_end;
 			     ++i) {
 				const char *instr;
 
-				switch ((fp->tex.
-					 inst[i] >> R300_FPITX_OPCODE_SHIFT) &
+				switch ((code->tex.
+					 inst[i] >> R300_TEX_INST_SHIFT) &
 					15) {
-				case R300_FPITX_OP_TEX:
+				case R300_TEX_OP_LD:
 					instr = "TEX";
 					break;
-				case R300_FPITX_OP_KIL:
+				case R300_TEX_OP_KIL:
 					instr = "KIL";
 					break;
-				case R300_FPITX_OP_TXP:
+				case R300_TEX_OP_TXP:
 					instr = "TXP";
 					break;
-				case R300_FPITX_OP_TXB:
+				case R300_TEX_OP_TXB:
 					instr = "TXB";
 					break;
 				default:
@@ -2297,22 +513,20 @@ static void dump_program(struct r300_fragment_program *fp)
 				fprintf(stderr,
 					"    %s t%i, %c%i, texture[%i]   (%08x)\n",
 					instr,
-					(fp->tex.
-					 inst[i] >> R300_FPITX_DST_SHIFT) & 31,
-					(fp->tex.
-					 inst[i] & R300_FPITX_SRC_CONST) ? 'c' :
+					(code->tex.
+					 inst[i] >> R300_DST_ADDR_SHIFT) & 31,
 					't',
-					(fp->tex.
-					 inst[i] >> R300_FPITX_SRC_SHIFT) & 31,
-					(fp->tex.
-					 inst[i] & R300_FPITX_IMAGE_MASK) >>
-					R300_FPITX_IMAGE_SHIFT,
-					fp->tex.inst[i]);
+					(code->tex.
+					 inst[i] >> R300_SRC_ADDR_SHIFT) & 31,
+					(code->tex.
+					 inst[i] & R300_TEX_ID_MASK) >>
+					R300_TEX_ID_SHIFT,
+					code->tex.inst[i]);
 			}
 		}
 
-		for (i = fp->node[n].alu_offset;
-		     i <= fp->node[n].alu_offset + fp->node[n].alu_end; ++i) {
+		for (i = code->node[n].alu_offset;
+		     i <= code->node[n].alu_offset + code->node[n].alu_end; ++i) {
 			char srcc[3][10], dstc[20];
 			char srca[3][10], dsta[20];
 			char argc[3][20];
@@ -2320,8 +534,8 @@ static void dump_program(struct r300_fragment_program *fp)
 			char flags[5], tmp[10];
 
 			for (j = 0; j < 3; ++j) {
-				int regc = fp->alu.inst[i].inst1 >> (j * 6);
-				int rega = fp->alu.inst[i].inst3 >> (j * 6);
+				int regc = code->alu.inst[i].inst1 >> (j * 6);
+				int rega = code->alu.inst[i].inst3 >> (j * 6);
 
 				sprintf(srcc[j], "%c%i",
 					(regc & 32) ? 'c' : 't', regc & 31);
@@ -2331,46 +545,46 @@ static void dump_program(struct r300_fragment_program *fp)
 
 			dstc[0] = 0;
 			sprintf(flags, "%s%s%s",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_REG_X) ? "x" : "",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_REG_Y) ? "y" : "",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_REG_Z) ? "z" : "");
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_REG_X) ? "x" : "",
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_REG_Y) ? "y" : "",
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_REG_Z) ? "z" : "");
 			if (flags[0] != 0) {
 				sprintf(dstc, "t%i.%s ",
-					(fp->alu.inst[i].
-					 inst1 >> R300_FPI1_DSTC_SHIFT) & 31,
+					(code->alu.inst[i].
+					 inst1 >> R300_ALU_DSTC_SHIFT) & 31,
 					flags);
 			}
 			sprintf(flags, "%s%s%s",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_OUTPUT_X) ? "x" : "",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_OUTPUT_Y) ? "y" : "",
-				(fp->alu.inst[i].
-				 inst1 & R300_FPI1_DSTC_OUTPUT_Z) ? "z" : "");
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_OUTPUT_X) ? "x" : "",
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_OUTPUT_Y) ? "y" : "",
+				(code->alu.inst[i].
+				 inst1 & R300_ALU_DSTC_OUTPUT_Z) ? "z" : "");
 			if (flags[0] != 0) {
 				sprintf(tmp, "o%i.%s",
-					(fp->alu.inst[i].
-					 inst1 >> R300_FPI1_DSTC_SHIFT) & 31,
+					(code->alu.inst[i].
+					 inst1 >> R300_ALU_DSTC_SHIFT) & 31,
 					flags);
 				strcat(dstc, tmp);
 			}
 
 			dsta[0] = 0;
-			if (fp->alu.inst[i].inst3 & R300_FPI3_DSTA_REG) {
+			if (code->alu.inst[i].inst3 & R300_ALU_DSTA_REG) {
 				sprintf(dsta, "t%i.w ",
-					(fp->alu.inst[i].
-					 inst3 >> R300_FPI3_DSTA_SHIFT) & 31);
+					(code->alu.inst[i].
+					 inst3 >> R300_ALU_DSTA_SHIFT) & 31);
 			}
-			if (fp->alu.inst[i].inst3 & R300_FPI3_DSTA_OUTPUT) {
+			if (code->alu.inst[i].inst3 & R300_ALU_DSTA_OUTPUT) {
 				sprintf(tmp, "o%i.w ",
-					(fp->alu.inst[i].
-					 inst3 >> R300_FPI3_DSTA_SHIFT) & 31);
+					(code->alu.inst[i].
+					 inst3 >> R300_ALU_DSTA_SHIFT) & 31);
 				strcat(dsta, tmp);
 			}
-			if (fp->alu.inst[i].inst3 & R300_FPI3_DSTA_DEPTH) {
+			if (code->alu.inst[i].inst3 & R300_ALU_DSTA_DEPTH) {
 				strcat(dsta, "Z");
 			}
 
@@ -2378,31 +592,31 @@ static void dump_program(struct r300_fragment_program *fp)
 				"%3i: xyz: %3s %3s %3s -> %-20s (%08x)\n"
 				"       w: %3s %3s %3s -> %-20s (%08x)\n", i,
 				srcc[0], srcc[1], srcc[2], dstc,
-				fp->alu.inst[i].inst1, srca[0], srca[1],
-				srca[2], dsta, fp->alu.inst[i].inst3);
+				code->alu.inst[i].inst1, srca[0], srca[1],
+				srca[2], dsta, code->alu.inst[i].inst3);
 
 			for (j = 0; j < 3; ++j) {
-				int regc = fp->alu.inst[i].inst0 >> (j * 7);
-				int rega = fp->alu.inst[i].inst2 >> (j * 7);
+				int regc = code->alu.inst[i].inst0 >> (j * 7);
+				int rega = code->alu.inst[i].inst2 >> (j * 7);
 				int d;
 				char buf[20];
 
 				d = regc & 31;
 				if (d < 12) {
 					switch (d % 4) {
-					case R300_FPI0_ARGC_SRC0C_XYZ:
+					case R300_ALU_ARGC_SRC0C_XYZ:
 						sprintf(buf, "%s.xyz",
 							srcc[d / 4]);
 						break;
-					case R300_FPI0_ARGC_SRC0C_XXX:
+					case R300_ALU_ARGC_SRC0C_XXX:
 						sprintf(buf, "%s.xxx",
 							srcc[d / 4]);
 						break;
-					case R300_FPI0_ARGC_SRC0C_YYY:
+					case R300_ALU_ARGC_SRC0C_YYY:
 						sprintf(buf, "%s.yyy",
 							srcc[d / 4]);
 						break;
-					case R300_FPI0_ARGC_SRC0C_ZZZ:
+					case R300_ALU_ARGC_SRC0C_ZZZ:
 						sprintf(buf, "%s.zzz",
 							srcc[d / 4]);
 						break;
@@ -2465,8 +679,8 @@ static void dump_program(struct r300_fragment_program *fp)
 			fprintf(stderr, "     xyz: %8s %8s %8s    op: %08x\n"
 				"       w: %8s %8s %8s    op: %08x\n",
 				argc[0], argc[1], argc[2],
-				fp->alu.inst[i].inst0, arga[0], arga[1],
-				arga[2], fp->alu.inst[i].inst2);
+				code->alu.inst[i].inst0, arga[0], arga[1],
+				arga[2], code->alu.inst[i].inst2);
 		}
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.h b/src/mesa/drivers/dri/r300/r300_fragprog.h
index 72fca778455..94fb554fb37 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.h
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.h
@@ -33,72 +33,100 @@
 #ifndef __R300_FRAGPROG_H_
 #define __R300_FRAGPROG_H_
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "shader/program.h"
 #include "shader/prog_instruction.h"
 
 #include "r300_context.h"
-
-typedef struct r300_fragment_program_swizzle {
-	GLuint length;
-	GLuint src[4];
-	GLuint inst[8];
-} r300_fragment_program_swizzle_t;
-
-/* supported hw opcodes */
-#define PFS_OP_MAD 0
-#define PFS_OP_DP3 1
-#define PFS_OP_DP4 2
-#define PFS_OP_MIN 3
-#define PFS_OP_MAX 4
-#define PFS_OP_CMP 5
-#define PFS_OP_FRC 6
-#define PFS_OP_EX2 7
-#define PFS_OP_LG2 8
-#define PFS_OP_RCP 9
-#define PFS_OP_RSQ 10
-#define PFS_OP_REPL_ALPHA 11
-#define PFS_OP_CMPH 12
-#define MAX_PFS_OP 12
-
-#define PFS_FLAG_SAT	(1 << 0)
-#define PFS_FLAG_ABS	(1 << 1)
-
-#define ARG_NEG			(1 << 5)
-#define ARG_ABS			(1 << 6)
-#define ARG_MASK		(127 << 0)
-#define ARG_STRIDE		7
-#define SRC_CONST		(1 << 5)
-#define SRC_MASK		(63 << 0)
-#define SRC_STRIDE		6
-
-#define NOP_INST0 (						 \
-		(R300_FPI0_OUTC_MAD) |				 \
-		(R300_FPI0_ARGC_ZERO << R300_FPI0_ARG0C_SHIFT) | \
-		(R300_FPI0_ARGC_ZERO << R300_FPI0_ARG1C_SHIFT) | \
-		(R300_FPI0_ARGC_ZERO << R300_FPI0_ARG2C_SHIFT))
-#define NOP_INST1 (					     \
-		((0 | SRC_CONST) << R300_FPI1_SRC0C_SHIFT) | \
-		((0 | SRC_CONST) << R300_FPI1_SRC1C_SHIFT) | \
-		((0 | SRC_CONST) << R300_FPI1_SRC2C_SHIFT))
-#define NOP_INST2 ( \
-		(R300_FPI2_OUTA_MAD) |				 \
-		(R300_FPI2_ARGA_ZERO << R300_FPI2_ARG0A_SHIFT) | \
-		(R300_FPI2_ARGA_ZERO << R300_FPI2_ARG1A_SHIFT) | \
-		(R300_FPI2_ARGA_ZERO << R300_FPI2_ARG2A_SHIFT))
-#define NOP_INST3 (					     \
-		((0 | SRC_CONST) << R300_FPI3_SRC0A_SHIFT) | \
-		((0 | SRC_CONST) << R300_FPI3_SRC1A_SHIFT) | \
-		((0 | SRC_CONST) << R300_FPI3_SRC2A_SHIFT))
+#include "radeon_program.h"
 
 #define DRI_CONF_FP_OPTIMIZATION_SPEED   0
 #define DRI_CONF_FP_OPTIMIZATION_QUALITY 1
 
+#if 1
+
+/**
+ * Fragment program helper macros
+ */
+
+/* Produce unshifted source selectors */
+#define FP_TMP(idx) (idx)
+#define FP_CONST(idx) ((idx) | (1 << 5))
+
+/* Produce source/dest selector dword */
+#define FP_SELC_MASK_NO		0
+#define FP_SELC_MASK_X		1
+#define FP_SELC_MASK_Y		2
+#define FP_SELC_MASK_XY		3
+#define FP_SELC_MASK_Z		4
+#define FP_SELC_MASK_XZ		5
+#define FP_SELC_MASK_YZ		6
+#define FP_SELC_MASK_XYZ	7
+
+#define FP_SELC(destidx,regmask,outmask,src0,src1,src2) \
+	(((destidx) << R300_ALU_DSTC_SHIFT) |		\
+	 (FP_SELC_MASK_##regmask << 23) |		\
+	 (FP_SELC_MASK_##outmask << 26) |		\
+	 ((src0) << R300_ALU_SRC0C_SHIFT) |		\
+	 ((src1) << R300_ALU_SRC1C_SHIFT) |		\
+	 ((src2) << R300_ALU_SRC2C_SHIFT))
+
+#define FP_SELA_MASK_NO		0
+#define FP_SELA_MASK_W		1
+
+#define FP_SELA(destidx,regmask,outmask,src0,src1,src2) \
+	(((destidx) << R300_ALU_DSTA_SHIFT) |		\
+	 (FP_SELA_MASK_##regmask << 23) |		\
+	 (FP_SELA_MASK_##outmask << 24) |		\
+	 ((src0) << R300_ALU_SRC0A_SHIFT) |		\
+	 ((src1) << R300_ALU_SRC1A_SHIFT) |		\
+	 ((src2) << R300_ALU_SRC2A_SHIFT))
+
+/* Produce unshifted argument selectors */
+#define FP_ARGC(source)	R300_ALU_ARGC_##source
+#define FP_ARGA(source) R300_ALU_ARGA_##source
+#define FP_ABS(arg) ((arg) | (1 << 6))
+#define FP_NEG(arg) ((arg) ^ (1 << 5))
+
+/* Produce instruction dword */
+#define FP_INSTRC(opcode,arg0,arg1,arg2) \
+	(R300_ALU_OUTC_##opcode | 		\
+	((arg0) << R300_ALU_ARG0C_SHIFT) |	\
+	((arg1) << R300_ALU_ARG1C_SHIFT) |	\
+	((arg2) << R300_ALU_ARG2C_SHIFT))
+
+#define FP_INSTRA(opcode,arg0,arg1,arg2) \
+	(R300_ALU_OUTA_##opcode | 		\
+	((arg0) << R300_ALU_ARG0A_SHIFT) |	\
+	((arg1) << R300_ALU_ARG1A_SHIFT) |	\
+	((arg2) << R300_ALU_ARG2A_SHIFT))
+
+#endif
+
 struct r300_fragment_program;
 
 extern void r300TranslateFragmentShader(r300ContextPtr r300,
 					struct r300_fragment_program *fp);
 
+
+/**
+ * Used internally by the r300 fragment program code to store compile-time
+ * only data.
+ */
+struct r300_fragment_program_compiler {
+	r300ContextPtr r300;
+	struct r300_fragment_program *fp;
+	struct r300_fragment_program_code *code;
+	struct gl_program *program;
+};
+
+extern GLboolean r300FragmentProgramEmit(struct r300_fragment_program_compiler *compiler);
+
+
+extern void r300FragmentProgramDump(
+	struct r300_fragment_program *fp,
+	struct r300_fragment_program_code *code);
+
 #endif
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_emit.c b/src/mesa/drivers/dri/r300/r300_fragprog_emit.c
new file mode 100644
index 00000000000..9f0b7e35349
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_emit.c
@@ -0,0 +1,344 @@
+/*
+ * Copyright (C) 2005 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * \file
+ *
+ * Emit the r300_fragment_program_code that can be understood by the hardware.
+ * Input is a pre-transformed radeon_program.
+ *
+ * \author Ben Skeggs <darktama@iinet.net.au>
+ *
+ * \author Jerome Glisse <j.glisse@gmail.com>
+ *
+ * \todo FogOption
+ */
+
+#include "r300_fragprog.h"
+
+#include "radeon_program_pair.h"
+#include "r300_fragprog_swizzle.h"
+#include "r300_reg.h"
+
+
+#define PROG_CODE \
+	struct r300_fragment_program_compiler *c = (struct r300_fragment_program_compiler*)data; \
+	struct r300_fragment_program_code *code = c->code
+
+#define error(fmt, args...) do {			\
+		fprintf(stderr, "%s::%s(): " fmt "\n",	\
+			__FILE__, __FUNCTION__, ##args);	\
+	} while(0)
+
+
+static GLboolean emit_const(void* data, GLuint file, GLuint index, GLuint *hwindex)
+{
+	PROG_CODE;
+
+	for (*hwindex = 0; *hwindex < code->const_nr; ++*hwindex) {
+		if (code->constant[*hwindex].File == file &&
+		    code->constant[*hwindex].Index == index)
+			break;
+	}
+
+	if (*hwindex >= code->const_nr) {
+		if (*hwindex >= PFS_NUM_CONST_REGS) {
+			error("Out of hw constants!\n");
+			return GL_FALSE;
+		}
+
+		code->const_nr++;
+		code->constant[*hwindex].File = file;
+		code->constant[*hwindex].Index = index;
+	}
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Mark a temporary register as used.
+ */
+static void use_temporary(struct r300_fragment_program_code *code, GLuint index)
+{
+	if (index > code->max_temp_idx)
+		code->max_temp_idx = index;
+}
+
+
+static GLuint translate_rgb_opcode(GLuint opcode)
+{
+	switch(opcode) {
+	case OPCODE_CMP: return R300_ALU_OUTC_CMP;
+	case OPCODE_DP3: return R300_ALU_OUTC_DP3;
+	case OPCODE_DP4: return R300_ALU_OUTC_DP4;
+	case OPCODE_FRC: return R300_ALU_OUTC_FRC;
+	default:
+		error("translate_rgb_opcode(%i): Unknown opcode", opcode);
+		/* fall through */
+	case OPCODE_NOP:
+		/* fall through */
+	case OPCODE_MAD: return R300_ALU_OUTC_MAD;
+	case OPCODE_MAX: return R300_ALU_OUTC_MAX;
+	case OPCODE_MIN: return R300_ALU_OUTC_MIN;
+	case OPCODE_REPL_ALPHA: return R300_ALU_OUTC_REPL_ALPHA;
+	}
+}
+
+static GLuint translate_alpha_opcode(GLuint opcode)
+{
+	switch(opcode) {
+	case OPCODE_CMP: return R300_ALU_OUTA_CMP;
+	case OPCODE_DP3: return R300_ALU_OUTA_DP4;
+	case OPCODE_DP4: return R300_ALU_OUTA_DP4;
+	case OPCODE_EX2: return R300_ALU_OUTA_EX2;
+	case OPCODE_FRC: return R300_ALU_OUTA_FRC;
+	case OPCODE_LG2: return R300_ALU_OUTA_LG2;
+	default:
+		error("translate_rgb_opcode(%i): Unknown opcode", opcode);
+		/* fall through */
+	case OPCODE_NOP:
+		/* fall through */
+	case OPCODE_MAD: return R300_ALU_OUTA_MAD;
+	case OPCODE_MAX: return R300_ALU_OUTA_MAX;
+	case OPCODE_MIN: return R300_ALU_OUTA_MIN;
+	case OPCODE_RCP: return R300_ALU_OUTA_RCP;
+	case OPCODE_RSQ: return R300_ALU_OUTA_RSQ;
+	}
+}
+
+/**
+ * Emit one paired ALU instruction.
+ */
+static GLboolean emit_alu(void* data, struct radeon_pair_instruction* inst)
+{
+	PROG_CODE;
+
+	if (code->alu.length >= PFS_MAX_ALU_INST) {
+		error("Too many ALU instructions");
+		return GL_FALSE;
+	}
+
+	int ip = code->alu.length++;
+	int j;
+	code->node[code->cur_node].alu_end++;
+
+	code->alu.inst[ip].inst0 = translate_rgb_opcode(inst->RGB.Opcode);
+	code->alu.inst[ip].inst2 = translate_alpha_opcode(inst->Alpha.Opcode);
+
+	for(j = 0; j < 3; ++j) {
+		GLuint src = inst->RGB.Src[j].Index | (inst->RGB.Src[j].Constant << 5);
+		if (!inst->RGB.Src[j].Constant)
+			use_temporary(code, inst->RGB.Src[j].Index);
+		code->alu.inst[ip].inst1 |= src << (6*j);
+
+		src = inst->Alpha.Src[j].Index | (inst->Alpha.Src[j].Constant << 5);
+		if (!inst->Alpha.Src[j].Constant)
+			use_temporary(code, inst->Alpha.Src[j].Index);
+		code->alu.inst[ip].inst3 |= src << (6*j);
+
+		GLuint arg = r300FPTranslateRGBSwizzle(inst->RGB.Arg[j].Source, inst->RGB.Arg[j].Swizzle);
+		arg |= inst->RGB.Arg[j].Abs << 6;
+		arg |= inst->RGB.Arg[j].Negate << 5;
+		code->alu.inst[ip].inst0 |= arg << (7*j);
+
+		arg = r300FPTranslateAlphaSwizzle(inst->Alpha.Arg[j].Source, inst->Alpha.Arg[j].Swizzle);
+		arg |= inst->Alpha.Arg[j].Abs << 6;
+		arg |= inst->Alpha.Arg[j].Negate << 5;
+		code->alu.inst[ip].inst2 |= arg << (7*j);
+	}
+
+	if (inst->RGB.Saturate)
+		code->alu.inst[ip].inst0 |= R300_ALU_OUTC_CLAMP;
+	if (inst->Alpha.Saturate)
+		code->alu.inst[ip].inst2 |= R300_ALU_OUTA_CLAMP;
+
+	if (inst->RGB.WriteMask) {
+		use_temporary(code, inst->RGB.DestIndex);
+		code->alu.inst[ip].inst1 |=
+			(inst->RGB.DestIndex << R300_ALU_DSTC_SHIFT) |
+			(inst->RGB.WriteMask << R300_ALU_DSTC_REG_MASK_SHIFT);
+	}
+	if (inst->RGB.OutputWriteMask) {
+		code->alu.inst[ip].inst1 |= (inst->RGB.OutputWriteMask << R300_ALU_DSTC_OUTPUT_MASK_SHIFT);
+		code->node[code->cur_node].flags |= R300_RGBA_OUT;
+	}
+
+	if (inst->Alpha.WriteMask) {
+		use_temporary(code, inst->Alpha.DestIndex);
+		code->alu.inst[ip].inst3 |=
+			(inst->Alpha.DestIndex << R300_ALU_DSTA_SHIFT) |
+			R300_ALU_DSTA_REG;
+	}
+	if (inst->Alpha.OutputWriteMask) {
+		code->alu.inst[ip].inst3 |= R300_ALU_DSTA_OUTPUT;
+		code->node[code->cur_node].flags |= R300_RGBA_OUT;
+	}
+	if (inst->Alpha.DepthWriteMask) {
+		code->alu.inst[ip].inst3 |= R300_ALU_DSTA_DEPTH;
+		code->node[code->cur_node].flags |= R300_W_OUT;
+		c->fp->WritesDepth = GL_TRUE;
+	}
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Finish the current node without advancing to the next one.
+ */
+static GLboolean finish_node(struct r300_fragment_program_compiler *c)
+{
+	struct r300_fragment_program_code *code = c->code;
+	struct r300_fragment_program_node *node = &code->node[code->cur_node];
+
+	if (node->alu_end < 0) {
+		/* Generate a single NOP for this node */
+		struct radeon_pair_instruction inst;
+		_mesa_bzero(&inst, sizeof(inst));
+		if (!emit_alu(c, &inst))
+			return GL_FALSE;
+	}
+
+	if (node->tex_end < 0) {
+		if (code->cur_node == 0) {
+			node->tex_end = 0;
+		} else {
+			error("Node %i has no TEX instructions", code->cur_node);
+			return GL_FALSE;
+		}
+	} else {
+		if (code->cur_node == 0)
+			code->first_node_has_tex = 1;
+	}
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Begin a block of texture instructions.
+ * Create the necessary indirection.
+ */
+static GLboolean begin_tex(void* data)
+{
+	PROG_CODE;
+
+	if (code->cur_node == 0) {
+		if (code->node[0].alu_end < 0 &&
+		    code->node[0].tex_end < 0)
+			return GL_TRUE;
+	}
+
+	if (code->cur_node == 3) {
+		error("Too many texture indirections");
+		return GL_FALSE;
+	}
+
+	if (!finish_node(c))
+		return GL_FALSE;
+
+	struct r300_fragment_program_node *node = &code->node[++code->cur_node];
+	node->alu_offset = code->alu.length;
+	node->alu_end = -1;
+	node->tex_offset = code->tex.length;
+	node->tex_end = -1;
+	return GL_TRUE;
+}
+
+
+static GLboolean emit_tex(void* data, struct prog_instruction* inst)
+{
+	PROG_CODE;
+
+	if (code->tex.length >= PFS_MAX_TEX_INST) {
+		error("Too many TEX instructions");
+		return GL_FALSE;
+	}
+
+	GLuint unit = inst->TexSrcUnit;
+	GLuint dest = inst->DstReg.Index;
+	GLuint opcode;
+
+	switch(inst->Opcode) {
+	case OPCODE_KIL: opcode = R300_TEX_OP_KIL; break;
+	case OPCODE_TEX: opcode = R300_TEX_OP_LD; break;
+	case OPCODE_TXB: opcode = R300_TEX_OP_TXB; break;
+	case OPCODE_TXP: opcode = R300_TEX_OP_TXP; break;
+	default:
+		error("Unknown texture opcode %i", inst->Opcode);
+		return GL_FALSE;
+	}
+
+	if (inst->Opcode == OPCODE_KIL) {
+		unit = 0;
+		dest = 0;
+	} else {
+		use_temporary(code, dest);
+	}
+
+	use_temporary(code, inst->SrcReg[0].Index);
+
+	code->node[code->cur_node].tex_end++;
+	code->tex.inst[code->tex.length++] =
+		(inst->SrcReg[0].Index << R300_SRC_ADDR_SHIFT) |
+		(dest << R300_DST_ADDR_SHIFT) |
+		(unit << R300_TEX_ID_SHIFT) |
+		(opcode << R300_TEX_INST_SHIFT);
+	return GL_TRUE;
+}
+
+
+static const struct radeon_pair_handler pair_handler = {
+	.EmitConst = &emit_const,
+	.EmitPaired = &emit_alu,
+	.EmitTex = &emit_tex,
+	.BeginTexBlock = &begin_tex,
+	.MaxHwTemps = PFS_NUM_TEMP_REGS
+};
+
+/**
+ * Final compilation step: Turn the intermediate radeon_program into
+ * machine-readable instructions.
+ */
+GLboolean r300FragmentProgramEmit(struct r300_fragment_program_compiler *compiler)
+{
+	struct r300_fragment_program_code *code = compiler->code;
+
+	_mesa_bzero(code, sizeof(struct r300_fragment_program_code));
+	code->node[0].alu_end = -1;
+	code->node[0].tex_end = -1;
+
+	if (!radeonPairProgram(compiler->r300->radeon.glCtx, compiler->program, &pair_handler, compiler))
+		return GL_FALSE;
+
+	if (!finish_node(compiler))
+		return GL_FALSE;
+
+	return GL_TRUE;
+}
+
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c
new file mode 100644
index 00000000000..a86d2bd4712
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c
@@ -0,0 +1,227 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * @file
+ * Utilities to deal with the somewhat odd restriction on R300 fragment
+ * program swizzles.
+ */
+
+#include "r300_fragprog_swizzle.h"
+
+#include "r300_reg.h"
+#include "radeon_nqssadce.h"
+
+#define MAKE_SWZ3(x, y, z) (MAKE_SWIZZLE4(SWIZZLE_##x, SWIZZLE_##y, SWIZZLE_##z, SWIZZLE_ZERO))
+
+struct swizzle_data {
+	GLuint hash; /**< swizzle value this matches */
+	GLuint base; /**< base value for hw swizzle */
+	GLuint stride; /**< difference in base between arg0/1/2 */
+};
+
+static const struct swizzle_data native_swizzles[] = {
+	{MAKE_SWZ3(X, Y, Z), R300_ALU_ARGC_SRC0C_XYZ, 4},
+	{MAKE_SWZ3(X, X, X), R300_ALU_ARGC_SRC0C_XXX, 4},
+	{MAKE_SWZ3(Y, Y, Y), R300_ALU_ARGC_SRC0C_YYY, 4},
+	{MAKE_SWZ3(Z, Z, Z), R300_ALU_ARGC_SRC0C_ZZZ, 4},
+	{MAKE_SWZ3(W, W, W), R300_ALU_ARGC_SRC0A, 1},
+	{MAKE_SWZ3(Y, Z, X), R300_ALU_ARGC_SRC0C_YZX, 1},
+	{MAKE_SWZ3(Z, X, Y), R300_ALU_ARGC_SRC0C_ZXY, 1},
+	{MAKE_SWZ3(W, Z, Y), R300_ALU_ARGC_SRC0CA_WZY, 1},
+	{MAKE_SWZ3(ONE, ONE, ONE), R300_ALU_ARGC_ONE, 0},
+	{MAKE_SWZ3(ZERO, ZERO, ZERO), R300_ALU_ARGC_ZERO, 0}
+};
+
+static const int num_native_swizzles = sizeof(native_swizzles)/sizeof(native_swizzles[0]);
+
+
+/**
+ * Find a native RGB swizzle that matches the given swizzle.
+ * Returns 0 if none found.
+ */
+static const struct swizzle_data* lookup_native_swizzle(GLuint swizzle)
+{
+	int i, comp;
+
+	for(i = 0; i < num_native_swizzles; ++i) {
+		const struct swizzle_data* sd = &native_swizzles[i];
+		for(comp = 0; comp < 3; ++comp) {
+			GLuint swz = GET_SWZ(swizzle, comp);
+			if (swz == SWIZZLE_NIL)
+				continue;
+			if (swz != GET_SWZ(sd->hash, comp))
+				break;
+		}
+		if (comp == 3)
+			return sd;
+	}
+
+	return 0;
+}
+
+
+/**
+ * Check whether the given instruction supports the swizzle and negate
+ * combinations in the given source register.
+ */
+GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
+{
+	if (reg.Abs)
+		reg.NegateBase = 0;
+
+	if (opcode == OPCODE_KIL ||
+	    opcode == OPCODE_TEX ||
+	    opcode == OPCODE_TXB ||
+	    opcode == OPCODE_TXP) {
+		int j;
+
+		if (reg.Abs || reg.NegateBase != (15*reg.NegateAbs))
+			return GL_FALSE;
+
+		for(j = 0; j < 4; ++j) {
+			GLuint swz = GET_SWZ(reg.Swizzle, j);
+			if (swz == SWIZZLE_NIL)
+				continue;
+			if (swz != j)
+				return GL_FALSE;
+		}
+
+		return GL_TRUE;
+	}
+
+	GLuint relevant = 0;
+	int j;
+
+	for(j = 0; j < 3; ++j)
+		if (GET_SWZ(reg.Swizzle, j) != SWIZZLE_NIL)
+			relevant |= 1 << j;
+
+	if ((reg.NegateBase & relevant) && (reg.NegateBase & relevant) != relevant)
+		return GL_FALSE;
+
+	if (!lookup_native_swizzle(reg.Swizzle))
+		return GL_FALSE;
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Generate MOV dst, src using only native swizzles.
+ */
+void r300FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst, struct prog_src_register src)
+{
+	if (src.Abs)
+		src.NegateBase = 0;
+
+	while(dst.WriteMask) {
+		const struct swizzle_data *best_swizzle = 0;
+		GLuint best_matchcount = 0;
+		GLuint best_matchmask = 0;
+		GLboolean rgbnegate;
+		int i, comp;
+
+		for(i = 0; i < num_native_swizzles; ++i) {
+			const struct swizzle_data *sd = &native_swizzles[i];
+			GLuint matchcount = 0;
+			GLuint matchmask = 0;
+			for(comp = 0; comp < 3; ++comp) {
+				if (!GET_BIT(dst.WriteMask, comp))
+					continue;
+				GLuint swz = GET_SWZ(src.Swizzle, comp);
+				if (swz == SWIZZLE_NIL)
+					continue;
+				if (swz == GET_SWZ(sd->hash, comp)) {
+					matchcount++;
+					matchmask |= 1 << comp;
+				}
+			}
+			if (matchcount > best_matchcount) {
+				best_swizzle = sd;
+				best_matchcount = matchcount;
+				best_matchmask = matchmask;
+				if (matchmask == (dst.WriteMask & WRITEMASK_XYZ))
+					break;
+			}
+		}
+
+		if ((src.NegateBase & best_matchmask) != 0) {
+			best_matchmask &= src.NegateBase;
+			rgbnegate = !src.NegateAbs;
+		} else {
+			rgbnegate = src.NegateAbs;
+		}
+
+		struct prog_instruction *inst;
+
+		_mesa_insert_instructions(s->Program, s->IP, 1);
+		inst = s->Program->Instructions + s->IP++;
+		inst->Opcode = OPCODE_MOV;
+		inst->DstReg = dst;
+		inst->DstReg.WriteMask &= (best_matchmask | WRITEMASK_W);
+		inst->SrcReg[0] = src;
+		/* Note: We rely on NqSSA/DCE to set unused swizzle components to NIL */
+
+		dst.WriteMask &= ~inst->DstReg.WriteMask;
+	}
+}
+
+
+/**
+ * Translate an RGB (XYZ) swizzle into the hardware code for the given
+ * instruction source.
+ */
+GLuint r300FPTranslateRGBSwizzle(GLuint src, GLuint swizzle)
+{
+	const struct swizzle_data* sd = lookup_native_swizzle(swizzle);
+
+	if (!sd) {
+		_mesa_printf("Not a native swizzle: %08x\n", swizzle);
+		return 0;
+	}
+
+	return sd->base + src*sd->stride;
+}
+
+
+/**
+ * Translate an Alpha (W) swizzle into the hardware code for the given
+ * instruction source.
+ */
+GLuint r300FPTranslateAlphaSwizzle(GLuint src, GLuint swizzle)
+{
+	if (swizzle < 3)
+		return swizzle + 3*src;
+
+	switch(swizzle) {
+	case SWIZZLE_W: return R300_ALU_ARGA_SRC0A + src;
+	case SWIZZLE_ONE: return R300_ALU_ARGA_ONE;
+	case SWIZZLE_ZERO: return R300_ALU_ARGA_ZERO;
+	default: return R300_ALU_ARGA_ONE;
+	}
+}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.h b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.h
index 3ded41417e1..231bf4eef5f 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_query.h
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2007 Ben Skeggs.
+ * Copyright (C) 2008 Nicolai Haehnle.
  *
  * All Rights Reserved.
  *
@@ -25,14 +25,18 @@
  *
  */
 
-#ifndef __NOUVEAU_QUERY_H__
-#define __NOUVEAU_QUERY_H__
+#ifndef __R300_FRAGPROG_SWIZZLE_H_
+#define __R300_FRAGPROG_SWIZZLE_H_
 
-typedef struct nouveau_query_object_t {
-	struct gl_query_object mesa;
+#include "main/glheader.h"
+#include "shader/prog_instruction.h"
 
-	int notifier_id;
-} nouveau_query_object;
+struct nqssadce_state;
 
-extern void nouveauQueryInitFuncs(GLcontext *ctx);
-#endif
+GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg);
+void r300FPBuildSwizzle(struct nqssadce_state*, struct prog_dst_register dst, struct prog_src_register src);
+
+GLuint r300FPTranslateRGBSwizzle(GLuint src, GLuint swizzle);
+GLuint r300FPTranslateAlphaSwizzle(GLuint src, GLuint swizzle);
+
+#endif /* __R300_FRAGPROG_SWIZZLE_H_ */
diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.c b/src/mesa/drivers/dri/r300/r300_ioctl.c
index 90f5027c9ad..ee85e229f0b 100644
--- a/src/mesa/drivers/dri/r300/r300_ioctl.c
+++ b/src/mesa/drivers/dri/r300/r300_ioctl.c
@@ -40,10 +40,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <sched.h>
 #include <errno.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "macros.h"
-#include "context.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
 
 #include "r300_context.h"
@@ -51,9 +51,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r300_ioctl.h"
 #include "r300_cmdbuf.h"
 #include "r300_state.h"
-#include "r300_program.h"
+#include "r300_vertprog.h"
 #include "radeon_reg.h"
 #include "r300_emit.h"
+#include "r300_fragprog.h"
 
 #include "vblank.h"
 
@@ -106,19 +107,19 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 	e32(cbpitch);
 
 	R300_STATECHANGE(r300, cmk);
-	reg_start(R300_RB3D_COLORMASK, 0);
+	reg_start(RB3D_COLOR_CHANNEL_MASK, 0);
 
 	if (flags & CLEARBUFFER_COLOR) {
-		e32((ctx->Color.ColorMask[BCOMP] ? R300_COLORMASK0_B : 0) |
-		    (ctx->Color.ColorMask[GCOMP] ? R300_COLORMASK0_G : 0) |
-		    (ctx->Color.ColorMask[RCOMP] ? R300_COLORMASK0_R : 0) |
-		    (ctx->Color.ColorMask[ACOMP] ? R300_COLORMASK0_A : 0));
+		e32((ctx->Color.ColorMask[BCOMP] ? RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0 : 0) |
+		    (ctx->Color.ColorMask[GCOMP] ? RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0 : 0) |
+		    (ctx->Color.ColorMask[RCOMP] ? RB3D_COLOR_CHANNEL_MASK_RED_MASK0 : 0) |
+		    (ctx->Color.ColorMask[ACOMP] ? RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0 : 0));
 	} else {
 		e32(0x0);
 	}
 
 	R300_STATECHANGE(r300, zs);
-	reg_start(R300_RB3D_ZSTENCIL_CNTL_0, 2);
+	reg_start(R300_ZB_CNTL, 2);
 
 	{
 		uint32_t t1, t2;
@@ -127,37 +128,28 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 		t2 = 0x0;
 
 		if (flags & CLEARBUFFER_DEPTH) {
-			t1 |= R300_RB3D_Z_WRITE_ONLY;
+			t1 |= R300_Z_ENABLE | R300_Z_WRITE_ENABLE;
 			t2 |=
-			    (R300_ZS_ALWAYS << R300_RB3D_ZS1_DEPTH_FUNC_SHIFT);
-		} else {
-			t1 |= R300_RB3D_Z_DISABLED_1;	// disable
+			    (R300_ZS_ALWAYS << R300_Z_FUNC_SHIFT);
 		}
 
 		if (flags & CLEARBUFFER_STENCIL) {
-			t1 |= R300_RB3D_STENCIL_ENABLE;
+			t1 |= R300_STENCIL_ENABLE;
 			t2 |=
 			    (R300_ZS_ALWAYS <<
-			     R300_RB3D_ZS1_FRONT_FUNC_SHIFT) |
+			     R300_S_FRONT_FUNC_SHIFT) |
 			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_FRONT_FAIL_OP_SHIFT) |
+			     R300_S_FRONT_SFAIL_OP_SHIFT) |
 			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT) |
+			     R300_S_FRONT_ZPASS_OP_SHIFT) |
 			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_FRONT_ZFAIL_OP_SHIFT) |
-			    (R300_ZS_ALWAYS <<
-			     R300_RB3D_ZS1_BACK_FUNC_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT);
+			     R300_S_FRONT_ZFAIL_OP_SHIFT);
 		}
 
 		e32(t1);
 		e32(t2);
-		e32(r300->state.stencil.clear);
+		e32(((ctx->Stencil.WriteMask[0] & R300_STENCILREF_MASK) << R300_STENCILWRITEMASK_SHIFT) |
+		    (ctx->Stencil.Clear & R300_STENCILREF_MASK));
 	}
 
 	cmd2 = (drm_r300_cmd_header_t *) r300AllocCmdBuf(r300, 9, __FUNCTION__);
@@ -186,10 +178,16 @@ static void r300EmitClearState(GLcontext * ctx)
 	int cmd_written = 0;
 	drm_radeon_cmd_header_t *cmd = NULL;
 	int has_tcl = 1;
+	int is_r500 = 0;
+	GLuint vap_cntl;
 
 	if (!(r300->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
 		has_tcl = 0;
 
+        if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+                is_r500 = 1;
+
+
 	/* FIXME: the values written to R300_VAP_INPUT_ROUTE_0_0 and
 	 * R300_VAP_INPUT_ROUTE_0_1 are in fact known, however, the values are
 	 * quite complex; see the functions in r300_emit.c.
@@ -199,25 +197,38 @@ static void r300EmitClearState(GLcontext * ctx)
 	 * these registers, as well as the actual values used for rendering.
 	 */
 	R300_STATECHANGE(r300, vir[0]);
-	reg_start(R300_VAP_INPUT_ROUTE_0_0, 0);
+	reg_start(R300_VAP_PROG_STREAM_CNTL_0, 0);
 	if (!has_tcl)
-		e32(0x22030003);
+	    e32(((((0 << R300_DST_VEC_LOC_SHIFT) | R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_0_SHIFT) |
+		 ((R300_LAST_VEC | (2 << R300_DST_VEC_LOC_SHIFT) | R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_1_SHIFT)));
 	else
-		e32(0x21030003);
+	    e32(((((0 << R300_DST_VEC_LOC_SHIFT) | R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_0_SHIFT) |
+		 ((R300_LAST_VEC | (1 << R300_DST_VEC_LOC_SHIFT) | R300_DATA_TYPE_FLOAT_4) << R300_DATA_TYPE_1_SHIFT)));
 
 	/* disable fog */
 	R300_STATECHANGE(r300, fogs);
-	reg_start(R300_RE_FOG_STATE, 0);
+	reg_start(R300_FG_FOG_BLEND, 0);
 	e32(0x0);
 
 	R300_STATECHANGE(r300, vir[1]);
-	reg_start(R300_VAP_INPUT_ROUTE_1_0, 0);
-	e32(0xF688F688);
+	reg_start(R300_VAP_PROG_STREAM_CNTL_EXT_0, 0);
+	e32(((((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) |
+	       (R300_SWIZZLE_SELECT_Y << R300_SWIZZLE_SELECT_Y_SHIFT) |
+	       (R300_SWIZZLE_SELECT_Z << R300_SWIZZLE_SELECT_Z_SHIFT) |
+	       (R300_SWIZZLE_SELECT_W << R300_SWIZZLE_SELECT_W_SHIFT) |
+	       ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y | R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT))
+	      << R300_SWIZZLE0_SHIFT) |
+	     (((R300_SWIZZLE_SELECT_X << R300_SWIZZLE_SELECT_X_SHIFT) |
+	       (R300_SWIZZLE_SELECT_Y << R300_SWIZZLE_SELECT_Y_SHIFT) |
+	       (R300_SWIZZLE_SELECT_Z << R300_SWIZZLE_SELECT_Z_SHIFT) |
+	       (R300_SWIZZLE_SELECT_W << R300_SWIZZLE_SELECT_W_SHIFT) |
+	       ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y | R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT))
+	      << R300_SWIZZLE1_SHIFT)));
 
 	/* R300_VAP_INPUT_CNTL_0, R300_VAP_INPUT_CNTL_1 */
 	R300_STATECHANGE(r300, vic);
-	reg_start(R300_VAP_INPUT_CNTL_0, 1);
-	e32(R300_INPUT_CNTL_0_COLOR);
+	reg_start(R300_VAP_VTX_STATE_CNTL, 1);
+	e32((R300_SEL_USER_COLOR_0 << R300_COLOR_0_ASSEMBLY_SHIFT));
 	e32(R300_INPUT_CNTL_POS | R300_INPUT_CNTL_COLOR | R300_INPUT_CNTL_TC0);
 
 	R300_STATECHANGE(r300, vte);
@@ -229,13 +240,13 @@ static void r300EmitClearState(GLcontext * ctx)
 	    R300_VPORT_Z_OFFSET_ENA);
 	e32(0x8);
 
-	reg_start(0x21dc, 0);
+	reg_start(R300_VAP_PSC_SGN_NORM_CNTL, 0);
 	e32(0xaaaaaaaa);
 
 	R300_STATECHANGE(r300, vof);
 	reg_start(R300_VAP_OUTPUT_VTX_FMT_0, 1);
 	e32(R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT |
-	    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT);
+	    R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT);
 	e32(0x0);		/* no textures */
 
 	R300_STATECHANGE(r300, txe);
@@ -252,7 +263,7 @@ static void r300EmitClearState(GLcontext * ctx)
 	efloat(0.0);
 
 	R300_STATECHANGE(r300, at);
-	reg_start(R300_PP_ALPHA_TEST, 0);
+	reg_start(R300_FG_ALPHA_FUNC, 0);
 	e32(0x0);
 
 	R300_STATECHANGE(r300, bld);
@@ -260,78 +271,192 @@ static void r300EmitClearState(GLcontext * ctx)
 	e32(0x0);
 	e32(0x0);
 
-	R300_STATECHANGE(r300, unk221C);
-	reg_start(R300_VAP_UNKNOWN_221C, 0);
-	e32(R300_221C_CLEAR);
+	if (has_tcl) {
+	    R300_STATECHANGE(r300, vap_clip_cntl);
+	    reg_start(R300_VAP_CLIP_CNTL, 0);
+	    e32(R300_PS_UCP_MODE_CLIP_AS_TRIFAN | R300_CLIP_DISABLE);
+        }
 
 	R300_STATECHANGE(r300, ps);
-	reg_start(R300_RE_POINTSIZE, 0);
+	reg_start(R300_GA_POINT_SIZE, 0);
 	e32(((dPriv->w * 6) << R300_POINTSIZE_X_SHIFT) |
 	    ((dPriv->h * 6) << R300_POINTSIZE_Y_SHIFT));
 
-	R300_STATECHANGE(r300, ri);
-	reg_start(R300_RS_INTERP_0, 8);
-	for (i = 0; i < 8; ++i) {
-		e32(R300_RS_INTERP_USED);
+	if (!is_r500) {
+		R300_STATECHANGE(r300, ri);
+		reg_start(R300_RS_IP_0, 7);
+		for (i = 0; i < 8; ++i) {
+			e32(R300_RS_SEL_T(1) | R300_RS_SEL_R(2) | R300_RS_SEL_Q(3));
+		}
+
+		R300_STATECHANGE(r300, rc);
+		/* The second constant is needed to get glxgears display anything .. */
+		reg_start(R300_RS_COUNT, 1);
+		e32((1 << R300_IC_COUNT_SHIFT) | R300_HIRES_EN);
+		e32(0x0);
+
+		R300_STATECHANGE(r300, rr);
+		reg_start(R300_RS_INST_0, 0);
+		e32(R300_RS_INST_COL_CN_WRITE);
+	} else {
+		R300_STATECHANGE(r300, ri);
+		reg_start(R500_RS_IP_0, 7);
+		for (i = 0; i < 8; ++i) {
+			e32((R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_S_SHIFT) |
+			    (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_T_SHIFT) |
+			    (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_R_SHIFT) |
+			    (R500_RS_IP_PTR_K1 << R500_RS_IP_TEX_PTR_Q_SHIFT));
+		}
+
+		R300_STATECHANGE(r300, rc);
+		/* The second constant is needed to get glxgears display anything .. */
+		reg_start(R300_RS_COUNT, 1);
+		e32((1 << R300_IC_COUNT_SHIFT) | R300_HIRES_EN);
+		e32(0x0);
+
+		R300_STATECHANGE(r300, rr);
+		reg_start(R500_RS_INST_0, 0);
+		e32(R500_RS_INST_COL_CN_WRITE);
+
 	}
 
-	R300_STATECHANGE(r300, rc);
-	/* The second constant is needed to get glxgears display anything .. */
-	reg_start(R300_RS_CNTL_0, 1);
-	e32((1 << R300_RS_CNTL_CI_CNT_SHIFT) | R300_RS_CNTL_0_UNKNOWN_18);
-	e32(0x0);
+	if (!is_r500) {
+		R300_STATECHANGE(r300, fp);
+		reg_start(R300_US_CONFIG, 2);
+		e32(0x0);
+		e32(0x0);
+		e32(0x0);
+		reg_start(R300_US_CODE_ADDR_0, 3);
+		e32(0x0);
+		e32(0x0);
+		e32(0x0);
+		e32(R300_RGBA_OUT);
 
-	R300_STATECHANGE(r300, rr);
-	reg_start(R300_RS_ROUTE_0, 0);
-	e32(R300_RS_ROUTE_0_COLOR);
+		R300_STATECHANGE(r300, fpi[0]);
+		R300_STATECHANGE(r300, fpi[1]);
+		R300_STATECHANGE(r300, fpi[2]);
+		R300_STATECHANGE(r300, fpi[3]);
 
-	R300_STATECHANGE(r300, fp);
-	reg_start(R300_PFS_CNTL_0, 2);
-	e32(0x0);
-	e32(0x0);
-	e32(0x0);
-	reg_start(R300_PFS_NODE_0, 3);
-	e32(0x0);
-	e32(0x0);
-	e32(0x0);
-	e32(R300_PFS_NODE_OUTPUT_COLOR);
+		reg_start(R300_US_ALU_RGB_INST_0, 0);
+		e32(FP_INSTRC(MAD, FP_ARGC(SRC0C_XYZ), FP_ARGC(ONE), FP_ARGC(ZERO)));
 
-	R300_STATECHANGE(r300, fpi[0]);
-	R300_STATECHANGE(r300, fpi[1]);
-	R300_STATECHANGE(r300, fpi[2]);
-	R300_STATECHANGE(r300, fpi[3]);
+		reg_start(R300_US_ALU_RGB_ADDR_0, 0);
+		e32(FP_SELC(0, NO, XYZ, FP_TMP(0), 0, 0));
 
-	reg_start(R300_PFS_INSTR0_0, 0);
-	e32(FP_INSTRC(MAD, FP_ARGC(SRC0C_XYZ), FP_ARGC(ONE), FP_ARGC(ZERO)));
+		reg_start(R300_US_ALU_ALPHA_INST_0, 0);
+		e32(FP_INSTRA(MAD, FP_ARGA(SRC0A), FP_ARGA(ONE), FP_ARGA(ZERO)));
 
-	reg_start(R300_PFS_INSTR1_0, 0);
-	e32(FP_SELC(0, NO, XYZ, FP_TMP(0), 0, 0));
+		reg_start(R300_US_ALU_ALPHA_ADDR_0, 0);
+		e32(FP_SELA(0, NO, W, FP_TMP(0), 0, 0));
+	} else {
+ 		R300_STATECHANGE(r300, fp);
+ 		reg_start(R500_US_CONFIG, 1);
+ 		e32(R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO);
+ 		e32(0x0);
+ 		reg_start(R500_US_CODE_ADDR, 2);
+ 		e32(R500_US_CODE_START_ADDR(0) | R500_US_CODE_END_ADDR(1));
+ 		e32(R500_US_CODE_RANGE_ADDR(0) | R500_US_CODE_RANGE_SIZE(1));
+ 		e32(R500_US_CODE_OFFSET_ADDR(0));
+
+		R300_STATECHANGE(r300, r500fp);
+		r500fp_start_fragment(0, 6);
+
+		e32(R500_INST_TYPE_OUT |
+		    R500_INST_TEX_SEM_WAIT |
+		    R500_INST_LAST |
+		    R500_INST_RGB_OMASK_R |
+		    R500_INST_RGB_OMASK_G |
+		    R500_INST_RGB_OMASK_B |
+		    R500_INST_ALPHA_OMASK |
+		    R500_INST_RGB_CLAMP |
+		    R500_INST_ALPHA_CLAMP);
+
+		e32(R500_RGB_ADDR0(0) |
+		    R500_RGB_ADDR1(0) |
+		    R500_RGB_ADDR1_CONST |
+		    R500_RGB_ADDR2(0) |
+		    R500_RGB_ADDR2_CONST);
+
+		e32(R500_ALPHA_ADDR0(0) |
+		    R500_ALPHA_ADDR1(0) |
+		    R500_ALPHA_ADDR1_CONST |
+		    R500_ALPHA_ADDR2(0) |
+		    R500_ALPHA_ADDR2_CONST);
+
+		e32(R500_ALU_RGB_SEL_A_SRC0 |
+		    R500_ALU_RGB_R_SWIZ_A_R |
+		    R500_ALU_RGB_G_SWIZ_A_G |
+		    R500_ALU_RGB_B_SWIZ_A_B |
+		    R500_ALU_RGB_SEL_B_SRC0 |
+		    R500_ALU_RGB_R_SWIZ_B_R |
+		    R500_ALU_RGB_B_SWIZ_B_G |
+		    R500_ALU_RGB_G_SWIZ_B_B);
+
+		e32(R500_ALPHA_OP_CMP |
+		    R500_ALPHA_SWIZ_A_A |
+		    R500_ALPHA_SWIZ_B_A);
+
+		e32(R500_ALU_RGBA_OP_CMP |
+		    R500_ALU_RGBA_R_SWIZ_0 |
+		    R500_ALU_RGBA_G_SWIZ_0 |
+		    R500_ALU_RGBA_B_SWIZ_0 |
+		    R500_ALU_RGBA_A_SWIZ_0);
+	}
 
-	reg_start(R300_PFS_INSTR2_0, 0);
-	e32(FP_INSTRA(MAD, FP_ARGA(SRC0A), FP_ARGA(ONE), FP_ARGA(ZERO)));
+	reg_start(R300_VAP_PVS_STATE_FLUSH_REG, 0);
+	e32(0x00000000);
+	if (has_tcl) {
+	    vap_cntl = ((10 << R300_PVS_NUM_SLOTS_SHIFT) |
+			(5 << R300_PVS_NUM_CNTLRS_SHIFT) |
+			(12 << R300_VF_MAX_VTX_NUM_SHIFT));
+	    if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+		vap_cntl |= R500_TCL_STATE_OPTIMIZATION;
+	} else
+	    vap_cntl = ((10 << R300_PVS_NUM_SLOTS_SHIFT) |
+			(5 << R300_PVS_NUM_CNTLRS_SHIFT) |
+			(5 << R300_VF_MAX_VTX_NUM_SHIFT));
+
+	if (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV515)
+	    vap_cntl |= (2 << R300_PVS_NUM_FPUS_SHIFT);
+	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV530) ||
+		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV560) ||
+		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV570))
+	    vap_cntl |= (5 << R300_PVS_NUM_FPUS_SHIFT);
+	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV410) ||
+		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R420))
+	    vap_cntl |= (6 << R300_PVS_NUM_FPUS_SHIFT);
+	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R520) ||
+		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R580))
+	    vap_cntl |= (8 << R300_PVS_NUM_FPUS_SHIFT);
+	else
+	    vap_cntl |= (4 << R300_PVS_NUM_FPUS_SHIFT);
 
-	reg_start(R300_PFS_INSTR3_0, 0);
-	e32(FP_SELA(0, NO, W, FP_TMP(0), 0, 0));
+	R300_STATECHANGE(rmesa, vap_cntl);
+	reg_start(R300_VAP_CNTL, 0);
+	e32(vap_cntl);
 
 	if (has_tcl) {
 		R300_STATECHANGE(r300, pvs);
-		reg_start(R300_VAP_PVS_CNTL_1, 2);
-		e32((0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-		    (0 << R300_PVS_CNTL_1_POS_END_SHIFT) |
-		    (1 << R300_PVS_CNTL_1_PROGRAM_END_SHIFT));
-		e32(0x0);
-		e32(1 << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT);
+		reg_start(R300_VAP_PVS_CODE_CNTL_0, 2);
+
+		e32((0 << R300_PVS_FIRST_INST_SHIFT) |
+		    (0 << R300_PVS_XYZW_VALID_INST_SHIFT) |
+		    (1 << R300_PVS_LAST_INST_SHIFT));
+		e32((0 << R300_PVS_CONST_BASE_OFFSET_SHIFT) |
+		    (0 << R300_PVS_MAX_CONST_ADDR_SHIFT));
+		e32(1 << R300_PVS_LAST_VTX_SRC_INST_SHIFT);
 
 		R300_STATECHANGE(r300, vpi);
 		vsf_start_fragment(0x0, 8);
-		e32(VP_OUT(ADD, OUT, 0, XYZW));
-		e32(VP_IN(IN, 0));
-		e32(VP_ZERO());
+
+		e32(PVS_OP_DST_OPERAND(VE_ADD, GL_FALSE, GL_FALSE, 0, 0xf, PVS_DST_REG_OUT));
+		e32(PVS_SRC_OPERAND(0, PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y, PVS_SRC_SELECT_Z, PVS_SRC_SELECT_W, PVS_SRC_REG_INPUT, VSF_FLAG_NONE));
+		e32(PVS_SRC_OPERAND(0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_REG_INPUT, VSF_FLAG_NONE));
 		e32(0x0);
 
-		e32(VP_OUT(ADD, OUT, 1, XYZW));
-		e32(VP_IN(IN, 1));
-		e32(VP_ZERO());
+		e32(PVS_OP_DST_OPERAND(VE_ADD, GL_FALSE, GL_FALSE, 1, 0xf, PVS_DST_REG_OUT));
+		e32(PVS_SRC_OPERAND(1, PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y, PVS_SRC_SELECT_Z, PVS_SRC_SELECT_W, PVS_SRC_REG_INPUT, VSF_FLAG_NONE));
+		e32(PVS_SRC_OPERAND(1, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_SELECT_FORCE_0, PVS_SRC_REG_INPUT, VSF_FLAG_NONE));
 		e32(0x0);
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/r300_program.h b/src/mesa/drivers/dri/r300/r300_program.h
deleted file mode 100644
index eddd783f073..00000000000
--- a/src/mesa/drivers/dri/r300/r300_program.h
+++ /dev/null
@@ -1,150 +0,0 @@
-/*
-Copyright (C) 2004 Nicolai Haehnle.  All Rights Reserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Nicolai Haehnle <prefect_@gmx.net>
- */
-
-#ifndef __R300_PROGRAM_H__
-#define __R300_PROGRAM_H__
-
-#include "r300_reg.h"
-
-/**
- * Vertex program helper macros
- */
-
-/* Produce out dword */
-#define VP_OUTCLASS_TMP		R300_VPI_OUT_REG_CLASS_TEMPORARY
-#define VP_OUTCLASS_OUT		R300_VPI_OUT_REG_CLASS_RESULT
-
-#define VP_OUTMASK_X	R300_VPI_OUT_WRITE_X
-#define VP_OUTMASK_Y	R300_VPI_OUT_WRITE_Y
-#define VP_OUTMASK_Z	R300_VPI_OUT_WRITE_Z
-#define VP_OUTMASK_W	R300_VPI_OUT_WRITE_W
-#define VP_OUTMASK_XY	(VP_OUTMASK_X|VP_OUTMASK_Y)
-#define VP_OUTMASK_XZ	(VP_OUTMASK_X|VP_OUTMASK_Z)
-#define VP_OUTMASK_XW	(VP_OUTMASK_X|VP_OUTMASK_W)
-#define VP_OUTMASK_XYZ	(VP_OUTMASK_XY|VP_OUTMASK_Z)
-#define VP_OUTMASK_XYW	(VP_OUTMASK_XY|VP_OUTMASK_W)
-#define VP_OUTMASK_XZW	(VP_OUTMASK_XZ|VP_OUTMASK_W)
-#define VP_OUTMASK_XYZW	(VP_OUTMASK_XYZ|VP_OUTMASK_W)
-#define VP_OUTMASK_YZ	(VP_OUTMASK_Y|VP_OUTMASK_Z)
-#define VP_OUTMASK_YW	(VP_OUTMASK_Y|VP_OUTMASK_W)
-#define VP_OUTMASK_YZW	(VP_OUTMASK_YZ|VP_OUTMASK_W)
-#define VP_OUTMASK_ZW	(VP_OUTMASK_Z|VP_OUTMASK_W)
-
-#define VP_OUT(instr,outclass,outidx,outmask) \
-	(R300_VPI_OUT_OP_##instr |				\
-	((outidx) << R300_VPI_OUT_REG_INDEX_SHIFT) |		\
-	VP_OUTCLASS_##outclass |				\
-	VP_OUTMASK_##outmask)
-
-/* Produce in dword */
-#define VP_INCLASS_TMP		R300_VPI_IN_REG_CLASS_TEMPORARY
-#define VP_INCLASS_IN		R300_VPI_IN_REG_CLASS_ATTRIBUTE
-#define VP_INCLASS_CONST	R300_VPI_IN_REG_CLASS_PARAMETER
-
-#define VP_IN(class,idx) \
-	(((idx) << R300_VPI_IN_REG_INDEX_SHIFT) |		\
-	VP_INCLASS_##class |					\
-	(R300_VPI_IN_SELECT_X << R300_VPI_IN_X_SHIFT) |		\
-	(R300_VPI_IN_SELECT_Y << R300_VPI_IN_Y_SHIFT) |		\
-	(R300_VPI_IN_SELECT_Z << R300_VPI_IN_Z_SHIFT) |		\
-	(R300_VPI_IN_SELECT_W << R300_VPI_IN_W_SHIFT))
-#define VP_ZERO() \
-	((R300_VPI_IN_SELECT_ZERO << R300_VPI_IN_X_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ZERO << R300_VPI_IN_Y_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ZERO << R300_VPI_IN_Z_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ZERO << R300_VPI_IN_W_SHIFT))
-#define VP_ONE() \
-	((R300_VPI_IN_SELECT_ONE << R300_VPI_IN_X_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ONE << R300_VPI_IN_Y_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ONE << R300_VPI_IN_Z_SHIFT) |	\
-	(R300_VPI_IN_SELECT_ONE << R300_VPI_IN_W_SHIFT))
-
-#define VP_NEG(in,comp)		((in) ^ (R300_VPI_IN_NEG_##comp))
-#define VP_NEGALL(in,comp)	VP_NEG(VP_NEG(VP_NEG(VP_NEG((in),X),Y),Z),W)
-
-/**
- * Fragment program helper macros
- */
-
-/* Produce unshifted source selectors */
-#define FP_TMP(idx) (idx)
-#define FP_CONST(idx) ((idx) | (1 << 5))
-
-/* Produce source/dest selector dword */
-#define FP_SELC_MASK_NO		0
-#define FP_SELC_MASK_X		1
-#define FP_SELC_MASK_Y		2
-#define FP_SELC_MASK_XY		3
-#define FP_SELC_MASK_Z		4
-#define FP_SELC_MASK_XZ		5
-#define FP_SELC_MASK_YZ		6
-#define FP_SELC_MASK_XYZ	7
-
-#define FP_SELC(destidx,regmask,outmask,src0,src1,src2) \
-	(((destidx) << R300_FPI1_DSTC_SHIFT) |		\
-	 (FP_SELC_MASK_##regmask << 23) |		\
-	 (FP_SELC_MASK_##outmask << 26) |		\
-	 ((src0) << R300_FPI1_SRC0C_SHIFT) |		\
-	 ((src1) << R300_FPI1_SRC1C_SHIFT) |		\
-	 ((src2) << R300_FPI1_SRC2C_SHIFT))
-
-#define FP_SELA_MASK_NO		0
-#define FP_SELA_MASK_W		1
-
-#define FP_SELA(destidx,regmask,outmask,src0,src1,src2) \
-	(((destidx) << R300_FPI3_DSTA_SHIFT) |		\
-	 (FP_SELA_MASK_##regmask << 23) |		\
-	 (FP_SELA_MASK_##outmask << 24) |		\
-	 ((src0) << R300_FPI3_SRC0A_SHIFT) |		\
-	 ((src1) << R300_FPI3_SRC1A_SHIFT) |		\
-	 ((src2) << R300_FPI3_SRC2A_SHIFT))
-
-/* Produce unshifted argument selectors */
-#define FP_ARGC(source)	R300_FPI0_ARGC_##source
-#define FP_ARGA(source) R300_FPI2_ARGA_##source
-#define FP_ABS(arg) ((arg) | (1 << 6))
-#define FP_NEG(arg) ((arg) ^ (1 << 5))
-
-/* Produce instruction dword */
-#define FP_INSTRC(opcode,arg0,arg1,arg2) \
-	(R300_FPI0_OUTC_##opcode | 		\
-	((arg0) << R300_FPI0_ARG0C_SHIFT) |	\
-	((arg1) << R300_FPI0_ARG1C_SHIFT) |	\
-	((arg2) << R300_FPI0_ARG2C_SHIFT))
-
-#define FP_INSTRA(opcode,arg0,arg1,arg2) \
-	(R300_FPI2_OUTA_##opcode | 		\
-	((arg0) << R300_FPI2_ARG0A_SHIFT) |	\
-	((arg1) << R300_FPI2_ARG1A_SHIFT) |	\
-	((arg2) << R300_FPI2_ARG2A_SHIFT))
-
-extern void debug_vp(GLcontext * ctx, struct gl_vertex_program *vp);
-
-#endif				/* __R300_PROGRAM_H__ */
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index 1baa74c5269..778db96cc1f 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -67,9 +67,15 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /*
  * Vertex Array Processing (VAP) Control
- * Stolen from r200 code from Christoph Brill (It's a guess!)
  */
 #define R300_VAP_CNTL	0x2080
+#       define R300_PVS_NUM_SLOTS_SHIFT                 0
+#       define R300_PVS_NUM_CNTLRS_SHIFT                4
+#       define R300_PVS_NUM_FPUS_SHIFT                  8
+#       define R300_VF_MAX_VTX_NUM_SHIFT                18
+#       define R300_GL_CLIP_SPACE_DEF                   (0 << 22)
+#       define R300_DX_CLIP_SPACE_DEF                   (1 << 22)
+#       define R500_TCL_STATE_OPTIMIZATION              (1 << 23)
 
 /* This register is written directly and also starts data section
  * in many 3d CP_PACKET3's
@@ -106,14 +112,15 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 	/* number of vertices */
 #	define	R300_VAP_VF_CNTL__NUM_VERTICES__SHIFT           16
 
-/* BEGIN: Wild guesses */
+#define R500_VAP_INDEX_OFFSET		    0x208c
+
 #define R300_VAP_OUTPUT_VTX_FMT_0           0x2090
 #       define R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT     (1<<0)
-#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT   (1<<1)
-#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT (1<<2)  /* GUESS */
-#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT (1<<3)  /* GUESS */
-#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT (1<<4)  /* GUESS */
-#       define R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT (1<<16) /* GUESS */
+#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT (1<<1)
+#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_1_PRESENT (1<<2)
+#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_2_PRESENT (1<<3)
+#       define R300_VAP_OUTPUT_VTX_FMT_0__COLOR_3_PRESENT (1<<4)
+#       define R300_VAP_OUTPUT_VTX_FMT_0__PT_SIZE_PRESENT (1<<16)
 
 #define R300_VAP_OUTPUT_VTX_FMT_1           0x2094
 	/* each of the following is 3 bits wide, specifies number
@@ -126,30 +133,64 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_5_COMP_CNT_SHIFT 15
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_6_COMP_CNT_SHIFT 18
 #       define R300_VAP_OUTPUT_VTX_FMT_1__TEX_7_COMP_CNT_SHIFT 21
-/* END: Wild guesses */
+#	define R300_VAP_OUTPUT_VTX_FMT_1__NOT_PRESENT  0
+#	define R300_VAP_OUTPUT_VTX_FMT_1__1_COMPONENT  1
+#	define R300_VAP_OUTPUT_VTX_FMT_1__2_COMPONENTS 2
+#	define R300_VAP_OUTPUT_VTX_FMT_1__3_COMPONENTS 3
+#	define R300_VAP_OUTPUT_VTX_FMT_1__4_COMPONENTS 4
 
 #define R300_SE_VTE_CNTL                  0x20b0
-#	define     R300_VPORT_X_SCALE_ENA                0x00000001
-#	define     R300_VPORT_X_OFFSET_ENA               0x00000002
-#	define     R300_VPORT_Y_SCALE_ENA                0x00000004
-#	define     R300_VPORT_Y_OFFSET_ENA               0x00000008
-#	define     R300_VPORT_Z_SCALE_ENA                0x00000010
-#	define     R300_VPORT_Z_OFFSET_ENA               0x00000020
-#	define     R300_VTX_XY_FMT                       0x00000100
-#	define     R300_VTX_Z_FMT                        0x00000200
-#	define     R300_VTX_W0_FMT                       0x00000400
-#	define     R300_VTX_W0_NORMALIZE                 0x00000800
-#	define     R300_VTX_ST_DENORMALIZED              0x00001000
+#	define     R300_VPORT_X_SCALE_ENA                (1 << 0)
+#	define     R300_VPORT_X_OFFSET_ENA               (1 << 1)
+#	define     R300_VPORT_Y_SCALE_ENA                (1 << 2)
+#	define     R300_VPORT_Y_OFFSET_ENA               (1 << 3)
+#	define     R300_VPORT_Z_SCALE_ENA                (1 << 4)
+#	define     R300_VPORT_Z_OFFSET_ENA               (1 << 5)
+#	define     R300_VTX_XY_FMT                       (1 << 8)
+#	define     R300_VTX_Z_FMT                        (1 << 9)
+#	define     R300_VTX_W0_FMT                       (1 << 10)
+#	define     R300_SERIAL_PROC_ENA                  (1 << 11)
 
 /* BEGIN: Vertex data assembly - lots of uncertainties */
 
 /* gap */
 
+/* Maximum Vertex Indx Clamp */
+#define R300_VAP_VF_MAX_VTX_INDX         0x2134
+/* Minimum Vertex Indx Clamp */
+#define R300_VAP_VF_MIN_VTX_INDX         0x2138
+
+/** Vertex assembler/processor control status */
 #define R300_VAP_CNTL_STATUS              0x2140
+/* No swap at all (default) */
 #	define R300_VC_NO_SWAP                  (0 << 0)
+/* 16-bit swap: 0xAABBCCDD becomes 0xBBAADDCC */
 #	define R300_VC_16BIT_SWAP               (1 << 0)
+/* 32-bit swap: 0xAABBCCDD becomes 0xDDCCBBAA */
 #	define R300_VC_32BIT_SWAP               (2 << 0)
+/* Half-dword swap: 0xAABBCCDD becomes 0xCCDDAABB */
+#	define R300_VC_HALF_DWORD_SWAP          (3 << 0)
+/* The TCL engine will not be used (as it is logically or even physically removed) */
 #	define R300_VAP_TCL_BYPASS		(1 << 8)
+/* Read only flag if TCL engine is busy. */
+#	define R300_VAP_PVS_BUSY                (1 << 11)
+/* TODO: gap for MAX_MPS */
+/* Read only flag if the vertex store is busy. */
+#	define R300_VAP_VS_BUSY                 (1 << 24)
+/* Read only flag if the reciprocal engine is busy. */
+#	define R300_VAP_RCP_BUSY                (1 << 25)
+/* Read only flag if the viewport transform engine is busy. */
+#	define R300_VAP_VTE_BUSY                (1 << 26)
+/* Read only flag if the memory interface unit is busy. */
+#	define R300_VAP_MUI_BUSY                (1 << 27)
+/* Read only flag if the vertex cache is busy. */
+#	define R300_VAP_VC_BUSY                 (1 << 28)
+/* Read only flag if the vertex fetcher is busy. */
+#	define R300_VAP_VF_BUSY                 (1 << 29)
+/* Read only flag if the register pipeline is busy. */
+#	define R300_VAP_REGPIPE_BUSY            (1 << 30)
+/* Read only flag if the VAP engine is busy. */
+#	define R300_VAP_VAP_BUSY                (1 << 31)
 
 /* gap */
 
@@ -177,27 +218,31 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Always set COMPONENTS_4 in immediate mode.
  */
 
-#define R300_VAP_INPUT_ROUTE_0_0            0x2150
-#       define R300_INPUT_ROUTE_COMPONENTS_1     (0 << 0)
-#       define R300_INPUT_ROUTE_COMPONENTS_2     (1 << 0)
-#       define R300_INPUT_ROUTE_COMPONENTS_3     (2 << 0)
-#       define R300_INPUT_ROUTE_COMPONENTS_4     (3 << 0)
-#       define R300_INPUT_ROUTE_COMPONENTS_RGBA  (4 << 0) /* GUESS */
-#       define R300_VAP_INPUT_ROUTE_IDX_SHIFT    8
-#       define R300_VAP_INPUT_ROUTE_IDX_MASK     (31 << 8) /* GUESS */
-#       define R300_VAP_INPUT_ROUTE_END          (1 << 13)
-#       define R300_INPUT_ROUTE_IMMEDIATE_MODE   (0 << 14) /* GUESS */
-#       define R300_INPUT_ROUTE_FLOAT            (1 << 14) /* GUESS */
-#       define R300_INPUT_ROUTE_UNSIGNED_BYTE    (2 << 14) /* GUESS */
-#       define R300_INPUT_ROUTE_FLOAT_COLOR      (3 << 14) /* GUESS */
-#define R300_VAP_INPUT_ROUTE_0_1            0x2154
-#define R300_VAP_INPUT_ROUTE_0_2            0x2158
-#define R300_VAP_INPUT_ROUTE_0_3            0x215C
-#define R300_VAP_INPUT_ROUTE_0_4            0x2160
-#define R300_VAP_INPUT_ROUTE_0_5            0x2164
-#define R300_VAP_INPUT_ROUTE_0_6            0x2168
-#define R300_VAP_INPUT_ROUTE_0_7            0x216C
-
+#define R300_VAP_PROG_STREAM_CNTL_0                     0x2150
+#       define R300_DATA_TYPE_0_SHIFT                   0
+#       define R300_DATA_TYPE_FLOAT_1                   0
+#       define R300_DATA_TYPE_FLOAT_2                   1
+#       define R300_DATA_TYPE_FLOAT_3                   2
+#       define R300_DATA_TYPE_FLOAT_4                   3
+#       define R300_DATA_TYPE_BYTE                      4
+#       define R300_DATA_TYPE_D3DCOLOR                  5
+#       define R300_DATA_TYPE_SHORT_2                   6
+#       define R300_DATA_TYPE_SHORT_4                   7
+#       define R300_DATA_TYPE_VECTOR_3_TTT              8
+#       define R300_DATA_TYPE_VECTOR_3_EET              9
+#       define R300_SKIP_DWORDS_SHIFT                   4
+#       define R300_DST_VEC_LOC_SHIFT                   8
+#       define R300_LAST_VEC                            (1 << 13)
+#       define R300_SIGNED                              (1 << 14)
+#       define R300_NORMALIZE                           (1 << 15)
+#       define R300_DATA_TYPE_1_SHIFT                   16
+#define R300_VAP_PROG_STREAM_CNTL_1                     0x2154
+#define R300_VAP_PROG_STREAM_CNTL_2                     0x2158
+#define R300_VAP_PROG_STREAM_CNTL_3                     0x215C
+#define R300_VAP_PROG_STREAM_CNTL_4                     0x2160
+#define R300_VAP_PROG_STREAM_CNTL_5                     0x2164
+#define R300_VAP_PROG_STREAM_CNTL_6                     0x2168
+#define R300_VAP_PROG_STREAM_CNTL_7                     0x216C
 /* gap */
 
 /* Notes:
@@ -205,9 +250,26 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *    if vertex program uses only position, fglrx will set normal, too
  *  - INPUT_CNTL_0_COLOR and INPUT_CNTL_COLOR bits are always equal.
  */
-#define R300_VAP_INPUT_CNTL_0               0x2180
-#       define R300_INPUT_CNTL_0_COLOR           0x00000001
-#define R300_VAP_INPUT_CNTL_1               0x2184
+#define R300_VAP_VTX_STATE_CNTL               0x2180
+#       define R300_COLOR_0_ASSEMBLY_SHIFT    0
+#       define R300_SEL_COLOR                 0
+#       define R300_SEL_USER_COLOR_0          1
+#       define R300_SEL_USER_COLOR_1          2
+#       define R300_COLOR_1_ASSEMBLY_SHIFT    2
+#       define R300_COLOR_2_ASSEMBLY_SHIFT    4
+#       define R300_COLOR_3_ASSEMBLY_SHIFT    6
+#       define R300_COLOR_4_ASSEMBLY_SHIFT    8
+#       define R300_COLOR_5_ASSEMBLY_SHIFT    10
+#       define R300_COLOR_6_ASSEMBLY_SHIFT    12
+#       define R300_COLOR_7_ASSEMBLY_SHIFT    14
+#       define R300_UPDATE_USER_COLOR_0_ENA   (1 << 16)
+
+/*
+ * Each bit in this field applies to the corresponding vector in the VSM
+ * memory (i.e. Bit 0 applies to VECTOR_0 (POSITION), etc.). If the bit
+ * is set, then the corresponding 4-Dword Vector is output into the Vertex Stream.
+ */
+#define R300_VAP_VSM_VTX_ASSM               0x2184
 #       define R300_INPUT_CNTL_POS               0x00000001
 #       define R300_INPUT_CNTL_NORMAL            0x00000002
 #       define R300_INPUT_CNTL_COLOR             0x00000004
@@ -220,6 +282,12 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_INPUT_CNTL_TC6               0x00010000 /* GUESS */
 #       define R300_INPUT_CNTL_TC7               0x00020000 /* GUESS */
 
+/* Programmable Stream Control Signed Normalize Control */
+#define R300_VAP_PSC_SGN_NORM_CNTL         0x21dc
+#	define SGN_NORM_ZERO                 0
+#	define SGN_NORM_ZERO_CLAMP_MINUS_ONE 1
+#	define SGN_NORM_NO_ZERO              2
+
 /* gap */
 
 /* Words parallel to INPUT_ROUTE_0; All words that are active in INPUT_ROUTE_0
@@ -229,26 +297,40 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * mode, the swizzling pattern is e.g. used to set zw components in texture
  * coordinates with only tweo components.
  */
-#define R300_VAP_INPUT_ROUTE_1_0            0x21E0
+#define R300_VAP_PROG_STREAM_CNTL_EXT_0                 0x21e0
+#       define R300_SWIZZLE0_SHIFT                      0
+#       define R300_SWIZZLE_SELECT_X_SHIFT              0
+#       define R300_SWIZZLE_SELECT_Y_SHIFT              3
+#       define R300_SWIZZLE_SELECT_Z_SHIFT              6
+#       define R300_SWIZZLE_SELECT_W_SHIFT              9
+
+#       define R300_SWIZZLE_SELECT_X                    0
+#       define R300_SWIZZLE_SELECT_Y                    1
+#       define R300_SWIZZLE_SELECT_Z                    2
+#       define R300_SWIZZLE_SELECT_W                    3
+#       define R300_SWIZZLE_SELECT_FP_ZERO              4
+#       define R300_SWIZZLE_SELECT_FP_ONE               5
+/* alternate forms for r300_emit.c */
 #       define R300_INPUT_ROUTE_SELECT_X    0
 #       define R300_INPUT_ROUTE_SELECT_Y    1
 #       define R300_INPUT_ROUTE_SELECT_Z    2
 #       define R300_INPUT_ROUTE_SELECT_W    3
 #       define R300_INPUT_ROUTE_SELECT_ZERO 4
 #       define R300_INPUT_ROUTE_SELECT_ONE  5
-#       define R300_INPUT_ROUTE_SELECT_MASK 7
-#       define R300_INPUT_ROUTE_X_SHIFT     0
-#       define R300_INPUT_ROUTE_Y_SHIFT     3
-#       define R300_INPUT_ROUTE_Z_SHIFT     6
-#       define R300_INPUT_ROUTE_W_SHIFT     9
-#       define R300_INPUT_ROUTE_ENABLE      (15 << 12)
-#define R300_VAP_INPUT_ROUTE_1_1            0x21E4
-#define R300_VAP_INPUT_ROUTE_1_2            0x21E8
-#define R300_VAP_INPUT_ROUTE_1_3            0x21EC
-#define R300_VAP_INPUT_ROUTE_1_4            0x21F0
-#define R300_VAP_INPUT_ROUTE_1_5            0x21F4
-#define R300_VAP_INPUT_ROUTE_1_6            0x21F8
-#define R300_VAP_INPUT_ROUTE_1_7            0x21FC
+
+#       define R300_WRITE_ENA_SHIFT                     12
+#       define R300_WRITE_ENA_X                         1
+#       define R300_WRITE_ENA_Y                         2
+#       define R300_WRITE_ENA_Z                         4
+#       define R300_WRITE_ENA_W                         8
+#       define R300_SWIZZLE1_SHIFT                      16
+#define R300_VAP_PROG_STREAM_CNTL_EXT_1                 0x21e4
+#define R300_VAP_PROG_STREAM_CNTL_EXT_2                 0x21e8
+#define R300_VAP_PROG_STREAM_CNTL_EXT_3                 0x21ec
+#define R300_VAP_PROG_STREAM_CNTL_EXT_4                 0x21f0
+#define R300_VAP_PROG_STREAM_CNTL_EXT_5                 0x21f4
+#define R300_VAP_PROG_STREAM_CNTL_EXT_6                 0x21f8
+#define R300_VAP_PROG_STREAM_CNTL_EXT_7                 0x21fc
 
 /* END: Vertex data assembly */
 
@@ -280,18 +362,20 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Multiple vertex programs and parameter sets can be loaded at once,
  * which could explain the size discrepancy.
  */
-#define R300_VAP_PVS_UPLOAD_ADDRESS         0x2200
-#       define R300_PVS_UPLOAD_PROGRAM           0x00000000
-/* gap */
-#       define R300_PVS_UPLOAD_PARAMETERS        0x00000200
-/* gap */
-#       define R300_PVS_UPLOAD_CLIP_PLANE0       0x00000400
-#       define R300_PVS_UPLOAD_CLIP_PLANE1       0x00000401
-#       define R300_PVS_UPLOAD_CLIP_PLANE2       0x00000402
-#       define R300_PVS_UPLOAD_CLIP_PLANE3       0x00000403
-#       define R300_PVS_UPLOAD_CLIP_PLANE4       0x00000404
-#       define R300_PVS_UPLOAD_CLIP_PLANE5       0x00000405
-#       define R300_PVS_UPLOAD_POINTSIZE         0x00000406
+#define R300_VAP_PVS_VECTOR_INDX_REG         0x2200
+#       define R300_PVS_CODE_START           0
+#       define R300_MAX_PVS_CODE_LINES       256
+#       define R500_MAX_PVS_CODE_LINES       1024
+#       define R300_PVS_CONST_START          512
+#       define R500_PVS_CONST_START          1024
+#       define R300_MAX_PVS_CONST_VECS       256
+#       define R500_MAX_PVS_CONST_VECS       1024
+#       define R300_PVS_UCP_START            1024
+#       define R500_PVS_UCP_START            1536
+#       define R300_POINT_VPORT_SCALE_OFFSET 1030
+#       define R500_POINT_VPORT_SCALE_OFFSET 1542
+#       define R300_POINT_GEN_TEX_OFFSET     1031
+#       define R500_POINT_GEN_TEX_OFFSET     1543
 
 /*
  * These are obsolete defines form r300_context.h, but they might give some
@@ -319,10 +403,28 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 /* I do not know the purpose of this register. However, I do know that
  * it is set to 221C_CLEAR for clear operations and to 221C_NORMAL
  * for normal rendering.
+ *
+ * 2007-11-05: This register is the user clip plane control register, but there
+ * also seems to be a rendering mode control; the NORMAL/CLEAR defines.
+ *
+ * See bug #9871. http://bugs.freedesktop.org/attachment.cgi?id=10672&action=view
  */
-#define R300_VAP_UNKNOWN_221C               0x221C
-#       define R300_221C_NORMAL                  0x00000000
-#       define R300_221C_CLEAR                   0x0001C000
+#define R300_VAP_CLIP_CNTL                       0x221C
+#       define R300_VAP_UCP_ENABLE_0             (1 << 0)
+#       define R300_VAP_UCP_ENABLE_1             (1 << 1)
+#       define R300_VAP_UCP_ENABLE_2             (1 << 2)
+#       define R300_VAP_UCP_ENABLE_3             (1 << 3)
+#       define R300_VAP_UCP_ENABLE_4             (1 << 4)
+#       define R300_VAP_UCP_ENABLE_5             (1 << 5)
+#       define R300_PS_UCP_MODE_DIST_COP         (0 << 14)
+#       define R300_PS_UCP_MODE_RADIUS_COP       (1 << 14)
+#       define R300_PS_UCP_MODE_RADIUS_COP_CLIP  (2 << 14)
+#       define R300_PS_UCP_MODE_CLIP_AS_TRIFAN   (3 << 14)
+#       define R300_CLIP_DISABLE                 (1 << 16)
+#       define R300_UCP_CULL_ONLY_ENABLE         (1 << 17)
+#       define R300_BOUNDARY_EDGE_FLAG_ENABLE    (1 << 18)
+#       define R500_COLOR2_IS_TEXTURE            (1 << 20)
+#       define R500_COLOR3_IS_TEXTURE            (1 << 21)
 
 /* These seem to be per-pixel and per-vertex X and Y clipping planes. The first
  * plane is per-pixel and the second plane is per-vertex.
@@ -331,10 +433,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  * These registers are called X_QUAD0_1_FL to X_QUAD0_4_FL by glxtest.
  */
-#define R300_VAP_CLIP_X_0                   0x2220
-#define R300_VAP_CLIP_X_1                   0x2224
-#define R300_VAP_CLIP_Y_0                   0x2228
-#define R300_VAP_CLIP_Y_1                   0x2230
+#define R300_VAP_GB_VERT_CLIP_ADJ                   0x2220
+#define R300_VAP_GB_VERT_DISC_ADJ                   0x2224
+#define R300_VAP_GB_HORZ_CLIP_ADJ                   0x2228
+#define R300_VAP_GB_HORZ_DISC_ADJ                   0x222c
 
 /* gap */
 
@@ -343,10 +445,14 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Therefore, I suspect writing zero to 0x2284 synchronizes the engine and
  * avoids bugs caused by still running shaders reading bad data from memory.
  */
-#define R300_VAP_PVS_WAITIDLE               0x2284 /* GUESS */
+#define R300_VAP_PVS_STATE_FLUSH_REG        0x2284
 
-/* Absolutely no clue what this register is about. */
-#define R300_VAP_UNKNOWN_2288               0x2288
+/* This register is used to define the number of core clocks to wait for a
+ * vertex to be received by the VAP input controller (while the primitive
+ * path is backed up) before forcing any accumulated vertices to be submitted
+ * to the vertex processing path.
+ */
+#define VAP_PVS_VTX_TIMEOUT_REG             0x2288
 #       define R300_2288_R300                    0x00750000 /* -- nh */
 #       define R300_2288_RV350                   0x0000FFFF /* -- Vladimir */
 
@@ -369,17 +475,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * is sometimes accepted other instruction that have no relationship with
  * position calculations.
  */
-#define R300_VAP_PVS_CNTL_1                 0x22D0
-#       define R300_PVS_CNTL_1_PROGRAM_START_SHIFT   0
-#       define R300_PVS_CNTL_1_POS_END_SHIFT         10
-#       define R300_PVS_CNTL_1_PROGRAM_END_SHIFT     20
+#define R300_VAP_PVS_CODE_CNTL_0            0x22D0
+#       define R300_PVS_FIRST_INST_SHIFT         0
+#       define R300_PVS_XYZW_VALID_INST_SHIFT    10
+#       define R300_PVS_LAST_INST_SHIFT          20
 /* Addresses are relative the the vertex program parameters area. */
-#define R300_VAP_PVS_CNTL_2                 0x22D4
-#       define R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT 0
-#       define R300_PVS_CNTL_2_PARAM_COUNT_SHIFT  16
-#define R300_VAP_PVS_CNTL_3	           0x22D8
-#       define R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT 10
-#       define R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT 0
+#define R300_VAP_PVS_CONST_CNTL             0x22D4
+#       define R300_PVS_CONST_BASE_OFFSET_SHIFT  0
+#       define R300_PVS_MAX_CONST_ADDR_SHIFT     16
+#define R300_VAP_PVS_CODE_CNTL_1	    0x22D8
+#       define R300_PVS_LAST_VTX_SRC_INST_SHIFT  0
+#define R300_VAP_PVS_FLOW_CNTL_OPC          0x22DC
 
 /* The entire range from 0x2300 to 0x2AC inclusive seems to be used for
  * immediate vertices
@@ -427,17 +533,23 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * (or something closely related to that).
  * This bit is rather fatal at the time being due to lackings at pixel
  * shader side
+ * Specifies top of Raster pipe specific enable controls.
  */
 #define R300_GB_ENABLE	0x4008
-#	define R300_GB_POINT_STUFF_ENABLE	(1<<0)
-#	define R300_GB_LINE_STUFF_ENABLE	(1<<1)
-#	define R300_GB_TRIANGLE_STUFF_ENABLE	(1<<2)
-#	define R300_GB_STENCIL_AUTO_ENABLE	(1<<4)
-#	define R300_GB_UNK31			(1<<31)
+#	define R300_GB_POINT_STUFF_DISABLE     (0 << 0)
+#	define R300_GB_POINT_STUFF_ENABLE      (1 << 0) /* Specifies if points will have stuffed texture coordinates. */
+#	define R300_GB_LINE_STUFF_DISABLE      (0 << 1)
+#	define R300_GB_LINE_STUFF_ENABLE       (1 << 1) /* Specifies if lines will have stuffed texture coordinates. */
+#	define R300_GB_TRIANGLE_STUFF_DISABLE  (0 << 2)
+#	define R300_GB_TRIANGLE_STUFF_ENABLE   (1 << 2) /* Specifies if triangles will have stuffed texture coordinates. */
+#	define R300_GB_STENCIL_AUTO_DISABLE    (0 << 4)
+#	define R300_GB_STENCIL_AUTO_ENABLE     (1 << 4) /* Enable stencil auto inc/dec based on triangle cw/ccw, force into dzy low bit. */
+#	define R300_GB_STENCIL_AUTO_FORCE      (2 << 4) /* Force 0 into dzy low bit. */
+
 	/* each of the following is 2 bits wide */
-#define R300_GB_TEX_REPLICATE	0
-#define R300_GB_TEX_ST		1
-#define R300_GB_TEX_STR		2
+#define R300_GB_TEX_REPLICATE	0 /* Replicate VAP source texture coordinates (S,T,[R,Q]). */
+#define R300_GB_TEX_ST		1 /* Stuff with source texture coordinates (S,T). */
+#define R300_GB_TEX_STR		2 /* Stuff with source texture coordinates (S,T,R). */
 #	define R300_GB_TEX0_SOURCE_SHIFT	16
 #	define R300_GB_TEX1_SOURCE_SHIFT	18
 #	define R300_GB_TEX2_SOURCE_SHIFT	20
@@ -448,7 +560,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_GB_TEX7_SOURCE_SHIFT	30
 
 /* MSPOS - positions for multisample antialiasing (?) */
-#define R300_GB_MSPOS0	0x4010
+#define R300_GB_MSPOS0                           0x4010
 	/* shifts - each of the fields is 4 bits */
 #	define R300_GB_MSPOS0__MS_X0_SHIFT	0
 #	define R300_GB_MSPOS0__MS_Y0_SHIFT	4
@@ -459,7 +571,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_GB_MSPOS0__MSBD0_Y		24
 #	define R300_GB_MSPOS0__MSBD0_X		28
 
-#define R300_GB_MSPOS1	0x4014
+#define R300_GB_MSPOS1                           0x4014
 #	define R300_GB_MSPOS1__MS_X3_SHIFT	0
 #	define R300_GB_MSPOS1__MS_Y3_SHIFT	4
 #	define R300_GB_MSPOS1__MS_X4_SHIFT	8
@@ -468,31 +580,47 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_GB_MSPOS1__MS_Y5_SHIFT	20
 #	define R300_GB_MSPOS1__MSBD1		24
 
-
-#define R300_GB_TILE_CONFIG	0x4018
-#	define R300_GB_TILE_ENABLE	(1<<0)
-#	define R300_GB_TILE_PIPE_COUNT_RV300	0
-#	define R300_GB_TILE_PIPE_COUNT_R300	(3<<1)
-#	define R300_GB_TILE_PIPE_COUNT_R420	(7<<1)
-#	define R300_GB_TILE_PIPE_COUNT_RV410	(3<<1)
-#	define R300_GB_TILE_SIZE_8		0
-#	define R300_GB_TILE_SIZE_16		(1<<4)
-#	define R300_GB_TILE_SIZE_32		(2<<4)
-#	define R300_GB_SUPER_SIZE_1		(0<<6)
-#	define R300_GB_SUPER_SIZE_2		(1<<6)
-#	define R300_GB_SUPER_SIZE_4		(2<<6)
-#	define R300_GB_SUPER_SIZE_8		(3<<6)
-#	define R300_GB_SUPER_SIZE_16		(4<<6)
-#	define R300_GB_SUPER_SIZE_32		(5<<6)
-#	define R300_GB_SUPER_SIZE_64		(6<<6)
-#	define R300_GB_SUPER_SIZE_128		(7<<6)
+/* Specifies the graphics pipeline configuration for rasterization. */
+#define R300_GB_TILE_CONFIG                      0x4018
+#	define R300_GB_TILE_DISABLE             (0 << 0)
+#	define R300_GB_TILE_ENABLE              (1 << 0)
+#	define R300_GB_TILE_PIPE_COUNT_RV300	(0 << 1) /* RV350 (1 pipe, 1 ctx) */
+#	define R300_GB_TILE_PIPE_COUNT_R300	(3 << 1) /* R300 (2 pipes, 1 ctx) */
+#	define R300_GB_TILE_PIPE_COUNT_R420_3P  (6 << 1) /* R420-3P (3 pipes, 1 ctx) */
+#	define R300_GB_TILE_PIPE_COUNT_R420	(7 << 1) /* R420 (4 pipes, 1 ctx) */
+#	define R300_GB_TILE_SIZE_8		(0 << 4)
+#	define R300_GB_TILE_SIZE_16		(1 << 4)
+#	define R300_GB_TILE_SIZE_32		(2 << 4)
+#	define R300_GB_SUPER_SIZE_1		(0 << 6)
+#	define R300_GB_SUPER_SIZE_2		(1 << 6)
+#	define R300_GB_SUPER_SIZE_4		(2 << 6)
+#	define R300_GB_SUPER_SIZE_8		(3 << 6)
+#	define R300_GB_SUPER_SIZE_16		(4 << 6)
+#	define R300_GB_SUPER_SIZE_32		(5 << 6)
+#	define R300_GB_SUPER_SIZE_64		(6 << 6)
+#	define R300_GB_SUPER_SIZE_128		(7 << 6)
 #	define R300_GB_SUPER_X_SHIFT		9	/* 3 bits wide */
 #	define R300_GB_SUPER_Y_SHIFT		12	/* 3 bits wide */
-#	define R300_GB_SUPER_TILE_A		0
-#	define R300_GB_SUPER_TILE_B		(1<<15)
-#	define R300_GB_SUBPIXEL_1_12		0
-#	define R300_GB_SUBPIXEL_1_16		(1<<16)
-
+#	define R300_GB_SUPER_TILE_A		(0 << 15)
+#	define R300_GB_SUPER_TILE_B		(1 << 15)
+#	define R300_GB_SUBPIXEL_1_12		(0 << 16)
+#	define R300_GB_SUBPIXEL_1_16		(1 << 16)
+#	define GB_TILE_CONFIG_QUADS_PER_RAS_4   (0 << 17)
+#	define GB_TILE_CONFIG_QUADS_PER_RAS_8   (1 << 17)
+#	define GB_TILE_CONFIG_QUADS_PER_RAS_16  (2 << 17)
+#	define GB_TILE_CONFIG_QUADS_PER_RAS_32  (3 << 17)
+#	define GB_TILE_CONFIG_BB_SCAN_INTERCEPT (0 << 19)
+#	define GB_TILE_CONFIG_BB_SCAN_BOUND_BOX (1 << 19)
+#	define GB_TILE_CONFIG_ALT_SCAN_EN_LR    (0 << 20)
+#	define GB_TILE_CONFIG_ALT_SCAN_EN_LRL   (1 << 20)
+#	define GB_TILE_CONFIG_ALT_OFFSET        (0 << 21)
+#	define GB_TILE_CONFIG_SUBPRECISION      (0 << 22)
+#	define GB_TILE_CONFIG_ALT_TILING_DEF    (0 << 23)
+#	define GB_TILE_CONFIG_ALT_TILING_3_2    (1 << 23)
+#	define GB_TILE_CONFIG_Z_EXTENDED_24_1   (0 << 24)
+#	define GB_TILE_CONFIG_Z_EXTENDED_S25_1  (1 << 24)
+
+/* Specifies the sizes of the various FIFO`s in the sc/rs/us. This register must be the first one written */
 #define R300_GB_FIFO_SIZE	0x4024
 	/* each of the following is 2 bits wide */
 #define R300_GB_FIFO_SIZE_32	0
@@ -516,30 +644,102 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_OFIFO_HIGHWATER_SHIFT	22	/* two bits only */
 #	define R300_CUBE_FIFO_HIGHWATER_COL_SHIFT	24
 
-#define R300_GB_SELECT	0x401C
-#	define R300_GB_FOG_SELECT_C0A		0
-#	define R300_GB_FOG_SELECT_C1A		1
-#	define R300_GB_FOG_SELECT_C2A		2
-#	define R300_GB_FOG_SELECT_C3A		3
-#	define R300_GB_FOG_SELECT_1_1_W	4
-#	define R300_GB_FOG_SELECT_Z		5
-#	define R300_GB_DEPTH_SELECT_Z		0
-#	define R300_GB_DEPTH_SELECT_1_1_W	(1<<3)
-#	define R300_GB_W_SELECT_1_W		0
-#	define R300_GB_W_SELECT_1		(1<<4)
-
-#define R300_GB_AA_CONFIG		0x4020
-#	define R300_AA_DISABLE			0x00
-#	define R300_AA_ENABLE			0x01
-#	define R300_AA_SUBSAMPLES_2		0
-#	define R300_AA_SUBSAMPLES_3		(1<<1)
-#	define R300_AA_SUBSAMPLES_4		(2<<1)
-#	define R300_AA_SUBSAMPLES_6		(3<<1)
+#define GB_Z_PEQ_CONFIG                          0x4028
+#	define GB_Z_PEQ_CONFIG_Z_PEQ_SIZE_4_4    (0 << 0)
+#	define GB_Z_PEQ_CONFIG_Z_PEQ_SIZE_8_8    (1 << 0)
+
+/* Specifies various polygon specific selects (fog, depth, perspective). */
+#define R300_GB_SELECT                           0x401c
+#	define R300_GB_FOG_SELECT_C0A		(0 << 0)
+#	define R300_GB_FOG_SELECT_C1A           (1 << 0)
+#	define R300_GB_FOG_SELECT_C2A           (2 << 0)
+#	define R300_GB_FOG_SELECT_C3A           (3 << 0)
+#	define R300_GB_FOG_SELECT_1_1_W         (4 << 0)
+#	define R300_GB_FOG_SELECT_Z		(5 << 0)
+#	define R300_GB_DEPTH_SELECT_Z		(0 << 3
+#	define R300_GB_DEPTH_SELECT_1_1_W	(1 << 3)
+#	define R300_GB_W_SELECT_1_W		(0 << 4)
+#	define R300_GB_W_SELECT_1		(1 << 4)
+#	define R300_GB_FOG_STUFF_DISABLE        (0 << 5)
+#	define R300_GB_FOG_STUFF_ENABLE         (1 << 5)
+#	define R300_GB_FOG_STUFF_TEX_SHIFT      6
+#	define R300_GB_FOG_STUFF_TEX_MASK       0x000003c0
+#	define R300_GB_FOG_STUFF_COMP_SHIFT     10
+#	define R300_GB_FOG_STUFF_COMP_MASK      0x00000c00
+
+/* Specifies the graphics pipeline configuration for antialiasing. */
+#define GB_AA_CONFIG   	                         0x4020
+#	define GB_AA_CONFIG_AA_DISABLE           (0 << 0)
+#	define GB_AA_CONFIG_AA_ENABLE            (1 << 0)
+#	define GB_AA_CONFIG_NUM_AA_SUBSAMPLES_2  (0 << 1)
+#	define GB_AA_CONFIG_NUM_AA_SUBSAMPLES_3  (1 << 1)
+#	define GB_AA_CONFIG_NUM_AA_SUBSAMPLES_4  (2 << 1)
+#	define GB_AA_CONFIG_NUM_AA_SUBSAMPLES_6  (3 << 1)
+
+/* Selects which of 4 pipes are active. */
+#define GB_PIPE_SELECT                           0x402c
+#	define GB_PIPE_SELECT_PIPE0_ID_SHIFT  0
+#	define GB_PIPE_SELECT_PIPE1_ID_SHIFT  2
+#	define GB_PIPE_SELECT_PIPE2_ID_SHIFT  4
+#	define GB_PIPE_SELECT_PIPE3_ID_SHIFT  6
+#	define GB_PIPE_SELECT_PIPE_MASK_SHIFT 8
+#	define GB_PIPE_SELECT_MAX_PIPE        12
+#	define GB_PIPE_SELECT_BAD_PIPES       14
+#	define GB_PIPE_SELECT_CONFIG_PIPES    18
+
+
+/* Specifies the sizes of the various FIFO`s in the sc/rs. */
+#define GB_FIFO_SIZE1                            0x4070
+/* High water mark for SC input fifo */
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_IFIFO_SHIFT 0
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_IFIFO_MASK  0x0000003f
+/* High water mark for SC input fifo (B) */
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_BFIFO_SHIFT 6
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_BFIFO_MASK  0x00000fc0
+/* High water mark for RS colors' fifo */
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_COL_SHIFT   12
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_COL_MASK    0x0003f000
+/* High water mark for RS textures' fifo */
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_TEX_SHIFT   18
+#	define GB_FIFO_SIZE1_SC_HIGHWATER_TEX_MASK    0x00fc0000
+
+/* This table specifies the source location and format for up to 16 texture
+ * addresses (i[0]:i[15]) and four colors (c[0]:c[3])
+ */
+#define R500_RS_IP_0					0x4074
+#define R500_RS_IP_1					0x4078
+#define R500_RS_IP_2					0x407C
+#define R500_RS_IP_3					0x4080
+#define R500_RS_IP_4					0x4084
+#define R500_RS_IP_5					0x4088
+#define R500_RS_IP_6					0x408C
+#define R500_RS_IP_7					0x4090
+#define R500_RS_IP_8					0x4094
+#define R500_RS_IP_9					0x4098
+#define R500_RS_IP_10					0x409C
+#define R500_RS_IP_11					0x40A0
+#define R500_RS_IP_12					0x40A4
+#define R500_RS_IP_13					0x40A8
+#define R500_RS_IP_14					0x40AC
+#define R500_RS_IP_15					0x40B0
+#define R500_RS_IP_PTR_K0                               62
+#define R500_RS_IP_PTR_K1                               63
+#define R500_RS_IP_TEX_PTR_S_SHIFT 			0
+#define R500_RS_IP_TEX_PTR_T_SHIFT 			6
+#define R500_RS_IP_TEX_PTR_R_SHIFT 			12
+#define R500_RS_IP_TEX_PTR_Q_SHIFT 			18
+#define R500_RS_IP_COL_PTR_SHIFT 			24
+#define R500_RS_IP_COL_FMT_SHIFT 			27
+#	define R500_RS_COL_PTR(x)		        (x << 24)
+#       define R500_RS_COL_FMT(x)                       (x << 27)
+/* gap */
+#define R500_RS_IP_OFFSET_DIS 				(0 << 31)
+#define R500_RS_IP_OFFSET_EN 				(1 << 31)
 
 /* gap */
 
 /* Zero to flush caches. */
-#define R300_TX_CNTL                        0x4100
+#define R300_TX_INVALTAGS                   0x4100
 #define R300_TX_FLUSH                       0x0
 
 /* The upper enable bits are guessed, based on fglrx reported limits. */
@@ -561,53 +761,335 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_TX_ENABLE_14                 (1 << 14)
 #       define R300_TX_ENABLE_15                 (1 << 15)
 
-/* The pointsize is given in multiples of 6. The pointsize can be
- * enormous: Clear() renders a single point that fills the entire
- * framebuffer.
+#define R500_TX_FILTER_4		    0x4110
+#	define R500_TX_WEIGHT_1_SHIFT            (0)
+#	define R500_TX_WEIGHT_0_SHIFT            (11)
+#	define R500_TX_WEIGHT_PAIR               (1<<22)
+#	define R500_TX_PHASE_SHIFT               (23)
+#	define R500_TX_DIRECTION_HORIZONTAL	 (0<<27)
+#	define R500_TX_DIRECTION_VERITCAL	 (1<<27)
+
+/* S Texture Coordinate of Vertex 0 for Point texture stuffing (LLC) */
+#define R300_GA_POINT_S0                              0x4200
+
+/* T Texture Coordinate of Vertex 0 for Point texture stuffing (LLC) */
+#define R300_GA_POINT_T0                              0x4204
+
+/* S Texture Coordinate of Vertex 2 for Point texture stuffing (URC) */
+#define R300_GA_POINT_S1                              0x4208
+
+/* T Texture Coordinate of Vertex 2 for Point texture stuffing (URC) */
+#define R300_GA_POINT_T1                              0x420c
+
+/* Specifies amount to shift integer position of vertex (screen space) before
+ * converting to float for triangle stipple.
+ */
+#define R300_GA_TRIANGLE_STIPPLE            0x4214
+#	define R300_GA_TRIANGLE_STIPPLE_X_SHIFT_SHIFT 0
+#	define R300_GA_TRIANGLE_STIPPLE_X_SHIFT_MASK  0x0000000f
+#	define R300_GA_TRIANGLE_STIPPLE_Y_SHIFT_SHIFT 16
+#	define R300_GA_TRIANGLE_STIPPLE_Y_SHIFT_MASK  0x000f0000
+
+/* The pointsize is given in multiples of 6. The pointsize can be enormous:
+ * Clear() renders a single point that fills the entire framebuffer.
+ * 1/2 Height of point; fixed (16.0), subpixel format (1/12 or 1/16, even if in
+ * 8b precision).
  */
-#define R300_RE_POINTSIZE                   0x421C
-#       define R300_POINTSIZE_Y_SHIFT            0
-#       define R300_POINTSIZE_Y_MASK             (0xFFFF << 0) /* GUESS */
-#       define R300_POINTSIZE_X_SHIFT            16
-#       define R300_POINTSIZE_X_MASK             (0xFFFF << 16) /* GUESS */
+#define R300_GA_POINT_SIZE                   0x421C
+#       define R300_POINTSIZE_Y_SHIFT         0
+#       define R300_POINTSIZE_Y_MASK          0x0000ffff
+#       define R300_POINTSIZE_X_SHIFT         16
+#       define R300_POINTSIZE_X_MASK          0xffff0000
 #       define R300_POINTSIZE_MAX             (R300_POINTSIZE_Y_MASK / 6)
 
-/* The line width is given in multiples of 6.
+/* Blue fill color */
+#define R500_GA_FILL_R                                0x4220
+
+/* Blue fill color */
+#define R500_GA_FILL_G                                0x4224
+
+/* Blue fill color */
+#define R500_GA_FILL_B                                0x4228
+
+/* Alpha fill color */
+#define R500_GA_FILL_A                                0x422c
+
+
+/* Specifies maximum and minimum point & sprite sizes for per vertex size
+ * specification. The lower part (15:0) is MIN and (31:16) is max.
+ */
+#define R300_GA_POINT_MINMAX                0x4230
+#       define R300_GA_POINT_MINMAX_MIN_SHIFT          0
+#       define R300_GA_POINT_MINMAX_MIN_MASK           (0xFFFF << 0)
+#       define R300_GA_POINT_MINMAX_MAX_SHIFT          16
+#       define R300_GA_POINT_MINMAX_MAX_MASK           (0xFFFF << 16)
+
+/* 1/2 width of line, in subpixels (1/12 or 1/16 only, even in 8b
+ * subprecision); (16.0) fixed format.
+ *
+ * The line width is given in multiples of 6.
  * In default mode lines are classified as vertical lines.
  * HO: horizontal
  * VE: vertical or horizontal
  * HO & VE: no classification
  */
-#define R300_RE_LINE_CNT                      0x4234
-#       define R300_LINESIZE_SHIFT            0
-#       define R300_LINESIZE_MASK             (0xFFFF << 0) /* GUESS */
-#       define R300_LINESIZE_MAX             (R300_LINESIZE_MASK / 6)
+#define R300_GA_LINE_CNTL                             0x4234
+#       define R300_GA_LINE_CNTL_WIDTH_SHIFT       0
+#       define R300_GA_LINE_CNTL_WIDTH_MASK        0x0000ffff
+#	define R300_GA_LINE_CNTL_END_TYPE_HOR      (0 << 16)
+#	define R300_GA_LINE_CNTL_END_TYPE_VER      (1 << 16)
+#	define R300_GA_LINE_CNTL_END_TYPE_SQR      (2 << 16) /* horizontal or vertical depending upon slope */
+#	define R300_GA_LINE_CNTL_END_TYPE_COMP     (3 << 16) /* Computed (perpendicular to slope) */
+#	define R500_GA_LINE_CNTL_SORT_NO           (0 << 18)
+#	define R500_GA_LINE_CNTL_SORT_MINX_MINY    (1 << 18)
+/** TODO: looks wrong */
+#       define R300_LINESIZE_MAX              (R300_GA_LINE_CNTL_WIDTH_MASK / 6)
+/** TODO: looks wrong */
 #       define R300_LINE_CNT_HO               (1 << 16)
+/** TODO: looks wrong */
 #       define R300_LINE_CNT_VE               (1 << 17)
 
-/* Some sort of scale or clamp value for texcoordless textures. */
-#define R300_RE_UNK4238                       0x4238
-
-/* Something shade related */
-#define R300_RE_SHADE                         0x4274
-
-#define R300_RE_SHADE_MODEL                   0x4278
-#	define R300_RE_SHADE_MODEL_SMOOTH     0x3aaaa
-#	define R300_RE_SHADE_MODEL_FLAT       0x39595
-
-/* Dangerous */
-#define R300_RE_POLYGON_MODE                  0x4288
-#	define R300_PM_ENABLED                (1 << 0)
-#	define R300_PM_FRONT_POINT            (0 << 0)
-#	define R300_PM_BACK_POINT             (0 << 0)
-#	define R300_PM_FRONT_LINE             (1 << 4)
-#	define R300_PM_FRONT_FILL             (1 << 5)
-#	define R300_PM_BACK_LINE              (1 << 7)
-#	define R300_PM_BACK_FILL              (1 << 8)
-
-/* Fog parameters */
-#define R300_RE_FOG_SCALE                     0x4294
-#define R300_RE_FOG_START                     0x4298
+/* Line Stipple configuration information. */
+#define R300_GA_LINE_STIPPLE_CONFIG                   0x4238
+#	define R300_GA_LINE_STIPPLE_CONFIG_LINE_RESET_NO     (0 << 0)
+#	define R300_GA_LINE_STIPPLE_CONFIG_LINE_RESET_LINE   (1 << 0)
+#	define R300_GA_LINE_STIPPLE_CONFIG_LINE_RESET_PACKET (2 << 0)
+#	define R300_GA_LINE_STIPPLE_CONFIG_STIPPLE_SCALE_SHIFT 2
+#	define R300_GA_LINE_STIPPLE_CONFIG_STIPPLE_SCALE_MASK  0xfffffffc
+
+/* Used to load US instructions and constants */
+#define R500_GA_US_VECTOR_INDEX               0x4250
+#	define R500_GA_US_VECTOR_INDEX_SHIFT       0
+#	define R500_GA_US_VECTOR_INDEX_MASK        0x000000ff
+#	define R500_GA_US_VECTOR_INDEX_TYPE_INSTR  (0 << 16)
+#	define R500_GA_US_VECTOR_INDEX_TYPE_CONST  (1 << 16)
+#	define R500_GA_US_VECTOR_INDEX_CLAMP_NO    (0 << 17)
+#	define R500_GA_US_VECTOR_INDEX_CLAMP_CONST (1 << 17)
+
+/* Data register for loading US instructions and constants */
+#define R500_GA_US_VECTOR_DATA                0x4254
+
+/* Specifies color properties and mappings of textures. */
+#define R500_GA_COLOR_CONTROL_PS3                     0x4258
+#	define R500_TEX0_SHADING_PS3_SOLID       (0 << 0)
+#	define R500_TEX0_SHADING_PS3_FLAT        (1 << 0)
+#	define R500_TEX0_SHADING_PS3_GOURAUD     (2 << 0)
+#	define R500_TEX1_SHADING_PS3_SOLID       (0 << 2)
+#	define R500_TEX1_SHADING_PS3_FLAT        (1 << 2)
+#	define R500_TEX1_SHADING_PS3_GOURAUD     (2 << 2)
+#	define R500_TEX2_SHADING_PS3_SOLID       (0 << 4)
+#	define R500_TEX2_SHADING_PS3_FLAT        (1 << 4)
+#	define R500_TEX2_SHADING_PS3_GOURAUD     (2 << 4)
+#	define R500_TEX3_SHADING_PS3_SOLID       (0 << 6)
+#	define R500_TEX3_SHADING_PS3_FLAT        (1 << 6)
+#	define R500_TEX3_SHADING_PS3_GOURAUD     (2 << 6)
+#	define R500_TEX4_SHADING_PS3_SOLID       (0 << 8)
+#	define R500_TEX4_SHADING_PS3_FLAT        (1 << 8)
+#	define R500_TEX4_SHADING_PS3_GOURAUD     (2 << 8)
+#	define R500_TEX5_SHADING_PS3_SOLID       (0 << 10)
+#	define R500_TEX5_SHADING_PS3_FLAT        (1 << 10)
+#	define R500_TEX5_SHADING_PS3_GOURAUD     (2 << 10)
+#	define R500_TEX6_SHADING_PS3_SOLID       (0 << 12)
+#	define R500_TEX6_SHADING_PS3_FLAT        (1 << 12)
+#	define R500_TEX6_SHADING_PS3_GOURAUD     (2 << 12)
+#	define R500_TEX7_SHADING_PS3_SOLID       (0 << 14)
+#	define R500_TEX7_SHADING_PS3_FLAT        (1 << 14)
+#	define R500_TEX7_SHADING_PS3_GOURAUD     (2 << 14)
+#	define R500_TEX8_SHADING_PS3_SOLID       (0 << 16)
+#	define R500_TEX8_SHADING_PS3_FLAT        (1 << 16)
+#	define R500_TEX8_SHADING_PS3_GOURAUD     (2 << 16)
+#	define R500_TEX9_SHADING_PS3_SOLID       (0 << 18)
+#	define R500_TEX9_SHADING_PS3_FLAT        (1 << 18)
+#	define R500_TEX9_SHADING_PS3_GOURAUD     (2 << 18)
+#	define R500_TEX10_SHADING_PS3_SOLID      (0 << 20)
+#	define R500_TEX10_SHADING_PS3_FLAT       (1 << 20)
+#	define R500_TEX10_SHADING_PS3_GOURAUD    (2 << 20)
+#	define R500_COLOR0_TEX_OVERRIDE_NO       (0 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_0    (1 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_1    (2 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_2    (3 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_3    (4 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_4    (5 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_5    (6 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_6    (7 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_7    (8 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_8_C2 (9 << 22)
+#	define R500_COLOR0_TEX_OVERRIDE_TEX_9_C3 (10 << 22)
+#	define R500_COLOR1_TEX_OVERRIDE_NO       (0 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_0    (1 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_1    (2 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_2    (3 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_3    (4 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_4    (5 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_5    (6 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_6    (7 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_7    (8 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_8_C2 (9 << 26)
+#	define R500_COLOR1_TEX_OVERRIDE_TEX_9_C3 (10 << 26)
+
+/* Returns idle status of various G3D block, captured when GA_IDLE written or
+ * when hard or soft reset asserted.
+ */
+#define R500_GA_IDLE                                  0x425c
+#	define R500_GA_IDLE_PIPE3_Z_IDLE  (0 << 0)
+#	define R500_GA_IDLE_PIPE2_Z_IDLE  (0 << 1)
+#	define R500_GA_IDLE_PIPE3_CD_IDLE (0 << 2)
+#	define R500_GA_IDLE_PIPE2_CD_IDLE (0 << 3)
+#	define R500_GA_IDLE_PIPE3_FG_IDLE (0 << 4)
+#	define R500_GA_IDLE_PIPE2_FG_IDLE (0 << 5)
+#	define R500_GA_IDLE_PIPE3_US_IDLE (0 << 6)
+#	define R500_GA_IDLE_PIPE2_US_IDLE (0 << 7)
+#	define R500_GA_IDLE_PIPE3_SC_IDLE (0 << 8)
+#	define R500_GA_IDLE_PIPE2_SC_IDLE (0 << 9)
+#	define R500_GA_IDLE_PIPE3_RS_IDLE (0 << 10)
+#	define R500_GA_IDLE_PIPE2_RS_IDLE (0 << 11)
+#	define R500_GA_IDLE_PIPE1_Z_IDLE  (0 << 12)
+#	define R500_GA_IDLE_PIPE0_Z_IDLE  (0 << 13)
+#	define R500_GA_IDLE_PIPE1_CD_IDLE (0 << 14)
+#	define R500_GA_IDLE_PIPE0_CD_IDLE (0 << 15)
+#	define R500_GA_IDLE_PIPE1_FG_IDLE (0 << 16)
+#	define R500_GA_IDLE_PIPE0_FG_IDLE (0 << 17)
+#	define R500_GA_IDLE_PIPE1_US_IDLE (0 << 18)
+#	define R500_GA_IDLE_PIPE0_US_IDLE (0 << 19)
+#	define R500_GA_IDLE_PIPE1_SC_IDLE (0 << 20)
+#	define R500_GA_IDLE_PIPE0_SC_IDLE (0 << 21)
+#	define R500_GA_IDLE_PIPE1_RS_IDLE (0 << 22)
+#	define R500_GA_IDLE_PIPE0_RS_IDLE (0 << 23)
+#	define R500_GA_IDLE_SU_IDLE       (0 << 24)
+#	define R500_GA_IDLE_GA_IDLE       (0 << 25)
+#	define R500_GA_IDLE_GA_UNIT2_IDLE (0 << 26)
+
+/* Current value of stipple accumulator. */
+#define R300_GA_LINE_STIPPLE_VALUE            0x4260
+
+/* S Texture Coordinate Value for Vertex 0 of Line (stuff textures -- i.e. AA) */
+#define R300_GA_LINE_S0                               0x4264
+/* S Texture Coordinate Value for Vertex 1 of Lines (V2 of parallelogram -- stuff textures -- i.e. AA) */
+#define R300_GA_LINE_S1                               0x4268
+
+/* GA Input fifo high water marks */
+#define R500_GA_FIFO_CNTL                             0x4270
+#	define R500_GA_FIFO_CNTL_VERTEX_FIFO_MASK   0x00000007
+#	define R500_GA_FIFO_CNTL_VERTEX_FIFO_SHIFT  0
+#	define R500_GA_FIFO_CNTL_VERTEX_INDEX_MASK  0x00000038
+#	define R500_GA_FIFO_CNTL_VERTEX_INDEX_SHIFT 3
+#	define R500_GA_FIFO_CNTL_VERTEX_REG_MASK    0x00003fc0
+#	define R500_GA_FIFO_CNTL_VERTEX_REG_SHIFT   6
+
+/* GA enhance/tweaks */
+#define R300_GA_ENHANCE                               0x4274
+#	define R300_GA_ENHANCE_DEADLOCK_CNTL_NO_EFFECT   (0 << 0)
+#	define R300_GA_ENHANCE_DEADLOCK_CNTL_PREVENT_TCL (1 << 0) /* Prevents TCL interface from deadlocking on GA side. */
+#	define R300_GA_ENHANCE_FASTSYNC_CNTL_NO_EFFECT   (0 << 1)
+#	define R300_GA_ENHANCE_FASTSYNC_CNTL_ENABLE      (1 << 1) /* Enables high-performance register/primitive switching. */
+#	define R500_GA_ENHANCE_REG_READWRITE_NO_EFFECT   (0 << 2) /* R520+ only */
+#	define R500_GA_ENHANCE_REG_READWRITE_ENABLE      (1 << 2) /* R520+ only, Enables GA support of simultaneous register reads and writes. */
+#	define R500_GA_ENHANCE_REG_NOSTALL_NO_EFFECT     (0 << 3)
+#	define R500_GA_ENHANCE_REG_NOSTALL_ENABLE        (1 << 3) /* Enables GA support of no-stall reads for register read back. */
+
+#define R300_GA_COLOR_CONTROL                   0x4278
+#	define R300_GA_COLOR_CONTROL_RGB0_SHADING_SOLID      (0 << 0)
+#	define R300_GA_COLOR_CONTROL_RGB0_SHADING_FLAT       (1 << 0)
+#	define R300_GA_COLOR_CONTROL_RGB0_SHADING_GOURAUD    (2 << 0)
+#	define R300_GA_COLOR_CONTROL_ALPHA0_SHADING_SOLID    (0 << 2)
+#	define R300_GA_COLOR_CONTROL_ALPHA0_SHADING_FLAT     (1 << 2)
+#	define R300_GA_COLOR_CONTROL_ALPHA0_SHADING_GOURAUD  (2 << 2)
+#	define R300_GA_COLOR_CONTROL_RGB1_SHADING_SOLID      (0 << 4)
+#	define R300_GA_COLOR_CONTROL_RGB1_SHADING_FLAT       (1 << 4)
+#	define R300_GA_COLOR_CONTROL_RGB1_SHADING_GOURAUD    (2 << 4)
+#	define R300_GA_COLOR_CONTROL_ALPHA1_SHADING_SOLID    (0 << 6)
+#	define R300_GA_COLOR_CONTROL_ALPHA1_SHADING_FLAT     (1 << 6)
+#	define R300_GA_COLOR_CONTROL_ALPHA1_SHADING_GOURAUD  (2 << 6)
+#	define R300_GA_COLOR_CONTROL_RGB2_SHADING_SOLID      (0 << 8)
+#	define R300_GA_COLOR_CONTROL_RGB2_SHADING_FLAT       (1 << 8)
+#	define R300_GA_COLOR_CONTROL_RGB2_SHADING_GOURAUD    (2 << 8)
+#	define R300_GA_COLOR_CONTROL_ALPHA2_SHADING_SOLID    (0 << 10)
+#	define R300_GA_COLOR_CONTROL_ALPHA2_SHADING_FLAT     (1 << 10)
+#	define R300_GA_COLOR_CONTROL_ALPHA2_SHADING_GOURAUD  (2 << 10)
+#	define R300_GA_COLOR_CONTROL_RGB3_SHADING_SOLID      (0 << 12)
+#	define R300_GA_COLOR_CONTROL_RGB3_SHADING_FLAT       (1 << 12)
+#	define R300_GA_COLOR_CONTROL_RGB3_SHADING_GOURAUD    (2 << 12)
+#	define R300_GA_COLOR_CONTROL_ALPHA3_SHADING_SOLID    (0 << 14)
+#	define R300_GA_COLOR_CONTROL_ALPHA3_SHADING_FLAT     (1 << 14)
+#	define R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD  (2 << 14)
+#	define R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_FIRST  (0 << 16)
+#	define R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_SECOND (1 << 16)
+#	define R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_THIRD  (2 << 16)
+#	define R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST   (3 << 16)
+
+/** TODO: might be candidate for removal */
+#	define R300_RE_SHADE_MODEL_SMOOTH     ( \
+	R300_GA_COLOR_CONTROL_RGB0_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA0_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_RGB1_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA1_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_RGB2_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA2_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_RGB3_SHADING_GOURAUD | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST )
+/** TODO: might be candidate for removal, the GOURAUD stuff also looks buggy to me */
+#	define R300_RE_SHADE_MODEL_FLAT     ( \
+	R300_GA_COLOR_CONTROL_RGB0_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA0_SHADING_FLAT | \
+	R300_GA_COLOR_CONTROL_RGB1_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA1_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_RGB2_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA2_SHADING_FLAT | \
+	R300_GA_COLOR_CONTROL_RGB3_SHADING_FLAT | R300_GA_COLOR_CONTROL_ALPHA3_SHADING_GOURAUD | \
+	R300_GA_COLOR_CONTROL_PROVOKING_VERTEX_LAST )
+
+/* Specifies red & green components of fill color -- S312 format -- Backwards comp. */
+#define R300_GA_SOLID_RG                         0x427c
+#	define GA_SOLID_RG_COLOR_GREEN_SHIFT 0
+#	define GA_SOLID_RG_COLOR_GREEN_MASK  0x0000ffff
+#	define GA_SOLID_RG_COLOR_RED_SHIFT   16
+#	define GA_SOLID_RG_COLOR_RED_MASK    0xffff0000
+/* Specifies blue & alpha components of fill color -- S312 format -- Backwards comp. */
+#define R300_GA_SOLID_BA                         0x4280
+#	define GA_SOLID_BA_COLOR_ALPHA_SHIFT 0
+#	define GA_SOLID_BA_COLOR_ALPHA_MASK  0x0000ffff
+#	define GA_SOLID_BA_COLOR_BLUE_SHIFT  16
+#	define GA_SOLID_BA_COLOR_BLUE_MASK   0xffff0000
+
+/* Polygon Mode
+ * Dangerous
+ */
+#define R300_GA_POLY_MODE                             0x4288
+#	define R300_GA_POLY_MODE_DISABLE           (0 << 0)
+#	define R300_GA_POLY_MODE_DUAL              (1 << 0) /* send 2 sets of 3 polys with specified poly type */
+/* reserved */
+#	define R300_GA_POLY_MODE_FRONT_PTYPE_POINT (0 << 4)
+#	define R300_GA_POLY_MODE_FRONT_PTYPE_LINE  (1 << 4)
+#	define R300_GA_POLY_MODE_FRONT_PTYPE_TRI   (2 << 4)
+/* reserved */
+#	define R300_GA_POLY_MODE_BACK_PTYPE_POINT  (0 << 7)
+#	define R300_GA_POLY_MODE_BACK_PTYPE_LINE   (1 << 7)
+#	define R300_GA_POLY_MODE_BACK_PTYPE_TRI    (2 << 7)
+/* reserved */
+
+/* Specifies the rouding mode for geometry & color SPFP to FP conversions. */
+#define R300_GA_ROUND_MODE                            0x428c
+#	define R300_GA_ROUND_MODE_GEOMETRY_ROUND_TRUNC   (0 << 0)
+#	define R300_GA_ROUND_MODE_GEOMETRY_ROUND_NEAREST (1 << 0)
+#	define R300_GA_ROUND_MODE_COLOR_ROUND_TRUNC      (0 << 2)
+#	define R300_GA_ROUND_MODE_COLOR_ROUND_NEAREST    (1 << 2)
+#	define R300_GA_ROUND_MODE_RGB_CLAMP_RGB          (0 << 4)
+#	define R300_GA_ROUND_MODE_RGB_CLAMP_FP20         (1 << 4)
+#	define R300_GA_ROUND_MODE_ALPHA_CLAMP_RGB        (0 << 5)
+#	define R300_GA_ROUND_MODE_ALPHA_CLAMP_FP20       (1 << 5)
+#	define R500_GA_ROUND_MODE_GEOMETRY_MASK_SHIFT    6
+#	define R500_GA_ROUND_MODE_GEOMETRY_MASK_MASK     0x000003c0
+
+/* Specifies x & y offsets for vertex data after conversion to FP.
+ * Offsets are in S15 format (subpixels -- 1/12 or 1/16, even in 8b
+ * subprecision).
+ */
+#define R300_GA_OFFSET                                0x4290
+#	define R300_GA_OFFSET_X_OFFSET_SHIFT 0
+#	define R300_GA_OFFSET_X_OFFSET_MASK  0x0000ffff
+#	define R300_GA_OFFSET_Y_OFFSET_SHIFT 16
+#	define R300_GA_OFFSET_Y_OFFSET_MASK  0xffff0000
+
+/* Specifies the scale to apply to fog. */
+#define R300_GA_FOG_SCALE                     0x4294
+/* Specifies the offset to apply to fog. */
+#define R300_GA_FOG_OFFSET                    0x4298
+/* Specifies number of cycles to assert reset, and also causes RB3D soft reset to assert. */
+#define R300_GA_SOFT_RESET                    0x429c
 
 /* Not sure why there are duplicate of factor and constant values.
  * My best guess so far is that there are seperate zbiases for test and write.
@@ -615,11 +1097,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Some of the tests indicate that fgl has a fallback implementation of zbias
  * via pixel shaders.
  */
-#define R300_RE_ZBIAS_CNTL                    0x42A0 /* GUESS */
-#define R300_RE_ZBIAS_T_FACTOR                0x42A4
-#define R300_RE_ZBIAS_T_CONSTANT              0x42A8
-#define R300_RE_ZBIAS_W_FACTOR                0x42AC
-#define R300_RE_ZBIAS_W_CONSTANT              0x42B0
+#define R300_SU_TEX_WRAP                      0x42A0
+#define R300_SU_POLY_OFFSET_FRONT_SCALE       0x42A4
+#define R300_SU_POLY_OFFSET_FRONT_OFFSET      0x42A8
+#define R300_SU_POLY_OFFSET_BACK_SCALE        0x42AC
+#define R300_SU_POLY_OFFSET_BACK_OFFSET       0x42B0
 
 /* This register needs to be set to (1<<1) for RV350 to correctly
  * perform depth test (see --vb-triangles in r300_demo)
@@ -630,31 +1112,44 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *  One to enable depth test and one for depth write.
  * Yet this doesnt explain why depth writes work ...
  */
-#define R300_RE_OCCLUSION_CNTL		    0x42B4
-#	define R300_OCCLUSION_ON		(1<<1)
+#define R300_SU_POLY_OFFSET_ENABLE	       0x42B4
+#	define R300_FRONT_ENABLE	       (1 << 0)
+#	define R300_BACK_ENABLE 	       (1 << 1)
+#	define R300_PARA_ENABLE 	       (1 << 2)
 
-#define R300_RE_CULL_CNTL                   0x42B8
+#define R300_SU_CULL_MODE                      0x42B8
 #       define R300_CULL_FRONT                   (1 << 0)
 #       define R300_CULL_BACK                    (1 << 1)
 #       define R300_FRONT_FACE_CCW               (0 << 2)
 #       define R300_FRONT_FACE_CW                (1 << 2)
 
+/* SU Depth Scale value */
+#define R300_SU_DEPTH_SCALE                 0x42c0
+/* SU Depth Offset value */
+#define R300_SU_DEPTH_OFFSET                0x42c4
+
 
 /* BEGIN: Rasterization / Interpolators - many guesses */
 
-/* 0_UNKNOWN_18 has always been set except for clear operations.
+/*
  * TC_CNT is the number of incoming texture coordinate sets (i.e. it depends
  * on the vertex program, *not* the fragment program)
  */
-#define R300_RS_CNTL_0                      0x4300
-#       define R300_RS_CNTL_TC_CNT_SHIFT         2
-#       define R300_RS_CNTL_TC_CNT_MASK          (7 << 2)
-	/* number of color interpolators used */
-#	define R300_RS_CNTL_CI_CNT_SHIFT         7
-#       define R300_RS_CNTL_0_UNKNOWN_18         (1 << 18)
-	/* Guess: RS_CNTL_1 holds the index of the highest used RS_ROUTE_n
-	   register. */
-#define R300_RS_CNTL_1                      0x4304
+#define R300_RS_COUNT                      0x4300
+#       define R300_IT_COUNT_SHIFT               0
+#       define R300_IT_COUNT_MASK                0x0000007f
+#       define R300_IC_COUNT_SHIFT               7
+#       define R300_IC_COUNT_MASK                0x00000780
+#       define R300_W_ADDR_SHIFT                 12
+#       define R300_W_ADDR_MASK                  0x0003f000
+#       define R300_HIRES_DIS                    (0 << 18)
+#       define R300_HIRES_EN                     (1 << 18)
+
+#define R300_RS_INST_COUNT                       0x4304
+#       define R300_RS_INST_COUNT_SHIFT          0
+#       define R300_RS_INST_COUNT_MASK           0x0000000f
+#       define R300_RS_TX_OFFSET_SHIFT           5
+#	define R300_RS_TX_OFFSET_MASK            0x000000e0
 
 /* gap */
 
@@ -671,63 +1166,108 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * Note: The _UNKNOWN constants are always set in their respective
  * register. I don't know if this is necessary.
  */
-#define R300_RS_INTERP_0                    0x4310
-#define R300_RS_INTERP_1                    0x4314
-#       define R300_RS_INTERP_1_UNKNOWN          0x40
-#define R300_RS_INTERP_2                    0x4318
-#       define R300_RS_INTERP_2_UNKNOWN          0x80
-#define R300_RS_INTERP_3                    0x431C
-#       define R300_RS_INTERP_3_UNKNOWN          0xC0
-#define R300_RS_INTERP_4                    0x4320
-#define R300_RS_INTERP_5                    0x4324
-#define R300_RS_INTERP_6                    0x4328
-#define R300_RS_INTERP_7                    0x432C
-#       define R300_RS_INTERP_SRC_SHIFT          2
-#       define R300_RS_INTERP_SRC_MASK           (7 << 2)
-#       define R300_RS_INTERP_USED               0x00D10000
+#define R300_RS_IP_0				        0x4310
+#define R300_RS_IP_1				        0x4314
+#define R300_RS_IP_2				        0x4318
+#define R300_RS_IP_3				        0x431C
+#       define R300_RS_INTERP_SRC_SHIFT          2 /* TODO: check for removal */
+#       define R300_RS_INTERP_SRC_MASK           (7 << 2) /* TODO: check for removal */
+#	define R300_RS_TEX_PTR(x)		        (x << 0)
+#	define R300_RS_COL_PTR(x)		        (x << 6)
+#	define R300_RS_COL_FMT(x)		        (x << 9)
+#	define R300_RS_COL_FMT_RGBA		        0
+#	define R300_RS_COL_FMT_RGB0		        1
+#	define R300_RS_COL_FMT_RGB1		        2
+#	define R300_RS_COL_FMT_000A		        4
+#	define R300_RS_COL_FMT_0000		        5
+#	define R300_RS_COL_FMT_0001		        6
+#	define R300_RS_COL_FMT_111A		        8
+#	define R300_RS_COL_FMT_1110		        9
+#	define R300_RS_COL_FMT_1111		        10
+#	define R300_RS_SEL_S(x)		                (x << 13)
+#	define R300_RS_SEL_T(x)		                (x << 16)
+#	define R300_RS_SEL_R(x)		                (x << 19)
+#	define R300_RS_SEL_Q(x)		                (x << 22)
+#	define R300_RS_SEL_C0		                0
+#	define R300_RS_SEL_C1		                1
+#	define R300_RS_SEL_C2		                2
+#	define R300_RS_SEL_C3		                3
+#	define R300_RS_SEL_K0		                4
+#	define R300_RS_SEL_K1		                5
+
+
+/*  */
+#define R500_RS_INST_0					0x4320
+#define R500_RS_INST_1					0x4324
+#define R500_RS_INST_2					0x4328
+#define R500_RS_INST_3					0x432c
+#define R500_RS_INST_4					0x4330
+#define R500_RS_INST_5					0x4334
+#define R500_RS_INST_6					0x4338
+#define R500_RS_INST_7					0x433c
+#define R500_RS_INST_8					0x4340
+#define R500_RS_INST_9					0x4344
+#define R500_RS_INST_10					0x4348
+#define R500_RS_INST_11					0x434c
+#define R500_RS_INST_12					0x4350
+#define R500_RS_INST_13					0x4354
+#define R500_RS_INST_14					0x4358
+#define R500_RS_INST_15					0x435c
+#define R500_RS_INST_TEX_ID_SHIFT			0
+#define R500_RS_INST_TEX_CN_WRITE			(1 << 4)
+#define R500_RS_INST_TEX_ADDR_SHIFT			5
+#define R500_RS_INST_COL_ID_SHIFT			12
+#define R500_RS_INST_COL_CN_NO_WRITE			(0 << 16)
+#define R500_RS_INST_COL_CN_WRITE			(1 << 16)
+#define R500_RS_INST_COL_CN_WRITE_FBUFFER		(2 << 16)
+#define R500_RS_INST_COL_CN_WRITE_BACKFACE		(3 << 16)
+#define R500_RS_INST_COL_ADDR_SHIFT			18
+#define R500_RS_INST_TEX_ADJ				(1 << 25)
+#define R500_RS_INST_W_CN				(1 << 26)
 
 /* These DWORDs control how vertex data is routed into fragment program
  * registers, after interpolators.
  */
-#define R300_RS_ROUTE_0                     0x4330
-#define R300_RS_ROUTE_1                     0x4334
-#define R300_RS_ROUTE_2                     0x4338
-#define R300_RS_ROUTE_3                     0x433C /* GUESS */
-#define R300_RS_ROUTE_4                     0x4340 /* GUESS */
-#define R300_RS_ROUTE_5                     0x4344 /* GUESS */
-#define R300_RS_ROUTE_6                     0x4348 /* GUESS */
-#define R300_RS_ROUTE_7                     0x434C /* GUESS */
-#       define R300_RS_ROUTE_SOURCE_INTERP_0     0
-#       define R300_RS_ROUTE_SOURCE_INTERP_1     1
-#       define R300_RS_ROUTE_SOURCE_INTERP_2     2
-#       define R300_RS_ROUTE_SOURCE_INTERP_3     3
-#       define R300_RS_ROUTE_SOURCE_INTERP_4     4
-#       define R300_RS_ROUTE_SOURCE_INTERP_5     5 /* GUESS */
-#       define R300_RS_ROUTE_SOURCE_INTERP_6     6 /* GUESS */
-#       define R300_RS_ROUTE_SOURCE_INTERP_7     7 /* GUESS */
-#       define R300_RS_ROUTE_ENABLE              (1 << 3) /* GUESS */
-#       define R300_RS_ROUTE_DEST_SHIFT          6
-#       define R300_RS_ROUTE_DEST_MASK           (31 << 6) /* GUESS */
-
-/* Special handling for color: When the fragment program uses color,
- * the ROUTE_0_COLOR bit is set and ROUTE_0_COLOR_DEST contains the
- * color register index.
- *
- * Apperently you may set the R300_RS_ROUTE_0_COLOR bit, but not provide any
- * R300_RS_ROUTE_0_COLOR_DEST value; this setup is used for clearing the state.
- * See r300_ioctl.c:r300EmitClearState. I'm not sure if this setup is strictly
- * correct or not. - Oliver.
- */
-#       define R300_RS_ROUTE_0_COLOR             (1 << 14)
-#       define R300_RS_ROUTE_0_COLOR_DEST_SHIFT  17
-#       define R300_RS_ROUTE_0_COLOR_DEST_MASK   (31 << 17) /* GUESS */
-/* As above, but for secondary color */
-#		define R300_RS_ROUTE_1_COLOR1            (1 << 14)
-#		define R300_RS_ROUTE_1_COLOR1_DEST_SHIFT 17
-#		define R300_RS_ROUTE_1_COLOR1_DEST_MASK  (31 << 17)
-#		define R300_RS_ROUTE_1_UNKNOWN11         (1 << 11)
+#define R300_RS_INST_0                     0x4330
+#define R300_RS_INST_1                     0x4334
+#define R300_RS_INST_2                     0x4338
+#define R300_RS_INST_3                     0x433C /* GUESS */
+#define R300_RS_INST_4                     0x4340 /* GUESS */
+#define R300_RS_INST_5                     0x4344 /* GUESS */
+#define R300_RS_INST_6                     0x4348 /* GUESS */
+#define R300_RS_INST_7                     0x434C /* GUESS */
+#	define R300_RS_INST_TEX_ID(x)  		((x) << 0)
+#	define R300_RS_INST_TEX_CN_WRITE 	(1 << 3)
+#	define R300_RS_INST_TEX_ADDR_SHIFT 	6
+#	define R300_RS_INST_COL_ID(x)		((x) << 11)
+#	define R300_RS_INST_COL_CN_WRITE	(1 << 14)
+#	define R300_RS_INST_COL_ADDR_SHIFT	17
+#	define R300_RS_INST_TEX_ADJ		(1 << 22)
+#	define R300_RS_COL_BIAS_UNUSED_SHIFT    23
+
 /* END: Rasterization / Interpolators - many guesses */
 
+/* Hierarchical Z Enable */
+#define R300_SC_HYPERZ                   0x43a4
+#	define R300_SC_HYPERZ_DISABLE     (0 << 0)
+#	define R300_SC_HYPERZ_ENABLE      (1 << 0)
+#	define R300_SC_HYPERZ_MIN         (0 << 1)
+#	define R300_SC_HYPERZ_MAX         (1 << 1)
+#	define R300_SC_HYPERZ_ADJ_256     (0 << 2)
+#	define R300_SC_HYPERZ_ADJ_128     (1 << 2)
+#	define R300_SC_HYPERZ_ADJ_64      (2 << 2)
+#	define R300_SC_HYPERZ_ADJ_32      (3 << 2)
+#	define R300_SC_HYPERZ_ADJ_16      (4 << 2)
+#	define R300_SC_HYPERZ_ADJ_8       (5 << 2)
+#	define R300_SC_HYPERZ_ADJ_4       (6 << 2)
+#	define R300_SC_HYPERZ_ADJ_2       (7 << 2)
+#	define R300_SC_HYPERZ_HZ_Z0MIN_NO (0 << 5)
+#	define R300_SC_HYPERZ_HZ_Z0MIN    (1 << 5)
+#	define R300_SC_HYPERZ_HZ_Z0MAX_NO (0 << 6)
+#	define R300_SC_HYPERZ_HZ_Z0MAX    (1 << 6)
+
+#define R300_SC_EDGERULE                 0x43a8
+
 /* BEGIN: Scissors and cliprects */
 
 /* There are four clipping rectangles. Their corner coordinates are inclusive.
@@ -744,21 +1284,21 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * For some reason, the top-left corner of the framebuffer is at (1440, 1440)
  * for the purpose of clipping and scissors.
  */
-#define R300_RE_CLIPRECT_TL_0               0x43B0
-#define R300_RE_CLIPRECT_BR_0               0x43B4
-#define R300_RE_CLIPRECT_TL_1               0x43B8
-#define R300_RE_CLIPRECT_BR_1               0x43BC
-#define R300_RE_CLIPRECT_TL_2               0x43C0
-#define R300_RE_CLIPRECT_BR_2               0x43C4
-#define R300_RE_CLIPRECT_TL_3               0x43C8
-#define R300_RE_CLIPRECT_BR_3               0x43CC
+#define R300_SC_CLIPRECT_TL_0               0x43B0
+#define R300_SC_CLIPRECT_BR_0               0x43B4
+#define R300_SC_CLIPRECT_TL_1               0x43B8
+#define R300_SC_CLIPRECT_BR_1               0x43BC
+#define R300_SC_CLIPRECT_TL_2               0x43C0
+#define R300_SC_CLIPRECT_BR_2               0x43C4
+#define R300_SC_CLIPRECT_TL_3               0x43C8
+#define R300_SC_CLIPRECT_BR_3               0x43CC
 #       define R300_CLIPRECT_OFFSET              1440
 #       define R300_CLIPRECT_MASK                0x1FFF
 #       define R300_CLIPRECT_X_SHIFT             0
 #       define R300_CLIPRECT_X_MASK              (0x1FFF << 0)
 #       define R300_CLIPRECT_Y_SHIFT             13
 #       define R300_CLIPRECT_Y_MASK              (0x1FFF << 13)
-#define R300_RE_CLIPRECT_CNTL               0x43D0
+#define R300_SC_CLIP_RULE                   0x43D0
 #       define R300_CLIP_OUT                     (1 << 0)
 #       define R300_CLIP_0                       (1 << 1)
 #       define R300_CLIP_1                       (1 << 2)
@@ -778,13 +1318,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* gap */
 
-#define R300_RE_SCISSORS_TL                 0x43E0
-#define R300_RE_SCISSORS_BR                 0x43E4
+#define R300_SC_SCISSORS_TL                 0x43E0
+#define R300_SC_SCISSORS_BR                 0x43E4
 #       define R300_SCISSORS_OFFSET              1440
 #       define R300_SCISSORS_X_SHIFT             0
 #       define R300_SCISSORS_X_MASK              (0x1FFF << 0)
 #       define R300_SCISSORS_Y_SHIFT             13
 #       define R300_SCISSORS_Y_MASK              (0x1FFF << 13)
+
+/* Screen door sample mask */
+#define R300_SC_SCREENDOOR                 0x43e8
+
 /* END: Scissors and cliprects */
 
 /* BEGIN: Texture specification */
@@ -794,43 +1338,55 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * unit. This means that e.g. the offset for texture image unit N is found in
  * register TX_OFFSET_0 + (4*N)
  */
-#define R300_TX_FILTER_0                    0x4400
+#define R300_TX_FILTER0_0                        0x4400
+#define R300_TX_FILTER0_1                        0x4404
+#define R300_TX_FILTER0_2                        0x4408
+#define R300_TX_FILTER0_3                        0x440c
+#define R300_TX_FILTER0_4                        0x4410
+#define R300_TX_FILTER0_5                        0x4414
+#define R300_TX_FILTER0_6                        0x4418
+#define R300_TX_FILTER0_7                        0x441c
+#define R300_TX_FILTER0_8                        0x4420
+#define R300_TX_FILTER0_9                        0x4424
+#define R300_TX_FILTER0_10                       0x4428
+#define R300_TX_FILTER0_11                       0x442c
+#define R300_TX_FILTER0_12                       0x4430
+#define R300_TX_FILTER0_13                       0x4434
+#define R300_TX_FILTER0_14                       0x4438
+#define R300_TX_FILTER0_15                       0x443c
 #       define R300_TX_REPEAT                    0
 #       define R300_TX_MIRRORED                  1
-#       define R300_TX_CLAMP                     4
 #       define R300_TX_CLAMP_TO_EDGE             2
+#	define R300_TX_MIRROR_ONCE_TO_EDGE       3
+#       define R300_TX_CLAMP                     4
+#	define R300_TX_MIRROR_ONCE               5
 #       define R300_TX_CLAMP_TO_BORDER           6
+#	define R300_TX_MIRROR_ONCE_TO_BORDER     7
 #       define R300_TX_WRAP_S_SHIFT              0
 #       define R300_TX_WRAP_S_MASK               (7 << 0)
 #       define R300_TX_WRAP_T_SHIFT              3
 #       define R300_TX_WRAP_T_MASK               (7 << 3)
-#       define R300_TX_WRAP_Q_SHIFT              6
-#       define R300_TX_WRAP_Q_MASK               (7 << 6)
+#       define R300_TX_WRAP_R_SHIFT              6
+#       define R300_TX_WRAP_R_MASK               (7 << 6)
+#	define R300_TX_MAG_FILTER_4              (0 << 9)
 #       define R300_TX_MAG_FILTER_NEAREST        (1 << 9)
 #       define R300_TX_MAG_FILTER_LINEAR         (2 << 9)
+#       define R300_TX_MAG_FILTER_ANISO          (3 << 9)
 #       define R300_TX_MAG_FILTER_MASK           (3 << 9)
 #       define R300_TX_MIN_FILTER_NEAREST        (1 << 11)
 #       define R300_TX_MIN_FILTER_LINEAR         (2 << 11)
-#	define R300_TX_MIN_FILTER_NEAREST_MIP_NEAREST       (5  <<  11)
-#	define R300_TX_MIN_FILTER_NEAREST_MIP_LINEAR        (9  <<  11)
-#	define R300_TX_MIN_FILTER_LINEAR_MIP_NEAREST        (6  <<  11)
-#	define R300_TX_MIN_FILTER_LINEAR_MIP_LINEAR         (10 <<  11)
-
-/* NOTE: NEAREST doesnt seem to exist.
- * Im not seting MAG_FILTER_MASK and (3 << 11) on for all
- * anisotropy modes because that would void selected mag filter
- */
-#	define R300_TX_MIN_FILTER_ANISO_NEAREST             (0 << 13)
-#	define R300_TX_MIN_FILTER_ANISO_LINEAR              (0 << 13)
-#	define R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (1 << 13)
-#	define R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR  (2 << 13)
-#       define R300_TX_MIN_FILTER_MASK   ( (15 << 11) | (3 << 13) )
-#	define R300_TX_MAX_ANISO_1_TO_1  (0 << 21)
-#	define R300_TX_MAX_ANISO_2_TO_1  (2 << 21)
-#	define R300_TX_MAX_ANISO_4_TO_1  (4 << 21)
-#	define R300_TX_MAX_ANISO_8_TO_1  (6 << 21)
-#	define R300_TX_MAX_ANISO_16_TO_1 (8 << 21)
-#	define R300_TX_MAX_ANISO_MASK    (14 << 21)
+#	define R300_TX_MIN_FILTER_ANISO          (3 << 11)
+#	define R300_TX_MIN_FILTER_MASK           (3 << 11)
+#	define R300_TX_MIN_FILTER_MIP_NONE       (0 << 13)
+#	define R300_TX_MIN_FILTER_MIP_NEAREST    (1 << 13)
+#	define R300_TX_MIN_FILTER_MIP_LINEAR     (2 << 13)
+#	define R300_TX_MIN_FILTER_MIP_MASK       (3 << 13)
+#	define R300_TX_MAX_ANISO_1_TO_1          (0 << 21)
+#	define R300_TX_MAX_ANISO_2_TO_1          (1 << 21)
+#	define R300_TX_MAX_ANISO_4_TO_1          (2 << 21)
+#	define R300_TX_MAX_ANISO_8_TO_1          (3 << 21)
+#	define R300_TX_MAX_ANISO_16_TO_1         (4 << 21)
+#	define R300_TX_MAX_ANISO_MASK            (7 << 21)
 
 #define R300_TX_FILTER1_0                      0x4440
 #	define R300_CHROMA_KEY_MODE_DISABLE    0
@@ -838,7 +1394,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_CHROMA_KEY_BLEND           2
 #	define R300_MC_ROUND_NORMAL            (0<<2)
 #	define R300_MC_ROUND_MPEG4             (1<<2)
-#	define R300_LOD_BIAS_MASK	    0x1fff
+#	define R300_LOD_BIAS_SHIFT             3
+#	define R300_LOD_BIAS_MASK	       0x1ff8
 #	define R300_EDGE_ANISO_EDGE_DIAG       (0<<13)
 #	define R300_EDGE_ANISO_EDGE_ONLY       (1<<13)
 #	define R300_MC_COORD_TRUNCATE_DISABLE  (0<<14)
@@ -849,12 +1406,16 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_TRI_PERF_3_8            (3<<15)
 #	define R300_ANISO_THRESHOLD_MASK       (7<<17)
 
+#	define R500_MACRO_SWITCH               (1<<22)
+#	define R500_BORDER_FIX                 (1<<31)
+
 #define R300_TX_SIZE_0                      0x4480
 #       define R300_TX_WIDTHMASK_SHIFT           0
 #       define R300_TX_WIDTHMASK_MASK            (2047 << 0)
 #       define R300_TX_HEIGHTMASK_SHIFT          11
 #       define R300_TX_HEIGHTMASK_MASK           (2047 << 11)
-#       define R300_TX_UNK23                     (1 << 23)
+#	define R300_TX_DEPTHMASK_SHIFT		 22
+#	define R300_TX_DEPTHMASK_MASK		 (0xf << 22)
 #       define R300_TX_MAX_MIP_LEVEL_SHIFT       26
 #       define R300_TX_MAX_MIP_LEVEL_MASK        (0xf << 26)
 #       define R300_TX_SIZE_PROJECTED            (1<<30)
@@ -865,7 +1426,9 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 	   They are given meanings as R, G, B and Alpha by the swizzle
 	   specification */
 #	define R300_TX_FORMAT_X8		    0x0
+#	define R500_TX_FORMAT_X1		    0x0 // bit set in format 2
 #	define R300_TX_FORMAT_X16		    0x1
+#	define R500_TX_FORMAT_X1_REV		    0x0 // bit set in format 2
 #	define R300_TX_FORMAT_Y4X4		    0x2
 #	define R300_TX_FORMAT_Y8X8		    0x3
 #	define R300_TX_FORMAT_Y16X16		    0x4
@@ -886,9 +1449,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_FORMAT_A8R8G8B8	    	    0x13     /* no swizzle */
 #	define R300_TX_FORMAT_B8G8_B8G8	    	    0x14     /* no swizzle */
 #	define R300_TX_FORMAT_G8R8_G8B8	    	    0x15     /* no swizzle */
+
+	/* These two values are wrong, but they're the only values that
+	 * produce any even vaguely correct results.  Can r300 only do 16-bit
+	 * depth textures?
+	 */
+#	define R300_TX_FORMAT_X24_Y8	    	    0x1e
+#	define R300_TX_FORMAT_X32	    	    0x1e
+
 	/* 0x16 - some 16 bit green format.. ?? */
-#	define R300_TX_FORMAT_UNK25		   (1 << 25) /* no swizzle */
-#	define R300_TX_FORMAT_CUBIC_MAP		   (1 << 26)
+#	define R300_TX_FORMAT_3D		   (1 << 25)
+#	define R300_TX_FORMAT_CUBIC_MAP		   (2 << 25)
 
 	/* gap */
 	/* Floating point formats */
@@ -941,7 +1512,18 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #	define R300_TX_FORMAT_YUV_MODE		0x00800000
 
-#define R300_TX_PITCH_0			    0x4500 /* obvious missing in gap */
+#define R300_TX_FORMAT2_0		    0x4500 /* obvious missing in gap */
+#       define R300_TX_PITCHMASK_SHIFT           0
+#       define R300_TX_PITCHMASK_MASK            (2047 << 0)
+#	define R500_TXFORMAT_MSB		 (1 << 14)
+#	define R500_TXWIDTH_BIT11	         (1 << 15)
+#	define R500_TXHEIGHT_BIT11	         (1 << 16)
+#	define R500_POW2FIX2FLT			 (1 << 17)
+#	define R500_SEL_FILTER4_TC0		 (0 << 18)
+#	define R500_SEL_FILTER4_TC1		 (1 << 18)
+#	define R500_SEL_FILTER4_TC2		 (2 << 18)
+#	define R500_SEL_FILTER4_TC3		 (3 << 18)
+
 #define R300_TX_OFFSET_0                    0x4540
 	/* BEGIN: Guess from R200 */
 #       define R300_TXO_ENDIAN_NO_SWAP           (0 << 0)
@@ -949,15 +1531,50 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_TXO_ENDIAN_WORD_SWAP         (2 << 0)
 #       define R300_TXO_ENDIAN_HALFDW_SWAP       (3 << 0)
 #       define R300_TXO_MACRO_TILE               (1 << 2)
+#       define R300_TXO_MICRO_TILE_LINEAR        (0 << 3)
 #       define R300_TXO_MICRO_TILE               (1 << 3)
+#       define R300_TXO_MICRO_TILE_SQUARE        (2 << 3)
 #       define R300_TXO_OFFSET_MASK              0xffffffe0
 #       define R300_TXO_OFFSET_SHIFT             5
 	/* END: Guess from R200 */
 
 /* 32 bit chroma key */
 #define R300_TX_CHROMA_KEY_0                      0x4580
+#define R300_TX_CHROMA_KEY_1                      0x4584
+#define R300_TX_CHROMA_KEY_2                      0x4588
+#define R300_TX_CHROMA_KEY_3                      0x458c
+#define R300_TX_CHROMA_KEY_4                      0x4590
+#define R300_TX_CHROMA_KEY_5                      0x4594
+#define R300_TX_CHROMA_KEY_6                      0x4598
+#define R300_TX_CHROMA_KEY_7                      0x459c
+#define R300_TX_CHROMA_KEY_8                      0x45a0
+#define R300_TX_CHROMA_KEY_9                      0x45a4
+#define R300_TX_CHROMA_KEY_10                     0x45a8
+#define R300_TX_CHROMA_KEY_11                     0x45ac
+#define R300_TX_CHROMA_KEY_12                     0x45b0
+#define R300_TX_CHROMA_KEY_13                     0x45b4
+#define R300_TX_CHROMA_KEY_14                     0x45b8
+#define R300_TX_CHROMA_KEY_15                     0x45bc
 /* ff00ff00 == { 0, 1.0, 0, 1.0 } */
-#define R300_TX_BORDER_COLOR_0              0x45C0
+
+/* Border Color */
+#define R300_TX_BORDER_COLOR_0              0x45c0
+#define R300_TX_BORDER_COLOR_1              0x45c4
+#define R300_TX_BORDER_COLOR_2              0x45c8
+#define R300_TX_BORDER_COLOR_3              0x45cc
+#define R300_TX_BORDER_COLOR_4              0x45d0
+#define R300_TX_BORDER_COLOR_5              0x45d4
+#define R300_TX_BORDER_COLOR_6              0x45d8
+#define R300_TX_BORDER_COLOR_7              0x45dc
+#define R300_TX_BORDER_COLOR_8              0x45e0
+#define R300_TX_BORDER_COLOR_9              0x45e4
+#define R300_TX_BORDER_COLOR_10             0x45e8
+#define R300_TX_BORDER_COLOR_11             0x45ec
+#define R300_TX_BORDER_COLOR_12             0x45f0
+#define R300_TX_BORDER_COLOR_13             0x45f4
+#define R300_TX_BORDER_COLOR_14             0x45f8
+#define R300_TX_BORDER_COLOR_15             0x45fc
+
 
 /* END: Texture specification */
 
@@ -980,23 +1597,23 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * offsets into the respective instruction streams, while *_END points to the
  * last instruction relative to this offset.
  */
-#define R300_PFS_CNTL_0                     0x4600
+#define R300_US_CONFIG                      0x4600
 #       define R300_PFS_CNTL_LAST_NODES_SHIFT    0
 #       define R300_PFS_CNTL_LAST_NODES_MASK     (3 << 0)
 #       define R300_PFS_CNTL_FIRST_NODE_HAS_TEX  (1 << 3)
-#define R300_PFS_CNTL_1                     0x4604
+#define R300_US_PIXSIZE                     0x4604
 /* There is an unshifted value here which has so far always been equal to the
  * index of the highest used temporary register.
  */
-#define R300_PFS_CNTL_2                     0x4608
+#define R300_US_CODE_OFFSET                 0x4608
 #       define R300_PFS_CNTL_ALU_OFFSET_SHIFT    0
 #       define R300_PFS_CNTL_ALU_OFFSET_MASK     (63 << 0)
 #       define R300_PFS_CNTL_ALU_END_SHIFT       6
 #       define R300_PFS_CNTL_ALU_END_MASK        (63 << 6)
-#       define R300_PFS_CNTL_TEX_OFFSET_SHIFT    12
-#       define R300_PFS_CNTL_TEX_OFFSET_MASK     (31 << 12) /* GUESS */
+#       define R300_PFS_CNTL_TEX_OFFSET_SHIFT    13
+#       define R300_PFS_CNTL_TEX_OFFSET_MASK     (31 << 13)
 #       define R300_PFS_CNTL_TEX_END_SHIFT       18
-#       define R300_PFS_CNTL_TEX_END_MASK        (31 << 18) /* GUESS */
+#       define R300_PFS_CNTL_TEX_END_MASK        (31 << 18)
 
 /* gap */
 
@@ -1007,45 +1624,65 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  * Offsets are relative to the master offset from PFS_CNTL_2.
  */
-#define R300_PFS_NODE_0                     0x4610
-#define R300_PFS_NODE_1                     0x4614
-#define R300_PFS_NODE_2                     0x4618
-#define R300_PFS_NODE_3                     0x461C
-#       define R300_PFS_NODE_ALU_OFFSET_SHIFT    0
-#       define R300_PFS_NODE_ALU_OFFSET_MASK     (63 << 0)
-#       define R300_PFS_NODE_ALU_END_SHIFT       6
-#       define R300_PFS_NODE_ALU_END_MASK        (63 << 6)
-#       define R300_PFS_NODE_TEX_OFFSET_SHIFT    12
-#       define R300_PFS_NODE_TEX_OFFSET_MASK     (31 << 12)
-#       define R300_PFS_NODE_TEX_END_SHIFT       17
-#       define R300_PFS_NODE_TEX_END_MASK        (31 << 17)
-#		define R300_PFS_NODE_OUTPUT_COLOR        (1 << 22)
-#		define R300_PFS_NODE_OUTPUT_DEPTH        (1 << 23)
+#define R300_US_CODE_ADDR_0                 0x4610
+#define R300_US_CODE_ADDR_1                 0x4614
+#define R300_US_CODE_ADDR_2                 0x4618
+#define R300_US_CODE_ADDR_3                 0x461C
+#       define R300_ALU_START_SHIFT         0
+#       define R300_ALU_START_MASK          (63 << 0)
+#       define R300_ALU_SIZE_SHIFT          6
+#       define R300_ALU_SIZE_MASK           (63 << 6)
+#       define R300_TEX_START_SHIFT         12
+#       define R300_TEX_START_MASK          (31 << 12)
+#       define R300_TEX_SIZE_SHIFT          17
+#       define R300_TEX_SIZE_MASK           (31 << 17)
+#	define R300_RGBA_OUT                (1 << 22)
+#	define R300_W_OUT                   (1 << 23)
 
 /* TEX
  * As far as I can tell, texture instructions cannot write into output
  * registers directly. A subsequent ALU instruction is always necessary,
  * even if it's just MAD o0, r0, 1, 0
  */
-#define R300_PFS_TEXI_0                     0x4620
-#	define R300_FPITX_SRC_SHIFT              0
-#	define R300_FPITX_SRC_MASK               (31 << 0)
-	/* GUESS */
-#	define R300_FPITX_SRC_CONST              (1 << 5)
-#	define R300_FPITX_DST_SHIFT              6
-#	define R300_FPITX_DST_MASK               (31 << 6)
-#	define R300_FPITX_IMAGE_SHIFT            11
-	/* GUESS based on layout and native limits */
-#       define R300_FPITX_IMAGE_MASK             (15 << 11)
-/* Unsure if these are opcodes, or some kind of bitfield, but this is how
- * they were set when I checked
- */
-#	define R300_FPITX_OPCODE_SHIFT		15
-#		define R300_FPITX_OP_TEX	1
-#		define R300_FPITX_OP_KIL	2
-#		define R300_FPITX_OP_TXP	3
-#		define R300_FPITX_OP_TXB	4
-#	define R300_FPITX_OPCODE_MASK           (7 << 15)
+#define R300_US_TEX_INST_0                  0x4620
+#	define R300_SRC_ADDR_SHIFT          0
+#	define R300_SRC_ADDR_MASK           (31 << 0)
+#	define R300_DST_ADDR_SHIFT          6
+#	define R300_DST_ADDR_MASK           (31 << 6)
+#	define R300_TEX_ID_SHIFT            11
+#       define R300_TEX_ID_MASK             (15 << 11)
+#	define R300_TEX_INST_SHIFT		15
+#		define R300_TEX_OP_NOP	        0
+#		define R300_TEX_OP_LD	        1
+#		define R300_TEX_OP_KIL	        2
+#		define R300_TEX_OP_TXP	        3
+#		define R300_TEX_OP_TXB	        4
+#	define R300_TEX_INST_MASK               (7 << 15)
+
+/* Output format from the unfied shader */
+#define R300_US_OUT_FMT                     0x46A4
+#	define R300_US_OUT_FMT_C4_8         (0 << 0)
+#	define R300_US_OUT_FMT_C4_10        (1 << 0)
+#	define R300_US_OUT_FMT_C4_10_GAMMA  (2 << 0)
+#	define R300_US_OUT_FMT_C_16         (3 << 0)
+#	define R300_US_OUT_FMT_C2_16        (4 << 0)
+#	define R300_US_OUT_FMT_C4_16        (5 << 0)
+#	define R300_US_OUT_FMT_C_16_MPEG    (6 << 0)
+#	define R300_US_OUT_FMT_C2_16_MPEG   (7 << 0)
+#	define R300_US_OUT_FMT_C2_4         (8 << 0)
+#	define R300_US_OUT_FMT_C_3_3_2      (9 << 0)
+#	define R300_US_OUT_FMT_C_6_5_6      (10 << 0)
+#	define R300_US_OUT_FMT_C_11_11_10   (11 << 0)
+#	define R300_US_OUT_FMT_C_10_11_11   (12 << 0)
+#	define R300_US_OUT_FMT_C_2_10_10_10 (13 << 0)
+/* reserved */
+#	define R300_US_OUT_FMT_UNUSED       (15 << 0)
+#	define R300_US_OUT_FMT_C_16_FP      (16 << 0)
+#	define R300_US_OUT_FMT_C2_16_FP     (17 << 0)
+#	define R300_US_OUT_FMT_C4_16_FP     (18 << 0)
+#	define R300_US_OUT_FMT_C_32_FP      (19 << 0)
+#	define R300_US_OUT_FMT_C2_32_FP     (20 << 0)
+#	define R300_US_OUT_FMT_C4_32_FP     (20 << 0)
 
 /* ALU
  * The ALU instructions register blocks are enumerated according to the order
@@ -1111,172 +1748,256 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *  - Set FPI0/FPI2_SPECIAL_LRP
  * Arbitrary LRP (including support for swizzling) requires vanilla MAD+MAD
  */
-#define R300_PFS_INSTR1_0                   0x46C0
-#       define R300_FPI1_SRC0C_SHIFT             0
-#       define R300_FPI1_SRC0C_MASK              (31 << 0)
-#       define R300_FPI1_SRC0C_CONST             (1 << 5)
-#       define R300_FPI1_SRC1C_SHIFT             6
-#       define R300_FPI1_SRC1C_MASK              (31 << 6)
-#       define R300_FPI1_SRC1C_CONST             (1 << 11)
-#       define R300_FPI1_SRC2C_SHIFT             12
-#       define R300_FPI1_SRC2C_MASK              (31 << 12)
-#       define R300_FPI1_SRC2C_CONST             (1 << 17)
-#       define R300_FPI1_SRC_MASK                0x0003ffff
-#       define R300_FPI1_DSTC_SHIFT              18
-#       define R300_FPI1_DSTC_MASK               (31 << 18)
-#		define R300_FPI1_DSTC_REG_MASK_SHIFT     23
-#       define R300_FPI1_DSTC_REG_X              (1 << 23)
-#       define R300_FPI1_DSTC_REG_Y              (1 << 24)
-#       define R300_FPI1_DSTC_REG_Z              (1 << 25)
-#		define R300_FPI1_DSTC_OUTPUT_MASK_SHIFT  26
-#       define R300_FPI1_DSTC_OUTPUT_X           (1 << 26)
-#       define R300_FPI1_DSTC_OUTPUT_Y           (1 << 27)
-#       define R300_FPI1_DSTC_OUTPUT_Z           (1 << 28)
-
-#define R300_PFS_INSTR3_0                   0x47C0
-#       define R300_FPI3_SRC0A_SHIFT             0
-#       define R300_FPI3_SRC0A_MASK              (31 << 0)
-#       define R300_FPI3_SRC0A_CONST             (1 << 5)
-#       define R300_FPI3_SRC1A_SHIFT             6
-#       define R300_FPI3_SRC1A_MASK              (31 << 6)
-#       define R300_FPI3_SRC1A_CONST             (1 << 11)
-#       define R300_FPI3_SRC2A_SHIFT             12
-#       define R300_FPI3_SRC2A_MASK              (31 << 12)
-#       define R300_FPI3_SRC2A_CONST             (1 << 17)
-#       define R300_FPI3_SRC_MASK                0x0003ffff
-#       define R300_FPI3_DSTA_SHIFT              18
-#       define R300_FPI3_DSTA_MASK               (31 << 18)
-#       define R300_FPI3_DSTA_REG                (1 << 23)
-#       define R300_FPI3_DSTA_OUTPUT             (1 << 24)
-#		define R300_FPI3_DSTA_DEPTH              (1 << 27)
-
-#define R300_PFS_INSTR0_0                   0x48C0
-#       define R300_FPI0_ARGC_SRC0C_XYZ          0
-#       define R300_FPI0_ARGC_SRC0C_XXX          1
-#       define R300_FPI0_ARGC_SRC0C_YYY          2
-#       define R300_FPI0_ARGC_SRC0C_ZZZ          3
-#       define R300_FPI0_ARGC_SRC1C_XYZ          4
-#       define R300_FPI0_ARGC_SRC1C_XXX          5
-#       define R300_FPI0_ARGC_SRC1C_YYY          6
-#       define R300_FPI0_ARGC_SRC1C_ZZZ          7
-#       define R300_FPI0_ARGC_SRC2C_XYZ          8
-#       define R300_FPI0_ARGC_SRC2C_XXX          9
-#       define R300_FPI0_ARGC_SRC2C_YYY          10
-#       define R300_FPI0_ARGC_SRC2C_ZZZ          11
-#       define R300_FPI0_ARGC_SRC0A              12
-#       define R300_FPI0_ARGC_SRC1A              13
-#       define R300_FPI0_ARGC_SRC2A              14
-#       define R300_FPI0_ARGC_SRC1C_LRP          15
-#       define R300_FPI0_ARGC_ZERO               20
-#       define R300_FPI0_ARGC_ONE                21
-	/* GUESS */
-#       define R300_FPI0_ARGC_HALF               22
-#       define R300_FPI0_ARGC_SRC0C_YZX          23
-#       define R300_FPI0_ARGC_SRC1C_YZX          24
-#       define R300_FPI0_ARGC_SRC2C_YZX          25
-#       define R300_FPI0_ARGC_SRC0C_ZXY          26
-#       define R300_FPI0_ARGC_SRC1C_ZXY          27
-#       define R300_FPI0_ARGC_SRC2C_ZXY          28
-#       define R300_FPI0_ARGC_SRC0CA_WZY         29
-#       define R300_FPI0_ARGC_SRC1CA_WZY         30
-#       define R300_FPI0_ARGC_SRC2CA_WZY         31
-
-#       define R300_FPI0_ARG0C_SHIFT             0
-#       define R300_FPI0_ARG0C_MASK              (31 << 0)
-#       define R300_FPI0_ARG0C_NEG               (1 << 5)
-#       define R300_FPI0_ARG0C_ABS               (1 << 6)
-#       define R300_FPI0_ARG1C_SHIFT             7
-#       define R300_FPI0_ARG1C_MASK              (31 << 7)
-#       define R300_FPI0_ARG1C_NEG               (1 << 12)
-#       define R300_FPI0_ARG1C_ABS               (1 << 13)
-#       define R300_FPI0_ARG2C_SHIFT             14
-#       define R300_FPI0_ARG2C_MASK              (31 << 14)
-#       define R300_FPI0_ARG2C_NEG               (1 << 19)
-#       define R300_FPI0_ARG2C_ABS               (1 << 20)
-#       define R300_FPI0_SPECIAL_LRP             (1 << 21)
-#       define R300_FPI0_OUTC_MAD                (0 << 23)
-#       define R300_FPI0_OUTC_DP3                (1 << 23)
-#       define R300_FPI0_OUTC_DP4                (2 << 23)
-#       define R300_FPI0_OUTC_MIN                (4 << 23)
-#       define R300_FPI0_OUTC_MAX                (5 << 23)
-#       define R300_FPI0_OUTC_CMPH               (7 << 23)
-#       define R300_FPI0_OUTC_CMP                (8 << 23)
-#       define R300_FPI0_OUTC_FRC                (9 << 23)
-#       define R300_FPI0_OUTC_REPL_ALPHA         (10 << 23)
-#       define R300_FPI0_OUTC_SAT                (1 << 30)
-#       define R300_FPI0_INSERT_NOP              (1 << 31)
-
-#define R300_PFS_INSTR2_0                   0x49C0
-#       define R300_FPI2_ARGA_SRC0C_X            0
-#       define R300_FPI2_ARGA_SRC0C_Y            1
-#       define R300_FPI2_ARGA_SRC0C_Z            2
-#       define R300_FPI2_ARGA_SRC1C_X            3
-#       define R300_FPI2_ARGA_SRC1C_Y            4
-#       define R300_FPI2_ARGA_SRC1C_Z            5
-#       define R300_FPI2_ARGA_SRC2C_X            6
-#       define R300_FPI2_ARGA_SRC2C_Y            7
-#       define R300_FPI2_ARGA_SRC2C_Z            8
-#       define R300_FPI2_ARGA_SRC0A              9
-#       define R300_FPI2_ARGA_SRC1A              10
-#       define R300_FPI2_ARGA_SRC2A              11
-#       define R300_FPI2_ARGA_SRC1A_LRP          15
-#       define R300_FPI2_ARGA_ZERO               16
-#       define R300_FPI2_ARGA_ONE                17
-	/* GUESS */
-#       define R300_FPI2_ARGA_HALF               18
-#       define R300_FPI2_ARG0A_SHIFT             0
-#       define R300_FPI2_ARG0A_MASK              (31 << 0)
-#       define R300_FPI2_ARG0A_NEG               (1 << 5)
-	/* GUESS */
-#	define R300_FPI2_ARG0A_ABS		 (1 << 6)
-#       define R300_FPI2_ARG1A_SHIFT             7
-#       define R300_FPI2_ARG1A_MASK              (31 << 7)
-#       define R300_FPI2_ARG1A_NEG               (1 << 12)
-	/* GUESS */
-#	define R300_FPI2_ARG1A_ABS		 (1 << 13)
-#       define R300_FPI2_ARG2A_SHIFT             14
-#       define R300_FPI2_ARG2A_MASK              (31 << 14)
-#       define R300_FPI2_ARG2A_NEG               (1 << 19)
-	/* GUESS */
-#	define R300_FPI2_ARG2A_ABS		 (1 << 20)
-#       define R300_FPI2_SPECIAL_LRP             (1 << 21)
-#       define R300_FPI2_OUTA_MAD                (0 << 23)
-#       define R300_FPI2_OUTA_DP4                (1 << 23)
-#       define R300_FPI2_OUTA_MIN                (2 << 23)
-#       define R300_FPI2_OUTA_MAX                (3 << 23)
-#       define R300_FPI2_OUTA_CMP                (6 << 23)
-#       define R300_FPI2_OUTA_FRC                (7 << 23)
-#       define R300_FPI2_OUTA_EX2                (8 << 23)
-#       define R300_FPI2_OUTA_LG2                (9 << 23)
-#       define R300_FPI2_OUTA_RCP                (10 << 23)
-#       define R300_FPI2_OUTA_RSQ                (11 << 23)
-#       define R300_FPI2_OUTA_SAT                (1 << 30)
-#       define R300_FPI2_UNKNOWN_31              (1 << 31)
+#define R300_US_ALU_RGB_ADDR_0                   0x46C0
+#       define R300_ALU_SRC0C_SHIFT             0
+#       define R300_ALU_SRC0C_MASK              (31 << 0)
+#       define R300_ALU_SRC0C_CONST             (1 << 5)
+#       define R300_ALU_SRC1C_SHIFT             6
+#       define R300_ALU_SRC1C_MASK              (31 << 6)
+#       define R300_ALU_SRC1C_CONST             (1 << 11)
+#       define R300_ALU_SRC2C_SHIFT             12
+#       define R300_ALU_SRC2C_MASK              (31 << 12)
+#       define R300_ALU_SRC2C_CONST             (1 << 17)
+#       define R300_ALU_SRC_MASK                0x0003ffff
+#       define R300_ALU_DSTC_SHIFT              18
+#       define R300_ALU_DSTC_MASK               (31 << 18)
+#		define R300_ALU_DSTC_REG_MASK_SHIFT     23
+#       define R300_ALU_DSTC_REG_X              (1 << 23)
+#       define R300_ALU_DSTC_REG_Y              (1 << 24)
+#       define R300_ALU_DSTC_REG_Z              (1 << 25)
+#		define R300_ALU_DSTC_OUTPUT_MASK_SHIFT  26
+#       define R300_ALU_DSTC_OUTPUT_X           (1 << 26)
+#       define R300_ALU_DSTC_OUTPUT_Y           (1 << 27)
+#       define R300_ALU_DSTC_OUTPUT_Z           (1 << 28)
+
+#define R300_US_ALU_ALPHA_ADDR_0                 0x47C0
+#       define R300_ALU_SRC0A_SHIFT             0
+#       define R300_ALU_SRC0A_MASK              (31 << 0)
+#       define R300_ALU_SRC0A_CONST             (1 << 5)
+#       define R300_ALU_SRC1A_SHIFT             6
+#       define R300_ALU_SRC1A_MASK              (31 << 6)
+#       define R300_ALU_SRC1A_CONST             (1 << 11)
+#       define R300_ALU_SRC2A_SHIFT             12
+#       define R300_ALU_SRC2A_MASK              (31 << 12)
+#       define R300_ALU_SRC2A_CONST             (1 << 17)
+#       define R300_ALU_SRC_MASK                0x0003ffff
+#       define R300_ALU_DSTA_SHIFT              18
+#       define R300_ALU_DSTA_MASK               (31 << 18)
+#       define R300_ALU_DSTA_REG                (1 << 23)
+#       define R300_ALU_DSTA_OUTPUT             (1 << 24)
+#		define R300_ALU_DSTA_DEPTH              (1 << 27)
+
+#define R300_US_ALU_RGB_INST_0                   0x48C0
+#       define R300_ALU_ARGC_SRC0C_XYZ          0
+#       define R300_ALU_ARGC_SRC0C_XXX          1
+#       define R300_ALU_ARGC_SRC0C_YYY          2
+#       define R300_ALU_ARGC_SRC0C_ZZZ          3
+#       define R300_ALU_ARGC_SRC1C_XYZ          4
+#       define R300_ALU_ARGC_SRC1C_XXX          5
+#       define R300_ALU_ARGC_SRC1C_YYY          6
+#       define R300_ALU_ARGC_SRC1C_ZZZ          7
+#       define R300_ALU_ARGC_SRC2C_XYZ          8
+#       define R300_ALU_ARGC_SRC2C_XXX          9
+#       define R300_ALU_ARGC_SRC2C_YYY          10
+#       define R300_ALU_ARGC_SRC2C_ZZZ          11
+#       define R300_ALU_ARGC_SRC0A              12
+#       define R300_ALU_ARGC_SRC1A              13
+#       define R300_ALU_ARGC_SRC2A              14
+#       define R300_ALU_ARGC_SRCP_XYZ           15
+#       define R300_ALU_ARGC_SRCP_XXX           16
+#       define R300_ALU_ARGC_SRCP_YYY           17
+#       define R300_ALU_ARGC_SRCP_ZZZ           18
+#       define R300_ALU_ARGC_SRCP_WWW           19
+#       define R300_ALU_ARGC_ZERO               20
+#       define R300_ALU_ARGC_ONE                21
+#       define R300_ALU_ARGC_HALF               22
+#       define R300_ALU_ARGC_SRC0C_YZX          23
+#       define R300_ALU_ARGC_SRC1C_YZX          24
+#       define R300_ALU_ARGC_SRC2C_YZX          25
+#       define R300_ALU_ARGC_SRC0C_ZXY          26
+#       define R300_ALU_ARGC_SRC1C_ZXY          27
+#       define R300_ALU_ARGC_SRC2C_ZXY          28
+#       define R300_ALU_ARGC_SRC0CA_WZY         29
+#       define R300_ALU_ARGC_SRC1CA_WZY         30
+#       define R300_ALU_ARGC_SRC2CA_WZY         31
+
+#       define R300_ALU_ARG0C_SHIFT             0
+#       define R300_ALU_ARG0C_MASK              (31 << 0)
+#       define R300_ALU_ARG0C_NOP               (0 << 5)
+#       define R300_ALU_ARG0C_NEG               (1 << 5)
+#       define R300_ALU_ARG0C_ABS               (2 << 5)
+#       define R300_ALU_ARG0C_NAB               (3 << 5)
+#       define R300_ALU_ARG1C_SHIFT             7
+#       define R300_ALU_ARG1C_MASK              (31 << 7)
+#       define R300_ALU_ARG1C_NOP               (0 << 12)
+#       define R300_ALU_ARG1C_NEG               (1 << 12)
+#       define R300_ALU_ARG1C_ABS               (2 << 12)
+#       define R300_ALU_ARG1C_NAB               (3 << 12)
+#       define R300_ALU_ARG2C_SHIFT             14
+#       define R300_ALU_ARG2C_MASK              (31 << 14)
+#       define R300_ALU_ARG2C_NOP               (0 << 19)
+#       define R300_ALU_ARG2C_NEG               (1 << 19)
+#       define R300_ALU_ARG2C_ABS               (2 << 19)
+#       define R300_ALU_ARG2C_NAB               (3 << 19)
+#       define R300_ALU_SRCP_1_MINUS_2_SRC0     (0 << 21)
+#       define R300_ALU_SRCP_SRC1_MINUS_SRC0    (1 << 21)
+#       define R300_ALU_SRCP_SRC1_PLUS_SRC0     (2 << 21)
+#       define R300_ALU_SRCP_1_MINUS_SRC0       (3 << 21)
+
+#       define R300_ALU_OUTC_MAD                (0 << 23)
+#       define R300_ALU_OUTC_DP3                (1 << 23)
+#       define R300_ALU_OUTC_DP4                (2 << 23)
+#       define R300_ALU_OUTC_D2A                (3 << 23)
+#       define R300_ALU_OUTC_MIN                (4 << 23)
+#       define R300_ALU_OUTC_MAX                (5 << 23)
+#       define R300_ALU_OUTC_CMPH               (7 << 23)
+#       define R300_ALU_OUTC_CMP                (8 << 23)
+#       define R300_ALU_OUTC_FRC                (9 << 23)
+#       define R300_ALU_OUTC_REPL_ALPHA         (10 << 23)
+
+#       define R300_ALU_OUTC_MOD_NOP            (0 << 27)
+#       define R300_ALU_OUTC_MOD_MUL2           (1 << 27)
+#       define R300_ALU_OUTC_MOD_MUL4           (2 << 27)
+#       define R300_ALU_OUTC_MOD_MUL8           (3 << 27)
+#       define R300_ALU_OUTC_MOD_DIV2           (4 << 27)
+#       define R300_ALU_OUTC_MOD_DIV4           (5 << 27)
+#       define R300_ALU_OUTC_MOD_DIV8           (6 << 27)
+
+#       define R300_ALU_OUTC_CLAMP              (1 << 30)
+#       define R300_ALU_INSERT_NOP              (1 << 31)
+
+#define R300_US_ALU_ALPHA_INST_0                 0x49C0
+#       define R300_ALU_ARGA_SRC0C_X            0
+#       define R300_ALU_ARGA_SRC0C_Y            1
+#       define R300_ALU_ARGA_SRC0C_Z            2
+#       define R300_ALU_ARGA_SRC1C_X            3
+#       define R300_ALU_ARGA_SRC1C_Y            4
+#       define R300_ALU_ARGA_SRC1C_Z            5
+#       define R300_ALU_ARGA_SRC2C_X            6
+#       define R300_ALU_ARGA_SRC2C_Y            7
+#       define R300_ALU_ARGA_SRC2C_Z            8
+#       define R300_ALU_ARGA_SRC0A              9
+#       define R300_ALU_ARGA_SRC1A              10
+#       define R300_ALU_ARGA_SRC2A              11
+#       define R300_ALU_ARGA_SRCP_X             12
+#       define R300_ALU_ARGA_SRCP_Y             13
+#       define R300_ALU_ARGA_SRCP_Z             14
+#       define R300_ALU_ARGA_SRCP_W             15
+
+#       define R300_ALU_ARGA_ZERO               16
+#       define R300_ALU_ARGA_ONE                17
+#       define R300_ALU_ARGA_HALF               18
+#       define R300_ALU_ARG0A_SHIFT             0
+#       define R300_ALU_ARG0A_MASK              (31 << 0)
+#       define R300_ALU_ARG0A_NOP               (0 << 5)
+#       define R300_ALU_ARG0A_NEG               (1 << 5)
+#	define R300_ALU_ARG0A_ABS		 (2 << 5)
+#	define R300_ALU_ARG0A_NAB		 (3 << 5)
+#       define R300_ALU_ARG1A_SHIFT             7
+#       define R300_ALU_ARG1A_MASK              (31 << 7)
+#       define R300_ALU_ARG1A_NOP               (0 << 12)
+#       define R300_ALU_ARG1A_NEG               (1 << 12)
+#	define R300_ALU_ARG1A_ABS		 (2 << 12)
+#	define R300_ALU_ARG1A_NAB		 (3 << 12)
+#       define R300_ALU_ARG2A_SHIFT             14
+#       define R300_ALU_ARG2A_MASK              (31 << 14)
+#       define R300_ALU_ARG2A_NOP               (0 << 19)
+#       define R300_ALU_ARG2A_NEG               (1 << 19)
+#	define R300_ALU_ARG2A_ABS		 (2 << 19)
+#	define R300_ALU_ARG2A_NAB		 (3 << 19)
+#       define R300_ALU_SRCP_1_MINUS_2_SRC0     (0 << 21)
+#       define R300_ALU_SRCP_SRC1_MINUS_SRC0    (1 << 21)
+#       define R300_ALU_SRCP_SRC1_PLUS_SRC0     (2 << 21)
+#       define R300_ALU_SRCP_1_MINUS_SRC0       (3 << 21)
+
+#       define R300_ALU_OUTA_MAD                (0 << 23)
+#       define R300_ALU_OUTA_DP4                (1 << 23)
+#       define R300_ALU_OUTA_MIN                (2 << 23)
+#       define R300_ALU_OUTA_MAX                (3 << 23)
+#       define R300_ALU_OUTA_CND                (5 << 23)
+#       define R300_ALU_OUTA_CMP                (6 << 23)
+#       define R300_ALU_OUTA_FRC                (7 << 23)
+#       define R300_ALU_OUTA_EX2                (8 << 23)
+#       define R300_ALU_OUTA_LG2                (9 << 23)
+#       define R300_ALU_OUTA_RCP                (10 << 23)
+#       define R300_ALU_OUTA_RSQ                (11 << 23)
+
+#       define R300_ALU_OUTA_MOD_NOP            (0 << 27)
+#       define R300_ALU_OUTA_MOD_MUL2           (1 << 27)
+#       define R300_ALU_OUTA_MOD_MUL4           (2 << 27)
+#       define R300_ALU_OUTA_MOD_MUL8           (3 << 27)
+#       define R300_ALU_OUTA_MOD_DIV2           (4 << 27)
+#       define R300_ALU_OUTA_MOD_DIV4           (5 << 27)
+#       define R300_ALU_OUTA_MOD_DIV8           (6 << 27)
+
+#       define R300_ALU_OUTA_CLAMP              (1 << 30)
 /* END: Fragment program instruction set */
 
-/* Fog state and color */
-#define R300_RE_FOG_STATE                   0x4BC0
-#       define R300_FOG_ENABLE                   (1 << 0)
-#	define R300_FOG_MODE_LINEAR              (0 << 1)
-#	define R300_FOG_MODE_EXP                 (1 << 1)
-#	define R300_FOG_MODE_EXP2                (2 << 1)
-#	define R300_FOG_MODE_MASK                (3 << 1)
-#define R300_FOG_COLOR_R                    0x4BC8
-#define R300_FOG_COLOR_G                    0x4BCC
-#define R300_FOG_COLOR_B                    0x4BD0
-
-#define R300_PP_ALPHA_TEST                  0x4BD4
-#       define R300_REF_ALPHA_MASK               0x000000ff
-#       define R300_ALPHA_TEST_FAIL              (0 << 8)
-#       define R300_ALPHA_TEST_LESS              (1 << 8)
-#       define R300_ALPHA_TEST_LEQUAL            (3 << 8)
-#       define R300_ALPHA_TEST_EQUAL             (2 << 8)
-#       define R300_ALPHA_TEST_GEQUAL            (6 << 8)
-#       define R300_ALPHA_TEST_GREATER           (4 << 8)
-#       define R300_ALPHA_TEST_NEQUAL            (5 << 8)
-#       define R300_ALPHA_TEST_PASS              (7 << 8)
-#       define R300_ALPHA_TEST_OP_MASK           (7 << 8)
-#       define R300_ALPHA_TEST_ENABLE            (1 << 11)
+/* Fog: Fog Blending Enable */
+#define R300_FG_FOG_BLEND                             0x4bc0
+#       define R300_FG_FOG_BLEND_DISABLE              (0 << 0)
+#       define R300_FG_FOG_BLEND_ENABLE               (1 << 0)
+#	define R300_FG_FOG_BLEND_FN_LINEAR            (0 << 1)
+#	define R300_FG_FOG_BLEND_FN_EXP               (1 << 1)
+#	define R300_FG_FOG_BLEND_FN_EXP2              (2 << 1)
+#	define R300_FG_FOG_BLEND_FN_CONSTANT          (3 << 1)
+#	define R300_FG_FOG_BLEND_FN_MASK              (3 << 1)
+
+/* Fog: Red Component of Fog Color */
+#define R300_FG_FOG_COLOR_R                           0x4bc8
+/* Fog: Green Component of Fog Color */
+#define R300_FG_FOG_COLOR_G                           0x4bcc
+/* Fog: Blue Component of Fog Color */
+#define R300_FG_FOG_COLOR_B                           0x4bd0
+#	define R300_FG_FOG_COLOR_MASK 0x000003ff
+
+/* Fog: Constant Factor for Fog Blending */
+#define R300_FG_FOG_FACTOR                            0x4bc4
+#	define FG_FOG_FACTOR_MASK 0x000003ff
+
+/* Fog: Alpha function */
+#define R300_FG_ALPHA_FUNC                            0x4bd4
+#       define R300_FG_ALPHA_FUNC_VAL_MASK               0x000000ff
+#       define R300_FG_ALPHA_FUNC_NEVER                     (0 << 8)
+#       define R300_FG_ALPHA_FUNC_LESS                      (1 << 8)
+#       define R300_FG_ALPHA_FUNC_EQUAL                     (2 << 8)
+#       define R300_FG_ALPHA_FUNC_LE                        (3 << 8)
+#       define R300_FG_ALPHA_FUNC_GREATER                   (4 << 8)
+#       define R300_FG_ALPHA_FUNC_NOTEQUAL                  (5 << 8)
+#       define R300_FG_ALPHA_FUNC_GE                        (6 << 8)
+#       define R300_FG_ALPHA_FUNC_ALWAYS                    (7 << 8)
+#       define R300_ALPHA_TEST_OP_MASK                      (7 << 8)
+#       define R300_FG_ALPHA_FUNC_DISABLE                   (0 << 11)
+#       define R300_FG_ALPHA_FUNC_ENABLE                    (1 << 11)
+
+#       define R500_FG_ALPHA_FUNC_10BIT                     (0 << 12)
+#       define R500_FG_ALPHA_FUNC_8BIT                      (1 << 12)
+
+#       define R300_FG_ALPHA_FUNC_MASK_DISABLE              (0 << 16)
+#       define R300_FG_ALPHA_FUNC_MASK_ENABLE               (1 << 16)
+#       define R300_FG_ALPHA_FUNC_CFG_2_OF_4                (0 << 17)
+#       define R300_FG_ALPHA_FUNC_CFG_3_OF_6                (1 << 17)
+
+#       define R300_FG_ALPHA_FUNC_DITH_DISABLE              (0 << 20)
+#       define R300_FG_ALPHA_FUNC_DITH_ENABLE               (1 << 20)
+
+#       define R500_FG_ALPHA_FUNC_OFFSET_DISABLE            (0 << 24)
+#       define R500_FG_ALPHA_FUNC_OFFSET_ENABLE             (1 << 24) /* Not supported in R520 */
+#       define R500_FG_ALPHA_FUNC_DISC_ZERO_MASK_DISABLE    (0 << 25)
+#       define R500_FG_ALPHA_FUNC_DISC_ZERO_MASK_ENABLE     (1 << 25)
+
+#       define R500_FG_ALPHA_FUNC_FP16_DISABLE              (0 << 28)
+#       define R500_FG_ALPHA_FUNC_FP16_ENABLE               (1 << 28)
+
+
+/* Fog: Where does the depth come from? */
+#define R300_FG_DEPTH_SRC                  0x4bd8
+#	define R300_FG_DEPTH_SRC_SCAN   (0 << 0)
+#	define R300_FG_DEPTH_SRC_SHADER (1 << 0)
+
+/* Fog: Alpha Compare Value */
+#define R500_FG_ALPHA_VALUE                0x4be0
+#	define R500_FG_ALPHA_VALUE_MASK 0x0000ffff
 
 /* gap */
 
@@ -1285,12 +2006,33 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_PFS_PARAM_0_Y                  0x4C04
 #define R300_PFS_PARAM_0_Z                  0x4C08
 #define R300_PFS_PARAM_0_W                  0x4C0C
-/* GUESS: PARAM_31 is last, based on native limits reported by fglrx */
+/* last consts */
 #define R300_PFS_PARAM_31_X                 0x4DF0
 #define R300_PFS_PARAM_31_Y                 0x4DF4
 #define R300_PFS_PARAM_31_Z                 0x4DF8
 #define R300_PFS_PARAM_31_W                 0x4DFC
 
+/* Unpipelined. */
+#define R300_RB3D_CCTL                      0x4e00
+#	define R300_RB3D_CCTL_NUM_MULTIWRITES_1_BUFFER                (0 << 5)
+#	define R300_RB3D_CCTL_NUM_MULTIWRITES_2_BUFFERS               (1 << 5)
+#	define R300_RB3D_CCTL_NUM_MULTIWRITES_3_BUFFERS               (2 << 5)
+#	define R300_RB3D_CCTL_NUM_MULTIWRITES_4_BUFFERS               (3 << 5)
+#	define R300_RB3D_CCTL_CLRCMP_FLIPE_DISABLE                    (0 << 7)
+#	define R300_RB3D_CCTL_CLRCMP_FLIPE_ENABLE                     (1 << 7)
+#	define R300_RB3D_CCTL_AA_COMPRESSION_DISABLE                  (0 << 9)
+#	define R300_RB3D_CCTL_AA_COMPRESSION_ENABLE                   (1 << 9)
+#	define R300_RB3D_CCTL_CMASK_DISABLE                           (0 << 10)
+#	define R300_RB3D_CCTL_CMASK_ENABLE                            (1 << 10)
+/* reserved */
+#	define R300_RB3D_CCTL_INDEPENDENT_COLOR_CHANNEL_MASK_DISABLE  (0 << 12)
+#	define R300_RB3D_CCTL_INDEPENDENT_COLOR_CHANNEL_MASK_ENABLE   (1 << 12)
+#	define R300_RB3D_CCTL_WRITE_COMPRESSION_ENABLE                (0 << 13)
+#	define R300_RB3D_CCTL_WRITE_COMPRESSION_DISABLE               (1 << 13)
+#	define R300_RB3D_CCTL_INDEPENDENT_COLORFORMAT_ENABLE_DISABLE  (0 << 14)
+#	define R300_RB3D_CCTL_INDEPENDENT_COLORFORMAT_ENABLE_ENABLE   (1 << 14)
+
+
 /* Notes:
  * - AFAIK fglrx always sets BLEND_UNKNOWN when blending is used in
  *   the application
@@ -1302,9 +2044,17 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_RB3D_CBLEND                    0x4E04
 #define R300_RB3D_ABLEND                    0x4E08
 /* the following only appear in CBLEND */
-#       define R300_BLEND_ENABLE                     (1 << 0)
-#       define R300_BLEND_UNKNOWN                    (3 << 1)
-#       define R300_BLEND_NO_SEPARATE                (1 << 3)
+#       define R300_ALPHA_BLEND_ENABLE         (1 << 0)
+#       define R300_SEPARATE_ALPHA_ENABLE      (1 << 1)
+#       define R300_READ_ENABLE                (1 << 2)
+#       define R300_DISCARD_SRC_PIXELS_DIS     (0 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_ALPHA_0     (1 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_COLOR_0     (2 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_ALPHA_COLOR_0     (3 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_ALPHA_1     (4 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_COLOR_1     (5 << 3)
+#       define R300_DISCARD_SRC_PIXELS_SRC_ALPHA_COLOR_1     (6 << 3)
+
 /* the following are shared between CBLEND and ABLEND */
 #       define R300_FCN_MASK                         (3  << 12)
 #       define R300_COMB_FCN_ADD_CLAMP               (0  << 12)
@@ -1333,67 +2083,213 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_BLEND_MASK                       (63)
 #       define R300_SRC_BLEND_SHIFT                  (16)
 #       define R300_DST_BLEND_SHIFT                  (24)
+
+/* Constant color used by the blender. Pipelined through the blender.
+ * Note: For R520, this field is ignored, use RB3D_CONSTANT_COLOR_GB__BLUE,
+ * RB3D_CONSTANT_COLOR_GB__GREEN, etc. instead.
+ */
 #define R300_RB3D_BLEND_COLOR               0x4E10
-#define R300_RB3D_COLORMASK                 0x4E0C
-#       define R300_COLORMASK0_B                 (1<<0)
-#       define R300_COLORMASK0_G                 (1<<1)
-#       define R300_COLORMASK0_R                 (1<<2)
-#       define R300_COLORMASK0_A                 (1<<3)
 
-/* gap */
 
-#define R300_RB3D_COLOROFFSET0              0x4E28
-#       define R300_COLOROFFSET_MASK             0xFFFFFFF0 /* GUESS */
-#define R300_RB3D_COLOROFFSET1              0x4E2C /* GUESS */
-#define R300_RB3D_COLOROFFSET2              0x4E30 /* GUESS */
-#define R300_RB3D_COLOROFFSET3              0x4E34 /* GUESS */
+/* 3D Color Channel Mask. If all the channels used in the current color format
+ * are disabled, then the cb will discard all the incoming quads. Pipelined
+ * through the blender.
+ */
+#define RB3D_COLOR_CHANNEL_MASK                  0x4E0C
+#	define RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0  (1 << 0)
+#	define RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0 (1 << 1)
+#	define RB3D_COLOR_CHANNEL_MASK_RED_MASK0   (1 << 2)
+#	define RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0 (1 << 3)
+#	define RB3D_COLOR_CHANNEL_MASK_BLUE_MASK1  (1 << 4)
+#	define RB3D_COLOR_CHANNEL_MASK_GREEN_MASK1 (1 << 5)
+#	define RB3D_COLOR_CHANNEL_MASK_RED_MASK1   (1 << 6)
+#	define RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK1 (1 << 7)
+#	define RB3D_COLOR_CHANNEL_MASK_BLUE_MASK2  (1 << 8)
+#	define RB3D_COLOR_CHANNEL_MASK_GREEN_MASK2 (1 << 9)
+#	define RB3D_COLOR_CHANNEL_MASK_RED_MASK2   (1 << 10)
+#	define RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK2 (1 << 11)
+#	define RB3D_COLOR_CHANNEL_MASK_BLUE_MASK3  (1 << 12)
+#	define RB3D_COLOR_CHANNEL_MASK_GREEN_MASK3 (1 << 13)
+#	define RB3D_COLOR_CHANNEL_MASK_RED_MASK3   (1 << 14)
+#	define RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK3 (1 << 15)
+
+/* Clear color that is used when the color mask is set to 00. Unpipelined.
+ * Program this register with a 32-bit value in ARGB8888 or ARGB2101010
+ * formats, ignoring the fields.
+ */
+#define RB3D_COLOR_CLEAR_VALUE                   0x4e14
 
 /* gap */
 
-/* Bit 16: Larger tiles
+/* Color Compare Color. Stalls the 2d/3d datapath until it is idle. */
+#define RB3D_CLRCMP_CLR                     0x4e20
+
+/* Color Compare Mask. Stalls the 2d/3d datapath until it is idle. */
+#define RB3D_CLRCMP_MSK                     0x4e24
+
+/* Color Buffer Address Offset of multibuffer 0. Unpipelined. */
+#define R300_RB3D_COLOROFFSET0              0x4E28
+#       define R300_COLOROFFSET_MASK             0xFFFFFFE0
+/* Color Buffer Address Offset of multibuffer 1. Unpipelined. */
+#define R300_RB3D_COLOROFFSET1              0x4E2C
+/* Color Buffer Address Offset of multibuffer 2. Unpipelined. */
+#define R300_RB3D_COLOROFFSET2              0x4E30
+/* Color Buffer Address Offset of multibuffer 3. Unpipelined. */
+#define R300_RB3D_COLOROFFSET3              0x4E34
+
+/* Color buffer format and tiling control for all the multibuffers and the
+ * pitch of multibuffer 0 to 3. Unpipelined. The cache must be empty before any
+ * of the registers are changed.
+ *
+ * Bit 16: Larger tiles
  * Bit 17: 4x2 tiles
  * Bit 18: Extremely weird tile like, but some pixels duplicated?
  */
 #define R300_RB3D_COLORPITCH0               0x4E38
-#       define R300_COLORPITCH_MASK              0x00001FF8 /* GUESS */
-#       define R300_COLOR_TILE_ENABLE            (1 << 16) /* GUESS */
-#       define R300_COLOR_MICROTILE_ENABLE       (1 << 17) /* GUESS */
-#       define R300_COLOR_ENDIAN_NO_SWAP         (0 << 18) /* GUESS */
-#       define R300_COLOR_ENDIAN_WORD_SWAP       (1 << 18) /* GUESS */
-#       define R300_COLOR_ENDIAN_DWORD_SWAP      (2 << 18) /* GUESS */
-#       define R300_COLOR_FORMAT_RGB565          (2 << 22)
-#       define R300_COLOR_FORMAT_ARGB8888        (3 << 22)
-#define R300_RB3D_COLORPITCH1               0x4E3C /* GUESS */
-#define R300_RB3D_COLORPITCH2               0x4E40 /* GUESS */
-#define R300_RB3D_COLORPITCH3               0x4E44 /* GUESS */
+#       define R300_COLORPITCH_MASK              0x00003FFE
+#       define R300_COLOR_TILE_DISABLE            (0 << 16)
+#       define R300_COLOR_TILE_ENABLE             (1 << 16)
+#       define R300_COLOR_MICROTILE_DISABLE       (0 << 17)
+#       define R300_COLOR_MICROTILE_ENABLE        (1 << 17)
+#       define R300_COLOR_MICROTILE_ENABLE_SQUARE (2 << 17) /* Only available in 16-bit */
+#       define R300_COLOR_ENDIAN_NO_SWAP          (0 << 19)
+#       define R300_COLOR_ENDIAN_WORD_SWAP        (1 << 19)
+#       define R300_COLOR_ENDIAN_DWORD_SWAP       (2 << 19)
+#       define R300_COLOR_ENDIAN_HALF_DWORD_SWAP  (3 << 19)
+#	define R500_COLOR_FORMAT_ARGB10101010     (0 << 21)
+#	define R500_COLOR_FORMAT_UV1010           (1 << 21)
+#	define R500_COLOR_FORMAT_CI8              (2 << 21) /* 2D only */
+#	define R300_COLOR_FORMAT_ARGB1555         (3 << 21)
+#       define R300_COLOR_FORMAT_RGB565           (4 << 21)
+#       define R500_COLOR_FORMAT_ARGB2101010      (5 << 21)
+#       define R300_COLOR_FORMAT_ARGB8888         (6 << 21)
+#       define R300_COLOR_FORMAT_ARGB32323232     (7 << 21)
+/* reserved */
+#       define R300_COLOR_FORMAT_I8               (9 << 21)
+#       define R300_COLOR_FORMAT_ARGB16161616     (10 << 21)
+#       define R300_COLOR_FORMAT_VYUY             (11 << 21)
+#       define R300_COLOR_FORMAT_YVYU             (12 << 21)
+#       define R300_COLOR_FORMAT_UV88             (13 << 21)
+#       define R500_COLOR_FORMAT_I10              (14 << 21)
+#       define R300_COLOR_FORMAT_ARGB4444         (15 << 21)
+#define R300_RB3D_COLORPITCH1               0x4E3C
+#define R300_RB3D_COLORPITCH2               0x4E40
+#define R300_RB3D_COLORPITCH3               0x4E44
 
 /* gap */
 
-/* Guess by Vladimir.
+/* Destination Color Buffer Cache Control/Status. If the cb is in e2 mode, then
+ * a flush or free will not occur upon a write to this register, but a sync
+ * will be immediately sent if one is requested. If both DC_FLUSH and DC_FREE
+ * are zero but DC_FINISH is one, then a sync will be sent immediately -- the
+ * cb will not wait for all the previous operations to complete before sending
+ * the sync. Unpipelined except when DC_FINISH and DC_FREE are both set to
+ * zero.
+ *
  * Set to 0A before 3D operations, set to 02 afterwards.
  */
-#define R300_RB3D_DSTCACHE_CTLSTAT          0x4E4C
-#       define R300_RB3D_DSTCACHE_UNKNOWN_02             0x00000002
-#       define R300_RB3D_DSTCACHE_UNKNOWN_0A             0x0000000A
+#define R300_RB3D_DSTCACHE_CTLSTAT               0x4e4c
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_NO_EFFECT         (0 << 0)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_NO_EFFECT_1       (1 << 0)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_FLUSH_DIRTY_3D    (2 << 0)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FLUSH_FLUSH_DIRTY_3D_1  (3 << 0)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_NO_EFFECT          (0 << 2)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_NO_EFFECT_1        (1 << 2)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_FREE_3D_TAGS       (2 << 2)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FREE_FREE_3D_TAGS_1     (3 << 2)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FINISH_NO_SIGNAL        (0 << 4)
+#	define R300_RB3D_DSTCACHE_CTLSTAT_DC_FINISH_SIGNAL           (1 << 4)
+
+#define R300_RB3D_DITHER_CTL 0x4E50
+#	define R300_RB3D_DITHER_CTL_DITHER_MODE_TRUNCATE         (0 << 0)
+#	define R300_RB3D_DITHER_CTL_DITHER_MODE_ROUND            (1 << 0)
+#	define R300_RB3D_DITHER_CTL_DITHER_MODE_LUT              (2 << 0)
+/* reserved */
+#	define R300_RB3D_DITHER_CTL_ALPHA_DITHER_MODE_TRUNCATE   (0 << 2)
+#	define R300_RB3D_DITHER_CTL_ALPHA_DITHER_MODE_ROUND      (1 << 2)
+#	define R300_RB3D_DITHER_CTL_ALPHA_DITHER_MODE_LUT        (2 << 2)
+/* reserved */
+
+/* Resolve buffer destination address. The cache must be empty before changing
+ * this register if the cb is in resolve mode. Unpipelined
+ */
+#define R300_RB3D_AARESOLVE_OFFSET        0x4e80
+#	define R300_RB3D_AARESOLVE_OFFSET_SHIFT 5
+#	define R300_RB3D_AARESOLVE_OFFSET_MASK 0xffffffe0 /* At least according to the calculations of Christoph Brill */
+
+/* Resolve Buffer Pitch and Tiling Control. The cache must be empty before
+ * changing this register if the cb is in resolve mode. Unpipelined
+ */
+#define R300_RB3D_AARESOLVE_PITCH         0x4e84
+#	define R300_RB3D_AARESOLVE_PITCH_SHIFT 1
+#	define R300_RB3D_AARESOLVE_PITCH_MASK  0x00003ffe /* At least according to the calculations of Christoph Brill */
+
+/* Resolve Buffer Control. Unpipelined */
+#define R300_RB3D_AARESOLVE_CTL           0x4e88
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_MODE_NORMAL   (0 << 0)
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_MODE_RESOLVE  (1 << 0)
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_GAMMA_10      (0 << 1)
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_GAMMA_22      (1 << 1)
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_ALPHA_SAMPLE0 (0 << 2)
+#	define R300_RB3D_AARESOLVE_CTL_AARESOLVE_ALPHA_AVERAGE (1 << 2)
+
+
+/* Discard src pixels less than or equal to threshold. */
+#define R500_RB3D_DISCARD_SRC_PIXEL_LTE_THRESHOLD 0x4ea0
+/* Discard src pixels greater than or equal to threshold. */
+#define R500_RB3D_DISCARD_SRC_PIXEL_GTE_THRESHOLD 0x4ea4
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_BLUE_SHIFT 0
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_BLUE_MASK 0x000000ff
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_GREEN_SHIFT 8
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_GREEN_MASK 0x0000ff00
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_RED_SHIFT 16
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_RED_MASK 0x00ff0000
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_ALPHA_SHIFT 24
+#	define R500_RB3D_DISCARD_SRC_PIXEL_THRESHOLD_ALPHA_MASK 0xff000000
+
+/* 3D ROP Control. Stalls the 2d/3d datapath until it is idle. */
+#define R300_RB3D_ROPCNTL                             0x4e18
+#	define R300_RB3D_ROPCNTL_ROP_ENABLE            0x00000004
+#	define R300_RB3D_ROPCNTL_ROP_MASK              (15 << 8)
+#	define R300_RB3D_ROPCNTL_ROP_SHIFT             8
+
+/* Color Compare Flip. Stalls the 2d/3d datapath until it is idle. */
+#define R300_RB3D_CLRCMP_FLIPE                        0x4e1c
+
+/* Sets the fifo sizes */
+#define R500_RB3D_FIFO_SIZE                           0x4ef4
+#	define R500_RB3D_FIFO_SIZE_OP_FIFO_SIZE_FULL   (0 << 0)
+#	define R500_RB3D_FIFO_SIZE_OP_FIFO_SIZE_HALF   (1 << 0)
+#	define R500_RB3D_FIFO_SIZE_OP_FIFO_SIZE_QUATER (2 << 0)
+#	define R500_RB3D_FIFO_SIZE_OP_FIFO_SIZE_EIGTHS (3 << 0)
+
+/* Constant color used by the blender. Pipelined through the blender. */
+#define R500_RB3D_CONSTANT_COLOR_AR                   0x4ef8
+#	define R500_RB3D_CONSTANT_COLOR_AR_RED_MASK    0x0000ffff
+#	define R500_RB3D_CONSTANT_COLOR_AR_RED_SHIFT   0
+#	define R500_RB3D_CONSTANT_COLOR_AR_ALPHA_MASK  0xffff0000
+#	define R500_RB3D_CONSTANT_COLOR_AR_ALPHA_SHIFT 16
+
+/* Constant color used by the blender. Pipelined through the blender. */
+#define R500_RB3D_CONSTANT_COLOR_GB                   0x4efc
+#	define R500_RB3D_CONSTANT_COLOR_AR_BLUE_MASK   0x0000ffff
+#	define R500_RB3D_CONSTANT_COLOR_AR_BLUE_SHIFT  0
+#	define R500_RB3D_CONSTANT_COLOR_AR_GREEN_MASK  0xffff0000
+#	define R500_RB3D_CONSTANT_COLOR_AR_GREEN_SHIFT 16
 
 /* gap */
 /* There seems to be no "write only" setting, so use Z-test = ALWAYS
  * for this.
  * Bit (1<<8) is the "test" bit. so plain write is 6  - vd
  */
-#define R300_RB3D_ZSTENCIL_CNTL_0                   0x4F00
-#       define R300_RB3D_Z_DISABLED_1            0x00000010
-#       define R300_RB3D_Z_DISABLED_2            0x00000014
-#       define R300_RB3D_Z_TEST                  0x00000012
-#       define R300_RB3D_Z_TEST_AND_WRITE        0x00000016
-#       define R300_RB3D_Z_WRITE_ONLY        	 0x00000006
-
-#       define R300_RB3D_Z_TEST                  0x00000012
-#       define R300_RB3D_Z_TEST_AND_WRITE        0x00000016
-#       define R300_RB3D_Z_WRITE_ONLY        	 0x00000006
-#	define R300_RB3D_STENCIL_ENABLE		 0x00000001
-
-#define R300_RB3D_ZSTENCIL_CNTL_1                   0x4F04
+#define R300_ZB_CNTL                             0x4F00
+#	define R300_STENCIL_ENABLE		 (1 << 0)
+#	define R300_Z_ENABLE		         (1 << 1)
+#	define R300_Z_WRITE_ENABLE		 (1 << 2)
+#	define R300_Z_SIGNED_COMPARE		 (1 << 3)
+#	define R300_STENCIL_FRONT_BACK		 (1 << 4)
+
+#define R300_ZB_ZSTENCILCNTL                   0x4f04
 	/* functions */
 #	define R300_ZS_NEVER			0
 #	define R300_ZS_LESS			1
@@ -1413,162 +2309,344 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_ZS_INVERT			5
 #	define R300_ZS_INCR_WRAP		6
 #	define R300_ZS_DECR_WRAP		7
+#	define R300_Z_FUNC_SHIFT		0
 	/* front and back refer to operations done for front
 	   and back faces, i.e. separate stencil function support */
-#	define R300_RB3D_ZS1_DEPTH_FUNC_SHIFT		0
-#	define R300_RB3D_ZS1_FRONT_FUNC_SHIFT		3
-#	define R300_RB3D_ZS1_FRONT_FAIL_OP_SHIFT	6
-#	define R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT	9
-#	define R300_RB3D_ZS1_FRONT_ZFAIL_OP_SHIFT      12
-#	define R300_RB3D_ZS1_BACK_FUNC_SHIFT           15
-#	define R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT        18
-#	define R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT       21
-#	define R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT       24
-
-#define R300_RB3D_ZSTENCIL_CNTL_2                   0x4F08
-#	define R300_RB3D_ZS2_STENCIL_REF_SHIFT		0
-#	define R300_RB3D_ZS2_STENCIL_MASK		0xFF
-#	define R300_RB3D_ZS2_STENCIL_MASK_SHIFT	        8
-#	define R300_RB3D_ZS2_STENCIL_WRITE_MASK_SHIFT	16
+#	define R300_S_FRONT_FUNC_SHIFT	        3
+#	define R300_S_FRONT_SFAIL_OP_SHIFT	6
+#	define R300_S_FRONT_ZPASS_OP_SHIFT	9
+#	define R300_S_FRONT_ZFAIL_OP_SHIFT      12
+#	define R300_S_BACK_FUNC_SHIFT           15
+#	define R300_S_BACK_SFAIL_OP_SHIFT       18
+#	define R300_S_BACK_ZPASS_OP_SHIFT       21
+#	define R300_S_BACK_ZFAIL_OP_SHIFT       24
+
+#define R300_ZB_STENCILREFMASK                        0x4f08
+#	define R300_STENCILREF_SHIFT       0
+#	define R300_STENCILREF_MASK        0x000000ff
+#	define R300_STENCILMASK_SHIFT      8
+#	define R300_STENCILMASK_MASK       0x0000ff00
+#	define R300_STENCILWRITEMASK_SHIFT 16
+#	define R300_STENCILWRITEMASK_MASK  0x00ff0000
 
 /* gap */
 
-#define R300_RB3D_ZSTENCIL_FORMAT                   0x4F10
-#	define R300_DEPTH_FORMAT_16BIT_INT_Z     (0 << 0)
-#	define R300_DEPTH_FORMAT_24BIT_INT_Z     (2 << 0)
-	/* 16 bit format or some aditional bit ? */
-#	define R300_DEPTH_FORMAT_UNK32          (32 << 0)
+#define R300_ZB_FORMAT                             0x4f10
+#	define R300_DEPTHFORMAT_16BIT_INT_Z   (0 << 0)
+#	define R300_DEPTHFORMAT_16BIT_13E3    (1 << 0)
+#	define R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL   (2 << 0)
+/* reserved up to (15 << 0) */
+#	define R300_INVERT_13E3_LEADING_ONES  (0 << 4)
+#	define R300_INVERT_13E3_LEADING_ZEROS (1 << 4)
 
-#define R300_RB3D_EARLY_Z                           0x4F14
-#	define R300_EARLY_Z_DISABLE              (0 << 0)
-#	define R300_EARLY_Z_ENABLE               (1 << 0)
+#define R300_ZB_ZTOP                             0x4F14
+#	define R300_ZTOP_DISABLE                 (0 << 0)
+#	define R300_ZTOP_ENABLE                  (1 << 0)
 
 /* gap */
 
-#define R300_RB3D_ZCACHE_CTLSTAT            0x4F18 /* GUESS */
-#       define R300_RB3D_ZCACHE_UNKNOWN_01  0x1
-#       define R300_RB3D_ZCACHE_UNKNOWN_03  0x3
+#define R300_ZB_ZCACHE_CTLSTAT            0x4f18
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_NO_EFFECT      (0 << 0)
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_FLUSH_FLUSH_AND_FREE (1 << 0)
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_NO_EFFECT       (0 << 1)
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_FREE_FREE            (1 << 1)
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_BUSY_IDLE            (0 << 31)
+#       define R300_ZB_ZCACHE_CTLSTAT_ZC_BUSY_BUSY            (1 << 31)
+
+#define R300_ZB_BW_CNTL                     0x4f1c
+#	define R300_HIZ_DISABLE                              (0 << 0)
+#	define R300_HIZ_ENABLE                               (1 << 0)
+#	define R300_HIZ_MIN                                  (0 << 1)
+#	define R300_HIZ_MAX                                  (1 << 1)
+#	define R300_FAST_FILL_DISABLE                        (0 << 2)
+#	define R300_FAST_FILL_ENABLE                         (1 << 2)
+#	define R300_RD_COMP_DISABLE                          (0 << 3)
+#	define R300_RD_COMP_ENABLE                           (1 << 3)
+#	define R300_WR_COMP_DISABLE                          (0 << 4)
+#	define R300_WR_COMP_ENABLE                           (1 << 4)
+#	define R300_ZB_CB_CLEAR_RMW                          (0 << 5)
+#	define R300_ZB_CB_CLEAR_CACHE_LINEAR                 (1 << 5)
+#	define R300_FORCE_COMPRESSED_STENCIL_VALUE_DISABLE   (0 << 6)
+#	define R300_FORCE_COMPRESSED_STENCIL_VALUE_ENABLE    (1 << 6)
+
+#	define R500_ZEQUAL_OPTIMIZE_ENABLE                   (0 << 7)
+#	define R500_ZEQUAL_OPTIMIZE_DISABLE                  (1 << 7)
+#	define R500_SEQUAL_OPTIMIZE_ENABLE                   (0 << 8)
+#	define R500_SEQUAL_OPTIMIZE_DISABLE                  (1 << 8)
+
+#	define R500_BMASK_ENABLE                             (0 << 10)
+#	define R500_BMASK_DISABLE                            (1 << 10)
+#	define R500_HIZ_EQUAL_REJECT_DISABLE                 (0 << 11)
+#	define R500_HIZ_EQUAL_REJECT_ENABLE                  (1 << 11)
+#	define R500_HIZ_FP_EXP_BITS_DISABLE                  (0 << 12)
+#	define R500_HIZ_FP_EXP_BITS_1                        (1 << 12)
+#	define R500_HIZ_FP_EXP_BITS_2                        (2 << 12)
+#	define R500_HIZ_FP_EXP_BITS_3                        (3 << 12)
+#	define R500_HIZ_FP_EXP_BITS_4                        (4 << 12)
+#	define R500_HIZ_FP_EXP_BITS_5                        (5 << 12)
+#	define R500_HIZ_FP_INVERT_LEADING_ONES               (0 << 15)
+#	define R500_HIZ_FP_INVERT_LEADING_ZEROS              (1 << 15)
+#	define R500_TILE_OVERWRITE_RECOMPRESSION_ENABLE      (0 << 16)
+#	define R500_TILE_OVERWRITE_RECOMPRESSION_DISABLE     (1 << 16)
+#	define R500_CONTIGUOUS_6XAA_SAMPLES_ENABLE           (0 << 17)
+#	define R500_CONTIGUOUS_6XAA_SAMPLES_DISABLE          (1 << 17)
+#	define R500_PEQ_PACKING_DISABLE                      (0 << 18)
+#	define R500_PEQ_PACKING_ENABLE                       (1 << 18)
+#	define R500_COVERED_PTR_MASKING_DISABLE              (0 << 18)
+#	define R500_COVERED_PTR_MASKING_ENABLE               (1 << 18)
+
 
 /* gap */
 
-#define R300_RB3D_DEPTHOFFSET               0x4F20
-#define R300_RB3D_DEPTHPITCH                0x4F24
-#       define R300_DEPTHPITCH_MASK              0x00001FF8 /* GUESS */
-#       define R300_DEPTH_TILE_ENABLE            (1 << 16) /* GUESS */
-#       define R300_DEPTH_MICROTILE_ENABLE       (1 << 17) /* GUESS */
-#       define R300_DEPTH_ENDIAN_NO_SWAP         (0 << 18) /* GUESS */
-#       define R300_DEPTH_ENDIAN_WORD_SWAP       (1 << 18) /* GUESS */
-#       define R300_DEPTH_ENDIAN_DWORD_SWAP      (2 << 18) /* GUESS */
-
-/* BEGIN: Vertex program instruction set */
-
-/* Every instruction is four dwords long:
- *  DWORD 0: output and opcode
- *  DWORD 1: first argument
- *  DWORD 2: second argument
- *  DWORD 3: third argument
- *
- * Notes:
- *  - ABS r, a is implemented as MAX r, a, -a
- *  - MOV is implemented as ADD to zero
- *  - XPD is implemented as MUL + MAD
- *  - FLR is implemented as FRC + ADD
- *  - apparently, fglrx tries to schedule instructions so that there is at
- *    least one instruction between the write to a temporary and the first
- *    read from said temporary; however, violations of this scheduling are
- *    allowed
- *  - register indices seem to be unrelated with OpenGL aliasing to
- *    conventional state
- *  - only one attribute and one parameter can be loaded at a time; however,
- *    the same attribute/parameter can be used for more than one argument
- *  - the second software argument for POW is the third hardware argument
- *    (no idea why)
- *  - MAD with only temporaries as input seems to use VPI_OUT_SELECT_MAD_2
- *
- * There is some magic surrounding LIT:
- *   The single argument is replicated across all three inputs, but swizzled:
- *     First argument: xyzy
- *     Second argument: xyzx
- *     Third argument: xyzw
- *   Whenever the result is used later in the fragment program, fglrx forces
- *   x and w to be 1.0 in the input selection; I don't know whether this is
- *   strictly necessary
+/* Z Buffer Address Offset.
+ * Bits 31 to 5 are used for aligned Z buffer address offset for macro tiles.
  */
-#define R300_VPI_OUT_OP_DOT                     (1 << 0)
-#define R300_VPI_OUT_OP_MUL                     (2 << 0)
-#define R300_VPI_OUT_OP_ADD                     (3 << 0)
-#define R300_VPI_OUT_OP_MAD                     (4 << 0)
-#define R300_VPI_OUT_OP_DST                     (5 << 0)
-#define R300_VPI_OUT_OP_FRC                     (6 << 0)
-#define R300_VPI_OUT_OP_MAX                     (7 << 0)
-#define R300_VPI_OUT_OP_MIN                     (8 << 0)
-#define R300_VPI_OUT_OP_SGE                     (9 << 0)
-#define R300_VPI_OUT_OP_SLT                     (10 << 0)
-	/* Used in GL_POINT_DISTANCE_ATTENUATION_ARB, vector(scalar, vector) */
-#define R300_VPI_OUT_OP_UNK12                   (12 << 0)
-#define R300_VPI_OUT_OP_ARL                     (13 << 0)
-#define R300_VPI_OUT_OP_EXP                     (65 << 0)
-#define R300_VPI_OUT_OP_LOG                     (66 << 0)
-	/* Used in fog computations, scalar(scalar) */
-#define R300_VPI_OUT_OP_UNK67                   (67 << 0)
-#define R300_VPI_OUT_OP_LIT                     (68 << 0)
-#define R300_VPI_OUT_OP_POW                     (69 << 0)
-#define R300_VPI_OUT_OP_RCP                     (70 << 0)
-#define R300_VPI_OUT_OP_RSQ                     (72 << 0)
-	/* Used in GL_POINT_DISTANCE_ATTENUATION_ARB, scalar(scalar) */
-#define R300_VPI_OUT_OP_UNK73                   (73 << 0)
-#define R300_VPI_OUT_OP_EX2                     (75 << 0)
-#define R300_VPI_OUT_OP_LG2                     (76 << 0)
-#define R300_VPI_OUT_OP_MAD_2                   (128 << 0)
-	/* all temps, vector(scalar, vector, vector) */
-#define R300_VPI_OUT_OP_UNK129                  (129 << 0)
-
-#define R300_VPI_OUT_REG_CLASS_TEMPORARY        (0 << 8)
-#define R300_VPI_OUT_REG_CLASS_ADDR             (1 << 8)
-#define R300_VPI_OUT_REG_CLASS_RESULT           (2 << 8)
-#define R300_VPI_OUT_REG_CLASS_MASK             (31 << 8)
-
-#define R300_VPI_OUT_REG_INDEX_SHIFT            13
-	/* GUESS based on fglrx native limits */
-#define R300_VPI_OUT_REG_INDEX_MASK             (31 << 13)
-
-#define R300_VPI_OUT_WRITE_X                    (1 << 20)
-#define R300_VPI_OUT_WRITE_Y                    (1 << 21)
-#define R300_VPI_OUT_WRITE_Z                    (1 << 22)
-#define R300_VPI_OUT_WRITE_W                    (1 << 23)
-
-#define R300_VPI_IN_REG_CLASS_TEMPORARY         (0 << 0)
-#define R300_VPI_IN_REG_CLASS_ATTRIBUTE         (1 << 0)
-#define R300_VPI_IN_REG_CLASS_PARAMETER         (2 << 0)
-#define R300_VPI_IN_REG_CLASS_NONE              (9 << 0)
-#define R300_VPI_IN_REG_CLASS_MASK              (31 << 0)
-
-#define R300_VPI_IN_REG_INDEX_SHIFT             5
-	/* GUESS based on fglrx native limits */
-#define R300_VPI_IN_REG_INDEX_MASK              (255 << 5)
-
-/* The R300 can select components from the input register arbitrarily.
- * Use the following constants, shifted by the component shift you
- * want to select
+#define R300_ZB_DEPTHOFFSET               0x4f20
+
+/* Z Buffer Pitch and Endian Control */
+#define R300_ZB_DEPTHPITCH                0x4f24
+#       define R300_DEPTHPITCH_MASK              0x00003FFC
+#       define R300_DEPTHMACROTILE_DISABLE      (0 << 16)
+#       define R300_DEPTHMACROTILE_ENABLE       (1 << 16)
+#       define R300_DEPTHMICROTILE_LINEAR       (0 << 17)
+#       define R300_DEPTHMICROTILE_TILED        (1 << 17)
+#       define R300_DEPTHMICROTILE_TILED_SQUARE (2 << 17)
+#       define R300_DEPTHENDIAN_NO_SWAP         (0 << 18)
+#       define R300_DEPTHENDIAN_WORD_SWAP       (1 << 18)
+#       define R300_DEPTHENDIAN_DWORD_SWAP      (2 << 18)
+#       define R300_DEPTHENDIAN_HALF_DWORD_SWAP (3 << 18)
+
+/* Z Buffer Clear Value */
+#define R300_ZB_DEPTHCLEARVALUE                  0x4f28
+
+/* Hierarchical Z Memory Offset */
+#define R300_ZB_HIZ_OFFSET                       0x4f44
+
+/* Hierarchical Z Write Index */
+#define R300_ZB_HIZ_WRINDEX                      0x4f48
+
+/* Hierarchical Z Data */
+#define R300_ZB_HIZ_DWORD                        0x4f4c
+
+/* Hierarchical Z Read Index */
+#define R300_ZB_HIZ_RDINDEX                      0x4f50
+
+/* Hierarchical Z Pitch */
+#define R300_ZB_HIZ_PITCH                        0x4f54
+
+/* Z Buffer Z Pass Counter Data */
+#define R300_ZB_ZPASS_DATA                       0x4f58
+
+/* Z Buffer Z Pass Counter Address */
+#define R300_ZB_ZPASS_ADDR                       0x4f5c
+
+/* Depth buffer X and Y coordinate offset */
+#define R300_ZB_DEPTHXY_OFFSET                   0x4f60
+#	define R300_DEPTHX_OFFSET_SHIFT  1
+#	define R300_DEPTHX_OFFSET_MASK   0x000007FE
+#	define R300_DEPTHY_OFFSET_SHIFT  17
+#	define R300_DEPTHY_OFFSET_MASK   0x07FE0000
+
+/* Sets the fifo sizes */
+#define R500_ZB_FIFO_SIZE                        0x4fd0
+#	define R500_OP_FIFO_SIZE_FULL   (0 << 0)
+#	define R500_OP_FIFO_SIZE_HALF   (1 << 0)
+#	define R500_OP_FIFO_SIZE_QUATER (2 << 0)
+#	define R500_OP_FIFO_SIZE_EIGTHS (4 << 0)
+
+/* Stencil Reference Value and Mask for backfacing quads */
+/* R300_ZB_STENCILREFMASK handles front face */
+#define R500_ZB_STENCILREFMASK_BF                0x4fd4
+#	define R500_STENCILREF_SHIFT       0
+#	define R500_STENCILREF_MASK        0x000000ff
+#	define R500_STENCILMASK_SHIFT      8
+#	define R500_STENCILMASK_MASK       0x0000ff00
+#	define R500_STENCILWRITEMASK_SHIFT 16
+#	define R500_STENCILWRITEMASK_MASK  0x00ff0000
+
+/**
+ * \defgroup R3XX_R5XX_PROGRAMMABLE_VERTEX_SHADER_DESCRIPTION R3XX-R5XX PROGRAMMABLE VERTEX SHADER DESCRIPTION
+ *
+ * The PVS_DST_MATH_INST is used to identify whether the instruction is a Vector
+ * Engine instruction or a Math Engine instruction.
  */
-#define R300_VPI_IN_SELECT_X    0
-#define R300_VPI_IN_SELECT_Y    1
-#define R300_VPI_IN_SELECT_Z    2
-#define R300_VPI_IN_SELECT_W    3
-#define R300_VPI_IN_SELECT_ZERO 4
-#define R300_VPI_IN_SELECT_ONE  5
-#define R300_VPI_IN_SELECT_MASK 7
-
-#define R300_VPI_IN_X_SHIFT                     13
-#define R300_VPI_IN_Y_SHIFT                     16
-#define R300_VPI_IN_Z_SHIFT                     19
-#define R300_VPI_IN_W_SHIFT                     22
-
-#define R300_VPI_IN_NEG_X                       (1 << 25)
-#define R300_VPI_IN_NEG_Y                       (1 << 26)
-#define R300_VPI_IN_NEG_Z                       (1 << 27)
-#define R300_VPI_IN_NEG_W                       (1 << 28)
-/* END: Vertex program instruction set */
+
+/*\{*/
+
+enum {
+	/* R3XX */
+	VECTOR_NO_OP			= 0,
+	VE_DOT_PRODUCT			= 1,
+	VE_MULTIPLY			= 2,
+	VE_ADD				= 3,
+	VE_MULTIPLY_ADD			= 4,
+	VE_DISTANCE_VECTOR		= 5,
+	VE_FRACTION			= 6,
+	VE_MAXIMUM			= 7,
+	VE_MINIMUM			= 8,
+	VE_SET_GREATER_THAN_EQUAL	= 9,
+	VE_SET_LESS_THAN		= 10,
+	VE_MULTIPLYX2_ADD		= 11,
+	VE_MULTIPLY_CLAMP		= 12,
+	VE_FLT2FIX_DX			= 13,
+	VE_FLT2FIX_DX_RND		= 14,
+	/* R5XX */
+	VE_PRED_SET_EQ_PUSH		= 15,
+	VE_PRED_SET_GT_PUSH		= 16,
+	VE_PRED_SET_GTE_PUSH		= 17,
+	VE_PRED_SET_NEQ_PUSH		= 18,
+	VE_COND_WRITE_EQ		= 19,
+	VE_COND_WRITE_GT		= 20,
+	VE_COND_WRITE_GTE		= 21,
+	VE_COND_WRITE_NEQ		= 22,
+	VE_COND_MUX_EQ			= 23,
+	VE_COND_MUX_GT			= 24,
+	VE_COND_MUX_GTE			= 25,
+	VE_SET_GREATER_THAN		= 26,
+	VE_SET_EQUAL			= 27,
+	VE_SET_NOT_EQUAL		= 28,
+};
+
+enum {
+	/* R3XX */
+	MATH_NO_OP			= 0,
+	ME_EXP_BASE2_DX			= 1,
+	ME_LOG_BASE2_DX			= 2,
+	ME_EXP_BASEE_FF			= 3,
+	ME_LIGHT_COEFF_DX		= 4,
+	ME_POWER_FUNC_FF		= 5,
+	ME_RECIP_DX			= 6,
+	ME_RECIP_FF			= 7,
+	ME_RECIP_SQRT_DX		= 8,
+	ME_RECIP_SQRT_FF		= 9,
+	ME_MULTIPLY			= 10,
+	ME_EXP_BASE2_FULL_DX		= 11,
+	ME_LOG_BASE2_FULL_DX		= 12,
+	ME_POWER_FUNC_FF_CLAMP_B	= 13,
+	ME_POWER_FUNC_FF_CLAMP_B1	= 14,
+	ME_POWER_FUNC_FF_CLAMP_01	= 15,
+	ME_SIN				= 16,
+	ME_COS				= 17,
+	/* R5XX */
+	ME_LOG_BASE2_IEEE		= 18,
+	ME_RECIP_IEEE			= 19,
+	ME_RECIP_SQRT_IEEE		= 20,
+	ME_PRED_SET_EQ			= 21,
+	ME_PRED_SET_GT			= 22,
+	ME_PRED_SET_GTE			= 23,
+	ME_PRED_SET_NEQ			= 24,
+	ME_PRED_SET_CLR			= 25,
+	ME_PRED_SET_INV			= 26,
+	ME_PRED_SET_POP			= 27,
+	ME_PRED_SET_RESTORE		= 28,
+};
+
+enum {
+	/* R3XX */
+	PVS_MACRO_OP_2CLK_MADD		= 0,
+	PVS_MACRO_OP_2CLK_M2X_ADD	= 1,
+};
+
+enum {
+	PVS_SRC_REG_TEMPORARY		= 0,	/* Intermediate Storage */
+	PVS_SRC_REG_INPUT		= 1,	/* Input Vertex Storage */
+	PVS_SRC_REG_CONSTANT		= 2,	/* Constant State Storage */
+	PVS_SRC_REG_ALT_TEMPORARY	= 3,	/* Alternate Intermediate Storage */
+};
+
+enum {
+	PVS_DST_REG_TEMPORARY		= 0,	/* Intermediate Storage */
+	PVS_DST_REG_A0			= 1,	/* Address Register Storage */
+	PVS_DST_REG_OUT			= 2,	/* Output Memory. Used for all outputs */
+	PVS_DST_REG_OUT_REPL_X		= 3,	/* Output Memory & Replicate X to all channels */
+	PVS_DST_REG_ALT_TEMPORARY	= 4,	/* Alternate Intermediate Storage */
+	PVS_DST_REG_INPUT		= 5,	/* Output Memory & Replicate X to all channels */
+};
+
+enum {
+	PVS_SRC_SELECT_X		= 0,	/* Select X Component */
+	PVS_SRC_SELECT_Y		= 1,	/* Select Y Component */
+	PVS_SRC_SELECT_Z		= 2,	/* Select Z Component */
+	PVS_SRC_SELECT_W		= 3,	/* Select W Component */
+	PVS_SRC_SELECT_FORCE_0		= 4,	/* Force Component to 0.0 */
+	PVS_SRC_SELECT_FORCE_1		= 5,	/* Force Component to 1.0 */
+};
+
+/* PVS Opcode & Destination Operand Description */
+
+enum {
+	PVS_DST_OPCODE_MASK		= 0x3f,
+	PVS_DST_OPCODE_SHIFT		= 0,
+	PVS_DST_MATH_INST_MASK		= 0x1,
+	PVS_DST_MATH_INST_SHIFT		= 6,
+	PVS_DST_MACRO_INST_MASK		= 0x1,
+	PVS_DST_MACRO_INST_SHIFT	= 7,
+	PVS_DST_REG_TYPE_MASK		= 0xf,
+	PVS_DST_REG_TYPE_SHIFT		= 8,
+	PVS_DST_ADDR_MODE_1_MASK	= 0x1,
+	PVS_DST_ADDR_MODE_1_SHIFT	= 12,
+	PVS_DST_OFFSET_MASK		= 0x7f,
+	PVS_DST_OFFSET_SHIFT		= 13,
+	PVS_DST_WE_X_MASK		= 0x1,
+	PVS_DST_WE_X_SHIFT		= 20,
+	PVS_DST_WE_Y_MASK		= 0x1,
+	PVS_DST_WE_Y_SHIFT		= 21,
+	PVS_DST_WE_Z_MASK		= 0x1,
+	PVS_DST_WE_Z_SHIFT		= 22,
+	PVS_DST_WE_W_MASK		= 0x1,
+	PVS_DST_WE_W_SHIFT		= 23,
+	PVS_DST_VE_SAT_MASK		= 0x1,
+	PVS_DST_VE_SAT_SHIFT		= 24,
+	PVS_DST_ME_SAT_MASK		= 0x1,
+	PVS_DST_ME_SAT_SHIFT		= 25,
+	PVS_DST_PRED_ENABLE_MASK	= 0x1,
+	PVS_DST_PRED_ENABLE_SHIFT	= 26,
+	PVS_DST_PRED_SENSE_MASK		= 0x1,
+	PVS_DST_PRED_SENSE_SHIFT	= 27,
+	PVS_DST_DUAL_MATH_OP_MASK	= 0x3,
+	PVS_DST_DUAL_MATH_OP_SHIFT	= 27,
+	PVS_DST_ADDR_SEL_MASK		= 0x3,
+	PVS_DST_ADDR_SEL_SHIFT		= 29,
+	PVS_DST_ADDR_MODE_0_MASK	= 0x1,
+	PVS_DST_ADDR_MODE_0_SHIFT	= 31,
+};
+
+/* PVS Source Operand Description */
+
+enum {
+	PVS_SRC_REG_TYPE_MASK		= 0x3,
+	PVS_SRC_REG_TYPE_SHIFT		= 0,
+	SPARE_0_MASK			= 0x1,
+	SPARE_0_SHIFT			= 2,
+	PVS_SRC_ABS_XYZW_MASK		= 0x1,
+	PVS_SRC_ABS_XYZW_SHIFT		= 3,
+	PVS_SRC_ADDR_MODE_0_MASK	= 0x1,
+	PVS_SRC_ADDR_MODE_0_SHIFT	= 4,
+	PVS_SRC_OFFSET_MASK		= 0xff,
+	PVS_SRC_OFFSET_SHIFT		= 5,
+	PVS_SRC_SWIZZLE_X_MASK		= 0x7,
+	PVS_SRC_SWIZZLE_X_SHIFT		= 13,
+	PVS_SRC_SWIZZLE_Y_MASK		= 0x7,
+	PVS_SRC_SWIZZLE_Y_SHIFT		= 16,
+	PVS_SRC_SWIZZLE_Z_MASK		= 0x7,
+	PVS_SRC_SWIZZLE_Z_SHIFT		= 19,
+	PVS_SRC_SWIZZLE_W_MASK		= 0x7,
+	PVS_SRC_SWIZZLE_W_SHIFT		= 22,
+	PVS_SRC_MODIFIER_X_MASK		= 0x1,
+	PVS_SRC_MODIFIER_X_SHIFT	= 25,
+	PVS_SRC_MODIFIER_Y_MASK		= 0x1,
+	PVS_SRC_MODIFIER_Y_SHIFT	= 26,
+	PVS_SRC_MODIFIER_Z_MASK		= 0x1,
+	PVS_SRC_MODIFIER_Z_SHIFT	= 27,
+	PVS_SRC_MODIFIER_W_MASK		= 0x1,
+	PVS_SRC_MODIFIER_W_SHIFT	= 28,
+	PVS_SRC_ADDR_SEL_MASK		= 0x3,
+	PVS_SRC_ADDR_SEL_SHIFT		= 29,
+	PVS_SRC_ADDR_MODE_1_MASK	= 0x0,
+	PVS_SRC_ADDR_MODE_1_SHIFT	= 32,
+};
+
+/*\}*/
 
 /* BEGIN: Packet 3 commands */
 
@@ -1601,13 +2679,511 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_PRIM_NUM_VERTICES_SHIFT            16
 #define R300_PRIM_NUM_VERTICES_MASK             0xffff
 
+
+
+/*
+ * The R500 unified shader (US) registers come in banks of 512 each, one
+ * for each instruction slot in the shader.  You can't touch them directly.
+ * R500_US_VECTOR_INDEX() sets the base instruction to modify; successive
+ * writes to R500_GA_US_VECTOR_DATA autoincrement the index after the
+ * instruction is fully specified.
+ */
+#define R500_US_ALU_ALPHA_INST_0			0xa800
+#   define R500_ALPHA_OP_MAD				0
+#   define R500_ALPHA_OP_DP				1
+#   define R500_ALPHA_OP_MIN				2
+#   define R500_ALPHA_OP_MAX				3
+/* #define R500_ALPHA_OP_RESERVED			4 */
+#   define R500_ALPHA_OP_CND				5
+#   define R500_ALPHA_OP_CMP				6
+#   define R500_ALPHA_OP_FRC				7
+#   define R500_ALPHA_OP_EX2				8
+#   define R500_ALPHA_OP_LN2				9
+#   define R500_ALPHA_OP_RCP				10
+#   define R500_ALPHA_OP_RSQ				11
+#   define R500_ALPHA_OP_SIN				12
+#   define R500_ALPHA_OP_COS				13
+#   define R500_ALPHA_OP_MDH				14
+#   define R500_ALPHA_OP_MDV				15
+#   define R500_ALPHA_ADDRD(x)				(x << 4)
+#   define R500_ALPHA_ADDRD_REL				(1 << 11)
+#  define R500_ALPHA_SEL_A_SHIFT			12
+#   define R500_ALPHA_SEL_A_SRC0			(0 << 12)
+#   define R500_ALPHA_SEL_A_SRC1			(1 << 12)
+#   define R500_ALPHA_SEL_A_SRC2			(2 << 12)
+#   define R500_ALPHA_SEL_A_SRCP			(3 << 12)
+#   define R500_ALPHA_SWIZ_A_R				(0 << 14)
+#   define R500_ALPHA_SWIZ_A_G				(1 << 14)
+#   define R500_ALPHA_SWIZ_A_B				(2 << 14)
+#   define R500_ALPHA_SWIZ_A_A				(3 << 14)
+#   define R500_ALPHA_SWIZ_A_0				(4 << 14)
+#   define R500_ALPHA_SWIZ_A_HALF			(5 << 14)
+#   define R500_ALPHA_SWIZ_A_1				(6 << 14)
+/* #define R500_ALPHA_SWIZ_A_UNUSED			(7 << 14) */
+#   define R500_ALPHA_MOD_A_NOP				(0 << 17)
+#   define R500_ALPHA_MOD_A_NEG				(1 << 17)
+#   define R500_ALPHA_MOD_A_ABS				(2 << 17)
+#   define R500_ALPHA_MOD_A_NAB				(3 << 17)
+#  define R500_ALPHA_SEL_B_SHIFT			19
+#   define R500_ALPHA_SEL_B_SRC0			(0 << 19)
+#   define R500_ALPHA_SEL_B_SRC1			(1 << 19)
+#   define R500_ALPHA_SEL_B_SRC2			(2 << 19)
+#   define R500_ALPHA_SEL_B_SRCP			(3 << 19)
+#   define R500_ALPHA_SWIZ_B_R				(0 << 21)
+#   define R500_ALPHA_SWIZ_B_G				(1 << 21)
+#   define R500_ALPHA_SWIZ_B_B				(2 << 21)
+#   define R500_ALPHA_SWIZ_B_A				(3 << 21)
+#   define R500_ALPHA_SWIZ_B_0				(4 << 21)
+#   define R500_ALPHA_SWIZ_B_HALF			(5 << 21)
+#   define R500_ALPHA_SWIZ_B_1				(6 << 21)
+/* #define R500_ALPHA_SWIZ_B_UNUSED			(7 << 21) */
+#   define R500_ALPHA_MOD_B_NOP				(0 << 24)
+#   define R500_ALPHA_MOD_B_NEG				(1 << 24)
+#   define R500_ALPHA_MOD_B_ABS				(2 << 24)
+#   define R500_ALPHA_MOD_B_NAB				(3 << 24)
+#   define R500_ALPHA_OMOD_IDENTITY			(0 << 26)
+#   define R500_ALPHA_OMOD_MUL_2			(1 << 26)
+#   define R500_ALPHA_OMOD_MUL_4			(2 << 26)
+#   define R500_ALPHA_OMOD_MUL_8			(3 << 26)
+#   define R500_ALPHA_OMOD_DIV_2			(4 << 26)
+#   define R500_ALPHA_OMOD_DIV_4			(5 << 26)
+#   define R500_ALPHA_OMOD_DIV_8			(6 << 26)
+#   define R500_ALPHA_OMOD_DISABLE			(7 << 26)
+#   define R500_ALPHA_TARGET(x)				(x << 29)
+#   define R500_ALPHA_W_OMASK				(1 << 31)
+#define R500_US_ALU_ALPHA_ADDR_0			0x9800
+#   define R500_ALPHA_ADDR0(x)				(x << 0)
+#   define R500_ALPHA_ADDR0_CONST			(1 << 8)
+#   define R500_ALPHA_ADDR0_REL				(1 << 9)
+#   define R500_ALPHA_ADDR1(x)				(x << 10)
+#   define R500_ALPHA_ADDR1_CONST			(1 << 18)
+#   define R500_ALPHA_ADDR1_REL				(1 << 19)
+#   define R500_ALPHA_ADDR2(x)				(x << 20)
+#   define R500_ALPHA_ADDR2_CONST			(1 << 28)
+#   define R500_ALPHA_ADDR2_REL				(1 << 29)
+#   define R500_ALPHA_SRCP_OP_1_MINUS_2A0		(0 << 30)
+#   define R500_ALPHA_SRCP_OP_A1_MINUS_A0		(1 << 30)
+#   define R500_ALPHA_SRCP_OP_A1_PLUS_A0		(2 << 30)
+#   define R500_ALPHA_SRCP_OP_1_MINUS_A0		(3 << 30)
+#define R500_US_ALU_RGBA_INST_0				0xb000
+#   define R500_ALU_RGBA_OP_MAD				(0 << 0)
+#   define R500_ALU_RGBA_OP_DP3				(1 << 0)
+#   define R500_ALU_RGBA_OP_DP4				(2 << 0)
+#   define R500_ALU_RGBA_OP_D2A				(3 << 0)
+#   define R500_ALU_RGBA_OP_MIN				(4 << 0)
+#   define R500_ALU_RGBA_OP_MAX				(5 << 0)
+/* #define R500_ALU_RGBA_OP_RESERVED			(6 << 0) */
+#   define R500_ALU_RGBA_OP_CND				(7 << 0)
+#   define R500_ALU_RGBA_OP_CMP				(8 << 0)
+#   define R500_ALU_RGBA_OP_FRC				(9 << 0)
+#   define R500_ALU_RGBA_OP_SOP				(10 << 0)
+#   define R500_ALU_RGBA_OP_MDH				(11 << 0)
+#   define R500_ALU_RGBA_OP_MDV				(12 << 0)
+#   define R500_ALU_RGBA_ADDRD(x)			(x << 4)
+#   define R500_ALU_RGBA_ADDRD_REL			(1 << 11)
+#  define R500_ALU_RGBA_SEL_C_SHIFT			12
+#   define R500_ALU_RGBA_SEL_C_SRC0			(0 << 12)
+#   define R500_ALU_RGBA_SEL_C_SRC1			(1 << 12)
+#   define R500_ALU_RGBA_SEL_C_SRC2			(2 << 12)
+#   define R500_ALU_RGBA_SEL_C_SRCP			(3 << 12)
+#   define R500_ALU_RGBA_R_SWIZ_R			(0 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_G			(1 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_B			(2 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_A			(3 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_0			(4 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_HALF			(5 << 14)
+#   define R500_ALU_RGBA_R_SWIZ_1			(6 << 14)
+/* #define R500_ALU_RGBA_R_SWIZ_UNUSED			(7 << 14) */
+#   define R500_ALU_RGBA_G_SWIZ_R			(0 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_G			(1 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_B			(2 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_A			(3 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_0			(4 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_HALF			(5 << 17)
+#   define R500_ALU_RGBA_G_SWIZ_1			(6 << 17)
+/* #define R500_ALU_RGBA_G_SWIZ_UNUSED			(7 << 17) */
+#   define R500_ALU_RGBA_B_SWIZ_R			(0 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_G			(1 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_B			(2 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_A			(3 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_0			(4 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_HALF			(5 << 20)
+#   define R500_ALU_RGBA_B_SWIZ_1			(6 << 20)
+/* #define R500_ALU_RGBA_B_SWIZ_UNUSED			(7 << 20) */
+#   define R500_ALU_RGBA_MOD_C_NOP			(0 << 23)
+#   define R500_ALU_RGBA_MOD_C_NEG			(1 << 23)
+#   define R500_ALU_RGBA_MOD_C_ABS			(2 << 23)
+#   define R500_ALU_RGBA_MOD_C_NAB			(3 << 23)
+#  define R500_ALU_RGBA_ALPHA_SEL_C_SHIFT		25
+#   define R500_ALU_RGBA_ALPHA_SEL_C_SRC0		(0 << 25)
+#   define R500_ALU_RGBA_ALPHA_SEL_C_SRC1		(1 << 25)
+#   define R500_ALU_RGBA_ALPHA_SEL_C_SRC2		(2 << 25)
+#   define R500_ALU_RGBA_ALPHA_SEL_C_SRCP		(3 << 25)
+#   define R500_ALU_RGBA_A_SWIZ_R			(0 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_G			(1 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_B			(2 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_A			(3 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_0			(4 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_HALF			(5 << 27)
+#   define R500_ALU_RGBA_A_SWIZ_1			(6 << 27)
+/* #define R500_ALU_RGBA_A_SWIZ_UNUSED			(7 << 27) */
+#   define R500_ALU_RGBA_ALPHA_MOD_C_NOP		(0 << 30)
+#   define R500_ALU_RGBA_ALPHA_MOD_C_NEG		(1 << 30)
+#   define R500_ALU_RGBA_ALPHA_MOD_C_ABS		(2 << 30)
+#   define R500_ALU_RGBA_ALPHA_MOD_C_NAB		(3 << 30)
+#define R500_US_ALU_RGB_INST_0				0xa000
+#  define R500_ALU_RGB_SEL_A_SHIFT			0
+#   define R500_ALU_RGB_SEL_A_SRC0			(0 << 0)
+#   define R500_ALU_RGB_SEL_A_SRC1			(1 << 0)
+#   define R500_ALU_RGB_SEL_A_SRC2			(2 << 0)
+#   define R500_ALU_RGB_SEL_A_SRCP			(3 << 0)
+#   define R500_ALU_RGB_R_SWIZ_A_R			(0 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_G			(1 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_B			(2 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_A			(3 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_0			(4 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_HALF			(5 << 2)
+#   define R500_ALU_RGB_R_SWIZ_A_1			(6 << 2)
+/* #define R500_ALU_RGB_R_SWIZ_A_UNUSED			(7 << 2) */
+#   define R500_ALU_RGB_G_SWIZ_A_R			(0 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_G			(1 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_B			(2 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_A			(3 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_0			(4 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_HALF			(5 << 5)
+#   define R500_ALU_RGB_G_SWIZ_A_1			(6 << 5)
+/* #define R500_ALU_RGB_G_SWIZ_A_UNUSED			(7 << 5) */
+#   define R500_ALU_RGB_B_SWIZ_A_R			(0 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_G			(1 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_B			(2 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_A			(3 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_0			(4 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_HALF			(5 << 8)
+#   define R500_ALU_RGB_B_SWIZ_A_1			(6 << 8)
+/* #define R500_ALU_RGB_B_SWIZ_A_UNUSED			(7 << 8) */
+#   define R500_ALU_RGB_MOD_A_NOP			(0 << 11)
+#   define R500_ALU_RGB_MOD_A_NEG			(1 << 11)
+#   define R500_ALU_RGB_MOD_A_ABS			(2 << 11)
+#   define R500_ALU_RGB_MOD_A_NAB			(3 << 11)
+#  define R500_ALU_RGB_SEL_B_SHIFT			13
+#   define R500_ALU_RGB_SEL_B_SRC0			(0 << 13)
+#   define R500_ALU_RGB_SEL_B_SRC1			(1 << 13)
+#   define R500_ALU_RGB_SEL_B_SRC2			(2 << 13)
+#   define R500_ALU_RGB_SEL_B_SRCP			(3 << 13)
+#   define R500_ALU_RGB_R_SWIZ_B_R			(0 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_G			(1 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_B			(2 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_A			(3 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_0			(4 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_HALF			(5 << 15)
+#   define R500_ALU_RGB_R_SWIZ_B_1			(6 << 15)
+/* #define R500_ALU_RGB_R_SWIZ_B_UNUSED			(7 << 15) */
+#   define R500_ALU_RGB_G_SWIZ_B_R			(0 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_G			(1 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_B			(2 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_A			(3 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_0			(4 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_HALF			(5 << 18)
+#   define R500_ALU_RGB_G_SWIZ_B_1			(6 << 18)
+/* #define R500_ALU_RGB_G_SWIZ_B_UNUSED			(7 << 18) */
+#   define R500_ALU_RGB_B_SWIZ_B_R			(0 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_G			(1 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_B			(2 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_A			(3 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_0			(4 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_HALF			(5 << 21)
+#   define R500_ALU_RGB_B_SWIZ_B_1			(6 << 21)
+/* #define R500_ALU_RGB_B_SWIZ_B_UNUSED			(7 << 21) */
+#   define R500_ALU_RGB_MOD_B_NOP			(0 << 24)
+#   define R500_ALU_RGB_MOD_B_NEG			(1 << 24)
+#   define R500_ALU_RGB_MOD_B_ABS			(2 << 24)
+#   define R500_ALU_RGB_MOD_B_NAB			(3 << 24)
+#   define R500_ALU_RGB_OMOD_IDENTITY			(0 << 26)
+#   define R500_ALU_RGB_OMOD_MUL_2			(1 << 26)
+#   define R500_ALU_RGB_OMOD_MUL_4			(2 << 26)
+#   define R500_ALU_RGB_OMOD_MUL_8			(3 << 26)
+#   define R500_ALU_RGB_OMOD_DIV_2			(4 << 26)
+#   define R500_ALU_RGB_OMOD_DIV_4			(5 << 26)
+#   define R500_ALU_RGB_OMOD_DIV_8			(6 << 26)
+#   define R500_ALU_RGB_OMOD_DISABLE			(7 << 26)
+#   define R500_ALU_RGB_TARGET(x)			(x << 29)
+#   define R500_ALU_RGB_WMASK				(1 << 31)
+#define R500_US_ALU_RGB_ADDR_0				0x9000
+#   define R500_RGB_ADDR0(x)				(x << 0)
+#   define R500_RGB_ADDR0_CONST				(1 << 8)
+#   define R500_RGB_ADDR0_REL				(1 << 9)
+#   define R500_RGB_ADDR1(x)				(x << 10)
+#   define R500_RGB_ADDR1_CONST				(1 << 18)
+#   define R500_RGB_ADDR1_REL				(1 << 19)
+#   define R500_RGB_ADDR2(x)				(x << 20)
+#   define R500_RGB_ADDR2_CONST				(1 << 28)
+#   define R500_RGB_ADDR2_REL				(1 << 29)
+#   define R500_RGB_SRCP_OP_1_MINUS_2RGB0		(0 << 30)
+#   define R500_RGB_SRCP_OP_RGB1_MINUS_RGB0		(1 << 30)
+#   define R500_RGB_SRCP_OP_RGB1_PLUS_RGB0		(2 << 30)
+#   define R500_RGB_SRCP_OP_1_MINUS_RGB0		(3 << 30)
+#define R500_US_CMN_INST_0				0xb800
+#  define R500_INST_TYPE_MASK				(3 << 0)
+#   define R500_INST_TYPE_ALU				(0 << 0)
+#   define R500_INST_TYPE_OUT				(1 << 0)
+#   define R500_INST_TYPE_FC				(2 << 0)
+#   define R500_INST_TYPE_TEX				(3 << 0)
+#   define R500_INST_TEX_SEM_WAIT			(1 << 2)
+#   define R500_INST_RGB_PRED_SEL_NONE			(0 << 3)
+#   define R500_INST_RGB_PRED_SEL_RGBA			(1 << 3)
+#   define R500_INST_RGB_PRED_SEL_RRRR			(2 << 3)
+#   define R500_INST_RGB_PRED_SEL_GGGG			(3 << 3)
+#   define R500_INST_RGB_PRED_SEL_BBBB			(4 << 3)
+#   define R500_INST_RGB_PRED_SEL_AAAA			(5 << 3)
+#   define R500_INST_RGB_PRED_INV			(1 << 6)
+#   define R500_INST_WRITE_INACTIVE			(1 << 7)
+#   define R500_INST_LAST				(1 << 8)
+#   define R500_INST_NOP				(1 << 9)
+#   define R500_INST_ALU_WAIT				(1 << 10)
+#   define R500_INST_RGB_WMASK_R			(1 << 11)
+#   define R500_INST_RGB_WMASK_G			(1 << 12)
+#   define R500_INST_RGB_WMASK_B			(1 << 13)
+#   define R500_INST_ALPHA_WMASK			(1 << 14)
+#   define R500_INST_RGB_OMASK_R			(1 << 15)
+#   define R500_INST_RGB_OMASK_G			(1 << 16)
+#   define R500_INST_RGB_OMASK_B			(1 << 17)
+#   define R500_INST_ALPHA_OMASK			(1 << 18)
+#   define R500_INST_RGB_CLAMP				(1 << 19)
+#   define R500_INST_ALPHA_CLAMP			(1 << 20)
+#   define R500_INST_ALU_RESULT_SEL			(1 << 21)
+#   define R500_INST_ALPHA_PRED_INV			(1 << 22)
+#   define R500_INST_ALU_RESULT_OP_EQ			(0 << 23)
+#   define R500_INST_ALU_RESULT_OP_LT			(1 << 23)
+#   define R500_INST_ALU_RESULT_OP_GE			(2 << 23)
+#   define R500_INST_ALU_RESULT_OP_NE			(3 << 23)
+#   define R500_INST_ALPHA_PRED_SEL_NONE		(0 << 25)
+#   define R500_INST_ALPHA_PRED_SEL_RGBA		(1 << 25)
+#   define R500_INST_ALPHA_PRED_SEL_RRRR		(2 << 25)
+#   define R500_INST_ALPHA_PRED_SEL_GGGG		(3 << 25)
+#   define R500_INST_ALPHA_PRED_SEL_BBBB		(4 << 25)
+#   define R500_INST_ALPHA_PRED_SEL_AAAA		(5 << 25)
+/* XXX next four are kind of guessed */
+#   define R500_INST_STAT_WE_R				(1 << 28)
+#   define R500_INST_STAT_WE_G				(1 << 29)
+#   define R500_INST_STAT_WE_B				(1 << 30)
+#   define R500_INST_STAT_WE_A				(1 << 31)
+
+/* note that these are 8 bit lengths, despite the offsets, at least for R500 */
+#define R500_US_CODE_ADDR				0x4630
+#   define R500_US_CODE_START_ADDR(x)			(x << 0)
+#   define R500_US_CODE_END_ADDR(x)			(x << 16)
+#define R500_US_CODE_OFFSET				0x4638
+#   define R500_US_CODE_OFFSET_ADDR(x)			(x << 0)
+#define R500_US_CODE_RANGE				0x4634
+#   define R500_US_CODE_RANGE_ADDR(x)			(x << 0)
+#   define R500_US_CODE_RANGE_SIZE(x)			(x << 16)
+#define R500_US_CONFIG					0x4600
+#   define R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO		(1 << 1)
+#define R500_US_FC_ADDR_0				0xa000
+#   define R500_FC_BOOL_ADDR(x)				(x << 0)
+#   define R500_FC_INT_ADDR(x)				(x << 8)
+#   define R500_FC_JUMP_ADDR(x)				(x << 16)
+#   define R500_FC_JUMP_GLOBAL				(1 << 31)
+#define R500_US_FC_BOOL_CONST				0x4620
+#   define R500_FC_KBOOL(x)				(x)
+#define R500_US_FC_CTRL					0x4624
+#   define R500_FC_TEST_EN				(1 << 30)
+#   define R500_FC_FULL_FC_EN				(1 << 31)
+#define R500_US_FC_INST_0				0x9800
+#   define R500_FC_OP_JUMP				(0 << 0)
+#   define R500_FC_OP_LOOP				(1 << 0)
+#   define R500_FC_OP_ENDLOOP				(2 << 0)
+#   define R500_FC_OP_REP				(3 << 0)
+#   define R500_FC_OP_ENDREP				(4 << 0)
+#   define R500_FC_OP_BREAKLOOP				(5 << 0)
+#   define R500_FC_OP_BREAKREP				(6 << 0)
+#   define R500_FC_OP_CONTINUE				(7 << 0)
+#   define R500_FC_B_ELSE				(1 << 4)
+#   define R500_FC_JUMP_ANY				(1 << 5)
+#   define R500_FC_A_OP_NONE				(0 << 6)
+#   define R500_FC_A_OP_POP				(1 << 6)
+#   define R500_FC_A_OP_PUSH				(2 << 6)
+#   define R500_FC_JUMP_FUNC(x)				(x << 8)
+#   define R500_FC_B_POP_CNT(x)				(x << 16)
+#   define R500_FC_B_OP0_NONE				(0 << 24)
+#   define R500_FC_B_OP0_DECR				(1 << 24)
+#   define R500_FC_B_OP0_INCR				(2 << 24)
+#   define R500_FC_B_OP1_DECR				(0 << 26)
+#   define R500_FC_B_OP1_NONE				(1 << 26)
+#   define R500_FC_B_OP1_INCR				(2 << 26)
+#   define R500_FC_IGNORE_UNCOVERED			(1 << 28)
+#define R500_US_FC_INT_CONST_0				0x4c00
+#   define R500_FC_INT_CONST_KR(x)			(x << 0)
+#   define R500_FC_INT_CONST_KG(x)			(x << 8)
+#   define R500_FC_INT_CONST_KB(x)			(x << 16)
+/* _0 through _15 */
+#define R500_US_FORMAT0_0				0x4640
+#   define R500_FORMAT_TXWIDTH(x)			(x << 0)
+#   define R500_FORMAT_TXHEIGHT(x)			(x << 11)
+#   define R500_FORMAT_TXDEPTH(x)			(x << 22)
+/* _0 through _3 */
+#define R500_US_OUT_FMT_0				0x46a4
+#   define R500_OUT_FMT_C4_8				(0 << 0)
+#   define R500_OUT_FMT_C4_10				(1 << 0)
+#   define R500_OUT_FMT_C4_10_GAMMA			(2 << 0)
+#   define R500_OUT_FMT_C_16				(3 << 0)
+#   define R500_OUT_FMT_C2_16				(4 << 0)
+#   define R500_OUT_FMT_C4_16				(5 << 0)
+#   define R500_OUT_FMT_C_16_MPEG			(6 << 0)
+#   define R500_OUT_FMT_C2_16_MPEG			(7 << 0)
+#   define R500_OUT_FMT_C2_4				(8 << 0)
+#   define R500_OUT_FMT_C_3_3_2				(9 << 0)
+#   define R500_OUT_FMT_C_6_5_6				(10 << 0)
+#   define R500_OUT_FMT_C_11_11_10			(11 << 0)
+#   define R500_OUT_FMT_C_10_11_11			(12 << 0)
+#   define R500_OUT_FMT_C_2_10_10_10			(13 << 0)
+/* #define R500_OUT_FMT_RESERVED			(14 << 0) */
+#   define R500_OUT_FMT_UNUSED				(15 << 0)
+#   define R500_OUT_FMT_C_16_FP				(16 << 0)
+#   define R500_OUT_FMT_C2_16_FP			(17 << 0)
+#   define R500_OUT_FMT_C4_16_FP			(18 << 0)
+#   define R500_OUT_FMT_C_32_FP				(19 << 0)
+#   define R500_OUT_FMT_C2_32_FP			(20 << 0)
+#   define R500_OUT_FMT_C4_32_FP			(21 << 0)
+#   define R500_C0_SEL_A				(0 << 8)
+#   define R500_C0_SEL_R				(1 << 8)
+#   define R500_C0_SEL_G				(2 << 8)
+#   define R500_C0_SEL_B				(3 << 8)
+#   define R500_C1_SEL_A				(0 << 10)
+#   define R500_C1_SEL_R				(1 << 10)
+#   define R500_C1_SEL_G				(2 << 10)
+#   define R500_C1_SEL_B				(3 << 10)
+#   define R500_C2_SEL_A				(0 << 12)
+#   define R500_C2_SEL_R				(1 << 12)
+#   define R500_C2_SEL_G				(2 << 12)
+#   define R500_C2_SEL_B				(3 << 12)
+#   define R500_C3_SEL_A				(0 << 14)
+#   define R500_C3_SEL_R				(1 << 14)
+#   define R500_C3_SEL_G				(2 << 14)
+#   define R500_C3_SEL_B				(3 << 14)
+#   define R500_OUT_SIGN(x)				(x << 16)
+#   define R500_ROUND_ADJ				(1 << 20)
+#define R500_US_PIXSIZE					0x4604
+#   define R500_PIX_SIZE(x)				(x)
+#define R500_US_TEX_ADDR_0				0x9800
+#   define R500_TEX_SRC_ADDR(x)				(x << 0)
+#   define R500_TEX_SRC_ADDR_REL			(1 << 7)
+#   define R500_TEX_SRC_S_SWIZ_R			(0 << 8)
+#   define R500_TEX_SRC_S_SWIZ_G			(1 << 8)
+#   define R500_TEX_SRC_S_SWIZ_B			(2 << 8)
+#   define R500_TEX_SRC_S_SWIZ_A			(3 << 8)
+#   define R500_TEX_SRC_T_SWIZ_R			(0 << 10)
+#   define R500_TEX_SRC_T_SWIZ_G			(1 << 10)
+#   define R500_TEX_SRC_T_SWIZ_B			(2 << 10)
+#   define R500_TEX_SRC_T_SWIZ_A			(3 << 10)
+#   define R500_TEX_SRC_R_SWIZ_R			(0 << 12)
+#   define R500_TEX_SRC_R_SWIZ_G			(1 << 12)
+#   define R500_TEX_SRC_R_SWIZ_B			(2 << 12)
+#   define R500_TEX_SRC_R_SWIZ_A			(3 << 12)
+#   define R500_TEX_SRC_Q_SWIZ_R			(0 << 14)
+#   define R500_TEX_SRC_Q_SWIZ_G			(1 << 14)
+#   define R500_TEX_SRC_Q_SWIZ_B			(2 << 14)
+#   define R500_TEX_SRC_Q_SWIZ_A			(3 << 14)
+#   define R500_TEX_DST_ADDR(x)				(x << 16)
+#   define R500_TEX_DST_ADDR_REL			(1 << 23)
+#   define R500_TEX_DST_R_SWIZ_R			(0 << 24)
+#   define R500_TEX_DST_R_SWIZ_G			(1 << 24)
+#   define R500_TEX_DST_R_SWIZ_B			(2 << 24)
+#   define R500_TEX_DST_R_SWIZ_A			(3 << 24)
+#   define R500_TEX_DST_G_SWIZ_R			(0 << 26)
+#   define R500_TEX_DST_G_SWIZ_G			(1 << 26)
+#   define R500_TEX_DST_G_SWIZ_B			(2 << 26)
+#   define R500_TEX_DST_G_SWIZ_A			(3 << 26)
+#   define R500_TEX_DST_B_SWIZ_R			(0 << 28)
+#   define R500_TEX_DST_B_SWIZ_G			(1 << 28)
+#   define R500_TEX_DST_B_SWIZ_B			(2 << 28)
+#   define R500_TEX_DST_B_SWIZ_A			(3 << 28)
+#   define R500_TEX_DST_A_SWIZ_R			(0 << 30)
+#   define R500_TEX_DST_A_SWIZ_G			(1 << 30)
+#   define R500_TEX_DST_A_SWIZ_B			(2 << 30)
+#   define R500_TEX_DST_A_SWIZ_A			(3 << 30)
+#define R500_US_TEX_ADDR_DXDY_0				0xa000
+#   define R500_DX_ADDR(x)				(x << 0)
+#   define R500_DX_ADDR_REL				(1 << 7)
+#   define R500_DX_S_SWIZ_R				(0 << 8)
+#   define R500_DX_S_SWIZ_G				(1 << 8)
+#   define R500_DX_S_SWIZ_B				(2 << 8)
+#   define R500_DX_S_SWIZ_A				(3 << 8)
+#   define R500_DX_T_SWIZ_R				(0 << 10)
+#   define R500_DX_T_SWIZ_G				(1 << 10)
+#   define R500_DX_T_SWIZ_B				(2 << 10)
+#   define R500_DX_T_SWIZ_A				(3 << 10)
+#   define R500_DX_R_SWIZ_R				(0 << 12)
+#   define R500_DX_R_SWIZ_G				(1 << 12)
+#   define R500_DX_R_SWIZ_B				(2 << 12)
+#   define R500_DX_R_SWIZ_A				(3 << 12)
+#   define R500_DX_Q_SWIZ_R				(0 << 14)
+#   define R500_DX_Q_SWIZ_G				(1 << 14)
+#   define R500_DX_Q_SWIZ_B				(2 << 14)
+#   define R500_DX_Q_SWIZ_A				(3 << 14)
+#   define R500_DY_ADDR(x)				(x << 16)
+#   define R500_DY_ADDR_REL				(1 << 17)
+#   define R500_DY_S_SWIZ_R				(0 << 24)
+#   define R500_DY_S_SWIZ_G				(1 << 24)
+#   define R500_DY_S_SWIZ_B				(2 << 24)
+#   define R500_DY_S_SWIZ_A				(3 << 24)
+#   define R500_DY_T_SWIZ_R				(0 << 26)
+#   define R500_DY_T_SWIZ_G				(1 << 26)
+#   define R500_DY_T_SWIZ_B				(2 << 26)
+#   define R500_DY_T_SWIZ_A				(3 << 26)
+#   define R500_DY_R_SWIZ_R				(0 << 28)
+#   define R500_DY_R_SWIZ_G				(1 << 28)
+#   define R500_DY_R_SWIZ_B				(2 << 28)
+#   define R500_DY_R_SWIZ_A				(3 << 28)
+#   define R500_DY_Q_SWIZ_R				(0 << 30)
+#   define R500_DY_Q_SWIZ_G				(1 << 30)
+#   define R500_DY_Q_SWIZ_B				(2 << 30)
+#   define R500_DY_Q_SWIZ_A				(3 << 30)
+#define R500_US_TEX_INST_0				0x9000
+#   define R500_TEX_ID(x)				(x << 16)
+#   define R500_TEX_INST_NOP				(0 << 22)
+#   define R500_TEX_INST_LD				(1 << 22)
+#   define R500_TEX_INST_TEXKILL			(2 << 22)
+#   define R500_TEX_INST_PROJ				(3 << 22)
+#   define R500_TEX_INST_LODBIAS			(4 << 22)
+#   define R500_TEX_INST_LOD				(5 << 22)
+#   define R500_TEX_INST_DXDY				(6 << 22)
+#   define R500_TEX_SEM_ACQUIRE				(1 << 25)
+#   define R500_TEX_IGNORE_UNCOVERED			(1 << 26)
+#   define R500_TEX_UNSCALED				(1 << 27)
+#define R300_US_W_FMT					0x46b4
+#   define R300_W_FMT_W0				(0 << 0)
+#   define R300_W_FMT_W24				(1 << 0)
+#   define R300_W_FMT_W24FP				(2 << 0)
+#   define R300_W_SRC_US				(0 << 2)
+#   define R300_W_SRC_RAS				(1 << 2)
+
+
 /* Draw a primitive from vertex data in arrays loaded via 3D_LOAD_VBPNTR.
  * Two parameter dwords:
- * 0. The first parameter appears to be always 0
- * 1. The second parameter is a standard primitive emission dword.
+ * 0. VAP_VTX_FMT: The first parameter is not written to hardware
+ * 1. VAP_VF_CTL: The second parameter is a standard primitive emission dword.
  */
 #define R300_PACKET3_3D_DRAW_VBUF           0x00002800
 
+/* Draw a primitive from immediate vertices in this packet
+ * Up to 16382 dwords:
+ * 0. VAP_VTX_FMT: The first parameter is not written to hardware
+ * 1. VAP_VF_CTL: The second parameter is a standard primitive emission dword.
+ * 2 to end: Up to 16380 dwords of vertex data.
+ */
+#define R300_PACKET3_3D_DRAW_IMMD           0x00002900
+
+/* Draw a primitive from vertex data in arrays loaded via 3D_LOAD_VBPNTR and
+ * immediate vertices in this packet
+ * Up to 16382 dwords:
+ * 0. VAP_VTX_FMT: The first parameter is not written to hardware
+ * 1. VAP_VF_CTL: The second parameter is a standard primitive emission dword.
+ * 2 to end: Up to 16380 dwords of vertex data.
+ */
+#define R300_PACKET3_3D_DRAW_INDX           0x00002A00
+
+
 /* Specify the full set of vertex arrays as (address, stride).
  * The first parameter is the number of vertex arrays specified.
  * The rest of the command is a variable length list of blocks, where
@@ -1628,9 +3204,29 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #    define R300_EB_UNK1_SHIFT                      24
 #    define R300_EB_UNK1                    (0x80<<24)
 #    define R300_EB_UNK2                        0x0810
+
+/* Same as R300_PACKET3_3D_DRAW_VBUF but without VAP_VTX_FMT */
 #define R300_PACKET3_3D_DRAW_VBUF_2         0x00003400
+/* Same as R300_PACKET3_3D_DRAW_IMMD but without VAP_VTX_FMT */
+#define R300_PACKET3_3D_DRAW_IMMD_2         0x00003500
+/* Same as R300_PACKET3_3D_DRAW_INDX but without VAP_VTX_FMT */
 #define R300_PACKET3_3D_DRAW_INDX_2         0x00003600
 
+/* Clears a portion of hierachical Z RAM
+ * 3 dword parameters
+ * 0. START
+ * 1. COUNT: 13:0 (max is 0x3FFF)
+ * 2. CLEAR_VALUE: Value to write into HIZ RAM.
+ */
+#define R300_PACKET3_3D_CLEAR_HIZ           0x00003700
+
+/* Draws a set of primitives using vertex buffers pointed by the state data.
+ * At least 2 Parameters:
+ * 0. VAP_VF_CNTL: The first parameter is a standard primitive emission dword.
+ * 2 to end: Data or indices (see other 3D_DRAW_* packets for details)
+ */
+#define R300_PACKET3_3D_DRAW_128            0x00003900
+
 /* END: Packet 3 commands */
 
 
@@ -1652,3 +3248,5 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #endif /* _R300_REG_H */
 
 /* *INDENT-ON* */
+
+/* vim: set foldenable foldmarker=\\{,\\} foldmethod=marker : */
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index c809679e6cd..292f87a2b1b 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -50,15 +50,15 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * no bugs...
  */
 
-#include "glheader.h"
-#include "state.h"
-#include "imports.h"
-#include "enums.h"
-#include "macros.h"
-#include "context.h"
-#include "dd.h"
-#include "simple_list.h"
-#include "api_arrayelt.h"
+#include "main/glheader.h"
+#include "main/state.h"
+#include "main/imports.h"
+#include "main/enums.h"
+#include "main/macros.h"
+#include "main/context.h"
+#include "main/dd.h"
+#include "main/simple_list.h"
+#include "main/api_arrayelt.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "vbo/vbo.h"
@@ -74,6 +74,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r300_reg.h"
 #include "r300_tex.h"
 #include "r300_emit.h"
+#include "r300_fragprog.h"
 extern int future_hw_tcl_on;
 
 /**
@@ -268,13 +269,24 @@ static void r300RunRenderPrimitive(r300ContextPtr rmesa, GLcontext * ctx,
 		return;
 
 	if (vb->Elts) {
-		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
 		if (num_verts > 65535) {
 			/* not implemented yet */
 			WARN_ONCE("Too many elts\n");
 			return;
 		}
+		/* Note: The following is incorrect, but it's the best I can do
+		 * without a major refactoring of how DMA memory is handled.
+		 * The problem: Ensuring that both vertex arrays *and* index
+		 * arrays are at the right position, and then ensuring that
+		 * the LOAD_VBPNTR, DRAW_INDX and INDX_BUFFER packets are emitted
+		 * at once.
+		 *
+		 * So why is the following incorrect? Well, it seems like
+		 * allocating the index array might actually evict the vertex
+		 * arrays. *sigh*
+		 */
 		r300EmitElts(ctx, vb->Elts, num_verts);
+		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
 		r300FireEB(rmesa, rmesa->state.elt_dma.aos_offset, num_verts, type);
 	} else {
 		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
@@ -334,13 +346,26 @@ static GLboolean r300RunRender(GLcontext * ctx,
 static int r300Fallback(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	struct r300_fragment_program *fp = (struct r300_fragment_program *)
+	/* Do we need to use new-style shaders?
+	 * Also is there a better way to do this? */
+	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+		struct r500_fragment_program *fp = (struct r500_fragment_program *)
 	    (char *)ctx->FragmentProgram._Current;
-
-	if (fp) {
-		if (!fp->translated)
-			r300TranslateFragmentShader(r300, fp);
-		FALLBACK_IF(!fp->translated);
+		if (fp) {
+			if (!fp->translated) {
+				r500TranslateFragmentShader(r300, fp);
+				FALLBACK_IF(!fp->translated);
+			}
+		}
+	} else {
+		struct r300_fragment_program *fp = (struct r300_fragment_program *)
+	    (char *)ctx->FragmentProgram._Current;
+		if (fp) {
+			if (!fp->translated) {
+				r300TranslateFragmentShader(r300, fp);
+				FALLBACK_IF(!fp->translated);
+			}
+		}
 	}
 
 	FALLBACK_IF(ctx->RenderMode != GL_RENDER);
@@ -352,8 +377,6 @@ static int r300Fallback(GLcontext * ctx)
 			|| ctx->Stencil.WriteMask[0] !=
 			ctx->Stencil.WriteMask[1]));
 
-	FALLBACK_IF(ctx->Color.ColorLogicOpEnabled);
-
 	if (ctx->Extensions.NV_point_sprite || ctx->Extensions.ARB_point_sprite)
 		FALLBACK_IF(ctx->Point.PointSprite);
 
diff --git a/src/mesa/drivers/dri/r300/r300_shader.c b/src/mesa/drivers/dri/r300/r300_shader.c
index 5f5ac7c4c71..f30fd986e0f 100644
--- a/src/mesa/drivers/dri/r300/r300_shader.c
+++ b/src/mesa/drivers/dri/r300/r300_shader.c
@@ -1,8 +1,7 @@
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
 
-#include "program.h"
+#include "main/glheader.h"
+
+#include "shader/program.h"
 #include "tnl/tnl.h"
 #include "r300_context.h"
 #include "r300_fragprog.h"
@@ -10,8 +9,10 @@
 static struct gl_program *r300NewProgram(GLcontext * ctx, GLenum target,
 					 GLuint id)
 {
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct r300_vertex_program_cont *vp;
-	struct r300_fragment_program *fp;
+	struct r300_fragment_program *r300_fp;
+	struct r500_fragment_program *r500_fp;
 
 	switch (target) {
 	case GL_VERTEX_STATE_PROGRAM_NV:
@@ -20,14 +21,27 @@ static struct gl_program *r300NewProgram(GLcontext * ctx, GLenum target,
 		return _mesa_init_vertex_program(ctx, &vp->mesa_program,
 						 target, id);
 	case GL_FRAGMENT_PROGRAM_ARB:
-		fp = CALLOC_STRUCT(r300_fragment_program);
-		fp->ctx = ctx;
-		return _mesa_init_fragment_program(ctx, &fp->mesa_program,
-						   target, id);
+		if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+			r500_fp = CALLOC_STRUCT(r500_fragment_program);
+			r500_fp->ctx = ctx;
+			return _mesa_init_fragment_program(ctx, &r500_fp->mesa_program,
+							   target, id);
+		} else {
+			r300_fp = CALLOC_STRUCT(r300_fragment_program);
+			return _mesa_init_fragment_program(ctx, &r300_fp->mesa_program,
+							   target, id);
+		}
+
 	case GL_FRAGMENT_PROGRAM_NV:
-		fp = CALLOC_STRUCT(r300_fragment_program);
-		return _mesa_init_fragment_program(ctx, &fp->mesa_program,
-						   target, id);
+		if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+			r500_fp = CALLOC_STRUCT(r500_fragment_program);
+			return _mesa_init_fragment_program(ctx, &r500_fp->mesa_program,
+							   target, id);
+		} else {
+			r300_fp = CALLOC_STRUCT(r300_fragment_program);
+			return _mesa_init_fragment_program(ctx, &r300_fp->mesa_program,
+							   target, id);
+		}
 	default:
 		_mesa_problem(ctx, "Bad target in r300NewProgram");
 	}
@@ -43,15 +57,20 @@ static void r300DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 static void
 r300ProgramStringNotify(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct r300_vertex_program_cont *vp = (void *)prog;
-	struct r300_fragment_program *fp = (struct r300_fragment_program *)prog;
+	struct r300_fragment_program *r300_fp = (struct r300_fragment_program *)prog;
+	struct r500_fragment_program *r500_fp = (struct r500_fragment_program *)prog;
 
 	switch (target) {
 	case GL_VERTEX_PROGRAM_ARB:
 		vp->progs = NULL;
 		break;
 	case GL_FRAGMENT_PROGRAM_ARB:
-		fp->translated = GL_FALSE;
+		if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+			r500_fp->translated = GL_FALSE;
+		else
+			r300_fp->translated = GL_FALSE;
 		break;
 	}
 
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index 088216c76ee..6a5c3633a23 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -35,23 +35,23 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Nicolai Haehnle <prefect_@gmx.net>
  */
 
-#include "glheader.h"
-#include "state.h"
-#include "imports.h"
-#include "enums.h"
-#include "macros.h"
-#include "context.h"
-#include "dd.h"
-#include "simple_list.h"
-
-#include "api_arrayelt.h"
+#include "main/glheader.h"
+#include "main/state.h"
+#include "main/imports.h"
+#include "main/enums.h"
+#include "main/macros.h"
+#include "main/context.h"
+#include "main/dd.h"
+#include "main/simple_list.h"
+#include "main/api_arrayelt.h"
+#include "main/texformat.h"
+
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
-#include "texformat.h"
 
 #include "radeon_ioctl.h"
 #include "radeon_state.h"
@@ -70,20 +70,28 @@ extern void _tnl_UpdateFixedFunctionProgram(GLcontext * ctx);
 
 static void r300BlendColor(GLcontext * ctx, const GLfloat cf[4])
 {
-	GLubyte color[4];
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 
 	R300_STATECHANGE(rmesa, blend_color);
 
-	CLAMPED_FLOAT_TO_UBYTE(color[0], cf[0]);
-	CLAMPED_FLOAT_TO_UBYTE(color[1], cf[1]);
-	CLAMPED_FLOAT_TO_UBYTE(color[2], cf[2]);
-	CLAMPED_FLOAT_TO_UBYTE(color[3], cf[3]);
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+		GLuint r = IROUND(cf[0]*1023.0f);
+		GLuint g = IROUND(cf[1]*1023.0f);
+		GLuint b = IROUND(cf[2]*1023.0f);
+		GLuint a = IROUND(cf[3]*1023.0f);
+
+		rmesa->hw.blend_color.cmd[1] = r | (a << 16);
+		rmesa->hw.blend_color.cmd[2] = b | (g << 16);
+	} else {
+		GLubyte color[4];
+		CLAMPED_FLOAT_TO_UBYTE(color[0], cf[0]);
+		CLAMPED_FLOAT_TO_UBYTE(color[1], cf[1]);
+		CLAMPED_FLOAT_TO_UBYTE(color[2], cf[2]);
+		CLAMPED_FLOAT_TO_UBYTE(color[3], cf[3]);
 
-	rmesa->hw.blend_color.cmd[1] = PACK_COLOR_8888(color[3], color[0],
-						       color[1], color[2]);
-	rmesa->hw.blend_color.cmd[2] = 0;
-	rmesa->hw.blend_color.cmd[3] = 0;
+		rmesa->hw.blend_color.cmd[1] = PACK_COLOR_8888(color[3], color[0],
+							color[1], color[2]);
+	}
 }
 
 /**
@@ -189,7 +197,7 @@ static void r300SetBlendCntl(r300ContextPtr r300, int func, int eqn,
 	 */
 #if 0
 	if (new_ablend == new_cblend) {
-		new_cblend |= R300_BLEND_NO_SEPARATE;
+		new_cblend |= R300_DISCARD_SRC_PIXELS_SRC_ALPHA_0;
 	}
 #endif
 	new_cblend |= cbits;
@@ -295,7 +303,9 @@ static void r300SetBlendState(GLcontext * ctx)
 
 	r300SetBlendCntl(r300,
 			 func, eqn,
-			 R300_BLEND_UNKNOWN | R300_BLEND_ENABLE, funcA, eqnA);
+			 (R300_SEPARATE_ALPHA_ENABLE |
+			  R300_READ_ENABLE |
+			  R300_ALPHA_BLEND_ENABLE), funcA, eqnA);
 }
 
 static void r300BlendEquationSeparate(GLcontext * ctx,
@@ -312,6 +322,83 @@ static void r300BlendFuncSeparate(GLcontext * ctx,
 }
 
 /**
+ * Translate LogicOp enums into hardware representation.
+ * Both use a very logical bit-wise layout, but unfortunately the order
+ * of bits is reversed.
+ */
+static GLuint translate_logicop(GLenum logicop)
+{
+	GLuint bits = logicop - GL_CLEAR;
+	bits = ((bits & 1) << 3) | ((bits & 2) << 1) | ((bits & 4) >> 1) | ((bits & 8) >> 3);
+	return bits << R300_RB3D_ROPCNTL_ROP_SHIFT;
+}
+
+/**
+ * Used internally to update the r300->hw hardware state to match the
+ * current OpenGL state.
+ */
+static void r300SetLogicOpState(GLcontext *ctx)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	R300_STATECHANGE(r300, rop);
+	if (RGBA_LOGICOP_ENABLED(ctx)) {
+		r300->hw.rop.cmd[1] = R300_RB3D_ROPCNTL_ROP_ENABLE |
+			translate_logicop(ctx->Color.LogicOp);
+	} else {
+		r300->hw.rop.cmd[1] = 0;
+	}
+}
+
+/**
+ * Called by Mesa when an application program changes the LogicOp state
+ * via glLogicOp.
+ */
+static void r300LogicOpcode(GLcontext *ctx, GLenum logicop)
+{
+	if (RGBA_LOGICOP_ENABLED(ctx))
+		r300SetLogicOpState(ctx);
+}
+
+static void r300ClipPlane( GLcontext *ctx, GLenum plane, const GLfloat *eq )
+{
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
+	GLint p;
+	GLint *ip;
+
+	/* no VAP UCP on non-TCL chipsets */
+	if (!(rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
+			return;
+
+	p = (GLint) plane - (GLint) GL_CLIP_PLANE0;
+	ip = (GLint *)ctx->Transform._ClipUserPlane[p];
+
+	R300_STATECHANGE( rmesa, vpucp[p] );
+	rmesa->hw.vpucp[p].cmd[R300_VPUCP_X] = ip[0];
+	rmesa->hw.vpucp[p].cmd[R300_VPUCP_Y] = ip[1];
+	rmesa->hw.vpucp[p].cmd[R300_VPUCP_Z] = ip[2];
+	rmesa->hw.vpucp[p].cmd[R300_VPUCP_W] = ip[3];
+}
+
+static void r300SetClipPlaneState(GLcontext * ctx, GLenum cap, GLboolean state)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	GLuint p;
+
+	/* no VAP UCP on non-TCL chipsets */
+	if (!(r300->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
+		return;
+
+	p = cap - GL_CLIP_PLANE0;
+	R300_STATECHANGE(r300, vap_clip_cntl);
+	if (state) {
+		r300->hw.vap_clip_cntl.cmd[1] |= (R300_VAP_UCP_ENABLE_0 << p);
+		r300ClipPlane(ctx, cap, NULL);
+	} else {
+		r300->hw.vap_clip_cntl.cmd[1] &= ~(R300_VAP_UCP_ENABLE_0 << p);
+	}
+}
+
+/**
  * Update our tracked culling state based on Mesa's state.
  */
 static void r300UpdateCulling(GLcontext * ctx)
@@ -350,44 +437,52 @@ static void r300UpdateCulling(GLcontext * ctx)
 	r300->hw.cul.cmd[R300_CUL_CULL] = val;
 }
 
-static void r300SetEarlyZState(GLcontext * ctx)
+static void r300SetPolygonOffsetState(GLcontext * ctx, GLboolean state)
 {
-	/* updates register R300_RB3D_EARLY_Z (0x4F14)
-	   if depth test is not enabled it should be R300_EARLY_Z_DISABLE
-	   if depth is enabled and alpha not it should be R300_EARLY_Z_ENABLE
-	   if depth and alpha is enabled it should be R300_EARLY_Z_DISABLE
-	 */
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
-	R300_STATECHANGE(r300, zstencil_format);
-	switch (ctx->Visual.depthBits) {
-	case 16:
-		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z;
-		break;
-	case 24:
-		r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z;
-		break;
-	default:
-		fprintf(stderr, "Error: Unsupported depth %d... exiting\n", ctx->Visual.depthBits);
-		_mesa_exit(-1);
+	R300_STATECHANGE(r300, occlusion_cntl);
+	if (state) {
+		r300->hw.occlusion_cntl.cmd[1] |= (3 << 0);
+	} else {
+		r300->hw.occlusion_cntl.cmd[1] &= ~(3 << 0);
 	}
+}
 
-	// r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32;
+static GLboolean current_fragment_program_writes_depth(GLcontext* ctx)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
-	if (ctx->Color.AlphaEnabled && ctx->Color.AlphaFunc != GL_ALWAYS)
-		/* disable early Z */
-		r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
-	else {
-		if (ctx->Depth.Test && ctx->Depth.Func != GL_NEVER)
-			/* enable early Z */
-			r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_ENABLE;
-		else
-			/* disable early Z */
-			r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
+	if (r300->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV515) {
+		struct r300_fragment_program *fp = (struct r300_fragment_program *)
+			(char *)ctx->FragmentProgram._Current;
+		return (fp && fp->WritesDepth);
+	} else {
+		struct r500_fragment_program* fp =
+			(struct r500_fragment_program*)(char*)
+			ctx->FragmentProgram._Current;
+		return (fp && fp->writes_depth);
 	}
+}
 
-	r300->hw.zstencil_format.cmd[3] = 0x00000003;
-	r300->hw.zstencil_format.cmd[4] = 0x00000000;
+static void r300SetEarlyZState(GLcontext * ctx)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	GLuint topZ = R300_ZTOP_ENABLE;
+
+	if (ctx->Color.AlphaEnabled && ctx->Color.AlphaFunc != GL_ALWAYS)
+		topZ = R300_ZTOP_DISABLE;
+	if (current_fragment_program_writes_depth(ctx))
+		topZ = R300_ZTOP_DISABLE;
+
+	if (topZ != r300->hw.zstencil_format.cmd[2]) {
+		/* Note: This completely reemits the stencil format.
+		 * I have not tested whether this is strictly necessary,
+		 * or if emitting a write to ZB_ZTOP is enough.
+		 */
+		R300_STATECHANGE(r300, zstencil_format);
+		r300->hw.zstencil_format.cmd[2] = topZ;
+	}
 }
 
 static void r300SetAlphaState(GLcontext * ctx)
@@ -401,35 +496,36 @@ static void r300SetAlphaState(GLcontext * ctx)
 
 	switch (ctx->Color.AlphaFunc) {
 	case GL_NEVER:
-		pp_misc |= R300_ALPHA_TEST_FAIL;
+		pp_misc |= R300_FG_ALPHA_FUNC_NEVER;
 		break;
 	case GL_LESS:
-		pp_misc |= R300_ALPHA_TEST_LESS;
+		pp_misc |= R300_FG_ALPHA_FUNC_LESS;
 		break;
 	case GL_EQUAL:
-		pp_misc |= R300_ALPHA_TEST_EQUAL;
+		pp_misc |= R300_FG_ALPHA_FUNC_EQUAL;
 		break;
 	case GL_LEQUAL:
-		pp_misc |= R300_ALPHA_TEST_LEQUAL;
+		pp_misc |= R300_FG_ALPHA_FUNC_LE;
 		break;
 	case GL_GREATER:
-		pp_misc |= R300_ALPHA_TEST_GREATER;
+		pp_misc |= R300_FG_ALPHA_FUNC_GREATER;
 		break;
 	case GL_NOTEQUAL:
-		pp_misc |= R300_ALPHA_TEST_NEQUAL;
+		pp_misc |= R300_FG_ALPHA_FUNC_NOTEQUAL;
 		break;
 	case GL_GEQUAL:
-		pp_misc |= R300_ALPHA_TEST_GEQUAL;
+		pp_misc |= R300_FG_ALPHA_FUNC_GE;
 		break;
 	case GL_ALWAYS:
-		/*pp_misc |= R300_ALPHA_TEST_PASS; */
+		/*pp_misc |= FG_ALPHA_FUNC_ALWAYS; */
 		really_enabled = GL_FALSE;
 		break;
 	}
 
 	if (really_enabled) {
-		pp_misc |= R300_ALPHA_TEST_ENABLE;
-		pp_misc |= (refByte & R300_REF_ALPHA_MASK);
+		pp_misc |= R300_FG_ALPHA_FUNC_ENABLE;
+		pp_misc |= R500_FG_ALPHA_FUNC_8BIT;
+		pp_misc |= (refByte & R300_FG_ALPHA_FUNC_VAL_MASK);
 	} else {
 		pp_misc = 0x0;
 	}
@@ -476,38 +572,53 @@ static void r300SetDepthState(GLcontext * ctx)
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 
 	R300_STATECHANGE(r300, zs);
-	r300->hw.zs.cmd[R300_ZS_CNTL_0] &= R300_RB3D_STENCIL_ENABLE;
-	r300->hw.zs.cmd[R300_ZS_CNTL_1] &=
-	    ~(R300_ZS_MASK << R300_RB3D_ZS1_DEPTH_FUNC_SHIFT);
+	r300->hw.zs.cmd[R300_ZS_CNTL_0] &= R300_STENCIL_ENABLE|R300_STENCIL_FRONT_BACK;
+	r300->hw.zs.cmd[R300_ZS_CNTL_1] &= ~(R300_ZS_MASK << R300_Z_FUNC_SHIFT);
 
-	if (ctx->Depth.Test && ctx->Depth.Func != GL_NEVER) {
+	if (ctx->Depth.Test) {
+		r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_Z_ENABLE;
 		if (ctx->Depth.Mask)
-			r300->hw.zs.cmd[R300_ZS_CNTL_0] |=
-			    R300_RB3D_Z_TEST_AND_WRITE;
-		else
-			r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_RB3D_Z_TEST;
-
-		r300->hw.zs.cmd[R300_ZS_CNTL_1] |=
-		    translate_func(ctx->Depth.
-				   Func) << R300_RB3D_ZS1_DEPTH_FUNC_SHIFT;
-	} else {
-		r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_RB3D_Z_DISABLED_1;
+			r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_Z_WRITE_ENABLE;
 		r300->hw.zs.cmd[R300_ZS_CNTL_1] |=
-		    translate_func(GL_NEVER) << R300_RB3D_ZS1_DEPTH_FUNC_SHIFT;
+		    translate_func(ctx->Depth.Func) << R300_Z_FUNC_SHIFT;
 	}
 
 	r300SetEarlyZState(ctx);
 }
 
+static void r300SetStencilState(GLcontext * ctx, GLboolean state)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+
+	if (r300->state.stencil.hw_stencil) {
+		R300_STATECHANGE(r300, zs);
+		if (state) {
+			r300->hw.zs.cmd[R300_ZS_CNTL_0] |=
+			    R300_STENCIL_ENABLE;
+		} else {
+			r300->hw.zs.cmd[R300_ZS_CNTL_0] &=
+			    ~R300_STENCIL_ENABLE;
+		}
+	} else {
+#if R200_MERGED
+		FALLBACK(&r300->radeon, RADEON_FALLBACK_STENCIL, state);
+#endif
+	}
+}
+
 static void r300UpdatePolygonMode(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	uint32_t hw_mode = 0;
+	uint32_t hw_mode = R300_GA_POLY_MODE_DISABLE;
 
+	/* Only do something if a polygon mode is wanted, default is GL_FILL */
 	if (ctx->Polygon.FrontMode != GL_FILL ||
 	    ctx->Polygon.BackMode != GL_FILL) {
 		GLenum f, b;
 
+		/* Handle GL_CW (clock wise and GL_CCW (counter clock wise)
+		 * correctly by selecting the correct front and back face
+		 */
 		if (ctx->Polygon.FrontFace == GL_CCW) {
 			f = ctx->Polygon.FrontMode;
 			b = ctx->Polygon.BackMode;
@@ -516,29 +627,30 @@ static void r300UpdatePolygonMode(GLcontext * ctx)
 			b = ctx->Polygon.FrontMode;
 		}
 
-		hw_mode |= R300_PM_ENABLED;
+		/* Enable polygon mode */
+		hw_mode |= R300_GA_POLY_MODE_DUAL;
 
 		switch (f) {
 		case GL_LINE:
-			hw_mode |= R300_PM_FRONT_LINE;
+			hw_mode |= R300_GA_POLY_MODE_FRONT_PTYPE_LINE;
 			break;
-		case GL_POINT:	/* noop */
-			hw_mode |= R300_PM_FRONT_POINT;
+		case GL_POINT:
+			hw_mode |= R300_GA_POLY_MODE_FRONT_PTYPE_POINT;
 			break;
 		case GL_FILL:
-			hw_mode |= R300_PM_FRONT_FILL;
+			hw_mode |= R300_GA_POLY_MODE_FRONT_PTYPE_TRI;
 			break;
 		}
 
 		switch (b) {
 		case GL_LINE:
-			hw_mode |= R300_PM_BACK_LINE;
+			hw_mode |= R300_GA_POLY_MODE_BACK_PTYPE_LINE;
 			break;
-		case GL_POINT:	/* noop */
-			hw_mode |= R300_PM_BACK_POINT;
+		case GL_POINT:
+			hw_mode |= R300_GA_POLY_MODE_BACK_PTYPE_POINT;
 			break;
 		case GL_FILL:
-			hw_mode |= R300_PM_BACK_FILL;
+			hw_mode |= R300_GA_POLY_MODE_BACK_PTYPE_TRI;
 			break;
 		}
 	}
@@ -606,9 +718,10 @@ static void r300ColorMask(GLcontext * ctx,
 			  GLboolean r, GLboolean g, GLboolean b, GLboolean a)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	int mask = (r ? R300_COLORMASK0_R : 0) |
-	    (g ? R300_COLORMASK0_G : 0) |
-	    (b ? R300_COLORMASK0_B : 0) | (a ? R300_COLORMASK0_A : 0);
+	int mask = (r ? RB3D_COLOR_CHANNEL_MASK_RED_MASK0 : 0) |
+	    (g ? RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0 : 0) |
+	    (b ? RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0 : 0) |
+	    (a ? RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0 : 0);
 
 	if (mask != r300->hw.cmk.cmd[R300_CMK_COLORMASK]) {
 		R300_STATECHANGE(r300, cmk);
@@ -634,15 +747,13 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 
 	switch (pname) {
 	case GL_FOG_MODE:
-		if (!ctx->Fog.Enabled)
-			return;
 		switch (ctx->Fog.Mode) {
 		case GL_LINEAR:
 			R300_STATECHANGE(r300, fogs);
 			r300->hw.fogs.cmd[R300_FOGS_STATE] =
 			    (r300->hw.fogs.
-			     cmd[R300_FOGS_STATE] & ~R300_FOG_MODE_MASK) |
-			    R300_FOG_MODE_LINEAR;
+			     cmd[R300_FOGS_STATE] & ~R300_FG_FOG_BLEND_FN_MASK) |
+			    R300_FG_FOG_BLEND_FN_LINEAR;
 
 			if (ctx->Fog.Start == ctx->Fog.End) {
 				fogScale.f = -1.0;
@@ -659,8 +770,8 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 			R300_STATECHANGE(r300, fogs);
 			r300->hw.fogs.cmd[R300_FOGS_STATE] =
 			    (r300->hw.fogs.
-			     cmd[R300_FOGS_STATE] & ~R300_FOG_MODE_MASK) |
-			    R300_FOG_MODE_EXP;
+			     cmd[R300_FOGS_STATE] & ~R300_FG_FOG_BLEND_FN_MASK) |
+			    R300_FG_FOG_BLEND_FN_EXP;
 			fogScale.f = 0.0933 * ctx->Fog.Density;
 			fogStart.f = 0.0;
 			break;
@@ -668,8 +779,8 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 			R300_STATECHANGE(r300, fogs);
 			r300->hw.fogs.cmd[R300_FOGS_STATE] =
 			    (r300->hw.fogs.
-			     cmd[R300_FOGS_STATE] & ~R300_FOG_MODE_MASK) |
-			    R300_FOG_MODE_EXP2;
+			     cmd[R300_FOGS_STATE] & ~R300_FG_FOG_BLEND_FN_MASK) |
+			    R300_FG_FOG_BLEND_FN_EXP2;
 			fogScale.f = 0.3 * ctx->Fog.Density;
 			fogStart.f = 0.0;
 		default:
@@ -727,6 +838,24 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 	}
 }
 
+static void r300SetFogState(GLcontext * ctx, GLboolean state)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+
+	R300_STATECHANGE(r300, fogs);
+	if (state) {
+		r300->hw.fogs.cmd[R300_FOGS_STATE] |= R300_FG_FOG_BLEND_ENABLE;
+
+		r300Fogfv(ctx, GL_FOG_MODE, NULL);
+		r300Fogfv(ctx, GL_FOG_DENSITY, &ctx->Fog.Density);
+		r300Fogfv(ctx, GL_FOG_START, &ctx->Fog.Start);
+		r300Fogfv(ctx, GL_FOG_END, &ctx->Fog.End);
+		r300Fogfv(ctx, GL_FOG_COLOR, ctx->Fog.Color);
+	} else {
+		r300->hw.fogs.cmd[R300_FOGS_STATE] &= ~R300_FG_FOG_BLEND_ENABLE;
+	}
+}
+
 /* =============================================================
  * Point state
  */
@@ -742,6 +871,31 @@ static void r300PointSize(GLcontext * ctx, GLfloat size)
 	    ((int)(size * 6) << R300_POINTSIZE_Y_SHIFT);
 }
 
+static void r300PointParameter(GLcontext * ctx, GLenum pname, const GLfloat * param)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+
+	switch (pname) {
+	case GL_POINT_SIZE_MIN:
+		R300_STATECHANGE(r300, ga_point_minmax);
+		r300->hw.ga_point_minmax.cmd[1] &= ~R300_GA_POINT_MINMAX_MIN_MASK;
+		r300->hw.ga_point_minmax.cmd[1] |= (GLuint)(ctx->Point.MinSize * 6.0);
+		break;
+	case GL_POINT_SIZE_MAX:
+		R300_STATECHANGE(r300, ga_point_minmax);
+		r300->hw.ga_point_minmax.cmd[1] &= ~R300_GA_POINT_MINMAX_MAX_MASK;
+		r300->hw.ga_point_minmax.cmd[1] |= (GLuint)(ctx->Point.MaxSize * 6.0)
+			<< R300_GA_POINT_MINMAX_MAX_SHIFT;
+		break;
+	case GL_POINT_DISTANCE_ATTENUATION:
+		break;
+	case GL_POINT_FADE_THRESHOLD_SIZE:
+		break;
+	default:
+		break;
+	}
+}
+
 /* =============================================================
  * Line state
  */
@@ -821,37 +975,36 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	GLuint refmask =
 	    (((ctx->Stencil.
-	       Ref[0] & 0xff) << R300_RB3D_ZS2_STENCIL_REF_SHIFT) | ((ctx->
-								      Stencil.
-								      ValueMask
-								      [0] &
-								      0xff)
-								     <<
-								     R300_RB3D_ZS2_STENCIL_MASK_SHIFT));
+	       Ref[0] & 0xff) << R300_STENCILREF_SHIFT) | ((ctx->
+							    Stencil.
+							    ValueMask
+							    [0] &
+							    0xff)
+							   <<
+							   R300_STENCILMASK_SHIFT));
 
 	GLuint flag;
 
 	R300_STATECHANGE(rmesa, zs);
-
+	rmesa->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_STENCIL_FRONT_BACK;
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] &= ~((R300_ZS_MASK <<
-					       R300_RB3D_ZS1_FRONT_FUNC_SHIFT)
+					       R300_S_FRONT_FUNC_SHIFT)
 					      | (R300_ZS_MASK <<
-						 R300_RB3D_ZS1_BACK_FUNC_SHIFT));
+						 R300_S_BACK_FUNC_SHIFT));
 
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] &=
-	    ~((R300_RB3D_ZS2_STENCIL_MASK <<
-	       R300_RB3D_ZS2_STENCIL_REF_SHIFT) |
-	      (R300_RB3D_ZS2_STENCIL_MASK << R300_RB3D_ZS2_STENCIL_MASK_SHIFT));
+	    ~((R300_STENCILREF_MASK << R300_STENCILREF_SHIFT) |
+	      (R300_STENCILREF_MASK << R300_STENCILMASK_SHIFT));
 
 	flag = translate_func(ctx->Stencil.Function[0]);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
-	    (flag << R300_RB3D_ZS1_FRONT_FUNC_SHIFT);
+	    (flag << R300_S_FRONT_FUNC_SHIFT);
 
 	if (ctx->Stencil._TestTwoSide)
 		flag = translate_func(ctx->Stencil.Function[1]);
 
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
-	    (flag << R300_RB3D_ZS1_BACK_FUNC_SHIFT);
+	    (flag << R300_S_BACK_FUNC_SHIFT);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] |= refmask;
 }
 
@@ -861,11 +1014,12 @@ static void r300StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 
 	R300_STATECHANGE(rmesa, zs);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] &=
-	    ~(R300_RB3D_ZS2_STENCIL_MASK <<
-	      R300_RB3D_ZS2_STENCIL_WRITE_MASK_SHIFT);
+	    ~(R300_STENCILREF_MASK <<
+	      R300_STENCILWRITEMASK_SHIFT);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] |=
 	    (ctx->Stencil.
-	     WriteMask[0] & 0xff) << R300_RB3D_ZS2_STENCIL_WRITE_MASK_SHIFT;
+	     WriteMask[0] & R300_STENCILREF_MASK) <<
+	     R300_STENCILWRITEMASK_SHIFT;
 }
 
 static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
@@ -876,49 +1030,37 @@ static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
 	R300_STATECHANGE(rmesa, zs);
 	/* It is easier to mask what's left.. */
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] &=
-	    (R300_ZS_MASK << R300_RB3D_ZS1_DEPTH_FUNC_SHIFT) |
-	    (R300_ZS_MASK << R300_RB3D_ZS1_FRONT_FUNC_SHIFT) |
-	    (R300_ZS_MASK << R300_RB3D_ZS1_BACK_FUNC_SHIFT);
+	    (R300_ZS_MASK << R300_Z_FUNC_SHIFT) |
+	    (R300_ZS_MASK << R300_S_FRONT_FUNC_SHIFT) |
+	    (R300_ZS_MASK << R300_S_BACK_FUNC_SHIFT);
 
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 	    (translate_stencil_op(ctx->Stencil.FailFunc[0]) <<
-	     R300_RB3D_ZS1_FRONT_FAIL_OP_SHIFT)
+	     R300_S_FRONT_SFAIL_OP_SHIFT)
 	    | (translate_stencil_op(ctx->Stencil.ZFailFunc[0]) <<
-	       R300_RB3D_ZS1_FRONT_ZFAIL_OP_SHIFT)
+	       R300_S_FRONT_ZFAIL_OP_SHIFT)
 	    | (translate_stencil_op(ctx->Stencil.ZPassFunc[0]) <<
-	       R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT);
+	       R300_S_FRONT_ZPASS_OP_SHIFT);
 
 	if (ctx->Stencil._TestTwoSide) {
 		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 		    (translate_stencil_op(ctx->Stencil.FailFunc[1]) <<
-		     R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT)
+		     R300_S_BACK_SFAIL_OP_SHIFT)
 		    | (translate_stencil_op(ctx->Stencil.ZFailFunc[1]) <<
-		       R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT)
+		       R300_S_BACK_ZFAIL_OP_SHIFT)
 		    | (translate_stencil_op(ctx->Stencil.ZPassFunc[1]) <<
-		       R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT);
+		       R300_S_BACK_ZPASS_OP_SHIFT);
 	} else {
 		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 		    (translate_stencil_op(ctx->Stencil.FailFunc[0]) <<
-		     R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT)
+		     R300_S_BACK_SFAIL_OP_SHIFT)
 		    | (translate_stencil_op(ctx->Stencil.ZFailFunc[0]) <<
-		       R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT)
+		       R300_S_BACK_ZFAIL_OP_SHIFT)
 		    | (translate_stencil_op(ctx->Stencil.ZPassFunc[0]) <<
-		       R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT);
+		       R300_S_BACK_ZPASS_OP_SHIFT);
 	}
 }
 
-static void r300ClearStencil(GLcontext * ctx, GLint s)
-{
-	r300ContextPtr rmesa = R300_CONTEXT(ctx);
-
-	rmesa->state.stencil.clear =
-	    ((GLuint) (ctx->Stencil.Clear & 0xff) |
-	     (R300_RB3D_ZS2_STENCIL_MASK <<
-	      R300_RB3D_ZS2_STENCIL_MASK_SHIFT) | ((ctx->Stencil.
-						    WriteMask[0] & 0xff) <<
-						   R300_RB3D_ZS2_STENCIL_WRITE_MASK_SHIFT));
-}
-
 /* =============================================================
  * Window position and viewport transformation
  */
@@ -1006,12 +1148,12 @@ void r300UpdateDrawBuffer(GLcontext * ctx)
 	struct gl_framebuffer *fb = ctx->DrawBuffer;
 	driRenderbuffer *drb;
 
-	if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
+	if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
 		/* draw to front */
 		drb =
 		    (driRenderbuffer *) fb->Attachment[BUFFER_FRONT_LEFT].
 		    Renderbuffer;
-	} else if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT) {
+	} else if (fb->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT) {
 		/* draw to back */
 		drb =
 		    (driRenderbuffer *) fb->Attachment[BUFFER_BACK_LEFT].
@@ -1183,8 +1325,8 @@ static unsigned long gen_fixed_filter(unsigned long f)
 	    (R300_TX_CLAMP << R300_TX_WRAP_T_SHIFT)) {
 		needs_fixing |= 2;
 	}
-	if ((f & ((7 - 1) << R300_TX_WRAP_Q_SHIFT)) ==
-	    (R300_TX_CLAMP << R300_TX_WRAP_Q_SHIFT)) {
+	if ((f & ((7 - 1) << R300_TX_WRAP_R_SHIFT)) ==
+	    (R300_TX_CLAMP << R300_TX_WRAP_R_SHIFT)) {
 		needs_fixing |= 4;
 	}
 
@@ -1192,7 +1334,7 @@ static unsigned long gen_fixed_filter(unsigned long f)
 		return f;
 
 	mag = f & R300_TX_MAG_FILTER_MASK;
-	min = f & R300_TX_MIN_FILTER_MASK;
+	min = f & (R300_TX_MIN_FILTER_MASK|R300_TX_MIN_FILTER_MIP_MASK);
 
 	/* TODO: Check for anisto filters too */
 	if ((mag != R300_TX_MAG_FILTER_NEAREST)
@@ -1224,12 +1366,100 @@ static unsigned long gen_fixed_filter(unsigned long f)
 		f |= R300_TX_CLAMP_TO_EDGE << R300_TX_WRAP_T_SHIFT;
 	}
 	if (needs_fixing & 4) {
-		f &= ~((7 - 1) << R300_TX_WRAP_Q_SHIFT);
-		f |= R300_TX_CLAMP_TO_EDGE << R300_TX_WRAP_Q_SHIFT;
+		f &= ~((7 - 1) << R300_TX_WRAP_R_SHIFT);
+		f |= R300_TX_CLAMP_TO_EDGE << R300_TX_WRAP_R_SHIFT;
 	}
 	return f;
 }
 
+static void r300SetupFragmentShaderTextures(GLcontext *ctx, int *tmu_mappings)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	int i;
+	struct r300_fragment_program *fp = (struct r300_fragment_program *)
+	    (char *)ctx->FragmentProgram._Current;
+	struct r300_fragment_program_code *code = &fp->code;
+
+	R300_STATECHANGE(r300, fpt);
+
+	for (i = 0; i < code->tex.length; i++) {
+		int unit;
+		int opcode;
+		unsigned long val;
+
+		unit = code->tex.inst[i] >> R300_TEX_ID_SHIFT;
+		unit &= 15;
+
+		val = code->tex.inst[i];
+		val &= ~R300_TEX_ID_MASK;
+
+		opcode =
+			(val & R300_TEX_INST_MASK) >> R300_TEX_INST_SHIFT;
+		if (opcode == R300_TEX_OP_KIL) {
+			r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
+		} else {
+			if (tmu_mappings[unit] >= 0) {
+				val |=
+					tmu_mappings[unit] <<
+					R300_TEX_ID_SHIFT;
+				r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
+			} else {
+				// We get here when the corresponding texture image is incomplete
+				// (e.g. incomplete mipmaps etc.)
+				r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
+			}
+		}
+	}
+
+	r300->hw.fpt.cmd[R300_FPT_CMD_0] =
+		cmdpacket0(R300_US_TEX_INST_0, code->tex.length);
+}
+
+static void r500SetupFragmentShaderTextures(GLcontext *ctx, int *tmu_mappings)
+{
+	int i;
+	struct r500_fragment_program *fp = (struct r500_fragment_program *)
+	    (char *)ctx->FragmentProgram._Current;
+	struct r500_fragment_program_code *code = &fp->code;
+
+	/* find all the texture instructions and relocate the texture units */
+	for (i = 0; i < code->inst_end + 1; i++) {
+		if ((code->inst[i].inst0 & 0x3) == R500_INST_TYPE_TEX) {
+			uint32_t val;
+			int unit, opcode, new_unit;
+
+			val = code->inst[i].inst1;
+
+			unit = (val >> 16) & 0xf;
+
+			val &= ~(0xf << 16);
+
+			opcode = val & (0x7 << 22);
+			if (opcode == R500_TEX_INST_TEXKILL) {
+				new_unit = 0;
+			} else {
+				if (tmu_mappings[unit] >= 0) {
+					new_unit = tmu_mappings[unit];
+				} else {
+					new_unit = 0;
+				}
+			}
+			val |= R500_TEX_ID(new_unit);
+			code->inst[i].inst1 = val;
+		}
+	}
+}
+
+static GLuint translate_lod_bias(GLfloat bias)
+{
+	GLint b = (int)(bias*32);
+	if (b >= (1 << 9))
+		b = (1 << 9)-1;
+	else if (b < -(1 << 9))
+		b = -(1 << 9);
+	return (((GLuint)b) << R300_LOD_BIAS_SHIFT) & R300_LOD_BIAS_MASK;
+}
+
 static void r300SetupTextures(GLcontext * ctx)
 {
 	int i, mtu;
@@ -1293,8 +1523,14 @@ static void r300SetupTextures(GLcontext * ctx)
 			r300->hw.tex.filter.cmd[R300_TEX_VALUE_0 +
 						hw_tmu] =
 			    gen_fixed_filter(t->filter) | (hw_tmu << 28);
-			/* Currently disabled! */
-			r300->hw.tex.filter_1.cmd[R300_TEX_VALUE_0 + hw_tmu] = 0x0;	//0x20501f80;
+			/* Note: There is a LOD bias per texture unit and a LOD bias
+			 * per texture object. We add them here to get the correct behaviour.
+			 * (The per-texture object LOD bias was introduced in OpenGL 1.4
+			 * and is not present in the EXT_texture_object extension).
+			 */
+			r300->hw.tex.filter_1.cmd[R300_TEX_VALUE_0 + hw_tmu] =
+				t->filter_1 |
+				translate_lod_bias(ctx->Texture.Unit[i].LodBias + t->base.tObj->LodBias);
 			r300->hw.tex.size.cmd[R300_TEX_VALUE_0 + hw_tmu] =
 			    t->size;
 			r300->hw.tex.format.cmd[R300_TEX_VALUE_0 +
@@ -1325,7 +1561,7 @@ static void r300SetupTextures(GLcontext * ctx)
 	}
 
 	r300->hw.tex.filter.cmd[R300_TEX_CMD_0] =
-	    cmdpacket0(R300_TX_FILTER_0, last_hw_tmu + 1);
+	    cmdpacket0(R300_TX_FILTER0_0, last_hw_tmu + 1);
 	r300->hw.tex.filter_1.cmd[R300_TEX_CMD_0] =
 	    cmdpacket0(R300_TX_FILTER1_0, last_hw_tmu + 1);
 	r300->hw.tex.size.cmd[R300_TEX_CMD_0] =
@@ -1333,7 +1569,7 @@ static void r300SetupTextures(GLcontext * ctx)
 	r300->hw.tex.format.cmd[R300_TEX_CMD_0] =
 	    cmdpacket0(R300_TX_FORMAT_0, last_hw_tmu + 1);
 	r300->hw.tex.pitch.cmd[R300_TEX_CMD_0] =
-	    cmdpacket0(R300_TX_PITCH_0, last_hw_tmu + 1);
+	    cmdpacket0(R300_TX_FORMAT2_0, last_hw_tmu + 1);
 	r300->hw.tex.offset.cmd[R300_TEX_CMD_0] =
 	    cmdpacket0(R300_TX_OFFSET_0, last_hw_tmu + 1);
 	r300->hw.tex.chroma_key.cmd[R300_TEX_CMD_0] =
@@ -1344,39 +1580,18 @@ static void r300SetupTextures(GLcontext * ctx)
 	if (!fp)		/* should only happenen once, just after context is created */
 		return;
 
-	R300_STATECHANGE(r300, fpt);
-
-	for (i = 0; i < fp->tex.length; i++) {
-		int unit;
-		int opcode;
-		unsigned long val;
-
-		unit = fp->tex.inst[i] >> R300_FPITX_IMAGE_SHIFT;
-		unit &= 15;
-
-		val = fp->tex.inst[i];
-		val &= ~R300_FPITX_IMAGE_MASK;
-
-		opcode =
-		    (val & R300_FPITX_OPCODE_MASK) >> R300_FPITX_OPCODE_SHIFT;
-		if (opcode == R300_FPITX_OP_KIL) {
-			r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
-		} else {
-			if (tmu_mappings[unit] >= 0) {
-				val |=
-				    tmu_mappings[unit] <<
-				    R300_FPITX_IMAGE_SHIFT;
-				r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
-			} else {
-				// We get here when the corresponding texture image is incomplete
-				// (e.g. incomplete mipmaps etc.)
-				r300->hw.fpt.cmd[R300_FPT_INSTR_0 + i] = val;
-			}
+	if (r300->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV515) {
+		if (fp->mesa_program.UsesKill && last_hw_tmu < 0) {
+			// The KILL operation requires the first texture unit
+			// to be enabled.
+			r300->hw.txe.cmd[R300_TXE_ENABLE] |= 1;
+			r300->hw.tex.filter.cmd[R300_TEX_VALUE_0] = 0;
+			r300->hw.tex.filter.cmd[R300_TEX_CMD_0] =
+				cmdpacket0(R300_TX_FILTER0_0, 1);
 		}
-	}
-
-	r300->hw.fpt.cmd[R300_FPT_CMD_0] =
-	    cmdpacket0(R300_PFS_TEXI_0, fp->tex.length);
+		r300SetupFragmentShaderTextures(ctx, tmu_mappings);
+	} else
+		r500SetupFragmentShaderTextures(ctx, tmu_mappings);
 
 	if (RADEON_DEBUG & DEBUG_STATE)
 		fprintf(stderr, "TX_ENABLE: %08x  last_hw_tmu=%d\n",
@@ -1396,21 +1611,17 @@ static void r300SetupRSUnit(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 	/* I'm still unsure if these are needed */
-	GLuint interp_magic[8] = {
-		0x00,
-		R300_RS_INTERP_1_UNKNOWN,
-		R300_RS_INTERP_2_UNKNOWN,
-		R300_RS_INTERP_3_UNKNOWN,
-		0x00,
-		0x00,
-		0x00,
-		0x00
-	};
+	GLuint interp_col[8];
+        TNLcontext *tnl = TNL_CONTEXT(ctx);
+	struct vertex_buffer *VB = &tnl->vb;
 	union r300_outputs_written OutputsWritten;
 	GLuint InputsRead;
 	int fp_reg, high_rr;
-	int in_texcoords, col_interp_nr;
-	int i;
+	int col_interp_nr;
+	int rs_tex_count = 0, rs_col_count = 0;
+	int i, count;
+
+	memset(interp_col, 0, sizeof(interp_col));
 
 	if (hw_tcl_on)
 		OutputsWritten.vp_outputs = CURRENT_VERTEX_SHADER(ctx)->key.OutputsWritten;
@@ -1428,9 +1639,9 @@ static void r300SetupRSUnit(GLcontext * ctx)
 	R300_STATECHANGE(r300, rc);
 	R300_STATECHANGE(r300, rr);
 
-	fp_reg = in_texcoords = col_interp_nr = high_rr = 0;
+	fp_reg = col_interp_nr = high_rr = 0;
 
-	r300->hw.rr.cmd[R300_RR_ROUTE_1] = 0;
+	r300->hw.rr.cmd[R300_RR_INST_1] = 0;
 
 	if (InputsRead & FRAG_BIT_WPOS) {
 		for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
@@ -1446,15 +1657,53 @@ static void r300SetupRSUnit(GLcontext * ctx)
 		InputsRead &= ~FRAG_BIT_WPOS;
 	}
 
+	if (InputsRead & FRAG_BIT_COL0) {
+		count = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
+		interp_col[0] |= R300_RS_COL_PTR(rs_col_count);
+		if (count == 3)
+			interp_col[0] |= R300_RS_COL_FMT(R300_RS_COL_FMT_RGB1);
+		rs_col_count += count;
+	}
+	else
+		interp_col[0] = R300_RS_COL_FMT(R300_RS_COL_FMT_0001);
+
+	if (InputsRead & FRAG_BIT_COL1) {
+		count = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->size;
+		if (count == 3)
+			interp_col[1] |= R300_RS_COL_FMT(R300_RS_COL_FMT_RGB0);
+		interp_col[1] |= R300_RS_COL_PTR(1);
+		rs_col_count += count;
+	}
+
+
 	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] = 0 | R300_RS_INTERP_USED | (in_texcoords << R300_RS_INTERP_SRC_SHIFT)
-		    | interp_magic[i];
+		int swiz;
+
+		/* with TCL we always seem to route 4 components */
+		if (hw_tcl_on)
+		  count = 4;
+		else
+		  count = VB->AttribPtr[_TNL_ATTRIB_TEX(i)]->size;
+
+		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] = interp_col[i] | rs_tex_count;
+		switch(count) {
+		case 4: swiz = R300_RS_SEL_S(0) | R300_RS_SEL_T(1) | R300_RS_SEL_R(2) | R300_RS_SEL_Q(3); break;
+		case 3: swiz = R300_RS_SEL_S(0) | R300_RS_SEL_T(1) | R300_RS_SEL_R(2) | R300_RS_SEL_Q(R300_RS_SEL_K1); break;
+		default:
+		case 1:
+		case 2: swiz = R300_RS_SEL_S(0) | R300_RS_SEL_T(1) | R300_RS_SEL_R(R300_RS_SEL_K0) | R300_RS_SEL_Q(R300_RS_SEL_K1); break;
+		};
+
+		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] |= swiz;
 
-		r300->hw.rr.cmd[R300_RR_ROUTE_0 + fp_reg] = 0;
+		r300->hw.rr.cmd[R300_RR_INST_0 + fp_reg] = 0;
 		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
+
+			rs_tex_count += count;
+
 			//assert(r300->state.texture.tc_count != 0);
-			r300->hw.rr.cmd[R300_RR_ROUTE_0 + fp_reg] |= R300_RS_ROUTE_ENABLE | i	/* source INTERP */
-			    | (fp_reg << R300_RS_ROUTE_DEST_SHIFT);
+			r300->hw.rr.cmd[R300_RR_INST_0 + fp_reg] |= R300_RS_INST_TEX_CN_WRITE | i	/* source INTERP */
+			    | (fp_reg << R300_RS_INST_TEX_ADDR_SHIFT);
 			high_rr = fp_reg;
 
 			/* Passing invalid data here can lock the GPU. */
@@ -1465,15 +1714,171 @@ static void r300SetupRSUnit(GLcontext * ctx)
 				WARN_ONCE("fragprog wants coords for tex%d, vp doesn't provide them!\n", i);
 			}
 		}
-		/* Need to count all coords enabled at vof */
-		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_TEX0 + i, _TNL_ATTRIB_TEX(i))) {
-			in_texcoords++;
+	}
+
+	if (InputsRead & FRAG_BIT_COL0) {
+		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL0, _TNL_ATTRIB_COLOR0)) {
+			r300->hw.rr.cmd[R300_RR_INST_0] |= R300_RS_INST_COL_ID(0) | R300_RS_INST_COL_CN_WRITE | (fp_reg++ << R300_RS_INST_COL_ADDR_SHIFT);
+			InputsRead &= ~FRAG_BIT_COL0;
+			col_interp_nr++;
+		} else {
+			WARN_ONCE("fragprog wants col0, vp doesn't provide it\n");
+		}
+	}
+
+	if (InputsRead & FRAG_BIT_COL1) {
+		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL1, _TNL_ATTRIB_COLOR1)) {
+			r300->hw.rr.cmd[R300_RR_INST_1] |= R300_RS_INST_COL_ID(1) | R300_RS_INST_COL_CN_WRITE | (fp_reg++ << R300_RS_INST_COL_ADDR_SHIFT);
+			InputsRead &= ~FRAG_BIT_COL1;
+			if (high_rr < 1)
+				high_rr = 1;
+			col_interp_nr++;
+		} else {
+			WARN_ONCE("fragprog wants col1, vp doesn't provide it\n");
+		}
+	}
+
+	/* Need at least one. This might still lock as the values are undefined... */
+	if (rs_tex_count == 0 && col_interp_nr == 0) {
+		r300->hw.rr.cmd[R300_RR_INST_0] |= R300_RS_INST_COL_ID(0) | R300_RS_INST_COL_CN_WRITE | (fp_reg++ << R300_RS_INST_COL_ADDR_SHIFT);
+		col_interp_nr++;
+	}
+
+	r300->hw.rc.cmd[1] = 0 | (rs_tex_count << R300_IT_COUNT_SHIFT)
+	  | (col_interp_nr << R300_IC_COUNT_SHIFT)
+	  | R300_HIRES_EN;
+
+	assert(high_rr >= 0);
+	r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R300_RS_INST_0, high_rr + 1);
+	r300->hw.rc.cmd[2] = high_rr;
+
+	if (InputsRead)
+		WARN_ONCE("Don't know how to satisfy InputsRead=0x%08x\n", InputsRead);
+}
+
+static void r500SetupRSUnit(GLcontext * ctx)
+{
+	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	/* I'm still unsure if these are needed */
+	GLuint interp_col[8];
+	union r300_outputs_written OutputsWritten;
+        TNLcontext *tnl = TNL_CONTEXT(ctx);
+	struct vertex_buffer *VB = &tnl->vb;
+	GLuint InputsRead;
+	int fp_reg, high_rr;
+	int rs_col_count = 0;
+	int in_texcoords, col_interp_nr;
+	int i, count;
+
+	memset(interp_col, 0, sizeof(interp_col));
+	if (hw_tcl_on)
+		OutputsWritten.vp_outputs = CURRENT_VERTEX_SHADER(ctx)->key.OutputsWritten;
+	else
+		RENDERINPUTS_COPY(OutputsWritten.index_bitset, r300->state.render_inputs_bitset);
+
+	if (ctx->FragmentProgram._Current)
+		InputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
+	else {
+		fprintf(stderr, "No ctx->FragmentProgram._Current!!\n");
+		return;		/* This should only ever happen once.. */
+	}
+
+	R300_STATECHANGE(r300, ri);
+	R300_STATECHANGE(r300, rc);
+	R300_STATECHANGE(r300, rr);
+
+	fp_reg = col_interp_nr = high_rr = in_texcoords = 0;
+
+	r300->hw.rr.cmd[R300_RR_INST_1] = 0;
+
+	if (InputsRead & FRAG_BIT_WPOS) {
+		for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
+			if (!(InputsRead & (FRAG_BIT_TEX0 << i)))
+				break;
+
+		if (i == ctx->Const.MaxTextureUnits) {
+			fprintf(stderr, "\tno free texcoord found...\n");
+			_mesa_exit(-1);
+		}
+
+		InputsRead |= (FRAG_BIT_TEX0 << i);
+		InputsRead &= ~FRAG_BIT_WPOS;
+	}
+
+	if (InputsRead & FRAG_BIT_COL0) {
+		count = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
+		interp_col[0] |= R500_RS_COL_PTR(rs_col_count);
+		if (count == 3)
+			interp_col[0] |= R500_RS_COL_FMT(R300_RS_COL_FMT_RGB1);
+		rs_col_count += count;
+	}
+	else
+		interp_col[0] = R500_RS_COL_FMT(R300_RS_COL_FMT_0001);
+
+	if (InputsRead & FRAG_BIT_COL1) {
+		count = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->size;
+		interp_col[1] |= R500_RS_COL_PTR(1);
+		if (count == 3)
+			interp_col[1] |= R500_RS_COL_FMT(R300_RS_COL_FMT_RGB0);
+		rs_col_count += count;
+	}
+
+	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+		GLuint swiz = 0;
+
+		/* with TCL we always seem to route 4 components */
+		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
+
+		  if (hw_tcl_on)
+		    count = 4;
+		  else
+		    count = VB->AttribPtr[_TNL_ATTRIB_TEX(i)]->size;
+
+		  /* always have on texcoord */
+		  swiz |= in_texcoords++ << R500_RS_IP_TEX_PTR_S_SHIFT;
+		  if (count >= 2)
+		    swiz |= in_texcoords++ << R500_RS_IP_TEX_PTR_T_SHIFT;
+		  else
+		    swiz |= R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_T_SHIFT;
+
+		  if (count >= 3)
+		    swiz |= in_texcoords++ << R500_RS_IP_TEX_PTR_R_SHIFT;
+		  else
+		    swiz |= R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_R_SHIFT;
+
+		  if (count == 4)
+		    swiz |= in_texcoords++ << R500_RS_IP_TEX_PTR_Q_SHIFT;
+		  else
+		    swiz |= R500_RS_IP_PTR_K1 << R500_RS_IP_TEX_PTR_Q_SHIFT;
+
+		} else
+		   swiz = (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_S_SHIFT) |
+		          (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_T_SHIFT) |
+		          (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_R_SHIFT) |
+		          (R500_RS_IP_PTR_K1 << R500_RS_IP_TEX_PTR_Q_SHIFT);
+
+		r300->hw.ri.cmd[R300_RI_INTERP_0 + i] = interp_col[i] | swiz;
+
+		r300->hw.rr.cmd[R300_RR_INST_0 + fp_reg] = 0;
+		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
+			//assert(r300->state.texture.tc_count != 0);
+			r300->hw.rr.cmd[R300_RR_INST_0 + fp_reg] |= R500_RS_INST_TEX_CN_WRITE | i	/* source INTERP */
+			    | (fp_reg << R500_RS_INST_TEX_ADDR_SHIFT);
+			high_rr = fp_reg;
+
+			/* Passing invalid data here can lock the GPU. */
+			if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_TEX0 + i, _TNL_ATTRIB_TEX(i))) {
+				InputsRead &= ~(FRAG_BIT_TEX0 << i);
+				fp_reg++;
+			} else {
+				WARN_ONCE("fragprog wants coords for tex%d, vp doesn't provide them!\n", i);
+			}
 		}
 	}
 
 	if (InputsRead & FRAG_BIT_COL0) {
 		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL0, _TNL_ATTRIB_COLOR0)) {
-			r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0 | R300_RS_ROUTE_0_COLOR | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
+			r300->hw.rr.cmd[R300_RR_INST_0] |= R500_RS_INST_COL_CN_WRITE | (fp_reg++ << R500_RS_INST_COL_ADDR_SHIFT);
 			InputsRead &= ~FRAG_BIT_COL0;
 			col_interp_nr++;
 		} else {
@@ -1483,7 +1888,7 @@ static void r300SetupRSUnit(GLcontext * ctx)
 
 	if (InputsRead & FRAG_BIT_COL1) {
 		if (R300_OUTPUTS_WRITTEN_TEST(OutputsWritten, VERT_RESULT_COL1, _TNL_ATTRIB_COLOR1)) {
-			r300->hw.rr.cmd[R300_RR_ROUTE_1] |= R300_RS_ROUTE_1_UNKNOWN11 | R300_RS_ROUTE_1_COLOR1 | (fp_reg++ << R300_RS_ROUTE_1_COLOR1_DEST_SHIFT);
+			r300->hw.rr.cmd[R300_RR_INST_1] |= (1 << 12) | R500_RS_INST_COL_CN_WRITE |  (fp_reg++ << R500_RS_INST_COL_ADDR_SHIFT);
 			InputsRead &= ~FRAG_BIT_COL1;
 			if (high_rr < 1)
 				high_rr = 1;
@@ -1495,22 +1900,25 @@ static void r300SetupRSUnit(GLcontext * ctx)
 
 	/* Need at least one. This might still lock as the values are undefined... */
 	if (in_texcoords == 0 && col_interp_nr == 0) {
-		r300->hw.rr.cmd[R300_RR_ROUTE_0] |= 0 | R300_RS_ROUTE_0_COLOR | (fp_reg++ << R300_RS_ROUTE_0_COLOR_DEST_SHIFT);
+		r300->hw.rr.cmd[R300_RR_INST_0] |= 0 | R500_RS_INST_COL_CN_WRITE | (fp_reg++ << R500_RS_INST_COL_ADDR_SHIFT);
 		col_interp_nr++;
 	}
 
-	r300->hw.rc.cmd[1] = 0 | (in_texcoords << R300_RS_CNTL_TC_CNT_SHIFT)
-	    | (col_interp_nr << R300_RS_CNTL_CI_CNT_SHIFT)
-	    | R300_RS_CNTL_0_UNKNOWN_18;
+	r300->hw.rc.cmd[1] = 0 | (in_texcoords << R300_IT_COUNT_SHIFT)
+	  | (col_interp_nr << R300_IC_COUNT_SHIFT)
+	  | R300_HIRES_EN;
 
 	assert(high_rr >= 0);
-	r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R300_RS_ROUTE_0, high_rr + 1);
+	r300->hw.rr.cmd[R300_RR_CMD_0] = cmdpacket0(R500_RS_INST_0, high_rr + 1);
 	r300->hw.rc.cmd[2] = 0xC0 | high_rr;
 
 	if (InputsRead)
 		WARN_ONCE("Don't know how to satisfy InputsRead=0x%08x\n", InputsRead);
 }
 
+
+
+
 #define bump_vpu_count(ptr, new_count)   do{\
 	drm_r300_cmd_header_t* _p=((drm_r300_cmd_header_t*)(ptr));\
 	int _nc=(new_count)/4; \
@@ -1518,7 +1926,7 @@ static void r300SetupRSUnit(GLcontext * ctx)
 	if(_nc>_p->vpu.count)_p->vpu.count=_nc;\
 	}while(0)
 
-static inline void r300SetupVertexProgramFragment(r300ContextPtr r300, int dest, struct r300_vertex_shader_fragment *vsf)
+static INLINE void r300SetupVertexProgramFragment(r300ContextPtr r300, int dest, struct r300_vertex_shader_fragment *vsf)
 {
 	int i;
 
@@ -1556,10 +1964,68 @@ static inline void r300SetupVertexProgramFragment(r300ContextPtr r300, int dest,
 	}
 }
 
+#define MIN3(a, b, c)	((a) < (b) ? MIN2(a, c) : MIN2(b, c))
+
+
+static void r300VapCntl(r300ContextPtr rmesa, GLuint input_count,
+			GLuint output_count, GLuint temp_count)
+{
+    int vtx_mem_size;
+    int pvs_num_slots;
+    int pvs_num_cntrls;
+
+    /* Flush PVS engine before changing PVS_NUM_SLOTS, PVS_NUM_CNTRLS.
+     * See r500 docs 6.5.2 - done in emit */
+
+    /* avoid division by zero */
+    if (input_count == 0) input_count = 1;
+    if (output_count == 0) output_count = 1;
+    if (temp_count == 0) temp_count = 1;
+
+    if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+	vtx_mem_size = 128;
+    else
+	vtx_mem_size = 72;
+
+    pvs_num_slots = MIN3(10, vtx_mem_size/input_count, vtx_mem_size/output_count);
+    pvs_num_cntrls = MIN2(6, vtx_mem_size/temp_count);
+
+    R300_STATECHANGE(rmesa, vap_cntl);
+    if (rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL) {
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] =
+	    (pvs_num_slots << R300_PVS_NUM_SLOTS_SHIFT) |
+	    (pvs_num_cntrls << R300_PVS_NUM_CNTLRS_SHIFT) |
+	    (12 << R300_VF_MAX_VTX_NUM_SHIFT);
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+	    rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= R500_TCL_STATE_OPTIMIZATION;
+    } else
+	/* not sure about non-tcl */
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] = ((10 << R300_PVS_NUM_SLOTS_SHIFT) |
+				    (5 << R300_PVS_NUM_CNTLRS_SHIFT) |
+				    (5 << R300_VF_MAX_VTX_NUM_SHIFT));
+
+    if (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV515)
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= (2 << R300_PVS_NUM_FPUS_SHIFT);
+    else if ((rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV530) ||
+	     (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV560) ||
+	     (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV570))
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= (5 << R300_PVS_NUM_FPUS_SHIFT);
+    else if ((rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV410) ||
+	     (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_R420))
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= (6 << R300_PVS_NUM_FPUS_SHIFT);
+    else if ((rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_R520) ||
+	     (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_R580))
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= (8 << R300_PVS_NUM_FPUS_SHIFT);
+    else
+	rmesa->hw.vap_cntl.cmd[R300_VAP_CNTL_INSTR] |= (4 << R300_PVS_NUM_FPUS_SHIFT);
+
+}
+
 static void r300SetupDefaultVertexProgram(r300ContextPtr rmesa)
 {
 	struct r300_vertex_shader_state *prog = &(rmesa->state.vertex_shader);
 	GLuint o_reg = 0;
+	GLuint i_reg = 0;
 	int i;
 	int inst_count = 0;
 	int param_count = 0;
@@ -1567,31 +2033,42 @@ static void r300SetupDefaultVertexProgram(r300ContextPtr rmesa)
 
 	for (i = VERT_ATTRIB_POS; i < VERT_ATTRIB_MAX; i++) {
 		if (rmesa->state.sw_tcl_inputs[i] != -1) {
-			prog->program.body.i[program_end + 0] = EASY_VSF_OP(MUL, o_reg++, ALL, RESULT);
-			prog->program.body.i[program_end + 1] = VSF_REG(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end + 2] = VSF_ATTR_UNITY(rmesa->state.sw_tcl_inputs[i]);
-			prog->program.body.i[program_end + 3] = VSF_UNITY(rmesa->state.sw_tcl_inputs[i]);
+			prog->program.body.i[program_end + 0] = PVS_OP_DST_OPERAND(VE_MULTIPLY, GL_FALSE, GL_FALSE, o_reg++, VSF_FLAG_ALL, PVS_DST_REG_OUT);
+			prog->program.body.i[program_end + 1] = PVS_SRC_OPERAND(rmesa->state.sw_tcl_inputs[i], PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y, PVS_SRC_SELECT_Z, PVS_SRC_SELECT_W, PVS_SRC_REG_INPUT, VSF_FLAG_NONE);
+			prog->program.body.i[program_end + 2] = PVS_SRC_OPERAND(rmesa->state.sw_tcl_inputs[i], PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_REG_INPUT, VSF_FLAG_NONE);
+			prog->program.body.i[program_end + 3] = PVS_SRC_OPERAND(rmesa->state.sw_tcl_inputs[i], PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_SELECT_FORCE_1, PVS_SRC_REG_INPUT, VSF_FLAG_NONE);
 			program_end += 4;
+			i_reg++;
 		}
 	}
 
 	prog->program.length = program_end;
 
-	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM,
+	r300SetupVertexProgramFragment(rmesa, R300_PVS_CODE_START,
 				       &(prog->program));
 	inst_count = (prog->program.length / 4) - 1;
 
+	r300VapCntl(rmesa, i_reg, o_reg, 0);
+
 	R300_STATECHANGE(rmesa, pvs);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_1] =
-	    (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-	    (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
-	    (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
+	    (0 << R300_PVS_FIRST_INST_SHIFT) |
+	    (inst_count << R300_PVS_XYZW_VALID_INST_SHIFT) |
+	    (inst_count << R300_PVS_LAST_INST_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_2] =
-	    (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
-	    (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
+	    (0 << R300_PVS_CONST_BASE_OFFSET_SHIFT) |
+	    (param_count << R300_PVS_MAX_CONST_ADDR_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_3] =
-	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
-	    (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
+	    (inst_count << R300_PVS_LAST_VTX_SRC_INST_SHIFT);
+}
+
+static int bit_count (int x)
+{
+    x = ((x & 0xaaaaaaaaU) >> 1) + (x & 0x55555555U);
+    x = ((x & 0xccccccccU) >> 2) + (x & 0x33333333U);
+    x = (x >> 16) + (x & 0xffff);
+    x = ((x & 0xf0f0) >> 4) + (x & 0x0f0f);
+    return (x >> 8) + (x & 0x00ff);
 }
 
 static void r300SetupRealVertexProgram(r300ContextPtr rmesa)
@@ -1612,20 +2089,22 @@ static void r300SetupRealVertexProgram(r300ContextPtr rmesa)
 	bump_vpu_count(rmesa->hw.vpp.cmd, param_count);
 	param_count /= 4;
 
-	r300SetupVertexProgramFragment(rmesa, R300_PVS_UPLOAD_PROGRAM, &(prog->program));
+	r300SetupVertexProgramFragment(rmesa, R300_PVS_CODE_START, &(prog->program));
 	inst_count = (prog->program.length / 4) - 1;
 
+	r300VapCntl(rmesa, bit_count(prog->key.InputsRead),
+		    bit_count(prog->key.OutputsWritten), prog->num_temporaries);
+
 	R300_STATECHANGE(rmesa, pvs);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_1] =
-	  (0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_POS_END_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
+	  (0 << R300_PVS_FIRST_INST_SHIFT) |
+	  (inst_count << R300_PVS_XYZW_VALID_INST_SHIFT) |
+	  (inst_count << R300_PVS_LAST_INST_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_2] =
-	  (0 << R300_PVS_CNTL_2_PARAM_OFFSET_SHIFT) |
-	  (param_count << R300_PVS_CNTL_2_PARAM_COUNT_SHIFT);
+	  (0 << R300_PVS_CONST_BASE_OFFSET_SHIFT) |
+	  (param_count << R300_PVS_MAX_CONST_ADDR_SHIFT);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_3] =
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT) |
-	  (inst_count << R300_PVS_CNTL_3_PROGRAM_UNKNOWN2_SHIFT);
+	  (inst_count << R300_PVS_LAST_VTX_SRC_INST_SHIFT);
 }
 
 static void r300SetupVertexProgram(r300ContextPtr rmesa)
@@ -1648,13 +2127,6 @@ static void r300SetupVertexProgram(r300ContextPtr rmesa)
 		r300SetupDefaultVertexProgram(rmesa);
 	}
 
-
-	/* FIXME: This is done for vertex shader fragments, but also needs to be
-	 * done for vap_pvs, so I leave it as a reminder. */
-#if 0
-	reg_start(R300_VAP_PVS_WAITIDLE, 0);
-	e32(0x00000000);
-#endif
 }
 
 /**
@@ -1664,84 +2136,54 @@ static void r300SetupVertexProgram(r300ContextPtr rmesa)
  */
 static void r300Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 {
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-
 	if (RADEON_DEBUG & DEBUG_STATE)
 		fprintf(stderr, "%s( %s = %s )\n", __FUNCTION__,
 			_mesa_lookup_enum_by_nr(cap),
 			state ? "GL_TRUE" : "GL_FALSE");
 
 	switch (cap) {
-		/* Fast track this one...
-		 */
 	case GL_TEXTURE_1D:
 	case GL_TEXTURE_2D:
 	case GL_TEXTURE_3D:
+		/* empty */
 		break;
-
 	case GL_FOG:
-		R300_STATECHANGE(r300, fogs);
-		if (state) {
-			r300->hw.fogs.cmd[R300_FOGS_STATE] |= R300_FOG_ENABLE;
-
-			r300Fogfv(ctx, GL_FOG_MODE, NULL);
-			r300Fogfv(ctx, GL_FOG_DENSITY, &ctx->Fog.Density);
-			r300Fogfv(ctx, GL_FOG_START, &ctx->Fog.Start);
-			r300Fogfv(ctx, GL_FOG_END, &ctx->Fog.End);
-			r300Fogfv(ctx, GL_FOG_COLOR, ctx->Fog.Color);
-		} else {
-			r300->hw.fogs.cmd[R300_FOGS_STATE] &= ~R300_FOG_ENABLE;
-		}
-
+		r300SetFogState(ctx, state);
 		break;
-
 	case GL_ALPHA_TEST:
 		r300SetAlphaState(ctx);
 		break;
-
-	case GL_BLEND:
 	case GL_COLOR_LOGIC_OP:
+		r300SetLogicOpState(ctx);
+		/* fall-through, because logic op overrides blending */
+	case GL_BLEND:
 		r300SetBlendState(ctx);
 		break;
-
+	case GL_CLIP_PLANE0:
+	case GL_CLIP_PLANE1:
+	case GL_CLIP_PLANE2:
+	case GL_CLIP_PLANE3:
+	case GL_CLIP_PLANE4:
+	case GL_CLIP_PLANE5:
+		r300SetClipPlaneState(ctx, cap, state);
+		break;
 	case GL_DEPTH_TEST:
 		r300SetDepthState(ctx);
 		break;
-
 	case GL_STENCIL_TEST:
-		if (r300->state.stencil.hw_stencil) {
-			R300_STATECHANGE(r300, zs);
-			if (state) {
-				r300->hw.zs.cmd[R300_ZS_CNTL_0] |=
-				    R300_RB3D_STENCIL_ENABLE;
-			} else {
-				r300->hw.zs.cmd[R300_ZS_CNTL_0] &=
-				    ~R300_RB3D_STENCIL_ENABLE;
-			}
-		} else {
-#if R200_MERGED
-			FALLBACK(&r300->radeon, RADEON_FALLBACK_STENCIL, state);
-#endif
-		}
+		r300SetStencilState(ctx, state);
 		break;
-
 	case GL_CULL_FACE:
 		r300UpdateCulling(ctx);
 		break;
-
 	case GL_POLYGON_OFFSET_POINT:
 	case GL_POLYGON_OFFSET_LINE:
 	case GL_POLYGON_OFFSET_FILL:
-		R300_STATECHANGE(r300, occlusion_cntl);
-		if (state) {
-			r300->hw.occlusion_cntl.cmd[1] |= (3 << 0);
-		} else {
-			r300->hw.occlusion_cntl.cmd[1] &= ~(3 << 0);
-		}
+		r300SetPolygonOffsetState(ctx, state);
 		break;
 	default:
 		radeonEnable(ctx, cap, state);
-		return;
+		break;
 	}
 }
 
@@ -1784,15 +2226,11 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300UpdateTextureState(ctx);
 
 	r300SetBlendState(ctx);
+	r300SetLogicOpState(ctx);
 
 	r300AlphaFunc(ctx, ctx->Color.AlphaFunc, ctx->Color.AlphaRef);
 	r300Enable(ctx, GL_ALPHA_TEST, ctx->Color.AlphaEnabled);
 
-	if (!has_tcl)
-		r300->hw.vap_cntl.cmd[1] = 0x0014045a;
-	else
-		r300->hw.vap_cntl.cmd[1] = 0x0030045A;	//0x0030065a /* Dangerous */
-
 	r300->hw.vte.cmd[1] = R300_VPORT_X_SCALE_ENA
 	    | R300_VPORT_X_OFFSET_ENA
 	    | R300_VPORT_Y_SCALE_ENA
@@ -1801,8 +2239,8 @@ static void r300ResetHwState(r300ContextPtr r300)
 	    | R300_VPORT_Z_OFFSET_ENA | R300_VTX_W0_FMT;
 	r300->hw.vte.cmd[2] = 0x00000008;
 
-	r300->hw.unk2134.cmd[1] = 0x00FFFFFF;
-	r300->hw.unk2134.cmd[2] = 0x00000000;
+	r300->hw.vap_vf_max_vtx_indx.cmd[1] = 0x00FFFFFF;
+	r300->hw.vap_vf_max_vtx_indx.cmd[2] = 0x00000000;
 
 #ifdef MESA_LITTLE_ENDIAN
 	r300->hw.vap_cntl_status.cmd[1] = R300_VC_NO_SWAP;
@@ -1814,54 +2252,53 @@ static void r300ResetHwState(r300ContextPtr r300)
 	if (!has_tcl)
 		r300->hw.vap_cntl_status.cmd[1] |= R300_VAP_TCL_BYPASS;
 
-	r300->hw.unk21DC.cmd[1] = 0xAAAAAAAA;
-
-	r300->hw.unk221C.cmd[1] = R300_221C_NORMAL;
-
-	r300->hw.vap_clip.cmd[1] = r300PackFloat32(1.0); /* X */
-	r300->hw.vap_clip.cmd[2] = r300PackFloat32(1.0); /* X */
-	r300->hw.vap_clip.cmd[3] = r300PackFloat32(1.0); /* Y */
-	r300->hw.vap_clip.cmd[4] = r300PackFloat32(1.0); /* Y */
+	r300->hw.vap_psc_sgn_norm_cntl.cmd[1] = 0xAAAAAAAA;
 
 	/* XXX: Other families? */
 	if (has_tcl) {
+		r300->hw.vap_clip_cntl.cmd[1] = R300_PS_UCP_MODE_DIST_COP;
+
+		r300->hw.vap_clip.cmd[1] = r300PackFloat32(1.0); /* X */
+		r300->hw.vap_clip.cmd[2] = r300PackFloat32(1.0); /* X */
+		r300->hw.vap_clip.cmd[3] = r300PackFloat32(1.0); /* Y */
+		r300->hw.vap_clip.cmd[4] = r300PackFloat32(1.0); /* Y */
+
 		switch (r300->radeon.radeonScreen->chip_family) {
 		case CHIP_FAMILY_R300:
-			r300->hw.unk2288.cmd[1] = R300_2288_R300;
+			r300->hw.vap_pvs_vtx_timeout_reg.cmd[1] = R300_2288_R300;
 			break;
 		default:
-			r300->hw.unk2288.cmd[1] = R300_2288_RV350;
+			r300->hw.vap_pvs_vtx_timeout_reg.cmd[1] = R300_2288_RV350;
 			break;
 		}
 	}
 
 	r300->hw.gb_enable.cmd[1] = R300_GB_POINT_STUFF_ENABLE
 	    | R300_GB_LINE_STUFF_ENABLE
-	    | R300_GB_TRIANGLE_STUFF_ENABLE /*| R300_GB_UNK31 */ ;
+	    | R300_GB_TRIANGLE_STUFF_ENABLE;
 
 	r300->hw.gb_misc.cmd[R300_GB_MISC_MSPOS_0] = 0x66666666;
 	r300->hw.gb_misc.cmd[R300_GB_MISC_MSPOS_1] = 0x06666666;
 
-	/* XXX: Other families? */
 	r300->hw.gb_misc.cmd[R300_GB_MISC_TILE_CONFIG] =
-	    R300_GB_TILE_ENABLE | R300_GB_TILE_SIZE_16;
-	switch (r300->radeon.radeonScreen->chip_family) {
-	case CHIP_FAMILY_R300:
-	case CHIP_FAMILY_R350:
+	    R300_GB_TILE_ENABLE | R300_GB_TILE_SIZE_16 /*| R300_GB_SUBPIXEL_1_16*/;
+	switch (r300->radeon.radeonScreen->num_gb_pipes) {
+	case 1:
+	default:
 		r300->hw.gb_misc.cmd[R300_GB_MISC_TILE_CONFIG] |=
-		    R300_GB_TILE_PIPE_COUNT_R300;
+		    R300_GB_TILE_PIPE_COUNT_RV300;
 		break;
-	case CHIP_FAMILY_RV410:
+	case 2:
 		r300->hw.gb_misc.cmd[R300_GB_MISC_TILE_CONFIG] |=
-		    R300_GB_TILE_PIPE_COUNT_RV410;
+		    R300_GB_TILE_PIPE_COUNT_R300;
 		break;
-	case CHIP_FAMILY_R420:
+	case 3:
 		r300->hw.gb_misc.cmd[R300_GB_MISC_TILE_CONFIG] |=
-		    R300_GB_TILE_PIPE_COUNT_R420;
+		    R300_GB_TILE_PIPE_COUNT_R420_3P;
 		break;
-	default:
+	case 4:
 		r300->hw.gb_misc.cmd[R300_GB_MISC_TILE_CONFIG] |=
-		    R300_GB_TILE_PIPE_COUNT_RV300;
+		    R300_GB_TILE_PIPE_COUNT_R420;
 		break;
 	}
 
@@ -1869,26 +2306,26 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300->hw.gb_misc.cmd[R300_GB_MISC_SELECT] = R300_GB_FOG_SELECT_1_1_W;
 
 	/* XXX: Enable anti-aliasing? */
-	r300->hw.gb_misc.cmd[R300_GB_MISC_AA_CONFIG] = R300_AA_DISABLE;
+	r300->hw.gb_misc.cmd[R300_GB_MISC_AA_CONFIG] = GB_AA_CONFIG_AA_DISABLE;
 
-	r300->hw.unk4200.cmd[1] = r300PackFloat32(0.0);
-	r300->hw.unk4200.cmd[2] = r300PackFloat32(0.0);
-	r300->hw.unk4200.cmd[3] = r300PackFloat32(1.0);
-	r300->hw.unk4200.cmd[4] = r300PackFloat32(1.0);
+	r300->hw.ga_point_s0.cmd[1] = r300PackFloat32(0.0);
+	r300->hw.ga_point_s0.cmd[2] = r300PackFloat32(0.0);
+	r300->hw.ga_point_s0.cmd[3] = r300PackFloat32(1.0);
+	r300->hw.ga_point_s0.cmd[4] = r300PackFloat32(1.0);
 
-	r300->hw.unk4214.cmd[1] = 0x00050005;
+	r300->hw.ga_triangle_stipple.cmd[1] = 0x00050005;
 
 	r300PointSize(ctx, 1.0);
 
-	r300->hw.unk4230.cmd[1] = 0x18000006;
-	r300->hw.unk4230.cmd[2] = 0x00020006;
-	r300->hw.unk4230.cmd[3] = r300PackFloat32(1.0 / 192.0);
+	r300->hw.ga_point_minmax.cmd[1] = 0x18000006;
+	r300->hw.ga_point_minmax.cmd[2] = 0x00020006;
+	r300->hw.ga_point_minmax.cmd[3] = r300PackFloat32(1.0 / 192.0);
 
 	r300LineWidth(ctx, 1.0);
 
-	r300->hw.unk4260.cmd[1] = 0;
-	r300->hw.unk4260.cmd[2] = r300PackFloat32(0.0);
-	r300->hw.unk4260.cmd[3] = r300PackFloat32(1.0);
+	r300->hw.ga_line_stipple.cmd[1] = 0;
+	r300->hw.ga_line_stipple.cmd[2] = r300PackFloat32(0.0);
+	r300->hw.ga_line_stipple.cmd[3] = r300PackFloat32(1.0);
 
 	r300ShadeModel(ctx, ctx->Light.ShadeModel);
 
@@ -1902,19 +2339,23 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300Enable(ctx, GL_POLYGON_OFFSET_LINE, ctx->Polygon.OffsetLine);
 	r300Enable(ctx, GL_POLYGON_OFFSET_FILL, ctx->Polygon.OffsetFill);
 
-	r300->hw.unk42C0.cmd[1] = 0x4B7FFFFF;
-	r300->hw.unk42C0.cmd[2] = 0x00000000;
+	r300->hw.su_depth_scale.cmd[1] = 0x4B7FFFFF;
+	r300->hw.su_depth_scale.cmd[2] = 0x00000000;
 
-	r300->hw.unk43A4.cmd[1] = 0x0000001C;
-	r300->hw.unk43A4.cmd[2] = 0x2DA49525;
+	r300->hw.sc_hyperz.cmd[1] = 0x0000001C;
+	r300->hw.sc_hyperz.cmd[2] = 0x2DA49525;
 
-	r300->hw.unk43E8.cmd[1] = 0x00FFFFFF;
+	r300->hw.sc_screendoor.cmd[1] = 0x00FFFFFF;
 
-	r300->hw.unk46A4.cmd[1] = 0x00001B01;
-	r300->hw.unk46A4.cmd[2] = 0x00001B0F;
-	r300->hw.unk46A4.cmd[3] = 0x00001B0F;
-	r300->hw.unk46A4.cmd[4] = 0x00001B0F;
-	r300->hw.unk46A4.cmd[5] = 0x00000001;
+	r300->hw.us_out_fmt.cmd[1] = R500_OUT_FMT_C4_8  |
+	  R500_C0_SEL_B | R500_C1_SEL_G | R500_C2_SEL_R | R500_C3_SEL_A;
+	r300->hw.us_out_fmt.cmd[2] = R500_OUT_FMT_UNUSED |
+	  R500_C0_SEL_B | R500_C1_SEL_G | R500_C2_SEL_R | R500_C3_SEL_A;
+	r300->hw.us_out_fmt.cmd[3] = R500_OUT_FMT_UNUSED |
+	  R500_C0_SEL_B | R500_C1_SEL_G | R500_C2_SEL_R | R500_C3_SEL_A;
+	r300->hw.us_out_fmt.cmd[4] = R500_OUT_FMT_UNUSED |
+	  R500_C0_SEL_B | R500_C1_SEL_G | R500_C2_SEL_R | R500_C3_SEL_A;
+	r300->hw.us_out_fmt.cmd[5] = R300_W_FMT_W24;
 
 	r300Enable(ctx, GL_FOG, ctx->Fog.Enabled);
 	r300Fogfv(ctx, GL_FOG_MODE, NULL);
@@ -1924,9 +2365,9 @@ static void r300ResetHwState(r300ContextPtr r300)
 	r300Fogfv(ctx, GL_FOG_COLOR, ctx->Fog.Color);
 	r300Fogfv(ctx, GL_FOG_COORDINATE_SOURCE_EXT, NULL);
 
-	r300->hw.unk4BD8.cmd[1] = 0;
+	r300->hw.fg_depth_src.cmd[1] = 0;
 
-	r300->hw.unk4E00.cmd[1] = 0;
+	r300->hw.rb3d_cctl.cmd[1] = 0;
 
 	r300BlendColor(ctx, ctx->Color.BlendColor);
 
@@ -1944,20 +2385,20 @@ static void r300ResetHwState(r300ContextPtr r300)
 	if (r300->radeon.sarea->tiling_enabled)
 		r300->hw.cb.cmd[R300_CB_PITCH] |= R300_COLOR_TILE_ENABLE;
 
-	r300->hw.unk4E50.cmd[1] = 0;
-	r300->hw.unk4E50.cmd[2] = 0;
-	r300->hw.unk4E50.cmd[3] = 0;
-	r300->hw.unk4E50.cmd[4] = 0;
-	r300->hw.unk4E50.cmd[5] = 0;
-	r300->hw.unk4E50.cmd[6] = 0;
-	r300->hw.unk4E50.cmd[7] = 0;
-	r300->hw.unk4E50.cmd[8] = 0;
-	r300->hw.unk4E50.cmd[9] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[1] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[2] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[3] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[4] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[5] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[6] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[7] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[8] = 0;
+	r300->hw.rb3d_dither_ctl.cmd[9] = 0;
 
-	r300->hw.unk4E88.cmd[1] = 0;
+	r300->hw.rb3d_aaresolve_ctl.cmd[1] = 0;
 
-	r300->hw.unk4EA0.cmd[1] = 0x00000000;
-	r300->hw.unk4EA0.cmd[2] = 0xffffffff;
+	r300->hw.rb3d_discard_src_pixel_lte_threshold.cmd[1] = 0x00000000;
+	r300->hw.rb3d_discard_src_pixel_lte_threshold.cmd[2] = 0xffffffff;
 
 	r300->hw.zb.cmd[R300_ZB_OFFSET] =
 	    r300->radeon.radeonScreen->depthOffset +
@@ -1966,22 +2407,40 @@ static void r300ResetHwState(r300ContextPtr r300)
 
 	if (r300->radeon.sarea->tiling_enabled) {
 		/* XXX: Turn off when clearing buffers ? */
-		r300->hw.zb.cmd[R300_ZB_PITCH] |= R300_DEPTH_TILE_ENABLE;
+		r300->hw.zb.cmd[R300_ZB_PITCH] |= R300_DEPTHMACROTILE_ENABLE;
 
 		if (ctx->Visual.depthBits == 24)
 			r300->hw.zb.cmd[R300_ZB_PITCH] |=
-			    R300_DEPTH_MICROTILE_ENABLE;
+			    R300_DEPTHMICROTILE_TILED;
 	}
 
-	r300->hw.unk4F28.cmd[1] = 0;
+	r300->hw.zb_depthclearvalue.cmd[1] = 0;
+
+	switch (ctx->Visual.depthBits) {
+	case 16:
+		r300->hw.zstencil_format.cmd[1] = R300_DEPTHFORMAT_16BIT_INT_Z;
+		break;
+	case 24:
+		r300->hw.zstencil_format.cmd[1] = R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
+		break;
+	default:
+		fprintf(stderr, "Error: Unsupported depth %d... exiting\n", ctx->Visual.depthBits);
+		_mesa_exit(-1);
+	}
+
+	r300->hw.zstencil_format.cmd[2] = R300_ZTOP_DISABLE;
+	r300->hw.zstencil_format.cmd[3] = 0x00000003;
+	r300->hw.zstencil_format.cmd[4] = 0x00000000;
+	r300SetEarlyZState(ctx);
 
 	r300->hw.unk4F30.cmd[1] = 0;
 	r300->hw.unk4F30.cmd[2] = 0;
 
-	r300->hw.unk4F44.cmd[1] = 0;
+	r300->hw.zb_hiz_offset.cmd[1] = 0;
 
-	r300->hw.unk4F54.cmd[1] = 0;
+	r300->hw.zb_hiz_pitch.cmd[1] = 0;
 
+	r300VapCntl(r300, 0, 0, 0);
 	if (has_tcl) {
 		r300->hw.vps.cmd[R300_VPS_ZERO_0] = 0;
 		r300->hw.vps.cmd[R300_VPS_ZERO_1] = 0;
@@ -2027,17 +2486,40 @@ void r300UpdateShaders(r300ContextPtr rmesa)
 			hw_tcl_on = future_hw_tcl_on = 0;
 			r300ResetHwState(rmesa);
 
+			r300UpdateStateParameters(ctx, _NEW_PROGRAM);
 			return;
 		}
-		r300UpdateStateParameters(ctx, _NEW_PROGRAM);
 	}
+	r300UpdateStateParameters(ctx, _NEW_PROGRAM);
 }
 
+static const GLfloat *get_fragmentprogram_constant(GLcontext *ctx,
+	struct gl_program *program, struct prog_src_register srcreg)
+{
+	static const GLfloat dummy[4] = { 0, 0, 0, 0 };
+
+	switch(srcreg.File) {
+	case PROGRAM_LOCAL_PARAM:
+		return program->LocalParams[srcreg.Index];
+	case PROGRAM_ENV_PARAM:
+		return ctx->FragmentProgram.Parameters[srcreg.Index];
+	case PROGRAM_STATE_VAR:
+	case PROGRAM_NAMED_PARAM:
+	case PROGRAM_CONSTANT:
+		return program->Parameters->ParameterValues[srcreg.Index];
+	default:
+		_mesa_problem(ctx, "get_fragmentprogram_constant: Unknown\n");
+		return dummy;
+	}
+}
+
+
 static void r300SetupPixelShader(r300ContextPtr rmesa)
 {
 	GLcontext *ctx = rmesa->radeon.glCtx;
 	struct r300_fragment_program *fp = (struct r300_fragment_program *)
 	    (char *)ctx->FragmentProgram._Current;
+	struct r300_fragment_program_code *code;
 	int i, k;
 
 	if (!fp)		/* should only happenen once, just after context is created */
@@ -2049,76 +2531,174 @@ static void r300SetupPixelShader(r300ContextPtr rmesa)
 			__FUNCTION__);
 		return;
 	}
+	code = &fp->code;
 
-	R300_STATECHANGE(rmesa, fpi[0]);
-	rmesa->hw.fpi[0].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR0_0, fp->alu_end + 1);
-	for (i = 0; i <= fp->alu_end; i++) {
-		rmesa->hw.fpi[0].cmd[R300_FPI_INSTR_0 + i] = fp->alu.inst[i].inst0;
-	}
+	r300SetupTextures(ctx);
 
+	R300_STATECHANGE(rmesa, fpi[0]);
 	R300_STATECHANGE(rmesa, fpi[1]);
-	rmesa->hw.fpi[1].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR1_0, fp->alu_end + 1);
-	for (i = 0; i <= fp->alu_end; i++) {
-		rmesa->hw.fpi[1].cmd[R300_FPI_INSTR_0 + i] = fp->alu.inst[i].inst1;
-	}
-
 	R300_STATECHANGE(rmesa, fpi[2]);
-	rmesa->hw.fpi[2].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR2_0, fp->alu_end + 1);
-	for (i = 0; i <= fp->alu_end; i++) {
-		rmesa->hw.fpi[2].cmd[R300_FPI_INSTR_0 + i] = fp->alu.inst[i].inst2;
-	}
-
 	R300_STATECHANGE(rmesa, fpi[3]);
-	rmesa->hw.fpi[3].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_PFS_INSTR3_0, fp->alu_end + 1);
-	for (i = 0; i <= fp->alu_end; i++) {
-		rmesa->hw.fpi[3].cmd[R300_FPI_INSTR_0 + i] = fp->alu.inst[i].inst3;
+	rmesa->hw.fpi[0].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_RGB_INST_0, code->alu.length);
+	rmesa->hw.fpi[1].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_RGB_ADDR_0, code->alu.length);
+	rmesa->hw.fpi[2].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_ALPHA_INST_0, code->alu.length);
+	rmesa->hw.fpi[3].cmd[R300_FPI_CMD_0] = cmdpacket0(R300_US_ALU_ALPHA_ADDR_0, code->alu.length);
+	for (i = 0; i < code->alu.length; i++) {
+		rmesa->hw.fpi[0].cmd[R300_FPI_INSTR_0 + i] = code->alu.inst[i].inst0;
+		rmesa->hw.fpi[1].cmd[R300_FPI_INSTR_0 + i] = code->alu.inst[i].inst1;
+		rmesa->hw.fpi[2].cmd[R300_FPI_INSTR_0 + i] = code->alu.inst[i].inst2;
+		rmesa->hw.fpi[3].cmd[R300_FPI_INSTR_0 + i] = code->alu.inst[i].inst3;
 	}
 
 	R300_STATECHANGE(rmesa, fp);
-	rmesa->hw.fp.cmd[R300_FP_CNTL0] = fp->cur_node | (fp->first_node_has_tex << 3);
-	rmesa->hw.fp.cmd[R300_FP_CNTL1] = fp->max_temp_idx;
+	rmesa->hw.fp.cmd[R300_FP_CNTL0] = code->cur_node | (code->first_node_has_tex << 3);
+	rmesa->hw.fp.cmd[R300_FP_CNTL1] = code->max_temp_idx;
 	rmesa->hw.fp.cmd[R300_FP_CNTL2] =
-	  (fp->alu_offset << R300_PFS_CNTL_ALU_OFFSET_SHIFT) |
-	  (fp->alu_end << R300_PFS_CNTL_ALU_END_SHIFT) |
-	  (fp->tex_offset << R300_PFS_CNTL_TEX_OFFSET_SHIFT) |
-	  (fp->tex_end << R300_PFS_CNTL_TEX_END_SHIFT);
+	  (0 << R300_PFS_CNTL_ALU_OFFSET_SHIFT) |
+	  ((code->alu.length-1) << R300_PFS_CNTL_ALU_END_SHIFT) |
+	  (0 << R300_PFS_CNTL_TEX_OFFSET_SHIFT) |
+	  ((code->tex.length ? code->tex.length-1 : 0) << R300_PFS_CNTL_TEX_END_SHIFT);
 	/* I just want to say, the way these nodes are stored.. weird.. */
-	for (i = 0, k = (4 - (fp->cur_node + 1)); i < 4; i++, k++) {
-		if (i < (fp->cur_node + 1)) {
+	for (i = 0, k = (4 - (code->cur_node + 1)); i < 4; i++, k++) {
+		if (i < (code->cur_node + 1)) {
 			rmesa->hw.fp.cmd[R300_FP_NODE0 + k] =
-			  (fp->node[i].alu_offset << R300_PFS_NODE_ALU_OFFSET_SHIFT) |
-			  (fp->node[i].alu_end << R300_PFS_NODE_ALU_END_SHIFT) |
-			  (fp->node[i].tex_offset << R300_PFS_NODE_TEX_OFFSET_SHIFT) |
-			  (fp->node[i].tex_end << R300_PFS_NODE_TEX_END_SHIFT) |
-			  fp->node[i].flags;
+			  (code->node[i].alu_offset << R300_ALU_START_SHIFT) |
+			  (code->node[i].alu_end << R300_ALU_SIZE_SHIFT) |
+			  (code->node[i].tex_offset << R300_TEX_START_SHIFT) |
+			  (code->node[i].tex_end << R300_TEX_SIZE_SHIFT) |
+			  code->node[i].flags;
 		} else {
 			rmesa->hw.fp.cmd[R300_FP_NODE0 + (3 - i)] = 0;
 		}
 	}
 
 	R300_STATECHANGE(rmesa, fpp);
-	rmesa->hw.fpp.cmd[R300_FPP_CMD_0] = cmdpacket0(R300_PFS_PARAM_0_X, fp->const_nr * 4);
-	for (i = 0; i < fp->const_nr; i++) {
-		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 0] = r300PackFloat24(fp->constant[i][0]);
-		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 1] = r300PackFloat24(fp->constant[i][1]);
-		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 2] = r300PackFloat24(fp->constant[i][2]);
-		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 3] = r300PackFloat24(fp->constant[i][3]);
+	rmesa->hw.fpp.cmd[R300_FPP_CMD_0] = cmdpacket0(R300_PFS_PARAM_0_X, code->const_nr * 4);
+	for (i = 0; i < code->const_nr; i++) {
+		const GLfloat *constant = get_fragmentprogram_constant(ctx,
+			&fp->mesa_program.Base, code->constant[i]);
+		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 0] = r300PackFloat24(constant[0]);
+		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 1] = r300PackFloat24(constant[1]);
+		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 2] = r300PackFloat24(constant[2]);
+		rmesa->hw.fpp.cmd[R300_FPP_PARAM_0 + 4 * i + 3] = r300PackFloat24(constant[3]);
 	}
 }
 
+#define bump_r500fp_count(ptr, new_count)   do{\
+	drm_r300_cmd_header_t* _p=((drm_r300_cmd_header_t*)(ptr));\
+	int _nc=(new_count)/6; \
+	assert(_nc < 256); \
+	if(_nc>_p->r500fp.count)_p->r500fp.count=_nc;\
+} while(0)
+
+#define bump_r500fp_const_count(ptr, new_count)   do{\
+	drm_r300_cmd_header_t* _p=((drm_r300_cmd_header_t*)(ptr));\
+	int _nc=(new_count)/4; \
+	assert(_nc < 256); \
+	if(_nc>_p->r500fp.count)_p->r500fp.count=_nc;\
+} while(0)
+
+static void r500SetupPixelShader(r300ContextPtr rmesa)
+{
+	GLcontext *ctx = rmesa->radeon.glCtx;
+	struct r500_fragment_program *fp = (struct r500_fragment_program *)
+	    (char *)ctx->FragmentProgram._Current;
+	int i;
+	struct r500_fragment_program_code *code;
+
+	if (!fp)		/* should only happenen once, just after context is created */
+		return;
+
+	((drm_r300_cmd_header_t *) rmesa->hw.r500fp.cmd)->r500fp.count = 0;
+	((drm_r300_cmd_header_t *) rmesa->hw.r500fp_const.cmd)->r500fp.count = 0;
+
+	r500TranslateFragmentShader(rmesa, fp);
+	if (!fp->translated) {
+		fprintf(stderr, "%s: No valid fragment shader, exiting\n",
+			__FUNCTION__);
+		return;
+	}
+	code = &fp->code;
+
+	if (fp->mesa_program.FogOption != GL_NONE) {
+		/* Enable HW fog. Try not to squish GL context.
+		 * (Anybody sane remembered to set glFog() opts first!) */
+		r300SetFogState(ctx, GL_TRUE);
+		ctx->Fog.Mode = fp->mesa_program.FogOption;
+		r300Fogfv(ctx, GL_FOG_MODE, NULL);
+	} else
+		/* Make sure HW is matching GL context. */
+		r300SetFogState(ctx, ctx->Fog.Enabled);
+
+	r300SetupTextures(ctx);
+
+	R300_STATECHANGE(rmesa, fp);
+	rmesa->hw.fp.cmd[R500_FP_PIXSIZE] = code->max_temp_idx;
+
+	rmesa->hw.fp.cmd[R500_FP_CODE_ADDR] =
+	    R500_US_CODE_START_ADDR(code->inst_offset) |
+	    R500_US_CODE_END_ADDR(code->inst_end);
+	rmesa->hw.fp.cmd[R500_FP_CODE_RANGE] =
+	    R500_US_CODE_RANGE_ADDR(code->inst_offset) |
+	    R500_US_CODE_RANGE_SIZE(code->inst_end);
+	rmesa->hw.fp.cmd[R500_FP_CODE_OFFSET] =
+	    R500_US_CODE_OFFSET_ADDR(0); /* FIXME when we add flow control */
+
+	R300_STATECHANGE(rmesa, r500fp);
+	/* Emit our shader... */
+	for (i = 0; i < code->inst_end+1; i++) {
+		rmesa->hw.r500fp.cmd[i*6+1] = code->inst[i].inst0;
+		rmesa->hw.r500fp.cmd[i*6+2] = code->inst[i].inst1;
+		rmesa->hw.r500fp.cmd[i*6+3] = code->inst[i].inst2;
+		rmesa->hw.r500fp.cmd[i*6+4] = code->inst[i].inst3;
+		rmesa->hw.r500fp.cmd[i*6+5] = code->inst[i].inst4;
+		rmesa->hw.r500fp.cmd[i*6+6] = code->inst[i].inst5;
+	}
+
+	bump_r500fp_count(rmesa->hw.r500fp.cmd, (code->inst_end + 1) * 6);
+
+	R300_STATECHANGE(rmesa, r500fp_const);
+	for (i = 0; i < code->const_nr; i++) {
+		const GLfloat *constant = get_fragmentprogram_constant(ctx,
+			&fp->mesa_program.Base, code->constant[i]);
+		rmesa->hw.r500fp_const.cmd[R300_FPP_PARAM_0 + 4 * i + 0] = r300PackFloat32(constant[0]);
+		rmesa->hw.r500fp_const.cmd[R300_FPP_PARAM_0 + 4 * i + 1] = r300PackFloat32(constant[1]);
+		rmesa->hw.r500fp_const.cmd[R300_FPP_PARAM_0 + 4 * i + 2] = r300PackFloat32(constant[2]);
+		rmesa->hw.r500fp_const.cmd[R300_FPP_PARAM_0 + 4 * i + 3] = r300PackFloat32(constant[3]);
+	}
+	bump_r500fp_const_count(rmesa->hw.r500fp_const.cmd, code->const_nr * 4);
+
+}
+
 void r300UpdateShaderStates(r300ContextPtr rmesa)
 {
 	GLcontext *ctx;
 	ctx = rmesa->radeon.glCtx;
 
 	r300UpdateTextureState(ctx);
+	r300SetEarlyZState(ctx);
 
-	r300SetupPixelShader(rmesa);
-	r300SetupTextures(ctx);
+	GLuint fgdepthsrc = R300_FG_DEPTH_SRC_SCAN;
+	if (current_fragment_program_writes_depth(ctx))
+		fgdepthsrc = R300_FG_DEPTH_SRC_SHADER;
+	if (fgdepthsrc != rmesa->hw.fg_depth_src.cmd[1]) {
+		R300_STATECHANGE(rmesa, fg_depth_src);
+		rmesa->hw.fg_depth_src.cmd[1] = fgdepthsrc;
+	}
+
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+		r500SetupPixelShader(rmesa);
+	else
+		r300SetupPixelShader(rmesa);
+
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+		r500SetupRSUnit(ctx);
+	else
+		r300SetupRSUnit(ctx);
 
 	if ((rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
 		r300SetupVertexProgram(rmesa);
-	r300SetupRSUnit(ctx);
+
 }
 
 /**
@@ -2158,13 +2738,11 @@ void r300InitState(r300ContextPtr r300)
 	switch (ctx->Visual.depthBits) {
 	case 16:
 		r300->state.depth.scale = 1.0 / (GLfloat) 0xffff;
-		depth_fmt = R300_DEPTH_FORMAT_16BIT_INT_Z;
-		r300->state.stencil.clear = 0x00000000;
+		depth_fmt = R300_DEPTHFORMAT_16BIT_INT_Z;
 		break;
 	case 24:
 		r300->state.depth.scale = 1.0 / (GLfloat) 0xffffff;
-		depth_fmt = R300_DEPTH_FORMAT_24BIT_INT_Z;
-		r300->state.stencil.clear = 0x00ff0000;
+		depth_fmt = R300_DEPTHFORMAT_24BIT_INT_Z_8BIT_STENCIL;
 		break;
 	default:
 		fprintf(stderr, "Error: Unsupported depth %d... exiting\n",
@@ -2188,6 +2766,24 @@ static void r300RenderMode(GLcontext * ctx, GLenum mode)
 	(void)mode;
 }
 
+void r300UpdateClipPlanes( GLcontext *ctx )
+{
+	r300ContextPtr rmesa = R300_CONTEXT(ctx);
+	GLuint p;
+
+	for (p = 0; p < ctx->Const.MaxClipPlanes; p++) {
+		if (ctx->Transform.ClipPlanesEnabled & (1 << p)) {
+			GLint *ip = (GLint *)ctx->Transform._ClipUserPlane[p];
+
+			R300_STATECHANGE( rmesa, vpucp[p] );
+			rmesa->hw.vpucp[p].cmd[R300_VPUCP_X] = ip[0];
+			rmesa->hw.vpucp[p].cmd[R300_VPUCP_Y] = ip[1];
+			rmesa->hw.vpucp[p].cmd[R300_VPUCP_Z] = ip[2];
+			rmesa->hw.vpucp[p].cmd[R300_VPUCP_W] = ip[3];
+		}
+	}
+}
+
 /**
  * Initialize driver's state callback functions
  */
@@ -2208,9 +2804,12 @@ void r300InitStateFuncs(struct dd_function_table *functions)
 	functions->Fogfv = r300Fogfv;
 	functions->FrontFace = r300FrontFace;
 	functions->ShadeModel = r300ShadeModel;
+	functions->LogicOpcode = r300LogicOpcode;
+
+	/* ARB_point_parameters */
+	functions->PointParameterfv = r300PointParameter;
 
 	/* Stencil related */
-	functions->ClearStencil = r300ClearStencil;
 	functions->StencilFuncSeparate = r300StencilFuncSeparate;
 	functions->StencilMaskSeparate = r300StencilMaskSeparate;
 	functions->StencilOpSeparate = r300StencilOpSeparate;
@@ -2225,4 +2824,6 @@ void r300InitStateFuncs(struct dd_function_table *functions)
 	functions->PolygonMode = r300PolygonMode;
 
 	functions->RenderMode = r300RenderMode;
+
+	functions->ClipPlane = r300ClipPlane;
 }
diff --git a/src/mesa/drivers/dri/r300/r300_state.h b/src/mesa/drivers/dri/r300/r300_state.h
index 365f7ecd0c6..0589ab7cad9 100644
--- a/src/mesa/drivers/dri/r300/r300_state.h
+++ b/src/mesa/drivers/dri/r300/r300_state.h
@@ -65,13 +65,16 @@ do {							\
     \
 } while (0)
 
-extern void r300UpdateStateParameters(GLcontext * ctx, GLuint new_state);
-extern void r300InitState(r300ContextPtr r300);
-extern void r300InitStateFuncs(struct dd_function_table *functions);
-extern void r300UpdateViewportOffset(GLcontext * ctx);
-extern void r300UpdateDrawBuffer(GLcontext * ctx);
-
-extern void r300UpdateShaders(r300ContextPtr rmesa);
-extern void r300UpdateShaderStates(r300ContextPtr rmesa);
+// r300_state.c
+extern int future_hw_tcl_on;
+void _tnl_UpdateFixedFunctionProgram (GLcontext * ctx);
+void r300UpdateViewportOffset (GLcontext * ctx);
+void r300UpdateDrawBuffer (GLcontext * ctx);
+void r300UpdateStateParameters (GLcontext * ctx, GLuint new_state);
+void r300UpdateShaders (r300ContextPtr rmesa);
+void r300UpdateShaderStates (r300ContextPtr rmesa);
+void r300InitState (r300ContextPtr r300);
+void r300UpdateClipPlanes (GLcontext * ctx);
+void r300InitStateFuncs (struct dd_function_table *functions);
 
 #endif				/* __R300_STATE_H__ */
diff --git a/src/mesa/drivers/dri/r300/r300_swtcl.c b/src/mesa/drivers/dri/r300/r300_swtcl.c
index c949f33bf33..b6e7ce1a1a6 100644
--- a/src/mesa/drivers/dri/r300/r300_swtcl.c
+++ b/src/mesa/drivers/dri/r300/r300_swtcl.c
@@ -34,13 +34,14 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "enums.h"
-#include "image.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/imports.h"
+#include "main/light.h"
+#include "main/macros.h"
 
 #include "swrast/s_context.h"
 #include "swrast/s_fog.h"
@@ -77,31 +78,6 @@ do {									\
    rmesa->swtcl.vertex_attr_count++;					\
 } while (0)
 
-/* this differs from the VIR0 in emit.c - TODO merge them using another option */
-static GLuint r300VAPInputRoute0(uint32_t * dst, GLvector4f ** attribptr,
-				 int *inputs, GLint * tab, GLuint nr)
-{
-	GLuint i, dw;
-
-	/* type, inputs, stop bit, size */
-	for (i = 0; i + 1 < nr; i += 2) {
-		dw = (inputs[tab[i]] << 8) | 0x3;
-		dw |= ((inputs[tab[i + 1]] << 8) | 0x3) << 16;
-		if (i + 2 == nr) {
-			dw |= (R300_VAP_INPUT_ROUTE_END << 16);
-		}
-		dst[i >> 1] = dw;
-	}
-
-	if (nr & 1) {
-		dw = (inputs[tab[nr - 1]] << 8) | 0x3;
-		dw |= R300_VAP_INPUT_ROUTE_END;
-		dst[nr >> 1] = dw;
-	}
-
-	return (nr + 1) >> 1;
-}
-
 static void r300SetVertexFormat( GLcontext *ctx )
 {
 	r300ContextPtr rmesa = R300_CONTEXT( ctx );
@@ -117,19 +93,24 @@ static void r300SetVertexFormat( GLcontext *ctx )
 	GLint tab[VERT_ATTRIB_MAX];
 	int swizzle[VERT_ATTRIB_MAX][4];
 	GLuint i, nr;
+	GLuint sz, vap_fmt_1 = 0;
 
 	DECLARE_RENDERINPUTS(render_inputs_bitset);
 	RENDERINPUTS_COPY(render_inputs_bitset, tnl->render_inputs_bitset);
 	RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 	RENDERINPUTS_COPY(rmesa->state.render_inputs_bitset, render_inputs_bitset);
 
+	vte = rmesa->hw.vte.cmd[1];
+	vte &= ~(R300_VTX_XY_FMT | R300_VTX_Z_FMT | R300_VTX_W0_FMT);
 	/* Important:
 	 */
 	if ( VB->NdcPtr != NULL ) {
 		VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
+		vte |= R300_VTX_XY_FMT | R300_VTX_Z_FMT;
 	}
 	else {
 		VB->AttribPtr[VERT_ATTRIB_POS] = VB->ClipPtr;
+		vte |= R300_VTX_W0_FMT;
 	}
 
 	assert( VB->AttribPtr[VERT_ATTRIB_POS] != NULL );
@@ -139,14 +120,15 @@ static void r300SetVertexFormat( GLcontext *ctx )
 	 * build up a hardware vertex.
 	 */
 	if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POS)) {
-		vap_vte_cntl |= R300_VTX_W0_FMT;
+		sz = VB->AttribPtr[VERT_ATTRIB_POS]->size;
 		InputsRead |= 1 << VERT_ATTRIB_POS;
 		OutputsWritten |= 1 << VERT_RESULT_HPOS;
-		EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F );
-	} else
+		EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_1F + sz - 1 );
+		offset = sz;
+	} else {
+		offset = 4;
 		EMIT_PAD(4 * sizeof(float));
-
-	offset = 4;
+	}
 
 	if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POINTSIZE )) {
 		EMIT_ATTR( _TNL_ATTRIB_POINTSIZE, EMIT_1F );
@@ -155,18 +137,19 @@ static void r300SetVertexFormat( GLcontext *ctx )
 	}
 
 	if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR0)) {
+		sz = VB->AttribPtr[VERT_ATTRIB_COLOR0]->size;
 	        rmesa->swtcl.coloroffset = offset;
 		InputsRead |= 1 << VERT_ATTRIB_COLOR0;
 		OutputsWritten |= 1 << VERT_RESULT_COL0;
-		EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4F );
+		EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_1F + sz - 1 );
+		offset += sz;
 	}
 
-	offset += 4;
-
 	rmesa->swtcl.specoffset = 0;
 	if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 )) {
+		sz = VB->AttribPtr[VERT_ATTRIB_COLOR1]->size;
 		rmesa->swtcl.specoffset = offset;
-		EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_4F );
+		EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_1F + sz - 1 );
 		InputsRead |= 1 << VERT_ATTRIB_COLOR1;
 		OutputsWritten |= 1 << VERT_RESULT_COL1;
 	}
@@ -176,9 +159,11 @@ static void r300SetVertexFormat( GLcontext *ctx )
 
 		for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 			if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
+				sz = VB->TexCoordPtr[i]->size;
 				InputsRead |= 1 << (VERT_ATTRIB_TEX0 + i);
 				OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i);
-				EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_4F );
+				EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_1F + sz - 1 );
+				vap_fmt_1 |= sz << (3 * i);
 			}
 		}
 	}
@@ -237,7 +222,7 @@ static void r300SetVertexFormat( GLcontext *ctx )
    
 	R300_STATECHANGE(rmesa, vof);
 	rmesa->hw.vof.cmd[R300_VOF_CNTL_0] = r300VAPOutputCntl0(ctx, OutputsWritten);
-	rmesa->hw.vof.cmd[R300_VOF_CNTL_1] = r300VAPOutputCntl1(ctx, OutputsWritten);
+	rmesa->hw.vof.cmd[R300_VOF_CNTL_1] = vap_fmt_1;
    
 	rmesa->swtcl.vertex_size =
 		_tnl_install_attrs( ctx,
@@ -249,7 +234,7 @@ static void r300SetVertexFormat( GLcontext *ctx )
 
 	RENDERINPUTS_COPY( rmesa->tnl_index_bitset, index_bitset );
 
-	vte = rmesa->hw.vte.cmd[1];
+
 	R300_STATECHANGE(rmesa, vte);
 	rmesa->hw.vte.cmd[1] = vte;
 	rmesa->hw.vte.cmd[2] = rmesa->swtcl.vertex_size;
@@ -590,6 +575,7 @@ static void r300RenderStart(GLcontext *ctx)
 	r300ChooseRenderState(ctx);	
 	r300SetVertexFormat(ctx);
 
+	r300UpdateShaders(rmesa);
 	r300UpdateShaderStates(rmesa);
 
 	r300EmitCacheFlush(rmesa);
diff --git a/src/mesa/drivers/dri/r300/r300_swtcl.h b/src/mesa/drivers/dri/r300/r300_swtcl.h
index 2ea6ceded7b..55df53c1adf 100644
--- a/src/mesa/drivers/dri/r300/r300_swtcl.h
+++ b/src/mesa/drivers/dri/r300/r300_swtcl.h
@@ -35,7 +35,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R300_SWTCL_H__
 #define __R300_SWTCL_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "r300_context.h"
 
diff --git a/src/mesa/drivers/dri/r300/r300_tex.c b/src/mesa/drivers/dri/r300/r300_tex.c
index 1805cecd0af..8ab382c83cc 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.c
+++ b/src/mesa/drivers/dri/r300/r300_tex.c
@@ -32,18 +32,19 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "colormac.h"
-#include "context.h"
-#include "enums.h"
-#include "image.h"
-#include "simple_list.h"
-#include "texformat.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+
 #include "texmem.h"
-#include "teximage.h"
-#include "texobj.h"
 
 #include "r300_context.h"
 #include "r300_state.h"
@@ -52,129 +53,59 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "xmlpool.h"
 
+
+static unsigned int translate_wrap_mode(GLenum wrapmode)
+{
+	switch(wrapmode) {
+	case GL_REPEAT: return R300_TX_REPEAT;
+	case GL_CLAMP: return R300_TX_CLAMP;
+	case GL_CLAMP_TO_EDGE: return R300_TX_CLAMP_TO_EDGE;
+	case GL_CLAMP_TO_BORDER: return R300_TX_CLAMP_TO_BORDER;
+	case GL_MIRRORED_REPEAT: return R300_TX_REPEAT | R300_TX_MIRRORED;
+	case GL_MIRROR_CLAMP_EXT: return R300_TX_CLAMP | R300_TX_MIRRORED;
+	case GL_MIRROR_CLAMP_TO_EDGE_EXT: return R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
+	case GL_MIRROR_CLAMP_TO_BORDER_EXT: return R300_TX_CLAMP_TO_BORDER | R300_TX_MIRRORED;
+	default:
+		_mesa_problem(NULL, "bad wrap mode in %s", __FUNCTION__);
+		return 0;
+	}
+}
+
+
 /**
- * Set the texture wrap modes.
+ * Update the cached hardware registers based on the current texture wrap modes.
  *
  * \param t Texture object whose wrap modes are to be set
- * \param swrap Wrap mode for the \a s texture coordinate
- * \param twrap Wrap mode for the \a t texture coordinate
  */
-
-static void r300SetTexWrap(r300TexObjPtr t, GLenum swrap, GLenum twrap,
-			   GLenum rwrap)
+static void r300UpdateTexWrap(r300TexObjPtr t)
 {
-	unsigned long hw_swrap = 0, hw_twrap = 0, hw_qwrap = 0;
+	struct gl_texture_object *tObj = t->base.tObj;
 
 	t->filter &=
-	    ~(R300_TX_WRAP_S_MASK | R300_TX_WRAP_T_MASK | R300_TX_WRAP_Q_MASK);
+	    ~(R300_TX_WRAP_S_MASK | R300_TX_WRAP_T_MASK | R300_TX_WRAP_R_MASK);
 
-	switch (swrap) {
-	case GL_REPEAT:
-		hw_swrap |= R300_TX_REPEAT;
-		break;
-	case GL_CLAMP:
-		hw_swrap |= R300_TX_CLAMP;
-		break;
-	case GL_CLAMP_TO_EDGE:
-		hw_swrap |= R300_TX_CLAMP_TO_EDGE;
-		break;
-	case GL_CLAMP_TO_BORDER:
-		hw_swrap |= R300_TX_CLAMP_TO_BORDER;
-		break;
-	case GL_MIRRORED_REPEAT:
-		hw_swrap |= R300_TX_REPEAT | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_EXT:
-		hw_swrap |= R300_TX_CLAMP | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-		hw_swrap |= R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-		hw_swrap |= R300_TX_CLAMP_TO_BORDER | R300_TX_MIRRORED;
-		break;
-	default:
-		_mesa_problem(NULL, "bad S wrap mode in %s", __FUNCTION__);
-	}
+	t->filter |= translate_wrap_mode(tObj->WrapS) << R300_TX_WRAP_S_SHIFT;
 
-	switch (twrap) {
-	case GL_REPEAT:
-		hw_twrap |= R300_TX_REPEAT;
-		break;
-	case GL_CLAMP:
-		hw_twrap |= R300_TX_CLAMP;
-		break;
-	case GL_CLAMP_TO_EDGE:
-		hw_twrap |= R300_TX_CLAMP_TO_EDGE;
-		break;
-	case GL_CLAMP_TO_BORDER:
-		hw_twrap |= R300_TX_CLAMP_TO_BORDER;
-		break;
-	case GL_MIRRORED_REPEAT:
-		hw_twrap |= R300_TX_REPEAT | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_EXT:
-		hw_twrap |= R300_TX_CLAMP | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-		hw_twrap |= R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-		hw_twrap |= R300_TX_CLAMP_TO_BORDER | R300_TX_MIRRORED;
-		break;
-	default:
-		_mesa_problem(NULL, "bad T wrap mode in %s", __FUNCTION__);
-	}
+	if (tObj->Target != GL_TEXTURE_1D) {
+		t->filter |= translate_wrap_mode(tObj->WrapT) << R300_TX_WRAP_T_SHIFT;
 
-	switch (rwrap) {
-	case GL_REPEAT:
-		hw_qwrap |= R300_TX_REPEAT;
-		break;
-	case GL_CLAMP:
-		hw_qwrap |= R300_TX_CLAMP;
-		break;
-	case GL_CLAMP_TO_EDGE:
-		hw_qwrap |= R300_TX_CLAMP_TO_EDGE;
-		break;
-	case GL_CLAMP_TO_BORDER:
-		hw_qwrap |= R300_TX_CLAMP_TO_BORDER;
-		break;
-	case GL_MIRRORED_REPEAT:
-		hw_qwrap |= R300_TX_REPEAT | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_EXT:
-		hw_qwrap |= R300_TX_CLAMP | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-		hw_qwrap |= R300_TX_CLAMP_TO_EDGE | R300_TX_MIRRORED;
-		break;
-	case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-		hw_qwrap |= R300_TX_CLAMP_TO_BORDER | R300_TX_MIRRORED;
-		break;
-	default:
-		_mesa_problem(NULL, "bad R wrap mode in %s", __FUNCTION__);
+		if (tObj->Target == GL_TEXTURE_3D)
+			t->filter |= translate_wrap_mode(tObj->WrapR) << R300_TX_WRAP_R_SHIFT;
 	}
-
-	t->filter |= hw_swrap << R300_TX_WRAP_S_SHIFT;
-	t->filter |= hw_twrap << R300_TX_WRAP_T_SHIFT;
-	t->filter |= hw_qwrap << R300_TX_WRAP_Q_SHIFT;
 }
 
-static void r300SetTexMaxAnisotropy(r300TexObjPtr t, GLfloat max)
+static GLuint aniso_filter(GLfloat anisotropy)
 {
-
-	t->filter &= ~R300_TX_MAX_ANISO_MASK;
-
-	if (max <= 1.0) {
-		t->filter |= R300_TX_MAX_ANISO_1_TO_1;
-	} else if (max <= 2.0) {
-		t->filter |= R300_TX_MAX_ANISO_2_TO_1;
-	} else if (max <= 4.0) {
-		t->filter |= R300_TX_MAX_ANISO_4_TO_1;
-	} else if (max <= 8.0) {
-		t->filter |= R300_TX_MAX_ANISO_8_TO_1;
+	if (anisotropy >= 16.0) {
+		return R300_TX_MAX_ANISO_16_TO_1;
+	} else if (anisotropy >= 8.0) {
+		return R300_TX_MAX_ANISO_8_TO_1;
+	} else if (anisotropy >= 4.0) {
+		return R300_TX_MAX_ANISO_4_TO_1;
+	} else if (anisotropy >= 2.0) {
+		return R300_TX_MAX_ANISO_2_TO_1;
 	} else {
-		t->filter |= R300_TX_MAX_ANISO_16_TO_1;
+		return R300_TX_MAX_ANISO_1_TO_1;
 	}
 }
 
@@ -184,54 +115,47 @@ static void r300SetTexMaxAnisotropy(r300TexObjPtr t, GLfloat max)
  * \param t Texture whose filter modes are to be set
  * \param minf Texture minification mode
  * \param magf Texture magnification mode
+ * \param anisotropy Maximum anisotropy level
  */
-
-static void r300SetTexFilter(r300TexObjPtr t, GLenum minf, GLenum magf)
+static void r300SetTexFilter(r300TexObjPtr t, GLenum minf, GLenum magf, GLfloat anisotropy)
 {
-	GLuint anisotropy = (t->filter & R300_TX_MAX_ANISO_MASK);
+	t->filter &= ~(R300_TX_MIN_FILTER_MASK | R300_TX_MIN_FILTER_MIP_MASK | R300_TX_MAG_FILTER_MASK | R300_TX_MAX_ANISO_MASK);
+	t->filter_1 &= ~R300_EDGE_ANISO_EDGE_ONLY;
 
-	t->filter &= ~(R300_TX_MIN_FILTER_MASK | R300_TX_MAG_FILTER_MASK);
+	/* Note that EXT_texture_filter_anisotropic is extremely vague about
+	 * how anisotropic filtering interacts with the "normal" filter modes.
+	 * When anisotropic filtering is enabled, we override min and mag
+	 * filter settings completely. This includes driconf's settings.
+	 */
+	if (anisotropy >= 2.0 && (minf != GL_NEAREST) && (magf != GL_NEAREST)) {
+		t->filter |= R300_TX_MAG_FILTER_ANISO
+			| R300_TX_MIN_FILTER_ANISO
+			| R300_TX_MIN_FILTER_MIP_LINEAR
+			| aniso_filter(anisotropy);
+		if (RADEON_DEBUG & DEBUG_TEXTURE)
+			fprintf(stderr, "Using maximum anisotropy of %f\n", anisotropy);
+		return;
+	}
 
-	if (anisotropy == R300_TX_MAX_ANISO_1_TO_1) {
-		switch (minf) {
-		case GL_NEAREST:
-			t->filter |= R300_TX_MIN_FILTER_NEAREST;
-			break;
-		case GL_LINEAR:
-			t->filter |= R300_TX_MIN_FILTER_LINEAR;
-			break;
-		case GL_NEAREST_MIPMAP_NEAREST:
-			t->filter |= R300_TX_MIN_FILTER_NEAREST_MIP_NEAREST;
-			break;
-		case GL_NEAREST_MIPMAP_LINEAR:
-			t->filter |= R300_TX_MIN_FILTER_NEAREST_MIP_LINEAR;
-			break;
-		case GL_LINEAR_MIPMAP_NEAREST:
-			t->filter |= R300_TX_MIN_FILTER_LINEAR_MIP_NEAREST;
-			break;
-		case GL_LINEAR_MIPMAP_LINEAR:
-			t->filter |= R300_TX_MIN_FILTER_LINEAR_MIP_LINEAR;
-			break;
-		}
-	} else {
-		switch (minf) {
-		case GL_NEAREST:
-			t->filter |= R300_TX_MIN_FILTER_ANISO_NEAREST;
-			break;
-		case GL_LINEAR:
-			t->filter |= R300_TX_MIN_FILTER_ANISO_LINEAR;
-			break;
-		case GL_NEAREST_MIPMAP_NEAREST:
-		case GL_LINEAR_MIPMAP_NEAREST:
-			t->filter |=
-			    R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST;
-			break;
-		case GL_NEAREST_MIPMAP_LINEAR:
-		case GL_LINEAR_MIPMAP_LINEAR:
-			t->filter |=
-			    R300_TX_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR;
-			break;
-		}
+	switch (minf) {
+	case GL_NEAREST:
+		t->filter |= R300_TX_MIN_FILTER_NEAREST;
+		break;
+	case GL_LINEAR:
+		t->filter |= R300_TX_MIN_FILTER_LINEAR;
+		break;
+	case GL_NEAREST_MIPMAP_NEAREST:
+		t->filter |= R300_TX_MIN_FILTER_NEAREST|R300_TX_MIN_FILTER_MIP_NEAREST;
+		break;
+	case GL_NEAREST_MIPMAP_LINEAR:
+		t->filter |= R300_TX_MIN_FILTER_NEAREST|R300_TX_MIN_FILTER_MIP_LINEAR;
+		break;
+	case GL_LINEAR_MIPMAP_NEAREST:
+		t->filter |= R300_TX_MIN_FILTER_LINEAR|R300_TX_MIN_FILTER_MIP_NEAREST;
+		break;
+	case GL_LINEAR_MIPMAP_LINEAR:
+		t->filter |= R300_TX_MIN_FILTER_LINEAR|R300_TX_MIN_FILTER_MIP_LINEAR;
+		break;
 	}
 
 	/* Note we don't have 3D mipmaps so only use the mag filter setting
@@ -249,7 +173,7 @@ static void r300SetTexFilter(r300TexObjPtr t, GLenum minf, GLenum magf)
 
 static void r300SetTexBorderColor(r300TexObjPtr t, GLubyte c[4])
 {
-	t->pp_border_color = PACK_COLOR_8888(c[0], c[1], c[2], c[3]);
+	t->pp_border_color = PACK_COLOR_8888(c[3], c[0], c[1], c[2]);
 }
 
 /**
@@ -277,9 +201,8 @@ static r300TexObjPtr r300AllocTexObj(struct gl_texture_object *texObj)
 
 		make_empty_list(&t->base);
 
-		r300SetTexWrap(t, texObj->WrapS, texObj->WrapT, texObj->WrapR);
-		r300SetTexMaxAnisotropy(t, texObj->MaxAnisotropy);
-		r300SetTexFilter(t, texObj->MinFilter, texObj->MagFilter);
+		r300UpdateTexWrap(t);
+		r300SetTexFilter(t, texObj->MinFilter, texObj->MagFilter, texObj->MaxAnisotropy);
 		r300SetTexBorderColor(t, texObj->_BorderChan);
 	}
 
@@ -482,6 +405,25 @@ static const struct gl_texture_format *r300ChooseTextureFormat(GLcontext * ctx,
 	case GL_RGBA32F_ARB:
 		return &_mesa_texformat_rgba_float32;
 
+	case GL_DEPTH_COMPONENT:
+	case GL_DEPTH_COMPONENT16:
+	case GL_DEPTH_COMPONENT24:
+	case GL_DEPTH_COMPONENT32:
+#if 0
+		switch (type) {
+		case GL_UNSIGNED_BYTE:
+		case GL_UNSIGNED_SHORT:
+			return &_mesa_texformat_z16;
+		case GL_UNSIGNED_INT:
+			return &_mesa_texformat_z32;
+		case GL_UNSIGNED_INT_24_8_EXT:
+		default:
+			return &_mesa_texformat_z24_s8;
+		}
+#else
+		return &_mesa_texformat_z16;
+#endif
+
 	default:
 		_mesa_problem(ctx,
 			      "unexpected internalFormat 0x%x in r300ChooseTextureFormat",
@@ -957,60 +899,6 @@ r300TexSubImage3D(GLcontext * ctx, GLenum target, GLint level,
 	t->dirty_images[0] |= (1 << level);
 }
 
-static void r300TexEnv(GLcontext * ctx, GLenum target,
-		       GLenum pname, const GLfloat * param)
-{
-	if (RADEON_DEBUG & DEBUG_STATE) {
-		fprintf(stderr, "%s( %s )\n",
-			__FUNCTION__, _mesa_lookup_enum_by_nr(pname));
-	}
-
-	/* This is incorrect: Need to maintain this data for each of
-	 * GL_TEXTURE_{123}D, GL_TEXTURE_RECTANGLE_NV, etc, and switch
-	 * between them according to _ReallyEnabled.
-	 */
-	switch (pname) {
-	case GL_TEXTURE_LOD_BIAS_EXT:{
-#if 0				/* Needs to be relocated in order to make sure we got the right tmu */
-			GLfloat bias, min;
-			GLuint b;
-
-			/* The R300's LOD bias is a signed 2's complement value with a
-			 * range of -16.0 <= bias < 16.0.
-			 *
-			 * NOTE: Add a small bias to the bias for conform mipsel.c test.
-			 */
-			bias = *param + .01;
-			min =
-			    driQueryOptionb(&rmesa->radeon.optionCache,
-					    "no_neg_lod_bias") ? 0.0 : -16.0;
-			bias = CLAMP(bias, min, 16.0);
-
-			/* 0.0 - 16.0 == 0x0 - 0x1000 */
-			/* 0.0 - -16.0 == 0x1001 - 0x1fff */
-			b = 0x1000 / 16.0 * bias;
-			b &= R300_LOD_BIAS_MASK;
-
-			if (b !=
-			    (rmesa->hw.tex.unknown1.
-			     cmd[R300_TEX_VALUE_0 +
-				 unit] & R300_LOD_BIAS_MASK)) {
-				R300_STATECHANGE(rmesa, tex.unknown1);
-				rmesa->hw.tex.unknown1.cmd[R300_TEX_VALUE_0 +
-							   unit] &=
-				    ~R300_LOD_BIAS_MASK;
-				rmesa->hw.tex.unknown1.cmd[R300_TEX_VALUE_0 +
-							   unit] |= b;
-			}
-#endif
-			break;
-		}
-
-	default:
-		return;
-	}
-}
-
 /**
  * Changes variables and flags for a state update, which will happen at the
  * next UpdateTextureState
@@ -1031,14 +919,13 @@ static void r300TexParameter(GLcontext * ctx, GLenum target,
 	case GL_TEXTURE_MIN_FILTER:
 	case GL_TEXTURE_MAG_FILTER:
 	case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-		r300SetTexMaxAnisotropy(t, texObj->MaxAnisotropy);
-		r300SetTexFilter(t, texObj->MinFilter, texObj->MagFilter);
+		r300SetTexFilter(t, texObj->MinFilter, texObj->MagFilter, texObj->MaxAnisotropy);
 		break;
 
 	case GL_TEXTURE_WRAP_S:
 	case GL_TEXTURE_WRAP_T:
 	case GL_TEXTURE_WRAP_R:
-		r300SetTexWrap(t, texObj->WrapS, texObj->WrapT, texObj->WrapR);
+		r300UpdateTexWrap(t);
 		break;
 
 	case GL_TEXTURE_BORDER_COLOR:
@@ -1057,13 +944,24 @@ static void r300TexParameter(GLcontext * ctx, GLenum target,
 		driSwapOutTextureObject((driTextureObject *) t);
 		break;
 
+	case GL_DEPTH_TEXTURE_MODE:
+		if (!texObj->Image[0][texObj->BaseLevel])
+			return;
+		if (texObj->Image[0][texObj->BaseLevel]->TexFormat->BaseFormat
+		    == GL_DEPTH_COMPONENT) {
+			r300SetDepthTexMode(texObj);
+			break;
+		} else {
+			/* If the texture isn't a depth texture, changing this
+			 * state won't cause any changes to the hardware.
+			 * Don't force a flush of texture state.
+			 */
+			return;
+		}
+
 	default:
 		return;
 	}
-
-	/* Mark this texobj as dirty (one bit per tex unit)
-	 */
-	t->dirty_state = TEX_ALL;
 }
 
 static void r300BindTexture(GLcontext * ctx, GLenum target,
@@ -1146,7 +1044,6 @@ void r300InitTextureFuncs(struct dd_function_table *functions)
 	functions->DeleteTexture = r300DeleteTexture;
 	functions->IsTextureResident = driIsTextureResident;
 
-	functions->TexEnv = r300TexEnv;
 	functions->TexParameter = r300TexParameter;
 
 	functions->CompressedTexImage2D = r300CompressedTexImage2D;
diff --git a/src/mesa/drivers/dri/r300/r300_tex.h b/src/mesa/drivers/dri/r300/r300_tex.h
index f67a8e6ba65..b86d45bfe05 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.h
+++ b/src/mesa/drivers/dri/r300/r300_tex.h
@@ -35,6 +35,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __r300_TEX_H__
 #define __r300_TEX_H__
 
+extern void r300SetDepthTexMode(struct gl_texture_object *tObj);
+
 extern void r300SetTexOffset(__DRIcontext *pDRICtx, GLint texname,
 			     unsigned long long offset, GLint depth,
 			     GLuint pitch);
diff --git a/src/mesa/drivers/dri/r300/r300_texmem.c b/src/mesa/drivers/dri/r300/r300_texmem.c
index 38f0da8b7c0..b03eefaa7c5 100644
--- a/src/mesa/drivers/dri/r300/r300_texmem.c
+++ b/src/mesa/drivers/dri/r300/r300_texmem.c
@@ -38,12 +38,12 @@ SOFTWARE.
 
 #include <errno.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "colormac.h"
-#include "macros.h"
-#include "simple_list.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/colormac.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
 #include "radeon_reg.h"		/* gets definition for usleep */
 #include "r300_context.h"
 #include "r300_state.h"
@@ -349,7 +349,7 @@ static void r300UploadSubImage(r300ContextPtr rmesa, r300TexObjPtr t,
 	imageWidth = texImage->Width;
 	imageHeight = texImage->Height;
 
-	offset = t->bufAddr + t->base.totalSize / 6 * face;
+	offset = t->bufAddr;
 
 	if (RADEON_DEBUG & (DEBUG_TEXTURE | DEBUG_IOCTL)) {
 		GLint imageX = 0;
@@ -505,7 +505,7 @@ int r300UploadTexImages(r300ContextPtr rmesa, r300TexObjPtr t, GLuint face)
 			t->base.lastLevel);
 	}
 
-	if (!t || t->base.totalSize == 0)
+	if (t->base.totalSize == 0)
 		return 0;
 
 	if (RADEON_DEBUG & DEBUG_SYNC) {
@@ -534,10 +534,6 @@ int r300UploadTexImages(r300ContextPtr rmesa, r300TexObjPtr t, GLuint face)
 			/* hope it's safe to add that here... */
 			t->offset |= t->tile_bits;
 		}
-
-		/* Mark this texobj as dirty on all units:
-		 */
-		t->dirty_state = TEX_ALL;
 	}
 
 	/* Let the world know we've used this memory recently.
diff --git a/src/mesa/drivers/dri/r300/r300_texstate.c b/src/mesa/drivers/dri/r300/r300_texstate.c
index 1d2909fd214..e2329f04ec7 100644
--- a/src/mesa/drivers/dri/r300/r300_texstate.c
+++ b/src/mesa/drivers/dri/r300/r300_texstate.c
@@ -35,14 +35,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \todo Enable R300 texture tiling code?
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texobj.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "main/enums.h"
 
 #include "r300_context.h"
 #include "r300_state.h"
@@ -115,11 +115,186 @@ static const struct tx_table {
 	_ASSIGN(LUMINANCE_ALPHA_FLOAT16, R300_EASY_TX_FORMAT(X, X, X, Y, FL_I16A16)),
 	_ASSIGN(INTENSITY_FLOAT32, R300_EASY_TX_FORMAT(X, X, X, X, FL_I32)),
 	_ASSIGN(INTENSITY_FLOAT16, R300_EASY_TX_FORMAT(X, X, X, X, FL_I16)),
+	_ASSIGN(Z16, R300_EASY_TX_FORMAT(X, X, X, X, X16)),
+	_ASSIGN(Z24_S8, R300_EASY_TX_FORMAT(X, X, X, X, X24_Y8)),
+	_ASSIGN(Z32, R300_EASY_TX_FORMAT(X, X, X, X, X32)),
 	/* *INDENT-ON* */
 };
 
 #undef _ASSIGN
 
+void r300SetDepthTexMode(struct gl_texture_object *tObj)
+{
+	static const GLuint formats[3][3] = {
+		{
+			R300_EASY_TX_FORMAT(X, X, X, ONE, X16),
+			R300_EASY_TX_FORMAT(X, X, X, X, X16),
+			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X16),
+		},
+		{
+			R300_EASY_TX_FORMAT(X, X, X, ONE, X24_Y8),
+			R300_EASY_TX_FORMAT(X, X, X, X, X24_Y8),
+			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X24_Y8),
+		},
+		{
+			R300_EASY_TX_FORMAT(X, X, X, ONE, X32),
+			R300_EASY_TX_FORMAT(X, X, X, X, X32),
+			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X32),
+		},
+	};
+	const GLuint *format;
+	r300TexObjPtr t;
+
+	if (!tObj)
+		return;
+
+	t = (r300TexObjPtr) tObj->DriverData;
+
+
+	switch (tObj->Image[0][tObj->BaseLevel]->TexFormat->MesaFormat) {
+	case MESA_FORMAT_Z16:
+		format = formats[0];
+		break;
+	case MESA_FORMAT_Z24_S8:
+		format = formats[1];
+		break;
+	case MESA_FORMAT_Z32:
+		format = formats[2];
+		break;
+	default:
+		/* Error...which should have already been caught by higher
+		 * levels of Mesa.
+		 */
+		ASSERT(0);
+		return;
+	}
+
+	switch (tObj->DepthMode) {
+	case GL_LUMINANCE:
+		t->format = format[0];
+		break;
+	case GL_INTENSITY:
+		t->format = format[1];
+		break;
+	case GL_ALPHA:
+		t->format = format[2];
+		break;
+	default:
+		/* Error...which should have already been caught by higher
+		 * levels of Mesa.
+		 */
+		ASSERT(0);
+		return;
+	}
+}
+
+
+/**
+ * Compute sizes and fill in offset and blit information for the given
+ * image (determined by \p face and \p level).
+ *
+ * \param curOffset points to the offset at which the image is to be stored
+ * and is updated by this function according to the size of the image.
+ */
+static void compute_tex_image_offset(
+	struct gl_texture_object *tObj,
+	GLuint face,
+	GLint level,
+	GLint* curOffset)
+{
+	r300TexObjPtr t = (r300TexObjPtr) tObj->DriverData;
+	const struct gl_texture_image* texImage;
+	GLuint blitWidth = R300_BLIT_WIDTH_BYTES;
+	GLuint texelBytes;
+	GLuint size;
+
+	texImage = tObj->Image[0][level + t->base.firstLevel];
+	if (!texImage)
+		return;
+
+	texelBytes = texImage->TexFormat->TexelBytes;
+
+	/* find image size in bytes */
+	if (texImage->IsCompressed) {
+		if ((t->format & R300_TX_FORMAT_DXT1) ==
+			R300_TX_FORMAT_DXT1) {
+			// fprintf(stderr,"DXT 1 %d %08X\n", texImage->Width, t->format);
+			if ((texImage->Width + 3) < 8)	/* width one block */
+				size = texImage->CompressedSize * 4;
+			else if ((texImage->Width + 3) < 16)
+				size = texImage->CompressedSize * 2;
+			else
+				size = texImage->CompressedSize;
+		} else {
+			/* DXT3/5, 16 bytes per block */
+			WARN_ONCE
+				("DXT 3/5 suffers from multitexturing problems!\n");
+			// fprintf(stderr,"DXT 3/5 %d\n", texImage->Width);
+			if ((texImage->Width + 3) < 8)
+				size = texImage->CompressedSize * 2;
+			else
+				size = texImage->CompressedSize;
+		}
+	} else if (tObj->Target == GL_TEXTURE_RECTANGLE_NV) {
+		size =
+			((texImage->Width * texelBytes +
+			63) & ~63) * texImage->Height;
+		blitWidth = 64 / texelBytes;
+	} else if (t->tile_bits & R300_TXO_MICRO_TILE) {
+		/* tile pattern is 16 bytes x2. mipmaps stay 32 byte aligned,
+			though the actual offset may be different (if texture is less than
+			32 bytes width) to the untiled case */
+		int w = (texImage->Width * texelBytes * 2 + 31) & ~31;
+		size =
+			(w * ((texImage->Height + 1) / 2)) *
+			texImage->Depth;
+		blitWidth = MAX2(texImage->Width, 64 / texelBytes);
+	} else {
+		int w = (texImage->Width * texelBytes + 31) & ~31;
+		size = w * texImage->Height * texImage->Depth;
+		blitWidth = MAX2(texImage->Width, 64 / texelBytes);
+	}
+	assert(size > 0);
+
+	if (RADEON_DEBUG & DEBUG_TEXTURE)
+		fprintf(stderr, "w=%d h=%d d=%d tb=%d intFormat=%d\n",
+			texImage->Width, texImage->Height,
+			texImage->Depth,
+			texImage->TexFormat->TexelBytes,
+			texImage->InternalFormat);
+
+	/* All images are aligned to a 32-byte offset */
+	*curOffset = (*curOffset + 0x1f) & ~0x1f;
+
+	if (texelBytes) {
+		/* fix x and y coords up later together with offset */
+		t->image[face][level].x = *curOffset;
+		t->image[face][level].y = 0;
+		t->image[face][level].width =
+			MIN2(size / texelBytes, blitWidth);
+		t->image[face][level].height =
+			(size / texelBytes) / t->image[face][level].width;
+	} else {
+		t->image[face][level].x = *curOffset % R300_BLIT_WIDTH_BYTES;
+		t->image[face][level].y = *curOffset / R300_BLIT_WIDTH_BYTES;
+		t->image[face][level].width =
+			MIN2(size, R300_BLIT_WIDTH_BYTES);
+		t->image[face][level].height = size / t->image[face][level].width;
+	}
+
+	if (RADEON_DEBUG & DEBUG_TEXTURE)
+		fprintf(stderr,
+			"level %d, face %d: %dx%d x=%d y=%d w=%d h=%d size=%d at %d\n",
+			level, face, texImage->Width, texImage->Height,
+			t->image[face][level].x, t->image[face][level].y,
+			t->image[face][level].width, t->image[face][level].height,
+			size, *curOffset);
+
+	*curOffset += size;
+}
+
+
+
 /**
  * This function computes the number of bytes of storage needed for
  * the given texture object (all mipmap levels, all cube faces).
@@ -137,7 +312,7 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 	r300TexObjPtr t = (r300TexObjPtr) tObj->DriverData;
 	const struct gl_texture_image *baseImage =
 	    tObj->Image[0][tObj->BaseLevel];
-	GLint curOffset, blitWidth;
+	GLint curOffset;
 	GLint i, texelBytes;
 	GLint numLevels;
 	GLint log2Width, log2Height, log2Depth;
@@ -146,7 +321,12 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 	 */
 	if (!t->image_override
 	    && VALID_FORMAT(baseImage->TexFormat->MesaFormat)) {
-		t->format = tx_table[baseImage->TexFormat->MesaFormat].format;
+		if (baseImage->TexFormat->BaseFormat == GL_DEPTH_COMPONENT) {
+			r300SetDepthTexMode(tObj);
+		} else {
+			t->format = tx_table[baseImage->TexFormat->MesaFormat].format;
+		}
+
 		t->filter |= tx_table[baseImage->TexFormat->MesaFormat].filter;
 	} else if (!t->image_override) {
 		_mesa_problem(NULL, "unexpected texture format in %s",
@@ -171,8 +351,6 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 	 * The idea is that we lay out the mipmap levels within a block of
 	 * memory organized as a rectangle of width BLIT_WIDTH_BYTES.
 	 */
-	curOffset = 0;
-	blitWidth = R300_BLIT_WIDTH_BYTES;
 	t->tile_bits = 0;
 
 	/* figure out if this texture is suitable for tiling. */
@@ -202,94 +380,23 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 	}
 #endif
 
-	for (i = 0; i < numLevels; i++) {
-		const struct gl_texture_image *texImage;
-		GLuint size;
-
-		texImage = tObj->Image[0][i + t->base.firstLevel];
-		if (!texImage)
-			break;
-
-		/* find image size in bytes */
-		if (texImage->IsCompressed) {
-			if ((t->format & R300_TX_FORMAT_DXT1) ==
-			    R300_TX_FORMAT_DXT1) {
-				// fprintf(stderr,"DXT 1 %d %08X\n", texImage->Width, t->format);
-				if ((texImage->Width + 3) < 8)	/* width one block */
-					size = texImage->CompressedSize * 4;
-				else if ((texImage->Width + 3) < 16)
-					size = texImage->CompressedSize * 2;
-				else
-					size = texImage->CompressedSize;
-			} else {
-				/* DXT3/5, 16 bytes per block */
-				WARN_ONCE
-				    ("DXT 3/5 suffers from multitexturing problems!\n");
-				// fprintf(stderr,"DXT 3/5 %d\n", texImage->Width);
-				if ((texImage->Width + 3) < 8)
-					size = texImage->CompressedSize * 2;
-				else
-					size = texImage->CompressedSize;
-			}
-		} else if (tObj->Target == GL_TEXTURE_RECTANGLE_NV) {
-			size =
-			    ((texImage->Width * texelBytes +
-			      63) & ~63) * texImage->Height;
-			blitWidth = 64 / texelBytes;
-		} else if (t->tile_bits & R300_TXO_MICRO_TILE) {
-			/* tile pattern is 16 bytes x2. mipmaps stay 32 byte aligned,
-			   though the actual offset may be different (if texture is less than
-			   32 bytes width) to the untiled case */
-			int w = (texImage->Width * texelBytes * 2 + 31) & ~31;
-			size =
-			    (w * ((texImage->Height + 1) / 2)) *
-			    texImage->Depth;
-			blitWidth = MAX2(texImage->Width, 64 / texelBytes);
-		} else {
-			int w = (texImage->Width * texelBytes + 31) & ~31;
-			size = w * texImage->Height * texImage->Depth;
-			blitWidth = MAX2(texImage->Width, 64 / texelBytes);
-		}
-		assert(size > 0);
-
-		if (RADEON_DEBUG & DEBUG_TEXTURE)
-			fprintf(stderr, "w=%d h=%d d=%d tb=%d intFormat=%d\n",
-				texImage->Width, texImage->Height,
-				texImage->Depth,
-				texImage->TexFormat->TexelBytes,
-				texImage->InternalFormat);
-
-		/* Align to 32-byte offset.  It is faster to do this unconditionally
-		 * (no branch penalty).
-		 */
+	curOffset = 0;
 
-		curOffset = (curOffset + 0x1f) & ~0x1f;
+	if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
+		ASSERT(log2Width == log2Height);
+		t->format |= R300_TX_FORMAT_CUBIC_MAP;
 
-		if (texelBytes) {
-			/* fix x and y coords up later together with offset */
-			t->image[0][i].x = curOffset;
-			t->image[0][i].y = 0;
-			t->image[0][i].width =
-			    MIN2(size / texelBytes, blitWidth);
-			t->image[0][i].height =
-			    (size / texelBytes) / t->image[0][i].width;
-		} else {
-			t->image[0][i].x = curOffset % R300_BLIT_WIDTH_BYTES;
-			t->image[0][i].y = curOffset / R300_BLIT_WIDTH_BYTES;
-			t->image[0][i].width =
-			    MIN2(size, R300_BLIT_WIDTH_BYTES);
-			t->image[0][i].height = size / t->image[0][i].width;
+		for(i = 0; i < numLevels; i++) {
+			GLuint face;
+			for(face = 0; face < 6; face++)
+				compute_tex_image_offset(tObj, face, i, &curOffset);
 		}
+	} else {
+		if (tObj->Target == GL_TEXTURE_3D)
+                	t->format |= R300_TX_FORMAT_3D;
 
-		if (RADEON_DEBUG & DEBUG_TEXTURE)
-			fprintf(stderr,
-				"level %d: %dx%d x=%d y=%d w=%d h=%d size=%d at %d\n",
-				i, texImage->Width, texImage->Height,
-				t->image[0][i].x, t->image[0][i].y,
-				t->image[0][i].width, t->image[0][i].height,
-				size, curOffset);
-
-		curOffset += size;
+		for (i = 0; i < numLevels; i++)
+			compute_tex_image_offset(tObj, 0, i, &curOffset);
 	}
 
 	/* Align the total size of texture memory block.
@@ -297,43 +404,27 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 	t->base.totalSize =
 	    (curOffset + RADEON_OFFSET_MASK) & ~RADEON_OFFSET_MASK;
 
-	/* Setup remaining cube face blits, if needed */
-	if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
-		GLuint face;
-		for (face = 1; face < 6; face++) {
-			for (i = 0; i < numLevels; i++) {
-				t->image[face][i].x = t->image[0][i].x;
-				t->image[face][i].y = t->image[0][i].y;
-				t->image[face][i].width = t->image[0][i].width;
-				t->image[face][i].height =
-				    t->image[0][i].height;
-			}
-		}
-		t->base.totalSize *= 6;	/* total texmem needed */
-	}
-
-	if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
-		ASSERT(log2Width == log2Height);
-		t->format |= R300_TX_FORMAT_CUBIC_MAP;
-	}
-
 	t->size =
 	    (((tObj->Image[0][t->base.firstLevel]->Width -
 	       1) << R300_TX_WIDTHMASK_SHIFT)
 	     | ((tObj->Image[0][t->base.firstLevel]->Height - 1) <<
-		R300_TX_HEIGHTMASK_SHIFT))
+		R300_TX_HEIGHTMASK_SHIFT)
+	     | ((tObj->Image[0][t->base.firstLevel]->DepthLog2) <<
+		R300_TX_DEPTHMASK_SHIFT))
 	    | ((numLevels - 1) << R300_TX_MAX_MIP_LEVEL_SHIFT);
 
+	t->pitch = 0;
+
 	/* Only need to round to nearest 32 for textures, but the blitter
 	 * requires 64-byte aligned pitches, and we may/may not need the
 	 * blitter.   NPOT only!
 	 */
 	if (baseImage->IsCompressed) {
-		t->pitch =
+		t->pitch |=
 		    (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
 	} else if (tObj->Target == GL_TEXTURE_RECTANGLE_NV) {
-		unsigned int align = blitWidth - 1;
-		t->pitch = ((tObj->Image[0][t->base.firstLevel]->Width *
+		unsigned int align = (64 / texelBytes) - 1;
+		t->pitch |= ((tObj->Image[0][t->base.firstLevel]->Width *
 			     texelBytes) + 63) & ~(63);
 		t->size |= R300_TX_SIZE_TXPITCH_EN;
 		if (!t->image_override)
@@ -341,14 +432,17 @@ static void r300SetTexImages(r300ContextPtr rmesa,
 			    (((tObj->Image[0][t->base.firstLevel]->Width) +
 			      align) & ~align) - 1;
 	} else {
-		t->pitch =
+		t->pitch |=
 		    ((tObj->Image[0][t->base.firstLevel]->Width *
 		      texelBytes) + 63) & ~(63);
 	}
 
-	t->dirty_state = TEX_ALL;
-
-	/* FYI: r300UploadTexImages( rmesa, t ) used to be called here */
+	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+	    if (tObj->Image[0][t->base.firstLevel]->Width > 2048)
+		t->pitch_reg |= R500_TXWIDTH_BIT11;
+	    if (tObj->Image[0][t->base.firstLevel]->Height > 2048)
+		t->pitch_reg |= R500_TXHEIGHT_BIT11;
+	}
 }
 
 /* ================================================================
@@ -485,7 +579,6 @@ static GLboolean r300UpdateTexture(GLcontext * ctx, int unit)
 
 		rmesa->state.texture.unit[unit].texobj = t;
 		t->base.bound |= (1 << unit);
-		t->dirty_state |= 1 << unit;
 		driUpdateTextureLRU((driTextureObject *) t);	/* XXX: should be locked! */
 	}
 
@@ -495,12 +588,11 @@ static GLboolean r300UpdateTexture(GLcontext * ctx, int unit)
 void r300SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 		      unsigned long long offset, GLint depth, GLuint pitch)
 {
-	r300ContextPtr rmesa =
-	    (r300ContextPtr) ((__DRIcontextPrivate *) pDRICtx->private)->
-	    driverPrivate;
+	r300ContextPtr rmesa = pDRICtx->driverPrivate;
 	struct gl_texture_object *tObj =
 	    _mesa_lookup_texture(rmesa->radeon.glCtx, texname);
 	r300TexObjPtr t;
+	uint32_t pitch_val;
 
 	if (!tObj)
 		return;
@@ -513,28 +605,30 @@ void r300SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 		return;
 
 	t->offset = offset;
-	t->pitch_reg = pitch;
+	t->pitch_reg &= (1 << 13) -1;
+	pitch_val = pitch;
 
 	switch (depth) {
 	case 32:
 		t->format = R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
 		t->filter |= tx_table[2].filter;
-		t->pitch_reg /= 4;
+		pitch_val /= 4;
 		break;
 	case 24:
 	default:
 		t->format = R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
 		t->filter |= tx_table[4].filter;
-		t->pitch_reg /= 4;
+		pitch_val /= 4;
 		break;
 	case 16:
 		t->format = R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
 		t->filter |= tx_table[5].filter;
-		t->pitch_reg /= 2;
+		pitch_val /= 2;
 		break;
 	}
+	pitch_val--;
 
-	t->pitch_reg--;
+	t->pitch_reg |= pitch_val;
 }
 
 static GLboolean r300UpdateTextureUnit(GLcontext * ctx, int unit)
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 7d4e8c95112..c4e325e6a76 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -1,6 +1,7 @@
 /**************************************************************************
 
-Copyright (C) 2005 Aapo Tahkola.
+Copyright (C) 2005  Aapo Tahkola <aet@rasterburn.org>
+Copyright (C) 2008  Oliver McFadden <z3ro.geek@gmail.com>
 
 All Rights Reserved.
 
@@ -25,20 +26,12 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 **************************************************************************/
 
-/**
- * \file
- *
- * \author Aapo Tahkola <aet@rasterburn.org>
- *
- * \author Oliver McFadden <z3ro.geek@gmail.com>
- *
- * For a description of the vertex program instruction set see r300_reg.h.
- */
+/* Radeon R5xx Acceleration, Revision 1.2 */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "program.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/program.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
@@ -46,71 +39,38 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "r300_context.h"
 
-#if SWIZZLE_X != VSF_IN_COMPONENT_X || \
-    SWIZZLE_Y != VSF_IN_COMPONENT_Y || \
-    SWIZZLE_Z != VSF_IN_COMPONENT_Z || \
-    SWIZZLE_W != VSF_IN_COMPONENT_W || \
-    SWIZZLE_ZERO != VSF_IN_COMPONENT_ZERO || \
-    SWIZZLE_ONE != VSF_IN_COMPONENT_ONE || \
-    WRITEMASK_X != VSF_FLAG_X || \
-    WRITEMASK_Y != VSF_FLAG_Y || \
-    WRITEMASK_Z != VSF_FLAG_Z || \
-    WRITEMASK_W != VSF_FLAG_W
-#error Cannot change these!
-#endif
-
 /* TODO: Get rid of t_src_class call */
 #define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \
-		       ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \
-			 t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \
-			(t_src_class(a.File) == VSF_IN_CLASS_ATTR && \
-			 t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \
-
-#define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
-
-#define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
-
-#define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    SWIZZLE_ZERO, SWIZZLE_ZERO, \
-				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
-
-#define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
-
-#define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
-
-#define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    SWIZZLE_ONE, SWIZZLE_ONE, \
-				    t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+		       ((t_src_class(a.File) == PVS_SRC_REG_CONSTANT && \
+			 t_src_class(b.File) == PVS_SRC_REG_CONSTANT) || \
+			(t_src_class(a.File) == PVS_SRC_REG_INPUT && \
+			 t_src_class(b.File) == PVS_SRC_REG_INPUT)))) \
 
-/* DP4 version seems to trigger some hw peculiarity */
-//#define PREFER_DP4
+/*
+ * Take an already-setup and valid source then swizzle it appropriately to
+ * obtain a constant ZERO or ONE source.
+ */
+#define __CONST(x, y)	\
+	(PVS_SRC_OPERAND(t_src_index(vp, &src[x]),	\
+			   t_swizzle(y),	\
+			   t_swizzle(y),	\
+			   t_swizzle(y),	\
+			   t_swizzle(y),	\
+			   t_src_class(src[x].File), \
+			   VSF_FLAG_NONE) | (src[x].RelAddr << 4))
 
 #define FREE_TEMPS() \
 	do { \
-		if(u_temp_i < vp->num_temporaries) { \
-			WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \
+		int u_temp_used = (VSF_MAX_FRAGMENT_TEMPS - 1) - u_temp_i; \
+		if((vp->num_temporaries + u_temp_used) > VSF_MAX_FRAGMENT_TEMPS) { \
+			WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_used); \
 			vp->native = GL_FALSE; \
 		} \
 		u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \
 	} while (0)
 
 int r300VertexProgUpdateParams(GLcontext * ctx,
-			       struct r300_vertex_program_cont *vp,
-			       float *dst)
+			       struct r300_vertex_program_cont *vp, float *dst)
 {
 	int pi;
 	struct gl_vertex_program *mesa_vp = &vp->mesa_program;
@@ -141,7 +101,6 @@ int r300VertexProgUpdateParams(GLcontext * ctx,
 	paramList = mesa_vp->Base.Parameters;
 	for (pi = 0; pi < paramList->NumParameters; pi++) {
 		switch (paramList->Parameters[pi].Type) {
-
 		case PROGRAM_STATE_VAR:
 		case PROGRAM_NAMED_PARAM:
 			//fprintf(stderr, "%s", vp->Parameters->Parameters[pi].Name);
@@ -151,7 +110,6 @@ int r300VertexProgUpdateParams(GLcontext * ctx,
 			*dst++ = paramList->ParameterValues[pi][2];
 			*dst++ = paramList->ParameterValues[pi][3];
 			break;
-
 		default:
 			_mesa_problem(NULL, "Bad param type in %s",
 				      __FUNCTION__);
@@ -173,11 +131,11 @@ static unsigned long t_dst_class(enum register_file file)
 
 	switch (file) {
 	case PROGRAM_TEMPORARY:
-		return VSF_OUT_CLASS_TMP;
+		return PVS_DST_REG_TEMPORARY;
 	case PROGRAM_OUTPUT:
-		return VSF_OUT_CLASS_RESULT;
+		return PVS_DST_REG_OUT;
 	case PROGRAM_ADDRESS:
-		return VSF_OUT_CLASS_ADDR;
+		return PVS_DST_REG_A0;
 		/*
 		   case PROGRAM_INPUT:
 		   case PROGRAM_LOCAL_PARAM:
@@ -205,19 +163,17 @@ static unsigned long t_dst_index(struct r300_vertex_program *vp,
 
 static unsigned long t_src_class(enum register_file file)
 {
-
 	switch (file) {
 	case PROGRAM_TEMPORARY:
-		return VSF_IN_CLASS_TMP;
-
+		return PVS_SRC_REG_TEMPORARY;
 	case PROGRAM_INPUT:
-		return VSF_IN_CLASS_ATTR;
-
+		return PVS_SRC_REG_INPUT;
 	case PROGRAM_LOCAL_PARAM:
 	case PROGRAM_ENV_PARAM:
 	case PROGRAM_NAMED_PARAM:
+	case PROGRAM_CONSTANT:
 	case PROGRAM_STATE_VAR:
-		return VSF_IN_CLASS_PARAM;
+		return PVS_SRC_REG_CONSTANT;
 		/*
 		   case PROGRAM_OUTPUT:
 		   case PROGRAM_WRITE_ONLY:
@@ -230,7 +186,7 @@ static unsigned long t_src_class(enum register_file file)
 	}
 }
 
-static inline unsigned long t_swizzle(GLubyte swizzle)
+static INLINE unsigned long t_swizzle(GLubyte swizzle)
 {
 /* this is in fact a NOP as the Mesa SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */
 	return swizzle;
@@ -242,8 +198,8 @@ static void vp_dump_inputs(struct r300_vertex_program *vp, char *caller)
 	int i;
 
 	if (vp == NULL) {
-		fprintf(stderr, "vp null in call to %s from %s\n",
-			__FUNCTION__, caller);
+		fprintf(stderr, "vp null in call to %s from %s\n", __FUNCTION__,
+			caller);
 		return;
 	}
 
@@ -292,7 +248,7 @@ static unsigned long t_src(struct r300_vertex_program *vp,
 	/* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
-	return MAKE_VSF_SOURCE(t_src_index(vp, src),
+	return PVS_SRC_OPERAND(t_src_index(vp, src),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 1)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 2)),
@@ -307,7 +263,7 @@ static unsigned long t_src_scalar(struct r300_vertex_program *vp,
 	/* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
-	return MAKE_VSF_SOURCE(t_src_index(vp, src),
+	return PVS_SRC_OPERAND(t_src_index(vp, src),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
@@ -330,397 +286,364 @@ static GLboolean valid_dst(struct r300_vertex_program *vp,
 	return GL_TRUE;
 }
 
-/*
- * Instruction    Inputs  Output   Description
- * -----------    ------  ------   --------------------------------
- * ABS            v       v        absolute value
- * ADD            v,v     v        add
- * ARL            s       a        address register load
- * DP3            v,v     ssss     3-component dot product
- * DP4            v,v     ssss     4-component dot product
- * DPH            v,v     ssss     homogeneous dot product
- * DST            v,v     v        distance vector
- * EX2            s       ssss     exponential base 2
- * EXP            s       v        exponential base 2 (approximate)
- * FLR            v       v        floor
- * FRC            v       v        fraction
- * LG2            s       ssss     logarithm base 2
- * LIT            v       v        compute light coefficients
- * LOG            s       v        logarithm base 2 (approximate)
- * MAD            v,v,v   v        multiply and add
- * MAX            v,v     v        maximum
- * MIN            v,v     v        minimum
- * MOV            v       v        move
- * MUL            v,v     v        multiply
- * POW            s,s     ssss     exponentiate
- * RCP            s       ssss     reciprocal
- * RSQ            s       ssss     reciprocal square root
- * SGE            v,v     v        set on greater than or equal
- * SLT            v,v     v        set on less than
- * SUB            v,v     v        subtract
- * SWZ            v       v        extended swizzle
- * XPD            v,v     v        cross product
- *
- * Table X.5:  Summary of vertex program instructions.  "v" indicates a
- * floating-point vector input or output, "s" indicates a floating-point
- * scalar input, "ssss" indicates a scalar output replicated across a
- * 4-component result vector, and "a" indicates a single address register
- * component.
- */
-
-static GLuint *t_opcode_abs(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeABS(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_MAXIMUM,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
-			    t_src_class(src[0].File),
-			    (!src[0].
-			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
+				  t_src_class(src[0].File),
+				  (!src[0].
+				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[3] = 0;
 
 	return inst;
 }
 
-static GLuint *t_opcode_add(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeADD(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	unsigned long hw_op;
-
-#if 1
-	hw_op = (src[0].File == PROGRAM_TEMPORARY
-		 && src[1].File ==
-		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-	inst[1] = ONE_SRC_0;
-	inst[2] = t_src(vp, &src[0]);
-	inst[3] = t_src(vp, &src[1]);
-#else
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
-
-#endif
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_arl(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeARL(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ARL, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_FLT2FIX_DX,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_dp3(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeDP3(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO}
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
-	inst[1] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
-			    SWIZZLE_ZERO, t_src_class(src[0].File),
-			    src[0].
-			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+	inst[0] = PVS_OP_DST_OPERAND(VE_DOT_PRODUCT,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+				  SWIZZLE_ZERO,
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
-
 	inst[2] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+	    PVS_SRC_OPERAND(t_src_index(vp, &src[1]),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
-			    SWIZZLE_ZERO, t_src_class(src[1].File),
+			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)), SWIZZLE_ZERO,
+			    t_src_class(src[1].File),
 			    src[1].
 			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
-
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_dp4(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeDP4(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_DOT_PRODUCT,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_dph(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeDPH(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W}
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
-	inst[1] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
-			    VSF_IN_COMPONENT_ONE, t_src_class(src[0].File),
-			    src[0].
-			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+	inst[0] = PVS_OP_DST_OPERAND(VE_DOT_PRODUCT,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+				  PVS_SRC_SELECT_FORCE_1,
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_dst(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeDST(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_DST, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_DISTANCE_VECTOR,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_ex2(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeEX2(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_EX2, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(ME_EXP_BASE2_FULL_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_exp(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeEXP(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_EXP, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(ME_EXP_BASE2_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_flr(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3], int *u_temp_i)
+static GLuint *r300TranslateOpcodeFLR(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3],
+				      int *u_temp_i)
 {
 	/* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W}
 	   ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, *u_temp_i,
-			t_dst_mask(vpi->DstReg.WriteMask),
-			VSF_OUT_CLASS_TMP);
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_FRACTION,
+				     GL_FALSE,
+				     GL_FALSE,
+				     *u_temp_i,
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     PVS_DST_REG_TEMPORARY);
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 	inst += 4;
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] =
-	    MAKE_VSF_SOURCE(*u_temp_i, VSF_IN_COMPONENT_X,
-			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
-			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
-			    /* Not 100% sure about this */
-			    (!src[0].
-			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE
-			    /*VSF_FLAG_ALL */ );
-
-	inst[3] = ZERO_SRC_0;
+	inst[2] = PVS_SRC_OPERAND(*u_temp_i,
+				  PVS_SRC_SELECT_X,
+				  PVS_SRC_SELECT_Y,
+				  PVS_SRC_SELECT_Z,
+				  PVS_SRC_SELECT_W, PVS_SRC_REG_TEMPORARY,
+				  /* Not 100% sure about this */
+				  (!src[0].
+				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE
+				  /*VSF_FLAG_ALL */ );
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 	(*u_temp_i)--;
 
 	return inst;
 }
 
-static GLuint *t_opcode_frc(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeFRC(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_FRACTION,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_lg2(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeLG2(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	// LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X}
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
-	inst[1] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
-			    t_src_class(src[0].File),
-			    src[0].
-			    NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	inst[0] = PVS_OP_DST_OPERAND(ME_LOG_BASE2_FULL_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+				  t_src_class(src[0].File),
+				  src[0].
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_lit(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeLIT(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(ME_LIGHT_COEFF_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	/* NOTE: Users swizzling might not work. */
-	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-				  VSF_IN_COMPONENT_ZERO,	// z
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
+				  PVS_SRC_SELECT_FORCE_0,	// Z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
 				  t_src_class(src[0].File),
 				  src[0].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
-	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
-				  VSF_IN_COMPONENT_ZERO,	// z
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
+				  PVS_SRC_SELECT_FORCE_0,	// Z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
 				  t_src_class(src[0].File),
 				  src[0].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
-	inst[3] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-				  VSF_IN_COMPONENT_ZERO,	// z
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[3] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
+				  PVS_SRC_SELECT_FORCE_0,	// Z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
 				  src[0].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
 
 	return inst;
 }
 
-static GLuint *t_opcode_log(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeLOG(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_LOG, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(ME_LOG_BASE2_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_mad(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeMAD(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	unsigned long hw_op;
-
-	hw_op = (src[0].File == PROGRAM_TEMPORARY
-		 && src[1].File == PROGRAM_TEMPORARY
-		 && src[2].File ==
-		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(PVS_MACRO_OP_2CLK_MADD,
+				     GL_FALSE,
+				     GL_TRUE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
 	inst[3] = t_src(vp, &src[2]);
@@ -728,323 +651,302 @@ static GLuint *t_opcode_mad(struct r300_vertex_program *vp,
 	return inst;
 }
 
-static GLuint *t_opcode_max(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeMAX(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_MAXIMUM,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_min(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeMIN(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_MIN, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_MINIMUM,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_mov(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeMOV(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
 
-#if 1
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
-#else
-	hw_op =
-	    (src[0].File ==
-	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ONE_SRC_0;
-	inst[3] = ZERO_SRC_0;
-#endif
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_mul(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeMUL(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	unsigned long hw_op;
-
-	// HW mul can take third arg but appears to have some other limitations.
-
-	hw_op = (src[0].File == PROGRAM_TEMPORARY
-		 && src[1].File ==
-		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(VE_MULTIPLY,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_pow(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodePOW(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_POW, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(ME_POWER_FUNC_FF,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
 	inst[3] = t_src_scalar(vp, &src[1]);
 
 	return inst;
 }
 
-static GLuint *t_opcode_rcp(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeRCP(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_RCP, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(ME_RECIP_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_rsq(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeRSQ(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_RSQ, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(ME_RECIP_SQRT_DX,
+				     GL_TRUE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src_scalar(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_sge(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeSGE(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_SGE, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_SET_GREATER_THAN_EQUAL,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_slt(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeSLT(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_SLT, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	inst[0] = PVS_OP_DST_OPERAND(VE_SET_LESS_THAN,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
 	inst[2] = t_src(vp, &src[1]);
-	inst[3] = ZERO_SRC_1;
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_sub(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeSUB(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
-	unsigned long hw_op;
-
 	//ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
 
-#if 1
-	hw_op = (src[0].File == PROGRAM_TEMPORARY
-		 && src[1].File ==
-		 PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+#if 0
+	inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ONE_SRC_0;
-	inst[3] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
-			    t_src_class(src[1].File),
-			    (!src[1].
-			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[1]),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+				  t_src_class(src[1].File),
+				  (!src[1].
+				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
+	inst[3] = 0;
 #else
 	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
+	    PVS_OP_DST_OPERAND(VE_MULTIPLY_ADD,
+			       GL_FALSE,
+			       GL_FALSE,
+			       t_dst_index(vp, &vpi->DstReg),
+			       t_dst_mask(vpi->DstReg.WriteMask),
+			       t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] =
-	    MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
-			    t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
-			    t_src_class(src[1].File),
-			    (!src[1].
-			     NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	inst[2] = __CONST(0, SWIZZLE_ONE);
+	inst[3] = PVS_SRC_OPERAND(t_src_index(vp, &src[1]),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+				  t_src_class(src[1].File),
+				  (!src[1].
+				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
-	inst[3] = 0;
 #endif
 
 	return inst;
 }
 
-static GLuint *t_opcode_swz(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3])
+static GLuint *r300TranslateOpcodeSWZ(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3])
 {
 	//ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
 
-#if 1
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
+	inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
 	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ZERO_SRC_0;
-	inst[3] = ZERO_SRC_0;
-#else
-	hw_op =
-	    (src[0].File ==
-	     PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 :
-	    R300_VPI_OUT_OP_MAD;
-
-	inst[0] =
-	    MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-	inst[1] = t_src(vp, &src[0]);
-	inst[2] = ONE_SRC_0;
-	inst[3] = ZERO_SRC_0;
-#endif
+	inst[2] = __CONST(0, SWIZZLE_ZERO);
+	inst[3] = __CONST(0, SWIZZLE_ZERO);
 
 	return inst;
 }
 
-static GLuint *t_opcode_xpd(struct r300_vertex_program *vp,
-			    struct prog_instruction *vpi, GLuint * inst,
-			    struct prog_src_register src[3], int *u_temp_i)
+static GLuint *r300TranslateOpcodeXPD(struct r300_vertex_program *vp,
+				      struct prog_instruction *vpi,
+				      GLuint * inst,
+				      struct prog_src_register src[3],
+				      int *u_temp_i)
 {
 	/* mul r0, r1.yzxw, r2.zxyw
 	   mad r0, -r2.yzxw, r1.zxyw, r0
-	   NOTE: might need MAD_2
 	 */
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, *u_temp_i,
-			t_dst_mask(vpi->DstReg.WriteMask),
-			VSF_OUT_CLASS_TMP);
-
-	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+	inst[0] = PVS_OP_DST_OPERAND(VE_MULTIPLY_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     *u_temp_i,
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     PVS_DST_REG_TEMPORARY);
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// Z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
 				  src[0].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
-
-	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
+	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// Z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// X
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// W
 				  t_src_class(src[1].File),
 				  src[1].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
-
-	inst[3] = ZERO_SRC_1;
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[1].RelAddr << 4);
+	inst[3] = __CONST(1, SWIZZLE_ZERO);
 	inst += 4;
-	(*u_temp_i)--;
 
-	inst[0] =
-	    MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, t_dst_index(vp, &vpi->DstReg),
-			t_dst_mask(vpi->DstReg.WriteMask),
-			t_dst_class(vpi->DstReg.File));
-
-	inst[1] = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// z
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// x
-				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// w
+	inst[0] = PVS_OP_DST_OPERAND(VE_MULTIPLY_ADD,
+				     GL_FALSE,
+				     GL_FALSE,
+				     t_dst_index(vp, &vpi->DstReg),
+				     t_dst_mask(vpi->DstReg.WriteMask),
+				     t_dst_class(vpi->DstReg.File));
+	inst[1] = PVS_SRC_OPERAND(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// Z
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// X
+				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// W
 				  t_src_class(src[1].File),
 				  (!src[1].
-				   NegateBase) ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[1].RelAddr << 4);
-
-	inst[2] = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// z
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// x
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// y
-				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// w
+				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[1].RelAddr << 4);
+	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// Z
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
+				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
 				  src[0].
-				  NegateBase ? VSF_FLAG_ALL :
-				  VSF_FLAG_NONE) | (src[0].RelAddr << 4);
-
+				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+	    (src[0].RelAddr << 4);
 	inst[3] =
-	    MAKE_VSF_SOURCE(*u_temp_i + 1, VSF_IN_COMPONENT_X,
-			    VSF_IN_COMPONENT_Y, VSF_IN_COMPONENT_Z,
-			    VSF_IN_COMPONENT_W, VSF_IN_CLASS_TMP,
-			    VSF_FLAG_NONE);
+	    PVS_SRC_OPERAND(*u_temp_i, PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y,
+			    PVS_SRC_SELECT_Z, PVS_SRC_SELECT_W,
+			    PVS_SRC_REG_TEMPORARY, VSF_FLAG_NONE);
+
+	(*u_temp_i)--;
 
 	return inst;
 }
@@ -1145,23 +1047,24 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 		if (num_operands == 3) {	/* TODO: scalars */
 			if (CMP_SRCS(src[1], src[2])
 			    || CMP_SRCS(src[0], src[2])) {
-				inst[0] =
-				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-						u_temp_i, VSF_FLAG_ALL,
-						VSF_OUT_CLASS_TMP);
-
+				inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+							     GL_FALSE,
+							     GL_FALSE,
+							     u_temp_i,
+							     VSF_FLAG_ALL,
+							     PVS_DST_REG_TEMPORARY);
 				inst[1] =
-				    MAKE_VSF_SOURCE(t_src_index
-						    (vp, &src[2]),
-						    SWIZZLE_X, SWIZZLE_Y,
-						    SWIZZLE_Z, SWIZZLE_W,
-						    t_src_class(src[2].
-								File),
-						    VSF_FLAG_NONE) |
-				    (src[2].RelAddr << 4);
-
-				inst[2] = ZERO_SRC_2;
-				inst[3] = ZERO_SRC_2;
+				    PVS_SRC_OPERAND(t_src_index(vp, &src[2]),
+						    SWIZZLE_X,
+						    SWIZZLE_Y,
+						    SWIZZLE_Z,
+						    SWIZZLE_W,
+						    t_src_class(src[2].File),
+						    VSF_FLAG_NONE) | (src[2].
+								      RelAddr <<
+								      4);
+				inst[2] = __CONST(2, SWIZZLE_ZERO);
+				inst[3] = __CONST(2, SWIZZLE_ZERO);
 				inst += 4;
 
 				src[2].File = PROGRAM_TEMPORARY;
@@ -1173,23 +1076,24 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 		if (num_operands >= 2) {
 			if (CMP_SRCS(src[1], src[0])) {
-				inst[0] =
-				    MAKE_VSF_OP(R300_VPI_OUT_OP_ADD,
-						u_temp_i, VSF_FLAG_ALL,
-						VSF_OUT_CLASS_TMP);
-
+				inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+							     GL_FALSE,
+							     GL_FALSE,
+							     u_temp_i,
+							     VSF_FLAG_ALL,
+							     PVS_DST_REG_TEMPORARY);
 				inst[1] =
-				    MAKE_VSF_SOURCE(t_src_index
-						    (vp, &src[0]),
-						    SWIZZLE_X, SWIZZLE_Y,
-						    SWIZZLE_Z, SWIZZLE_W,
-						    t_src_class(src[0].
-								File),
-						    VSF_FLAG_NONE) |
-				    (src[0].RelAddr << 4);
-
-				inst[2] = ZERO_SRC_0;
-				inst[3] = ZERO_SRC_0;
+				    PVS_SRC_OPERAND(t_src_index(vp, &src[0]),
+						    SWIZZLE_X,
+						    SWIZZLE_Y,
+						    SWIZZLE_Z,
+						    SWIZZLE_W,
+						    t_src_class(src[0].File),
+						    VSF_FLAG_NONE) | (src[0].
+								      RelAddr <<
+								      4);
+				inst[2] = __CONST(0, SWIZZLE_ZERO);
+				inst[3] = __CONST(0, SWIZZLE_ZERO);
 				inst += 4;
 
 				src[0].File = PROGRAM_TEMPORARY;
@@ -1201,89 +1105,87 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 
 		switch (vpi->Opcode) {
 		case OPCODE_ABS:
-			inst = t_opcode_abs(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeABS(vp, vpi, inst, src);
 			break;
 		case OPCODE_ADD:
-			inst = t_opcode_add(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeADD(vp, vpi, inst, src);
 			break;
 		case OPCODE_ARL:
-			inst = t_opcode_arl(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeARL(vp, vpi, inst, src);
 			break;
 		case OPCODE_DP3:
-			inst = t_opcode_dp3(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeDP3(vp, vpi, inst, src);
 			break;
 		case OPCODE_DP4:
-			inst = t_opcode_dp4(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeDP4(vp, vpi, inst, src);
 			break;
 		case OPCODE_DPH:
-			inst = t_opcode_dph(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeDPH(vp, vpi, inst, src);
 			break;
 		case OPCODE_DST:
-			inst = t_opcode_dst(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeDST(vp, vpi, inst, src);
 			break;
 		case OPCODE_EX2:
-			inst = t_opcode_ex2(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeEX2(vp, vpi, inst, src);
 			break;
 		case OPCODE_EXP:
-			inst = t_opcode_exp(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeEXP(vp, vpi, inst, src);
 			break;
 		case OPCODE_FLR:
-			inst =
-			    t_opcode_flr(vp, vpi, inst, src, /* FIXME */
-					 &u_temp_i);
+			inst = r300TranslateOpcodeFLR(vp, vpi, inst, src,	/* FIXME */
+						      &u_temp_i);
 			break;
 		case OPCODE_FRC:
-			inst = t_opcode_frc(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeFRC(vp, vpi, inst, src);
 			break;
 		case OPCODE_LG2:
-			inst = t_opcode_lg2(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeLG2(vp, vpi, inst, src);
 			break;
 		case OPCODE_LIT:
-			inst = t_opcode_lit(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeLIT(vp, vpi, inst, src);
 			break;
 		case OPCODE_LOG:
-			inst = t_opcode_log(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeLOG(vp, vpi, inst, src);
 			break;
 		case OPCODE_MAD:
-			inst = t_opcode_mad(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeMAD(vp, vpi, inst, src);
 			break;
 		case OPCODE_MAX:
-			inst = t_opcode_max(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeMAX(vp, vpi, inst, src);
 			break;
 		case OPCODE_MIN:
-			inst = t_opcode_min(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeMIN(vp, vpi, inst, src);
 			break;
 		case OPCODE_MOV:
-			inst = t_opcode_mov(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeMOV(vp, vpi, inst, src);
 			break;
 		case OPCODE_MUL:
-			inst = t_opcode_mul(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeMUL(vp, vpi, inst, src);
 			break;
 		case OPCODE_POW:
-			inst = t_opcode_pow(vp, vpi, inst, src);
+			inst = r300TranslateOpcodePOW(vp, vpi, inst, src);
 			break;
 		case OPCODE_RCP:
-			inst = t_opcode_rcp(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeRCP(vp, vpi, inst, src);
 			break;
 		case OPCODE_RSQ:
-			inst = t_opcode_rsq(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeRSQ(vp, vpi, inst, src);
 			break;
 		case OPCODE_SGE:
-			inst = t_opcode_sge(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeSGE(vp, vpi, inst, src);
 			break;
 		case OPCODE_SLT:
-			inst = t_opcode_slt(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeSLT(vp, vpi, inst, src);
 			break;
 		case OPCODE_SUB:
-			inst = t_opcode_sub(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeSUB(vp, vpi, inst, src);
 			break;
 		case OPCODE_SWZ:
-			inst = t_opcode_swz(vp, vpi, inst, src);
+			inst = r300TranslateOpcodeSWZ(vp, vpi, inst, src);
 			break;
 		case OPCODE_XPD:
-			inst =
-			    t_opcode_xpd(vp, vpi, inst, src, /* FIXME */
-					 &u_temp_i);
+			inst = r300TranslateOpcodeXPD(vp, vpi, inst, src,	/* FIXME */
+						      &u_temp_i);
 			break;
 		default:
 			assert(0);
@@ -1291,6 +1193,23 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 		}
 	}
 
+	/* Some outputs may be artificially added, to match the inputs
+	   of the fragment program. Blank the outputs here. */
+	for (i = 0; i < VERT_RESULT_MAX; i++) {
+		if (vp->key.OutputsAdded & (1 << i)) {
+			inst[0] = PVS_OP_DST_OPERAND(VE_ADD,
+						     GL_FALSE,
+						     GL_FALSE,
+						     vp->outputs[i],
+						     VSF_FLAG_ALL,
+						     PVS_DST_REG_OUT);
+			inst[1] = __CONST(0, SWIZZLE_ZERO);
+			inst[2] = __CONST(0, SWIZZLE_ZERO);
+			inst[3] = __CONST(0, SWIZZLE_ZERO);
+			inst += 4;
+		}
+	}
+
 	vp->program.length = (inst - vp->program.body.i);
 	if (vp->program.length >= VSF_MAX_FRAGMENT_LENGTH) {
 		vp->program.length = 0;
@@ -1303,14 +1222,16 @@ static void r300TranslateVertexShader(struct r300_vertex_program *vp,
 #endif
 }
 
+/* DP4 version seems to trigger some hw peculiarity */
+//#define PREFER_DP4
+
 static void position_invariant(struct gl_program *prog)
 {
 	struct prog_instruction *vpi;
 	struct gl_program_parameter_list *paramList;
 	int i;
 
-	gl_state_index tokens[STATE_LENGTH] =
-	    { STATE_MVP_MATRIX, 0, 0, 0, 0 };
+	gl_state_index tokens[STATE_LENGTH] = { STATE_MVP_MATRIX, 0, 0, 0, 0 };
 
 	/* tokens[4] = matrix modifier */
 #ifdef PREFER_DP4
@@ -1390,8 +1311,8 @@ static void position_invariant(struct gl_program *prog)
 	assert(vpi->Opcode == OPCODE_END);
 }
 
-static void insert_wpos(struct r300_vertex_program *vp,
-			struct gl_program *prog, GLuint temp_index)
+static void insert_wpos(struct r300_vertex_program *vp, struct gl_program *prog,
+			GLuint temp_index)
 {
 	struct prog_instruction *vpi;
 	struct prog_instruction *vpi_insert;
@@ -1404,8 +1325,8 @@ static void insert_wpos(struct r300_vertex_program *vp,
 				prog->NumInstructions - 1);
 	/* END */
 	_mesa_copy_instructions(&vpi[prog->NumInstructions + 1],
-				&prog->Instructions[prog->NumInstructions -
-						    1], 1);
+				&prog->Instructions[prog->NumInstructions - 1],
+				1);
 	vpi_insert = &vpi[prog->NumInstructions - 1];
 
 	vpi_insert[i].Opcode = OPCODE_MOV;
@@ -1485,6 +1406,15 @@ static struct r300_vertex_program *build_program(struct r300_vertex_program_key
 	return vp;
 }
 
+static void add_outputs(struct r300_vertex_program_key *key, GLint vert)
+{
+	if (key->OutputsWritten & (1 << vert))
+		return;
+
+	key->OutputsWritten |= 1 << vert;
+	key->OutputsAdded |= 1 << vert;
+}
+
 void r300SelectVertexShader(r300ContextPtr r300)
 {
 	GLcontext *ctx = ctx = r300->radeon.glCtx;
@@ -1495,8 +1425,9 @@ void r300SelectVertexShader(r300ContextPtr r300)
 	struct r300_vertex_program *vp;
 	GLint wpos_idx;
 
-	vpc =
-	    (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
+	vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
+	wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
+	wanted_key.OutputsWritten = vpc->mesa_program.Base.OutputsWritten;
 	InputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
 
 	wpos_idx = -1;
@@ -1510,33 +1441,29 @@ void r300SelectVertexShader(r300ContextPtr r300)
 			_mesa_exit(-1);
 		}
 
-		InputsRead |= (FRAG_BIT_TEX0 << i);
+		wanted_key.OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i);
 		wpos_idx = i;
 	}
-	wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
-	wanted_key.OutputsWritten = vpc->mesa_program.Base.OutputsWritten;
 
-	wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS;
+	add_outputs(&wanted_key, VERT_RESULT_HPOS);
 
 	if (InputsRead & FRAG_BIT_COL0) {
-		wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL0;
+		add_outputs(&wanted_key, VERT_RESULT_COL0);
 	}
 
-	if ((InputsRead & FRAG_BIT_COL1)) {
-		wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL1;
+	if (InputsRead & FRAG_BIT_COL1) {
+		add_outputs(&wanted_key, VERT_RESULT_COL1);
 	}
 
 	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 		if (InputsRead & (FRAG_BIT_TEX0 << i)) {
-			wanted_key.OutputsWritten |=
-			    1 << (VERT_RESULT_TEX0 + i);
+			add_outputs(&wanted_key, VERT_RESULT_TEX0 + i);
 		}
 	}
 
 	if (vpc->mesa_program.IsPositionInvariant) {
 		/* we wan't position don't we ? */
 		wanted_key.InputsRead |= (1 << VERT_ATTRIB_POS);
-		wanted_key.OutputsWritten |= (1 << VERT_RESULT_HPOS);
 	}
 
 	for (vp = vpc->progs; vp; vp = vp->next)
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.h b/src/mesa/drivers/dri/r300/r300_vertprog.h
index 3df0eee7991..2f35f02bc84 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.h
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.h
@@ -3,6 +3,25 @@
 
 #include "r300_reg.h"
 
+#define PVS_OP_DST_OPERAND(opcode, math_inst, macro_inst, reg_index, reg_writemask, reg_class)	\
+	 (((opcode & PVS_DST_OPCODE_MASK) << PVS_DST_OPCODE_SHIFT)	\
+	 | ((math_inst & PVS_DST_MATH_INST_MASK) << PVS_DST_MATH_INST_SHIFT)	\
+	 | ((macro_inst & PVS_DST_MACRO_INST_MASK) << PVS_DST_MACRO_INST_SHIFT)	\
+	 | ((reg_index & PVS_DST_OFFSET_MASK) << PVS_DST_OFFSET_SHIFT)	\
+	 | ((reg_writemask & 0xf) << PVS_DST_WE_X_SHIFT)	/* X Y Z W */	\
+	 | ((reg_class & PVS_DST_REG_TYPE_MASK) << PVS_DST_REG_TYPE_SHIFT))
+
+#define PVS_SRC_OPERAND(in_reg_index, comp_x, comp_y, comp_z, comp_w, reg_class, negate)	\
+	(((in_reg_index & PVS_SRC_OFFSET_MASK) << PVS_SRC_OFFSET_SHIFT)				\
+	 | ((comp_x & PVS_SRC_SWIZZLE_X_MASK) << PVS_SRC_SWIZZLE_X_SHIFT)			\
+	 | ((comp_y & PVS_SRC_SWIZZLE_Y_MASK) << PVS_SRC_SWIZZLE_Y_SHIFT)			\
+	 | ((comp_z & PVS_SRC_SWIZZLE_Z_MASK) << PVS_SRC_SWIZZLE_Z_SHIFT)			\
+	 | ((comp_w & PVS_SRC_SWIZZLE_W_MASK) << PVS_SRC_SWIZZLE_W_SHIFT)			\
+	 | ((negate & 0xf) << PVS_SRC_MODIFIER_X_SHIFT)	/* X Y Z W */				\
+	 | ((reg_class & PVS_SRC_REG_TYPE_MASK) << PVS_SRC_REG_TYPE_SHIFT))
+
+#if 1
+
 #define VSF_FLAG_X	1
 #define VSF_FLAG_Y	2
 #define VSF_FLAG_Z	4
@@ -11,74 +30,6 @@
 #define VSF_FLAG_ALL  0xf
 #define VSF_FLAG_NONE  0
 
-#define VSF_OUT_CLASS_TMP	0
-#define VSF_OUT_CLASS_ADDR	1
-#define VSF_OUT_CLASS_RESULT	2
-
-/* first DWORD of an instruction */
-
-/* possible operations: 
-    DOT, MUL, ADD, MAD, FRC, MAX, MIN, SGE, SLT, EXP, LOG, LIT, POW, RCP, RSQ, EX2,
-    LG2, MAD_2 */
-
-#define MAKE_VSF_OP(op, out_reg_index, out_reg_fields, class) \
-   ((op)  \
-  	| ((out_reg_index) << R300_VPI_OUT_REG_INDEX_SHIFT) 	\
- 	 | ((out_reg_fields) << 20) 	\
-  	| ( (class) << 8 ) )
-
-#define EASY_VSF_OP(op, out_reg_index, out_reg_fields, class) \
-	MAKE_VSF_OP(R300_VPI_OUT_OP_##op, out_reg_index, VSF_FLAG_##out_reg_fields, VSF_OUT_CLASS_##class) \
-
-/* according to Nikolai, the subsequent 3 DWORDs are sources, use same define for each */
-
-#define VSF_IN_CLASS_TMP	0
-#define VSF_IN_CLASS_ATTR	1
-#define VSF_IN_CLASS_PARAM	2
-#define VSF_IN_CLASS_NONE	9
-
-#define VSF_IN_COMPONENT_X	0
-#define VSF_IN_COMPONENT_Y	1
-#define VSF_IN_COMPONENT_Z	2
-#define VSF_IN_COMPONENT_W	3
-#define VSF_IN_COMPONENT_ZERO	4
-#define VSF_IN_COMPONENT_ONE	5
-
-#define MAKE_VSF_SOURCE(in_reg_index, comp_x, comp_y, comp_z, comp_w, class, negate) \
-	( ((in_reg_index)<<R300_VPI_IN_REG_INDEX_SHIFT) \
-	   | ((comp_x)<<R300_VPI_IN_X_SHIFT) \
-	   | ((comp_y)<<R300_VPI_IN_Y_SHIFT) \
-	   | ((comp_z)<<R300_VPI_IN_Z_SHIFT) \
-	   | ((comp_w)<<R300_VPI_IN_W_SHIFT) \
-	   | ((negate)<<25) | ((class)))
-
-#define EASY_VSF_SOURCE(in_reg_index, comp_x, comp_y, comp_z, comp_w, class, negate) \
-	MAKE_VSF_SOURCE(in_reg_index, \
-		VSF_IN_COMPONENT_##comp_x, \
-		VSF_IN_COMPONENT_##comp_y, \
-		VSF_IN_COMPONENT_##comp_z, \
-		VSF_IN_COMPONENT_##comp_w, \
-		VSF_IN_CLASS_##class, VSF_FLAG_##negate)
-
-/* special sources: */
-
-/* (1.0,1.0,1.0,1.0) vector (ATTR, plain ) */
-#define VSF_ATTR_UNITY(reg) 	EASY_VSF_SOURCE(reg, ONE, ONE, ONE, ONE, ATTR, NONE)
-#define VSF_UNITY(reg) 	EASY_VSF_SOURCE(reg, ONE, ONE, ONE, ONE, NONE, NONE)
-
-/* contents of unmodified register */
-#define VSF_REG(reg) 	EASY_VSF_SOURCE(reg, X, Y, Z, W, ATTR, NONE)
-
-/* contents of unmodified parameter */
-#define VSF_PARAM(reg) 	EASY_VSF_SOURCE(reg, X, Y, Z, W, PARAM, NONE)
-
-/* contents of unmodified temporary register */
-#define VSF_TMP(reg) 	EASY_VSF_SOURCE(reg, X, Y, Z, W, TMP, NONE)
-
-/* components of ATTR register */
-#define VSF_ATTR_X(reg) EASY_VSF_SOURCE(reg, X, X, X, X, ATTR, NONE)
-#define VSF_ATTR_Y(reg) EASY_VSF_SOURCE(reg, Y, Y, Y, Y, ATTR, NONE)
-#define VSF_ATTR_Z(reg) EASY_VSF_SOURCE(reg, Z, Z, Z, Z, ATTR, NONE)
-#define VSF_ATTR_W(reg) EASY_VSF_SOURCE(reg, W, W, W, W, ATTR, NONE)
+#endif
 
 #endif
diff --git a/src/mesa/drivers/dri/r300/r500_fragprog.c b/src/mesa/drivers/dri/r300/r500_fragprog.c
new file mode 100644
index 00000000000..75dae86fa81
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r500_fragprog.c
@@ -0,0 +1,700 @@
+/*
+ * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "r500_fragprog.h"
+
+#include "radeon_nqssadce.h"
+#include "radeon_program_alu.h"
+
+
+static struct prog_src_register shadow_ambient(struct gl_program *program, int tmu)
+{
+	gl_state_index fail_value_tokens[STATE_LENGTH] = {
+		STATE_INTERNAL, STATE_SHADOW_AMBIENT, 0, 0, 0
+	};
+	struct prog_src_register reg = { 0, };
+
+	fail_value_tokens[2] = tmu;
+	reg.File = PROGRAM_STATE_VAR;
+	reg.Index = _mesa_add_state_reference(program->Parameters, fail_value_tokens);
+	reg.Swizzle = SWIZZLE_WWWW;
+	return reg;
+}
+
+/**
+ * Transform TEX, TXP, TXB, and KIL instructions in the following way:
+ *  - premultiply texture coordinates for RECT
+ *  - extract operand swizzles
+ *  - introduce a temporary register when write masks are needed
+ *
+ */
+static GLboolean transform_TEX(
+	struct radeon_transform_context *t,
+	struct prog_instruction* orig_inst, void* data)
+{
+	struct r500_fragment_program_compiler *compiler =
+		(struct r500_fragment_program_compiler*)data;
+	struct prog_instruction inst = *orig_inst;
+	struct prog_instruction* tgt;
+	GLboolean destredirect = GL_FALSE;
+
+	if (inst.Opcode != OPCODE_TEX &&
+	    inst.Opcode != OPCODE_TXB &&
+	    inst.Opcode != OPCODE_TXP &&
+	    inst.Opcode != OPCODE_KIL)
+		return GL_FALSE;
+
+	/* ARB_shadow & EXT_shadow_funcs */
+	if (inst.Opcode != OPCODE_KIL &&
+	    t->Program->ShadowSamplers & (1 << inst.TexSrcUnit)) {
+		GLuint comparefunc = GL_NEVER + compiler->fp->state.unit[inst.TexSrcUnit].texture_compare_func;
+
+		if (comparefunc == GL_NEVER || comparefunc == GL_ALWAYS) {
+			tgt = radeonAppendInstructions(t->Program, 1);
+
+			tgt->Opcode = OPCODE_MOV;
+			tgt->DstReg = inst.DstReg;
+			if (comparefunc == GL_ALWAYS) {
+				tgt->SrcReg[0].File = PROGRAM_BUILTIN;
+				tgt->SrcReg[0].Swizzle = SWIZZLE_1111;
+			} else {
+				tgt->SrcReg[0] = shadow_ambient(t->Program, inst.TexSrcUnit);
+			}
+			return GL_TRUE;
+		}
+
+		inst.DstReg.File = PROGRAM_TEMPORARY;
+		inst.DstReg.Index = radeonFindFreeTemporary(t);
+		inst.DstReg.WriteMask = WRITEMASK_XYZW;
+	} else if (inst.Opcode != OPCODE_KIL && inst.DstReg.File != PROGRAM_TEMPORARY) {
+		int tempreg = radeonFindFreeTemporary(t);
+
+		inst.DstReg.File = PROGRAM_TEMPORARY;
+		inst.DstReg.Index = tempreg;
+		inst.DstReg.WriteMask = WRITEMASK_XYZW;
+		destredirect = GL_TRUE;
+	}
+
+	tgt = radeonAppendInstructions(t->Program, 1);
+	_mesa_copy_instructions(tgt, &inst, 1);
+
+	if (inst.Opcode != OPCODE_KIL &&
+	    t->Program->ShadowSamplers & (1 << inst.TexSrcUnit)) {
+		GLuint comparefunc = GL_NEVER + compiler->fp->state.unit[inst.TexSrcUnit].texture_compare_func;
+		GLuint depthmode = compiler->fp->state.unit[inst.TexSrcUnit].depth_texture_mode;
+		int rcptemp = radeonFindFreeTemporary(t);
+		int pass, fail;
+
+		tgt = radeonAppendInstructions(t->Program, 3);
+
+		tgt[0].Opcode = OPCODE_RCP;
+		tgt[0].DstReg.File = PROGRAM_TEMPORARY;
+		tgt[0].DstReg.Index = rcptemp;
+		tgt[0].DstReg.WriteMask = WRITEMASK_W;
+		tgt[0].SrcReg[0] = inst.SrcReg[0];
+		tgt[0].SrcReg[0].Swizzle = SWIZZLE_WWWW;
+
+		tgt[1].Opcode = OPCODE_MAD;
+		tgt[1].DstReg = inst.DstReg;
+		tgt[1].DstReg.WriteMask = orig_inst->DstReg.WriteMask;
+		tgt[1].SrcReg[0] = inst.SrcReg[0];
+		tgt[1].SrcReg[0].Swizzle = SWIZZLE_ZZZZ;
+		tgt[1].SrcReg[1].File = PROGRAM_TEMPORARY;
+		tgt[1].SrcReg[1].Index = rcptemp;
+		tgt[1].SrcReg[1].Swizzle = SWIZZLE_WWWW;
+		tgt[1].SrcReg[2].File = PROGRAM_TEMPORARY;
+		tgt[1].SrcReg[2].Index = inst.DstReg.Index;
+		if (depthmode == 0) /* GL_LUMINANCE */
+			tgt[1].SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_Z);
+		else if (depthmode == 2) /* GL_ALPHA */
+			tgt[1].SrcReg[2].Swizzle = SWIZZLE_WWWW;
+
+		/* Recall that SrcReg[0] is tex, SrcReg[2] is r and:
+		 *   r  < tex  <=>      -tex+r < 0
+		 *   r >= tex  <=> not (-tex+r < 0 */
+		if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
+			tgt[1].SrcReg[2].NegateBase = tgt[0].SrcReg[2].NegateBase ^ NEGATE_XYZW;
+		else
+			tgt[1].SrcReg[0].NegateBase = tgt[0].SrcReg[0].NegateBase ^ NEGATE_XYZW;
+
+		tgt[2].Opcode = OPCODE_CMP;
+		tgt[2].DstReg = orig_inst->DstReg;
+		tgt[2].SrcReg[0].File = PROGRAM_TEMPORARY;
+		tgt[2].SrcReg[0].Index = tgt[1].DstReg.Index;
+
+		if (comparefunc == GL_LESS || comparefunc == GL_GREATER) {
+			pass = 1;
+			fail = 2;
+		} else {
+			pass = 2;
+			fail = 1;
+		}
+
+		tgt[2].SrcReg[pass].File = PROGRAM_BUILTIN;
+		tgt[2].SrcReg[pass].Swizzle = SWIZZLE_1111;
+		tgt[2].SrcReg[fail] = shadow_ambient(t->Program, inst.TexSrcUnit);
+	} else if (destredirect) {
+		tgt = radeonAppendInstructions(t->Program, 1);
+
+		tgt->Opcode = OPCODE_MOV;
+		tgt->DstReg = orig_inst->DstReg;
+		tgt->SrcReg[0].File = PROGRAM_TEMPORARY;
+		tgt->SrcReg[0].Index = inst.DstReg.Index;
+	}
+
+	return GL_TRUE;
+}
+
+
+static void update_params(r300ContextPtr r300, struct r500_fragment_program *fp)
+{
+	struct gl_fragment_program *mp = &fp->mesa_program;
+
+	/* Ask Mesa nicely to fill in ParameterValues for us */
+	if (mp->Base.Parameters)
+		_mesa_load_state_parameters(r300->radeon.glCtx, mp->Base.Parameters);
+}
+
+
+/**
+ * Transform the program to support fragment.position.
+ *
+ * Introduce a small fragment at the start of the program that will be
+ * the only code that directly reads the FRAG_ATTRIB_WPOS input.
+ * All other code pieces that reference that input will be rewritten
+ * to read from a newly allocated temporary.
+ *
+ * \todo if/when r5xx supports the radeon_program architecture, this is a
+ * likely candidate for code sharing.
+ */
+static void insert_WPOS_trailer(struct r500_fragment_program_compiler *compiler)
+{
+	GLuint InputsRead = compiler->fp->mesa_program.Base.InputsRead;
+
+	if (!(InputsRead & FRAG_BIT_WPOS))
+		return;
+
+	static gl_state_index tokens[STATE_LENGTH] = {
+		STATE_INTERNAL, STATE_R300_WINDOW_DIMENSION, 0, 0, 0
+	};
+	struct prog_instruction *fpi;
+	GLuint window_index;
+	int i = 0;
+	GLuint tempregi = _mesa_find_free_register(compiler->program, PROGRAM_TEMPORARY);
+
+	_mesa_insert_instructions(compiler->program, 0, 3);
+	fpi = compiler->program->Instructions;
+
+	/* perspective divide */
+	fpi[i].Opcode = OPCODE_RCP;
+
+	fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+	fpi[i].DstReg.Index = tempregi;
+	fpi[i].DstReg.WriteMask = WRITEMASK_W;
+	fpi[i].DstReg.CondMask = COND_TR;
+
+	fpi[i].SrcReg[0].File = PROGRAM_INPUT;
+	fpi[i].SrcReg[0].Index = FRAG_ATTRIB_WPOS;
+	fpi[i].SrcReg[0].Swizzle = SWIZZLE_WWWW;
+	i++;
+
+	fpi[i].Opcode = OPCODE_MUL;
+
+	fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+	fpi[i].DstReg.Index = tempregi;
+	fpi[i].DstReg.WriteMask = WRITEMASK_XYZ;
+	fpi[i].DstReg.CondMask = COND_TR;
+
+	fpi[i].SrcReg[0].File = PROGRAM_INPUT;
+	fpi[i].SrcReg[0].Index = FRAG_ATTRIB_WPOS;
+	fpi[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+
+	fpi[i].SrcReg[1].File = PROGRAM_TEMPORARY;
+	fpi[i].SrcReg[1].Index = tempregi;
+	fpi[i].SrcReg[1].Swizzle = SWIZZLE_WWWW;
+	i++;
+
+	/* viewport transformation */
+	window_index = _mesa_add_state_reference(compiler->program->Parameters, tokens);
+
+	fpi[i].Opcode = OPCODE_MAD;
+
+	fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+	fpi[i].DstReg.Index = tempregi;
+	fpi[i].DstReg.WriteMask = WRITEMASK_XYZ;
+	fpi[i].DstReg.CondMask = COND_TR;
+
+	fpi[i].SrcReg[0].File = PROGRAM_TEMPORARY;
+	fpi[i].SrcReg[0].Index = tempregi;
+	fpi[i].SrcReg[0].Swizzle =
+	    MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+
+	fpi[i].SrcReg[1].File = PROGRAM_STATE_VAR;
+	fpi[i].SrcReg[1].Index = window_index;
+	fpi[i].SrcReg[1].Swizzle =
+	    MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+
+	fpi[i].SrcReg[2].File = PROGRAM_STATE_VAR;
+	fpi[i].SrcReg[2].Index = window_index;
+	fpi[i].SrcReg[2].Swizzle =
+	    MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+	i++;
+
+	for (; i < compiler->program->NumInstructions; ++i) {
+		int reg;
+		for (reg = 0; reg < 3; reg++) {
+			if (fpi[i].SrcReg[reg].File == PROGRAM_INPUT &&
+			    fpi[i].SrcReg[reg].Index == FRAG_ATTRIB_WPOS) {
+				fpi[i].SrcReg[reg].File = PROGRAM_TEMPORARY;
+				fpi[i].SrcReg[reg].Index = tempregi;
+			}
+		}
+	}
+}
+
+
+static void nqssadce_init(struct nqssadce_state* s)
+{
+	s->Outputs[FRAG_RESULT_COLR].Sourced = WRITEMASK_XYZW;
+	s->Outputs[FRAG_RESULT_DEPR].Sourced = WRITEMASK_W;
+}
+
+static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
+{
+	GLuint relevant;
+	int i;
+
+	if (opcode == OPCODE_TEX ||
+	    opcode == OPCODE_TXB ||
+	    opcode == OPCODE_TXP ||
+	    opcode == OPCODE_KIL) {
+		if (reg.Abs)
+			return GL_FALSE;
+
+		if (reg.NegateAbs)
+			reg.NegateBase ^= 15;
+
+		if (opcode == OPCODE_KIL) {
+			if (reg.Swizzle != SWIZZLE_NOOP)
+				return GL_FALSE;
+		} else {
+			for(i = 0; i < 4; ++i) {
+				GLuint swz = GET_SWZ(reg.Swizzle, i);
+				if (swz == SWIZZLE_NIL) {
+					reg.NegateBase &= ~(1 << i);
+					continue;
+				}
+				if (swz >= 4)
+					return GL_FALSE;
+			}
+		}
+
+		if (reg.NegateBase)
+			return GL_FALSE;
+
+		return GL_TRUE;
+	} else if (opcode == OPCODE_DDX || opcode == OPCODE_DDY) {
+		/* DDX/MDH and DDY/MDV explicitly ignore incoming swizzles;
+		 * if it doesn't fit perfectly into a .xyzw case... */
+		if (reg.Swizzle == SWIZZLE_NOOP && !reg.Abs
+				&& !reg.NegateBase && !reg.NegateAbs)
+			return GL_TRUE;
+
+		return GL_FALSE;
+	} else {
+		/* ALU instructions support almost everything */
+		if (reg.Abs)
+			return GL_TRUE;
+
+		relevant = 0;
+		for(i = 0; i < 3; ++i) {
+			GLuint swz = GET_SWZ(reg.Swizzle, i);
+			if (swz != SWIZZLE_NIL && swz != SWIZZLE_ZERO)
+				relevant |= 1 << i;
+		}
+		if ((reg.NegateBase & relevant) && ((reg.NegateBase & relevant) != relevant))
+			return GL_FALSE;
+
+		return GL_TRUE;
+	}
+}
+
+/**
+ * Implement a MOV with a potentially non-native swizzle.
+ *
+ * The only thing we *cannot* do in an ALU instruction is per-component
+ * negation. Therefore, we split the MOV into two instructions when necessary.
+ */
+static void nqssadce_build_swizzle(struct nqssadce_state *s,
+	struct prog_dst_register dst, struct prog_src_register src)
+{
+	struct prog_instruction *inst;
+	GLuint negatebase[2] = { 0, 0 };
+	int i;
+
+	for(i = 0; i < 4; ++i) {
+		GLuint swz = GET_SWZ(src.Swizzle, i);
+		if (swz == SWIZZLE_NIL)
+			continue;
+		negatebase[GET_BIT(src.NegateBase, i)] |= 1 << i;
+	}
+
+	_mesa_insert_instructions(s->Program, s->IP, (negatebase[0] ? 1 : 0) + (negatebase[1] ? 1 : 0));
+	inst = s->Program->Instructions + s->IP;
+
+	for(i = 0; i <= 1; ++i) {
+		if (!negatebase[i])
+			continue;
+
+		inst->Opcode = OPCODE_MOV;
+		inst->DstReg = dst;
+		inst->DstReg.WriteMask = negatebase[i];
+		inst->SrcReg[0] = src;
+		inst++;
+		s->IP++;
+	}
+}
+
+static GLuint build_dtm(GLuint depthmode)
+{
+	switch(depthmode) {
+	default:
+	case GL_LUMINANCE: return 0;
+	case GL_INTENSITY: return 1;
+	case GL_ALPHA: return 2;
+	}
+}
+
+static GLuint build_func(GLuint comparefunc)
+{
+	return comparefunc - GL_NEVER;
+}
+
+
+/**
+ * Collect all external state that is relevant for compiling the given
+ * fragment program.
+ */
+static void build_state(
+	r300ContextPtr r300,
+	struct r500_fragment_program *fp,
+	struct r500_fragment_program_external_state *state)
+{
+	int unit;
+
+	_mesa_bzero(state, sizeof(*state));
+
+	for(unit = 0; unit < 16; ++unit) {
+		if (fp->mesa_program.Base.ShadowSamplers & (1 << unit)) {
+			struct gl_texture_object* tex = r300->radeon.glCtx->Texture.Unit[unit]._Current;
+
+			state->unit[unit].depth_texture_mode = build_dtm(tex->DepthMode);
+			state->unit[unit].texture_compare_func = build_func(tex->CompareFunc);
+		}
+	}
+}
+
+static void dump_program(struct r500_fragment_program_code *code);
+
+void r500TranslateFragmentShader(r300ContextPtr r300,
+				 struct r500_fragment_program *fp)
+{
+	struct r500_fragment_program_external_state state;
+
+	build_state(r300, fp, &state);
+	if (_mesa_memcmp(&fp->state, &state, sizeof(state))) {
+		/* TODO: cache compiled programs */
+		fp->translated = GL_FALSE;
+		_mesa_memcpy(&fp->state, &state, sizeof(state));
+	}
+
+	if (!fp->translated) {
+		struct r500_fragment_program_compiler compiler;
+
+		compiler.r300 = r300;
+		compiler.fp = fp;
+		compiler.code = &fp->code;
+		compiler.program = _mesa_clone_program(r300->radeon.glCtx, &fp->mesa_program.Base);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Compiler: Initial program:\n");
+			_mesa_print_program(compiler.program);
+		}
+
+		insert_WPOS_trailer(&compiler);
+
+		struct radeon_program_transformation transformations[] = {
+			{ &transform_TEX, &compiler },
+			{ &radeonTransformALU, 0 },
+			{ &radeonTransformDeriv, 0 },
+			{ &radeonTransformTrigScale, 0 }
+		};
+		radeonLocalTransform(r300->radeon.glCtx, compiler.program,
+			4, transformations);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Compiler: after native rewrite:\n");
+			_mesa_print_program(compiler.program);
+		}
+
+		struct radeon_nqssadce_descr nqssadce = {
+			.Init = &nqssadce_init,
+			.IsNativeSwizzle = &is_native_swizzle,
+			.BuildSwizzle = &nqssadce_build_swizzle,
+			.RewriteDepthOut = GL_TRUE
+		};
+		radeonNqssaDce(r300->radeon.glCtx, compiler.program, &nqssadce);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			_mesa_printf("Compiler: after NqSSA-DCE:\n");
+			_mesa_print_program(compiler.program);
+		}
+
+		fp->translated = r500FragmentProgramEmit(&compiler);
+
+		/* Subtle: Rescue any parameters that have been added during transformations */
+		_mesa_free_parameter_list(fp->mesa_program.Base.Parameters);
+		fp->mesa_program.Base.Parameters = compiler.program->Parameters;
+		compiler.program->Parameters = 0;
+
+		_mesa_reference_program(r300->radeon.glCtx, &compiler.program, 0);
+
+		r300UpdateStateParameters(r300->radeon.glCtx, _NEW_PROGRAM);
+
+		if (RADEON_DEBUG & DEBUG_PIXEL) {
+			if (fp->translated) {
+				_mesa_printf("Machine-readable code:\n");
+				dump_program(&fp->code);
+			}
+		}
+
+	}
+
+	update_params(r300, fp);
+
+}
+
+static char *toswiz(int swiz_val) {
+  switch(swiz_val) {
+  case 0: return "R";
+  case 1: return "G";
+  case 2: return "B";
+  case 3: return "A";
+  case 4: return "0";
+  case 5: return "1/2";
+  case 6: return "1";
+  case 7: return "U";
+  }
+  return NULL;
+}
+
+static char *toop(int op_val)
+{
+  char *str = NULL;
+  switch (op_val) {
+  case 0: str = "MAD"; break;
+  case 1: str = "DP3"; break;
+  case 2: str = "DP4"; break;
+  case 3: str = "D2A"; break;
+  case 4: str = "MIN"; break;
+  case 5: str = "MAX"; break;
+  case 6: str = "Reserved"; break;
+  case 7: str = "CND"; break;
+  case 8: str = "CMP"; break;
+  case 9: str = "FRC"; break;
+  case 10: str = "SOP"; break;
+  case 11: str = "MDH"; break;
+  case 12: str = "MDV"; break;
+  }
+  return str;
+}
+
+static char *to_alpha_op(int op_val)
+{
+  char *str = NULL;
+  switch (op_val) {
+  case 0: str = "MAD"; break;
+  case 1: str = "DP"; break;
+  case 2: str = "MIN"; break;
+  case 3: str = "MAX"; break;
+  case 4: str = "Reserved"; break;
+  case 5: str = "CND"; break;
+  case 6: str = "CMP"; break;
+  case 7: str = "FRC"; break;
+  case 8: str = "EX2"; break;
+  case 9: str = "LN2"; break;
+  case 10: str = "RCP"; break;
+  case 11: str = "RSQ"; break;
+  case 12: str = "SIN"; break;
+  case 13: str = "COS"; break;
+  case 14: str = "MDH"; break;
+  case 15: str = "MDV"; break;
+  }
+  return str;
+}
+
+static char *to_mask(int val)
+{
+  char *str = NULL;
+  switch(val) {
+  case 0: str = "NONE"; break;
+  case 1: str = "R"; break;
+  case 2: str = "G"; break;
+  case 3: str = "RG"; break;
+  case 4: str = "B"; break;
+  case 5: str = "RB"; break;
+  case 6: str = "GB"; break;
+  case 7: str = "RGB"; break;
+  case 8: str = "A"; break;
+  case 9: str = "AR"; break;
+  case 10: str = "AG"; break;
+  case 11: str = "ARG"; break;
+  case 12: str = "AB"; break;
+  case 13: str = "ARB"; break;
+  case 14: str = "AGB"; break;
+  case 15: str = "ARGB"; break;
+  }
+  return str;
+}
+
+static char *to_texop(int val)
+{
+  switch(val) {
+  case 0: return "NOP";
+  case 1: return "LD";
+  case 2: return "TEXKILL";
+  case 3: return "PROJ";
+  case 4: return "LODBIAS";
+  case 5: return "LOD";
+  case 6: return "DXDY";
+  }
+  return NULL;
+}
+
+static void dump_program(struct r500_fragment_program_code *code)
+{
+
+  fprintf(stderr, "R500 Fragment Program:\n--------\n");
+
+  int n;
+  uint32_t inst;
+  uint32_t inst0;
+  char *str = NULL;
+
+  if (code->const_nr) {
+    fprintf(stderr, "--------\nConstants:\n");
+    for (n = 0; n < code->const_nr; n++) {
+      fprintf(stderr, "Constant %d: %i[%i]\n", n,
+        code->constant[n].File, code->constant[n].Index);
+    }
+    fprintf(stderr, "--------\n");
+  }
+
+  for (n = 0; n < code->inst_end+1; n++) {
+    inst0 = inst = code->inst[n].inst0;
+    fprintf(stderr,"%d\t0:CMN_INST   0x%08x:", n, inst);
+    switch(inst & 0x3) {
+    case R500_INST_TYPE_ALU: str = "ALU"; break;
+    case R500_INST_TYPE_OUT: str = "OUT"; break;
+    case R500_INST_TYPE_FC: str = "FC"; break;
+    case R500_INST_TYPE_TEX: str = "TEX"; break;
+    };
+    fprintf(stderr,"%s %s %s %s %s ", str,
+	    inst & R500_INST_TEX_SEM_WAIT ? "TEX_WAIT" : "",
+	    inst & R500_INST_LAST ? "LAST" : "",
+	    inst & R500_INST_NOP ? "NOP" : "",
+	    inst & R500_INST_ALU_WAIT ? "ALU WAIT" : "");
+    fprintf(stderr,"wmask: %s omask: %s\n", to_mask((inst >> 11) & 0xf),
+	    to_mask((inst >> 15) & 0xf));
+
+    switch(inst0 & 0x3) {
+    case 0:
+    case 1:
+      fprintf(stderr,"\t1:RGB_ADDR   0x%08x:", code->inst[n].inst1);
+      inst = code->inst[n].inst1;
+
+      fprintf(stderr,"Addr0: %d%c, Addr1: %d%c, Addr2: %d%c, srcp:%d\n",
+	      inst & 0xff, (inst & (1<<8)) ? 'c' : 't',
+	      (inst >> 10) & 0xff, (inst & (1<<18)) ? 'c' : 't',
+	      (inst >> 20) & 0xff, (inst & (1<<28)) ? 'c' : 't',
+	      (inst >> 30));
+
+      fprintf(stderr,"\t2:ALPHA_ADDR 0x%08x:", code->inst[n].inst2);
+      inst = code->inst[n].inst2;
+      fprintf(stderr,"Addr0: %d%c, Addr1: %d%c, Addr2: %d%c, srcp:%d\n",
+	      inst & 0xff, (inst & (1<<8)) ? 'c' : 't',
+	      (inst >> 10) & 0xff, (inst & (1<<18)) ? 'c' : 't',
+	      (inst >> 20) & 0xff, (inst & (1<<28)) ? 'c' : 't',
+	      (inst >> 30));
+      fprintf(stderr,"\t3 RGB_INST:  0x%08x:", code->inst[n].inst3);
+      inst = code->inst[n].inst3;
+      fprintf(stderr,"rgb_A_src:%d %s/%s/%s %d rgb_B_src:%d %s/%s/%s %d\n",
+	      (inst) & 0x3, toswiz((inst >> 2) & 0x7), toswiz((inst >> 5) & 0x7), toswiz((inst >> 8) & 0x7),
+	      (inst >> 11) & 0x3,
+	      (inst >> 13) & 0x3, toswiz((inst >> 15) & 0x7), toswiz((inst >> 18) & 0x7), toswiz((inst >> 21) & 0x7),
+	      (inst >> 24) & 0x3);
+
+
+      fprintf(stderr,"\t4 ALPHA_INST:0x%08x:", code->inst[n].inst4);
+      inst = code->inst[n].inst4;
+      fprintf(stderr,"%s dest:%d%s alp_A_src:%d %s %d alp_B_src:%d %s %d w:%d\n", to_alpha_op(inst & 0xf),
+	      (inst >> 4) & 0x7f, inst & (1<<11) ? "(rel)":"",
+	      (inst >> 12) & 0x3, toswiz((inst >> 14) & 0x7), (inst >> 17) & 0x3,
+	      (inst >> 19) & 0x3, toswiz((inst >> 21) & 0x7), (inst >> 24) & 0x3,
+	      (inst >> 31) & 0x1);
+
+      fprintf(stderr,"\t5 RGBA_INST: 0x%08x:", code->inst[n].inst5);
+      inst = code->inst[n].inst5;
+      fprintf(stderr,"%s dest:%d%s rgb_C_src:%d %s/%s/%s %d alp_C_src:%d %s %d\n", toop(inst & 0xf),
+	      (inst >> 4) & 0x7f, inst & (1<<11) ? "(rel)":"",
+	      (inst >> 12) & 0x3, toswiz((inst >> 14) & 0x7), toswiz((inst >> 17) & 0x7), toswiz((inst >> 20) & 0x7),
+	      (inst >> 23) & 0x3,
+	      (inst >> 25) & 0x3, toswiz((inst >> 27) & 0x7), (inst >> 30) & 0x3);
+      break;
+    case 2:
+      break;
+    case 3:
+      inst = code->inst[n].inst1;
+      fprintf(stderr,"\t1:TEX_INST:  0x%08x: id: %d op:%s, %s, %s %s\n", inst, (inst >> 16) & 0xf,
+	      to_texop((inst >> 22) & 0x7), (inst & (1<<25)) ? "ACQ" : "",
+	      (inst & (1<<26)) ? "IGNUNC" : "", (inst & (1<<27)) ? "UNSCALED" : "SCALED");
+      inst = code->inst[n].inst2;
+      fprintf(stderr,"\t2:TEX_ADDR:  0x%08x: src: %d%s %s/%s/%s/%s dst: %d%s %s/%s/%s/%s\n", inst,
+	      inst & 127, inst & (1<<7) ? "(rel)" : "",
+	      toswiz((inst >> 8) & 0x3), toswiz((inst >> 10) & 0x3),
+	      toswiz((inst >> 12) & 0x3), toswiz((inst >> 14) & 0x3),
+	      (inst >> 16) & 127, inst & (1<<23) ? "(rel)" : "",
+	      toswiz((inst >> 24) & 0x3), toswiz((inst >> 26) & 0x3),
+	      toswiz((inst >> 28) & 0x3), toswiz((inst >> 30) & 0x3));
+
+      fprintf(stderr,"\t3:TEX_DXDY:  0x%08x\n", code->inst[n].inst3);
+      break;
+    }
+    fprintf(stderr,"\n");
+  }
+
+}
diff --git a/src/mesa/drivers/dri/r300/r500_fragprog.h b/src/mesa/drivers/dri/r300/r500_fragprog.h
new file mode 100644
index 00000000000..1e45538f807
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r500_fragprog.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (C) 2005 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/*
+ * Authors:
+ *   Ben Skeggs <darktama@iinet.net.au>
+ *   Jerome Glisse <j.glisse@gmail.com>
+ */
+#ifndef __R500_FRAGPROG_H_
+#define __R500_FRAGPROG_H_
+
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+
+#include "r300_context.h"
+#include "r300_state.h"
+#include "radeon_program.h"
+
+struct r500_fragment_program;
+
+extern void r500TranslateFragmentShader(r300ContextPtr r300,
+					struct r500_fragment_program *fp);
+
+struct r500_fragment_program_compiler {
+	r300ContextPtr r300;
+	struct r500_fragment_program *fp;
+	struct r500_fragment_program_code *code;
+	struct gl_program *program;
+};
+
+extern GLboolean r500FragmentProgramEmit(struct r500_fragment_program_compiler *compiler);
+
+#endif
diff --git a/src/mesa/drivers/dri/r300/r500_fragprog_emit.c b/src/mesa/drivers/dri/r300/r500_fragprog_emit.c
new file mode 100644
index 00000000000..4631235f0d3
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r500_fragprog_emit.c
@@ -0,0 +1,327 @@
+/*
+ * Copyright (C) 2005 Ben Skeggs.
+ *
+ * Copyright 2008 Corbin Simpson <MostAwesomeDude@gmail.com>
+ * Adaptation and modification for ATI/AMD Radeon R500 GPU chipsets.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * \file
+ *
+ * \author Ben Skeggs <darktama@iinet.net.au>
+ *
+ * \author Jerome Glisse <j.glisse@gmail.com>
+ *
+ * \author Corbin Simpson <MostAwesomeDude@gmail.com>
+ *
+ * \todo Depth write, WPOS/FOGC inputs
+ *
+ * \todo FogOption
+ *
+ */
+
+#include "r500_fragprog.h"
+
+#include "radeon_program_pair.h"
+
+
+#define PROG_CODE \
+	struct r500_fragment_program_compiler *c = (struct r500_fragment_program_compiler*)data; \
+	struct r500_fragment_program_code *code = c->code
+
+#define error(fmt, args...) do {			\
+		fprintf(stderr, "%s::%s(): " fmt "\n",	\
+			__FILE__, __FUNCTION__, ##args);	\
+	} while(0)
+
+
+/**
+ * Callback to register hardware constants.
+ */
+static GLboolean emit_const(void *data, GLuint file, GLuint idx, GLuint *hwindex)
+{
+	PROG_CODE;
+
+	for (*hwindex = 0; *hwindex < code->const_nr; ++*hwindex) {
+		if (code->constant[*hwindex].File == file &&
+		    code->constant[*hwindex].Index == idx)
+			break;
+	}
+
+	if (*hwindex >= code->const_nr) {
+		if (*hwindex >= PFS_NUM_CONST_REGS) {
+			error("Out of hw constants!\n");
+			return GL_FALSE;
+		}
+
+		code->const_nr++;
+		code->constant[*hwindex].File = file;
+		code->constant[*hwindex].Index = idx;
+	}
+
+	return GL_TRUE;
+}
+
+static GLuint translate_rgb_op(GLuint opcode)
+{
+	switch(opcode) {
+	case OPCODE_CMP: return R500_ALU_RGBA_OP_CMP;
+	case OPCODE_DDX: return R500_ALU_RGBA_OP_MDH;
+	case OPCODE_DDY: return R500_ALU_RGBA_OP_MDV;
+	case OPCODE_DP3: return R500_ALU_RGBA_OP_DP3;
+	case OPCODE_DP4: return R500_ALU_RGBA_OP_DP4;
+	case OPCODE_FRC: return R500_ALU_RGBA_OP_FRC;
+	default:
+		error("translate_rgb_op(%d): unknown opcode\n", opcode);
+		/* fall through */
+	case OPCODE_NOP:
+		/* fall through */
+	case OPCODE_MAD: return R500_ALU_RGBA_OP_MAD;
+	case OPCODE_MAX: return R500_ALU_RGBA_OP_MAX;
+	case OPCODE_MIN: return R500_ALU_RGBA_OP_MIN;
+	case OPCODE_REPL_ALPHA: return R500_ALU_RGBA_OP_SOP;
+	}
+}
+
+static GLuint translate_alpha_op(GLuint opcode)
+{
+	switch(opcode) {
+	case OPCODE_CMP: return R500_ALPHA_OP_CMP;
+	case OPCODE_COS: return R500_ALPHA_OP_COS;
+	case OPCODE_DDX: return R500_ALPHA_OP_MDH;
+	case OPCODE_DDY: return R500_ALPHA_OP_MDV;
+	case OPCODE_DP3: return R500_ALPHA_OP_DP;
+	case OPCODE_DP4: return R500_ALPHA_OP_DP;
+	case OPCODE_EX2: return R500_ALPHA_OP_EX2;
+	case OPCODE_FRC: return R500_ALPHA_OP_FRC;
+	case OPCODE_LG2: return R500_ALPHA_OP_LN2;
+	default:
+		error("translate_alpha_op(%d): unknown opcode\n", opcode);
+		/* fall through */
+	case OPCODE_NOP:
+		/* fall through */
+	case OPCODE_MAD: return R500_ALPHA_OP_MAD;
+	case OPCODE_MAX: return R500_ALPHA_OP_MAX;
+	case OPCODE_MIN: return R500_ALPHA_OP_MIN;
+	case OPCODE_RCP: return R500_ALPHA_OP_RCP;
+	case OPCODE_RSQ: return R500_ALPHA_OP_RSQ;
+	case OPCODE_SIN: return R500_ALPHA_OP_SIN;
+	}
+}
+
+static GLuint fix_hw_swizzle(GLuint swz)
+{
+	if (swz == 5) swz = 6;
+	if (swz == SWIZZLE_NIL) swz = 4;
+	return swz;
+}
+
+static GLuint translate_arg_rgb(struct radeon_pair_instruction *inst, int arg)
+{
+	GLuint t = inst->RGB.Arg[arg].Source;
+	int comp;
+	t |= inst->RGB.Arg[arg].Negate << 11;
+	t |= inst->RGB.Arg[arg].Abs << 12;
+
+	for(comp = 0; comp < 3; ++comp)
+		t |= fix_hw_swizzle(GET_SWZ(inst->RGB.Arg[arg].Swizzle, comp)) << (3*comp + 2);
+
+	return t;
+}
+
+static GLuint translate_arg_alpha(struct radeon_pair_instruction *inst, int i)
+{
+	GLuint t = inst->Alpha.Arg[i].Source;
+	t |= fix_hw_swizzle(inst->Alpha.Arg[i].Swizzle) << 2;
+	t |= inst->Alpha.Arg[i].Negate << 5;
+	t |= inst->Alpha.Arg[i].Abs << 6;
+	return t;
+}
+
+static void use_temporary(struct r500_fragment_program_code* code, GLuint index)
+{
+	if (index > code->max_temp_idx)
+		code->max_temp_idx = index;
+}
+
+static GLuint use_source(struct r500_fragment_program_code* code, struct radeon_pair_instruction_source src)
+{
+	if (!src.Constant)
+		use_temporary(code, src.Index);
+	return src.Index | src.Constant << 8;
+}
+
+
+/**
+ * Emit a paired ALU instruction.
+ */
+static GLboolean emit_paired(void *data, struct radeon_pair_instruction *inst)
+{
+	PROG_CODE;
+
+	if (code->inst_end >= 511) {
+		error("emit_alu: Too many instructions");
+		return GL_FALSE;
+	}
+
+	int ip = ++code->inst_end;
+
+	code->inst[ip].inst5 = translate_rgb_op(inst->RGB.Opcode);
+	code->inst[ip].inst4 = translate_alpha_op(inst->Alpha.Opcode);
+
+	if (inst->RGB.OutputWriteMask || inst->Alpha.OutputWriteMask || inst->Alpha.DepthWriteMask)
+		code->inst[ip].inst0 = R500_INST_TYPE_OUT;
+	else
+		code->inst[ip].inst0 = R500_INST_TYPE_ALU;
+	code->inst[ip].inst0 |= R500_INST_TEX_SEM_WAIT;
+
+	code->inst[ip].inst0 |= (inst->RGB.WriteMask << 11) | (inst->Alpha.WriteMask << 14);
+	code->inst[ip].inst0 |= (inst->RGB.OutputWriteMask << 15) | (inst->Alpha.OutputWriteMask << 18);
+	if (inst->Alpha.DepthWriteMask) {
+		code->inst[ip].inst4 |= R500_ALPHA_W_OMASK;
+		c->fp->writes_depth = GL_TRUE;
+	}
+
+	code->inst[ip].inst4 |= R500_ALPHA_ADDRD(inst->Alpha.DestIndex);
+	code->inst[ip].inst5 |= R500_ALU_RGBA_ADDRD(inst->RGB.DestIndex);
+	use_temporary(code, inst->Alpha.DestIndex);
+	use_temporary(code, inst->RGB.DestIndex);
+
+	if (inst->RGB.Saturate)
+		code->inst[ip].inst0 |= R500_INST_RGB_CLAMP;
+	if (inst->Alpha.Saturate)
+		code->inst[ip].inst0 |= R500_INST_ALPHA_CLAMP;
+
+	code->inst[ip].inst1 |= R500_RGB_ADDR0(use_source(code, inst->RGB.Src[0]));
+	code->inst[ip].inst1 |= R500_RGB_ADDR1(use_source(code, inst->RGB.Src[1]));
+	code->inst[ip].inst1 |= R500_RGB_ADDR2(use_source(code, inst->RGB.Src[2]));
+
+	code->inst[ip].inst2 |= R500_ALPHA_ADDR0(use_source(code, inst->Alpha.Src[0]));
+	code->inst[ip].inst2 |= R500_ALPHA_ADDR1(use_source(code, inst->Alpha.Src[1]));
+	code->inst[ip].inst2 |= R500_ALPHA_ADDR2(use_source(code, inst->Alpha.Src[2]));
+
+	code->inst[ip].inst3 |= translate_arg_rgb(inst, 0) << R500_ALU_RGB_SEL_A_SHIFT;
+	code->inst[ip].inst3 |= translate_arg_rgb(inst, 1) << R500_ALU_RGB_SEL_B_SHIFT;
+	code->inst[ip].inst5 |= translate_arg_rgb(inst, 2) << R500_ALU_RGBA_SEL_C_SHIFT;
+
+	code->inst[ip].inst4 |= translate_arg_alpha(inst, 0) << R500_ALPHA_SEL_A_SHIFT;
+	code->inst[ip].inst4 |= translate_arg_alpha(inst, 1) << R500_ALPHA_SEL_B_SHIFT;
+	code->inst[ip].inst5 |= translate_arg_alpha(inst, 2) << R500_ALU_RGBA_ALPHA_SEL_C_SHIFT;
+
+	return GL_TRUE;
+}
+
+static GLuint translate_strq_swizzle(struct prog_src_register src)
+{
+	GLuint swiz = 0;
+	int i;
+	for (i = 0; i < 4; i++)
+		swiz |= (GET_SWZ(src.Swizzle, i) & 0x3) << i*2;
+	return swiz;
+}
+
+/**
+ * Emit a single TEX instruction
+ */
+static GLboolean emit_tex(void *data, struct prog_instruction *inst)
+{
+	PROG_CODE;
+
+	if (code->inst_end >= 511) {
+		error("emit_tex: Too many instructions");
+		return GL_FALSE;
+	}
+
+	int ip = ++code->inst_end;
+
+	code->inst[ip].inst0 = R500_INST_TYPE_TEX
+		| (inst->DstReg.WriteMask << 11)
+		| R500_INST_TEX_SEM_WAIT;
+	code->inst[ip].inst1 = R500_TEX_ID(inst->TexSrcUnit)
+		| R500_TEX_SEM_ACQUIRE | R500_TEX_IGNORE_UNCOVERED;
+
+	if (inst->TexSrcTarget == TEXTURE_RECT_INDEX)
+	        code->inst[ip].inst1 |= R500_TEX_UNSCALED;
+
+	switch (inst->Opcode) {
+	case OPCODE_KIL:
+		code->inst[ip].inst1 |= R500_TEX_INST_TEXKILL;
+		break;
+	case OPCODE_TEX:
+		code->inst[ip].inst1 |= R500_TEX_INST_LD;
+		break;
+	case OPCODE_TXB:
+		code->inst[ip].inst1 |= R500_TEX_INST_LODBIAS;
+		break;
+	case OPCODE_TXP:
+		code->inst[ip].inst1 |= R500_TEX_INST_PROJ;
+		break;
+	default:
+		error("emit_tex can't handle opcode %x\n", inst->Opcode);
+	}
+
+	code->inst[ip].inst2 = R500_TEX_SRC_ADDR(inst->SrcReg[0].Index)
+		| (translate_strq_swizzle(inst->SrcReg[0]) << 8)
+		| R500_TEX_DST_ADDR(inst->DstReg.Index)
+		| R500_TEX_DST_R_SWIZ_R | R500_TEX_DST_G_SWIZ_G
+		| R500_TEX_DST_B_SWIZ_B | R500_TEX_DST_A_SWIZ_A;
+
+	return GL_TRUE;
+}
+
+static const struct radeon_pair_handler pair_handler = {
+	.EmitConst = emit_const,
+	.EmitPaired = emit_paired,
+	.EmitTex = emit_tex,
+	.MaxHwTemps = 128
+};
+
+GLboolean r500FragmentProgramEmit(struct r500_fragment_program_compiler *compiler)
+{
+	struct r500_fragment_program_code *code = compiler->code;
+
+	_mesa_bzero(code, sizeof(*code));
+	code->max_temp_idx = 1;
+	code->inst_offset = 0;
+	code->inst_end = -1;
+
+	if (!radeonPairProgram(compiler->r300->radeon.glCtx, compiler->program, &pair_handler, compiler))
+		return GL_FALSE;
+
+	if ((code->inst[code->inst_end].inst0 & R500_INST_TYPE_MASK) != R500_INST_TYPE_OUT) {
+		/* This may happen when dead-code elimination is disabled or
+		 * when most of the fragment program logic is leading to a KIL */
+		if (code->inst_end >= 511) {
+			error("Introducing fake OUT: Too many instructions");
+			return GL_FALSE;
+		}
+
+		int ip = ++code->inst_end;
+		code->inst[ip].inst0 = R500_INST_TYPE_OUT | R500_INST_TEX_SEM_WAIT;
+	}
+
+	return GL_TRUE;
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_context.c b/src/mesa/drivers/dri/r300/radeon_context.c
index e9634b427a6..5267fe9a774 100644
--- a/src/mesa/drivers/dri/r300/radeon_context.c
+++ b/src/mesa/drivers/dri/r300/radeon_context.c
@@ -36,12 +36,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include <dlfcn.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "state.h"
-#include "matrix.h"
-#include "framebuffer.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/state.h"
+#include "main/matrix.h"
+#include "main/framebuffer.h"
 
 #include "drivers/common/driverfuncs.h"
 #include "swrast/swrast.h"
@@ -135,6 +135,7 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 	/* Fill in additional standard functions. */
 	radeonInitDriverFuncs(functions);
 
+	radeon->radeonScreen = screen;
 	/* Allocate and initialize the Mesa context */
 	if (sharedContextPrivate)
 		shareCtx = ((radeonContextPtr)sharedContextPrivate)->glCtx;
@@ -156,9 +157,8 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 	radeon->dri.hwContext = driContextPriv->hHWContext;
 	radeon->dri.hwLock = &sPriv->pSAREA->lock;
 	radeon->dri.fd = sPriv->fd;
-	radeon->dri.drmMinor = sPriv->drmMinor;
+	radeon->dri.drmMinor = sPriv->drm_version.minor;
 
-	radeon->radeonScreen = screen;
 	radeon->sarea = (drm_radeon_sarea_t *) ((GLubyte *) sPriv->pSAREA +
 					       screen->sarea_priv_offset);
 
@@ -177,10 +177,7 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 			radeon->do_usleeps ? "usleeps" : "busy waits",
 			fthrottle_mode, radeon->radeonScreen->irq);
 
-	radeon->vblank_flags = (radeon->radeonScreen->irq != 0)
-	    ? driGetDefaultVBlankFlags(&radeon->optionCache) : VBLANK_FLAG_NO_IRQ;
-
-	(*dri_interface->getUST) (&radeon->swap_ust);
+	(*sPriv->systemTime->getUST) (&radeon->swap_ust);
 
 	return GL_TRUE;
 }
@@ -277,9 +274,15 @@ GLboolean radeonMakeCurrent(__DRIcontextPrivate * driContextPriv,
 				radeon->glCtx);
 
 		if (radeon->dri.drawable != driDrawPriv) {
-			driDrawableInitVBlank(driDrawPriv,
-					      radeon->vblank_flags,
-					      &radeon->vbl_seq);
+			if (driDrawPriv->swap_interval == (unsigned)-1) {
+				driDrawPriv->vblFlags =
+					(radeon->radeonScreen->irq != 0)
+					? driGetDefaultVBlankFlags(&radeon->
+								   optionCache)
+					: VBLANK_FLAG_NO_IRQ;
+
+				driDrawableInitVBlank(driDrawPriv);
+			}
 		}
 
 		radeon->dri.readable = driReadPriv;
diff --git a/src/mesa/drivers/dri/r300/radeon_context.h b/src/mesa/drivers/dri/r300/radeon_context.h
index 2f239417a97..47cbc22a725 100644
--- a/src/mesa/drivers/dri/r300/radeon_context.h
+++ b/src/mesa/drivers/dri/r300/radeon_context.h
@@ -43,26 +43,16 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __RADEON_CONTEXT_H__
 #define __RADEON_CONTEXT_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
 #include "radeon_screen.h"
 #include "drm.h"
 #include "dri_util.h"
-#include "colormac.h"
 
 struct radeon_context;
 typedef struct radeon_context radeonContextRec;
 typedef struct radeon_context *radeonContextPtr;
 
-#define TEX_0   0x1
-#define TEX_1   0x2
-#define TEX_2	0x4
-#define TEX_3	0x8
-#define TEX_4	0x10
-#define TEX_5	0x20
-#define TEX_6	0x40
-#define TEX_7	0x80
-#define TEX_ALL 0xff
-
 /* Rasterizing fallbacks */
 /* See correponding strings in r200_swtcl.c */
 #define RADEON_FALLBACK_TEXTURE		0x0001
@@ -182,10 +172,7 @@ struct radeon_context {
 	GLuint irqsEmitted;
 	drm_radeon_irq_wait_t iw;
 
-	/* VBI / buffer swap */
-	GLuint vbl_seq;
-	GLuint vblank_flags;
-
+	/* buffer swap */
 	int64_t swap_ust;
 	int64_t swap_missed_ust;
 
diff --git a/src/mesa/drivers/dri/r300/radeon_ioctl.c b/src/mesa/drivers/dri/r300/radeon_ioctl.c
index 0b8656b9c1d..36502eb42de 100644
--- a/src/mesa/drivers/dri/r300/radeon_ioctl.c
+++ b/src/mesa/drivers/dri/r300/radeon_ioctl.c
@@ -35,10 +35,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <sched.h>
 #include <errno.h>
 
-#include "glheader.h"
-#include "imports.h"
-#include "macros.h"
-#include "context.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
 #include "r300_context.h"
 #include "radeon_ioctl.h"
@@ -157,13 +157,14 @@ static void radeonWaitForFrameCompletion(radeonContextPtr radeon)
 
 /* Copy the back color buffer to the front color buffer.
  */
-void radeonCopyBuffer(const __DRIdrawablePrivate * dPriv,
+void radeonCopyBuffer(__DRIdrawablePrivate * dPriv,
 		      const drm_clip_rect_t	 * rect)
 {
 	radeonContextPtr radeon;
 	GLint nbox, i, ret;
 	GLboolean missed_target;
 	int64_t ust;
+	__DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
 	assert(dPriv);
 	assert(dPriv->driContextPriv);
@@ -187,8 +188,7 @@ void radeonCopyBuffer(const __DRIdrawablePrivate * dPriv,
 	if (!rect)
 	{
 	    UNLOCK_HARDWARE(radeon);
-	    driWaitForVBlank(dPriv, &radeon->vbl_seq, radeon->vblank_flags,
-			     &missed_target);
+	    driWaitForVBlank(dPriv, &missed_target);
 	    LOCK_HARDWARE(radeon);
 	}
 
@@ -215,16 +215,18 @@ void radeonCopyBuffer(const __DRIdrawablePrivate * dPriv,
 			if (rect->y2 < b->y2)
 			    b->y2 = rect->y2;
 
-			if (b->x1 < b->x2 && b->y1 < b->y2)
-			    b++;
+			if (b->x1 >= b->x2 || b->y1 >= b->y2)
+			    continue;
 		    }
-		    else
-			b++;
 
+		    b++;
 		    n++;
 		}
 		radeon->sarea->nbox = n;
 
+		if (!n)
+		   continue;
+
 		ret = drmCommandNone(radeon->dri.fd, DRM_RADEON_SWAP);
 
 		if (ret) {
@@ -241,7 +243,7 @@ void radeonCopyBuffer(const __DRIdrawablePrivate * dPriv,
 	    ((r300ContextPtr)radeon)->hw.all_dirty = GL_TRUE;
 
 	    radeon->swap_count++;
-	    (*dri_interface->getUST) (&ust);
+	    (*psp->systemTime->getUST) (&ust);
 	    if (missed_target) {
 		radeon->swap_missed_count++;
 		radeon->swap_missed_ust = ust - radeon->swap_ust;
@@ -253,11 +255,12 @@ void radeonCopyBuffer(const __DRIdrawablePrivate * dPriv,
 	}
 }
 
-void radeonPageFlip(const __DRIdrawablePrivate * dPriv)
+void radeonPageFlip(__DRIdrawablePrivate * dPriv)
 {
 	radeonContextPtr radeon;
 	GLint ret;
 	GLboolean missed_target;
+	__DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
 	assert(dPriv);
 	assert(dPriv->driContextPriv);
@@ -293,11 +296,10 @@ void radeonPageFlip(const __DRIdrawablePrivate * dPriv)
 	 */
 	radeonWaitForFrameCompletion(radeon);
 	UNLOCK_HARDWARE(radeon);
-	driWaitForVBlank(dPriv, &radeon->vbl_seq, radeon->vblank_flags,
-			 &missed_target);
+	driWaitForVBlank(dPriv, &missed_target);
 	if (missed_target) {
 		radeon->swap_missed_count++;
-		(void)(*dri_interface->getUST) (&radeon->swap_missed_ust);
+		(void)(*psp->systemTime->getUST) (&radeon->swap_missed_ust);
 	}
 	LOCK_HARDWARE(radeon);
 
@@ -311,7 +313,7 @@ void radeonPageFlip(const __DRIdrawablePrivate * dPriv)
 	}
 
 	radeon->swap_count++;
-	(void)(*dri_interface->getUST) (&radeon->swap_ust);
+	(void)(*psp->systemTime->getUST) (&radeon->swap_ust);
 
         driFlipRenderbuffers(radeon->glCtx->WinSysDrawBuffer, 
                              radeon->sarea->pfCurrentPage);
diff --git a/src/mesa/drivers/dri/r300/radeon_ioctl.h b/src/mesa/drivers/dri/r300/radeon_ioctl.h
index 3a80d36c622..3add775b822 100644
--- a/src/mesa/drivers/dri/r300/radeon_ioctl.h
+++ b/src/mesa/drivers/dri/r300/radeon_ioctl.h
@@ -35,20 +35,20 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __RADEON_IOCTL_H__
 #define __RADEON_IOCTL_H__
 
-#include "simple_list.h"
+#include "main/simple_list.h"
 #include "radeon_dri.h"
 #include "radeon_lock.h"
 
 #include "xf86drm.h"
 #include "drm.h"
 #if 0
-#include "r200_context.h"
+#include "r200context.h"
 #endif
 #include "radeon_drm.h"
 
-extern void radeonCopyBuffer(const __DRIdrawablePrivate * drawable,
+extern void radeonCopyBuffer(__DRIdrawablePrivate * drawable,
 			     const drm_clip_rect_t	* rect);
-extern void radeonPageFlip(const __DRIdrawablePrivate * drawable);
+extern void radeonPageFlip(__DRIdrawablePrivate * drawable);
 extern void radeonFlush(GLcontext * ctx);
 extern void radeonFinish(GLcontext * ctx);
 extern void radeonWaitForIdleLocked(radeonContextPtr radeon);
diff --git a/src/mesa/drivers/dri/r300/radeon_lock.c b/src/mesa/drivers/dri/r300/radeon_lock.c
index bc3c2d6c6b3..4f47afd5dc6 100644
--- a/src/mesa/drivers/dri/r300/radeon_lock.c
+++ b/src/mesa/drivers/dri/r300/radeon_lock.c
@@ -45,7 +45,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r300_context.h"
 #include "r300_state.h"
 
-#include "framebuffer.h"
+#include "main/framebuffer.h"
 
 #include "drirenderbuffer.h"
 
@@ -68,8 +68,8 @@ void radeonUpdatePageFlipping(radeonContextPtr rmesa)
 	}
 
 	use_back = rmesa->glCtx->DrawBuffer ?
-	    (rmesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0] ==
-	     BUFFER_BIT_BACK_LEFT) : 1;
+	    (rmesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0] ==
+	     BUFFER_BACK_LEFT) : 1;
 	use_back ^= (rmesa->sarea->pfCurrentPage == 1);
 
 	if (use_back) {
diff --git a/src/mesa/drivers/dri/r300/radeon_nqssadce.c b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
new file mode 100644
index 00000000000..97ce016c99c
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
@@ -0,0 +1,284 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * @file
+ *
+ * "Not-quite SSA" and Dead-Code Elimination.
+ *
+ * @note This code uses SWIZZLE_NIL in a source register to indicate that
+ * the corresponding component is ignored by the corresponding instruction.
+ */
+
+#include "radeon_nqssadce.h"
+
+
+/**
+ * Return the @ref register_state for the given register (or 0 for untracked
+ * registers, i.e. constants).
+ */
+static struct register_state *get_reg_state(struct nqssadce_state* s, GLuint file, GLuint index)
+{
+	switch(file) {
+	case PROGRAM_TEMPORARY: return &s->Temps[index];
+	case PROGRAM_OUTPUT: return &s->Outputs[index];
+	default: return 0;
+	}
+}
+
+
+/**
+ * Left multiplication of a register with a swizzle
+ *
+ * @note Works correctly only for X, Y, Z, W swizzles, not for constant swizzles.
+ */
+static struct prog_src_register lmul_swizzle(GLuint swizzle, struct prog_src_register srcreg)
+{
+	struct prog_src_register tmp = srcreg;
+	int i;
+	tmp.Swizzle = 0;
+	tmp.NegateBase = 0;
+	for(i = 0; i < 4; ++i) {
+		GLuint swz = GET_SWZ(swizzle, i);
+		if (swz < 4) {
+			tmp.Swizzle |= GET_SWZ(srcreg.Swizzle, swz) << (i*3);
+			tmp.NegateBase |= GET_BIT(srcreg.NegateBase, swz) << i;
+		} else {
+			tmp.Swizzle |= swz << (i*3);
+		}
+	}
+	return tmp;
+}
+
+
+static struct prog_instruction* track_used_srcreg(struct nqssadce_state* s,
+	struct prog_instruction *inst, GLint src, GLuint sourced)
+{
+	int i;
+	GLuint deswz_source = 0;
+
+	for(i = 0; i < 4; ++i) {
+		if (GET_BIT(sourced, i)) {
+			GLuint swz = GET_SWZ(inst->SrcReg[src].Swizzle, i);
+			deswz_source |= 1 << swz;
+		} else {
+			inst->SrcReg[src].Swizzle &= ~(7 << (3*i));
+			inst->SrcReg[src].Swizzle |= SWIZZLE_NIL << (3*i);
+		}
+	}
+
+	if (!s->Descr->IsNativeSwizzle(inst->Opcode, inst->SrcReg[src])) {
+		struct prog_dst_register dstreg = inst->DstReg;
+		dstreg.File = PROGRAM_TEMPORARY;
+		dstreg.Index = _mesa_find_free_register(s->Program, PROGRAM_TEMPORARY);
+		dstreg.WriteMask = sourced;
+
+		s->Descr->BuildSwizzle(s, dstreg, inst->SrcReg[src]);
+
+		inst = s->Program->Instructions + s->IP;
+		inst->SrcReg[src].File = PROGRAM_TEMPORARY;
+		inst->SrcReg[src].Index = dstreg.Index;
+		inst->SrcReg[src].Swizzle = 0;
+		inst->SrcReg[src].NegateBase = 0;
+		inst->SrcReg[src].Abs = 0;
+		inst->SrcReg[src].NegateAbs = 0;
+		for(i = 0; i < 4; ++i) {
+			if (GET_BIT(sourced, i))
+				inst->SrcReg[src].Swizzle |= i << (3*i);
+			else
+				inst->SrcReg[src].Swizzle |= SWIZZLE_NIL << (3*i);
+		}
+		deswz_source = sourced;
+	}
+
+	struct register_state *regstate = get_reg_state(s, inst->SrcReg[src].File, inst->SrcReg[src].Index);
+	if (regstate)
+		regstate->Sourced |= deswz_source & 0xf;
+
+	return inst;
+}
+
+
+static void rewrite_depth_out(struct prog_instruction *inst)
+{
+	if (inst->DstReg.WriteMask & WRITEMASK_Z) {
+		inst->DstReg.WriteMask = WRITEMASK_W;
+	} else {
+		inst->DstReg.WriteMask = 0;
+		return;
+	}
+
+	switch (inst->Opcode) {
+	case OPCODE_FRC:
+	case OPCODE_MOV:
+		inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
+		break;
+	case OPCODE_ADD:
+	case OPCODE_MAX:
+	case OPCODE_MIN:
+	case OPCODE_MUL:
+		inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
+		inst->SrcReg[1] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[1]);
+		break;
+	case OPCODE_CMP:
+	case OPCODE_MAD:
+		inst->SrcReg[0] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[0]);
+		inst->SrcReg[1] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[1]);
+		inst->SrcReg[2] = lmul_swizzle(SWIZZLE_ZZZZ, inst->SrcReg[2]);
+		break;
+	default:
+		// Scalar instructions needn't be reswizzled
+		break;
+	}
+}
+
+static void unalias_srcregs(struct prog_instruction *inst, GLuint oldindex, GLuint newindex)
+{
+	int nsrc = _mesa_num_inst_src_regs(inst->Opcode);
+	int i;
+	for(i = 0; i < nsrc; ++i)
+		if (inst->SrcReg[i].File == PROGRAM_TEMPORARY && inst->SrcReg[i].Index == oldindex)
+			inst->SrcReg[i].Index = newindex;
+}
+
+static void unalias_temporary(struct nqssadce_state* s, GLuint oldindex)
+{
+	GLuint newindex = _mesa_find_free_register(s->Program, PROGRAM_TEMPORARY);
+	int ip;
+	for(ip = 0; ip < s->IP; ++ip) {
+		struct prog_instruction* inst = s->Program->Instructions + ip;
+		if (inst->DstReg.File == PROGRAM_TEMPORARY && inst->DstReg.Index == oldindex)
+			inst->DstReg.Index = newindex;
+		unalias_srcregs(inst, oldindex, newindex);
+	}
+	unalias_srcregs(s->Program->Instructions + s->IP, oldindex, newindex);
+}
+
+
+/**
+ * Handle one instruction.
+ */
+static void process_instruction(struct nqssadce_state* s)
+{
+	struct prog_instruction *inst = s->Program->Instructions + s->IP;
+
+	if (inst->Opcode == OPCODE_END)
+		return;
+
+	if (inst->Opcode != OPCODE_KIL) {
+		if (s->Descr->RewriteDepthOut) {
+			if (inst->DstReg.File == PROGRAM_OUTPUT && inst->DstReg.Index == FRAG_RESULT_DEPR)
+				rewrite_depth_out(inst);
+		}
+
+		struct register_state *regstate = get_reg_state(s, inst->DstReg.File, inst->DstReg.Index);
+		if (!regstate) {
+			_mesa_problem(s->Ctx, "NqssaDce: bad destination register (%i[%i])\n",
+				inst->DstReg.File, inst->DstReg.Index);
+			return;
+		}
+
+		inst->DstReg.WriteMask &= regstate->Sourced;
+		regstate->Sourced &= ~inst->DstReg.WriteMask;
+
+		if (inst->DstReg.WriteMask == 0) {
+			_mesa_delete_instructions(s->Program, s->IP, 1);
+			return;
+		}
+
+		if (inst->DstReg.File == PROGRAM_TEMPORARY && !regstate->Sourced)
+			unalias_temporary(s, inst->DstReg.Index);
+	}
+
+	/* Attention: Due to swizzle emulation code, the following
+	 * might change the instruction stream under us, so we have
+	 * to be careful with the inst pointer. */
+	switch (inst->Opcode) {
+	case OPCODE_DDX:
+	case OPCODE_DDY:
+	case OPCODE_FRC:
+	case OPCODE_MOV:
+		inst = track_used_srcreg(s, inst, 0, inst->DstReg.WriteMask);
+		break;
+	case OPCODE_ADD:
+	case OPCODE_MAX:
+	case OPCODE_MIN:
+	case OPCODE_MUL:
+		inst = track_used_srcreg(s, inst, 0, inst->DstReg.WriteMask);
+		inst = track_used_srcreg(s, inst, 1, inst->DstReg.WriteMask);
+		break;
+	case OPCODE_CMP:
+	case OPCODE_MAD:
+		inst = track_used_srcreg(s, inst, 0, inst->DstReg.WriteMask);
+		inst = track_used_srcreg(s, inst, 1, inst->DstReg.WriteMask);
+		inst = track_used_srcreg(s, inst, 2, inst->DstReg.WriteMask);
+		break;
+	case OPCODE_COS:
+	case OPCODE_EX2:
+	case OPCODE_LG2:
+	case OPCODE_RCP:
+	case OPCODE_RSQ:
+	case OPCODE_SIN:
+		inst = track_used_srcreg(s, inst, 0, 0x1);
+		break;
+	case OPCODE_DP3:
+		inst = track_used_srcreg(s, inst, 0, 0x7);
+		inst = track_used_srcreg(s, inst, 1, 0x7);
+		break;
+	case OPCODE_DP4:
+		inst = track_used_srcreg(s, inst, 0, 0xf);
+		inst = track_used_srcreg(s, inst, 1, 0xf);
+		break;
+	case OPCODE_KIL:
+	case OPCODE_TEX:
+	case OPCODE_TXB:
+	case OPCODE_TXP:
+		inst = track_used_srcreg(s, inst, 0, 0xf);
+		break;
+	default:
+		_mesa_problem(s->Ctx, "NqssaDce: Unknown opcode %d\n", inst->Opcode);
+		return;
+	}
+}
+
+
+void radeonNqssaDce(GLcontext *ctx, struct gl_program *p, struct radeon_nqssadce_descr* descr)
+{
+	struct nqssadce_state s;
+
+	_mesa_bzero(&s, sizeof(s));
+	s.Ctx = ctx;
+	s.Program = p;
+	s.Descr = descr;
+	s.Descr->Init(&s);
+	s.IP = p->NumInstructions;
+
+	while(s.IP > 0) {
+		s.IP--;
+		process_instruction(&s);
+	}
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_nqssadce.h b/src/mesa/drivers/dri/r300/radeon_nqssadce.h
new file mode 100644
index 00000000000..a4f94abcb62
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_nqssadce.h
@@ -0,0 +1,96 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __RADEON_PROGRAM_NQSSADCE_H_
+#define __RADEON_PROGRAM_NQSSADCE_H_
+
+#include "radeon_program.h"
+
+
+struct register_state {
+	/**
+	 * Bitmask indicating which components of the register are sourced
+	 * by later instructions.
+	 */
+	GLuint Sourced : 4;
+};
+
+/**
+ * Maintain state such as which registers are used, which registers are
+ * read from, etc.
+ */
+struct nqssadce_state {
+	GLcontext *Ctx;
+	struct gl_program *Program;
+	struct radeon_nqssadce_descr *Descr;
+
+	/**
+	 * All instructions after this instruction pointer have been dealt with.
+	 */
+	int IP;
+
+	/**
+	 * Which registers are read by subsequent instructions?
+	 */
+	struct register_state Temps[MAX_PROGRAM_TEMPS];
+	struct register_state Outputs[VERT_RESULT_MAX];
+};
+
+
+/**
+ * This structure contains a description of the hardware in-so-far as
+ * it is required for the NqSSA-DCE pass.
+ */
+struct radeon_nqssadce_descr {
+	/**
+	 * Fill in which outputs
+	 */
+	void (*Init)(struct nqssadce_state *);
+
+	/**
+	 * Check whether the given swizzle, absolute and negate combination
+	 * can be implemented natively by the hardware for this opcode.
+	 */
+	GLboolean (*IsNativeSwizzle)(GLuint opcode, struct prog_src_register reg);
+
+	/**
+	 * Emit (at the current IP) the instruction MOV dst, src;
+	 * The transformation will work recursively on the emitted instruction(s).
+	 */
+	void (*BuildSwizzle)(struct nqssadce_state*, struct prog_dst_register dst, struct prog_src_register src);
+
+	/**
+	 * Rewrite instructions that write to DEPR.z to write to DEPR.w
+	 * instead (rewriting is done *before* the WriteMask test).
+	 */
+	GLboolean RewriteDepthOut;
+	void *Data;
+};
+
+void radeonNqssaDce(GLcontext *ctx, struct gl_program *p, struct radeon_nqssadce_descr* descr);
+
+#endif /* __RADEON_PROGRAM_NQSSADCE_H_ */
diff --git a/src/mesa/drivers/dri/r300/radeon_program.c b/src/mesa/drivers/dri/r300/radeon_program.c
new file mode 100644
index 00000000000..da5e7aefce5
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program.c
@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "radeon_program.h"
+
+#include "shader/prog_print.h"
+
+
+/**
+ * Transform the given clause in the following way:
+ *  1. Replace it with an empty clause
+ *  2. For every instruction in the original clause, try the given
+ *     transformations in order.
+ *  3. If one of the transformations returns GL_TRUE, assume that it
+ *     has emitted the appropriate instruction(s) into the new clause;
+ *     otherwise, copy the instruction verbatim.
+ *
+ * \note The transformation is currently not recursive; in other words,
+ * instructions emitted by transformations are not transformed.
+ *
+ * \note The transform is called 'local' because it can only look at
+ * one instruction at a time.
+ */
+void radeonLocalTransform(
+	GLcontext *Ctx,
+	struct gl_program *program,
+	int num_transformations,
+	struct radeon_program_transformation* transformations)
+{
+	struct radeon_transform_context ctx;
+	int ip;
+
+	ctx.Ctx = Ctx;
+	ctx.Program = program;
+	ctx.OldInstructions = program->Instructions;
+	ctx.OldNumInstructions = program->NumInstructions;
+
+	program->Instructions = 0;
+	program->NumInstructions = 0;
+
+	for(ip = 0; ip < ctx.OldNumInstructions; ++ip) {
+		struct prog_instruction *instr = ctx.OldInstructions + ip;
+		int i;
+
+		for(i = 0; i < num_transformations; ++i) {
+			struct radeon_program_transformation* t = transformations + i;
+
+			if (t->function(&ctx, instr, t->userData))
+				break;
+		}
+
+		if (i >= num_transformations) {
+			struct prog_instruction* dest = radeonAppendInstructions(program, 1);
+			_mesa_copy_instructions(dest, instr, 1);
+		}
+	}
+
+	_mesa_free_instructions(ctx.OldInstructions, ctx.OldNumInstructions);
+}
+
+
+static void scan_instructions(GLboolean* used, const struct prog_instruction* insts, GLuint count)
+{
+	GLuint i;
+	for (i = 0; i < count; i++) {
+		const struct prog_instruction *inst = insts + i;
+		const GLuint n = _mesa_num_inst_src_regs(inst->Opcode);
+		GLuint k;
+
+		for (k = 0; k < n; k++) {
+			if (inst->SrcReg[k].File == PROGRAM_TEMPORARY)
+				used[inst->SrcReg[k].Index] = GL_TRUE;
+		}
+	}
+}
+
+GLint radeonFindFreeTemporary(struct radeon_transform_context *t)
+{
+	GLboolean used[MAX_PROGRAM_TEMPS];
+	GLuint i;
+
+	_mesa_memset(used, 0, sizeof(used));
+	scan_instructions(used, t->Program->Instructions, t->Program->NumInstructions);
+	scan_instructions(used, t->OldInstructions, t->OldNumInstructions);
+
+	for (i = 0; i < MAX_PROGRAM_TEMPS; i++) {
+		if (!used[i])
+			return i;
+	}
+
+	return -1;
+}
+
+
+/**
+ * Append the given number of instructions to the program and return a
+ * pointer to the first new instruction.
+ */
+struct prog_instruction *radeonAppendInstructions(struct gl_program *program, int count)
+{
+	int oldnum = program->NumInstructions;
+	_mesa_insert_instructions(program, oldnum, count);
+	return program->Instructions + oldnum;
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_program.h b/src/mesa/drivers/dri/r300/radeon_program.h
new file mode 100644
index 00000000000..b411f69bc88
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __RADEON_PROGRAM_H_
+#define __RADEON_PROGRAM_H_
+
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+
+
+enum {
+	CLAUSE_MIXED = 0,
+	CLAUSE_ALU,
+	CLAUSE_TEX
+};
+
+enum {
+	PROGRAM_BUILTIN = PROGRAM_FILE_MAX /**< not a real register, but a special swizzle constant */
+};
+
+enum {
+	OPCODE_REPL_ALPHA = MAX_OPCODE /**< used in paired instructions */
+};
+
+#define SWIZZLE_0000 MAKE_SWIZZLE4(SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO)
+#define SWIZZLE_1111 MAKE_SWIZZLE4(SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE, SWIZZLE_ONE)
+
+/**
+ * Transformation context that is passed to local transformations.
+ *
+ * Care must be taken with some operations during transformation,
+ * e.g. finding new temporary registers must use @ref radeonFindFreeTemporary
+ */
+struct radeon_transform_context {
+	GLcontext *Ctx;
+	struct gl_program *Program;
+	struct prog_instruction *OldInstructions;
+	GLuint OldNumInstructions;
+};
+
+/**
+ * A transformation that can be passed to \ref radeonLocalTransform.
+ *
+ * The function will be called once for each instruction.
+ * It has to either emit the appropriate transformed code for the instruction
+ * and return GL_TRUE, or return GL_FALSE if it doesn't understand the
+ * instruction.
+ *
+ * The function gets passed the userData as last parameter.
+ */
+struct radeon_program_transformation {
+	GLboolean (*function)(
+		struct radeon_transform_context*,
+		struct prog_instruction*,
+		void*);
+	void *userData;
+};
+
+void radeonLocalTransform(
+	GLcontext* ctx,
+	struct gl_program *program,
+	int num_transformations,
+	struct radeon_program_transformation* transformations);
+
+/**
+ * Find a usable free temporary register during program transformation
+ */
+GLint radeonFindFreeTemporary(struct radeon_transform_context *ctx);
+
+struct prog_instruction *radeonAppendInstructions(struct gl_program *program, int count);
+
+#endif
diff --git a/src/mesa/drivers/dri/r300/radeon_program_alu.c b/src/mesa/drivers/dri/r300/radeon_program_alu.c
new file mode 100644
index 00000000000..1ef71e74dcf
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program_alu.c
@@ -0,0 +1,658 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * @file
+ *
+ * Shareable transformations that transform "special" ALU instructions
+ * into ALU instructions that are supported by hardware.
+ *
+ */
+
+#include "radeon_program_alu.h"
+
+#include "shader/prog_parameter.h"
+
+
+static struct prog_instruction *emit1(struct gl_program* p,
+	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
+	struct prog_src_register SrcReg)
+{
+	struct prog_instruction *fpi = radeonAppendInstructions(p, 1);
+
+	fpi->Opcode = Opcode;
+	fpi->SaturateMode = Saturate;
+	fpi->DstReg = DstReg;
+	fpi->SrcReg[0] = SrcReg;
+	return fpi;
+}
+
+static struct prog_instruction *emit2(struct gl_program* p,
+	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
+	struct prog_src_register SrcReg0, struct prog_src_register SrcReg1)
+{
+	struct prog_instruction *fpi = radeonAppendInstructions(p, 1);
+
+	fpi->Opcode = Opcode;
+	fpi->SaturateMode = Saturate;
+	fpi->DstReg = DstReg;
+	fpi->SrcReg[0] = SrcReg0;
+	fpi->SrcReg[1] = SrcReg1;
+	return fpi;
+}
+
+static struct prog_instruction *emit3(struct gl_program* p,
+	gl_inst_opcode Opcode, GLuint Saturate, struct prog_dst_register DstReg,
+	struct prog_src_register SrcReg0, struct prog_src_register SrcReg1,
+	struct prog_src_register SrcReg2)
+{
+	struct prog_instruction *fpi = radeonAppendInstructions(p, 1);
+
+	fpi->Opcode = Opcode;
+	fpi->SaturateMode = Saturate;
+	fpi->DstReg = DstReg;
+	fpi->SrcReg[0] = SrcReg0;
+	fpi->SrcReg[1] = SrcReg1;
+	fpi->SrcReg[2] = SrcReg2;
+	return fpi;
+}
+
+static void set_swizzle(struct prog_src_register *SrcReg, int coordinate, int swz)
+{
+	SrcReg->Swizzle &= ~(7 << (3*coordinate));
+	SrcReg->Swizzle |= swz << (3*coordinate);
+}
+
+static void set_negate_base(struct prog_src_register *SrcReg, int coordinate, int negate)
+{
+	SrcReg->NegateBase &= ~(1 << coordinate);
+	SrcReg->NegateBase |= (negate << coordinate);
+}
+
+static struct prog_dst_register dstreg(int file, int index)
+{
+	struct prog_dst_register dst;
+	dst.File = file;
+	dst.Index = index;
+	dst.WriteMask = WRITEMASK_XYZW;
+	dst.CondMask = COND_TR;
+	dst.CondSwizzle = SWIZZLE_NOOP;
+	dst.CondSrc = 0;
+	dst.pad = 0;
+	return dst;
+}
+
+static struct prog_dst_register dstregtmpmask(int index, int mask)
+{
+	struct prog_dst_register dst;
+	dst.File = PROGRAM_TEMPORARY;
+	dst.Index = index;
+	dst.WriteMask = mask;
+	dst.CondMask = COND_TR;
+	dst.CondSwizzle = SWIZZLE_NOOP;
+	dst.CondSrc = 0;
+	dst.pad = 0;
+	return dst;
+}
+
+static const struct prog_src_register builtin_zero = {
+	.File = PROGRAM_BUILTIN,
+	.Index = 0,
+	.Swizzle = SWIZZLE_0000
+};
+static const struct prog_src_register builtin_one = {
+	.File = PROGRAM_BUILTIN,
+	.Index = 0,
+	.Swizzle = SWIZZLE_1111
+};
+static const struct prog_src_register srcreg_undefined = {
+	.File = PROGRAM_UNDEFINED,
+	.Index = 0,
+	.Swizzle = SWIZZLE_NOOP
+};
+
+static struct prog_src_register srcreg(int file, int index)
+{
+	struct prog_src_register src = srcreg_undefined;
+	src.File = file;
+	src.Index = index;
+	return src;
+}
+
+static struct prog_src_register srcregswz(int file, int index, int swz)
+{
+	struct prog_src_register src = srcreg_undefined;
+	src.File = file;
+	src.Index = index;
+	src.Swizzle = swz;
+	return src;
+}
+
+static struct prog_src_register absolute(struct prog_src_register reg)
+{
+	struct prog_src_register newreg = reg;
+	newreg.Abs = 1;
+	newreg.NegateBase = 0;
+	newreg.NegateAbs = 0;
+	return newreg;
+}
+
+static struct prog_src_register negate(struct prog_src_register reg)
+{
+	struct prog_src_register newreg = reg;
+	newreg.NegateAbs = !newreg.NegateAbs;
+	return newreg;
+}
+
+static struct prog_src_register swizzle(struct prog_src_register reg, GLuint x, GLuint y, GLuint z, GLuint w)
+{
+	struct prog_src_register swizzled = reg;
+	swizzled.Swizzle = MAKE_SWIZZLE4(
+		x >= 4 ? x : GET_SWZ(reg.Swizzle, x),
+		y >= 4 ? y : GET_SWZ(reg.Swizzle, y),
+		z >= 4 ? z : GET_SWZ(reg.Swizzle, z),
+		w >= 4 ? w : GET_SWZ(reg.Swizzle, w));
+	return swizzled;
+}
+
+static struct prog_src_register scalar(struct prog_src_register reg)
+{
+	return swizzle(reg, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X);
+}
+
+static void transform_ABS(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	struct prog_src_register src = inst->SrcReg[0];
+	src.Abs = 1;
+	src.NegateBase = 0;
+	src.NegateAbs = 0;
+	emit1(t->Program, OPCODE_MOV, inst->SaturateMode, inst->DstReg, src);
+}
+
+static void transform_DPH(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	struct prog_src_register src0 = inst->SrcReg[0];
+	if (src0.NegateAbs) {
+		if (src0.Abs) {
+			int tempreg = radeonFindFreeTemporary(t);
+			emit1(t->Program, OPCODE_MOV, 0, dstreg(PROGRAM_TEMPORARY, tempreg), src0);
+			src0 = srcreg(src0.File, src0.Index);
+		} else {
+			src0.NegateAbs = 0;
+			src0.NegateBase ^= NEGATE_XYZW;
+		}
+	}
+	set_swizzle(&src0, 3, SWIZZLE_ONE);
+	set_negate_base(&src0, 3, 0);
+	emit2(t->Program, OPCODE_DP4, inst->SaturateMode, inst->DstReg, src0, inst->SrcReg[1]);
+}
+
+/**
+ * [1, src0.y*src1.y, src0.z, src1.w]
+ * So basically MUL with lotsa swizzling.
+ */
+static void transform_DST(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	emit2(t->Program, OPCODE_MUL, inst->SaturateMode, inst->DstReg,
+		swizzle(inst->SrcReg[0], SWIZZLE_ONE, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ONE),
+		swizzle(inst->SrcReg[1], SWIZZLE_ONE, SWIZZLE_Y, SWIZZLE_ONE, SWIZZLE_W));
+}
+
+static void transform_FLR(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+	emit1(t->Program, OPCODE_FRC, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->SrcReg[0]);
+	emit2(t->Program, OPCODE_ADD, inst->SaturateMode, inst->DstReg,
+		inst->SrcReg[0], negate(srcreg(PROGRAM_TEMPORARY, tempreg)));
+}
+
+/**
+ * Definition of LIT (from ARB_fragment_program):
+ *
+ *  tmp = VectorLoad(op0);
+ *  if (tmp.x < 0) tmp.x = 0;
+ *  if (tmp.y < 0) tmp.y = 0;
+ *  if (tmp.w < -(128.0-epsilon)) tmp.w = -(128.0-epsilon);
+ *  else if (tmp.w > 128-epsilon) tmp.w = 128-epsilon;
+ *  result.x = 1.0;
+ *  result.y = tmp.x;
+ *  result.z = (tmp.x > 0) ? RoughApproxPower(tmp.y, tmp.w) : 0.0;
+ *  result.w = 1.0;
+ *
+ * The longest path of computation is the one leading to result.z,
+ * consisting of 5 operations. This implementation of LIT takes
+ * 5 slots, if the subsequent optimization passes are clever enough
+ * to pair instructions correctly.
+ */
+static void transform_LIT(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	static const GLfloat LitConst[4] = { -127.999999 };
+
+	GLuint constant;
+	GLuint constant_swizzle;
+	GLuint temp;
+	int needTemporary = 0;
+	struct prog_src_register srctemp;
+
+	constant = _mesa_add_unnamed_constant(t->Program->Parameters, LitConst, 1, &constant_swizzle);
+
+	if (inst->DstReg.WriteMask != WRITEMASK_XYZW) {
+		needTemporary = 1;
+	} else if (inst->DstReg.File != PROGRAM_TEMPORARY) {
+		// LIT is typically followed by DP3/DP4, so there's no point
+		// in creating special code for this case
+		needTemporary = 1;
+	}
+
+	if (needTemporary) {
+		temp = radeonFindFreeTemporary(t);
+	} else {
+		temp = inst->DstReg.Index;
+	}
+	srctemp = srcreg(PROGRAM_TEMPORARY, temp);
+
+	// tmp.x = max(0.0, Src.x);
+	// tmp.y = max(0.0, Src.y);
+	// tmp.w = clamp(Src.z, -128+eps, 128-eps);
+	emit2(t->Program, OPCODE_MAX, 0,
+		dstregtmpmask(temp, WRITEMASK_XYW),
+		inst->SrcReg[0],
+		swizzle(srcreg(PROGRAM_CONSTANT, constant),
+			SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO, constant_swizzle&3));
+	emit2(t->Program, OPCODE_MIN, 0,
+		dstregtmpmask(temp, WRITEMASK_Z),
+		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+		negate(srcregswz(PROGRAM_CONSTANT, constant, constant_swizzle)));
+
+	// tmp.w = Pow(tmp.y, tmp.w)
+	emit1(t->Program, OPCODE_LG2, 0,
+		dstregtmpmask(temp, WRITEMASK_W),
+		swizzle(srctemp, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y));
+	emit2(t->Program, OPCODE_MUL, 0,
+		dstregtmpmask(temp, WRITEMASK_W),
+		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+		swizzle(srctemp, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z));
+	emit1(t->Program, OPCODE_EX2, 0,
+		dstregtmpmask(temp, WRITEMASK_W),
+		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
+
+	// tmp.z = (tmp.x > 0) ? tmp.w : 0.0
+	emit3(t->Program, OPCODE_CMP, inst->SaturateMode,
+		dstregtmpmask(temp, WRITEMASK_Z),
+		negate(swizzle(srctemp, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
+		swizzle(srctemp, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+		builtin_zero);
+
+	// tmp.x, tmp.y, tmp.w = 1.0, tmp.x, 1.0
+	emit1(t->Program, OPCODE_MOV, inst->SaturateMode,
+		dstregtmpmask(temp, WRITEMASK_XYW),
+		swizzle(srctemp, SWIZZLE_ONE, SWIZZLE_X, SWIZZLE_ONE, SWIZZLE_ONE));
+
+	if (needTemporary)
+		emit1(t->Program, OPCODE_MOV, 0, inst->DstReg, srctemp);
+}
+
+static void transform_LRP(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+
+	emit2(t->Program, OPCODE_ADD, 0,
+		dstreg(PROGRAM_TEMPORARY, tempreg),
+		inst->SrcReg[1], negate(inst->SrcReg[2]));
+	emit3(t->Program, OPCODE_MAD, inst->SaturateMode,
+		inst->DstReg,
+		inst->SrcReg[0], srcreg(PROGRAM_TEMPORARY, tempreg), inst->SrcReg[2]);
+}
+
+static void transform_POW(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+	struct prog_dst_register tempdst = dstreg(PROGRAM_TEMPORARY, tempreg);
+	struct prog_src_register tempsrc = srcreg(PROGRAM_TEMPORARY, tempreg);
+	tempdst.WriteMask = WRITEMASK_W;
+	tempsrc.Swizzle = SWIZZLE_WWWW;
+
+	emit1(t->Program, OPCODE_LG2, 0, tempdst, scalar(inst->SrcReg[0]));
+	emit2(t->Program, OPCODE_MUL, 0, tempdst, tempsrc, scalar(inst->SrcReg[1]));
+	emit1(t->Program, OPCODE_EX2, inst->SaturateMode, inst->DstReg, tempsrc);
+}
+
+static void transform_RSQ(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	emit1(t->Program, OPCODE_RSQ, inst->SaturateMode, inst->DstReg, absolute(inst->SrcReg[0]));
+}
+
+static void transform_SGE(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+
+	emit2(t->Program, OPCODE_ADD, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->SrcReg[0], negate(inst->SrcReg[1]));
+	emit3(t->Program, OPCODE_CMP, inst->SaturateMode, inst->DstReg,
+		srcreg(PROGRAM_TEMPORARY, tempreg), builtin_zero, builtin_one);
+}
+
+static void transform_SLT(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+
+	emit2(t->Program, OPCODE_ADD, 0, dstreg(PROGRAM_TEMPORARY, tempreg), inst->SrcReg[0], negate(inst->SrcReg[1]));
+	emit3(t->Program, OPCODE_CMP, inst->SaturateMode, inst->DstReg,
+		srcreg(PROGRAM_TEMPORARY, tempreg), builtin_one, builtin_zero);
+}
+
+static void transform_SUB(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	emit2(t->Program, OPCODE_ADD, inst->SaturateMode, inst->DstReg, inst->SrcReg[0], negate(inst->SrcReg[1]));
+}
+
+static void transform_SWZ(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	emit1(t->Program, OPCODE_MOV, inst->SaturateMode, inst->DstReg, inst->SrcReg[0]);
+}
+
+static void transform_XPD(struct radeon_transform_context* t,
+	struct prog_instruction* inst)
+{
+	int tempreg = radeonFindFreeTemporary(t);
+
+	emit2(t->Program, OPCODE_MUL, 0, dstreg(PROGRAM_TEMPORARY, tempreg),
+		swizzle(inst->SrcReg[0], SWIZZLE_Z, SWIZZLE_X, SWIZZLE_Y, SWIZZLE_W),
+		swizzle(inst->SrcReg[1], SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_X, SWIZZLE_W));
+	emit3(t->Program, OPCODE_MAD, inst->SaturateMode, inst->DstReg,
+		swizzle(inst->SrcReg[0], SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_X, SWIZZLE_W),
+		swizzle(inst->SrcReg[1], SWIZZLE_Z, SWIZZLE_X, SWIZZLE_Y, SWIZZLE_W),
+		negate(srcreg(PROGRAM_TEMPORARY, tempreg)));
+}
+
+
+/**
+ * Can be used as a transformation for @ref radeonClauseLocalTransform,
+ * no userData necessary.
+ *
+ * Eliminates the following ALU instructions:
+ *  ABS, DPH, DST, FLR, LIT, LRP, POW, SGE, SLT, SUB, SWZ, XPD
+ * using:
+ *  MOV, ADD, MUL, MAD, FRC, DP3, LG2, EX2, CMP
+ *
+ * Transforms RSQ to Radeon's native RSQ by explicitly setting
+ * absolute value.
+ *
+ * @note should be applicable to R300 and R500 fragment programs.
+ */
+GLboolean radeonTransformALU(struct radeon_transform_context* t,
+	struct prog_instruction* inst,
+	void* unused)
+{
+	switch(inst->Opcode) {
+	case OPCODE_ABS: transform_ABS(t, inst); return GL_TRUE;
+	case OPCODE_DPH: transform_DPH(t, inst); return GL_TRUE;
+	case OPCODE_DST: transform_DST(t, inst); return GL_TRUE;
+	case OPCODE_FLR: transform_FLR(t, inst); return GL_TRUE;
+	case OPCODE_LIT: transform_LIT(t, inst); return GL_TRUE;
+	case OPCODE_LRP: transform_LRP(t, inst); return GL_TRUE;
+	case OPCODE_POW: transform_POW(t, inst); return GL_TRUE;
+	case OPCODE_RSQ: transform_RSQ(t, inst); return GL_TRUE;
+	case OPCODE_SGE: transform_SGE(t, inst); return GL_TRUE;
+	case OPCODE_SLT: transform_SLT(t, inst); return GL_TRUE;
+	case OPCODE_SUB: transform_SUB(t, inst); return GL_TRUE;
+	case OPCODE_SWZ: transform_SWZ(t, inst); return GL_TRUE;
+	case OPCODE_XPD: transform_XPD(t, inst); return GL_TRUE;
+	default:
+		return GL_FALSE;
+	}
+}
+
+
+static void sincos_constants(struct radeon_transform_context* t, GLuint *constants)
+{
+	static const GLfloat SinCosConsts[2][4] = {
+		{
+			1.273239545,		// 4/PI
+			-0.405284735,		// -4/(PI*PI)
+			3.141592654,		// PI
+			0.2225			// weight
+		},
+		{
+			0.75,
+			0.5,
+			0.159154943,		// 1/(2*PI)
+			6.283185307		// 2*PI
+		}
+	};
+	int i;
+
+	for(i = 0; i < 2; ++i) {
+		GLuint swz;
+		constants[i] = _mesa_add_unnamed_constant(t->Program->Parameters, SinCosConsts[i], 4, &swz);
+		ASSERT(swz == SWIZZLE_NOOP);
+	}
+}
+
+/**
+ * Approximate sin(x), where x is clamped to (-pi/2, pi/2).
+ *
+ * MUL tmp.xy, src, { 4/PI, -4/(PI^2) }
+ * MAD tmp.x, tmp.y, |src|, tmp.x
+ * MAD tmp.y, tmp.x, |tmp.x|, -tmp.x
+ * MAD dest, tmp.y, weight, tmp.x
+ */
+static void sin_approx(struct radeon_transform_context* t,
+	struct prog_dst_register dst, struct prog_src_register src, const GLuint* constants)
+{
+	GLuint tempreg = radeonFindFreeTemporary(t);
+
+	emit2(t->Program, OPCODE_MUL, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
+		swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+		srcreg(PROGRAM_CONSTANT, constants[0]));
+	emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_X),
+		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
+		absolute(swizzle(src, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
+		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
+	emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_Y),
+		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+		absolute(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)),
+		negate(swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)));
+	emit3(t->Program, OPCODE_MAD, 0, dst,
+		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
+		swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+		swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
+}
+
+/**
+ * Translate the trigonometric functions COS, SIN, and SCS
+ * using only the basic instructions
+ *  MOV, ADD, MUL, MAD, FRC
+ */
+GLboolean radeonTransformTrigSimple(struct radeon_transform_context* t,
+	struct prog_instruction* inst,
+	void* unused)
+{
+	if (inst->Opcode != OPCODE_COS &&
+	    inst->Opcode != OPCODE_SIN &&
+	    inst->Opcode != OPCODE_SCS)
+		return GL_FALSE;
+
+	GLuint constants[2];
+	GLuint tempreg = radeonFindFreeTemporary(t);
+
+	sincos_constants(t, constants);
+
+	if (inst->Opcode == OPCODE_COS) {
+		// MAD tmp.x, src, 1/(2*PI), 0.75
+		// FRC tmp.x, tmp.x
+		// MAD tmp.z, tmp.x, 2*PI, -PI
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(inst->SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X));
+		emit1(t->Program, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
+
+		sin_approx(t, inst->DstReg,
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			constants);
+	} else if (inst->Opcode == OPCODE_SIN) {
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(inst->SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y));
+		emit1(t->Program, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W));
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_W),
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
+
+		sin_approx(t, inst->DstReg,
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			constants);
+	} else {
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
+			swizzle(inst->SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W));
+		emit1(t->Program, OPCODE_FRC, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
+			srcreg(PROGRAM_TEMPORARY, tempreg));
+		emit3(t->Program, OPCODE_MAD, 0, dstregtmpmask(tempreg, WRITEMASK_XY),
+			srcreg(PROGRAM_TEMPORARY, tempreg),
+			swizzle(srcreg(PROGRAM_CONSTANT, constants[1]), SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W),
+			negate(swizzle(srcreg(PROGRAM_CONSTANT, constants[0]), SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)));
+
+		struct prog_dst_register dst = inst->DstReg;
+
+		dst.WriteMask = inst->DstReg.WriteMask & WRITEMASK_X;
+		sin_approx(t, dst,
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+			constants);
+
+		dst.WriteMask = inst->DstReg.WriteMask & WRITEMASK_Y;
+		sin_approx(t, dst,
+			swizzle(srcreg(PROGRAM_TEMPORARY, tempreg), SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y),
+			constants);
+	}
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Transform the trigonometric functions COS, SIN, and SCS
+ * to include pre-scaling by 1/(2*PI) and taking the fractional
+ * part, so that the input to COS and SIN is always in the range [0,1).
+ * SCS is replaced by one COS and one SIN instruction.
+ *
+ * @warning This transformation implicitly changes the semantics of SIN and COS!
+ */
+GLboolean radeonTransformTrigScale(struct radeon_transform_context* t,
+	struct prog_instruction* inst,
+	void* unused)
+{
+	if (inst->Opcode != OPCODE_COS &&
+	    inst->Opcode != OPCODE_SIN &&
+	    inst->Opcode != OPCODE_SCS)
+		return GL_FALSE;
+
+	static const GLfloat RCP_2PI[] = { 0.15915494309189535 };
+	GLuint temp;
+	GLuint constant;
+	GLuint constant_swizzle;
+
+	temp = radeonFindFreeTemporary(t);
+	constant = _mesa_add_unnamed_constant(t->Program->Parameters, RCP_2PI, 1, &constant_swizzle);
+
+	emit2(t->Program, OPCODE_MUL, 0, dstregtmpmask(temp, WRITEMASK_W),
+		swizzle(inst->SrcReg[0], SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X),
+		srcregswz(PROGRAM_CONSTANT, constant, constant_swizzle));
+	emit1(t->Program, OPCODE_FRC, 0, dstregtmpmask(temp, WRITEMASK_W),
+		srcreg(PROGRAM_TEMPORARY, temp));
+
+	if (inst->Opcode == OPCODE_COS) {
+		emit1(t->Program, OPCODE_COS, inst->SaturateMode, inst->DstReg,
+			srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+	} else if (inst->Opcode == OPCODE_SIN) {
+		emit1(t->Program, OPCODE_SIN, inst->SaturateMode,
+			inst->DstReg, srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+	} else if (inst->Opcode == OPCODE_SCS) {
+		struct prog_dst_register moddst = inst->DstReg;
+
+		if (inst->DstReg.WriteMask & WRITEMASK_X) {
+			moddst.WriteMask = WRITEMASK_X;
+			emit1(t->Program, OPCODE_COS, inst->SaturateMode, moddst,
+				srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+		}
+		if (inst->DstReg.WriteMask & WRITEMASK_Y) {
+			moddst.WriteMask = WRITEMASK_Y;
+			emit1(t->Program, OPCODE_SIN, inst->SaturateMode, moddst,
+				srcregswz(PROGRAM_TEMPORARY, temp, SWIZZLE_WWWW));
+		}
+	}
+
+	return GL_TRUE;
+}
+
+/**
+ * Rewrite DDX/DDY instructions to properly work with r5xx shaders.
+ * The r5xx MDH/MDV instruction provides per-quad partial derivatives.
+ * It takes the form A*B+C. A and C are set by setting src0. B should be -1.
+ *
+ * @warning This explicitly changes the form of DDX and DDY!
+ */
+
+GLboolean radeonTransformDeriv(struct radeon_transform_context* t,
+	struct prog_instruction* inst,
+	void* unused)
+{
+	if (inst->Opcode != OPCODE_DDX && inst->Opcode != OPCODE_DDY)
+		return GL_FALSE;
+
+	struct prog_src_register B = inst->SrcReg[1];
+
+	B.Swizzle = MAKE_SWIZZLE4(SWIZZLE_ONE, SWIZZLE_ONE,
+						SWIZZLE_ONE, SWIZZLE_ONE);
+	B.NegateBase = NEGATE_XYZW;
+
+	emit2(t->Program, inst->Opcode, inst->SaturateMode, inst->DstReg,
+		inst->SrcReg[0], B);
+
+	return GL_TRUE;
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_program_alu.h b/src/mesa/drivers/dri/r300/radeon_program_alu.h
new file mode 100644
index 00000000000..b45958115cf
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program_alu.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __RADEON_PROGRAM_ALU_H_
+#define __RADEON_PROGRAM_ALU_H_
+
+#include "radeon_program.h"
+
+GLboolean radeonTransformALU(
+	struct radeon_transform_context *t,
+	struct prog_instruction*,
+	void*);
+
+GLboolean radeonTransformTrigSimple(
+	struct radeon_transform_context *t,
+	struct prog_instruction*,
+	void*);
+
+GLboolean radeonTransformTrigScale(
+	struct radeon_transform_context *t,
+	struct prog_instruction*,
+	void*);
+
+GLboolean radeonTransformDeriv(
+	struct radeon_transform_context *t,
+	struct prog_instruction*,
+	void*);
+
+#endif /* __RADEON_PROGRAM_ALU_H_ */
diff --git a/src/mesa/drivers/dri/r300/radeon_program_pair.c b/src/mesa/drivers/dri/r300/radeon_program_pair.c
new file mode 100644
index 00000000000..5ad50d2863a
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program_pair.c
@@ -0,0 +1,1001 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/**
+ * @file
+ *
+ * Perform temporary register allocation and attempt to pair off instructions
+ * in RGB and Alpha pairs. Also attempts to optimize the TEX instruction
+ * vs. ALU instruction scheduling.
+ */
+
+#include "radeon_program_pair.h"
+
+#include "radeon_context.h"
+
+#include "shader/prog_print.h"
+
+#define error(fmt, args...) do { \
+	_mesa_problem(s->Ctx, "%s::%s(): " fmt "\n",	\
+		__FILE__, __FUNCTION__, ##args);	\
+	s->Error = GL_TRUE;				\
+} while(0)
+
+struct pair_state_instruction {
+	GLuint IsTex:1; /**< Is a texture instruction */
+	GLuint NeedRGB:1; /**< Needs the RGB ALU */
+	GLuint NeedAlpha:1; /**< Needs the Alpha ALU */
+	GLuint IsTranscendent:1; /**< Is a special transcendent instruction */
+
+	/**
+	 * Number of (read and write) dependencies that must be resolved before
+	 * this instruction can be scheduled.
+	 */
+	GLuint NumDependencies:5;
+
+	/**
+	 * Next instruction in the linked list of ready instructions.
+	 */
+	struct pair_state_instruction *NextReady;
+
+	/**
+	 * Values that this instruction writes
+	 */
+	struct reg_value *Values[4];
+};
+
+
+/**
+ * Used to keep track of which instructions read a value.
+ */
+struct reg_value_reader {
+	GLuint IP; /**< IP of the instruction that performs this access */
+	struct reg_value_reader *Next;
+};
+
+/**
+ * Used to keep track which values are stored in each component of a
+ * PROGRAM_TEMPORARY.
+ */
+struct reg_value {
+	GLuint IP; /**< IP of the instruction that writes this value */
+	struct reg_value *Next; /**< Pointer to the next value to be written to the same PROGRAM_TEMPORARY component */
+
+	/**
+	 * Unordered linked list of instructions that read from this value.
+	 */
+	struct reg_value_reader *Readers;
+
+	/**
+	 * Number of readers of this value. This is calculated during @ref scan_instructions
+	 * and continually decremented during code emission.
+	 * When this count reaches zero, the instruction that writes the @ref Next value
+	 * can be scheduled.
+	 */
+	GLuint NumReaders;
+};
+
+/**
+ * Used to translate a PROGRAM_INPUT or PROGRAM_TEMPORARY Mesa register
+ * to the proper hardware temporary.
+ */
+struct pair_register_translation {
+	GLuint Allocated:1;
+	GLuint HwIndex:8;
+	GLuint RefCount:23; /**< # of times this occurs in an unscheduled instruction SrcReg or DstReg */
+
+	/**
+	 * Notes the value that is currently contained in each component
+	 * (only used for PROGRAM_TEMPORARY registers).
+	 */
+	struct reg_value *Value[4];
+};
+
+struct pair_state {
+	GLcontext *Ctx;
+	struct gl_program *Program;
+	const struct radeon_pair_handler *Handler;
+	GLboolean Error;
+	GLboolean Debug;
+	GLboolean Verbose;
+	void *UserData;
+
+	/**
+	 * Translate Mesa registers to hardware registers
+	 */
+	struct pair_register_translation Inputs[FRAG_ATTRIB_MAX];
+	struct pair_register_translation Temps[MAX_PROGRAM_TEMPS];
+
+	/**
+	 * Derived information about program instructions.
+	 */
+	struct pair_state_instruction *Instructions;
+
+	struct {
+		GLuint RefCount; /**< # of times this occurs in an unscheduled SrcReg or DstReg */
+	} HwTemps[128];
+
+	/**
+	 * Linked list of instructions that can be scheduled right now,
+	 * based on which ALU/TEX resources they require.
+	 */
+	struct pair_state_instruction *ReadyFullALU;
+	struct pair_state_instruction *ReadyRGB;
+	struct pair_state_instruction *ReadyAlpha;
+	struct pair_state_instruction *ReadyTEX;
+
+	/**
+	 * Pool of @ref reg_value structures for fast allocation.
+	 */
+	struct reg_value *ValuePool;
+	GLuint ValuePoolUsed;
+	struct reg_value_reader *ReaderPool;
+	GLuint ReaderPoolUsed;
+};
+
+
+static struct pair_register_translation *get_register(struct pair_state *s, GLuint file, GLuint index)
+{
+	switch(file) {
+	case PROGRAM_TEMPORARY: return &s->Temps[index];
+	case PROGRAM_INPUT: return &s->Inputs[index];
+	default: return 0;
+	}
+}
+
+static void alloc_hw_reg(struct pair_state *s, GLuint file, GLuint index, GLuint hwindex)
+{
+	struct pair_register_translation *t = get_register(s, file, index);
+	ASSERT(!s->HwTemps[hwindex].RefCount);
+	ASSERT(!t->Allocated);
+	s->HwTemps[hwindex].RefCount = t->RefCount;
+	t->Allocated = 1;
+	t->HwIndex = hwindex;
+}
+
+static GLuint get_hw_reg(struct pair_state *s, GLuint file, GLuint index)
+{
+	GLuint hwindex;
+
+	struct pair_register_translation *t = get_register(s, file, index);
+	if (!t) {
+		_mesa_problem(s->Ctx, "get_hw_reg: %i[%i]\n", file, index);
+		return 0;
+	}
+
+	if (t->Allocated)
+		return t->HwIndex;
+
+	for(hwindex = 0; hwindex < s->Handler->MaxHwTemps; ++hwindex)
+		if (!s->HwTemps[hwindex].RefCount)
+			break;
+
+	if (hwindex >= s->Handler->MaxHwTemps) {
+		error("Ran out of hardware temporaries");
+		return 0;
+	}
+
+	alloc_hw_reg(s, file, index, hwindex);
+	return hwindex;
+}
+
+
+static void deref_hw_reg(struct pair_state *s, GLuint hwindex)
+{
+	if (!s->HwTemps[hwindex].RefCount) {
+		error("Hwindex %i refcount error", hwindex);
+		return;
+	}
+
+	s->HwTemps[hwindex].RefCount--;
+}
+
+static void add_pairinst_to_list(struct pair_state_instruction **list, struct pair_state_instruction *pairinst)
+{
+	pairinst->NextReady = *list;
+	*list = pairinst;
+}
+
+/**
+ * The instruction at the given IP has become ready. Link it into the ready
+ * instructions.
+ */
+static void instruction_ready(struct pair_state *s, int ip)
+{
+	struct pair_state_instruction *pairinst = s->Instructions + ip;
+
+	if (s->Verbose)
+		_mesa_printf("instruction_ready(%i)\n", ip);
+
+	if (pairinst->IsTex)
+		add_pairinst_to_list(&s->ReadyTEX, pairinst);
+	else if (!pairinst->NeedAlpha)
+		add_pairinst_to_list(&s->ReadyRGB, pairinst);
+	else if (!pairinst->NeedRGB)
+		add_pairinst_to_list(&s->ReadyAlpha, pairinst);
+	else
+		add_pairinst_to_list(&s->ReadyFullALU, pairinst);
+}
+
+
+/**
+ * Finally rewrite ADD, MOV, MUL as the appropriate native instruction
+ * and reverse the order of arguments for CMP.
+ */
+static void final_rewrite(struct pair_state *s, struct prog_instruction *inst)
+{
+	struct prog_src_register tmp;
+
+	switch(inst->Opcode) {
+	case OPCODE_ADD:
+		inst->SrcReg[2] = inst->SrcReg[1];
+		inst->SrcReg[1].File = PROGRAM_BUILTIN;
+		inst->SrcReg[1].Swizzle = SWIZZLE_1111;
+		inst->SrcReg[1].NegateBase = 0;
+		inst->SrcReg[1].NegateAbs = 0;
+		inst->Opcode = OPCODE_MAD;
+		break;
+	case OPCODE_CMP:
+		tmp = inst->SrcReg[2];
+		inst->SrcReg[2] = inst->SrcReg[0];
+		inst->SrcReg[0] = tmp;
+		break;
+	case OPCODE_MOV:
+		/* AMD say we should use CMP.
+		 * However, when we transform
+		 *  KIL -r0;
+		 * into
+		 *  CMP tmp, -r0, -r0, 0;
+		 *  KIL tmp;
+		 * we get incorrect behaviour on R500 when r0 == 0.0.
+		 * It appears that the R500 KIL hardware treats -0.0 as less
+		 * than zero.
+		 */
+		inst->SrcReg[1].File = PROGRAM_BUILTIN;
+		inst->SrcReg[1].Swizzle = SWIZZLE_1111;
+		inst->SrcReg[2].File = PROGRAM_BUILTIN;
+		inst->SrcReg[2].Swizzle = SWIZZLE_0000;
+		inst->Opcode = OPCODE_MAD;
+		break;
+	case OPCODE_MUL:
+		inst->SrcReg[2].File = PROGRAM_BUILTIN;
+		inst->SrcReg[2].Swizzle = SWIZZLE_0000;
+		inst->Opcode = OPCODE_MAD;
+		break;
+	default:
+		/* nothing to do */
+		break;
+	}
+}
+
+
+/**
+ * Classify an instruction according to which ALUs etc. it needs
+ */
+static void classify_instruction(struct pair_state *s,
+	struct prog_instruction *inst, struct pair_state_instruction *pairinst)
+{
+	pairinst->NeedRGB = (inst->DstReg.WriteMask & WRITEMASK_XYZ) ? 1 : 0;
+	pairinst->NeedAlpha = (inst->DstReg.WriteMask & WRITEMASK_W) ? 1 : 0;
+
+	switch(inst->Opcode) {
+	case OPCODE_ADD:
+	case OPCODE_CMP:
+	case OPCODE_DDX:
+	case OPCODE_DDY:
+	case OPCODE_FRC:
+	case OPCODE_MAD:
+	case OPCODE_MAX:
+	case OPCODE_MIN:
+	case OPCODE_MOV:
+	case OPCODE_MUL:
+		break;
+	case OPCODE_COS:
+	case OPCODE_EX2:
+	case OPCODE_LG2:
+	case OPCODE_RCP:
+	case OPCODE_RSQ:
+	case OPCODE_SIN:
+		pairinst->IsTranscendent = 1;
+		pairinst->NeedAlpha = 1;
+		break;
+	case OPCODE_DP4:
+		pairinst->NeedAlpha = 1;
+		/* fall through */
+	case OPCODE_DP3:
+		pairinst->NeedRGB = 1;
+		break;
+	case OPCODE_KIL:
+	case OPCODE_TEX:
+	case OPCODE_TXB:
+	case OPCODE_TXP:
+	case OPCODE_END:
+		pairinst->IsTex = 1;
+		break;
+	default:
+		error("Unknown opcode %d\n", inst->Opcode);
+		break;
+	}
+}
+
+
+/**
+ * Count which (input, temporary) register is read and written how often,
+ * and scan the instruction stream to find dependencies.
+ */
+static void scan_instructions(struct pair_state *s)
+{
+	struct prog_instruction *inst;
+	struct pair_state_instruction *pairinst;
+	GLuint ip;
+
+	for(inst = s->Program->Instructions, pairinst = s->Instructions, ip = 0;
+	    inst->Opcode != OPCODE_END;
+	    ++inst, ++pairinst, ++ip) {
+		final_rewrite(s, inst);
+		classify_instruction(s, inst, pairinst);
+
+		int nsrc = _mesa_num_inst_src_regs(inst->Opcode);
+		int j;
+		for(j = 0; j < nsrc; j++) {
+			struct pair_register_translation *t =
+				get_register(s, inst->SrcReg[j].File, inst->SrcReg[j].Index);
+			if (!t)
+				continue;
+
+			t->RefCount++;
+
+			if (inst->SrcReg[j].File == PROGRAM_TEMPORARY) {
+				int i;
+				for(i = 0; i < 4; ++i) {
+					GLuint swz = GET_SWZ(inst->SrcReg[j].Swizzle, i);
+					if (swz >= 4)
+						continue; /* constant or NIL swizzle */
+					if (!t->Value[swz])
+						continue; /* this is an undefined read */
+
+					/* Do not add a dependency if this instruction
+					 * also rewrites the value. The code below adds
+					 * a dependency for the DstReg, which is a superset
+					 * of the SrcReg dependency. */
+					if (inst->DstReg.File == PROGRAM_TEMPORARY &&
+					    inst->DstReg.Index == inst->SrcReg[j].Index &&
+					    GET_BIT(inst->DstReg.WriteMask, swz))
+						continue;
+
+					struct reg_value_reader* r = &s->ReaderPool[s->ReaderPoolUsed++];
+					pairinst->NumDependencies++;
+					t->Value[swz]->NumReaders++;
+					r->IP = ip;
+					r->Next = t->Value[swz]->Readers;
+					t->Value[swz]->Readers = r;
+				}
+			}
+		}
+
+		int ndst = _mesa_num_inst_dst_regs(inst->Opcode);
+		if (ndst) {
+			struct pair_register_translation *t =
+				get_register(s, inst->DstReg.File, inst->DstReg.Index);
+			if (t) {
+				t->RefCount++;
+
+				if (inst->DstReg.File == PROGRAM_TEMPORARY) {
+					int j;
+					for(j = 0; j < 4; ++j) {
+						if (!GET_BIT(inst->DstReg.WriteMask, j))
+							continue;
+
+						struct reg_value* v = &s->ValuePool[s->ValuePoolUsed++];
+						v->IP = ip;
+						if (t->Value[j]) {
+							pairinst->NumDependencies++;
+							t->Value[j]->Next = v;
+						}
+						t->Value[j] = v;
+						pairinst->Values[j] = v;
+					}
+				}
+			}
+		}
+
+		if (s->Verbose)
+			_mesa_printf("scan(%i): NumDeps = %i\n", ip, pairinst->NumDependencies);
+
+		if (!pairinst->NumDependencies)
+			instruction_ready(s, ip);
+	}
+
+	/* Clear the PROGRAM_TEMPORARY state */
+	int i, j;
+	for(i = 0; i < MAX_PROGRAM_TEMPS; ++i) {
+		for(j = 0; j < 4; ++j)
+			s->Temps[i].Value[j] = 0;
+	}
+}
+
+
+/**
+ * Reserve hardware temporary registers for the program inputs.
+ *
+ * @note This allocation is performed explicitly, because the order of inputs
+ * is determined by the RS hardware.
+ */
+static void allocate_input_registers(struct pair_state *s)
+{
+	GLuint InputsRead = s->Program->InputsRead;
+	int i;
+	GLuint hwindex = 0;
+
+	/* Texcoords come first */
+	for (i = 0; i < s->Ctx->Const.MaxTextureUnits; i++) {
+		if (InputsRead & (FRAG_BIT_TEX0 << i))
+			alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_TEX0+i, hwindex++);
+	}
+	InputsRead &= ~FRAG_BITS_TEX_ANY;
+
+	/* fragment position treated as a texcoord */
+	if (InputsRead & FRAG_BIT_WPOS)
+		alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_WPOS, hwindex++);
+	InputsRead &= ~FRAG_BIT_WPOS;
+
+	/* Then primary colour */
+	if (InputsRead & FRAG_BIT_COL0)
+		alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_COL0, hwindex++);
+	InputsRead &= ~FRAG_BIT_COL0;
+
+	/* Secondary color */
+	if (InputsRead & FRAG_BIT_COL1)
+		alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_COL1, hwindex++);
+	InputsRead &= ~FRAG_BIT_COL1;
+
+	/* Anything else */
+	if (InputsRead)
+		error("Don't know how to handle inputs 0x%x\n", InputsRead);
+}
+
+
+static void decrement_dependencies(struct pair_state *s, int ip)
+{
+	struct pair_state_instruction *pairinst = s->Instructions + ip;
+	ASSERT(pairinst->NumDependencies > 0);
+	if (!--pairinst->NumDependencies)
+		instruction_ready(s, ip);
+}
+
+/**
+ * Update the dependency tracking state based on what the instruction
+ * at the given IP does.
+ */
+static void commit_instruction(struct pair_state *s, int ip)
+{
+	struct prog_instruction *inst = s->Program->Instructions + ip;
+	struct pair_state_instruction *pairinst = s->Instructions + ip;
+
+	if (s->Verbose)
+		_mesa_printf("commit_instruction(%i)\n", ip);
+
+	if (inst->DstReg.File == PROGRAM_TEMPORARY) {
+		struct pair_register_translation *t = &s->Temps[inst->DstReg.Index];
+		deref_hw_reg(s, t->HwIndex);
+
+		int i;
+		for(i = 0; i < 4; ++i) {
+			if (!GET_BIT(inst->DstReg.WriteMask, i))
+				continue;
+
+			t->Value[i] = pairinst->Values[i];
+			if (t->Value[i]->NumReaders) {
+				struct reg_value_reader *r;
+				for(r = pairinst->Values[i]->Readers; r; r = r->Next)
+					decrement_dependencies(s, r->IP);
+			} else if (t->Value[i]->Next) {
+				/* This happens when the only reader writes
+				 * the register at the same time */
+				decrement_dependencies(s, t->Value[i]->Next->IP);
+			}
+		}
+	}
+
+	int nsrc = _mesa_num_inst_src_regs(inst->Opcode);
+	int i;
+	for(i = 0; i < nsrc; i++) {
+		struct pair_register_translation *t = get_register(s, inst->SrcReg[i].File, inst->SrcReg[i].Index);
+		if (!t)
+			continue;
+
+		deref_hw_reg(s, get_hw_reg(s, inst->SrcReg[i].File, inst->SrcReg[i].Index));
+
+		if (inst->SrcReg[i].File != PROGRAM_TEMPORARY)
+			continue;
+
+		int j;
+		for(j = 0; j < 4; ++j) {
+			GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, j);
+			if (swz >= 4)
+				continue;
+			if (!t->Value[swz])
+				continue;
+
+			/* Do not free a dependency if this instruction
+			 * also rewrites the value. See scan_instructions. */
+			if (inst->DstReg.File == PROGRAM_TEMPORARY &&
+			    inst->DstReg.Index == inst->SrcReg[i].Index &&
+			    GET_BIT(inst->DstReg.WriteMask, swz))
+				continue;
+
+			if (!--t->Value[swz]->NumReaders) {
+				if (t->Value[swz]->Next)
+					decrement_dependencies(s, t->Value[swz]->Next->IP);
+			}
+		}
+	}
+}
+
+
+/**
+ * Emit all ready texture instructions in a single block.
+ *
+ * Emit as a single block to (hopefully) sample many textures in parallel,
+ * and to avoid hardware indirections on R300.
+ *
+ * In R500, we don't really know when the result of a texture instruction
+ * arrives. So allocate all destinations first, to make sure they do not
+ * arrive early and overwrite a texture coordinate we're going to use later
+ * in the block.
+ */
+static void emit_all_tex(struct pair_state *s)
+{
+	struct pair_state_instruction *readytex;
+	struct pair_state_instruction *pairinst;
+
+	ASSERT(s->ReadyTEX);
+
+	// Don't let the ready list change under us!
+	readytex = s->ReadyTEX;
+	s->ReadyTEX = 0;
+
+	// Allocate destination hardware registers in one block to avoid conflicts.
+	for(pairinst = readytex; pairinst; pairinst = pairinst->NextReady) {
+		int ip = pairinst - s->Instructions;
+		struct prog_instruction *inst = s->Program->Instructions + ip;
+		if (inst->Opcode != OPCODE_KIL)
+			get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
+	}
+
+	if (s->Debug)
+		_mesa_printf(" BEGIN_TEX\n");
+
+	if (s->Handler->BeginTexBlock)
+		s->Error = s->Error || !s->Handler->BeginTexBlock(s->UserData);
+
+	for(pairinst = readytex; pairinst; pairinst = pairinst->NextReady) {
+		int ip = pairinst - s->Instructions;
+		struct prog_instruction *inst = s->Program->Instructions + ip;
+		commit_instruction(s, ip);
+
+		if (inst->Opcode != OPCODE_KIL)
+			inst->DstReg.Index = get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
+		inst->SrcReg[0].Index = get_hw_reg(s, inst->SrcReg[0].File, inst->SrcReg[0].Index);
+
+		if (s->Debug) {
+			_mesa_printf("   ");
+			_mesa_print_instruction(inst);
+		}
+		s->Error = s->Error || !s->Handler->EmitTex(s->UserData, inst);
+	}
+
+	if (s->Debug)
+		_mesa_printf(" END_TEX\n");
+}
+
+
+static int alloc_pair_source(struct pair_state *s, struct radeon_pair_instruction *pair,
+	struct prog_src_register src, GLboolean rgb, GLboolean alpha)
+{
+	int candidate = -1;
+	int candidate_quality = -1;
+	int i;
+
+	if (!rgb && !alpha)
+		return 0;
+
+	GLuint constant;
+	GLuint index;
+
+	if (src.File == PROGRAM_TEMPORARY || src.File == PROGRAM_INPUT) {
+		constant = 0;
+		index = get_hw_reg(s, src.File, src.Index);
+	} else {
+		constant = 1;
+		s->Error |= !s->Handler->EmitConst(s->UserData, src.File, src.Index, &index);
+	}
+
+	for(i = 0; i < 3; ++i) {
+		int q = 0;
+		if (rgb) {
+			if (pair->RGB.Src[i].Used) {
+				if (pair->RGB.Src[i].Constant != constant ||
+				    pair->RGB.Src[i].Index != index)
+					continue;
+				q++;
+			}
+		}
+		if (alpha) {
+			if (pair->Alpha.Src[i].Used) {
+				if (pair->Alpha.Src[i].Constant != constant ||
+				    pair->Alpha.Src[i].Index != index)
+					continue;
+				q++;
+			}
+		}
+		if (q > candidate_quality) {
+			candidate_quality = q;
+			candidate = i;
+		}
+	}
+
+	if (candidate >= 0) {
+		if (rgb) {
+			pair->RGB.Src[candidate].Used = 1;
+			pair->RGB.Src[candidate].Constant = constant;
+			pair->RGB.Src[candidate].Index = index;
+		}
+		if (alpha) {
+			pair->Alpha.Src[candidate].Used = 1;
+			pair->Alpha.Src[candidate].Constant = constant;
+			pair->Alpha.Src[candidate].Index = index;
+		}
+	}
+
+	return candidate;
+}
+
+/**
+ * Fill the given ALU instruction's opcodes and source operands into the given pair,
+ * if possible.
+ */
+static GLboolean fill_instruction_into_pair(struct pair_state *s, struct radeon_pair_instruction *pair, int ip)
+{
+	struct pair_state_instruction *pairinst = s->Instructions + ip;
+	struct prog_instruction *inst = s->Program->Instructions + ip;
+
+	ASSERT(!pairinst->NeedRGB || pair->RGB.Opcode == OPCODE_NOP);
+	ASSERT(!pairinst->NeedAlpha || pair->Alpha.Opcode == OPCODE_NOP);
+
+	if (pairinst->NeedRGB) {
+		if (pairinst->IsTranscendent)
+			pair->RGB.Opcode = OPCODE_REPL_ALPHA;
+		else
+			pair->RGB.Opcode = inst->Opcode;
+		if (inst->SaturateMode == SATURATE_ZERO_ONE)
+			pair->RGB.Saturate = 1;
+	}
+	if (pairinst->NeedAlpha) {
+		pair->Alpha.Opcode = inst->Opcode;
+		if (inst->SaturateMode == SATURATE_ZERO_ONE)
+			pair->Alpha.Saturate = 1;
+	}
+
+	int nargs = _mesa_num_inst_src_regs(inst->Opcode);
+	int i;
+
+	/* Special case for DDX/DDY (MDH/MDV). */
+	if (inst->Opcode == OPCODE_DDX || inst->Opcode == OPCODE_DDY) {
+		if (pair->RGB.Src[0].Used || pair->Alpha.Src[0].Used)
+			return GL_FALSE;
+		else
+			nargs++;
+	}
+
+	for(i = 0; i < nargs; ++i) {
+		int source;
+		if (pairinst->NeedRGB && !pairinst->IsTranscendent) {
+			GLboolean srcrgb = GL_FALSE;
+			GLboolean srcalpha = GL_FALSE;
+			GLuint negatebase = 0;
+			int j;
+			for(j = 0; j < 3; ++j) {
+				GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, j);
+				if (swz < 3)
+					srcrgb = GL_TRUE;
+				else if (swz < 4)
+					srcalpha = GL_TRUE;
+				if (swz != SWIZZLE_NIL && GET_BIT(inst->SrcReg[i].NegateBase, j))
+					negatebase = 1;
+			}
+			source = alloc_pair_source(s, pair, inst->SrcReg[i], srcrgb, srcalpha);
+			if (source < 0)
+				return GL_FALSE;
+			pair->RGB.Arg[i].Source = source;
+			pair->RGB.Arg[i].Swizzle = inst->SrcReg[i].Swizzle & 0x1ff;
+			pair->RGB.Arg[i].Abs = inst->SrcReg[i].Abs;
+			pair->RGB.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].NegateAbs;
+		}
+		if (pairinst->NeedAlpha) {
+			GLboolean srcrgb = GL_FALSE;
+			GLboolean srcalpha = GL_FALSE;
+			GLuint negatebase = GET_BIT(inst->SrcReg[i].NegateBase, pairinst->IsTranscendent ? 0 : 3);
+			GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, pairinst->IsTranscendent ? 0 : 3);
+			if (swz < 3)
+				srcrgb = GL_TRUE;
+			else if (swz < 4)
+				srcalpha = GL_TRUE;
+			source = alloc_pair_source(s, pair, inst->SrcReg[i], srcrgb, srcalpha);
+			if (source < 0)
+				return GL_FALSE;
+			pair->Alpha.Arg[i].Source = source;
+			pair->Alpha.Arg[i].Swizzle = swz;
+			pair->Alpha.Arg[i].Abs = inst->SrcReg[i].Abs;
+			pair->Alpha.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].NegateAbs;
+		}
+	}
+
+	return GL_TRUE;
+}
+
+
+/**
+ * Fill in the destination register information.
+ *
+ * This is split from filling in source registers because we want
+ * to avoid allocating hardware temporaries for destinations until
+ * we are absolutely certain that we're going to emit a certain
+ * instruction pairing.
+ */
+static void fill_dest_into_pair(struct pair_state *s, struct radeon_pair_instruction *pair, int ip)
+{
+	struct pair_state_instruction *pairinst = s->Instructions + ip;
+	struct prog_instruction *inst = s->Program->Instructions + ip;
+
+	if (inst->DstReg.File == PROGRAM_OUTPUT) {
+		if (inst->DstReg.Index == FRAG_RESULT_COLR) {
+			pair->RGB.OutputWriteMask |= inst->DstReg.WriteMask & WRITEMASK_XYZ;
+			pair->Alpha.OutputWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		} else if (inst->DstReg.Index == FRAG_RESULT_DEPR) {
+			pair->Alpha.DepthWriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		}
+	} else {
+		GLuint hwindex = get_hw_reg(s, inst->DstReg.File, inst->DstReg.Index);
+		if (pairinst->NeedRGB) {
+			pair->RGB.DestIndex = hwindex;
+			pair->RGB.WriteMask |= inst->DstReg.WriteMask & WRITEMASK_XYZ;
+		}
+		if (pairinst->NeedAlpha) {
+			pair->Alpha.DestIndex = hwindex;
+			pair->Alpha.WriteMask |= GET_BIT(inst->DstReg.WriteMask, 3);
+		}
+	}
+}
+
+
+/**
+ * Find a good ALU instruction or pair of ALU instruction and emit it.
+ *
+ * Prefer emitting full ALU instructions, so that when we reach a point
+ * where no full ALU instruction can be emitted, we have more candidates
+ * for RGB/Alpha pairing.
+ */
+static void emit_alu(struct pair_state *s)
+{
+	struct radeon_pair_instruction pair;
+
+	if (s->ReadyFullALU || !(s->ReadyRGB && s->ReadyAlpha)) {
+		int ip;
+		if (s->ReadyFullALU) {
+			ip = s->ReadyFullALU - s->Instructions;
+			s->ReadyFullALU = s->ReadyFullALU->NextReady;
+		} else if (s->ReadyRGB) {
+			ip = s->ReadyRGB - s->Instructions;
+			s->ReadyRGB = s->ReadyRGB->NextReady;
+		} else {
+			ip = s->ReadyAlpha - s->Instructions;
+			s->ReadyAlpha = s->ReadyAlpha->NextReady;
+		}
+
+		_mesa_bzero(&pair, sizeof(pair));
+		fill_instruction_into_pair(s, &pair, ip);
+		fill_dest_into_pair(s, &pair, ip);
+		commit_instruction(s, ip);
+	} else {
+		struct pair_state_instruction **prgb;
+		struct pair_state_instruction **palpha;
+
+		/* Some pairings might fail because they require too
+		 * many source slots; try all possible pairings if necessary */
+		for(prgb = &s->ReadyRGB; *prgb; prgb = &(*prgb)->NextReady) {
+			for(palpha = &s->ReadyAlpha; *palpha; palpha = &(*palpha)->NextReady) {
+				int rgbip = *prgb - s->Instructions;
+				int alphaip = *palpha - s->Instructions;
+				_mesa_bzero(&pair, sizeof(pair));
+				fill_instruction_into_pair(s, &pair, rgbip);
+				if (!fill_instruction_into_pair(s, &pair, alphaip))
+					continue;
+				*prgb = (*prgb)->NextReady;
+				*palpha = (*palpha)->NextReady;
+				fill_dest_into_pair(s, &pair, rgbip);
+				fill_dest_into_pair(s, &pair, alphaip);
+				commit_instruction(s, rgbip);
+				commit_instruction(s, alphaip);
+				goto success;
+			}
+		}
+
+		/* No success in pairing; just take the first RGB instruction */
+		int ip = s->ReadyRGB - s->Instructions;
+		s->ReadyRGB = s->ReadyRGB->NextReady;
+		_mesa_bzero(&pair, sizeof(pair));
+		fill_instruction_into_pair(s, &pair, ip);
+		fill_dest_into_pair(s, &pair, ip);
+		commit_instruction(s, ip);
+	success: ;
+	}
+
+	if (s->Debug)
+		radeonPrintPairInstruction(&pair);
+
+	s->Error = s->Error || !s->Handler->EmitPaired(s->UserData, &pair);
+}
+
+
+GLboolean radeonPairProgram(GLcontext *ctx, struct gl_program *program,
+	const struct radeon_pair_handler* handler, void *userdata)
+{
+	struct pair_state s;
+
+	_mesa_bzero(&s, sizeof(s));
+	s.Ctx = ctx;
+	s.Program = program;
+	s.Handler = handler;
+	s.UserData = userdata;
+	s.Debug = (RADEON_DEBUG & DEBUG_PIXEL) ? GL_TRUE : GL_FALSE;
+	s.Verbose = GL_FALSE && s.Debug;
+
+	s.Instructions = (struct pair_state_instruction*)_mesa_calloc(
+		sizeof(struct pair_state_instruction)*s.Program->NumInstructions);
+	s.ValuePool = (struct reg_value*)_mesa_calloc(sizeof(struct reg_value)*s.Program->NumInstructions*4);
+	s.ReaderPool = (struct reg_value_reader*)_mesa_calloc(
+		sizeof(struct reg_value_reader)*s.Program->NumInstructions*12);
+
+	if (s.Debug)
+		_mesa_printf("Emit paired program\n");
+
+	scan_instructions(&s);
+	allocate_input_registers(&s);
+
+	while(!s.Error &&
+	      (s.ReadyTEX || s.ReadyRGB || s.ReadyAlpha || s.ReadyFullALU)) {
+		if (s.ReadyTEX)
+			emit_all_tex(&s);
+
+		while(s.ReadyFullALU || s.ReadyRGB || s.ReadyAlpha)
+			emit_alu(&s);
+	}
+
+	if (s.Debug)
+		_mesa_printf(" END\n");
+
+	_mesa_free(s.Instructions);
+	_mesa_free(s.ValuePool);
+	_mesa_free(s.ReaderPool);
+
+	return !s.Error;
+}
+
+
+static void print_pair_src(int i, struct radeon_pair_instruction_source* src)
+{
+	_mesa_printf("  Src%i = %s[%i]", i, src->Constant ? "CNST" : "TEMP", src->Index);
+}
+
+static const char* opcode_string(GLuint opcode)
+{
+	if (opcode == OPCODE_REPL_ALPHA)
+		return "SOP";
+	else
+		return _mesa_opcode_string(opcode);
+}
+
+static int num_pairinst_args(GLuint opcode)
+{
+	if (opcode == OPCODE_REPL_ALPHA)
+		return 0;
+	else
+		return _mesa_num_inst_src_regs(opcode);
+}
+
+static char swizzle_char(GLuint swz)
+{
+	switch(swz) {
+	case SWIZZLE_X: return 'x';
+	case SWIZZLE_Y: return 'y';
+	case SWIZZLE_Z: return 'z';
+	case SWIZZLE_W: return 'w';
+	case SWIZZLE_ZERO: return '0';
+	case SWIZZLE_ONE: return '1';
+	case SWIZZLE_NIL: return '_';
+	default: return '?';
+	}
+}
+
+void radeonPrintPairInstruction(struct radeon_pair_instruction *inst)
+{
+	int nargs;
+	int i;
+
+	_mesa_printf("       RGB:  ");
+	for(i = 0; i < 3; ++i) {
+		if (inst->RGB.Src[i].Used)
+			print_pair_src(i, inst->RGB.Src + i);
+	}
+	_mesa_printf("\n");
+	_mesa_printf("       Alpha:");
+	for(i = 0; i < 3; ++i) {
+		if (inst->Alpha.Src[i].Used)
+			print_pair_src(i, inst->Alpha.Src + i);
+	}
+	_mesa_printf("\n");
+
+	_mesa_printf("  %s%s", opcode_string(inst->RGB.Opcode), inst->RGB.Saturate ? "_SAT" : "");
+	if (inst->RGB.WriteMask)
+		_mesa_printf(" TEMP[%i].%s%s%s", inst->RGB.DestIndex,
+			(inst->RGB.WriteMask & 1) ? "x" : "",
+			(inst->RGB.WriteMask & 2) ? "y" : "",
+			(inst->RGB.WriteMask & 4) ? "z" : "");
+	if (inst->RGB.OutputWriteMask)
+		_mesa_printf(" COLOR.%s%s%s",
+			(inst->RGB.OutputWriteMask & 1) ? "x" : "",
+			(inst->RGB.OutputWriteMask & 2) ? "y" : "",
+			(inst->RGB.OutputWriteMask & 4) ? "z" : "");
+	nargs = num_pairinst_args(inst->RGB.Opcode);
+	for(i = 0; i < nargs; ++i) {
+		const char* abs = inst->RGB.Arg[i].Abs ? "|" : "";
+		const char* neg = inst->RGB.Arg[i].Negate ? "-" : "";
+		_mesa_printf(", %s%sSrc%i.%c%c%c%s", neg, abs, inst->RGB.Arg[i].Source,
+			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 0)),
+			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 1)),
+			swizzle_char(GET_SWZ(inst->RGB.Arg[i].Swizzle, 2)),
+			abs);
+	}
+	_mesa_printf("\n");
+
+	_mesa_printf("  %s%s", opcode_string(inst->Alpha.Opcode), inst->Alpha.Saturate ? "_SAT" : "");
+	if (inst->Alpha.WriteMask)
+		_mesa_printf(" TEMP[%i].w", inst->Alpha.DestIndex);
+	if (inst->Alpha.OutputWriteMask)
+		_mesa_printf(" COLOR.w");
+	if (inst->Alpha.DepthWriteMask)
+		_mesa_printf(" DEPTH.w");
+	nargs = num_pairinst_args(inst->Alpha.Opcode);
+	for(i = 0; i < nargs; ++i) {
+		const char* abs = inst->Alpha.Arg[i].Abs ? "|" : "";
+		const char* neg = inst->Alpha.Arg[i].Negate ? "-" : "";
+		_mesa_printf(", %s%sSrc%i.%c%s", neg, abs, inst->Alpha.Arg[i].Source,
+			swizzle_char(inst->Alpha.Arg[i].Swizzle), abs);
+	}
+	_mesa_printf("\n");
+}
diff --git a/src/mesa/drivers/dri/r300/radeon_program_pair.h b/src/mesa/drivers/dri/r300/radeon_program_pair.h
new file mode 100644
index 00000000000..4624a246298
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_program_pair.h
@@ -0,0 +1,126 @@
+/*
+ * Copyright (C) 2008 Nicolai Haehnle.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __RADEON_PROGRAM_PAIR_H_
+#define __RADEON_PROGRAM_PAIR_H_
+
+#include "radeon_program.h"
+
+
+/**
+ * Represents a paired instruction, as found in R300 and R500
+ * fragment programs.
+ */
+struct radeon_pair_instruction_source {
+	GLuint Index:8;
+	GLuint Constant:1;
+	GLuint Used:1;
+};
+
+struct radeon_pair_instruction_rgb {
+	GLuint Opcode:8;
+	GLuint DestIndex:8;
+	GLuint WriteMask:3;
+	GLuint OutputWriteMask:3;
+	GLuint Saturate:1;
+
+	struct radeon_pair_instruction_source Src[3];
+
+	struct {
+		GLuint Source:2;
+		GLuint Swizzle:9;
+		GLuint Abs:1;
+		GLuint Negate:1;
+	} Arg[3];
+};
+
+struct radeon_pair_instruction_alpha {
+	GLuint Opcode:8;
+	GLuint DestIndex:8;
+	GLuint WriteMask:1;
+	GLuint OutputWriteMask:1;
+	GLuint DepthWriteMask:1;
+	GLuint Saturate:1;
+
+	struct radeon_pair_instruction_source Src[3];
+
+	struct {
+		GLuint Source:2;
+		GLuint Swizzle:3;
+		GLuint Abs:1;
+		GLuint Negate:1;
+	} Arg[3];
+};
+
+struct radeon_pair_instruction {
+	struct radeon_pair_instruction_rgb RGB;
+	struct radeon_pair_instruction_alpha Alpha;
+};
+
+
+/**
+ *
+ */
+struct radeon_pair_handler {
+	/**
+	 * Fill in the proper hardware index for the given constant register.
+	 *
+	 * @return GL_FALSE on error.
+	 */
+	GLboolean (*EmitConst)(void*, GLuint file, GLuint index, GLuint *hwindex);
+
+	/**
+	 * Write a paired instruction to the hardware.
+	 *
+	 * @return GL_FALSE on error.
+	 */
+	GLboolean (*EmitPaired)(void*, struct radeon_pair_instruction*);
+
+	/**
+	 * Write a texture instruction to the hardware.
+	 * Register indices have already been rewritten to the allocated
+	 * hardware register numbers.
+	 *
+	 * @return GL_FALSE on error.
+	 */
+	GLboolean (*EmitTex)(void*, struct prog_instruction*);
+
+	/**
+	 * Called before a block of contiguous, independent texture
+	 * instructions is emitted.
+	 */
+	GLboolean (*BeginTexBlock)(void*);
+
+	GLuint MaxHwTemps;
+};
+
+GLboolean radeonPairProgram(GLcontext *ctx, struct gl_program *program,
+	const struct radeon_pair_handler*, void *userdata);
+
+void radeonPrintPairInstruction(struct radeon_pair_instruction *inst);
+
+#endif /* __RADEON_PROGRAM_PAIR_H_ */
diff --git a/src/mesa/drivers/dri/r300/radeon_span.c b/src/mesa/drivers/dri/r300/radeon_span.c
index eae09d6b35e..16f9fb99e67 100644
--- a/src/mesa/drivers/dri/r300/radeon_span.c
+++ b/src/mesa/drivers/dri/r300/radeon_span.c
@@ -40,7 +40,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "swrast/swrast.h"
 
 #include "r300_state.h"
@@ -172,6 +172,8 @@ radeon_mba_z16(const driRenderbuffer * drb, GLint x, GLint y)
 
 /* 16-bit depth buffer functions
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)(buf + radeon_mba_z16( drb, _x + xo, _y + yo )) = d;
 
@@ -186,6 +188,8 @@ radeon_mba_z16(const driRenderbuffer * drb, GLint x, GLint y)
  * Careful: It looks like the R300 uses ZZZS byte order while the R200
  * uses SZZZ for 24 bit depth, 8 bit stencil mode.
  */
+#define VALUE_TYPE GLuint
+
 #ifdef COMPILE_R300
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
@@ -282,6 +286,30 @@ static void radeonSpanRenderStart(GLcontext * ctx)
 #endif
 	LOCK_HARDWARE(rmesa);
 	radeonWaitForIdleLocked(rmesa);
+
+	/* Read the first pixel in the frame buffer.  This should
+	 * be a noop, right?  In fact without this conform fails as reading
+	 * from the framebuffer sometimes produces old results -- the
+	 * on-card read cache gets mixed up and doesn't notice that the
+	 * framebuffer has been updated.
+	 *
+	 * Note that we should probably be reading some otherwise unused
+	 * region of VRAM, otherwise we might get incorrect results when
+	 * reading pixels from the top left of the screen.
+	 *
+	 * I found this problem on an R420 with glean's texCube test.
+	 * Note that the R200 span code also *writes* the first pixel in the
+	 * framebuffer, but I've found this to be unnecessary.
+	 *  -- Nicolai Hähnle, June 2008
+	 */
+	{
+		int p;
+		driRenderbuffer *drb =
+			(driRenderbuffer *) ctx->WinSysDrawBuffer->_ColorDrawBuffers[0];
+		volatile int *buf =
+			(volatile int *)(rmesa->dri.screen->pFB + drb->offset);
+		p = *buf;
+	}
 }
 
 static void radeonSpanRenderFinish(GLcontext * ctx)
diff --git a/src/mesa/drivers/dri/r300/radeon_state.c b/src/mesa/drivers/dri/r300/radeon_state.c
index 82bfd951b92..c401da6c544 100644
--- a/src/mesa/drivers/dri/r300/radeon_state.c
+++ b/src/mesa/drivers/dri/r300/radeon_state.c
@@ -33,12 +33,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "api_arrayelt.h"
-#include "enums.h"
-#include "colormac.h"
-#include "light.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/api_arrayelt.h"
+#include "main/enums.h"
+#include "main/framebuffer.h"
+#include "main/colormac.h"
+#include "main/light.h"
 
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
@@ -49,7 +50,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_ioctl.h"
 #include "radeon_state.h"
 #include "r300_ioctl.h"
-#include "framebuffer.h"
+
 
 /* =============================================================
  * Scissoring
@@ -125,8 +126,8 @@ void radeonUpdateScissor(GLcontext* ctx)
 
 		radeon->state.scissor.rect.x1 = x1;
 		radeon->state.scissor.rect.y1 = y1;
-		radeon->state.scissor.rect.x2 = x1 + ctx->Scissor.Width - 1;
-		radeon->state.scissor.rect.y2 = y1 + ctx->Scissor.Height - 1;
+		radeon->state.scissor.rect.x2 = x1 + ctx->Scissor.Width;
+		radeon->state.scissor.rect.y2 = y1 + ctx->Scissor.Height;
 
 		radeonRecalcScissorRects(radeon);
 	}
@@ -152,7 +153,7 @@ void radeonSetCliprects(radeonContextPtr radeon)
 	GLframebuffer *const draw_fb = (GLframebuffer*)drawable->driverPrivate;
 	GLframebuffer *const read_fb = (GLframebuffer*)readable->driverPrivate;
 
-	if (draw_fb->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT) {
+	if (draw_fb->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT) {
 		/* Can't ignore 2d windows if we are page flipping. */
 		if (drawable->numBackClipRects == 0 || radeon->doPageFlip ||
 		    radeon->sarea->pfCurrentPage == 1) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_chipset.h b/src/mesa/drivers/dri/radeon/radeon_chipset.h
index 7318099093a..55a73eab209 100644
--- a/src/mesa/drivers/dri/radeon/radeon_chipset.h
+++ b/src/mesa/drivers/dri/radeon/radeon_chipset.h
@@ -106,6 +106,7 @@
 #define PCI_CHIP_RV410_564F		0x564F
 #define PCI_CHIP_RV410_5652		0x5652
 #define PCI_CHIP_RV410_5653		0x5653
+#define PCI_CHIP_RV410_5657		0x5657
 #define PCI_CHIP_RS300_5834		0x5834
 #define PCI_CHIP_RS300_5835		0x5835
 #define PCI_CHIP_RS480_5954		0x5954
@@ -145,8 +146,112 @@
 #define PCI_CHIP_RV410_5E4C		0x5E4C
 #define PCI_CHIP_RV410_5E4D		0x5E4D
 #define PCI_CHIP_RV410_5E4F		0x5E4F
+
+#define PCI_CHIP_R520_7100              0x7100
+#define PCI_CHIP_R520_7101              0x7101
+#define PCI_CHIP_R520_7102              0x7102
+#define PCI_CHIP_R520_7103              0x7103
+#define PCI_CHIP_R520_7104              0x7104
+#define PCI_CHIP_R520_7105              0x7105
+#define PCI_CHIP_R520_7106              0x7106
+#define PCI_CHIP_R520_7108              0x7108
+#define PCI_CHIP_R520_7109              0x7109
+#define PCI_CHIP_R520_710A              0x710A
+#define PCI_CHIP_R520_710B              0x710B
+#define PCI_CHIP_R520_710C              0x710C
+#define PCI_CHIP_R520_710E              0x710E
+#define PCI_CHIP_R520_710F              0x710F
+#define PCI_CHIP_RV515_7140             0x7140
+#define PCI_CHIP_RV515_7141             0x7141
+#define PCI_CHIP_RV515_7142             0x7142
+#define PCI_CHIP_RV515_7143             0x7143
+#define PCI_CHIP_RV515_7144             0x7144
+#define PCI_CHIP_RV515_7145             0x7145
+#define PCI_CHIP_RV515_7146             0x7146
+#define PCI_CHIP_RV515_7147             0x7147
+#define PCI_CHIP_RV515_7149             0x7149
+#define PCI_CHIP_RV515_714A             0x714A
+#define PCI_CHIP_RV515_714B             0x714B
+#define PCI_CHIP_RV515_714C             0x714C
+#define PCI_CHIP_RV515_714D             0x714D
+#define PCI_CHIP_RV515_714E             0x714E
+#define PCI_CHIP_RV515_714F             0x714F
+#define PCI_CHIP_RV515_7151             0x7151
+#define PCI_CHIP_RV515_7152             0x7152
+#define PCI_CHIP_RV515_7153             0x7153
+#define PCI_CHIP_RV515_715E             0x715E
+#define PCI_CHIP_RV515_715F             0x715F
+#define PCI_CHIP_RV515_7180             0x7180
+#define PCI_CHIP_RV515_7181             0x7181
+#define PCI_CHIP_RV515_7183             0x7183
+#define PCI_CHIP_RV515_7186             0x7186
+#define PCI_CHIP_RV515_7187             0x7187
+#define PCI_CHIP_RV515_7188             0x7188
+#define PCI_CHIP_RV515_718A             0x718A
+#define PCI_CHIP_RV515_718B             0x718B
+#define PCI_CHIP_RV515_718C             0x718C
+#define PCI_CHIP_RV515_718D             0x718D
+#define PCI_CHIP_RV515_718F             0x718F
+#define PCI_CHIP_RV515_7193             0x7193
+#define PCI_CHIP_RV515_7196             0x7196
+#define PCI_CHIP_RV515_719B             0x719B
+#define PCI_CHIP_RV515_719F             0x719F
+#define PCI_CHIP_RV530_71C0             0x71C0
+#define PCI_CHIP_RV530_71C1             0x71C1
+#define PCI_CHIP_RV530_71C2             0x71C2
+#define PCI_CHIP_RV530_71C3             0x71C3
+#define PCI_CHIP_RV530_71C4             0x71C4
+#define PCI_CHIP_RV530_71C5             0x71C5
+#define PCI_CHIP_RV530_71C6             0x71C6
+#define PCI_CHIP_RV530_71C7             0x71C7
+#define PCI_CHIP_RV530_71CD             0x71CD
+#define PCI_CHIP_RV530_71CE             0x71CE
+#define PCI_CHIP_RV530_71D2             0x71D2
+#define PCI_CHIP_RV530_71D4             0x71D4
+#define PCI_CHIP_RV530_71D5             0x71D5
+#define PCI_CHIP_RV530_71D6             0x71D6
+#define PCI_CHIP_RV530_71DA             0x71DA
+#define PCI_CHIP_RV530_71DE             0x71DE
+#define PCI_CHIP_RV515_7200             0x7200
+#define PCI_CHIP_RV515_7210             0x7210
+#define PCI_CHIP_RV515_7211             0x7211
+#define PCI_CHIP_R580_7240              0x7240
+#define PCI_CHIP_R580_7243              0x7243
+#define PCI_CHIP_R580_7244              0x7244
+#define PCI_CHIP_R580_7245              0x7245
+#define PCI_CHIP_R580_7246              0x7246
+#define PCI_CHIP_R580_7247              0x7247
+#define PCI_CHIP_R580_7248              0x7248
+#define PCI_CHIP_R580_7249              0x7249
+#define PCI_CHIP_R580_724A              0x724A
+#define PCI_CHIP_R580_724B              0x724B
+#define PCI_CHIP_R580_724C              0x724C
+#define PCI_CHIP_R580_724D              0x724D
+#define PCI_CHIP_R580_724E              0x724E
+#define PCI_CHIP_R580_724F              0x724F
+#define PCI_CHIP_RV570_7280             0x7280
+#define PCI_CHIP_RV560_7281             0x7281
+#define PCI_CHIP_RV560_7283             0x7283
+#define PCI_CHIP_R580_7284              0x7284
+#define PCI_CHIP_RV560_7287             0x7287
+#define PCI_CHIP_RV570_7288             0x7288
+#define PCI_CHIP_RV570_7289             0x7289
+#define PCI_CHIP_RV570_728B             0x728B
+#define PCI_CHIP_RV570_728C             0x728C
+#define PCI_CHIP_RV560_7290             0x7290
+#define PCI_CHIP_RV560_7291             0x7291
+#define PCI_CHIP_RV560_7293             0x7293
+#define PCI_CHIP_RV560_7297             0x7297
+
 #define PCI_CHIP_RS350_7834		0x7834
 #define PCI_CHIP_RS350_7835		0x7835
+#define PCI_CHIP_RS690_791E             0x791E
+#define PCI_CHIP_RS690_791F             0x791F
+#define PCI_CHIP_RS740_796C             0x796C
+#define PCI_CHIP_RS740_796D             0x796D
+#define PCI_CHIP_RS740_796E             0x796E
+#define PCI_CHIP_RS740_796F             0x796F
+
 
 enum {
    CHIP_FAMILY_R100,
@@ -165,6 +270,14 @@ enum {
    CHIP_FAMILY_R420,
    CHIP_FAMILY_RV410,
    CHIP_FAMILY_RS400,
+   CHIP_FAMILY_RS690,
+   CHIP_FAMILY_RS740,
+   CHIP_FAMILY_RV515,
+   CHIP_FAMILY_R520,
+   CHIP_FAMILY_RV530,
+   CHIP_FAMILY_R580,
+   CHIP_FAMILY_RV560,
+   CHIP_FAMILY_RV570,
    CHIP_FAMILY_LAST
 };
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_compat.c b/src/mesa/drivers/dri/radeon/radeon_compat.c
index bd467fb15b0..46b490d61f7 100644
--- a/src/mesa/drivers/dri/radeon/radeon_compat.c
+++ b/src/mesa/drivers/dri/radeon/radeon_compat.c
@@ -32,8 +32,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  */
 
-#include "glheader.h"
-#include "imports.h"
+#include "main/glheader.h"
+#include "main/imports.h"
 
 #include "radeon_context.h"
 #include "radeon_state.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index ba93a054ae2..1e992c0b3d6 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -34,14 +34,15 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
-#include "framebuffer.h"
+#include "main/glheader.h"
+#include "main/api_arrayelt.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/state.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -231,14 +232,14 @@ radeonCreateContext( const __GLcontextModes *glVisual,
                                                  "def_max_anisotropy");
 
    if ( driQueryOptionb( &rmesa->optionCache, "hyperz" ) ) {
-      if ( sPriv->drmMinor < 13 )
+      if ( sPriv->drm_version.minor < 13 )
 	 fprintf( stderr, "DRM version 1.%d too old to support HyperZ, "
-			  "disabling.\n",sPriv->drmMinor );
+			  "disabling.\n", sPriv->drm_version.minor );
       else
 	 rmesa->using_hyperz = GL_TRUE;
    }
 
-   if ( sPriv->drmMinor >= 15 )
+   if ( sPriv->drm_version.minor >= 15 )
       rmesa->texmicrotile = GL_TRUE;
 
    /* Init default driver functions then plug in our Radeon-specific functions
@@ -269,7 +270,7 @@ radeonCreateContext( const __GLcontextModes *glVisual,
    rmesa->dri.hwContext = driContextPriv->hHWContext;
    rmesa->dri.hwLock = &sPriv->pSAREA->lock;
    rmesa->dri.fd = sPriv->fd;
-   rmesa->dri.drmMinor = sPriv->drmMinor;
+   rmesa->dri.drmMinor = sPriv->drm_version.minor;
 
    rmesa->radeonScreen = screen;
    rmesa->sarea = (drm_radeon_sarea_t *)((GLubyte *)sPriv->pSAREA +
@@ -422,10 +423,7 @@ radeonCreateContext( const __GLcontextModes *glVisual,
 
    rmesa->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
 
-   rmesa->vblank_flags = (rmesa->radeonScreen->irq != 0)
-       ? driGetDefaultVBlankFlags(&rmesa->optionCache) : VBLANK_FLAG_NO_IRQ;
-
-   (*dri_interface->getUST)( & rmesa->swap_ust );
+   (*sPriv->systemTime->getUST)( & rmesa->swap_ust );
 
 
 #if DO_DEBUG
@@ -590,16 +588,18 @@ radeonMakeCurrent( __DRIcontextPrivate *driContextPriv,
       if (RADEON_DEBUG & DEBUG_DRI)
 	 fprintf(stderr, "%s ctx %p\n", __FUNCTION__, (void *) newCtx->glCtx);
 
-      if ( newCtx->dri.drawable != driDrawPriv ) {
-         /* XXX we may need to validate the drawable here!!! */
-	 driDrawableInitVBlank( driDrawPriv, newCtx->vblank_flags,
-				&newCtx->vbl_seq );
-      }
-
       newCtx->dri.readable = driReadPriv;
 
       if ( (newCtx->dri.drawable != driDrawPriv) ||
            newCtx->lastStamp != driDrawPriv->lastStamp ) {
+	 if (driDrawPriv->swap_interval == (unsigned)-1) {
+	    driDrawPriv->vblFlags = (newCtx->radeonScreen->irq != 0)
+	       ? driGetDefaultVBlankFlags(&newCtx->optionCache)
+	       : VBLANK_FLAG_NO_IRQ;
+
+	    driDrawableInitVBlank( driDrawPriv );
+	 }
+
 	 newCtx->dri.drawable = driDrawPriv;
 
 	 radeonSetCliprects(newCtx);
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.h b/src/mesa/drivers/dri/radeon/radeon_context.h
index 8dedd66f563..53df766f8cb 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -49,9 +49,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_drm.h"
 #include "texmem.h"
 
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
 
 struct radeon_context;
 typedef struct radeon_context radeonContextRec;
@@ -66,7 +66,7 @@ typedef union {
 
 #include "radeon_lock.h"
 #include "radeon_screen.h"
-#include "mm.h"
+#include "main/mm.h"
 
 #include "math/m_vector.h"
 
@@ -161,6 +161,8 @@ struct radeon_tex_obj {
 	drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
 	/* Six, for the cube faces */
 
+	GLboolean image_override; /* Image overridden by GLX_EXT_tfp */
+
 	GLuint pp_txfilter;	/* hardware register values */
 	GLuint pp_txformat;
 	GLuint pp_txoffset;	/* Image location in texmem.
@@ -667,9 +669,6 @@ struct radeon_context {
 
 	/* VBI
 	 */
-	GLuint vbl_seq;
-	GLuint vblank_flags;
-
 	int64_t swap_ust;
 	int64_t swap_missed_ust;
 
@@ -708,9 +707,9 @@ struct radeon_context {
 
 #define RADEON_CONTEXT(ctx)		((radeonContextPtr)(ctx->DriverCtx))
 
-static __inline GLuint radeonPackColor(GLuint cpp,
-				       GLubyte r, GLubyte g,
-				       GLubyte b, GLubyte a)
+static INLINE GLuint radeonPackColor(GLuint cpp,
+                                     GLubyte r, GLubyte g,
+                                     GLubyte b, GLubyte a)
 {
 	switch (cpp) {
 	case 2:
diff --git a/src/mesa/drivers/dri/radeon/radeon_ioctl.c b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
index a0c563c7cba..09acf6b4f85 100644
--- a/src/mesa/drivers/dri/radeon/radeon_ioctl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
@@ -37,9 +37,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <sched.h>
 #include <errno.h> 
 
-#include "glheader.h"
-#include "imports.h"
-#include "simple_list.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
 #include "swrast/swrast.h"
 
 #include "radeon_context.h"
@@ -862,13 +862,14 @@ static void radeonWaitForFrameCompletion( radeonContextPtr rmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void radeonCopyBuffer( const __DRIdrawablePrivate *dPriv,
+void radeonCopyBuffer( __DRIdrawablePrivate *dPriv,
 		       const drm_clip_rect_t	  *rect)
 {
    radeonContextPtr rmesa;
    GLint nbox, i, ret;
    GLboolean   missed_target;
    int64_t ust;
+   __DRIscreenPrivate *psp;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
@@ -890,7 +891,7 @@ void radeonCopyBuffer( const __DRIdrawablePrivate *dPriv,
    if (!rect)
    {
        UNLOCK_HARDWARE( rmesa );
-       driWaitForVBlank( dPriv, & rmesa->vbl_seq, rmesa->vblank_flags, & missed_target );
+       driWaitForVBlank( dPriv, & missed_target );
        LOCK_HARDWARE( rmesa );
    }
 
@@ -917,16 +918,18 @@ void radeonCopyBuffer( const __DRIdrawablePrivate *dPriv,
 	      if (rect->y2 < b->y2)
 		  b->y2 = rect->y2;
 
-	      if (b->x1 < b->x2 && b->y1 < b->y2)
-		  b++;
+	      if (b->x1 >= b->x2 || b->y1 >= b->y2)
+		  continue;
 	  }
-	  else
-	      b++;
 
+	  b++;
 	  n++;
       }
       rmesa->sarea->nbox = n;
 
+      if (!n)
+	 continue;
+
       ret = drmCommandNone( rmesa->dri.fd, DRM_RADEON_SWAP );
 
       if ( ret ) {
@@ -939,8 +942,9 @@ void radeonCopyBuffer( const __DRIdrawablePrivate *dPriv,
    UNLOCK_HARDWARE( rmesa );
    if (!rect)
    {
+       psp = dPriv->driScreenPriv;
        rmesa->swap_count++;
-       (*dri_interface->getUST)( & ust );
+       (*psp->systemTime->getUST)( & ust );
        if ( missed_target ) {
 	   rmesa->swap_missed_count++;
 	   rmesa->swap_missed_ust = ust - rmesa->swap_ust;
@@ -951,17 +955,19 @@ void radeonCopyBuffer( const __DRIdrawablePrivate *dPriv,
    }
 }
 
-void radeonPageFlip( const __DRIdrawablePrivate *dPriv )
+void radeonPageFlip( __DRIdrawablePrivate *dPriv )
 {
    radeonContextPtr rmesa;
    GLint ret;
    GLboolean   missed_target;
+   __DRIscreenPrivate *psp;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
    assert(dPriv->driContextPriv->driverPrivate);
 
    rmesa = (radeonContextPtr) dPriv->driContextPriv->driverPrivate;
+   psp = dPriv->driScreenPriv;
 
    if ( RADEON_DEBUG & DEBUG_IOCTL ) {
       fprintf(stderr, "%s: pfCurrentPage: %d\n", __FUNCTION__,
@@ -986,10 +992,10 @@ void radeonPageFlip( const __DRIdrawablePrivate *dPriv )
     */
    radeonWaitForFrameCompletion( rmesa );
    UNLOCK_HARDWARE( rmesa );
-   driWaitForVBlank( dPriv, & rmesa->vbl_seq, rmesa->vblank_flags, & missed_target );
+   driWaitForVBlank( dPriv, & missed_target );
    if ( missed_target ) {
       rmesa->swap_missed_count++;
-      (void) (*dri_interface->getUST)( & rmesa->swap_missed_ust );
+      (void) (*psp->systemTime->getUST)( & rmesa->swap_missed_ust );
    }
    LOCK_HARDWARE( rmesa );
 
@@ -1003,7 +1009,7 @@ void radeonPageFlip( const __DRIdrawablePrivate *dPriv )
    }
 
    rmesa->swap_count++;
-   (void) (*dri_interface->getUST)( & rmesa->swap_ust );
+   (void) (*psp->systemTime->getUST)( & rmesa->swap_ust );
 
    /* Get ready for drawing next frame.  Update the renderbuffers'
     * flippedOffset/Pitch fields so we draw into the right place.
diff --git a/src/mesa/drivers/dri/radeon/radeon_ioctl.h b/src/mesa/drivers/dri/radeon/radeon_ioctl.h
index 020a5c21e2d..4e3a44df075 100644
--- a/src/mesa/drivers/dri/radeon/radeon_ioctl.h
+++ b/src/mesa/drivers/dri/radeon/radeon_ioctl.h
@@ -36,7 +36,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __RADEON_IOCTL_H__
 #define __RADEON_IOCTL_H__
 
-#include "simple_list.h"
+#include "main/simple_list.h"
 #include "radeon_lock.h"
 
 
@@ -86,9 +86,9 @@ extern void radeonReleaseDmaRegion( radeonContextPtr rmesa,
 				    struct radeon_dma_region *region,
 				    const char *caller );
 
-extern void radeonCopyBuffer( const __DRIdrawablePrivate *drawable,
+extern void radeonCopyBuffer( __DRIdrawablePrivate *drawable,
 			      const drm_clip_rect_t	 *rect);
-extern void radeonPageFlip( const __DRIdrawablePrivate *drawable );
+extern void radeonPageFlip( __DRIdrawablePrivate *drawable );
 extern void radeonFlush( GLcontext *ctx );
 extern void radeonFinish( GLcontext *ctx );
 extern void radeonWaitForIdleLocked( radeonContextPtr rmesa );
@@ -123,7 +123,7 @@ do {								\
    memcpy( rmesa->hw.ATOM.lastcmd, rmesa->hw.ATOM.cmd,	\
 	   rmesa->hw.ATOM.cmd_size * 4)
 
-static __inline int RADEON_DB_STATECHANGE( 
+static INLINE int RADEON_DB_STATECHANGE( 
    radeonContextPtr rmesa,
    struct radeon_state_atom *atom )
 {
@@ -176,7 +176,7 @@ do {							\
  * and hang on to the lock until the critical section is finished and we flush
  * the buffer again and unlock.
  */
-static __inline void radeonEnsureCmdBufSpace( radeonContextPtr rmesa,
+static INLINE void radeonEnsureCmdBufSpace( radeonContextPtr rmesa,
 					      int bytes )
 {
    if (rmesa->store.cmd_used + bytes > RADEON_CMD_BUF_SZ)
@@ -186,7 +186,7 @@ static __inline void radeonEnsureCmdBufSpace( radeonContextPtr rmesa,
 
 /* Alloc space in the command buffer
  */
-static __inline char *radeonAllocCmdBuf( radeonContextPtr rmesa,
+static INLINE char *radeonAllocCmdBuf( radeonContextPtr rmesa,
 					 int bytes, const char *where )
 {
    if (rmesa->store.cmd_used + bytes > RADEON_CMD_BUF_SZ)
diff --git a/src/mesa/drivers/dri/radeon/radeon_lighting.c b/src/mesa/drivers/dri/radeon/radeon_lighting.c
index 5e9b9c3051b..6d9ccfa24d6 100644
--- a/src/mesa/drivers/dri/radeon/radeon_lighting.c
+++ b/src/mesa/drivers/dri/radeon/radeon_lighting.c
@@ -27,11 +27,11 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
+#include "main/glheader.h"
+#include "main/imports.h"
 #include "api_arrayelt.h"
 /* #include "mmath.h" */
-#include "enums.h"
+#include "main/enums.h"
 #include "colormac.h"
 
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_lock.c b/src/mesa/drivers/dri/radeon/radeon_lock.c
index 30a0c3863c2..64bb3ca103f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_lock.c
+++ b/src/mesa/drivers/dri/radeon/radeon_lock.c
@@ -39,8 +39,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Kevin E. Martin <martin@valinux.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
 #include "radeon_context.h"
 #include "radeon_lock.h"
 #include "radeon_tex.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
index b61f5e0f3e4..de3c3a15a7f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
@@ -32,15 +32,15 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "math/m_translate.h"
 #include "tnl/tnl.h"
-#include "tnl/t_context.h"
+#include "tnl/tcontext.h"
 
 #include "radeon_context.h"
 #include "radeon_ioctl.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index d5ceedfa24e..126d0727c63 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -32,9 +32,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "vbo/vbo.h"
 #include "math/m_translate.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_sanity.c b/src/mesa/drivers/dri/radeon/radeon_sanity.c
index bdfb7240d72..6613757fcea 100644
--- a/src/mesa/drivers/dri/radeon/radeon_sanity.c
+++ b/src/mesa/drivers/dri/radeon/radeon_sanity.c
@@ -33,7 +33,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 #include <errno.h> 
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 #include "radeon_context.h"
 #include "radeon_ioctl.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 762848fc260..5f32dd575e3 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -35,11 +35,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * \author  Gareth Hughes <gareth@valinux.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #define STANDALONE_MMIO
 #include "radeon_chipset.h"
@@ -48,10 +48,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #if !RADEON_COMMON
 #include "radeon_context.h"
 #include "radeon_span.h"
+#include "radeon_tex.h"
 #elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_span.h"
+#include "r200_tex.h"
 #elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
 #include "r300_context.h"
 #include "r300_fragprog.h"
@@ -60,7 +62,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #endif
 
 #include "utils.h"
-#include "context.h"
 #include "vblank.h"
 #include "drirenderbuffer.h"
 
@@ -88,7 +89,7 @@ DRI_CONF_BEGIN
         DRI_CONF_COLOR_REDUCTION(DRI_CONF_COLOR_REDUCTION_DITHER)
         DRI_CONF_ROUND_MODE(DRI_CONF_ROUND_TRUNC)
         DRI_CONF_DITHER_MODE(DRI_CONF_DITHER_XERRORDIFF)
-        DRI_CONF_ALLOW_LARGE_TEXTURES(1)
+        DRI_CONF_ALLOW_LARGE_TEXTURES(2)
     DRI_CONF_SECTION_END
     DRI_CONF_SECTION_DEBUG
         DRI_CONF_NO_RAST(false)
@@ -115,7 +116,7 @@ DRI_CONF_BEGIN
         DRI_CONF_COLOR_REDUCTION(DRI_CONF_COLOR_REDUCTION_DITHER)
         DRI_CONF_ROUND_MODE(DRI_CONF_ROUND_TRUNC)
         DRI_CONF_DITHER_MODE(DRI_CONF_DITHER_XERRORDIFF)
-        DRI_CONF_ALLOW_LARGE_TEXTURES(1)
+        DRI_CONF_ALLOW_LARGE_TEXTURES(2)
         DRI_CONF_TEXTURE_BLEND_QUALITY(1.0,"0.0:1.0")
     DRI_CONF_SECTION_END
     DRI_CONF_SECTION_DEBUG
@@ -177,7 +178,7 @@ DRI_CONF_OPT_BEGIN_V(fp_optimization,enum,def,"0:1") \
         DRI_CONF_DESC_END \
 DRI_CONF_OPT_END
 
-const char __driConfigOptions[] =
+PUBLIC const char __driConfigOptions[] =
 DRI_CONF_BEGIN
 	DRI_CONF_SECTION_PERFORMANCE
 		DRI_CONF_TCL_MODE(DRI_CONF_TCL_CODEGEN)
@@ -186,14 +187,13 @@ DRI_CONF_BEGIN
 		DRI_CONF_MAX_TEXTURE_IMAGE_UNITS(8, 2, 8)
 		DRI_CONF_MAX_TEXTURE_COORD_UNITS(8, 2, 8)
 		DRI_CONF_COMMAND_BUFFER_SIZE(8, 8, 32)
-		DRI_CONF_DISABLE_FALLBACK(false)
+		DRI_CONF_DISABLE_FALLBACK(true)
 		DRI_CONF_DISABLE_DOUBLE_SIDE_STENCIL(false)
 	DRI_CONF_SECTION_END
 	DRI_CONF_SECTION_QUALITY
 		DRI_CONF_TEXTURE_DEPTH(DRI_CONF_TEXTURE_DEPTH_FB)
 		DRI_CONF_DEF_MAX_ANISOTROPY(1.0, "1.0,2.0,4.0,8.0,16.0")
-		DRI_CONF_NO_NEG_LOD_BIAS(false)
-                DRI_CONF_FORCE_S3TC_ENABLE(false)
+		DRI_CONF_FORCE_S3TC_ENABLE(false)
 		DRI_CONF_DISABLE_S3TC(false)
 		DRI_CONF_COLOR_REDUCTION(DRI_CONF_COLOR_REDUCTION_DITHER)
 		DRI_CONF_ROUND_MODE(DRI_CONF_ROUND_TRUNC)
@@ -204,7 +204,7 @@ DRI_CONF_BEGIN
 		DRI_CONF_NO_RAST(false)
 	DRI_CONF_SECTION_END
 DRI_CONF_END;
-static const GLuint __driNConfigOptions = 18;
+static const GLuint __driNConfigOptions = 17;
 
 #ifndef RADEON_DEBUG
 int RADEON_DEBUG = 0;
@@ -242,25 +242,26 @@ radeonGetParam(int fd, int param, void *value)
 {
   int ret;
   drm_radeon_getparam_t gp;
-  
+
   gp.param = param;
   gp.value = value;
-  
+
   ret = drmCommandWriteRead( fd, DRM_RADEON_GETPARAM, &gp, sizeof(gp));
   return ret;
 }
 
-static __GLcontextModes *
-radeonFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
+static const __DRIconfig **
+radeonFillInModes( __DRIscreenPrivate *psp,
+		   unsigned pixel_bits, unsigned depth_bits,
+		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
-    __GLcontextModes * modes;
-    __GLcontextModes * m;
-    unsigned num_modes;
+    __DRIconfig **configs;
+    __GLcontextModes *m;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     GLenum fb_format;
     GLenum fb_type;
+    int i;
 
     /* Right now GLX_SWAP_COPY_OML isn't supported, but it would be easy
      * enough to add support.  Basically, if a context is created with an
@@ -277,7 +278,7 @@ radeonFillInModes( unsigned pixel_bits, unsigned depth_bits,
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
-    
+
     /* Just like with the accumulation buffer, always provide some modes
      * with a stencil buffer.  It will be a sw fallback, but some apps won't
      * care about that.
@@ -288,8 +289,6 @@ radeonFillInModes( unsigned pixel_bits, unsigned depth_bits,
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -299,21 +298,11 @@ radeonFillInModes( unsigned pixel_bits, unsigned depth_bits,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
-
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR ) ) {
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor,
+			       back_buffer_modes, back_buffer_factor);
+    if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
 	return NULL;
@@ -321,15 +310,43 @@ radeonFillInModes( unsigned pixel_bits, unsigned depth_bits,
 
     /* Mark the visual as slow if there are "fake" stencil bits.
      */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-	if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
+    for (i = 0; configs[i]; i++) {
+	m = &configs[i]->modes;
+	if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
 	    m->visualRating = GLX_SLOW_CONFIG;
 	}
     }
 
-    return modes;
+    return (const __DRIconfig **) configs;
 }
 
+#if !RADEON_COMMON
+static const __DRItexOffsetExtension radeonTexOffsetExtension = {
+    { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
+    radeonSetTexOffset,
+};
+#endif
+
+#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
+static const __DRIallocateExtension r200AllocateExtension = {
+    { __DRI_ALLOCATE, __DRI_ALLOCATE_VERSION },
+    r200AllocateMemoryMESA,
+    r200FreeMemoryMESA,
+    r200GetMemoryOffsetMESA
+};
+
+static const __DRItexOffsetExtension r200texOffsetExtension = {
+    { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
+   r200SetTexOffset,
+};
+#endif
+
+#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+static const __DRItexOffsetExtension r300texOffsetExtension = {
+    { __DRI_TEX_OFFSET, __DRI_TEX_OFFSET_VERSION },
+   r300SetTexOffset,
+};
+#endif
 
 /* Create the device specific screen private data struct.
  */
@@ -339,9 +356,9 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
    radeonScreenPtr screen;
    RADEONDRIPtr dri_priv = (RADEONDRIPtr)sPriv->pDevPriv;
    unsigned char *RADEONMMIO;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
+   int i;
+   int ret;
+   uint32_t temp;
 
    if (sPriv->devPrivSize != sizeof(RADEONDRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(RADEONDRIRec) does not match passed size from device driver\n");
@@ -372,7 +389,7 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
       int ret;
       ret = radeonGetParam( sPriv->fd, RADEON_PARAM_GART_BUFFER_OFFSET,
 			    &screen->gart_buffer_offset);
-	
+
       if (ret) {
 	 FREE( screen );
 	 fprintf(stderr, "drm_radeon_getparam_t (RADEON_PARAM_GART_BUFFER_OFFSET): %d\n", ret);
@@ -394,13 +411,13 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
 	 fprintf(stderr, "drm_radeon_getparam_t (RADEON_PARAM_IRQ_NR): %d\n", ret);
 	 return NULL;
       }
-      screen->drmSupportsCubeMapsR200 = (sPriv->drmMinor >= 7);
-      screen->drmSupportsBlendColor = (sPriv->drmMinor >= 11);
-      screen->drmSupportsTriPerf = (sPriv->drmMinor >= 16);
-      screen->drmSupportsFragShader = (sPriv->drmMinor >= 18);
-      screen->drmSupportsPointSprites = (sPriv->drmMinor >= 13);
-      screen->drmSupportsCubeMapsR100 = (sPriv->drmMinor >= 15);
-      screen->drmSupportsVertexProgram = (sPriv->drmMinor >= 25);
+      screen->drmSupportsCubeMapsR200 = (sPriv->drm_version.minor >= 7);
+      screen->drmSupportsBlendColor = (sPriv->drm_version.minor >= 11);
+      screen->drmSupportsTriPerf = (sPriv->drm_version.minor >= 16);
+      screen->drmSupportsFragShader = (sPriv->drm_version.minor >= 18);
+      screen->drmSupportsPointSprites = (sPriv->drm_version.minor >= 13);
+      screen->drmSupportsCubeMapsR100 = (sPriv->drm_version.minor >= 15);
+      screen->drmSupportsVertexProgram = (sPriv->drm_version.minor >= 25);
    }
 
    screen->mmio.handle = dri_priv->registerHandle;
@@ -534,7 +551,11 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
       screen->chip_family = CHIP_FAMILY_RS300;
       break;
 
+      /* 9500 with 1 pipe verified by: Reid Linnemann <lreid@cs.okstate.edu> */
    case PCI_CHIP_R300_AD:
+      screen->chip_family = CHIP_FAMILY_RV350;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
+      break;
    case PCI_CHIP_R300_AE:
    case PCI_CHIP_R300_AF:
    case PCI_CHIP_R300_AG:
@@ -632,17 +653,18 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
       screen->chip_flags = RADEON_CHIPSET_TCL;
       break;
 
+   case PCI_CHIP_RV410_5E4C:
+   case PCI_CHIP_RV410_5E4F:
    case PCI_CHIP_RV410_564A:
    case PCI_CHIP_RV410_564B:
    case PCI_CHIP_RV410_564F:
    case PCI_CHIP_RV410_5652:
    case PCI_CHIP_RV410_5653:
+   case PCI_CHIP_RV410_5657:
    case PCI_CHIP_RV410_5E48:
    case PCI_CHIP_RV410_5E4A:
    case PCI_CHIP_RV410_5E4B:
-   case PCI_CHIP_RV410_5E4C:
    case PCI_CHIP_RV410_5E4D:
-   case PCI_CHIP_RV410_5E4F:
       screen->chip_family = CHIP_FAMILY_RV410;
       screen->chip_flags = RADEON_CHIPSET_TCL;
       break;
@@ -656,7 +678,132 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
    case PCI_CHIP_RC410_5A61:
    case PCI_CHIP_RC410_5A62:
       screen->chip_family = CHIP_FAMILY_RS400;
-      fprintf(stderr, "Warning, xpress200 detected.\n");
+      break;
+
+   case PCI_CHIP_RS690_791E:
+   case PCI_CHIP_RS690_791F:
+      screen->chip_family = CHIP_FAMILY_RS690;
+      break;
+   case PCI_CHIP_RS740_796C:
+   case PCI_CHIP_RS740_796D:
+   case PCI_CHIP_RS740_796E:
+   case PCI_CHIP_RS740_796F:
+      screen->chip_family = CHIP_FAMILY_RS740;
+      break;
+
+   case PCI_CHIP_R520_7100:
+   case PCI_CHIP_R520_7101:
+   case PCI_CHIP_R520_7102:
+   case PCI_CHIP_R520_7103:
+   case PCI_CHIP_R520_7104:
+   case PCI_CHIP_R520_7105:
+   case PCI_CHIP_R520_7106:
+   case PCI_CHIP_R520_7108:
+   case PCI_CHIP_R520_7109:
+   case PCI_CHIP_R520_710A:
+   case PCI_CHIP_R520_710B:
+   case PCI_CHIP_R520_710C:
+   case PCI_CHIP_R520_710E:
+   case PCI_CHIP_R520_710F:
+      screen->chip_family = CHIP_FAMILY_R520;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
+      break;
+
+   case PCI_CHIP_RV515_7140:
+   case PCI_CHIP_RV515_7141:
+   case PCI_CHIP_RV515_7142:
+   case PCI_CHIP_RV515_7143:
+   case PCI_CHIP_RV515_7144:
+   case PCI_CHIP_RV515_7145:
+   case PCI_CHIP_RV515_7146:
+   case PCI_CHIP_RV515_7147:
+   case PCI_CHIP_RV515_7149:
+   case PCI_CHIP_RV515_714A:
+   case PCI_CHIP_RV515_714B:
+   case PCI_CHIP_RV515_714C:
+   case PCI_CHIP_RV515_714D:
+   case PCI_CHIP_RV515_714E:
+   case PCI_CHIP_RV515_714F:
+   case PCI_CHIP_RV515_7151:
+   case PCI_CHIP_RV515_7152:
+   case PCI_CHIP_RV515_7153:
+   case PCI_CHIP_RV515_715E:
+   case PCI_CHIP_RV515_715F:
+   case PCI_CHIP_RV515_7180:
+   case PCI_CHIP_RV515_7181:
+   case PCI_CHIP_RV515_7183:
+   case PCI_CHIP_RV515_7186:
+   case PCI_CHIP_RV515_7187:
+   case PCI_CHIP_RV515_7188:
+   case PCI_CHIP_RV515_718A:
+   case PCI_CHIP_RV515_718B:
+   case PCI_CHIP_RV515_718C:
+   case PCI_CHIP_RV515_718D:
+   case PCI_CHIP_RV515_718F:
+   case PCI_CHIP_RV515_7193:
+   case PCI_CHIP_RV515_7196:
+   case PCI_CHIP_RV515_719B:
+   case PCI_CHIP_RV515_719F:
+   case PCI_CHIP_RV515_7200:
+   case PCI_CHIP_RV515_7210:
+   case PCI_CHIP_RV515_7211:
+      screen->chip_family = CHIP_FAMILY_RV515;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
+      break;
+
+   case PCI_CHIP_RV530_71C0:
+   case PCI_CHIP_RV530_71C1:
+   case PCI_CHIP_RV530_71C2:
+   case PCI_CHIP_RV530_71C3:
+   case PCI_CHIP_RV530_71C4:
+   case PCI_CHIP_RV530_71C5:
+   case PCI_CHIP_RV530_71C6:
+   case PCI_CHIP_RV530_71C7:
+   case PCI_CHIP_RV530_71CD:
+   case PCI_CHIP_RV530_71CE:
+   case PCI_CHIP_RV530_71D2:
+   case PCI_CHIP_RV530_71D4:
+   case PCI_CHIP_RV530_71D5:
+   case PCI_CHIP_RV530_71D6:
+   case PCI_CHIP_RV530_71DA:
+   case PCI_CHIP_RV530_71DE:
+      screen->chip_family = CHIP_FAMILY_RV530;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
+      break;
+
+   case PCI_CHIP_R580_7240:
+   case PCI_CHIP_R580_7243:
+   case PCI_CHIP_R580_7244:
+   case PCI_CHIP_R580_7245:
+   case PCI_CHIP_R580_7246:
+   case PCI_CHIP_R580_7247:
+   case PCI_CHIP_R580_7248:
+   case PCI_CHIP_R580_7249:
+   case PCI_CHIP_R580_724A:
+   case PCI_CHIP_R580_724B:
+   case PCI_CHIP_R580_724C:
+   case PCI_CHIP_R580_724D:
+   case PCI_CHIP_R580_724E:
+   case PCI_CHIP_R580_724F:
+   case PCI_CHIP_R580_7284:
+      screen->chip_family = CHIP_FAMILY_R580;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
+      break;
+
+   case PCI_CHIP_RV570_7280:
+   case PCI_CHIP_RV560_7281:
+   case PCI_CHIP_RV560_7283:
+   case PCI_CHIP_RV560_7287:
+   case PCI_CHIP_RV570_7288:
+   case PCI_CHIP_RV570_7289:
+   case PCI_CHIP_RV570_728B:
+   case PCI_CHIP_RV570_728C:
+   case PCI_CHIP_RV560_7290:
+   case PCI_CHIP_RV560_7291:
+   case PCI_CHIP_RV560_7293:
+   case PCI_CHIP_RV560_7297:
+      screen->chip_family = CHIP_FAMILY_RV560;
+      screen->chip_flags = RADEON_CHIPSET_TCL;
       break;
 
    default:
@@ -665,11 +812,19 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
       return NULL;
    }
    if ((screen->chip_family == CHIP_FAMILY_R350 || screen->chip_family == CHIP_FAMILY_R300) &&
-       sPriv->ddxMinor < 2) {
+       sPriv->ddx_version.minor < 2) {
       fprintf(stderr, "xf86-video-ati-6.6.2 or newer needed for Radeon 9500/9700/9800 cards.\n");
       return NULL;
    }
 
+   if ((sPriv->drm_version.minor < 29) && (screen->chip_family >= CHIP_FAMILY_RV515)) {
+      fprintf(stderr, "R500 support requires a newer drm.\n");
+      return NULL;
+   }
+
+   if (getenv("R300_NO_TCL"))
+     screen->chip_flags &= ~RADEON_CHIPSET_TCL;
+
    if (screen->chip_family <= CHIP_FAMILY_RS200)
       screen->chip_flags |= RADEON_CLASS_R100;
    else if (screen->chip_family <= CHIP_FAMILY_RV280)
@@ -680,9 +835,51 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
    screen->cpp = dri_priv->bpp / 8;
    screen->AGPMode = dri_priv->AGPMode;
 
-   screen->fbLocation	= ( INREG( RADEON_MC_FB_LOCATION ) & 0xffff ) << 16;
+   ret = radeonGetParam( sPriv->fd, RADEON_PARAM_FB_LOCATION,
+                         &temp);
+   if (ret) {
+       if (screen->chip_family < CHIP_FAMILY_RS690)
+	   screen->fbLocation      = ( INREG( RADEON_MC_FB_LOCATION ) & 0xffff) << 16;
+       else {
+           FREE( screen );
+           fprintf(stderr, "Unable to get fb location need newer drm\n");
+           return NULL;
+       }
+   } else {
+       screen->fbLocation = (temp & 0xffff) << 16;
+   }
+
+   if (screen->chip_family >= CHIP_FAMILY_RV515) {
+       ret = radeonGetParam( sPriv->fd, RADEON_PARAM_NUM_GB_PIPES,
+			     &temp);
+       if (ret) {
+	   fprintf(stderr, "Unable to get num_pipes, need newer drm\n");
+	   switch (screen->chip_family) {
+	   case CHIP_FAMILY_R300:
+	   case CHIP_FAMILY_R350:
+	       screen->num_gb_pipes = 2;
+	       break;
+	   case CHIP_FAMILY_R420:
+	   case CHIP_FAMILY_R520:
+	   case CHIP_FAMILY_R580:
+	   case CHIP_FAMILY_RV560:
+	   case CHIP_FAMILY_RV570:
+	       screen->num_gb_pipes = 4;
+	       break;
+	   case CHIP_FAMILY_RV350:
+	   case CHIP_FAMILY_RV515:
+	   case CHIP_FAMILY_RV530:
+	   case CHIP_FAMILY_RV410:
+	   default:
+	       screen->num_gb_pipes = 1;
+	       break;
+	   }
+       } else {
+	   screen->num_gb_pipes = temp;
+       }
+   }
 
-   if ( sPriv->drmMinor >= 10 ) {
+   if ( sPriv->drm_version.minor >= 10 ) {
       drm_radeon_setparam_t sp;
 
       sp.param = RADEON_SETPARAM_FB_LOCATION;
@@ -700,8 +897,11 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
    screen->depthPitch	= dri_priv->depthPitch;
 
    /* Check if ddx has set up a surface reg to cover depth buffer */
-   screen->depthHasSurface = ((sPriv->ddxMajor > 4) &&
-      (screen->chip_flags & RADEON_CHIPSET_TCL));
+   screen->depthHasSurface = (sPriv->ddx_version.major > 4) ||
+      /* these chips don't use tiled z without hyperz. So always pretend
+         we have set up a surface which will cause linear reads/writes */
+      (IS_R100_CLASS(screen) &&
+      !(screen->chip_flags & RADEON_CHIPSET_TCL));
 
    if ( dri_priv->textureSize == 0 ) {
       screen->texOffset[RADEON_LOCAL_TEX_HEAP] = screen->gart_texture_offset;
@@ -730,29 +930,34 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
 	 dri_priv->log2GARTTexGran;
    }
 
-   if ( glx_enable_extension != NULL ) {
-      if ( screen->irq != 0 ) {
-	 (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-	 (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-	 (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      }
-
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
-      if (IS_R200_CLASS(screen))
-	 (*glx_enable_extension)( psc, "GLX_MESA_allocate_memory" );
+   i = 0;
+   screen->extensions[i++] = &driCopySubBufferExtension.base;
+   screen->extensions[i++] = &driFrameTrackingExtension.base;
+   screen->extensions[i++] = &driReadDrawableExtension;
 
-      (*glx_enable_extension)( psc, "GLX_MESA_copy_sub_buffer" );
-      (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
+   if ( screen->irq != 0 ) {
+       screen->extensions[i++] = &driSwapControlExtension.base;
+       screen->extensions[i++] = &driMediaStreamCounterExtension.base;
    }
 
+#if !RADEON_COMMON
+   screen->extensions[i++] = &radeonTexOffsetExtension.base;
+#endif
+
 #if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
-   if (IS_R200_CLASS(screen)) {
-      sPriv->psc->allocateMemory = (void *) r200AllocateMemoryMESA;
-      sPriv->psc->freeMemory     = (void *) r200FreeMemoryMESA;
-      sPriv->psc->memoryOffset   = (void *) r200GetMemoryOffsetMESA;
-   }
+   if (IS_R200_CLASS(screen))
+       screen->extensions[i++] = &r200AllocateExtension.base;
+
+   screen->extensions[i++] = &r200texOffsetExtension.base;
 #endif
 
+#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
+   screen->extensions[i++] = &r300texOffsetExtension.base;
+#endif
+
+   screen->extensions[i++] = NULL;
+   sPriv->extensions = screen->extensions;
+
    screen->driScreen = sPriv;
    screen->sarea_priv_offset = dri_priv->sarea_priv_offset;
    return screen;
@@ -935,71 +1140,16 @@ static void radeonDestroyContext(__DRIcontextPrivate * driContextPriv)
 
 #endif
 
-#if !RADEON_COMMON || (RADEON_COMMON && defined(RADEON_COMMON_FOR_R300))
-static struct __DriverAPIRec radeonAPI = {
-   .InitDriver      = radeonInitDriver,
-   .DestroyScreen   = radeonDestroyScreen,
-   .CreateContext   = radeonCreateContext,
-   .DestroyContext  = radeonDestroyContext,
-   .CreateBuffer    = radeonCreateBuffer,
-   .DestroyBuffer   = radeonDestroyBuffer,
-   .SwapBuffers     = radeonSwapBuffers,
-   .MakeCurrent     = radeonMakeCurrent,
-   .UnbindContext   = radeonUnbindContext,
-   .GetSwapInfo     = getSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = radeonCopySubBuffer,
-#if RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
-   .setTexOffset    = r300SetTexOffset,
-#endif
-};
-#else
-static const struct __DriverAPIRec r200API = {
-   .InitDriver      = radeonInitDriver,
-   .DestroyScreen   = radeonDestroyScreen,
-   .CreateContext   = r200CreateContext,
-   .DestroyContext  = r200DestroyContext,
-   .CreateBuffer    = radeonCreateBuffer,
-   .DestroyBuffer   = radeonDestroyBuffer,
-   .SwapBuffers     = r200SwapBuffers,
-   .MakeCurrent     = r200MakeCurrent,
-   .UnbindContext   = r200UnbindContext,
-   .GetSwapInfo     = getSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = r200CopySubBuffer
-};
-#endif
-
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  *
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC void *
-__driCreateNewScreen_20050727( __DRInativeDisplay *dpy,
-                             int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd,
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
+static const __DRIconfig **
+radeonInitScreen(__DRIscreenPrivate *psp)
 {
-   __DRIscreenPrivate *psp;
 #if !RADEON_COMMON
    static const char *driver_name = "Radeon";
    static const __DRIutilversion2 ddx_expected = { 4, 5, 0, 0 };
@@ -1016,57 +1166,42 @@ __driCreateNewScreen_20050727( __DRInativeDisplay *dpy,
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 24, 0 };
 #endif
-
-   dri_interface = interface;
+   RADEONDRIPtr dri_priv = (RADEONDRIPtr) psp->pDevPriv;
 
    if ( ! driCheckDriDdxDrmVersions3( driver_name,
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) ) {
       return NULL;
    }
-#if !RADEON_COMMON || (RADEON_COMMON && defined(RADEON_COMMON_FOR_R300))
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &radeonAPI);
-#elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &r200API);
-#endif
 
-   if ( psp != NULL ) {
-      RADEONDRIPtr dri_priv = (RADEONDRIPtr) psp->pDevPriv;
-      if (driver_modes) {
-         *driver_modes = radeonFillInModes( dri_priv->bpp,
-                                            (dri_priv->bpp == 16) ? 16 : 24,
-                                            (dri_priv->bpp == 16) ? 0  : 8,
-                                            (dri_priv->backOffset != dri_priv->depthOffset) );
-      }
-
-      /* Calling driInitExtensions here, with a NULL context pointer,
-       * does not actually enable the extensions.  It just makes sure
-       * that all the dispatch offsets for all the extensions that
-       * *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create
-       * is called, but we can't enable the extensions until we have a
-       * context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create
+    * is called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
 #if RADEON_COMMON && defined(RADEON_COMMON_FOR_R200)
-      driInitExtensions( NULL, blend_extensions, GL_FALSE );
-      driInitSingleExtension( NULL, ARB_vp_extension );
-      driInitSingleExtension( NULL, NV_vp_extension );
-      driInitSingleExtension( NULL, ATI_fs_extension );
-      driInitExtensions( NULL, point_extensions, GL_FALSE );
+   driInitExtensions( NULL, blend_extensions, GL_FALSE );
+   driInitSingleExtension( NULL, ARB_vp_extension );
+   driInitSingleExtension( NULL, NV_vp_extension );
+   driInitSingleExtension( NULL, ATI_fs_extension );
+   driInitExtensions( NULL, point_extensions, GL_FALSE );
 #endif
-   }
 
-   return (void *) psp;
+   if (!radeonInitDriver(psp))
+       return NULL;
+
+   return radeonFillInModes( psp,
+			     dri_priv->bpp,
+			     (dri_priv->bpp == 16) ? 16 : 24,
+			     (dri_priv->bpp == 16) ? 0  : 8,
+			     (dri_priv->backOffset != dri_priv->depthOffset) );
 }
 
 
@@ -1099,3 +1234,41 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
 
    return 0;
 }
+
+#if !RADEON_COMMON || (RADEON_COMMON && defined(RADEON_COMMON_FOR_R300))
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = radeonInitScreen,
+   .DestroyScreen   = radeonDestroyScreen,
+   .CreateContext   = radeonCreateContext,
+   .DestroyContext  = radeonDestroyContext,
+   .CreateBuffer    = radeonCreateBuffer,
+   .DestroyBuffer   = radeonDestroyBuffer,
+   .SwapBuffers     = radeonSwapBuffers,
+   .MakeCurrent     = radeonMakeCurrent,
+   .UnbindContext   = radeonUnbindContext,
+   .GetSwapInfo     = getSwapInfo,
+   .GetDrawableMSC  = driDrawableGetMSC32,
+   .WaitForMSC      = driWaitForMSC32,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL,
+   .CopySubBuffer   = radeonCopySubBuffer,
+};
+#else
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = radeonInitScreen,
+   .DestroyScreen   = radeonDestroyScreen,
+   .CreateContext   = r200CreateContext,
+   .DestroyContext  = r200DestroyContext,
+   .CreateBuffer    = radeonCreateBuffer,
+   .DestroyBuffer   = radeonDestroyBuffer,
+   .SwapBuffers     = r200SwapBuffers,
+   .MakeCurrent     = r200MakeCurrent,
+   .UnbindContext   = r200UnbindContext,
+   .GetSwapInfo     = getSwapInfo,
+   .GetDrawableMSC  = driDrawableGetMSC32,
+   .WaitForMSC      = driWaitForMSC32,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL,
+   .CopySubBuffer   = r200CopySubBuffer,
+};
+#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.h b/src/mesa/drivers/dri/radeon/radeon_screen.h
index c9b0c3af125..b84c70bfae9 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.h
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.h
@@ -102,6 +102,10 @@ typedef struct {
 
    /* Configuration cache with default values for all contexts */
    driOptionCache optionCache;
+
+   const __DRIextension *extensions[8];
+
+   int num_gb_pipes;
 } radeonScreenRec, *radeonScreenPtr;
 
 #define IS_R100_CLASS(screen) \
diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c
index 732a85ecf0b..12051ff1c81 100644
--- a/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -40,7 +40,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  */
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "swrast/swrast.h"
 
 #include "radeon_context.h"
@@ -173,6 +173,8 @@ radeon_mba_z16(const driRenderbuffer * drb, GLint x, GLint y)
 
 /* 16-bit depth buffer functions
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )					\
    *(GLushort *)(buf + radeon_mba_z16( drb, _x + xo, _y + yo )) = d;
 
@@ -187,6 +189,8 @@ radeon_mba_z16(const driRenderbuffer * drb, GLint x, GLint y)
  * Careful: It looks like the R300 uses ZZZS byte order while the R200
  * uses SZZZ for 24 bit depth, 8 bit stencil mode.
  */
+#define VALUE_TYPE GLuint
+
 #ifdef COMPILE_R300
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c
index 1fe5c244cdb..32bcff33602 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state.c
@@ -32,14 +32,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "api_arrayelt.h"
-#include "enums.h"
-#include "light.h"
-#include "state.h"
-#include "context.h"
-#include "framebuffer.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/api_arrayelt.h"
+#include "main/enums.h"
+#include "main/light.h"
+#include "main/state.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
 
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -1639,8 +1639,7 @@ void radeonSetCliprects( radeonContextPtr rmesa )
    GLframebuffer *const draw_fb = (GLframebuffer*) drawable->driverPrivate;
    GLframebuffer *const read_fb = (GLframebuffer*) readable->driverPrivate;
 
-   if (draw_fb->_ColorDrawBufferMask[0]
-       == BUFFER_BIT_BACK_LEFT) {
+   if (draw_fb->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT) {
       /* Can't ignore 2d windows if we are page flipping.
        */
       if ( drawable->numBackClipRects == 0 || rmesa->doPageFlip ) {
@@ -1692,17 +1691,18 @@ static void radeonDrawBuffer( GLcontext *ctx, GLenum mode )
 
    RADEON_FIREVERTICES(rmesa);	/* don't pipeline cliprect changes */
 
-   /*
-    * _ColorDrawBufferMask is easier to cope with than <mode>.
-    * Check for software fallback, update cliprects.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-   case BUFFER_BIT_BACK_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      /* 0 (GL_NONE) buffers or multiple color drawing buffers */
+      FALLBACK( rmesa, RADEON_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
+   case BUFFER_BACK_LEFT:
       FALLBACK( rmesa, RADEON_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
    default:
-      /* 0 (GL_NONE) buffers or multiple color drawing buffers */
       FALLBACK( rmesa, RADEON_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
@@ -2221,11 +2221,11 @@ radeonUpdateDrawBuffer(GLcontext *ctx)
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    driRenderbuffer *drb;
 
-   if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
+   if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
       /* draw to front */
       drb = (driRenderbuffer *) fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
    }
-   else if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT) {
+   else if (fb->_ColorDrawBufferIndexes[0] == BUFFER_BACK_LEFT) {
       /* draw to back */
       drb = (driRenderbuffer *) fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
    }
diff --git a/src/mesa/drivers/dri/radeon/radeon_state_init.c b/src/mesa/drivers/dri/radeon/radeon_state_init.c
index c876a596e6e..57dc3800501 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -27,9 +27,9 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "api_arrayelt.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/api_arrayelt.h"
 
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index 2b3ae14ff74..ebea1fecdca 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -32,12 +32,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "enums.h"
-#include "imports.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/enums.h"
+#include "main/imports.h"
+#include "main/macros.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "math/m_translate.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.h b/src/mesa/drivers/dri/radeon/radeon_swtcl.h
index 1feedf185dd..e485052ad77 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.h
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.h
@@ -35,7 +35,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __RADEON_TRIS_H__
 #define __RADEON_TRIS_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "radeon_context.h"
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_tcl.c b/src/mesa/drivers/dri/radeon/radeon_tcl.c
index d35be1ca884..779e9ae5df0 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tcl.c
@@ -32,11 +32,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "light.h"
-#include "mtypes.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/light.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
 
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -417,7 +417,7 @@ static GLboolean radeon_run_tcl_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_tex.c b/src/mesa/drivers/dri/radeon/radeon_tex.c
index f3eb9d8eef8..b0aec216706 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tex.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tex.c
@@ -31,18 +31,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *    Brian Paul <brianp@valinux.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "colormac.h"
-#include "context.h"
-#include "enums.h"
-#include "image.h"
-#include "simple_list.h"
-#include "texformat.h"
-#include "texstore.h"
-#include "teximage.h"
-#include "texobj.h"
-
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/simple_list.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
 
 #include "radeon_context.h"
 #include "radeon_state.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_tex.h b/src/mesa/drivers/dri/radeon/radeon_tex.h
index bdf086dfeec..80008808289 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tex.h
+++ b/src/mesa/drivers/dri/radeon/radeon_tex.h
@@ -37,6 +37,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __RADEON_TEX_H__
 #define __RADEON_TEX_H__
 
+extern void radeonSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+                               unsigned long long offset, GLint depth,
+                               GLuint pitch);
+
 extern void radeonUpdateTextureState( GLcontext *ctx );
 
 extern int radeonUploadTexImages( radeonContextPtr rmesa, radeonTexObjPtr t,
diff --git a/src/mesa/drivers/dri/radeon/radeon_texmem.c b/src/mesa/drivers/dri/radeon/radeon_texmem.c
index f7520f1dea3..5f7bbe6a4cb 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texmem.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texmem.c
@@ -36,10 +36,10 @@ SOFTWARE.
  */
 #include <errno.h> 
 
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
 
 #include "radeon_context.h"
 #include "radeon_ioctl.h"
@@ -333,7 +333,7 @@ int radeonUploadTexImages( radeonContextPtr rmesa, radeonTexObjPtr t, GLuint fac
 {
    int numLevels;
 
-   if ( !t || t->base.totalSize == 0 )
+   if ( !t || t->base.totalSize == 0 || t->image_override )
       return 0;
 
    if ( RADEON_DEBUG & (DEBUG_TEXTURE|DEBUG_IOCTL) ) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index ae8d527cf46..1e2f654addf 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -33,13 +33,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Gareth Hughes <gareth@valinux.com>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "colormac.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
+#include "main/texobj.h"
+#include "main/enums.h"
 
 #include "radeon_context.h"
 #include "radeon_state.h"
@@ -83,7 +84,7 @@ tx_table[] =
    _ALPHA_REV(RGBA8888),
    _ALPHA(ARGB8888),
    _ALPHA_REV(ARGB8888),
-   _INVALID(RGB888),
+   [ MESA_FORMAT_RGB888 ] = { RADEON_TXFORMAT_ARGB8888, 0 },
    _COLOR(RGB565),
    _COLOR_REV(RGB565),
    _ALPHA(ARGB4444),
@@ -133,18 +134,19 @@ static void radeonSetTexImages( radeonContextPtr rmesa,
 
    /* Set the hardware texture format
     */
-
-   t->pp_txformat &= ~(RADEON_TXFORMAT_FORMAT_MASK |
-		       RADEON_TXFORMAT_ALPHA_IN_MAP);
-   t->pp_txfilter &= ~RADEON_YUV_TO_RGB;
-
-   if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
-      t->pp_txformat |= tx_table[ baseImage->TexFormat->MesaFormat ].format;
-      t->pp_txfilter |= tx_table[ baseImage->TexFormat->MesaFormat ].filter;
-   }
-   else {
-      _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
-      return;
+   if ( !t->image_override ) {
+      t->pp_txformat &= ~(RADEON_TXFORMAT_FORMAT_MASK |
+                          RADEON_TXFORMAT_ALPHA_IN_MAP);
+      t->pp_txfilter &= ~RADEON_YUV_TO_RGB;
+
+      if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
+         t->pp_txformat |= tx_table[ baseImage->TexFormat->MesaFormat ].format;
+         t->pp_txfilter |= tx_table[ baseImage->TexFormat->MesaFormat ].filter;
+      }
+      else {
+         _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
+         return;
+      }
    }
 
    texelBytes = baseImage->TexFormat->TexelBytes;
@@ -340,11 +342,13 @@ static void radeonSetTexImages( radeonContextPtr rmesa,
     * requires 64-byte aligned pitches, and we may/may not need the
     * blitter.   NPOT only!
     */
-   if (baseImage->IsCompressed)
-      t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
-   else
-      t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
-   t->pp_txpitch -= 32;
+   if ( !t->image_override ) {
+      if (baseImage->IsCompressed)
+         t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
+      else
+         t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
+      t->pp_txpitch -= 32;
+   }
 
    t->dirty_state = TEX_ALL;
 
@@ -839,6 +843,44 @@ static GLboolean radeonUpdateTextureEnv( GLcontext *ctx, int unit )
    return GL_TRUE;
 }
 
+void radeonSetTexOffset(__DRIcontext * pDRICtx, GLint texname,
+                        unsigned long long offset, GLint depth, GLuint pitch)
+{
+	radeonContextPtr rmesa = pDRICtx->driverPrivate;
+	struct gl_texture_object *tObj =
+	    _mesa_lookup_texture(rmesa->glCtx, texname);
+	radeonTexObjPtr t;
+
+	if (tObj == NULL)
+		return;
+
+	t = (radeonTexObjPtr) tObj->DriverData;
+
+	t->image_override = GL_TRUE;
+
+	if (!offset)
+		return;
+
+	t->pp_txoffset = offset;
+	t->pp_txpitch = pitch - 32;
+
+	switch (depth) {
+	case 32:
+		t->pp_txformat = tx_table[MESA_FORMAT_ARGB8888].format;
+		t->pp_txfilter |= tx_table[MESA_FORMAT_ARGB8888].filter;
+		break;
+	case 24:
+	default:
+		t->pp_txformat = tx_table[MESA_FORMAT_RGB888].format;
+		t->pp_txfilter |= tx_table[MESA_FORMAT_RGB888].filter;
+		break;
+	case 16:
+		t->pp_txformat = tx_table[MESA_FORMAT_RGB565].format;
+		t->pp_txfilter |= tx_table[MESA_FORMAT_RGB565].filter;
+		break;
+	}
+}
+
 #define TEXOBJ_TXFILTER_MASK (RADEON_MAX_MIP_LEVEL_MASK |	\
 			      RADEON_MIN_FILTER_MASK | 		\
 			      RADEON_MAG_FILTER_MASK |		\
@@ -1135,7 +1177,7 @@ static GLboolean enable_tex_2d( GLcontext *ctx, int unit )
       RADEON_FIREVERTICES( rmesa );
       radeonSetTexImages( rmesa, tObj );
       radeonUploadTexImages( rmesa, (radeonTexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock ) 
+      if ( !t->base.memBlock && !t->image_override ) 
 	return GL_FALSE;
    }
 
@@ -1202,7 +1244,8 @@ static GLboolean enable_tex_rect( GLcontext *ctx, int unit )
       RADEON_FIREVERTICES( rmesa );
       radeonSetTexImages( rmesa, tObj );
       radeonUploadTexImages( rmesa, (radeonTexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock /* && !rmesa->prefer_gart_client_texturing  FIXME */ ) {
+      if ( !t->base.memBlock &&
+           !t->image_override /* && !rmesa->prefer_gart_client_texturing  FIXME */ ) {
 	 fprintf(stderr, "%s: upload failed\n", __FUNCTION__);
 	 return GL_FALSE;
       }
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.c b/src/mesa/drivers/dri/s3v/s3v_context.c
index 2d2f704ad77..14502f95aec 100644
--- a/src/mesa/drivers/dri/s3v/s3v_context.c
+++ b/src/mesa/drivers/dri/s3v/s3v_context.c
@@ -11,15 +11,15 @@
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 
-#include "context.h"
-#include "simple_list.h"
-#include "matrix.h"
-#include "extensions.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
 #if defined(USE_X86_ASM)
 #include "x86/common_x86_asm.h"
 #endif
-#include "simple_list.h"
-#include "mm.h"
+#include "main/simple_list.h"
+#include "main/mm.h"
 
 #include "drivers/common/driverfuncs.h"
 #include "s3v_vb.h"
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.h b/src/mesa/drivers/dri/s3v/s3v_context.h
index 6f527f3a834..671ba90d781 100644
--- a/src/mesa/drivers/dri/s3v/s3v_context.h
+++ b/src/mesa/drivers/dri/s3v/s3v_context.h
@@ -11,11 +11,11 @@
 #include "s3v_regs.h"
 #include "s3v_macros.h"
 #include "s3v_screen.h"
-#include "colormac.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/colormac.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 #include "drm.h"
-#include "mm.h"
+#include "main/mm.h"
 #include "drirenderbuffer.h"
 
 /* Flags for context */
@@ -419,9 +419,9 @@ struct s3v_context {
 #define S3VIRGEPACKCOLOR4444( r, g, b, a ) \
     ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
 
-static __inline GLuint s3vPackColor( GLuint cpp,
-					GLubyte r, GLubyte g,
-					GLubyte b, GLubyte a )
+static INLINE GLuint s3vPackColor( GLuint cpp,
+                                   GLubyte r, GLubyte g,
+                                   GLubyte b, GLubyte a )
 {
 	unsigned int ret;
 	DEBUG(("cpp = %i, r=0x%x, g=0x%x, b=0x%x, a=0x%x\n", cpp, r, g, b, a));
diff --git a/src/mesa/drivers/dri/s3v/s3v_dd.c b/src/mesa/drivers/dri/s3v/s3v_dd.c
index 1cbe890319a..e340116f5e1 100644
--- a/src/mesa/drivers/dri/s3v/s3v_dd.c
+++ b/src/mesa/drivers/dri/s3v/s3v_dd.c
@@ -9,8 +9,8 @@
 #include "x86/common_x86_asm.h"
 #endif
 
-#include "context.h"
-#include "framebuffer.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
 #include "swrast/swrast.h"
 
 #define S3V_DATE "20020207"
diff --git a/src/mesa/drivers/dri/s3v/s3v_macros.h b/src/mesa/drivers/dri/s3v/s3v_macros.h
index 3bc871b7eaf..7e9b4529df4 100644
--- a/src/mesa/drivers/dri/s3v/s3v_macros.h
+++ b/src/mesa/drivers/dri/s3v/s3v_macros.h
@@ -19,6 +19,7 @@
 #include <stdio.h>
 #endif
 
+#undef DEBUG
 #if GENERIC_DEBUG
 #define DEBUG(str) printf str
 #else
diff --git a/src/mesa/drivers/dri/s3v/s3v_render.c b/src/mesa/drivers/dri/s3v/s3v_render.c
index 6aaa94976e3..5023f3c4641 100644
--- a/src/mesa/drivers/dri/s3v/s3v_render.c
+++ b/src/mesa/drivers/dri/s3v/s3v_render.c
@@ -2,10 +2,10 @@
  * Author: Max Lingua <sunmax@libero.it>
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 
 #include "tnl/t_context.h"
 
@@ -66,7 +66,7 @@ static const GLuint hw_prim[GL_POLYGON+1] = {
 	PrimType_Polygon
 };
 
-static __inline void s3vStartPrimitive( s3vContextPtr vmesa, GLenum prim )
+static INLINE void s3vStartPrimitive( s3vContextPtr vmesa, GLenum prim )
 {
 	__DRIdrawablePrivate *dPriv = vmesa->driDrawable;
 
@@ -110,7 +110,7 @@ static __inline void s3vStartPrimitive( s3vContextPtr vmesa, GLenum prim )
 	vmesa->restore_primitive = _hw_prim;
 }
 
-static __inline void s3vEndPrimitive( s3vContextPtr vmesa )
+static INLINE void s3vEndPrimitive( s3vContextPtr vmesa )
 {
 /*	GLcontext *ctx = vmesa->glCtx; */
 	DEBUG(("s3vEndPrimitive\n"));
@@ -170,7 +170,7 @@ static GLboolean s3v_run_render( GLcontext *ctx,
 
 	for (i = 0 ; i < VB->PrimitiveCount ; i++ )
 	{
-                GLuint prim = VB->Primitive[i].mode;
+		GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
 		GLuint start = VB->Primitive[i].start;
 		GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/s3v/s3v_screen.h b/src/mesa/drivers/dri/s3v/s3v_screen.h
index 0c4f69efac3..c49bc8587da 100644
--- a/src/mesa/drivers/dri/s3v/s3v_screen.h
+++ b/src/mesa/drivers/dri/s3v/s3v_screen.h
@@ -2,7 +2,7 @@
  * Author: Max Lingua <sunmax@libero.it>
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 typedef struct _s3vRegion {
     drm_handle_t  handle;
diff --git a/src/mesa/drivers/dri/s3v/s3v_span.c b/src/mesa/drivers/dri/s3v/s3v_span.c
index de78f9f6b1b..f9f7c0d1eee 100644
--- a/src/mesa/drivers/dri/s3v/s3v_span.c
+++ b/src/mesa/drivers/dri/s3v/s3v_span.c
@@ -128,6 +128,8 @@ do { \
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d ) \
    *(GLushort *)(buf + _x*2 + _y*dPriv->w*2) = d
 
@@ -143,6 +145,8 @@ do { \
 /* 32 bit depthbuffer functions.
  */
 #if 0
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLuint *)(buf + _x*4 + _y*pitch) = d;
 
@@ -157,6 +161,8 @@ do { \
 /* 24/8 bit interleaved depth/stencil functions
  */
 #if 0
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) { \
    GLuint tmp = *(GLuint *)(buf + _x*4 + _y*pitch);	\
    tmp &= 0xff; \
diff --git a/src/mesa/drivers/dri/s3v/s3v_state.c b/src/mesa/drivers/dri/s3v/s3v_state.c
index b86b618c117..c71c89a3e1b 100644
--- a/src/mesa/drivers/dri/s3v/s3v_state.c
+++ b/src/mesa/drivers/dri/s3v/s3v_state.c
@@ -5,9 +5,9 @@
 #include <X11/Xlibint.h>
 #include "s3v_context.h"
 #include "s3v_macros.h"
-#include "macros.h"
 #include "s3v_dri.h"
-#include "colormac.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "vbo/vbo.h"
diff --git a/src/mesa/drivers/dri/s3v/s3v_tex.c b/src/mesa/drivers/dri/s3v/s3v_tex.c
index 5bee051b092..8bf2ea98783 100644
--- a/src/mesa/drivers/dri/s3v/s3v_tex.c
+++ b/src/mesa/drivers/dri/s3v/s3v_tex.c
@@ -5,16 +5,16 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
+#include "main/mm.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
 #include "s3v_context.h"
 #include "s3v_tex.h"
 
diff --git a/src/mesa/drivers/dri/s3v/s3v_texmem.c b/src/mesa/drivers/dri/s3v/s3v_texmem.c
index 5b44340d19b..705d105f55f 100644
--- a/src/mesa/drivers/dri/s3v/s3v_texmem.c
+++ b/src/mesa/drivers/dri/s3v/s3v_texmem.c
@@ -5,13 +5,13 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "s3v_context.h"
 #include "s3v_lock.h"
 #include "s3v_tex.h"
diff --git a/src/mesa/drivers/dri/s3v/s3v_texstate.c b/src/mesa/drivers/dri/s3v/s3v_texstate.c
index 2719de3663a..455bae6301d 100644
--- a/src/mesa/drivers/dri/s3v/s3v_texstate.c
+++ b/src/mesa/drivers/dri/s3v/s3v_texstate.c
@@ -5,13 +5,13 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "s3v_context.h"
 #include "s3v_tex.h"
 
diff --git a/src/mesa/drivers/dri/s3v/s3v_tris.c b/src/mesa/drivers/dri/s3v/s3v_tris.c
index d6cceddd454..fafd38480c9 100644
--- a/src/mesa/drivers/dri/s3v/s3v_tris.c
+++ b/src/mesa/drivers/dri/s3v/s3v_tris.c
@@ -11,10 +11,10 @@
 #include "s3v_vb.h"
 #include "s3v_tris.h"
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/s3v/s3v_vb.c b/src/mesa/drivers/dri/s3v/s3v_vb.c
index dadf2b48911..00e375c6c43 100644
--- a/src/mesa/drivers/dri/s3v/s3v_vb.c
+++ b/src/mesa/drivers/dri/s3v/s3v_vb.c
@@ -2,10 +2,10 @@
  * Author: Max Lingua <sunmax@libero.it>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
diff --git a/src/mesa/drivers/dri/s3v/s3v_vb.h b/src/mesa/drivers/dri/s3v/s3v_vb.h
index b35d804e625..0fd54373805 100644
--- a/src/mesa/drivers/dri/s3v/s3v_vb.h
+++ b/src/mesa/drivers/dri/s3v/s3v_vb.h
@@ -5,7 +5,7 @@
 #ifndef S3VVB_INC
 #define S3VVB_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 
 #define _S3V_NEW_VERTEX (_NEW_TEXTURE |		\
diff --git a/src/mesa/drivers/dri/s3v/s3v_xmesa.c b/src/mesa/drivers/dri/s3v/s3v_xmesa.c
index c66fd6dac31..b18c8763c3a 100644
--- a/src/mesa/drivers/dri/s3v/s3v_xmesa.c
+++ b/src/mesa/drivers/dri/s3v/s3v_xmesa.c
@@ -4,11 +4,11 @@
 
 #include "s3v_context.h"
 #include "s3v_vb.h"
-#include "context.h"
-#include "matrix.h"
+#include "main/context.h"
+#include "main/matrix.h"
 #include "s3v_dri.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -17,8 +17,8 @@
 
 /* #define DEBUG(str) printf str */
 
-static GLboolean 
-s3vInitDriver(__DRIscreenPrivate *sPriv)
+static const __DRIconfig **
+s3vInitScreen(__DRIscreen *sPriv)
 {
     sPriv->private = (void *) s3vCreateScreen( sPriv );
 
@@ -27,7 +27,7 @@ s3vInitDriver(__DRIscreenPrivate *sPriv)
 	return GL_FALSE;
     }
 
-    return GL_TRUE;
+   return NULL;
 }
 
 static void 
@@ -327,34 +327,14 @@ s3vUnbindContext( __DRIcontextPrivate *driContextPriv )
    return GL_TRUE;
 }
 
-
-static struct __DriverAPIRec s3vAPI = {
-   s3vInitDriver,
-   s3vDestroyScreen,
-   s3vCreateContext,
-   s3vDestroyContext,
-   s3vCreateBuffer,
-   s3vDestroyBuffer,
-   s3vSwapBuffers,
-   s3vMakeCurrent,
-   s3vUnbindContext,
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen	   = s3vInitScreen,
+   .DestroyScreen  = s3vDestroyScreen,
+   .CreateContext  = s3vCreateContext,
+   .DestroyContext = s3vDestroyContext,
+   .CreateBuffer   = s3vCreateBuffer,
+   .DestroyBuffer  = s3vDestroyBuffer,
+   .SwapBuffers	   = s3vSwapBuffers,
+   .MakeCurrent	   = s3vMakeCurrent,
+   .UnbindContext  = s3vUnbindContext,
 };
-
-
-#if 0
-/*
- * This is the bootstrap function for the driver.
- * The __driCreateScreen name is the symbol that libGL.so fetches.
- * Return:  pointer to a __DRIscreenPrivate.
- */
-void *__driCreateScreen(Display *dpy, int scrn, __DRIscreen *psc,
-                        int numConfigs, __GLXvisualConfig *config)
-{
-   __DRIscreenPrivate *psp=NULL;
-
-   DEBUG(("__driCreateScreen: psp = %p\n", psp));
-   psp = __driUtilCreateScreen(dpy, scrn, psc, numConfigs, config, &s3vAPI);
-   DEBUG(("__driCreateScreen: psp = %p\n", psp));
-   return (void *) psp;
-}
-#endif
diff --git a/src/mesa/drivers/dri/savage/savage_init.h b/src/mesa/drivers/dri/savage/savage_init.h
index 43fb969c693..abb8440fc4c 100644
--- a/src/mesa/drivers/dri/savage/savage_init.h
+++ b/src/mesa/drivers/dri/savage/savage_init.h
@@ -28,7 +28,7 @@
 
 #include <sys/time.h>
 #include "dri_util.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 #include "xmlconfig.h"
 
diff --git a/src/mesa/drivers/dri/savage/savage_xmesa.c b/src/mesa/drivers/dri/savage/savage_xmesa.c
index 00171253290..a344aab71bb 100644
--- a/src/mesa/drivers/dri/savage/savage_xmesa.c
+++ b/src/mesa/drivers/dri/savage/savage_xmesa.c
@@ -26,16 +26,16 @@
 #include <X11/Xlibint.h>
 #include <stdio.h>
 
-#include "savagecontext.h"
-#include "context.h"
-#include "matrix.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
+#include "main/context.h"
+#include "main/context.h"
+#include "main/matrix.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
+#include "main/simple_list.h"
 
 #include "utils.h"
 
-#include "extensions.h"
+#include "main/extensions.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -168,16 +168,17 @@ static const struct tnl_pipeline_stage *savage_pipeline[] = {
 };
 
 
-/* this is first function called in dirver*/
+PUBLIC const __DRIextension *savageScreenExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    &driReadDrawableExtension,
+};
 
 static GLboolean
 savageInitDriver(__DRIscreenPrivate *sPriv)
 {
   savageScreenPrivate *savageScreen;
   SAVAGEDRIPtr         gDRIPriv = (SAVAGEDRIPtr)sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-
 
    if (sPriv->devPrivSize != sizeof(SAVAGEDRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(SAVAGEDRIRec) does not match passed size from device driver\n");
@@ -265,10 +266,7 @@ savageInitDriver(__DRIscreenPrivate *sPriv)
    driParseOptionInfo (&savageScreen->optionCache,
 		       __driConfigOptions, __driNConfigOptions);
 
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension)(sPriv->psc->screenConfigs,
-			      "GLX_SGI_make_current_read");
-   }
+   sPriv->extensions = savageScreenExtensions;
 
 #if 0
    savageDDFastPathInit();
@@ -295,34 +293,6 @@ savageDestroyScreen(__DRIscreenPrivate *sPriv)
    sPriv->private = NULL;
 }
 
-#if 0
-GLvisual *XMesaCreateVisual(Display *dpy,
-                            __DRIscreenPrivate *driScrnPriv,
-                            const XVisualInfo *visinfo,
-                            const __GLXvisualConfig *config)
-{
-   /* Drivers may change the args to _mesa_create_visual() in order to
-    * setup special visuals.
-    */
-   return _mesa_create_visual( config->rgba,
-                               config->doubleBuffer,
-                               config->stereo,
-                               _mesa_bitcount(visinfo->red_mask),
-                               _mesa_bitcount(visinfo->green_mask),
-                               _mesa_bitcount(visinfo->blue_mask),
-                               config->alphaSize,
-                               0, /* index bits */
-                               config->depthSize,
-                               config->stencilSize,
-                               config->accumRedSize,
-                               config->accumGreenSize,
-                               config->accumBlueSize,
-                               config->accumAlphaSize,
-                               0 /* num samples */ );
-}
-#endif
-
-
 static GLboolean
 savageCreateContext( const __GLcontextModes *mesaVis,
 		     __DRIcontextPrivate *driContextPriv,
@@ -525,7 +495,7 @@ savageCreateContext( const __GLcontextModes *mesaVis,
 					    "enable_fastpath");
    /* DRM versions before 2.1.3 would only render triangle lists. ELTS
     * support was added in 2.2.0. */
-   if (imesa->enable_fastpath && sPriv->drmMinor < 2) {
+   if (imesa->enable_fastpath && sPriv->drm_version.minor < 2) {
       fprintf (stderr,
 	       "*** Disabling fast path because your DRM version is buggy "
 	       "or doesn't\n*** support ELTS. You need at least Savage DRM "
@@ -732,7 +702,7 @@ void savageXMesaSetClipRects(savageContextPtr imesa)
    __DRIdrawablePrivate *dPriv = imesa->driDrawable;
 
    if ((dPriv->numBackClipRects == 0)
-       || (imesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT)) {
+       || (imesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT)) {
       imesa->numClipRects = dPriv->numClipRects;
       imesa->pClipRects = dPriv->pClipRects;
       imesa->drawX = dPriv->x;
@@ -917,32 +887,18 @@ void savageGetLock( savageContextPtr imesa, GLuint flags )
    }
 }
 
-
-
-static const struct __DriverAPIRec savageAPI = {
-   savageInitDriver,
-   savageDestroyScreen,
-   savageCreateContext,
-   savageDestroyContext,
-   savageCreateBuffer,
-   savageDestroyBuffer,
-   savageSwapBuffers,
-   savageMakeCurrent,
-   savageUnbindContext
-};
-
-
-static __GLcontextModes *
-savageFillInModes( unsigned pixel_bits, unsigned depth_bits,
+static const  __DRIconfig **
+savageFillInModes( __DRIscreenPrivate *psp,
+		   unsigned pixel_bits, unsigned depth_bits,
 		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
-    __GLcontextModes * modes;
+    __DRIconfig **configs;
     __GLcontextModes * m;
-    unsigned num_modes;
     unsigned depth_buffer_factor;
     unsigned back_buffer_factor;
     GLenum fb_format;
     GLenum fb_type;
+    int i;
 
     /* Right now GLX_SWAP_COPY_OML isn't supported, but it would be easy
      * enough to add support.  Basically, if a context is created with an
@@ -973,8 +929,6 @@ savageFillInModes( unsigned pixel_bits, unsigned depth_bits,
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -984,21 +938,11 @@ savageFillInModes( unsigned pixel_bits, unsigned depth_bits,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
-
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR ) ) {
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor,
+			       back_buffer_modes, back_buffer_factor);
+    if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
 	return NULL;
@@ -1006,74 +950,68 @@ savageFillInModes( unsigned pixel_bits, unsigned depth_bits,
 
     /* Mark the visual as slow if there are "fake" stencil bits.
      */
-    for ( m = modes ; m != NULL ; m = m->next ) {
-	if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
+    for (i = 0; configs[i]; i++) {
+	m = &configs[i]->modes;
+	if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
 	    m->visualRating = GLX_SLOW_CONFIG;
 	}
     }
 
-    return modes;
+    return (const __DRIconfig **) configs;
 }
 
 
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+static const __DRIconfig **
+savageInitScreen(__DRIscreenPrivate *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 2, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 2, 1, 0 };
-
-   dri_interface = interface;
+   SAVAGEDRIPtr dri_priv = (SAVAGEDRIPtr)psp->pDevPriv;
 
    if ( ! driCheckDriDdxDrmVersions2( "Savage",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
-      
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &savageAPI);
-   if ( psp != NULL ) {
-      SAVAGEDRIPtr dri_priv = (SAVAGEDRIPtr)psp->pDevPriv;
-      *driver_modes = savageFillInModes( dri_priv->cpp*8,
-					 (dri_priv->cpp == 2) ? 16 : 24,
-					 (dri_priv->cpp == 2) ? 0  : 8,
-					 (dri_priv->backOffset != dri_priv->depthOffset) );
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
 
-   return (void *) psp;
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+   if (!savageInitDriver(psp))
+       return NULL;
+
+   return savageFillInModes( psp,
+			     dri_priv->cpp*8,
+			     (dri_priv->cpp == 2) ? 16 : 24,
+			     (dri_priv->cpp == 2) ? 0  : 8,
+			     (dri_priv->backOffset != dri_priv->depthOffset) );
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   savageInitScreen, 
+   savageDestroyScreen,
+   savageCreateContext,
+   savageDestroyContext,
+   savageCreateBuffer,
+   savageDestroyBuffer,
+   savageSwapBuffers,
+   savageMakeCurrent,
+   savageUnbindContext
+};
diff --git a/src/mesa/drivers/dri/savage/savagecontext.h b/src/mesa/drivers/dri/savage/savagecontext.h
index 3fe3d71d4ec..fd6399d6a6f 100644
--- a/src/mesa/drivers/dri/savage/savagecontext.h
+++ b/src/mesa/drivers/dri/savage/savagecontext.h
@@ -33,13 +33,13 @@ typedef struct savage_texture_object_t *savageTextureObjectPtr;
 
 #include <X11/Xlibint.h>
 #include "dri_util.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "xf86drm.h"
 #include "drm.h"
 #include "savage_drm.h"
 #include "savage_init.h"
 #include "savage_3d_reg.h"
-#include "mm.h"
+#include "main/mm.h"
 #include "tnl/t_vertex.h"
 
 #include "texmem.h"
diff --git a/src/mesa/drivers/dri/savage/savagedd.c b/src/mesa/drivers/dri/savage/savagedd.c
index a5c5310e287..32ca86de8ae 100644
--- a/src/mesa/drivers/dri/savage/savagedd.c
+++ b/src/mesa/drivers/dri/savage/savagedd.c
@@ -23,12 +23,12 @@
  */
 
 
-#include "mtypes.h"
-#include "framebuffer.h"
+#include "main/mtypes.h"
+#include "main/framebuffer.h"
 
 #include <stdio.h>
 
-#include "mm.h"
+#include "main/mm.h"
 #include "swrast/swrast.h"
 
 #include "savagedd.h"
@@ -37,7 +37,7 @@
 #include "savagetex.h"
 #include "savagetris.h"
 #include "savagecontext.h"
-#include "extensions.h"
+#include "main/extensions.h"
 
 #include "utils.h"
 
diff --git a/src/mesa/drivers/dri/savage/savagedd.h b/src/mesa/drivers/dri/savage/savagedd.h
index ae167be7008..698a8d5de92 100644
--- a/src/mesa/drivers/dri/savage/savagedd.h
+++ b/src/mesa/drivers/dri/savage/savagedd.h
@@ -26,7 +26,7 @@
 #ifndef SAVAGEDD_INC
 #define SAVAGEDD_INC
 
-#include "context.h"
+#include "main/context.h"
 
 void savageDDInitDriverFuncs( GLcontext *ctx );
 #endif
diff --git a/src/mesa/drivers/dri/savage/savageioctl.c b/src/mesa/drivers/dri/savage/savageioctl.c
index bea9a3ea535..948ed18419e 100644
--- a/src/mesa/drivers/dri/savage/savageioctl.c
+++ b/src/mesa/drivers/dri/savage/savageioctl.c
@@ -27,14 +27,14 @@
 #include <unistd.h>
 #include <sys/mman.h>
 
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
-#include "context.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/dd.h"
+#include "main/context.h"
+#include "main/colormac.h"
+#include "main/mm.h"
 #include "swrast/swrast.h"
-#include "colormac.h"
 
-#include "mm.h"
 #include "savagecontext.h"
 #include "savageioctl.h"
 #include "savage_bci.h"
diff --git a/src/mesa/drivers/dri/savage/savageioctl.h b/src/mesa/drivers/dri/savage/savageioctl.h
index 98629048aef..639605cc517 100644
--- a/src/mesa/drivers/dri/savage/savageioctl.h
+++ b/src/mesa/drivers/dri/savage/savageioctl.h
@@ -64,19 +64,19 @@ void savageSwapBuffers( __DRIdrawablePrivate *dPriv );
 
 extern void savageGetDMABuffer( savageContextPtr imesa );
 
-static __inline
+static INLINE
 void savageReleaseIndexedVerts( savageContextPtr imesa )
 {
     imesa->firstElt = -1;
 }
 
-static __inline
+static INLINE
 GLboolean savageHaveIndexedVerts( savageContextPtr imesa )
 {
     return (imesa->firstElt != -1);
 }
 
-static __inline
+static INLINE
 uint32_t *savageAllocVtxBuf( savageContextPtr imesa, GLuint words )
 {
    struct savage_vtxbuf_t *buffer = imesa->vtxBuf;
@@ -115,7 +115,7 @@ uint32_t *savageAllocVtxBuf( savageContextPtr imesa, GLuint words )
    return head;
 }
 
-static __inline
+static INLINE
 uint32_t *savageAllocIndexedVerts( savageContextPtr imesa, GLuint n )
 {
     uint32_t *ret;
@@ -131,7 +131,7 @@ uint32_t *savageAllocIndexedVerts( savageContextPtr imesa, GLuint n )
  * - Actually allocate entries for the indices in the command buffer.
  *   (This allocation must succeed without wrapping the cmd buffer!)
  */
-static __inline
+static INLINE
 void savageFlushElts( savageContextPtr imesa )
 {
     if (imesa->elts.cmd) {
@@ -148,7 +148,7 @@ void savageFlushElts( savageContextPtr imesa )
 /* Allocate a command buffer entry with <bytes> bytes of arguments:
  * - implies savageFlushElts
  */
-static __inline
+static INLINE
 drm_savage_cmd_header_t *savageAllocCmdBuf( savageContextPtr imesa, GLuint bytes )
 {
     drm_savage_cmd_header_t *ret;
@@ -171,7 +171,7 @@ drm_savage_cmd_header_t *savageAllocCmdBuf( savageContextPtr imesa, GLuint bytes
  *   incomplete indexed drawing command yet
  * - increments the number of elts. Final allocation is done in savageFlushElts
  */
-static __inline
+static INLINE
 uint16_t *savageAllocElts( savageContextPtr imesa, GLuint n )
 {
     uint16_t *ret;
diff --git a/src/mesa/drivers/dri/savage/savagerender.c b/src/mesa/drivers/dri/savage/savagerender.c
index 514434c4271..32c74f9467e 100644
--- a/src/mesa/drivers/dri/savage/savagerender.c
+++ b/src/mesa/drivers/dri/savage/savagerender.c
@@ -27,11 +27,11 @@
  * dma buffers.  Use strip/fan hardware primitives where possible.
  * Simulate missing primitives with indexed vertices.
  */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "tnl/t_context.h"
 
@@ -198,7 +198,7 @@ static GLboolean savage_run_render( GLcontext *ctx,
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      GLuint prim = VB->Primitive[i].mode;
+      GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
diff --git a/src/mesa/drivers/dri/savage/savagespan.c b/src/mesa/drivers/dri/savage/savagespan.c
index 61ab9e6d64a..9615e34013c 100644
--- a/src/mesa/drivers/dri/savage/savagespan.c
+++ b/src/mesa/drivers/dri/savage/savagespan.c
@@ -22,7 +22,7 @@
  * DEALINGS IN THE SOFTWARE.
  */
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 #include "savagedd.h"
 #include "savagespan.h"
 #include "savageioctl.h"
@@ -93,6 +93,8 @@
 /* 16 bit integer depthbuffer functions
  * Depth range is reversed. See also savageCalcViewport.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d ) \
     *(GLushort *)(buf + ((_x)<<1) + (_y)*pitch) = 0xFFFF - d
 
@@ -107,6 +109,8 @@
 
 /* 16 bit float depthbuffer functions
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d ) \
     *(GLushort *)(buf + ((_x)<<1) + (_y)*pitch) = \
         savageEncodeFloat16( 1.0 - (GLfloat)d/65535.0 )
@@ -125,6 +129,8 @@
 /* 8-bit stencil /24-bit integer depth depthbuffer functions.
  * Depth range is reversed. See also savageCalcViewport.
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) do {				\
    GLuint tmp = *(GLuint *)(buf + ((_x)<<2) + (_y)*pitch);	\
    tmp &= 0xFF000000;						\
@@ -143,6 +149,8 @@
 
 /* 24 bit float depthbuffer functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) do {				\
     GLuint tmp = *(GLuint *)(buf + ((_x)<<2) + (_y)*pitch);	\
     tmp &= 0xFF000000;						\
diff --git a/src/mesa/drivers/dri/savage/savagespan.h b/src/mesa/drivers/dri/savage/savagespan.h
index f6a312e820e..53a7f8b97c8 100644
--- a/src/mesa/drivers/dri/savage/savagespan.h
+++ b/src/mesa/drivers/dri/savage/savagespan.h
@@ -55,7 +55,7 @@ savageSetSpanFunctions(driRenderbuffer *rb, const GLvisual *vis,
  *
  * Note that there is no encoding for numbers < 2^-16.
  */
-static __inline GLuint savageEncodeFloat16( GLdouble x )
+static INLINE GLuint savageEncodeFloat16( GLdouble x )
 {
     GLint r = (GLint)(x * 0x10000000);
     GLint exp = 0;
@@ -67,7 +67,7 @@ static __inline GLuint savageEncodeFloat16( GLdouble x )
     }
     return exp > 0xf ? 0xffff : (r - 0x1000) | (exp << 12);
 }
-static __inline GLdouble savageDecodeFloat16( GLuint x )
+static INLINE GLdouble savageDecodeFloat16( GLuint x )
 {
     static const GLdouble pow2[16] = {
 	1.0/(1<<28), 1.0/(1<<27), 1.0/(1<<26), 1.0/(1<<25),
@@ -92,7 +92,7 @@ static __inline GLdouble savageDecodeFloat16( GLuint x )
  *
  * Details analogous to the 16-bit format.
  */
-static __inline GLuint savageEncodeFloat24( GLdouble x )
+static INLINE GLuint savageEncodeFloat24( GLdouble x )
 {
     int64_t r = (int64_t)(x * ((int64_t)1 << (19+32)));
     GLint exp = 0;
@@ -105,7 +105,7 @@ static __inline GLuint savageEncodeFloat24( GLdouble x )
     return exp > 0x1f ? 0xffffff : (r - 0x80000) | (exp << 19);
 }
 #define _1 (int64_t)1
-static __inline GLdouble savageDecodeFloat24( GLuint x )
+static INLINE GLdouble savageDecodeFloat24( GLuint x )
 {
     static const GLdouble pow2[32] = {
 	1.0/(_1<<51), 1.0/(_1<<50), 1.0/(_1<<49), 1.0/(_1<<48),
diff --git a/src/mesa/drivers/dri/savage/savagestate.c b/src/mesa/drivers/dri/savage/savagestate.c
index 84fd3157edf..73d85ed57b4 100644
--- a/src/mesa/drivers/dri/savage/savagestate.c
+++ b/src/mesa/drivers/dri/savage/savagestate.c
@@ -25,12 +25,12 @@
 
 #include <stdio.h>
 
-#include "mtypes.h"
-#include "enums.h"
-#include "macros.h"
-#include "dd.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/macros.h"
+#include "main/dd.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "savagedd.h"
 #include "savagecontext.h"
 
@@ -76,7 +76,7 @@
 static void savageBlendFunc_s4(GLcontext *);
 static void savageBlendFunc_s3d(GLcontext *);
 
-static __inline__ GLuint savagePackColor(GLuint format, 
+static INLINE GLuint savagePackColor(GLuint format, 
                                          GLubyte r, GLubyte g, 
                                          GLubyte b, GLubyte a)
 {
@@ -640,15 +640,17 @@ static void savageDDDrawBuffer(GLcontext *ctx, GLenum mode )
     savageContextPtr imesa = SAVAGE_CONTEXT(ctx);
     uint32_t destCtrl = imesa->regs.s4.destCtrl.ui;
 
-    /*
-     * _DrawDestMask is easier to cope with than <mode>.
-     */
-    switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-    case BUFFER_BIT_FRONT_LEFT:
+    if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+	FALLBACK( ctx, SAVAGE_FALLBACK_DRAW_BUFFER, GL_TRUE );
+        return;
+    }
+
+    switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+    case BUFFER_FRONT_LEFT:
         imesa->IsDouble = GL_FALSE;
 	imesa->regs.s4.destCtrl.ni.offset = imesa->savageScreen->frontOffset>>11;
 	break;
-    case BUFFER_BIT_BACK_LEFT:
+    case BUFFER_BACK_LEFT:
         imesa->IsDouble = GL_TRUE;
 	imesa->regs.s4.destCtrl.ni.offset = imesa->savageScreen->backOffset>>11;
 	break;
diff --git a/src/mesa/drivers/dri/savage/savagetex.c b/src/mesa/drivers/dri/savage/savagetex.c
index dbe30d4c947..a3bebfa8cf7 100644
--- a/src/mesa/drivers/dri/savage/savagetex.c
+++ b/src/mesa/drivers/dri/savage/savagetex.c
@@ -28,22 +28,21 @@
 
 #include <GL/gl.h>
 
-#include "mm.h"
+#include "main/mm.h"
 #include "savagecontext.h"
 #include "savagetex.h"
 #include "savagetris.h"
 #include "savageioctl.h"
-#include "simple_list.h"
-#include "enums.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 #include "savage_bci.h"
 
-#include "macros.h"
-#include "texformat.h"
-#include "texstore.h"
-#include "texobj.h"
-
-#include "convolve.h"
-#include "colormac.h"
+#include "main/macros.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/texobj.h"
+#include "main/convolve.h"
+#include "main/colormac.h"
 
 #include "swrast/swrast.h"
 
@@ -101,7 +100,7 @@ static const savageTileInfo tileInfo_s3d_s4[5] = {
  * \param w   width in bytes
  */
 #define SUBTILE_FUNC(w,h)					\
-static __inline GLubyte *savageUploadSubtile_##w##x##h		\
+static INLINE GLubyte *savageUploadSubtile_##w##x##h		\
 (GLubyte *dest, GLubyte *src, GLuint srcStride)			\
 {								\
     GLuint y;							\
@@ -1016,7 +1015,7 @@ static void savageUploadTexImages( savageContextPtr imesa, savageTexObjPtr t )
       /* Heap timestamps are only reliable with Savage DRM 2.3.x or
        * later. Earlier versions had only 16 bit time stamps which
        * would wrap too frequently. */
-      if (imesa->savageScreen->driScrnPriv->drmMinor >= 3) {
+      if (imesa->savageScreen->driScrnPriv->drm_version.minor >= 3) {
 	  unsigned int heap = t->base.heap->heapId;
 	  LOCK_HARDWARE(imesa);
 	  savageWaitEvent (imesa, imesa->textureHeaps[heap]->timestamp);
@@ -1713,7 +1712,7 @@ static void savageTimestampTextures( savageContextPtr imesa )
     * Only useful with long-lived 32-bit event tags available
     * with Savage DRM 2.3.x or later. */
    if ((imesa->CurrentTexObj[0] || imesa->CurrentTexObj[1]) &&
-       imesa->savageScreen->driScrnPriv->drmMinor >= 3) {
+       imesa->savageScreen->driScrnPriv->drm_version.minor >= 3) {
        unsigned int e;
        FLUSH_BATCH(imesa);
        e = savageEmitEvent(imesa, SAVAGE_WAIT_3D);
diff --git a/src/mesa/drivers/dri/savage/savagetex.h b/src/mesa/drivers/dri/savage/savagetex.h
index f1ee7224145..e5f8a80f85d 100644
--- a/src/mesa/drivers/dri/savage/savagetex.h
+++ b/src/mesa/drivers/dri/savage/savagetex.h
@@ -26,7 +26,7 @@
 #ifndef SAVAGETEX_INC
 #define SAVAGETEX_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 #include "savagecontext.h"
 #include "texmem.h"
diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c
index a1c7bb167e7..c04763b40e1 100644
--- a/src/mesa/drivers/dri/savage/savagetris.c
+++ b/src/mesa/drivers/dri/savage/savagetris.c
@@ -37,10 +37,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include <stdio.h>
 #include <math.h>
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -95,7 +95,7 @@ do {						\
 } while (0)
 #endif
 
-static void __inline__ savage_draw_triangle (savageContextPtr imesa,
+static void INLINE savage_draw_triangle (savageContextPtr imesa,
 					     savageVertexPtr v0,
 					     savageVertexPtr v1,
 					     savageVertexPtr v2) {
@@ -108,7 +108,7 @@ static void __inline__ savage_draw_triangle (savageContextPtr imesa,
    EMIT_VERT (j, vb, vertsize, 0, v2);
 }
 
-static void __inline__ savage_draw_quad (savageContextPtr imesa,
+static void INLINE savage_draw_quad (savageContextPtr imesa,
 					 savageVertexPtr v0,
 					 savageVertexPtr v1,
 					 savageVertexPtr v2,
@@ -125,7 +125,7 @@ static void __inline__ savage_draw_quad (savageContextPtr imesa,
    EMIT_VERT (j, vb, vertsize, 0, v3);
 }
 
-static __inline__ void savage_draw_point (savageContextPtr imesa,
+static INLINE void savage_draw_point (savageContextPtr imesa,
 					  savageVertexPtr tmp) {
    GLuint vertsize = imesa->HwVertexSize;
    uint32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
@@ -161,7 +161,7 @@ static __inline__ void savage_draw_point (savageContextPtr imesa,
    EMIT_VERT (j, vb, vertsize, 2, tmp);
 }
 
-static __inline__ void savage_draw_line (savageContextPtr imesa,
+static INLINE void savage_draw_line (savageContextPtr imesa,
 					 savageVertexPtr v0,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
@@ -219,7 +219,7 @@ do {							\
    tmp.f[vertex_size-1] *= rhw;				\
 } while (0)
 
-static void __inline__ savage_ptex_tri (savageContextPtr imesa,
+static void INLINE savage_ptex_tri (savageContextPtr imesa,
 					savageVertexPtr v0,
 					savageVertexPtr v1,
 					savageVertexPtr v2) {
@@ -233,7 +233,7 @@ static void __inline__ savage_ptex_tri (savageContextPtr imesa,
    PTEX_VERTEX (j, tmp, vertsize, 0, v2); EMIT_VERT (j, vb, vertsize, 0, &tmp);
 }
 
-static __inline__ void savage_ptex_line (savageContextPtr imesa,
+static INLINE void savage_ptex_line (savageContextPtr imesa,
 					 savageVertexPtr v0,
 					 savageVertexPtr v1 ) {
    GLuint vertsize = imesa->HwVertexSize;
@@ -281,7 +281,7 @@ static __inline__ void savage_ptex_line (savageContextPtr imesa,
    EMIT_VERT (j, vb, vertsize, 2, &tmp1);
 } 
 
-static __inline__ void savage_ptex_point (savageContextPtr imesa,
+static INLINE void savage_ptex_point (savageContextPtr imesa,
 					  savageVertexPtr v0) {
    GLuint vertsize = imesa->HwVertexSize;
    uint32_t *vb = savageAllocVtxBuf (imesa, 6*vertsize);
@@ -934,7 +934,7 @@ do {									\
 #define SAVAGE_EMIT_ST1  0x0300
 
 
-static __inline__ GLuint savageChooseVertexFormat_s3d( GLcontext *ctx )
+static INLINE GLuint savageChooseVertexFormat_s3d( GLcontext *ctx )
 {
    savageContextPtr imesa = SAVAGE_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
@@ -997,7 +997,7 @@ static __inline__ GLuint savageChooseVertexFormat_s3d( GLcontext *ctx )
 }
 
 
-static __inline__ GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
+static INLINE GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
 {
    savageContextPtr imesa = SAVAGE_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
diff --git a/src/mesa/drivers/dri/savage/savagetris.h b/src/mesa/drivers/dri/savage/savagetris.h
index b2b3d951c68..a2a9375ed54 100644
--- a/src/mesa/drivers/dri/savage/savagetris.h
+++ b/src/mesa/drivers/dri/savage/savagetris.h
@@ -36,7 +36,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R128_TRIS_H__
 #define __R128_TRIS_H__
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void savageInitTriFuncs( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/sis/sis6326_clear.c b/src/mesa/drivers/dri/sis/sis6326_clear.c
index 48db19566c6..d46ecc9cd27 100644
--- a/src/mesa/drivers/dri/sis/sis6326_clear.c
+++ b/src/mesa/drivers/dri/sis/sis6326_clear.c
@@ -32,7 +32,7 @@
 #include "sis_reg.h"
 
 #include "swrast/swrast.h"
-#include "macros.h"
+#include "main/macros.h"
 
 static void sis_clear_front_buffer(GLcontext *ctx, GLenum mask, GLint x,
 				   GLint y, GLint width, GLint height);
diff --git a/src/mesa/drivers/dri/sis/sis6326_state.c b/src/mesa/drivers/dri/sis/sis6326_state.c
index 08402fb3e2a..65d4c064660 100644
--- a/src/mesa/drivers/dri/sis/sis6326_state.c
+++ b/src/mesa/drivers/dri/sis/sis6326_state.c
@@ -33,9 +33,9 @@
 #include "sis_tex.h"
 #include "sis_reg.h"
 
-#include "context.h"
-#include "enums.h"
-#include "colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -497,25 +497,27 @@ void sis6326DDDrawBuffer( GLcontext *ctx, GLenum mode )
    __GLSiSHardware *current = &smesa->current;
 
    if(getenv("SIS_DRAW_FRONT"))
-      ctx->DrawBuffer->_ColorDrawBufferMask[0] = GL_FRONT_LEFT;
+      ctx->DrawBuffer->_ColorDrawBufferIndexes[0] = BUFFER_FRONT_LEFT;
+
+   if (ctx->DrawBuffer->_NumColorDrawBuffers > 1) {
+      FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
 
-   /*
-    * _DrawDestMask is easier to cope with than <mode>.
-    */
    current->hwDstSet &= ~MASK_DstBufferPitch;
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       current->hwOffsetDest = smesa->front.offset;
       current->hwDstSet |= smesa->front.pitch;
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       current->hwOffsetDest = smesa->back.offset;
       current->hwDstSet |= smesa->back.pitch;
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
    default:
-      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
diff --git a/src/mesa/drivers/dri/sis/sis_clear.c b/src/mesa/drivers/dri/sis/sis_clear.c
index 174f3c07686..323383da62a 100644
--- a/src/mesa/drivers/dri/sis/sis_clear.c
+++ b/src/mesa/drivers/dri/sis/sis_clear.c
@@ -36,7 +36,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_lock.h"
 
 #include "swrast/swrast.h"
-#include "macros.h"
+#include "main/macros.h"
 
 static GLbitfield sis_3D_Clear( GLcontext * ctx, GLbitfield mask,
 				GLint x, GLint y, GLint width,
diff --git a/src/mesa/drivers/dri/sis/sis_context.c b/src/mesa/drivers/dri/sis/sis_context.c
index 04c7464c5e2..00d17da3bad 100644
--- a/src/mesa/drivers/dri/sis/sis_context.c
+++ b/src/mesa/drivers/dri/sis/sis_context.c
@@ -42,11 +42,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_tris.h"
 #include "sis_alloc.h"
 
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
 #include "utils.h"
-#include "framebuffer.h"
+#include "main/framebuffer.h"
 
 #include "drivers/common/driverfuncs.h"
 
diff --git a/src/mesa/drivers/dri/sis/sis_context.h b/src/mesa/drivers/dri/sis/sis_context.h
index b81812d6cec..bc53cb5efa1 100644
--- a/src/mesa/drivers/dri/sis/sis_context.h
+++ b/src/mesa/drivers/dri/sis/sis_context.h
@@ -34,7 +34,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef _sis_ctx_h_
 #define _sis_ctx_h_
 
-#include "context.h"
+#include "main/context.h"
 #include "dri_util.h"
 #include "drm.h"
 #include "drm_sarea.h"
@@ -400,8 +400,10 @@ struct sis_context
 #define MMIO_READ(reg) *(volatile GLint *)(smesa->IOBase + (reg))
 #define MMIO_READf(reg) *(volatile GLfloat *)(smesa->IOBase + (reg))
 
-#if defined(__i386__) || defined(__amd64__)
+#if defined(__i386__) || defined(__x86_64__)
 #define MMIO_WMB()	__asm __volatile("" : : : "memory")
+#elif defined(__ia64__)
+#define MMIO_WMB()	__asm __volatile("mf" : : : "memory")
 #else
 #error platform needs WMB
 #endif
diff --git a/src/mesa/drivers/dri/sis/sis_dd.c b/src/mesa/drivers/dri/sis/sis_dd.c
index 989c159a809..bddc4a9285c 100644
--- a/src/mesa/drivers/dri/sis/sis_dd.c
+++ b/src/mesa/drivers/dri/sis/sis_dd.c
@@ -41,8 +41,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_tris.h"
 
 #include "swrast/swrast.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "utils.h"
 
diff --git a/src/mesa/drivers/dri/sis/sis_fog.c b/src/mesa/drivers/dri/sis/sis_fog.c
index ba5ac908512..517d5722e6a 100644
--- a/src/mesa/drivers/dri/sis/sis_fog.c
+++ b/src/mesa/drivers/dri/sis/sis_fog.c
@@ -35,7 +35,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_state.h"
 #include "swrast/swrast.h"
 
-#include "macros.h"
+#include "main/macros.h"
 
 static GLint convertFtToFogFt( GLfloat dwInValue );
 static GLint doFPtoFixedNoRound( GLfloat dwInValue, int nFraction );
diff --git a/src/mesa/drivers/dri/sis/sis_lock.c b/src/mesa/drivers/dri/sis/sis_lock.c
index 0ea64e34988..806110cad4d 100644
--- a/src/mesa/drivers/dri/sis/sis_lock.c
+++ b/src/mesa/drivers/dri/sis/sis_lock.c
@@ -28,7 +28,7 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Eric Anholt <anholt@FreeBSD.org>
  */
 
-#include "context.h"
+#include "main/context.h"
 #include "sis_context.h"
 #include "sis_lock.h"
 #include "sis_dd.h"
diff --git a/src/mesa/drivers/dri/sis/sis_screen.c b/src/mesa/drivers/dri/sis/sis_screen.c
index f55027e0945..b1a5d15236f 100644
--- a/src/mesa/drivers/dri/sis/sis_screen.c
+++ b/src/mesa/drivers/dri/sis/sis_screen.c
@@ -30,11 +30,11 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "dri_util.h"
 
-#include "context.h"
+#include "main/context.h"
 #include "utils.h"
-#include "imports.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/imports.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "sis_context.h"
 #include "sis_dri.h"
@@ -64,12 +64,10 @@ static const GLuint __driNConfigOptions = 3;
 
 extern const struct dri_extension card_extensions[];
 
-static __GLcontextModes *
-sisFillInModes(int bpp)
+static const __DRIconfig **
+sisFillInModes(__DRIscreenPrivate *psp, int bpp)
 {
-   __GLcontextModes *modes;
-   __GLcontextModes *m;
-   unsigned num_modes;
+   __DRIconfig **configs;
    unsigned depth_buffer_factor;
    unsigned back_buffer_factor;
    GLenum fb_format;
@@ -92,9 +90,6 @@ sisFillInModes(int bpp)
    depth_buffer_factor = 4;
    back_buffer_factor = 2;
 
-   /* Last 4 is for GLX_TRUE_COLOR & GLX_DIRECT_COLOR, with/without accum */
-   num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
    if (bpp == 16) {
       fb_format = GL_RGB;
       fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -103,25 +98,15 @@ sisFillInModes(int bpp)
       fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
    }
 
-   modes = (*dri_interface->createContextModes)(num_modes, sizeof(__GLcontextModes));
-   m = modes;
-   if (!driFillInModes(&m, fb_format, fb_type, depth_bits_array,
-		       stencil_bits_array, depth_buffer_factor,
-		       back_buffer_modes, back_buffer_factor,
-		       GLX_TRUE_COLOR)) {
-      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__);
-      return NULL;
-   }
-
-   if (!driFillInModes(&m, fb_format, fb_type, depth_bits_array,
-		       stencil_bits_array, depth_buffer_factor,
-		       back_buffer_modes, back_buffer_factor,
-		       GLX_DIRECT_COLOR)) {
+   configs = driCreateConfigs(fb_format, fb_type, depth_bits_array,
+			      stencil_bits_array, depth_buffer_factor,
+			      back_buffer_modes, back_buffer_factor);
+   if (configs == NULL) {
       fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__);
       return NULL;
    }
 
-   return modes;
+   return (const __DRIconfig **) configs;
 }
 
 
@@ -287,23 +272,52 @@ sisSwapBuffers(__DRIdrawablePrivate *dPriv)
 }
 
 
-/* Initialize the driver specific screen private data.
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-static GLboolean
-sisInitDriver( __DRIscreenPrivate *sPriv )
+static const __DRIconfig **
+sisInitScreen(__DRIscreenPrivate *psp)
 {
-   sPriv->private = (void *) sisCreateScreen( sPriv );
+   static const __DRIversion ddx_expected = {0, 8, 0};
+   static const __DRIversion dri_expected = {4, 0, 0};
+   static const __DRIversion drm_expected = {1, 0, 0};
+   static const char *driver_name = "SiS";
+   SISDRIPtr dri_priv = (SISDRIPtr)psp->pDevPriv;
 
-   if ( !sPriv->private ) {
-      sisDestroyScreen( sPriv );
-      return GL_FALSE;
+   if (!driCheckDriDdxDrmVersions2(driver_name,
+				   &psp->dri_version, &dri_expected,
+				   &psp->ddx_version, &ddx_expected,
+				   &psp->drm_version, &drm_expected))
+      return NULL;
+
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+   psp->private = sisCreateScreen(psp);
+
+   if (!psp->private) {
+      sisDestroyScreen(psp);
+      return NULL;
    }
 
-   return GL_TRUE;
+   return sisFillInModes(psp, dri_priv->bytesPerPixel * 8);
 }
 
-static struct __DriverAPIRec sisAPI = {
-   .InitDriver      = sisInitDriver,
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = sisInitScreen,
    .DestroyScreen   = sisDestroyScreen,
    .CreateContext   = sisCreateContext,
    .DestroyContext  = sisDestroyContext,
@@ -313,69 +327,9 @@ static struct __DriverAPIRec sisAPI = {
    .MakeCurrent     = sisMakeCurrent,
    .UnbindContext   = sisUnbindContext,
    .GetSwapInfo     = NULL,
-   .GetMSC          = NULL,
+   .GetDrawableMSC  = NULL,
    .WaitForMSC      = NULL,
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 
 };
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- *
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn,
-			     __DRIscreen *psc,
-			     const __GLcontextModes *modes,
-			     const __DRIversion *ddx_version,
-			     const __DRIversion *dri_version,
-			     const __DRIversion *drm_version,
-			     const __DRIframebuffer *frame_buffer,
-			     drmAddress pSAREA, int fd,
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes **driver_modes )
-
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = {0, 8, 0};
-   static const __DRIversion dri_expected = {4, 0, 0};
-   static const __DRIversion drm_expected = {1, 0, 0};
-   static const char *driver_name = "SiS";
-   dri_interface = interface;
-
-   if (!driCheckDriDdxDrmVersions2(driver_name, dri_version, &dri_expected,
-				   ddx_version, &ddx_expected,
-				   drm_version, &drm_expected)) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &sisAPI);
-   if (psp != NULL) {
-      SISDRIPtr dri_priv = (SISDRIPtr)psp->pDevPriv;
-      *driver_modes = sisFillInModes(dri_priv->bytesPerPixel * 8);
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
-
-   return (void *)psp;
-}
diff --git a/src/mesa/drivers/dri/sis/sis_span.c b/src/mesa/drivers/dri/sis/sis_span.c
index dc50bda877b..9e9a509755d 100644
--- a/src/mesa/drivers/dri/sis/sis_span.c
+++ b/src/mesa/drivers/dri/sis/sis_span.c
@@ -84,6 +84,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* 16 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLushort *)(buf + (_x)*2 + (_y)*srb->pitch) = d;
 
@@ -96,6 +98,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* 32 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d )	\
    *(GLuint *)(buf + (_x)*4 + (_y)*srb->pitch) = d;
 
@@ -108,6 +112,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* 8/24 bit interleaved depth/stencil functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) {				\
    GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*srb->pitch);	\
    tmp &= 0xff000000;						\
diff --git a/src/mesa/drivers/dri/sis/sis_state.c b/src/mesa/drivers/dri/sis/sis_state.c
index 305c63f73f7..98e8d02fabe 100644
--- a/src/mesa/drivers/dri/sis/sis_state.c
+++ b/src/mesa/drivers/dri/sis/sis_state.c
@@ -37,9 +37,9 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_lock.h"
 #include "sis_tex.h"
 
-#include "context.h"
-#include "enums.h"
-#include "colormac.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
 #include "tnl/tnl.h"
@@ -546,23 +546,24 @@ void sisDDDrawBuffer( GLcontext *ctx, GLenum mode )
    __GLSiSHardware *prev = &smesa->prev;
    __GLSiSHardware *current = &smesa->current;
 
-   /*
-    * _DrawDestMask is easier to cope with than <mode>.
-    */
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
    current->hwDstSet &= ~MASK_DstBufferPitch;
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_FALSE );
       current->hwOffsetDest = smesa->front.offset >> 1;
       current->hwDstSet |= smesa->front.pitch >> 2;
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_FALSE );
       current->hwOffsetDest = smesa->back.offset >> 1;
       current->hwDstSet |= smesa->back.pitch >> 2;
       break;
    default:
-      /* GL_NONE or GL_FRONT_AND_BACK or stereo left&right, etc */
       FALLBACK( smesa, SIS_FALLBACK_DRAW_BUFFER, GL_TRUE );
       return;
    }
diff --git a/src/mesa/drivers/dri/sis/sis_tex.c b/src/mesa/drivers/dri/sis/sis_tex.c
index 5e10c610f80..28ced6cfd5f 100644
--- a/src/mesa/drivers/dri/sis/sis_tex.c
+++ b/src/mesa/drivers/dri/sis/sis_tex.c
@@ -33,11 +33,11 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_tex.h"
 
 #include "swrast/swrast.h"
-#include "imports.h"
-#include "texformat.h"
-#include "texstore.h"
-#include "teximage.h"
-#include "texobj.h"
+#include "main/imports.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
 
 #include "xmlpool.h"
 
@@ -129,6 +129,8 @@ sisAllocTexImage( sisContextPtr smesa, sisTexObjPtr t, int level,
 static void
 sisFreeTexImage( sisContextPtr smesa, sisTexObjPtr t, int level )
 {
+   assert(level >= 0);
+   assert(level < SIS_MAX_TEXTURE_LEVELS);
    if (t->image[level].Data == NULL)
       return;
 
@@ -212,7 +214,7 @@ sisDeleteTexture( GLcontext * ctx, struct gl_texture_object *texObj )
        */
       return;
    }
-   for (i = 0; i < MAX_TEXTURE_LEVELS; i++) {
+   for (i = 0; i < SIS_MAX_TEXTURE_LEVELS; i++) {
       sisFreeTexImage( smesa, t, i );
    }
 
diff --git a/src/mesa/drivers/dri/sis/sis_texstate.c b/src/mesa/drivers/dri/sis/sis_texstate.c
index 4f813bb81c3..63f23fc0149 100644
--- a/src/mesa/drivers/dri/sis/sis_texstate.c
+++ b/src/mesa/drivers/dri/sis/sis_texstate.c
@@ -31,12 +31,12 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Eric Anholt <anholt@FreeBSD.org>
  */
 
-#include "glheader.h"
-#include "imports.h"
-#include "colormac.h"
-#include "context.h"
-#include "macros.h"
-#include "texformat.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/texformat.h"
 
 #include "sis_context.h"
 #include "sis_state.h"
diff --git a/src/mesa/drivers/dri/sis/sis_tris.c b/src/mesa/drivers/dri/sis/sis_tris.c
index a0e39dcd3c9..095941aea2e 100644
--- a/src/mesa/drivers/dri/sis/sis_tris.c
+++ b/src/mesa/drivers/dri/sis/sis_tris.c
@@ -32,10 +32,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Eric Anholt <anholt@FreeBSD.org>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "macros.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/macros.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -848,7 +848,7 @@ static void sisRenderStart( GLcontext *ctx )
 
    RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
-   if (ctx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT && 
+   if (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT && 
       smesa->driDrawable->numClipRects != 0)
    {
       multipass_cliprect(ctx, 0);
diff --git a/src/mesa/drivers/dri/sis/sis_tris.h b/src/mesa/drivers/dri/sis/sis_tris.h
index 499eb4d24d4..b34fe8c7c98 100644
--- a/src/mesa/drivers/dri/sis/sis_tris.h
+++ b/src/mesa/drivers/dri/sis/sis_tris.h
@@ -32,7 +32,7 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define __SIS_TRIS_H__
 
 #include "sis_lock.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void sisInitTriFuncs( GLcontext *ctx );
 extern void sisFlushPrims( sisContextPtr smesa );
@@ -47,7 +47,7 @@ do {							\
       sisFlushPrims(smesa);				\
 } while (0)
 
-static __inline GLuint *sisAllocDmaLow(sisContextPtr smesa, int bytes)
+static INLINE GLuint *sisAllocDmaLow(sisContextPtr smesa, int bytes)
 {
    GLuint *start;
 
diff --git a/src/mesa/drivers/dri/sis/sis_tritmp.h b/src/mesa/drivers/dri/sis/sis_tritmp.h
index e670a5bf760..f75e17318fc 100644
--- a/src/mesa/drivers/dri/sis/sis_tritmp.h
+++ b/src/mesa/drivers/dri/sis/sis_tritmp.h
@@ -234,7 +234,7 @@ static void TAG(sis6326_draw_point_mmio)(sisContextPtr smesa, char *verts)
 }
 #endif
 
-static __inline void TAG(sis_vert_init)( void )
+static INLINE void TAG(sis_vert_init)( void )
 {
    sis_tri_func_mmio[SIS_STATES] = TAG(sis_draw_tri_mmio);
    sis_line_func_mmio[SIS_STATES] = TAG(sis_draw_line_mmio);
diff --git a/src/mesa/drivers/dri/swrast/Makefile b/src/mesa/drivers/dri/swrast/Makefile
new file mode 100644
index 00000000000..5f3a4f2191b
--- /dev/null
+++ b/src/mesa/drivers/dri/swrast/Makefile
@@ -0,0 +1,24 @@
+# src/mesa/drivers/dri/swrast/Makefile
+
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = swrast_dri.so
+
+DRIVER_SOURCES = \
+	swrast.c \
+	swrast_span.c
+
+C_SOURCES = \
+	$(SWRAST_COMMON_SOURCES) \
+	$(DRIVER_SOURCES)
+
+ASM_SOURCES =
+
+SWRAST_COMMON_SOURCES = \
+	../../common/driverfuncs.c \
+	../common/utils.c
+
+include ../Makefile.template
+
+symlinks:
diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c
new file mode 100644
index 00000000000..49f1b8b9ee7
--- /dev/null
+++ b/src/mesa/drivers/dri/swrast/swrast.c
@@ -0,0 +1,730 @@
+/*
+ * Copyright (C) 2008 George Sapountzis <gsap7@yahoo.gr>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * DRI software rasterizer
+ *
+ * This is the mesa swrast module packaged into a DRI driver structure.
+ *
+ * The front-buffer is allocated by the loader. The loader provides read/write
+ * callbacks for access to the front-buffer. The driver uses a scratch row for
+ * front-buffer rendering to avoid repeated calls to the loader.
+ *
+ * The back-buffer is allocated by the driver and is private.
+ */
+
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/imports.h"
+#include "main/renderbuffer.h"
+#include "swrast/swrast.h"
+#include "swrast_setup/swrast_setup.h"
+#include "tnl/tnl.h"
+#include "tnl/t_context.h"
+#include "tnl/t_pipeline.h"
+#include "vbo/vbo.h"
+#include "drivers/common/driverfuncs.h"
+#include "utils.h"
+
+#include "swrast_priv.h"
+
+
+#define need_GL_VERSION_1_3
+#define need_GL_VERSION_1_4
+#define need_GL_VERSION_1_5
+#define need_GL_VERSION_2_0
+#define need_GL_VERSION_2_1
+
+/* sw extensions for imaging */
+#define need_GL_EXT_blend_color
+#define need_GL_EXT_blend_minmax
+#define need_GL_EXT_convolution
+#define need_GL_EXT_histogram
+#define need_GL_SGI_color_table
+
+/* sw extensions not associated with some GL version */
+#define need_GL_ARB_shader_objects
+#define need_GL_ARB_vertex_program
+#define need_GL_APPLE_vertex_array_object
+#define need_GL_ATI_fragment_shader
+#define need_GL_EXT_depth_bounds_test
+#define need_GL_EXT_framebuffer_object
+#define need_GL_EXT_framebuffer_blit
+#define need_GL_EXT_gpu_program_parameters
+#define need_GL_EXT_paletted_texture
+#define need_GL_IBM_multimode_draw_arrays
+#define need_GL_MESA_resize_buffers
+#define need_GL_NV_vertex_program
+#define need_GL_NV_fragment_program
+
+#include "extension_helper.h"
+
+const struct dri_extension card_extensions[] =
+{
+    { "GL_VERSION_1_3",			GL_VERSION_1_3_functions },
+    { "GL_VERSION_1_4",			GL_VERSION_1_4_functions },
+    { "GL_VERSION_1_5",			GL_VERSION_1_5_functions },
+    { "GL_VERSION_2_0",			GL_VERSION_2_0_functions },
+    { "GL_VERSION_2_1",			GL_VERSION_2_1_functions },
+
+    { "GL_EXT_blend_color",		GL_EXT_blend_color_functions },
+    { "GL_EXT_blend_minmax",		GL_EXT_blend_minmax_functions },
+    { "GL_EXT_convolution",		GL_EXT_convolution_functions },
+    { "GL_EXT_histogram",		GL_EXT_histogram_functions },
+    { "GL_SGI_color_table",		GL_SGI_color_table_functions },
+
+    { "GL_ARB_shader_objects",		GL_ARB_shader_objects_functions },
+    { "GL_ARB_vertex_program",		GL_ARB_vertex_program_functions },
+    { "GL_APPLE_vertex_array_object",	GL_APPLE_vertex_array_object_functions },
+    { "GL_ATI_fragment_shader",		GL_ATI_fragment_shader_functions },
+    { "GL_EXT_depth_bounds_test",	GL_EXT_depth_bounds_test_functions },
+    { "GL_EXT_framebuffer_object",	GL_EXT_framebuffer_object_functions },
+    { "GL_EXT_framebuffer_blit",	GL_EXT_framebuffer_blit_functions },
+    { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
+    { "GL_EXT_paletted_texture",	GL_EXT_paletted_texture_functions },
+    { "GL_IBM_multimode_draw_arrays",	GL_IBM_multimode_draw_arrays_functions },
+    { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
+    { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
+    { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
+    { NULL,				NULL }
+};
+
+
+/**
+ * Screen and config-related functions
+ */
+
+static void
+setupLoaderExtensions(__DRIscreen *psp,
+		      const __DRIextension **extensions)
+{
+    int i;
+
+    for (i = 0; extensions[i]; i++) {
+	if (strcmp(extensions[i]->name, __DRI_SWRAST_LOADER) == 0)
+	    psp->swrast_loader = (__DRIswrastLoaderExtension *) extensions[i];
+    }
+}
+
+static __DRIconfig **
+swrastFillInModes(__DRIscreen *psp,
+		  unsigned pixel_bits, unsigned depth_bits,
+		  unsigned stencil_bits, GLboolean have_back_buffer)
+{
+    __DRIconfig **configs;
+    unsigned depth_buffer_factor;
+    unsigned back_buffer_factor;
+    GLenum fb_format;
+    GLenum fb_type;
+
+    /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
+     * support pageflipping at all.
+     */
+    static const GLenum back_buffer_modes[] = {
+	GLX_NONE, GLX_SWAP_UNDEFINED_OML
+    };
+
+    uint8_t depth_bits_array[4];
+    uint8_t stencil_bits_array[4];
+
+    depth_bits_array[0] = 0;
+    depth_bits_array[1] = 0;
+    depth_bits_array[2] = depth_bits;
+    depth_bits_array[3] = depth_bits;
+
+    /* Just like with the accumulation buffer, always provide some modes
+     * with a stencil buffer.
+     */
+    stencil_bits_array[0] = 0;
+    stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
+    stencil_bits_array[2] = 0;
+    stencil_bits_array[3] = (stencil_bits == 0) ? 8 : stencil_bits;
+
+    depth_buffer_factor = 4;
+    back_buffer_factor = 2;
+
+    if (pixel_bits == 8) {
+	fb_format = GL_RGB;
+	fb_type = GL_UNSIGNED_BYTE_2_3_3_REV;
+    }
+    else if (pixel_bits == 16) {
+	fb_format = GL_RGB;
+	fb_type = GL_UNSIGNED_SHORT_5_6_5;
+    }
+    else {
+	fb_format = GL_BGRA;
+	fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+    }
+
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor, back_buffer_modes,
+			       back_buffer_factor);
+    if (configs == NULL) {
+	fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+		__LINE__);
+	return NULL;
+    }
+
+    return configs;
+}
+
+static __DRIscreen *
+driCreateNewScreen(int scrn, const __DRIextension **extensions,
+		   const __DRIconfig ***driver_configs, void *data)
+{
+    static const __DRIextension *emptyExtensionList[] = { NULL };
+    __DRIscreen *psp;
+    __DRIconfig **configs8, **configs16, **configs32;
+
+    (void) data;
+
+    TRACE;
+
+    psp = _mesa_calloc(sizeof(*psp));
+    if (!psp)
+	return NULL;
+
+    setupLoaderExtensions(psp, extensions);
+
+    psp->num = scrn;
+    psp->extensions = emptyExtensionList;
+
+    configs8  = swrastFillInModes(psp,  8,  8, 0, 1);
+    configs16 = swrastFillInModes(psp, 16, 16, 0, 1);
+    configs32 = swrastFillInModes(psp, 32, 24, 8, 1);
+
+    configs16 = (__DRIconfig **)driConcatConfigs(configs8, configs16);
+
+    *driver_configs = driConcatConfigs(configs16, configs32);
+
+    driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+    return psp;
+}
+
+static void driDestroyScreen(__DRIscreen *psp)
+{
+    TRACE;
+
+    if (psp) {
+	_mesa_free(psp);
+    }
+}
+
+static const __DRIextension **driGetExtensions(__DRIscreen *psp)
+{
+    TRACE;
+
+    return psp->extensions;
+}
+
+
+/**
+ * Framebuffer and renderbuffer-related functions.
+ */
+
+static GLuint
+choose_pixel_format(const GLvisual *v)
+{
+    if (v->rgbMode) {
+	int bpp = v->rgbBits;
+
+	if (bpp == 32
+	    && v->redMask   == 0xff0000
+	    && v->greenMask == 0x00ff00
+	    && v->blueMask  == 0x0000ff)
+	    return PF_A8R8G8B8;
+	else if (bpp == 16
+	    && v->redMask   == 0xf800
+	    && v->greenMask == 0x07e0
+	    && v->blueMask  == 0x001f)
+	    return PF_R5G6B5;
+	else if (bpp == 8
+	    && v->redMask   == 0x07
+	    && v->greenMask == 0x38
+	    && v->blueMask  == 0xc0)
+	    return PF_R3G3B2;
+    }
+    else {
+	if (v->indexBits == 8)
+	    return PF_CI8;
+    }
+
+    _mesa_problem( NULL, "unexpected format in %s", __FUNCTION__ );
+    return 0;
+}
+
+static void
+swrast_delete_renderbuffer(struct gl_renderbuffer *rb)
+{
+    TRACE;
+
+    _mesa_free(rb->Data);
+    _mesa_free(rb);
+}
+
+static GLboolean
+swrast_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
+			   GLenum internalFormat, GLuint width, GLuint height)
+{
+    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+    int bpp;
+    unsigned mask = PITCH_ALIGN_BITS - 1;
+
+    TRACE;
+
+    rb->Data = NULL;
+    rb->Width = width;
+    rb->Height = height;
+
+    switch (internalFormat) {
+    case GL_RGB:
+	bpp = rb->RedBits + rb->GreenBits + rb->BlueBits;
+	break;
+    case GL_RGBA:
+	bpp = rb->RedBits + rb->GreenBits + rb->BlueBits + rb->AlphaBits;
+	break;
+    case GL_COLOR_INDEX8_EXT:
+	bpp = rb->IndexBits;
+	break;
+    default:
+	_mesa_problem( NULL, "unexpected format in %s", __FUNCTION__ );
+	return GL_FALSE;
+    }
+
+    /* always pad to PITCH_ALIGN_BITS */
+    xrb->pitch = ((width * bpp + mask) & ~mask) / 8;
+
+    return GL_TRUE;
+}
+
+static GLboolean
+swrast_alloc_back_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
+			  GLenum internalFormat, GLuint width, GLuint height)
+{
+    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+
+    TRACE;
+
+    _mesa_free(rb->Data);
+
+    swrast_alloc_front_storage(ctx, rb, internalFormat, width, height);
+
+    rb->Data = _mesa_malloc(height * xrb->pitch);
+
+    return GL_TRUE;
+}
+
+static struct swrast_renderbuffer *
+swrast_new_renderbuffer(const GLvisual *visual, GLboolean front)
+{
+    struct swrast_renderbuffer *xrb = _mesa_calloc(sizeof *xrb);
+    GLuint pixel_format;
+
+    TRACE;
+
+    if (!xrb)
+	return NULL;
+
+    _mesa_init_renderbuffer(&xrb->Base, 0);
+
+    pixel_format = choose_pixel_format(visual);
+
+    xrb->Base.Delete = swrast_delete_renderbuffer;
+    if (front) {
+	xrb->Base.AllocStorage = swrast_alloc_front_storage;
+	swrast_set_span_funcs_front(xrb, pixel_format);
+    }
+    else {
+	xrb->Base.AllocStorage = swrast_alloc_back_storage;
+	swrast_set_span_funcs_back(xrb, pixel_format);
+    }
+
+    switch (pixel_format) {
+    case PF_A8R8G8B8:
+	xrb->Base.InternalFormat = GL_RGBA;
+	xrb->Base._BaseFormat = GL_RGBA;
+	xrb->Base.DataType = GL_UNSIGNED_BYTE;
+	xrb->Base.RedBits   = 8 * sizeof(GLubyte);
+	xrb->Base.GreenBits = 8 * sizeof(GLubyte);
+	xrb->Base.BlueBits  = 8 * sizeof(GLubyte);
+	xrb->Base.AlphaBits = 8 * sizeof(GLubyte);
+	break;
+    case PF_R5G6B5:
+	xrb->Base.InternalFormat = GL_RGB;
+	xrb->Base._BaseFormat = GL_RGB;
+	xrb->Base.DataType = GL_UNSIGNED_BYTE;
+	xrb->Base.RedBits   = 5 * sizeof(GLubyte);
+	xrb->Base.GreenBits = 6 * sizeof(GLubyte);
+	xrb->Base.BlueBits  = 5 * sizeof(GLubyte);
+	xrb->Base.AlphaBits = 0;
+	break;
+    case PF_R3G3B2:
+	xrb->Base.InternalFormat = GL_RGB;
+	xrb->Base._BaseFormat = GL_RGB;
+	xrb->Base.DataType = GL_UNSIGNED_BYTE;
+	xrb->Base.RedBits   = 3 * sizeof(GLubyte);
+	xrb->Base.GreenBits = 3 * sizeof(GLubyte);
+	xrb->Base.BlueBits  = 2 * sizeof(GLubyte);
+	xrb->Base.AlphaBits = 0;
+	break;
+    case PF_CI8:
+	xrb->Base.InternalFormat = GL_COLOR_INDEX8_EXT;
+	xrb->Base._BaseFormat = GL_COLOR_INDEX;
+	xrb->Base.DataType = GL_UNSIGNED_BYTE;
+	xrb->Base.IndexBits = 8 * sizeof(GLubyte);
+	break;
+    default:
+	return NULL;
+    }
+
+    return xrb;
+}
+
+static __DRIdrawable *
+driCreateNewDrawable(__DRIscreen *screen,
+		     const __DRIconfig *config, void *data)
+{
+    __DRIdrawable *buf;
+    struct swrast_renderbuffer *frontrb, *backrb;
+
+    TRACE;
+
+    buf = _mesa_calloc(sizeof *buf);
+    if (!buf)
+	return NULL;
+
+    buf->loaderPrivate = data;
+
+    buf->driScreenPriv = screen;
+
+    buf->row = _mesa_malloc(MAX_WIDTH * 4);
+
+    /* basic framebuffer setup */
+    _mesa_initialize_framebuffer(&buf->Base, &config->modes);
+
+    /* add front renderbuffer */
+    frontrb = swrast_new_renderbuffer(&config->modes, GL_TRUE);
+    _mesa_add_renderbuffer(&buf->Base, BUFFER_FRONT_LEFT, &frontrb->Base);
+
+    /* add back renderbuffer */
+    if (config->modes.doubleBufferMode) {
+	backrb = swrast_new_renderbuffer(&config->modes, GL_FALSE);
+	_mesa_add_renderbuffer(&buf->Base, BUFFER_BACK_LEFT, &backrb->Base);
+    }
+
+    /* add software renderbuffers */
+    _mesa_add_soft_renderbuffers(&buf->Base,
+				 GL_FALSE, /* color */
+				 config->modes.haveDepthBuffer,
+				 config->modes.haveStencilBuffer,
+				 config->modes.haveAccumBuffer,
+				 GL_FALSE, /* alpha */
+				 GL_FALSE /* aux bufs */);
+
+    return buf;
+}
+
+static void
+driDestroyDrawable(__DRIdrawable *buf)
+{
+    TRACE;
+
+    if (buf) {
+	struct gl_framebuffer *fb = &buf->Base;
+
+	_mesa_free(buf->row);
+
+	fb->DeletePending = GL_TRUE;
+	_mesa_unreference_framebuffer(&fb);
+    }
+}
+
+static void driSwapBuffers(__DRIdrawable *buf)
+{
+    GET_CURRENT_CONTEXT(ctx);
+
+    struct swrast_renderbuffer *frontrb =
+	swrast_renderbuffer(buf->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer);
+    struct swrast_renderbuffer *backrb =
+	swrast_renderbuffer(buf->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer);
+
+    __DRIscreen *screen = buf->driScreenPriv;
+
+    TRACE;
+
+    /* check for signle-buffered */
+    if (backrb == NULL)
+	return;
+
+    /* check if swapping currently bound buffer */
+    if (ctx && ctx->DrawBuffer == &(buf->Base)) {
+	/* flush pending rendering */
+	_mesa_notifySwapBuffers(ctx);
+    }
+
+    screen->swrast_loader->putImage(buf, __DRI_SWRAST_IMAGE_OP_SWAP,
+				    0, 0,
+				    frontrb->Base.Width,
+				    frontrb->Base.Height,
+				    backrb->Base.Data,
+				    buf->loaderPrivate);
+}
+
+
+/**
+ * General device driver functions.
+ */
+
+static void
+get_window_size( GLframebuffer *fb, GLsizei *w, GLsizei *h )
+{
+    __DRIdrawable *buf = swrast_drawable(fb);
+    __DRIscreen *screen = buf->driScreenPriv;
+    int x, y;
+
+    screen->swrast_loader->getDrawableInfo(buf,
+					   &x, &y, w, h,
+					   buf->loaderPrivate);
+}
+
+static void
+swrast_check_and_update_window_size( GLcontext *ctx, GLframebuffer *fb )
+{
+    GLsizei width, height;
+
+    get_window_size(fb, &width, &height);
+    if (fb->Width != width || fb->Height != height) {
+	_mesa_resize_framebuffer(ctx, fb, width, height);
+    }
+}
+
+static const GLubyte *
+get_string(GLcontext *ctx, GLenum pname)
+{
+    (void) ctx;
+    switch (pname) {
+	case GL_VENDOR:
+	    return (const GLubyte *) "Mesa Project";
+	case GL_RENDERER:
+	    return (const GLubyte *) "Software Rasterizer";
+	default:
+	    return NULL;
+    }
+}
+
+static void
+update_state( GLcontext *ctx, GLuint new_state )
+{
+    /* not much to do here - pass it on */
+    _swrast_InvalidateState( ctx, new_state );
+    _swsetup_InvalidateState( ctx, new_state );
+    _vbo_InvalidateState( ctx, new_state );
+    _tnl_InvalidateState( ctx, new_state );
+}
+
+static void
+viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+{
+    GLframebuffer *draw = ctx->WinSysDrawBuffer;
+    GLframebuffer *read = ctx->WinSysReadBuffer;
+
+    swrast_check_and_update_window_size(ctx, draw);
+    swrast_check_and_update_window_size(ctx, read);
+}
+
+static void
+swrast_init_driver_functions(struct dd_function_table *driver)
+{
+    driver->GetString = get_string;
+    driver->UpdateState = update_state;
+    driver->GetBufferSize = NULL;
+    driver->Viewport = viewport;
+}
+
+
+/**
+ * Context-related functions.
+ */
+
+static __DRIcontext *
+driCreateNewContext(__DRIscreen *screen, const __DRIconfig *config,
+		    __DRIcontext *shared, void *data)
+{
+    __DRIcontext *ctx;
+    GLcontext *mesaCtx;
+    struct dd_function_table functions;
+
+    TRACE;
+
+    ctx = _mesa_calloc(sizeof *ctx);
+    if (!ctx)
+	return NULL;
+
+    ctx->loaderPrivate = data;
+
+    ctx->driScreenPriv = screen;
+
+    /* build table of device driver functions */
+    _mesa_init_driver_functions(&functions);
+    swrast_init_driver_functions(&functions);
+
+    if (!_mesa_initialize_context(&ctx->Base, &config->modes,
+				  shared ? &shared->Base : NULL,
+				  &functions, (void *) ctx)) {
+      _mesa_free(ctx);
+      return NULL;
+    }
+
+    mesaCtx = &ctx->Base;
+
+    /* do bounds checking to prevent segfaults and server crashes! */
+    mesaCtx->Const.CheckArrayBounds = GL_TRUE;
+
+    /* create module contexts */
+    _swrast_CreateContext( mesaCtx );
+    _vbo_CreateContext( mesaCtx );
+    _tnl_CreateContext( mesaCtx );
+    _swsetup_CreateContext( mesaCtx );
+    _swsetup_Wakeup( mesaCtx );
+
+    /* use default TCL pipeline */
+    {
+       TNLcontext *tnl = TNL_CONTEXT(mesaCtx);
+       tnl->Driver.RunPipeline = _tnl_run_pipeline;
+    }
+
+    _mesa_enable_sw_extensions(mesaCtx);
+    _mesa_enable_1_3_extensions(mesaCtx);
+    _mesa_enable_1_4_extensions(mesaCtx);
+    _mesa_enable_1_5_extensions(mesaCtx);
+    _mesa_enable_2_0_extensions(mesaCtx);
+    _mesa_enable_2_1_extensions(mesaCtx);
+
+    return ctx;
+}
+
+static void
+driDestroyContext(__DRIcontext *ctx)
+{
+    GLcontext *mesaCtx;
+    TRACE;
+
+    if (ctx) {
+	mesaCtx = &ctx->Base;
+	_swsetup_DestroyContext( mesaCtx );
+	_swrast_DestroyContext( mesaCtx );
+	_tnl_DestroyContext( mesaCtx );
+	_vbo_DestroyContext( mesaCtx );
+	_mesa_destroy_context( mesaCtx );
+    }
+}
+
+static int
+driCopyContext(__DRIcontext *dst, __DRIcontext *src, unsigned long mask)
+{
+    TRACE;
+
+    _mesa_copy_context(&src->Base, &dst->Base, mask);
+    return GL_TRUE;
+}
+
+static int driBindContext(__DRIcontext *ctx,
+			  __DRIdrawable *draw,
+			  __DRIdrawable *read)
+{
+    GLcontext *mesaCtx;
+    GLframebuffer *mesaDraw;
+    GLframebuffer *mesaRead;
+    TRACE;
+
+    if (ctx) {
+	if (!draw || !read)
+	    return GL_FALSE;
+
+	mesaCtx = &ctx->Base;
+	mesaDraw = &draw->Base;
+	mesaRead = &read->Base;
+
+	/* check for same context and buffer */
+	if (mesaCtx == _mesa_get_current_context()
+	    && mesaCtx->DrawBuffer == mesaDraw
+	    && mesaCtx->ReadBuffer == mesaRead) {
+	    return GL_TRUE;
+	}
+
+	_glapi_check_multithread();
+
+	swrast_check_and_update_window_size(mesaCtx, mesaDraw);
+	if (read != draw)
+	    swrast_check_and_update_window_size(mesaCtx, mesaRead);
+
+	_mesa_make_current( mesaCtx,
+			    mesaDraw,
+			    mesaRead );
+    }
+    else {
+	/* unbind */
+	_mesa_make_current( NULL, NULL, NULL );
+    }
+
+    return GL_TRUE;
+}
+
+static int driUnbindContext(__DRIcontext *ctx)
+{
+    TRACE;
+    (void) ctx;
+    return GL_TRUE;
+}
+
+
+static const __DRIcoreExtension driCoreExtension = {
+    { __DRI_CORE, __DRI_CORE_VERSION },
+    NULL, /* driCreateNewScreen */
+    driDestroyScreen,
+    driGetExtensions,
+    driGetConfigAttrib,
+    driIndexConfigAttrib,
+    NULL, /* driCreateNewDrawable */
+    driDestroyDrawable,
+    driSwapBuffers,
+    driCreateNewContext,
+    driCopyContext,
+    driDestroyContext,
+    driBindContext,
+    driUnbindContext
+};
+
+static const __DRIswrastExtension driSWRastExtension = {
+    { __DRI_SWRAST, __DRI_SWRAST_VERSION },
+    driCreateNewScreen,
+    driCreateNewDrawable
+};
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driSWRastExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/swrast/swrast_priv.h b/src/mesa/drivers/dri/swrast/swrast_priv.h
new file mode 100644
index 00000000000..a707ffc40a8
--- /dev/null
+++ b/src/mesa/drivers/dri/swrast/swrast_priv.h
@@ -0,0 +1,142 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.1
+ *
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Authors:
+ *    George Sapountzis <gsap7@yahoo.gr>
+ */
+
+
+#ifndef _SWRAST_PRIV_H
+#define _SWRAST_PRIV_H
+
+#include <GL/gl.h>
+#include <GL/internal/dri_interface.h>
+#include "main/mtypes.h"
+
+
+/**
+ * Debugging
+ */
+#define DEBUG_CORE	0
+#define DEBUG_SPAN	0
+
+#if DEBUG_CORE
+#define TRACE _mesa_printf("--> %s\n", __FUNCTION__)
+#else
+#define TRACE
+#endif
+
+#if DEBUG_SPAN
+#define TRACE_SPAN _mesa_printf("--> %s\n", __FUNCTION__)
+#else
+#define TRACE_SPAN
+#endif
+
+
+/**
+ * Data types
+ */
+struct __DRIscreenRec {
+    int num;
+
+    const __DRIextension **extensions;
+
+    const __DRIswrastLoaderExtension *swrast_loader;
+};
+
+struct __DRIcontextRec {
+    GLcontext Base;
+
+    void *loaderPrivate;
+
+    __DRIscreen *driScreenPriv;
+};
+
+struct __DRIdrawableRec {
+    GLframebuffer Base;
+
+    void *loaderPrivate;
+
+    __DRIscreen *driScreenPriv;
+
+    /* scratch row for optimized front-buffer rendering */
+    char *row;
+};
+
+struct swrast_renderbuffer {
+    struct gl_renderbuffer Base;
+
+    /* renderbuffer pitch (in bytes) */
+    GLuint pitch;
+};
+
+static INLINE __DRIcontext *
+swrast_context(GLcontext *ctx)
+{
+    return (__DRIcontext *) ctx;
+}
+
+static INLINE __DRIdrawable *
+swrast_drawable(GLframebuffer *fb)
+{
+    return (__DRIdrawable *) fb;
+}
+
+static INLINE struct swrast_renderbuffer *
+swrast_renderbuffer(struct gl_renderbuffer *rb)
+{
+    return (struct swrast_renderbuffer *) rb;
+}
+
+
+/**
+ * Pixel formats we support
+ */
+#define PF_CI8        1		/**< Color Index mode */
+#define PF_A8R8G8B8   2		/**< 32-bit TrueColor:  8-A, 8-R, 8-G, 8-B bits */
+#define PF_R5G6B5     3		/**< 16-bit TrueColor:  5-R, 6-G, 5-B bits */
+#define PF_R3G3B2     4		/**<  8-bit TrueColor:  3-R, 3-G, 2-B bits */
+
+
+/**
+ * Renderbuffer pitch alignment (in bits).
+ *
+ * The xorg loader requires padding images to 32 bits. However, this should
+ * become a screen/drawable parameter XXX
+ */
+#define PITCH_ALIGN_BITS 32
+
+
+/* swrast_span.c */
+
+extern void
+swrast_set_span_funcs_back(struct swrast_renderbuffer *xrb,
+			   GLuint pixel_format);
+
+extern void
+swrast_set_span_funcs_front(struct swrast_renderbuffer *xrb,
+			    GLuint pixel_format);
+
+#endif /* _SWRAST_PRIV_H_ */
diff --git a/src/mesa/drivers/dri/swrast/swrast_span.c b/src/mesa/drivers/dri/swrast/swrast_span.c
new file mode 100644
index 00000000000..5e990368b2e
--- /dev/null
+++ b/src/mesa/drivers/dri/swrast/swrast_span.c
@@ -0,0 +1,367 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.1
+ *
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Authors:
+ *    George Sapountzis <gsap7@yahoo.gr>
+ */
+
+#include "swrast_priv.h"
+
+#define YFLIP(_xrb, Y) ((_xrb)->Base.Height - (Y) - 1)
+
+/*
+ * Dithering support takes the "computation" extreme in the "computation vs.
+ * storage" trade-off. This approach is very simple to implement and any
+ * computational overhead should be acceptable. XMesa uses table lookups for
+ * around 8KB of storage overhead per visual.
+ */
+#define DITHER 1
+
+static const GLubyte kernel[16] = {
+    0*16,  8*16,  2*16, 10*16,
+   12*16,  4*16, 14*16,  6*16,
+    3*16, 11*16,  1*16,  9*16,
+   15*16,  7*16, 13*16,  5*16,
+};
+
+#if DITHER
+#define DITHER_COMP(X, Y) kernel[((X) & 0x3) | (((Y) & 0x3) << 2)]
+
+#define DITHER_CLAMP(X) (((X) < CHAN_MAX) ? (X) : CHAN_MAX)
+#else
+#define DITHER_COMP(X, Y) 0
+
+#define DITHER_CLAMP(X) (X)
+#endif
+
+
+/*
+ * Pixel macros shared across front/back buffer span functions.
+ */
+
+/* 32-bit BGRA */
+#define STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE) \
+   DST[3] = VALUE[ACOMP]; \
+   DST[2] = VALUE[RCOMP]; \
+   DST[1] = VALUE[GCOMP]; \
+   DST[0] = VALUE[BCOMP]
+#define STORE_PIXEL_RGB_A8R8G8B8(DST, X, Y, VALUE) \
+   DST[3] = 0xff; \
+   DST[2] = VALUE[RCOMP]; \
+   DST[1] = VALUE[GCOMP]; \
+   DST[0] = VALUE[BCOMP]
+#define FETCH_PIXEL_A8R8G8B8(DST, SRC) \
+   DST[ACOMP] = SRC[3]; \
+   DST[RCOMP] = SRC[2]; \
+   DST[GCOMP] = SRC[1]; \
+   DST[BCOMP] = SRC[0]
+
+
+/* 16-bit BGR */
+#define STORE_PIXEL_R5G6B5(DST, X, Y, VALUE) \
+   do { \
+   int d = DITHER_COMP(X, Y) >> 6; \
+   GLushort *p = (GLushort *)DST; \
+   *p = ( ((DITHER_CLAMP((VALUE[RCOMP]) + d) & 0xf8) << 8) | \
+	  ((DITHER_CLAMP((VALUE[GCOMP]) + d) & 0xfc) << 3) | \
+	  ((DITHER_CLAMP((VALUE[BCOMP]) + d) & 0xf8) >> 3) ); \
+   } while(0)
+#define FETCH_PIXEL_R5G6B5(DST, SRC) \
+   do { \
+   GLushort p = *(GLushort *)SRC; \
+   DST[ACOMP] = 0xff; \
+   DST[RCOMP] = ((p >> 8) & 0xf8) * 255 / 0xf8; \
+   DST[GCOMP] = ((p >> 3) & 0xfc) * 255 / 0xfc; \
+   DST[BCOMP] = ((p << 3) & 0xf8) * 255 / 0xf8; \
+   } while(0)
+
+
+/* 8-bit BGR */
+#define STORE_PIXEL_R3G3B2(DST, X, Y, VALUE) \
+   do { \
+   int d = DITHER_COMP(X, Y) >> 3; \
+   GLubyte *p = (GLubyte *)DST; \
+   *p = ( ((DITHER_CLAMP((VALUE[RCOMP]) + d) & 0xe0) >> 5) | \
+	  ((DITHER_CLAMP((VALUE[GCOMP]) + d) & 0xe0) >> 2) | \
+	  ((DITHER_CLAMP((VALUE[BCOMP]) + d) & 0xc0) >> 0) ); \
+   } while(0)
+#define FETCH_PIXEL_R3G3B2(DST, SRC) \
+   do { \
+   GLubyte p = *(GLubyte *)SRC; \
+   DST[ACOMP] = 0xff; \
+   DST[RCOMP] = ((p << 5) & 0xe0) * 255 / 0xe0; \
+   DST[GCOMP] = ((p << 2) & 0xe0) * 255 / 0xe0; \
+   DST[BCOMP] = ((p << 0) & 0xc0) * 255 / 0xc0; \
+   } while(0)
+
+
+/*
+ * Generate code for back-buffer span functions.
+ */
+
+/* 32-bit BGRA */
+#define NAME(FUNC) FUNC##_A8R8G8B8
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 4;
+#define INC_PIXEL_PTR(P) P += 4
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE)
+#define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
+   STORE_PIXEL_RGB_A8R8G8B8(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_A8R8G8B8(DST, SRC)
+
+#include "swrast/s_spantemp.h"
+
+
+/* 16-bit BGR */
+#define NAME(FUNC) FUNC##_R5G6B5
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 2;
+#define INC_PIXEL_PTR(P) P += 2
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_R5G6B5(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_R5G6B5(DST, SRC)
+
+#include "swrast/s_spantemp.h"
+
+
+/* 8-bit BGR */
+#define NAME(FUNC) FUNC##_R3G3B2
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 1;
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_R3G3B2(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_R3G3B2(DST, SRC)
+
+#include "swrast/s_spantemp.h"
+
+
+/* 8-bit color index */
+#define NAME(FUNC) FUNC##_CI8
+#define CI_MODE
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X);
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   *DST = VALUE[0]
+#define FETCH_PIXEL(DST, SRC) \
+   DST = SRC[0]
+
+#include "swrast/s_spantemp.h"
+
+
+/*
+ * Generate code for front-buffer span functions.
+ */
+
+/* 32-bit BGRA */
+#define NAME(FUNC) FUNC##_A8R8G8B8_front
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)row;
+#define INC_PIXEL_PTR(P) P += 4
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE)
+#define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
+   STORE_PIXEL_RGB_A8R8G8B8(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_A8R8G8B8(DST, SRC)
+
+#include "swrast_spantemp.h"
+
+
+/* 16-bit BGR */
+#define NAME(FUNC) FUNC##_R5G6B5_front
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)row;
+#define INC_PIXEL_PTR(P) P += 2
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_R5G6B5(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_R5G6B5(DST, SRC)
+
+#include "swrast_spantemp.h"
+
+
+/* 8-bit BGR */
+#define NAME(FUNC) FUNC##_R3G3B2_front
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)row;
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_R3G3B2(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_R3G3B2(DST, SRC)
+
+#include "swrast_spantemp.h"
+
+
+/* 8-bit color index */
+#define NAME(FUNC) FUNC##_CI8_front
+#define CI_MODE
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)row;
+#define INC_PIXEL_PTR(P) P += 1
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   *DST = VALUE[0]
+#define FETCH_PIXEL(DST, SRC) \
+   DST = SRC[0]
+
+#include "swrast_spantemp.h"
+
+
+/*
+ * Back-buffers are malloced memory and always private.
+ *
+ * BACK_PIXMAP (not supported)
+ * BACK_XIMAGE
+ */
+void
+swrast_set_span_funcs_back(struct swrast_renderbuffer *xrb,
+			   GLuint pixel_format)
+{
+    switch (pixel_format) {
+    case PF_A8R8G8B8:
+	xrb->Base.GetRow = get_row_A8R8G8B8;
+	xrb->Base.GetValues = get_values_A8R8G8B8;
+	xrb->Base.PutRow = put_row_A8R8G8B8;
+	xrb->Base.PutRowRGB = put_row_rgb_A8R8G8B8;
+	xrb->Base.PutMonoRow = put_mono_row_A8R8G8B8;
+	xrb->Base.PutValues = put_values_A8R8G8B8;
+	xrb->Base.PutMonoValues = put_mono_values_A8R8G8B8;
+	break;
+    case PF_R5G6B5:
+	xrb->Base.GetRow = get_row_R5G6B5;
+	xrb->Base.GetValues = get_values_R5G6B5;
+	xrb->Base.PutRow = put_row_R5G6B5;
+	xrb->Base.PutRowRGB = put_row_rgb_R5G6B5;
+	xrb->Base.PutMonoRow = put_mono_row_R5G6B5;
+	xrb->Base.PutValues = put_values_R5G6B5;
+	xrb->Base.PutMonoValues = put_mono_values_R5G6B5;
+	break;
+    case PF_R3G3B2:
+	xrb->Base.GetRow = get_row_R3G3B2;
+	xrb->Base.GetValues = get_values_R3G3B2;
+	xrb->Base.PutRow = put_row_R3G3B2;
+	xrb->Base.PutRowRGB = put_row_rgb_R3G3B2;
+	xrb->Base.PutMonoRow = put_mono_row_R3G3B2;
+	xrb->Base.PutValues = put_values_R3G3B2;
+	xrb->Base.PutMonoValues = put_mono_values_R3G3B2;
+	break;
+    case PF_CI8:
+	xrb->Base.GetRow = get_row_CI8;
+	xrb->Base.GetValues = get_values_CI8;
+	xrb->Base.PutRow = put_row_CI8;
+	xrb->Base.PutMonoRow = put_mono_row_CI8;
+	xrb->Base.PutValues = put_values_CI8;
+	xrb->Base.PutMonoValues = put_mono_values_CI8;
+	break;
+    default:
+	assert(0);
+	return;
+    }
+}
+
+
+/*
+ * Front-buffers are provided by the loader, the xorg loader uses pixmaps.
+ *
+ * WINDOW,          An X window
+ * GLXWINDOW,       GLX window
+ * PIXMAP,          GLX pixmap
+ * PBUFFER          GLX Pbuffer
+ */
+void
+swrast_set_span_funcs_front(struct swrast_renderbuffer *xrb,
+			    GLuint pixel_format)
+{
+    switch (pixel_format) {
+    case PF_A8R8G8B8:
+	xrb->Base.GetRow = get_row_A8R8G8B8_front;
+	xrb->Base.GetValues = get_values_A8R8G8B8_front;
+	xrb->Base.PutRow = put_row_A8R8G8B8_front;
+	xrb->Base.PutRowRGB = put_row_rgb_A8R8G8B8_front;
+	xrb->Base.PutMonoRow = put_mono_row_A8R8G8B8_front;
+	xrb->Base.PutValues = put_values_A8R8G8B8_front;
+	xrb->Base.PutMonoValues = put_mono_values_A8R8G8B8_front;
+	break;
+    case PF_R5G6B5:
+	xrb->Base.GetRow = get_row_R5G6B5_front;
+	xrb->Base.GetValues = get_values_R5G6B5_front;
+	xrb->Base.PutRow = put_row_R5G6B5_front;
+	xrb->Base.PutRowRGB = put_row_rgb_R5G6B5_front;
+	xrb->Base.PutMonoRow = put_mono_row_R5G6B5_front;
+	xrb->Base.PutValues = put_values_R5G6B5_front;
+	xrb->Base.PutMonoValues = put_mono_values_R5G6B5_front;
+	break;
+    case PF_R3G3B2:
+	xrb->Base.GetRow = get_row_R3G3B2_front;
+	xrb->Base.GetValues = get_values_R3G3B2_front;
+	xrb->Base.PutRow = put_row_R3G3B2_front;
+	xrb->Base.PutRowRGB = put_row_rgb_R3G3B2_front;
+	xrb->Base.PutMonoRow = put_mono_row_R3G3B2_front;
+	xrb->Base.PutValues = put_values_R3G3B2_front;
+	xrb->Base.PutMonoValues = put_mono_values_R3G3B2_front;
+	break;
+    case PF_CI8:
+	xrb->Base.GetRow = get_row_CI8_front;
+	xrb->Base.GetValues = get_values_CI8_front;
+	xrb->Base.PutRow = put_row_CI8_front;
+	xrb->Base.PutMonoRow = put_mono_row_CI8_front;
+	xrb->Base.PutValues = put_values_CI8_front;
+	xrb->Base.PutMonoValues = put_mono_values_CI8_front;
+	break;
+    default:
+	assert(0);
+	return;
+    }
+}
diff --git a/src/mesa/drivers/dri/swrast/swrast_spantemp.h b/src/mesa/drivers/dri/swrast/swrast_spantemp.h
new file mode 100644
index 00000000000..e0cb2414294
--- /dev/null
+++ b/src/mesa/drivers/dri/swrast/swrast_spantemp.h
@@ -0,0 +1,328 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  6.5.1
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/*
+ * Modified version of swrast/s_spantemp.h for front-buffer rendering. The
+ * no-mask paths use a scratch row to avoid repeated calls to the loader.
+ *
+ * For the mask paths we always use an array of 4 elements of RB_TYPE. This is
+ * to satisfy the xorg loader requirement of an image pitch of 32 bits and
+ * should be ok for other loaders also.
+ */
+
+
+#ifndef _SWRAST_SPANTEMP_ONCE
+#define _SWRAST_SPANTEMP_ONCE
+
+static INLINE void
+PUT_PIXEL( GLcontext *glCtx, GLint x, GLint y, GLubyte *p )
+{
+    __DRIcontext *ctx = swrast_context(glCtx);
+    __DRIdrawable *draw = swrast_drawable(glCtx->DrawBuffer);
+
+    __DRIscreen *screen = ctx->driScreenPriv;
+
+    screen->swrast_loader->putImage(draw, __DRI_SWRAST_IMAGE_OP_DRAW,
+				    x, y, 1, 1, (char *)p,
+				    draw->loaderPrivate);
+}
+
+
+static INLINE void
+GET_PIXEL( GLcontext *glCtx, GLint x, GLint y, GLubyte *p )
+{
+    __DRIcontext *ctx = swrast_context(glCtx);
+    __DRIdrawable *read = swrast_drawable(glCtx->ReadBuffer);
+
+    __DRIscreen *screen = ctx->driScreenPriv;
+
+    screen->swrast_loader->getImage(read, x, y, 1, 1, (char *)p,
+				    read->loaderPrivate);
+}
+
+static INLINE void
+PUT_ROW( GLcontext *glCtx, GLint x, GLint y, GLuint n, char *row )
+{
+    __DRIcontext *ctx = swrast_context(glCtx);
+    __DRIdrawable *draw = swrast_drawable(glCtx->DrawBuffer);
+
+    __DRIscreen *screen = ctx->driScreenPriv;
+
+    screen->swrast_loader->putImage(draw, __DRI_SWRAST_IMAGE_OP_DRAW,
+				    x, y, n, 1, row,
+				    draw->loaderPrivate);
+}
+
+static INLINE void
+GET_ROW( GLcontext *glCtx, GLint x, GLint y, GLuint n, char *row )
+{
+    __DRIcontext *ctx = swrast_context(glCtx);
+    __DRIdrawable *read = swrast_drawable(glCtx->ReadBuffer);
+
+    __DRIscreen *screen = ctx->driScreenPriv;
+
+    screen->swrast_loader->getImage(read, x, y, n, 1, row,
+				    read->loaderPrivate);
+}
+
+#endif /* _SWRAST_SPANTEMP_ONCE */
+
+
+/*
+ * Templates for the span/pixel-array write/read functions called via
+ * the gl_renderbuffer's GetRow, GetValues, PutRow, PutMonoRow, PutValues
+ * and PutMonoValues functions.
+ *
+ * Define the following macros before including this file:
+ *   NAME(BASE)  to generate the function name (i.e. add prefix or suffix)
+ *   RB_TYPE  the renderbuffer DataType
+ *   CI_MODE  if set, color index mode, else RGBA
+ *   SPAN_VARS  to declare any local variables
+ *   INIT_PIXEL_PTR(P, X, Y)  to initialize a pointer to a pixel
+ *   INC_PIXEL_PTR(P)  to increment a pixel pointer by one pixel
+ *   STORE_PIXEL(DST, X, Y, VALUE)  to store pixel values in buffer
+ *   FETCH_PIXEL(DST, SRC)  to fetch pixel values from buffer
+ *
+ * Note that in the STORE_PIXEL macros, we also pass in the (X,Y) coordinates
+ * for the pixels to be stored.  This is useful when dithering and probably
+ * ignored otherwise.
+ */
+
+#include "main/macros.h"
+
+
+#ifdef CI_MODE
+#define RB_COMPONENTS 1
+#elif !defined(RB_COMPONENTS)
+#define RB_COMPONENTS 4
+#endif
+
+
+static void
+NAME(get_row)( GLcontext *ctx, struct gl_renderbuffer *rb,
+               GLuint count, GLint x, GLint y, void *values )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+#ifdef CI_MODE
+   RB_TYPE *dest = (RB_TYPE *) values;
+#else
+   RB_TYPE (*dest)[RB_COMPONENTS] = (RB_TYPE (*)[RB_COMPONENTS]) values;
+#endif
+   GLuint i;
+   char *row = swrast_drawable(ctx->ReadBuffer)->row;
+   INIT_PIXEL_PTR(pixel, x, y);
+   GET_ROW( ctx, x, YFLIP(xrb, y), count, row );
+   for (i = 0; i < count; i++) {
+      FETCH_PIXEL(dest[i], pixel);
+      INC_PIXEL_PTR(pixel);
+   }
+   (void) rb;
+}
+
+
+static void
+NAME(get_values)( GLcontext *ctx, struct gl_renderbuffer *rb,
+                  GLuint count, const GLint x[], const GLint y[], void *values )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+#ifdef CI_MODE
+   RB_TYPE *dest = (RB_TYPE *) values;
+#else
+   RB_TYPE (*dest)[RB_COMPONENTS] = (RB_TYPE (*)[RB_COMPONENTS]) values;
+#endif
+   GLuint i;
+   for (i = 0; i < count; i++) {
+      RB_TYPE pixel[4];
+      GET_PIXEL(ctx, x[i], YFLIP(xrb, y[i]), pixel);
+      FETCH_PIXEL(dest[i], pixel);
+   }
+   (void) rb;
+}
+
+
+static void
+NAME(put_row)( GLcontext *ctx, struct gl_renderbuffer *rb,
+               GLuint count, GLint x, GLint y,
+               const void *values, const GLubyte mask[] )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+   const RB_TYPE (*src)[RB_COMPONENTS] = (const RB_TYPE (*)[RB_COMPONENTS]) values;
+   GLuint i;
+   if (mask) {
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            RB_TYPE pixel[4];
+            STORE_PIXEL(pixel, x + i, y, src[i]);
+            PUT_PIXEL(ctx, x + i, YFLIP(xrb, y), pixel);
+         }
+      }
+   }
+   else {
+      char *row = swrast_drawable(ctx->DrawBuffer)->row;
+      INIT_PIXEL_PTR(pixel, x, y);
+      for (i = 0; i < count; i++) {
+         STORE_PIXEL(pixel, x + i, y, src[i]);
+         INC_PIXEL_PTR(pixel);
+      }
+      PUT_ROW( ctx, x, YFLIP(xrb, y), count, row );
+   }
+   (void) rb;
+}
+
+
+#if !defined(CI_MODE)
+static void
+NAME(put_row_rgb)( GLcontext *ctx, struct gl_renderbuffer *rb,
+                   GLuint count, GLint x, GLint y,
+                   const void *values, const GLubyte mask[] )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+   const RB_TYPE (*src)[3] = (const RB_TYPE (*)[3]) values;
+   GLuint i;
+   if (mask) {
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            RB_TYPE pixel[4];
+#ifdef STORE_PIXEL_RGB
+            STORE_PIXEL_RGB(pixel, x + i, y, src[i]);
+#else
+            STORE_PIXEL(pixel, x + i, y, src[i]);
+#endif
+            PUT_PIXEL(ctx, x + i, YFLIP(xrb, y), pixel);
+         }
+      }
+   }
+   else {
+      char *row = swrast_drawable(ctx->DrawBuffer)->row;
+      INIT_PIXEL_PTR(pixel, x, y);
+      for (i = 0; i < count; i++) {
+#ifdef STORE_PIXEL_RGB
+         STORE_PIXEL_RGB(pixel, x + i, y, src[i]);
+#else
+         STORE_PIXEL(pixel, x + i, y, src[i]);
+#endif
+         INC_PIXEL_PTR(pixel);
+      }
+      PUT_ROW( ctx, x, YFLIP(xrb, y), count, row );
+   }
+   (void) rb;
+}
+#endif
+
+
+static void
+NAME(put_mono_row)( GLcontext *ctx, struct gl_renderbuffer *rb,
+                    GLuint count, GLint x, GLint y,
+                    const void *value, const GLubyte mask[] )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+   const RB_TYPE *src = (const RB_TYPE *) value;
+   GLuint i;
+   if (mask) {
+      for (i = 0; i < count; i++) {
+         if (mask[i]) {
+            RB_TYPE pixel[4];
+            STORE_PIXEL(pixel, x + i, y, src);
+            PUT_PIXEL(ctx, x + i, YFLIP(xrb, y), pixel);
+         }
+      }
+   }
+   else {
+      char *row = swrast_drawable(ctx->DrawBuffer)->row;
+      INIT_PIXEL_PTR(pixel, x, y);
+      for (i = 0; i < count; i++) {
+         STORE_PIXEL(pixel, x + i, y, src);
+         INC_PIXEL_PTR(pixel);
+      }
+      PUT_ROW( ctx, x, YFLIP(xrb, y), count, row );
+   }
+   (void) rb;
+}
+
+
+static void
+NAME(put_values)( GLcontext *ctx, struct gl_renderbuffer *rb,
+                  GLuint count, const GLint x[], const GLint y[],
+                  const void *values, const GLubyte mask[] )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+   const RB_TYPE (*src)[RB_COMPONENTS] = (const RB_TYPE (*)[RB_COMPONENTS]) values;
+   GLuint i;
+   ASSERT(mask);
+   for (i = 0; i < count; i++) {
+      if (mask[i]) {
+         RB_TYPE pixel[4];
+         STORE_PIXEL(pixel, x[i], y[i], src[i]);
+         PUT_PIXEL(ctx, x[i], YFLIP(xrb, y[i]), pixel);
+      }
+   }
+   (void) rb;
+}
+
+
+static void
+NAME(put_mono_values)( GLcontext *ctx, struct gl_renderbuffer *rb,
+                       GLuint count, const GLint x[], const GLint y[],
+                       const void *value, const GLubyte mask[] )
+{
+#ifdef SPAN_VARS
+   SPAN_VARS
+#endif
+   const RB_TYPE *src = (const RB_TYPE *) value;
+   GLuint i;
+   ASSERT(mask);
+   for (i = 0; i < count; i++) {
+      if (mask[i]) {
+         RB_TYPE pixel[4];
+         STORE_PIXEL(pixel, x[i], y[i], src);
+         PUT_PIXEL(ctx, x[i], YFLIP(xrb, y[i]), pixel);
+      }
+   }
+   (void) rb;
+}
+
+
+#undef NAME
+#undef RB_TYPE
+#undef RB_COMPONENTS
+#undef CI_MODE
+#undef SPAN_VARS
+#undef INIT_PIXEL_PTR
+#undef INC_PIXEL_PTR
+#undef STORE_PIXEL
+#undef STORE_PIXEL_RGB
+#undef FETCH_PIXEL
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.c b/src/mesa/drivers/dri/tdfx/tdfx_context.c
index b4eea2566f6..ef688d103d3 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.c
@@ -43,9 +43,9 @@
 #include "tdfx_render.h"
 #include "tdfx_span.h"
 #include "tdfx_texman.h"
-#include "extensions.h"
-#include "hash.h"
-#include "texobj.h"
+#include "main/extensions.h"
+#include "main/hash.h"
+#include "main/texobj.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.h b/src/mesa/drivers/dri/tdfx/tdfx_context.h
index 05673cd1867..3bcb5451193 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.h
@@ -44,7 +44,7 @@
 #ifdef XFree86Server
 #include "GL/xf86glx.h"
 #else
-#include "glheader.h"
+#include "main/glheader.h"
 #endif
 #if defined(__linux__)
 #include <signal.h>
@@ -55,12 +55,12 @@
 #include "tdfx_glide.h"
 #include "xmlconfig.h"
 
-#include "clip.h"
-#include "context.h"
-#include "macros.h"
-#include "matrix.h"
-#include "imports.h"
-#include "mtypes.h"
+#include "main/clip.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/matrix.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 
 #include "tdfx_screen.h"
 
@@ -983,9 +983,9 @@ FX_grColorMaskv_NoLock(GLcontext *ctx, const GLboolean rgba[4]);
 #define TDFXPACKCOLOR4444( r, g, b, a )					   \
    ((((a) & 0xf0) << 8) | (((b) & 0xf0) << 4) | ((g) & 0xf0) | ((r) >> 4))
 
-static __inline__ GrColor_t tdfxPackColor( GLuint cpp,
-					   GLubyte r, GLubyte g,
-					   GLubyte b, GLubyte a )
+static INLINE GrColor_t tdfxPackColor( GLuint cpp,
+                                       GLubyte r, GLubyte g,
+                                       GLubyte b, GLubyte a )
 {
    switch ( cpp ) {
    case 2:
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_dd.c b/src/mesa/drivers/dri/tdfx/tdfx_dd.c
index adbe0c0f33f..2cef0795153 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_dd.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_dd.c
@@ -39,9 +39,9 @@
 #include "tdfx_pixels.h"
 
 #include "utils.h"
-#include "context.h"
-#include "enums.h"
-#include "framebuffer.h"
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/framebuffer.h"
 #include "swrast/swrast.h"
 #if defined(USE_X86_ASM)
 #include "x86/common_x86_asm.h"
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_dd.h b/src/mesa/drivers/dri/tdfx/tdfx_dd.h
index bd61e106052..f419c8426af 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_dd.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_dd.h
@@ -36,7 +36,7 @@
 #ifndef __TDFX_DD_H__
 #define __TDFX_DD_H__
 
-#include "context.h"
+#include "main/context.h"
 
 extern void tdfxDDInitDriverFuncs( const __GLcontextModes *visual,
                                    struct dd_function_table *functions );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_pixels.c b/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
index b5c01f6ef21..9ab9c05f2bd 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
@@ -44,7 +44,7 @@
 
 #include "swrast/swrast.h"
 
-#include "image.h"
+#include "main/image.h"
 
 
 #define FX_grLfbWriteRegion(fxMesa,dst_buffer,dst_x,dst_y,src_format,src_width,src_height,src_stride,src_data)		\
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_pixels.h b/src/mesa/drivers/dri/tdfx/tdfx_pixels.h
index 55f7eedef89..f5e5427653e 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_pixels.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_pixels.h
@@ -38,7 +38,7 @@
 #ifndef __TDFX_PIXELS_H__
 #define __TDFX_PIXELS_H__
 
-#include "context.h"
+#include "main/context.h"
 
 extern void
 tdfx_bitmap_R5G6B5( GLcontext *ctx, GLint px, GLint py,
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_render.c b/src/mesa/drivers/dri/tdfx/tdfx_render.c
index e374f09df3a..cf840c57a7a 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_render.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_render.c
@@ -274,7 +274,7 @@ static void tdfxClear( GLcontext *ctx, GLbitfield mask )
                                         fxMesa->Color.ClearAlpha,
                                         fxMesa->Depth.Clear);
 	 FX_grColorMaskv_NoLock(ctx, true4);
-	 if (ctx->DrawBuffer->_ColorDrawBufferMask[0] & BUFFER_BIT_FRONT_LEFT)
+	 if (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT)
             fxMesa->Glide.grRenderBuffer(GR_BUFFER_FRONTBUFFER);
 	 if (!ctx->Depth.Test || !ctx->Depth.Mask)
 	    fxMesa->Glide.grDepthMask(FXFALSE);
@@ -294,7 +294,7 @@ static void tdfxClear( GLcontext *ctx, GLbitfield mask )
                fxMesa->Glide.grDepthMask(FXTRUE);
             }
             FX_grColorMaskv_NoLock(ctx, true4);
-            if (ctx->DrawBuffer->_ColorDrawBufferMask[0] & BUFFER_BIT_FRONT_LEFT)
+            if (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT)
                fxMesa->Glide.grRenderBuffer(GR_BUFFER_FRONTBUFFER);
          }
       }
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_screen.c b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
index 77616643944..cd22b849511 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_screen.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
@@ -40,8 +40,8 @@
 #include "tdfx_span.h"
 #include "tdfx_tris.h"
 
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 #include "xmlpool.h"
 
 #include "utils.h"
@@ -62,6 +62,11 @@ DRI_CONF_BEGIN
     DRI_CONF_SECTION_END
 DRI_CONF_END;
 
+static const __DRIextension *tdfxExtensions[] = {
+    &driReadDrawableExtension,
+    NULL
+};
+
 static const GLuint __driNConfigOptions = 1;
 
 extern const struct dri_extension card_extensions[];
@@ -72,9 +77,6 @@ tdfxCreateScreen( __DRIscreenPrivate *sPriv )
 {
    tdfxScreenPrivate *fxScreen;
    TDFXDRIPtr fxDRIPriv = (TDFXDRIPtr) sPriv->pDevPriv;
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void *const psc = sPriv->psc->screenConfigs;
 
    if (sPriv->devPrivSize != sizeof(TDFXDRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(TDFXDRIRec) does not match passed size from device driver\n");
@@ -115,9 +117,7 @@ tdfxCreateScreen( __DRIscreenPrivate *sPriv )
       return GL_FALSE;
    }
 
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension)(psc, "GLX_SGI_make_current_read");
-   }
+   sPriv->extensions = tdfxExtensions;
 
    return GL_TRUE;
 }
@@ -343,36 +343,14 @@ tdfxSwapBuffers( __DRIdrawablePrivate *driDrawPriv )
    }
 }
 
-
-static const struct __DriverAPIRec tdfxAPI = {
-   .InitDriver      = tdfxInitDriver,
-   .DestroyScreen   = tdfxDestroyScreen,
-   .CreateContext   = tdfxCreateContext,
-   .DestroyContext  = tdfxDestroyContext,
-   .CreateBuffer    = tdfxCreateBuffer,
-   .DestroyBuffer   = tdfxDestroyBuffer,
-   .SwapBuffers     = tdfxSwapBuffers,
-   .MakeCurrent     = tdfxMakeCurrent,
-   .UnbindContext   = tdfxUnbindContext,
-   .GetSwapInfo     = NULL,
-   .GetMSC          = NULL,
-   .WaitForMSC      = NULL,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL
-};
-
-
-static __GLcontextModes *tdfxFillInModes(unsigned pixel_bits,
-					 unsigned depth_bits,
-					 unsigned stencil_bits,
-					 GLboolean have_back_buffer)
+static const __DRIconfig **
+tdfxFillInModes(__DRIscreenPrivate *psp,
+		unsigned pixel_bits,
+		unsigned depth_bits,
+		unsigned stencil_bits,
+		GLboolean have_back_buffer)
 {
-	__GLcontextModes *modes;
-	__GLcontextModes *m;
-	unsigned num_modes;
-	unsigned vis[2] = { GLX_TRUE_COLOR, GLX_DIRECT_COLOR };
 	unsigned deep = (depth_bits > 17);
-	unsigned i, db, depth, accum, stencil;
 
 	/* Right now GLX_SWAP_COPY_OML isn't supported, but it would be easy
 	 * enough to add support.  Basically, if a context is created with an
@@ -380,120 +358,96 @@ static __GLcontextModes *tdfxFillInModes(unsigned pixel_bits,
 	 * will never be used.
 	 */
 
-	num_modes = (depth_bits == 16) ? 32 : 16;
-
-	modes = (*dri_interface->createContextModes)(num_modes, sizeof(__GLcontextModes));
-	m = modes;
-
-	for (i = 0; i <= 1; i++) {
-	    for (db = 0; db <= 1; db++) {
-		for (depth = 0; depth <= 1; depth++) {
-		    for (accum = 0; accum <= 1; accum++) {
-			for (stencil = 0; stencil <= !deep; stencil++) {
-			    if (deep) stencil = depth;
-			    m->redBits		= deep ? 8 : 5;
-			    m->greenBits	= deep ? 8 : 6;
-			    m->blueBits		= deep ? 8 : 5;
-			    m->alphaBits	= deep ? 8 : 0;
-			    m->redMask		= deep ?0xFF000000 :0x0000F800;
-			    m->greenMask	= deep ?0x00FF0000 :0x000007E0;
-			    m->blueMask		= deep ?0x0000FF00 :0x0000001F;
-			    m->alphaMask	= deep ? 0x000000FF : 0;
-			    m->rgbBits		= m->redBits + m->greenBits +
-			    			  m->blueBits + m->alphaBits;
-			    m->accumRedBits	= accum ? 16 : 0;
-			    m->accumGreenBits	= accum ? 16 : 0;
-			    m->accumBlueBits	= accum ? 16 : 0;
-			    m->accumAlphaBits	= (accum && deep) ? 16 : 0;
-			    m->stencilBits	= stencil ? 8 : 0;
-			    m->depthBits	= deep
-			    			  ? (depth ? 24 : 0)
-			    			  : (depth ? 0 : depth_bits);
-			    m->visualType	= vis[i];
-			    m->renderType	= GLX_RGBA_BIT;
-			    m->drawableType	= GLX_WINDOW_BIT;
-			    m->rgbMode		= GL_TRUE;
-			    m->doubleBufferMode = db ? GL_TRUE : GL_FALSE;
-			    if (db)
-			    	m->swapMethod = GLX_SWAP_UNDEFINED_OML;
-			    m->visualRating	= ((stencil && !deep) || accum)
-			    			  ? GLX_SLOW_CONFIG
-						  : GLX_NONE;
-			    m = m->next;
-			    if (deep) stencil = 0;
-			}
-		    }
-		}
-	    }
+	static const GLenum db_modes[2] = { GLX_NONE, GLX_SWAP_UNDEFINED_OML };
+	uint8_t depth_bits_array[4];
+	uint8_t stencil_bits_array[4];
+	if(deep) {
+		depth_bits_array[0] = 0;
+		depth_bits_array[1] = 24;
+		stencil_bits_array[0] = 0;
+		stencil_bits_array[1] = 8;
+	} else {
+		depth_bits_array[0] = depth_bits;
+		depth_bits_array[1] = 0;
+		depth_bits_array[2] = depth_bits;
+		depth_bits_array[3] = 0;
+		stencil_bits_array[0] = 0;
+		stencil_bits_array[1] = 0;
+		stencil_bits_array[2] = 8;
+		stencil_bits_array[3] = 8;
 	}
 
-	return modes;
+	return driCreateConfigs(
+		deep ? GL_RGBA : GL_RGB,
+		deep ? GL_UNSIGNED_INT_8_8_8_8 : GL_UNSIGNED_SHORT_5_6_5,
+		depth_bits_array,
+		stencil_bits_array,
+		deep ? 2 : 4,
+		db_modes, 2);
 }
 
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \todo maybe fold this into intelInitDriver
  *
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on
- *         failure.
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd,
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
+static const __DRIconfig **
+tdfxInitScreen(__DRIscreen *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 1, 1, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 0, 0 };
 
-   dri_interface = interface;
+   /* divined from tdfx_dri.c, sketchy */
+   TDFXDRIPtr dri_priv = (TDFXDRIPtr) psp->pDevPriv;
+
+   /* XXX i wish it was like this */
+   /* bpp = dri_priv->bpp */
+   int bpp = (dri_priv->cpp > 2) ? 24 : 16;
 
    if ( ! driCheckDriDdxDrmVersions2( "tdfx",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-   				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &tdfxAPI);
 
-   if (psp != NULL) {
-      /* divined from tdfx_dri.c, sketchy */
-      TDFXDRIPtr dri_priv = (TDFXDRIPtr) psp->pDevPriv;
-      int bpp = (dri_priv->cpp > 2) ? 24 : 16;
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+   driInitExtensions( NULL, napalm_extensions, GL_FALSE );
 
-      /* XXX i wish it was like this */
-      /* bpp = dri_priv->bpp */
+   if (!tdfxInitDriver(psp))
+      return NULL;
       
-      *driver_modes = tdfxFillInModes(bpp, (bpp == 16) ? 16 : 24,
-				(bpp == 16) ? 0 : 8,
-				(dri_priv->backOffset!=dri_priv->depthOffset));
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-      driInitExtensions( NULL, napalm_extensions, GL_FALSE );
-   }
-
-   return (void *)psp;
+   return tdfxFillInModes(psp,
+			  bpp, (bpp == 16) ? 16 : 24,
+			  (bpp == 16) ? 0 : 8,
+			  (dri_priv->backOffset!=dri_priv->depthOffset));
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = tdfxInitScreen,
+   .DestroyScreen   = tdfxDestroyScreen,
+   .CreateContext   = tdfxCreateContext,
+   .DestroyContext  = tdfxDestroyContext,
+   .CreateBuffer    = tdfxCreateBuffer,
+   .DestroyBuffer   = tdfxDestroyBuffer,
+   .SwapBuffers     = tdfxSwapBuffers,
+   .MakeCurrent     = tdfxMakeCurrent,
+   .UnbindContext   = tdfxUnbindContext,
+   .GetSwapInfo     = NULL,
+   .GetDrawableMSC  = NULL,
+   .WaitForMSC      = NULL,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL
+};
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_span.h b/src/mesa/drivers/dri/tdfx/tdfx_span.h
index 5af9f9b3014..6973f8d1407 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_span.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_span.h
@@ -37,7 +37,7 @@
 #ifndef __TDFX_SPAN_H__
 #define __TDFX_SPAN_H__
 
-#include "context.h"
+#include "main/context.h"
 #include "drirenderbuffer.h"
 
 extern void tdfxDDInitSpanFuncs( GLcontext *ctx );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_state.c b/src/mesa/drivers/dri/tdfx/tdfx_state.c
index 3688c76a5c9..a2d7bcd97d9 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_state.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_state.c
@@ -38,11 +38,11 @@
  *
  */
 
-#include "mtypes.h"
-#include "colormac.h"
-#include "texformat.h"
-#include "texstore.h"
-#include "teximage.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
 
 #include "swrast/swrast.h"
 #include "vbo/vbo.h"
@@ -1032,21 +1032,23 @@ static void tdfxDDDrawBuffer( GLcontext *ctx, GLenum mode )
 
    FLUSH_BATCH( fxMesa );
 
-   /*
-    * _ColorDrawBufferMask is easier to cope with than <mode>.
-    */
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers > 1) {
+      FALLBACK( fxMesa, TDFX_FALLBACK_DRAW_BUFFER, GL_TRUE );
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       fxMesa->DrawBuffer = fxMesa->ReadBuffer = GR_BUFFER_FRONTBUFFER;
       fxMesa->new_state |= TDFX_NEW_RENDER;
       FALLBACK( fxMesa, TDFX_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       fxMesa->DrawBuffer = fxMesa->ReadBuffer = GR_BUFFER_BACKBUFFER;
       fxMesa->new_state |= TDFX_NEW_RENDER;
       FALLBACK( fxMesa, TDFX_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
-   case 0:
+   case -1:
       FX_grColorMaskv( ctx, false4 );
       FALLBACK( fxMesa, TDFX_FALLBACK_DRAW_BUFFER, GL_FALSE );
       break;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_state.h b/src/mesa/drivers/dri/tdfx/tdfx_state.h
index 591ea5b083f..4880b990fcd 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_state.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_state.h
@@ -37,7 +37,7 @@
 #ifndef __TDFX_STATE_H__
 #define __TDFX_STATE_H__
 
-#include "context.h"
+#include "main/context.h"
 #include "tdfx_context.h"
 
 extern void tdfxDDInitStateFuncs( GLcontext *ctx );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
index 70ef7264379..1f7257eaead 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
@@ -38,14 +38,14 @@
  */
 
 
-#include "enums.h"
-#include "image.h"
-#include "mipmap.h"
-#include "texcompress.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texstore.h"
-#include "texobj.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/mipmap.h"
+#include "main/texcompress.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
+#include "main/texobj.h"
 #include "tdfx_context.h"
 #include "tdfx_tex.h"
 #include "tdfx_texman.h"
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_texman.c b/src/mesa/drivers/dri/tdfx/tdfx_texman.c
index f9b2726da29..35636ee5efb 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_texman.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_texman.c
@@ -37,8 +37,8 @@
 #include "tdfx_context.h"
 #include "tdfx_tex.h"
 #include "tdfx_texman.h"
-#include "texobj.h"
-#include "hash.h"
+#include "main/texobj.h"
+#include "main/hash.h"
 
 
 #define BAD_ADDRESS ((FxU32) -1)
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tris.c b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
index 59ff35a7fa9..88249888952 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tris.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tris.c
@@ -31,10 +31,10 @@
  *    Keith Whitwell <keith@tungstengraphics.com>
  */
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tris.h b/src/mesa/drivers/dri/tdfx/tdfx_tris.h
index a591decf1dc..ec48a486927 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tris.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tris.h
@@ -33,7 +33,7 @@
 #ifndef TDFX_TRIS_INC
 #define TDFX_TRIS_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void tdfxDDInitTriFuncs( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vb.c b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
index 62885daaa5b..49288022328 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vb.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
@@ -23,11 +23,11 @@
  *
  */
  
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "math/m_translate.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vb.h b/src/mesa/drivers/dri/tdfx/tdfx_vb.h
index 6389ec95b17..1e190e85f64 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vb.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vb.h
@@ -26,7 +26,7 @@
 #ifndef TDFXVB_INC
 #define TDFXVB_INC
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
diff --git a/src/mesa/drivers/dri/trident/trident_context.c b/src/mesa/drivers/dri/trident/trident_context.c
index 8dc7f0dc781..e134cfcf8e9 100644
--- a/src/mesa/drivers/dri/trident/trident_context.c
+++ b/src/mesa/drivers/dri/trident/trident_context.c
@@ -35,17 +35,17 @@
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 
-#include "context.h"
-#include "simple_list.h"
-#include "matrix.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 #if defined(USE_X86_ASM)
 #include "x86/common_x86_asm.h"
 #endif
-#include "simple_list.h"
-#include "mm.h"
+#include "main/simple_list.h"
+#include "main/mm.h"
 #include "drirenderbuffer.h"
 
 #include "drivers/common/driverfuncs.h"
@@ -417,56 +417,46 @@ tridentInitDriver(__DRIscreenPrivate *sPriv)
     return GL_TRUE;
 }
 
-static struct __DriverAPIRec tridentAPI = {
-   tridentInitDriver,
-   tridentDestroyScreen,
-   tridentCreateContext,
-   tridentDestroyContext,
-   tridentCreateBuffer,
-   tridentDestroyBuffer,
-   tridentSwapBuffers,
-   tridentMakeCurrent,
-   tridentUnbindContext,
-};
-
-
-PUBLIC void *__driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn,
-                                   __DRIscreen *psc,
-                                   const __GLcontextModes * modes,
-                                   const __DRIversion * ddx_version,
-                                   const __DRIversion * dri_version,
-                                   const __DRIversion * drm_version,
-                                   const __DRIframebuffer * frame_buffer,
-                                   drmAddress pSAREA, int fd,
-                                   int internal_api_version,
-				   const __DRIinterfaceMethods * interface,
-                                   __GLcontextModes ** driver_modes )
+/**
+ * This is the driver specific part of the createNewScreen entry point.
+ * 
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
+ */
+const __DRIconfig **tridentInitScreen(__DRIscreenPrivate *psp)
 {
-    __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 4, 0, 0 };
    static const __DRIversion dri_expected = { 3, 1, 0 };
    static const __DRIversion drm_expected = { 1, 0, 0 };
-
-   dri_interface = interface;
-
+   
    if ( ! driCheckDriDdxDrmVersions2( "Trident",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected ) )
       return NULL;
-   }
 
-    psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-                                   ddx_version, dri_version, drm_version,
-                                   frame_buffer, pSAREA, fd,
-                                   internal_api_version, &tridentAPI);
+   if (!tridentInitDriver(psp))
+	return NULL;
 
-    if ( psp != NULL ) {
+    /* Wait... what?  This driver doesn't report any modes... */
 #if 0
-       TRIDENTDRIPtr dri_priv = (TRIDENTDRIPtr) psp->pDevPriv;
-       *driver_modes = tridentFillInModes( dri_priv->bytesPerPixel * 8,
-					   GL_TRUE );
+   TRIDENTDRIPtr dri_priv = (TRIDENTDRIPtr) psp->pDevPriv;
+   *driver_modes = tridentFillInModes( dri_priv->bytesPerPixel * 8,
+				       GL_TRUE );
 #endif
-    }
-    return (void *) psp;
+
+   return NULL;
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   tridentInitScreen,
+   tridentDestroyScreen,
+   tridentCreateContext,
+   tridentDestroyContext,
+   tridentCreateBuffer,
+   tridentDestroyBuffer,
+   tridentSwapBuffers,
+   tridentMakeCurrent,
+   tridentUnbindContext,
+};
diff --git a/src/mesa/drivers/dri/trident/trident_context.h b/src/mesa/drivers/dri/trident/trident_context.h
index 1d3ca84400d..fbbb4a96e78 100644
--- a/src/mesa/drivers/dri/trident/trident_context.h
+++ b/src/mesa/drivers/dri/trident/trident_context.h
@@ -28,10 +28,10 @@
 #define _TRIDENT_CONTEXT_H_
 
 #include "dri_util.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 #include "drm.h"
-#include "mm.h"
+#include "main/mm.h"
 
 #define SUBPIXEL_X (0.0F)
 #define SUBPIXEL_Y (0.125F)
@@ -54,14 +54,14 @@
 #undef TAG
 
 /* these require that base be dword-aligned */
-static inline void MMIO_OUT32(unsigned char *base, unsigned int offset,
+static INLINE void MMIO_OUT32(unsigned char *base, unsigned int offset,
                               unsigned int val)
 {
     unsigned int *addr = (unsigned int *)(base + offset);
     *addr = val;
 }
 
-static inline unsigned int MMIO_IN32(unsigned char *base, unsigned int offset)
+static INLINE unsigned int MMIO_IN32(unsigned char *base, unsigned int offset)
 {
     unsigned int *addr = (unsigned int *)(base + offset);
     return *addr;
diff --git a/src/mesa/drivers/dri/trident/trident_dd.c b/src/mesa/drivers/dri/trident/trident_dd.c
index 4639b3a15e7..faa40c36a26 100644
--- a/src/mesa/drivers/dri/trident/trident_dd.c
+++ b/src/mesa/drivers/dri/trident/trident_dd.c
@@ -31,8 +31,8 @@
 #endif
 
 #include "swrast/swrast.h"
-#include "context.h"
-#include "framebuffer.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
 
 #define TRIDENT_DATE	"20041223"
 
diff --git a/src/mesa/drivers/dri/trident/trident_state.c b/src/mesa/drivers/dri/trident/trident_state.c
index 5303bd422ed..e68d3a73c64 100644
--- a/src/mesa/drivers/dri/trident/trident_state.c
+++ b/src/mesa/drivers/dri/trident/trident_state.c
@@ -30,7 +30,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "framebuffer.h"
+#include "main/framebuffer.h"
 
 #define TRIDENTPACKCOLOR332(r, g, b)					\
    (((r) & 0xe0) | (((g) & 0xe0) >> 3) | (((b) & 0xc0) >> 6))
@@ -51,9 +51,9 @@
 #define TRIDENTPACKCOLOR4444(r, g, b, a)					\
    ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
 
-static __inline__ GLuint tridentPackColor( GLuint cpp,
-					  GLubyte r, GLubyte g,
-					  GLubyte b, GLubyte a )
+static INLINE GLuint tridentPackColor( GLuint cpp,
+                                       GLubyte r, GLubyte g,
+                                       GLubyte b, GLubyte a )
 {
    switch ( cpp ) {
    case 2:
diff --git a/src/mesa/drivers/dri/trident/trident_vb.c b/src/mesa/drivers/dri/trident/trident_vb.c
index 77e4d9b76e0..b231f5ef15e 100644
--- a/src/mesa/drivers/dri/trident/trident_vb.c
+++ b/src/mesa/drivers/dri/trident/trident_vb.c
@@ -24,10 +24,10 @@
  * Trident CyberBladeXP driver.
  *
  */
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/swrast.h"
diff --git a/src/mesa/drivers/dri/unichrome/via_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index 4d25d328e33..f5bdb65eb08 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -30,14 +30,14 @@
  * \author Others at S3 Graphics?
  */
 
-#include "glheader.h"
-#include "context.h"
-#include "matrix.h"
-#include "state.h"
-#include "simple_list.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/matrix.h"
+#include "main/state.h"
+#include "main/simple_list.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -59,7 +59,7 @@
 #include "via_fb.h"
 
 #include <stdio.h>
-#include "macros.h"
+#include "main/macros.h"
 #include "drirenderbuffer.h"
 
 #define need_GL_ARB_multisample
@@ -123,7 +123,7 @@ static const GLubyte *viaGetString(GLcontext *ctx, GLenum name)
  * 
  * \returns A pixel width that meets the alignment requirements.
  */
-static __inline__ unsigned
+static INLINE unsigned
 buffer_align( unsigned width )
 {
     return (width + 0x0f) & ~0x0f;
@@ -601,9 +601,6 @@ viaCreateContext(const __GLcontextModes *visual,
     _tnl_allow_pixel_fog(ctx, GL_FALSE);
     _tnl_allow_vertex_fog(ctx, GL_TRUE);
 
-/*     vmesa->display = dpy; */
-    vmesa->display = sPriv->display;
-    
     vmesa->hHWContext = driContextPriv->hHWContext;
     vmesa->driFd = sPriv->fd;
     vmesa->driHwLock = &sPriv->pSAREA->lock;
@@ -661,14 +658,10 @@ viaCreateContext(const __GLcontextModes *visual,
         driQueryOptionb(&vmesa->optionCache, "no_rast"))
        FALLBACK(vmesa, VIA_FALLBACK_USER_DISABLE, 1);
 
-    vmesa->vblank_flags =
-       vmesa->viaScreen->irqEnabled ?
-        driGetDefaultVBlankFlags(&vmesa->optionCache) : VBLANK_FLAG_NO_IRQ;
-
     if (getenv("VIA_PAGEFLIP"))
        vmesa->allowPageFlip = 1;
 
-    (*dri_interface->getUST)( &vmesa->swap_ust );
+    (*sPriv->systemTime->getUST)( &vmesa->swap_ust );
 
 
     vmesa->regMMIOBase = (GLuint *)((unsigned long)viaScreen->reg);
@@ -686,46 +679,48 @@ void
 viaDestroyContext(__DRIcontextPrivate *driContextPriv)
 {
     GET_CURRENT_CONTEXT(ctx);
-    struct via_context *vmesa = 
+    struct via_context *vmesa =
        (struct via_context *)driContextPriv->driverPrivate;
     struct via_context *current = ctx ? VIA_CONTEXT(ctx) : NULL;
+
     assert(vmesa); /* should never be null */
 
+    if (vmesa->driDrawable) {
+       viaWaitIdle(vmesa, GL_FALSE);
+
+       if (vmesa->doPageFlip) {
+	  LOCK_HARDWARE(vmesa);
+	  if (vmesa->pfCurrentOffset != 0) {
+	     fprintf(stderr, "%s - reset pf\n", __FUNCTION__);
+	     viaResetPageFlippingLocked(vmesa);
+	  }
+	  UNLOCK_HARDWARE(vmesa);
+       }
+    }
+
     /* check if we're deleting the currently bound context */
     if (vmesa == current) {
       VIA_FLUSH_DMA(vmesa);
       _mesa_make_current(NULL, NULL, NULL);
     }
 
-    if (vmesa) {
-        viaWaitIdle(vmesa, GL_FALSE);
-	if (vmesa->doPageFlip) {
-	   LOCK_HARDWARE(vmesa);
-	   if (vmesa->pfCurrentOffset != 0) {
-	      fprintf(stderr, "%s - reset pf\n", __FUNCTION__);
-	      viaResetPageFlippingLocked(vmesa);
-	   }
-	   UNLOCK_HARDWARE(vmesa);
-	}
-	
-	_swsetup_DestroyContext(vmesa->glCtx);
-        _tnl_DestroyContext(vmesa->glCtx);
-        _vbo_DestroyContext(vmesa->glCtx);
-        _swrast_DestroyContext(vmesa->glCtx);
-        /* free the Mesa context */
-	_mesa_destroy_context(vmesa->glCtx);
-	/* release our data */
-	FreeBuffer(vmesa);
+    _swsetup_DestroyContext(vmesa->glCtx);
+    _tnl_DestroyContext(vmesa->glCtx);
+    _vbo_DestroyContext(vmesa->glCtx);
+    _swrast_DestroyContext(vmesa->glCtx);
+    /* free the Mesa context */
+    _mesa_destroy_context(vmesa->glCtx);
+    /* release our data */
+    FreeBuffer(vmesa);
 
-	assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_AGP]));
-	assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_VIDEO]));
-	assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_SYSTEM]));
-	assert (is_empty_list(&vmesa->freed_tex_buffers));
+    assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_AGP]));
+    assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_VIDEO]));
+    assert (is_empty_list(&vmesa->tex_image_list[VIA_MEM_SYSTEM]));
+    assert (is_empty_list(&vmesa->freed_tex_buffers));
 
-	driDestroyOptionCache(&vmesa->optionCache);
+    driDestroyOptionCache(&vmesa->optionCache);
 
-	FREE(vmesa);
-    }
+    FREE(vmesa);
 }
 
 
@@ -733,17 +728,18 @@ void viaXMesaWindowMoved(struct via_context *vmesa)
 {
    __DRIdrawablePrivate *const drawable = vmesa->driDrawable;
    __DRIdrawablePrivate *const readable = vmesa->driReadable;
-   struct via_renderbuffer *const draw_buffer = 
-     (struct via_renderbuffer *) drawable->driverPrivate;
-   struct via_renderbuffer *const read_buffer =
-     (struct via_renderbuffer *) readable->driverPrivate;
+   struct via_renderbuffer * draw_buffer;
+   struct via_renderbuffer * read_buffer;
    GLuint bytePerPixel = vmesa->viaScreen->bitsPerPixel >> 3;
 
    if (!drawable)
       return;
+
+   draw_buffer =  (struct via_renderbuffer *) drawable->driverPrivate;
+   read_buffer =  (struct via_renderbuffer *) readable->driverPrivate;
    
-   switch (vmesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0]) {
-   case BUFFER_BIT_BACK_LEFT: 
+   switch (vmesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0]) {
+   case BUFFER_BACK_LEFT: 
       if (drawable->numBackClipRects == 0) {
 	 vmesa->numClipRects = drawable->numClipRects;
 	 vmesa->pClipRects = drawable->pClipRects;
@@ -753,7 +749,7 @@ void viaXMesaWindowMoved(struct via_context *vmesa)
 	 vmesa->pClipRects = drawable->pBackClipRects;
       }
       break;
-   case BUFFER_BIT_FRONT_LEFT:
+   case BUFFER_FRONT_LEFT:
       vmesa->numClipRects = drawable->numClipRects;
       vmesa->pClipRects = drawable->pClipRects;
       break;
@@ -839,13 +835,17 @@ viaMakeCurrent(__DRIcontextPrivate *driContextPriv,
         drawBuffer = (GLframebuffer *)driDrawPriv->driverPrivate;
         readBuffer = (GLframebuffer *)driReadPriv->driverPrivate;
 
-	if (vmesa->driDrawable != driDrawPriv) {
-	   driDrawableInitVBlank(driDrawPriv, vmesa->vblank_flags,
-				 &vmesa->vbl_seq);
-	}
-
        if ((vmesa->driDrawable != driDrawPriv)
 	   || (vmesa->driReadable != driReadPriv)) {
+	  if (driDrawPriv->swap_interval == (unsigned)-1) {
+	     driDrawPriv->vblFlags =
+		vmesa->viaScreen->irqEnabled ?
+		driGetDefaultVBlankFlags(&vmesa->optionCache) :
+		VBLANK_FLAG_NO_IRQ;
+
+	     driDrawableInitVBlank(driDrawPriv);
+	  }
+
 	  vmesa->driDrawable = driDrawPriv;
 	  vmesa->driReadable = driReadPriv;
 
diff --git a/src/mesa/drivers/dri/unichrome/via_context.h b/src/mesa/drivers/dri/unichrome/via_context.h
index fecd2782fba..4cc9e475c21 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.h
+++ b/src/mesa/drivers/dri/unichrome/via_context.h
@@ -28,9 +28,10 @@
 
 #include "dri_util.h"
 
-#include "mtypes.h"
 #include "drm.h"
-#include "mm.h"
+
+#include "main/mtypes.h"
+#include "main/mm.h"
 #include "tnl/t_vertex.h"
 
 #include "via_screen.h"
@@ -289,7 +290,6 @@ struct via_context {
    drm_context_t hHWContext;
    drm_hw_lock_t *driHwLock;
    int driFd;
-   __DRInativeDisplay *display;
 
    /**
     * DRI drawable bound to this context for drawing.
@@ -322,9 +322,6 @@ struct via_context {
     */
    driOptionCache optionCache;
 
-   GLuint vblank_flags;
-   GLuint vbl_seq;
-
    int64_t swap_ust;
    int64_t swap_missed_ust;
 
diff --git a/src/mesa/drivers/dri/unichrome/via_fb.c b/src/mesa/drivers/dri/unichrome/via_fb.c
index 48c7ed4caa5..e4fb29f6c60 100644
--- a/src/mesa/drivers/dri/unichrome/via_fb.c
+++ b/src/mesa/drivers/dri/unichrome/via_fb.c
@@ -28,8 +28,8 @@
 #include "via_ioctl.h"
 #include "via_fb.h"
 #include "xf86drm.h"
-#include "imports.h"
-#include "simple_list.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
 #include <sys/ioctl.h>
 
 GLboolean
diff --git a/src/mesa/drivers/dri/unichrome/via_ioctl.c b/src/mesa/drivers/dri/unichrome/via_ioctl.c
index 4a733fb00c2..6746f552ae6 100644
--- a/src/mesa/drivers/dri/unichrome/via_ioctl.c
+++ b/src/mesa/drivers/dri/unichrome/via_ioctl.c
@@ -24,13 +24,13 @@
 #include <stdio.h>
 #include <unistd.h>
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "dd.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/dd.h"
 #include "swrast/swrast.h"
 
-#include "mm.h"
+#include "main/mm.h"
 #include "via_context.h"
 #include "via_tris.h"
 #include "via_ioctl.h"
@@ -507,11 +507,12 @@ void viaWaitIdleLocked( struct via_context *vmesa, GLboolean light )
  * except that WAIT_IDLE() will spin the CPU polling, while this is
  * IRQ driven.
  */
-static void viaWaitIdleVBlank( const __DRIdrawablePrivate *dPriv, 
+static void viaWaitIdleVBlank(  __DRIdrawablePrivate *dPriv, 
 			       struct via_context *vmesa,
 			       GLuint value )
 {
    GLboolean missed_target;
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
    VIA_FLUSH_DMA(vmesa); 
 
@@ -523,11 +524,10 @@ static void viaWaitIdleVBlank( const __DRIdrawablePrivate *dPriv,
 	  vmesa->thrashing)
 	 viaSwapOutWork(vmesa);
 
-      driWaitForVBlank( dPriv, & vmesa->vbl_seq, 
-			vmesa->vblank_flags, & missed_target );
+      driWaitForVBlank( dPriv, & missed_target );
       if ( missed_target ) {
 	 vmesa->swap_missed_count++;
-	 (*dri_interface->getUST)( &vmesa->swap_missed_ust );
+	 (*psp->systemTime->getUST)( &vmesa->swap_missed_ust );
       }
    } 
    while (!viaCheckBreadcrumb(vmesa, value));	 
@@ -591,10 +591,11 @@ void viaResetPageFlippingLocked(struct via_context *vmesa)
 /*
  * Copy the back buffer to the front buffer. 
  */
-void viaCopyBuffer(const __DRIdrawablePrivate *dPriv)
+void viaCopyBuffer(__DRIdrawablePrivate *dPriv)
 {
    struct via_context *vmesa = 
       (struct via_context *)dPriv->driContextPriv->driverPrivate;
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
    if (VIA_DEBUG & DEBUG_IOCTL)
       fprintf(stderr, 
@@ -607,7 +608,7 @@ void viaCopyBuffer(const __DRIdrawablePrivate *dPriv)
 
    VIA_FLUSH_DMA(vmesa);
 
-   if (vmesa->vblank_flags == VBLANK_FLAG_SYNC &&
+   if (dPriv->vblFlags == VBLANK_FLAG_SYNC &&
        vmesa->lastBreadcrumbWrite > 1)
       viaWaitIdleVBlank(dPriv, vmesa, vmesa->lastBreadcrumbWrite-1);
    else
@@ -630,18 +631,19 @@ void viaCopyBuffer(const __DRIdrawablePrivate *dPriv)
    viaEmitBreadcrumbLocked(vmesa);
    UNLOCK_HARDWARE(vmesa);
 
-   (*dri_interface->getUST)( &vmesa->swap_ust );
+   (*psp->systemTime->getUST)( &vmesa->swap_ust );
 }
 
 
-void viaPageFlip(const __DRIdrawablePrivate *dPriv)
+void viaPageFlip(__DRIdrawablePrivate *dPriv)
 {
     struct via_context *vmesa = 
        (struct via_context *)dPriv->driContextPriv->driverPrivate;
     struct via_renderbuffer buffer_tmp;
+    __DRIscreenPrivate *psp = dPriv->driScreenPriv;
 
     VIA_FLUSH_DMA(vmesa);
-   if (vmesa->vblank_flags == VBLANK_FLAG_SYNC &&
+   if (dPriv->vblFlags == VBLANK_FLAG_SYNC &&
        vmesa->lastBreadcrumbWrite > 1)
       viaWaitIdleVBlank(dPriv, vmesa, vmesa->lastBreadcrumbWrite - 1);
    else
@@ -654,7 +656,7 @@ void viaPageFlip(const __DRIdrawablePrivate *dPriv)
     viaEmitBreadcrumbLocked(vmesa);
     UNLOCK_HARDWARE(vmesa);
 
-    (*dri_interface->getUST)( &vmesa->swap_ust );
+    (*psp->systemTime->getUST)( &vmesa->swap_ust );
 
 
     /* KW: FIXME: When buffers are freed, could free frontbuffer by
diff --git a/src/mesa/drivers/dri/unichrome/via_ioctl.h b/src/mesa/drivers/dri/unichrome/via_ioctl.h
index a81b427d807..14a833a97d0 100644
--- a/src/mesa/drivers/dri/unichrome/via_ioctl.h
+++ b/src/mesa/drivers/dri/unichrome/via_ioctl.h
@@ -33,8 +33,8 @@ void viaFlushDma(struct via_context *vmesa);
 void viaFlushDmaLocked(struct via_context *vmesa, GLuint flags);
 
 void viaInitIoctlFuncs(GLcontext *ctx);
-void viaCopyBuffer(const __DRIdrawablePrivate *dpriv);
-void viaPageFlip(const __DRIdrawablePrivate *dpriv);
+void viaCopyBuffer(__DRIdrawablePrivate *dpriv);
+void viaPageFlip(__DRIdrawablePrivate *dpriv);
 void viaCheckDma(struct via_context *vmesa, GLuint bytes);
 void viaResetPageFlippingLocked(struct via_context *vmesa);
 void viaWaitIdle(struct via_context *vmesa, GLboolean light);
@@ -58,7 +58,7 @@ void viaEmitBreadcrumb( struct via_context *vmesa );
 
 void viaWrapPrimitive( struct via_context *vmesa );
 
-static __inline__ GLuint *viaAllocDma(struct via_context *vmesa, int bytes)
+static INLINE GLuint *viaAllocDma(struct via_context *vmesa, int bytes)
 {
    if (vmesa->dmaLow + bytes > VIA_DMA_HIGHWATER) {
       viaFlushDma(vmesa);
@@ -72,7 +72,7 @@ static __inline__ GLuint *viaAllocDma(struct via_context *vmesa, int bytes)
 }
 
 
-static GLuint __inline__ *viaExtendPrimitive(struct via_context *vmesa, int bytes)
+static GLuint INLINE *viaExtendPrimitive(struct via_context *vmesa, int bytes)
 {
    if (vmesa->dmaLow + bytes > VIA_DMA_HIGHWATER) {
       viaWrapPrimitive(vmesa);
diff --git a/src/mesa/drivers/dri/unichrome/via_render.c b/src/mesa/drivers/dri/unichrome/via_render.c
index 387473ef522..f676cc13c81 100644
--- a/src/mesa/drivers/dri/unichrome/via_render.c
+++ b/src/mesa/drivers/dri/unichrome/via_render.c
@@ -28,10 +28,10 @@
  * dma buffers.  Use strip/fan hardware acceleration where possible.
  *
  */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 
 #include "tnl/t_context.h"
 
@@ -106,7 +106,7 @@ static GLboolean via_run_fastrender(GLcontext *ctx,
     tnl->clipspace.new_inputs |= VERT_BIT_POS;
 
     for (i = 0; i < VB->PrimitiveCount; ++i) {
-        GLuint mode = VB->Primitive[i].mode;
+        GLuint mode = _tnl_translate_prim(&VB->Primitive[i]);
         GLuint start = VB->Primitive[i].start;
         GLuint length = VB->Primitive[i].count;
         if (length)
diff --git a/src/mesa/drivers/dri/unichrome/via_screen.c b/src/mesa/drivers/dri/unichrome/via_screen.c
index b81549df0e9..988f9935ac1 100644
--- a/src/mesa/drivers/dri/unichrome/via_screen.c
+++ b/src/mesa/drivers/dri/unichrome/via_screen.c
@@ -24,14 +24,14 @@
 
 #include <stdio.h>
 
-#include "utils.h"
 #include "dri_util.h"
-#include "glheader.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "matrix.h"
-#include "simple_list.h"
+#include "utils.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
+#include "main/matrix.h"
+#include "main/simple_list.h"
 #include "vblank.h"
 
 #include "via_state.h"
@@ -48,7 +48,7 @@
 
 #include "xmlpool.h"
 
-const char __driConfigOptions[] =
+PUBLIC const char __driConfigOptions[] =
 DRI_CONF_BEGIN
     DRI_CONF_SECTION_PERFORMANCE
         DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
@@ -64,8 +64,6 @@ static const GLuint __driNConfigOptions = 3;
 
 extern const struct dri_extension card_extensions[];
 
-static int getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo );
-
 static drmBufMapPtr via_create_empty_buffers(void)
 {
     drmBufMapPtr retval;
@@ -98,9 +96,7 @@ viaInitDriver(__DRIscreenPrivate *sPriv)
 {
     viaScreenPrivate *viaScreen;
     VIADRIPtr gDRIPriv = (VIADRIPtr)sPriv->pDevPriv;
-    PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-      (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-    void * const psc = sPriv->psc->screenConfigs;
+    int i;
 
     if (sPriv->devPrivSize != sizeof(VIADRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(VIADRIRec) does not match passed size from device driver\n");
@@ -175,17 +171,17 @@ viaInitDriver(__DRIscreenPrivate *sPriv)
 
     viaScreen->sareaPrivOffset = gDRIPriv->sarea_priv_offset;
 
-    if ( glx_enable_extension != NULL ) {
-       if ( viaScreen->irqEnabled ) {
-	  (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-	  (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-	  (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-       }
-
-       (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
-       (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
+    i = 0;
+    viaScreen->extensions[i++] = &driFrameTrackingExtension.base;
+    viaScreen->extensions[i++] = &driReadDrawableExtension;
+    if ( viaScreen->irqEnabled ) {
+	viaScreen->extensions[i++] = &driSwapControlExtension.base;
+	viaScreen->extensions[i++] = &driMediaStreamCounterExtension.base;
     }
 
+    viaScreen->extensions[i++] = NULL;
+    sPriv->extensions = viaScreen->extensions;
+
     return GL_TRUE;
 }
 
@@ -323,32 +319,11 @@ viaDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
    _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
 }
 
-
-
-static struct __DriverAPIRec viaAPI = {
-   .InitDriver      = viaInitDriver,
-   .DestroyScreen   = viaDestroyScreen,
-   .CreateContext   = viaCreateContext,
-   .DestroyContext  = viaDestroyContext,
-   .CreateBuffer    = viaCreateBuffer,
-   .DestroyBuffer   = viaDestroyBuffer,
-   .SwapBuffers     = viaSwapBuffers,
-   .MakeCurrent     = viaMakeCurrent,
-   .UnbindContext   = viaUnbindContext,
-   .GetSwapInfo     = getSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL
-};
-
-
-static __GLcontextModes *
-viaFillInModes( unsigned pixel_bits, GLboolean have_back_buffer )
+static const __DRIconfig **
+viaFillInModes( __DRIscreenPrivate *psp,
+		unsigned pixel_bits, GLboolean have_back_buffer )
 {
-    __GLcontextModes * modes;
-    __GLcontextModes * m;
-    unsigned num_modes;
+    __DRIconfig **configs;
     const unsigned back_buffer_factor = (have_back_buffer) ? 2 : 1;
     GLenum fb_format;
     GLenum fb_type;
@@ -369,9 +344,6 @@ viaFillInModes( unsigned pixel_bits, GLboolean have_back_buffer )
     static const uint8_t stencil_bits_array[4] = { 0,  0,  8,  0 };
     const unsigned depth_buffer_factor = 3;
 
-
-    num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
     if ( pixel_bits == 16 ) {
         fb_format = GL_RGB;
         fb_type = GL_UNSIGNED_SHORT_5_6_5;
@@ -381,94 +353,61 @@ viaFillInModes( unsigned pixel_bits, GLboolean have_back_buffer )
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-    modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
-    m = modes;
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, 
-			   depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
+    configs = driCreateConfigs(fb_format, fb_type,
+			       depth_bits_array, stencil_bits_array,
+			       depth_buffer_factor, back_buffer_modes,
+			       back_buffer_factor);
+    if (configs == NULL) {
+	fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+		__LINE__);
 	return NULL;
     }
 
-    if ( ! driFillInModes( & m, fb_format, fb_type,
-			   depth_bits_array, stencil_bits_array, 
-			   depth_buffer_factor,
-			   back_buffer_modes, back_buffer_factor,
-			   GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-    }
-
-    return modes;
+    return (const __DRIconfig **) configs;
 }
 
 
 /**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
+ * This is the driver specific part of the createNewScreen entry point.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
+ * \todo maybe fold this into intelInitDriver
+ *
+ * \return the __GLcontextModes supported by this driver
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn,
-			     __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+static const __DRIconfig **
+viaInitScreen(__DRIscreenPrivate *psp)
 {
-   __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { VIA_DRIDDX_VERSION_MAJOR,
                                               VIA_DRIDDX_VERSION_MINOR,
                                               VIA_DRIDDX_VERSION_PATCH };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 2, 3, 0 };
    static const char *driver_name = "Unichrome";
-
-   dri_interface = interface;
+   VIADRIPtr dri_priv = (VIADRIPtr) psp->pDevPriv;
 
    if ( ! driCheckDriDdxDrmVersions2( driver_name,
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected) ) {
+				      &psp->dri_version, & dri_expected,
+				      &psp->ddx_version, & ddx_expected,
+				      &psp->drm_version, & drm_expected) )
       return NULL;
-   }
-      
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &viaAPI);
-   if ( psp != NULL ) {
-      VIADRIPtr dri_priv = (VIADRIPtr) psp->pDevPriv;
-      *driver_modes = viaFillInModes( dri_priv->bytesPerPixel * 8,
-				      GL_TRUE );
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
-   }
 
-   return (void *) psp;
+   /* Calling driInitExtensions here, with a NULL context pointer,
+    * does not actually enable the extensions.  It just makes sure
+    * that all the dispatch offsets for all the extensions that
+    * *might* be enables are known.  This is needed because the
+    * dispatch offsets need to be known when _mesa_context_create is
+    * called, but we can't enable the extensions until we have a
+    * context pointer.
+    *
+    * Hello chicken.  Hello egg.  How are you two today?
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+
+   if (!viaInitDriver(psp))
+       return NULL;
+
+   return viaFillInModes( psp, dri_priv->bytesPerPixel * 8, GL_TRUE );
+
 }
 
 
@@ -497,3 +436,20 @@ getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
 
    return 0;
 }
+
+const struct __DriverAPIRec driDriverAPI = {
+   .InitScreen      = viaInitScreen,
+   .DestroyScreen   = viaDestroyScreen,
+   .CreateContext   = viaCreateContext,
+   .DestroyContext  = viaDestroyContext,
+   .CreateBuffer    = viaCreateBuffer,
+   .DestroyBuffer   = viaDestroyBuffer,
+   .SwapBuffers     = viaSwapBuffers,
+   .MakeCurrent     = viaMakeCurrent,
+   .UnbindContext   = viaUnbindContext,
+   .GetSwapInfo     = getSwapInfo,
+   .GetDrawableMSC  = driDrawableGetMSC32,
+   .WaitForMSC      = driWaitForMSC32,
+   .WaitForSBC      = NULL,
+   .SwapBuffersMSC  = NULL
+};
diff --git a/src/mesa/drivers/dri/unichrome/via_screen.h b/src/mesa/drivers/dri/unichrome/via_screen.h
index 84aa5aef88e..c3ef722ff07 100644
--- a/src/mesa/drivers/dri/unichrome/via_screen.h
+++ b/src/mesa/drivers/dri/unichrome/via_screen.h
@@ -70,6 +70,8 @@ typedef struct {
 
     /* Configuration cache with default values for all contexts */
     driOptionCache optionCache;
+
+    const __DRIextension *extensions[5];
 } viaScreenPrivate;
 
 
diff --git a/src/mesa/drivers/dri/unichrome/via_span.c b/src/mesa/drivers/dri/unichrome/via_span.c
index 3a16dadd238..b908f0fb23a 100644
--- a/src/mesa/drivers/dri/unichrome/via_span.c
+++ b/src/mesa/drivers/dri/unichrome/via_span.c
@@ -22,10 +22,10 @@
  * DEALINGS IN THE SOFTWARE.
  */
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/colormac.h"
 #include "via_context.h"
 #include "via_span.h"
 #include "via_ioctl.h"
@@ -86,6 +86,7 @@
 
 #define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS 
 
+#define VALUE_TYPE GLushort
 
 #define WRITE_DEPTH(_x, _y, d)                      \
     *(GLushort *)(buf + (_x) * 2 + (_y) * depth_pitch) = d;
@@ -98,6 +99,8 @@
 
 /* 32 bit depthbuffer functions.
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH(_x, _y, d)                      \
     *(GLuint *)(buf + (_x) * 4 + (_y) * depth_pitch) = d;
 
@@ -111,6 +114,8 @@
 
 /* 24/8 bit interleaved depth/stencil functions
  */
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d ) {			\
    GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*depth_pitch);	\
    tmp &= 0x000000ff;					\
diff --git a/src/mesa/drivers/dri/unichrome/via_state.c b/src/mesa/drivers/dri/unichrome/via_state.c
index 5b0453423ef..1cef01ab033 100644
--- a/src/mesa/drivers/dri/unichrome/via_state.c
+++ b/src/mesa/drivers/dri/unichrome/via_state.c
@@ -24,14 +24,14 @@
 
 #include <stdio.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "colormac.h"
-#include "enums.h"
-#include "dd.h"
-
-#include "mm.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/colormac.h"
+#include "main/enums.h"
+#include "main/dd.h"
+#include "main/mm.h"
+
 #include "via_context.h"
 #include "via_state.h"
 #include "via_tex.h"
@@ -511,9 +511,9 @@ void viaEmitState(struct via_context *vmesa)
 }
 
 
-static __inline__ GLuint viaPackColor(GLuint bpp,
-                                      GLubyte r, GLubyte g,
-                                      GLubyte b, GLubyte a)
+static INLINE GLuint viaPackColor(GLuint bpp,
+                                  GLubyte r, GLubyte g,
+                                  GLubyte b, GLubyte a)
 {
     switch (bpp) {
     case 16:
@@ -656,13 +656,18 @@ static void viaDrawBuffer(GLcontext *ctx, GLenum mode)
    if (!ctx->DrawBuffer)
       return;
 
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
+      FALLBACK(vmesa, VIA_FALLBACK_DRAW_BUFFER, GL_TRUE);
+      return;
+   }
+
+   switch ( ctx->DrawBuffer->_ColorDrawBufferIndexes[0] ) {
+   case BUFFER_FRONT_LEFT:
       VIA_FLUSH_DMA(vmesa);
       vmesa->drawBuffer = &vmesa->front;
       FALLBACK(vmesa, VIA_FALLBACK_DRAW_BUFFER, GL_FALSE);
       break;
-   case BUFFER_BIT_BACK_LEFT:
+   case BUFFER_BACK_LEFT:
       VIA_FLUSH_DMA(vmesa);
       vmesa->drawBuffer = &vmesa->back;
       FALLBACK(vmesa, VIA_FALLBACK_DRAW_BUFFER, GL_FALSE);
diff --git a/src/mesa/drivers/dri/unichrome/via_tex.c b/src/mesa/drivers/dri/unichrome/via_tex.c
index 0261a3ff177..d2010f09074 100644
--- a/src/mesa/drivers/dri/unichrome/via_tex.c
+++ b/src/mesa/drivers/dri/unichrome/via_tex.c
@@ -26,21 +26,21 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "enums.h"
-#include "colortab.h"
-#include "convolve.h"
-#include "context.h"
-#include "mipmap.h"
-#include "simple_list.h"
-#include "texcompress.h"
-#include "texformat.h"
-#include "texobj.h"
-#include "texstore.h"
-
-#include "mm.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/enums.h"
+#include "main/colortab.h"
+#include "main/convolve.h"
+#include "main/context.h"
+#include "main/mipmap.h"
+#include "main/simple_list.h"
+#include "main/texcompress.h"
+#include "main/texformat.h"
+#include "main/texobj.h"
+#include "main/texstore.h"
+
+#include "main/mm.h"
 #include "via_context.h"
 #include "via_fb.h"
 #include "via_tex.h"
@@ -820,9 +820,7 @@ static void viaTexImage(GLcontext *ctx,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target,
-                            &ctx->Texture.Unit[ctx->Texture.CurrentUnit],
-                            texObj);
+      _mesa_generate_mipmap(ctx, target, texObj);
    }
 
    _mesa_unmap_teximage_pbo(ctx, packing);
diff --git a/src/mesa/drivers/dri/unichrome/via_tex.h b/src/mesa/drivers/dri/unichrome/via_tex.h
index 73cfa91addc..25eeee32f3d 100644
--- a/src/mesa/drivers/dri/unichrome/via_tex.h
+++ b/src/mesa/drivers/dri/unichrome/via_tex.h
@@ -26,7 +26,7 @@
 #ifndef _VIATEX_H
 #define _VIATEX_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 struct via_context;
 
diff --git a/src/mesa/drivers/dri/unichrome/via_texcombine.c b/src/mesa/drivers/dri/unichrome/via_texcombine.c
index d604457bfd8..b6468978486 100644
--- a/src/mesa/drivers/dri/unichrome/via_texcombine.c
+++ b/src/mesa/drivers/dri/unichrome/via_texcombine.c
@@ -31,11 +31,11 @@
 
 #include <stdio.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "colormac.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/colormac.h"
+#include "main/enums.h"
 
 #include "via_context.h"
 #include "via_state.h"
diff --git a/src/mesa/drivers/dri/unichrome/via_tris.c b/src/mesa/drivers/dri/unichrome/via_tris.c
index 4cc7942b1b6..79e67620c9e 100644
--- a/src/mesa/drivers/dri/unichrome/via_tris.c
+++ b/src/mesa/drivers/dri/unichrome/via_tris.c
@@ -25,12 +25,12 @@
 #include <stdio.h>
 #include <math.h>
 
-#include "glheader.h"
-#include "context.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
-#include "enums.h"
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
+#include "main/enums.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -625,13 +625,12 @@ static void viaFastRenderClippedPoly(GLcontext *ctx, const GLuint *elts,
     }
 }
 
+
 /**********************************************************************/
 /*                    Choose render functions                         */
 /**********************************************************************/
 
 
-
-
 #define _VIA_NEW_VERTEX (_NEW_TEXTURE |                         \
                          _DD_NEW_SEPARATE_SPECULAR |            \
                          _DD_NEW_TRI_UNFILLED |                 \
@@ -665,14 +664,17 @@ static void viaChooseRenderState(GLcontext *ctx)
       vmesa->drawTri = via_draw_triangle;
    }
 
-   if (flags & (ANY_FALLBACK_FLAGS|ANY_RASTER_FLAGS)) {
-      if (flags & DD_TRI_LIGHT_TWOSIDE)    index |= VIA_TWOSIDE_BIT;
-      if (flags & DD_TRI_OFFSET)           index |= VIA_OFFSET_BIT;
-      if (flags & DD_TRI_UNFILLED)         index |= VIA_UNFILLED_BIT;
-      if (flags & ANY_FALLBACK_FLAGS)      index |= VIA_FALLBACK_BIT;
-
-      /* Hook in fallbacks for specific primitives.
-       */
+   if (flags & (ANY_FALLBACK_FLAGS | ANY_RASTER_FLAGS)) {
+      if (ctx->Light.Enabled && ctx->Light.Model.TwoSide)
+         index |= VIA_TWOSIDE_BIT;
+      if (ctx->Polygon.FrontMode != GL_FILL || ctx->Polygon.BackMode != GL_FILL)
+         index |= VIA_UNFILLED_BIT;
+      if (flags & DD_TRI_OFFSET)
+         index |= VIA_OFFSET_BIT;
+      if (flags & ANY_FALLBACK_FLAGS)
+         index |= VIA_FALLBACK_BIT;
+
+      /* Hook in fallbacks for specific primitives. */
       if (flags & POINT_FALLBACK)
 	 vmesa->drawPoint = via_fallback_point;
       
@@ -683,11 +685,8 @@ static void viaChooseRenderState(GLcontext *ctx)
 	 vmesa->drawTri = via_fallback_tri;
    }
 
-
-   if ((flags & DD_SEPARATE_SPECULAR) &&
-       ctx->Light.ShadeModel == GL_FLAT) {
+   if ((flags & DD_SEPARATE_SPECULAR) && ctx->Light.ShadeModel == GL_FLAT)
       index = VIA_MAX_TRIFUNC;	/* flat specular */
-   }
 
    if (vmesa->renderIndex != index) {
       vmesa->renderIndex = index;
diff --git a/src/mesa/drivers/dri/unichrome/via_tris.h b/src/mesa/drivers/dri/unichrome/via_tris.h
index b4f0880a7ab..bc6ef4e4eba 100644
--- a/src/mesa/drivers/dri/unichrome/via_tris.h
+++ b/src/mesa/drivers/dri/unichrome/via_tris.h
@@ -25,7 +25,7 @@
 #ifndef _VIATRIS_H
 #define _VIATRIS_H
 
-#include "mtypes.h"
+#include "main/mtypes.h"
 
 extern void viaPrintRenderState(const char *msg, GLuint state);
 extern void viaInitTriFuncs(GLcontext *ctx);
diff --git a/src/mesa/drivers/fbdev/Makefile b/src/mesa/drivers/fbdev/Makefile
index c0ef54f604f..ee73f29a46a 100644
--- a/src/mesa/drivers/fbdev/Makefile
+++ b/src/mesa/drivers/fbdev/Makefile
@@ -25,8 +25,7 @@ default: $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)
 
 
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(CORE_MESA) $(OBJECTS)
-	@ $(TOP)/bin/mklib -o $(GL_LIB) \
-		-linker '$(CC)' -ldflags '$(LDFLAGS)' \
+	@ $(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
 		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
 		$(CORE_MESA) $(OBJECTS) $(GL_LIB_DEPS)
diff --git a/src/mesa/drivers/fbdev/glfbdev.c b/src/mesa/drivers/fbdev/glfbdev.c
index e95a424698c..3c874ba57ab 100644
--- a/src/mesa/drivers/fbdev/glfbdev.c
+++ b/src/mesa/drivers/fbdev/glfbdev.c
@@ -41,19 +41,19 @@
 
 #ifdef USE_GLFBDEV_DRIVER
 
-#include "glheader.h"
 #include <linux/fb.h>
 #include "GL/glfbdev.h"
-#include "buffers.h"
-#include "context.h"
-#include "extensions.h"
-#include "fbobject.h"
-#include "framebuffer.h"
-#include "imports.h"
-#include "renderbuffer.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texstore.h"
+#include "main/glheader.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/fbobject.h"
+#include "main/framebuffer.h"
+#include "main/imports.h"
+#include "main/renderbuffer.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
 #include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -790,6 +790,11 @@ glFBDevCreateContext( const GLFBDevVisualPtr visual, GLFBDevContextPtr share )
    }
 
    _mesa_enable_sw_extensions(glctx);
+   _mesa_enable_1_3_extensions(glctx);
+   _mesa_enable_1_4_extensions(glctx);
+   _mesa_enable_1_5_extensions(glctx);
+   _mesa_enable_2_0_extensions(glctx);
+   _mesa_enable_2_1_extensions(glctx);
 
    return ctx;
 }
diff --git a/src/mesa/drivers/ggi/ggimesa.c b/src/mesa/drivers/ggi/ggimesa.c
index 85592d98dab..bc08144d663 100644
--- a/src/mesa/drivers/ggi/ggimesa.c
+++ b/src/mesa/drivers/ggi/ggimesa.c
@@ -29,20 +29,20 @@
 
 #include <ggi/mesa/ggimesa_int.h>
 #include <ggi/mesa/debug.h>
-#include "extensions.h"
-#include "buffers.h"
-#include "colormac.h"
-#include "imports.h"
-#include "matrix.h"
+#include "main/extensions.h"
+#include "main/buffers.h"
+#include "main/colormac.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/teximage.h"
+#include "main/texformat.h"
+#include "main/texstore.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 #include "vbo/vbo.h"
-#include "teximage.h"
-#include "texformat.h"
-#include "texstore.h"
 
 /* We use LibGG to manage config files */
 #include <ggi/gg.h>
diff --git a/src/mesa/drivers/glide/fxapi.c b/src/mesa/drivers/glide/fxapi.c
index 00b9d29fac1..b7708fd636a 100644
--- a/src/mesa/drivers/glide/fxapi.c
+++ b/src/mesa/drivers/glide/fxapi.c
@@ -43,7 +43,7 @@
 #include "fxdrv.h"
 
 #include "drivers/common/driverfuncs.h"
-#include "framebuffer.h"
+#include "main/framebuffer.h"
 
 #ifndef TDFX_DEBUG
 int TDFX_DEBUG = (0
diff --git a/src/mesa/drivers/glide/fxdd.c b/src/mesa/drivers/glide/fxdd.c
index 4e56f7bde50..213ef2382fe 100644
--- a/src/mesa/drivers/glide/fxdd.c
+++ b/src/mesa/drivers/glide/fxdd.c
@@ -40,15 +40,15 @@
 
 #if defined(FX)
 
-#include "image.h"
-#include "mtypes.h"
+#include "main/image.h"
+#include "main/mtypes.h"
 #include "fxdrv.h"
-#include "buffers.h"
-#include "enums.h"
-#include "extensions.h"
-#include "macros.h"
-#include "texstore.h"
-#include "teximage.h"
+#include "main/buffers.h"
+#include "main/enums.h"
+#include "main/extensions.h"
+#include "main/macros.h"
+#include "main/texstore.h"
+#include "main/teximage.h"
 #include "swrast/swrast.h"
 #include "swrast/s_context.h"
 #include "swrast_setup/swrast_setup.h"
@@ -1955,8 +1955,8 @@ fx_check_IsInHardware(GLcontext * ctx)
       return FX_FALLBACK_STENCIL;
    }
 
-   if (ctx->DrawBuffer->_ColorDrawBufferMask[0] != BUFFER_BIT_FRONT_LEFT &&
-       ctx->DrawBuffer->_ColorDrawBufferMask[0] != BUFFER_BIT_BACK_LEFT) {
+   if ((ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_FRONT_LEFT) &&
+       (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_BACK_LEFT)) {
       return FX_FALLBACK_DRAW_BUFFER;
    }
 
diff --git a/src/mesa/drivers/glide/fxddspan.c b/src/mesa/drivers/glide/fxddspan.c
index 3ea9f73dd20..d3a58a301c2 100644
--- a/src/mesa/drivers/glide/fxddspan.c
+++ b/src/mesa/drivers/glide/fxddspan.c
@@ -275,6 +275,8 @@
 #undef BYTESPERPIXEL
 #define BYTESPERPIXEL 2
 
+#define VALUE_TYPE GLushort
+
 #define WRITE_DEPTH( _x, _y, d )					\
     *(GLushort *)(buf + _x*BYTESPERPIXEL + _y*pitch) = d
 
@@ -302,6 +304,8 @@
 #undef BYTESPERPIXEL
 #define BYTESPERPIXEL 4
 
+#define VALUE_TYPE GLuint
+
 #define WRITE_DEPTH( _x, _y, d )					\
     *(GLuint *)(buf + _x*BYTESPERPIXEL + _y*pitch) = d << 8
 
diff --git a/src/mesa/drivers/glide/fxddtex.c b/src/mesa/drivers/glide/fxddtex.c
index cb2c8554b1c..f3cd9081810 100644
--- a/src/mesa/drivers/glide/fxddtex.c
+++ b/src/mesa/drivers/glide/fxddtex.c
@@ -39,13 +39,13 @@
 #if defined(FX)
 
 #include "fxdrv.h"
-#include "enums.h"
-#include "image.h"
-#include "teximage.h"
-#include "texformat.h"
-#include "texcompress.h"
-#include "texobj.h"
-#include "texstore.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/teximage.h"
+#include "main/texformat.h"
+#include "main/texcompress.h"
+#include "main/texobj.h"
+#include "main/texstore.h"
 
 
 /* no borders! can't halve 1x1! (stride > width * comp) not allowed */
diff --git a/src/mesa/drivers/glide/fxdrv.h b/src/mesa/drivers/glide/fxdrv.h
index dc19a912da5..bee10de2f49 100644
--- a/src/mesa/drivers/glide/fxdrv.h
+++ b/src/mesa/drivers/glide/fxdrv.h
@@ -48,11 +48,11 @@
 #include <signal.h>
 #endif
 
-#include "context.h"
-#include "imports.h"
-#include "macros.h"
-#include "matrix.h"
-#include "mtypes.h"
+#include "main/context.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/matrix.h"
+#include "main/mtypes.h"
 
 #include "GL/fxmesa.h"
 #include "fxglidew.h"
diff --git a/src/mesa/drivers/glide/fxsetup.c b/src/mesa/drivers/glide/fxsetup.c
index 240e5e0b597..d48726a62ae 100644
--- a/src/mesa/drivers/glide/fxsetup.c
+++ b/src/mesa/drivers/glide/fxsetup.c
@@ -41,11 +41,11 @@
 #if defined(FX)
 
 #include "fxdrv.h"
-#include "enums.h"
-#include "tnl.h"
+#include "main/enums.h"
+#include "main/texstore.h"
+#include "tnl/tnl.h"
 #include "tnl/t_context.h"
-#include "swrast.h"
-#include "texstore.h"
+#include "swrast/swrast.h"
 
 
 static void
diff --git a/src/mesa/drivers/glide/fxtris.c b/src/mesa/drivers/glide/fxtris.c
index 0b9b3be5829..aff91fe7d4c 100644
--- a/src/mesa/drivers/glide/fxtris.c
+++ b/src/mesa/drivers/glide/fxtris.c
@@ -31,10 +31,10 @@
 
 #ifdef FX
 
-#include "imports.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/t_context.h"
diff --git a/src/mesa/drivers/glide/fxvb.c b/src/mesa/drivers/glide/fxvb.c
index 34ada61f4e4..1dc5f9891a1 100644
--- a/src/mesa/drivers/glide/fxvb.c
+++ b/src/mesa/drivers/glide/fxvb.c
@@ -34,11 +34,11 @@
 
 #ifdef FX
 
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "math/m_translate.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/glslcompiler/Makefile b/src/mesa/drivers/glslcompiler/Makefile
index 858457ddd40..dc4abd44d49 100644
--- a/src/mesa/drivers/glslcompiler/Makefile
+++ b/src/mesa/drivers/glslcompiler/Makefile
@@ -41,4 +41,4 @@ glslcompiler.o: glslcompiler.c
 
 
 clean:
-	rm -f *.o *~ $(PROGRAM)
+	-rm -f *.o *~ $(PROGRAM)
diff --git a/src/mesa/drivers/glslcompiler/glslcompiler.c b/src/mesa/drivers/glslcompiler/glslcompiler.c
index 016b53d57aa..34cce977c86 100644
--- a/src/mesa/drivers/glslcompiler/glslcompiler.c
+++ b/src/mesa/drivers/glslcompiler/glslcompiler.c
@@ -45,11 +45,11 @@
  */
 
 
-#include "imports.h"
-#include "context.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "shaders.h"
+#include "main/imports.h"
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/shaders.h"
 #include "shader/shader_api.h"
 #include "shader/prog_print.h"
 #include "drivers/common/driverfuncs.h"
@@ -226,12 +226,8 @@ PrintShaderInstructions(GLuint shader, FILE *f)
 {
    GET_CURRENT_CONTEXT(ctx);
    struct gl_shader *sh = _mesa_lookup_shader(ctx, shader);
-   GLuint i;
-
-   for (i = 0; i < sh->NumPrograms; i++) {
-      struct gl_program *prog = sh->Programs[i];
-      _mesa_print_program_opt(prog, Options.Mode, Options.LineNumbers);
-   }
+   struct gl_program *prog = sh->Program;
+   _mesa_print_program_opt(prog, Options.Mode, Options.LineNumbers);
 }
 
 
diff --git a/src/mesa/drivers/osmesa/Makefile b/src/mesa/drivers/osmesa/Makefile
index fa8dffcb3ef..d6a18b1d755 100644
--- a/src/mesa/drivers/osmesa/Makefile
+++ b/src/mesa/drivers/osmesa/Makefile
@@ -45,11 +45,10 @@ default:
 osmesa8: $(TOP)/lib/$(OSMESA_LIB_NAME)
 
 $(TOP)/lib/$(OSMESA_LIB_NAME): $(OBJECTS)
-	$(TOP)/bin/mklib -o $(OSMESA_LIB) \
-		-linker "$(CC)" \
+	$(MKLIB) -o $(OSMESA_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
-		-install $(TOP)/$(LIB_DIR) \
-		$(MKLIB_OPTIONS) \
+		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
+		-id $(INSTALL_LIB_DIR)/lib$(OSMESA_LIB).$(MESA_MAJOR).dylib \
 		$(OSMESA_LIB_DEPS) $(OBJECTS)
 
 
@@ -58,11 +57,10 @@ $(TOP)/lib/$(OSMESA_LIB_NAME): $(OBJECTS)
 # The libOSMesa16/libOSMesa32 libraries do not use libGL but rather are built
 # with all the other Mesa sources (compiled with -DCHAN_BITS=16/32
 osmesa16: $(OBJECTS) $(CORE_MESA)
-	$(TOP)/bin/mklib -o $(OSMESA_LIB) \
-		-linker "$(CC)" \
+	$(MKLIB) -o $(OSMESA_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
-		-install $(TOP)/$(LIB_DIR) \
-		$(MKLIB_OPTIONS) \
+		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
+		-id $(INSTALL_LIB_DIR)/lib$(OSMESA_LIB).$(MESA_MAJOR).dylib \
 		$(OSMESA_LIB_DEPS) $(OBJECTS) $(CORE_MESA)
 
 
diff --git a/src/mesa/drivers/osmesa/descrip.mms b/src/mesa/drivers/osmesa/descrip.mms
new file mode 100644
index 00000000000..6c2f3226f64
--- /dev/null
+++ b/src/mesa/drivers/osmesa/descrip.mms
@@ -0,0 +1,45 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [----.include.gl]
+	define math [--.math]
+	define tnl [--.tnl]
+	define vbo [--.vbo]
+	define swrast [--.swrast]
+	define swrast_setup [--.swrast_setup]
+	define array_cache [--.array_cache]
+	define drivers [-]
+	define glapi [--.glapi]
+	define main [--.main]
+	define shader [--.shader]
+
+.include [----]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [----.include],[--.main],[--.glapi],[--.shader]
+LIBDIR = [----.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = osmesa.c
+
+OBJECTS = osmesa.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+osmesa.obj : osmesa.c
diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c
index 399997bdecb..f2367bbbb70 100644
--- a/src/mesa/drivers/osmesa/osmesa.c
+++ b/src/mesa/drivers/osmesa/osmesa.c
@@ -1226,6 +1226,8 @@ OSMesaCreateContextExt( GLenum format, GLint depthBits, GLint stencilBits,
       _mesa_enable_1_3_extensions(&(osmesa->mesa));
       _mesa_enable_1_4_extensions(&(osmesa->mesa));
       _mesa_enable_1_5_extensions(&(osmesa->mesa));
+      _mesa_enable_2_0_extensions(&(osmesa->mesa));
+      _mesa_enable_2_1_extensions(&(osmesa->mesa));
 
       osmesa->gl_buffer = _mesa_create_framebuffer(osmesa->gl_visual);
       if (!osmesa->gl_buffer) {
diff --git a/src/mesa/drivers/svga/svgamesa.c b/src/mesa/drivers/svga/svgamesa.c
index 1e4e185d659..5afa8e188bb 100644
--- a/src/mesa/drivers/svga/svgamesa.c
+++ b/src/mesa/drivers/svga/svgamesa.c
@@ -38,12 +38,12 @@
 #include <string.h>
 #include <vga.h>
 #include "GL/svgamesa.h"
-#include "buffers.h"
-#include "context.h"
-#include "extensions.h"
-#include "imports.h"
-#include "matrix.h"
-#include "mtypes.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/mtypes.h"
 #include "swrast/swrast.h"
 #include "svgapix.h"
 #include "svgamesa8.h"
diff --git a/src/mesa/drivers/svga/svgamesa24.c b/src/mesa/drivers/svga/svgamesa24.c
index c7c095333fe..f2ec9c0364a 100644
--- a/src/mesa/drivers/svga/svgamesa24.c
+++ b/src/mesa/drivers/svga/svgamesa24.c
@@ -39,7 +39,7 @@
 
 #if 0
 /* this doesn't compile with GCC on RedHat 6.1 */
-static inline int RGB2BGR24(int c)
+static INLINE int RGB2BGR24(int c)
 {
 	asm("rorw  $8, %0\n"	 
 	    "rorl $16, %0\n"	 
diff --git a/src/mesa/drivers/svga/svgamesa32.c b/src/mesa/drivers/svga/svgamesa32.c
index d089c20c050..8eea3cbe648 100644
--- a/src/mesa/drivers/svga/svgamesa32.c
+++ b/src/mesa/drivers/svga/svgamesa32.c
@@ -39,7 +39,7 @@
 
 #if 0
 /* this doesn't compile with GCC on RedHat 6.1 */
-static inline int RGB2BGR32(int c)
+static INLINE int RGB2BGR32(int c)
 {
 	asm("rorw  $8, %0\n"	 
 	    "rorl $16, %0\n"	 
diff --git a/src/mesa/drivers/svga/svgapix.h b/src/mesa/drivers/svga/svgapix.h
index 19cb74487da..c8cee37ca68 100644
--- a/src/mesa/drivers/svga/svgapix.h
+++ b/src/mesa/drivers/svga/svgapix.h
@@ -32,8 +32,8 @@
 
 #include "GL/gl.h"
 #include "GL/svgamesa.h"
-#include "context.h"
-#include "colormac.h"
+#include "main/context.h"
+#include "main/colormac.h"
 #include "vga.h"
 
 struct svgamesa_context {
diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def
index c525945426f..3f2d644e86e 100644
--- a/src/mesa/drivers/windows/gdi/mesa.def
+++ b/src/mesa/drivers/windows/gdi/mesa.def
@@ -867,8 +867,6 @@ EXPORTS
 	_glapi_get_proc_address
 	_mesa_add_soft_renderbuffers
 	_mesa_add_renderbuffer
-	_mesa_attach_shader
-	_mesa_bind_attrib_location
 	_mesa_buffer_data
 	_mesa_buffer_get_subdata
 	_mesa_buffer_map
@@ -877,24 +875,20 @@ EXPORTS
 	_mesa_bzero
 	_mesa_calloc
 	_mesa_choose_tex_format
-	_mesa_compile_shader
 	_mesa_compressed_texture_size
 	_mesa_create_framebuffer
-	_mesa_create_program
-	_mesa_create_shader
 	_mesa_create_visual
 	_mesa_delete_array_object
 	_mesa_delete_buffer_object
 	_mesa_delete_program
-	_mesa_delete_program2
-	_mesa_delete_shader
 	_mesa_delete_texture_object
 	_mesa_destroy_framebuffer
 	_mesa_destroy_visual
-	_mesa_detach_shader
 	_mesa_enable_1_3_extensions
 	_mesa_enable_1_4_extensions
 	_mesa_enable_1_5_extensions
+	_mesa_enable_2_0_extensions
+	_mesa_enable_2_1_extensions
 	_mesa_enable_sw_extensions
 	_mesa_error
 	_mesa_finish_render_texture
@@ -902,28 +896,15 @@ EXPORTS
 	_mesa_free
 	_mesa_free_context_data
 	_mesa_free_texture_image_data
-	_mesa_get_active_attrib
-	_mesa_get_active_uniform 
-	_mesa_get_attached_shaders
-	_mesa_get_attrib_location
+	_mesa_generate_mipmap
 	_mesa_get_compressed_teximage
 	_mesa_get_current_context
-	_mesa_get_handle
-	_mesa_get_programiv
-	_mesa_get_program_info_log
 	_mesa_get_program_register
-	_mesa_get_shaderiv
-	_mesa_get_shader_info_log
-	_mesa_get_shader_source
 	_mesa_get_teximage
-	_mesa_get_uniformfv
-	_mesa_get_uniform_location
 	_mesa_init_driver_functions
+	_mesa_init_glsl_driver_functions
 	_mesa_init_renderbuffer
 	_mesa_initialize_context
-	_mesa_is_program
-	_mesa_is_shader
-	_mesa_link_program
 	_mesa_make_current
 	_mesa_memcpy
 	_mesa_memset
@@ -942,7 +923,6 @@ EXPORTS
 	_mesa_render_texture
 	_mesa_ResizeBuffersMESA
 	_mesa_resize_framebuffer
-	_mesa_shader_source
 	_mesa_store_compressed_teximage1d
 	_mesa_store_compressed_teximage2d
 	_mesa_store_compressed_teximage3d
@@ -957,12 +937,9 @@ EXPORTS
 	_mesa_store_texsubimage3d
 	_mesa_strcmp
 	_mesa_test_proxy_teximage
-	_mesa_uniform
-	_mesa_uniform_matrix
 	_mesa_unreference_framebuffer
 	_mesa_update_framebuffer_visual
 	_mesa_use_program
-	_mesa_validate_program
 	_mesa_Viewport
 	_swrast_Accum
 	_swrast_Bitmap
diff --git a/src/mesa/drivers/windows/gdi/wgl.c b/src/mesa/drivers/windows/gdi/wgl.c
index 6e00d08abaf..9f0bb9122a9 100644
--- a/src/mesa/drivers/windows/gdi/wgl.c
+++ b/src/mesa/drivers/windows/gdi/wgl.c
@@ -54,7 +54,7 @@
 #include <windows.h>
 
 #endif
-
+#include "config.h"
 #include "glapi.h"
 #include "GL/wmesa.h"   /* protos for wmesa* functions */
 
@@ -69,10 +69,12 @@ struct __pixelformat__
     GLboolean doubleBuffered;
 };
 
+
+
 /* These are the PFD's supported by this driver. */
 struct __pixelformat__	pfd[] =
 {
-#if 0
+#if 0 
     /* Double Buffer, alpha */
     {	
 	{	
@@ -86,7 +88,7 @@ struct __pixelformat__	pfd[] =
 	    8, 16,	
 	    8, 24,
 	    0, 0, 0, 0, 0,	
-	    16,	8,	
+	    DEFAULT_SOFTWARE_DEPTH_BITS,	8,	
 	    0, 0, 0,	
 	    0, 0, 0 
 	},
@@ -105,13 +107,13 @@ struct __pixelformat__	pfd[] =
 	    8, 16,	
 	    8, 24,
 	    0, 0, 0, 0,	0,	
-	    16,	8,
+	    DEFAULT_SOFTWARE_DEPTH_BITS,	8,	
 	    0, 0, 0,	
 	    0, 0, 0
 	},
         GL_FALSE
     },
-#endif
+#endif 
     /* Double Buffer, no alpha */
     {	
 	{	
@@ -125,7 +127,7 @@ struct __pixelformat__	pfd[] =
 	    8, 16,
 	    0, 0,
 	    0, 0, 0, 0,	0,
-	    16,	8,
+	    DEFAULT_SOFTWARE_DEPTH_BITS,	8,	
 	    0, 0, 0, 
 	    0, 0, 0 
 	},
@@ -144,7 +146,7 @@ struct __pixelformat__	pfd[] =
 	    8, 16,
 	    0, 0,
 	    0, 0, 0, 0,	0,
-	    16,	8,
+	    DEFAULT_SOFTWARE_DEPTH_BITS,	8,	
 	    0, 0, 0,
 	    0, 0, 0 
 	},
@@ -578,6 +580,13 @@ WINGDIAPI BOOL GLAPIENTRY wglUseFontBitmapsA(HDC hdc, DWORD first,
     return success;
 }
 
+WINGDIAPI BOOL GLAPIENTRY wglShareLists(HGLRC hglrc1,
+					HGLRC hglrc2)
+{
+    WMesaShareLists((WMesaContext)hglrc1, (WMesaContext)hglrc2);
+    return(TRUE);
+}
+
 
 
 /* NOT IMPLEMENTED YET */
@@ -597,13 +606,6 @@ WINGDIAPI HGLRC GLAPIENTRY wglCreateLayerContext(HDC hdc,
     return(NULL);
 }
 
-WINGDIAPI BOOL GLAPIENTRY wglShareLists(HGLRC hglrc1,
-					HGLRC hglrc2)
-{
-    (void) hglrc1; (void) hglrc2;
-    return(TRUE);
-}
-
 
 WINGDIAPI BOOL GLAPIENTRY wglUseFontBitmapsW(HDC hdc,
 					     DWORD first,
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 5b67439f0f2..80746950c71 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -56,11 +56,13 @@ wmesa_free_framebuffer(HDC hdc)
 	prev = pwfb;
     }
     if (pwfb) {
+        struct gl_framebuffer *fb;
 	if (pwfb == FirstFramebuffer)
 	    FirstFramebuffer = pwfb->next;
 	else
 	    prev->next = pwfb->next;
-	free(pwfb);
+        fb = &pwfb->Base;
+        _mesa_unreference_framebuffer(&fb); 
     }
 }
 
@@ -118,6 +120,7 @@ static void wmSetPixelFormat(WMesaFramebuffer pwfb, HDC hDC)
     /* Only 16 and 32 bit targets are supported now */
     assert(pwfb->cColorBits == 0 ||
 	   pwfb->cColorBits == 16 || 
+	   pwfb->cColorBits == 24 || 
 	   pwfb->cColorBits == 32);
 
     switch(pwfb->cColorBits){
@@ -127,6 +130,7 @@ static void wmSetPixelFormat(WMesaFramebuffer pwfb, HDC hDC)
     case 16:
 	pwfb->pixelformat = PF_5R6G5B;
 	break;
+    case 24:
     case 32:
 	pwfb->pixelformat = PF_8R8G8B;
 	break;
@@ -458,23 +462,84 @@ static void write_rgba_span_front(const GLcontext *ctx,
 				   const GLubyte rgba[][4], 
 				   const GLubyte mask[] )
 {
-    WMesaContext pwc = wmesa_context(ctx);
-    GLuint i;
-    
-    (void) ctx;
-    y=FLIP(y);
-    if (mask) {
-	for (i=0; i<n; i++)
-	    if (mask[i])
-		SetPixel(pwc->hDC, x+i, y, RGB(rgba[i][RCOMP], rgba[i][GCOMP], 
-					       rgba[i][BCOMP]));
-    }
-    else {
-	for (i=0; i<n; i++)
-	    SetPixel(pwc->hDC, x+i, y, RGB(rgba[i][RCOMP], rgba[i][GCOMP], 
-					   rgba[i][BCOMP]));
-    }
-    
+   WMesaContext pwc = wmesa_context(ctx);
+   WMesaFramebuffer pwfb = wmesa_lookup_framebuffer(pwc->hDC);
+   CONST BITMAPINFO bmi=
+   {
+      {
+         sizeof(BITMAPINFOHEADER),
+         n, 1, 1, 32, BI_RGB, 0, 1, 1, 0, 0
+      }
+   };
+   HBITMAP bmp=0;
+   HDC mdc=0;
+   typedef union
+   {
+      unsigned i;
+      struct {
+         unsigned b:8, g:8, r:8, a:8;
+      };
+   } BGRA;
+   BGRA *bgra, c;
+   int i;
+
+   if (n < 16) {   // the value 16 is just guessed
+      y=FLIP(y);
+      if (mask) {
+         for (i=0; i<n; i++)
+            if (mask[i])
+               SetPixel(pwc->hDC, x+i, y,
+                        RGB(rgba[i][RCOMP], rgba[i][GCOMP], rgba[i][BCOMP]));
+      }
+      else {
+         for (i=0; i<n; i++)
+            SetPixel(pwc->hDC, x+i, y,
+                     RGB(rgba[i][RCOMP], rgba[i][GCOMP], rgba[i][BCOMP]));
+      }
+   }
+   else {
+      if (!pwfb) {
+         _mesa_problem(NULL, "wmesa: write_rgba_span_front on unknown hdc");
+         return;
+      }
+      bgra=malloc(n*sizeof(BGRA));
+      if (!bgra) {
+         _mesa_problem(NULL, "wmesa: write_rgba_span_front: out of memory");
+         return;
+      }
+      c.a=0;
+      if (mask) {
+         for (i=0; i<n; i++) {
+            if (mask[i]) {
+               c.r=rgba[i][RCOMP];
+               c.g=rgba[i][GCOMP];
+               c.b=rgba[i][BCOMP];
+               c.a=rgba[i][ACOMP];
+               bgra[i]=c;
+            }
+            else
+               bgra[i].i=0;
+         }
+      }
+      else {
+         for (i=0; i<n; i++) {
+            c.r=rgba[i][RCOMP];
+            c.g=rgba[i][GCOMP];
+            c.b=rgba[i][BCOMP];
+            c.a=rgba[i][ACOMP];
+            bgra[i]=c;
+         }
+      }
+      bmp=CreateBitmap(n, 1,  1, 32, bgra);
+      mdc=CreateCompatibleDC(pwfb->hDC);
+      SelectObject(mdc, bmp);
+      y=FLIP(y);
+      BitBlt(pwfb->hDC, x, y, n, 1, mdc, 0, 0, SRCCOPY);
+      SelectObject(mdc, 0);
+      DeleteObject(bmp);
+      DeleteDC(mdc);
+      free(bgra);
+   }
 }
 
 /* Write a horizontal span of RGB color pixels with a boolean mask. */
@@ -795,6 +860,195 @@ static void read_rgba_pixels_32(const GLcontext *ctx,
 
 /*********************************************************************/
 
+/* DOUBLE BUFFER 24-bit */
+
+#define WMSETPIXEL24(pwc, y, x, r, g, b) { \
+LPBYTE lpb = ((LPBYTE)((pwc)->pbPixels + (pwc)->ScanWidth * (y)) + (3 * x)); \
+lpb[0] = (b); \
+lpb[1] = (g); \
+lpb[2] = (r); }
+
+/* Write a horizontal span of RGBA color pixels with a boolean mask. */
+static void write_rgba_span_24(const GLcontext *ctx, 
+			       struct gl_renderbuffer *rb, 
+			       GLuint n, GLint x, GLint y,
+			       const GLubyte rgba[][4], 
+			       const GLubyte mask[] )
+{
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    GLuint i;
+    LPBYTE lpb;
+
+    (void) ctx;
+    
+    y=FLIP(y);
+    lpb = ((LPBYTE)(pwfb->pbPixels + pwfb->ScanWidth * y)) + (3 * x);
+    if (mask) {
+	for (i=0; i<n; i++)
+	    if (mask[i]) {
+                lpb[3*i] = rgba[i][BCOMP];
+                lpb[3*i+1] = rgba[i][GCOMP];
+                lpb[3*i+2] = rgba[i][RCOMP];
+	    }
+    }
+    else {
+	    for (i=0; i<n; i++) {
+            *lpb++ = rgba[i][BCOMP];
+            *lpb++ = rgba[i][GCOMP];
+            *lpb++ = rgba[i][RCOMP];
+	    }
+    }
+}
+
+
+/* Write a horizontal span of RGB color pixels with a boolean mask. */
+static void write_rgb_span_24(const GLcontext *ctx, 
+			      struct gl_renderbuffer *rb, 
+			      GLuint n, GLint x, GLint y,
+			      const GLubyte rgb[][3], 
+			      const GLubyte mask[] )
+{
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    GLuint i;
+    LPBYTE lpb;
+
+    (void) ctx;
+    
+    y=FLIP(y);
+    lpb = ((LPBYTE)(pwfb->pbPixels + pwfb->ScanWidth * y)) + (3 * x);
+    if (mask) {
+	for (i=0; i<n; i++)
+	    if (mask[i]) {
+            lpb[3*i] = rgb[i][BCOMP];
+            lpb[3*i+1] = rgb[i][GCOMP];
+            lpb[3*i+2] = rgb[i][RCOMP];
+	    }
+    }
+    else {
+    	for (i=0; i<n; i++) {
+    		*lpb++ = rgb[i][BCOMP];
+    		*lpb++ = rgb[i][GCOMP];
+    		*lpb++ = rgb[i][RCOMP];
+    	}
+    }
+}
+
+/*
+ * Write a horizontal span of pixels with a boolean mask.  The current color
+ * is used for all pixels.
+ */
+static void write_mono_rgba_span_24(const GLcontext *ctx, 
+				    struct gl_renderbuffer *rb,
+				    GLuint n, GLint x, GLint y,
+				    const GLchan color[4], 
+				    const GLubyte mask[])
+{
+    LPBYTE lpb;
+    GLuint i;
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    lpb = ((LPBYTE)(pwfb->pbPixels + pwfb->ScanWidth * y)) + (3 * x);
+    y=FLIP(y);
+    if (mask) {
+	for (i=0; i<n; i++)
+	    if (mask[i]) {
+	    	lpb[3*i] = color[BCOMP];
+	    	lpb[3*i+1] = color[GCOMP];
+	    	lpb[3*i+2] = color[RCOMP];
+	    }
+    }
+    else
+	for (i=0; i<n; i++) {
+		*lpb++ = color[BCOMP];
+		*lpb++ = color[GCOMP];
+		*lpb++ = color[RCOMP];		
+	}
+}
+
+/* Write an array of RGBA pixels with a boolean mask. */
+static void write_rgba_pixels_24(const GLcontext *ctx, 
+				 struct gl_renderbuffer *rb,
+				 GLuint n, const GLint x[], const GLint y[],
+				 const GLubyte rgba[][4], 
+				 const GLubyte mask[])
+{
+    GLuint i;
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    for (i=0; i<n; i++)
+	if (mask[i])
+	    WMSETPIXEL24(pwfb, FLIP(y[i]), x[i],
+			 rgba[i][RCOMP], rgba[i][GCOMP], rgba[i][BCOMP]);
+}
+
+/*
+ * Write an array of pixels with a boolean mask.  The current color
+ * is used for all pixels.
+ */
+static void write_mono_rgba_pixels_24(const GLcontext *ctx, 
+				      struct gl_renderbuffer *rb,
+				      GLuint n,
+				      const GLint x[], const GLint y[],
+				      const GLchan color[4],
+				      const GLubyte mask[])
+{
+    GLuint i;
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    for (i=0; i<n; i++)
+	if (mask[i])
+	    WMSETPIXEL24(pwfb, FLIP(y[i]),x[i],color[RCOMP],
+			 color[GCOMP], color[BCOMP]);
+}
+
+/* Read a horizontal span of color pixels. */
+static void read_rgba_span_24(const GLcontext *ctx, 
+			      struct gl_renderbuffer *rb,
+			      GLuint n, GLint x, GLint y,
+			      GLubyte rgba[][4] )
+{
+    GLuint i;
+    LPBYTE lpb;
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+    
+    y = FLIP(y);
+    lpb = ((LPBYTE)(pwfb->pbPixels + pwfb->ScanWidth * y)) + (3 * x);
+    for (i=0; i<n; i++) {
+	rgba[i][RCOMP] = lpb[3*i+2];
+	rgba[i][GCOMP] = lpb[3*i+1];
+	rgba[i][BCOMP] = lpb[3*i];
+	rgba[i][ACOMP] = 255;
+    }
+}
+
+
+/* Read an array of color pixels. */
+static void read_rgba_pixels_24(const GLcontext *ctx, 
+				struct gl_renderbuffer *rb,
+				GLuint n, const GLint x[], const GLint y[],
+				GLubyte rgba[][4])
+{
+    GLuint i;
+    LPBYTE lpb;
+    WMesaContext pwc = wmesa_context(ctx);
+    WMesaFramebuffer pwfb = wmesa_framebuffer(ctx->DrawBuffer);
+
+    for (i=0; i<n; i++) {
+	GLint y2 = FLIP(y[i]);
+	lpb = ((LPBYTE)(pwfb->pbPixels + pwfb->ScanWidth * y2)) + (3 * x[i]);
+	rgba[i][RCOMP] = lpb[3*i+2];
+	rgba[i][GCOMP] = lpb[3*i+1];
+	rgba[i][BCOMP] = lpb[3*i];
+	rgba[i][ACOMP] = 255;
+  }
+}
+
+
+/*********************************************************************/
+
 /* DOUBLE BUFFER 16-bit */
 
 #define WMSETPIXEL16(pwc, y, x, r, g, b) { \
@@ -1016,7 +1270,7 @@ wmesa_renderbuffer_storage(GLcontext *ctx,
  * on if we're drawing to the front or back color buffer.
  */
 void wmesa_set_renderbuffer_funcs(struct gl_renderbuffer *rb, int pixelformat,
-                                  int double_buffer)
+                                  BYTE cColorBits, int double_buffer)
 {
     if (double_buffer) {
         /* back buffer */
@@ -1036,16 +1290,32 @@ void wmesa_set_renderbuffer_funcs(struct gl_renderbuffer *rb, int pixelformat,
             rb->BlueBits = 5;
 	    break;
 	case PF_8R8G8B:
-	    rb->PutRow = write_rgba_span_32;
-	    rb->PutRowRGB = write_rgb_span_32;
-	    rb->PutMonoRow = write_mono_rgba_span_32;
-	    rb->PutValues = write_rgba_pixels_32;
-	    rb->PutMonoValues = write_mono_rgba_pixels_32;
-	    rb->GetRow = read_rgba_span_32;
-	    rb->GetValues = read_rgba_pixels_32;
+		if (cColorBits == 24)
+		{
+		    rb->PutRow = write_rgba_span_24;
+		    rb->PutRowRGB = write_rgb_span_24;
+		    rb->PutMonoRow = write_mono_rgba_span_24;
+		    rb->PutValues = write_rgba_pixels_24;
+		    rb->PutMonoValues = write_mono_rgba_pixels_24;
+		    rb->GetRow = read_rgba_span_24;
+		    rb->GetValues = read_rgba_pixels_24;
+	        rb->RedBits = 8;
+	        rb->GreenBits = 8;
+	        rb->BlueBits = 8;		
+		}
+		else
+		{
+	        rb->PutRow = write_rgba_span_32;
+	        rb->PutRowRGB = write_rgb_span_32;
+	        rb->PutMonoRow = write_mono_rgba_span_32;
+	        rb->PutValues = write_rgba_pixels_32;
+	        rb->PutMonoValues = write_mono_rgba_pixels_32;
+	        rb->GetRow = read_rgba_span_32;
+	        rb->GetValues = read_rgba_pixels_32;
             rb->RedBits = 8;
             rb->GreenBits = 8;
             rb->BlueBits = 8;
+		}
 	    break;
 	default:
 	    break;
@@ -1235,6 +1505,9 @@ WMesaContext WMesaCreateContext(HDC hDC,
     ctx = &c->gl_ctx;
     _mesa_initialize_context(ctx, visual, NULL, &functions, (void *)c);
 
+    /* visual no longer needed - it was copied by _mesa_initialize_context() */
+    _mesa_destroy_visual(visual);
+
     _mesa_enable_sw_extensions(ctx);
     _mesa_enable_1_3_extensions(ctx);
     _mesa_enable_1_4_extensions(ctx);
@@ -1352,11 +1625,11 @@ void WMesaMakeCurrent(WMesaContext c, HDC hdc)
         if (visual->doubleBufferMode == 1) {
             rb = wmesa_new_renderbuffer();
             _mesa_add_renderbuffer(&pwfb->Base, BUFFER_BACK_LEFT, rb);
-            wmesa_set_renderbuffer_funcs(rb, pwfb->pixelformat, 1);
+            wmesa_set_renderbuffer_funcs(rb, pwfb->pixelformat, pwfb->cColorBits, 1);
 	}
         rb = wmesa_new_renderbuffer();
         _mesa_add_renderbuffer(&pwfb->Base, BUFFER_FRONT_LEFT, rb);
-        wmesa_set_renderbuffer_funcs(rb, pwfb->pixelformat, 0);
+        wmesa_set_renderbuffer_funcs(rb, pwfb->pixelformat, pwfb->cColorBits, 0);
 
 	/* Let Mesa own the Depth, Stencil, and Accum buffers */
         _mesa_add_soft_renderbuffers(&pwfb->Base,
@@ -1401,6 +1674,11 @@ void WMesaSwapBuffers( HDC hdc )
     }
 }
 
+void WMesaShareLists(WMesaContext ctx_to_share, WMesaContext ctx)
+{
+	_mesa_share_state(&ctx->gl_ctx, &ctx_to_share->gl_ctx);	
+}
+
 /* This is hopefully a temporary hack to define some needed dispatch
  * table entries.  Hopefully, I'll find a better solution.  The
  * dispatch table generation scripts ought to be making these dummy
@@ -1439,15 +1717,19 @@ void gl_dispatch_stub_650(void){}
 void gl_dispatch_stub_651(void){}
 void gl_dispatch_stub_652(void){}
 void gl_dispatch_stub_653(void){}
+void gl_dispatch_stub_733(void){}
 void gl_dispatch_stub_734(void){}
 void gl_dispatch_stub_735(void){}
 void gl_dispatch_stub_736(void){}
 void gl_dispatch_stub_737(void){}
 void gl_dispatch_stub_738(void){}
+void gl_dispatch_stub_744(void){}
 void gl_dispatch_stub_745(void){}
 void gl_dispatch_stub_746(void){}
 void gl_dispatch_stub_760(void){}
 void gl_dispatch_stub_761(void){}
+void gl_dispatch_stub_763(void){}
+void gl_dispatch_stub_765(void){}
 void gl_dispatch_stub_766(void){}
 void gl_dispatch_stub_767(void){}
 void gl_dispatch_stub_768(void){}
diff --git a/src/mesa/drivers/x11/Makefile b/src/mesa/drivers/x11/Makefile
index 51226eeae11..d2780e62c97 100644
--- a/src/mesa/drivers/x11/Makefile
+++ b/src/mesa/drivers/x11/Makefile
@@ -55,11 +55,11 @@ default: $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME)
 
 
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(OBJECTS) $(CORE_MESA)
-	@ $(TOP)/bin/mklib -o $(GL_LIB) \
-		-linker '$(CC)' -ldflags '$(LDFLAGS)' \
+	@ $(MKLIB) -o $(GL_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(GL_MAJOR) -minor $(GL_MINOR) -patch $(GL_TINY) \
-		-install $(TOP)/$(LIB_DIR) \
-		$(MKLIB_OPTIONS) $(GL_LIB_DEPS) $(OBJECTS) $(CORE_MESA)
+		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
+		-id $(INSTALL_LIB_DIR)/lib$(GL_LIB).$(GL_MAJOR).dylib \
+		$(GL_LIB_DEPS) $(OBJECTS) $(CORE_MESA)
 
 
 
diff --git a/src/mesa/drivers/x11/descrip.mms b/src/mesa/drivers/x11/descrip.mms
new file mode 100644
index 00000000000..648f683dfa4
--- /dev/null
+++ b/src/mesa/drivers/x11/descrip.mms
@@ -0,0 +1,55 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [----.include.gl]
+	define math [--.math]
+	define tnl [--.tnl]
+	define vbo [--.vbo]
+	define swrast [--.swrast]
+	define swrast_setup [--.swrast_setup]
+	define array_cache [--.array_cache]
+	define drivers [-]
+	define glapi [--.glapi]
+	define main [--.main]
+	define shader [--.shader]
+
+.include [----]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [----.include],[--.main],[--.glapi],[--.shader]
+LIBDIR = [----.lib]
+CFLAGS =/include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = fakeglx.c glxapi.c xfonts.c xm_api.c xm_dd.c xm_line.c xm_span.c\
+	xm_tri.c xm_buffer.c
+
+OBJECTS =fakeglx.obj,glxapi.obj,xfonts.obj,xm_api.obj,xm_dd.obj,xm_line.obj,\
+	xm_span.obj,xm_tri.obj,xm_buffer.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+fakeglx.obj : fakeglx.c
+glxapi.obj : glxapi.c
+xfonts.obj : xfonts.c
+xm_api.obj : xm_api.c
+xm_buffer.obj : xm_buffer.c
+xm_dd.obj : xm_dd.c
+xm_line.obj : xm_line.c
+xm_span.obj : xm_span.c
+xm_tri.obj : xm_tri.c
diff --git a/src/mesa/drivers/x11/fakeglx.c b/src/mesa/drivers/x11/fakeglx.c
index 7a170b4d3d1..827d39f9957 100644
--- a/src/mesa/drivers/x11/fakeglx.c
+++ b/src/mesa/drivers/x11/fakeglx.c
@@ -43,12 +43,12 @@
 #include "glxheader.h"
 #include "glxapi.h"
 #include "GL/xmesa.h"
-#include "context.h"
-#include "config.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "version.h"
+#include "main/context.h"
+#include "main/config.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/version.h"
 #include "xfonts.h"
 #include "xmesaP.h"
 
@@ -1183,11 +1183,12 @@ choose_visual( Display *dpy, int screen, const int *list, GLboolean fbConfig )
           * GLX_ARB_multisample
           */
          case GLX_SAMPLE_BUFFERS_ARB:
-            /* ms not supported */
-            return NULL;
          case GLX_SAMPLES_ARB:
-            /* ms not supported */
-            return NULL;
+	    parselist++;
+	    if (*parselist++ != 0)
+	       /* ms not supported */
+	       return NULL;
+	    break;
 
          /*
           * FBConfig attribs.
diff --git a/src/mesa/drivers/x11/glxapi.c b/src/mesa/drivers/x11/glxapi.c
index 309a0008d76..c2ccce6f520 100644
--- a/src/mesa/drivers/x11/glxapi.c
+++ b/src/mesa/drivers/x11/glxapi.c
@@ -34,8 +34,8 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
-#include "glheader.h"
-#include "glapi.h"
+#include "main/glheader.h"
+#include "glapi/glapi.h"
 #include "glxapi.h"
 
 
diff --git a/src/mesa/drivers/x11/glxheader.h b/src/mesa/drivers/x11/glxheader.h
index a402191f131..d88afba20e7 100644
--- a/src/mesa/drivers/x11/glxheader.h
+++ b/src/mesa/drivers/x11/glxheader.h
@@ -30,10 +30,11 @@
 #include <GL/vms_x_fix.h>
 #endif
 
-#include "glheader.h"
+#include "main/glheader.h"
 
 #ifdef XFree86Server
 
+# include "xorg-server.h"
 # include "resource.h"
 # include "windowstr.h"
 
diff --git a/src/mesa/drivers/x11/sources b/src/mesa/drivers/x11/sources
deleted file mode 100644
index d76d65eaad1..00000000000
--- a/src/mesa/drivers/x11/sources
+++ /dev/null
@@ -1,8 +0,0 @@
-# Note: only listing sources needed for X server renderer
-MESA_DRIVER_X11_SOURCES = \
-xm_api.c \
-xm_buffer.c \
-xm_dd.c \
-xm_line.c \
-xm_span.c \
-xm_tri.c
diff --git a/src/mesa/drivers/x11/xfonts.c b/src/mesa/drivers/x11/xfonts.c
index d72c600bd1b..f732c945865 100644
--- a/src/mesa/drivers/x11/xfonts.c
+++ b/src/mesa/drivers/x11/xfonts.c
@@ -33,8 +33,8 @@
 #endif
 
 #include "glxheader.h"
-#include "context.h"
-#include "imports.h"
+#include "main/context.h"
+#include "main/imports.h"
 #include "xfonts.h"
 
 
diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c
index 18b033666f0..c9009bad031 100644
--- a/src/mesa/drivers/x11/xm_api.c
+++ b/src/mesa/drivers/x11/xm_api.c
@@ -65,29 +65,22 @@
 #include "glxheader.h"
 #include "GL/xmesa.h"
 #include "xmesaP.h"
-#include "context.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "glthread.h"
-#include "imports.h"
-#include "macros.h"
-#include "renderbuffer.h"
-#include "teximage.h"
+#include "main/context.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/renderbuffer.h"
+#include "main/teximage.h"
+#include "glapi/glthread.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "vbo/vbo.h"
-#if 0
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
-#endif
 #include "drivers/common/driverfuncs.h"
 
-#include "state_tracker/st_public.h"
-#include "state_tracker/st_context.h"
-#include "softpipe/sp_context.h"
-#include "pipe/p_defines.h"
-
 /**
  * Global X driver lock
  */
@@ -388,7 +381,7 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type,
    /*
     * Front renderbuffer
     */
-   b->frontxrb = xmesa_create_renderbuffer(NULL, 0, &vis->mesa_visual, GL_FALSE);
+   b->frontxrb = xmesa_new_renderbuffer(NULL, 0, &vis->mesa_visual, GL_FALSE);
    if (!b->frontxrb) {
       _mesa_free(b);
       return NULL;
@@ -397,13 +390,13 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type,
    b->frontxrb->drawable = d;
    b->frontxrb->pixmap = (XMesaPixmap) d;
    _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_FRONT_LEFT,
-                          &b->frontxrb->St.Base);
+                          &b->frontxrb->Base);
 
    /*
     * Back renderbuffer
     */
    if (vis->mesa_visual.doubleBufferMode) {
-      b->backxrb = xmesa_create_renderbuffer(NULL, 0, &vis->mesa_visual, GL_TRUE);
+      b->backxrb = xmesa_new_renderbuffer(NULL, 0, &vis->mesa_visual, GL_TRUE);
       if (!b->backxrb) {
          /* XXX free front xrb too */
          _mesa_free(b);
@@ -414,7 +407,7 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type,
       b->db_mode = vis->ximage_flag ? BACK_XIMAGE : BACK_PIXMAP;
       
       _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_BACK_LEFT,
-                             &b->backxrb->St.Base);
+                             &b->backxrb->Base);
    }
 
    /*
@@ -432,43 +425,14 @@ create_xmesa_buffer(XMesaDrawable d, BufferType type,
       b->swAlpha = GL_FALSE;
    }
 
-   if (vis->mesa_visual.depthBits > 0 &&
-       vis->mesa_visual.stencilBits > 0) {
-      /* combined depth/stencil */
-      struct gl_renderbuffer *rb
-         = st_new_renderbuffer_fb(GL_DEPTH24_STENCIL8_EXT);
-      _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_DEPTH, rb);
-      _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_STENCIL, rb);
-   }
-   else {
-      if (vis->mesa_visual.depthBits > 0) {
-         struct gl_renderbuffer *rb
-         = st_new_renderbuffer_fb(GL_DEPTH_COMPONENT32);
-         _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_DEPTH, rb);
-      }
-
-      if (vis->mesa_visual.stencilBits > 0) {
-         struct gl_renderbuffer *rb
-            = st_new_renderbuffer_fb(GL_STENCIL_INDEX8_EXT);
-         _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_STENCIL, rb);
-      }
-   }
-
-   if (vis->mesa_visual.accumRedBits > 0) {
-      struct gl_renderbuffer *rb
-         = st_new_renderbuffer_fb(GL_RGBA16);
-      _mesa_add_renderbuffer(&b->mesa_buffer, BUFFER_ACCUM, rb);
-   }
-
-
    /*
     * Other renderbuffer (depth, stencil, etc)
     */
    _mesa_add_soft_renderbuffers(&b->mesa_buffer,
-                                GL_FALSE, /* color */
-                                GL_FALSE, /*vis->mesa_visual.haveDepthBuffer,*/
-                                GL_FALSE, /* stencil */
-                                GL_FALSE, /* accum */
+                                GL_FALSE,  /* color */
+                                vis->mesa_visual.haveDepthBuffer,
+                                vis->mesa_visual.haveStencilBuffer,
+                                vis->mesa_visual.haveAccumBuffer,
                                 b->swAlpha,
                                 vis->mesa_visual.numAuxBuffers > 0 );
 
@@ -1339,6 +1303,67 @@ xmesa_convert_from_x_visual_type( int visualType )
 /**********************************************************************/
 
 
+#ifdef IN_DRI_DRIVER
+#define need_GL_VERSION_1_3
+#define need_GL_VERSION_1_4
+#define need_GL_VERSION_1_5
+#define need_GL_VERSION_2_0
+
+/* sw extensions for imaging */
+#define need_GL_EXT_blend_color
+#define need_GL_EXT_blend_minmax
+#define need_GL_EXT_convolution
+#define need_GL_EXT_histogram
+#define need_GL_SGI_color_table
+
+/* sw extensions not associated with some GL version */
+#define need_GL_ARB_shader_objects
+#define need_GL_ARB_vertex_program
+#define need_GL_APPLE_vertex_array_object
+#define need_GL_ATI_fragment_shader
+#define need_GL_EXT_depth_bounds_test
+#define need_GL_EXT_framebuffer_object
+#define need_GL_EXT_framebuffer_blit
+#define need_GL_EXT_gpu_program_parameters
+#define need_GL_EXT_paletted_texture
+#define need_GL_IBM_multimode_draw_arrays
+#define need_GL_MESA_resize_buffers
+#define need_GL_NV_vertex_program
+#define need_GL_NV_fragment_program
+
+#include "extension_helper.h"
+#include "utils.h"
+
+const struct dri_extension card_extensions[] =
+{
+   { "GL_VERSION_1_3",			GL_VERSION_1_3_functions },
+   { "GL_VERSION_1_4",			GL_VERSION_1_4_functions },
+   { "GL_VERSION_1_5",			GL_VERSION_1_5_functions },
+   { "GL_VERSION_2_0",			GL_VERSION_2_0_functions },
+
+   { "GL_EXT_blend_color",		GL_EXT_blend_color_functions },
+   { "GL_EXT_blend_minmax",		GL_EXT_blend_minmax_functions },
+   { "GL_EXT_convolution",		GL_EXT_convolution_functions },
+   { "GL_EXT_histogram",		GL_EXT_histogram_functions },
+   { "GL_SGI_color_table",		GL_SGI_color_table_functions },
+
+   { "GL_ARB_shader_objects",		GL_ARB_shader_objects_functions },
+   { "GL_ARB_vertex_program",		GL_ARB_vertex_program_functions },
+   { "GL_APPLE_vertex_array_object",	GL_APPLE_vertex_array_object_functions },
+   { "GL_ATI_fragment_shader",		GL_ATI_fragment_shader_functions },
+   { "GL_EXT_depth_bounds_test",	GL_EXT_depth_bounds_test_functions },
+   { "GL_EXT_framebuffer_object",	GL_EXT_framebuffer_object_functions },
+   { "GL_EXT_framebuffer_blit",		GL_EXT_framebuffer_blit_functions },
+   { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
+   { "GL_EXT_paletted_texture",		GL_EXT_paletted_texture_functions },
+   { "GL_IBM_multimode_draw_arrays",	GL_IBM_multimode_draw_arrays_functions },
+   { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
+   { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
+   { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
+   { NULL,				NULL }
+};
+#endif
+
 /*
  * Create a new X/Mesa visual.
  * Input:  display - X11 display
@@ -1384,6 +1409,14 @@ XMesaVisual XMesaCreateVisual( XMesaDisplay *display,
    XMesaVisual v;
    GLint red_bits, green_bits, blue_bits, alpha_bits;
 
+#ifdef IN_DRI_DRIVER
+   /* driInitExtensions() should be called once per screen to setup extension
+    * indices.  There is no need to call it when the context is created since
+    * XMesa enables mesa sw extensions on its own.
+    */
+   driInitExtensions( NULL, card_extensions, GL_FALSE );
+#endif
+
 #ifndef XFree86Server
    /* For debugging only */
    if (_mesa_getenv("MESA_XSYNC")) {
@@ -1530,9 +1563,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
    XMesaContext c;
    GLcontext *mesaCtx;
    struct dd_function_table functions;
-#if 0
    TNLcontext *tnl;
-#endif
 
    if (firstTime) {
       _glthread_INIT_MUTEX(_xmesa_lock);
@@ -1549,15 +1580,6 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
    /* initialize with default driver functions, then plug in XMesa funcs */
    _mesa_init_driver_functions(&functions);
    xmesa_init_driver_functions(v, &functions);
-   st_init_driver_functions(&functions);
-
-   /* override st's function */
-   functions.UpdateState = xmesa_update_state;
-
-   /*
-   functions.NewRenderbuffer = xmesa_new_renderbuffer;
-   */
-
    if (!_mesa_initialize_context(mesaCtx, &v->mesa_visual,
                       share_list ? &(share_list->mesa) : (GLcontext *) NULL,
                       &functions, (void *) c)) {
@@ -1570,8 +1592,9 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
    _mesa_enable_1_4_extensions(mesaCtx);
    _mesa_enable_1_5_extensions(mesaCtx);
    _mesa_enable_2_0_extensions(mesaCtx);
+   _mesa_enable_2_1_extensions(mesaCtx);
 #if ENABLE_EXT_texure_compression_s3tc
-    if (c->Mesa_DXTn) {
+    if (mesaCtx->Mesa_DXTn) {
        _mesa_enable_extension(mesaCtx, "GL_EXT_texture_compression_s3tc");
        _mesa_enable_extension(mesaCtx, "GL_S3_s3tc");
     }
@@ -1597,49 +1620,22 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
 
    /* Initialize the software rasterizer and helper modules.
     */
-   if (!_swrast_CreateContext( mesaCtx )
-#if 0
-       || !_vbo_CreateContext( mesaCtx ) ||
+   if (!_swrast_CreateContext( mesaCtx ) ||
+       !_vbo_CreateContext( mesaCtx ) ||
        !_tnl_CreateContext( mesaCtx ) ||
-       !_swsetup_CreateContext( mesaCtx )
-#endif
-       ) {
+       !_swsetup_CreateContext( mesaCtx )) {
       _mesa_free_context_data(&c->mesa);
       _mesa_free(c);
       return NULL;
    }
 
-#if 0
    /* tnl setup */
    tnl = TNL_CONTEXT(mesaCtx);
    tnl->Driver.RunPipeline = _tnl_run_pipeline;
-#endif
-
    /* swrast setup */
    xmesa_register_swrast_functions( mesaCtx );
-
-
-   st_create_context( mesaCtx,
-                      xmesa_create_softpipe( c ) );
-
-   _swsetup_CreateContext( mesaCtx );
    _swsetup_Wakeup(mesaCtx);
 
-   /* override these functions, as if the xlib driver were derived from
-    * the softpipe driver.
-    */
-#if 0
-   mesaCtx->st->pipe->surface_alloc = xmesa_surface_alloc;
-#endif
-   mesaCtx->st->pipe->is_format_supported = xmesa_is_format_supported;
-   mesaCtx->st->pipe->get_tile_rgba = xmesa_get_tile_rgba;
-   mesaCtx->st->pipe->put_tile_rgba = xmesa_put_tile_rgba;
-
-   mesaCtx->st->haveFramebufferRegions = GL_FALSE;
-
-   /* special pipe->clear function */
-   mesaCtx->st->pipe->clear = xmesa_clear;
-
    return c;
 }
 
@@ -1656,10 +1652,8 @@ void XMesaDestroyContext( XMesaContext c )
 
    _swsetup_DestroyContext( mesaCtx );
    _swrast_DestroyContext( mesaCtx );
-#if 0
    _tnl_DestroyContext( mesaCtx );
    _vbo_DestroyContext( mesaCtx );
-#endif
    _mesa_free_context_data( mesaCtx );
    _mesa_free( c );
 }
diff --git a/src/mesa/drivers/x11/xm_buffer.c b/src/mesa/drivers/x11/xm_buffer.c
index 09356c7329f..f104d44d051 100644
--- a/src/mesa/drivers/x11/xm_buffer.c
+++ b/src/mesa/drivers/x11/xm_buffer.c
@@ -32,13 +32,9 @@
 #include "glxheader.h"
 #include "GL/xmesa.h"
 #include "xmesaP.h"
-#include "imports.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "pipe/p_state.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_winsys.h"
-#include "state_tracker/st_context.h"
+#include "main/imports.h"
+#include "main/framebuffer.h"
+#include "main/renderbuffer.h"
 
 
 #if defined(USE_XSHM) && !defined(XFree86Server)
@@ -249,18 +245,6 @@ xmesa_delete_renderbuffer(struct gl_renderbuffer *rb)
 }
 
 
-static void
-finish_surface_init(GLcontext *ctx, struct xmesa_renderbuffer *xrb)
-{
-   struct pipe_context *pipe = ctx->st->pipe;
-   if (!xrb->St.surface->region) {
-      int w = 1, h = 1;
-      xrb->St.surface->region = pipe->winsys->region_alloc(pipe->winsys,
-                                                           1, w, h, 0x0);
-   }
-}
-
-
 /**
  * Reallocate renderbuffer storage for front color buffer.
  * Called via gl_renderbuffer::AllocStorage()
@@ -284,12 +268,6 @@ xmesa_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
    rb->Height = height;
    rb->InternalFormat = internalFormat;
 
-   if (!xrb->St.surface || !xrb->St.surface->region)
-      finish_surface_init(ctx, xrb);
-
-   xrb->St.surface->width = width;
-   xrb->St.surface->height = height;
-
    return GL_TRUE;
 }
 
@@ -339,103 +317,46 @@ xmesa_alloc_back_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
       xrb->origin4 = NULL;
    }
 
-   if (!xrb->St.surface || !xrb->St.surface->region)
-      finish_surface_init(ctx, xrb);
-
-   xrb->St.surface->width = width;
-   xrb->St.surface->height = height;
-
    return GL_TRUE;
 }
 
 
-/**
- * Called to create the front/back color renderbuffers, not user-created
- * renderbuffers.
- */
 struct xmesa_renderbuffer *
-xmesa_create_renderbuffer(GLcontext *ctx, GLuint name, const GLvisual *visual,
-                          GLboolean backBuffer)
+xmesa_new_renderbuffer(GLcontext *ctx, GLuint name, const GLvisual *visual,
+                       GLboolean backBuffer)
 {
    struct xmesa_renderbuffer *xrb = CALLOC_STRUCT(xmesa_renderbuffer);
-   struct pipe_context *pipe = NULL;/*ctx->st->pipe;*/
    if (xrb) {
       GLuint name = 0;
-      GLuint pipeFormat = 0;
-      struct xmesa_surface *xms;
-
-      _mesa_init_renderbuffer(&xrb->St.Base, name);
+      _mesa_init_renderbuffer(&xrb->Base, name);
 
-      xrb->St.Base.Delete = xmesa_delete_renderbuffer;
+      xrb->Base.Delete = xmesa_delete_renderbuffer;
       if (backBuffer)
-         xrb->St.Base.AllocStorage = xmesa_alloc_back_storage;
+         xrb->Base.AllocStorage = xmesa_alloc_back_storage;
       else
-         xrb->St.Base.AllocStorage = xmesa_alloc_front_storage;
+         xrb->Base.AllocStorage = xmesa_alloc_front_storage;
 
       if (visual->rgbMode) {
-         xrb->St.Base.InternalFormat = GL_RGBA;
-         xrb->St.Base._BaseFormat = GL_RGBA;
-         xrb->St.Base.DataType = GL_UNSIGNED_BYTE;
-         xrb->St.Base.RedBits = visual->redBits;
-         xrb->St.Base.GreenBits = visual->greenBits;
-         xrb->St.Base.BlueBits = visual->blueBits;
-         xrb->St.Base.AlphaBits = visual->alphaBits;
-         pipeFormat = PIPE_FORMAT_U_A8_R8_G8_B8;
+         xrb->Base.InternalFormat = GL_RGBA;
+         xrb->Base._BaseFormat = GL_RGBA;
+         xrb->Base.DataType = GL_UNSIGNED_BYTE;
+         xrb->Base.RedBits = visual->redBits;
+         xrb->Base.GreenBits = visual->greenBits;
+         xrb->Base.BlueBits = visual->blueBits;
+         xrb->Base.AlphaBits = visual->alphaBits;
       }
       else {
-         xrb->St.Base.InternalFormat = GL_COLOR_INDEX;
-         xrb->St.Base._BaseFormat = GL_COLOR_INDEX;
-         xrb->St.Base.DataType = GL_UNSIGNED_INT;
-         xrb->St.Base.IndexBits = visual->indexBits;
+         xrb->Base.InternalFormat = GL_COLOR_INDEX;
+         xrb->Base._BaseFormat = GL_COLOR_INDEX;
+         xrb->Base.DataType = GL_UNSIGNED_INT;
+         xrb->Base.IndexBits = visual->indexBits;
       }
       /* only need to set Red/Green/EtcBits fields for user-created RBs */
-
-      xrb->St.surface = xmesa_new_color_surface(pipe, pipeFormat);
-      xms = (struct xmesa_surface *) xrb->St.surface;
-      xms->xrb = xrb;
    }
    return xrb;
 }
 
 
-#if 0
-struct gl_renderbuffer *
-xmesa_new_renderbuffer(GLcontext *ctx, struct gl_renderbuffer *rb,
-                       GLenum internalFormat, GLuint width, GLuint height)
-{
-   struct xmesa_renderbuffer *xrb = CALLOC_STRUCT(xmesa_renderbuffer);
-   if (xrb) {
-      GLuint name = 0;
-      _mesa_init_renderbuffer(&xrb->St.Base, name);
-
-      xrb->St.Base.Delete = xmesa_delete_renderbuffer;
-      if (backBuffer)
-         xrb->St.Base.AllocStorage = xmesa_alloc_back_storage;
-      else
-         xrb->St.Base.AllocStorage = xmesa_alloc_front_storage;
-
-      if (visual->rgbMode) {
-         xrb->St.Base.InternalFormat = GL_RGBA;
-         xrb->St.Base._BaseFormat = GL_RGBA;
-         xrb->St.Base.DataType = GL_UNSIGNED_BYTE;
-         xrb->St.Base.RedBits = visual->redBits;
-         xrb->St.Base.GreenBits = visual->greenBits;
-         xrb->St.Base.BlueBits = visual->blueBits;
-         xrb->St.Base.AlphaBits = visual->alphaBits;
-      }
-      else {
-         xrb->St.Base.InternalFormat = GL_COLOR_INDEX;
-         xrb->St.Base._BaseFormat = GL_COLOR_INDEX;
-         xrb->St.Base.DataType = GL_UNSIGNED_INT;
-         xrb->St.Base.IndexBits = visual->indexBits;
-      }
-      /* only need to set Red/Green/EtcBits fields for user-created RBs */
-   }
-   return xrb;
-}
-#endif
-
-
 /**
  * Called via gl_framebuffer::Delete() method when this buffer
  * is _really_ being deleted.
diff --git a/src/mesa/drivers/x11/xm_dd.c b/src/mesa/drivers/x11/xm_dd.c
index 34287effe1c..305df548fa2 100644
--- a/src/mesa/drivers/x11/xm_dd.c
+++ b/src/mesa/drivers/x11/xm_dd.c
@@ -29,34 +29,30 @@
  */
 
 #include "glxheader.h"
-#include "bufferobj.h"
-#include "buffers.h"
-#include "context.h"
-#include "colormac.h"
-#include "depth.h"
-#include "drawpix.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "macros.h"
-#include "image.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "state.h"
-#include "texobj.h"
-#include "teximage.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "xmesaP.h"
+#include "main/bufferobj.h"
+#include "main/buffers.h"
+#include "main/context.h"
+#include "main/colormac.h"
+#include "main/depth.h"
+#include "main/drawpix.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/macros.h"
+#include "main/image.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/state.h"
+#include "main/texobj.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
+#include "main/texformat.h"
 #include "swrast/swrast.h"
 #include "swrast/s_context.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
+#include "xmesaP.h"
 
-#include "softpipe/sp_context.h"
-#include "state_tracker/st_public.h"
-#include "state_tracker/st_context.h"
-#include "state_tracker/st_draw.h"
 
 
 /*
@@ -222,7 +218,7 @@ clear_pixmap(GLcontext *ctx, struct xmesa_renderbuffer *xrb,
    assert(xmbuf->cleargc);
 
    XMesaFillRectangle( xmesa->display, xrb->pixmap, xmbuf->cleargc,
-                       x, xrb->St.Base.Height - y - height,
+                       x, xrb->Base.Height - y - height,
                        width, height );
 }
 
@@ -333,9 +329,9 @@ clear_32bit_ximage(GLcontext *ctx, struct xmesa_renderbuffer *xrb,
             | ((pixel << 24) & 0xff000000);
    }
 
-   if (width == xrb->St.Base.Width && height == xrb->St.Base.Height) {
+   if (width == xrb->Base.Width && height == xrb->Base.Height) {
       /* clearing whole buffer */
-      const GLuint n = xrb->St.Base.Width * xrb->St.Base.Height;
+      const GLuint n = xrb->Base.Width * xrb->Base.Height;
       GLuint *ptr4 = (GLuint *) xrb->ximage->data;
       if (pixel == 0) {
          /* common case */
@@ -379,8 +375,8 @@ clear_nbit_ximage(GLcontext *ctx, struct xmesa_renderbuffer *xrb,
 
 
 
-void
-xmesa_clear_buffers(GLcontext *ctx, GLbitfield buffers)
+static void
+clear_buffers(GLcontext *ctx, GLbitfield buffers)
 {
    if (ctx->DrawBuffer->Name == 0) {
       /* this is a window system framebuffer */
@@ -424,7 +420,42 @@ xmesa_clear_buffers(GLcontext *ctx, GLbitfield buffers)
 
 
 #ifndef XFree86Server
-/* XXX this was never tested in the Xserver environment */
+/* XXX these functions haven't been tested in the Xserver environment */
+
+
+/**
+ * Check if we can do an optimized glDrawPixels into an 8R8G8B visual.
+ */
+static GLboolean
+can_do_DrawPixels_8R8G8B(GLcontext *ctx, GLenum format, GLenum type)
+{
+   if (format == GL_BGRA &&
+       type == GL_UNSIGNED_BYTE &&
+       ctx->DrawBuffer &&
+       ctx->DrawBuffer->Name == 0 &&
+       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
+       ctx->Pixel.ZoomY == 1.0 &&
+       ctx->_ImageTransferState == 0 /* no color tables, scale/bias, etc */) {
+      const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+
+      if (swrast->NewState)
+         _swrast_validate_derived( ctx );
+      
+      if ((swrast->_RasterMask & ~CLIP_BIT) == 0) /* no blend, z-test, etc */ {
+         struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
+         if (rb) {
+            struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb->Wrapped);
+            if (xrb &&
+                xrb->pixmap && /* drawing to pixmap or window */
+                xrb->Base.AlphaBits == 0) {
+               return GL_TRUE;
+            }
+         }
+      }
+   }
+   return GL_FALSE;
+}
+
 
 /**
  * This function implements glDrawPixels() with an XPutImage call when
@@ -438,37 +469,16 @@ xmesa_DrawPixels_8R8G8B( GLcontext *ctx,
                          const struct gl_pixelstore_attrib *unpack,
                          const GLvoid *pixels )
 {
-   const SWcontext *swrast = SWRAST_CONTEXT( ctx );
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
-   struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb->Wrapped);
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   if (ctx->DrawBuffer->Name == 0 &&
-       format == GL_BGRA &&
-       type == GL_UNSIGNED_BYTE &&
-       (swrast->_RasterMask & ~CLIP_BIT) == 0 && /* no blend, z-test, etc */
-       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
-       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
-       ctx->Pixel.ZoomY == 1.0 &&
-       xrb->pixmap &&                    /* drawing to pixmap or window */
-       xrb->St.Base.AlphaBits == 0)
-   {
-      const XMesaContext xmesa = XMESA_CONTEXT(ctx);
-      XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
-      XMesaDisplay *dpy = xmesa->xm_visual->display;
-      const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
+   if (can_do_DrawPixels_8R8G8B(ctx, format, type)) {
+      const SWcontext *swrast = SWRAST_CONTEXT( ctx );
+      struct gl_pixelstore_attrib clippedUnpack = *unpack;
       int dstX = x;
       int dstY = y;
       int w = width;
       int h = height;
-      struct gl_pixelstore_attrib clippedUnpack = *unpack;
 
-      ASSERT(xmesa->xm_visual->dithered_pf == PF_8R8G8B);
-      ASSERT(xmesa->xm_visual->undithered_pf == PF_8R8G8B);
-      ASSERT(dpy);
-      ASSERT(gc);
+      if (swrast->NewState)
+         _swrast_validate_derived( ctx );
 
       if (unpack->BufferObj->Name) {
          /* unpack from PBO */
@@ -493,14 +503,26 @@ xmesa_DrawPixels_8R8G8B( GLcontext *ctx,
       }
 
       if (_mesa_clip_drawpixels(ctx, &dstX, &dstY, &w, &h, &clippedUnpack)) {
+         const XMesaContext xmesa = XMESA_CONTEXT(ctx);
+         XMesaDisplay *dpy = xmesa->xm_visual->display;
+         XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
+         const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
+         struct xmesa_renderbuffer *xrb
+            = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
+         const int srcX = clippedUnpack.SkipPixels;
+         const int srcY = clippedUnpack.SkipRows;
+         const int rowLength = clippedUnpack.RowLength;
+         XMesaImage ximage;
+
+         ASSERT(xmesa->xm_visual->dithered_pf == PF_8R8G8B);
+         ASSERT(xmesa->xm_visual->undithered_pf == PF_8R8G8B);
+         ASSERT(dpy);
+         ASSERT(gc);
+
          /* This is a little tricky since all coordinates up to now have
           * been in the OpenGL bottom-to-top orientation.  X is top-to-bottom
           * so we have to carefully compute the Y coordinates/addresses here.
           */
-         int srcX = clippedUnpack.SkipPixels;
-         int srcY = clippedUnpack.SkipRows;
-         int rowLength = clippedUnpack.RowLength;
-         XMesaImage ximage;
          MEMSET(&ximage, 0, sizeof(XMesaImage));
          ximage.width = width;
          ximage.height = height;
@@ -511,9 +533,9 @@ xmesa_DrawPixels_8R8G8B( GLcontext *ctx,
          ximage.bitmap_unit = 32;
          ximage.bitmap_bit_order = LSBFirst;
          ximage.bitmap_pad = 32;
-         ximage.depth = 24;
-         ximage.bytes_per_line = -rowLength * 4; /* negative to flip image */
+         ximage.depth = 32;
          ximage.bits_per_pixel = 32;
+         ximage.bytes_per_line = -rowLength * 4; /* negative to flip image */
          /* it seems we don't need to set the ximage.red/green/blue_mask fields */
          /* flip Y axis for dest position */
          dstY = YFLIP(xrb, dstY) - h + 1;
@@ -535,6 +557,41 @@ xmesa_DrawPixels_8R8G8B( GLcontext *ctx,
 
 
 /**
+ * Check if we can do an optimized glDrawPixels into an 5R6G5B visual.
+ */
+static GLboolean
+can_do_DrawPixels_5R6G5B(GLcontext *ctx, GLenum format, GLenum type)
+{
+   if (format == GL_RGB &&
+       type == GL_UNSIGNED_SHORT_5_6_5 &&
+       !ctx->Color.DitherFlag &&  /* no dithering */
+       ctx->DrawBuffer &&
+       ctx->DrawBuffer->Name == 0 &&
+       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
+       ctx->Pixel.ZoomY == 1.0 &&
+       ctx->_ImageTransferState == 0 /* no color tables, scale/bias, etc */) {
+      const SWcontext *swrast = SWRAST_CONTEXT(ctx);
+
+      if (swrast->NewState)
+         _swrast_validate_derived( ctx );
+      
+      if ((swrast->_RasterMask & ~CLIP_BIT) == 0) /* no blend, z-test, etc */ {
+         struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
+         if (rb) {
+            struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb->Wrapped);
+            if (xrb &&
+                xrb->pixmap && /* drawing to pixmap or window */
+                xrb->Base.AlphaBits == 0) {
+               return GL_TRUE;
+            }
+         }
+      }
+   }
+   return GL_FALSE;
+}
+
+
+/**
  * This function implements glDrawPixels() with an XPutImage call when
  * drawing to the front buffer (X Window drawable).  The image format
  * must be GL_RGB and image type must be GL_UNSIGNED_SHORT_5_6_5 to
@@ -547,35 +604,17 @@ xmesa_DrawPixels_5R6G5B( GLcontext *ctx,
                          const struct gl_pixelstore_attrib *unpack,
                          const GLvoid *pixels )
 {
-   struct xmesa_renderbuffer *xrb
-      = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped);
-   const XMesaContext xmesa = XMESA_CONTEXT(ctx);
-   const SWcontext *swrast = SWRAST_CONTEXT( ctx );
-   XMesaDisplay *dpy = xmesa->xm_visual->display;
-   XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
-   const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
-
-   ASSERT(dpy);
-   ASSERT(gc);
-   ASSERT(xmesa->xm_visual->undithered_pf == PF_5R6G5B);
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   if (xrb->pixmap &&       /* drawing to pixmap or window */
-       format == GL_RGB &&
-       type == GL_UNSIGNED_SHORT_5_6_5 &&
-       !ctx->Color.DitherFlag &&  /* no dithering */
-       (swrast->_RasterMask & ~CLIP_BIT) == 0 && /* no blend, z-test, etc */
-       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
-       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
-       ctx->Pixel.ZoomY == 1.0) {
+   if (can_do_DrawPixels_5R6G5B(ctx, format, type)) {
+      const SWcontext *swrast = SWRAST_CONTEXT( ctx );
+      struct gl_pixelstore_attrib clippedUnpack = *unpack;
       int dstX = x;
       int dstY = y;
       int w = width;
       int h = height;
-      struct gl_pixelstore_attrib clippedUnpack = *unpack;
 
+      if (swrast->NewState)
+         _swrast_validate_derived( ctx );
+      
       if (unpack->BufferObj->Name) {
          /* unpack from PBO */
          GLubyte *buf;
@@ -599,14 +638,25 @@ xmesa_DrawPixels_5R6G5B( GLcontext *ctx,
       }
 
       if (_mesa_clip_drawpixels(ctx, &dstX, &dstY, &w, &h, &clippedUnpack)) {
+         const XMesaContext xmesa = XMESA_CONTEXT(ctx);
+         XMesaDisplay *dpy = xmesa->xm_visual->display;
+         XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
+         const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
+         struct xmesa_renderbuffer *xrb
+            = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
+         const int srcX = clippedUnpack.SkipPixels;
+         const int srcY = clippedUnpack.SkipRows;
+         const int rowLength = clippedUnpack.RowLength;
+         XMesaImage ximage;
+
+         ASSERT(xmesa->xm_visual->undithered_pf == PF_5R6G5B);
+         ASSERT(dpy);
+         ASSERT(gc);
+
          /* This is a little tricky since all coordinates up to now have
           * been in the OpenGL bottom-to-top orientation.  X is top-to-bottom
           * so we have to carefully compute the Y coordinates/addresses here.
           */
-         int srcX = clippedUnpack.SkipPixels;
-         int srcY = clippedUnpack.SkipRows;
-         int rowLength = clippedUnpack.RowLength;
-         XMesaImage ximage;
          MEMSET(&ximage, 0, sizeof(XMesaImage));
          ximage.width = width;
          ximage.height = height;
@@ -618,8 +668,8 @@ xmesa_DrawPixels_5R6G5B( GLcontext *ctx,
          ximage.bitmap_bit_order = LSBFirst;
          ximage.bitmap_pad = 16;
          ximage.depth = 16;
-         ximage.bytes_per_line = -rowLength * 2; /* negative to flip image */
          ximage.bits_per_pixel = 16;
+         ximage.bytes_per_line = -rowLength * 2; /* negative to flip image */
          /* it seems we don't need to set the ximage.red/green/blue_mask fields */
          /* flip Y axis for dest position */
          dstY = YFLIP(xrb, dstY) - h + 1;
@@ -639,6 +689,42 @@ xmesa_DrawPixels_5R6G5B( GLcontext *ctx,
 }
 
 
+/**
+ * Determine if we can do an optimized glCopyPixels.
+ */
+static GLboolean
+can_do_CopyPixels(GLcontext *ctx, GLenum type)
+{
+   if (type == GL_COLOR &&
+       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
+       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
+       ctx->Pixel.ZoomY == 1.0 &&
+       ctx->Color.DrawBuffer[0] == GL_FRONT &&  /* copy to front buf */
+       ctx->Pixel.ReadBuffer == GL_FRONT &&    /* copy from front buf */
+       ctx->ReadBuffer->_ColorReadBuffer &&
+       ctx->DrawBuffer->_ColorDrawBuffers[0]) {
+      const SWcontext *swrast = SWRAST_CONTEXT( ctx );
+
+      if (swrast->NewState)
+         _swrast_validate_derived( ctx );
+
+      if ((swrast->_RasterMask & ~CLIP_BIT) == 0x0 &&
+          ctx->ReadBuffer &&
+          ctx->ReadBuffer->_ColorReadBuffer &&
+          ctx->DrawBuffer &&
+          ctx->DrawBuffer->_ColorDrawBuffers[0]) {
+         struct xmesa_renderbuffer *srcXrb
+            = xmesa_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer->Wrapped);
+         struct xmesa_renderbuffer *dstXrb
+            = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
+         if (srcXrb->pixmap && dstXrb->pixmap) {
+            return GL_TRUE;
+         }
+      }
+   }
+   return GL_FALSE;
+}
+
 
 /**
  * Implement glCopyPixels for the front color buffer (or back buffer Pixmap)
@@ -650,31 +736,19 @@ xmesa_CopyPixels( GLcontext *ctx,
                   GLint srcx, GLint srcy, GLsizei width, GLsizei height,
                   GLint destx, GLint desty, GLenum type )
 {
-   const XMesaContext xmesa = XMESA_CONTEXT(ctx);
-   const SWcontext *swrast = SWRAST_CONTEXT( ctx );
-   XMesaDisplay *dpy = xmesa->xm_visual->display;
-   XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
-   const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
-   struct xmesa_renderbuffer *srcXrb
-      = xmesa_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer->Wrapped);
-   struct xmesa_renderbuffer *dstXrb
-      = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped);
-
-   ASSERT(dpy);
-   ASSERT(gc);
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   if (ctx->Color.DrawBuffer[0] == GL_FRONT &&
-       ctx->Pixel.ReadBuffer == GL_FRONT &&
-       srcXrb->pixmap &&
-       dstXrb->pixmap &&
-       type == GL_COLOR &&
-       (swrast->_RasterMask & ~CLIP_BIT) == 0 && /* no blend, z-test, etc */
-       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
-       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
-       ctx->Pixel.ZoomY == 1.0) {
+   if (can_do_CopyPixels(ctx, type)) {
+      const XMesaContext xmesa = XMESA_CONTEXT(ctx);
+      XMesaDisplay *dpy = xmesa->xm_visual->display;
+      XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
+      const XMesaGC gc = xmbuf->cleargc;  /* effected by glColorMask */
+      struct xmesa_renderbuffer *srcXrb
+         = xmesa_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer->Wrapped);
+      struct xmesa_renderbuffer *dstXrb
+         = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
+
+      ASSERT(dpy);
+      ASSERT(gc);
+
       /* Note: we don't do any special clipping work here.  We could,
        * but X will do it for us.
        */
@@ -687,6 +761,7 @@ xmesa_CopyPixels( GLcontext *ctx,
       _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type );
    }
 }
+
 #endif /* XFree86Server */
 
 
@@ -704,7 +779,7 @@ get_string( GLcontext *ctx, GLenum name )
 #ifdef XFree86Server
          return (const GLubyte *) "Mesa GLX Indirect";
 #else
-         return (const GLubyte *) "Mesa X11 (softpipe)";
+         return (const GLubyte *) "Mesa X11";
 #endif
       case GL_VENDOR:
 #ifdef XFree86Server
@@ -831,9 +906,6 @@ xmesa_update_state( GLcontext *ctx, GLbitfield new_state )
    _vbo_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
 
-   st_invalidate_state( ctx, new_state );
-
-
    if (ctx->DrawBuffer->Name != 0)
       return;
 
@@ -841,7 +913,7 @@ xmesa_update_state( GLcontext *ctx, GLbitfield new_state )
     * GL_DITHER, GL_READ/DRAW_BUFFER, buffer binding state, etc. effect
     * renderbuffer span/clear funcs.
     */
-   if (new_state & (_NEW_COLOR | _NEW_BUFFERS)) {
+   if (new_state & (_NEW_COLOR | _NEW_PIXEL | _NEW_BUFFERS)) {
       XMesaBuffer xmbuf = XMESA_BUFFER(ctx->DrawBuffer);
       struct xmesa_renderbuffer *front_xrb, *back_xrb;
 
@@ -1074,12 +1146,13 @@ xmesa_init_driver_functions( XMesaVisual xmvisual,
    driver->IndexMask = index_mask;
    driver->ColorMask = color_mask;
    driver->Enable = enable;
-   driver->Clear = xmesa_clear_buffers;
+   driver->Clear = clear_buffers;
    driver->Viewport = xmesa_viewport;
 #ifndef XFree86Server
    driver->CopyPixels = xmesa_CopyPixels;
    if (xmvisual->undithered_pf == PF_8R8G8B &&
-       xmvisual->dithered_pf == PF_8R8G8B) {
+       xmvisual->dithered_pf == PF_8R8G8B &&
+       xmvisual->BitsPerPixel == 32) {
       driver->DrawPixels = xmesa_DrawPixels_8R8G8B;
    }
    else if (xmvisual->undithered_pf == PF_5R6G5B) {
@@ -1098,7 +1171,6 @@ xmesa_init_driver_functions( XMesaVisual xmvisual,
    driver->BeginQuery = xmesa_begin_query;
    driver->EndQuery = xmesa_end_query;
 #endif
-
 }
 
 
diff --git a/src/mesa/drivers/x11/xm_glide.c b/src/mesa/drivers/x11/xm_glide.c
index ae4f4282dbb..cbd69b011a1 100644
--- a/src/mesa/drivers/x11/xm_glide.c
+++ b/src/mesa/drivers/x11/xm_glide.c
@@ -64,7 +64,8 @@ FXcreateContext(XMesaVisual v, XMesaWindow w, XMesaContext c, XMesaBuffer b)
        attribs[numAttribs++] = FXMESA_NONE;
 
        /* [dBorca] we should take an envvar for `fxMesaSelectCurrentBoard'!!! */
-       hw = fxMesaSelectCurrentBoard(0);
+/*       hw = fxMesaSelectCurrentBoard(0); */
+       hw = GR_SSTTYPE_Voodoo2;
 
        /* if these fail, there's a new bug somewhere */
        ASSERT(b->mesa_buffer.Width > 0);
diff --git a/src/mesa/drivers/x11/xm_line.c b/src/mesa/drivers/x11/xm_line.c
index deeae5019c4..f643b6d3a76 100644
--- a/src/mesa/drivers/x11/xm_line.c
+++ b/src/mesa/drivers/x11/xm_line.c
@@ -31,9 +31,9 @@
 
 
 #include "glxheader.h"
-#include "depth.h"
-#include "macros.h"
-#include "mtypes.h"
+#include "main/depth.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 #include "xmesaP.h"
 
 /* Internal swrast includes:
@@ -121,7 +121,7 @@ void xmesa_choose_point( GLcontext *ctx )
 
 
 #define GET_XRB(XRB)  struct xmesa_renderbuffer *XRB = \
-   xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped)
+   xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped)
 
 
 /*
@@ -587,8 +587,8 @@ get_line_func(GLcontext *ctx)
    const int depth = GET_VISUAL_DEPTH(xmesa->xm_visual);
    const struct xmesa_renderbuffer *xrb;
 
-   if ((ctx->DrawBuffer->_ColorDrawBufferMask[0]
-        & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT)) == 0)
+   if ((ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_FRONT_LEFT) &&
+       (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_BACK_LEFT))
       return (swrast_line_func) NULL;
    if (ctx->RenderMode != GL_RENDER)      return (swrast_line_func) NULL;
    if (ctx->Line.SmoothFlag)              return (swrast_line_func) NULL;
@@ -598,7 +598,7 @@ get_line_func(GLcontext *ctx)
    if (swrast->_RasterMask & MULTI_DRAW_BIT) return (swrast_line_func) NULL;
    if (xmbuf->swAlpha)                    return (swrast_line_func) NULL;
 
-   xrb = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped);
+   xrb = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
 
    if (xrb->ximage
        && swrast->_RasterMask==DEPTH_BIT
@@ -661,8 +661,8 @@ get_line_func(GLcontext *ctx)
    }
 
 #ifndef XFree86Server
-   if (ctx->DrawBuffer->_NumColorDrawBuffers[0] == 1
-       && ctx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT
+   if (ctx->DrawBuffer->_NumColorDrawBuffers == 1
+       && ctx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT
        && swrast->_RasterMask == LOGIC_OP_BIT
        && ctx->Color.LogicOp == GL_XOR
        && !ctx->Line.StippleFlag
diff --git a/src/mesa/drivers/x11/xm_span.c b/src/mesa/drivers/x11/xm_span.c
index ce54a18a273..57b57494483 100644
--- a/src/mesa/drivers/x11/xm_span.c
+++ b/src/mesa/drivers/x11/xm_span.c
@@ -23,15 +23,15 @@
  */
 
 #include "glxheader.h"
-#include "colormac.h"
-#include "context.h"
-#include "depth.h"
-#include "drawpix.h"
-#include "extensions.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "state.h"
+#include "main/colormac.h"
+#include "main/context.h"
+#include "main/depth.h"
+#include "main/drawpix.h"
+#include "main/extensions.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "main/state.h"
 #include "xmesaP.h"
 
 #include "swrast/swrast.h"
@@ -1303,17 +1303,6 @@ static void put_row_rgb_TRUEDITHER_ximage( RGB_SPAN_ARGS )
 }
 
 
-
-static void *get_pointer_4_ximage( GLcontext *ctx, 
-				   struct gl_renderbuffer *rb, 
-				   GLint x, GLint y )
-{
-   GET_XRB(xrb);
-   return PIXEL_ADDR4(xrb, x, y);
-}
-
-
-
 /*
  * Write a span of PF_8A8B8G8R-format pixels to an ximage.
  */
@@ -4539,260 +4528,257 @@ xmesa_set_renderbuffer_funcs(struct xmesa_renderbuffer *xrb,
                              enum pixel_format pixelformat, GLint depth)
 {
    const GLboolean pixmap = xrb->pixmap ? GL_TRUE : GL_FALSE;
-   struct gl_renderbuffer *rb  = &xrb->St.Base;
 
    switch (pixelformat) {
    case PF_Index:
-      ASSERT(rb->DataType == GL_UNSIGNED_INT);
+      ASSERT(xrb->Base.DataType == GL_UNSIGNED_INT);
       if (pixmap) {
-         rb->PutRow        = put_row_ci_pixmap;
-         rb->PutRowRGB     = NULL;
-         rb->PutMonoRow    = put_mono_row_ci_pixmap;
-         rb->PutValues     = put_values_ci_pixmap;
-         rb->PutMonoValues = put_mono_values_ci_pixmap;
+         xrb->Base.PutRow        = put_row_ci_pixmap;
+         xrb->Base.PutRowRGB     = NULL;
+         xrb->Base.PutMonoRow    = put_mono_row_ci_pixmap;
+         xrb->Base.PutValues     = put_values_ci_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_ci_pixmap;
       }
       else {
-         rb->PutRow        = put_row_ci_ximage;
-         rb->PutRowRGB     = NULL;
-         rb->PutMonoRow    = put_mono_row_ci_ximage;
-         rb->PutValues     = put_values_ci_ximage;
-         rb->PutMonoValues = put_mono_values_ci_ximage;
+         xrb->Base.PutRow        = put_row_ci_ximage;
+         xrb->Base.PutRowRGB     = NULL;
+         xrb->Base.PutMonoRow    = put_mono_row_ci_ximage;
+         xrb->Base.PutValues     = put_values_ci_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_ci_ximage;
       }
       break;
    case PF_Truecolor:
       if (pixmap) {
-         rb->PutRow        = put_row_TRUECOLOR_pixmap;
-         rb->PutRowRGB     = put_row_rgb_TRUECOLOR_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_TRUECOLOR_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_TRUECOLOR_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_TRUECOLOR_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_TRUECOLOR_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_TRUECOLOR_ximage;
-         rb->PutRowRGB     = put_row_rgb_TRUECOLOR_ximage;
-         rb->PutMonoRow    = put_mono_row_ximage;
-         rb->PutValues     = put_values_TRUECOLOR_ximage;
-         rb->PutMonoValues = put_mono_values_ximage;
+         xrb->Base.PutRow        = put_row_TRUECOLOR_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_TRUECOLOR_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_ximage;
+         xrb->Base.PutValues     = put_values_TRUECOLOR_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_ximage;
       }
       break;
    case PF_Dither_True:
       if (pixmap) {
-         rb->PutRow        = put_row_TRUEDITHER_pixmap;
-         rb->PutRowRGB     = put_row_rgb_TRUEDITHER_pixmap;
-         rb->PutMonoRow    = put_mono_row_TRUEDITHER_pixmap;
-         rb->PutValues     = put_values_TRUEDITHER_pixmap;
-         rb->PutMonoValues = put_mono_values_TRUEDITHER_pixmap;
+         xrb->Base.PutRow        = put_row_TRUEDITHER_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_TRUEDITHER_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_TRUEDITHER_pixmap;
+         xrb->Base.PutValues     = put_values_TRUEDITHER_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_TRUEDITHER_pixmap;
       }
       else {
-         rb->PutRow        = put_row_TRUEDITHER_ximage;
-         rb->PutRowRGB     = put_row_rgb_TRUEDITHER_ximage;
-         rb->PutMonoRow    = put_mono_row_TRUEDITHER_ximage;
-         rb->PutValues     = put_values_TRUEDITHER_ximage;
-         rb->PutMonoValues = put_mono_values_TRUEDITHER_ximage;
+         xrb->Base.PutRow        = put_row_TRUEDITHER_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_TRUEDITHER_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_TRUEDITHER_ximage;
+         xrb->Base.PutValues     = put_values_TRUEDITHER_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_TRUEDITHER_ximage;
       }
       break;
    case PF_8A8B8G8R:
       if (pixmap) {
-         rb->PutRow        = put_row_8A8B8G8R_pixmap;
-         rb->PutRowRGB     = put_row_rgb_8A8B8G8R_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_8A8B8G8R_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_8A8B8G8R_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_8A8B8G8R_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_8A8B8G8R_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_8A8B8G8R_ximage;
-         rb->PutRowRGB     = put_row_rgb_8A8B8G8R_ximage;
-         rb->PutMonoRow    = put_mono_row_8A8B8G8R_ximage;
-         rb->PutValues     = put_values_8A8B8G8R_ximage;
-         rb->PutMonoValues = put_mono_values_8A8B8G8R_ximage;
-	 rb->GetPointer    = get_pointer_4_ximage;
+         xrb->Base.PutRow        = put_row_8A8B8G8R_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_8A8B8G8R_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_8A8B8G8R_ximage;
+         xrb->Base.PutValues     = put_values_8A8B8G8R_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_8A8B8G8R_ximage;
       }
       break;
    case PF_8A8R8G8B:
       if (pixmap) {
-         rb->PutRow        = put_row_8A8R8G8B_pixmap;
-         rb->PutRowRGB     = put_row_rgb_8A8R8G8B_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_8A8R8G8B_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_8A8R8G8B_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_8A8R8G8B_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_8A8R8G8B_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_8A8R8G8B_ximage;
-         rb->PutRowRGB     = put_row_rgb_8A8R8G8B_ximage;
-         rb->PutMonoRow    = put_mono_row_8A8R8G8B_ximage;
-         rb->PutValues     = put_values_8A8R8G8B_ximage;
-         rb->PutMonoValues = put_mono_values_8A8R8G8B_ximage;
-	 rb->GetPointer    = get_pointer_4_ximage;
+         xrb->Base.PutRow        = put_row_8A8R8G8B_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_8A8R8G8B_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_8A8R8G8B_ximage;
+         xrb->Base.PutValues     = put_values_8A8R8G8B_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_8A8R8G8B_ximage;
       }
       break;
    case PF_8R8G8B:
       if (pixmap) {
-         rb->PutRow        = put_row_8R8G8B_pixmap;
-         rb->PutRowRGB     = put_row_rgb_8R8G8B_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_8R8G8B_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_8R8G8B_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_8R8G8B_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_8R8G8B_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_8R8G8B_ximage;
-         rb->PutRowRGB     = put_row_rgb_8R8G8B_ximage;
-         rb->PutMonoRow    = put_mono_row_8R8G8B_ximage;
-         rb->PutValues     = put_values_8R8G8B_ximage;
-         rb->PutMonoValues = put_mono_values_8R8G8B_ximage;
+         xrb->Base.PutRow        = put_row_8R8G8B_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_8R8G8B_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_8R8G8B_ximage;
+         xrb->Base.PutValues     = put_values_8R8G8B_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_8R8G8B_ximage;
       }
       break;
    case PF_8R8G8B24:
       if (pixmap) {
-         rb->PutRow        = put_row_8R8G8B24_pixmap;
-         rb->PutRowRGB     = put_row_rgb_8R8G8B24_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_8R8G8B24_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_8R8G8B24_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_8R8G8B24_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_8R8G8B24_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_8R8G8B24_ximage;
-         rb->PutRowRGB     = put_row_rgb_8R8G8B24_ximage;
-         rb->PutMonoRow    = put_mono_row_8R8G8B24_ximage;
-         rb->PutValues     = put_values_8R8G8B24_ximage;
-         rb->PutMonoValues = put_mono_values_8R8G8B24_ximage;
+         xrb->Base.PutRow        = put_row_8R8G8B24_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_8R8G8B24_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_8R8G8B24_ximage;
+         xrb->Base.PutValues     = put_values_8R8G8B24_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_8R8G8B24_ximage;
       }
       break;
    case PF_5R6G5B:
       if (pixmap) {
-         rb->PutRow        = put_row_5R6G5B_pixmap;
-         rb->PutRowRGB     = put_row_rgb_5R6G5B_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_5R6G5B_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_5R6G5B_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_5R6G5B_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_5R6G5B_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_5R6G5B_ximage;
-         rb->PutRowRGB     = put_row_rgb_5R6G5B_ximage;
-         rb->PutMonoRow    = put_mono_row_ximage;
-         rb->PutValues     = put_values_5R6G5B_ximage;
-         rb->PutMonoValues = put_mono_values_ximage;
+         xrb->Base.PutRow        = put_row_5R6G5B_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_5R6G5B_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_ximage;
+         xrb->Base.PutValues     = put_values_5R6G5B_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_ximage;
       }
       break;
    case PF_Dither_5R6G5B:
       if (pixmap) {
-         rb->PutRow        = put_row_DITHER_5R6G5B_pixmap;
-         rb->PutRowRGB     = put_row_rgb_DITHER_5R6G5B_pixmap;
-         rb->PutMonoRow    = put_mono_row_TRUEDITHER_pixmap;
-         rb->PutValues     = put_values_DITHER_5R6G5B_pixmap;
-         rb->PutMonoValues = put_mono_values_TRUEDITHER_pixmap;
+         xrb->Base.PutRow        = put_row_DITHER_5R6G5B_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_DITHER_5R6G5B_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_TRUEDITHER_pixmap;
+         xrb->Base.PutValues     = put_values_DITHER_5R6G5B_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_TRUEDITHER_pixmap;
       }
       else {
-         rb->PutRow        = put_row_DITHER_5R6G5B_ximage;
-         rb->PutRowRGB     = put_row_rgb_DITHER_5R6G5B_ximage;
-         rb->PutMonoRow    = put_mono_row_DITHER_5R6G5B_ximage;
-         rb->PutValues     = put_values_DITHER_5R6G5B_ximage;
-         rb->PutMonoValues = put_mono_values_DITHER_5R6G5B_ximage;
+         xrb->Base.PutRow        = put_row_DITHER_5R6G5B_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_DITHER_5R6G5B_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_DITHER_5R6G5B_ximage;
+         xrb->Base.PutValues     = put_values_DITHER_5R6G5B_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_DITHER_5R6G5B_ximage;
       }
       break;
    case PF_Dither:
       if (pixmap) {
-         rb->PutRow        = put_row_DITHER_pixmap;
-         rb->PutRowRGB     = put_row_rgb_DITHER_pixmap;
-         rb->PutMonoRow    = put_mono_row_DITHER_pixmap;
-         rb->PutValues     = put_values_DITHER_pixmap;
-         rb->PutMonoValues = put_mono_values_DITHER_pixmap;
+         xrb->Base.PutRow        = put_row_DITHER_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_DITHER_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_DITHER_pixmap;
+         xrb->Base.PutValues     = put_values_DITHER_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_DITHER_pixmap;
       }
       else {
          if (depth == 8) {
-            rb->PutRow        = put_row_DITHER8_ximage;
-            rb->PutRowRGB     = put_row_rgb_DITHER8_ximage;
-            rb->PutMonoRow    = put_mono_row_DITHER8_ximage;
-            rb->PutValues     = put_values_DITHER8_ximage;
-            rb->PutMonoValues = put_mono_values_DITHER8_ximage;
+            xrb->Base.PutRow        = put_row_DITHER8_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_DITHER8_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_DITHER8_ximage;
+            xrb->Base.PutValues     = put_values_DITHER8_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_DITHER8_ximage;
          }
          else {
-            rb->PutRow        = put_row_DITHER_ximage;
-            rb->PutRowRGB     = put_row_rgb_DITHER_ximage;
-            rb->PutMonoRow    = put_mono_row_DITHER_ximage;
-            rb->PutValues     = put_values_DITHER_ximage;
-            rb->PutMonoValues = put_mono_values_DITHER_ximage;
+            xrb->Base.PutRow        = put_row_DITHER_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_DITHER_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_DITHER_ximage;
+            xrb->Base.PutValues     = put_values_DITHER_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_DITHER_ximage;
          }
       }
       break;
    case PF_1Bit:
       if (pixmap) {
-         rb->PutRow        = put_row_1BIT_pixmap;
-         rb->PutRowRGB     = put_row_rgb_1BIT_pixmap;
-         rb->PutMonoRow    = put_mono_row_1BIT_pixmap;
-         rb->PutValues     = put_values_1BIT_pixmap;
-         rb->PutMonoValues = put_mono_values_1BIT_pixmap;
+         xrb->Base.PutRow        = put_row_1BIT_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_1BIT_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_1BIT_pixmap;
+         xrb->Base.PutValues     = put_values_1BIT_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_1BIT_pixmap;
       }
       else {
-         rb->PutRow        = put_row_1BIT_ximage;
-         rb->PutRowRGB     = put_row_rgb_1BIT_ximage;
-         rb->PutMonoRow    = put_mono_row_1BIT_ximage;
-         rb->PutValues     = put_values_1BIT_ximage;
-         rb->PutMonoValues = put_mono_values_1BIT_ximage;
+         xrb->Base.PutRow        = put_row_1BIT_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_1BIT_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_1BIT_ximage;
+         xrb->Base.PutValues     = put_values_1BIT_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_1BIT_ximage;
       }
       break;
    case PF_HPCR:
       if (pixmap) {
-         rb->PutRow        = put_row_HPCR_pixmap;
-         rb->PutRowRGB     = put_row_rgb_HPCR_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_HPCR_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_HPCR_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_HPCR_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_HPCR_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
-         rb->PutRow        = put_row_HPCR_ximage;
-         rb->PutRowRGB     = put_row_rgb_HPCR_ximage;
-         rb->PutMonoRow    = put_mono_row_HPCR_ximage;
-         rb->PutValues     = put_values_HPCR_ximage;
-         rb->PutMonoValues = put_mono_values_HPCR_ximage;
+         xrb->Base.PutRow        = put_row_HPCR_ximage;
+         xrb->Base.PutRowRGB     = put_row_rgb_HPCR_ximage;
+         xrb->Base.PutMonoRow    = put_mono_row_HPCR_ximage;
+         xrb->Base.PutValues     = put_values_HPCR_ximage;
+         xrb->Base.PutMonoValues = put_mono_values_HPCR_ximage;
       }
       break;
    case PF_Lookup:
       if (pixmap) {
-         rb->PutRow        = put_row_LOOKUP_pixmap;
-         rb->PutRowRGB     = put_row_rgb_LOOKUP_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_LOOKUP_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_LOOKUP_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_LOOKUP_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_LOOKUP_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
          if (depth==8) {
-            rb->PutRow        = put_row_LOOKUP8_ximage;
-            rb->PutRowRGB     = put_row_rgb_LOOKUP8_ximage;
-            rb->PutMonoRow    = put_mono_row_LOOKUP8_ximage;
-            rb->PutValues     = put_values_LOOKUP8_ximage;
-            rb->PutMonoValues = put_mono_values_LOOKUP8_ximage;
+            xrb->Base.PutRow        = put_row_LOOKUP8_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_LOOKUP8_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_LOOKUP8_ximage;
+            xrb->Base.PutValues     = put_values_LOOKUP8_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_LOOKUP8_ximage;
          }
          else {
-            rb->PutRow        = put_row_LOOKUP_ximage;
-            rb->PutRowRGB     = put_row_rgb_LOOKUP_ximage;
-            rb->PutMonoRow    = put_mono_row_ximage;
-            rb->PutValues     = put_values_LOOKUP_ximage;
-            rb->PutMonoValues = put_mono_values_ximage;
+            xrb->Base.PutRow        = put_row_LOOKUP_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_LOOKUP_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_ximage;
+            xrb->Base.PutValues     = put_values_LOOKUP_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_ximage;
          }
       }
       break;
    case PF_Grayscale:
       if (pixmap) {
-         rb->PutRow        = put_row_GRAYSCALE_pixmap;
-         rb->PutRowRGB     = put_row_rgb_GRAYSCALE_pixmap;
-         rb->PutMonoRow    = put_mono_row_pixmap;
-         rb->PutValues     = put_values_GRAYSCALE_pixmap;
-         rb->PutMonoValues = put_mono_values_pixmap;
+         xrb->Base.PutRow        = put_row_GRAYSCALE_pixmap;
+         xrb->Base.PutRowRGB     = put_row_rgb_GRAYSCALE_pixmap;
+         xrb->Base.PutMonoRow    = put_mono_row_pixmap;
+         xrb->Base.PutValues     = put_values_GRAYSCALE_pixmap;
+         xrb->Base.PutMonoValues = put_mono_values_pixmap;
       }
       else {
          if (depth == 8) {
-            rb->PutRow        = put_row_GRAYSCALE8_ximage;
-            rb->PutRowRGB     = put_row_rgb_GRAYSCALE8_ximage;
-            rb->PutMonoRow    = put_mono_row_GRAYSCALE8_ximage;
-            rb->PutValues     = put_values_GRAYSCALE8_ximage;
-            rb->PutMonoValues = put_mono_values_GRAYSCALE8_ximage;
+            xrb->Base.PutRow        = put_row_GRAYSCALE8_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_GRAYSCALE8_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_GRAYSCALE8_ximage;
+            xrb->Base.PutValues     = put_values_GRAYSCALE8_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_GRAYSCALE8_ximage;
          }
          else {
-            rb->PutRow        = put_row_GRAYSCALE_ximage;
-            rb->PutRowRGB     = put_row_rgb_GRAYSCALE_ximage;
-            rb->PutMonoRow    = put_mono_row_ximage;
-            rb->PutValues     = put_values_GRAYSCALE_ximage;
-            rb->PutMonoValues = put_mono_values_ximage;
+            xrb->Base.PutRow        = put_row_GRAYSCALE_ximage;
+            xrb->Base.PutRowRGB     = put_row_rgb_GRAYSCALE_ximage;
+            xrb->Base.PutMonoRow    = put_mono_row_ximage;
+            xrb->Base.PutValues     = put_values_GRAYSCALE_ximage;
+            xrb->Base.PutMonoValues = put_mono_values_ximage;
          }
       }
       break;
@@ -4804,12 +4790,12 @@ xmesa_set_renderbuffer_funcs(struct xmesa_renderbuffer *xrb,
 
    /* Get functions */
    if (pixelformat == PF_Index) {
-      rb->GetRow = get_row_ci;
-      rb->GetValues = get_values_ci;
+      xrb->Base.GetRow = get_row_ci;
+      xrb->Base.GetValues = get_values_ci;
    }
    else {
-      rb->GetRow = get_row_rgba;
-      rb->GetValues = get_values_rgba;
+      xrb->Base.GetRow = get_row_rgba;
+      xrb->Base.GetValues = get_values_rgba;
    }
 }
 
diff --git a/src/mesa/drivers/x11/xm_surface.c b/src/mesa/drivers/x11/xm_surface.c
deleted file mode 100644
index a3f2fe7d685..00000000000
--- a/src/mesa/drivers/x11/xm_surface.c
+++ /dev/null
@@ -1,251 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.1
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/**
- * \file xm_surface.c
- * Code to allow the softpipe code to write to X windows/buffers.
- * This is a bit of a hack for now.  We've basically got two different
- * abstractions for color buffers: gl_renderbuffer and pipe_surface.
- * They'll need to get merged someday...
- * For now, they're separate things that point to each other.
- */
-
-
-#include "glxheader.h"
-#include "GL/xmesa.h"
-#include "xmesaP.h"
-#include "context.h"
-#include "imports.h"
-#include "macros.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_winsys.h"
-#include "softpipe/sp_context.h"
-#include "softpipe/sp_clear.h"
-#include "softpipe/sp_tile_cache.h"
-#include "softpipe/sp_surface.h"
-#include "state_tracker/st_context.h"
-
-
-#define CLIP_TILE \
-   do { \
-      if (x + w > ps->width) \
-         w = ps->width - x; \
-      if (y + h > ps->height) \
-         h = ps->height -y; \
-   } while(0)
-
-
-static INLINE struct xmesa_surface *
-xmesa_surface(struct pipe_surface *ps)
-{
-   return (struct xmesa_surface *) ps;
-}
-
-
-static INLINE struct xmesa_renderbuffer *
-xmesa_rb(struct pipe_surface *ps)
-{
-   struct xmesa_surface *xms = xmesa_surface(ps);
-   return xms->xrb;
-}
-
-
-#define FLIP(Y) Y = xrb->St.Base.Height - (Y) - 1;
-
-
-void
-xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps,
-                    uint x, uint y, uint w, uint h, float *p)
-{
-   struct xmesa_surface *xms = xmesa_surface(ps);
-   struct xmesa_renderbuffer *xrb = xms->xrb;
-
-   if (xrb) {
-      /* this is a front/back color buffer */
-      GLubyte tmp[MAX_WIDTH * 4];
-      GLuint i, j;
-      uint w0 = w;
-      GET_CURRENT_CONTEXT(ctx);
-
-      CLIP_TILE;
-
-      FLIP(y);
-      for (i = 0; i < h; i++) {
-         xrb->St.Base.GetRow(ctx, &xrb->St.Base, w, x, y - i, tmp);
-         for (j = 0; j < w * 4; j++) {
-            p[j] = UBYTE_TO_FLOAT(tmp[j]);
-         }
-         p += w0 * 4;
-      }
-   }
-   else {
-      /* other softpipe surface */
-      softpipe_get_tile_rgba(ps, x, y, w, h, p);
-   }
-}
-
-
-void
-xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps,
-                    uint x, uint y, uint w, uint h, const float *p)
-{
-   struct xmesa_surface *xms = xmesa_surface(ps);
-   struct xmesa_renderbuffer *xrb = xms->xrb;
-
-   if (xrb) {
-      /* this is a front/back color buffer */
-      GLubyte tmp[MAX_WIDTH * 4];
-      GLuint i, j;
-      uint w0 = w;
-      GET_CURRENT_CONTEXT(ctx);
-      CLIP_TILE;
-      FLIP(y);
-      for (i = 0; i < h; i++) {
-         for (j = 0; j < w * 4; j++) {
-            UNCLAMPED_FLOAT_TO_UBYTE(tmp[j], p[j]);
-         }
-         xrb->St.Base.PutRow(ctx, &xrb->St.Base, w, x, y - i, tmp, NULL);
-         p += w0 * 4;
-      }
-#if 0 /* debug: flush */
-      {
-         XMesaContext xm = XMESA_CONTEXT(ctx);
-         XSync(xm->display, 0);
-      }
-#endif
-   }
-   else {
-      /* other softpipe surface */
-      softpipe_put_tile_rgba(ps, x, y, w, h, p);
-   }
-}
-
-
-
-/**
- * Called to create a pipe_surface for each X renderbuffer.
- * Note: this is being used instead of pipe->surface_alloc() since we
- * have special/unique quad read/write functions for X.
- */
-struct pipe_surface *
-xmesa_new_color_surface(struct pipe_context *pipe, GLuint pipeFormat)
-{
-   struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface);
-
-   assert(pipeFormat);
-
-   xms->surface.format = pipeFormat;
-   xms->surface.refcount = 1;
-
-   /* Note, the region we allocate doesn't actually have any storage
-    * since we're drawing into an XImage or Pixmap.
-    * The region's size will get set in the xmesa_alloc_front/back_storage()
-    * functions.
-    */
-   if (pipe)
-      xms->surface.region = pipe->winsys->region_alloc(pipe->winsys,
-                                                       1, 0, 0, 0x0);
-
-   return &xms->surface;
-}
-
-
-/**
- * Called via pipe->surface_alloc() to create new surfaces (textures,
- * renderbuffers, etc.
- */
-struct pipe_surface *
-xmesa_surface_alloc(struct pipe_context *pipe, GLuint pipeFormat)
-{
-   struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface);
-
-   assert(pipe);
-   assert(pipeFormat);
-
-   xms->surface.format = pipeFormat;
-   xms->surface.refcount = 1;
-
-   return &xms->surface;
-}
-
-
-boolean
-xmesa_is_format_supported(struct pipe_context *pipe, uint format)
-{
-   switch( format ) {
-   case PIPE_FORMAT_U_A8_R8_G8_B8:
-   case PIPE_FORMAT_S_R16_G16_B16_A16:
-   case PIPE_FORMAT_S8_Z24:
-      return TRUE;
-   };
-   return FALSE;
-}
-
-
-/**
- * Called via pipe->clear()
- */
-void
-xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, GLuint value)
-{
-   struct xmesa_renderbuffer *xrb = xmesa_rb(ps);
-
-   /* XXX actually, we should just discard any cached tiles from this
-    * surface since we don't want to accidentally re-use them after clearing.
-    */
-   pipe->flush(pipe, 0);
-
-   {
-      struct softpipe_context *sp = softpipe_context(pipe);
-      if (ps == sp_tile_cache_get_surface(sp->cbuf_cache[0])) {
-         float clear[4];
-         clear[0] = 0.2; /* XXX hack */
-         clear[1] = 0.2;
-         clear[2] = 0.2;
-         clear[3] = 0.2;
-         sp_tile_cache_clear(sp->cbuf_cache[0], clear);
-      }
-   }
-
-   if (xrb && xrb->ximage) {
-      /* clearing back color buffer */
-      GET_CURRENT_CONTEXT(ctx);
-      xmesa_clear_buffers(ctx, BUFFER_BIT_BACK_LEFT);
-   }
-   else if (xrb && xrb->pixmap) {
-      /* clearing front color buffer */
-      GET_CURRENT_CONTEXT(ctx);
-      xmesa_clear_buffers(ctx, BUFFER_BIT_FRONT_LEFT);
-   }
-   else {
-      /* clearing other buffer */
-      softpipe_clear(pipe, ps, value);
-   }
-}
-
diff --git a/src/mesa/drivers/x11/xm_tri.c b/src/mesa/drivers/x11/xm_tri.c
index 9f17083f901..3a0cf801397 100644
--- a/src/mesa/drivers/x11/xm_tri.c
+++ b/src/mesa/drivers/x11/xm_tri.c
@@ -30,11 +30,11 @@
  */
 
 
+#include "main/depth.h"
+#include "main/macros.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
 #include "glxheader.h"
-#include "depth.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
 #include "xmesaP.h"
 
 /* Internal swrast includes:
@@ -45,7 +45,7 @@
 
 
 #define GET_XRB(XRB)  struct xmesa_renderbuffer *XRB = \
-   xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped)
+   xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped)
 
 
 /**********************************************************************/
@@ -66,7 +66,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
 
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    GLuint i;							\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -82,7 +82,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 
 #include "swrast/s_tritemp.h"
 
@@ -102,7 +102,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       const DEPTH_TYPE z = FixedToDepth(span.z);		\
@@ -117,7 +117,7 @@
       span.blue += span.blueStep;				\
       span.alpha += span.alphaStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 
 #include "swrast/s_tritemp.h"
 
@@ -137,7 +137,7 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);
 
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       const DEPTH_TYPE z = FixedToDepth(span.z);		\
@@ -152,7 +152,7 @@
       span.blue += span.blueStep;				\
       span.alpha += span.alphaStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 
 #include "swrast/s_tritemp.h"
 
@@ -171,7 +171,7 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);
 
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       const DEPTH_TYPE z = FixedToDepth(span.z);		\
@@ -184,7 +184,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 
 #include "swrast/s_tritemp.h"
 
@@ -202,7 +202,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span ) 					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       const DEPTH_TYPE z = FixedToDepth(span.z);		\
@@ -217,7 +217,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -232,7 +232,7 @@
 #define SETUP_CODE						\
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -248,7 +248,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -265,7 +265,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       const DEPTH_TYPE z = FixedToDepth(span.z);		\
@@ -278,7 +278,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -296,7 +296,7 @@
 #define SETUP_CODE						\
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -310,7 +310,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -327,7 +327,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    XDITHER_SETUP(y);						\
@@ -342,7 +342,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -357,7 +357,7 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    XDITHER_SETUP(y);						\
@@ -373,7 +373,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -390,7 +390,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    LOOKUP_SETUP;						\
    for (i = 0; i < span.end; i++) {				\
@@ -404,7 +404,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -422,7 +422,7 @@
 #define SETUP_CODE						\
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -436,7 +436,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -453,7 +453,7 @@
    XMesaImage *img = xrb->ximage;				\
    unsigned long pixel;						\
    PACK_TRUECOLOR(pixel, v2->color[0], v2->color[1], v2->color[2]);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -463,7 +463,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -481,7 +481,7 @@
    GET_XRB(xrb);					\
    GLuint p = PACK_8A8B8G8R( v2->color[0], v2->color[1],\
                              v2->color[2], v2->color[3]);
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       const DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -490,7 +490,7 @@
          zRow[i] = z;					\
       }							\
       span.z += span.zStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -508,7 +508,7 @@
    GET_XRB(xrb);					\
    GLuint p = PACK_8A8R8G8B(v2->color[0], v2->color[1],	\
                             v2->color[2], v2->color[3]);
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       const DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -517,7 +517,7 @@
          zRow[i] = z;					\
       }							\
       span.z += span.zStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -534,7 +534,7 @@
 #define SETUP_CODE					\
    GET_XRB(xrb);					\
    GLuint p = PACK_8R8G8B( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )			\
+#define RENDER_SPAN( span ) {			\
    GLuint i;					\
    for (i = 0; i < span.end; i++) {		\
       DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -543,7 +543,7 @@
          zRow[i] = z;				\
       }						\
       span.z += span.zStep;			\
-   }
+   } }
 
 #include "swrast/s_tritemp.h"
 
@@ -561,7 +561,7 @@
 #define SETUP_CODE					\
    GET_XRB(xrb);					\
    const GLubyte *color = v2->color;
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       const DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -573,7 +573,7 @@
          zRow[i] = z;					\
       }							\
       span.z += span.zStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -588,7 +588,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -601,7 +601,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -618,7 +618,7 @@
 #define SETUP_CODE					\
    GET_XRB(xrb);					\
    GLushort p = PACK_5R6G5B( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       const DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -627,7 +627,7 @@
          zRow[i] = z;					\
       }							\
       span.z += span.zStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -645,7 +645,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    const GLubyte *color = v2->color;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -656,7 +656,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -673,7 +673,7 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);						\
    FLAT_DITHER_SETUP( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);			\
    FLAT_DITHER_ROW_SETUP(YFLIP(xrb, y));			\
@@ -684,7 +684,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -699,7 +699,7 @@
    GET_XRB(xrb);					\
    XMesaImage *img = xrb->ximage;		\
    FLAT_DITHER_SETUP( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    FLAT_DITHER_ROW_SETUP(y);					\
@@ -711,7 +711,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -731,7 +731,7 @@
    GLubyte r = v2->color[0];					\
    GLubyte g = v2->color[1];					\
    GLubyte b = v2->color[2];
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -741,7 +741,7 @@
          zRow[i] = z;						\
       }								\
       span.z += span.zStep;					\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -762,7 +762,7 @@
    GLubyte g = v2->color[1];				\
    GLubyte b = v2->color[2];				\
    GLubyte p = LOOKUP(r,g,b);
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       const DEPTH_TYPE z = FixedToDepth(span.z);	\
@@ -771,7 +771,7 @@
          zRow[i] = z;					\
       }							\
       span.z += span.zStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -785,7 +785,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -796,7 +796,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -812,7 +812,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       pRow[i] = PACK_8A8B8G8R(FixedToInt(span.red),		\
@@ -822,7 +822,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.alpha += span.alphaStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -838,7 +838,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       pRow[i] = PACK_8A8R8G8B(FixedToInt(span.red),		\
@@ -848,7 +848,7 @@
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
       span.alpha += span.alphaStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -863,7 +863,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       pRow[i] = PACK_8R8G8B(FixedToInt(span.red),		\
@@ -871,7 +871,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -886,7 +886,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE					\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    PIXEL_TYPE *pixel = pRow;				\
    for (i = 0; i < span.end; i++, pixel++) {		\
@@ -896,7 +896,7 @@
       span.red += span.redStep;				\
       span.green += span.greenStep;			\
       span.blue += span.blueStep;			\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -910,7 +910,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -921,7 +921,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -936,7 +936,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    for (i = 0; i < span.end; i++) {				\
       pRow[i] = (PIXEL_TYPE) PACK_5R6G5B(FixedToInt(span.red),	\
@@ -944,7 +944,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -960,7 +960,7 @@
 #define SETUP_CODE						\
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -969,7 +969,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -984,7 +984,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    XDITHER_SETUP(y);						\
@@ -994,7 +994,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1007,7 +1007,7 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    XDITHER_SETUP(y);						\
@@ -1018,7 +1018,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1033,7 +1033,7 @@
 #define BYTES_PER_ROW (xrb->ximage->bytes_per_line)
 #define SETUP_CODE						\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    LOOKUP_SETUP;						\
    for (i = 0; i < span.end; i++) {				\
@@ -1042,7 +1042,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1058,7 +1058,7 @@
 #define SETUP_CODE						\
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -1067,7 +1067,7 @@
       span.red += span.redStep;					\
       span.green += span.greenStep;				\
       span.blue += span.blueStep;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1082,12 +1082,12 @@
    XMesaImage *img = xrb->ximage;				\
    unsigned long pixel;						\
    PACK_TRUECOLOR(pixel, v2->color[0], v2->color[1], v2->color[2]);
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
       XMesaPutPixel(img, x, y, pixel);				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1103,11 +1103,11 @@
    GET_XRB(xrb);					\
    unsigned long p = PACK_8B8G8R( v2->color[0],		\
 		 v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       pRow[i] = (PIXEL_TYPE) p;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1123,11 +1123,11 @@
    GET_XRB(xrb);					\
    unsigned long p = PACK_8R8G8B( v2->color[0],		\
 		 v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       pRow[i] = (PIXEL_TYPE) p;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1143,11 +1143,11 @@
    GET_XRB(xrb);					\
    unsigned long p = PACK_8R8G8B( v2->color[0],		\
 		 v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       pRow[i] = (PIXEL_TYPE) p;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1162,14 +1162,14 @@
 #define SETUP_CODE					\
    GET_XRB(xrb);					\
    const GLubyte *color = v2->color;
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    PIXEL_TYPE *pixel = pRow;				\
    for (i = 0; i < span.end; i++, pixel++) {		\
       pixel->r = color[RCOMP];				\
       pixel->g = color[GCOMP];				\
       pixel->b = color[BCOMP];				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1182,7 +1182,7 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
@@ -1190,7 +1190,7 @@
       PACK_TRUEDITHER(p, x, y, v2->color[0],			\
                v2->color[1], v2->color[2] );			\
       XMesaPutPixel(img, x, y, p);				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1206,11 +1206,11 @@
    GET_XRB(xrb);					\
    unsigned long p = PACK_5R6G5B( v2->color[0],		\
 		 v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )				\
+#define RENDER_SPAN( span ) {				\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       pRow[i] = (PIXEL_TYPE) p;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1226,13 +1226,13 @@
    XMesaContext xmesa = XMESA_CONTEXT(ctx);			\
    GET_XRB(xrb);						\
    const GLubyte *color = v2->color;
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
       PACK_TRUEDITHER(pRow[i], x, y, color[RCOMP],		\
          color[GCOMP], color[BCOMP]);				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1247,13 +1247,13 @@
 #define SETUP_CODE						\
    GET_XRB(xrb);						\
    FLAT_DITHER_SETUP( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    FLAT_DITHER_ROW_SETUP(YFLIP(xrb, y));		\
    for (i = 0; i < span.end; i++, x++) {			\
       pRow[i] = (PIXEL_TYPE) FLAT_DITHER(x);			\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1266,14 +1266,14 @@
    GET_XRB(xrb);						\
    XMesaImage *img = xrb->ximage;				\
    FLAT_DITHER_SETUP( v2->color[0], v2->color[1], v2->color[2] );
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    FLAT_DITHER_ROW_SETUP(y);					\
    for (i = 0; i < span.end; i++, x++) {			\
       unsigned long p = FLAT_DITHER(x);				\
       XMesaPutPixel(img, x, y, p );				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1291,12 +1291,12 @@
    GLubyte r = v2->color[0];					\
    GLubyte g = v2->color[1];					\
    GLubyte b = v2->color[2];
-#define RENDER_SPAN( span )					\
+#define RENDER_SPAN( span ) {					\
    GLuint i;							\
    GLint x = span.x, y = YFLIP(xrb, span.y);	\
    for (i = 0; i < span.end; i++, x++) {			\
       pRow[i] = (PIXEL_TYPE) DITHER_HPCR(x, y, r, g, b);	\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1315,11 +1315,11 @@
    GLubyte g = v2->color[1];				\
    GLubyte b = v2->color[2];				\
    GLubyte p = LOOKUP(r,g,b);
-#define RENDER_SPAN( span )         	 		\
+#define RENDER_SPAN( span ) {         	 		\
    GLuint i;						\
    for (i = 0; i < span.end; i++) {			\
       pRow[i] = (PIXEL_TYPE) p;				\
-   }
+   } }
 #include "swrast/s_tritemp.h"
 
 
@@ -1443,46 +1443,6 @@ do {                                   \
 #endif
 
 
-#if 0
-GLboolean xmesa_get_cbuf_details( GLcontext *ctx,
-				  void **ptr,
-				  GLuint *cpp,
-				  GLint *stride,
-				  GLuint *format )
-{
-   XMesaContext xmesa = XMESA_CONTEXT(ctx);
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_renderbuffer *crb = fb->_ColorDrawBuffers[0][0];
-   struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(crb->Wrapped);
-
-   *ptr = crb->GetPointer(ctx, crb, 0, 0);
-   *stride = ((GLubyte *)crb->GetPointer(ctx, crb, 0, 1) - 
-	      (GLubyte *)crb->GetPointer(ctx, crb, 0, 0));
-
-   if (!ptr) 
-      goto bad;
-
-   switch (xmesa->pixelformat) {
-   case PF_8A8B8G8R:
-   case PF_8A8R8G8B:
-      *format = 1;		/* whatever */
-      *cpp = 4;
-      break;
-   default:
-      goto bad;
-   }
-
-   return GL_TRUE;
-
- bad:
-   *ptr = NULL;
-   *stride = 0;
-   *format = 0;
-   return GL_FALSE;   
-}
-#endif
-
-
 /**
  * Return pointer to line drawing function, or NULL if we should use a
  * swrast fallback.
@@ -1502,8 +1462,8 @@ get_triangle_func(GLcontext *ctx)
 #endif
 
    /* trivial fallback tests */
-   if ((ctx->DrawBuffer->_ColorDrawBufferMask[0]
-        & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT)) == 0)
+   if ((ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_FRONT_LEFT) &&
+       (ctx->DrawBuffer->_ColorDrawBufferIndexes[0] != BUFFER_BIT_BACK_LEFT))
       return (swrast_tri_func) NULL;
    if (ctx->RenderMode != GL_RENDER)
       return (swrast_tri_func) NULL;
@@ -1519,7 +1479,7 @@ get_triangle_func(GLcontext *ctx)
    if (xmbuf->swAlpha)
       return (swrast_tri_func) NULL;
 
-   xrb = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]->Wrapped);
+   xrb = xmesa_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]->Wrapped);
 
    if (xrb->ximage) {
       if (   ctx->Light.ShadeModel==GL_SMOOTH
diff --git a/src/mesa/drivers/x11/xm_winsys.c b/src/mesa/drivers/x11/xm_winsys.c
deleted file mode 100644
index eab9fd38521..00000000000
--- a/src/mesa/drivers/x11/xm_winsys.c
+++ /dev/null
@@ -1,362 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell
- *   Brian Paul
- */
-
-
-#include "glxheader.h"
-#include "xmesaP.h"
-#include "main/macros.h"
-
-#include "pipe/p_winsys.h"
-#include "softpipe/sp_winsys.h"
-
-
-/**
- * XMesa winsys, derived from softpipe winsys.
- * NOTE: there's nothing really X-specific in this winsys layer so
- * we could probably lift it up somewhere.
- */
-struct xm_winsys
-{
-   struct softpipe_winsys sws;
-   int foo; /* placeholder */
-};
-
-
-/**
- * Low-level OS/window system memory buffer
- */
-struct xm_buffer
-{
-   boolean userBuffer;  /** Is this a user-space buffer? */
-   int refcount;
-   unsigned size;
-   void *data;
-   void *mapped;
-};
-
-
-
-/* Turn the softpipe opaque buffer pointer into a dri_bufmgr opaque
- * buffer pointer...
- */
-static inline struct xm_buffer *
-xm_bo( struct pipe_buffer *bo )
-{
-   return (struct xm_buffer *) bo;
-}
-
-static inline struct pipe_buffer *
-pipe_bo( struct xm_buffer *bo )
-{
-   return (struct pipe_buffer *) bo;
-}
-
-/* Turn a softpipe winsys into an xm/softpipe winsys:
- */
-static inline struct xm_winsys *
-xm_winsys(struct softpipe_winsys *sws)
-{
-   return (struct xm_winsys *) sws;
-}
-
-
-/* Most callbacks map direcly onto dri_bufmgr operations:
- */
-static void *
-xm_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buf,
-              unsigned flags)
-{
-   struct xm_buffer *xm_buf = xm_bo(buf);
-   xm_buf->mapped = xm_buf->data;
-   return xm_buf->mapped;
-}
-
-static void
-xm_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buf)
-{
-   struct xm_buffer *xm_buf = xm_bo(buf);
-   xm_buf->mapped = NULL;
-}
-
-static void
-xm_buffer_reference(struct pipe_winsys *pws,
-                    struct pipe_buffer **ptr,
-                    struct pipe_buffer *buf)
-{
-   if (*ptr) {
-      struct xm_buffer *oldBuf = xm_bo(*ptr);
-      oldBuf->refcount--;
-      assert(oldBuf->refcount >= 0);
-      if (oldBuf->refcount == 0) {
-         if (oldBuf->data) {
-            if (!oldBuf->userBuffer)
-               free(oldBuf->data);
-            oldBuf->data = NULL;
-         }
-         free(oldBuf);
-      }
-      *ptr = NULL;
-   }
-
-   assert(!(*ptr));
-
-   if (buf) {
-      struct xm_buffer *newBuf = xm_bo(buf);
-      newBuf->refcount++;
-      *ptr = buf;
-   }
-}
-
-static void
-xm_buffer_data(struct pipe_winsys *pws, struct pipe_buffer *buf,
-               unsigned size, const void *data, unsigned usage)
-{
-   struct xm_buffer *xm_buf = xm_bo(buf);
-   assert(!xm_buf->userBuffer);
-   if (xm_buf->size != size) {
-      if (xm_buf->data)
-         free(xm_buf->data);
-      xm_buf->data = malloc(size);
-      xm_buf->size = size;
-   }
-   if (data)
-      memcpy(xm_buf->data, data, size);
-}
-
-static void
-xm_buffer_subdata(struct pipe_winsys *pws, struct pipe_buffer *buf,
-                  unsigned long offset, unsigned long size, const void *data)
-{
-   struct xm_buffer *xm_buf = xm_bo(buf);
-   GLubyte *b = (GLubyte *) xm_buf->data;
-   assert(!xm_buf->userBuffer);
-   assert(b);
-   memcpy(b + offset, data, size);
-}
-
-static void
-xm_buffer_get_subdata(struct pipe_winsys *pws, struct pipe_buffer *buf,
-                      unsigned long offset, unsigned long size, void *data)
-{
-   const struct xm_buffer *xm_buf = xm_bo(buf);
-   const GLubyte *b = (GLubyte *) xm_buf->data;
-   assert(!xm_buf->userBuffer);
-   assert(b);
-   memcpy(data, b + offset, size);
-}
-
-static void
-xm_flush_frontbuffer(struct pipe_winsys *pws)
-{
-   /*
-   struct intel_context *intel = intel_pipe_winsys(sws)->intel;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   
-   intelCopyBuffer(dPriv, NULL);
-   */
-}
-
-static void
-xm_wait_idle(struct pipe_winsys *pws)
-{
-   /* no-op */
-}
-
-static const char *
-xm_get_name(struct pipe_winsys *pws)
-{
-   return "Xlib";
-}
-
-
-static struct pipe_buffer *
-xm_buffer_create(struct pipe_winsys *pws, 
-                 unsigned alignment, 
-                 unsigned flags, 
-                 unsigned hint)
-{
-   struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer);
-   buffer->refcount = 1;
-   return pipe_bo(buffer);
-}
-
-
-/**
- * Create buffer which wraps user-space data.
- */
-static struct pipe_buffer *
-xm_user_buffer_create(struct pipe_winsys *pws, void *ptr, unsigned bytes)
-{
-   struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer);
-   buffer->userBuffer = TRUE;
-   buffer->refcount = 1;
-   buffer->data = ptr;
-   buffer->size = bytes;
-   return pipe_bo(buffer);
-}
-
-
-
-/**
- * Round n up to next multiple.
- */
-static INLINE unsigned
-round_up(unsigned n, unsigned multiple)
-{
-   return (n + multiple - 1) & ~(multiple - 1);
-}
-
-
-static struct pipe_region *
-xm_region_alloc(struct pipe_winsys *winsys,
-                unsigned cpp, unsigned width, unsigned height, unsigned flags)
-{
-   struct pipe_region *region = CALLOC_STRUCT(pipe_region);
-   const unsigned alignment = 64;
-
-   region->cpp = cpp;
-   region->pitch = round_up(width, alignment / cpp);
-   region->height = height;
-   region->refcount = 1;
-
-   assert(region->pitch > 0);
-
-   region->buffer = winsys->buffer_create( winsys, alignment, 0, 0 )
-;
-
-   /* NULL data --> just allocate the space */
-   winsys->buffer_data( winsys,
-                        region->buffer, 
-                        region->pitch * cpp * height, 
-                        NULL,
-                        PIPE_BUFFER_USAGE_PIXEL );
-   return region;
-}
-
-
-static void
-xm_region_release(struct pipe_winsys *winsys, struct pipe_region **region)
-{
-   if (!*region)
-      return;
-
-   assert((*region)->refcount > 0);
-   (*region)->refcount--;
-
-   if ((*region)->refcount == 0) {
-      assert((*region)->map_refcount == 0);
-
-      winsys->buffer_reference( winsys, &((*region)->buffer), NULL );
-      free(*region);
-   }
-   *region = NULL;
-}
-
-
-/**
- * Called via pipe->surface_alloc() to create new surfaces (textures,
- * renderbuffers, etc.
- */
-static struct pipe_surface *
-xm_surface_alloc(struct pipe_winsys *ws, GLuint pipeFormat)
-{
-   struct xmesa_surface *xms = CALLOC_STRUCT(xmesa_surface);
-
-   assert(ws);
-   assert(pipeFormat);
-
-   xms->surface.format = pipeFormat;
-   xms->surface.refcount = 1;
-#if 0
-   /*
-    * This is really just a softpipe surface, not an XImage/Pixmap surface.
-    */
-   softpipe_init_surface_funcs(&xms->surface);
-#endif
-   return &xms->surface;
-}
-
-
-
-
-struct xmesa_pipe_winsys
-{
-   struct pipe_winsys winsys;
-   XMesaContext xmesa;
-};
-
-static struct pipe_winsys *
-xmesa_create_pipe_winsys( XMesaContext xmesa )
-{
-   struct xmesa_pipe_winsys *xws = CALLOC_STRUCT(xmesa_pipe_winsys);
-   
-   /* Fill in this struct with callbacks that pipe will need to
-    * communicate with the window system, buffer manager, etc. 
-    *
-    * Pipe would be happy with a malloc based memory manager, but
-    * the SwapBuffers implementation in this winsys driver requires
-    * that rendering be done to an appropriate _DriBufferObject.  
-    */
-   xws->winsys.buffer_create = xm_buffer_create;
-   xws->winsys.user_buffer_create = xm_user_buffer_create;
-   xws->winsys.buffer_map = xm_buffer_map;
-   xws->winsys.buffer_unmap = xm_buffer_unmap;
-   xws->winsys.buffer_reference = xm_buffer_reference;
-   xws->winsys.buffer_data = xm_buffer_data;
-   xws->winsys.buffer_subdata = xm_buffer_subdata;
-   xws->winsys.buffer_get_subdata = xm_buffer_get_subdata;
-
-   xws->winsys.region_alloc = xm_region_alloc;
-   xws->winsys.region_release = xm_region_release;
-
-   xws->winsys.surface_alloc = xm_surface_alloc;
-
-   xws->winsys.flush_frontbuffer = xm_flush_frontbuffer;
-   xws->winsys.wait_idle = xm_wait_idle;
-   xws->winsys.get_name = xm_get_name;
-   xws->xmesa = xmesa;
-
-   return &xws->winsys;
-}
-
-
-struct pipe_context *
-xmesa_create_softpipe(XMesaContext xmesa)
-{
-   struct xm_winsys *xm_ws = CALLOC_STRUCT( xm_winsys );
-   
-   /* Create the softpipe context:
-    */
-   return softpipe_create( xmesa_create_pipe_winsys(xmesa), &xm_ws->sws );
-}
diff --git a/src/mesa/drivers/x11/xmesaP.h b/src/mesa/drivers/x11/xmesaP.h
index fd2dfcd79aa..98867ac7106 100644
--- a/src/mesa/drivers/x11/xmesaP.h
+++ b/src/mesa/drivers/x11/xmesaP.h
@@ -28,7 +28,7 @@
 
 
 #include "GL/xmesa.h"
-#include "mtypes.h"
+#include "main/mtypes.h"
 #if defined(FX)
 #include "GL/fxmesa.h"
 #include "xm_glide.h"
@@ -36,9 +36,6 @@
 #ifdef XFree86Server
 #include "xm_image.h"
 #endif
-#include "state_tracker/st_cb_fbo.h"
-#include "softpipe/sp_context.h"
-#include "softpipe/sp_surface.h"
 
 
 extern _glthread_Mutex _xmesa_lock;
@@ -130,7 +127,7 @@ struct xmesa_visual {
 
 
 /**
- * Context info, dDerived from GLcontext.
+ * Context info, derived from GLcontext.
  * Basically corresponds to a GLXContext.
  */
 struct xmesa_context {
@@ -180,11 +177,7 @@ typedef enum {
  */
 struct xmesa_renderbuffer
 {
-#if 0
    struct gl_renderbuffer Base;  /* Base class */
-#else
-   struct st_renderbuffer St; /**< Base class */
-#endif
 
    XMesaBuffer Parent;  /**< The XMesaBuffer this renderbuffer belongs to */
    XMesaDrawable drawable;	/* Usually the X window ID */
@@ -203,8 +196,6 @@ struct xmesa_renderbuffer
    GLint bottom;	/* used for FLIP macro, equals height - 1 */
 
    ClearFunc clearFunc;
-
-   void *pSurface;      /** pipe surface */
 };
 
 
@@ -500,8 +491,8 @@ extern const int xmesa_kernel1[16];
  */
 
 extern struct xmesa_renderbuffer *
-xmesa_create_renderbuffer(GLcontext *ctx, GLuint name, const GLvisual *visual,
-                          GLboolean backBuffer);
+xmesa_new_renderbuffer(GLcontext *ctx, GLuint name, const GLvisual *visual,
+                       GLboolean backBuffer);
 
 extern void
 xmesa_delete_framebuffer(struct gl_framebuffer *fb);
@@ -549,7 +540,6 @@ xmesa_renderbuffer(struct gl_renderbuffer *rb)
 /**
  * Return pointer to XMesaContext corresponding to a Mesa GLcontext.
  * Since we're using structure containment, it's just a cast!.
- * XXX should use inlined function for better type safety.
  */
 static INLINE XMesaContext
 XMESA_CONTEXT(GLcontext *ctx)
@@ -561,7 +551,6 @@ XMESA_CONTEXT(GLcontext *ctx)
 /**
  * Return pointer to XMesaBuffer corresponding to a Mesa GLframebuffer.
  * Since we're using structure containment, it's just a cast!.
- * XXX should use inlined function for better type safety.
  */
 static INLINE XMesaBuffer
 XMESA_BUFFER(GLframebuffer *b)
@@ -592,41 +581,4 @@ extern void xmesa_register_swrast_functions( GLcontext *ctx );
 #define ENABLE_EXT_timer_query 0 /* may not have 64-bit GLuint64EXT */
 #endif
 
-
-struct pipe_surface;
-struct pipe_context;
-
-struct xmesa_surface
-{
-   struct pipe_surface surface;
-   struct xmesa_renderbuffer *xrb;
-};
-
-
-extern void
-xmesa_clear(struct pipe_context *pipe, struct pipe_surface *ps, GLuint value);
-
-extern void
-xmesa_clear_buffers(GLcontext *ctx, GLbitfield buffers);
-
-extern struct pipe_context *
-xmesa_create_softpipe(XMesaContext xm);
-
-extern struct pipe_surface *
-xmesa_surface_alloc(struct pipe_context *pipe, GLuint format);
-
-extern struct pipe_surface *
-xmesa_new_color_surface(struct pipe_context *pipe, GLuint format);
-
-extern boolean
-xmesa_is_format_supported(struct pipe_context *pipe, uint format);
-
-extern void
-xmesa_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps,
-                    uint x, uint y, uint w, uint h, float *p);
-
-extern void
-xmesa_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *ps,
-                    uint x, uint y, uint w, uint h, const float *p);
-
 #endif
diff --git a/src/mesa/gl.pc.in b/src/mesa/gl.pc.in
new file mode 100644
index 00000000000..0462b9fca2b
--- /dev/null
+++ b/src/mesa/gl.pc.in
@@ -0,0 +1,12 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: gl
+Description: Mesa OpenGL library
+Requires.private: @GL_PC_REQ_PRIV@
+Version: @VERSION@
+Libs: -L${libdir} -lGL
+Libs.private: @GL_PC_LIB_PRIV@
+Cflags: -I${includedir} @GL_PC_CFLAGS@
diff --git a/src/mesa/glapi/EXT_framebuffer_object.xml b/src/mesa/glapi/EXT_framebuffer_object.xml
index 66f250c003f..1b0de2ad231 100644
--- a/src/mesa/glapi/EXT_framebuffer_object.xml
+++ b/src/mesa/glapi/EXT_framebuffer_object.xml
@@ -59,12 +59,18 @@
     <enum name="FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT"   value="0x8CDC"/>
     <enum name="FRAMEBUFFER_UNSUPPORTED_EXT"              value="0x8CDD"/>
     <enum name="FRAMEBUFFER_STATUS_ERROR_EXT"             value="0x8CDE"/>
-
-    <enum name="FRAMEBUFFER_BINDING_EXT"                  value="0x8CA6"/>
-    <enum name="RENDERBUFFER_BINDING_EXT"                 value="0x8CA7"/>
-    <enum name="MAX_COLOR_ATTACHMENTS_EXT"                value="0x8CDF"/>
-    <enum name="MAX_RENDERBUFFER_SIZE_EXT"                value="0x84E8"/>
-
+    <enum name="FRAMEBUFFER_BINDING_EXT"        count="1" value="0x8CA6">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="RENDERBUFFER_BINDING_EXT"       count="1" value="0x8CA7">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="MAX_COLOR_ATTACHMENTS_EXT"      count="1" value="0x8CDF">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="MAX_RENDERBUFFER_SIZE_EXT"      count="1" value="0x84E8">
+        <size name="Get" mode="get"/>
+    </enum>
     <enum name="INVALID_FRAMEBUFFER_OPERATION_EXT"        value="0x0506"/>
 
     <function name="IsRenderbufferEXT" offset="assign">
diff --git a/src/mesa/glapi/Makefile b/src/mesa/glapi/Makefile
index c52c0903478..adc53d92494 100644
--- a/src/mesa/glapi/Makefile
+++ b/src/mesa/glapi/Makefile
@@ -20,17 +20,30 @@ OUTPUTS = glprocs.h glapitemp.h glapioffsets.h glapitable.h dispatch.h \
 	../../glx/x11/indirect_size.c
 
 
-#XORG_BASE = /home/idr/devel/graphics/Xorg/xserver/xorg
-GLX_DIR   = $(XORG_BASE)/GL/glx
+GLX_DIR = $(XORG_BASE)/glx
 
-SERVER_OUTPUTS = $(GLX_DIR)/indirect_dispatch.c \
+SERVER_GLAPI_FILES = \
+	$(GLX_DIR)/glapi.h \
+	$(GLX_DIR)/glapi.c \
+	$(GLX_DIR)/glthread.c \
+	$(GLX_DIR)/glthread.h
+
+SERVER_OUTPUTS = \
+	$(GLX_DIR)/indirect_dispatch.c \
 	$(GLX_DIR)/indirect_dispatch_swap.c \
 	$(GLX_DIR)/indirect_dispatch.h \
 	$(GLX_DIR)/indirect_reqsize.c \
 	$(GLX_DIR)/indirect_reqsize.h \
+	$(GLX_DIR)/indirect_size.h \
 	$(GLX_DIR)/indirect_size_get.c \
 	$(GLX_DIR)/indirect_size_get.h \
-	$(GLX_DIR)/indirect_table.c
+	$(GLX_DIR)/indirect_table.c \
+	$(GLX_DIR)/glapitemp.h \
+	$(GLX_DIR)/glapitable.h \
+	$(GLX_DIR)/glapioffsets.h \
+	$(GLX_DIR)/glprocs.h \
+	$(GLX_DIR)/dispatch.h \
+	$(SERVER_GLAPI_FILES)
 
 API_XML = gl_API.xml \
 	EXT_framebuffer_object.xml \
@@ -39,82 +52,94 @@ API_XML = gl_API.xml \
 COMMON = gl_XML.py glX_XML.py license.py $(API_XML) typeexpr.py
 COMMON_GLX = $(COMMON) glX_API.xml glX_XML.py glX_proto_common.py
 
-INDENT_FLAGS = -i4 -nut -br -brs -npcs -ce -T GLubyte -T GLbyte -T Bool
+all: check-xorg-source $(OUTPUTS) $(SERVER_OUTPUTS)
+
+check-xorg-source:
+	@if ! test -d $(GLX_DIR); then \
+		echo "ERROR: Must specify path to xserver checkout; set XORG_BASE."; \
+		exit 1; \
+	fi
 
-all: $(OUTPUTS)
+$(GLX_DIR)/%.c: %.c
+	cp $< $@
 
-server: $(SERVER_OUTPUTS)
+$(GLX_DIR)/%.h: %.h
+	cp $< $@
 
-glprocs.h: $(COMMON) gl_procs.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_procs.py > glprocs.h
+glprocs.h $(GLX_DIR)/glprocs.h: gl_procs.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-glapitemp.h: $(COMMON) gl_apitemp.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_apitemp.py > glapitemp.h
+glapitemp.h $(GLX_DIR)/glapitemp.h: gl_apitemp.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-glapioffsets.h: $(COMMON) gl_offsets.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_offsets.py > glapioffsets.h
+glapioffsets.h $(GLX_DIR)/glapioffsets.h: gl_offsets.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-glapitable.h: $(COMMON) gl_table.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_table.py > glapitable.h
+glapitable.h $(GLX_DIR)/glapitable.h: gl_table.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-dispatch.h: $(COMMON) gl_table.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_table.py -m remap_table > dispatch.h
+dispatch.h $(GLX_DIR)/dispatch.h: gl_table.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m remap_table > $@
 
-../main/enums.c: $(COMMON) gl_enums.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_enums.py > ../main/enums.c
+../main/enums.c: gl_enums.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-../x86/glapi_x86.S: $(COMMON) gl_x86_asm.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_x86_asm.py > ../x86/glapi_x86.S
+../x86/glapi_x86.S: gl_x86_asm.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-../x86-64/glapi_x86-64.S: $(COMMON) gl_x86-64_asm.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_x86-64_asm.py > ../x86-64/glapi_x86-64.S
+../x86-64/glapi_x86-64.S: gl_x86-64_asm.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-../sparc/glapi_sparc.S: $(COMMON) gl_SPARC_asm.py
-	$(PYTHON2) $(PYTHON_FLAGS) gl_SPARC_asm.py > ../sparc/glapi_sparc.S
+../sparc/glapi_sparc.S: gl_SPARC_asm.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
 
-../drivers/dri/common/extension_helper.h: $(COMMON) extension_helper.py
-	$(PYTHON2) $(PYTHON_FLAGS) extension_helper.py > ../drivers/dri/common/extension_helper.h
+../drivers/dri/common/extension_helper.h: extension_helper.py $(COMMON)
+	$(PYTHON2) $(PYTHON_FLAGS) $< > $@
 
-../../glx/x11/indirect.c: $(COMMON_GLX) glX_proto_send.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_send.py -m proto | indent $(INDENT_FLAGS) > ../../glx/x11/indirect.c
+../../glx/x11/indirect.c: glX_proto_send.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m proto | $(INDENT) $(INDENT_FLAGS) > $@
 
-../../glx/x11/indirect.h: $(COMMON_GLX) glX_proto_send.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_send.py -m init_h > ../../glx/x11/indirect.h
+../../glx/x11/indirect.h: glX_proto_send.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m init_h > $@
 
-../../glx/x11/indirect_init.c: $(COMMON_GLX) glX_proto_send.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_send.py -m init_c > ../../glx/x11/indirect_init.c
+../../glx/x11/indirect_init.c: glX_proto_send.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m init_c > $@
 
-../../glx/x11/indirect_size.h: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m size_h --only-set -h _INDIRECT_SIZE_H_ | indent $(INDENT_FLAGS) > ../../glx/x11/indirect_size.h
+../../glx/x11/indirect_size.h $(GLX_DIR)/indirect_size.h: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m size_h --only-set -h _INDIRECT_SIZE_H_ \
+	  | $(INDENT) $(INDENT_FLAGS) > $@
 
-../../glx/x11/indirect_size.c: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m size_c --only-set | indent $(INDENT_FLAGS) > ../../glx/x11/indirect_size.c
+../../glx/x11/indirect_size.c: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m size_c --only-set \
+	  | $(INDENT) $(INDENT_FLAGS) > $@
 
-$(GLX_DIR)/indirect_dispatch.c: $(COMMON_GLX) glX_proto_recv.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_recv.py -m dispatch_c > $@
+$(GLX_DIR)/indirect_dispatch.c: glX_proto_recv.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m dispatch_c > $@
 
-$(GLX_DIR)/indirect_dispatch_swap.c: $(COMMON_GLX) glX_proto_recv.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_recv.py -m dispatch_c -s > $@
+$(GLX_DIR)/indirect_dispatch_swap.c: glX_proto_recv.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m dispatch_c -s > $@
 
-$(GLX_DIR)/indirect_dispatch.h: $(COMMON_GLX) glX_proto_recv.py glX_API.xml
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_recv.py -m dispatch_h -f gl_and_glX_API.xml -s > $@
+$(GLX_DIR)/indirect_dispatch.h: glX_proto_recv.py gl_and_glX_API.xml $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m dispatch_h -f gl_and_glX_API.xml -s > $@
 
-$(GLX_DIR)/indirect_size_get.h: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m size_h --only-get -h '_INDIRECT_SIZE_GET_H_' | indent $(INDENT_FLAGS) > $@
+$(GLX_DIR)/indirect_size_get.h: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m size_h --only-get -h '_INDIRECT_SIZE_GET_H_' \
+	  | $(INDENT) $(INDENT_FLAGS) > $@
 
-$(GLX_DIR)/indirect_size_get.c: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m size_c | indent $(INDENT_FLAGS) > $@
+$(GLX_DIR)/indirect_size_get.c: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m size_c | $(INDENT) $(INDENT_FLAGS) > $@
 
-$(GLX_DIR)/indirect_reqsize.h: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m reqsize_h --only-get -h '_INDIRECT_SIZE_GET_H_' | indent $(INDENT_FLAGS) -l200 > $@
+$(GLX_DIR)/indirect_reqsize.h: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m reqsize_h --only-get -h '_INDIRECT_SIZE_GET_H_' \
+	  | $(INDENT) $(INDENT_FLAGS) -l200 > $@
 
-$(GLX_DIR)/indirect_reqsize.c: $(COMMON_GLX) glX_proto_size.py
-	$(PYTHON2) $(PYTHON_FLAGS) glX_proto_size.py -m reqsize_c | indent $(INDENT_FLAGS) > $@
+$(GLX_DIR)/indirect_reqsize.c: glX_proto_size.py $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -m reqsize_c | $(INDENT) $(INDENT_FLAGS) > $@
 
-$(GLX_DIR)/indirect_table.c: $(COMMON_GLX) glX_server_table.py glX_API.xml
-	$(PYTHON2) $(PYTHON_FLAGS) glX_server_table.py -f gl_and_glX_API.xml > $@
+$(GLX_DIR)/indirect_table.c: glX_server_table.py gl_and_glX_API.xml $(COMMON_GLX)
+	$(PYTHON2) $(PYTHON_FLAGS) $< -f gl_and_glX_API.xml > $@
 
 clean:
-	rm -f *~ *.pyo
-	rm -f $(OUTPUTS)
+	-rm -f *~ *.pyo
+	-rm -f $(OUTPUTS)
diff --git a/src/mesa/glapi/dispatch.h b/src/mesa/glapi/dispatch.h
index 98f654f4022..f019aa6aafe 100644
--- a/src/mesa/glapi/dispatch.h
+++ b/src/mesa/glapi/dispatch.h
@@ -28,9 +28,8 @@
 #if !defined( _DISPATCH_H_ )
 #  define _DISPATCH_H_
 
-#include "glapitable.h"
-
 
+#include "glapitable.h"
 /**
  * \file dispatch.h
  * Macros for handling GL dispatch tables.
diff --git a/src/mesa/glapi/extension_helper.py b/src/mesa/glapi/extension_helper.py
index f7fa90c3900..64f64a2fd86 100644
--- a/src/mesa/glapi/extension_helper.py
+++ b/src/mesa/glapi/extension_helper.py
@@ -151,7 +151,7 @@ class PrintGlExtensionGlue(gl_XML.gl_print_base):
 
 	def printRealHeader(self):
 		print '#include "utils.h"'
-		print '#include "dispatch.h"'
+		print '#include "glapi/dispatch.h"'
 		print ''
 		return
 
@@ -174,6 +174,9 @@ class PrintGlExtensionGlue(gl_XML.gl_print_base):
 
 				parameter_signature = ''
 				for p in f.parameterIterator():
+					if p.is_padding:
+						continue
+
 					# FIXME: This is a *really* ugly hack. :(
 
 					tn = p.type_expr.get_base_type_node()
diff --git a/src/mesa/glapi/glX_API.xml b/src/mesa/glapi/glX_API.xml
index 455fc6a5676..9e02aa498ee 100644
--- a/src/mesa/glapi/glX_API.xml
+++ b/src/mesa/glapi/glX_API.xml
@@ -183,7 +183,6 @@
     </function>
 </category>
 
-<!--
 <category name="GLX_SGIX_pbuffer" number="50" window_system="glX">
     <function name="CreateGLXPbufferSGIX">
         <glx vendorpriv="65543"/>
@@ -201,7 +200,6 @@
         <glx vendorpriv="65546"/>
     </function>
 </category>
--->
 
 <category name="GLX_MESA_copy_sub_buffer" number="215">
     <function name="CopySubBufferMESA">
diff --git a/src/mesa/glapi/glX_XML.py b/src/mesa/glapi/glX_XML.py
index 1b5c3239e01..4c2e0f95bf4 100644
--- a/src/mesa/glapi/glX_XML.py
+++ b/src/mesa/glapi/glX_XML.py
@@ -115,7 +115,12 @@ class glx_function(gl_XML.gl_function):
 	def process_element(self, element):
 		gl_XML.gl_function.process_element(self, element)
 
-		self.vectorequiv = element.nsProp( "vectorequiv", None )
+		# If the function already has a vector equivalent set, don't
+		# set it again.  This can happen if an alias to a function
+		# appears after the function that it aliases.
+
+		if not self.vectorequiv:
+			self.vectorequiv = element.nsProp("vectorequiv", None)
 
 
 		name = element.nsProp("name", None)
@@ -132,48 +137,47 @@ class glx_function(gl_XML.gl_function):
 
 		child = element.children
 		while child:
-			if child.type == "element":
-				if child.name == "glx":
-					rop = child.nsProp( 'rop', None )
-					sop = child.nsProp( 'sop', None )
-					vop = child.nsProp( 'vendorpriv', None )
-
-					if rop:
-						self.glx_rop = int(rop)
-
-					if sop:
-						self.glx_sop = int(sop)
-
-					if vop:
-						self.glx_vendorpriv = int(vop)
-						self.glx_vendorpriv_names.append(name)
-
-					self.img_reset = child.nsProp( 'img_reset', None )
-
-					# The 'handcode' attribute can be one of 'true',
-					# 'false', 'client', or 'server'.
-
-					handcode = child.nsProp( 'handcode', None )
-					if handcode == "false":
-						self.server_handcode = 0
-						self.client_handcode = 0
-					elif handcode == "true":
-						self.server_handcode = 1
-						self.client_handcode = 1
-					elif handcode == "client":
-						self.server_handcode = 0
-						self.client_handcode = 1
-					elif handcode == "server":
-						self.server_handcode = 1
-						self.client_handcode = 0
-					else:
-						raise RuntimeError('Invalid handcode mode "%s" in function "%s".' % (handcode, self.name))
-
-					self.ignore               = gl_XML.is_attr_true( child, 'ignore' )
-					self.can_be_large         = gl_XML.is_attr_true( child, 'large' )
-					self.glx_doubles_in_order = gl_XML.is_attr_true( child, 'doubles_in_order' )
-					self.reply_always_array   = gl_XML.is_attr_true( child, 'always_array' )
-					self.dimensions_in_reply  = gl_XML.is_attr_true( child, 'dimensions_in_reply' )
+			if child.type == "element" and child.name == "glx":
+				rop = child.nsProp( 'rop', None )
+				sop = child.nsProp( 'sop', None )
+				vop = child.nsProp( 'vendorpriv', None )
+
+				if rop:
+					self.glx_rop = int(rop)
+
+				if sop:
+					self.glx_sop = int(sop)
+
+				if vop:
+					self.glx_vendorpriv = int(vop)
+					self.glx_vendorpriv_names.append(name)
+
+				self.img_reset = child.nsProp( 'img_reset', None )
+
+				# The 'handcode' attribute can be one of 'true',
+				# 'false', 'client', or 'server'.
+
+				handcode = child.nsProp( 'handcode', None )
+				if handcode == "false":
+					self.server_handcode = 0
+					self.client_handcode = 0
+				elif handcode == "true":
+					self.server_handcode = 1
+					self.client_handcode = 1
+				elif handcode == "client":
+					self.server_handcode = 0
+					self.client_handcode = 1
+				elif handcode == "server":
+					self.server_handcode = 1
+					self.client_handcode = 0
+				else:
+					raise RuntimeError('Invalid handcode mode "%s" in function "%s".' % (handcode, self.name))
+
+				self.ignore               = gl_XML.is_attr_true( child, 'ignore' )
+				self.can_be_large         = gl_XML.is_attr_true( child, 'large' )
+				self.glx_doubles_in_order = gl_XML.is_attr_true( child, 'doubles_in_order' )
+				self.reply_always_array   = gl_XML.is_attr_true( child, 'always_array' )
+				self.dimensions_in_reply  = gl_XML.is_attr_true( child, 'dimensions_in_reply' )
 
 			child = child.next
 
@@ -553,6 +557,7 @@ class glx_function_iterator:
 
 	def next(self):
 		f = self.iterator.next()
+
 		if f.client_supported_for_indirect():
 			return f
 		else:
diff --git a/src/mesa/glapi/glX_proto_recv.py b/src/mesa/glapi/glX_proto_recv.py
index 20f75575cf1..923c1958f0f 100644
--- a/src/mesa/glapi/glX_proto_recv.py
+++ b/src/mesa/glapi/glX_proto_recv.py
@@ -89,8 +89,8 @@ class PrintGlxDispatchFunctions(glX_proto_common.glx_print_proto):
 		print '#include "glxbyteorder.h"'
 		print '#include "indirect_util.h"'
 		print '#include "singlesize.h"'
-		print '#include "glapitable.h"'
 		print '#include "glapi.h"'
+		print '#include "glapitable.h"'
 		print '#include "glthread.h"'
 		print '#include "dispatch.h"'
 		print ''
@@ -225,6 +225,8 @@ class PrintGlxDispatchFunctions(glX_proto_common.glx_print_proto):
 		list = []
 
 		for param in f.parameterIterator():
+			if param.is_padding:
+				continue
 
 			if param.is_counter or param.is_image() or param.is_output or param.name in f.count_parameter_list or len(param.count_parameter_list):
 				location = param.name
diff --git a/src/mesa/glapi/glX_proto_send.py b/src/mesa/glapi/glX_proto_send.py
index 1b6a5f3890c..501706acc77 100644
--- a/src/mesa/glapi/glX_proto_send.py
+++ b/src/mesa/glapi/glX_proto_send.py
@@ -333,7 +333,7 @@ const GLuint __glXDefaultPixelStore[9] = { 0, 0, 0, 0, 0, 0, 0, 0, 1 };
 					if image.img_pad_dimensions:
 						do_it = 1
 						break
-			
+
 
 				if do_it:
 					[h, n] = hash_pixel_function(func)
@@ -373,7 +373,7 @@ const GLuint __glXDefaultPixelStore[9] = { 0, 0, 0, 0, 0, 0, 0, 0, 1 };
 				print '{'
 				print '    __GLXcontext * const gc = __glXGetCurrentContext();'
 				print ''
-				print '    if (gc->isDirect) {'
+				print '    if (gc->driContext) {'
 				print '    %sCALL_%s(GET_DISPATCH(), (%s));' % (ret_string, func.name, func.get_called_parameter_string())
 				print '    } else {'
 				footer = '}\n}\n'
@@ -422,7 +422,10 @@ generic_%u_byte( GLint rop, const void * ptr )
 		else:
 			src_ptr = "&" + p.name
 
-		if not extra_offset:
+		if p.is_padding:
+			print '(void) memset((void *)(%s + %u), 0, %s);' \
+			    % (pc, p.offset + adjust, p.size_string() )
+		elif not extra_offset:
 			print '(void) memcpy((void *)(%s + %u), (void *)(%s), %s);' \
 			    % (pc, p.offset + adjust, src_ptr, p.size_string() )
 		else:
@@ -472,6 +475,10 @@ generic_%u_byte( GLint rop, const void * ptr )
 				else:
 					dim_str = str(dim)
 
+				if param.is_padding:
+					print '(void) memset((void *)(%s + %u), 0, %s);' \
+					% (pc, (param.offset - 4) + adjust, param.size_string() )
+
 				if param.img_null_flag:
 					if large:
 						print '(void) memcpy((void *)(%s + %u), zero, 4);' % (pc, (param.offset - 4) + adjust)
@@ -511,7 +518,7 @@ generic_%u_byte( GLint rop, const void * ptr )
 		return
 
 
-	def common_func_print_just_start(self, f):
+	def common_func_print_just_start(self, f, name):
 		print '    __GLXcontext * const gc = __glXGetCurrentContext();'
 
 		# The only reason that single and vendor private commands need
@@ -529,7 +536,7 @@ generic_%u_byte( GLint rop, const void * ptr )
 
 		if not f.glx_rop:
 			for p in f.parameterIterateOutputs():
-				if p.is_image():
+				if p.is_image() and (p.img_format != "GL_COLOR_INDEX" or p.img_type != "GL_BITMAP"):
 					print '    const __GLXattribute * const state = gc->client_state_private;'
 					break
 
@@ -545,11 +552,23 @@ generic_%u_byte( GLint rop, const void * ptr )
 			print '    %s retval = (%s) 0;' % (f.return_type, f.return_type)
 
 
+		if name != None and name not in f.glx_vendorpriv_names:
+			print '#ifndef USE_XCB'
 		self.emit_packet_size_calculation(f, 0)
+		if name != None and name not in f.glx_vendorpriv_names:
+			print '#endif'
 
 		condition_list = []
 		for p in f.parameterIterateCounters():
 			condition_list.append( "%s >= 0" % (p.name) )
+			# 'counter' parameters cannot be negative
+			print "    if (%s < 0) {" % p.name
+			print "        __glXSetError(gc, GL_INVALID_VALUE);"
+			if f.return_type != 'void':
+				print "        return 0;"
+			else:
+				print "        return;"
+			print "    }"
 
 		if skip_condition:
 			condition_list.append( skip_condition )
@@ -567,7 +586,7 @@ generic_%u_byte( GLint rop, const void * ptr )
 
 
 	def printSingleFunction(self, f, name):
-		self.common_func_print_just_start(f)
+		self.common_func_print_just_start(f, name)
 
 		if self.debug:
 			print '        printf( "Enter %%s...\\n", "gl%s" );' % (f.name)
@@ -727,6 +746,9 @@ generic_%u_byte( GLint rop, const void * ptr )
 
 			p_string = ""
 			for param in f.parameterIterateGlxSend():
+				if param.is_padding:
+					continue
+
 				p_string += ", " + param.name
 
 				if param.is_image():
@@ -739,7 +761,7 @@ generic_%u_byte( GLint rop, const void * ptr )
 			return
 
 
-		if self.common_func_print_just_start(f):
+		if self.common_func_print_just_start(f, None):
 			trailer = "    }"
 		else:
 			trailer = None
@@ -791,7 +813,7 @@ generic_%u_byte( GLint rop, const void * ptr )
 					print '    generic_%u_byte( %s, %s );' % (cmdlen, f.opcode_real_name(), p.name)
 					return
 
-		if self.common_func_print_just_start(f):
+		if self.common_func_print_just_start(f, None):
 			trailer = "    }"
 		else:
 			trailer = None
diff --git a/src/mesa/glapi/glX_proto_size.py b/src/mesa/glapi/glX_proto_size.py
index 2b9a6433622..95cb5110cc3 100644
--- a/src/mesa/glapi/glX_proto_size.py
+++ b/src/mesa/glapi/glX_proto_size.py
@@ -581,6 +581,11 @@ class PrintGlxReqSize_c(PrintGlxReqSize_common):
 
 		self.common_emit_fixups(fixup)
 
+		if img.img_null_flag:
+			print ''
+			print '	   if (*(CARD32 *) (pc + %s))' % (img.offset - 4)
+			print '	       return 0;'
+
 		print ''
 		print '    return __glXImageSize(%s, %s, %s, %s, %s, %s,' % (img.img_format, img.img_type, img.img_target, w, h, d )
 		print '                          image_height, row_length, skip_images,'
diff --git a/src/mesa/glapi/gl_API.dtd b/src/mesa/glapi/gl_API.dtd
index f89d3818663..30c646c9244 100644
--- a/src/mesa/glapi/gl_API.dtd
+++ b/src/mesa/glapi/gl_API.dtd
@@ -45,6 +45,7 @@
                    counter             (true | false) "false"
                    count_scale         NMTOKEN "1"
                    output              (true | false) "false"
+                   padding             (true | false) "false"
                    img_width           NMTOKEN #IMPLIED
                    img_height          NMTOKEN #IMPLIED
                    img_depth           NMTOKEN #IMPLIED
diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml
index ef4a309cd6b..951fd957994 100644
--- a/src/mesa/glapi/gl_API.xml
+++ b/src/mesa/glapi/gl_API.xml
@@ -3267,7 +3267,8 @@
         <param name="width" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
-        <param name="pixels" type="const GLvoid *" img_width="width" img_xoff="xoffset" img_format="format" img_type="type" img_target="target" img_null_flag="true" img_pad_dimensions="true"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
+        <param name="pixels" type="const GLvoid *" img_width="width" img_xoff="xoffset" img_format="format" img_type="type" img_target="target" img_pad_dimensions="true"/>
         <glx rop="4099" large="true"/>
     </function>
 
@@ -3280,7 +3281,8 @@
         <param name="height" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
-        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_xoff="xoffset" img_yoff="yoffset" img_format="format" img_type="type" img_target="target" img_null_flag="true" img_pad_dimensions="true"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
+        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_xoff="xoffset" img_yoff="yoffset" img_format="format" img_type="type" img_target="target" img_pad_dimensions="true"/>
         <glx rop="4100" large="true"/>
     </function>
 
@@ -3289,7 +3291,7 @@
         <param name="textures" type="const GLuint *" count="n"/>
         <param name="residences" type="GLboolean *" output="true" count="n"/>
         <return type="GLboolean"/>
-        <glx sop="143" always_array="true"/>
+        <glx sop="143" handcode="client" always_array="true"/>
     </function>
 
     <function name="BindTexture" offset="307">
@@ -3380,7 +3382,9 @@
         <size name="GetTexParameterfv" mode="get"/>
         <size name="GetTexParameteriv" mode="get"/>
     </enum>
-    <enum name="MAX_3D_TEXTURE_SIZE"                      value="0x8073"/>
+    <enum name="MAX_3D_TEXTURE_SIZE"           count="1"  value="0x8073">
+        <size name="Get" mode="get"/>
+    </enum>
     <enum name="UNSIGNED_BYTE_2_3_3_REV"                  value="0x8362"/>
     <enum name="UNSIGNED_SHORT_5_6_5"                     value="0x8363"/>
     <enum name="UNSIGNED_SHORT_5_6_5_REV"                 value="0x8364"/>
@@ -3992,7 +3996,8 @@
         <param name="depth" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
-        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_depth="depth" img_xoff="xoffset" img_yoff="yoffset" img_zoff="zoffset" img_format="format" img_type="type" img_target="target" img_null_flag="true" img_pad_dimensions="true"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
+        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_depth="depth" img_xoff="xoffset" img_yoff="yoffset" img_zoff="zoffset" img_format="format" img_type="type" img_target="target" img_pad_dimensions="true"/>
         <glx rop="4115" large="true"/>
     </function>
 
@@ -4527,15 +4532,19 @@
     </enum>
     <enum name="POINT_SIZE_MIN"                count="1"  value="0x8126">
         <size name="PointParameterfvEXT"/>
+        <size name="Get" mode="get"/>
     </enum>
     <enum name="POINT_SIZE_MAX"                count="1"  value="0x8127">
         <size name="PointParameterfvEXT"/>
+        <size name="Get" mode="get"/>
     </enum>
     <enum name="POINT_FADE_THRESHOLD_SIZE"     count="1"  value="0x8128">
         <size name="PointParameterfvEXT"/>
+        <size name="Get" mode="get"/>
     </enum>
     <enum name="POINT_DISTANCE_ATTENUATION"    count="3"  value="0x8129">
         <size name="PointParameterfvEXT"/>
+        <size name="Get" mode="get"/>
     </enum>
     <enum name="GENERATE_MIPMAP"               count="1"  value="0x8191">
         <size name="TexParameterfv"/>
@@ -7285,27 +7294,59 @@
 </category>
 
 <category name="GL_ARB_vertex_buffer_object" number="28">
-    <enum name="BUFFER_SIZE_ARB"                          value="0x8764"/>
-    <enum name="BUFFER_USAGE_ARB"                         value="0x8765"/>
-    <enum name="ARRAY_BUFFER_ARB"                         value="0x8892"/>
-    <enum name="ELEMENT_ARRAY_BUFFER_ARB"                 value="0x8893"/>
-    <enum name="ARRAY_BUFFER_BINDING_ARB"                 value="0x8894"/>
-    <enum name="ELEMENT_ARRAY_BUFFER_BINDING_ARB"         value="0x8895"/>
-    <enum name="VERTEX_ARRAY_BUFFER_BINDING_ARB"          value="0x8896"/>
-    <enum name="NORMAL_ARRAY_BUFFER_BINDING_ARB"          value="0x8897"/>
-    <enum name="COLOR_ARRAY_BUFFER_BINDING_ARB"           value="0x8898"/>
-    <enum name="INDEX_ARRAY_BUFFER_BINDING_ARB"           value="0x8899"/>
-    <enum name="TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB"   value="0x889A"/>
-    <enum name="EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB"       value="0x889B"/>
-    <enum name="SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB" value="0x889C"/>
-    <enum name="FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB"  value="0x889D"/>
-    <enum name="WEIGHT_ARRAY_BUFFER_BINDING_ARB"          value="0x889E"/>
-    <enum name="VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB"   value="0x889F"/>
+    <enum name="BUFFER_SIZE_ARB"               count="1"  value="0x8764">
+        <size name="GetBufferParameterivARB" mode="get"/>
+    </enum>
+    <enum name="BUFFER_USAGE_ARB"              count="1"  value="0x8765">
+        <size name="GetBufferParameterivARB" mode="get"/>
+    </enum>
+    <enum name="ARRAY_BUFFER_BINDING_ARB"      count="1"  value="0x8894">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="ELEMENT_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x8895">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="VERTEX_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x8896">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="NORMAL_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x8897">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="COLOR_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x8898">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="INDEX_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x8899">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889A">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889B">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889C">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889D">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="WEIGHT_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889E">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB" count="1" value="0x889F">
+        <size name="GetVertexAttribdvARB" mode="get"/>
+        <size name="GetVertexAttribfvARB" mode="get"/>
+        <size name="GetVertexAttribivARB" mode="get"/>
+    </enum>
     <enum name="READ_ONLY_ARB"                            value="0x88B8"/>
     <enum name="WRITE_ONLY_ARB"                           value="0x88B9"/>
     <enum name="READ_WRITE_ARB"                           value="0x88BA"/>
-    <enum name="BUFFER_ACCESS_ARB"                        value="0x88BB"/>
-    <enum name="BUFFER_MAPPED_ARB"                        value="0x88BC"/>
+    <enum name="BUFFER_ACCESS_ARB"             count="1"  value="0x88BB">
+        <size name="GetBufferParameterivARB" mode="get"/>
+    </enum>
+    <enum name="BUFFER_MAPPED_ARB"             count="1"  value="0x88BC">
+        <size name="GetBufferParameterivARB" mode="get"/>
+    </enum>
     <enum name="BUFFER_MAP_POINTER_ARB"                   value="0x88BD"/>
     <enum name="STREAM_DRAW_ARB"                          value="0x88E0"/>
     <enum name="STREAM_READ_ARB"                          value="0x88E1"/>
@@ -7317,8 +7358,8 @@
     <enum name="DYNAMIC_READ_ARB"                         value="0x88E9"/>
     <enum name="DYNAMIC_COPY_ARB"                         value="0x88EA"/>
 
-    <type name="intptrARB"                   size="4"/>
-    <type name="sizeiptrARB" unsigned="true" size="4"/>
+    <type name="intptrARB"   size="4"                  glx_name="CARD32"/>
+    <type name="sizeiptrARB" size="4"  unsigned="true" glx_name="CARD32"/>
 
     <function name="BindBufferARB" offset="assign">
         <param name="target" type="GLenum"/>
@@ -7328,8 +7369,8 @@
 
     <function name="BufferDataARB" offset="assign">
         <param name="target" type="GLenum"/>
-        <param name="size" type="GLsizeiptrARB"/>
-        <param name="data" type="const GLvoid *"/>
+        <param name="size" type="GLsizeiptrARB" counter="true"/>
+        <param name="data" type="const GLvoid *" count="size" img_null_flag="true"/>
         <param name="usage" type="GLenum"/>
         <glx ignore="true"/>
     </function>
@@ -7337,14 +7378,14 @@
     <function name="BufferSubDataARB" offset="assign">
         <param name="target" type="GLenum"/>
         <param name="offset" type="GLintptrARB"/>
-        <param name="size" type="GLsizeiptrARB"/>
-        <param name="data" type="const GLvoid *"/>
+        <param name="size" type="GLsizeiptrARB" counter="true"/>
+        <param name="data" type="const GLvoid *" count="size"/>
         <glx ignore="true"/>
     </function>
 
     <function name="DeleteBuffersARB" offset="assign">
-        <param name="n" type="GLsizei"/>
-        <param name="buffer" type="const GLuint *"/>
+        <param name="n" type="GLsizei" counter="true"/>
+        <param name="buffer" type="const GLuint *" count="n"/>
         <glx ignore="true"/>
     </function>
 
@@ -7357,7 +7398,7 @@
     <function name="GetBufferParameterivARB" offset="assign">
         <param name="target" type="GLenum"/>
         <param name="pname" type="GLenum"/>
-        <param name="params" type="GLint *" output="true"/>
+        <param name="params" type="GLint *" output="true" variable_param="pname"/>
         <glx ignore="true"/>
     </function>
 
@@ -7371,8 +7412,8 @@
     <function name="GetBufferSubDataARB" offset="assign">
         <param name="target" type="GLenum"/>
         <param name="offset" type="GLintptrARB"/>
-        <param name="size" type="GLsizeiptrARB"/>
-        <param name="data" type="GLvoid *" output="true"/>
+        <param name="size" type="GLsizeiptrARB" counter="true"/>
+        <param name="data" type="GLvoid *" output="true" count="size"/>
         <glx ignore="true"/>
     </function>
 
@@ -8023,6 +8064,7 @@
         <param name="depth" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
         <param name="pixels" type="const GLvoid *"/>
     </function>
 </category>
@@ -8054,6 +8096,7 @@
         <param name="width" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
         <param name="pixels" type="const GLvoid *"/>
     </function>
 
@@ -8066,6 +8109,7 @@
         <param name="height" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
         <param name="pixels" type="const GLvoid *"/>
     </function>
 </category>
@@ -8589,7 +8633,8 @@
         <param name="size4d" type="GLsizei"/>
         <param name="format" type="GLenum"/>
         <param name="type" type="GLenum"/>
-        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_depth="depth" img_extent="size4d" img_xoff="xoffset" img_yoff="yoffset" img_zoff="zoffset" img_woff="woffset" img_format="format" img_type="type" img_target="target" img_null_flag="true" img_pad_dimensions="true"/>
+        <param name="UNUSED" type="GLuint" padding="true"/>
+        <param name="pixels" type="const GLvoid *" img_width="width" img_height="height" img_depth="depth" img_extent="size4d" img_xoff="xoffset" img_yoff="yoffset" img_zoff="zoffset" img_woff="woffset" img_format="format" img_type="type" img_target="target" img_pad_dimensions="true"/>
         <glx rop="2058" ignore="true"/>
     </function>
 </category>
@@ -8607,7 +8652,7 @@
         <param name="textures" type="const GLuint *" count="n"/>
         <param name="residences" type="GLboolean *" output="true" count="n"/>
         <return type="GLboolean"/>
-        <glx vendorpriv="11" always_array="true"/>
+        <glx vendorpriv="11" handcode="client" always_array="true"/>
     </function>
 
     <function name="BindTextureEXT" alias="BindTexture">
@@ -12129,8 +12174,12 @@
 <category name="GL_EXT_framebuffer_blit" number="316">
     <enum name="READ_FRAMEBUFFER_EXT" value="0x8CA8"/>
     <enum name="DRAW_FRAMEBUFFER_EXT" value="0x8CA9"/>
-    <enum name="DRAW_FRAMEBUFFER_BINDING_EXT" value="0x8CA6"/>
-    <enum name="READ_FRAMEBUFFER_BINDING_EXT" value="0x8CAA"/>
+    <enum name="DRAW_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CA6">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="READ_FRAMEBUFFER_BINDING_EXT" count="1" value="0x8CAA">
+        <size name="Get" mode="get"/>
+    </enum>
     <function name="BlitFramebufferEXT" offset="assign" static_dispatch="false">
         <param name="srcX0" type="GLint"/>
         <param name="srcY0" type="GLint"/>
diff --git a/src/mesa/glapi/gl_XML.py b/src/mesa/glapi/gl_XML.py
index b7a7388400d..b98919134fb 100644
--- a/src/mesa/glapi/gl_XML.py
+++ b/src/mesa/glapi/gl_XML.py
@@ -309,6 +309,9 @@ def create_parameter_string(parameters, include_names):
 
 	list = []
 	for p in parameters:
+		if p.is_padding:
+			continue
+
 		if include_names:
 			list.append( p.string() )
 		else:
@@ -463,6 +466,7 @@ class gl_parameter:
 		self.img_null_flag      = is_attr_true( element, 'img_null_flag' )
 		self.img_send_null      = is_attr_true( element, 'img_send_null' )
 
+		self.is_padding = is_attr_true( element, 'padding' )
 		return
 
 
diff --git a/src/mesa/glapi/gl_apitemp.py b/src/mesa/glapi/gl_apitemp.py
index 6e35571e143..a37c08d6ce1 100644
--- a/src/mesa/glapi/gl_apitemp.py
+++ b/src/mesa/glapi/gl_apitemp.py
@@ -63,6 +63,9 @@ class PrintGlOffsets(gl_XML.gl_print_base):
 		n = f.static_name(name)
 
 		for p in f.parameterIterator():
+			if p.is_padding:
+				continue
+
 			if p.is_pointer():
 				cast = "(const void *) "
 			else:
diff --git a/src/mesa/glapi/gl_enums.py b/src/mesa/glapi/gl_enums.py
index 0e3c57042a6..67fec7968a9 100644
--- a/src/mesa/glapi/gl_enums.py
+++ b/src/mesa/glapi/gl_enums.py
@@ -96,8 +96,10 @@ const char *_mesa_lookup_enum_by_nr( int nr )
 {
    unsigned * i;
 
-   i = (unsigned *)bsearch( & nr, reduced_enums, Elements(reduced_enums),
-                            sizeof(reduced_enums[0]), (cfunc) compar_nr );
+   i = (unsigned *) _mesa_bsearch(& nr, reduced_enums,
+                                  Elements(reduced_enums),
+                                  sizeof(reduced_enums[0]),
+                                  (cfunc) compar_nr);
 
    if ( i != NULL ) {
       return & enum_string_table[ all_enums[ *i ].offset ];
@@ -114,8 +116,10 @@ int _mesa_lookup_enum_by_name( const char *symbol )
    enum_elt * f = NULL;
 
    if ( symbol != NULL ) {
-      f = (enum_elt *)bsearch( symbol, all_enums, Elements(all_enums),
-			       sizeof( enum_elt ), (cfunc) compar_name );
+      f = (enum_elt *) _mesa_bsearch(symbol, all_enums,
+                                     Elements(all_enums),
+                                     sizeof( enum_elt ),
+                                     (cfunc) compar_name);
    }
 
    return (f != NULL) ? f->n : -1;
diff --git a/src/mesa/glapi/gl_table.py b/src/mesa/glapi/gl_table.py
index 7023a4b71a3..55a33748aed 100644
--- a/src/mesa/glapi/gl_table.py
+++ b/src/mesa/glapi/gl_table.py
@@ -78,7 +78,9 @@ class PrintRemapTable(gl_XML.gl_print_base):
 
 
 	def printRealHeader(self):
-		print """/**
+		print """
+#include "glapitable.h"
+/**
  * \\file dispatch.h
  * Macros for handling GL dispatch tables.
  *
diff --git a/src/mesa/glapi/gl_x86_asm.py b/src/mesa/glapi/gl_x86_asm.py
index 651cb03f14d..0dbf3ebe0ab 100644
--- a/src/mesa/glapi/gl_x86_asm.py
+++ b/src/mesa/glapi/gl_x86_asm.py
@@ -44,6 +44,9 @@ class PrintGenericStubs(gl_XML.gl_print_base):
 	def get_stack_size(self, f):
 		size = 0
 		for p in f.parameterIterator():
+			if p.is_padding:
+				continue
+
 			size += p.get_stack_size()
 
 		return size
diff --git a/src/mesa/glapi/glapi.c b/src/mesa/glapi/glapi.c
index 53efd7eef47..9b5144a88b6 100644
--- a/src/mesa/glapi/glapi.c
+++ b/src/mesa/glapi/glapi.c
@@ -50,7 +50,23 @@
 
 
 
+#ifdef HAVE_DIX_CONFIG_H
+
+#include <dix-config.h>
+#define PUBLIC
+
+#else
+
 #include "main/glheader.h"
+
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+#ifdef DEBUG
+#include <assert.h>
+#endif
+
 #include "glapi.h"
 #include "glapioffsets.h"
 #include "glapitable.h"
@@ -278,8 +294,32 @@ _glapi_get_context(void)
 #endif
 }
 
+#ifdef USE_X86_ASM
+
+#if defined( GLX_USE_TLS )
+extern       GLubyte gl_dispatch_functions_start[];
+extern       GLubyte gl_dispatch_functions_end[];
+#else
+extern const GLubyte gl_dispatch_functions_start[];
+#endif
+
+#endif /* USE_X86_ASM */
 
 
+#if defined(USE_X64_64_ASM) && defined(GLX_USE_TLS)
+# define DISPATCH_FUNCTION_SIZE  16
+#elif defined(USE_X86_ASM)
+# if defined(THREADS) && !defined(GLX_USE_TLS)
+#  define DISPATCH_FUNCTION_SIZE  32
+# else
+#  define DISPATCH_FUNCTION_SIZE  16
+# endif
+#endif
+
+#if !defined(DISPATCH_FUNCTION_SIZE) && !defined(XFree86Server) && !defined(XGLServer)
+# define NEED_FUNCTION_POINTER
+#endif
+
 #if defined(PTHREADS) || defined(GLX_USE_TLS)
 /**
  * Perform platform-specific GL API entry-point fixups.
diff --git a/src/mesa/glapi/glthread.c b/src/mesa/glapi/glthread.c
index 09cc8cfcde2..56ddf7c25ac 100644
--- a/src/mesa/glapi/glthread.c
+++ b/src/mesa/glapi/glthread.c
@@ -28,6 +28,9 @@
  * truly reusable outside of Mesa.  First, the glheader.h include must go.
  */
 
+#ifdef HAVE_DIX_CONFIG_H
+#include <dix-config.h>
+#endif
 
 #include "main/glheader.h"
 #include "glthread.h"
diff --git a/src/mesa/glapi/sources b/src/mesa/glapi/sources
deleted file mode 100644
index 1d5c2528211..00000000000
--- a/src/mesa/glapi/sources
+++ /dev/null
@@ -1,12 +0,0 @@
-MESA_GLAPI_SOURCES = \
-glapi.c \
-glthread.c
-
-MESA_GLAPI_HEADERS = \
-dispatch.h \
-glapi.h \
-glapioffsets.h \
-glapitable.h \
-glapitemp.h \
-glprocs.h \
-glthread.h
diff --git a/src/mesa/main/api_loopback.c b/src/mesa/main/api_loopback.c
index 924d7134a27..0e3f5ff9570 100644
--- a/src/mesa/main/api_loopback.c
+++ b/src/mesa/main/api_loopback.c
@@ -31,7 +31,6 @@
 
 #include "glheader.h"
 #include "macros.h"
-#include "colormac.h"
 #include "api_loopback.h"
 #include "mtypes.h"
 #include "glapi/glapi.h"
diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c
index 2eb96ae3f4e..bbc5933ab9f 100644
--- a/src/mesa/main/api_validate.c
+++ b/src/mesa/main/api_validate.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.0.1
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -30,6 +30,55 @@
 #include "state.h"
 
 
+/**
+ * Find the max index in the given element/index buffer
+ */
+static GLuint
+max_buffer_index(GLcontext *ctx, GLuint count, GLenum type,
+                 const void *indices,
+                 struct gl_buffer_object *elementBuf)
+{
+   const GLubyte *map = NULL;
+   GLuint max = 0;
+   GLint i;
+
+   if (elementBuf->Name) {
+      /* elements are in a user-defined buffer object.  need to map it */
+      map = ctx->Driver.MapBuffer(ctx,
+                                  GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                  GL_READ_ONLY,
+                                  elementBuf);
+      /* Actual address is the sum of pointers */
+      indices = (const GLvoid *) ADD_POINTERS(map, (const GLubyte *) indices);
+   }
+
+   if (type == GL_UNSIGNED_INT) {
+      for (i = 0; i < count; i++)
+         if (((GLuint *) indices)[i] > max)
+            max = ((GLuint *) indices)[i];
+   }
+   else if (type == GL_UNSIGNED_SHORT) {
+      for (i = 0; i < count; i++)
+         if (((GLushort *) indices)[i] > max)
+            max = ((GLushort *) indices)[i];
+   }
+   else {
+      ASSERT(type == GL_UNSIGNED_BYTE);
+      for (i = 0; i < count; i++)
+         if (((GLubyte *) indices)[i] > max)
+            max = ((GLubyte *) indices)[i];
+   }
+
+   if (map) {
+      ctx->Driver.UnmapBuffer(ctx,
+                              GL_ELEMENT_ARRAY_BUFFER_ARB,
+                              ctx->Array.ElementArrayBufferObj);
+   }
+
+   return max;
+}
+
+
 GLboolean
 _mesa_validate_DrawElements(GLcontext *ctx,
 			    GLenum mode, GLsizei count, GLenum type,
@@ -67,16 +116,15 @@ _mesa_validate_DrawElements(GLcontext *ctx,
 
    /* Vertex buffer object tests */
    if (ctx->Array.ElementArrayBufferObj->Name) {
+      /* use indices in the buffer object */
       GLuint indexBytes;
 
-      /* use indices in the buffer object */
       if (!ctx->Array.ElementArrayBufferObj->Size) {
          _mesa_warning(ctx,
                        "glDrawElements called with empty array elements buffer");
          return GL_FALSE;
       }
 
-      /* make sure count doesn't go outside buffer bounds */
       if (type == GL_UNSIGNED_INT) {
          indexBytes = count * sizeof(GLuint);
       }
@@ -88,6 +136,7 @@ _mesa_validate_DrawElements(GLcontext *ctx,
          indexBytes = count * sizeof(GLushort);
       }
 
+      /* make sure count doesn't go outside buffer bounds */
       if (indexBytes > (GLuint) ctx->Array.ElementArrayBufferObj->Size) {
          _mesa_warning(ctx, "glDrawElements index out of buffer bounds");
          return GL_FALSE;
@@ -101,39 +150,8 @@ _mesa_validate_DrawElements(GLcontext *ctx,
 
    if (ctx->Const.CheckArrayBounds) {
       /* find max array index */
-      const GLubyte *map;
-      GLuint max = 0;
-      GLint i;
-
-      map = ctx->Driver.MapBuffer(ctx,
-                                  GL_ELEMENT_ARRAY_BUFFER_ARB,
-                                  GL_READ_ONLY,
-                                  ctx->Array.ElementArrayBufferObj);
-
-      /* Actual address is the sum of pointers */
-      indices = (const GLvoid *) ADD_POINTERS(map, (const GLubyte *) indices);
-
-      if (type == GL_UNSIGNED_INT) {
-         for (i = 0; i < count; i++)
-            if (((GLuint *) indices)[i] > max)
-               max = ((GLuint *) indices)[i];
-      }
-      else if (type == GL_UNSIGNED_SHORT) {
-         for (i = 0; i < count; i++)
-            if (((GLushort *) indices)[i] > max)
-               max = ((GLushort *) indices)[i];
-      }
-      else {
-         ASSERT(type == GL_UNSIGNED_BYTE);
-         for (i = 0; i < count; i++)
-            if (((GLubyte *) indices)[i] > max)
-               max = ((GLubyte *) indices)[i];
-      }
-
-      ctx->Driver.UnmapBuffer(ctx,
-                              GL_ELEMENT_ARRAY_BUFFER_ARB,
-                              ctx->Array.ElementArrayBufferObj);
-
+      GLuint max = max_buffer_index(ctx, count, type, indices,
+                                    ctx->Array.ElementArrayBufferObj);
       if (max >= ctx->Array._MaxElement) {
          /* the max element is out of bounds of one or more enabled arrays */
          return GL_FALSE;
@@ -186,36 +204,35 @@ _mesa_validate_DrawRangeElements(GLcontext *ctx, GLenum mode,
 
    /* Vertex buffer object tests */
    if (ctx->Array.ElementArrayBufferObj->Name) {
-      /* XXX re-use code from above? */
+      /* use indices in the buffer object */
+      GLuint indexBytes;
+
+      if (type == GL_UNSIGNED_INT) {
+         indexBytes = count * sizeof(GLuint);
+      }
+      else if (type == GL_UNSIGNED_BYTE) {
+         indexBytes = count * sizeof(GLubyte);
+      }
+      else {
+         ASSERT(type == GL_UNSIGNED_SHORT);
+         indexBytes = count * sizeof(GLushort);
+      }
+
+      /* make sure count doesn't go outside buffer bounds */
+      if (indexBytes > ctx->Array.ElementArrayBufferObj->Size) {
+         _mesa_warning(ctx, "glDrawRangeElements index out of buffer bounds");
+         return GL_FALSE;
+      }
    }
    else {
-      /* not using VBO */
+      /* not using a VBO */
       if (!indices)
          return GL_FALSE;
    }
 
    if (ctx->Const.CheckArrayBounds) {
-      /* Find max array index.
-       * We don't trust the user's start and end values.
-       */
-      GLuint max = 0;
-      GLint i;
-      if (type == GL_UNSIGNED_INT) {
-         for (i = 0; i < count; i++)
-            if (((GLuint *) indices)[i] > max)
-               max = ((GLuint *) indices)[i];
-      }
-      else if (type == GL_UNSIGNED_SHORT) {
-         for (i = 0; i < count; i++)
-            if (((GLushort *) indices)[i] > max)
-               max = ((GLushort *) indices)[i];
-      }
-      else {
-         ASSERT(type == GL_UNSIGNED_BYTE);
-         for (i = 0; i < count; i++)
-            if (((GLubyte *) indices)[i] > max)
-               max = ((GLubyte *) indices)[i];
-      }
+      GLuint max = max_buffer_index(ctx, count, type, indices,
+                                    ctx->Array.ElementArrayBufferObj);
       if (max >= ctx->Array._MaxElement) {
          /* the max element is out of bounds of one or more enabled arrays */
          return GL_FALSE;
@@ -236,8 +253,9 @@ _mesa_validate_DrawArrays(GLcontext *ctx,
 {
    ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);
 
-   if (count < 0) {
-      _mesa_error(ctx, GL_INVALID_VALUE, "glDrawArrays(count)" );
+   if (count <= 0) {
+      if (count < 0)
+         _mesa_error(ctx, GL_INVALID_VALUE, "glDrawArrays(count)" );
       return GL_FALSE;
    }
 
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index 7179fba3db3..dc85da2518f 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -53,9 +53,27 @@
 #include "texobj.h"
 #include "texparam.h"
 #include "texstate.h"
+#include "varray.h"
 #include "mtypes.h"
 #include "math/m_xform.h"
 
+/**
+ * Special struct for saving/restoring texture state (GL_TEXTURE_BIT)
+ */
+struct texture_state
+{
+   struct gl_texture_attrib Texture;  /**< The usual context state */
+
+   /** to save per texture object state (wrap modes, filters, etc): */
+   struct gl_texture_object SavedObj[MAX_TEXTURE_UNITS][NUM_TEXTURE_TARGETS];
+
+   /**
+    * To save references to texture objects (so they don't get accidentally
+    * deleted while saved in the attribute stack).
+    */
+   struct gl_texture_object *SavedTexRef[MAX_TEXTURE_UNITS][NUM_TEXTURE_TARGETS];
+};
+
 
 /**
  * Allocate a new attribute state node.  These nodes have a
@@ -345,47 +363,61 @@ _mesa_PushAttrib(GLbitfield mask)
    }
 
    if (mask & GL_TEXTURE_BIT) {
-      struct gl_texture_attrib *attr;
+      struct texture_state *texstate = CALLOC_STRUCT(texture_state);
       GLuint u;
 
+      if (!texstate) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glPushAttrib(GL_TEXTURE_BIT)");
+         goto end;
+      }
+
       _mesa_lock_context_textures(ctx);
-      /* Bump the texture object reference counts so that they don't
-       * inadvertantly get deleted.
+
+      /* copy/save the bulk of texture state here */
+      _mesa_memcpy(&texstate->Texture, &ctx->Texture, sizeof(ctx->Texture));
+
+      /* Save references to the currently bound texture objects so they don't
+       * accidentally get deleted while referenced in the attribute stack.
        */
       for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-	 ctx->Texture.Unit[u].Current1D->RefCount++;
-	 ctx->Texture.Unit[u].Current2D->RefCount++;
-	 ctx->Texture.Unit[u].Current3D->RefCount++;
-	 ctx->Texture.Unit[u].CurrentCubeMap->RefCount++;
-	 ctx->Texture.Unit[u].CurrentRect->RefCount++;
-	 ctx->Texture.Unit[u].Current1DArray->RefCount++;
-	 ctx->Texture.Unit[u].Current2DArray->RefCount++;
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_1D_INDEX],
+                                ctx->Texture.Unit[u].Current1D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_2D_INDEX],
+                                ctx->Texture.Unit[u].Current2D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_3D_INDEX],
+                                ctx->Texture.Unit[u].Current3D);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_CUBE_INDEX],
+                                ctx->Texture.Unit[u].CurrentCubeMap);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_RECT_INDEX],
+                                ctx->Texture.Unit[u].CurrentRect);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_1D_ARRAY_INDEX],
+                                ctx->Texture.Unit[u].Current1DArray);
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][TEXTURE_2D_ARRAY_INDEX],
+                                ctx->Texture.Unit[u].Current2DArray);
       }
 
-      attr = MALLOC_STRUCT( gl_texture_attrib );
-      MEMCPY( attr, &ctx->Texture, sizeof(struct gl_texture_attrib) );
-      /* copy state of the currently bound texture objects */
+      /* copy state/contents of the currently bound texture objects */
       for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-         _mesa_copy_texture_object(&attr->Unit[u].Saved1D,
-                                   attr->Unit[u].Current1D);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved2D,
-                                   attr->Unit[u].Current2D);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved3D,
-                                   attr->Unit[u].Current3D);
-         _mesa_copy_texture_object(&attr->Unit[u].SavedCubeMap,
-                                   attr->Unit[u].CurrentCubeMap);
-         _mesa_copy_texture_object(&attr->Unit[u].SavedRect,
-                                   attr->Unit[u].CurrentRect);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved1DArray,
-                                   attr->Unit[u].Current1DArray);
-         _mesa_copy_texture_object(&attr->Unit[u].Saved2DArray,
-                                   attr->Unit[u].Current2DArray);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_1D_INDEX],
+                                   ctx->Texture.Unit[u].Current1D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_2D_INDEX],
+                                   ctx->Texture.Unit[u].Current2D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_3D_INDEX],
+                                   ctx->Texture.Unit[u].Current3D);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_CUBE_INDEX],
+                                   ctx->Texture.Unit[u].CurrentCubeMap);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_RECT_INDEX],
+                                   ctx->Texture.Unit[u].CurrentRect);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_1D_ARRAY_INDEX],
+                                   ctx->Texture.Unit[u].Current1DArray);
+         _mesa_copy_texture_object(&texstate->SavedObj[u][TEXTURE_2D_ARRAY_INDEX],
+                                   ctx->Texture.Unit[u].Current2DArray);
       }
 
       _mesa_unlock_context_textures(ctx);
 
       newnode = new_attrib_node( GL_TEXTURE_BIT );
-      newnode->data = attr;
+      newnode->data = texstate;
       newnode->next = head;
       head = newnode;
    }
@@ -421,6 +453,7 @@ _mesa_PushAttrib(GLbitfield mask)
       head = newnode;
    }
 
+end:
    ctx->AttribStack[ctx->AttribStackDepth] = head;
    ctx->AttribStackDepth++;
 }
@@ -630,14 +663,19 @@ pop_enable_group(GLcontext *ctx, const struct gl_enable_attrib *enable)
 }
 
 
+/**
+ * Pop/restore texture attribute/group state.
+ */
 static void
-pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
+pop_texture_group(GLcontext *ctx, struct texture_state *texstate)
 {
    GLuint u;
 
+   _mesa_lock_context_textures(ctx);
+
    for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-      const struct gl_texture_unit *unit = &texAttrib->Unit[u];
-      GLuint i;
+      const struct gl_texture_unit *unit = &texstate->Texture.Unit[u];
+      GLuint tgt;
 
       _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + u);
       _mesa_set_enable(ctx, GL_TEXTURE_1D,
@@ -730,53 +768,33 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
                        1 << unit->Combine.ScaleShiftA);
       }
 
-      /* Restore texture object state */
-      for (i = 0; i < NUM_TEXTURE_TARGETS; i++) {
-         GLenum target = 0;
+      /* Restore texture object state for each target */
+      for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
          const struct gl_texture_object *obj = NULL;
          GLfloat bordColor[4];
+         GLenum target;
 
-         switch (i) {
-         case 0:
-            target = GL_TEXTURE_1D;
-            obj = &unit->Saved1D;
-            break;
-         case 1:
-            target = GL_TEXTURE_2D;
-            obj = &unit->Saved2D;
-            break;
-         case 2:
-            target = GL_TEXTURE_3D;
-            obj = &unit->Saved3D;
-            break;
-         case 3:
-            if (!ctx->Extensions.ARB_texture_cube_map)
-               continue;
-            target = GL_TEXTURE_CUBE_MAP_ARB;
-            obj = &unit->SavedCubeMap;
-            break;
-         case 4:
-            if (!ctx->Extensions.NV_texture_rectangle)
-               continue;
-            target = GL_TEXTURE_RECTANGLE_NV;
-            obj = &unit->SavedRect;
-            break;
-         case 5:
-            if (!ctx->Extensions.MESA_texture_array)
-               continue;
-            target = GL_TEXTURE_1D_ARRAY_EXT;
-            obj = &unit->Saved1DArray;
-            break;
-         case 6:
-            if (!ctx->Extensions.MESA_texture_array)
-               continue;
-            target = GL_TEXTURE_2D_ARRAY_EXT;
-            obj = &unit->Saved2DArray;
-            break;
-         default:
-            ; /* silence warnings */
+         obj = &texstate->SavedObj[u][tgt];
+
+         /* don't restore state for unsupported targets to prevent
+          * raising GL errors.
+          */
+         if (obj->Target == GL_TEXTURE_CUBE_MAP_ARB &&
+             !ctx->Extensions.ARB_texture_cube_map) {
+            continue;
+         }
+         else if (obj->Target == GL_TEXTURE_RECTANGLE_NV &&
+                  !ctx->Extensions.NV_texture_rectangle) {
+            continue;
+         }
+         else if ((obj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
+                   obj->Target == GL_TEXTURE_2D_ARRAY_EXT) &&
+                  !ctx->Extensions.MESA_texture_array) {
+            continue;
          }
 
+         target = obj->Target;
+
          _mesa_BindTexture(target, obj->Name);
 
          bordColor[0] = CHAN_TO_FLOAT(obj->BorderColor[0]);
@@ -784,8 +802,8 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
          bordColor[2] = CHAN_TO_FLOAT(obj->BorderColor[2]);
          bordColor[3] = CHAN_TO_FLOAT(obj->BorderColor[3]);
 
-         _mesa_TexParameterf(target, GL_TEXTURE_PRIORITY, obj->Priority);
          _mesa_TexParameterfv(target, GL_TEXTURE_BORDER_COLOR, bordColor);
+         _mesa_TexParameterf(target, GL_TEXTURE_PRIORITY, obj->Priority);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, obj->WrapS);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, obj->WrapT);
          _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, obj->WrapR);
@@ -811,25 +829,17 @@ pop_texture_group(GLcontext *ctx, const struct gl_texture_attrib *texAttrib)
             _mesa_TexParameterf(target, GL_SHADOW_AMBIENT_SGIX,
                                 obj->ShadowAmbient);
          }
+      }
 
+      /* remove saved references to the texture objects */
+      for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
+         _mesa_reference_texobj(&texstate->SavedTexRef[u][tgt], NULL);
       }
    }
-   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB
-                          + texAttrib->CurrentUnit);
 
-   /* "un-bump" the texture object reference counts.  We did that so they
-    * wouldn't inadvertantly get deleted while they were still referenced
-    * inside the attribute state stack.
-    */
-   for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
-      ctx->Texture.Unit[u].Current1D->RefCount--;
-      ctx->Texture.Unit[u].Current2D->RefCount--;
-      ctx->Texture.Unit[u].Current3D->RefCount--;
-      ctx->Texture.Unit[u].CurrentCubeMap->RefCount--;
-      ctx->Texture.Unit[u].CurrentRect->RefCount--;
-      ctx->Texture.Unit[u].Current1DArray->RefCount--;
-      ctx->Texture.Unit[u].Current2DArray->RefCount--;
-   }
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + texstate->Texture.CurrentUnit);
+
+   _mesa_unlock_context_textures(ctx);
 }
 
 
@@ -1098,8 +1108,9 @@ _mesa_PopAttrib(void)
                                    (GLint) point->CoordReplace[u]);
                   }
                   _mesa_set_enable(ctx, GL_POINT_SPRITE_NV,point->PointSprite);
-                  _mesa_PointParameteri(GL_POINT_SPRITE_R_MODE_NV,
-                                        ctx->Point.SpriteRMode);
+                  if (ctx->Extensions.NV_point_sprite)
+                     _mesa_PointParameteri(GL_POINT_SPRITE_R_MODE_NV,
+                                           ctx->Point.SpriteRMode);
                   _mesa_PointParameterf(GL_POINT_SPRITE_COORD_ORIGIN,
                                         (GLfloat)ctx->Point.SpriteOrigin);
                }
@@ -1208,9 +1219,9 @@ _mesa_PopAttrib(void)
          case GL_TEXTURE_BIT:
             /* Take care of texture object reference counters */
             {
-               const struct gl_texture_attrib *texture;
-               texture = (const struct gl_texture_attrib *) attr->data;
-               pop_texture_group(ctx, texture);
+               struct texture_state *texstate
+                  = (struct texture_state *) attr->data;
+               pop_texture_group(ctx, texstate);
 	       ctx->NewState |= _NEW_TEXTURE;
             }
             break;
@@ -1409,8 +1420,10 @@ _mesa_PopClientAttrib(void)
             adjust_buffer_object_ref_counts(&ctx->Array, -1);
 	 
             ctx->Array.ActiveTexture = data->ActiveTexture;
-	    ctx->Array.LockFirst = data->LockFirst;
-	    ctx->Array.LockCount = data->LockCount;
+	    if (data->LockCount != 0)
+	       _mesa_LockArraysEXT(data->LockFirst, data->LockCount);
+	    else if (ctx->Array.LockCount)
+	       _mesa_UnlockArraysEXT();
 
 	    _mesa_BindVertexArrayAPPLE( data->ArrayObj->Name );
 	    
@@ -1447,6 +1460,42 @@ _mesa_PopClientAttrib(void)
 }
 
 
+/**
+ * Free any attribute state data that might be attached to the context.
+ */
+void
+_mesa_free_attrib_data(GLcontext *ctx)
+{
+   while (ctx->AttribStackDepth > 0) {
+      struct gl_attrib_node *attr, *next;
+
+      ctx->AttribStackDepth--;
+      attr = ctx->AttribStack[ctx->AttribStackDepth];
+
+      while (attr) {
+         if (attr->kind == GL_TEXTURE_BIT) {
+            struct texture_state *texstate = (struct texture_state*)attr->data;
+            GLuint u, tgt;
+            /* clear references to the saved texture objects */
+            for (u = 0; u < ctx->Const.MaxTextureUnits; u++) {
+               for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) {
+                  _mesa_reference_texobj(&texstate->SavedTexRef[u][tgt], NULL);
+               }
+            }
+         }
+         else {
+            /* any other chunks of state that requires special handling? */
+         }
+
+         next = attr->next;
+         _mesa_free(attr->data);
+         _mesa_free(attr);
+         attr = next;
+      }
+   }
+}
+
+
 void _mesa_init_attrib( GLcontext *ctx )
 {
    /* Renderer and client attribute stacks */
diff --git a/src/mesa/main/attrib.h b/src/mesa/main/attrib.h
index 09d75196b23..2cf8fe69345 100644
--- a/src/mesa/main/attrib.h
+++ b/src/mesa/main/attrib.h
@@ -1,18 +1,8 @@
-/**
- * \file attrib.h
- * Attribute stacks.
- * 
- * \if subset
- * (No-op)
- *
- * \endif
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  5.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -32,8 +22,6 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
-
-
 #ifndef ATTRIB_H
 #define ATTRIB_H
 
@@ -58,10 +46,14 @@ _mesa_PopClientAttrib( void );
 extern void 
 _mesa_init_attrib( GLcontext *ctx );
 
+extern void 
+_mesa_free_attrib_data( GLcontext *ctx );
+
 #else
 
 /** No-op */
 #define _mesa_init_attrib( c ) ((void)0)
+#define _mesa_free_attrib_data( c ) ((void)0)
 
 #endif
 
diff --git a/src/mesa/main/blend.c b/src/mesa/main/blend.c
index 4d4a8971419..39cf6153e28 100644
--- a/src/mesa/main/blend.c
+++ b/src/mesa/main/blend.c
@@ -31,7 +31,6 @@
 
 #include "glheader.h"
 #include "blend.h"
-#include "colormac.h"
 #include "context.h"
 #include "enums.h"
 #include "macros.h"
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index dd4ac4679e2..190e6ab5641 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -446,6 +446,89 @@ _mesa_init_buffer_objects( GLcontext *ctx )
    ctx->Array.ElementArrayBufferObj = ctx->Array.NullBufferObj;
 }
 
+/**
+ * Bind the specified target to buffer for the specified context.
+ */
+static void
+bind_buffer_object(GLcontext *ctx, GLenum target, GLuint buffer)
+{
+   struct gl_buffer_object *oldBufObj;
+   struct gl_buffer_object *newBufObj = NULL;
+   struct gl_buffer_object **bindTarget = NULL;
+
+   switch (target) {
+   case GL_ARRAY_BUFFER_ARB:
+      bindTarget = &ctx->Array.ArrayBufferObj;
+      break;
+   case GL_ELEMENT_ARRAY_BUFFER_ARB:
+      bindTarget = &ctx->Array.ElementArrayBufferObj;
+      break;
+   case GL_PIXEL_PACK_BUFFER_EXT:
+      bindTarget = &ctx->Pack.BufferObj;
+      break;
+   case GL_PIXEL_UNPACK_BUFFER_EXT:
+      bindTarget = &ctx->Unpack.BufferObj;
+      break;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glBindBufferARB(target)");
+      return;
+   }
+
+   /* Get pointer to old buffer object (to be unbound) */
+   oldBufObj = get_buffer(ctx, target);
+   if (oldBufObj && oldBufObj->Name == buffer)
+      return;   /* rebinding the same buffer object- no change */
+
+   /*
+    * Get pointer to new buffer object (newBufObj)
+    */
+   if (buffer == 0) {
+      /* The spec says there's not a buffer object named 0, but we use
+       * one internally because it simplifies things.
+       */
+      newBufObj = ctx->Array.NullBufferObj;
+   }
+   else {
+      /* non-default buffer object */
+      newBufObj = _mesa_lookup_bufferobj(ctx, buffer);
+      if (!newBufObj) {
+         /* if this is a new buffer object id, allocate a buffer object now */
+         ASSERT(ctx->Driver.NewBufferObject);
+         newBufObj = ctx->Driver.NewBufferObject(ctx, buffer, target);
+         if (!newBufObj) {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBindBufferARB");
+            return;
+         }
+         _mesa_HashInsert(ctx->Shared->BufferObjects, buffer, newBufObj);
+      }
+   }
+   
+   /* bind new buffer */
+   _mesa_reference_buffer_object(ctx, bindTarget, newBufObj);
+
+   /* Pass BindBuffer call to device driver */
+   if (ctx->Driver.BindBuffer && newBufObj)
+      ctx->Driver.BindBuffer( ctx, target, newBufObj );
+}
+
+
+/**
+ * Update the default buffer objects in the given context to reference those
+ * specified in the shared state and release those referencing the old 
+ * shared state.
+ */
+void
+_mesa_update_default_objects_buffer_objects(GLcontext *ctx)
+{
+   /* Bind the NullBufferObj to remove references to those
+    * in the shared context hash table.
+    */
+   bind_buffer_object( ctx, GL_ARRAY_BUFFER_ARB, 0);
+   bind_buffer_object( ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, 0);
+   bind_buffer_object( ctx, GL_PIXEL_PACK_BUFFER_ARB, 0);
+   bind_buffer_object( ctx, GL_PIXEL_UNPACK_BUFFER_ARB, 0);
+}
+
 
 /**
  * When we're about to read pixel data out of a PBO (via glDrawPixels,
@@ -684,64 +767,9 @@ void GLAPIENTRY
 _mesa_BindBufferARB(GLenum target, GLuint buffer)
 {
    GET_CURRENT_CONTEXT(ctx);
-   struct gl_buffer_object *oldBufObj;
-   struct gl_buffer_object *newBufObj = NULL;
-   struct gl_buffer_object **bindTarget = NULL;
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
-   switch (target) {
-      case GL_ARRAY_BUFFER_ARB:
-         bindTarget = &ctx->Array.ArrayBufferObj;
-         break;
-      case GL_ELEMENT_ARRAY_BUFFER_ARB:
-         bindTarget = &ctx->Array.ElementArrayBufferObj;
-         break;
-      case GL_PIXEL_PACK_BUFFER_EXT:
-         bindTarget = &ctx->Pack.BufferObj;
-         break;
-      case GL_PIXEL_UNPACK_BUFFER_EXT:
-         bindTarget = &ctx->Unpack.BufferObj;
-         break;
-      default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glBindBufferARB(target)");
-         return;
-   }
-
-   /* Get pointer to old buffer object (to be unbound) */
-   oldBufObj = get_buffer(ctx, target);
-   if (oldBufObj && oldBufObj->Name == buffer)
-      return;   /* rebinding the same buffer object- no change */
-
-   /*
-    * Get pointer to new buffer object (newBufObj)
-    */
-   if (buffer == 0) {
-      /* The spec says there's not a buffer object named 0, but we use
-       * one internally because it simplifies things.
-       */
-      newBufObj = ctx->Array.NullBufferObj;
-   }
-   else {
-      /* non-default buffer object */
-      newBufObj = _mesa_lookup_bufferobj(ctx, buffer);
-      if (!newBufObj) {
-         /* if this is a new buffer object id, allocate a buffer object now */
-         ASSERT(ctx->Driver.NewBufferObject);
-	 newBufObj = ctx->Driver.NewBufferObject(ctx, buffer, target);
-         if (!newBufObj) {
-            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBindBufferARB");
-            return;
-         }
-          _mesa_HashInsert(ctx->Shared->BufferObjects, buffer, newBufObj);
-      }
-   }
-   
-   /* bind new buffer */
-   _mesa_reference_buffer_object(ctx, bindTarget, newBufObj);
-
-   /* Pass BindBuffer call to device driver */
-   if (ctx->Driver.BindBuffer && newBufObj)
-      ctx->Driver.BindBuffer( ctx, target, newBufObj );
+   bind_buffer_object(ctx, target, buffer);
 }
 
 
@@ -801,7 +829,7 @@ _mesa_DeleteBuffersARB(GLsizei n, const GLuint *ids)
             _mesa_BindBufferARB( GL_PIXEL_UNPACK_BUFFER_EXT, 0 );
          }
 
-	 /* The ID is immediately freed for re-use */
+         /* The ID is immediately freed for re-use */
          _mesa_HashRemove(ctx->Shared->BufferObjects, bufObj->Name);
          _mesa_reference_buffer_object(ctx, &bufObj, NULL);
       }
diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h
index 2f908c5c35d..3c08f0083cf 100644
--- a/src/mesa/main/bufferobj.h
+++ b/src/mesa/main/bufferobj.h
@@ -38,6 +38,9 @@
 extern void
 _mesa_init_buffer_objects( GLcontext *ctx );
 
+extern void
+_mesa_update_default_objects_buffer_objects(GLcontext *ctx);
+
 extern struct gl_buffer_object *
 _mesa_new_buffer_object( GLcontext *ctx, GLuint name, GLenum target );
 
@@ -103,7 +106,6 @@ extern void
 _mesa_unmap_drawpix_pbo(GLcontext *ctx,
                         const struct gl_pixelstore_attrib *unpack);
 
-
 extern void *
 _mesa_map_readpix_pbo(GLcontext *ctx,
                       const struct gl_pixelstore_attrib *pack,
diff --git a/src/mesa/main/buffers.c b/src/mesa/main/buffers.c
index 1d07c68633c..5dd85de272a 100644
--- a/src/mesa/main/buffers.c
+++ b/src/mesa/main/buffers.c
@@ -133,6 +133,14 @@ draw_buffer_enum_to_bitmask(GLenum buffer)
          return BUFFER_BIT_COLOR2;
       case GL_COLOR_ATTACHMENT3_EXT:
          return BUFFER_BIT_COLOR3;
+      case GL_COLOR_ATTACHMENT4_EXT:
+         return BUFFER_BIT_COLOR4;
+      case GL_COLOR_ATTACHMENT5_EXT:
+         return BUFFER_BIT_COLOR5;
+      case GL_COLOR_ATTACHMENT6_EXT:
+         return BUFFER_BIT_COLOR6;
+      case GL_COLOR_ATTACHMENT7_EXT:
+         return BUFFER_BIT_COLOR7;
       default:
          /* error */
          return BAD_MASK;
@@ -182,6 +190,14 @@ read_buffer_enum_to_index(GLenum buffer)
          return BUFFER_COLOR2;
       case GL_COLOR_ATTACHMENT3_EXT:
          return BUFFER_COLOR3;
+      case GL_COLOR_ATTACHMENT4_EXT:
+         return BUFFER_COLOR4;
+      case GL_COLOR_ATTACHMENT5_EXT:
+         return BUFFER_COLOR5;
+      case GL_COLOR_ATTACHMENT6_EXT:
+         return BUFFER_COLOR6;
+      case GL_COLOR_ATTACHMENT7_EXT:
+         return BUFFER_COLOR7;
       default:
          /* error */
          return -1;
@@ -196,6 +212,20 @@ read_buffer_enum_to_index(GLenum buffer)
  * \sa _mesa_DrawBuffersARB
  *
  * \param buffer  buffer token such as GL_LEFT or GL_FRONT_AND_BACK, etc.
+ *
+ * Note that the behaviour of this function depends on whether the
+ * current ctx->DrawBuffer is a window-system framebuffer (Name=0) or
+ * a user-created framebuffer object (Name!=0).
+ *   In the former case, we update the per-context ctx->Color.DrawBuffer
+ *   state var _and_ the FB's ColorDrawBuffer state.
+ *   In the later case, we update the FB's ColorDrawBuffer state only.
+ *
+ * Furthermore, upon a MakeCurrent() or BindFramebuffer() call, if the
+ * new FB is a window system FB, we need to re-update the FB's
+ * ColorDrawBuffer state to match the context.  This is handled in
+ * _mesa_update_framebuffer().
+ *
+ * See the GL_EXT_framebuffer_object spec for more info.
  */
 void GLAPIENTRY
 _mesa_DrawBuffer(GLenum buffer)
@@ -315,45 +345,16 @@ _mesa_DrawBuffersARB(GLsizei n, const GLenum *buffers)
 
 
 /**
- * Set color output state.  Traditionally, there was only one color
- * output, but fragment programs can now have several distinct color
- * outputs (see GL_ARB_draw_buffers).  This function sets the state
- * for one such color output.
- * \param ctx  current context
- * \param output  which fragment program output
- * \param buffer  buffer to write to (like GL_LEFT)
- * \param destMask  BUFFER_* bitmask
- *                  (like BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT).
- */
-static void
-set_color_output(GLcontext *ctx, GLuint output, GLenum buffer,
-                 GLbitfield destMask)
-{
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-   ASSERT(output < ctx->Const.MaxDrawBuffers);
-
-   /* Set per-FBO state */
-   fb->ColorDrawBuffer[output] = buffer;
-   fb->_ColorDrawBufferMask[output] = destMask;
-   /* not really needed, will be set later */
-   fb->_NumColorDrawBuffers[output] = 0;
-
-   if (fb->Name == 0)
-   /* Set traditional state var */
-      ctx->Color.DrawBuffer[output] = buffer;
-}
-
-
-/**
- * Helper routine used by _mesa_DrawBuffer, _mesa_DrawBuffersARB and
- * other places (window fbo fixup) to set fbo (and the old ctx) fields.
+ * Helper function to set the GL_DRAW_BUFFER state in the context and
+ * current FBO.
+ *
  * All error checking will have been done prior to calling this function
  * so nothing should go wrong at this point.
+ *
  * \param ctx  current context
  * \param n    number of color outputs to set
  * \param buffers  array[n] of colorbuffer names, like GL_LEFT.
- * \param destMask  array[n] of BUFFER_* bitmasks which correspond to the
+ * \param destMask  array[n] of BUFFER_BIT_* bitmasks which correspond to the
  *                  colorbuffer names.  (i.e. GL_FRONT_AND_BACK =>
  *                  BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT).
  */
@@ -361,13 +362,13 @@ void
 _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
                   const GLbitfield *destMask)
 {
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
    GLbitfield mask[MAX_DRAW_BUFFERS];
-   GLuint output;
 
    if (!destMask) {
       /* compute destMask values now */
-      const GLbitfield supportedMask
-         = supported_buffer_bitmask(ctx, ctx->DrawBuffer);
+      const GLbitfield supportedMask = supported_buffer_bitmask(ctx, fb);
+      GLuint output;
       for (output = 0; output < n; output++) {
          mask[output] = draw_buffer_enum_to_bitmask(buffers[output]);
          ASSERT(mask[output] != BAD_MASK);
@@ -376,56 +377,75 @@ _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
       destMask = mask;
    }
 
-   for (output = 0; output < n; output++) {
-      set_color_output(ctx, output, buffers[output], destMask[output]);
+   if (n == 1) {
+      GLuint buf, count = 0;
+      /* init to -1 to help catch errors */
+      fb->_ColorDrawBufferIndexes[0] = -1;
+      for (buf = 0; buf < BUFFER_COUNT; buf++) {
+         if (destMask[0] & (1 << buf)) {
+            fb->_ColorDrawBufferIndexes[count] = buf;
+            count++;
+         }
+      }
+      fb->ColorDrawBuffer[0] = buffers[0];
+      fb->_NumColorDrawBuffers = count;
+   }
+   else {
+      GLuint buf, count = 0;
+      for (buf = 0; buf < n; buf++ ) {
+         if (destMask[buf]) {
+            fb->_ColorDrawBufferIndexes[buf] = _mesa_ffs(destMask[buf]) - 1;
+            fb->ColorDrawBuffer[buf] = buffers[buf];
+            count = buf + 1;
+         }
+         else {
+            fb->_ColorDrawBufferIndexes[buf] = -1;
+         }
+      }
+      /* set remaining outputs to -1 (GL_NONE) */
+      while (buf < ctx->Const.MaxDrawBuffers) {
+         fb->_ColorDrawBufferIndexes[buf] = -1;
+         fb->ColorDrawBuffer[buf] = GL_NONE;
+         buf++;
+      }
+      fb->_NumColorDrawBuffers = count;
    }
 
-   /* set remaining color outputs to NONE */
-   for (output = n; output < ctx->Const.MaxDrawBuffers; output++) {
-      set_color_output(ctx, output, GL_NONE, 0x0);
+   if (fb->Name == 0) {
+      /* also set context drawbuffer state */
+      GLuint buf;
+      for (buf = 0; buf < ctx->Const.MaxDrawBuffers; buf++) {
+         ctx->Color.DrawBuffer[buf] = fb->ColorDrawBuffer[buf];
+      }
    }
 
    ctx->NewState |= _NEW_BUFFERS;
 }
 
 
-GLboolean
-_mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer)
+/**
+ * Like \sa _mesa_drawbuffers(), this is a helper function for setting
+ * GL_READ_BUFFER state in the context and current FBO.
+ * \param ctx  the rendering context
+ * \param buffer  GL_FRONT, GL_BACK, GL_COLOR_ATTACHMENT0, etc.
+ * \param bufferIndex  the numerical index corresponding to 'buffer'
+ */
+void
+_mesa_readbuffer(GLcontext *ctx, GLenum buffer, GLint bufferIndex)
 {
-   struct gl_framebuffer *fb;
-   GLbitfield supportedMask;
-   GLint srcBuffer;
-
-   fb = ctx->ReadBuffer;
-
-   if (MESA_VERBOSE & VERBOSE_API)
-      _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
-
-   if (fb->Name > 0 && buffer == GL_NONE) {
-      /* This is legal for user-created framebuffer objects */
-      srcBuffer = -1;
-   }
-   else {
-      /* general case / window-system framebuffer */
-      srcBuffer = read_buffer_enum_to_index(buffer);
-      if (srcBuffer == -1) {
-         _mesa_error(ctx, GL_INVALID_ENUM, "glReadBuffer(buffer=0x%x)", buffer);
-         return GL_FALSE;
-      }
-      supportedMask = supported_buffer_bitmask(ctx, fb);
-      if (((1 << srcBuffer) & supportedMask) == 0) {
-         _mesa_error(ctx, GL_INVALID_OPERATION, "glReadBuffer(buffer=0x%x)", buffer);
-         return GL_FALSE;
-      }
-   }
+   struct gl_framebuffer *fb = ctx->ReadBuffer;
 
    if (fb->Name == 0) {
+      /* Only update the per-context READ_BUFFER state if we're bound to
+       * a window-system framebuffer.
+       */
       ctx->Pixel.ReadBuffer = buffer;
    }
+
    fb->ColorReadBuffer = buffer;
-   fb->_ColorReadBufferIndex = srcBuffer;
+   fb->_ColorReadBufferIndex = bufferIndex;
 
-   return GL_TRUE;
+   ctx->NewState |= _NEW_PIXEL;
 }
 
 
@@ -437,15 +457,43 @@ _mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer)
 void GLAPIENTRY
 _mesa_ReadBuffer(GLenum buffer)
 {
+   struct gl_framebuffer *fb;
+   GLbitfield supportedMask;
+   GLint srcBuffer;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
 
-   if (!_mesa_readbuffer_update_fields(ctx, buffer))
-      return;
+   fb = ctx->ReadBuffer;
+
+   if (MESA_VERBOSE & VERBOSE_API)
+      _mesa_debug(ctx, "glReadBuffer %s\n", _mesa_lookup_enum_by_nr(buffer));
+
+   if (fb->Name > 0 && buffer == GL_NONE) {
+      /* This is legal for user-created framebuffer objects */
+      srcBuffer = -1;
+   }
+   else {
+      /* general case / window-system framebuffer */
+      srcBuffer = read_buffer_enum_to_index(buffer);
+      if (srcBuffer == -1) {
+         _mesa_error(ctx, GL_INVALID_ENUM,
+                     "glReadBuffer(buffer=0x%x)", buffer);
+         return;
+      }
+      supportedMask = supported_buffer_bitmask(ctx, fb);
+      if (((1 << srcBuffer) & supportedMask) == 0) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glReadBuffer(buffer=0x%x)", buffer);
+         return;
+      }
+   }
+
+   /* OK, all error checking has been completed now */
 
+   _mesa_readbuffer(ctx, buffer, srcBuffer);
    ctx->NewState |= _NEW_BUFFERS;
 
    /*
diff --git a/src/mesa/main/buffers.h b/src/mesa/main/buffers.h
index 53d5fb80d40..8a7e7b5c1f0 100644
--- a/src/mesa/main/buffers.h
+++ b/src/mesa/main/buffers.h
@@ -5,9 +5,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -46,8 +46,8 @@ extern void
 _mesa_drawbuffers(GLcontext *ctx, GLuint n, const GLenum *buffers,
                   const GLbitfield *destMask);
 
-extern GLboolean
-_mesa_readbuffer_update_fields(GLcontext *ctx, GLenum buffer);
+extern void
+_mesa_readbuffer(GLcontext *ctx, GLenum buffer, GLint bufferIndex);
 
 extern void GLAPIENTRY
 _mesa_ReadBuffer( GLenum mode );
diff --git a/src/mesa/main/clear.c b/src/mesa/main/clear.c
index 434685984d1..63388f42ee4 100644
--- a/src/mesa/main/clear.c
+++ b/src/mesa/main/clear.c
@@ -138,7 +138,9 @@ _mesa_Clear( GLbitfield mask )
       return;
    }
 
-   if (ctx->DrawBuffer->Width == 0 || ctx->DrawBuffer->Height == 0)
+   if (ctx->DrawBuffer->Width == 0 || ctx->DrawBuffer->Height == 0 ||
+       ctx->DrawBuffer->_Xmin >= ctx->DrawBuffer->_Xmax ||
+       ctx->DrawBuffer->_Ymin >= ctx->DrawBuffer->_Ymax)
       return;
 
    if (ctx->RenderMode == GL_RENDER) {
@@ -155,7 +157,10 @@ _mesa_Clear( GLbitfield mask )
        */
       bufferMask = 0;
       if (mask & GL_COLOR_BUFFER_BIT) {
-         bufferMask |= ctx->DrawBuffer->_ColorDrawBufferMask[0];
+         GLuint i;
+         for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
+            bufferMask |= (1 << ctx->DrawBuffer->_ColorDrawBufferIndexes[i]);
+         }
       }
 
       if ((mask & GL_DEPTH_BUFFER_BIT)
diff --git a/src/mesa/main/colormac.h b/src/mesa/main/colormac.h
index a19521fc85b..a34bd2ed388 100644
--- a/src/mesa/main/colormac.h
+++ b/src/mesa/main/colormac.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.1
+ * Version:  7.3
  *
- * Copyright (C) 1999-2004  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -180,20 +180,20 @@ do {						\
  */
 /*@{*/
 
-#define PACK_COLOR_8888( R, G, B, A )					\
-   (((R) << 24) | ((G) << 16) | ((B) << 8) | (A))
+#define PACK_COLOR_8888( X, Y, Z, W ) \
+   (((X) << 24) | ((Y) << 16) | ((Z) << 8) | (W))
 
-#define PACK_COLOR_8888_REV( R, G, B, A )				\
-   (((A) << 24) | ((B) << 16) | ((G) << 8) | (R))
+#define PACK_COLOR_8888_REV( X, Y, Z, W ) \
+   (((W) << 24) | ((Z) << 16) | ((Y) << 8) | (X))
 
-#define PACK_COLOR_888( R, G, B )					\
-   (((R) << 16) | ((G) << 8) | (B))
+#define PACK_COLOR_888( X, Y, Z ) \
+   (((X) << 16) | ((Y) << 8) | (Z))
 
-#define PACK_COLOR_565( R, G, B )					\
-   ((((R) & 0xf8) << 8) | (((G) & 0xfc) << 3) | (((B) & 0xf8) >> 3))
+#define PACK_COLOR_565( X, Y, Z )                                  \
+   ((((X) & 0xf8) << 8) | (((Y) & 0xfc) << 3) | (((Z) & 0xf8) >> 3))
 
-#define PACK_COLOR_565_REV( R, G, B )					\
-   (((R) & 0xf8) | ((G) & 0xe0) >> 5 | (((G) & 0x1c) << 11) | (((B) & 0xf8) << 5))
+#define PACK_COLOR_565_REV( X, Y, Z ) \
+   (((X) & 0xf8) | ((Y) & 0xe0) >> 5 | (((Y) & 0x1c) << 11) | (((Z) & 0xf8) << 5))
 
 #define PACK_COLOR_1555( A, B, G, R )					\
    ((((B) & 0xf8) << 7) | (((G) & 0xf8) << 2) | (((R) & 0xf8) >> 3) |	\
diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c
index 97120398f9b..bd9cf438b4f 100644
--- a/src/mesa/main/colortab.c
+++ b/src/mesa/main/colortab.c
@@ -383,7 +383,7 @@ _mesa_ColorTable( GLenum target, GLenum internalFormat,
       return;
    }
 
-   if (width < 0 || (width != 0 && _mesa_bitcount(width) != 1)) {
+   if (width < 0 || (width != 0 && !_mesa_is_pow_two(width))) {
       /* error */
       if (proxy) {
          table->Size = 0;
diff --git a/src/mesa/main/config.h b/src/mesa/main/config.h
index 5e9a4f89394..e29964a1e86 100644
--- a/src/mesa/main/config.h
+++ b/src/mesa/main/config.h
@@ -191,6 +191,8 @@
 #define MAX_UNIFORMS 128   /**< number of float components */
 #define MAX_VARYING 8      /**< number of float[4] vectors */
 #define MAX_SAMPLERS 8
+#define MAX_PROGRAM_INPUTS 32
+#define MAX_PROGRAM_OUTPUTS 32
 /*@}*/
 
 /** For GL_NV_vertex_program */
@@ -216,7 +218,7 @@
 /** For GL_ARB_vertex_shader */
 /*@{*/
 #define MAX_VERTEX_ATTRIBS 16
-#define MAX_VERTEX_TEXTURE_IMAGE_UNITS 0
+#define MAX_VERTEX_TEXTURE_IMAGE_UNITS MAX_TEXTURE_UNITS
 #define MAX_COMBINED_TEXTURE_IMAGE_UNITS (MAX_TEXTURE_IMAGE_UNITS + MAX_VERTEX_TEXTURE_IMAGE_UNITS)
 /*@}*/
 
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index e5ec35c77fd..61c0861cbd4 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -6,9 +6,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -809,7 +809,7 @@ _mesa_init_current(GLcontext *ctx)
    }
 
    /* redo special cases: */
-   ASSIGN_4V( ctx->Current.Attrib[VERT_ATTRIB_WEIGHT], 1.0, 0.0, 0.0, 1.0 );
+   ASSIGN_4V( ctx->Current.Attrib[VERT_ATTRIB_WEIGHT], 1.0, 0.0, 0.0, 0.0 );
    ASSIGN_4V( ctx->Current.Attrib[VERT_ATTRIB_NORMAL], 0.0, 0.0, 1.0, 1.0 );
    ASSIGN_4V( ctx->Current.Attrib[VERT_ATTRIB_COLOR0], 1.0, 1.0, 1.0, 1.0 );
    ASSIGN_4V( ctx->Current.Attrib[VERT_ATTRIB_COLOR1], 0.0, 0.0, 0.0, 1.0 );
@@ -1063,7 +1063,6 @@ init_attrib_groups(GLcontext *ctx)
    /* Miscellaneous */
    ctx->NewState = _NEW_ALL;
    ctx->ErrorValue = (GLenum) GL_NO_ERROR;
-   ctx->_Facing = 0;
    ctx->varying_vp_inputs = ~0;
 
    return GL_TRUE;
@@ -1071,6 +1070,28 @@ init_attrib_groups(GLcontext *ctx)
 
 
 /**
+ * Update default objects in a GL context with respect to shared state.
+ *
+ * \param ctx GL context.
+ *
+ * Removes references to old default objects, (texture objects, program
+ * objects, etc.) and changes to reference those from the current shared
+ * state.
+ */
+static GLboolean
+update_default_objects(GLcontext *ctx)
+{
+   assert(ctx);
+
+   _mesa_update_default_objects_program(ctx);
+   _mesa_update_default_objects_texture(ctx);
+   _mesa_update_default_objects_buffer_objects(ctx);
+
+   return GL_TRUE;
+}
+
+
+/**
  * This is the default function we plug into all dispatch table slots
  * This helps prevents a segfault when someone calls a GL function without
  * first checking if the extension's supported.
@@ -1274,18 +1295,19 @@ _mesa_create_context(const GLvisual *visual,
 void
 _mesa_free_context_data( GLcontext *ctx )
 {
-   /* if we're destroying the current context, unbind it first */
-   if (ctx == _mesa_get_current_context()) {
-      _mesa_make_current(NULL, NULL, NULL);
-   }
-   else {
-      /* unreference WinSysDraw/Read buffers */
-      _mesa_unreference_framebuffer(&ctx->WinSysDrawBuffer);
-      _mesa_unreference_framebuffer(&ctx->WinSysReadBuffer);
-      _mesa_unreference_framebuffer(&ctx->DrawBuffer);
-      _mesa_unreference_framebuffer(&ctx->ReadBuffer);
+   if (!_mesa_get_current_context()){
+      /* No current context, but we may need one in order to delete
+       * texture objs, etc.  So temporarily bind the context now.
+       */
+      _mesa_make_current(ctx, NULL, NULL);
    }
 
+   /* unreference WinSysDraw/Read buffers */
+   _mesa_unreference_framebuffer(&ctx->WinSysDrawBuffer);
+   _mesa_unreference_framebuffer(&ctx->WinSysReadBuffer);
+   _mesa_unreference_framebuffer(&ctx->DrawBuffer);
+   _mesa_unreference_framebuffer(&ctx->ReadBuffer);
+
    _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current, NULL);
    _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, NULL);
    _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram, NULL);
@@ -1294,6 +1316,9 @@ _mesa_free_context_data( GLcontext *ctx )
    _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL);
    _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, NULL);
 
+#if FEATURE_attrib_stack
+   _mesa_free_attrib_data(ctx);
+#endif
    _mesa_free_lighting_data( ctx );
 #if FEATURE_evaluators
    _mesa_free_eval_data( ctx );
@@ -1331,6 +1356,11 @@ _mesa_free_context_data( GLcontext *ctx )
 
    if (ctx->Extensions.String)
       _mesa_free((void *) ctx->Extensions.String);
+
+   /* unbind the context if it's currently bound */
+   if (ctx == _mesa_get_current_context()) {
+      _mesa_make_current(NULL, NULL, NULL);
+   }
 }
 
 
@@ -1562,8 +1592,6 @@ void
 _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer,
                     GLframebuffer *readBuffer )
 {
-   GET_CURRENT_CONTEXT(oldCtx);
-
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(newCtx, "_mesa_make_current()\n");
 
@@ -1588,13 +1616,6 @@ _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer,
    _glapi_set_context((void *) newCtx);
    ASSERT(_mesa_get_current_context() == newCtx);
 
-   if (oldCtx) {
-      _mesa_unreference_framebuffer(&oldCtx->WinSysDrawBuffer);
-      _mesa_unreference_framebuffer(&oldCtx->WinSysReadBuffer);
-      _mesa_unreference_framebuffer(&oldCtx->DrawBuffer);
-      _mesa_unreference_framebuffer(&oldCtx->ReadBuffer);
-   }
-         
    if (!newCtx) {
       _glapi_set_dispatch(NULL);  /* none current */
    }
@@ -1614,10 +1635,12 @@ _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer,
           * or not bound to a user-created FBO.
           */
          if (!newCtx->DrawBuffer || newCtx->DrawBuffer->Name == 0) {
-         /* fix up the fb fields - these will end up wrong otherwise
-          * if the DRIdrawable changes, and everything relies on them.
-          * This is a bit messy (same as needed in _mesa_BindFramebufferEXT)
-	  */
+            /* KW: merge conflict here, revisit. 
+             */
+            /* fix up the fb fields - these will end up wrong otherwise
+             * if the DRIdrawable changes, and everything relies on them.
+             * This is a bit messy (same as needed in _mesa_BindFramebufferEXT)
+             */
             unsigned int i;
             GLenum buffers[MAX_DRAW_BUFFERS];
 
@@ -1631,9 +1654,11 @@ _mesa_make_current( GLcontext *newCtx, GLframebuffer *drawBuffer,
          }
          if (!newCtx->ReadBuffer || newCtx->ReadBuffer->Name == 0) {
             _mesa_reference_framebuffer(&newCtx->ReadBuffer, readBuffer);
-            _mesa_readbuffer_update_fields(newCtx, newCtx->Pixel.ReadBuffer);
          }
 
+         /* XXX only set this flag if we're really changing the draw/read
+          * framebuffer bindings.
+          */
 	 newCtx->NewState |= _NEW_BUFFERS;
 
 #if 1
@@ -1701,12 +1726,18 @@ GLboolean
 _mesa_share_state(GLcontext *ctx, GLcontext *ctxToShare)
 {
    if (ctx && ctxToShare && ctx->Shared && ctxToShare->Shared) {
-      ctx->Shared->RefCount--;
-      if (ctx->Shared->RefCount == 0) {
-         free_shared_state(ctx, ctx->Shared);
-      }
+      struct gl_shared_state *oldSharedState = ctx->Shared;
+
       ctx->Shared = ctxToShare->Shared;
       ctx->Shared->RefCount++;
+
+      update_default_objects(ctx);
+
+      oldSharedState->RefCount--;
+      if (oldSharedState->RefCount == 0) {
+         free_shared_state(ctx, oldSharedState);
+      }
+
       return GL_TRUE;
    }
    else {
diff --git a/src/mesa/main/context.h b/src/mesa/main/context.h
index 9423b66a7d8..54f1af911d3 100644
--- a/src/mesa/main/context.h
+++ b/src/mesa/main/context.h
@@ -275,10 +275,12 @@ do {									\
    (((CTX)->Light.Enabled &&						\
      (CTX)->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)	\
     || (CTX)->Fog.ColorSumEnabled					\
-    || ((CTX)->VertexProgram._Enabled &&				\
-        ((CTX)->VertexProgram.Current->Base.InputsRead & VERT_BIT_COLOR1)) \
-    || ((CTX)->FragmentProgram._Enabled &&				\
-        ((CTX)->FragmentProgram.Current->Base.InputsRead & FRAG_BIT_COL1)) \
+    || ((CTX)->VertexProgram._Current &&				\
+        ((CTX)->VertexProgram._Current != (CTX)->VertexProgram._TnlProgram) &&    \
+        ((CTX)->VertexProgram._Current->Base.InputsRead & VERT_BIT_COLOR1)) \
+    || ((CTX)->FragmentProgram._Current &&				\
+        ((CTX)->FragmentProgram._Current != (CTX)->FragmentProgram._TexEnvProgram) &&  \
+        ((CTX)->FragmentProgram._Current->Base.InputsRead & FRAG_BIT_COL1)) \
    )
 
 
diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h
index 7fb0a211d7a..ddb38030bf7 100644
--- a/src/mesa/main/dd.h
+++ b/src/mesa/main/dd.h
@@ -456,8 +456,8 @@ struct dd_function_table {
     */
    void (*GetCompressedTexImage)(GLcontext *ctx, GLenum target, GLint level,
                                  GLvoid *img,
-                                 const struct gl_texture_object *texObj,
-                                 const struct gl_texture_image *texImage);
+                                 struct gl_texture_object *texObj,
+                                 struct gl_texture_image *texImage);
 
    /**
     * Called to query number of bytes of storage needed to store the
@@ -504,6 +504,11 @@ struct dd_function_table {
     */
    void (*FreeTexImageData)( GLcontext *ctx, struct gl_texture_image *tImage );
 
+   /** Map texture image data into user space */
+   void (*MapTexture)( GLcontext *ctx, struct gl_texture_object *tObj );
+   /** Unmap texture images from user space */
+   void (*UnmapTexture)( GLcontext *ctx, struct gl_texture_object *tObj );
+
    /**
     * Note: no context argument.  This function doesn't initially look
     * like it belongs here, except that the driver is the only entity
diff --git a/src/mesa/main/depthstencil.c b/src/mesa/main/depthstencil.c
index d4990bb795f..9d208e2997d 100644
--- a/src/mesa/main/depthstencil.c
+++ b/src/mesa/main/depthstencil.c
@@ -213,7 +213,7 @@ put_values_z24(GLcontext *ctx, struct gl_renderbuffer *z24rb, GLuint count,
                const void *values, const GLubyte *mask)
 {
    struct gl_renderbuffer *dsrb = z24rb->Wrapped;
-   const GLubyte *src = (const GLubyte *) values;
+   const GLuint *src = (const GLuint *) values;
    ASSERT(z24rb->DataType == GL_UNSIGNED_INT);
    ASSERT(dsrb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT);
    ASSERT(dsrb->DataType == GL_UNSIGNED_INT_24_8_EXT);
@@ -282,8 +282,8 @@ _mesa_new_z24_renderbuffer_wrapper(GLcontext *ctx,
    z24rb->RefCount = 1;
    z24rb->Width = dsrb->Width;
    z24rb->Height = dsrb->Height;
-   z24rb->InternalFormat = GL_DEPTH_COMPONENT24_ARB;
-   z24rb->_ActualFormat = GL_DEPTH_COMPONENT24_ARB;
+   z24rb->InternalFormat = GL_DEPTH_COMPONENT24;
+   z24rb->_ActualFormat = GL_DEPTH_COMPONENT24;
    z24rb->_BaseFormat = GL_DEPTH_COMPONENT;
    z24rb->DataType = GL_UNSIGNED_INT;
    z24rb->DepthBits = 24;
diff --git a/src/mesa/main/descrip.mms b/src/mesa/main/descrip.mms
new file mode 100644
index 00000000000..e49ec65d42d
--- /dev/null
+++ b/src/mesa/main/descrip.mms
@@ -0,0 +1,258 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 29 September 2008
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define shader [-.shader]
+	define glapi [-.glapi]
+	define main [-.main]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.glapi],[-.shader]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES =accum.c \
+	api_arrayelt.c \
+	api_exec.c \
+	api_loopback.c \
+	api_noop.c \
+	api_validate.c \
+ 	attrib.c \
+	arrayobj.c \
+	blend.c \
+	bufferobj.c \
+	buffers.c \
+	clear.c \
+	clip.c \
+	colortab.c \
+	context.c \
+	convolve.c \
+	debug.c \
+	depth.c \
+	depthstencil.c \
+	dispatch.c \
+	dlist.c \
+	drawpix.c \
+	enable.c \
+	enums.c \
+	eval.c \
+	execmem.c \
+	extensions.c \
+	fbobject.c \
+	feedback.c \
+	ffvertex_prog.c \
+	fog.c \
+	framebuffer.c \
+	get.c \
+	getstring.c \
+	hash.c \
+	hint.c \
+	histogram.c \
+	image.c \
+	imports.c \
+	light.c \
+	lines.c \
+	matrix.c \
+	mipmap.c \
+	mm.c \
+	multisample.c \
+	pixel.c \
+	pixelstore.c \
+	points.c \
+	polygon.c \
+	rastpos.c \
+	rbadaptors.c \
+	readpix.c \
+	renderbuffer.c \
+	scissor.c \
+	shaders.c \
+	state.c \
+	stencil.c \
+	texcompress.c \
+	texcompress_fxt1.c \
+	texcompress_s3tc.c \
+	texenv.c \
+	texenvprogram.c \
+	texformat.c \
+	texgen.c \
+	teximage.c \
+	texobj.c \
+	texparam.c \
+	texrender.c \
+	texstate.c \
+	texstore.c \
+	varray.c \
+	vtxfmt.c \
+	queryobj.c \
+	rbadaptors.c
+
+OBJECTS=accum.obj,\
+api_arrayelt.obj,\
+api_exec.obj,\
+api_loopback.obj,\
+api_noop.obj,\
+api_validate.obj,\
+arrayobj.obj,\
+attrib.obj,\
+blend.obj,\
+bufferobj.obj,\
+buffers.obj,\
+clear.obj,\
+clip.obj,\
+colortab.obj,\
+context.obj,\
+convolve.obj,\
+debug.obj,\
+depth.obj,\
+depthstencil.obj,\
+dispatch.obj,\
+dlist.obj,\
+drawpix.obj,\
+enable.obj,\
+enums.obj,\
+eval.obj,\
+execmem.obj,\
+extensions.obj,\
+fbobject.obj,\
+feedback.obj,\
+ffvertex_prog.obj,\
+fog.obj,\
+framebuffer.obj,\
+get.obj,\
+getstring.obj,\
+hash.obj,\
+hint.obj,\
+histogram.obj,\
+image.obj,\
+imports.obj,\
+light.obj,\
+lines.obj,\
+matrix.obj,\
+mipmap.obj,\
+mm.obj,\
+multisample.obj,\
+pixel.obj,\
+pixelstore.obj,\
+points.obj,\
+polygon.obj,\
+rastpos.obj,\
+readpix.obj,\
+renderbuffer.obj,\
+scissor.obj,\
+shaders.obj,\
+state.obj,\
+stencil.obj,\
+texcompress.obj,\
+texcompress_fxt1.obj,\
+texcompress_s3tc.obj,\
+texenv.obj,\
+texenvprogram.obj,\
+texformat.obj,\
+texgen.obj,\
+teximage.obj,\
+texobj.obj,\
+texparam.obj,\
+texrender.obj,\
+texstate.obj,\
+texstore.obj,\
+varray.obj,\
+vtxfmt.obj,\
+queryobj.obj,\
+rbadaptors.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ $(MAKELIB) $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+accum.obj : accum.c
+api_arrayelt.obj : api_arrayelt.c
+api_loopback.obj : api_loopback.c
+api_noop.obj : api_noop.c
+api_validate.obj : api_validate.c
+arrayobj.obj : arrayobj.c
+attrib.obj : attrib.c
+blend.obj : blend.c
+bufferobj.obj : bufferobj.c
+buffers.obj : buffers.c
+clip.obj : clip.c
+colortab.obj : colortab.c
+context.obj : context.c
+convolve.obj : convolve.c
+debug.obj : debug.c
+depth.obj : depth.c
+depthstencil.obj : depthstencil.c
+dispatch.obj : dispatch.c
+dlist.obj : dlist.c
+drawpix.obj : drawpix.c
+enable.obj : enable.c
+enums.obj : enums.c
+eval.obj : eval.c
+execmem.obj : execmem.c
+extensions.obj : extensions.c
+fbobject.obj : fbobject.c
+feedback.obj : feedback.c
+fog.obj : fog.c
+framebuffer.obj : framebuffer.c
+get.obj : get.c
+getstring.obj : getstring.c
+hash.obj : hash.c
+hint.obj : hint.c
+histogram.obj : histogram.c
+image.obj : image.c
+imports.obj : imports.c vsnprintf.c
+light.obj : light.c
+lines.obj : lines.c
+matrix.obj : matrix.c
+mipmap.obj : mipmap.c
+mm.obj : mm.c
+pixel.obj : pixel.c
+points.obj : points.c
+polygon.obj : polygon.c
+rastpos.obj : rastpos.c
+rbadaptors.obj : rbadaptors.c
+renderbuffer.obj : renderbuffer.c
+state.obj : state.c
+stencil.obj : stencil.c
+texcompress.obj : texcompress.c
+texcompress_fxt1.obj : texcompress_fxt1.c
+	cc$(CFLAGS)/warn=(disable=SHIFTCOUNT) texcompress_fxt1.c
+texcompress_s3tc.obj : texcompress_s3tc.c
+texenvprogram.obj : texenvprogram.c
+texformat.obj : texformat.c
+teximage.obj : teximage.c
+texobj.obj : texobj.c
+texrender.obj : texrender.c
+texstate.obj : texstate.c
+texstore.obj : texstore.c
+varray.obj : varray.c
+vtxfmt.obj : vtxfmt.c
+shaders.obj : shaders.c
+queryobj.obj : queryobj.c
+rbadaptors.obj : rbadaptors.c
+clear.obj : clear.c
+multisample.obj : multisample.c
+scissor.obj : scissor.c
+texenv.obj : texenv.c
+texgen.obj : texgen.c
+texparam.obj : texparam.c
+readpix.obj : readpix.c
+ffvertex_prog.obj : ffvertex_prog.c
+api_exec.obj : api_exec.c
+pixelstore.obj : pixelstore.c
diff --git a/src/mesa/main/dispatch.c b/src/mesa/main/dispatch.c
index c12f55a7a1f..34127cb248a 100644
--- a/src/mesa/main/dispatch.c
+++ b/src/mesa/main/dispatch.c
@@ -39,7 +39,7 @@
 
 #ifndef GLX_USE_APPLEGL
 
-#include "glheader.h"
+#include "main/glheader.h"
 #include "glapi/glapi.h"
 #include "glapi/glapitable.h"
 #include "glapi/glthread.h"
diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c
index ffe6dbfe08a..d3aee196c74 100644
--- a/src/mesa/main/dlist.c
+++ b/src/mesa/main/dlist.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -41,7 +41,6 @@
 #endif
 #include "arrayobj.h"
 #include "clip.h"
-#include "colormac.h"
 #include "colortab.h"
 #include "context.h"
 #include "convolve.h"
@@ -612,9 +611,9 @@ destroy_list(GLcontext *ctx, GLuint list)
 
 
 /*
- * Translate the nth element of list from type to GLuint.
+ * Translate the nth element of list from <type> to GLint.
  */
-static GLuint
+static GLint
 translate_id(GLsizei n, GLenum type, const GLvoid * list)
 {
    GLbyte *bptr;
@@ -628,37 +627,40 @@ translate_id(GLsizei n, GLenum type, const GLvoid * list)
    switch (type) {
    case GL_BYTE:
       bptr = (GLbyte *) list;
-      return (GLuint) *(bptr + n);
+      return (GLint) bptr[n];
    case GL_UNSIGNED_BYTE:
       ubptr = (GLubyte *) list;
-      return (GLuint) *(ubptr + n);
+      return (GLint) ubptr[n];
    case GL_SHORT:
       sptr = (GLshort *) list;
-      return (GLuint) *(sptr + n);
+      return (GLint) sptr[n];
    case GL_UNSIGNED_SHORT:
       usptr = (GLushort *) list;
-      return (GLuint) *(usptr + n);
+      return (GLint) usptr[n];
    case GL_INT:
       iptr = (GLint *) list;
-      return (GLuint) *(iptr + n);
+      return iptr[n];
    case GL_UNSIGNED_INT:
       uiptr = (GLuint *) list;
-      return (GLuint) *(uiptr + n);
+      return (GLint) uiptr[n];
    case GL_FLOAT:
       fptr = (GLfloat *) list;
-      return (GLuint) *(fptr + n);
+      return (GLint) FLOORF(fptr[n]);
    case GL_2_BYTES:
       ubptr = ((GLubyte *) list) + 2 * n;
-      return (GLuint) *ubptr * 256 + (GLuint) * (ubptr + 1);
+      return (GLint) ubptr[0] * 256
+           + (GLint) ubptr[1];
    case GL_3_BYTES:
       ubptr = ((GLubyte *) list) + 3 * n;
-      return (GLuint) * ubptr * 65536
-           + (GLuint) *(ubptr + 1) * 256 + (GLuint) * (ubptr + 2);
+      return (GLint) ubptr[0] * 65536
+           + (GLint) ubptr[1] * 256
+           + (GLint) ubptr[2];
    case GL_4_BYTES:
       ubptr = ((GLubyte *) list) + 4 * n;
-      return (GLuint) *ubptr * 16777216
-           + (GLuint) *(ubptr + 1) * 65536
-           + (GLuint) *(ubptr + 2) * 256 + (GLuint) * (ubptr + 3);
+      return (GLint) ubptr[0] * 16777216
+           + (GLint) ubptr[1] * 65536
+           + (GLint) ubptr[2] * 256
+           + (GLint) ubptr[3];
    default:
       return 0;
    }
@@ -1000,10 +1002,10 @@ _mesa_save_CallLists(GLsizei n, GLenum type, const GLvoid * lists)
    }
 
    for (i = 0; i < n; i++) {
-      GLuint list = translate_id(i, type, lists);
+      GLint list = translate_id(i, type, lists);
       Node *n = ALLOC_INSTRUCTION(ctx, OPCODE_CALL_LIST_OFFSET, 2);
       if (n) {
-         n[1].ui = list;
+         n[1].i = list;
          n[2].b = typeErrorFlag;
       }
    }
@@ -2002,7 +2004,7 @@ save_Lightfv(GLenum light, GLenum pname, const GLfloat *params)
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
    n = ALLOC_INSTRUCTION(ctx, OPCODE_LIGHT, 6);
-   if (OPCODE_LIGHT) {
+   if (n) {
       GLint i, nParams;
       n[1].e = light;
       n[2].e = pname;
@@ -2723,21 +2725,20 @@ save_PolygonMode(GLenum face, GLenum mode)
 }
 
 
-/*
- * Polygon stipple must have been upacked already!
- */
 static void GLAPIENTRY
 save_PolygonStipple(const GLubyte * pattern)
 {
    GET_CURRENT_CONTEXT(ctx);
+   GLvoid *image = unpack_image(2, 32, 32, 1, GL_COLOR_INDEX, GL_BITMAP,
+                                pattern, &ctx->Unpack);
    Node *n;
    ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
    n = ALLOC_INSTRUCTION(ctx, OPCODE_POLYGON_STIPPLE, 1);
    if (n) {
-      void *data;
-      n[1].data = _mesa_malloc(32 * 4);
-      data = n[1].data;         /* This needed for Acorn compiler */
-      MEMCPY(data, pattern, 32 * 4);
+      n[1].data = image; 
+   }
+   else if (image) {
+      _mesa_free(image);
    }
    if (ctx->ExecuteFlag) {
       CALL_PolygonStipple(ctx->Exec, ((GLubyte *) pattern));
@@ -3254,6 +3255,36 @@ save_StencilFuncSeparate(GLenum face, GLenum func, GLint ref, GLuint mask)
 
 
 static void GLAPIENTRY
+save_StencilFuncSeparateATI(GLenum frontfunc, GLenum backfunc, GLint ref,
+                            GLuint mask)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   Node *n;
+   ASSERT_OUTSIDE_SAVE_BEGIN_END_AND_FLUSH(ctx);
+   /* GL_FRONT */
+   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
+   if (n) {
+      n[1].e = GL_FRONT;
+      n[2].e = frontfunc;
+      n[3].i = ref;
+      n[4].ui = mask;
+   }
+   /* GL_BACK */
+   n = ALLOC_INSTRUCTION(ctx, OPCODE_STENCIL_FUNC_SEPARATE, 4);
+   if (n) {
+      n[1].e = GL_BACK;
+      n[2].e = backfunc;
+      n[3].i = ref;
+      n[4].ui = mask;
+   }
+   if (ctx->ExecuteFlag) {
+      CALL_StencilFuncSeparate(ctx->Exec, (GL_FRONT, frontfunc, ref, mask));
+      CALL_StencilFuncSeparate(ctx->Exec, (GL_BACK, backfunc, ref, mask));
+   }
+}
+
+
+static void GLAPIENTRY
 save_StencilMaskSeparate(GLenum face, GLuint mask)
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -5587,6 +5618,27 @@ save_VertexAttrib4fvARB(GLuint index, const GLfloat * v)
 }
 
 
+/* GL_ARB_shader_objects, GL_ARB_vertex/fragment_shader */
+
+static void GLAPIENTRY
+exec_BindAttribLocationARB(GLuint program, GLuint index, const GLchar *name)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   FLUSH_VERTICES(ctx, 0);
+   CALL_BindAttribLocationARB(ctx->Exec, (program, index, name));
+}
+
+static GLint GLAPIENTRY
+exec_GetAttribLocationARB(GLuint program, const GLchar *name)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   FLUSH_VERTICES(ctx, 0);
+   return CALL_GetAttribLocationARB(ctx->Exec, (program, name));
+}
+/* XXX more shader functions needed here */
+
+
+
 #if FEATURE_EXT_framebuffer_blit
 static void GLAPIENTRY
 save_BlitFramebufferEXT(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
@@ -5695,7 +5747,7 @@ execute_list(GLcontext *ctx, GLuint list)
    if (!dlist)
       return;
 
-   ctx->ListState.CallStack[ctx->ListState.CallDepth++] = dlist;
+   ctx->ListState.CallDepth++;
 
    if (ctx->Driver.BeginCallList)
       ctx->Driver.BeginCallList(ctx, dlist);
@@ -5762,7 +5814,8 @@ execute_list(GLcontext *ctx, GLuint list)
                _mesa_error(ctx, GL_INVALID_ENUM, "glCallLists(type)");
             }
             else if (ctx->ListState.CallDepth < MAX_LIST_NESTING) {
-               execute_list(ctx, ctx->List.ListBase + n[1].ui);
+               GLuint list = (GLuint) (ctx->List.ListBase + n[1].i);
+               execute_list(ctx, list);
             }
             break;
          case OPCODE_CLEAR:
@@ -6126,7 +6179,12 @@ execute_list(GLcontext *ctx, GLuint list)
             CALL_PolygonMode(ctx->Exec, (n[1].e, n[2].e));
             break;
          case OPCODE_POLYGON_STIPPLE:
-            CALL_PolygonStipple(ctx->Exec, ((GLubyte *) n[1].data));
+            {
+               const struct gl_pixelstore_attrib save = ctx->Unpack;
+               ctx->Unpack = ctx->DefaultPacking;
+               CALL_PolygonStipple(ctx->Exec, ((GLubyte *) n[1].data));
+               ctx->Unpack = save;      /* restore */
+            }
             break;
          case OPCODE_POLYGON_OFFSET:
             CALL_PolygonOffset(ctx->Exec, (n[1].f, n[2].f));
@@ -6577,7 +6635,7 @@ execute_list(GLcontext *ctx, GLuint list)
    if (ctx->Driver.EndCallList)
       ctx->Driver.EndCallList(ctx);
 
-   ctx->ListState.CallStack[ctx->ListState.CallDepth--] = NULL;
+   ctx->ListState.CallDepth--;
 }
 
 
@@ -6808,7 +6866,6 @@ void GLAPIENTRY
 _mesa_CallLists(GLsizei n, GLenum type, const GLvoid * lists)
 {
    GET_CURRENT_CONTEXT(ctx);
-   GLuint list;
    GLint i;
    GLboolean save_compile_flag;
 
@@ -6840,8 +6897,8 @@ _mesa_CallLists(GLsizei n, GLenum type, const GLvoid * lists)
    ctx->CompileFlag = GL_FALSE;
 
    for (i = 0; i < n; i++) {
-      list = translate_id(i, type, lists);
-      execute_list(ctx, ctx->List.ListBase + list);
+      GLuint list = (GLuint) (ctx->List.ListBase + translate_id(i, type, lists));
+      execute_list(ctx, list);
    }
 
    ctx->CompileFlag = save_compile_flag;
@@ -7821,6 +7878,9 @@ _mesa_init_dlist_table(struct _glapi_table *table)
    SET_StencilMaskSeparate(table, save_StencilMaskSeparate);
    SET_StencilOpSeparate(table, save_StencilOpSeparate);
 
+   /* ATI_separate_stencil */ 
+   SET_StencilFuncSeparateATI(table, save_StencilFuncSeparateATI);
+
    /* GL_ARB_imaging */
    /* Not all are supported */
    SET_BlendColor(table, save_BlendColor);
@@ -8118,6 +8178,11 @@ _mesa_init_dlist_table(struct _glapi_table *table)
    SET_BlitFramebufferEXT(table, save_BlitFramebufferEXT);
 #endif
 
+   /* ARB 30/31/32. GL_ARB_shader_objects, GL_ARB_vertex/fragment_shader */
+   SET_BindAttribLocationARB(table, exec_BindAttribLocationARB);
+   SET_GetAttribLocationARB(table, exec_GetAttribLocationARB);
+   /* XXX additional functions need to be implemented here! */
+
    /* 299. GL_EXT_blend_equation_separate */
    SET_BlendEquationSeparateEXT(table, save_BlendEquationSeparateEXT);
 
diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c
index 6db198ea3bb..13cfa0e756e 100644
--- a/src/mesa/main/drawpix.c
+++ b/src/mesa/main/drawpix.c
@@ -218,7 +218,7 @@ _mesa_Bitmap( GLsizei width, GLsizei height,
 
    if (ctx->RenderMode == GL_RENDER) {
       /* Truncate, to satisfy conformance tests (matches SGI's OpenGL). */
-      const GLfloat epsilon = (const GLfloat)0.0001;
+      const GLfloat epsilon = 0.0001F;
       GLint x = IFLOOR(ctx->Current.RasterPos[0] + epsilon - xorig);
       GLint y = IFLOOR(ctx->Current.RasterPos[1] + epsilon - yorig);
 
diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c
index 9dc55d4e69f..248df1badca 100644
--- a/src/mesa/main/enable.c
+++ b/src/mesa/main/enable.c
@@ -5,9 +5,9 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.0.3
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -38,6 +38,7 @@
 #include "enums.h"
 #include "math/m_matrix.h"
 #include "math/m_xform.h"
+#include "api_arrayelt.h"
 
 
 
@@ -136,6 +137,9 @@ client_state(GLcontext *ctx, GLenum cap, GLboolean state)
 
    FLUSH_VERTICES(ctx, _NEW_ARRAY);
    ctx->Array.NewState |= flag;
+
+   _ae_invalidate_state(ctx, _NEW_ARRAY);
+
    *var = state;
 
    if (state)
@@ -373,18 +377,24 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
             return;
          FLUSH_VERTICES(ctx, _NEW_LIGHT);
          ctx->Light.Enabled = state;
+         if (ctx->Light.Enabled && ctx->Light.Model.TwoSide)
+            ctx->_TriangleCaps |= DD_TRI_LIGHT_TWOSIDE;
+         else
+            ctx->_TriangleCaps &= ~DD_TRI_LIGHT_TWOSIDE;
          break;
       case GL_LINE_SMOOTH:
          if (ctx->Line.SmoothFlag == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_LINE);
          ctx->Line.SmoothFlag = state;
+         ctx->_TriangleCaps ^= DD_LINE_SMOOTH;
          break;
       case GL_LINE_STIPPLE:
          if (ctx->Line.StippleFlag == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_LINE);
          ctx->Line.StippleFlag = state;
+         ctx->_TriangleCaps ^= DD_LINE_STIPPLE;
          break;
       case GL_INDEX_LOGIC_OP:
          if (ctx->Color.IndexLogicOpEnabled == state)
@@ -523,18 +533,21 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
             return;
          FLUSH_VERTICES(ctx, _NEW_POINT);
          ctx->Point.SmoothFlag = state;
+         ctx->_TriangleCaps ^= DD_POINT_SMOOTH;
          break;
       case GL_POLYGON_SMOOTH:
          if (ctx->Polygon.SmoothFlag == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_POLYGON);
          ctx->Polygon.SmoothFlag = state;
+         ctx->_TriangleCaps ^= DD_TRI_SMOOTH;
          break;
       case GL_POLYGON_STIPPLE:
          if (ctx->Polygon.StippleFlag == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_POLYGON);
          ctx->Polygon.StippleFlag = state;
+         ctx->_TriangleCaps ^= DD_TRI_STIPPLE;
          break;
       case GL_POLYGON_OFFSET_POINT:
          if (ctx->Polygon.OffsetPoint == state)
@@ -885,6 +898,10 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
             return;
          FLUSH_VERTICES(ctx, _NEW_STENCIL);
          ctx->Stencil.TestTwoSide = state;
+         if (state)
+            ctx->_TriangleCaps |= DD_TRI_TWOSTENCIL;
+         else
+            ctx->_TriangleCaps &= ~DD_TRI_TWOSTENCIL;
          break;
 
 #if FEATURE_ARB_fragment_program
diff --git a/src/mesa/main/enums.c b/src/mesa/main/enums.c
index 8ce6b51d17c..4796f3027ad 100644
--- a/src/mesa/main/enums.c
+++ b/src/mesa/main/enums.c
@@ -88,7 +88,6 @@ LONGSTRING static const char enum_string_table[] =
    "GL_AND_INVERTED\0"
    "GL_AND_REVERSE\0"
    "GL_ARRAY_BUFFER\0"
-   "GL_ARRAY_BUFFER_ARB\0"
    "GL_ARRAY_BUFFER_BINDING\0"
    "GL_ARRAY_BUFFER_BINDING_ARB\0"
    "GL_ATTACHED_SHADERS\0"
@@ -456,7 +455,6 @@ LONGSTRING static const char enum_string_table[] =
    "GL_EDGE_FLAG_ARRAY_POINTER\0"
    "GL_EDGE_FLAG_ARRAY_STRIDE\0"
    "GL_ELEMENT_ARRAY_BUFFER\0"
-   "GL_ELEMENT_ARRAY_BUFFER_ARB\0"
    "GL_ELEMENT_ARRAY_BUFFER_BINDING\0"
    "GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB\0"
    "GL_EMISSION\0"
@@ -1787,7 +1785,7 @@ LONGSTRING static const char enum_string_table[] =
    "GL_ZOOM_Y\0"
    ;
 
-static const enum_elt all_enums[1750] =
+static const enum_elt all_enums[1748] =
 {
    {     0, 0x00000600 }, /* GL_2D */
    {     6, 0x00001407 }, /* GL_2_BYTES */
@@ -1842,2990 +1840,2988 @@ static const enum_elt all_enums[1750] =
    {   830, 0x00001504 }, /* GL_AND_INVERTED */
    {   846, 0x00001502 }, /* GL_AND_REVERSE */
    {   861, 0x00008892 }, /* GL_ARRAY_BUFFER */
-   {   877, 0x00008892 }, /* GL_ARRAY_BUFFER_ARB */
-   {   897, 0x00008894 }, /* GL_ARRAY_BUFFER_BINDING */
-   {   921, 0x00008894 }, /* GL_ARRAY_BUFFER_BINDING_ARB */
-   {   949, 0x00008B85 }, /* GL_ATTACHED_SHADERS */
-   {   969, 0x00008645 }, /* GL_ATTRIB_ARRAY_POINTER_NV */
-   {   996, 0x00008623 }, /* GL_ATTRIB_ARRAY_SIZE_NV */
-   {  1020, 0x00008624 }, /* GL_ATTRIB_ARRAY_STRIDE_NV */
-   {  1046, 0x00008625 }, /* GL_ATTRIB_ARRAY_TYPE_NV */
-   {  1070, 0x00000BB0 }, /* GL_ATTRIB_STACK_DEPTH */
-   {  1092, 0x00000D80 }, /* GL_AUTO_NORMAL */
-   {  1107, 0x00000409 }, /* GL_AUX0 */
-   {  1115, 0x0000040A }, /* GL_AUX1 */
-   {  1123, 0x0000040B }, /* GL_AUX2 */
-   {  1131, 0x0000040C }, /* GL_AUX3 */
-   {  1139, 0x00000C00 }, /* GL_AUX_BUFFERS */
-   {  1154, 0x00000405 }, /* GL_BACK */
-   {  1162, 0x00000402 }, /* GL_BACK_LEFT */
-   {  1175, 0x00000403 }, /* GL_BACK_RIGHT */
-   {  1189, 0x000080E0 }, /* GL_BGR */
-   {  1196, 0x000080E1 }, /* GL_BGRA */
-   {  1204, 0x00001A00 }, /* GL_BITMAP */
-   {  1214, 0x00000704 }, /* GL_BITMAP_TOKEN */
-   {  1230, 0x00000BE2 }, /* GL_BLEND */
-   {  1239, 0x00008005 }, /* GL_BLEND_COLOR */
-   {  1254, 0x00008005 }, /* GL_BLEND_COLOR_EXT */
-   {  1273, 0x00000BE0 }, /* GL_BLEND_DST */
-   {  1286, 0x000080CA }, /* GL_BLEND_DST_ALPHA */
-   {  1305, 0x000080C8 }, /* GL_BLEND_DST_RGB */
-   {  1322, 0x00008009 }, /* GL_BLEND_EQUATION */
-   {  1340, 0x0000883D }, /* GL_BLEND_EQUATION_ALPHA */
-   {  1364, 0x0000883D }, /* GL_BLEND_EQUATION_ALPHA_EXT */
-   {  1392, 0x00008009 }, /* GL_BLEND_EQUATION_EXT */
-   {  1414, 0x00008009 }, /* GL_BLEND_EQUATION_RGB_EXT */
-   {  1440, 0x00000BE1 }, /* GL_BLEND_SRC */
-   {  1453, 0x000080CB }, /* GL_BLEND_SRC_ALPHA */
-   {  1472, 0x000080C9 }, /* GL_BLEND_SRC_RGB */
-   {  1489, 0x00001905 }, /* GL_BLUE */
-   {  1497, 0x00000D1B }, /* GL_BLUE_BIAS */
-   {  1510, 0x00000D54 }, /* GL_BLUE_BITS */
-   {  1523, 0x00000D1A }, /* GL_BLUE_SCALE */
-   {  1537, 0x00008B56 }, /* GL_BOOL */
-   {  1545, 0x00008B56 }, /* GL_BOOL_ARB */
-   {  1557, 0x00008B57 }, /* GL_BOOL_VEC2 */
-   {  1570, 0x00008B57 }, /* GL_BOOL_VEC2_ARB */
-   {  1587, 0x00008B58 }, /* GL_BOOL_VEC3 */
-   {  1600, 0x00008B58 }, /* GL_BOOL_VEC3_ARB */
-   {  1617, 0x00008B59 }, /* GL_BOOL_VEC4 */
-   {  1630, 0x00008B59 }, /* GL_BOOL_VEC4_ARB */
-   {  1647, 0x000088BB }, /* GL_BUFFER_ACCESS */
-   {  1664, 0x000088BB }, /* GL_BUFFER_ACCESS_ARB */
-   {  1685, 0x000088BC }, /* GL_BUFFER_MAPPED */
-   {  1702, 0x000088BC }, /* GL_BUFFER_MAPPED_ARB */
-   {  1723, 0x000088BD }, /* GL_BUFFER_MAP_POINTER */
-   {  1745, 0x000088BD }, /* GL_BUFFER_MAP_POINTER_ARB */
-   {  1771, 0x00008764 }, /* GL_BUFFER_SIZE */
-   {  1786, 0x00008764 }, /* GL_BUFFER_SIZE_ARB */
-   {  1805, 0x00008765 }, /* GL_BUFFER_USAGE */
-   {  1821, 0x00008765 }, /* GL_BUFFER_USAGE_ARB */
-   {  1841, 0x00001400 }, /* GL_BYTE */
-   {  1849, 0x00002A24 }, /* GL_C3F_V3F */
-   {  1860, 0x00002A26 }, /* GL_C4F_N3F_V3F */
-   {  1875, 0x00002A22 }, /* GL_C4UB_V2F */
-   {  1887, 0x00002A23 }, /* GL_C4UB_V3F */
-   {  1899, 0x00000901 }, /* GL_CCW */
-   {  1906, 0x00002900 }, /* GL_CLAMP */
-   {  1915, 0x0000812D }, /* GL_CLAMP_TO_BORDER */
-   {  1934, 0x0000812D }, /* GL_CLAMP_TO_BORDER_ARB */
-   {  1957, 0x0000812D }, /* GL_CLAMP_TO_BORDER_SGIS */
-   {  1981, 0x0000812F }, /* GL_CLAMP_TO_EDGE */
-   {  1998, 0x0000812F }, /* GL_CLAMP_TO_EDGE_SGIS */
-   {  2020, 0x00001500 }, /* GL_CLEAR */
-   {  2029, 0x000084E1 }, /* GL_CLIENT_ACTIVE_TEXTURE */
-   {  2054, 0x000084E1 }, /* GL_CLIENT_ACTIVE_TEXTURE_ARB */
-   {  2083, 0xFFFFFFFF }, /* GL_CLIENT_ALL_ATTRIB_BITS */
-   {  2109, 0x00000BB1 }, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
-   {  2138, 0x00000001 }, /* GL_CLIENT_PIXEL_STORE_BIT */
-   {  2164, 0x00000002 }, /* GL_CLIENT_VERTEX_ARRAY_BIT */
-   {  2191, 0x00003000 }, /* GL_CLIP_PLANE0 */
-   {  2206, 0x00003001 }, /* GL_CLIP_PLANE1 */
-   {  2221, 0x00003002 }, /* GL_CLIP_PLANE2 */
-   {  2236, 0x00003003 }, /* GL_CLIP_PLANE3 */
-   {  2251, 0x00003004 }, /* GL_CLIP_PLANE4 */
-   {  2266, 0x00003005 }, /* GL_CLIP_PLANE5 */
-   {  2281, 0x000080F0 }, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
-   {  2314, 0x00000A00 }, /* GL_COEFF */
-   {  2323, 0x00001800 }, /* GL_COLOR */
-   {  2332, 0x00008076 }, /* GL_COLOR_ARRAY */
-   {  2347, 0x00008898 }, /* GL_COLOR_ARRAY_BUFFER_BINDING */
-   {  2377, 0x00008898 }, /* GL_COLOR_ARRAY_BUFFER_BINDING_ARB */
-   {  2411, 0x00008090 }, /* GL_COLOR_ARRAY_POINTER */
-   {  2434, 0x00008081 }, /* GL_COLOR_ARRAY_SIZE */
-   {  2454, 0x00008083 }, /* GL_COLOR_ARRAY_STRIDE */
-   {  2476, 0x00008082 }, /* GL_COLOR_ARRAY_TYPE */
-   {  2496, 0x00008CE0 }, /* GL_COLOR_ATTACHMENT0_EXT */
-   {  2521, 0x00008CEA }, /* GL_COLOR_ATTACHMENT10_EXT */
-   {  2547, 0x00008CEB }, /* GL_COLOR_ATTACHMENT11_EXT */
-   {  2573, 0x00008CEC }, /* GL_COLOR_ATTACHMENT12_EXT */
-   {  2599, 0x00008CED }, /* GL_COLOR_ATTACHMENT13_EXT */
-   {  2625, 0x00008CEE }, /* GL_COLOR_ATTACHMENT14_EXT */
-   {  2651, 0x00008CEF }, /* GL_COLOR_ATTACHMENT15_EXT */
-   {  2677, 0x00008CE1 }, /* GL_COLOR_ATTACHMENT1_EXT */
-   {  2702, 0x00008CE2 }, /* GL_COLOR_ATTACHMENT2_EXT */
-   {  2727, 0x00008CE3 }, /* GL_COLOR_ATTACHMENT3_EXT */
-   {  2752, 0x00008CE4 }, /* GL_COLOR_ATTACHMENT4_EXT */
-   {  2777, 0x00008CE5 }, /* GL_COLOR_ATTACHMENT5_EXT */
-   {  2802, 0x00008CE6 }, /* GL_COLOR_ATTACHMENT6_EXT */
-   {  2827, 0x00008CE7 }, /* GL_COLOR_ATTACHMENT7_EXT */
-   {  2852, 0x00008CE8 }, /* GL_COLOR_ATTACHMENT8_EXT */
-   {  2877, 0x00008CE9 }, /* GL_COLOR_ATTACHMENT9_EXT */
-   {  2902, 0x00004000 }, /* GL_COLOR_BUFFER_BIT */
-   {  2922, 0x00000C22 }, /* GL_COLOR_CLEAR_VALUE */
-   {  2943, 0x00001900 }, /* GL_COLOR_INDEX */
-   {  2958, 0x00001603 }, /* GL_COLOR_INDEXES */
-   {  2975, 0x00000BF2 }, /* GL_COLOR_LOGIC_OP */
-   {  2993, 0x00000B57 }, /* GL_COLOR_MATERIAL */
-   {  3011, 0x00000B55 }, /* GL_COLOR_MATERIAL_FACE */
-   {  3034, 0x00000B56 }, /* GL_COLOR_MATERIAL_PARAMETER */
-   {  3062, 0x000080B1 }, /* GL_COLOR_MATRIX */
-   {  3078, 0x000080B1 }, /* GL_COLOR_MATRIX_SGI */
-   {  3098, 0x000080B2 }, /* GL_COLOR_MATRIX_STACK_DEPTH */
-   {  3126, 0x000080B2 }, /* GL_COLOR_MATRIX_STACK_DEPTH_SGI */
-   {  3158, 0x00008458 }, /* GL_COLOR_SUM */
-   {  3171, 0x00008458 }, /* GL_COLOR_SUM_ARB */
-   {  3188, 0x000080D0 }, /* GL_COLOR_TABLE */
-   {  3203, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE */
-   {  3229, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE_EXT */
-   {  3259, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE_SGI */
-   {  3289, 0x000080D7 }, /* GL_COLOR_TABLE_BIAS */
-   {  3309, 0x000080D7 }, /* GL_COLOR_TABLE_BIAS_SGI */
-   {  3333, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE */
-   {  3358, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE_EXT */
-   {  3387, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE_SGI */
-   {  3416, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT */
-   {  3438, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT_EXT */
-   {  3464, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT_SGI */
-   {  3490, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE */
-   {  3516, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE_EXT */
-   {  3546, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE_SGI */
-   {  3576, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE */
-   {  3606, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE_EXT */
-   {  3640, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE_SGI */
-   {  3674, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE */
-   {  3704, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE_EXT */
-   {  3738, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE_SGI */
-   {  3772, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE */
-   {  3796, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE_EXT */
-   {  3824, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE_SGI */
-   {  3852, 0x000080D6 }, /* GL_COLOR_TABLE_SCALE */
-   {  3873, 0x000080D6 }, /* GL_COLOR_TABLE_SCALE_SGI */
-   {  3898, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH */
-   {  3919, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH_EXT */
-   {  3944, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH_SGI */
-   {  3969, 0x00000C23 }, /* GL_COLOR_WRITEMASK */
-   {  3988, 0x00008570 }, /* GL_COMBINE */
-   {  3999, 0x00008503 }, /* GL_COMBINE4 */
-   {  4011, 0x00008572 }, /* GL_COMBINE_ALPHA */
-   {  4028, 0x00008572 }, /* GL_COMBINE_ALPHA_ARB */
-   {  4049, 0x00008572 }, /* GL_COMBINE_ALPHA_EXT */
-   {  4070, 0x00008570 }, /* GL_COMBINE_ARB */
-   {  4085, 0x00008570 }, /* GL_COMBINE_EXT */
-   {  4100, 0x00008571 }, /* GL_COMBINE_RGB */
-   {  4115, 0x00008571 }, /* GL_COMBINE_RGB_ARB */
-   {  4134, 0x00008571 }, /* GL_COMBINE_RGB_EXT */
-   {  4153, 0x0000884E }, /* GL_COMPARE_REF_DEPTH_TO_TEXTURE_EXT */
-   {  4189, 0x0000884E }, /* GL_COMPARE_R_TO_TEXTURE */
-   {  4213, 0x0000884E }, /* GL_COMPARE_R_TO_TEXTURE_ARB */
-   {  4241, 0x00001300 }, /* GL_COMPILE */
-   {  4252, 0x00001301 }, /* GL_COMPILE_AND_EXECUTE */
-   {  4275, 0x00008B81 }, /* GL_COMPILE_STATUS */
-   {  4293, 0x000084E9 }, /* GL_COMPRESSED_ALPHA */
-   {  4313, 0x000084E9 }, /* GL_COMPRESSED_ALPHA_ARB */
-   {  4337, 0x000084EC }, /* GL_COMPRESSED_INTENSITY */
-   {  4361, 0x000084EC }, /* GL_COMPRESSED_INTENSITY_ARB */
-   {  4389, 0x000084EA }, /* GL_COMPRESSED_LUMINANCE */
-   {  4413, 0x000084EB }, /* GL_COMPRESSED_LUMINANCE_ALPHA */
-   {  4443, 0x000084EB }, /* GL_COMPRESSED_LUMINANCE_ALPHA_ARB */
-   {  4477, 0x000084EA }, /* GL_COMPRESSED_LUMINANCE_ARB */
-   {  4505, 0x000084ED }, /* GL_COMPRESSED_RGB */
-   {  4523, 0x000084EE }, /* GL_COMPRESSED_RGBA */
-   {  4542, 0x000084EE }, /* GL_COMPRESSED_RGBA_ARB */
-   {  4565, 0x000086B1 }, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
-   {  4594, 0x000083F1 }, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
-   {  4627, 0x000083F2 }, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
-   {  4660, 0x000083F3 }, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
-   {  4693, 0x000084ED }, /* GL_COMPRESSED_RGB_ARB */
-   {  4715, 0x000086B0 }, /* GL_COMPRESSED_RGB_FXT1_3DFX */
-   {  4743, 0x000083F0 }, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
-   {  4775, 0x000086A3 }, /* GL_COMPRESSED_TEXTURE_FORMATS */
-   {  4805, 0x00008576 }, /* GL_CONSTANT */
-   {  4817, 0x00008003 }, /* GL_CONSTANT_ALPHA */
-   {  4835, 0x00008003 }, /* GL_CONSTANT_ALPHA_EXT */
-   {  4857, 0x00008576 }, /* GL_CONSTANT_ARB */
-   {  4873, 0x00001207 }, /* GL_CONSTANT_ATTENUATION */
-   {  4897, 0x00008151 }, /* GL_CONSTANT_BORDER_HP */
-   {  4919, 0x00008001 }, /* GL_CONSTANT_COLOR */
-   {  4937, 0x00008001 }, /* GL_CONSTANT_COLOR_EXT */
-   {  4959, 0x00008576 }, /* GL_CONSTANT_EXT */
-   {  4975, 0x00008010 }, /* GL_CONVOLUTION_1D */
-   {  4993, 0x00008011 }, /* GL_CONVOLUTION_2D */
-   {  5011, 0x00008154 }, /* GL_CONVOLUTION_BORDER_COLOR */
-   {  5039, 0x00008154 }, /* GL_CONVOLUTION_BORDER_COLOR_HP */
-   {  5070, 0x00008013 }, /* GL_CONVOLUTION_BORDER_MODE */
-   {  5097, 0x00008013 }, /* GL_CONVOLUTION_BORDER_MODE_EXT */
-   {  5128, 0x00008015 }, /* GL_CONVOLUTION_FILTER_BIAS */
-   {  5155, 0x00008015 }, /* GL_CONVOLUTION_FILTER_BIAS_EXT */
-   {  5186, 0x00008014 }, /* GL_CONVOLUTION_FILTER_SCALE */
-   {  5214, 0x00008014 }, /* GL_CONVOLUTION_FILTER_SCALE_EXT */
-   {  5246, 0x00008017 }, /* GL_CONVOLUTION_FORMAT */
-   {  5268, 0x00008017 }, /* GL_CONVOLUTION_FORMAT_EXT */
-   {  5294, 0x00008019 }, /* GL_CONVOLUTION_HEIGHT */
-   {  5316, 0x00008019 }, /* GL_CONVOLUTION_HEIGHT_EXT */
-   {  5342, 0x00008018 }, /* GL_CONVOLUTION_WIDTH */
-   {  5363, 0x00008018 }, /* GL_CONVOLUTION_WIDTH_EXT */
-   {  5388, 0x00008862 }, /* GL_COORD_REPLACE */
-   {  5405, 0x00008862 }, /* GL_COORD_REPLACE_ARB */
-   {  5426, 0x00008862 }, /* GL_COORD_REPLACE_NV */
-   {  5446, 0x00001503 }, /* GL_COPY */
-   {  5454, 0x0000150C }, /* GL_COPY_INVERTED */
-   {  5471, 0x00000706 }, /* GL_COPY_PIXEL_TOKEN */
-   {  5491, 0x00000B44 }, /* GL_CULL_FACE */
-   {  5504, 0x00000B45 }, /* GL_CULL_FACE_MODE */
-   {  5522, 0x000081AA }, /* GL_CULL_VERTEX_EXT */
-   {  5541, 0x000081AC }, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
-   {  5573, 0x000081AB }, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
-   {  5608, 0x00008626 }, /* GL_CURRENT_ATTRIB_NV */
-   {  5629, 0x00000001 }, /* GL_CURRENT_BIT */
-   {  5644, 0x00000B00 }, /* GL_CURRENT_COLOR */
-   {  5661, 0x00008453 }, /* GL_CURRENT_FOG_COORD */
-   {  5682, 0x00008453 }, /* GL_CURRENT_FOG_COORDINATE */
-   {  5708, 0x00000B01 }, /* GL_CURRENT_INDEX */
-   {  5725, 0x00008641 }, /* GL_CURRENT_MATRIX_ARB */
-   {  5747, 0x00008845 }, /* GL_CURRENT_MATRIX_INDEX_ARB */
-   {  5775, 0x00008641 }, /* GL_CURRENT_MATRIX_NV */
-   {  5796, 0x00008640 }, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
-   {  5830, 0x00008640 }, /* GL_CURRENT_MATRIX_STACK_DEPTH_NV */
-   {  5863, 0x00000B02 }, /* GL_CURRENT_NORMAL */
-   {  5881, 0x00008843 }, /* GL_CURRENT_PALETTE_MATRIX_ARB */
-   {  5911, 0x00008B8D }, /* GL_CURRENT_PROGRAM */
-   {  5930, 0x00008865 }, /* GL_CURRENT_QUERY */
-   {  5947, 0x00008865 }, /* GL_CURRENT_QUERY_ARB */
-   {  5968, 0x00000B04 }, /* GL_CURRENT_RASTER_COLOR */
-   {  5992, 0x00000B09 }, /* GL_CURRENT_RASTER_DISTANCE */
-   {  6019, 0x00000B05 }, /* GL_CURRENT_RASTER_INDEX */
-   {  6043, 0x00000B07 }, /* GL_CURRENT_RASTER_POSITION */
-   {  6070, 0x00000B08 }, /* GL_CURRENT_RASTER_POSITION_VALID */
-   {  6103, 0x00000B06 }, /* GL_CURRENT_RASTER_TEXTURE_COORDS */
-   {  6136, 0x00008459 }, /* GL_CURRENT_SECONDARY_COLOR */
-   {  6163, 0x00000B03 }, /* GL_CURRENT_TEXTURE_COORDS */
-   {  6189, 0x00008626 }, /* GL_CURRENT_VERTEX_ATTRIB */
-   {  6214, 0x00008626 }, /* GL_CURRENT_VERTEX_ATTRIB_ARB */
-   {  6243, 0x000086A8 }, /* GL_CURRENT_WEIGHT_ARB */
-   {  6265, 0x00000900 }, /* GL_CW */
-   {  6271, 0x0000875B }, /* GL_DEBUG_ASSERT_MESA */
-   {  6292, 0x00008759 }, /* GL_DEBUG_OBJECT_MESA */
-   {  6313, 0x0000875A }, /* GL_DEBUG_PRINT_MESA */
-   {  6333, 0x00002101 }, /* GL_DECAL */
-   {  6342, 0x00001E03 }, /* GL_DECR */
-   {  6350, 0x00008508 }, /* GL_DECR_WRAP */
-   {  6363, 0x00008508 }, /* GL_DECR_WRAP_EXT */
-   {  6380, 0x00008B80 }, /* GL_DELETE_STATUS */
-   {  6397, 0x00001801 }, /* GL_DEPTH */
-   {  6406, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT_EXT */
-   {  6430, 0x00000D1F }, /* GL_DEPTH_BIAS */
-   {  6444, 0x00000D56 }, /* GL_DEPTH_BITS */
-   {  6458, 0x00008891 }, /* GL_DEPTH_BOUNDS_EXT */
-   {  6478, 0x00008890 }, /* GL_DEPTH_BOUNDS_TEST_EXT */
-   {  6503, 0x00000100 }, /* GL_DEPTH_BUFFER_BIT */
-   {  6523, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
-   {  6541, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
-   {  6562, 0x00001902 }, /* GL_DEPTH_COMPONENT */
-   {  6581, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
-   {  6602, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
-   {  6627, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
-   {  6653, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
-   {  6674, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
-   {  6699, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
-   {  6725, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
-   {  6746, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
-   {  6771, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
-   {  6797, 0x00000B74 }, /* GL_DEPTH_FUNC */
-   {  6811, 0x00000B70 }, /* GL_DEPTH_RANGE */
-   {  6826, 0x00000D1E }, /* GL_DEPTH_SCALE */
-   {  6841, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
-   {  6861, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-   {  6889, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-   {  6917, 0x00000B71 }, /* GL_DEPTH_TEST */
-   {  6931, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
-   {  6953, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
-   {  6979, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
-   {  6998, 0x00001201 }, /* GL_DIFFUSE */
-   {  7009, 0x00000BD0 }, /* GL_DITHER */
-   {  7019, 0x00000A02 }, /* GL_DOMAIN */
-   {  7029, 0x00001100 }, /* GL_DONT_CARE */
-   {  7042, 0x000086AE }, /* GL_DOT3_RGB */
-   {  7054, 0x000086AF }, /* GL_DOT3_RGBA */
-   {  7067, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
-   {  7084, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
-   {  7101, 0x000086AE }, /* GL_DOT3_RGB_ARB */
-   {  7117, 0x00008740 }, /* GL_DOT3_RGB_EXT */
-   {  7133, 0x0000140A }, /* GL_DOUBLE */
-   {  7143, 0x00000C32 }, /* GL_DOUBLEBUFFER */
-   {  7159, 0x00000C01 }, /* GL_DRAW_BUFFER */
-   {  7174, 0x00008825 }, /* GL_DRAW_BUFFER0 */
-   {  7190, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
-   {  7210, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
-   {  7230, 0x00008826 }, /* GL_DRAW_BUFFER1 */
-   {  7246, 0x0000882F }, /* GL_DRAW_BUFFER10 */
-   {  7263, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
-   {  7284, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
-   {  7305, 0x00008830 }, /* GL_DRAW_BUFFER11 */
-   {  7322, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
-   {  7343, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
-   {  7364, 0x00008831 }, /* GL_DRAW_BUFFER12 */
-   {  7381, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
-   {  7402, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
-   {  7423, 0x00008832 }, /* GL_DRAW_BUFFER13 */
-   {  7440, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
-   {  7461, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
-   {  7482, 0x00008833 }, /* GL_DRAW_BUFFER14 */
-   {  7499, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
-   {  7520, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
-   {  7541, 0x00008834 }, /* GL_DRAW_BUFFER15 */
-   {  7558, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
-   {  7579, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
-   {  7600, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
-   {  7620, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
-   {  7640, 0x00008827 }, /* GL_DRAW_BUFFER2 */
-   {  7656, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
-   {  7676, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
-   {  7696, 0x00008828 }, /* GL_DRAW_BUFFER3 */
-   {  7712, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
-   {  7732, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
-   {  7752, 0x00008829 }, /* GL_DRAW_BUFFER4 */
-   {  7768, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
-   {  7788, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
-   {  7808, 0x0000882A }, /* GL_DRAW_BUFFER5 */
-   {  7824, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
-   {  7844, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
-   {  7864, 0x0000882B }, /* GL_DRAW_BUFFER6 */
-   {  7880, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
-   {  7900, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
-   {  7920, 0x0000882C }, /* GL_DRAW_BUFFER7 */
-   {  7936, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
-   {  7956, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
-   {  7976, 0x0000882D }, /* GL_DRAW_BUFFER8 */
-   {  7992, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
-   {  8012, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
-   {  8032, 0x0000882E }, /* GL_DRAW_BUFFER9 */
-   {  8048, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
-   {  8068, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
-   {  8088, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-   {  8120, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
-   {  8144, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
-   {  8164, 0x00000304 }, /* GL_DST_ALPHA */
-   {  8177, 0x00000306 }, /* GL_DST_COLOR */
-   {  8190, 0x000088EA }, /* GL_DYNAMIC_COPY */
-   {  8206, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
-   {  8226, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
-   {  8242, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
-   {  8262, 0x000088E9 }, /* GL_DYNAMIC_READ */
-   {  8278, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
-   {  8298, 0x00000B43 }, /* GL_EDGE_FLAG */
-   {  8311, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
-   {  8330, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-   {  8364, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
-   {  8402, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
-   {  8429, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-   {  8455, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
-   {  8479, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER_ARB */
-   {  8507, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-   {  8539, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
-   {  8575, 0x00001600 }, /* GL_EMISSION */
-   {  8587, 0x00002000 }, /* GL_ENABLE_BIT */
-   {  8601, 0x00000202 }, /* GL_EQUAL */
-   {  8610, 0x00001509 }, /* GL_EQUIV */
-   {  8619, 0x00010000 }, /* GL_EVAL_BIT */
-   {  8631, 0x00000800 }, /* GL_EXP */
-   {  8638, 0x00000801 }, /* GL_EXP2 */
-   {  8646, 0x00001F03 }, /* GL_EXTENSIONS */
-   {  8660, 0x00002400 }, /* GL_EYE_LINEAR */
-   {  8674, 0x00002502 }, /* GL_EYE_PLANE */
-   {  8687, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
-   {  8712, 0x0000855B }, /* GL_EYE_RADIAL_NV */
-   {  8729, 0x00000000 }, /* GL_FALSE */
-   {  8738, 0x00001101 }, /* GL_FASTEST */
-   {  8749, 0x00001C01 }, /* GL_FEEDBACK */
-   {  8761, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
-   {  8788, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
-   {  8812, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
-   {  8836, 0x00001B02 }, /* GL_FILL */
-   {  8844, 0x00001D00 }, /* GL_FLAT */
-   {  8852, 0x00001406 }, /* GL_FLOAT */
-   {  8861, 0x00008B5A }, /* GL_FLOAT_MAT2 */
-   {  8875, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
-   {  8893, 0x00008B5B }, /* GL_FLOAT_MAT3 */
-   {  8907, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
-   {  8925, 0x00008B5C }, /* GL_FLOAT_MAT4 */
-   {  8939, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
-   {  8957, 0x00008B50 }, /* GL_FLOAT_VEC2 */
-   {  8971, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
-   {  8989, 0x00008B51 }, /* GL_FLOAT_VEC3 */
-   {  9003, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
-   {  9021, 0x00008B52 }, /* GL_FLOAT_VEC4 */
-   {  9035, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
-   {  9053, 0x00000B60 }, /* GL_FOG */
-   {  9060, 0x00000080 }, /* GL_FOG_BIT */
-   {  9071, 0x00000B66 }, /* GL_FOG_COLOR */
-   {  9084, 0x00008451 }, /* GL_FOG_COORD */
-   {  9097, 0x00008451 }, /* GL_FOG_COORDINATE */
-   {  9115, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
-   {  9139, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-   {  9178, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
-   {  9221, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-   {  9253, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-   {  9284, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-   {  9313, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
-   {  9338, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
-   {  9357, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
-   {  9391, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
-   {  9418, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
-   {  9444, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
-   {  9468, 0x00008450 }, /* GL_FOG_COORD_SRC */
-   {  9485, 0x00000B62 }, /* GL_FOG_DENSITY */
-   {  9500, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
-   {  9524, 0x00000B64 }, /* GL_FOG_END */
-   {  9535, 0x00000C54 }, /* GL_FOG_HINT */
-   {  9547, 0x00000B61 }, /* GL_FOG_INDEX */
-   {  9560, 0x00000B65 }, /* GL_FOG_MODE */
-   {  9572, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
-   {  9591, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
-   {  9616, 0x00000B63 }, /* GL_FOG_START */
-   {  9629, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
-   {  9647, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
-   {  9671, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
-   {  9690, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
-   {  9713, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-   {  9748, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
-   {  9790, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
-   {  9832, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
-   {  9881, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
-   {  9933, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
-   {  9977, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
-   { 10021, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
-   { 10048, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
-   { 10076, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
-   { 10095, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
-   { 10136, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-   { 10177, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-   { 10219, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-   { 10270, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-   { 10308, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
-   { 10357, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-   { 10399, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-   { 10431, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
-   { 10462, 0x00000404 }, /* GL_FRONT */
-   { 10471, 0x00000408 }, /* GL_FRONT_AND_BACK */
-   { 10489, 0x00000B46 }, /* GL_FRONT_FACE */
-   { 10503, 0x00000400 }, /* GL_FRONT_LEFT */
-   { 10517, 0x00000401 }, /* GL_FRONT_RIGHT */
-   { 10532, 0x00008006 }, /* GL_FUNC_ADD */
-   { 10544, 0x00008006 }, /* GL_FUNC_ADD_EXT */
-   { 10560, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
-   { 10585, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
-   { 10614, 0x0000800A }, /* GL_FUNC_SUBTRACT */
-   { 10631, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
-   { 10652, 0x00008191 }, /* GL_GENERATE_MIPMAP */
-   { 10671, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
-   { 10695, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
-   { 10724, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
-   { 10748, 0x00000206 }, /* GL_GEQUAL */
-   { 10758, 0x00008009 }, /* GL_GL_BLEND_EQUATION_RGB */
-   { 10783, 0x00008C4A }, /* GL_GL_COMPRESSED_SLUMINANCE */
-   { 10811, 0x00008C4B }, /* GL_GL_COMPRESSED_SLUMINANCE_ALPHA */
-   { 10845, 0x00008C48 }, /* GL_GL_COMPRESSED_SRGB */
-   { 10867, 0x00008C49 }, /* GL_GL_COMPRESSED_SRGB_ALPHA */
-   { 10895, 0x0000845F }, /* GL_GL_CURRENT_RASTER_SECONDARY_COLOR */
-   { 10932, 0x00008B65 }, /* GL_GL_FLOAT_MAT2x3 */
-   { 10951, 0x00008B66 }, /* GL_GL_FLOAT_MAT2x4 */
-   { 10970, 0x00008B67 }, /* GL_GL_FLOAT_MAT3x2 */
-   { 10989, 0x00008B68 }, /* GL_GL_FLOAT_MAT3x4 */
-   { 11008, 0x00008B69 }, /* GL_GL_FLOAT_MAT4x2 */
-   { 11027, 0x00008B6A }, /* GL_GL_FLOAT_MAT4x3 */
-   { 11046, 0x000088EB }, /* GL_GL_PIXEL_PACK_BUFFER */
-   { 11070, 0x000088ED }, /* GL_GL_PIXEL_PACK_BUFFER_BINDING */
-   { 11102, 0x000088EC }, /* GL_GL_PIXEL_UNPACK_BUFFER */
-   { 11128, 0x000088EF }, /* GL_GL_PIXEL_UNPACK_BUFFER_BINDING */
-   { 11162, 0x00008C46 }, /* GL_GL_SLUMINANCE */
-   { 11179, 0x00008C47 }, /* GL_GL_SLUMINANCE8 */
-   { 11197, 0x00008C45 }, /* GL_GL_SLUMINANCE8_ALPHA8 */
-   { 11222, 0x00008C44 }, /* GL_GL_SLUMINANCE_ALPHA */
-   { 11245, 0x00008C40 }, /* GL_GL_SRGB */
-   { 11256, 0x00008C41 }, /* GL_GL_SRGB8 */
-   { 11268, 0x00008C43 }, /* GL_GL_SRGB8_ALPHA8 */
-   { 11287, 0x00008C42 }, /* GL_GL_SRGB_ALPHA */
-   { 11304, 0x00000204 }, /* GL_GREATER */
-   { 11315, 0x00001904 }, /* GL_GREEN */
-   { 11324, 0x00000D19 }, /* GL_GREEN_BIAS */
-   { 11338, 0x00000D53 }, /* GL_GREEN_BITS */
-   { 11352, 0x00000D18 }, /* GL_GREEN_SCALE */
-   { 11367, 0x00008000 }, /* GL_HINT_BIT */
-   { 11379, 0x00008024 }, /* GL_HISTOGRAM */
-   { 11392, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
-   { 11416, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
-   { 11444, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
-   { 11467, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
-   { 11494, 0x00008024 }, /* GL_HISTOGRAM_EXT */
-   { 11511, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
-   { 11531, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
-   { 11555, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
-   { 11579, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
-   { 11607, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-   { 11635, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
-   { 11667, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
-   { 11689, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
-   { 11715, 0x0000802D }, /* GL_HISTOGRAM_SINK */
-   { 11733, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
-   { 11755, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
-   { 11774, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
-   { 11797, 0x0000862A }, /* GL_IDENTITY_NV */
-   { 11812, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
-   { 11832, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-   { 11872, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-   { 11910, 0x00001E02 }, /* GL_INCR */
-   { 11918, 0x00008507 }, /* GL_INCR_WRAP */
-   { 11931, 0x00008507 }, /* GL_INCR_WRAP_EXT */
-   { 11948, 0x00008077 }, /* GL_INDEX_ARRAY */
-   { 11963, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-   { 11993, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
-   { 12027, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
-   { 12050, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
-   { 12072, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
-   { 12092, 0x00000D51 }, /* GL_INDEX_BITS */
-   { 12106, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
-   { 12127, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
-   { 12145, 0x00000C30 }, /* GL_INDEX_MODE */
-   { 12159, 0x00000D13 }, /* GL_INDEX_OFFSET */
-   { 12175, 0x00000D12 }, /* GL_INDEX_SHIFT */
-   { 12190, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
-   { 12209, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
-   { 12228, 0x00001404 }, /* GL_INT */
-   { 12235, 0x00008049 }, /* GL_INTENSITY */
-   { 12248, 0x0000804C }, /* GL_INTENSITY12 */
-   { 12263, 0x0000804C }, /* GL_INTENSITY12_EXT */
-   { 12282, 0x0000804D }, /* GL_INTENSITY16 */
-   { 12297, 0x0000804D }, /* GL_INTENSITY16_EXT */
-   { 12316, 0x0000804A }, /* GL_INTENSITY4 */
-   { 12330, 0x0000804A }, /* GL_INTENSITY4_EXT */
-   { 12348, 0x0000804B }, /* GL_INTENSITY8 */
-   { 12362, 0x0000804B }, /* GL_INTENSITY8_EXT */
-   { 12380, 0x00008049 }, /* GL_INTENSITY_EXT */
-   { 12397, 0x00008575 }, /* GL_INTERPOLATE */
-   { 12412, 0x00008575 }, /* GL_INTERPOLATE_ARB */
-   { 12431, 0x00008575 }, /* GL_INTERPOLATE_EXT */
-   { 12450, 0x00008B53 }, /* GL_INT_VEC2 */
-   { 12462, 0x00008B53 }, /* GL_INT_VEC2_ARB */
-   { 12478, 0x00008B54 }, /* GL_INT_VEC3 */
-   { 12490, 0x00008B54 }, /* GL_INT_VEC3_ARB */
-   { 12506, 0x00008B55 }, /* GL_INT_VEC4 */
-   { 12518, 0x00008B55 }, /* GL_INT_VEC4_ARB */
-   { 12534, 0x00000500 }, /* GL_INVALID_ENUM */
-   { 12550, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
-   { 12587, 0x00000502 }, /* GL_INVALID_OPERATION */
-   { 12608, 0x00000501 }, /* GL_INVALID_VALUE */
-   { 12625, 0x0000862B }, /* GL_INVERSE_NV */
-   { 12639, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
-   { 12663, 0x0000150A }, /* GL_INVERT */
-   { 12673, 0x00001E00 }, /* GL_KEEP */
-   { 12681, 0x00000406 }, /* GL_LEFT */
-   { 12689, 0x00000203 }, /* GL_LEQUAL */
-   { 12699, 0x00000201 }, /* GL_LESS */
-   { 12707, 0x00004000 }, /* GL_LIGHT0 */
-   { 12717, 0x00004001 }, /* GL_LIGHT1 */
-   { 12727, 0x00004002 }, /* GL_LIGHT2 */
-   { 12737, 0x00004003 }, /* GL_LIGHT3 */
-   { 12747, 0x00004004 }, /* GL_LIGHT4 */
-   { 12757, 0x00004005 }, /* GL_LIGHT5 */
-   { 12767, 0x00004006 }, /* GL_LIGHT6 */
-   { 12777, 0x00004007 }, /* GL_LIGHT7 */
-   { 12787, 0x00000B50 }, /* GL_LIGHTING */
-   { 12799, 0x00000040 }, /* GL_LIGHTING_BIT */
-   { 12815, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
-   { 12838, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-   { 12867, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
-   { 12900, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-   { 12928, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
-   { 12952, 0x00001B01 }, /* GL_LINE */
-   { 12960, 0x00002601 }, /* GL_LINEAR */
-   { 12970, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
-   { 12992, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-   { 13022, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-   { 13053, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
-   { 13077, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
-   { 13102, 0x00000001 }, /* GL_LINES */
-   { 13111, 0x00000004 }, /* GL_LINE_BIT */
-   { 13123, 0x00000002 }, /* GL_LINE_LOOP */
-   { 13136, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
-   { 13156, 0x00000B20 }, /* GL_LINE_SMOOTH */
-   { 13171, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
-   { 13191, 0x00000B24 }, /* GL_LINE_STIPPLE */
-   { 13207, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
-   { 13231, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
-   { 13254, 0x00000003 }, /* GL_LINE_STRIP */
-   { 13268, 0x00000702 }, /* GL_LINE_TOKEN */
-   { 13282, 0x00000B21 }, /* GL_LINE_WIDTH */
-   { 13296, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
-   { 13322, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
-   { 13342, 0x00008B82 }, /* GL_LINK_STATUS */
-   { 13357, 0x00000B32 }, /* GL_LIST_BASE */
-   { 13370, 0x00020000 }, /* GL_LIST_BIT */
-   { 13382, 0x00000B33 }, /* GL_LIST_INDEX */
-   { 13396, 0x00000B30 }, /* GL_LIST_MODE */
-   { 13409, 0x00000101 }, /* GL_LOAD */
-   { 13417, 0x00000BF1 }, /* GL_LOGIC_OP */
-   { 13429, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
-   { 13446, 0x00008CA1 }, /* GL_LOWER_LEFT */
-   { 13460, 0x00001909 }, /* GL_LUMINANCE */
-   { 13473, 0x00008041 }, /* GL_LUMINANCE12 */
-   { 13488, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
-   { 13511, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
-   { 13538, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
-   { 13560, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
-   { 13586, 0x00008041 }, /* GL_LUMINANCE12_EXT */
-   { 13605, 0x00008042 }, /* GL_LUMINANCE16 */
-   { 13620, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
-   { 13643, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
-   { 13670, 0x00008042 }, /* GL_LUMINANCE16_EXT */
-   { 13689, 0x0000803F }, /* GL_LUMINANCE4 */
-   { 13703, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
-   { 13724, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
-   { 13749, 0x0000803F }, /* GL_LUMINANCE4_EXT */
-   { 13767, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
-   { 13788, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
-   { 13813, 0x00008040 }, /* GL_LUMINANCE8 */
-   { 13827, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
-   { 13848, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
-   { 13873, 0x00008040 }, /* GL_LUMINANCE8_EXT */
-   { 13891, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
-   { 13910, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
-   { 13926, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
-   { 13946, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
-   { 13968, 0x00000D91 }, /* GL_MAP1_INDEX */
-   { 13982, 0x00000D92 }, /* GL_MAP1_NORMAL */
-   { 13997, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
-   { 14021, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
-   { 14045, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
-   { 14069, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
-   { 14093, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
-   { 14110, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
-   { 14127, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-   { 14155, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-   { 14184, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-   { 14213, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-   { 14242, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-   { 14271, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-   { 14300, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-   { 14329, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-   { 14357, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-   { 14385, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-   { 14413, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-   { 14441, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-   { 14469, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-   { 14497, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-   { 14525, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-   { 14553, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-   { 14581, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
-   { 14597, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
-   { 14617, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
-   { 14639, 0x00000DB1 }, /* GL_MAP2_INDEX */
-   { 14653, 0x00000DB2 }, /* GL_MAP2_NORMAL */
-   { 14668, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
-   { 14692, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
-   { 14716, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
-   { 14740, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
-   { 14764, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
-   { 14781, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
-   { 14798, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-   { 14826, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-   { 14855, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-   { 14884, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-   { 14913, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-   { 14942, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-   { 14971, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-   { 15000, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-   { 15028, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-   { 15056, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-   { 15084, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-   { 15112, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-   { 15140, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-   { 15168, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
-   { 15196, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-   { 15224, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-   { 15252, 0x00000D10 }, /* GL_MAP_COLOR */
-   { 15265, 0x00000D11 }, /* GL_MAP_STENCIL */
-   { 15280, 0x000088C0 }, /* GL_MATRIX0_ARB */
-   { 15295, 0x00008630 }, /* GL_MATRIX0_NV */
-   { 15309, 0x000088CA }, /* GL_MATRIX10_ARB */
-   { 15325, 0x000088CB }, /* GL_MATRIX11_ARB */
-   { 15341, 0x000088CC }, /* GL_MATRIX12_ARB */
-   { 15357, 0x000088CD }, /* GL_MATRIX13_ARB */
-   { 15373, 0x000088CE }, /* GL_MATRIX14_ARB */
-   { 15389, 0x000088CF }, /* GL_MATRIX15_ARB */
-   { 15405, 0x000088D0 }, /* GL_MATRIX16_ARB */
-   { 15421, 0x000088D1 }, /* GL_MATRIX17_ARB */
-   { 15437, 0x000088D2 }, /* GL_MATRIX18_ARB */
-   { 15453, 0x000088D3 }, /* GL_MATRIX19_ARB */
-   { 15469, 0x000088C1 }, /* GL_MATRIX1_ARB */
-   { 15484, 0x00008631 }, /* GL_MATRIX1_NV */
-   { 15498, 0x000088D4 }, /* GL_MATRIX20_ARB */
-   { 15514, 0x000088D5 }, /* GL_MATRIX21_ARB */
-   { 15530, 0x000088D6 }, /* GL_MATRIX22_ARB */
-   { 15546, 0x000088D7 }, /* GL_MATRIX23_ARB */
-   { 15562, 0x000088D8 }, /* GL_MATRIX24_ARB */
-   { 15578, 0x000088D9 }, /* GL_MATRIX25_ARB */
-   { 15594, 0x000088DA }, /* GL_MATRIX26_ARB */
-   { 15610, 0x000088DB }, /* GL_MATRIX27_ARB */
-   { 15626, 0x000088DC }, /* GL_MATRIX28_ARB */
-   { 15642, 0x000088DD }, /* GL_MATRIX29_ARB */
-   { 15658, 0x000088C2 }, /* GL_MATRIX2_ARB */
-   { 15673, 0x00008632 }, /* GL_MATRIX2_NV */
-   { 15687, 0x000088DE }, /* GL_MATRIX30_ARB */
-   { 15703, 0x000088DF }, /* GL_MATRIX31_ARB */
-   { 15719, 0x000088C3 }, /* GL_MATRIX3_ARB */
-   { 15734, 0x00008633 }, /* GL_MATRIX3_NV */
-   { 15748, 0x000088C4 }, /* GL_MATRIX4_ARB */
-   { 15763, 0x00008634 }, /* GL_MATRIX4_NV */
-   { 15777, 0x000088C5 }, /* GL_MATRIX5_ARB */
-   { 15792, 0x00008635 }, /* GL_MATRIX5_NV */
-   { 15806, 0x000088C6 }, /* GL_MATRIX6_ARB */
-   { 15821, 0x00008636 }, /* GL_MATRIX6_NV */
-   { 15835, 0x000088C7 }, /* GL_MATRIX7_ARB */
-   { 15850, 0x00008637 }, /* GL_MATRIX7_NV */
-   { 15864, 0x000088C8 }, /* GL_MATRIX8_ARB */
-   { 15879, 0x000088C9 }, /* GL_MATRIX9_ARB */
-   { 15894, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
-   { 15920, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-   { 15954, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-   { 15985, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-   { 16018, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-   { 16049, 0x00000BA0 }, /* GL_MATRIX_MODE */
-   { 16064, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
-   { 16086, 0x00008008 }, /* GL_MAX */
-   { 16093, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
-   { 16116, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-   { 16148, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
-   { 16174, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-   { 16207, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-   { 16233, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 16267, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
-   { 16286, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
-   { 16315, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-   { 16347, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
-   { 16383, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-   { 16419, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
-   { 16459, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
-   { 16485, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
-   { 16515, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
-   { 16540, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
-   { 16569, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-   { 16598, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
-   { 16631, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
-   { 16651, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
-   { 16675, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
-   { 16699, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
-   { 16723, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
-   { 16748, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
-   { 16766, 0x00008008 }, /* GL_MAX_EXT */
-   { 16777, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-   { 16812, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
-   { 16851, 0x00000D31 }, /* GL_MAX_LIGHTS */
-   { 16865, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
-   { 16885, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-   { 16923, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-   { 16952, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
-   { 16976, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
-   { 17004, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
-   { 17027, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 17064, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 17100, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-   { 17127, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-   { 17156, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-   { 17190, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
-   { 17226, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-   { 17253, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-   { 17285, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-   { 17321, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-   { 17350, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-   { 17379, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
-   { 17407, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-   { 17445, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 17489, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 17532, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 17566, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 17605, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 17642, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 17680, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 17723, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 17766, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-   { 17796, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-   { 17827, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 17863, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 17899, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
-   { 17929, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-   { 17963, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
-   { 17996, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
-   { 18025, 0x00008504 }, /* GL_MAX_SHININESS_NV */
-   { 18045, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
-   { 18069, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
-   { 18091, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
-   { 18117, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-   { 18144, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
-   { 18175, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
-   { 18199, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 18233, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
-   { 18253, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
-   { 18280, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
-   { 18301, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
-   { 18326, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
-   { 18351, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
-   { 18386, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
-   { 18408, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
-   { 18434, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
-   { 18456, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
-   { 18482, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-   { 18516, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
-   { 18554, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-   { 18587, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
-   { 18624, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
-   { 18648, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
-   { 18669, 0x00008007 }, /* GL_MIN */
-   { 18676, 0x0000802E }, /* GL_MINMAX */
-   { 18686, 0x0000802E }, /* GL_MINMAX_EXT */
-   { 18700, 0x0000802F }, /* GL_MINMAX_FORMAT */
-   { 18717, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
-   { 18738, 0x00008030 }, /* GL_MINMAX_SINK */
-   { 18753, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
-   { 18772, 0x00008007 }, /* GL_MIN_EXT */
-   { 18783, 0x00008370 }, /* GL_MIRRORED_REPEAT */
-   { 18802, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
-   { 18825, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
-   { 18848, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
-   { 18868, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
-   { 18888, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-   { 18918, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
-   { 18946, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-   { 18974, 0x00001700 }, /* GL_MODELVIEW */
-   { 18987, 0x00001700 }, /* GL_MODELVIEW0_ARB */
-   { 19005, 0x0000872A }, /* GL_MODELVIEW10_ARB */
-   { 19024, 0x0000872B }, /* GL_MODELVIEW11_ARB */
-   { 19043, 0x0000872C }, /* GL_MODELVIEW12_ARB */
-   { 19062, 0x0000872D }, /* GL_MODELVIEW13_ARB */
-   { 19081, 0x0000872E }, /* GL_MODELVIEW14_ARB */
-   { 19100, 0x0000872F }, /* GL_MODELVIEW15_ARB */
-   { 19119, 0x00008730 }, /* GL_MODELVIEW16_ARB */
-   { 19138, 0x00008731 }, /* GL_MODELVIEW17_ARB */
-   { 19157, 0x00008732 }, /* GL_MODELVIEW18_ARB */
-   { 19176, 0x00008733 }, /* GL_MODELVIEW19_ARB */
-   { 19195, 0x0000850A }, /* GL_MODELVIEW1_ARB */
-   { 19213, 0x00008734 }, /* GL_MODELVIEW20_ARB */
-   { 19232, 0x00008735 }, /* GL_MODELVIEW21_ARB */
-   { 19251, 0x00008736 }, /* GL_MODELVIEW22_ARB */
-   { 19270, 0x00008737 }, /* GL_MODELVIEW23_ARB */
-   { 19289, 0x00008738 }, /* GL_MODELVIEW24_ARB */
-   { 19308, 0x00008739 }, /* GL_MODELVIEW25_ARB */
-   { 19327, 0x0000873A }, /* GL_MODELVIEW26_ARB */
-   { 19346, 0x0000873B }, /* GL_MODELVIEW27_ARB */
-   { 19365, 0x0000873C }, /* GL_MODELVIEW28_ARB */
-   { 19384, 0x0000873D }, /* GL_MODELVIEW29_ARB */
-   { 19403, 0x00008722 }, /* GL_MODELVIEW2_ARB */
-   { 19421, 0x0000873E }, /* GL_MODELVIEW30_ARB */
-   { 19440, 0x0000873F }, /* GL_MODELVIEW31_ARB */
-   { 19459, 0x00008723 }, /* GL_MODELVIEW3_ARB */
-   { 19477, 0x00008724 }, /* GL_MODELVIEW4_ARB */
-   { 19495, 0x00008725 }, /* GL_MODELVIEW5_ARB */
-   { 19513, 0x00008726 }, /* GL_MODELVIEW6_ARB */
-   { 19531, 0x00008727 }, /* GL_MODELVIEW7_ARB */
-   { 19549, 0x00008728 }, /* GL_MODELVIEW8_ARB */
-   { 19567, 0x00008729 }, /* GL_MODELVIEW9_ARB */
-   { 19585, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
-   { 19605, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
-   { 19632, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
-   { 19657, 0x00002100 }, /* GL_MODULATE */
-   { 19669, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
-   { 19689, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
-   { 19716, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
-   { 19741, 0x00000103 }, /* GL_MULT */
-   { 19749, 0x0000809D }, /* GL_MULTISAMPLE */
-   { 19764, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
-   { 19784, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
-   { 19803, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
-   { 19822, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
-   { 19846, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
-   { 19869, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-   { 19899, 0x00002A25 }, /* GL_N3F_V3F */
-   { 19910, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
-   { 19930, 0x0000150E }, /* GL_NAND */
-   { 19938, 0x00002600 }, /* GL_NEAREST */
-   { 19949, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-   { 19980, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-   { 20012, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
-   { 20037, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
-   { 20063, 0x00000200 }, /* GL_NEVER */
-   { 20072, 0x00001102 }, /* GL_NICEST */
-   { 20082, 0x00000000 }, /* GL_NONE */
-   { 20090, 0x00001505 }, /* GL_NOOP */
-   { 20098, 0x00001508 }, /* GL_NOR */
-   { 20105, 0x00000BA1 }, /* GL_NORMALIZE */
-   { 20118, 0x00008075 }, /* GL_NORMAL_ARRAY */
-   { 20134, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
-   { 20165, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
-   { 20200, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
-   { 20224, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
-   { 20247, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
-   { 20268, 0x00008511 }, /* GL_NORMAL_MAP */
-   { 20282, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
-   { 20300, 0x00008511 }, /* GL_NORMAL_MAP_NV */
-   { 20317, 0x00000205 }, /* GL_NOTEQUAL */
-   { 20329, 0x00000000 }, /* GL_NO_ERROR */
-   { 20341, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
-   { 20375, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
-   { 20413, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
-   { 20445, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
-   { 20487, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
-   { 20517, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
-   { 20557, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
-   { 20588, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
-   { 20617, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
-   { 20645, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
-   { 20675, 0x00002401 }, /* GL_OBJECT_LINEAR */
-   { 20692, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
-   { 20718, 0x00002501 }, /* GL_OBJECT_PLANE */
-   { 20734, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
-   { 20769, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
-   { 20791, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
-   { 20810, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
-   { 20840, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
-   { 20861, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
-   { 20889, 0x00000001 }, /* GL_ONE */
-   { 20896, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
-   { 20924, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
-   { 20956, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
-   { 20984, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
-   { 21016, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
-   { 21039, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
-   { 21062, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
-   { 21085, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
-   { 21108, 0x00008598 }, /* GL_OPERAND0_ALPHA */
-   { 21126, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
-   { 21148, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
-   { 21170, 0x00008590 }, /* GL_OPERAND0_RGB */
-   { 21186, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
-   { 21206, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
-   { 21226, 0x00008599 }, /* GL_OPERAND1_ALPHA */
-   { 21244, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
-   { 21266, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
-   { 21288, 0x00008591 }, /* GL_OPERAND1_RGB */
-   { 21304, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
-   { 21324, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
-   { 21344, 0x0000859A }, /* GL_OPERAND2_ALPHA */
-   { 21362, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
-   { 21384, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
-   { 21406, 0x00008592 }, /* GL_OPERAND2_RGB */
-   { 21422, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
-   { 21442, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
-   { 21462, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
-   { 21483, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
-   { 21502, 0x00001507 }, /* GL_OR */
-   { 21508, 0x00000A01 }, /* GL_ORDER */
-   { 21517, 0x0000150D }, /* GL_OR_INVERTED */
-   { 21532, 0x0000150B }, /* GL_OR_REVERSE */
-   { 21546, 0x00000505 }, /* GL_OUT_OF_MEMORY */
-   { 21563, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
-   { 21581, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
-   { 21602, 0x00008758 }, /* GL_PACK_INVERT_MESA */
-   { 21622, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
-   { 21640, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
-   { 21659, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
-   { 21679, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
-   { 21699, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
-   { 21717, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
-   { 21736, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
-   { 21761, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
-   { 21785, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
-   { 21806, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
-   { 21828, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
-   { 21850, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
-   { 21875, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
-   { 21899, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
-   { 21920, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
-   { 21942, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
-   { 21964, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
-   { 21986, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
-   { 22017, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
-   { 22037, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-   { 22062, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
-   { 22082, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-   { 22107, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
-   { 22127, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-   { 22152, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
-   { 22172, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-   { 22197, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
-   { 22217, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-   { 22242, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
-   { 22262, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-   { 22287, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
-   { 22307, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-   { 22332, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
-   { 22352, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-   { 22377, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
-   { 22397, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-   { 22422, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
-   { 22442, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-   { 22467, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
-   { 22485, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
-   { 22518, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
-   { 22543, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
-   { 22578, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
-   { 22605, 0x00001B00 }, /* GL_POINT */
-   { 22614, 0x00000000 }, /* GL_POINTS */
-   { 22624, 0x00000002 }, /* GL_POINT_BIT */
-   { 22637, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
-   { 22667, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
-   { 22701, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
-   { 22735, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
-   { 22770, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
-   { 22799, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
-   { 22832, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
-   { 22865, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
-   { 22899, 0x00000B11 }, /* GL_POINT_SIZE */
-   { 22913, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
-   { 22939, 0x00008127 }, /* GL_POINT_SIZE_MAX */
-   { 22957, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
-   { 22979, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
-   { 23001, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
-   { 23024, 0x00008126 }, /* GL_POINT_SIZE_MIN */
-   { 23042, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
-   { 23064, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
-   { 23086, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
-   { 23109, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
-   { 23129, 0x00000B10 }, /* GL_POINT_SMOOTH */
-   { 23145, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
-   { 23166, 0x00008861 }, /* GL_POINT_SPRITE */
-   { 23182, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
-   { 23202, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
-   { 23231, 0x00008861 }, /* GL_POINT_SPRITE_NV */
-   { 23250, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
-   { 23276, 0x00000701 }, /* GL_POINT_TOKEN */
-   { 23291, 0x00000009 }, /* GL_POLYGON */
-   { 23302, 0x00000008 }, /* GL_POLYGON_BIT */
-   { 23317, 0x00000B40 }, /* GL_POLYGON_MODE */
-   { 23333, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
-   { 23356, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
-   { 23381, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
-   { 23404, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
-   { 23427, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
-   { 23451, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
-   { 23475, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
-   { 23493, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
-   { 23516, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
-   { 23535, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
-   { 23558, 0x00000703 }, /* GL_POLYGON_TOKEN */
-   { 23575, 0x00001203 }, /* GL_POSITION */
-   { 23587, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-   { 23619, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
-   { 23655, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-   { 23688, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
-   { 23725, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-   { 23756, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
-   { 23791, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-   { 23823, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
-   { 23859, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 23892, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-   { 23924, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
-   { 23960, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-   { 23993, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
-   { 24030, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-   { 24060, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
-   { 24094, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-   { 24125, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
-   { 24160, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-   { 24191, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
-   { 24226, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-   { 24258, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
-   { 24294, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-   { 24324, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
-   { 24358, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-   { 24389, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
-   { 24424, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-   { 24456, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-   { 24487, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
-   { 24522, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-   { 24554, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
-   { 24590, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
-   { 24619, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
-   { 24652, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
-   { 24682, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
-   { 24716, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-   { 24755, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-   { 24788, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-   { 24828, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-   { 24862, 0x00008578 }, /* GL_PREVIOUS */
-   { 24874, 0x00008578 }, /* GL_PREVIOUS_ARB */
-   { 24890, 0x00008578 }, /* GL_PREVIOUS_EXT */
-   { 24906, 0x00008577 }, /* GL_PRIMARY_COLOR */
-   { 24923, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
-   { 24944, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
-   { 24965, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-   { 24998, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-   { 25030, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
-   { 25053, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
-   { 25076, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
-   { 25106, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
-   { 25135, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
-   { 25163, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
-   { 25185, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-   { 25213, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-   { 25241, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
-   { 25263, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
-   { 25284, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-   { 25324, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-   { 25363, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-   { 25393, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-   { 25428, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-   { 25461, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-   { 25495, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-   { 25534, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-   { 25573, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
-   { 25595, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
-   { 25621, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
-   { 25645, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
-   { 25668, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
-   { 25690, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
-   { 25711, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
-   { 25732, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
-   { 25759, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-   { 25791, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-   { 25823, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-   { 25858, 0x00001701 }, /* GL_PROJECTION */
-   { 25872, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
-   { 25893, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
-   { 25919, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
-   { 25940, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
-   { 25959, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
-   { 25982, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
-   { 26021, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-   { 26059, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
-   { 26079, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-   { 26109, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
-   { 26133, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
-   { 26153, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-   { 26183, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
-   { 26207, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
-   { 26227, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-   { 26260, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
-   { 26286, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
-   { 26316, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-   { 26347, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
-   { 26377, 0x00002003 }, /* GL_Q */
-   { 26382, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
-   { 26407, 0x00000007 }, /* GL_QUADS */
-   { 26416, 0x00008614 }, /* GL_QUAD_MESH_SUN */
-   { 26433, 0x00000008 }, /* GL_QUAD_STRIP */
-   { 26447, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
-   { 26469, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
-   { 26495, 0x00008866 }, /* GL_QUERY_RESULT */
-   { 26511, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
-   { 26531, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
-   { 26557, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
-   { 26587, 0x00002002 }, /* GL_R */
-   { 26592, 0x00002A10 }, /* GL_R3_G3_B2 */
-   { 26604, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-   { 26637, 0x00000C02 }, /* GL_READ_BUFFER */
-   { 26652, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-   { 26684, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
-   { 26708, 0x000088B8 }, /* GL_READ_ONLY */
-   { 26721, 0x000088B8 }, /* GL_READ_ONLY_ARB */
-   { 26738, 0x000088BA }, /* GL_READ_WRITE */
-   { 26752, 0x000088BA }, /* GL_READ_WRITE_ARB */
-   { 26770, 0x00001903 }, /* GL_RED */
-   { 26777, 0x00008016 }, /* GL_REDUCE */
-   { 26787, 0x00008016 }, /* GL_REDUCE_EXT */
-   { 26801, 0x00000D15 }, /* GL_RED_BIAS */
-   { 26813, 0x00000D52 }, /* GL_RED_BITS */
-   { 26825, 0x00000D14 }, /* GL_RED_SCALE */
-   { 26838, 0x00008512 }, /* GL_REFLECTION_MAP */
-   { 26856, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
-   { 26878, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
-   { 26899, 0x00001C00 }, /* GL_RENDER */
-   { 26909, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
-   { 26937, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
-   { 26957, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
-   { 26984, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
-   { 27020, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
-   { 27046, 0x00001F01 }, /* GL_RENDERER */
-   { 27058, 0x00000C40 }, /* GL_RENDER_MODE */
-   { 27073, 0x00002901 }, /* GL_REPEAT */
-   { 27083, 0x00001E01 }, /* GL_REPLACE */
-   { 27094, 0x00008062 }, /* GL_REPLACE_EXT */
-   { 27109, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
-   { 27132, 0x0000803A }, /* GL_RESCALE_NORMAL */
-   { 27150, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
-   { 27172, 0x00000102 }, /* GL_RETURN */
-   { 27182, 0x00001907 }, /* GL_RGB */
-   { 27189, 0x00008052 }, /* GL_RGB10 */
-   { 27198, 0x00008059 }, /* GL_RGB10_A2 */
-   { 27210, 0x00008059 }, /* GL_RGB10_A2_EXT */
-   { 27226, 0x00008052 }, /* GL_RGB10_EXT */
-   { 27239, 0x00008053 }, /* GL_RGB12 */
-   { 27248, 0x00008053 }, /* GL_RGB12_EXT */
-   { 27261, 0x00008054 }, /* GL_RGB16 */
-   { 27270, 0x00008054 }, /* GL_RGB16_EXT */
-   { 27283, 0x0000804E }, /* GL_RGB2_EXT */
-   { 27295, 0x0000804F }, /* GL_RGB4 */
-   { 27303, 0x0000804F }, /* GL_RGB4_EXT */
-   { 27315, 0x000083A1 }, /* GL_RGB4_S3TC */
-   { 27328, 0x00008050 }, /* GL_RGB5 */
-   { 27336, 0x00008057 }, /* GL_RGB5_A1 */
-   { 27347, 0x00008057 }, /* GL_RGB5_A1_EXT */
-   { 27362, 0x00008050 }, /* GL_RGB5_EXT */
-   { 27374, 0x00008051 }, /* GL_RGB8 */
-   { 27382, 0x00008051 }, /* GL_RGB8_EXT */
-   { 27394, 0x00001908 }, /* GL_RGBA */
-   { 27402, 0x0000805A }, /* GL_RGBA12 */
-   { 27412, 0x0000805A }, /* GL_RGBA12_EXT */
-   { 27426, 0x0000805B }, /* GL_RGBA16 */
-   { 27436, 0x0000805B }, /* GL_RGBA16_EXT */
-   { 27450, 0x00008055 }, /* GL_RGBA2 */
-   { 27459, 0x00008055 }, /* GL_RGBA2_EXT */
-   { 27472, 0x00008056 }, /* GL_RGBA4 */
-   { 27481, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
-   { 27500, 0x00008056 }, /* GL_RGBA4_EXT */
-   { 27513, 0x000083A3 }, /* GL_RGBA4_S3TC */
-   { 27527, 0x00008058 }, /* GL_RGBA8 */
-   { 27536, 0x00008058 }, /* GL_RGBA8_EXT */
-   { 27549, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
-   { 27567, 0x00000C31 }, /* GL_RGBA_MODE */
-   { 27580, 0x000083A2 }, /* GL_RGBA_S3TC */
-   { 27593, 0x000083A0 }, /* GL_RGB_S3TC */
-   { 27605, 0x00008573 }, /* GL_RGB_SCALE */
-   { 27618, 0x00008573 }, /* GL_RGB_SCALE_ARB */
-   { 27635, 0x00008573 }, /* GL_RGB_SCALE_EXT */
-   { 27652, 0x00000407 }, /* GL_RIGHT */
-   { 27661, 0x00002000 }, /* GL_S */
-   { 27666, 0x00008B5D }, /* GL_SAMPLER_1D */
-   { 27680, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
-   { 27701, 0x00008B5E }, /* GL_SAMPLER_2D */
-   { 27715, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
-   { 27736, 0x00008B5F }, /* GL_SAMPLER_3D */
-   { 27750, 0x00008B60 }, /* GL_SAMPLER_CUBE */
-   { 27766, 0x000080A9 }, /* GL_SAMPLES */
-   { 27777, 0x000086B4 }, /* GL_SAMPLES_3DFX */
-   { 27793, 0x000080A9 }, /* GL_SAMPLES_ARB */
-   { 27808, 0x00008914 }, /* GL_SAMPLES_PASSED */
-   { 27826, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
-   { 27848, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-   { 27876, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
-   { 27908, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
-   { 27931, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
-   { 27958, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
-   { 27976, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
-   { 27999, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
-   { 28021, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
-   { 28040, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
-   { 28063, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
-   { 28089, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
-   { 28119, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
-   { 28144, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
-   { 28173, 0x00080000 }, /* GL_SCISSOR_BIT */
-   { 28188, 0x00000C10 }, /* GL_SCISSOR_BOX */
-   { 28203, 0x00000C11 }, /* GL_SCISSOR_TEST */
-   { 28219, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
-   { 28244, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-   { 28284, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
-   { 28328, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-   { 28361, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-   { 28391, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-   { 28423, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-   { 28453, 0x00001C02 }, /* GL_SELECT */
-   { 28463, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
-   { 28491, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
-   { 28516, 0x00008012 }, /* GL_SEPARABLE_2D */
-   { 28532, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
-   { 28559, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
-   { 28590, 0x0000150F }, /* GL_SET */
-   { 28597, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
-   { 28618, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
-   { 28642, 0x00008B4F }, /* GL_SHADER_TYPE */
-   { 28657, 0x00000B54 }, /* GL_SHADE_MODEL */
-   { 28672, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
-   { 28700, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
-   { 28723, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-   { 28753, 0x00001601 }, /* GL_SHININESS */
-   { 28766, 0x00001402 }, /* GL_SHORT */
-   { 28775, 0x000081F9 }, /* GL_SINGLE_COLOR */
-   { 28791, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
-   { 28811, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
-   { 28830, 0x00001D01 }, /* GL_SMOOTH */
-   { 28840, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
-   { 28873, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
-   { 28900, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
-   { 28933, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
-   { 28960, 0x00008588 }, /* GL_SOURCE0_ALPHA */
-   { 28977, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
-   { 28998, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
-   { 29019, 0x00008580 }, /* GL_SOURCE0_RGB */
-   { 29034, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
-   { 29053, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
-   { 29072, 0x00008589 }, /* GL_SOURCE1_ALPHA */
-   { 29089, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
-   { 29110, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
-   { 29131, 0x00008581 }, /* GL_SOURCE1_RGB */
-   { 29146, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
-   { 29165, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
-   { 29184, 0x0000858A }, /* GL_SOURCE2_ALPHA */
-   { 29201, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
-   { 29222, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
-   { 29243, 0x00008582 }, /* GL_SOURCE2_RGB */
-   { 29258, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
-   { 29277, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
-   { 29296, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
-   { 29316, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
-   { 29334, 0x00001202 }, /* GL_SPECULAR */
-   { 29346, 0x00002402 }, /* GL_SPHERE_MAP */
-   { 29360, 0x00001206 }, /* GL_SPOT_CUTOFF */
-   { 29375, 0x00001204 }, /* GL_SPOT_DIRECTION */
-   { 29393, 0x00001205 }, /* GL_SPOT_EXPONENT */
-   { 29410, 0x00008588 }, /* GL_SRC0_ALPHA */
-   { 29424, 0x00008580 }, /* GL_SRC0_RGB */
-   { 29436, 0x00008589 }, /* GL_SRC1_ALPHA */
-   { 29450, 0x00008581 }, /* GL_SRC1_RGB */
-   { 29462, 0x0000858A }, /* GL_SRC2_ALPHA */
-   { 29476, 0x00008582 }, /* GL_SRC2_RGB */
-   { 29488, 0x00000302 }, /* GL_SRC_ALPHA */
-   { 29501, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
-   { 29523, 0x00000300 }, /* GL_SRC_COLOR */
-   { 29536, 0x00000503 }, /* GL_STACK_OVERFLOW */
-   { 29554, 0x00000504 }, /* GL_STACK_UNDERFLOW */
-   { 29573, 0x000088E6 }, /* GL_STATIC_COPY */
-   { 29588, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
-   { 29607, 0x000088E4 }, /* GL_STATIC_DRAW */
-   { 29622, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
-   { 29641, 0x000088E5 }, /* GL_STATIC_READ */
-   { 29656, 0x000088E5 }, /* GL_STATIC_READ_ARB */
-   { 29675, 0x00001802 }, /* GL_STENCIL */
-   { 29686, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
-   { 29712, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
-   { 29733, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
-   { 29758, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
-   { 29779, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
-   { 29804, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-   { 29836, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
-   { 29872, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-   { 29904, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
-   { 29940, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
-   { 29960, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
-   { 29987, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
-   { 30013, 0x00000D57 }, /* GL_STENCIL_BITS */
-   { 30029, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
-   { 30051, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
-   { 30074, 0x00000B94 }, /* GL_STENCIL_FAIL */
-   { 30090, 0x00000B92 }, /* GL_STENCIL_FUNC */
-   { 30106, 0x00001901 }, /* GL_STENCIL_INDEX */
-   { 30123, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
-   { 30146, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
-   { 30168, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
-   { 30190, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
-   { 30212, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
-   { 30233, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
-   { 30260, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
-   { 30287, 0x00000B97 }, /* GL_STENCIL_REF */
-   { 30302, 0x00000B90 }, /* GL_STENCIL_TEST */
-   { 30318, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
-   { 30347, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
-   { 30369, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
-   { 30390, 0x00000C33 }, /* GL_STEREO */
-   { 30400, 0x000088E2 }, /* GL_STREAM_COPY */
-   { 30415, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
-   { 30434, 0x000088E0 }, /* GL_STREAM_DRAW */
-   { 30449, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
-   { 30468, 0x000088E1 }, /* GL_STREAM_READ */
-   { 30483, 0x000088E1 }, /* GL_STREAM_READ_ARB */
-   { 30502, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
-   { 30519, 0x000084E7 }, /* GL_SUBTRACT */
-   { 30531, 0x000084E7 }, /* GL_SUBTRACT_ARB */
-   { 30547, 0x00002001 }, /* GL_T */
-   { 30552, 0x00002A2A }, /* GL_T2F_C3F_V3F */
-   { 30567, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
-   { 30586, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
-   { 30602, 0x00002A2B }, /* GL_T2F_N3F_V3F */
-   { 30617, 0x00002A27 }, /* GL_T2F_V3F */
-   { 30628, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
-   { 30647, 0x00002A28 }, /* GL_T4F_V4F */
-   { 30658, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
-   { 30681, 0x00001702 }, /* GL_TEXTURE */
-   { 30692, 0x000084C0 }, /* GL_TEXTURE0 */
-   { 30704, 0x000084C0 }, /* GL_TEXTURE0_ARB */
-   { 30720, 0x000084C1 }, /* GL_TEXTURE1 */
-   { 30732, 0x000084CA }, /* GL_TEXTURE10 */
-   { 30745, 0x000084CA }, /* GL_TEXTURE10_ARB */
-   { 30762, 0x000084CB }, /* GL_TEXTURE11 */
-   { 30775, 0x000084CB }, /* GL_TEXTURE11_ARB */
-   { 30792, 0x000084CC }, /* GL_TEXTURE12 */
-   { 30805, 0x000084CC }, /* GL_TEXTURE12_ARB */
-   { 30822, 0x000084CD }, /* GL_TEXTURE13 */
-   { 30835, 0x000084CD }, /* GL_TEXTURE13_ARB */
-   { 30852, 0x000084CE }, /* GL_TEXTURE14 */
-   { 30865, 0x000084CE }, /* GL_TEXTURE14_ARB */
-   { 30882, 0x000084CF }, /* GL_TEXTURE15 */
-   { 30895, 0x000084CF }, /* GL_TEXTURE15_ARB */
-   { 30912, 0x000084D0 }, /* GL_TEXTURE16 */
-   { 30925, 0x000084D0 }, /* GL_TEXTURE16_ARB */
-   { 30942, 0x000084D1 }, /* GL_TEXTURE17 */
-   { 30955, 0x000084D1 }, /* GL_TEXTURE17_ARB */
-   { 30972, 0x000084D2 }, /* GL_TEXTURE18 */
-   { 30985, 0x000084D2 }, /* GL_TEXTURE18_ARB */
-   { 31002, 0x000084D3 }, /* GL_TEXTURE19 */
-   { 31015, 0x000084D3 }, /* GL_TEXTURE19_ARB */
-   { 31032, 0x000084C1 }, /* GL_TEXTURE1_ARB */
-   { 31048, 0x000084C2 }, /* GL_TEXTURE2 */
-   { 31060, 0x000084D4 }, /* GL_TEXTURE20 */
-   { 31073, 0x000084D4 }, /* GL_TEXTURE20_ARB */
-   { 31090, 0x000084D5 }, /* GL_TEXTURE21 */
-   { 31103, 0x000084D5 }, /* GL_TEXTURE21_ARB */
-   { 31120, 0x000084D6 }, /* GL_TEXTURE22 */
-   { 31133, 0x000084D6 }, /* GL_TEXTURE22_ARB */
-   { 31150, 0x000084D7 }, /* GL_TEXTURE23 */
-   { 31163, 0x000084D7 }, /* GL_TEXTURE23_ARB */
-   { 31180, 0x000084D8 }, /* GL_TEXTURE24 */
-   { 31193, 0x000084D8 }, /* GL_TEXTURE24_ARB */
-   { 31210, 0x000084D9 }, /* GL_TEXTURE25 */
-   { 31223, 0x000084D9 }, /* GL_TEXTURE25_ARB */
-   { 31240, 0x000084DA }, /* GL_TEXTURE26 */
-   { 31253, 0x000084DA }, /* GL_TEXTURE26_ARB */
-   { 31270, 0x000084DB }, /* GL_TEXTURE27 */
-   { 31283, 0x000084DB }, /* GL_TEXTURE27_ARB */
-   { 31300, 0x000084DC }, /* GL_TEXTURE28 */
-   { 31313, 0x000084DC }, /* GL_TEXTURE28_ARB */
-   { 31330, 0x000084DD }, /* GL_TEXTURE29 */
-   { 31343, 0x000084DD }, /* GL_TEXTURE29_ARB */
-   { 31360, 0x000084C2 }, /* GL_TEXTURE2_ARB */
-   { 31376, 0x000084C3 }, /* GL_TEXTURE3 */
-   { 31388, 0x000084DE }, /* GL_TEXTURE30 */
-   { 31401, 0x000084DE }, /* GL_TEXTURE30_ARB */
-   { 31418, 0x000084DF }, /* GL_TEXTURE31 */
-   { 31431, 0x000084DF }, /* GL_TEXTURE31_ARB */
-   { 31448, 0x000084C3 }, /* GL_TEXTURE3_ARB */
-   { 31464, 0x000084C4 }, /* GL_TEXTURE4 */
-   { 31476, 0x000084C4 }, /* GL_TEXTURE4_ARB */
-   { 31492, 0x000084C5 }, /* GL_TEXTURE5 */
-   { 31504, 0x000084C5 }, /* GL_TEXTURE5_ARB */
-   { 31520, 0x000084C6 }, /* GL_TEXTURE6 */
-   { 31532, 0x000084C6 }, /* GL_TEXTURE6_ARB */
-   { 31548, 0x000084C7 }, /* GL_TEXTURE7 */
-   { 31560, 0x000084C7 }, /* GL_TEXTURE7_ARB */
-   { 31576, 0x000084C8 }, /* GL_TEXTURE8 */
-   { 31588, 0x000084C8 }, /* GL_TEXTURE8_ARB */
-   { 31604, 0x000084C9 }, /* GL_TEXTURE9 */
-   { 31616, 0x000084C9 }, /* GL_TEXTURE9_ARB */
-   { 31632, 0x00000DE0 }, /* GL_TEXTURE_1D */
-   { 31646, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
-   { 31670, 0x00000DE1 }, /* GL_TEXTURE_2D */
-   { 31684, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
-   { 31708, 0x0000806F }, /* GL_TEXTURE_3D */
-   { 31722, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
-   { 31744, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
-   { 31770, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
-   { 31792, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
-   { 31814, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-   { 31846, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
-   { 31868, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-   { 31900, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
-   { 31922, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
-   { 31950, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
-   { 31982, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-   { 32015, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
-   { 32047, 0x00040000 }, /* GL_TEXTURE_BIT */
-   { 32062, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
-   { 32083, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
-   { 32108, 0x00001005 }, /* GL_TEXTURE_BORDER */
-   { 32126, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
-   { 32150, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-   { 32181, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-   { 32211, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-   { 32241, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-   { 32276, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-   { 32307, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-   { 32345, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
-   { 32372, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-   { 32404, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
-   { 32438, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
-   { 32462, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
-   { 32490, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
-   { 32514, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
-   { 32542, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-   { 32575, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
-   { 32599, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
-   { 32621, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
-   { 32643, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
-   { 32669, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
-   { 32703, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-   { 32736, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
-   { 32773, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
-   { 32801, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
-   { 32833, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
-   { 32856, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-   { 32894, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
-   { 32936, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-   { 32967, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-   { 32995, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-   { 33025, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-   { 33053, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
-   { 33073, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
-   { 33097, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-   { 33128, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
-   { 33163, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-   { 33194, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
-   { 33229, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-   { 33260, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
-   { 33295, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-   { 33326, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
-   { 33361, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-   { 33392, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
-   { 33427, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-   { 33458, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
-   { 33493, 0x00008071 }, /* GL_TEXTURE_DEPTH */
-   { 33510, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
-   { 33532, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
-   { 33558, 0x00002300 }, /* GL_TEXTURE_ENV */
-   { 33573, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
-   { 33594, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
-   { 33614, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
-   { 33640, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
-   { 33660, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
-   { 33677, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
-   { 33694, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
-   { 33711, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
-   { 33728, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
-   { 33753, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
-   { 33775, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
-   { 33801, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
-   { 33819, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
-   { 33845, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
-   { 33871, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
-   { 33901, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
-   { 33928, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
-   { 33953, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
-   { 33973, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
-   { 33997, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-   { 34024, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-   { 34051, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-   { 34078, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
-   { 34104, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
-   { 34134, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
-   { 34156, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
-   { 34174, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-   { 34204, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-   { 34232, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-   { 34260, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-   { 34288, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
-   { 34309, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
-   { 34328, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
-   { 34350, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
-   { 34369, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
-   { 34389, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
-   { 34414, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
-   { 34438, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
-   { 34458, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
-   { 34482, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
-   { 34502, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
-   { 34525, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
-   { 34550, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
-   { 34584, 0x00001000 }, /* GL_TEXTURE_WIDTH */
-   { 34601, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
-   { 34619, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
-   { 34637, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
-   { 34655, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
-   { 34675, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
-   { 34694, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-   { 34723, 0x00001000 }, /* GL_TRANSFORM_BIT */
-   { 34740, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
-   { 34766, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
-   { 34796, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-   { 34828, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-   { 34858, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
-   { 34892, 0x0000862C }, /* GL_TRANSPOSE_NV */
-   { 34908, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-   { 34939, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
-   { 34974, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-   { 35002, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
-   { 35034, 0x00000004 }, /* GL_TRIANGLES */
-   { 35047, 0x00000006 }, /* GL_TRIANGLE_FAN */
-   { 35063, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
-   { 35084, 0x00000005 }, /* GL_TRIANGLE_STRIP */
-   { 35102, 0x00000001 }, /* GL_TRUE */
-   { 35110, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
-   { 35130, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
-   { 35153, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
-   { 35173, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
-   { 35194, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
-   { 35216, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
-   { 35238, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
-   { 35258, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
-   { 35279, 0x00001401 }, /* GL_UNSIGNED_BYTE */
-   { 35296, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-   { 35323, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
-   { 35346, 0x00001405 }, /* GL_UNSIGNED_INT */
-   { 35362, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
-   { 35389, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
-   { 35413, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-   { 35444, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
-   { 35468, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-   { 35496, 0x00001403 }, /* GL_UNSIGNED_SHORT */
-   { 35514, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-   { 35544, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-   { 35570, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-   { 35600, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-   { 35626, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
-   { 35650, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-   { 35678, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-   { 35706, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
-   { 35733, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-   { 35765, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
-   { 35796, 0x00008CA2 }, /* GL_UPPER_LEFT */
-   { 35810, 0x00002A20 }, /* GL_V2F */
-   { 35817, 0x00002A21 }, /* GL_V3F */
-   { 35824, 0x00008B83 }, /* GL_VALIDATE_STATUS */
-   { 35843, 0x00001F00 }, /* GL_VENDOR */
-   { 35853, 0x00001F02 }, /* GL_VERSION */
-   { 35864, 0x00008074 }, /* GL_VERTEX_ARRAY */
-   { 35880, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
-   { 35910, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-   { 35941, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
-   { 35976, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
-   { 36000, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
-   { 36021, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
-   { 36044, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
-   { 36065, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-   { 36092, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-   { 36120, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-   { 36148, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-   { 36176, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-   { 36204, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-   { 36232, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-   { 36260, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-   { 36287, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-   { 36314, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-   { 36341, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-   { 36368, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-   { 36395, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-   { 36422, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-   { 36449, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-   { 36476, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-   { 36503, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-   { 36541, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
-   { 36583, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-   { 36614, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
-   { 36649, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-   { 36683, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
-   { 36721, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-   { 36752, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
-   { 36787, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-   { 36815, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
-   { 36847, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-   { 36877, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
-   { 36911, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
-   { 36939, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
-   { 36971, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
-   { 36991, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
-   { 37013, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
-   { 37042, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
-   { 37063, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-   { 37092, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
-   { 37125, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
-   { 37157, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-   { 37184, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
-   { 37215, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
-   { 37245, 0x00008B31 }, /* GL_VERTEX_SHADER */
-   { 37262, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
-   { 37283, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
-   { 37310, 0x00000BA2 }, /* GL_VIEWPORT */
-   { 37322, 0x00000800 }, /* GL_VIEWPORT_BIT */
-   { 37338, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
-   { 37358, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-   { 37389, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
-   { 37424, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-   { 37452, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-   { 37477, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-   { 37504, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-   { 37529, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
-   { 37553, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
-   { 37572, 0x000088B9 }, /* GL_WRITE_ONLY */
-   { 37586, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
-   { 37604, 0x00001506 }, /* GL_XOR */
-   { 37611, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
-   { 37630, 0x00008757 }, /* GL_YCBCR_MESA */
-   { 37644, 0x00000000 }, /* GL_ZERO */
-   { 37652, 0x00000D16 }, /* GL_ZOOM_X */
-   { 37662, 0x00000D17 }, /* GL_ZOOM_Y */
+   {   877, 0x00008894 }, /* GL_ARRAY_BUFFER_BINDING */
+   {   901, 0x00008894 }, /* GL_ARRAY_BUFFER_BINDING_ARB */
+   {   929, 0x00008B85 }, /* GL_ATTACHED_SHADERS */
+   {   949, 0x00008645 }, /* GL_ATTRIB_ARRAY_POINTER_NV */
+   {   976, 0x00008623 }, /* GL_ATTRIB_ARRAY_SIZE_NV */
+   {  1000, 0x00008624 }, /* GL_ATTRIB_ARRAY_STRIDE_NV */
+   {  1026, 0x00008625 }, /* GL_ATTRIB_ARRAY_TYPE_NV */
+   {  1050, 0x00000BB0 }, /* GL_ATTRIB_STACK_DEPTH */
+   {  1072, 0x00000D80 }, /* GL_AUTO_NORMAL */
+   {  1087, 0x00000409 }, /* GL_AUX0 */
+   {  1095, 0x0000040A }, /* GL_AUX1 */
+   {  1103, 0x0000040B }, /* GL_AUX2 */
+   {  1111, 0x0000040C }, /* GL_AUX3 */
+   {  1119, 0x00000C00 }, /* GL_AUX_BUFFERS */
+   {  1134, 0x00000405 }, /* GL_BACK */
+   {  1142, 0x00000402 }, /* GL_BACK_LEFT */
+   {  1155, 0x00000403 }, /* GL_BACK_RIGHT */
+   {  1169, 0x000080E0 }, /* GL_BGR */
+   {  1176, 0x000080E1 }, /* GL_BGRA */
+   {  1184, 0x00001A00 }, /* GL_BITMAP */
+   {  1194, 0x00000704 }, /* GL_BITMAP_TOKEN */
+   {  1210, 0x00000BE2 }, /* GL_BLEND */
+   {  1219, 0x00008005 }, /* GL_BLEND_COLOR */
+   {  1234, 0x00008005 }, /* GL_BLEND_COLOR_EXT */
+   {  1253, 0x00000BE0 }, /* GL_BLEND_DST */
+   {  1266, 0x000080CA }, /* GL_BLEND_DST_ALPHA */
+   {  1285, 0x000080C8 }, /* GL_BLEND_DST_RGB */
+   {  1302, 0x00008009 }, /* GL_BLEND_EQUATION */
+   {  1320, 0x0000883D }, /* GL_BLEND_EQUATION_ALPHA */
+   {  1344, 0x0000883D }, /* GL_BLEND_EQUATION_ALPHA_EXT */
+   {  1372, 0x00008009 }, /* GL_BLEND_EQUATION_EXT */
+   {  1394, 0x00008009 }, /* GL_BLEND_EQUATION_RGB_EXT */
+   {  1420, 0x00000BE1 }, /* GL_BLEND_SRC */
+   {  1433, 0x000080CB }, /* GL_BLEND_SRC_ALPHA */
+   {  1452, 0x000080C9 }, /* GL_BLEND_SRC_RGB */
+   {  1469, 0x00001905 }, /* GL_BLUE */
+   {  1477, 0x00000D1B }, /* GL_BLUE_BIAS */
+   {  1490, 0x00000D54 }, /* GL_BLUE_BITS */
+   {  1503, 0x00000D1A }, /* GL_BLUE_SCALE */
+   {  1517, 0x00008B56 }, /* GL_BOOL */
+   {  1525, 0x00008B56 }, /* GL_BOOL_ARB */
+   {  1537, 0x00008B57 }, /* GL_BOOL_VEC2 */
+   {  1550, 0x00008B57 }, /* GL_BOOL_VEC2_ARB */
+   {  1567, 0x00008B58 }, /* GL_BOOL_VEC3 */
+   {  1580, 0x00008B58 }, /* GL_BOOL_VEC3_ARB */
+   {  1597, 0x00008B59 }, /* GL_BOOL_VEC4 */
+   {  1610, 0x00008B59 }, /* GL_BOOL_VEC4_ARB */
+   {  1627, 0x000088BB }, /* GL_BUFFER_ACCESS */
+   {  1644, 0x000088BB }, /* GL_BUFFER_ACCESS_ARB */
+   {  1665, 0x000088BC }, /* GL_BUFFER_MAPPED */
+   {  1682, 0x000088BC }, /* GL_BUFFER_MAPPED_ARB */
+   {  1703, 0x000088BD }, /* GL_BUFFER_MAP_POINTER */
+   {  1725, 0x000088BD }, /* GL_BUFFER_MAP_POINTER_ARB */
+   {  1751, 0x00008764 }, /* GL_BUFFER_SIZE */
+   {  1766, 0x00008764 }, /* GL_BUFFER_SIZE_ARB */
+   {  1785, 0x00008765 }, /* GL_BUFFER_USAGE */
+   {  1801, 0x00008765 }, /* GL_BUFFER_USAGE_ARB */
+   {  1821, 0x00001400 }, /* GL_BYTE */
+   {  1829, 0x00002A24 }, /* GL_C3F_V3F */
+   {  1840, 0x00002A26 }, /* GL_C4F_N3F_V3F */
+   {  1855, 0x00002A22 }, /* GL_C4UB_V2F */
+   {  1867, 0x00002A23 }, /* GL_C4UB_V3F */
+   {  1879, 0x00000901 }, /* GL_CCW */
+   {  1886, 0x00002900 }, /* GL_CLAMP */
+   {  1895, 0x0000812D }, /* GL_CLAMP_TO_BORDER */
+   {  1914, 0x0000812D }, /* GL_CLAMP_TO_BORDER_ARB */
+   {  1937, 0x0000812D }, /* GL_CLAMP_TO_BORDER_SGIS */
+   {  1961, 0x0000812F }, /* GL_CLAMP_TO_EDGE */
+   {  1978, 0x0000812F }, /* GL_CLAMP_TO_EDGE_SGIS */
+   {  2000, 0x00001500 }, /* GL_CLEAR */
+   {  2009, 0x000084E1 }, /* GL_CLIENT_ACTIVE_TEXTURE */
+   {  2034, 0x000084E1 }, /* GL_CLIENT_ACTIVE_TEXTURE_ARB */
+   {  2063, 0xFFFFFFFF }, /* GL_CLIENT_ALL_ATTRIB_BITS */
+   {  2089, 0x00000BB1 }, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
+   {  2118, 0x00000001 }, /* GL_CLIENT_PIXEL_STORE_BIT */
+   {  2144, 0x00000002 }, /* GL_CLIENT_VERTEX_ARRAY_BIT */
+   {  2171, 0x00003000 }, /* GL_CLIP_PLANE0 */
+   {  2186, 0x00003001 }, /* GL_CLIP_PLANE1 */
+   {  2201, 0x00003002 }, /* GL_CLIP_PLANE2 */
+   {  2216, 0x00003003 }, /* GL_CLIP_PLANE3 */
+   {  2231, 0x00003004 }, /* GL_CLIP_PLANE4 */
+   {  2246, 0x00003005 }, /* GL_CLIP_PLANE5 */
+   {  2261, 0x000080F0 }, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
+   {  2294, 0x00000A00 }, /* GL_COEFF */
+   {  2303, 0x00001800 }, /* GL_COLOR */
+   {  2312, 0x00008076 }, /* GL_COLOR_ARRAY */
+   {  2327, 0x00008898 }, /* GL_COLOR_ARRAY_BUFFER_BINDING */
+   {  2357, 0x00008898 }, /* GL_COLOR_ARRAY_BUFFER_BINDING_ARB */
+   {  2391, 0x00008090 }, /* GL_COLOR_ARRAY_POINTER */
+   {  2414, 0x00008081 }, /* GL_COLOR_ARRAY_SIZE */
+   {  2434, 0x00008083 }, /* GL_COLOR_ARRAY_STRIDE */
+   {  2456, 0x00008082 }, /* GL_COLOR_ARRAY_TYPE */
+   {  2476, 0x00008CE0 }, /* GL_COLOR_ATTACHMENT0_EXT */
+   {  2501, 0x00008CEA }, /* GL_COLOR_ATTACHMENT10_EXT */
+   {  2527, 0x00008CEB }, /* GL_COLOR_ATTACHMENT11_EXT */
+   {  2553, 0x00008CEC }, /* GL_COLOR_ATTACHMENT12_EXT */
+   {  2579, 0x00008CED }, /* GL_COLOR_ATTACHMENT13_EXT */
+   {  2605, 0x00008CEE }, /* GL_COLOR_ATTACHMENT14_EXT */
+   {  2631, 0x00008CEF }, /* GL_COLOR_ATTACHMENT15_EXT */
+   {  2657, 0x00008CE1 }, /* GL_COLOR_ATTACHMENT1_EXT */
+   {  2682, 0x00008CE2 }, /* GL_COLOR_ATTACHMENT2_EXT */
+   {  2707, 0x00008CE3 }, /* GL_COLOR_ATTACHMENT3_EXT */
+   {  2732, 0x00008CE4 }, /* GL_COLOR_ATTACHMENT4_EXT */
+   {  2757, 0x00008CE5 }, /* GL_COLOR_ATTACHMENT5_EXT */
+   {  2782, 0x00008CE6 }, /* GL_COLOR_ATTACHMENT6_EXT */
+   {  2807, 0x00008CE7 }, /* GL_COLOR_ATTACHMENT7_EXT */
+   {  2832, 0x00008CE8 }, /* GL_COLOR_ATTACHMENT8_EXT */
+   {  2857, 0x00008CE9 }, /* GL_COLOR_ATTACHMENT9_EXT */
+   {  2882, 0x00004000 }, /* GL_COLOR_BUFFER_BIT */
+   {  2902, 0x00000C22 }, /* GL_COLOR_CLEAR_VALUE */
+   {  2923, 0x00001900 }, /* GL_COLOR_INDEX */
+   {  2938, 0x00001603 }, /* GL_COLOR_INDEXES */
+   {  2955, 0x00000BF2 }, /* GL_COLOR_LOGIC_OP */
+   {  2973, 0x00000B57 }, /* GL_COLOR_MATERIAL */
+   {  2991, 0x00000B55 }, /* GL_COLOR_MATERIAL_FACE */
+   {  3014, 0x00000B56 }, /* GL_COLOR_MATERIAL_PARAMETER */
+   {  3042, 0x000080B1 }, /* GL_COLOR_MATRIX */
+   {  3058, 0x000080B1 }, /* GL_COLOR_MATRIX_SGI */
+   {  3078, 0x000080B2 }, /* GL_COLOR_MATRIX_STACK_DEPTH */
+   {  3106, 0x000080B2 }, /* GL_COLOR_MATRIX_STACK_DEPTH_SGI */
+   {  3138, 0x00008458 }, /* GL_COLOR_SUM */
+   {  3151, 0x00008458 }, /* GL_COLOR_SUM_ARB */
+   {  3168, 0x000080D0 }, /* GL_COLOR_TABLE */
+   {  3183, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE */
+   {  3209, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE_EXT */
+   {  3239, 0x000080DD }, /* GL_COLOR_TABLE_ALPHA_SIZE_SGI */
+   {  3269, 0x000080D7 }, /* GL_COLOR_TABLE_BIAS */
+   {  3289, 0x000080D7 }, /* GL_COLOR_TABLE_BIAS_SGI */
+   {  3313, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE */
+   {  3338, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE_EXT */
+   {  3367, 0x000080DC }, /* GL_COLOR_TABLE_BLUE_SIZE_SGI */
+   {  3396, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT */
+   {  3418, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT_EXT */
+   {  3444, 0x000080D8 }, /* GL_COLOR_TABLE_FORMAT_SGI */
+   {  3470, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE */
+   {  3496, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE_EXT */
+   {  3526, 0x000080DB }, /* GL_COLOR_TABLE_GREEN_SIZE_SGI */
+   {  3556, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE */
+   {  3586, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE_EXT */
+   {  3620, 0x000080DF }, /* GL_COLOR_TABLE_INTENSITY_SIZE_SGI */
+   {  3654, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE */
+   {  3684, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE_EXT */
+   {  3718, 0x000080DE }, /* GL_COLOR_TABLE_LUMINANCE_SIZE_SGI */
+   {  3752, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE */
+   {  3776, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE_EXT */
+   {  3804, 0x000080DA }, /* GL_COLOR_TABLE_RED_SIZE_SGI */
+   {  3832, 0x000080D6 }, /* GL_COLOR_TABLE_SCALE */
+   {  3853, 0x000080D6 }, /* GL_COLOR_TABLE_SCALE_SGI */
+   {  3878, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH */
+   {  3899, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH_EXT */
+   {  3924, 0x000080D9 }, /* GL_COLOR_TABLE_WIDTH_SGI */
+   {  3949, 0x00000C23 }, /* GL_COLOR_WRITEMASK */
+   {  3968, 0x00008570 }, /* GL_COMBINE */
+   {  3979, 0x00008503 }, /* GL_COMBINE4 */
+   {  3991, 0x00008572 }, /* GL_COMBINE_ALPHA */
+   {  4008, 0x00008572 }, /* GL_COMBINE_ALPHA_ARB */
+   {  4029, 0x00008572 }, /* GL_COMBINE_ALPHA_EXT */
+   {  4050, 0x00008570 }, /* GL_COMBINE_ARB */
+   {  4065, 0x00008570 }, /* GL_COMBINE_EXT */
+   {  4080, 0x00008571 }, /* GL_COMBINE_RGB */
+   {  4095, 0x00008571 }, /* GL_COMBINE_RGB_ARB */
+   {  4114, 0x00008571 }, /* GL_COMBINE_RGB_EXT */
+   {  4133, 0x0000884E }, /* GL_COMPARE_REF_DEPTH_TO_TEXTURE_EXT */
+   {  4169, 0x0000884E }, /* GL_COMPARE_R_TO_TEXTURE */
+   {  4193, 0x0000884E }, /* GL_COMPARE_R_TO_TEXTURE_ARB */
+   {  4221, 0x00001300 }, /* GL_COMPILE */
+   {  4232, 0x00001301 }, /* GL_COMPILE_AND_EXECUTE */
+   {  4255, 0x00008B81 }, /* GL_COMPILE_STATUS */
+   {  4273, 0x000084E9 }, /* GL_COMPRESSED_ALPHA */
+   {  4293, 0x000084E9 }, /* GL_COMPRESSED_ALPHA_ARB */
+   {  4317, 0x000084EC }, /* GL_COMPRESSED_INTENSITY */
+   {  4341, 0x000084EC }, /* GL_COMPRESSED_INTENSITY_ARB */
+   {  4369, 0x000084EA }, /* GL_COMPRESSED_LUMINANCE */
+   {  4393, 0x000084EB }, /* GL_COMPRESSED_LUMINANCE_ALPHA */
+   {  4423, 0x000084EB }, /* GL_COMPRESSED_LUMINANCE_ALPHA_ARB */
+   {  4457, 0x000084EA }, /* GL_COMPRESSED_LUMINANCE_ARB */
+   {  4485, 0x000084ED }, /* GL_COMPRESSED_RGB */
+   {  4503, 0x000084EE }, /* GL_COMPRESSED_RGBA */
+   {  4522, 0x000084EE }, /* GL_COMPRESSED_RGBA_ARB */
+   {  4545, 0x000086B1 }, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
+   {  4574, 0x000083F1 }, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
+   {  4607, 0x000083F2 }, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
+   {  4640, 0x000083F3 }, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
+   {  4673, 0x000084ED }, /* GL_COMPRESSED_RGB_ARB */
+   {  4695, 0x000086B0 }, /* GL_COMPRESSED_RGB_FXT1_3DFX */
+   {  4723, 0x000083F0 }, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
+   {  4755, 0x000086A3 }, /* GL_COMPRESSED_TEXTURE_FORMATS */
+   {  4785, 0x00008576 }, /* GL_CONSTANT */
+   {  4797, 0x00008003 }, /* GL_CONSTANT_ALPHA */
+   {  4815, 0x00008003 }, /* GL_CONSTANT_ALPHA_EXT */
+   {  4837, 0x00008576 }, /* GL_CONSTANT_ARB */
+   {  4853, 0x00001207 }, /* GL_CONSTANT_ATTENUATION */
+   {  4877, 0x00008151 }, /* GL_CONSTANT_BORDER_HP */
+   {  4899, 0x00008001 }, /* GL_CONSTANT_COLOR */
+   {  4917, 0x00008001 }, /* GL_CONSTANT_COLOR_EXT */
+   {  4939, 0x00008576 }, /* GL_CONSTANT_EXT */
+   {  4955, 0x00008010 }, /* GL_CONVOLUTION_1D */
+   {  4973, 0x00008011 }, /* GL_CONVOLUTION_2D */
+   {  4991, 0x00008154 }, /* GL_CONVOLUTION_BORDER_COLOR */
+   {  5019, 0x00008154 }, /* GL_CONVOLUTION_BORDER_COLOR_HP */
+   {  5050, 0x00008013 }, /* GL_CONVOLUTION_BORDER_MODE */
+   {  5077, 0x00008013 }, /* GL_CONVOLUTION_BORDER_MODE_EXT */
+   {  5108, 0x00008015 }, /* GL_CONVOLUTION_FILTER_BIAS */
+   {  5135, 0x00008015 }, /* GL_CONVOLUTION_FILTER_BIAS_EXT */
+   {  5166, 0x00008014 }, /* GL_CONVOLUTION_FILTER_SCALE */
+   {  5194, 0x00008014 }, /* GL_CONVOLUTION_FILTER_SCALE_EXT */
+   {  5226, 0x00008017 }, /* GL_CONVOLUTION_FORMAT */
+   {  5248, 0x00008017 }, /* GL_CONVOLUTION_FORMAT_EXT */
+   {  5274, 0x00008019 }, /* GL_CONVOLUTION_HEIGHT */
+   {  5296, 0x00008019 }, /* GL_CONVOLUTION_HEIGHT_EXT */
+   {  5322, 0x00008018 }, /* GL_CONVOLUTION_WIDTH */
+   {  5343, 0x00008018 }, /* GL_CONVOLUTION_WIDTH_EXT */
+   {  5368, 0x00008862 }, /* GL_COORD_REPLACE */
+   {  5385, 0x00008862 }, /* GL_COORD_REPLACE_ARB */
+   {  5406, 0x00008862 }, /* GL_COORD_REPLACE_NV */
+   {  5426, 0x00001503 }, /* GL_COPY */
+   {  5434, 0x0000150C }, /* GL_COPY_INVERTED */
+   {  5451, 0x00000706 }, /* GL_COPY_PIXEL_TOKEN */
+   {  5471, 0x00000B44 }, /* GL_CULL_FACE */
+   {  5484, 0x00000B45 }, /* GL_CULL_FACE_MODE */
+   {  5502, 0x000081AA }, /* GL_CULL_VERTEX_EXT */
+   {  5521, 0x000081AC }, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
+   {  5553, 0x000081AB }, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
+   {  5588, 0x00008626 }, /* GL_CURRENT_ATTRIB_NV */
+   {  5609, 0x00000001 }, /* GL_CURRENT_BIT */
+   {  5624, 0x00000B00 }, /* GL_CURRENT_COLOR */
+   {  5641, 0x00008453 }, /* GL_CURRENT_FOG_COORD */
+   {  5662, 0x00008453 }, /* GL_CURRENT_FOG_COORDINATE */
+   {  5688, 0x00000B01 }, /* GL_CURRENT_INDEX */
+   {  5705, 0x00008641 }, /* GL_CURRENT_MATRIX_ARB */
+   {  5727, 0x00008845 }, /* GL_CURRENT_MATRIX_INDEX_ARB */
+   {  5755, 0x00008641 }, /* GL_CURRENT_MATRIX_NV */
+   {  5776, 0x00008640 }, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
+   {  5810, 0x00008640 }, /* GL_CURRENT_MATRIX_STACK_DEPTH_NV */
+   {  5843, 0x00000B02 }, /* GL_CURRENT_NORMAL */
+   {  5861, 0x00008843 }, /* GL_CURRENT_PALETTE_MATRIX_ARB */
+   {  5891, 0x00008B8D }, /* GL_CURRENT_PROGRAM */
+   {  5910, 0x00008865 }, /* GL_CURRENT_QUERY */
+   {  5927, 0x00008865 }, /* GL_CURRENT_QUERY_ARB */
+   {  5948, 0x00000B04 }, /* GL_CURRENT_RASTER_COLOR */
+   {  5972, 0x00000B09 }, /* GL_CURRENT_RASTER_DISTANCE */
+   {  5999, 0x00000B05 }, /* GL_CURRENT_RASTER_INDEX */
+   {  6023, 0x00000B07 }, /* GL_CURRENT_RASTER_POSITION */
+   {  6050, 0x00000B08 }, /* GL_CURRENT_RASTER_POSITION_VALID */
+   {  6083, 0x00000B06 }, /* GL_CURRENT_RASTER_TEXTURE_COORDS */
+   {  6116, 0x00008459 }, /* GL_CURRENT_SECONDARY_COLOR */
+   {  6143, 0x00000B03 }, /* GL_CURRENT_TEXTURE_COORDS */
+   {  6169, 0x00008626 }, /* GL_CURRENT_VERTEX_ATTRIB */
+   {  6194, 0x00008626 }, /* GL_CURRENT_VERTEX_ATTRIB_ARB */
+   {  6223, 0x000086A8 }, /* GL_CURRENT_WEIGHT_ARB */
+   {  6245, 0x00000900 }, /* GL_CW */
+   {  6251, 0x0000875B }, /* GL_DEBUG_ASSERT_MESA */
+   {  6272, 0x00008759 }, /* GL_DEBUG_OBJECT_MESA */
+   {  6293, 0x0000875A }, /* GL_DEBUG_PRINT_MESA */
+   {  6313, 0x00002101 }, /* GL_DECAL */
+   {  6322, 0x00001E03 }, /* GL_DECR */
+   {  6330, 0x00008508 }, /* GL_DECR_WRAP */
+   {  6343, 0x00008508 }, /* GL_DECR_WRAP_EXT */
+   {  6360, 0x00008B80 }, /* GL_DELETE_STATUS */
+   {  6377, 0x00001801 }, /* GL_DEPTH */
+   {  6386, 0x00008D00 }, /* GL_DEPTH_ATTACHMENT_EXT */
+   {  6410, 0x00000D1F }, /* GL_DEPTH_BIAS */
+   {  6424, 0x00000D56 }, /* GL_DEPTH_BITS */
+   {  6438, 0x00008891 }, /* GL_DEPTH_BOUNDS_EXT */
+   {  6458, 0x00008890 }, /* GL_DEPTH_BOUNDS_TEST_EXT */
+   {  6483, 0x00000100 }, /* GL_DEPTH_BUFFER_BIT */
+   {  6503, 0x0000864F }, /* GL_DEPTH_CLAMP_NV */
+   {  6521, 0x00000B73 }, /* GL_DEPTH_CLEAR_VALUE */
+   {  6542, 0x00001902 }, /* GL_DEPTH_COMPONENT */
+   {  6561, 0x000081A5 }, /* GL_DEPTH_COMPONENT16 */
+   {  6582, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_ARB */
+   {  6607, 0x000081A5 }, /* GL_DEPTH_COMPONENT16_SGIX */
+   {  6633, 0x000081A6 }, /* GL_DEPTH_COMPONENT24 */
+   {  6654, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_ARB */
+   {  6679, 0x000081A6 }, /* GL_DEPTH_COMPONENT24_SGIX */
+   {  6705, 0x000081A7 }, /* GL_DEPTH_COMPONENT32 */
+   {  6726, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_ARB */
+   {  6751, 0x000081A7 }, /* GL_DEPTH_COMPONENT32_SGIX */
+   {  6777, 0x00000B74 }, /* GL_DEPTH_FUNC */
+   {  6791, 0x00000B70 }, /* GL_DEPTH_RANGE */
+   {  6806, 0x00000D1E }, /* GL_DEPTH_SCALE */
+   {  6821, 0x000084F9 }, /* GL_DEPTH_STENCIL_NV */
+   {  6841, 0x0000886F }, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+   {  6869, 0x0000886E }, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+   {  6897, 0x00000B71 }, /* GL_DEPTH_TEST */
+   {  6911, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE */
+   {  6933, 0x0000884B }, /* GL_DEPTH_TEXTURE_MODE_ARB */
+   {  6959, 0x00000B72 }, /* GL_DEPTH_WRITEMASK */
+   {  6978, 0x00001201 }, /* GL_DIFFUSE */
+   {  6989, 0x00000BD0 }, /* GL_DITHER */
+   {  6999, 0x00000A02 }, /* GL_DOMAIN */
+   {  7009, 0x00001100 }, /* GL_DONT_CARE */
+   {  7022, 0x000086AE }, /* GL_DOT3_RGB */
+   {  7034, 0x000086AF }, /* GL_DOT3_RGBA */
+   {  7047, 0x000086AF }, /* GL_DOT3_RGBA_ARB */
+   {  7064, 0x00008741 }, /* GL_DOT3_RGBA_EXT */
+   {  7081, 0x000086AE }, /* GL_DOT3_RGB_ARB */
+   {  7097, 0x00008740 }, /* GL_DOT3_RGB_EXT */
+   {  7113, 0x0000140A }, /* GL_DOUBLE */
+   {  7123, 0x00000C32 }, /* GL_DOUBLEBUFFER */
+   {  7139, 0x00000C01 }, /* GL_DRAW_BUFFER */
+   {  7154, 0x00008825 }, /* GL_DRAW_BUFFER0 */
+   {  7170, 0x00008825 }, /* GL_DRAW_BUFFER0_ARB */
+   {  7190, 0x00008825 }, /* GL_DRAW_BUFFER0_ATI */
+   {  7210, 0x00008826 }, /* GL_DRAW_BUFFER1 */
+   {  7226, 0x0000882F }, /* GL_DRAW_BUFFER10 */
+   {  7243, 0x0000882F }, /* GL_DRAW_BUFFER10_ARB */
+   {  7264, 0x0000882F }, /* GL_DRAW_BUFFER10_ATI */
+   {  7285, 0x00008830 }, /* GL_DRAW_BUFFER11 */
+   {  7302, 0x00008830 }, /* GL_DRAW_BUFFER11_ARB */
+   {  7323, 0x00008830 }, /* GL_DRAW_BUFFER11_ATI */
+   {  7344, 0x00008831 }, /* GL_DRAW_BUFFER12 */
+   {  7361, 0x00008831 }, /* GL_DRAW_BUFFER12_ARB */
+   {  7382, 0x00008831 }, /* GL_DRAW_BUFFER12_ATI */
+   {  7403, 0x00008832 }, /* GL_DRAW_BUFFER13 */
+   {  7420, 0x00008832 }, /* GL_DRAW_BUFFER13_ARB */
+   {  7441, 0x00008832 }, /* GL_DRAW_BUFFER13_ATI */
+   {  7462, 0x00008833 }, /* GL_DRAW_BUFFER14 */
+   {  7479, 0x00008833 }, /* GL_DRAW_BUFFER14_ARB */
+   {  7500, 0x00008833 }, /* GL_DRAW_BUFFER14_ATI */
+   {  7521, 0x00008834 }, /* GL_DRAW_BUFFER15 */
+   {  7538, 0x00008834 }, /* GL_DRAW_BUFFER15_ARB */
+   {  7559, 0x00008834 }, /* GL_DRAW_BUFFER15_ATI */
+   {  7580, 0x00008826 }, /* GL_DRAW_BUFFER1_ARB */
+   {  7600, 0x00008826 }, /* GL_DRAW_BUFFER1_ATI */
+   {  7620, 0x00008827 }, /* GL_DRAW_BUFFER2 */
+   {  7636, 0x00008827 }, /* GL_DRAW_BUFFER2_ARB */
+   {  7656, 0x00008827 }, /* GL_DRAW_BUFFER2_ATI */
+   {  7676, 0x00008828 }, /* GL_DRAW_BUFFER3 */
+   {  7692, 0x00008828 }, /* GL_DRAW_BUFFER3_ARB */
+   {  7712, 0x00008828 }, /* GL_DRAW_BUFFER3_ATI */
+   {  7732, 0x00008829 }, /* GL_DRAW_BUFFER4 */
+   {  7748, 0x00008829 }, /* GL_DRAW_BUFFER4_ARB */
+   {  7768, 0x00008829 }, /* GL_DRAW_BUFFER4_ATI */
+   {  7788, 0x0000882A }, /* GL_DRAW_BUFFER5 */
+   {  7804, 0x0000882A }, /* GL_DRAW_BUFFER5_ARB */
+   {  7824, 0x0000882A }, /* GL_DRAW_BUFFER5_ATI */
+   {  7844, 0x0000882B }, /* GL_DRAW_BUFFER6 */
+   {  7860, 0x0000882B }, /* GL_DRAW_BUFFER6_ARB */
+   {  7880, 0x0000882B }, /* GL_DRAW_BUFFER6_ATI */
+   {  7900, 0x0000882C }, /* GL_DRAW_BUFFER7 */
+   {  7916, 0x0000882C }, /* GL_DRAW_BUFFER7_ARB */
+   {  7936, 0x0000882C }, /* GL_DRAW_BUFFER7_ATI */
+   {  7956, 0x0000882D }, /* GL_DRAW_BUFFER8 */
+   {  7972, 0x0000882D }, /* GL_DRAW_BUFFER8_ARB */
+   {  7992, 0x0000882D }, /* GL_DRAW_BUFFER8_ATI */
+   {  8012, 0x0000882E }, /* GL_DRAW_BUFFER9 */
+   {  8028, 0x0000882E }, /* GL_DRAW_BUFFER9_ARB */
+   {  8048, 0x0000882E }, /* GL_DRAW_BUFFER9_ATI */
+   {  8068, 0x00008CA6 }, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
+   {  8100, 0x00008CA9 }, /* GL_DRAW_FRAMEBUFFER_EXT */
+   {  8124, 0x00000705 }, /* GL_DRAW_PIXEL_TOKEN */
+   {  8144, 0x00000304 }, /* GL_DST_ALPHA */
+   {  8157, 0x00000306 }, /* GL_DST_COLOR */
+   {  8170, 0x000088EA }, /* GL_DYNAMIC_COPY */
+   {  8186, 0x000088EA }, /* GL_DYNAMIC_COPY_ARB */
+   {  8206, 0x000088E8 }, /* GL_DYNAMIC_DRAW */
+   {  8222, 0x000088E8 }, /* GL_DYNAMIC_DRAW_ARB */
+   {  8242, 0x000088E9 }, /* GL_DYNAMIC_READ */
+   {  8258, 0x000088E9 }, /* GL_DYNAMIC_READ_ARB */
+   {  8278, 0x00000B43 }, /* GL_EDGE_FLAG */
+   {  8291, 0x00008079 }, /* GL_EDGE_FLAG_ARRAY */
+   {  8310, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+   {  8344, 0x0000889B }, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING_ARB */
+   {  8382, 0x00008093 }, /* GL_EDGE_FLAG_ARRAY_POINTER */
+   {  8409, 0x0000808C }, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+   {  8435, 0x00008893 }, /* GL_ELEMENT_ARRAY_BUFFER */
+   {  8459, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+   {  8491, 0x00008895 }, /* GL_ELEMENT_ARRAY_BUFFER_BINDING_ARB */
+   {  8527, 0x00001600 }, /* GL_EMISSION */
+   {  8539, 0x00002000 }, /* GL_ENABLE_BIT */
+   {  8553, 0x00000202 }, /* GL_EQUAL */
+   {  8562, 0x00001509 }, /* GL_EQUIV */
+   {  8571, 0x00010000 }, /* GL_EVAL_BIT */
+   {  8583, 0x00000800 }, /* GL_EXP */
+   {  8590, 0x00000801 }, /* GL_EXP2 */
+   {  8598, 0x00001F03 }, /* GL_EXTENSIONS */
+   {  8612, 0x00002400 }, /* GL_EYE_LINEAR */
+   {  8626, 0x00002502 }, /* GL_EYE_PLANE */
+   {  8639, 0x0000855C }, /* GL_EYE_PLANE_ABSOLUTE_NV */
+   {  8664, 0x0000855B }, /* GL_EYE_RADIAL_NV */
+   {  8681, 0x00000000 }, /* GL_FALSE */
+   {  8690, 0x00001101 }, /* GL_FASTEST */
+   {  8701, 0x00001C01 }, /* GL_FEEDBACK */
+   {  8713, 0x00000DF0 }, /* GL_FEEDBACK_BUFFER_POINTER */
+   {  8740, 0x00000DF1 }, /* GL_FEEDBACK_BUFFER_SIZE */
+   {  8764, 0x00000DF2 }, /* GL_FEEDBACK_BUFFER_TYPE */
+   {  8788, 0x00001B02 }, /* GL_FILL */
+   {  8796, 0x00001D00 }, /* GL_FLAT */
+   {  8804, 0x00001406 }, /* GL_FLOAT */
+   {  8813, 0x00008B5A }, /* GL_FLOAT_MAT2 */
+   {  8827, 0x00008B5A }, /* GL_FLOAT_MAT2_ARB */
+   {  8845, 0x00008B5B }, /* GL_FLOAT_MAT3 */
+   {  8859, 0x00008B5B }, /* GL_FLOAT_MAT3_ARB */
+   {  8877, 0x00008B5C }, /* GL_FLOAT_MAT4 */
+   {  8891, 0x00008B5C }, /* GL_FLOAT_MAT4_ARB */
+   {  8909, 0x00008B50 }, /* GL_FLOAT_VEC2 */
+   {  8923, 0x00008B50 }, /* GL_FLOAT_VEC2_ARB */
+   {  8941, 0x00008B51 }, /* GL_FLOAT_VEC3 */
+   {  8955, 0x00008B51 }, /* GL_FLOAT_VEC3_ARB */
+   {  8973, 0x00008B52 }, /* GL_FLOAT_VEC4 */
+   {  8987, 0x00008B52 }, /* GL_FLOAT_VEC4_ARB */
+   {  9005, 0x00000B60 }, /* GL_FOG */
+   {  9012, 0x00000080 }, /* GL_FOG_BIT */
+   {  9023, 0x00000B66 }, /* GL_FOG_COLOR */
+   {  9036, 0x00008451 }, /* GL_FOG_COORD */
+   {  9049, 0x00008451 }, /* GL_FOG_COORDINATE */
+   {  9067, 0x00008457 }, /* GL_FOG_COORDINATE_ARRAY */
+   {  9091, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+   {  9130, 0x0000889D }, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING_ARB */
+   {  9173, 0x00008456 }, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+   {  9205, 0x00008455 }, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+   {  9236, 0x00008454 }, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+   {  9265, 0x00008450 }, /* GL_FOG_COORDINATE_SOURCE */
+   {  9290, 0x00008457 }, /* GL_FOG_COORD_ARRAY */
+   {  9309, 0x0000889D }, /* GL_FOG_COORD_ARRAY_BUFFER_BINDING */
+   {  9343, 0x00008456 }, /* GL_FOG_COORD_ARRAY_POINTER */
+   {  9370, 0x00008455 }, /* GL_FOG_COORD_ARRAY_STRIDE */
+   {  9396, 0x00008454 }, /* GL_FOG_COORD_ARRAY_TYPE */
+   {  9420, 0x00008450 }, /* GL_FOG_COORD_SRC */
+   {  9437, 0x00000B62 }, /* GL_FOG_DENSITY */
+   {  9452, 0x0000855A }, /* GL_FOG_DISTANCE_MODE_NV */
+   {  9476, 0x00000B64 }, /* GL_FOG_END */
+   {  9487, 0x00000C54 }, /* GL_FOG_HINT */
+   {  9499, 0x00000B61 }, /* GL_FOG_INDEX */
+   {  9512, 0x00000B65 }, /* GL_FOG_MODE */
+   {  9524, 0x00008198 }, /* GL_FOG_OFFSET_SGIX */
+   {  9543, 0x00008199 }, /* GL_FOG_OFFSET_VALUE_SGIX */
+   {  9568, 0x00000B63 }, /* GL_FOG_START */
+   {  9581, 0x00008452 }, /* GL_FRAGMENT_DEPTH */
+   {  9599, 0x00008804 }, /* GL_FRAGMENT_PROGRAM_ARB */
+   {  9623, 0x00008B30 }, /* GL_FRAGMENT_SHADER */
+   {  9642, 0x00008B30 }, /* GL_FRAGMENT_SHADER_ARB */
+   {  9665, 0x00008B8B }, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+   {  9700, 0x00008CD1 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
+   {  9742, 0x00008CD0 }, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
+   {  9784, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
+   {  9833, 0x00008CD3 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
+   {  9885, 0x00008CD4 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER_EXT */
+   {  9929, 0x00008CD2 }, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
+   {  9973, 0x00008CA6 }, /* GL_FRAMEBUFFER_BINDING_EXT */
+   { 10000, 0x00008CD5 }, /* GL_FRAMEBUFFER_COMPLETE_EXT */
+   { 10028, 0x00008D40 }, /* GL_FRAMEBUFFER_EXT */
+   { 10047, 0x00008CD6 }, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
+   { 10088, 0x00008CD9 }, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+   { 10129, 0x00008CDB }, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
+   { 10171, 0x00008CD8 }, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+   { 10222, 0x00008CDA }, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+   { 10260, 0x00008CD7 }, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
+   { 10309, 0x00008CDC }, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
+   { 10351, 0x00008CDE }, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+   { 10383, 0x00008CDD }, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
+   { 10414, 0x00000404 }, /* GL_FRONT */
+   { 10423, 0x00000408 }, /* GL_FRONT_AND_BACK */
+   { 10441, 0x00000B46 }, /* GL_FRONT_FACE */
+   { 10455, 0x00000400 }, /* GL_FRONT_LEFT */
+   { 10469, 0x00000401 }, /* GL_FRONT_RIGHT */
+   { 10484, 0x00008006 }, /* GL_FUNC_ADD */
+   { 10496, 0x00008006 }, /* GL_FUNC_ADD_EXT */
+   { 10512, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT */
+   { 10537, 0x0000800B }, /* GL_FUNC_REVERSE_SUBTRACT_EXT */
+   { 10566, 0x0000800A }, /* GL_FUNC_SUBTRACT */
+   { 10583, 0x0000800A }, /* GL_FUNC_SUBTRACT_EXT */
+   { 10604, 0x00008191 }, /* GL_GENERATE_MIPMAP */
+   { 10623, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT */
+   { 10647, 0x00008192 }, /* GL_GENERATE_MIPMAP_HINT_SGIS */
+   { 10676, 0x00008191 }, /* GL_GENERATE_MIPMAP_SGIS */
+   { 10700, 0x00000206 }, /* GL_GEQUAL */
+   { 10710, 0x00008009 }, /* GL_GL_BLEND_EQUATION_RGB */
+   { 10735, 0x00008C4A }, /* GL_GL_COMPRESSED_SLUMINANCE */
+   { 10763, 0x00008C4B }, /* GL_GL_COMPRESSED_SLUMINANCE_ALPHA */
+   { 10797, 0x00008C48 }, /* GL_GL_COMPRESSED_SRGB */
+   { 10819, 0x00008C49 }, /* GL_GL_COMPRESSED_SRGB_ALPHA */
+   { 10847, 0x0000845F }, /* GL_GL_CURRENT_RASTER_SECONDARY_COLOR */
+   { 10884, 0x00008B65 }, /* GL_GL_FLOAT_MAT2x3 */
+   { 10903, 0x00008B66 }, /* GL_GL_FLOAT_MAT2x4 */
+   { 10922, 0x00008B67 }, /* GL_GL_FLOAT_MAT3x2 */
+   { 10941, 0x00008B68 }, /* GL_GL_FLOAT_MAT3x4 */
+   { 10960, 0x00008B69 }, /* GL_GL_FLOAT_MAT4x2 */
+   { 10979, 0x00008B6A }, /* GL_GL_FLOAT_MAT4x3 */
+   { 10998, 0x000088EB }, /* GL_GL_PIXEL_PACK_BUFFER */
+   { 11022, 0x000088ED }, /* GL_GL_PIXEL_PACK_BUFFER_BINDING */
+   { 11054, 0x000088EC }, /* GL_GL_PIXEL_UNPACK_BUFFER */
+   { 11080, 0x000088EF }, /* GL_GL_PIXEL_UNPACK_BUFFER_BINDING */
+   { 11114, 0x00008C46 }, /* GL_GL_SLUMINANCE */
+   { 11131, 0x00008C47 }, /* GL_GL_SLUMINANCE8 */
+   { 11149, 0x00008C45 }, /* GL_GL_SLUMINANCE8_ALPHA8 */
+   { 11174, 0x00008C44 }, /* GL_GL_SLUMINANCE_ALPHA */
+   { 11197, 0x00008C40 }, /* GL_GL_SRGB */
+   { 11208, 0x00008C41 }, /* GL_GL_SRGB8 */
+   { 11220, 0x00008C43 }, /* GL_GL_SRGB8_ALPHA8 */
+   { 11239, 0x00008C42 }, /* GL_GL_SRGB_ALPHA */
+   { 11256, 0x00000204 }, /* GL_GREATER */
+   { 11267, 0x00001904 }, /* GL_GREEN */
+   { 11276, 0x00000D19 }, /* GL_GREEN_BIAS */
+   { 11290, 0x00000D53 }, /* GL_GREEN_BITS */
+   { 11304, 0x00000D18 }, /* GL_GREEN_SCALE */
+   { 11319, 0x00008000 }, /* GL_HINT_BIT */
+   { 11331, 0x00008024 }, /* GL_HISTOGRAM */
+   { 11344, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE */
+   { 11368, 0x0000802B }, /* GL_HISTOGRAM_ALPHA_SIZE_EXT */
+   { 11396, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE */
+   { 11419, 0x0000802A }, /* GL_HISTOGRAM_BLUE_SIZE_EXT */
+   { 11446, 0x00008024 }, /* GL_HISTOGRAM_EXT */
+   { 11463, 0x00008027 }, /* GL_HISTOGRAM_FORMAT */
+   { 11483, 0x00008027 }, /* GL_HISTOGRAM_FORMAT_EXT */
+   { 11507, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE */
+   { 11531, 0x00008029 }, /* GL_HISTOGRAM_GREEN_SIZE_EXT */
+   { 11559, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+   { 11587, 0x0000802C }, /* GL_HISTOGRAM_LUMINANCE_SIZE_EXT */
+   { 11619, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE */
+   { 11641, 0x00008028 }, /* GL_HISTOGRAM_RED_SIZE_EXT */
+   { 11667, 0x0000802D }, /* GL_HISTOGRAM_SINK */
+   { 11685, 0x0000802D }, /* GL_HISTOGRAM_SINK_EXT */
+   { 11707, 0x00008026 }, /* GL_HISTOGRAM_WIDTH */
+   { 11726, 0x00008026 }, /* GL_HISTOGRAM_WIDTH_EXT */
+   { 11749, 0x0000862A }, /* GL_IDENTITY_NV */
+   { 11764, 0x00008150 }, /* GL_IGNORE_BORDER_HP */
+   { 11784, 0x00008B9B }, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+   { 11824, 0x00008B9A }, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+   { 11862, 0x00001E02 }, /* GL_INCR */
+   { 11870, 0x00008507 }, /* GL_INCR_WRAP */
+   { 11883, 0x00008507 }, /* GL_INCR_WRAP_EXT */
+   { 11900, 0x00008077 }, /* GL_INDEX_ARRAY */
+   { 11915, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+   { 11945, 0x00008899 }, /* GL_INDEX_ARRAY_BUFFER_BINDING_ARB */
+   { 11979, 0x00008091 }, /* GL_INDEX_ARRAY_POINTER */
+   { 12002, 0x00008086 }, /* GL_INDEX_ARRAY_STRIDE */
+   { 12024, 0x00008085 }, /* GL_INDEX_ARRAY_TYPE */
+   { 12044, 0x00000D51 }, /* GL_INDEX_BITS */
+   { 12058, 0x00000C20 }, /* GL_INDEX_CLEAR_VALUE */
+   { 12079, 0x00000BF1 }, /* GL_INDEX_LOGIC_OP */
+   { 12097, 0x00000C30 }, /* GL_INDEX_MODE */
+   { 12111, 0x00000D13 }, /* GL_INDEX_OFFSET */
+   { 12127, 0x00000D12 }, /* GL_INDEX_SHIFT */
+   { 12142, 0x00000C21 }, /* GL_INDEX_WRITEMASK */
+   { 12161, 0x00008B84 }, /* GL_INFO_LOG_LENGTH */
+   { 12180, 0x00001404 }, /* GL_INT */
+   { 12187, 0x00008049 }, /* GL_INTENSITY */
+   { 12200, 0x0000804C }, /* GL_INTENSITY12 */
+   { 12215, 0x0000804C }, /* GL_INTENSITY12_EXT */
+   { 12234, 0x0000804D }, /* GL_INTENSITY16 */
+   { 12249, 0x0000804D }, /* GL_INTENSITY16_EXT */
+   { 12268, 0x0000804A }, /* GL_INTENSITY4 */
+   { 12282, 0x0000804A }, /* GL_INTENSITY4_EXT */
+   { 12300, 0x0000804B }, /* GL_INTENSITY8 */
+   { 12314, 0x0000804B }, /* GL_INTENSITY8_EXT */
+   { 12332, 0x00008049 }, /* GL_INTENSITY_EXT */
+   { 12349, 0x00008575 }, /* GL_INTERPOLATE */
+   { 12364, 0x00008575 }, /* GL_INTERPOLATE_ARB */
+   { 12383, 0x00008575 }, /* GL_INTERPOLATE_EXT */
+   { 12402, 0x00008B53 }, /* GL_INT_VEC2 */
+   { 12414, 0x00008B53 }, /* GL_INT_VEC2_ARB */
+   { 12430, 0x00008B54 }, /* GL_INT_VEC3 */
+   { 12442, 0x00008B54 }, /* GL_INT_VEC3_ARB */
+   { 12458, 0x00008B55 }, /* GL_INT_VEC4 */
+   { 12470, 0x00008B55 }, /* GL_INT_VEC4_ARB */
+   { 12486, 0x00000500 }, /* GL_INVALID_ENUM */
+   { 12502, 0x00000506 }, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
+   { 12539, 0x00000502 }, /* GL_INVALID_OPERATION */
+   { 12560, 0x00000501 }, /* GL_INVALID_VALUE */
+   { 12577, 0x0000862B }, /* GL_INVERSE_NV */
+   { 12591, 0x0000862D }, /* GL_INVERSE_TRANSPOSE_NV */
+   { 12615, 0x0000150A }, /* GL_INVERT */
+   { 12625, 0x00001E00 }, /* GL_KEEP */
+   { 12633, 0x00000406 }, /* GL_LEFT */
+   { 12641, 0x00000203 }, /* GL_LEQUAL */
+   { 12651, 0x00000201 }, /* GL_LESS */
+   { 12659, 0x00004000 }, /* GL_LIGHT0 */
+   { 12669, 0x00004001 }, /* GL_LIGHT1 */
+   { 12679, 0x00004002 }, /* GL_LIGHT2 */
+   { 12689, 0x00004003 }, /* GL_LIGHT3 */
+   { 12699, 0x00004004 }, /* GL_LIGHT4 */
+   { 12709, 0x00004005 }, /* GL_LIGHT5 */
+   { 12719, 0x00004006 }, /* GL_LIGHT6 */
+   { 12729, 0x00004007 }, /* GL_LIGHT7 */
+   { 12739, 0x00000B50 }, /* GL_LIGHTING */
+   { 12751, 0x00000040 }, /* GL_LIGHTING_BIT */
+   { 12767, 0x00000B53 }, /* GL_LIGHT_MODEL_AMBIENT */
+   { 12790, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+   { 12819, 0x000081F8 }, /* GL_LIGHT_MODEL_COLOR_CONTROL_EXT */
+   { 12852, 0x00000B51 }, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+   { 12880, 0x00000B52 }, /* GL_LIGHT_MODEL_TWO_SIDE */
+   { 12904, 0x00001B01 }, /* GL_LINE */
+   { 12912, 0x00002601 }, /* GL_LINEAR */
+   { 12922, 0x00001208 }, /* GL_LINEAR_ATTENUATION */
+   { 12944, 0x00008170 }, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+   { 12974, 0x0000844F }, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+   { 13005, 0x00002703 }, /* GL_LINEAR_MIPMAP_LINEAR */
+   { 13029, 0x00002701 }, /* GL_LINEAR_MIPMAP_NEAREST */
+   { 13054, 0x00000001 }, /* GL_LINES */
+   { 13063, 0x00000004 }, /* GL_LINE_BIT */
+   { 13075, 0x00000002 }, /* GL_LINE_LOOP */
+   { 13088, 0x00000707 }, /* GL_LINE_RESET_TOKEN */
+   { 13108, 0x00000B20 }, /* GL_LINE_SMOOTH */
+   { 13123, 0x00000C52 }, /* GL_LINE_SMOOTH_HINT */
+   { 13143, 0x00000B24 }, /* GL_LINE_STIPPLE */
+   { 13159, 0x00000B25 }, /* GL_LINE_STIPPLE_PATTERN */
+   { 13183, 0x00000B26 }, /* GL_LINE_STIPPLE_REPEAT */
+   { 13206, 0x00000003 }, /* GL_LINE_STRIP */
+   { 13220, 0x00000702 }, /* GL_LINE_TOKEN */
+   { 13234, 0x00000B21 }, /* GL_LINE_WIDTH */
+   { 13248, 0x00000B23 }, /* GL_LINE_WIDTH_GRANULARITY */
+   { 13274, 0x00000B22 }, /* GL_LINE_WIDTH_RANGE */
+   { 13294, 0x00008B82 }, /* GL_LINK_STATUS */
+   { 13309, 0x00000B32 }, /* GL_LIST_BASE */
+   { 13322, 0x00020000 }, /* GL_LIST_BIT */
+   { 13334, 0x00000B33 }, /* GL_LIST_INDEX */
+   { 13348, 0x00000B30 }, /* GL_LIST_MODE */
+   { 13361, 0x00000101 }, /* GL_LOAD */
+   { 13369, 0x00000BF1 }, /* GL_LOGIC_OP */
+   { 13381, 0x00000BF0 }, /* GL_LOGIC_OP_MODE */
+   { 13398, 0x00008CA1 }, /* GL_LOWER_LEFT */
+   { 13412, 0x00001909 }, /* GL_LUMINANCE */
+   { 13425, 0x00008041 }, /* GL_LUMINANCE12 */
+   { 13440, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12 */
+   { 13463, 0x00008047 }, /* GL_LUMINANCE12_ALPHA12_EXT */
+   { 13490, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4 */
+   { 13512, 0x00008046 }, /* GL_LUMINANCE12_ALPHA4_EXT */
+   { 13538, 0x00008041 }, /* GL_LUMINANCE12_EXT */
+   { 13557, 0x00008042 }, /* GL_LUMINANCE16 */
+   { 13572, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16 */
+   { 13595, 0x00008048 }, /* GL_LUMINANCE16_ALPHA16_EXT */
+   { 13622, 0x00008042 }, /* GL_LUMINANCE16_EXT */
+   { 13641, 0x0000803F }, /* GL_LUMINANCE4 */
+   { 13655, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4 */
+   { 13676, 0x00008043 }, /* GL_LUMINANCE4_ALPHA4_EXT */
+   { 13701, 0x0000803F }, /* GL_LUMINANCE4_EXT */
+   { 13719, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2 */
+   { 13740, 0x00008044 }, /* GL_LUMINANCE6_ALPHA2_EXT */
+   { 13765, 0x00008040 }, /* GL_LUMINANCE8 */
+   { 13779, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8 */
+   { 13800, 0x00008045 }, /* GL_LUMINANCE8_ALPHA8_EXT */
+   { 13825, 0x00008040 }, /* GL_LUMINANCE8_EXT */
+   { 13843, 0x0000190A }, /* GL_LUMINANCE_ALPHA */
+   { 13862, 0x00000D90 }, /* GL_MAP1_COLOR_4 */
+   { 13878, 0x00000DD0 }, /* GL_MAP1_GRID_DOMAIN */
+   { 13898, 0x00000DD1 }, /* GL_MAP1_GRID_SEGMENTS */
+   { 13920, 0x00000D91 }, /* GL_MAP1_INDEX */
+   { 13934, 0x00000D92 }, /* GL_MAP1_NORMAL */
+   { 13949, 0x00000D93 }, /* GL_MAP1_TEXTURE_COORD_1 */
+   { 13973, 0x00000D94 }, /* GL_MAP1_TEXTURE_COORD_2 */
+   { 13997, 0x00000D95 }, /* GL_MAP1_TEXTURE_COORD_3 */
+   { 14021, 0x00000D96 }, /* GL_MAP1_TEXTURE_COORD_4 */
+   { 14045, 0x00000D97 }, /* GL_MAP1_VERTEX_3 */
+   { 14062, 0x00000D98 }, /* GL_MAP1_VERTEX_4 */
+   { 14079, 0x00008660 }, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+   { 14107, 0x0000866A }, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+   { 14136, 0x0000866B }, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+   { 14165, 0x0000866C }, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+   { 14194, 0x0000866D }, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+   { 14223, 0x0000866E }, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+   { 14252, 0x0000866F }, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+   { 14281, 0x00008661 }, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+   { 14309, 0x00008662 }, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+   { 14337, 0x00008663 }, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+   { 14365, 0x00008664 }, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+   { 14393, 0x00008665 }, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+   { 14421, 0x00008666 }, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+   { 14449, 0x00008667 }, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+   { 14477, 0x00008668 }, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+   { 14505, 0x00008669 }, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+   { 14533, 0x00000DB0 }, /* GL_MAP2_COLOR_4 */
+   { 14549, 0x00000DD2 }, /* GL_MAP2_GRID_DOMAIN */
+   { 14569, 0x00000DD3 }, /* GL_MAP2_GRID_SEGMENTS */
+   { 14591, 0x00000DB1 }, /* GL_MAP2_INDEX */
+   { 14605, 0x00000DB2 }, /* GL_MAP2_NORMAL */
+   { 14620, 0x00000DB3 }, /* GL_MAP2_TEXTURE_COORD_1 */
+   { 14644, 0x00000DB4 }, /* GL_MAP2_TEXTURE_COORD_2 */
+   { 14668, 0x00000DB5 }, /* GL_MAP2_TEXTURE_COORD_3 */
+   { 14692, 0x00000DB6 }, /* GL_MAP2_TEXTURE_COORD_4 */
+   { 14716, 0x00000DB7 }, /* GL_MAP2_VERTEX_3 */
+   { 14733, 0x00000DB8 }, /* GL_MAP2_VERTEX_4 */
+   { 14750, 0x00008670 }, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+   { 14778, 0x0000867A }, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+   { 14807, 0x0000867B }, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+   { 14836, 0x0000867C }, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+   { 14865, 0x0000867D }, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+   { 14894, 0x0000867E }, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+   { 14923, 0x0000867F }, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+   { 14952, 0x00008671 }, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+   { 14980, 0x00008672 }, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+   { 15008, 0x00008673 }, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+   { 15036, 0x00008674 }, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+   { 15064, 0x00008675 }, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+   { 15092, 0x00008676 }, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+   { 15120, 0x00008677 }, /* GL_MAP2_VERTEX_ATTRIB7_4_NV */
+   { 15148, 0x00008678 }, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+   { 15176, 0x00008679 }, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+   { 15204, 0x00000D10 }, /* GL_MAP_COLOR */
+   { 15217, 0x00000D11 }, /* GL_MAP_STENCIL */
+   { 15232, 0x000088C0 }, /* GL_MATRIX0_ARB */
+   { 15247, 0x00008630 }, /* GL_MATRIX0_NV */
+   { 15261, 0x000088CA }, /* GL_MATRIX10_ARB */
+   { 15277, 0x000088CB }, /* GL_MATRIX11_ARB */
+   { 15293, 0x000088CC }, /* GL_MATRIX12_ARB */
+   { 15309, 0x000088CD }, /* GL_MATRIX13_ARB */
+   { 15325, 0x000088CE }, /* GL_MATRIX14_ARB */
+   { 15341, 0x000088CF }, /* GL_MATRIX15_ARB */
+   { 15357, 0x000088D0 }, /* GL_MATRIX16_ARB */
+   { 15373, 0x000088D1 }, /* GL_MATRIX17_ARB */
+   { 15389, 0x000088D2 }, /* GL_MATRIX18_ARB */
+   { 15405, 0x000088D3 }, /* GL_MATRIX19_ARB */
+   { 15421, 0x000088C1 }, /* GL_MATRIX1_ARB */
+   { 15436, 0x00008631 }, /* GL_MATRIX1_NV */
+   { 15450, 0x000088D4 }, /* GL_MATRIX20_ARB */
+   { 15466, 0x000088D5 }, /* GL_MATRIX21_ARB */
+   { 15482, 0x000088D6 }, /* GL_MATRIX22_ARB */
+   { 15498, 0x000088D7 }, /* GL_MATRIX23_ARB */
+   { 15514, 0x000088D8 }, /* GL_MATRIX24_ARB */
+   { 15530, 0x000088D9 }, /* GL_MATRIX25_ARB */
+   { 15546, 0x000088DA }, /* GL_MATRIX26_ARB */
+   { 15562, 0x000088DB }, /* GL_MATRIX27_ARB */
+   { 15578, 0x000088DC }, /* GL_MATRIX28_ARB */
+   { 15594, 0x000088DD }, /* GL_MATRIX29_ARB */
+   { 15610, 0x000088C2 }, /* GL_MATRIX2_ARB */
+   { 15625, 0x00008632 }, /* GL_MATRIX2_NV */
+   { 15639, 0x000088DE }, /* GL_MATRIX30_ARB */
+   { 15655, 0x000088DF }, /* GL_MATRIX31_ARB */
+   { 15671, 0x000088C3 }, /* GL_MATRIX3_ARB */
+   { 15686, 0x00008633 }, /* GL_MATRIX3_NV */
+   { 15700, 0x000088C4 }, /* GL_MATRIX4_ARB */
+   { 15715, 0x00008634 }, /* GL_MATRIX4_NV */
+   { 15729, 0x000088C5 }, /* GL_MATRIX5_ARB */
+   { 15744, 0x00008635 }, /* GL_MATRIX5_NV */
+   { 15758, 0x000088C6 }, /* GL_MATRIX6_ARB */
+   { 15773, 0x00008636 }, /* GL_MATRIX6_NV */
+   { 15787, 0x000088C7 }, /* GL_MATRIX7_ARB */
+   { 15802, 0x00008637 }, /* GL_MATRIX7_NV */
+   { 15816, 0x000088C8 }, /* GL_MATRIX8_ARB */
+   { 15831, 0x000088C9 }, /* GL_MATRIX9_ARB */
+   { 15846, 0x00008844 }, /* GL_MATRIX_INDEX_ARRAY_ARB */
+   { 15872, 0x00008849 }, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+   { 15906, 0x00008846 }, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+   { 15937, 0x00008848 }, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+   { 15970, 0x00008847 }, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+   { 16001, 0x00000BA0 }, /* GL_MATRIX_MODE */
+   { 16016, 0x00008840 }, /* GL_MATRIX_PALETTE_ARB */
+   { 16038, 0x00008008 }, /* GL_MAX */
+   { 16045, 0x00008073 }, /* GL_MAX_3D_TEXTURE_SIZE */
+   { 16068, 0x000088FF }, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+   { 16100, 0x00000D35 }, /* GL_MAX_ATTRIB_STACK_DEPTH */
+   { 16126, 0x00000D3B }, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+   { 16159, 0x00008177 }, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+   { 16185, 0x00008178 }, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 16219, 0x00000D32 }, /* GL_MAX_CLIP_PLANES */
+   { 16238, 0x00008CDF }, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+   { 16267, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+   { 16299, 0x000080B3 }, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH_SGI */
+   { 16335, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+   { 16371, 0x00008B4D }, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB */
+   { 16411, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT */
+   { 16437, 0x0000801B }, /* GL_MAX_CONVOLUTION_HEIGHT_EXT */
+   { 16467, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH */
+   { 16492, 0x0000801A }, /* GL_MAX_CONVOLUTION_WIDTH_EXT */
+   { 16521, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+   { 16550, 0x0000851C }, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE_ARB */
+   { 16583, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS */
+   { 16603, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ARB */
+   { 16627, 0x00008824 }, /* GL_MAX_DRAW_BUFFERS_ATI */
+   { 16651, 0x000080E9 }, /* GL_MAX_ELEMENTS_INDICES */
+   { 16675, 0x000080E8 }, /* GL_MAX_ELEMENTS_VERTICES */
+   { 16700, 0x00000D30 }, /* GL_MAX_EVAL_ORDER */
+   { 16718, 0x00008008 }, /* GL_MAX_EXT */
+   { 16729, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+   { 16764, 0x00008B49 }, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB */
+   { 16803, 0x00000D31 }, /* GL_MAX_LIGHTS */
+   { 16817, 0x00000B31 }, /* GL_MAX_LIST_NESTING */
+   { 16837, 0x00008841 }, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+   { 16875, 0x00000D36 }, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+   { 16904, 0x00000D37 }, /* GL_MAX_NAME_STACK_DEPTH */
+   { 16928, 0x00008842 }, /* GL_MAX_PALETTE_MATRICES_ARB */
+   { 16956, 0x00000D34 }, /* GL_MAX_PIXEL_MAP_TABLE */
+   { 16979, 0x000088B1 }, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 17016, 0x0000880B }, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 17052, 0x000088AD }, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+   { 17079, 0x000088F5 }, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+   { 17108, 0x000088B5 }, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+   { 17142, 0x000088F4 }, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
+   { 17178, 0x000088F6 }, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+   { 17205, 0x000088A1 }, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+   { 17237, 0x000088B4 }, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+   { 17273, 0x000088F8 }, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+   { 17302, 0x000088F7 }, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+   { 17331, 0x0000862F }, /* GL_MAX_PROGRAM_MATRICES_ARB */
+   { 17359, 0x0000862E }, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+   { 17397, 0x000088B3 }, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 17441, 0x0000880E }, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 17484, 0x000088AF }, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 17518, 0x000088A3 }, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 17557, 0x000088AB }, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 17594, 0x000088A7 }, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 17632, 0x00008810 }, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 17675, 0x0000880F }, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 17718, 0x000088A9 }, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+   { 17748, 0x000088A5 }, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+   { 17779, 0x0000880D }, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 17815, 0x0000880C }, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 17851, 0x00000D38 }, /* GL_MAX_PROJECTION_STACK_DEPTH */
+   { 17881, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+   { 17915, 0x000084F8 }, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_NV */
+   { 17948, 0x000084E8 }, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+   { 17977, 0x00008504 }, /* GL_MAX_SHININESS_NV */
+   { 17997, 0x00008505 }, /* GL_MAX_SPOT_EXPONENT_NV */
+   { 18021, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS */
+   { 18043, 0x00008871 }, /* GL_MAX_TEXTURE_COORDS_ARB */
+   { 18069, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+   { 18096, 0x00008872 }, /* GL_MAX_TEXTURE_IMAGE_UNITS_ARB */
+   { 18127, 0x000084FD }, /* GL_MAX_TEXTURE_LOD_BIAS */
+   { 18151, 0x000084FF }, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 18185, 0x00000D33 }, /* GL_MAX_TEXTURE_SIZE */
+   { 18205, 0x00000D39 }, /* GL_MAX_TEXTURE_STACK_DEPTH */
+   { 18232, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS */
+   { 18253, 0x000084E2 }, /* GL_MAX_TEXTURE_UNITS_ARB */
+   { 18278, 0x0000862F }, /* GL_MAX_TRACK_MATRICES_NV */
+   { 18303, 0x0000862E }, /* GL_MAX_TRACK_MATRIX_STACK_DEPTH_NV */
+   { 18338, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS */
+   { 18360, 0x00008B4B }, /* GL_MAX_VARYING_FLOATS_ARB */
+   { 18386, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS */
+   { 18408, 0x00008869 }, /* GL_MAX_VERTEX_ATTRIBS_ARB */
+   { 18434, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+   { 18468, 0x00008B4C }, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB */
+   { 18506, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+   { 18539, 0x00008B4A }, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS_ARB */
+   { 18576, 0x000086A4 }, /* GL_MAX_VERTEX_UNITS_ARB */
+   { 18600, 0x00000D3A }, /* GL_MAX_VIEWPORT_DIMS */
+   { 18621, 0x00008007 }, /* GL_MIN */
+   { 18628, 0x0000802E }, /* GL_MINMAX */
+   { 18638, 0x0000802E }, /* GL_MINMAX_EXT */
+   { 18652, 0x0000802F }, /* GL_MINMAX_FORMAT */
+   { 18669, 0x0000802F }, /* GL_MINMAX_FORMAT_EXT */
+   { 18690, 0x00008030 }, /* GL_MINMAX_SINK */
+   { 18705, 0x00008030 }, /* GL_MINMAX_SINK_EXT */
+   { 18724, 0x00008007 }, /* GL_MIN_EXT */
+   { 18735, 0x00008370 }, /* GL_MIRRORED_REPEAT */
+   { 18754, 0x00008370 }, /* GL_MIRRORED_REPEAT_ARB */
+   { 18777, 0x00008370 }, /* GL_MIRRORED_REPEAT_IBM */
+   { 18800, 0x00008742 }, /* GL_MIRROR_CLAMP_ATI */
+   { 18820, 0x00008742 }, /* GL_MIRROR_CLAMP_EXT */
+   { 18840, 0x00008912 }, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+   { 18870, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_ATI */
+   { 18898, 0x00008743 }, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+   { 18926, 0x00001700 }, /* GL_MODELVIEW */
+   { 18939, 0x00001700 }, /* GL_MODELVIEW0_ARB */
+   { 18957, 0x0000872A }, /* GL_MODELVIEW10_ARB */
+   { 18976, 0x0000872B }, /* GL_MODELVIEW11_ARB */
+   { 18995, 0x0000872C }, /* GL_MODELVIEW12_ARB */
+   { 19014, 0x0000872D }, /* GL_MODELVIEW13_ARB */
+   { 19033, 0x0000872E }, /* GL_MODELVIEW14_ARB */
+   { 19052, 0x0000872F }, /* GL_MODELVIEW15_ARB */
+   { 19071, 0x00008730 }, /* GL_MODELVIEW16_ARB */
+   { 19090, 0x00008731 }, /* GL_MODELVIEW17_ARB */
+   { 19109, 0x00008732 }, /* GL_MODELVIEW18_ARB */
+   { 19128, 0x00008733 }, /* GL_MODELVIEW19_ARB */
+   { 19147, 0x0000850A }, /* GL_MODELVIEW1_ARB */
+   { 19165, 0x00008734 }, /* GL_MODELVIEW20_ARB */
+   { 19184, 0x00008735 }, /* GL_MODELVIEW21_ARB */
+   { 19203, 0x00008736 }, /* GL_MODELVIEW22_ARB */
+   { 19222, 0x00008737 }, /* GL_MODELVIEW23_ARB */
+   { 19241, 0x00008738 }, /* GL_MODELVIEW24_ARB */
+   { 19260, 0x00008739 }, /* GL_MODELVIEW25_ARB */
+   { 19279, 0x0000873A }, /* GL_MODELVIEW26_ARB */
+   { 19298, 0x0000873B }, /* GL_MODELVIEW27_ARB */
+   { 19317, 0x0000873C }, /* GL_MODELVIEW28_ARB */
+   { 19336, 0x0000873D }, /* GL_MODELVIEW29_ARB */
+   { 19355, 0x00008722 }, /* GL_MODELVIEW2_ARB */
+   { 19373, 0x0000873E }, /* GL_MODELVIEW30_ARB */
+   { 19392, 0x0000873F }, /* GL_MODELVIEW31_ARB */
+   { 19411, 0x00008723 }, /* GL_MODELVIEW3_ARB */
+   { 19429, 0x00008724 }, /* GL_MODELVIEW4_ARB */
+   { 19447, 0x00008725 }, /* GL_MODELVIEW5_ARB */
+   { 19465, 0x00008726 }, /* GL_MODELVIEW6_ARB */
+   { 19483, 0x00008727 }, /* GL_MODELVIEW7_ARB */
+   { 19501, 0x00008728 }, /* GL_MODELVIEW8_ARB */
+   { 19519, 0x00008729 }, /* GL_MODELVIEW9_ARB */
+   { 19537, 0x00000BA6 }, /* GL_MODELVIEW_MATRIX */
+   { 19557, 0x00008629 }, /* GL_MODELVIEW_PROJECTION_NV */
+   { 19584, 0x00000BA3 }, /* GL_MODELVIEW_STACK_DEPTH */
+   { 19609, 0x00002100 }, /* GL_MODULATE */
+   { 19621, 0x00008744 }, /* GL_MODULATE_ADD_ATI */
+   { 19641, 0x00008745 }, /* GL_MODULATE_SIGNED_ADD_ATI */
+   { 19668, 0x00008746 }, /* GL_MODULATE_SUBTRACT_ATI */
+   { 19693, 0x00000103 }, /* GL_MULT */
+   { 19701, 0x0000809D }, /* GL_MULTISAMPLE */
+   { 19716, 0x000086B2 }, /* GL_MULTISAMPLE_3DFX */
+   { 19736, 0x0000809D }, /* GL_MULTISAMPLE_ARB */
+   { 19755, 0x20000000 }, /* GL_MULTISAMPLE_BIT */
+   { 19774, 0x20000000 }, /* GL_MULTISAMPLE_BIT_3DFX */
+   { 19798, 0x20000000 }, /* GL_MULTISAMPLE_BIT_ARB */
+   { 19821, 0x00008534 }, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+   { 19851, 0x00002A25 }, /* GL_N3F_V3F */
+   { 19862, 0x00000D70 }, /* GL_NAME_STACK_DEPTH */
+   { 19882, 0x0000150E }, /* GL_NAND */
+   { 19890, 0x00002600 }, /* GL_NEAREST */
+   { 19901, 0x0000844E }, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+   { 19932, 0x0000844D }, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+   { 19964, 0x00002702 }, /* GL_NEAREST_MIPMAP_LINEAR */
+   { 19989, 0x00002700 }, /* GL_NEAREST_MIPMAP_NEAREST */
+   { 20015, 0x00000200 }, /* GL_NEVER */
+   { 20024, 0x00001102 }, /* GL_NICEST */
+   { 20034, 0x00000000 }, /* GL_NONE */
+   { 20042, 0x00001505 }, /* GL_NOOP */
+   { 20050, 0x00001508 }, /* GL_NOR */
+   { 20057, 0x00000BA1 }, /* GL_NORMALIZE */
+   { 20070, 0x00008075 }, /* GL_NORMAL_ARRAY */
+   { 20086, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+   { 20117, 0x00008897 }, /* GL_NORMAL_ARRAY_BUFFER_BINDING_ARB */
+   { 20152, 0x0000808F }, /* GL_NORMAL_ARRAY_POINTER */
+   { 20176, 0x0000807F }, /* GL_NORMAL_ARRAY_STRIDE */
+   { 20199, 0x0000807E }, /* GL_NORMAL_ARRAY_TYPE */
+   { 20220, 0x00008511 }, /* GL_NORMAL_MAP */
+   { 20234, 0x00008511 }, /* GL_NORMAL_MAP_ARB */
+   { 20252, 0x00008511 }, /* GL_NORMAL_MAP_NV */
+   { 20269, 0x00000205 }, /* GL_NOTEQUAL */
+   { 20281, 0x00000000 }, /* GL_NO_ERROR */
+   { 20293, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+   { 20327, 0x000086A2 }, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB */
+   { 20365, 0x00008B89 }, /* GL_OBJECT_ACTIVE_ATTRIBUTES_ARB */
+   { 20397, 0x00008B8A }, /* GL_OBJECT_ACTIVE_ATTRIBUTE_MAX_LENGTH_ARB */
+   { 20439, 0x00008B86 }, /* GL_OBJECT_ACTIVE_UNIFORMS_ARB */
+   { 20469, 0x00008B87 }, /* GL_OBJECT_ACTIVE_UNIFORM_MAX_LENGTH_ARB */
+   { 20509, 0x00008B85 }, /* GL_OBJECT_ATTACHED_OBJECTS_ARB */
+   { 20540, 0x00008B81 }, /* GL_OBJECT_COMPILE_STATUS_ARB */
+   { 20569, 0x00008B80 }, /* GL_OBJECT_DELETE_STATUS_ARB */
+   { 20597, 0x00008B84 }, /* GL_OBJECT_INFO_LOG_LENGTH_ARB */
+   { 20627, 0x00002401 }, /* GL_OBJECT_LINEAR */
+   { 20644, 0x00008B82 }, /* GL_OBJECT_LINK_STATUS_ARB */
+   { 20670, 0x00002501 }, /* GL_OBJECT_PLANE */
+   { 20686, 0x00008B88 }, /* GL_OBJECT_SHADER_SOURCE_LENGTH_ARB */
+   { 20721, 0x00008B4F }, /* GL_OBJECT_SUBTYPE_ARB */
+   { 20743, 0x00008B4E }, /* GL_OBJECT_TYPE_ARB */
+   { 20762, 0x00008B83 }, /* GL_OBJECT_VALIDATE_STATUS_ARB */
+   { 20792, 0x00008165 }, /* GL_OCCLUSION_TEST_HP */
+   { 20813, 0x00008166 }, /* GL_OCCLUSION_TEST_RESULT_HP */
+   { 20841, 0x00000001 }, /* GL_ONE */
+   { 20848, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+   { 20876, 0x00008004 }, /* GL_ONE_MINUS_CONSTANT_ALPHA_EXT */
+   { 20908, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR */
+   { 20936, 0x00008002 }, /* GL_ONE_MINUS_CONSTANT_COLOR_EXT */
+   { 20968, 0x00000305 }, /* GL_ONE_MINUS_DST_ALPHA */
+   { 20991, 0x00000307 }, /* GL_ONE_MINUS_DST_COLOR */
+   { 21014, 0x00000303 }, /* GL_ONE_MINUS_SRC_ALPHA */
+   { 21037, 0x00000301 }, /* GL_ONE_MINUS_SRC_COLOR */
+   { 21060, 0x00008598 }, /* GL_OPERAND0_ALPHA */
+   { 21078, 0x00008598 }, /* GL_OPERAND0_ALPHA_ARB */
+   { 21100, 0x00008598 }, /* GL_OPERAND0_ALPHA_EXT */
+   { 21122, 0x00008590 }, /* GL_OPERAND0_RGB */
+   { 21138, 0x00008590 }, /* GL_OPERAND0_RGB_ARB */
+   { 21158, 0x00008590 }, /* GL_OPERAND0_RGB_EXT */
+   { 21178, 0x00008599 }, /* GL_OPERAND1_ALPHA */
+   { 21196, 0x00008599 }, /* GL_OPERAND1_ALPHA_ARB */
+   { 21218, 0x00008599 }, /* GL_OPERAND1_ALPHA_EXT */
+   { 21240, 0x00008591 }, /* GL_OPERAND1_RGB */
+   { 21256, 0x00008591 }, /* GL_OPERAND1_RGB_ARB */
+   { 21276, 0x00008591 }, /* GL_OPERAND1_RGB_EXT */
+   { 21296, 0x0000859A }, /* GL_OPERAND2_ALPHA */
+   { 21314, 0x0000859A }, /* GL_OPERAND2_ALPHA_ARB */
+   { 21336, 0x0000859A }, /* GL_OPERAND2_ALPHA_EXT */
+   { 21358, 0x00008592 }, /* GL_OPERAND2_RGB */
+   { 21374, 0x00008592 }, /* GL_OPERAND2_RGB_ARB */
+   { 21394, 0x00008592 }, /* GL_OPERAND2_RGB_EXT */
+   { 21414, 0x0000859B }, /* GL_OPERAND3_ALPHA_NV */
+   { 21435, 0x00008593 }, /* GL_OPERAND3_RGB_NV */
+   { 21454, 0x00001507 }, /* GL_OR */
+   { 21460, 0x00000A01 }, /* GL_ORDER */
+   { 21469, 0x0000150D }, /* GL_OR_INVERTED */
+   { 21484, 0x0000150B }, /* GL_OR_REVERSE */
+   { 21498, 0x00000505 }, /* GL_OUT_OF_MEMORY */
+   { 21515, 0x00000D05 }, /* GL_PACK_ALIGNMENT */
+   { 21533, 0x0000806C }, /* GL_PACK_IMAGE_HEIGHT */
+   { 21554, 0x00008758 }, /* GL_PACK_INVERT_MESA */
+   { 21574, 0x00000D01 }, /* GL_PACK_LSB_FIRST */
+   { 21592, 0x00000D02 }, /* GL_PACK_ROW_LENGTH */
+   { 21611, 0x0000806B }, /* GL_PACK_SKIP_IMAGES */
+   { 21631, 0x00000D04 }, /* GL_PACK_SKIP_PIXELS */
+   { 21651, 0x00000D03 }, /* GL_PACK_SKIP_ROWS */
+   { 21669, 0x00000D00 }, /* GL_PACK_SWAP_BYTES */
+   { 21688, 0x00008B92 }, /* GL_PALETTE4_R5_G6_B5_OES */
+   { 21713, 0x00008B94 }, /* GL_PALETTE4_RGB5_A1_OES */
+   { 21737, 0x00008B90 }, /* GL_PALETTE4_RGB8_OES */
+   { 21758, 0x00008B93 }, /* GL_PALETTE4_RGBA4_OES */
+   { 21780, 0x00008B91 }, /* GL_PALETTE4_RGBA8_OES */
+   { 21802, 0x00008B97 }, /* GL_PALETTE8_R5_G6_B5_OES */
+   { 21827, 0x00008B99 }, /* GL_PALETTE8_RGB5_A1_OES */
+   { 21851, 0x00008B95 }, /* GL_PALETTE8_RGB8_OES */
+   { 21872, 0x00008B98 }, /* GL_PALETTE8_RGBA4_OES */
+   { 21894, 0x00008B96 }, /* GL_PALETTE8_RGBA8_OES */
+   { 21916, 0x00000700 }, /* GL_PASS_THROUGH_TOKEN */
+   { 21938, 0x00000C50 }, /* GL_PERSPECTIVE_CORRECTION_HINT */
+   { 21969, 0x00000C79 }, /* GL_PIXEL_MAP_A_TO_A */
+   { 21989, 0x00000CB9 }, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+   { 22014, 0x00000C78 }, /* GL_PIXEL_MAP_B_TO_B */
+   { 22034, 0x00000CB8 }, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+   { 22059, 0x00000C77 }, /* GL_PIXEL_MAP_G_TO_G */
+   { 22079, 0x00000CB7 }, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+   { 22104, 0x00000C75 }, /* GL_PIXEL_MAP_I_TO_A */
+   { 22124, 0x00000CB5 }, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+   { 22149, 0x00000C74 }, /* GL_PIXEL_MAP_I_TO_B */
+   { 22169, 0x00000CB4 }, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+   { 22194, 0x00000C73 }, /* GL_PIXEL_MAP_I_TO_G */
+   { 22214, 0x00000CB3 }, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+   { 22239, 0x00000C70 }, /* GL_PIXEL_MAP_I_TO_I */
+   { 22259, 0x00000CB0 }, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+   { 22284, 0x00000C72 }, /* GL_PIXEL_MAP_I_TO_R */
+   { 22304, 0x00000CB2 }, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+   { 22329, 0x00000C76 }, /* GL_PIXEL_MAP_R_TO_R */
+   { 22349, 0x00000CB6 }, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+   { 22374, 0x00000C71 }, /* GL_PIXEL_MAP_S_TO_S */
+   { 22394, 0x00000CB1 }, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+   { 22419, 0x00000020 }, /* GL_PIXEL_MODE_BIT */
+   { 22437, 0x000088ED }, /* GL_PIXEL_PACK_BUFFER_BINDING_EXT */
+   { 22470, 0x000088EB }, /* GL_PIXEL_PACK_BUFFER_EXT */
+   { 22495, 0x000088EF }, /* GL_PIXEL_UNPACK_BUFFER_BINDING_EXT */
+   { 22530, 0x000088EC }, /* GL_PIXEL_UNPACK_BUFFER_EXT */
+   { 22557, 0x00001B00 }, /* GL_POINT */
+   { 22566, 0x00000000 }, /* GL_POINTS */
+   { 22576, 0x00000002 }, /* GL_POINT_BIT */
+   { 22589, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION */
+   { 22619, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_ARB */
+   { 22653, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_EXT */
+   { 22687, 0x00008129 }, /* GL_POINT_DISTANCE_ATTENUATION_SGIS */
+   { 22722, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE */
+   { 22751, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_ARB */
+   { 22784, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_EXT */
+   { 22817, 0x00008128 }, /* GL_POINT_FADE_THRESHOLD_SIZE_SGIS */
+   { 22851, 0x00000B11 }, /* GL_POINT_SIZE */
+   { 22865, 0x00000B13 }, /* GL_POINT_SIZE_GRANULARITY */
+   { 22891, 0x00008127 }, /* GL_POINT_SIZE_MAX */
+   { 22909, 0x00008127 }, /* GL_POINT_SIZE_MAX_ARB */
+   { 22931, 0x00008127 }, /* GL_POINT_SIZE_MAX_EXT */
+   { 22953, 0x00008127 }, /* GL_POINT_SIZE_MAX_SGIS */
+   { 22976, 0x00008126 }, /* GL_POINT_SIZE_MIN */
+   { 22994, 0x00008126 }, /* GL_POINT_SIZE_MIN_ARB */
+   { 23016, 0x00008126 }, /* GL_POINT_SIZE_MIN_EXT */
+   { 23038, 0x00008126 }, /* GL_POINT_SIZE_MIN_SGIS */
+   { 23061, 0x00000B12 }, /* GL_POINT_SIZE_RANGE */
+   { 23081, 0x00000B10 }, /* GL_POINT_SMOOTH */
+   { 23097, 0x00000C51 }, /* GL_POINT_SMOOTH_HINT */
+   { 23118, 0x00008861 }, /* GL_POINT_SPRITE */
+   { 23134, 0x00008861 }, /* GL_POINT_SPRITE_ARB */
+   { 23154, 0x00008CA0 }, /* GL_POINT_SPRITE_COORD_ORIGIN */
+   { 23183, 0x00008861 }, /* GL_POINT_SPRITE_NV */
+   { 23202, 0x00008863 }, /* GL_POINT_SPRITE_R_MODE_NV */
+   { 23228, 0x00000701 }, /* GL_POINT_TOKEN */
+   { 23243, 0x00000009 }, /* GL_POLYGON */
+   { 23254, 0x00000008 }, /* GL_POLYGON_BIT */
+   { 23269, 0x00000B40 }, /* GL_POLYGON_MODE */
+   { 23285, 0x00008039 }, /* GL_POLYGON_OFFSET_BIAS */
+   { 23308, 0x00008038 }, /* GL_POLYGON_OFFSET_FACTOR */
+   { 23333, 0x00008037 }, /* GL_POLYGON_OFFSET_FILL */
+   { 23356, 0x00002A02 }, /* GL_POLYGON_OFFSET_LINE */
+   { 23379, 0x00002A01 }, /* GL_POLYGON_OFFSET_POINT */
+   { 23403, 0x00002A00 }, /* GL_POLYGON_OFFSET_UNITS */
+   { 23427, 0x00000B41 }, /* GL_POLYGON_SMOOTH */
+   { 23445, 0x00000C53 }, /* GL_POLYGON_SMOOTH_HINT */
+   { 23468, 0x00000B42 }, /* GL_POLYGON_STIPPLE */
+   { 23487, 0x00000010 }, /* GL_POLYGON_STIPPLE_BIT */
+   { 23510, 0x00000703 }, /* GL_POLYGON_TOKEN */
+   { 23527, 0x00001203 }, /* GL_POSITION */
+   { 23539, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+   { 23571, 0x000080BB }, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS_SGI */
+   { 23607, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+   { 23640, 0x000080B7 }, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE_SGI */
+   { 23677, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+   { 23708, 0x000080BA }, /* GL_POST_COLOR_MATRIX_BLUE_BIAS_SGI */
+   { 23743, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+   { 23775, 0x000080B6 }, /* GL_POST_COLOR_MATRIX_BLUE_SCALE_SGI */
+   { 23811, 0x000080D2 }, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 23844, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+   { 23876, 0x000080B9 }, /* GL_POST_COLOR_MATRIX_GREEN_BIAS_SGI */
+   { 23912, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+   { 23945, 0x000080B5 }, /* GL_POST_COLOR_MATRIX_GREEN_SCALE_SGI */
+   { 23982, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+   { 24012, 0x000080B8 }, /* GL_POST_COLOR_MATRIX_RED_BIAS_SGI */
+   { 24046, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+   { 24077, 0x000080B4 }, /* GL_POST_COLOR_MATRIX_RED_SCALE_SGI */
+   { 24112, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+   { 24143, 0x00008023 }, /* GL_POST_CONVOLUTION_ALPHA_BIAS_EXT */
+   { 24178, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+   { 24210, 0x0000801F }, /* GL_POST_CONVOLUTION_ALPHA_SCALE_EXT */
+   { 24246, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+   { 24276, 0x00008022 }, /* GL_POST_CONVOLUTION_BLUE_BIAS_EXT */
+   { 24310, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+   { 24341, 0x0000801E }, /* GL_POST_CONVOLUTION_BLUE_SCALE_EXT */
+   { 24376, 0x000080D1 }, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+   { 24408, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+   { 24439, 0x00008021 }, /* GL_POST_CONVOLUTION_GREEN_BIAS_EXT */
+   { 24474, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+   { 24506, 0x0000801D }, /* GL_POST_CONVOLUTION_GREEN_SCALE_EXT */
+   { 24542, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS */
+   { 24571, 0x00008020 }, /* GL_POST_CONVOLUTION_RED_BIAS_EXT */
+   { 24604, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE */
+   { 24634, 0x0000801C }, /* GL_POST_CONVOLUTION_RED_SCALE_EXT */
+   { 24668, 0x0000817B }, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+   { 24707, 0x00008179 }, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+   { 24740, 0x0000817C }, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+   { 24780, 0x0000817A }, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+   { 24814, 0x00008578 }, /* GL_PREVIOUS */
+   { 24826, 0x00008578 }, /* GL_PREVIOUS_ARB */
+   { 24842, 0x00008578 }, /* GL_PREVIOUS_EXT */
+   { 24858, 0x00008577 }, /* GL_PRIMARY_COLOR */
+   { 24875, 0x00008577 }, /* GL_PRIMARY_COLOR_ARB */
+   { 24896, 0x00008577 }, /* GL_PRIMARY_COLOR_EXT */
+   { 24917, 0x000088B0 }, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+   { 24950, 0x00008805 }, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+   { 24982, 0x000088AC }, /* GL_PROGRAM_ATTRIBS_ARB */
+   { 25005, 0x00008677 }, /* GL_PROGRAM_BINDING_ARB */
+   { 25028, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_ARB */
+   { 25058, 0x0000864B }, /* GL_PROGRAM_ERROR_POSITION_NV */
+   { 25087, 0x00008874 }, /* GL_PROGRAM_ERROR_STRING_ARB */
+   { 25115, 0x00008876 }, /* GL_PROGRAM_FORMAT_ARB */
+   { 25137, 0x00008875 }, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+   { 25165, 0x000088A0 }, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+   { 25193, 0x00008627 }, /* GL_PROGRAM_LENGTH_ARB */
+   { 25215, 0x00008627 }, /* GL_PROGRAM_LENGTH_NV */
+   { 25236, 0x000088B2 }, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+   { 25276, 0x00008808 }, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+   { 25315, 0x000088AE }, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+   { 25345, 0x000088A2 }, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+   { 25380, 0x000088AA }, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+   { 25413, 0x000088A6 }, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+   { 25447, 0x0000880A }, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+   { 25486, 0x00008809 }, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+   { 25525, 0x00008B40 }, /* GL_PROGRAM_OBJECT_ARB */
+   { 25547, 0x000088A8 }, /* GL_PROGRAM_PARAMETERS_ARB */
+   { 25573, 0x00008644 }, /* GL_PROGRAM_PARAMETER_NV */
+   { 25597, 0x00008647 }, /* GL_PROGRAM_RESIDENT_NV */
+   { 25620, 0x00008628 }, /* GL_PROGRAM_STRING_ARB */
+   { 25642, 0x00008628 }, /* GL_PROGRAM_STRING_NV */
+   { 25663, 0x00008646 }, /* GL_PROGRAM_TARGET_NV */
+   { 25684, 0x000088A4 }, /* GL_PROGRAM_TEMPORARIES_ARB */
+   { 25711, 0x00008807 }, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+   { 25743, 0x00008806 }, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+   { 25775, 0x000088B6 }, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+   { 25810, 0x00001701 }, /* GL_PROJECTION */
+   { 25824, 0x00000BA7 }, /* GL_PROJECTION_MATRIX */
+   { 25845, 0x00000BA4 }, /* GL_PROJECTION_STACK_DEPTH */
+   { 25871, 0x000080D3 }, /* GL_PROXY_COLOR_TABLE */
+   { 25892, 0x00008025 }, /* GL_PROXY_HISTOGRAM */
+   { 25911, 0x00008025 }, /* GL_PROXY_HISTOGRAM_EXT */
+   { 25934, 0x000080D5 }, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+   { 25973, 0x000080D4 }, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+   { 26011, 0x00008063 }, /* GL_PROXY_TEXTURE_1D */
+   { 26031, 0x00008C19 }, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+   { 26061, 0x00008063 }, /* GL_PROXY_TEXTURE_1D_EXT */
+   { 26085, 0x00008064 }, /* GL_PROXY_TEXTURE_2D */
+   { 26105, 0x00008C1B }, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+   { 26135, 0x00008064 }, /* GL_PROXY_TEXTURE_2D_EXT */
+   { 26159, 0x00008070 }, /* GL_PROXY_TEXTURE_3D */
+   { 26179, 0x000080BD }, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+   { 26212, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP */
+   { 26238, 0x0000851B }, /* GL_PROXY_TEXTURE_CUBE_MAP_ARB */
+   { 26268, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+   { 26299, 0x000084F7 }, /* GL_PROXY_TEXTURE_RECTANGLE_NV */
+   { 26329, 0x00002003 }, /* GL_Q */
+   { 26334, 0x00001209 }, /* GL_QUADRATIC_ATTENUATION */
+   { 26359, 0x00000007 }, /* GL_QUADS */
+   { 26368, 0x00008614 }, /* GL_QUAD_MESH_SUN */
+   { 26385, 0x00000008 }, /* GL_QUAD_STRIP */
+   { 26399, 0x00008864 }, /* GL_QUERY_COUNTER_BITS */
+   { 26421, 0x00008864 }, /* GL_QUERY_COUNTER_BITS_ARB */
+   { 26447, 0x00008866 }, /* GL_QUERY_RESULT */
+   { 26463, 0x00008866 }, /* GL_QUERY_RESULT_ARB */
+   { 26483, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE */
+   { 26509, 0x00008867 }, /* GL_QUERY_RESULT_AVAILABLE_ARB */
+   { 26539, 0x00002002 }, /* GL_R */
+   { 26544, 0x00002A10 }, /* GL_R3_G3_B2 */
+   { 26556, 0x00019262 }, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+   { 26589, 0x00000C02 }, /* GL_READ_BUFFER */
+   { 26604, 0x00008CAA }, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
+   { 26636, 0x00008CA8 }, /* GL_READ_FRAMEBUFFER_EXT */
+   { 26660, 0x000088B8 }, /* GL_READ_ONLY */
+   { 26673, 0x000088B8 }, /* GL_READ_ONLY_ARB */
+   { 26690, 0x000088BA }, /* GL_READ_WRITE */
+   { 26704, 0x000088BA }, /* GL_READ_WRITE_ARB */
+   { 26722, 0x00001903 }, /* GL_RED */
+   { 26729, 0x00008016 }, /* GL_REDUCE */
+   { 26739, 0x00008016 }, /* GL_REDUCE_EXT */
+   { 26753, 0x00000D15 }, /* GL_RED_BIAS */
+   { 26765, 0x00000D52 }, /* GL_RED_BITS */
+   { 26777, 0x00000D14 }, /* GL_RED_SCALE */
+   { 26790, 0x00008512 }, /* GL_REFLECTION_MAP */
+   { 26808, 0x00008512 }, /* GL_REFLECTION_MAP_ARB */
+   { 26830, 0x00008512 }, /* GL_REFLECTION_MAP_NV */
+   { 26851, 0x00001C00 }, /* GL_RENDER */
+   { 26861, 0x00008CA7 }, /* GL_RENDERBUFFER_BINDING_EXT */
+   { 26889, 0x00008D41 }, /* GL_RENDERBUFFER_EXT */
+   { 26909, 0x00008D43 }, /* GL_RENDERBUFFER_HEIGHT_EXT */
+   { 26936, 0x00008D44 }, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
+   { 26972, 0x00008D42 }, /* GL_RENDERBUFFER_WIDTH_EXT */
+   { 26998, 0x00001F01 }, /* GL_RENDERER */
+   { 27010, 0x00000C40 }, /* GL_RENDER_MODE */
+   { 27025, 0x00002901 }, /* GL_REPEAT */
+   { 27035, 0x00001E01 }, /* GL_REPLACE */
+   { 27046, 0x00008062 }, /* GL_REPLACE_EXT */
+   { 27061, 0x00008153 }, /* GL_REPLICATE_BORDER_HP */
+   { 27084, 0x0000803A }, /* GL_RESCALE_NORMAL */
+   { 27102, 0x0000803A }, /* GL_RESCALE_NORMAL_EXT */
+   { 27124, 0x00000102 }, /* GL_RETURN */
+   { 27134, 0x00001907 }, /* GL_RGB */
+   { 27141, 0x00008052 }, /* GL_RGB10 */
+   { 27150, 0x00008059 }, /* GL_RGB10_A2 */
+   { 27162, 0x00008059 }, /* GL_RGB10_A2_EXT */
+   { 27178, 0x00008052 }, /* GL_RGB10_EXT */
+   { 27191, 0x00008053 }, /* GL_RGB12 */
+   { 27200, 0x00008053 }, /* GL_RGB12_EXT */
+   { 27213, 0x00008054 }, /* GL_RGB16 */
+   { 27222, 0x00008054 }, /* GL_RGB16_EXT */
+   { 27235, 0x0000804E }, /* GL_RGB2_EXT */
+   { 27247, 0x0000804F }, /* GL_RGB4 */
+   { 27255, 0x0000804F }, /* GL_RGB4_EXT */
+   { 27267, 0x000083A1 }, /* GL_RGB4_S3TC */
+   { 27280, 0x00008050 }, /* GL_RGB5 */
+   { 27288, 0x00008057 }, /* GL_RGB5_A1 */
+   { 27299, 0x00008057 }, /* GL_RGB5_A1_EXT */
+   { 27314, 0x00008050 }, /* GL_RGB5_EXT */
+   { 27326, 0x00008051 }, /* GL_RGB8 */
+   { 27334, 0x00008051 }, /* GL_RGB8_EXT */
+   { 27346, 0x00001908 }, /* GL_RGBA */
+   { 27354, 0x0000805A }, /* GL_RGBA12 */
+   { 27364, 0x0000805A }, /* GL_RGBA12_EXT */
+   { 27378, 0x0000805B }, /* GL_RGBA16 */
+   { 27388, 0x0000805B }, /* GL_RGBA16_EXT */
+   { 27402, 0x00008055 }, /* GL_RGBA2 */
+   { 27411, 0x00008055 }, /* GL_RGBA2_EXT */
+   { 27424, 0x00008056 }, /* GL_RGBA4 */
+   { 27433, 0x000083A5 }, /* GL_RGBA4_DXT5_S3TC */
+   { 27452, 0x00008056 }, /* GL_RGBA4_EXT */
+   { 27465, 0x000083A3 }, /* GL_RGBA4_S3TC */
+   { 27479, 0x00008058 }, /* GL_RGBA8 */
+   { 27488, 0x00008058 }, /* GL_RGBA8_EXT */
+   { 27501, 0x000083A4 }, /* GL_RGBA_DXT5_S3TC */
+   { 27519, 0x00000C31 }, /* GL_RGBA_MODE */
+   { 27532, 0x000083A2 }, /* GL_RGBA_S3TC */
+   { 27545, 0x000083A0 }, /* GL_RGB_S3TC */
+   { 27557, 0x00008573 }, /* GL_RGB_SCALE */
+   { 27570, 0x00008573 }, /* GL_RGB_SCALE_ARB */
+   { 27587, 0x00008573 }, /* GL_RGB_SCALE_EXT */
+   { 27604, 0x00000407 }, /* GL_RIGHT */
+   { 27613, 0x00002000 }, /* GL_S */
+   { 27618, 0x00008B5D }, /* GL_SAMPLER_1D */
+   { 27632, 0x00008B61 }, /* GL_SAMPLER_1D_SHADOW */
+   { 27653, 0x00008B5E }, /* GL_SAMPLER_2D */
+   { 27667, 0x00008B62 }, /* GL_SAMPLER_2D_SHADOW */
+   { 27688, 0x00008B5F }, /* GL_SAMPLER_3D */
+   { 27702, 0x00008B60 }, /* GL_SAMPLER_CUBE */
+   { 27718, 0x000080A9 }, /* GL_SAMPLES */
+   { 27729, 0x000086B4 }, /* GL_SAMPLES_3DFX */
+   { 27745, 0x000080A9 }, /* GL_SAMPLES_ARB */
+   { 27760, 0x00008914 }, /* GL_SAMPLES_PASSED */
+   { 27778, 0x00008914 }, /* GL_SAMPLES_PASSED_ARB */
+   { 27800, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+   { 27828, 0x0000809E }, /* GL_SAMPLE_ALPHA_TO_COVERAGE_ARB */
+   { 27860, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE */
+   { 27883, 0x0000809F }, /* GL_SAMPLE_ALPHA_TO_ONE_ARB */
+   { 27910, 0x000080A8 }, /* GL_SAMPLE_BUFFERS */
+   { 27928, 0x000086B3 }, /* GL_SAMPLE_BUFFERS_3DFX */
+   { 27951, 0x000080A8 }, /* GL_SAMPLE_BUFFERS_ARB */
+   { 27973, 0x000080A0 }, /* GL_SAMPLE_COVERAGE */
+   { 27992, 0x000080A0 }, /* GL_SAMPLE_COVERAGE_ARB */
+   { 28015, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT */
+   { 28041, 0x000080AB }, /* GL_SAMPLE_COVERAGE_INVERT_ARB */
+   { 28071, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE */
+   { 28096, 0x000080AA }, /* GL_SAMPLE_COVERAGE_VALUE_ARB */
+   { 28125, 0x00080000 }, /* GL_SCISSOR_BIT */
+   { 28140, 0x00000C10 }, /* GL_SCISSOR_BOX */
+   { 28155, 0x00000C11 }, /* GL_SCISSOR_TEST */
+   { 28171, 0x0000845E }, /* GL_SECONDARY_COLOR_ARRAY */
+   { 28196, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+   { 28236, 0x0000889C }, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING_ARB */
+   { 28280, 0x0000845D }, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+   { 28313, 0x0000845A }, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+   { 28343, 0x0000845C }, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+   { 28375, 0x0000845B }, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+   { 28405, 0x00001C02 }, /* GL_SELECT */
+   { 28415, 0x00000DF3 }, /* GL_SELECTION_BUFFER_POINTER */
+   { 28443, 0x00000DF4 }, /* GL_SELECTION_BUFFER_SIZE */
+   { 28468, 0x00008012 }, /* GL_SEPARABLE_2D */
+   { 28484, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR */
+   { 28511, 0x000081FA }, /* GL_SEPARATE_SPECULAR_COLOR_EXT */
+   { 28542, 0x0000150F }, /* GL_SET */
+   { 28549, 0x00008B48 }, /* GL_SHADER_OBJECT_ARB */
+   { 28570, 0x00008B88 }, /* GL_SHADER_SOURCE_LENGTH */
+   { 28594, 0x00008B4F }, /* GL_SHADER_TYPE */
+   { 28609, 0x00000B54 }, /* GL_SHADE_MODEL */
+   { 28624, 0x00008B8C }, /* GL_SHADING_LANGUAGE_VERSION */
+   { 28652, 0x000080BF }, /* GL_SHADOW_AMBIENT_SGIX */
+   { 28675, 0x000081FB }, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+   { 28705, 0x00001601 }, /* GL_SHININESS */
+   { 28718, 0x00001402 }, /* GL_SHORT */
+   { 28727, 0x000081F9 }, /* GL_SINGLE_COLOR */
+   { 28743, 0x000081F9 }, /* GL_SINGLE_COLOR_EXT */
+   { 28763, 0x000085CC }, /* GL_SLICE_ACCUM_SUN */
+   { 28782, 0x00001D01 }, /* GL_SMOOTH */
+   { 28792, 0x00000B23 }, /* GL_SMOOTH_LINE_WIDTH_GRANULARITY */
+   { 28825, 0x00000B22 }, /* GL_SMOOTH_LINE_WIDTH_RANGE */
+   { 28852, 0x00000B13 }, /* GL_SMOOTH_POINT_SIZE_GRANULARITY */
+   { 28885, 0x00000B12 }, /* GL_SMOOTH_POINT_SIZE_RANGE */
+   { 28912, 0x00008588 }, /* GL_SOURCE0_ALPHA */
+   { 28929, 0x00008588 }, /* GL_SOURCE0_ALPHA_ARB */
+   { 28950, 0x00008588 }, /* GL_SOURCE0_ALPHA_EXT */
+   { 28971, 0x00008580 }, /* GL_SOURCE0_RGB */
+   { 28986, 0x00008580 }, /* GL_SOURCE0_RGB_ARB */
+   { 29005, 0x00008580 }, /* GL_SOURCE0_RGB_EXT */
+   { 29024, 0x00008589 }, /* GL_SOURCE1_ALPHA */
+   { 29041, 0x00008589 }, /* GL_SOURCE1_ALPHA_ARB */
+   { 29062, 0x00008589 }, /* GL_SOURCE1_ALPHA_EXT */
+   { 29083, 0x00008581 }, /* GL_SOURCE1_RGB */
+   { 29098, 0x00008581 }, /* GL_SOURCE1_RGB_ARB */
+   { 29117, 0x00008581 }, /* GL_SOURCE1_RGB_EXT */
+   { 29136, 0x0000858A }, /* GL_SOURCE2_ALPHA */
+   { 29153, 0x0000858A }, /* GL_SOURCE2_ALPHA_ARB */
+   { 29174, 0x0000858A }, /* GL_SOURCE2_ALPHA_EXT */
+   { 29195, 0x00008582 }, /* GL_SOURCE2_RGB */
+   { 29210, 0x00008582 }, /* GL_SOURCE2_RGB_ARB */
+   { 29229, 0x00008582 }, /* GL_SOURCE2_RGB_EXT */
+   { 29248, 0x0000858B }, /* GL_SOURCE3_ALPHA_NV */
+   { 29268, 0x00008583 }, /* GL_SOURCE3_RGB_NV */
+   { 29286, 0x00001202 }, /* GL_SPECULAR */
+   { 29298, 0x00002402 }, /* GL_SPHERE_MAP */
+   { 29312, 0x00001206 }, /* GL_SPOT_CUTOFF */
+   { 29327, 0x00001204 }, /* GL_SPOT_DIRECTION */
+   { 29345, 0x00001205 }, /* GL_SPOT_EXPONENT */
+   { 29362, 0x00008588 }, /* GL_SRC0_ALPHA */
+   { 29376, 0x00008580 }, /* GL_SRC0_RGB */
+   { 29388, 0x00008589 }, /* GL_SRC1_ALPHA */
+   { 29402, 0x00008581 }, /* GL_SRC1_RGB */
+   { 29414, 0x0000858A }, /* GL_SRC2_ALPHA */
+   { 29428, 0x00008582 }, /* GL_SRC2_RGB */
+   { 29440, 0x00000302 }, /* GL_SRC_ALPHA */
+   { 29453, 0x00000308 }, /* GL_SRC_ALPHA_SATURATE */
+   { 29475, 0x00000300 }, /* GL_SRC_COLOR */
+   { 29488, 0x00000503 }, /* GL_STACK_OVERFLOW */
+   { 29506, 0x00000504 }, /* GL_STACK_UNDERFLOW */
+   { 29525, 0x000088E6 }, /* GL_STATIC_COPY */
+   { 29540, 0x000088E6 }, /* GL_STATIC_COPY_ARB */
+   { 29559, 0x000088E4 }, /* GL_STATIC_DRAW */
+   { 29574, 0x000088E4 }, /* GL_STATIC_DRAW_ARB */
+   { 29593, 0x000088E5 }, /* GL_STATIC_READ */
+   { 29608, 0x000088E5 }, /* GL_STATIC_READ_ARB */
+   { 29627, 0x00001802 }, /* GL_STENCIL */
+   { 29638, 0x00008D20 }, /* GL_STENCIL_ATTACHMENT_EXT */
+   { 29664, 0x00008801 }, /* GL_STENCIL_BACK_FAIL */
+   { 29685, 0x00008801 }, /* GL_STENCIL_BACK_FAIL_ATI */
+   { 29710, 0x00008800 }, /* GL_STENCIL_BACK_FUNC */
+   { 29731, 0x00008800 }, /* GL_STENCIL_BACK_FUNC_ATI */
+   { 29756, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+   { 29788, 0x00008802 }, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL_ATI */
+   { 29824, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+   { 29856, 0x00008803 }, /* GL_STENCIL_BACK_PASS_DEPTH_PASS_ATI */
+   { 29892, 0x00008CA3 }, /* GL_STENCIL_BACK_REF */
+   { 29912, 0x00008CA4 }, /* GL_STENCIL_BACK_VALUE_MASK */
+   { 29939, 0x00008CA5 }, /* GL_STENCIL_BACK_WRITEMASK */
+   { 29965, 0x00000D57 }, /* GL_STENCIL_BITS */
+   { 29981, 0x00000400 }, /* GL_STENCIL_BUFFER_BIT */
+   { 30003, 0x00000B91 }, /* GL_STENCIL_CLEAR_VALUE */
+   { 30026, 0x00000B94 }, /* GL_STENCIL_FAIL */
+   { 30042, 0x00000B92 }, /* GL_STENCIL_FUNC */
+   { 30058, 0x00001901 }, /* GL_STENCIL_INDEX */
+   { 30075, 0x00008D49 }, /* GL_STENCIL_INDEX16_EXT */
+   { 30098, 0x00008D46 }, /* GL_STENCIL_INDEX1_EXT */
+   { 30120, 0x00008D47 }, /* GL_STENCIL_INDEX4_EXT */
+   { 30142, 0x00008D48 }, /* GL_STENCIL_INDEX8_EXT */
+   { 30164, 0x00008D45 }, /* GL_STENCIL_INDEX_EXT */
+   { 30185, 0x00000B95 }, /* GL_STENCIL_PASS_DEPTH_FAIL */
+   { 30212, 0x00000B96 }, /* GL_STENCIL_PASS_DEPTH_PASS */
+   { 30239, 0x00000B97 }, /* GL_STENCIL_REF */
+   { 30254, 0x00000B90 }, /* GL_STENCIL_TEST */
+   { 30270, 0x00008910 }, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+   { 30299, 0x00000B93 }, /* GL_STENCIL_VALUE_MASK */
+   { 30321, 0x00000B98 }, /* GL_STENCIL_WRITEMASK */
+   { 30342, 0x00000C33 }, /* GL_STEREO */
+   { 30352, 0x000088E2 }, /* GL_STREAM_COPY */
+   { 30367, 0x000088E2 }, /* GL_STREAM_COPY_ARB */
+   { 30386, 0x000088E0 }, /* GL_STREAM_DRAW */
+   { 30401, 0x000088E0 }, /* GL_STREAM_DRAW_ARB */
+   { 30420, 0x000088E1 }, /* GL_STREAM_READ */
+   { 30435, 0x000088E1 }, /* GL_STREAM_READ_ARB */
+   { 30454, 0x00000D50 }, /* GL_SUBPIXEL_BITS */
+   { 30471, 0x000084E7 }, /* GL_SUBTRACT */
+   { 30483, 0x000084E7 }, /* GL_SUBTRACT_ARB */
+   { 30499, 0x00002001 }, /* GL_T */
+   { 30504, 0x00002A2A }, /* GL_T2F_C3F_V3F */
+   { 30519, 0x00002A2C }, /* GL_T2F_C4F_N3F_V3F */
+   { 30538, 0x00002A29 }, /* GL_T2F_C4UB_V3F */
+   { 30554, 0x00002A2B }, /* GL_T2F_N3F_V3F */
+   { 30569, 0x00002A27 }, /* GL_T2F_V3F */
+   { 30580, 0x00002A2D }, /* GL_T4F_C4F_N3F_V4F */
+   { 30599, 0x00002A28 }, /* GL_T4F_V4F */
+   { 30610, 0x00008031 }, /* GL_TABLE_TOO_LARGE_EXT */
+   { 30633, 0x00001702 }, /* GL_TEXTURE */
+   { 30644, 0x000084C0 }, /* GL_TEXTURE0 */
+   { 30656, 0x000084C0 }, /* GL_TEXTURE0_ARB */
+   { 30672, 0x000084C1 }, /* GL_TEXTURE1 */
+   { 30684, 0x000084CA }, /* GL_TEXTURE10 */
+   { 30697, 0x000084CA }, /* GL_TEXTURE10_ARB */
+   { 30714, 0x000084CB }, /* GL_TEXTURE11 */
+   { 30727, 0x000084CB }, /* GL_TEXTURE11_ARB */
+   { 30744, 0x000084CC }, /* GL_TEXTURE12 */
+   { 30757, 0x000084CC }, /* GL_TEXTURE12_ARB */
+   { 30774, 0x000084CD }, /* GL_TEXTURE13 */
+   { 30787, 0x000084CD }, /* GL_TEXTURE13_ARB */
+   { 30804, 0x000084CE }, /* GL_TEXTURE14 */
+   { 30817, 0x000084CE }, /* GL_TEXTURE14_ARB */
+   { 30834, 0x000084CF }, /* GL_TEXTURE15 */
+   { 30847, 0x000084CF }, /* GL_TEXTURE15_ARB */
+   { 30864, 0x000084D0 }, /* GL_TEXTURE16 */
+   { 30877, 0x000084D0 }, /* GL_TEXTURE16_ARB */
+   { 30894, 0x000084D1 }, /* GL_TEXTURE17 */
+   { 30907, 0x000084D1 }, /* GL_TEXTURE17_ARB */
+   { 30924, 0x000084D2 }, /* GL_TEXTURE18 */
+   { 30937, 0x000084D2 }, /* GL_TEXTURE18_ARB */
+   { 30954, 0x000084D3 }, /* GL_TEXTURE19 */
+   { 30967, 0x000084D3 }, /* GL_TEXTURE19_ARB */
+   { 30984, 0x000084C1 }, /* GL_TEXTURE1_ARB */
+   { 31000, 0x000084C2 }, /* GL_TEXTURE2 */
+   { 31012, 0x000084D4 }, /* GL_TEXTURE20 */
+   { 31025, 0x000084D4 }, /* GL_TEXTURE20_ARB */
+   { 31042, 0x000084D5 }, /* GL_TEXTURE21 */
+   { 31055, 0x000084D5 }, /* GL_TEXTURE21_ARB */
+   { 31072, 0x000084D6 }, /* GL_TEXTURE22 */
+   { 31085, 0x000084D6 }, /* GL_TEXTURE22_ARB */
+   { 31102, 0x000084D7 }, /* GL_TEXTURE23 */
+   { 31115, 0x000084D7 }, /* GL_TEXTURE23_ARB */
+   { 31132, 0x000084D8 }, /* GL_TEXTURE24 */
+   { 31145, 0x000084D8 }, /* GL_TEXTURE24_ARB */
+   { 31162, 0x000084D9 }, /* GL_TEXTURE25 */
+   { 31175, 0x000084D9 }, /* GL_TEXTURE25_ARB */
+   { 31192, 0x000084DA }, /* GL_TEXTURE26 */
+   { 31205, 0x000084DA }, /* GL_TEXTURE26_ARB */
+   { 31222, 0x000084DB }, /* GL_TEXTURE27 */
+   { 31235, 0x000084DB }, /* GL_TEXTURE27_ARB */
+   { 31252, 0x000084DC }, /* GL_TEXTURE28 */
+   { 31265, 0x000084DC }, /* GL_TEXTURE28_ARB */
+   { 31282, 0x000084DD }, /* GL_TEXTURE29 */
+   { 31295, 0x000084DD }, /* GL_TEXTURE29_ARB */
+   { 31312, 0x000084C2 }, /* GL_TEXTURE2_ARB */
+   { 31328, 0x000084C3 }, /* GL_TEXTURE3 */
+   { 31340, 0x000084DE }, /* GL_TEXTURE30 */
+   { 31353, 0x000084DE }, /* GL_TEXTURE30_ARB */
+   { 31370, 0x000084DF }, /* GL_TEXTURE31 */
+   { 31383, 0x000084DF }, /* GL_TEXTURE31_ARB */
+   { 31400, 0x000084C3 }, /* GL_TEXTURE3_ARB */
+   { 31416, 0x000084C4 }, /* GL_TEXTURE4 */
+   { 31428, 0x000084C4 }, /* GL_TEXTURE4_ARB */
+   { 31444, 0x000084C5 }, /* GL_TEXTURE5 */
+   { 31456, 0x000084C5 }, /* GL_TEXTURE5_ARB */
+   { 31472, 0x000084C6 }, /* GL_TEXTURE6 */
+   { 31484, 0x000084C6 }, /* GL_TEXTURE6_ARB */
+   { 31500, 0x000084C7 }, /* GL_TEXTURE7 */
+   { 31512, 0x000084C7 }, /* GL_TEXTURE7_ARB */
+   { 31528, 0x000084C8 }, /* GL_TEXTURE8 */
+   { 31540, 0x000084C8 }, /* GL_TEXTURE8_ARB */
+   { 31556, 0x000084C9 }, /* GL_TEXTURE9 */
+   { 31568, 0x000084C9 }, /* GL_TEXTURE9_ARB */
+   { 31584, 0x00000DE0 }, /* GL_TEXTURE_1D */
+   { 31598, 0x00008C18 }, /* GL_TEXTURE_1D_ARRAY_EXT */
+   { 31622, 0x00000DE1 }, /* GL_TEXTURE_2D */
+   { 31636, 0x00008C1A }, /* GL_TEXTURE_2D_ARRAY_EXT */
+   { 31660, 0x0000806F }, /* GL_TEXTURE_3D */
+   { 31674, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE */
+   { 31696, 0x0000805F }, /* GL_TEXTURE_ALPHA_SIZE_EXT */
+   { 31722, 0x0000813C }, /* GL_TEXTURE_BASE_LEVEL */
+   { 31744, 0x00008068 }, /* GL_TEXTURE_BINDING_1D */
+   { 31766, 0x00008C1C }, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+   { 31798, 0x00008069 }, /* GL_TEXTURE_BINDING_2D */
+   { 31820, 0x00008C1D }, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+   { 31852, 0x0000806A }, /* GL_TEXTURE_BINDING_3D */
+   { 31874, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP */
+   { 31902, 0x00008514 }, /* GL_TEXTURE_BINDING_CUBE_MAP_ARB */
+   { 31934, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+   { 31967, 0x000084F6 }, /* GL_TEXTURE_BINDING_RECTANGLE_NV */
+   { 31999, 0x00040000 }, /* GL_TEXTURE_BIT */
+   { 32014, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE */
+   { 32035, 0x0000805E }, /* GL_TEXTURE_BLUE_SIZE_EXT */
+   { 32060, 0x00001005 }, /* GL_TEXTURE_BORDER */
+   { 32078, 0x00001004 }, /* GL_TEXTURE_BORDER_COLOR */
+   { 32102, 0x00008171 }, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+   { 32133, 0x00008176 }, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+   { 32163, 0x00008172 }, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+   { 32193, 0x00008175 }, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+   { 32228, 0x00008173 }, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+   { 32259, 0x00008174 }, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+   { 32297, 0x000080BC }, /* GL_TEXTURE_COLOR_TABLE_SGI */
+   { 32324, 0x000081EF }, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+   { 32356, 0x000080BF }, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+   { 32390, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC */
+   { 32414, 0x0000884D }, /* GL_TEXTURE_COMPARE_FUNC_ARB */
+   { 32442, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE */
+   { 32466, 0x0000884C }, /* GL_TEXTURE_COMPARE_MODE_ARB */
+   { 32494, 0x0000819B }, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+   { 32527, 0x0000819A }, /* GL_TEXTURE_COMPARE_SGIX */
+   { 32551, 0x00001003 }, /* GL_TEXTURE_COMPONENTS */
+   { 32573, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED */
+   { 32595, 0x000086A1 }, /* GL_TEXTURE_COMPRESSED_ARB */
+   { 32621, 0x000086A3 }, /* GL_TEXTURE_COMPRESSED_FORMATS_ARB */
+   { 32655, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+   { 32688, 0x000086A0 }, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB */
+   { 32725, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT */
+   { 32753, 0x000084EF }, /* GL_TEXTURE_COMPRESSION_HINT_ARB */
+   { 32785, 0x00008078 }, /* GL_TEXTURE_COORD_ARRAY */
+   { 32808, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+   { 32846, 0x0000889A }, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING_ARB */
+   { 32888, 0x00008092 }, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+   { 32919, 0x00008088 }, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+   { 32947, 0x0000808A }, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+   { 32977, 0x00008089 }, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+   { 33005, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP */
+   { 33025, 0x00008513 }, /* GL_TEXTURE_CUBE_MAP_ARB */
+   { 33049, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+   { 33080, 0x00008516 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X_ARB */
+   { 33115, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+   { 33146, 0x00008518 }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y_ARB */
+   { 33181, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+   { 33212, 0x0000851A }, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB */
+   { 33247, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+   { 33278, 0x00008515 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB */
+   { 33313, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+   { 33344, 0x00008517 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y_ARB */
+   { 33379, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+   { 33410, 0x00008519 }, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB */
+   { 33445, 0x00008071 }, /* GL_TEXTURE_DEPTH */
+   { 33462, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE */
+   { 33484, 0x0000884A }, /* GL_TEXTURE_DEPTH_SIZE_ARB */
+   { 33510, 0x00002300 }, /* GL_TEXTURE_ENV */
+   { 33525, 0x00002201 }, /* GL_TEXTURE_ENV_COLOR */
+   { 33546, 0x00002200 }, /* GL_TEXTURE_ENV_MODE */
+   { 33566, 0x00008500 }, /* GL_TEXTURE_FILTER_CONTROL */
+   { 33592, 0x00002500 }, /* GL_TEXTURE_GEN_MODE */
+   { 33612, 0x00000C63 }, /* GL_TEXTURE_GEN_Q */
+   { 33629, 0x00000C62 }, /* GL_TEXTURE_GEN_R */
+   { 33646, 0x00000C60 }, /* GL_TEXTURE_GEN_S */
+   { 33663, 0x00000C61 }, /* GL_TEXTURE_GEN_T */
+   { 33680, 0x0000819D }, /* GL_TEXTURE_GEQUAL_R_SGIX */
+   { 33705, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE */
+   { 33727, 0x0000805D }, /* GL_TEXTURE_GREEN_SIZE_EXT */
+   { 33753, 0x00001001 }, /* GL_TEXTURE_HEIGHT */
+   { 33771, 0x000080ED }, /* GL_TEXTURE_INDEX_SIZE_EXT */
+   { 33797, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE */
+   { 33823, 0x00008061 }, /* GL_TEXTURE_INTENSITY_SIZE_EXT */
+   { 33853, 0x00001003 }, /* GL_TEXTURE_INTERNAL_FORMAT */
+   { 33880, 0x0000819C }, /* GL_TEXTURE_LEQUAL_R_SGIX */
+   { 33905, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS */
+   { 33925, 0x00008501 }, /* GL_TEXTURE_LOD_BIAS_EXT */
+   { 33949, 0x00008190 }, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+   { 33976, 0x0000818E }, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+   { 34003, 0x0000818F }, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+   { 34030, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE */
+   { 34056, 0x00008060 }, /* GL_TEXTURE_LUMINANCE_SIZE_EXT */
+   { 34086, 0x00002800 }, /* GL_TEXTURE_MAG_FILTER */
+   { 34108, 0x00000BA8 }, /* GL_TEXTURE_MATRIX */
+   { 34126, 0x000084FE }, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+   { 34156, 0x0000836B }, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+   { 34184, 0x00008369 }, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+   { 34212, 0x0000836A }, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+   { 34240, 0x0000813D }, /* GL_TEXTURE_MAX_LEVEL */
+   { 34261, 0x0000813B }, /* GL_TEXTURE_MAX_LOD */
+   { 34280, 0x00002801 }, /* GL_TEXTURE_MIN_FILTER */
+   { 34302, 0x0000813A }, /* GL_TEXTURE_MIN_LOD */
+   { 34321, 0x00008066 }, /* GL_TEXTURE_PRIORITY */
+   { 34341, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_ARB */
+   { 34366, 0x000084F5 }, /* GL_TEXTURE_RECTANGLE_NV */
+   { 34390, 0x0000805C }, /* GL_TEXTURE_RED_SIZE */
+   { 34410, 0x0000805C }, /* GL_TEXTURE_RED_SIZE_EXT */
+   { 34434, 0x00008067 }, /* GL_TEXTURE_RESIDENT */
+   { 34454, 0x00000BA5 }, /* GL_TEXTURE_STACK_DEPTH */
+   { 34477, 0x00008065 }, /* GL_TEXTURE_TOO_LARGE_EXT */
+   { 34502, 0x0000888F }, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+   { 34536, 0x00001000 }, /* GL_TEXTURE_WIDTH */
+   { 34553, 0x00008072 }, /* GL_TEXTURE_WRAP_R */
+   { 34571, 0x00002802 }, /* GL_TEXTURE_WRAP_S */
+   { 34589, 0x00002803 }, /* GL_TEXTURE_WRAP_T */
+   { 34607, 0x000088BF }, /* GL_TIME_ELAPSED_EXT */
+   { 34627, 0x00008648 }, /* GL_TRACK_MATRIX_NV */
+   { 34646, 0x00008649 }, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+   { 34675, 0x00001000 }, /* GL_TRANSFORM_BIT */
+   { 34692, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX */
+   { 34718, 0x000084E6 }, /* GL_TRANSPOSE_COLOR_MATRIX_ARB */
+   { 34748, 0x000088B7 }, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+   { 34780, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+   { 34810, 0x000084E3 }, /* GL_TRANSPOSE_MODELVIEW_MATRIX_ARB */
+   { 34844, 0x0000862C }, /* GL_TRANSPOSE_NV */
+   { 34860, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+   { 34891, 0x000084E4 }, /* GL_TRANSPOSE_PROJECTION_MATRIX_ARB */
+   { 34926, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+   { 34954, 0x000084E5 }, /* GL_TRANSPOSE_TEXTURE_MATRIX_ARB */
+   { 34986, 0x00000004 }, /* GL_TRIANGLES */
+   { 34999, 0x00000006 }, /* GL_TRIANGLE_FAN */
+   { 35015, 0x00008615 }, /* GL_TRIANGLE_MESH_SUN */
+   { 35036, 0x00000005 }, /* GL_TRIANGLE_STRIP */
+   { 35054, 0x00000001 }, /* GL_TRUE */
+   { 35062, 0x00000CF5 }, /* GL_UNPACK_ALIGNMENT */
+   { 35082, 0x0000806E }, /* GL_UNPACK_IMAGE_HEIGHT */
+   { 35105, 0x00000CF1 }, /* GL_UNPACK_LSB_FIRST */
+   { 35125, 0x00000CF2 }, /* GL_UNPACK_ROW_LENGTH */
+   { 35146, 0x0000806D }, /* GL_UNPACK_SKIP_IMAGES */
+   { 35168, 0x00000CF4 }, /* GL_UNPACK_SKIP_PIXELS */
+   { 35190, 0x00000CF3 }, /* GL_UNPACK_SKIP_ROWS */
+   { 35210, 0x00000CF0 }, /* GL_UNPACK_SWAP_BYTES */
+   { 35231, 0x00001401 }, /* GL_UNSIGNED_BYTE */
+   { 35248, 0x00008362 }, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+   { 35275, 0x00008032 }, /* GL_UNSIGNED_BYTE_3_3_2 */
+   { 35298, 0x00001405 }, /* GL_UNSIGNED_INT */
+   { 35314, 0x00008036 }, /* GL_UNSIGNED_INT_10_10_10_2 */
+   { 35341, 0x000084FA }, /* GL_UNSIGNED_INT_24_8_NV */
+   { 35365, 0x00008368 }, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+   { 35396, 0x00008035 }, /* GL_UNSIGNED_INT_8_8_8_8 */
+   { 35420, 0x00008367 }, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+   { 35448, 0x00001403 }, /* GL_UNSIGNED_SHORT */
+   { 35466, 0x00008366 }, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+   { 35496, 0x00008033 }, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+   { 35522, 0x00008365 }, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+   { 35552, 0x00008034 }, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+   { 35578, 0x00008363 }, /* GL_UNSIGNED_SHORT_5_6_5 */
+   { 35602, 0x00008364 }, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+   { 35630, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+   { 35658, 0x000085BA }, /* GL_UNSIGNED_SHORT_8_8_MESA */
+   { 35685, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+   { 35717, 0x000085BB }, /* GL_UNSIGNED_SHORT_8_8_REV_MESA */
+   { 35748, 0x00008CA2 }, /* GL_UPPER_LEFT */
+   { 35762, 0x00002A20 }, /* GL_V2F */
+   { 35769, 0x00002A21 }, /* GL_V3F */
+   { 35776, 0x00008B83 }, /* GL_VALIDATE_STATUS */
+   { 35795, 0x00001F00 }, /* GL_VENDOR */
+   { 35805, 0x00001F02 }, /* GL_VERSION */
+   { 35816, 0x00008074 }, /* GL_VERTEX_ARRAY */
+   { 35832, 0x000085B5 }, /* GL_VERTEX_ARRAY_BINDING_APPLE */
+   { 35862, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+   { 35893, 0x00008896 }, /* GL_VERTEX_ARRAY_BUFFER_BINDING_ARB */
+   { 35928, 0x0000808E }, /* GL_VERTEX_ARRAY_POINTER */
+   { 35952, 0x0000807A }, /* GL_VERTEX_ARRAY_SIZE */
+   { 35973, 0x0000807C }, /* GL_VERTEX_ARRAY_STRIDE */
+   { 35996, 0x0000807B }, /* GL_VERTEX_ARRAY_TYPE */
+   { 36017, 0x00008650 }, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+   { 36044, 0x0000865A }, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+   { 36072, 0x0000865B }, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+   { 36100, 0x0000865C }, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+   { 36128, 0x0000865D }, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+   { 36156, 0x0000865E }, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+   { 36184, 0x0000865F }, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+   { 36212, 0x00008651 }, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+   { 36239, 0x00008652 }, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+   { 36266, 0x00008653 }, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+   { 36293, 0x00008654 }, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+   { 36320, 0x00008655 }, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+   { 36347, 0x00008656 }, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+   { 36374, 0x00008657 }, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+   { 36401, 0x00008658 }, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+   { 36428, 0x00008659 }, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+   { 36455, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+   { 36493, 0x0000889F }, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB */
+   { 36535, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+   { 36566, 0x00008622 }, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED_ARB */
+   { 36601, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+   { 36635, 0x0000886A }, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED_ARB */
+   { 36673, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+   { 36704, 0x00008645 }, /* GL_VERTEX_ATTRIB_ARRAY_POINTER_ARB */
+   { 36739, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+   { 36767, 0x00008623 }, /* GL_VERTEX_ATTRIB_ARRAY_SIZE_ARB */
+   { 36799, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+   { 36829, 0x00008624 }, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE_ARB */
+   { 36863, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+   { 36891, 0x00008625 }, /* GL_VERTEX_ATTRIB_ARRAY_TYPE_ARB */
+   { 36923, 0x000086A7 }, /* GL_VERTEX_BLEND_ARB */
+   { 36943, 0x00008620 }, /* GL_VERTEX_PROGRAM_ARB */
+   { 36965, 0x0000864A }, /* GL_VERTEX_PROGRAM_BINDING_NV */
+   { 36994, 0x00008620 }, /* GL_VERTEX_PROGRAM_NV */
+   { 37015, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+   { 37044, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_ARB */
+   { 37077, 0x00008642 }, /* GL_VERTEX_PROGRAM_POINT_SIZE_NV */
+   { 37109, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+   { 37136, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_ARB */
+   { 37167, 0x00008643 }, /* GL_VERTEX_PROGRAM_TWO_SIDE_NV */
+   { 37197, 0x00008B31 }, /* GL_VERTEX_SHADER */
+   { 37214, 0x00008B31 }, /* GL_VERTEX_SHADER_ARB */
+   { 37235, 0x00008621 }, /* GL_VERTEX_STATE_PROGRAM_NV */
+   { 37262, 0x00000BA2 }, /* GL_VIEWPORT */
+   { 37274, 0x00000800 }, /* GL_VIEWPORT_BIT */
+   { 37290, 0x000086AD }, /* GL_WEIGHT_ARRAY_ARB */
+   { 37310, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+   { 37341, 0x0000889E }, /* GL_WEIGHT_ARRAY_BUFFER_BINDING_ARB */
+   { 37376, 0x000086AC }, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+   { 37404, 0x000086AB }, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+   { 37429, 0x000086AA }, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+   { 37456, 0x000086A9 }, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+   { 37481, 0x000086A6 }, /* GL_WEIGHT_SUM_UNITY_ARB */
+   { 37505, 0x000081D4 }, /* GL_WRAP_BORDER_SUN */
+   { 37524, 0x000088B9 }, /* GL_WRITE_ONLY */
+   { 37538, 0x000088B9 }, /* GL_WRITE_ONLY_ARB */
+   { 37556, 0x00001506 }, /* GL_XOR */
+   { 37563, 0x000085B9 }, /* GL_YCBCR_422_APPLE */
+   { 37582, 0x00008757 }, /* GL_YCBCR_MESA */
+   { 37596, 0x00000000 }, /* GL_ZERO */
+   { 37604, 0x00000D16 }, /* GL_ZOOM_X */
+   { 37614, 0x00000D17 }, /* GL_ZOOM_Y */
 };
 
 static const unsigned reduced_enums[1284] =
 {
-       436, /* GL_FALSE */
-       645, /* GL_LINES */
-       647, /* GL_LINE_LOOP */
-       654, /* GL_LINE_STRIP */
-      1641, /* GL_TRIANGLES */
-      1644, /* GL_TRIANGLE_STRIP */
-      1642, /* GL_TRIANGLE_FAN */
-      1211, /* GL_QUADS */
-      1213, /* GL_QUAD_STRIP */
-      1099, /* GL_POLYGON */
-      1111, /* GL_POLYGON_STIPPLE_BIT */
-      1064, /* GL_PIXEL_MODE_BIT */
-       632, /* GL_LIGHTING_BIT */
-       458, /* GL_FOG_BIT */
+       434, /* GL_FALSE */
+       643, /* GL_LINES */
+       645, /* GL_LINE_LOOP */
+       652, /* GL_LINE_STRIP */
+      1639, /* GL_TRIANGLES */
+      1642, /* GL_TRIANGLE_STRIP */
+      1640, /* GL_TRIANGLE_FAN */
+      1209, /* GL_QUADS */
+      1211, /* GL_QUAD_STRIP */
+      1097, /* GL_POLYGON */
+      1109, /* GL_POLYGON_STIPPLE_BIT */
+      1062, /* GL_PIXEL_MODE_BIT */
+       630, /* GL_LIGHTING_BIT */
+       456, /* GL_FOG_BIT */
          8, /* GL_ACCUM */
-       664, /* GL_LOAD */
-      1253, /* GL_RETURN */
-       937, /* GL_MULT */
+       662, /* GL_LOAD */
+      1251, /* GL_RETURN */
+       935, /* GL_MULT */
         23, /* GL_ADD */
-       953, /* GL_NEVER */
-       622, /* GL_LESS */
-       426, /* GL_EQUAL */
-       621, /* GL_LEQUAL */
-       547, /* GL_GREATER */
-       968, /* GL_NOTEQUAL */
-       522, /* GL_GEQUAL */
+       951, /* GL_NEVER */
+       620, /* GL_LESS */
+       424, /* GL_EQUAL */
+       619, /* GL_LEQUAL */
+       545, /* GL_GREATER */
+       966, /* GL_NOTEQUAL */
+       520, /* GL_GEQUAL */
         46, /* GL_ALWAYS */
-      1386, /* GL_SRC_COLOR */
-       997, /* GL_ONE_MINUS_SRC_COLOR */
-      1384, /* GL_SRC_ALPHA */
-       996, /* GL_ONE_MINUS_SRC_ALPHA */
-       406, /* GL_DST_ALPHA */
-       994, /* GL_ONE_MINUS_DST_ALPHA */
-       407, /* GL_DST_COLOR */
-       995, /* GL_ONE_MINUS_DST_COLOR */
-      1385, /* GL_SRC_ALPHA_SATURATE */
-       510, /* GL_FRONT_LEFT */
-       511, /* GL_FRONT_RIGHT */
-        69, /* GL_BACK_LEFT */
-        70, /* GL_BACK_RIGHT */
-       507, /* GL_FRONT */
-        68, /* GL_BACK */
-       620, /* GL_LEFT */
-      1293, /* GL_RIGHT */
-       508, /* GL_FRONT_AND_BACK */
-        63, /* GL_AUX0 */
-        64, /* GL_AUX1 */
-        65, /* GL_AUX2 */
-        66, /* GL_AUX3 */
-       612, /* GL_INVALID_ENUM */
-       615, /* GL_INVALID_VALUE */
-       614, /* GL_INVALID_OPERATION */
-      1387, /* GL_STACK_OVERFLOW */
-      1388, /* GL_STACK_UNDERFLOW */
-      1022, /* GL_OUT_OF_MEMORY */
-       613, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
+      1384, /* GL_SRC_COLOR */
+       995, /* GL_ONE_MINUS_SRC_COLOR */
+      1382, /* GL_SRC_ALPHA */
+       994, /* GL_ONE_MINUS_SRC_ALPHA */
+       405, /* GL_DST_ALPHA */
+       992, /* GL_ONE_MINUS_DST_ALPHA */
+       406, /* GL_DST_COLOR */
+       993, /* GL_ONE_MINUS_DST_COLOR */
+      1383, /* GL_SRC_ALPHA_SATURATE */
+       508, /* GL_FRONT_LEFT */
+       509, /* GL_FRONT_RIGHT */
+        68, /* GL_BACK_LEFT */
+        69, /* GL_BACK_RIGHT */
+       505, /* GL_FRONT */
+        67, /* GL_BACK */
+       618, /* GL_LEFT */
+      1291, /* GL_RIGHT */
+       506, /* GL_FRONT_AND_BACK */
+        62, /* GL_AUX0 */
+        63, /* GL_AUX1 */
+        64, /* GL_AUX2 */
+        65, /* GL_AUX3 */
+       610, /* GL_INVALID_ENUM */
+       613, /* GL_INVALID_VALUE */
+       612, /* GL_INVALID_OPERATION */
+      1385, /* GL_STACK_OVERFLOW */
+      1386, /* GL_STACK_UNDERFLOW */
+      1020, /* GL_OUT_OF_MEMORY */
+       611, /* GL_INVALID_FRAMEBUFFER_OPERATION_EXT */
          0, /* GL_2D */
          2, /* GL_3D */
          3, /* GL_3D_COLOR */
          4, /* GL_3D_COLOR_TEXTURE */
          6, /* GL_4D_COLOR_TEXTURE */
-      1042, /* GL_PASS_THROUGH_TOKEN */
-      1098, /* GL_POINT_TOKEN */
-       655, /* GL_LINE_TOKEN */
-      1112, /* GL_POLYGON_TOKEN */
-        74, /* GL_BITMAP_TOKEN */
-       405, /* GL_DRAW_PIXEL_TOKEN */
-       271, /* GL_COPY_PIXEL_TOKEN */
-       648, /* GL_LINE_RESET_TOKEN */
-       429, /* GL_EXP */
-       430, /* GL_EXP2 */
-       304, /* GL_CW */
-       116, /* GL_CCW */
-       137, /* GL_COEFF */
-      1019, /* GL_ORDER */
-       344, /* GL_DOMAIN */
-       279, /* GL_CURRENT_COLOR */
-       282, /* GL_CURRENT_INDEX */
-       288, /* GL_CURRENT_NORMAL */
-       300, /* GL_CURRENT_TEXTURE_COORDS */
-       293, /* GL_CURRENT_RASTER_COLOR */
-       295, /* GL_CURRENT_RASTER_INDEX */
-       298, /* GL_CURRENT_RASTER_TEXTURE_COORDS */
-       296, /* GL_CURRENT_RASTER_POSITION */
-       297, /* GL_CURRENT_RASTER_POSITION_VALID */
-       294, /* GL_CURRENT_RASTER_DISTANCE */
-      1091, /* GL_POINT_SMOOTH */
-      1080, /* GL_POINT_SIZE */
-      1090, /* GL_POINT_SIZE_RANGE */
-      1081, /* GL_POINT_SIZE_GRANULARITY */
-       649, /* GL_LINE_SMOOTH */
-       656, /* GL_LINE_WIDTH */
-       658, /* GL_LINE_WIDTH_RANGE */
-       657, /* GL_LINE_WIDTH_GRANULARITY */
-       651, /* GL_LINE_STIPPLE */
-       652, /* GL_LINE_STIPPLE_PATTERN */
-       653, /* GL_LINE_STIPPLE_REPEAT */
-       663, /* GL_LIST_MODE */
-       822, /* GL_MAX_LIST_NESTING */
-       660, /* GL_LIST_BASE */
-       662, /* GL_LIST_INDEX */
-      1101, /* GL_POLYGON_MODE */
-      1108, /* GL_POLYGON_SMOOTH */
-      1110, /* GL_POLYGON_STIPPLE */
-       414, /* GL_EDGE_FLAG */
-       272, /* GL_CULL_FACE */
-       273, /* GL_CULL_FACE_MODE */
-       509, /* GL_FRONT_FACE */
-       631, /* GL_LIGHTING */
-       636, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
-       637, /* GL_LIGHT_MODEL_TWO_SIDE */
-       633, /* GL_LIGHT_MODEL_AMBIENT */
-      1339, /* GL_SHADE_MODEL */
-       168, /* GL_COLOR_MATERIAL_FACE */
-       169, /* GL_COLOR_MATERIAL_PARAMETER */
-       167, /* GL_COLOR_MATERIAL */
-       457, /* GL_FOG */
-       479, /* GL_FOG_INDEX */
-       475, /* GL_FOG_DENSITY */
-       483, /* GL_FOG_START */
-       477, /* GL_FOG_END */
-       480, /* GL_FOG_MODE */
-       459, /* GL_FOG_COLOR */
-       333, /* GL_DEPTH_RANGE */
-       338, /* GL_DEPTH_TEST */
-       341, /* GL_DEPTH_WRITEMASK */
-       321, /* GL_DEPTH_CLEAR_VALUE */
-       332, /* GL_DEPTH_FUNC */
+      1040, /* GL_PASS_THROUGH_TOKEN */
+      1096, /* GL_POINT_TOKEN */
+       653, /* GL_LINE_TOKEN */
+      1110, /* GL_POLYGON_TOKEN */
+        73, /* GL_BITMAP_TOKEN */
+       404, /* GL_DRAW_PIXEL_TOKEN */
+       270, /* GL_COPY_PIXEL_TOKEN */
+       646, /* GL_LINE_RESET_TOKEN */
+       427, /* GL_EXP */
+       428, /* GL_EXP2 */
+       303, /* GL_CW */
+       115, /* GL_CCW */
+       136, /* GL_COEFF */
+      1017, /* GL_ORDER */
+       343, /* GL_DOMAIN */
+       278, /* GL_CURRENT_COLOR */
+       281, /* GL_CURRENT_INDEX */
+       287, /* GL_CURRENT_NORMAL */
+       299, /* GL_CURRENT_TEXTURE_COORDS */
+       292, /* GL_CURRENT_RASTER_COLOR */
+       294, /* GL_CURRENT_RASTER_INDEX */
+       297, /* GL_CURRENT_RASTER_TEXTURE_COORDS */
+       295, /* GL_CURRENT_RASTER_POSITION */
+       296, /* GL_CURRENT_RASTER_POSITION_VALID */
+       293, /* GL_CURRENT_RASTER_DISTANCE */
+      1089, /* GL_POINT_SMOOTH */
+      1078, /* GL_POINT_SIZE */
+      1088, /* GL_POINT_SIZE_RANGE */
+      1079, /* GL_POINT_SIZE_GRANULARITY */
+       647, /* GL_LINE_SMOOTH */
+       654, /* GL_LINE_WIDTH */
+       656, /* GL_LINE_WIDTH_RANGE */
+       655, /* GL_LINE_WIDTH_GRANULARITY */
+       649, /* GL_LINE_STIPPLE */
+       650, /* GL_LINE_STIPPLE_PATTERN */
+       651, /* GL_LINE_STIPPLE_REPEAT */
+       661, /* GL_LIST_MODE */
+       820, /* GL_MAX_LIST_NESTING */
+       658, /* GL_LIST_BASE */
+       660, /* GL_LIST_INDEX */
+      1099, /* GL_POLYGON_MODE */
+      1106, /* GL_POLYGON_SMOOTH */
+      1108, /* GL_POLYGON_STIPPLE */
+       413, /* GL_EDGE_FLAG */
+       271, /* GL_CULL_FACE */
+       272, /* GL_CULL_FACE_MODE */
+       507, /* GL_FRONT_FACE */
+       629, /* GL_LIGHTING */
+       634, /* GL_LIGHT_MODEL_LOCAL_VIEWER */
+       635, /* GL_LIGHT_MODEL_TWO_SIDE */
+       631, /* GL_LIGHT_MODEL_AMBIENT */
+      1337, /* GL_SHADE_MODEL */
+       167, /* GL_COLOR_MATERIAL_FACE */
+       168, /* GL_COLOR_MATERIAL_PARAMETER */
+       166, /* GL_COLOR_MATERIAL */
+       455, /* GL_FOG */
+       477, /* GL_FOG_INDEX */
+       473, /* GL_FOG_DENSITY */
+       481, /* GL_FOG_START */
+       475, /* GL_FOG_END */
+       478, /* GL_FOG_MODE */
+       457, /* GL_FOG_COLOR */
+       332, /* GL_DEPTH_RANGE */
+       337, /* GL_DEPTH_TEST */
+       340, /* GL_DEPTH_WRITEMASK */
+       320, /* GL_DEPTH_CLEAR_VALUE */
+       331, /* GL_DEPTH_FUNC */
         12, /* GL_ACCUM_CLEAR_VALUE */
-      1422, /* GL_STENCIL_TEST */
-      1410, /* GL_STENCIL_CLEAR_VALUE */
-      1412, /* GL_STENCIL_FUNC */
-      1424, /* GL_STENCIL_VALUE_MASK */
-      1411, /* GL_STENCIL_FAIL */
-      1419, /* GL_STENCIL_PASS_DEPTH_FAIL */
-      1420, /* GL_STENCIL_PASS_DEPTH_PASS */
-      1421, /* GL_STENCIL_REF */
-      1425, /* GL_STENCIL_WRITEMASK */
-       791, /* GL_MATRIX_MODE */
-       958, /* GL_NORMALIZE */
-      1731, /* GL_VIEWPORT */
-       932, /* GL_MODELVIEW_STACK_DEPTH */
-      1191, /* GL_PROJECTION_STACK_DEPTH */
-      1620, /* GL_TEXTURE_STACK_DEPTH */
-       930, /* GL_MODELVIEW_MATRIX */
-      1190, /* GL_PROJECTION_MATRIX */
-      1605, /* GL_TEXTURE_MATRIX */
-        61, /* GL_ATTRIB_STACK_DEPTH */
-       127, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
+      1420, /* GL_STENCIL_TEST */
+      1408, /* GL_STENCIL_CLEAR_VALUE */
+      1410, /* GL_STENCIL_FUNC */
+      1422, /* GL_STENCIL_VALUE_MASK */
+      1409, /* GL_STENCIL_FAIL */
+      1417, /* GL_STENCIL_PASS_DEPTH_FAIL */
+      1418, /* GL_STENCIL_PASS_DEPTH_PASS */
+      1419, /* GL_STENCIL_REF */
+      1423, /* GL_STENCIL_WRITEMASK */
+       789, /* GL_MATRIX_MODE */
+       956, /* GL_NORMALIZE */
+      1729, /* GL_VIEWPORT */
+       930, /* GL_MODELVIEW_STACK_DEPTH */
+      1189, /* GL_PROJECTION_STACK_DEPTH */
+      1618, /* GL_TEXTURE_STACK_DEPTH */
+       928, /* GL_MODELVIEW_MATRIX */
+      1188, /* GL_PROJECTION_MATRIX */
+      1603, /* GL_TEXTURE_MATRIX */
+        60, /* GL_ATTRIB_STACK_DEPTH */
+       126, /* GL_CLIENT_ATTRIB_STACK_DEPTH */
         43, /* GL_ALPHA_TEST */
         44, /* GL_ALPHA_TEST_FUNC */
         45, /* GL_ALPHA_TEST_REF */
-       343, /* GL_DITHER */
-        78, /* GL_BLEND_DST */
-        86, /* GL_BLEND_SRC */
-        75, /* GL_BLEND */
-       666, /* GL_LOGIC_OP_MODE */
-       586, /* GL_INDEX_LOGIC_OP */
-       166, /* GL_COLOR_LOGIC_OP */
-        67, /* GL_AUX_BUFFERS */
-       354, /* GL_DRAW_BUFFER */
-      1223, /* GL_READ_BUFFER */
-      1320, /* GL_SCISSOR_BOX */
-      1321, /* GL_SCISSOR_TEST */
-       585, /* GL_INDEX_CLEAR_VALUE */
-       590, /* GL_INDEX_WRITEMASK */
-       163, /* GL_COLOR_CLEAR_VALUE */
-       205, /* GL_COLOR_WRITEMASK */
-       587, /* GL_INDEX_MODE */
-      1287, /* GL_RGBA_MODE */
-       353, /* GL_DOUBLEBUFFER */
-      1426, /* GL_STEREO */
-      1246, /* GL_RENDER_MODE */
-      1043, /* GL_PERSPECTIVE_CORRECTION_HINT */
-      1092, /* GL_POINT_SMOOTH_HINT */
-       650, /* GL_LINE_SMOOTH_HINT */
-      1109, /* GL_POLYGON_SMOOTH_HINT */
-       478, /* GL_FOG_HINT */
-      1586, /* GL_TEXTURE_GEN_S */
-      1587, /* GL_TEXTURE_GEN_T */
-      1585, /* GL_TEXTURE_GEN_R */
-      1584, /* GL_TEXTURE_GEN_Q */
-      1056, /* GL_PIXEL_MAP_I_TO_I */
-      1062, /* GL_PIXEL_MAP_S_TO_S */
-      1058, /* GL_PIXEL_MAP_I_TO_R */
-      1054, /* GL_PIXEL_MAP_I_TO_G */
-      1052, /* GL_PIXEL_MAP_I_TO_B */
-      1050, /* GL_PIXEL_MAP_I_TO_A */
-      1060, /* GL_PIXEL_MAP_R_TO_R */
-      1048, /* GL_PIXEL_MAP_G_TO_G */
-      1046, /* GL_PIXEL_MAP_B_TO_B */
-      1044, /* GL_PIXEL_MAP_A_TO_A */
-      1057, /* GL_PIXEL_MAP_I_TO_I_SIZE */
-      1063, /* GL_PIXEL_MAP_S_TO_S_SIZE */
-      1059, /* GL_PIXEL_MAP_I_TO_R_SIZE */
-      1055, /* GL_PIXEL_MAP_I_TO_G_SIZE */
-      1053, /* GL_PIXEL_MAP_I_TO_B_SIZE */
-      1051, /* GL_PIXEL_MAP_I_TO_A_SIZE */
-      1061, /* GL_PIXEL_MAP_R_TO_R_SIZE */
-      1049, /* GL_PIXEL_MAP_G_TO_G_SIZE */
-      1047, /* GL_PIXEL_MAP_B_TO_B_SIZE */
-      1045, /* GL_PIXEL_MAP_A_TO_A_SIZE */
-      1653, /* GL_UNPACK_SWAP_BYTES */
-      1648, /* GL_UNPACK_LSB_FIRST */
-      1649, /* GL_UNPACK_ROW_LENGTH */
-      1652, /* GL_UNPACK_SKIP_ROWS */
-      1651, /* GL_UNPACK_SKIP_PIXELS */
-      1646, /* GL_UNPACK_ALIGNMENT */
-      1031, /* GL_PACK_SWAP_BYTES */
-      1026, /* GL_PACK_LSB_FIRST */
-      1027, /* GL_PACK_ROW_LENGTH */
-      1030, /* GL_PACK_SKIP_ROWS */
-      1029, /* GL_PACK_SKIP_PIXELS */
-      1023, /* GL_PACK_ALIGNMENT */
-       744, /* GL_MAP_COLOR */
-       745, /* GL_MAP_STENCIL */
-       589, /* GL_INDEX_SHIFT */
-       588, /* GL_INDEX_OFFSET */
-      1235, /* GL_RED_SCALE */
-      1233, /* GL_RED_BIAS */
-      1748, /* GL_ZOOM_X */
-      1749, /* GL_ZOOM_Y */
-       551, /* GL_GREEN_SCALE */
-       549, /* GL_GREEN_BIAS */
-        92, /* GL_BLUE_SCALE */
-        90, /* GL_BLUE_BIAS */
+       342, /* GL_DITHER */
+        77, /* GL_BLEND_DST */
+        85, /* GL_BLEND_SRC */
+        74, /* GL_BLEND */
+       664, /* GL_LOGIC_OP_MODE */
+       584, /* GL_INDEX_LOGIC_OP */
+       165, /* GL_COLOR_LOGIC_OP */
+        66, /* GL_AUX_BUFFERS */
+       353, /* GL_DRAW_BUFFER */
+      1221, /* GL_READ_BUFFER */
+      1318, /* GL_SCISSOR_BOX */
+      1319, /* GL_SCISSOR_TEST */
+       583, /* GL_INDEX_CLEAR_VALUE */
+       588, /* GL_INDEX_WRITEMASK */
+       162, /* GL_COLOR_CLEAR_VALUE */
+       204, /* GL_COLOR_WRITEMASK */
+       585, /* GL_INDEX_MODE */
+      1285, /* GL_RGBA_MODE */
+       352, /* GL_DOUBLEBUFFER */
+      1424, /* GL_STEREO */
+      1244, /* GL_RENDER_MODE */
+      1041, /* GL_PERSPECTIVE_CORRECTION_HINT */
+      1090, /* GL_POINT_SMOOTH_HINT */
+       648, /* GL_LINE_SMOOTH_HINT */
+      1107, /* GL_POLYGON_SMOOTH_HINT */
+       476, /* GL_FOG_HINT */
+      1584, /* GL_TEXTURE_GEN_S */
+      1585, /* GL_TEXTURE_GEN_T */
+      1583, /* GL_TEXTURE_GEN_R */
+      1582, /* GL_TEXTURE_GEN_Q */
+      1054, /* GL_PIXEL_MAP_I_TO_I */
+      1060, /* GL_PIXEL_MAP_S_TO_S */
+      1056, /* GL_PIXEL_MAP_I_TO_R */
+      1052, /* GL_PIXEL_MAP_I_TO_G */
+      1050, /* GL_PIXEL_MAP_I_TO_B */
+      1048, /* GL_PIXEL_MAP_I_TO_A */
+      1058, /* GL_PIXEL_MAP_R_TO_R */
+      1046, /* GL_PIXEL_MAP_G_TO_G */
+      1044, /* GL_PIXEL_MAP_B_TO_B */
+      1042, /* GL_PIXEL_MAP_A_TO_A */
+      1055, /* GL_PIXEL_MAP_I_TO_I_SIZE */
+      1061, /* GL_PIXEL_MAP_S_TO_S_SIZE */
+      1057, /* GL_PIXEL_MAP_I_TO_R_SIZE */
+      1053, /* GL_PIXEL_MAP_I_TO_G_SIZE */
+      1051, /* GL_PIXEL_MAP_I_TO_B_SIZE */
+      1049, /* GL_PIXEL_MAP_I_TO_A_SIZE */
+      1059, /* GL_PIXEL_MAP_R_TO_R_SIZE */
+      1047, /* GL_PIXEL_MAP_G_TO_G_SIZE */
+      1045, /* GL_PIXEL_MAP_B_TO_B_SIZE */
+      1043, /* GL_PIXEL_MAP_A_TO_A_SIZE */
+      1651, /* GL_UNPACK_SWAP_BYTES */
+      1646, /* GL_UNPACK_LSB_FIRST */
+      1647, /* GL_UNPACK_ROW_LENGTH */
+      1650, /* GL_UNPACK_SKIP_ROWS */
+      1649, /* GL_UNPACK_SKIP_PIXELS */
+      1644, /* GL_UNPACK_ALIGNMENT */
+      1029, /* GL_PACK_SWAP_BYTES */
+      1024, /* GL_PACK_LSB_FIRST */
+      1025, /* GL_PACK_ROW_LENGTH */
+      1028, /* GL_PACK_SKIP_ROWS */
+      1027, /* GL_PACK_SKIP_PIXELS */
+      1021, /* GL_PACK_ALIGNMENT */
+       742, /* GL_MAP_COLOR */
+       743, /* GL_MAP_STENCIL */
+       587, /* GL_INDEX_SHIFT */
+       586, /* GL_INDEX_OFFSET */
+      1233, /* GL_RED_SCALE */
+      1231, /* GL_RED_BIAS */
+      1746, /* GL_ZOOM_X */
+      1747, /* GL_ZOOM_Y */
+       549, /* GL_GREEN_SCALE */
+       547, /* GL_GREEN_BIAS */
+        91, /* GL_BLUE_SCALE */
+        89, /* GL_BLUE_BIAS */
         42, /* GL_ALPHA_SCALE */
         40, /* GL_ALPHA_BIAS */
-       334, /* GL_DEPTH_SCALE */
-       315, /* GL_DEPTH_BIAS */
-       817, /* GL_MAX_EVAL_ORDER */
-       821, /* GL_MAX_LIGHTS */
-       800, /* GL_MAX_CLIP_PLANES */
-       865, /* GL_MAX_TEXTURE_SIZE */
-       827, /* GL_MAX_PIXEL_MAP_TABLE */
-       796, /* GL_MAX_ATTRIB_STACK_DEPTH */
-       824, /* GL_MAX_MODELVIEW_STACK_DEPTH */
-       825, /* GL_MAX_NAME_STACK_DEPTH */
-       853, /* GL_MAX_PROJECTION_STACK_DEPTH */
-       866, /* GL_MAX_TEXTURE_STACK_DEPTH */
-       880, /* GL_MAX_VIEWPORT_DIMS */
-       797, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
-      1433, /* GL_SUBPIXEL_BITS */
-       584, /* GL_INDEX_BITS */
-      1234, /* GL_RED_BITS */
-       550, /* GL_GREEN_BITS */
-        91, /* GL_BLUE_BITS */
+       333, /* GL_DEPTH_SCALE */
+       314, /* GL_DEPTH_BIAS */
+       815, /* GL_MAX_EVAL_ORDER */
+       819, /* GL_MAX_LIGHTS */
+       798, /* GL_MAX_CLIP_PLANES */
+       863, /* GL_MAX_TEXTURE_SIZE */
+       825, /* GL_MAX_PIXEL_MAP_TABLE */
+       794, /* GL_MAX_ATTRIB_STACK_DEPTH */
+       822, /* GL_MAX_MODELVIEW_STACK_DEPTH */
+       823, /* GL_MAX_NAME_STACK_DEPTH */
+       851, /* GL_MAX_PROJECTION_STACK_DEPTH */
+       864, /* GL_MAX_TEXTURE_STACK_DEPTH */
+       878, /* GL_MAX_VIEWPORT_DIMS */
+       795, /* GL_MAX_CLIENT_ATTRIB_STACK_DEPTH */
+      1431, /* GL_SUBPIXEL_BITS */
+       582, /* GL_INDEX_BITS */
+      1232, /* GL_RED_BITS */
+       548, /* GL_GREEN_BITS */
+        90, /* GL_BLUE_BITS */
         41, /* GL_ALPHA_BITS */
-       316, /* GL_DEPTH_BITS */
-      1408, /* GL_STENCIL_BITS */
+       315, /* GL_DEPTH_BITS */
+      1406, /* GL_STENCIL_BITS */
         14, /* GL_ACCUM_RED_BITS */
         13, /* GL_ACCUM_GREEN_BITS */
         10, /* GL_ACCUM_BLUE_BITS */
          9, /* GL_ACCUM_ALPHA_BITS */
-       946, /* GL_NAME_STACK_DEPTH */
-        62, /* GL_AUTO_NORMAL */
-       690, /* GL_MAP1_COLOR_4 */
-       693, /* GL_MAP1_INDEX */
-       694, /* GL_MAP1_NORMAL */
-       695, /* GL_MAP1_TEXTURE_COORD_1 */
-       696, /* GL_MAP1_TEXTURE_COORD_2 */
-       697, /* GL_MAP1_TEXTURE_COORD_3 */
-       698, /* GL_MAP1_TEXTURE_COORD_4 */
-       699, /* GL_MAP1_VERTEX_3 */
-       700, /* GL_MAP1_VERTEX_4 */
-       717, /* GL_MAP2_COLOR_4 */
-       720, /* GL_MAP2_INDEX */
-       721, /* GL_MAP2_NORMAL */
-       722, /* GL_MAP2_TEXTURE_COORD_1 */
-       723, /* GL_MAP2_TEXTURE_COORD_2 */
-       724, /* GL_MAP2_TEXTURE_COORD_3 */
-       725, /* GL_MAP2_TEXTURE_COORD_4 */
-       726, /* GL_MAP2_VERTEX_3 */
-       727, /* GL_MAP2_VERTEX_4 */
-       691, /* GL_MAP1_GRID_DOMAIN */
-       692, /* GL_MAP1_GRID_SEGMENTS */
-       718, /* GL_MAP2_GRID_DOMAIN */
-       719, /* GL_MAP2_GRID_SEGMENTS */
-      1510, /* GL_TEXTURE_1D */
-      1512, /* GL_TEXTURE_2D */
-       439, /* GL_FEEDBACK_BUFFER_POINTER */
-       440, /* GL_FEEDBACK_BUFFER_SIZE */
-       441, /* GL_FEEDBACK_BUFFER_TYPE */
-      1330, /* GL_SELECTION_BUFFER_POINTER */
-      1331, /* GL_SELECTION_BUFFER_SIZE */
-      1623, /* GL_TEXTURE_WIDTH */
-      1591, /* GL_TEXTURE_HEIGHT */
-      1547, /* GL_TEXTURE_COMPONENTS */
-      1531, /* GL_TEXTURE_BORDER_COLOR */
-      1530, /* GL_TEXTURE_BORDER */
-       345, /* GL_DONT_CARE */
-       437, /* GL_FASTEST */
-       954, /* GL_NICEST */
+       944, /* GL_NAME_STACK_DEPTH */
+        61, /* GL_AUTO_NORMAL */
+       688, /* GL_MAP1_COLOR_4 */
+       691, /* GL_MAP1_INDEX */
+       692, /* GL_MAP1_NORMAL */
+       693, /* GL_MAP1_TEXTURE_COORD_1 */
+       694, /* GL_MAP1_TEXTURE_COORD_2 */
+       695, /* GL_MAP1_TEXTURE_COORD_3 */
+       696, /* GL_MAP1_TEXTURE_COORD_4 */
+       697, /* GL_MAP1_VERTEX_3 */
+       698, /* GL_MAP1_VERTEX_4 */
+       715, /* GL_MAP2_COLOR_4 */
+       718, /* GL_MAP2_INDEX */
+       719, /* GL_MAP2_NORMAL */
+       720, /* GL_MAP2_TEXTURE_COORD_1 */
+       721, /* GL_MAP2_TEXTURE_COORD_2 */
+       722, /* GL_MAP2_TEXTURE_COORD_3 */
+       723, /* GL_MAP2_TEXTURE_COORD_4 */
+       724, /* GL_MAP2_VERTEX_3 */
+       725, /* GL_MAP2_VERTEX_4 */
+       689, /* GL_MAP1_GRID_DOMAIN */
+       690, /* GL_MAP1_GRID_SEGMENTS */
+       716, /* GL_MAP2_GRID_DOMAIN */
+       717, /* GL_MAP2_GRID_SEGMENTS */
+      1508, /* GL_TEXTURE_1D */
+      1510, /* GL_TEXTURE_2D */
+       437, /* GL_FEEDBACK_BUFFER_POINTER */
+       438, /* GL_FEEDBACK_BUFFER_SIZE */
+       439, /* GL_FEEDBACK_BUFFER_TYPE */
+      1328, /* GL_SELECTION_BUFFER_POINTER */
+      1329, /* GL_SELECTION_BUFFER_SIZE */
+      1621, /* GL_TEXTURE_WIDTH */
+      1589, /* GL_TEXTURE_HEIGHT */
+      1545, /* GL_TEXTURE_COMPONENTS */
+      1529, /* GL_TEXTURE_BORDER_COLOR */
+      1528, /* GL_TEXTURE_BORDER */
+       344, /* GL_DONT_CARE */
+       435, /* GL_FASTEST */
+       952, /* GL_NICEST */
         47, /* GL_AMBIENT */
-       342, /* GL_DIFFUSE */
-      1373, /* GL_SPECULAR */
-      1113, /* GL_POSITION */
-      1376, /* GL_SPOT_DIRECTION */
-      1377, /* GL_SPOT_EXPONENT */
-      1375, /* GL_SPOT_CUTOFF */
-       245, /* GL_CONSTANT_ATTENUATION */
-       640, /* GL_LINEAR_ATTENUATION */
-      1210, /* GL_QUADRATIC_ATTENUATION */
-       219, /* GL_COMPILE */
-       220, /* GL_COMPILE_AND_EXECUTE */
-       111, /* GL_BYTE */
-      1654, /* GL_UNSIGNED_BYTE */
-      1344, /* GL_SHORT */
-      1663, /* GL_UNSIGNED_SHORT */
-       592, /* GL_INT */
-      1657, /* GL_UNSIGNED_INT */
-       444, /* GL_FLOAT */
+       341, /* GL_DIFFUSE */
+      1371, /* GL_SPECULAR */
+      1111, /* GL_POSITION */
+      1374, /* GL_SPOT_DIRECTION */
+      1375, /* GL_SPOT_EXPONENT */
+      1373, /* GL_SPOT_CUTOFF */
+       244, /* GL_CONSTANT_ATTENUATION */
+       638, /* GL_LINEAR_ATTENUATION */
+      1208, /* GL_QUADRATIC_ATTENUATION */
+       218, /* GL_COMPILE */
+       219, /* GL_COMPILE_AND_EXECUTE */
+       110, /* GL_BYTE */
+      1652, /* GL_UNSIGNED_BYTE */
+      1342, /* GL_SHORT */
+      1661, /* GL_UNSIGNED_SHORT */
+       590, /* GL_INT */
+      1655, /* GL_UNSIGNED_INT */
+       442, /* GL_FLOAT */
          1, /* GL_2_BYTES */
          5, /* GL_3_BYTES */
          7, /* GL_4_BYTES */
-       352, /* GL_DOUBLE */
-       123, /* GL_CLEAR */
+       351, /* GL_DOUBLE */
+       122, /* GL_CLEAR */
         49, /* GL_AND */
         51, /* GL_AND_REVERSE */
-       269, /* GL_COPY */
+       268, /* GL_COPY */
         50, /* GL_AND_INVERTED */
-       956, /* GL_NOOP */
-      1744, /* GL_XOR */
-      1018, /* GL_OR */
-       957, /* GL_NOR */
-       427, /* GL_EQUIV */
-       618, /* GL_INVERT */
-      1021, /* GL_OR_REVERSE */
-       270, /* GL_COPY_INVERTED */
-      1020, /* GL_OR_INVERTED */
-       947, /* GL_NAND */
-      1335, /* GL_SET */
-       424, /* GL_EMISSION */
-      1343, /* GL_SHININESS */
+       954, /* GL_NOOP */
+      1742, /* GL_XOR */
+      1016, /* GL_OR */
+       955, /* GL_NOR */
+       425, /* GL_EQUIV */
+       616, /* GL_INVERT */
+      1019, /* GL_OR_REVERSE */
+       269, /* GL_COPY_INVERTED */
+      1018, /* GL_OR_INVERTED */
+       945, /* GL_NAND */
+      1333, /* GL_SET */
+       422, /* GL_EMISSION */
+      1341, /* GL_SHININESS */
         48, /* GL_AMBIENT_AND_DIFFUSE */
-       165, /* GL_COLOR_INDEXES */
-       897, /* GL_MODELVIEW */
-      1189, /* GL_PROJECTION */
-      1445, /* GL_TEXTURE */
-       138, /* GL_COLOR */
-       313, /* GL_DEPTH */
-      1395, /* GL_STENCIL */
-       164, /* GL_COLOR_INDEX */
-      1413, /* GL_STENCIL_INDEX */
-       322, /* GL_DEPTH_COMPONENT */
-      1230, /* GL_RED */
-       548, /* GL_GREEN */
-        89, /* GL_BLUE */
+       164, /* GL_COLOR_INDEXES */
+       895, /* GL_MODELVIEW */
+      1187, /* GL_PROJECTION */
+      1443, /* GL_TEXTURE */
+       137, /* GL_COLOR */
+       312, /* GL_DEPTH */
+      1393, /* GL_STENCIL */
+       163, /* GL_COLOR_INDEX */
+      1411, /* GL_STENCIL_INDEX */
+       321, /* GL_DEPTH_COMPONENT */
+      1228, /* GL_RED */
+       546, /* GL_GREEN */
+        88, /* GL_BLUE */
         31, /* GL_ALPHA */
-      1254, /* GL_RGB */
-      1273, /* GL_RGBA */
-       668, /* GL_LUMINANCE */
-       689, /* GL_LUMINANCE_ALPHA */
-        73, /* GL_BITMAP */
-      1069, /* GL_POINT */
-       638, /* GL_LINE */
-       442, /* GL_FILL */
-      1239, /* GL_RENDER */
-       438, /* GL_FEEDBACK */
-      1329, /* GL_SELECT */
-       443, /* GL_FLAT */
-      1348, /* GL_SMOOTH */
-       619, /* GL_KEEP */
-      1248, /* GL_REPLACE */
-       575, /* GL_INCR */
-       309, /* GL_DECR */
-      1678, /* GL_VENDOR */
-      1245, /* GL_RENDERER */
-      1679, /* GL_VERSION */
-       431, /* GL_EXTENSIONS */
-      1294, /* GL_S */
-      1436, /* GL_T */
-      1220, /* GL_R */
-      1209, /* GL_Q */
-       933, /* GL_MODULATE */
-       308, /* GL_DECAL */
-      1581, /* GL_TEXTURE_ENV_MODE */
-      1580, /* GL_TEXTURE_ENV_COLOR */
-      1579, /* GL_TEXTURE_ENV */
-       432, /* GL_EYE_LINEAR */
-       980, /* GL_OBJECT_LINEAR */
-      1374, /* GL_SPHERE_MAP */
-      1583, /* GL_TEXTURE_GEN_MODE */
-       982, /* GL_OBJECT_PLANE */
-       433, /* GL_EYE_PLANE */
-       948, /* GL_NEAREST */
-       639, /* GL_LINEAR */
-       952, /* GL_NEAREST_MIPMAP_NEAREST */
-       644, /* GL_LINEAR_MIPMAP_NEAREST */
-       951, /* GL_NEAREST_MIPMAP_LINEAR */
-       643, /* GL_LINEAR_MIPMAP_LINEAR */
-      1604, /* GL_TEXTURE_MAG_FILTER */
-      1612, /* GL_TEXTURE_MIN_FILTER */
-      1625, /* GL_TEXTURE_WRAP_S */
-      1626, /* GL_TEXTURE_WRAP_T */
-       117, /* GL_CLAMP */
-      1247, /* GL_REPEAT */
-      1107, /* GL_POLYGON_OFFSET_UNITS */
-      1106, /* GL_POLYGON_OFFSET_POINT */
-      1105, /* GL_POLYGON_OFFSET_LINE */
-      1221, /* GL_R3_G3_B2 */
-      1675, /* GL_V2F */
-      1676, /* GL_V3F */
-       114, /* GL_C4UB_V2F */
-       115, /* GL_C4UB_V3F */
-       112, /* GL_C3F_V3F */
-       945, /* GL_N3F_V3F */
-       113, /* GL_C4F_N3F_V3F */
-      1441, /* GL_T2F_V3F */
-      1443, /* GL_T4F_V4F */
-      1439, /* GL_T2F_C4UB_V3F */
-      1437, /* GL_T2F_C3F_V3F */
-      1440, /* GL_T2F_N3F_V3F */
-      1438, /* GL_T2F_C4F_N3F_V3F */
-      1442, /* GL_T4F_C4F_N3F_V4F */
-       130, /* GL_CLIP_PLANE0 */
-       131, /* GL_CLIP_PLANE1 */
-       132, /* GL_CLIP_PLANE2 */
-       133, /* GL_CLIP_PLANE3 */
-       134, /* GL_CLIP_PLANE4 */
-       135, /* GL_CLIP_PLANE5 */
-       623, /* GL_LIGHT0 */
-       624, /* GL_LIGHT1 */
-       625, /* GL_LIGHT2 */
-       626, /* GL_LIGHT3 */
-       627, /* GL_LIGHT4 */
-       628, /* GL_LIGHT5 */
-       629, /* GL_LIGHT6 */
-       630, /* GL_LIGHT7 */
-       552, /* GL_HINT_BIT */
-       247, /* GL_CONSTANT_COLOR */
-       992, /* GL_ONE_MINUS_CONSTANT_COLOR */
-       242, /* GL_CONSTANT_ALPHA */
-       990, /* GL_ONE_MINUS_CONSTANT_ALPHA */
-        76, /* GL_BLEND_COLOR */
-       512, /* GL_FUNC_ADD */
-       881, /* GL_MIN */
-       793, /* GL_MAX */
-        81, /* GL_BLEND_EQUATION */
-       516, /* GL_FUNC_SUBTRACT */
-       514, /* GL_FUNC_REVERSE_SUBTRACT */
-       250, /* GL_CONVOLUTION_1D */
-       251, /* GL_CONVOLUTION_2D */
-      1332, /* GL_SEPARABLE_2D */
-       254, /* GL_CONVOLUTION_BORDER_MODE */
-       258, /* GL_CONVOLUTION_FILTER_SCALE */
-       256, /* GL_CONVOLUTION_FILTER_BIAS */
-      1231, /* GL_REDUCE */
-       260, /* GL_CONVOLUTION_FORMAT */
-       264, /* GL_CONVOLUTION_WIDTH */
-       262, /* GL_CONVOLUTION_HEIGHT */
-       808, /* GL_MAX_CONVOLUTION_WIDTH */
-       806, /* GL_MAX_CONVOLUTION_HEIGHT */
-      1146, /* GL_POST_CONVOLUTION_RED_SCALE */
-      1142, /* GL_POST_CONVOLUTION_GREEN_SCALE */
-      1137, /* GL_POST_CONVOLUTION_BLUE_SCALE */
-      1133, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
-      1144, /* GL_POST_CONVOLUTION_RED_BIAS */
-      1140, /* GL_POST_CONVOLUTION_GREEN_BIAS */
-      1135, /* GL_POST_CONVOLUTION_BLUE_BIAS */
-      1131, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
-       553, /* GL_HISTOGRAM */
-      1193, /* GL_PROXY_HISTOGRAM */
-       569, /* GL_HISTOGRAM_WIDTH */
-       559, /* GL_HISTOGRAM_FORMAT */
-       565, /* GL_HISTOGRAM_RED_SIZE */
-       561, /* GL_HISTOGRAM_GREEN_SIZE */
-       556, /* GL_HISTOGRAM_BLUE_SIZE */
-       554, /* GL_HISTOGRAM_ALPHA_SIZE */
-       563, /* GL_HISTOGRAM_LUMINANCE_SIZE */
-       567, /* GL_HISTOGRAM_SINK */
-       882, /* GL_MINMAX */
-       884, /* GL_MINMAX_FORMAT */
-       886, /* GL_MINMAX_SINK */
-      1444, /* GL_TABLE_TOO_LARGE_EXT */
-      1656, /* GL_UNSIGNED_BYTE_3_3_2 */
-      1665, /* GL_UNSIGNED_SHORT_4_4_4_4 */
-      1667, /* GL_UNSIGNED_SHORT_5_5_5_1 */
-      1661, /* GL_UNSIGNED_INT_8_8_8_8 */
-      1658, /* GL_UNSIGNED_INT_10_10_10_2 */
-      1104, /* GL_POLYGON_OFFSET_FILL */
-      1103, /* GL_POLYGON_OFFSET_FACTOR */
-      1102, /* GL_POLYGON_OFFSET_BIAS */
-      1251, /* GL_RESCALE_NORMAL */
+      1252, /* GL_RGB */
+      1271, /* GL_RGBA */
+       666, /* GL_LUMINANCE */
+       687, /* GL_LUMINANCE_ALPHA */
+        72, /* GL_BITMAP */
+      1067, /* GL_POINT */
+       636, /* GL_LINE */
+       440, /* GL_FILL */
+      1237, /* GL_RENDER */
+       436, /* GL_FEEDBACK */
+      1327, /* GL_SELECT */
+       441, /* GL_FLAT */
+      1346, /* GL_SMOOTH */
+       617, /* GL_KEEP */
+      1246, /* GL_REPLACE */
+       573, /* GL_INCR */
+       308, /* GL_DECR */
+      1676, /* GL_VENDOR */
+      1243, /* GL_RENDERER */
+      1677, /* GL_VERSION */
+       429, /* GL_EXTENSIONS */
+      1292, /* GL_S */
+      1434, /* GL_T */
+      1218, /* GL_R */
+      1207, /* GL_Q */
+       931, /* GL_MODULATE */
+       307, /* GL_DECAL */
+      1579, /* GL_TEXTURE_ENV_MODE */
+      1578, /* GL_TEXTURE_ENV_COLOR */
+      1577, /* GL_TEXTURE_ENV */
+       430, /* GL_EYE_LINEAR */
+       978, /* GL_OBJECT_LINEAR */
+      1372, /* GL_SPHERE_MAP */
+      1581, /* GL_TEXTURE_GEN_MODE */
+       980, /* GL_OBJECT_PLANE */
+       431, /* GL_EYE_PLANE */
+       946, /* GL_NEAREST */
+       637, /* GL_LINEAR */
+       950, /* GL_NEAREST_MIPMAP_NEAREST */
+       642, /* GL_LINEAR_MIPMAP_NEAREST */
+       949, /* GL_NEAREST_MIPMAP_LINEAR */
+       641, /* GL_LINEAR_MIPMAP_LINEAR */
+      1602, /* GL_TEXTURE_MAG_FILTER */
+      1610, /* GL_TEXTURE_MIN_FILTER */
+      1623, /* GL_TEXTURE_WRAP_S */
+      1624, /* GL_TEXTURE_WRAP_T */
+       116, /* GL_CLAMP */
+      1245, /* GL_REPEAT */
+      1105, /* GL_POLYGON_OFFSET_UNITS */
+      1104, /* GL_POLYGON_OFFSET_POINT */
+      1103, /* GL_POLYGON_OFFSET_LINE */
+      1219, /* GL_R3_G3_B2 */
+      1673, /* GL_V2F */
+      1674, /* GL_V3F */
+       113, /* GL_C4UB_V2F */
+       114, /* GL_C4UB_V3F */
+       111, /* GL_C3F_V3F */
+       943, /* GL_N3F_V3F */
+       112, /* GL_C4F_N3F_V3F */
+      1439, /* GL_T2F_V3F */
+      1441, /* GL_T4F_V4F */
+      1437, /* GL_T2F_C4UB_V3F */
+      1435, /* GL_T2F_C3F_V3F */
+      1438, /* GL_T2F_N3F_V3F */
+      1436, /* GL_T2F_C4F_N3F_V3F */
+      1440, /* GL_T4F_C4F_N3F_V4F */
+       129, /* GL_CLIP_PLANE0 */
+       130, /* GL_CLIP_PLANE1 */
+       131, /* GL_CLIP_PLANE2 */
+       132, /* GL_CLIP_PLANE3 */
+       133, /* GL_CLIP_PLANE4 */
+       134, /* GL_CLIP_PLANE5 */
+       621, /* GL_LIGHT0 */
+       622, /* GL_LIGHT1 */
+       623, /* GL_LIGHT2 */
+       624, /* GL_LIGHT3 */
+       625, /* GL_LIGHT4 */
+       626, /* GL_LIGHT5 */
+       627, /* GL_LIGHT6 */
+       628, /* GL_LIGHT7 */
+       550, /* GL_HINT_BIT */
+       246, /* GL_CONSTANT_COLOR */
+       990, /* GL_ONE_MINUS_CONSTANT_COLOR */
+       241, /* GL_CONSTANT_ALPHA */
+       988, /* GL_ONE_MINUS_CONSTANT_ALPHA */
+        75, /* GL_BLEND_COLOR */
+       510, /* GL_FUNC_ADD */
+       879, /* GL_MIN */
+       791, /* GL_MAX */
+        80, /* GL_BLEND_EQUATION */
+       514, /* GL_FUNC_SUBTRACT */
+       512, /* GL_FUNC_REVERSE_SUBTRACT */
+       249, /* GL_CONVOLUTION_1D */
+       250, /* GL_CONVOLUTION_2D */
+      1330, /* GL_SEPARABLE_2D */
+       253, /* GL_CONVOLUTION_BORDER_MODE */
+       257, /* GL_CONVOLUTION_FILTER_SCALE */
+       255, /* GL_CONVOLUTION_FILTER_BIAS */
+      1229, /* GL_REDUCE */
+       259, /* GL_CONVOLUTION_FORMAT */
+       263, /* GL_CONVOLUTION_WIDTH */
+       261, /* GL_CONVOLUTION_HEIGHT */
+       806, /* GL_MAX_CONVOLUTION_WIDTH */
+       804, /* GL_MAX_CONVOLUTION_HEIGHT */
+      1144, /* GL_POST_CONVOLUTION_RED_SCALE */
+      1140, /* GL_POST_CONVOLUTION_GREEN_SCALE */
+      1135, /* GL_POST_CONVOLUTION_BLUE_SCALE */
+      1131, /* GL_POST_CONVOLUTION_ALPHA_SCALE */
+      1142, /* GL_POST_CONVOLUTION_RED_BIAS */
+      1138, /* GL_POST_CONVOLUTION_GREEN_BIAS */
+      1133, /* GL_POST_CONVOLUTION_BLUE_BIAS */
+      1129, /* GL_POST_CONVOLUTION_ALPHA_BIAS */
+       551, /* GL_HISTOGRAM */
+      1191, /* GL_PROXY_HISTOGRAM */
+       567, /* GL_HISTOGRAM_WIDTH */
+       557, /* GL_HISTOGRAM_FORMAT */
+       563, /* GL_HISTOGRAM_RED_SIZE */
+       559, /* GL_HISTOGRAM_GREEN_SIZE */
+       554, /* GL_HISTOGRAM_BLUE_SIZE */
+       552, /* GL_HISTOGRAM_ALPHA_SIZE */
+       561, /* GL_HISTOGRAM_LUMINANCE_SIZE */
+       565, /* GL_HISTOGRAM_SINK */
+       880, /* GL_MINMAX */
+       882, /* GL_MINMAX_FORMAT */
+       884, /* GL_MINMAX_SINK */
+      1442, /* GL_TABLE_TOO_LARGE_EXT */
+      1654, /* GL_UNSIGNED_BYTE_3_3_2 */
+      1663, /* GL_UNSIGNED_SHORT_4_4_4_4 */
+      1665, /* GL_UNSIGNED_SHORT_5_5_5_1 */
+      1659, /* GL_UNSIGNED_INT_8_8_8_8 */
+      1656, /* GL_UNSIGNED_INT_10_10_10_2 */
+      1102, /* GL_POLYGON_OFFSET_FILL */
+      1101, /* GL_POLYGON_OFFSET_FACTOR */
+      1100, /* GL_POLYGON_OFFSET_BIAS */
+      1249, /* GL_RESCALE_NORMAL */
         36, /* GL_ALPHA4 */
         38, /* GL_ALPHA8 */
         32, /* GL_ALPHA12 */
         34, /* GL_ALPHA16 */
-       679, /* GL_LUMINANCE4 */
-       685, /* GL_LUMINANCE8 */
-       669, /* GL_LUMINANCE12 */
-       675, /* GL_LUMINANCE16 */
-       680, /* GL_LUMINANCE4_ALPHA4 */
-       683, /* GL_LUMINANCE6_ALPHA2 */
-       686, /* GL_LUMINANCE8_ALPHA8 */
-       672, /* GL_LUMINANCE12_ALPHA4 */
-       670, /* GL_LUMINANCE12_ALPHA12 */
-       676, /* GL_LUMINANCE16_ALPHA16 */
-       593, /* GL_INTENSITY */
-       598, /* GL_INTENSITY4 */
-       600, /* GL_INTENSITY8 */
-       594, /* GL_INTENSITY12 */
-       596, /* GL_INTENSITY16 */
-      1263, /* GL_RGB2_EXT */
-      1264, /* GL_RGB4 */
-      1267, /* GL_RGB5 */
-      1271, /* GL_RGB8 */
-      1255, /* GL_RGB10 */
-      1259, /* GL_RGB12 */
-      1261, /* GL_RGB16 */
-      1278, /* GL_RGBA2 */
-      1280, /* GL_RGBA4 */
-      1268, /* GL_RGB5_A1 */
-      1284, /* GL_RGBA8 */
-      1256, /* GL_RGB10_A2 */
-      1274, /* GL_RGBA12 */
-      1276, /* GL_RGBA16 */
-      1617, /* GL_TEXTURE_RED_SIZE */
-      1589, /* GL_TEXTURE_GREEN_SIZE */
-      1528, /* GL_TEXTURE_BLUE_SIZE */
-      1515, /* GL_TEXTURE_ALPHA_SIZE */
-      1602, /* GL_TEXTURE_LUMINANCE_SIZE */
-      1593, /* GL_TEXTURE_INTENSITY_SIZE */
-      1249, /* GL_REPLACE_EXT */
-      1197, /* GL_PROXY_TEXTURE_1D */
-      1200, /* GL_PROXY_TEXTURE_2D */
-      1621, /* GL_TEXTURE_TOO_LARGE_EXT */
-      1614, /* GL_TEXTURE_PRIORITY */
-      1619, /* GL_TEXTURE_RESIDENT */
-      1518, /* GL_TEXTURE_BINDING_1D */
-      1520, /* GL_TEXTURE_BINDING_2D */
-      1522, /* GL_TEXTURE_BINDING_3D */
-      1028, /* GL_PACK_SKIP_IMAGES */
-      1024, /* GL_PACK_IMAGE_HEIGHT */
-      1650, /* GL_UNPACK_SKIP_IMAGES */
-      1647, /* GL_UNPACK_IMAGE_HEIGHT */
-      1514, /* GL_TEXTURE_3D */
-      1203, /* GL_PROXY_TEXTURE_3D */
-      1576, /* GL_TEXTURE_DEPTH */
-      1624, /* GL_TEXTURE_WRAP_R */
-       794, /* GL_MAX_3D_TEXTURE_SIZE */
-      1680, /* GL_VERTEX_ARRAY */
-       959, /* GL_NORMAL_ARRAY */
-       139, /* GL_COLOR_ARRAY */
-       578, /* GL_INDEX_ARRAY */
-      1555, /* GL_TEXTURE_COORD_ARRAY */
-       415, /* GL_EDGE_FLAG_ARRAY */
-      1685, /* GL_VERTEX_ARRAY_SIZE */
-      1687, /* GL_VERTEX_ARRAY_TYPE */
-      1686, /* GL_VERTEX_ARRAY_STRIDE */
-       964, /* GL_NORMAL_ARRAY_TYPE */
-       963, /* GL_NORMAL_ARRAY_STRIDE */
-       143, /* GL_COLOR_ARRAY_SIZE */
-       145, /* GL_COLOR_ARRAY_TYPE */
-       144, /* GL_COLOR_ARRAY_STRIDE */
-       583, /* GL_INDEX_ARRAY_TYPE */
-       582, /* GL_INDEX_ARRAY_STRIDE */
-      1559, /* GL_TEXTURE_COORD_ARRAY_SIZE */
-      1561, /* GL_TEXTURE_COORD_ARRAY_TYPE */
-      1560, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
-       419, /* GL_EDGE_FLAG_ARRAY_STRIDE */
-      1684, /* GL_VERTEX_ARRAY_POINTER */
-       962, /* GL_NORMAL_ARRAY_POINTER */
-       142, /* GL_COLOR_ARRAY_POINTER */
-       581, /* GL_INDEX_ARRAY_POINTER */
-      1558, /* GL_TEXTURE_COORD_ARRAY_POINTER */
-       418, /* GL_EDGE_FLAG_ARRAY_POINTER */
-       938, /* GL_MULTISAMPLE */
-      1306, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
-      1308, /* GL_SAMPLE_ALPHA_TO_ONE */
-      1313, /* GL_SAMPLE_COVERAGE */
-      1310, /* GL_SAMPLE_BUFFERS */
-      1301, /* GL_SAMPLES */
-      1317, /* GL_SAMPLE_COVERAGE_VALUE */
-      1315, /* GL_SAMPLE_COVERAGE_INVERT */
-       170, /* GL_COLOR_MATRIX */
-       172, /* GL_COLOR_MATRIX_STACK_DEPTH */
-       802, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
-      1129, /* GL_POST_COLOR_MATRIX_RED_SCALE */
-      1125, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
-      1120, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
-      1116, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
-      1127, /* GL_POST_COLOR_MATRIX_RED_BIAS */
-      1123, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
-      1118, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
-      1114, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
-      1538, /* GL_TEXTURE_COLOR_TABLE_SGI */
-      1204, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
-      1540, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
-        80, /* GL_BLEND_DST_RGB */
-        88, /* GL_BLEND_SRC_RGB */
-        79, /* GL_BLEND_DST_ALPHA */
-        87, /* GL_BLEND_SRC_ALPHA */
-       176, /* GL_COLOR_TABLE */
-      1139, /* GL_POST_CONVOLUTION_COLOR_TABLE */
-      1122, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
-      1192, /* GL_PROXY_COLOR_TABLE */
-      1196, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
-      1195, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
-       200, /* GL_COLOR_TABLE_SCALE */
-       180, /* GL_COLOR_TABLE_BIAS */
-       185, /* GL_COLOR_TABLE_FORMAT */
-       202, /* GL_COLOR_TABLE_WIDTH */
-       197, /* GL_COLOR_TABLE_RED_SIZE */
-       188, /* GL_COLOR_TABLE_GREEN_SIZE */
-       182, /* GL_COLOR_TABLE_BLUE_SIZE */
-       177, /* GL_COLOR_TABLE_ALPHA_SIZE */
-       194, /* GL_COLOR_TABLE_LUMINANCE_SIZE */
-       191, /* GL_COLOR_TABLE_INTENSITY_SIZE */
-        71, /* GL_BGR */
-        72, /* GL_BGRA */
-       816, /* GL_MAX_ELEMENTS_VERTICES */
-       815, /* GL_MAX_ELEMENTS_INDICES */
-      1592, /* GL_TEXTURE_INDEX_SIZE_EXT */
-       136, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
-      1086, /* GL_POINT_SIZE_MIN */
-      1082, /* GL_POINT_SIZE_MAX */
-      1076, /* GL_POINT_FADE_THRESHOLD_SIZE */
-      1072, /* GL_POINT_DISTANCE_ATTENUATION */
-       118, /* GL_CLAMP_TO_BORDER */
-       121, /* GL_CLAMP_TO_EDGE */
-      1613, /* GL_TEXTURE_MIN_LOD */
-      1611, /* GL_TEXTURE_MAX_LOD */
-      1517, /* GL_TEXTURE_BASE_LEVEL */
-      1610, /* GL_TEXTURE_MAX_LEVEL */
-       572, /* GL_IGNORE_BORDER_HP */
-       246, /* GL_CONSTANT_BORDER_HP */
-      1250, /* GL_REPLICATE_BORDER_HP */
-       252, /* GL_CONVOLUTION_BORDER_COLOR */
-       987, /* GL_OCCLUSION_TEST_HP */
-       988, /* GL_OCCLUSION_TEST_RESULT_HP */
-       641, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
-      1532, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
-      1534, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
-      1536, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
-      1537, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1535, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
-      1533, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
-       798, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
-       799, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
-      1149, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
-      1151, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
-      1148, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
-      1150, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
-      1600, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
-      1601, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
-      1599, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
-       518, /* GL_GENERATE_MIPMAP */
-       519, /* GL_GENERATE_MIPMAP_HINT */
-       481, /* GL_FOG_OFFSET_SGIX */
-       482, /* GL_FOG_OFFSET_VALUE_SGIX */
-      1546, /* GL_TEXTURE_COMPARE_SGIX */
-      1545, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
-      1596, /* GL_TEXTURE_LEQUAL_R_SGIX */
-      1588, /* GL_TEXTURE_GEQUAL_R_SGIX */
-       323, /* GL_DEPTH_COMPONENT16 */
-       326, /* GL_DEPTH_COMPONENT24 */
-       329, /* GL_DEPTH_COMPONENT32 */
-       274, /* GL_CULL_VERTEX_EXT */
-       276, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
-       275, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
-      1741, /* GL_WRAP_BORDER_SUN */
-      1539, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
-       634, /* GL_LIGHT_MODEL_COLOR_CONTROL */
-      1345, /* GL_SINGLE_COLOR */
-      1333, /* GL_SEPARATE_SPECULAR_COLOR */
-      1342, /* GL_SHARED_TEXTURE_PALETTE_EXT */
-      1655, /* GL_UNSIGNED_BYTE_2_3_3_REV */
-      1668, /* GL_UNSIGNED_SHORT_5_6_5 */
-      1669, /* GL_UNSIGNED_SHORT_5_6_5_REV */
-      1666, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
-      1664, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
-      1662, /* GL_UNSIGNED_INT_8_8_8_8_REV */
-      1660, /* GL_UNSIGNED_INT_2_10_10_10_REV */
-      1608, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
-      1609, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
-      1607, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
-       889, /* GL_MIRRORED_REPEAT */
-      1289, /* GL_RGB_S3TC */
-      1266, /* GL_RGB4_S3TC */
-      1288, /* GL_RGBA_S3TC */
-      1283, /* GL_RGBA4_S3TC */
-      1286, /* GL_RGBA_DXT5_S3TC */
-      1281, /* GL_RGBA4_DXT5_S3TC */
-       239, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
-       234, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
-       235, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
-       236, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
-       950, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
-       949, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
-       642, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
-       468, /* GL_FOG_COORDINATE_SOURCE */
-       460, /* GL_FOG_COORD */
-       484, /* GL_FRAGMENT_DEPTH */
-       280, /* GL_CURRENT_FOG_COORD */
-       467, /* GL_FOG_COORDINATE_ARRAY_TYPE */
-       466, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
-       465, /* GL_FOG_COORDINATE_ARRAY_POINTER */
-       462, /* GL_FOG_COORDINATE_ARRAY */
-       174, /* GL_COLOR_SUM */
-       299, /* GL_CURRENT_SECONDARY_COLOR */
-      1326, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
-      1328, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
-      1327, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
-      1325, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
-      1322, /* GL_SECONDARY_COLOR_ARRAY */
-       528, /* GL_GL_CURRENT_RASTER_SECONDARY_COLOR */
+       677, /* GL_LUMINANCE4 */
+       683, /* GL_LUMINANCE8 */
+       667, /* GL_LUMINANCE12 */
+       673, /* GL_LUMINANCE16 */
+       678, /* GL_LUMINANCE4_ALPHA4 */
+       681, /* GL_LUMINANCE6_ALPHA2 */
+       684, /* GL_LUMINANCE8_ALPHA8 */
+       670, /* GL_LUMINANCE12_ALPHA4 */
+       668, /* GL_LUMINANCE12_ALPHA12 */
+       674, /* GL_LUMINANCE16_ALPHA16 */
+       591, /* GL_INTENSITY */
+       596, /* GL_INTENSITY4 */
+       598, /* GL_INTENSITY8 */
+       592, /* GL_INTENSITY12 */
+       594, /* GL_INTENSITY16 */
+      1261, /* GL_RGB2_EXT */
+      1262, /* GL_RGB4 */
+      1265, /* GL_RGB5 */
+      1269, /* GL_RGB8 */
+      1253, /* GL_RGB10 */
+      1257, /* GL_RGB12 */
+      1259, /* GL_RGB16 */
+      1276, /* GL_RGBA2 */
+      1278, /* GL_RGBA4 */
+      1266, /* GL_RGB5_A1 */
+      1282, /* GL_RGBA8 */
+      1254, /* GL_RGB10_A2 */
+      1272, /* GL_RGBA12 */
+      1274, /* GL_RGBA16 */
+      1615, /* GL_TEXTURE_RED_SIZE */
+      1587, /* GL_TEXTURE_GREEN_SIZE */
+      1526, /* GL_TEXTURE_BLUE_SIZE */
+      1513, /* GL_TEXTURE_ALPHA_SIZE */
+      1600, /* GL_TEXTURE_LUMINANCE_SIZE */
+      1591, /* GL_TEXTURE_INTENSITY_SIZE */
+      1247, /* GL_REPLACE_EXT */
+      1195, /* GL_PROXY_TEXTURE_1D */
+      1198, /* GL_PROXY_TEXTURE_2D */
+      1619, /* GL_TEXTURE_TOO_LARGE_EXT */
+      1612, /* GL_TEXTURE_PRIORITY */
+      1617, /* GL_TEXTURE_RESIDENT */
+      1516, /* GL_TEXTURE_BINDING_1D */
+      1518, /* GL_TEXTURE_BINDING_2D */
+      1520, /* GL_TEXTURE_BINDING_3D */
+      1026, /* GL_PACK_SKIP_IMAGES */
+      1022, /* GL_PACK_IMAGE_HEIGHT */
+      1648, /* GL_UNPACK_SKIP_IMAGES */
+      1645, /* GL_UNPACK_IMAGE_HEIGHT */
+      1512, /* GL_TEXTURE_3D */
+      1201, /* GL_PROXY_TEXTURE_3D */
+      1574, /* GL_TEXTURE_DEPTH */
+      1622, /* GL_TEXTURE_WRAP_R */
+       792, /* GL_MAX_3D_TEXTURE_SIZE */
+      1678, /* GL_VERTEX_ARRAY */
+       957, /* GL_NORMAL_ARRAY */
+       138, /* GL_COLOR_ARRAY */
+       576, /* GL_INDEX_ARRAY */
+      1553, /* GL_TEXTURE_COORD_ARRAY */
+       414, /* GL_EDGE_FLAG_ARRAY */
+      1683, /* GL_VERTEX_ARRAY_SIZE */
+      1685, /* GL_VERTEX_ARRAY_TYPE */
+      1684, /* GL_VERTEX_ARRAY_STRIDE */
+       962, /* GL_NORMAL_ARRAY_TYPE */
+       961, /* GL_NORMAL_ARRAY_STRIDE */
+       142, /* GL_COLOR_ARRAY_SIZE */
+       144, /* GL_COLOR_ARRAY_TYPE */
+       143, /* GL_COLOR_ARRAY_STRIDE */
+       581, /* GL_INDEX_ARRAY_TYPE */
+       580, /* GL_INDEX_ARRAY_STRIDE */
+      1557, /* GL_TEXTURE_COORD_ARRAY_SIZE */
+      1559, /* GL_TEXTURE_COORD_ARRAY_TYPE */
+      1558, /* GL_TEXTURE_COORD_ARRAY_STRIDE */
+       418, /* GL_EDGE_FLAG_ARRAY_STRIDE */
+      1682, /* GL_VERTEX_ARRAY_POINTER */
+       960, /* GL_NORMAL_ARRAY_POINTER */
+       141, /* GL_COLOR_ARRAY_POINTER */
+       579, /* GL_INDEX_ARRAY_POINTER */
+      1556, /* GL_TEXTURE_COORD_ARRAY_POINTER */
+       417, /* GL_EDGE_FLAG_ARRAY_POINTER */
+       936, /* GL_MULTISAMPLE */
+      1304, /* GL_SAMPLE_ALPHA_TO_COVERAGE */
+      1306, /* GL_SAMPLE_ALPHA_TO_ONE */
+      1311, /* GL_SAMPLE_COVERAGE */
+      1308, /* GL_SAMPLE_BUFFERS */
+      1299, /* GL_SAMPLES */
+      1315, /* GL_SAMPLE_COVERAGE_VALUE */
+      1313, /* GL_SAMPLE_COVERAGE_INVERT */
+       169, /* GL_COLOR_MATRIX */
+       171, /* GL_COLOR_MATRIX_STACK_DEPTH */
+       800, /* GL_MAX_COLOR_MATRIX_STACK_DEPTH */
+      1127, /* GL_POST_COLOR_MATRIX_RED_SCALE */
+      1123, /* GL_POST_COLOR_MATRIX_GREEN_SCALE */
+      1118, /* GL_POST_COLOR_MATRIX_BLUE_SCALE */
+      1114, /* GL_POST_COLOR_MATRIX_ALPHA_SCALE */
+      1125, /* GL_POST_COLOR_MATRIX_RED_BIAS */
+      1121, /* GL_POST_COLOR_MATRIX_GREEN_BIAS */
+      1116, /* GL_POST_COLOR_MATRIX_BLUE_BIAS */
+      1112, /* GL_POST_COLOR_MATRIX_ALPHA_BIAS */
+      1536, /* GL_TEXTURE_COLOR_TABLE_SGI */
+      1202, /* GL_PROXY_TEXTURE_COLOR_TABLE_SGI */
+      1538, /* GL_TEXTURE_COMPARE_FAIL_VALUE_ARB */
+        79, /* GL_BLEND_DST_RGB */
+        87, /* GL_BLEND_SRC_RGB */
+        78, /* GL_BLEND_DST_ALPHA */
+        86, /* GL_BLEND_SRC_ALPHA */
+       175, /* GL_COLOR_TABLE */
+      1137, /* GL_POST_CONVOLUTION_COLOR_TABLE */
+      1120, /* GL_POST_COLOR_MATRIX_COLOR_TABLE */
+      1190, /* GL_PROXY_COLOR_TABLE */
+      1194, /* GL_PROXY_POST_CONVOLUTION_COLOR_TABLE */
+      1193, /* GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE */
+       199, /* GL_COLOR_TABLE_SCALE */
+       179, /* GL_COLOR_TABLE_BIAS */
+       184, /* GL_COLOR_TABLE_FORMAT */
+       201, /* GL_COLOR_TABLE_WIDTH */
+       196, /* GL_COLOR_TABLE_RED_SIZE */
+       187, /* GL_COLOR_TABLE_GREEN_SIZE */
+       181, /* GL_COLOR_TABLE_BLUE_SIZE */
+       176, /* GL_COLOR_TABLE_ALPHA_SIZE */
+       193, /* GL_COLOR_TABLE_LUMINANCE_SIZE */
+       190, /* GL_COLOR_TABLE_INTENSITY_SIZE */
+        70, /* GL_BGR */
+        71, /* GL_BGRA */
+       814, /* GL_MAX_ELEMENTS_VERTICES */
+       813, /* GL_MAX_ELEMENTS_INDICES */
+      1590, /* GL_TEXTURE_INDEX_SIZE_EXT */
+       135, /* GL_CLIP_VOLUME_CLIPPING_HINT_EXT */
+      1084, /* GL_POINT_SIZE_MIN */
+      1080, /* GL_POINT_SIZE_MAX */
+      1074, /* GL_POINT_FADE_THRESHOLD_SIZE */
+      1070, /* GL_POINT_DISTANCE_ATTENUATION */
+       117, /* GL_CLAMP_TO_BORDER */
+       120, /* GL_CLAMP_TO_EDGE */
+      1611, /* GL_TEXTURE_MIN_LOD */
+      1609, /* GL_TEXTURE_MAX_LOD */
+      1515, /* GL_TEXTURE_BASE_LEVEL */
+      1608, /* GL_TEXTURE_MAX_LEVEL */
+       570, /* GL_IGNORE_BORDER_HP */
+       245, /* GL_CONSTANT_BORDER_HP */
+      1248, /* GL_REPLICATE_BORDER_HP */
+       251, /* GL_CONVOLUTION_BORDER_COLOR */
+       985, /* GL_OCCLUSION_TEST_HP */
+       986, /* GL_OCCLUSION_TEST_RESULT_HP */
+       639, /* GL_LINEAR_CLIPMAP_LINEAR_SGIX */
+      1530, /* GL_TEXTURE_CLIPMAP_CENTER_SGIX */
+      1532, /* GL_TEXTURE_CLIPMAP_FRAME_SGIX */
+      1534, /* GL_TEXTURE_CLIPMAP_OFFSET_SGIX */
+      1535, /* GL_TEXTURE_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1533, /* GL_TEXTURE_CLIPMAP_LOD_OFFSET_SGIX */
+      1531, /* GL_TEXTURE_CLIPMAP_DEPTH_SGIX */
+       796, /* GL_MAX_CLIPMAP_DEPTH_SGIX */
+       797, /* GL_MAX_CLIPMAP_VIRTUAL_DEPTH_SGIX */
+      1147, /* GL_POST_TEXTURE_FILTER_BIAS_SGIX */
+      1149, /* GL_POST_TEXTURE_FILTER_SCALE_SGIX */
+      1146, /* GL_POST_TEXTURE_FILTER_BIAS_RANGE_SGIX */
+      1148, /* GL_POST_TEXTURE_FILTER_SCALE_RANGE_SGIX */
+      1598, /* GL_TEXTURE_LOD_BIAS_S_SGIX */
+      1599, /* GL_TEXTURE_LOD_BIAS_T_SGIX */
+      1597, /* GL_TEXTURE_LOD_BIAS_R_SGIX */
+       516, /* GL_GENERATE_MIPMAP */
+       517, /* GL_GENERATE_MIPMAP_HINT */
+       479, /* GL_FOG_OFFSET_SGIX */
+       480, /* GL_FOG_OFFSET_VALUE_SGIX */
+      1544, /* GL_TEXTURE_COMPARE_SGIX */
+      1543, /* GL_TEXTURE_COMPARE_OPERATOR_SGIX */
+      1594, /* GL_TEXTURE_LEQUAL_R_SGIX */
+      1586, /* GL_TEXTURE_GEQUAL_R_SGIX */
+       322, /* GL_DEPTH_COMPONENT16 */
+       325, /* GL_DEPTH_COMPONENT24 */
+       328, /* GL_DEPTH_COMPONENT32 */
+       273, /* GL_CULL_VERTEX_EXT */
+       275, /* GL_CULL_VERTEX_OBJECT_POSITION_EXT */
+       274, /* GL_CULL_VERTEX_EYE_POSITION_EXT */
+      1739, /* GL_WRAP_BORDER_SUN */
+      1537, /* GL_TEXTURE_COLOR_WRITEMASK_SGIS */
+       632, /* GL_LIGHT_MODEL_COLOR_CONTROL */
+      1343, /* GL_SINGLE_COLOR */
+      1331, /* GL_SEPARATE_SPECULAR_COLOR */
+      1340, /* GL_SHARED_TEXTURE_PALETTE_EXT */
+      1653, /* GL_UNSIGNED_BYTE_2_3_3_REV */
+      1666, /* GL_UNSIGNED_SHORT_5_6_5 */
+      1667, /* GL_UNSIGNED_SHORT_5_6_5_REV */
+      1664, /* GL_UNSIGNED_SHORT_4_4_4_4_REV */
+      1662, /* GL_UNSIGNED_SHORT_1_5_5_5_REV */
+      1660, /* GL_UNSIGNED_INT_8_8_8_8_REV */
+      1658, /* GL_UNSIGNED_INT_2_10_10_10_REV */
+      1606, /* GL_TEXTURE_MAX_CLAMP_S_SGIX */
+      1607, /* GL_TEXTURE_MAX_CLAMP_T_SGIX */
+      1605, /* GL_TEXTURE_MAX_CLAMP_R_SGIX */
+       887, /* GL_MIRRORED_REPEAT */
+      1287, /* GL_RGB_S3TC */
+      1264, /* GL_RGB4_S3TC */
+      1286, /* GL_RGBA_S3TC */
+      1281, /* GL_RGBA4_S3TC */
+      1284, /* GL_RGBA_DXT5_S3TC */
+      1279, /* GL_RGBA4_DXT5_S3TC */
+       238, /* GL_COMPRESSED_RGB_S3TC_DXT1_EXT */
+       233, /* GL_COMPRESSED_RGBA_S3TC_DXT1_EXT */
+       234, /* GL_COMPRESSED_RGBA_S3TC_DXT3_EXT */
+       235, /* GL_COMPRESSED_RGBA_S3TC_DXT5_EXT */
+       948, /* GL_NEAREST_CLIPMAP_NEAREST_SGIX */
+       947, /* GL_NEAREST_CLIPMAP_LINEAR_SGIX */
+       640, /* GL_LINEAR_CLIPMAP_NEAREST_SGIX */
+       466, /* GL_FOG_COORDINATE_SOURCE */
+       458, /* GL_FOG_COORD */
+       482, /* GL_FRAGMENT_DEPTH */
+       279, /* GL_CURRENT_FOG_COORD */
+       465, /* GL_FOG_COORDINATE_ARRAY_TYPE */
+       464, /* GL_FOG_COORDINATE_ARRAY_STRIDE */
+       463, /* GL_FOG_COORDINATE_ARRAY_POINTER */
+       460, /* GL_FOG_COORDINATE_ARRAY */
+       173, /* GL_COLOR_SUM */
+       298, /* GL_CURRENT_SECONDARY_COLOR */
+      1324, /* GL_SECONDARY_COLOR_ARRAY_SIZE */
+      1326, /* GL_SECONDARY_COLOR_ARRAY_TYPE */
+      1325, /* GL_SECONDARY_COLOR_ARRAY_STRIDE */
+      1323, /* GL_SECONDARY_COLOR_ARRAY_POINTER */
+      1320, /* GL_SECONDARY_COLOR_ARRAY */
+       526, /* GL_GL_CURRENT_RASTER_SECONDARY_COLOR */
         28, /* GL_ALIASED_POINT_SIZE_RANGE */
         27, /* GL_ALIASED_LINE_WIDTH_RANGE */
-      1446, /* GL_TEXTURE0 */
-      1448, /* GL_TEXTURE1 */
-      1470, /* GL_TEXTURE2 */
-      1492, /* GL_TEXTURE3 */
-      1498, /* GL_TEXTURE4 */
-      1500, /* GL_TEXTURE5 */
-      1502, /* GL_TEXTURE6 */
-      1504, /* GL_TEXTURE7 */
-      1506, /* GL_TEXTURE8 */
-      1508, /* GL_TEXTURE9 */
-      1449, /* GL_TEXTURE10 */
-      1451, /* GL_TEXTURE11 */
-      1453, /* GL_TEXTURE12 */
-      1455, /* GL_TEXTURE13 */
-      1457, /* GL_TEXTURE14 */
-      1459, /* GL_TEXTURE15 */
-      1461, /* GL_TEXTURE16 */
-      1463, /* GL_TEXTURE17 */
-      1465, /* GL_TEXTURE18 */
-      1467, /* GL_TEXTURE19 */
-      1471, /* GL_TEXTURE20 */
-      1473, /* GL_TEXTURE21 */
-      1475, /* GL_TEXTURE22 */
-      1477, /* GL_TEXTURE23 */
-      1479, /* GL_TEXTURE24 */
-      1481, /* GL_TEXTURE25 */
-      1483, /* GL_TEXTURE26 */
-      1485, /* GL_TEXTURE27 */
-      1487, /* GL_TEXTURE28 */
-      1489, /* GL_TEXTURE29 */
-      1493, /* GL_TEXTURE30 */
-      1495, /* GL_TEXTURE31 */
+      1444, /* GL_TEXTURE0 */
+      1446, /* GL_TEXTURE1 */
+      1468, /* GL_TEXTURE2 */
+      1490, /* GL_TEXTURE3 */
+      1496, /* GL_TEXTURE4 */
+      1498, /* GL_TEXTURE5 */
+      1500, /* GL_TEXTURE6 */
+      1502, /* GL_TEXTURE7 */
+      1504, /* GL_TEXTURE8 */
+      1506, /* GL_TEXTURE9 */
+      1447, /* GL_TEXTURE10 */
+      1449, /* GL_TEXTURE11 */
+      1451, /* GL_TEXTURE12 */
+      1453, /* GL_TEXTURE13 */
+      1455, /* GL_TEXTURE14 */
+      1457, /* GL_TEXTURE15 */
+      1459, /* GL_TEXTURE16 */
+      1461, /* GL_TEXTURE17 */
+      1463, /* GL_TEXTURE18 */
+      1465, /* GL_TEXTURE19 */
+      1469, /* GL_TEXTURE20 */
+      1471, /* GL_TEXTURE21 */
+      1473, /* GL_TEXTURE22 */
+      1475, /* GL_TEXTURE23 */
+      1477, /* GL_TEXTURE24 */
+      1479, /* GL_TEXTURE25 */
+      1481, /* GL_TEXTURE26 */
+      1483, /* GL_TEXTURE27 */
+      1485, /* GL_TEXTURE28 */
+      1487, /* GL_TEXTURE29 */
+      1491, /* GL_TEXTURE30 */
+      1493, /* GL_TEXTURE31 */
         18, /* GL_ACTIVE_TEXTURE */
-       124, /* GL_CLIENT_ACTIVE_TEXTURE */
-       867, /* GL_MAX_TEXTURE_UNITS */
-      1634, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
-      1637, /* GL_TRANSPOSE_PROJECTION_MATRIX */
-      1639, /* GL_TRANSPOSE_TEXTURE_MATRIX */
-      1631, /* GL_TRANSPOSE_COLOR_MATRIX */
-      1434, /* GL_SUBTRACT */
-       856, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
-       222, /* GL_COMPRESSED_ALPHA */
-       226, /* GL_COMPRESSED_LUMINANCE */
-       227, /* GL_COMPRESSED_LUMINANCE_ALPHA */
-       224, /* GL_COMPRESSED_INTENSITY */
-       230, /* GL_COMPRESSED_RGB */
-       231, /* GL_COMPRESSED_RGBA */
-      1553, /* GL_TEXTURE_COMPRESSION_HINT */
-      1615, /* GL_TEXTURE_RECTANGLE_ARB */
-      1525, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
-      1207, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
-       854, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
-       335, /* GL_DEPTH_STENCIL_NV */
-      1659, /* GL_UNSIGNED_INT_24_8_NV */
-       863, /* GL_MAX_TEXTURE_LOD_BIAS */
-      1606, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
-       864, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
-      1582, /* GL_TEXTURE_FILTER_CONTROL */
-      1597, /* GL_TEXTURE_LOD_BIAS */
-       207, /* GL_COMBINE4 */
-       857, /* GL_MAX_SHININESS_NV */
-       858, /* GL_MAX_SPOT_EXPONENT_NV */
-       576, /* GL_INCR_WRAP */
-       310, /* GL_DECR_WRAP */
-       909, /* GL_MODELVIEW1_ARB */
-       965, /* GL_NORMAL_MAP */
-      1236, /* GL_REFLECTION_MAP */
-      1562, /* GL_TEXTURE_CUBE_MAP */
-      1523, /* GL_TEXTURE_BINDING_CUBE_MAP */
-      1570, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
-      1564, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
-      1572, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
-      1566, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
-      1574, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
-      1568, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
-      1205, /* GL_PROXY_TEXTURE_CUBE_MAP */
-       810, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
-       944, /* GL_MULTISAMPLE_FILTER_HINT_NV */
-       476, /* GL_FOG_DISTANCE_MODE_NV */
-       435, /* GL_EYE_RADIAL_NV */
-       434, /* GL_EYE_PLANE_ABSOLUTE_NV */
-       206, /* GL_COMBINE */
-       213, /* GL_COMBINE_RGB */
-       208, /* GL_COMBINE_ALPHA */
-      1290, /* GL_RGB_SCALE */
+       123, /* GL_CLIENT_ACTIVE_TEXTURE */
+       865, /* GL_MAX_TEXTURE_UNITS */
+      1632, /* GL_TRANSPOSE_MODELVIEW_MATRIX */
+      1635, /* GL_TRANSPOSE_PROJECTION_MATRIX */
+      1637, /* GL_TRANSPOSE_TEXTURE_MATRIX */
+      1629, /* GL_TRANSPOSE_COLOR_MATRIX */
+      1432, /* GL_SUBTRACT */
+       854, /* GL_MAX_RENDERBUFFER_SIZE_EXT */
+       221, /* GL_COMPRESSED_ALPHA */
+       225, /* GL_COMPRESSED_LUMINANCE */
+       226, /* GL_COMPRESSED_LUMINANCE_ALPHA */
+       223, /* GL_COMPRESSED_INTENSITY */
+       229, /* GL_COMPRESSED_RGB */
+       230, /* GL_COMPRESSED_RGBA */
+      1551, /* GL_TEXTURE_COMPRESSION_HINT */
+      1613, /* GL_TEXTURE_RECTANGLE_ARB */
+      1523, /* GL_TEXTURE_BINDING_RECTANGLE_ARB */
+      1205, /* GL_PROXY_TEXTURE_RECTANGLE_ARB */
+       852, /* GL_MAX_RECTANGLE_TEXTURE_SIZE_ARB */
+       334, /* GL_DEPTH_STENCIL_NV */
+      1657, /* GL_UNSIGNED_INT_24_8_NV */
+       861, /* GL_MAX_TEXTURE_LOD_BIAS */
+      1604, /* GL_TEXTURE_MAX_ANISOTROPY_EXT */
+       862, /* GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT */
+      1580, /* GL_TEXTURE_FILTER_CONTROL */
+      1595, /* GL_TEXTURE_LOD_BIAS */
+       206, /* GL_COMBINE4 */
+       855, /* GL_MAX_SHININESS_NV */
+       856, /* GL_MAX_SPOT_EXPONENT_NV */
+       574, /* GL_INCR_WRAP */
+       309, /* GL_DECR_WRAP */
+       907, /* GL_MODELVIEW1_ARB */
+       963, /* GL_NORMAL_MAP */
+      1234, /* GL_REFLECTION_MAP */
+      1560, /* GL_TEXTURE_CUBE_MAP */
+      1521, /* GL_TEXTURE_BINDING_CUBE_MAP */
+      1568, /* GL_TEXTURE_CUBE_MAP_POSITIVE_X */
+      1562, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_X */
+      1570, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Y */
+      1564, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Y */
+      1572, /* GL_TEXTURE_CUBE_MAP_POSITIVE_Z */
+      1566, /* GL_TEXTURE_CUBE_MAP_NEGATIVE_Z */
+      1203, /* GL_PROXY_TEXTURE_CUBE_MAP */
+       808, /* GL_MAX_CUBE_MAP_TEXTURE_SIZE */
+       942, /* GL_MULTISAMPLE_FILTER_HINT_NV */
+       474, /* GL_FOG_DISTANCE_MODE_NV */
+       433, /* GL_EYE_RADIAL_NV */
+       432, /* GL_EYE_PLANE_ABSOLUTE_NV */
+       205, /* GL_COMBINE */
+       212, /* GL_COMBINE_RGB */
+       207, /* GL_COMBINE_ALPHA */
+      1288, /* GL_RGB_SCALE */
         24, /* GL_ADD_SIGNED */
-       603, /* GL_INTERPOLATE */
-       241, /* GL_CONSTANT */
-      1155, /* GL_PRIMARY_COLOR */
-      1152, /* GL_PREVIOUS */
-      1356, /* GL_SOURCE0_RGB */
-      1362, /* GL_SOURCE1_RGB */
-      1368, /* GL_SOURCE2_RGB */
-      1372, /* GL_SOURCE3_RGB_NV */
-      1353, /* GL_SOURCE0_ALPHA */
-      1359, /* GL_SOURCE1_ALPHA */
-      1365, /* GL_SOURCE2_ALPHA */
-      1371, /* GL_SOURCE3_ALPHA_NV */
-      1001, /* GL_OPERAND0_RGB */
-      1007, /* GL_OPERAND1_RGB */
-      1013, /* GL_OPERAND2_RGB */
-      1017, /* GL_OPERAND3_RGB_NV */
-       998, /* GL_OPERAND0_ALPHA */
-      1004, /* GL_OPERAND1_ALPHA */
-      1010, /* GL_OPERAND2_ALPHA */
-      1016, /* GL_OPERAND3_ALPHA_NV */
-      1681, /* GL_VERTEX_ARRAY_BINDING_APPLE */
-      1745, /* GL_YCBCR_422_APPLE */
-      1670, /* GL_UNSIGNED_SHORT_8_8_APPLE */
-      1672, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
-      1347, /* GL_SLICE_ACCUM_SUN */
-      1212, /* GL_QUAD_MESH_SUN */
-      1643, /* GL_TRIANGLE_MESH_SUN */
-      1719, /* GL_VERTEX_PROGRAM_ARB */
-      1730, /* GL_VERTEX_STATE_PROGRAM_NV */
-      1706, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
-      1712, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
-      1714, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
-      1716, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
-       301, /* GL_CURRENT_VERTEX_ATTRIB */
-      1168, /* GL_PROGRAM_LENGTH_ARB */
-      1182, /* GL_PROGRAM_STRING_ARB */
-       931, /* GL_MODELVIEW_PROJECTION_NV */
-       571, /* GL_IDENTITY_NV */
-       616, /* GL_INVERSE_NV */
-      1636, /* GL_TRANSPOSE_NV */
-       617, /* GL_INVERSE_TRANSPOSE_NV */
-       840, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
-       839, /* GL_MAX_PROGRAM_MATRICES_ARB */
-       747, /* GL_MATRIX0_NV */
-       759, /* GL_MATRIX1_NV */
-       771, /* GL_MATRIX2_NV */
-       775, /* GL_MATRIX3_NV */
-       777, /* GL_MATRIX4_NV */
-       779, /* GL_MATRIX5_NV */
-       781, /* GL_MATRIX6_NV */
-       783, /* GL_MATRIX7_NV */
-       286, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
-       283, /* GL_CURRENT_MATRIX_ARB */
-      1722, /* GL_VERTEX_PROGRAM_POINT_SIZE */
-      1725, /* GL_VERTEX_PROGRAM_TWO_SIDE */
-      1180, /* GL_PROGRAM_PARAMETER_NV */
-      1710, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
-      1184, /* GL_PROGRAM_TARGET_NV */
-      1181, /* GL_PROGRAM_RESIDENT_NV */
-      1628, /* GL_TRACK_MATRIX_NV */
-      1629, /* GL_TRACK_MATRIX_TRANSFORM_NV */
-      1720, /* GL_VERTEX_PROGRAM_BINDING_NV */
-      1162, /* GL_PROGRAM_ERROR_POSITION_ARB */
-       320, /* GL_DEPTH_CLAMP_NV */
-      1688, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
-      1695, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
-      1696, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
-      1697, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
-      1698, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
-      1699, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
-      1700, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
-      1701, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
-      1702, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
-      1703, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
-      1689, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
-      1690, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
-      1691, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
-      1692, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
-      1693, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
-      1694, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
-       701, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
-       708, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
-       709, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
-       710, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
-       711, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
-       712, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
-       713, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
-       714, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
-       715, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
-       716, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
-       702, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
-       703, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
-       704, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
-       705, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
-       706, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
-       707, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
-       728, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
-       735, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
-       736, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
-       737, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
-       738, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
-       739, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
-       740, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
-      1161, /* GL_PROGRAM_BINDING_ARB */
-       742, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
-       743, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
-       729, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
-       730, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
-       731, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
-       732, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
-       733, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
-       734, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
-      1551, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
-      1548, /* GL_TEXTURE_COMPRESSED */
-       970, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
-       240, /* GL_COMPRESSED_TEXTURE_FORMATS */
-       879, /* GL_MAX_VERTEX_UNITS_ARB */
+       601, /* GL_INTERPOLATE */
+       240, /* GL_CONSTANT */
+      1153, /* GL_PRIMARY_COLOR */
+      1150, /* GL_PREVIOUS */
+      1354, /* GL_SOURCE0_RGB */
+      1360, /* GL_SOURCE1_RGB */
+      1366, /* GL_SOURCE2_RGB */
+      1370, /* GL_SOURCE3_RGB_NV */
+      1351, /* GL_SOURCE0_ALPHA */
+      1357, /* GL_SOURCE1_ALPHA */
+      1363, /* GL_SOURCE2_ALPHA */
+      1369, /* GL_SOURCE3_ALPHA_NV */
+       999, /* GL_OPERAND0_RGB */
+      1005, /* GL_OPERAND1_RGB */
+      1011, /* GL_OPERAND2_RGB */
+      1015, /* GL_OPERAND3_RGB_NV */
+       996, /* GL_OPERAND0_ALPHA */
+      1002, /* GL_OPERAND1_ALPHA */
+      1008, /* GL_OPERAND2_ALPHA */
+      1014, /* GL_OPERAND3_ALPHA_NV */
+      1679, /* GL_VERTEX_ARRAY_BINDING_APPLE */
+      1743, /* GL_YCBCR_422_APPLE */
+      1668, /* GL_UNSIGNED_SHORT_8_8_APPLE */
+      1670, /* GL_UNSIGNED_SHORT_8_8_REV_APPLE */
+      1345, /* GL_SLICE_ACCUM_SUN */
+      1210, /* GL_QUAD_MESH_SUN */
+      1641, /* GL_TRIANGLE_MESH_SUN */
+      1717, /* GL_VERTEX_PROGRAM_ARB */
+      1728, /* GL_VERTEX_STATE_PROGRAM_NV */
+      1704, /* GL_VERTEX_ATTRIB_ARRAY_ENABLED */
+      1710, /* GL_VERTEX_ATTRIB_ARRAY_SIZE */
+      1712, /* GL_VERTEX_ATTRIB_ARRAY_STRIDE */
+      1714, /* GL_VERTEX_ATTRIB_ARRAY_TYPE */
+       300, /* GL_CURRENT_VERTEX_ATTRIB */
+      1166, /* GL_PROGRAM_LENGTH_ARB */
+      1180, /* GL_PROGRAM_STRING_ARB */
+       929, /* GL_MODELVIEW_PROJECTION_NV */
+       569, /* GL_IDENTITY_NV */
+       614, /* GL_INVERSE_NV */
+      1634, /* GL_TRANSPOSE_NV */
+       615, /* GL_INVERSE_TRANSPOSE_NV */
+       838, /* GL_MAX_PROGRAM_MATRIX_STACK_DEPTH_ARB */
+       837, /* GL_MAX_PROGRAM_MATRICES_ARB */
+       745, /* GL_MATRIX0_NV */
+       757, /* GL_MATRIX1_NV */
+       769, /* GL_MATRIX2_NV */
+       773, /* GL_MATRIX3_NV */
+       775, /* GL_MATRIX4_NV */
+       777, /* GL_MATRIX5_NV */
+       779, /* GL_MATRIX6_NV */
+       781, /* GL_MATRIX7_NV */
+       285, /* GL_CURRENT_MATRIX_STACK_DEPTH_ARB */
+       282, /* GL_CURRENT_MATRIX_ARB */
+      1720, /* GL_VERTEX_PROGRAM_POINT_SIZE */
+      1723, /* GL_VERTEX_PROGRAM_TWO_SIDE */
+      1178, /* GL_PROGRAM_PARAMETER_NV */
+      1708, /* GL_VERTEX_ATTRIB_ARRAY_POINTER */
+      1182, /* GL_PROGRAM_TARGET_NV */
+      1179, /* GL_PROGRAM_RESIDENT_NV */
+      1626, /* GL_TRACK_MATRIX_NV */
+      1627, /* GL_TRACK_MATRIX_TRANSFORM_NV */
+      1718, /* GL_VERTEX_PROGRAM_BINDING_NV */
+      1160, /* GL_PROGRAM_ERROR_POSITION_ARB */
+       319, /* GL_DEPTH_CLAMP_NV */
+      1686, /* GL_VERTEX_ATTRIB_ARRAY0_NV */
+      1693, /* GL_VERTEX_ATTRIB_ARRAY1_NV */
+      1694, /* GL_VERTEX_ATTRIB_ARRAY2_NV */
+      1695, /* GL_VERTEX_ATTRIB_ARRAY3_NV */
+      1696, /* GL_VERTEX_ATTRIB_ARRAY4_NV */
+      1697, /* GL_VERTEX_ATTRIB_ARRAY5_NV */
+      1698, /* GL_VERTEX_ATTRIB_ARRAY6_NV */
+      1699, /* GL_VERTEX_ATTRIB_ARRAY7_NV */
+      1700, /* GL_VERTEX_ATTRIB_ARRAY8_NV */
+      1701, /* GL_VERTEX_ATTRIB_ARRAY9_NV */
+      1687, /* GL_VERTEX_ATTRIB_ARRAY10_NV */
+      1688, /* GL_VERTEX_ATTRIB_ARRAY11_NV */
+      1689, /* GL_VERTEX_ATTRIB_ARRAY12_NV */
+      1690, /* GL_VERTEX_ATTRIB_ARRAY13_NV */
+      1691, /* GL_VERTEX_ATTRIB_ARRAY14_NV */
+      1692, /* GL_VERTEX_ATTRIB_ARRAY15_NV */
+       699, /* GL_MAP1_VERTEX_ATTRIB0_4_NV */
+       706, /* GL_MAP1_VERTEX_ATTRIB1_4_NV */
+       707, /* GL_MAP1_VERTEX_ATTRIB2_4_NV */
+       708, /* GL_MAP1_VERTEX_ATTRIB3_4_NV */
+       709, /* GL_MAP1_VERTEX_ATTRIB4_4_NV */
+       710, /* GL_MAP1_VERTEX_ATTRIB5_4_NV */
+       711, /* GL_MAP1_VERTEX_ATTRIB6_4_NV */
+       712, /* GL_MAP1_VERTEX_ATTRIB7_4_NV */
+       713, /* GL_MAP1_VERTEX_ATTRIB8_4_NV */
+       714, /* GL_MAP1_VERTEX_ATTRIB9_4_NV */
+       700, /* GL_MAP1_VERTEX_ATTRIB10_4_NV */
+       701, /* GL_MAP1_VERTEX_ATTRIB11_4_NV */
+       702, /* GL_MAP1_VERTEX_ATTRIB12_4_NV */
+       703, /* GL_MAP1_VERTEX_ATTRIB13_4_NV */
+       704, /* GL_MAP1_VERTEX_ATTRIB14_4_NV */
+       705, /* GL_MAP1_VERTEX_ATTRIB15_4_NV */
+       726, /* GL_MAP2_VERTEX_ATTRIB0_4_NV */
+       733, /* GL_MAP2_VERTEX_ATTRIB1_4_NV */
+       734, /* GL_MAP2_VERTEX_ATTRIB2_4_NV */
+       735, /* GL_MAP2_VERTEX_ATTRIB3_4_NV */
+       736, /* GL_MAP2_VERTEX_ATTRIB4_4_NV */
+       737, /* GL_MAP2_VERTEX_ATTRIB5_4_NV */
+       738, /* GL_MAP2_VERTEX_ATTRIB6_4_NV */
+      1159, /* GL_PROGRAM_BINDING_ARB */
+       740, /* GL_MAP2_VERTEX_ATTRIB8_4_NV */
+       741, /* GL_MAP2_VERTEX_ATTRIB9_4_NV */
+       727, /* GL_MAP2_VERTEX_ATTRIB10_4_NV */
+       728, /* GL_MAP2_VERTEX_ATTRIB11_4_NV */
+       729, /* GL_MAP2_VERTEX_ATTRIB12_4_NV */
+       730, /* GL_MAP2_VERTEX_ATTRIB13_4_NV */
+       731, /* GL_MAP2_VERTEX_ATTRIB14_4_NV */
+       732, /* GL_MAP2_VERTEX_ATTRIB15_4_NV */
+      1549, /* GL_TEXTURE_COMPRESSED_IMAGE_SIZE */
+      1546, /* GL_TEXTURE_COMPRESSED */
+       968, /* GL_NUM_COMPRESSED_TEXTURE_FORMATS */
+       239, /* GL_COMPRESSED_TEXTURE_FORMATS */
+       877, /* GL_MAX_VERTEX_UNITS_ARB */
         22, /* GL_ACTIVE_VERTEX_UNITS_ARB */
-      1740, /* GL_WEIGHT_SUM_UNITY_ARB */
-      1718, /* GL_VERTEX_BLEND_ARB */
-       303, /* GL_CURRENT_WEIGHT_ARB */
-      1739, /* GL_WEIGHT_ARRAY_TYPE_ARB */
-      1738, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
-      1737, /* GL_WEIGHT_ARRAY_SIZE_ARB */
-      1736, /* GL_WEIGHT_ARRAY_POINTER_ARB */
-      1733, /* GL_WEIGHT_ARRAY_ARB */
-       346, /* GL_DOT3_RGB */
-       347, /* GL_DOT3_RGBA */
-       238, /* GL_COMPRESSED_RGB_FXT1_3DFX */
-       233, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
-       939, /* GL_MULTISAMPLE_3DFX */
-      1311, /* GL_SAMPLE_BUFFERS_3DFX */
-      1302, /* GL_SAMPLES_3DFX */
-       920, /* GL_MODELVIEW2_ARB */
-       923, /* GL_MODELVIEW3_ARB */
-       924, /* GL_MODELVIEW4_ARB */
-       925, /* GL_MODELVIEW5_ARB */
-       926, /* GL_MODELVIEW6_ARB */
-       927, /* GL_MODELVIEW7_ARB */
-       928, /* GL_MODELVIEW8_ARB */
-       929, /* GL_MODELVIEW9_ARB */
-       899, /* GL_MODELVIEW10_ARB */
-       900, /* GL_MODELVIEW11_ARB */
-       901, /* GL_MODELVIEW12_ARB */
-       902, /* GL_MODELVIEW13_ARB */
-       903, /* GL_MODELVIEW14_ARB */
-       904, /* GL_MODELVIEW15_ARB */
-       905, /* GL_MODELVIEW16_ARB */
-       906, /* GL_MODELVIEW17_ARB */
-       907, /* GL_MODELVIEW18_ARB */
-       908, /* GL_MODELVIEW19_ARB */
-       910, /* GL_MODELVIEW20_ARB */
-       911, /* GL_MODELVIEW21_ARB */
-       912, /* GL_MODELVIEW22_ARB */
-       913, /* GL_MODELVIEW23_ARB */
-       914, /* GL_MODELVIEW24_ARB */
-       915, /* GL_MODELVIEW25_ARB */
-       916, /* GL_MODELVIEW26_ARB */
-       917, /* GL_MODELVIEW27_ARB */
-       918, /* GL_MODELVIEW28_ARB */
-       919, /* GL_MODELVIEW29_ARB */
-       921, /* GL_MODELVIEW30_ARB */
-       922, /* GL_MODELVIEW31_ARB */
-       351, /* GL_DOT3_RGB_EXT */
-       349, /* GL_DOT3_RGBA_EXT */
-       893, /* GL_MIRROR_CLAMP_EXT */
-       896, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
-       934, /* GL_MODULATE_ADD_ATI */
-       935, /* GL_MODULATE_SIGNED_ADD_ATI */
-       936, /* GL_MODULATE_SUBTRACT_ATI */
-      1746, /* GL_YCBCR_MESA */
-      1025, /* GL_PACK_INVERT_MESA */
-       306, /* GL_DEBUG_OBJECT_MESA */
-       307, /* GL_DEBUG_PRINT_MESA */
-       305, /* GL_DEBUG_ASSERT_MESA */
-       107, /* GL_BUFFER_SIZE */
-       109, /* GL_BUFFER_USAGE */
-      1399, /* GL_STENCIL_BACK_FUNC */
-      1397, /* GL_STENCIL_BACK_FAIL */
-      1401, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
-      1403, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
-       485, /* GL_FRAGMENT_PROGRAM_ARB */
-      1159, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
-      1187, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
-      1186, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
-      1171, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-      1177, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-      1176, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       829, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
-       852, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
-       851, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
-       842, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
-       848, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
-       847, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
-       812, /* GL_MAX_DRAW_BUFFERS */
-       355, /* GL_DRAW_BUFFER0 */
-       358, /* GL_DRAW_BUFFER1 */
-       379, /* GL_DRAW_BUFFER2 */
-       382, /* GL_DRAW_BUFFER3 */
-       385, /* GL_DRAW_BUFFER4 */
-       388, /* GL_DRAW_BUFFER5 */
-       391, /* GL_DRAW_BUFFER6 */
-       394, /* GL_DRAW_BUFFER7 */
-       397, /* GL_DRAW_BUFFER8 */
-       400, /* GL_DRAW_BUFFER9 */
-       359, /* GL_DRAW_BUFFER10 */
-       362, /* GL_DRAW_BUFFER11 */
-       365, /* GL_DRAW_BUFFER12 */
-       368, /* GL_DRAW_BUFFER13 */
-       371, /* GL_DRAW_BUFFER14 */
-       374, /* GL_DRAW_BUFFER15 */
-        82, /* GL_BLEND_EQUATION_ALPHA */
-       792, /* GL_MATRIX_PALETTE_ARB */
-       823, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
-       826, /* GL_MAX_PALETTE_MATRICES_ARB */
-       289, /* GL_CURRENT_PALETTE_MATRIX_ARB */
-       786, /* GL_MATRIX_INDEX_ARRAY_ARB */
-       284, /* GL_CURRENT_MATRIX_INDEX_ARB */
-       788, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
-       790, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
-       789, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
-       787, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
-      1577, /* GL_TEXTURE_DEPTH_SIZE */
-       339, /* GL_DEPTH_TEXTURE_MODE */
-      1543, /* GL_TEXTURE_COMPARE_MODE */
-      1541, /* GL_TEXTURE_COMPARE_FUNC */
-       217, /* GL_COMPARE_R_TO_TEXTURE */
-      1093, /* GL_POINT_SPRITE */
-       266, /* GL_COORD_REPLACE */
-      1097, /* GL_POINT_SPRITE_R_MODE_NV */
-      1214, /* GL_QUERY_COUNTER_BITS */
-       291, /* GL_CURRENT_QUERY */
-      1216, /* GL_QUERY_RESULT */
-      1218, /* GL_QUERY_RESULT_AVAILABLE */
-       873, /* GL_MAX_VERTEX_ATTRIBS */
-      1708, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
-       337, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
-       336, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
-       859, /* GL_MAX_TEXTURE_COORDS */
-       861, /* GL_MAX_TEXTURE_IMAGE_UNITS */
-      1164, /* GL_PROGRAM_ERROR_STRING_ARB */
-      1166, /* GL_PROGRAM_FORMAT_ASCII_ARB */
-      1165, /* GL_PROGRAM_FORMAT_ARB */
-      1622, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
-       318, /* GL_DEPTH_BOUNDS_TEST_EXT */
-       317, /* GL_DEPTH_BOUNDS_EXT */
+      1738, /* GL_WEIGHT_SUM_UNITY_ARB */
+      1716, /* GL_VERTEX_BLEND_ARB */
+       302, /* GL_CURRENT_WEIGHT_ARB */
+      1737, /* GL_WEIGHT_ARRAY_TYPE_ARB */
+      1736, /* GL_WEIGHT_ARRAY_STRIDE_ARB */
+      1735, /* GL_WEIGHT_ARRAY_SIZE_ARB */
+      1734, /* GL_WEIGHT_ARRAY_POINTER_ARB */
+      1731, /* GL_WEIGHT_ARRAY_ARB */
+       345, /* GL_DOT3_RGB */
+       346, /* GL_DOT3_RGBA */
+       237, /* GL_COMPRESSED_RGB_FXT1_3DFX */
+       232, /* GL_COMPRESSED_RGBA_FXT1_3DFX */
+       937, /* GL_MULTISAMPLE_3DFX */
+      1309, /* GL_SAMPLE_BUFFERS_3DFX */
+      1300, /* GL_SAMPLES_3DFX */
+       918, /* GL_MODELVIEW2_ARB */
+       921, /* GL_MODELVIEW3_ARB */
+       922, /* GL_MODELVIEW4_ARB */
+       923, /* GL_MODELVIEW5_ARB */
+       924, /* GL_MODELVIEW6_ARB */
+       925, /* GL_MODELVIEW7_ARB */
+       926, /* GL_MODELVIEW8_ARB */
+       927, /* GL_MODELVIEW9_ARB */
+       897, /* GL_MODELVIEW10_ARB */
+       898, /* GL_MODELVIEW11_ARB */
+       899, /* GL_MODELVIEW12_ARB */
+       900, /* GL_MODELVIEW13_ARB */
+       901, /* GL_MODELVIEW14_ARB */
+       902, /* GL_MODELVIEW15_ARB */
+       903, /* GL_MODELVIEW16_ARB */
+       904, /* GL_MODELVIEW17_ARB */
+       905, /* GL_MODELVIEW18_ARB */
+       906, /* GL_MODELVIEW19_ARB */
+       908, /* GL_MODELVIEW20_ARB */
+       909, /* GL_MODELVIEW21_ARB */
+       910, /* GL_MODELVIEW22_ARB */
+       911, /* GL_MODELVIEW23_ARB */
+       912, /* GL_MODELVIEW24_ARB */
+       913, /* GL_MODELVIEW25_ARB */
+       914, /* GL_MODELVIEW26_ARB */
+       915, /* GL_MODELVIEW27_ARB */
+       916, /* GL_MODELVIEW28_ARB */
+       917, /* GL_MODELVIEW29_ARB */
+       919, /* GL_MODELVIEW30_ARB */
+       920, /* GL_MODELVIEW31_ARB */
+       350, /* GL_DOT3_RGB_EXT */
+       348, /* GL_DOT3_RGBA_EXT */
+       891, /* GL_MIRROR_CLAMP_EXT */
+       894, /* GL_MIRROR_CLAMP_TO_EDGE_EXT */
+       932, /* GL_MODULATE_ADD_ATI */
+       933, /* GL_MODULATE_SIGNED_ADD_ATI */
+       934, /* GL_MODULATE_SUBTRACT_ATI */
+      1744, /* GL_YCBCR_MESA */
+      1023, /* GL_PACK_INVERT_MESA */
+       305, /* GL_DEBUG_OBJECT_MESA */
+       306, /* GL_DEBUG_PRINT_MESA */
+       304, /* GL_DEBUG_ASSERT_MESA */
+       106, /* GL_BUFFER_SIZE */
+       108, /* GL_BUFFER_USAGE */
+      1397, /* GL_STENCIL_BACK_FUNC */
+      1395, /* GL_STENCIL_BACK_FAIL */
+      1399, /* GL_STENCIL_BACK_PASS_DEPTH_FAIL */
+      1401, /* GL_STENCIL_BACK_PASS_DEPTH_PASS */
+       483, /* GL_FRAGMENT_PROGRAM_ARB */
+      1157, /* GL_PROGRAM_ALU_INSTRUCTIONS_ARB */
+      1185, /* GL_PROGRAM_TEX_INSTRUCTIONS_ARB */
+      1184, /* GL_PROGRAM_TEX_INDIRECTIONS_ARB */
+      1169, /* GL_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+      1175, /* GL_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+      1174, /* GL_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       827, /* GL_MAX_PROGRAM_ALU_INSTRUCTIONS_ARB */
+       850, /* GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB */
+       849, /* GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB */
+       840, /* GL_MAX_PROGRAM_NATIVE_ALU_INSTRUCTIONS_ARB */
+       846, /* GL_MAX_PROGRAM_NATIVE_TEX_INSTRUCTIONS_ARB */
+       845, /* GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB */
+       810, /* GL_MAX_DRAW_BUFFERS */
+       354, /* GL_DRAW_BUFFER0 */
+       357, /* GL_DRAW_BUFFER1 */
+       378, /* GL_DRAW_BUFFER2 */
+       381, /* GL_DRAW_BUFFER3 */
+       384, /* GL_DRAW_BUFFER4 */
+       387, /* GL_DRAW_BUFFER5 */
+       390, /* GL_DRAW_BUFFER6 */
+       393, /* GL_DRAW_BUFFER7 */
+       396, /* GL_DRAW_BUFFER8 */
+       399, /* GL_DRAW_BUFFER9 */
+       358, /* GL_DRAW_BUFFER10 */
+       361, /* GL_DRAW_BUFFER11 */
+       364, /* GL_DRAW_BUFFER12 */
+       367, /* GL_DRAW_BUFFER13 */
+       370, /* GL_DRAW_BUFFER14 */
+       373, /* GL_DRAW_BUFFER15 */
+        81, /* GL_BLEND_EQUATION_ALPHA */
+       790, /* GL_MATRIX_PALETTE_ARB */
+       821, /* GL_MAX_MATRIX_PALETTE_STACK_DEPTH_ARB */
+       824, /* GL_MAX_PALETTE_MATRICES_ARB */
+       288, /* GL_CURRENT_PALETTE_MATRIX_ARB */
+       784, /* GL_MATRIX_INDEX_ARRAY_ARB */
+       283, /* GL_CURRENT_MATRIX_INDEX_ARB */
+       786, /* GL_MATRIX_INDEX_ARRAY_SIZE_ARB */
+       788, /* GL_MATRIX_INDEX_ARRAY_TYPE_ARB */
+       787, /* GL_MATRIX_INDEX_ARRAY_STRIDE_ARB */
+       785, /* GL_MATRIX_INDEX_ARRAY_POINTER_ARB */
+      1575, /* GL_TEXTURE_DEPTH_SIZE */
+       338, /* GL_DEPTH_TEXTURE_MODE */
+      1541, /* GL_TEXTURE_COMPARE_MODE */
+      1539, /* GL_TEXTURE_COMPARE_FUNC */
+       216, /* GL_COMPARE_R_TO_TEXTURE */
+      1091, /* GL_POINT_SPRITE */
+       265, /* GL_COORD_REPLACE */
+      1095, /* GL_POINT_SPRITE_R_MODE_NV */
+      1212, /* GL_QUERY_COUNTER_BITS */
+       290, /* GL_CURRENT_QUERY */
+      1214, /* GL_QUERY_RESULT */
+      1216, /* GL_QUERY_RESULT_AVAILABLE */
+       871, /* GL_MAX_VERTEX_ATTRIBS */
+      1706, /* GL_VERTEX_ATTRIB_ARRAY_NORMALIZED */
+       336, /* GL_DEPTH_STENCIL_TO_RGBA_NV */
+       335, /* GL_DEPTH_STENCIL_TO_BGRA_NV */
+       857, /* GL_MAX_TEXTURE_COORDS */
+       859, /* GL_MAX_TEXTURE_IMAGE_UNITS */
+      1162, /* GL_PROGRAM_ERROR_STRING_ARB */
+      1164, /* GL_PROGRAM_FORMAT_ASCII_ARB */
+      1163, /* GL_PROGRAM_FORMAT_ARB */
+      1620, /* GL_TEXTURE_UNSIGNED_REMAP_MODE_NV */
+       317, /* GL_DEPTH_BOUNDS_TEST_EXT */
+       316, /* GL_DEPTH_BOUNDS_EXT */
         52, /* GL_ARRAY_BUFFER */
-       420, /* GL_ELEMENT_ARRAY_BUFFER */
-        54, /* GL_ARRAY_BUFFER_BINDING */
-       422, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
-      1682, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
-       960, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
-       140, /* GL_COLOR_ARRAY_BUFFER_BINDING */
-       579, /* GL_INDEX_ARRAY_BUFFER_BINDING */
-      1556, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
-       416, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
-      1323, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
-       463, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
-      1734, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
-      1704, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
-      1167, /* GL_PROGRAM_INSTRUCTIONS_ARB */
-       835, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
-      1173, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-       844, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
-      1185, /* GL_PROGRAM_TEMPORARIES_ARB */
-       850, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
-      1175, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
-       846, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
-      1179, /* GL_PROGRAM_PARAMETERS_ARB */
-       849, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
-      1174, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
-       845, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
-      1160, /* GL_PROGRAM_ATTRIBS_ARB */
-       830, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
-      1172, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
-       843, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
-      1158, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
-       828, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
-      1170, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       841, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
-       836, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
-       832, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
-      1188, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
-      1633, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
-      1226, /* GL_READ_ONLY */
-      1742, /* GL_WRITE_ONLY */
-      1228, /* GL_READ_WRITE */
-       101, /* GL_BUFFER_ACCESS */
-       103, /* GL_BUFFER_MAPPED */
-       105, /* GL_BUFFER_MAP_POINTER */
-      1627, /* GL_TIME_ELAPSED_EXT */
-       746, /* GL_MATRIX0_ARB */
-       758, /* GL_MATRIX1_ARB */
-       770, /* GL_MATRIX2_ARB */
-       774, /* GL_MATRIX3_ARB */
-       776, /* GL_MATRIX4_ARB */
-       778, /* GL_MATRIX5_ARB */
-       780, /* GL_MATRIX6_ARB */
-       782, /* GL_MATRIX7_ARB */
-       784, /* GL_MATRIX8_ARB */
-       785, /* GL_MATRIX9_ARB */
-       748, /* GL_MATRIX10_ARB */
-       749, /* GL_MATRIX11_ARB */
-       750, /* GL_MATRIX12_ARB */
-       751, /* GL_MATRIX13_ARB */
-       752, /* GL_MATRIX14_ARB */
-       753, /* GL_MATRIX15_ARB */
-       754, /* GL_MATRIX16_ARB */
-       755, /* GL_MATRIX17_ARB */
-       756, /* GL_MATRIX18_ARB */
-       757, /* GL_MATRIX19_ARB */
-       760, /* GL_MATRIX20_ARB */
-       761, /* GL_MATRIX21_ARB */
-       762, /* GL_MATRIX22_ARB */
-       763, /* GL_MATRIX23_ARB */
-       764, /* GL_MATRIX24_ARB */
-       765, /* GL_MATRIX25_ARB */
-       766, /* GL_MATRIX26_ARB */
-       767, /* GL_MATRIX27_ARB */
-       768, /* GL_MATRIX28_ARB */
-       769, /* GL_MATRIX29_ARB */
-       772, /* GL_MATRIX30_ARB */
-       773, /* GL_MATRIX31_ARB */
-      1429, /* GL_STREAM_DRAW */
-      1431, /* GL_STREAM_READ */
-      1427, /* GL_STREAM_COPY */
-      1391, /* GL_STATIC_DRAW */
-      1393, /* GL_STATIC_READ */
-      1389, /* GL_STATIC_COPY */
-       410, /* GL_DYNAMIC_DRAW */
-       412, /* GL_DYNAMIC_READ */
-       408, /* GL_DYNAMIC_COPY */
-       535, /* GL_GL_PIXEL_PACK_BUFFER */
-       537, /* GL_GL_PIXEL_UNPACK_BUFFER */
-       536, /* GL_GL_PIXEL_PACK_BUFFER_BINDING */
-       538, /* GL_GL_PIXEL_UNPACK_BUFFER_BINDING */
-       833, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
-       831, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
-       834, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
-       838, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
-       837, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
-       795, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
-      1423, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
+       419, /* GL_ELEMENT_ARRAY_BUFFER */
+        53, /* GL_ARRAY_BUFFER_BINDING */
+       420, /* GL_ELEMENT_ARRAY_BUFFER_BINDING */
+      1680, /* GL_VERTEX_ARRAY_BUFFER_BINDING */
+       958, /* GL_NORMAL_ARRAY_BUFFER_BINDING */
+       139, /* GL_COLOR_ARRAY_BUFFER_BINDING */
+       577, /* GL_INDEX_ARRAY_BUFFER_BINDING */
+      1554, /* GL_TEXTURE_COORD_ARRAY_BUFFER_BINDING */
+       415, /* GL_EDGE_FLAG_ARRAY_BUFFER_BINDING */
+      1321, /* GL_SECONDARY_COLOR_ARRAY_BUFFER_BINDING */
+       461, /* GL_FOG_COORDINATE_ARRAY_BUFFER_BINDING */
+      1732, /* GL_WEIGHT_ARRAY_BUFFER_BINDING */
+      1702, /* GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING */
+      1165, /* GL_PROGRAM_INSTRUCTIONS_ARB */
+       833, /* GL_MAX_PROGRAM_INSTRUCTIONS_ARB */
+      1171, /* GL_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+       842, /* GL_MAX_PROGRAM_NATIVE_INSTRUCTIONS_ARB */
+      1183, /* GL_PROGRAM_TEMPORARIES_ARB */
+       848, /* GL_MAX_PROGRAM_TEMPORARIES_ARB */
+      1173, /* GL_PROGRAM_NATIVE_TEMPORARIES_ARB */
+       844, /* GL_MAX_PROGRAM_NATIVE_TEMPORARIES_ARB */
+      1177, /* GL_PROGRAM_PARAMETERS_ARB */
+       847, /* GL_MAX_PROGRAM_PARAMETERS_ARB */
+      1172, /* GL_PROGRAM_NATIVE_PARAMETERS_ARB */
+       843, /* GL_MAX_PROGRAM_NATIVE_PARAMETERS_ARB */
+      1158, /* GL_PROGRAM_ATTRIBS_ARB */
+       828, /* GL_MAX_PROGRAM_ATTRIBS_ARB */
+      1170, /* GL_PROGRAM_NATIVE_ATTRIBS_ARB */
+       841, /* GL_MAX_PROGRAM_NATIVE_ATTRIBS_ARB */
+      1156, /* GL_PROGRAM_ADDRESS_REGISTERS_ARB */
+       826, /* GL_MAX_PROGRAM_ADDRESS_REGISTERS_ARB */
+      1168, /* GL_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       839, /* GL_MAX_PROGRAM_NATIVE_ADDRESS_REGISTERS_ARB */
+       834, /* GL_MAX_PROGRAM_LOCAL_PARAMETERS_ARB */
+       830, /* GL_MAX_PROGRAM_ENV_PARAMETERS_ARB */
+      1186, /* GL_PROGRAM_UNDER_NATIVE_LIMITS_ARB */
+      1631, /* GL_TRANSPOSE_CURRENT_MATRIX_ARB */
+      1224, /* GL_READ_ONLY */
+      1740, /* GL_WRITE_ONLY */
+      1226, /* GL_READ_WRITE */
+       100, /* GL_BUFFER_ACCESS */
+       102, /* GL_BUFFER_MAPPED */
+       104, /* GL_BUFFER_MAP_POINTER */
+      1625, /* GL_TIME_ELAPSED_EXT */
+       744, /* GL_MATRIX0_ARB */
+       756, /* GL_MATRIX1_ARB */
+       768, /* GL_MATRIX2_ARB */
+       772, /* GL_MATRIX3_ARB */
+       774, /* GL_MATRIX4_ARB */
+       776, /* GL_MATRIX5_ARB */
+       778, /* GL_MATRIX6_ARB */
+       780, /* GL_MATRIX7_ARB */
+       782, /* GL_MATRIX8_ARB */
+       783, /* GL_MATRIX9_ARB */
+       746, /* GL_MATRIX10_ARB */
+       747, /* GL_MATRIX11_ARB */
+       748, /* GL_MATRIX12_ARB */
+       749, /* GL_MATRIX13_ARB */
+       750, /* GL_MATRIX14_ARB */
+       751, /* GL_MATRIX15_ARB */
+       752, /* GL_MATRIX16_ARB */
+       753, /* GL_MATRIX17_ARB */
+       754, /* GL_MATRIX18_ARB */
+       755, /* GL_MATRIX19_ARB */
+       758, /* GL_MATRIX20_ARB */
+       759, /* GL_MATRIX21_ARB */
+       760, /* GL_MATRIX22_ARB */
+       761, /* GL_MATRIX23_ARB */
+       762, /* GL_MATRIX24_ARB */
+       763, /* GL_MATRIX25_ARB */
+       764, /* GL_MATRIX26_ARB */
+       765, /* GL_MATRIX27_ARB */
+       766, /* GL_MATRIX28_ARB */
+       767, /* GL_MATRIX29_ARB */
+       770, /* GL_MATRIX30_ARB */
+       771, /* GL_MATRIX31_ARB */
+      1427, /* GL_STREAM_DRAW */
+      1429, /* GL_STREAM_READ */
+      1425, /* GL_STREAM_COPY */
+      1389, /* GL_STATIC_DRAW */
+      1391, /* GL_STATIC_READ */
+      1387, /* GL_STATIC_COPY */
+       409, /* GL_DYNAMIC_DRAW */
+       411, /* GL_DYNAMIC_READ */
+       407, /* GL_DYNAMIC_COPY */
+       533, /* GL_GL_PIXEL_PACK_BUFFER */
+       535, /* GL_GL_PIXEL_UNPACK_BUFFER */
+       534, /* GL_GL_PIXEL_PACK_BUFFER_BINDING */
+       536, /* GL_GL_PIXEL_UNPACK_BUFFER_BINDING */
+       831, /* GL_MAX_PROGRAM_EXEC_INSTRUCTIONS_NV */
+       829, /* GL_MAX_PROGRAM_CALL_DEPTH_NV */
+       832, /* GL_MAX_PROGRAM_IF_DEPTH_NV */
+       836, /* GL_MAX_PROGRAM_LOOP_DEPTH_NV */
+       835, /* GL_MAX_PROGRAM_LOOP_COUNT_NV */
+       793, /* GL_MAX_ARRAY_TEXTURE_LAYERS_EXT */
+      1421, /* GL_STENCIL_TEST_TWO_SIDE_EXT */
         17, /* GL_ACTIVE_STENCIL_FACE_EXT */
-       894, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
-      1304, /* GL_SAMPLES_PASSED */
-       486, /* GL_FRAGMENT_SHADER */
-      1728, /* GL_VERTEX_SHADER */
-      1178, /* GL_PROGRAM_OBJECT_ARB */
-      1336, /* GL_SHADER_OBJECT_ARB */
-       819, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
-       877, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
-       871, /* GL_MAX_VARYING_FLOATS */
-       875, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
-       804, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
-       985, /* GL_OBJECT_TYPE_ARB */
-      1338, /* GL_SHADER_TYPE */
-       451, /* GL_FLOAT_VEC2 */
-       453, /* GL_FLOAT_VEC3 */
-       455, /* GL_FLOAT_VEC4 */
-       606, /* GL_INT_VEC2 */
-       608, /* GL_INT_VEC3 */
-       610, /* GL_INT_VEC4 */
-        93, /* GL_BOOL */
-        95, /* GL_BOOL_VEC2 */
-        97, /* GL_BOOL_VEC3 */
-        99, /* GL_BOOL_VEC4 */
-       445, /* GL_FLOAT_MAT2 */
-       447, /* GL_FLOAT_MAT3 */
-       449, /* GL_FLOAT_MAT4 */
-      1295, /* GL_SAMPLER_1D */
-      1297, /* GL_SAMPLER_2D */
-      1299, /* GL_SAMPLER_3D */
-      1300, /* GL_SAMPLER_CUBE */
-      1296, /* GL_SAMPLER_1D_SHADOW */
-      1298, /* GL_SAMPLER_2D_SHADOW */
-       529, /* GL_GL_FLOAT_MAT2x3 */
-       530, /* GL_GL_FLOAT_MAT2x4 */
-       531, /* GL_GL_FLOAT_MAT3x2 */
-       532, /* GL_GL_FLOAT_MAT3x4 */
-       533, /* GL_GL_FLOAT_MAT4x2 */
-       534, /* GL_GL_FLOAT_MAT4x3 */
-       312, /* GL_DELETE_STATUS */
-       221, /* GL_COMPILE_STATUS */
-       659, /* GL_LINK_STATUS */
-      1677, /* GL_VALIDATE_STATUS */
-       591, /* GL_INFO_LOG_LENGTH */
-        56, /* GL_ATTACHED_SHADERS */
+       892, /* GL_MIRROR_CLAMP_TO_BORDER_EXT */
+      1302, /* GL_SAMPLES_PASSED */
+       484, /* GL_FRAGMENT_SHADER */
+      1726, /* GL_VERTEX_SHADER */
+      1176, /* GL_PROGRAM_OBJECT_ARB */
+      1334, /* GL_SHADER_OBJECT_ARB */
+       817, /* GL_MAX_FRAGMENT_UNIFORM_COMPONENTS */
+       875, /* GL_MAX_VERTEX_UNIFORM_COMPONENTS */
+       869, /* GL_MAX_VARYING_FLOATS */
+       873, /* GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS */
+       802, /* GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS */
+       983, /* GL_OBJECT_TYPE_ARB */
+      1336, /* GL_SHADER_TYPE */
+       449, /* GL_FLOAT_VEC2 */
+       451, /* GL_FLOAT_VEC3 */
+       453, /* GL_FLOAT_VEC4 */
+       604, /* GL_INT_VEC2 */
+       606, /* GL_INT_VEC3 */
+       608, /* GL_INT_VEC4 */
+        92, /* GL_BOOL */
+        94, /* GL_BOOL_VEC2 */
+        96, /* GL_BOOL_VEC3 */
+        98, /* GL_BOOL_VEC4 */
+       443, /* GL_FLOAT_MAT2 */
+       445, /* GL_FLOAT_MAT3 */
+       447, /* GL_FLOAT_MAT4 */
+      1293, /* GL_SAMPLER_1D */
+      1295, /* GL_SAMPLER_2D */
+      1297, /* GL_SAMPLER_3D */
+      1298, /* GL_SAMPLER_CUBE */
+      1294, /* GL_SAMPLER_1D_SHADOW */
+      1296, /* GL_SAMPLER_2D_SHADOW */
+       527, /* GL_GL_FLOAT_MAT2x3 */
+       528, /* GL_GL_FLOAT_MAT2x4 */
+       529, /* GL_GL_FLOAT_MAT3x2 */
+       530, /* GL_GL_FLOAT_MAT3x4 */
+       531, /* GL_GL_FLOAT_MAT4x2 */
+       532, /* GL_GL_FLOAT_MAT4x3 */
+       311, /* GL_DELETE_STATUS */
+       220, /* GL_COMPILE_STATUS */
+       657, /* GL_LINK_STATUS */
+      1675, /* GL_VALIDATE_STATUS */
+       589, /* GL_INFO_LOG_LENGTH */
+        55, /* GL_ATTACHED_SHADERS */
         20, /* GL_ACTIVE_UNIFORMS */
         21, /* GL_ACTIVE_UNIFORM_MAX_LENGTH */
-      1337, /* GL_SHADER_SOURCE_LENGTH */
+      1335, /* GL_SHADER_SOURCE_LENGTH */
         15, /* GL_ACTIVE_ATTRIBUTES */
         16, /* GL_ACTIVE_ATTRIBUTE_MAX_LENGTH */
-       488, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
-      1340, /* GL_SHADING_LANGUAGE_VERSION */
-       290, /* GL_CURRENT_PROGRAM */
-      1034, /* GL_PALETTE4_RGB8_OES */
-      1036, /* GL_PALETTE4_RGBA8_OES */
-      1032, /* GL_PALETTE4_R5_G6_B5_OES */
-      1035, /* GL_PALETTE4_RGBA4_OES */
-      1033, /* GL_PALETTE4_RGB5_A1_OES */
-      1039, /* GL_PALETTE8_RGB8_OES */
-      1041, /* GL_PALETTE8_RGBA8_OES */
-      1037, /* GL_PALETTE8_R5_G6_B5_OES */
-      1040, /* GL_PALETTE8_RGBA4_OES */
-      1038, /* GL_PALETTE8_RGB5_A1_OES */
-       574, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
-       573, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
-      1511, /* GL_TEXTURE_1D_ARRAY_EXT */
-      1198, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
-      1513, /* GL_TEXTURE_2D_ARRAY_EXT */
-      1201, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
-      1519, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
-      1521, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
-       543, /* GL_GL_SRGB */
-       544, /* GL_GL_SRGB8 */
-       546, /* GL_GL_SRGB_ALPHA */
-       545, /* GL_GL_SRGB8_ALPHA8 */
-       542, /* GL_GL_SLUMINANCE_ALPHA */
-       541, /* GL_GL_SLUMINANCE8_ALPHA8 */
-       539, /* GL_GL_SLUMINANCE */
-       540, /* GL_GL_SLUMINANCE8 */
-       526, /* GL_GL_COMPRESSED_SRGB */
-       527, /* GL_GL_COMPRESSED_SRGB_ALPHA */
-       524, /* GL_GL_COMPRESSED_SLUMINANCE */
-       525, /* GL_GL_COMPRESSED_SLUMINANCE_ALPHA */
-      1095, /* GL_POINT_SPRITE_COORD_ORIGIN */
-       667, /* GL_LOWER_LEFT */
-      1674, /* GL_UPPER_LEFT */
-      1405, /* GL_STENCIL_BACK_REF */
-      1406, /* GL_STENCIL_BACK_VALUE_MASK */
-      1407, /* GL_STENCIL_BACK_WRITEMASK */
-       403, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
-      1240, /* GL_RENDERBUFFER_BINDING_EXT */
-      1225, /* GL_READ_FRAMEBUFFER_EXT */
-       404, /* GL_DRAW_FRAMEBUFFER_EXT */
-      1224, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
-       490, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
-       489, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
-       494, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
-       492, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
-       491, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
-       496, /* GL_FRAMEBUFFER_COMPLETE_EXT */
-       498, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
-       503, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
-       501, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
-       499, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
-       502, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
-       500, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
-       504, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
-       506, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
-       505, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
-       801, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
-       146, /* GL_COLOR_ATTACHMENT0_EXT */
-       153, /* GL_COLOR_ATTACHMENT1_EXT */
-       154, /* GL_COLOR_ATTACHMENT2_EXT */
-       155, /* GL_COLOR_ATTACHMENT3_EXT */
-       156, /* GL_COLOR_ATTACHMENT4_EXT */
-       157, /* GL_COLOR_ATTACHMENT5_EXT */
-       158, /* GL_COLOR_ATTACHMENT6_EXT */
-       159, /* GL_COLOR_ATTACHMENT7_EXT */
-       160, /* GL_COLOR_ATTACHMENT8_EXT */
-       161, /* GL_COLOR_ATTACHMENT9_EXT */
-       147, /* GL_COLOR_ATTACHMENT10_EXT */
-       148, /* GL_COLOR_ATTACHMENT11_EXT */
-       149, /* GL_COLOR_ATTACHMENT12_EXT */
-       150, /* GL_COLOR_ATTACHMENT13_EXT */
-       151, /* GL_COLOR_ATTACHMENT14_EXT */
-       152, /* GL_COLOR_ATTACHMENT15_EXT */
-       314, /* GL_DEPTH_ATTACHMENT_EXT */
-      1396, /* GL_STENCIL_ATTACHMENT_EXT */
-       497, /* GL_FRAMEBUFFER_EXT */
-      1241, /* GL_RENDERBUFFER_EXT */
-      1244, /* GL_RENDERBUFFER_WIDTH_EXT */
-      1242, /* GL_RENDERBUFFER_HEIGHT_EXT */
-      1243, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
-      1418, /* GL_STENCIL_INDEX_EXT */
-      1415, /* GL_STENCIL_INDEX1_EXT */
-      1416, /* GL_STENCIL_INDEX4_EXT */
-      1417, /* GL_STENCIL_INDEX8_EXT */
-      1414, /* GL_STENCIL_INDEX16_EXT */
-       428, /* GL_EVAL_BIT */
-      1222, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
-       661, /* GL_LIST_BIT */
-      1527, /* GL_TEXTURE_BIT */
-      1319, /* GL_SCISSOR_BIT */
+       486, /* GL_FRAGMENT_SHADER_DERIVATIVE_HINT */
+      1338, /* GL_SHADING_LANGUAGE_VERSION */
+       289, /* GL_CURRENT_PROGRAM */
+      1032, /* GL_PALETTE4_RGB8_OES */
+      1034, /* GL_PALETTE4_RGBA8_OES */
+      1030, /* GL_PALETTE4_R5_G6_B5_OES */
+      1033, /* GL_PALETTE4_RGBA4_OES */
+      1031, /* GL_PALETTE4_RGB5_A1_OES */
+      1037, /* GL_PALETTE8_RGB8_OES */
+      1039, /* GL_PALETTE8_RGBA8_OES */
+      1035, /* GL_PALETTE8_R5_G6_B5_OES */
+      1038, /* GL_PALETTE8_RGBA4_OES */
+      1036, /* GL_PALETTE8_RGB5_A1_OES */
+       572, /* GL_IMPLEMENTATION_COLOR_READ_TYPE_OES */
+       571, /* GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES */
+      1509, /* GL_TEXTURE_1D_ARRAY_EXT */
+      1196, /* GL_PROXY_TEXTURE_1D_ARRAY_EXT */
+      1511, /* GL_TEXTURE_2D_ARRAY_EXT */
+      1199, /* GL_PROXY_TEXTURE_2D_ARRAY_EXT */
+      1517, /* GL_TEXTURE_BINDING_1D_ARRAY_EXT */
+      1519, /* GL_TEXTURE_BINDING_2D_ARRAY_EXT */
+       541, /* GL_GL_SRGB */
+       542, /* GL_GL_SRGB8 */
+       544, /* GL_GL_SRGB_ALPHA */
+       543, /* GL_GL_SRGB8_ALPHA8 */
+       540, /* GL_GL_SLUMINANCE_ALPHA */
+       539, /* GL_GL_SLUMINANCE8_ALPHA8 */
+       537, /* GL_GL_SLUMINANCE */
+       538, /* GL_GL_SLUMINANCE8 */
+       524, /* GL_GL_COMPRESSED_SRGB */
+       525, /* GL_GL_COMPRESSED_SRGB_ALPHA */
+       522, /* GL_GL_COMPRESSED_SLUMINANCE */
+       523, /* GL_GL_COMPRESSED_SLUMINANCE_ALPHA */
+      1093, /* GL_POINT_SPRITE_COORD_ORIGIN */
+       665, /* GL_LOWER_LEFT */
+      1672, /* GL_UPPER_LEFT */
+      1403, /* GL_STENCIL_BACK_REF */
+      1404, /* GL_STENCIL_BACK_VALUE_MASK */
+      1405, /* GL_STENCIL_BACK_WRITEMASK */
+       402, /* GL_DRAW_FRAMEBUFFER_BINDING_EXT */
+      1238, /* GL_RENDERBUFFER_BINDING_EXT */
+      1223, /* GL_READ_FRAMEBUFFER_EXT */
+       403, /* GL_DRAW_FRAMEBUFFER_EXT */
+      1222, /* GL_READ_FRAMEBUFFER_BINDING_EXT */
+       488, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT */
+       487, /* GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT */
+       492, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LEVEL_EXT */
+       490, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_CUBE_MAP_FACE_EXT */
+       489, /* GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_3D_ZOFFSET_EXT */
+       494, /* GL_FRAMEBUFFER_COMPLETE_EXT */
+       496, /* GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT */
+       501, /* GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT */
+       499, /* GL_FRAMEBUFFER_INCOMPLETE_DUPLICATE_ATTACHMENT_EXT */
+       497, /* GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT */
+       500, /* GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT */
+       498, /* GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT */
+       502, /* GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT */
+       504, /* GL_FRAMEBUFFER_UNSUPPORTED_EXT */
+       503, /* GL_FRAMEBUFFER_STATUS_ERROR_EXT */
+       799, /* GL_MAX_COLOR_ATTACHMENTS_EXT */
+       145, /* GL_COLOR_ATTACHMENT0_EXT */
+       152, /* GL_COLOR_ATTACHMENT1_EXT */
+       153, /* GL_COLOR_ATTACHMENT2_EXT */
+       154, /* GL_COLOR_ATTACHMENT3_EXT */
+       155, /* GL_COLOR_ATTACHMENT4_EXT */
+       156, /* GL_COLOR_ATTACHMENT5_EXT */
+       157, /* GL_COLOR_ATTACHMENT6_EXT */
+       158, /* GL_COLOR_ATTACHMENT7_EXT */
+       159, /* GL_COLOR_ATTACHMENT8_EXT */
+       160, /* GL_COLOR_ATTACHMENT9_EXT */
+       146, /* GL_COLOR_ATTACHMENT10_EXT */
+       147, /* GL_COLOR_ATTACHMENT11_EXT */
+       148, /* GL_COLOR_ATTACHMENT12_EXT */
+       149, /* GL_COLOR_ATTACHMENT13_EXT */
+       150, /* GL_COLOR_ATTACHMENT14_EXT */
+       151, /* GL_COLOR_ATTACHMENT15_EXT */
+       313, /* GL_DEPTH_ATTACHMENT_EXT */
+      1394, /* GL_STENCIL_ATTACHMENT_EXT */
+       495, /* GL_FRAMEBUFFER_EXT */
+      1239, /* GL_RENDERBUFFER_EXT */
+      1242, /* GL_RENDERBUFFER_WIDTH_EXT */
+      1240, /* GL_RENDERBUFFER_HEIGHT_EXT */
+      1241, /* GL_RENDERBUFFER_INTERNAL_FORMAT_EXT */
+      1416, /* GL_STENCIL_INDEX_EXT */
+      1413, /* GL_STENCIL_INDEX1_EXT */
+      1414, /* GL_STENCIL_INDEX4_EXT */
+      1415, /* GL_STENCIL_INDEX8_EXT */
+      1412, /* GL_STENCIL_INDEX16_EXT */
+       426, /* GL_EVAL_BIT */
+      1220, /* GL_RASTER_POSITION_UNCLIPPED_IBM */
+       659, /* GL_LIST_BIT */
+      1525, /* GL_TEXTURE_BIT */
+      1317, /* GL_SCISSOR_BIT */
         29, /* GL_ALL_ATTRIB_BITS */
-       941, /* GL_MULTISAMPLE_BIT */
+       939, /* GL_MULTISAMPLE_BIT */
         30, /* GL_ALL_CLIENT_ATTRIB_BITS */
 };
 
@@ -4870,8 +4866,10 @@ const char *_mesa_lookup_enum_by_nr( int nr )
 {
    unsigned * i;
 
-   i = (unsigned *)_mesa_bsearch( & nr, reduced_enums, Elements(reduced_enums),
-                                  sizeof(reduced_enums[0]), (cfunc) compar_nr );
+   i = (unsigned *) _mesa_bsearch(& nr, reduced_enums,
+                                  Elements(reduced_enums),
+                                  sizeof(reduced_enums[0]),
+                                  (cfunc) compar_nr);
 
    if ( i != NULL ) {
       return & enum_string_table[ all_enums[ *i ].offset ];
@@ -4888,8 +4886,10 @@ int _mesa_lookup_enum_by_name( const char *symbol )
    enum_elt * f = NULL;
 
    if ( symbol != NULL ) {
-      f = (enum_elt *)_mesa_bsearch( symbol, all_enums, Elements(all_enums),
-                                     sizeof( enum_elt ), (cfunc) compar_name );
+      f = (enum_elt *) _mesa_bsearch(symbol, all_enums,
+                                     Elements(all_enums),
+                                     sizeof( enum_elt ),
+                                     (cfunc) compar_name);
    }
 
    return (f != NULL) ? f->n : -1;
diff --git a/src/mesa/main/execmem.c b/src/mesa/main/execmem.c
index 0fe85af93e0..f95c31862a5 100644
--- a/src/mesa/main/execmem.c
+++ b/src/mesa/main/execmem.c
@@ -65,9 +65,17 @@ static struct mem_block *exec_heap = NULL;
 static unsigned char *exec_mem = NULL;
 
 
-static void
+static int
 init_heap(void)
 {
+#ifdef MESA_SELINUX
+   if (is_selinux_enabled()) {
+      if (!security_get_boolean_active("allow_execmem") ||
+	  !security_get_boolean_pending("allow_execmem"))
+         return 0;
+   }
+#endif
+
    if (!exec_heap)
       exec_heap = mmInit( 0, EXEC_HEAP_SIZE );
    
@@ -75,6 +83,8 @@ init_heap(void)
       exec_mem = (unsigned char *) mmap(0, EXEC_HEAP_SIZE, 
 					PROT_EXEC | PROT_READ | PROT_WRITE, 
 					MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+
+   return (exec_mem != NULL);
 }
 
 
@@ -86,7 +96,8 @@ _mesa_exec_malloc(GLuint size)
 
    _glthread_LOCK_MUTEX(exec_mutex);
 
-   init_heap();
+   if (!init_heap())
+      goto bail;
 
    if (exec_heap) {
       size = (size + 31) & ~31;
@@ -97,7 +108,8 @@ _mesa_exec_malloc(GLuint size)
       addr = exec_mem + block->ofs;
    else 
       _mesa_printf("_mesa_exec_malloc failed\n");
-   
+
+bail:
    _glthread_UNLOCK_MUTEX(exec_mutex);
    
    return addr;
diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
index e5279e7f3eb..aa5b1721234 100644
--- a/src/mesa/main/extensions.c
+++ b/src/mesa/main/extensions.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.3
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -165,7 +165,7 @@ static const struct {
    { OFF, "GL_SGIS_texture_border_clamp",      F(ARB_texture_border_clamp) },
    { ON,  "GL_SGIS_texture_edge_clamp",        F(SGIS_texture_edge_clamp) },
    { ON,  "GL_SGIS_texture_lod",               F(SGIS_texture_lod) },
-   { OFF, "GL_SGIX_depth_texture",             F(SGIX_depth_texture) },
+   { OFF, "GL_SGIX_depth_texture",             F(ARB_depth_texture) },
    { OFF, "GL_SGIX_shadow",                    F(SGIX_shadow) },
    { OFF, "GL_SGIX_shadow_ambient",            F(SGIX_shadow_ambient) },
    { OFF, "GL_SUN_multi_draw_arrays",          F(EXT_multi_draw_arrays) },
@@ -204,7 +204,7 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
    ctx->Extensions.ARB_shading_language_100 = GL_TRUE;
 #endif
 #if FEATURE_ARB_shading_language_120
-   ctx->Extensions.ARB_shading_language_120 = GL_TRUE;
+   ctx->Extensions.ARB_shading_language_120 = GL_FALSE; /* not quite done */
 #endif
    ctx->Extensions.ARB_shadow = GL_TRUE;
    ctx->Extensions.ARB_texture_border_clamp = GL_TRUE;
@@ -292,12 +292,20 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
    ctx->Extensions.SGI_texture_color_table = GL_TRUE;
    ctx->Extensions.SGIS_generate_mipmap = GL_TRUE;
    ctx->Extensions.SGIS_texture_edge_clamp = GL_TRUE;
-   ctx->Extensions.SGIX_depth_texture = GL_TRUE;
    ctx->Extensions.SGIX_shadow = GL_TRUE;
    ctx->Extensions.SGIX_shadow_ambient = GL_TRUE;
 #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program
    ctx->Extensions.EXT_gpu_program_parameters = GL_TRUE;
 #endif
+#if FEATURE_texture_fxt1
+   _mesa_enable_extension(ctx, "GL_3DFX_texture_compression_FXT1");
+#endif
+#if FEATURE_texture_s3tc
+   if (ctx->Mesa_DXTn) {
+      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+      _mesa_enable_extension(ctx, "GL_S3_s3tc");
+   }
+#endif
 }
 
 
@@ -309,6 +317,7 @@ _mesa_enable_imaging_extensions(GLcontext *ctx)
 {
    ctx->Extensions.ARB_imaging = GL_TRUE;
    ctx->Extensions.EXT_blend_color = GL_TRUE;
+   ctx->Extensions.EXT_blend_logic_op = GL_TRUE;
    ctx->Extensions.EXT_blend_minmax = GL_TRUE;
    ctx->Extensions.EXT_blend_subtract = GL_TRUE;
    ctx->Extensions.EXT_convolution = GL_TRUE;
@@ -353,7 +362,6 @@ _mesa_enable_1_4_extensions(GLcontext *ctx)
    ctx->Extensions.ARB_window_pos = GL_TRUE;
    ctx->Extensions.EXT_blend_color = GL_TRUE;
    ctx->Extensions.EXT_blend_func_separate = GL_TRUE;
-   ctx->Extensions.EXT_blend_logic_op = GL_TRUE;
    ctx->Extensions.EXT_blend_minmax = GL_TRUE;
    ctx->Extensions.EXT_blend_subtract = GL_TRUE;
    ctx->Extensions.EXT_fog_coord = GL_TRUE;
@@ -419,7 +427,7 @@ _mesa_enable_2_1_extensions(GLcontext *ctx)
    ctx->Extensions.EXT_texture_sRGB = GL_TRUE;
 #endif
 #ifdef FEATURE_ARB_shading_language_120
-   ctx->Extensions.ARB_shading_language_120 = GL_TRUE;
+   ctx->Extensions.ARB_shading_language_120 = GL_FALSE; /* not quite done */
 #endif
 }
 
diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c
index b5605a199c5..718c97568b5 100644
--- a/src/mesa/main/fbobject.c
+++ b/src/mesa/main/fbobject.c
@@ -983,9 +983,11 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
    }
 
    FLUSH_VERTICES(ctx, _NEW_BUFFERS);
+
    if (ctx->Driver.Flush) {  
       ctx->Driver.Flush(ctx);
    }
+
    if (framebuffer) {
       /* Binding a user-created framebuffer object */
       newFb = _mesa_lookup_framebuffer(ctx, framebuffer);
@@ -1019,32 +1021,18 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer)
     * XXX check if re-binding same buffer and skip some of this code.
     */
 
-   /* for window-framebuffers, re-initialize the fbo values, as they
-      could be wrong (makecurrent with a new drawable while still a fbo
-      was bound will lead to default init fbo values).
-      note that therefore the context ReadBuffer/DrawBuffer values are not
-      valid while fbo's are bound!!! */
    if (bindReadBuf) {
       _mesa_reference_framebuffer(&ctx->ReadBuffer, newFbread);
-      if (!newFbread->Name) {
-         _mesa_readbuffer_update_fields(ctx, ctx->Pixel.ReadBuffer);
-      }
    }
 
    if (bindDrawBuf) {
       /* check if old FB had any texture attachments */
       check_end_texture_render(ctx, ctx->DrawBuffer);
+
       /* check if time to delete this framebuffer */
       _mesa_reference_framebuffer(&ctx->DrawBuffer, newFb);
-      if (!newFb->Name) {
-         GLuint i;
-         GLenum buffers[MAX_DRAW_BUFFERS];
-         for(i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-            buffers[i] = ctx->Color.DrawBuffer[i];
-         }
-         _mesa_drawbuffers(ctx, ctx->Const.MaxDrawBuffers, buffers, NULL);
-      }
-      else {
+
+      if (newFb->Name != 0) {
          /* check if newly bound framebuffer has any texture attachments */
          check_begin_texture_render(ctx, newFb);
       }
@@ -1586,9 +1574,17 @@ _mesa_GenerateMipmapEXT(GLenum target)
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
    texObj = _mesa_select_tex_object(ctx, texUnit, target);
 
-   /* XXX this might not handle cube maps correctly */
    _mesa_lock_texture(ctx, texObj);
-   ctx->Driver.GenerateMipmap(ctx, target, texObj);
+   if (target == GL_TEXTURE_CUBE_MAP) {
+      int face;
+
+      for (face = 0; face < 6; face++)
+	 ctx->Driver.GenerateMipmap(ctx,
+				    GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB + face,
+				    texObj);
+   } else {
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
+   }
    _mesa_unlock_texture(ctx, texObj);
 }
 
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index b87c443fecf..ec0a5e3896d 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -180,12 +180,12 @@ static GLboolean check_active_shininess( GLcontext *ctx,
 
 
 
-static struct state_key *make_state_key( GLcontext *ctx )
+static void make_state_key( GLcontext *ctx, struct state_key *key )
 {
    const struct gl_fragment_program *fp;
-   struct state_key *key = CALLOC_STRUCT(state_key);
    GLuint i;
 
+   memset(key, 0, sizeof(struct state_key));
    fp = ctx->FragmentProgram._Current;
 
    /* This now relies on texenvprogram.c being active:
@@ -304,8 +304,6 @@ static struct state_key *make_state_key( GLcontext *ctx )
 			      texUnit->GenModeQ );
       }
    }
-   
-   return key;
 }
 
 
@@ -738,11 +736,16 @@ static void emit_normalize_vec3( struct tnl_program *p,
 				 struct ureg dest,
 				 struct ureg src )
 {
+#if 0
+   /* XXX use this when drivers are ready for NRM3 */
+   emit_op1(p, OPCODE_NRM3, dest, WRITEMASK_XYZ, src);
+#else
    struct ureg tmp = get_temp(p);
    emit_op2(p, OPCODE_DP3, tmp, WRITEMASK_X, src, src);
    emit_op1(p, OPCODE_RSQ, tmp, WRITEMASK_X, tmp);
    emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(tmp, X));
    release_temp(p, tmp);
+#endif
 }
 
 static void emit_passthrough( struct tnl_program *p, 
@@ -1726,16 +1729,16 @@ struct gl_vertex_program *
 _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
 {
    struct gl_vertex_program *prog;
-   struct state_key *key;
+   struct state_key key;
 
    /* Grab all the relevent state and put it in a single structure:
     */
-   key = make_state_key(ctx);
+   make_state_key(ctx, &key);
 
    /* Look for an already-prepared program for this state:
     */
    prog = (struct gl_vertex_program *)
-      _mesa_search_program_cache(ctx->VertexProgram.Cache, key, sizeof(*key));
+      _mesa_search_program_cache(ctx->VertexProgram.Cache, &key, sizeof(key));
    
    if (!prog) {
       /* OK, we'll have to build a new one */
@@ -1747,7 +1750,7 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
       if (!prog)
          return NULL;
 
-      create_new_program( key, prog,
+      create_new_program( &key, prog,
                           ctx->Const.VertexProgram.MaxTemps );
 
 #if 0
@@ -1756,10 +1759,8 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
                                           &prog->Base );
 #endif
       _mesa_program_cache_insert(ctx, ctx->VertexProgram.Cache,
-                                 key, sizeof(*key), &prog->Base);
+                                 &key, sizeof(key), &prog->Base);
    }
 
-   _mesa_free(key);
-
    return prog;
 }
diff --git a/src/mesa/main/framebuffer.c b/src/mesa/main/framebuffer.c
index 494743f1349..af78363ad3a 100644
--- a/src/mesa/main/framebuffer.c
+++ b/src/mesa/main/framebuffer.c
@@ -32,6 +32,7 @@
 
 #include "glheader.h"
 #include "imports.h"
+#include "buffers.h"
 #include "context.h"
 #include "depthstencil.h"
 #include "mtypes.h"
@@ -108,8 +109,9 @@ _mesa_new_framebuffer(GLcontext *ctx, GLuint name)
    if (fb) {
       fb->Name = name;
       fb->RefCount = 1;
+      fb->_NumColorDrawBuffers = 1;
       fb->ColorDrawBuffer[0] = GL_COLOR_ATTACHMENT0_EXT;
-      fb->_ColorDrawBufferMask[0] = BUFFER_BIT_COLOR0;
+      fb->_ColorDrawBufferIndexes[0] = BUFFER_COLOR0;
       fb->ColorReadBuffer = GL_COLOR_ATTACHMENT0_EXT;
       fb->_ColorReadBufferIndex = BUFFER_COLOR0;
       fb->Delete = _mesa_destroy_framebuffer;
@@ -138,16 +140,18 @@ _mesa_initialize_framebuffer(struct gl_framebuffer *fb, const GLvisual *visual)
    /* save the visual */
    fb->Visual = *visual;
 
-   /* Init glRead/DrawBuffer state */
+   /* Init read/draw renderbuffer state */
    if (visual->doubleBufferMode) {
+      fb->_NumColorDrawBuffers = 1;
       fb->ColorDrawBuffer[0] = GL_BACK;
-      fb->_ColorDrawBufferMask[0] = BUFFER_BIT_BACK_LEFT;
+      fb->_ColorDrawBufferIndexes[0] = BUFFER_BACK_LEFT;
       fb->ColorReadBuffer = GL_BACK;
       fb->_ColorReadBufferIndex = BUFFER_BACK_LEFT;
    }
    else {
+      fb->_NumColorDrawBuffers = 1;
       fb->ColorDrawBuffer[0] = GL_FRONT;
-      fb->_ColorDrawBufferMask[0] = BUFFER_BIT_FRONT_LEFT;
+      fb->_ColorDrawBufferIndexes[0] = BUFFER_FRONT_LEFT;
       fb->ColorReadBuffer = GL_FRONT;
       fb->_ColorReadBufferIndex = BUFFER_FRONT_LEFT;
    }
@@ -660,8 +664,53 @@ _mesa_update_stencil_buffer(GLcontext *ctx,
 }
 
 
+/*
+ * Example DrawBuffers scenarios:
+ *
+ * 1. glDrawBuffer(GL_FRONT_AND_BACK), fixed-func or shader writes to
+ * "gl_FragColor" or program writes to the "result.color" register:
+ *
+ *   fragment color output   renderbuffer
+ *   ---------------------   ---------------
+ *   color[0]                Front, Back
+ *
+ *
+ * 2. glDrawBuffers(3, [GL_FRONT, GL_AUX0, GL_AUX1]), shader writes to
+ * gl_FragData[i] or program writes to result.color[i] registers:
+ *
+ *   fragment color output   renderbuffer
+ *   ---------------------   ---------------
+ *   color[0]                Front
+ *   color[1]                Aux0
+ *   color[3]                Aux1
+ *
+ *
+ * 3. glDrawBuffers(3, [GL_FRONT, GL_AUX0, GL_AUX1]) and shader writes to
+ * gl_FragColor, or fixed function:
+ *
+ *   fragment color output   renderbuffer
+ *   ---------------------   ---------------
+ *   color[0]                Front, Aux0, Aux1
+ *
+ *
+ * In either case, the list of renderbuffers is stored in the
+ * framebuffer->_ColorDrawBuffers[] array and
+ * framebuffer->_NumColorDrawBuffers indicates the number of buffers.
+ * The renderer (like swrast) has to look at the current fragment shader
+ * to see if it writes to gl_FragColor vs. gl_FragData[i] to determine
+ * how to map color outputs to renderbuffers.
+ *
+ * Note that these two calls are equivalent (for fixed function fragment
+ * shading anyway):
+ *   a)  glDrawBuffer(GL_FRONT_AND_BACK);  (assuming non-stereo framebuffer)
+ *   b)  glDrawBuffers(2, [GL_FRONT_LEFT, GL_BACK_LEFT]);
+ */
+
+
+
+
 /**
- * Update the list of color drawing renderbuffer pointers.
+ * Update the (derived) list of color drawing renderbuffer pointers.
  * Later, when we're rendering we'll loop from 0 to _NumColorDrawBuffers
  * writing colors.
  */
@@ -670,42 +719,23 @@ update_color_draw_buffers(GLcontext *ctx, struct gl_framebuffer *fb)
 {
    GLuint output;
 
-   /*
-    * Fragment programs can write to multiple colorbuffers with
-    * the GL_ARB_draw_buffers extension.
-    */
-   for (output = 0; output < ctx->Const.MaxDrawBuffers; output++) {
-      GLbitfield bufferMask = fb->_ColorDrawBufferMask[output];
-      GLuint count = 0;
-      GLuint i;
-      if (!fb->DeletePending) {
-         /* We need the inner loop here because glDrawBuffer(GL_FRONT_AND_BACK)
-          * can specify writing to two or four color buffers (for example).
-          */
-         for (i = 0; bufferMask && i < BUFFER_COUNT; i++) {
-            const GLuint bufferBit = 1 << i;
-            if (bufferBit & bufferMask) {
-               struct gl_renderbuffer *rb = fb->Attachment[i].Renderbuffer;
-               if (rb && rb->Width > 0 && rb->Height > 0) {
-                  fb->_ColorDrawBuffers[output][count] = rb;
-                  count++;
-               }
-               else {
-                  /*
-                  _mesa_warning(ctx, "DrawBuffer names a missing buffer!\n");
-                  */
-               }
-               bufferMask &= ~bufferBit;
-            }
-         }
+   /* set 0th buffer to NULL now in case _NumColorDrawBuffers is zero */
+   fb->_ColorDrawBuffers[0] = NULL;
+
+   for (output = 0; output < fb->_NumColorDrawBuffers; output++) {
+      GLint buf = fb->_ColorDrawBufferIndexes[output];
+      if (buf >= 0) {
+         fb->_ColorDrawBuffers[output] = fb->Attachment[buf].Renderbuffer;
+      }
+      else {
+         fb->_ColorDrawBuffers[output] = NULL;
       }
-      fb->_NumColorDrawBuffers[output] = count;
    }
 }
 
 
 /**
- * Update the color read renderbuffer pointer.
+ * Update the (derived) color read renderbuffer pointer.
  * Unlike the DrawBuffer, we can only read from one (or zero) color buffers.
  */
 static void
@@ -727,19 +757,51 @@ update_color_read_buffer(GLcontext *ctx, struct gl_framebuffer *fb)
 }
 
 
+/**
+ * Update a gl_framebuffer's derived state.
+ *
+ * Specifically, update these framebuffer fields:
+ *    _ColorDrawBuffers
+ *    _NumColorDrawBuffers
+ *    _ColorReadBuffer
+ *    _DepthBuffer
+ *    _StencilBuffer
+ *
+ * If the framebuffer is user-created, make sure it's complete.
+ *
+ * The following functions (at least) can effect framebuffer state:
+ * glReadBuffer, glDrawBuffer, glDrawBuffersARB, glFramebufferRenderbufferEXT,
+ * glRenderbufferStorageEXT.
+ */
 static void
 update_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
 {
-   /* Completeness only matters for user-created framebuffers */
-   if (fb->Name != 0) {
-      /* XXX: EXT_framebuffer_blit:
-         framebuffer must still be complete wrt read/draw? */
+   if (fb->Name == 0) {
+      /* This is a window-system framebuffer */
+      /* Need to update the FB's GL_DRAW_BUFFER state to match the
+       * context state (GL_READ_BUFFER too).
+       */
+      if (fb->ColorDrawBuffer[0] != ctx->Color.DrawBuffer[0]) {
+         _mesa_drawbuffers(ctx, ctx->Const.MaxDrawBuffers,
+                           ctx->Color.DrawBuffer, NULL);
+      }
+      if (fb->ColorReadBuffer != ctx->Pixel.ReadBuffer) {
+         
+      }
+   }
+   else {
+      /* This is a user-created framebuffer.
+       * Completeness only matters for user-created framebuffers.
+       */
       _mesa_test_framebuffer_completeness(ctx, fb);
       _mesa_update_framebuffer_visual(fb);
    }
 
-   /* update_color_draw/read_buffers not needed for
-      read/draw only fb, but shouldn't hurt ??? */
+   /* Strictly speaking, we don't need to update the draw-state
+    * if this FB is bound as ctx->ReadBuffer (and conversely, the
+    * read-state if this FB is bound as ctx->DrawBuffer), but no
+    * harm.
+    */
    update_color_draw_buffers(ctx, fb);
    update_color_read_buffer(ctx, fb);
    _mesa_update_depth_buffer(ctx, fb, BUFFER_DEPTH);
@@ -748,28 +810,19 @@ update_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
    compute_depth_max(fb);
 }
 
+
 /**
  * Update state related to the current draw/read framebuffers.
- * Specifically, update these framebuffer fields:
- *    _ColorDrawBuffers
- *    _NumColorDrawBuffers
- *    _ColorReadBuffer
- *    _DepthBuffer
- *    _StencilBuffer
- * If the current framebuffer is user-created, make sure it's complete.
- * The following functions can effect this state:  glReadBuffer,
- * glDrawBuffer, glDrawBuffersARB, glFramebufferRenderbufferEXT,
- * glRenderbufferStorageEXT.
  */
 void
 _mesa_update_framebuffer(GLcontext *ctx)
 {
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_framebuffer *fbread = ctx->ReadBuffer;
+   struct gl_framebuffer *drawFb = ctx->DrawBuffer;
+   struct gl_framebuffer *readFb = ctx->ReadBuffer;
 
-   update_framebuffer(ctx, fb);
-   if (fbread != fb)
-      update_framebuffer(ctx, fbread);
+   update_framebuffer(ctx, drawFb);
+   if (readFb != drawFb)
+      update_framebuffer(ctx, readFb);
 }
 
 
diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c
index 4674c44a3f2..f72aa6a2882 100644
--- a/src/mesa/main/get.c
+++ b/src/mesa/main/get.c
@@ -1092,7 +1092,7 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params )
          GLuint i, n = _mesa_get_compressed_formats(ctx, formats, GL_FALSE);
          ASSERT(n <= 100);
          for (i = 0; i < n; i++)
-            params[i] = ENUM_TO_INT(formats[i]);
+            params[i] = ENUM_TO_BOOLEAN(formats[i]);
          }
          break;
       case GL_ARRAY_ELEMENT_LOCK_FIRST_EXT:
@@ -1873,6 +1873,10 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params )
          CHECK_EXT1(EXT_framebuffer_object, "GetBooleanv");
          params[0] = INT_TO_BOOLEAN(ctx->Const.MaxRenderbufferSize);
          break;
+      case GL_READ_FRAMEBUFFER_BINDING_EXT:
+         CHECK_EXT1(EXT_framebuffer_blit, "GetBooleanv");
+         params[0] = INT_TO_BOOLEAN(ctx->ReadBuffer->Name);
+         break;
       case GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB:
          CHECK_EXT1(ARB_fragment_shader, "GetBooleanv");
          params[0] = INT_TO_BOOLEAN(ctx->Const.FragmentProgram.MaxUniformComponents);
@@ -2936,7 +2940,7 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params )
          GLuint i, n = _mesa_get_compressed_formats(ctx, formats, GL_FALSE);
          ASSERT(n <= 100);
          for (i = 0; i < n; i++)
-            params[i] = ENUM_TO_INT(formats[i]);
+            params[i] = ENUM_TO_FLOAT(formats[i]);
          }
          break;
       case GL_ARRAY_ELEMENT_LOCK_FIRST_EXT:
@@ -3717,6 +3721,10 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params )
          CHECK_EXT1(EXT_framebuffer_object, "GetFloatv");
          params[0] = (GLfloat)(ctx->Const.MaxRenderbufferSize);
          break;
+      case GL_READ_FRAMEBUFFER_BINDING_EXT:
+         CHECK_EXT1(EXT_framebuffer_blit, "GetFloatv");
+         params[0] = (GLfloat)(ctx->ReadBuffer->Name);
+         break;
       case GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB:
          CHECK_EXT1(ARB_fragment_shader, "GetFloatv");
          params[0] = (GLfloat)(ctx->Const.FragmentProgram.MaxUniformComponents);
@@ -3977,7 +3985,7 @@ _mesa_GetIntegerv( GLenum pname, GLint *params )
          params[0] = ctx->DrawBuffer->Visual.depthBits;
          break;
       case GL_DEPTH_CLEAR_VALUE:
-         params[0] = IROUND(ctx->Depth.Clear);
+         params[0] = FLOAT_TO_INT(ctx->Depth.Clear);
          break;
       case GL_DEPTH_FUNC:
          params[0] = ENUM_TO_INT(ctx->Depth.Func);
@@ -5561,6 +5569,10 @@ _mesa_GetIntegerv( GLenum pname, GLint *params )
          CHECK_EXT1(EXT_framebuffer_object, "GetIntegerv");
          params[0] = ctx->Const.MaxRenderbufferSize;
          break;
+      case GL_READ_FRAMEBUFFER_BINDING_EXT:
+         CHECK_EXT1(EXT_framebuffer_blit, "GetIntegerv");
+         params[0] = ctx->ReadBuffer->Name;
+         break;
       case GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB:
          CHECK_EXT1(ARB_fragment_shader, "GetIntegerv");
          params[0] = ctx->Const.FragmentProgram.MaxUniformComponents;
diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py
index 653deefa726..152e378b4fe 100644
--- a/src/mesa/main/get_gen.py
+++ b/src/mesa/main/get_gen.py
@@ -50,7 +50,8 @@ TypeStrings = {
 #  - the GL state name, such as GL_CURRENT_COLOR
 #  - the state datatype, one of GLint, GLfloat, GLboolean or GLenum
 #  - list of code fragments to get the state, such as ["ctx->Foo.Bar"]
-#  - optional extra code or empty string
+#  - optional extra code or empty string.  If present, "CONVERSION" will be
+#    replaced by ENUM_TO_FLOAT, INT_TO_FLOAT, etc.
 #  - optional extensions to check, or None
 #
 StateVars = [
@@ -179,7 +180,7 @@ StateVars = [
 	( "GL_DEPTH_BIAS", GLfloat, ["ctx->Pixel.DepthBias"], "", None ),
 	( "GL_DEPTH_BITS", GLint, ["ctx->DrawBuffer->Visual.depthBits"],
 	  "", None ),
-	( "GL_DEPTH_CLEAR_VALUE", GLfloat, ["ctx->Depth.Clear"], "", None ),
+	( "GL_DEPTH_CLEAR_VALUE", GLfloatN, ["ctx->Depth.Clear"], "", None ),
 	( "GL_DEPTH_FUNC", GLenum, ["ctx->Depth.Func"], "", None ),
 	( "GL_DEPTH_RANGE", GLfloatN,
 	  [ "ctx->Viewport.Near", "ctx->Viewport.Far" ], "", None ),
@@ -532,7 +533,7 @@ StateVars = [
          GLuint i, n = _mesa_get_compressed_formats(ctx, formats, GL_FALSE);
          ASSERT(n <= 100);
          for (i = 0; i < n; i++)
-            params[i] = ENUM_TO_INT(formats[i]);""",
+            params[i] = CONVERSION(formats[i]);""",
 	  ["ARB_texture_compression"] ),
 
 	# GL_EXT_compiled_vertex_array
@@ -980,6 +981,11 @@ StateVars = [
 	  ["ctx->Const.MaxRenderbufferSize"], "",
 	  ["EXT_framebuffer_object"] ),
 
+	# GL_EXT_framebuffer_blit
+	# NOTE: GL_DRAW_FRAMEBUFFER_BINDING_EXT == GL_FRAMEBUFFER_BINDING_EXT
+	( "GL_READ_FRAMEBUFFER_BINDING_EXT", GLint, ["ctx->ReadBuffer->Name"], "",
+	  ["EXT_framebuffer_blit"] ),
+
 	# GL_ARB_fragment_shader
 	( "GL_MAX_FRAGMENT_UNIFORM_COMPONENTS_ARB", GLint,
 	  ["ctx->Const.FragmentProgram.MaxUniformComponents"], "",
@@ -1078,10 +1084,11 @@ def EmitGetFunction(stateVars, returnType):
 				assert len(extensions) == 4
 				print ('         CHECK_EXT4(%s, %s, %s, %s, "%s");' %
 					   (extensions[0], extensions[1], extensions[2], extensions[3], function))
+		conversion = ConversionFunc(varType, returnType)
 		if optionalCode:
+			optionalCode = string.replace(optionalCode, "CONVERSION", conversion);	
 			print "         {"
 			print "         " + optionalCode
-		conversion = ConversionFunc(varType, returnType)
 		n = len(state)
 		for i in range(n):
 			if conversion:
diff --git a/src/mesa/main/getstring.c b/src/mesa/main/getstring.c
index 89daa0e21db..94bf5de1e8b 100644
--- a/src/mesa/main/getstring.c
+++ b/src/mesa/main/getstring.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -25,7 +25,6 @@
 
 
 #include "glheader.h"
-#include "colormac.h"
 #include "context.h"
 #include "get.h"
 #include "version.h"
@@ -85,7 +84,7 @@ compute_version(const GLcontext *ctx)
                               ctx->Extensions.ARB_texture_non_power_of_two &&
                               ctx->Extensions.EXT_blend_equation_separate);
    const GLboolean ver_2_1 = (ver_2_0 &&
-                              ctx->Extensions.ARB_shading_language_120 &&
+                              /*ctx->Extensions.ARB_shading_language_120 &&*/
                               ctx->Extensions.EXT_pixel_buffer_object &&
                               ctx->Extensions.EXT_texture_sRGB);
    if (ver_2_1)
@@ -121,12 +120,6 @@ _mesa_GetString( GLenum name )
    static const char *vendor = "Brian Paul";
    static const char *renderer = "Mesa";
 
-#if FEATURE_ARB_shading_language_120_foo /* support not complete! */
-   static const char *sl_version = "1.20";
-#elif FEATURE_ARB_shading_language_100
-   static const char *sl_version = "1.10";
-#endif
-
    if (!ctx)
       return NULL;
 
@@ -154,8 +147,10 @@ _mesa_GetString( GLenum name )
          return (const GLubyte *) ctx->Extensions.String;
 #if FEATURE_ARB_shading_language_100
       case GL_SHADING_LANGUAGE_VERSION_ARB:
-         if (ctx->Extensions.ARB_shading_language_100)
-            return (const GLubyte *) sl_version;
+         if (ctx->Extensions.ARB_shading_language_120)
+            return (const GLubyte *) "1.20";
+         else if (ctx->Extensions.ARB_shading_language_100)
+            return (const GLubyte *) "1.10";
          goto error;
 #endif
 #if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program || \
diff --git a/src/mesa/main/glheader.h b/src/mesa/main/glheader.h
index f0f97c218c2..d69c7bbb21d 100644
--- a/src/mesa/main/glheader.h
+++ b/src/mesa/main/glheader.h
@@ -46,11 +46,6 @@
 #ifndef GLHEADER_H
 #define GLHEADER_H
 
-/* This allows Mesa to be integrated into XFree86 */
-#ifdef HAVE_DIX_CONFIG_H
-#include "dix-config.h"
-#endif
-
 #include <assert.h>
 #include <ctype.h>
 #if defined(__alpha__) && defined(CCPML)
@@ -98,6 +93,21 @@
 #  include <stdint.h>
 #endif
 
+
+/* Sun compilers define __i386 instead of the gcc-style __i386__ */
+#ifdef __SUNPRO_C
+# if !defined(__i386__) && defined(__i386)
+#  define __i386__
+# elif !defined(__amd64__) && defined(__amd64)
+#  define __amd64__
+# elif !defined(__sparc__) && defined(__sparc)
+#  define __sparc__
+# endif
+# if !defined(__volatile)
+#  define __volatile volatile
+# endif
+#endif
+
 #if defined(_WIN32) && !defined(__WIN32__) && !defined(__CYGWIN__) && !defined(BUILD_FOR_SNAP)
 #  define __WIN32__
 #  define finite _finite
@@ -236,6 +246,12 @@
 #  define INLINE inline
 #elif defined(__WATCOMC__) && (__WATCOMC__ >= 1100)
 #  define INLINE __inline
+#elif defined(__SUNPRO_C) && defined(__C99FEATURES__)
+#  define INLINE inline
+#  define __inline inline
+#  define __inline__ inline
+#elif (__STDC_VERSION__ >= 199901L) /* C99 */
+#  define INLINE inline
 #else
 #  define INLINE
 #endif
@@ -275,7 +291,7 @@
 #endif
 
 
-#if !defined __GNUC__ || __GNUC__ < 3
+#if (!defined(__GNUC__) || __GNUC__ < 3) && (!defined(__IBMC__) || __IBMC__ < 900)
 #  define __builtin_expect(x, y) x
 #endif
 
@@ -283,12 +299,17 @@
  * If we're not using gcc, define __FUNCTION__ as a cpp symbol here.
  * Don't define it if using a newer Windows compiler.
  */
-#if defined(__VMS)
-# define __FUNCTION__ "VMS$NL:"
-#elif __STDC_VERSION__ < 199901L
-# if ((!defined __GNUC__) || (__GNUC__ < 2)) && (!defined __xlC__) && \
+#ifndef __FUNCTION__
+# if defined(__VMS)
+#  define __FUNCTION__ "VMS$NL:"
+# elif ((!defined __GNUC__) || (__GNUC__ < 2)) && (!defined __xlC__) && \
       (!defined(_MSC_VER) || _MSC_VER < 1300)
-#  define __FUNCTION__ "<unknown>"
+#  if (__STDC_VERSION__ >= 199901L) /* C99 */ || \
+    (defined(__SUNPRO_C) && defined(__C99FEATURES__))
+#   define __FUNCTION__ __func__
+#  else
+#   define __FUNCTION__ "<unknown>"
+#  endif
 # endif
 #endif
 
diff --git a/src/mesa/main/histogram.c b/src/mesa/main/histogram.c
index 77c458d540c..905c1ad8301 100644
--- a/src/mesa/main/histogram.c
+++ b/src/mesa/main/histogram.c
@@ -955,7 +955,7 @@ _mesa_Histogram(GLenum target, GLsizei width, GLenum internalFormat, GLboolean s
       }
    }
 
-   if (width != 0 && _mesa_bitcount(width) != 1) {
+   if (width != 0 && !_mesa_is_pow_two(width)) {
       if (target == GL_PROXY_HISTOGRAM) {
          error = GL_TRUE;
       }
diff --git a/src/mesa/main/image.c b/src/mesa/main/image.c
index 52c4999e16e..d3282d42e99 100644
--- a/src/mesa/main/image.c
+++ b/src/mesa/main/image.c
@@ -849,7 +849,7 @@ _mesa_unpack_bitmap( GLint width, GLint height, const GLubyte *pixels,
          return NULL;
       }
 
-      if (packing->SkipPixels == 0) {
+      if ((packing->SkipPixels & 7) == 0) {
          _mesa_memcpy( dst, src, width_in_bytes );
          if (packing->LsbFirst) {
             flip_bytes( dst, width_in_bytes );
@@ -941,7 +941,7 @@ _mesa_pack_bitmap( GLint width, GLint height, const GLubyte *source,
       if (!dst)
          return;
 
-      if (packing->SkipPixels == 0) {
+      if ((packing->SkipPixels & 7) == 0) {
          _mesa_memcpy( dst, src, width_in_bytes );
          if (packing->LsbFirst) {
             flip_bytes( dst, width_in_bytes );
@@ -960,20 +960,20 @@ _mesa_pack_bitmap( GLint width, GLint height, const GLubyte *source,
                if (*s & srcMask) {
                   *d |= dstMask;
                }
-               if (srcMask == 128) {
-                  srcMask = 1;
+               if (srcMask == 1) {
+                  srcMask = 128;
                   s++;
                }
                else {
-                  srcMask = srcMask << 1;
+                  srcMask = srcMask >> 1;
                }
-               if (dstMask == 1) {
-                  dstMask = 128;
+               if (dstMask == 128) {
+                  dstMask = 1;
                   d++;
                   *d = 0;
                }
                else {
-                  dstMask = dstMask >> 1;
+                  dstMask = dstMask << 1;
                }
             }
          }
@@ -1689,7 +1689,7 @@ _mesa_pack_rgba_span_float(GLcontext *ctx, GLuint n, GLfloat rgba[][4],
 
    if (dstFormat == GL_LUMINANCE || dstFormat == GL_LUMINANCE_ALPHA) {
       /* compute luminance values */
-      if (dstType != GL_FLOAT || ctx->Color.ClampReadColor == GL_TRUE) {
+      if (transferOps & IMAGE_CLAMP_BIT) {
          for (i = 0; i < n; i++) {
             GLfloat sum = rgba[i][RCOMP] + rgba[i][GCOMP] + rgba[i][BCOMP];
             luminance[i] = CLAMP(sum, 0.0F, 1.0F);
@@ -4214,7 +4214,7 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
 
    switch (dstType) {
    case GL_UNSIGNED_BYTE:
-      if (sizeof(GLstencil) == 8) {
+      if (sizeof(GLstencil) == 1) {
          _mesa_memcpy( dest, source, n );
       }
       else {
@@ -4226,14 +4226,11 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
       }
       break;
    case GL_BYTE:
-      if (sizeof(GLstencil) == 8) {
-         _mesa_memcpy( dest, source, n );
-      }
-      else {
+      {
          GLbyte *dst = (GLbyte *) dest;
          GLuint i;
          for (i=0;i<n;i++) {
-            dst[i] = (GLbyte) source[i];
+            dst[i] = (GLbyte) (source[i] & 0x7f);
          }
       }
       break;
@@ -4278,7 +4275,7 @@ _mesa_pack_stencil_span( const GLcontext *ctx, GLuint n,
          GLint *dst = (GLint *) dest;
          GLuint i;
          for (i=0;i<n;i++) {
-            *dst++ = (GLint) source[i];
+            dst[i] = (GLint) source[i];
          }
          if (dstPacking->SwapBytes) {
             _mesa_swap4( (GLuint *) dst, n );
diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c
index 7231fa699cb..6cfd7ccc723 100644
--- a/src/mesa/main/imports.c
+++ b/src/mesa/main/imports.c
@@ -104,6 +104,8 @@ _mesa_align_malloc(size_t bytes, unsigned long alignment)
 
    (void) posix_memalign(& mem, alignment, bytes);
    return mem;
+#elif defined(_WIN32) && defined(_MSC_VER)
+   return _aligned_malloc(bytes, alignment);
 #else
    uintptr_t ptr, buf;
 
@@ -144,6 +146,15 @@ _mesa_align_calloc(size_t bytes, unsigned long alignment)
    }
 
    return mem;
+#elif defined(_WIN32) && defined(_MSC_VER)
+   void *mem;
+
+   mem = _aligned_malloc(bytes, alignment);
+   if (mem != NULL) {
+      (void) memset(mem, 0, bytes);
+   }
+
+   return mem;
 #else
    uintptr_t ptr, buf;
 
@@ -180,6 +191,8 @@ _mesa_align_free(void *ptr)
 {
 #if defined(HAVE_POSIX_MEMALIGN)
    free(ptr);
+#elif defined(_WIN32) && defined(_MSC_VER)
+   _aligned_free(ptr);
 #else
    void **cubbyHole = (void **) ((char *) ptr - sizeof(void *));
    void *realAddr = *cubbyHole;
@@ -194,6 +207,10 @@ void *
 _mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
                     unsigned long alignment)
 {
+#if defined(_WIN32) && defined(_MSC_VER)
+   (void) oldSize;
+   return _aligned_realloc(oldBuffer, newSize, alignment);
+#else
    const size_t copySize = (oldSize < newSize) ? oldSize : newSize;
    void *newBuf = _mesa_align_malloc(newSize, alignment);
    if (newBuf && oldBuffer && copySize > 0) {
@@ -202,6 +219,7 @@ _mesa_align_realloc(void *oldBuffer, size_t oldSize, size_t newSize,
    if (oldBuffer)
       _mesa_align_free(oldBuffer);
    return newBuf;
+#endif
 }
 
 
@@ -541,25 +559,28 @@ _mesa_pow(double x, double y)
 int
 _mesa_ffs(int32_t i)
 {
-#if (defined(_WIN32) && !defined(__MINGW32__) ) || defined(__IBMC__) || defined(__IBMCPP__)
-   register int32_t bit = 1;
-   if ((i & 0xffff) == 0) {
-      bit += 16;
-      i >>= 16;
-   }
-   if ((i & 0xff) == 0) {
-      bit += 8;
-      i >>= 8;
-   }
-   if ((i & 0xf) == 0) {
-      bit += 4;
-      i >>= 4;
-   }
-   if ((i & 0x3) == 0) {
-      bit += 2;
-      i >>= 2;
+#if (defined(_WIN32) ) || defined(__IBMC__) || defined(__IBMCPP__)
+   register int bit = 0;
+   if (i != 0) {
+      if ((i & 0xffff) == 0) {
+         bit += 16;
+         i >>= 16;
+      }
+      if ((i & 0xff) == 0) {
+         bit += 8;
+         i >>= 8;
+      }
+      if ((i & 0xf) == 0) {
+         bit += 4;
+         i >>= 4;
+      }
+      while ((i & 1) == 0) {
+         bit++;
+         i >>= 1;
+      }
+      bit++;
    }
-   return (i) ? (bit + ((i + 1) & 0x01)) : 0;
+   return bit;
 #else
    return ffs(i);
 #endif
diff --git a/src/mesa/main/imports.h b/src/mesa/main/imports.h
index d91a1366b07..bab00420710 100644
--- a/src/mesa/main/imports.h
+++ b/src/mesa/main/imports.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -159,12 +159,13 @@ typedef union { GLfloat f; GLint i; } fi_type;
  ***/
 #if defined(__i386__) || defined(__386__) || defined(__sparc__) || \
     defined(__s390x__) || defined(__powerpc__) || \
-    defined(__amd64__) || \
+    defined(__x86_64__) || \
     defined(ia64) || defined(__ia64__) || \
     defined(__hppa__) || defined(hpux) || \
     defined(__mips) || defined(_MIPS_ARCH) || \
     defined(__arm__) || \
     defined(__sh__) || defined(__m32r__) || \
+    (defined(__sun) && defined(_IEEE_754)) || \
     (defined(__alpha__) && (defined(__IEEE_FLOAT) || !defined(VMS)))
 #define USE_IEEE
 #define IEEE_ONE 0x3f800000
@@ -461,6 +462,16 @@ static INLINE int iceil(float f)
 #endif
 
 
+/**
+ * Is x a power of two?
+ */
+static INLINE int
+_mesa_is_pow_two(int x)
+{
+   return !(x & (x - 1));
+}
+
+
 /***
  *** UNCLAMPED_FLOAT_TO_UBYTE: clamp float to [0,1] and map to ubyte in [0,255]
  *** CLAMPED_FLOAT_TO_UBYTE: map float known to be in [0,1] to ubyte in [0,255]
diff --git a/src/mesa/main/light.c b/src/mesa/main/light.c
index f37d1f216fc..d4db960f1b9 100644
--- a/src/mesa/main/light.c
+++ b/src/mesa/main/light.c
@@ -53,6 +53,11 @@ _mesa_ShadeModel( GLenum mode )
 
    FLUSH_VERTICES(ctx, _NEW_LIGHT);
    ctx->Light.ShadeModel = mode;
+   if (mode == GL_FLAT)
+      ctx->_TriangleCaps |= DD_FLATSHADE;
+   else
+      ctx->_TriangleCaps &= ~DD_FLATSHADE;
+
    if (ctx->Driver.ShadeModel)
       ctx->Driver.ShadeModel( ctx, mode );
 }
@@ -179,6 +184,7 @@ _mesa_Lightfv( GLenum light, GLenum pname, const GLfloat *params )
    GET_CURRENT_CONTEXT(ctx);
    GLint i = (GLint) (light - GL_LIGHT0);
    GLfloat temp[4];
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    if (i < 0 || i >= (GLint) ctx->Const.MaxLights) {
       _mesa_error( ctx, GL_INVALID_ENUM, "glLight(light=0x%x)", light );
@@ -441,6 +447,10 @@ _mesa_LightModelfv( GLenum pname, const GLfloat *params )
 	    return;
 	 FLUSH_VERTICES(ctx, _NEW_LIGHT);
 	 ctx->Light.Model.TwoSide = newbool;
+         if (ctx->Light.Enabled && ctx->Light.Model.TwoSide)
+            ctx->_TriangleCaps |= DD_TRI_LIGHT_TWOSIDE;
+         else
+            ctx->_TriangleCaps &= ~DD_TRI_LIGHT_TWOSIDE;
          break;
       case GL_LIGHT_MODEL_COLOR_CONTROL:
          if (params[0] == (GLfloat) GL_SINGLE_COLOR)
diff --git a/src/mesa/main/lines.c b/src/mesa/main/lines.c
index 0c2dbf915a1..81d0d33abb4 100644
--- a/src/mesa/main/lines.c
+++ b/src/mesa/main/lines.c
@@ -56,6 +56,11 @@ _mesa_LineWidth( GLfloat width )
    FLUSH_VERTICES(ctx, _NEW_LINE);
    ctx->Line.Width = width;
 
+   if (width != 1.0F)
+      ctx->_TriangleCaps |= DD_LINE_WIDTH;
+   else
+      ctx->_TriangleCaps &= ~DD_LINE_WIDTH;
+
    if (ctx->Driver.LineWidth)
       ctx->Driver.LineWidth(ctx, width);
 }
diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h
index 487493f88e1..3819da3d680 100644
--- a/src/mesa/main/mfeatures.h
+++ b/src/mesa/main/mfeatures.h
@@ -44,7 +44,6 @@
 #define FEATURE_dlist  _HAVE_FULL_GL
 #define FEATURE_draw_read_buffer  _HAVE_FULL_GL
 #define FEATURE_drawpix  _HAVE_FULL_GL
-#define FEATURE_es2_glsl 0
 #define FEATURE_evaluators  _HAVE_FULL_GL
 #define FEATURE_feedback  _HAVE_FULL_GL
 #define FEATURE_fixedpt 0
@@ -56,6 +55,8 @@
 #define FEATURE_texture_s3tc  _HAVE_FULL_GL
 #define FEATURE_userclip  _HAVE_FULL_GL
 #define FEATURE_vertex_array_byte 0
+#define FEATURE_windowpos  _HAVE_FULL_GL
+#define FEATURE_es2_glsl 0
 
 #define FEATURE_ARB_occlusion_query  _HAVE_FULL_GL
 #define FEATURE_ARB_fragment_program  _HAVE_FULL_GL
diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c
index 061378f3b77..13d90e78fe4 100644
--- a/src/mesa/main/mipmap.c
+++ b/src/mesa/main/mipmap.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -446,34 +446,27 @@ make_1d_mipmap(GLenum datatype, GLuint comps, GLint border,
 }
 
 
-/**
- * Strides are in bytes.  If zero, it'll be computed as width * bpp.
- */
 static void
 make_2d_mipmap(GLenum datatype, GLuint comps, GLint border,
                GLint srcWidth, GLint srcHeight,
-               GLint srcRowStride, const GLubyte *srcPtr,
+	       const GLubyte *srcPtr, GLint srcRowStride,
                GLint dstWidth, GLint dstHeight,
-               GLint dstRowStride, GLubyte *dstPtr)
+	       GLubyte *dstPtr, GLint dstRowStride)
 {
    const GLint bpt = bytes_per_pixel(datatype, comps);
    const GLint srcWidthNB = srcWidth - 2 * border;  /* sizes w/out border */
    const GLint dstWidthNB = dstWidth - 2 * border;
    const GLint dstHeightNB = dstHeight - 2 * border;
+   const GLint srcRowBytes = bpt * srcRowStride;
+   const GLint dstRowBytes = bpt * dstRowStride;
    const GLubyte *srcA, *srcB;
    GLubyte *dst;
    GLint row;
 
-   if (!srcRowStride)
-      srcRowStride = bpt * srcWidth;
-
-   if (!dstRowStride)
-      dstRowStride = bpt * dstWidth;
-
    /* Compute src and dst pointers, skipping any border */
    srcA = srcPtr + border * ((srcWidth + 1) * bpt);
    if (srcHeight > 1) 
-      srcB = srcA + srcRowStride;
+      srcB = srcA + srcRowBytes;
    else
       srcB = srcA;
    dst = dstPtr + border * ((dstWidth + 1) * bpt);
@@ -481,9 +474,9 @@ make_2d_mipmap(GLenum datatype, GLuint comps, GLint border,
    for (row = 0; row < dstHeightNB; row++) {
       do_row(datatype, comps, srcWidthNB, srcA, srcB,
              dstWidthNB, dst);
-      srcA += 2 * srcRowStride;
-      srcB += 2 * srcRowStride;
-      dst += dstRowStride;
+      srcA += 2 * srcRowBytes;
+      srcB += 2 * srcRowBytes;
+      dst += dstRowBytes;
    }
 
    /* This is ugly but probably won't be used much */
@@ -541,11 +534,9 @@ make_2d_mipmap(GLenum datatype, GLuint comps, GLint border,
 static void
 make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
                GLint srcWidth, GLint srcHeight, GLint srcDepth,
-               GLint srcRowStride,
-               const GLubyte *srcPtr,
+               const GLubyte *srcPtr, GLint srcRowStride,
                GLint dstWidth, GLint dstHeight, GLint dstDepth,
-               GLint dstRowStride,
-               GLubyte *dstPtr)
+               GLubyte *dstPtr, GLint dstRowStride)
 {
    const GLint bpt = bytes_per_pixel(datatype, comps);
    const GLint srcWidthNB = srcWidth - 2 * border;  /* sizes w/out border */
@@ -556,6 +547,7 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
    GLvoid *tmpRowA, *tmpRowB;
    GLint img, row;
    GLint bytesPerSrcImage, bytesPerDstImage;
+   GLint bytesPerSrcRow, bytesPerDstRow;
    GLint srcImageOffset, srcRowOffset;
 
    (void) srcDepthNB; /* silence warnings */
@@ -573,10 +565,8 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
    bytesPerSrcImage = srcWidth * srcHeight * bpt;
    bytesPerDstImage = dstWidth * dstHeight * bpt;
 
-   if (!srcRowStride)
-      srcRowStride = srcWidth * bpt;
-   if (!dstRowStride)
-      dstRowStride = dstWidth * bpt;
+   bytesPerSrcRow = srcWidth * bpt;
+   bytesPerDstRow = dstWidth * bpt;
 
    /* Offset between adjacent src images to be averaged together */
    srcImageOffset = (srcDepth == dstDepth) ? 0 : bytesPerSrcImage;
@@ -600,13 +590,13 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
    for (img = 0; img < dstDepthNB; img++) {
       /* first source image pointer, skipping border */
       const GLubyte *imgSrcA = srcPtr
-         + (bytesPerSrcImage + srcRowStride + border) * bpt * border
+         + (bytesPerSrcImage + bytesPerSrcRow + border) * bpt * border
          + img * (bytesPerSrcImage + srcImageOffset);
       /* second source image pointer, skipping border */
       const GLubyte *imgSrcB = imgSrcA + srcImageOffset;
       /* address of the dest image, skipping border */
       GLubyte *imgDst = dstPtr
-         + (bytesPerDstImage + dstRowStride + border) * bpt * border
+         + (bytesPerDstImage + bytesPerDstRow + border) * bpt * border
          + img * bytesPerDstImage;
 
       /* setup the four source row pointers and the dest row pointer */
@@ -627,11 +617,11 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
          do_row(datatype, comps, srcWidthNB, tmpRowA, tmpRowB,
                 dstWidthNB, dstImgRow);
          /* advance to next rows */
-         srcImgARowA += srcRowStride + srcRowOffset;
-         srcImgARowB += srcRowStride + srcRowOffset;
-         srcImgBRowA += srcRowStride + srcRowOffset;
-         srcImgBRowB += srcRowStride + srcRowOffset;
-         dstImgRow += dstRowStride;
+         srcImgARowA += bytesPerSrcRow + srcRowOffset;
+         srcImgARowB += bytesPerSrcRow + srcRowOffset;
+         srcImgBRowA += bytesPerSrcRow + srcRowOffset;
+         srcImgBRowB += bytesPerSrcRow + srcRowOffset;
+         dstImgRow += bytesPerDstRow;
       }
    }
 
@@ -641,15 +631,13 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
    /* Luckily we can leverage the make_2d_mipmap() function here! */
    if (border > 0) {
       /* do front border image */
-      make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, 0, srcPtr,
-                     dstWidth, dstHeight, 0, dstPtr);
+      make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight, srcPtr, srcRowStride,
+                     dstWidth, dstHeight, dstPtr, dstRowStride);
       /* do back border image */
       make_2d_mipmap(datatype, comps, 1, srcWidth, srcHeight,
-                     0, 
-                     srcPtr + bytesPerSrcImage * (srcDepth - 1),
+                     srcPtr + bytesPerSrcImage * (srcDepth - 1), srcRowStride,
                      dstWidth, dstHeight,
-                     0,
-                     dstPtr + bytesPerDstImage * (dstDepth - 1));
+                     dstPtr + bytesPerDstImage * (dstDepth - 1), dstRowStride);
       /* do four remaining border edges that span the image slices */
       if (srcDepth == dstDepth) {
          /* just copy border pixels from src to dst */
@@ -664,9 +652,9 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
 
             /* do border along [img][row=dstHeight-1][col=0] */
             src = srcPtr + (img * 2 + 1) * bytesPerSrcImage
-                         + (srcHeight - 1) * srcRowStride;
+                         + (srcHeight - 1) * bytesPerSrcRow;
             dst = dstPtr + (img + 1) * bytesPerDstImage
-                         + (dstHeight - 1) * dstRowStride;
+                         + (dstHeight - 1) * bytesPerDstRow;
             MEMCPY(dst, src, bpt);
 
             /* do border along [img][row=0][col=dstWidth-1] */
@@ -698,9 +686,9 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
 
             /* do border along [img][row=dstHeight-1][col=0] */
             src = srcPtr + (img * 2 + 1) * bytesPerSrcImage
-                         + (srcHeight - 1) * srcRowStride;
+                         + (srcHeight - 1) * bytesPerSrcRow;
             dst = dstPtr + (img + 1) * bytesPerDstImage
-                         + (dstHeight - 1) * dstRowStride;
+                         + (dstHeight - 1) * bytesPerDstRow;
             do_row(datatype, comps, 1, src, src + srcImageOffset, 1, dst);
 
             /* do border along [img][row=0][col=dstWidth-1] */
@@ -724,15 +712,16 @@ make_3d_mipmap(GLenum datatype, GLuint comps, GLint border,
 
 static void
 make_1d_stack_mipmap(GLenum datatype, GLuint comps, GLint border,
-                     GLint srcWidth, const GLubyte *srcPtr,
-                     GLint dstWidth, GLint dstHeight, GLubyte *dstPtr)
+                     GLint srcWidth, const GLubyte *srcPtr, GLuint srcRowStride,
+                     GLint dstWidth, GLint dstHeight,
+		     GLubyte *dstPtr, GLuint dstRowStride )
 {
    const GLint bpt = bytes_per_pixel(datatype, comps);
    const GLint srcWidthNB = srcWidth - 2 * border;  /* sizes w/out border */
    const GLint dstWidthNB = dstWidth - 2 * border;
    const GLint dstHeightNB = dstHeight - 2 * border;
-   const GLint srcRowStride = bpt * srcWidth;
-   const GLint dstRowStride = bpt * dstWidth;
+   const GLint srcRowBytes = bpt * srcRowStride;
+   const GLint dstRowBytes = bpt * dstRowStride;
    const GLubyte *src;
    GLubyte *dst;
    GLint row;
@@ -744,8 +733,8 @@ make_1d_stack_mipmap(GLenum datatype, GLuint comps, GLint border,
    for (row = 0; row < dstHeightNB; row++) {
       do_row(datatype, comps, srcWidthNB, src, src,
              dstWidthNB, dst);
-      src += srcRowStride;
-      dst += dstRowStride;
+      src += srcRowBytes;
+      dst += dstRowBytes;
    }
 
    if (border) {
@@ -767,32 +756,26 @@ make_1d_stack_mipmap(GLenum datatype, GLuint comps, GLint border,
 static void
 make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border,
                      GLint srcWidth, GLint srcHeight,
-                     GLint srcRowStride,
-                     const GLubyte *srcPtr,
+		     const GLubyte *srcPtr, GLint srcRowStride,
                      GLint dstWidth, GLint dstHeight, GLint dstDepth,
-                     GLint dstRowStride,
-                     GLubyte *dstPtr)
+                     GLubyte *dstPtr, GLint dstRowStride)
 {
    const GLint bpt = bytes_per_pixel(datatype, comps);
    const GLint srcWidthNB = srcWidth - 2 * border;  /* sizes w/out border */
    const GLint dstWidthNB = dstWidth - 2 * border;
    const GLint dstHeightNB = dstHeight - 2 * border;
    const GLint dstDepthNB = dstDepth - 2 * border;
+   const GLint srcRowBytes = bpt * srcRowStride;
+   const GLint dstRowBytes = bpt * dstRowStride;
    const GLubyte *srcA, *srcB;
    GLubyte *dst;
    GLint layer;
    GLint row;
 
-   if (!srcRowStride)
-      srcRowStride = bpt * srcWidth;
-
-   if (!dstRowStride)
-      dstRowStride = bpt * dstWidth;
-
    /* Compute src and dst pointers, skipping any border */
    srcA = srcPtr + border * ((srcWidth + 1) * bpt);
    if (srcHeight > 1) 
-      srcB = srcA + srcRowStride;
+      srcB = srcA + srcRowBytes;
    else
       srcB = srcA;
    dst = dstPtr + border * ((dstWidth + 1) * bpt);
@@ -801,9 +784,9 @@ make_2d_stack_mipmap(GLenum datatype, GLuint comps, GLint border,
       for (row = 0; row < dstHeightNB; row++) {
          do_row(datatype, comps, srcWidthNB, srcA, srcB,
                 dstWidthNB, dst);
-         srcA += 2 * srcRowStride;
-         srcB += 2 * srcRowStride;
-         dst += dstRowStride;
+         srcA += 2 * srcRowBytes;
+         srcB += 2 * srcRowBytes;
+         dst += dstRowBytes;
       }
 
       /* This is ugly but probably won't be used much */
@@ -867,12 +850,15 @@ _mesa_generate_mipmap_level(GLenum target,
                             GLenum datatype, GLuint comps,
                             GLint border,
                             GLint srcWidth, GLint srcHeight, GLint srcDepth,
-                            GLint srcRowStride,
                             const GLubyte *srcData,
+                            GLint srcRowStride,
                             GLint dstWidth, GLint dstHeight, GLint dstDepth,
-                            GLint dstRowStride,
-                            GLubyte *dstData)
+                            GLubyte *dstData,
+                            GLint dstRowStride)
 {
+   /*
+    * We use simple 2x2 averaging to compute the next mipmap level.
+    */
    switch (target) {
    case GL_TEXTURE_1D:
       make_1d_mipmap(datatype, comps, border,
@@ -887,29 +873,35 @@ _mesa_generate_mipmap_level(GLenum target,
    case GL_TEXTURE_CUBE_MAP_POSITIVE_Z_ARB:
    case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z_ARB:
       make_2d_mipmap(datatype, comps, border,
-                     srcWidth, srcHeight, srcRowStride, srcData,
-                     dstWidth, dstHeight, dstRowStride, dstData);
+                     srcWidth, srcHeight, srcData, srcRowStride,
+                     dstWidth, dstHeight, dstData, dstRowStride);
       break;
    case GL_TEXTURE_3D:
       make_3d_mipmap(datatype, comps, border,
-                     srcWidth, srcHeight, srcDepth, srcRowStride, srcData,
-                     dstWidth, dstHeight, dstDepth, dstRowStride, dstData);
+                     srcWidth, srcHeight, srcDepth,
+                     srcData, srcRowStride,
+                     dstWidth, dstHeight, dstDepth,
+                     dstData, dstRowStride);
       break;
    case GL_TEXTURE_1D_ARRAY_EXT:
       make_1d_stack_mipmap(datatype, comps, border,
-                           srcWidth, srcData,
-                           dstWidth, dstHeight, dstData);
+                           srcWidth, srcData, srcRowStride,
+                           dstWidth, dstHeight,
+                           dstData, dstRowStride);
       break;
    case GL_TEXTURE_2D_ARRAY_EXT:
       make_2d_stack_mipmap(datatype, comps, border,
-                           srcWidth, srcHeight, srcRowStride, srcData,
-                           dstWidth, dstHeight, dstDepth, dstRowStride, dstData);
+                           srcWidth, srcHeight,
+                           srcData, srcRowStride,
+                           dstWidth, dstHeight,
+                           dstDepth, dstData, dstRowStride);
       break;
    case GL_TEXTURE_RECTANGLE_NV:
       /* no mipmaps, do nothing */
       break;
    default:
-      _mesa_problem(NULL, "bad target in _mesa_generate_mipmap_level");
+      _mesa_problem(NULL, "bad dimensions in _mesa_generate_mipmaps");
+      return;
    }
 }
 
@@ -991,7 +983,8 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target,
       GLint  components, size;
       GLchan *dst;
 
-      assert(texObj->Target == GL_TEXTURE_2D);
+      assert(texObj->Target == GL_TEXTURE_2D ||
+             texObj->Target == GL_TEXTURE_CUBE_MAP_ARB);
 
       if (srcImage->_BaseFormat == GL_RGB) {
          convertFormat = &_mesa_texformat_rgb;
@@ -1130,10 +1123,12 @@ _mesa_generate_mipmap(GLcontext *ctx, GLenum target,
          dstData = (GLubyte *) dstImage->Data;
       }
 
-      /* Note, 0 indicates default row strides */
       _mesa_generate_mipmap_level(target, datatype, comps, border,
-                                  srcWidth, srcHeight, srcDepth, 0, srcData,
-                                  dstWidth, dstHeight, dstDepth, 0, dstData);
+                                  srcWidth, srcHeight, srcDepth, 
+                                  srcData, srcImage->RowStride,
+                                  dstWidth, dstHeight, dstDepth, 
+                                  dstData, dstImage->RowStride);
+
 
       if (dstImage->IsCompressed) {
          GLubyte *temp;
diff --git a/src/mesa/main/mipmap.h b/src/mesa/main/mipmap.h
index 44ecdddb277..22094c34372 100644
--- a/src/mesa/main/mipmap.h
+++ b/src/mesa/main/mipmap.h
@@ -34,11 +34,11 @@ _mesa_generate_mipmap_level(GLenum target,
                             GLenum datatype, GLuint comps,
                             GLint border,
                             GLint srcWidth, GLint srcHeight, GLint srcDepth,
-                            GLint srcRowStride,
                             const GLubyte *srcData,
+                            GLint srcRowStride,
                             GLint dstWidth, GLint dstHeight, GLint dstDepth,
-                            GLint dstRowStride,
-                            GLubyte *dstData);
+                            GLubyte *dstData,
+                            GLint dstRowStride);
 
 
 extern void
diff --git a/src/mesa/main/mm.c b/src/mesa/main/mm.c
index 6f381b02a7f..d430bcdb840 100644
--- a/src/mesa/main/mm.c
+++ b/src/mesa/main/mm.c
@@ -167,7 +167,7 @@ mmAllocMem(struct mem_block *heap, unsigned size, unsigned align2, unsigned star
    unsigned startofs = 0;
    unsigned endofs;
 
-   if (!heap || !align2 || !size)
+   if (!heap || !size)
       return NULL;
 
    for (p = heap->next_free; p != heap; p = p->next_free) {
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 732de2bb432..556b4ed0164 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1369,7 +1369,7 @@ struct gl_texture_image
    GLboolean IsCompressed;	/**< GL_ARB_texture_compression */
    GLuint CompressedSize;	/**< GL_ARB_texture_compression */
 
-   GLuint RowStride;		/**< == Width unless IsClientData and padded */
+   GLuint RowStride;		/**< Padded width in units of texels */
    GLuint *ImageOffsets;        /**< if 3D texture: array [Depth] of offsets to
                                      each 2D slice in 'Data', in texels */
    GLvoid *Data;		/**< Image data, accessed via FetchTexel() */
@@ -1540,17 +1540,6 @@ struct gl_texture_unit
 
    struct gl_texture_object *_Current; /**< Points to really enabled tex obj */
 
-   /** These are used for glPush/PopAttrib */
-   /*@{*/
-   struct gl_texture_object Saved1D;
-   struct gl_texture_object Saved2D;
-   struct gl_texture_object Saved3D;
-   struct gl_texture_object SavedCubeMap;
-   struct gl_texture_object SavedRect;
-   struct gl_texture_object Saved1DArray;
-   struct gl_texture_object Saved2DArray;
-   /*@}*/
-
    /** GL_SGI_texture_color_table */
    /*@{*/
    struct gl_color_table ColorTable;
@@ -1560,6 +1549,7 @@ struct gl_texture_unit
 };
 
 
+
 /**
  * Texture attribute group (GL_TEXTURE_BIT).
  */
@@ -1579,17 +1569,7 @@ struct gl_texture_attrib
 
    struct gl_texture_unit Unit[MAX_TEXTURE_UNITS];
 
-#if 0
-   struct gl_texture_object *Proxy1D;
-   struct gl_texture_object *Proxy2D;
-   struct gl_texture_object *Proxy3D;
-   struct gl_texture_object *ProxyCubeMap;
-   struct gl_texture_object *ProxyRect;
-   struct gl_texture_object *Proxy1DArray;
-   struct gl_texture_object *Proxy2DArray;
-#else
    struct gl_texture_object *ProxyTex[NUM_TEXTURE_TARGETS];
-#endif
 
    /** GL_EXT_shared_texture_palette */
    GLboolean SharedPalette;
@@ -1894,10 +1874,13 @@ struct gl_program
 
    GLbitfield InputsRead;     /**< Bitmask of which input regs are read */
    GLbitfield OutputsWritten; /**< Bitmask of which output regs are written to */
+   GLbitfield InputFlags[MAX_PROGRAM_INPUTS];   /**< PROG_PARAM_BIT_x flags */
+   GLbitfield OutputFlags[MAX_PROGRAM_OUTPUTS]; /**< PROG_PARAM_BIT_x flags */
    GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS];  /**< TEXTURE_x_BIT bitmask */
    GLbitfield SamplersUsed;   /**< Bitfield of which samplers are used */
    GLbitfield ShadowSamplers; /**< Texture units used for shadow sampling. */
 
+
    /** Named parameters, constants, etc. from program text */
    struct gl_program_parameter_list *Parameters;
    /** Numbered local parameters */
@@ -2428,13 +2411,11 @@ struct gl_framebuffer
    GLenum ColorDrawBuffer[MAX_DRAW_BUFFERS];
    GLenum ColorReadBuffer;
 
-   /* These are computed from ColorDrawBuffer and ColorReadBuffer */
-   GLbitfield _ColorDrawBufferMask[MAX_DRAW_BUFFERS]; /* Mask of BUFFER_BIT_* flags */
+   /** Computed from ColorDraw/ReadBuffer above */
+   GLuint _NumColorDrawBuffers;
+   GLint _ColorDrawBufferIndexes[MAX_DRAW_BUFFERS]; /**< BUFFER_x or -1 */
    GLint _ColorReadBufferIndex; /* -1 = None */
-
-   /* These are computed from _ColorDrawBufferMask and _ColorReadBufferIndex */
-   GLuint _NumColorDrawBuffers[MAX_DRAW_BUFFERS];
-   struct gl_renderbuffer *_ColorDrawBuffers[MAX_DRAW_BUFFERS][4];
+   struct gl_renderbuffer *_ColorDrawBuffers[MAX_DRAW_BUFFERS];
    struct gl_renderbuffer *_ColorReadBuffer;
 
    /** The Actual depth/stencil buffers to use.  May be wrappers around the
@@ -2652,7 +2633,6 @@ struct gl_extensions
    GLboolean SGIS_generate_mipmap;
    GLboolean SGIS_texture_edge_clamp;
    GLboolean SGIS_texture_lod;
-   GLboolean SGIX_depth_texture;
    GLboolean SGIX_shadow;
    GLboolean SGIX_shadow_ambient; /* or GL_ARB_shadow_ambient */
    GLboolean TDFX_texture_compression_FXT1;
@@ -2902,7 +2882,6 @@ struct mesa_display_list
  */
 struct gl_dlist_state
 {
-   struct mesa_display_list *CallStack[MAX_LIST_NESTING];
    GLuint CallDepth;		/**< Current recursion calling depth */
 
    struct mesa_display_list *CurrentList;
@@ -3079,7 +3058,10 @@ struct __GLcontextRec
 
    /** \name Derived state */
    /*@{*/
-   GLbitfield _TriangleCaps;      /**< bitwise-or of DD_* flags */
+   /** Bitwise-or of DD_* flags.  Note that this bitfield may be used before
+    * state validation so they need to always be current.
+    */
+   GLbitfield _TriangleCaps;
    GLbitfield _ImageTransferState;/**< bitwise-or of IMAGE_*_BIT flags */
    GLfloat _EyeZDir[3];
    GLfloat _ModelViewInvScale;
@@ -3095,12 +3077,6 @@ struct __GLcontextRec
 
    struct gl_list_extensions ListExt; /**< driver dlist extensions */
 
-
-   GLuint _Facing; /**< This is a hack for 2-sided stencil test.
-		    *
-		    * We don't have a better way to communicate this value from
-		    * swrast_setup to swrast. */
-
    /** \name For debugging/development only */
    /*@{*/
    GLboolean FirstTimeCurrent;
diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c
index c98506b2bb0..8d24a201f07 100644
--- a/src/mesa/main/pixel.c
+++ b/src/mesa/main/pixel.c
@@ -150,7 +150,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values )
 
    if (map >= GL_PIXEL_MAP_S_TO_S && map <= GL_PIXEL_MAP_I_TO_A) {
       /* test that mapsize is a power of two */
-      if (_mesa_bitcount((GLuint) mapsize) != 1) {
+      if (!_mesa_is_pow_two(mapsize)) {
 	 _mesa_error( ctx, GL_INVALID_VALUE, "glPixelMapfv(mapsize)" );
          return;
       }
@@ -209,7 +209,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values )
 
    if (map >= GL_PIXEL_MAP_S_TO_S && map <= GL_PIXEL_MAP_I_TO_A) {
       /* test that mapsize is a power of two */
-      if (_mesa_bitcount((GLuint) mapsize) != 1) {
+      if (!_mesa_is_pow_two(mapsize)) {
 	 _mesa_error( ctx, GL_INVALID_VALUE, "glPixelMapuiv(mapsize)" );
          return;
       }
@@ -282,7 +282,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values )
 
    if (map >= GL_PIXEL_MAP_S_TO_S && map <= GL_PIXEL_MAP_I_TO_A) {
       /* test that mapsize is a power of two */
-      if (_mesa_bitcount((GLuint) mapsize) != 1) {
+      if (!_mesa_is_pow_two(mapsize)) {
 	 _mesa_error( ctx, GL_INVALID_VALUE, "glPixelMapuiv(mapsize)" );
          return;
       }
diff --git a/src/mesa/main/points.c b/src/mesa/main/points.c
index fbedbcb22c4..1fe697033f1 100644
--- a/src/mesa/main/points.c
+++ b/src/mesa/main/points.c
@@ -5,7 +5,7 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  7.0
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -77,8 +77,13 @@ _mesa_PointParameteri( GLenum pname, GLint param )
 void GLAPIENTRY
 _mesa_PointParameteriv( GLenum pname, const GLint *params )
 {
-   const GLfloat value = (GLfloat) params[0];
-   _mesa_PointParameterfv(pname, &value);
+   GLfloat p[3];
+   p[0] = (GLfloat) params[0];
+   if (pname == GL_DISTANCE_ATTENUATION_EXT) {
+      p[1] = (GLfloat) params[1];
+      p[2] = (GLfloat) params[2];
+   }
+   _mesa_PointParameterfv(pname, p);
 }
 
 
@@ -105,6 +110,11 @@ _mesa_PointParameterfv( GLenum pname, const GLfloat *params)
             ctx->Point._Attenuated = (ctx->Point.Params[0] != 1.0 ||
                                       ctx->Point.Params[1] != 0.0 ||
                                       ctx->Point.Params[2] != 0.0);
+
+            if (ctx->Point._Attenuated)
+               ctx->_TriangleCaps |= DD_POINT_ATTEN;
+            else
+               ctx->_TriangleCaps &= ~DD_POINT_ATTEN;
          }
          else {
             _mesa_error(ctx, GL_INVALID_ENUM,
@@ -190,7 +200,7 @@ _mesa_PointParameterfv( GLenum pname, const GLfloat *params)
          }
          break;
       case GL_POINT_SPRITE_COORD_ORIGIN:
-         if (ctx->Extensions.ARB_point_sprite) {
+         if (ctx->Extensions.ARB_point_sprite || ctx->Extensions.NV_point_sprite) {
             GLenum value = (GLenum) params[0];
             if (value != GL_LOWER_LEFT && value != GL_UPPER_LEFT) {
                _mesa_error(ctx, GL_INVALID_VALUE,
diff --git a/src/mesa/main/polygon.c b/src/mesa/main/polygon.c
index fd02e5a652d..564250b881c 100644
--- a/src/mesa/main/polygon.c
+++ b/src/mesa/main/polygon.c
@@ -167,6 +167,11 @@ _mesa_PolygonMode( GLenum face, GLenum mode )
       return;
    }
 
+   if (ctx->Polygon.FrontMode == GL_FILL && ctx->Polygon.BackMode == GL_FILL)
+      ctx->_TriangleCaps &= ~DD_TRI_UNFILLED;
+   else
+      ctx->_TriangleCaps |= DD_TRI_UNFILLED;
+
    if (ctx->Driver.PolygonMode)
       ctx->Driver.PolygonMode(ctx, face, mode);
 }
diff --git a/src/mesa/main/queryobj.c b/src/mesa/main/queryobj.c
index 2d060300300..554e0b0d181 100644
--- a/src/mesa/main/queryobj.c
+++ b/src/mesa/main/queryobj.c
@@ -90,9 +90,8 @@ _mesa_wait_query(GLcontext *ctx, struct gl_query_object *q)
 
 
 /**
- * Delete an occlusion query object.
+ * Delete a query object.  Called via ctx->Driver.DeleteQuery().
  * Not removed from hash table here.
- * XXX maybe add Delete() method to gl_query_object class and call that instead
  */
 void
 _mesa_delete_query(GLcontext *ctx, struct gl_query_object *q)
@@ -546,6 +545,6 @@ delete_queryobj_cb(GLuint id, void *data, void *userData)
 void
 _mesa_free_query_data(GLcontext *ctx)
 {
-   _mesa_HashDeleteAll(ctx->Query.QueryObjects, delete_queryobj_cb, NULL);
+   _mesa_HashDeleteAll(ctx->Query.QueryObjects, delete_queryobj_cb, ctx);
    _mesa_DeleteHashTable(ctx->Query.QueryObjects);
 }
diff --git a/src/mesa/main/rastpos.c b/src/mesa/main/rastpos.c
index 155140f3ccc..9842172f460 100644
--- a/src/mesa/main/rastpos.c
+++ b/src/mesa/main/rastpos.c
@@ -50,12 +50,12 @@ rasterpos(GLfloat x, GLfloat y, GLfloat z, GLfloat w)
    p[2] = z;
    p[3] = w;
 
-   if (ctx->NewState)
-      _mesa_update_state( ctx );
-
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
    FLUSH_CURRENT(ctx, 0);
 
+   if (ctx->NewState)
+      _mesa_update_state( ctx );
+
    ctx->Driver.RasterPos(ctx, p);
 }
 
diff --git a/src/mesa/main/readpix.c b/src/mesa/main/readpix.c
index dc22808e5bb..dfdd297b6e7 100644
--- a/src/mesa/main/readpix.c
+++ b/src/mesa/main/readpix.c
@@ -105,8 +105,7 @@ _mesa_error_check_format_type(GLcontext *ctx, GLenum format, GLenum type,
       }
       break;
    case GL_DEPTH_COMPONENT:
-      if ((drawing && !_mesa_dest_buffer_exists(ctx, format)) ||
-          (!drawing && !_mesa_source_buffer_exists(ctx, format))) {
+      if ((drawing && !_mesa_dest_buffer_exists(ctx, format))) {
          _mesa_error(ctx, GL_INVALID_OPERATION,
                      "gl%sPixels(no depth buffer)", readDraw);
          return GL_TRUE;
diff --git a/src/mesa/main/shaders.c b/src/mesa/main/shaders.c
index f0db0d2a818..7491d00c357 100644
--- a/src/mesa/main/shaders.c
+++ b/src/mesa/main/shaders.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.3
  *
  * Copyright (C) 2004-2008  Brian Paul   All Rights Reserved.
  *
@@ -233,13 +233,23 @@ _mesa_GetObjectParameterivARB(GLhandleARB object, GLenum pname, GLint *params)
    GET_CURRENT_CONTEXT(ctx);
    /* Implement in terms of GetProgramiv, GetShaderiv */
    if (ctx->Driver.IsProgram(ctx, object)) {
-      ctx->Driver.GetProgramiv(ctx, object, pname, params);
+      if (pname == GL_OBJECT_TYPE_ARB) {
+	 *params = GL_PROGRAM_OBJECT_ARB;
+      }
+      else {
+	 ctx->Driver.GetProgramiv(ctx, object, pname, params);
+      }
    }
    else if (ctx->Driver.IsShader(ctx, object)) {
-      ctx->Driver.GetShaderiv(ctx, object, pname, params);
+      if (pname == GL_OBJECT_TYPE_ARB) {
+	 *params = GL_SHADER_OBJECT_ARB;
+      }
+      else {
+	 ctx->Driver.GetShaderiv(ctx, object, pname, params);
+      }
    }
    else {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetObjectParameterivARB");
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetObjectParameterivARB");
    }
 }
 
diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c
index a962f1cb41d..0d452fd8791 100644
--- a/src/mesa/main/state.c
+++ b/src/mesa/main/state.c
@@ -51,6 +51,16 @@
 #include "texstate.h"
 
 
+static void
+update_separate_specular(GLcontext *ctx)
+{
+   if (NEED_SECONDARY_COLOR(ctx))
+      ctx->_TriangleCaps |= DD_SEPARATE_SPECULAR;
+   else
+      ctx->_TriangleCaps &= ~DD_SEPARATE_SPECULAR;
+}
+
+
 /**
  * Update state dependent on vertex arrays.
  */
@@ -176,7 +186,7 @@ update_program(GLcontext *ctx)
    ctx->FragmentProgram._Enabled = ctx->FragmentProgram.Enabled
       && ctx->FragmentProgram.Current->Base.Instructions;
    ctx->ATIFragmentShader._Enabled = ctx->ATIFragmentShader.Enabled
-      && ctx->ATIFragmentShader.Current->Instructions;
+      && ctx->ATIFragmentShader.Current->Instructions[0];
 
    /*
     * Set the ctx->VertexProgram._Current and ctx->FragmentProgram._Current
@@ -194,56 +204,39 @@ update_program(GLcontext *ctx)
     * come up, or matter.
     */
 
-   /**
-    ** Fragment program
-    **/
-#if 1
-   /* XXX get rid of this someday? */
-   ctx->FragmentProgram._Active = GL_FALSE;
-#endif
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL);
+
    if (shProg && shProg->LinkStatus && shProg->FragmentProgram) {
-      /* user-defined fragment shader */
+      /* Use shader programs */
       _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current,
                                shProg->FragmentProgram);
    }
    else if (ctx->FragmentProgram._Enabled) {
-      /* use user-defined fragment program */
+      /* use user-defined vertex program */
       _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current,
                                ctx->FragmentProgram.Current);
    }
    else if (ctx->FragmentProgram._MaintainTexEnvProgram) {
-      /* fragment program generated from fixed-function state */
+      /* Use fragment program generated from fixed-function state.
+       */
       _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current,
                                _mesa_get_fixed_func_fragment_program(ctx));
       _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram,
                                ctx->FragmentProgram._Current);
-
-      /* XXX get rid of this confusing stuff someday? */
-      ctx->FragmentProgram._Active = ctx->FragmentProgram._Enabled;
-      if (ctx->FragmentProgram._UseTexEnvProgram)
-         ctx->FragmentProgram._Active = GL_TRUE;
    }
    else {
       /* no fragment program */
       _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL);
    }
 
-   if (ctx->FragmentProgram._Current != prevFP && ctx->Driver.BindProgram) {
-      ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB,
-                         (struct gl_program *) ctx->FragmentProgram._Current);
-   }
-
-   /**
-    ** Vertex program
-    **/
-#if 1
-   /* XXX get rid of this someday? */
-   _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram, NULL);
-#endif
+   /* Examine vertex program after fragment program as
+    * _mesa_get_fixed_func_vertex_program() needs to know active
+    * fragprog inputs.
+    */
    if (shProg && shProg->LinkStatus && shProg->VertexProgram) {
-      /* user-defined vertex shader */
+      /* Use shader programs */
       _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current,
-                               shProg->VertexProgram);
+                            shProg->VertexProgram);
    }
    else if (ctx->VertexProgram._Enabled) {
       /* use user-defined vertex program */
@@ -251,20 +244,39 @@ update_program(GLcontext *ctx)
                                ctx->VertexProgram.Current);
    }
    else if (ctx->VertexProgram._MaintainTnlProgram) {
-      /* vertex program generated from fixed-function state */
+      /* Use vertex program generated from fixed-function state.
+       */
       _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current,
                                _mesa_get_fixed_func_vertex_program(ctx));
       _mesa_reference_vertprog(ctx, &ctx->VertexProgram._TnlProgram,
                                ctx->VertexProgram._Current);
    }
    else {
-      /* no vertex program / used fixed-function code */
+      /* no vertex program */
       _mesa_reference_vertprog(ctx, &ctx->VertexProgram._Current, NULL);
    }
 
+   /* XXX: get rid of _Active flag.
+    */
+#if 1
+   ctx->FragmentProgram._Active = ctx->FragmentProgram._Enabled;
+   if (ctx->FragmentProgram._MaintainTexEnvProgram &&
+       !ctx->FragmentProgram._Enabled) {
+      if (ctx->FragmentProgram._UseTexEnvProgram)
+	 ctx->FragmentProgram._Active = GL_TRUE;
+   }
+#endif
+
+   /* Let the driver know what's happening:
+    */
+   if (ctx->FragmentProgram._Current != prevFP && ctx->Driver.BindProgram) {
+      ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB,
+                              (struct gl_program *) ctx->FragmentProgram._Current);
+   }
+   
    if (ctx->VertexProgram._Current != prevVP && ctx->Driver.BindProgram) {
       ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
-                           (struct gl_program *) ctx->VertexProgram._Current);
+                              (struct gl_program *) ctx->VertexProgram._Current);
    }
 }
 
@@ -315,6 +327,24 @@ update_color(GLcontext *ctx)
 }
 
 
+/*
+ * Check polygon state and set DD_TRI_CULL_FRONT_BACK and/or DD_TRI_OFFSET
+ * in ctx->_TriangleCaps if needed.
+ */
+static void
+update_polygon(GLcontext *ctx)
+{
+   ctx->_TriangleCaps &= ~(DD_TRI_CULL_FRONT_BACK | DD_TRI_OFFSET);
+
+   if (ctx->Polygon.CullFlag && ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK)
+      ctx->_TriangleCaps |= DD_TRI_CULL_FRONT_BACK;
+
+   if (   ctx->Polygon.OffsetPoint
+       || ctx->Polygon.OffsetLine
+       || ctx->Polygon.OffsetFill)
+      ctx->_TriangleCaps |= DD_TRI_OFFSET;
+}
+
 
 /**
  * Update the ctx->_TriangleCaps bitfield.
@@ -322,6 +352,7 @@ update_color(GLcontext *ctx)
  * This function must be called after other update_*() functions since
  * there are dependencies on some other derived values.
  */
+#if 0
 static void
 update_tricaps(GLcontext *ctx, GLbitfield new_state)
 {
@@ -387,6 +418,7 @@ update_tricaps(GLcontext *ctx, GLbitfield new_state)
    if (ctx->Stencil._TestTwoSide)
       ctx->_TriangleCaps |= DD_TRI_TWOSTENCIL;
 }
+#endif
 
 
 /**
@@ -425,6 +457,9 @@ _mesa_update_state_locked( GLcontext *ctx )
    if (new_state & (_NEW_SCISSOR | _NEW_BUFFERS | _NEW_VIEWPORT))
       _mesa_update_draw_buffer_bounds( ctx );
 
+   if (new_state & _NEW_POLYGON)
+      update_polygon( ctx );
+
    if (new_state & _NEW_LIGHT)
       _mesa_update_lighting( ctx );
 
@@ -436,6 +471,9 @@ _mesa_update_state_locked( GLcontext *ctx )
       _mesa_update_pixel( ctx, new_state );
 #endif
 
+   if (new_state & _DD_NEW_SEPARATE_SPECULAR)
+      update_separate_specular( ctx );
+
    if (new_state & (_NEW_ARRAY | _NEW_PROGRAM))
       update_arrays( ctx );
 
@@ -448,9 +486,11 @@ _mesa_update_state_locked( GLcontext *ctx )
    if (new_state & _NEW_COLOR)
       update_color( ctx );
 
+#if 0
    if (new_state & (_NEW_POINT | _NEW_LINE | _NEW_POLYGON | _NEW_LIGHT
                     | _NEW_STENCIL | _DD_NEW_SEPARATE_SPECULAR))
       update_tricaps( ctx, new_state );
+#endif
 
    /* ctx->_NeedEyeCoords is now up to date.
     *
diff --git a/src/mesa/main/texcompress.c b/src/mesa/main/texcompress.c
index 5ad936419b6..04f0f3ab138 100644
--- a/src/mesa/main/texcompress.c
+++ b/src/mesa/main/texcompress.c
@@ -3,6 +3,7 @@
  * Version:  6.5.1
  *
  * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -84,6 +85,25 @@ _mesa_get_compressed_formats(GLcontext *ctx, GLint *formats, GLboolean all)
             if (all)
                n += 1;
          }
+#if FEATURE_EXT_texture_sRGB
+         if (ctx->Extensions.EXT_texture_sRGB) {
+            if (formats) {
+               if (all) {
+                  /* according to sRGB spec, these should not be returned
+                     via the GL_COMPRESSED_TEXTURE_FORMATS query as they
+                     aren't really general purpose */
+                  formats[n++] = GL_COMPRESSED_SRGB_S3TC_DXT1_EXT;
+                  formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT;
+                  formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT;
+                  formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT;
+               }
+            }
+            else {
+               if (all)
+                  n += 4;
+            }
+         }
+#endif /* FEATURE_EXT_texture_sRGB */
       }
       if (ctx->Extensions.S3_s3tc) {
          if (formats) {
@@ -96,19 +116,6 @@ _mesa_get_compressed_formats(GLcontext *ctx, GLint *formats, GLboolean all)
             n += 4;
          }
       }
-#if FEATURE_EXT_texture_sRGB
-      if (ctx->Extensions.EXT_texture_sRGB) {
-         if (formats) {
-            formats[n++] = GL_COMPRESSED_SRGB_S3TC_DXT1_EXT;
-            formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT;
-            formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT;
-            formats[n++] = GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT;
-         }
-         else {
-            n += 4;
-         }
-      }
-#endif /* FEATURE_EXT_texture_sRGB */
    }
    return n;
 }
@@ -151,13 +158,15 @@ _mesa_compressed_texture_size( GLcontext *ctx,
       /* Textures smaller than 8x4 will effectively be made into 8x4 and
        * take 16 bytes.
        */
-      if (size < 16)
-         size = 16;
       return size;
 #endif
 #if FEATURE_texture_s3tc
    case MESA_FORMAT_RGB_DXT1:
    case MESA_FORMAT_RGBA_DXT1:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGB_DXT1:
+   case MESA_FORMAT_SRGBA_DXT1:
+#endif
       /* round up width, height to next multiple of 4 */
       width = (width + 3) & ~3;
       height = (height + 3) & ~3;
@@ -166,11 +175,13 @@ _mesa_compressed_texture_size( GLcontext *ctx,
       /* Textures smaller than 4x4 will effectively be made into 4x4 and
        * take 8 bytes.
        */
-      if (size < 8)
-         size = 8;
       return size;
    case MESA_FORMAT_RGBA_DXT3:
    case MESA_FORMAT_RGBA_DXT5:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGBA_DXT3:
+   case MESA_FORMAT_SRGBA_DXT5:
+#endif
       /* round up width, height to next multiple of 4 */
       width = (width + 3) & ~3;
       height = (height + 3) & ~3;
@@ -179,8 +190,6 @@ _mesa_compressed_texture_size( GLcontext *ctx,
       /* Textures smaller than 4x4 will effectively be made into 4x4 and
        * take 16 bytes.
        */
-      if (size < 16)
-         size = 16;
       return size;
 #endif
    default:
@@ -232,6 +241,20 @@ _mesa_compressed_texture_size_glenum(GLcontext *ctx,
    case GL_RGBA4_S3TC:
       mesaFormat = MESA_FORMAT_RGBA_DXT5;
       break;
+#if FEATURE_EXT_texture_sRGB
+   case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+      mesaFormat = MESA_FORMAT_SRGB_DXT1;
+      break;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+      mesaFormat = MESA_FORMAT_SRGBA_DXT1;
+      break;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+      mesaFormat = MESA_FORMAT_SRGBA_DXT3;
+      break;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
+      mesaFormat = MESA_FORMAT_SRGBA_DXT5;
+      break;
+#endif
 #endif
    default:
       return 0;
@@ -263,10 +286,18 @@ _mesa_compressed_row_stride(GLuint mesaFormat, GLsizei width)
 #if FEATURE_texture_s3tc
    case MESA_FORMAT_RGB_DXT1:
    case MESA_FORMAT_RGBA_DXT1:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGB_DXT1:
+   case MESA_FORMAT_SRGBA_DXT1:
+#endif
       stride = ((width + 3) / 4) * 8; /* 8 bytes per 4x4 tile */
       break;
    case MESA_FORMAT_RGBA_DXT3:
    case MESA_FORMAT_RGBA_DXT5:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGBA_DXT3:
+   case MESA_FORMAT_SRGBA_DXT5:
+#endif
       stride = ((width + 3) / 4) * 16; /* 16 bytes per 4x4 tile */
       break;
 #endif
@@ -315,10 +346,18 @@ _mesa_compressed_image_address(GLint col, GLint row, GLint img,
 #if FEATURE_texture_s3tc
    case MESA_FORMAT_RGB_DXT1:
    case MESA_FORMAT_RGBA_DXT1:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGB_DXT1:
+   case MESA_FORMAT_SRGBA_DXT1:
+#endif
       addr = (GLubyte *) image + 8 * (((width + 3) / 4) * (row / 4) + col / 4);
       break;
    case MESA_FORMAT_RGBA_DXT3:
    case MESA_FORMAT_RGBA_DXT5:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGBA_DXT3:
+   case MESA_FORMAT_SRGBA_DXT5:
+#endif
       addr = (GLubyte *) image + 16 * (((width + 3) / 4) * (row / 4) + col / 4);
       break;
 #endif
diff --git a/src/mesa/main/texcompress_fxt1.c b/src/mesa/main/texcompress_fxt1.c
index 16a3ba076fd..fc151605c9e 100644
--- a/src/mesa/main/texcompress_fxt1.c
+++ b/src/mesa/main/texcompress_fxt1.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/src/mesa/main/texcompress_s3tc.c b/src/mesa/main/texcompress_s3tc.c
index c48063d9192..ccc007c24d1 100644
--- a/src/mesa/main/texcompress_s3tc.c
+++ b/src/mesa/main/texcompress_s3tc.c
@@ -3,6 +3,7 @@
  * Version:  6.5.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -56,6 +57,34 @@
 #define DXTN_LIBNAME "libtxc_dxtn.so"
 #endif
 
+#if FEATURE_EXT_texture_sRGB
+/**
+ * Convert an 8-bit sRGB value from non-linear space to a
+ * linear RGB value in [0, 1].
+ * Implemented with a 256-entry lookup table.
+ */
+static INLINE GLfloat
+nonlinear_to_linear(GLubyte cs8)
+{
+   static GLfloat table[256];
+   static GLboolean tableReady = GL_FALSE;
+   if (!tableReady) {
+      /* compute lookup table now */
+      GLuint i;
+      for (i = 0; i < 256; i++) {
+         const GLfloat cs = UBYTE_TO_FLOAT(i);
+         if (cs <= 0.04045) {
+            table[i] = cs / 12.92f;
+         }
+         else {
+            table[i] = (GLfloat) _mesa_pow((cs + 0.055) / 1.055, 2.4);
+         }
+      }
+      tableReady = GL_TRUE;
+   }
+   return table[cs8];
+}
+#endif /* FEATURE_EXT_texture_sRGB */
 
 typedef void (*dxtFetchTexelFuncExt)( GLint srcRowstride, GLubyte *pixdata, GLint col, GLint row, GLvoid *texelOut );
 
@@ -552,6 +581,59 @@ fetch_texel_2d_f_rgba_dxt5( const struct gl_texture_image *texImage,
    texel[ACOMP] = CHAN_TO_FLOAT(rgba[ACOMP]);
 }
 
+#if FEATURE_EXT_texture_sRGB
+static void
+fetch_texel_2d_f_srgb_dxt1( const struct gl_texture_image *texImage,
+                            GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   /* just sample as GLchan and convert to float here */
+   GLchan rgba[4];
+   fetch_texel_2d_rgb_dxt1(texImage, i, j, k, rgba);
+   texel[RCOMP] = nonlinear_to_linear(rgba[RCOMP]);
+   texel[GCOMP] = nonlinear_to_linear(rgba[GCOMP]);
+   texel[BCOMP] = nonlinear_to_linear(rgba[BCOMP]);
+   texel[ACOMP] = CHAN_TO_FLOAT(rgba[ACOMP]);
+}
+
+static void
+fetch_texel_2d_f_srgba_dxt1( const struct gl_texture_image *texImage,
+                             GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   /* just sample as GLchan and convert to float here */
+   GLchan rgba[4];
+   fetch_texel_2d_rgba_dxt1(texImage, i, j, k, rgba);
+   texel[RCOMP] = nonlinear_to_linear(rgba[RCOMP]);
+   texel[GCOMP] = nonlinear_to_linear(rgba[GCOMP]);
+   texel[BCOMP] = nonlinear_to_linear(rgba[BCOMP]);
+   texel[ACOMP] = CHAN_TO_FLOAT(rgba[ACOMP]);
+}
+
+static void
+fetch_texel_2d_f_srgba_dxt3( const struct gl_texture_image *texImage,
+                             GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   /* just sample as GLchan and convert to float here */
+   GLchan rgba[4];
+   fetch_texel_2d_rgba_dxt3(texImage, i, j, k, rgba);
+   texel[RCOMP] = nonlinear_to_linear(rgba[RCOMP]);
+   texel[GCOMP] = nonlinear_to_linear(rgba[GCOMP]);
+   texel[BCOMP] = nonlinear_to_linear(rgba[BCOMP]);
+   texel[ACOMP] = CHAN_TO_FLOAT(rgba[ACOMP]);
+}
+
+static void
+fetch_texel_2d_f_srgba_dxt5( const struct gl_texture_image *texImage,
+                             GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   /* just sample as GLchan and convert to float here */
+   GLchan rgba[4];
+   fetch_texel_2d_rgba_dxt5(texImage, i, j, k, rgba);
+   texel[RCOMP] = nonlinear_to_linear(rgba[RCOMP]);
+   texel[GCOMP] = nonlinear_to_linear(rgba[GCOMP]);
+   texel[BCOMP] = nonlinear_to_linear(rgba[BCOMP]);
+   texel[ACOMP] = CHAN_TO_FLOAT(rgba[ACOMP]);
+}
+#endif
 
 const struct gl_texture_format _mesa_texformat_rgb_dxt1 = {
    MESA_FORMAT_RGB_DXT1,		/* MesaFormat */
@@ -648,3 +730,101 @@ const struct gl_texture_format _mesa_texformat_rgba_dxt5 = {
    NULL, /*impossible*/ 		/* FetchTexel3Df */
    NULL					/* StoreTexel */
 };
+
+#if FEATURE_EXT_texture_sRGB
+const struct gl_texture_format _mesa_texformat_srgb_dxt1 = {
+   MESA_FORMAT_SRGB_DXT1,		/* MesaFormat */
+   GL_RGB,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   4, /*approx*/			/* RedBits */
+   4, /*approx*/			/* GreenBits */
+   4, /*approx*/			/* BlueBits */
+   0,					/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   0,					/* TexelBytes */
+   texstore_rgb_dxt1,			/* StoreTexImageFunc */
+   NULL, /*impossible*/ 		/* FetchTexel1D */
+   NULL, 				/* FetchTexel2D */
+   NULL, /*impossible*/ 		/* FetchTexel3D */
+   NULL, /*impossible*/ 		/* FetchTexel1Df */
+   fetch_texel_2d_f_srgb_dxt1, 		/* FetchTexel2Df */
+   NULL, /*impossible*/ 		/* FetchTexel3Df */
+   NULL					/* StoreTexel */
+};
+
+const struct gl_texture_format _mesa_texformat_srgba_dxt1 = {
+   MESA_FORMAT_SRGBA_DXT1,		/* MesaFormat */
+   GL_RGBA,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   4, /*approx*/			/* RedBits */
+   4, /*approx*/			/* GreenBits */
+   4, /*approx*/			/* BlueBits */
+   1, /*approx*/			/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   0,					/* TexelBytes */
+   texstore_rgba_dxt1,			/* StoreTexImageFunc */
+   NULL, /*impossible*/ 		/* FetchTexel1D */
+   NULL, 				/* FetchTexel2D */
+   NULL, /*impossible*/ 		/* FetchTexel3D */
+   NULL, /*impossible*/ 		/* FetchTexel1Df */
+   fetch_texel_2d_f_srgba_dxt1, 	/* FetchTexel2Df */
+   NULL, /*impossible*/ 		/* FetchTexel3Df */
+   NULL					/* StoreTexel */
+};
+
+const struct gl_texture_format _mesa_texformat_srgba_dxt3 = {
+   MESA_FORMAT_SRGBA_DXT3,		/* MesaFormat */
+   GL_RGBA,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   4, /*approx*/			/* RedBits */
+   4, /*approx*/			/* GreenBits */
+   4, /*approx*/			/* BlueBits */
+   4, /*approx*/			/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   0,					/* TexelBytes */
+   texstore_rgba_dxt3,			/* StoreTexImageFunc */
+   NULL, /*impossible*/ 		/* FetchTexel1D */
+   NULL, 				/* FetchTexel2D */
+   NULL, /*impossible*/ 		/* FetchTexel3D */
+   NULL, /*impossible*/ 		/* FetchTexel1Df */
+   fetch_texel_2d_f_srgba_dxt3, 	/* FetchTexel2Df */
+   NULL, /*impossible*/ 		/* FetchTexel3Df */
+   NULL					/* StoreTexel */
+};
+
+const struct gl_texture_format _mesa_texformat_srgba_dxt5 = {
+   MESA_FORMAT_SRGBA_DXT5,		/* MesaFormat */
+   GL_RGBA,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   4,/*approx*/				/* RedBits */
+   4,/*approx*/				/* GreenBits */
+   4,/*approx*/				/* BlueBits */
+   4,/*approx*/				/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   0,					/* TexelBytes */
+   texstore_rgba_dxt5,			/* StoreTexImageFunc */
+   NULL, /*impossible*/ 		/* FetchTexel1D */
+   NULL, 				/* FetchTexel2D */
+   NULL, /*impossible*/ 		/* FetchTexel3D */
+   NULL, /*impossible*/ 		/* FetchTexel1Df */
+   fetch_texel_2d_f_srgba_dxt5, 	/* FetchTexel2Df */
+   NULL, /*impossible*/ 		/* FetchTexel3Df */
+   NULL					/* StoreTexel */
+};
+#endif
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index c279956f2a2..3cb103f51f8 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -28,6 +28,7 @@
 #include "glheader.h"
 #include "macros.h"
 #include "enums.h"
+#include "shader/program.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_cache.h"
 #include "shader/prog_instruction.h"
@@ -36,6 +37,16 @@
 #include "shader/programopt.h"
 #include "texenvprogram.h"
 
+
+struct texenvprog_cache_item
+{
+   GLuint hash;
+   void *key;
+   struct gl_fragment_program *data;
+   struct texenvprog_cache_item *next;
+};
+
+
 /**
  * Up to nine instructions per tex unit, plus fog, specular color.
  */
@@ -59,6 +70,7 @@ struct state_key {
    struct {
       GLuint enabled:1;
       GLuint source_index:3;   /* one of TEXTURE_1D/2D/3D/CUBE/RECT_INDEX */
+      GLuint shadow:1;
       GLuint ScaleShiftRGB:2;
       GLuint ScaleShiftA:2;
 
@@ -285,10 +297,13 @@ static void make_state_key( GLcontext *ctx,  struct state_key *key )
 
    for (i=0;i<MAX_TEXTURE_UNITS;i++) {
       const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
-		
-      if (!texUnit->_ReallyEnabled) 
+      GLenum format;
+
+      if (!texUnit->_ReallyEnabled || !texUnit->Enabled)
          continue;
 
+      format = texUnit->_Current->Image[0][texUnit->_Current->BaseLevel]->_BaseFormat;
+
       key->unit[i].enabled = 1;
       key->enabled_units |= (1<<i);
       key->nr_enabled_units = i+1;
@@ -296,6 +311,9 @@ static void make_state_key( GLcontext *ctx,  struct state_key *key )
 
       key->unit[i].source_index = 
 	 translate_tex_src_bit(texUnit->_ReallyEnabled);		
+      key->unit[i].shadow = ((texUnit->_Current->CompareMode == GL_COMPARE_R_TO_TEXTURE) && 
+                             ((format == GL_DEPTH_COMPONENT) || 
+                              (format == GL_DEPTH_STENCIL_EXT)));
 
       key->unit[i].NumArgsRGB = texUnit->_CurrentCombine->_NumArgsRGB;
       key->unit[i].NumArgsA = texUnit->_CurrentCombine->_NumArgsA;
@@ -597,7 +615,7 @@ emit_op(struct texenv_fragment_program *p,
 {
    GLuint nr = p->program->Base.NumInstructions++;
    struct prog_instruction *inst = &p->program->Base.Instructions[nr];
-      
+
    assert(nr < MAX_INSTRUCTIONS);
 
    _mesa_init_instructions(inst, 1);
@@ -1066,6 +1084,10 @@ static void load_texture( struct texenv_fragment_program *p, GLuint unit )
 	 p->src_texture[unit] = emit_texld( p, OPCODE_TXP,
 					    tmp, WRITEMASK_XYZW, 
 					    unit, dim, texcoord );
+
+	 if (p->state->unit[unit].shadow)
+	    p->program->Base.ShadowSamplers |= 1 << unit;
+
          p->program->Base.SamplersUsed |= (1 << unit);
          /* This identity mapping should already be in place
           * (see _mesa_init_program_struct()) but let's be safe.
@@ -1300,6 +1322,9 @@ _mesa_get_fixed_func_fragment_program(GLcontext *ctx)
  * If _MaintainTexEnvProgram is set we'll generate a fragment program that
  * implements the current texture env/combine mode.
  * This function generates that program and puts it into effect.
+ *
+ * XXX: remove this function.  currently only called by some drivers,
+ * not by mesa core.  We now handle this properly from inside mesa.
  */
 void
 _mesa_UpdateTexEnvProgram( GLcontext *ctx )
@@ -1311,11 +1336,14 @@ _mesa_UpdateTexEnvProgram( GLcontext *ctx )
    /* If a conventional fragment program/shader isn't in effect... */
    if (!ctx->FragmentProgram._Enabled &&
        (!ctx->Shader.CurrentProgram ||
-        !ctx->Shader.CurrentProgram->FragmentProgram) ) {
+        !ctx->Shader.CurrentProgram->FragmentProgram) ) 
+   {
+      struct gl_fragment_program *newProg;
+
+      newProg = _mesa_get_fixed_func_fragment_program(ctx);
 
-      ctx->FragmentProgram._Current
-         = ctx->FragmentProgram._TexEnvProgram
-         = _mesa_get_fixed_func_fragment_program(ctx);
+      _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, newProg);
+      _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, newProg);
    } 
 
    /* Tell the driver about the change.  Could define a new target for
diff --git a/src/mesa/main/texformat.c b/src/mesa/main/texformat.c
index 60f36c4a87f..db3525666a2 100644
--- a/src/mesa/main/texformat.c
+++ b/src/mesa/main/texformat.c
@@ -3,6 +3,7 @@
  * Version:  6.5.1
  *
  * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -333,6 +334,30 @@ const struct gl_texture_format _mesa_texformat_srgba8 = {
    store_texel_srgba8			/* StoreTexel */
 };
 
+const struct gl_texture_format _mesa_texformat_sargb8 = {
+   MESA_FORMAT_SARGB8,			/* MesaFormat */
+   GL_RGBA,				/* BaseFormat */
+   GL_UNSIGNED_NORMALIZED_ARB,		/* DataType */
+   8,					/* RedBits */
+   8,					/* GreenBits */
+   8,					/* BlueBits */
+   8,					/* AlphaBits */
+   0,					/* LuminanceBits */
+   0,					/* IntensityBits */
+   0,					/* IndexBits */
+   0,					/* DepthBits */
+   0,					/* StencilBits */
+   4,					/* TexelBytes */
+   _mesa_texstore_sargb8,		/* StoreTexImageFunc */
+   NULL,				/* FetchTexel1D */
+   NULL,				/* FetchTexel2D */
+   NULL,				/* FetchTexel3D */
+   fetch_texel_1d_sargb8,		/* FetchTexel1Df */
+   fetch_texel_2d_sargb8,		/* FetchTexel2Df */
+   fetch_texel_3d_sargb8,		/* FetchTexel3Df */
+   store_texel_sargb8			/* StoreTexel */
+};
+
 const struct gl_texture_format _mesa_texformat_sl8 = {
    MESA_FORMAT_SL8,			/* MesaFormat */
    GL_LUMINANCE,			/* BaseFormat */
@@ -1420,14 +1445,13 @@ _mesa_choose_tex_format( GLcontext *ctx, GLint internalFormat,
          ; /* fallthrough */
    }
 
-   if (ctx->Extensions.SGIX_depth_texture ||
-       ctx->Extensions.ARB_depth_texture) {
+   if (ctx->Extensions.ARB_depth_texture) {
       switch (internalFormat) {
          case GL_DEPTH_COMPONENT:
-         case GL_DEPTH_COMPONENT24_SGIX:
-         case GL_DEPTH_COMPONENT32_SGIX:
+         case GL_DEPTH_COMPONENT24:
+         case GL_DEPTH_COMPONENT32:
             return &_mesa_texformat_z32;
-         case GL_DEPTH_COMPONENT16_SGIX:
+         case GL_DEPTH_COMPONENT16:
             return &_mesa_texformat_z16;
          default:
             ; /* fallthrough */
@@ -1579,21 +1603,40 @@ _mesa_choose_tex_format( GLcontext *ctx, GLint internalFormat,
          case GL_SLUMINANCE_ALPHA_EXT:
          case GL_SLUMINANCE8_ALPHA8_EXT:
             return &_mesa_texformat_sla8;
-         /* NOTE: not supporting any compression of sRGB at this time */
-         case GL_COMPRESSED_SRGB_EXT:
-            return &_mesa_texformat_srgb8;
-         case GL_COMPRESSED_SRGB_ALPHA_EXT:
-            return &_mesa_texformat_srgba8;
          case GL_COMPRESSED_SLUMINANCE_EXT:
             return &_mesa_texformat_sl8;
          case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
             return &_mesa_texformat_sla8;
-         case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+         case GL_COMPRESSED_SRGB_EXT:
+#if FEATURE_texture_s3tc
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgb_dxt1;
+#endif
             return &_mesa_texformat_srgb8;
+         case GL_COMPRESSED_SRGB_ALPHA_EXT:
+#if FEATURE_texture_s3tc
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgba_dxt3; /* Not srgba_dxt1, see spec */
+#endif
+            return &_mesa_texformat_srgba8;
+#if FEATURE_texture_s3tc
+         case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgb_dxt1;
+            break;
          case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgba_dxt1;
+            break;
          case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgba_dxt3;
+            break;
          case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
-            return &_mesa_texformat_srgba8;
+            if (ctx->Extensions.EXT_texture_compression_s3tc)
+               return &_mesa_texformat_srgba_dxt5;
+            break;
+#endif
          default:
             ; /* fallthrough */
       }
@@ -1695,6 +1738,7 @@ _mesa_format_to_type_and_comps(const struct gl_texture_format *format,
       *comps = 3;
       return;
    case MESA_FORMAT_SRGBA8:
+   case MESA_FORMAT_SARGB8:
       *datatype = GL_UNSIGNED_BYTE;
       *comps = 4;
       return;
@@ -1717,6 +1761,12 @@ _mesa_format_to_type_and_comps(const struct gl_texture_format *format,
    case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGBA_DXT3:
    case MESA_FORMAT_RGBA_DXT5:
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGB_DXT1:
+   case MESA_FORMAT_SRGBA_DXT1:
+   case MESA_FORMAT_SRGBA_DXT3:
+   case MESA_FORMAT_SRGBA_DXT5:
+#endif
       /* XXX generate error instead? */
       *datatype = GL_UNSIGNED_BYTE;
       *comps = 0;
diff --git a/src/mesa/main/texformat.h b/src/mesa/main/texformat.h
index 3b6d05612dd..c7a754b0b75 100644
--- a/src/mesa/main/texformat.h
+++ b/src/mesa/main/texformat.h
@@ -3,6 +3,7 @@
  * Version:  6.5.1
  *
  * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -96,8 +97,15 @@ enum _format {
    /*@{*/
    MESA_FORMAT_SRGB8,
    MESA_FORMAT_SRGBA8,
+   MESA_FORMAT_SARGB8,
    MESA_FORMAT_SL8,
    MESA_FORMAT_SLA8,
+#if FEATURE_texture_s3tc
+   MESA_FORMAT_SRGB_DXT1,
+   MESA_FORMAT_SRGBA_DXT1,
+   MESA_FORMAT_SRGBA_DXT3,
+   MESA_FORMAT_SRGBA_DXT5,
+#endif
    /*@}*/
 #endif
 
@@ -171,8 +179,15 @@ extern const struct gl_texture_format _mesa_texformat_intensity;
 /*@{*/
 extern const struct gl_texture_format _mesa_texformat_srgb8;
 extern const struct gl_texture_format _mesa_texformat_srgba8;
+extern const struct gl_texture_format _mesa_texformat_sargb8;
 extern const struct gl_texture_format _mesa_texformat_sl8;
 extern const struct gl_texture_format _mesa_texformat_sla8;
+#if FEATURE_texture_s3tc
+extern const struct gl_texture_format _mesa_texformat_srgb_dxt1;
+extern const struct gl_texture_format _mesa_texformat_srgba_dxt1;
+extern const struct gl_texture_format _mesa_texformat_srgba_dxt3;
+extern const struct gl_texture_format _mesa_texformat_srgba_dxt5;
+#endif
 /*@}*/
 #endif
 
diff --git a/src/mesa/main/texformat_tmp.h b/src/mesa/main/texformat_tmp.h
index 63939f40111..b1031b0cfe3 100644
--- a/src/mesa/main/texformat_tmp.h
+++ b/src/mesa/main/texformat_tmp.h
@@ -3,6 +3,7 @@
  * Version:  6.5.1
  *
  * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -1117,6 +1118,7 @@ static void FETCH(ci8)( const struct gl_texture_image *texImage,
          break;;
       default:
          _mesa_problem(ctx, "Bad palette format in fetch_texel_ci8");
+         return;
       }
 #if CHAN_TYPE == GL_UNSIGNED_BYTE
       COPY_4UBV(texel, texelUB);
@@ -1190,6 +1192,31 @@ static void store_texel_srgba8(struct gl_texture_image *texImage,
    dst[0] = rgba[RCOMP];
    dst[1] = rgba[GCOMP];
    dst[2] = rgba[BCOMP];
+   dst[3] = rgba[ACOMP];
+}
+#endif
+
+/* Fetch texel from 1D, 2D or 3D sargb8 texture, return 4 GLfloats */
+static void FETCH(sargb8)(const struct gl_texture_image *texImage,
+                          GLint i, GLint j, GLint k, GLfloat *texel )
+{
+   const GLubyte *src = TEXEL_ADDR(GLubyte, texImage, i, j, k, 4);
+   texel[RCOMP] = nonlinear_to_linear(src[1]);
+   texel[GCOMP] = nonlinear_to_linear(src[2]);
+   texel[BCOMP] = nonlinear_to_linear(src[3]);
+   texel[ACOMP] = UBYTE_TO_FLOAT(src[0]); /* linear! */
+}
+
+#if DIM == 3
+static void store_texel_sargb8(struct gl_texture_image *texImage,
+                               GLint i, GLint j, GLint k, const void *texel)
+{
+   const GLubyte *rgba = (const GLubyte *) texel;
+   GLubyte *dst = TEXEL_ADDR(GLubyte, texImage, i, j, k, 4);
+   dst[0] = rgba[ACOMP];
+   dst[1] = rgba[RCOMP];
+   dst[2] = rgba[GCOMP];
+   dst[3] = rgba[BCOMP];
 }
 #endif
 
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index 36557f19fd4..a5e0db736ed 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -1,8 +1,8 @@
 /*
- * Mesa 3-D graphics library
+ * mesa 3-D graphics library
  * Version:  7.1
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -241,13 +241,12 @@ _mesa_base_tex_format( GLcontext *ctx, GLint internalFormat )
       }
    }
 
-   if (ctx->Extensions.SGIX_depth_texture ||
-       ctx->Extensions.ARB_depth_texture) {
+   if (ctx->Extensions.ARB_depth_texture) {
       switch (internalFormat) {
          case GL_DEPTH_COMPONENT:
-         case GL_DEPTH_COMPONENT16_SGIX:
-         case GL_DEPTH_COMPONENT24_SGIX:
-         case GL_DEPTH_COMPONENT32_SGIX:
+         case GL_DEPTH_COMPONENT16:
+         case GL_DEPTH_COMPONENT24:
+         case GL_DEPTH_COMPONENT32:
             return GL_DEPTH_COMPONENT;
          default:
             ; /* fallthrough */
@@ -389,9 +388,10 @@ _mesa_base_tex_format( GLcontext *ctx, GLint internalFormat )
  * index, depth, stencil, etc).
  * \param format  the image format value (may by an internal texture format)
  * \return GL_TRUE if its a color/RGBA format, GL_FALSE otherwise.
+ * XXX maybe move this func to image.c
  */
-static GLboolean
-is_color_format(GLenum format)
+GLboolean
+_mesa_is_color_format(GLenum format)
 {
    switch (format) {
       case GL_RED:
@@ -492,6 +492,7 @@ is_color_format(GLenum format)
 #endif /* FEATURE_EXT_texture_sRGB */
          return GL_TRUE;
       case GL_YCBCR_MESA:  /* not considered to be RGB */
+         /* fall-through */
       default:
          return GL_FALSE;
    }
@@ -526,9 +527,9 @@ static GLboolean
 is_depth_format(GLenum format)
 {
    switch (format) {
-      case GL_DEPTH_COMPONENT16_ARB:
-      case GL_DEPTH_COMPONENT24_ARB:
-      case GL_DEPTH_COMPONENT32_ARB:
+      case GL_DEPTH_COMPONENT16:
+      case GL_DEPTH_COMPONENT24:
+      case GL_DEPTH_COMPONENT32:
       case GL_DEPTH_COMPONENT:
          return GL_TRUE;
       default:
@@ -1243,9 +1244,9 @@ _mesa_init_teximage_fields(GLcontext *ctx, GLenum target,
    img->IsCompressed = GL_FALSE;
    img->CompressedSize = 0;
 
-   if ((width == 1 || _mesa_bitcount(img->Width2) == 1) &&
-       (height == 1 || _mesa_bitcount(img->Height2) == 1) &&
-       (depth == 1 || _mesa_bitcount(img->Depth2) == 1))
+   if ((width == 1 || _mesa_is_pow_two(img->Width2)) &&
+       (height == 1 || _mesa_is_pow_two(img->Height2)) &&
+       (depth == 1 || _mesa_is_pow_two(img->Depth2)))
       img->_IsPowerOfTwo = GL_TRUE;
    else
       img->_IsPowerOfTwo = GL_FALSE;
@@ -1316,7 +1317,7 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width >0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width >0 && !_mesa_is_pow_two(width - 2 * border)) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or level */
          return GL_FALSE;
@@ -1326,10 +1327,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && !_mesa_is_pow_two(width - 2 * border)) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && !_mesa_is_pow_two(height - 2 * border)) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or height or level */
          return GL_FALSE;
@@ -1339,13 +1340,13 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.Max3DTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && !_mesa_is_pow_two(width - 2 * border)) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && !_mesa_is_pow_two(height - 2 * border)) ||
           depth < 2 * border || depth > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           depth > 0 && _mesa_bitcount(depth - 2 * border) != 1) ||
+           depth > 0 && !_mesa_is_pow_two(depth - 2 * border)) ||
           level >= ctx->Const.Max3DTextureLevels) {
          /* bad width or height or depth or level */
          return GL_FALSE;
@@ -1363,10 +1364,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxCubeTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && !_mesa_is_pow_two(width - 2 * border)) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && !_mesa_is_pow_two(height - 2 * border)) ||
           level >= ctx->Const.MaxCubeTextureLevels) {
          /* bad width or height */
          return GL_FALSE;
@@ -1376,7 +1377,7 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && !_mesa_is_pow_two(width - 2 * border)) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or level */
          return GL_FALSE;
@@ -1390,10 +1391,10 @@ _mesa_test_proxy_teximage(GLcontext *ctx, GLenum target, GLint level,
       maxSize = 1 << (ctx->Const.MaxTextureLevels - 1);
       if (width < 2 * border || width > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           width > 0 && _mesa_bitcount(width - 2 * border) != 1) ||
+           width > 0 && !_mesa_is_pow_two(width - 2 * border)) ||
           height < 2 * border || height > 2 + maxSize ||
           (!ctx->Extensions.ARB_texture_non_power_of_two &&
-           height > 0 && _mesa_bitcount(height - 2 * border) != 1) ||
+           height > 0 && !_mesa_is_pow_two(height - 2 * border)) ||
           level >= ctx->Const.MaxTextureLevels) {
          /* bad width or height or level */
          return GL_FALSE;
@@ -1588,9 +1589,9 @@ texture_error_check( GLcontext *ctx, GLenum target,
    }
 
    /* make sure internal format and format basically agree */
-   colorFormat = is_color_format(format);
+   colorFormat = _mesa_is_color_format(format);
    indexFormat = is_index_format(format);
-   if ((is_color_format(internalFormat) && !colorFormat && !indexFormat) ||
+   if ((_mesa_is_color_format(internalFormat) && !colorFormat && !indexFormat) ||
        (is_index_format(internalFormat) && !indexFormat) ||
        (is_depth_format(internalFormat) != is_depth_format(format)) ||
        (is_ycbcr_format(internalFormat) != is_ycbcr_format(format)) ||
@@ -2017,7 +2018,7 @@ copytexture_error_check( GLcontext *ctx, GLuint dimensions,
    }
 
    if (is_compressed_format(ctx, internalFormat)) {
-      if (target != GL_TEXTURE_2D) {
+      if (!target_can_be_compressed(ctx, target)) {
          _mesa_error(ctx, GL_INVALID_ENUM,
                      "glCopyTexImage%d(target)", dimensions);
          return GL_TRUE;
@@ -2052,30 +2053,20 @@ copytexture_error_check( GLcontext *ctx, GLuint dimensions,
 
 /**
  * Test glCopyTexSubImage[12]D() parameters for errors.
+ * Note that this is the first part of error checking.
+ * See also copytexsubimage_error_check2() below for the second part.
  * 
  * \param ctx GL context.
  * \param dimensions texture image dimensions (must be 1, 2 or 3).
  * \param target texture target given by the user.
  * \param level image level given by the user.
- * \param xoffset sub-image x offset given by the user.
- * \param yoffset sub-image y offset given by the user.
- * \param zoffset sub-image z offset given by the user.
- * \param width image width given by the user.
- * \param height image height given by the user.
  * 
  * \return GL_TRUE if an error was detected, or GL_FALSE if no errors.
- * 
- * Verifies each of the parameters against the constants specified in
- * __GLcontextRec::Const and the supported extensions, and according to the
- * OpenGL specification.
  */
 static GLboolean
-copytexsubimage_error_check( GLcontext *ctx, GLuint dimensions,
-                             GLenum target, GLint level,
-                             GLint xoffset, GLint yoffset, GLint zoffset,
-                             GLsizei width, GLsizei height)
+copytexsubimage_error_check1( GLcontext *ctx, GLuint dimensions,
+                              GLenum target, GLint level)
 {
-   /* Check target */
    /* Check that the source buffer is complete */
    if (ctx->ReadBuffer->Name) {
       _mesa_test_framebuffer_completeness(ctx, ctx->ReadBuffer);
@@ -2086,6 +2077,7 @@ copytexsubimage_error_check( GLcontext *ctx, GLuint dimensions,
       }
    }
 
+   /* Check target */
    if (dimensions == 1) {
       if (target != GL_TEXTURE_1D) {
          _mesa_error( ctx, GL_INVALID_ENUM, "glCopyTexSubImage1D(target)" );
@@ -2133,21 +2125,18 @@ copytexsubimage_error_check( GLcontext *ctx, GLuint dimensions,
       return GL_TRUE;
    }
 
-   /* Check size */
-   if (width < 0) {
-      _mesa_error(ctx, GL_INVALID_VALUE,
-                  "glCopyTexSubImage%dD(width=%d)", dimensions, width);
-      return GL_TRUE;
-   }
-   if (dimensions > 1 && height < 0) {
-      _mesa_error(ctx, GL_INVALID_VALUE,
-                  "glCopyTexSubImage%dD(height=%d)", dimensions, height);
-      return GL_TRUE;
-   }
-
    return GL_FALSE;
 }
 
+
+/**
+ * Second part of error checking for glCopyTexSubImage[12]D().
+ * \param xoffset sub-image x offset given by the user.
+ * \param yoffset sub-image y offset given by the user.
+ * \param zoffset sub-image z offset given by the user.
+ * \param width image width given by the user.
+ * \param height image height given by the user.
+ */
 static GLboolean
 copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
 			      GLenum target, GLint level,
@@ -2155,6 +2144,7 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
 			      GLsizei width, GLsizei height,
 			      const struct gl_texture_image *teximage )
 {
+   /* check that dest tex image exists */
    if (!teximage) {
       _mesa_error(ctx, GL_INVALID_OPERATION,
                   "glCopyTexSubImage%dD(undefined texture level: %d)",
@@ -2162,6 +2152,19 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
       return GL_TRUE;
    }
 
+   /* Check size */
+   if (width < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glCopyTexSubImage%dD(width=%d)", dimensions, width);
+      return GL_TRUE;
+   }
+   if (dimensions > 1 && height < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glCopyTexSubImage%dD(height=%d)", dimensions, height);
+      return GL_TRUE;
+   }
+
+   /* check x/y offsets */
    if (xoffset < -((GLint)teximage->Border)) {
       _mesa_error(ctx, GL_INVALID_VALUE,
                   "glCopyTexSubImage%dD(xoffset=%d)", dimensions, xoffset);
@@ -2186,6 +2189,7 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
       }
    }
 
+   /* check z offset */
    if (dimensions > 2) {
       if (zoffset < -((GLint)teximage->Border)) {
          _mesa_error(ctx, GL_INVALID_VALUE,
@@ -2200,7 +2204,7 @@ copytexsubimage_error_check2( GLcontext *ctx, GLuint dimensions,
    }
 
    if (teximage->IsCompressed) {
-      if (target != GL_TEXTURE_2D) {
+      if (!target_can_be_compressed(ctx, target)) {
          _mesa_error(ctx, GL_INVALID_ENUM,
                      "glCopyTexSubImage%d(target)", dimensions);
          return GL_TRUE;
@@ -2308,8 +2312,7 @@ _mesa_GetTexImage( GLenum target, GLint level, GLenum format,
       return;
    }
 
-   if (!ctx->Extensions.SGIX_depth_texture &&
-       !ctx->Extensions.ARB_depth_texture && is_depth_format(format)) {
+   if (!ctx->Extensions.ARB_depth_texture && is_depth_format(format)) {
       _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexImage(format)");
       return;
    }
@@ -2338,8 +2341,8 @@ _mesa_GetTexImage( GLenum target, GLint level, GLenum format,
        * texture's format.  Note that a color index texture can be converted
        * to RGBA so that combo is allowed.
        */
-      if (is_color_format(format)
-	  && !is_color_format(texImage->TexFormat->BaseFormat)
+      if (_mesa_is_color_format(format)
+	  && !_mesa_is_color_format(texImage->TexFormat->BaseFormat)
 	  && !is_index_format(texImage->TexFormat->BaseFormat)) {
 	 _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTexImage(format mismatch)");
 	 goto out;
@@ -2432,7 +2435,7 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat,
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
 #if FEATURE_convolve
-   if (is_color_format(internalFormat)) {
+   if (_mesa_is_color_format(internalFormat)) {
       _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
    }
 #endif
@@ -2529,7 +2532,7 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat,
    ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
 
 #if FEATURE_convolve
-   if (is_color_format(internalFormat)) {
+   if (_mesa_is_color_format(internalFormat)) {
       _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth,
 					 &postConvHeight);
    }
@@ -2711,8 +2714,8 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat,
       }
       else {
          /* no error, set the tex image parameters */
-         _mesa_init_teximage_fields(ctx, target, texImage, width, height, 1,
-                                    border, internalFormat);
+         _mesa_init_teximage_fields(ctx, target, texImage, width, height,
+                                    depth, border, internalFormat);
          texImage->TexFormat = (*ctx->Driver.ChooseTextureFormat)(ctx,
                                           internalFormat, format, type);
       }
@@ -2754,7 +2757,7 @@ _mesa_TexSubImage1D( GLenum target, GLint level,
 
 #if FEATURE_convolve
    /* XXX should test internal format */
-   if (is_color_format(format)) {
+   if (_mesa_is_color_format(format)) {
       _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
    }
 #endif
@@ -2814,7 +2817,7 @@ _mesa_TexSubImage2D( GLenum target, GLint level,
 
 #if FEATURE_convolve
    /* XXX should test internal format */
-   if (is_color_format(format)) {
+   if (_mesa_is_color_format(format)) {
       _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth,
                                          &postConvHeight);
    }
@@ -2929,7 +2932,7 @@ _mesa_CopyTexImage1D( GLenum target, GLint level,
       _mesa_update_state(ctx);
 
 #if FEATURE_convolve
-   if (is_color_format(internalFormat)) {
+   if (_mesa_is_color_format(internalFormat)) {
       _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
    }
 #endif
@@ -2994,11 +2997,12 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat,
       _mesa_update_state(ctx);
 
 #if FEATURE_convolve
-   if (is_color_format(internalFormat)) {
+   if (_mesa_is_color_format(internalFormat)) {
       _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth,
                                          &postConvHeight);
    }
 #endif
+
    if (copytexture_error_check(ctx, 2, target, level, internalFormat,
                                postConvWidth, postConvHeight, border))
       return;
@@ -3060,13 +3064,7 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level,
    if (ctx->NewState & _IMAGE_NEW_TRANSFER_STATE)
       _mesa_update_state(ctx);
 
-#if FEATURE_convolve
-   /* XXX should test internal format */
-   _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
-#endif
-
-   if (copytexsubimage_error_check(ctx, 1, target, level,
-                                   xoffset, 0, 0, postConvWidth, 1))
+   if (copytexsubimage_error_check1(ctx, 1, target, level))
       return;
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
@@ -3076,6 +3074,12 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level,
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
+#if FEATURE_convolve
+      if (texImage && _mesa_is_color_format(texImage->InternalFormat)) {
+         _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
+      }
+#endif
+
       if (copytexsubimage_error_check2(ctx, 1, target, level,
 				       xoffset, 0, 0, postConvWidth, 1,
 				       texImage))
@@ -3115,13 +3119,7 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level,
    if (ctx->NewState & _IMAGE_NEW_TRANSFER_STATE)
       _mesa_update_state(ctx);
 
-#if FEATURE_convolve
-   /* XXX should test internal format */
-   _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth, &postConvHeight);
-#endif
-
-   if (copytexsubimage_error_check(ctx, 2, target, level, xoffset, yoffset, 0,
-                                   postConvWidth, postConvHeight))
+   if (copytexsubimage_error_check1(ctx, 2, target, level))
       return;
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
@@ -3131,6 +3129,13 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level,
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
+#if FEATURE_convolve
+      if (texImage && _mesa_is_color_format(texImage->InternalFormat)) {
+         _mesa_adjust_image_for_convolution(ctx, 2,
+                                            &postConvWidth, &postConvHeight);
+      }
+#endif
+
       if (copytexsubimage_error_check2(ctx, 2, target, level, xoffset, yoffset, 0,
 				       postConvWidth, postConvHeight, texImage))
 	 goto out;
@@ -3169,13 +3174,7 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level,
    if (ctx->NewState & _IMAGE_NEW_TRANSFER_STATE)
       _mesa_update_state(ctx);
 
-#if FEATURE_convolve
-   /* XXX should test internal format */
-   _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth, &postConvHeight);
-#endif
-
-   if (copytexsubimage_error_check(ctx, 3, target, level, xoffset, yoffset,
-                                   zoffset, postConvWidth, postConvHeight))
+   if (copytexsubimage_error_check1(ctx, 3, target, level))
       return;
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
@@ -3185,6 +3184,13 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level,
    {
       texImage = _mesa_select_tex_image(ctx, texObj, target, level);
 
+#if FEATURE_convolve
+      if (texImage && _mesa_is_color_format(texImage->InternalFormat)) {
+         _mesa_adjust_image_for_convolution(ctx, 2,
+                                            &postConvWidth, &postConvHeight);
+      }
+#endif
+
       if (copytexsubimage_error_check2(ctx, 3, target, level, xoffset, yoffset,
 				       zoffset, postConvWidth, postConvHeight,
 				       texImage))
@@ -3278,16 +3284,16 @@ compressed_texture_error_check(GLcontext *ctx, GLint dimensions,
     * XXX We should probably use the proxy texture error check function here.
     */
    if (width < 1 || width > maxTextureSize ||
-       (!ctx->Extensions.ARB_texture_non_power_of_two && _mesa_bitcount(width) != 1))
+       (!ctx->Extensions.ARB_texture_non_power_of_two && !_mesa_is_pow_two(width)))
       return GL_INVALID_VALUE;
 
    if ((height < 1 || height > maxTextureSize ||
-       (!ctx->Extensions.ARB_texture_non_power_of_two && _mesa_bitcount(height) != 1))
+       (!ctx->Extensions.ARB_texture_non_power_of_two && !_mesa_is_pow_two(height)))
        && dimensions > 1)
       return GL_INVALID_VALUE;
 
    if ((depth < 1 || depth > maxTextureSize ||
-       (!ctx->Extensions.ARB_texture_non_power_of_two && _mesa_bitcount(depth) != 1))
+       (!ctx->Extensions.ARB_texture_non_power_of_two && !_mesa_is_pow_two(depth)))
        && dimensions > 2)
       return GL_INVALID_VALUE;
 
diff --git a/src/mesa/main/teximage.h b/src/mesa/main/teximage.h
index b718c0046d8..eb60a1fa8fa 100644
--- a/src/mesa/main/teximage.h
+++ b/src/mesa/main/teximage.h
@@ -111,6 +111,9 @@ extern GLuint
 _mesa_tex_target_to_face(GLenum target);
 
 
+extern GLboolean
+_mesa_is_color_format(GLenum format);
+
 
 /**
  * Lock a texture for updating.  See also _mesa_lock_context_textures().
diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c
index 2a54ff7ff90..d8e8b559f53 100644
--- a/src/mesa/main/texobj.c
+++ b/src/mesa/main/texobj.c
@@ -383,6 +383,18 @@ _mesa_test_texobj_completeness( const GLcontext *ctx,
 
    t->_Complete = GL_TRUE;  /* be optimistic */
 
+   /* Detect cases where the application set the base level to an invalid
+    * value.
+    */
+   if ((baseLevel < 0) || (baseLevel > MAX_TEXTURE_LEVELS)) {
+      char s[100];
+      _mesa_sprintf(s, "obj %p (%d) base level = %d is invalid",
+              (void *) t, t->Name, baseLevel);
+      incomplete(t, s);
+      t->_Complete = GL_FALSE;
+      return;
+   }
+
    /* Always need the base level image */
    if (!t->Image[0][baseLevel]) {
       char s[100];
diff --git a/src/mesa/main/texparam.c b/src/mesa/main/texparam.c
index 3f3b448dbc6..a9e752a6371 100644
--- a/src/mesa/main/texparam.c
+++ b/src/mesa/main/texparam.c
@@ -324,6 +324,7 @@ _mesa_TexParameterfv( GLenum target, GLenum pname, const GLfloat *params )
          break;
       case GL_GENERATE_MIPMAP_SGIS:
          if (ctx->Extensions.SGIS_generate_mipmap) {
+            FLUSH_VERTICES(ctx, _NEW_TEXTURE);
             texObj->GenerateMipmap = params[0] ? GL_TRUE : GL_FALSE;
          }
          else {
@@ -641,8 +642,7 @@ _mesa_GetTexLevelParameteriv( GLenum target, GLint level,
             *params = 0;
          break;
       case GL_TEXTURE_DEPTH_SIZE_ARB:
-         if (ctx->Extensions.SGIX_depth_texture ||
-             ctx->Extensions.ARB_depth_texture)
+         if (ctx->Extensions.ARB_depth_texture)
             *params = img->TexFormat->DepthBits;
          else
             _mesa_error(ctx, GL_INVALID_ENUM,
@@ -902,9 +902,9 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
 #ifdef FEATURE_OES_draw_texture
       case GL_TEXTURE_CROP_RECT_OES:
          params[0] = obj->CropRect[0];
-         params[0] = obj->CropRect[1];
-         params[0] = obj->CropRect[2];
-         params[0] = obj->CropRect[3];
+         params[1] = obj->CropRect[1];
+         params[2] = obj->CropRect[2];
+         params[3] = obj->CropRect[3];
          break;
 #endif
       default:
@@ -1052,9 +1052,9 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
 #ifdef FEATURE_OES_draw_texture
       case GL_TEXTURE_CROP_RECT_OES:
          params[0] = obj->CropRect[0];
-         params[0] = obj->CropRect[1];
-         params[0] = obj->CropRect[2];
-         params[0] = obj->CropRect[3];
+         params[1] = obj->CropRect[1];
+         params[2] = obj->CropRect[2];
+         params[3] = obj->CropRect[3];
          break;
 #endif
       default:
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index 20f9c4512c4..f0193770412 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -118,20 +118,20 @@ _mesa_copy_texture_state( const GLcontext *src, GLcontext *dst )
       /* copy texture object bindings, not contents of texture objects */
       _mesa_lock_context_textures(dst);
 
-       _mesa_reference_texobj(&dst->Texture.Unit[i].Current1D,
-                              src->Texture.Unit[i].Current1D);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].Current2D,
-                              src->Texture.Unit[i].Current2D);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].Current3D,
-                              src->Texture.Unit[i].Current3D);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentCubeMap,
-                              src->Texture.Unit[i].CurrentCubeMap);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentRect,
-                              src->Texture.Unit[i].CurrentRect);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].Current1DArray,
-                              src->Texture.Unit[i].Current1DArray);
-       _mesa_reference_texobj(&dst->Texture.Unit[i].Current2DArray,
-                              src->Texture.Unit[i].Current2DArray);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current1D,
+                             src->Texture.Unit[i].Current1D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current2D,
+                             src->Texture.Unit[i].Current2D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current3D,
+                             src->Texture.Unit[i].Current3D);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentCubeMap,
+                             src->Texture.Unit[i].CurrentCubeMap);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].CurrentRect,
+                             src->Texture.Unit[i].CurrentRect);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current1DArray,
+                             src->Texture.Unit[i].Current1DArray);
+      _mesa_reference_texobj(&dst->Texture.Unit[i].Current2DArray,
+                             src->Texture.Unit[i].Current2DArray);
 
       _mesa_unlock_context_textures(dst);
    }
@@ -608,6 +608,24 @@ update_texture_state( GLcontext *ctx )
          _mesa_problem(ctx, "invalid Alpha combine mode in update_texture_state");
 	 break;
       }
+   }
+
+   /* Determine which texture coordinate sets are actually needed */
+   if (fprog) {
+      const GLuint coordMask = (1 << MAX_TEXTURE_COORD_UNITS) - 1;
+      ctx->Texture._EnabledCoordUnits
+         = (fprog->Base.InputsRead >> FRAG_ATTRIB_TEX0) & coordMask;
+   }
+   else {
+      ctx->Texture._EnabledCoordUnits = ctx->Texture._EnabledUnits;
+   }
+
+   /* Setup texgen for those texture coordinate sets that are in use */
+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+
+      if (!(ctx->Texture._EnabledCoordUnits & (1 << unit)))
+	 continue;
 
       if (texUnit->TexGenEnabled) {
 	 if (texUnit->TexGenEnabled & S_BIT) {
@@ -630,16 +648,6 @@ update_texture_state( GLcontext *ctx )
       if (ctx->TextureMatrixStack[unit].Top->type != MATRIX_IDENTITY)
 	 ctx->Texture._TexMatEnabled |= ENABLE_TEXMAT(unit);
    }
-
-   /* Determine which texture coordinate sets are actually needed */
-   if (fprog) {
-      const GLuint coordMask = (1 << MAX_TEXTURE_COORD_UNITS) - 1;
-      ctx->Texture._EnabledCoordUnits
-         = (fprog->Base.InputsRead >> FRAG_ATTRIB_TEX0) & coordMask;
-   }
-   else {
-      ctx->Texture._EnabledCoordUnits = ctx->Texture._EnabledUnits;
-   }
 }
 
 
@@ -811,7 +819,6 @@ _mesa_free_texture_data(GLcontext *ctx)
    for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++)
       ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]);
 
-
 #if FEATURE_colortable
    {
       GLuint i;
@@ -820,3 +827,27 @@ _mesa_free_texture_data(GLcontext *ctx)
    }
 #endif
 }
+
+
+/**
+ * Update the default texture objects in the given context to reference those
+ * specified in the shared state and release those referencing the old 
+ * shared state.
+ */
+void
+_mesa_update_default_objects_texture(GLcontext *ctx)
+{
+   GLuint i;
+
+   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
+
+      _mesa_reference_texobj(&texUnit->Current1D, ctx->Shared->Default1D);
+      _mesa_reference_texobj(&texUnit->Current2D, ctx->Shared->Default2D);
+      _mesa_reference_texobj(&texUnit->Current3D, ctx->Shared->Default3D);
+      _mesa_reference_texobj(&texUnit->CurrentCubeMap, ctx->Shared->DefaultCubeMap);
+      _mesa_reference_texobj(&texUnit->CurrentRect, ctx->Shared->DefaultRect);
+      _mesa_reference_texobj(&texUnit->Current1DArray, ctx->Shared->Default1DArray);
+      _mesa_reference_texobj(&texUnit->Current2DArray, ctx->Shared->Default2DArray);
+   }
+}
diff --git a/src/mesa/main/texstate.h b/src/mesa/main/texstate.h
index b5003d5d6ec..a7d7088c621 100644
--- a/src/mesa/main/texstate.h
+++ b/src/mesa/main/texstate.h
@@ -73,6 +73,9 @@ _mesa_init_texture( GLcontext *ctx );
 extern void 
 _mesa_free_texture_data( GLcontext *ctx );
 
+extern void
+_mesa_update_default_objects_texture(GLcontext *ctx);
+
 /*@}*/
 
 #endif
diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c
index 0fd6a2daae2..8afb947fa17 100644
--- a/src/mesa/main/texstore.c
+++ b/src/mesa/main/texstore.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.3
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -1536,10 +1537,10 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
          for (row = 0; row < srcHeight; row++) {
             GLuint *d4 = (GLuint *) dstRow;
             for (col = 0; col < srcWidth; col++) {
-               d4[col] = ((0xff                    << 24) |
-                          (srcRow[col * 3 + RCOMP] << 16) |
-                          (srcRow[col * 3 + GCOMP] <<  8) |
-                          (srcRow[col * 3 + BCOMP] <<  0));
+               d4[col] = PACK_COLOR_8888(0xff,
+                                         srcRow[col * 3 + RCOMP],
+                                         srcRow[col * 3 + GCOMP],
+                                         srcRow[col * 3 + BCOMP]);
             }
             dstRow += dstRowStride;
             srcRow += srcRowStride;
@@ -1551,8 +1552,7 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
 	    dstFormat == &_mesa_texformat_argb8888 &&
             srcFormat == GL_RGBA &&
 	    baseInternalFormat == GL_RGBA &&
-            srcType == GL_UNSIGNED_BYTE &&
-            littleEndian) {
+            srcType == GL_UNSIGNED_BYTE) {
       /* same as above case, but src data has alpha too */
       GLint img, row, col;
       /* For some reason, streaming copies to write-combined regions
@@ -1573,39 +1573,10 @@ _mesa_texstore_argb8888(TEXSTORE_PARAMS)
          for (row = 0; row < srcHeight; row++) {
             GLuint *d4 = (GLuint *) dstRow;
             for (col = 0; col < srcWidth; col++) {
-               d4[col] = ((srcRow[col * 4 + ACOMP] << 24) |
-                          (srcRow[col * 4 + RCOMP] << 16) |
-                          (srcRow[col * 4 + GCOMP] <<  8) |
-                          (srcRow[col * 4 + BCOMP] <<  0));
-            }
-            dstRow += dstRowStride;
-            srcRow += srcRowStride;
-         }
-      }
-   }
-   else if (!ctx->_ImageTransferState &&
-            !srcPacking->SwapBytes &&
-	    dstFormat == &_mesa_texformat_argb8888 &&
-            srcFormat == GL_RGBA &&
-	    baseInternalFormat == GL_RGBA &&
-            srcType == GL_UNSIGNED_BYTE) {
-
-      GLint img, row, col;
-      for (img = 0; img < srcDepth; img++) {
-         const GLint srcRowStride = _mesa_image_row_stride(srcPacking,
-                                                 srcWidth, srcFormat, srcType);
-         GLubyte *srcRow = (GLubyte *) _mesa_image_address(dims, srcPacking,
-                  srcAddr, srcWidth, srcHeight, srcFormat, srcType, img, 0, 0);
-         GLubyte *dstRow = (GLubyte *) dstAddr
-            + dstImageOffsets[dstZoffset + img] * dstFormat->TexelBytes
-            + dstYoffset * dstRowStride
-            + dstXoffset * dstFormat->TexelBytes;
-         for (row = 0; row < srcHeight; row++) {
-            for (col = 0; col < srcWidth; col++) {
-               dstRow[col * 4 + 0] = srcRow[col * 4 + BCOMP];
-               dstRow[col * 4 + 1] = srcRow[col * 4 + GCOMP];
-               dstRow[col * 4 + 2] = srcRow[col * 4 + RCOMP];
-               dstRow[col * 4 + 3] = srcRow[col * 4 + ACOMP];
+               d4[col] = PACK_COLOR_8888(srcRow[col * 4 + ACOMP],
+                                         srcRow[col * 4 + RCOMP],
+                                         srcRow[col * 4 + GCOMP],
+                                         srcRow[col * 4 + BCOMP]);
             }
             dstRow += dstRowStride;
             srcRow += srcRowStride;
@@ -2692,7 +2663,6 @@ _mesa_texstore_rgba_float16(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_srgb8(TEXSTORE_PARAMS)
 {
-   const GLboolean littleEndian = _mesa_little_endian();
    const struct gl_texture_format *newDstFormat;
    StoreTexImageFunc store;
    GLboolean k;
@@ -2700,14 +2670,8 @@ _mesa_texstore_srgb8(TEXSTORE_PARAMS)
    ASSERT(dstFormat == &_mesa_texformat_srgb8);
 
    /* reuse normal rgb texstore code */
-   if (littleEndian) {
-      newDstFormat = &_mesa_texformat_bgr888;
-      store = _mesa_texstore_bgr888;
-   }
-   else {
-      newDstFormat = &_mesa_texformat_rgb888;
-      store = _mesa_texstore_rgb888;
-   }
+   newDstFormat = &_mesa_texformat_rgb888;
+   store = _mesa_texstore_rgb888;
 
    k = store(ctx, dims, baseInternalFormat,
              newDstFormat, dstAddr,
@@ -2723,17 +2687,13 @@ _mesa_texstore_srgb8(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_srgba8(TEXSTORE_PARAMS)
 {
-   const GLboolean littleEndian = _mesa_little_endian();
    const struct gl_texture_format *newDstFormat;
    GLboolean k;
 
    ASSERT(dstFormat == &_mesa_texformat_srgba8);
 
    /* reuse normal rgba texstore code */
-   if (littleEndian)
-      newDstFormat = &_mesa_texformat_rgba8888_rev;
-   else
-      newDstFormat = &_mesa_texformat_rgba8888;
+   newDstFormat = &_mesa_texformat_rgba8888;
 
    k = _mesa_texstore_rgba8888(ctx, dims, baseInternalFormat,
                                newDstFormat, dstAddr,
@@ -2747,6 +2707,28 @@ _mesa_texstore_srgba8(TEXSTORE_PARAMS)
 
 
 GLboolean
+_mesa_texstore_sargb8(TEXSTORE_PARAMS)
+{
+   const struct gl_texture_format *newDstFormat;
+   GLboolean k;
+
+   ASSERT(dstFormat == &_mesa_texformat_sargb8);
+
+   /* reuse normal rgba texstore code */
+   newDstFormat = &_mesa_texformat_argb8888;
+
+   k = _mesa_texstore_argb8888(ctx, dims, baseInternalFormat,
+                               newDstFormat, dstAddr,
+                               dstXoffset, dstYoffset, dstZoffset,
+                               dstRowStride, dstImageOffsets,
+                               srcWidth, srcHeight, srcDepth,
+                               srcFormat, srcType,
+                               srcAddr, srcPacking);
+   return k;
+}
+
+
+GLboolean
 _mesa_texstore_sl8(TEXSTORE_PARAMS)
 {
    const struct gl_texture_format *newDstFormat;
@@ -2771,17 +2753,13 @@ _mesa_texstore_sl8(TEXSTORE_PARAMS)
 GLboolean
 _mesa_texstore_sla8(TEXSTORE_PARAMS)
 {
-   const GLboolean littleEndian = _mesa_little_endian();
    const struct gl_texture_format *newDstFormat;
    GLboolean k;
 
    ASSERT(dstFormat == &_mesa_texformat_sla8);
 
    /* reuse normal luminance/alpha texstore code */
-   if (littleEndian)
-      newDstFormat = &_mesa_texformat_al88;
-   else
-      newDstFormat = &_mesa_texformat_al88_rev;
+   newDstFormat = &_mesa_texformat_al88;
 
    k = _mesa_texstore_al88(ctx, dims, baseInternalFormat,
                            newDstFormat, dstAddr,
@@ -3018,10 +2996,13 @@ choose_texture_format(GLcontext *ctx, struct gl_texture_image *texImage,
 
 
 
-/*
+/**
  * This is the software fallback for Driver.TexImage1D()
  * and Driver.CopyTexImage1D().
  * \sa _mesa_store_teximage2d()
+ * Note that the width may not be the actual texture width since it may
+ * be changed by convolution w/ GL_REDUCE.  The texImage->Width field will
+ * have the actual texture size.
  */
 void
 _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level,
@@ -3032,21 +3013,16 @@ _mesa_store_teximage1d(GLcontext *ctx, GLenum target, GLint level,
                        struct gl_texture_object *texObj,
                        struct gl_texture_image *texImage)
 {
-   GLint postConvWidth = width;
    GLint sizeInBytes;
    (void) border;
 
-   if (ctx->_ImageTransferState & IMAGE_CONVOLUTION_BIT) {
-      _mesa_adjust_image_for_convolution(ctx, 1, &postConvWidth, NULL);
-   }
-
    choose_texture_format(ctx, texImage, 1, format, type, internalFormat);
 
    /* allocate memory */
    if (texImage->IsCompressed)
       sizeInBytes = texImage->CompressedSize;
    else
-      sizeInBytes = postConvWidth * texImage->TexFormat->TexelBytes;
+      sizeInBytes = texImage->Width * texImage->TexFormat->TexelBytes;
    texImage->Data = _mesa_alloc_texmemory(sizeInBytes);
    if (!texImage->Data) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage1D");
@@ -3106,15 +3082,9 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level,
                        struct gl_texture_object *texObj,
                        struct gl_texture_image *texImage)
 {
-   GLint postConvWidth = width, postConvHeight = height;
    GLint texelBytes, sizeInBytes;
    (void) border;
 
-   if (ctx->_ImageTransferState & IMAGE_CONVOLUTION_BIT) {
-      _mesa_adjust_image_for_convolution(ctx, 2, &postConvWidth,
-                                         &postConvHeight);
-   }
-
    choose_texture_format(ctx, texImage, 2, format, type, internalFormat);
 
    texelBytes = texImage->TexFormat->TexelBytes;
@@ -3123,7 +3093,7 @@ _mesa_store_teximage2d(GLcontext *ctx, GLenum target, GLint level,
    if (texImage->IsCompressed)
       sizeInBytes = texImage->CompressedSize;
    else
-      sizeInBytes = postConvWidth * postConvHeight * texelBytes;
+      sizeInBytes = texImage->Width * texImage->Height * texelBytes;
    texImage->Data = _mesa_alloc_texmemory(sizeInBytes);
    if (!texImage->Data) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage2D");
@@ -3619,6 +3589,7 @@ is_srgb_teximage(const struct gl_texture_image *texImage)
    switch (texImage->TexFormat->MesaFormat) {
    case MESA_FORMAT_SRGB8:
    case MESA_FORMAT_SRGBA8:
+   case MESA_FORMAT_SARGB8:
    case MESA_FORMAT_SL8:
    case MESA_FORMAT_SLA8:
       return GL_TRUE;
@@ -3751,12 +3722,24 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level,
                MEMCPY(dest,
                       (const GLubyte *) texImage->Data + row * rowstride,
                       comps * width * sizeof(GLubyte));
+               /* FIXME: isn't it necessary to still do component assigning
+                  according to format/type? */
+               /* FIXME: need to do something else for compressed srgb textures
+                         (currently will return values converted to linear) */
             }
 #endif /* FEATURE_EXT_texture_sRGB */
             else {
                /* general case:  convert row to RGBA format */
                GLfloat rgba[MAX_WIDTH][4];
                GLint col;
+               GLbitfield transferOps = 0x0;
+
+               if (type == GL_FLOAT && 
+                   ((ctx->Color.ClampReadColor == GL_TRUE) ||
+                    (ctx->Color.ClampReadColor == GL_FIXED_ONLY_ARB &&
+                     texImage->TexFormat->DataType != GL_FLOAT)))
+                  transferOps |= IMAGE_CLAMP_BIT;
+
                for (col = 0; col < width; col++) {
                   (*texImage->FetchTexelf)(texImage, col, row, img, rgba[col]);
                   if (texImage->TexFormat->BaseFormat == GL_ALPHA) {
@@ -3781,7 +3764,7 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level,
                }
                _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba,
                                           format, type, dest,
-                                          &ctx->Pack, 0x0 /*image xfer ops*/);
+                                          &ctx->Pack, transferOps /*image xfer ops*/);
             } /* format */
          } /* row */
       } /* img */
@@ -3802,8 +3785,8 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level,
 void
 _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
                               GLvoid *img,
-                              const struct gl_texture_object *texObj,
-                              const struct gl_texture_image *texImage)
+                              struct gl_texture_object *texObj,
+                              struct gl_texture_image *texImage)
 {
    GLuint size;
 
diff --git a/src/mesa/main/texstore.h b/src/mesa/main/texstore.h
index c5813fbeeec..8dc1c963cda 100644
--- a/src/mesa/main/texstore.h
+++ b/src/mesa/main/texstore.h
@@ -3,6 +3,7 @@
  * Version:  6.5.1
  *
  * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (c) 2008 VMware, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -71,6 +72,7 @@ extern GLboolean _mesa_texstore_rgba_dxt5(TEXSTORE_PARAMS);
 #if FEATURE_EXT_texture_sRGB
 extern GLboolean _mesa_texstore_srgb8(TEXSTORE_PARAMS);
 extern GLboolean _mesa_texstore_srgba8(TEXSTORE_PARAMS);
+extern GLboolean _mesa_texstore_sargb8(TEXSTORE_PARAMS);
 extern GLboolean _mesa_texstore_sl8(TEXSTORE_PARAMS);
 extern GLboolean _mesa_texstore_sla8(TEXSTORE_PARAMS);
 #endif
@@ -215,8 +217,8 @@ _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level,
 extern void
 _mesa_get_compressed_teximage(GLcontext *ctx, GLenum target, GLint level,
                               GLvoid *img,
-                              const struct gl_texture_object *texObj,
-                              const struct gl_texture_image *texImage);
+                              struct gl_texture_object *texObj,
+                              struct gl_texture_image *texImage);
 
 extern const GLvoid *
 _mesa_validate_pbo_teximage(GLcontext *ctx, GLuint dimensions,
diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index 9f1fcd9d262..9d9b28b5189 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -499,7 +499,7 @@ void GLAPIENTRY
 _mesa_VertexAttribPointerNV(GLuint index, GLint size, GLenum type,
                             GLsizei stride, const GLvoid *ptr)
 {
-   const GLboolean normalized = GL_FALSE;
+   GLboolean normalized = GL_FALSE;
    GLsizei elementSize;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
@@ -527,6 +527,7 @@ _mesa_VertexAttribPointerNV(GLuint index, GLint size, GLenum type,
    /* check for valid 'type' and compute StrideB right away */
    switch (type) {
       case GL_UNSIGNED_BYTE:
+         normalized = GL_TRUE;
          elementSize = size * sizeof(GLubyte);
          break;
       case GL_SHORT:
@@ -865,16 +866,22 @@ _mesa_LockArraysEXT(GLint first, GLsizei count)
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(ctx, "glLockArrays %d %d\n", first, count);
 
-   if (first == 0 && count > 0 &&
-       count <= (GLint) ctx->Const.MaxArrayLockSize) {
-      ctx->Array.LockFirst = first;
-      ctx->Array.LockCount = count;
+   if (first < 0) {
+      _mesa_error( ctx, GL_INVALID_VALUE, "glLockArraysEXT(first)" );
+      return;
    }
-   else {
-      ctx->Array.LockFirst = 0;
-      ctx->Array.LockCount = 0;
+   if (count <= 0) {
+      _mesa_error( ctx, GL_INVALID_VALUE, "glLockArraysEXT(count)" );
+      return;
+   }
+   if (ctx->Array.LockCount != 0) {
+      _mesa_error( ctx, GL_INVALID_OPERATION, "glLockArraysEXT(reentry)" );
+      return;
    }
 
+   ctx->Array.LockFirst = first;
+   ctx->Array.LockCount = count;
+
    ctx->NewState |= _NEW_ARRAY;
    ctx->Array.NewState |= _NEW_ARRAY_ALL;
 
@@ -892,6 +899,11 @@ _mesa_UnlockArraysEXT( void )
    if (MESA_VERBOSE & VERBOSE_API)
       _mesa_debug(ctx, "glUnlockArrays\n");
 
+   if (ctx->Array.LockCount == 0) {
+      _mesa_error( ctx, GL_INVALID_OPERATION, "glUnlockArraysEXT(reexit)" );
+      return;
+   }
+
    ctx->Array.LockFirst = 0;
    ctx->Array.LockCount = 0;
    ctx->NewState |= _NEW_ARRAY;
diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h
index f557940738e..97d5c8219d0 100644
--- a/src/mesa/main/varray.h
+++ b/src/mesa/main/varray.h
@@ -150,6 +150,11 @@ _mesa_MultiModeDrawElementsIBM( const GLenum * mode, const GLsizei * count,
 				GLenum type, const GLvoid * const * indices,
 				GLsizei primcount, GLint modestride );
 
+extern void GLAPIENTRY
+_mesa_LockArraysEXT(GLint first, GLsizei count);
+
+extern void GLAPIENTRY
+_mesa_UnlockArraysEXT( void );
 
 
 extern void GLAPIENTRY
diff --git a/src/mesa/main/version.h b/src/mesa/main/version.h
index 9229077f42c..81034a35d1e 100644
--- a/src/mesa/main/version.h
+++ b/src/mesa/main/version.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -29,9 +29,9 @@
 
 /* Mesa version */
 #define MESA_MAJOR 7
-#define MESA_MINOR 1
+#define MESA_MINOR 3
 #define MESA_PATCH 0
-#define MESA_VERSION_STRING "7.1"
+#define MESA_VERSION_STRING "7.3-devel"
 
 /* To make version comparison easy */
 #define MESA_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
diff --git a/src/mesa/math/descrip.mms b/src/mesa/math/descrip.mms
new file mode 100644
index 00000000000..3aaa6eb8b4e
--- /dev/null
+++ b/src/mesa/math/descrip.mms
@@ -0,0 +1,47 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define glapi [-.glapi]
+	define main [-.main]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.main],[-.glapi]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = m_debug_clip.c m_debug_norm.c m_debug_xform.c m_eval.c m_matrix.c\
+	m_translate.c m_vector.c m_xform.c
+
+OBJECTS = m_debug_clip.obj,m_debug_norm.obj,m_debug_xform.obj,m_eval.obj,\
+	m_matrix.obj,m_translate.obj,m_vector.obj,m_xform.obj
+ 
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+m_debug_clip.obj : m_debug_clip.c
+m_debug_norm.obj : m_debug_norm.c
+m_debug_xform.obj : m_debug_xform.c
+m_eval.obj : m_eval.c
+m_matrix.obj : m_matrix.c
+m_translate.obj : m_translate.c
+m_vector.obj : m_vector.c
+m_xform.obj : m_xform.c
diff --git a/src/mesa/math/m_debug_util.h b/src/mesa/math/m_debug_util.h
index 776c26475e7..7abe6f2565a 100644
--- a/src/mesa/math/m_debug_util.h
+++ b/src/mesa/math/m_debug_util.h
@@ -185,7 +185,7 @@ extern char *mesa_profile;
 
 #endif
 
-#elif defined(__amd64__)
+#elif defined(__x86_64__)
 
 #define rdtscll(val) do { \
      unsigned int a,d; \
diff --git a/src/mesa/math/m_vector.h b/src/mesa/math/m_vector.h
index b1a65b33569..647388ac7d0 100644
--- a/src/mesa/math/m_vector.h
+++ b/src/mesa/math/m_vector.h
@@ -1,9 +1,8 @@
-
 /*
  * Mesa 3-D graphics library
- * Version:  3.5
+ * Version:  7.3
  *
- * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/src/mesa/math/m_xform.h b/src/mesa/math/m_xform.h
index d1b974f0433..24e8ddbd57c 100644
--- a/src/mesa/math/m_xform.h
+++ b/src/mesa/math/m_xform.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.3
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/src/mesa/math/sources b/src/mesa/math/sources
deleted file mode 100644
index 7c7dcccedfd..00000000000
--- a/src/mesa/math/sources
+++ /dev/null
@@ -1,25 +0,0 @@
-MESA_MATH_SOURCES = \
-m_debug_clip.c \
-m_debug_norm.c \
-m_debug_xform.c \
-m_eval.c \
-m_matrix.c \
-m_translate.c \
-m_vector.c \
-m_xform.c
-
-MESA_MATH_HEADERS = \
-m_clip_tmp.h \
-m_copy_tmp.h \
-m_debug.h \
-m_debug_util.h \
-m_dotprod_tmp.h \
-m_eval.h \
-m_matrix.h \
-m_norm_tmp.h \
-m_trans_tmp.h \
-m_translate.h \
-m_vector.h \
-m_xform.h \
-m_xform_tmp.h \
-mathmod.h
diff --git a/src/mesa/osmesa.pc.in b/src/mesa/osmesa.pc.in
new file mode 100644
index 00000000000..05327f40aa0
--- /dev/null
+++ b/src/mesa/osmesa.pc.in
@@ -0,0 +1,12 @@
+prefix=@INSTALL_DIR@
+exec_prefix=${prefix}
+libdir=@INSTALL_LIB_DIR@
+includedir=@INSTALL_INC_DIR@
+
+Name: osmesa
+Description: Mesa Off-screen Rendering library
+Requires: @OSMESA_PC_REQ@
+Version: @VERSION@
+Libs: -L${libdir} -l@OSMESA_LIB@
+Libs.private: @OSMESA_PC_LIB_PRIV@
+Cflags: -I${includedir}
diff --git a/src/mesa/ppc/common_ppc.c b/src/mesa/ppc/common_ppc.c
index 13526df78c9..c7cec96e5d5 100644
--- a/src/mesa/ppc/common_ppc.c
+++ b/src/mesa/ppc/common_ppc.c
@@ -38,6 +38,9 @@
 #include <elf.h>
 #endif
 
+#include "common_ppc_features.h"
+
+
 unsigned long _mesa_ppc_cpu_features = 0;
 
 /**
@@ -82,7 +85,7 @@ void _mesa_init_all_ppc_transform_asm( void )
    }
    
 # ifndef USE_VMX_ASM
-   _mesa_ppc_cpu_features &= ~PPC_FEATURES_HAS_ALTIVEC;
+   _mesa_ppc_cpu_features &= ~PPC_FEATURE_HAS_ALTIVEC;
 # endif
 #endif
 }
diff --git a/src/mesa/ppc/common_ppc_features.h b/src/mesa/ppc/common_ppc_features.h
index 4d46ca04c9d..9cde422d72d 100644
--- a/src/mesa/ppc/common_ppc_features.h
+++ b/src/mesa/ppc/common_ppc_features.h
@@ -48,4 +48,7 @@ extern unsigned long _mesa_ppc_cpu_features;
 #define cpu_has_fpu ((_mesa_ppc_cpu_features & PPC_FEATURE_HAS_FPU) != 0)
 
 #endif /* USE_PPC_ASM */
+
+extern void _mesa_init_all_ppc_transform_asm( void );
+
 #endif /* COMMON_PPC_FEATURES_H */
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index 4d89d057c7d..536404bf978 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -53,23 +53,17 @@ having three separate program parameter arrays.
 
 #include "main/glheader.h"
 #include "main/imports.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
 #include "shader/grammar/grammar_mesa.h"
 #include "arbprogparse.h"
 #include "program.h"
 #include "programopt.h"
 #include "prog_parameter.h"
 #include "prog_statevars.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
 #include "prog_instruction.h"
 
-
-/* For ARB programs, use the NV instruction limits */
-#define MAX_INSTRUCTIONS MAX2(MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS, \
-                              MAX_NV_VERTEX_PROGRAM_INSTRUCTIONS)
-
-
 /**
  * This is basically a union of the vertex_program and fragment_program
  * structs that we can use to parse the program into
@@ -3443,6 +3437,8 @@ parse_instructions(GLcontext * ctx, const GLubyte * inst,
       : ctx->Const.VertexProgram.MaxInstructions;
    GLint err = 0;
 
+   ASSERT(MAX_PROGRAM_INSTRUCTIONS >= maxInst);
+
    Program->MajorVersion = (GLuint) * inst++;
    Program->MinorVersion = (GLuint) * inst++;
 
@@ -3796,7 +3792,7 @@ _mesa_parse_arb_program(GLcontext *ctx, GLenum target,
 
    /* Initialize the arb_program struct */
    program->Base.String = strz;
-   program->Base.Instructions = _mesa_alloc_instructions(MAX_INSTRUCTIONS);
+   program->Base.Instructions = _mesa_alloc_instructions(MAX_PROGRAM_INSTRUCTIONS);
    program->Base.NumInstructions =
    program->Base.NumTemporaries =
    program->Base.NumParameters =
@@ -3841,12 +3837,12 @@ _mesa_parse_arb_program(GLcontext *ctx, GLenum target,
 
    _mesa_free (parsed);
 
-   /* Reallocate the instruction array from size [MAX_INSTRUCTIONS]
+   /* Reallocate the instruction array from size [MAX_PROGRAM_INSTRUCTIONS]
     * to size [ap.Base.NumInstructions].
     */
    program->Base.Instructions
       = _mesa_realloc_instructions(program->Base.Instructions,
-                                   MAX_INSTRUCTIONS,
+                                   MAX_PROGRAM_INSTRUCTIONS,
                                    program->Base.NumInstructions);
 
    return !err;
@@ -3899,6 +3895,9 @@ _mesa_parse_arb_fragment_program(GLcontext* ctx, GLenum target,
    program->FogOption          = ap.FogOption;
    program->UsesKill          = ap.UsesKill;
 
+   if (program->FogOption)
+      program->Base.InputsRead |= FRAG_BIT_FOGC;
+      
    if (program->Base.Instructions)
       _mesa_free(program->Base.Instructions);
    program->Base.Instructions = ap.Base.Instructions;
diff --git a/src/mesa/shader/arbprogram.c b/src/mesa/shader/arbprogram.c
index beb5deea505..760dac23993 100644
--- a/src/mesa/shader/arbprogram.c
+++ b/src/mesa/shader/arbprogram.c
@@ -30,13 +30,13 @@
 
 
 #include "main/glheader.h"
-#include "arbprogram.h"
-#include "arbprogparse.h"
 #include "main/context.h"
 #include "main/hash.h"
 #include "main/imports.h"
 #include "main/macros.h"
 #include "main/mtypes.h"
+#include "arbprogram.h"
+#include "arbprogparse.h"
 #include "program.h"
 
 
diff --git a/src/mesa/shader/descrip.mms b/src/mesa/shader/descrip.mms
new file mode 100644
index 00000000000..19bafd48302
--- /dev/null
+++ b/src/mesa/shader/descrip.mms
@@ -0,0 +1,95 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 29 September 2008
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define swrast [-.swrast]
+	define array_cache [-.array_cache]
+	define glapi [-.glapi]
+	define main [-.main]
+	define shader [-.shader]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[.grammar],[-.main],[-.glapi],[.slang]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1,"__extension__=")/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = \
+	atifragshader.c \
+	arbprogparse.c \
+	arbprogram.c \
+	nvfragparse.c \
+	nvprogram.c \
+	nvvertparse.c \
+	program.c \
+	programopt.c \
+	prog_debug.c \
+	prog_execute.c \
+	prog_instruction.c \
+	prog_parameter.c \
+	prog_print.c \
+	prog_cache.c \
+	prog_statevars.c \
+	shader_api.c prog_uniform.c
+
+OBJECTS = \
+	atifragshader.obj,\
+	arbprogparse.obj,\
+	arbprogram.obj,\
+	nvfragparse.obj,\
+	nvprogram.obj,\
+	nvvertparse.obj,\
+	program.obj,\
+	programopt.obj,\
+	prog_debug.obj,\
+	prog_execute.obj,\
+	prog_instruction.obj,\
+	prog_parameter.obj,\
+	prog_print.obj,\
+	prog_statevars.obj,\
+	shader_api.obj,prog_uniform.obj,prog_cache.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+all : 
+	$(MMS)$(MMSQUALIFIERS) $(LIBDIR)$(GL_LIB)
+	set def [.slang]
+	$(MMS)$(MMSQUALIFIERS)
+	set def [-.grammar]
+	$(MMS)$(MMSQUALIFIERS)
+	set def [-]
+
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+atifragshader.obj : atifragshader.c
+arbprogparse.obj : arbprogparse.c
+arbprogram.obj : arbprogram.c
+nvfragparse.obj : nvfragparse.c
+nvprogram.obj : nvprogram.c
+nvvertparse.obj : nvvertparse.c
+program.obj : program.c
+programopt. obj : programopt.c
+prog_debug.obj : prog_debug.c
+prog_execute.obj : prog_execute.c
+prog_instruction.obj : prog_instruction.c
+prog_parameter.obj : prog_parameter.c
+prog_print.obj : prog_print.c
+prog_statevars.obj : prog_statevars.c
+shader_api.obj : shader_api.c
+prog_uniform.obj : prog_uniform.c
+prog_cache.obj : prog_cache.c
diff --git a/src/mesa/shader/grammar/grammar_crt.c b/src/mesa/shader/grammar/grammar_crt.c
index bdf2da9b2e2..d2c95d1c8e7 100644
--- a/src/mesa/shader/grammar/grammar_crt.c
+++ b/src/mesa/shader/grammar/grammar_crt.c
@@ -10,17 +10,17 @@ void grammar_alloc_free (void *ptr)
     free (ptr);
 }
 
-void *grammar_alloc_malloc (unsigned int size)
+void *grammar_alloc_malloc (size_t size)
 {
     return malloc (size);
 }
 
-void *grammar_alloc_realloc (void *ptr, unsigned int old_size, unsigned int size)
+void *grammar_alloc_realloc (void *ptr, size_t old_size, size_t size)
 {
     return realloc (ptr, size);
 }
 
-void *grammar_memory_copy (void *dst, const void * src, unsigned int size)
+void *grammar_memory_copy (void *dst, const void * src, size_t size)
 {
     return memcpy (dst, src, size);
 }
@@ -30,7 +30,7 @@ int grammar_string_compare (const byte *str1, const byte *str2)
     return strcmp ((const char *) str1, (const char *) str2);
 }
 
-int grammar_string_compare_n (const byte *str1, const byte *str2, unsigned int n)
+int grammar_string_compare_n (const byte *str1, const byte *str2, size_t n)
 {
     return strncmp ((const char *) str1, (const char *) str2, n);
 }
@@ -40,7 +40,7 @@ byte *grammar_string_copy (byte *dst, const byte *src)
     return (byte *) strcpy ((char *) dst, (const char *) src);
 }
 
-byte *grammar_string_copy_n (byte *dst, const byte *src, unsigned int n)
+byte *grammar_string_copy_n (byte *dst, const byte *src, size_t n)
 {
     return (byte *) strncpy ((char *) dst, (const char *) src, n);
 }
diff --git a/src/mesa/shader/grammar/sources b/src/mesa/shader/grammar/sources
deleted file mode 100644
index a6bbfd3ffd3..00000000000
--- a/src/mesa/shader/grammar/sources
+++ /dev/null
@@ -1,8 +0,0 @@
-MESA_SHADER_GRAMMAR_SOURCES = \
-grammar_mesa.c
-
-MESA_SHADER_GRAMMAR_HEADERS = \
-grammar.c \
-grammar.h \
-grammar_mesa.h \
-grammar_syn.h
diff --git a/src/mesa/shader/nvfragparse.c b/src/mesa/shader/nvfragparse.c
index a2a7a5f3f49..20e47813729 100644
--- a/src/mesa/shader/nvfragparse.c
+++ b/src/mesa/shader/nvfragparse.c
@@ -41,10 +41,10 @@
 #include "main/context.h"
 #include "main/imports.h"
 #include "main/macros.h"
+#include "program.h"
 #include "prog_parameter.h"
 #include "prog_instruction.h"
 #include "nvfragparse.h"
-#include "program.h"
 
 
 #define INPUT_1V     1
diff --git a/src/mesa/shader/nvprogram.c b/src/mesa/shader/nvprogram.c
index d656d4b28b6..88272fff3ff 100644
--- a/src/mesa/shader/nvprogram.c
+++ b/src/mesa/shader/nvprogram.c
@@ -42,12 +42,12 @@
 #include "main/hash.h"
 #include "main/imports.h"
 #include "main/macros.h"
+#include "program.h"
 #include "prog_parameter.h"
 #include "prog_instruction.h"
 #include "nvfragparse.h"
 #include "nvvertparse.h"
 #include "nvprogram.h"
-#include "program.h"
 
 
 
diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c
index 768e936d5f6..923611b7975 100644
--- a/src/mesa/shader/prog_execute.c
+++ b/src/mesa/shader/prog_execute.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.0.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -76,56 +76,52 @@ static const GLfloat ZeroVec[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
  * source register.
  */
 static INLINE const GLfloat *
-get_register_pointer(const struct prog_src_register *source,
-                     const struct gl_program_machine *machine)
+get_src_register_pointer(const struct prog_src_register *source,
+                         const struct gl_program_machine *machine)
 {
+   const struct gl_program *prog = machine->CurProgram;
+   GLint reg = source->Index;
+
    if (source->RelAddr) {
-      const GLint reg = source->Index + machine->AddressReg[0][0];
-      if (source->File == PROGRAM_ENV_PARAM) {
-         if (reg < 0 || reg >= MAX_PROGRAM_ENV_PARAMS)
-            return ZeroVec;
-         else
-            return machine->EnvParams[reg];
-      }
-      else {
-         const struct gl_program_parameter_list *params;
-         ASSERT(source->File == PROGRAM_LOCAL_PARAM ||
-                source->File == PROGRAM_CONSTANT ||
-                source->File == PROGRAM_STATE_VAR);
-         params = machine->CurProgram->Parameters;
-         if (reg < 0 || reg >= (GLint)params->NumParameters)
-            return ZeroVec;
-         else
-            return params->ParameterValues[reg];
+      /* add address register value to src index/offset */
+      reg += machine->AddressReg[0][0];
+      if (reg < 0) {
+         return ZeroVec;
       }
    }
 
    switch (source->File) {
    case PROGRAM_TEMPORARY:
-      ASSERT(source->Index < MAX_PROGRAM_TEMPS);
-      return machine->Temporaries[source->Index];
+      if (reg >= MAX_PROGRAM_TEMPS)
+         return ZeroVec;
+      return machine->Temporaries[reg];
 
    case PROGRAM_INPUT:
-      if (machine->CurProgram->Target == GL_VERTEX_PROGRAM_ARB) {
-         ASSERT(source->Index < VERT_ATTRIB_MAX);
-         return machine->VertAttribs[source->Index];
+      if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
+         if (reg >= VERT_ATTRIB_MAX)
+            return ZeroVec;
+         return machine->VertAttribs[reg];
       }
       else {
-         ASSERT(source->Index < FRAG_ATTRIB_MAX);
-         return machine->Attribs[source->Index][machine->CurElement];
+         if (reg >= FRAG_ATTRIB_MAX)
+            return ZeroVec;
+         return machine->Attribs[reg][machine->CurElement];
       }
 
    case PROGRAM_OUTPUT:
-      ASSERT(source->Index < MAX_PROGRAM_OUTPUTS);
-      return machine->Outputs[source->Index];
+      if (reg >= MAX_PROGRAM_OUTPUTS)
+         return ZeroVec;
+      return machine->Outputs[reg];
 
    case PROGRAM_LOCAL_PARAM:
-      ASSERT(source->Index < MAX_PROGRAM_LOCAL_PARAMS);
-      return machine->CurProgram->LocalParams[source->Index];
+      if (reg >= MAX_PROGRAM_LOCAL_PARAMS)
+         return ZeroVec;
+      return machine->CurProgram->LocalParams[reg];
 
    case PROGRAM_ENV_PARAM:
-      ASSERT(source->Index < MAX_PROGRAM_ENV_PARAMS);
-      return machine->EnvParams[source->Index];
+      if (reg >= MAX_PROGRAM_ENV_PARAMS)
+         return ZeroVec;
+      return machine->EnvParams[reg];
 
    case PROGRAM_STATE_VAR:
       /* Fallthrough */
@@ -134,19 +130,62 @@ get_register_pointer(const struct prog_src_register *source,
    case PROGRAM_UNIFORM:
       /* Fallthrough */
    case PROGRAM_NAMED_PARAM:
-      ASSERT(source->Index <
-             (GLint) machine->CurProgram->Parameters->NumParameters);
-      return machine->CurProgram->Parameters->ParameterValues[source->Index];
+      if (reg >= (GLint) prog->Parameters->NumParameters)
+         return ZeroVec;
+      return prog->Parameters->ParameterValues[reg];
+
+   default:
+      _mesa_problem(NULL,
+         "Invalid src register file %d in get_src_register_pointer()",
+         source->File);
+      return NULL;
+   }
+}
+
+
+/**
+ * Return a pointer to the 4-element float vector specified by the given
+ * destination register.
+ */
+static INLINE GLfloat *
+get_dst_register_pointer(const struct prog_dst_register *dest,
+                         struct gl_program_machine *machine)
+{
+   static GLfloat dummyReg[4];
+   GLint reg = dest->Index;
+
+   if (dest->RelAddr) {
+      /* add address register value to src index/offset */
+      reg += machine->AddressReg[0][0];
+      if (reg < 0) {
+         return dummyReg;
+      }
+   }
+
+   switch (dest->File) {
+   case PROGRAM_TEMPORARY:
+      if (reg >= MAX_PROGRAM_TEMPS)
+         return dummyReg;
+      return machine->Temporaries[reg];
+
+   case PROGRAM_OUTPUT:
+      if (reg >= MAX_PROGRAM_OUTPUTS)
+         return dummyReg;
+      return machine->Outputs[reg];
+
+   case PROGRAM_WRITE_ONLY:
+      return dummyReg;
 
    default:
       _mesa_problem(NULL,
-                    "Invalid input register file %d in get_register_pointer()",
-                    source->File);
+         "Invalid dest register file %d in get_dst_register_pointer()",
+         dest->File);
       return NULL;
    }
 }
 
 
+
 #if FEATURE_MESA_program_debug
 static struct gl_program_machine *CurrentMachine = NULL;
 
@@ -160,12 +199,12 @@ _mesa_get_program_register(GLcontext *ctx, enum register_file file,
                            GLuint index, GLfloat val[4])
 {
    if (CurrentMachine) {
-      struct prog_src_register src;
-      const GLfloat *reg;
-      src.File = file;
-      src.Index = index;
-      reg = get_register_pointer(&src, CurrentMachine);
-      COPY_4V(val, reg);
+      struct prog_src_register srcReg;
+      const GLfloat *src;
+      srcReg.File = file;
+      srcReg.Index = index;
+      src = get_src_register_pointer(&srcReg, CurrentMachine);
+      COPY_4V(val, src);
    }
 }
 #endif /* FEATURE_MESA_program_debug */
@@ -179,7 +218,7 @@ static void
 fetch_vector4(const struct prog_src_register *source,
               const struct gl_program_machine *machine, GLfloat result[4])
 {
-   const GLfloat *src = get_register_pointer(source, machine);
+   const GLfloat *src = get_src_register_pointer(source, machine);
    ASSERT(src);
 
    if (source->Swizzle == SWIZZLE_NOOP) {
@@ -219,6 +258,37 @@ fetch_vector4(const struct prog_src_register *source,
 
 
 /**
+ * Fetch a 4-element uint vector from the given source register.
+ * Apply swizzling but not negation/abs.
+ */
+static void
+fetch_vector4ui(const struct prog_src_register *source,
+                const struct gl_program_machine *machine, GLuint result[4])
+{
+   const GLuint *src = (GLuint *) get_src_register_pointer(source, machine);
+   ASSERT(src);
+
+   if (source->Swizzle == SWIZZLE_NOOP) {
+      /* no swizzling */
+      COPY_4V(result, src);
+   }
+   else {
+      ASSERT(GET_SWZ(source->Swizzle, 0) <= 3);
+      ASSERT(GET_SWZ(source->Swizzle, 1) <= 3);
+      ASSERT(GET_SWZ(source->Swizzle, 2) <= 3);
+      ASSERT(GET_SWZ(source->Swizzle, 3) <= 3);
+      result[0] = src[GET_SWZ(source->Swizzle, 0)];
+      result[1] = src[GET_SWZ(source->Swizzle, 1)];
+      result[2] = src[GET_SWZ(source->Swizzle, 2)];
+      result[3] = src[GET_SWZ(source->Swizzle, 3)];
+   }
+
+   /* Note: no NegateBase, Abs, NegateAbs here */
+}
+
+
+
+/**
  * Fetch the derivative with respect to X or Y for the given register.
  * XXX this currently only works for fragment program input attribs.
  */
@@ -228,7 +298,8 @@ fetch_vector4_deriv(GLcontext * ctx,
                     const struct gl_program_machine *machine,
                     char xOrY, GLfloat result[4])
 {
-   if (source->File == PROGRAM_INPUT && source->Index < (GLint)machine->NumDeriv) {
+   if (source->File == PROGRAM_INPUT &&
+       source->Index < (GLint) machine->NumDeriv) {
       const GLint col = machine->CurElement;
       const GLfloat w = machine->Attribs[FRAG_ATTRIB_WPOS][col][3];
       const GLfloat invQ = 1.0f / w;
@@ -284,7 +355,7 @@ static void
 fetch_vector1(const struct prog_src_register *source,
               const struct gl_program_machine *machine, GLfloat result[4])
 {
-   const GLfloat *src = get_register_pointer(source, machine);
+   const GLfloat *src = get_src_register_pointer(source, machine);
    ASSERT(src);
 
    result[0] = src[GET_SWZ(source->Swizzle, 0)];
@@ -399,29 +470,11 @@ static void
 store_vector4(const struct prog_instruction *inst,
               struct gl_program_machine *machine, const GLfloat value[4])
 {
-   const struct prog_dst_register *dest = &(inst->DstReg);
+   const struct prog_dst_register *dstReg = &(inst->DstReg);
    const GLboolean clamp = inst->SaturateMode == SATURATE_ZERO_ONE;
-   GLfloat *dstReg;
-   GLfloat dummyReg[4];
+   GLuint writeMask = dstReg->WriteMask;
    GLfloat clampedValue[4];
-   GLuint writeMask = dest->WriteMask;
-
-   switch (dest->File) {
-   case PROGRAM_OUTPUT:
-      ASSERT(dest->Index < MAX_PROGRAM_OUTPUTS);
-      dstReg = machine->Outputs[dest->Index];
-      break;
-   case PROGRAM_TEMPORARY:
-      ASSERT(dest->Index < MAX_PROGRAM_TEMPS);
-      dstReg = machine->Temporaries[dest->Index];
-      break;
-   case PROGRAM_WRITE_ONLY:
-      dstReg = dummyReg;
-      return;
-   default:
-      _mesa_problem(NULL, "bad register file in store_vector4(fp)");
-      return;
-   }
+   GLfloat *dst = get_dst_register_pointer(dstReg, machine);
 
 #if 0
    if (value[0] > 1.0e10 ||
@@ -439,38 +492,38 @@ store_vector4(const struct prog_instruction *inst,
       value = clampedValue;
    }
 
-   if (dest->CondMask != COND_TR) {
+   if (dstReg->CondMask != COND_TR) {
       /* condition codes may turn off some writes */
       if (writeMask & WRITEMASK_X) {
-         if (!test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 0)],
-                      dest->CondMask))
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
+                      dstReg->CondMask))
             writeMask &= ~WRITEMASK_X;
       }
       if (writeMask & WRITEMASK_Y) {
-         if (!test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 1)],
-                      dest->CondMask))
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
+                      dstReg->CondMask))
             writeMask &= ~WRITEMASK_Y;
       }
       if (writeMask & WRITEMASK_Z) {
-         if (!test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 2)],
-                      dest->CondMask))
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
+                      dstReg->CondMask))
             writeMask &= ~WRITEMASK_Z;
       }
       if (writeMask & WRITEMASK_W) {
-         if (!test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 3)],
-                      dest->CondMask))
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
+                      dstReg->CondMask))
             writeMask &= ~WRITEMASK_W;
       }
    }
 
    if (writeMask & WRITEMASK_X)
-      dstReg[0] = value[0];
+      dst[0] = value[0];
    if (writeMask & WRITEMASK_Y)
-      dstReg[1] = value[1];
+      dst[1] = value[1];
    if (writeMask & WRITEMASK_Z)
-      dstReg[2] = value[2];
+      dst[2] = value[2];
    if (writeMask & WRITEMASK_W)
-      dstReg[3] = value[3];
+      dst[3] = value[3];
 
    if (inst->CondUpdate) {
       if (writeMask & WRITEMASK_X)
@@ -493,6 +546,71 @@ store_vector4(const struct prog_instruction *inst,
 
 
 /**
+ * Store 4 uints into a register.  Observe the set-condition-code flags.
+ */
+static void
+store_vector4ui(const struct prog_instruction *inst,
+                struct gl_program_machine *machine, const GLuint value[4])
+{
+   const struct prog_dst_register *dstReg = &(inst->DstReg);
+   GLuint writeMask = dstReg->WriteMask;
+   GLuint *dst = (GLuint *) get_dst_register_pointer(dstReg, machine);
+
+   if (dstReg->CondMask != COND_TR) {
+      /* condition codes may turn off some writes */
+      if (writeMask & WRITEMASK_X) {
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
+                      dstReg->CondMask))
+            writeMask &= ~WRITEMASK_X;
+      }
+      if (writeMask & WRITEMASK_Y) {
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
+                      dstReg->CondMask))
+            writeMask &= ~WRITEMASK_Y;
+      }
+      if (writeMask & WRITEMASK_Z) {
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
+                      dstReg->CondMask))
+            writeMask &= ~WRITEMASK_Z;
+      }
+      if (writeMask & WRITEMASK_W) {
+         if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
+                      dstReg->CondMask))
+            writeMask &= ~WRITEMASK_W;
+      }
+   }
+
+   if (writeMask & WRITEMASK_X)
+      dst[0] = value[0];
+   if (writeMask & WRITEMASK_Y)
+      dst[1] = value[1];
+   if (writeMask & WRITEMASK_Z)
+      dst[2] = value[2];
+   if (writeMask & WRITEMASK_W)
+      dst[3] = value[3];
+
+   if (inst->CondUpdate) {
+      if (writeMask & WRITEMASK_X)
+         machine->CondCodes[0] = generate_cc(value[0]);
+      if (writeMask & WRITEMASK_Y)
+         machine->CondCodes[1] = generate_cc(value[1]);
+      if (writeMask & WRITEMASK_Z)
+         machine->CondCodes[2] = generate_cc(value[2]);
+      if (writeMask & WRITEMASK_W)
+         machine->CondCodes[3] = generate_cc(value[3]);
+#if DEBUG_PROG
+      printf("CondCodes=(%s,%s,%s,%s) for:\n",
+             _mesa_condcode_string(machine->CondCodes[0]),
+             _mesa_condcode_string(machine->CondCodes[1]),
+             _mesa_condcode_string(machine->CondCodes[2]),
+             _mesa_condcode_string(machine->CondCodes[3]));
+#endif
+   }
+}
+
+
+
+/**
  * Execute the given vertex/fragment program.
  *
  * \param ctx  rendering context
@@ -571,6 +689,18 @@ _mesa_execute_program(GLcontext * ctx,
             }
          }
          break;
+      case OPCODE_AND:     /* bitwise AND */
+         {
+            GLuint a[4], b[4], result[4];
+            fetch_vector4ui(&inst->SrcReg[0], machine, a);
+            fetch_vector4ui(&inst->SrcReg[1], machine, b);
+            result[0] = a[0] & b[0];
+            result[1] = a[1] & b[1];
+            result[2] = a[2] & b[2];
+            result[3] = a[3] & b[3];
+            store_vector4ui(inst, machine, result);
+         }
+         break;
       case OPCODE_ARL:
          {
             GLfloat t[4];
@@ -649,6 +779,33 @@ _mesa_execute_program(GLcontext * ctx,
             store_vector4(inst, machine, result);
          }
          break;
+      case OPCODE_DP2:
+         {
+            GLfloat a[4], b[4], result[4];
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            fetch_vector4(&inst->SrcReg[1], machine, b);
+            result[0] = result[1] = result[2] = result[3] = DOT2(a, b);
+            store_vector4(inst, machine, result);
+            if (DEBUG_PROG) {
+               printf("DP2 %g = (%g %g) . (%g %g)\n",
+                      result[0], a[0], a[1], b[0], b[1]);
+            }
+         }
+         break;
+      case OPCODE_DP2A:
+         {
+            GLfloat a[4], b[4], c, result[4];
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            fetch_vector4(&inst->SrcReg[1], machine, b);
+            fetch_vector1(&inst->SrcReg[1], machine, &c);
+            result[0] = result[1] = result[2] = result[3] = DOT2(a, b) + c;
+            store_vector4(inst, machine, result);
+            if (DEBUG_PROG) {
+               printf("DP2A %g = (%g %g) . (%g %g) + %g\n",
+                      result[0], a[0], a[1], b[0], b[1], c);
+            }
+         }
+         break;
       case OPCODE_DP3:
          {
             GLfloat a[4], b[4], result[4];
@@ -681,8 +838,7 @@ _mesa_execute_program(GLcontext * ctx,
             GLfloat a[4], b[4], result[4];
             fetch_vector4(&inst->SrcReg[0], machine, a);
             fetch_vector4(&inst->SrcReg[1], machine, b);
-            result[0] = result[1] = result[2] = result[3] =
-               a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + b[3];
+            result[0] = result[1] = result[2] = result[3] = DOT3(a, b) + b[3];
             store_vector4(inst, machine, result);
          }
          break;
@@ -789,17 +945,6 @@ _mesa_execute_program(GLcontext * ctx,
       case OPCODE_ENDIF:
          /* nothing */
          break;
-      case OPCODE_INT:         /* float to int */
-         {
-            GLfloat a[4], result[4];
-            fetch_vector4(&inst->SrcReg[0], machine, a);
-            result[0] = (GLfloat) (GLint) a[0];
-            result[1] = (GLfloat) (GLint) a[1];
-            result[2] = (GLfloat) (GLint) a[2];
-            result[3] = (GLfloat) (GLint) a[3];
-            store_vector4(inst, machine, result);
-         }
-         break;
       case OPCODE_KIL_NV:      /* NV_f_p only (conditional) */
          if (eval_condition(machine, inst)) {
             return GL_FALSE;
@@ -1030,39 +1175,94 @@ _mesa_execute_program(GLcontext * ctx,
          break;
       case OPCODE_NOP:
          break;
-      case OPCODE_PK2H:        /* pack two 16-bit floats in one 32-bit float */
+      case OPCODE_NOT:         /* bitwise NOT */
+         {
+            GLuint a[4], result[4];
+            fetch_vector4ui(&inst->SrcReg[0], machine, a);
+            result[0] = ~a[0];
+            result[1] = ~a[1];
+            result[2] = ~a[2];
+            result[3] = ~a[3];
+            store_vector4ui(inst, machine, result);
+         }
+         break;
+      case OPCODE_NRM3:        /* 3-component normalization */
          {
             GLfloat a[4], result[4];
+            GLfloat tmp;
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2];
+            if (tmp != 0.0F)
+               tmp = INV_SQRTF(tmp);
+            result[0] = tmp * a[0];
+            result[1] = tmp * a[1];
+            result[2] = tmp * a[2];
+            result[3] = 0.0;  /* undefined, but prevent valgrind warnings */
+            store_vector4(inst, machine, result);
+         }
+         break;
+      case OPCODE_NRM4:        /* 4-component normalization */
+         {
+            GLfloat a[4], result[4];
+            GLfloat tmp;
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2] + a[3] * a[3];
+            if (tmp != 0.0F)
+               tmp = INV_SQRTF(tmp);
+            result[0] = tmp * a[0];
+            result[1] = tmp * a[1];
+            result[2] = tmp * a[2];
+            result[3] = tmp * a[3];
+            store_vector4(inst, machine, result);
+         }
+         break;
+      case OPCODE_OR:          /* bitwise OR */
+         {
+            GLuint a[4], b[4], result[4];
+            fetch_vector4ui(&inst->SrcReg[0], machine, a);
+            fetch_vector4ui(&inst->SrcReg[1], machine, b);
+            result[0] = a[0] | b[0];
+            result[1] = a[1] | b[1];
+            result[2] = a[2] | b[2];
+            result[3] = a[3] | b[3];
+            store_vector4ui(inst, machine, result);
+         }
+         break;
+      case OPCODE_PK2H:        /* pack two 16-bit floats in one 32-bit float */
+         {
+            GLfloat a[4];
+            GLuint result[4];
             GLhalfNV hx, hy;
-            GLuint *rawResult = (GLuint *) result;
-            GLuint twoHalves;
             fetch_vector4(&inst->SrcReg[0], machine, a);
             hx = _mesa_float_to_half(a[0]);
             hy = _mesa_float_to_half(a[1]);
-            twoHalves = hx | (hy << 16);
-            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
-               = twoHalves;
-            store_vector4(inst, machine, result);
+            result[0] =
+            result[1] =
+            result[2] =
+            result[3] = hx | (hy << 16);
+            store_vector4ui(inst, machine, result);
          }
          break;
       case OPCODE_PK2US:       /* pack two GLushorts into one 32-bit float */
          {
-            GLfloat a[4], result[4];
-            GLuint usx, usy, *rawResult = (GLuint *) result;
+            GLfloat a[4];
+            GLuint result[4], usx, usy;
             fetch_vector4(&inst->SrcReg[0], machine, a);
             a[0] = CLAMP(a[0], 0.0F, 1.0F);
             a[1] = CLAMP(a[1], 0.0F, 1.0F);
             usx = IROUND(a[0] * 65535.0F);
             usy = IROUND(a[1] * 65535.0F);
-            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
-               = usx | (usy << 16);
-            store_vector4(inst, machine, result);
+            result[0] =
+            result[1] =
+            result[2] =
+            result[3] = usx | (usy << 16);
+            store_vector4ui(inst, machine, result);
          }
          break;
       case OPCODE_PK4B:        /* pack four GLbytes into one 32-bit float */
          {
-            GLfloat a[4], result[4];
-            GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
+            GLfloat a[4];
+            GLuint result[4], ubx, uby, ubz, ubw;
             fetch_vector4(&inst->SrcReg[0], machine, a);
             a[0] = CLAMP(a[0], -128.0F / 127.0F, 1.0F);
             a[1] = CLAMP(a[1], -128.0F / 127.0F, 1.0F);
@@ -1072,15 +1272,17 @@ _mesa_execute_program(GLcontext * ctx,
             uby = IROUND(127.0F * a[1] + 128.0F);
             ubz = IROUND(127.0F * a[2] + 128.0F);
             ubw = IROUND(127.0F * a[3] + 128.0F);
-            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
-               = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
-            store_vector4(inst, machine, result);
+            result[0] =
+            result[1] =
+            result[2] =
+            result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
+            store_vector4ui(inst, machine, result);
          }
          break;
       case OPCODE_PK4UB:       /* pack four GLubytes into one 32-bit float */
          {
-            GLfloat a[4], result[4];
-            GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
+            GLfloat a[4];
+            GLuint result[4], ubx, uby, ubz, ubw;
             fetch_vector4(&inst->SrcReg[0], machine, a);
             a[0] = CLAMP(a[0], 0.0F, 1.0F);
             a[1] = CLAMP(a[1], 0.0F, 1.0F);
@@ -1090,9 +1292,11 @@ _mesa_execute_program(GLcontext * ctx,
             uby = IROUND(255.0F * a[1]);
             ubz = IROUND(255.0F * a[2]);
             ubw = IROUND(255.0F * a[3]);
-            rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
-               = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
-            store_vector4(inst, machine, result);
+            result[0] =
+            result[1] =
+            result[2] =
+            result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
+            store_vector4ui(inst, machine, result);
          }
          break;
       case OPCODE_POW:
@@ -1288,6 +1492,17 @@ _mesa_execute_program(GLcontext * ctx,
             }
          }
          break;
+      case OPCODE_SSG:         /* set sign (-1, 0 or +1) */
+         {
+            GLfloat a[4], result[4];
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            result[0] = (GLfloat) ((a[0] > 0.0F) - (a[0] < 0.0F));
+            result[1] = (GLfloat) ((a[1] > 0.0F) - (a[1] < 0.0F));
+            result[2] = (GLfloat) ((a[2] > 0.0F) - (a[2] < 0.0F));
+            result[3] = (GLfloat) ((a[3] > 0.0F) - (a[3] < 0.0F));
+            store_vector4(inst, machine, result);
+         }
+         break;
       case OPCODE_STR:         /* set true, operands ignored */
          {
             static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F };
@@ -1314,7 +1529,7 @@ _mesa_execute_program(GLcontext * ctx,
       case OPCODE_SWZ:         /* extended swizzle */
          {
             const struct prog_src_register *source = &inst->SrcReg[0];
-            const GLfloat *src = get_register_pointer(source, machine);
+            const GLfloat *src = get_src_register_pointer(source, machine);
             GLfloat result[4];
             GLuint i;
             for (i = 0; i < 4; i++) {
@@ -1425,6 +1640,17 @@ _mesa_execute_program(GLcontext * ctx,
             store_vector4(inst, machine, color);
          }
          break;
+      case OPCODE_TRUNC:       /* truncate toward zero */
+         {
+            GLfloat a[4], result[4];
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            result[0] = (GLfloat) (GLint) a[0];
+            result[1] = (GLfloat) (GLint) a[1];
+            result[2] = (GLfloat) (GLint) a[2];
+            result[3] = (GLfloat) (GLint) a[3];
+            store_vector4(inst, machine, result);
+         }
+         break;
       case OPCODE_UP2H:        /* unpack two 16-bit floats */
          {
             GLfloat a[4], result[4];
@@ -1475,6 +1701,18 @@ _mesa_execute_program(GLcontext * ctx,
             store_vector4(inst, machine, result);
          }
          break;
+      case OPCODE_XOR:         /* bitwise XOR */
+         {
+            GLuint a[4], b[4], result[4];
+            fetch_vector4ui(&inst->SrcReg[0], machine, a);
+            fetch_vector4ui(&inst->SrcReg[1], machine, b);
+            result[0] = a[0] ^ b[0];
+            result[1] = a[1] ^ b[1];
+            result[2] = a[2] ^ b[2];
+            result[3] = a[3] ^ b[3];
+            store_vector4ui(inst, machine, result);
+         }
+         break;
       case OPCODE_XPD:         /* cross product */
          {
             GLfloat a[4], b[4], result[4];
diff --git a/src/mesa/shader/prog_execute.h b/src/mesa/shader/prog_execute.h
index 18b13e11a41..8ceb7b092e8 100644
--- a/src/mesa/shader/prog_execute.h
+++ b/src/mesa/shader/prog_execute.h
@@ -25,6 +25,8 @@
 #ifndef PROG_EXECUTE_H
 #define PROG_EXECUTE_H
 
+#include "main/config.h"
+
 
 typedef void (*FetchTexelLodFunc)(GLcontext *ctx, const GLfloat texcoord[4],
                                   GLfloat lambda, GLuint unit, GLfloat color[4]);
@@ -36,10 +38,6 @@ typedef void (*FetchTexelDerivFunc)(GLcontext *ctx, const GLfloat texcoord[4],
                                     GLuint unit, GLfloat color[4]);
 
 
-/** The larger of VERT_RESULT_MAX, FRAG_RESULT_MAX */
-#define MAX_PROGRAM_OUTPUTS VERT_RESULT_MAX
-
-
 /**
  * Virtual machine state used during execution of vertex/fragment programs.
  */
diff --git a/src/mesa/shader/prog_instruction.c b/src/mesa/shader/prog_instruction.c
index 1033496d978..f5c0a498fb0 100644
--- a/src/mesa/shader/prog_instruction.c
+++ b/src/mesa/shader/prog_instruction.c
@@ -154,6 +154,7 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_NOP,    "NOP",     0, 0 },
    { OPCODE_ABS,    "ABS",     1, 1 },
    { OPCODE_ADD,    "ADD",     2, 1 },
+   { OPCODE_AND,    "AND",     2, 1 },
    { OPCODE_ARA,    "ARA",     1, 1 },
    { OPCODE_ARL,    "ARL",     1, 1 },
    { OPCODE_ARL_NV, "ARL",     1, 1 },
@@ -168,6 +169,8 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_COS,    "COS",     1, 1 },
    { OPCODE_DDX,    "DDX",     1, 1 },
    { OPCODE_DDY,    "DDY",     1, 1 },
+   { OPCODE_DP2,    "DP2",     2, 1 },
+   { OPCODE_DP2A,   "DP2A",    3, 1 },
    { OPCODE_DP3,    "DP3",     2, 1 },
    { OPCODE_DP4,    "DP4",     2, 1 },
    { OPCODE_DPH,    "DPH",     2, 1 },
@@ -182,7 +185,6 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_FLR,    "FLR",     1, 1 },
    { OPCODE_FRC,    "FRC",     1, 1 },
    { OPCODE_IF,     "IF",      1, 0 },
-   { OPCODE_INT,    "INT",     1, 1 },
    { OPCODE_KIL,    "KIL",     1, 0 },
    { OPCODE_KIL_NV, "KIL",     0, 0 },
    { OPCODE_LG2,    "LG2",     1, 1 },
@@ -198,6 +200,10 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_NOISE2, "NOISE2",  1, 1 },
    { OPCODE_NOISE3, "NOISE3",  1, 1 },
    { OPCODE_NOISE4, "NOISE4",  1, 1 },
+   { OPCODE_NOT,    "NOT",     1, 1 },
+   { OPCODE_NRM3,   "NRM3",    1, 1 },
+   { OPCODE_NRM4,   "NRM4",    1, 1 },
+   { OPCODE_OR,     "OR",      2, 1 },
    { OPCODE_PK2H,   "PK2H",    1, 1 },
    { OPCODE_PK2US,  "PK2US",   1, 1 },
    { OPCODE_PK4B,   "PK4B",    1, 1 },
@@ -230,11 +236,13 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_TXL,    "TXL",     1, 1 },
    { OPCODE_TXP,    "TXP",     1, 1 },
    { OPCODE_TXP_NV, "TXP",     1, 1 },
+   { OPCODE_TRUNC,  "TRUNC",   1, 1 },
    { OPCODE_UP2H,   "UP2H",    1, 1 },
    { OPCODE_UP2US,  "UP2US",   1, 1 },
    { OPCODE_UP4B,   "UP4B",    1, 1 },
    { OPCODE_UP4UB,  "UP4UB",   1, 1 },
    { OPCODE_X2D,    "X2D",     3, 1 },
+   { OPCODE_XOR,    "XOR",     2, 1 },
    { OPCODE_XPD,    "XPD",     2, 1 }
 };
 
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index 711166f9ddb..f978334ab22 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -145,6 +145,7 @@ typedef enum prog_opcode {
    OPCODE_NOP = 0,   /*                                      X   */
    OPCODE_ABS,       /*   X        X       1.1               X   */
    OPCODE_ADD,       /*   X        X       X       X         X   */
+   OPCODE_AND,       /*                                          */
    OPCODE_ARA,       /*                    2                     */
    OPCODE_ARL,       /*   X                X                     */
    OPCODE_ARL_NV,    /*                    2                     */
@@ -159,6 +160,8 @@ typedef enum prog_opcode {
    OPCODE_COS,       /*            X       2       X         X   */
    OPCODE_DDX,       /*                            X         X   */
    OPCODE_DDY,       /*                            X         X   */
+   OPCODE_DP2,       /*                            2             */
+   OPCODE_DP2A,      /*                            2             */
    OPCODE_DP3,       /*   X        X       X       X         X   */
    OPCODE_DP4,       /*   X        X       X       X         X   */
    OPCODE_DPH,       /*   X        X       1.1                   */
@@ -173,7 +176,6 @@ typedef enum prog_opcode {
    OPCODE_FLR,       /*   X        X       2       X         X   */
    OPCODE_FRC,       /*   X        X       2       X         X   */
    OPCODE_IF,        /*                                     opt  */
-   OPCODE_INT,       /*                                      X   */
    OPCODE_KIL,       /*            X                             */
    OPCODE_KIL_NV,    /*                            X         X   */
    OPCODE_LG2,       /*   X        X       2       X         X   */
@@ -189,6 +191,10 @@ typedef enum prog_opcode {
    OPCODE_NOISE2,    /*                                      X   */
    OPCODE_NOISE3,    /*                                      X   */
    OPCODE_NOISE4,    /*                                      X   */
+   OPCODE_NOT,       /*                                          */
+   OPCODE_NRM3,      /*                                          */
+   OPCODE_NRM4,      /*                                          */
+   OPCODE_OR,        /*                                          */
    OPCODE_PK2H,      /*                            X             */
    OPCODE_PK2US,     /*                            X             */
    OPCODE_PK4B,      /*                            X             */
@@ -221,11 +227,13 @@ typedef enum prog_opcode {
    OPCODE_TXL,       /*                    3       2         X   */
    OPCODE_TXP,       /*            X                         X   */
    OPCODE_TXP_NV,    /*                    3       X             */
+   OPCODE_TRUNC,     /*                                      X   */
    OPCODE_UP2H,      /*                            X             */
    OPCODE_UP2US,     /*                            X             */
    OPCODE_UP4B,      /*                            X             */
    OPCODE_UP4UB,     /*                            X             */
    OPCODE_X2D,       /*                            X             */
+   OPCODE_XOR,       /*                                          */
    OPCODE_XPD,       /*   X        X                         X   */
    MAX_OPCODE
 } gl_inst_opcode;
@@ -280,13 +288,10 @@ struct prog_src_register
  */
 struct prog_dst_register
 {
-   /**
-    * One of the PROGRAM_* register file values.
-    */
-   GLuint File:4;
-
+   GLuint File:4;      /**< One of the PROGRAM_* register file values */
    GLuint Index:8;
    GLuint WriteMask:4;
+   GLuint RelAddr:1;
 
    /**
     * \name Conditional destination update control.
@@ -318,7 +323,7 @@ struct prog_dst_register
    GLuint CondSrc:1;
    /*@}*/
 
-   GLuint pad:31;
+   GLuint pad:30;
 };
 
 
diff --git a/src/mesa/shader/prog_parameter.c b/src/mesa/shader/prog_parameter.c
index bfe27d2f638..6744ad93b81 100644
--- a/src/mesa/shader/prog_parameter.c
+++ b/src/mesa/shader/prog_parameter.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -80,7 +80,8 @@ GLint
 _mesa_add_parameter(struct gl_program_parameter_list *paramList,
                     enum register_file type, const char *name,
                     GLuint size, GLenum datatype, const GLfloat *values,
-                    const gl_state_index state[STATE_LENGTH])
+                    const gl_state_index state[STATE_LENGTH],
+                    GLbitfield flags)
 {
    const GLuint oldNum = paramList->NumParameters;
    const GLuint sz4 = (size + 3) / 4; /* no. of new param slots needed */
@@ -125,9 +126,11 @@ _mesa_add_parameter(struct gl_program_parameter_list *paramList,
          p->Type = type;
          p->Size = size;
          p->DataType = datatype;
+         p->Flags = flags;
          if (values) {
             COPY_4V(paramList->ParameterValues[oldNum + i], values);
             values += 4;
+            p->Initialized = GL_TRUE;
          }
          else {
             /* silence valgrind */
@@ -155,7 +158,7 @@ _mesa_add_named_parameter(struct gl_program_parameter_list *paramList,
                           const char *name, const GLfloat values[4])
 {
    return _mesa_add_parameter(paramList, PROGRAM_NAMED_PARAM, name,
-                              4, GL_NONE, values, NULL);
+                              4, GL_NONE, values, NULL, 0x0);
                               
 }
 
@@ -186,7 +189,7 @@ _mesa_add_named_constant(struct gl_program_parameter_list *paramList,
 #endif
    size = 4; /** XXX fix */
    return _mesa_add_parameter(paramList, PROGRAM_CONSTANT, name,
-                              size, GL_NONE, values, NULL);
+                              size, GL_NONE, values, NULL, 0x0);
 }
 
 
@@ -238,7 +241,7 @@ _mesa_add_unnamed_constant(struct gl_program_parameter_list *paramList,
 
    /* add a new parameter to store this constant */
    pos = _mesa_add_parameter(paramList, PROGRAM_CONSTANT, NULL,
-                             size, GL_NONE, values, NULL);
+                             size, GL_NONE, values, NULL, 0x0);
    if (pos >= 0 && swizzleOut) {
       if (size == 1)
          *swizzleOut = SWIZZLE_XXXX;
@@ -259,7 +262,8 @@ _mesa_add_unnamed_constant(struct gl_program_parameter_list *paramList,
  */
 GLint
 _mesa_add_uniform(struct gl_program_parameter_list *paramList,
-                  const char *name, GLuint size, GLenum datatype)
+                  const char *name, GLuint size, GLenum datatype,
+                  const GLfloat *values)
 {
    GLint i = _mesa_lookup_parameter_index(paramList, -1, name);
    ASSERT(datatype != GL_NONE);
@@ -271,7 +275,7 @@ _mesa_add_uniform(struct gl_program_parameter_list *paramList,
    }
    else {
       i = _mesa_add_parameter(paramList, PROGRAM_UNIFORM, name,
-                              size, datatype, NULL, NULL);
+                              size, datatype, values, NULL, 0x0);
       return i;
    }
 }
@@ -326,7 +330,7 @@ _mesa_add_sampler(struct gl_program_parameter_list *paramList,
       }
       value = (GLfloat) numSamplers;
       (void) _mesa_add_parameter(paramList, PROGRAM_SAMPLER, name,
-                                 size, datatype, &value, NULL);
+                                 size, datatype, &value, NULL, 0x0);
       return numSamplers;
    }
 }
@@ -337,7 +341,7 @@ _mesa_add_sampler(struct gl_program_parameter_list *paramList,
  */
 GLint
 _mesa_add_varying(struct gl_program_parameter_list *paramList,
-                  const char *name, GLuint size)
+                  const char *name, GLuint size, GLbitfield flags)
 {
    GLint i = _mesa_lookup_parameter_index(paramList, -1, name);
    if (i >= 0 && paramList->Parameters[i].Type == PROGRAM_VARYING) {
@@ -347,7 +351,7 @@ _mesa_add_varying(struct gl_program_parameter_list *paramList,
    else {
       /*assert(size == 4);*/
       i = _mesa_add_parameter(paramList, PROGRAM_VARYING, name,
-                              size, GL_NONE, NULL, NULL);
+                              size, GL_NONE, NULL, NULL, flags);
       return i;
    }
 }
@@ -376,7 +380,7 @@ _mesa_add_attribute(struct gl_program_parameter_list *paramList,
       if (size < 0)
          size = 4;
       i = _mesa_add_parameter(paramList, PROGRAM_INPUT, name,
-                              size, datatype, NULL, state);
+                              size, datatype, NULL, state, 0x0);
    }
    return i;
 }
@@ -443,7 +447,7 @@ _mesa_add_state_reference(struct gl_program_parameter_list *paramList,
    name = _mesa_program_state_string(stateTokens);
    index = _mesa_add_parameter(paramList, PROGRAM_STATE_VAR, name,
                                size, GL_NONE,
-                               NULL, (gl_state_index *) stateTokens);
+                               NULL, (gl_state_index *) stateTokens, 0x0);
    paramList->StateFlags |= _mesa_program_state_flags(stateTokens);
 
    /* free name string here since we duplicated it in add_parameter() */
@@ -614,10 +618,11 @@ _mesa_clone_parameter_list(const struct gl_program_parameter_list *list)
       struct gl_program_parameter *pCopy;
       GLuint size = MIN2(p->Size, 4);
       GLint j = _mesa_add_parameter(clone, p->Type, p->Name, size, p->DataType,
-                                    list->ParameterValues[i], NULL);
+                                    list->ParameterValues[i], NULL, 0x0);
       ASSERT(j >= 0);
       pCopy = clone->Parameters + j;
       pCopy->Used = p->Used;
+      pCopy->Flags = p->Flags;
       /* copy state indexes */
       if (p->Type == PROGRAM_STATE_VAR) {
          GLint k;
@@ -655,7 +660,8 @@ _mesa_combine_parameter_lists(const struct gl_program_parameter_list *listA,
             _mesa_add_parameter(list, param->Type, param->Name, param->Size,
                                 param->DataType,
                                 listB->ParameterValues[i],
-                                param->StateIndexes);
+                                param->StateIndexes,
+                                param->Flags);
          }
       }
    }
diff --git a/src/mesa/shader/prog_parameter.h b/src/mesa/shader/prog_parameter.h
index e8d3e9ec336..200f2c00458 100644
--- a/src/mesa/shader/prog_parameter.h
+++ b/src/mesa/shader/prog_parameter.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -36,11 +36,20 @@
 
 
 /**
+ * Program parameter flags
+ */
+/*@{*/
+#define PROG_PARAM_BIT_CENTROID   0x1  /**< for varying vars (GLSL 1.20) */
+#define PROG_PARAM_BIT_INVARIANT  0x2  /**< for varying vars (GLSL 1.20) */
+#define PROG_PARAM_BIT_FLAT       0x4  /**< for varying vars (GLSL 1.30) */
+#define PROG_PARAM_BIT_LINEAR     0x8  /**< for varying vars (GLSL 1.30) */
+/*@}*/
+
+
+
+/**
  * Program parameter.
- * Used for NV_fragment_program for "DEFINE"d constants and "DECLARE"d
- * parameters.
- * Also used by ARB_vertex/fragment_programs for state variables, etc.
- * Used by shaders for uniforms, constants, varying vars, etc.
+ * Used by shaders/programs for uniforms, constants, varying vars, etc.
  */
 struct gl_program_parameter
 {
@@ -49,6 +58,8 @@ struct gl_program_parameter
    GLenum DataType;         /**< GL_FLOAT, GL_FLOAT_VEC2, etc */
    GLuint Size;             /**< Number of components (1..4) */
    GLboolean Used;          /**< Helper flag for GLSL uniform tracking */
+   GLboolean Initialized;   /**< Has the ParameterValue[] been set? */
+   GLbitfield Flags;        /**< Bitmask of PROG_PARAM_*_BIT */
    /**
     * A sequence of STATE_* tokens and integers to identify GL state.
     */
@@ -93,7 +104,8 @@ extern GLint
 _mesa_add_parameter(struct gl_program_parameter_list *paramList,
                     enum register_file type, const char *name,
                     GLuint size, GLenum datatype, const GLfloat *values,
-                    const gl_state_index state[STATE_LENGTH]);
+                    const gl_state_index state[STATE_LENGTH],
+                    GLbitfield flags);
 
 extern GLint
 _mesa_add_named_parameter(struct gl_program_parameter_list *paramList,
@@ -111,7 +123,8 @@ _mesa_add_unnamed_constant(struct gl_program_parameter_list *paramList,
 
 extern GLint
 _mesa_add_uniform(struct gl_program_parameter_list *paramList,
-                  const char *name, GLuint size, GLenum datatype);
+                  const char *name, GLuint size, GLenum datatype,
+                  const GLfloat *values);
 
 extern void
 _mesa_use_uniform(struct gl_program_parameter_list *paramList,
@@ -123,7 +136,7 @@ _mesa_add_sampler(struct gl_program_parameter_list *paramList,
 
 extern GLint
 _mesa_add_varying(struct gl_program_parameter_list *paramList,
-                  const char *name, GLuint size);
+                  const char *name, GLuint size, GLbitfield flags);
 
 extern GLint
 _mesa_add_attribute(struct gl_program_parameter_list *paramList,
diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c
index ec260f18a9f..29b4d90a3e8 100644
--- a/src/mesa/shader/prog_print.c
+++ b/src/mesa/shader/prog_print.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -215,7 +215,7 @@ reg_string(enum register_file f, GLint index, gl_prog_print_mode mode,
    switch (mode) {
    case PROG_PRINT_DEBUG:
       if (relAddr)
-         sprintf(str, "%s[ADDR%s%d]", file_string(f, mode), (index > 0) ? "+" : "", index);
+         sprintf(str, "%s[ADDR+%d]", file_string(f, mode), index);
       else
          sprintf(str, "%s[%d]", file_string(f, mode), index);
       break;
@@ -371,8 +371,8 @@ _mesa_print_swizzle(GLuint swizzle)
 }
 
 
-static const char *
-writemask_string(GLuint writeMask)
+const char *
+_mesa_writemask_string(GLuint writeMask)
 {
    static char s[10];
    GLuint i = 0;
@@ -419,8 +419,8 @@ print_dst_reg(const struct prog_dst_register *dstReg, gl_prog_print_mode mode,
 {
    _mesa_printf("%s%s",
                 reg_string((enum register_file) dstReg->File,
-                           dstReg->Index, mode, GL_FALSE, prog),
-                writemask_string(dstReg->WriteMask));
+                           dstReg->Index, mode, dstReg->RelAddr, prog),
+                _mesa_writemask_string(dstReg->WriteMask));
 
    if (dstReg->CondMask != COND_TR) {
       _mesa_printf(" (%s.%s)",
@@ -432,7 +432,7 @@ print_dst_reg(const struct prog_dst_register *dstReg, gl_prog_print_mode mode,
    _mesa_printf("%s[%d]%s",
                 file_string((enum register_file) dstReg->File, mode),
                 dstReg->Index,
-                writemask_string(dstReg->WriteMask));
+                _mesa_writemask_string(dstReg->WriteMask));
 #endif
 }
 
@@ -814,9 +814,18 @@ _mesa_print_parameter_list(const struct gl_program_parameter_list *list)
    for (i = 0; i < list->NumParameters; i++){
       struct gl_program_parameter *param = list->Parameters + i;
       const GLfloat *v = list->ParameterValues[i];
-      _mesa_printf("param[%d] sz=%d %s %s = {%.3g, %.3g, %.3g, %.3g};\n",
+      _mesa_printf("param[%d] sz=%d %s %s = {%.3g, %.3g, %.3g, %.3g}",
                    i, param->Size,
                    file_string(list->Parameters[i].Type, mode),
                    param->Name, v[0], v[1], v[2], v[3]);
+      if (param->Flags & PROG_PARAM_BIT_CENTROID)
+         _mesa_printf(" Centroid");
+      if (param->Flags & PROG_PARAM_BIT_INVARIANT)
+         _mesa_printf(" Invariant");
+      if (param->Flags & PROG_PARAM_BIT_FLAT)
+         _mesa_printf(" Flat");
+      if (param->Flags & PROG_PARAM_BIT_LINEAR)
+         _mesa_printf(" Linear");
+      _mesa_printf("\n");
    }
 }
diff --git a/src/mesa/shader/prog_print.h b/src/mesa/shader/prog_print.h
index 3cdb1b195e0..3966909aed0 100644
--- a/src/mesa/shader/prog_print.h
+++ b/src/mesa/shader/prog_print.h
@@ -43,6 +43,9 @@ _mesa_condcode_string(GLuint condcode);
 extern const char *
 _mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended);
 
+const char *
+_mesa_writemask_string(GLuint writeMask);
+
 extern void
 _mesa_print_swizzle(GLuint swizzle);
 
diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c
index 9cc33fa2c11..8d29acac8ba 100644
--- a/src/mesa/shader/prog_statevars.c
+++ b/src/mesa/shader/prog_statevars.c
@@ -131,7 +131,7 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
 	       ADD_3V(value, p, eye_z);
 	       NORMALIZE_3FV(value);
 	       value[3] = 1.0;
-            }						  
+            }
             return;
          default:
             _mesa_problem(ctx, "Invalid light state in fetch_state");
@@ -235,11 +235,11 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
          }
       }
    case STATE_TEXENV_COLOR:
-      {		
+      {
          /* state[1] is the texture unit */
          const GLuint unit = (GLuint) state[1];
          COPY_4V(value, ctx->Texture.Unit[unit].EnvColor);
-      }			
+      }
       return;
    case STATE_FOG_COLOR:
       COPY_4V(value, ctx->Fog.Color);
@@ -369,7 +369,7 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
          }
       }
       return;
-		
+
    case STATE_VERTEX_PROGRAM:
       {
          /* state[1] = {STATE_ENV, STATE_LOCAL} */
@@ -475,7 +475,6 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
       }						  
 
 
-
       case STATE_PT_SCALE:
          value[0] = ctx->Pixel.RedScale;
          value[1] = ctx->Pixel.GreenScale;
@@ -494,6 +493,19 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
       case STATE_PCM_BIAS:
          COPY_4V(value, ctx->Pixel.PostColorMatrixBias);
          break;
+      case STATE_SHADOW_AMBIENT:
+         {
+            const int unit = (int) state[2];
+            const struct gl_texture_object *texObj
+               = ctx->Texture.Unit[unit]._Current;
+            if (texObj) {
+               value[0] = texObj->ShadowAmbient;
+               value[1] = texObj->ShadowAmbient;
+               value[2] = texObj->ShadowAmbient;
+               value[3] = texObj->ShadowAmbient;
+            }
+         }
+         return;
 
          /* XXX: make sure new tokens added here are also handled in the 
           * _mesa_program_state_flags() switch, below.
@@ -578,6 +590,7 @@ _mesa_program_state_flags(const gl_state_index state[STATE_LENGTH])
          return _NEW_MODELVIEW;
 
       case STATE_TEXRECT_SCALE:
+      case STATE_SHADOW_AMBIENT:
 	 return _NEW_TEXTURE;
       case STATE_FOG_PARAMS_OPTIMIZED:
 	 return _NEW_FOG;
@@ -618,6 +631,9 @@ append(char *dst, const char *src)
 }
 
 
+/**
+ * Convert token 'k' to a string, append it onto 'dst' string.
+ */
 static void
 append_token(char *dst, gl_state_index k)
 {
@@ -750,11 +766,30 @@ append_token(char *dst, gl_state_index k)
    case STATE_LOCAL:
       append(dst, "local");
       break;
+   /* BEGIN internal state vars */
+   case STATE_INTERNAL:
+      append(dst, "(internal)");
+      break;
    case STATE_NORMAL_SCALE:
       append(dst, "normalScale");
       break;
-   case STATE_INTERNAL:
-      append(dst, "(internal)");
+   case STATE_TEXRECT_SCALE:
+      append(dst, "texrectScale");
+      break;
+   case STATE_FOG_PARAMS_OPTIMIZED:
+      append(dst, "fogParamsOptimized");
+      break;
+   case STATE_LIGHT_SPOT_DIR_NORMALIZED:
+      append(dst, "lightSpotDirNormalized");
+      break;
+   case STATE_LIGHT_POSITION:
+      append(dst, "lightPosition");
+      break;
+   case STATE_LIGHT_POSITION_NORMALIZED:
+      append(dst, "light.position.normalized");
+      break;
+   case STATE_LIGHT_HALF_VECTOR:
+      append(dst, "lightHalfVector");
       break;
    case STATE_PT_SCALE:
       append(dst, "PTscale");
@@ -768,8 +803,12 @@ append_token(char *dst, gl_state_index k)
    case STATE_PCM_BIAS:
       append(dst, "PCMbias");
       break;
+   case STATE_SHADOW_AMBIENT:
+      append(dst, "ShadowAmbient");
+      break;
    default:
-      ;
+      /* probably STATE_INTERNAL_DRIVER+i (driver private state) */
+      append(dst, "driverState");
    }
 }
 
@@ -802,16 +841,16 @@ _mesa_program_state_string(const gl_state_index state[STATE_LENGTH])
    char tmp[30];
 
    append(str, "state.");
-   append_token(str, (gl_state_index) state[0]);
+   append_token(str, state[0]);
 
    switch (state[0]) {
    case STATE_MATERIAL:
       append_face(str, state[1]);
-      append_token(str, (gl_state_index) state[2]);
+      append_token(str, state[2]);
       break;
    case STATE_LIGHT:
       append_index(str, state[1]); /* light number [i]. */
-      append_token(str, (gl_state_index) state[2]); /* coefficients */
+      append_token(str, state[2]); /* coefficients */
       break;
    case STATE_LIGHTMODEL_AMBIENT:
       append(str, "lightmodel.ambient");
@@ -827,11 +866,11 @@ _mesa_program_state_string(const gl_state_index state[STATE_LENGTH])
    case STATE_LIGHTPROD:
       append_index(str, state[1]); /* light number [i]. */
       append_face(str, state[2]);
-      append_token(str, (gl_state_index) state[3]);
+      append_token(str, state[3]);
       break;
    case STATE_TEXGEN:
       append_index(str, state[1]); /* tex unit [i] */
-      append_token(str, (gl_state_index) state[2]); /* plane coef */
+      append_token(str, state[2]); /* plane coef */
       break;
    case STATE_TEXENV_COLOR:
       append_index(str, state[1]); /* tex unit [i] */
@@ -853,11 +892,11 @@ _mesa_program_state_string(const gl_state_index state[STATE_LENGTH])
          /* state[2] = first row to fetch */
          /* state[3] = last row to fetch */
          /* state[4] = transpose, inverse or invtrans */
-         const gl_state_index mat = (gl_state_index) state[0];
+         const gl_state_index mat = state[0];
          const GLuint index = (GLuint) state[1];
          const GLuint firstRow = (GLuint) state[2];
          const GLuint lastRow = (GLuint) state[3];
-         const gl_state_index modifier = (gl_state_index) state[4];
+         const gl_state_index modifier = state[4];
          if (index ||
              mat == STATE_TEXTURE_MATRIX ||
              mat == STATE_PROGRAM_MATRIX)
@@ -885,10 +924,11 @@ _mesa_program_state_string(const gl_state_index state[STATE_LENGTH])
    case STATE_VERTEX_PROGRAM:
       /* state[1] = {STATE_ENV, STATE_LOCAL} */
       /* state[2] = parameter index          */
-      append_token(str, (gl_state_index) state[1]);
+      append_token(str, state[1]);
       append_index(str, state[2]);
       break;
    case STATE_INTERNAL:
+      append_token(str, state[1]);
       break;
    default:
       _mesa_problem(NULL, "Invalid state in _mesa_program_state_string");
@@ -918,10 +958,101 @@ _mesa_load_state_parameters(GLcontext *ctx,
 
    for (i = 0; i < paramList->NumParameters; i++) {
       if (paramList->Parameters[i].Type == PROGRAM_STATE_VAR) {
-         _mesa_fetch_state(ctx, 
+         _mesa_fetch_state(ctx,
 			   (gl_state_index *) paramList->Parameters[i].StateIndexes,
                            paramList->ParameterValues[i]);
       }
    }
 }
 
+
+/**
+ * Copy the 16 elements of a matrix into four consecutive program
+ * registers starting at 'pos'.
+ */
+static void
+load_matrix(GLfloat registers[][4], GLuint pos, const GLfloat mat[16])
+{
+   GLuint i;
+   for (i = 0; i < 4; i++) {
+      registers[pos + i][0] = mat[0 + i];
+      registers[pos + i][1] = mat[4 + i];
+      registers[pos + i][2] = mat[8 + i];
+      registers[pos + i][3] = mat[12 + i];
+   }
+}
+
+
+/**
+ * As above, but transpose the matrix.
+ */
+static void
+load_transpose_matrix(GLfloat registers[][4], GLuint pos,
+                      const GLfloat mat[16])
+{
+   MEMCPY(registers[pos], mat, 16 * sizeof(GLfloat));
+}
+
+
+/**
+ * Load current vertex program's parameter registers with tracked
+ * matrices (if NV program).  This only needs to be done per
+ * glBegin/glEnd, not per-vertex.
+ */
+void
+_mesa_load_tracked_matrices(GLcontext *ctx)
+{
+   GLuint i;
+
+   for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS / 4; i++) {
+      /* point 'mat' at source matrix */
+      GLmatrix *mat;
+      if (ctx->VertexProgram.TrackMatrix[i] == GL_MODELVIEW) {
+         mat = ctx->ModelviewMatrixStack.Top;
+      }
+      else if (ctx->VertexProgram.TrackMatrix[i] == GL_PROJECTION) {
+         mat = ctx->ProjectionMatrixStack.Top;
+      }
+      else if (ctx->VertexProgram.TrackMatrix[i] == GL_TEXTURE) {
+         mat = ctx->TextureMatrixStack[ctx->Texture.CurrentUnit].Top;
+      }
+      else if (ctx->VertexProgram.TrackMatrix[i] == GL_COLOR) {
+         mat = ctx->ColorMatrixStack.Top;
+      }
+      else if (ctx->VertexProgram.TrackMatrix[i]==GL_MODELVIEW_PROJECTION_NV) {
+         /* XXX verify the combined matrix is up to date */
+         mat = &ctx->_ModelProjectMatrix;
+      }
+      else if (ctx->VertexProgram.TrackMatrix[i] >= GL_MATRIX0_NV &&
+               ctx->VertexProgram.TrackMatrix[i] <= GL_MATRIX7_NV) {
+         GLuint n = ctx->VertexProgram.TrackMatrix[i] - GL_MATRIX0_NV;
+         ASSERT(n < MAX_PROGRAM_MATRICES);
+         mat = ctx->ProgramMatrixStack[n].Top;
+      }
+      else {
+         /* no matrix is tracked, but we leave the register values as-is */
+         assert(ctx->VertexProgram.TrackMatrix[i] == GL_NONE);
+         continue;
+      }
+
+      /* load the matrix values into sequential registers */
+      if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_IDENTITY_NV) {
+         load_matrix(ctx->VertexProgram.Parameters, i*4, mat->m);
+      }
+      else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_INVERSE_NV) {
+         _math_matrix_analyse(mat); /* update the inverse */
+         ASSERT(!_math_matrix_is_dirty(mat));
+         load_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv);
+      }
+      else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_TRANSPOSE_NV) {
+         load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->m);
+      }
+      else {
+         assert(ctx->VertexProgram.TrackMatrixTransform[i]
+                == GL_INVERSE_TRANSPOSE_NV);
+         _math_matrix_analyse(mat); /* update the inverse */
+         ASSERT(!_math_matrix_is_dirty(mat));
+         load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv);
+      }
+   }
+}
diff --git a/src/mesa/shader/prog_statevars.h b/src/mesa/shader/prog_statevars.h
index 1f728c64e80..72e51f40314 100644
--- a/src/mesa/shader/prog_statevars.h
+++ b/src/mesa/shader/prog_statevars.h
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -77,7 +77,7 @@ typedef enum gl_state_index_ {
    STATE_SPECULAR,
    STATE_EMISSION,
    STATE_SHININESS,
-   STATE_HALF_VECTOR,	
+   STATE_HALF_VECTOR,
 
    STATE_POSITION,
    STATE_ATTENUATION,
@@ -94,7 +94,7 @@ typedef enum gl_state_index_ {
    STATE_TEXGEN_OBJECT_Q,
 
    STATE_TEXENV_COLOR,
-	
+
    STATE_DEPTH_RANGE,
 
    STATE_VERTEX_PROGRAM,
@@ -116,6 +116,7 @@ typedef enum gl_state_index_ {
    STATE_PT_BIAS,               /**< Pixel transfer RGBA bias */
    STATE_PCM_SCALE,             /**< Post color matrix RGBA scale */
    STATE_PCM_BIAS,              /**< Post color matrix RGBA bias */
+   STATE_SHADOW_AMBIENT,        /**< ARB_shadow_ambient fail value; token[2] is texture unit index */
    STATE_INTERNAL_DRIVER	/* first available state index for drivers (must be last) */
 } gl_state_index;
 
@@ -134,4 +135,8 @@ extern char *
 _mesa_program_state_string(const gl_state_index state[STATE_LENGTH]);
 
 
+extern void
+_mesa_load_tracked_matrices(GLcontext *ctx);
+
+
 #endif /* PROG_STATEVARS_H */
diff --git a/src/mesa/shader/prog_uniform.c b/src/mesa/shader/prog_uniform.c
index a0aa615c5f6..0642713148c 100644
--- a/src/mesa/shader/prog_uniform.c
+++ b/src/mesa/shader/prog_uniform.c
@@ -52,11 +52,12 @@ _mesa_free_uniform_list(struct gl_uniform_list *list)
 }
 
 
-GLboolean
+struct gl_uniform *
 _mesa_append_uniform(struct gl_uniform_list *list,
                      const char *name, GLenum target, GLuint progPos)
 {
    const GLuint oldNum = list->NumUniforms;
+   struct gl_uniform *uniform;
    GLint index;
 
    assert(target == GL_VERTEX_PROGRAM_ARB ||
@@ -84,31 +85,37 @@ _mesa_append_uniform(struct gl_uniform_list *list,
          return GL_FALSE;
       }
 
-      list->Uniforms[oldNum].Name = _mesa_strdup(name);
-      list->Uniforms[oldNum].VertPos = -1;
-      list->Uniforms[oldNum].FragPos = -1;
-      list->Uniforms[oldNum].Initialized = GL_FALSE;
-      index = oldNum;
+      uniform = list->Uniforms + oldNum;
+
+      uniform->Name = _mesa_strdup(name);
+      uniform->VertPos = -1;
+      uniform->FragPos = -1;
+      uniform->Initialized = GL_FALSE;
+
       list->NumUniforms++;
    }
+   else {
+      /* found */
+      uniform = list->Uniforms + index;
+   }
 
    /* update position for the vertex or fragment program */
    if (target == GL_VERTEX_PROGRAM_ARB) {
-      if (list->Uniforms[index].VertPos != -1) {
+      if (uniform->VertPos != -1) {
          /* this uniform is already in the list - that shouldn't happen */
          return GL_FALSE;
       }
-      list->Uniforms[index].VertPos = progPos;
+      uniform->VertPos = progPos;
    }
    else {
-      if (list->Uniforms[index].FragPos != -1) {
+      if (uniform->FragPos != -1) {
          /* this uniform is already in the list - that shouldn't happen */
          return GL_FALSE;
       }
-      list->Uniforms[index].FragPos = progPos;
+      uniform->FragPos = progPos;
    }
 
-   return GL_TRUE;
+   return uniform;
 }
 
 
@@ -135,8 +142,8 @@ _mesa_longest_uniform_name(const struct gl_uniform_list *list)
    GLint max = 0;
    GLuint i;
    for (i = 0; list && i < list->NumUniforms; i++) {
-      GLuint len = _mesa_strlen(list->Uniforms[i].Name);
-      if (len > (GLuint)max)
+      GLint len = (GLint)_mesa_strlen(list->Uniforms[i].Name);
+      if (len > max)
          max = len;
    }
    return max;
diff --git a/src/mesa/shader/prog_uniform.h b/src/mesa/shader/prog_uniform.h
index deea7329912..22a2bfd9701 100644
--- a/src/mesa/shader/prog_uniform.h
+++ b/src/mesa/shader/prog_uniform.h
@@ -75,7 +75,7 @@ _mesa_new_uniform_list(void);
 extern void
 _mesa_free_uniform_list(struct gl_uniform_list *list);
 
-extern GLboolean
+extern struct gl_uniform *
 _mesa_append_uniform(struct gl_uniform_list *list,
                      const char *name, GLenum target, GLuint progPos);
 
diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c
index 723c46ee8cf..37962f0e9bb 100644
--- a/src/mesa/shader/program.c
+++ b/src/mesa/shader/program.c
@@ -121,6 +121,43 @@ _mesa_free_program_data(GLcontext *ctx)
 
 
 /**
+ * Update the default program objects in the given context to reference those
+ * specified in the shared state and release those referencing the old
+ * shared state.
+ */
+void
+_mesa_update_default_objects_program(GLcontext *ctx)
+{
+#if FEATURE_NV_vertex_program || FEATURE_ARB_vertex_program
+   _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current,
+                            (struct gl_vertex_program *)
+                            ctx->Shared->DefaultVertexProgram);
+   assert(ctx->VertexProgram.Current);
+#endif
+
+#if FEATURE_NV_fragment_program || FEATURE_ARB_fragment_program
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current,
+                            (struct gl_fragment_program *)
+                            ctx->Shared->DefaultFragmentProgram);
+   assert(ctx->FragmentProgram.Current);
+#endif
+
+   /* XXX probably move this stuff */
+#if FEATURE_ATI_fragment_shader
+   if (ctx->ATIFragmentShader.Current) {
+      ctx->ATIFragmentShader.Current->RefCount--;
+      if (ctx->ATIFragmentShader.Current->RefCount <= 0) {
+         _mesa_free(ctx->ATIFragmentShader.Current);
+      }
+   }
+   ctx->ATIFragmentShader.Current = (struct ati_fragment_shader *) ctx->Shared->DefaultFragmentShader;
+   assert(ctx->ATIFragmentShader.Current);
+   ctx->ATIFragmentShader.Current->RefCount++;
+#endif
+}
+
+
+/**
  * Set the vertex/fragment program error state (position and error string).
  * This is generally called from within the parsers.
  */
@@ -211,7 +248,7 @@ struct gl_program *
 _mesa_init_fragment_program( GLcontext *ctx, struct gl_fragment_program *prog,
                              GLenum target, GLuint id)
 {
-   if (prog) 
+   if (prog)
       return _mesa_init_program_struct( ctx, &prog->Base, target, id );
    else
       return NULL;
@@ -225,7 +262,7 @@ struct gl_program *
 _mesa_init_vertex_program( GLcontext *ctx, struct gl_vertex_program *prog,
                            GLenum target, GLuint id)
 {
-   if (prog) 
+   if (prog)
       return _mesa_init_program_struct( ctx, &prog->Base, target, id );
    else
       return NULL;
@@ -238,7 +275,7 @@ _mesa_init_vertex_program( GLcontext *ctx, struct gl_vertex_program *prog,
  * ctx->Driver.NewProgram.  May be overridden (ie. replaced) by a
  * device driver function to implement OO deriviation with additional
  * types not understood by this function.
- * 
+ *
  * \param ctx  context
  * \param id   program id/number
  * \param target  program target/type
@@ -247,19 +284,23 @@ _mesa_init_vertex_program( GLcontext *ctx, struct gl_vertex_program *prog,
 struct gl_program *
 _mesa_new_program(GLcontext *ctx, GLenum target, GLuint id)
 {
+   struct gl_program *prog;
    switch (target) {
    case GL_VERTEX_PROGRAM_ARB: /* == GL_VERTEX_PROGRAM_NV */
-      return _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
+      prog = _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
                                        target, id );
+      break;
    case GL_FRAGMENT_PROGRAM_NV:
    case GL_FRAGMENT_PROGRAM_ARB:
-      return _mesa_init_fragment_program(ctx,
+      prog =_mesa_init_fragment_program(ctx,
                                          CALLOC_STRUCT(gl_fragment_program),
                                          target, id );
+      break;
    default:
       _mesa_problem(ctx, "bad target in _mesa_new_program");
-      return NULL;
+      prog = NULL;
    }
+   return prog;
 }
 
 
@@ -278,7 +319,7 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog)
 
    if (prog == &_mesa_DummyProgram)
       return;
-                 
+
    if (prog->String)
       _mesa_free(prog->String);
 
@@ -331,7 +372,11 @@ _mesa_reference_program(GLcontext *ctx,
    assert(ptr);
    if (*ptr && prog) {
       /* sanity check */
-      ASSERT((*ptr)->Target == prog->Target);
+      if ((*ptr)->Target == GL_VERTEX_PROGRAM_ARB)
+         ASSERT(prog->Target == GL_VERTEX_PROGRAM_ARB);
+      else if ((*ptr)->Target == GL_FRAGMENT_PROGRAM_ARB)
+         ASSERT(prog->Target == GL_FRAGMENT_PROGRAM_ARB ||
+                prog->Target == GL_FRAGMENT_PROGRAM_NV);
    }
    if (*ptr == prog) {
       return;  /* no change */
@@ -341,15 +386,17 @@ _mesa_reference_program(GLcontext *ctx,
 
       /*_glthread_LOCK_MUTEX((*ptr)->Mutex);*/
 #if 0
-      printf("Program %p %u 0x%x  Refcount-- to %d\n",
-             *ptr, (*ptr)->Id, (*ptr)->Target, (*ptr)->RefCount - 1);
+      printf("Program %p ID=%u Target=%s  Refcount-- to %d\n",
+             *ptr, (*ptr)->Id,
+             ((*ptr)->Target == GL_VERTEX_PROGRAM_ARB ? "VP" : "FP"),
+             (*ptr)->RefCount - 1);
 #endif
       ASSERT((*ptr)->RefCount > 0);
       (*ptr)->RefCount--;
 
       deleteFlag = ((*ptr)->RefCount == 0);
       /*_glthread_UNLOCK_MUTEX((*ptr)->Mutex);*/
-      
+
       if (deleteFlag) {
          ASSERT(ctx);
          ctx->Driver.DeleteProgram(ctx, *ptr);
@@ -363,8 +410,10 @@ _mesa_reference_program(GLcontext *ctx,
       /*_glthread_LOCK_MUTEX(prog->Mutex);*/
       prog->RefCount++;
 #if 0
-      printf("Program %p %u 0x%x  Refcount++ to %d\n",
-             prog, prog->Id, prog->Target, prog->RefCount);
+      printf("Program %p ID=%u Target=%s  Refcount++ to %d\n",
+             prog, prog->Id,
+             (prog->Target == GL_VERTEX_PROGRAM_ARB ? "VP" : "FP"),
+             prog->RefCount);
 #endif
       /*_glthread_UNLOCK_MUTEX(prog->Mutex);*/
    }
@@ -402,6 +451,7 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog)
    clone->InputsRead = prog->InputsRead;
    clone->OutputsWritten = prog->OutputsWritten;
    clone->SamplersUsed = prog->SamplersUsed;
+   clone->ShadowSamplers = prog->ShadowSamplers;
    memcpy(clone->TexturesUsed, prog->TexturesUsed, sizeof(prog->TexturesUsed));
 
    if (prog->Parameters)
@@ -504,6 +554,52 @@ _mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count)
    return GL_TRUE;
 }
 
+/**
+ * Delete 'count' instructions at 'start' in the given program.
+ * Adjust branch targets accordingly.
+ */
+GLboolean
+_mesa_delete_instructions(struct gl_program *prog, GLuint start, GLuint count)
+{
+   const GLuint origLen = prog->NumInstructions;
+   const GLuint newLen = origLen - count;
+   struct prog_instruction *newInst;
+   GLuint i;
+
+   /* adjust branches */
+   for (i = 0; i < prog->NumInstructions; i++) {
+      struct prog_instruction *inst = prog->Instructions + i;
+      if (inst->BranchTarget > 0) {
+         if (inst->BranchTarget >= start) {
+            inst->BranchTarget -= count;
+         }
+      }
+   }
+
+   /* Alloc storage for new instructions */
+   newInst = _mesa_alloc_instructions(newLen);
+   if (!newInst) {
+      return GL_FALSE;
+   }
+
+   /* Copy 'start' instructions into new instruction buffer */
+   _mesa_copy_instructions(newInst, prog->Instructions, start);
+
+   /* Copy the remaining/tail instructions to new inst buffer */
+   _mesa_copy_instructions(newInst + start,
+                           prog->Instructions + start + count,
+                           newLen - start);
+
+   /* free old instructions */
+   _mesa_free_instructions(prog->Instructions, origLen);
+
+   /* install new instructions */
+   prog->Instructions = newInst;
+   prog->NumInstructions = newLen;
+
+   return GL_TRUE;
+}
+
 
 /**
  * Search instructions for registers that match (oldFile, oldIndex),
diff --git a/src/mesa/shader/program.h b/src/mesa/shader/program.h
index f8bd63233ed..48176162c35 100644
--- a/src/mesa/shader/program.h
+++ b/src/mesa/shader/program.h
@@ -53,6 +53,9 @@ extern void
 _mesa_free_program_data(GLcontext *ctx);
 
 extern void
+_mesa_update_default_objects_program(GLcontext *ctx);
+
+extern void
 _mesa_set_program_error(GLcontext *ctx, GLint pos, const char *string);
 
 extern const GLubyte *
@@ -60,13 +63,13 @@ _mesa_find_line_column(const GLubyte *string, const GLubyte *pos,
                        GLint *line, GLint *col);
 
 
-extern struct gl_program * 
-_mesa_init_vertex_program(GLcontext *ctx, 
-                          struct gl_vertex_program *prog, 
+extern struct gl_program *
+_mesa_init_vertex_program(GLcontext *ctx,
+                          struct gl_vertex_program *prog,
                           GLenum target, GLuint id);
 
-extern struct gl_program * 
-_mesa_init_fragment_program(GLcontext *ctx, 
+extern struct gl_program *
+_mesa_init_fragment_program(GLcontext *ctx,
                             struct gl_fragment_program *prog,
                             GLenum target, GLuint id);
 
@@ -108,6 +111,9 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog);
 extern  GLboolean
 _mesa_insert_instructions(struct gl_program *prog, GLuint start, GLuint count);
 
+extern  GLboolean
+_mesa_delete_instructions(struct gl_program *prog, GLuint start, GLuint count);
+
 extern struct gl_program *
 _mesa_combine_programs(GLcontext *ctx,
                        const struct gl_program *progA,
diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c
index 7af502a84cb..122688826cd 100644
--- a/src/mesa/shader/shader_api.c
+++ b/src/mesa/shader/shader_api.c
@@ -39,15 +39,15 @@
 #include "main/context.h"
 #include "main/hash.h"
 #include "main/macros.h"
-#include "program.h"
-#include "prog_parameter.h"
-#include "prog_print.h"
-#include "prog_statevars.h"
-#include "prog_uniform.h"
+#include "shader/program.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
+#include "shader/prog_statevars.h"
+#include "shader/prog_uniform.h"
 #include "shader/shader_api.h"
 #include "shader/slang/slang_compile.h"
 #include "shader/slang/slang_link.h"
-
+#include "glapi/dispatch.h"
 
 
 #ifndef GL_PROGRAM_BINARY_LENGTH_OES
@@ -455,7 +455,13 @@ _mesa_attach_shader(GLcontext *ctx, GLuint program, GLuint shader)
    n = shProg->NumShaders;
    for (i = 0; i < n; i++) {
       if (shProg->Shaders[i] == sh) {
-         /* already attached */
+         /* The shader is already attched to this program.  The
+          * GL_ARB_shader_objects spec says:
+          *
+          *     "The error INVALID_OPERATION is generated by AttachObjectARB
+          *     if <obj> is already attached to <containerObj>."
+          */
+         _mesa_error(ctx, GL_INVALID_OPERATION, "glAttachShader");
          return;
       }
    }
@@ -517,7 +523,7 @@ _mesa_bind_attrib_location(GLcontext *ctx, GLuint program, GLuint index,
 {
    struct gl_shader_program *shProg;
    const GLint size = -1; /* unknown size */
-   GLint i;
+   GLint i, oldIndex;
    GLenum datatype = GL_FLOAT_VEC4;
 
    shProg = _mesa_lookup_shader_program_err(ctx, program,
@@ -540,6 +546,14 @@ _mesa_bind_attrib_location(GLcontext *ctx, GLuint program, GLuint index,
       return;
    }
 
+   if (shProg->LinkStatus) {
+      /* get current index/location for the attribute */
+      oldIndex = _mesa_get_attrib_location(ctx, program, name);
+   }
+   else {
+      oldIndex = -1;
+   }
+
    /* this will replace the current value if it's already in the list */
    i = _mesa_add_attribute(shProg->Attributes, name, size, datatype, index);
    if (i < 0) {
@@ -911,24 +925,15 @@ _mesa_get_attached_shaders(GLcontext *ctx, GLuint program, GLsizei maxCount,
 static GLuint
 _mesa_get_handle(GLcontext *ctx, GLenum pname)
 {
-#if 0
-   GET_CURRENT_CONTEXT(ctx);
-
-   switch (pname) {
-   case GL_PROGRAM_OBJECT_ARB:
-      {
-         struct gl2_program_intf **pro = ctx->Shader.CurrentProgram;
-
-         if (pro != NULL)
-            return (**pro)._container._generic.
-               GetName((struct gl2_generic_intf **) (pro));
-      }
-      break;
-   default:
+   GLint handle = 0;
+   
+   if (pname == GL_PROGRAM_OBJECT_ARB) {
+      CALL_GetIntegerv(ctx->Exec, (GL_CURRENT_PROGRAM, &handle));
+   } else {
       _mesa_error(ctx, GL_INVALID_ENUM, "glGetHandleARB");
    }
-#endif
-   return 0;
+
+   return handle;
 }
 
 
@@ -1610,6 +1615,7 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count,
               const GLvoid *values, GLenum type)
 {
    struct gl_shader_program *shProg = ctx->Shader.CurrentProgram;
+   struct gl_uniform *uniform;
    GLint elems, offset;
 
    if (!shProg || !shProg->LinkStatus) {
@@ -1656,12 +1662,14 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count,
 
    FLUSH_VERTICES(ctx, _NEW_PROGRAM);
 
+   uniform = &shProg->Uniforms->Uniforms[location];
+
    /* A uniform var may be used by both a vertex shader and a fragment
     * shader.  We may need to update one or both shader's uniform here:
     */
    if (shProg->VertexProgram) {
       /* convert uniform location to program parameter index */
-      GLint index = shProg->Uniforms->Uniforms[location].VertPos;
+      GLint index = uniform->VertPos;
       if (index >= 0) {
          set_program_uniform(ctx, &shProg->VertexProgram->Base,
                              index, offset, type, count, elems, values);
@@ -1670,14 +1678,14 @@ _mesa_uniform(GLcontext *ctx, GLint location, GLsizei count,
 
    if (shProg->FragmentProgram) {
       /* convert uniform location to program parameter index */
-      GLint index = shProg->Uniforms->Uniforms[location].FragPos;
+      GLint index = uniform->FragPos;
       if (index >= 0) {
          set_program_uniform(ctx, &shProg->FragmentProgram->Base,
                              index, offset, type, count, elems, values);
       }
    }
 
-   shProg->Uniforms->Uniforms[location].Initialized = GL_TRUE;
+   uniform->Initialized = GL_TRUE;
 }
 
 
@@ -1744,8 +1752,9 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows,
                      GLenum matrixType, GLint location, GLsizei count,
                      GLboolean transpose, const GLfloat *values)
 {
-   GLint offset;
    struct gl_shader_program *shProg = ctx->Shader.CurrentProgram;
+   struct gl_uniform *uniform;
+   GLint offset;
 
    if (!shProg || !shProg->LinkStatus) {
       _mesa_error(ctx, GL_INVALID_OPERATION,
@@ -1769,9 +1778,11 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows,
 
    FLUSH_VERTICES(ctx, _NEW_PROGRAM);
 
+   uniform = &shProg->Uniforms->Uniforms[location];
+
    if (shProg->VertexProgram) {
       /* convert uniform location to program parameter index */
-      GLint index = shProg->Uniforms->Uniforms[location].VertPos;
+      GLint index = uniform->VertPos;
       if (index >= 0) {
          set_program_uniform_matrix(ctx, &shProg->VertexProgram->Base,
                                     index, offset,
@@ -1781,7 +1792,7 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows,
 
    if (shProg->FragmentProgram) {
       /* convert uniform location to program parameter index */
-      GLint index = shProg->Uniforms->Uniforms[location].FragPos;
+      GLint index = uniform->FragPos;
       if (index >= 0) {
          set_program_uniform_matrix(ctx, &shProg->FragmentProgram->Base,
                                     index, offset,
@@ -1789,7 +1800,7 @@ _mesa_uniform_matrix(GLcontext *ctx, GLint cols, GLint rows,
       }
    }
 
-   shProg->Uniforms->Uniforms[location].Initialized = GL_TRUE;
+   uniform->Initialized = GL_TRUE;
 }
 
 
diff --git a/src/mesa/shader/slang/descrip.mms b/src/mesa/shader/slang/descrip.mms
new file mode 100644
index 00000000000..6eefbcf5bdd
--- /dev/null
+++ b/src/mesa/shader/slang/descrip.mms
@@ -0,0 +1,68 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [----.include.gl]
+	define math [--.math]
+	define swrast [--.swrast]
+	define array_cache [--.array_cache]
+	define main [--.main]
+	define glapi [--.glapi]
+	define shader [--.shader]
+
+.include [----]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [----.include],[--.main],[--.glapi],[-.slang],[-.grammar],[-]
+LIBDIR = [----.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = \
+	slang_compile.c,slang_preprocess.c
+
+OBJECTS = slang_builtin.obj,slang_codegen.obj,slang_compile.obj,\
+	slang_compile_function.obj,slang_compile_operation.obj,\
+	slang_compile_struct.obj,slang_compile_variable.obj,slang_emit.obj,\
+	slang_ir.obj,slang_label.obj,slang_library_noise.obj,slang_link.obj,\
+	slang_log.obj,slang_mem.obj,slang_preprocess.obj,slang_print.obj,\
+	slang_simplify.obj,slang_storage.obj,slang_typeinfo.obj,\
+	slang_utility.obj,slang_vartable.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+slang_builtin.obj : slang_builtin.c
+slang_codegen.obj : slang_codegen.c
+slang_compile.obj : slang_compile.c
+slang_compile_function.obj : slang_compile_function.c
+slang_compile_operation.obj : slang_compile_operation.c
+slang_compile_struct.obj : slang_compile_struct.c
+slang_compile_variable.obj : slang_compile_variable.c
+slang_emit.obj : slang_emit.c
+slang_ir.obj : slang_ir.c
+slang_label.obj : slang_label.c
+slang_library_noise.obj : slang_library_noise.c
+slang_link.obj : slang_link.c
+slang_log.obj : slang_log.c
+slang_mem.obj : slang_mem.c
+slang_preprocess.obj : slang_preprocess.c
+slang_print.obj : slang_print.c
+slang_simplify.obj : slang_simplify.c
+slang_storage.obj : slang_storage.c
+slang_typeinfo.obj : slang_typeinfo.c
+slang_utility.obj : slang_utility.c
+slang_vartable.obj : slang_vartable.c
diff --git a/src/mesa/shader/slang/library/slang_120_core_gc.h b/src/mesa/shader/slang/library/slang_120_core_gc.h
index 15e3a3f5775..ad2ef34909a 100644
--- a/src/mesa/shader/slang/library/slang_120_core_gc.h
+++ b/src/mesa/shader/slang/library/slang_120_core_gc.h
@@ -2,725 +2,756 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_120_core.gc */
 
-4,1,0,0,26,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,
-48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,
-18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,48,0,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,
-0,57,59,122,0,18,102,50,49,0,20,0,0,1,0,0,26,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,58,109,97,116,50,120,51,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,
-102,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,
-111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,102,
-0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,
-0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,102,0,0,0,20,0,0,1,0,0,26,1,
-1,1,0,0,11,99,48,0,0,1,1,0,0,11,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
-99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,0,0,28,1,1,1,0,0,
-9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,51,48,0,0,1,1,0,0,9,
-102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,1,1,0,0,9,102,51,49,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,8,48,0,57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,
-122,0,18,102,50,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,119,0,18,102,51,48,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,49,0,57,59,122,0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,
-0,18,102,51,49,0,20,0,0,1,0,0,28,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
+4,1,90,95,0,0,26,221,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,
+0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,48,
+0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,49,0,57,59,122,0,18,102,50,49,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,9,102,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,
+0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,5,105,0,0,0,1,
+3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,
+97,108,0,58,109,97,116,50,120,51,0,18,102,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,1,98,0,0,0,1,3,
+2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,97,
+108,0,58,109,97,116,50,120,51,0,18,102,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,11,99,48,0,0,1,1,
+0,0,11,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,9,102,48,48,0,
+0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,51,48,0,0,1,1,0,0,9,102,48,49,0,0,
+1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,1,1,0,0,9,102,51,49,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,
+48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,119,0,18,102,51,48,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,
+59,122,0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,0,18,102,51,
+49,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
 97,116,50,120,52,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,
-18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,
-102,0,2,58,102,108,111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,
-50,120,52,0,18,102,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,
-97,116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,102,0,0,0,
-20,0,0,1,0,0,28,1,1,1,0,0,12,99,48,0,0,1,1,0,0,12,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,
-0,0,27,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,
-49,0,0,1,1,0,0,9,102,48,50,0,0,1,1,0,0,9,102,49,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
-48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,
-102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,50,0,57,59,120,0,18,102,48,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,
-0,57,59,121,0,18,102,49,50,0,20,0,0,1,0,0,27,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,58,109,97,116,51,120,50,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,
-0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,
-111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,102,
-0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,
-0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,102,0,0,0,20,0,0,1,0,0,27,1,
-1,1,0,0,10,99,48,0,0,1,1,0,0,10,99,49,0,0,1,1,0,0,10,99,50,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,
-0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,0,1,0,0,30,1,1,1,0,0,9,102,48,
-48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,51,48,0,0,1,1,0,0,9,102,48,49,
-0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,1,1,0,0,9,102,51,49,0,0,1,1,0,0,9,102,48,50,0,
-0,1,1,0,0,9,102,49,50,0,0,1,1,0,0,9,102,50,50,0,0,1,1,0,0,9,102,51,50,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,
-48,0,57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,
-102,50,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,119,0,18,102,51,48,0,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,59,122,0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,0,18,102,
-51,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,18,102,48,50,0,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,102,49,50,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,50,0,57,59,122,0,18,102,50,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,
-59,119,0,18,102,51,50,0,20,0,0,1,0,0,30,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,51,120,52,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,
-0,0,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,
-48,0,48,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,
-0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,102,0,0,0,20,0,
-0,1,0,0,30,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,
-95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,102,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,12,
-99,48,0,0,1,1,0,0,12,99,49,0,0,1,1,0,0,12,99,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,0,1,0,0,29,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,
-9,102,49,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,48,50,0,0,1,1,0,0,9,
-102,49,50,0,0,1,1,0,0,9,102,48,51,0,0,1,1,0,0,9,102,49,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,
-0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,
-0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,50,0,57,59,120,0,18,102,48,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,59,121,0,18,102,49,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,120,0,
-18,102,48,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,121,0,18,102,49,51,0,20,0,
-0,1,0,0,29,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,
-102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,16,10,52,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,
-48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,
-111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,102,
-0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,
-0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,102,0,0,0,20,0,0,1,0,0,29,1,
-1,1,0,0,10,99,48,0,0,1,1,0,0,10,99,49,0,0,1,1,0,0,10,99,50,0,0,1,1,0,0,10,99,51,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,51,0,57,18,99,51,0,20,0,0,1,0,0,31,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,
-102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,
-102,50,49,0,0,1,1,0,0,9,102,48,50,0,0,1,1,0,0,9,102,49,50,0,0,1,1,0,0,9,102,50,50,0,0,1,1,0,0,9,
-102,48,51,0,0,1,1,0,0,9,102,49,51,0,0,1,1,0,0,9,102,50,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,
-0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,48,0,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,49,0,57,59,122,0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,
+18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,5,105,0,0,0,1,3,
+2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,
+108,0,58,109,97,116,50,120,52,0,18,102,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,1,98,0,0,0,1,3,2,
+90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,
+0,58,109,97,116,50,120,52,0,18,102,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,12,99,48,0,0,1,1,0,0,
+12,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,9,102,48,48,0,0,1,
+1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,48,50,0,0,1,1,
+0,0,9,102,49,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,
+0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,
 18,102,48,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,102,49,50,0,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,122,0,18,102,50,50,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,51,0,57,59,120,0,18,102,48,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,51,0,57,59,121,0,18,102,49,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,122,0,
-18,102,50,51,0,20,0,0,1,0,0,31,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
-97,116,52,120,51,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,
-0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,
-48,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,5,105,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,
-105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,102,0,0,0,20,0,0,1,0,
-0,31,1,1,1,0,0,1,98,0,0,0,1,3,2,1,0,9,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,102,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,11,99,48,
-0,0,1,1,0,0,11,99,49,0,0,1,1,0,0,11,99,50,0,0,1,1,0,0,11,99,51,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,51,0,57,18,99,51,0,20,0,0,1,0,0,13,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,18,109,0,20,0,0,1,0,0,13,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
-109,97,116,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,13,1,1,1,0,0,29,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,0,18,109,0,
-16,10,49,0,57,0,0,20,0,0,1,0,0,13,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
-109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,
-0,1,0,0,13,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,
-16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,0,0,13,1,1,1,0,0,14,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,
-0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,0,0,13,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,
-0,57,59,120,121,0,0,0,20,0,0,1,0,0,13,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,
-0,0,1,0,0,13,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,
-0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,26,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,26,1,1,1,0,0,14,109,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,
-49,0,57,0,0,20,0,0,1,0,0,26,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
-116,50,120,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,26,1,1,1,0,0,28,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,
-120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,30,109,0,0,0,
-1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,121,
-122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,15,109,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,
-18,109,0,16,10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,
-48,0,57,59,121,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,
-121,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
+0,1,90,95,0,0,27,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,
+120,50,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,
+0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,5,105,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,
+97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,102,0,0,
+0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,1,98,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,
+116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,102,0,0,0,20,
+0,0,1,90,95,0,0,27,221,1,1,1,0,0,10,99,48,0,0,1,1,0,0,10,99,49,0,0,1,1,0,0,10,99,50,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,49,0,57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,0,1,90,
+95,0,0,30,221,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,
+102,51,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,1,1,0,0,9,
+102,51,49,0,0,1,1,0,0,9,102,48,50,0,0,1,1,0,0,9,102,49,50,0,0,1,1,0,0,9,102,50,50,0,0,1,1,0,0,9,
+102,51,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,59,119,0,18,102,51,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,
+48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,49,0,57,59,119,0,18,102,51,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,
+59,120,0,18,102,48,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,102,49,
+50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,122,0,18,102,50,50,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,50,0,57,59,119,0,18,102,51,50,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,
+9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,102,0,0,17,48,0,48,0,
+0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,
+0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,
+0,0,5,105,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,105,0,0,0,0,0,9,18,95,
+95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,102,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,
+0,0,1,98,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,95,95,
+114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,102,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,
+12,99,48,0,0,1,1,0,0,12,99,49,0,0,1,1,0,0,12,99,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,9,102,48,
+48,0,0,1,1,0,0,9,102,49,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,48,50,
+0,0,1,1,0,0,9,102,49,50,0,0,1,1,0,0,9,102,48,51,0,0,1,1,0,0,9,102,49,51,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,59,121,0,18,102,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,
+102,48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,18,102,48,50,0,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,50,0,57,59,121,0,18,102,49,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,
+0,57,59,120,0,18,102,48,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,121,0,18,102,
+49,51,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
+109,97,116,52,120,50,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,16,10,52,0,0,17,48,0,48,0,0,0,
+17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,5,105,0,0,
+0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,
+86,97,108,0,58,109,97,116,52,120,50,0,18,102,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,1,98,0,0,0,
+1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,
+97,108,0,58,109,97,116,52,120,50,0,18,102,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,10,99,48,0,0,1,
+1,0,0,10,99,49,0,0,1,1,0,0,10,99,50,0,0,1,1,0,0,10,99,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,51,0,57,18,99,51,0,20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,9,102,48,48,0,0,1,1,0,0,9,102,49,48,0,
+0,1,1,0,0,9,102,50,48,0,0,1,1,0,0,9,102,48,49,0,0,1,1,0,0,9,102,49,49,0,0,1,1,0,0,9,102,50,49,0,0,
+1,1,0,0,9,102,48,50,0,0,1,1,0,0,9,102,49,50,0,0,1,1,0,0,9,102,50,50,0,0,1,1,0,0,9,102,48,51,0,0,1,
+1,0,0,9,102,49,51,0,0,1,1,0,0,9,102,50,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+59,120,0,18,102,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,102,49,48,
+0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,102,50,48,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,102,48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,59,121,0,18,102,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,
+0,18,102,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,18,102,48,50,0,20,
+0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,102,49,50,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,50,0,57,59,122,0,18,102,50,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,51,0,57,59,120,0,18,102,48,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,121,0,
+18,102,49,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,122,0,18,102,50,51,0,20,0,
+0,1,90,95,0,0,31,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,
+120,51,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,
+17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,102,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,
+20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,5,105,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,
+116,0,18,105,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,102,0,0,0,
+20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,1,98,0,0,0,1,3,2,90,95,1,0,9,221,1,102,0,2,58,102,108,111,97,
+116,0,18,98,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,102,0,0,0,20,
+0,0,1,90,95,0,0,31,221,1,1,1,0,0,11,99,48,0,0,1,1,0,0,11,99,49,0,0,1,1,0,0,11,99,50,0,0,1,1,0,0,11,
+99,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
+99,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,99,51,0,20,0,0,1,90,95,0,0,13,221,
+1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,13,221,1,1,
+1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,0,
+18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,
+95,0,0,13,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,
+109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,13,
+221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,
+48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,
+14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,
+121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,30,109,0,0,0,1,
+9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,
+0,16,10,49,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,
+59,120,121,0,0,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,58,109,97,116,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,
+0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,
+20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
+116,50,120,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,
+0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,
+57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,
+10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,
+10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,
+10,49,0,57,59,120,121,122,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,
+57,59,121,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,
+17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
 108,0,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,
 17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,
-0,0,0,20,0,0,1,0,0,26,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,
-120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,17,48,0,48,0,0,0,18,109,
-0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,
-1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,28,1,1,1,0,0,30,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,0,
-18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,28,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,28,
-1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,
-8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,17,48,0,48,
-0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,49,0,57,
-59,122,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,
-0,18,109,0,16,8,48,0,57,59,122,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,
-10,49,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,122,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,
-0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,
-57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,17,48,0,48,0,0,0,
-18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,122,
-0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,
-0,48,0,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,
-48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,
+0,0,0,20,0,0,1,90,95,0,0,26,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
+97,116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,17,48,0,48,0,
+0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,0,20,0,0,
+1,90,95,0,0,28,221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,
+90,95,0,0,28,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,
+52,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,15,109,
+0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,0,18,
+109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,
 86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,
-121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,
-59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,28,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,
-48,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,
-10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,27,109,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,27,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,
-18,109,0,16,10,50,0,57,0,0,20,0,0,1,0,0,27,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,18,109,0,16,10,50,
-0,57,0,0,20,0,0,1,0,0,27,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,
-51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,
-57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,
-50,0,57,59,121,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,
-0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,
-109,0,16,10,50,0,57,59,121,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,
-121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,
-0,48,0,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
-116,51,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,58,118,101,99,50,0,17,48,0,48,0,0,
-0,0,0,0,20,0,0,1,0,0,27,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,
-51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,
-57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,
-27,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,
-16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,
-0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,0,0,14,1,1,1,0,0,14,109,0,
-0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,14,1,1,1,0,0,31,109,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,18,
-109,0,16,10,50,0,57,0,0,20,0,0,1,0,0,14,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,
-122,0,0,18,109,0,16,10,50,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,14,1,1,1,0,0,15,109,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,
-16,10,49,0,57,59,120,121,122,0,0,18,109,0,16,10,50,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,14,1,1,1,
-0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,
-18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,14,1,1,1,0,0,28,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,121,
-122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,
-0,1,0,0,14,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,
+121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,
+0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,122,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,
+28,221,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,
+109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,17,
+48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,
+49,0,57,59,122,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,31,109,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,
+8,48,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,
+120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,122,0,0,17,48,0,48,0,0,0,0,20,
+0,0,1,90,95,0,0,28,221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,
+50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,
+48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,
+17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,
+17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,
+0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,28,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,
+8,48,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,
+16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,
+27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,29,
+109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,0,
+18,109,0,16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,14,109,0,0,
+0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,
+16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,30,109,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,
+16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,
+0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,121,0,0,0,20,0,0,1,90,95,0,0,27,221,1,1,1,0,0,
+31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,
+59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,49,0,
+57,59,121,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,121,0,0,0,20,0,0,1,90,95,
+0,0,27,221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,
+18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,
+0,18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,27,221,
+1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,
+8,48,0,57,0,18,109,0,16,10,49,0,57,0,58,118,101,99,50,0,17,48,0,48,0,0,0,0,0,0,20,0,0,1,90,95,0,0,
+27,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,
+109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,
+18,109,0,16,10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,27,221,1,
+1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,50,0,18,109,0,16,8,
+48,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,
+10,49,0,57,59,121,0,0,17,48,0,48,0,0,0,17,48,0,48,0,0,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,14,
+109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,31,109,
+0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,
+10,49,0,57,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,
+10,49,0,57,59,120,121,122,0,0,18,109,0,16,10,50,0,57,59,120,121,122,0,0,0,20,0,0,1,90,95,0,0,14,
+221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,
+48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,18,109,0,16,10,50,0,57,59,120,
+121,122,0,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,
+17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,
+121,122,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,27,109,
+0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,
+0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,
+0,14,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,0,18,109,0,
 16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,
-0,0,0,0,0,20,0,0,1,0,0,14,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
-116,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,
-10,50,0,57,0,17,49,0,0,0,0,0,20,0,0,1,0,0,14,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,58,109,97,116,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,
-0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,30,109,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,30,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,18,109,
-0,16,10,50,0,57,0,0,20,0,0,1,0,0,30,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
-109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,
-0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,
-49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,28,109,
-0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,18,
-109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,30,1,
-1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,
-48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,
-0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,
-57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,
-0,30,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,
-0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,
-18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,30,1,1,1,0,0,13,109,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,
-17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,
-49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,18,109,0,20,0,0,1,0,0,29,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
+0,0,0,0,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
+109,97,116,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,
+0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,
+18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,
+0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,
+31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,
+0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,0,20,
+0,0,1,90,95,0,0,30,221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,
+51,120,52,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,
+0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,
+17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,
+1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,
+8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,
+109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,29,109,0,0,
+0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,
+0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,
+49,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,30,221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,
+109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,
+0,0,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
+109,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
+97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,18,
+109,0,16,10,50,0,57,59,120,121,0,0,18,109,0,16,10,51,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,29,
+221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,
+16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,18,109,0,16,10,50,0,57,59,120,
+121,0,0,18,109,0,16,10,51,0,57,59,120,121,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,27,109,0,0,0,1,
+9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,
+10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,14,109,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,
+109,0,16,10,49,0,57,59,120,121,0,0,18,109,0,16,10,50,0,57,59,120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,
+0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
 97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,18,
-109,0,16,10,50,0,57,59,120,121,0,0,18,109,0,16,10,51,0,57,59,120,121,0,0,0,20,0,0,1,0,0,29,1,1,1,0,
-0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,
-57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,18,109,0,16,10,50,0,57,59,120,121,0,0,18,
-109,0,16,10,51,0,57,59,120,121,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,
-0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,
-0,18,109,0,16,10,50,0,57,59,120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,30,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,
-120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,18,109,0,16,10,50,0,57,59,120,121,0,0,17,48,0,0,
-0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
-109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,
-0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,
-120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,29,1,1,1,0,0,28,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,
-120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,
-0,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,
-0,0,1,0,0,31,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,
-18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,18,109,0,16,10,
-50,0,57,59,120,121,122,0,0,18,109,0,16,10,51,0,57,59,120,121,122,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,
-14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,
-0,18,109,0,16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,
-0,0,1,0,0,31,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,
-18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,18,109,0,16,10,
-50,0,57,59,120,121,122,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,
-29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,
-0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,18,
-109,0,16,10,51,0,57,0,17,48,0,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,
-0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,31,1,1,
-1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,
-0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,
-17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,
-0,16,10,49,0,57,59,120,121,122,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,
-0,0,0,17,48,0,0,0,0,0,20,0,0,1,0,0,31,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,
-0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,
-0,0,15,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,15,1,1,1,
-0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,
-18,109,0,16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,
-0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
-116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,
-10,50,0,57,0,17,48,0,0,0,0,18,109,0,16,10,51,0,57,0,17,49,0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,28,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,18,109,0,
-16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,
-0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,
-0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,18,109,0,
-16,10,51,0,57,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,
-0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,
-0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
-109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,
-0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,
-0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
-116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,
-0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,
-17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,15,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,
-0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,
-0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,0,0,0,2,1,1,0,2,0,26,109,0,0,1,1,0,0,26,
-110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,
-16,10,49,0,57,21,0,0,1,0,0,0,2,1,1,0,2,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,109,0,16,8,48,0,57,
-18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,0,0,0,2,1,1,0,
-2,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,
-0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,0,
-1,0,0,0,2,1,1,0,2,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,
-57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,
-10,50,0,57,21,0,0,1,0,0,0,2,1,1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
-110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,
-57,18,110,0,16,10,50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,0,0,0,2,1,
-1,0,2,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,
-109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,
-0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,0,0,0,2,2,1,0,2,0,26,109,0,0,1,1,0,0,26,
-110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,
-16,10,49,0,57,22,0,0,1,0,0,0,2,2,1,0,2,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,109,0,16,8,48,0,57,
-18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,0,1,0,0,0,2,2,1,0,
-2,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,
-0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,0,
-1,0,0,0,2,2,1,0,2,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,
-57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,
-10,50,0,57,22,0,0,1,0,0,0,2,2,1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
-110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,
-57,18,110,0,16,10,50,0,57,22,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,0,0,0,2,2,
-1,0,2,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,
+109,0,16,10,50,0,57,59,120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,
+0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,
+57,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,
+90,95,0,0,29,221,1,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,
+50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,109,0,16,10,49,0,57,59,120,121,0,0,17,48,0,0,0,0,17,
+48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,29,221,1,1,1,0,0,28,109,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,121,0,0,18,
+109,0,16,10,49,0,57,59,120,121,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,
+0,1,90,95,0,0,31,221,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,
+90,95,0,0,31,221,1,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,
+51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,18,109,0,
+16,10,50,0,57,59,120,121,122,0,0,18,109,0,16,10,51,0,57,59,120,121,122,0,0,0,20,0,0,1,90,95,0,0,31,
+221,1,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,
+16,8,48,0,57,0,18,109,0,16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,
+0,0,0,0,0,20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
+109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,
+122,0,0,18,109,0,16,10,50,0,57,59,120,121,122,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,
+0,1,90,95,0,0,31,221,1,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,
+120,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,
+10,50,0,57,0,17,49,0,0,0,0,18,109,0,16,10,51,0,57,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,31,221,1,1,
+1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,
+0,57,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,
+0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,
+17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,
+20,0,0,1,90,95,0,0,31,221,1,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,
+116,52,120,51,0,18,109,0,16,8,48,0,57,59,120,121,122,0,0,18,109,0,16,10,49,0,57,59,120,121,122,0,0,
+17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,
+95,0,0,31,221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,120,51,
+0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,
+0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,1,
+0,0,15,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,
+30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,18,
+109,0,16,10,49,0,57,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,
+0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,
+97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,
+16,10,50,0,57,0,17,48,0,0,0,0,18,109,0,16,10,51,0,57,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,
+1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,
+0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,
+17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,29,109,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,48,0,
+0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,0,0,
+17,48,0,0,0,0,18,109,0,16,10,51,0,57,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,
+1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,
+17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,48,0,0,0,0,17,48,
+0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,26,109,0,0,
+0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,18,
+109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,
+0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,27,109,0,0,
+0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,48,0,57,0,17,48,0,0,0,0,17,
+48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,50,0,57,0,17,49,0,0,
+0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,0,20,0,0,1,90,95,0,0,15,
+221,1,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,109,97,116,52,0,18,109,0,16,8,
+48,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,18,109,0,16,10,49,0,57,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,
+0,0,0,17,48,0,0,0,0,17,49,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,48,0,0,0,0,17,49,0,
+0,0,0,0,20,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,9,18,109,0,16,8,48,
+0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,90,95,0,0,
+0,221,2,1,1,0,2,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
+21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,27,109,0,
+0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,
+57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,0,1,90,95,0,0,
+0,221,2,1,1,0,2,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
+21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,
+50,0,57,21,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,109,0,16,8,48,
+0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,
+10,50,0,57,18,110,0,16,10,50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,90,
+95,0,0,0,221,2,1,1,0,2,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,
+48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,
+0,16,10,50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,90,95,0,0,0,221,2,2,
+1,0,2,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,
+109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,28,109,0,0,1,1,0,0,
+28,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,
+0,16,10,49,0,57,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,
+16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,
+109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,30,109,0,0,1,1,0,0,
+30,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,
+0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,0,1,90,95,0,0,0,221,2,2,
+1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,
 109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,
-0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,0,0,0,2,4,1,0,2,0,26,109,0,0,1,1,0,0,26,
-110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,
-16,10,49,0,57,24,0,0,1,0,0,0,2,4,1,0,2,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,109,0,16,8,48,0,57,
-18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,0,1,0,0,0,2,4,1,0,
-2,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,
-0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,0,0,
-1,0,0,0,2,4,1,0,2,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,
-57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,
-10,50,0,57,24,0,0,1,0,0,0,2,4,1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
-110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,
-57,18,110,0,16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,0,0,1,0,0,0,2,4,
-1,0,2,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,
-109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,
-0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,0,0,1,0,0,11,2,21,1,1,0,0,26,109,0,0,1,1,0,0,
-10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,
-59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,
-97,108,0,59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,
-16,10,49,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,118,0,59,120,0,18,
-109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,48,46,20,0,0,1,0,
-0,12,2,21,1,1,0,0,28,109,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,
-118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,
-48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,
-121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,
+0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,31,109,0,0,
+1,1,0,0,31,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,
+18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,9,18,109,0,16,10,
+51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,26,109,0,0,1,1,0,0,26,110,0,0,
+0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,
+0,57,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,109,0,16,8,48,0,
+57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,0,1,90,95,0,0,0,
+221,2,4,1,0,2,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,
+0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,
+0,57,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,109,0,16,8,48,0,
+57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,
+10,50,0,57,18,110,0,16,10,50,0,57,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,29,109,0,0,1,1,0,0,29,110,0,
+0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,
+49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,
+0,16,10,51,0,57,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,109,0,
+16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,
+109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,
+0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,26,109,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,
+10,49,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,118,0,59,120,0,18,
+109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,48,46,20,0,9,18,
+95,95,114,101,116,86,97,108,0,59,122,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,48,18,118,
+0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,48,46,20,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,28,109,0,0,
+1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,
+48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,46,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,
+18,109,0,16,10,49,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,118,0,59,
+120,0,18,109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,48,46,20,
+0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,119,0,48,
+18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,119,0,48,46,20,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,27,
+109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,18,
+109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,46,18,118,0,59,
+122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,
+118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,
+48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,20,0,0,1,90,95,0,0,12,221,2,21,1,1,0,
+0,30,109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,
+18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,46,18,118,0,
+59,122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,
+118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,
+48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,
 108,0,59,122,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,109,0,16,
-10,49,0,57,59,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,118,0,59,120,0,18,
-109,0,16,8,48,0,57,59,119,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,119,0,48,46,20,0,0,1,0,
-0,10,2,21,1,1,0,0,27,109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,
-118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,
-48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,
-108,0,59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,
-10,49,0,57,59,121,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,20,0,0,1,0,0,12,
-2,21,1,1,0,0,30,109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,118,
-0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,
-46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,
-59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,
-57,59,121,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,122,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,
-109,0,16,10,49,0,57,59,122,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,122,0,48,46,20,0,9,
-18,95,95,114,101,116,86,97,108,0,59,119,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,119,0,48,18,
-118,0,59,121,0,18,109,0,16,10,49,0,57,59,119,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,
-119,0,48,46,20,0,0,1,0,0,10,2,21,1,1,0,0,29,109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,
+10,49,0,57,59,122,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,122,0,48,46,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,119,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,119,0,48,18,118,0,59,
+121,0,18,109,0,16,10,49,0,57,59,119,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,119,0,48,
+46,20,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,29,109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,
 86,97,108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,18,109,
 0,16,10,49,0,57,59,120,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,18,118,0,59,
 119,0,18,109,0,16,10,51,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,
 118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,121,0,
 48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,18,118,0,59,119,0,18,109,0,16,10,51,
-0,57,59,121,0,48,46,20,0,0,1,0,0,11,2,21,1,1,0,0,31,109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,0,59,121,0,
-18,109,0,16,10,49,0,57,59,120,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,48,46,18,
-118,0,59,119,0,18,109,0,16,10,51,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,
-121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,
-59,121,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,18,118,0,59,119,0,18,109,0,
-16,10,51,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,118,0,59,120,0,18,
-109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,48,46,18,118,0,59,
-122,0,18,109,0,16,10,50,0,57,59,122,0,48,46,18,118,0,59,119,0,18,109,0,16,10,51,0,57,59,122,0,48,
-46,20,0,0,1,0,0,27,2,21,1,1,0,0,13,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,
-0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
-109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,29,2,21,1,1,0,0,13,109,0,0,1,1,0,0,29,110,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,26,2,21,1,1,0,0,26,109,0,0,1,1,
-0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,
-48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,
-1,0,0,14,2,21,1,1,0,0,26,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
-109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,
-110,0,16,10,50,0,57,48,20,0,0,1,0,0,31,2,21,1,1,0,0,26,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,
+0,57,59,121,0,48,46,20,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,31,109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,59,120,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,120,0,48,18,118,
+0,59,121,0,18,109,0,16,10,49,0,57,59,120,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,120,0,
+48,46,18,118,0,59,119,0,18,109,0,16,10,51,0,57,59,120,0,48,46,20,0,9,18,95,95,114,101,116,86,97,
+108,0,59,121,0,18,118,0,59,120,0,18,109,0,16,8,48,0,57,59,121,0,48,18,118,0,59,121,0,18,109,0,16,
+10,49,0,57,59,121,0,48,46,18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,121,0,48,46,18,118,0,59,119,
+0,18,109,0,16,10,51,0,57,59,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,118,0,
+59,120,0,18,109,0,16,8,48,0,57,59,122,0,48,18,118,0,59,121,0,18,109,0,16,10,49,0,57,59,122,0,48,46,
+18,118,0,59,122,0,18,109,0,16,10,50,0,57,59,122,0,48,46,18,118,0,59,119,0,18,109,0,16,10,51,0,57,
+59,122,0,48,46,20,0,0,1,90,95,0,0,27,221,2,21,1,1,0,0,13,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,95,95,
 114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
 86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,
-0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
-51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,28,2,21,1,1,0,0,28,109,0,0,1,1,0,0,13,110,
-0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,30,
-2,21,1,1,0,0,28,109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,29,221,2,21,1,1,0,0,13,109,0,
+0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,
+0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,
+20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,
+95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,
+26,221,2,21,1,1,0,0,26,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
+0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,14,221,2,21,1,1,0,0,26,109,0,0,1,1,0,0,27,110,0,0,0,
+1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,31,221,2,
+21,1,1,0,0,26,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
 109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,
 110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,
-10,50,0,57,48,20,0,0,1,0,0,15,2,21,1,1,0,0,28,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,114,101,
+10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,
+57,48,20,0,0,1,90,95,0,0,28,221,2,21,1,1,0,0,28,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,
 116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,
-57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,13,2,21,1,1,0,0,27,109,0,0,1,1,0,0,26,110,0,0,0,
-1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,27,2,21,1,
-1,0,0,27,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,
-18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,
-16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,
-0,57,48,20,0,0,1,0,0,29,2,21,1,1,0,0,27,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,
-57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,
-0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,26,2,21,1,1,0,0,14,109,0,0,1,1,0,0,26,110,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,31,2,21,1,1,0,0,14,
-109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,
+108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,30,221,2,21,1,1,0,0,28,
+109,0,0,1,1,0,0,27,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,
 16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,
 57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,
-0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,
-28,2,21,1,1,0,0,30,109,0,0,1,1,0,0,26,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
-18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,
-18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,30,2,21,1,1,0,0,30,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,15,2,21,1,1,0,0,30,109,0,0,1,1,
-0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,
-48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,13,2,21,1,1,
-0,0,29,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,
-110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,
-10,49,0,57,48,20,0,0,1,0,0,27,2,21,1,1,0,0,29,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,29,2,21,1,1,0,0,29,109,0,0,1,1,0,0,15,
-110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,26,2,21,1,1,0,0,
-31,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,
-0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,
-0,57,48,20,0,0,1,0,0,14,2,21,1,1,0,0,31,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,
-57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,31,2,21,1,1,0,0,31,109,0,0,1,1,0,0,15,110,0,0,0,
+0,0,1,90,95,0,0,15,221,2,21,1,1,0,0,28,109,0,0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,
+18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,
+18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,13,221,2,21,1,1,0,0,27,109,0,0,1,1,0,0,26,110,0,0,0,1,
+9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,27,221,
+2,21,1,1,0,0,27,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,
+110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,
+10,50,0,57,48,20,0,0,1,90,95,0,0,29,221,2,21,1,1,0,0,27,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,26,221,2,21,1,1,0,0,14,109,0,0,1,1,0,
+0,26,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,
+20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,
+90,95,0,0,31,221,2,21,1,1,0,0,14,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
+16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
+57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,
+0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,
+0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,28,221,2,21,1,1,0,0,30,109,0,0,1,1,0,0,26,110,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,30,221,2,
+21,1,1,0,0,30,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,
+110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,
+10,50,0,57,48,20,0,0,1,90,95,0,0,15,221,2,21,1,1,0,0,30,109,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,13,221,2,21,1,1,0,0,29,109,0,0,1,1,0,
+0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,
+20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,0,1,
+90,95,0,0,27,221,2,21,1,1,0,0,29,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
+16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
+57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,
+0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,29,221,2,21,1,1,0,0,29,109,0,0,1,1,0,0,15,110,0,0,0,
 1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,
 95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,
 101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,28,2,21,1,1,0,0,15,109,0,
-0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,
+86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,26,221,2,21,1,1,0,
+0,31,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,
+110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,
+10,49,0,57,48,20,0,0,1,90,95,0,0,14,221,2,21,1,1,0,0,31,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,31,221,2,21,1,1,0,0,31,109,0,
+0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,
 0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,
-20,0,0,1,0,0,30,2,21,1,1,0,0,15,109,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,
-0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,0,2,3,1,0,2,0,26,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,
-0,18,109,0,18,110,0,48,20,0,0,1,0,0,0,2,3,1,0,2,0,28,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,18,
-109,0,18,110,0,48,20,0,0,1,0,0,0,2,3,1,0,2,0,27,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,18,109,0,
-18,110,0,48,20,0,0,1,0,0,0,2,3,1,0,2,0,30,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18,
-110,0,48,20,0,0,1,0,0,0,2,3,1,0,2,0,29,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,
-48,20,0,0,1,0,0,0,2,3,1,0,2,0,31,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,
-0,0,1,0,0,11,2,21,1,1,0,0,10,118,0,0,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
+20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,9,18,
+95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,
+28,221,2,21,1,1,0,0,15,109,0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
+0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,30,221,2,21,1,1,0,0,15,109,0,0,1,1,0,0,30,110,0,0,0,
+1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,50,0,57,18,109,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,0,221,2,3,
+1,0,2,0,26,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,0,0,221,
+2,3,1,0,2,0,28,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,0,0,
+221,2,3,1,0,2,0,27,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,
+0,0,221,2,3,1,0,2,0,30,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,
+95,0,0,0,221,2,3,1,0,2,0,29,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,
+1,90,95,0,0,0,221,2,3,1,0,2,0,31,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,
+0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,10,118,0,0,1,1,0,0,27,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,
+86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,
+0,0,12,221,2,21,1,1,0,0,10,118,0,0,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
 120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,
 0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,
-97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,0,0,12,2,21,1,1,
-0,0,10,118,0,0,1,1,0,0,29,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,
+97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,59,119,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,51,0,57,0,0,20,0,0,1,90,95,0,0,
+10,221,2,21,1,1,0,0,11,118,0,0,1,1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,
+58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,
+121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,12,221,2,21,1,1,0,
+0,11,118,0,0,1,1,0,0,31,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,
 18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,
 116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,
 100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,
-0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,51,0,57,0,0,20,0,0,1,0,0,10,2,21,1,1,0,0,11,118,0,0,1,
-1,0,0,26,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,
-109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,
-0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,12,2,21,1,1,0,0,11,118,0,0,1,1,0,0,31,109,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,
-9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,
-57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,
-51,0,57,0,0,20,0,0,1,0,0,10,2,21,1,1,0,0,12,118,0,0,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,11,2,
-21,1,1,0,0,12,118,0,0,1,1,0,0,30,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,
-111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,
-100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,
-0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,0,0,0,2,1,1,0,2,0,26,109,0,0,1,1,
-0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,0,1,0,
-0,0,2,1,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,
-10,49,0,57,18,97,0,21,0,0,1,0,0,0,2,1,1,0,2,0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,
-57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,0,
-0,0,2,1,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,
-10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,0,0,0,2,1,1,0,2,0,29,109,0,0,1,1,
-0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,
-109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,0,0,0,2,1,1,0,2,0,31,
-109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,
-21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,0,0,0,2,2,1,0,
-2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,
-18,97,0,22,0,0,1,0,0,0,2,2,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,
-22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,0,0,0,2,2,1,0,2,0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,
-18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,
-18,97,0,22,0,0,1,0,0,0,2,2,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,
-22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,0,0,0,2,2,1,0,
-2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,
-18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,0,0,0,
-2,2,1,0,2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,
-49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,
-1,0,0,0,2,3,1,0,2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,
-0,16,10,49,0,57,18,97,0,23,0,0,1,0,0,0,2,3,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,
-48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,0,0,0,2,3,1,0,2,0,27,109,0,0,1,1,0,
-0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,
-0,16,10,50,0,57,18,97,0,23,0,0,1,0,0,0,2,3,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,
-48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,0,
-1,0,0,0,2,3,1,0,2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,
-0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,
-0,23,0,0,1,0,0,0,2,3,1,0,2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,
-9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,
-57,18,97,0,23,0,0,1,0,0,0,2,4,1,0,2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
-97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,0,1,0,0,0,2,4,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,
-0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,0,1,0,0,0,2,4,1,0,2,
-0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,
-97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,0,1,0,0,0,2,4,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,
-0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,
-0,57,18,97,0,24,0,0,1,0,0,0,2,4,1,0,2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
-97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,18,109,0,
-16,10,51,0,57,18,97,0,24,0,0,1,0,0,0,2,4,1,0,2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,
-48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,
-18,109,0,16,10,51,0,57,18,97,0,24,0,0,1,0,0,26,2,26,1,1,0,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,8,58,
-109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,
-110,0,16,10,49,0,57,46,0,0,0,0,1,0,0,28,2,26,1,1,0,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,
-116,50,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,
-16,10,49,0,57,46,0,0,0,0,1,0,0,27,2,26,1,1,0,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,
-51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,
-49,0,57,46,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,0,0,0,0,1,0,0,30,2,26,1,1,0,0,30,109,
-0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
-46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,
-57,46,0,0,0,0,1,0,0,29,2,26,1,1,0,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,
-18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,
-0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,
-46,0,0,0,0,1,0,0,31,2,26,1,1,0,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,
-109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,
-18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,
-46,0,0,0,0,1,0,0,26,2,27,1,1,0,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,
-109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,0,
-0,0,1,0,0,28,2,27,1,1,0,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,
+0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,51,0,57,0,0,20,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,12,
+118,0,0,1,1,0,0,28,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,
+0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,
+118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,12,118,0,0,1,1,0,0,30,
+109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,
+48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,
+16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,
+109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,
+109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,0,1,90,95,0,0,0,221,2,1,1,0,
+2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,
+18,97,0,21,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,
+57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,90,
+95,0,0,0,221,2,1,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,
+109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,90,95,0,0,0,221,2,1,1,0,
+2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,
+18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,90,95,
+0,0,0,221,2,1,1,0,2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,
+109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,
+97,0,21,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,
+18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,28,109,0,0,1,1,
+0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,90,
+95,0,0,0,221,2,2,1,0,2,0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,
+109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,90,95,0,0,0,221,2,2,1,0,
+2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,
+18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,29,109,0,0,1,1,
+0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,
+109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,90,95,0,0,0,221,2,2,1,0,
+2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,
+18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,90,95,
+0,0,0,221,2,3,1,0,2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,
+109,0,16,10,49,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,
+18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,3,1,
+0,2,0,27,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,
+57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,30,109,0,0,1,
+1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,
+109,0,16,10,50,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,
+18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,
+18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,31,109,0,0,1,1,
+0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,
+109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,4,1,0,
+2,0,26,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,
+18,97,0,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,28,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,
+57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,27,109,0,0,1,
+1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,
+109,0,16,10,50,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,30,109,0,0,1,1,0,0,9,97,0,0,0,1,9,
+18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,
+18,97,0,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,29,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,
+57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,18,
+109,0,16,10,51,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,31,109,0,0,1,1,0,0,9,97,0,0,0,1,9,
+18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,
+18,97,0,24,0,9,18,109,0,16,10,51,0,57,18,97,0,24,0,0,1,90,95,0,0,26,221,2,26,1,1,0,0,26,109,0,0,1,
+1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,
+18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,0,0,0,1,90,95,0,0,28,221,2,26,1,1,0,0,28,109,0,
+0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
+46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,0,0,0,1,90,95,0,0,27,221,2,26,1,1,0,0,27,
+109,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,
+0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,18,109,0,16,10,50,0,57,18,110,0,16,10,
+50,0,57,46,0,0,0,0,1,90,95,0,0,30,221,2,26,1,1,0,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,
+116,51,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,
+16,10,49,0,57,46,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,0,0,0,0,1,90,95,0,0,29,221,2,
+26,1,1,0,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,
+110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,18,109,0,16,10,50,0,57,
+18,110,0,16,10,50,0,57,46,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,46,0,0,0,0,1,90,95,0,0,
+31,221,2,26,1,1,0,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,
+0,57,18,110,0,16,8,48,0,57,46,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,0,18,109,0,16,10,
+50,0,57,18,110,0,16,10,50,0,57,46,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,46,0,0,0,0,1,90,
+95,0,0,26,221,2,27,1,1,0,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,
 16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,0,0,0,1,
-0,0,27,2,27,1,1,0,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,
-0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,18,109,0,16,10,
-50,0,57,18,110,0,16,10,50,0,57,47,0,0,0,0,1,0,0,30,2,27,1,1,0,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,
-8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,
-57,18,110,0,16,10,49,0,57,47,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,0,0,0,0,1,0,0,29,2,
-27,1,1,0,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,
-110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,18,109,0,16,10,50,0,57,
-18,110,0,16,10,50,0,57,47,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,47,0,0,0,0,1,0,0,31,2,27,
-1,1,0,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,110,
-0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,18,109,0,16,10,50,0,57,18,
-110,0,16,10,50,0,57,47,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,47,0,0,0,0,1,0,0,26,2,22,1,
-1,0,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,
-16,8,48,0,57,49,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,0,0,0,1,0,0,28,2,22,1,1,0,0,
-28,109,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,
-48,0,57,49,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,0,0,0,1,0,0,27,2,22,1,1,0,0,27,109,
-0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
-49,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,
-57,49,0,0,0,0,1,0,0,30,2,22,1,1,0,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,
-18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,
-0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,0,0,0,0,1,0,0,29,2,22,1,1,0,0,29,109,0,0,1,1,0,
-0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,
-109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,0,
-18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,0,0,0,0,1,0,0,31,2,22,1,1,0,0,31,109,0,0,1,1,0,0,
-31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,
+90,95,0,0,28,221,2,27,1,1,0,0,28,109,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,
+0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,0,0,0,
+1,90,95,0,0,27,221,2,27,1,1,0,0,27,109,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,
+109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,
+18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,0,0,0,0,1,90,95,0,0,30,221,2,27,1,1,0,0,30,109,0,
+0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
+47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,
+57,47,0,0,0,0,1,90,95,0,0,29,221,2,27,1,1,0,0,29,109,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,116,52,
+120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,
+0,57,47,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,0,18,109,0,16,10,51,0,57,18,110,0,16,10,
+51,0,57,47,0,0,0,0,1,90,95,0,0,31,221,2,27,1,1,0,0,31,109,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,
+116,52,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,0,18,109,0,16,10,49,0,57,18,110,0,
+16,10,49,0,57,47,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,0,18,109,0,16,10,51,0,57,18,
+110,0,16,10,51,0,57,47,0,0,0,0,1,90,95,0,0,26,221,2,22,1,1,0,0,26,109,0,0,1,1,0,0,26,110,0,0,0,1,8,
+58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,0,16,10,49,0,57,
+18,110,0,16,10,49,0,57,49,0,0,0,0,1,90,95,0,0,28,221,2,22,1,1,0,0,28,109,0,0,1,1,0,0,28,110,0,0,0,
+1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,0,16,10,49,0,
+57,18,110,0,16,10,49,0,57,49,0,0,0,0,1,90,95,0,0,27,221,2,22,1,1,0,0,27,109,0,0,1,1,0,0,27,110,0,0,
+0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,0,16,10,49,
+0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,0,0,0,0,1,90,95,
+0,0,30,221,2,22,1,1,0,0,30,109,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,
+8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,
+16,10,50,0,57,18,110,0,16,10,50,0,57,49,0,0,0,0,1,90,95,0,0,29,221,2,22,1,1,0,0,29,109,0,0,1,1,0,0,
+29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,18,109,
 0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,0,18,
-109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,0,0,0,0,1,0,0,26,2,26,1,1,0,0,9,97,0,0,1,1,0,0,26,
-110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,
-49,0,57,46,0,0,0,0,1,0,0,26,2,26,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,50,120,51,
-0,18,109,0,16,8,48,0,57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,0,0,0,1,0,0,28,2,26,1,1,0,
-0,9,97,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,97,0,18,110,0,16,8,48,0,57,46,0,
-18,97,0,18,110,0,16,10,49,0,57,46,0,0,0,0,1,0,0,28,2,26,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,8,
-58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,0,
-0,0,1,0,0,27,2,26,1,1,0,0,9,97,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,97,0,18,
-110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,110,0,16,10,50,0,57,46,0,0,
-0,0,1,0,0,27,2,26,1,1,0,0,27,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,
-8,48,0,57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,0,0,
-0,1,0,0,30,2,26,1,1,0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,97,0,18,110,
-0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,110,0,16,10,50,0,57,46,0,0,0,0,
-1,0,0,30,2,26,1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,
-0,57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,0,0,0,1,
-0,0,29,2,26,1,1,0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,
-57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,18,109,0,
-16,10,51,0,57,18,98,0,46,0,0,0,0,1,0,0,29,2,26,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,
-116,52,120,50,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,
-110,0,16,10,50,0,57,46,0,18,97,0,18,110,0,16,10,51,0,57,46,0,0,0,0,1,0,0,31,2,26,1,1,0,0,31,109,0,
-0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,98,0,46,0,18,109,0,16,
-10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,18,109,0,16,10,51,0,57,18,98,0,46,0,0,
-0,0,1,0,0,31,2,26,1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,97,0,18,
-110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,110,0,16,10,50,0,57,46,0,18,
-97,0,18,110,0,16,10,51,0,57,46,0,0,0,0,1,0,0,26,2,27,1,1,0,0,9,97,0,0,1,1,0,0,26,110,0,0,0,1,8,58,
-109,97,116,50,120,51,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,0,0,
-0,1,0,0,26,2,27,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,
-48,0,57,18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,0,0,0,1,0,0,28,2,27,1,1,0,0,9,97,0,0,1,1,
-0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,
-16,10,49,0,57,47,0,0,0,0,1,0,0,28,2,27,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,50,
-120,52,0,18,109,0,16,8,48,0,57,18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,0,0,0,1,0,0,27,2,
-27,1,1,0,0,9,97,0,0,1,1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,97,0,18,110,0,16,8,48,0,
-57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,57,47,0,0,0,0,1,0,0,27,2,
-27,1,1,0,0,27,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,98,
-0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,0,0,0,1,0,0,30,2,27,
-1,1,0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,97,0,18,110,0,16,8,48,0,57,
-47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,57,47,0,0,0,0,1,0,0,30,2,27,1,
-1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,98,0,47,
-0,18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,0,0,0,1,0,0,29,2,27,1,1,
-0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,98,0,47,0,
-18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,18,109,0,16,10,51,0,57,18,
-98,0,47,0,0,0,0,1,0,0,29,2,27,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,
-18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,
-57,47,0,18,97,0,18,110,0,16,10,51,0,57,47,0,0,0,0,1,0,0,31,2,27,1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,
-0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,
-0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,18,109,0,16,10,51,0,57,18,98,0,47,0,0,0,0,1,0,0,31,2,27,
-1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,97,0,18,110,0,16,8,48,0,57,
-47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,57,47,0,18,97,0,18,110,0,16,
-10,51,0,57,47,0,0,0,0,1,0,0,26,2,21,1,1,0,0,9,97,0,0,1,1,0,0,26,110,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,26,2,21,1,1,0,0,26,109,0,0,1,1,0,0,9,
-98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,0,0,28,2,
-21,1,1,0,0,9,97,0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,
-18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,
-10,49,0,57,48,20,0,0,1,0,0,28,2,21,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,0,0,27,2,21,1,1,0,0,9,97,0,0,1,1,0,0,27,
-110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,27,2,21,1,1,0,
-0,27,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,
-0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,
-0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,
-0,1,0,0,30,2,21,1,1,0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,
-0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,
-16,10,50,0,57,48,20,0,0,1,0,0,30,2,21,1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,
+109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,0,0,0,0,1,90,95,0,0,31,221,2,22,1,1,0,0,31,109,0,0,1,
+1,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,0,
+18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,
+49,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,0,0,0,0,1,90,95,0,0,26,221,2,26,1,1,0,0,9,97,
+0,0,1,1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,
+18,110,0,16,10,49,0,57,46,0,0,0,0,1,90,95,0,0,26,221,2,26,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,
+8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,
+0,0,0,1,90,95,0,0,28,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,
+18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,0,0,0,1,90,95,0,0,28,221,2,
+26,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,18,98,
+0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,0,0,0,1,90,95,0,0,27,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,
+27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,
+10,49,0,57,46,0,18,97,0,18,110,0,16,10,50,0,57,46,0,0,0,0,1,90,95,0,0,27,221,2,26,1,1,0,0,27,109,0,
+0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,98,0,46,0,18,109,0,16,
+10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,0,0,0,1,90,95,0,0,30,221,2,26,1,1,0,0,
+9,97,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,
+97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,110,0,16,10,50,0,57,46,0,0,0,0,1,90,95,0,0,30,221,2,26,
+1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,98,0,
+46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,0,0,0,1,90,95,0,0,29,
+221,2,26,1,1,0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,
+18,98,0,46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,18,109,0,16,
+10,51,0,57,18,98,0,46,0,0,0,0,1,90,95,0,0,29,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,8,58,
+109,97,116,52,120,50,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,
+97,0,18,110,0,16,10,50,0,57,46,0,18,97,0,18,110,0,16,10,51,0,57,46,0,0,0,0,1,90,95,0,0,31,221,2,26,
+1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,98,0,
+46,0,18,109,0,16,10,49,0,57,18,98,0,46,0,18,109,0,16,10,50,0,57,18,98,0,46,0,18,109,0,16,10,51,0,
+57,18,98,0,46,0,0,0,0,1,90,95,0,0,31,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,
+116,52,120,51,0,18,97,0,18,110,0,16,8,48,0,57,46,0,18,97,0,18,110,0,16,10,49,0,57,46,0,18,97,0,18,
+110,0,16,10,50,0,57,46,0,18,97,0,18,110,0,16,10,51,0,57,46,0,0,0,0,1,90,95,0,0,26,221,2,27,1,1,0,0,
+9,97,0,0,1,1,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,
+97,0,18,110,0,16,10,49,0,57,47,0,0,0,0,1,90,95,0,0,26,221,2,27,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,
+0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,0,
+47,0,0,0,0,1,90,95,0,0,28,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,
+52,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,0,0,0,1,90,95,0,0,28,
+221,2,27,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,
+18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,0,0,0,1,90,95,0,0,27,221,2,27,1,1,0,0,9,97,0,0,1,
+1,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,
+0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,57,47,0,0,0,0,1,90,95,0,0,27,221,2,27,1,1,0,0,27,
+109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,98,0,47,0,18,109,
+0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,0,0,0,1,90,95,0,0,30,221,2,27,1,1,
+0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,97,0,18,110,0,16,8,48,0,57,47,0,
+18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,110,0,16,10,50,0,57,47,0,0,0,0,1,90,95,0,0,30,221,2,
+27,1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,98,
+0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,0,0,0,1,90,95,0,0,
+29,221,2,27,1,1,0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,
+57,18,98,0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,18,109,0,
+16,10,51,0,57,18,98,0,47,0,0,0,0,1,90,95,0,0,29,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,8,
+58,109,97,116,52,120,50,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,
+18,97,0,18,110,0,16,10,50,0,57,47,0,18,97,0,18,110,0,16,10,51,0,57,47,0,0,0,0,1,90,95,0,0,31,221,2,
+27,1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,98,
+0,47,0,18,109,0,16,10,49,0,57,18,98,0,47,0,18,109,0,16,10,50,0,57,18,98,0,47,0,18,109,0,16,10,51,0,
+57,18,98,0,47,0,0,0,0,1,90,95,0,0,31,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,8,58,109,97,
+116,52,120,51,0,18,97,0,18,110,0,16,8,48,0,57,47,0,18,97,0,18,110,0,16,10,49,0,57,47,0,18,97,0,18,
+110,0,16,10,50,0,57,47,0,18,97,0,18,110,0,16,10,51,0,57,47,0,0,0,0,1,90,95,0,0,26,221,2,21,1,1,0,0,
+9,97,0,0,1,1,0,0,26,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,
+16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,
+57,48,20,0,0,1,90,95,0,0,26,221,2,21,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,
 116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,0,1,0,0,29,2,21,1,1,0,0,29,109,0,0,1,1,0,0,9,98,
-0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,48,20,0,0,1,0,0,29,2,21,1,1,0,0,9,97,0,0,1,1,
-0,0,29,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,
+108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,90,95,0,0,28,221,2,21,1,1,0,0,9,97,
+0,0,1,1,0,0,28,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,
+48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,
+20,0,0,1,90,95,0,0,28,221,2,21,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,90,95,0,0,27,221,2,21,1,1,0,0,9,97,0,0,1,1,0,
+0,27,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,
+20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,
+95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,
+27,221,2,21,1,1,0,0,27,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,
+16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,
+57,18,98,0,48,20,0,0,1,90,95,0,0,30,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,30,221,2,21,1,1,0,0,30,109,0,
+0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,
+0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,
+9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,0,1,90,95,
+0,0,29,221,2,21,1,1,0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
+0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
+0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,
+0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,
+0,48,20,0,0,1,90,95,0,0,29,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,
+57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,
+18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,31,221,2,21,1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,
+97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,48,20,0,0,1,90,95,0,0,31,221,2,21,1,1,0,0,9,97,0,0,1,
+1,0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,
 48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,
 18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,31,2,21,1,1,0,
-0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,
-0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,
-0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,48,20,0,0,1,0,0,31,
-2,21,1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,
-0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,
-16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,
-57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,48,20,0,
-0,1,0,0,26,2,22,1,1,0,0,9,97,0,0,1,1,0,0,26,110,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,
+114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,26,221,
+2,22,1,1,0,0,9,97,0,0,1,1,0,0,26,110,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,
 18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,
 57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,
-57,48,20,0,0,1,0,0,26,2,22,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,
-49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,
-18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,
-18,105,110,118,0,48,20,0,0,1,0,0,28,2,22,1,1,0,0,9,97,0,0,1,1,0,0,28,110,0,0,0,1,3,2,1,0,9,1,105,
-110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,
-110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,105,
-110,118,0,18,110,0,16,10,49,0,57,48,20,0,0,1,0,0,28,2,22,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,3,
-2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,18,109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,0,1,0,0,27,2,22,1,1,0,0,9,97,0,0,1,1,0,0,27,110,
-0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,
-0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,50,0,57,18,105,110,118,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,27,2,22,1,1,0,0,27,109,0,0,1,
-1,0,0,9,98,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,0,1,0,0,30,2,22,1,1,0,0,
-9,97,0,0,1,1,0,0,30,110,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,
-95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,50,0,57,18,105,110,118,0,18,110,0,16,10,50,0,57,48,20,0,0,1,0,0,30,2,
-22,1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,
-0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,0,1,
-0,0,29,2,22,1,1,0,0,29,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,0,48,0,0,18,
+57,48,20,0,0,1,90,95,0,0,26,221,2,22,1,1,0,0,26,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,
+105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,0,1,90,95,0,0,28,221,2,22,1,1,0,0,9,97,0,0,1,1,0,0,28,
+110,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,
+116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,57,48,20,0,0,1,90,95,0,0,28,221,2,
+22,1,1,0,0,28,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,18,
 98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,
 48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,105,110,118,0,
-48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,105,110,118,0,
-48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,105,110,118,0,
-48,20,0,0,1,0,0,29,2,22,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,3,2,1,0,9,1,105,110,118,0,2,17,49,
-0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,
-16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,
-10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,105,110,118,0,18,110,0,16,
-10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,105,110,118,0,18,110,0,16,
-10,51,0,57,48,20,0,0,1,0,0,31,2,22,1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,1,0,9,1,105,110,118,
-0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,
-0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,
-57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,
-57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,
-57,18,105,110,118,0,48,20,0,0,1,0,0,31,2,22,1,1,0,0,9,97,0,0,1,1,0,0,31,110,0,0,0,1,3,2,1,0,9,1,
+48,20,0,0,1,90,95,0,0,27,221,2,22,1,1,0,0,9,97,0,0,1,1,0,0,27,110,0,0,0,1,3,2,90,95,1,0,9,221,1,
 105,110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
 105,110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
 105,110,118,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
-105,110,118,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,
-105,110,118,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,26,2,27,1,1,0,0,26,109,0,0,0,1,8,58,109,97,
-116,50,120,51,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,0,0,0,1,0,0,28,2,27,1,1,0,0,
-28,109,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,
-0,0,0,1,0,0,27,2,27,1,1,0,0,27,109,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,54,0,
-18,109,0,16,10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,0,0,0,1,0,0,30,2,27,1,1,0,0,30,109,0,0,0,1,
-8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,18,109,0,16,10,
-50,0,57,54,0,0,0,0,1,0,0,29,2,27,1,1,0,0,29,109,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,
-48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,18,109,0,16,10,51,0,57,54,0,0,
-0,0,1,0,0,31,2,27,1,1,0,0,31,109,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,54,0,18,
-109,0,16,10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,18,109,0,16,10,51,0,57,54,0,0,0,0,1,0,0,0,2,
-25,1,0,2,0,26,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,0,1,0,0,0,2,
-25,1,0,2,0,28,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,0,1,0,0,0,2,
-25,1,0,2,0,27,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,9,18,109,0,16,
-10,50,0,57,52,0,0,1,0,0,0,2,25,1,0,2,0,30,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,
-10,49,0,57,52,0,9,18,109,0,16,10,50,0,57,52,0,0,1,0,0,0,2,25,1,0,2,0,29,109,0,0,0,1,9,18,109,0,16,
-8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,9,18,109,0,16,10,50,0,57,52,0,9,18,109,0,16,10,51,0,
-57,52,0,0,1,0,0,0,2,25,1,0,2,0,31,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,
-57,52,0,9,18,109,0,16,10,50,0,57,52,0,9,18,109,0,16,10,51,0,57,52,0,0,1,0,0,0,2,24,1,0,2,0,26,109,
-0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,0,1,0,0,0,2,24,1,0,2,0,28,109,0,
-0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,0,1,0,0,0,2,24,1,0,2,0,27,109,0,0,
-0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,109,0,16,10,50,0,57,51,0,0,1,0,
-0,0,2,24,1,0,2,0,30,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,
-109,0,16,10,50,0,57,51,0,0,1,0,0,0,2,24,1,0,2,0,29,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,
-109,0,16,10,49,0,57,51,0,9,18,109,0,16,10,50,0,57,51,0,9,18,109,0,16,10,51,0,57,51,0,0,1,0,0,0,2,
-24,1,0,2,0,31,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,109,0,16,
-10,50,0,57,51,0,9,18,109,0,16,10,51,0,57,51,0,0,0
+105,110,118,0,18,110,0,16,10,50,0,57,48,20,0,0,1,90,95,0,0,27,221,2,22,1,1,0,0,27,109,0,0,1,1,0,0,
+9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,0,1,90,95,0,0,30,221,
+2,22,1,1,0,0,9,97,0,0,1,1,0,0,30,110,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,
+18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,
+57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,
+57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,105,110,118,0,18,110,0,16,10,50,0,
+57,48,20,0,0,1,90,95,0,0,30,221,2,22,1,1,0,0,30,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,
+105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
+109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,0,1,90,95,0,0,29,221,2,22,1,1,0,0,29,109,0,0,1,1,0,0,
+9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,105,110,118,0,48,20,0,0,1,90,95,0,0,29,221,
+2,22,1,1,0,0,9,97,0,0,1,1,0,0,29,110,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,
+18,97,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,
+57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,
+57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,105,110,118,0,18,110,0,16,10,50,0,
+57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,105,110,118,0,18,110,0,16,10,51,0,
+57,48,20,0,0,1,90,95,0,0,31,221,2,22,1,1,0,0,31,109,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,1,0,9,221,1,
+105,110,118,0,2,17,49,0,48,0,0,18,98,0,49,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,16,8,48,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+109,0,16,10,49,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
+109,0,16,10,50,0,57,18,105,110,118,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,
+109,0,16,10,51,0,57,18,105,110,118,0,48,20,0,0,1,90,95,0,0,31,221,2,22,1,1,0,0,9,97,0,0,1,1,0,0,31,
+110,0,0,0,1,3,2,90,95,1,0,9,221,1,105,110,118,0,2,17,49,0,48,0,0,18,97,0,49,0,0,9,18,95,95,114,101,
+116,86,97,108,0,16,8,48,0,57,18,105,110,118,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,105,110,118,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,50,0,57,18,105,110,118,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,51,0,57,18,105,110,118,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,26,221,2,
+27,1,1,0,0,26,109,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,
+0,57,54,0,0,0,0,1,90,95,0,0,28,221,2,27,1,1,0,0,28,109,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,
+0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,0,0,0,1,90,95,0,0,27,221,2,27,1,1,0,0,27,109,0,0,0,
+1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,18,109,0,16,
+10,50,0,57,54,0,0,0,0,1,90,95,0,0,30,221,2,27,1,1,0,0,30,109,0,0,0,1,8,58,109,97,116,51,120,52,0,
+18,109,0,16,8,48,0,57,54,0,18,109,0,16,10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,0,0,0,1,90,95,0,
+0,29,221,2,27,1,1,0,0,29,109,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,54,0,18,109,
+0,16,10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,18,109,0,16,10,51,0,57,54,0,0,0,0,1,90,95,0,0,31,
+221,2,27,1,1,0,0,31,109,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,54,0,18,109,0,16,
+10,49,0,57,54,0,18,109,0,16,10,50,0,57,54,0,18,109,0,16,10,51,0,57,54,0,0,0,0,1,90,95,0,0,0,221,2,
+25,1,0,2,0,26,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,0,1,90,95,0,0,
+0,221,2,25,1,0,2,0,28,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,0,1,
+90,95,0,0,0,221,2,25,1,0,2,0,27,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,
+52,0,9,18,109,0,16,10,50,0,57,52,0,0,1,90,95,0,0,0,221,2,25,1,0,2,0,30,109,0,0,0,1,9,18,109,0,16,8,
+48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,9,18,109,0,16,10,50,0,57,52,0,0,1,90,95,0,0,0,221,2,25,
+1,0,2,0,29,109,0,0,0,1,9,18,109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,9,18,109,0,16,10,
+50,0,57,52,0,9,18,109,0,16,10,51,0,57,52,0,0,1,90,95,0,0,0,221,2,25,1,0,2,0,31,109,0,0,0,1,9,18,
+109,0,16,8,48,0,57,52,0,9,18,109,0,16,10,49,0,57,52,0,9,18,109,0,16,10,50,0,57,52,0,9,18,109,0,16,
+10,51,0,57,52,0,0,1,90,95,0,0,0,221,2,24,1,0,2,0,26,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,
+109,0,16,10,49,0,57,51,0,0,1,90,95,0,0,0,221,2,24,1,0,2,0,28,109,0,0,0,1,9,18,109,0,16,8,48,0,57,
+51,0,9,18,109,0,16,10,49,0,57,51,0,0,1,90,95,0,0,0,221,2,24,1,0,2,0,27,109,0,0,0,1,9,18,109,0,16,8,
+48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,109,0,16,10,50,0,57,51,0,0,1,90,95,0,0,0,221,2,24,
+1,0,2,0,30,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,109,0,16,10,
+50,0,57,51,0,0,1,90,95,0,0,0,221,2,24,1,0,2,0,29,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,
+0,16,10,49,0,57,51,0,9,18,109,0,16,10,50,0,57,51,0,9,18,109,0,16,10,51,0,57,51,0,0,1,90,95,0,0,0,
+221,2,24,1,0,2,0,31,109,0,0,0,1,9,18,109,0,16,8,48,0,57,51,0,9,18,109,0,16,10,49,0,57,51,0,9,18,
+109,0,16,10,50,0,57,51,0,9,18,109,0,16,10,51,0,57,51,0,0,0
diff --git a/src/mesa/shader/slang/library/slang_builtin_120_common_gc.h b/src/mesa/shader/slang/library/slang_builtin_120_common_gc.h
index d9399d1a537..2ff49ef45f6 100644
--- a/src/mesa/shader/slang/library/slang_builtin_120_common_gc.h
+++ b/src/mesa/shader/slang/library/slang_builtin_120_common_gc.h
@@ -2,104 +2,106 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_builtin_120_common.gc */
 
-4,1,0,0,26,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,26,109,0,0,1,0,0,0,26,
-110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,
-16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,0,0,0,1,0,0,28,0,109,97,116,114,105,120,67,111,109,112,
-77,117,108,116,0,1,0,0,0,28,109,0,0,1,0,0,0,28,110,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,
-8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,0,0,0,1,0,
-0,27,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,27,109,0,0,1,0,0,0,27,110,0,
-0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,
-49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,0,0,0,1,0,0,
-30,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,30,109,0,0,1,0,0,0,30,110,0,0,
-0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,
-0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,0,0,0,1,0,0,29,
-0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,29,109,0,0,1,0,0,0,29,110,0,0,0,1,
-8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,
-57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,18,109,0,16,10,
-51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,0,0,31,0,109,97,116,114,105,120,67,111,109,112,77,117,
-108,116,0,1,0,0,0,31,109,0,0,1,0,0,0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,
-57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,
-0,57,18,110,0,16,10,50,0,57,48,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,0,0,13,
-0,111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,10,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,
-109,97,116,50,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,
-99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,0,0,0,1,0,0,14,0,111,
-117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,11,114,0,0,0,1,8,58,109,97,
-116,51,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,
+4,1,90,95,0,0,26,221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,26,109,0,0,1,
+0,0,0,26,110,0,0,0,1,8,58,109,97,116,50,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,
+18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,0,0,0,1,90,95,0,0,28,221,0,109,97,116,114,105,
+120,67,111,109,112,77,117,108,116,0,1,0,0,0,28,109,0,0,1,0,0,0,28,110,0,0,0,1,8,58,109,97,116,50,
+120,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,
+0,57,48,0,0,0,0,1,90,95,0,0,27,221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,
+0,27,109,0,0,1,0,0,0,27,110,0,0,0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,18,110,0,16,
+8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,
+16,10,50,0,57,48,0,0,0,0,1,90,95,0,0,30,221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,
+0,1,0,0,0,30,109,0,0,1,0,0,0,30,110,0,0,0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,18,
+110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,
+18,110,0,16,10,50,0,57,48,0,0,0,0,1,90,95,0,0,29,221,0,109,97,116,114,105,120,67,111,109,112,77,
+117,108,116,0,1,0,0,0,29,109,0,0,1,0,0,0,29,110,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,
+48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,
+10,50,0,57,18,110,0,16,10,50,0,57,48,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,
+90,95,0,0,31,221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,31,109,0,0,1,0,0,
+0,31,110,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,
+109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,
+18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,90,95,0,0,13,221,0,111,117,116,101,114,
+80,114,111,100,117,99,116,0,1,0,0,0,10,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,109,97,116,50,0,18,99,0,
+59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,
+59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,0,0,0,1,90,95,0,0,14,221,0,111,117,116,101,
+114,80,114,111,100,117,99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,11,114,0,0,0,1,8,58,109,97,116,51,0,18,
+99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,122,0,18,
+114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,
+0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,
+18,114,0,59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,0,0,0,1,90,95,0,0,15,221,0,111,117,
+116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,12,99,0,0,1,0,0,0,12,114,0,0,0,1,8,58,109,97,116,
+52,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,
+122,0,18,114,0,59,120,0,48,0,18,99,0,59,119,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,
+121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,
+59,119,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,114,0,
+59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,59,119,0,18,114,0,59,122,0,48,0,18,
+99,0,59,120,0,18,114,0,59,119,0,48,0,18,99,0,59,121,0,18,114,0,59,119,0,48,0,18,99,0,59,122,0,18,
+114,0,59,119,0,48,0,18,99,0,59,119,0,18,114,0,59,119,0,48,0,0,0,0,1,90,95,0,0,26,221,0,111,117,116,
+101,114,80,114,111,100,117,99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,109,97,116,50,
+120,51,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,
 59,122,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,
-59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,
-99,0,59,121,0,18,114,0,59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,0,0,0,1,0,0,15,0,111,
-117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,12,99,0,0,1,0,0,0,12,114,0,0,0,1,8,58,109,97,
-116,52,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,
-59,122,0,18,114,0,59,120,0,48,0,18,99,0,59,119,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,
-59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,
-99,0,59,119,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,
-114,0,59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,59,119,0,18,114,0,59,122,0,48,
-0,18,99,0,59,120,0,18,114,0,59,119,0,48,0,18,99,0,59,121,0,18,114,0,59,119,0,48,0,18,99,0,59,122,0,
-18,114,0,59,119,0,48,0,18,99,0,59,119,0,18,114,0,59,119,0,48,0,0,0,0,1,0,0,26,0,111,117,116,101,
-114,80,114,111,100,117,99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,109,97,116,50,120,51,
-0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,122,0,
-18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,
-48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,0,0,0,1,0,0,27,0,111,117,116,101,114,80,114,111,100,
-117,99,116,0,1,0,0,0,10,99,0,0,1,0,0,0,11,114,0,0,0,1,8,58,109,97,116,51,120,50,0,18,99,0,59,120,0,
-18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,
-48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,
-121,0,18,114,0,59,122,0,48,0,0,0,0,1,0,0,28,0,111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,
-0,0,12,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,109,97,116,50,120,52,0,18,99,0,59,120,0,18,114,0,59,120,
-0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,122,0,18,114,0,59,120,0,48,0,18,99,0,59,
-119,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,
-121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,59,119,0,18,114,0,59,121,0,48,0,0,0,0,1,
-0,0,29,0,111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,10,99,0,0,1,0,0,0,12,114,0,0,0,1,
-8,58,109,97,116,52,120,50,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,
-120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,
-59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,114,0,59,122,0,48,0,18,99,0,59,120,0,18,114,0,
-59,119,0,48,0,18,99,0,59,121,0,18,114,0,59,119,0,48,0,0,0,0,1,0,0,30,0,111,117,116,101,114,80,114,
-111,100,117,99,116,0,1,0,0,0,12,99,0,0,1,0,0,0,11,114,0,0,0,1,8,58,109,97,116,51,120,52,0,18,99,0,
+59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,0,0,0,1,90,95,0,0,27,221,0,111,117,116,101,
+114,80,114,111,100,117,99,116,0,1,0,0,0,10,99,0,0,1,0,0,0,11,114,0,0,0,1,8,58,109,97,116,51,120,50,
+0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,
+18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,
+48,0,18,99,0,59,121,0,18,114,0,59,122,0,48,0,0,0,0,1,90,95,0,0,28,221,0,111,117,116,101,114,80,114,
+111,100,117,99,116,0,1,0,0,0,12,99,0,0,1,0,0,0,10,114,0,0,0,1,8,58,109,97,116,50,120,52,0,18,99,0,
 59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,122,0,18,114,0,
 59,120,0,48,0,18,99,0,59,119,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,
 99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,59,119,0,18,
-114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,114,0,59,122,0,48,
-0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,59,119,0,18,114,0,59,122,0,48,0,0,0,0,1,0,0,31,0,
-111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,12,114,0,0,0,1,8,58,109,
-97,116,52,120,51,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,
-18,99,0,59,122,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,
-18,114,0,59,121,0,48,0,18,99,0,59,122,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,
-48,0,18,99,0,59,121,0,18,114,0,59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,59,
-120,0,18,114,0,59,119,0,48,0,18,99,0,59,121,0,18,114,0,59,119,0,48,0,18,99,0,59,122,0,18,114,0,59,
-119,0,48,0,0,0,0,1,0,0,13,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,13,109,0,0,0,1,8,58,109,
-97,116,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,8,48,0,
-57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,0,0,0,1,0,0,14,0,116,114,97,110,115,112,111,115,
-101,0,1,0,0,0,14,109,0,0,0,1,8,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,
-49,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,
-10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,
-16,10,49,0,57,59,122,0,0,18,109,0,16,10,50,0,57,59,122,0,0,0,0,0,1,0,0,15,0,116,114,97,110,115,112,
-111,115,101,0,1,0,0,0,15,109,0,0,0,1,8,58,109,97,116,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,
-0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,
-109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,
-18,109,0,16,10,51,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,
-0,18,109,0,16,10,50,0,57,59,122,0,0,18,109,0,16,10,51,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,
-0,0,18,109,0,16,10,49,0,57,59,119,0,0,18,109,0,16,10,50,0,57,59,119,0,0,18,109,0,16,10,51,0,57,59,
-119,0,0,0,0,0,1,0,0,26,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,27,109,0,0,0,1,8,58,109,97,
-116,50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,
-50,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,
-10,50,0,57,59,121,0,0,0,0,0,1,0,0,27,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,26,109,0,0,0,1,
-8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,
-109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,
-18,109,0,16,10,49,0,57,59,122,0,0,0,0,0,1,0,0,28,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,29,
-109,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,
-120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,109,0,16,8,48,0,57,
-59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,10,51,0,
-57,59,121,0,0,0,0,0,1,0,0,29,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,28,109,0,0,0,1,8,58,
-109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,
-16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,
-0,16,10,49,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,0,0,18,109,0,16,10,49,0,57,59,119,0,0,0,0,
-0,1,0,0,30,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,31,109,0,0,0,1,8,58,109,97,116,51,120,52,
-0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,
-0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,
-121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,10,51,0,57,59,121,0,0,18,109,0,16,8,48,0,57,
-59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,0,18,109,0,16,10,50,0,57,59,122,0,0,18,109,0,16,10,51,0,
-57,59,122,0,0,0,0,0,1,0,0,31,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,30,109,0,0,0,1,8,58,
-109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,
-16,10,50,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,
-0,16,10,50,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,0,18,
-109,0,16,10,50,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,0,0,18,109,0,16,10,49,0,57,59,119,0,0,
-18,109,0,16,10,50,0,57,59,119,0,0,0,0,0,0
+114,0,59,121,0,48,0,0,0,0,1,90,95,0,0,29,221,0,111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,
+0,0,10,99,0,0,1,0,0,0,12,114,0,0,0,1,8,58,109,97,116,52,120,50,0,18,99,0,59,120,0,18,114,0,59,120,
+0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,
+121,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,114,0,59,
+122,0,48,0,18,99,0,59,120,0,18,114,0,59,119,0,48,0,18,99,0,59,121,0,18,114,0,59,119,0,48,0,0,0,0,1,
+90,95,0,0,30,221,0,111,117,116,101,114,80,114,111,100,117,99,116,0,1,0,0,0,12,99,0,0,1,0,0,0,11,
+114,0,0,0,1,8,58,109,97,116,51,120,52,0,18,99,0,59,120,0,18,114,0,59,120,0,48,0,18,99,0,59,121,0,
+18,114,0,59,120,0,48,0,18,99,0,59,122,0,18,114,0,59,120,0,48,0,18,99,0,59,119,0,18,114,0,59,120,0,
+48,0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,
+122,0,18,114,0,59,121,0,48,0,18,99,0,59,119,0,18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,
+122,0,48,0,18,99,0,59,121,0,18,114,0,59,122,0,48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,
+59,119,0,18,114,0,59,122,0,48,0,0,0,0,1,90,95,0,0,31,221,0,111,117,116,101,114,80,114,111,100,117,
+99,116,0,1,0,0,0,11,99,0,0,1,0,0,0,12,114,0,0,0,1,8,58,109,97,116,52,120,51,0,18,99,0,59,120,0,18,
+114,0,59,120,0,48,0,18,99,0,59,121,0,18,114,0,59,120,0,48,0,18,99,0,59,122,0,18,114,0,59,120,0,48,
+0,18,99,0,59,120,0,18,114,0,59,121,0,48,0,18,99,0,59,121,0,18,114,0,59,121,0,48,0,18,99,0,59,122,0,
+18,114,0,59,121,0,48,0,18,99,0,59,120,0,18,114,0,59,122,0,48,0,18,99,0,59,121,0,18,114,0,59,122,0,
+48,0,18,99,0,59,122,0,18,114,0,59,122,0,48,0,18,99,0,59,120,0,18,114,0,59,119,0,48,0,18,99,0,59,
+121,0,18,114,0,59,119,0,48,0,18,99,0,59,122,0,18,114,0,59,119,0,48,0,0,0,0,1,90,95,0,0,13,221,0,
+116,114,97,110,115,112,111,115,101,0,1,0,0,0,13,109,0,0,0,1,8,58,109,97,116,50,0,18,109,0,16,8,48,
+0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,
+49,0,57,59,121,0,0,0,0,0,1,90,95,0,0,14,221,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,14,109,
+0,0,0,1,8,58,109,97,116,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,
+109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,
+18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,
+0,18,109,0,16,10,50,0,57,59,122,0,0,0,0,0,1,90,95,0,0,15,221,0,116,114,97,110,115,112,111,115,101,
+0,1,0,0,0,15,109,0,0,0,1,8,58,109,97,116,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,
+57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,109,0,16,8,48,
+0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,10,
+51,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,0,18,109,0,16,
+10,50,0,57,59,122,0,0,18,109,0,16,10,51,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,0,0,18,109,0,
+16,10,49,0,57,59,119,0,0,18,109,0,16,10,50,0,57,59,119,0,0,18,109,0,16,10,51,0,57,59,119,0,0,0,0,0,
+1,90,95,0,0,26,221,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,27,109,0,0,0,1,8,58,109,97,116,
+50,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,50,0,
+57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,
+0,57,59,121,0,0,0,0,0,1,90,95,0,0,27,221,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,26,109,0,0,
+0,1,8,58,109,97,116,51,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,
+18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,
+0,18,109,0,16,10,49,0,57,59,122,0,0,0,0,0,1,90,95,0,0,28,221,0,116,114,97,110,115,112,111,115,101,
+0,1,0,0,0,29,109,0,0,0,1,8,58,109,97,116,50,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,
+10,49,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,109,0,
+16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,
+0,16,10,51,0,57,59,121,0,0,0,0,0,1,90,95,0,0,29,221,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,
+28,109,0,0,0,1,8,58,109,97,116,52,120,50,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,
+59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,8,48,0,
+57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,0,0,18,109,0,16,10,49,
+0,57,59,119,0,0,0,0,0,1,90,95,0,0,30,221,0,116,114,97,110,115,112,111,115,101,0,1,0,0,0,31,109,0,0,
+0,1,8,58,109,97,116,51,120,52,0,18,109,0,16,8,48,0,57,59,120,0,0,18,109,0,16,10,49,0,57,59,120,0,0,
+18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,10,51,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,
+0,18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,10,51,0,57,59,
+121,0,0,18,109,0,16,8,48,0,57,59,122,0,0,18,109,0,16,10,49,0,57,59,122,0,0,18,109,0,16,10,50,0,57,
+59,122,0,0,18,109,0,16,10,51,0,57,59,122,0,0,0,0,0,1,90,95,0,0,31,221,0,116,114,97,110,115,112,111,
+115,101,0,1,0,0,0,30,109,0,0,0,1,8,58,109,97,116,52,120,51,0,18,109,0,16,8,48,0,57,59,120,0,0,18,
+109,0,16,10,49,0,57,59,120,0,0,18,109,0,16,10,50,0,57,59,120,0,0,18,109,0,16,8,48,0,57,59,121,0,0,
+18,109,0,16,10,49,0,57,59,121,0,0,18,109,0,16,10,50,0,57,59,121,0,0,18,109,0,16,8,48,0,57,59,122,0,
+0,18,109,0,16,10,49,0,57,59,122,0,0,18,109,0,16,10,50,0,57,59,122,0,0,18,109,0,16,8,48,0,57,59,119,
+0,0,18,109,0,16,10,49,0,57,59,119,0,0,18,109,0,16,10,50,0,57,59,119,0,0,0,0,0,0
diff --git a/src/mesa/shader/slang/library/slang_builtin_120_fragment_gc.h b/src/mesa/shader/slang/library/slang_builtin_120_fragment_gc.h
index 72d86fb25dd..5532678fe55 100644
--- a/src/mesa/shader/slang/library/slang_builtin_120_fragment_gc.h
+++ b/src/mesa/shader/slang/library/slang_builtin_120_fragment_gc.h
@@ -2,4 +2,4 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_builtin_120_fragment.gc */
 
-4,2,2,3,0,10,1,103,108,95,80,111,105,110,116,67,111,111,114,100,0,0,0,0
+4,2,2,90,95,3,0,10,221,1,103,108,95,80,111,105,110,116,67,111,111,114,100,0,0,0,0
diff --git a/src/mesa/shader/slang/library/slang_common_builtin.gc b/src/mesa/shader/slang/library/slang_common_builtin.gc
index 561e94f6bab..a051c53eea3 100644
--- a/src/mesa/shader/slang/library/slang_common_builtin.gc
+++ b/src/mesa/shader/slang/library/slang_common_builtin.gc
@@ -47,7 +47,6 @@ uniform mat4 gl_ModelViewProjectionMatrix;
 uniform mat4 gl_TextureMatrix[gl_MaxTextureCoords];
 
 uniform mat3 gl_NormalMatrix;
-uniform mat3 __NormalMatrixTranspose;  // Mesa only
 
 uniform mat4 gl_ModelViewMatrixInverse;
 uniform mat4 gl_ProjectionMatrixInverse;
diff --git a/src/mesa/shader/slang/library/slang_common_builtin_gc.h b/src/mesa/shader/slang/library/slang_common_builtin_gc.h
index 69da18e268e..1ee776984fb 100644
--- a/src/mesa/shader/slang/library/slang_common_builtin_gc.h
+++ b/src/mesa/shader/slang/library/slang_common_builtin_gc.h
@@ -2,823 +2,863 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_common_builtin.gc */
 
-4,2,2,1,0,5,1,103,108,95,77,97,120,76,105,103,104,116,115,0,2,16,10,56,0,0,0,2,2,1,0,5,1,103,108,
-95,77,97,120,67,108,105,112,80,108,97,110,101,115,0,2,16,10,54,0,0,0,2,2,1,0,5,1,103,108,95,77,97,
-120,84,101,120,116,117,114,101,85,110,105,116,115,0,2,16,10,56,0,0,0,2,2,1,0,5,1,103,108,95,77,97,
-120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,2,16,10,56,0,0,0,2,2,1,0,5,1,103,108,95,77,
-97,120,86,101,114,116,101,120,65,116,116,114,105,98,115,0,2,16,10,49,54,0,0,0,2,2,1,0,5,1,103,108,
-95,77,97,120,86,101,114,116,101,120,85,110,105,102,111,114,109,67,111,109,112,111,110,101,110,116,
-115,0,2,16,10,53,49,50,0,0,0,2,2,1,0,5,1,103,108,95,77,97,120,86,97,114,121,105,110,103,70,108,111,
-97,116,115,0,2,16,10,51,50,0,0,0,2,2,1,0,5,1,103,108,95,77,97,120,86,101,114,116,101,120,84,101,
-120,116,117,114,101,73,109,97,103,101,85,110,105,116,115,0,2,16,8,48,0,0,0,2,2,1,0,5,1,103,108,95,
-77,97,120,67,111,109,98,105,110,101,100,84,101,120,116,117,114,101,73,109,97,103,101,85,110,105,
-116,115,0,2,16,10,50,0,0,0,2,2,1,0,5,1,103,108,95,77,97,120,84,101,120,116,117,114,101,73,109,97,
-103,101,85,110,105,116,115,0,2,16,10,50,0,0,0,2,2,1,0,5,1,103,108,95,77,97,120,70,114,97,103,109,
-101,110,116,85,110,105,102,111,114,109,67,111,109,112,111,110,101,110,116,115,0,2,16,10,54,52,0,0,
-0,2,2,1,0,5,1,103,108,95,77,97,120,68,114,97,119,66,117,102,102,101,114,115,0,2,16,10,49,0,0,0,2,2,
-4,0,15,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,105,120,0,0,0,2,2,4,0,15,1,103,
-108,95,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,0,0,0,2,2,4,0,15,1,103,108,95,
-77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,0,0,
-0,2,2,4,0,15,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,120,0,3,18,103,108,95,77,97,
-120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,4,0,14,1,103,108,95,78,111,114,109,
-97,108,77,97,116,114,105,120,0,0,0,2,2,4,0,14,1,95,95,78,111,114,109,97,108,77,97,116,114,105,120,
-84,114,97,110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,77,111,100,101,108,86,105,101,119,
-77,97,116,114,105,120,73,110,118,101,114,115,101,0,0,0,2,2,4,0,15,1,103,108,95,80,114,111,106,101,
-99,116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,0,0,0,2,2,4,0,15,1,103,108,95,
-77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,73,
-110,118,101,114,115,101,0,0,0,2,2,4,0,15,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,
-120,73,110,118,101,114,115,101,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,
-114,100,115,0,0,0,2,2,4,0,15,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,105,120,
-84,114,97,110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,80,114,111,106,101,99,116,105,111,
-110,77,97,116,114,105,120,84,114,97,110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,77,111,
-100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,84,114,97,
-110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,
-120,84,114,97,110,115,112,111,115,101,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,
-111,111,114,100,115,0,0,0,2,2,4,0,15,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,
-105,120,73,110,118,101,114,115,101,84,114,97,110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,
-80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,84,114,97,
-110,115,112,111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,
-106,101,99,116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,84,114,97,110,115,112,
-111,115,101,0,0,0,2,2,4,0,15,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,120,73,110,
-118,101,114,115,101,84,114,97,110,115,112,111,115,101,0,3,18,103,108,95,77,97,120,84,101,120,116,
-117,114,101,67,111,111,114,100,115,0,0,0,2,2,4,0,9,1,103,108,95,78,111,114,109,97,108,83,99,97,108,
-101,0,0,0,2,2,0,0,24,103,108,95,68,101,112,116,104,82,97,110,103,101,80,97,114,97,109,101,116,101,
-114,115,0,9,110,101,97,114,0,0,0,1,9,102,97,114,0,0,0,1,9,100,105,102,102,0,0,0,0,0,0,2,2,4,0,25,
-103,108,95,68,101,112,116,104,82,97,110,103,101,80,97,114,97,109,101,116,101,114,115,0,1,103,108,
-95,68,101,112,116,104,82,97,110,103,101,0,0,0,2,2,4,0,12,1,103,108,95,67,108,105,112,80,108,97,110,
-101,0,3,18,103,108,95,77,97,120,67,108,105,112,80,108,97,110,101,115,0,0,0,2,2,0,0,24,103,108,95,
-80,111,105,110,116,80,97,114,97,109,101,116,101,114,115,0,9,115,105,122,101,0,0,0,1,9,115,105,122,
-101,77,105,110,0,0,0,1,9,115,105,122,101,77,97,120,0,0,0,1,9,102,97,100,101,84,104,114,101,115,104,
-111,108,100,83,105,122,101,0,0,0,1,9,100,105,115,116,97,110,99,101,67,111,110,115,116,97,110,116,
-65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,100,105,115,116,97,110,99,101,76,105,110,101,
-97,114,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,100,105,115,116,97,110,99,101,81,117,97,
-100,114,97,116,105,99,65,116,116,101,110,117,97,116,105,111,110,0,0,0,0,0,0,2,2,4,0,25,103,108,95,
-80,111,105,110,116,80,97,114,97,109,101,116,101,114,115,0,1,103,108,95,80,111,105,110,116,0,0,0,2,
-2,0,0,24,103,108,95,77,97,116,101,114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,12,101,109,
-105,115,115,105,111,110,0,0,0,1,12,97,109,98,105,101,110,116,0,0,0,1,12,100,105,102,102,117,115,
-101,0,0,0,1,12,115,112,101,99,117,108,97,114,0,0,0,1,9,115,104,105,110,105,110,101,115,115,0,0,0,0,
-0,0,2,2,4,0,25,103,108,95,77,97,116,101,114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,1,
-103,108,95,70,114,111,110,116,77,97,116,101,114,105,97,108,0,0,0,2,2,4,0,25,103,108,95,77,97,116,
-101,114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,1,103,108,95,66,97,99,107,77,97,116,101,
-114,105,97,108,0,0,0,2,2,0,0,24,103,108,95,76,105,103,104,116,83,111,117,114,99,101,80,97,114,97,
-109,101,116,101,114,115,0,12,97,109,98,105,101,110,116,0,0,0,1,12,100,105,102,102,117,115,101,0,0,
-0,1,12,115,112,101,99,117,108,97,114,0,0,0,1,12,112,111,115,105,116,105,111,110,0,0,0,1,12,104,97,
-108,102,86,101,99,116,111,114,0,0,0,1,11,115,112,111,116,68,105,114,101,99,116,105,111,110,0,0,0,1,
-9,115,112,111,116,69,120,112,111,110,101,110,116,0,0,0,1,9,115,112,111,116,67,117,116,111,102,102,
-0,0,0,1,9,115,112,111,116,67,111,115,67,117,116,111,102,102,0,0,0,1,9,99,111,110,115,116,97,110,
-116,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,108,105,110,101,97,114,65,116,116,101,110,
-117,97,116,105,111,110,0,0,0,1,9,113,117,97,100,114,97,116,105,99,65,116,116,101,110,117,97,116,
-105,111,110,0,0,0,0,0,0,2,2,4,0,25,103,108,95,76,105,103,104,116,83,111,117,114,99,101,80,97,114,
-97,109,101,116,101,114,115,0,1,103,108,95,76,105,103,104,116,83,111,117,114,99,101,0,3,18,103,108,
-95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,0,0,24,103,108,95,76,105,103,104,116,77,111,100,101,
-108,80,97,114,97,109,101,116,101,114,115,0,12,97,109,98,105,101,110,116,0,0,0,0,0,0,2,2,4,0,25,103,
-108,95,76,105,103,104,116,77,111,100,101,108,80,97,114,97,109,101,116,101,114,115,0,1,103,108,95,
-76,105,103,104,116,77,111,100,101,108,0,0,0,2,2,0,0,24,103,108,95,76,105,103,104,116,77,111,100,
-101,108,80,114,111,100,117,99,116,115,0,12,115,99,101,110,101,67,111,108,111,114,0,0,0,0,0,0,2,2,4,
-0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,1,103,108,95,
-70,114,111,110,116,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,2,4,0,
-25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,1,103,108,95,
-66,97,99,107,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,2,0,0,24,103,
-108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,12,97,109,98,105,101,110,116,0,0,0,1,12,
-100,105,102,102,117,115,101,0,0,0,1,12,115,112,101,99,117,108,97,114,0,0,0,0,0,0,2,2,4,0,25,103,
-108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,1,103,108,95,70,114,111,110,116,76,105,
-103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,
-4,0,25,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,1,103,108,95,66,97,99,107,76,
-105,103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,97,120,76,105,103,104,116,115,0,0,0,
-2,2,4,0,12,1,103,108,95,84,101,120,116,117,114,101,69,110,118,67,111,108,111,114,0,3,18,103,108,95,
-77,97,120,84,101,120,116,117,114,101,73,109,97,103,101,85,110,105,116,115,0,0,0,2,2,4,0,12,1,103,
-108,95,69,121,101,80,108,97,110,101,83,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,
-111,111,114,100,115,0,0,0,2,2,4,0,12,1,103,108,95,69,121,101,80,108,97,110,101,84,0,3,18,103,108,
-95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,4,0,12,1,103,108,95,69,
-121,101,80,108,97,110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,
-100,115,0,0,0,2,2,4,0,12,1,103,108,95,69,121,101,80,108,97,110,101,81,0,3,18,103,108,95,77,97,120,
-84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,4,0,12,1,103,108,95,79,98,106,101,99,
-116,80,108,97,110,101,83,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,
-115,0,0,0,2,2,4,0,12,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,84,0,3,18,103,108,95,77,
-97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,4,0,12,1,103,108,95,79,98,106,
-101,99,116,80,108,97,110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,
-114,100,115,0,0,0,2,2,4,0,12,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,81,0,3,18,103,108,
-95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,0,0,24,103,108,95,70,111,
-103,80,97,114,97,109,101,116,101,114,115,0,12,99,111,108,111,114,0,0,0,1,9,100,101,110,115,105,116,
-121,0,0,0,1,9,115,116,97,114,116,0,0,0,1,9,101,110,100,0,0,0,1,9,115,99,97,108,101,0,0,0,0,0,0,2,2,
-4,0,25,103,108,95,70,111,103,80,97,114,97,109,101,116,101,114,115,0,1,103,108,95,70,111,103,0,0,0,
-1,0,0,9,0,114,97,100,105,97,110,115,0,1,1,0,0,9,100,101,103,0,0,0,1,3,2,1,0,9,1,99,0,2,17,51,0,49,
-52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,
-121,0,18,95,95,114,101,116,86,97,108,0,0,18,100,101,103,0,0,18,99,0,0,0,0,1,0,0,10,0,114,97,100,
-105,97,110,115,0,1,1,0,0,10,100,101,103,0,0,0,1,3,2,1,0,9,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,
-0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,100,101,103,0,59,120,121,0,0,18,99,0,59,120,120,0,0,0,0,1,0,
-0,11,0,114,97,100,105,97,110,115,0,1,1,0,0,11,100,101,103,0,0,0,1,3,2,1,0,9,1,99,0,2,17,51,0,49,52,
-49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,100,101,103,0,59,120,121,122,0,0,18,99,0,
-59,120,120,120,0,0,0,0,1,0,0,12,0,114,97,100,105,97,110,115,0,1,1,0,0,12,100,101,103,0,0,0,1,3,2,1,
-0,9,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,
-117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,100,101,103,0,0,18,99,0,59,120,
-120,120,120,0,0,0,0,1,0,0,9,0,100,101,103,114,101,101,115,0,1,1,0,0,9,114,97,100,0,0,0,1,3,2,1,0,9,
-1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,118,101,99,52,95,109,117,
-108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,100,0,0,18,99,0,0,0,0,1,0,0,
-10,0,100,101,103,114,101,101,115,0,1,1,0,0,10,114,97,100,0,0,0,1,3,2,1,0,9,1,99,0,2,17,49,56,48,0,
-48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,114,97,100,0,59,120,121,0,0,18,99,0,59,120,
-120,0,0,0,0,1,0,0,11,0,100,101,103,114,101,101,115,0,1,1,0,0,11,114,97,100,0,0,0,1,3,2,1,0,9,1,99,
+4,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,76,105,103,104,116,115,0,2,16,10,56,0,0,0,2,2,90,95,1,
+0,5,221,1,103,108,95,77,97,120,67,108,105,112,80,108,97,110,101,115,0,2,16,10,54,0,0,0,2,2,90,95,1,
+0,5,221,1,103,108,95,77,97,120,84,101,120,116,117,114,101,85,110,105,116,115,0,2,16,10,56,0,0,0,2,
+2,90,95,1,0,5,221,1,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,2,16,
+10,56,0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,86,101,114,116,101,120,65,116,116,114,105,
+98,115,0,2,16,10,49,54,0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,86,101,114,116,101,120,85,
+110,105,102,111,114,109,67,111,109,112,111,110,101,110,116,115,0,2,16,10,53,49,50,0,0,0,2,2,90,95,
+1,0,5,221,1,103,108,95,77,97,120,86,97,114,121,105,110,103,70,108,111,97,116,115,0,2,16,10,51,50,0,
+0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,86,101,114,116,101,120,84,101,120,116,117,114,101,
+73,109,97,103,101,85,110,105,116,115,0,2,16,8,48,0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,
+67,111,109,98,105,110,101,100,84,101,120,116,117,114,101,73,109,97,103,101,85,110,105,116,115,0,2,
+16,10,50,0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,84,101,120,116,117,114,101,73,109,97,103,
+101,85,110,105,116,115,0,2,16,10,50,0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,70,114,97,103,
+109,101,110,116,85,110,105,102,111,114,109,67,111,109,112,111,110,101,110,116,115,0,2,16,10,54,52,
+0,0,0,2,2,90,95,1,0,5,221,1,103,108,95,77,97,120,68,114,97,119,66,117,102,102,101,114,115,0,2,16,
+10,49,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,105,
+120,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,80,114,111,106,101,99,116,105,111,110,77,97,116,114,
+105,120,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,
+101,99,116,105,111,110,77,97,116,114,105,120,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,84,101,120,
+116,117,114,101,77,97,116,114,105,120,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,
+111,111,114,100,115,0,0,0,2,2,90,95,4,0,14,221,1,103,108,95,78,111,114,109,97,108,77,97,116,114,
+105,120,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,
+105,120,73,110,118,101,114,115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,80,114,111,106,101,99,
+116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,
+108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,
+120,73,110,118,101,114,115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,84,101,120,116,117,114,101,
+77,97,116,114,105,120,73,110,118,101,114,115,101,0,3,18,103,108,95,77,97,120,84,101,120,116,117,
+114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,
+101,119,77,97,116,114,105,120,84,114,97,110,115,112,111,115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,
+108,95,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,120,84,114,97,110,115,112,111,115,
+101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,
+99,116,105,111,110,77,97,116,114,105,120,84,114,97,110,115,112,111,115,101,0,0,0,2,2,90,95,4,0,15,
+221,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,120,84,114,97,110,115,112,111,115,
+101,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,
+4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,77,97,116,114,105,120,73,110,118,101,114,
+115,101,84,114,97,110,115,112,111,115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,80,114,111,106,
+101,99,116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,84,114,97,110,115,112,111,
+115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,
+101,99,116,105,111,110,77,97,116,114,105,120,73,110,118,101,114,115,101,84,114,97,110,115,112,111,
+115,101,0,0,0,2,2,90,95,4,0,15,221,1,103,108,95,84,101,120,116,117,114,101,77,97,116,114,105,120,
+73,110,118,101,114,115,101,84,114,97,110,115,112,111,115,101,0,3,18,103,108,95,77,97,120,84,101,
+120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,9,221,1,103,108,95,78,111,114,109,
+97,108,83,99,97,108,101,0,0,0,2,2,90,95,0,0,24,103,108,95,68,101,112,116,104,82,97,110,103,101,80,
+97,114,97,109,101,116,101,114,115,0,9,110,101,97,114,0,0,0,1,9,102,97,114,0,0,0,1,9,100,105,102,
+102,0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,68,101,112,116,104,82,97,110,103,101,80,97,114,97,
+109,101,116,101,114,115,0,221,1,103,108,95,68,101,112,116,104,82,97,110,103,101,0,0,0,2,2,90,95,4,
+0,12,221,1,103,108,95,67,108,105,112,80,108,97,110,101,0,3,18,103,108,95,77,97,120,67,108,105,112,
+80,108,97,110,101,115,0,0,0,2,2,90,95,0,0,24,103,108,95,80,111,105,110,116,80,97,114,97,109,101,
+116,101,114,115,0,9,115,105,122,101,0,0,0,1,9,115,105,122,101,77,105,110,0,0,0,1,9,115,105,122,101,
+77,97,120,0,0,0,1,9,102,97,100,101,84,104,114,101,115,104,111,108,100,83,105,122,101,0,0,0,1,9,100,
+105,115,116,97,110,99,101,67,111,110,115,116,97,110,116,65,116,116,101,110,117,97,116,105,111,110,
+0,0,0,1,9,100,105,115,116,97,110,99,101,76,105,110,101,97,114,65,116,116,101,110,117,97,116,105,
+111,110,0,0,0,1,9,100,105,115,116,97,110,99,101,81,117,97,100,114,97,116,105,99,65,116,116,101,110,
+117,97,116,105,111,110,0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,80,111,105,110,116,80,97,114,97,
+109,101,116,101,114,115,0,221,1,103,108,95,80,111,105,110,116,0,0,0,2,2,90,95,0,0,24,103,108,95,77,
+97,116,101,114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,12,101,109,105,115,115,105,111,
+110,0,0,0,1,12,97,109,98,105,101,110,116,0,0,0,1,12,100,105,102,102,117,115,101,0,0,0,1,12,115,112,
+101,99,117,108,97,114,0,0,0,1,9,115,104,105,110,105,110,101,115,115,0,0,0,0,221,0,0,2,2,90,95,4,0,
+25,103,108,95,77,97,116,101,114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,221,1,103,108,95,
+70,114,111,110,116,77,97,116,101,114,105,97,108,0,0,0,2,2,90,95,4,0,25,103,108,95,77,97,116,101,
+114,105,97,108,80,97,114,97,109,101,116,101,114,115,0,221,1,103,108,95,66,97,99,107,77,97,116,101,
+114,105,97,108,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,83,111,117,114,99,101,80,97,
+114,97,109,101,116,101,114,115,0,12,97,109,98,105,101,110,116,0,0,0,1,12,100,105,102,102,117,115,
+101,0,0,0,1,12,115,112,101,99,117,108,97,114,0,0,0,1,12,112,111,115,105,116,105,111,110,0,0,0,1,12,
+104,97,108,102,86,101,99,116,111,114,0,0,0,1,11,115,112,111,116,68,105,114,101,99,116,105,111,110,
+0,0,0,1,9,115,112,111,116,69,120,112,111,110,101,110,116,0,0,0,1,9,115,112,111,116,67,117,116,111,
+102,102,0,0,0,1,9,115,112,111,116,67,111,115,67,117,116,111,102,102,0,0,0,1,9,99,111,110,115,116,
+97,110,116,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,108,105,110,101,97,114,65,116,116,
+101,110,117,97,116,105,111,110,0,0,0,1,9,113,117,97,100,114,97,116,105,99,65,116,116,101,110,117,
+97,116,105,111,110,0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,83,111,117,114,
+99,101,80,97,114,97,109,101,116,101,114,115,0,221,1,103,108,95,76,105,103,104,116,83,111,117,114,
+99,101,0,3,18,103,108,95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,
+103,104,116,77,111,100,101,108,80,97,114,97,109,101,116,101,114,115,0,12,97,109,98,105,101,110,116,
+0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,97,114,97,109,
+101,116,101,114,115,0,221,1,103,108,95,76,105,103,104,116,77,111,100,101,108,0,0,0,2,2,90,95,0,0,
+24,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,12,115,99,101,
+110,101,67,111,108,111,114,0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,77,111,
+100,101,108,80,114,111,100,117,99,116,115,0,221,1,103,108,95,70,114,111,110,116,76,105,103,104,116,
+77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,
+77,111,100,101,108,80,114,111,100,117,99,116,115,0,221,1,103,108,95,66,97,99,107,76,105,103,104,
+116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,
+116,80,114,111,100,117,99,116,115,0,12,97,109,98,105,101,110,116,0,0,0,1,12,100,105,102,102,117,
+115,101,0,0,0,1,12,115,112,101,99,117,108,97,114,0,0,0,0,221,0,0,2,2,90,95,4,0,25,103,108,95,76,
+105,103,104,116,80,114,111,100,117,99,116,115,0,221,1,103,108,95,70,114,111,110,116,76,105,103,104,
+116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,90,95,4,
+0,25,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,221,1,103,108,95,66,97,99,107,
+76,105,103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,97,120,76,105,103,104,116,115,0,
+0,0,2,2,90,95,4,0,12,221,1,103,108,95,84,101,120,116,117,114,101,69,110,118,67,111,108,111,114,0,3,
+18,103,108,95,77,97,120,84,101,120,116,117,114,101,73,109,97,103,101,85,110,105,116,115,0,0,0,2,2,
+90,95,4,0,12,221,1,103,108,95,69,121,101,80,108,97,110,101,83,0,3,18,103,108,95,77,97,120,84,101,
+120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,221,1,103,108,95,69,121,101,80,
+108,97,110,101,84,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,
+0,0,2,2,90,95,4,0,12,221,1,103,108,95,69,121,101,80,108,97,110,101,82,0,3,18,103,108,95,77,97,120,
+84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,221,1,103,108,95,69,121,
+101,80,108,97,110,101,81,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,
+115,0,0,0,2,2,90,95,4,0,12,221,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,83,0,3,18,103,
+108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,221,1,
+103,108,95,79,98,106,101,99,116,80,108,97,110,101,84,0,3,18,103,108,95,77,97,120,84,101,120,116,
+117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,221,1,103,108,95,79,98,106,101,99,116,80,
+108,97,110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,
+0,0,2,2,90,95,4,0,12,221,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,81,0,3,18,103,108,95,
+77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,0,0,24,103,108,95,70,
+111,103,80,97,114,97,109,101,116,101,114,115,0,12,99,111,108,111,114,0,0,0,1,9,100,101,110,115,105,
+116,121,0,0,0,1,9,115,116,97,114,116,0,0,0,1,9,101,110,100,0,0,0,1,9,115,99,97,108,101,0,0,0,0,221,
+0,0,2,2,90,95,4,0,25,103,108,95,70,111,103,80,97,114,97,109,101,116,101,114,115,0,221,1,103,108,95,
+70,111,103,0,0,0,1,90,95,0,0,9,221,0,114,97,100,105,97,110,115,0,1,1,0,0,9,100,101,103,0,0,0,1,3,2,
+90,95,1,0,9,221,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,
+52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,100,101,103,0,0,18,
+99,0,0,0,0,1,90,95,0,0,10,221,0,114,97,100,105,97,110,115,0,1,1,0,0,10,100,101,103,0,0,0,1,3,2,90,
+95,1,0,9,221,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,
+95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,100,101,
+103,0,59,120,121,0,0,18,99,0,59,120,120,0,0,0,0,1,90,95,0,0,11,221,0,114,97,100,105,97,110,115,0,1,
+1,0,0,11,100,101,103,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,
+56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,
+97,108,0,59,120,121,122,0,0,18,100,101,103,0,59,120,121,122,0,0,18,99,0,59,120,120,120,0,0,0,0,1,
+90,95,0,0,12,221,0,114,97,100,105,97,110,115,0,1,1,0,0,12,100,101,103,0,0,0,1,3,2,90,95,1,0,9,221,
+1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,
+108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,100,101,103,0,0,18,99,0,59,120,120,
+120,120,0,0,0,0,1,90,95,0,0,9,221,0,100,101,103,114,101,101,115,0,1,1,0,0,9,114,97,100,0,0,0,1,3,2,
+90,95,1,0,9,221,1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,118,101,99,
+52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,100,0,0,18,99,
+0,0,0,0,1,90,95,0,0,10,221,0,100,101,103,114,101,101,115,0,1,1,0,0,10,114,97,100,0,0,0,1,3,2,90,95,
+1,0,9,221,1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,118,101,99,52,95,
+109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,114,97,100,0,
+59,120,121,0,0,18,99,0,59,120,120,0,0,0,0,1,90,95,0,0,11,221,0,100,101,103,114,101,101,115,0,1,1,0,
+0,11,114,97,100,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,
+50,54,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,
+108,0,59,120,121,122,0,0,18,114,97,100,0,59,120,121,122,0,0,18,99,0,59,120,120,120,0,0,0,0,1,90,95,
+0,0,12,221,0,100,101,103,114,101,101,115,0,1,1,0,0,12,114,97,100,0,0,0,1,3,2,90,95,1,0,9,221,1,99,
 0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,118,101,99,52,95,109,117,108,
-116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,114,97,100,0,59,120,
-121,122,0,0,18,99,0,59,120,120,120,0,0,0,0,1,0,0,12,0,100,101,103,114,101,101,115,0,1,1,0,0,12,114,
-97,100,0,0,0,1,3,2,1,0,9,1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,0,49,0,0,4,
-118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,
-100,0,0,18,99,0,59,120,120,120,120,0,0,0,0,1,0,0,9,0,115,105,110,0,1,1,0,0,9,114,97,100,105,97,110,
-115,0,0,0,1,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,
-100,105,97,110,115,0,0,0,0,1,0,0,10,0,115,105,110,0,1,1,0,0,10,114,97,100,105,97,110,115,0,0,0,1,4,
-102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,97,100,
-105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,
-108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,0,1,0,0,11,0,115,105,110,0,1,1,0,0,11,
-114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,
-86,97,108,0,59,120,0,0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,115,105,
-110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,
-4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,97,
-100,105,97,110,115,0,59,122,0,0,0,0,1,0,0,12,0,115,105,110,0,1,1,0,0,12,114,97,100,105,97,110,115,
-0,0,0,1,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
-114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,
-101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,4,102,108,111,97,116,95,
-115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,97,100,105,97,110,115,0,59,
-122,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,
-114,97,100,105,97,110,115,0,59,119,0,0,0,0,1,0,0,9,0,99,111,115,0,1,1,0,0,9,114,97,100,105,97,110,
-115,0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,
-114,97,100,105,97,110,115,0,0,0,0,1,0,0,10,0,99,111,115,0,1,1,0,0,10,114,97,100,105,97,110,115,0,0,
-0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,
-18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,
-95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,0,1,0,0,11,0,99,
-111,115,0,1,1,0,0,11,114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,
-101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,
-102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,
-97,100,105,97,110,115,0,59,121,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,
-101,116,86,97,108,0,59,122,0,0,18,114,97,100,105,97,110,115,0,59,122,0,0,0,0,1,0,0,12,0,99,111,115,
-0,1,1,0,0,12,114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,
+116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,100,0,0,18,99,0,59,120,120,120,
+120,0,0,0,0,1,90,95,0,0,9,221,0,115,105,110,0,1,1,0,0,9,114,97,100,105,97,110,115,0,0,0,1,4,102,
+108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,114,97,100,105,97,110,
+115,0,0,0,0,1,90,95,0,0,10,221,0,115,105,110,0,1,1,0,0,10,114,97,100,105,97,110,115,0,0,0,1,4,102,
+108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,97,100,105,
+97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,
+0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,0,1,90,95,0,0,11,221,0,115,105,110,0,1,1,
+0,0,11,114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,
+101,116,86,97,108,0,59,120,0,0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,
+115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,
+121,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,
+114,97,100,105,97,110,115,0,59,122,0,0,0,0,1,90,95,0,0,12,221,0,115,105,110,0,1,1,0,0,12,114,97,
+100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,
+108,0,59,120,0,0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,115,105,110,
+101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,4,
+102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,97,100,
+105,97,110,115,0,59,122,0,0,0,4,102,108,111,97,116,95,115,105,110,101,0,18,95,95,114,101,116,86,97,
+108,0,59,119,0,0,18,114,97,100,105,97,110,115,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,99,111,115,0,1,
+1,0,0,9,114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,
+95,114,101,116,86,97,108,0,0,18,114,97,100,105,97,110,115,0,0,0,0,1,90,95,0,0,10,221,0,99,111,115,
+0,1,1,0,0,10,114,97,100,105,97,110,115,0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,
 95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,
 97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,
-97,110,115,0,59,121,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,
-97,108,0,59,122,0,0,18,114,97,100,105,97,110,115,0,59,122,0,0,0,4,102,108,111,97,116,95,99,111,115,
-105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,114,97,100,105,97,110,115,0,59,119,0,
-0,0,0,1,0,0,9,0,116,97,110,0,1,1,0,0,9,97,110,103,108,101,0,0,0,1,3,2,1,0,9,1,115,0,2,58,115,105,
-110,0,18,97,110,103,108,101,0,0,0,0,0,3,2,1,0,9,1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,
-0,0,0,8,18,115,0,18,99,0,49,0,0,1,0,0,10,0,116,97,110,0,1,1,0,0,10,97,110,103,108,101,0,0,0,1,3,2,
-1,0,10,1,115,0,2,58,115,105,110,0,18,97,110,103,108,101,0,0,0,0,0,3,2,1,0,10,1,99,0,2,58,99,111,
-115,0,18,97,110,103,108,101,0,0,0,0,0,8,18,115,0,18,99,0,49,0,0,1,0,0,11,0,116,97,110,0,1,1,0,0,11,
-97,110,103,108,101,0,0,0,1,3,2,1,0,11,1,115,0,2,58,115,105,110,0,18,97,110,103,108,101,0,0,0,0,0,3,
-2,1,0,11,1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,0,0,0,8,18,115,0,18,99,0,49,0,0,1,0,0,
-12,0,116,97,110,0,1,1,0,0,12,97,110,103,108,101,0,0,0,1,3,2,1,0,12,1,115,0,2,58,115,105,110,0,18,
-97,110,103,108,101,0,0,0,0,0,3,2,1,0,12,1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,0,0,0,8,
-18,115,0,18,99,0,49,0,0,1,0,0,9,0,97,115,105,110,0,1,1,0,0,9,120,0,0,0,1,3,2,1,0,9,1,97,48,0,2,17,
-49,0,53,55,48,55,50,56,56,0,0,0,0,3,2,1,0,9,1,97,49,0,2,17,48,0,50,49,50,49,49,52,52,0,0,54,0,0,3,
-2,1,0,9,1,97,50,0,2,17,48,0,48,55,52,50,54,49,48,0,0,0,0,3,2,1,0,9,1,104,97,108,102,80,105,0,2,17,
-51,0,49,52,49,53,57,50,54,0,0,17,48,0,53,0,0,48,0,0,3,2,1,0,9,1,121,0,2,58,97,98,115,0,18,120,0,0,
-0,0,0,9,18,95,95,114,101,116,86,97,108,0,18,104,97,108,102,80,105,0,58,115,113,114,116,0,17,49,0,
-48,0,0,18,121,0,47,0,0,18,97,48,0,18,121,0,18,97,49,0,18,97,50,0,18,121,0,48,46,48,46,48,47,58,115,
-105,103,110,0,18,120,0,0,0,48,20,0,0,1,0,0,10,0,97,115,105,110,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,59,120,0,58,97,115,105,110,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,121,0,58,97,115,105,110,0,18,118,0,59,121,0,0,0,20,0,0,1,0,0,11,0,97,115,105,
-110,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,115,105,110,0,18,
-118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,115,105,110,0,18,118,0,
-59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,115,105,110,0,18,118,0,59,122,
-0,0,0,20,0,0,1,0,0,12,0,97,115,105,110,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-59,120,0,58,97,115,105,110,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,
-0,58,97,115,105,110,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,
-115,105,110,0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,115,105,
-110,0,18,118,0,59,119,0,0,0,20,0,0,1,0,0,9,0,97,99,111,115,0,1,1,0,0,9,120,0,0,0,1,3,2,1,0,9,1,104,
+97,110,115,0,59,121,0,0,0,0,1,90,95,0,0,11,221,0,99,111,115,0,1,1,0,0,11,114,97,100,105,97,110,115,
+0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,
+0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,
+95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,4,102,108,111,
+97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,97,100,105,
+97,110,115,0,59,122,0,0,0,0,1,90,95,0,0,12,221,0,99,111,115,0,1,1,0,0,12,114,97,100,105,97,110,115,
+0,0,0,1,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,
+0,18,114,97,100,105,97,110,115,0,59,120,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,
+95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,97,100,105,97,110,115,0,59,121,0,0,0,4,102,108,111,
+97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,97,100,105,
+97,110,115,0,59,122,0,0,0,4,102,108,111,97,116,95,99,111,115,105,110,101,0,18,95,95,114,101,116,86,
+97,108,0,59,119,0,0,18,114,97,100,105,97,110,115,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,116,97,110,0,
+1,1,0,0,9,97,110,103,108,101,0,0,0,1,3,2,90,95,1,0,9,221,1,115,0,2,58,115,105,110,0,18,97,110,103,
+108,101,0,0,0,0,0,3,2,90,95,1,0,9,221,1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,0,0,0,8,
+18,115,0,18,99,0,49,0,0,1,90,95,0,0,10,221,0,116,97,110,0,1,1,0,0,10,97,110,103,108,101,0,0,0,1,3,
+2,90,95,1,0,10,221,1,115,0,2,58,115,105,110,0,18,97,110,103,108,101,0,0,0,0,0,3,2,90,95,1,0,10,221,
+1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,0,0,0,8,18,115,0,18,99,0,49,0,0,1,90,95,0,0,11,
+221,0,116,97,110,0,1,1,0,0,11,97,110,103,108,101,0,0,0,1,3,2,90,95,1,0,11,221,1,115,0,2,58,115,105,
+110,0,18,97,110,103,108,101,0,0,0,0,0,3,2,90,95,1,0,11,221,1,99,0,2,58,99,111,115,0,18,97,110,103,
+108,101,0,0,0,0,0,8,18,115,0,18,99,0,49,0,0,1,90,95,0,0,12,221,0,116,97,110,0,1,1,0,0,12,97,110,
+103,108,101,0,0,0,1,3,2,90,95,1,0,12,221,1,115,0,2,58,115,105,110,0,18,97,110,103,108,101,0,0,0,0,
+0,3,2,90,95,1,0,12,221,1,99,0,2,58,99,111,115,0,18,97,110,103,108,101,0,0,0,0,0,8,18,115,0,18,99,0,
+49,0,0,1,90,95,0,0,9,221,0,97,115,105,110,0,1,1,0,0,9,120,0,0,0,1,3,2,90,95,1,0,9,221,1,97,48,0,2,
+17,49,0,53,55,48,55,50,56,56,0,0,0,0,3,2,90,95,1,0,9,221,1,97,49,0,2,17,48,0,50,49,50,49,49,52,52,
+0,0,54,0,0,3,2,90,95,1,0,9,221,1,97,50,0,2,17,48,0,48,55,52,50,54,49,48,0,0,0,0,3,2,90,95,1,0,9,
+221,1,104,97,108,102,80,105,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,48,0,53,0,0,48,0,0,3,2,90,95,1,
+0,9,221,1,121,0,2,58,97,98,115,0,18,120,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,18,104,97,108,
+102,80,105,0,58,115,113,114,116,0,17,49,0,48,0,0,18,121,0,47,0,0,18,97,48,0,18,121,0,18,97,49,0,18,
+97,50,0,18,121,0,48,46,48,46,48,47,58,115,105,103,110,0,18,120,0,0,0,48,20,0,0,1,90,95,0,0,10,221,
+0,97,115,105,110,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,115,
+105,110,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,115,105,110,
+0,18,118,0,59,121,0,0,0,20,0,0,1,90,95,0,0,11,221,0,97,115,105,110,0,1,1,0,0,11,118,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,59,120,0,58,97,115,105,110,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,121,0,58,97,115,105,110,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,59,122,0,58,97,115,105,110,0,18,118,0,59,122,0,0,0,20,0,0,1,90,95,0,0,12,221,0,97,
+115,105,110,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,115,105,110,
+0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,115,105,110,0,18,
+118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,115,105,110,0,18,118,0,
+59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,115,105,110,0,18,118,0,59,119,
+0,0,0,20,0,0,1,90,95,0,0,9,221,0,97,99,111,115,0,1,1,0,0,9,120,0,0,0,1,3,2,90,95,1,0,9,221,1,104,
 97,108,102,80,105,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,48,0,53,0,0,48,0,0,9,18,95,95,114,101,
-116,86,97,108,0,18,104,97,108,102,80,105,0,58,97,115,105,110,0,18,120,0,0,0,47,20,0,0,1,0,0,10,0,
-97,99,111,115,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,
-115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,
-118,0,59,121,0,0,0,20,0,0,1,0,0,11,0,97,99,111,115,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,
-122,0,58,97,99,111,115,0,18,118,0,59,122,0,0,0,20,0,0,1,0,0,12,0,97,99,111,115,0,1,1,0,0,12,118,0,
-0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,
-18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,59,122,0,58,97,99,111,115,0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,119,0,58,97,99,111,115,0,18,118,0,59,119,0,0,0,20,0,0,1,0,0,9,0,97,116,97,110,0,
-1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,97,115,105,110,0,18,120,0,58,105,110,
-118,101,114,115,101,115,113,114,116,0,18,120,0,18,120,0,48,17,49,0,48,0,0,46,0,0,48,0,0,20,0,0,1,0,
-0,10,0,97,116,97,110,0,1,1,0,0,10,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,
-95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,
-0,0,20,0,0,1,0,0,11,0,97,116,97,110,0,1,1,0,0,11,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,
-0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,
-120,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,
-111,118,101,114,95,120,0,59,122,0,0,0,20,0,0,1,0,0,12,0,97,116,97,110,0,1,1,0,0,12,121,95,111,118,
+116,86,97,108,0,18,104,97,108,102,80,105,0,58,97,115,105,110,0,18,120,0,0,0,47,20,0,0,1,90,95,0,0,
+10,221,0,97,99,111,115,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,
+99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,
+115,0,18,118,0,59,121,0,0,0,20,0,0,1,90,95,0,0,11,221,0,97,99,111,115,0,1,1,0,0,11,118,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,115,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,59,122,0,58,97,99,111,115,0,18,118,0,59,122,0,0,0,20,0,0,1,90,95,0,0,12,221,0,97,
+99,111,115,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,99,111,115,0,
+18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,99,111,115,0,18,118,0,
+59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,99,111,115,0,18,118,0,59,122,0,
+0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,99,111,115,0,18,118,0,59,119,0,0,0,20,0,
+0,1,90,95,0,0,9,221,0,97,116,97,110,0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
+97,115,105,110,0,18,120,0,58,105,110,118,101,114,115,101,115,113,114,116,0,18,120,0,18,120,0,48,17,
+49,0,48,0,0,46,0,0,48,0,0,20,0,0,1,90,95,0,0,10,221,0,97,116,97,110,0,1,1,0,0,10,121,95,111,118,
 101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,
 111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,
-97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
-59,122,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,122,0,0,0,20,0,9,18,95,95,114,
-101,116,86,97,108,0,59,119,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,119,0,0,0,20,
-0,0,1,0,0,9,0,97,116,97,110,0,1,1,0,0,9,121,0,0,1,1,0,0,9,120,0,0,0,1,3,2,0,0,9,1,114,0,0,0,10,58,
-97,98,115,0,18,120,0,0,0,17,49,0,48,0,45,52,0,41,0,2,9,18,114,0,58,97,116,97,110,0,18,121,0,18,120,
-0,49,0,0,20,0,10,18,120,0,17,48,0,48,0,0,40,0,2,9,18,114,0,18,114,0,58,115,105,103,110,0,18,121,0,
-0,0,17,51,0,49,52,49,53,57,51,0,0,48,46,20,0,0,9,14,0,0,2,9,18,114,0,58,115,105,103,110,0,18,121,0,
-0,0,17,49,0,53,55,48,55,57,54,53,0,0,48,20,0,0,8,18,114,0,0,0,1,0,0,10,0,97,116,97,110,0,1,1,0,0,
-10,117,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,
-18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,
-116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,0,1,0,0,11,0,97,116,97,110,0,1,1,0,0,
-11,117,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,
-18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,
-116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,
-122,0,58,97,116,97,110,0,18,117,0,59,122,0,0,18,118,0,59,122,0,0,0,20,0,0,1,0,0,12,0,97,116,97,110,
-0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,
-97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,
-0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0,0,18,118,0,59,122,0,0,0,20,0,9,18,95,95,114,
-101,116,86,97,108,0,59,119,0,58,97,116,97,110,0,18,117,0,59,119,0,0,18,118,0,59,119,0,0,0,20,0,0,1,
-0,0,9,0,112,111,119,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,102,108,111,97,116,95,112,111,119,
-101,114,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,10,0,112,111,119,0,1,1,
-0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,
-116,86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,112,111,
-119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,18,98,0,59,121,0,0,0,
-0,1,0,0,11,0,112,111,119,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,0,0,1,4,102,108,111,97,116,95,112,111,
+97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,0,1,90,95,0,0,11,221,0,97,116,97,110,
+0,1,1,0,0,11,121,95,111,118,101,114,95,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,
+97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,121,0,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,122,0,0,
+0,20,0,0,1,90,95,0,0,12,221,0,97,116,97,110,0,1,1,0,0,12,121,95,111,118,101,114,95,120,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,
+120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,116,97,110,0,18,121,95,111,118,
+101,114,95,120,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,
+18,121,95,111,118,101,114,95,120,0,59,122,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,
+58,97,116,97,110,0,18,121,95,111,118,101,114,95,120,0,59,119,0,0,0,20,0,0,1,90,95,0,0,9,221,0,97,
+116,97,110,0,1,1,0,0,9,121,0,0,1,1,0,0,9,120,0,0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,10,58,97,98,
+115,0,18,120,0,0,0,17,49,0,48,0,45,52,0,41,0,2,9,18,114,0,58,97,116,97,110,0,18,121,0,18,120,0,49,
+0,0,20,0,10,18,120,0,17,48,0,48,0,0,40,0,2,9,18,114,0,18,114,0,58,115,105,103,110,0,18,121,0,0,0,
+17,51,0,49,52,49,53,57,51,0,0,48,46,20,0,0,9,14,0,0,2,9,18,114,0,58,115,105,103,110,0,18,121,0,0,0,
+17,49,0,53,55,48,55,57,54,53,0,0,48,20,0,0,8,18,114,0,0,0,1,90,95,0,0,10,221,0,97,116,97,110,0,1,1,
+0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,110,
+0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,97,
+116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,0,1,90,95,0,0,11,221,0,97,116,97,110,0,
+1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,97,116,97,
+110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,
+58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,
+0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0,0,18,118,0,59,122,0,0,0,20,0,0,1,90,95,0,0,12,221,
+0,97,116,97,110,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
+120,0,58,97,116,97,110,0,18,117,0,59,120,0,0,18,118,0,59,120,0,0,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,59,121,0,58,97,116,97,110,0,18,117,0,59,121,0,0,18,118,0,59,121,0,0,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,122,0,58,97,116,97,110,0,18,117,0,59,122,0,0,18,118,0,59,122,0,0,0,20,0,9,
+18,95,95,114,101,116,86,97,108,0,59,119,0,58,97,116,97,110,0,18,117,0,59,119,0,0,18,118,0,59,119,0,
+0,0,20,0,0,1,90,95,0,0,9,221,0,112,111,119,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,102,108,111,
+97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,
+0,0,10,221,0,112,111,119,0,1,1,0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,4,102,108,111,97,116,95,112,111,
 119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,
 4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,97,0,
-59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,
-86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,18,98,0,59,122,0,0,0,0,1,0,0,12,0,112,111,119,0,1,1,0,0,
-12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,
-86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,112,111,119,
-101,114,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,18,98,0,59,121,0,0,0,4,
-102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,97,0,59,
-122,0,0,18,98,0,59,122,0,0,0,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,
-97,108,0,59,119,0,0,18,97,0,59,119,0,0,18,98,0,59,119,0,0,0,0,1,0,0,9,0,101,120,112,0,1,1,0,0,9,97,
-0,0,0,1,3,2,0,0,9,1,116,0,2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,108,111,97,
-116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,0,18,116,0,0,0,0,1,0,0,10,0,101,120,112,0,
-1,1,0,0,10,97,0,0,0,1,3,2,0,0,10,1,116,0,2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,
-102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,116,0,59,120,
-0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,116,
-0,59,121,0,0,0,0,1,0,0,11,0,101,120,112,0,1,1,0,0,11,97,0,0,0,1,3,2,0,0,11,1,116,0,2,18,97,0,17,49,
-0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,
-86,97,108,0,59,120,0,0,18,116,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,
-101,116,86,97,108,0,59,121,0,0,18,116,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,
-95,95,114,101,116,86,97,108,0,59,122,0,0,18,116,0,59,122,0,0,0,0,1,0,0,12,0,101,120,112,0,1,1,0,0,
-12,97,0,0,0,1,3,2,0,0,12,1,116,0,2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,108,
-111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,116,0,59,120,0,0,0,4,
-102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,116,0,59,121,
-0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,116,
-0,59,122,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,119,0,
-0,18,116,0,59,119,0,0,0,0,1,0,0,9,0,108,111,103,50,0,1,1,0,0,9,120,0,0,0,1,4,102,108,111,97,116,95,
-108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,10,0,108,111,103,50,0,1,1,
-0,0,10,118,0,0,0,1,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,
-120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,
-108,0,59,121,0,0,18,118,0,59,121,0,0,0,0,1,0,0,11,0,108,111,103,50,0,1,1,0,0,11,118,0,0,0,1,4,102,
-108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,
-0,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,
-121,0,0,0,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,
-118,0,59,122,0,0,0,0,1,0,0,12,0,108,111,103,50,0,1,1,0,0,12,118,0,0,0,1,4,102,108,111,97,116,95,
-108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,
-97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,
-108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,
-0,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,118,0,59,
-119,0,0,0,0,1,0,0,9,0,108,111,103,0,1,1,0,0,9,120,0,0,0,1,3,2,1,0,9,1,99,0,2,17,48,0,54,57,51,49,
-52,55,49,56,49,0,0,0,0,8,58,108,111,103,50,0,18,120,0,0,0,18,99,0,48,0,0,1,0,0,10,0,108,111,103,0,
-1,1,0,0,10,118,0,0,0,1,3,2,1,0,9,1,99,0,2,17,48,0,54,57,51,49,52,55,49,56,49,0,0,0,0,8,58,108,111,
-103,50,0,18,118,0,0,0,18,99,0,48,0,0,1,0,0,11,0,108,111,103,0,1,1,0,0,11,118,0,0,0,1,3,2,1,0,9,1,
-99,0,2,17,48,0,54,57,51,49,52,55,49,56,49,0,0,0,0,8,58,108,111,103,50,0,18,118,0,0,0,18,99,0,48,0,
-0,1,0,0,12,0,108,111,103,0,1,1,0,0,12,118,0,0,0,1,3,2,1,0,9,1,99,0,2,17,48,0,54,57,51,49,52,55,49,
-56,49,0,0,0,0,8,58,108,111,103,50,0,18,118,0,0,0,18,99,0,48,0,0,1,0,0,9,0,101,120,112,50,0,1,1,0,0,
-9,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
-0,0,1,0,0,10,0,101,120,112,50,0,1,1,0,0,10,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,
-95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,
-50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,0,0,1,0,0,11,0,101,120,112,50,
-0,1,1,0,0,11,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,
-59,120,0,0,18,97,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,
-97,108,0,59,121,0,0,18,97,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,
-116,86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,0,0,1,0,0,12,0,101,120,112,50,0,1,1,0,0,12,97,0,0,0,
-1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,59,
-120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,
-97,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,122,
-0,0,18,97,0,59,122,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,
-59,119,0,0,18,97,0,59,119,0,0,0,0,1,0,0,9,0,115,113,114,116,0,1,1,0,0,9,120,0,0,0,1,3,2,0,0,9,1,
-114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,120,0,0,0,4,102,108,111,97,116,95,
-114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,0,0,1,0,0,10,0,115,113,114,116,0,1,1,0,
-0,10,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,
-59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
-114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,
-116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,0,1,0,0,11,0,115,113,
-114,116,0,1,1,0,0,11,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,
-114,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,
-0,59,120,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,
-102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,
-108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,
-112,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,114,0,0,0,0,1,0,0,12,0,115,113,114,116,0,1,1,
-0,0,12,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,
-59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
-114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,
-116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,108,111,97,116,
-95,114,115,113,0,18,114,0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,
-114,101,116,86,97,108,0,59,122,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,
-18,118,0,59,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,119,
-0,0,18,114,0,0,0,0,1,0,0,9,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,9,120,0,0,0,1,4,
-102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,120,0,0,0,0,1,0,
-0,10,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,10,118,0,0,0,1,4,102,108,111,97,116,
-95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,
-97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,0,1,0,0,
-11,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,11,118,0,0,0,1,4,102,108,111,97,116,95,
-114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,
-116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,108,
-111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0,0,1,
-0,0,12,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,12,118,0,0,0,1,4,102,108,111,97,116,
-95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,
-97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,
-108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0,
-4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,118,0,59,119,
-0,0,0,0,1,0,0,9,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,17,49,0,48,0,0,20,0,0,1,0,0,10,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,10,
-118,0,0,0,1,3,2,1,0,9,1,115,0,2,58,105,110,118,101,114,115,101,115,113,114,116,0,58,100,111,116,0,
-18,118,0,0,18,118,0,0,0,0,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,115,0,0,0,0,1,0,0,11,0,110,111,114,109,97,108,105,
-122,101,0,1,1,0,0,11,118,0,0,0,1,3,2,0,0,9,1,116,109,112,0,0,0,4,118,101,99,51,95,100,111,116,0,18,
-116,109,112,0,0,18,118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,
-18,116,109,112,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,
-97,108,0,59,120,121,122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,0,0,12,0,110,111,114,109,97,108,
-105,122,101,0,1,1,0,0,12,118,0,0,0,1,3,2,0,0,9,1,116,109,112,0,0,0,4,118,101,99,52,95,100,111,116,
-0,18,116,109,112,0,0,18,118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,
-0,0,18,116,109,112,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,
-86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,0,0,9,0,97,98,115,0,1,1,0,0,9,
-97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,0,0,
-10,0,97,98,115,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,
-108,0,59,120,121,0,0,18,97,0,0,0,0,1,0,0,11,0,97,98,115,0,1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,
-97,98,115,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,0,0,1,0,0,12,0,97,98,115,
-0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,
-0,0,0,1,0,0,9,0,115,105,103,110,0,1,1,0,0,9,120,0,0,0,1,3,2,0,0,9,1,112,0,0,1,1,110,0,0,0,4,118,
-101,99,52,95,115,103,116,0,18,112,0,0,18,120,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,
-0,18,110,0,0,17,48,0,48,0,0,0,18,120,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,
-95,114,101,116,86,97,108,0,0,18,112,0,0,18,110,0,0,0,0,1,0,0,10,0,115,105,103,110,0,1,1,0,0,10,118,
-0,0,0,1,3,2,0,0,10,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,59,120,121,0,
-0,18,118,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,59,120,121,0,0,17,48,0,
-48,0,0,0,18,118,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,
-108,0,59,120,121,0,0,18,112,0,0,18,110,0,0,0,0,1,0,0,11,0,115,105,103,110,0,1,1,0,0,11,118,0,0,0,1,
-3,2,0,0,11,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,59,120,121,122,0,0,18,
-118,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,59,120,121,122,0,0,17,48,0,48,
-0,0,0,18,118,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,
-108,0,59,120,121,122,0,0,18,112,0,0,18,110,0,0,0,0,1,0,0,12,0,115,105,103,110,0,1,1,0,0,12,118,0,0,
-0,1,3,2,0,0,12,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,0,18,118,0,0,17,
-48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,0,17,48,0,48,0,0,0,18,118,0,0,0,4,118,
-101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,112,0,0,18,110,0,
-0,0,0,1,0,0,9,0,102,108,111,111,114,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,
-0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,0,0,10,0,102,108,111,111,114,0,1,1,0,0,10,97,
-0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,
-18,97,0,0,0,0,1,0,0,11,0,102,108,111,111,114,0,1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,102,108,
-111,111,114,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,0,0,1,0,0,12,0,102,108,
-111,111,114,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,0,18,95,95,114,101,116,
-86,97,108,0,0,18,97,0,0,0,0,1,0,0,9,0,99,101,105,108,0,1,1,0,0,9,97,0,0,0,1,3,2,0,0,9,1,98,0,2,18,
+59,121,0,0,18,98,0,59,121,0,0,0,0,1,90,95,0,0,11,221,0,112,111,119,0,1,1,0,0,11,97,0,0,1,1,0,0,11,
+98,0,0,0,1,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,120,0,
+0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,
+114,101,116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,
+112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,18,98,0,59,
+122,0,0,0,0,1,90,95,0,0,12,221,0,112,111,119,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,102,108,
+111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,
+18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,
+0,59,121,0,0,18,97,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,112,111,119,101,114,0,
+18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,18,98,0,59,122,0,0,0,4,102,108,111,
+97,116,95,112,111,119,101,114,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,97,0,59,119,0,0,18,
+98,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,101,120,112,0,1,1,0,0,9,97,0,0,0,1,3,2,90,95,0,0,9,221,1,
+116,0,2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,108,111,97,116,95,101,120,112,50,
+0,18,95,95,114,101,116,86,97,108,0,0,18,116,0,0,0,0,1,90,95,0,0,10,221,0,101,120,112,0,1,1,0,0,10,
+97,0,0,0,1,3,2,90,95,0,0,10,221,1,116,0,2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,
+108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,116,0,59,120,0,0,
+0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,116,0,59,
+121,0,0,0,0,1,90,95,0,0,11,221,0,101,120,112,0,1,1,0,0,11,97,0,0,0,1,3,2,90,95,0,0,11,221,1,116,0,
+2,18,97,0,17,49,0,52,52,50,54,57,53,48,50,0,0,48,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,
+95,95,114,101,116,86,97,108,0,59,120,0,0,18,116,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,
+50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,116,0,59,121,0,0,0,4,102,108,111,97,116,95,101,
+120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,116,0,59,122,0,0,0,0,1,90,95,0,0,12,
+221,0,101,120,112,0,1,1,0,0,12,97,0,0,0,1,3,2,90,95,0,0,12,221,1,116,0,2,18,97,0,17,49,0,52,52,50,
+54,57,53,48,50,0,0,48,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,
+0,59,120,0,0,18,116,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,
+86,97,108,0,59,121,0,0,18,116,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,
+101,116,86,97,108,0,59,122,0,0,18,116,0,59,122,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,
+95,95,114,101,116,86,97,108,0,59,119,0,0,18,116,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,108,111,103,
+50,0,1,1,0,0,9,120,0,0,0,1,4,102,108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,
+0,0,18,120,0,0,0,0,1,90,95,0,0,10,221,0,108,111,103,50,0,1,1,0,0,10,118,0,0,0,1,4,102,108,111,97,
+116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,
+108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,
+0,0,1,90,95,0,0,11,221,0,108,111,103,50,0,1,1,0,0,11,118,0,0,0,1,4,102,108,111,97,116,95,108,111,
+103,50,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,
+108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,108,111,
+97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0,0,1,
+90,95,0,0,12,221,0,108,111,103,50,0,1,1,0,0,12,118,0,0,0,1,4,102,108,111,97,116,95,108,111,103,50,
+0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,108,
+111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,
+116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0,4,102,
+108,111,97,116,95,108,111,103,50,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,118,0,59,119,0,0,
+0,0,1,90,95,0,0,9,221,0,108,111,103,0,1,1,0,0,9,120,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,48,0,
+54,57,51,49,52,55,49,56,49,0,0,0,0,8,58,108,111,103,50,0,18,120,0,0,0,18,99,0,48,0,0,1,90,95,0,0,
+10,221,0,108,111,103,0,1,1,0,0,10,118,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,48,0,54,57,51,49,52,
+55,49,56,49,0,0,0,0,8,58,108,111,103,50,0,18,118,0,0,0,18,99,0,48,0,0,1,90,95,0,0,11,221,0,108,111,
+103,0,1,1,0,0,11,118,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,48,0,54,57,51,49,52,55,49,56,49,0,0,0,
+0,8,58,108,111,103,50,0,18,118,0,0,0,18,99,0,48,0,0,1,90,95,0,0,12,221,0,108,111,103,0,1,1,0,0,12,
+118,0,0,0,1,3,2,90,95,1,0,9,221,1,99,0,2,17,48,0,54,57,51,49,52,55,49,56,49,0,0,0,0,8,58,108,111,
+103,50,0,18,118,0,0,0,18,99,0,48,0,0,1,90,95,0,0,9,221,0,101,120,112,50,0,1,1,0,0,9,97,0,0,0,1,4,
+102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,
+0,10,221,0,101,120,112,50,0,1,1,0,0,10,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,
+95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,
+0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,0,0,1,90,95,0,0,11,221,0,101,120,
+112,50,0,1,1,0,0,11,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,
+108,0,59,120,0,0,18,97,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,
+116,86,97,108,0,59,121,0,0,18,97,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,
+114,101,116,86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,0,0,1,90,95,0,0,12,221,0,101,120,112,50,0,1,
+1,0,0,12,97,0,0,0,1,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0,59,
+120,0,0,18,97,0,59,120,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,
+108,0,59,121,0,0,18,97,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,
+116,86,97,108,0,59,122,0,0,18,97,0,59,122,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,
+114,101,116,86,97,108,0,59,119,0,0,18,97,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,115,113,114,116,0,1,
+1,0,0,9,120,0,0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,
+0,18,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,
+0,0,1,90,95,0,0,10,221,0,115,113,114,116,0,1,1,0,0,10,118,0,0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,
+4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,114,
+99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,
+113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,
+86,97,108,0,59,121,0,0,18,114,0,0,0,0,1,90,95,0,0,11,221,0,115,113,114,116,0,1,1,0,0,11,118,0,0,0,
+1,3,2,90,95,0,0,9,221,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,120,
+0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,
+0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,
+114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,
+115,113,0,18,114,0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,
+116,86,97,108,0,59,122,0,0,18,114,0,0,0,0,1,90,95,0,0,12,221,0,115,113,114,116,0,1,1,0,0,12,118,0,
+0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,
+120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,
+0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,
+95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,108,111,97,116,95,
+114,115,113,0,18,114,0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,
+101,116,86,97,108,0,59,122,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,
+118,0,59,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,119,0,
+0,18,114,0,0,0,0,1,90,95,0,0,9,221,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,9,120,0,
+0,0,1,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,120,0,0,
+0,0,1,90,95,0,0,10,221,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,10,118,0,0,0,1,4,
+102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,
+0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,
+121,0,0,0,0,1,90,95,0,0,11,221,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,11,118,0,0,
+0,1,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,
+120,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,
+0,59,121,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,
+18,118,0,59,122,0,0,0,0,1,90,95,0,0,12,221,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,
+12,118,0,0,0,1,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,
+18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,
+121,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,
+0,59,122,0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,
+97,108,0,59,119,0,0,18,118,0,59,119,0,0,0,0,1,90,95,0,0,9,221,0,110,111,114,109,97,108,105,122,101,
+0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,17,49,0,48,0,0,20,0,0,1,90,95,0,0,10,
+221,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,10,118,0,0,0,1,3,2,90,95,1,0,9,221,1,115,0,2,58,
+105,110,118,101,114,115,101,115,113,114,116,0,58,100,111,116,0,18,118,0,0,18,118,0,0,0,0,0,0,0,4,
+118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,
+18,118,0,0,18,115,0,0,0,0,1,90,95,0,0,11,221,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,11,118,
+0,0,0,1,3,2,90,95,0,0,9,221,1,116,109,112,0,0,0,4,118,101,99,51,95,100,111,116,0,18,116,109,112,0,
+0,18,118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,18,116,109,112,
+0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,
+121,122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,90,95,0,0,12,221,0,110,111,114,109,97,108,105,122,
+101,0,1,1,0,0,12,118,0,0,0,1,3,2,90,95,0,0,9,221,1,116,109,112,0,0,0,4,118,101,99,52,95,100,111,
+116,0,18,116,109,112,0,0,18,118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,
+112,0,0,18,116,109,112,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,90,95,0,0,9,221,0,97,98,115,
+0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
+0,0,1,90,95,0,0,10,221,0,97,98,115,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,0,0,1,90,95,0,0,11,221,0,97,98,115,0,1,1,0,0,11,
+97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,
+97,0,0,0,0,1,90,95,0,0,12,221,0,97,98,115,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,
+18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,0,9,221,0,115,105,103,110,0,1,1,0,0,9,
+120,0,0,0,1,3,2,90,95,0,0,9,221,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,
+0,18,120,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,0,17,48,0,48,0,0,0,18,
+120,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,
+112,0,0,18,110,0,0,0,0,1,90,95,0,0,10,221,0,115,105,103,110,0,1,1,0,0,10,118,0,0,0,1,3,2,90,95,0,0,
+10,221,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,59,120,121,0,0,18,118,0,0,
+17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,59,120,121,0,0,17,48,0,48,0,0,0,18,
+118,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,
+120,121,0,0,18,112,0,0,18,110,0,0,0,0,1,90,95,0,0,11,221,0,115,105,103,110,0,1,1,0,0,11,118,0,0,0,
+1,3,2,90,95,0,0,11,221,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,103,116,0,18,112,0,59,120,
+121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,59,120,121,122,
+0,0,17,48,0,48,0,0,0,18,118,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,
+101,116,86,97,108,0,59,120,121,122,0,0,18,112,0,0,18,110,0,0,0,0,1,90,95,0,0,12,221,0,115,105,103,
+110,0,1,1,0,0,12,118,0,0,0,1,3,2,90,95,0,0,12,221,1,112,0,0,1,1,110,0,0,0,4,118,101,99,52,95,115,
+103,116,0,18,112,0,0,18,118,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,103,116,0,18,110,0,0,17,
+48,0,48,0,0,0,18,118,0,0,0,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,
+86,97,108,0,0,18,112,0,0,18,110,0,0,0,0,1,90,95,0,0,9,221,0,102,108,111,111,114,0,1,1,0,0,9,97,0,0,
+0,1,4,118,101,99,52,95,102,108,111,111,114,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,
+95,0,0,10,221,0,102,108,111,111,114,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,
+0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,0,0,1,90,95,0,0,11,221,0,102,108,111,
+111,114,0,1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,0,18,95,95,114,101,116,86,
+97,108,0,59,120,121,122,0,0,18,97,0,0,0,0,1,90,95,0,0,12,221,0,102,108,111,111,114,0,1,1,0,0,12,97,
+0,0,0,1,4,118,101,99,52,95,102,108,111,111,114,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,
+1,90,95,0,0,9,221,0,99,101,105,108,0,1,1,0,0,9,97,0,0,0,1,3,2,90,95,0,0,9,221,1,98,0,2,18,97,0,54,
+0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,98,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,
+108,0,18,98,0,54,20,0,0,1,90,95,0,0,10,221,0,99,101,105,108,0,1,1,0,0,10,97,0,0,0,1,3,2,90,95,0,0,
+10,221,1,98,0,2,18,97,0,54,0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,98,0,0,18,98,0,0,0,9,18,
+95,95,114,101,116,86,97,108,0,59,120,121,0,18,98,0,54,20,0,0,1,90,95,0,0,11,221,0,99,101,105,108,0,
+1,1,0,0,11,97,0,0,0,1,3,2,90,95,0,0,11,221,1,98,0,2,18,97,0,54,0,0,4,118,101,99,52,95,102,108,111,
+111,114,0,18,98,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,18,98,0,54,20,
+0,0,1,90,95,0,0,12,221,0,99,101,105,108,0,1,1,0,0,12,97,0,0,0,1,3,2,90,95,0,0,12,221,1,98,0,2,18,
 97,0,54,0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,98,0,0,18,98,0,0,0,9,18,95,95,114,101,116,
-86,97,108,0,18,98,0,54,20,0,0,1,0,0,10,0,99,101,105,108,0,1,1,0,0,10,97,0,0,0,1,3,2,0,0,10,1,98,0,
-2,18,97,0,54,0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,98,0,0,18,98,0,0,0,9,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,18,98,0,54,20,0,0,1,0,0,11,0,99,101,105,108,0,1,1,0,0,11,97,0,0,0,1,3,
-2,0,0,11,1,98,0,2,18,97,0,54,0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,98,0,0,18,98,0,0,0,9,
-18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,18,98,0,54,20,0,0,1,0,0,12,0,99,101,105,108,0,1,
-1,0,0,12,97,0,0,0,1,3,2,0,0,12,1,98,0,2,18,97,0,54,0,0,4,118,101,99,52,95,102,108,111,111,114,0,18,
-98,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,18,98,0,54,20,0,0,1,0,0,9,0,102,114,97,99,
-116,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,102,114,97,99,0,18,95,95,114,101,116,86,97,108,0,0,
-18,97,0,0,0,0,1,0,0,10,0,102,114,97,99,116,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,102,114,97,
-99,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,0,0,1,0,0,11,0,102,114,97,99,116,0,
-1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,102,114,97,99,0,18,95,95,114,101,116,86,97,108,0,59,120,
-121,122,0,0,18,97,0,0,0,0,1,0,0,12,0,102,114,97,99,116,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,
-102,114,97,99,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,0,0,9,0,109,111,100,0,1,1,0,0,9,
-97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,
-95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,
-0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,
-20,0,0,1,0,0,10,0,109,111,100,0,1,1,0,0,10,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,111,110,101,79,
-118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,0,18,
-98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,59,120,121,0,18,97,0,18,98,0,58,102,108,111,111,114,0,
-18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,0,0,11,0,109,111,100,0,1,1,0,0,11,
-97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,
-95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,
-0,59,120,121,122,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,
-66,0,48,0,0,48,47,20,0,0,1,0,0,12,0,109,111,100,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,
-1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,
-101,114,66,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,
-114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,0,0,10,0,109,111,100,0,1,1,
-0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,3,2,0,0,10,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,
-97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,
-108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,
-9,18,95,95,114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,
-79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,0,0,11,0,109,111,100,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,
-0,0,1,3,2,0,0,11,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,
-110,101,79,118,101,114,66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,
-18,111,110,101,79,118,101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,
-112,0,18,111,110,101,79,118,101,114,66,0,59,122,0,0,18,98,0,59,122,0,0,0,9,18,95,95,114,101,116,86,
-97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,
-0,48,47,20,0,0,1,0,0,12,0,109,111,100,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,3,2,0,0,12,1,111,
-110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,
-66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,
-101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,
-79,118,101,114,66,0,59,122,0,0,18,98,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,
-110,101,79,118,101,114,66,0,59,119,0,0,18,98,0,59,119,0,0,0,9,18,95,95,114,101,116,86,97,108,0,18,
-97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,20,0,
-0,1,0,0,9,0,109,105,110,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,
-18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,10,0,109,105,110,0,1,1,0,0,10,97,
-0,0,1,1,0,0,10,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,59,120,
-121,0,0,18,97,0,59,120,121,0,0,18,98,0,59,120,121,0,0,0,0,1,0,0,11,0,109,105,110,0,1,1,0,0,11,97,0,
-0,1,1,0,0,11,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,59,120,
-121,122,0,0,18,97,0,59,120,121,122,0,0,18,98,0,59,120,121,122,0,0,0,0,1,0,0,12,0,109,105,110,0,1,1,
-0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,
-108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,10,0,109,105,110,0,1,1,0,0,10,97,0,0,1,1,0,0,9,98,0,0,0,1,4,
-118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,0,0,18,98,0,0,
-0,0,1,0,0,11,0,109,105,110,0,1,1,0,0,11,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,105,110,
-0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,12,0,109,105,
-110,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,
-86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,9,0,109,97,120,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,
-4,118,101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,
-10,0,109,97,120,0,1,1,0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,59,120,121,0,0,18,98,0,59,120,121,0,0,0,0,1,0,0,11,
-0,109,97,120,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,59,120,121,122,0,0,18,98,0,59,120,121,122,0,0,0,
-0,1,0,0,12,0,109,97,120,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,
-18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,10,0,109,97,120,0,1,1,0,0,10,97,0,
-0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,
-59,120,121,0,0,18,98,0,0,0,0,1,0,0,11,0,109,97,120,0,1,1,0,0,11,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,
-101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,122,0,0,18,98,0,0,
-0,0,1,0,0,12,0,109,97,120,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,
-18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,9,0,99,108,97,109,112,0,1,1,0,0,9,
+86,97,108,0,18,98,0,54,20,0,0,1,90,95,0,0,9,221,0,102,114,97,99,116,0,1,1,0,0,9,97,0,0,0,1,4,118,
+101,99,52,95,102,114,97,99,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,0,10,221,0,
+102,114,97,99,116,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,102,114,97,99,0,18,95,95,114,101,116,
+86,97,108,0,59,120,121,0,0,18,97,0,0,0,0,1,90,95,0,0,11,221,0,102,114,97,99,116,0,1,1,0,0,11,97,0,
+0,0,1,4,118,101,99,52,95,102,114,97,99,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,
+0,0,0,0,1,90,95,0,0,12,221,0,102,114,97,99,116,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,102,114,
+97,99,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,0,9,221,0,109,111,100,0,1,1,0,0,
+9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,
+111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,0,18,98,0,0,0,9,18,95,95,114,101,116,
+86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,
+0,0,48,47,20,0,0,1,90,95,0,0,10,221,0,109,111,100,0,1,1,0,0,10,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,
+95,0,0,9,221,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,
+101,79,118,101,114,66,0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,59,120,121,0,18,97,0,18,
+98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,90,
+95,0,0,11,221,0,109,111,100,0,1,1,0,0,11,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,111,110,
+101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,
+0,0,18,98,0,0,0,9,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,18,97,0,18,98,0,58,102,108,111,
+111,114,0,18,97,0,18,111,110,101,79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,90,95,0,0,12,221,0,109,
+111,100,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,111,110,101,79,118,101,114,
+66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,0,18,98,0,0,0,9,
+18,95,95,114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,
+79,118,101,114,66,0,48,0,0,48,47,20,0,0,1,90,95,0,0,10,221,0,109,111,100,0,1,1,0,0,10,97,0,0,1,1,0,
+0,10,98,0,0,0,1,3,2,90,95,0,0,10,221,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,
+114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,9,18,95,95,
+114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,
+114,66,0,48,0,0,48,47,20,0,0,1,90,95,0,0,11,221,0,109,111,100,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,
+0,0,1,3,2,90,95,0,0,11,221,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,
+112,0,18,111,110,101,79,118,101,114,66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,
+114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,122,0,0,18,98,0,59,122,0,0,0,9,18,95,95,
+114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,118,101,
+114,66,0,48,0,0,48,47,20,0,0,1,90,95,0,0,12,221,0,109,111,100,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,
+0,0,1,3,2,90,95,0,0,12,221,1,111,110,101,79,118,101,114,66,0,0,0,4,102,108,111,97,116,95,114,99,
+112,0,18,111,110,101,79,118,101,114,66,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,
+114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,122,0,0,18,98,0,59,122,0,0,0,4,102,108,
+111,97,116,95,114,99,112,0,18,111,110,101,79,118,101,114,66,0,59,119,0,0,18,98,0,59,119,0,0,0,9,18,
+95,95,114,101,116,86,97,108,0,18,97,0,18,98,0,58,102,108,111,111,114,0,18,97,0,18,111,110,101,79,
+118,101,114,66,0,48,0,0,48,47,20,0,0,1,90,95,0,0,9,221,0,109,105,110,0,1,1,0,0,9,97,0,0,1,1,0,0,9,
+98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,
+0,0,1,90,95,0,0,10,221,0,109,105,110,0,1,1,0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,4,118,101,99,52,95,
+109,105,110,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,59,120,121,0,0,18,98,0,59,
+120,121,0,0,0,0,1,90,95,0,0,11,221,0,109,105,110,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,0,0,1,4,118,
+101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,59,120,121,
+122,0,0,18,98,0,59,120,121,122,0,0,0,0,1,90,95,0,0,12,221,0,109,105,110,0,1,1,0,0,12,97,0,0,1,1,0,
+0,12,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,
+98,0,0,0,0,1,90,95,0,0,10,221,0,109,105,110,0,1,1,0,0,10,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,
+52,95,109,105,110,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,0,0,18,98,0,0,0,0,1,90,
+95,0,0,11,221,0,109,105,110,0,1,1,0,0,11,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,105,
+110,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,12,
+221,0,109,105,110,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,105,110,0,18,95,
+95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,9,221,0,109,97,120,0,1,1,0,0,9,97,
+0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,
+0,0,18,98,0,0,0,0,1,90,95,0,0,10,221,0,109,97,120,0,1,1,0,0,10,97,0,0,1,1,0,0,10,98,0,0,0,1,4,118,
+101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,59,120,121,0,0,
+18,98,0,59,120,121,0,0,0,0,1,90,95,0,0,11,221,0,109,97,120,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,0,0,
+1,4,118,101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,59,
+120,121,122,0,0,18,98,0,59,120,121,122,0,0,0,0,1,90,95,0,0,12,221,0,109,97,120,0,1,1,0,0,12,97,0,0,
+1,1,0,0,12,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
+18,98,0,0,0,0,1,90,95,0,0,10,221,0,109,97,120,0,1,1,0,0,10,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,
+99,52,95,109,97,120,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,0,0,18,98,0,0,0,0,1,90,
+95,0,0,11,221,0,109,97,120,0,1,1,0,0,11,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,97,120,
+0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,12,221,0,
+109,97,120,0,1,1,0,0,12,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,97,120,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,9,221,0,99,108,97,109,112,0,1,1,0,0,9,
 118,97,108,0,0,1,1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,
 101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,
-110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,0,0,10,0,99,108,97,109,112,0,1,1,0,0,10,118,97,
-108,0,0,1,1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,
-95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,
-108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,0,0,11,0,99,108,97,109,112,0,1,1,0,0,11,118,97,108,0,0,1,
-1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,95,99,108,
-97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,108,0,0,18,
-109,97,120,86,97,108,0,0,0,0,1,0,0,12,0,99,108,97,109,112,0,1,1,0,0,12,118,97,108,0,0,1,1,0,0,9,
-109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,95,99,108,97,109,
-112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,108,0,0,18,109,97,
-120,86,97,108,0,0,0,0,1,0,0,10,0,99,108,97,109,112,0,1,1,0,0,10,118,97,108,0,0,1,1,0,0,10,109,105,
-110,86,97,108,0,0,1,1,0,0,10,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,95,99,108,97,109,112,0,
-18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,108,0,0,18,109,97,120,86,
-97,108,0,0,0,0,1,0,0,11,0,99,108,97,109,112,0,1,1,0,0,11,118,97,108,0,0,1,1,0,0,11,109,105,110,86,
-97,108,0,0,1,1,0,0,11,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,95,99,108,97,109,112,0,18,95,95,
-114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,108,0,0,18,109,97,120,86,97,108,0,
-0,0,0,1,0,0,12,0,99,108,97,109,112,0,1,1,0,0,12,118,97,108,0,0,1,1,0,0,12,109,105,110,86,97,108,0,
-0,1,1,0,0,12,109,97,120,86,97,108,0,0,0,1,4,118,101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,
-116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,
-0,0,9,0,109,105,120,0,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,
-108,114,112,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,10,0,
-109,105,120,0,1,1,0,0,10,120,0,0,1,1,0,0,10,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,
-114,112,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,11,0,109,
-105,120,0,1,1,0,0,11,120,0,0,1,1,0,0,11,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,
-112,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,12,0,109,105,
-120,0,1,1,0,0,12,120,0,0,1,1,0,0,12,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,
-18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,10,0,109,105,120,0,1,
-1,0,0,10,120,0,0,1,1,0,0,10,121,0,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,
-95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,11,0,109,105,120,0,1,1,0,0,
-11,120,0,0,1,1,0,0,11,121,0,0,1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,
-101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,12,0,109,105,120,0,1,1,0,0,12,120,
-0,0,1,1,0,0,12,121,0,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,101,116,
-86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,0,0,9,0,115,116,101,112,0,1,1,0,0,9,101,100,
-103,101,0,0,1,1,0,0,9,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,
-0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,0,0,10,0,115,116,101,112,0,1,1,0,0,10,101,100,103,101,
-0,0,1,1,0,0,10,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,
-120,121,0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,0,0,11,0,115,116,101,112,0,1,1,0,0,11,101,100,
-103,101,0,0,1,1,0,0,11,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,
-0,59,120,121,122,0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,0,0,12,0,115,116,101,112,0,1,1,0,0,12,
-101,100,103,101,0,0,1,1,0,0,12,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,
-86,97,108,0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,0,0,10,0,115,116,101,112,0,1,1,0,0,9,101,100,
-103,101,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,
-0,59,120,121,0,0,18,118,0,0,18,101,100,103,101,0,0,0,0,1,0,0,11,0,115,116,101,112,0,1,1,0,0,9,101,
-100,103,101,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,
-108,0,59,120,121,122,0,0,18,118,0,0,18,101,100,103,101,0,0,0,0,1,0,0,12,0,115,116,101,112,0,1,1,0,
-0,9,101,100,103,101,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,
-116,86,97,108,0,0,18,118,0,0,18,101,100,103,101,0,0,0,0,1,0,0,9,0,115,109,111,111,116,104,115,116,
+110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,10,221,0,99,108,97,109,112,0,1,1,0,0,
+10,118,97,108,0,0,1,1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,
+101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,
+110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,11,221,0,99,108,97,109,112,0,1,1,0,0,
+11,118,97,108,0,0,1,1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,
+101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,
+110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,12,221,0,99,108,97,109,112,0,1,1,0,0,
+12,118,97,108,0,0,1,1,0,0,9,109,105,110,86,97,108,0,0,1,1,0,0,9,109,97,120,86,97,108,0,0,0,1,4,118,
+101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,105,
+110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,10,221,0,99,108,97,109,112,0,1,1,0,0,
+10,118,97,108,0,0,1,1,0,0,10,109,105,110,86,97,108,0,0,1,1,0,0,10,109,97,120,86,97,108,0,0,0,1,4,
+118,101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,
+105,110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,11,221,0,99,108,97,109,112,0,1,1,
+0,0,11,118,97,108,0,0,1,1,0,0,11,109,105,110,86,97,108,0,0,1,1,0,0,11,109,97,120,86,97,108,0,0,0,1,
+4,118,101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,
+105,110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,12,221,0,99,108,97,109,112,0,1,1,
+0,0,12,118,97,108,0,0,1,1,0,0,12,109,105,110,86,97,108,0,0,1,1,0,0,12,109,97,120,86,97,108,0,0,0,1,
+4,118,101,99,52,95,99,108,97,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,118,97,108,0,0,18,109,
+105,110,86,97,108,0,0,18,109,97,120,86,97,108,0,0,0,0,1,90,95,0,0,9,221,0,109,105,120,0,1,1,0,0,9,
+120,0,0,1,1,0,0,9,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,101,
+116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,10,221,0,109,105,120,0,1,1,0,0,
+10,120,0,0,1,1,0,0,10,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,11,221,0,109,105,120,0,1,1,0,
+0,11,120,0,0,1,1,0,0,11,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,12,221,0,109,105,120,0,1,1,0,
+0,12,120,0,0,1,1,0,0,12,121,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,10,221,0,109,105,120,0,1,1,0,
+0,10,120,0,0,1,1,0,0,10,121,0,0,1,1,0,0,10,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,95,
+114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,11,221,0,109,105,120,0,1,
+1,0,0,11,120,0,0,1,1,0,0,11,121,0,0,1,1,0,0,11,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,95,
+95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,12,221,0,109,105,120,
+0,1,1,0,0,12,120,0,0,1,1,0,0,12,121,0,0,1,1,0,0,12,97,0,0,0,1,4,118,101,99,52,95,108,114,112,0,18,
+95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,121,0,0,18,120,0,0,0,0,1,90,95,0,0,9,221,0,115,116,
+101,112,0,1,1,0,0,9,101,100,103,101,0,0,1,1,0,0,9,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,
+95,95,114,101,116,86,97,108,0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,90,95,0,0,10,221,0,115,116,
+101,112,0,1,1,0,0,10,101,100,103,101,0,0,1,1,0,0,10,120,0,0,0,1,4,118,101,99,52,95,115,103,101,0,
+18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,120,0,0,18,101,100,103,101,0,0,0,0,1,90,95,0,0,
+11,221,0,115,116,101,112,0,1,1,0,0,11,101,100,103,101,0,0,1,1,0,0,11,120,0,0,0,1,4,118,101,99,52,
+95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,120,0,0,18,101,100,103,101,
+0,0,0,0,1,90,95,0,0,12,221,0,115,116,101,112,0,1,1,0,0,12,101,100,103,101,0,0,1,1,0,0,12,120,0,0,0,
+1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,18,101,100,103,
+101,0,0,0,0,1,90,95,0,0,10,221,0,115,116,101,112,0,1,1,0,0,9,101,100,103,101,0,0,1,1,0,0,10,118,0,
+0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,
+18,101,100,103,101,0,0,0,0,1,90,95,0,0,11,221,0,115,116,101,112,0,1,1,0,0,9,101,100,103,101,0,0,1,
+1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,
+122,0,0,18,118,0,0,18,101,100,103,101,0,0,0,0,1,90,95,0,0,12,221,0,115,116,101,112,0,1,1,0,0,9,101,
+100,103,101,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,
+108,0,0,18,118,0,0,18,101,100,103,101,0,0,0,0,1,90,95,0,0,9,221,0,115,109,111,111,116,104,115,116,
 101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,0,9,120,0,0,0,1,
-3,2,0,0,9,1,116,0,2,58,99,108,97,109,112,0,18,120,0,18,101,100,103,101,48,0,47,18,101,100,103,101,
-49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,116,0,
-48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,10,0,115,109,111,111,116,104,115,116,
-101,112,0,1,1,0,0,10,101,100,103,101,48,0,0,1,1,0,0,10,101,100,103,101,49,0,0,1,1,0,0,10,118,0,0,0,
-1,3,2,0,0,10,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,100,103,
-101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,116,
-0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,11,0,115,109,111,111,116,104,115,
-116,101,112,0,1,1,0,0,11,101,100,103,101,48,0,0,1,1,0,0,11,101,100,103,101,49,0,0,1,1,0,0,11,118,0,
-0,0,1,3,2,0,0,11,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,100,
-103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,
-116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,12,0,115,109,111,111,116,104,
-115,116,101,112,0,1,1,0,0,12,101,100,103,101,48,0,0,1,1,0,0,12,101,100,103,101,49,0,0,1,1,0,0,12,
-118,0,0,0,1,3,2,0,0,12,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,
-100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,
-0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,10,0,115,109,111,111,116,
-104,115,116,101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,0,10,
-118,0,0,0,1,3,2,0,0,10,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,
-100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,
-0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,11,0,115,109,111,111,116,
-104,115,116,101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,0,11,
-118,0,0,0,1,3,2,0,0,11,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,
+3,2,90,95,0,0,9,221,1,116,0,2,58,99,108,97,109,112,0,18,120,0,18,101,100,103,101,48,0,47,18,101,
 100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,
-0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,12,0,115,109,111,111,116,
-104,115,116,101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,0,12,
-118,0,0,0,1,3,2,0,0,12,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,
-100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,
-0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,0,0,9,0,108,101,110,103,116,
-104,0,1,1,0,0,9,120,0,0,0,1,8,58,97,98,115,0,18,120,0,0,0,0,0,1,0,0,9,0,108,101,110,103,116,104,0,
-1,1,0,0,10,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,3,2,1,0,9,1,112,0,2,58,100,111,116,0,18,118,0,0,18,
-118,0,0,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,112,0,0,0,4,102,108,111,97,116,
-95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,0,1,0,0,9,0,108,101,110,
-103,116,104,0,1,1,0,0,11,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,3,2,1,0,9,1,112,0,2,58,100,111,116,0,18,
-118,0,0,18,118,0,0,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,112,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,0,0,1,0,0,9,0,108,101,110,
-103,116,104,0,1,1,0,0,12,118,0,0,0,1,3,2,0,0,9,1,114,0,0,0,3,2,1,0,9,1,112,0,2,58,100,111,116,0,18,
+0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,90,95,0,0,10,221,0,115,109,111,
+111,116,104,115,116,101,112,0,1,1,0,0,10,101,100,103,101,48,0,0,1,1,0,0,10,101,100,103,101,49,0,0,
+1,1,0,0,10,118,0,0,0,1,3,2,90,95,0,0,10,221,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,
+103,101,48,0,47,18,101,100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,
+48,0,0,0,0,0,0,8,18,116,0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,90,95,
+0,0,11,221,0,115,109,111,111,116,104,115,116,101,112,0,1,1,0,0,11,101,100,103,101,48,0,0,1,1,0,0,
+11,101,100,103,101,49,0,0,1,1,0,0,11,118,0,0,0,1,3,2,90,95,0,0,11,221,1,116,0,2,58,99,108,97,109,
+112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,
+17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,
+116,0,48,47,48,0,0,1,90,95,0,0,12,221,0,115,109,111,111,116,104,115,116,101,112,0,1,1,0,0,12,101,
+100,103,101,48,0,0,1,1,0,0,12,101,100,103,101,49,0,0,1,1,0,0,12,118,0,0,0,1,3,2,90,95,0,0,12,221,1,
+116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,100,103,101,49,0,18,101,
+100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,116,0,48,17,51,0,48,
+0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,90,95,0,0,10,221,0,115,109,111,111,116,104,115,116,101,
+112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,0,10,118,0,0,0,1,3,2,
+90,95,0,0,10,221,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,48,0,47,18,101,100,
+103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,
+116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,90,95,0,0,11,221,0,115,109,111,111,
+116,104,115,116,101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,103,101,49,0,0,1,1,0,
+0,11,118,0,0,0,1,3,2,90,95,0,0,11,221,1,116,0,2,58,99,108,97,109,112,0,18,118,0,18,101,100,103,101,
+48,0,47,18,101,100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,0,0,17,49,0,48,0,0,0,
+0,0,0,8,18,116,0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,0,0,1,90,95,0,0,12,
+221,0,115,109,111,111,116,104,115,116,101,112,0,1,1,0,0,9,101,100,103,101,48,0,0,1,1,0,0,9,101,100,
+103,101,49,0,0,1,1,0,0,12,118,0,0,0,1,3,2,90,95,0,0,12,221,1,116,0,2,58,99,108,97,109,112,0,18,118,
+0,18,101,100,103,101,48,0,47,18,101,100,103,101,49,0,18,101,100,103,101,48,0,47,49,0,17,48,0,48,0,
+0,0,17,49,0,48,0,0,0,0,0,0,8,18,116,0,18,116,0,48,17,51,0,48,0,0,17,50,0,48,0,0,18,116,0,48,47,48,
+0,0,1,90,95,0,0,9,221,0,108,101,110,103,116,104,0,1,1,0,0,9,120,0,0,0,1,8,58,97,98,115,0,18,120,0,
+0,0,0,0,1,90,95,0,0,9,221,0,108,101,110,103,116,104,0,1,1,0,0,10,118,0,0,0,1,3,2,90,95,0,0,9,221,1,
+114,0,0,0,3,2,90,95,1,0,9,221,1,112,0,2,58,100,111,116,0,18,118,0,0,18,118,0,0,0,0,0,4,102,108,111,
+97,116,95,114,115,113,0,18,114,0,0,18,112,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,
+101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,0,1,90,95,0,0,9,221,0,108,101,110,103,116,104,0,1,1,0,
+0,11,118,0,0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,3,2,90,95,1,0,9,221,1,112,0,2,58,100,111,116,0,18,
 118,0,0,18,118,0,0,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,112,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,0,0,1,0,0,9,0,100,105,115,
-116,97,110,99,101,0,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,0,1,3,2,1,0,9,1,100,0,2,18,120,0,18,121,0,
-47,0,0,9,18,95,95,114,101,116,86,97,108,0,58,108,101,110,103,116,104,0,18,100,0,0,0,20,0,0,1,0,0,9,
-0,100,105,115,116,97,110,99,101,0,1,1,0,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,3,2,1,0,10,1,100,50,0,
-2,18,118,0,18,117,0,47,0,0,9,18,95,95,114,101,116,86,97,108,0,58,108,101,110,103,116,104,0,18,100,
-50,0,0,0,20,0,0,1,0,0,9,0,100,105,115,116,97,110,99,101,0,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,
-1,3,2,1,0,11,1,100,51,0,2,18,118,0,18,117,0,47,0,0,9,18,95,95,114,101,116,86,97,108,0,58,108,101,
-110,103,116,104,0,18,100,51,0,0,0,20,0,0,1,0,0,9,0,100,105,115,116,97,110,99,101,0,1,1,0,0,12,118,
-0,0,1,1,0,0,12,117,0,0,0,1,3,2,1,0,12,1,100,52,0,2,18,118,0,18,117,0,47,0,0,9,18,95,95,114,101,116,
-86,97,108,0,58,108,101,110,103,116,104,0,18,100,52,0,0,0,20,0,0,1,0,0,11,0,99,114,111,115,115,0,1,
-1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,51,95,99,114,111,115,115,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,9,0,102,97,99,101,102,111,114,
-119,97,114,100,0,1,1,0,0,9,78,0,0,1,1,0,0,9,73,0,0,1,1,0,0,9,78,114,101,102,0,0,0,1,3,2,1,0,9,1,
-100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,115,0,0,0,4,118,101,99,
-52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,18,100,0,0,0,8,58,109,105,120,0,18,78,0,54,0,18,78,
-0,0,18,115,0,0,0,0,0,1,0,0,10,0,102,97,99,101,102,111,114,119,97,114,100,0,1,1,0,0,10,78,0,0,1,1,0,
-0,10,73,0,0,1,1,0,0,10,78,114,101,102,0,0,0,1,3,2,1,0,9,1,100,0,2,58,100,111,116,0,18,78,114,101,
-102,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,
-48,0,0,0,18,100,0,0,0,8,58,109,105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,0,0,11,0,102,97,
-99,101,102,111,114,119,97,114,100,0,1,1,0,0,11,78,0,0,1,1,0,0,11,73,0,0,1,1,0,0,11,78,114,101,102,
-0,0,0,1,3,2,1,0,9,1,100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,115,
-0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,18,100,0,0,0,8,58,109,105,120,0,
-18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,0,0,12,0,102,97,99,101,102,111,114,119,97,114,100,0,1,1,
-0,0,12,78,0,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,114,101,102,0,0,0,1,3,2,1,0,9,1,100,0,2,58,100,111,
-116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,
-18,115,0,0,17,48,0,48,0,0,0,18,100,0,0,0,8,58,109,105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,
-0,1,0,0,9,0,114,101,102,108,101,99,116,0,1,1,0,0,9,73,0,0,1,1,0,0,9,78,0,0,0,1,8,18,73,0,17,50,0,
-48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,0,1,0,0,10,0,114,101,102,108,101,
-99,116,0,1,1,0,0,10,73,0,0,1,1,0,0,10,78,0,0,0,1,8,18,73,0,17,50,0,48,0,0,58,100,111,116,0,18,78,0,
-0,18,73,0,0,0,48,18,78,0,48,47,0,0,1,0,0,11,0,114,101,102,108,101,99,116,0,1,1,0,0,11,73,0,0,1,1,0,
-0,11,78,0,0,0,1,8,18,73,0,17,50,0,48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,
-0,1,0,0,12,0,114,101,102,108,101,99,116,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,0,0,0,1,8,18,73,0,17,50,
-0,48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,0,1,0,0,9,0,114,101,102,114,97,
-99,116,0,1,1,0,0,9,73,0,0,1,1,0,0,9,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,0,0,9,1,110,95,100,111,
-116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,107,0,2,17,49,0,48,0,0,18,
-101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,110,95,100,111,
-116,95,105,0,48,47,48,47,0,0,3,2,0,0,9,1,114,101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,0,
-40,0,9,18,114,101,116,118,97,108,0,17,48,0,48,0,0,20,0,9,18,114,101,116,118,97,108,0,18,101,116,97,
-0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,113,114,116,0,18,107,0,0,0,
-46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,0,0,10,0,114,101,102,114,97,99,116,0,1,1,
-0,0,10,73,0,0,1,1,0,0,10,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,0,0,9,1,110,95,100,111,116,95,105,
-0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,107,0,2,17,49,0,48,0,0,18,101,116,97,0,
-18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,110,95,100,111,116,95,105,0,48,
-47,48,47,0,0,3,2,0,0,10,1,114,101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,0,40,0,9,18,114,
-101,116,118,97,108,0,58,118,101,99,50,0,17,48,0,48,0,0,0,0,20,0,9,18,114,101,116,118,97,108,0,18,
-101,116,97,0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,113,114,116,0,18,
-107,0,0,0,46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,0,0,11,0,114,101,102,114,97,99,
-116,0,1,1,0,0,11,73,0,0,1,1,0,0,11,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,0,0,9,1,110,95,100,111,
-116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,107,0,2,17,49,0,48,0,0,18,
-101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,110,95,100,111,
-116,95,105,0,48,47,48,47,0,0,3,2,0,0,11,1,114,101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,0,
-40,0,9,18,114,101,116,118,97,108,0,58,118,101,99,51,0,17,48,0,48,0,0,0,0,20,0,9,18,114,101,116,118,
-97,108,0,18,101,116,97,0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,113,
-114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,0,0,12,0,114,101,
-102,114,97,99,116,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,0,0,9,1,
-110,95,100,111,116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,0,0,9,1,107,0,2,17,49,
-0,48,0,0,18,101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,110,
-95,100,111,116,95,105,0,48,47,48,47,0,0,3,2,0,0,12,1,114,101,116,118,97,108,0,0,0,10,18,107,0,17,
-48,0,48,0,0,40,0,9,18,114,101,116,118,97,108,0,58,118,101,99,52,0,17,48,0,48,0,0,0,0,20,0,9,18,114,
-101,116,118,97,108,0,18,101,116,97,0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,
-58,115,113,114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,0,0,13,
-0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,13,109,0,0,1,0,0,0,13,110,0,0,0,1,
-8,58,109,97,116,50,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,
-110,0,16,10,49,0,57,48,0,0,0,0,1,0,0,14,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,
-0,0,0,14,109,0,0,1,0,0,0,14,110,0,0,0,1,8,58,109,97,116,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,
-48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,
-10,50,0,57,48,0,0,0,0,1,0,0,15,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,15,
-109,0,0,1,0,0,0,15,110,0,0,0,1,8,58,109,97,116,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,
-0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,
-48,0,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,0,0,2,0,108,101,115,115,84,104,97,
-110,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,108,101,115,115,84,104,97,110,0,
+111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,0,0,1,90,95,0,0,9,221,0,
+108,101,110,103,116,104,0,1,1,0,0,12,118,0,0,0,1,3,2,90,95,0,0,9,221,1,114,0,0,0,3,2,90,95,1,0,9,
+221,1,112,0,2,58,100,111,116,0,18,118,0,0,18,118,0,0,0,0,0,4,102,108,111,97,116,95,114,115,113,0,
+18,114,0,0,18,112,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,
+114,0,0,0,0,1,90,95,0,0,9,221,0,100,105,115,116,97,110,99,101,0,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,
+0,0,1,3,2,90,95,1,0,9,221,1,100,0,2,18,120,0,18,121,0,47,0,0,9,18,95,95,114,101,116,86,97,108,0,58,
+108,101,110,103,116,104,0,18,100,0,0,0,20,0,0,1,90,95,0,0,9,221,0,100,105,115,116,97,110,99,101,0,
+1,1,0,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,3,2,90,95,1,0,10,221,1,100,50,0,2,18,118,0,18,117,0,47,0,
+0,9,18,95,95,114,101,116,86,97,108,0,58,108,101,110,103,116,104,0,18,100,50,0,0,0,20,0,0,1,90,95,0,
+0,9,221,0,100,105,115,116,97,110,99,101,0,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,3,2,90,95,1,0,
+11,221,1,100,51,0,2,18,118,0,18,117,0,47,0,0,9,18,95,95,114,101,116,86,97,108,0,58,108,101,110,103,
+116,104,0,18,100,51,0,0,0,20,0,0,1,90,95,0,0,9,221,0,100,105,115,116,97,110,99,101,0,1,1,0,0,12,
+118,0,0,1,1,0,0,12,117,0,0,0,1,3,2,90,95,1,0,12,221,1,100,52,0,2,18,118,0,18,117,0,47,0,0,9,18,95,
+95,114,101,116,86,97,108,0,58,108,101,110,103,116,104,0,18,100,52,0,0,0,20,0,0,1,90,95,0,0,11,221,
+0,99,114,111,115,115,0,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,51,95,99,114,111,115,
+115,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,9,
+221,0,102,97,99,101,102,111,114,119,97,114,100,0,1,1,0,0,9,78,0,0,1,1,0,0,9,73,0,0,1,1,0,0,9,78,
+114,101,102,0,0,0,1,3,2,90,95,1,0,9,221,1,100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,
+0,0,0,3,2,90,95,0,0,9,221,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,
+18,100,0,0,0,8,58,109,105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,90,95,0,0,10,221,0,102,
+97,99,101,102,111,114,119,97,114,100,0,1,1,0,0,10,78,0,0,1,1,0,0,10,73,0,0,1,1,0,0,10,78,114,101,
+102,0,0,0,1,3,2,90,95,1,0,9,221,1,100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,
+2,90,95,0,0,9,221,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,18,100,
+0,0,0,8,58,109,105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,90,95,0,0,11,221,0,102,97,99,
+101,102,111,114,119,97,114,100,0,1,1,0,0,11,78,0,0,1,1,0,0,11,73,0,0,1,1,0,0,11,78,114,101,102,0,0,
+0,1,3,2,90,95,1,0,9,221,1,100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,2,90,95,
+0,0,9,221,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,18,100,0,0,0,8,
+58,109,105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,90,95,0,0,12,221,0,102,97,99,101,102,
+111,114,119,97,114,100,0,1,1,0,0,12,78,0,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,114,101,102,0,0,0,1,3,2,
+90,95,1,0,9,221,1,100,0,2,58,100,111,116,0,18,78,114,101,102,0,0,18,73,0,0,0,0,0,3,2,90,95,0,0,9,
+221,1,115,0,0,0,4,118,101,99,52,95,115,103,116,0,18,115,0,0,17,48,0,48,0,0,0,18,100,0,0,0,8,58,109,
+105,120,0,18,78,0,54,0,18,78,0,0,18,115,0,0,0,0,0,1,90,95,0,0,9,221,0,114,101,102,108,101,99,116,0,
+1,1,0,0,9,73,0,0,1,1,0,0,9,78,0,0,0,1,8,18,73,0,17,50,0,48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,
+0,0,48,18,78,0,48,47,0,0,1,90,95,0,0,10,221,0,114,101,102,108,101,99,116,0,1,1,0,0,10,73,0,0,1,1,0,
+0,10,78,0,0,0,1,8,18,73,0,17,50,0,48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,
+0,1,90,95,0,0,11,221,0,114,101,102,108,101,99,116,0,1,1,0,0,11,73,0,0,1,1,0,0,11,78,0,0,0,1,8,18,
+73,0,17,50,0,48,0,0,58,100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,0,1,90,95,0,0,12,221,
+0,114,101,102,108,101,99,116,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,0,0,0,1,8,18,73,0,17,50,0,48,0,0,58,
+100,111,116,0,18,78,0,0,18,73,0,0,0,48,18,78,0,48,47,0,0,1,90,95,0,0,9,221,0,114,101,102,114,97,99,
+116,0,1,1,0,0,9,73,0,0,1,1,0,0,9,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,90,95,0,0,9,221,1,110,95,
+100,111,116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,90,95,0,0,9,221,1,107,0,2,17,
+49,0,48,0,0,18,101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,
+110,95,100,111,116,95,105,0,48,47,48,47,0,0,3,2,90,95,0,0,9,221,1,114,101,116,118,97,108,0,0,0,10,
+18,107,0,17,48,0,48,0,0,40,0,9,18,114,101,116,118,97,108,0,17,48,0,48,0,0,20,0,9,18,114,101,116,
+118,97,108,0,18,101,116,97,0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,
+113,114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,90,95,0,0,10,
+221,0,114,101,102,114,97,99,116,0,1,1,0,0,10,73,0,0,1,1,0,0,10,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,
+3,2,90,95,0,0,9,221,1,110,95,100,111,116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,
+90,95,0,0,9,221,1,107,0,2,17,49,0,48,0,0,18,101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,
+95,100,111,116,95,105,0,18,110,95,100,111,116,95,105,0,48,47,48,47,0,0,3,2,90,95,0,0,10,221,1,114,
+101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,0,40,0,9,18,114,101,116,118,97,108,0,58,118,101,
+99,50,0,17,48,0,48,0,0,0,0,20,0,9,18,114,101,116,118,97,108,0,18,101,116,97,0,18,73,0,48,18,101,
+116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,113,114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,
+8,18,114,101,116,118,97,108,0,0,0,1,90,95,0,0,11,221,0,114,101,102,114,97,99,116,0,1,1,0,0,11,73,0,
+0,1,1,0,0,11,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,3,2,90,95,0,0,9,221,1,110,95,100,111,116,95,105,0,
+2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,90,95,0,0,9,221,1,107,0,2,17,49,0,48,0,0,18,101,
+116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,95,100,111,116,95,105,0,18,110,95,100,111,116,95,
+105,0,48,47,48,47,0,0,3,2,90,95,0,0,11,221,1,114,101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,
+0,40,0,9,18,114,101,116,118,97,108,0,58,118,101,99,51,0,17,48,0,48,0,0,0,0,20,0,9,18,114,101,116,
+118,97,108,0,18,101,116,97,0,18,73,0,48,18,101,116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,
+113,114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,8,18,114,101,116,118,97,108,0,0,0,1,90,95,0,0,12,
+221,0,114,101,102,114,97,99,116,0,1,1,0,0,12,73,0,0,1,1,0,0,12,78,0,0,1,1,0,0,9,101,116,97,0,0,0,1,
+3,2,90,95,0,0,9,221,1,110,95,100,111,116,95,105,0,2,58,100,111,116,0,18,78,0,0,18,73,0,0,0,0,0,3,2,
+90,95,0,0,9,221,1,107,0,2,17,49,0,48,0,0,18,101,116,97,0,18,101,116,97,0,48,17,49,0,48,0,0,18,110,
+95,100,111,116,95,105,0,18,110,95,100,111,116,95,105,0,48,47,48,47,0,0,3,2,90,95,0,0,12,221,1,114,
+101,116,118,97,108,0,0,0,10,18,107,0,17,48,0,48,0,0,40,0,9,18,114,101,116,118,97,108,0,58,118,101,
+99,52,0,17,48,0,48,0,0,0,0,20,0,9,18,114,101,116,118,97,108,0,18,101,116,97,0,18,73,0,48,18,101,
+116,97,0,18,110,95,100,111,116,95,105,0,48,58,115,113,114,116,0,18,107,0,0,0,46,18,78,0,48,47,20,0,
+8,18,114,101,116,118,97,108,0,0,0,1,90,95,0,0,13,221,0,109,97,116,114,105,120,67,111,109,112,77,
+117,108,116,0,1,0,0,0,13,109,0,0,1,0,0,0,13,110,0,0,0,1,8,58,109,97,116,50,0,18,109,0,16,8,48,0,57,
+18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,48,0,0,0,0,1,90,95,0,0,14,
+221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,14,109,0,0,1,0,0,0,14,110,0,0,
+0,1,8,58,109,97,116,51,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,57,
+18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,0,0,0,1,90,95,0,0,
+15,221,0,109,97,116,114,105,120,67,111,109,112,77,117,108,116,0,1,0,0,0,15,109,0,0,1,0,0,0,15,110,
+0,0,0,1,8,58,109,97,116,52,0,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,48,0,18,109,0,16,10,49,0,
+57,18,110,0,16,10,49,0,57,48,0,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,48,0,18,109,0,16,10,
+51,0,57,18,110,0,16,10,51,0,57,48,0,0,0,0,1,90,95,0,0,2,221,0,108,101,115,115,84,104,97,110,0,1,1,
+0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,97,
+108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,108,101,115,115,84,104,97,110,0,
 1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,
-97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,108,101,115,115,84,104,97,110,0,1,
-1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,
-97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,108,101,115,115,84,104,97,110,0,1,1,0,0,6,117,0,0,1,
-1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,
-0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,108,101,115,115,84,104,97,110,0,1,1,0,0,7,117,0,0,1,1,0,0,7,
-118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,
-18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,108,101,115,115,84,104,97,110,0,1,1,0,0,8,117,0,0,1,1,0,0,8,
-118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,
-0,0,0,0,1,0,0,2,0,108,101,115,115,84,104,97,110,69,113,117,97,108,0,1,1,0,0,10,117,0,0,1,1,0,0,10,
-118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,
-117,0,0,18,118,0,0,0,0,1,0,0,3,0,108,101,115,115,84,104,97,110,69,113,117,97,108,0,1,1,0,0,11,117,
-0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,116,86,97,108,0,59,
-120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,108,101,115,115,84,104,97,110,69,113,117,97,
-108,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,
-116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,108,101,115,115,84,104,97,110,69,113,117,97,
-108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,108,101,115,115,84,104,97,110,
-69,113,117,97,108,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,
-95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,108,101,115,115,
+97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,108,101,115,115,84,104,
+97,110,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,
+101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,108,101,115,115,84,104,97,110,
+0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,116,86,
+97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,108,101,115,115,84,104,97,
+110,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,108,101,115,115,
+84,104,97,110,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,108,116,0,18,95,95,
+114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,108,101,115,115,84,104,97,
+110,69,113,117,97,108,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,
+18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,108,
+101,115,115,84,104,97,110,69,113,117,97,108,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,
+99,52,95,115,108,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,
+0,1,90,95,0,0,4,221,0,108,101,115,115,84,104,97,110,69,113,117,97,108,0,1,1,0,0,12,117,0,0,1,1,0,0,
+12,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,
+118,0,0,0,0,1,90,95,0,0,2,221,0,108,101,115,115,84,104,97,110,69,113,117,97,108,0,1,1,0,0,6,117,0,
+0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,116,86,97,108,0,59,120,
+121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,108,101,115,115,84,104,97,110,69,113,117,97,
+108,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,108,101,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,108,101,115,115,
 84,104,97,110,69,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,
-108,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,103,114,101,97,
-116,101,114,84,104,97,110,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,103,
-116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,103,114,
-101,97,116,101,114,84,104,97,110,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,
-115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,
-4,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,
-99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,103,
-114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,
-115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,59,120,121,0,0,18,118,0,59,
-120,121,0,0,0,0,1,0,0,3,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,7,117,0,0,1,1,0,0,7,
-118,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,
-18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,8,117,0,0,1,
-1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,
-18,118,0,0,0,0,1,0,0,2,0,103,114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,10,
-117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,
-59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,103,114,101,97,116,101,114,84,104,97,110,69,113,
-117,97,108,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,103,114,101,97,116,
-101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,
-52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,103,114,
+108,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,103,114,
+101,97,116,101,114,84,104,97,110,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,
+115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,
+0,3,221,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,
+118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,
+118,0,0,0,0,1,90,95,0,0,4,221,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,12,117,0,0,1,1,
+0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,
+18,118,0,0,0,0,1,90,95,0,0,2,221,0,103,114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,6,117,0,0,1,
+1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,
+0,18,117,0,59,120,121,0,0,18,118,0,59,120,121,0,0,0,0,1,90,95,0,0,3,221,0,103,114,101,97,116,101,
+114,84,104,97,110,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,
+95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,103,
+114,101,97,116,101,114,84,104,97,110,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,
+115,103,116,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,103,
+114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,
+4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,
+0,0,0,0,1,90,95,0,0,3,221,0,103,114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,
+11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,
+0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,103,114,101,97,116,101,114,84,
+104,97,110,69,113,117,97,108,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,
+103,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,103,114,
 101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,
 101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,
-0,1,0,0,3,0,103,114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,7,117,0,0,1,1,0,0,
-7,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,
-18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,103,114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,
-1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,
-108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,101,113,117,97,108,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,
+0,1,90,95,0,0,3,221,0,103,114,101,97,116,101,114,84,104,97,110,69,113,117,97,108,0,1,1,0,0,7,117,0,
+0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,95,114,101,116,86,97,108,0,59,120,
+121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,103,114,101,97,116,101,114,84,104,97,110,
+69,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,103,101,0,18,95,
+95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,101,113,117,97,108,0,1,
+1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,
+97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,101,113,117,97,108,0,1,1,0,0,
+11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,
+0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,101,113,117,97,108,0,1,1,0,0,12,
+117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,0,
+18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,101,113,117,97,108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,
 0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,
-0,18,118,0,0,0,0,1,0,0,3,0,101,113,117,97,108,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,
-101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,
-0,0,0,1,0,0,4,0,101,113,117,97,108,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,
-115,101,113,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,101,113,117,
-97,108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,101,113,117,97,108,0,1,1,0,0,7,
-117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,
-120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,101,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,
-8,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,
-118,0,0,0,0,1,0,0,2,0,101,113,117,97,108,0,1,1,0,0,2,117,0,0,1,1,0,0,2,118,0,0,0,1,4,118,101,99,52,
-95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,
+0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,101,113,117,97,108,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,
+4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,
+118,0,0,0,0,1,90,95,0,0,4,221,0,101,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,
+101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,
+0,2,221,0,101,113,117,97,108,0,1,1,0,0,2,117,0,0,1,1,0,0,2,118,0,0,0,1,4,118,101,99,52,95,115,101,
+113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,
 0,101,113,117,97,108,0,1,1,0,0,3,117,0,0,1,1,0,0,3,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,
-95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,101,113,117,
-97,108,0,1,1,0,0,4,117,0,0,1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,
-116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,2,0,110,111,116,69,113,117,97,108,0,1,1,0,0,10,
-117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,
-59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,110,111,116,69,113,117,97,108,0,1,1,0,0,11,117,
-0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,
-120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,110,111,116,69,113,117,97,108,0,1,1,0,0,12,117,
-0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,
-117,0,0,18,118,0,0,0,0,1,0,0,2,0,110,111,116,69,113,117,97,108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,
-0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,
-18,118,0,0,0,0,1,0,0,3,0,110,111,116,69,113,117,97,108,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,
-118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,
-118,0,0,0,0,1,0,0,4,0,110,111,116,69,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,
-118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,
-2,0,110,111,116,69,113,117,97,108,0,1,1,0,0,2,117,0,0,1,1,0,0,2,118,0,0,0,1,4,118,101,99,52,95,115,
-110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,3,0,110,
-111,116,69,113,117,97,108,0,1,1,0,0,3,117,0,0,1,1,0,0,3,118,0,0,0,1,4,118,101,99,52,95,115,110,101,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,4,0,110,111,
-116,69,113,117,97,108,0,1,1,0,0,4,117,0,0,1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,
-18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,0,0,1,0,97,110,121,0,1,1,0,0,2,118,
-0,0,0,1,3,2,0,0,9,1,115,117,109,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,
-18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,
-97,108,0,59,120,0,0,18,115,117,109,0,59,120,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,0,97,110,121,0,1,1,0,
-0,3,118,0,0,0,1,3,2,0,0,9,1,115,117,109,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,59,
-120,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,
-59,120,0,0,18,115,117,109,0,59,120,0,0,18,118,0,59,122,0,0,0,4,118,101,99,52,95,115,110,101,0,18,
-95,95,114,101,116,86,97,108,0,59,120,0,0,18,115,117,109,0,59,120,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,
-0,97,110,121,0,1,1,0,0,4,118,0,0,0,1,3,2,0,0,9,1,115,117,109,0,0,0,4,118,101,99,52,95,97,100,100,0,
-18,115,117,109,0,59,120,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,97,100,
-100,0,18,115,117,109,0,59,120,0,0,18,115,117,109,0,59,120,0,0,18,118,0,59,122,0,0,0,4,118,101,99,
-52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,18,115,117,109,0,59,120,0,0,18,118,0,59,119,0,0,0,4,
-118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,115,117,109,0,59,120,
-0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,0,97,108,108,0,1,1,0,0,2,118,0,0,0,1,3,2,0,0,9,1,112,114,111,100,
-0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,118,0,59,120,
-0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,
-112,114,111,100,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,0,97,108,108,0,1,1,0,0,3,118,0,0,0,1,3,2,0,0,9,1,
-112,114,111,100,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,
-18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
-112,114,111,100,0,0,18,112,114,111,100,0,0,18,118,0,59,122,0,0,0,4,118,101,99,52,95,115,110,101,0,
-18,95,95,114,101,116,86,97,108,0,0,18,112,114,111,100,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,0,97,108,
-108,0,1,1,0,0,4,118,0,0,0,1,3,2,0,0,9,1,112,114,111,100,0,0,0,4,118,101,99,52,95,109,117,108,116,
-105,112,108,121,0,18,112,114,111,100,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,
-95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,112,114,111,100,0,0,18,118,0,59,122,
-0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,112,114,111,
-100,0,0,18,118,0,59,119,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,
-18,112,114,111,100,0,0,17,48,0,48,0,0,0,0,0,1,0,0,2,0,110,111,116,0,1,1,0,0,2,118,0,0,0,1,4,118,
-101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,17,48,0,48,0,
-0,0,0,0,1,0,0,3,0,110,111,116,0,1,1,0,0,3,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,4,0,110,111,116,0,
-1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,
-0,17,48,0,48,0,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,49,68,0,1,1,0,0,16,115,97,109,112,
-108,101,114,0,0,1,1,0,0,9,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,49,100,0,18,95,
-95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,
-12,0,116,101,120,116,117,114,101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,
-1,1,0,0,10,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,49,100,0,18,95,95,114,101,
-116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,59,120,121,121,121,0,0,
-0,0,1,0,0,12,0,116,101,120,116,117,114,101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,
-101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,49,100,0,18,
-95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,
-0,0,12,0,116,101,120,116,117,114,101,50,68,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,10,
-99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,50,100,0,18,95,95,114,101,116,86,97,108,
-0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,
-114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,
-100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,
-97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,59,120,121,122,122,0,0,0,0,1,0,0,12,0,116,101,
-120,116,117,114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,
-111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,50,100,0,18,95,95,114,101,116,86,97,108,
-0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,
-114,101,51,68,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,
-118,101,99,52,95,116,101,120,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,
-114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,51,68,80,114,111,106,
-0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,
-95,116,101,120,112,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
-18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,67,117,98,101,0,1,1,0,0,19,
-115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,
-120,99,117,98,101,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,
-111,114,100,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,49,68,0,1,1,0,0,20,115,97,109,112,108,101,
-114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,49,100,0,18,95,95,114,
-101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,
-115,104,97,100,111,119,49,68,80,114,111,106,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,12,
-99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,49,100,0,18,95,95,114,101,116,86,97,
-108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,115,104,97,100,
-111,119,50,68,0,1,1,0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,
-118,101,99,52,95,116,101,120,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,
-114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,50,68,80,114,111,106,0,1,1,
-0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,
-101,120,112,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,
-111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,50,68,82,101,99,116,0,1,1,0,0,22,
+95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,101,
+113,117,97,108,0,1,1,0,0,4,117,0,0,1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,
+114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,110,111,116,69,113,117,97,
+108,0,1,1,0,0,10,117,0,0,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,110,111,116,69,113,
+117,97,108,0,1,1,0,0,11,117,0,0,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,110,111,
+116,69,113,117,97,108,0,1,1,0,0,12,117,0,0,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,
+18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,110,111,116,69,
+113,117,97,108,0,1,1,0,0,6,117,0,0,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,110,111,116,
+69,113,117,97,108,0,1,1,0,0,7,117,0,0,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,
+95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,110,
+111,116,69,113,117,97,108,0,1,1,0,0,8,117,0,0,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,110,101,
+0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,2,221,0,110,111,116,69,
+113,117,97,108,0,1,1,0,0,2,117,0,0,1,1,0,0,2,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,3,221,0,110,111,116,
+69,113,117,97,108,0,1,1,0,0,3,117,0,0,1,1,0,0,3,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,
+95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,4,221,0,110,
+111,116,69,113,117,97,108,0,1,1,0,0,4,117,0,0,1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,110,101,
+0,18,95,95,114,101,116,86,97,108,0,0,18,117,0,0,18,118,0,0,0,0,1,90,95,0,0,1,221,0,97,110,121,0,1,
+1,0,0,2,118,0,0,0,1,3,2,90,95,0,0,9,221,1,115,117,109,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,
+117,109,0,59,120,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,115,110,101,0,18,
+95,95,114,101,116,86,97,108,0,59,120,0,0,18,115,117,109,0,59,120,0,0,17,48,0,48,0,0,0,0,0,1,90,95,
+0,0,1,221,0,97,110,121,0,1,1,0,0,3,118,0,0,0,1,3,2,90,95,0,0,9,221,1,115,117,109,0,0,0,4,118,101,
+99,52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,
+101,99,52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,18,115,117,109,0,59,120,0,0,18,118,0,59,122,
+0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,115,117,109,
+0,59,120,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,1,221,0,97,110,121,0,1,1,0,0,4,118,0,0,0,1,3,2,90,95,
+0,0,9,221,1,115,117,109,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,18,118,0,
+59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,59,120,0,0,18,
+115,117,109,0,59,120,0,0,18,118,0,59,122,0,0,0,4,118,101,99,52,95,97,100,100,0,18,115,117,109,0,59,
+120,0,0,18,115,117,109,0,59,120,0,0,18,118,0,59,119,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,
+95,114,101,116,86,97,108,0,59,120,0,0,18,115,117,109,0,59,120,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,
+1,221,0,97,108,108,0,1,1,0,0,2,118,0,0,0,1,3,2,90,95,0,0,9,221,1,112,114,111,100,0,0,0,4,118,101,
+99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,118,0,59,120,0,0,18,118,0,59,
+121,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,112,114,111,100,0,
+0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,1,221,0,97,108,108,0,1,1,0,0,3,118,0,0,0,1,3,2,90,95,0,0,9,221,
+1,112,114,111,100,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,
+0,18,118,0,59,120,0,0,18,118,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,
+18,112,114,111,100,0,0,18,112,114,111,100,0,0,18,118,0,59,122,0,0,0,4,118,101,99,52,95,115,110,101,
+0,18,95,95,114,101,116,86,97,108,0,0,18,112,114,111,100,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,1,221,
+0,97,108,108,0,1,1,0,0,4,118,0,0,0,1,3,2,90,95,0,0,9,221,1,112,114,111,100,0,0,0,4,118,101,99,52,
+95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,118,0,59,120,0,0,18,118,0,59,121,0,
+0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,0,0,18,112,114,111,100,
+0,0,18,118,0,59,122,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,112,114,111,100,
+0,0,18,112,114,111,100,0,0,18,118,0,59,119,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,
+116,86,97,108,0,0,18,112,114,111,100,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,2,221,0,110,111,116,0,1,
+1,0,0,2,118,0,0,0,1,4,118,101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,
+0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,3,221,0,110,111,116,0,1,1,0,0,3,118,0,0,0,1,4,118,
+101,99,52,95,115,101,113,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,17,48,0,
+48,0,0,0,0,0,1,90,95,0,0,4,221,0,110,111,116,0,1,1,0,0,4,118,0,0,0,1,4,118,101,99,52,95,115,101,
+113,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,12,221,0,116,
+101,120,116,117,114,101,49,68,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,9,99,111,111,114,
+100,0,0,0,1,4,118,101,99,52,95,116,101,120,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
+109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,
+101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,
+0,0,0,1,4,118,101,99,52,95,116,101,120,112,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
+109,112,108,101,114,0,0,18,99,111,111,114,100,0,59,120,121,121,121,0,0,0,0,1,90,95,0,0,12,221,0,
+116,101,120,116,117,114,101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,
+0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,49,100,0,18,95,95,114,101,116,
+86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,
+116,101,120,116,117,114,101,50,68,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,
+111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,
+115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,
+117,114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,
+114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,
+115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,59,120,121,122,122,0,0,0,0,1,90,95,0,0,12,
+221,0,116,101,120,116,117,114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,
+1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,50,100,0,18,95,95,114,101,
+116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,
+221,0,116,101,120,116,117,114,101,51,68,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,
+111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,51,100,0,18,95,95,114,101,116,86,97,108,0,0,
+18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,
+116,117,114,101,51,68,80,114,111,106,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,
+111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,51,100,0,18,95,95,114,101,116,86,97,108,0,0,
+18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,
+116,117,114,101,67,117,98,101,0,1,1,0,0,19,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,
+114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,99,117,98,101,0,18,95,95,114,101,116,86,97,108,0,0,
+18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,
+100,111,119,49,68,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,
+1,4,118,101,99,52,95,116,101,120,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,
+101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,49,68,80,114,
+111,106,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,
+101,99,52,95,116,101,120,112,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,
+114,0,0,18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,50,68,0,1,1,0,0,
+21,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,
+120,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,
+114,100,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,50,68,80,114,111,106,0,1,1,0,0,21,115,
+97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,
+112,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,
+114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,50,68,82,101,99,116,0,1,1,0,0,22,
 115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,
 120,95,114,101,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,
-111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,50,68,82,101,99,116,80,114,111,106,
-0,1,1,0,0,22,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,118,101,99,52,
-95,116,101,120,112,95,114,101,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,
-101,114,0,0,18,99,111,111,114,100,0,59,120,121,122,122,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,
-101,50,68,82,101,99,116,80,114,111,106,0,1,1,0,0,22,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,
-111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,95,114,101,99,116,0,18,95,95,114,101,
-116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,99,111,111,114,100,0,0,0,0,1,0,0,12,0,
-115,104,97,100,111,119,50,68,82,101,99,116,0,1,1,0,0,23,115,97,109,112,108,101,114,0,0,1,1,0,0,11,
-99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,95,114,101,99,116,0,18,95,95,114,101,116,
-86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,12,0,115,104,
-97,100,111,119,50,68,82,101,99,116,80,114,111,106,0,1,1,0,0,23,115,97,109,112,108,101,114,0,0,1,1,
-0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,95,114,101,99,116,0,18,95,95,
-114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,0,0,0,1,0,0,9,
-0,110,111,105,115,101,49,0,1,1,0,0,9,120,0,0,0,1,4,102,108,111,97,116,95,110,111,105,115,101,49,0,
-18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,9,0,110,111,105,115,101,49,0,1,1,0,0,10,
-120,0,0,0,1,4,102,108,111,97,116,95,110,111,105,115,101,50,0,18,95,95,114,101,116,86,97,108,0,0,18,
-120,0,0,0,0,1,0,0,9,0,110,111,105,115,101,49,0,1,1,0,0,11,120,0,0,0,1,4,102,108,111,97,116,95,110,
-111,105,115,101,51,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,9,0,110,111,105,115,
-101,49,0,1,1,0,0,12,120,0,0,0,1,4,102,108,111,97,116,95,110,111,105,115,101,52,0,18,95,95,114,101,
-116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,10,0,110,111,105,115,101,50,0,1,1,0,0,9,120,0,0,0,1,9,18,95,
+111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,50,68,82,101,99,116,80,
+114,111,106,0,1,1,0,0,22,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,
+118,101,99,52,95,116,101,120,112,95,114,101,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
+109,112,108,101,114,0,0,18,99,111,111,114,100,0,59,120,121,122,122,0,0,0,0,1,90,95,0,0,12,221,0,
+116,101,120,116,117,114,101,50,68,82,101,99,116,80,114,111,106,0,1,1,0,0,22,115,97,109,112,108,101,
+114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,112,95,114,101,99,116,
+0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,99,111,111,114,100,0,
+0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,50,68,82,101,99,116,0,1,1,0,0,23,115,97,109,112,
+108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,120,95,114,101,99,
+116,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,0,
+0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,50,68,82,101,99,116,80,114,111,106,0,1,1,0,0,23,
+115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,0,1,4,118,101,99,52,95,116,101,
+120,112,95,114,101,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
+18,99,111,111,114,100,0,0,0,0,1,90,95,0,0,9,221,0,110,111,105,115,101,49,0,1,1,0,0,9,120,0,0,0,1,4,
+102,108,111,97,116,95,110,111,105,115,101,49,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,
+90,95,0,0,9,221,0,110,111,105,115,101,49,0,1,1,0,0,10,120,0,0,0,1,4,102,108,111,97,116,95,110,111,
+105,115,101,50,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,0,0,9,221,0,110,111,105,
+115,101,49,0,1,1,0,0,11,120,0,0,0,1,4,102,108,111,97,116,95,110,111,105,115,101,51,0,18,95,95,114,
+101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,0,0,9,221,0,110,111,105,115,101,49,0,1,1,0,0,12,120,0,
+0,0,1,4,102,108,111,97,116,95,110,111,105,115,101,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,
+0,0,0,1,90,95,0,0,10,221,0,110,111,105,115,101,50,0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,17,49,57,0,51,52,0,0,46,0,0,20,0,0,1,90,95,0,0,
+10,221,0,110,111,105,115,101,50,0,1,1,0,0,10,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,
+0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,
+111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,0,46,0,
+0,20,0,0,1,90,95,0,0,10,221,0,110,111,105,115,101,50,0,1,1,0,0,11,120,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,49,57,0,51,52,0,0,0,
+17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,20,0,0,1,90,95,0,0,10,221,0,110,111,105,115,101,
+50,0,1,1,0,0,12,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,
+0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,
+0,58,118,101,99,52,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,17,50,0,55,55,
+0,0,0,0,46,0,0,20,0,0,1,90,95,0,0,11,221,0,110,111,105,115,101,51,0,1,1,0,0,9,120,0,0,0,1,9,18,95,
 95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,
 101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,17,49,57,0,51,52,0,0,46,0,0,20,0,
-0,1,0,0,10,0,110,111,105,115,101,50,0,1,1,0,0,10,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
-120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,
-110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,0,
-46,0,0,20,0,0,1,0,0,10,0,110,111,105,115,101,50,0,1,1,0,0,11,120,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
-59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,49,57,0,51,52,0,0,0,17,55,0,54,
-54,0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,20,0,0,1,0,0,10,0,110,111,105,115,101,50,0,1,1,0,0,12,120,0,
-0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,
-18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,52,0,
-17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,17,50,0,55,55,0,0,0,0,46,0,0,20,0,0,
-1,0,0,11,0,110,111,105,115,101,51,0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
-120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,
-110,111,105,115,101,49,0,18,120,0,17,49,57,0,51,52,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,17,53,0,52,55,0,0,46,0,0,20,0,0,1,0,0,11,0,110,
-111,105,115,101,51,0,1,1,0,0,10,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,
-105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,
-101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,0,46,0,0,20,0,9,18,
-95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,
-53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,0,46,0,0,20,0,0,1,0,0,11,0,110,111,105,115,101,51,0,1,1,0,
-0,11,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,
-0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,
-101,99,51,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,20,0,9,18,95,
-95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,53,
-0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,0,46,0,0,20,0,0,1,0,0,11,0,110,111,105,
-115,101,51,0,1,1,0,0,12,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,
-101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,
-18,120,0,58,118,101,99,52,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,17,50,0,
-55,55,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,
-18,120,0,58,118,101,99,52,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,17,
-49,51,0,49,57,0,0,0,0,46,0,0,20,0,0,1,0,0,12,0,110,111,105,115,101,52,0,1,1,0,0,9,120,0,0,0,1,9,18,
+9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,17,53,0,52,55,0,0,
+46,0,0,20,0,0,1,90,95,0,0,11,221,0,110,111,105,115,101,51,0,1,1,0,0,10,120,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,
+86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,0,51,52,0,0,
+0,17,55,0,54,54,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,
+101,49,0,18,120,0,58,118,101,99,50,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,0,46,0,0,20,0,0,1,
+90,95,0,0,11,221,0,110,111,105,115,101,51,0,1,1,0,0,11,120,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,
+121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,
+0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,
+115,101,49,0,18,120,0,58,118,101,99,51,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,
+52,0,0,0,0,46,0,0,20,0,0,1,90,95,0,0,11,221,0,110,111,105,115,101,51,0,1,1,0,0,12,120,0,0,0,1,9,18,
 95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,17,49,57,0,51,52,0,0,46,0,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,17,53,0,52,
-55,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,
-0,17,50,51,0,53,52,0,0,46,0,0,20,0,0,1,0,0,12,0,110,111,105,115,101,52,0,1,1,0,0,10,120,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,
-0,51,52,0,0,0,17,55,0,54,54,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,
-111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,0,46,0,
-0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,
-99,50,0,17,50,51,0,53,52,0,0,0,17,50,57,0,49,49,0,0,0,0,46,0,0,20,0,0,1,0,0,12,0,110,111,105,115,
-101,52,0,1,1,0,0,11,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,
-49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,
-120,0,58,118,101,99,51,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,
-99,51,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,0,46,0,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,50,51,
-0,53,52,0,0,0,17,50,57,0,49,49,0,0,0,17,51,49,0,57,49,0,0,0,0,46,0,0,20,0,0,1,0,0,12,0,110,111,105,
-115,101,52,0,1,1,0,0,12,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,
-101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,
-18,120,0,58,118,101,99,52,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,17,50,0,
-55,55,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,
-18,120,0,58,118,101,99,52,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,17,
-49,51,0,49,57,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,
-101,49,0,18,120,0,58,118,101,99,52,0,17,50,51,0,53,52,0,0,0,17,50,57,0,49,49,0,0,0,17,51,49,0,57,
-49,0,0,0,17,51,55,0,52,56,0,0,0,0,46,0,0,20,0,0,0
+114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,52,0,17,49,57,
+0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,17,50,0,55,55,0,0,0,0,46,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,52,0,17,53,0,
+52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,17,49,51,0,49,57,0,0,0,0,46,0,0,20,0,0,1,
+90,95,0,0,12,221,0,110,111,105,115,101,52,0,1,1,0,0,9,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,
+0,58,110,111,105,115,101,49,0,18,120,0,17,49,57,0,51,52,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,17,53,0,52,55,0,0,46,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,0,17,50,51,0,53,52,0,0,46,0,0,
+20,0,0,1,90,95,0,0,12,221,0,110,111,105,115,101,52,0,1,1,0,0,10,120,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,59,120,0,58,110,111,105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,
+108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,49,57,0,51,52,0,0,0,17,
+55,0,54,54,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,
+49,0,18,120,0,58,118,101,99,50,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,0,46,0,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,50,0,17,50,
+51,0,53,52,0,0,0,17,50,57,0,49,49,0,0,0,0,46,0,0,20,0,0,1,90,95,0,0,12,221,0,110,111,105,115,101,
+52,0,1,1,0,0,11,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,105,115,101,49,
+0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,101,49,0,18,120,
+0,58,118,101,99,51,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,0,0,46,0,0,20,0,
+9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,
+0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,0,0,0,46,0,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,119,0,58,110,111,105,115,101,49,0,18,120,0,58,118,101,99,51,0,17,50,51,0,53,
+52,0,0,0,17,50,57,0,49,49,0,0,0,17,51,49,0,57,49,0,0,0,0,46,0,0,20,0,0,1,90,95,0,0,12,221,0,110,
+111,105,115,101,52,0,1,1,0,0,12,120,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,110,111,
+105,115,101,49,0,18,120,0,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,110,111,105,115,
+101,49,0,18,120,0,58,118,101,99,52,0,17,49,57,0,51,52,0,0,0,17,55,0,54,54,0,0,0,17,51,0,50,51,0,0,
+0,17,50,0,55,55,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,110,111,105,115,
+101,49,0,18,120,0,58,118,101,99,52,0,17,53,0,52,55,0,0,0,17,49,55,0,56,53,0,0,0,17,49,49,0,48,52,0,
+0,0,17,49,51,0,49,57,0,0,0,0,46,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,110,111,
+105,115,101,49,0,18,120,0,58,118,101,99,52,0,17,50,51,0,53,52,0,0,0,17,50,57,0,49,49,0,0,0,17,51,
+49,0,57,49,0,0,0,17,51,55,0,52,56,0,0,0,0,46,0,0,20,0,0,0
diff --git a/src/mesa/shader/slang/library/slang_core_gc.h b/src/mesa/shader/slang/library/slang_core_gc.h
index 125dd8dcce5..3cf1911a5d8 100644
--- a/src/mesa/shader/slang/library/slang_core_gc.h
+++ b/src/mesa/shader/slang/library/slang_core_gc.h
@@ -2,127 +2,134 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_core.gc */
 
-4,1,0,0,5,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,
-101,116,86,97,108,0,0,18,102,0,0,0,0,1,0,0,5,1,1,1,0,0,1,98,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,18,98,0,20,0,0,1,0,0,5,1,1,1,0,0,5,105,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,105,0,
-20,0,0,1,0,0,1,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,
-108,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,
-101,0,18,95,95,114,101,116,86,97,108,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,0,0,1,1,1,1,0,0,1,98,0,
-0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,98,0,20,0,0,1,0,0,9,1,1,1,0,0,5,105,0,0,0,1,4,105,118,
-101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,105,0,0,0,0,1,0,0,9,
-1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,
-97,108,0,0,18,98,0,0,0,0,1,0,0,9,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,102,
-0,20,0,0,1,0,0,10,1,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
-120,0,18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,0,1,0,0,10,1,1,1,0,0,
-9,102,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,
-18,102,0,0,0,0,1,0,0,10,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,
-18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,105,0,0,0,0,1,0,0,10,1,1,1,0,0,1,98,0,0,0,1,4,
-105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,
-98,0,0,0,0,1,0,0,10,1,1,1,0,0,2,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,
-95,114,101,116,86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,10,1,1,1,0,0,11,118,0,0,0,1,4,118,
-101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,
-0,0,0,0,1,0,0,10,1,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,0,0,1,0,0,11,1,1,1,0,0,9,120,0,0,1,1,0,0,9,
-121,0,0,1,1,0,0,9,122,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,120,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,122,
-0,20,0,0,1,0,0,11,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,122,0,0,18,102,0,0,0,0,1,0,0,11,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,
-99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,105,0,0,
-0,0,1,0,0,11,1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,11,1,1,1,0,0,3,98,0,0,0,1,4,105,118,101,
-99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,
-0,0,1,0,0,11,1,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,
-97,108,0,59,120,121,122,0,0,18,118,0,0,0,0,1,0,0,12,1,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,1,1,0,0,
-9,122,0,0,1,1,0,0,9,119,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,120,0,20,0,9,18,95,
-95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,
-122,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,119,0,20,0,0,1,0,0,12,1,1,1,0,0,9,102,0,
-0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,0,18,102,0,0,0,0,1,0,0,
-12,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,
-86,97,108,0,0,18,105,0,0,0,0,1,0,0,12,1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,
-101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,98,0,0,0,0,1,0,0,12,1,1,1,0,0,4,98,0,0,0,1,4,105,
-118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,98,0,0,0,0,1,0,0,
-12,1,1,1,0,0,8,105,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,
-86,97,108,0,0,18,105,0,0,0,0,1,0,0,12,1,1,1,0,0,11,118,51,0,0,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,
+4,1,90,95,0,0,5,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,
+95,95,114,101,116,86,97,108,0,0,18,102,0,0,0,0,1,90,95,0,0,5,221,1,1,1,0,0,1,98,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,18,98,0,20,0,0,1,90,95,0,0,5,221,1,1,1,0,0,5,105,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,18,105,0,20,0,0,1,90,95,0,0,1,221,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,
+115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,1,221,
+1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,0,18,102,
+0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,1,221,1,1,1,0,0,1,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,18,98,0,20,0,0,1,90,95,0,0,9,221,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,
+101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,105,0,0,0,0,1,90,95,0,0,9,221,1,1,1,0,0,1,98,0,0,
+0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,98,0,0,
+0,0,1,90,95,0,0,9,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,102,0,20,0,0,1,
+90,95,0,0,10,221,1,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
+120,0,18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,0,1,90,95,0,0,10,221,
+1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,
+120,121,0,0,18,102,0,0,0,0,1,90,95,0,0,10,221,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,52,95,116,
+111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,105,0,0,0,0,1,90,95,0,0,
+10,221,1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,10,221,1,1,1,0,0,2,98,0,0,0,1,4,105,118,
+101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,98,0,0,
+0,0,1,90,95,0,0,10,221,1,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,
+101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,0,0,1,90,95,0,0,10,221,1,1,1,0,0,12,118,
+0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,
+118,0,59,120,121,0,0,0,0,1,90,95,0,0,11,221,1,1,1,0,0,9,120,0,0,1,1,0,0,9,121,0,0,1,1,0,0,9,122,0,
+0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
+59,121,0,18,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,122,0,20,0,0,1,90,95,0,0,11,
+221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,
+59,120,121,122,0,0,18,102,0,0,0,0,1,90,95,0,0,11,221,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,52,
+95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,105,0,0,0,0,1,
+90,95,0,0,11,221,1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,11,221,1,1,1,0,0,3,98,0,0,0,1,
+4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,
+0,0,18,98,0,0,0,0,1,90,95,0,0,11,221,1,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,
+18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,0,0,1,90,95,0,0,12,221,1,1,1,0,0,9,
+120,0,0,1,1,0,0,9,121,0,0,1,1,0,0,9,122,0,0,1,1,0,0,9,119,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,59,120,0,18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,122,0,18,122,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,119,0,20,
+0,0,1,90,95,0,0,12,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,
+101,116,86,97,108,0,0,18,102,0,0,0,0,1,90,95,0,0,12,221,1,1,1,0,0,5,105,0,0,0,1,4,105,118,101,99,
+52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,105,0,0,0,0,1,90,95,0,0,12,
+221,1,1,1,0,0,1,98,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,
+86,97,108,0,0,18,98,0,0,0,0,1,90,95,0,0,12,221,1,1,1,0,0,4,98,0,0,0,1,4,105,118,101,99,52,95,116,
+111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,98,0,0,0,0,1,90,95,0,0,12,221,1,1,1,0,
+0,8,105,0,0,0,1,4,105,118,101,99,52,95,116,111,95,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,
+0,18,105,0,0,0,0,1,90,95,0,0,12,221,1,1,1,0,0,11,118,51,0,0,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,
 101,116,86,97,108,0,59,120,121,122,0,18,118,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,
-18,102,0,20,0,0,1,0,0,12,1,1,1,0,0,10,118,50,0,0,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,50,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,59,120,121,0,18,118,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
-59,122,0,18,102,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,102,50,0,20,0,0,1,0,0,6,1,
-1,1,0,0,5,105,0,0,1,1,0,0,5,106,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,105,0,20,0,
-9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,106,0,20,0,0,1,0,0,6,1,1,1,0,0,5,105,0,0,0,1,4,118,
-101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,105,0,0,0,0,1,0,
-0,6,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,
-86,97,108,0,59,120,121,0,0,18,102,0,0,0,0,1,0,0,6,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,95,116,
-111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,7,1,
-1,1,0,0,5,105,0,0,1,1,0,0,5,106,0,0,1,1,0,0,5,107,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
-120,0,18,105,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,106,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,122,0,18,107,0,20,0,0,1,0,0,7,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,109,
-111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,105,0,0,0,0,1,0,0,7,1,1,1,0,0,
-9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,
-59,120,121,122,0,0,18,102,0,0,0,0,1,0,0,7,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,95,109,111,118,
-101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,8,1,1,1,0,0,5,120,0,
-0,1,1,0,0,5,121,0,0,1,1,0,0,5,122,0,0,1,1,0,0,5,119,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
-120,0,18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,9,18,95,95,114,101,
-116,86,97,108,0,59,122,0,18,122,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,119,0,20,0,0,
-1,0,0,8,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,
-108,0,0,18,105,0,0,0,0,1,0,0,8,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,
-99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,102,0,0,0,0,1,0,0,8,1,1,1,0,0,1,98,0,0,0,1,4,118,101,
-99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,98,0,0,0,0,1,0,0,2,1,
-1,1,0,0,1,98,49,0,0,1,1,0,0,1,98,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,98,49,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,98,50,0,20,0,0,1,0,0,2,1,1,1,0,0,5,105,49,0,0,
-1,1,0,0,5,105,50,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,
-0,0,18,105,49,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,
-108,0,59,121,0,0,18,105,50,0,0,17,48,0,48,0,0,0,0,0,1,0,0,2,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,
-95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,2,1,1,1,0,
-0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,
-18,102,0,0,17,48,0,48,0,0,0,0,0,1,0,0,2,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,110,101,0,
-18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,0,0,2,1,1,1,0,0,
-10,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,
-118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,2,1,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,
-95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,3,1,1,1,0,0,1,
-98,49,0,0,1,1,0,0,1,98,50,0,0,1,1,0,0,1,98,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,
-18,98,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,98,50,0,20,0,9,18,95,95,114,101,116,
-86,97,108,0,59,122,0,18,98,51,0,20,0,0,1,0,0,3,1,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,50,0,0,1,1,0,0,
-9,102,51,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
-102,49,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,
-121,0,0,18,102,50,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,
-97,108,0,59,122,0,0,18,102,51,0,0,17,48,0,48,0,0,0,0,0,1,0,0,3,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,
-52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,3,
-1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,
-121,122,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,0,0,3,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,
-110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,0,
-0,3,1,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,
-120,121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,3,1,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,
-115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,
-1,0,0,4,1,1,1,0,0,1,98,49,0,0,1,1,0,0,1,98,50,0,0,1,1,0,0,1,98,51,0,0,1,1,0,0,1,98,52,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,59,120,0,18,98,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,
-18,98,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,98,51,0,20,0,9,18,95,95,114,101,116,
-86,97,108,0,59,119,0,18,98,52,0,20,0,0,1,0,0,4,1,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,50,0,0,1,1,0,0,
-9,102,51,0,0,1,1,0,0,9,102,52,0,0,0,1,3,2,1,0,9,1,122,101,114,111,0,2,17,48,0,48,0,0,0,0,4,118,101,
-99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,102,49,0,0,18,122,101,114,
-111,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,102,50,0,
-0,18,122,101,114,111,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,
-122,0,0,18,102,51,0,0,18,122,101,114,111,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,
-116,86,97,108,0,59,119,0,0,18,102,52,0,0,18,122,101,114,111,0,0,0,0,1,0,0,4,1,1,1,0,0,1,98,0,0,0,1,
+18,102,0,20,0,0,1,90,95,0,0,12,221,1,1,1,0,0,10,118,50,0,0,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,50,0,
+0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,121,0,18,118,50,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,59,122,0,18,102,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,102,50,0,20,0,0,
+1,90,95,0,0,6,221,1,1,1,0,0,5,105,0,0,1,1,0,0,5,106,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,
+120,0,18,105,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,106,0,20,0,0,1,90,95,0,0,6,221,
+1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,
+120,121,0,0,18,105,0,0,0,0,1,90,95,0,0,6,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,
+105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,102,0,0,0,0,1,90,95,0,0,6,
+221,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,
+86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,7,221,1,1,1,0,0,5,105,0,0,1,1,0,0,5,106,0,0,1,
+1,0,0,5,107,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,18,105,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,59,121,0,18,106,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,18,107,0,20,0,0,
+1,90,95,0,0,7,221,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,122,0,0,18,105,0,0,0,0,1,90,95,0,0,7,221,1,1,1,0,0,9,102,0,0,0,1,4,118,
+101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,
+102,0,0,0,0,1,90,95,0,0,7,221,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,8,221,1,1,1,0,0,5,120,0,0,1,1,
+0,0,5,121,0,0,1,1,0,0,5,122,0,0,1,1,0,0,5,119,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,
+18,120,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,121,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,59,122,0,18,122,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,119,0,20,0,0,1,90,
+95,0,0,8,221,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,
+97,108,0,0,18,105,0,0,0,0,1,90,95,0,0,8,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,116,111,95,
+105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,102,0,0,0,0,1,90,95,0,0,8,221,1,1,1,0,0,
+1,98,0,0,0,1,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,
+18,98,0,0,0,0,1,90,95,0,0,2,221,1,1,1,0,0,1,98,49,0,0,1,1,0,0,1,98,50,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,59,120,0,18,98,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,98,50,0,20,
+0,0,1,90,95,0,0,2,221,1,1,1,0,0,5,105,49,0,0,1,1,0,0,5,105,50,0,0,0,1,4,118,101,99,52,95,115,110,
+101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,105,49,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,
+95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,105,50,0,0,17,48,0,48,0,0,0,0,0,1,
+90,95,0,0,2,221,1,1,1,0,0,1,98,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,
+86,97,108,0,59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,2,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,
+95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,
+90,95,0,0,2,221,1,1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,
+97,108,0,59,120,121,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,2,221,1,1,1,0,0,10,118,0,0,0,1,
+4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,17,48,
+0,48,0,0,0,0,0,1,90,95,0,0,2,221,1,1,1,0,0,6,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,
+114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,3,221,1,1,1,0,0,
+1,98,49,0,0,1,1,0,0,1,98,50,0,0,1,1,0,0,1,98,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,
+0,18,98,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,98,50,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,59,122,0,18,98,51,0,20,0,0,1,90,95,0,0,3,221,1,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,
+50,0,0,1,1,0,0,9,102,51,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,
+59,120,0,0,18,102,49,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,
+86,97,108,0,59,121,0,0,18,102,50,0,0,17,48,0,48,0,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,
+114,101,116,86,97,108,0,59,122,0,0,18,102,51,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,3,221,1,1,1,0,0,
+1,98,0,0,0,1,4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,
+0,0,18,98,0,0,0,0,1,90,95,0,0,3,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,
+95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,3,221,1,
+1,1,0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,
+122,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,3,221,1,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,
+95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,
+0,1,90,95,0,0,3,221,1,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,
+86,97,108,0,59,120,121,122,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,4,221,1,1,1,0,0,1,98,49,
+0,0,1,1,0,0,1,98,50,0,0,1,1,0,0,1,98,51,0,0,1,1,0,0,1,98,52,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,59,120,0,18,98,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,18,98,50,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,59,122,0,18,98,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,18,
+98,52,0,20,0,0,1,90,95,0,0,4,221,1,1,1,0,0,9,102,49,0,0,1,1,0,0,9,102,50,0,0,1,1,0,0,9,102,51,0,0,
+1,1,0,0,9,102,52,0,0,0,1,3,2,90,95,1,0,9,221,1,122,101,114,111,0,2,17,48,0,48,0,0,0,0,4,118,101,99,
+52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,102,49,0,0,18,122,101,114,111,0,
+0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,102,50,0,0,18,
+122,101,114,111,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,
+18,102,51,0,0,18,122,101,114,111,0,0,0,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,
+108,0,59,119,0,0,18,102,52,0,0,18,122,101,114,111,0,0,0,0,1,90,95,0,0,4,221,1,1,1,0,0,1,98,0,0,0,1,
 4,118,101,99,52,95,109,111,118,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,119,0,0,18,98,
-0,0,0,0,1,0,0,4,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,
-97,108,0,59,120,121,122,119,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,0,0,4,1,1,1,0,0,5,105,0,0,0,1,4,
-118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,119,0,0,18,105,0,0,
-17,48,0,48,0,0,0,0,0,1,0,0,4,1,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,119,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,4,1,1,1,0,0,8,
-118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,119,0,
-0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,0,0,13,1,1,1,0,0,9,109,48,48,0,0,1,1,0,0,9,109,49,48,0,0,1,1,0,
-0,9,109,48,49,0,0,1,1,0,0,9,109,49,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,
-120,0,18,109,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,109,49,48,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,109,48,49,0,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,109,49,49,0,20,0,0,1,0,0,13,1,1,1,0,0,9,102,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,102,0,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,8,48,0,57,59,121,0,17,48,0,48,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,59,120,0,17,48,0,48,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,0,
-20,0,0,1,0,0,13,1,1,1,0,0,5,105,0,0,0,1,8,58,109,97,116,50,0,58,102,108,111,97,116,0,18,105,0,0,0,
-0,0,0,0,1,0,0,13,1,1,1,0,0,1,98,0,0,0,1,8,58,109,97,116,50,0,58,102,108,111,97,116,0,18,98,0,0,0,0,
-0,0,0,1,0,0,13,1,1,1,0,0,10,99,48,0,0,1,1,0,0,10,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,
-0,0,14,1,1,1,0,0,9,109,48,48,0,0,1,1,0,0,9,109,49,48,0,0,1,1,0,0,9,109,50,48,0,0,1,1,0,0,9,109,48,
+0,0,0,0,1,90,95,0,0,4,221,1,1,1,0,0,9,102,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,
+101,116,86,97,108,0,59,120,121,122,119,0,0,18,102,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,4,221,1,1,1,
+0,0,5,105,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,
+119,0,0,18,105,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,4,221,1,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,
+95,115,110,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,119,0,0,18,118,0,0,17,48,0,48,0,0,
+0,0,0,1,90,95,0,0,4,221,1,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,115,110,101,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,122,119,0,0,18,118,0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,13,221,1,1,1,0,
+0,9,109,48,48,0,0,1,1,0,0,9,109,49,48,0,0,1,1,0,0,9,109,48,49,0,0,1,1,0,0,9,109,49,49,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,109,48,48,0,20,0,9,18,95,95,114,101,116,86,
+97,108,0,16,8,48,0,57,59,121,0,18,109,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
+57,59,120,0,18,109,48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,109,
+49,49,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,9,102,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,59,120,0,18,102,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,17,48,0,48,
+0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,17,48,0,48,0,0,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,49,0,57,59,121,0,18,102,0,20,0,0,1,90,95,0,0,13,221,1,1,1,0,0,5,105,
+0,0,0,1,8,58,109,97,116,50,0,58,102,108,111,97,116,0,18,105,0,0,0,0,0,0,0,1,90,95,0,0,13,221,1,1,1,
+0,0,1,98,0,0,0,1,8,58,109,97,116,50,0,58,102,108,111,97,116,0,18,98,0,0,0,0,0,0,0,1,90,95,0,0,13,
+221,1,1,1,0,0,10,99,48,0,0,1,1,0,0,10,99,49,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,0,1,90,95,0,0,
+14,221,1,1,1,0,0,9,109,48,48,0,0,1,1,0,0,9,109,49,48,0,0,1,1,0,0,9,109,50,48,0,0,1,1,0,0,9,109,48,
 49,0,0,1,1,0,0,9,109,49,49,0,0,1,1,0,0,9,109,50,49,0,0,1,1,0,0,9,109,48,50,0,0,1,1,0,0,9,109,49,50,
 0,0,1,1,0,0,9,109,50,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,120,0,18,109,48,
 48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,109,49,48,0,20,0,9,18,95,95,
@@ -131,703 +138,738 @@
 121,0,18,109,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,0,18,109,50,49,0,
 20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,18,109,48,50,0,20,0,9,18,95,95,114,
 101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,109,49,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,50,0,57,59,122,0,18,109,50,50,0,20,0,0,1,0,0,14,1,1,1,0,0,9,102,0,0,0,1,3,2,0,0,10,1,118,0,2,
-58,118,101,99,50,0,18,102,0,0,17,48,0,48,0,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,118,0,59,120,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,118,0,59,121,
-120,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,118,0,59,121,121,120,0,20,0,0,1,
-0,0,14,1,1,1,0,0,5,105,0,0,0,1,8,58,109,97,116,51,0,58,102,108,111,97,116,0,18,105,0,0,0,0,0,0,0,1,
-0,0,14,1,1,1,0,0,1,98,0,0,0,1,8,58,109,97,116,51,0,58,102,108,111,97,116,0,18,98,0,0,0,0,0,0,0,1,0,
-0,14,1,1,1,0,0,11,99,48,0,0,1,1,0,0,11,99,49,0,0,1,1,0,0,11,99,50,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,
-0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,0,0,1,0,0,15,1,1,1,0,0,9,109,
-48,48,0,0,1,1,0,0,9,109,49,48,0,0,1,1,0,0,9,109,50,48,0,0,1,1,0,0,9,109,51,48,0,0,1,1,0,0,9,109,48,
-49,0,0,1,1,0,0,9,109,49,49,0,0,1,1,0,0,9,109,50,49,0,0,1,1,0,0,9,109,51,49,0,0,1,1,0,0,9,109,48,50,
-0,0,1,1,0,0,9,109,49,50,0,0,1,1,0,0,9,109,50,50,0,0,1,1,0,0,9,109,51,50,0,0,1,1,0,0,9,109,48,51,0,
-0,1,1,0,0,9,109,49,51,0,0,1,1,0,0,9,109,50,51,0,0,1,1,0,0,9,109,51,51,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,59,120,0,18,109,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,
-48,0,57,59,121,0,18,109,49,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,
-109,50,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,119,0,18,109,51,48,0,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,120,0,18,109,48,49,0,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,10,49,0,57,59,121,0,18,109,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,59,122,0,18,109,50,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,0,18,109,
-51,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,120,0,18,109,48,50,0,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,50,0,57,59,121,0,18,109,49,50,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,50,0,57,59,122,0,18,109,50,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,
-59,119,0,18,109,51,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,120,0,18,109,48,
-51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,121,0,18,109,49,51,0,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,51,0,57,59,122,0,18,109,50,51,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,51,0,57,59,119,0,18,109,51,51,0,20,0,0,1,0,0,15,1,1,1,0,0,9,102,0,0,0,1,3,2,0,0,10,1,
-118,0,2,58,118,101,99,50,0,18,102,0,0,17,48,0,48,0,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,16,
-8,48,0,57,18,118,0,59,120,121,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
-118,0,59,121,120,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,118,0,59,121,
-121,120,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,118,0,59,121,121,121,120,0,
-20,0,0,1,0,0,15,1,1,1,0,0,5,105,0,0,0,1,8,58,109,97,116,52,0,58,102,108,111,97,116,0,18,105,0,0,0,
-0,0,0,0,1,0,0,15,1,1,1,0,0,1,98,0,0,0,1,8,58,109,97,116,52,0,58,102,108,111,97,116,0,18,98,0,0,0,0,
-0,0,0,1,0,0,15,1,1,1,0,0,12,99,48,0,0,1,1,0,0,12,99,49,0,0,1,1,0,0,12,99,50,0,0,1,1,0,0,12,99,51,0,
-0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,99,50,0,20,
-0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,99,51,0,20,0,0,1,0,0,5,2,26,1,1,0,0,5,97,0,0,
-1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
-18,98,0,0,0,0,1,0,0,5,2,27,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,
-114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,5,2,21,1,1,0,0,5,
-97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,
-116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,5,2,22,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,3,2,0,
-0,9,1,98,73,110,118,0,0,1,1,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,0,18,
-98,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,
-118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,
-120,0,0,0,0,1,0,0,6,2,26,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,
-95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,6,2,27,1,1,0,0,6,97,0,0,1,1,0,0,6,98,
+16,10,50,0,57,59,122,0,18,109,50,50,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,9,102,0,0,0,1,3,2,90,95,
+0,0,10,221,1,118,0,2,58,118,101,99,50,0,18,102,0,0,17,48,0,48,0,0,0,0,0,0,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,18,118,0,59,120,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+49,0,57,18,118,0,59,121,120,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,118,0,
+59,121,121,120,0,20,0,0,1,90,95,0,0,14,221,1,1,1,0,0,5,105,0,0,0,1,8,58,109,97,116,51,0,58,102,108,
+111,97,116,0,18,105,0,0,0,0,0,0,0,1,90,95,0,0,14,221,1,1,1,0,0,1,98,0,0,0,1,8,58,109,97,116,51,0,
+58,102,108,111,97,116,0,18,98,0,0,0,0,0,0,0,1,90,95,0,0,14,221,1,1,1,0,0,11,99,48,0,0,1,1,0,0,11,
+99,49,0,0,1,1,0,0,11,99,50,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,
+9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,99,50,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,9,109,48,48,0,0,1,1,0,0,9,109,49,48,
+0,0,1,1,0,0,9,109,50,48,0,0,1,1,0,0,9,109,51,48,0,0,1,1,0,0,9,109,48,49,0,0,1,1,0,0,9,109,49,49,0,
+0,1,1,0,0,9,109,50,49,0,0,1,1,0,0,9,109,51,49,0,0,1,1,0,0,9,109,48,50,0,0,1,1,0,0,9,109,49,50,0,0,
+1,1,0,0,9,109,50,50,0,0,1,1,0,0,9,109,51,50,0,0,1,1,0,0,9,109,48,51,0,0,1,1,0,0,9,109,49,51,0,0,1,
+1,0,0,9,109,50,51,0,0,1,1,0,0,9,109,51,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+59,120,0,18,109,48,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,121,0,18,109,49,48,
+0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,59,122,0,18,109,50,48,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,59,119,0,18,109,51,48,0,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,59,120,0,18,109,48,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,121,
+0,18,109,49,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,122,0,18,109,50,49,0,20,
+0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,59,119,0,18,109,51,49,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,50,0,57,59,120,0,18,109,48,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,50,0,57,59,121,0,18,109,49,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,122,0,
+18,109,50,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,59,119,0,18,109,51,50,0,20,0,
+9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,120,0,18,109,48,51,0,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,51,0,57,59,121,0,18,109,49,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,51,0,57,59,122,0,18,109,50,51,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,59,119,0,
+18,109,51,51,0,20,0,0,1,90,95,0,0,15,221,1,1,1,0,0,9,102,0,0,0,1,3,2,90,95,0,0,10,221,1,118,0,2,58,
+118,101,99,50,0,18,102,0,0,17,48,0,48,0,0,0,0,0,0,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+18,118,0,59,120,121,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,118,0,59,
+121,120,121,121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,118,0,59,121,121,120,
+121,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,118,0,59,121,121,121,120,0,20,0,0,1,
+90,95,0,0,15,221,1,1,1,0,0,5,105,0,0,0,1,8,58,109,97,116,52,0,58,102,108,111,97,116,0,18,105,0,0,0,
+0,0,0,0,1,90,95,0,0,15,221,1,1,1,0,0,1,98,0,0,0,1,8,58,109,97,116,52,0,58,102,108,111,97,116,0,18,
+98,0,0,0,0,0,0,0,1,90,95,0,0,15,221,1,1,1,0,0,12,99,48,0,0,1,1,0,0,12,99,49,0,0,1,1,0,0,12,99,50,0,
+0,1,1,0,0,12,99,51,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,99,48,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,18,99,49,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+50,0,57,18,99,50,0,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,99,51,0,20,0,0,1,90,95,
+0,0,5,221,2,26,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,5,221,2,27,1,1,0,0,5,97,0,0,1,1,0,0,5,98,
 0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,
-0,0,18,98,0,0,0,0,1,0,0,6,2,21,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,4,118,101,99,52,95,109,117,
-108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,6,2,22,
-1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,3,2,0,0,10,1,98,73,110,118,0,0,1,1,120,0,0,0,4,102,108,111,
-97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,
-114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,98,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,
-105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,
-118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,7,2,26,1,1,0,0,7,97,0,0,1,
-1,0,0,7,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,
-98,0,0,0,0,1,0,0,7,2,27,1,1,0,0,7,97,0,0,1,1,0,0,7,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,
-114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,7,2,21,1,1,0,0,7,
-97,0,0,1,1,0,0,7,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,
-116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,7,2,22,1,1,0,0,7,97,0,0,1,1,0,0,7,98,0,0,0,1,3,2,0,
-0,11,1,98,73,110,118,0,0,1,1,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,
-120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,
-98,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,122,0,0,18,98,0,59,
-122,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,
-118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,
-120,0,0,0,0,1,0,0,8,2,26,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,
-95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,8,2,27,1,1,0,0,8,97,0,0,1,1,0,0,8,98,
+0,0,18,98,0,0,0,0,1,90,95,0,0,5,221,2,21,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,
+109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,
+95,0,0,5,221,2,22,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,3,2,90,95,0,0,9,221,1,98,73,110,118,0,0,1,
+1,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,0,18,98,0,0,0,4,118,101,99,52,
+95,109,117,108,116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,99,52,
+95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,0,0,6,
+221,2,26,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,
+116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,6,221,2,27,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,
+0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
+18,98,0,0,0,0,1,90,95,0,0,6,221,2,21,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,4,118,101,99,52,95,109,
+117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,
+0,6,221,2,22,1,1,0,0,6,97,0,0,1,1,0,0,6,98,0,0,0,1,3,2,90,95,0,0,10,221,1,98,73,110,118,0,0,1,1,
+120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,59,120,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,98,0,59,121,0,0,0,4,118,101,
+99,52,95,109,117,108,116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,
+99,52,95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,
+0,0,7,221,2,26,1,1,0,0,7,97,0,0,1,1,0,0,7,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,
+101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,7,221,2,27,1,1,0,0,7,97,0,0,1,1,0,0,7,98,
 0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,
-0,0,18,98,0,0,0,0,1,0,0,8,2,21,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,4,118,101,99,52,95,109,117,
-108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,8,2,22,
-1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,3,2,0,0,12,1,98,73,110,118,0,0,1,1,120,0,0,0,4,102,108,111,
-97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,59,120,0,0,0,4,102,108,111,97,116,95,
-114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,
-0,18,98,73,110,118,0,59,122,0,0,18,98,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,
-110,118,0,59,119,0,0,18,98,0,59,119,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
-120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,95,
-95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,0,0,9,2,26,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,
-118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,9,2,
-27,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,
-95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,9,2,21,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,
-0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,
-97,0,0,18,98,0,0,0,0,1,0,0,9,2,22,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,98,73,110,118,
-0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,0,0,4,118,101,99,
-52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,73,110,
-118,0,0,0,0,1,0,0,10,2,26,1,1,0,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,97,100,100,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,10,2,27,1,1,0,0,
-10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,10,2,21,1,1,0,0,10,118,0,0,1,1,
-0,0,10,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,
-108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,10,2,22,1,1,0,0,10,118,0,0,1,1,0,0,10,117,0,0,
-0,1,3,2,0,0,10,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,
-120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,
-99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,
-0,0,18,119,0,0,0,0,1,0,0,11,2,26,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,97,
-100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,11,2,
-27,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,
-95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,11,2,21,1,1,0,0,
-11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,11,2,22,1,1,0,0,11,118,
-0,0,1,1,0,0,11,117,0,0,0,1,3,2,0,0,11,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,
-120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,
-121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,118,101,
-99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,
-118,0,0,18,119,0,0,0,0,1,0,0,12,2,26,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,
-97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,12,2,27,1,1,0,0,12,
-118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,
-116,86,97,108,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,12,2,21,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,0,
-1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,
-0,18,117,0,0,0,0,1,0,0,12,2,22,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,3,2,0,0,12,1,119,0,0,0,4,
-102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,
-95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,
-119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,119,0,0,18,
-117,0,59,119,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,
-108,0,0,18,118,0,0,18,119,0,0,0,0,1,0,0,10,2,26,1,1,0,0,9,97,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,
-99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,18,117,0,59,120,
-121,0,0,0,0,1,0,0,10,2,26,1,1,0,0,10,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,
-18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,10,2,
-27,1,1,0,0,9,97,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,
-95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,18,117,0,59,120,121,0,0,0,0,1,0,0,10,2,27,1,1,
-0,0,10,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,10,2,21,1,1,0,0,9,
-97,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,18,117,0,59,120,121,0,0,0,0,1,0,0,10,2,21,1,1,0,0,10,
-118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,
-116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,98,0,0,0,0,1,0,0,10,2,22,1,1,0,0,9,97,0,
-0,1,1,0,0,10,117,0,0,0,1,3,2,0,0,10,1,105,110,118,85,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,
+0,0,18,98,0,0,0,0,1,90,95,0,0,7,221,2,21,1,1,0,0,7,97,0,0,1,1,0,0,7,98,0,0,0,1,4,118,101,99,52,95,
+109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,
+95,0,0,7,221,2,22,1,1,0,0,7,97,0,0,1,1,0,0,7,98,0,0,0,1,3,2,90,95,0,0,11,221,1,98,73,110,118,0,0,1,
+1,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,59,120,0,0,
+0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,
+108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,122,0,0,18,98,0,59,122,0,0,0,4,118,101,99,52,
+95,109,117,108,116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,99,52,
+95,116,111,95,105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,0,0,8,
+221,2,26,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,
+116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,8,221,2,27,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,
+0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,
+18,98,0,0,0,0,1,90,95,0,0,8,221,2,21,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,4,118,101,99,52,95,109,
+117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,
+0,8,221,2,22,1,1,0,0,8,97,0,0,1,1,0,0,8,98,0,0,0,1,3,2,90,95,0,0,12,221,1,98,73,110,118,0,0,1,1,
+120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,59,120,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,121,0,0,18,98,0,59,121,0,0,0,4,102,108,
+111,97,116,95,114,99,112,0,18,98,73,110,118,0,59,122,0,0,18,98,0,59,122,0,0,0,4,102,108,111,97,116,
+95,114,99,112,0,18,98,73,110,118,0,59,119,0,0,18,98,0,59,119,0,0,0,4,118,101,99,52,95,109,117,108,
+116,105,112,108,121,0,18,120,0,0,18,97,0,0,18,98,73,110,118,0,0,0,4,118,101,99,52,95,116,111,95,
+105,118,101,99,52,0,18,95,95,114,101,116,86,97,108,0,0,18,120,0,0,0,0,1,90,95,0,0,9,221,2,26,1,1,0,
+0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,
+18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,9,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,
+52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,
+90,95,0,0,9,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,
+112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,9,221,2,22,1,
+1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,98,73,110,118,0,0,0,4,102,108,111,97,116,
+95,114,99,112,0,18,98,73,110,118,0,59,120,0,0,18,98,0,0,0,4,118,101,99,52,95,109,117,108,116,105,
+112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,73,110,118,0,0,0,0,1,90,95,0,0,10,
+221,2,26,1,1,0,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,
+101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,10,221,2,27,1,1,0,0,10,
+118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,
+116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,10,118,0,
+0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,
+86,97,108,0,59,120,121,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,10,221,2,22,1,1,0,0,10,118,0,0,1,
+1,0,0,10,117,0,0,0,1,3,2,90,95,0,0,10,221,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,
+0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,
+0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,
+0,59,120,121,0,0,18,118,0,0,18,119,0,0,0,0,1,90,95,0,0,11,221,2,26,1,1,0,0,11,118,0,0,1,1,0,0,11,
+117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,
+118,0,0,18,117,0,0,0,0,1,90,95,0,0,11,221,2,27,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,
+99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,
+0,0,18,117,0,0,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,
+52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,
+0,0,18,117,0,0,0,0,1,90,95,0,0,11,221,2,22,1,1,0,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,3,2,90,95,0,0,
+11,221,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,118,101,99,52,95,109,117,108,116,
+105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0,18,119,0,0,0,0,1,
+90,95,0,0,12,221,2,26,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,
+95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,27,1,1,0,0,12,118,0,
+0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,
+97,108,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,
+0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,
+0,0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,22,1,1,0,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,3,2,90,95,0,0,
+12,221,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,
+18,119,0,59,119,0,0,18,117,0,59,119,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
+95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,119,0,0,0,0,1,90,95,0,0,10,221,2,26,1,1,0,0,9,97,0,0,
+1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,
+0,0,18,97,0,0,18,117,0,59,120,121,0,0,0,0,1,90,95,0,0,10,221,2,26,1,1,0,0,10,118,0,0,1,1,0,0,9,98,
+0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,
+59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,10,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,10,117,0,0,0,1,4,118,
+101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,
+0,0,18,117,0,59,120,121,0,0,0,0,1,90,95,0,0,10,221,2,27,1,1,0,0,10,118,0,0,1,1,0,0,9,98,0,0,0,1,4,
+118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,
+118,0,59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,10,117,0,0,0,1,
+4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,
+0,18,97,0,0,18,117,0,59,120,121,0,0,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,10,118,0,0,1,1,0,0,9,98,0,
+0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,
+121,0,0,18,118,0,59,120,121,0,0,18,98,0,0,0,0,1,90,95,0,0,10,221,2,22,1,1,0,0,9,97,0,0,1,1,0,0,10,
+117,0,0,0,1,3,2,90,95,0,0,10,221,1,105,110,118,85,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,
 105,110,118,85,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,
 118,85,0,59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
-95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,18,105,110,118,85,0,59,120,121,0,0,0,0,1,0,
-0,10,2,22,1,1,0,0,10,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,105,110,118,66,0,0,0,4,102,108,111,
-97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,118,101,99,52,95,109,117,108,116,105,
-112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,105,110,
-118,66,0,0,0,0,1,0,0,11,2,26,1,1,0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,97,100,100,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,117,0,59,120,121,122,0,0,0,0,1,
-0,0,11,2,26,1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,11,2,27,1,1,
-0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,117,0,59,120,121,122,0,0,0,0,1,0,0,11,2,27,1,1,
-0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,11,2,21,1,1,
-0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,
-114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,117,0,59,120,121,122,0,0,0,0,1,0,0,11,2,21,
-1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,
-95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,98,0,0,0,0,1,0,0,11,2,
-22,1,1,0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,3,2,0,0,11,1,105,110,118,85,0,0,0,4,102,108,111,97,116,
+95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,97,0,0,18,105,110,118,85,0,59,120,121,0,0,0,0,1,90,
+95,0,0,10,221,2,22,1,1,0,0,10,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,105,110,118,66,0,
+0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,118,101,99,52,95,109,
+117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,
+0,0,18,105,110,118,66,0,0,0,0,1,90,95,0,0,11,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,4,
+118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,117,
+0,59,120,121,122,0,0,0,0,1,90,95,0,0,11,221,2,26,1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,
+99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,
+0,0,18,98,0,0,0,0,1,90,95,0,0,11,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,
+95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,
+117,0,59,120,121,122,0,0,0,0,1,90,95,0,0,11,221,2,27,1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,
+101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,
+118,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,11,117,0,0,
+0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,
+121,122,0,0,18,97,0,0,18,117,0,59,120,121,122,0,0,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,11,118,0,0,1,
+1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,
+108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,98,0,0,0,0,1,90,95,0,0,11,221,2,22,1,1,0,0,
+9,97,0,0,1,1,0,0,11,117,0,0,0,1,3,2,90,95,0,0,11,221,1,105,110,118,85,0,0,0,4,102,108,111,97,116,
 95,114,99,112,0,18,105,110,118,85,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,
 99,112,0,18,105,110,118,85,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,
 18,105,110,118,85,0,59,122,0,0,18,117,0,59,122,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,
 108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,18,105,110,118,85,0,59,120,
-121,122,0,0,0,0,1,0,0,11,2,22,1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,105,110,118,66,0,
-0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,118,101,99,52,95,109,
-117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,
-121,122,0,0,18,105,110,118,66,0,0,0,0,1,0,0,12,2,26,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,1,4,118,
-101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,117,0,0,0,0,1,0,0,12,2,
-26,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,
-97,108,0,0,18,118,0,0,18,98,0,0,0,0,1,0,0,12,2,27,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,1,4,118,
-101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,117,0,
-0,0,0,1,0,0,12,2,27,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,
-97,99,116,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,98,0,0,0,0,1,0,0,12,2,21,1,1,0,0,9,97,
-0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,
-116,86,97,108,0,0,18,97,0,0,18,117,0,0,0,0,1,0,0,12,2,21,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,4,
-118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,
-18,98,0,0,0,0,1,0,0,12,2,22,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,1,3,2,0,0,12,1,105,110,118,85,0,
-0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,120,0,0,18,117,0,59,120,0,0,0,4,
-102,108,111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,122,0,0,18,117,0,59,122,0,0,0,4,102,108,111,97,
-116,95,114,99,112,0,18,105,110,118,85,0,59,119,0,0,18,117,0,59,119,0,0,0,4,118,101,99,52,95,109,
+121,122,0,0,0,0,1,90,95,0,0,11,221,2,22,1,1,0,0,11,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,
+221,1,105,110,118,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,
+4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,
+122,0,0,18,118,0,59,120,121,122,0,0,18,105,110,118,66,0,0,0,0,1,90,95,0,0,12,221,2,26,1,1,0,0,9,97,
+0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,
+97,0,0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,26,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,
+52,95,97,100,100,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,98,0,0,0,0,1,90,95,0,0,12,221,
+2,27,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,
+95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,27,1,1,0,0,12,118,0,
+0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,95,95,114,101,116,86,
+97,108,0,0,18,118,0,0,18,98,0,0,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,
+1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,
+0,18,117,0,0,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,
+95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,98,0,0,0,0,1,
+90,95,0,0,12,221,2,22,1,1,0,0,9,97,0,0,1,1,0,0,12,117,0,0,0,1,3,2,90,95,0,0,12,221,1,105,110,118,
+85,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,120,0,0,18,117,0,59,120,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,
+108,111,97,116,95,114,99,112,0,18,105,110,118,85,0,59,122,0,0,18,117,0,59,122,0,0,0,4,102,108,111,
+97,116,95,114,99,112,0,18,105,110,118,85,0,59,119,0,0,18,117,0,59,119,0,0,0,4,118,101,99,52,95,109,
 117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,105,110,118,85,0,0,0,
-0,1,0,0,12,2,22,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,105,110,118,66,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,118,101,99,52,95,109,117,108,116,
-105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,105,110,118,66,0,0,0,0,1,0,0,6,
-2,26,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,
-50,0,18,97,0,0,0,18,117,0,46,20,0,0,1,0,0,6,2,26,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,50,0,18,98,0,0,0,46,20,0,0,1,0,0,6,2,27,1,1,0,0,
-5,97,0,0,1,1,0,0,6,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,
-0,18,117,0,47,20,0,0,1,0,0,6,2,27,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,18,118,0,58,105,118,101,99,50,0,18,98,0,0,0,47,20,0,0,1,0,0,6,2,21,1,1,0,0,5,97,0,0,1,1,0,
-0,6,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,48,
-20,0,0,1,0,0,6,2,21,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
-118,0,58,105,118,101,99,50,0,18,98,0,0,0,48,20,0,0,1,0,0,6,2,22,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,0,
-0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,49,20,0,0,1,0,0,
-6,2,22,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,
-118,101,99,50,0,18,98,0,0,0,49,20,0,0,1,0,0,7,2,26,1,1,0,0,5,97,0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,0,0,18,117,0,46,20,0,0,1,0,0,7,2,26,1,1,
-0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,
-51,0,18,98,0,0,0,46,20,0,0,1,0,0,7,2,27,1,1,0,0,5,97,0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,0,0,18,117,0,47,20,0,0,1,0,0,7,2,27,1,1,0,0,7,118,0,
-0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,
-0,0,47,20,0,0,1,0,0,7,2,21,1,1,0,0,5,97,0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,58,105,118,101,99,51,0,18,97,0,0,0,18,117,0,48,20,0,0,1,0,0,7,2,21,1,1,0,0,7,118,0,0,1,1,0,0,5,
-98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,48,20,0,
-0,1,0,0,7,2,22,1,1,0,0,5,97,0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,
-118,101,99,51,0,18,97,0,0,0,18,117,0,49,20,0,0,1,0,0,7,2,22,1,1,0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,49,20,0,0,1,0,0,8,2,
+0,1,90,95,0,0,12,221,2,22,1,1,0,0,12,118,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,105,110,
+118,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,118,101,99,
+52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,18,105,110,
+118,66,0,0,0,0,1,90,95,0,0,6,221,2,26,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,46,20,0,0,1,90,95,0,0,6,221,2,26,1,1,0,
+0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,50,
+0,18,98,0,0,0,46,20,0,0,1,90,95,0,0,6,221,2,27,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,47,20,0,0,1,90,95,0,0,6,221,2,
+27,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,
+101,99,50,0,18,98,0,0,0,47,20,0,0,1,90,95,0,0,6,221,2,21,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,48,20,0,0,1,90,95,0,0,
+6,221,2,21,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,
+105,118,101,99,50,0,18,98,0,0,0,48,20,0,0,1,90,95,0,0,6,221,2,22,1,1,0,0,5,97,0,0,1,1,0,0,6,117,0,
+0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,50,0,18,97,0,0,0,18,117,0,49,20,0,0,1,
+90,95,0,0,6,221,2,22,1,1,0,0,6,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
+118,0,58,105,118,101,99,50,0,18,98,0,0,0,49,20,0,0,1,90,95,0,0,7,221,2,26,1,1,0,0,5,97,0,0,1,1,0,0,
+7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,0,0,18,117,0,46,20,
+0,0,1,90,95,0,0,7,221,2,26,1,1,0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
+0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,46,20,0,0,1,90,95,0,0,7,221,2,27,1,1,0,0,5,97,0,0,1,
+1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,0,0,18,117,0,
+47,20,0,0,1,90,95,0,0,7,221,2,27,1,1,0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,47,20,0,0,1,90,95,0,0,7,221,2,21,1,1,0,0,5,97,
+0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,0,0,18,
+117,0,48,20,0,0,1,90,95,0,0,7,221,2,21,1,1,0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,48,20,0,0,1,90,95,0,0,7,221,2,22,1,1,0,
+0,5,97,0,0,1,1,0,0,7,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,51,0,18,97,0,
+0,0,18,117,0,49,20,0,0,1,90,95,0,0,7,221,2,22,1,1,0,0,7,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,51,0,18,98,0,0,0,49,20,0,0,1,90,95,0,0,8,221,2,
 26,1,1,0,0,5,97,0,0,1,1,0,0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,52,
-0,18,97,0,0,0,18,117,0,46,20,0,0,1,0,0,8,2,26,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,0,0,46,20,0,0,1,0,0,8,2,27,1,1,0,0,
-5,97,0,0,1,1,0,0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,52,0,18,97,0,0,
-0,18,117,0,47,20,0,0,1,0,0,8,2,27,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,
-97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,0,0,47,20,0,0,1,0,0,8,2,21,1,1,0,0,5,97,0,0,1,1,0,
-0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,52,0,18,97,0,0,0,18,117,0,48,
-20,0,0,1,0,0,8,2,21,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
-118,0,58,105,118,101,99,52,0,18,98,0,0,0,48,20,0,0,1,0,0,8,2,22,1,1,0,0,5,97,0,0,1,1,0,0,8,117,0,0,
-0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,101,99,52,0,18,97,0,0,0,18,117,0,49,20,0,0,1,0,0,
-8,2,22,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,
-118,101,99,52,0,18,98,0,0,0,49,20,0,0,1,0,0,5,2,27,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,110,101,
-103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,0,0,0,1,0,0,6,2,27,1,1,0,0,6,
+0,18,97,0,0,0,18,117,0,46,20,0,0,1,90,95,0,0,8,221,2,26,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,0,0,46,20,0,0,1,90,95,0,0,
+8,221,2,27,1,1,0,0,5,97,0,0,1,1,0,0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,105,118,
+101,99,52,0,18,97,0,0,0,18,117,0,47,20,0,0,1,90,95,0,0,8,221,2,27,1,1,0,0,8,118,0,0,1,1,0,0,5,98,0,
+0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,0,0,47,20,0,0,1,
+90,95,0,0,8,221,2,21,1,1,0,0,5,97,0,0,1,1,0,0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,58,
+105,118,101,99,52,0,18,97,0,0,0,18,117,0,48,20,0,0,1,90,95,0,0,8,221,2,21,1,1,0,0,8,118,0,0,1,1,0,
+0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,0,0,48,
+20,0,0,1,90,95,0,0,8,221,2,22,1,1,0,0,5,97,0,0,1,1,0,0,8,117,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,58,105,118,101,99,52,0,18,97,0,0,0,18,117,0,49,20,0,0,1,90,95,0,0,8,221,2,22,1,1,0,0,8,118,0,
+0,1,1,0,0,5,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,58,105,118,101,99,52,0,18,98,0,
+0,0,49,20,0,0,1,90,95,0,0,5,221,2,27,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,
+101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,97,0,0,0,0,1,90,95,0,0,6,221,2,27,1,1,0,0,6,
 118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,0,18,118,
-0,0,0,0,1,0,0,7,2,27,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,
-114,101,116,86,97,108,0,0,18,118,0,0,0,0,1,0,0,8,2,27,1,1,0,0,8,118,0,0,0,1,4,118,101,99,52,95,110,
-101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,0,0,1,0,0,9,2,27,1,1,0,0,9,97,0,
-0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
-97,0,0,0,0,1,0,0,10,2,27,1,1,0,0,10,118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,
-95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,120,121,0,0,0,0,1,0,0,11,2,27,1,1,0,0,11,118,
-0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,
-122,0,0,18,118,0,59,120,121,122,0,0,0,0,1,0,0,12,2,27,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,
-110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,0,0,1,0,0,13,2,27,1,1,0,0,
-13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,0,1,0,0,14,2,27,1,1,0,0,
-14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,54,20,0,0,1,0,0,15,2,27,1,1,0,0,15,109,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
-50,0,57,18,109,0,16,10,50,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,
-16,10,51,0,57,54,20,0,0,1,0,0,9,0,100,111,116,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,18,97,0,18,98,0,48,20,0,0,1,0,0,9,0,100,111,116,0,1,1,0,0,10,97,0,0,1,1,0,
-0,10,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,59,120,0,18,98,0,59,120,0,48,18,97,0,59,
-121,0,18,98,0,59,121,0,48,46,20,0,0,1,0,0,9,0,100,111,116,0,1,1,0,0,11,97,0,0,1,1,0,0,11,98,0,0,0,
-1,4,118,101,99,51,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,
-0,9,0,100,111,116,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,118,101,99,52,95,100,111,116,0,18,95,
-95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,0,2,1,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,
-0,1,4,118,101,99,52,95,97,100,100,0,18,97,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,0,2,2,1,0,2,0,5,97,0,0,
-1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,97,0,0,18,97,0,0,18,98,0,
-0,0,0,1,0,0,0,2,3,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,
-108,121,0,18,97,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,0,2,4,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,3,2,
-0,0,9,1,105,110,118,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,
-0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,97,0,0,18,97,0,0,18,105,110,118,66,0,0,0,
-4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,97,0,0,18,97,0,0,0,0,1,0,0,0,2,1,1,0,2,0,6,
-118,0,0,1,1,0,0,6,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,
-1,0,0,0,2,2,1,0,2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,
-116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,3,1,0,2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,4,
-118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,
-4,1,0,2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,3,2,0,0,6,1,105,110,118,0,0,1,1,122,0,0,0,4,102,108,111,
-97,116,95,114,99,112,0,18,105,110,118,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,
-114,99,112,0,18,105,110,118,0,59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,
-105,112,108,121,0,18,122,0,0,18,118,0,0,18,105,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,
-101,99,52,0,18,118,0,0,18,122,0,0,0,0,1,0,0,0,2,1,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,4,118,
-101,99,52,95,97,100,100,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,2,1,0,2,0,7,118,0,0,1,1,0,
-0,7,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,
-0,0,1,0,0,0,2,3,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,
-108,121,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,4,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,
-3,2,0,0,7,1,105,110,118,0,0,1,1,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,
-120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,121,0,0,18,
-117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,122,0,0,18,118,0,0,18,
-105,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,118,0,0,18,122,0,0,0,0,1,0,
-0,0,2,1,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,118,
-0,0,18,117,0,0,0,0,1,0,0,0,2,2,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,4,118,101,99,52,95,115,117,
-98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,3,1,0,2,0,8,118,0,0,1,1,0,0,
-8,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,117,0,
-0,0,0,1,0,0,0,2,4,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,3,2,0,0,8,1,105,110,118,0,0,1,1,122,0,0,
-0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,105,110,118,0,59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,99,52,95,
-109,117,108,116,105,112,108,121,0,18,122,0,0,18,118,0,0,18,105,110,118,0,0,0,4,118,101,99,52,95,
-116,111,95,105,118,101,99,52,0,18,118,0,0,18,122,0,0,0,0,1,0,0,0,2,1,1,0,2,0,9,97,0,0,1,1,0,0,9,98,
-0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,97,0,59,120,0,0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,
-0,1,0,0,0,2,2,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,
-116,0,18,97,0,59,120,0,0,18,97,0,0,18,98,0,0,0,0,1,0,0,0,2,3,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,
-4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,97,0,59,120,0,0,18,97,0,0,18,98,0,0,0,0,1,
-0,0,0,2,4,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,9,1,119,0,0,0,4,102,108,111,97,116,95,114,
-99,112,0,18,119,0,59,120,0,0,18,98,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
-97,0,59,120,0,0,18,97,0,0,18,119,0,0,0,0,1,0,0,0,2,1,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,
-118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,0,59,120,121,
-0,0,0,0,1,0,0,0,2,2,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,
-114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,0,59,120,121,0,0,0,0,1,0,0,
-0,2,3,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
-0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,0,59,120,121,0,0,0,0,1,0,0,0,2,4,1,0,2,0,
-10,118,0,0,1,1,0,0,10,117,0,0,0,1,3,2,0,0,10,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,
-119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,
-117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,0,0,18,
-118,0,59,120,121,0,0,18,119,0,59,120,121,0,0,0,0,1,0,0,0,2,1,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,
-0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,
-2,2,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,
-118,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,3,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,
-0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,0,
-18,117,0,0,0,0,1,0,0,0,2,4,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,3,2,0,0,11,1,119,0,0,0,4,102,
-108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,
-114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,
-0,59,122,0,0,18,117,0,59,122,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,
-59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,119,0,59,120,121,122,0,0,0,0,1,0,0,0,2,1,1,0,2,0,
-12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,118,0,0,18,117,0,0,
-0,0,1,0,0,0,2,2,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,
-99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,0,2,3,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,
-1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,0,0,
-0,2,4,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,3,2,0,0,12,1,119,0,0,0,4,102,108,111,97,116,95,114,
-99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,
-59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,122,0,0,18,117,0,
-59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,119,0,0,18,117,0,59,119,0,0,0,4,118,
-101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,119,0,0,0,0,1,0,0,0,2,1,1,
-0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,0,0,18,
-118,0,59,120,121,0,0,18,97,0,0,0,0,1,0,0,0,2,2,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,
-52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,97,0,0,0,0,
-1,0,0,0,2,3,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,
-121,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,97,0,0,0,9,18,118,0,59,120,0,18,97,0,23,0,
-9,18,118,0,59,121,0,18,97,0,23,0,0,1,0,0,0,2,4,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,9,18,118,0,
-59,120,0,18,97,0,24,0,9,18,118,0,59,121,0,18,97,0,24,0,0,1,0,0,0,2,1,1,0,2,0,7,118,0,0,1,1,0,0,5,
-97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,
-18,97,0,0,0,0,1,0,0,0,2,2,1,0,2,0,7,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,115,117,98,116,
-114,97,99,116,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,97,0,0,0,0,1,0,0,0,2,3,
-1,0,2,0,7,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,
-0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,97,0,0,0,0,1,0,0,0,2,4,1,0,2,0,7,118,0,0,1,1,0,
-0,5,97,0,0,0,1,9,18,118,0,59,120,0,18,97,0,24,0,9,18,118,0,59,121,0,18,97,0,24,0,9,18,118,0,59,122,
-0,18,97,0,24,0,0,1,0,0,0,2,1,1,0,2,0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,97,100,100,
-0,18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,2,1,0,2,0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,
-99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,3,1,0,2,0,8,
-118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,
-118,0,0,18,97,0,0,0,0,1,0,0,0,2,4,1,0,2,0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,9,18,118,0,59,120,0,18,97,
-0,24,0,9,18,118,0,59,121,0,18,97,0,24,0,9,18,118,0,59,122,0,18,97,0,24,0,9,18,118,0,59,119,0,18,97,
-0,24,0,0,1,0,0,0,2,1,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,
-118,0,59,120,121,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,2,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,
-4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,0,18,97,0,0,0,0,
-1,0,0,0,2,3,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,
-121,0,18,118,0,59,120,121,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,4,1,0,2,0,10,118,0,0,1,1,0,0,9,97,
-0,0,0,1,3,2,0,0,9,1,105,110,118,65,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,65,0,
-0,18,97,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,0,0,18,118,
-0,59,120,121,0,0,18,105,110,118,65,0,0,0,0,1,0,0,0,2,1,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,4,
-118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,2,1,0,
-2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,
-120,121,122,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,3,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,
-101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,97,0,0,0,
-0,1,0,0,0,2,4,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,3,2,0,0,9,1,105,110,118,65,0,0,0,4,102,108,
-111,97,116,95,114,99,112,0,18,105,110,118,65,0,0,18,97,0,0,0,4,118,101,99,52,95,109,117,108,116,
-105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,105,110,118,65,0,0,0,
-0,1,0,0,0,2,1,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,
-18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,2,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,
-115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,0,0,0,2,3,1,0,2,0,12,118,0,0,
-1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,
-97,0,0,0,0,1,0,0,0,2,4,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,3,2,0,0,9,1,105,110,118,65,0,0,0,4,
-102,108,111,97,116,95,114,99,112,0,18,105,110,118,65,0,0,18,97,0,0,0,4,118,101,99,52,95,109,117,
-108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,105,110,118,65,0,0,0,0,1,0,0,13,2,26,1,1,0,0,13,
-109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,
-57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,
-49,0,57,18,110,0,16,10,49,0,57,46,20,0,0,1,0,0,13,2,27,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,
-9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,
-20,0,0,1,0,0,13,2,21,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
-16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,0,48,18,109,0,16,10,49,0,57,18,
-110,0,16,8,48,0,57,59,121,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
-0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,
-57,59,121,121,0,48,46,20,0,0,1,0,0,13,2,22,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,0,1,
-0,0,14,2,26,1,1,0,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,
-0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,20,0,0,1,0,0,14,2,27,1,1,0,0,14,109,0,
-0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,
-110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,
-18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,
-0,57,18,110,0,16,10,50,0,57,47,20,0,0,1,0,0,14,2,21,1,1,0,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,
-120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,121,121,121,0,48,46,18,109,0,16,10,50,0,
-57,18,110,0,16,8,48,0,57,59,122,122,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
-57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,
-0,16,10,49,0,57,59,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,49,0,57,59,122,122,
-122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,8,48,0,57,18,110,0,
-16,10,50,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,50,0,57,59,121,121,121,0,
-48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,122,122,122,0,48,46,20,0,0,1,0,0,14,2,22,1,
-1,0,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,
-16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
-0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
-109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,0,1,0,0,15,2,26,1,1,0,0,15,109,0,0,1,1,0,0,15,
-110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,
-0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,
-10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,
-0,16,10,50,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,
-110,0,16,10,51,0,57,46,20,0,0,1,0,0,15,2,27,1,1,0,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,
-95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,
-0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,
-47,20,0,0,1,0,0,15,2,21,1,1,0,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,120,120,0,48,18,109,0,16,10,
-49,0,57,18,110,0,16,8,48,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,8,48,0,
-57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,8,48,0,57,59,119,119,119,119,0,48,
-46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,
-57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,59,121,121,121,121,0,48,
-46,18,109,0,16,10,50,0,57,18,110,0,16,10,49,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,
-18,110,0,16,10,49,0,57,59,119,119,119,119,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,
-0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,50,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,
-18,110,0,16,10,50,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,
-122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,10,50,0,57,59,119,119,119,119,0,48,46,
-20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,51,0,57,
-59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,51,0,57,59,121,121,121,121,0,48,46,
-18,109,0,16,10,50,0,57,18,110,0,16,10,51,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,
-110,0,16,10,51,0,57,59,119,119,119,119,0,48,46,20,0,0,1,0,0,15,2,22,1,1,0,0,15,109,0,0,1,1,0,0,15,
-110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,
-0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,
-10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,
-0,16,10,50,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,
-110,0,16,10,51,0,57,49,20,0,0,1,0,0,13,2,26,1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,0,1,0,0,13,2,26,1,1,0,0,13,109,0,0,1,
-1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,
-46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,0,
-1,0,0,13,2,27,1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,
-18,110,0,16,10,49,0,57,47,20,0,0,1,0,0,13,2,27,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47,20,0,0,1,0,0,13,2,21,1,1,0,0,9,97,0,0,
-1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,
-57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,
-0,1,0,0,13,2,21,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
-0,16,10,49,0,57,18,98,0,48,20,0,0,1,0,0,13,2,22,1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,
-86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,0,1,0,0,13,2,22,1,1,0,0,13,109,0,
+0,0,0,0,1,90,95,0,0,7,221,2,27,1,1,0,0,7,118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,
+18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,0,0,1,90,95,0,0,8,221,2,27,1,1,0,0,8,118,0,0,0,1,4,
+118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,0,18,118,0,0,0,0,1,90,
+95,0,0,9,221,2,27,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,
+101,116,86,97,108,0,59,120,0,0,18,97,0,0,0,0,1,90,95,0,0,10,221,2,27,1,1,0,0,10,118,0,0,0,1,4,118,
+101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,59,
+120,121,0,0,0,0,1,90,95,0,0,11,221,2,27,1,1,0,0,11,118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,
+116,101,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,0,0,1,90,
+95,0,0,12,221,2,27,1,1,0,0,12,118,0,0,0,1,4,118,101,99,52,95,110,101,103,97,116,101,0,18,95,95,114,
+101,116,86,97,108,0,0,18,118,0,0,0,0,1,90,95,0,0,13,221,2,27,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,0,1,90,95,0,0,14,221,2,27,1,1,0,0,14,109,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,95,95,114,101,116,86,
+97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,
+0,57,18,109,0,16,10,50,0,57,54,20,0,0,1,90,95,0,0,15,221,2,27,1,1,0,0,15,109,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,54,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,
+57,18,109,0,16,10,50,0,57,54,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,
+51,0,57,54,20,0,0,1,90,95,0,0,9,221,0,100,111,116,0,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,18,97,0,18,98,0,48,20,0,0,1,90,95,0,0,9,221,0,100,111,116,0,1,1,0,0,10,
+97,0,0,1,1,0,0,10,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,59,120,0,18,98,0,59,120,0,
+48,18,97,0,59,121,0,18,98,0,59,121,0,48,46,20,0,0,1,90,95,0,0,9,221,0,100,111,116,0,1,1,0,0,11,97,
+0,0,1,1,0,0,11,98,0,0,0,1,4,118,101,99,51,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,0,18,
+97,0,0,18,98,0,0,0,0,1,90,95,0,0,9,221,0,100,111,116,0,1,1,0,0,12,97,0,0,1,1,0,0,12,98,0,0,0,1,4,
+118,101,99,52,95,100,111,116,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,
+0,0,0,221,2,1,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,97,0,0,18,
+97,0,0,18,98,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,
+95,115,117,98,116,114,97,99,116,0,18,97,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,
+5,97,0,0,1,1,0,0,5,98,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,97,0,0,18,97,
+0,0,18,98,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,3,2,90,95,0,0,9,221,
+1,105,110,118,66,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,66,0,0,18,98,0,0,0,4,
+118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,97,0,0,18,97,0,0,18,105,110,118,66,0,0,0,4,
+118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,97,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,1,1,0,
+2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,118,0,0,18,117,0,
+0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,4,118,101,99,52,95,115,117,98,
+116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,6,118,0,0,1,
+1,0,0,6,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,
+117,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,6,118,0,0,1,1,0,0,6,117,0,0,0,1,3,2,90,95,0,0,6,221,1,
+105,110,118,0,0,1,1,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,120,0,0,18,
+117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,121,0,0,18,117,0,59,
+121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,122,0,0,18,118,0,0,18,105,110,
+118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,118,0,0,18,122,0,0,0,0,1,90,95,0,0,
+0,221,2,1,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,
+118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,4,118,101,99,
+52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,3,1,
+0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,
+0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,7,118,0,0,1,1,0,0,7,117,0,0,0,1,3,2,90,
+95,0,0,7,221,1,105,110,118,0,0,1,1,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,
+59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,59,121,0,0,
+18,117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,122,0,0,18,118,0,0,
+18,105,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,118,0,0,18,122,0,0,0,0,1,
+90,95,0,0,0,221,2,1,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,
+0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,4,118,
+101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,
+2,3,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,
+18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,8,118,0,0,1,1,0,0,8,117,0,0,0,1,
+3,2,90,95,0,0,8,221,1,105,110,118,0,0,1,1,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,
+110,118,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,0,
+59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,122,0,0,
+18,118,0,0,18,105,110,118,0,0,0,4,118,101,99,52,95,116,111,95,105,118,101,99,52,0,18,118,0,0,18,
+122,0,0,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,97,100,
+100,0,18,97,0,59,120,0,0,18,97,0,59,120,0,0,18,98,0,59,120,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,9,
+97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,97,0,59,120,0,0,
+18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,
+52,95,109,117,108,116,105,112,108,121,0,18,97,0,59,120,0,0,18,97,0,0,18,98,0,0,0,0,1,90,95,0,0,0,
+221,2,4,1,0,2,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,9,221,1,119,0,0,0,4,102,108,111,97,116,
+95,114,99,112,0,18,119,0,59,120,0,0,18,98,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
+0,18,97,0,59,120,0,0,18,97,0,0,18,119,0,0,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,10,118,0,0,1,1,0,0,10,
+117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,
+0,59,120,121,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,
+52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,0,59,
+120,121,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,4,118,101,99,52,95,
+109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,117,0,59,120,
+121,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,10,118,0,0,1,1,0,0,10,117,0,0,0,1,3,2,90,95,0,0,10,221,1,
+119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,
+111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,118,101,99,52,95,109,117,
+108,116,105,112,108,121,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,119,0,59,120,121,0,0,
+0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,97,100,100,
+0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,11,118,0,0,1,
+1,0,0,11,117,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,122,0,0,
+18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,4,118,
+101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,117,0,0,0,
+0,1,90,95,0,0,0,221,2,4,1,0,2,0,11,118,0,0,1,1,0,0,11,117,0,0,0,1,3,2,90,95,0,0,11,221,1,119,0,0,0,
+4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,59,120,0,0,0,4,102,108,111,97,
+116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,
+18,119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
+118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,119,0,59,120,121,122,0,0,0,0,1,90,95,0,0,0,
+221,2,1,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,
+118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,
+99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,3,
+1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,
+118,0,0,18,118,0,0,18,117,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,12,118,0,0,1,1,0,0,12,117,0,0,0,1,
+3,2,90,95,0,0,12,221,1,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,120,0,0,18,117,0,
+59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,119,0,59,121,0,0,18,117,0,59,121,0,0,0,4,102,
+108,111,97,116,95,114,99,112,0,18,119,0,59,122,0,0,18,117,0,59,122,0,0,0,4,102,108,111,97,116,95,
+114,99,112,0,18,119,0,59,119,0,0,18,117,0,59,119,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,
+108,121,0,18,118,0,0,18,118,0,0,18,119,0,0,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,6,118,0,0,1,1,0,0,5,
+97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,97,0,
+0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,115,117,98,
+116,114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,97,0,0,0,0,1,90,95,0,0,0,
+221,2,3,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
+0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,97,0,0,0,9,18,118,0,59,120,0,18,97,0,23,0,9,
+18,118,0,59,121,0,18,97,0,23,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,6,118,0,0,1,1,0,0,5,97,0,0,0,1,9,18,
+118,0,59,120,0,18,97,0,24,0,9,18,118,0,59,121,0,18,97,0,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,7,118,
+0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,120,121,122,0,0,18,118,0,59,
+120,121,122,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,7,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,
+101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,
+0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,7,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,
+109,117,108,116,105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,97,0,
+0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,7,118,0,0,1,1,0,0,5,97,0,0,0,1,9,18,118,0,59,120,0,18,97,0,24,
+0,9,18,118,0,59,121,0,18,97,0,24,0,9,18,118,0,59,122,0,18,97,0,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,
+0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,0,18,118,0,0,18,97,0,0,0,
+0,1,90,95,0,0,0,221,2,2,1,0,2,0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,4,118,101,99,52,95,115,117,98,116,
+114,97,99,116,0,18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,8,118,0,0,1,1,0,
+0,5,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,97,0,
+0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,8,118,0,0,1,1,0,0,5,97,0,0,0,1,9,18,118,0,59,120,0,18,97,0,24,
+0,9,18,118,0,59,121,0,18,97,0,24,0,9,18,118,0,59,122,0,18,97,0,24,0,9,18,118,0,59,119,0,18,97,0,24,
+0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,
+18,118,0,59,120,121,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,10,118,0,0,1,1,0,0,
+9,97,0,0,0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,0,0,18,118,0,0,
+18,97,0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,109,
+117,108,116,105,112,108,121,0,18,118,0,59,120,121,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,
+4,1,0,2,0,10,118,0,0,1,1,0,0,9,97,0,0,0,1,3,2,90,95,0,0,9,221,1,105,110,118,65,0,0,0,4,102,108,111,
+97,116,95,114,99,112,0,18,105,110,118,65,0,0,18,97,0,0,0,4,118,101,99,52,95,109,117,108,116,105,
+112,108,121,0,18,118,0,59,120,121,0,0,18,118,0,59,120,121,0,0,18,105,110,118,65,0,0,0,0,1,90,95,0,
+0,0,221,2,1,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,0,59,
+120,121,122,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,
+0,1,4,118,101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,97,
+0,0,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,109,117,
+108,116,105,112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,
+4,1,0,2,0,11,118,0,0,1,1,0,0,9,97,0,0,0,1,3,2,90,95,0,0,9,221,1,105,110,118,65,0,0,0,4,102,108,111,
+97,116,95,114,99,112,0,18,105,110,118,65,0,0,18,97,0,0,0,4,118,101,99,52,95,109,117,108,116,105,
+112,108,121,0,18,118,0,59,120,121,122,0,0,18,118,0,59,120,121,122,0,0,18,105,110,118,65,0,0,0,0,1,
+90,95,0,0,0,221,2,1,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,97,100,100,0,18,118,
+0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,
+101,99,52,95,115,117,98,116,114,97,99,116,0,18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,
+2,3,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,
+18,118,0,0,18,118,0,0,18,97,0,0,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,12,118,0,0,1,1,0,0,9,97,0,0,0,1,
+3,2,90,95,0,0,9,221,1,105,110,118,65,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,105,110,118,65,
+0,0,18,97,0,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,118,0,0,18,118,0,0,18,105,
+110,118,65,0,0,0,0,1,90,95,0,0,13,221,2,26,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,0,1,
+90,95,0,0,13,221,2,27,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
+16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,47,20,0,0,1,90,95,0,0,13,221,2,21,1,
+1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,
+16,8,48,0,57,18,110,0,16,8,48,0,57,59,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,
+121,121,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,
+0,16,10,49,0,57,59,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,59,121,121,0,48,46,
+20,0,0,1,90,95,0,0,13,221,2,22,1,1,0,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,0,1,90,95,0,0,14,
+221,2,26,1,1,0,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,
+57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,20,0,0,1,90,95,0,0,14,221,2,27,1,1,0,0,14,
+109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,
+57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,
+49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,
+16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,0,1,90,95,0,0,14,221,2,21,1,1,0,0,14,109,0,0,1,1,0,0,
+14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,
+48,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,121,121,121,0,48,46,18,
+109,0,16,10,50,0,57,18,110,0,16,8,48,0,57,59,122,122,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,120,0,48,18,109,0,16,
+10,49,0,57,18,110,0,16,10,49,0,57,59,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,49,
+0,57,59,122,122,122,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,8,48,
+0,57,18,110,0,16,10,50,0,57,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,50,0,57,59,
+121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,122,122,122,0,48,46,20,0,0,1,
+90,95,0,0,14,221,2,22,1,1,0,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
+16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,
+108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,0,1,90,95,0,0,15,221,2,
+26,1,1,0,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,
+109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,
+18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,
+0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
+10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,46,20,0,0,1,90,95,0,0,15,221,2,27,1,1,0,0,
+15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,
+0,57,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,
+49,0,57,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,
+16,10,50,0,57,18,110,0,16,10,50,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,
+109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,47,20,0,0,1,90,95,0,0,15,221,2,21,1,1,0,0,15,109,0,0,1,
+1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,
+0,16,8,48,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,8,48,0,57,59,121,121,121,
+121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,8,48,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,
+51,0,57,18,110,0,16,8,48,0,57,59,119,119,119,119,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,49,0,57,59,120,120,120,120,0,48,18,109,0,16,10,
+49,0,57,18,110,0,16,10,49,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,49,
+0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,10,49,0,57,59,119,119,119,119,0,
+48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,
+50,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,110,0,16,10,50,0,57,59,121,121,121,121,0,
+48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,
+57,18,110,0,16,10,50,0,57,59,119,119,119,119,0,48,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
+51,0,57,18,109,0,16,8,48,0,57,18,110,0,16,10,51,0,57,59,120,120,120,120,0,48,18,109,0,16,10,49,0,
+57,18,110,0,16,10,51,0,57,59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,110,0,16,10,51,0,57,
+59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,59,119,119,119,119,0,48,
+46,20,0,0,1,90,95,0,0,15,221,2,22,1,1,0,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,49,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,49,20,0,0,1,90,
+95,0,0,13,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+97,0,18,110,0,16,10,49,0,57,46,20,0,0,1,90,95,0,0,13,221,2,26,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,
+0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,
+95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,0,1,90,95,0,0,13,
+221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,
+18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,
+110,0,16,10,49,0,57,47,20,0,0,1,90,95,0,0,13,221,2,27,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,47,20,0,9,18,95,95,114,
+101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47,20,0,0,1,90,95,0,0,13,221,2,21,
+1,1,0,0,9,97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,
+110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,
+49,0,57,48,20,0,0,1,90,95,0,0,13,221,2,21,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,
+97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,0,1,90,95,0,0,13,221,2,22,1,1,0,0,9,
+97,0,0,1,1,0,0,13,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,
+8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,
+49,20,0,0,1,90,95,0,0,13,221,2,22,1,1,0,0,13,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,
+86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,49,20,0,0,1,90,95,0,0,14,221,2,26,1,1,0,0,9,97,0,0,1,
+1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,
+46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,46,20,0,0,1,90,95,0,
+0,14,221,2,26,1,1,0,0,14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
+57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,
+16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,
+57,18,98,0,46,20,0,0,1,90,95,0,0,14,221,2,27,1,1,0,0,9,97,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,0,0,1,90,95,0,0,14,221,2,27,1,1,0,0,14,109,0,
 0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,
-0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,49,20,0,
-0,1,0,0,14,2,26,1,1,0,0,9,97,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
-0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,
-0,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,
-16,10,50,0,57,46,20,0,0,1,0,0,14,2,26,1,1,0,0,14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0,0,1,0,0,14,2,27,1,1,0,0,9,97,0,0,1,1,0,0,14,110,
-0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,0,0,1,0,0,14,2,27,1,1,0,0,
-14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,
-57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,
-47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,47,20,0,0,
-1,0,0,14,2,21,1,1,0,0,9,97,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,
-18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,
-16,10,50,0,57,48,20,0,0,1,0,0,14,2,21,1,1,0,0,14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,
-10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,0,1,0,0,14,2,22,1,1,0,0,9,97,0,0,1,1,0,0,14,110,
-0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,49,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,49,20,0,0,1,0,0,14,2,22,1,1,0,0,
-14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,
-57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,
-49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,49,20,0,0,
-1,0,0,15,2,26,1,1,0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,
-18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,
-16,10,50,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,
-57,46,20,0,0,1,0,0,15,2,26,1,1,0,0,15,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,
-49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,
-109,0,16,10,50,0,57,18,98,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,
-10,51,0,57,18,98,0,46,20,0,0,1,0,0,15,2,27,1,1,0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,
-97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,
-16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,
-57,18,97,0,18,110,0,16,10,51,0,57,47,20,0,0,1,0,0,15,2,27,1,1,0,0,15,109,0,0,1,1,0,0,9,98,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,47,20,0,9,18,95,95,
-114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47,20,0,9,18,95,95,114,101,
-116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,
-108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,47,20,0,0,1,0,0,15,2,21,1,1,0,0,9,97,0,0,1,1,0,
-0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,
-20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,
-95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,
-101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,48,20,0,0,1,0,0,15,2,21,1,1,0,0,
-15,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,
-57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,
-48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,9,
-18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,48,20,0,0,1,0,0,15,2,
-22,1,1,0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,
-18,110,0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,
-10,49,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,
-49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,49,20,0,0,
-1,0,0,15,2,22,1,1,0,0,15,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,
-57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,
-16,10,49,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,
-57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,
-49,20,0,0,1,0,0,10,2,21,1,1,0,0,13,109,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,18,109,0,16,8,48,0,57,18,118,0,59,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,121,0,48,
-46,20,0,0,1,0,0,10,2,21,1,1,0,0,10,118,0,0,1,1,0,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,
-0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,
-108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,0,1,0,0,11,2,21,1,1,0,0,
-14,109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,16,8,48,0,57,18,118,
-0,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,121,121,0,48,46,18,109,0,16,10,50,0,
-57,18,118,0,59,122,122,122,0,48,46,20,0,0,1,0,0,11,2,21,1,1,0,0,11,118,0,0,1,1,0,0,14,109,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,
-20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,
-57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,
-50,0,57,0,0,20,0,0,1,0,0,12,2,21,1,1,0,0,15,109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,18,109,0,16,8,48,0,57,18,118,0,59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,
-59,121,121,121,121,0,48,46,18,109,0,16,10,50,0,57,18,118,0,59,122,122,122,122,0,48,46,18,109,0,16,
-10,51,0,57,18,118,0,59,119,119,119,119,0,48,46,20,0,0,1,0,0,12,2,21,1,1,0,0,12,118,0,0,1,1,0,0,15,
-109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,
-48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,
-16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,
-109,0,16,10,50,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,119,0,58,100,111,116,0,18,118,0,
-0,18,109,0,16,10,51,0,57,0,0,20,0,0,1,0,0,0,2,1,1,0,2,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,
-0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,
-0,0,0,2,2,1,0,2,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,
-22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,0,1,0,0,0,2,3,1,0,2,0,13,109,0,0,1,1,0,0,
-13,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,0,0,0,2,4,1,0,2,0,13,109,0,0,1,1,0,0,13,
-110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,
-16,10,49,0,57,24,0,0,1,0,0,0,2,1,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,
-18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,
-50,0,57,18,110,0,16,10,50,0,57,21,0,0,1,0,0,0,2,2,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,
-109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,
-9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,0,1,0,0,0,2,3,1,0,2,0,14,109,0,0,1,1,0,0,14,
-110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,0,0,0,2,4,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,
-0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,
-49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,24,0,0,1,0,0,0,2,1,1,0,2,0,15,109,0,0,
-1,1,0,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,
-18,110,0,16,10,49,0,57,21,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,9,18,109,0,16,10,
-51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,0,0,0,2,2,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,
-109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,
-9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,
-57,22,0,0,1,0,0,0,2,3,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,
-0,0,1,0,0,0,2,4,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,
-0,57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,
-16,10,50,0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,0,0,1,0,0,0,2,1,1,0,2,0,13,
+0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,47,20,0,
+9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,47,20,0,0,1,90,95,
+0,0,14,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
+0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,
+0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,
+16,10,50,0,57,48,20,0,0,1,90,95,0,0,14,221,2,21,1,1,0,0,14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,0,1,90,95,0,0,14,221,2,22,1,1,0,0,9,97,0,0,
+1,1,0,0,14,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,
+57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,49,20,0,
+9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,49,20,0,0,1,90,95,
+0,0,14,221,2,22,1,1,0,0,14,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,
+0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,
+0,16,10,49,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,
+0,57,18,98,0,49,20,0,0,1,90,95,0,0,15,221,2,26,1,1,0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,
+114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,46,20,0,9,18,95,95,114,101,116,
+86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,
+0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,46,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,
+0,57,18,97,0,18,110,0,16,10,51,0,57,46,20,0,0,1,90,95,0,0,15,221,2,26,1,1,0,0,15,109,0,0,1,1,0,0,9,
+98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,46,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,46,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,46,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,46,20,0,0,1,90,95,0,0,15,221,2,27,1,1,
+0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,
+0,16,8,48,0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,
+0,57,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,47,20,
+0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,47,20,0,0,1,90,
+95,0,0,15,221,2,27,1,1,0,0,15,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,18,109,0,16,8,48,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+109,0,16,10,49,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,
+10,50,0,57,18,98,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,
+18,98,0,47,20,0,0,1,90,95,0,0,15,221,2,21,1,1,0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,0,16,8,48,0,57,48,20,0,9,18,95,95,114,101,116,86,
+97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,
+16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,48,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,
+57,18,97,0,18,110,0,16,10,51,0,57,48,20,0,0,1,90,95,0,0,15,221,2,21,1,1,0,0,15,109,0,0,1,1,0,0,9,
+98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,109,0,16,8,48,0,57,18,98,0,48,20,0,9,
+18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,109,0,16,10,49,0,57,18,98,0,48,20,0,9,18,95,95,
+114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,10,50,0,57,18,98,0,48,20,0,9,18,95,95,114,101,
+116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,18,98,0,48,20,0,0,1,90,95,0,0,15,221,2,22,1,1,
+0,0,9,97,0,0,1,1,0,0,15,110,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,48,0,57,18,97,0,18,110,
+0,16,8,48,0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,97,0,18,110,0,16,10,49,
+0,57,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,97,0,18,110,0,16,10,50,0,57,49,20,
+0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,97,0,18,110,0,16,10,51,0,57,49,20,0,0,1,90,
+95,0,0,15,221,2,22,1,1,0,0,15,109,0,0,1,1,0,0,9,98,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,16,8,
+48,0,57,18,109,0,16,8,48,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,49,0,57,18,
+109,0,16,10,49,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,50,0,57,18,109,0,16,
+10,50,0,57,18,98,0,49,20,0,9,18,95,95,114,101,116,86,97,108,0,16,10,51,0,57,18,109,0,16,10,51,0,57,
+18,98,0,49,20,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,13,109,0,0,1,1,0,0,10,118,0,0,0,1,9,18,95,95,114,
+101,116,86,97,108,0,18,109,0,16,8,48,0,57,18,118,0,59,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,
+59,121,121,0,48,46,20,0,0,1,90,95,0,0,10,221,2,21,1,1,0,0,10,118,0,0,1,1,0,0,13,109,0,0,0,1,9,18,
+95,95,114,101,116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,
+9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,
+20,0,0,1,90,95,0,0,11,221,2,21,1,1,0,0,14,109,0,0,1,1,0,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,18,109,0,16,8,48,0,57,18,118,0,59,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,
+121,121,0,48,46,18,109,0,16,10,50,0,57,18,118,0,59,122,122,122,0,48,46,20,0,0,1,90,95,0,0,11,221,2,
+21,1,1,0,0,11,118,0,0,1,1,0,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,59,120,0,58,100,
+111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,121,0,58,
+100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,95,114,101,116,86,97,108,0,59,122,
+0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,15,
+109,0,0,1,1,0,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,16,8,48,0,57,18,118,0,
+59,120,120,120,120,0,48,18,109,0,16,10,49,0,57,18,118,0,59,121,121,121,121,0,48,46,18,109,0,16,10,
+50,0,57,18,118,0,59,122,122,122,122,0,48,46,18,109,0,16,10,51,0,57,18,118,0,59,119,119,119,119,0,
+48,46,20,0,0,1,90,95,0,0,12,221,2,21,1,1,0,0,12,118,0,0,1,1,0,0,15,109,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,59,120,0,58,100,111,116,0,18,118,0,0,18,109,0,16,8,48,0,57,0,0,20,0,9,18,95,95,114,
+101,116,86,97,108,0,59,121,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,49,0,57,0,0,20,0,9,18,95,
+95,114,101,116,86,97,108,0,59,122,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,50,0,57,0,0,20,0,9,
+18,95,95,114,101,116,86,97,108,0,59,119,0,58,100,111,116,0,18,118,0,0,18,109,0,16,10,51,0,57,0,0,
+20,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,
+110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,0,1,90,95,0,0,0,221,2,
+2,1,0,2,0,13,109,0,0,1,1,0,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,
+18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,13,109,0,0,1,1,
+0,0,13,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,13,109,0,0,
+1,1,0,0,13,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,0,9,18,109,0,16,10,49,0,57,
+18,110,0,16,10,49,0,57,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,
+109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,
+9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,14,109,0,0,1,
+1,0,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,
+18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,0,57,22,0,0,1,90,95,0,0,0,
+221,2,3,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,
+0,0,221,2,4,1,0,2,0,14,109,0,0,1,1,0,0,14,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,
+57,24,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,
+10,50,0,57,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,16,8,
+48,0,57,18,110,0,16,8,48,0,57,21,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,21,0,9,18,109,0,
+16,10,50,0,57,18,110,0,16,10,50,0,57,21,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,21,0,0,1,
+90,95,0,0,0,221,2,2,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,
+8,48,0,57,22,0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,22,0,9,18,109,0,16,10,50,0,57,18,
+110,0,16,10,50,0,57,22,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,22,0,0,1,90,95,0,0,0,221,
+2,3,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,18,109,0,18,110,0,48,20,0,0,1,90,95,0,0,0,
+221,2,4,1,0,2,0,15,109,0,0,1,1,0,0,15,110,0,0,0,1,9,18,109,0,16,8,48,0,57,18,110,0,16,8,48,0,57,24,
+0,9,18,109,0,16,10,49,0,57,18,110,0,16,10,49,0,57,24,0,9,18,109,0,16,10,50,0,57,18,110,0,16,10,50,
+0,57,24,0,9,18,109,0,16,10,51,0,57,18,110,0,16,10,51,0,57,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,13,
 109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,
-21,0,0,1,0,0,0,2,2,1,0,2,0,13,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,
-18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,0,0,0,2,3,1,0,2,0,13,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,
-0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,0,0,0,2,4,1,0,2,0,13,109,0,0,
-1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,0,
-1,0,0,0,2,1,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,
-0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,0,1,0,0,0,2,2,1,0,2,0,14,109,0,
-0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,
-9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,0,0,0,2,3,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,
-109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,
-97,0,23,0,0,1,0,0,0,2,4,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,
-0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,0,1,0,0,0,2,1,1,0,2,
-0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,
-97,0,21,0,9,18,109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,0,0,0,2,
-2,1,0,2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,
-0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,
-0,0,0,2,3,1,0,2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,
-16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,97,0,
-23,0,0,1,0,0,0,2,4,1,0,2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,
-18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,18,109,0,16,10,51,0,57,
-18,97,0,24,0,0,1,0,0,0,2,3,1,0,2,0,10,118,0,0,1,1,0,0,13,109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,
-48,20,0,0,1,0,0,0,2,3,1,0,2,0,11,118,0,0,1,1,0,0,14,109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,
-0,0,1,0,0,0,2,3,1,0,2,0,12,118,0,0,1,1,0,0,15,109,0,0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,
-0,0,5,2,25,1,0,2,0,5,97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,47,20,0,9,18,95,95,114,101,116,86,97,
-108,0,18,97,0,20,0,0,1,0,0,6,2,25,1,0,2,0,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,50,0,
-16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,7,2,25,1,0,2,0,7,
-118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,
-116,86,97,108,0,18,118,0,20,0,0,1,0,0,8,2,25,1,0,2,0,8,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,
-101,99,52,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,9,2,25,
-1,0,2,0,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,
-18,97,0,20,0,0,1,0,0,10,2,25,1,0,2,0,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,
-48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,11,2,25,1,0,2,0,11,118,
-0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,
-86,97,108,0,18,118,0,20,0,0,1,0,0,12,2,25,1,0,2,0,12,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,
-52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,13,2,25,1,
-0,2,0,13,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,
-0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,
-0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,14,2,25,1,0,2,0,14,109,0,0,0,1,
-9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,
-109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,
-0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,
-114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,15,2,25,1,0,2,0,15,109,0,0,0,1,9,18,109,0,16,8,48,0,
-57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,
-109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,
-0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,
-10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,
-0,20,0,0,1,0,0,5,2,24,1,0,2,0,5,97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,46,20,0,9,18,95,95,114,101,
-116,86,97,108,0,18,97,0,20,0,0,1,0,0,6,2,24,1,0,2,0,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,
-101,99,50,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,7,2,24,
-1,0,2,0,7,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,9,18,95,95,
-114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,8,2,24,1,0,2,0,8,118,0,0,0,1,9,18,118,0,18,118,0,58,
-105,118,101,99,52,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,
-0,9,2,24,1,0,2,0,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,46,20,0,9,18,95,95,114,101,116,86,
-97,108,0,18,97,0,20,0,0,1,0,0,10,2,24,1,0,2,0,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,
-0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,11,2,24,1,0,
-2,0,11,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,
-114,101,116,86,97,108,0,18,118,0,20,0,0,1,0,0,12,2,24,1,0,2,0,12,118,0,0,0,1,9,18,118,0,18,118,0,
-58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,
-0,0,13,2,24,1,0,2,0,13,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,
-0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,
-49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,14,2,24,1,0,2,0,
-14,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,
-46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,
-20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,
-9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,0,0,15,2,24,1,0,2,0,15,109,0,0,0,1,9,18,109,0,
-16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,
-49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,
-57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,
-18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,
-108,0,18,109,0,20,0,0,1,0,0,5,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,5,97,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,47,20,0,0,1,0,0,6,0,95,95,112,
+21,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,13,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,
+0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,13,109,0,0,1,1,0,0,9,
+97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,109,0,16,10,49,0,57,18,97,0,23,0,0,1,90,95,0,
+0,0,221,2,4,1,0,2,0,13,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,
+0,16,10,49,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,
+109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,109,0,16,10,50,0,57,18,
+97,0,21,0,0,1,90,95,0,0,0,221,2,2,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,
+18,97,0,22,0,9,18,109,0,16,10,49,0,57,18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,0,1,90,95,
+0,0,0,221,2,3,1,0,2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,
+109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,0,1,90,95,0,0,0,221,2,4,1,0,
+2,0,14,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,24,0,9,18,109,0,16,10,49,0,57,
+18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,1,1,0,2,0,15,109,0,0,1,1,
+0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,21,0,9,18,109,0,16,10,49,0,57,18,97,0,21,0,9,18,
+109,0,16,10,50,0,57,18,97,0,21,0,9,18,109,0,16,10,51,0,57,18,97,0,21,0,0,1,90,95,0,0,0,221,2,2,1,0,
+2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,22,0,9,18,109,0,16,10,49,0,57,
+18,97,0,22,0,9,18,109,0,16,10,50,0,57,18,97,0,22,0,9,18,109,0,16,10,51,0,57,18,97,0,22,0,0,1,90,95,
+0,0,0,221,2,3,1,0,2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,18,97,0,23,0,9,18,
+109,0,16,10,49,0,57,18,97,0,23,0,9,18,109,0,16,10,50,0,57,18,97,0,23,0,9,18,109,0,16,10,51,0,57,18,
+97,0,23,0,0,1,90,95,0,0,0,221,2,4,1,0,2,0,15,109,0,0,1,1,0,0,9,97,0,0,0,1,9,18,109,0,16,8,48,0,57,
+18,97,0,24,0,9,18,109,0,16,10,49,0,57,18,97,0,24,0,9,18,109,0,16,10,50,0,57,18,97,0,24,0,9,18,109,
+0,16,10,51,0,57,18,97,0,24,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,10,118,0,0,1,1,0,0,13,109,0,0,0,1,9,
+18,118,0,18,118,0,18,109,0,48,20,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,11,118,0,0,1,1,0,0,14,109,0,0,0,
+1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,90,95,0,0,0,221,2,3,1,0,2,0,12,118,0,0,1,1,0,0,15,109,0,
+0,0,1,9,18,118,0,18,118,0,18,109,0,48,20,0,0,1,90,95,0,0,5,221,2,25,1,0,2,0,5,97,0,0,0,1,9,18,97,0,
+18,97,0,16,10,49,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,1,90,95,0,0,6,221,2,
+25,1,0,2,0,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,47,20,0,9,18,95,
+95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,7,221,2,25,1,0,2,0,7,118,0,0,0,1,9,18,118,0,
+18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,
+20,0,0,1,90,95,0,0,8,221,2,25,1,0,2,0,8,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,
+10,49,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,9,221,2,25,1,0,
+2,0,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97,
+0,20,0,0,1,90,95,0,0,10,221,2,25,1,0,2,0,10,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,50,0,17,
+49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,11,221,2,25,
+1,0,2,0,11,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,
+95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,12,221,2,25,1,0,2,0,12,118,0,0,0,1,9,18,118,
+0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,
+0,20,0,0,1,90,95,0,0,13,221,2,25,1,0,2,0,13,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,
+57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,
+58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,
+90,95,0,0,14,221,2,25,1,0,2,0,14,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,
+101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,
+99,51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,
+51,0,17,49,0,48,0,0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,15,
+221,2,25,1,0,2,0,15,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,
+17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,
+49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,
+48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,
+0,0,0,47,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,5,221,2,24,1,0,2,0,5,
+97,0,0,0,1,9,18,97,0,18,97,0,16,10,49,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,0,
+1,90,95,0,0,6,221,2,24,1,0,2,0,6,118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,
+0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,7,221,2,24,1,0,2,0,7,
+118,0,0,0,1,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,
+116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,8,221,2,24,1,0,2,0,8,118,0,0,0,1,9,18,118,0,18,118,0,
+58,105,118,101,99,52,0,16,10,49,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,
+90,95,0,0,9,221,2,24,1,0,2,0,9,97,0,0,0,1,9,18,97,0,18,97,0,17,49,0,48,0,0,46,20,0,9,18,95,95,114,
+101,116,86,97,108,0,18,97,0,20,0,0,1,90,95,0,0,10,221,2,24,1,0,2,0,10,118,0,0,0,1,9,18,118,0,18,
+118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,
+0,0,1,90,95,0,0,11,221,2,24,1,0,2,0,11,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,
+48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,12,221,2,24,1,0,
+2,0,12,118,0,0,0,1,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,
+114,101,116,86,97,108,0,18,118,0,20,0,0,1,90,95,0,0,13,221,2,24,1,0,2,0,13,109,0,0,0,1,9,18,109,0,
+16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,
+49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,
+116,86,97,108,0,18,109,0,20,0,0,1,90,95,0,0,14,221,2,24,1,0,2,0,14,109,0,0,0,1,9,18,109,0,16,8,48,
+0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,
+18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,
+109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,
+0,18,109,0,20,0,0,1,90,95,0,0,15,221,2,24,1,0,2,0,15,109,0,0,0,1,9,18,109,0,16,8,48,0,57,18,109,0,
+16,8,48,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,
+49,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,
+57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,
+58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,0,1,
+90,95,0,0,5,221,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,5,97,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,47,20,0,0,1,90,95,0,0,6,221,0,95,95,112,
 111,115,116,68,101,99,114,0,1,0,2,0,6,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,
-9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,47,20,0,0,1,0,0,7,0,95,95,112,111,115,
-116,68,101,99,114,0,1,0,2,0,7,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,
-118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,0,1,0,0,8,0,95,95,112,111,115,116,68,
-101,99,114,0,1,0,2,0,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,
-118,0,58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,0,1,0,0,9,0,95,95,112,111,115,116,68,101,99,
-114,0,1,0,2,0,9,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,17,49,
-0,48,0,0,47,20,0,0,1,0,0,10,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,10,118,0,0,0,1,9,18,95,
-95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,
-47,20,0,0,1,0,0,11,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,11,118,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,
-0,0,1,0,0,12,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,12,118,0,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,0,1,0,
-0,13,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,
-108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,
-0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,
-0,47,20,0,0,1,0,0,14,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,14,109,0,0,0,1,9,18,95,95,114,
-101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,
-17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,
-49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,
-48,0,0,0,0,47,20,0,0,1,0,0,15,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,15,109,0,0,0,1,9,18,
-95,95,114,101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,
-101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,
-99,52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,
-52,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,
-17,49,0,48,0,0,0,0,47,20,0,0,1,0,0,9,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,9,97,0,0,0,1,
-9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,0,0,10,0,
-95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
-118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,0,11,0,95,95,
-112,111,115,116,73,110,99,114,0,1,0,2,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,
-20,0,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,0,12,0,95,95,112,111,
-115,116,73,110,99,114,0,1,0,2,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,
-18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,0,1,0,0,5,0,95,95,112,111,115,116,
-73,110,99,114,0,1,0,2,0,5,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,
-97,0,16,10,49,0,46,20,0,0,1,0,0,6,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,6,118,0,0,0,1,9,
-18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,
-0,0,0,46,20,0,0,1,0,0,7,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,7,118,0,0,0,1,9,18,95,95,
-114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,
-20,0,0,1,0,0,8,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,8,118,0,0,0,1,9,18,95,95,114,101,
-116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,
-1,0,0,13,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,13,109,0,0,0,1,3,2,0,0,13,1,110,0,2,18,
-109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,
-20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,
-8,18,110,0,0,0,1,0,0,14,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,14,109,0,0,0,1,3,2,0,0,14,
-1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,
-0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,
-0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,
-46,20,0,8,18,110,0,0,0,1,0,0,15,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,15,109,0,0,0,1,3,2,
-0,0,15,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,
-49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,
-48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,
-0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,
-0,46,20,0,8,18,110,0,0,0,1,0,0,1,2,15,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,
-103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,98,0,0,18,97,0,0,0,0,1,0,0,1,2,15,1,1,0,0,
-5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,
-0,0,0,40,0,0,1,0,0,1,2,16,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,1,1,99,0,0,0,4,102,108,111,
-97,116,95,108,101,115,115,0,18,99,0,0,18,98,0,0,18,97,0,0,0,8,18,99,0,0,0,1,0,0,1,2,16,1,1,0,0,5,
-97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,
-0,0,41,0,0,1,0,0,1,2,18,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,1,1,103,0,0,1,1,101,0,0,0,4,
-102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,98,0,0,18,97,0,0,0,4,102,108,111,97,116,95,
-101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,0,0,1,2,18,1,1,
-0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,
-98,0,0,0,43,0,0,1,0,0,1,2,17,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,0,0,1,1,103,0,0,1,1,101,0,0,
-0,4,102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,97,0,0,18,98,0,0,0,4,102,108,111,97,116,
-95,101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,0,0,1,2,17,1,
-1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,
-18,98,0,0,0,42,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,9,102,0,0,0,1,4,102,108,111,
-97,116,95,112,114,105,110,116,0,18,102,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,
-5,105,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,105,0,0,0,0,1,0,0,0,0,112,114,105,110,116,
-77,69,83,65,0,1,1,0,0,1,98,0,0,0,1,4,98,111,111,108,95,112,114,105,110,116,0,18,98,0,0,0,0,1,0,0,0,
-0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,10,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,
-18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,0,
-112,114,105,110,116,77,69,83,65,0,1,1,0,0,11,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,
-118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,
-105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,
-1,0,0,12,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,
-105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,
-59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,0,112,114,
-105,110,116,77,69,83,65,0,1,1,0,0,6,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,
-120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,0,112,114,105,
-110,116,77,69,83,65,0,1,1,0,0,7,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,
-0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,
-69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,8,118,0,0,
-0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,
-83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,
-58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,
-83,65,0,1,1,0,0,2,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,
-112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,
-65,0,1,1,0,0,3,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,
-114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,
-118,0,59,122,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,4,118,0,0,0,1,9,58,112,
-114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,
-118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,
-105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,
-1,0,0,13,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,
-114,105,110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,
+9,18,118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,47,20,0,0,1,90,95,0,0,7,221,0,95,95,112,
+111,115,116,68,101,99,114,0,1,0,2,0,7,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,
+9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,47,20,0,0,1,90,95,0,0,8,221,0,95,95,112,
+111,115,116,68,101,99,114,0,1,0,2,0,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,
+9,18,118,0,18,118,0,58,105,118,101,99,52,0,16,10,49,0,0,0,47,20,0,0,1,90,95,0,0,9,221,0,95,95,112,
+111,115,116,68,101,99,114,0,1,0,2,0,9,97,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,97,0,20,0,9,
+18,97,0,18,97,0,17,49,0,48,0,0,47,20,0,0,1,90,95,0,0,10,221,0,95,95,112,111,115,116,68,101,99,114,
+0,1,0,2,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,
+118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,0,1,90,95,0,0,11,221,0,95,95,112,111,115,116,68,101,99,
+114,0,1,0,2,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,
+58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,0,1,90,95,0,0,12,221,0,95,95,112,111,115,116,68,101,
+99,114,0,1,0,2,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,118,0,18,118,
+0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,47,20,0,0,1,90,95,0,0,13,221,0,95,95,112,111,115,116,68,
+101,99,114,0,1,0,2,0,13,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,
+8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,
+0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,47,20,0,0,1,90,95,0,0,14,221,0,
+95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,14,109,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,
+109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,
+20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,
+9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,0,0,0,47,20,0,0,1,
+90,95,0,0,15,221,0,95,95,112,111,115,116,68,101,99,114,0,1,0,2,0,15,109,0,0,0,1,9,18,95,95,114,101,
+116,86,97,108,0,18,109,0,20,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,52,0,17,
+49,0,48,0,0,0,0,47,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,52,0,17,49,0,
+48,0,0,0,0,47,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,52,0,17,49,0,48,0,
+0,0,0,47,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,101,99,52,0,17,49,0,48,0,0,0,
+0,47,20,0,0,1,90,95,0,0,9,221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,9,97,0,0,0,1,9,18,95,
+95,114,101,116,86,97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,90,95,0,0,10,221,
+0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,10,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,
+18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,0,1,90,95,0,0,11,
+221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,11,118,0,0,0,1,9,18,95,95,114,101,116,86,97,
+108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,51,0,17,49,0,48,0,0,0,0,46,20,0,0,1,90,95,0,
+0,12,221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,12,118,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,18,118,0,20,0,9,18,118,0,18,118,0,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,0,1,90,95,
+0,0,5,221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,5,97,0,0,0,1,9,18,95,95,114,101,116,86,
+97,108,0,18,97,0,20,0,9,18,97,0,18,97,0,16,10,49,0,46,20,0,0,1,90,95,0,0,6,221,0,95,95,112,111,115,
+116,73,110,99,114,0,1,0,2,0,6,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,18,
+118,0,18,118,0,58,105,118,101,99,50,0,16,10,49,0,0,0,46,20,0,0,1,90,95,0,0,7,221,0,95,95,112,111,
+115,116,73,110,99,114,0,1,0,2,0,7,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,9,
+18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,90,95,0,0,8,221,0,95,95,112,
+111,115,116,73,110,99,114,0,1,0,2,0,8,118,0,0,0,1,9,18,95,95,114,101,116,86,97,108,0,18,118,0,20,0,
+9,18,118,0,18,118,0,58,105,118,101,99,51,0,16,10,49,0,0,0,46,20,0,0,1,90,95,0,0,13,221,0,95,95,112,
+111,115,116,73,110,99,114,0,1,0,2,0,13,109,0,0,0,1,3,2,90,95,0,0,13,221,1,110,0,2,18,109,0,0,0,9,
+18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,
+0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,50,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,
+0,1,90,95,0,0,14,221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,14,109,0,0,0,1,3,2,90,95,0,0,
+14,221,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,57,58,118,101,99,51,0,17,
+49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,58,118,101,99,51,0,17,49,0,
+48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,118,101,99,51,0,17,49,0,48,0,
+0,0,0,46,20,0,8,18,110,0,0,0,1,90,95,0,0,15,221,0,95,95,112,111,115,116,73,110,99,114,0,1,0,2,0,15,
+109,0,0,0,1,3,2,90,95,0,0,15,221,1,110,0,2,18,109,0,0,0,9,18,109,0,16,8,48,0,57,18,109,0,16,8,48,0,
+57,58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,49,0,57,18,109,0,16,10,49,0,57,
+58,118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,50,0,57,18,109,0,16,10,50,0,57,58,
+118,101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,9,18,109,0,16,10,51,0,57,18,109,0,16,10,51,0,57,58,118,
+101,99,52,0,17,49,0,48,0,0,0,0,46,20,0,8,18,110,0,0,0,1,90,95,0,0,1,221,2,15,1,1,0,0,9,97,0,0,1,1,
+0,0,9,98,0,0,0,1,4,118,101,99,52,95,115,103,116,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,
+98,0,0,18,97,0,0,0,0,1,90,95,0,0,1,221,2,15,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,
+97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,0,0,40,0,0,1,90,95,0,0,1,221,2,16,1,1,0,0,9,
+97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,1,221,1,99,0,0,0,4,102,108,111,97,116,95,108,101,115,115,
+0,18,99,0,0,18,98,0,0,18,97,0,0,0,8,18,99,0,0,0,1,90,95,0,0,1,221,2,16,1,1,0,0,5,97,0,0,1,1,0,0,5,
+98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,0,0,41,0,0,1,90,
+95,0,0,1,221,2,18,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,90,95,0,0,1,221,1,103,0,0,1,1,101,0,0,
+0,4,102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,98,0,0,18,97,0,0,0,4,102,108,111,97,116,
+95,101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,103,0,18,101,0,32,0,0,1,90,95,0,0,1,
+221,2,18,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,111,97,116,0,18,97,0,0,0,58,102,108,
+111,97,116,0,18,98,0,0,0,43,0,0,1,90,95,0,0,1,221,2,17,1,1,0,0,9,97,0,0,1,1,0,0,9,98,0,0,0,1,3,2,
+90,95,0,0,1,221,1,103,0,0,1,1,101,0,0,0,4,102,108,111,97,116,95,108,101,115,115,0,18,103,0,0,18,97,
+0,0,18,98,0,0,0,4,102,108,111,97,116,95,101,113,117,97,108,0,18,101,0,0,18,97,0,0,18,98,0,0,0,8,18,
+103,0,18,101,0,32,0,0,1,90,95,0,0,1,221,2,17,1,1,0,0,5,97,0,0,1,1,0,0,5,98,0,0,0,1,8,58,102,108,
+111,97,116,0,18,97,0,0,0,58,102,108,111,97,116,0,18,98,0,0,0,42,0,0,1,90,95,0,0,0,221,0,112,114,
+105,110,116,77,69,83,65,0,1,1,0,0,9,102,0,0,0,1,4,102,108,111,97,116,95,112,114,105,110,116,0,18,
+102,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,5,105,0,0,0,1,4,105,110,
+116,95,112,114,105,110,116,0,18,105,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,
+1,1,0,0,1,98,0,0,0,1,4,98,111,111,108,95,112,114,105,110,116,0,18,98,0,0,0,0,1,90,95,0,0,0,221,0,
+112,114,105,110,116,77,69,83,65,0,1,1,0,0,10,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,
+118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,0,1,90,95,0,0,
+0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,11,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,
+65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,
+112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,
+116,77,69,83,65,0,1,1,0,0,12,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,
+0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,
+69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,119,0,0,0,0,
+0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,6,118,0,0,0,1,9,58,112,114,105,110,
+116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,
+0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,7,118,0,0,0,1,9,58,112,114,
+105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,
+59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,0,1,90,95,0,0,0,221,
+0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,8,118,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,
+18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,
+114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,
+118,0,59,119,0,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,2,118,0,0,0,1,
+9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,
+65,0,18,118,0,59,121,0,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,3,118,
+0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,
+69,83,65,0,18,118,0,59,121,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,
+0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,4,118,0,0,0,1,9,58,112,114,105,110,
+116,77,69,83,65,0,18,118,0,59,120,0,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,121,0,
+0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,118,0,59,122,0,0,0,0,9,58,112,114,105,110,116,77,
+69,83,65,0,18,118,0,59,119,0,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,
+13,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,
+110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,
 83,65,0,1,1,0,0,14,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,
 9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,9,58,112,114,105,110,116,77,69,
-83,65,0,18,109,0,16,10,50,0,57,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,15,109,
-0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,105,110,
-116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,
-10,50,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,51,0,57,0,0,0,0,1,0,0,0,0,
-112,114,105,110,116,77,69,83,65,0,1,1,0,0,16,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,
-101,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,17,101,0,0,0,1,4,105,110,116,95,
-112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,18,101,0,
-0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,
-65,0,1,1,0,0,19,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,0,0,0,0,112,
-114,105,110,116,77,69,83,65,0,1,1,0,0,20,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,
-0,0,0,0,1,0,0,0,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,21,101,0,0,0,1,4,105,110,116,95,112,
-114,105,110,116,0,18,101,0,0,0,0,0
+83,65,0,18,109,0,16,10,50,0,57,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,
+0,15,109,0,0,0,1,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,8,48,0,57,0,0,0,9,58,112,114,
+105,110,116,77,69,83,65,0,18,109,0,16,10,49,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,
+109,0,16,10,50,0,57,0,0,0,9,58,112,114,105,110,116,77,69,83,65,0,18,109,0,16,10,51,0,57,0,0,0,0,1,
+90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,16,101,0,0,0,1,4,105,110,116,95,112,
+114,105,110,116,0,18,101,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,17,
+101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,
+110,116,77,69,83,65,0,1,1,0,0,18,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,
+1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,19,101,0,0,0,1,4,105,110,116,95,112,
+114,105,110,116,0,18,101,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,110,116,77,69,83,65,0,1,1,0,0,20,
+101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,1,90,95,0,0,0,221,0,112,114,105,
+110,116,77,69,83,65,0,1,1,0,0,21,101,0,0,0,1,4,105,110,116,95,112,114,105,110,116,0,18,101,0,0,0,0,
+0
diff --git a/src/mesa/shader/slang/library/slang_fragment_builtin_gc.h b/src/mesa/shader/slang/library/slang_fragment_builtin_gc.h
index ff1a85d4cee..e2a135fe0dc 100644
--- a/src/mesa/shader/slang/library/slang_fragment_builtin_gc.h
+++ b/src/mesa/shader/slang/library/slang_fragment_builtin_gc.h
@@ -2,100 +2,105 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_fragment_builtin.gc */
 
-4,2,2,6,0,12,1,103,108,95,70,114,97,103,67,111,111,114,100,0,0,0,2,2,6,0,1,1,103,108,95,70,114,111,
-110,116,70,97,99,105,110,103,0,0,0,2,2,5,0,12,1,103,108,95,70,114,97,103,67,111,108,111,114,0,0,0,
-2,2,5,0,12,1,103,108,95,70,114,97,103,68,97,116,97,0,3,18,103,108,95,77,97,120,68,114,97,119,66,
-117,102,102,101,114,115,0,0,0,2,2,5,0,9,1,103,108,95,70,114,97,103,68,101,112,116,104,0,0,0,2,2,3,
-0,12,1,103,108,95,67,111,108,111,114,0,0,0,2,2,3,0,12,1,103,108,95,83,101,99,111,110,100,97,114,
-121,67,111,108,111,114,0,0,0,2,2,3,0,12,1,103,108,95,84,101,120,67,111,111,114,100,0,3,18,103,108,
-95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,3,0,9,1,103,108,95,70,111,
-103,70,114,97,103,67,111,111,114,100,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,49,68,0,1,1,0,0,
-16,115,97,109,112,108,101,114,0,0,1,1,0,0,9,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,
-3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,0,18,99,111,111,114,
-100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,
-101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,
-111,114,100,52,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,49,68,80,114,111,106,0,1,1,0,0,16,
-115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,
-2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,0,18,99,111,111,114,
-100,0,59,120,0,18,99,111,111,114,100,0,59,121,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,
-98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,
-18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,
-114,101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,
-100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,
-111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,
-18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,
-100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,
-100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,50,68,0,1,1,0,0,17,115,97,109,112,108,101,114,0,
-0,1,1,0,0,10,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,99,111,111,114,
-100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,20,0,9,
-18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,
-100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,
-52,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,
-108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,112,
-99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,
-121,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,
-115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
-109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,
+4,2,2,90,95,6,0,12,221,1,103,108,95,70,114,97,103,67,111,111,114,100,0,0,0,2,2,90,95,6,0,1,221,1,
+103,108,95,70,114,111,110,116,70,97,99,105,110,103,0,0,0,2,2,90,95,5,0,12,221,1,103,108,95,70,114,
+97,103,67,111,108,111,114,0,0,0,2,2,90,95,5,0,12,221,1,103,108,95,70,114,97,103,68,97,116,97,0,3,
+18,103,108,95,77,97,120,68,114,97,119,66,117,102,102,101,114,115,0,0,0,2,2,90,95,5,0,9,221,1,103,
+108,95,70,114,97,103,68,101,112,116,104,0,0,0,2,2,90,95,3,0,12,221,1,103,108,95,67,111,108,111,114,
+0,0,0,2,2,90,95,3,0,12,221,1,103,108,95,83,101,99,111,110,100,97,114,121,67,111,108,111,114,0,0,0,
+2,2,90,95,3,0,12,221,1,103,108,95,84,101,120,67,111,111,114,100,0,3,18,103,108,95,77,97,120,84,101,
+120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,3,0,9,221,1,103,108,95,70,111,103,70,
+114,97,103,67,111,111,114,100,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,0,1,1,0,
+0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,9,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,
+1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,0,18,99,
+111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,
+52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
+18,99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,80,114,111,
+106,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,0,0,1,1,0,0,9,98,105,
+97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,
+59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,121,0,49,20,0,9,18,112,99,111,
+111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,
+114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,
+95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,80,114,111,106,0,1,1,0,0,16,115,97,109,112,108,
+101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,
+1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,
+120,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,
+115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
+109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,
+114,101,50,68,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,0,0,1,1,0,
+0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,
+100,52,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,20,0,9,18,99,111,111,114,100,52,0,59,
+119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,
+97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,
+116,101,120,116,117,114,101,50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,
+0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,
+114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,18,
+99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,
+0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,
+108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,
 50,68,80,114,111,106,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,
-1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,
-100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,
-18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,
-100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,
-100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,51,68,0,1,1,0,0,18,115,97,109,112,108,101,114,0,
-0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,99,111,111,114,
-100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,59,120,121,122,
-0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,
-120,98,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,
-114,100,52,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,51,68,80,114,111,106,0,1,1,0,0,18,115,97,
-109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,
-12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,122,0,18,99,111,111,114,
-100,0,59,120,121,122,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,59,
-119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,51,100,0,18,95,95,114,101,116,86,
-97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,
-120,116,117,114,101,67,117,98,101,0,1,1,0,0,19,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,
-111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,18,99,
-111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,
-119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,99,117,98,101,0,18,95,95,114,101,116,
-86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,0,0,12,0,115,
-104,97,100,111,119,49,68,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,
-0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,
-100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,
-105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,
-115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,
-49,68,80,114,111,106,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,
-1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,
-100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,
-99,111,111,114,100,0,59,122,0,18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,
-59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,
-86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,115,
-104,97,100,111,119,50,68,0,1,1,0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,
-0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,
-100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,
-105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,
-115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,
-50,68,80,114,111,106,0,1,1,0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,
-1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,
-100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,
-18,112,99,111,111,114,100,0,59,122,0,18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,
-100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,
-116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,9,0,
-100,70,100,120,0,1,1,0,0,9,112,0,0,0,1,4,118,101,99,52,95,100,100,120,0,18,95,95,114,101,116,86,97,
-108,0,59,120,0,0,18,112,0,59,120,120,120,120,0,0,0,0,1,0,0,10,0,100,70,100,120,0,1,1,0,0,10,112,0,
-0,0,1,4,118,101,99,52,95,100,100,120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,112,0,59,
-120,121,121,121,0,0,0,0,1,0,0,11,0,100,70,100,120,0,1,1,0,0,11,112,0,0,0,1,4,118,101,99,52,95,100,
-100,120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,112,0,59,120,121,122,122,0,0,0,0,
-1,0,0,12,0,100,70,100,120,0,1,1,0,0,12,112,0,0,0,1,4,118,101,99,52,95,100,100,120,0,18,95,95,114,
-101,116,86,97,108,0,0,18,112,0,0,0,0,1,0,0,9,0,100,70,100,121,0,1,1,0,0,9,112,0,0,0,1,4,118,101,99,
-52,95,100,100,121,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,112,0,59,120,120,120,120,0,0,0,
-0,1,0,0,10,0,100,70,100,121,0,1,1,0,0,10,112,0,0,0,1,4,118,101,99,52,95,100,100,121,0,18,95,95,114,
-101,116,86,97,108,0,59,120,121,0,0,18,112,0,59,120,121,121,121,0,0,0,0,1,0,0,11,0,100,70,100,121,0,
-1,1,0,0,11,112,0,0,0,1,4,118,101,99,52,95,100,100,121,0,18,95,95,114,101,116,86,97,108,0,59,120,
-121,122,0,0,18,112,0,59,120,121,122,122,0,0,0,0,1,0,0,12,0,100,70,100,121,0,1,1,0,0,12,112,0,0,0,1,
-4,118,101,99,52,95,100,100,121,0,18,95,95,114,101,116,86,97,108,0,0,18,112,0,0,0,0,1,0,0,9,0,102,
-119,105,100,116,104,0,1,1,0,0,9,112,0,0,0,1,8,58,97,98,115,0,58,100,70,100,120,0,18,112,0,0,0,0,0,
-58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,0,0,0,46,0,0,1,0,0,10,0,102,119,105,100,116,104,0,1,
-1,0,0,10,112,0,0,0,1,8,58,97,98,115,0,58,100,70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,
-70,100,121,0,18,112,0,0,0,0,0,46,0,0,1,0,0,11,0,102,119,105,100,116,104,0,1,1,0,0,11,112,0,0,0,1,8,
-58,97,98,115,0,58,100,70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,
-0,0,0,46,0,0,1,0,0,12,0,102,119,105,100,116,104,0,1,1,0,0,12,112,0,0,0,1,8,58,97,98,115,0,58,100,
-70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,0,0,0,46,0,0,0
+1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,
+111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,
+0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,
+101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,
+111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,51,68,0,1,1,0,0,18,115,97,
+109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,
+95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,
+111,111,114,100,0,59,120,121,122,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,
+20,0,4,118,101,99,52,95,116,101,120,98,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,
+112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,
+101,51,68,80,114,111,106,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,
+0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,
+99,111,111,114,100,0,59,120,121,122,0,18,99,111,111,114,100,0,59,120,121,122,0,18,99,111,111,114,
+100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,
+52,95,116,101,120,98,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
+18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,67,117,98,101,0,
+1,1,0,0,19,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,
+115,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,
+121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,
+0,4,118,101,99,52,95,116,101,120,99,117,98,101,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,
+112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,
+49,68,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,98,
+105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,
+59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,98,105,97,
+115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
+109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,
+119,49,68,80,114,111,106,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,
+0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,
+99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,119,0,49,
+20,0,9,18,112,99,111,111,114,100,0,59,122,0,18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,
+111,114,100,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,
+114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,
+95,0,0,12,221,0,115,104,97,100,111,119,50,68,0,1,1,0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,
+11,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,
+100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,
+111,111,114,100,52,0,59,119,0,18,98,105,97,115,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,
+18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,
+0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,50,68,80,114,111,106,0,1,1,0,0,21,115,97,109,112,
+108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,98,105,97,115,0,0,0,1,3,2,90,95,0,0,12,
+221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,
+100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,59,122,0,
+18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,98,105,97,115,0,20,
+0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,
+108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,9,221,0,100,70,100,120,0,1,1,0,0,9,
+112,0,0,0,1,4,118,101,99,52,95,100,100,120,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,112,0,
+59,120,120,120,120,0,0,0,0,1,90,95,0,0,10,221,0,100,70,100,120,0,1,1,0,0,10,112,0,0,0,1,4,118,101,
+99,52,95,100,100,120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,112,0,59,120,121,121,121,
+0,0,0,0,1,90,95,0,0,11,221,0,100,70,100,120,0,1,1,0,0,11,112,0,0,0,1,4,118,101,99,52,95,100,100,
+120,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,112,0,59,120,121,122,122,0,0,0,0,1,90,
+95,0,0,12,221,0,100,70,100,120,0,1,1,0,0,12,112,0,0,0,1,4,118,101,99,52,95,100,100,120,0,18,95,95,
+114,101,116,86,97,108,0,0,18,112,0,0,0,0,1,90,95,0,0,9,221,0,100,70,100,121,0,1,1,0,0,9,112,0,0,0,
+1,4,118,101,99,52,95,100,100,121,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,112,0,59,120,120,
+120,120,0,0,0,0,1,90,95,0,0,10,221,0,100,70,100,121,0,1,1,0,0,10,112,0,0,0,1,4,118,101,99,52,95,
+100,100,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,112,0,59,120,121,121,121,0,0,0,0,
+1,90,95,0,0,11,221,0,100,70,100,121,0,1,1,0,0,11,112,0,0,0,1,4,118,101,99,52,95,100,100,121,0,18,
+95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,112,0,59,120,121,122,122,0,0,0,0,1,90,95,0,0,
+12,221,0,100,70,100,121,0,1,1,0,0,12,112,0,0,0,1,4,118,101,99,52,95,100,100,121,0,18,95,95,114,101,
+116,86,97,108,0,0,18,112,0,0,0,0,1,90,95,0,0,9,221,0,102,119,105,100,116,104,0,1,1,0,0,9,112,0,0,0,
+1,8,58,97,98,115,0,58,100,70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,
+0,0,0,0,0,46,0,0,1,90,95,0,0,10,221,0,102,119,105,100,116,104,0,1,1,0,0,10,112,0,0,0,1,8,58,97,98,
+115,0,58,100,70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,0,0,0,46,
+0,0,1,90,95,0,0,11,221,0,102,119,105,100,116,104,0,1,1,0,0,11,112,0,0,0,1,8,58,97,98,115,0,58,100,
+70,100,120,0,18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,0,0,0,46,0,0,1,90,95,0,
+0,12,221,0,102,119,105,100,116,104,0,1,1,0,0,12,112,0,0,0,1,8,58,97,98,115,0,58,100,70,100,120,0,
+18,112,0,0,0,0,0,58,97,98,115,0,58,100,70,100,121,0,18,112,0,0,0,0,0,46,0,0,0
diff --git a/src/mesa/shader/slang/library/slang_pp_directives.syn b/src/mesa/shader/slang/library/slang_pp_directives.syn
index d4a321034d1..d4a321034d1 100755..100644
--- a/src/mesa/shader/slang/library/slang_pp_directives.syn
+++ b/src/mesa/shader/slang/library/slang_pp_directives.syn
diff --git a/src/mesa/shader/slang/library/slang_pp_expression.syn b/src/mesa/shader/slang/library/slang_pp_expression.syn
index bfdb220bf5c..bfdb220bf5c 100755..100644
--- a/src/mesa/shader/slang/library/slang_pp_expression.syn
+++ b/src/mesa/shader/slang/library/slang_pp_expression.syn
diff --git a/src/mesa/shader/slang/library/slang_shader.syn b/src/mesa/shader/slang/library/slang_shader.syn
index ef4b410669c..760dfbcea7e 100644
--- a/src/mesa/shader/slang/library/slang_shader.syn
+++ b/src/mesa/shader/slang/library/slang_shader.syn
@@ -32,18 +32,28 @@
  * usage:
  *   syn2c slang_shader.syn > slang_shader_syn.h
  *
- * when modifying or extending this file, several things must be taken into consideration:
- * - when adding new operators that were marked as reserved in the initial specification,
- *   one must only uncomment particular lines of code that refer to operators being added;
- * - when adding new shader target, one must reserve new value for shader_type register and
- *   use it in .if constructs for symbols that are exclusive for that shader;
- * - some symbols mimic output of other symbols - the best example is the "for" construct:
- *   expression "for (foo(); ; bar())" is seen as "for (foo(); true; bar())" by the output
- *   processor - hence, special care must be taken when rearranging output of essential symbols;
- * - order of single-quoted tokens does matter in alternatives - so do not parse "<" operator
- *   before "<<" and "<<" before "<<=";
- * - all double-quoted tokens are internally preprocessed to eliminate problems with parsing
- *   strings that are prefixes of other strings, like "sampler1D" and "sampler1DShadow";
+ * when modifying or extending this file, several things must be taken into
+ * consideration:
+ *
+ * - when adding new operators that were marked as reserved in the
+ *   initial specification, one must only uncomment particular lines of
+ *   code that refer to operators being added;
+ *
+ * - when adding new shader targets, one must reserve a new value for
+ *   shader_type register and use it in .if constructs for symbols that
+ *   are exclusive for that shader;
+ *
+ * - some symbols mimic output of other symbols - the best example is
+ *   the "for" construct: expression "for (foo(); ; bar())" is seen as
+ *   "for (foo(); true; bar())" by the output processor - hence, special
+ *   care must be taken when rearranging output of essential symbols;
+ *
+ * - order of single-quoted tokens does matter in alternatives - so do not
+ *   parse "<" operator before "<<" and "<<" before "<<=";
+ *
+ * - all double-quoted tokens are internally preprocessed to eliminate
+ *   problems with parsing strings that are prefixes of other strings,
+ *   like "sampler1D" and "sampler1DShadow";
  */
 
 .syntax translation_unit;
@@ -124,6 +134,14 @@
 .emtcode TYPE_QUALIFIER_FIXEDOUTPUT                 5
 .emtcode TYPE_QUALIFIER_FIXEDINPUT                  6
 
+/* invariant qualifier */
+.emtcode TYPE_VARIANT                               90
+.emtcode TYPE_INVARIANT                             91
+
+/* centroid qualifier */
+.emtcode TYPE_CENTER                                95
+.emtcode TYPE_CENTROID                              96
+
 /* type specifier */
 .emtcode TYPE_SPECIFIER_VOID                        0
 .emtcode TYPE_SPECIFIER_BOOL                        1
@@ -160,6 +178,9 @@
 .emtcode TYPE_SPECIFIER_MAT34                       30
 .emtcode TYPE_SPECIFIER_MAT43                       31
 
+/* optional array size.  Ex: "float[6] myarray;" */
+.emtcode TYPE_ARRAY_SIZE                           220
+.emtcode TYPE_NO_ARRAY_SIZE                        221
 
 /* structure field */
 .emtcode FIELD_NONE                                 0
@@ -230,6 +251,7 @@
 .emtcode OP_POSTINCREMENT                           60
 .emtcode OP_POSTDECREMENT                           61
 .emtcode OP_PRECISION                               62
+.emtcode OP_METHOD                                  63
 
 /* parameter qualifier */
 .emtcode PARAM_QUALIFIER_IN                         0
@@ -255,44 +277,49 @@
 .errtext INVALID_PRECISION_TYPE                     "2007: Invalid precision type '$err_token$'."
 
 
-/* tells whether the shader that is being parsed is a built-in shader or not */
-/*   0 - normal behaviour */
-/*   1 - accepts constructor and operator definitions and __asm statements */
-/* the implementation will set it to 1 when compiling internal built-in shaders */
+/*
+ * tells whether the shader that is being parsed is a built-in shader or not
+ *  0 - normal behaviour
+ *  1 - accepts constructor and operator definitions and __asm statements
+ * the implementation will set it to 1 when compiling internal built-in shaders
+ */
 .regbyte parsing_builtin                            0
 
-/* holds the type of the shader being parsed; possible values are listed below */
-/*   FRAGMENT_SHADER            1 */
-/*   VERTEX_SHADER              2 */
-/* shader type is set by the caller before parsing */
+/*
+ * holds the type of the shader being parsed; possible values are
+ * listed below.
+ *   FRAGMENT_SHADER            1
+ *   VERTEX_SHADER              2
+ * shader type is set by the caller before parsing
+ */
 .regbyte shader_type                                0
 
 /*
-    <variable_identifier>               ::= <identifier>
-*/
+ * <variable_identifier> ::= <identifier>
+ */
 variable_identifier
     identifier .emit OP_PUSH_IDENTIFIER;
 
 /*
-    <primary_expression>                ::= <variable_identifier>
-                                          | <intconstant>
-                                          | <floatconstant>
-                                          | <boolconstant>
-                                          | "(" <expression> ")"
-*/
+ *  <primary_expression> ::= <variable_identifier>
+ *                         | <intconstant>
+ *                         | <floatconstant>
+ *                         | <boolconstant>
+ *                         | "(" <expression> ")"
+ */
 primary_expression
     floatconstant .or boolconstant .or intconstant .or variable_identifier .or primary_expression_1;
 primary_expression_1
     lparen .and expression .and rparen;
 
 /*
-    <postfix_expression>                ::= <primary_expression>
-                                          | <postfix_expression> "[" <integer_expression> "]"
-                                          | <function_call>
-                                          | <postfix_expression> "." <field_selection>
-                                          | <postfix_expression> "++"
-                                          | <postfix_expression> "--"
-*/
+ * <postfix_expression> ::= <primary_expression>
+ *                        | <postfix_expression> "[" <integer_expression> "]"
+ *                        | <function_call>
+ *                        | <postfix_expression> "." <field_selection>
+ *                        | <postfix_expression> "++"
+ *                        | <postfix_expression> "--"
+ */
 postfix_expression
     postfix_expression_1 .and .loop postfix_expression_2;
 postfix_expression_1
@@ -307,21 +334,40 @@ postfix_expression_4
     dot .and field_selection .emit OP_FIELD;
 
 /*
-    <integer_expression>                ::= <expression>
-*/
+ * <integer_expression> ::= <expression>
+ */
 integer_expression
     expression;
 
 /*
-    <function_call>                     ::= <function_call_generic>
-*/
+ * <function_call> ::= <function_call_generic>
+ */
 function_call
+    function_call_or_method;
+
+/*
+ * <function_call_or_method> ::= <regular_function_call>
+ *                             | <postfix_expression> "." <function_call_generic>
+ */
+function_call_or_method
+    regular_function_call .or method_call;
+
+/*
+ * <method_call> ::= <identifier> "." <function_call_generic>
+ */
+method_call
+    identifier .emit OP_METHOD .and dot .and function_call_generic .and .true .emit OP_END;
+
+/*
+ * <regular_function_call> ::= <function_call_generic>
+ */
+regular_function_call
     function_call_generic .emit OP_CALL .and .true .emit OP_END;
 
 /*
-    <function_call_generic>             ::= <function_call_header_with_parameters> ")"
-                                          | <function_call_header_no_parameters> ")"
-*/
+ * <function_call_generic> ::= <function_call_header_with_parameters> ")"
+ *                           | <function_call_header_no_parameters> ")"
+ */
 function_call_generic
     function_call_generic_1 .or function_call_generic_2;
 function_call_generic_1
@@ -330,19 +376,18 @@ function_call_generic_2
     function_call_header_no_parameters .and rparen .error RPAREN_EXPECTED;
 
 /*
-    <function_call_header_no_parameters>::= <function_call_header> "void"
-                                          | <function_call_header>
-*/
+ * <function_call_header_no_parameters>::= <function_call_header> "void"
+ *                                        | <function_call_header>
+ */
 function_call_header_no_parameters
     function_call_header .and function_call_header_no_parameters_1;
 function_call_header_no_parameters_1
     "void" .or .true;
 
 /*
-    <function_call_header_with_parameters>::= <function_call_header> <assignment_expression>
-                                            | <function_call_header_with_parameters> ","
-                                              <assignment_expression>
-*/
+ * <function_call_header_with_parameters> ::= <function_call_header> <assignment_expression>
+ *                                          | <function_call_header_with_parameters> "," <assignment_expression>
+ */
 function_call_header_with_parameters
     function_call_header .and assignment_expression .and .true .emit OP_END .and
     .loop function_call_header_with_parameters_1;
@@ -350,31 +395,31 @@ function_call_header_with_parameters_1
     comma .and assignment_expression .and .true .emit OP_END;
 
 /*
-    <function_call_header>              ::= <function_identifier> "("
-*/
+ * <function_call_header> ::= <function_identifier> "("
+ */
 function_call_header
     function_identifier .and lparen;
 
 /*
-    <function_identifier>               ::= <constructor_identifier>
-                                          | <identifier>
-
-note: <constructor_identifier> has been deleted
-*/
+ * <function_identifier> ::= <constructor_identifier>
+ *                         | <identifier>
+ *
+ * note: <constructor_identifier> has been deleted
+ */
 function_identifier
     identifier;
 
 /*
-    <unary_expression>                  ::= <postfix_expression>
-                                          | "++" <unary_expression>
-                                          | "--" <unary_expression>
-                                          | <unary_operator> <unary_expression>
-
-    <unary_operator>                    ::= "+"
-                                          | "-"
-                                          | "!"
-                                          | "~" // reserved
-*/
+ * <unary_expression> ::= <postfix_expression>
+ *                      | "++" <unary_expression>
+ *                      | "--" <unary_expression>
+ *                      | <unary_operator> <unary_expression>
+ *
+ * <unary_operator>   ::= "+"
+ *                      | "-"
+ *                      | "!"
+ *                      | "~" // reserved
+ */
 unary_expression
     postfix_expression .or unary_expression_1 .or unary_expression_2 .or unary_expression_3 .or
     unary_expression_4 .or unary_expression_5/* .or unary_expression_6*/;
@@ -392,11 +437,11 @@ unary_expression_5
     tilde .and unary_expression .and .true .emit OP_COMPLEMENT;*/
 
 /*
-    <multiplicative_expression>         ::= <unary_expression>
-                                          | <multiplicative_expression> "*" <unary_expression>
-                                          | <multiplicative_expression> "/" <unary_expression>
-                                          | <multiplicative_expression> "%" <unary_expression> // reserved
-*/
+ * <multiplicative_expression> ::= <unary_expression>
+ *                               | <multiplicative_expression> "*" <unary_expression>
+ *                               | <multiplicative_expression> "/" <unary_expression>
+ *                               | <multiplicative_expression> "%" <unary_expression> // reserved
+ */
 multiplicative_expression
     unary_expression .and .loop multiplicative_expression_1;
 multiplicative_expression_1
@@ -409,10 +454,10 @@ multiplicative_expression_3
     percent .and unary_expression .and .true .emit OP_MODULUS;*/
 
 /*
-    <additive_expression>               ::= <multiplicative_expression>
-                                          | <additive_expression> "+" <multiplicative_expression>
-                                          | <additive_expression> "-" <multiplicative_expression>
-*/
+ * <additive_expression> ::= <multiplicative_expression>
+ *                         | <additive_expression> "+" <multiplicative_expression>
+ *                         | <additive_expression> "-" <multiplicative_expression>
+ */
 additive_expression
     multiplicative_expression .and .loop additive_expression_1;
 additive_expression_1
@@ -423,10 +468,10 @@ additive_expression_3
     minus .and multiplicative_expression .and .true .emit OP_SUBTRACT;
 
 /*
-    <shift_expression>                  ::= <additive_expression>
-                                          | <shift_expression> "<<" <additive_expression> // reserved
-                                          | <shift_expression> ">>" <additive_expression> // reserved
-*/
+ * <shift_expression> ::= <additive_expression>
+ *                      | <shift_expression> "<<" <additive_expression> // reserved
+ *                      | <shift_expression> ">>" <additive_expression> // reserved
+ */
 shift_expression
     additive_expression/* .and .loop shift_expression_1*/;
 /*shift_expression_1
@@ -437,12 +482,12 @@ shift_expression
     greatergreater .and additive_expression .and .true .emit OP_RSHIFT;*/
 
 /*
-    <relational_expression>             ::= <shift_expression>
-                                          | <relational_expression> "<" <shift_expression>
-                                          | <relational_expression> ">" <shift_expression>
-                                          | <relational_expression> "<=" <shift_expression>
-                                          | <relational_expression> ">=" <shift_expression>
-*/
+ * <relational_expression> ::= <shift_expression>
+ *                           | <relational_expression> "<" <shift_expression>
+ *                           | <relational_expression> ">" <shift_expression>
+ *                           | <relational_expression> "<=" <shift_expression>
+ *                           | <relational_expression> ">=" <shift_expression>
+ */
 relational_expression
     shift_expression .and .loop relational_expression_1;
 relational_expression_1
@@ -458,10 +503,10 @@ relational_expression_5
     greater .and shift_expression .and .true .emit OP_GREATER;
 
 /*
-    <equality_expression>               ::= <relational_expression>
-                                          | <equality_expression> "==" <relational_expression>
-                                          | <equality_expression> "!=" <relational_expression>
-*/
+ * <equality_expression> ::= <relational_expression>
+ *                         | <equality_expression> "==" <relational_expression>
+ *                         | <equality_expression> "!=" <relational_expression>
+ */
 equality_expression
     relational_expression .and .loop equality_expression_1;
 equality_expression_1
@@ -472,86 +517,84 @@ equality_expression_3
     bangequals .and relational_expression .and .true .emit OP_NOTEQUAL;
 
 /*
-    <and_expression>                    ::= <equality_expression>
-                                          | <and_expression> "&" <equality_expression> // reserved
-*/
+ * <and_expression> ::= <equality_expression>
+ *                    | <and_expression> "&" <equality_expression> // reserved
+ */
 and_expression
     equality_expression/* .and .loop and_expression_1*/;
 /*and_expression_1
     ampersand .and equality_expression .and .true .emit OP_BITAND;*/
 
 /*
-    <exclusive_or_expression>           ::= <and_expression>
-                                          | <exclusive_or_expression> "^" <and_expression> // reserved
-*/
+ * <exclusive_or_expression> ::= <and_expression>
+ *                             | <exclusive_or_expression> "^" <and_expression> // reserved
+ */
 exclusive_or_expression
     and_expression/* .and .loop exclusive_or_expression_1*/;
 /*exclusive_or_expression_1
     caret .and and_expression .and .true .emit OP_BITXOR;*/
 
 /*
-    <inclusive_or_expression>           ::= <exclusive_or_expression>
-                                          | <inclusive_or_expression> "|" <exclusive_or_expression> // reserved
-*/
+ * <inclusive_or_expression> ::= <exclusive_or_expression>
+ *                             | <inclusive_or_expression> "|" <exclusive_or_expression> // reserved
+ */
 inclusive_or_expression
     exclusive_or_expression/* .and .loop inclusive_or_expression_1*/;
 /*inclusive_or_expression_1
     bar .and exclusive_or_expression .and .true .emit OP_BITOR;*/
 
 /*
-    <logical_and_expression>            ::= <inclusive_or_expression>
-                                          | <logical_and_expression> "&&" <inclusive_or_expression>
-*/
+ * <logical_and_expression> ::= <inclusive_or_expression>
+ *                            | <logical_and_expression> "&&" <inclusive_or_expression>
+ */
 logical_and_expression
     inclusive_or_expression .and .loop logical_and_expression_1;
 logical_and_expression_1
     ampersandampersand .and inclusive_or_expression .and .true .emit OP_LOGICALAND;
 
 /*
-    <logical_xor_expression>            ::= <logical_and_expression>
-                                          | <logical_xor_expression> "^^" <logical_and_expression>
-*/
+ * <logical_xor_expression> ::= <logical_and_expression>
+ *                            | <logical_xor_expression> "^^" <logical_and_expression>
+ */
 logical_xor_expression
     logical_and_expression .and .loop logical_xor_expression_1;
 logical_xor_expression_1
     caretcaret .and logical_and_expression .and .true .emit OP_LOGICALXOR;
 
 /*
-    <logical_or_expression>             ::= <logical_xor_expression>
-                                          | <logical_or_expression> "||" <logical_xor_expression>
-*/
+ * <logical_or_expression> ::= <logical_xor_expression>
+ *                           | <logical_or_expression> "||" <logical_xor_expression>
+ */
 logical_or_expression
     logical_xor_expression .and .loop logical_or_expression_1;
 logical_or_expression_1
     barbar .and logical_xor_expression .and .true .emit OP_LOGICALOR;
 
 /*
-    <conditional_expression>            ::= <logical_or_expression>
-                                          | <logical_or_expression> "?" <expression> ":"
-                                            <conditional_expression>
-*/
+ * <conditional_expression> ::= <logical_or_expression>
+ *                            | <logical_or_expression> "?" <expression> ":" <conditional_expression>
+ */
 conditional_expression
     logical_or_expression .and .loop conditional_expression_1;
 conditional_expression_1
     question .and expression .and colon .and conditional_expression .and .true .emit OP_SELECT;
 
 /*
-    <assignment_expression>             ::= <conditional_expression>
-                                          | <unary_expression> <assignment_operator>
-                                            <assignment_expression>
-
-    <assignment_operator>               ::= "="
-                                          | "*="
-                                          | "/="
-                                          | "+="
-                                          | "-="
-                                          | "%=" // reserved
-                                          | "<<=" // reserved
-                                          | ">>=" // reserved
-                                          | "&=" // reserved
-                                          | "^=" // reserved
-                                          | "|=" // reserved
-*/
+ * <assignment_expression> ::= <conditional_expression>
+ *                           | <unary_expression> <assignment_operator> <assignment_expression>
+ *
+ * <assignment_operator> ::= "="
+ *                         | "*="
+ *                         | "/="
+ *                         | "+="
+ *                         | "-="
+ *                         | "%=" // reserved
+ *                         | "<<=" // reserved
+ *                         | ">>=" // reserved
+ *                         | "&=" // reserved
+ *                         | "^=" // reserved
+ *                         | "|=" // reserved
+ */
 assignment_expression
     assignment_expression_1 .or assignment_expression_2 .or assignment_expression_3 .or
     assignment_expression_4 .or assignment_expression_5/* .or assignment_expression_6 .or
@@ -582,24 +625,24 @@ assignment_expression_5
     unary_expression .and barequals .and assignment_expression .and .true .emit OP_ORASSIGN;*/
 
 /*
-    <expression>                        ::= <assignment_expression>
-                                          | <expression> "," <assignment_expression>
-*/
+ * <expression> ::= <assignment_expression>
+ *                | <expression> "," <assignment_expression>
+ */
 expression
     assignment_expression .and .loop expression_1;
 expression_1
     comma .and assignment_expression .and .true .emit OP_SEQUENCE;
 
 /*
-    <constant_expression>               ::= <conditional_expression>
-*/
+ * <constant_expression> ::= <conditional_expression>
+ */
 constant_expression
     conditional_expression .and .true .emit OP_END;
 
 /*
-    <declaration>                       ::= <function_prototype> ";"
-                                          | <init_declarator_list> ";"
-*/
+ * <declaration> ::= <function_prototype> ";"
+ *                 | <init_declarator_list> ";"
+ */
 declaration
     declaration_1 .or declaration_2;
 declaration_1
@@ -608,9 +651,9 @@ declaration_2
     init_declarator_list .emit DECLARATION_INIT_DECLARATOR_LIST .and semicolon;
 
 /*
-    <function_prototype>                ::= <function_header> "void" ")"
-                                          | <function_declarator> ")"
-*/
+ * <function_prototype> ::= <function_header> "void" ")"
+ *                        | <function_declarator> ")"
+ */
 function_prototype
     function_prototype_1 .or function_prototype_2;
 function_prototype_1
@@ -619,25 +662,25 @@ function_prototype_2
     function_declarator .and rparen .error RPAREN_EXPECTED .emit PARAMETER_NONE;
 
 /*
-    <function_declarator>               ::= <function_header>
-                                          | <function_header_with_parameters>
-*/
+ * <function_declarator> ::= <function_header>
+ *                         | <function_header_with_parameters>
+ */
 function_declarator
     function_header_with_parameters .or function_header;
 
 /*
-    <function_header_with_parameters>   ::= <function_header> <parameter_declaration>
-                                          | <function_header_with_parameters> ","
-                                            <parameter_declaration>
-*/
+ * <function_header_with_parameters> ::= <function_header> <parameter_declaration>
+ *                                     | <function_header_with_parameters> ","
+ *                                       <parameter_declaration>
+ */
 function_header_with_parameters
     function_header .and parameter_declaration .and .loop function_header_with_parameters_1;
 function_header_with_parameters_1
     comma .and parameter_declaration;
 
 /*
-    <function_header>                   ::= <fully_specified_type> <identifier> "("
-*/
+ * <function_header> ::= <fully_specified_type> <identifier> "("
+ */
 function_header
     function_header_nospace .or function_header_space;
 function_header_space
@@ -646,64 +689,64 @@ function_header_nospace
     fully_specified_type_nospace .and function_decl_identifier .and lparen;
 
 /*
-    <function_decl_identifier>          ::= "__constructor"
-                                          | <__operator>
-                                          | <identifier>
-
-note: this is an extension to the standard language specification - normally slang disallows
-      operator and constructor prototypes and definitions
-*/
+ * <function_decl_identifier> ::= "__constructor"
+ *                              | <__operator>
+ *                              | <identifier>
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows operator and constructor prototypes and definitions
+ */
 function_decl_identifier
     .if (parsing_builtin != 0) __operator .emit FUNCTION_OPERATOR .or
     .if (parsing_builtin != 0) "__constructor" .emit FUNCTION_CONSTRUCTOR .or
     identifier .emit FUNCTION_ORDINARY;
 
 /*
-    <__operator>                        ::= "__operator" <overriden_op>
-
-note: this is an extension to the standard language specification - normally slang disallows
-      operator prototypes and definitions
-*/
+ * <__operator> ::= "__operator" <overriden_op>
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows operator prototypes and definitions
+ */
 __operator
     "__operator" .and overriden_operator .error INVALID_OPERATOR_OVERRIDE;
 
 /*
-    <overriden_op>                      ::= "="
-                                          | "+="
-                                          | "-="
-                                          | "*="
-                                          | "/="
-                                          | "%=" // reserved
-                                          | "<<=" // reserved
-                                          | ">>=" // reserved
-                                          | "&=" // reserved
-                                          | "^=" // reserved
-                                          | "|=" // reserved
-                                          | "^^"
-                                          | "|" // reserved
-                                          | "^" // reserved
-                                          | "&" // reserved
-                                          | "=="
-                                          | "!="
-                                          | "<"
-                                          | ">"
-                                          | "<="
-                                          | ">="
-                                          | "<<" // reserved
-                                          | ">>" // reserved
-                                          | "*"
-                                          | "/"
-                                          | "%" // reserved
-                                          | "++"
-                                          | "--"
-                                          | "+"
-                                          | "-"
-                                          | "~" // reserved
-                                          | "!"
-
-note: this is an extension to the standard language specification - normally slang disallows
-      operator prototypes and definitions
-*/
+ * <overriden_op> ::= "="
+ *                  | "+="
+ *                  | "-="
+ *                  | "*="
+ *                  | "/="
+ *                  | "%=" // reserved
+ *                  | "<<=" // reserved
+ *                  | ">>=" // reserved
+ *                  | "&=" // reserved
+ *                  | "^=" // reserved
+ *                  | "|=" // reserved
+ *                  | "^^"
+ *                  | "|" // reserved
+ *                  | "^" // reserved
+ *                  | "&" // reserved
+ *                  | "=="
+ *                  | "!="
+ *                  | "<"
+ *                  | ">"
+ *                  | "<="
+ *                  | ">="
+ *                  | "<<" // reserved
+ *                  | ">>" // reserved
+ *                  | "*"
+ *                  | "/"
+ *                  | "%" // reserved
+ *                  | "++"
+ *                  | "--"
+ *                  | "+"
+ *                  | "-"
+ *                  | "~" // reserved
+ *                  | "!"
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows operator prototypes and definitions
+ */
 overriden_operator
     plusplus .emit OPERATOR_INCREMENT .or
     plusequals .emit OPERATOR_ADDASSIGN .or
@@ -736,10 +779,9 @@ overriden_operator
     caret .emit OPERATOR_BITXOR*/;
 
 /*
-    <parameter_declarator>              ::= <type_specifier> <identifier>
-                                          | <type_specifier> <identifier> "[" <constant_expression>
-                                            "]"
-*/
+ * <parameter_declarator> ::= <type_specifier> <identifier>
+ *                          | <type_specifier> <identifier> "[" <constant_expression> "]"
+ */
 parameter_declarator
     parameter_declarator_nospace .or parameter_declarator_space;
 parameter_declarator_nospace
@@ -753,21 +795,21 @@ parameter_declarator_2
     lbracket .and constant_expression .and rbracket;
 
 /*
-    <parameter_declaration>             ::= <type_qualifier> <parameter_qualifier>
-                                            <precision> <parameter_declarator>
-                                          | <type_qualifier> <parameter_qualifier>
-                                            <precision> <parameter_type_specifier>
-                                          | <type_qualifier> <parameter_qualifier>
-                                            <parameter_declarator>
-                                          | <type_qualifier> <parameter_qualifier>
-                                            <parameter_type_specifier>
-                                          | <parameter_qualifier> <precision>
-                                            <parameter_declarator>
-                                          | <parameter_qualifier> <precision>
-                                            <parameter_type_specifier>
-                                          | <parameter_qualifier> <parameter_declarator>
-                                          | <parameter_qualifier> <parameter_type_specifier>
-*/
+ * <parameter_declaration> ::= <type_qualifier> <parameter_qualifier>
+ *                             <precision> <parameter_declarator>
+ *                           | <type_qualifier> <parameter_qualifier>
+ *                             <precision> <parameter_type_specifier>
+ *                           | <type_qualifier> <parameter_qualifier>
+ *                             <parameter_declarator>
+ *                           | <type_qualifier> <parameter_qualifier>
+ *                             <parameter_type_specifier>
+ *                           | <parameter_qualifier> <precision>
+ *                             <parameter_declarator>
+ *                           | <parameter_qualifier> <precision>
+ *                             <parameter_type_specifier>
+ *                           | <parameter_qualifier> <parameter_declarator>
+ *                           | <parameter_qualifier> <parameter_type_specifier>
+ */
 parameter_declaration
     parameter_declaration_1 .emit PARAMETER_NEXT;
 parameter_declaration_1
@@ -786,11 +828,11 @@ parameter_declaration_rest
     parameter_declarator .or parameter_type_specifier;
 
 /*
-    <parameter_qualifier>               ::= "in"
-                                          | "out"
-                                          | "inout"
-                                          | ""
-*/
+ * <parameter_qualifier> ::= "in"
+ *                         | "out"
+ *                         | "inout"
+ *                         | ""
+ */
 parameter_qualifier
     parameter_qualifier_1 .or .true .emit PARAM_QUALIFIER_IN;
 parameter_qualifier_1
@@ -801,9 +843,9 @@ parameter_qualifier_2
     "inout" .emit PARAM_QUALIFIER_INOUT;
 
 /*
-    <parameter_type_specifier>          ::= <type_specifier>
-                                          | <type_specifier> "[" <constant_expression> "]"
-*/
+ * <parameter_type_specifier> ::= <type_specifier>
+ *                              | <type_specifier> "[" <constant_expression> "]"
+ */
 parameter_type_specifier
     parameter_type_specifier_1 .and .true .emit '\0' .and parameter_type_specifier_2;
 parameter_type_specifier_1
@@ -815,14 +857,12 @@ parameter_type_specifier_3
     lbracket .and constant_expression .and rbracket;
 
 /*
-    <init_declarator_list>              ::= <single_declaration>
-                                          | <init_declarator_list> "," <identifier>
-                                          | <init_declarator_list> "," <identifier> "[" "]"
-                                          | <init_declarator_list> "," <identifier> "["
-                                            <constant_expression> "]"
-                                          | <init_declarator_list> "," <identifier> "="
-                                            <initializer>
-*/
+ * <init_declarator_list> ::= <single_declaration>
+ *                          | <init_declarator_list> "," <identifier>
+ *                          | <init_declarator_list> "," <identifier> "[" "]"
+ *                          | <init_declarator_list> "," <identifier> "[" <constant_expression> "]"
+ *                          | <init_declarator_list> "," <identifier> "=" <initializer>
+ */
 init_declarator_list
     single_declaration .and .loop init_declarator_list_1 .emit DECLARATOR_NEXT .and
     .true .emit DECLARATOR_NONE;
@@ -838,13 +878,12 @@ init_declarator_list_5
     constant_expression .emit VARIABLE_ARRAY_EXPLICIT .or .true .emit VARIABLE_ARRAY_UNKNOWN;
 
 /*
-    <single_declaration>                ::= <fully_specified_type>
-                                          | <fully_specified_type> <identifier>
-                                          | <fully_specified_type> <identifier> "[" "]"
-                                          | <fully_specified_type> <identifier> "["
-                                            <constant_expression> "]"
-                                          | <fully_specified_type> <identifier> "=" <initializer>
-*/
+ * <single_declaration> ::= <fully_specified_type>
+ *                        | <fully_specified_type> <identifier>
+ *                        | <fully_specified_type> <identifier> "[" "]"
+ *                        | <fully_specified_type> <identifier> "[" <constant_expression> "]"
+ *                        | <fully_specified_type> <identifier> "=" <initializer>
+ */
 single_declaration
     single_declaration_nospace .or single_declaration_space;
 single_declaration_space
@@ -869,15 +908,30 @@ single_declaration_6
     constant_expression .emit VARIABLE_ARRAY_EXPLICIT .or .true .emit VARIABLE_ARRAY_UNKNOWN;
 
 /*
-    <fully_specified_type>              ::= <type_specifier>
-                                          | <type_qualifier> <type_specifier>
-                                          | <precision> <type_specifier>
-                                          | <type_qualifier> <precision> <type_specifier>
-*/
+ * <fully_specified_type> ::= <opt_invariant> <opt_centroid> <opt_qualifer> <opt_precision> <type_specifier> <opt_array_size>
+ *
+ * Example: "invariant varying highp vec3 [4]"
+ *
+ * Note that the *_space version doesn't parse the array size, make sure it is being invoked after the *_nospace version.
+ */
 fully_specified_type_space
-    fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_space;
+    fully_specified_type_optinvariant .and fully_specified_type_optcentroid .and fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_space .and .true .emit TYPE_NO_ARRAY_SIZE;
 fully_specified_type_nospace
-    fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_nospace;
+    fully_specified_type_optinvariant .and fully_specified_type_optcentroid .and fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_array;
+type_specifier_array
+    type_specifier_array_1 .or type_specifier_array_2;
+type_specifier_array_1
+    type_specifier_space .and array_size;
+type_specifier_array_2
+    type_specifier_nospace .and opt_array_size;
+fully_specified_type_optinvariant
+    fully_specified_type_invariant .or .true .emit TYPE_VARIANT;
+fully_specified_type_invariant
+    invariant_qualifier .and space;
+fully_specified_type_optcentroid
+    fully_specified_type_centroid .or .true .emit TYPE_CENTER;
+fully_specified_type_centroid
+    centroid_qualifier .and space;
 fully_specified_type_optqual
     fully_specified_type_qual .or .true .emit TYPE_QUALIFIER_NONE;
 fully_specified_type_qual
@@ -886,60 +940,75 @@ fully_specified_type_optprec
     fully_specified_type_prec .or .true .emit PRECISION_DEFAULT;
 fully_specified_type_prec
     precision .and space;
+opt_array_size
+    array_size .or .true .emit TYPE_NO_ARRAY_SIZE;
+array_size
+    lbracket .emit TYPE_ARRAY_SIZE .and constant_expression .and rbracket;
+
 
 /*
-    <type_qualifier>                    ::= "const"
-                                          | "attribute" // Vertex only.
-                                          | "varying"
-                                          | "uniform"
-										  | "__fixed_output"
-										  | "__fixed_input"
+ * <invariant_qualifier> ::= "invariant"
+ */
+invariant_qualifier
+    "invariant" .emit TYPE_INVARIANT;
+
+centroid_qualifier
+    "centroid" .emit TYPE_CENTROID;
 
-note: this is an extension to the standard language specification - normally slang disallows
-      __fixed_output and __fixed_input type qualifiers
-*/
+
+/*
+ * <type_qualifier> ::= "const"
+ *                    | "attribute" // Vertex only.
+ *                    | "varying"
+ *                    | "uniform"
+ *                    | "__fixed_output"
+ *                    | "__fixed_input"
+ *
+ * note: this is an extension to the standard language specification,
+ * normally slang disallows __fixed_output and __fixed_input type qualifiers
+ */
 type_qualifier
     "const" .emit TYPE_QUALIFIER_CONST .or
     .if (shader_type == 2) "attribute" .emit TYPE_QUALIFIER_ATTRIBUTE .or
     "varying" .emit TYPE_QUALIFIER_VARYING .or
     "uniform" .emit TYPE_QUALIFIER_UNIFORM .or
-	.if (parsing_builtin != 0) "__fixed_output" .emit TYPE_QUALIFIER_FIXEDOUTPUT .or
-	.if (parsing_builtin != 0) "__fixed_input" .emit TYPE_QUALIFIER_FIXEDINPUT;
-
-/*
-    <type_specifier>                    ::= "void"
-                                          | "float"
-                                          | "int"
-                                          | "bool"
-                                          | "vec2"
-                                          | "vec3"
-                                          | "vec4"
-                                          | "bvec2"
-                                          | "bvec3"
-                                          | "bvec4"
-                                          | "ivec2"
-                                          | "ivec3"
-                                          | "ivec4"
-                                          | "mat2"
-                                          | "mat3"
-                                          | "mat4"
-                                          | "mat2x3"
-                                          | "mat3x2"
-                                          | "mat2x4"
-                                          | "mat4x2"
-                                          | "mat3x4"
-                                          | "mat4x3"
-                                          | "sampler1D"
-                                          | "sampler2D"
-                                          | "sampler3D"
-                                          | "samplerCube"
-                                          | "sampler1DShadow"
-                                          | "sampler2DShadow"
-					  | "sampler2DRect"
-					  | "sampler2DRectShadow"
-                                          | <struct_specifier>
-                                          | <type_name>
-*/
+    .if (parsing_builtin != 0) "__fixed_output" .emit TYPE_QUALIFIER_FIXEDOUTPUT .or
+    .if (parsing_builtin != 0) "__fixed_input" .emit TYPE_QUALIFIER_FIXEDINPUT;
+
+/*
+ * <type_specifier> ::= "void"
+ *                    | "float"
+ *                    | "int"
+ *                    | "bool"
+ *                    | "vec2"
+ *                    | "vec3"
+ *                    | "vec4"
+ *                    | "bvec2"
+ *                    | "bvec3"
+ *                    | "bvec4"
+ *                    | "ivec2"
+ *                    | "ivec3"
+ *                    | "ivec4"
+ *                    | "mat2"
+ *                    | "mat3"
+ *                    | "mat4"
+ *                    | "mat2x3"
+ *                    | "mat3x2"
+ *                    | "mat2x4"
+ *                    | "mat4x2"
+ *                    | "mat3x4"
+ *                    | "mat4x3"
+ *                    | "sampler1D"
+ *                    | "sampler2D"
+ *                    | "sampler3D"
+ *                    | "samplerCube"
+ *                    | "sampler1DShadow"
+ *                    | "sampler2DShadow"
+ *                    | "sampler2DRect"
+ *                    | "sampler2DRectShadow"
+ *                    | <struct_specifier>
+ *                    | <type_name>
+ */
 type_specifier_space
     "void" .emit TYPE_SPECIFIER_VOID .or
     "float" .emit TYPE_SPECIFIER_FLOAT .or
@@ -976,9 +1045,9 @@ type_specifier_nospace
     struct_specifier .emit TYPE_SPECIFIER_STRUCT;
 
 /*
-    <struct_specifier>                  ::= "struct" <identifier> "{" <struct_declaration_list> "}"
-                                          | "struct" "{" <struct_declaration_list> "}"
-*/
+ * <struct_specifier> ::= "struct" <identifier> "{" <struct_declaration_list> "}"
+ *                      | "struct" "{" <struct_declaration_list> "}"
+ */
 struct_specifier
     "struct" .and struct_specifier_1 .and optional_space .and lbrace .error LBRACE_EXPECTED .and
     struct_declaration_list .and rbrace .emit FIELD_NONE;
@@ -988,15 +1057,15 @@ struct_specifier_2
     space .and identifier;
 
 /*
-    <struct_declaration_list>           ::= <struct_declaration>
-                                          | <struct_declaration_list> <struct_declaration>
-*/
+ * <struct_declaration_list> ::= <struct_declaration>
+ *                             | <struct_declaration_list> <struct_declaration>
+ */
 struct_declaration_list
     struct_declaration .and .loop struct_declaration .emit FIELD_NEXT;
 
 /*
-    <struct_declaration>                ::= <type_specifier> <struct_declarator_list> ";"
-*/
+ * <struct_declaration> ::= <type_specifier> <struct_declarator_list> ";"
+ */
 struct_declaration
     struct_declaration_nospace .or struct_declaration_space;
 struct_declaration_space
@@ -1005,18 +1074,18 @@ struct_declaration_nospace
     type_specifier_nospace .and struct_declarator_list .and semicolon .emit FIELD_NONE;
 
 /*
-    <struct_declarator_list>            ::= <struct_declarator>
-                                          | <struct_declarator_list> "," <struct_declarator>
-*/
+ * <struct_declarator_list> ::= <struct_declarator>
+ *                            | <struct_declarator_list> "," <struct_declarator>
+ */
 struct_declarator_list
     struct_declarator .and .loop struct_declarator_list_1 .emit FIELD_NEXT;
 struct_declarator_list_1
     comma .and struct_declarator;
 
 /*
-    <struct_declarator>                 ::= <identifier>
-                                          | <identifier> "[" <constant_expression> "]"
-*/
+ * <struct_declarator> ::= <identifier>
+ *                       | <identifier> "[" <constant_expression> "]"
+ */
 struct_declarator
     identifier .and struct_declarator_1;
 struct_declarator_1
@@ -1025,21 +1094,21 @@ struct_declarator_2
     lbracket .and constant_expression .and rbracket;
 
 /*
-    <initializer>                       ::= <assignment_expression>
-*/
+ * <initializer> ::= <assignment_expression>
+ */
 initializer
     assignment_expression .and .true .emit OP_END;
 
 /*
-    <declaration_statement>             ::= <declaration>
-*/
+ * <declaration_statement> ::= <declaration>
+ */
 declaration_statement
     declaration;
 
 /*
-    <statement>                         ::= <compound_statement>
-                                          | <simple_statement>
-*/
+ * <statement> ::= <compound_statement>
+ *               | <simple_statement>
+ */
 statement
     compound_statement .or simple_statement;
 statement_space
@@ -1048,16 +1117,16 @@ statement_space_1
     space .and simple_statement;
 
 /*
-    <simple_statement>                  ::= <__asm_statement>
-                                          | <selection_statement>
-                                          | <iteration_statement>
-                                          | <jump_statement>
-                                          | <expression_statement>
-                                          | <declaration_statement>
-
-note: this is an extension to the standard language specification - normally slang disallows
-      use of __asm statements
-*/
+ * <simple_statement> ::= <__asm_statement>
+ *                      | <selection_statement>
+ *                      | <iteration_statement>
+ *                      | <jump_statement>
+ *                      | <expression_statement>
+ *                      | <declaration_statement>
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows use of __asm statements
+ */
 simple_statement
     .if (parsing_builtin != 0) __asm_statement .emit OP_ASM .or
     selection_statement .or
@@ -1068,9 +1137,9 @@ simple_statement
     declaration_statement .emit OP_DECLARE;
 
 /*
-    <compound_statement>                ::= "{" "}"
-                                          | "{" <statement_list> "}"
-*/
+ * <compound_statement> ::= "{" "}"
+ *                        | "{" <statement_list> "}"
+ */
 compound_statement
     compound_statement_1 .emit OP_BLOCK_BEGIN_NEW_SCOPE .and .true .emit OP_END;
 compound_statement_1
@@ -1081,16 +1150,16 @@ compound_statement_3
     lbrace .and statement_list .and rbrace;
 
 /*
-    <statement_no_new_scope>            ::= <compound_statement_no_new_scope>
-                                          | <simple_statement>
-*/
+ * <statement_no_new_scope> ::= <compound_statement_no_new_scope>
+ *                            | <simple_statement>
+ */
 statement_no_new_scope
     compound_statement_no_new_scope .or simple_statement;
 
 /*
-    <compound_statement_no_new_scope>   ::= "{" "}"
-                                          | "{" <statement_list> "}"
-*/
+ * <compound_statement_no_new_scope> ::= "{" "}"
+ *                                     | "{" <statement_list> "}"
+ */
 compound_statement_no_new_scope
     compound_statement_no_new_scope_1 .emit OP_BLOCK_BEGIN_NO_NEW_SCOPE .and .true .emit OP_END;
 compound_statement_no_new_scope_1
@@ -1101,16 +1170,16 @@ compound_statement_no_new_scope_3
     lbrace .and statement_list .and rbrace;
 
 /*
-    <statement_list>                    ::= <statement>
-                                          | <statement_list> <statement>
-*/
+ * <statement_list> ::= <statement>
+ *                    | <statement_list> <statement>
+ */
 statement_list
     statement .and .loop statement;
 
 /*
-    <expression_statement>              ::= ";"
-                                          | <expression> ";"
-*/
+ * <expression_statement> ::= ";"
+ *                          | <expression> ";"
+ */
 expression_statement
     expression_statement_1 .or expression_statement_2;
 expression_statement_1
@@ -1119,16 +1188,16 @@ expression_statement_2
     expression .and semicolon .emit OP_END;
 
 /*
-    <selection_statement>               ::= "if" "(" <expression> ")" <selection_rest_statement>
-*/
+ * <selection_statement> ::= "if" "(" <expression> ")" <selection_rest_statement>
+ */
 selection_statement
     "if" .emit OP_IF .and lparen .error LPAREN_EXPECTED .and expression .and
     rparen .error RPAREN_EXPECTED .emit OP_END .and selection_rest_statement;
 
 /*
-    <selection_rest_statement>          ::= <statement> "else" <statement>
-                                          | <statement>
-*/
+ * <selection_rest_statement> ::= <statement> "else" <statement>
+ *                              | <statement>
+ */
 selection_rest_statement
     statement .and selection_rest_statement_1;
 selection_rest_statement_1
@@ -1137,11 +1206,12 @@ selection_rest_statement_2
     "else" .and optional_space .and statement;
 
 /*
-    <condition>                         ::= <expression>
-                                          | <fully_specified_type> <identifier> "=" <initializer>
-
-note: if <condition_1> is executed, the emit format must match <declaration> emit format
-*/
+ * <condition> ::= <expression>
+ *               | <fully_specified_type> <identifier> "=" <initializer>
+ *
+ * note: if <condition_1> is executed, the emit format must
+ * match <declaration> emit format
+ */
 condition
     condition_1 .emit OP_DECLARE .emit DECLARATION_INIT_DECLARATOR_LIST .or
     condition_3 .emit OP_EXPRESSION;
@@ -1158,11 +1228,11 @@ condition_3
     expression .and .true .emit OP_END;
 
 /*
-    <iteration_statement>               ::= "while" "(" <condition> ")" <statement>
-                                          | "do" <statement> "while" "(" <expression> ")" ";"
-                                          | "for" "(" <for_init_statement> <for_rest_statement> ")"
-                                            <statement_no_new_scope>
-*/
+ * <iteration_statement> ::= "while" "(" <condition> ")" <statement>
+ *                         | "do" <statement> "while" "(" <expression> ")" ";"
+ *                         | "for" "(" <for_init_statement> <for_rest_statement> ")"
+ *                            <statement_no_new_scope>
+ */
 iteration_statement
     iteration_statement_1 .or iteration_statement_2 .or iteration_statement_3;
 iteration_statement_1
@@ -1176,27 +1246,28 @@ iteration_statement_3
     for_rest_statement .and rparen .error RPAREN_EXPECTED .and statement_no_new_scope;
 
 /*
-    <for_init_statement>                ::= <expression_statement>
-                                          | <declaration_statement>
-*/
+ * <for_init_statement> ::= <expression_statement>
+ *                        | <declaration_statement>
+ */
 for_init_statement
     expression_statement .emit OP_EXPRESSION .or declaration_statement .emit OP_DECLARE;
 
 /*
-    <conditionopt>                      ::= <condition>
-                                          | ""
-
-note: <conditionopt> is used only by "for" statement - if <condition> is ommitted, parser
-      simulates default behaviour, that is simulates "true" expression
-*/
+ * <conditionopt> ::= <condition>
+ *                  | ""
+ *
+ * note: <conditionopt> is used only by "for" statement.
+ * if <condition> is ommitted, parser simulates default behaviour,
+ * that is simulates "true" expression
+ */
 conditionopt
     condition .or
     .true .emit OP_EXPRESSION .emit OP_PUSH_BOOL .emit 2 .emit '1' .emit '\0' .emit OP_END;
 
 /*
-    <for_rest_statement>                ::= <conditionopt> ";"
-                                          | <conditionopt> ";" <expression>
-*/
+ * <for_rest_statement> ::= <conditionopt> ";"
+ *                        | <conditionopt> ";" <expression>
+ */
 for_rest_statement
     conditionopt .and semicolon .and for_rest_statement_1;
 for_rest_statement_1
@@ -1205,12 +1276,12 @@ for_rest_statement_2
     expression .and .true .emit OP_END;
 
 /*
-    <jump_statement>                    ::= "continue" ";"
-                                          | "break" ";"
-                                          | "return" ";"
-                                          | "return" <expression> ";"
-                                          | "discard" ";" // Fragment shader only.
-*/
+ * <jump_statement> ::= "continue" ";"
+ *                    | "break" ";"
+ *                    | "return" ";"
+ *                    | "return" <expression> ";"
+ *                    | "discard" ";" // Fragment shader only.
+ */
 jump_statement
     jump_statement_1 .or jump_statement_2 .or jump_statement_3 .or jump_statement_4 .or
     .if (shader_type == 1) jump_statement_5;
@@ -1226,33 +1297,33 @@ jump_statement_5
     "discard" .and semicolon .emit OP_DISCARD;
 
 /*
-    <__asm_statement>                   ::= "__asm" <identifier> <asm_arguments> ";"
-
-note: this is an extension to the standard language specification - normally slang disallows
-      __asm statements
-*/
+ * <__asm_statement> ::= "__asm" <identifier> <asm_arguments> ";"
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows __asm statements
+ */
 __asm_statement
     "__asm" .and space .and identifier .and space .and asm_arguments .and semicolon .emit OP_END;
 
 /*
-    <asm_arguments>                     ::= <asm_argument>
-                                          | <asm_arguments> "," <asm_argument>
-
-note: this is an extension to the standard language specification - normally slang disallows
-      __asm statements
-*/
+ * <asm_arguments> ::= <asm_argument>
+ *                   | <asm_arguments> "," <asm_argument>
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows __asm statements
+ */
 asm_arguments
     asm_argument .and .true .emit OP_END .and .loop asm_arguments_1;
 asm_arguments_1
     comma .and asm_argument .and .true .emit OP_END;
 
 /*
-    <asm_argument>                      ::= <variable_identifier>
-                                          | <floatconstant>
-
-note: this is an extension to the standard language specification - normally slang disallows
-      __asm statements
-*/
+ * <asm_argument> ::= <variable_identifier>
+ *                  | <floatconstant>
+ *
+ * note: this is an extension to the standard language specification.
+ * normally slang disallows __asm statements
+ */
 asm_argument
     var_with_field .or
     variable_identifier .or
@@ -1263,8 +1334,8 @@ var_with_field
 
 
 /*
- *  <translation_unit>                  ::= <external_declaration>
- *                                        | <translation_unit> <external_declaration>
+ * <translation_unit> ::= <external_declaration>
+ *                      | <translation_unit> <external_declaration>
  */
 translation_unit
     optional_space .emit REVISION .and external_declaration .error INVALID_EXTERNAL_DECLARATION .and
@@ -1273,26 +1344,26 @@ translation_unit
 
 
 /*
- *  <external_declaration>              ::= <function_definition>
- *                                        | <declaration>
+ * <external_declaration> ::= <function_definition>
+ *                          | <declaration>
  */
 external_declaration
     precision_stmt .emit DEFAULT_PRECISION .or
-    invariant_stmt .emit INVARIANT_STMT .or
     function_definition .emit EXTERNAL_FUNCTION_DEFINITION .or
+    invariant_stmt .emit INVARIANT_STMT .or
     declaration .emit EXTERNAL_DECLARATION;
 
 
 /*
- * <precision_stmt>    ::= "precision" <precision> <prectype>
+ * <precision_stmt> ::= "precision" <precision> <prectype>
  */
 precision_stmt
     "precision" .and space .and precision .error INVALID_PRECISION .and space .and prectype .error INVALID_PRECISION_TYPE .and semicolon;
 
 /*
- * <precision>           ::= "lowp"
- *                         | "mediump"
- *                         | "highp"
+ * <precision> ::= "lowp"
+ *               | "mediump"
+ *               | "highp"
  */
 precision
     "lowp" .emit PRECISION_LOW .or
@@ -1318,20 +1389,23 @@ prectype
 
 
 /*
- * <invariant_stmt>    ::= "invariant" identifier;
+ * <invariant_stmt> ::= "invariant" identifier;
  */
 invariant_stmt
     "invariant" .and space .and identifier .and semicolon;
 
 
-
 /*
-    <function_definition>               :: <function_prototype> <compound_statement_no_new_scope>
-*/
+ * <function_definition> :: <function_prototype> <compound_statement_no_new_scope>
+ */
 function_definition
     function_prototype .and compound_statement_no_new_scope;
 
-/* helper rulez, not part of the official language syntax */
+
+
+/*
+ * helper rules, not part of the official language syntax
+ */
 
 digit_oct
     '0'-'7';
@@ -1352,11 +1426,13 @@ identifier
     id_character_first .emit * .and .loop id_character_next .emit * .and .true .emit '\0';
 
 float
-    float_1 .or float_2;
+    float_1 .or float_2 .or float_3;
 float_1
-    float_fractional_constant .and float_optional_exponent_part;
+    float_fractional_constant .and float_optional_exponent_part .and optional_f_suffix;
 float_2
-    float_digit_sequence .and .true .emit '\0' .and float_exponent_part;
+    float_digit_sequence .and .true .emit '\0' .and float_exponent_part .and optional_f_suffix;
+float_3
+    float_digit_sequence .and .true .emit '\0' .and 'f' .emit '\0';
 
 float_fractional_constant
     float_fractional_constant_1 .or float_fractional_constant_2 .or float_fractional_constant_3;
@@ -1386,6 +1462,10 @@ float_optional_sign
 float_sign
     '+' .or '-' .emit '-';
 
+optional_f_suffix
+    'f' .or .true;
+
+
 integer
     integer_hex .or integer_oct .or integer_dec;
 
@@ -1469,7 +1549,7 @@ cpp_style_comment_block_3
 cpp_style_comment_char
     '\x0E'-'\xFF' .or '\x01'-'\x09' .or '\x0B'-'\x0C';
 
-/* lexical rulez */
+/* lexical rules */
 
 /*ampersand
     optional_space .and '&' .and optional_space;*/
@@ -1606,7 +1686,7 @@ starequals
 /*tilde
     optional_space .and '~' .and optional_space;*/
 
-/* string rulez - these are used internally by the parser when parsing quoted strings */
+/* string rules - these are used internally by the parser when parsing quoted strings */
 
 .string string_lexer;
 
@@ -1619,7 +1699,7 @@ lex_first_identifier_character
 lex_next_identifier_character
     'a'-'z' .or 'A'-'Z' .or '0'-'9' .or '_';
 
-/* error rulez - these are used by error messages */
+/* error rules - these are used by error messages */
 
 err_token
     '~' .or '`' .or '!' .or '@' .or '#' .or '$' .or '%' .or '^' .or '&' .or '*' .or '(' .or ')' .or
diff --git a/src/mesa/shader/slang/library/slang_shader_syn.h b/src/mesa/shader/slang/library/slang_shader_syn.h
index 6457bb69bd4..42ff92b65a8 100644
--- a/src/mesa/shader/slang/library/slang_shader_syn.h
+++ b/src/mesa/shader/slang/library/slang_shader_syn.h
@@ -47,6 +47,10 @@
 ".emtcode TYPE_QUALIFIER_UNIFORM 4\n"
 ".emtcode TYPE_QUALIFIER_FIXEDOUTPUT 5\n"
 ".emtcode TYPE_QUALIFIER_FIXEDINPUT 6\n"
+".emtcode TYPE_VARIANT 90\n"
+".emtcode TYPE_INVARIANT 91\n"
+".emtcode TYPE_CENTER 95\n"
+".emtcode TYPE_CENTROID 96\n"
 ".emtcode TYPE_SPECIFIER_VOID 0\n"
 ".emtcode TYPE_SPECIFIER_BOOL 1\n"
 ".emtcode TYPE_SPECIFIER_BVEC2 2\n"
@@ -79,6 +83,8 @@
 ".emtcode TYPE_SPECIFIER_MAT42 29\n"
 ".emtcode TYPE_SPECIFIER_MAT34 30\n"
 ".emtcode TYPE_SPECIFIER_MAT43 31\n"
+".emtcode TYPE_ARRAY_SIZE 220\n"
+".emtcode TYPE_NO_ARRAY_SIZE 221\n"
 ".emtcode FIELD_NONE 0\n"
 ".emtcode FIELD_NEXT 1\n"
 ".emtcode FIELD_ARRAY 2\n"
@@ -132,6 +138,7 @@
 ".emtcode OP_POSTINCREMENT 60\n"
 ".emtcode OP_POSTDECREMENT 61\n"
 ".emtcode OP_PRECISION 62\n"
+".emtcode OP_METHOD 63\n"
 ".emtcode PARAM_QUALIFIER_IN 0\n"
 ".emtcode PARAM_QUALIFIER_OUT 1\n"
 ".emtcode PARAM_QUALIFIER_INOUT 2\n"
@@ -169,6 +176,12 @@
 "integer_expression\n"
 " expression;\n"
 "function_call\n"
+" function_call_or_method;\n"
+"function_call_or_method\n"
+" regular_function_call .or method_call;\n"
+"method_call\n"
+" identifier .emit OP_METHOD .and dot .and function_call_generic .and .true .emit OP_END;\n"
+"regular_function_call\n"
 " function_call_generic .emit OP_CALL .and .true .emit OP_END;\n"
 "function_call_generic\n"
 " function_call_generic_1 .or function_call_generic_2;\n"
@@ -421,9 +434,23 @@
 "single_declaration_6\n"
 " constant_expression .emit VARIABLE_ARRAY_EXPLICIT .or .true .emit VARIABLE_ARRAY_UNKNOWN;\n"
 "fully_specified_type_space\n"
-" fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_space;\n"
+" fully_specified_type_optinvariant .and fully_specified_type_optcentroid .and fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_space .and .true .emit TYPE_NO_ARRAY_SIZE;\n"
 "fully_specified_type_nospace\n"
-" fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_nospace;\n"
+" fully_specified_type_optinvariant .and fully_specified_type_optcentroid .and fully_specified_type_optqual .and fully_specified_type_optprec .and type_specifier_array;\n"
+"type_specifier_array\n"
+" type_specifier_array_1 .or type_specifier_array_2;\n"
+"type_specifier_array_1\n"
+" type_specifier_space .and array_size;\n"
+"type_specifier_array_2\n"
+" type_specifier_nospace .and opt_array_size;\n"
+"fully_specified_type_optinvariant\n"
+" fully_specified_type_invariant .or .true .emit TYPE_VARIANT;\n"
+"fully_specified_type_invariant\n"
+" invariant_qualifier .and space;\n"
+"fully_specified_type_optcentroid\n"
+" fully_specified_type_centroid .or .true .emit TYPE_CENTER;\n"
+"fully_specified_type_centroid\n"
+" centroid_qualifier .and space;\n"
 "fully_specified_type_optqual\n"
 " fully_specified_type_qual .or .true .emit TYPE_QUALIFIER_NONE;\n"
 "fully_specified_type_qual\n"
@@ -432,13 +459,21 @@
 " fully_specified_type_prec .or .true .emit PRECISION_DEFAULT;\n"
 "fully_specified_type_prec\n"
 " precision .and space;\n"
+"opt_array_size\n"
+" array_size .or .true .emit TYPE_NO_ARRAY_SIZE;\n"
+"array_size\n"
+" lbracket .emit TYPE_ARRAY_SIZE .and constant_expression .and rbracket;\n"
+"invariant_qualifier\n"
+" \"invariant\" .emit TYPE_INVARIANT;\n"
+"centroid_qualifier\n"
+" \"centroid\" .emit TYPE_CENTROID;\n"
 "type_qualifier\n"
 " \"const\" .emit TYPE_QUALIFIER_CONST .or\n"
 " .if (shader_type == 2) \"attribute\" .emit TYPE_QUALIFIER_ATTRIBUTE .or\n"
 " \"varying\" .emit TYPE_QUALIFIER_VARYING .or\n"
 " \"uniform\" .emit TYPE_QUALIFIER_UNIFORM .or\n"
-"	.if (parsing_builtin != 0) \"__fixed_output\" .emit TYPE_QUALIFIER_FIXEDOUTPUT .or\n"
-"	.if (parsing_builtin != 0) \"__fixed_input\" .emit TYPE_QUALIFIER_FIXEDINPUT;\n"
+" .if (parsing_builtin != 0) \"__fixed_output\" .emit TYPE_QUALIFIER_FIXEDOUTPUT .or\n"
+" .if (parsing_builtin != 0) \"__fixed_input\" .emit TYPE_QUALIFIER_FIXEDINPUT;\n"
 "type_specifier_space\n"
 " \"void\" .emit TYPE_SPECIFIER_VOID .or\n"
 " \"float\" .emit TYPE_SPECIFIER_FLOAT .or\n"
@@ -618,8 +653,8 @@
 " '\\0' .error INVALID_EXTERNAL_DECLARATION .emit EXTERNAL_NULL;\n"
 "external_declaration\n"
 " precision_stmt .emit DEFAULT_PRECISION .or\n"
-" invariant_stmt .emit INVARIANT_STMT .or\n"
 " function_definition .emit EXTERNAL_FUNCTION_DEFINITION .or\n"
+" invariant_stmt .emit INVARIANT_STMT .or\n"
 " declaration .emit EXTERNAL_DECLARATION;\n"
 "precision_stmt\n"
 " \"precision\" .and space .and precision .error INVALID_PRECISION .and space .and prectype .error INVALID_PRECISION_TYPE .and semicolon;\n"
@@ -655,11 +690,13 @@
 "identifier\n"
 " id_character_first .emit * .and .loop id_character_next .emit * .and .true .emit '\\0';\n"
 "float\n"
-" float_1 .or float_2;\n"
+" float_1 .or float_2 .or float_3;\n"
 "float_1\n"
-" float_fractional_constant .and float_optional_exponent_part;\n"
+" float_fractional_constant .and float_optional_exponent_part .and optional_f_suffix;\n"
 "float_2\n"
-" float_digit_sequence .and .true .emit '\\0' .and float_exponent_part;\n"
+" float_digit_sequence .and .true .emit '\\0' .and float_exponent_part .and optional_f_suffix;\n"
+"float_3\n"
+" float_digit_sequence .and .true .emit '\\0' .and 'f' .emit '\\0';\n"
 "float_fractional_constant\n"
 " float_fractional_constant_1 .or float_fractional_constant_2 .or float_fractional_constant_3;\n"
 "float_fractional_constant_1\n"
@@ -682,6 +719,8 @@
 " float_sign .or .true;\n"
 "float_sign\n"
 " '+' .or '-' .emit '-';\n"
+"optional_f_suffix\n"
+" 'f' .or .true;\n"
 "integer\n"
 " integer_hex .or integer_oct .or integer_dec;\n"
 "integer_hex\n"
diff --git a/src/mesa/shader/slang/library/slang_version.syn b/src/mesa/shader/slang/library/slang_version.syn
index aaf8bef342f..aaf8bef342f 100755..100644
--- a/src/mesa/shader/slang/library/slang_version.syn
+++ b/src/mesa/shader/slang/library/slang_version.syn
diff --git a/src/mesa/shader/slang/library/slang_vertex_builtin_gc.h b/src/mesa/shader/slang/library/slang_vertex_builtin_gc.h
index 2c2aecb916f..01980826a95 100644
--- a/src/mesa/shader/slang/library/slang_vertex_builtin_gc.h
+++ b/src/mesa/shader/slang/library/slang_vertex_builtin_gc.h
@@ -2,100 +2,105 @@
 /* DO NOT EDIT - THIS FILE IS AUTOMATICALLY GENERATED FROM THE FOLLOWING FILE: */
 /* slang_vertex_builtin.gc */
 
-4,2,2,5,0,12,1,103,108,95,80,111,115,105,116,105,111,110,0,0,0,2,2,5,0,9,1,103,108,95,80,111,105,
-110,116,83,105,122,101,0,0,0,2,2,5,0,12,1,103,108,95,67,108,105,112,86,101,114,116,101,120,0,0,0,2,
-2,2,0,12,1,103,108,95,67,111,108,111,114,0,0,0,2,2,2,0,12,1,103,108,95,83,101,99,111,110,100,97,
-114,121,67,111,108,111,114,0,0,0,2,2,2,0,11,1,103,108,95,78,111,114,109,97,108,0,0,0,2,2,2,0,12,1,
-103,108,95,86,101,114,116,101,120,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,105,84,101,120,67,
-111,111,114,100,48,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,
-49,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,50,0,0,0,2,2,2,0,
-12,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,51,0,0,0,2,2,2,0,12,1,103,108,95,
-77,117,108,116,105,84,101,120,67,111,111,114,100,52,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,
-105,84,101,120,67,111,111,114,100,53,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,105,84,101,120,
-67,111,111,114,100,54,0,0,0,2,2,2,0,12,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,
-100,55,0,0,0,2,2,2,0,9,1,103,108,95,70,111,103,67,111,111,114,100,0,0,0,2,2,3,0,12,1,103,108,95,70,
-114,111,110,116,67,111,108,111,114,0,0,0,2,2,3,0,12,1,103,108,95,66,97,99,107,67,111,108,111,114,0,
-0,0,2,2,3,0,12,1,103,108,95,70,114,111,110,116,83,101,99,111,110,100,97,114,121,67,111,108,111,114,
-0,0,0,2,2,3,0,12,1,103,108,95,66,97,99,107,83,101,99,111,110,100,97,114,121,67,111,108,111,114,0,0,
-0,2,2,3,0,12,1,103,108,95,84,101,120,67,111,111,114,100,0,3,18,103,108,95,77,97,120,84,101,120,116,
-117,114,101,67,111,111,114,100,115,0,0,0,2,2,3,0,9,1,103,108,95,70,111,103,70,114,97,103,67,111,
-111,114,100,0,0,0,1,0,0,12,0,102,116,114,97,110,115,102,111,114,109,0,0,1,9,18,95,95,114,101,116,
-86,97,108,0,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,
-77,97,116,114,105,120,0,16,8,48,0,57,18,103,108,95,86,101,114,116,101,120,0,59,120,120,120,120,0,
-48,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,
-114,105,120,0,16,10,49,0,57,18,103,108,95,86,101,114,116,101,120,0,59,121,121,121,121,0,48,46,18,
-103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,
-105,120,0,16,10,50,0,57,18,103,108,95,86,101,114,116,101,120,0,59,122,122,122,122,0,48,46,18,103,
-108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,110,77,97,116,114,105,
-120,0,16,10,51,0,57,18,103,108,95,86,101,114,116,101,120,0,59,119,119,119,119,0,48,46,20,0,0,1,0,0,
-12,0,116,101,120,116,117,114,101,49,68,76,111,100,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,
-0,0,9,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,99,111,111,114,100,52,0,0,
-0,9,18,99,111,111,114,100,52,0,59,120,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,
+4,2,2,90,95,5,0,12,221,1,103,108,95,80,111,115,105,116,105,111,110,0,0,0,2,2,90,95,5,0,9,221,1,103,
+108,95,80,111,105,110,116,83,105,122,101,0,0,0,2,2,90,95,5,0,12,221,1,103,108,95,67,108,105,112,86,
+101,114,116,101,120,0,0,0,2,2,90,95,2,0,12,221,1,103,108,95,67,111,108,111,114,0,0,0,2,2,90,95,2,0,
+12,221,1,103,108,95,83,101,99,111,110,100,97,114,121,67,111,108,111,114,0,0,0,2,2,90,95,2,0,11,221,
+1,103,108,95,78,111,114,109,97,108,0,0,0,2,2,90,95,2,0,12,221,1,103,108,95,86,101,114,116,101,120,
+0,0,0,2,2,90,95,2,0,12,221,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,48,0,0,0,
+2,2,90,95,2,0,12,221,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,49,0,0,0,2,2,90,
+95,2,0,12,221,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,50,0,0,0,2,2,90,95,2,0,
+12,221,1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,51,0,0,0,2,2,90,95,2,0,12,221,
+1,103,108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,52,0,0,0,2,2,90,95,2,0,12,221,1,103,
+108,95,77,117,108,116,105,84,101,120,67,111,111,114,100,53,0,0,0,2,2,90,95,2,0,12,221,1,103,108,95,
+77,117,108,116,105,84,101,120,67,111,111,114,100,54,0,0,0,2,2,90,95,2,0,12,221,1,103,108,95,77,117,
+108,116,105,84,101,120,67,111,111,114,100,55,0,0,0,2,2,90,95,2,0,9,221,1,103,108,95,70,111,103,67,
+111,111,114,100,0,0,0,2,2,90,95,3,0,12,221,1,103,108,95,70,114,111,110,116,67,111,108,111,114,0,0,
+0,2,2,90,95,3,0,12,221,1,103,108,95,66,97,99,107,67,111,108,111,114,0,0,0,2,2,90,95,3,0,12,221,1,
+103,108,95,70,114,111,110,116,83,101,99,111,110,100,97,114,121,67,111,108,111,114,0,0,0,2,2,90,95,
+3,0,12,221,1,103,108,95,66,97,99,107,83,101,99,111,110,100,97,114,121,67,111,108,111,114,0,0,0,2,2,
+90,95,3,0,12,221,1,103,108,95,84,101,120,67,111,111,114,100,0,3,18,103,108,95,77,97,120,84,101,120,
+116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,3,0,9,221,1,103,108,95,70,111,103,70,114,97,
+103,67,111,111,114,100,0,0,0,1,90,95,0,0,12,221,0,102,116,114,97,110,115,102,111,114,109,0,0,1,9,
+18,95,95,114,101,116,86,97,108,0,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,
+101,99,116,105,111,110,77,97,116,114,105,120,0,16,8,48,0,57,18,103,108,95,86,101,114,116,101,120,0,
+59,120,120,120,120,0,48,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,
+105,111,110,77,97,116,114,105,120,0,16,10,49,0,57,18,103,108,95,86,101,114,116,101,120,0,59,121,
+121,121,121,0,48,46,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,
+111,110,77,97,116,114,105,120,0,16,10,50,0,57,18,103,108,95,86,101,114,116,101,120,0,59,122,122,
+122,122,0,48,46,18,103,108,95,77,111,100,101,108,86,105,101,119,80,114,111,106,101,99,116,105,111,
+110,77,97,116,114,105,120,0,16,10,51,0,57,18,103,108,95,86,101,114,116,101,120,0,59,119,119,119,
+119,0,48,46,20,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,76,111,100,0,1,1,0,0,16,
+115,97,109,112,108,101,114,0,0,1,1,0,0,9,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,
+90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,0,18,99,111,111,
+114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,
+101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,
+111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,80,114,111,106,76,
+111,100,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,10,99,111,111,114,100,0,0,1,1,0,0,9,
+108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,
+100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,121,0,49,20,0,9,18,112,
+99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,
+95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,
+0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,49,68,80,114,111,106,76,111,100,0,1,1,0,0,16,
+115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,
+90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,0,18,99,111,
+111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,
+119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,
+108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,
+116,101,120,116,117,114,101,50,68,76,111,100,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,
+10,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,
+52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,20,0,9,18,
+99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,
+18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,
+0,0,1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,50,68,80,114,111,106,76,111,100,0,1,1,0,0,17,
+115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,
+90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,
+111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,
+100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,
+116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,
+12,221,0,116,101,120,116,117,114,101,50,68,80,114,111,106,76,111,100,0,1,1,0,0,17,115,97,109,112,
+108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,
+221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,
+100,0,59,120,121,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,
+18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,
+0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,0,116,101,
+120,116,117,114,101,51,68,76,111,100,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,
+111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,
+18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,59,120,121,122,0,20,0,9,18,99,
+111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,51,100,0,18,
+95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,
+1,90,95,0,0,12,221,0,116,101,120,116,117,114,101,51,68,80,114,111,106,76,111,100,0,1,1,0,0,18,115,
+97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,
+95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,122,0,18,99,
+111,111,114,100,0,59,120,121,122,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,
+114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,51,100,0,18,95,95,114,
+101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,
+0,0,12,221,0,116,101,120,116,117,114,101,67,117,98,101,76,111,100,0,1,1,0,0,19,115,97,109,112,108,
+101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,
+99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,
+20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,99,
+117,98,101,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,
+114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,49,68,76,111,100,0,1,1,0,0,20,115,
+97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,
+95,0,0,12,221,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,
+111,111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,
+95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,
+99,111,111,114,100,52,0,0,0,0,1,90,95,0,0,12,221,0,115,104,97,100,111,119,49,68,80,114,111,106,76,
+111,100,0,1,1,0,0,20,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,
+108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,
+100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,
+99,111,111,114,100,0,59,122,0,18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,
 59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,
-97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,0,0,12,0,116,101,
-120,116,117,114,101,49,68,80,114,111,106,76,111,100,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,
-1,0,0,10,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,
-0,0,0,9,18,112,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,
-100,0,59,121,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,
-52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
-18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,49,68,80,114,111,106,76,
-111,100,0,1,1,0,0,16,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,
-108,111,100,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,
-0,18,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,
-114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,49,100,0,18,95,95,114,
-101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,
-0,116,101,120,116,117,114,101,50,68,76,111,100,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,
-10,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,
-18,99,111,111,114,100,52,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,20,0,9,18,99,111,111,
+97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,90,95,0,0,12,221,
+0,115,104,97,100,111,119,50,68,76,111,100,0,1,1,0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,11,
+99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,221,1,99,111,111,114,100,52,
+0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,0,9,18,99,111,111,
 114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,
-114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,0,0,
-12,0,116,101,120,116,117,114,101,50,68,80,114,111,106,76,111,100,0,1,1,0,0,17,115,97,109,112,108,
-101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,112,99,
-111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,
-121,0,18,99,111,111,114,100,0,59,122,0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,
-100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
-109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,
-50,68,80,114,111,106,76,111,100,0,1,1,0,0,17,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,
-114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,
-111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,122,
-0,49,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,
-120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,
-111,114,100,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,51,68,76,111,100,0,1,1,0,0,18,115,97,
-109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,
-1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,
-0,59,120,121,122,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,
-52,95,116,101,120,98,51,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,
-18,99,111,111,114,100,52,0,0,0,0,1,0,0,12,0,116,101,120,116,117,114,101,51,68,80,114,111,106,76,
-111,100,0,1,1,0,0,18,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,
-108,111,100,0,0,0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,
-121,122,0,18,99,111,111,114,100,0,59,120,121,122,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,
-112,99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,51,100,0,
-18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,
-0,0,1,0,0,12,0,116,101,120,116,117,114,101,67,117,98,101,76,111,100,0,1,1,0,0,19,115,97,109,112,
-108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,99,
-111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,
-0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,99,
-117,98,101,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,
-114,100,52,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,49,68,76,111,100,0,1,1,0,0,20,115,97,109,112,
-108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,1,99,
-111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,111,114,100,0,20,
-0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,116,101,120,98,49,
-100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,
-52,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,49,68,80,114,111,106,76,111,100,0,1,1,0,0,20,115,97,
-109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,0,0,12,
-1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,0,18,99,111,111,114,100,0,59,
-120,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,59,122,0,18,99,111,
-111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,4,118,101,
-99,52,95,116,101,120,98,49,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,
-0,0,18,112,99,111,111,114,100,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,50,68,76,111,100,0,1,1,0,0,
-21,115,97,109,112,108,101,114,0,0,1,1,0,0,11,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,
-3,2,0,0,12,1,99,111,111,114,100,52,0,0,0,9,18,99,111,111,114,100,52,0,59,120,121,122,0,18,99,111,
-111,114,100,0,20,0,9,18,99,111,111,114,100,52,0,59,119,0,18,108,111,100,0,20,0,4,118,101,99,52,95,
-116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,
-111,111,114,100,52,0,0,0,0,1,0,0,12,0,115,104,97,100,111,119,50,68,80,114,111,106,76,111,100,0,1,1,
-0,0,21,115,97,109,112,108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,
-0,1,3,2,0,0,12,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,
-111,114,100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,
-59,122,0,18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,
-100,0,20,0,4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,
-109,112,108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,0
+114,101,116,86,97,108,0,0,18,115,97,109,112,108,101,114,0,0,18,99,111,111,114,100,52,0,0,0,0,1,90,
+95,0,0,12,221,0,115,104,97,100,111,119,50,68,80,114,111,106,76,111,100,0,1,1,0,0,21,115,97,109,112,
+108,101,114,0,0,1,1,0,0,12,99,111,111,114,100,0,0,1,1,0,0,9,108,111,100,0,0,0,1,3,2,90,95,0,0,12,
+221,1,112,99,111,111,114,100,0,0,0,9,18,112,99,111,111,114,100,0,59,120,121,0,18,99,111,111,114,
+100,0,59,120,121,0,18,99,111,111,114,100,0,59,119,0,49,20,0,9,18,112,99,111,111,114,100,0,59,122,0,
+18,99,111,111,114,100,0,59,122,0,20,0,9,18,112,99,111,111,114,100,0,59,119,0,18,108,111,100,0,20,0,
+4,118,101,99,52,95,116,101,120,98,50,100,0,18,95,95,114,101,116,86,97,108,0,0,18,115,97,109,112,
+108,101,114,0,0,18,112,99,111,111,114,100,0,0,0,0,0
diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c
index ed6de40d4b6..db00c54b8a2 100644
--- a/src/mesa/shader/slang/slang_builtin.c
+++ b/src/mesa/shader/slang/slang_builtin.c
@@ -80,9 +80,7 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
       { "gl_TextureMatrixTranspose", STATE_TEXTURE_MATRIX, 0 },
       { "gl_TextureMatrixInverseTranspose", STATE_TEXTURE_MATRIX, STATE_MATRIX_INVERSE },
 
-      /* XXX verify these!!! */
       { "gl_NormalMatrix", STATE_MODELVIEW_MATRIX, STATE_MATRIX_TRANSPOSE },
-      { "__NormalMatrixTranspose", STATE_MODELVIEW_MATRIX, 0 },
 
       { NULL, 0, 0 }
    };
@@ -109,10 +107,14 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
    if (isMatrix) {
       if (tokens[0] == STATE_TEXTURE_MATRIX) {
          if (index1 >= 0) {
-            tokens[1] = index1;
+            tokens[1] = index1; /* which texture matrix */
             index1 = 0; /* prevent extra addition at end of function */
          }
       }
+      if (index1 < 0) {
+         /* index1 is unused: prevent extra addition at end of function */
+         index1 = 0;
+      }
    }
    else if (strcmp(var, "gl_DepthRange") == 0) {
       tokens[0] = STATE_DEPTH_RANGE;
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index d83e3b01e6b..8abb642a728 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -238,9 +238,9 @@ _slang_attach_storage(slang_ir_node *n, slang_variable *var)
 
    if (!n->Store) {
       /* need to setup storage */
-      if (n->Var && n->Var->aux) {
+      if (n->Var && n->Var->store) {
          /* node storage info = var storage info */
-         n->Store = (slang_ir_storage *) n->Var->aux;
+         n->Store = n->Var->store;
       }
       else {
          /* alloc new storage info */
@@ -251,8 +251,8 @@ _slang_attach_storage(slang_ir_node *n, slang_variable *var)
                 (void*) n->Store, n->Store->Size);
 #endif
          if (n->Var)
-            n->Var->aux = n->Store;
-         assert(n->Var->aux);
+            n->Var->store = n->Store;
+         assert(n->Var->store);
       }
    }
 }
@@ -413,6 +413,9 @@ static slang_asm_info AsmInfo[] = {
    { "vec4_multiply", IR_MUL, 1, 2 },
    { "vec4_dot", IR_DOT4, 1, 2 },
    { "vec3_dot", IR_DOT3, 1, 2 },
+   { "vec2_dot", IR_DOT2, 1, 2 },
+   { "vec3_nrm", IR_NRM3, 1, 1 },
+   { "vec4_nrm", IR_NRM4, 1, 1 },
    { "vec3_cross", IR_CROSS, 1, 2 },
    { "vec4_lrp", IR_LRP, 1, 3 },
    { "vec4_min", IR_MIN, 1, 2 },
@@ -477,7 +480,6 @@ new_node3(slang_ir_opcode op,
       n->Children[0] = c0;
       n->Children[1] = c1;
       n->Children[2] = c2;
-      n->Writemask = WRITEMASK_XYZW;
       n->InstLocation = -1;
    }
    return n;
@@ -1475,6 +1477,8 @@ _slang_simple_writemask(GLuint writemask, GLuint swizzle)
  * Convert the given swizzle into a writemask.  In some cases this
  * is trivial, in other cases, we'll need to also swizzle the right
  * hand side to put components in the right places.
+ * See comment above for more info.
+ * XXX this function could be simplified and should probably be renamed.
  * \param swizzle  the incoming swizzle
  * \param writemaskOut  returns the writemask
  * \param swizzleOut  swizzle to apply to the right-hand-side
@@ -1599,19 +1603,6 @@ resolve_swizzle(const slang_operation *oper)
 
 
 /**
- * As above, but produce a writemask.
- */
-static GLuint
-resolve_writemask(slang_assemble_ctx *A, const slang_operation *oper)
-{
-   GLuint swizzle = resolve_swizzle(oper);
-   GLuint writemask, swizzleOut;
-   swizzle_to_writemask(A, swizzle, &writemask, &swizzleOut);
-   return writemask;
-}
-
-
-/**
  * Recursively descend through swizzle nodes to find the node's storage info.
  */
 static slang_ir_storage *
@@ -1674,14 +1665,11 @@ _slang_gen_asm(slang_assemble_ctx *A, slang_operation *oper,
       /* Setup n->Store to be a particular location.  Otherwise, storage
        * for the result (a temporary) will be allocated later.
        */
-      GLuint writemask = WRITEMASK_XYZW;
       slang_operation *dest_oper;
       slang_ir_node *n0;
 
       dest_oper = &oper->children[0];
 
-      writemask = resolve_writemask(A, dest_oper);
-
       n0 = _slang_gen_operation(A, dest_oper);
       if (!n0)
          return NULL;
@@ -2032,6 +2020,28 @@ _slang_gen_function_call_name(slang_assemble_ctx *A, const char *name,
                            name);
       return NULL;
    }
+   if (!fun->body) {
+      slang_info_log_error(A->log,
+                           "Function '%s' prototyped but not defined.  "
+                           "Separate compilation units not supported.",
+                           name);
+      return NULL;
+   }
+
+   /* type checking to be sure function's return type matches 'dest' type */
+   if (dest) {
+      slang_typeinfo t0;
+
+      slang_typeinfo_construct(&t0);
+      _slang_typeof_operation(A, dest, &t0);
+
+      if (!slang_type_specifier_equal(&t0.spec, &fun->header.type.specifier)) {
+         slang_info_log_error(A->log,
+                              "Incompatible type returned by call to '%s'",
+                              name);
+         return NULL;
+      }
+   }
 
    n = _slang_gen_function_call(A, fun, oper, dest);
 
@@ -2047,6 +2057,46 @@ _slang_gen_function_call_name(slang_assemble_ctx *A, const char *name,
 }
 
 
+static slang_ir_node *
+_slang_gen_method_call(slang_assemble_ctx *A, slang_operation *oper)
+{
+   slang_atom *a_length = slang_atom_pool_atom(A->atoms, "length");
+   slang_ir_node *n;
+   slang_variable *var;
+
+   /* NOTE: In GLSL 1.20, there's only one kind of method
+    * call: array.length().  Anything else is an error.
+    */
+   if (oper->a_id != a_length) {
+      slang_info_log_error(A->log,
+                           "Undefined method call '%s'", (char *) oper->a_id);
+      return NULL;
+   }
+
+   /* length() takes no arguments */
+   if (oper->num_children > 0) {
+      slang_info_log_error(A->log, "Invalid arguments to length() method");
+      return NULL;
+   }
+
+   /* lookup the object/variable */
+   var = _slang_locate_variable(oper->locals, oper->a_obj, GL_TRUE);
+   if (!var || var->type.specifier.type != SLANG_SPEC_ARRAY) {
+      slang_info_log_error(A->log,
+                           "Undefined object '%s'", (char *) oper->a_obj);
+      return NULL;
+   }
+
+   /* Create a float/literal IR node encoding the array length */
+   n = new_node0(IR_FLOAT);
+   if (n) {
+      n->Value[0] = (float) var->array_len;
+      n->Store = _slang_new_ir_storage(PROGRAM_CONSTANT, -1, 1);
+   }
+   return n;
+}
+
+
 static GLboolean
 _slang_is_constant_cond(const slang_operation *oper, GLboolean *value)
 {
@@ -2442,8 +2492,8 @@ _slang_gen_var_decl(slang_assemble_ctx *A, slang_variable *var)
    n = new_node0(IR_VAR_DECL);
    if (n) {
       _slang_attach_storage(n, var);
-      assert(var->aux);
-      assert(n->Store == var->aux);
+      assert(var->store);
+      assert(n->Store == var->store);
       assert(n->Store);
       assert(n->Store->Index < 0);
 
@@ -3061,7 +3111,6 @@ _slang_gen_assignment(slang_assemble_ctx * A, slang_operation *oper)
             rhs = _slang_gen_swizzle(rhs, newSwizzle);
          }
          n = new_node2(IR_COPY, lhs, rhs);
-         n->Writemask = writemask;
          return n;
       }
       else {
@@ -3129,7 +3178,7 @@ _slang_gen_struct_field(slang_assemble_ctx * A, slang_operation *oper)
       /* oper->a_id is the field name */
       slang_ir_node *base, *n;
       slang_typeinfo field_ti;
-      GLint fieldSize, fieldOffset = -1, swz;
+      GLint fieldSize, fieldOffset = -1;
 
       /* type of field */
       slang_typeinfo_construct(&field_ti);
@@ -3162,22 +3211,12 @@ _slang_gen_struct_field(slang_assemble_ctx * A, slang_operation *oper)
       if (!n)
          return NULL;
 
-
-      /* setup the storage info for this node */
-      swz = fieldOffset % 4;
-
       n->Field = (char *) oper->a_id;
-      n->Store = _slang_new_ir_storage_relative(fieldOffset / 4,
-                                                fieldSize,
-                                                base->Store);
-      if (fieldSize == 1)
-         n->Store->Swizzle = MAKE_SWIZZLE4(swz, swz, swz, swz);
-      else if (fieldSize == 2)
-         n->Store->Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y,
-                                           SWIZZLE_NIL, SWIZZLE_NIL);
-      else if (fieldSize == 3)
-         n->Store->Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y,
-                                           SWIZZLE_Z, SWIZZLE_NIL);
+
+      /* Store the field's offset in storage->Index */
+      n->Store = _slang_new_ir_storage(base->Store->File,
+                                       fieldOffset,
+                                       fieldSize);
 
       return n;
    }
@@ -3218,8 +3257,6 @@ _slang_gen_array_element(slang_assemble_ctx * A, slang_operation *oper)
                                         SWIZZLE_NIL,
                                         SWIZZLE_NIL);
          n = _slang_gen_swizzle(n, swizzle);
-         /*n->Store = _slang_clone_ir_storage_swz(n->Store, */
-         n->Writemask = WRITEMASK_X << index;
       }
       assert(n->Store);
       return n;
@@ -3272,12 +3309,12 @@ _slang_gen_array_element(slang_assemble_ctx * A, slang_operation *oper)
          }
 
          elem = new_node2(IR_ELEMENT, array, index);
-         elem->Store = _slang_new_ir_storage_relative(constIndex,
-                                                      elemSize,
-                                                      array->Store);
 
-         assert(elem->Store->Parent);
-         /* XXX try to do some array bounds checking here */
+         /* The storage info here will be updated during code emit */
+         elem->Store = _slang_new_ir_storage(array->Store->File,
+                                             array->Store->Index,
+                                             elemSize);
+
          return elem;
       }
       else {
@@ -3552,6 +3589,8 @@ _slang_gen_operation(slang_assemble_ctx * A, slang_operation *oper)
    case SLANG_OPER_CALL:
       return _slang_gen_function_call_name(A, (const char *) oper->a_id,
                                            oper, NULL);
+   case SLANG_OPER_METHOD:
+      return _slang_gen_method_call(A, oper);
    case SLANG_OPER_RETURN:
       return _slang_gen_return(A, oper);
    case SLANG_OPER_LABEL:
@@ -3705,11 +3744,6 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
       const GLint totalSize = array_size(size, var->array_len);
       const GLuint swizzle = _slang_var_swizzle(totalSize, 0);
 
-      if (var->initializer) {
-         slang_info_log_error(A->log, "illegal initializer for uniform '%s'", varName);
-         return GL_FALSE;
-      }
-
       if (prog) {
          /* user-defined uniform */
          if (datatype == GL_NONE) {
@@ -3717,7 +3751,7 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
                /* temporary work-around */
                GLenum datatype = GL_FLOAT;
                GLint uniformLoc = _mesa_add_uniform(prog->Parameters, varName,
-                                                    totalSize, datatype);
+                                                    totalSize, datatype, NULL);
                store = _slang_new_ir_storage_swz(PROGRAM_UNIFORM, uniformLoc,
                                                  totalSize, swizzle);
 
@@ -3734,6 +3768,12 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
                 * "f.a"  (GL_FLOAT_VEC3)
                 * "f.b"  (GL_FLOAT_VEC4)
                 */
+
+               if (var->initializer) {
+                  slang_info_log_error(A->log,
+                     "unsupported initializer for uniform '%s'", varName);
+                  return GL_FALSE;
+               }
             }
             else {
                slang_info_log_error(A->log,
@@ -3743,8 +3783,25 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
             }
          }
          else {
-            GLint uniformLoc = _mesa_add_uniform(prog->Parameters, varName,
-                                                 totalSize, datatype);
+            GLint uniformLoc;
+            const GLfloat *initialValues = NULL;
+            if (var->initializer) {
+               _slang_simplify(var->initializer, &A->space, A->atoms);
+               if (var->initializer->type == SLANG_OPER_LITERAL_FLOAT ||
+                   var->initializer->type == SLANG_OPER_LITERAL_INT) {
+                  /* simple float/vector initializer */
+                  initialValues = var->initializer->literal;
+               }
+               else {
+                  /* complex initializer */
+                  slang_info_log_error(A->log,
+                     "unsupported initializer for uniform '%s'", varName);
+                  return GL_FALSE;
+               }
+            }
+
+            uniformLoc = _mesa_add_uniform(prog->Parameters, varName,
+                                           totalSize, datatype, initialValues);
             store = _slang_new_ir_storage_swz(PROGRAM_UNIFORM, uniformLoc,
                                               totalSize, swizzle);
          }
@@ -3779,8 +3836,19 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
 
       if (prog) {
          /* user-defined varying */
-         GLint varyingLoc = _mesa_add_varying(prog->Varying, varName, totalSize);
-         GLuint swizzle = _slang_var_swizzle(size, 0);
+         GLbitfield flags;
+         GLint varyingLoc;
+         GLuint swizzle;
+
+         flags = 0x0;
+         if (var->type.centroid == SLANG_CENTROID)
+            flags |= PROG_PARAM_BIT_CENTROID;
+         if (var->type.variant == SLANG_INVARIANT)
+            flags |= PROG_PARAM_BIT_INVARIANT;
+
+         varyingLoc = _mesa_add_varying(prog->Varying, varName,
+                                        totalSize, flags);
+         swizzle = _slang_var_swizzle(size, 0);
          store = _slang_new_ir_storage_swz(PROGRAM_VARYING, varyingLoc,
                                            totalSize, swizzle);
       }
@@ -3897,7 +3965,7 @@ _slang_codegen_global_variable(slang_assemble_ctx *A, slang_variable *var,
                    store ? store->Index : -2);
 
    if (store)
-      var->aux = store;  /* save var's storage info */
+      var->store = store;  /* save var's storage info */
 
    var->declared = GL_TRUE;
 
diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c
index 294e46235c2..457a6b92c11 100644
--- a/src/mesa/shader/slang/slang_compile.c
+++ b/src/mesa/shader/slang/slang_compile.c
@@ -160,10 +160,24 @@ typedef struct slang_output_ctx_
    struct gl_program *program;
    slang_var_table *vartable;
    GLuint default_precision[TYPE_SPECIFIER_COUNT];
+   GLboolean allow_precision;
+   GLboolean allow_invariant;
+   GLboolean allow_centroid;
+   GLboolean allow_array_types;  /* float[] syntax */
 } slang_output_ctx;
 
 /* _slang_compile() */
 
+
+/* Debugging aid, print file/line where parsing error is detected */
+#define RETURN0 \
+   do { \
+      if (0) \
+         printf("slang error at %s:%d\n", __FILE__, __LINE__); \
+      return 0; \
+   } while (0)
+
+
 static void
 parse_identifier_str(slang_parse_ctx * C, char **id)
 {
@@ -220,7 +234,7 @@ parse_float(slang_parse_ctx * C, float *number)
                                   _mesa_strlen(exponent) + 3) * sizeof(char));
    if (whole == NULL) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
 
    slang_string_copy(whole, integral);
@@ -244,7 +258,7 @@ check_revision(slang_parse_ctx * C)
 {
    if (*C->I != REVISION) {
       slang_info_log_error(C->L, "Internal compiler error.");
-      return 0;
+      RETURN0;
    }
    C->I++;
    return 1;
@@ -331,8 +345,8 @@ static GLboolean
 convert_to_array(slang_parse_ctx * C, slang_variable * var,
                  const slang_type_specifier * sp)
 {
-   /* sized array - mark it as array, copy the specifier to the array element and
-    * parse the expression */
+   /* sized array - mark it as array, copy the specifier to the array element
+    * and parse the expression */
    var->type.specifier.type = SLANG_SPEC_ARRAY;
    var->type.specifier._array = (slang_type_specifier *)
       _slang_alloc(sizeof(slang_type_specifier));
@@ -384,23 +398,23 @@ parse_struct_field(slang_parse_ctx * C, slang_output_ctx * O,
 
    o.structs = st->structs;
    if (!parse_type_specifier(C, &o, sp))
-      return 0;
+      RETURN0;
 
    do {
       slang_atom a_name;
       slang_variable *var = slang_variable_scope_grow(st->fields);
       if (!var) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       a_name = parse_identifier(C);
       if (_slang_locate_variable(st->fields, a_name, GL_FALSE)) {
          slang_info_log_error(C->L, "duplicate field '%s'", (char *) a_name);
-         return 0;
+         RETURN0;
       }
 
       if (!parse_struct_field_var(C, &o, var, a_name, sp))
-         return 0;
+         RETURN0;
    }
    while (*C->I++ != FIELD_NONE);
 
@@ -416,26 +430,26 @@ parse_struct(slang_parse_ctx * C, slang_output_ctx * O, slang_struct ** st)
    /* parse struct name (if any) and make sure it is unique in current scope */
    a_name = parse_identifier(C);
    if (a_name == SLANG_ATOM_NULL)
-      return 0;
+      RETURN0;
 
    name = slang_atom_pool_id(C->atoms, a_name);
    if (name[0] != '\0'
        && slang_struct_scope_find(O->structs, a_name, 0) != NULL) {
       slang_info_log_error(C->L, "%s: duplicate type name.", name);
-      return 0;
+      RETURN0;
    }
 
    /* set-up a new struct */
    *st = (slang_struct *) _slang_alloc(sizeof(slang_struct));
    if (*st == NULL) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
    if (!slang_struct_construct(*st)) {
       _slang_free(*st);
       *st = NULL;
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
    (**st).a_name = a_name;
    (**st).structs->outer_scope = O->structs;
@@ -447,7 +461,7 @@ parse_struct(slang_parse_ctx * C, slang_output_ctx * O, slang_struct ** st)
       slang_type_specifier_ctr(&sp);
       if (!parse_struct_field(C, O, *st, &sp)) {
          slang_type_specifier_dtr(&sp);
-         return 0;
+         RETURN0;
       }
       slang_type_specifier_dtr(&sp);
    }
@@ -465,20 +479,62 @@ parse_struct(slang_parse_ctx * C, slang_output_ctx * O, slang_struct ** st)
                                          * sizeof(slang_struct));
       if (O->structs->structs == NULL) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       s = &O->structs->structs[O->structs->num_structs];
       if (!slang_struct_construct(s))
-         return 0;
+         RETURN0;
       O->structs->num_structs++;
       if (!slang_struct_copy(s, *st))
-         return 0;
+         RETURN0;
    }
 
    return 1;
 }
 
 
+/* invariant qualifer */
+#define TYPE_VARIANT    90
+#define TYPE_INVARIANT  91
+
+static int
+parse_type_variant(slang_parse_ctx * C, slang_type_variant *variant)
+{
+   GLuint invariant = *C->I++;
+   switch (invariant) {
+   case TYPE_VARIANT:
+      *variant = SLANG_VARIANT;
+      return 1;
+   case TYPE_INVARIANT:
+      *variant = SLANG_INVARIANT;
+      return 1;
+   default:
+      RETURN0;
+   }
+}
+
+
+/* centroid qualifer */
+#define TYPE_CENTER    95
+#define TYPE_CENTROID  96
+
+static int
+parse_type_centroid(slang_parse_ctx * C, slang_type_centroid *centroid)
+{
+   GLuint c = *C->I++;
+   switch (c) {
+   case TYPE_CENTER:
+      *centroid = SLANG_CENTER;
+      return 1;
+   case TYPE_CENTROID:
+      *centroid = SLANG_CENTROID;
+      return 1;
+   default:
+      RETURN0;
+   }
+}
+
+
 /* type qualifier */
 #define TYPE_QUALIFIER_NONE 0
 #define TYPE_QUALIFIER_CONST 1
@@ -491,7 +547,8 @@ parse_struct(slang_parse_ctx * C, slang_output_ctx * O, slang_struct ** st)
 static int
 parse_type_qualifier(slang_parse_ctx * C, slang_type_qualifier * qual)
 {
-   switch (*C->I++) {
+   GLuint qualifier = *C->I++;
+   switch (qualifier) {
    case TYPE_QUALIFIER_NONE:
       *qual = SLANG_QUAL_NONE;
       break;
@@ -514,7 +571,7 @@ parse_type_qualifier(slang_parse_ctx * C, slang_type_qualifier * qual)
       *qual = SLANG_QUAL_FIXEDINPUT;
       break;
    default:
-      return 0;
+      RETURN0;
    }
    return 1;
 }
@@ -652,7 +709,7 @@ parse_type_specifier(slang_parse_ctx * C, slang_output_ctx * O,
    case TYPE_SPECIFIER_STRUCT:
       spec->type = SLANG_SPEC_STRUCT;
       if (!parse_struct(C, O, &spec->_struct))
-         return 0;
+         RETURN0;
       break;
    case TYPE_SPECIFIER_TYPENAME:
       spec->type = SLANG_SPEC_STRUCT;
@@ -662,77 +719,163 @@ parse_type_specifier(slang_parse_ctx * C, slang_output_ctx * O,
 
          a_name = parse_identifier(C);
          if (a_name == NULL)
-            return 0;
+            RETURN0;
 
          stru = slang_struct_scope_find(O->structs, a_name, 1);
          if (stru == NULL) {
             slang_info_log_error(C->L, "undeclared type name '%s'",
                                  slang_atom_pool_id(C->atoms, a_name));
-            return 0;
+            RETURN0;
          }
 
          spec->_struct = (slang_struct *) _slang_alloc(sizeof(slang_struct));
          if (spec->_struct == NULL) {
             slang_info_log_memory(C->L);
-            return 0;
+            RETURN0;
          }
          if (!slang_struct_construct(spec->_struct)) {
             _slang_free(spec->_struct);
             spec->_struct = NULL;
-            return 0;
+            RETURN0;
          }
          if (!slang_struct_copy(spec->_struct, stru))
-            return 0;
+            RETURN0;
       }
       break;
    default:
-      return 0;
+      RETURN0;
    }
    return 1;
 }
 
+
 #define PRECISION_DEFAULT 0
 #define PRECISION_LOW     1
 #define PRECISION_MEDIUM  2
 #define PRECISION_HIGH    3
 
 static int
+parse_type_precision(slang_parse_ctx *C,
+                     slang_type_precision *precision)
+{
+   GLint prec = *C->I++;
+   switch (prec) {
+   case PRECISION_DEFAULT:
+      *precision = SLANG_PREC_DEFAULT;
+      return 1;
+   case PRECISION_LOW:
+      *precision = SLANG_PREC_LOW;
+      return 1;
+   case PRECISION_MEDIUM:
+      *precision = SLANG_PREC_MEDIUM;
+      return 1;
+   case PRECISION_HIGH:
+      *precision = SLANG_PREC_HIGH;
+      return 1;
+   default:
+      RETURN0;
+   }
+}
+
+
+#define TYPE_ARRAY_SIZE      220
+#define TYPE_NO_ARRAY_SIZE   221
+
+
+/*
+ * Parse array size (if present) in something like "uniform float [6] var;".
+ */
+static int
+parse_type_array_size(slang_parse_ctx *C, slang_output_ctx * O,
+                      GLint *size)
+{
+   GLint arr = *C->I++;
+   GLuint sz;
+
+   switch (arr) {
+   case TYPE_ARRAY_SIZE:
+      if (!parse_array_len(C, O, &sz))
+         RETURN0;
+      *size = sz;
+      return 1;
+   case TYPE_NO_ARRAY_SIZE:
+      *size = -1; /* -1 = not an array */
+      return 1;
+   default:
+      assert(0);
+      RETURN0;
+   }
+}
+
+static int
 parse_fully_specified_type(slang_parse_ctx * C, slang_output_ctx * O,
                            slang_fully_specified_type * type)
 {
-   GLuint precision;
+   if (!parse_type_variant(C, &type->variant))
+      RETURN0;
+  
+   if (!parse_type_centroid(C, &type->centroid))
+      RETURN0;
 
    if (!parse_type_qualifier(C, &type->qualifier))
-      return 0;
-   precision = *C->I++;
+      RETURN0;
+
+   if (!parse_type_precision(C, &type->precision))
+      RETURN0;
+
    if (!parse_type_specifier(C, O, &type->specifier))
-      return 0;
+      RETURN0;
 
-   switch (precision) {
-   case PRECISION_DEFAULT:
-      assert(type->specifier.type < TYPE_SPECIFIER_COUNT);
-      if (type->specifier.type < TYPE_SPECIFIER_COUNT)
+   if (!parse_type_array_size(C, O, &type->array_len))
+      RETURN0;
+
+   if (!O->allow_invariant && type->variant == SLANG_INVARIANT) {
+      slang_info_log_error(C->L,
+         "'invariant' keyword not allowed (perhaps set #version 120)");
+      RETURN0;
+   }
+
+   if (!O->allow_centroid && type->centroid == SLANG_CENTROID) {
+      slang_info_log_error(C->L,
+         "'centroid' keyword not allowed (perhaps set #version 120)");
+      RETURN0;
+   }
+   else if (type->centroid == SLANG_CENTROID &&
+            type->qualifier != SLANG_QUAL_VARYING) {
+      slang_info_log_error(C->L,
+         "'centroid' keyword only allowed for varying vars");
+      RETURN0;
+   }
+
+
+   /* need this?
+   if (type->qualifier != SLANG_QUAL_VARYING &&
+       type->variant == SLANG_INVARIANT) {
+      slang_info_log_error(C->L,
+                           "invariant qualifer only allowed for varying vars");
+      RETURN0;
+   }
+   */
+
+   if (O->allow_precision) {
+      if (type->precision == SLANG_PREC_DEFAULT) {
+         assert(type->specifier.type < TYPE_SPECIFIER_COUNT);
+         /* use the default precision for this datatype */
          type->precision = O->default_precision[type->specifier.type];
-      break;
-   case PRECISION_LOW:
-      type->precision = SLANG_PREC_LOW;
-      break;
-   case PRECISION_MEDIUM:
-      type->precision = SLANG_PREC_MEDIUM;
-      break;
-   case PRECISION_HIGH:
-      type->precision = SLANG_PREC_HIGH;
-      break;
-   default:
-      return 0;
+      }
+   }
+   else {
+      /* only default is allowed */
+      if (type->precision != SLANG_PREC_DEFAULT) {
+         slang_info_log_error(C->L, "precision qualifiers not allowed");
+         RETURN0;
+      }
    }
 
-#if !FEATURE_es2_glsl
-   if (precision != PRECISION_DEFAULT) {
-      slang_info_log_error(C->L, "precision qualifiers not allowed");
-      return 0;
+   if (!O->allow_array_types && type->array_len >= 0) {
+      slang_info_log_error(C->L, "first-class array types not allowed");
+      RETURN0;
    }
-#endif
 
    return 1;
 }
@@ -801,6 +944,7 @@ parse_fully_specified_type(slang_parse_ctx * C, slang_output_ctx * O,
 #define OP_POSTINCREMENT 60
 #define OP_POSTDECREMENT 61
 #define OP_PRECISION 62
+#define OP_METHOD 63
 
 
 /**
@@ -841,7 +985,7 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
       oper->type = SLANG_OPER_BLOCK_NO_NEW_SCOPE;
       while (*C->I != OP_END)
          if (!parse_child_operation(C, O, oper, 1))
-            return 0;
+            RETURN0;
       C->I++;
       break;
    case OP_BLOCK_BEGIN_NEW_SCOPE:
@@ -853,7 +997,7 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
          o.vars = oper->locals;
          while (*C->I != OP_END)
             if (!parse_child_operation(C, &o, oper, 1))
-               return 0;
+               RETURN0;
          C->I++;
       }
       break;
@@ -869,7 +1013,7 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
           * than one declarators
           */
          if (!parse_declaration(C, O))
-            return 0;
+            RETURN0;
          if (first_var < O->vars->num_variables) {
             const unsigned int num_vars = O->vars->num_variables - first_var;
             unsigned int i;
@@ -878,18 +1022,23 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
             oper->children = slang_operation_new(num_vars);
             if (oper->children == NULL) {
                slang_info_log_memory(C->L);
-               return 0;
+               RETURN0;
             }
             for (i = first_var; i < O->vars->num_variables; i++) {
                slang_operation *o = &oper->children[i - first_var];
+               slang_variable *var = O->vars->variables[i];
                o->type = SLANG_OPER_VARIABLE_DECL;
                o->locals->outer_scope = O->vars;
-               o->a_id = O->vars->variables[i]->a_name;
+               o->a_id = var->a_name;
+
+               /* new/someday...
+               calculate_var_size(C, O, var);
+               */
 
                if (!legal_identifier(o->a_id)) {
                   slang_info_log_error(C->L, "illegal variable name '%s'",
                                        (char *) o->a_id);
-                  return 0;
+                  RETURN0;
                }
             }
          }
@@ -902,10 +1051,10 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
       oper->type = SLANG_OPER_ASM;
       oper->a_id = parse_identifier(C);
       if (oper->a_id == SLANG_ATOM_NULL)
-         return 0;
+         RETURN0;
       while (*C->I != OP_END) {
          if (!parse_child_operation(C, O, oper, 0))
-            return 0;
+            RETURN0;
       }
       C->I++;
       break;
@@ -921,21 +1070,21 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
    case OP_RETURN:
       oper->type = SLANG_OPER_RETURN;
       if (!parse_child_operation(C, O, oper, 0))
-         return 0;
+         RETURN0;
       break;
    case OP_EXPRESSION:
       oper->type = SLANG_OPER_EXPRESSION;
       if (!parse_child_operation(C, O, oper, 0))
-         return 0;
+         RETURN0;
       break;
    case OP_IF:
       oper->type = SLANG_OPER_IF;
       if (!parse_child_operation(C, O, oper, 0))
-         return 0;
+         RETURN0;
       if (!parse_child_operation(C, O, oper, 1))
-         return 0;
+         RETURN0;
       if (!parse_child_operation(C, O, oper, 1))
-         return 0;
+         RETURN0;
       break;
    case OP_WHILE:
       {
@@ -944,17 +1093,17 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
          oper->type = SLANG_OPER_WHILE;
          o.vars = oper->locals;
          if (!parse_child_operation(C, &o, oper, 1))
-            return 0;
+            RETURN0;
          if (!parse_child_operation(C, &o, oper, 1))
-            return 0;
+            RETURN0;
       }
       break;
    case OP_DO:
       oper->type = SLANG_OPER_DO;
       if (!parse_child_operation(C, O, oper, 1))
-         return 0;
+         RETURN0;
       if (!parse_child_operation(C, O, oper, 0))
-         return 0;
+         RETURN0;
       break;
    case OP_FOR:
       {
@@ -963,13 +1112,13 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
          oper->type = SLANG_OPER_FOR;
          o.vars = oper->locals;
          if (!parse_child_operation(C, &o, oper, 1))
-            return 0;
+            RETURN0;
          if (!parse_child_operation(C, &o, oper, 1))
-            return 0;
+            RETURN0;
          if (!parse_child_operation(C, &o, oper, 0))
-            return 0;
+            RETURN0;
          if (!parse_child_operation(C, &o, oper, 1))
-            return 0;
+            RETURN0;
       }
       break;
    case OP_PRECISION:
@@ -983,7 +1132,7 @@ parse_statement(slang_parse_ctx * C, slang_output_ctx * O,
       }
       break;
    default:
-      return 0;
+      RETURN0;
    }
    return 1;
 }
@@ -998,7 +1147,7 @@ handle_nary_expression(slang_parse_ctx * C, slang_operation * op,
    op->children = slang_operation_new(n);
    if (op->children == NULL) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
    op->num_children = n;
 
@@ -1016,7 +1165,7 @@ handle_nary_expression(slang_parse_ctx * C, slang_operation * op,
                      *total_ops * sizeof(slang_operation));
    if (*ops == NULL) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
    return 1;
 }
@@ -1049,12 +1198,12 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
                         (num_ops + 1) * sizeof(slang_operation));
       if (ops == NULL) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       op = &ops[num_ops];
       if (!slang_operation_construct(op)) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       num_ops++;
       op->locals->outer_scope = O->vars;
@@ -1066,7 +1215,7 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
       case OP_PUSH_BOOL:
          op->type = SLANG_OPER_LITERAL_BOOL;
          if (!parse_number(C, &number))
-            return 0;
+            RETURN0;
          op->literal[0] =
          op->literal[1] =
          op->literal[2] =
@@ -1076,7 +1225,7 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
       case OP_PUSH_INT:
          op->type = SLANG_OPER_LITERAL_INT;
          if (!parse_number(C, &number))
-            return 0;
+            RETURN0;
          op->literal[0] =
          op->literal[1] =
          op->literal[2] =
@@ -1086,7 +1235,7 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
       case OP_PUSH_FLOAT:
          op->type = SLANG_OPER_LITERAL_FLOAT;
          if (!parse_float(C, &op->literal[0]))
-            return 0;
+            RETURN0;
          op->literal[1] =
          op->literal[2] =
          op->literal[3] = op->literal[0];
@@ -1096,37 +1245,37 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
          op->type = SLANG_OPER_IDENTIFIER;
          op->a_id = parse_identifier(C);
          if (op->a_id == SLANG_ATOM_NULL)
-            return 0;
+            RETURN0;
          break;
       case OP_SEQUENCE:
          op->type = SLANG_OPER_SEQUENCE;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_ASSIGN:
          op->type = SLANG_OPER_ASSIGN;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_ADDASSIGN:
          op->type = SLANG_OPER_ADDASSIGN;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_SUBASSIGN:
          op->type = SLANG_OPER_SUBASSIGN;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_MULASSIGN:
          op->type = SLANG_OPER_MULASSIGN;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_DIVASSIGN:
          op->type = SLANG_OPER_DIVASSIGN;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
          /*case OP_MODASSIGN: */
          /*case OP_LSHASSIGN: */
@@ -1137,22 +1286,22 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
       case OP_SELECT:
          op->type = SLANG_OPER_SELECT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 3))
-            return 0;
+            RETURN0;
          break;
       case OP_LOGICALOR:
          op->type = SLANG_OPER_LOGICALOR;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_LOGICALXOR:
          op->type = SLANG_OPER_LOGICALXOR;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_LOGICALAND:
          op->type = SLANG_OPER_LOGICALAND;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
          /*case OP_BITOR: */
          /*case OP_BITXOR: */
@@ -1160,95 +1309,123 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
       case OP_EQUAL:
          op->type = SLANG_OPER_EQUAL;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_NOTEQUAL:
          op->type = SLANG_OPER_NOTEQUAL;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_LESS:
          op->type = SLANG_OPER_LESS;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_GREATER:
          op->type = SLANG_OPER_GREATER;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_LESSEQUAL:
          op->type = SLANG_OPER_LESSEQUAL;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_GREATEREQUAL:
          op->type = SLANG_OPER_GREATEREQUAL;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
          /*case OP_LSHIFT: */
          /*case OP_RSHIFT: */
       case OP_ADD:
          op->type = SLANG_OPER_ADD;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_SUBTRACT:
          op->type = SLANG_OPER_SUBTRACT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_MULTIPLY:
          op->type = SLANG_OPER_MULTIPLY;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
       case OP_DIVIDE:
          op->type = SLANG_OPER_DIVIDE;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
          break;
          /*case OP_MODULUS: */
       case OP_PREINCREMENT:
          op->type = SLANG_OPER_PREINCREMENT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_PREDECREMENT:
          op->type = SLANG_OPER_PREDECREMENT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_PLUS:
          op->type = SLANG_OPER_PLUS;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_MINUS:
          op->type = SLANG_OPER_MINUS;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_NOT:
          op->type = SLANG_OPER_NOT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
          /*case OP_COMPLEMENT: */
       case OP_SUBSCRIPT:
          op->type = SLANG_OPER_SUBSCRIPT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 2))
-            return 0;
+            RETURN0;
+         break;
+      case OP_METHOD:
+         op->type = SLANG_OPER_METHOD;
+         op->a_obj = parse_identifier(C);
+         if (op->a_obj == SLANG_ATOM_NULL)
+            RETURN0;
+
+         op->a_id = parse_identifier(C);
+         if (op->a_id == SLANG_ATOM_NULL)
+            RETURN0;
+
+         while (*C->I != OP_END)
+            if (!parse_child_operation(C, O, op, 0))
+               RETURN0;
+         C->I++;
+#if 0
+         /* don't lookup the method (not yet anyway) */
+         if (!C->parsing_builtin
+             && !slang_function_scope_find_by_name(O->funs, op->a_id, 1)) {
+            const char *id;
+
+            id = slang_atom_pool_id(C->atoms, op->a_id);
+            if (!is_constructor_name(id, op->a_id, O->structs)) {
+               slang_info_log_error(C->L, "%s: undeclared function name.", id);
+               RETURN0;
+            }
+         }
+#endif
          break;
       case OP_CALL:
          op->type = SLANG_OPER_CALL;
          op->a_id = parse_identifier(C);
          if (op->a_id == SLANG_ATOM_NULL)
-            return 0;
+            RETURN0;
          while (*C->I != OP_END)
             if (!parse_child_operation(C, O, op, 0))
-               return 0;
+               RETURN0;
          C->I++;
 
          if (!C->parsing_builtin
@@ -1258,7 +1435,7 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
             id = slang_atom_pool_id(C->atoms, op->a_id);
             if (!is_constructor_name(id, op->a_id, O->structs)) {
                slang_info_log_error(C->L, "%s: undeclared function name.", id);
-               return 0;
+               RETURN0;
             }
          }
          break;
@@ -1266,22 +1443,22 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
          op->type = SLANG_OPER_FIELD;
          op->a_id = parse_identifier(C);
          if (op->a_id == SLANG_ATOM_NULL)
-            return 0;
+            RETURN0;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_POSTINCREMENT:
          op->type = SLANG_OPER_POSTINCREMENT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       case OP_POSTDECREMENT:
          op->type = SLANG_OPER_POSTDECREMENT;
          if (!handle_nary_expression(C, op, &ops, &num_ops, 1))
-            return 0;
+            RETURN0;
          break;
       default:
-         return 0;
+         RETURN0;
       }
    }
    C->I++;
@@ -1312,7 +1489,7 @@ parse_parameter_declaration(slang_parse_ctx * C, slang_output_ctx * O,
     * two at most) because not all combinations are valid
     */
    if (!parse_type_qualifier(C, &param->type.qualifier))
-      return 0;
+      RETURN0;
 
    param_qual = *C->I++;
    switch (param_qual) {
@@ -1320,7 +1497,7 @@ parse_parameter_declaration(slang_parse_ctx * C, slang_output_ctx * O,
       if (param->type.qualifier != SLANG_QUAL_CONST
           && param->type.qualifier != SLANG_QUAL_NONE) {
          slang_info_log_error(C->L, "Invalid type qualifier.");
-         return 0;
+         RETURN0;
       }
       break;
    case PARAM_QUALIFIER_OUT:
@@ -1328,7 +1505,7 @@ parse_parameter_declaration(slang_parse_ctx * C, slang_output_ctx * O,
          param->type.qualifier = SLANG_QUAL_OUT;
       else {
          slang_info_log_error(C->L, "Invalid type qualifier.");
-         return 0;
+         RETURN0;
       }
       break;
    case PARAM_QUALIFIER_INOUT:
@@ -1336,11 +1513,11 @@ parse_parameter_declaration(slang_parse_ctx * C, slang_output_ctx * O,
          param->type.qualifier = SLANG_QUAL_INOUT;
       else {
          slang_info_log_error(C->L, "Invalid type qualifier.");
-         return 0;
+         RETURN0;
       }
       break;
    default:
-      return 0;
+      RETURN0;
    }
 
    /* parse precision qualifier (lowp, mediump, highp */
@@ -1350,10 +1527,10 @@ parse_parameter_declaration(slang_parse_ctx * C, slang_output_ctx * O,
 
    /* parse parameter's type specifier and name */
    if (!parse_type_specifier(C, O, &param->type.specifier))
-      return 0;
+      RETURN0;
    param->a_name = parse_identifier(C);
    if (param->a_name == SLANG_ATOM_NULL)
-      return 0;
+      RETURN0;
 
    /* if the parameter is an array, parse its size (the size must be
     * explicitly defined
@@ -1470,13 +1647,13 @@ parse_operator_name(slang_parse_ctx * C)
             slang_atom_pool_atom(C->atoms, operator_names[i].o_name);
          if (atom == SLANG_ATOM_NULL) {
             slang_info_log_memory(C->L);
-            return 0;
+            RETURN0;
          }
          C->I++;
          return atom;
       }
    }
-   return 0;
+   RETURN0;
 }
 
 
@@ -1484,43 +1661,46 @@ static int
 parse_function_prototype(slang_parse_ctx * C, slang_output_ctx * O,
                          slang_function * func)
 {
+   GLuint functype;
    /* parse function type and name */
    if (!parse_fully_specified_type(C, O, &func->header.type))
-      return 0;
-   switch (*C->I++) {
+      RETURN0;
+
+   functype = *C->I++;
+   switch (functype) {
    case FUNCTION_ORDINARY:
       func->kind = SLANG_FUNC_ORDINARY;
       func->header.a_name = parse_identifier(C);
       if (func->header.a_name == SLANG_ATOM_NULL)
-         return 0;
+         RETURN0;
       break;
    case FUNCTION_CONSTRUCTOR:
       func->kind = SLANG_FUNC_CONSTRUCTOR;
       if (func->header.type.specifier.type == SLANG_SPEC_STRUCT)
-         return 0;
+         RETURN0;
       func->header.a_name =
          slang_atom_pool_atom(C->atoms,
                               slang_type_specifier_type_to_string
                               (func->header.type.specifier.type));
       if (func->header.a_name == SLANG_ATOM_NULL) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       break;
    case FUNCTION_OPERATOR:
       func->kind = SLANG_FUNC_OPERATOR;
       func->header.a_name = parse_operator_name(C);
       if (func->header.a_name == SLANG_ATOM_NULL)
-         return 0;
+         RETURN0;
       break;
    default:
-      return 0;
+      RETURN0;
    }
 
    if (!legal_identifier(func->header.a_name)) {
       slang_info_log_error(C->L, "illegal function name '%s'",
                            (char *) func->header.a_name);
-      return 0;
+      RETURN0;
    }
 
    /* parse function parameters */
@@ -1528,10 +1708,10 @@ parse_function_prototype(slang_parse_ctx * C, slang_output_ctx * O,
       slang_variable *p = slang_variable_scope_grow(func->parameters);
       if (!p) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       if (!parse_parameter_declaration(C, O, p))
-         return 0;
+         RETURN0;
    }
 
    /* if the function returns a value, append a hidden __retVal 'out'
@@ -1565,19 +1745,19 @@ parse_function_definition(slang_parse_ctx * C, slang_output_ctx * O,
    slang_output_ctx o = *O;
 
    if (!parse_function_prototype(C, O, func))
-      return 0;
+      RETURN0;
 
    /* create function's body operation */
    func->body = (slang_operation *) _slang_alloc(sizeof(slang_operation));
    if (func->body == NULL) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
    if (!slang_operation_construct(func->body)) {
       _slang_free(func->body);
       func->body = NULL;
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
 
    /* to parse the body the parse context is modified in order to
@@ -1586,7 +1766,7 @@ parse_function_definition(slang_parse_ctx * C, slang_output_ctx * O,
    C->global_scope = GL_FALSE;
    o.vars = func->parameters;
    if (!parse_statement(C, &o, func->body))
-      return 0;
+      RETURN0;
 
    C->global_scope = GL_TRUE;
    return 1;
@@ -1682,46 +1862,49 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O,
       slang_info_log_error(C->L,
                    "declaration of '%s' conflicts with previous declaration",
                    (char *) a_name);
-      return 0;
+      RETURN0;
    }
 
    /* make room for the new variable and initialize it */
    var = slang_variable_scope_grow(O->vars);
    if (!var) {
       slang_info_log_memory(C->L);
-      return 0;
+      RETURN0;
    }
 
-   /* copy the declarator qualifier type, parse the identifier */
+   /* copy the declarator type qualifier/etc info, parse the identifier */
    var->type.qualifier = type->qualifier;
+   var->type.centroid = type->centroid;
+   var->type.precision = type->precision;
+   var->type.variant = type->variant;
    var->a_name = a_name;
    if (var->a_name == SLANG_ATOM_NULL)
-      return 0;
+      RETURN0;
 
    switch (*C->I++) {
    case VARIABLE_NONE:
       /* simple variable declarator - just copy the specifier */
       if (!slang_type_specifier_copy(&var->type.specifier, &type->specifier))
-         return 0;
+         RETURN0;
       break;
    case VARIABLE_INITIALIZER:
       /* initialized variable - copy the specifier and parse the expression */
       if (!slang_type_specifier_copy(&var->type.specifier, &type->specifier))
-         return 0;
+         RETURN0;
       var->initializer =
          (slang_operation *) _slang_alloc(sizeof(slang_operation));
       if (var->initializer == NULL) {
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       if (!slang_operation_construct(var->initializer)) {
          _slang_free(var->initializer);
          var->initializer = NULL;
          slang_info_log_memory(C->L);
-         return 0;
+         RETURN0;
       }
       if (!parse_expression(C, O, var->initializer))
-         return 0;
+         RETURN0;
       break;
    case VARIABLE_ARRAY_UNKNOWN:
       /* unsized array - mark it as array and copy the specifier to
@@ -1737,7 +1920,17 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O,
          return GL_FALSE;
       break;
    default:
-      return 0;
+      RETURN0;
+   }
+
+   if (type->array_len >= 0) {
+      /* The type was something like "float[4]" */
+      if (var->array_len != 0) {
+         slang_info_log_error(C->L, "multi-dimensional arrays not allowed");
+         RETURN0;
+      }
+      convert_to_array(C, var, &type->specifier);
+      var->array_len = type->array_len;
    }
 
    /* allocate global address space for a variable with a known size */
@@ -1761,7 +1954,7 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O,
       A.log = C->L;
       A.curFuncEndLabel = NULL;
       if (!_slang_codegen_global_variable(&A, var, C->type))
-         return 0;
+         RETURN0;
    }
 
    /* initialize global variable */
@@ -1774,7 +1967,7 @@ parse_init_declarator(slang_parse_ctx * C, slang_output_ctx * O,
          A.space.structs = O->structs;
          A.space.vars = O->vars;
          if (!initialize_global(&A, var))
-            return 0;
+            RETURN0;
       }
    }
    return 1;
@@ -1791,17 +1984,17 @@ parse_init_declarator_list(slang_parse_ctx * C, slang_output_ctx * O)
 
    /* parse the fully specified type, common to all declarators */
    if (!slang_fully_specified_type_construct(&type))
-      return 0;
+      RETURN0;
    if (!parse_fully_specified_type(C, O, &type)) {
       slang_fully_specified_type_destruct(&type);
-      return 0;
+      RETURN0;
    }
 
    /* parse declarators, pass-in the parsed type */
    do {
       if (!parse_init_declarator(C, O, &type)) {
          slang_fully_specified_type_destruct(&type);
-         return 0;
+         RETURN0;
       }
    }
    while (*C->I++ == DECLARATOR_NEXT);
@@ -1908,18 +2101,18 @@ parse_declaration(slang_parse_ctx * C, slang_output_ctx * O)
    switch (*C->I++) {
    case DECLARATION_INIT_DECLARATOR_LIST:
       if (!parse_init_declarator_list(C, O))
-         return 0;
+         RETURN0;
       break;
    case DECLARATION_FUNCTION_PROTOTYPE:
       {
          slang_function *dummy_func;
 
          if (!parse_function(C, O, 0, &dummy_func))
-            return 0;
+            RETURN0;
       }
       break;
    default:
-      return 0;
+      RETURN0;
    }
    return 1;
 }
@@ -1927,9 +2120,13 @@ parse_declaration(slang_parse_ctx * C, slang_output_ctx * O)
 static int
 parse_default_precision(slang_parse_ctx * C, slang_output_ctx * O)
 {
-#if FEATURE_es2_glsl
    int precision, type;
 
+   if (!O->allow_precision) {
+      slang_info_log_error(C->L, "syntax error at \"precision\"");
+      RETURN0;
+   }
+
    precision = *C->I++;
    switch (precision) {
    case PRECISION_LOW:
@@ -1940,7 +2137,7 @@ parse_default_precision(slang_parse_ctx * C, slang_output_ctx * O)
    default:
       _mesa_problem(NULL, "unexpected precision %d at %s:%d\n",
                     precision, __FILE__, __LINE__);
-      return 0;
+      RETURN0;
    }
 
    type = *C->I++;
@@ -1960,17 +2157,13 @@ parse_default_precision(slang_parse_ctx * C, slang_output_ctx * O)
    default:
       _mesa_problem(NULL, "unexpected type %d at %s:%d\n",
                     type, __FILE__, __LINE__);
-      return 0;
+      RETURN0;
    }
 
    assert(type < TYPE_SPECIFIER_COUNT);
    O->default_precision[type] = precision;
 
    return 1;
-#else
-   slang_info_log_error(C->L, "syntax error at \"precision\"");
-   return 0;
-#endif
 }
 
 
@@ -1989,7 +2182,7 @@ init_default_precision(slang_output_ctx *O, slang_unit_type type)
       O->default_precision[i] = PRECISION_HIGH;
 #endif
    }
-#if FEATURE_es2_glsl
+
    if (type == SLANG_UNIT_VERTEX_SHADER) {
       O->default_precision[TYPE_SPECIFIER_FLOAT] = PRECISION_HIGH;
       O->default_precision[TYPE_SPECIFIER_INT] = PRECISION_HIGH;
@@ -1997,14 +2190,13 @@ init_default_precision(slang_output_ctx *O, slang_unit_type type)
    else {
       O->default_precision[TYPE_SPECIFIER_INT] = PRECISION_MEDIUM;
    }
-#endif
 }
 
 
 static int
 parse_invariant(slang_parse_ctx * C, slang_output_ctx * O)
 {
-   if (C->version >= 120 || FEATURE_es2_glsl) {
+   if (O->allow_invariant) {
       slang_atom *a = parse_identifier(C);
       /* XXX not doing anything with this var yet */
       /*printf("ID: %s\n", (char*) a);*/
@@ -2012,7 +2204,7 @@ parse_invariant(slang_parse_ctx * C, slang_output_ctx * O)
    }
    else {
       slang_info_log_error(C->L, "syntax error at \"invariant\"");
-      return 0;
+      RETURN0;
    }
 }
       
@@ -2046,7 +2238,6 @@ parse_code_unit(slang_parse_ctx * C, slang_code_unit * unit,
    }
 
    /* setup output context */
-   init_default_precision(&o, unit->type);
    o.funs = &unit->funs;
    o.structs = &unit->structs;
    o.vars = &unit->vars;
@@ -2055,6 +2246,27 @@ parse_code_unit(slang_parse_ctx * C, slang_code_unit * unit,
    o.vartable = _slang_new_var_table(maxRegs);
    _slang_push_var_table(o.vartable);
 
+   /* allow 'invariant' keyword? */
+#if FEATURE_es2_glsl
+   o.allow_invariant = GL_TRUE;
+#else
+   o.allow_invariant = (C->version >= 120) ? GL_TRUE : GL_FALSE;
+#endif
+
+   /* allow 'centroid' keyword? */
+   o.allow_centroid = (C->version >= 120) ? GL_TRUE : GL_FALSE;
+
+   /* allow 'lowp/mediump/highp' keywords? */
+#if FEATURE_es2_glsl
+   o.allow_precision = GL_TRUE;
+#else
+   o.allow_precision = (C->version >= 120) ? GL_TRUE : GL_FALSE;
+#endif
+   init_default_precision(&o, unit->type);
+
+   /* allow 'float[]' keyword? */
+   o.allow_array_types = (C->version >= 120) ? GL_TRUE : GL_FALSE;
+
    /* parse individual functions and declarations */
    while (*C->I != EXTERNAL_NULL) {
       switch (*C->I++) {
diff --git a/src/mesa/shader/slang/slang_compile_operation.h b/src/mesa/shader/slang/slang_compile_operation.h
index 4f92aa9a086..ec99338cb88 100644
--- a/src/mesa/shader/slang/slang_compile_operation.h
+++ b/src/mesa/shader/slang/slang_compile_operation.h
@@ -94,6 +94,7 @@ typedef enum slang_operation_type_
    SLANG_OPER_SUBSCRIPT,        /* [expr] "[" [expr] "]" */
    SLANG_OPER_CALL,             /* [func name] [param] [param] [...] */
    SLANG_OPER_NON_INLINED_CALL, /* a real function call */
+   SLANG_OPER_METHOD,           /* method call, such as  v.length() */
    SLANG_OPER_FIELD,            /* i.e.: ".next" or ".xzy" or ".xxx" etc */
    SLANG_OPER_POSTINCREMENT,    /* [var] "++" */
    SLANG_OPER_POSTDECREMENT     /* [var] "--" */
@@ -115,6 +116,7 @@ typedef struct slang_operation_
    GLfloat literal[4];           /**< Used for float, int and bool values */
    GLuint literal_size;          /**< 1, 2, 3, or 4 */
    slang_atom a_id;              /**< type: asm, identifier, call, field */
+   slang_atom a_obj;             /**< object in a method call */
    slang_variable_scope *locals; /**< local vars for scope */
    struct slang_function_ *fun;  /**< If type == SLANG_OPER_CALL */
    struct slang_variable_ *var;  /**< If type == slang_oper_identier */
diff --git a/src/mesa/shader/slang/slang_compile_variable.c b/src/mesa/shader/slang/slang_compile_variable.c
index 3428b49e166..abe9c55d799 100644
--- a/src/mesa/shader/slang/slang_compile_variable.c
+++ b/src/mesa/shader/slang/slang_compile_variable.c
@@ -122,6 +122,9 @@ slang_fully_specified_type_copy(slang_fully_specified_type * x,
    if (!slang_fully_specified_type_construct(&z))
       return 0;
    z.qualifier = y->qualifier;
+   z.precision = y->precision;
+   z.variant = y->variant;
+   z.centroid = y->centroid;
    if (!slang_type_specifier_copy(&z.specifier, &y->specifier)) {
       slang_fully_specified_type_destruct(&z);
       return 0;
@@ -266,7 +269,7 @@ slang_variable_construct(slang_variable * var)
    var->address = ~0;
    var->size = 0;
    var->isTemp = GL_FALSE;
-   var->aux = NULL;
+   var->store = NULL;
    var->declared = 0;
    return 1;
 }
diff --git a/src/mesa/shader/slang/slang_compile_variable.h b/src/mesa/shader/slang/slang_compile_variable.h
index d81a3d28699..d2a86b772c6 100644
--- a/src/mesa/shader/slang/slang_compile_variable.h
+++ b/src/mesa/shader/slang/slang_compile_variable.h
@@ -30,6 +30,23 @@ extern "C" {
 #endif
 
 
+struct slang_ir_storage_;
+
+
+typedef enum slang_type_variant_
+{
+   SLANG_VARIANT,    /* the default */
+   SLANG_INVARIANT   /* indicates the "invariant" keyword */
+} slang_type_variant;
+
+
+typedef enum slang_type_centroid_
+{
+   SLANG_CENTER,    /* the default */
+   SLANG_CENTROID   /* indicates the "centroid" keyword */
+} slang_type_centroid;
+
+
 typedef enum slang_type_qualifier_
 {
    SLANG_QUAL_NONE,
@@ -65,6 +82,9 @@ typedef struct slang_fully_specified_type_
    slang_type_qualifier qualifier;
    slang_type_specifier specifier;
    slang_type_precision precision;
+   slang_type_variant variant;
+   slang_type_centroid centroid;
+   GLint array_len;           /**< -1 if not an array type */
 } slang_fully_specified_type;
 
 extern int
@@ -91,7 +111,7 @@ typedef struct slang_variable_
    GLuint size;                     /**< Variable's size in bytes */
    GLboolean isTemp;                /**< a named temporary (__resultTmp) */
    GLboolean declared;              /**< for debug */
-   void *aux;                       /**< Used during code gen */
+   struct slang_ir_storage_ *store; /**< Storage for this var */
 } slang_variable;
 
 
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 672ec4bd606..e3cb252a3d7 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -60,6 +60,8 @@ typedef struct
    struct gl_program **Subroutines;
    GLuint NumSubroutines;
 
+   GLuint MaxInstructions;  /**< size of prog->Instructions[] buffer */
+
    /* code-gen options */
    GLboolean EmitHighLevelInstructions;
    GLboolean EmitCondCodes;
@@ -108,6 +110,30 @@ writemask_to_swizzle(GLuint writemask)
 
 
 /**
+ * Convert a swizzle mask to a writemask.
+ * Note that the slang_ir_storage->Swizzle field can represent either a
+ * swizzle mask or a writemask, depending on how it's used.  For example,
+ * when we parse "direction.yz" alone, we don't know whether .yz is a
+ * writemask or a swizzle.  In this case, we encode ".yz" in store->Swizzle
+ * as a swizzle mask (.yz?? actually).  Later, if direction.yz is used as
+ * an R-value, we use store->Swizzle as-is.  Otherwise, if direction.yz is
+ * used as an L-value, we convert it to a writemask.
+ */
+static GLuint
+swizzle_to_writemask(GLuint swizzle)
+{
+   GLuint i, writemask = 0x0;
+   for (i = 0; i < 4; i++) {
+      GLuint swz = GET_SWZ(swizzle, i);
+      if (swz <= SWIZZLE_W) {
+         writemask |= (1 << swz);
+      }
+   }
+   return writemask;
+}
+
+
+/**
  * Swizzle a swizzle (function composition).
  * That is, return swz2(swz1), or said another way: swz1.szw2
  * Example: swizzle_swizzle(".zwxx", ".xxyw") yields ".zzwx"
@@ -233,16 +259,18 @@ fix_swizzle(GLuint swizzle)
  * Convert IR storage to an instruction dst register.
  */
 static void
-storage_to_dst_reg(struct prog_dst_register *dst, const slang_ir_storage *st,
-                   GLuint writemask)
+storage_to_dst_reg(struct prog_dst_register *dst, const slang_ir_storage *st)
 {
+   const GLboolean relAddr = st->RelAddr;
    const GLint size = st->Size;
    GLint index = st->Index;
    GLuint swizzle = st->Swizzle;
 
+   assert(index >= 0);
    /* if this is storage relative to some parent storage, walk up the tree */
    while (st->Parent) {
       st = st->Parent;
+      assert(st->Index >= 0);
       index += st->Index;
       swizzle = _slang_swizzle_swizzle(st->Swizzle, swizzle);
    }
@@ -256,14 +284,31 @@ storage_to_dst_reg(struct prog_dst_register *dst, const slang_ir_storage *st,
    assert(size >= 1);
    assert(size <= 4);
 
-   if (size == 1) {
-      GLuint comp = GET_SWZ(swizzle, 0);
-      assert(comp < 4);
-      dst->WriteMask = WRITEMASK_X << comp;
+   if (swizzle != SWIZZLE_XYZW) {
+      dst->WriteMask = swizzle_to_writemask(swizzle);
    }
    else {
+      GLuint writemask;
+      switch (size) {
+      case 1:
+         writemask = WRITEMASK_X << GET_SWZ(st->Swizzle, 0);
+         break;
+      case 2:
+         writemask = WRITEMASK_XY;
+         break;
+      case 3:
+         writemask = WRITEMASK_XYZ;
+         break;
+      case 4:
+         writemask = WRITEMASK_XYZW;
+         break;
+      default:
+         ; /* error would have been caught above */
+      }
       dst->WriteMask = writemask;
    }
+
+   dst->RelAddr = relAddr;
 }
 
 
@@ -278,8 +323,10 @@ storage_to_src_reg(struct prog_src_register *src, const slang_ir_storage *st)
    GLuint swizzle = st->Swizzle;
 
    /* if this is storage relative to some parent storage, walk up the tree */
+   assert(index >= 0);
    while (st->Parent) {
       st = st->Parent;
+      assert(st->Index >= 0);
       index += st->Index;
       swizzle = _slang_swizzle_swizzle(fix_swizzle(st->Swizzle), swizzle);
    }
@@ -309,38 +356,28 @@ storage_to_src_reg(struct prog_src_register *src, const slang_ir_storage *st)
 
 
 /*
- * Setup an instrucion src register to point to a scalar constant.
+ * Setup storage pointing to a scalar constant/literal.
  */
 static void
-constant_to_src_reg(struct prog_src_register *src, GLfloat val,
-                    slang_emit_info *emitInfo)
+constant_to_storage(slang_emit_info *emitInfo,
+                    GLfloat val,
+                    slang_ir_storage *store)
 {
-   GLuint zeroSwizzle;
-   GLint zeroReg;
+   GLuint swizzle;
+   GLint reg;
    GLfloat value[4];
 
    value[0] = val;
-   zeroReg = _mesa_add_unnamed_constant(emitInfo->prog->Parameters,
-                                        value, 1, &zeroSwizzle);
-   assert(zeroReg >= 0);
+   reg = _mesa_add_unnamed_constant(emitInfo->prog->Parameters,
+                                        value, 1, &swizzle);
 
-   src->File = PROGRAM_CONSTANT;
-   src->Index = zeroReg;
-   src->Swizzle = zeroSwizzle;
-}
-
-
-static void
-address_to_dst_reg(struct prog_dst_register *dst, GLuint index)
-{
-   assert(index == 0); /* only one address reg at this time */
-   dst->File = PROGRAM_ADDRESS;
-   dst->Index = index;
-   dst->WriteMask = WRITEMASK_X;
+   memset(store, 0, sizeof(*store));
+   store->File = PROGRAM_CONSTANT;
+   store->Index = reg;
+   store->Swizzle = swizzle;
 }
 
 
-
 /**
  * Add new instruction at end of given program.
  * \param prog  the program to append instruction onto
@@ -359,9 +396,17 @@ new_instruction(slang_emit_info *emitInfo, gl_inst_opcode opcode)
       _mesa_print_instruction(prog->Instructions + prog->NumInstructions - 1);
    }
 #endif
-   prog->Instructions = _mesa_realloc_instructions(prog->Instructions,
-                                                   prog->NumInstructions,
-                                                   prog->NumInstructions + 1);
+   assert(prog->NumInstructions <= emitInfo->MaxInstructions);
+
+   if (prog->NumInstructions == emitInfo->MaxInstructions) {
+      /* grow the instruction buffer */
+      emitInfo->MaxInstructions += 20;
+      prog->Instructions =
+         _mesa_realloc_instructions(prog->Instructions,
+                                    prog->NumInstructions,
+                                    emitInfo->MaxInstructions);
+   }
+
    inst = prog->Instructions + prog->NumInstructions;
    prog->NumInstructions++;
    _mesa_init_instructions(inst, 1);
@@ -375,6 +420,166 @@ new_instruction(slang_emit_info *emitInfo, gl_inst_opcode opcode)
 }
 
 
+static struct prog_instruction *
+emit_arl_load(slang_emit_info *emitInfo,
+              enum register_file file, GLint index, GLuint swizzle)
+{
+   struct prog_instruction *inst = new_instruction(emitInfo, OPCODE_ARL);
+   inst->SrcReg[0].File = file;
+   inst->SrcReg[0].Index = index;
+   inst->SrcReg[0].Swizzle = swizzle;
+   inst->DstReg.File = PROGRAM_ADDRESS;
+   inst->DstReg.Index = 0;
+   inst->DstReg.WriteMask = WRITEMASK_X;
+   return inst;
+}
+
+
+/**
+ * Emit a new instruction with given opcode, operands.
+ * At this point the instruction may have multiple indirect register
+ * loads/stores.  We convert those into ARL loads and address-relative
+ * operands.  See comments inside.
+ * At some point in the future we could directly emit indirectly addressed
+ * registers in Mesa GPU instructions.
+ */
+static struct prog_instruction *
+emit_instruction(slang_emit_info *emitInfo,
+                 gl_inst_opcode opcode,
+                 const slang_ir_storage *dst,
+                 const slang_ir_storage *src0,
+                 const slang_ir_storage *src1,
+                 const slang_ir_storage *src2)
+{
+   struct prog_instruction *inst;
+   GLuint numIndirect = 0;
+   const slang_ir_storage *src[3];
+   slang_ir_storage newSrc[3], newDst;
+   GLuint i;
+   GLboolean isTemp[3];
+
+   isTemp[0] = isTemp[1] = isTemp[2] = GL_FALSE;
+
+   src[0] = src0;
+   src[1] = src1;
+   src[2] = src2;
+
+   /* count up how many operands are indirect loads */
+   for (i = 0; i < 3; i++) {
+      if (src[i] && src[i]->IsIndirect)
+         numIndirect++;
+   }
+   if (dst && dst->IsIndirect)
+      numIndirect++;
+
+   /* Take special steps for indirect register loads.
+    * If we had multiple address registers this would be simpler.
+    * For example, this GLSL code:
+    *    x[i] = y[j] + z[k];
+    * would translate into something like:
+    *    ARL ADDR.x, i;
+    *    ARL ADDR.y, j;
+    *    ARL ADDR.z, k;
+    *    ADD TEMP[ADDR.x+5], TEMP[ADDR.y+9], TEMP[ADDR.z+4];
+    * But since we currently only have one address register we have to do this:
+    *    ARL ADDR.x, i;
+    *    MOV t1, TEMP[ADDR.x+9];
+    *    ARL ADDR.x, j;
+    *    MOV t2, TEMP[ADDR.x+4];
+    *    ARL ADDR.x, k;
+    *    ADD TEMP[ADDR.x+5], t1, t2;
+    * The code here figures this out...
+    */
+   if (numIndirect > 0) {
+      for (i = 0; i < 3; i++) {
+         if (src[i] && src[i]->IsIndirect) {
+            /* load the ARL register with the indirect register */
+            emit_arl_load(emitInfo,
+                          src[i]->IndirectFile,
+                          src[i]->IndirectIndex,
+                          src[i]->IndirectSwizzle);
+
+            if (numIndirect > 1) {
+               /* Need to load src[i] into a temporary register */
+               slang_ir_storage srcRelAddr;
+               alloc_local_temp(emitInfo, &newSrc[i], src[i]->Size);
+               isTemp[i] = GL_TRUE;
+
+               /* set RelAddr flag on src register */
+               srcRelAddr = *src[i];
+               srcRelAddr.RelAddr = GL_TRUE;
+               srcRelAddr.IsIndirect = GL_FALSE; /* not really needed */
+
+               /* MOV newSrc, srcRelAddr; */
+               inst = emit_instruction(emitInfo,
+                                       OPCODE_MOV,
+                                       &newSrc[i],
+                                       &srcRelAddr,
+                                       NULL,
+                                       NULL);
+
+               src[i] = &newSrc[i];
+            }
+            else {
+               /* just rewrite the src[i] storage to be ARL-relative */
+               newSrc[i] = *src[i];
+               newSrc[i].RelAddr = GL_TRUE;
+               newSrc[i].IsIndirect = GL_FALSE; /* not really needed */
+               src[i] = &newSrc[i];
+            }
+         }
+      }
+   }
+
+   /* Take special steps for indirect dest register write */
+   if (dst && dst->IsIndirect) {
+      /* load the ARL register with the indirect register */
+      emit_arl_load(emitInfo,
+                    dst->IndirectFile,
+                    dst->IndirectIndex,
+                    dst->IndirectSwizzle);
+      newDst = *dst;
+      newDst.RelAddr = GL_TRUE;
+      newDst.IsIndirect = GL_FALSE;
+      dst = &newDst;
+   }
+
+   /* OK, emit the instruction and its dst, src regs */
+   inst = new_instruction(emitInfo, opcode);
+   if (!inst)
+      return NULL;
+
+   if (dst)
+      storage_to_dst_reg(&inst->DstReg, dst);
+
+   for (i = 0; i < 3; i++) {
+      if (src[i])
+         storage_to_src_reg(&inst->SrcReg[i], src[i]);
+   }
+
+   /* Free any temp registers that we allocated above */
+   for (i = 0; i < 3; i++) {
+      if (isTemp[i])
+         _slang_free_temp(emitInfo->vt, &newSrc[i]);
+   }
+
+   return inst;
+}
+
+
+
+/**
+ * Put a comment on the given instruction.
+ */
+static void
+inst_comment(struct prog_instruction *inst, const char *comment)
+{
+   if (inst)
+      inst->Comment = _mesa_strdup(comment);
+}
+
+
+
 /**
  * Return pointer to last instruction in program.
  */
@@ -488,6 +693,9 @@ instruction_annotation(gl_inst_opcode opcode, char *dstAnnot,
    case OPCODE_MUL:
       operator = "*";
       break;
+   case OPCODE_DP2:
+      operator = "DP2";
+      break;
    case OPCODE_DP3:
       operator = "DP3";
       break;
@@ -528,12 +736,10 @@ instruction_annotation(gl_inst_opcode opcode, char *dstAnnot,
  * Emit an instruction that's just a comment.
  */
 static struct prog_instruction *
-emit_comment(slang_emit_info *emitInfo, const char *s)
+emit_comment(slang_emit_info *emitInfo, const char *comment)
 {
    struct prog_instruction *inst = new_instruction(emitInfo, OPCODE_NOP);
-   if (inst) {
-      inst->Comment = _mesa_strdup(s);
-   }
+   inst_comment(inst, comment);
    return inst;
 }
 
@@ -545,22 +751,13 @@ emit_comment(slang_emit_info *emitInfo, const char *s)
 static struct prog_instruction *
 emit_arith(slang_emit_info *emitInfo, slang_ir_node *n)
 {
-   struct prog_instruction *inst;
    const slang_ir_info *info = _slang_ir_info(n->Opcode);
-   char *srcAnnot[3], *dstAnnot;
+   struct prog_instruction *inst;
    GLuint i;
-   slang_ir_node *temps[3];
-
-   /* we'll save pointers to nodes/storage to free in temps[] until
-    * the very end.
-    */
-   temps[0] = temps[1] = temps[2] = NULL;
 
    assert(info);
    assert(info->InstOpcode != OPCODE_NOP);
 
-   srcAnnot[0] = srcAnnot[1] = srcAnnot[2] = dstAnnot = NULL;
-
 #if PEEPHOLE_OPTIMIZATIONS
    /* Look for MAD opportunity */
    if (info->NumParams == 2 &&
@@ -569,85 +766,67 @@ emit_arith(slang_emit_info *emitInfo, slang_ir_node *n)
       emit(emitInfo, n->Children[0]->Children[0]);  /* A */
       emit(emitInfo, n->Children[0]->Children[1]);  /* B */
       emit(emitInfo, n->Children[1]);  /* C */
-      /* generate MAD instruction */
-      inst = new_instruction(emitInfo, OPCODE_MAD);
-      /* operands: A, B, C: */
-      storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Children[0]->Store);
-      storage_to_src_reg(&inst->SrcReg[1], n->Children[0]->Children[1]->Store);
-      storage_to_src_reg(&inst->SrcReg[2], n->Children[1]->Store);
-      temps[0] = n->Children[0]->Children[0];
-      temps[1] = n->Children[0]->Children[1];
-      temps[2] = n->Children[1];
-   }
-   else if (info->NumParams == 2 &&
-            n->Opcode == IR_ADD && n->Children[1]->Opcode == IR_MUL) {
+      alloc_node_storage(emitInfo, n, -1);  /* dest */
+
+      inst = emit_instruction(emitInfo,
+                              OPCODE_MAD,
+                              n->Store,
+                              n->Children[0]->Children[0]->Store,
+                              n->Children[0]->Children[1]->Store,
+                              n->Children[1]->Store);
+
+      free_node_storage(emitInfo->vt, n->Children[0]->Children[0]);
+      free_node_storage(emitInfo->vt, n->Children[0]->Children[1]);
+      free_node_storage(emitInfo->vt, n->Children[1]);
+      return inst;
+   }
+
+   if (info->NumParams == 2 &&
+       n->Opcode == IR_ADD && n->Children[1]->Opcode == IR_MUL) {
       /* found pattern IR_ADD(A, IR_MUL(B, C)) */
       emit(emitInfo, n->Children[0]);  /* A */
       emit(emitInfo, n->Children[1]->Children[0]);  /* B */
       emit(emitInfo, n->Children[1]->Children[1]);  /* C */
-      /* generate MAD instruction */
-      inst = new_instruction(emitInfo, OPCODE_MAD);
-      /* operands: B, C, A */
-      storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Children[0]->Store);
-      storage_to_src_reg(&inst->SrcReg[1], n->Children[1]->Children[1]->Store);
-      storage_to_src_reg(&inst->SrcReg[2], n->Children[0]->Store);
-      temps[0] = n->Children[1]->Children[0];
-      temps[1] = n->Children[1]->Children[1];
-      temps[2] = n->Children[0];
+      alloc_node_storage(emitInfo, n, -1);  /* dest */
+
+      inst = emit_instruction(emitInfo,
+                              OPCODE_MAD,
+                              n->Store,
+                              n->Children[1]->Children[0]->Store,
+                              n->Children[1]->Children[1]->Store,
+                              n->Children[0]->Store);
+
+      free_node_storage(emitInfo->vt, n->Children[1]->Children[0]);
+      free_node_storage(emitInfo->vt, n->Children[1]->Children[1]);
+      free_node_storage(emitInfo->vt, n->Children[0]);
+      return inst;
    }
-   else
 #endif
-   {
-      /* normal case */
 
-      /* gen code for children */
-      for (i = 0; i < info->NumParams; i++) {
-         emit(emitInfo, n->Children[i]);
-         if (!n->Children[i] || !n->Children[i]->Store) {
-            /* error recovery */
-            return NULL;
-         }
+   /* gen code for children, may involve temp allocation */
+   for (i = 0; i < info->NumParams; i++) {
+      emit(emitInfo, n->Children[i]);
+      if (!n->Children[i] || !n->Children[i]->Store) {
+         /* error recovery */
+         return NULL;
       }
-
-      /* gen this instruction and src registers */
-      inst = new_instruction(emitInfo, info->InstOpcode);
-      for (i = 0; i < info->NumParams; i++)
-         storage_to_src_reg(&inst->SrcReg[i], n->Children[i]->Store);
-
-      /* annotation */
-      for (i = 0; i < info->NumParams; i++)
-         srcAnnot[i] = storage_annotation(n->Children[i], emitInfo->prog);
-
-      /* record (potential) temps to free */
-      for (i = 0; i < info->NumParams; i++)
-         temps[i] = n->Children[i];
    }
 
    /* result storage */
    alloc_node_storage(emitInfo, n, -1);
 
-   assert(n->Store->Index >= 0);
-   if (n->Store->Size == 2)
-      n->Writemask = WRITEMASK_XY;
-   else if (n->Store->Size == 3)
-      n->Writemask = WRITEMASK_XYZ;
-   else if (n->Store->Size == 1)
-      n->Writemask = WRITEMASK_X << GET_SWZ(n->Store->Swizzle, 0);
-
-
-   storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
+   inst = emit_instruction(emitInfo,
+                           info->InstOpcode,
+                           n->Store,  /* dest */
+                           (info->NumParams > 0 ? n->Children[0]->Store : NULL),
+                           (info->NumParams > 1 ? n->Children[1]->Store : NULL),
+                           (info->NumParams > 2 ? n->Children[2]->Store : NULL)
+                           );
 
-   dstAnnot = storage_annotation(n, emitInfo->prog);
-
-   inst->Comment = instruction_annotation(inst->Opcode, dstAnnot, srcAnnot[0],
-                                          srcAnnot[1], srcAnnot[2]);
-
-   /* really free temps now */
-   for (i = 0; i < 3; i++)
-      if (temps[i])
-         free_node_storage(emitInfo->vt, temps[i]);
+   /* free temps */
+   for (i = 0; i < info->NumParams; i++)
+      free_node_storage(emitInfo->vt, n->Children[i]);
 
-   /*_mesa_print_instruction(inst);*/
    return inst;
 }
 
@@ -659,7 +838,7 @@ emit_arith(slang_emit_info *emitInfo, slang_ir_node *n)
 static struct prog_instruction *
 emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
 {
-   struct prog_instruction *inst;
+   struct prog_instruction *inst = NULL;
    GLint size;
 
    assert(n->Opcode == IR_EQUAL || n->Opcode == IR_NOTEQUAL);
@@ -680,15 +859,20 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
    size = n->Children[0]->Store->Size;
 
    if (size == 1) {
-      gl_inst_opcode opcode;
-
-      opcode = n->Opcode == IR_EQUAL ? OPCODE_SEQ : OPCODE_SNE;
-      inst = new_instruction(emitInfo, opcode);
-      storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
-      storage_to_src_reg(&inst->SrcReg[1], n->Children[1]->Store);
-      storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
+      gl_inst_opcode opcode = n->Opcode == IR_EQUAL ? OPCODE_SEQ : OPCODE_SNE;
+      inst =  emit_instruction(emitInfo,
+                               opcode,
+                               n->Store, /* dest */
+                               n->Children[0]->Store,
+                               n->Children[1]->Store,
+                               NULL);
    }
    else if (size <= 4) {
+      /* compare two vectors.
+       * Unfortunately, there's no instruction to compare vectors and
+       * return a scalar result.  Do it with some compare and dot product
+       * instructions...
+       */
       GLuint swizzle;
       gl_inst_opcode dotOp;
       slang_ir_storage tempStore;
@@ -708,34 +892,42 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
       }
       else {
          assert(size == 2);
-         dotOp = OPCODE_DP3;
+         dotOp = OPCODE_DP3; /* XXX use OPCODE_DP2 eventually */
          swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y);
       }
 
       /* Compute inequality (temp = (A != B)) */
-      inst = new_instruction(emitInfo, OPCODE_SNE);
-      storage_to_dst_reg(&inst->DstReg, &tempStore, n->Writemask);
-      storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
-      storage_to_src_reg(&inst->SrcReg[1], n->Children[1]->Store);
-      inst->Comment = _mesa_strdup("Compare values");
+      inst = emit_instruction(emitInfo,
+                              OPCODE_SNE,
+                              &tempStore,
+                              n->Children[0]->Store,
+                              n->Children[1]->Store,
+                              NULL);
+      inst_comment(inst, "Compare values");
 
       /* Compute val = DOT(temp, temp)  (reduction) */
-      inst = new_instruction(emitInfo, dotOp);
-      storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-      storage_to_src_reg(&inst->SrcReg[0], &tempStore);
-      storage_to_src_reg(&inst->SrcReg[1], &tempStore);
+      inst = emit_instruction(emitInfo,
+                              dotOp,
+                              n->Store,
+                              &tempStore,
+                              &tempStore,
+                              NULL);
       inst->SrcReg[0].Swizzle = inst->SrcReg[1].Swizzle = swizzle; /*override*/
-      inst->Comment = _mesa_strdup("Reduce vec to bool");
+      inst_comment(inst, "Reduce vec to bool");
 
       _slang_free_temp(emitInfo->vt, &tempStore); /* free temp */
 
       if (n->Opcode == IR_EQUAL) {
          /* compute val = !val.x  with SEQ val, val, 0; */
-         inst = new_instruction(emitInfo, OPCODE_SEQ);
-         storage_to_src_reg(&inst->SrcReg[0], n->Store);
-         constant_to_src_reg(&inst->SrcReg[1], 0.0, emitInfo);
-         storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-         inst->Comment = _mesa_strdup("Invert true/false");
+         slang_ir_storage zero;
+         constant_to_storage(emitInfo, 0.0, &zero);
+         inst = emit_instruction(emitInfo,
+                                 OPCODE_SEQ,
+                                 n->Store, /* dest */
+                                 n->Store,
+                                 &zero,
+                                 NULL);
+         inst_comment(inst, "Invert true/false");
       }
    }
    else {
@@ -752,41 +944,54 @@ emit_compare(slang_emit_info *emitInfo, slang_ir_node *n)
          return NULL;
 
       for (i = 0; i < num; i++) {
-         /* SNE sneTemp, left[i], right[i] */
-         inst = new_instruction(emitInfo, OPCODE_SNE);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
-         storage_to_src_reg(&inst->SrcReg[1], n->Children[1]->Store);
-         inst->SrcReg[0].Index += i;
-         inst->SrcReg[1].Index += i;
+         slang_ir_storage srcStore0 = *n->Children[0]->Store;
+         slang_ir_storage srcStore1 = *n->Children[1]->Store;
+         srcStore0.Index += i;
+         srcStore1.Index += i;
+
          if (i == 0) {
-            storage_to_dst_reg(&inst->DstReg, &accTemp, WRITEMASK_XYZW);
-            inst->Comment = _mesa_strdup("Begin struct/array comparison");
+            /* SNE accTemp, left[i], right[i] */
+            inst = emit_instruction(emitInfo, OPCODE_SNE,
+                                    &accTemp, /* dest */
+                                    &srcStore0,
+                                    &srcStore1,
+                                    NULL);
+            inst_comment(inst, "Begin struct/array comparison");
          }
          else {
-            storage_to_dst_reg(&inst->DstReg, &sneTemp, WRITEMASK_XYZW);
-
+            /* SNE sneTemp, left[i], right[i] */
+            inst = emit_instruction(emitInfo, OPCODE_SNE,
+                                    &sneTemp, /* dest */
+                                    &srcStore0,
+                                    &srcStore1,
+                                    NULL);
             /* ADD accTemp, accTemp, sneTemp; # like logical-OR */
-            inst = new_instruction(emitInfo, OPCODE_ADD);
-            storage_to_dst_reg(&inst->DstReg, &accTemp, WRITEMASK_XYZW);
-            storage_to_src_reg(&inst->SrcReg[0], &accTemp);
-            storage_to_src_reg(&inst->SrcReg[1], &sneTemp);
+            inst = emit_instruction(emitInfo, OPCODE_ADD,
+                                    &accTemp, /* dest */
+                                    &accTemp,
+                                    &sneTemp,
+                                    NULL);
          }
       }
 
       /* compute accTemp.x || accTemp.y || accTemp.z || accTemp.w with DOT4 */
-      inst = new_instruction(emitInfo, OPCODE_DP4);
-      storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-      storage_to_src_reg(&inst->SrcReg[0], &accTemp);
-      storage_to_src_reg(&inst->SrcReg[1], &accTemp);
-      inst->Comment = _mesa_strdup("End struct/array comparison");
+      inst = emit_instruction(emitInfo, OPCODE_DP4,
+                              n->Store,
+                              &accTemp,
+                              &accTemp,
+                              NULL);
+      inst_comment(inst, "End struct/array comparison");
 
       if (n->Opcode == IR_EQUAL) {
          /* compute tmp.x = !tmp.x  via tmp.x = (tmp.x == 0) */
-         inst = new_instruction(emitInfo, OPCODE_SEQ);
-         storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-         storage_to_src_reg(&inst->SrcReg[0], n->Store);
-         constant_to_src_reg(&inst->SrcReg[1], 0.0, emitInfo);
-         inst->Comment = _mesa_strdup("Invert true/false");
+         slang_ir_storage zero;
+         constant_to_storage(emitInfo, 0.0, &zero);
+         inst = emit_instruction(emitInfo, OPCODE_SEQ,
+                                 n->Store, /* dest */
+                                 n->Store,
+                                 &zero,
+                                 NULL);
+         inst_comment(inst, "Invert true/false");
       }
 
       _slang_free_temp(emitInfo->vt, &accTemp);
@@ -862,16 +1067,18 @@ emit_clamp(slang_emit_info *emitInfo, slang_ir_node *n)
    alloc_node_storage(emitInfo, &tmpNode, n->Store->Size);
 
    /* tmp = max(ch[0], ch[1]) */
-   inst = new_instruction(emitInfo, OPCODE_MAX);
-   storage_to_dst_reg(&inst->DstReg, tmpNode.Store, n->Writemask);
-   storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
-   storage_to_src_reg(&inst->SrcReg[1], n->Children[1]->Store);
+   inst = emit_instruction(emitInfo, OPCODE_MAX,
+                           tmpNode.Store, /* dest */
+                           n->Children[0]->Store,
+                           n->Children[1]->Store,
+                           NULL);
 
    /* n->dest = min(tmp, ch[2]) */
-   inst = new_instruction(emitInfo, OPCODE_MIN);
-   storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-   storage_to_src_reg(&inst->SrcReg[0], tmpNode.Store);
-   storage_to_src_reg(&inst->SrcReg[1], n->Children[2]->Store);
+   inst = emit_instruction(emitInfo, OPCODE_MIN,
+                           n->Store, /* dest */
+                           tmpNode.Store,
+                           n->Children[2]->Store,
+                           NULL);
 
    free_node_storage(emitInfo->vt, &tmpNode);
 
@@ -893,9 +1100,12 @@ emit_negation(slang_emit_info *emitInfo, slang_ir_node *n)
    if (!alloc_node_storage(emitInfo, n, n->Children[0]->Store->Size))
       return NULL;
 
-   inst = new_instruction(emitInfo, OPCODE_MOV);
-   storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-   storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
+   inst = emit_instruction(emitInfo,
+                           OPCODE_MOV,
+                           n->Store, /* dest */
+                           n->Children[0]->Store,
+                           NULL,
+                           NULL);
    inst->SrcReg[0].NegateBase = NEGATE_XYZW;
    return inst;
 }
@@ -930,13 +1140,17 @@ emit_fcall(slang_emit_info *emitInfo, slang_ir_node *n)
    struct gl_program *progSave;
    struct prog_instruction *inst;
    GLuint subroutineId;
+   GLuint maxInstSave;
 
    assert(n->Opcode == IR_CALL);
    assert(n->Label);
 
    /* save/push cur program */
+   maxInstSave = emitInfo->MaxInstructions;
    progSave = emitInfo->prog;
+
    emitInfo->prog = new_subroutine(emitInfo, &subroutineId);
+   emitInfo->MaxInstructions = emitInfo->prog->NumInstructions;
 
    _slang_label_set_location(n->Label, emitInfo->prog->NumInstructions,
                              emitInfo->prog);
@@ -948,7 +1162,7 @@ emit_fcall(slang_emit_info *emitInfo, slang_ir_node *n)
        * really just a NOP to attach the label to.
        */
       inst = new_instruction(emitInfo, OPCODE_BGNSUB);
-      inst->Comment = _mesa_strdup(n->Label->Name);
+      inst_comment(inst, n->Label->Name);
    }
 
    /* body of function: */
@@ -963,17 +1177,18 @@ emit_fcall(slang_emit_info *emitInfo, slang_ir_node *n)
 
    if (emitInfo->EmitBeginEndSub) {
       inst = new_instruction(emitInfo, OPCODE_ENDSUB);
-      inst->Comment = _mesa_strdup(n->Label->Name);
+      inst_comment(inst, n->Label->Name);
    }
 
    /* pop/restore cur program */
    emitInfo->prog = progSave;
+   emitInfo->MaxInstructions = maxInstSave;
 
    /* emit the function call */
    inst = new_instruction(emitInfo, OPCODE_CAL);
    /* The branch target is just the subroutine number (changed later) */
    inst->BranchTarget = subroutineId;
-   inst->Comment = _mesa_strdup(n->Label->Name);
+   inst_comment(inst, n->Label->Name);
    assert(inst->BranchTarget >= 0);
 
    return inst;
@@ -1019,28 +1234,33 @@ static struct prog_instruction *
 emit_tex(slang_emit_info *emitInfo, slang_ir_node *n)
 {
    struct prog_instruction *inst;
-
-   (void) emit(emitInfo, n->Children[1]);
+   gl_inst_opcode opcode;
 
    if (n->Opcode == IR_TEX) {
-      inst = new_instruction(emitInfo, OPCODE_TEX);
+      opcode = OPCODE_TEX;
    }
    else if (n->Opcode == IR_TEXB) {
-      inst = new_instruction(emitInfo, OPCODE_TXB);
+      opcode = OPCODE_TXB;
    }
    else {
       assert(n->Opcode == IR_TEXP);
-      inst = new_instruction(emitInfo, OPCODE_TXP);
+      opcode = OPCODE_TXP;
    }
 
+   /* emit code for the texcoord operand */
+   (void) emit(emitInfo, n->Children[1]);
+
+   /* alloc storage for result of texture fetch */
    if (!alloc_node_storage(emitInfo, n, 4))
       return NULL;
 
-   storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-
-   /* Child[1] is the coord */
-   assert(n->Children[1]->Store->Index >= 0);
-   storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
+   /* emit TEX instruction;  Child[1] is the texcoord */
+   inst = emit_instruction(emitInfo,
+                           opcode,
+                           n->Store,
+                           n->Children[1]->Store,
+                           NULL,
+                           NULL);
 
    /* Child[0] is the sampler (a uniform which'll indicate the texture unit) */
    assert(n->Children[0]->Store);
@@ -1051,14 +1271,8 @@ emit_tex(slang_emit_info *emitInfo, slang_ir_node *n)
    assert(n->Children[0]->Store->Size <= TEXTURE_RECT_INDEX);
 
    inst->TexSrcTarget = n->Children[0]->Store->Size;
-#if 0
-   inst->TexSrcUnit = 27; /* Dummy value; the TexSrcUnit will be computed at
-                           * link time, using the sampler uniform's value.
-                           */
-   inst->Sampler = n->Children[0]->Store->Index; /* i.e. uniform's index */
-#else
    inst->TexSrcUnit = n->Children[0]->Store->Index; /* i.e. uniform's index */
-#endif
+
    return inst;
 }
 
@@ -1109,12 +1323,19 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
    if (inst &&
        _slang_is_temp(emitInfo->vt, n->Children[1]->Store) &&
        (inst->DstReg.File == n->Children[1]->Store->File) &&
-       (inst->DstReg.Index == n->Children[1]->Store->Index)) {
+       (inst->DstReg.Index == n->Children[1]->Store->Index) &&
+       !n->Children[0]->Store->IsIndirect &&
+       n->Children[0]->Store->Size <= 4) {
       /* Peephole optimization:
        * The Right-Hand-Side has its results in a temporary place.
        * Modify the RHS (and the prev instruction) to store its results
        * in the destination specified by n->Children[0].
        * Then, this MOVE is a no-op.
+       * Ex:
+       *   MUL tmp, x, y;
+       *   MOV a, tmp;
+       * becomes:
+       *   MUL a, x, y;
        */
       if (n->Children[1]->Opcode != IR_SWIZZLE)
          _slang_free_temp(emitInfo->vt, n->Children[1]->Store);
@@ -1123,11 +1344,7 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
       /* fixup the previous instruction (which stored the RHS result) */
       assert(n->Children[0]->Store->Index >= 0);
 
-      /* use tighter writemask when possible */
-      if (n->Writemask == WRITEMASK_XYZW)
-         n->Writemask = inst->DstReg.WriteMask;
-
-      storage_to_dst_reg(&inst->DstReg, n->Children[0]->Store, n->Writemask);
+      storage_to_dst_reg(&inst->DstReg, n->Children[0]->Store);
       return inst;
    }
    else
@@ -1138,15 +1355,16 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
          slang_ir_storage dstStore = *n->Children[0]->Store;
          slang_ir_storage srcStore = *n->Children[1]->Store;
          GLint size = srcStore.Size;
-         ASSERT(n->Children[0]->Writemask == WRITEMASK_XYZW);
          ASSERT(n->Children[1]->Store->Swizzle == SWIZZLE_NOOP);
          dstStore.Size = 4;
          srcStore.Size = 4;
          while (size >= 4) {
-            inst = new_instruction(emitInfo, OPCODE_MOV);
-            inst->Comment = _mesa_strdup("IR_COPY block");
-            storage_to_dst_reg(&inst->DstReg, &dstStore, n->Writemask);
-            storage_to_src_reg(&inst->SrcReg[0], &srcStore);
+            inst = emit_instruction(emitInfo, OPCODE_MOV,
+                                    &dstStore,
+                                    &srcStore,
+                                    NULL,
+                                    NULL);
+            inst_comment(inst, "IR_COPY block");
             srcStore.Index++;
             dstStore.Index++;
             size -= 4;
@@ -1155,10 +1373,12 @@ emit_copy(slang_emit_info *emitInfo, slang_ir_node *n)
       else {
          /* single register move */
          char *srcAnnot, *dstAnnot;
-         inst = new_instruction(emitInfo, OPCODE_MOV);
          assert(n->Children[0]->Store->Index >= 0);
-         storage_to_dst_reg(&inst->DstReg, n->Children[0]->Store, n->Writemask);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
+         inst = emit_instruction(emitInfo, OPCODE_MOV,
+                                 n->Children[0]->Store, /* dest */
+                                 n->Children[1]->Store,
+                                 NULL,
+                                 NULL);
          dstAnnot = storage_annotation(n->Children[0], emitInfo->prog);
          srcAnnot = storage_annotation(n->Children[1], emitInfo->prog);
          inst->Comment = instruction_annotation(inst->Opcode, dstAnnot,
@@ -1215,12 +1435,14 @@ emit_cond(slang_emit_info *emitInfo, slang_ir_node *n)
           */
          if (!alloc_node_storage(emitInfo, n, 1))
             return NULL;
-         inst = new_instruction(emitInfo, OPCODE_MOV);
+         inst = emit_instruction(emitInfo, OPCODE_MOV,
+                                 n->Store, /* dest */
+                                 n->Children[0]->Store,
+                                 NULL,
+                                 NULL);
          inst->CondUpdate = GL_TRUE;
-         storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
+         inst_comment(inst, "COND expr");
          _slang_free_temp(emitInfo->vt, n->Store);
-         inst->Comment = _mesa_strdup("COND expr");
          return inst;
       }
    }
@@ -1250,6 +1472,7 @@ emit_not(slang_emit_info *emitInfo, slang_ir_node *n)
       { 0, 0 }
    };
    struct prog_instruction *inst;
+   slang_ir_storage zero;
    GLuint i;
 
    /* child expr */
@@ -1272,13 +1495,17 @@ emit_not(slang_emit_info *emitInfo, slang_ir_node *n)
    if (!alloc_node_storage(emitInfo, n, n->Children[0]->Store->Size))
       return NULL;
 
-   inst = new_instruction(emitInfo, OPCODE_SEQ);
-   storage_to_dst_reg(&inst->DstReg, n->Store, n->Writemask);
-   storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
-   constant_to_src_reg(&inst->SrcReg[1], 0.0, emitInfo);
+   constant_to_storage(emitInfo, 0.0, &zero);
+   inst = emit_instruction(emitInfo,
+                           OPCODE_SEQ,
+                           n->Store,
+                           n->Children[0]->Store,
+                           &zero,
+                           NULL);
+   inst_comment(inst, "NOT");
+
    free_node_storage(emitInfo->vt, n->Children[0]);
 
-   inst->Comment = _mesa_strdup("NOT");
    return inst;
 }
 
@@ -1312,8 +1539,10 @@ emit_if(slang_emit_info *emitInfo, slang_ir_node *n)
 
    ifInstLoc = prog->NumInstructions;
    if (emitInfo->EmitHighLevelInstructions) {
-      struct prog_instruction *ifInst = new_instruction(emitInfo, OPCODE_IF);
       if (emitInfo->EmitCondCodes) {
+         /* IF condcode THEN ... */
+         struct prog_instruction *ifInst;
+         ifInst = new_instruction(emitInfo, OPCODE_IF);
          ifInst->DstReg.CondMask = COND_NE;  /* if cond is non-zero */
          /* only test the cond code (1 of 4) that was updated by the
           * previous instruction.
@@ -1321,15 +1550,19 @@ emit_if(slang_emit_info *emitInfo, slang_ir_node *n)
          ifInst->DstReg.CondSwizzle = writemask_to_swizzle(condWritemask);
       }
       else {
-         /* test reg.x */
-         storage_to_src_reg(&ifInst->SrcReg[0], n->Children[0]->Store);
+         /* IF src[0] THEN ... */
+         emit_instruction(emitInfo, OPCODE_IF,
+                          NULL, /* dst */
+                          n->Children[0]->Store, /* op0 */
+                          NULL,
+                          NULL);
       }
    }
    else {
       /* conditional jump to else, or endif */
       struct prog_instruction *ifInst = new_instruction(emitInfo, OPCODE_BRA);
       ifInst->DstReg.CondMask = COND_EQ;  /* BRA if cond is zero */
-      ifInst->Comment = _mesa_strdup("if zero");
+      inst_comment(ifInst, "if zero");
       ifInst->DstReg.CondSwizzle = writemask_to_swizzle(condWritemask);
    }
 
@@ -1346,7 +1579,7 @@ emit_if(slang_emit_info *emitInfo, slang_ir_node *n)
          /* jump to endif instruction */
          struct prog_instruction *inst;
          inst = new_instruction(emitInfo, OPCODE_BRA);
-         inst->Comment = _mesa_strdup("else");
+         inst_comment(inst, "else");
          inst->DstReg.CondMask = COND_TR;  /* always branch */
       }
       prog->Instructions[ifInstLoc].BranchTarget = prog->NumInstructions;
@@ -1517,8 +1750,11 @@ emit_cont_break_if_true(slang_emit_info *emitInfo, slang_ir_node *n)
           */
          GLint ifInstLoc;
          ifInstLoc = emitInfo->prog->NumInstructions;
-         inst = new_instruction(emitInfo, OPCODE_IF);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[0]->Store);
+         inst = emit_instruction(emitInfo, OPCODE_IF,
+                                 NULL, /* dest */
+                                 n->Children[0]->Store,
+                                 NULL,
+                                 NULL);
          n->InstLocation = emitInfo->prog->NumInstructions;
 
          inst = new_instruction(emitInfo, opcode);
@@ -1547,158 +1783,140 @@ emit_swizzle(slang_emit_info *emitInfo, slang_ir_node *n)
 
    inst = emit(emitInfo, n->Children[0]);
 
-   /* setup storage info, if needed */
-   if (!n->Store->Parent)
-      n->Store->Parent = n->Children[0]->Store;
-
+#if 0
    assert(n->Store->Parent);
-
-   return inst;
-}
-
-
-/**
- * Move a block registers from src to dst (or move a single register).
- * \param size  size of block, in floats (<=4 means one register)
- */
-static struct prog_instruction *
-move_block(slang_emit_info *emitInfo,
-           GLuint size, GLboolean relAddr,
-           const slang_ir_storage *dst,
-           const slang_ir_storage *src)
-{
-   struct prog_instruction *inst;
-
-   if (size > 4) {
-      /* move matrix/struct etc (block of registers) */
-      slang_ir_storage dstStore = *dst;
-      slang_ir_storage srcStore = *src;
-      //GLint size = srcStore.Size;
-      /*ASSERT(n->Children[0]->Writemask == WRITEMASK_XYZW);
-      ASSERT(n->Children[1]->Store->Swizzle == SWIZZLE_NOOP);
-      */
-      dstStore.Size = 4;
-      srcStore.Size = 4;
-      while (size >= 4) {
-         inst = new_instruction(emitInfo, OPCODE_MOV);
-         inst->Comment = _mesa_strdup("IR_COPY block");
-         storage_to_dst_reg(&inst->DstReg, &dstStore, WRITEMASK_XYZW);
-         storage_to_src_reg(&inst->SrcReg[0], &srcStore);
-         inst->SrcReg[0].RelAddr = relAddr;
-         srcStore.Index++;
-         dstStore.Index++;
-         size -= 4;
-      }
-   }
-   else {
-      /* single register move */
-      GLuint writemask;
-      if (size == 1) {
-         GLuint comp = GET_SWZ(src->Swizzle, 0);
-         assert(comp < 4);
-         writemask = WRITEMASK_X << comp;
-      }
-      else {
-         writemask = WRITEMASK_XYZW;
-      }
-      inst = new_instruction(emitInfo, OPCODE_MOV);
-      storage_to_dst_reg(&inst->DstReg, dst, writemask);
-      storage_to_src_reg(&inst->SrcReg[0], src);
-      inst->SrcReg[0].RelAddr = relAddr;
-   }
+   /* Apply this node's swizzle to parent's storage */
+   GLuint swizzle = n->Store->Swizzle;
+   _slang_copy_ir_storage(n->Store, n->Store->Parent);
+   n->Store->Swizzle = _slang_swizzle_swizzle(n->Store->Swizzle, swizzle);
+   assert(!n->Store->Parent);
+#endif
    return inst;
 }
 
 
-
 /**
- * Dereference array element.  Just resolve storage for the array
- * element represented by this node.
+ * Dereference array element:  element == array[index]
+ * This basically involves emitting code for computing the array index
+ * and updating the node/element's storage info.
  */
 static struct prog_instruction *
 emit_array_element(slang_emit_info *emitInfo, slang_ir_node *n)
 {
-   slang_ir_storage *root;
+   slang_ir_storage *arrayStore, *indexStore;
+   const int elemSize = n->Store->Size;           /* number of floats */
+   const GLint elemSizeVec = (elemSize + 3) / 4;  /* number of vec4 */
+   struct prog_instruction *inst;
 
    assert(n->Opcode == IR_ELEMENT);
-   assert(n->Store);
-   assert(n->Store->File == PROGRAM_UNDEFINED);
-   assert(n->Store->Parent);
-   assert(n->Store->Size > 0);
-
-   root = n->Store;
-   while (root->Parent)
-      root = root->Parent;
+   assert(elemSize > 0);
 
-   if (root->File == PROGRAM_STATE_VAR) {
-      GLint index = _slang_alloc_statevar(n, emitInfo->prog->Parameters);
-      assert(n->Store->Index == index);
-      return NULL;
+   /* special case for built-in state variables, like light state */
+   {
+      slang_ir_storage *root = n->Store;
+      assert(!root->Parent);
+      while (root->Parent)
+         root = root->Parent;
+
+      if (root->File == PROGRAM_STATE_VAR) {
+         GLint index = _slang_alloc_statevar(n, emitInfo->prog->Parameters);
+         assert(n->Store->Index == index);
+         return NULL;
+      }
    }
 
-   /* do codegen for array */
+   /* do codegen for array itself */
    emit(emitInfo, n->Children[0]);
+   arrayStore = n->Children[0]->Store;
+
+   /* The initial array element storage is the array's storage,
+    * then modified below.
+    */
+   _slang_copy_ir_storage(n->Store, arrayStore);
+
 
    if (n->Children[1]->Opcode == IR_FLOAT) {
-      /* Constant array index.
-       * Set Store's index to be the offset of the array element in
-       * the register file.
-       */
+      /* Constant array index */
       const GLint element = (GLint) n->Children[1]->Value[0];
-      const GLint sz = (n->Store->Size + 3) / 4; /* size in slots/registers */
 
-      n->Store->Index = sz * element;
-      assert(n->Store->Parent);
+      /* this element's storage is the array's storage, plus constant offset */
+      n->Store->Index += elemSizeVec * element;
    }
    else {
       /* Variable array index */
-      struct prog_instruction *inst;
 
       /* do codegen for array index expression */
       emit(emitInfo, n->Children[1]);
+      indexStore = n->Children[1]->Store;
+
+      if (indexStore->IsIndirect) {
+         /* need to put the array index into a temporary since we can't
+          * directly support a[b[i]] constructs.
+          */
+
+
+         /*indexStore = tempstore();*/
+      }
 
-      /* allocate temp storage for the array element */
-      assert(n->Store->Index < 0);
-      n->Store->File = PROGRAM_TEMPORARY;
-      n->Store->Parent = NULL;
-      alloc_node_storage(emitInfo, n, -1);
 
-      if (n->Store->Size > 4) {
-         /* need to multiply the index by the element size */
-         GLint elemSize = (n->Store->Size + 3) / 4;
-         slang_ir_storage indexTemp;
+      if (elemSize > 4) {
+         /* need to multiply array index by array element size */
+         struct prog_instruction *inst;
+         slang_ir_storage *indexTemp;
+         slang_ir_storage elemSizeStore;
 
          /* allocate 1 float indexTemp */
-         alloc_local_temp(emitInfo, &indexTemp, 1);
+         indexTemp = _slang_new_ir_storage(PROGRAM_TEMPORARY, -1, 1);
+         _slang_alloc_temp(emitInfo->vt, indexTemp);
 
-         /* MUL temp, index, elemSize */
-         inst = new_instruction(emitInfo, OPCODE_MUL);
-         storage_to_dst_reg(&inst->DstReg, &indexTemp, WRITEMASK_X);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
-         constant_to_src_reg(&inst->SrcReg[1], elemSize, emitInfo);
+         /* allocate a constant containing the element size */
+         constant_to_storage(emitInfo, (float) elemSizeVec, &elemSizeStore);
 
-         /* load ADDR[0].X = temp */
-         inst = new_instruction(emitInfo, OPCODE_ARL);
-         storage_to_src_reg(&inst->SrcReg[0], &indexTemp);
-         address_to_dst_reg(&inst->DstReg, 0);
+         /* multiply array index by element size */
+         inst = emit_instruction(emitInfo,
+                                 OPCODE_MUL,
+                                 indexTemp, /* dest */
+                                 indexStore, /* the index */
+                                 &elemSizeStore,
+                                 NULL);
 
-         _slang_free_temp(emitInfo->vt, &indexTemp);
+         indexStore = indexTemp;
       }
-      else {
-         /* simply load address reg w/ array index */
-         inst = new_instruction(emitInfo, OPCODE_ARL);
-         storage_to_src_reg(&inst->SrcReg[0], n->Children[1]->Store);
-         address_to_dst_reg(&inst->DstReg, 0);
+
+      if (arrayStore->IsIndirect) {
+         /* ex: in a[i][j], a[i] (the arrayStore) is indirect */
+         /* Need to add indexStore to arrayStore->Indirect store */
+         slang_ir_storage indirectArray;
+         slang_ir_storage *indexTemp;
+
+         _slang_init_ir_storage(&indirectArray,
+                                arrayStore->IndirectFile,
+                                arrayStore->IndirectIndex,
+                                1,
+                                arrayStore->IndirectSwizzle);
+
+         /* allocate 1 float indexTemp */
+         indexTemp = _slang_new_ir_storage(PROGRAM_TEMPORARY, -1, 1);
+         _slang_alloc_temp(emitInfo->vt, indexTemp);
+
+         inst = emit_instruction(emitInfo,
+                                 OPCODE_ADD,
+                                 indexTemp,      /* dest */
+                                 indexStore,     /* the index */
+                                 &indirectArray, /* indirect array base */
+                                 NULL);
+
+         indexStore = indexTemp;
       }
 
-      /* copy from array element to temp storage */
-      move_block(emitInfo, n->Store->Size, GL_TRUE,
-                 n->Store, n->Children[0]->Store);
+      /* update the array element storage info */
+      n->Store->IsIndirect = GL_TRUE;
+      n->Store->IndirectFile = indexStore->File;
+      n->Store->IndirectIndex = indexStore->Index;
+      n->Store->IndirectSwizzle = indexStore->Swizzle;
    }
 
-   /* if array element size is one, make sure we only access X */
-   if (n->Store->Size == 1)
-      n->Store->Swizzle = SWIZZLE_XXXX;
+   n->Store->Size = elemSize;
 
    return NULL; /* no instruction */
 }
@@ -1711,9 +1929,11 @@ static struct prog_instruction *
 emit_struct_field(slang_emit_info *emitInfo, slang_ir_node *n)
 {
    slang_ir_storage *root = n->Store;
+   GLint fieldOffset, fieldSize;
 
    assert(n->Opcode == IR_FIELD);
 
+   assert(!root->Parent);
    while (root->Parent)
       root = root->Parent;
 
@@ -1729,12 +1949,45 @@ emit_struct_field(slang_emit_info *emitInfo, slang_ir_node *n)
          slang_info_log_error(emitInfo->log, "Error parsing state variable");
          return NULL;
       }
+      return NULL;
    }
    else {
       /* do codegen for struct */
       emit(emitInfo, n->Children[0]);
+      assert(n->Children[0]->Store->Index >= 0);
+   }
+
+   fieldOffset = n->Store->Index;
+   fieldSize = n->Store->Size;
+
+   _slang_copy_ir_storage(n->Store, n->Children[0]->Store);
+
+   n->Store->Index = n->Children[0]->Store->Index + fieldOffset / 4;
+   /* XXX test this:
+   n->Store->Index += fieldOffset / 4;
+   */
+
+   switch (fieldSize) {
+   case 1:
+      {
+         GLint swz = fieldOffset % 4;
+         n->Store->Swizzle = MAKE_SWIZZLE4(swz, swz, swz, swz);
+      }
+      break;
+   case 2:
+      n->Store->Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y,
+                                        SWIZZLE_NIL, SWIZZLE_NIL);
+      break;
+   case 3:
+      n->Store->Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y,
+                                        SWIZZLE_Z, SWIZZLE_NIL);
+      break;
+   default:
+      n->Store->Swizzle = SWIZZLE_XYZW;
    }
 
+   assert(n->Store->Index >= 0);
+
    return NULL; /* no instruction */
 }
 
@@ -1747,8 +2000,6 @@ emit_struct_field(slang_emit_info *emitInfo, slang_ir_node *n)
 static struct prog_instruction *
 emit_var_decl(slang_emit_info *emitInfo, slang_ir_node *n)
 {
-   struct prog_instruction *inst;
-
    assert(n->Store);
    assert(n->Store->File != PROGRAM_UNDEFINED);
    assert(n->Store->Size > 0);
@@ -1775,7 +2026,7 @@ emit_var_decl(slang_emit_info *emitInfo, slang_ir_node *n)
         printf("IR_VAR_DECL %s %d store %p\n",
         (char*) n->Var->a_name, n->Store->Index, (void*) n->Store);
       */
-      assert(n->Var->aux == n->Store);
+      assert(n->Var->store == n->Store);
    }
    if (emitInfo->EmitComments) {
       /* emit NOP with comment describing the variable's storage location */
@@ -1785,8 +2036,7 @@ emit_var_decl(slang_emit_info *emitInfo, slang_ir_node *n)
               _mesa_swizzle_string(n->Store->Swizzle, 0, GL_FALSE), 
               (n->Var ? (char *) n->Var->a_name : "anonymous"),
               n->Store->Size);
-      inst = emit_comment(emitInfo, s);
-      return inst;
+      emit_comment(emitInfo, s);
    }
    return NULL;
 }
@@ -1794,7 +2044,7 @@ emit_var_decl(slang_emit_info *emitInfo, slang_ir_node *n)
 
 /**
  * Emit code for a reference to a variable.
- * Actually, no code is generated but we may do some memory alloation.
+ * Actually, no code is generated but we may do some memory allocation.
  * In particular, state vars (uniforms) are allocated on an as-needed basis.
  */
 static struct prog_instruction *
@@ -1893,12 +2143,15 @@ emit(slang_emit_info *emitInfo, slang_ir_node *n)
    case IR_NOISE2:
    case IR_NOISE3:
    case IR_NOISE4:
+   case IR_NRM4:
+   case IR_NRM3:
    /* binary */
    case IR_ADD:
    case IR_SUB:
    case IR_MUL:
    case IR_DOT4:
    case IR_DOT3:
+   case IR_DOT2:
    case IR_CROSS:
    case IR_MIN:
    case IR_MAX:
@@ -2012,7 +2265,7 @@ _slang_resolve_subroutines(slang_emit_info *emitInfo)
       total += emitInfo->Subroutines[i]->NumInstructions;
    }
 
-   /* adjust BrancTargets within the functions */
+   /* adjust BranchTargets within the functions */
    for (i = 0; i < emitInfo->NumSubroutines; i++) {
       struct gl_program *sub = emitInfo->Subroutines[i];
       GLuint j;
@@ -2081,6 +2334,7 @@ _slang_emit_code(slang_ir_node *n, slang_var_table *vt,
    emitInfo.prog = prog;
    emitInfo.Subroutines = NULL;
    emitInfo.NumSubroutines = 0;
+   emitInfo.MaxInstructions = prog->NumInstructions;
 
    emitInfo.EmitHighLevelInstructions = ctx->Shader.EmitHighLevelInstructions;
    emitInfo.EmitCondCodes = ctx->Shader.EmitCondCodes;
diff --git a/src/mesa/shader/slang/slang_ir.c b/src/mesa/shader/slang/slang_ir.c
index 3a0b8bf3a0d..c33c80da999 100644
--- a/src/mesa/shader/slang/slang_ir.c
+++ b/src/mesa/shader/slang/slang_ir.c
@@ -27,6 +27,7 @@
 #include "main/context.h"
 #include "slang_ir.h"
 #include "slang_mem.h"
+#include "shader/prog_instruction.h"
 #include "shader/prog_print.h"
 
 
@@ -36,8 +37,11 @@ static const slang_ir_info IrInfo[] = {
    { IR_SUB, "IR_SUB", OPCODE_SUB, 4, 2 },
    { IR_MUL, "IR_MUL", OPCODE_MUL, 4, 2 },
    { IR_DIV, "IR_DIV", OPCODE_NOP, 0, 2 }, /* XXX broke */
-   { IR_DOT4, "IR_DOT_4", OPCODE_DP4, 1, 2 },
-   { IR_DOT3, "IR_DOT_3", OPCODE_DP3, 1, 2 },
+   { IR_DOT4, "IR_DOT4", OPCODE_DP4, 1, 2 },
+   { IR_DOT3, "IR_DOT3", OPCODE_DP3, 1, 2 },
+   { IR_DOT2, "IR_DOT2", OPCODE_DP2, 1, 2 },
+   { IR_NRM4, "IR_NRM4", OPCODE_NRM4, 1, 1 },
+   { IR_NRM3, "IR_NRM3", OPCODE_NRM3, 1, 1 },
    { IR_CROSS, "IR_CROSS", OPCODE_XPD, 3, 2 },
    { IR_LRP, "IR_LRP", OPCODE_LRP, 4, 3 },
    { IR_MIN, "IR_MIN", OPCODE_MIN, 4, 2 },
@@ -56,7 +60,7 @@ static const slang_ir_info IrInfo[] = {
    /* unary ops */
    { IR_MOVE, "IR_MOVE", OPCODE_MOV, 4, 1 },
    { IR_I_TO_F, "IR_I_TO_F", OPCODE_MOV, 4, 1 },  /* int[4] to float[4] */
-   { IR_F_TO_I, "IR_F_TO_I", OPCODE_INT, 4, 1 }, /* 4 floats to 4 ints */
+   { IR_F_TO_I, "IR_F_TO_I", OPCODE_TRUNC, 4, 1 },
    { IR_EXP, "IR_EXP", OPCODE_EXP, 1, 1 },
    { IR_EXP2, "IR_EXP2", OPCODE_EX2, 1, 1 },
    { IR_LOG2, "IR_LOG2", OPCODE_LG2, 1, 1 },
@@ -112,6 +116,20 @@ _slang_ir_info(slang_ir_opcode opcode)
 }
 
 
+void
+_slang_init_ir_storage(slang_ir_storage *st,
+                       enum register_file file, GLint index, GLint size,
+                       GLuint swizzle)
+{
+   st->File = file;
+   st->Index = index;
+   st->Size = size;
+   st->Swizzle = swizzle;
+   st->Parent = NULL;
+   st->IsIndirect = GL_FALSE;
+}
+
+
 /**
  * Return a new slang_ir_storage object.
  */
@@ -126,6 +144,7 @@ _slang_new_ir_storage(enum register_file file, GLint index, GLint size)
       st->Size = size;
       st->Swizzle = SWIZZLE_NOOP;
       st->Parent = NULL;
+      st->IsIndirect = GL_FALSE;
    }
    return st;
 }
@@ -146,6 +165,7 @@ _slang_new_ir_storage_swz(enum register_file file, GLint index, GLint size,
       st->Size = size;
       st->Swizzle = swizzle;
       st->Parent = NULL;
+      st->IsIndirect = GL_FALSE;
    }
    return st;
 }
@@ -166,11 +186,45 @@ _slang_new_ir_storage_relative(GLint index, GLint size,
       st->Size = size;
       st->Swizzle = SWIZZLE_NOOP;
       st->Parent = parent;
+      st->IsIndirect = GL_FALSE;
+   }
+   return st;
+}
+
+
+slang_ir_storage *
+_slang_new_ir_storage_indirect(enum register_file file,
+                               GLint index,
+                               GLint size,
+                               enum register_file indirectFile,
+                               GLint indirectIndex,
+                               GLuint indirectSwizzle)
+{
+   slang_ir_storage *st;
+   st = (slang_ir_storage *) _slang_alloc(sizeof(slang_ir_storage));
+   if (st) {
+      st->File = file;
+      st->Index = index;
+      st->Size = size;
+      st->Swizzle = SWIZZLE_NOOP;
+      st->IsIndirect = GL_TRUE;
+      st->IndirectFile = indirectFile;
+      st->IndirectIndex = indirectIndex;
+      st->IndirectSwizzle = indirectSwizzle;
    }
    return st;
 }
 
 
+/* XXX temporary function */
+void
+_slang_copy_ir_storage(slang_ir_storage *dst, const slang_ir_storage *src)
+{
+   *dst = *src;
+   dst->Parent = NULL;
+}
+
+
 
 static const char *
 _slang_ir_name(slang_ir_opcode opcode)
@@ -239,22 +293,6 @@ _slang_free_ir_tree(slang_ir_node *n)
 }
 
 
-
-static const char *
-writemask_string(GLuint writemask)
-{
-   static char s[6];
-   GLuint i, j = 0;
-   s[j++] = '.';
-   for (i = 0; i < 4; i++) {
-      if (writemask & (1 << i))
-         s[j++] = "xyzw"[i];
-   }
-   s[j] = 0;
-   return s;
-}
-
-
 static const char *
 storage_string(const slang_ir_storage *st)
 {
@@ -328,7 +366,7 @@ _slang_print_ir_tree(const slang_ir_node *n, int indent)
       _slang_print_ir_tree(n->Children[0], indent + 3);
       break;
    case IR_COPY:
-      printf("COPY (writemask = %s)\n", writemask_string(n->Writemask));
+      printf("COPY\n");
       _slang_print_ir_tree(n->Children[0], indent+3);
       _slang_print_ir_tree(n->Children[1], indent+3);
       break;
diff --git a/src/mesa/shader/slang/slang_ir.h b/src/mesa/shader/slang/slang_ir.h
index f64f9a93b78..a258e92e06f 100644
--- a/src/mesa/shader/slang/slang_ir.h
+++ b/src/mesa/shader/slang/slang_ir.h
@@ -83,6 +83,9 @@ typedef enum
    IR_DIV,
    IR_DOT4,
    IR_DOT3,
+   IR_DOT2,
+   IR_NRM4,
+   IR_NRM3,
    IR_CROSS,   /* vec3 cross product */
    IR_LRP,
    IR_CLAMP,
@@ -137,24 +140,53 @@ typedef enum
 
 
 /**
- * Describes where data storage is allocated.
+ * Describes where data/variables are stored in the various register files.
+ *
+ * In the simple case, the File, Index and Size fields indicate where
+ * a variable is stored.  For example, a vec3 variable may be stored
+ * as (File=PROGRAM_TEMPORARY, Index=6, Size=3).  Or, File[Index].
+ * Or, a program input like color may be stored as
+ * (File=PROGRAM_INPUT,Index=3,Size=4);
+ *
+ * For single-float values, the Swizzle field indicates which component
+ * of the vector contains the float.
+ *
+ * If IsIndirect is set, the storage is accessed through an indirect
+ * register lookup.  The value in question will be located at:
+ *   File[Index + IndirectFile[IndirectIndex]]
+ *
+ * This is primary used for indexing arrays.  For example, consider this
+ * GLSL code:
+ *   uniform int i;
+ *   float a[10];
+ *   float x = a[i];
+ *
+ * here, storage for a[i] would be described by (File=PROGRAM_TEMPORAY,
+ * Index=aPos, IndirectFile=PROGRAM_UNIFORM, IndirectIndex=iPos), which
+ * would mean TEMP[aPos + UNIFORM[iPos]]
  */
-struct _slang_ir_storage
+struct slang_ir_storage_
 {
    enum register_file File;  /**< PROGRAM_TEMPORARY, PROGRAM_INPUT, etc */
    GLint Index;  /**< -1 means unallocated */
    GLint Size;  /**< number of floats */
-   GLuint Swizzle;
+   GLuint Swizzle; /**< Swizzle AND writemask info */
    GLint RefCount; /**< Used during IR tree delete */
-   GLboolean RelAddr;
+
+   GLboolean RelAddr; /* we'll remove this eventually */
+
+   GLboolean IsIndirect;
+   enum register_file IndirectFile;
+   GLint IndirectIndex;
+   GLuint IndirectSwizzle;
 
    /** If Parent is non-null, Index is relative to parent.
     * The other fields are ignored.
     */
-   struct _slang_ir_storage *Parent;
+   struct slang_ir_storage_ *Parent;
 };
 
-typedef struct _slang_ir_storage slang_ir_storage;
+typedef struct slang_ir_storage_ slang_ir_storage;
 
 
 /**
@@ -170,7 +202,6 @@ typedef struct slang_ir_node_
 
    /** special fields depending on Opcode: */
    const char *Field;  /**< If Opcode == IR_FIELD */
-   GLuint Writemask;  /**< If Opcode == IR_MOVE */
    GLfloat Value[4];    /**< If Opcode == IR_FLOAT */
    slang_variable *Var;  /**< If Opcode == IR_VAR or IR_VAR_DECL */
    struct slang_ir_node_ *List;  /**< For various linked lists */
@@ -197,6 +228,11 @@ extern const slang_ir_info *
 _slang_ir_info(slang_ir_opcode opcode);
 
 
+extern void
+_slang_init_ir_storage(slang_ir_storage *st,
+                       enum register_file file, GLint index, GLint size,
+                       GLuint swizzle);
+
 extern slang_ir_storage *
 _slang_new_ir_storage(enum register_file file, GLint index, GLint size);
 
@@ -210,6 +246,17 @@ _slang_new_ir_storage_relative(GLint index, GLint size,
                                slang_ir_storage *parent);
 
 
+extern slang_ir_storage *
+_slang_new_ir_storage_indirect(enum register_file file,
+                               GLint index,
+                               GLint size,
+                               enum register_file indirectFile,
+                               GLint indirectIndex,
+                               GLuint indirectSwizzle);
+
+extern void
+_slang_copy_ir_storage(slang_ir_storage *dst, const slang_ir_storage *src);
+
 
 extern void
 _slang_free_ir_tree(slang_ir_node *n);
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index 4361efc56e5..2cd02d72140 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -76,42 +76,93 @@ link_error(struct gl_shader_program *shProg, const char *msg)
 
 
 /**
+ * Check if the given bit is either set or clear in both bitfields.
+ */
+static GLboolean
+bits_agree(GLbitfield flags1, GLbitfield flags2, GLbitfield bit)
+{
+   return (flags1 & bit) == (flags2 & bit);
+}
+
+
+/**
  * Linking varying vars involves rearranging varying vars so that the
  * vertex program's output varyings matches the order of the fragment
  * program's input varyings.
+ * We'll then rewrite instructions to replace PROGRAM_VARYING with either
+ * PROGRAM_INPUT or PROGRAM_OUTPUT depending on whether it's a vertex or
+ * fragment shader.
+ * This is also where we set program Input/OutputFlags to indicate
+ * which inputs are centroid-sampled, invariant, etc.
  */
 static GLboolean
 link_varying_vars(struct gl_shader_program *shProg, struct gl_program *prog)
 {
    GLuint *map, i, firstVarying, newFile;
+   GLbitfield *inOutFlags;
 
    map = (GLuint *) malloc(prog->Varying->NumParameters * sizeof(GLuint));
    if (!map)
       return GL_FALSE;
 
+   /* Varying variables are treated like other vertex program outputs
+    * (and like other fragment program inputs).  The position of the
+    * first varying differs for vertex/fragment programs...
+    * Also, replace File=PROGRAM_VARYING with File=PROGRAM_INPUT/OUTPUT.
+    */
+   if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
+      firstVarying = VERT_RESULT_VAR0;
+      newFile = PROGRAM_OUTPUT;
+      inOutFlags = prog->OutputFlags;
+   }
+   else {
+      assert(prog->Target == GL_FRAGMENT_PROGRAM_ARB);
+      firstVarying = FRAG_ATTRIB_VAR0;
+      newFile = PROGRAM_INPUT;
+      inOutFlags = prog->InputFlags;
+   }
+
    for (i = 0; i < prog->Varying->NumParameters; i++) {
       /* see if this varying is in the linked varying list */
       const struct gl_program_parameter *var = prog->Varying->Parameters + i;
       GLint j = _mesa_lookup_parameter_index(shProg->Varying, -1, var->Name);
       if (j >= 0) {
-         /* already in list, check size */
-         if (var->Size != shProg->Varying->Parameters[j].Size) {
-            /* error */
+         /* varying is already in list, do some error checking */
+         const struct gl_program_parameter *v =
+            &shProg->Varying->Parameters[j];
+         if (var->Size != v->Size) {
             link_error(shProg, "mismatched varying variable types");
             return GL_FALSE;
          }
+         if (!bits_agree(var->Flags, v->Flags, PROG_PARAM_BIT_CENTROID)) {
+            char msg[100];
+            snprintf(msg, sizeof(msg),
+                     "centroid modifier mismatch for '%s'", var->Name);
+            link_error(shProg, msg);
+            return GL_FALSE;
+         }
+         if (!bits_agree(var->Flags, v->Flags, PROG_PARAM_BIT_INVARIANT)) {
+            char msg[100];
+            snprintf(msg, sizeof(msg),
+                     "invariant modifier mismatch for '%s'", var->Name);
+            link_error(shProg, msg);
+            return GL_FALSE;
+         }
       }
       else {
          /* not already in linked list */
-         j = _mesa_add_varying(shProg->Varying, var->Name, var->Size);
+         j = _mesa_add_varying(shProg->Varying, var->Name, var->Size,
+                               var->Flags);
       }
 
-      /* map varying[i] to varying[j].
+      /* Map varying[i] to varying[j].
+       * Plus, set prog->Input/OutputFlags[] as described above.
        * Note: the loop here takes care of arrays or large (sz>4) vars.
        */
       {
          GLint sz = var->Size;
          while (sz > 0) {
+            inOutFlags[firstVarying + j] = var->Flags;
             /*printf("Link varying from %d to %d\n", i, j);*/
             map[i++] = j++;
             sz -= 4;
@@ -121,21 +172,6 @@ link_varying_vars(struct gl_shader_program *shProg, struct gl_program *prog)
    }
 
 
-   /* Varying variables are treated like other vertex program outputs
-    * (and like other fragment program inputs).  The position of the
-    * first varying differs for vertex/fragment programs...
-    * Also, replace File=PROGRAM_VARYING with File=PROGRAM_INPUT/OUTPUT.
-    */
-   if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
-      firstVarying = VERT_RESULT_VAR0;
-      newFile = PROGRAM_OUTPUT;
-   }
-   else {
-      assert(prog->Target == GL_FRAGMENT_PROGRAM_ARB);
-      firstVarying = FRAG_ATTRIB_VAR0;
-      newFile = PROGRAM_INPUT;
-   }
-
    /* OK, now scan the program/shader instructions looking for varying vars,
     * replacing the old index with the new index.
     */
@@ -193,7 +229,10 @@ link_uniform_vars(struct gl_shader_program *shProg,
 
       if ((p->Type == PROGRAM_UNIFORM && p->Used) ||
           p->Type == PROGRAM_SAMPLER) {
-         _mesa_append_uniform(shProg->Uniforms, p->Name, prog->Target, i);
+         struct gl_uniform *uniform =
+            _mesa_append_uniform(shProg->Uniforms, p->Name, prog->Target, i);
+         if (uniform)
+            uniform->Initialized = p->Initialized;
       }
 
       if (p->Type == PROGRAM_SAMPLER) {
@@ -419,7 +458,6 @@ _slang_update_inputs_outputs(struct gl_program *prog)
          maxAddrReg = MAX2(maxAddrReg, inst->DstReg.Index + 1);
       }
    }
-
    prog->NumAddressRegs = maxAddrReg;
 }
 
@@ -602,6 +640,11 @@ _slang_link(GLcontext *ctx,
       }
    }
 
+   if (MESA_VERBOSE & VERBOSE_GLSL_DUMP) {
+      printf("Varying vars:\n");
+      _mesa_print_parameter_list(shProg->Varying);
+   }
+
    shProg->LinkStatus = (shProg->VertexProgram || shProg->FragmentProgram);
 }
 
diff --git a/src/mesa/shader/slang/slang_print.c b/src/mesa/shader/slang/slang_print.c
index b88cefc283b..1194d0b4b99 100644
--- a/src/mesa/shader/slang/slang_print.c
+++ b/src/mesa/shader/slang/slang_print.c
@@ -653,6 +653,11 @@ slang_print_tree(const slang_operation *op, int indent)
       printf(")\n");
       break;
 
+   case SLANG_OPER_METHOD:
+      spaces(indent);
+      printf("METHOD CALL %s.%s\n", (char *) op->a_obj, (char *) op->a_id);
+      break;
+
    case SLANG_OPER_FIELD:
       spaces(indent);
       printf("FIELD %s of\n", (char*) op->a_id);
diff --git a/src/mesa/shader/slang/slang_typeinfo.c b/src/mesa/shader/slang/slang_typeinfo.c
index b2f0f149956..b1afd969d95 100644
--- a/src/mesa/shader/slang/slang_typeinfo.c
+++ b/src/mesa/shader/slang/slang_typeinfo.c
@@ -630,6 +630,12 @@ _slang_typeof_operation_(slang_operation * op,
          }
       }
       break;
+   case SLANG_OPER_METHOD:
+      /* at this time, GLSL 1.20 only has one method: array.length()
+       * which returns an integer.
+       */
+      ti->spec.type = SLANG_SPEC_INT;
+      break;
    case SLANG_OPER_FIELD:
       {
          slang_typeinfo _ti;
diff --git a/src/mesa/shader/slang/slang_vartable.c b/src/mesa/shader/slang/slang_vartable.c
index 95971a70a93..de0c93957b4 100644
--- a/src/mesa/shader/slang/slang_vartable.c
+++ b/src/mesa/shader/slang/slang_vartable.c
@@ -107,7 +107,7 @@ _slang_pop_var_table(slang_var_table *vt)
 
    /* free the storage allocated for each variable */
    for (i = 0; i < t->NumVars; i++) {
-      slang_ir_storage *store = (slang_ir_storage *) t->Vars[i]->aux;
+      slang_ir_storage *store = t->Vars[i]->store;
       GLint j;
       GLuint comp;
       if (dbg) printf("  Free var %s, size %d at %d.%s\n",
@@ -165,7 +165,7 @@ _slang_add_variable(slang_var_table *vt, slang_variable *v)
    assert(vt);
    t = vt->Top;
    assert(t);
-   if (dbg) printf("Adding var %s, store %p\n", (char *) v->a_name, v->aux);
+   if (dbg) printf("Adding var %s, store %p\n", (char *) v->a_name, (void *) v->store);
    t->Vars = (slang_variable **)
       _slang_realloc(t->Vars,
                      t->NumVars * sizeof(slang_variable *),
diff --git a/src/mesa/shader/slang/slang_vartable.h b/src/mesa/shader/slang/slang_vartable.h
index 8a3b992c969..94bcd63f45a 100644
--- a/src/mesa/shader/slang/slang_vartable.h
+++ b/src/mesa/shader/slang/slang_vartable.h
@@ -2,7 +2,7 @@
 #ifndef SLANG_VARTABLE_H
 #define SLANG_VARTABLE_H
 
-struct _slang_ir_storage;
+struct slang_ir_storage_;
 
 typedef struct slang_var_table_ slang_var_table;
 
@@ -27,16 +27,16 @@ extern struct slang_variable_ *
 _slang_find_variable(const slang_var_table *t, slang_atom name);
 
 extern GLboolean
-_slang_alloc_var(slang_var_table *t, struct _slang_ir_storage *store);
+_slang_alloc_var(slang_var_table *t, struct slang_ir_storage_ *store);
 
 extern GLboolean
-_slang_alloc_temp(slang_var_table *t, struct _slang_ir_storage *store);
+_slang_alloc_temp(slang_var_table *t, struct slang_ir_storage_ *store);
 
 extern void
-_slang_free_temp(slang_var_table *t, struct _slang_ir_storage *store);
+_slang_free_temp(slang_var_table *t, struct slang_ir_storage_ *store);
 
 extern GLboolean
-_slang_is_temp(const slang_var_table *t, const struct _slang_ir_storage *store);
+_slang_is_temp(const slang_var_table *t, const struct slang_ir_storage_ *store);
 
 
 #endif /* SLANG_VARTABLE_H */
diff --git a/src/mesa/shader/slang/sources b/src/mesa/shader/slang/sources
deleted file mode 100644
index 00d617fa8af..00000000000
--- a/src/mesa/shader/slang/sources
+++ /dev/null
@@ -1,44 +0,0 @@
-MESA_SHADER_SLANG_SOURCES = \
-slang_analyse.c \
-slang_assemble_assignment.c \
-slang_assemble.c \
-slang_assemble_conditional.c \
-slang_assemble_constructor.c \
-slang_assemble_typeinfo.c \
-slang_compile.c \
-slang_compile_function.c \
-slang_compile_operation.c \
-slang_compile_struct.c \
-slang_compile_variable.c \
-slang_execute.c \
-slang_execute_x86.c \
-slang_export.c \
-slang_library_texsample.c \
-slang_library_noise.c \
-slang_link.c \
-slang_preprocess.c \
-slang_storage.c \
-slang_utility.c
-
-MESA_SHADER_SLANG_HEADERS = \
-slang_analyse.h \
-slang_assemble.h \
-slang_assemble_assignment.h \
-slang_assemble_conditional.h \
-slang_assemble_constructor.h \
-slang_assemble_typeinfo.h \
-slang_compile.h \
-slang_compile_function.h \
-slang_compile_operation.h \
-slang_compile_struct.h \
-slang_compile_variable.h \
-slang_execute.h \
-slang_export.h \
-slang_library_noise.h \
-slang_library_texsample.h \
-slang_link.h \
-slang_mesa.h \
-slang_preprocess.h \
-slang_storage.h \
-slang_utility.h \
-traverse_wrap.h
diff --git a/src/mesa/shader/sources b/src/mesa/shader/sources
deleted file mode 100644
index 27871872766..00000000000
--- a/src/mesa/shader/sources
+++ /dev/null
@@ -1,28 +0,0 @@
-# List of source files in this directory used for X.org xserver build
-MESA_SHADER_SOURCES = \
-arbprogparse.c \
-arbprogram.c \
-atifragshader.c \
-nvfragparse.c \
-nvprogram.c \
-nvvertexec.c \
-nvvertparse.c \
-program.c \
-programopt.c \
-shaderobjects.c \
-shaderobjects_3dlabs.c
-
-MESA_SHADER_HEADERS = \
-arbprogparse.h \
-arbprogram.h \
-arbprogram_syn.h \
-atifragshader.h \
-nvfragparse.h \
-nvprogram.h \
-nvvertexec.h \
-nvvertparse.h \
-programopt.h \
-program.h \
-program_instruction.h \
-shaderobjects.h \
-shaderobjects_3dlabs.h
diff --git a/src/mesa/sources b/src/mesa/sources
new file mode 100644
index 00000000000..9a34d1a989f
--- /dev/null
+++ b/src/mesa/sources
@@ -0,0 +1,302 @@
+### Lists of source files, included by Makefiles
+
+MAIN_SOURCES = \
+	main/api_arrayelt.c \
+	main/api_exec.c \
+	main/api_loopback.c \
+	main/api_noop.c \
+	main/api_validate.c \
+	main/accum.c \
+	main/attrib.c \
+	main/arrayobj.c \
+	main/blend.c \
+	main/bufferobj.c \
+	main/buffers.c \
+	main/clear.c \
+	main/clip.c \
+	main/colortab.c \
+	main/context.c \
+	main/convolve.c \
+	main/debug.c \
+	main/depth.c \
+	main/depthstencil.c \
+	main/dlist.c \
+	main/drawpix.c \
+	main/enable.c \
+	main/enums.c \
+	main/eval.c \
+	main/execmem.c \
+	main/extensions.c \
+	main/fbobject.c \
+	main/feedback.c \
+	main/ffvertex_prog.c \
+	main/fog.c \
+	main/framebuffer.c \
+	main/get.c \
+	main/getstring.c \
+	main/hash.c \
+	main/hint.c \
+	main/histogram.c \
+	main/image.c \
+	main/imports.c \
+	main/light.c \
+	main/lines.c \
+	main/matrix.c \
+	main/mipmap.c \
+	main/mm.c \
+	main/multisample.c \
+	main/pixel.c \
+	main/pixelstore.c \
+	main/points.c \
+	main/polygon.c \
+	main/queryobj.c \
+	main/rastpos.c \
+	main/rbadaptors.c \
+	main/readpix.c \
+	main/renderbuffer.c \
+	main/scissor.c \
+	main/shaders.c \
+	main/state.c \
+	main/stencil.c \
+	main/texcompress.c \
+	main/texcompress_s3tc.c \
+	main/texcompress_fxt1.c \
+	main/texenv.c \
+	main/texenvprogram.c \
+	main/texformat.c \
+	main/texgen.c \
+	main/teximage.c \
+	main/texobj.c \
+	main/texparam.c \
+	main/texrender.c \
+	main/texstate.c \
+	main/texstore.c \
+	main/varray.c \
+	main/vtxfmt.c
+
+GLAPI_SOURCES = \
+	main/dispatch.c \
+	glapi/glapi.c \
+	glapi/glapi_getproc.c \
+	glapi/glthread.c
+
+MATH_SOURCES = \
+	math/m_debug_clip.c \
+	math/m_debug_norm.c \
+	math/m_debug_xform.c \
+	math/m_eval.c \
+	math/m_matrix.c \
+	math/m_translate.c \
+	math/m_vector.c \
+	math/m_xform.c
+
+SWRAST_SOURCES = \
+	swrast/s_aaline.c \
+	swrast/s_aatriangle.c \
+	swrast/s_accum.c \
+	swrast/s_alpha.c \
+	swrast/s_atifragshader.c \
+	swrast/s_bitmap.c \
+	swrast/s_blend.c \
+	swrast/s_blit.c \
+	swrast/s_buffers.c \
+	swrast/s_copypix.c \
+	swrast/s_context.c \
+	swrast/s_depth.c \
+	swrast/s_drawpix.c \
+	swrast/s_feedback.c \
+	swrast/s_fog.c \
+	swrast/s_fragprog.c \
+	swrast/s_imaging.c \
+	swrast/s_lines.c \
+	swrast/s_logic.c \
+	swrast/s_masking.c \
+	swrast/s_points.c \
+	swrast/s_readpix.c \
+	swrast/s_span.c \
+	swrast/s_stencil.c \
+	swrast/s_texcombine.c \
+	swrast/s_texfilter.c \
+	swrast/s_texstore.c \
+	swrast/s_triangle.c \
+	swrast/s_zoom.c
+
+SWRAST_SETUP_SOURCES = \
+	swrast_setup/ss_context.c \
+	swrast_setup/ss_triangle.c 
+
+TNL_SOURCES = \
+	tnl/t_context.c \
+	tnl/t_pipeline.c \
+	tnl/t_draw.c \
+	tnl/t_rasterpos.c \
+	tnl/t_vb_program.c \
+	tnl/t_vb_render.c \
+	tnl/t_vb_texgen.c \
+	tnl/t_vb_texmat.c \
+	tnl/t_vb_vertex.c \
+	tnl/t_vb_cull.c \
+	tnl/t_vb_fog.c \
+	tnl/t_vb_light.c \
+	tnl/t_vb_normals.c \
+	tnl/t_vb_points.c \
+	tnl/t_vp_build.c \
+	tnl/t_vertex.c \
+	tnl/t_vertex_sse.c \
+	tnl/t_vertex_generic.c 
+
+VBO_SOURCES = \
+	vbo/vbo_context.c \
+	vbo/vbo_exec.c \
+	vbo/vbo_exec_api.c \
+	vbo/vbo_exec_array.c \
+	vbo/vbo_exec_draw.c \
+	vbo/vbo_exec_eval.c \
+	vbo/vbo_rebase.c \
+	vbo/vbo_split.c \
+	vbo/vbo_split_copy.c \
+	vbo/vbo_split_inplace.c \
+	vbo/vbo_save.c \
+	vbo/vbo_save_api.c \
+	vbo/vbo_save_draw.c \
+	vbo/vbo_save_loopback.c 
+
+
+SHADER_SOURCES = \
+	shader/arbprogparse.c \
+	shader/arbprogram.c \
+	shader/atifragshader.c \
+	shader/grammar/grammar_mesa.c \
+	shader/nvfragparse.c \
+	shader/nvprogram.c \
+	shader/nvvertparse.c \
+	shader/program.c \
+	shader/prog_cache.c \
+	shader/prog_debug.c \
+	shader/prog_execute.c \
+	shader/prog_instruction.c \
+	shader/prog_parameter.c \
+	shader/prog_print.c \
+	shader/prog_statevars.c \
+	shader/prog_uniform.c \
+	shader/programopt.c \
+	shader/shader_api.c \
+
+SLANG_SOURCES =	\
+	shader/slang/slang_builtin.c	\
+	shader/slang/slang_codegen.c	\
+	shader/slang/slang_compile.c	\
+	shader/slang/slang_compile_function.c	\
+	shader/slang/slang_compile_operation.c	\
+	shader/slang/slang_compile_struct.c	\
+	shader/slang/slang_compile_variable.c	\
+	shader/slang/slang_emit.c	\
+	shader/slang/slang_ir.c	\
+	shader/slang/slang_label.c	\
+	shader/slang/slang_library_noise.c	\
+	shader/slang/slang_link.c	\
+	shader/slang/slang_log.c	\
+	shader/slang/slang_mem.c	\
+	shader/slang/slang_preprocess.c	\
+	shader/slang/slang_print.c	\
+	shader/slang/slang_simplify.c	\
+	shader/slang/slang_storage.c	\
+	shader/slang/slang_typeinfo.c	\
+	shader/slang/slang_vartable.c	\
+	shader/slang/slang_utility.c
+
+ASM_C_SOURCES =	\
+	x86/common_x86.c \
+	x86/x86.c \
+	x86/3dnow.c \
+	x86/sse.c \
+	x86/rtasm/x86sse.c \
+	sparc/sparc.c \
+	ppc/common_ppc.c \
+	x86-64/x86-64.c
+
+X86_SOURCES =			\
+	x86/common_x86_asm.S	\
+	x86/x86_xform2.S	\
+	x86/x86_xform3.S	\
+	x86/x86_xform4.S	\
+	x86/x86_cliptest.S	\
+	x86/mmx_blend.S		\
+	x86/3dnow_xform1.S	\
+	x86/3dnow_xform2.S	\
+	x86/3dnow_xform3.S	\
+	x86/3dnow_xform4.S	\
+	x86/3dnow_normal.S	\
+	x86/sse_xform1.S	\
+	x86/sse_xform2.S	\
+	x86/sse_xform3.S	\
+	x86/sse_xform4.S	\
+	x86/sse_normal.S	\
+	x86/read_rgba_span_x86.S
+
+X86_API =			\
+	x86/glapi_x86.S
+
+X86-64_SOURCES =		\
+	x86-64/xform4.S
+
+X86-64_API =			\
+	x86-64/glapi_x86-64.S
+
+SPARC_SOURCES =			\
+	sparc/clip.S		\
+	sparc/norm.S		\
+	sparc/xform.S
+
+SPARC_API =			\
+	sparc/glapi_sparc.S
+
+COMMON_DRIVER_SOURCES =			\
+	drivers/common/driverfuncs.c
+
+
+
+### All the core C sources
+
+MESA_SOURCES = \
+	$(MAIN_SOURCES)		\
+	$(MATH_SOURCES)		\
+	$(VBO_SOURCES)		\
+	$(TNL_SOURCES)		\
+	$(SHADER_SOURCES)	\
+	$(SWRAST_SOURCES)	\
+	$(SWRAST_SETUP_SOURCES)	\
+	$(COMMON_DRIVER_SOURCES) \
+	$(ASM_C_SOURCES)	\
+	$(SLANG_SOURCES)
+
+ALL_SOURCES = \
+	$(MESA_SOURCES)		\
+	$(GLAPI_SOURCES)	\
+	$(MESA_ASM_SOURCES)		\
+	$(COMMON_DRIVER_SOURCES)
+
+
+### Object files
+
+MESA_OBJECTS = \
+	$(MESA_SOURCES:.c=.o) \
+	$(MESA_ASM_SOURCES:.S=.o)
+
+GLAPI_OBJECTS = \
+	$(GLAPI_SOURCES:.c=.o) \
+	$(GLAPI_ASM_SOURCES:.S=.o)
+
+
+COMMON_DRIVER_OBJECTS = $(COMMON_DRIVER_SOURCES:.c=.o)
+
+
+
+### Include directories
+### XXX we should be able to trim this down to just -Iinclude/ and -Isrc/mesa/
+### since most #includes have been changed from #include "mtypes.h" to
+### #include "main/mtypes.h", etc.
+
+INCLUDE_DIRS = \
+	-I$(TOP)/include \
+	-I$(TOP)/src/mesa
diff --git a/src/mesa/state_tracker/acc2.c b/src/mesa/state_tracker/acc2.c
deleted file mode 100644
index fa5de2b764c..00000000000
--- a/src/mesa/state_tracker/acc2.c
+++ /dev/null
@@ -1,319 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
- /*
-  * Authors:
-  *   Brian Paul
-  */
-
-#include "main/imports.h"
-#include "main/image.h"
-#include "main/macros.h"
-
-#include "st_context.h"
-#include "st_cb_accum.h"
-#include "st_cb_fbo.h"
-#include "st_draw.h"
-#include "st_format.h"
-#include "pipe/p_context.h"
-#include "pipe/p_defines.h"
-#include "pipe/p_inlines.h"
-#include "util/p_tile.h"
-
-
-#define UNCLAMPED_FLOAT_TO_SHORT(us, f)  \
-   us = ( (short) ( CLAMP((f), -1.0, 1.0) * 32767.0F) )
-
-
-/**
- * For hardware that supports deep color buffers, we could accelerate
- * most/all the accum operations with blending/texturing.
- * For now, just use the get/put_tile() functions and do things in software.
- */
-
-
-static void
-acc_get_tile_rgba(struct pipe_context *pipe, struct pipe_surface *acc_ps,
-                  uint x, uint y, uint w, uint h, float *p)
-{
-   const enum pipe_format f = acc_ps->format;
-   const int cpp = acc_ps->cpp;
-
-   acc_ps->format = PIPE_FORMAT_R16G16B16A16_SNORM;
-   acc_ps->cpp = 8;
-
-   pipe_get_tile_rgba(pipe, acc_ps, x, y, w, h, p);
-
-   acc_ps->format = f;
-   acc_ps->cpp = cpp;
-}
-
-
-static void
-acc_put_tile_rgba(struct pipe_context *pipe, struct pipe_surface *acc_ps,
-                  uint x, uint y, uint w, uint h, const float *p)
-{
-   enum pipe_format f = acc_ps->format;
-   const int cpp = acc_ps->cpp;
-
-   acc_ps->format = PIPE_FORMAT_R16G16B16A16_SNORM;
-   acc_ps->cpp = 8;
-
-   pipe_put_tile_rgba(pipe, acc_ps, x, y, w, h, p);
-
-   acc_ps->format = f;
-   acc_ps->cpp = cpp;
-}
-
-
-
-void
-st_clear_accum_buffer(GLcontext *ctx, struct gl_renderbuffer *rb)
-{
-   struct pipe_context *pipe = ctx->st->pipe;
-   struct st_renderbuffer *acc_strb = st_renderbuffer(rb);
-   struct pipe_surface *acc_ps = acc_strb->surface;
-   const GLint xpos = ctx->DrawBuffer->_Xmin;
-   const GLint ypos = ctx->DrawBuffer->_Ymin;
-   const GLint width = ctx->DrawBuffer->_Xmax - xpos;
-   const GLint height = ctx->DrawBuffer->_Ymax - ypos;
-   const GLfloat r = ctx->Accum.ClearColor[0];
-   const GLfloat g = ctx->Accum.ClearColor[1];
-   const GLfloat b = ctx->Accum.ClearColor[2];
-   const GLfloat a = ctx->Accum.ClearColor[3];
-   GLfloat *accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-   int i;
-
-#if 1
-   GLvoid *map;
-
-   map = pipe_surface_map(acc_ps);
-   switch (acc_strb->format) {
-   case PIPE_FORMAT_R16G16B16A16_SNORM:
-      {
-         GLshort r = FLOAT_TO_SHORT(ctx->Accum.ClearColor[0]);
-         GLshort g = FLOAT_TO_SHORT(ctx->Accum.ClearColor[1]);
-         GLshort b = FLOAT_TO_SHORT(ctx->Accum.ClearColor[2]);
-         GLshort a = FLOAT_TO_SHORT(ctx->Accum.ClearColor[3]);
-         int i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *dst = ((GLshort *) map
-                            + ((ypos + i) * acc_ps->pitch + xpos) * 4);
-            for (j = 0; j < width; j++) {
-               dst[0] = r;
-               dst[1] = g;
-               dst[2] = b;
-               dst[3] = a;
-               dst += 4;
-            }
-         }
-      }
-      break;
-   default:
-      _mesa_problem(ctx, "unexpected format in st_clear_accum_buffer()");
-   }
-
-   pipe_surface_unmap(acc_ps);
-
-#else
-   for (i = 0; i < width * height; i++) {
-      accBuf[i*4+0] = r;
-      accBuf[i*4+1] = g;
-      accBuf[i*4+2] = b;
-      accBuf[i*4+3] = a;
-   }
-
-   acc_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf);
-#endif
-}
-
-
-/** For ADD/MULT */
-static void
-accum_mad(struct pipe_context *pipe, GLfloat scale, GLfloat bias,
-          GLint xpos, GLint ypos, GLint width, GLint height,
-          struct pipe_surface *acc_ps)
-{
-   GLfloat *accBuf;
-   GLint i;
-
-   accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   pipe_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf);
-
-   for (i = 0; i < 4 * width * height; i++) {
-      accBuf[i] = accBuf[i] * scale + bias;
-   }
-
-   pipe_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf);
-
-   free(accBuf);
-}
-
-
-static void
-accum_accum(struct pipe_context *pipe, GLfloat value,
-            GLint xpos, GLint ypos, GLint width, GLint height,
-            struct pipe_surface *acc_ps,
-            struct pipe_surface *color_ps)
-{
-   GLfloat *colorBuf, *accBuf;
-   GLint i;
-
-   colorBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-   accBuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   pipe_get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, colorBuf);
-   acc_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf);
-
-   for (i = 0; i < 4 * width * height; i++) {
-      accBuf[i] = accBuf[i] + colorBuf[i] * value;
-   }
-
-   acc_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, accBuf);
-
-   free(colorBuf);
-   free(accBuf);
-}
-
-
-static void
-accum_load(struct pipe_context *pipe, GLfloat value,
-           GLint xpos, GLint ypos, GLint width, GLint height,
-           struct pipe_surface *acc_ps,
-           struct pipe_surface *color_ps)
-{
-   GLfloat *buf;
-   GLint i;
-
-   buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   pipe_get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, buf);
-
-   for (i = 0; i < 4 * width * height; i++) {
-      buf[i] = buf[i] * value;
-   }
-
-   acc_put_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, buf);
-
-   free(buf);
-}
-
-
-static void
-accum_return(GLcontext *ctx, GLfloat value,
-             GLint xpos, GLint ypos, GLint width, GLint height,
-             struct pipe_surface *acc_ps,
-             struct pipe_surface *color_ps)
-{
-   struct pipe_context *pipe = ctx->st->pipe;
-   const GLubyte *colormask = ctx->Color.ColorMask;
-   GLfloat *abuf, *cbuf = NULL;
-   GLint i, ch;
-
-   abuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-
-   acc_get_tile_rgba(pipe, acc_ps, xpos, ypos, width, height, abuf);
-
-   if (!colormask[0] || !colormask[1] || !colormask[2] || !colormask[3]) {
-      cbuf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat));
-      pipe_get_tile_rgba(pipe, color_ps, xpos, ypos, width, height, cbuf);
-   }
-
-   for (i = 0; i < width * height; i++) {
-      for (ch = 0; ch < 4; ch++) {
-         if (colormask[ch]) {
-            GLfloat val = abuf[i * 4 + ch] * value;
-            abuf[i * 4 + ch] = CLAMP(val, 0.0, 1.0);
-         }
-         else {
-            abuf[i * 4 + ch] = cbuf[i * 4 + ch];
-         }
-      }
-   }
-
-   pipe_put_tile_rgba(pipe, color_ps, xpos, ypos, width, height, abuf);
-
-   free(abuf);
-   if (cbuf)
-      free(cbuf);
-}
-
-
-static void
-st_Accum(GLcontext *ctx, GLenum op, GLfloat value)
-{
-   struct st_context *st = ctx->st;
-   struct pipe_context *pipe = st->pipe;
-   struct st_renderbuffer *acc_strb
-     = st_renderbuffer(ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer);
-   struct st_renderbuffer *color_strb
-      = st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
-   struct pipe_surface *acc_ps = acc_strb->surface;
-   struct pipe_surface *color_ps = color_strb->surface;
-
-   const GLint xpos = ctx->DrawBuffer->_Xmin;
-   const GLint ypos = ctx->DrawBuffer->_Ymin;
-   const GLint width = ctx->DrawBuffer->_Xmax - xpos;
-   const GLint height = ctx->DrawBuffer->_Ymax - ypos;
-
-   /* make sure color bufs aren't cached */
-   pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL);
-
-   switch (op) {
-   case GL_ADD:
-      if (value != 0.0F) {
-         accum_mad(pipe, 1.0, value, xpos, ypos, width, height, acc_ps);
-      }
-      break;
-   case GL_MULT:
-      if (value != 1.0F) {
-         accum_mad(pipe, value, 0.0, xpos, ypos, width, height, acc_ps);
-      }
-      break;
-   case GL_ACCUM:
-      if (value != 0.0F) {
-         accum_accum(pipe, value, xpos, ypos, width, height, acc_ps, color_ps);
-      }
-      break;
-   case GL_LOAD:
-      accum_load(pipe, value, xpos, ypos, width, height, acc_ps, color_ps);
-      break;
-   case GL_RETURN:
-      accum_return(ctx, value, xpos, ypos, width, height, acc_ps, color_ps);
-      break;
-   default:
-      assert(0);
-   }
-}
-
-
-
-void st_init_accum_functions(struct dd_function_table *functions)
-{
-   functions->Accum = st_Accum;
-}
diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c
index 2916886610d..ca1a719a9ac 100644
--- a/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -95,7 +95,7 @@ update_framebuffer_state( struct st_context *st )
    struct pipe_framebuffer_state *framebuffer = &st->state.framebuffer;
    struct gl_framebuffer *fb = st->ctx->DrawBuffer;
    struct st_renderbuffer *strb;
-   GLuint i, j;
+   GLuint i;
 
    memset(framebuffer, 0, sizeof(*framebuffer));
 
@@ -108,20 +108,18 @@ update_framebuffer_state( struct st_context *st )
     * to determine which surfaces to draw to
     */
    framebuffer->num_cbufs = 0;
-   for (j = 0; j < MAX_DRAW_BUFFERS; j++) {
-      for (i = 0; i < fb->_NumColorDrawBuffers[j]; i++) {
-         strb = st_renderbuffer(fb->_ColorDrawBuffers[j][i]);
-
-         /*printf("--------- framebuffer surface rtt %p\n", strb->rtt);*/
-         if (strb->rtt) {
-            /* rendering to a GL texture, may have to update surface */
-            update_renderbuffer_surface(st, strb);
-         }
+   for (i = 0; i < fb->_NumColorDrawBuffers; i++) {
+      strb = st_renderbuffer(fb->_ColorDrawBuffers[i]);
 
-         if (strb->surface) {
-            framebuffer->cbufs[framebuffer->num_cbufs] = strb->surface;
-            framebuffer->num_cbufs++;
-         }
+      /*printf("--------- framebuffer surface rtt %p\n", strb->rtt);*/
+      if (strb->rtt) {
+         /* rendering to a GL texture, may have to update surface */
+         update_renderbuffer_surface(st, strb);
+      }
+
+      if (strb->surface) {
+         framebuffer->cbufs[framebuffer->num_cbufs] = strb->surface;
+         framebuffer->num_cbufs++;
       }
    }
 
@@ -146,7 +144,7 @@ update_framebuffer_state( struct st_context *st )
 
    cso_set_framebuffer(st->cso_context, framebuffer);
 
-   if (fb->_ColorDrawBufferMask[0] & BUFFER_BIT_FRONT_LEFT) {
+   if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
       if (st->frontbuffer_status == FRONT_STATUS_COPY_OF_BACK) {
          /* XXX copy back buf to front? */
       }
diff --git a/src/mesa/state_tracker/st_atom_sampler.c b/src/mesa/state_tracker/st_atom_sampler.c
index cef61fb55c5..d7b904354f2 100644
--- a/src/mesa/state_tracker/st_atom_sampler.c
+++ b/src/mesa/state_tracker/st_atom_sampler.c
@@ -121,24 +121,30 @@ gl_filter_to_img_filter(GLenum filter)
 static void 
 update_samplers(struct st_context *st)
 {
-   const struct st_fragment_program *fs = st->fp;
+   struct gl_vertex_program *vprog = st->ctx->VertexProgram._Current;
+   struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current;
+   const GLbitfield samplersUsed = (vprog->Base.SamplersUsed |
+                                    fprog->Base.SamplersUsed);
    GLuint su;
 
    st->state.num_samplers = 0;
 
-   /*printf("%s samplers used = 0x%x\n", __FUNCTION__, fs->Base.Base.SamplersUsed);*/
-
    /* loop over sampler units (aka tex image units) */
    for (su = 0; su < st->ctx->Const.MaxTextureImageUnits; su++) {
       struct pipe_sampler_state *sampler = st->state.samplers + su;
 
       memset(sampler, 0, sizeof(*sampler));
 
-      if (fs->Base.Base.SamplersUsed & (1 << su)) {
-         GLuint texUnit = fs->Base.Base.SamplerUnits[su];
-         const struct gl_texture_object *texobj
-            = st->ctx->Texture.Unit[texUnit]._Current;
+      if (samplersUsed & (1 << su)) {
+         struct gl_texture_object *texobj;
+         GLuint texUnit;
+
+         if (fprog->Base.SamplersUsed & (1 << su))
+            texUnit = fprog->Base.SamplerUnits[su];
+         else
+            texUnit = vprog->Base.SamplerUnits[su];
 
+         texobj = st->ctx->Texture.Unit[texUnit]._Current;
          if (!texobj) {
             texobj = st_get_default_texture(st);
          }
diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c
index fb03766ff59..18ffc08c6b1 100644
--- a/src/mesa/state_tracker/st_atom_texture.c
+++ b/src/mesa/state_tracker/st_atom_texture.c
@@ -44,22 +44,30 @@
 static void 
 update_textures(struct st_context *st)
 {
+   struct gl_vertex_program *vprog = st->ctx->VertexProgram._Current;
    struct gl_fragment_program *fprog = st->ctx->FragmentProgram._Current;
+   const GLbitfield samplersUsed = (vprog->Base.SamplersUsed |
+                                    fprog->Base.SamplersUsed);
    GLuint su;
 
    st->state.num_textures = 0;
 
-   /*printf("%s samplers used = 0x%x\n", __FUNCTION__, fprog->Base.SamplersUsed);*/
-
+   /* loop over sampler units (aka tex image units) */
    for (su = 0; su < st->ctx->Const.MaxTextureCoordUnits; su++) {
       struct pipe_texture *pt = NULL;
 
-      if (fprog->Base.SamplersUsed & (1 << su)) {
-         const GLuint texUnit = fprog->Base.SamplerUnits[su];
-         struct gl_texture_object *texObj
-            = st->ctx->Texture.Unit[texUnit]._Current;
+      if (samplersUsed & (1 << su)) {
+         struct gl_texture_object *texObj;
          struct st_texture_object *stObj;
          GLboolean flush, retval;
+         GLuint texUnit;
+
+         if (fprog->Base.SamplersUsed & (1 << su))
+            texUnit = fprog->Base.SamplerUnits[su];
+         else
+            texUnit = vprog->Base.SamplerUnits[su];
+
+         texObj = st->ctx->Texture.Unit[texUnit]._Current;
 
          if (!texObj) {
             texObj = st_get_default_texture(st);
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index bc05ca6a2ff..f14e562400b 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -821,7 +821,7 @@ st_destroy_bitmap(struct st_context *st)
    }
 
    if (st->bitmap.vbuf) {
-      pipe_buffer_destroy(pipe->screen, st->bitmap.vbuf);
+      pipe_buffer_reference(pipe->screen, &st->bitmap.vbuf, NULL);
       st->bitmap.vbuf = NULL;
    }
 
diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c
index 327bafeb986..28526234720 100644
--- a/src/mesa/state_tracker/st_cb_blit.c
+++ b/src/mesa/state_tracker/st_cb_blit.c
@@ -79,7 +79,7 @@ st_BlitFramebuffer(GLcontext *ctx,
       struct st_renderbuffer *srcRb = 
          st_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
       struct st_renderbuffer *dstRb = 
-         st_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]);
+         st_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
       struct pipe_surface *srcSurf = srcRb->surface;
       struct pipe_surface *dstSurf = dstRb->surface;
 
diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index b6cea161634..fca1107d72e 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -116,7 +116,7 @@ st_destroy_clear(struct st_context *st)
       st->clear.vs = NULL;
    }
    if (st->clear.vbuf) {
-      pipe_buffer_destroy(pipe->screen, st->clear.vbuf);
+      pipe_buffer_reference(pipe->screen, &st->clear.vbuf, NULL);
       st->clear.vbuf = NULL;
    }
 }
diff --git a/src/mesa/state_tracker/st_cb_strings.c b/src/mesa/state_tracker/st_cb_strings.c
index 066124f8f30..09545aa8fb8 100644
--- a/src/mesa/state_tracker/st_cb_strings.c
+++ b/src/mesa/state_tracker/st_cb_strings.c
@@ -40,7 +40,7 @@
 #include "st_context.h"
 #include "st_cb_strings.h"
 
-#define ST_VERSION_STRING "0.1"
+#define ST_VERSION_STRING "0.2"
 
 static const GLubyte *
 st_get_string(GLcontext * ctx, GLenum name)
diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c
index 0b2b639a5bd..d08229b57a3 100644
--- a/src/mesa/state_tracker/st_cb_texture.c
+++ b/src/mesa/state_tracker/st_cb_texture.c
@@ -749,8 +749,8 @@ st_GetTexImage(GLcontext * ctx, GLenum target, GLint level,
 static void
 st_GetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
                          GLvoid *pixels,
-                         const struct gl_texture_object *texObj,
-                         const struct gl_texture_image *texImage)
+                         struct gl_texture_object *texObj,
+                         struct gl_texture_image *texImage)
 {
    st_get_tex_image(ctx, target, level, 0, 0, pixels,
                     (struct gl_texture_object *) texObj,
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index e78d98f1e91..31383b48874 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -301,18 +301,17 @@ is_interleaved_arrays(const struct st_vertex_program *vp,
    }
 
    *userSpace = (num_client_arrays == vp->num_inputs);
-   /*printf("user space: %d\n", (int) *userSpace);*/
+   /* printf("user space: %d (%d %d)\n", (int) *userSpace,num_client_arrays,vp->num_inputs); */
 
    return GL_TRUE;
 }
 
 
 /**
- * Once we know all the arrays are in user space, this function
- * computes the memory range occupied by the arrays.
+ * Compute the memory range occupied by the arrays.
  */
 static void
-get_user_arrays_bounds(const struct st_vertex_program *vp,
+get_arrays_bounds(const struct st_vertex_program *vp,
                        const struct gl_client_array **arrays,
                        GLuint max_index,
                        const GLubyte **low, const GLubyte **high)
@@ -366,20 +365,21 @@ setup_interleaved_attribs(GLcontext *ctx,
       /*printf("stobj %u = %p\n", attr, (void*)stobj);*/
 
       if (attr == 0) {
+         const GLubyte *low, *high;
+
+         get_arrays_bounds(vp, arrays, max_index, &low, &high);
+         /*printf("buffer range: %p %p  %d\n", low, high, high-low);*/
+
+         offset0 = low;
          if (userSpace) {
-            const GLubyte *low, *high;
-            get_user_arrays_bounds(vp, arrays, max_index, &low, &high);
-            /*printf("user buffer range: %p %p  %d\n", low, high, high-low);*/
             vbuffer->buffer =
                pipe_user_buffer_create(pipe->screen, (void *) low, high - low);
             vbuffer->buffer_offset = 0;
-            offset0 = low;
          }
          else {
             vbuffer->buffer = NULL;
             pipe_buffer_reference(pipe->screen, &vbuffer->buffer, stobj->buffer);
             vbuffer->buffer_offset = (unsigned) arrays[mesaAttr]->Ptr;
-            offset0 = arrays[mesaAttr]->Ptr;
          }
          vbuffer->pitch = stride; /* in bytes */
          vbuffer->max_index = max_index;
diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c
index 60fd440ef77..5ff0c611477 100644
--- a/src/mesa/state_tracker/st_extensions.c
+++ b/src/mesa/state_tracker/st_extensions.c
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright (c) 2008 VMware, Inc.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -89,6 +90,9 @@ void st_init_limits(struct st_context *st)
       = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS),
             MAX_TEXTURE_IMAGE_UNITS);
 
+   c->MaxVertexTextureImageUnits
+      = screen->get_param(screen, PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS);
+
    c->MaxDrawBuffers
       = _clamp(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS),
               1, MAX_DRAW_BUFFERS);
@@ -240,6 +244,9 @@ void st_init_extensions(struct st_context *st)
 
    if (screen->is_format_supported(screen, PIPE_FORMAT_R8G8B8A8_SRGB,
                                    PIPE_TEXTURE_2D, 
+                                   PIPE_TEXTURE_USAGE_SAMPLER, 0) ||
+      screen->is_format_supported(screen, PIPE_FORMAT_A8R8G8B8_SRGB,
+                                   PIPE_TEXTURE_2D, 
                                    PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
       ctx->Extensions.EXT_texture_sRGB = GL_TRUE;
    }
diff --git a/src/mesa/state_tracker/st_format.c b/src/mesa/state_tracker/st_format.c
index a8ae30a454f..9e2d60c9269 100644
--- a/src/mesa/state_tracker/st_format.c
+++ b/src/mesa/state_tracker/st_format.c
@@ -1,6 +1,7 @@
 /**************************************************************************
  * 
  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright (c) 2008 VMware, Inc.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -263,6 +264,28 @@ st_mesa_format_to_pipe_format(GLuint mesaFormat)
       return PIPE_FORMAT_DXT3_RGBA;
    case MESA_FORMAT_RGBA_DXT5:
       return PIPE_FORMAT_DXT5_RGBA;
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SRGB_DXT1:
+      return PIPE_FORMAT_DXT1_SRGB;
+   case MESA_FORMAT_SRGBA_DXT1:
+      return PIPE_FORMAT_DXT1_SRGBA;
+   case MESA_FORMAT_SRGBA_DXT3:
+      return PIPE_FORMAT_DXT3_SRGBA;
+   case MESA_FORMAT_SRGBA_DXT5:
+      return PIPE_FORMAT_DXT5_SRGBA;
+#endif
+#endif
+#if FEATURE_EXT_texture_sRGB
+   case MESA_FORMAT_SLA8:
+      return PIPE_FORMAT_A8L8_SRGB;
+   case MESA_FORMAT_SL8:
+      return PIPE_FORMAT_L8_SRGB;
+   case MESA_FORMAT_SRGB8:
+      return PIPE_FORMAT_R8G8B8_SRGB;
+   case MESA_FORMAT_SRGBA8:
+      return PIPE_FORMAT_R8G8B8A8_SRGB;
+   case MESA_FORMAT_SARGB8:
+      return PIPE_FORMAT_A8R8G8B8_SRGB;
 #endif
    default:
       assert(0);
@@ -294,6 +317,28 @@ default_rgba_format(struct pipe_screen *screen,
    return PIPE_FORMAT_NONE;
 }
 
+/**
+ * Find an sRGBA format supported by the context/winsys.
+ */
+static enum pipe_format
+default_srgba_format(struct pipe_screen *screen, 
+                    enum pipe_texture_target target,
+                    unsigned tex_usage, 
+                    unsigned geom_flags)
+{
+   static const enum pipe_format colorFormats[] = {
+      PIPE_FORMAT_A8R8G8B8_SRGB,
+      PIPE_FORMAT_B8G8R8A8_SRGB,
+      PIPE_FORMAT_R8G8B8A8_SRGB,
+   };
+   uint i;
+   for (i = 0; i < Elements(colorFormats); i++) {
+      if (screen->is_format_supported( screen, colorFormats[i], target, tex_usage, geom_flags )) {
+         return colorFormats[i];
+      }
+   }
+   return PIPE_FORMAT_NONE;
+}
 
 /**
  * Search list of formats for first RGBA format with >8 bits/channel.
@@ -512,6 +557,36 @@ st_choose_format(struct pipe_context *pipe, GLint internalFormat,
          return PIPE_FORMAT_Z24S8_UNORM;
       return PIPE_FORMAT_NONE;
 
+   case GL_SRGB_EXT:
+   case GL_SRGB8_EXT:
+   case GL_COMPRESSED_SRGB_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_EXT:
+   case GL_SRGB_ALPHA_EXT:
+   case GL_SRGB8_ALPHA8_EXT:
+      return default_srgba_format( screen, target, tex_usage, geom_flags );
+   case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+      return PIPE_FORMAT_DXT1_SRGB;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+      return PIPE_FORMAT_DXT1_SRGBA;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+      return PIPE_FORMAT_DXT3_SRGBA;
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
+      return PIPE_FORMAT_DXT5_SRGBA;
+
+   case GL_SLUMINANCE_ALPHA_EXT:
+   case GL_SLUMINANCE8_ALPHA8_EXT:
+   case GL_COMPRESSED_SLUMINANCE_EXT:
+   case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
+      if (screen->is_format_supported( screen, PIPE_FORMAT_A8L8_SRGB, target, tex_usage, geom_flags ))
+         return PIPE_FORMAT_A8L8_SRGB;
+      return default_srgba_format( screen, target, tex_usage, geom_flags );
+
+   case GL_SLUMINANCE_EXT:
+   case GL_SLUMINANCE8_EXT:
+      if (screen->is_format_supported( screen, PIPE_FORMAT_L8_SRGB, target, tex_usage, geom_flags ))
+         return PIPE_FORMAT_L8_SRGB;
+      return default_srgba_format( screen, target, tex_usage, geom_flags );
+
    default:
       return PIPE_FORMAT_NONE;
    }
@@ -591,6 +666,28 @@ translate_gallium_format_to_mesa_format(enum pipe_format format)
       return &_mesa_texformat_rgba_dxt3;
    case PIPE_FORMAT_DXT5_RGBA:
       return &_mesa_texformat_rgba_dxt5;
+#if FEATURE_EXT_texture_sRGB
+   case PIPE_FORMAT_DXT1_SRGB:
+      return &_mesa_texformat_srgb_dxt1;
+   case PIPE_FORMAT_DXT1_SRGBA:
+      return &_mesa_texformat_srgba_dxt1;
+   case PIPE_FORMAT_DXT3_SRGBA:
+      return &_mesa_texformat_srgba_dxt3;
+   case PIPE_FORMAT_DXT5_SRGBA:
+      return &_mesa_texformat_srgba_dxt5;
+#endif
+#endif
+#if FEATURE_EXT_texture_sRGB
+   case PIPE_FORMAT_A8L8_SRGB:
+      return &_mesa_texformat_sla8;
+   case PIPE_FORMAT_L8_SRGB:
+      return &_mesa_texformat_sl8;
+   case PIPE_FORMAT_R8G8B8_SRGB:
+      return &_mesa_texformat_srgb8;
+   case PIPE_FORMAT_R8G8B8A8_SRGB:
+      return &_mesa_texformat_srgba8;
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      return &_mesa_texformat_sargb8;
 #endif
    /* XXX add additional cases */
    default:
diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c
index 7d270a32729..aaa7d42ae06 100644
--- a/src/mesa/state_tracker/st_framebuffer.c
+++ b/src/mesa/state_tracker/st_framebuffer.c
@@ -153,9 +153,9 @@ void st_resize_framebuffer( struct st_framebuffer *stfb,
 }
 
 
-void st_unreference_framebuffer( struct st_framebuffer **stfb )
+void st_unreference_framebuffer( struct st_framebuffer *stfb )
 {
-   _mesa_unreference_framebuffer((struct gl_framebuffer **) stfb);
+   _mesa_unreference_framebuffer((struct gl_framebuffer **) &stfb);
 }
 
 
@@ -176,7 +176,9 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb,
    assert(surfIndex < BUFFER_COUNT);
 
    strb = st_renderbuffer(stfb->Base.Attachment[surfIndex].Renderbuffer);
-   assert(strb);
+
+   /* fail */
+   if (!strb) return;
 
    /* replace the renderbuffer's surface/texture pointers */
    pipe_surface_reference( &strb->surface, surf );
@@ -304,3 +306,10 @@ void *st_framebuffer_private( struct st_framebuffer *stfb )
    return stfb->Private;
 }
 
+void st_get_framebuffer_dimensions( struct st_framebuffer *stfb,
+				    uint *width,
+				    uint *height)
+{
+   *width = stfb->Base.Width;
+   *height = stfb->Base.Height;
+}
diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c
index b9d114b1c97..a15faf732ca 100644
--- a/src/mesa/state_tracker/st_gen_mipmap.c
+++ b/src/mesa/state_tracker/st_gen_mipmap.c
@@ -138,11 +138,11 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target,
       _mesa_generate_mipmap_level(target, datatype, comps,
                    0 /*border*/,
                    pt->width[srcLevel], pt->height[srcLevel], pt->depth[srcLevel],
-                   srcSurf->stride, /* stride in bytes */
                    srcData,
+                   srcSurf->stride, /* stride in bytes */
                    pt->width[dstLevel], pt->height[dstLevel], pt->depth[dstLevel],
-                   dstSurf->stride, /* stride in bytes */
-                   dstData);
+                   dstData,
+                   dstSurf->stride); /* stride in bytes */
 
       pipe_buffer_unmap(pipe->screen, srcSurf->buffer);
       pipe_buffer_unmap(pipe->screen, dstSurf->buffer);
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 33574ddf409..ce8b4411940 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -346,6 +346,12 @@ compile_instruction(
    case OPCODE_DDY:
       fullinst->Instruction.Opcode = TGSI_OPCODE_DDY;
       break;
+   case OPCODE_DP2:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_DP2;
+      break;
+   case OPCODE_DP2A:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_DP2A;
+      break;
    case OPCODE_DP3:
       fullinst->Instruction.Opcode = TGSI_OPCODE_DP3;
       break;
@@ -389,8 +395,8 @@ compile_instruction(
       fullinst->Instruction.Opcode = TGSI_OPCODE_IF;
       fullinst->InstructionExtLabel.Label = inst->BranchTarget + preamble_size;
       break;
-   case OPCODE_INT:
-      fullinst->Instruction.Opcode = TGSI_OPCODE_INT;
+   case OPCODE_TRUNC:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_TRUNC;
       break;
    case OPCODE_KIL:
       /* conditional */
@@ -443,6 +449,12 @@ compile_instruction(
    case OPCODE_NOP:
       fullinst->Instruction.Opcode = TGSI_OPCODE_NOP;
       break;
+   case OPCODE_NRM3:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_NRM;
+      break;
+   case OPCODE_NRM4:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_NRM4;
+      break;
    case OPCODE_POW:
       fullinst->Instruction.Opcode = TGSI_OPCODE_POW;
       break;
@@ -498,6 +510,9 @@ compile_instruction(
    case OPCODE_SNE:
       fullinst->Instruction.Opcode = TGSI_OPCODE_SNE;
       break;
+   case OPCODE_SSG:
+      fullinst->Instruction.Opcode = TGSI_OPCODE_SSG;
+      break;
    case OPCODE_SUB:
       fullinst->Instruction.Opcode = TGSI_OPCODE_SUB;
       break;
@@ -569,7 +584,8 @@ make_input_decl(
    GLuint usage_mask,
    GLboolean semantic_info,
    GLuint semantic_name,
-   GLbitfield semantic_index )
+   GLbitfield semantic_index,
+   GLbitfield input_flags)
 {
    struct tgsi_full_declaration decl;
 
@@ -588,6 +604,10 @@ make_input_decl(
    if (interpolate_info) {
       decl.Declaration.Interpolate = interpolate;
    }
+   if (input_flags & PROG_PARAM_BIT_CENTROID)
+      decl.Declaration.Centroid = 1;
+   if (input_flags & PROG_PARAM_BIT_INVARIANT)
+      decl.Declaration.Invariant = 1;
 
    return decl;
 }
@@ -600,7 +620,8 @@ make_output_decl(
    GLuint index,
    GLuint semantic_name,
    GLuint semantic_index,
-   GLbitfield usage_mask )
+   GLuint usage_mask,
+   GLbitfield output_flags)
 {
    struct tgsi_full_declaration decl;
 
@@ -614,6 +635,10 @@ make_output_decl(
    decl.DeclarationRange.Last = index;
    decl.Semantic.SemanticName = semantic_name;
    decl.Semantic.SemanticIndex = semantic_index;
+   if (output_flags & PROG_PARAM_BIT_CENTROID)
+      decl.Declaration.Centroid = 1;
+   if (output_flags & PROG_PARAM_BIT_INVARIANT)
+      decl.Declaration.Invariant = 1;
 
    return decl;
 }
@@ -727,10 +752,12 @@ st_translate_mesa_program(
    const ubyte inputSemanticName[],
    const ubyte inputSemanticIndex[],
    const GLuint interpMode[],
+   const GLbitfield inputFlags[],
    GLuint numOutputs,
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
    const ubyte outputSemanticIndex[],
+   const GLbitfield outputFlags[],
    struct tgsi_token *tokens,
    GLuint maxTokens )
 {
@@ -768,7 +795,8 @@ st_translate_mesa_program(
                                     GL_TRUE, interpMode[i],
                                     TGSI_WRITEMASK_XYZW,
                                     GL_TRUE, inputSemanticName[i],
-                                    inputSemanticIndex[i]);
+                                    inputSemanticIndex[i],
+                                    inputFlags[i]);
          ti += tgsi_build_full_declaration(&fulldecl,
                                            &tokens[ti],
                                            header,
@@ -785,7 +813,8 @@ st_translate_mesa_program(
          fulldecl = make_input_decl(i,
                                     GL_FALSE, 0,
                                     TGSI_WRITEMASK_XYZW,
-                                    GL_FALSE, 0, 0);
+                                    GL_FALSE, 0, 0,
+                                    inputFlags[i]);
          ti += tgsi_build_full_declaration(&fulldecl,
                                            &tokens[ti],
                                            header,
@@ -804,13 +833,15 @@ st_translate_mesa_program(
             fulldecl = make_output_decl(i,
                                         TGSI_SEMANTIC_POSITION, /* Z / Depth */
                                         outputSemanticIndex[i],
-                                        TGSI_WRITEMASK_Z );
+                                        TGSI_WRITEMASK_Z,
+                                        outputFlags[i]);
             break;
          case TGSI_SEMANTIC_COLOR:
             fulldecl = make_output_decl(i,
                                         TGSI_SEMANTIC_COLOR,
                                         outputSemanticIndex[i],
-                                        TGSI_WRITEMASK_XYZW );
+                                        TGSI_WRITEMASK_XYZW,
+                                        outputFlags[i]);
             break;
          default:
             assert(0);
@@ -829,7 +860,8 @@ st_translate_mesa_program(
          fulldecl = make_output_decl(i,
                                      outputSemanticName[i],
                                      outputSemanticIndex[i],
-                                     TGSI_WRITEMASK_XYZW );
+                                     TGSI_WRITEMASK_XYZW,
+                                     outputFlags[i]);
          ti += tgsi_build_full_declaration(&fulldecl,
                                            &tokens[ti],
                                            header,
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h
index 77c74644b8b..7b2bee1ab70 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.h
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h
@@ -48,10 +48,12 @@ st_translate_mesa_program(
    const ubyte inputSemanticName[],
    const ubyte inputSemanticIndex[],
    const GLuint interpMode[],
+   const GLbitfield inputFlags[],
    GLuint numOutputs,
    const GLuint outputMapping[],
    const ubyte outputSemanticName[],
    const ubyte outputSemanticIndex[],
+   const GLbitfield outputFlags[],
    struct tgsi_token *tokens,
    GLuint maxTokens );
 
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index af0df22dc59..cf4b39cee40 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -99,7 +99,12 @@ st_translate_vertex_program(struct st_context *st,
    ubyte vs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
    uint vs_num_outputs = 0;
 
+   GLbitfield input_flags[MAX_PROGRAM_INPUTS];
+   GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
+
    memset(&vs, 0, sizeof(vs));
+   memset(input_flags, 0, sizeof(input_flags));
+   memset(output_flags, 0, sizeof(output_flags));
 
    if (stvp->Base.IsPositionInvariant)
       _mesa_insert_mvp_code(st->ctx, &stvp->Base);
@@ -171,6 +176,8 @@ st_translate_vertex_program(struct st_context *st,
          default:
             assert(0);
          }
+
+         input_flags[slot] = stvp->Base.Base.InputFlags[attr];
       }
    }
 
@@ -192,6 +199,7 @@ st_translate_vertex_program(struct st_context *st,
    for (i = 0; i < PIPE_MAX_SHADER_OUTPUTS; i++) {
       vs_output_semantic_name[i] = TGSI_SEMANTIC_GENERIC;
       vs_output_semantic_index[i] = 0;
+      output_flags[i] = 0x0;
    }
 
    num_generic = 0;
@@ -276,6 +284,8 @@ st_translate_vertex_program(struct st_context *st,
                vs_output_semantic_index[slot] = num_generic++;
             }
          }
+
+         output_flags[slot] = stvp->Base.Base.OutputFlags[attr];
       }
    }
 
@@ -307,21 +317,23 @@ st_translate_vertex_program(struct st_context *st,
 
    /* XXX: fix static allocation of tokens:
     */
-   num_tokens = st_translate_mesa_program( TGSI_PROCESSOR_VERTEX,
-                                &stvp->Base.Base,
-                                /* inputs */
-                                vs_num_inputs,
-                                stvp->input_to_index,
-                                vs_input_semantic_name,
-                                vs_input_semantic_index,
-                                NULL,
-                                /* outputs */
-                                vs_num_outputs,
-                                outputMapping,
-                                vs_output_semantic_name,
-                                vs_output_semantic_index,
-                                /* tokenized result */
-                                tokens, ST_MAX_SHADER_TOKENS);
+   num_tokens = st_translate_mesa_program(TGSI_PROCESSOR_VERTEX,
+                                          &stvp->Base.Base,
+                                          /* inputs */
+                                          vs_num_inputs,
+                                          stvp->input_to_index,
+                                          vs_input_semantic_name,
+                                          vs_input_semantic_index,
+                                          NULL,
+                                          input_flags,
+                                          /* outputs */
+                                          vs_num_outputs,
+                                          outputMapping,
+                                          vs_output_semantic_name,
+                                          vs_output_semantic_index,
+                                          output_flags,
+                                          /* tokenized result */
+                                          tokens, ST_MAX_SHADER_TOKENS);
 
    assert(num_tokens < ST_MAX_SHADER_TOKENS);
 
@@ -371,7 +383,12 @@ st_translate_fragment_program(struct st_context *st,
    ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS];
    uint fs_num_outputs = 0;
 
+   GLbitfield input_flags[MAX_PROGRAM_INPUTS];
+   GLbitfield output_flags[MAX_PROGRAM_OUTPUTS];
+
    memset(&fs, 0, sizeof(fs));
+   memset(input_flags, 0, sizeof(input_flags));
+   memset(output_flags, 0, sizeof(output_flags));
 
    /* which vertex output goes to the first fragment input: */
    if (inputsRead & FRAG_BIT_WPOS)
@@ -435,6 +452,8 @@ st_translate_fragment_program(struct st_context *st,
             stfp->input_semantic_index[slot] = num_generic++;
             interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE;
          }
+
+         input_flags[slot] = stfp->Base.Base.InputFlags[attr];
       }
    }
 
@@ -471,6 +490,9 @@ st_translate_fragment_program(struct st_context *st,
             default:
                assert(0);
             }
+
+            output_flags[fs_num_outputs] = stfp->Base.Base.OutputFlags[attr];
+
             fs_num_outputs++;
          }
       }
@@ -481,21 +503,23 @@ st_translate_fragment_program(struct st_context *st,
 
    /* XXX: fix static allocation of tokens:
     */
-   num_tokens = st_translate_mesa_program( TGSI_PROCESSOR_FRAGMENT,
-                                &stfp->Base.Base,
-                                /* inputs */
-                                fs_num_inputs,
-                                inputMapping,
-                                stfp->input_semantic_name,
-                                stfp->input_semantic_index,
-                                interpMode,
-                                /* outputs */
-                                fs_num_outputs,
-                                outputMapping,
-                                fs_output_semantic_name,
-                                fs_output_semantic_index,
-                                /* tokenized result */
-                                tokens, ST_MAX_SHADER_TOKENS);
+   num_tokens = st_translate_mesa_program(TGSI_PROCESSOR_FRAGMENT,
+                                          &stfp->Base.Base,
+                                          /* inputs */
+                                          fs_num_inputs,
+                                          inputMapping,
+                                          stfp->input_semantic_name,
+                                          stfp->input_semantic_index,
+                                          interpMode,
+                                          input_flags,
+                                          /* outputs */
+                                          fs_num_outputs,
+                                          outputMapping,
+                                          fs_output_semantic_name,
+                                          fs_output_semantic_index,
+                                          output_flags,
+                                          /* tokenized result */
+                                          tokens, ST_MAX_SHADER_TOKENS);
 
    assert(num_tokens < ST_MAX_SHADER_TOKENS);
 
diff --git a/src/mesa/state_tracker/st_public.h b/src/mesa/state_tracker/st_public.h
index 5cfb2e41f24..88995aa8747 100644
--- a/src/mesa/state_tracker/st_public.h
+++ b/src/mesa/state_tracker/st_public.h
@@ -42,6 +42,8 @@
 #define ST_SURFACE_DEPTH        8
 
 #define ST_TEXTURE_2D    0x2
+#define ST_TEXTURE_RECT  0x4
+
 #define ST_TEXTURE_RGB   0x1
 #define ST_TEXTURE_RGBA  0x2
 
@@ -75,6 +77,9 @@ void st_resize_framebuffer( struct st_framebuffer *stfb,
 void st_set_framebuffer_surface(struct st_framebuffer *stfb,
                                 uint surfIndex, struct pipe_surface *surf);
 
+void st_get_framebuffer_dimensions( struct st_framebuffer *stfb,
+				    uint *width, uint *height);
+
 struct pipe_surface *st_get_framebuffer_surface(struct st_framebuffer *stfb,
                                                 uint surfIndex);
 
@@ -83,7 +88,7 @@ struct pipe_texture *st_get_framebuffer_texture(struct st_framebuffer *stfb,
 
 void *st_framebuffer_private( struct st_framebuffer *stfb );
 
-void st_unreference_framebuffer( struct st_framebuffer **stfb );
+void st_unreference_framebuffer( struct st_framebuffer *stfb );
 
 void st_make_current(struct st_context *st,
                      struct st_framebuffer *draw,
@@ -96,6 +101,7 @@ void st_finish( struct st_context *st );
 void st_notify_swapbuffers(struct st_framebuffer *stfb);
 void st_notify_swapbuffers_complete(struct st_framebuffer *stfb);
 
+int st_set_teximage(struct pipe_texture *pt, int target);
 
 /** Redirect rendering into stfb's surface to a texture image */
 int st_bind_teximage(struct st_framebuffer *stfb, uint surfIndex,
diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c
index 29b16347625..63cfe5fc163 100644
--- a/src/mesa/state_tracker/st_texture.c
+++ b/src/mesa/state_tracker/st_texture.c
@@ -32,6 +32,7 @@
 #include "st_cb_fbo.h"
 #include "main/enums.h"
 #include "main/teximage.h"
+#include "main/texstore.h"
 
 #undef Elements  /* fix re-defined macro warning */
 
@@ -352,6 +353,52 @@ st_texture_image_copy(struct pipe_context *pipe,
    }
 }
 
+/** Bind a pipe surface for use as a texture image */
+int
+st_set_teximage(struct pipe_texture *pt, int target)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   const GLuint unit = ctx->Texture.CurrentUnit;
+   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+   struct st_texture_image *stImage;
+   int internalFormat;
+
+   switch (pt->format) {
+   case PIPE_FORMAT_A8R8G8B8_UNORM:
+      internalFormat = GL_RGBA8;
+      break;
+   default:
+      return 0;
+   };
+
+   switch (target) {
+   case ST_TEXTURE_2D:
+      target = GL_TEXTURE_2D;
+      break;
+   case ST_TEXTURE_RECT:
+      target = GL_TEXTURE_RECTANGLE_ARB;
+      break;
+   default:
+      return 0;
+   }
+
+   texObj = _mesa_select_tex_object(ctx, texUnit, target);
+   texImage = _mesa_get_tex_image(ctx, texObj, target, 0);
+   stImage = st_texture_image(texImage);
+   
+   _mesa_init_teximage_fields(ctx, GL_TEXTURE_2D, texImage, pt->width[0],
+                              pt->height[0], 1, 0, internalFormat);
+
+   texImage->TexFormat = st_ChooseTextureFormat(ctx, internalFormat, GL_RGBA,
+                                                GL_UNSIGNED_BYTE);
+   _mesa_set_fetch_functions(texImage, 2);
+
+   pipe_texture_reference(&stImage->pt, pt);
+
+   return 1;
+}
 
 /** Redirect rendering into stfb's surface to a texture image */
 int
diff --git a/src/mesa/swrast/descrip.mms b/src/mesa/swrast/descrip.mms
new file mode 100644
index 00000000000..0b23deb3c2f
--- /dev/null
+++ b/src/mesa/swrast/descrip.mms
@@ -0,0 +1,82 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define swrast [-.swrast]
+	define array_cache [-.array_cache]
+	define glapi [-.glapi]
+	define main [-.main]
+	define shader [-.shader]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.main],[-.glapi],[-.shader],[-.shader.slang]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = s_aaline.c s_aatriangle.c s_accum.c s_alpha.c \
+	s_bitmap.c s_blend.c s_blit.c s_buffers.c s_context.c \
+	s_copypix.c s_depth.c s_fragprog.c \
+        s_drawpix.c s_feedback.c s_fog.c s_imaging.c s_lines.c s_logic.c \
+	s_masking.c s_points.c s_readpix.c \
+	s_span.c s_stencil.c s_texstore.c s_texcombine.c s_texfilter.c \
+	s_triangle.c s_zoom.c s_atifragshader.c
+ 
+OBJECTS = s_aaline.obj,s_aatriangle.obj,s_accum.obj,s_alpha.obj,\
+	s_bitmap.obj,s_blend.obj,s_blit.obj,s_fragprog.obj,\
+	s_buffers.obj,s_context.obj,s_atifragshader.obj,\
+	s_copypix.obj,s_depth.obj,s_drawpix.obj,s_feedback.obj,s_fog.obj,\
+	s_imaging.obj,s_lines.obj,s_logic.obj,s_masking.obj,\
+	s_points.obj,s_readpix.obj,s_span.obj,s_stencil.obj,\
+	s_texstore.obj,s_texcombine.obj,s_texfilter.obj,s_triangle.obj,\
+	s_zoom.obj
+ 
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+s_atifragshader.obj : s_atifragshader.c
+s_aaline.obj : s_aaline.c
+s_aatriangle.obj : s_aatriangle.c
+s_accum.obj : s_accum.c
+s_alpha.obj : s_alpha.c
+s_bitmap.obj : s_bitmap.c
+s_blend.obj : s_blend.c
+s_blit.obj : s_blit.c
+s_buffers.obj : s_buffers.c
+s_context.obj : s_context.c
+s_copypix.obj : s_copypix.c
+s_depth.obj : s_depth.c
+s_drawpix.obj : s_drawpix.c
+s_feedback.obj : s_feedback.c
+s_fog.obj : s_fog.c
+s_imaging.obj : s_imaging.c
+s_lines.obj : s_lines.c
+s_logic.obj : s_logic.c
+s_masking.obj : s_masking.c
+s_points.obj : s_points.c
+s_readpix.obj : s_readpix.c
+s_span.obj : s_span.c
+s_stencil.obj : s_stencil.c
+s_texstore.obj : s_texstore.c
+s_texcombine.obj : s_texcombine.c
+s_texfilter.obj : s_texfilter.c
+s_triangle.obj : s_triangle.c
+s_zoom.obj : s_zoom.c
+s_fragprog.obj : s_fragprog.c
diff --git a/src/mesa/swrast/s_aaline.c b/src/mesa/swrast/s_aaline.c
index ee65a71d7e6..9bfa8f2e619 100644
--- a/src/mesa/swrast/s_aaline.c
+++ b/src/mesa/swrast/s_aaline.c
@@ -26,11 +26,11 @@
 #include "main/glheader.h"
 #include "main/imports.h"
 #include "main/macros.h"
+#include "main/mtypes.h"
 #include "swrast/s_aaline.h"
 #include "swrast/s_context.h"
 #include "swrast/s_span.h"
 #include "swrast/swrast.h"
-#include "main/mtypes.h"
 
 
 #define SUB_PIXEL 4
diff --git a/src/mesa/swrast/s_aaline.h b/src/mesa/swrast/s_aaline.h
index 9fb4959f917..f1d708ec801 100644
--- a/src/mesa/swrast/s_aaline.h
+++ b/src/mesa/swrast/s_aaline.h
@@ -28,7 +28,6 @@
 #define S_AALINE_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_aalinetemp.h b/src/mesa/swrast/s_aalinetemp.h
index e7911fec3b5..ca08203d831 100644
--- a/src/mesa/swrast/s_aalinetemp.h
+++ b/src/mesa/swrast/s_aalinetemp.h
@@ -141,6 +141,7 @@ NAME(line)(GLcontext *ctx, const SWvertex *v0, const SWvertex *v1)
 
    INIT_SPAN(line.span, GL_LINE);
    line.span.arrayMask = SPAN_XY | SPAN_COVERAGE;
+   line.span.facing = swrast->PointLineFacing;
    line.xAdj = line.dx / line.len * line.halfWidth;
    line.yAdj = line.dy / line.len * line.halfWidth;
 
diff --git a/src/mesa/swrast/s_aatriangle.h b/src/mesa/swrast/s_aatriangle.h
index 734d420b628..4b57fa73a27 100644
--- a/src/mesa/swrast/s_aatriangle.h
+++ b/src/mesa/swrast/s_aatriangle.h
@@ -28,7 +28,6 @@
 #define S_AATRIANGLE_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_aatritemp.h b/src/mesa/swrast/s_aatritemp.h
index 42d74a16328..0827b3db9eb 100644
--- a/src/mesa/swrast/s_aatritemp.h
+++ b/src/mesa/swrast/s_aatritemp.h
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.0.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -65,7 +65,7 @@
    GLfloat attrPlane[FRAG_ATTRIB_MAX][4][4];
    GLfloat wPlane[4];  /* win[3] */
 #endif
-   GLfloat bf = SWRAST_CONTEXT(ctx)->_BackfaceSign;
+   GLfloat bf = SWRAST_CONTEXT(ctx)->_BackfaceCullSign;
    
    (void) swrast;
 
@@ -104,6 +104,7 @@
    majDx = vMax->attrib[FRAG_ATTRIB_WPOS][0] - vMin->attrib[FRAG_ATTRIB_WPOS][0];
    majDy = vMax->attrib[FRAG_ATTRIB_WPOS][1] - vMin->attrib[FRAG_ATTRIB_WPOS][1];
 
+   /* front/back-face determination and cullling */
    {
       const GLfloat botDx = vMid->attrib[FRAG_ATTRIB_WPOS][0] - vMin->attrib[FRAG_ATTRIB_WPOS][0];
       const GLfloat botDy = vMid->attrib[FRAG_ATTRIB_WPOS][1] - vMin->attrib[FRAG_ATTRIB_WPOS][1];
@@ -112,6 +113,8 @@
       if (area * bf < 0 || area == 0 || IS_INF_OR_NAN(area))
 	 return;
       ltor = (GLboolean) (area < 0.0F);
+
+      span.facing = area * swrast->_BackfaceSign > 0.0F;
    }
 
    /* Plane equation setup:
@@ -286,7 +289,7 @@
          }
 
          /* skip fragments with zero coverage */
-         while (startX >= 0) {
+         while (startX > 0) {
             coverage = compute_coveragef(pMin, pMax, pMid, startX, iy);
             if (coverage > 0.0F)
                break;
@@ -300,6 +303,7 @@
             /* (cx,cy) = center of fragment */
             const GLfloat cx = ix + 0.5F, cy = iy + 0.5F;
             SWspanarrays *array = span.array;
+            ASSERT(ix >= 0);
 #ifdef DO_INDEX
             array->coverage[ix] = (GLfloat) compute_coveragei(pMin, pMax, pMid, ix, iy);
 #else
diff --git a/src/mesa/swrast/s_accum.c b/src/mesa/swrast/s_accum.c
index 13a42fdf53e..ff741777e75 100644
--- a/src/mesa/swrast/s_accum.c
+++ b/src/mesa/swrast/s_accum.c
@@ -27,7 +27,7 @@
 #include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "fbobject.h"
+#include "main/fbobject.h"
 
 #include "s_accum.h"
 #include "s_context.h"
@@ -525,8 +525,8 @@ accum_return(GLcontext *ctx, GLfloat value,
          }
 
          /* store colors */
-         for (buffer = 0; buffer < fb->_NumColorDrawBuffers[0]; buffer++) {
-            struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[0][buffer];
+         for (buffer = 0; buffer < fb->_NumColorDrawBuffers; buffer++) {
+            struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buffer];
             if (masking) {
                _swrast_mask_rgba_span(ctx, rb, &span);
             }
diff --git a/src/mesa/swrast/s_alpha.h b/src/mesa/swrast/s_alpha.h
index 92cb01b18ad..7a5b72e650a 100644
--- a/src/mesa/swrast/s_alpha.h
+++ b/src/mesa/swrast/s_alpha.h
@@ -28,7 +28,6 @@
 #define S_ALPHA_H
 
 
-#include "main/mtypes.h"
 #include "s_context.h"
 
 
diff --git a/src/mesa/swrast/s_atifragshader.c b/src/mesa/swrast/s_atifragshader.c
index 9fa352c36bf..21cbd3cf37f 100644
--- a/src/mesa/swrast/s_atifragshader.c
+++ b/src/mesa/swrast/s_atifragshader.c
@@ -1,5 +1,4 @@
 /*
- *
  * Copyright (C) 2004  David Airlie   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
diff --git a/src/mesa/swrast/s_bitmap.c b/src/mesa/swrast/s_bitmap.c
index 2308acae1cd..35b34e654ff 100644
--- a/src/mesa/swrast/s_bitmap.c
+++ b/src/mesa/swrast/s_bitmap.c
@@ -29,10 +29,10 @@
  */
 
 #include "main/glheader.h"
-#include "bufferobj.h"
-#include "image.h"
+#include "main/bufferobj.h"
+#include "main/image.h"
 #include "main/macros.h"
-#include "pixel.h"
+#include "main/pixel.h"
 
 #include "s_context.h"
 #include "s_span.h"
diff --git a/src/mesa/swrast/s_blend.c b/src/mesa/swrast/s_blend.c
index 7fd89453545..95c83432a9c 100644
--- a/src/mesa/swrast/s_blend.c
+++ b/src/mesa/swrast/s_blend.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -880,7 +880,8 @@ blend_general(GLcontext *ctx, GLuint n, const GLubyte mask[],
       }
    }
    else {
-      blend_general_float(ctx, n, mask, rgbaF, destF, chanType);
+      blend_general_float(ctx, n, mask, (GLfloat (*)[4]) src,
+                          (GLfloat (*)[4]) dst, chanType);
    }
 }
 
diff --git a/src/mesa/swrast/s_blend.h b/src/mesa/swrast/s_blend.h
index c8abecc099b..8d5a81635d5 100644
--- a/src/mesa/swrast/s_blend.h
+++ b/src/mesa/swrast/s_blend.h
@@ -27,7 +27,6 @@
 #define S_BLEND_H
 
 
-#include "main/mtypes.h"
 #include "s_context.h"
 
 
diff --git a/src/mesa/swrast/s_blit.c b/src/mesa/swrast/s_blit.c
index 0e12dfa1da6..bc4b2ac6251 100644
--- a/src/mesa/swrast/s_blit.c
+++ b/src/mesa/swrast/s_blit.c
@@ -135,7 +135,7 @@ blit_nearest(GLcontext *ctx,
    switch (buffer) {
    case GL_COLOR_BUFFER_BIT:
       readRb = ctx->ReadBuffer->_ColorReadBuffer;
-      drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+      drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0];
       comps = 4;
       break;
    case GL_DEPTH_BUFFER_BIT:
@@ -319,7 +319,7 @@ blit_linear(GLcontext *ctx,
             GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1)
 {
    struct gl_renderbuffer *readRb = ctx->ReadBuffer->_ColorReadBuffer;
-   struct gl_renderbuffer *drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+   struct gl_renderbuffer *drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0];
 
    const GLint srcWidth = ABS(srcX1 - srcX0);
    const GLint dstWidth = ABS(dstX1 - dstX0);
@@ -493,7 +493,7 @@ simple_blit(GLcontext *ctx,
    switch (buffer) {
    case GL_COLOR_BUFFER_BIT:
       readRb = ctx->ReadBuffer->_ColorReadBuffer;
-      drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+      drawRb = ctx->DrawBuffer->_ColorDrawBuffers[0];
       comps = 4;
       break;
    case GL_DEPTH_BUFFER_BIT:
@@ -745,6 +745,9 @@ _swrast_BlitFramebuffer(GLcontext *ctx,
    };
    GLint i;
 
+   if (!ctx->DrawBuffer->_NumColorDrawBuffers)
+      return;
+
    if (!clip_blit(ctx, &srcX0, &srcY0, &srcX1, &srcY1,
                   &dstX0, &dstY0, &dstX1, &dstY1)) {
       return;
diff --git a/src/mesa/swrast/s_buffers.c b/src/mesa/swrast/s_buffers.c
index a70f4745872..9e87d6d4857 100644
--- a/src/mesa/swrast/s_buffers.c
+++ b/src/mesa/swrast/s_buffers.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -251,7 +251,7 @@ static void
 clear_color_buffers(GLcontext *ctx)
 {
    GLboolean masking;
-   GLuint i;
+   GLuint buf;
 
    if (ctx->Visual.rgbMode) {
       if (ctx->Color.ColorMask[0] && 
@@ -265,7 +265,7 @@ clear_color_buffers(GLcontext *ctx)
       }
    }
    else {
-      struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+      struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
       const GLuint indexBits = (1 << rb->IndexBits) - 1;
       if ((ctx->Color.IndexMask & indexBits) == indexBits) {
          masking = GL_FALSE;
@@ -275,8 +275,8 @@ clear_color_buffers(GLcontext *ctx)
       }
    }
 
-   for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers[0]; i++) {
-      struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][i];
+   for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
+      struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[buf];
       if (ctx->Visual.rgbMode) {
          if (masking) {
             clear_rgba_buffer_with_masking(ctx, rb);
@@ -331,7 +331,8 @@ _swrast_Clear(GLcontext *ctx, GLbitfield buffers)
 
    /* do software clearing here */
    if (buffers) {
-      if (buffers & ctx->DrawBuffer->_ColorDrawBufferMask[0]) {
+      if ((buffers & BUFFER_BITS_COLOR)
+          && (ctx->DrawBuffer->_NumColorDrawBuffers > 0)) {
          clear_color_buffers(ctx);
       }
       if (buffers & BUFFER_BIT_DEPTH) {
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index 4d9b956f856..297940adbd6 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.c
@@ -2,7 +2,7 @@
  * Mesa 3-D graphics library
  * Version:  7.1
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,13 +27,14 @@
  */
 
 #include "main/imports.h"
-#include "bufferobj.h"
+#include "main/bufferobj.h"
 #include "main/context.h"
 #include "main/colormac.h"
 #include "main/mtypes.h"
-#include "teximage.h"
-#include "swrast.h"
+#include "main/teximage.h"
+#include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
+#include "swrast.h"
 #include "s_blend.h"
 #include "s_context.h"
 #include "s_lines.h"
@@ -87,7 +88,7 @@ _swrast_update_rasterflags( GLcontext *ctx )
     * MULTI_DRAW_BIT flag.  Also set it if we're drawing to no
     * buffers or the RGBA or CI mask disables all writes.
     */
-   if (ctx->DrawBuffer->_NumColorDrawBuffers[0] != 1) {
+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
       /* more than one color buffer designated for writing (or zero buffers) */
       rasterMask |= MULTI_DRAW_BIT;
    }
@@ -117,8 +118,8 @@ _swrast_update_rasterflags( GLcontext *ctx )
 
 
 /**
- * Examine polycon culls tate to compute the _BackfaceSign field.
- * _BackfaceSign will be 0 if no culling, -1 if culling back-faces,
+ * Examine polygon cull state to compute the _BackfaceCullSign field.
+ * _BackfaceCullSign will be 0 if no culling, -1 if culling back-faces,
  * and 1 if culling front-faces.  The Polygon FrontFace state also
  * factors in.
  */
@@ -128,31 +129,32 @@ _swrast_update_polygon( GLcontext *ctx )
    GLfloat backface_sign;
 
    if (ctx->Polygon.CullFlag) {
-      backface_sign = 1.0;
       switch (ctx->Polygon.CullFaceMode) {
       case GL_BACK:
-	 if (ctx->Polygon.FrontFace == GL_CCW)
-	    backface_sign = -1.0;
+         backface_sign = -1.0;
 	 break;
       case GL_FRONT:
-	 if (ctx->Polygon.FrontFace != GL_CCW)
-	    backface_sign = -1.0;
+         backface_sign = 1.0;
 	 break;
       case GL_FRONT_AND_BACK:
          /* fallthrough */
       default:
 	 backface_sign = 0.0;
-	 break;
       }
    }
    else {
       backface_sign = 0.0;
    }
 
-   SWRAST_CONTEXT(ctx)->_BackfaceSign = backface_sign;
+   SWRAST_CONTEXT(ctx)->_BackfaceCullSign = backface_sign;
+
+   /* This is for front/back-face determination, but not for culling */
+   SWRAST_CONTEXT(ctx)->_BackfaceSign
+      = (ctx->Polygon.FrontFace == GL_CW) ? -1.0 : 1.0;
 }
 
 
+
 /**
  * Update the _PreferPixelFog field to indicate if we need to compute
  * fog blend factors (from the fog coords) per-fragment.
@@ -500,6 +502,13 @@ _swrast_invalidate_state( GLcontext *ctx, GLbitfield new_state )
       new_state = ~0;
    }
 
+   {
+      const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
+      if (fp && (fp->Base.Parameters->StateFlags & new_state)) {
+         _mesa_load_state_parameters(ctx, fp->Base.Parameters);
+      }
+   }
+
    if (new_state & swrast->InvalidateTriangleMask)
       swrast->Triangle = _swrast_validate_triangle;
 
@@ -598,44 +607,6 @@ _swrast_update_active_attribs(GLcontext *ctx)
 }
 
 
-/**
- * Update the swrast->_ColorOutputsMask which indicates which color
- * renderbuffers (aka rendertargets) are being written to by the current
- * fragment program.
- * We also take glDrawBuffers() into account to skip outputs that are
- * set to GL_NONE.
- */
-static void
-_swrast_update_color_outputs(GLcontext *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-   swrast->_ColorOutputsMask = 0;
-   swrast->_NumColorOutputs = 0;
-
-   if (ctx->FragmentProgram._Current) {
-      const GLbitfield outputsWritten
-         = ctx->FragmentProgram._Current->Base.OutputsWritten;
-      GLuint output;
-      for (output = 0; output < ctx->Const.MaxDrawBuffers; output++) {
-         if ((outputsWritten & (1 << (FRAG_RESULT_DATA0 + output)))
-             && (fb->_NumColorDrawBuffers[output] > 0)) {
-            swrast->_ColorOutputsMask |= (1 << output);
-            swrast->_NumColorOutputs = output + 1;
-         }
-      }
-   }
-   if (swrast->_ColorOutputsMask == 0x0) {
-      /* no fragment program, or frag prog didn't write to gl_FragData[] */
-      if (fb->_NumColorDrawBuffers[0] > 0) {
-         swrast->_ColorOutputsMask = 0x1;
-         swrast->_NumColorOutputs = 1;
-      }
-   }
-}
-
-
 void
 _swrast_validate_derived( GLcontext *ctx )
 {
@@ -685,9 +656,6 @@ _swrast_validate_derived( GLcontext *ctx )
                               _NEW_TEXTURE))
          _swrast_update_active_attribs(ctx);
 
-      if (swrast->NewState & (_NEW_PROGRAM | _NEW_BUFFERS))
-         _swrast_update_color_outputs(ctx);
-
       swrast->NewState = 0;
       swrast->StateChanges = 0;
       swrast->InvalidateState = _swrast_invalidate_state;
@@ -767,6 +735,12 @@ _swrast_ResetLineStipple( GLcontext *ctx )
 }
 
 void
+_swrast_SetFacing(GLcontext *ctx, GLuint facing)
+{
+   SWRAST_CONTEXT(ctx)->PointLineFacing = facing;
+}
+
+void
 _swrast_allow_vertex_fog( GLcontext *ctx, GLboolean value )
 {
    if (SWRAST_DEBUG) {
@@ -869,6 +843,8 @@ _swrast_DestroyContext( GLcontext *ctx )
    }
 
    FREE( swrast->SpanArrays );
+   if (swrast->ZoomedArrays)
+      FREE( swrast->ZoomedArrays );
    FREE( swrast->TexelBuffer );
    FREE( swrast );
 
diff --git a/src/mesa/swrast/s_context.h b/src/mesa/swrast/s_context.h
index daa07e15783..a511d1c9a17 100644
--- a/src/mesa/swrast/s_context.h
+++ b/src/mesa/swrast/s_context.h
@@ -128,17 +128,14 @@ typedef struct
     * _swrast_validate_derived():
     */
    GLbitfield _RasterMask;
-   GLfloat _BackfaceSign;
+   GLfloat _BackfaceSign;      /** +1 or -1 */
+   GLfloat _BackfaceCullSign;  /** +1, 0, or -1 */
    GLboolean _PreferPixelFog;    /* Compute fog blend factor per fragment? */
    GLboolean _AnyTextureCombine;
    GLboolean _FogEnabled;
    GLboolean _DeferredTexture;
    GLenum _FogMode;  /* either GL_FOG_MODE or fragment program's fog mode */
 
-   /** Multiple render targets */
-   GLbitfield _ColorOutputsMask;
-   GLuint _NumColorOutputs;
-
    /** List/array of the fragment attributes to interpolate */
    GLuint _ActiveAttribs[FRAG_ATTRIB_MAX];
    /** Same info, but as a bitmask */
@@ -156,6 +153,7 @@ typedef struct
    /* Working values:
     */
    GLuint StippleCounter;    /**< Line stipple counter */
+   GLuint PointLineFacing;
    GLbitfield NewState;
    GLuint StateChanges;
    GLenum Primitive;    /* current primitive being drawn (ala glBegin) */
@@ -208,6 +206,7 @@ typedef struct
     * on some systems.
     */
    SWspanarrays *SpanArrays;
+   SWspanarrays *ZoomedArrays;  /**< For pixel zooming */
 
    /**
     * Used to buffer N GL_POINTS, instead of rendering one by one.
diff --git a/src/mesa/swrast/s_copypix.c b/src/mesa/swrast/s_copypix.c
index 7385a9942c9..fc5990b261c 100644
--- a/src/mesa/swrast/s_copypix.c
+++ b/src/mesa/swrast/s_copypix.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -26,12 +26,12 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/colormac.h"
-#include "convolve.h"
-#include "histogram.h"
-#include "image.h"
+#include "main/convolve.h"
+#include "main/histogram.h"
+#include "main/image.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "pixel.h"
+#include "main/pixel.h"
 
 #include "s_context.h"
 #include "s_depth.h"
@@ -71,13 +71,20 @@ regions_overlap(GLint srcx, GLint srcy,
    }
    else {
       /* add one pixel of slop when zooming, just to be safe */
-      if ((srcx > dstx + (width * zoomX) + 1) || (srcx + width + 1 < dstx)) {
+      if (srcx > (dstx + ((zoomX > 0.0F) ? (width * zoomX + 1.0F) : 0.0F))) {
+         /* src is completely right of dest */
+         return GL_FALSE;
+      }
+      else if (srcx + width + 1.0F < dstx + ((zoomX > 0.0F) ? 0.0F : (width * zoomX))) {
+         /* src is completely left of dest */
          return GL_FALSE;
       }
       else if ((srcy < dsty) && (srcy + height < dsty + (height * zoomY))) {
+         /* src is completely below dest */
          return GL_FALSE;
       }
       else if ((srcy > dsty) && (srcy + height > dsty + (height * zoomY))) {
+         /* src is completely above dest */
          return GL_FALSE;
       }
       else {
@@ -823,10 +830,10 @@ fast_copy_pixels(GLcontext *ctx,
    }
 
    if (type == GL_COLOR) {
-      if (dstFb->_NumColorDrawBuffers[0] != 1)
+      if (dstFb->_NumColorDrawBuffers != 1)
          return GL_FALSE;
       srcRb = srcFb->_ColorReadBuffer;
-      dstRb = dstFb->_ColorDrawBuffers[0][0];
+      dstRb = dstFb->_ColorDrawBuffers[0];
    }
    else if (type == GL_STENCIL) {
       srcRb = srcFb->_StencilBuffer;
diff --git a/src/mesa/swrast/s_depth.c b/src/mesa/swrast/s_depth.c
index a9d3e9d98e7..26e23f02d59 100644
--- a/src/mesa/swrast/s_depth.c
+++ b/src/mesa/swrast/s_depth.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.1
+ * Version:  7.2.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,7 +27,7 @@
 #include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "fbobject.h"
+#include "main/fbobject.h"
 
 #include "s_depth.h"
 #include "s_context.h"
@@ -534,15 +534,15 @@ depth_test_span( GLcontext *ctx, SWspan *span)
       if (rb->DataType == GL_UNSIGNED_SHORT) {
          GLushort zbuffer[MAX_WIDTH];
          rb->GetRow(ctx, rb, count, x, y, zbuffer);
-         passed = depth_test_span16(ctx, count, zbuffer, zValues, mask );
-         rb->PutRow(ctx, rb, count, x, y, zbuffer, NULL);
+         passed = depth_test_span16(ctx, count, zbuffer, zValues, mask);
+         rb->PutRow(ctx, rb, count, x, y, zbuffer, mask);
       }
       else {
          GLuint zbuffer[MAX_WIDTH];
          ASSERT(rb->DataType == GL_UNSIGNED_INT);
          rb->GetRow(ctx, rb, count, x, y, zbuffer);
-         passed = depth_test_span32(ctx, count, zbuffer, zValues, mask );
-         rb->PutRow(ctx, rb, count, x, y, zbuffer, NULL);
+         passed = depth_test_span32(ctx, count, zbuffer, zValues, mask);
+         rb->PutRow(ctx, rb, count, x, y, zbuffer, mask);
       }
    }
 
@@ -1080,15 +1080,15 @@ depth_test_pixels( GLcontext *ctx, SWspan *span )
       if (rb->DataType == GL_UNSIGNED_SHORT) {
          GLushort zbuffer[MAX_WIDTH];
          _swrast_get_values(ctx, rb, count, x, y, zbuffer, sizeof(GLushort));
-         depth_test_span16(ctx, count, zbuffer, z, mask );
-         rb->PutValues(ctx, rb, count, x, y, zbuffer, NULL);
+         depth_test_span16(ctx, count, zbuffer, z, mask);
+         rb->PutValues(ctx, rb, count, x, y, zbuffer, mask);
       }
       else {
          GLuint zbuffer[MAX_WIDTH];
          ASSERT(rb->DataType == GL_UNSIGNED_INT);
          _swrast_get_values(ctx, rb, count, x, y, zbuffer, sizeof(GLuint));
-         depth_test_span32(ctx, count, zbuffer, z, mask );
-         rb->PutValues(ctx, rb, count, x, y, zbuffer, NULL);
+         depth_test_span32(ctx, count, zbuffer, z, mask);
+         rb->PutValues(ctx, rb, count, x, y, zbuffer, mask);
       }
    }
 
diff --git a/src/mesa/swrast/s_depth.h b/src/mesa/swrast/s_depth.h
index 484cc73f493..36886256830 100644
--- a/src/mesa/swrast/s_depth.h
+++ b/src/mesa/swrast/s_depth.h
@@ -27,7 +27,6 @@
 #define S_DEPTH_H
 
 
-#include "main/mtypes.h"
 #include "s_context.h"
 
 
diff --git a/src/mesa/swrast/s_drawpix.c b/src/mesa/swrast/s_drawpix.c
index d1120d2cee5..7af3e3dad1c 100644
--- a/src/mesa/swrast/s_drawpix.c
+++ b/src/mesa/swrast/s_drawpix.c
@@ -24,17 +24,16 @@
 
 
 #include "main/glheader.h"
-#include "bufferobj.h"
+#include "main/bufferobj.h"
 #include "main/context.h"
-#include "convolve.h"
-#include "image.h"
+#include "main/convolve.h"
+#include "main/image.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "pixel.h"
-#include "state.h"
+#include "main/pixel.h"
+#include "main/state.h"
 
 #include "s_context.h"
-#include "s_drawpix.h"
 #include "s_span.h"
 #include "s_stencil.h"
 #include "s_zoom.h"
@@ -53,8 +52,8 @@ fast_draw_rgba_pixels(GLcontext *ctx, GLint x, GLint y,
                       const GLvoid *pixels)
 {
    const GLint imgX = x, imgY = y;
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];
-   const GLenum rbType = rb->DataType;
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
+   GLenum rbType;
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    SWspan span;
    GLboolean simpleZoom;
@@ -62,6 +61,11 @@ fast_draw_rgba_pixels(GLcontext *ctx, GLint x, GLint y,
    struct gl_pixelstore_attrib unpack;
    GLint destX, destY, drawWidth, drawHeight; /* post clipping */
 
+   if (!rb)
+      return GL_TRUE; /* no-op */
+
+   rbType = rb->DataType;
+
    if ((swrast->_RasterMask & ~CLIP_BIT) ||
        ctx->Texture._EnabledCoordUnits ||
        userUnpack->SwapBytes ||
@@ -608,8 +612,8 @@ draw_rgba_pixels( GLcontext *ctx, GLint x, GLint y,
                        IMAGE_POST_CONVOLUTION_SCALE_BIAS);
    }
 
-   if (ctx->DrawBuffer->_NumColorDrawBuffers[0] > 0 &&
-       ctx->DrawBuffer->_ColorDrawBuffers[0][0]->DataType != GL_FLOAT &&
+   if (ctx->DrawBuffer->_NumColorDrawBuffers > 0 &&
+       ctx->DrawBuffer->_ColorDrawBuffers[0]->DataType != GL_FLOAT &&
        ctx->Color.ClampFragmentColor != GL_FALSE) {
       /* need to clamp colors before applying fragment ops */
       transferOps |= IMAGE_CLAMP_BIT;
@@ -834,9 +838,11 @@ _swrast_DrawPixels( GLcontext *ctx,
    if (swrast->NewState)
       _swrast_validate_derived( ctx );
 
-   pixels = _mesa_map_drawpix_pbo(ctx, unpack, pixels);
-   if (!pixels)
-      return;
+    pixels = _mesa_map_drawpix_pbo(ctx, unpack, pixels);
+    if (!pixels) {
+       RENDER_FINISH(swrast,ctx);
+       return;
+    }
 
    switch (format) {
    case GL_STENCIL_INDEX:
@@ -873,11 +879,9 @@ _swrast_DrawPixels( GLcontext *ctx,
       /* don't return yet, clean-up */
    }
 
-end:
-
    RENDER_FINISH(swrast,ctx);
 
-   _mesa_unmap_drawpix_pbo(ctx, unpack);
+   _mesa_unmap_drapix_pbo(ctx, unpack);
 }
 
 
diff --git a/src/mesa/swrast/s_feedback.c b/src/mesa/swrast/s_feedback.c
index 31cea8e41f3..aa79531277c 100644
--- a/src/mesa/swrast/s_feedback.c
+++ b/src/mesa/swrast/s_feedback.c
@@ -26,7 +26,7 @@
 #include "main/colormac.h"
 #include "main/context.h"
 #include "main/enums.h"
-#include "feedback.h"
+#include "main/feedback.h"
 #include "main/macros.h"
 
 #include "s_context.h"
diff --git a/src/mesa/swrast/s_feedback.h b/src/mesa/swrast/s_feedback.h
index 6484f1dc751..9feab75dbb0 100644
--- a/src/mesa/swrast/s_feedback.h
+++ b/src/mesa/swrast/s_feedback.h
@@ -28,7 +28,6 @@
 #define S_FEEDBACK_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_fog.h b/src/mesa/swrast/s_fog.h
index 2346dd1734d..50760d88af7 100644
--- a/src/mesa/swrast/s_fog.h
+++ b/src/mesa/swrast/s_fog.h
@@ -28,7 +28,6 @@
 #define S_FOG_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 8eeb40f7c53..525cf9d7245 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -44,7 +44,8 @@ fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
 
-   lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
+   if (texObj)
+      lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
 
    /* XXX use a float-valued TextureSample routine here!!! */
    swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord,
@@ -68,20 +69,23 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4],
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-   const struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel];
-   const GLfloat texW = (GLfloat) texImg->WidthScale;
-   const GLfloat texH = (GLfloat) texImg->HeightScale;
+   GLfloat lambda;
    GLchan rgba[4];
 
-   GLfloat lambda
-      = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
-                               texdx[1], texdy[1], /* dt/dx, dt/dy */
-                               texdx[3], texdy[2], /* dq/dx, dq/dy */
-                               texW, texH,
-                               texcoord[0], texcoord[1], texcoord[3],
-                               1.0F / texcoord[3]) + lodBias;
+   if (texObj) {
+      const struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel];
+      const GLfloat texW = (GLfloat) texImg->WidthScale;
+      const GLfloat texH = (GLfloat) texImg->HeightScale;
 
-   lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
+      lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
+                                      texdx[1], texdy[1], /* dt/dx, dt/dy */
+                                      texdx[3], texdy[2], /* dq/dx, dq/dy */
+                                      texW, texH,
+                                      texcoord[0], texcoord[1], texcoord[3],
+                                      1.0F / texcoord[3]) + lodBias;
+
+      lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod);
+   }
 
    swrast->TextureSample[unit](ctx, texObj, 1, (const GLfloat (*)[4]) texcoord,
                                &lambda, &rgba);
@@ -124,7 +128,8 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine,
    /* if running a GLSL program (not ARB_fragment_program) */
    if (ctx->Shader.CurrentProgram) {
       /* Store front/back facing value in register FOGC.Y */
-      machine->Attribs[FRAG_ATTRIB_FOGC][col][1] = (GLfloat) ctx->_Facing;
+      machine->Attribs[FRAG_ATTRIB_FOGC][col][1] = 1.0 - span->facing;
+      /* Note FOGC.ZW is gl_PointCoord if drawing a sprite */
    }
 
    machine->CurElement = col;
@@ -171,11 +176,11 @@ run_program(GLcontext *ctx, SWspan *span, GLuint start, GLuint end)
                 * Note that colors beyond 0 and 1 will overwrite other
                 * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.
                 */
-               GLuint output;
-               for (output = 0; output < swrast->_NumColorOutputs; output++) {
-                  if (outputsWritten & (1 << (FRAG_RESULT_DATA0 + output))) {
-                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0+output][i],
-                             machine->Outputs[FRAG_RESULT_DATA0 + output]);
+               GLuint buf;
+               for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
+                  if (outputsWritten & (1 << (FRAG_RESULT_DATA0 + buf))) {
+                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
+                             machine->Outputs[FRAG_RESULT_DATA0 + buf]);
                   }
                }
             }
diff --git a/src/mesa/swrast/s_imaging.c b/src/mesa/swrast/s_imaging.c
index 73aaba1ec9f..591857c3423 100644
--- a/src/mesa/swrast/s_imaging.c
+++ b/src/mesa/swrast/s_imaging.c
@@ -27,10 +27,11 @@
  * extensions into either swrast or a sister module.  
  */
 
+#include "main/glheader.h"
+#include "main/colortab.h"
+#include "main/convolve.h"
 #include "s_context.h"
 #include "s_span.h"
-#include "colortab.h"
-#include "convolve.h"
 
 
 void
diff --git a/src/mesa/swrast/s_lines.h b/src/mesa/swrast/s_lines.h
index 6c629ca2d45..22979a02b60 100644
--- a/src/mesa/swrast/s_lines.h
+++ b/src/mesa/swrast/s_lines.h
@@ -27,7 +27,7 @@
 #ifndef S_LINES_H
 #define S_LINES_H
 
-#include "main/mtypes.h"
+#include "swrast.h"
 
 void
 _swrast_choose_line( GLcontext *ctx );
diff --git a/src/mesa/swrast/s_linetemp.h b/src/mesa/swrast/s_linetemp.h
index 1accfc67e2c..1abf8d6c7f3 100644
--- a/src/mesa/swrast/s_linetemp.h
+++ b/src/mesa/swrast/s_linetemp.h
@@ -308,6 +308,9 @@ NAME( GLcontext *ctx, const SWvertex *vert0, const SWvertex *vert1 )
    span.interpMask = interpFlags;
    span.arrayMask = SPAN_XY;
 
+   span.facing = swrast->PointLineFacing;
+
+
    /*
     * Draw
     */
diff --git a/src/mesa/swrast/s_logic.h b/src/mesa/swrast/s_logic.h
index 04ef00bb996..ba20cd7b325 100644
--- a/src/mesa/swrast/s_logic.h
+++ b/src/mesa/swrast/s_logic.h
@@ -27,7 +27,6 @@
 #define S_LOGIC_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_masking.h b/src/mesa/swrast/s_masking.h
index 688c07c7ae7..3260ca34e32 100644
--- a/src/mesa/swrast/s_masking.h
+++ b/src/mesa/swrast/s_masking.h
@@ -27,7 +27,6 @@
 #define S_MASKING_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_points.c b/src/mesa/swrast/s_points.c
index 9f52da980c7..61ff4d0b84c 100644
--- a/src/mesa/swrast/s_points.c
+++ b/src/mesa/swrast/s_points.c
@@ -27,7 +27,7 @@
 #include "main/colormac.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "texstate.h"
+#include "main/texstate.h"
 #include "s_context.h"
 #include "s_feedback.h"
 #include "s_points.h"
@@ -46,6 +46,38 @@
    } while(0)
 
 
+
+/**
+ * Get/compute the point size.
+ * The size may come from a vertex shader, or computed with attentuation
+ * or just the glPointSize value.
+ * Must also clamp to user-defined range and implmentation limits.
+ */
+static INLINE GLfloat
+get_size(const GLcontext *ctx, const SWvertex *vert, GLboolean smoothed)
+{
+   GLfloat size;
+
+   if (ctx->Point._Attenuated || ctx->VertexProgram.PointSizeEnabled) {
+      /* use vertex's point size */
+      size = vert->pointSize;
+   }
+   else {
+      /* use constant point size */
+      size = ctx->Point.Size;
+   }
+   /* always clamp to user-specified limits */
+   size = CLAMP(size, ctx->Point.MinSize, ctx->Point.MaxSize);
+   /* clamp to implementation limits */
+   if (smoothed)
+      size = CLAMP(size, ctx->Const.MinPointSizeAA, ctx->Const.MaxPointSizeAA);
+   else
+      size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
+
+   return size;
+}
+
+
 /**
  * Draw a point sprite
  */
@@ -68,23 +100,14 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       span.z = (GLuint) (vert->attrib[FRAG_ATTRIB_WPOS][2] + 0.5F);
    span.zStep = 0;
 
-   /* compute size */
-   if (ctx->Point._Attenuated || ctx->VertexProgram.PointSizeEnabled) {
-      /* use vertex's point size */
-      /* first, clamp attenuated size to the user-specifed range */
-      size = CLAMP(vert->pointSize, ctx->Point.MinSize, ctx->Point.MaxSize);
-   }
-   else {
-      /* use constant point size */
-      size = ctx->Point.Size;
-   }
-   /* clamp to non-AA implementation limits */
-   size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
+   size = get_size(ctx, vert, GL_FALSE);
 
    /* span init */
    INIT_SPAN(span, GL_POINT);
    span.interpMask = SPAN_Z | SPAN_RGBA;
 
+   span.facing = swrast->PointLineFacing;
+
    span.red   = ChanToFixed(vert->color[0]);
    span.green = ChanToFixed(vert->color[1]);
    span.blue  = ChanToFixed(vert->color[2]);
@@ -106,13 +129,13 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       s = 0.0;
       dsdx = 1.0 / size;
       if (ctx->Point.SpriteOrigin == GL_LOWER_LEFT) {
-         t0 = 0.0;
          dtdy = 1.0 / size;
+         t0 = 0.5 * dtdy;
       }
       else {
          /* GL_UPPER_LEFT */
-         t0 = 1.0;
          dtdy = -1.0 / size;
+         t0 = 1.0 + 0.5 * dtdy;
       }
 
       ATTRIB_LOOP_BEGIN
@@ -186,9 +209,10 @@ sprite_point(GLcontext *ctx, const SWvertex *vert)
       }
       else {
          /* even size */
-         xmin = (GLint) x - iRadius + 1;
+         /* 0.501 factor allows conformance to pass */
+         xmin = (GLint) (x + 0.501) - iRadius;
          xmax = xmin + iSize - 1;
-         ymin = (GLint) y - iRadius + 1;
+         ymin = (GLint) (y + 0.501) - iRadius;
          ymax = ymin + iSize - 1;
       }
 
@@ -236,18 +260,7 @@ smooth_point(GLcontext *ctx, const SWvertex *vert)
       span.z = (GLuint) (vert->attrib[FRAG_ATTRIB_WPOS][2] + 0.5F);
    span.zStep = 0;
 
-   /* compute size */
-   if (ctx->Point._Attenuated || ctx->VertexProgram.PointSizeEnabled) {
-      /* use vertex's point size */
-      /* first, clamp attenuated size to the user-specifed range */
-      size = CLAMP(vert->pointSize, ctx->Point.MinSize, ctx->Point.MaxSize);
-   }
-   else {
-      /* use constant point size */
-      size = ctx->Point.Size;
-   }
-   /* clamp to AA implementation limits */
-   size = CLAMP(size, ctx->Const.MinPointSizeAA, ctx->Const.MaxPointSizeAA);
+   size = get_size(ctx, vert, GL_TRUE);
 
    /* alpha attenuation / fade factor */
    if (ctx->Multisample._Enabled) {
@@ -269,6 +282,8 @@ smooth_point(GLcontext *ctx, const SWvertex *vert)
    span.interpMask = SPAN_Z | SPAN_RGBA;
    span.arrayMask = SPAN_COVERAGE | SPAN_MASK;
 
+   span.facing = swrast->PointLineFacing;
+
    span.red   = ChanToFixed(vert->color[0]);
    span.green = ChanToFixed(vert->color[1]);
    span.blue  = ChanToFixed(vert->color[2]);
@@ -370,22 +385,12 @@ large_point(GLcontext *ctx, const SWvertex *vert)
       span.z = (GLuint) (vert->attrib[FRAG_ATTRIB_WPOS][2] + 0.5F);
    span.zStep = 0;
 
-   /* compute size */
-   if (ctx->Point._Attenuated || ctx->VertexProgram.PointSizeEnabled) {
-      /* use vertex's point size */
-      /* first, clamp attenuated size to the user-specifed range */
-      size = CLAMP(vert->pointSize, ctx->Point.MinSize, ctx->Point.MaxSize);
-   }
-   else {
-      /* use constant point size */
-      size = ctx->Point.Size;
-   }
-   /* clamp to non-AA implementation limits */
-   size = CLAMP(size, ctx->Const.MinPointSize, ctx->Const.MaxPointSize);
+   size = get_size(ctx, vert, GL_FALSE);
 
    /* span init */
    INIT_SPAN(span, GL_POINT);
    span.arrayMask = SPAN_XY;
+   span.facing = swrast->PointLineFacing;
 
    if (ciMode) {
       span.interpMask = SPAN_Z | SPAN_INDEX;
@@ -435,9 +440,10 @@ large_point(GLcontext *ctx, const SWvertex *vert)
       }
       else {
          /* even size */
-         xmin = (GLint) x - iRadius + 1;
+         /* 0.501 factor allows conformance to pass */
+         xmin = (GLint) (x + 0.501) - iRadius;
          xmax = xmin + iSize - 1;
-         ymin = (GLint) y - iRadius + 1;
+         ymin = (GLint) (y + 0.501) - iRadius;
          ymax = ymin + iSize - 1;
       }
 
@@ -491,16 +497,21 @@ pixel_point(GLcontext *ctx, const SWvertex *vert)
 
    /* check if we need to flush */
    if (span->end >= MAX_WIDTH ||
-       (swrast->_RasterMask & (BLEND_BIT | LOGIC_OP_BIT | MASKING_BIT))) {
-      if (ciMode)
-         _swrast_write_index_span(ctx, span);
-      else
-         _swrast_write_rgba_span(ctx, span);
-      span->end = 0;
+       (swrast->_RasterMask & (BLEND_BIT | LOGIC_OP_BIT | MASKING_BIT)) ||
+       span->facing != swrast->PointLineFacing) {
+      if (span->end > 0) {
+         if (ciMode)
+            _swrast_write_index_span(ctx, span);
+         else
+            _swrast_write_rgba_span(ctx, span);
+         span->end = 0;
+      }
    }
 
    count = span->end;
 
+   span->facing = swrast->PointLineFacing;
+
    /* fragment attributes */
    if (ciMode) {
       span->array->index[count] = (GLuint) vert->attrib[FRAG_ATTRIB_CI][0];
diff --git a/src/mesa/swrast/s_points.h b/src/mesa/swrast/s_points.h
index 3fda115c0df..9e39c601efb 100644
--- a/src/mesa/swrast/s_points.h
+++ b/src/mesa/swrast/s_points.h
@@ -27,7 +27,7 @@
 #ifndef S_POINTS_H
 #define S_POINTS_H
 
-#include "main/mtypes.h"
+#include "swrast.h"
 
 extern void
 _swrast_choose_point( GLcontext *ctx );
diff --git a/src/mesa/swrast/s_readpix.c b/src/mesa/swrast/s_readpix.c
index 6186f92899e..f2630451707 100644
--- a/src/mesa/swrast/s_readpix.c
+++ b/src/mesa/swrast/s_readpix.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.0.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -24,16 +24,16 @@
 
 
 #include "main/glheader.h"
-#include "bufferobj.h"
+#include "main/bufferobj.h"
 #include "main/colormac.h"
-#include "convolve.h"
+#include "main/convolve.h"
 #include "main/context.h"
-#include "feedback.h"
-#include "image.h"
+#include "main/feedback.h"
+#include "main/image.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "pixel.h"
-#include "state.h"
+#include "main/pixel.h"
+#include "main/state.h"
 
 #include "s_context.h"
 #include "s_depth.h"
@@ -129,7 +129,8 @@ read_depth_pixels( GLcontext *ctx,
          rb->GetRow(ctx, rb, width, x, y, dest);
          /* convert range from 24-bit to 32-bit */
          for (k = 0; k < width; k++) {
-            dest[k] = (dest[k] << 8) | (dest[k] >> 24);
+            /* Note: put MSByte of 24-bit value into LSByte */
+            dest[k] = (dest[k] << 8) | ((dest[k] >> 16) & 0xff);
          }
       }
    }
@@ -569,13 +570,14 @@ _swrast_ReadPixels( GLcontext *ctx,
    /* Do all needed clipping here, so that we can forget about it later */
    if (!_mesa_clip_readpixels(ctx, &x, &y, &width, &height, &clippedPacking)) {
       /* The ReadPixels region is totally outside the window bounds */
-      goto end;
+      RENDER_FINISH(swrast, ctx);
+      return;
    }
 
    pixels = _mesa_map_readpix_pbo(ctx, &clippedPacking, pixels);
    if (!pixels)
       return;
-
+  
    switch (format) {
       case GL_COLOR_INDEX:
          read_index_pixels(ctx, x, y, width, height, type, pixels,
@@ -612,8 +614,6 @@ _swrast_ReadPixels( GLcontext *ctx,
          /* don't return yet, clean-up */
    }
 
-
-end:
    RENDER_FINISH(swrast, ctx);
 
    _mesa_unmap_readpix_pbo(ctx, &clippedPacking);
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 457dc4a6ce2..214c2a1b6f3 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -35,7 +35,7 @@
 #include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "image.h"
+#include "main/image.h"
 
 #include "s_atifragshader.h"
 #include "s_alpha.h"
@@ -64,8 +64,11 @@ _swrast_span_default_attribs(GLcontext *ctx, SWspan *span)
       const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;
       if (ctx->DrawBuffer->Visual.depthBits <= 16)
          span->z = FloatToFixed(ctx->Current.RasterPos[2] * depthMax + 0.5F);
-      else
-         span->z = (GLint) (ctx->Current.RasterPos[2] * depthMax + 0.5F);
+      else {
+         GLfloat tmpf = ctx->Current.RasterPos[2] * depthMax; 
+         tmpf = MIN2(tmpf, depthMax);
+         span->z = (GLint)tmpf;
+      }
       span->zStep = 0;
       span->interpMask |= SPAN_Z;
    }
@@ -789,6 +792,7 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
    const SWcontext *swrast = SWRAST_CONTEXT(ctx);
    const GLbitfield origInterpMask = span->interpMask;
    const GLbitfield origArrayMask = span->arrayMask;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
 
    ASSERT(span->end <= MAX_WIDTH);
    ASSERT(span->primitive == GL_POINT  ||  span->primitive == GL_LINE ||
@@ -815,7 +819,7 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
    }
 
    /* Depth bounds test */
-   if (ctx->Depth.BoundsTest && ctx->DrawBuffer->Visual.depthBits > 0) {
+   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
       if (!_swrast_depth_bounds_test(ctx, span)) {
          return;
       }
@@ -827,10 +831,10 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
       GLuint i;
       for (i = 0; i < span->end; i++) {
          if (span->array->mask[i]) {
-            assert(span->array->x[i] >= ctx->DrawBuffer->_Xmin);
-            assert(span->array->x[i] < ctx->DrawBuffer->_Xmax);
-            assert(span->array->y[i] >= ctx->DrawBuffer->_Ymin);
-            assert(span->array->y[i] < ctx->DrawBuffer->_Ymax);
+            assert(span->array->x[i] >= fb->_Xmin);
+            assert(span->array->x[i] < fb->_Xmax);
+            assert(span->array->y[i] >= fb->_Ymin);
+            assert(span->array->y[i] < fb->_Ymax);
          }
       }
    }
@@ -873,7 +877,7 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
 #endif
 
    /* we have to wait until after occlusion to do this test */
-   if (ctx->Color.DrawBuffer == GL_NONE || ctx->Color.IndexMask == 0) {
+   if (ctx->Color.IndexMask == 0) {
       /* write no pixels */
       span->arrayMask = origArrayMask;
       return;
@@ -909,22 +913,21 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
     * Write to renderbuffers
     */
    {
-      struct gl_framebuffer *fb = ctx->DrawBuffer;
-      const GLuint output = 0; /* only frag progs can write to other outputs */
-      const GLuint numDrawBuffers = fb->_NumColorDrawBuffers[output];
-      GLuint indexSave[MAX_WIDTH];
+      const GLuint numBuffers = fb->_NumColorDrawBuffers;
       GLuint buf;
 
-      if (numDrawBuffers > 1) {
-         /* save indexes for second, third renderbuffer writes */
-         _mesa_memcpy(indexSave, span->array->index,
-                      span->end * sizeof(indexSave[0]));
-      }
+      for (buf = 0; buf < numBuffers; buf++) {
+         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];
+         GLuint indexSave[MAX_WIDTH];
 
-      for (buf = 0; buf < fb->_NumColorDrawBuffers[output]; buf++) {
-         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[output][buf];
          ASSERT(rb->_BaseFormat == GL_COLOR_INDEX);
 
+         if (numBuffers > 1) {
+            /* save indexes for second, third renderbuffer writes */
+            _mesa_memcpy(indexSave, span->array->index,
+                         span->end * sizeof(indexSave[0]));
+         }
+
          if (ctx->Color.IndexLogicOpEnabled) {
             _swrast_logicop_ci_span(ctx, rb, span);
          }
@@ -999,7 +1002,7 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span)
             }
          }
 
-         if (buf + 1 < numDrawBuffers) {
+         if (buf + 1 < numBuffers) {
             /* restore original span values */
             _mesa_memcpy(span->array->index, indexSave,
                          span->end * sizeof(indexSave[0]));
@@ -1116,7 +1119,7 @@ clamp_colors(SWspan *span)
 
 /**
  * Convert the span's color arrays to the given type.
- * The only way 'output' can be greater than one is when we have a fragment
+ * The only way 'output' can be greater than zero is when we have a fragment
  * program that writes to gl_FragData[1] or higher.
  * \param output  which fragment program color output is being processed
  */
@@ -1180,8 +1183,10 @@ shade_texture_span(GLcontext *ctx, SWspan *span)
       if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {
          convert_color_type(span, GL_FLOAT, 0);
       }
-      if (span->primitive != GL_POINT || ctx->Point.PointSprite) {
-         /* for points, we populated the arrays already */
+      if (span->primitive != GL_POINT ||
+	  (span->interpMask & SPAN_RGBA) ||
+	  ctx->Point.PointSprite) {
+         /* for single-pixel points, we populated the arrays already */
          interpolate_active_attribs(ctx, span, ~0);
       }
       span->array->ChanType = GL_FLOAT;
@@ -1246,7 +1251,6 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
                              || ctx->ATIFragmentShader._Enabled);
    const GLboolean shaderOrTexture = shader || ctx->Texture._EnabledUnits;
    struct gl_framebuffer *fb = ctx->DrawBuffer;
-   GLuint output;
 
    /*
    printf("%s()  interp 0x%x  array 0x%x\n", __FUNCTION__,
@@ -1396,67 +1400,67 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span)
    /*
     * Write to renderbuffers
     */
-   /* Loop over color outputs (GL_ARB_draw_buffers) written by frag prog */
-   for (output = 0; output < swrast->_NumColorOutputs; output++) {
-      if (swrast->_ColorOutputsMask & (1 << output)) {
-        const GLuint numDrawBuffers = fb->_NumColorDrawBuffers[output];
-        GLchan rgbaSave[MAX_WIDTH][4];
-        GLuint buf;
-
-        ASSERT(numDrawBuffers > 0);
-
-        if (fb->_ColorDrawBuffers[output][0]->DataType
-            != span->array->ChanType || output > 0) {
-           convert_color_type(span,
-                              fb->_ColorDrawBuffers[output][0]->DataType,
-                              output);
-        }
-
-        if (numDrawBuffers > 1) {
-           /* save colors for second, third renderbuffer writes */
-           _mesa_memcpy(rgbaSave, span->array->rgba,
-                        4 * span->end * sizeof(GLchan));
-        }
-
-        /* Loop over renderbuffers (i.e. GL_FRONT_AND_BACK) */
-        for (buf = 0; buf < numDrawBuffers; buf++) {
-           struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[output][buf];
-           ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB);
-
-           if (ctx->Color._LogicOpEnabled) {
-              _swrast_logicop_rgba_span(ctx, rb, span);
-           }
-           else if (ctx->Color.BlendEnabled) {
-              _swrast_blend_span(ctx, rb, span);
-           }
-
-           if (colorMask != 0xffffffff) {
-              _swrast_mask_rgba_span(ctx, rb, span);
-           }
-
-           if (span->arrayMask & SPAN_XY) {
-              /* array of pixel coords */
-              ASSERT(rb->PutValues);
-              rb->PutValues(ctx, rb, span->end,
-                            span->array->x, span->array->y,
-                            span->array->rgba, span->array->mask);
-           }
-           else {
-              /* horizontal run of pixels */
-              ASSERT(rb->PutRow);
-              rb->PutRow(ctx, rb, span->end, span->x, span->y,
-                         span->array->rgba,
-                         span->writeAll ? NULL: span->array->mask);
-           }
-
-           if (buf + 1 < numDrawBuffers) {
-              /* restore original span values */
-              _mesa_memcpy(span->array->rgba, rgbaSave,
-                           4 * span->end * sizeof(GLchan));
-           }
-        } /* for buf */
-      } /* if output is written to */
-   } /* for output */
+   {
+      const GLuint numBuffers = fb->_NumColorDrawBuffers;
+      const GLboolean multiFragOutputs = numBuffers > 1;
+      GLuint buf;
+
+      for (buf = 0; buf < numBuffers; buf++) {
+         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];
+
+         /* color[fragOutput] will be written to buffer[buf] */
+
+         if (rb) {
+            GLchan rgbaSave[MAX_WIDTH][4];
+            const GLuint fragOutput = multiFragOutputs ? buf : 0;
+
+            if (rb->DataType != span->array->ChanType || fragOutput > 0) {
+               convert_color_type(span, rb->DataType, fragOutput);
+            }
+
+            if (!multiFragOutputs && numBuffers > 1) {
+               /* save colors for second, third renderbuffer writes */
+               _mesa_memcpy(rgbaSave, span->array->rgba,
+                            4 * span->end * sizeof(GLchan));
+            }
+
+            ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB);
+
+            if (ctx->Color._LogicOpEnabled) {
+               _swrast_logicop_rgba_span(ctx, rb, span);
+            }
+            else if (ctx->Color.BlendEnabled) {
+               _swrast_blend_span(ctx, rb, span);
+            }
+
+            if (colorMask != 0xffffffff) {
+               _swrast_mask_rgba_span(ctx, rb, span);
+            }
+
+            if (span->arrayMask & SPAN_XY) {
+               /* array of pixel coords */
+               ASSERT(rb->PutValues);
+               rb->PutValues(ctx, rb, span->end,
+                             span->array->x, span->array->y,
+                             span->array->rgba, span->array->mask);
+            }
+            else {
+               /* horizontal run of pixels */
+               ASSERT(rb->PutRow);
+               rb->PutRow(ctx, rb, span->end, span->x, span->y,
+                          span->array->rgba,
+                          span->writeAll ? NULL: span->array->mask);
+            }
+
+            if (!multiFragOutputs && numBuffers > 1) {
+               /* restore original span values */
+               _mesa_memcpy(span->array->rgba, rgbaSave,
+                            4 * span->end * sizeof(GLchan));
+            }
+
+         } /* if rb */
+      } /* for buf */
+   }
 
 end:
    /* restore these values before returning */
diff --git a/src/mesa/swrast/s_span.h b/src/mesa/swrast/s_span.h
index 6b814fc8fb1..c4b47df58ff 100644
--- a/src/mesa/swrast/s_span.h
+++ b/src/mesa/swrast/s_span.h
@@ -27,7 +27,6 @@
 #define S_SPAN_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_stencil.c b/src/mesa/swrast/s_stencil.c
index cdb7e4669ce..c925922463e 100644
--- a/src/mesa/swrast/s_stencil.c
+++ b/src/mesa/swrast/s_stencil.c
@@ -923,6 +923,8 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
       ASSERT(rb->DataType == GL_UNSIGNED_BYTE);
       _swrast_get_values(ctx, rb, n, x, y, stencil, sizeof(GLubyte));
 
+      _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));          
+
       (void) do_stencil_test(ctx, face, n, stencil, mask);
 
       if (ctx->Depth.Test == GL_FALSE) {
@@ -930,11 +932,12 @@ stencil_and_ztest_pixels( GLcontext *ctx, SWspan *span, GLuint face )
                           n, stencil, mask);
       }
       else {
-         _mesa_memcpy(origMask, mask, n * sizeof(GLubyte));
+         GLubyte tmpMask[MAX_WIDTH]; 
+         _mesa_memcpy(tmpMask, mask, n * sizeof(GLubyte));
 
          _swrast_depth_test_span(ctx, span);
 
-         compute_pass_fail_masks(n, origMask, mask, passMask, failMask);
+         compute_pass_fail_masks(n, tmpMask, mask, passMask, failMask);
 
          if (ctx->Stencil.ZFailFunc[face] != GL_KEEP) {
             apply_stencil_op(ctx, ctx->Stencil.ZFailFunc[face], face,
diff --git a/src/mesa/swrast/s_stencil.h b/src/mesa/swrast/s_stencil.h
index 742f8d4c946..cd6cbc57b0b 100644
--- a/src/mesa/swrast/s_stencil.h
+++ b/src/mesa/swrast/s_stencil.h
@@ -27,7 +27,6 @@
 #define S_STENCIL_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_texcombine.c b/src/mesa/swrast/s_texcombine.c
index b7724f7b20a..632d650007e 100644
--- a/src/mesa/swrast/s_texcombine.c
+++ b/src/mesa/swrast/s_texcombine.c
@@ -26,9 +26,10 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/colormac.h"
+#include "main/image.h"
 #include "main/imports.h"
 #include "main/macros.h"
-#include "image.h"
+#include "main/pixel.h"
 
 #include "s_context.h"
 #include "s_texcombine.h"
diff --git a/src/mesa/swrast/s_texcombine.h b/src/mesa/swrast/s_texcombine.h
index 20cd2bd8ad1..9ed96efb879 100644
--- a/src/mesa/swrast/s_texcombine.h
+++ b/src/mesa/swrast/s_texcombine.h
@@ -27,7 +27,6 @@
 #define S_TEXCOMBINE_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 extern void
diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c
index bb0fc823e7e..6b1dfd5d502 100644
--- a/src/mesa/swrast/s_texfilter.c
+++ b/src/mesa/swrast/s_texfilter.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,12 +27,22 @@
 #include "main/context.h"
 #include "main/colormac.h"
 #include "main/imports.h"
-#include "texformat.h"
+#include "main/texformat.h"
 
 #include "s_context.h"
 #include "s_texfilter.h"
 
 
+/*
+ * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
+ * see 1-pixel bands of improperly weighted linear-filtered textures.
+ * The tests/texwrap.c demo is a good test.
+ * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
+ * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
+ */
+#define FRAC(f)  ((f) - IFLOOR(f))
+
+
 /**
  * Constants for integer linear interpolation.
  */
@@ -213,272 +223,281 @@ lerp_rgba_3d(GLchan result[4], GLfloat a, GLfloat b, GLfloat c,
 
 
 /**
- * Compute the remainder of a divided by b, but be careful with
- * negative values so that GL_REPEAT mode works right.
+ * If A is a signed integer, A % B doesn't give the right value for A < 0
+ * (in terms of texture repeat).  Just casting to unsigned fixes that.
  */
-static INLINE GLint
-repeat_remainder(GLint a, GLint b)
-{
-   if (a >= 0)
-      return a % b;
-   else
-      return (a + 1) % b + b - 1;
-}
+#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B))
 
 
 /**
  * Used to compute texel locations for linear sampling.
  * Input:
  *    wrapMode = GL_REPEAT, GL_CLAMP, GL_CLAMP_TO_EDGE, GL_CLAMP_TO_BORDER
- *    S = texcoord in [0,1]
- *    SIZE = width (or height or depth) of texture
+ *    s = texcoord in [0,1]
+ *    size = width (or height or depth) of texture
  * Output:
- *    U = texcoord in [0, width]
- *    I0, I1 = two nearest texel indexes
+ *    i0, i1 = returns two nearest texel indexes
+ *    weight = returns blend factor between texels
  */
-#define COMPUTE_LINEAR_TEXEL_LOCATIONS(wrapMode, S, U, SIZE, I0, I1)	\
-{									\
-   switch (wrapMode) {							\
-   case GL_REPEAT:							\
-      U = S * SIZE - 0.5F;						\
-      if (img->_IsPowerOfTwo) {						\
-         I0 = IFLOOR(U) & (SIZE - 1);					\
-         I1 = (I0 + 1) & (SIZE - 1);					\
-      }									\
-      else {								\
-         I0 = repeat_remainder(IFLOOR(U), SIZE);			\
-         I1 = repeat_remainder(I0 + 1, SIZE);				\
-      }									\
-      break;								\
-   case GL_CLAMP_TO_EDGE:						\
-      if (S <= 0.0F)							\
-         U = 0.0F;							\
-      else if (S >= 1.0F)						\
-         U = (GLfloat) SIZE;						\
-      else								\
-         U = S * SIZE;							\
-      U -= 0.5F;							\
-      I0 = IFLOOR(U);							\
-      I1 = I0 + 1;							\
-      if (I0 < 0)							\
-         I0 = 0;							\
-      if (I1 >= (GLint) SIZE)						\
-         I1 = SIZE - 1;							\
-      break;								\
-   case GL_CLAMP_TO_BORDER:						\
-      {									\
-         const GLfloat min = -1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         if (S <= min)							\
-            U = min * SIZE;						\
-         else if (S >= max)						\
-            U = max * SIZE;						\
-         else								\
-            U = S * SIZE;						\
-         U -= 0.5F;							\
-         I0 = IFLOOR(U);						\
-         I1 = I0 + 1;							\
-      }									\
-      break;								\
-   case GL_MIRRORED_REPEAT:						\
-      {									\
-         const GLint flr = IFLOOR(S);					\
-         if (flr & 1)							\
-            U = 1.0F - (S - (GLfloat) flr);	/* flr is odd */	\
-         else								\
-            U = S - (GLfloat) flr;		/* flr is even */	\
-         U = (U * SIZE) - 0.5F;						\
-         I0 = IFLOOR(U);						\
-         I1 = I0 + 1;							\
-         if (I0 < 0)							\
-            I0 = 0;							\
-         if (I1 >= (GLint) SIZE)					\
-            I1 = SIZE - 1;						\
-      }									\
-      break;								\
-   case GL_MIRROR_CLAMP_EXT:						\
-      U = FABSF(S);							\
-      if (U >= 1.0F)							\
-         U = (GLfloat) SIZE;						\
-      else								\
-         U *= SIZE;							\
-      U -= 0.5F;							\
-      I0 = IFLOOR(U);							\
-      I1 = I0 + 1;							\
-      break;								\
-   case GL_MIRROR_CLAMP_TO_EDGE_EXT:					\
-      U = FABSF(S);							\
-      if (U >= 1.0F)							\
-         U = (GLfloat) SIZE;						\
-      else								\
-         U *= SIZE;							\
-      U -= 0.5F;							\
-      I0 = IFLOOR(U);							\
-      I1 = I0 + 1;							\
-      if (I0 < 0)							\
-         I0 = 0;							\
-      if (I1 >= (GLint) SIZE)						\
-         I1 = SIZE - 1;							\
-      break;								\
-   case GL_MIRROR_CLAMP_TO_BORDER_EXT:					\
-      {									\
-         const GLfloat min = -1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         U = FABSF(S);							\
-         if (U <= min)							\
-            U = min * SIZE;						\
-         else if (U >= max)						\
-            U = max * SIZE;						\
-         else								\
-            U *= SIZE;							\
-         U -= 0.5F;							\
-         I0 = IFLOOR(U);						\
-         I1 = I0 + 1;							\
-      }									\
-      break;								\
-   case GL_CLAMP:							\
-      if (S <= 0.0F)							\
-         U = 0.0F;							\
-      else if (S >= 1.0F)						\
-         U = (GLfloat) SIZE;						\
-      else								\
-         U = S * SIZE;							\
-      U -= 0.5F;							\
-      I0 = IFLOOR(U);							\
-      I1 = I0 + 1;							\
-      break;								\
-   default:								\
-      _mesa_problem(ctx, "Bad wrap mode");				\
-   }									\
+static INLINE void
+linear_texel_locations(GLenum wrapMode,
+                       const struct gl_texture_image *img,
+                       GLint size, GLfloat s,
+                       GLint *i0, GLint *i1, GLfloat *weight)
+{
+   GLfloat u;
+   switch (wrapMode) {
+   case GL_REPEAT:
+      u = s * size - 0.5F;
+      if (img->_IsPowerOfTwo) {
+         *i0 = IFLOOR(u) & (size - 1);
+         *i1 = (*i0 + 1) & (size - 1);
+      }
+      else {
+         *i0 = REMAINDER(IFLOOR(u), size);
+         *i1 = REMAINDER(*i0 + 1, size);
+      }
+      break;
+   case GL_CLAMP_TO_EDGE:
+      if (s <= 0.0F)
+         u = 0.0F;
+      else if (s >= 1.0F)
+         u = (GLfloat) size;
+      else
+         u = s * size;
+      u -= 0.5F;
+      *i0 = IFLOOR(u);
+      *i1 = *i0 + 1;
+      if (*i0 < 0)
+         *i0 = 0;
+      if (*i1 >= (GLint) size)
+         *i1 = size - 1;
+      break;
+   case GL_CLAMP_TO_BORDER:
+      {
+         const GLfloat min = -1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         if (s <= min)
+            u = min * size;
+         else if (s >= max)
+            u = max * size;
+         else
+            u = s * size;
+         u -= 0.5F;
+         *i0 = IFLOOR(u);
+         *i1 = *i0 + 1;
+      }
+      break;
+   case GL_MIRRORED_REPEAT:
+      {
+         const GLint flr = IFLOOR(s);
+         if (flr & 1)
+            u = 1.0F - (s - (GLfloat) flr);
+         else
+            u = s - (GLfloat) flr;
+         u = (u * size) - 0.5F;
+         *i0 = IFLOOR(u);
+         *i1 = *i0 + 1;
+         if (*i0 < 0)
+            *i0 = 0;
+         if (*i1 >= (GLint) size)
+            *i1 = size - 1;
+      }
+      break;
+   case GL_MIRROR_CLAMP_EXT:
+      u = FABSF(s);
+      if (u >= 1.0F)
+         u = (GLfloat) size;
+      else
+         u *= size;
+      u -= 0.5F;
+      *i0 = IFLOOR(u);
+      *i1 = *i0 + 1;
+      break;
+   case GL_MIRROR_CLAMP_TO_EDGE_EXT:
+      u = FABSF(s);
+      if (u >= 1.0F)
+         u = (GLfloat) size;
+      else
+         u *= size;
+      u -= 0.5F;
+      *i0 = IFLOOR(u);
+      *i1 = *i0 + 1;
+      if (*i0 < 0)
+         *i0 = 0;
+      if (*i1 >= (GLint) size)
+         *i1 = size - 1;
+      break;
+   case GL_MIRROR_CLAMP_TO_BORDER_EXT:
+      {
+         const GLfloat min = -1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         u = FABSF(s);
+         if (u <= min)
+            u = min * size;
+         else if (u >= max)
+            u = max * size;
+         else
+            u *= size;
+         u -= 0.5F;
+         *i0 = IFLOOR(u);
+         *i1 = *i0 + 1;
+      }
+      break;
+   case GL_CLAMP:
+      if (s <= 0.0F)
+         u = 0.0F;
+      else if (s >= 1.0F)
+         u = (GLfloat) size;
+      else
+         u = s * size;
+      u -= 0.5F;
+      *i0 = IFLOOR(u);
+      *i1 = *i0 + 1;
+      break;
+   default:
+      _mesa_problem(NULL, "Bad wrap mode");
+      u = 0.0F;
+   }
+   *weight = FRAC(u);
 }
 
 
 /**
  * Used to compute texel location for nearest sampling.
  */
-#define COMPUTE_NEAREST_TEXEL_LOCATION(wrapMode, S, SIZE, I)		\
-{									\
-   switch (wrapMode) {							\
-   case GL_REPEAT:							\
-      /* s limited to [0,1) */						\
-      /* i limited to [0,size-1] */					\
-      I = IFLOOR(S * SIZE);						\
-      if (img->_IsPowerOfTwo)						\
-         I &= (SIZE - 1);						\
-      else								\
-         I = repeat_remainder(I, SIZE);					\
-      break;								\
-   case GL_CLAMP_TO_EDGE:						\
-      {									\
-         /* s limited to [min,max] */					\
-         /* i limited to [0, size-1] */					\
-         const GLfloat min = 1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         if (S < min)							\
-            I = 0;							\
-         else if (S > max)						\
-            I = SIZE - 1;						\
-         else								\
-            I = IFLOOR(S * SIZE);					\
-      }									\
-      break;								\
-   case GL_CLAMP_TO_BORDER:						\
-      {									\
-         /* s limited to [min,max] */					\
-         /* i limited to [-1, size] */					\
-         const GLfloat min = -1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         if (S <= min)							\
-            I = -1;							\
-         else if (S >= max)						\
-            I = SIZE;							\
-         else								\
-            I = IFLOOR(S * SIZE);					\
-      }									\
-      break;								\
-   case GL_MIRRORED_REPEAT:						\
-      {									\
-         const GLfloat min = 1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         const GLint flr = IFLOOR(S);					\
-         GLfloat u;							\
-         if (flr & 1)							\
-            u = 1.0F - (S - (GLfloat) flr);	/* flr is odd */	\
-         else								\
-            u = S - (GLfloat) flr;		/* flr is even */	\
-         if (u < min)							\
-            I = 0;							\
-         else if (u > max)						\
-            I = SIZE - 1;						\
-         else								\
-            I = IFLOOR(u * SIZE);					\
-      }									\
-      break;								\
-   case GL_MIRROR_CLAMP_EXT:						\
-      {									\
-         /* s limited to [0,1] */					\
-         /* i limited to [0,size-1] */					\
-         const GLfloat u = FABSF(S);					\
-         if (u <= 0.0F)							\
-            I = 0;							\
-         else if (u >= 1.0F)						\
-            I = SIZE - 1;						\
-         else								\
-            I = IFLOOR(u * SIZE);					\
-      }									\
-      break;								\
-   case GL_MIRROR_CLAMP_TO_EDGE_EXT:					\
-      {									\
-         /* s limited to [min,max] */					\
-         /* i limited to [0, size-1] */					\
-         const GLfloat min = 1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         const GLfloat u = FABSF(S);					\
-         if (u < min)							\
-            I = 0;							\
-         else if (u > max)						\
-            I = SIZE - 1;						\
-         else								\
-            I = IFLOOR(u * SIZE);					\
-      }									\
-      break;								\
-   case GL_MIRROR_CLAMP_TO_BORDER_EXT:					\
-      {									\
-         /* s limited to [min,max] */					\
-         /* i limited to [0, size-1] */					\
-         const GLfloat min = -1.0F / (2.0F * SIZE);			\
-         const GLfloat max = 1.0F - min;				\
-         const GLfloat u = FABSF(S);					\
-         if (u < min)							\
-            I = -1;							\
-         else if (u > max)						\
-            I = SIZE;							\
-         else								\
-            I = IFLOOR(u * SIZE);					\
-      }									\
-      break;								\
-   case GL_CLAMP:							\
-      /* s limited to [0,1] */						\
-      /* i limited to [0,size-1] */					\
-      if (S <= 0.0F)							\
-         I = 0;								\
-      else if (S >= 1.0F)						\
-         I = SIZE - 1;							\
-      else								\
-         I = IFLOOR(S * SIZE);						\
-      break;								\
-   default:								\
-      _mesa_problem(ctx, "Bad wrap mode");				\
-   }									\
+static INLINE GLint
+nearest_texel_location(GLenum wrapMode,
+                       const struct gl_texture_image *img,
+                       GLint size, GLfloat s)
+{
+   GLint i;
+
+   switch (wrapMode) {
+   case GL_REPEAT:
+      /* s limited to [0,1) */
+      /* i limited to [0,size-1] */
+      i = IFLOOR(s * size);
+      if (img->_IsPowerOfTwo)
+         i &= (size - 1);
+      else
+         i = REMAINDER(i, size);
+      return i;
+   case GL_CLAMP_TO_EDGE:
+      {
+         /* s limited to [min,max] */
+         /* i limited to [0, size-1] */
+         const GLfloat min = 1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         if (s < min)
+            i = 0;
+         else if (s > max)
+            i = size - 1;
+         else
+            i = IFLOOR(s * size);
+      }
+      return i;
+   case GL_CLAMP_TO_BORDER:
+      {
+         /* s limited to [min,max] */
+         /* i limited to [-1, size] */
+         const GLfloat min = -1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         if (s <= min)
+            i = -1;
+         else if (s >= max)
+            i = size;
+         else
+            i = IFLOOR(s * size);
+      }
+      return i;
+   case GL_MIRRORED_REPEAT:
+      {
+         const GLfloat min = 1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         const GLint flr = IFLOOR(s);
+         GLfloat u;
+         if (flr & 1)
+            u = 1.0F - (s - (GLfloat) flr);
+         else
+            u = s - (GLfloat) flr;
+         if (u < min)
+            i = 0;
+         else if (u > max)
+            i = size - 1;
+         else
+            i = IFLOOR(u * size);
+      }
+      return i;
+   case GL_MIRROR_CLAMP_EXT:
+      {
+         /* s limited to [0,1] */
+         /* i limited to [0,size-1] */
+         const GLfloat u = FABSF(s);
+         if (u <= 0.0F)
+            i = 0;
+         else if (u >= 1.0F)
+            i = size - 1;
+         else
+            i = IFLOOR(u * size);
+      }
+      return i;
+   case GL_MIRROR_CLAMP_TO_EDGE_EXT:
+      {
+         /* s limited to [min,max] */
+         /* i limited to [0, size-1] */
+         const GLfloat min = 1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         const GLfloat u = FABSF(s);
+         if (u < min)
+            i = 0;
+         else if (u > max)
+            i = size - 1;
+         else
+            i = IFLOOR(u * size);
+      }
+      return i;
+   case GL_MIRROR_CLAMP_TO_BORDER_EXT:
+      {
+         /* s limited to [min,max] */
+         /* i limited to [0, size-1] */
+         const GLfloat min = -1.0F / (2.0F * size);
+         const GLfloat max = 1.0F - min;
+         const GLfloat u = FABSF(s);
+         if (u < min)
+            i = -1;
+         else if (u > max)
+            i = size;
+         else
+            i = IFLOOR(u * size);
+      }
+      return i;
+   case GL_CLAMP:
+      /* s limited to [0,1] */
+      /* i limited to [0,size-1] */
+      if (s <= 0.0F)
+         i = 0;
+      else if (s >= 1.0F)
+         i = size - 1;
+      else
+         i = IFLOOR(s * size);
+      return i;
+   default:
+      _mesa_problem(NULL, "Bad wrap mode");
+      return 0;
+   }
 }
 
 
 /* Power of two image sizes only */
-#define COMPUTE_LINEAR_REPEAT_TEXEL_LOCATION(S, U, SIZE, I0, I1)	\
-{									\
-   U = S * SIZE - 0.5F;							\
-   I0 = IFLOOR(U) & (SIZE - 1);						\
-   I1 = (I0 + 1) & (SIZE - 1);						\
+static INLINE void
+linear_repeat_texel_location(GLuint size, GLfloat s,
+                             GLint *i0, GLint *i1, GLfloat *weight)
+{
+   GLfloat u = s * size - 0.5F;
+   *i0 = IFLOOR(u) & (size - 1);
+   *i1 = (*i0 + 1) & (size - 1);
+   *weight = FRAC(u);
 }
 
 
@@ -521,17 +540,6 @@ nearest_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)
 
 
 /*
- * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
- * see 1-pixel bands of improperly weighted linear-filtered textures.
- * The tests/texwrap.c demo is a good test.
- * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
- * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
- */
-#define FRAC(f)  ((f) - IFLOOR(f))
-
-
-
-/*
  * Bitflags for texture border color sampling.
  */
 #define I0BIT   1
@@ -543,7 +551,7 @@ nearest_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)
 
 
 
-/*
+/**
  * The lambda[] array values are always monotonic.  Either the whole span
  * will be minified, magnified, or split between the two.  This function
  * determines the subranges in [0, n-1] that are to be minified or magnified.
@@ -656,10 +664,10 @@ compute_min_mag_ranges(const struct gl_texture_object *tObj,
 /*                    1-D Texture Sampling Functions                  */
 /**********************************************************************/
 
-/*
+/**
  * Return the texture sample for coordinate (s) using GL_NEAREST filter.
  */
-static void
+static INLINE void
 sample_1d_nearest(GLcontext *ctx,
                   const struct gl_texture_object *tObj,
                   const struct gl_texture_image *img,
@@ -667,7 +675,7 @@ sample_1d_nearest(GLcontext *ctx,
 {
    const GLint width = img->Width2;  /* without border, power of two */
    GLint i;
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoord[0], width, i);
+   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
    /* skip over the border, if any */
    i += img->Border;
    if (i < 0 || i >= (GLint) img->Width) {
@@ -680,10 +688,10 @@ sample_1d_nearest(GLcontext *ctx,
 }
 
 
-/*
+/**
  * Return the texture sample for coordinate (s) using GL_LINEAR filter.
  */
-static void
+static INLINE void
 sample_1d_linear(GLcontext *ctx,
                  const struct gl_texture_object *tObj,
                  const struct gl_texture_image *img,
@@ -691,12 +699,11 @@ sample_1d_linear(GLcontext *ctx,
 {
    const GLint width = img->Width2;
    GLint i0, i1;
-   GLfloat u;
    GLbitfield useBorderColor = 0x0;
    GLfloat a;
    GLchan t0[4], t1[4];  /* texels */
 
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoord[0], u, width, i0, i1);
+   linear_texel_locations(tObj->WrapS, img, width, texcoord[0], &i0, &i1, &a);
 
    if (img->Border) {
       i0 += img->Border;
@@ -721,7 +728,6 @@ sample_1d_linear(GLcontext *ctx,
       img->FetchTexelc(img, i1, 0, 0, t1);
    }
 
-   a = FRAC(u);
    lerp_rgba(rgba, a, t0, t1);
 }
 
@@ -781,7 +787,6 @@ sample_1d_nearest_mipmap_linear(GLcontext *ctx,
 }
 
 
-
 static void
 sample_1d_linear_mipmap_linear(GLcontext *ctx,
                                const struct gl_texture_object *tObj,
@@ -807,7 +812,7 @@ sample_1d_linear_mipmap_linear(GLcontext *ctx,
 }
 
 
-
+/** Sample 1D texture, nearest filtering for both min/magnification */
 static void
 sample_nearest_1d( GLcontext *ctx,
                    const struct gl_texture_object *tObj, GLuint n,
@@ -817,13 +822,13 @@ sample_nearest_1d( GLcontext *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_1d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-
+/** Sample 1D texture, linear filtering for both min/magnification */
 static void
 sample_linear_1d( GLcontext *ctx,
                   const struct gl_texture_object *tObj, GLuint n,
@@ -833,17 +838,13 @@ sample_linear_1d( GLcontext *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_1d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-/*
- * Given an (s) texture coordinate and lambda (level of detail) value,
- * return a texture sample.
- *
- */
+/** Sample 1D texture, using lambda to choose between min/magnification */
 static void
 sample_lambda_1d( GLcontext *ctx,
                   const struct gl_texture_object *tObj, GLuint n,
@@ -920,7 +921,7 @@ sample_lambda_1d( GLcontext *ctx,
 /**********************************************************************/
 
 
-/*
+/**
  * Return the texture sample for coordinate (s,t) using GL_NEAREST filter.
  */
 static INLINE void
@@ -935,8 +936,8 @@ sample_2d_nearest(GLcontext *ctx,
    GLint i, j;
    (void) ctx;
 
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoord[0], width,  i);
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapT, texcoord[1], height, j);
+   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
 
    /* skip over the border, if any */
    i += img->Border;
@@ -952,7 +953,6 @@ sample_2d_nearest(GLcontext *ctx,
 }
 
 
-
 /**
  * Return the texture sample for coordinate (s,t) using GL_LINEAR filter.
  * New sampling code contributed by Lynn Quam <quam@ai.sri.com>.
@@ -968,12 +968,11 @@ sample_2d_linear(GLcontext *ctx,
    const GLint height = img->Height2;
    GLint i0, j0, i1, j1;
    GLbitfield useBorderColor = 0x0;
-   GLfloat u, v;
    GLfloat a, b;
    GLchan t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */
 
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoord[0], u, width,  i0, i1);
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapT, texcoord[1], v, height, j0, j1);
+   linear_texel_locations(tObj->WrapS, img, width, texcoord[0],  &i0, &i1, &a);
+   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
 
    if (img->Border) {
       i0 += img->Border;
@@ -1014,13 +1013,11 @@ sample_2d_linear(GLcontext *ctx,
       img->FetchTexelc(img, i1, j1, 0, t11);
    }
 
-   a = FRAC(u);
-   b = FRAC(v);
    lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);
 }
 
 
-/*
+/**
  * As above, but we know WRAP_S == REPEAT and WRAP_T == REPEAT.
  * We don't have to worry about the texture border.
  */
@@ -1034,8 +1031,7 @@ sample_2d_linear_repeat(GLcontext *ctx,
    const GLint width = img->Width2;
    const GLint height = img->Height2;
    GLint i0, j0, i1, j1;
-   GLfloat u, v;
-   GLfloat a, b;
+   GLfloat wi, wj;
    GLchan t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */
 
    (void) ctx;
@@ -1043,24 +1039,21 @@ sample_2d_linear_repeat(GLcontext *ctx,
    ASSERT(tObj->WrapS == GL_REPEAT);
    ASSERT(tObj->WrapT == GL_REPEAT);
    ASSERT(img->Border == 0);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
+   ASSERT(img->TexFormat->BaseFormat != GL_COLOR_INDEX);
    ASSERT(img->_IsPowerOfTwo);
 
-   COMPUTE_LINEAR_REPEAT_TEXEL_LOCATION(texcoord[0], u, width,  i0, i1);
-   COMPUTE_LINEAR_REPEAT_TEXEL_LOCATION(texcoord[1], v, height, j0, j1);
+   linear_repeat_texel_location(width,  texcoord[0], &i0, &i1, &wi);
+   linear_repeat_texel_location(height, texcoord[1], &j0, &j1, &wj);
 
    img->FetchTexelc(img, i0, j0, 0, t00);
    img->FetchTexelc(img, i1, j0, 0, t10);
    img->FetchTexelc(img, i0, j1, 0, t01);
    img->FetchTexelc(img, i1, j1, 0, t11);
 
-   a = FRAC(u);
-   b = FRAC(v);
-   lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);
+   lerp_rgba_2d(rgba, wi, wj, t00, t10, t01, t11);
 }
 
 
-
 static void
 sample_2d_nearest_mipmap_nearest(GLcontext *ctx,
                                  const struct gl_texture_object *tObj,
@@ -1075,7 +1068,6 @@ sample_2d_nearest_mipmap_nearest(GLcontext *ctx,
 }
 
 
-
 static void
 sample_2d_linear_mipmap_nearest(GLcontext *ctx,
                                 const struct gl_texture_object *tObj,
@@ -1091,7 +1083,6 @@ sample_2d_linear_mipmap_nearest(GLcontext *ctx,
 }
 
 
-
 static void
 sample_2d_nearest_mipmap_linear(GLcontext *ctx,
                                 const struct gl_texture_object *tObj,
@@ -1117,8 +1108,6 @@ sample_2d_nearest_mipmap_linear(GLcontext *ctx,
 }
 
 
-
-/* Trilinear filtering */
 static void
 sample_2d_linear_mipmap_linear( GLcontext *ctx,
                                 const struct gl_texture_object *tObj,
@@ -1145,10 +1134,10 @@ sample_2d_linear_mipmap_linear( GLcontext *ctx,
 
 
 static void
-sample_2d_linear_mipmap_linear_repeat( GLcontext *ctx,
-                                       const struct gl_texture_object *tObj,
-                                       GLuint n, const GLfloat texcoord[][4],
-                                       const GLfloat lambda[], GLchan rgba[][4] )
+sample_2d_linear_mipmap_linear_repeat(GLcontext *ctx,
+                                      const struct gl_texture_object *tObj,
+                                      GLuint n, const GLfloat texcoord[][4],
+                                      const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    ASSERT(lambda != NULL);
@@ -1163,55 +1152,59 @@ sample_2d_linear_mipmap_linear_repeat( GLcontext *ctx,
       else {
          GLchan t0[4], t1[4];  /* texels */
          const GLfloat f = FRAC(lambda[i]);
-         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
+         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level  ],
+                                 texcoord[i], t0);
+         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level+1],
+                                 texcoord[i], t1);
          lerp_rgba(rgba[i], f, t0, t1);
       }
    }
 }
 
 
+/** Sample 2D texture, nearest filtering for both min/magnification */
 static void
-sample_nearest_2d( GLcontext *ctx,
-                   const struct gl_texture_object *tObj, GLuint n,
-                   const GLfloat texcoords[][4],
-                   const GLfloat lambda[], GLchan rgba[][4] )
+sample_nearest_2d(GLcontext *ctx,
+                  const struct gl_texture_object *tObj, GLuint n,
+                  const GLfloat texcoords[][4],
+                  const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_2d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-
+/** Sample 2D texture, linear filtering for both min/magnification */
 static void
-sample_linear_2d( GLcontext *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4],
-                  const GLfloat lambda[], GLchan rgba[][4] )
+sample_linear_2d(GLcontext *ctx,
+                 const struct gl_texture_object *tObj, GLuint n,
+                 const GLfloat texcoords[][4],
+                 const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
    if (tObj->WrapS == GL_REPEAT &&
        tObj->WrapT == GL_REPEAT &&
-       image->_IsPowerOfTwo) {
-      for (i=0;i<n;i++) {
+       image->_IsPowerOfTwo &&
+       image->Border == 0) {
+      for (i = 0; i < n; i++) {
          sample_2d_linear_repeat(ctx, tObj, image, texcoords[i], rgba[i]);
       }
    }
    else {
-      for (i=0;i<n;i++) {
+      for (i = 0; i < n; i++) {
          sample_2d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
       }
    }
 }
 
 
-/*
+/**
  * Optimized 2-D texture sampling:
  *    S and T wrap mode == GL_REPEAT
  *    GL_NEAREST min/mag filter
@@ -1220,10 +1213,10 @@ sample_linear_2d( GLcontext *ctx,
  *    Format = GL_RGB
  */
 static void
-opt_sample_rgb_2d( GLcontext *ctx,
-                   const struct gl_texture_object *tObj,
-                   GLuint n, const GLfloat texcoords[][4],
-                   const GLfloat lambda[], GLchan rgba[][4] )
+opt_sample_rgb_2d(GLcontext *ctx,
+                  const struct gl_texture_object *tObj,
+                  GLuint n, const GLfloat texcoords[][4],
+                  const GLfloat lambda[], GLchan rgba[][4])
 {
    const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];
    const GLfloat width = (GLfloat) img->Width;
@@ -1237,7 +1230,7 @@ opt_sample_rgb_2d( GLcontext *ctx,
    ASSERT(tObj->WrapS==GL_REPEAT);
    ASSERT(tObj->WrapT==GL_REPEAT);
    ASSERT(img->Border==0);
-   ASSERT(img->_BaseFormat==GL_RGB);
+   ASSERT(img->TexFormat->MesaFormat==MESA_FORMAT_RGB);
    ASSERT(img->_IsPowerOfTwo);
 
    for (k=0; k<n; k++) {
@@ -1252,7 +1245,7 @@ opt_sample_rgb_2d( GLcontext *ctx,
 }
 
 
-/*
+/**
  * Optimized 2-D texture sampling:
  *    S and T wrap mode == GL_REPEAT
  *    GL_NEAREST min/mag filter
@@ -1261,10 +1254,10 @@ opt_sample_rgb_2d( GLcontext *ctx,
  *    Format = GL_RGBA
  */
 static void
-opt_sample_rgba_2d( GLcontext *ctx,
-                    const struct gl_texture_object *tObj,
-                    GLuint n, const GLfloat texcoords[][4],
-                    const GLfloat lambda[], GLchan rgba[][4] )
+opt_sample_rgba_2d(GLcontext *ctx,
+                   const struct gl_texture_object *tObj,
+                   GLuint n, const GLfloat texcoords[][4],
+                   const GLfloat lambda[], GLchan rgba[][4])
 {
    const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];
    const GLfloat width = (GLfloat) img->Width;
@@ -1278,7 +1271,7 @@ opt_sample_rgba_2d( GLcontext *ctx,
    ASSERT(tObj->WrapS==GL_REPEAT);
    ASSERT(tObj->WrapT==GL_REPEAT);
    ASSERT(img->Border==0);
-   ASSERT(img->_BaseFormat==GL_RGBA);
+   ASSERT(img->TexFormat->MesaFormat==MESA_FORMAT_RGBA);
    ASSERT(img->_IsPowerOfTwo);
 
    for (i = 0; i < n; i++) {
@@ -1291,15 +1284,12 @@ opt_sample_rgba_2d( GLcontext *ctx,
 }
 
 
-/*
- * Given an array of texture coordinate and lambda (level of detail)
- * values, return an array of texture sample.
- */
+/** Sample 2D texture, using lambda to choose between min/magnification */
 static void
-sample_lambda_2d( GLcontext *ctx,
-                  const struct gl_texture_object *tObj,
-                  GLuint n, const GLfloat texcoords[][4],
-                  const GLfloat lambda[], GLchan rgba[][4] )
+sample_lambda_2d(GLcontext *ctx,
+                 const struct gl_texture_object *tObj,
+                 GLuint n, const GLfloat texcoords[][4],
+                 const GLfloat lambda[], GLchan rgba[][4])
 {
    const struct gl_texture_image *tImg = tObj->Image[0][tObj->BaseLevel];
    GLuint minStart, minEnd;  /* texels with minification */
@@ -1308,7 +1298,7 @@ sample_lambda_2d( GLcontext *ctx,
    const GLboolean repeatNoBorderPOT = (tObj->WrapS == GL_REPEAT)
       && (tObj->WrapT == GL_REPEAT)
       && (tImg->Border == 0 && (tImg->Width == tImg->RowStride))
-      && (tImg->_BaseFormat != GL_COLOR_INDEX)
+      && (tImg->TexFormat->BaseFormat != GL_COLOR_INDEX)
       && tImg->_IsPowerOfTwo;
 
    ASSERT(lambda != NULL);
@@ -1323,16 +1313,10 @@ sample_lambda_2d( GLcontext *ctx,
          if (repeatNoBorderPOT) {
             switch (tImg->TexFormat->MesaFormat) {
             case MESA_FORMAT_RGB:
-            case MESA_FORMAT_RGB888:
-            /*case MESA_FORMAT_BGR888:*/
                opt_sample_rgb_2d(ctx, tObj, m, texcoords + minStart,
                                  NULL, rgba + minStart);
                break;
             case MESA_FORMAT_RGBA:
-            case MESA_FORMAT_RGBA8888:
-            case MESA_FORMAT_ARGB8888:
-            /*case MESA_FORMAT_ABGR8888:*/
-            /*case MESA_FORMAT_BGRA8888:*/
 	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + minStart,
                                   NULL, rgba + minStart);
                break;
@@ -1386,16 +1370,10 @@ sample_lambda_2d( GLcontext *ctx,
          if (repeatNoBorderPOT) {
             switch (tImg->TexFormat->MesaFormat) {
             case MESA_FORMAT_RGB:
-            case MESA_FORMAT_RGB888:
-            /*case MESA_FORMAT_BGR888:*/
                opt_sample_rgb_2d(ctx, tObj, m, texcoords + magStart,
                                  NULL, rgba + magStart);
                break;
             case MESA_FORMAT_RGBA:
-            case MESA_FORMAT_RGBA8888:
-            case MESA_FORMAT_ARGB8888:
-            /*case MESA_FORMAT_ABGR8888:*/
-            /*case MESA_FORMAT_BGRA8888:*/
 	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + magStart,
                                   NULL, rgba + magStart);
                break;
@@ -1425,10 +1403,10 @@ sample_lambda_2d( GLcontext *ctx,
 /*                    3-D Texture Sampling Functions                  */
 /**********************************************************************/
 
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
  */
-static void
+static INLINE void
 sample_3d_nearest(GLcontext *ctx,
                   const struct gl_texture_object *tObj,
                   const struct gl_texture_image *img,
@@ -1441,9 +1419,9 @@ sample_3d_nearest(GLcontext *ctx,
    GLint i, j, k;
    (void) ctx;
 
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoord[0], width,  i);
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapT, texcoord[1], height, j);
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapR, texcoord[2], depth,  k);
+   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
+   k = nearest_texel_location(tObj->WrapR, img, depth, texcoord[2]);
 
    if (i < 0 || i >= (GLint) img->Width ||
        j < 0 || j >= (GLint) img->Height ||
@@ -1457,8 +1435,7 @@ sample_3d_nearest(GLcontext *ctx,
 }
 
 
-
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
  */
 static void
@@ -1473,14 +1450,13 @@ sample_3d_linear(GLcontext *ctx,
    const GLint depth = img->Depth2;
    GLint i0, j0, k0, i1, j1, k1;
    GLbitfield useBorderColor = 0x0;
-   GLfloat u, v, w;
    GLfloat a, b, c;
    GLchan t000[4], t010[4], t001[4], t011[4];
    GLchan t100[4], t110[4], t101[4], t111[4];
 
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoord[0], u, width,  i0, i1);
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapT, texcoord[1], v, height, j0, j1);
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapR, texcoord[2], w, depth,  k0, k1);
+   linear_texel_locations(tObj->WrapS, img, width, texcoord[0],  &i0, &i1, &a);
+   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
+   linear_texel_locations(tObj->WrapR, img, depth, texcoord[2],  &k0, &k1, &c);
 
    if (img->Border) {
       i0 += img->Border;
@@ -1552,14 +1528,10 @@ sample_3d_linear(GLcontext *ctx,
    }
 
    /* trilinear interpolation of samples */
-   a = FRAC(u);
-   b = FRAC(v);
-   c = FRAC(w);
    lerp_rgba_3d(rgba, a, b, c, t000, t100, t010, t110, t001, t101, t011, t111);
 }
 
 
-
 static void
 sample_3d_nearest_mipmap_nearest(GLcontext *ctx,
                                  const struct gl_texture_object *tObj,
@@ -1639,6 +1611,7 @@ sample_3d_linear_mipmap_linear(GLcontext *ctx,
 }
 
 
+/** Sample 3D texture, nearest filtering for both min/magnification */
 static void
 sample_nearest_3d(GLcontext *ctx,
                   const struct gl_texture_object *tObj, GLuint n,
@@ -1648,37 +1621,34 @@ sample_nearest_3d(GLcontext *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_3d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-
+/** Sample 3D texture, linear filtering for both min/magnification */
 static void
-sample_linear_3d( GLcontext *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4],
-		  const GLfloat lambda[], GLchan rgba[][4] )
+sample_linear_3d(GLcontext *ctx,
+                 const struct gl_texture_object *tObj, GLuint n,
+                 const GLfloat texcoords[][4],
+		 const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_3d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-/*
- * Given an (s,t,r) texture coordinate and lambda (level of detail) value,
- * return a texture sample.
- */
+/** Sample 3D texture, using lambda to choose between min/magnification */
 static void
-sample_lambda_3d( GLcontext *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4], const GLfloat lambda[],
-                  GLchan rgba[][4] )
+sample_lambda_3d(GLcontext *ctx,
+                 const struct gl_texture_object *tObj, GLuint n,
+                 const GLfloat texcoords[][4], const GLfloat lambda[],
+                 GLchan rgba[][4])
 {
    GLuint minStart, minEnd;  /* texels with minification */
    GLuint magStart, magEnd;  /* texels with magnification */
@@ -1872,8 +1842,19 @@ sample_cube_nearest_mipmap_nearest(GLcontext *ctx,
    for (i = 0; i < n; i++) {
       const struct gl_texture_image **images;
       GLfloat newCoord[4];
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
+      GLint level;
       images = choose_cube_face(tObj, texcoord[i], newCoord);
+
+      /* XXX we actually need to recompute lambda here based on the newCoords.
+       * But we would need the texcoords of adjacent fragments to compute that
+       * properly, and we don't have those here.
+       * For now, do an approximation:  subtracting 1 from the chosen mipmap
+       * level seems to work in some test cases.
+       * The same adjustment is done in the next few functions.
+      */
+      level = nearest_mipmap_level(tObj, lambda[i]);
+      level = MAX2(level - 1, 0);
+
       sample_2d_nearest(ctx, tObj, images[level], newCoord, rgba[i]);
    }
 }
@@ -1891,6 +1872,7 @@ sample_cube_linear_mipmap_nearest(GLcontext *ctx,
       const struct gl_texture_image **images;
       GLfloat newCoord[4];
       GLint level = nearest_mipmap_level(tObj, lambda[i]);
+      level = MAX2(level - 1, 0); /* see comment above */
       images = choose_cube_face(tObj, texcoord[i], newCoord);
       sample_2d_linear(ctx, tObj, images[level], newCoord, rgba[i]);
    }
@@ -1909,6 +1891,7 @@ sample_cube_nearest_mipmap_linear(GLcontext *ctx,
       const struct gl_texture_image **images;
       GLfloat newCoord[4];
       GLint level = linear_mipmap_level(tObj, lambda[i]);
+      level = MAX2(level - 1, 0); /* see comment above */
       images = choose_cube_face(tObj, texcoord[i], newCoord);
       if (level >= tObj->_MaxLevel) {
          sample_2d_nearest(ctx, tObj, images[tObj->_MaxLevel],
@@ -1937,6 +1920,7 @@ sample_cube_linear_mipmap_linear(GLcontext *ctx,
       const struct gl_texture_image **images;
       GLfloat newCoord[4];
       GLint level = linear_mipmap_level(tObj, lambda[i]);
+      level = MAX2(level - 1, 0); /* see comment above */
       images = choose_cube_face(tObj, texcoord[i], newCoord);
       if (level >= tObj->_MaxLevel) {
          sample_2d_linear(ctx, tObj, images[tObj->_MaxLevel],
@@ -1953,11 +1937,12 @@ sample_cube_linear_mipmap_linear(GLcontext *ctx,
 }
 
 
+/** Sample cube texture, using lambda to choose between min/magnification */
 static void
-sample_lambda_cube( GLcontext *ctx,
-		    const struct gl_texture_object *tObj, GLuint n,
-		    const GLfloat texcoords[][4], const GLfloat lambda[],
-		    GLchan rgba[][4])
+sample_lambda_cube(GLcontext *ctx,
+		   const struct gl_texture_object *tObj, GLuint n,
+		   const GLfloat texcoords[][4], const GLfloat lambda[],
+		   GLchan rgba[][4])
 {
    GLuint minStart, minEnd;  /* texels with minification */
    GLuint magStart, magEnd;  /* texels with magnification */
@@ -2033,50 +2018,56 @@ sample_lambda_cube( GLcontext *ctx,
 static INLINE GLint
 clamp_rect_coord_nearest(GLenum wrapMode, GLfloat coord, GLint max)
 {
-   if (wrapMode == GL_CLAMP) {
+   switch (wrapMode) {
+   case GL_CLAMP:
       return IFLOOR( CLAMP(coord, 0.0F, max - 1) );
-   }
-   else if (wrapMode == GL_CLAMP_TO_EDGE) {
+   case GL_CLAMP_TO_EDGE:
       return IFLOOR( CLAMP(coord, 0.5F, max - 0.5F) );
-   }
-   else {
+   case GL_CLAMP_TO_BORDER:
       return IFLOOR( CLAMP(coord, -0.5F, max + 0.5F) );
+   default:
+      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_nearest");
+      return 0;
    }
 }
 
 
-/*
+/**
  * As above, but GL_LINEAR filtering.
  */
 static INLINE void
 clamp_rect_coord_linear(GLenum wrapMode, GLfloat coord, GLint max,
-                        GLint *i0out, GLint *i1out)
+                        GLint *i0out, GLint *i1out, GLfloat *weight)
 {
    GLfloat fcol;
    GLint i0, i1;
-   if (wrapMode == GL_CLAMP) {
+   switch (wrapMode) {
+   case GL_CLAMP:
       /* Not exactly what the spec says, but it matches NVIDIA output */
       fcol = CLAMP(coord - 0.5F, 0.0, max-1);
       i0 = IFLOOR(fcol);
       i1 = i0 + 1;
-   }
-   else if (wrapMode == GL_CLAMP_TO_EDGE) {
+      break;
+   case GL_CLAMP_TO_EDGE:
       fcol = CLAMP(coord, 0.5F, max - 0.5F);
       fcol -= 0.5F;
       i0 = IFLOOR(fcol);
       i1 = i0 + 1;
       if (i1 > max - 1)
          i1 = max - 1;
-   }
-   else {
-      ASSERT(wrapMode == GL_CLAMP_TO_BORDER);
+      break;
+   case GL_CLAMP_TO_BORDER:
       fcol = CLAMP(coord, -0.5F, max + 0.5F);
       fcol -= 0.5F;
       i0 = IFLOOR(fcol);
       i1 = i0 + 1;
+   default:
+      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_linear");
+      i0 = i1 = 0;
    }
    *i0out = i0;
    *i1out = i1;
+   *weight = FRAC(fcol);
 }
 
 
@@ -2087,10 +2078,8 @@ sample_nearest_rect(GLcontext *ctx,
                     GLchan rgba[][4])
 {
    const struct gl_texture_image *img = tObj->Image[0][0];
-   const GLfloat width = (GLfloat) img->Width;
-   const GLfloat height = (GLfloat) img->Height;
-   const GLint width_minus_1 = img->Width - 1;
-   const GLint height_minus_1 = img->Height - 1;
+   const GLint width = img->Width;
+   const GLint height = img->Height;
    GLuint i;
 
    (void) ctx;
@@ -2102,13 +2091,13 @@ sample_nearest_rect(GLcontext *ctx,
    ASSERT(tObj->WrapT == GL_CLAMP ||
           tObj->WrapT == GL_CLAMP_TO_EDGE ||
           tObj->WrapT == GL_CLAMP_TO_BORDER);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
+   ASSERT(img->TexFormat->BaseFormat != GL_COLOR_INDEX);
 
    for (i = 0; i < n; i++) {
       GLint row, col;
       col = clamp_rect_coord_nearest(tObj->WrapS, texcoords[i][0], width);
       row = clamp_rect_coord_nearest(tObj->WrapT, texcoords[i][1], height);
-      if (col < 0 || col > width_minus_1 || row < 0 || row > height_minus_1)
+      if (col < 0 || col >= width || row < 0 || row >= height)
          COPY_CHAN4(rgba[i], tObj->_BorderChan);
       else
          img->FetchTexelc(img, col, row, 0, rgba[i]);
@@ -2123,10 +2112,8 @@ sample_linear_rect(GLcontext *ctx,
 		   const GLfloat lambda[], GLchan rgba[][4])
 {
    const struct gl_texture_image *img = tObj->Image[0][0];
-   const GLfloat width = (GLfloat) img->Width;
-   const GLfloat height = (GLfloat) img->Height;
-   const GLint width_minus_1 = img->Width - 1;
-   const GLint height_minus_1 = img->Height - 1;
+   const GLint width = img->Width;
+   const GLint height = img->Height;
    GLuint i;
 
    (void) ctx;
@@ -2138,66 +2125,24 @@ sample_linear_rect(GLcontext *ctx,
    ASSERT(tObj->WrapT == GL_CLAMP ||
           tObj->WrapT == GL_CLAMP_TO_EDGE ||
           tObj->WrapT == GL_CLAMP_TO_BORDER);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
+   ASSERT(img->TexFormat->BaseFormat != GL_COLOR_INDEX);
 
-   /* XXX lots of opportunity for optimization in this loop */
    for (i = 0; i < n; i++) {
-      GLfloat frow, fcol;
       GLint i0, j0, i1, j1;
       GLchan t00[4], t01[4], t10[4], t11[4];
       GLfloat a, b;
       GLbitfield useBorderColor = 0x0;
 
-      /* NOTE: we DO NOT use [0, 1] texture coordinates! */
-      if (tObj->WrapS == GL_CLAMP) {
-         /* Not exactly what the spec says, but it matches NVIDIA output */
-         fcol = CLAMP(texcoords[i][0] - 0.5F, 0.0, width_minus_1);
-         i0 = IFLOOR(fcol);
-         i1 = i0 + 1;
-      }
-      else if (tObj->WrapS == GL_CLAMP_TO_EDGE) {
-         fcol = CLAMP(texcoords[i][0], 0.5F, width - 0.5F);
-         fcol -= 0.5F;
-         i0 = IFLOOR(fcol);
-         i1 = i0 + 1;
-         if (i1 > width_minus_1)
-            i1 = width_minus_1;
-      }
-      else {
-         ASSERT(tObj->WrapS == GL_CLAMP_TO_BORDER);
-         fcol = CLAMP(texcoords[i][0], -0.5F, width + 0.5F);
-         fcol -= 0.5F;
-         i0 = IFLOOR(fcol);
-         i1 = i0 + 1;
-      }
-
-      if (tObj->WrapT == GL_CLAMP) {
-         /* Not exactly what the spec says, but it matches NVIDIA output */
-         frow = CLAMP(texcoords[i][1] - 0.5F, 0.0, width_minus_1);
-         j0 = IFLOOR(frow);
-         j1 = j0 + 1;
-      }
-      else if (tObj->WrapT == GL_CLAMP_TO_EDGE) {
-         frow = CLAMP(texcoords[i][1], 0.5F, height - 0.5F);
-         frow -= 0.5F;
-         j0 = IFLOOR(frow);
-         j1 = j0 + 1;
-         if (j1 > height_minus_1)
-            j1 = height_minus_1;
-      }
-      else {
-         ASSERT(tObj->WrapT == GL_CLAMP_TO_BORDER);
-         frow = CLAMP(texcoords[i][1], -0.5F, height + 0.5F);
-         frow -= 0.5F;
-         j0 = IFLOOR(frow);
-         j1 = j0 + 1;
-      }
+      clamp_rect_coord_linear(tObj->WrapS, texcoords[i][0], width,
+                              &i0, &i1, &a);
+      clamp_rect_coord_linear(tObj->WrapT, texcoords[i][1], height,
+                              &j0, &j1, &b);
 
       /* compute integer rows/columns */
-      if (i0 < 0 || i0 > width_minus_1)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 > width_minus_1)   useBorderColor |= I1BIT;
-      if (j0 < 0 || j0 > height_minus_1)  useBorderColor |= J0BIT;
-      if (j1 < 0 || j1 > height_minus_1)  useBorderColor |= J1BIT;
+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
+      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;
+      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;
 
       /* get four texel samples */
       if (useBorderColor & (I0BIT | J0BIT))
@@ -2220,20 +2165,17 @@ sample_linear_rect(GLcontext *ctx,
       else
          img->FetchTexelc(img, i1, j1, 0, t11);
 
-      /* compute interpolants */
-      a = FRAC(fcol);
-      b = FRAC(frow);
-
       lerp_rgba_2d(rgba[i], a, b, t00, t10, t01, t11);
    }
 }
 
 
+/** Sample Rect texture, using lambda to choose between min/magnification */
 static void
-sample_lambda_rect( GLcontext *ctx,
-		    const struct gl_texture_object *tObj, GLuint n,
-		    const GLfloat texcoords[][4], const GLfloat lambda[],
-		    GLchan rgba[][4])
+sample_lambda_rect(GLcontext *ctx,
+		   const struct gl_texture_object *tObj, GLuint n,
+		   const GLfloat texcoords[][4], const GLfloat lambda[],
+		   GLchan rgba[][4])
 {
    GLuint minStart, minEnd, magStart, magEnd;
 
@@ -2245,22 +2187,22 @@ sample_lambda_rect( GLcontext *ctx,
 
    if (minStart < minEnd) {
       if (tObj->MinFilter == GL_NEAREST) {
-         sample_nearest_rect( ctx, tObj, minEnd - minStart,
-                              texcoords + minStart, NULL, rgba + minStart);
+         sample_nearest_rect(ctx, tObj, minEnd - minStart,
+                             texcoords + minStart, NULL, rgba + minStart);
       }
       else {
-         sample_linear_rect( ctx, tObj, minEnd - minStart,
-                             texcoords + minStart, NULL, rgba + minStart);
+         sample_linear_rect(ctx, tObj, minEnd - minStart,
+                            texcoords + minStart, NULL, rgba + minStart);
       }
    }
    if (magStart < magEnd) {
       if (tObj->MagFilter == GL_NEAREST) {
-         sample_nearest_rect( ctx, tObj, magEnd - magStart,
-                              texcoords + magStart, NULL, rgba + magStart);
+         sample_nearest_rect(ctx, tObj, magEnd - magStart,
+                             texcoords + magStart, NULL, rgba + magStart);
       }
       else {
-         sample_linear_rect( ctx, tObj, magEnd - magStart,
-                             texcoords + magStart, NULL, rgba + magStart);
+         sample_linear_rect(ctx, tObj, magEnd - magStart,
+                            texcoords + magStart, NULL, rgba + magStart);
       }
    }
 }
@@ -2271,7 +2213,7 @@ sample_lambda_rect( GLcontext *ctx,
 /*                2D Texture Array Sampling Functions                 */
 /**********************************************************************/
 
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
  */
 static void
@@ -2288,8 +2230,8 @@ sample_2d_array_nearest(GLcontext *ctx,
    GLint array;
    (void) ctx;
 
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoord[0], width,  i);
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapT, texcoord[1], height, j);
+   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
+   j = nearest_texel_location(tObj->WrapT, img, height, texcoord[1]);
    array = clamp_rect_coord_nearest(tObj->WrapR, texcoord[2], depth);
 
    if (i < 0 || i >= (GLint) img->Width ||
@@ -2304,8 +2246,7 @@ sample_2d_array_nearest(GLcontext *ctx,
 }
 
 
-
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
  */
 static void
@@ -2321,12 +2262,11 @@ sample_2d_array_linear(GLcontext *ctx,
    GLint i0, j0, i1, j1;
    GLint array;
    GLbitfield useBorderColor = 0x0;
-   GLfloat u, v;
    GLfloat a, b;
    GLchan t00[4], t01[4], t10[4], t11[4];
 
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoord[0], u, width,  i0, i1);
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapT, texcoord[1], v, height, j0, j1);
+   linear_texel_locations(tObj->WrapS, img, width,  texcoord[0], &i0, &i1, &a);
+   linear_texel_locations(tObj->WrapT, img, height, texcoord[1], &j0, &j1, &b);
    array = clamp_rect_coord_nearest(tObj->WrapR, texcoord[2], depth);
 
    if (array < 0 || array >= depth) {
@@ -2374,19 +2314,16 @@ sample_2d_array_linear(GLcontext *ctx,
       }
       
       /* trilinear interpolation of samples */
-      a = FRAC(u);
-      b = FRAC(v);
       lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);
    }
 }
 
 
-
 static void
 sample_2d_array_nearest_mipmap_nearest(GLcontext *ctx,
                                        const struct gl_texture_object *tObj,
                                        GLuint n, const GLfloat texcoord[][4],
-                                       const GLfloat lambda[], GLchan rgba[][4] )
+                                       const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    for (i = 0; i < n; i++) {
@@ -2430,8 +2367,10 @@ sample_2d_array_nearest_mipmap_linear(GLcontext *ctx,
       else {
          GLchan t0[4], t1[4];  /* texels */
          const GLfloat f = FRAC(lambda[i]);
-         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
+         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level  ],
+                                 texcoord[i], t0);
+         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level+1],
+                                 texcoord[i], t1);
          lerp_rgba(rgba[i], f, t0, t1);
       }
    }
@@ -2440,9 +2379,9 @@ sample_2d_array_nearest_mipmap_linear(GLcontext *ctx,
 
 static void
 sample_2d_array_linear_mipmap_linear(GLcontext *ctx,
-                               const struct gl_texture_object *tObj,
-                               GLuint n, const GLfloat texcoord[][4],
-                               const GLfloat lambda[], GLchan rgba[][4])
+                                     const struct gl_texture_object *tObj,
+                                     GLuint n, const GLfloat texcoord[][4],
+                                     const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    ASSERT(lambda != NULL);
@@ -2455,30 +2394,34 @@ sample_2d_array_linear_mipmap_linear(GLcontext *ctx,
       else {
          GLchan t0[4], t1[4];  /* texels */
          const GLfloat f = FRAC(lambda[i]);
-         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
+         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level  ],
+                                texcoord[i], t0);
+         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level+1],
+                                texcoord[i], t1);
          lerp_rgba(rgba[i], f, t0, t1);
       }
    }
 }
 
 
+/** Sample 2D Array texture, nearest filtering for both min/magnification */
 static void
 sample_nearest_2d_array(GLcontext *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4], const GLfloat lambda[],
-                  GLchan rgba[][4])
+                        const struct gl_texture_object *tObj, GLuint n,
+                        const GLfloat texcoords[][4], const GLfloat lambda[],
+                        GLchan rgba[][4])
 {
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_2d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
 
+/** Sample 2D Array texture, linear filtering for both min/magnification */
 static void
 sample_linear_2d_array(GLcontext *ctx,
                        const struct gl_texture_object *tObj, GLuint n,
@@ -2488,16 +2431,13 @@ sample_linear_2d_array(GLcontext *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_2d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-/*
- * Given an (s,t,r) texture coordinate and lambda (level of detail) value,
- * return a texture sample.
- */
+/** Sample 2D Array texture, using lambda to choose between min/magnification */
 static void
 sample_lambda_2d_array(GLcontext *ctx,
                        const struct gl_texture_object *tObj, GLuint n,
@@ -2527,8 +2467,10 @@ sample_lambda_2d_array(GLcontext *ctx,
                                    texcoords[i], rgba[i]);
          break;
       case GL_NEAREST_MIPMAP_NEAREST:
-         sample_2d_array_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                                lambda + minStart, rgba + minStart);
+         sample_2d_array_nearest_mipmap_nearest(ctx, tObj, m,
+                                                texcoords + minStart,
+                                                lambda + minStart,
+                                                rgba + minStart);
          break;
       case GL_LINEAR_MIPMAP_NEAREST:
          sample_2d_array_linear_mipmap_nearest(ctx, tObj, m, 
@@ -2537,8 +2479,10 @@ sample_lambda_2d_array(GLcontext *ctx,
                                                rgba + minStart);
          break;
       case GL_NEAREST_MIPMAP_LINEAR:
-         sample_2d_array_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                               lambda + minStart, rgba + minStart);
+         sample_2d_array_nearest_mipmap_linear(ctx, tObj, m,
+                                               texcoords + minStart,
+                                               lambda + minStart,
+                                               rgba + minStart);
          break;
       case GL_LINEAR_MIPMAP_LINEAR:
          sample_2d_array_linear_mipmap_linear(ctx, tObj, m, 
@@ -2579,7 +2523,7 @@ sample_lambda_2d_array(GLcontext *ctx,
 /*                1D Texture Array Sampling Functions                 */
 /**********************************************************************/
 
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
  */
 static void
@@ -2595,7 +2539,7 @@ sample_1d_array_nearest(GLcontext *ctx,
    GLint array;
    (void) ctx;
 
-   COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoord[0], width,  i);
+   i = nearest_texel_location(tObj->WrapS, img, width, texcoord[0]);
    array = clamp_rect_coord_nearest(tObj->WrapT, texcoord[1], height);
 
    if (i < 0 || i >= (GLint) img->Width ||
@@ -2609,8 +2553,7 @@ sample_1d_array_nearest(GLcontext *ctx,
 }
 
 
-
-/*
+/**
  * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
  */
 static void
@@ -2625,11 +2568,10 @@ sample_1d_array_linear(GLcontext *ctx,
    GLint i0, i1;
    GLint array;
    GLbitfield useBorderColor = 0x0;
-   GLfloat u;
    GLfloat a;
    GLchan t0[4], t1[4];
 
-   COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoord[0], u, width,  i0, i1);
+   linear_texel_locations(tObj->WrapS, img, width, texcoord[0], &i0, &i1, &a);
    array = clamp_rect_coord_nearest(tObj->WrapT, texcoord[1], height);
 
    if (img->Border) {
@@ -2659,17 +2601,15 @@ sample_1d_array_linear(GLcontext *ctx,
    }
 
    /* bilinear interpolation of samples */
-   a = FRAC(u);
    lerp_rgba(rgba, a, t0, t1);
 }
 
 
-
 static void
 sample_1d_array_nearest_mipmap_nearest(GLcontext *ctx,
                                        const struct gl_texture_object *tObj,
                                        GLuint n, const GLfloat texcoord[][4],
-                                       const GLfloat lambda[], GLchan rgba[][4] )
+                                       const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    for (i = 0; i < n; i++) {
@@ -2723,9 +2663,9 @@ sample_1d_array_nearest_mipmap_linear(GLcontext *ctx,
 
 static void
 sample_1d_array_linear_mipmap_linear(GLcontext *ctx,
-                               const struct gl_texture_object *tObj,
-                               GLuint n, const GLfloat texcoord[][4],
-                               const GLfloat lambda[], GLchan rgba[][4])
+                                     const struct gl_texture_object *tObj,
+                                     GLuint n, const GLfloat texcoord[][4],
+                                     const GLfloat lambda[], GLchan rgba[][4])
 {
    GLuint i;
    ASSERT(lambda != NULL);
@@ -2746,22 +2686,23 @@ sample_1d_array_linear_mipmap_linear(GLcontext *ctx,
 }
 
 
+/** Sample 1D Array texture, nearest filtering for both min/magnification */
 static void
 sample_nearest_1d_array(GLcontext *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4], const GLfloat lambda[],
-                  GLchan rgba[][4])
+                        const struct gl_texture_object *tObj, GLuint n,
+                        const GLfloat texcoords[][4], const GLfloat lambda[],
+                        GLchan rgba[][4])
 {
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_1d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-
+/** Sample 1D Array texture, linear filtering for both min/magnification */
 static void
 sample_linear_1d_array(GLcontext *ctx,
                        const struct gl_texture_object *tObj, GLuint n,
@@ -2771,16 +2712,13 @@ sample_linear_1d_array(GLcontext *ctx,
    GLuint i;
    struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
    (void) lambda;
-   for (i=0;i<n;i++) {
+   for (i = 0; i < n; i++) {
       sample_1d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);
    }
 }
 
 
-/*
- * Given an (s,t,r) texture coordinate and lambda (level of detail) value,
- * return a texture sample.
- */
+/** Sample 1D Array texture, using lambda to choose between min/magnification */
 static void
 sample_lambda_1d_array(GLcontext *ctx,
                        const struct gl_texture_object *tObj, GLuint n,
@@ -2856,9 +2794,7 @@ sample_lambda_1d_array(GLcontext *ctx,
 }
 
 
-
-
-/*
+/**
  * Sample a shadow/depth texture.
  */
 static void
@@ -2880,8 +2816,8 @@ sample_depth_texture( GLcontext *ctx,
 
    (void) lambda;
 
-   ASSERT(tObj->Image[0][tObj->BaseLevel]->_BaseFormat == GL_DEPTH_COMPONENT ||
-          tObj->Image[0][tObj->BaseLevel]->_BaseFormat == GL_DEPTH_STENCIL_EXT);
+   ASSERT(img->TexFormat->BaseFormat == GL_DEPTH_COMPONENT ||
+          img->TexFormat->BaseFormat == GL_DEPTH_STENCIL_EXT);
 
    ASSERT(tObj->Target == GL_TEXTURE_1D ||
           tObj->Target == GL_TEXTURE_2D ||
@@ -2908,31 +2844,32 @@ sample_depth_texture( GLcontext *ctx,
             break;
             
          case GL_TEXTURE_1D:
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoords[i][0],
-                                           width, col);
+            col = nearest_texel_location(tObj->WrapS, img, width,
+                                         texcoords[i][0]);
             row = 0;
             slice = 0;
             break;
 
          case GL_TEXTURE_2D:
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoords[i][0],
-                                           width, col);
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapT, texcoords[i][1],
-                                           height, row);
+            col = nearest_texel_location(tObj->WrapS, img, width,
+                                         texcoords[i][0]);
+            row = nearest_texel_location(tObj->WrapT, img, height,
+                                         texcoords[i][1]);
             slice = 0;
             break;
 
          case GL_TEXTURE_1D_ARRAY_EXT:
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoords[i][0],
-                                           width, col);
+            col = nearest_texel_location(tObj->WrapS, img, width,
+                                         texcoords[i][0]);
             row = clamp_rect_coord_nearest(tObj->WrapT, texcoords[i][1], height);
             slice = 0;
+            break;
 
          case GL_TEXTURE_2D_ARRAY_EXT:
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapS, texcoords[i][0],
-                                           width, col);
-            COMPUTE_NEAREST_TEXEL_LOCATION(tObj->WrapT, texcoords[i][1],
-                                           height, row);
+            col = nearest_texel_location(tObj->WrapS, img, width,
+                                         texcoords[i][0]);
+            row = nearest_texel_location(tObj->WrapT, img, height,
+                                         texcoords[i][1]);
             slice = clamp_rect_coord_nearest(tObj->WrapR, texcoords[i][2], depth);
             break;
          }
@@ -3009,39 +2946,40 @@ sample_depth_texture( GLcontext *ctx,
          GLfloat depth00, depth01, depth10, depth11;
          GLint i0, i1, j0, j1;
          GLint slice;
-         GLfloat u, v;
+         GLfloat a, b;
          GLuint useBorderTexel;
 
          switch (tObj->Target) {
          case GL_TEXTURE_RECTANGLE_ARB:
             clamp_rect_coord_linear(tObj->WrapS, texcoords[i][0],
-                                    width, &i0, &i1);
+                                    width, &i0, &i1, &a);
             clamp_rect_coord_linear(tObj->WrapT, texcoords[i][1],
-                                    height, &j0, &j1);
+                                    height, &j0, &j1, &b);
             slice = 0;
             break;
 
          case GL_TEXTURE_1D:
          case GL_TEXTURE_2D:
-            COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoords[i][0],
-                                           u, width, i0, i1);
-            COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapT, texcoords[i][1],
-                                           v, height,j0, j1);
+            linear_texel_locations(tObj->WrapS, img, width,
+                                   texcoords[i][0], &i0, &i1, &a);
+            linear_texel_locations(tObj->WrapT, img, height,
+                                   texcoords[i][1], &j0, &j1, &b);
             slice = 0;
             break;
 
          case GL_TEXTURE_1D_ARRAY_EXT:
-            COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoords[i][0],
-                                           u, width, i0, i1);
+            linear_texel_locations(tObj->WrapS, img, width,
+                                   texcoords[i][0], &i0, &i1, &a);
             j0 = clamp_rect_coord_nearest(tObj->WrapT, texcoords[i][1], height);
             j1 = j0;
             slice = 0;
+            break;
 
          case GL_TEXTURE_2D_ARRAY_EXT:
-            COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapS, texcoords[i][0],
-                                           u, width, i0, i1);
-            COMPUTE_LINEAR_TEXEL_LOCATIONS(tObj->WrapT, texcoords[i][1],
-                                           v, height,j0, j1);
+            linear_texel_locations(tObj->WrapS, img, width,
+                                   texcoords[i][0], &i0, &i1, &a);
+            linear_texel_locations(tObj->WrapT, img, height,
+                                   texcoords[i][1], &j0, &j1, &b);
             slice = clamp_rect_coord_nearest(tObj->WrapR, texcoords[i][2], depth);
             break;
          }
@@ -3105,8 +3043,6 @@ sample_depth_texture( GLcontext *ctx,
 
          if (0) {
             /* compute a single weighted depth sample and do one comparison */
-            const GLfloat a = FRAC(u + 1.0F);
-            const GLfloat b = FRAC(v + 1.0F);
             const GLfloat depthSample
                = lerp_2d(a, b, depth00, depth10, depth01, depth11);
             if ((depthSample <= texcoords[i][compare_coord] && function == GL_LEQUAL) ||
@@ -3177,8 +3113,6 @@ sample_depth_texture( GLcontext *ctx,
             case GL_NONE:
                /* ordinary bilinear filtering */
                {
-                  const GLfloat a = FRAC(u + 1.0F);
-                  const GLfloat b = FRAC(v + 1.0F);
                   const GLfloat depthSample
                      = lerp_2d(a, b, depth00, depth10, depth01, depth11);
                   CLAMPED_FLOAT_TO_CHAN(result, depthSample);
@@ -3249,7 +3183,7 @@ sample_depth_texture2(const GLcontext *ctx,
     * GL_TEXTURE_COMPARE_SGIX == GL_TRUE but the current texture object
     * isn't a depth texture.
     */
-   if (texImage->_BaseFormat != GL_DEPTH_COMPONENT) {
+   if (texImage->TexFormat->BaseFormat != GL_DEPTH_COMPONENT) {
       _mesa_problem(ctx,"GL_TEXTURE_COMPARE_SGIX enabled with non-depth texture");
       return;
    }
@@ -3272,10 +3206,10 @@ sample_depth_texture2(const GLcontext *ctx,
          GLint col, row, ii, jj, imin, imax, jmin, jmax, samples, count;
          GLfloat w;
          GLchan lum;
-         COMPUTE_NEAREST_TEXEL_LOCATION(texObj->WrapS, texcoords[i][0],
-					width, col);
-         COMPUTE_NEAREST_TEXEL_LOCATION(texObj->WrapT, texcoords[i][1],
-					height, row);
+         col = nearest_texel_location(texObj->WrapS, img, width,
+                                      texcoords[i][0]);
+         row = nearest_texel_location(texObj->WrapT, img, height,
+                                      texcoords[i][1]);
 
          imin = col - K;
          imax = col + K;
@@ -3352,7 +3286,7 @@ _swrast_choose_texture_sample_func( GLcontext *ctx,
    }
    else {
       const GLboolean needLambda = (GLboolean) (t->MinFilter != t->MagFilter);
-      const GLenum format = t->Image[0][t->BaseLevel]->_BaseFormat;
+      const GLenum format = t->Image[0][t->BaseLevel]->TexFormat->BaseFormat;
 
       switch (t->Target) {
       case GL_TEXTURE_1D:
diff --git a/src/mesa/swrast/s_texfilter.h b/src/mesa/swrast/s_texfilter.h
index 6267ad17eb8..2e265d685c5 100644
--- a/src/mesa/swrast/s_texfilter.h
+++ b/src/mesa/swrast/s_texfilter.h
@@ -27,7 +27,6 @@
 #define S_TEXFILTER_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_texstore.c b/src/mesa/swrast/s_texstore.c
index f5d081d7a3f..16b00b9fa1c 100644
--- a/src/mesa/swrast/s_texstore.c
+++ b/src/mesa/swrast/s_texstore.c
@@ -40,13 +40,13 @@
 #include "main/imports.h"
 #include "main/colormac.h"
 #include "main/context.h"
-#include "convolve.h"
-#include "image.h"
+#include "main/convolve.h"
+#include "main/image.h"
 #include "main/macros.h"
-#include "mipmap.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texstore.h"
+#include "main/mipmap.h"
+#include "main/texformat.h"
+#include "main/teximage.h"
+#include "main/texstore.h"
 
 #include "s_context.h"
 #include "s_depth.h"
@@ -216,9 +216,9 @@ is_depth_format(GLenum format)
 {
    switch (format) {
       case GL_DEPTH_COMPONENT:
-      case GL_DEPTH_COMPONENT16_SGIX:
-      case GL_DEPTH_COMPONENT24_SGIX:
-      case GL_DEPTH_COMPONENT32_SGIX:
+      case GL_DEPTH_COMPONENT16:
+      case GL_DEPTH_COMPONENT24:
+      case GL_DEPTH_COMPONENT32:
          return GL_TRUE;
       default:
          return GL_FALSE;
@@ -305,7 +305,7 @@ _swrast_copy_teximage1d( GLcontext *ctx, GLenum target, GLint level,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texUnit, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 }
 
@@ -381,7 +381,7 @@ _swrast_copy_teximage2d( GLcontext *ctx, GLenum target, GLint level,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texUnit, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 }
 
@@ -450,7 +450,7 @@ _swrast_copy_texsubimage1d( GLcontext *ctx, GLenum target, GLint level,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texUnit, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 }
 
@@ -526,7 +526,7 @@ _swrast_copy_texsubimage2d( GLcontext *ctx,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texUnit, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 }
 
@@ -599,6 +599,6 @@ _swrast_copy_texsubimage3d( GLcontext *ctx,
 
    /* GL_SGIS_generate_mipmap */
    if (level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      _mesa_generate_mipmap(ctx, target, texUnit, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 }
diff --git a/src/mesa/swrast/s_triangle.c b/src/mesa/swrast/s_triangle.c
index 00cff6635f1..2033ab55291 100644
--- a/src/mesa/swrast/s_triangle.c
+++ b/src/mesa/swrast/s_triangle.c
@@ -34,7 +34,7 @@
 #include "main/colormac.h"
 #include "main/imports.h"
 #include "main/macros.h"
-#include "texformat.h"
+#include "main/texformat.h"
 
 #include "s_aatriangle.h"
 #include "s_context.h"
@@ -58,7 +58,7 @@ _swrast_culltriangle( GLcontext *ctx,
    GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
    GLfloat c = ex*fy-ey*fx;
 
-   if (c * SWRAST_CONTEXT(ctx)->_BackfaceSign > 0)
+   if (c * SWRAST_CONTEXT(ctx)->_BackfaceCullSign > 0)
       return 0;
 
    return 1;
@@ -130,7 +130,7 @@ _swrast_culltriangle( GLcontext *ctx,
 #define T_SCALE theight
 
 #define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];\
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
    struct gl_texture_object *obj = ctx->Texture.Unit[0].Current2D;	\
    const GLint b = obj->BaseLevel;					\
    const GLfloat twidth = (GLfloat) obj->Image[0][b]->Width;		\
@@ -139,8 +139,7 @@ _swrast_culltriangle( GLcontext *ctx,
    const GLchan *texture = (const GLchan *) obj->Image[0][b]->Data;	\
    const GLint smask = obj->Image[0][b]->Width - 1;			\
    const GLint tmask = obj->Image[0][b]->Height - 1;			\
-   if (!texture) {							\
-      /* this shouldn't happen */					\
+   if (!rb || !texture) {						\
       return;								\
    }
 
@@ -182,7 +181,7 @@ _swrast_culltriangle( GLcontext *ctx,
 #define T_SCALE theight
 
 #define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0][0];\
+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
    struct gl_texture_object *obj = ctx->Texture.Unit[0].Current2D;	\
    const GLint b = obj->BaseLevel;					\
    const GLfloat twidth = (GLfloat) obj->Image[0][b]->Width;		\
@@ -191,8 +190,7 @@ _swrast_culltriangle( GLcontext *ctx,
    const GLchan *texture = (const GLchan *) obj->Image[0][b]->Data;	\
    const GLint smask = obj->Image[0][b]->Width - 1;			\
    const GLint tmask = obj->Image[0][b]->Height - 1;			\
-   if (!texture) {							\
-      /* this shouldn't happen */					\
+   if (!rb || !texture) {						\
       return;								\
    }
 
diff --git a/src/mesa/swrast/s_triangle.h b/src/mesa/swrast/s_triangle.h
index c3cadae2d4e..b81932c7304 100644
--- a/src/mesa/swrast/s_triangle.h
+++ b/src/mesa/swrast/s_triangle.h
@@ -28,7 +28,6 @@
 #define S_TRIANGLES_H
 
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/s_tritemp.h b/src/mesa/swrast/s_tritemp.h
index cded4a6c1c4..8e3c5b5eebb 100644
--- a/src/mesa/swrast/s_tritemp.h
+++ b/src/mesa/swrast/s_tritemp.h
@@ -234,18 +234,18 @@ static void NAME(GLcontext *ctx, const SWvertex *v0,
    /* compute area, oneOverArea and perform backface culling */
    {
       const GLfloat area = eMaj.dx * eBot.dy - eBot.dx * eMaj.dy;
-      /* Do backface culling */
-      if (area * bf < 0.0)
-         return;
 
       if (IS_INF_OR_NAN(area) || area == 0.0F)
          return;
 
-      oneOverArea = 1.0F / area;
-   }
+      if (area * bf * swrast->_BackfaceCullSign < 0.0)
+         return;
 
+      oneOverArea = 1.0F / area;
 
-   span.facing = ctx->_Facing; /* for 2-sided stencil test */
+      /* 0 = front, 1 = back */
+      span.facing = oneOverArea * bf > 0.0F;
+   }
 
    /* Edge setup.  For a triangle strip these could be reused... */
    {
diff --git a/src/mesa/swrast/s_zoom.c b/src/mesa/swrast/s_zoom.c
index 48b4d1e2407..a48eae19259 100644
--- a/src/mesa/swrast/s_zoom.c
+++ b/src/mesa/swrast/s_zoom.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -130,8 +130,8 @@ static void
 zoom_span( GLcontext *ctx, GLint imgX, GLint imgY, const SWspan *span,
            const GLvoid *src, GLenum format )
 {
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    SWspan zoomed;
-   SWspanarrays zoomed_arrays;  /* this is big! */
    GLint x0, x1, y0, y1;
    GLint zoomedWidth;
 
@@ -140,6 +140,13 @@ zoom_span( GLcontext *ctx, GLint imgX, GLint imgY, const SWspan *span,
       return;  /* totally clipped */
    }
 
+   if (!swrast->ZoomedArrays) {
+      /* allocate on demand */
+      swrast->ZoomedArrays = (SWspanarrays *) CALLOC(sizeof(SWspanarrays));
+      if (!swrast->ZoomedArrays)
+         return;
+   }
+
    zoomedWidth = x1 - x0;
    ASSERT(zoomedWidth > 0);
    ASSERT(zoomedWidth <= MAX_WIDTH);
@@ -151,14 +158,14 @@ zoom_span( GLcontext *ctx, GLint imgX, GLint imgY, const SWspan *span,
    INIT_SPAN(zoomed, GL_BITMAP);
    zoomed.x = x0;
    zoomed.end = zoomedWidth;
-   zoomed.array = &zoomed_arrays;
-   zoomed_arrays.ChanType = span->array->ChanType;
-   if (zoomed_arrays.ChanType == GL_UNSIGNED_BYTE)
-      zoomed_arrays.rgba = (GLchan (*)[4]) zoomed_arrays.rgba8;
-   else if (zoomed_arrays.ChanType == GL_UNSIGNED_SHORT)
-      zoomed_arrays.rgba = (GLchan (*)[4]) zoomed_arrays.rgba16;
+   zoomed.array = swrast->ZoomedArrays;
+   zoomed.array->ChanType = span->array->ChanType;
+   if (zoomed.array->ChanType == GL_UNSIGNED_BYTE)
+      zoomed.array->rgba = (GLchan (*)[4]) zoomed.array->rgba8;
+   else if (zoomed.array->ChanType == GL_UNSIGNED_SHORT)
+      zoomed.array->rgba = (GLchan (*)[4]) zoomed.array->rgba16;
    else
-      zoomed_arrays.rgba = (GLchan (*)[4]) zoomed_arrays.attribs[FRAG_ATTRIB_COL0];
+      zoomed.array->rgba = (GLchan (*)[4]) zoomed.array->attribs[FRAG_ATTRIB_COL0];
 
    COPY_4V(zoomed.attrStart[FRAG_ATTRIB_WPOS], span->attrStart[FRAG_ATTRIB_WPOS]);
    COPY_4V(zoomed.attrStepX[FRAG_ATTRIB_WPOS], span->attrStepX[FRAG_ATTRIB_WPOS]);
diff --git a/src/mesa/swrast/s_zoom.h b/src/mesa/swrast/s_zoom.h
index 7701515476b..d2815b41a0e 100644
--- a/src/mesa/swrast/s_zoom.h
+++ b/src/mesa/swrast/s_zoom.h
@@ -25,7 +25,6 @@
 #ifndef S_ZOOM_H
 #define S_ZOOM_H
 
-#include "main/mtypes.h"
 #include "swrast.h"
 
 
diff --git a/src/mesa/swrast/sources b/src/mesa/swrast/sources
deleted file mode 100644
index 9ffd4cca722..00000000000
--- a/src/mesa/swrast/sources
+++ /dev/null
@@ -1,65 +0,0 @@
-# List of source files in this directory used for X.org xserver build
-MESA_SWRAST_SOURCES = \
-s_aaline.c \
-s_aatriangle.c \
-s_accum.c \
-s_alpha.c \
-s_arbshader.c \
-s_atifragshader.c \
-s_bitmap.c \
-s_blend.c \
-s_blit.c \
-s_buffers.c \
-s_context.c \
-s_copypix.c \
-s_depth.c \
-s_drawpix.c \
-s_feedback.c \
-s_fog.c \
-s_imaging.c \
-s_lines.c \
-s_logic.c \
-s_masking.c \
-s_nvfragprog.c \
-s_points.c \
-s_readpix.c \
-s_span.c \
-s_stencil.c \
-s_texcombine.c \
-s_texfilter.c \
-s_texstore.c \
-s_triangle.c \
-s_zoom.c
-
-MESA_SWRAST_HEADERS = \
-s_aaline.h \
-s_aalinetemp.h \
-s_aatriangle.h \
-s_aatritemp.h \
-s_accum.h \
-s_alpha.h \
-s_arbshader.h \
-s_atifragshader.h \
-s_blend.h \
-s_context.h \
-s_depth.h \
-s_drawpix.h \
-s_feedback.h \
-s_fog.h \
-s_lines.h \
-s_linetemp.h \
-s_logic.h \
-s_masking.h \
-s_nvfragprog.h \
-s_points.h \
-s_pointtemp.h \
-s_span.h \
-s_spantemp.h \
-s_stencil.h \
-s_texcombine.h \
-s_texfilter.h \
-s_triangle.h \
-s_trispan.h \
-s_tritemp.h \
-s_zoom.h \
-swrast.h
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index 85a27fd55bf..047f7991e64 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -143,6 +143,13 @@ _swrast_Accum(GLcontext *ctx, GLenum op, GLfloat value);
 extern void
 _swrast_ResetLineStipple( GLcontext *ctx );
 
+/**
+ * Indicates front/back facing for subsequent points/lines when drawing
+ * unfilled polygons.  Needed for two-side stencil.
+ */
+extern void
+_swrast_SetFacing(GLcontext *ctx, GLuint facing);
+
 /* These will always render the correct point/line/triangle for the
  * current state.
  *
diff --git a/src/mesa/swrast_setup/descrip.mms b/src/mesa/swrast_setup/descrip.mms
new file mode 100644
index 00000000000..32ffd80732d
--- /dev/null
+++ b/src/mesa/swrast_setup/descrip.mms
@@ -0,0 +1,42 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define tnl [-.tnl]
+	define vbo [-.vbo]
+	define swrast [-.swrast]
+	define array_cache [-.array_cache]
+	define glapi [-.glapi]
+	define main [-.main]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.main],[-.glapi]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = ss_context.c ss_triangle.c
+
+OBJECTS =  ss_context.obj,ss_triangle.obj
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+ss_context.obj : ss_context.c
+ss_triangle.obj : ss_triangle.c
diff --git a/src/mesa/swrast_setup/sources b/src/mesa/swrast_setup/sources
deleted file mode 100644
index dee14b6774e..00000000000
--- a/src/mesa/swrast_setup/sources
+++ /dev/null
@@ -1,10 +0,0 @@
-MESA_SWRAST_SETUP_SOURCES = \
-ss_context.c \
-ss_triangle.c
-
-MESA_SWRAST_SETUP_HEADERS = \
-ss_context.h \
-ss_triangle.h \
-ss_tritmp.h \
-ss_vb.h \
-swrast_setup.h
diff --git a/src/mesa/swrast_setup/ss_context.c b/src/mesa/swrast_setup/ss_context.c
index fd6935e7be8..61172f9979b 100644
--- a/src/mesa/swrast_setup/ss_context.c
+++ b/src/mesa/swrast_setup/ss_context.c
@@ -28,13 +28,14 @@
 #include "main/glheader.h"
 #include "main/imports.h"
 #include "main/colormac.h"
-#include "ss_context.h"
-#include "ss_triangle.h"
-#include "swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 #include "tnl/t_vertex.h"
+#include "swrast_setup.h"
+#include "ss_context.h"
+#include "ss_triangle.h"
+
 
 /* Need to check lighting state and vertex program state to know
  * if two-sided lighting is in effect.
@@ -111,22 +112,25 @@ setup_vertex_format(GLcontext *ctx)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    SScontext *swsetup = SWSETUP_CONTEXT(ctx);
+   GLboolean intColors = !ctx->FragmentProgram._Current
+                      && !ctx->ATIFragmentShader._Enabled
+                      && ctx->RenderMode == GL_RENDER
+                      && CHAN_TYPE == GL_UNSIGNED_BYTE;
 
-   if (!RENDERINPUTS_EQUAL(tnl->render_inputs_bitset,
+   if (intColors != swsetup->intColors ||
+       !RENDERINPUTS_EQUAL(tnl->render_inputs_bitset,
                            swsetup->last_index_bitset)) {
       DECLARE_RENDERINPUTS(index_bitset);
       struct tnl_attr_map map[_TNL_ATTRIB_MAX];
       int i, e = 0;
 
+      swsetup->intColors = intColors;
+
       RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
       EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, attrib[FRAG_ATTRIB_WPOS] );
 
       if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR0 )) {
-         swsetup->intColors = !ctx->FragmentProgram._Current
-                           && !ctx->ATIFragmentShader._Enabled
-                           && ctx->RenderMode == GL_RENDER
-                           && CHAN_TYPE == GL_UNSIGNED_BYTE;
          if (swsetup->intColors)
             EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4CHAN_4F_RGBA, color );
          else
@@ -201,6 +205,9 @@ _swsetup_RenderStart( GLcontext *ctx )
 
    swsetup->NewState = 0;
 
+   /* This will change if drawing unfilled tris */
+   _swrast_SetFacing(ctx, 0);
+
    _swrast_render_start(ctx);
 
    /* Important */
diff --git a/src/mesa/swrast_setup/ss_triangle.c b/src/mesa/swrast_setup/ss_triangle.c
index 9fef7a52eca..54e24c4c44e 100644
--- a/src/mesa/swrast_setup/ss_triangle.c
+++ b/src/mesa/swrast_setup/ss_triangle.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.1
+ * Version:  7.1
  *
- * Copyright (C) 1999-2004  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -68,6 +68,8 @@ static void _swsetup_render_line_tri( GLcontext *ctx,
          return;
    }
 
+   _swrast_SetFacing(ctx, facing);
+
    if (ctx->Light.ShadeModel == GL_FLAT) {
       COPY_CHAN4(c[0], v0->color);
       COPY_CHAN4(c[1], v1->color);
@@ -127,6 +129,8 @@ static void _swsetup_render_point_tri( GLcontext *ctx,
          return;
    }
 
+   _swrast_SetFacing(ctx, facing);
+
    if (ctx->Light.ShadeModel == GL_FLAT) {
       /* save colors/indexes for v0, v1 vertices */
       COPY_CHAN4(c[0], v0->color);
@@ -290,10 +294,8 @@ void _swsetup_choose_trifuncs( GLcontext *ctx )
        ctx->Polygon.OffsetFill)
       ind |= SS_OFFSET_BIT;
 
-   /* Note: gl_FrontFacing lives in fragment input FOGC.Y at this time */
    if ((ctx->Light.Enabled && ctx->Light.Model.TwoSide) ||
-       (ctx->VertexProgram._Enabled && ctx->VertexProgram.TwoSideEnabled) ||
-       (ctx->FragmentProgram._Current && ctx->FragmentProgram._Current->Base.InputsRead & (1 << FRAG_ATTRIB_FOGC)))
+       (ctx->VertexProgram._Current && ctx->VertexProgram.TwoSideEnabled))
       ind |= SS_TWOSIDE_BIT;
 
    /* We piggyback the two-sided stencil front/back determination on the
@@ -311,6 +313,4 @@ void _swsetup_choose_trifuncs( GLcontext *ctx )
    tnl->Driver.Render.Quad = quad_tab[ind];
    tnl->Driver.Render.Line = swsetup_line;
    tnl->Driver.Render.Points = swsetup_points;
-
-   ctx->_Facing = 0;
 }
diff --git a/src/mesa/swrast_setup/ss_triangle.h b/src/mesa/swrast_setup/ss_triangle.h
index 863e744607b..007fa2e9141 100644
--- a/src/mesa/swrast_setup/ss_triangle.h
+++ b/src/mesa/swrast_setup/ss_triangle.h
@@ -29,7 +29,6 @@
 #ifndef SS_TRIANGLE_H
 #define SS_TRIANGLE_H
 
-#include "main/mtypes.h"
 #include "ss_context.h"
 
 
diff --git a/src/mesa/swrast_setup/ss_tritmp.h b/src/mesa/swrast_setup/ss_tritmp.h
index f6b738d60df..97d2f4a16b1 100644
--- a/src/mesa/swrast_setup/ss_tritmp.h
+++ b/src/mesa/swrast_setup/ss_tritmp.h
@@ -49,7 +49,6 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
    v[1] = &verts[e1];
    v[2] = &verts[e2];
 
-
    if (IND & (SS_TWOSIDE_BIT | SS_OFFSET_BIT | SS_UNFILLED_BIT))
    {
       GLfloat ex = v[0]->attrib[FRAG_ATTRIB_WPOS][0] - v[2]->attrib[FRAG_ATTRIB_WPOS][0];
@@ -61,7 +60,6 @@ static void TAG(triangle)(GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
       if (IND & (SS_TWOSIDE_BIT | SS_UNFILLED_BIT))
       {
 	 facing = (cc < 0.0) ^ ctx->Polygon._FrontBit;
-         ctx->_Facing = facing;
 
 	 if (IND & SS_UNFILLED_BIT)
 	    mode = facing ? ctx->Polygon.BackMode : ctx->Polygon.FrontMode;
@@ -238,14 +236,16 @@ static void TAG(quadfunc)( GLcontext *ctx, GLuint v0,
 {
    if (IND & SS_UNFILLED_BIT) {
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-      GLubyte ef1 = VB->EdgeFlag[v1];
-      GLubyte ef3 = VB->EdgeFlag[v3];
-      VB->EdgeFlag[v1] = 0;
-      TAG(triangle)( ctx, v0, v1, v3 );
-      VB->EdgeFlag[v1] = ef1;
-      VB->EdgeFlag[v3] = 0;
-      TAG(triangle)( ctx, v1, v2, v3 );
-      VB->EdgeFlag[v3] = ef3;
+      if (VB->EdgeFlag) { /* XXX this test shouldn't be needed (bug 12614) */
+         GLubyte ef1 = VB->EdgeFlag[v1];
+         GLubyte ef3 = VB->EdgeFlag[v3];
+         VB->EdgeFlag[v1] = 0;
+         TAG(triangle)( ctx, v0, v1, v3 );
+         VB->EdgeFlag[v1] = ef1;
+         VB->EdgeFlag[v3] = 0;
+         TAG(triangle)( ctx, v1, v2, v3 );
+         VB->EdgeFlag[v3] = ef3;
+      }
    } else {
       TAG(triangle)( ctx, v0, v1, v3 );
       TAG(triangle)( ctx, v1, v2, v3 );
diff --git a/src/mesa/tnl/descrip.mms b/src/mesa/tnl/descrip.mms
new file mode 100644
index 00000000000..25dd1aecb13
--- /dev/null
+++ b/src/mesa/tnl/descrip.mms
@@ -0,0 +1,68 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 39 September 2008
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define vbo [-.vbo]
+	define shader [-.shader]
+	define swrast [-.swrast]
+	define array_cache [-.array_cache]
+	define main [-.main]
+	define glapi [-.glapi]
+	define tnl [-.tnl]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.main],[-.glapi],[-.shader],[-.shader.slang]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES = t_context.c t_draw.c \
+	t_pipeline.c t_vb_fog.c \
+	t_vb_light.c t_vb_normals.c t_vb_points.c t_vb_program.c \
+	t_vb_render.c t_vb_texgen.c t_vb_texmat.c t_vb_vertex.c \
+	t_vertex.c t_rasterpos.c\
+	t_vertex_generic.c t_vp_build.c
+
+OBJECTS = t_context.obj,t_draw.obj,\
+	t_pipeline.obj,t_vb_fog.obj,t_vb_light.obj,t_vb_normals.obj,\
+	t_vb_points.obj,t_vb_program.obj,t_vb_render.obj,t_vb_texgen.obj,\
+	t_vb_texmat.obj,t_vb_vertex.obj,t_rasterpos.obj,\
+	t_vertex.obj,t_vertex_generic.obj,\
+	t_vp_build.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+t_context.obj : t_context.c
+t_draw.obj : t_draw.c
+t_pipeline.obj : t_pipeline.c
+t_vb_fog.obj : t_vb_fog.c
+t_vb_light.obj : t_vb_light.c
+t_vb_normals.obj : t_vb_normals.c
+t_vb_points.obj : t_vb_points.c
+t_vb_program.obj : t_vb_program.c
+t_vb_render.obj : t_vb_render.c
+t_vb_texgen.obj : t_vb_texgen.c
+t_vb_texmat.obj : t_vb_texmat.c
+t_vb_vertex.obj : t_vb_vertex.c
+t_vertex.obj : t_vertex.c
+t_vertex_generic.obj : t_vertex_generic.c
+t_vp_build.obj : t_vp_build.c
+t_rasterpos.obj : t_rasterpos.c
diff --git a/src/mesa/tnl/sources b/src/mesa/tnl/sources
deleted file mode 100644
index a0888be11d6..00000000000
--- a/src/mesa/tnl/sources
+++ /dev/null
@@ -1,34 +0,0 @@
-# List of source files in this directory used for X.org xserver build
-MESA_TNL_SOURCES = \
-t_context.c \
-t_pipeline.c \
-t_vb_arbprogram.c \
-t_vb_arbprogram_sse.c \
-t_vb_arbshader.c \
-t_vb_cull.c \
-t_vb_fog.c \
-t_vb_light.c \
-t_vb_normals.c \
-t_vb_points.c \
-t_vb_program.c \
-t_vb_render.c \
-t_vb_texgen.c \
-t_vb_texmat.c \
-t_vb_vertex.c \
-t_vertex.c \
-t_vertex_generic.c \
-t_vertex_sse.c \
-t_vp_build.c 
-
-MESA_TNL_HEADERS = \
-t_array_api.h \
-t_array_import.h \
-t_context.h \
-t_pipeline.h \
-t_vb_arbprogram.h \
-t_vb_cliptmp.h \
-t_vb_lighttmp.h \
-t_vb_rendertmp.h \
-t_vertex.h \
-t_vp_build.h \
-tnl.h
diff --git a/src/mesa/tnl/t_context.c b/src/mesa/tnl/t_context.c
index 0ace5c2d6f1..ce37dc04282 100644
--- a/src/mesa/tnl/t_context.c
+++ b/src/mesa/tnl/t_context.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.2
+ * Version:  7.2
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -104,10 +104,10 @@ _tnl_InvalidateState( GLcontext *ctx, GLuint new_state )
    const struct gl_vertex_program *vp = ctx->VertexProgram._Current;
    const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
 
-   if (new_state & (_NEW_HINT)) {
+   if (new_state & (_NEW_HINT | _NEW_PROGRAM)) {
       ASSERT(tnl->AllowVertexFog || tnl->AllowPixelFog);
-      tnl->_DoVertexFog = (tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
-         || !tnl->AllowPixelFog;
+      tnl->_DoVertexFog = ((tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
+         || !tnl->AllowPixelFog) && !fp;
    }
 
    tnl->pipeline.new_state |= new_state;
@@ -136,11 +136,19 @@ _tnl_InvalidateState( GLcontext *ctx, GLuint new_state )
       RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_COLOR_INDEX );
    }
 
-   if (ctx->Fog.Enabled ||
-       ((ctx->FragmentProgram._Active || ctx->FragmentProgram._Current) &&
-        (ctx->FragmentProgram._Current->FogOption != GL_NONE ||
-         (ctx->FragmentProgram._Current->Base.InputsRead & FRAG_BIT_FOGC))))
+   if (ctx->Fog.Enabled) {
+      /* fixed-function fog */
       RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_FOG );
+   }
+   else if (ctx->FragmentProgram._Active || ctx->FragmentProgram._Current) {
+      struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
+      if (fp) {
+         if (fp->FogOption != GL_NONE || (fp->Base.InputsRead & FRAG_BIT_FOGC)) {
+            /* fragment program needs fog coord */
+            RENDERINPUTS_SET( tnl->render_inputs_bitset, _TNL_ATTRIB_FOG );
+         }
+      }
+   }
 
    if (ctx->Polygon.FrontMode != GL_FILL || 
        ctx->Polygon.BackMode != GL_FILL)
@@ -201,8 +209,8 @@ _tnl_allow_vertex_fog( GLcontext *ctx, GLboolean value )
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    tnl->AllowVertexFog = value;
-   tnl->_DoVertexFog = (tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
-      || !tnl->AllowPixelFog;
+   tnl->_DoVertexFog = ((tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
+      || !tnl->AllowPixelFog) && !ctx->FragmentProgram._Current;
 
 }
 
@@ -211,7 +219,7 @@ _tnl_allow_pixel_fog( GLcontext *ctx, GLboolean value )
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    tnl->AllowPixelFog = value;
-   tnl->_DoVertexFog = (tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
-      || !tnl->AllowPixelFog;
+   tnl->_DoVertexFog = ((tnl->AllowVertexFog && (ctx->Hint.Fog != GL_NICEST))
+      || !tnl->AllowPixelFog) && !ctx->FragmentProgram._Current;
 }
 
diff --git a/src/mesa/tnl/t_draw.c b/src/mesa/tnl/t_draw.c
index a1a46f6b827..fd647c1f4ab 100644
--- a/src/mesa/tnl/t_draw.c
+++ b/src/mesa/tnl/t_draw.c
@@ -1,9 +1,8 @@
-
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -29,7 +28,7 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/imports.h"
-#include "state.h"
+#include "main/state.h"
 #include "main/mtypes.h"
 #include "main/macros.h"
 #include "main/enums.h"
@@ -368,7 +367,7 @@ void _tnl_draw_prims( GLcontext *ctx,
 			_tnl_draw_prims );
       return;
    }
-   else if (max_index >= max) {
+   else if (max_index > max) {
       /* The software TNL pipeline has a fixed amount of storage for
        * vertices and it is necessary to split incoming drawing commands
        * if they exceed that limit.
diff --git a/src/mesa/tnl/t_pipeline.c b/src/mesa/tnl/t_pipeline.c
index 2a0ed8852a2..357ef1e24b5 100644
--- a/src/mesa/tnl/t_pipeline.c
+++ b/src/mesa/tnl/t_pipeline.c
@@ -199,11 +199,11 @@ const struct tnl_pipeline_stage *_tnl_default_pipeline[] = {
    &_tnl_vertex_transform_stage,
    &_tnl_normal_transform_stage,
    &_tnl_lighting_stage,
-   &_tnl_fog_coordinate_stage,
    &_tnl_texgen_stage,
    &_tnl_texture_transform_stage,
    &_tnl_point_attenuation_stage,
    &_tnl_vertex_program_stage, 
+   &_tnl_fog_coordinate_stage,
    &_tnl_render_stage,
    NULL 
 };
diff --git a/src/mesa/tnl/t_vb_fog.c b/src/mesa/tnl/t_vb_fog.c
index 00c0979f3f6..f3a7bd49f40 100644
--- a/src/mesa/tnl/t_vb_fog.c
+++ b/src/mesa/tnl/t_vb_fog.c
@@ -41,7 +41,6 @@
 
 struct fog_stage_data {
    GLvector4f fogcoord;		/* has actual storage allocated */
-   GLvector4f input;		/* points into VB->EyePtr Z values */
 };
 
 #define FOG_STAGE_DATA(stage) ((struct fog_stage_data *)stage->privatePtr)
@@ -91,7 +90,8 @@ init_static_data( void )
  * evaluating the GL_LINEAR, GL_EXP or GL_EXP2 fog function.
  * Fog coordinates are distances from the eye (typically between the
  * near and far clip plane distances).
- * Note the fog (eye Z) coords may be negative so we use ABS(z) below.
+ * Note that fogcoords may be negative, if eye z is source absolute
+ * value must be taken earlier.
  * Fog blend factors are in the range [0,1].
  */
 static void
@@ -148,11 +148,11 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
    struct fog_stage_data *store = FOG_STAGE_DATA(stage);
    GLvector4f *input;
 
-   if (!ctx->Fog.Enabled || ctx->VertexProgram._Current)
-      return GL_TRUE;
 
+   if (!ctx->Fog.Enabled)
+      return GL_TRUE;
 
-   if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT) {
+   if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT && !ctx->VertexProgram._Current) {
       GLuint i;
       GLfloat *coord;
       /* Fog is computed from vertex or fragment Z values */
@@ -169,13 +169,10 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
 	  */
 	 input = &store->fogcoord;
 
-         /* NOTE: negate plane here so we get positive fog coords! */
-	 /* NOTE2: this doesn't always work (tests/fog - all frag depth fog
-	    coords will be negative). */
-	 plane[0] = -m[2];
-	 plane[1] = -m[6];
-	 plane[2] = -m[10];
-	 plane[3] = -m[14];
+	 plane[0] = m[2];
+	 plane[1] = m[6];
+	 plane[2] = m[10];
+	 plane[3] = m[14];
 	 /* Full eye coords weren't required, just calculate the
 	  * eye Z values.
 	  */
@@ -189,12 +186,12 @@ run_fog_stage(GLcontext *ctx, struct tnl_pipeline_stage *stage)
 	    NOTE should avoid going through array twice */
 	 coord = input->start;
 	 for (i = 0; i < input->count; i++) {
-	    input->data[i][0] = FABSF(*coord);
+	    *coord = FABSF(*coord);
 	    STRIDE_F(coord, input->stride);
 	 }
       }
       else {
-         /* fog coordinates = eye Z coordinates (use ABS later) */
+         /* fog coordinates = eye Z coordinates - need to copy for ABS */
 	 input = &store->fogcoord;
 
 	 if (VB->EyePtr->size < 2)
@@ -249,7 +246,6 @@ alloc_fog_data(GLcontext *ctx, struct tnl_pipeline_stage *stage)
       return GL_FALSE;
 
    _mesa_vector4f_alloc( &store->fogcoord, 0, tnl->vb.Size, 32 );
-   _mesa_vector4f_init( &store->input, 0, NULL );
 
    if (!inited)
       init_static_data();
diff --git a/src/mesa/tnl/t_vb_light.c b/src/mesa/tnl/t_vb_light.c
index ebd3412d20f..f47f99397c9 100644
--- a/src/mesa/tnl/t_vb_light.c
+++ b/src/mesa/tnl/t_vb_light.c
@@ -26,10 +26,10 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "light.h"
+#include "main/light.h"
 #include "main/macros.h"
 #include "main/imports.h"
-#include "simple_list.h"
+#include "main/simple_list.h"
 #include "main/mtypes.h"
 
 #include "math/m_translate.h"
diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c
index c778f5d2488..f99401ca6d8 100644
--- a/src/mesa/tnl/t_vb_program.c
+++ b/src/mesa/tnl/t_vb_program.c
@@ -1,6 +1,6 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.1
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -41,9 +41,129 @@
 #include "swrast/s_context.h"
 #include "swrast/s_texfilter.h"
 
-#include "tnl.h"
-#include "t_context.h"
-#include "t_pipeline.h"
+#include "tnl/tnl.h"
+#include "tnl/t_context.h"
+#include "tnl/t_pipeline.h"
+
+
+
+/*!
+ * Private storage for the vertex program pipeline stage.
+ */
+struct vp_stage_data {
+   /** The results of running the vertex program go into these arrays. */
+   GLvector4f results[VERT_RESULT_MAX];
+
+   GLvector4f ndcCoords;              /**< normalized device coords */
+   GLubyte *clipmask;                 /**< clip flags */
+   GLubyte ormask, andmask;           /**< for clipping */
+};
+
+
+#define VP_STAGE_DATA(stage) ((struct vp_stage_data *)(stage->privatePtr))
+
+
+static void
+userclip( GLcontext *ctx,
+          GLvector4f *clip,
+          GLubyte *clipmask,
+          GLubyte *clipormask,
+          GLubyte *clipandmask )
+{
+   GLuint p;
+
+   for (p = 0; p < ctx->Const.MaxClipPlanes; p++) {
+      if (ctx->Transform.ClipPlanesEnabled & (1 << p)) {
+	 GLuint nr, i;
+	 const GLfloat a = ctx->Transform._ClipUserPlane[p][0];
+	 const GLfloat b = ctx->Transform._ClipUserPlane[p][1];
+	 const GLfloat c = ctx->Transform._ClipUserPlane[p][2];
+	 const GLfloat d = ctx->Transform._ClipUserPlane[p][3];
+         GLfloat *coord = (GLfloat *)clip->data;
+         GLuint stride = clip->stride;
+         GLuint count = clip->count;
+
+	 for (nr = 0, i = 0 ; i < count ; i++) {
+	    GLfloat dp = (coord[0] * a + 
+			  coord[1] * b +
+			  coord[2] * c +
+			  coord[3] * d);
+
+	    if (dp < 0) {
+	       nr++;
+	       clipmask[i] |= CLIP_USER_BIT;
+	    }
+
+	    STRIDE_F(coord, stride);
+	 }
+
+	 if (nr > 0) {
+	    *clipormask |= CLIP_USER_BIT;
+	    if (nr == count) {
+	       *clipandmask |= CLIP_USER_BIT;
+	       return;
+	    }
+	 }
+      }
+   }
+}
+
+
+static GLboolean
+do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store)
+{
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   struct vertex_buffer *VB = &tnl->vb;
+   /* Cliptest and perspective divide.  Clip functions must clear
+    * the clipmask.
+    */
+   store->ormask = 0;
+   store->andmask = CLIP_FRUSTUM_BITS;
+
+   if (tnl->NeedNdcCoords) {
+      VB->NdcPtr =
+         _mesa_clip_tab[VB->ClipPtr->size]( VB->ClipPtr,
+                                            &store->ndcCoords,
+                                            store->clipmask,
+                                            &store->ormask,
+                                            &store->andmask );
+   }
+   else {
+      VB->NdcPtr = NULL;
+      _mesa_clip_np_tab[VB->ClipPtr->size]( VB->ClipPtr,
+                                            NULL,
+                                            store->clipmask,
+                                            &store->ormask,
+                                            &store->andmask );
+   }
+
+   if (store->andmask) {
+      /* All vertices are outside the frustum */
+      return GL_FALSE;
+   }
+
+   /* Test userclip planes.  This contributes to VB->ClipMask.
+    */
+   /** XXX NEW_SLANG _Enabled ??? */
+   if (ctx->Transform.ClipPlanesEnabled && (!ctx->VertexProgram._Enabled ||
+      ctx->VertexProgram.Current->IsPositionInvariant)) {
+      userclip( ctx,
+		VB->ClipPtr,
+		store->clipmask,
+		&store->ormask,
+		&store->andmask );
+
+      if (store->andmask) {
+	 return GL_FALSE;
+      }
+   }
+
+   VB->ClipAndMask = store->andmask;
+   VB->ClipOrMask = store->ormask;
+   VB->ClipMask = store->clipmask;
+
+   return GL_TRUE;
+}
 
 
 /**
@@ -52,7 +172,6 @@
  * real dependencies on the rest of swrast.  It should probably be
  * moved into main/ someday.
  */
-
 static void
 vp_fetch_texel(GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
                GLuint unit, GLfloat color[4])
@@ -85,22 +204,6 @@ _tnl_program_string(GLcontext *ctx, GLenum target, struct gl_program *program)
 }
 
 
-/*!
- * Private storage for the vertex program pipeline stage.
- */
-struct vp_stage_data {
-   /** The results of running the vertex program go into these arrays. */
-   GLvector4f results[VERT_RESULT_MAX];
-
-   GLvector4f ndcCoords;              /**< normalized device coords */
-   GLubyte *clipmask;                 /**< clip flags */
-   GLubyte ormask, andmask;           /**< for clipping */
-};
-
-
-#define VP_STAGE_DATA(stage) ((struct vp_stage_data *)(stage->privatePtr))
-
-
 /**
  * Initialize virtual machine state prior to executing vertex program.
  */
@@ -139,96 +242,50 @@ init_machine(GLcontext *ctx, struct gl_program_machine *machine)
 
    machine->FetchTexelLod = vp_fetch_texel;
    machine->FetchTexelDeriv = NULL; /* not used by vertex programs */
+
+   machine->Samplers = ctx->VertexProgram._Current->Base.SamplerUnits;
 }
 
 
 /**
- * Copy the 16 elements of a matrix into four consecutive program
- * registers starting at 'pos'.
+ * Map the texture images which the vertex program will access (if any).
  */
 static void
-load_matrix(GLfloat registers[][4], GLuint pos, const GLfloat mat[16])
+map_textures(GLcontext *ctx, const struct gl_vertex_program *vp)
 {
-   GLuint i;
-   for (i = 0; i < 4; i++) {
-      registers[pos + i][0] = mat[0 + i];
-      registers[pos + i][1] = mat[4 + i];
-      registers[pos + i][2] = mat[8 + i];
-      registers[pos + i][3] = mat[12 + i];
-   }
-}
+   GLuint u;
 
+   if (!ctx->Driver.MapTexture)
+      return;
 
-/**
- * As above, but transpose the matrix.
- */
-static void
-load_transpose_matrix(GLfloat registers[][4], GLuint pos,
-                      const GLfloat mat[16])
-{
-   MEMCPY(registers[pos], mat, 16 * sizeof(GLfloat));
+   for (u = 0; u < ctx->Const.MaxVertexTextureImageUnits; u++) {
+      if (vp->Base.TexturesUsed[u]) {
+         /* Note: _Current *should* correspond to the target indicated
+          * in TexturesUsed[u].
+          */
+         ctx->Driver.MapTexture(ctx, ctx->Texture.Unit[u]._Current);
+      }
+   }
 }
 
 
 /**
- * Load current vertex program's parameter registers with tracked
- * matrices (if NV program).  This only needs to be done per
- * glBegin/glEnd, not per-vertex.
+ * Unmap the texture images which were used by the vertex program (if any).
  */
-void
-_mesa_load_tracked_matrices(GLcontext *ctx)
+static void
+unmap_textures(GLcontext *ctx, const struct gl_vertex_program *vp)
 {
-   GLuint i;
+   GLuint u;
 
-   for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS / 4; i++) {
-      /* point 'mat' at source matrix */
-      GLmatrix *mat;
-      if (ctx->VertexProgram.TrackMatrix[i] == GL_MODELVIEW) {
-         mat = ctx->ModelviewMatrixStack.Top;
-      }
-      else if (ctx->VertexProgram.TrackMatrix[i] == GL_PROJECTION) {
-         mat = ctx->ProjectionMatrixStack.Top;
-      }
-      else if (ctx->VertexProgram.TrackMatrix[i] == GL_TEXTURE) {
-         mat = ctx->TextureMatrixStack[ctx->Texture.CurrentUnit].Top;
-      }
-      else if (ctx->VertexProgram.TrackMatrix[i] == GL_COLOR) {
-         mat = ctx->ColorMatrixStack.Top;
-      }
-      else if (ctx->VertexProgram.TrackMatrix[i]==GL_MODELVIEW_PROJECTION_NV) {
-         /* XXX verify the combined matrix is up to date */
-         mat = &ctx->_ModelProjectMatrix;
-      }
-      else if (ctx->VertexProgram.TrackMatrix[i] >= GL_MATRIX0_NV &&
-               ctx->VertexProgram.TrackMatrix[i] <= GL_MATRIX7_NV) {
-         GLuint n = ctx->VertexProgram.TrackMatrix[i] - GL_MATRIX0_NV;
-         ASSERT(n < MAX_PROGRAM_MATRICES);
-         mat = ctx->ProgramMatrixStack[n].Top;
-      }
-      else {
-         /* no matrix is tracked, but we leave the register values as-is */
-         assert(ctx->VertexProgram.TrackMatrix[i] == GL_NONE);
-         continue;
-      }
+   if (!ctx->Driver.MapTexture)
+      return;
 
-      /* load the matrix values into sequential registers */
-      if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_IDENTITY_NV) {
-         load_matrix(ctx->VertexProgram.Parameters, i*4, mat->m);
-      }
-      else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_INVERSE_NV) {
-         _math_matrix_analyse(mat); /* update the inverse */
-         ASSERT(!_math_matrix_is_dirty(mat));
-         load_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv);
-      }
-      else if (ctx->VertexProgram.TrackMatrixTransform[i] == GL_TRANSPOSE_NV) {
-         load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->m);
-      }
-      else {
-         assert(ctx->VertexProgram.TrackMatrixTransform[i]
-                == GL_INVERSE_TRANSPOSE_NV);
-         _math_matrix_analyse(mat); /* update the inverse */
-         ASSERT(!_math_matrix_is_dirty(mat));
-         load_transpose_matrix(ctx->VertexProgram.Parameters, i*4, mat->inv);
+   for (u = 0; u < ctx->Const.MaxVertexTextureImageUnits; u++) {
+      if (vp->Base.TexturesUsed[u]) {
+         /* Note: _Current *should* correspond to the target indicated
+          * in TexturesUsed[u].
+          */
+         ctx->Driver.UnmapTexture(ctx, ctx->Texture.Unit[u]._Current);
       }
    }
 }
@@ -259,6 +316,7 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
       _mesa_load_state_parameters(ctx, program->Base.Parameters);
    }
 
+   /* make list of outputs to save some time below */
    numOutputs = 0;
    for (i = 0; i < VERT_RESULT_MAX; i++) {
       if (program->Base.OutputsWritten & (1 << i)) {
@@ -266,6 +324,8 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
       }
    }
 
+   map_textures(ctx, program);
+
    for (i = 0; i < VB->Count; i++) {
       GLuint attr;
 
@@ -317,6 +377,8 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
 #endif
    }
 
+   unmap_textures(ctx, program);
+
    /* Fixup fog and point size results if needed */
    if (program->IsNVProgram) {
       if (ctx->Fog.Enabled &&
@@ -334,12 +396,39 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
       }
    }
 
-   /* Setup the VB pointers so that the next pipeline stages get
-    * their data from the right place (the program output arrays).
-    */
-   VB->ClipPtr = &store->results[VERT_RESULT_HPOS];
-   VB->ClipPtr->size = 4;
-   VB->ClipPtr->count = VB->Count;
+   if (program->IsPositionInvariant) {
+      /* We need the exact same transform as in the fixed function path here
+       * to guarantee invariance, depending on compiler optimization flags
+       * results could be different otherwise.
+       */
+      VB->ClipPtr = TransformRaw( &store->results[0],
+				  &ctx->_ModelProjectMatrix,
+				  VB->AttribPtr[0] );
+
+      /* Drivers expect this to be clean to element 4...
+       */
+      switch (VB->ClipPtr->size) {
+      case 1:
+	 /* impossible */
+      case 2:
+	 _mesa_vector4f_clean_elem( VB->ClipPtr, VB->Count, 2 );
+	 /* fall-through */
+      case 3:
+	 _mesa_vector4f_clean_elem( VB->ClipPtr, VB->Count, 3 );
+	 /* fall-through */
+      case 4:
+	 break;
+      }
+   }
+   else {
+      /* Setup the VB pointers so that the next pipeline stages get
+       * their data from the right place (the program output arrays).
+       */
+      VB->ClipPtr = &store->results[VERT_RESULT_HPOS];
+      VB->ClipPtr->size = 4;
+      VB->ClipPtr->count = VB->Count;
+   }
+
    VB->ColorPtr[0] = &store->results[VERT_RESULT_COL0];
    VB->ColorPtr[1] = &store->results[VERT_RESULT_BFC0];
    VB->SecondaryColorPtr[0] = &store->results[VERT_RESULT_COL1];
@@ -365,41 +454,10 @@ run_vp( GLcontext *ctx, struct tnl_pipeline_stage *stage )
       }
    }
 
-   /* Cliptest and perspective divide.  Clip functions must clear
-    * the clipmask.
-    */
-   store->ormask = 0;
-   store->andmask = CLIP_FRUSTUM_BITS;
-
-   if (tnl->NeedNdcCoords) {
-      VB->NdcPtr =
-         _mesa_clip_tab[VB->ClipPtr->size]( VB->ClipPtr,
-                                            &store->ndcCoords,
-                                            store->clipmask,
-                                            &store->ormask,
-                                            &store->andmask );
-   }
-   else {
-      VB->NdcPtr = NULL;
-      _mesa_clip_np_tab[VB->ClipPtr->size]( VB->ClipPtr,
-                                            NULL,
-                                            store->clipmask,
-                                            &store->ormask,
-                                            &store->andmask );
-   }
-
-   if (store->andmask)  /* All vertices are outside the frustum */
-      return GL_FALSE;
 
-
-   /* This is where we'd do clip testing against the user-defined
-    * clipping planes, but they're not supported by vertex programs.
+   /* Perform NDC and cliptest operations:
     */
-
-   VB->ClipOrMask = store->ormask;
-   VB->ClipMask = store->clipmask;
-
-   return GL_TRUE;
+   return do_ndc_cliptest(ctx, store);
 }
 
 
diff --git a/src/mesa/tnl/t_vertex_generic.c b/src/mesa/tnl/t_vertex_generic.c
index c399423b9ec..9812f8c8081 100644
--- a/src/mesa/tnl/t_vertex_generic.c
+++ b/src/mesa/tnl/t_vertex_generic.c
@@ -29,11 +29,17 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/colormac.h"
+#include "main/simple_list.h"
 #include "t_context.h"
 #include "t_vertex.h"
-#include "simple_list.h"
 
 
+#if 0
+#define DEBUG_INSERT printf("%s\n", __FUNCTION__)
+#else
+#define DEBUG_INSERT
+#endif
+
 
 /*
  * These functions take the NDC coordinates pointed to by 'in', apply the
@@ -45,7 +51,7 @@ static INLINE void insert_4f_viewport_4( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
    out[2] = vp[10] * in[2] + vp[14];
@@ -57,7 +63,7 @@ static INLINE void insert_4f_viewport_3( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
    out[2] = vp[10] * in[2] + vp[14];
@@ -69,7 +75,7 @@ static INLINE void insert_4f_viewport_2( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
    out[2] = vp[14];
@@ -81,7 +87,7 @@ static INLINE void insert_4f_viewport_1( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[13];
    out[2] = vp[14];
@@ -93,7 +99,7 @@ static INLINE void insert_3f_viewport_3( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
    out[2] = vp[10] * in[2] + vp[14];
@@ -104,10 +110,10 @@ static INLINE void insert_3f_viewport_2( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
-   out[2] = vp[10] * in[2] + vp[14];
+   out[2] = vp[14];
 }
 
 static INLINE void insert_3f_viewport_1( const struct tnl_clipspace_attr *a, GLubyte *v,
@@ -115,7 +121,7 @@ static INLINE void insert_3f_viewport_1( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[13];
    out[2] = vp[14];
@@ -126,7 +132,7 @@ static INLINE void insert_2f_viewport_2( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[5] * in[1] + vp[13];
 }
@@ -136,7 +142,7 @@ static INLINE void insert_2f_viewport_1( const struct tnl_clipspace_attr *a, GLu
 {
    GLfloat *out = (GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = vp[0] * in[0] + vp[12];
    out[1] = vp[13];
 }
@@ -150,7 +156,7 @@ static INLINE void insert_4f_4( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = in[2];
@@ -161,7 +167,7 @@ static INLINE void insert_4f_3( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = in[2];
@@ -172,7 +178,7 @@ static INLINE void insert_4f_2( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = 0;
@@ -183,7 +189,7 @@ static INLINE void insert_4f_1( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = 0;
    out[2] = 0;
@@ -194,7 +200,7 @@ static INLINE void insert_3f_xyw_4( const struct tnl_clipspace_attr *a, GLubyte
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = in[3];
@@ -203,6 +209,7 @@ static INLINE void insert_3f_xyw_4( const struct tnl_clipspace_attr *a, GLubyte
 static INLINE void insert_3f_xyw_err( const struct tnl_clipspace_attr *a, GLubyte *v, const GLfloat *in )
 {
    (void) a; (void) v; (void) in;
+   DEBUG_INSERT;
    _mesa_exit(1);
 }
 
@@ -210,7 +217,7 @@ static INLINE void insert_3f_3( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = in[2];
@@ -220,7 +227,7 @@ static INLINE void insert_3f_2( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
    out[2] = 0;
@@ -230,7 +237,7 @@ static INLINE void insert_3f_1( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = 0;
    out[2] = 0;
@@ -241,7 +248,7 @@ static INLINE void insert_2f_2( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = in[1];
 }
@@ -250,7 +257,7 @@ static INLINE void insert_2f_1( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-   
+   DEBUG_INSERT;
    out[0] = in[0];
    out[1] = 0;
 }
@@ -259,12 +266,13 @@ static INLINE void insert_1f_1( const struct tnl_clipspace_attr *a, GLubyte *v,
 {
    GLfloat *out = (GLfloat *)(v);
    (void) a;
-
+   DEBUG_INSERT;
    out[0] = in[0];
 }
 
 static INLINE void insert_null( const struct tnl_clipspace_attr *a, GLubyte *v, const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a; (void) v; (void) in;
 }
 
@@ -272,6 +280,7 @@ static INLINE void insert_4chan_4f_rgba_4( const struct tnl_clipspace_attr *a, G
 				  const GLfloat *in )
 {
    GLchan *c = (GLchan *)v;
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); 
    UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); 
@@ -283,6 +292,7 @@ static INLINE void insert_4chan_4f_rgba_3( const struct tnl_clipspace_attr *a, G
 				  const GLfloat *in )
 {
    GLchan *c = (GLchan *)v;
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); 
    UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); 
@@ -294,6 +304,7 @@ static INLINE void insert_4chan_4f_rgba_2( const struct tnl_clipspace_attr *a, G
 				  const GLfloat *in )
 {
    GLchan *c = (GLchan *)v;
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); 
    UNCLAMPED_FLOAT_TO_CHAN(c[1], in[1]); 
@@ -305,6 +316,7 @@ static INLINE void insert_4chan_4f_rgba_1( const struct tnl_clipspace_attr *a, G
 				  const GLfloat *in )
 {
    GLchan *c = (GLchan *)v;
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_CHAN(c[0], in[0]); 
    c[1] = 0;
@@ -315,6 +327,7 @@ static INLINE void insert_4chan_4f_rgba_1( const struct tnl_clipspace_attr *a, G
 static INLINE void insert_4ub_4f_rgba_4( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -325,6 +338,7 @@ static INLINE void insert_4ub_4f_rgba_4( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_rgba_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -335,6 +349,7 @@ static INLINE void insert_4ub_4f_rgba_3( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_rgba_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -345,6 +360,7 @@ static INLINE void insert_4ub_4f_rgba_2( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_rgba_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    v[1] = 0;
@@ -355,6 +371,7 @@ static INLINE void insert_4ub_4f_rgba_1( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_bgra_4( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -365,6 +382,7 @@ static INLINE void insert_4ub_4f_bgra_4( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_bgra_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -375,6 +393,7 @@ static INLINE void insert_4ub_4f_bgra_3( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_bgra_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -385,6 +404,7 @@ static INLINE void insert_4ub_4f_bgra_2( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_bgra_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    v[1] = 0;
@@ -395,6 +415,7 @@ static INLINE void insert_4ub_4f_bgra_1( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_argb_4( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -405,6 +426,7 @@ static INLINE void insert_4ub_4f_argb_4( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_argb_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -415,6 +437,7 @@ static INLINE void insert_4ub_4f_argb_3( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_argb_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -425,6 +448,7 @@ static INLINE void insert_4ub_4f_argb_2( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_argb_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[0]);
    v[2] = 0x00;
@@ -435,6 +459,7 @@ static INLINE void insert_4ub_4f_argb_1( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_abgr_4( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -445,6 +470,7 @@ static INLINE void insert_4ub_4f_abgr_4( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_abgr_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -455,6 +481,7 @@ static INLINE void insert_4ub_4f_abgr_3( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_abgr_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[1]);
@@ -465,6 +492,7 @@ static INLINE void insert_4ub_4f_abgr_2( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_4ub_4f_abgr_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[3], in[0]);
    v[2] = 0x00;
@@ -475,6 +503,7 @@ static INLINE void insert_4ub_4f_abgr_1( const struct tnl_clipspace_attr *a, GLu
 static INLINE void insert_3ub_3f_rgb_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 			       const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -484,6 +513,7 @@ static INLINE void insert_3ub_3f_rgb_3( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_3ub_3f_rgb_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 			       const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -493,6 +523,7 @@ static INLINE void insert_3ub_3f_rgb_2( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_3ub_3f_rgb_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 			       const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
    v[1] = 0;
@@ -502,6 +533,7 @@ static INLINE void insert_3ub_3f_rgb_1( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_3ub_3f_bgr_3( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				 const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -511,6 +543,7 @@ static INLINE void insert_3ub_3f_bgr_3( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_3ub_3f_bgr_2( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				 const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(v[1], in[1]);
@@ -520,6 +553,7 @@ static INLINE void insert_3ub_3f_bgr_2( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_3ub_3f_bgr_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 				 const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[2], in[0]);
    v[1] = 0;
@@ -530,6 +564,7 @@ static INLINE void insert_3ub_3f_bgr_1( const struct tnl_clipspace_attr *a, GLub
 static INLINE void insert_1ub_1f_1( const struct tnl_clipspace_attr *a, GLubyte *v, 
 			   const GLfloat *in )
 {
+   DEBUG_INSERT;
    (void) a;
    UNCLAMPED_FLOAT_TO_UBYTE(v[0], in[0]);
 }
@@ -551,6 +586,7 @@ static void extract_4f_viewport( const struct tnl_clipspace_attr *a, GLfloat *ou
    /* Although included for completeness, the position coordinate is
     * usually handled differently during clipping.
     */
+   DEBUG_INSERT;
    out[0] = (in[0] - vp[12]) / vp[0];
    out[1] = (in[1] - vp[13]) / vp[5];
    out[2] = (in[2] - vp[14]) / vp[10];
@@ -562,7 +598,7 @@ static void extract_3f_viewport( const struct tnl_clipspace_attr *a, GLfloat *ou
 {
    const GLfloat *in = (const GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = (in[0] - vp[12]) / vp[0];
    out[1] = (in[1] - vp[13]) / vp[5];
    out[2] = (in[2] - vp[14]) / vp[10];
@@ -575,7 +611,7 @@ static void extract_2f_viewport( const struct tnl_clipspace_attr *a, GLfloat *ou
 {
    const GLfloat *in = (const GLfloat *)v;
    const GLfloat * const vp = a->vp;
-   
+   DEBUG_INSERT;
    out[0] = (in[0] - vp[12]) / vp[0];
    out[1] = (in[1] - vp[13]) / vp[5];
    out[2] = 0;
diff --git a/src/mesa/tnl/t_vertex_sse.c b/src/mesa/tnl/t_vertex_sse.c
index 96def25206b..7a255d680a1 100644
--- a/src/mesa/tnl/t_vertex_sse.c
+++ b/src/mesa/tnl/t_vertex_sse.c
@@ -28,10 +28,10 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/colormac.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 #include "t_context.h"
 #include "t_vertex.h"
-#include "simple_list.h"
-#include "main/enums.h"
 
 #if defined(USE_SSE_ASM)
 
@@ -39,6 +39,12 @@
 #include "x86/common_x86_asm.h"
 
 
+/**
+ * Number of bytes to allocate for generated SSE functions
+ */
+#define MAX_SSE_CODE_SIZE 1024
+
+
 #define X    0
 #define Y    1
 #define Z    2
@@ -140,7 +146,8 @@ static void emit_load3f_1( struct x86_program *p,
 			   struct x86_reg dest,
 			   struct x86_reg arg0 )
 {
-   emit_load4f_1(p, dest, arg0);
+   /* Loading from memory erases the upper bits. */
+   sse_movss(&p->func, dest, arg0);
 }
 
 static void emit_load2f_2( struct x86_program *p, 
@@ -154,7 +161,8 @@ static void emit_load2f_1( struct x86_program *p,
 			   struct x86_reg dest,
 			   struct x86_reg arg0 )
 {
-   emit_load4f_1(p, dest, arg0);
+   /* Loading from memory erases the upper bits. */
+   sse_movss(&p->func, dest, arg0);
 }
 
 static void emit_load1f_1( struct x86_program *p, 
@@ -346,6 +354,7 @@ static GLboolean build_vertex_emit( struct x86_program *p )
    struct x86_reg temp = x86_make_reg(file_XMM, 0);
    struct x86_reg vp0 = x86_make_reg(file_XMM, 1);
    struct x86_reg vp1 = x86_make_reg(file_XMM, 2);
+   struct x86_reg temp2 = x86_make_reg(file_XMM, 3);
    GLubyte *fixup, *label;
 
    /* Push a few regs?
@@ -518,7 +527,8 @@ static GLboolean build_vertex_emit( struct x86_program *p )
 	    sse_shufps(&p->func, temp, temp, SHUF(W,X,Y,Z));
 
 	    get_src_ptr(p, srcECX, vtxESI, &a[1]);
-	    emit_load(p, temp, 1, x86_deref(srcECX), a[1].inputsize);
+	    emit_load(p, temp2, 1, x86_deref(srcECX), a[1].inputsize);
+	    sse_movss(&p->func, temp, temp2);
 	    update_src_ptr(p, srcECX, vtxESI, &a[1]);
 
 	    /* Rearrange and possibly do BGR conversion:
@@ -533,8 +543,8 @@ static GLboolean build_vertex_emit( struct x86_program *p )
 	 }
 	 else {
 	    _mesa_printf("Can't emit 3ub\n");
+	    return GL_FALSE;	/* add this later */
 	 }
-	 return GL_FALSE;	/* add this later */
 	 break;
 
       case EMIT_4UB_4F_RGBA:
@@ -619,7 +629,10 @@ static GLboolean build_vertex_emit( struct x86_program *p )
    x86_pop(&p->func, countEBP);
    x86_ret(&p->func);
 
+   assert(!vtx->emit);
    vtx->emit = (tnl_emit_func)x86_get_func(&p->func);
+
+   assert( (char *) p->func.csr - (char *) p->func.store <= MAX_SSE_CODE_SIZE );
    return GL_TRUE;
 }
 
@@ -644,7 +657,10 @@ void _tnl_generate_sse_emit( GLcontext *ctx )
    p.identity = x86_make_reg(file_XMM, 6);
    p.chan0 = x86_make_reg(file_XMM, 7);
 
-   x86_init_func(&p.func);
+   if (!x86_init_func_size(&p.func, MAX_SSE_CODE_SIZE)) {
+      vtx->emit = NULL;
+      return;
+   }
 
    if (build_vertex_emit(&p)) {
       _tnl_register_fastpath( vtx, GL_TRUE );
diff --git a/src/mesa/tnl/t_vp_build.c b/src/mesa/tnl/t_vp_build.c
index b3b63cd3e4a..7be4d95af6b 100644
--- a/src/mesa/tnl/t_vp_build.c
+++ b/src/mesa/tnl/t_vp_build.c
@@ -1,8 +1,8 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.1
  *
- * Copyright (C) 2006  Tungsten Graphics   All Rights Reserved.
+ * Copyright (C) 2007  Tungsten Graphics   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/src/mesa/tnl/tnl.h b/src/mesa/tnl/tnl.h
index 4bdbed92df8..4d628aa9a60 100644
--- a/src/mesa/tnl/tnl.h
+++ b/src/mesa/tnl/tnl.h
@@ -1,9 +1,8 @@
-
 /*
  * Mesa 3-D graphics library
- * Version:  3.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
diff --git a/src/mesa/tnl_dd/t_dd_tritmp.h b/src/mesa/tnl_dd/t_dd_tritmp.h
index 0bf32ff3443..1ae70f4059f 100644
--- a/src/mesa/tnl_dd/t_dd_tritmp.h
+++ b/src/mesa/tnl_dd/t_dd_tritmp.h
@@ -155,10 +155,6 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
       {
 	 facing = AREA_IS_CCW( cc ) ^ ctx->Polygon._FrontBit;
 
-         if (DO_TWOSTENCIL && ctx->Stencil.TestTwoSide) {
-            ctx->_Facing = facing; /* mixed mode rendering: for 2-sided stencil test */
-         }
-
 	 if (DO_UNFILLED) {
 	    if (facing) {
 	       mode = ctx->Polygon.BackMode;
@@ -271,7 +267,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 	    GLfloat bc	= b * ic;
 	    if ( ac < 0.0f ) ac = -ac;
 	    if ( bc < 0.0f ) bc = -bc;
-	    offset += MAX2( ac, bc ) * ctx->Polygon.OffsetFactor;
+	    offset += MAX2( ac, bc ) * ctx->Polygon.OffsetFactor / ctx->DrawBuffer->_MRD;
 	 }
 	 offset *= ctx->DrawBuffer->_MRD * (REVERSE_DEPTH ? -1.0 : 1.0);
       }
@@ -393,7 +389,7 @@ static void TAG(triangle)( GLcontext *ctx, GLuint e0, GLuint e1, GLuint e2 )
 
 #if DO_QUAD
 #if DO_FULL_QUAD
-static void TAG(quad)( GLcontext *ctx,
+static void TAG(quadr)( GLcontext *ctx,
 		       GLuint e0, GLuint e1, GLuint e2, GLuint e3 )
 {
    struct vertex_buffer *VB = &TNL_CONTEXT( ctx )->vb;
@@ -421,10 +417,6 @@ static void TAG(quad)( GLcontext *ctx,
       {
 	 facing = AREA_IS_CCW( cc ) ^ ctx->Polygon._FrontBit;
 
-         if (DO_TWOSTENCIL && ctx->Stencil.TestTwoSide) {
-            ctx->_Facing = facing; /* mixed mode rendering: for 2-sided stencil test */
-         }
-
 	 if (DO_UNFILLED) {
 	    if (facing) {
 	       mode = ctx->Polygon.BackMode;
@@ -547,7 +539,7 @@ static void TAG(quad)( GLcontext *ctx,
 	    GLfloat bc	= b * ic;
 	    if ( ac < 0.0f ) ac = -ac;
 	    if ( bc < 0.0f ) bc = -bc;
-	    offset += MAX2( ac, bc ) * ctx->Polygon.OffsetFactor;
+	    offset += MAX2( ac, bc ) * ctx->Polygon.OffsetFactor / ctx->DrawBuffer->_MRD;
 	 }
 	 offset *= ctx->DrawBuffer->_MRD * (REVERSE_DEPTH ? -1.0 : 1.0);
       }
@@ -681,7 +673,7 @@ static void TAG(quad)( GLcontext *ctx,
    }
 }
 #else
-static void TAG(quad)( GLcontext *ctx, GLuint e0,
+static void TAG(quadr)( GLcontext *ctx, GLuint e0,
 		       GLuint e1, GLuint e2, GLuint e3 )
 {
    if (DO_UNFILLED) {
@@ -773,7 +765,7 @@ static void TAG(points)( GLcontext *ctx, GLuint first, GLuint last )
 static void TAG(init)( void )
 {
 #if DO_QUAD
-   TAB[IND].quad = TAG(quad);
+   TAB[IND].quad = TAG(quadr);
 #endif
 #if DO_TRI
    TAB[IND].triangle = TAG(triangle);
diff --git a/src/mesa/vbo/descrip.mms b/src/mesa/vbo/descrip.mms
new file mode 100644
index 00000000000..c2a16a04f50
--- /dev/null
+++ b/src/mesa/vbo/descrip.mms
@@ -0,0 +1,62 @@
+# Makefile for core library for VMS
+# contributed by Jouk Jansen  joukj@hrem.nano.tudelft.nl
+# Last revision : 3 October 2007
+
+.first
+	define gl [---.include.gl]
+	define math [-.math]
+	define vbo [-.vbo]
+	define tnl [-.tnl]
+	define shader [-.shader]
+	define swrast [-.swrast]
+	define swrast_setup [-.swrast_setup]
+	define main [-.main]
+	define glapi [-.glapi]
+
+.include [---]mms-config.
+
+##### MACROS #####
+
+VPATH = RCS
+
+INCDIR = [---.include],[-.main],[-.glapi],[-.shader],[-.shader.slang]
+LIBDIR = [---.lib]
+CFLAGS = /include=($(INCDIR),[])/define=(PTHREADS=1)/name=(as_is,short)/float=ieee/ieee=denorm
+
+SOURCES =vbo_context.c,vbo_exec.c,vbo_exec_api.c,vbo_exec_array.c,\
+	vbo_exec_draw.c,vbo_exec_eval.c,vbo_rebase.c,vbo_save.c,\
+	vbo_save_api.c,vbo_save_draw.c,vbo_save_loopback.c,\
+	vbo_split.c,vbo_split_copy.c,vbo_split_inplace.c
+
+OBJECTS =vbo_context.obj,vbo_exec.obj,vbo_exec_api.obj,vbo_exec_array.obj,\
+	vbo_exec_draw.obj,vbo_exec_eval.obj,vbo_rebase.obj,vbo_save.obj,\
+	vbo_save_api.obj,vbo_save_draw.obj,vbo_save_loopback.obj,\
+	vbo_split.obj,vbo_split_copy.obj,vbo_split_inplace.obj
+
+##### RULES #####
+
+VERSION=Mesa V3.4
+
+##### TARGETS #####
+# Make the library
+$(LIBDIR)$(GL_LIB) : $(OBJECTS)
+  @ library $(LIBDIR)$(GL_LIB) $(OBJECTS)
+
+clean :
+	purge
+	delete *.obj;*
+
+vbo_context.obj : vbo_context.c
+vbo_exec.obj : vbo_exec.c
+vbo_exec_api.obj : vbo_exec_api.c
+vbo_exec_array.obj : vbo_exec_array.c
+vbo_exec_draw.obj : vbo_exec_draw.c
+vbo_exec_eval.obj : vbo_exec_eval.c
+vbo_rebase.obj : vbo_rebase.c
+vbo_save.obj : vbo_save.c
+vbo_save_api.obj : vbo_save_api.c
+vbo_save_draw.obj : vbo_save_draw.c
+vbo_save_loopback.obj : vbo_save_loopback.c
+vbo_split.obj : vbo_split.c
+vbo_split_copy.obj : vbo_split_copy.c
+vbo_split_inplace.obj : vbo_split_inplace.c
diff --git a/src/mesa/vbo/vbo_context.c b/src/mesa/vbo/vbo_context.c
index b452ac8a38e..bbf745b0c62 100644
--- a/src/mesa/vbo/vbo_context.c
+++ b/src/mesa/vbo/vbo_context.c
@@ -114,6 +114,8 @@ static void init_mat_currval(GLcontext *ctx)
    struct gl_client_array *arrays = vbo->mat_currval;
    GLuint i;
 
+   ASSERT(NR_MAT_ATTRIBS == MAT_ATTRIB_MAX);
+
    memset(arrays, 0, sizeof(*arrays) * NR_MAT_ATTRIBS);
 
    /* Set up a constant (StrideB == 0) array for each current
@@ -139,11 +141,7 @@ static void init_mat_currval(GLcontext *ctx)
 	 break;
       }
 
-      if (i < MAT_ATTRIB_MAX)
-	 cl->Ptr = (const void *)ctx->Light.Material.Attrib[i];
-      else 
-	 cl->Ptr = (const void *)ctx->Current.Attrib[VERT_ATTRIB_GENERIC0 + i];
-
+      cl->Ptr = (const void *)ctx->Light.Material.Attrib[i];
       cl->Type = GL_FLOAT;
       cl->Stride = 0;
       cl->StrideB = 0;
diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c
index 23f4f8331e3..a6ce26ffed8 100644
--- a/src/mesa/vbo/vbo_exec_api.c
+++ b/src/mesa/vbo/vbo_exec_api.c
@@ -708,8 +708,8 @@ void vbo_exec_vtx_init( struct vbo_exec_context *exec )
                                  &exec->vtx.bufferobj,
                                  ctx->Array.NullBufferObj);
 
+   ASSERT(!exec->vtx.buffer_map);
    exec->vtx.buffer_map = ALIGN_MALLOC(VBO_VERT_BUFFER_SIZE * sizeof(GLfloat), 64);
-
    vbo_exec_vtxfmt_init( exec );
 
    /* Hook our functions into the dispatch table.
@@ -734,13 +734,13 @@ void vbo_exec_vtx_init( struct vbo_exec_context *exec )
 
 void vbo_exec_vtx_destroy( struct vbo_exec_context *exec )
 {
-   GLcontext *ctx = exec->ctx;
    if (exec->vtx.bufferobj->Name) {
-      ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER_ARB, exec->vtx.bufferobj);
-      ctx->Driver.DeleteBuffer(ctx, exec->vtx.bufferobj);
-      exec->vtx.bufferobj = NULL;
+      /* using a real VBO for vertex data */
+      GLcontext *ctx = exec->ctx;
+      _mesa_reference_buffer_object(ctx, &exec->vtx.bufferobj, NULL);
    }
    else {
+      /* just using malloc'd space for vertex data */
       if (exec->vtx.buffer_map) {
          ALIGN_FREE(exec->vtx.buffer_map);
          exec->vtx.buffer_map = NULL;
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index ae43857c8ae..5bf3d836db5 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -249,9 +249,11 @@ void vbo_exec_vtx_flush( struct vbo_exec_context *exec )
          if (ctx->NewState)
             _mesa_update_state( ctx );
 
-
-	 ctx->Driver.UnmapBuffer(ctx, target, exec->vtx.bufferobj);
-	 exec->vtx.buffer_map = NULL;
+         /* if using a real VBO, unmap it before drawing */
+         if (exec->vtx.bufferobj->Name) {
+            ctx->Driver.UnmapBuffer(ctx, target, exec->vtx.bufferobj);
+            exec->vtx.buffer_map = NULL;
+         }
 
 	 vbo_context(ctx)->draw_prims( ctx, 
 				       exec->vtx.inputs, 
@@ -261,11 +263,12 @@ void vbo_exec_vtx_flush( struct vbo_exec_context *exec )
 				       0,
 				       exec->vtx.vert_count - 1);
 
-	 /* Get new data:
-	  */
-	 ctx->Driver.BufferData(ctx, target, size, NULL, usage, exec->vtx.bufferobj);
-	 exec->vtx.buffer_map
-	    = ctx->Driver.MapBuffer(ctx, target, access, exec->vtx.bufferobj);
+	 /* If using a real VBO, get new storage */
+         if (exec->vtx.bufferobj->Name) {
+            ctx->Driver.BufferData(ctx, target, size, NULL, usage, exec->vtx.bufferobj);
+            exec->vtx.buffer_map = 
+               ctx->Driver.MapBuffer(ctx, target, access, exec->vtx.bufferobj);
+         }
       }
    }
 
diff --git a/src/mesa/vbo/vbo_split_copy.c b/src/mesa/vbo/vbo_split_copy.c
index e5c4429350e..685cc0fdf6e 100644
--- a/src/mesa/vbo/vbo_split_copy.c
+++ b/src/mesa/vbo/vbo_split_copy.c
@@ -129,6 +129,13 @@ static GLuint attr_size( const struct gl_client_array *array )
  */
 static GLboolean check_flush( struct copy_context *copy )
 {
+   GLenum mode = copy->dstprim[copy->dstprim_nr].mode; 
+
+   if (GL_TRIANGLE_STRIP == mode &&
+       copy->dstelt_nr & 1) { /* see bug9962 */
+       return GL_FALSE;
+   }
+
    if (copy->dstbuf_nr + 4 > copy->dstbuf_size)
       return GL_TRUE;
 
@@ -458,7 +465,7 @@ static void replay_init( struct copy_context *copy )
       dst->StrideB = copy->vertex_size;
       dst->Ptr = copy->dstbuf + offset;
       dst->Enabled = GL_TRUE;
-      dst->Normalized = GL_TRUE;
+      dst->Normalized = src->Normalized; 
       dst->BufferObj = ctx->Array.NullBufferObj;
       dst->_MaxElement = copy->dstbuf_size; /* may be less! */
 
diff --git a/src/mesa/x86-64/Makefile b/src/mesa/x86-64/Makefile
index ab58aa6c23d..9c3e9d2adf8 100644
--- a/src/mesa/x86-64/Makefile
+++ b/src/mesa/x86-64/Makefile
@@ -19,7 +19,7 @@ INCLUDE_DIRS = \
 default: matypes.h
 
 clean:
-	rm -f matypes.h
+	-rm -f matypes.h
 
 
 # need some special rules here, unfortunately
diff --git a/src/mesa/x86-64/glapi_x86-64.S b/src/mesa/x86-64/glapi_x86-64.S
index f8337ff93ee..cb34061b36c 100644
--- a/src/mesa/x86-64/glapi_x86-64.S
+++ b/src/mesa/x86-64/glapi_x86-64.S
@@ -73,7 +73,7 @@ _x86_64_get_dispatch:
 
 	.p2align	4,,15
 _x86_64_get_dispatch:
-	movq	_gl_DispatchTSD@GOTPCREL(%rip), %rdi
+	movq	_gl_DispatchTSD(%rip), %rdi
 	jmp	pthread_getspecific@PLT
 
 #elif defined(THREADS)
diff --git a/src/mesa/x86-64/x86-64.c b/src/mesa/x86-64/x86-64.c
index 9ec43c841d3..96f8da87f05 100644
--- a/src/mesa/x86-64/x86-64.c
+++ b/src/mesa/x86-64/x86-64.c
@@ -41,7 +41,10 @@
 #include "math/m_debug.h"
 #endif
 
+extern void _mesa_x86_64_cpuid(unsigned int *regs);
+
 DECLARE_XFORM_GROUP( x86_64, 4 )
+DECLARE_XFORM_GROUP( 3dnow, 4 )
 
 #else
 /* just to silence warning below */
@@ -81,6 +84,7 @@ static void message( const char *msg )
 void _mesa_init_all_x86_64_transform_asm(void)
 {
 #ifdef USE_X86_64_ASM
+   unsigned int regs[4];
 
    if ( _mesa_getenv( "MESA_NO_ASM" ) ) {
      return;
@@ -88,24 +92,32 @@ void _mesa_init_all_x86_64_transform_asm(void)
 
    message("Initializing x86-64 optimizations\n");
 
-   ASSIGN_XFORM_GROUP( x86_64, 4 );
 
-   /*
    _mesa_transform_tab[4][MATRIX_GENERAL] =
       _mesa_x86_64_transform_points4_general;
    _mesa_transform_tab[4][MATRIX_IDENTITY] =
       _mesa_x86_64_transform_points4_identity;
    _mesa_transform_tab[4][MATRIX_3D] =
       _mesa_x86_64_transform_points4_3d;
-   _mesa_transform_tab[4][MATRIX_3D_NO_ROT] =
-      _mesa_x86_64_transform_points4_3d_no_rot;
-   _mesa_transform_tab[4][MATRIX_PERSPECTIVE] =
-      _mesa_x86_64_transform_points4_perspective;
-   _mesa_transform_tab[4][MATRIX_2D_NO_ROT] =
-      _mesa_x86_64_transform_points4_2d_no_rot;
-   _mesa_transform_tab[4][MATRIX_2D] =
-      _mesa_x86_64_transform_points4_2d;
-   */
+
+   regs[0] = 0x80000001;
+   regs[1] = 0x00000000;
+   regs[2] = 0x00000000;
+   regs[3] = 0x00000000;
+   _mesa_x86_64_cpuid(regs);
+   if (regs[3] & (1U << 31)) {
+      message("3Dnow! detected\n");
+      _mesa_transform_tab[4][MATRIX_3D_NO_ROT] =
+	  _mesa_3dnow_transform_points4_3d_no_rot;
+      _mesa_transform_tab[4][MATRIX_PERSPECTIVE] =
+	  _mesa_3dnow_transform_points4_perspective;
+      _mesa_transform_tab[4][MATRIX_2D_NO_ROT] =
+	  _mesa_3dnow_transform_points4_2d_no_rot;
+      _mesa_transform_tab[4][MATRIX_2D] =
+	  _mesa_3dnow_transform_points4_2d;
+
+   }
+
    
 #ifdef DEBUG_MATH
    _math_test_all_transform_functions("x86_64");
diff --git a/src/mesa/x86-64/xform4.S b/src/mesa/x86-64/xform4.S
index 667ecf6e589..805969127db 100644
--- a/src/mesa/x86-64/xform4.S
+++ b/src/mesa/x86-64/xform4.S
@@ -1,9 +1,8 @@
-
 /*
  * Mesa 3-D graphics library
- * Version:  3.5
+ * Version:  7.1
  *
- * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -30,7 +29,22 @@
 .text
 
 .align 16
+.globl _mesa_x86_64_cpuid
+_mesa_x86_64_cpuid:
+	pushq	%rbx
+	movl	(%rdi), %eax
+	movl	8(%rdi), %ecx
+
+	cpuid
+
+	movl	%ebx, 4(%rdi)
+	movl	%eax, (%rdi)
+	movl	%ecx, 8(%rdi)
+	movl	%edx, 12(%rdi)
+	popq	%rbx
+	ret
 
+.align 16
 .globl _mesa_x86_64_transform_points4_general
 _mesa_x86_64_transform_points4_general:
 /*
@@ -63,7 +77,7 @@ _mesa_x86_64_transform_points4_general:
 
 p4_general_loop:
 
-	movaps (%rdx), %xmm8		/* ox | oy | oz | ow */
+	movups (%rdx), %xmm8		/* ox | oy | oz | ow */
 	prefetchw 16(%rdi)
 
 	pshufd $0x00, %xmm8, %xmm0	/* ox | ox | ox | ox */
@@ -148,7 +162,7 @@ _mesa_x86_64_transform_points4_3d:
 
 p4_3d_loop:
 
-	movaps (%rdx), %xmm8		/* ox | oy | oz | ow */
+	movups (%rdx), %xmm8		/* ox | oy | oz | ow */
 	prefetchw 16(%rdi)
 
 	pshufd $0x00, %xmm8, %xmm0	/* ox | ox | ox | ox */
@@ -205,8 +219,8 @@ p4_identity_done:
 
 	
 .align 16
-.globl _mesa_x86_64_transform_points4_3d_no_rot
-_mesa_x86_64_transform_points4_3d_no_rot:
+.globl _mesa_3dnow_transform_points4_3d_no_rot
+_mesa_3dnow_transform_points4_3d_no_rot:
 
 	movl V4F_COUNT(%rdx), %ecx	/* count */
 	movzx V4F_STRIDE(%rdx), %eax	/* stride */
@@ -269,8 +283,8 @@ p4_3d_no_rot_done:
 
 	
 .align 16
-.globl _mesa_x86_64_transform_points4_perspective
-_mesa_x86_64_transform_points4_perspective:
+.globl _mesa_3dnow_transform_points4_perspective
+_mesa_3dnow_transform_points4_perspective:
 
 	movl V4F_COUNT(%rdx), %ecx	/* count */
 	movzx V4F_STRIDE(%rdx), %eax	/* stride */
@@ -335,8 +349,8 @@ p4_perspective_done:
 	ret
 
 .align 16
-.globl _mesa_x86_64_transform_points4_2d_no_rot
-_mesa_x86_64_transform_points4_2d_no_rot:
+.globl _mesa_3dnow_transform_points4_2d_no_rot
+_mesa_3dnow_transform_points4_2d_no_rot:
 
 	movl V4F_COUNT(%rdx), %ecx	/* count */
 	movzx V4F_STRIDE(%rdx), %eax	/* stride */
@@ -390,8 +404,8 @@ p4_2d_no_rot_done:
 
 	
 .align 16
-.globl _mesa_x86_64_transform_points4_2d
-_mesa_x86_64_transform_points4_2d:
+.globl _mesa_3dnow_transform_points4_2d
+_mesa_3dnow_transform_points4_2d:
 
 	movl V4F_COUNT(%rdx), %ecx	/* count */
 	movzx V4F_STRIDE(%rdx), %eax	/* stride */
diff --git a/src/mesa/x86/Makefile b/src/mesa/x86/Makefile
index 09481dc5094..aa49a9134aa 100644
--- a/src/mesa/x86/Makefile
+++ b/src/mesa/x86/Makefile
@@ -17,7 +17,7 @@ INCLUDE_DIRS = \
 default: gen_matypes matypes.h
 
 clean:
-	rm -f matypes.h gen_matypes
+	-rm -f matypes.h gen_matypes
 
 
 gen_matypes: gen_matypes.c
diff --git a/src/mesa/x86/common_x86.c b/src/mesa/x86/common_x86.c
index dc80d26fa9b..53215479358 100644
--- a/src/mesa/x86/common_x86.c
+++ b/src/mesa/x86/common_x86.c
@@ -48,8 +48,8 @@
 #include <machine/cpu.h>
 #endif
 
-#include "common_x86_asm.h"
 #include "main/imports.h"
+#include "common_x86_asm.h"
 
 
 int _mesa_x86_cpu_features = 0;
diff --git a/src/mesa/x86/common_x86_asm.S b/src/mesa/x86/common_x86_asm.S
index 09c86b05ba8..ea4047a0e14 100644
--- a/src/mesa/x86/common_x86_asm.S
+++ b/src/mesa/x86/common_x86_asm.S
@@ -39,6 +39,7 @@
  * in there will break the build on some platforms.
  */
 
+#include "matypes.h"
 #include "assyntax.h"
 #include "common_x86_features.h"
 
diff --git a/src/mesa/x86/read_rgba_span_x86.S b/src/mesa/x86/read_rgba_span_x86.S
index 3cbcd719960..80144b889c7 100644
--- a/src/mesa/x86/read_rgba_span_x86.S
+++ b/src/mesa/x86/read_rgba_span_x86.S
@@ -33,20 +33,7 @@
 	.file	"read_rgba_span_x86.S"
 #if !defined(__DJGPP__) && !defined(__MINGW32__) /* this one cries for assyntax.h */
 /* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	.section	.rodata
-	.align 16
-	.type	mask, @object
-	.size	mask, 32
-mask:
-	.long	0xff00ff00
-	.long	0xff00ff00
-	.long	0xff00ff00
-	.long	0xff00ff00
-	.long	0x00ff0000
-	.long	0x00ff0000
-	.long	0x00ff0000
-	.long	0x00ff0000
+ * Replaced data segment constants with text-segment instructions.
  */
 #define	LOAD_MASK(mvins,m1,m2) \
    	pushl	$0xff00ff00 ;\
@@ -61,8 +48,7 @@ mask:
 	mvins	(%esp), m2	;\
 	addl	$32, %esp
 
-
-/* I implemented these as macros because the appear in quite a few places,
+/* I implemented these as macros because they appear in several places,
  * and I've tweaked them a number of times.  I got tired of changing every
  * place they appear. :)
  */
@@ -99,11 +85,6 @@ _generic_read_RGBA_span_BGRA8888_REV_MMX:
 #ifdef USE_INNER_EMMS
 	emms
 #endif
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	movq	mask, %mm1
-	movq	mask+16, %mm2
- */
 	LOAD_MASK(movq,%mm1,%mm2)
 
 	movl	8(%esp), %ebx	/* source pointer */
@@ -201,11 +182,7 @@ _generic_read_RGBA_span_BGRA8888_REV_SSE:
 #ifdef USE_INNER_EMMS
 	emms
 #endif
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	movq	mask, %mm1
-	movq	mask+16, %mm2
- */
+
 	LOAD_MASK(movq,%mm1,%mm2)
 
 	movl	16(%esp), %ebx	/* source pointer */
@@ -364,11 +341,6 @@ _generic_read_RGBA_span_BGRA8888_REV_SSE2:
 	pushl	%esi
 	pushl	%ebx
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	movdqa	mask, %xmm1
-	movdqa	mask+16, %xmm2
- */
 	LOAD_MASK(movdqu,%xmm1,%xmm2)
 
 	movl	12(%esp), %ebx	/* source pointer */
@@ -462,7 +434,8 @@ _generic_read_RGBA_span_BGRA8888_REV_SSE2:
 	je	.L47
 
 	movq	(%ebx), %xmm0
-
+	addl	$8, %ebx
+        
 	movdqa	%xmm0, %xmm3
 	movdqa	%xmm0, %xmm4
 	andps	%xmm1, %xmm0
@@ -476,6 +449,7 @@ _generic_read_RGBA_span_BGRA8888_REV_SSE2:
 	orps	%xmm3, %xmm0
 
 	movq	%xmm0, (%ecx)
+	addl	$8, %ecx        
 .L47:
 
 	testl	$1, %edx
@@ -491,60 +465,12 @@ _generic_read_RGBA_span_BGRA8888_REV_SSE2:
 
 
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
- */
-#if 0
-	.section	.rodata
-
-	.align	16
-mask_565:
-	.word	0xf800
-	.word	0x07e0
-	.word	0x001f
-	.word	0x0000
-
-/* Setting SCALE_ADJUST to 5 gives a perfect match with the classic C
- * implementation in Mesa.  Setting SCALE_ADJUST to 0 is slightly faster but
- * at a small cost to accuracy.
- */
-
-#define SCALE_ADJUST	5
-#if SCALE_ADJUST == 5
-prescale:
-	.word	0x0001
-	.word	0x0010
-	.word	0x0200
-	.word	0x0000
-
-scale:
-	.word	0x20e8		/* (0x00ff0000 / 0x000007c0) + 1 */
-	.word	0x40c5		/* (0x00ff0000 / 0x000003f0) + 1 */
-	.word	0x839d		/* (0x00ff0000 / 0x000001f0) + 1 */
-	.word	0x0000
-#elif SCALE_ADJUST == 0
-prescale:
-	.word	0x0001
-	.word	0x0020
-	.word	0x0800
-	.word	0x0000
-
-scale:
-	.word	0x0108		/* (0x00ff0000 / 0x0000f800) + 1 */
-	.word	0x0104		/* (0x00ff0000 / 0x0000fc00) + 1 */
-	.word	0x0108		/* (0x00ff0000 / 0x0000f800) + 1 */
-	.word	0x0000
-#else
-#error SCALE_ADJUST must either be 5 or 0.
-#endif
-
-
-alpha:	.long	0x00000000
-	.long	0x00ff0000
-#endif
-
 #define MASK_565_L	0x07e0f800
 #define MASK_565_H	0x0000001f
+/* Setting SCALE_ADJUST to 5 gives a perfect match with the
+ * classic C implementation in Mesa.  Setting SCALE_ADJUST
+ * to 0 is slightly faster but at a small cost to accuracy.
+ */
 #define SCALE_ADJUST	5
 #if SCALE_ADJUST == 5
 #define PRESCALE_L 0x00100001
@@ -581,23 +507,17 @@ _generic_read_RGBA_span_RGB565_MMX:
 	movl	8(%esp), %edx	/* destination pointer */
 	movl	12(%esp), %ecx	/* number of pixels to copy */
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	movq	mask_565, %mm5
-	movq	prescale, %mm6
-	movq	scale, %mm7
- */
- 	pushl	MASK_565_H
- 	pushl	MASK_565_L
+	pushl	$MASK_565_H
+	pushl	$MASK_565_L
 	movq	(%esp), %mm5
- 	pushl	PRESCALE_H
- 	pushl	PRESCALE_L
+	pushl	$PRESCALE_H
+	pushl	$PRESCALE_L
 	movq	(%esp), %mm6
- 	pushl	SCALE_H
- 	pushl	SCALE_L
+	pushl	$SCALE_H
+	pushl	$SCALE_L
 	movq	(%esp), %mm7
- 	pushl	ALPHA_H
- 	pushl	ALPHA_L
+	pushl	$ALPHA_H
+	pushl	$ALPHA_L
 	movq	(%esp), %mm3
 	addl	$32,%esp
 
@@ -648,11 +568,6 @@ _generic_read_RGBA_span_RGB565_MMX:
 	/* Always set the alpha value to 0xff.
 	 */
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	por	alpha, %mm0
-	por	alpha, %mm2
- */
  	por %mm3, %mm0
  	por %mm3, %mm2
 
@@ -665,8 +580,6 @@ _generic_read_RGBA_span_RGB565_MMX:
 	movq	%mm0, (%edx)
 	addl	$8, %edx
 
-
-
 	pshufw	$0xaa, %mm4, %mm0
 	pshufw	$0xff, %mm4, %mm2
 
@@ -681,11 +594,6 @@ _generic_read_RGBA_span_RGB565_MMX:
 	pmulhuw	%mm7, %mm0
 	pmulhuw	%mm7, %mm2
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	por	alpha, %mm0
-	por	alpha, %mm2
- */
  	por %mm3, %mm0
  	por %mm3, %mm2
 
@@ -724,11 +632,6 @@ _generic_read_RGBA_span_RGB565_MMX:
 	pmulhuw	%mm7, %mm0
 	pmulhuw	%mm7, %mm2
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	por	alpha, %mm0
-	por	alpha, %mm2
- */
  	por %mm3, %mm0
  	por %mm3, %mm2
 
@@ -757,10 +660,6 @@ _generic_read_RGBA_span_RGB565_MMX:
 #endif
 	pmulhuw	%mm7, %mm0
 
-/* Kevin F. Quinn 2nd July 2006
- * Replace data segment constants with text-segment instructions
-	por	alpha, %mm0
- */
  	por %mm3, %mm0
 
 	packuswb	%mm0, %mm0
diff --git a/src/mesa/x86/rtasm/x86sse.c b/src/mesa/x86/rtasm/x86sse.c
index 5c4bab73312..5aedf5b04be 100644
--- a/src/mesa/x86/rtasm/x86sse.c
+++ b/src/mesa/x86/rtasm/x86sse.c
@@ -1162,11 +1162,12 @@ void x86_init_func( struct x86_function *p )
    p->csr = p->store;
 }
 
-void x86_init_func_size( struct x86_function *p, unsigned code_size )
+int x86_init_func_size( struct x86_function *p, unsigned code_size )
 {
    p->size = code_size;
    p->store = _mesa_exec_malloc(code_size);
    p->csr = p->store;
+   return p->store != NULL;
 }
 
 void x86_release_func( struct x86_function *p )
diff --git a/src/mesa/x86/rtasm/x86sse.h b/src/mesa/x86/rtasm/x86sse.h
index c2aa416492e..f6282f5bd41 100644
--- a/src/mesa/x86/rtasm/x86sse.h
+++ b/src/mesa/x86/rtasm/x86sse.h
@@ -80,7 +80,7 @@ enum sse_cc {
 
 
 void x86_init_func( struct x86_function *p );
-void x86_init_func_size( struct x86_function *p, unsigned code_size );
+int x86_init_func_size( struct x86_function *p, unsigned code_size );
 void x86_release_func( struct x86_function *p );
 void (*x86_get_func( struct x86_function *p ))( void );